diff --git a/.gitattributes b/.gitattributes
index 3408a80ee5684749df039103ff0dbc0fc63b144c..79c5f6eef6d8ea529a5101de3ff81c6427ad2e09 100644
--- a/.gitattributes
+++ b/.gitattributes
@@ -219,3 +219,6 @@ logs/13849650-bench_80G_dp16_tp1_pp16_acc16_mbs1_seq4096_zero1_tpmodeRED_vocab13
 logs/14099270-bench_3.57G_dp2_tp8_pp16_acc64_mbs2_seq4096_zero1_tpmodeRED_vocab131k.out filter=lfs diff=lfs merge=lfs -text
 logs/13849961-bench_8.86G_dp8_tp32_pp2_acc8_mbs4_seq4096_zero1_tpmodeRED_vocab131k.out filter=lfs diff=lfs merge=lfs -text
 logs/13850100-bench_80G_dp256_tp1_pp2_acc1_mbs1_seq4096_zero1_tpmodeRED_vocab131k.out filter=lfs diff=lfs merge=lfs -text
+logs/13849632-bench_8.86G_dp8_tp16_pp2_acc4_mbs8_seq4096_zero1_tpmodeRED_vocab131k.out filter=lfs diff=lfs merge=lfs -text
+logs/13849414-bench_3.57G_dp2_tp16_pp8_acc32_mbs4_seq4096_zero1_tpmodeRED_vocab131k.out filter=lfs diff=lfs merge=lfs -text
+logs/14099236-bench_3.57G_dp16_tp4_pp4_acc8_mbs2_seq4096_zero1_tpmodeRED_vocab131k.out filter=lfs diff=lfs merge=lfs -text
diff --git a/logs/12179115-bench_1.07G_dp512_tp1_pp1_acc1_mbs2_seq4096_zero0_l15_h2048_heads16.out b/logs/12179115-bench_1.07G_dp512_tp1_pp1_acc1_mbs2_seq4096_zero0_l15_h2048_heads16.out
new file mode 100644
index 0000000000000000000000000000000000000000..3d656485f889f261d94a2a364b4f219e9611e1f6
--- /dev/null
+++ b/logs/12179115-bench_1.07G_dp512_tp1_pp1_acc1_mbs2_seq4096_zero0_l15_h2048_heads16.out
@@ -0,0 +1,3247 @@
++ source /etc/profile.d/modules.sh
+++ . /usr/share/modules/init/bash
++++ unset _mlshdbg
++++ '[' 0 = 1 ']'
++++ unset _mlre _mlIFS
++++ '[' -n x ']'
++++ _mlIFS=' 	
+'
++++ IFS=' '
++++ '[' -n '' ']'
+++++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash autoinit
++++ _mlcode='module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+MODULEPATH_modshare=/etc/environment-modules/modules:1:/usr/share/modules/\$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1; export MODULEPATH_modshare;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+LOADEDMODULES=; export LOADEDMODULES;
+MODULEPATH=/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/\$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles; export MODULEPATH;
+test 0;'
++++ _mlret=0
++++ '[' -n x ']'
++++ IFS=' 	
+'
++++ unset _mlIFS
++++ unset _mlre _mlv _mlrv
++++ '[' 0 -eq 0 ']'
++++ eval 'module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+MODULEPATH_modshare=/etc/environment-modules/modules:1:/usr/share/modules/\$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1; export MODULEPATH_modshare;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+LOADEDMODULES=; export LOADEDMODULES;
+MODULEPATH=/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/\$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles; export MODULEPATH;
+test 0;'
+++++ MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl
+++++ export MODULES_CMD
+++++ ENV=/usr/share/modules/init/profile.sh
+++++ export ENV
+++++ MODULEPATH_modshare='/etc/environment-modules/modules:1:/usr/share/modules/$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1'
+++++ export MODULEPATH_modshare
+++++ BASH_ENV=/usr/share/modules/init/bash
+++++ export BASH_ENV
+++++ MODULESHOME=/usr/share/modules
+++++ export MODULESHOME
+++++ LOADEDMODULES=
+++++ export LOADEDMODULES
+++++ MODULEPATH='/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles'
+++++ export MODULEPATH
+++++ test 0
++++ '[' 0 = 1 ']'
++++ '[' -t 2 ']'
++++ export -f module
++++ export -f switchml
++++ '[' 5 -ge 3 ']'
++++ [[ ehxB =~ i ]]
++++ [[ ! :/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin: =~ :/usr/bin: ]]
++++ '[' '!' -n '' ']'
++++ MANPATH=:
++++ export MANPATH
+++++ manpath
++++ [[ ! :/admin/home/nouamane/.local/share/man:/fsx/nouamane/miniconda/envs/2-1-cu121/man:/fsx/nouamane/miniconda/envs/2-1-cu121/share/man:/fsx/nouamane/miniconda/man:/fsx/nouamane/miniconda/share/man:/opt/amazon/openmpi/share/man:/opt/amazon/efa/share/man:/opt/slurm/share/man:/usr/local/man:/usr/local/share/man:/usr/share/man:/admin/home/nouamane/.fzf/man:: =~ :/usr/share/man: ]]
++++ unset _mlcode _mlret
++++ '[' -n '' ']'
++ module load cuda/12.1
++ unset _mlshdbg
++ '[' 0 = 1 ']'
++ unset _mlre _mlIFS
++ '[' -n x ']'
++ _mlIFS=' 	
+'
++ IFS=' '
++ '[' -n '' ']'
+++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash load cuda/12.1
++ eval 'CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include;' export 'CPATH;
+LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:;' export 'LD_LIBRARY_PATH;
+FI_EFA_FORK_SAFE=1;' export 'FI_EFA_FORK_SAFE;
+MANPATH=/usr/local/cuda-12.1/share/man::;' export 'MANPATH;
+LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64;' export 'LIBRARY_PATH;
+_LMFILES_=/usr/share/modules/modulefiles/cuda/12.1;' export '_LMFILES_;
+LOADEDMODULES=cuda/12.1;' export 'LOADEDMODULES;
+MPI_PATH=/opt/amazon/openmpi;' export 'MPI_PATH;
+NCCL_HOME_modshare=/opt/nccl/build:1;' export 'NCCL_HOME_modshare;
+NCCL_PROTO=simple;' export 'NCCL_PROTO;
+MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1;' export 'MANPATH_modshare;
+LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1;' export 'LIBRARY_PATH_modshare;
+NCCL_SOCKET_IFNAME=enp;' export 'NCCL_SOCKET_IFNAME;
+AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl;' export 'AWS_OFI_NCCL_HOME;
+NCCL_HOME=/opt/nccl/build;' export 'NCCL_HOME;
+FI_PROVIDER=efa;' export 'FI_PROVIDER;
+AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1;' export 'AWS_OFI_NCCL_HOME_modshare;
+CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1;' export 'CPATH_modshare;
+LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2::1:/opt/nccl/build/lib:1:/opt/aws-ofi-nccl/lib:2:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/efa/lib:1;' export 'LD_LIBRARY_PATH_modshare;
+FI_EFA_ENABLE_SHM_TRANSFER=1;' export 'FI_EFA_ENABLE_SHM_TRANSFER;
+_LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1;' export '_LMFILES__modshare;
+LOADEDMODULES_modshare=cuda/12.1:1;' export 'LOADEDMODULES_modshare;
+MPI_PATH_modshare=/opt/amazon/openmpi:1;' export 'MPI_PATH_modshare;
+PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin;' export 'PATH;
+CUDA_HOME=/usr/local/cuda-12.1;' export 'CUDA_HOME;
+PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/usr/local/cuda-12.1/include:1:/opt/amazon/efa/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/usr/sbin:1:/fsx/nouamane/miniconda/bin:1:/opt/slurm/bin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:2:/opt/amazon/openmpi/bin:1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1;' export 'PATH_modshare;
+test' '0;'
+++ CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include
+++ export CPATH
+++ LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:
+++ export LD_LIBRARY_PATH
+++ FI_EFA_FORK_SAFE=1
+++ export FI_EFA_FORK_SAFE
+++ MANPATH=/usr/local/cuda-12.1/share/man::
+++ export MANPATH
+++ LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64
+++ export LIBRARY_PATH
+++ _LMFILES_=/usr/share/modules/modulefiles/cuda/12.1
+++ export _LMFILES_
+++ LOADEDMODULES=cuda/12.1
+++ export LOADEDMODULES
+++ MPI_PATH=/opt/amazon/openmpi
+++ export MPI_PATH
+++ NCCL_HOME_modshare=/opt/nccl/build:1
+++ export NCCL_HOME_modshare
+++ NCCL_PROTO=simple
+++ export NCCL_PROTO
+++ MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1
+++ export MANPATH_modshare
+++ LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1
+++ export LIBRARY_PATH_modshare
+++ NCCL_SOCKET_IFNAME=enp
+++ export NCCL_SOCKET_IFNAME
+++ AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl
+++ export AWS_OFI_NCCL_HOME
+++ NCCL_HOME=/opt/nccl/build
+++ export NCCL_HOME
+++ FI_PROVIDER=efa
+++ export FI_PROVIDER
+++ AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1
+++ export AWS_OFI_NCCL_HOME_modshare
+++ CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1
+++ export CPATH_modshare
+++ LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2::1:/opt/nccl/build/lib:1:/opt/aws-ofi-nccl/lib:2:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/efa/lib:1
+++ export LD_LIBRARY_PATH_modshare
+++ FI_EFA_ENABLE_SHM_TRANSFER=1
+++ export FI_EFA_ENABLE_SHM_TRANSFER
+++ _LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1
+++ export _LMFILES__modshare
+++ LOADEDMODULES_modshare=cuda/12.1:1
+++ export LOADEDMODULES_modshare
+++ MPI_PATH_modshare=/opt/amazon/openmpi:1
+++ export MPI_PATH_modshare
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ export PATH
+++ CUDA_HOME=/usr/local/cuda-12.1
+++ export CUDA_HOME
+++ PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/usr/local/cuda-12.1/include:1:/opt/amazon/efa/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/usr/sbin:1:/fsx/nouamane/miniconda/bin:1:/opt/slurm/bin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:2:/opt/amazon/openmpi/bin:1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1
+++ export PATH_modshare
+++ test 0
++ _mlstatus=0
++ '[' -n x ']'
++ IFS=' 	
+'
++ unset _mlre _mlv _mlrv _mlIFS
++ '[' -n '' ']'
++ unset _mlshdbg
++ return 0
++ source /fsx/nouamane/miniconda/bin/activate
+++ _CONDA_ROOT=/fsx/nouamane/miniconda
+++ . /fsx/nouamane/miniconda/etc/profile.d/conda.sh
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ '[' -z x ']'
+++ conda activate
+++ local cmd=activate
+++ case "$cmd" in
+++ __conda_activate activate
+++ '[' -n '' ']'
+++ local ask_conda
++++ PS1=
++++ __conda_exe shell.posix activate
++++ /fsx/nouamane/miniconda/bin/conda shell.posix activate
+++ ask_conda='. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_2='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\'''
+++ eval '. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_2='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\'''
++++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh
+++++ test -n ''
+++++ unset XML_CATALOG_FILES
+++++ unset xml_catalog_files_libxml2
++++ PS1='(base) '
++++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++++ export CONDA_PREFIX=/fsx/nouamane/miniconda
++++ CONDA_PREFIX=/fsx/nouamane/miniconda
++++ export CONDA_SHLVL=3
++++ CONDA_SHLVL=3
++++ export CONDA_DEFAULT_ENV=base
++++ CONDA_DEFAULT_ENV=base
++++ export 'CONDA_PROMPT_MODIFIER=(base) '
++++ CONDA_PROMPT_MODIFIER='(base) '
++++ export CONDA_PREFIX_2=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ CONDA_PREFIX_2=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ __conda_hashr
+++ '[' -n '' ']'
+++ '[' -n '' ']'
+++ hash -r
++ conda activate 2-1-cu121
++ local cmd=activate
++ case "$cmd" in
++ __conda_activate activate 2-1-cu121
++ '[' -n '' ']'
++ local ask_conda
+++ PS1='(base) '
+++ __conda_exe shell.posix activate 2-1-cu121
+++ /fsx/nouamane/miniconda/bin/conda shell.posix activate 2-1-cu121
++ ask_conda='PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''4'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_3='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
++ eval 'PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''4'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_3='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
+++ PS1='(2-1-cu121) '
+++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ export CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ export CONDA_SHLVL=4
+++ CONDA_SHLVL=4
+++ export CONDA_DEFAULT_ENV=2-1-cu121
+++ CONDA_DEFAULT_ENV=2-1-cu121
+++ export 'CONDA_PROMPT_MODIFIER=(2-1-cu121) '
+++ CONDA_PROMPT_MODIFIER='(2-1-cu121) '
+++ export CONDA_PREFIX_3=/fsx/nouamane/miniconda
+++ CONDA_PREFIX_3=/fsx/nouamane/miniconda
+++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ export _CE_M=
+++ _CE_M=
+++ export _CE_CONDA=
+++ _CE_CONDA=
+++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh
++++ test -n ''
++++ xml_catalog_files_libxml2=
++++ XML_CATALOG_FILES=
++++ conda_catalog_files=
++++ ifs_libxml2=' 	
+'
++++ IFS=' '
++++ rem=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ for pre in ${rem}
++++ test '' = /fsx/nouamane/miniconda/envs/2-1-cu121
++++ conda_catalog_files=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ rem=
++++ IFS=' 	
+'
++++ conda_catalog_files='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ export 'XML_CATALOG_FILES=file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ XML_CATALOG_FILES='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ unset conda_catalog_files ifs_libxml2 rem
++ __conda_hashr
++ '[' -n '' ']'
++ '[' -n '' ']'
++ hash -r
++ export PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++ PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ scontrol show hostnames 'ip-26-0-160-[225,242],ip-26-0-161-[78,138],ip-26-0-162-[79,180,233],ip-26-0-163-[43,58,226,236],ip-26-0-164-[0,18,45,75,187,207,236],ip-26-0-165-[24,38,59,131,164,202,213],ip-26-0-166-[15,36,125,214,244],ip-26-0-167-[9,51,175,177,217],ip-26-0-168-[30,34,52,95,120,238],ip-26-0-170-[132,143,160],ip-26-0-171-21,ip-26-0-172-[57,73,116,142,147],ip-26-0-173-[7,202,246],ip-26-0-174-[36,100,186,196,240],ip-26-0-175-[19,34,132,165,170,241]'
++ export 'NODELIST=ip-26-0-160-225
+ip-26-0-160-242
+ip-26-0-161-78
+ip-26-0-161-138
+ip-26-0-162-79
+ip-26-0-162-180
+ip-26-0-162-233
+ip-26-0-163-43
+ip-26-0-163-58
+ip-26-0-163-226
+ip-26-0-163-236
+ip-26-0-164-0
+ip-26-0-164-18
+ip-26-0-164-45
+ip-26-0-164-75
+ip-26-0-164-187
+ip-26-0-164-207
+ip-26-0-164-236
+ip-26-0-165-24
+ip-26-0-165-38
+ip-26-0-165-59
+ip-26-0-165-131
+ip-26-0-165-164
+ip-26-0-165-202
+ip-26-0-165-213
+ip-26-0-166-15
+ip-26-0-166-36
+ip-26-0-166-125
+ip-26-0-166-214
+ip-26-0-166-244
+ip-26-0-167-9
+ip-26-0-167-51
+ip-26-0-167-175
+ip-26-0-167-177
+ip-26-0-167-217
+ip-26-0-168-30
+ip-26-0-168-34
+ip-26-0-168-52
+ip-26-0-168-95
+ip-26-0-168-120
+ip-26-0-168-238
+ip-26-0-170-132
+ip-26-0-170-143
+ip-26-0-170-160
+ip-26-0-171-21
+ip-26-0-172-57
+ip-26-0-172-73
+ip-26-0-172-116
+ip-26-0-172-142
+ip-26-0-172-147
+ip-26-0-173-7
+ip-26-0-173-202
+ip-26-0-173-246
+ip-26-0-174-36
+ip-26-0-174-100
+ip-26-0-174-186
+ip-26-0-174-196
+ip-26-0-174-240
+ip-26-0-175-19
+ip-26-0-175-34
+ip-26-0-175-132
+ip-26-0-175-165
+ip-26-0-175-170
+ip-26-0-175-241'
++ NODELIST='ip-26-0-160-225
+ip-26-0-160-242
+ip-26-0-161-78
+ip-26-0-161-138
+ip-26-0-162-79
+ip-26-0-162-180
+ip-26-0-162-233
+ip-26-0-163-43
+ip-26-0-163-58
+ip-26-0-163-226
+ip-26-0-163-236
+ip-26-0-164-0
+ip-26-0-164-18
+ip-26-0-164-45
+ip-26-0-164-75
+ip-26-0-164-187
+ip-26-0-164-207
+ip-26-0-164-236
+ip-26-0-165-24
+ip-26-0-165-38
+ip-26-0-165-59
+ip-26-0-165-131
+ip-26-0-165-164
+ip-26-0-165-202
+ip-26-0-165-213
+ip-26-0-166-15
+ip-26-0-166-36
+ip-26-0-166-125
+ip-26-0-166-214
+ip-26-0-166-244
+ip-26-0-167-9
+ip-26-0-167-51
+ip-26-0-167-175
+ip-26-0-167-177
+ip-26-0-167-217
+ip-26-0-168-30
+ip-26-0-168-34
+ip-26-0-168-52
+ip-26-0-168-95
+ip-26-0-168-120
+ip-26-0-168-238
+ip-26-0-170-132
+ip-26-0-170-143
+ip-26-0-170-160
+ip-26-0-171-21
+ip-26-0-172-57
+ip-26-0-172-73
+ip-26-0-172-116
+ip-26-0-172-142
+ip-26-0-172-147
+ip-26-0-173-7
+ip-26-0-173-202
+ip-26-0-173-246
+ip-26-0-174-36
+ip-26-0-174-100
+ip-26-0-174-186
+ip-26-0-174-196
+ip-26-0-174-240
+ip-26-0-175-19
+ip-26-0-175-34
+ip-26-0-175-132
+ip-26-0-175-165
+ip-26-0-175-170
+ip-26-0-175-241'
+++ head -n1
+++ scontrol show hostnames 'ip-26-0-160-[225,242],ip-26-0-161-[78,138],ip-26-0-162-[79,180,233],ip-26-0-163-[43,58,226,236],ip-26-0-164-[0,18,45,75,187,207,236],ip-26-0-165-[24,38,59,131,164,202,213],ip-26-0-166-[15,36,125,214,244],ip-26-0-167-[9,51,175,177,217],ip-26-0-168-[30,34,52,95,120,238],ip-26-0-170-[132,143,160],ip-26-0-171-21,ip-26-0-172-[57,73,116,142,147],ip-26-0-173-[7,202,246],ip-26-0-174-[36,100,186,196,240],ip-26-0-175-[19,34,132,165,170,241]'
++ export MASTER_NODE=ip-26-0-160-225
++ MASTER_NODE=ip-26-0-160-225
++ export MASTER_PORT=12356
++ MASTER_PORT=12356
++ export NNODES=64
++ NNODES=64
++ export GPUS_PER_NODE=8
++ GPUS_PER_NODE=8
++ export WORLD_SIZE=512
++ WORLD_SIZE=512
++ export CUDA_DEVICE_MAX_CONNECTIONS=1
++ CUDA_DEVICE_MAX_CONNECTIONS=1
++ export NANOTRON_BENCHMARK=1
++ NANOTRON_BENCHMARK=1
++ echo 'Master node: ip-26-0-160-225'
+Master node: ip-26-0-160-225
++ echo 'All nodes: ip-26-0-160-225
+ip-26-0-160-242
+ip-26-0-161-78
+ip-26-0-161-138
+ip-26-0-162-79
+ip-26-0-162-180
+ip-26-0-162-233
+ip-26-0-163-43
+ip-26-0-163-58
+ip-26-0-163-226
+ip-26-0-163-236
+ip-26-0-164-0
+ip-26-0-164-18
+ip-26-0-164-45
+ip-26-0-164-75
+ip-26-0-164-187
+ip-26-0-164-207
+ip-26-0-164-236
+ip-26-0-165-24
+ip-26-0-165-38
+ip-26-0-165-59
+ip-26-0-165-131
+ip-26-0-165-164
+ip-26-0-165-202
+ip-26-0-165-213
+ip-26-0-166-15
+ip-26-0-166-36
+ip-26-0-166-125
+ip-26-0-166-214
+ip-26-0-166-244
+ip-26-0-167-9
+ip-26-0-167-51
+ip-26-0-167-175
+ip-26-0-167-177
+ip-26-0-167-217
+ip-26-0-168-30
+ip-26-0-168-34
+ip-26-0-168-52
+ip-26-0-168-95
+ip-26-0-168-120
+ip-26-0-168-238
+ip-26-0-170-132
+ip-26-0-170-143
+ip-26-0-170-160
+ip-26-0-171-21
+ip-26-0-172-57
+ip-26-0-172-73
+ip-26-0-172-116
+ip-26-0-172-142
+ip-26-0-172-147
+ip-26-0-173-7
+ip-26-0-173-202
+ip-26-0-173-246
+ip-26-0-174-36
+ip-26-0-174-100
+ip-26-0-174-186
+ip-26-0-174-196
+ip-26-0-174-240
+ip-26-0-175-19
+ip-26-0-175-34
+ip-26-0-175-132
+ip-26-0-175-165
+ip-26-0-175-170
+ip-26-0-175-241'
+All nodes: ip-26-0-160-225
+ip-26-0-160-242
+ip-26-0-161-78
+ip-26-0-161-138
+ip-26-0-162-79
+ip-26-0-162-180
+ip-26-0-162-233
+ip-26-0-163-43
+ip-26-0-163-58
+ip-26-0-163-226
+ip-26-0-163-236
+ip-26-0-164-0
+ip-26-0-164-18
+ip-26-0-164-45
+ip-26-0-164-75
+ip-26-0-164-187
+ip-26-0-164-207
+ip-26-0-164-236
+ip-26-0-165-24
+ip-26-0-165-38
+ip-26-0-165-59
+ip-26-0-165-131
+ip-26-0-165-164
+ip-26-0-165-202
+ip-26-0-165-213
+ip-26-0-166-15
+ip-26-0-166-36
+ip-26-0-166-125
+ip-26-0-166-214
+ip-26-0-166-244
+ip-26-0-167-9
+ip-26-0-167-51
+ip-26-0-167-175
+ip-26-0-167-177
+ip-26-0-167-217
+ip-26-0-168-30
+ip-26-0-168-34
+ip-26-0-168-52
+ip-26-0-168-95
+ip-26-0-168-120
+ip-26-0-168-238
+ip-26-0-170-132
+ip-26-0-170-143
+ip-26-0-170-160
+ip-26-0-171-21
+ip-26-0-172-57
+ip-26-0-172-73
+ip-26-0-172-116
+ip-26-0-172-142
+ip-26-0-172-147
+ip-26-0-173-7
+ip-26-0-173-202
+ip-26-0-173-246
+ip-26-0-174-36
+ip-26-0-174-100
+ip-26-0-174-186
+ip-26-0-174-196
+ip-26-0-174-240
+ip-26-0-175-19
+ip-26-0-175-34
+ip-26-0-175-132
+ip-26-0-175-165
+ip-26-0-175-170
+ip-26-0-175-241
++ echo 'World size: 512'
+World size: 512
++ srun torchrun --nnodes=64 --nproc_per_node=8 --rdzv_id=12179115 --rdzv_backend=c10d --rdzv_endpoint=ip-26-0-160-225:12356 run_train.py --config-file benchmark/configs/config_1.07G_dp512_tp1_pp1_acc1_mbs2_seq4096_zero0_l15_h2048_heads16.yaml
+[2024-12-02 11:52:28,025] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-02 11:52:28,025] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-02 11:52:28,025] torch.distributed.run: [WARNING] 
+[2024-12-02 11:52:28,025] torch.distributed.run: [WARNING] *****************************************
+[2024-12-02 11:52:28,025] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-02 11:52:28,025] torch.distributed.run: [WARNING] *****************************************
+[2024-12-02 11:52:28,025] torch.distributed.run: [WARNING] 
+[2024-12-02 11:52:28,025] torch.distributed.run: [WARNING] *****************************************
+[2024-12-02 11:52:28,025] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-02 11:52:28,025] torch.distributed.run: [WARNING] *****************************************
+[2024-12-02 11:52:28,033] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-02 11:52:28,033] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-02 11:52:28,033] torch.distributed.run: [WARNING] 
+[2024-12-02 11:52:28,033] torch.distributed.run: [WARNING] *****************************************
+[2024-12-02 11:52:28,033] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-02 11:52:28,033] torch.distributed.run: [WARNING] *****************************************
+[2024-12-02 11:52:28,033] torch.distributed.run: [WARNING] 
+[2024-12-02 11:52:28,033] torch.distributed.run: [WARNING] *****************************************
+[2024-12-02 11:52:28,033] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-02 11:52:28,033] torch.distributed.run: [WARNING] *****************************************
+[2024-12-02 11:52:28,033] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-02 11:52:28,033] torch.distributed.run: [WARNING] 
+[2024-12-02 11:52:28,033] torch.distributed.run: [WARNING] *****************************************
+[2024-12-02 11:52:28,033] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-02 11:52:28,033] torch.distributed.run: [WARNING] *****************************************
+[2024-12-02 11:52:28,034] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-02 11:52:28,034] torch.distributed.run: [WARNING] 
+[2024-12-02 11:52:28,034] torch.distributed.run: [WARNING] *****************************************
+[2024-12-02 11:52:28,034] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-02 11:52:28,034] torch.distributed.run: [WARNING] *****************************************
+[2024-12-02 11:52:28,033] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-02 11:52:28,033] torch.distributed.run: [WARNING] 
+[2024-12-02 11:52:28,033] torch.distributed.run: [WARNING] *****************************************
+[2024-12-02 11:52:28,033] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-02 11:52:28,033] torch.distributed.run: [WARNING] *****************************************
+[2024-12-02 11:52:28,034] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-02 11:52:28,034] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-02 11:52:28,034] torch.distributed.run: [WARNING] 
+[2024-12-02 11:52:28,034] torch.distributed.run: [WARNING] *****************************************
+[2024-12-02 11:52:28,034] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-02 11:52:28,034] torch.distributed.run: [WARNING] *****************************************
+[2024-12-02 11:52:28,034] torch.distributed.run: [WARNING] 
+[2024-12-02 11:52:28,034] torch.distributed.run: [WARNING] *****************************************
+[2024-12-02 11:52:28,034] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-02 11:52:28,034] torch.distributed.run: [WARNING] *****************************************
+[2024-12-02 11:52:28,035] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-02 11:52:28,035] torch.distributed.run: [WARNING] 
+[2024-12-02 11:52:28,035] torch.distributed.run: [WARNING] *****************************************
+[2024-12-02 11:52:28,035] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-02 11:52:28,035] torch.distributed.run: [WARNING] *****************************************
+[2024-12-02 11:52:28,037] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-02 11:52:28,037] torch.distributed.run: [WARNING] 
+[2024-12-02 11:52:28,037] torch.distributed.run: [WARNING] *****************************************
+[2024-12-02 11:52:28,037] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-02 11:52:28,037] torch.distributed.run: [WARNING] *****************************************
+[2024-12-02 11:52:28,037] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-02 11:52:28,037] torch.distributed.run: [WARNING] 
+[2024-12-02 11:52:28,037] torch.distributed.run: [WARNING] *****************************************
+[2024-12-02 11:52:28,037] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-02 11:52:28,037] torch.distributed.run: [WARNING] *****************************************
+[2024-12-02 11:52:28,044] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-02 11:52:28,044] torch.distributed.run: [WARNING] 
+[2024-12-02 11:52:28,044] torch.distributed.run: [WARNING] *****************************************
+[2024-12-02 11:52:28,044] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-02 11:52:28,044] torch.distributed.run: [WARNING] *****************************************
+[2024-12-02 11:52:28,044] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-02 11:52:28,044] torch.distributed.run: [WARNING] 
+[2024-12-02 11:52:28,044] torch.distributed.run: [WARNING] *****************************************
+[2024-12-02 11:52:28,044] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-02 11:52:28,044] torch.distributed.run: [WARNING] *****************************************
+[2024-12-02 11:52:28,047] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-02 11:52:28,047] torch.distributed.run: [WARNING] 
+[2024-12-02 11:52:28,047] torch.distributed.run: [WARNING] *****************************************
+[2024-12-02 11:52:28,047] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-02 11:52:28,047] torch.distributed.run: [WARNING] *****************************************
+[2024-12-02 11:52:28,048] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-02 11:52:28,048] torch.distributed.run: [WARNING] 
+[2024-12-02 11:52:28,048] torch.distributed.run: [WARNING] *****************************************
+[2024-12-02 11:52:28,048] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-02 11:52:28,048] torch.distributed.run: [WARNING] *****************************************
+[2024-12-02 11:52:28,049] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-02 11:52:28,049] torch.distributed.run: [WARNING] 
+[2024-12-02 11:52:28,049] torch.distributed.run: [WARNING] *****************************************
+[2024-12-02 11:52:28,049] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-02 11:52:28,049] torch.distributed.run: [WARNING] *****************************************
+[2024-12-02 11:52:28,057] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-02 11:52:28,057] torch.distributed.run: [WARNING] 
+[2024-12-02 11:52:28,057] torch.distributed.run: [WARNING] *****************************************
+[2024-12-02 11:52:28,057] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-02 11:52:28,057] torch.distributed.run: [WARNING] *****************************************
+[2024-12-02 11:52:28,061] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-02 11:52:28,061] torch.distributed.run: [WARNING] 
+[2024-12-02 11:52:28,061] torch.distributed.run: [WARNING] *****************************************
+[2024-12-02 11:52:28,061] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-02 11:52:28,061] torch.distributed.run: [WARNING] *****************************************
+[2024-12-02 11:52:28,061] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-02 11:52:28,061] torch.distributed.run: [WARNING] 
+[2024-12-02 11:52:28,061] torch.distributed.run: [WARNING] *****************************************
+[2024-12-02 11:52:28,061] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-02 11:52:28,061] torch.distributed.run: [WARNING] *****************************************
+[2024-12-02 11:52:28,063] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-02 11:52:28,063] torch.distributed.run: [WARNING] 
+[2024-12-02 11:52:28,063] torch.distributed.run: [WARNING] *****************************************
+[2024-12-02 11:52:28,063] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-02 11:52:28,063] torch.distributed.run: [WARNING] *****************************************
+[2024-12-02 11:52:28,064] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-02 11:52:28,064] torch.distributed.run: [WARNING] 
+[2024-12-02 11:52:28,064] torch.distributed.run: [WARNING] *****************************************
+[2024-12-02 11:52:28,064] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-02 11:52:28,064] torch.distributed.run: [WARNING] *****************************************
+[2024-12-02 11:52:28,064] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-02 11:52:28,064] torch.distributed.run: [WARNING] 
+[2024-12-02 11:52:28,064] torch.distributed.run: [WARNING] *****************************************
+[2024-12-02 11:52:28,064] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-02 11:52:28,064] torch.distributed.run: [WARNING] *****************************************
+[2024-12-02 11:52:28,066] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-02 11:52:28,066] torch.distributed.run: [WARNING] 
+[2024-12-02 11:52:28,066] torch.distributed.run: [WARNING] *****************************************
+[2024-12-02 11:52:28,066] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-02 11:52:28,066] torch.distributed.run: [WARNING] *****************************************
+[2024-12-02 11:52:28,067] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-02 11:52:28,068] torch.distributed.run: [WARNING] 
+[2024-12-02 11:52:28,068] torch.distributed.run: [WARNING] *****************************************
+[2024-12-02 11:52:28,068] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-02 11:52:28,068] torch.distributed.run: [WARNING] *****************************************
+[2024-12-02 11:52:28,069] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-02 11:52:28,069] torch.distributed.run: [WARNING] 
+[2024-12-02 11:52:28,069] torch.distributed.run: [WARNING] *****************************************
+[2024-12-02 11:52:28,069] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-02 11:52:28,069] torch.distributed.run: [WARNING] *****************************************
+[2024-12-02 11:52:28,076] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-02 11:52:28,076] torch.distributed.run: [WARNING] 
+[2024-12-02 11:52:28,076] torch.distributed.run: [WARNING] *****************************************
+[2024-12-02 11:52:28,076] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-02 11:52:28,076] torch.distributed.run: [WARNING] *****************************************
+[2024-12-02 11:52:28,075] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-02 11:52:28,075] torch.distributed.run: [WARNING] 
+[2024-12-02 11:52:28,075] torch.distributed.run: [WARNING] *****************************************
+[2024-12-02 11:52:28,075] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-02 11:52:28,075] torch.distributed.run: [WARNING] *****************************************
+[2024-12-02 11:52:28,076] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-02 11:52:28,076] torch.distributed.run: [WARNING] 
+[2024-12-02 11:52:28,076] torch.distributed.run: [WARNING] *****************************************
+[2024-12-02 11:52:28,076] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-02 11:52:28,076] torch.distributed.run: [WARNING] *****************************************
+[2024-12-02 11:52:28,075] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-02 11:52:28,075] torch.distributed.run: [WARNING] 
+[2024-12-02 11:52:28,075] torch.distributed.run: [WARNING] *****************************************
+[2024-12-02 11:52:28,075] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-02 11:52:28,075] torch.distributed.run: [WARNING] *****************************************
+[2024-12-02 11:52:28,080] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-02 11:52:28,080] torch.distributed.run: [WARNING] 
+[2024-12-02 11:52:28,080] torch.distributed.run: [WARNING] *****************************************
+[2024-12-02 11:52:28,080] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-02 11:52:28,080] torch.distributed.run: [WARNING] *****************************************
+[2024-12-02 11:52:28,081] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-02 11:52:28,081] torch.distributed.run: [WARNING] 
+[2024-12-02 11:52:28,081] torch.distributed.run: [WARNING] *****************************************
+[2024-12-02 11:52:28,081] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-02 11:52:28,081] torch.distributed.run: [WARNING] *****************************************
+[2024-12-02 11:52:28,082] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-02 11:52:28,083] torch.distributed.run: [WARNING] 
+[2024-12-02 11:52:28,083] torch.distributed.run: [WARNING] *****************************************
+[2024-12-02 11:52:28,083] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-02 11:52:28,083] torch.distributed.run: [WARNING] *****************************************
+[2024-12-02 11:52:28,084] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-02 11:52:28,084] torch.distributed.run: [WARNING] 
+[2024-12-02 11:52:28,084] torch.distributed.run: [WARNING] *****************************************
+[2024-12-02 11:52:28,084] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-02 11:52:28,084] torch.distributed.run: [WARNING] *****************************************
+[2024-12-02 11:52:28,084] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-02 11:52:28,084] torch.distributed.run: [WARNING] 
+[2024-12-02 11:52:28,084] torch.distributed.run: [WARNING] *****************************************
+[2024-12-02 11:52:28,084] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-02 11:52:28,084] torch.distributed.run: [WARNING] *****************************************
+[2024-12-02 11:52:28,087] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-02 11:52:28,087] torch.distributed.run: [WARNING] 
+[2024-12-02 11:52:28,087] torch.distributed.run: [WARNING] *****************************************
+[2024-12-02 11:52:28,087] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-02 11:52:28,087] torch.distributed.run: [WARNING] *****************************************
+[2024-12-02 11:52:28,087] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-02 11:52:28,087] torch.distributed.run: [WARNING] 
+[2024-12-02 11:52:28,087] torch.distributed.run: [WARNING] *****************************************
+[2024-12-02 11:52:28,087] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-02 11:52:28,087] torch.distributed.run: [WARNING] *****************************************
+[2024-12-02 11:52:28,088] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-02 11:52:28,088] torch.distributed.run: [WARNING] 
+[2024-12-02 11:52:28,088] torch.distributed.run: [WARNING] *****************************************
+[2024-12-02 11:52:28,088] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-02 11:52:28,088] torch.distributed.run: [WARNING] *****************************************
+[2024-12-02 11:52:28,090] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-02 11:52:28,090] torch.distributed.run: [WARNING] 
+[2024-12-02 11:52:28,090] torch.distributed.run: [WARNING] *****************************************
+[2024-12-02 11:52:28,090] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-02 11:52:28,090] torch.distributed.run: [WARNING] *****************************************
+[2024-12-02 11:52:28,091] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-02 11:52:28,091] torch.distributed.run: [WARNING] 
+[2024-12-02 11:52:28,091] torch.distributed.run: [WARNING] *****************************************
+[2024-12-02 11:52:28,091] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-02 11:52:28,091] torch.distributed.run: [WARNING] *****************************************
+[2024-12-02 11:52:28,094] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-02 11:52:28,094] torch.distributed.run: [WARNING] 
+[2024-12-02 11:52:28,094] torch.distributed.run: [WARNING] *****************************************
+[2024-12-02 11:52:28,094] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-02 11:52:28,094] torch.distributed.run: [WARNING] *****************************************
+[2024-12-02 11:52:28,100] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-02 11:52:28,100] torch.distributed.run: [WARNING] 
+[2024-12-02 11:52:28,100] torch.distributed.run: [WARNING] *****************************************
+[2024-12-02 11:52:28,100] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-02 11:52:28,100] torch.distributed.run: [WARNING] *****************************************
+[2024-12-02 11:52:28,101] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-02 11:52:28,101] torch.distributed.run: [WARNING] 
+[2024-12-02 11:52:28,101] torch.distributed.run: [WARNING] *****************************************
+[2024-12-02 11:52:28,101] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-02 11:52:28,101] torch.distributed.run: [WARNING] *****************************************
+[2024-12-02 11:52:28,109] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-02 11:52:28,109] torch.distributed.run: [WARNING] 
+[2024-12-02 11:52:28,109] torch.distributed.run: [WARNING] *****************************************
+[2024-12-02 11:52:28,109] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-02 11:52:28,109] torch.distributed.run: [WARNING] *****************************************
+[2024-12-02 11:52:28,112] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-02 11:52:28,112] torch.distributed.run: [WARNING] 
+[2024-12-02 11:52:28,112] torch.distributed.run: [WARNING] *****************************************
+[2024-12-02 11:52:28,112] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-02 11:52:28,112] torch.distributed.run: [WARNING] *****************************************
+[2024-12-02 11:52:28,112] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-02 11:52:28,112] torch.distributed.run: [WARNING] 
+[2024-12-02 11:52:28,112] torch.distributed.run: [WARNING] *****************************************
+[2024-12-02 11:52:28,112] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-02 11:52:28,112] torch.distributed.run: [WARNING] *****************************************
+[2024-12-02 11:52:28,114] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-02 11:52:28,114] torch.distributed.run: [WARNING] 
+[2024-12-02 11:52:28,114] torch.distributed.run: [WARNING] *****************************************
+[2024-12-02 11:52:28,114] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-02 11:52:28,114] torch.distributed.run: [WARNING] *****************************************
+[2024-12-02 11:52:28,123] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-02 11:52:28,123] torch.distributed.run: [WARNING] 
+[2024-12-02 11:52:28,123] torch.distributed.run: [WARNING] *****************************************
+[2024-12-02 11:52:28,123] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-02 11:52:28,123] torch.distributed.run: [WARNING] *****************************************
+[2024-12-02 11:52:28,123] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-02 11:52:28,123] torch.distributed.run: [WARNING] 
+[2024-12-02 11:52:28,123] torch.distributed.run: [WARNING] *****************************************
+[2024-12-02 11:52:28,123] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-02 11:52:28,123] torch.distributed.run: [WARNING] *****************************************
+[2024-12-02 11:52:28,127] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-02 11:52:28,127] torch.distributed.run: [WARNING] 
+[2024-12-02 11:52:28,127] torch.distributed.run: [WARNING] *****************************************
+[2024-12-02 11:52:28,127] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-02 11:52:28,127] torch.distributed.run: [WARNING] *****************************************
+[2024-12-02 11:52:28,128] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-02 11:52:28,128] torch.distributed.run: [WARNING] 
+[2024-12-02 11:52:28,128] torch.distributed.run: [WARNING] *****************************************
+[2024-12-02 11:52:28,128] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-02 11:52:28,128] torch.distributed.run: [WARNING] *****************************************
+[2024-12-02 11:52:28,126] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-02 11:52:28,126] torch.distributed.run: [WARNING] 
+[2024-12-02 11:52:28,126] torch.distributed.run: [WARNING] *****************************************
+[2024-12-02 11:52:28,126] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-02 11:52:28,126] torch.distributed.run: [WARNING] *****************************************
+[2024-12-02 11:52:28,132] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-02 11:52:28,132] torch.distributed.run: [WARNING] 
+[2024-12-02 11:52:28,132] torch.distributed.run: [WARNING] *****************************************
+[2024-12-02 11:52:28,132] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-02 11:52:28,132] torch.distributed.run: [WARNING] *****************************************
+[2024-12-02 11:52:28,136] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-02 11:52:28,136] torch.distributed.run: [WARNING] 
+[2024-12-02 11:52:28,136] torch.distributed.run: [WARNING] *****************************************
+[2024-12-02 11:52:28,136] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-02 11:52:28,136] torch.distributed.run: [WARNING] *****************************************
+[2024-12-02 11:52:28,136] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-02 11:52:28,136] torch.distributed.run: [WARNING] 
+[2024-12-02 11:52:28,136] torch.distributed.run: [WARNING] *****************************************
+[2024-12-02 11:52:28,136] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-02 11:52:28,136] torch.distributed.run: [WARNING] *****************************************
+[2024-12-02 11:52:28,139] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-02 11:52:28,139] torch.distributed.run: [WARNING] 
+[2024-12-02 11:52:28,139] torch.distributed.run: [WARNING] *****************************************
+[2024-12-02 11:52:28,139] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-02 11:52:28,139] torch.distributed.run: [WARNING] *****************************************
+[2024-12-02 11:52:28,142] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-02 11:52:28,142] torch.distributed.run: [WARNING] 
+[2024-12-02 11:52:28,142] torch.distributed.run: [WARNING] *****************************************
+[2024-12-02 11:52:28,142] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-02 11:52:28,142] torch.distributed.run: [WARNING] *****************************************
+[2024-12-02 11:52:28,145] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-02 11:52:28,145] torch.distributed.run: [WARNING] 
+[2024-12-02 11:52:28,145] torch.distributed.run: [WARNING] *****************************************
+[2024-12-02 11:52:28,145] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-02 11:52:28,145] torch.distributed.run: [WARNING] *****************************************
+[2024-12-02 11:52:28,147] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-02 11:52:28,147] torch.distributed.run: [WARNING] 
+[2024-12-02 11:52:28,147] torch.distributed.run: [WARNING] *****************************************
+[2024-12-02 11:52:28,147] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-02 11:52:28,147] torch.distributed.run: [WARNING] *****************************************
+[2024-12-02 11:52:28,147] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-02 11:52:28,147] torch.distributed.run: [WARNING] 
+[2024-12-02 11:52:28,147] torch.distributed.run: [WARNING] *****************************************
+[2024-12-02 11:52:28,147] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-02 11:52:28,147] torch.distributed.run: [WARNING] *****************************************
+[2024-12-02 11:52:28,148] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-02 11:52:28,149] torch.distributed.run: [WARNING] 
+[2024-12-02 11:52:28,149] torch.distributed.run: [WARNING] *****************************************
+[2024-12-02 11:52:28,149] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-02 11:52:28,149] torch.distributed.run: [WARNING] *****************************************
+[2024-12-02 11:52:28,201] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-02 11:52:28,201] torch.distributed.run: [WARNING] 
+[2024-12-02 11:52:28,201] torch.distributed.run: [WARNING] *****************************************
+[2024-12-02 11:52:28,201] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-02 11:52:28,201] torch.distributed.run: [WARNING] *****************************************
+[2024-12-02 11:52:28,250] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-02 11:52:28,250] torch.distributed.run: [WARNING] 
+[2024-12-02 11:52:28,250] torch.distributed.run: [WARNING] *****************************************
+[2024-12-02 11:52:28,250] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-02 11:52:28,250] torch.distributed.run: [WARNING] *****************************************
+[2024-12-02 11:52:32,656] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-02 11:52:32,656] torch.distributed.run: [WARNING] 
+[2024-12-02 11:52:32,656] torch.distributed.run: [WARNING] *****************************************
+[2024-12-02 11:52:32,656] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-02 11:52:32,656] torch.distributed.run: [WARNING] *****************************************
+12/02/2024 11:53:27 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: Config:
+12/02/2024 11:53:27 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: Config(general=GeneralArgs(project='debug',
+12/02/2024 11:53:27 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                            run='1.07G_dp512_tp1_pp1_acc1_mbs2_seq4096_zero0_l15_h2048_heads16',
+12/02/2024 11:53:27 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                            seed=42,
+12/02/2024 11:53:27 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                            step=None,
+12/02/2024 11:53:27 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                            consumed_train_samples=None,
+12/02/2024 11:53:27 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                            benchmark_csv_path=PosixPath('bench.csv'),
+12/02/2024 11:53:27 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                            ignore_sanity_checks=True),
+12/02/2024 11:53:27 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:        parallelism=ParallelismArgs(dp=512,
+12/02/2024 11:53:27 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                    pp=1,
+12/02/2024 11:53:27 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                    tp=1,
+12/02/2024 11:53:27 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                    pp_engine=<nanotron.parallel.pipeline_parallel.engine.OneForwardOneBackwardPipelineEngine object at 0x7f4367510df0>,
+12/02/2024 11:53:27 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                    tp_mode=<TensorParallelLinearMode.REDUCE_SCATTER: 2>,
+12/02/2024 11:53:27 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                    tp_linear_async_communication=True,
+12/02/2024 11:53:27 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                    recompute_layer=False,
+12/02/2024 11:53:27 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                    tp_recompute_allgather=True,
+12/02/2024 11:53:27 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                    expert_parallel_size=1),
+12/02/2024 11:53:27 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:        model=ModelArgs(model_config=LlamaConfig(bos_token_id=0,
+12/02/2024 11:53:27 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 eos_token_id=0,
+12/02/2024 11:53:27 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 hidden_act='silu',
+12/02/2024 11:53:27 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 hidden_size=2048,
+12/02/2024 11:53:27 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 initializer_range=0.02,
+12/02/2024 11:53:27 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 intermediate_size=8192,
+12/02/2024 11:53:27 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 is_llama_config=True,
+12/02/2024 11:53:27 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 max_position_embeddings=4096,
+12/02/2024 11:53:27 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 num_attention_heads=16,
+12/02/2024 11:53:27 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 num_hidden_layers=15,
+12/02/2024 11:53:27 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 num_key_value_heads=16,
+12/02/2024 11:53:27 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 pad_token_id=None,
+12/02/2024 11:53:27 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 pretraining_tp=1,
+12/02/2024 11:53:27 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 rms_norm_eps=1e-05,
+12/02/2024 11:53:27 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 rope_scaling=None,
+12/02/2024 11:53:27 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 rope_theta=10000.0,
+12/02/2024 11:53:27 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 rope_interleaved=False,
+12/02/2024 11:53:27 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 tie_word_embeddings=True,
+12/02/2024 11:53:27 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 use_cache=True,
+12/02/2024 11:53:27 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 vocab_size=32768),
+12/02/2024 11:53:27 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                        init_method=RandomInit(std=0.02),
+12/02/2024 11:53:27 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                        dtype=torch.bfloat16,
+12/02/2024 11:53:27 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                        make_vocab_size_divisible_by=1,
+12/02/2024 11:53:27 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                        ddp_bucket_cap_mb=25),
+12/02/2024 11:53:27 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:        tokenizer=TokenizerArgs(tokenizer_name_or_path='robot-test/dummy-tokenizer-wordlevel',
+12/02/2024 11:53:27 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                tokenizer_revision=None,
+12/02/2024 11:53:27 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                tokenizer_max_length=None),
+12/02/2024 11:53:27 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:        checkpoints=CheckpointsArgs(checkpoints_path=PosixPath('checkpoints'),
+12/02/2024 11:53:27 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                    checkpoint_interval=10000,
+12/02/2024 11:53:27 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                    save_initial_state=False,
+12/02/2024 11:53:27 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                    save_final_state=False,
+12/02/2024 11:53:27 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                    resume_checkpoint_path=None,
+12/02/2024 11:53:27 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                    checkpoints_path_is_shared_file_system=False),
+12/02/2024 11:53:27 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:        logging=LoggingArgs(log_level='info',
+12/02/2024 11:53:27 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                            log_level_replica='info',
+12/02/2024 11:53:27 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                            iteration_step_info_interval=1),
+12/02/2024 11:53:27 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:        tokens=TokensArgs(sequence_length=4096,
+12/02/2024 11:53:27 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                          train_steps=100,
+12/02/2024 11:53:27 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                          micro_batch_size=2,
+12/02/2024 11:53:27 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                          batch_accumulation_per_replica=1,
+12/02/2024 11:53:27 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                          val_check_interval=100,
+12/02/2024 11:53:27 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                          limit_val_batches=0,
+12/02/2024 11:53:27 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                          limit_test_batches=0),
+12/02/2024 11:53:27 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:        optimizer=OptimizerArgs(optimizer_factory=AdamWOptimizerArgs(adam_eps=1e-08,
+12/02/2024 11:53:27 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                                     adam_beta1=0.9,
+12/02/2024 11:53:27 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                                     adam_beta2=0.95,
+12/02/2024 11:53:27 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                                     torch_adam_is_fused=True,
+12/02/2024 11:53:27 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                                     name='adamW'),
+12/02/2024 11:53:27 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                zero_stage=0,
+12/02/2024 11:53:27 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                weight_decay=0.01,
+12/02/2024 11:53:27 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                clip_grad=1.0,
+12/02/2024 11:53:27 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                accumulate_grad_in_fp32=True,
+12/02/2024 11:53:27 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                learning_rate_scheduler=LRSchedulerArgs(learning_rate=0.0003,
+12/02/2024 11:53:27 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                                        lr_warmup_steps=2,
+12/02/2024 11:53:27 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                                        lr_warmup_style='linear',
+12/02/2024 11:53:27 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                                        lr_decay_style='cosine',
+12/02/2024 11:53:27 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                                        lr_decay_steps=13,
+12/02/2024 11:53:27 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                                        lr_decay_starting_step=None,
+12/02/2024 11:53:27 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                                        min_decay_lr=1e-05)),
+12/02/2024 11:53:27 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:        data_stages=[DatasetStageArgs(name='Stable Training Stage',
+12/02/2024 11:53:27 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                      start_training_step=1,
+12/02/2024 11:53:27 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                      data=DataArgs(dataset=None,
+12/02/2024 11:53:27 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                    seed=42,
+12/02/2024 11:53:27 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                    num_loading_workers=1))],
+12/02/2024 11:53:27 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:        profiler=None,
+12/02/2024 11:53:27 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:        lighteval=None,
+12/02/2024 11:53:27 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:        s3_upload=None)
+12/02/2024 11:53:27 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: Model Config:
+12/02/2024 11:53:27 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: LlamaConfig(bos_token_id=0,
+12/02/2024 11:53:27 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             eos_token_id=0,
+12/02/2024 11:53:27 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             hidden_act='silu',
+12/02/2024 11:53:27 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             hidden_size=2048,
+12/02/2024 11:53:27 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             initializer_range=0.02,
+12/02/2024 11:53:27 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             intermediate_size=8192,
+12/02/2024 11:53:27 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             is_llama_config=True,
+12/02/2024 11:53:27 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             max_position_embeddings=4096,
+12/02/2024 11:53:27 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             num_attention_heads=16,
+12/02/2024 11:53:27 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             num_hidden_layers=15,
+12/02/2024 11:53:27 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             num_key_value_heads=16,
+12/02/2024 11:53:27 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             pad_token_id=None,
+12/02/2024 11:53:27 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             pretraining_tp=1,
+12/02/2024 11:53:27 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             rms_norm_eps=1e-05,
+12/02/2024 11:53:27 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             rope_scaling=None,
+12/02/2024 11:53:27 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             rope_theta=10000.0,
+12/02/2024 11:53:27 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             rope_interleaved=False,
+12/02/2024 11:53:27 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             tie_word_embeddings=True,
+12/02/2024 11:53:27 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             use_cache=True,
+12/02/2024 11:53:27 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             vocab_size=32768)
+12/02/2024 11:53:27 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: Building model..
+12/02/2024 11:53:27 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: Initialize RoPE Theta = 10000.0
+12/02/2024 11:53:27 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: Setting PP block ranks...
+12/02/2024 11:53:30 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: Total number of parameters: 1.07G (2048.12MiB)
+12/02/2024 11:53:30 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: Local number of parameters: 1.07G (2048.12MiB)
+12/02/2024 11:53:30 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [After model building] Memory usage: 2048.15MiB. Peak allocated: 5380.00MiB Peak reserved: 13590.00MiB
+12/02/2024 11:53:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: No checkpoint path provided.
+12/02/2024 11:53:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: Parametrizing model parameters using StandardParametrizator
+12/02/2024 11:53:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [Optimizer Building] Using LearningRateForSP as learning rate
+12/02/2024 11:53:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [Training Plan] Stage Stable Training Stage has 99 remaining training steps and has consumed 0 samples
+12/02/2024 11:53:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: Using dummy data generator
+12/02/2024 11:53:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [Training Plan] There are 1 training stages 
+12/02/2024 11:53:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [Stage Stable Training Stage] start from step 1 
+12/02/2024 11:53:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: 
+12/02/2024 11:53:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [Start training] datetime: 2024-12-02 11:53:41.483506 | mbs: 2 | grad_accum: 1 | global_batch_size: 1024 | sequence_length: 4096 | train_steps: 100 | start_iteration_step: 0 | consumed_train_samples: 0
+wandb: Tracking run with wandb version 0.16.0
+wandb: W&B syncing is set to `offline` in this directory.  
+wandb: Run `wandb online` or set WANDB_MODE=online to enable cloud syncing.
+12/02/2024 11:53:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: Resuming training from stage Stable Training Stage, it has trained for 0 samples and has 99 remaining train steps
+12/02/2024 11:53:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:  Memory usage: 12288.75MiB. Peak allocated 12288.75MiB. Peak reserved: 23836.00MiB
+12/02/2024 11:53:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:  Memory usage: 12367.89MiB. Peak allocated 26936.73MiB. Peak reserved: 27562.00MiB
+12/02/2024 11:53:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: iteration: 1 / 100 | consumed_tokens: 4.19M | elapsed_time_per_iteration_ms: 2.13K | tokens_per_sec: 1.97M | tokens_per_sec_per_gpu: 3.85K | global_batch_size: 1.02K | lm_loss: 10.8 | lr: 0.00015 | model_tflops_per_gpu: 30.6 | hardware_tflops_per_gpu: 30.6 | grad_norm: 0.539 | cuda_memory_allocated: 21.6G | cuda_max_memory_reserved: 28.9G | hd_total_memory_tb: 312G | hd_used_memory_tb: 69.4G | hd_free_memory_tb: 243G
+12/02/2024 11:53:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:  Memory usage: 20560.42MiB. Peak allocated 20560.42MiB. Peak reserved: 27562.00MiB
+12/02/2024 11:53:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:  Memory usage: 20560.42MiB. Peak allocated 35161.26MiB. Peak reserved: 36074.00MiB
+12/02/2024 11:53:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: iteration: 2 / 100 | consumed_tokens: 8.39M | elapsed_time_per_iteration_ms: 667 | tokens_per_sec: 6.29M | tokens_per_sec_per_gpu: 12.3K | global_batch_size: 1.02K | lm_loss: 10.8 | lr: 0.0003 | model_tflops_per_gpu: 97.7 | hardware_tflops_per_gpu: 97.7 | grad_norm: 0.539 | cuda_memory_allocated: 21.6G | cuda_max_memory_reserved: 37.8G | hd_total_memory_tb: 312G | hd_used_memory_tb: 69.4G | hd_free_memory_tb: 243G
+12/02/2024 11:53:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:  Memory usage: 20560.42MiB. Peak allocated 20560.47MiB. Peak reserved: 36074.00MiB
+12/02/2024 11:53:52 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:  Memory usage: 20560.42MiB. Peak allocated 35161.26MiB. Peak reserved: 37098.00MiB
+num_paramsnum_paramsnum_paramsnum_params
+
+num_params
+{'total': 1073805312, 'local': 1073805312}{'total': 1073805312, 'local': 1073805312}
+
+
+{'total': 1073805312, 'local': 1073805312}
+num_params
+{'total': 1073805312, 'local': 1073805312}{'total': 1073805312, 'local': 1073805312}
+
+
+num_paramsnum_paramsnum_params
+
+
+{'total': 1073805312, 'local': 1073805312}{'total': 1073805312, 'local': 1073805312}{'total': 1073805312, 'local': 1073805312}
+
+num_params
+
+num_params{'total': 1073805312, 'local': 1073805312}
+
+{'total': 1073805312, 'local': 1073805312}
+num_params
+{'total': 1073805312, 'local': 1073805312}
+num_params
+num_params{'total': 1073805312, 'local': 1073805312}
+num_paramsnum_params
+
+
+{'total': 1073805312, 'local': 1073805312}
+{'total': 1073805312, 'local': 1073805312}{'total': 1073805312, 'local': 1073805312}
+
+num_params{'total': 1073805312, 'local': 1073805312}
+
+{'total': 1073805312, 'local': 1073805312}
+num_paramsnum_params
+
+{'total': 1073805312, 'local': 1073805312}{'total': 1073805312, 'local': 1073805312}
+
+num_params
+num_params{'total': 1073805312, 'local': 1073805312}num_params
+
+
+{'total': 1073805312, 'local': 1073805312}
+num_paramsnum_params
+
+{'total': 1073805312, 'local': 1073805312}{'total': 1073805312, 'local': 1073805312}
+
+num_paramsnum_paramsnum_params
+
+num_paramsnum_params
+
+{'total': 1073805312, 'local': 1073805312}
+{'total': 1073805312, 'local': 1073805312}
+num_params
+{'total': 1073805312, 'local': 1073805312}
+num_params
+{'total': 1073805312, 'local': 1073805312}
+num_paramsnum_paramsnum_paramsnum_params
+
+
+
+{'total': 1073805312, 'local': 1073805312}{'total': 1073805312, 'local': 1073805312}
+
+num_params
+num_params{'total': 1073805312, 'local': 1073805312}
+num_params
+num_params{'total': 1073805312, 'local': 1073805312}
+num_params
+
+{'total': 1073805312, 'local': 1073805312}
+{'total': 1073805312, 'local': 1073805312}
+
+{'total': 1073805312, 'local': 1073805312}
+num_params
+{'total': 1073805312, 'local': 1073805312}
+num_paramsnum_params
+
+num_paramsnum_paramsnum_paramsnum_params
+
+
+{'total': 1073805312, 'local': 1073805312}
+{'total': 1073805312, 'local': 1073805312}
+{'total': 1073805312, 'local': 1073805312}
+
+num_params{'total': 1073805312, 'local': 1073805312}
+
+{'total': 1073805312, 'local': 1073805312}
+num_paramsnum_paramsnum_params
+
+
+{'total': 1073805312, 'local': 1073805312}{'total': 1073805312, 'local': 1073805312}{'total': 1073805312, 'local': 1073805312}
+
+
+num_paramsnum_params
+num_params
+{'total': 1073805312, 'local': 1073805312}
+
+{'total': 1073805312, 'local': 1073805312}
+{'total': 1073805312, 'local': 1073805312}
+num_paramsnum_paramsnum_paramsnum_params
+num_params
+
+
+num_paramsnum_paramsnum_paramsnum_paramsnum_params
+
+
+
+{'total': 1073805312, 'local': 1073805312}
+{'total': 1073805312, 'local': 1073805312}num_params{'total': 1073805312, 'local': 1073805312}num_paramsnum_params
+{'total': 1073805312, 'local': 1073805312}
+
+
+{'total': 1073805312, 'local': 1073805312}
+
+
+
+{'total': 1073805312, 'local': 1073805312}{'total': 1073805312, 'local': 1073805312}
+{'total': 1073805312, 'local': 1073805312}
+
+num_params
+{'total': 1073805312, 'local': 1073805312}
+num_params
+{'total': 1073805312, 'local': 1073805312}
+num_paramsnum_paramsnum_params
+
+
+num_params{'total': 1073805312, 'local': 1073805312}{'total': 1073805312, 'local': 1073805312}num_params{'total': 1073805312, 'local': 1073805312}
+
+
+num_params
+
+{'total': 1073805312, 'local': 1073805312}
+{'total': 1073805312, 'local': 1073805312}
+
+{'total': 1073805312, 'local': 1073805312}
+num_paramsnum_paramsnum_params
+
+
+{'total': 1073805312, 'local': 1073805312}{'total': 1073805312, 'local': 1073805312}
+{'total': 1073805312, 'local': 1073805312}
+
+num_params
+num_params{'total': 1073805312, 'local': 1073805312}
+
+num_params
+{'total': 1073805312, 'local': 1073805312}num_params
+num_params
+
+{'total': 1073805312, 'local': 1073805312}
+{'total': 1073805312, 'local': 1073805312}
+num_paramsnum_paramsnum_paramsnum_params
+
+
+
+{'total': 1073805312, 'local': 1073805312}{'total': 1073805312, 'local': 1073805312}
+{'total': 1073805312, 'local': 1073805312}{'total': 1073805312, 'local': 1073805312}
+num_params
+
+num_paramsnum_params
+
+num_params
+{'total': 1073805312, 'local': 1073805312}
+num_paramsnum_paramsnum_params
+
+
+{'total': 1073805312, 'local': 1073805312}
+{'total': 1073805312, 'local': 1073805312}{'total': 1073805312, 'local': 1073805312}
+
+num_paramsnum_params
+
+{'total': 1073805312, 'local': 1073805312}num_params{'total': 1073805312, 'local': 1073805312}
+num_paramsnum_paramsnum_params
+
+num_params{'total': 1073805312, 'local': 1073805312}{'total': 1073805312, 'local': 1073805312}
+
+
+
+{'total': 1073805312, 'local': 1073805312}{'total': 1073805312, 'local': 1073805312}
+
+num_paramsnum_paramsnum_paramsnum_params
+
+num_paramsnum_params
+num_params
+{'total': 1073805312, 'local': 1073805312}
+{'total': 1073805312, 'local': 1073805312}
+
+num_params{'total': 1073805312, 'local': 1073805312}
+
+num_params{'total': 1073805312, 'local': 1073805312}
+
+num_params
+{'total': 1073805312, 'local': 1073805312}
+num_params
+{'total': 1073805312, 'local': 1073805312}
+num_paramsnum_params
+
+num_paramsnum_paramsnum_paramsnum_paramsnum_params
+
+
+num_paramsnum_params{'total': 1073805312, 'local': 1073805312}
+
+
+
+{'total': 1073805312, 'local': 1073805312}{'total': 1073805312, 'local': 1073805312}{'total': 1073805312, 'local': 1073805312}{'total': 1073805312, 'local': 1073805312}{'total': 1073805312, 'local': 1073805312}
+
+num_params
+
+{'total': 1073805312, 'local': 1073805312}
+
+
+
+{'total': 1073805312, 'local': 1073805312}
+num_paramsnum_params
+
+{'total': 1073805312, 'local': 1073805312}
+num_params{'total': 1073805312, 'local': 1073805312}
+
+num_paramsnum_params{'total': 1073805312, 'local': 1073805312}num_params
+
+
+
+num_params
+{'total': 1073805312, 'local': 1073805312}
+num_paramsnum_params
+{'total': 1073805312, 'local': 1073805312}
+
+{'total': 1073805312, 'local': 1073805312}
+num_paramsnum_params
+
+num_paramsnum_paramsnum_paramsnum_paramsnum_params
+num_paramsnum_params
+
+{'total': 1073805312, 'local': 1073805312}
+
+
+
+{'total': 1073805312, 'local': 1073805312}
+{'total': 1073805312, 'local': 1073805312}{'total': 1073805312, 'local': 1073805312}{'total': 1073805312, 'local': 1073805312}
+{'total': 1073805312, 'local': 1073805312}
+
+
+{'total': 1073805312, 'local': 1073805312}
+
+num_params
+{'total': 1073805312, 'local': 1073805312}
+num_paramsnum_paramsnum_params
+
+num_params
+{'total': 1073805312, 'local': 1073805312}{'total': 1073805312, 'local': 1073805312}
+
+
+{'total': 1073805312, 'local': 1073805312}
+num_params{'total': 1073805312, 'local': 1073805312}
+
+num_paramsnum_params{'total': 1073805312, 'local': 1073805312}
+num_params
+num_paramsnum_params{'total': 1073805312, 'local': 1073805312}
+
+
+{'total': 1073805312, 'local': 1073805312}
+{'total': 1073805312, 'local': 1073805312}
+num_params
+num_params{'total': 1073805312, 'local': 1073805312}
+
+{'total': 1073805312, 'local': 1073805312}
+num_params
+{'total': 1073805312, 'local': 1073805312}
+num_paramsnum_params
+{'total': 1073805312, 'local': 1073805312}
+
+num_paramsnum_paramsnum_paramsnum_params
+
+
+
+{'total': 1073805312, 'local': 1073805312}{'total': 1073805312, 'local': 1073805312}
+{'total': 1073805312, 'local': 1073805312}{'total': 1073805312, 'local': 1073805312}
+
+
+num_paramsnum_params
+num_params
+num_params{'total': 1073805312, 'local': 1073805312}
+
+{'total': 1073805312, 'local': 1073805312}
+num_paramsnum_params
+
+{'total': 1073805312, 'local': 1073805312}
+{'total': 1073805312, 'local': 1073805312}
+num_paramsnum_params
+
+num_paramsnum_paramsnum_params
+
+
+{'total': 1073805312, 'local': 1073805312}{'total': 1073805312, 'local': 1073805312}
+{'total': 1073805312, 'local': 1073805312}
+
+num_params
+num_params
+{'total': 1073805312, 'local': 1073805312}
+num_params
+num_params{'total': 1073805312, 'local': 1073805312}
+num_params
+num_paramsnum_params
+num_params{'total': 1073805312, 'local': 1073805312}
+
+
+num_params{'total': 1073805312, 'local': 1073805312}num_params{'total': 1073805312, 'local': 1073805312}num_params
+
+
+
+
+{'total': 1073805312, 'local': 1073805312}{'total': 1073805312, 'local': 1073805312}
+{'total': 1073805312, 'local': 1073805312}
+
+num_paramsnum_params
+
+{'total': 1073805312, 'local': 1073805312}{'total': 1073805312, 'local': 1073805312}
+num_params
+
+num_params{'total': 1073805312, 'local': 1073805312}
+
+num_params{'total': 1073805312, 'local': 1073805312}
+num_params
+{'total': 1073805312, 'local': 1073805312}
+num_paramsnum_paramsnum_paramsnum_params
+num_params
+{'total': 1073805312, 'local': 1073805312}
+num_paramsnum_params
+num_params{'total': 1073805312, 'local': 1073805312}
+num_params
+
+num_params
+num_params{'total': 1073805312, 'local': 1073805312}
+
+num_params{'total': 1073805312, 'local': 1073805312}
+num_params
+{'total': 1073805312, 'local': 1073805312}
+
+{'total': 1073805312, 'local': 1073805312}
+num_params
+num_params{'total': 1073805312, 'local': 1073805312}num_params
+
+
+{'total': 1073805312, 'local': 1073805312}{'total': 1073805312, 'local': 1073805312}
+
+num_paramsnum_paramsnum_params
+num_params
+
+{'total': 1073805312, 'local': 1073805312}
+
+{'total': 1073805312, 'local': 1073805312}{'total': 1073805312, 'local': 1073805312}
+
+{'total': 1073805312, 'local': 1073805312}num_params
+
+{'total': 1073805312, 'local': 1073805312}
+num_paramsnum_params
+
+{'total': 1073805312, 'local': 1073805312}{'total': 1073805312, 'local': 1073805312}
+
+num_params
+{'total': 1073805312, 'local': 1073805312}
+num_params
+num_params
+{'total': 1073805312, 'local': 1073805312}
+num_paramsnum_params
+num_paramsnum_paramsnum_params
+
+{'total': 1073805312, 'local': 1073805312}
+{'total': 1073805312, 'local': 1073805312}
+num_params
+{'total': 1073805312, 'local': 1073805312}num_paramsnum_params
+
+
+{'total': 1073805312, 'local': 1073805312}
+
+{'total': 1073805312, 'local': 1073805312}
+{'total': 1073805312, 'local': 1073805312}
+num_params
+num_params{'total': 1073805312, 'local': 1073805312}
+
+num_params{'total': 1073805312, 'local': 1073805312}
+
+num_params{'total': 1073805312, 'local': 1073805312}
+num_paramsnum_params
+
+
+{'total': 1073805312, 'local': 1073805312}{'total': 1073805312, 'local': 1073805312}
+num_paramsnum_paramsnum_paramsnum_params
+
+
+num_paramsnum_paramsnum_params
+{'total': 1073805312, 'local': 1073805312}{'total': 1073805312, 'local': 1073805312}
+
+{'total': 1073805312, 'local': 1073805312}
+
+{'total': 1073805312, 'local': 1073805312}{'total': 1073805312, 'local': 1073805312}
+
+{'total': 1073805312, 'local': 1073805312}
+
+
+{'total': 1073805312, 'local': 1073805312}
+num_paramsnum_params
+num_params
+num_params{'total': 1073805312, 'local': 1073805312}num_paramsnum_params
+{'total': 1073805312, 'local': 1073805312}
+
+
+num_params
+{'total': 1073805312, 'local': 1073805312}{'total': 1073805312, 'local': 1073805312}
+
+
+
+{'total': 1073805312, 'local': 1073805312}
+{'total': 1073805312, 'local': 1073805312}{'total': 1073805312, 'local': 1073805312}
+
+num_params
+{'total': 1073805312, 'local': 1073805312}num_params
+num_params
+num_params{'total': 1073805312, 'local': 1073805312}
+num_params
+
+num_paramsnum_params{'total': 1073805312, 'local': 1073805312}
+{'total': 1073805312, 'local': 1073805312}
+
+
+
+{'total': 1073805312, 'local': 1073805312}
+num_params
+{'total': 1073805312, 'local': 1073805312}
+num_paramsnum_paramsnum_params
+
+num_params
+num_params{'total': 1073805312, 'local': 1073805312}{'total': 1073805312, 'local': 1073805312}
+{'total': 1073805312, 'local': 1073805312}
+
+
+num_params
+{'total': 1073805312, 'local': 1073805312}
+num_params
+num_params{'total': 1073805312, 'local': 1073805312}
+
+num_paramsnum_params
+num_params
+num_params{'total': 1073805312, 'local': 1073805312}
+
+{'total': 1073805312, 'local': 1073805312}
+
+num_params{'total': 1073805312, 'local': 1073805312}{'total': 1073805312, 'local': 1073805312}num_params
+
+num_params
+
+{'total': 1073805312, 'local': 1073805312}
+{'total': 1073805312, 'local': 1073805312}
+
+{'total': 1073805312, 'local': 1073805312}
+num_paramsnum_paramsnum_paramsnum_params
+
+
+
+num_params{'total': 1073805312, 'local': 1073805312}{'total': 1073805312, 'local': 1073805312}{'total': 1073805312, 'local': 1073805312}num_params{'total': 1073805312, 'local': 1073805312}
+
+
+
+
+
+{'total': 1073805312, 'local': 1073805312}
+{'total': 1073805312, 'local': 1073805312}
+num_paramsnum_params
+
+{'total': 1073805312, 'local': 1073805312}{'total': 1073805312, 'local': 1073805312}
+
+num_paramsnum_paramsnum_paramsnum_params
+
+
+num_params
+{'total': 1073805312, 'local': 1073805312}{'total': 1073805312, 'local': 1073805312}{'total': 1073805312, 'local': 1073805312}
+
+
+{'total': 1073805312, 'local': 1073805312}num_params
+{'total': 1073805312, 'local': 1073805312}
+num_params
+{'total': 1073805312, 'local': 1073805312}
+num_params
+{'total': 1073805312, 'local': 1073805312}
+num_paramsnum_paramsnum_params
+
+{'total': 1073805312, 'local': 1073805312}
+{'total': 1073805312, 'local': 1073805312}
+num_params
+{'total': 1073805312, 'local': 1073805312}num_params
+
+num_params
+{'total': 1073805312, 'local': 1073805312}
+{'total': 1073805312, 'local': 1073805312}
+
+num_params{'total': 1073805312, 'local': 1073805312}
+
+{'total': 1073805312, 'local': 1073805312}
+num_params
+{'total': 1073805312, 'local': 1073805312}
+num_params
+{'total': 1073805312, 'local': 1073805312}
+num_paramsnum_paramsnum_paramsnum_params
+
+num_params
+{'total': 1073805312, 'local': 1073805312}
+{'total': 1073805312, 'local': 1073805312}
+{'total': 1073805312, 'local': 1073805312}
+
+num_paramsnum_paramsnum_params
+
+
+{'total': 1073805312, 'local': 1073805312}num_params
+{'total': 1073805312, 'local': 1073805312}{'total': 1073805312, 'local': 1073805312}
+
+
+{'total': 1073805312, 'local': 1073805312}
+num_params
+{'total': 1073805312, 'local': 1073805312}
+num_paramsnum_params
+
+{'total': 1073805312, 'local': 1073805312}
+num_params{'total': 1073805312, 'local': 1073805312}
+
+{'total': 1073805312, 'local': 1073805312}{'total': 1073805312, 'local': 1073805312}
+num_params
+
+{'total': 1073805312, 'local': 1073805312}num_params
+
+
+{'total': 1073805312, 'local': 1073805312}
+{'total': 1073805312, 'local': 1073805312}
+num_params
+{'total': 1073805312, 'local': 1073805312}
+num_paramsnum_params
+
+{'total': 1073805312, 'local': 1073805312}{'total': 1073805312, 'local': 1073805312}
+
+{'total': 1073805312, 'local': 1073805312}{'total': 1073805312, 'local': 1073805312}
+
+num_paramsnum_paramsnum_params
+
+
+{'total': 1073805312, 'local': 1073805312}
+{'total': 1073805312, 'local': 1073805312}
+{'total': 1073805312, 'local': 1073805312}
+num_params
+{'total': 1073805312, 'local': 1073805312}num_params
+num_params
+{'total': 1073805312, 'local': 1073805312}
+
+{'total': 1073805312, 'local': 1073805312}
+num_paramsnum_paramsnum_params
+
+
+{'total': 1073805312, 'local': 1073805312}{'total': 1073805312, 'local': 1073805312}
+
+{'total': 1073805312, 'local': 1073805312}
+num_params
+num_params{'total': 1073805312, 'local': 1073805312}{'total': 1073805312, 'local': 1073805312}
+
+
+num_params{'total': 1073805312, 'local': 1073805312}
+num_params
+num_params
+num_params
+{'total': 1073805312, 'local': 1073805312}
+num_paramsnum_paramsnum_params
+
+{'total': 1073805312, 'local': 1073805312}
+
+{'total': 1073805312, 'local': 1073805312}
+{'total': 1073805312, 'local': 1073805312}
+num_params
+num_params{'total': 1073805312, 'local': 1073805312}
+num_params
+{'total': 1073805312, 'local': 1073805312}
+
+num_params{'total': 1073805312, 'local': 1073805312}
+num_params
+{'total': 1073805312, 'local': 1073805312}
+num_params
+{'total': 1073805312, 'local': 1073805312}
+num_params{'total': 1073805312, 'local': 1073805312}
+num_params
+{'total': 1073805312, 'local': 1073805312}
+
+{'total': 1073805312, 'local': 1073805312}
+num_paramsnum_paramsnum_params
+
+{'total': 1073805312, 'local': 1073805312}
+
+{'total': 1073805312, 'local': 1073805312}
+{'total': 1073805312, 'local': 1073805312}
+
+{'total': 1073805312, 'local': 1073805312}{'total': 1073805312, 'local': 1073805312}
+
+{'total': 1073805312, 'local': 1073805312}
+num_paramsnum_paramsnum_params
+
+
+{'total': 1073805312, 'local': 1073805312}{'total': 1073805312, 'local': 1073805312}
+
+{'total': 1073805312, 'local': 1073805312}num_params
+num_params
+
+
+{'total': 1073805312, 'local': 1073805312}
+{'total': 1073805312, 'local': 1073805312}{'total': 1073805312, 'local': 1073805312}
+
+
+
+{'total': 1073805312, 'local': 1073805312}
+{'total': 1073805312, 'local': 1073805312}
+num_paramsnum_params
+num_params
+num_paramsnum_params{'total': 1073805312, 'local': 1073805312}num_params
+
+
+
+{'total': 1073805312, 'local': 1073805312}
+{'total': 1073805312, 'local': 1073805312}{'total': 1073805312, 'local': 1073805312}
+{'total': 1073805312, 'local': 1073805312}
+
+
+{'total': 1073805312, 'local': 1073805312}
+{'total': 1073805312, 'local': 1073805312}
+num_params
+{'total': 1073805312, 'local': 1073805312}
+{'total': 1073805312, 'local': 1073805312}
+{'total': 1073805312, 'local': 1073805312}
+num_paramsnum_params
+
+{'total': 1073805312, 'local': 1073805312}{'total': 1073805312, 'local': 1073805312}num_params
+
+
+{'total': 1073805312, 'local': 1073805312}
+num_params
+{'total': 1073805312, 'local': 1073805312}
+{'total': 1073805312, 'local': 1073805312}{'total': 1073805312, 'local': 1073805312}{'total': 1073805312, 'local': 1073805312}
+
+
+num_params
+{'total': 1073805312, 'local': 1073805312}
+{'total': 1073805312, 'local': 1073805312}
+{'total': 1073805312, 'local': 1073805312}
+num_paramsnum_params
+
+{'total': 1073805312, 'local': 1073805312}
+{'total': 1073805312, 'local': 1073805312}
+
+
+num_params{'total': 1073805312, 'local': 1073805312}{'total': 1073805312, 'local': 1073805312}
+
+
+{'total': 1073805312, 'local': 1073805312}
+num_params
+{'total': 1073805312, 'local': 1073805312}num_params
+num_paramsnum_params
+
+num_params{'total': 1073805312, 'local': 1073805312}
+num_paramsnum_params
+{'total': 1073805312, 'local': 1073805312}
+
+
+{'total': 1073805312, 'local': 1073805312}{'total': 1073805312, 'local': 1073805312}
+
+{'total': 1073805312, 'local': 1073805312}
+num_params
+{'total': 1073805312, 'local': 1073805312}
+{'total': 1073805312, 'local': 1073805312}{'total': 1073805312, 'local': 1073805312}
+num_params
+num_params
+num_params
+{'total': 1073805312, 'local': 1073805312}
+
+{'total': 1073805312, 'local': 1073805312}
+num_paramsnum_params
+
+num_paramsnum_params{'total': 1073805312, 'local': 1073805312}{'total': 1073805312, 'local': 1073805312}
+
+
+
+num_params{'total': 1073805312, 'local': 1073805312}{'total': 1073805312, 'local': 1073805312}
+
+
+num_params{'total': 1073805312, 'local': 1073805312}
+
+num_params
+{'total': 1073805312, 'local': 1073805312}
+num_params
+num_params{'total': 1073805312, 'local': 1073805312}num_params
+
+
+num_paramsnum_params{'total': 1073805312, 'local': 1073805312}
+
+
+{'total': 1073805312, 'local': 1073805312}num_params{'total': 1073805312, 'local': 1073805312}
+
+
+{'total': 1073805312, 'local': 1073805312}num_params
+{'total': 1073805312, 'local': 1073805312}
+
+{'total': 1073805312, 'local': 1073805312}
+num_paramsnum_params
+num_params
+{'total': 1073805312, 'local': 1073805312}num_params
+
+num_params
+num_params
+{'total': 1073805312, 'local': 1073805312}
+
+{'total': 1073805312, 'local': 1073805312}num_params
+{'total': 1073805312, 'local': 1073805312}
+
+
+
+num_paramsnum_params
+{'total': 1073805312, 'local': 1073805312}{'total': 1073805312, 'local': 1073805312}{'total': 1073805312, 'local': 1073805312}
+
+
+
+
+{'total': 1073805312, 'local': 1073805312}
+{'total': 1073805312, 'local': 1073805312}{'total': 1073805312, 'local': 1073805312}
+
+num_params{'total': 1073805312, 'local': 1073805312}
+{'total': 1073805312, 'local': 1073805312}
+
+num_params
+{'total': 1073805312, 'local': 1073805312}{'total': 1073805312, 'local': 1073805312}
+
+num_params
+
+{'total': 1073805312, 'local': 1073805312}
+{'total': 1073805312, 'local': 1073805312}
+num_params
+{'total': 1073805312, 'local': 1073805312}
+num_params
+{'total': 1073805312, 'local': 1073805312}
+num_params
+num_params{'total': 1073805312, 'local': 1073805312}
+num_params
+num_params{'total': 1073805312, 'local': 1073805312}
+
+
+{'total': 1073805312, 'local': 1073805312}
+{'total': 1073805312, 'local': 1073805312}
+num_paramsnum_params
+
+{'total': 1073805312, 'local': 1073805312}{'total': 1073805312, 'local': 1073805312}
+
+num_paramsnum_paramsnum_params
+num_params
+
+{'total': 1073805312, 'local': 1073805312}
+{'total': 1073805312, 'local': 1073805312}
+
+{'total': 1073805312, 'local': 1073805312}
+{'total': 1073805312, 'local': 1073805312}
+num_params
+{'total': 1073805312, 'local': 1073805312}
+num_params
+num_params{'total': 1073805312, 'local': 1073805312}
+num_params
+{'total': 1073805312, 'local': 1073805312}
+
+num_params{'total': 1073805312, 'local': 1073805312}num_params{'total': 1073805312, 'local': 1073805312}
+
+
+
+{'total': 1073805312, 'local': 1073805312}
+{'total': 1073805312, 'local': 1073805312}
+num_params
+{'total': 1073805312, 'local': 1073805312}
+
+{'total': 1073805312, 'local': 1073805312}
+num_params
+{'total': 1073805312, 'local': 1073805312}
+num_params
+{'total': 1073805312, 'local': 1073805312}
+{'total': 1073805312, 'local': 1073805312}{'total': 1073805312, 'local': 1073805312}
+
+
+{'total': 1073805312, 'local': 1073805312}
+{'total': 1073805312, 'local': 1073805312}
+{'total': 1073805312, 'local': 1073805312}
+num_params
+num_paramsnum_paramsnum_params{'total': 1073805312, 'local': 1073805312}
+
+num_params
+{'total': 1073805312, 'local': 1073805312}
+num_params
+{'total': 1073805312, 'local': 1073805312}
+
+
+{'total': 1073805312, 'local': 1073805312}
+num_params
+num_params{'total': 1073805312, 'local': 1073805312}
+
+{'total': 1073805312, 'local': 1073805312}
+{'total': 1073805312, 'local': 1073805312}
+num_paramsnum_params
+{'total': 1073805312, 'local': 1073805312}
+
+{'total': 1073805312, 'local': 1073805312}
+num_params
+{'total': 1073805312, 'local': 1073805312}
+num_paramsnum_paramsnum_params
+
+{'total': 1073805312, 'local': 1073805312}
+
+{'total': 1073805312, 'local': 1073805312}
+{'total': 1073805312, 'local': 1073805312}
+num_params
+{'total': 1073805312, 'local': 1073805312}
+num_params
+{'total': 1073805312, 'local': 1073805312}num_params
+
+{'total': 1073805312, 'local': 1073805312}
+num_paramsnum_paramsnum_params{'total': 1073805312, 'local': 1073805312}
+
+
+
+{'total': 1073805312, 'local': 1073805312}{'total': 1073805312, 'local': 1073805312}
+
+{'total': 1073805312, 'local': 1073805312}
+{'total': 1073805312, 'local': 1073805312}num_params
+
+{'total': 1073805312, 'local': 1073805312}
+
+{'total': 1073805312, 'local': 1073805312}
+{'total': 1073805312, 'local': 1073805312}
+num_params
+num_params
+{'total': 1073805312, 'local': 1073805312}
+num_params
+{'total': 1073805312, 'local': 1073805312}
+
+num_params{'total': 1073805312, 'local': 1073805312}
+
+{'total': 1073805312, 'local': 1073805312}
+num_params
+{'total': 1073805312, 'local': 1073805312}
+num_params
+{'total': 1073805312, 'local': 1073805312}
+num_params
+{'total': 1073805312, 'local': 1073805312}
+{'total': 1073805312, 'local': 1073805312}
+
+num_params{'total': 1073805312, 'local': 1073805312}
+
+{'total': 1073805312, 'local': 1073805312}
+num_params
+{'total': 1073805312, 'local': 1073805312}
+num_params
+{'total': 1073805312, 'local': 1073805312}
+num_params
+{'total': 1073805312, 'local': 1073805312}
+num_paramsnum_params
+
+{'total': 1073805312, 'local': 1073805312}
+{'total': 1073805312, 'local': 1073805312}
+num_params
+{'total': 1073805312, 'local': 1073805312}
+num_params
+{'total': 1073805312, 'local': 1073805312}
+num_params
+{'total': 1073805312, 'local': 1073805312}
+num_paramsnum_params
+
+num_params{'total': 1073805312, 'local': 1073805312}{'total': 1073805312, 'local': 1073805312}
+
+
+{'total': 1073805312, 'local': 1073805312}
+num_paramsnum_params
+{'total': 1073805312, 'local': 1073805312}
+
+{'total': 1073805312, 'local': 1073805312}
+num_params
+{'total': 1073805312, 'local': 1073805312}
+num_params
+{'total': 1073805312, 'local': 1073805312}num_params
+
+{'total': 1073805312, 'local': 1073805312}
+{'total': 1073805312, 'local': 1073805312}
+{'total': 1073805312, 'local': 1073805312}
+
+{'total': 1073805312, 'local': 1073805312}
+
+{'total': 1073805312, 'local': 1073805312}
+{'total': 1073805312, 'local': 1073805312}
+{'total': 1073805312, 'local': 1073805312}num_params
+num_params
+
+{'total': 1073805312, 'local': 1073805312}
+{'total': 1073805312, 'local': 1073805312}
+num_params{'total': 1073805312, 'local': 1073805312}{'total': 1073805312, 'local': 1073805312}
+
+
+num_params{'total': 1073805312, 'local': 1073805312}
+
+num_params{'total': 1073805312, 'local': 1073805312}
+
+{'total': 1073805312, 'local': 1073805312}
+num_params{'total': 1073805312, 'local': 1073805312}
+
+
+{'total': 1073805312, 'local': 1073805312}{'total': 1073805312, 'local': 1073805312}
+
+{'total': 1073805312, 'local': 1073805312}
+{'total': 1073805312, 'local': 1073805312}
+num_params
+num_params
+{'total': 1073805312, 'local': 1073805312}
+num_params
+{'total': 1073805312, 'local': 1073805312}
+num_params
+{'total': 1073805312, 'local': 1073805312}
+num_paramsnum_params
+
+{'total': 1073805312, 'local': 1073805312}
+{'total': 1073805312, 'local': 1073805312}
+num_params
+{'total': 1073805312, 'local': 1073805312}
+
+{'total': 1073805312, 'local': 1073805312}
+num_params
+{'total': 1073805312, 'local': 1073805312}
+num_params
+{'total': 1073805312, 'local': 1073805312}
+num_params
+{'total': 1073805312, 'local': 1073805312}
+num_params
+{'total': 1073805312, 'local': 1073805312}
+num_params
+{'total': 1073805312, 'local': 1073805312}
+
+
+{'total': 1073805312, 'local': 1073805312}
+{'total': 1073805312, 'local': 1073805312}{'total': 1073805312, 'local': 1073805312}
+
+num_params
+{'total': 1073805312, 'local': 1073805312}
+12/02/2024 11:53:52 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: iteration: 3 / 100 | consumed_tokens: 12.6M | elapsed_time_per_iteration_ms: 267 | tokens_per_sec: 15.7M | tokens_per_sec_per_gpu: 30.7K | global_batch_size: 1.02K | lm_loss: 10.8 | lr: 0.000296 | model_tflops_per_gpu: 244 | hardware_tflops_per_gpu: 244 | grad_norm: 0.53 | cuda_memory_allocated: 21.6G | cuda_max_memory_reserved: 38.9G | hd_total_memory_tb: 312G | hd_used_memory_tb: 69.4G | hd_free_memory_tb: 243G
+num_paramsnum_params
+
+{'total': 1073805312, 'local': 1073805312}
+{'total': 1073805312, 'local': 1073805312}
+{'total': 1073805312, 'local': 1073805312}
+num_params
+{'total': 1073805312, 'local': 1073805312}
+{'total': 1073805312, 'local': 1073805312}
+num_paramsnum_params
+
+{'total': 1073805312, 'local': 1073805312}
+num_params{'total': 1073805312, 'local': 1073805312}
+
+{'total': 1073805312, 'local': 1073805312}
+num_params
+{'total': 1073805312, 'local': 1073805312}
+num_params
+{'total': 1073805312, 'local': 1073805312}
+{'total': 1073805312, 'local': 1073805312}
+num_params
+{'total': 1073805312, 'local': 1073805312}
+num_params
+num_params
+{'total': 1073805312, 'local': 1073805312}
+num_params
+{'total': 1073805312, 'local': 1073805312}
+num_params
+{'total': 1073805312, 'local': 1073805312}
+num_params
+{'total': 1073805312, 'local': 1073805312}
+num_params
+{'total': 1073805312, 'local': 1073805312}
+num_params
+
+{'total': 1073805312, 'local': 1073805312}
+num_params
+{'total': 1073805312, 'local': 1073805312}
+num_params
+{'total': 1073805312, 'local': 1073805312}
+{'total': 1073805312, 'local': 1073805312}
+num_params
+{'total': 1073805312, 'local': 1073805312}
+{'total': 1073805312, 'local': 1073805312}
+num_params
+{'total': 1073805312, 'local': 1073805312}
+num_params
+{'total': 1073805312, 'local': 1073805312}
+num_params
+{'total': 1073805312, 'local': 1073805312}
+num_params
+num_params
+{'total': 1073805312, 'local': 1073805312}
+num_params
+{'total': 1073805312, 'local': 1073805312}
+num_params
+{'total': 1073805312, 'local': 1073805312}
+
+{'total': 1073805312, 'local': 1073805312}
+num_params
+{'total': 1073805312, 'local': 1073805312}
+num_params
+{'total': 1073805312, 'local': 1073805312}
+{'total': 1073805312, 'local': 1073805312}
+num_params
+num_params
+{'total': 1073805312, 'local': 1073805312}
+{'total': 1073805312, 'local': 1073805312}
+num_params
+num_params
+{'total': 1073805312, 'local': 1073805312}
+{'total': 1073805312, 'local': 1073805312}
+{'total': 1073805312, 'local': 1073805312}
+{'total': 1073805312, 'local': 1073805312}
+num_params
+{'total': 1073805312, 'local': 1073805312}
+12/02/2024 11:53:52 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: | job_id   | name                                                          | nodes | seq_len | mbs | batch_accum | gbs  | mTFLOPs | hTFLOPs | tok/s/gpu | AllReduce (GB/s) | AllGather (GB/s) | ReduceScatter (GB/s) | AR Intra-node (GB/s) | AG Intra-node (GB/s) | RS Intra-node (GB/s) | Mem Alloc (GB) | Mem Res (GB) | dp  | pp | tp | pp_engine                                                                                                 | tp_mode                                 | tp_async_comm | hidden_size | hidden_act | num_layers | num_heads | num_kv_heads | max_pos | vocab_size | tie_word_embeddings | dtype          | zero_stage | ddp_bucket_cap_mb | accumulate_grad_in_fp32 | Total Params | Local Params |
+12/02/2024 11:53:52 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: | -------- | ------------------------------------------------------------- | ----- | ------- | --- | ----------- | ---- | ------- | ------- | --------- | ---------------- | ---------------- | -------------------- | -------------------- | -------------------- | -------------------- | -------------- | ------------ | --- | -- | -- | --------------------------------------------------------------------------------------------------------- | --------------------------------------- | ------------- | ----------- | ---------- | ---------- | --------- | ------------ | ------- | ---------- | ------------------- | -------------- | ---------- | ----------------- | ----------------------- | ------------ | ------------ |
+{'total': 1073805312, 'local': 1073805312}
+{'total': 1073805312, 'local': 1073805312}
+{'total': 1073805312, 'local': 1073805312}{'total': 1073805312, 'local': 1073805312}
+
+{'total': 1073805312, 'local': 1073805312}
+
+12/02/2024 11:53:52 [INFO|DP=387|PP=0|TP=0|ip-26-0-172-57]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=388|PP=0|TP=0|ip-26-0-172-57]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=263|PP=0|TP=0|ip-26-0-167-217]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=320|PP=0|TP=0|ip-26-0-168-95]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=322|PP=0|TP=0|ip-26-0-168-95]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=64|PP=0|TP=0|ip-26-0-163-236]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=70|PP=0|TP=0|ip-26-0-163-236]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=225|PP=0|TP=0|ip-26-0-166-244]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=16|PP=0|TP=0|ip-26-0-161-138]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=20|PP=0|TP=0|ip-26-0-161-138]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=358|PP=0|TP=0|ip-26-0-171-21]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=384|PP=0|TP=0|ip-26-0-172-57]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=385|PP=0|TP=0|ip-26-0-172-57]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=390|PP=0|TP=0|ip-26-0-172-57]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=32|PP=0|TP=0|ip-26-0-162-180]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=289|PP=0|TP=0|ip-26-0-168-238]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=294|PP=0|TP=0|ip-26-0-168-238]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=120|PP=0|TP=0|ip-26-0-164-236]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=124|PP=0|TP=0|ip-26-0-164-236]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=497|PP=0|TP=0|ip-26-0-175-241]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=502|PP=0|TP=0|ip-26-0-175-241]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=501|PP=0|TP=0|ip-26-0-175-241]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=498|PP=0|TP=0|ip-26-0-175-241]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=452|PP=0|TP=0|ip-26-0-174-240]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=454|PP=0|TP=0|ip-26-0-174-240]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=432|PP=0|TP=0|ip-26-0-174-186]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=137|PP=0|TP=0|ip-26-0-164-75]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=196|PP=0|TP=0|ip-26-0-165-59]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=197|PP=0|TP=0|ip-26-0-165-59]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=199|PP=0|TP=0|ip-26-0-165-59]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=340|PP=0|TP=0|ip-26-0-170-143]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=327|PP=0|TP=0|ip-26-0-168-95]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=257|PP=0|TP=0|ip-26-0-167-217]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=259|PP=0|TP=0|ip-26-0-167-217]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=256|PP=0|TP=0|ip-26-0-167-217]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=261|PP=0|TP=0|ip-26-0-167-217]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=260|PP=0|TP=0|ip-26-0-167-217]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=262|PP=0|TP=0|ip-26-0-167-217]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=470|PP=0|TP=0|ip-26-0-175-132]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=72|PP=0|TP=0|ip-26-0-163-43]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=73|PP=0|TP=0|ip-26-0-163-43]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=38|PP=0|TP=0|ip-26-0-162-180]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=165|PP=0|TP=0|ip-26-0-165-202]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=162|PP=0|TP=0|ip-26-0-165-202]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=167|PP=0|TP=0|ip-26-0-165-202]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=108|PP=0|TP=0|ip-26-0-164-187]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=66|PP=0|TP=0|ip-26-0-163-236]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=67|PP=0|TP=0|ip-26-0-163-236]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=65|PP=0|TP=0|ip-26-0-163-236]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=69|PP=0|TP=0|ip-26-0-163-236]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=68|PP=0|TP=0|ip-26-0-163-236]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=50|PP=0|TP=0|ip-26-0-162-79]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=307|PP=0|TP=0|ip-26-0-168-34]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=102|PP=0|TP=0|ip-26-0-164-18]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=133|PP=0|TP=0|ip-26-0-164-45]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=134|PP=0|TP=0|ip-26-0-164-45]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=132|PP=0|TP=0|ip-26-0-164-45]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=135|PP=0|TP=0|ip-26-0-164-45]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=210|PP=0|TP=0|ip-26-0-166-15]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=228|PP=0|TP=0|ip-26-0-166-244]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=229|PP=0|TP=0|ip-26-0-166-244]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=226|PP=0|TP=0|ip-26-0-166-244]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=224|PP=0|TP=0|ip-26-0-166-244]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=227|PP=0|TP=0|ip-26-0-166-244]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=230|PP=0|TP=0|ip-26-0-166-244]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=8|PP=0|TP=0|ip-26-0-160-242]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=13|PP=0|TP=0|ip-26-0-160-242]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=88|PP=0|TP=0|ip-26-0-164-0]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=148|PP=0|TP=0|ip-26-0-165-131]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=147|PP=0|TP=0|ip-26-0-165-131]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=202|PP=0|TP=0|ip-26-0-166-125]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=204|PP=0|TP=0|ip-26-0-166-125]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=203|PP=0|TP=0|ip-26-0-166-125]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=453|PP=0|TP=0|ip-26-0-174-240]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=450|PP=0|TP=0|ip-26-0-174-240]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=451|PP=0|TP=0|ip-26-0-174-240]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=455|PP=0|TP=0|ip-26-0-174-240]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=449|PP=0|TP=0|ip-26-0-174-240]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=437|PP=0|TP=0|ip-26-0-174-186]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=281|PP=0|TP=0|ip-26-0-168-120]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=286|PP=0|TP=0|ip-26-0-168-120]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=285|PP=0|TP=0|ip-26-0-168-120]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=21|PP=0|TP=0|ip-26-0-161-138]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=403|PP=0|TP=0|ip-26-0-173-202]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=195|PP=0|TP=0|ip-26-0-165-59]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=198|PP=0|TP=0|ip-26-0-165-59]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=486|PP=0|TP=0|ip-26-0-175-170]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=481|PP=0|TP=0|ip-26-0-175-170]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=485|PP=0|TP=0|ip-26-0-175-170]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=325|PP=0|TP=0|ip-26-0-168-95]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=181|PP=0|TP=0|ip-26-0-165-24]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=180|PP=0|TP=0|ip-26-0-165-24]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=182|PP=0|TP=0|ip-26-0-165-24]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=84|PP=0|TP=0|ip-26-0-163-58]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=87|PP=0|TP=0|ip-26-0-163-58]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=83|PP=0|TP=0|ip-26-0-163-58]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=373|PP=0|TP=0|ip-26-0-172-142]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=370|PP=0|TP=0|ip-26-0-172-142]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=372|PP=0|TP=0|ip-26-0-172-142]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=465|PP=0|TP=0|ip-26-0-175-132]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=468|PP=0|TP=0|ip-26-0-175-132]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=168|PP=0|TP=0|ip-26-0-165-213]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=175|PP=0|TP=0|ip-26-0-165-213]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=356|PP=0|TP=0|ip-26-0-171-21]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=296|PP=0|TP=0|ip-26-0-168-30]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=300|PP=0|TP=0|ip-26-0-168-30]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=389|PP=0|TP=0|ip-26-0-172-57]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=391|PP=0|TP=0|ip-26-0-172-57]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=76|PP=0|TP=0|ip-26-0-163-43]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=504|PP=0|TP=0|ip-26-0-175-34]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=36|PP=0|TP=0|ip-26-0-162-180]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=344|PP=0|TP=0|ip-26-0-170-160]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=360|PP=0|TP=0|ip-26-0-172-116]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=26|PP=0|TP=0|ip-26-0-161-78]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=29|PP=0|TP=0|ip-26-0-161-78]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=290|PP=0|TP=0|ip-26-0-168-238]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=293|PP=0|TP=0|ip-26-0-168-238]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=408|PP=0|TP=0|ip-26-0-173-246]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=49|PP=0|TP=0|ip-26-0-162-79]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=52|PP=0|TP=0|ip-26-0-162-79]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=114|PP=0|TP=0|ip-26-0-164-207]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=112|PP=0|TP=0|ip-26-0-164-207]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=117|PP=0|TP=0|ip-26-0-164-207]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=305|PP=0|TP=0|ip-26-0-168-34]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=96|PP=0|TP=0|ip-26-0-164-18]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=100|PP=0|TP=0|ip-26-0-164-18]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=99|PP=0|TP=0|ip-26-0-164-18]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=98|PP=0|TP=0|ip-26-0-164-18]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=103|PP=0|TP=0|ip-26-0-164-18]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=101|PP=0|TP=0|ip-26-0-164-18]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=44|PP=0|TP=0|ip-26-0-162-233]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=130|PP=0|TP=0|ip-26-0-164-45]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=274|PP=0|TP=0|ip-26-0-167-9]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=276|PP=0|TP=0|ip-26-0-167-9]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=272|PP=0|TP=0|ip-26-0-167-9]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=426|PP=0|TP=0|ip-26-0-174-100]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=427|PP=0|TP=0|ip-26-0-174-100]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=123|PP=0|TP=0|ip-26-0-164-236]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=126|PP=0|TP=0|ip-26-0-164-236]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=127|PP=0|TP=0|ip-26-0-164-236]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=121|PP=0|TP=0|ip-26-0-164-236]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=122|PP=0|TP=0|ip-26-0-164-236]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=125|PP=0|TP=0|ip-26-0-164-236]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=231|PP=0|TP=0|ip-26-0-166-244]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=503|PP=0|TP=0|ip-26-0-175-241]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=496|PP=0|TP=0|ip-26-0-175-241]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=500|PP=0|TP=0|ip-26-0-175-241]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=11|PP=0|TP=0|ip-26-0-160-242]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=333|PP=0|TP=0|ip-26-0-170-132]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=331|PP=0|TP=0|ip-26-0-170-132]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=329|PP=0|TP=0|ip-26-0-170-132]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=91|PP=0|TP=0|ip-26-0-164-0]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=144|PP=0|TP=0|ip-26-0-165-131]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=206|PP=0|TP=0|ip-26-0-166-125]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=316|PP=0|TP=0|ip-26-0-168-52]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=319|PP=0|TP=0|ip-26-0-168-52]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=313|PP=0|TP=0|ip-26-0-168-52]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=448|PP=0|TP=0|ip-26-0-174-240]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=434|PP=0|TP=0|ip-26-0-174-186]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=433|PP=0|TP=0|ip-26-0-174-186]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=436|PP=0|TP=0|ip-26-0-174-186]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=57|PP=0|TP=0|ip-26-0-163-226]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=61|PP=0|TP=0|ip-26-0-163-226]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=63|PP=0|TP=0|ip-26-0-163-226]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=280|PP=0|TP=0|ip-26-0-168-120]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=282|PP=0|TP=0|ip-26-0-168-120]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=18|PP=0|TP=0|ip-26-0-161-138]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=23|PP=0|TP=0|ip-26-0-161-138]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=404|PP=0|TP=0|ip-26-0-173-202]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=405|PP=0|TP=0|ip-26-0-173-202]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=140|PP=0|TP=0|ip-26-0-164-75]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=142|PP=0|TP=0|ip-26-0-164-75]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=141|PP=0|TP=0|ip-26-0-164-75]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=143|PP=0|TP=0|ip-26-0-164-75]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=138|PP=0|TP=0|ip-26-0-164-75]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=397|PP=0|TP=0|ip-26-0-172-73]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=192|PP=0|TP=0|ip-26-0-165-59]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=193|PP=0|TP=0|ip-26-0-165-59]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=336|PP=0|TP=0|ip-26-0-170-143]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=339|PP=0|TP=0|ip-26-0-170-143]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=338|PP=0|TP=0|ip-26-0-170-143]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=343|PP=0|TP=0|ip-26-0-170-143]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=484|PP=0|TP=0|ip-26-0-175-170]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=482|PP=0|TP=0|ip-26-0-175-170]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=480|PP=0|TP=0|ip-26-0-175-170]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=326|PP=0|TP=0|ip-26-0-168-95]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=323|PP=0|TP=0|ip-26-0-168-95]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=178|PP=0|TP=0|ip-26-0-165-24]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=183|PP=0|TP=0|ip-26-0-165-24]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=81|PP=0|TP=0|ip-26-0-163-58]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=85|PP=0|TP=0|ip-26-0-163-58]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=368|PP=0|TP=0|ip-26-0-172-142]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=369|PP=0|TP=0|ip-26-0-172-142]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=374|PP=0|TP=0|ip-26-0-172-142]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=258|PP=0|TP=0|ip-26-0-167-217]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=488|PP=0|TP=0|ip-26-0-175-19]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=489|PP=0|TP=0|ip-26-0-175-19]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=491|PP=0|TP=0|ip-26-0-175-19]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=467|PP=0|TP=0|ip-26-0-175-132]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=471|PP=0|TP=0|ip-26-0-175-132]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=171|PP=0|TP=0|ip-26-0-165-213]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=357|PP=0|TP=0|ip-26-0-171-21]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=421|PP=0|TP=0|ip-26-0-173-7]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=422|PP=0|TP=0|ip-26-0-173-7]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=419|PP=0|TP=0|ip-26-0-173-7]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=423|PP=0|TP=0|ip-26-0-173-7]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=420|PP=0|TP=0|ip-26-0-173-7]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=418|PP=0|TP=0|ip-26-0-173-7]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=416|PP=0|TP=0|ip-26-0-173-7]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=299|PP=0|TP=0|ip-26-0-168-30]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=301|PP=0|TP=0|ip-26-0-168-30]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=302|PP=0|TP=0|ip-26-0-168-30]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=303|PP=0|TP=0|ip-26-0-168-30]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=386|PP=0|TP=0|ip-26-0-172-57]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=79|PP=0|TP=0|ip-26-0-163-43]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=458|PP=0|TP=0|ip-26-0-174-36]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=457|PP=0|TP=0|ip-26-0-174-36]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=509|PP=0|TP=0|ip-26-0-175-34]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=508|PP=0|TP=0|ip-26-0-175-34]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=507|PP=0|TP=0|ip-26-0-175-34]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=477|PP=0|TP=0|ip-26-0-175-165]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=473|PP=0|TP=0|ip-26-0-175-165]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=187|PP=0|TP=0|ip-26-0-165-38]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=188|PP=0|TP=0|ip-26-0-165-38]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=189|PP=0|TP=0|ip-26-0-165-38]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=190|PP=0|TP=0|ip-26-0-165-38]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=186|PP=0|TP=0|ip-26-0-165-38]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=34|PP=0|TP=0|ip-26-0-162-180]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=37|PP=0|TP=0|ip-26-0-162-180]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=39|PP=0|TP=0|ip-26-0-162-180]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=35|PP=0|TP=0|ip-26-0-162-180]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=33|PP=0|TP=0|ip-26-0-162-180]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=248|PP=0|TP=0|ip-26-0-167-177]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=251|PP=0|TP=0|ip-26-0-167-177]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=345|PP=0|TP=0|ip-26-0-170-160]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=351|PP=0|TP=0|ip-26-0-170-160]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=350|PP=0|TP=0|ip-26-0-170-160]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=348|PP=0|TP=0|ip-26-0-170-160]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=347|PP=0|TP=0|ip-26-0-170-160]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=349|PP=0|TP=0|ip-26-0-170-160]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=364|PP=0|TP=0|ip-26-0-172-116]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=160|PP=0|TP=0|ip-26-0-165-202]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=163|PP=0|TP=0|ip-26-0-165-202]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=161|PP=0|TP=0|ip-26-0-165-202]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=269|PP=0|TP=0|ip-26-0-167-51]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=270|PP=0|TP=0|ip-26-0-167-51]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=111|PP=0|TP=0|ip-26-0-164-187]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=104|PP=0|TP=0|ip-26-0-164-187]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=106|PP=0|TP=0|ip-26-0-164-187]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=376|PP=0|TP=0|ip-26-0-172-147]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=378|PP=0|TP=0|ip-26-0-172-147]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=383|PP=0|TP=0|ip-26-0-172-147]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=71|PP=0|TP=0|ip-26-0-163-236]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=295|PP=0|TP=0|ip-26-0-168-238]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=288|PP=0|TP=0|ip-26-0-168-238]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=292|PP=0|TP=0|ip-26-0-168-238]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=291|PP=0|TP=0|ip-26-0-168-238]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=222|PP=0|TP=0|ip-26-0-166-214]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=220|PP=0|TP=0|ip-26-0-166-214]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=221|PP=0|TP=0|ip-26-0-166-214]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=216|PP=0|TP=0|ip-26-0-166-214]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=54|PP=0|TP=0|ip-26-0-162-79]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=53|PP=0|TP=0|ip-26-0-162-79]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=116|PP=0|TP=0|ip-26-0-164-207]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=118|PP=0|TP=0|ip-26-0-164-207]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=159|PP=0|TP=0|ip-26-0-165-164]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=157|PP=0|TP=0|ip-26-0-165-164]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=156|PP=0|TP=0|ip-26-0-165-164]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=446|PP=0|TP=0|ip-26-0-174-196]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=46|PP=0|TP=0|ip-26-0-162-233]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=233|PP=0|TP=0|ip-26-0-166-36]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=238|PP=0|TP=0|ip-26-0-166-36]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=234|PP=0|TP=0|ip-26-0-166-36]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=236|PP=0|TP=0|ip-26-0-166-36]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=239|PP=0|TP=0|ip-26-0-166-36]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=240|PP=0|TP=0|ip-26-0-167-175]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=245|PP=0|TP=0|ip-26-0-167-175]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=128|PP=0|TP=0|ip-26-0-164-45]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=129|PP=0|TP=0|ip-26-0-164-45]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=212|PP=0|TP=0|ip-26-0-166-15]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=211|PP=0|TP=0|ip-26-0-166-15]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=213|PP=0|TP=0|ip-26-0-166-15]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=209|PP=0|TP=0|ip-26-0-166-15]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=215|PP=0|TP=0|ip-26-0-166-15]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=425|PP=0|TP=0|ip-26-0-174-100]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=499|PP=0|TP=0|ip-26-0-175-241]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=12|PP=0|TP=0|ip-26-0-160-242]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=14|PP=0|TP=0|ip-26-0-160-242]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=15|PP=0|TP=0|ip-26-0-160-242]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=330|PP=0|TP=0|ip-26-0-170-132]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=332|PP=0|TP=0|ip-26-0-170-132]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=328|PP=0|TP=0|ip-26-0-170-132]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=95|PP=0|TP=0|ip-26-0-164-0]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=151|PP=0|TP=0|ip-26-0-165-131]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=149|PP=0|TP=0|ip-26-0-165-131]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=145|PP=0|TP=0|ip-26-0-165-131]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=150|PP=0|TP=0|ip-26-0-165-131]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=200|PP=0|TP=0|ip-26-0-166-125]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=318|PP=0|TP=0|ip-26-0-168-52]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=439|PP=0|TP=0|ip-26-0-174-186]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=58|PP=0|TP=0|ip-26-0-163-226]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=60|PP=0|TP=0|ip-26-0-163-226]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=59|PP=0|TP=0|ip-26-0-163-226]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=284|PP=0|TP=0|ip-26-0-168-120]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=287|PP=0|TP=0|ip-26-0-168-120]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=19|PP=0|TP=0|ip-26-0-161-138]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=401|PP=0|TP=0|ip-26-0-173-202]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=400|PP=0|TP=0|ip-26-0-173-202]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=406|PP=0|TP=0|ip-26-0-173-202]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=407|PP=0|TP=0|ip-26-0-173-202]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=136|PP=0|TP=0|ip-26-0-164-75]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=396|PP=0|TP=0|ip-26-0-172-73]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=392|PP=0|TP=0|ip-26-0-172-73]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=393|PP=0|TP=0|ip-26-0-172-73]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=398|PP=0|TP=0|ip-26-0-172-73]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=194|PP=0|TP=0|ip-26-0-165-59]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=337|PP=0|TP=0|ip-26-0-170-143]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=483|PP=0|TP=0|ip-26-0-175-170]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=321|PP=0|TP=0|ip-26-0-168-95]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=177|PP=0|TP=0|ip-26-0-165-24]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=179|PP=0|TP=0|ip-26-0-165-24]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=82|PP=0|TP=0|ip-26-0-163-58]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=80|PP=0|TP=0|ip-26-0-163-58]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=371|PP=0|TP=0|ip-26-0-172-142]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=375|PP=0|TP=0|ip-26-0-172-142]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=493|PP=0|TP=0|ip-26-0-175-19]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=492|PP=0|TP=0|ip-26-0-175-19]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=494|PP=0|TP=0|ip-26-0-175-19]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=466|PP=0|TP=0|ip-26-0-175-132]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=173|PP=0|TP=0|ip-26-0-165-213]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=172|PP=0|TP=0|ip-26-0-165-213]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=169|PP=0|TP=0|ip-26-0-165-213]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=174|PP=0|TP=0|ip-26-0-165-213]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=353|PP=0|TP=0|ip-26-0-171-21]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=417|PP=0|TP=0|ip-26-0-173-7]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=297|PP=0|TP=0|ip-26-0-168-30]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=74|PP=0|TP=0|ip-26-0-163-43]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=462|PP=0|TP=0|ip-26-0-174-36]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=459|PP=0|TP=0|ip-26-0-174-36]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=461|PP=0|TP=0|ip-26-0-174-36]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=463|PP=0|TP=0|ip-26-0-174-36]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=511|PP=0|TP=0|ip-26-0-175-34]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=510|PP=0|TP=0|ip-26-0-175-34]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=478|PP=0|TP=0|ip-26-0-175-165]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=476|PP=0|TP=0|ip-26-0-175-165]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=479|PP=0|TP=0|ip-26-0-175-165]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=184|PP=0|TP=0|ip-26-0-165-38]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=191|PP=0|TP=0|ip-26-0-165-38]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=250|PP=0|TP=0|ip-26-0-167-177]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=249|PP=0|TP=0|ip-26-0-167-177]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=252|PP=0|TP=0|ip-26-0-167-177]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=254|PP=0|TP=0|ip-26-0-167-177]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=255|PP=0|TP=0|ip-26-0-167-177]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=253|PP=0|TP=0|ip-26-0-167-177]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=346|PP=0|TP=0|ip-26-0-170-160]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=361|PP=0|TP=0|ip-26-0-172-116]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=365|PP=0|TP=0|ip-26-0-172-116]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=164|PP=0|TP=0|ip-26-0-165-202]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=166|PP=0|TP=0|ip-26-0-165-202]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=268|PP=0|TP=0|ip-26-0-167-51]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=105|PP=0|TP=0|ip-26-0-164-187]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=109|PP=0|TP=0|ip-26-0-164-187]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=107|PP=0|TP=0|ip-26-0-164-187]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=380|PP=0|TP=0|ip-26-0-172-147]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=381|PP=0|TP=0|ip-26-0-172-147]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=24|PP=0|TP=0|ip-26-0-161-78]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=28|PP=0|TP=0|ip-26-0-161-78]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=413|PP=0|TP=0|ip-26-0-173-246]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=411|PP=0|TP=0|ip-26-0-173-246]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=219|PP=0|TP=0|ip-26-0-166-214]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=218|PP=0|TP=0|ip-26-0-166-214]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=51|PP=0|TP=0|ip-26-0-162-79]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=115|PP=0|TP=0|ip-26-0-164-207]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=113|PP=0|TP=0|ip-26-0-164-207]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=306|PP=0|TP=0|ip-26-0-168-34]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=311|PP=0|TP=0|ip-26-0-168-34]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=309|PP=0|TP=0|ip-26-0-168-34]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=308|PP=0|TP=0|ip-26-0-168-34]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=442|PP=0|TP=0|ip-26-0-174-196]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=445|PP=0|TP=0|ip-26-0-174-196]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=97|PP=0|TP=0|ip-26-0-164-18]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=41|PP=0|TP=0|ip-26-0-162-233]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=40|PP=0|TP=0|ip-26-0-162-233]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=237|PP=0|TP=0|ip-26-0-166-36]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=214|PP=0|TP=0|ip-26-0-166-15]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=278|PP=0|TP=0|ip-26-0-167-9]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=277|PP=0|TP=0|ip-26-0-167-9]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=424|PP=0|TP=0|ip-26-0-174-100]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=10|PP=0|TP=0|ip-26-0-160-242]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=335|PP=0|TP=0|ip-26-0-170-132]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=92|PP=0|TP=0|ip-26-0-164-0]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=93|PP=0|TP=0|ip-26-0-164-0]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=89|PP=0|TP=0|ip-26-0-164-0]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=90|PP=0|TP=0|ip-26-0-164-0]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=146|PP=0|TP=0|ip-26-0-165-131]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=201|PP=0|TP=0|ip-26-0-166-125]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=317|PP=0|TP=0|ip-26-0-168-52]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=312|PP=0|TP=0|ip-26-0-168-52]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=314|PP=0|TP=0|ip-26-0-168-52]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=438|PP=0|TP=0|ip-26-0-174-186]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=62|PP=0|TP=0|ip-26-0-163-226]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=283|PP=0|TP=0|ip-26-0-168-120]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=22|PP=0|TP=0|ip-26-0-161-138]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=402|PP=0|TP=0|ip-26-0-173-202]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=139|PP=0|TP=0|ip-26-0-164-75]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=394|PP=0|TP=0|ip-26-0-172-73]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=399|PP=0|TP=0|ip-26-0-172-73]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=341|PP=0|TP=0|ip-26-0-170-143]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=487|PP=0|TP=0|ip-26-0-175-170]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=324|PP=0|TP=0|ip-26-0-168-95]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=176|PP=0|TP=0|ip-26-0-165-24]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=86|PP=0|TP=0|ip-26-0-163-58]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=495|PP=0|TP=0|ip-26-0-175-19]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=490|PP=0|TP=0|ip-26-0-175-19]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=469|PP=0|TP=0|ip-26-0-175-132]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=170|PP=0|TP=0|ip-26-0-165-213]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=352|PP=0|TP=0|ip-26-0-171-21]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=298|PP=0|TP=0|ip-26-0-168-30]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=77|PP=0|TP=0|ip-26-0-163-43]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=456|PP=0|TP=0|ip-26-0-174-36]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=460|PP=0|TP=0|ip-26-0-174-36]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=506|PP=0|TP=0|ip-26-0-175-34]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=472|PP=0|TP=0|ip-26-0-175-165]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=185|PP=0|TP=0|ip-26-0-165-38]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=363|PP=0|TP=0|ip-26-0-172-116]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=366|PP=0|TP=0|ip-26-0-172-116]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=271|PP=0|TP=0|ip-26-0-167-51]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=264|PP=0|TP=0|ip-26-0-167-51]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=266|PP=0|TP=0|ip-26-0-167-51]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=265|PP=0|TP=0|ip-26-0-167-51]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=110|PP=0|TP=0|ip-26-0-164-187]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=377|PP=0|TP=0|ip-26-0-172-147]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=30|PP=0|TP=0|ip-26-0-161-78]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=27|PP=0|TP=0|ip-26-0-161-78]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=410|PP=0|TP=0|ip-26-0-173-246]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=412|PP=0|TP=0|ip-26-0-173-246]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=409|PP=0|TP=0|ip-26-0-173-246]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=223|PP=0|TP=0|ip-26-0-166-214]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=158|PP=0|TP=0|ip-26-0-165-164]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=153|PP=0|TP=0|ip-26-0-165-164]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=155|PP=0|TP=0|ip-26-0-165-164]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=304|PP=0|TP=0|ip-26-0-168-34]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=310|PP=0|TP=0|ip-26-0-168-34]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=440|PP=0|TP=0|ip-26-0-174-196]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=447|PP=0|TP=0|ip-26-0-174-196]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=444|PP=0|TP=0|ip-26-0-174-196]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=441|PP=0|TP=0|ip-26-0-174-196]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=45|PP=0|TP=0|ip-26-0-162-233]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=232|PP=0|TP=0|ip-26-0-166-36]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=246|PP=0|TP=0|ip-26-0-167-175]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=247|PP=0|TP=0|ip-26-0-167-175]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=243|PP=0|TP=0|ip-26-0-167-175]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=244|PP=0|TP=0|ip-26-0-167-175]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=131|PP=0|TP=0|ip-26-0-164-45]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=275|PP=0|TP=0|ip-26-0-167-9]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=9|PP=0|TP=0|ip-26-0-160-242]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=334|PP=0|TP=0|ip-26-0-170-132]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=94|PP=0|TP=0|ip-26-0-164-0]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=207|PP=0|TP=0|ip-26-0-166-125]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=315|PP=0|TP=0|ip-26-0-168-52]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=435|PP=0|TP=0|ip-26-0-174-186]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=56|PP=0|TP=0|ip-26-0-163-226]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=17|PP=0|TP=0|ip-26-0-161-138]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=395|PP=0|TP=0|ip-26-0-172-73]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=342|PP=0|TP=0|ip-26-0-170-143]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=464|PP=0|TP=0|ip-26-0-175-132]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=355|PP=0|TP=0|ip-26-0-171-21]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=78|PP=0|TP=0|ip-26-0-163-43]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=505|PP=0|TP=0|ip-26-0-175-34]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=474|PP=0|TP=0|ip-26-0-175-165]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=367|PP=0|TP=0|ip-26-0-172-116]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=362|PP=0|TP=0|ip-26-0-172-116]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=267|PP=0|TP=0|ip-26-0-167-51]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=379|PP=0|TP=0|ip-26-0-172-147]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=31|PP=0|TP=0|ip-26-0-161-78]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=415|PP=0|TP=0|ip-26-0-173-246]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=217|PP=0|TP=0|ip-26-0-166-214]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=55|PP=0|TP=0|ip-26-0-162-79]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=48|PP=0|TP=0|ip-26-0-162-79]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=119|PP=0|TP=0|ip-26-0-164-207]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=152|PP=0|TP=0|ip-26-0-165-164]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=47|PP=0|TP=0|ip-26-0-162-233]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=241|PP=0|TP=0|ip-26-0-167-175]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=208|PP=0|TP=0|ip-26-0-166-15]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=279|PP=0|TP=0|ip-26-0-167-9]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=431|PP=0|TP=0|ip-26-0-174-100]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=428|PP=0|TP=0|ip-26-0-174-100]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=205|PP=0|TP=0|ip-26-0-166-125]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=359|PP=0|TP=0|ip-26-0-171-21]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=75|PP=0|TP=0|ip-26-0-163-43]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=475|PP=0|TP=0|ip-26-0-175-165]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=382|PP=0|TP=0|ip-26-0-172-147]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=414|PP=0|TP=0|ip-26-0-173-246]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=154|PP=0|TP=0|ip-26-0-165-164]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=443|PP=0|TP=0|ip-26-0-174-196]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=43|PP=0|TP=0|ip-26-0-162-233]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=42|PP=0|TP=0|ip-26-0-162-233]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=235|PP=0|TP=0|ip-26-0-166-36]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=273|PP=0|TP=0|ip-26-0-167-9]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=429|PP=0|TP=0|ip-26-0-174-100]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=354|PP=0|TP=0|ip-26-0-171-21]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=25|PP=0|TP=0|ip-26-0-161-78]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=242|PP=0|TP=0|ip-26-0-167-175]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=430|PP=0|TP=0|ip-26-0-174-100]: Throughput logging complete
+[2024-12-02 11:53:52,300] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-02 11:53:52,299] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-02 11:53:52,301] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2739265 closing signal SIGTERM
+[2024-12-02 11:53:52,300] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2609724 closing signal SIGTERM
+[2024-12-02 11:53:52,299] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-02 11:53:52,301] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2739266 closing signal SIGTERM
+[2024-12-02 11:53:52,299] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3437308 closing signal SIGTERM
+[2024-12-02 11:53:52,300] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2609725 closing signal SIGTERM
+[2024-12-02 11:53:52,301] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2739267 closing signal SIGTERM
+[2024-12-02 11:53:52,301] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-02 11:53:52,301] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3344182 closing signal SIGTERM
+[2024-12-02 11:53:52,301] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3344183 closing signal SIGTERM
+srun: Job step aborted: Waiting up to 32 seconds for job step to finish.
+[2024-12-02 11:53:52,301] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3344184 closing signal SIGTERM
+[2024-12-02 11:53:52,300] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+slurmstepd: error: *** JOB 12179115 ON ip-26-0-160-225 CANCELLED AT 2024-12-02T11:53:52 ***
+[2024-12-02 11:53:52,299] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3437309 closing signal SIGTERM
+[2024-12-02 11:53:52,301] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3344185 closing signal SIGTERM
+[2024-12-02 11:53:52,299] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3437310 closing signal SIGTERM
+[2024-12-02 11:53:52,301] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3344186 closing signal SIGTERM
+[2024-12-02 11:53:52,300] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1404837 closing signal SIGTERM
+[2024-12-02 11:53:52,300] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-02 11:53:52,302] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-02 11:53:52,300] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-02 11:53:52,302] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2739268 closing signal SIGTERM
+[2024-12-02 11:53:52,301] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-02 11:53:52,302] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-02 11:53:52,300] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2609726 closing signal SIGTERM
+[2024-12-02 11:53:52,301] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-02 11:53:52,301] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-02 11:53:52,302] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 681854 closing signal SIGTERM
+[2024-12-02 11:53:52,300] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3437311 closing signal SIGTERM
+[2024-12-02 11:53:52,303] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 681855 closing signal SIGTERM
+[2024-12-02 11:53:52,301] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3437312 closing signal SIGTERM
+[2024-12-02 11:53:52,301] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3437313 closing signal SIGTERM
+[2024-12-02 11:53:52,303] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 681856 closing signal SIGTERM
+[2024-12-02 11:53:52,300] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-02 11:53:52,301] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-02 11:53:52,301] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-02 11:53:52,301] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-02 11:53:52,301] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 325515 closing signal SIGTERM
+[2024-12-02 11:53:52,302] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-02 11:53:52,301] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 325516 closing signal SIGTERM
+[2024-12-02 11:53:52,301] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-02 11:53:52,302] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3162018 closing signal SIGTERM
+[2024-12-02 11:53:52,301] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 325517 closing signal SIGTERM
+[2024-12-02 11:53:52,301] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-02 11:53:52,302] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-02 11:53:52,302] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1994454 closing signal SIGTERM
+[2024-12-02 11:53:52,302] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-02 11:53:52,301] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3989975 closing signal SIGTERM
+[2024-12-02 11:53:52,301] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2602018 closing signal SIGTERM
+[2024-12-02 11:53:52,302] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3162019 closing signal SIGTERM
+[2024-12-02 11:53:52,303] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 159630 closing signal SIGTERM
+[2024-12-02 11:53:52,300] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1404838 closing signal SIGTERM
+[2024-12-02 11:53:52,301] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3989976 closing signal SIGTERM
+[2024-12-02 11:53:52,301] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2602019 closing signal SIGTERM
+[2024-12-02 11:53:52,302] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3274207 closing signal SIGTERM
+[2024-12-02 11:53:52,302] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1994455 closing signal SIGTERM
+[2024-12-02 11:53:52,301] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3989977 closing signal SIGTERM
+[2024-12-02 11:53:52,302] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-02 11:53:52,301] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2602020 closing signal SIGTERM
+[2024-12-02 11:53:52,303] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-02 11:53:52,301] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3437314 closing signal SIGTERM
+[2024-12-02 11:53:52,302] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3284410 closing signal SIGTERM
+[2024-12-02 11:53:52,302] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3274208 closing signal SIGTERM
+[2024-12-02 11:53:52,301] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3989978 closing signal SIGTERM
+[2024-12-02 11:53:52,301] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-02 11:53:52,301] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2602021 closing signal SIGTERM
+[2024-12-02 11:53:52,302] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3284411 closing signal SIGTERM
+[2024-12-02 11:53:52,303] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 159631 closing signal SIGTERM
+[2024-12-02 11:53:52,301] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-02 11:53:52,301] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-02 11:53:52,302] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3989979 closing signal SIGTERM
+[2024-12-02 11:53:52,302] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 260974 closing signal SIGTERM
+[2024-12-02 11:53:52,302] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2602022 closing signal SIGTERM
+[2024-12-02 11:53:52,303] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2740061 closing signal SIGTERM
+[2024-12-02 11:53:52,303] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-02 11:53:52,301] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3437315 closing signal SIGTERM
+[2024-12-02 11:53:52,302] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3284412 closing signal SIGTERM
+[2024-12-02 11:53:52,301] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 172270 closing signal SIGTERM
+[2024-12-02 11:53:52,302] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 260975 closing signal SIGTERM
+[2024-12-02 11:53:52,301] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-02 11:53:52,303] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2740062 closing signal SIGTERM
+[2024-12-02 11:53:52,302] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3284413 closing signal SIGTERM
+[2024-12-02 11:53:52,301] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+12/02/2024 11:53:52 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: | 12179115 | 1.07G_dp512_tp1_pp1_acc1_mbs2_seq4096_zero0_l15_h2048_heads16 | 64    | 4096    | 2   | 1           | 1024 | 244.34  | 244.34  | 30725.66  | 182.42           | 42.90            | 43.41                | 460.65               | 263.66               | 264.83               | 20.08          | 36.23        | 512 | 1  | 1  | <nanotron.parallel.pipeline_parallel.engine.OneForwardOneBackwardPipelineEngine object at 0x7f4367510df0> | TensorParallelLinearMode.REDUCE_SCATTER | True          | 2048        | silu       | 15         | 16        | 16           | 4096    | 32768      | True                | torch.bfloat16 | 0          | 25                | True                    | 1.07G        | 1.07G        |
+[2024-12-02 11:53:52,302] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 4129965 closing signal SIGTERM
+12/02/2024 11:53:52 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: Throughput logging complete
+[2024-12-02 11:53:52,303] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2740063 closing signal SIGTERM
+12/02/2024 11:53:52 [INFO|DP=5|PP=0|TP=0|ip-26-0-160-225]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=2|PP=0|TP=0|ip-26-0-160-225]: Throughput logging complete
+12/02/2024 11:53:52 [INFO|DP=4|PP=0|TP=0|ip-26-0-160-225]: Throughput logging complete
+[2024-12-02 11:53:52,302] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 4129966 closing signal SIGTERM
+12/02/2024 11:53:52 [INFO|DP=7|PP=0|TP=0|ip-26-0-160-225]: Throughput logging complete
+[2024-12-02 11:53:52,303] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 875542 closing signal SIGTERM
+12/02/2024 11:53:52 [INFO|DP=1|PP=0|TP=0|ip-26-0-160-225]: Throughput logging complete
+[2024-12-02 11:53:52,301] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+12/02/2024 11:53:52 [INFO|DP=3|PP=0|TP=0|ip-26-0-160-225]: Throughput logging complete
+[2024-12-02 11:53:52,301] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 172271 closing signal SIGTERM
+12/02/2024 11:53:52 [INFO|DP=6|PP=0|TP=0|ip-26-0-160-225]: Throughput logging complete
+[2024-12-02 11:53:52,301] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-02 11:53:52,302] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-02 11:53:52,302] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 682379 closing signal SIGTERM
+[2024-12-02 11:53:52,301] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 172272 closing signal SIGTERM
+[2024-12-02 11:53:52,302] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-02 11:53:52,301] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-02 11:53:52,301] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-02 11:53:52,302] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-02 11:53:52,301] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-02 11:53:52,301] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 172273 closing signal SIGTERM
+[2024-12-02 11:53:52,302] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 4129967 closing signal SIGTERM
+[2024-12-02 11:53:52,301] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 61691 closing signal SIGTERM
+[2024-12-02 11:53:52,303] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 875543 closing signal SIGTERM
+[2024-12-02 11:53:52,302] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 60626 closing signal SIGTERM
+[2024-12-02 11:53:52,301] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-02 11:53:52,301] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 257502 closing signal SIGTERM
+[2024-12-02 11:53:52,300] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3544949 closing signal SIGTERM
+[2024-12-02 11:53:52,302] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-02 11:53:52,302] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 4129968 closing signal SIGTERM
+[2024-12-02 11:53:52,301] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 61692 closing signal SIGTERM
+[2024-12-02 11:53:52,303] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 875544 closing signal SIGTERM
+[2024-12-02 11:53:52,302] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-02 11:53:52,302] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3162020 closing signal SIGTERM
+[2024-12-02 11:53:52,303] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3344187 closing signal SIGTERM
+[2024-12-02 11:53:52,302] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 344088 closing signal SIGTERM
+[2024-12-02 11:53:52,300] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3544950 closing signal SIGTERM
+[2024-12-02 11:53:52,301] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 526861 closing signal SIGTERM
+[2024-12-02 11:53:52,302] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 325518 closing signal SIGTERM
+[2024-12-02 11:53:52,302] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1231876 closing signal SIGTERM
+[2024-12-02 11:53:52,302] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2739269 closing signal SIGTERM
+[2024-12-02 11:53:52,301] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 61693 closing signal SIGTERM
+[2024-12-02 11:53:52,302] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 60627 closing signal SIGTERM
+[2024-12-02 11:53:52,301] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 257503 closing signal SIGTERM
+[2024-12-02 11:53:52,302] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 682380 closing signal SIGTERM
+[2024-12-02 11:53:52,301] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3544951 closing signal SIGTERM
+[2024-12-02 11:53:52,302] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-02 11:53:52,301] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 526862 closing signal SIGTERM
+[2024-12-02 11:53:52,301] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2670903 closing signal SIGTERM
+[2024-12-02 11:53:52,301] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-02 11:53:52,302] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 4145618 closing signal SIGTERM
+[2024-12-02 11:53:52,302] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-02 11:53:52,302] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1231877 closing signal SIGTERM
+[2024-12-02 11:53:52,302] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 61694 closing signal SIGTERM
+[2024-12-02 11:53:52,302] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2527119 closing signal SIGTERM
+[2024-12-02 11:53:52,303] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3344188 closing signal SIGTERM
+[2024-12-02 11:53:52,302] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 682381 closing signal SIGTERM
+[2024-12-02 11:53:52,301] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3544952 closing signal SIGTERM
+[2024-12-02 11:53:52,302] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 526863 closing signal SIGTERM
+[2024-12-02 11:53:52,301] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2670904 closing signal SIGTERM
+[2024-12-02 11:53:52,301] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2670905 closing signal SIGTERM
+[2024-12-02 11:53:52,302] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 4145619 closing signal SIGTERM
+[2024-12-02 11:53:52,302] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1231878 closing signal SIGTERM
+[2024-12-02 11:53:52,302] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2739270 closing signal SIGTERM
+[2024-12-02 11:53:52,303] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2740064 closing signal SIGTERM
+[2024-12-02 11:53:52,301] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2609727 closing signal SIGTERM
+[2024-12-02 11:53:52,302] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 60628 closing signal SIGTERM
+[2024-12-02 11:53:52,302] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3274209 closing signal SIGTERM
+[2024-12-02 11:53:52,303] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3274210 closing signal SIGTERM
+[2024-12-02 11:53:52,302] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 344089 closing signal SIGTERM
+[2024-12-02 11:53:52,302] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 682382 closing signal SIGTERM
+[2024-12-02 11:53:52,303] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-02 11:53:52,302] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 325519 closing signal SIGTERM
+[2024-12-02 11:53:52,302] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 4145620 closing signal SIGTERM
+[2024-12-02 11:53:52,303] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 4145621 closing signal SIGTERM
+[2024-12-02 11:53:52,302] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1231879 closing signal SIGTERM
+[2024-12-02 11:53:52,301] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 812466 closing signal SIGTERM
+[2024-12-02 11:53:52,302] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 260976 closing signal SIGTERM
+[2024-12-02 11:53:52,302] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2739271 closing signal SIGTERM
+[2024-12-02 11:53:52,300] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-02 11:53:52,301] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2609728 closing signal SIGTERM
+[2024-12-02 11:53:52,302] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 60629 closing signal SIGTERM
+[2024-12-02 11:53:52,301] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3566644 closing signal SIGTERM
+[2024-12-02 11:53:52,302] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2527120 closing signal SIGTERM
+[2024-12-02 11:53:52,302] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 344090 closing signal SIGTERM
+[2024-12-02 11:53:52,303] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 682383 closing signal SIGTERM
+[2024-12-02 11:53:52,301] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3544953 closing signal SIGTERM
+[2024-12-02 11:53:52,302] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 295149 closing signal SIGTERM
+[2024-12-02 11:53:52,303] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 159632 closing signal SIGTERM
+[2024-12-02 11:53:52,302] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1994456 closing signal SIGTERM
+[2024-12-02 11:53:52,302] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1994457 closing signal SIGTERM
+[2024-12-02 11:53:52,302] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 810379 closing signal SIGTERM
+[2024-12-02 11:53:52,302] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1231880 closing signal SIGTERM
+[2024-12-02 11:53:52,301] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 812467 closing signal SIGTERM
+[2024-12-02 11:53:52,302] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2739272 closing signal SIGTERM
+[2024-12-02 11:53:52,304] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2740065 closing signal SIGTERM
+[2024-12-02 11:53:52,302] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2609729 closing signal SIGTERM
+[2024-12-02 11:53:52,302] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-02 11:53:52,301] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3566645 closing signal SIGTERM
+[2024-12-02 11:53:52,302] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3544954 closing signal SIGTERM
+[2024-12-02 11:53:52,302] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 295150 closing signal SIGTERM
+[2024-12-02 11:53:52,303] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 303079 closing signal SIGTERM
+[2024-12-02 11:53:52,301] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 812468 closing signal SIGTERM
+[2024-12-02 11:53:52,301] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3566646 closing signal SIGTERM
+[2024-12-02 11:53:52,302] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-02 11:53:52,304] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 682384 closing signal SIGTERM
+[2024-12-02 11:53:52,302] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3544955 closing signal SIGTERM
+[2024-12-02 11:53:52,302] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-02 11:53:52,302] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2670906 closing signal SIGTERM
+[2024-12-02 11:53:52,301] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 812469 closing signal SIGTERM
+[2024-12-02 11:53:52,302] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2609730 closing signal SIGTERM
+[2024-12-02 11:53:52,302] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 53178 closing signal SIGTERM
+[2024-12-02 11:53:52,304] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 682385 closing signal SIGTERM
+[2024-12-02 11:53:52,302] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2609731 closing signal SIGTERM
+[2024-12-02 11:53:52,304] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 682386 closing signal SIGTERM
+[2024-12-02 11:53:52,303] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 303080 closing signal SIGTERM
+[2024-12-02 11:53:52,302] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 359415 closing signal SIGTERM
+[2024-12-02 11:53:52,303] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 61695 closing signal SIGTERM
+[2024-12-02 11:53:52,303] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 53179 closing signal SIGTERM
+[2024-12-02 11:53:52,303] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2970503 closing signal SIGTERM
+[2024-12-02 11:53:52,303] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 303081 closing signal SIGTERM
+[2024-12-02 11:53:52,303] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3274211 closing signal SIGTERM
+[2024-12-02 11:53:52,303] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 303082 closing signal SIGTERM
+[2024-12-02 11:53:52,303] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-02 11:53:52,302] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 359416 closing signal SIGTERM
+[2024-12-02 11:53:52,303] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-02 11:53:52,303] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2970504 closing signal SIGTERM
+[2024-12-02 11:53:52,303] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 53180 closing signal SIGTERM
+[2024-12-02 11:53:52,302] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2527121 closing signal SIGTERM
+[2024-12-02 11:53:52,301] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 45947 closing signal SIGTERM
+[2024-12-02 11:53:52,303] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 106905 closing signal SIGTERM
+[2024-12-02 11:53:52,303] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 53181 closing signal SIGTERM
+[2024-12-02 11:53:52,302] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 172274 closing signal SIGTERM
+[2024-12-02 11:53:52,302] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 45948 closing signal SIGTERM
+[2024-12-02 11:53:52,304] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 681857 closing signal SIGTERM
+[2024-12-02 11:53:52,303] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2970505 closing signal SIGTERM
+[2024-12-02 11:53:52,302] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 810380 closing signal SIGTERM
+[2024-12-02 11:53:52,304] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3458435 closing signal SIGTERM
+[2024-12-02 11:53:52,301] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 169347 closing signal SIGTERM
+[2024-12-02 11:53:52,302] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 45949 closing signal SIGTERM
+[2024-12-02 11:53:52,303] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-02 11:53:52,302] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-02 11:53:52,303] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3344189 closing signal SIGTERM
+[2024-12-02 11:53:52,302] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 810381 closing signal SIGTERM
+[2024-12-02 11:53:52,301] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 169348 closing signal SIGTERM
+[2024-12-02 11:53:52,302] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 45950 closing signal SIGTERM
+[2024-12-02 11:53:52,303] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 106906 closing signal SIGTERM
+[2024-12-02 11:53:52,302] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-02 11:53:52,301] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 257504 closing signal SIGTERM
+[2024-12-02 11:53:52,303] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 159633 closing signal SIGTERM
+[2024-12-02 11:53:52,303] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2670907 closing signal SIGTERM
+[2024-12-02 11:53:52,301] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 247698 closing signal SIGTERM
+[2024-12-02 11:53:52,303] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 172275 closing signal SIGTERM
+[2024-12-02 11:53:52,301] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 169349 closing signal SIGTERM
+[2024-12-02 11:53:52,302] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 45951 closing signal SIGTERM
+[2024-12-02 11:53:52,304] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 681858 closing signal SIGTERM
+[2024-12-02 11:53:52,303] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2602023 closing signal SIGTERM
+[2024-12-02 11:53:52,303] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3566647 closing signal SIGTERM
+[2024-12-02 11:53:52,303] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2970506 closing signal SIGTERM
+[2024-12-02 11:53:52,303] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 303083 closing signal SIGTERM
+[2024-12-02 11:53:52,301] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 247699 closing signal SIGTERM
+[2024-12-02 11:53:52,304] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3458436 closing signal SIGTERM
+[2024-12-02 11:53:52,302] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 169350 closing signal SIGTERM
+[2024-12-02 11:53:52,303] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 106907 closing signal SIGTERM
+[2024-12-02 11:53:52,303] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 48036 closing signal SIGTERM
+[2024-12-02 11:53:52,303] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 186757 closing signal SIGTERM
+[2024-12-02 11:53:52,301] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 997429 closing signal SIGTERM
+[2024-12-02 11:53:52,302] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 257505 closing signal SIGTERM
+[2024-12-02 11:53:52,303] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 65269 closing signal SIGTERM
+[2024-12-02 11:53:52,303] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2670908 closing signal SIGTERM
+[2024-12-02 11:53:52,301] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 247700 closing signal SIGTERM
+[2024-12-02 11:53:52,303] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 172276 closing signal SIGTERM
+[2024-12-02 11:53:52,304] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3458437 closing signal SIGTERM
+[2024-12-02 11:53:52,303] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2602024 closing signal SIGTERM
+[2024-12-02 11:53:52,303] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-02 11:53:52,303] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 48037 closing signal SIGTERM
+[2024-12-02 11:53:52,303] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 186758 closing signal SIGTERM
+[2024-12-02 11:53:52,303] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 60630 closing signal SIGTERM
+[2024-12-02 11:53:52,301] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 997430 closing signal SIGTERM
+[2024-12-02 11:53:52,301] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-02 11:53:52,302] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 359417 closing signal SIGTERM
+[2024-12-02 11:53:52,303] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2670909 closing signal SIGTERM
+[2024-12-02 11:53:52,300] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1404839 closing signal SIGTERM
+[2024-12-02 11:53:52,302] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 169351 closing signal SIGTERM
+[2024-12-02 11:53:52,303] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2602025 closing signal SIGTERM
+[2024-12-02 11:53:52,303] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 186759 closing signal SIGTERM
+[2024-12-02 11:53:52,301] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 997431 closing signal SIGTERM
+[2024-12-02 11:53:52,303] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 65270 closing signal SIGTERM
+[2024-12-02 11:53:52,302] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 526864 closing signal SIGTERM
+[2024-12-02 11:53:52,303] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2670910 closing signal SIGTERM
+[2024-12-02 11:53:52,302] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 247701 closing signal SIGTERM
+[2024-12-02 11:53:52,301] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1404840 closing signal SIGTERM
+[2024-12-02 11:53:52,303] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-02 11:53:52,303] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 169352 closing signal SIGTERM
+[2024-12-02 11:53:52,303] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1539635 closing signal SIGTERM
+[2024-12-02 11:53:52,303] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 48038 closing signal SIGTERM
+[2024-12-02 11:53:52,303] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 186760 closing signal SIGTERM
+[2024-12-02 11:53:52,301] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 997432 closing signal SIGTERM
+[2024-12-02 11:53:52,303] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3162021 closing signal SIGTERM
+[2024-12-02 11:53:52,301] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1030821 closing signal SIGTERM
+[2024-12-02 11:53:52,303] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3544956 closing signal SIGTERM
+[2024-12-02 11:53:52,302] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 526865 closing signal SIGTERM
+[2024-12-02 11:53:52,302] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 192989 closing signal SIGTERM
+[2024-12-02 11:53:52,302] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1404841 closing signal SIGTERM
+[2024-12-02 11:53:52,303] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 169353 closing signal SIGTERM
+[2024-12-02 11:53:52,301] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 997433 closing signal SIGTERM
+[2024-12-02 11:53:52,303] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 65271 closing signal SIGTERM
+[2024-12-02 11:53:52,303] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3162022 closing signal SIGTERM
+[2024-12-02 11:53:52,302] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 192990 closing signal SIGTERM
+[2024-12-02 11:53:52,302] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1404842 closing signal SIGTERM
+[2024-12-02 11:53:52,303] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 169354 closing signal SIGTERM
+[2024-12-02 11:53:52,303] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 875545 closing signal SIGTERM
+[2024-12-02 11:53:52,302] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 997434 closing signal SIGTERM
+[2024-12-02 11:53:52,302] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 192991 closing signal SIGTERM
+[2024-12-02 11:53:52,302] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1404843 closing signal SIGTERM
+[2024-12-02 11:53:52,303] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-02 11:53:52,303] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3389527 closing signal SIGTERM
+[2024-12-02 11:53:52,304] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 45952 closing signal SIGTERM
+[2024-12-02 11:53:52,304] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 61696 closing signal SIGTERM
+[2024-12-02 11:53:52,303] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1539636 closing signal SIGTERM
+[2024-12-02 11:53:52,303] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 875546 closing signal SIGTERM
+[2024-12-02 11:53:52,303] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 53182 closing signal SIGTERM
+[2024-12-02 11:53:52,303] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 53183 closing signal SIGTERM
+[2024-12-02 11:53:52,303] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-02 11:53:52,301] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1030822 closing signal SIGTERM
+[2024-12-02 11:53:52,302] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 192992 closing signal SIGTERM
+[2024-12-02 11:53:52,302] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1404844 closing signal SIGTERM
+[2024-12-02 11:53:52,303] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1539637 closing signal SIGTERM
+[2024-12-02 11:53:52,303] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3566648 closing signal SIGTERM
+[2024-12-02 11:53:52,303] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 325520 closing signal SIGTERM
+[2024-12-02 11:53:52,303] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 192993 closing signal SIGTERM
+[2024-12-02 11:53:52,303] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 172277 closing signal SIGTERM
+[2024-12-02 11:53:52,303] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 371159 closing signal SIGTERM
+[2024-12-02 11:53:52,304] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3458438 closing signal SIGTERM
+[2024-12-02 11:53:52,303] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3389528 closing signal SIGTERM
+[2024-12-02 11:53:52,304] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 45953 closing signal SIGTERM
+[2024-12-02 11:53:52,303] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1539638 closing signal SIGTERM
+[2024-12-02 11:53:52,302] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3284414 closing signal SIGTERM
+[2024-12-02 11:53:52,303] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 236651 closing signal SIGTERM
+[2024-12-02 11:53:52,303] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 257506 closing signal SIGTERM
+[2024-12-02 11:53:52,301] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1030823 closing signal SIGTERM
+[2024-12-02 11:53:52,302] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 247702 closing signal SIGTERM
+[2024-12-02 11:53:52,303] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 192994 closing signal SIGTERM
+[2024-12-02 11:53:52,303] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-02 11:53:52,303] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3566649 closing signal SIGTERM
+[2024-12-02 11:53:52,304] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 303084 closing signal SIGTERM
+[2024-12-02 11:53:52,303] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-02 11:53:52,303] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 325521 closing signal SIGTERM
+[2024-12-02 11:53:52,303] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 192995 closing signal SIGTERM
+[2024-12-02 11:53:52,303] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 371160 closing signal SIGTERM
+[2024-12-02 11:53:52,304] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1231881 closing signal SIGTERM
+[2024-12-02 11:53:52,303] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3389529 closing signal SIGTERM
+[2024-12-02 11:53:52,304] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 106908 closing signal SIGTERM
+[2024-12-02 11:53:52,303] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 236652 closing signal SIGTERM
+[2024-12-02 11:53:52,303] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3566650 closing signal SIGTERM
+[2024-12-02 11:53:52,303] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3274212 closing signal SIGTERM
+[2024-12-02 11:53:52,303] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 247703 closing signal SIGTERM
+[2024-12-02 11:53:52,303] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 192996 closing signal SIGTERM
+[2024-12-02 11:53:52,303] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3389530 closing signal SIGTERM
+[2024-12-02 11:53:52,303] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 140306 closing signal SIGTERM
+[2024-12-02 11:53:52,303] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 236653 closing signal SIGTERM
+[2024-12-02 11:53:52,301] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 663495 closing signal SIGTERM
+[2024-12-02 11:53:52,303] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 634286 closing signal SIGTERM
+[2024-12-02 11:53:52,303] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 325522 closing signal SIGTERM
+[2024-12-02 11:53:52,303] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 247704 closing signal SIGTERM
+[2024-12-02 11:53:52,304] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 4145622 closing signal SIGTERM
+[2024-12-02 11:53:52,303] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 140307 closing signal SIGTERM
+[2024-12-02 11:53:52,304] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 681859 closing signal SIGTERM
+[2024-12-02 11:53:52,304] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 681860 closing signal SIGTERM
+[2024-12-02 11:53:52,303] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3566651 closing signal SIGTERM
+[2024-12-02 11:53:52,301] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 663496 closing signal SIGTERM
+[2024-12-02 11:53:52,303] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3162023 closing signal SIGTERM
+[2024-12-02 11:53:52,303] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2970507 closing signal SIGTERM
+[2024-12-02 11:53:52,304] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 159634 closing signal SIGTERM
+[2024-12-02 11:53:52,303] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1994458 closing signal SIGTERM
+[2024-12-02 11:53:52,303] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 247705 closing signal SIGTERM
+[2024-12-02 11:53:52,303] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 48039 closing signal SIGTERM
+[2024-12-02 11:53:52,304] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-02 11:53:52,301] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 663497 closing signal SIGTERM
+[2024-12-02 11:53:52,303] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3162024 closing signal SIGTERM
+[2024-12-02 11:53:52,302] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1030824 closing signal SIGTERM
+[2024-12-02 11:53:52,304] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 634287 closing signal SIGTERM
+[2024-12-02 11:53:52,304] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 61697 closing signal SIGTERM
+[2024-12-02 11:53:52,301] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 663498 closing signal SIGTERM
+[2024-12-02 11:53:52,304] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 159635 closing signal SIGTERM
+[2024-12-02 11:53:52,303] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1994459 closing signal SIGTERM
+[2024-12-02 11:53:52,304] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 4129969 closing signal SIGTERM
+[2024-12-02 11:53:52,304] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 45954 closing signal SIGTERM
+[2024-12-02 11:53:52,304] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 61698 closing signal SIGTERM
+[2024-12-02 11:53:52,304] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3224094 closing signal SIGTERM
+[2024-12-02 11:53:52,303] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2527122 closing signal SIGTERM
+[2024-12-02 11:53:52,304] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 634288 closing signal SIGTERM
+[2024-12-02 11:53:52,303] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-02 11:53:52,304] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 663499 closing signal SIGTERM
+slurmstepd: error: *** STEP 12179115.0 ON ip-26-0-160-225 CANCELLED AT 2024-12-02T11:53:52 ***
+[2024-12-02 11:53:52,304] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3989980 closing signal SIGTERM
+[2024-12-02 11:53:52,304] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 4129970 closing signal SIGTERM
+[2024-12-02 11:53:52,304] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 663500 closing signal SIGTERM
+[2024-12-02 11:53:52,298] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-02 11:53:52,304] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3989981 closing signal SIGTERM
+[2024-12-02 11:53:52,304] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 4129971 closing signal SIGTERM
+[2024-12-02 11:53:52,304] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2527123 closing signal SIGTERM
+[2024-12-02 11:53:52,304] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 65272 closing signal SIGTERM
+[2024-12-02 11:53:52,298] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2278524 closing signal SIGTERM
+[2024-12-02 11:53:52,304] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3989982 closing signal SIGTERM
+[2024-12-02 11:53:52,303] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 479673 closing signal SIGTERM
+[2024-12-02 11:53:52,304] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3224095 closing signal SIGTERM
+[2024-12-02 11:53:52,298] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2278525 closing signal SIGTERM
+[2024-12-02 11:53:52,303] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 479674 closing signal SIGTERM
+[2024-12-02 11:53:52,304] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3224096 closing signal SIGTERM
+[2024-12-02 11:53:52,304] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 663501 closing signal SIGTERM
+[2024-12-02 11:53:52,298] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2278526 closing signal SIGTERM
+[2024-12-02 11:53:52,304] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 4129972 closing signal SIGTERM
+[2024-12-02 11:53:52,301] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 58967 closing signal SIGTERM
+[2024-12-02 11:53:52,304] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3284415 closing signal SIGTERM
+[2024-12-02 11:53:52,303] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-02 11:53:52,298] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2278527 closing signal SIGTERM
+[2024-12-02 11:53:52,301] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 58968 closing signal SIGTERM
+[2024-12-02 11:53:52,304] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 106909 closing signal SIGTERM
+[2024-12-02 11:53:52,299] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2278528 closing signal SIGTERM
+[2024-12-02 11:53:52,304] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 479675 closing signal SIGTERM
+[2024-12-02 11:53:52,304] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 140308 closing signal SIGTERM
+[2024-12-02 11:53:52,305] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2740066 closing signal SIGTERM
+[2024-12-02 11:53:52,301] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 58969 closing signal SIGTERM
+[2024-12-02 11:53:52,303] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 257507 closing signal SIGTERM
+[2024-12-02 11:53:52,301] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2278529 closing signal SIGTERM
+[2024-12-02 11:53:52,304] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 479676 closing signal SIGTERM
+[2024-12-02 11:53:52,304] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 140309 closing signal SIGTERM
+[2024-12-02 11:53:52,301] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 58970 closing signal SIGTERM
+[2024-12-02 11:53:52,303] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 257508 closing signal SIGTERM
+[2024-12-02 11:53:52,301] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2278530 closing signal SIGTERM
+[2024-12-02 11:53:52,301] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 58971 closing signal SIGTERM
+[2024-12-02 11:53:52,303] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 257509 closing signal SIGTERM
+[2024-12-02 11:53:52,301] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2278531 closing signal SIGTERM
+[2024-12-02 11:53:52,303] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 58972 closing signal SIGTERM
+[2024-12-02 11:53:52,304] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 48040 closing signal SIGTERM
+[2024-12-02 11:53:52,301] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 283197 closing signal SIGTERM
+[2024-12-02 11:53:52,305] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2740067 closing signal SIGTERM
+[2024-12-02 11:53:52,304] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3428698 closing signal SIGTERM
+[2024-12-02 11:53:52,301] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 283198 closing signal SIGTERM
+[2024-12-02 11:53:52,305] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2740068 closing signal SIGTERM
+[2024-12-02 11:53:52,304] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 60631 closing signal SIGTERM
+[2024-12-02 11:53:52,303] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 236654 closing signal SIGTERM
+[2024-12-02 11:53:52,302] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 283199 closing signal SIGTERM
+[2024-12-02 11:53:52,304] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 236655 closing signal SIGTERM
+[2024-12-02 11:53:52,302] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 283200 closing signal SIGTERM
+[2024-12-02 11:53:52,302] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1046178 closing signal SIGTERM
+[2024-12-02 11:53:52,302] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 283201 closing signal SIGTERM
+[2024-12-02 11:53:52,302] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1046179 closing signal SIGTERM
+[2024-12-02 11:53:52,304] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3428699 closing signal SIGTERM
+[2024-12-02 11:53:52,302] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 283202 closing signal SIGTERM
+[2024-12-02 11:53:52,305] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 875547 closing signal SIGTERM
+[2024-12-02 11:53:52,302] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1046180 closing signal SIGTERM
+[2024-12-02 11:53:52,302] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 810382 closing signal SIGTERM
+[2024-12-02 11:53:52,303] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 810383 closing signal SIGTERM
+[2024-12-02 11:53:52,302] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1046181 closing signal SIGTERM
+[2024-12-02 11:53:52,302] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 260977 closing signal SIGTERM
+[2024-12-02 11:53:52,304] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1046182 closing signal SIGTERM
+[2024-12-02 11:53:52,304] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1994460 closing signal SIGTERM
+[2024-12-02 11:53:52,304] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 260978 closing signal SIGTERM
+[2024-12-02 11:53:52,305] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 681861 closing signal SIGTERM
+[2024-12-02 11:53:52,305] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3274213 closing signal SIGTERM
+[2024-12-02 11:53:52,304] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3428700 closing signal SIGTERM
+[2024-12-02 11:53:52,304] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 65273 closing signal SIGTERM
+[2024-12-02 11:53:52,304] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 260979 closing signal SIGTERM
+[2024-12-02 11:53:52,304] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3428701 closing signal SIGTERM
+[2024-12-02 11:53:52,304] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3162025 closing signal SIGTERM
+[2024-12-02 11:53:52,304] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 344091 closing signal SIGTERM
+[2024-12-02 11:53:52,301] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3629796 closing signal SIGTERM
+[2024-12-02 11:53:52,304] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 810384 closing signal SIGTERM
+[2024-12-02 11:53:52,304] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1994461 closing signal SIGTERM
+[2024-12-02 11:53:52,301] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3629797 closing signal SIGTERM
+[2024-12-02 11:53:52,304] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 810385 closing signal SIGTERM
+[2024-12-02 11:53:52,305] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3274214 closing signal SIGTERM
+[2024-12-02 11:53:52,304] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 65274 closing signal SIGTERM
+[2024-12-02 11:53:52,301] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3629798 closing signal SIGTERM
+[2024-12-02 11:53:52,304] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 58973 closing signal SIGTERM
+[2024-12-02 11:53:52,304] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 58974 closing signal SIGTERM
+[2024-12-02 11:53:52,301] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3629799 closing signal SIGTERM
+[2024-12-02 11:53:52,301] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 812470 closing signal SIGTERM
+[2024-12-02 11:53:52,305] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 260980 closing signal SIGTERM
+[2024-12-02 11:53:52,301] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3629800 closing signal SIGTERM
+[2024-12-02 11:53:52,304] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 810386 closing signal SIGTERM
+[2024-12-02 11:53:52,304] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 812471 closing signal SIGTERM
+[2024-12-02 11:53:52,305] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 875548 closing signal SIGTERM
+[2024-12-02 11:53:52,304] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3629801 closing signal SIGTERM
+[2024-12-02 11:53:52,304] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 48041 closing signal SIGTERM
+[2024-12-02 11:53:52,305] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3284416 closing signal SIGTERM
+[2024-12-02 11:53:52,305] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 303085 closing signal SIGTERM
+[2024-12-02 11:53:52,304] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 526866 closing signal SIGTERM
+[2024-12-02 11:53:52,305] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 283203 closing signal SIGTERM
+[2024-12-02 11:53:52,304] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2527124 closing signal SIGTERM
+[2024-12-02 11:53:52,305] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 344092 closing signal SIGTERM
+[2024-12-02 11:53:52,302] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 295151 closing signal SIGTERM
+[2024-12-02 11:53:52,304] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 4145623 closing signal SIGTERM
+[2024-12-02 11:53:52,305] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 106910 closing signal SIGTERM
+[2024-12-02 11:53:52,304] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2527125 closing signal SIGTERM
+[2024-12-02 11:53:52,304] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2970508 closing signal SIGTERM
+[2024-12-02 11:53:52,302] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 295152 closing signal SIGTERM
+[2024-12-02 11:53:52,305] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 159636 closing signal SIGTERM
+[2024-12-02 11:53:52,305] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3629802 closing signal SIGTERM
+[2024-12-02 11:53:52,304] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1231882 closing signal SIGTERM
+[2024-12-02 11:53:52,305] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 48042 closing signal SIGTERM
+[2024-12-02 11:53:52,304] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2527126 closing signal SIGTERM
+[2024-12-02 11:53:52,305] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 344093 closing signal SIGTERM
+[2024-12-02 11:53:52,305] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 303086 closing signal SIGTERM
+[2024-12-02 11:53:52,304] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 526867 closing signal SIGTERM
+[2024-12-02 11:53:52,305] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 283204 closing signal SIGTERM
+[2024-12-02 11:53:52,305] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 48043 closing signal SIGTERM
+[2024-12-02 11:53:52,304] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 65275 closing signal SIGTERM
+[2024-12-02 11:53:52,304] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 65276 closing signal SIGTERM
+[2024-12-02 11:53:52,305] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 4145624 closing signal SIGTERM
+[2024-12-02 11:53:52,305] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3629804 closing signal SIGTERM
+[2024-12-02 11:53:52,305] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 106911 closing signal SIGTERM
+[2024-12-02 11:53:52,304] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3224097 closing signal SIGTERM
+[2024-12-02 11:53:52,305] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 344094 closing signal SIGTERM
+[2024-12-02 11:53:52,306] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3458439 closing signal SIGTERM
+[2024-12-02 11:53:52,305] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 260981 closing signal SIGTERM
+[2024-12-02 11:53:52,304] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 526868 closing signal SIGTERM
+[2024-12-02 11:53:52,305] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 4145625 closing signal SIGTERM
+[2024-12-02 11:53:52,305] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 812472 closing signal SIGTERM
+[2024-12-02 11:53:52,306] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 159637 closing signal SIGTERM
+[2024-12-02 11:53:52,302] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 359418 closing signal SIGTERM
+[2024-12-02 11:53:52,303] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 359419 closing signal SIGTERM
+[2024-12-02 11:53:52,304] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 479677 closing signal SIGTERM
+[2024-12-02 11:53:52,303] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 359420 closing signal SIGTERM
+[2024-12-02 11:53:52,306] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3458440 closing signal SIGTERM
+[2024-12-02 11:53:52,304] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 60632 closing signal SIGTERM
+[2024-12-02 11:53:52,303] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 359421 closing signal SIGTERM
+[2024-12-02 11:53:52,306] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3458441 closing signal SIGTERM
+[2024-12-02 11:53:52,305] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1539639 closing signal SIGTERM
+[2024-12-02 11:53:52,304] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 186761 closing signal SIGTERM
+[2024-12-02 11:53:52,304] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 60633 closing signal SIGTERM
+[2024-12-02 11:53:52,305] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 634289 closing signal SIGTERM
+[2024-12-02 11:53:52,303] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 371161 closing signal SIGTERM
+[2024-12-02 11:53:52,303] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 371162 closing signal SIGTERM
+[2024-12-02 11:53:52,304] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 371163 closing signal SIGTERM
+[2024-12-02 11:53:52,305] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1046183 closing signal SIGTERM
+[2024-12-02 11:53:52,305] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1539640 closing signal SIGTERM
+[2024-12-02 11:53:52,305] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 634290 closing signal SIGTERM
+[2024-12-02 11:53:52,305] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1046184 closing signal SIGTERM
+[2024-12-02 11:53:52,305] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1046185 closing signal SIGTERM
+[2024-12-02 11:53:52,305] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 663502 closing signal SIGTERM
+[2024-12-02 11:53:52,305] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2970509 closing signal SIGTERM
+[2024-12-02 11:53:52,305] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 295153 closing signal SIGTERM
+[2024-12-02 11:53:52,305] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 106912 closing signal SIGTERM
+[2024-12-02 11:53:52,305] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3284417 closing signal SIGTERM
+[2024-12-02 11:53:52,305] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2970510 closing signal SIGTERM
+[2024-12-02 11:53:52,305] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 295154 closing signal SIGTERM
+[2024-12-02 11:53:52,305] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1231883 closing signal SIGTERM
+[2024-12-02 11:53:52,305] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 295155 closing signal SIGTERM
+[2024-12-02 11:53:52,305] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 295156 closing signal SIGTERM
+[2024-12-02 11:53:52,304] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3389531 closing signal SIGTERM
+[2024-12-02 11:53:52,304] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3389532 closing signal SIGTERM
+[2024-12-02 11:53:52,306] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 186762 closing signal SIGTERM
+[2024-12-02 11:53:52,305] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 812473 closing signal SIGTERM
+[2024-12-02 11:53:52,306] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 186763 closing signal SIGTERM
+[2024-12-02 11:53:52,306] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 875549 closing signal SIGTERM
+[2024-12-02 11:53:52,305] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 140310 closing signal SIGTERM
+[2024-12-02 11:53:52,305] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 140311 closing signal SIGTERM
+[2024-12-02 11:53:52,305] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 140312 closing signal SIGTERM
+[2024-12-02 11:53:52,304] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3428702 closing signal SIGTERM
+[2024-12-02 11:53:52,304] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1030825 closing signal SIGTERM
+[2024-12-02 11:53:52,306] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3458442 closing signal SIGTERM
+[2024-12-02 11:53:52,304] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1030826 closing signal SIGTERM
+[2024-12-02 11:53:52,304] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1030827 closing signal SIGTERM
+[2024-12-02 11:53:52,304] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1030828 closing signal SIGTERM
+[2024-12-02 11:53:52,306] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 186764 closing signal SIGTERM
+[2024-12-02 11:53:52,306] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 140313 closing signal SIGTERM
+[2024-12-02 11:53:52,306] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3389533 closing signal SIGTERM
+[2024-12-02 11:53:52,306] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 479678 closing signal SIGTERM
+[2024-12-02 11:53:52,306] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3389534 closing signal SIGTERM
+[2024-12-02 11:53:52,306] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 344095 closing signal SIGTERM
+[2024-12-02 11:53:52,306] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 479679 closing signal SIGTERM
+[2024-12-02 11:53:52,306] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 479680 closing signal SIGTERM
+[2024-12-02 11:53:52,306] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 371164 closing signal SIGTERM
+[2024-12-02 11:53:52,306] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 53184 closing signal SIGTERM
+[2024-12-02 11:53:52,306] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 53185 closing signal SIGTERM
+[2024-12-02 11:53:52,305] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1539641 closing signal SIGTERM
+[2024-12-02 11:53:52,305] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1539642 closing signal SIGTERM
+[2024-12-02 11:53:52,306] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 236656 closing signal SIGTERM
+[2024-12-02 11:53:52,306] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3428703 closing signal SIGTERM
+[2024-12-02 11:53:52,306] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3428704 closing signal SIGTERM
+[2024-12-02 11:53:52,306] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 359422 closing signal SIGTERM
+[2024-12-02 11:53:52,306] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3428705 closing signal SIGTERM
+[2024-12-02 11:53:52,306] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3224098 closing signal SIGTERM
+[2024-12-02 11:53:52,306] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3224099 closing signal SIGTERM
+[2024-12-02 11:53:52,306] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3224100 closing signal SIGTERM
+[2024-12-02 11:53:52,308] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3224101 closing signal SIGTERM
+[2024-12-02 11:53:52,305] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 634291 closing signal SIGTERM
+[2024-12-02 11:53:52,305] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 634292 closing signal SIGTERM
+[2024-12-02 11:53:52,306] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 236657 closing signal SIGTERM
+[2024-12-02 11:53:52,307] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 236658 closing signal SIGTERM
+[2024-12-02 11:53:52,307] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 634293 closing signal SIGTERM
+[2024-12-02 11:53:52,306] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 371165 closing signal SIGTERM
+[2024-12-02 11:53:52,306] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 371166 closing signal SIGTERM
+[2024-12-02 11:53:52,305] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 997435 closing signal SIGTERM
+[2024-12-02 11:53:52,305] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 997436 closing signal SIGTERM
diff --git a/logs/12344322-bench_3.56G_dp16_tp8_pp1_acc1_mbs16_seq2048_zero0_l28_h3072_heads24.out b/logs/12344322-bench_3.56G_dp16_tp8_pp1_acc1_mbs16_seq2048_zero0_l28_h3072_heads24.out
new file mode 100644
index 0000000000000000000000000000000000000000..7d3b0deed222ef13fd9d1cf72009d7d7eebf54d2
--- /dev/null
+++ b/logs/12344322-bench_3.56G_dp16_tp8_pp1_acc1_mbs16_seq2048_zero0_l28_h3072_heads24.out
@@ -0,0 +1,1245 @@
++ source /etc/profile.d/modules.sh
+++ . /usr/share/modules/init/bash
++++ unset _mlshdbg
++++ '[' 0 = 1 ']'
++++ unset _mlre _mlIFS
++++ '[' -n x ']'
++++ _mlIFS=' 	
+'
++++ IFS=' '
++++ '[' -n '' ']'
+++++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash autoinit
++++ _mlcode='module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+MODULEPATH_modshare=/etc/environment-modules/modules:1:/usr/share/modules/\$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1; export MODULEPATH_modshare;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+LOADEDMODULES=; export LOADEDMODULES;
+MODULEPATH=/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/\$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles; export MODULEPATH;
+test 0;'
++++ _mlret=0
++++ '[' -n x ']'
++++ IFS=' 	
+'
++++ unset _mlIFS
++++ unset _mlre _mlv _mlrv
++++ '[' 0 -eq 0 ']'
++++ eval 'module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+MODULEPATH_modshare=/etc/environment-modules/modules:1:/usr/share/modules/\$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1; export MODULEPATH_modshare;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+LOADEDMODULES=; export LOADEDMODULES;
+MODULEPATH=/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/\$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles; export MODULEPATH;
+test 0;'
+++++ MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl
+++++ export MODULES_CMD
+++++ ENV=/usr/share/modules/init/profile.sh
+++++ export ENV
+++++ MODULEPATH_modshare='/etc/environment-modules/modules:1:/usr/share/modules/$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1'
+++++ export MODULEPATH_modshare
+++++ BASH_ENV=/usr/share/modules/init/bash
+++++ export BASH_ENV
+++++ MODULESHOME=/usr/share/modules
+++++ export MODULESHOME
+++++ LOADEDMODULES=
+++++ export LOADEDMODULES
+++++ MODULEPATH='/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles'
+++++ export MODULEPATH
+++++ test 0
++++ '[' 0 = 1 ']'
++++ '[' -t 2 ']'
++++ export -f module
++++ export -f switchml
++++ '[' 5 -ge 3 ']'
++++ [[ ehxB =~ i ]]
++++ [[ ! :/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin: =~ :/usr/bin: ]]
++++ '[' '!' -n '' ']'
++++ MANPATH=:
++++ export MANPATH
+++++ manpath
++++ [[ ! :/admin/home/nouamane/.local/share/man:/fsx/nouamane/miniconda/envs/2-1-cu121/man:/fsx/nouamane/miniconda/envs/2-1-cu121/share/man:/fsx/nouamane/miniconda/man:/fsx/nouamane/miniconda/share/man:/opt/amazon/openmpi/share/man:/opt/amazon/efa/share/man:/opt/slurm/share/man:/usr/local/man:/usr/local/share/man:/usr/share/man:/admin/home/nouamane/.fzf/man:: =~ :/usr/share/man: ]]
++++ unset _mlcode _mlret
++++ '[' -n '' ']'
++ module load cuda/12.1
++ unset _mlshdbg
++ '[' 0 = 1 ']'
++ unset _mlre _mlIFS
++ '[' -n x ']'
++ _mlIFS=' 	
+'
++ IFS=' '
++ '[' -n '' ']'
+++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash load cuda/12.1
++ eval 'CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include;' export 'CPATH;
+LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:;' export 'LD_LIBRARY_PATH;
+FI_EFA_FORK_SAFE=1;' export 'FI_EFA_FORK_SAFE;
+MANPATH=/usr/local/cuda-12.1/share/man::;' export 'MANPATH;
+LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64;' export 'LIBRARY_PATH;
+_LMFILES_=/usr/share/modules/modulefiles/cuda/12.1;' export '_LMFILES_;
+LOADEDMODULES=cuda/12.1;' export 'LOADEDMODULES;
+MPI_PATH=/opt/amazon/openmpi;' export 'MPI_PATH;
+NCCL_HOME_modshare=/opt/nccl/build:1;' export 'NCCL_HOME_modshare;
+NCCL_PROTO=simple;' export 'NCCL_PROTO;
+MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1;' export 'MANPATH_modshare;
+LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1;' export 'LIBRARY_PATH_modshare;
+NCCL_SOCKET_IFNAME=enp;' export 'NCCL_SOCKET_IFNAME;
+AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl;' export 'AWS_OFI_NCCL_HOME;
+NCCL_HOME=/opt/nccl/build;' export 'NCCL_HOME;
+FI_PROVIDER=efa;' export 'FI_PROVIDER;
+AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1;' export 'AWS_OFI_NCCL_HOME_modshare;
+CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1;' export 'CPATH_modshare;
+LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2::1:/opt/nccl/build/lib:1:/opt/aws-ofi-nccl/lib:2:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/efa/lib:1;' export 'LD_LIBRARY_PATH_modshare;
+FI_EFA_ENABLE_SHM_TRANSFER=1;' export 'FI_EFA_ENABLE_SHM_TRANSFER;
+_LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1;' export '_LMFILES__modshare;
+LOADEDMODULES_modshare=cuda/12.1:1;' export 'LOADEDMODULES_modshare;
+MPI_PATH_modshare=/opt/amazon/openmpi:1;' export 'MPI_PATH_modshare;
+PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin;' export 'PATH;
+CUDA_HOME=/usr/local/cuda-12.1;' export 'CUDA_HOME;
+PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/usr/local/cuda-12.1/include:1:/opt/amazon/efa/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/usr/sbin:1:/fsx/nouamane/miniconda/bin:1:/opt/slurm/bin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:2:/opt/amazon/openmpi/bin:1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1;' export 'PATH_modshare;
+test' '0;'
+++ CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include
+++ export CPATH
+++ LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:
+++ export LD_LIBRARY_PATH
+++ FI_EFA_FORK_SAFE=1
+++ export FI_EFA_FORK_SAFE
+++ MANPATH=/usr/local/cuda-12.1/share/man::
+++ export MANPATH
+++ LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64
+++ export LIBRARY_PATH
+++ _LMFILES_=/usr/share/modules/modulefiles/cuda/12.1
+++ export _LMFILES_
+++ LOADEDMODULES=cuda/12.1
+++ export LOADEDMODULES
+++ MPI_PATH=/opt/amazon/openmpi
+++ export MPI_PATH
+++ NCCL_HOME_modshare=/opt/nccl/build:1
+++ export NCCL_HOME_modshare
+++ NCCL_PROTO=simple
+++ export NCCL_PROTO
+++ MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1
+++ export MANPATH_modshare
+++ LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1
+++ export LIBRARY_PATH_modshare
+++ NCCL_SOCKET_IFNAME=enp
+++ export NCCL_SOCKET_IFNAME
+++ AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl
+++ export AWS_OFI_NCCL_HOME
+++ NCCL_HOME=/opt/nccl/build
+++ export NCCL_HOME
+++ FI_PROVIDER=efa
+++ export FI_PROVIDER
+++ AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1
+++ export AWS_OFI_NCCL_HOME_modshare
+++ CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1
+++ export CPATH_modshare
+++ LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2::1:/opt/nccl/build/lib:1:/opt/aws-ofi-nccl/lib:2:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/efa/lib:1
+++ export LD_LIBRARY_PATH_modshare
+++ FI_EFA_ENABLE_SHM_TRANSFER=1
+++ export FI_EFA_ENABLE_SHM_TRANSFER
+++ _LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1
+++ export _LMFILES__modshare
+++ LOADEDMODULES_modshare=cuda/12.1:1
+++ export LOADEDMODULES_modshare
+++ MPI_PATH_modshare=/opt/amazon/openmpi:1
+++ export MPI_PATH_modshare
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ export PATH
+++ CUDA_HOME=/usr/local/cuda-12.1
+++ export CUDA_HOME
+++ PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/usr/local/cuda-12.1/include:1:/opt/amazon/efa/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/usr/sbin:1:/fsx/nouamane/miniconda/bin:1:/opt/slurm/bin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:2:/opt/amazon/openmpi/bin:1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1
+++ export PATH_modshare
+++ test 0
++ _mlstatus=0
++ '[' -n x ']'
++ IFS=' 	
+'
++ unset _mlre _mlv _mlrv _mlIFS
++ '[' -n '' ']'
++ unset _mlshdbg
++ return 0
++ source /fsx/nouamane/miniconda/bin/activate
+++ _CONDA_ROOT=/fsx/nouamane/miniconda
+++ . /fsx/nouamane/miniconda/etc/profile.d/conda.sh
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ '[' -z x ']'
+++ conda activate
+++ local cmd=activate
+++ case "$cmd" in
+++ __conda_activate activate
+++ '[' -n '' ']'
+++ local ask_conda
++++ PS1=
++++ __conda_exe shell.posix activate
++++ /fsx/nouamane/miniconda/bin/conda shell.posix activate
+++ ask_conda='. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_2='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\'''
+++ eval '. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_2='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\'''
++++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh
+++++ test -n ''
+++++ unset XML_CATALOG_FILES
+++++ unset xml_catalog_files_libxml2
++++ PS1='(base) '
++++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++++ export CONDA_PREFIX=/fsx/nouamane/miniconda
++++ CONDA_PREFIX=/fsx/nouamane/miniconda
++++ export CONDA_SHLVL=3
++++ CONDA_SHLVL=3
++++ export CONDA_DEFAULT_ENV=base
++++ CONDA_DEFAULT_ENV=base
++++ export 'CONDA_PROMPT_MODIFIER=(base) '
++++ CONDA_PROMPT_MODIFIER='(base) '
++++ export CONDA_PREFIX_2=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ CONDA_PREFIX_2=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ __conda_hashr
+++ '[' -n '' ']'
+++ '[' -n '' ']'
+++ hash -r
++ conda activate 2-1-cu121
++ local cmd=activate
++ case "$cmd" in
++ __conda_activate activate 2-1-cu121
++ '[' -n '' ']'
++ local ask_conda
+++ PS1='(base) '
+++ __conda_exe shell.posix activate 2-1-cu121
+++ /fsx/nouamane/miniconda/bin/conda shell.posix activate 2-1-cu121
++ ask_conda='PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''4'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_3='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
++ eval 'PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''4'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_3='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
+++ PS1='(2-1-cu121) '
+++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ export CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ export CONDA_SHLVL=4
+++ CONDA_SHLVL=4
+++ export CONDA_DEFAULT_ENV=2-1-cu121
+++ CONDA_DEFAULT_ENV=2-1-cu121
+++ export 'CONDA_PROMPT_MODIFIER=(2-1-cu121) '
+++ CONDA_PROMPT_MODIFIER='(2-1-cu121) '
+++ export CONDA_PREFIX_3=/fsx/nouamane/miniconda
+++ CONDA_PREFIX_3=/fsx/nouamane/miniconda
+++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ export _CE_M=
+++ _CE_M=
+++ export _CE_CONDA=
+++ _CE_CONDA=
+++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh
++++ test -n ''
++++ xml_catalog_files_libxml2=
++++ XML_CATALOG_FILES=
++++ conda_catalog_files=
++++ ifs_libxml2=' 	
+'
++++ IFS=' '
++++ rem=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ for pre in ${rem}
++++ test '' = /fsx/nouamane/miniconda/envs/2-1-cu121
++++ conda_catalog_files=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ rem=
++++ IFS=' 	
+'
++++ conda_catalog_files='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ export 'XML_CATALOG_FILES=file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ XML_CATALOG_FILES='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ unset conda_catalog_files ifs_libxml2 rem
++ __conda_hashr
++ '[' -n '' ']'
++ '[' -n '' ']'
++ hash -r
++ export PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++ PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ scontrol show hostnames 'ip-26-0-161-[78,142,178],ip-26-0-162-180,ip-26-0-165-[24,59],ip-26-0-166-125,ip-26-0-167-245,ip-26-0-168-[95,120],ip-26-0-170-31,ip-26-0-171-[21,230],ip-26-0-173-[202,246],ip-26-0-174-36'
++ export 'NODELIST=ip-26-0-161-78
+ip-26-0-161-142
+ip-26-0-161-178
+ip-26-0-162-180
+ip-26-0-165-24
+ip-26-0-165-59
+ip-26-0-166-125
+ip-26-0-167-245
+ip-26-0-168-95
+ip-26-0-168-120
+ip-26-0-170-31
+ip-26-0-171-21
+ip-26-0-171-230
+ip-26-0-173-202
+ip-26-0-173-246
+ip-26-0-174-36'
++ NODELIST='ip-26-0-161-78
+ip-26-0-161-142
+ip-26-0-161-178
+ip-26-0-162-180
+ip-26-0-165-24
+ip-26-0-165-59
+ip-26-0-166-125
+ip-26-0-167-245
+ip-26-0-168-95
+ip-26-0-168-120
+ip-26-0-170-31
+ip-26-0-171-21
+ip-26-0-171-230
+ip-26-0-173-202
+ip-26-0-173-246
+ip-26-0-174-36'
+++ head -n1
+++ scontrol show hostnames 'ip-26-0-161-[78,142,178],ip-26-0-162-180,ip-26-0-165-[24,59],ip-26-0-166-125,ip-26-0-167-245,ip-26-0-168-[95,120],ip-26-0-170-31,ip-26-0-171-[21,230],ip-26-0-173-[202,246],ip-26-0-174-36'
++ export MASTER_NODE=ip-26-0-161-78
++ MASTER_NODE=ip-26-0-161-78
++ export MASTER_PORT=12356
++ MASTER_PORT=12356
++ export NNODES=16
++ NNODES=16
++ export GPUS_PER_NODE=8
++ GPUS_PER_NODE=8
++ export WORLD_SIZE=128
++ WORLD_SIZE=128
++ export CUDA_DEVICE_MAX_CONNECTIONS=1
++ CUDA_DEVICE_MAX_CONNECTIONS=1
++ export NANOTRON_BENCHMARK=1
++ NANOTRON_BENCHMARK=1
++ echo 'Master node: ip-26-0-161-78'
+Master node: ip-26-0-161-78
++ echo 'All nodes: ip-26-0-161-78
+ip-26-0-161-142
+ip-26-0-161-178
+ip-26-0-162-180
+ip-26-0-165-24
+ip-26-0-165-59
+ip-26-0-166-125
+ip-26-0-167-245
+ip-26-0-168-95
+ip-26-0-168-120
+ip-26-0-170-31
+ip-26-0-171-21
+ip-26-0-171-230
+ip-26-0-173-202
+ip-26-0-173-246
+ip-26-0-174-36'
+All nodes: ip-26-0-161-78
+ip-26-0-161-142
+ip-26-0-161-178
+ip-26-0-162-180
+ip-26-0-165-24
+ip-26-0-165-59
+ip-26-0-166-125
+ip-26-0-167-245
+ip-26-0-168-95
+ip-26-0-168-120
+ip-26-0-170-31
+ip-26-0-171-21
+ip-26-0-171-230
+ip-26-0-173-202
+ip-26-0-173-246
+ip-26-0-174-36
++ echo 'World size: 128'
+World size: 128
++ srun torchrun --nnodes=16 --nproc_per_node=8 --rdzv_id=12344322 --rdzv_backend=c10d --rdzv_endpoint=ip-26-0-161-78:12356 run_train.py --config-file benchmark/configs/config_3.56G_dp16_tp8_pp1_acc1_mbs16_seq2048_zero0_l28_h3072_heads24.yaml
+[2024-12-03 08:58:10,311] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-03 08:58:10,312] torch.distributed.run: [WARNING] 
+[2024-12-03 08:58:10,312] torch.distributed.run: [WARNING] *****************************************
+[2024-12-03 08:58:10,312] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-03 08:58:10,312] torch.distributed.run: [WARNING] *****************************************
+[2024-12-03 08:58:10,313] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-03 08:58:10,313] torch.distributed.run: [WARNING] 
+[2024-12-03 08:58:10,313] torch.distributed.run: [WARNING] *****************************************
+[2024-12-03 08:58:10,313] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-03 08:58:10,313] torch.distributed.run: [WARNING] *****************************************
+[2024-12-03 08:58:10,314] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-03 08:58:10,314] torch.distributed.run: [WARNING] 
+[2024-12-03 08:58:10,314] torch.distributed.run: [WARNING] *****************************************
+[2024-12-03 08:58:10,314] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-03 08:58:10,314] torch.distributed.run: [WARNING] *****************************************
+[2024-12-03 08:58:10,314] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-03 08:58:10,314] torch.distributed.run: [WARNING] 
+[2024-12-03 08:58:10,314] torch.distributed.run: [WARNING] *****************************************
+[2024-12-03 08:58:10,314] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-03 08:58:10,314] torch.distributed.run: [WARNING] *****************************************
+[2024-12-03 08:58:10,314] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-03 08:58:10,314] torch.distributed.run: [WARNING] 
+[2024-12-03 08:58:10,314] torch.distributed.run: [WARNING] *****************************************
+[2024-12-03 08:58:10,314] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-03 08:58:10,314] torch.distributed.run: [WARNING] *****************************************
+[2024-12-03 08:58:10,317] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-03 08:58:10,318] torch.distributed.run: [WARNING] 
+[2024-12-03 08:58:10,318] torch.distributed.run: [WARNING] *****************************************
+[2024-12-03 08:58:10,318] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-03 08:58:10,318] torch.distributed.run: [WARNING] *****************************************
+[2024-12-03 08:58:10,318] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-03 08:58:10,318] torch.distributed.run: [WARNING] 
+[2024-12-03 08:58:10,318] torch.distributed.run: [WARNING] *****************************************
+[2024-12-03 08:58:10,318] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-03 08:58:10,318] torch.distributed.run: [WARNING] *****************************************
+[2024-12-03 08:58:10,319] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-03 08:58:10,319] torch.distributed.run: [WARNING] 
+[2024-12-03 08:58:10,319] torch.distributed.run: [WARNING] *****************************************
+[2024-12-03 08:58:10,319] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-03 08:58:10,319] torch.distributed.run: [WARNING] *****************************************
+[2024-12-03 08:58:10,322] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-03 08:58:10,322] torch.distributed.run: [WARNING] 
+[2024-12-03 08:58:10,322] torch.distributed.run: [WARNING] *****************************************
+[2024-12-03 08:58:10,322] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-03 08:58:10,322] torch.distributed.run: [WARNING] *****************************************
+[2024-12-03 08:58:10,319] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-03 08:58:10,319] torch.distributed.run: [WARNING] 
+[2024-12-03 08:58:10,319] torch.distributed.run: [WARNING] *****************************************
+[2024-12-03 08:58:10,319] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-03 08:58:10,319] torch.distributed.run: [WARNING] *****************************************
+[2024-12-03 08:58:10,340] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-03 08:58:10,340] torch.distributed.run: [WARNING] 
+[2024-12-03 08:58:10,340] torch.distributed.run: [WARNING] *****************************************
+[2024-12-03 08:58:10,340] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-03 08:58:10,340] torch.distributed.run: [WARNING] *****************************************
+[2024-12-03 08:58:10,343] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-03 08:58:10,343] torch.distributed.run: [WARNING] 
+[2024-12-03 08:58:10,343] torch.distributed.run: [WARNING] *****************************************
+[2024-12-03 08:58:10,343] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-03 08:58:10,343] torch.distributed.run: [WARNING] *****************************************
+[2024-12-03 08:58:10,345] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-03 08:58:10,345] torch.distributed.run: [WARNING] 
+[2024-12-03 08:58:10,345] torch.distributed.run: [WARNING] *****************************************
+[2024-12-03 08:58:10,345] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-03 08:58:10,345] torch.distributed.run: [WARNING] *****************************************
+[2024-12-03 08:58:10,357] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-03 08:58:10,357] torch.distributed.run: [WARNING] 
+[2024-12-03 08:58:10,357] torch.distributed.run: [WARNING] *****************************************
+[2024-12-03 08:58:10,357] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-03 08:58:10,357] torch.distributed.run: [WARNING] *****************************************
+[2024-12-03 08:58:10,370] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-03 08:58:10,371] torch.distributed.run: [WARNING] 
+[2024-12-03 08:58:10,371] torch.distributed.run: [WARNING] *****************************************
+[2024-12-03 08:58:10,371] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-03 08:58:10,371] torch.distributed.run: [WARNING] *****************************************
+[2024-12-03 08:58:10,447] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-03 08:58:10,447] torch.distributed.run: [WARNING] 
+[2024-12-03 08:58:10,447] torch.distributed.run: [WARNING] *****************************************
+[2024-12-03 08:58:10,447] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-03 08:58:10,447] torch.distributed.run: [WARNING] *****************************************
+12/03/2024 08:58:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-142]: Config:
+12/03/2024 08:58:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-142]: Config(general=GeneralArgs(project='debug',
+12/03/2024 08:58:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-142]:                            run='3.56G_dp16_tp8_pp1_acc1_mbs16_seq2048_zero0_l28_h3072_heads24',
+12/03/2024 08:58:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-142]:                            seed=42,
+12/03/2024 08:58:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-142]:                            step=None,
+12/03/2024 08:58:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-142]:                            consumed_train_samples=None,
+12/03/2024 08:58:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-142]:                            benchmark_csv_path=PosixPath('bench_elie.csv'),
+12/03/2024 08:58:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-142]:                            ignore_sanity_checks=True),
+12/03/2024 08:58:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-142]:        parallelism=ParallelismArgs(dp=16,
+12/03/2024 08:58:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-142]:                                    pp=1,
+12/03/2024 08:58:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-142]:                                    tp=8,
+12/03/2024 08:58:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-142]:                                    pp_engine=<nanotron.parallel.pipeline_parallel.engine.OneForwardOneBackwardPipelineEngine object at 0x7f8256244df0>,
+12/03/2024 08:58:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-142]:                                    tp_mode=<TensorParallelLinearMode.REDUCE_SCATTER: 2>,
+12/03/2024 08:58:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-142]:                                    tp_linear_async_communication=True,
+12/03/2024 08:58:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-142]:                                    recompute_layer=False,
+12/03/2024 08:58:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-142]:                                    tp_recompute_allgather=True,
+12/03/2024 08:58:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-142]:                                    expert_parallel_size=1),
+12/03/2024 08:58:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-142]:        model=ModelArgs(model_config=LlamaConfig(bos_token_id=0,
+12/03/2024 08:58:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-142]:                                                 eos_token_id=0,
+12/03/2024 08:58:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-142]:                                                 hidden_act='silu',
+12/03/2024 08:58:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-142]:                                                 hidden_size=3072,
+12/03/2024 08:58:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-142]:                                                 initializer_range=0.02,
+12/03/2024 08:58:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-142]:                                                 intermediate_size=8192,
+12/03/2024 08:58:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-142]:                                                 is_llama_config=True,
+12/03/2024 08:58:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-142]:                                                 max_position_embeddings=2048,
+12/03/2024 08:58:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-142]:                                                 num_attention_heads=24,
+12/03/2024 08:58:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-142]:                                                 num_hidden_layers=28,
+12/03/2024 08:58:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-142]:                                                 num_key_value_heads=8,
+12/03/2024 08:58:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-142]:                                                 pad_token_id=None,
+12/03/2024 08:58:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-142]:                                                 pretraining_tp=1,
+12/03/2024 08:58:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-142]:                                                 rms_norm_eps=1e-05,
+12/03/2024 08:58:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-142]:                                                 rope_scaling=None,
+12/03/2024 08:58:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-142]:                                                 rope_theta=10000.0,
+12/03/2024 08:58:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-142]:                                                 rope_interleaved=False,
+12/03/2024 08:58:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-142]:                                                 tie_word_embeddings=True,
+12/03/2024 08:58:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-142]:                                                 use_cache=True,
+12/03/2024 08:58:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-142]:                                                 vocab_size=128256),
+12/03/2024 08:58:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-142]:                        init_method=RandomInit(std=0.02),
+12/03/2024 08:58:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-142]:                        dtype=torch.bfloat16,
+12/03/2024 08:58:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-142]:                        make_vocab_size_divisible_by=1,
+12/03/2024 08:58:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-142]:                        ddp_bucket_cap_mb=25),
+12/03/2024 08:58:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-142]:        tokenizer=TokenizerArgs(tokenizer_name_or_path='robot-test/dummy-tokenizer-wordlevel',
+12/03/2024 08:58:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-142]:                                tokenizer_revision=None,
+12/03/2024 08:58:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-142]:                                tokenizer_max_length=None),
+12/03/2024 08:58:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-142]:        checkpoints=CheckpointsArgs(checkpoints_path=PosixPath('checkpoints'),
+12/03/2024 08:58:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-142]:                                    checkpoint_interval=10000,
+12/03/2024 08:58:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-142]:                                    save_initial_state=False,
+12/03/2024 08:58:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-142]:                                    save_final_state=False,
+12/03/2024 08:58:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-142]:                                    resume_checkpoint_path=None,
+12/03/2024 08:58:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-142]:                                    checkpoints_path_is_shared_file_system=False),
+12/03/2024 08:58:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-142]:        logging=LoggingArgs(log_level='info',
+12/03/2024 08:58:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-142]:                            log_level_replica='info',
+12/03/2024 08:58:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-142]:                            iteration_step_info_interval=1),
+12/03/2024 08:58:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-142]:        tokens=TokensArgs(sequence_length=2048,
+12/03/2024 08:58:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-142]:                          train_steps=100,
+12/03/2024 08:58:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-142]:                          micro_batch_size=16,
+12/03/2024 08:58:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-142]:                          batch_accumulation_per_replica=1,
+12/03/2024 08:58:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-142]:                          val_check_interval=100,
+12/03/2024 08:58:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-142]:                          limit_val_batches=0,
+12/03/2024 08:58:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-142]:                          limit_test_batches=0),
+12/03/2024 08:58:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-142]:        optimizer=OptimizerArgs(optimizer_factory=AdamWOptimizerArgs(adam_eps=1e-08,
+12/03/2024 08:58:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-142]:                                                                     adam_beta1=0.9,
+12/03/2024 08:58:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-142]:                                                                     adam_beta2=0.95,
+12/03/2024 08:58:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-142]:                                                                     torch_adam_is_fused=True,
+12/03/2024 08:58:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-142]:                                                                     name='adamW'),
+12/03/2024 08:58:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-142]:                                zero_stage=0,
+12/03/2024 08:58:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-142]:                                weight_decay=0.01,
+12/03/2024 08:58:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-142]:                                clip_grad=1.0,
+12/03/2024 08:58:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-142]:                                accumulate_grad_in_fp32=True,
+12/03/2024 08:58:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-142]:                                learning_rate_scheduler=LRSchedulerArgs(learning_rate=0.0003,
+12/03/2024 08:58:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-142]:                                                                        lr_warmup_steps=2,
+12/03/2024 08:58:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-142]:                                                                        lr_warmup_style='linear',
+12/03/2024 08:58:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-142]:                                                                        lr_decay_style='cosine',
+12/03/2024 08:58:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-142]:                                                                        lr_decay_steps=13,
+12/03/2024 08:58:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-142]:                                                                        lr_decay_starting_step=None,
+12/03/2024 08:58:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-142]:                                                                        min_decay_lr=1e-05)),
+12/03/2024 08:58:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-142]:        data_stages=[DatasetStageArgs(name='Stable Training Stage',
+12/03/2024 08:58:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-142]:                                      start_training_step=1,
+12/03/2024 08:58:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-142]:                                      data=DataArgs(dataset=None,
+12/03/2024 08:58:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-142]:                                                    seed=42,
+12/03/2024 08:58:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-142]:                                                    num_loading_workers=1))],
+12/03/2024 08:58:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-142]:        profiler=None,
+12/03/2024 08:58:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-142]:        lighteval=None,
+12/03/2024 08:58:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-142]:        s3_upload=None)
+12/03/2024 08:58:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-142]: Model Config:
+12/03/2024 08:58:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-142]: LlamaConfig(bos_token_id=0,
+12/03/2024 08:58:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-142]:             eos_token_id=0,
+12/03/2024 08:58:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-142]:             hidden_act='silu',
+12/03/2024 08:58:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-142]:             hidden_size=3072,
+12/03/2024 08:58:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-142]:             initializer_range=0.02,
+12/03/2024 08:58:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-142]:             intermediate_size=8192,
+12/03/2024 08:58:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-142]:             is_llama_config=True,
+12/03/2024 08:58:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-142]:             max_position_embeddings=2048,
+12/03/2024 08:58:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-142]:             num_attention_heads=24,
+12/03/2024 08:58:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-142]:             num_hidden_layers=28,
+12/03/2024 08:58:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-142]:             num_key_value_heads=8,
+12/03/2024 08:58:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-142]:             pad_token_id=None,
+12/03/2024 08:58:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-142]:             pretraining_tp=1,
+12/03/2024 08:58:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-142]:             rms_norm_eps=1e-05,
+12/03/2024 08:58:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-142]:             rope_scaling=None,
+12/03/2024 08:58:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-142]:             rope_theta=10000.0,
+12/03/2024 08:58:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-142]:             rope_interleaved=False,
+12/03/2024 08:58:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-142]:             tie_word_embeddings=True,
+12/03/2024 08:58:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-142]:             use_cache=True,
+12/03/2024 08:58:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-142]:             vocab_size=128256)
+12/03/2024 08:58:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-142]: Building model..
+12/03/2024 08:58:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-142]: Initialize RoPE Theta = 10000.0
+12/03/2024 08:58:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-142]: Setting PP block ranks...
+12/03/2024 08:58:52 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-142]: Total number of parameters: 3.21G (6130.17MiB)
+12/03/2024 08:58:52 [INFO|DP=0|PP=0|TP=5|ip-26-0-161-142]: Local number of parameters: 402M (766.27MiB)
+12/03/2024 08:58:52 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-142]: Local number of parameters: 402M (766.27MiB)
+12/03/2024 08:58:52 [INFO|DP=0|PP=0|TP=4|ip-26-0-161-142]: Local number of parameters: 402M (766.27MiB)
+12/03/2024 08:58:52 [INFO|DP=0|PP=0|TP=7|ip-26-0-161-142]: Local number of parameters: 402M (766.27MiB)
+12/03/2024 08:58:52 [INFO|DP=0|PP=0|TP=2|ip-26-0-161-142]: Local number of parameters: 402M (766.27MiB)
+12/03/2024 08:58:52 [INFO|DP=0|PP=0|TP=6|ip-26-0-161-142]: Local number of parameters: 402M (766.27MiB)
+12/03/2024 08:58:52 [INFO|DP=0|PP=0|TP=1|ip-26-0-161-142]: Local number of parameters: 402M (766.27MiB)
+12/03/2024 08:58:52 [INFO|DP=0|PP=0|TP=3|ip-26-0-161-142]: Local number of parameters: 402M (766.27MiB)
+12/03/2024 08:58:52 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-142]: [After model building] Memory usage: 766.62MiB. Peak allocated: 5392.00MiB Peak reserved: 29974.00MiB
+12/03/2024 08:58:52 [INFO|DP=0|PP=0|TP=4|ip-26-0-161-142]: [After model building] Memory usage: 766.62MiB. Peak allocated: 5392.00MiB Peak reserved: 29974.00MiB
+12/03/2024 08:58:52 [INFO|DP=0|PP=0|TP=7|ip-26-0-161-142]: [After model building] Memory usage: 766.62MiB. Peak allocated: 5392.00MiB Peak reserved: 29974.00MiB
+12/03/2024 08:58:52 [INFO|DP=0|PP=0|TP=5|ip-26-0-161-142]: [After model building] Memory usage: 766.62MiB. Peak allocated: 5392.00MiB Peak reserved: 29974.00MiB
+12/03/2024 08:58:52 [INFO|DP=0|PP=0|TP=2|ip-26-0-161-142]: [After model building] Memory usage: 766.62MiB. Peak allocated: 5392.00MiB Peak reserved: 29974.00MiB
+12/03/2024 08:58:52 [INFO|DP=0|PP=0|TP=6|ip-26-0-161-142]: [After model building] Memory usage: 766.62MiB. Peak allocated: 5392.00MiB Peak reserved: 29974.00MiB
+12/03/2024 08:58:52 [INFO|DP=0|PP=0|TP=1|ip-26-0-161-142]: [After model building] Memory usage: 766.62MiB. Peak allocated: 5392.00MiB Peak reserved: 29974.00MiB
+12/03/2024 08:58:52 [INFO|DP=0|PP=0|TP=3|ip-26-0-161-142]: [After model building] Memory usage: 766.62MiB. Peak allocated: 5392.00MiB Peak reserved: 29974.00MiB
+12/03/2024 08:58:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-142]: No checkpoint path provided.
+12/03/2024 08:58:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-142]: Parametrizing model parameters using StandardParametrizator
+12/03/2024 08:58:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-142]: [Optimizer Building] Using LearningRateForSP as learning rate
+12/03/2024 08:58:54 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-142]: [Training Plan] Stage Stable Training Stage has 99 remaining training steps and has consumed 0 samples
+12/03/2024 08:58:54 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-142]: Using dummy data generator
+12/03/2024 08:58:54 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-142]: [Training Plan] There are 1 training stages 
+12/03/2024 08:58:54 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-142]: [Stage Stable Training Stage] start from step 1 
+12/03/2024 08:58:54 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-142]: 
+12/03/2024 08:58:54 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-142]: [Start training] datetime: 2024-12-03 08:58:54.322590 | mbs: 16 | grad_accum: 1 | global_batch_size: 256 | sequence_length: 2048 | train_steps: 100 | start_iteration_step: 0 | consumed_train_samples: 0
+wandb: Tracking run with wandb version 0.16.0
+wandb: W&B syncing is set to `offline` in this directory.  
+wandb: Run `wandb online` or set WANDB_MODE=online to enable cloud syncing.
+12/03/2024 08:59:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-142]: Resuming training from stage Stable Training Stage, it has trained for 0 samples and has 99 remaining train steps
+12/03/2024 08:59:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-142]:  Memory usage: 4597.98MiB. Peak allocated 5392.00MiB. Peak reserved: 33042.00MiB
+12/03/2024 08:59:04 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-142]:  Memory usage: 5060.54MiB. Peak allocated 20553.66MiB. Peak reserved: 23178.00MiB
+12/03/2024 08:59:04 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-142]: iteration: 1 / 100 | consumed_tokens: 524K | elapsed_time_per_iteration_ms: 2.55K | tokens_per_sec: 206K | tokens_per_sec_per_gpu: 1.61K | global_batch_size: 256 | lm_loss: 12.4 | lr: 0.00015 | model_tflops_per_gpu: 34.4 | hardware_tflops_per_gpu: 34.4 | grad_norm: 1.66 | cuda_memory_allocated: 8.52G | cuda_max_memory_reserved: 24.3G | hd_total_memory_tb: 312G | hd_used_memory_tb: 68.9G | hd_free_memory_tb: 243G
+12/03/2024 08:59:04 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-142]:  Memory usage: 8126.83MiB. Peak allocated 8126.83MiB. Peak reserved: 23178.00MiB
+12/03/2024 08:59:04 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-142]:  Memory usage: 8126.83MiB. Peak allocated 23843.95MiB. Peak reserved: 29916.00MiB
+12/03/2024 08:59:04 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-142]: iteration: 2 / 100 | consumed_tokens: 1.05M | elapsed_time_per_iteration_ms: 368 | tokens_per_sec: 1.42M | tokens_per_sec_per_gpu: 11.1K | global_batch_size: 256 | lm_loss: 12.4 | lr: 0.0003 | model_tflops_per_gpu: 238 | hardware_tflops_per_gpu: 238 | grad_norm: 1.66 | cuda_memory_allocated: 8.52G | cuda_max_memory_reserved: 31.4G | hd_total_memory_tb: 312G | hd_used_memory_tb: 68.9G | hd_free_memory_tb: 243G
+12/03/2024 08:59:04 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-142]:  Memory usage: 8126.83MiB. Peak allocated 8126.92MiB. Peak reserved: 29916.00MiB
+12/03/2024 08:59:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-142]:  Memory usage: 8126.83MiB. Peak allocated 23843.95MiB. Peak reserved: 31920.00MiB
+num_paramsnum_params
+
+{'total': 3213975552, 'local': 401746944}
+{'total': 3213975552, 'local': 401746944}
+num_params
+{'total': 3213975552, 'local': 401746944}num_params
+num_params
+num_params
+{'total': 3213975552, 'local': 401746944}
+{'total': 3213975552, 'local': 401746944}
+
+{'total': 3213975552, 'local': 401746944}
+num_paramsnum_paramsnum_paramsnum_paramsnum_params
+
+
+
+
+{'total': 3213975552, 'local': 401746944}{'total': 3213975552, 'local': 401746944}{'total': 3213975552, 'local': 401746944}
+{'total': 3213975552, 'local': 401746944}{'total': 3213975552, 'local': 401746944}
+num_params
+
+
+num_params
+{'total': 3213975552, 'local': 401746944}
+
+num_params
+{'total': 3213975552, 'local': 401746944}
+num_params
+num_params{'total': 3213975552, 'local': 401746944}
+
+{'total': 3213975552, 'local': 401746944}
+num_params
+{'total': 3213975552, 'local': 401746944}
+num_paramsnum_paramsnum_params
+num_params
+{'total': 3213975552, 'local': 401746944}
+
+{'total': 3213975552, 'local': 401746944}
+
+num_params{'total': 3213975552, 'local': 401746944}{'total': 3213975552, 'local': 401746944}
+
+num_params
+{'total': 3213975552, 'local': 401746944}
+
+num_params
+{'total': 3213975552, 'local': 401746944}
+num_paramsnum_params
+num_params
+{'total': 3213975552, 'local': 401746944}
+num_paramsnum_paramsnum_params
+
+num_params
+num_params{'total': 3213975552, 'local': 401746944}{'total': 3213975552, 'local': 401746944}
+
+{'total': 3213975552, 'local': 401746944}
+
+
+{'total': 3213975552, 'local': 401746944}
+num_params{'total': 3213975552, 'local': 401746944}
+
+num_params{'total': 3213975552, 'local': 401746944}
+num_paramsnum_paramsnum_params
+
+{'total': 3213975552, 'local': 401746944}
+
+{'total': 3213975552, 'local': 401746944}
+{'total': 3213975552, 'local': 401746944}num_params
+
+num_params{'total': 3213975552, 'local': 401746944}
+
+num_paramsnum_params
+
+{'total': 3213975552, 'local': 401746944}num_params
+{'total': 3213975552, 'local': 401746944}
+
+{'total': 3213975552, 'local': 401746944}num_params
+
+num_params{'total': 3213975552, 'local': 401746944}
+num_paramsnum_params
+num_paramsnum_params
+num_params
+{'total': 3213975552, 'local': 401746944}
+
+num_params{'total': 3213975552, 'local': 401746944}
+{'total': 3213975552, 'local': 401746944}num_paramsnum_params
+num_paramsnum_params
+num_params
+num_params{'total': 3213975552, 'local': 401746944}
+
+{'total': 3213975552, 'local': 401746944}
+
+{'total': 3213975552, 'local': 401746944}
+{'total': 3213975552, 'local': 401746944}
+num_params
+num_params{'total': 3213975552, 'local': 401746944}
+num_paramsnum_paramsnum_params
+num_params
+
+{'total': 3213975552, 'local': 401746944}
+
+{'total': 3213975552, 'local': 401746944}
+num_paramsnum_params
+
+{'total': 3213975552, 'local': 401746944}{'total': 3213975552, 'local': 401746944}
+
+num_params
+{'total': 3213975552, 'local': 401746944}
+num_params
+{'total': 3213975552, 'local': 401746944}
+num_paramsnum_params
+
+num_paramsnum_params{'total': 3213975552, 'local': 401746944}{'total': 3213975552, 'local': 401746944}
+
+
+
+{'total': 3213975552, 'local': 401746944}
+{'total': 3213975552, 'local': 401746944}
+num_paramsnum_params
+num_params
+num_params{'total': 3213975552, 'local': 401746944}
+
+{'total': 3213975552, 'local': 401746944}
+num_params
+{'total': 3213975552, 'local': 401746944}
+num_params
+{'total': 3213975552, 'local': 401746944}
+
+{'total': 3213975552, 'local': 401746944}num_params
+
+{'total': 3213975552, 'local': 401746944}
+num_params{'total': 3213975552, 'local': 401746944}
+num_params
+
+{'total': 3213975552, 'local': 401746944}
+{'total': 3213975552, 'local': 401746944}
+
+{'total': 3213975552, 'local': 401746944}
+
+{'total': 3213975552, 'local': 401746944}
+{'total': 3213975552, 'local': 401746944}
+
+
+num_params
+{'total': 3213975552, 'local': 401746944}{'total': 3213975552, 'local': 401746944}
+
+{'total': 3213975552, 'local': 401746944}
+
+{'total': 3213975552, 'local': 401746944}
+num_paramsnum_paramsnum_params
+
+num_params{'total': 3213975552, 'local': 401746944}{'total': 3213975552, 'local': 401746944}
+
+
+
+{'total': 3213975552, 'local': 401746944}num_params
+{'total': 3213975552, 'local': 401746944}
+
+{'total': 3213975552, 'local': 401746944}
+num_paramsnum_params
+
+{'total': 3213975552, 'local': 401746944}num_params
+num_params{'total': 3213975552, 'local': 401746944}
+{'total': 3213975552, 'local': 401746944}{'total': 3213975552, 'local': 401746944}
+
+
+
+num_params{'total': 3213975552, 'local': 401746944}
+
+{'total': 3213975552, 'local': 401746944}
+{'total': 3213975552, 'local': 401746944}
+num_paramsnum_params
+
+{'total': 3213975552, 'local': 401746944}
+{'total': 3213975552, 'local': 401746944}
+num_params
+{'total': 3213975552, 'local': 401746944}
+num_params
+{'total': 3213975552, 'local': 401746944}
+num_params{'total': 3213975552, 'local': 401746944}
+
+
+num_params{'total': 3213975552, 'local': 401746944}{'total': 3213975552, 'local': 401746944}
+
+
+{'total': 3213975552, 'local': 401746944}
+num_paramsnum_paramsnum_params
+
+
+{'total': 3213975552, 'local': 401746944}
+{'total': 3213975552, 'local': 401746944}{'total': 3213975552, 'local': 401746944}
+
+num_params{'total': 3213975552, 'local': 401746944}
+
+
+{'total': 3213975552, 'local': 401746944}num_params
+{'total': 3213975552, 'local': 401746944}num_params
+
+
+{'total': 3213975552, 'local': 401746944}
+num_paramsnum_paramsnum_params
+num_params
+num_params
+{'total': 3213975552, 'local': 401746944}{'total': 3213975552, 'local': 401746944}
+
+
+{'total': 3213975552, 'local': 401746944}{'total': 3213975552, 'local': 401746944}
+
+
+{'total': 3213975552, 'local': 401746944}
+num_params
+{'total': 3213975552, 'local': 401746944}
+num_params
+{'total': 3213975552, 'local': 401746944}
+num_params
+{'total': 3213975552, 'local': 401746944}
+num_params
+{'total': 3213975552, 'local': 401746944}
+num_params
+{'total': 3213975552, 'local': 401746944}
+num_params
+{'total': 3213975552, 'local': 401746944}
+
+{'total': 3213975552, 'local': 401746944}
+num_params
+{'total': 3213975552, 'local': 401746944}
+num_paramsnum_paramsnum_params
+
+num_params
+{'total': 3213975552, 'local': 401746944}num_paramsnum_params{'total': 3213975552, 'local': 401746944}{'total': 3213975552, 'local': 401746944}
+
+
+
+
+
+{'total': 3213975552, 'local': 401746944}
+{'total': 3213975552, 'local': 401746944}{'total': 3213975552, 'local': 401746944}
+
+num_params
+{'total': 3213975552, 'local': 401746944}
+num_params
+{'total': 3213975552, 'local': 401746944}
+{'total': 3213975552, 'local': 401746944}
+num_params
+{'total': 3213975552, 'local': 401746944}
+num_params
+num_params{'total': 3213975552, 'local': 401746944}
+
+{'total': 3213975552, 'local': 401746944}
+{'total': 3213975552, 'local': 401746944}
+num_params
+{'total': 3213975552, 'local': 401746944}
+{'total': 3213975552, 'local': 401746944}
+num_params
+{'total': 3213975552, 'local': 401746944}
+12/03/2024 08:59:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-142]: iteration: 3 / 100 | consumed_tokens: 1.57M | elapsed_time_per_iteration_ms: 368 | tokens_per_sec: 1.42M | tokens_per_sec_per_gpu: 11.1K | global_batch_size: 256 | lm_loss: 12.4 | lr: 0.000296 | model_tflops_per_gpu: 238 | hardware_tflops_per_gpu: 238 | grad_norm: 1.6 | cuda_memory_allocated: 8.52G | cuda_max_memory_reserved: 33.5G | hd_total_memory_tb: 312G | hd_used_memory_tb: 68.9G | hd_free_memory_tb: 243G
+num_params
+{'total': 3213975552, 'local': 401746944}
+12/03/2024 08:59:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-142]: | job_id   | name                                                          | nodes | seq_len | mbs | batch_accum | gbs | mTFLOPs | hTFLOPs | tok/s/gpu | AllReduce (GB/s) | AllGather (GB/s) | ReduceScatter (GB/s) | AR Intra-node (GB/s) | AG Intra-node (GB/s) | RS Intra-node (GB/s) | Mem Alloc (GB) | Mem Res (GB) | dp | pp | tp | pp_engine                                                                                                 | tp_mode                                 | tp_async_comm | hidden_size | hidden_act | num_layers | num_heads | num_kv_heads | max_pos | vocab_size | tie_word_embeddings | dtype          | zero_stage | ddp_bucket_cap_mb | accumulate_grad_in_fp32 | Total Params | Local Params |
+12/03/2024 08:59:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-142]: | -------- | ------------------------------------------------------------- | ----- | ------- | --- | ----------- | --- | ------- | ------- | --------- | ---------------- | ---------------- | -------------------- | -------------------- | -------------------- | -------------------- | -------------- | ------------ | -- | -- | -- | --------------------------------------------------------------------------------------------------------- | --------------------------------------- | ------------- | ----------- | ---------- | ---------- | --------- | ------------ | ------- | ---------- | ------------------- | -------------- | ---------- | ----------------- | ----------------------- | ------------ | ------------ |
+12/03/2024 08:59:05 [INFO|DP=4|PP=0|TP=0|ip-26-0-165-24]: Throughput logging complete
+12/03/2024 08:59:05 [INFO|DP=2|PP=0|TP=2|ip-26-0-161-78]: Throughput logging complete
+12/03/2024 08:59:05 [INFO|DP=8|PP=0|TP=4|ip-26-0-168-120]: Throughput logging complete
+12/03/2024 08:59:05 [INFO|DP=8|PP=0|TP=5|ip-26-0-168-120]: Throughput logging complete
+12/03/2024 08:59:05 [INFO|DP=14|PP=0|TP=6|ip-26-0-173-246]: Throughput logging complete
+12/03/2024 08:59:05 [INFO|DP=2|PP=0|TP=1|ip-26-0-161-78]: Throughput logging complete
+12/03/2024 08:59:05 [INFO|DP=4|PP=0|TP=2|ip-26-0-165-24]: Throughput logging complete
+12/03/2024 08:59:05 [INFO|DP=2|PP=0|TP=4|ip-26-0-161-78]: Throughput logging complete
+12/03/2024 08:59:05 [INFO|DP=2|PP=0|TP=3|ip-26-0-161-78]: Throughput logging complete
+12/03/2024 08:59:05 [INFO|DP=8|PP=0|TP=7|ip-26-0-168-120]: Throughput logging complete
+12/03/2024 08:59:05 [INFO|DP=8|PP=0|TP=0|ip-26-0-168-120]: Throughput logging complete
+12/03/2024 08:59:05 [INFO|DP=6|PP=0|TP=3|ip-26-0-166-125]: Throughput logging complete
+12/03/2024 08:59:05 [INFO|DP=12|PP=0|TP=7|ip-26-0-171-230]: Throughput logging complete
+12/03/2024 08:59:05 [INFO|DP=12|PP=0|TP=4|ip-26-0-171-230]: Throughput logging complete
+12/03/2024 08:59:05 [INFO|DP=13|PP=0|TP=0|ip-26-0-173-202]: Throughput logging complete
+12/03/2024 08:59:05 [INFO|DP=11|PP=0|TP=4|ip-26-0-171-21]: Throughput logging complete
+12/03/2024 08:59:05 [INFO|DP=15|PP=0|TP=1|ip-26-0-174-36]: Throughput logging complete
+12/03/2024 08:59:05 [INFO|DP=15|PP=0|TP=5|ip-26-0-174-36]: Throughput logging complete
+12/03/2024 08:59:05 [INFO|DP=1|PP=0|TP=0|ip-26-0-161-178]: Throughput logging complete
+12/03/2024 08:59:05 [INFO|DP=14|PP=0|TP=0|ip-26-0-173-246]: Throughput logging complete
+12/03/2024 08:59:05 [INFO|DP=14|PP=0|TP=5|ip-26-0-173-246]: Throughput logging complete
+12/03/2024 08:59:05 [INFO|DP=14|PP=0|TP=3|ip-26-0-173-246]: Throughput logging complete
+12/03/2024 08:59:05 [INFO|DP=14|PP=0|TP=4|ip-26-0-173-246]: Throughput logging complete
+12/03/2024 08:59:05 [INFO|DP=4|PP=0|TP=5|ip-26-0-165-24]: Throughput logging complete
+12/03/2024 08:59:05 [INFO|DP=4|PP=0|TP=4|ip-26-0-165-24]: Throughput logging complete
+12/03/2024 08:59:05 [INFO|DP=10|PP=0|TP=3|ip-26-0-170-31]: Throughput logging complete
+12/03/2024 08:59:05 [INFO|DP=10|PP=0|TP=1|ip-26-0-170-31]: Throughput logging complete
+12/03/2024 08:59:05 [INFO|DP=14|PP=0|TP=2|ip-26-0-173-246]: Throughput logging complete
+12/03/2024 08:59:05 [INFO|DP=6|PP=0|TP=1|ip-26-0-166-125]: Throughput logging complete
+12/03/2024 08:59:05 [INFO|DP=12|PP=0|TP=3|ip-26-0-171-230]: Throughput logging complete
+12/03/2024 08:59:05 [INFO|DP=12|PP=0|TP=2|ip-26-0-171-230]: Throughput logging complete
+12/03/2024 08:59:05 [INFO|DP=12|PP=0|TP=0|ip-26-0-171-230]: Throughput logging complete
+12/03/2024 08:59:05 [INFO|DP=12|PP=0|TP=6|ip-26-0-171-230]: Throughput logging complete
+12/03/2024 08:59:05 [INFO|DP=12|PP=0|TP=1|ip-26-0-171-230]: Throughput logging complete
+12/03/2024 08:59:05 [INFO|DP=12|PP=0|TP=5|ip-26-0-171-230]: Throughput logging complete
+12/03/2024 08:59:05 [INFO|DP=7|PP=0|TP=5|ip-26-0-167-245]: Throughput logging complete
+12/03/2024 08:59:05 [INFO|DP=4|PP=0|TP=1|ip-26-0-165-24]: Throughput logging complete
+12/03/2024 08:59:05 [INFO|DP=4|PP=0|TP=3|ip-26-0-165-24]: Throughput logging complete
+12/03/2024 08:59:05 [INFO|DP=11|PP=0|TP=0|ip-26-0-171-21]: Throughput logging complete
+12/03/2024 08:59:05 [INFO|DP=8|PP=0|TP=3|ip-26-0-168-120]: Throughput logging complete
+12/03/2024 08:59:05 [INFO|DP=8|PP=0|TP=6|ip-26-0-168-120]: Throughput logging complete
+12/03/2024 08:59:05 [INFO|DP=15|PP=0|TP=0|ip-26-0-174-36]: Throughput logging complete
+12/03/2024 08:59:05 [INFO|DP=3|PP=0|TP=2|ip-26-0-162-180]: Throughput logging complete
+12/03/2024 08:59:05 [INFO|DP=3|PP=0|TP=3|ip-26-0-162-180]: Throughput logging complete
+12/03/2024 08:59:05 [INFO|DP=1|PP=0|TP=5|ip-26-0-161-178]: Throughput logging complete
+12/03/2024 08:59:05 [INFO|DP=14|PP=0|TP=7|ip-26-0-173-246]: Throughput logging complete
+12/03/2024 08:59:05 [INFO|DP=2|PP=0|TP=0|ip-26-0-161-78]: Throughput logging complete
+12/03/2024 08:59:05 [INFO|DP=2|PP=0|TP=5|ip-26-0-161-78]: Throughput logging complete
+12/03/2024 08:59:05 [INFO|DP=2|PP=0|TP=7|ip-26-0-161-78]: Throughput logging complete
+12/03/2024 08:59:05 [INFO|DP=6|PP=0|TP=6|ip-26-0-166-125]: Throughput logging complete
+12/03/2024 08:59:05 [INFO|DP=6|PP=0|TP=5|ip-26-0-166-125]: Throughput logging complete
+12/03/2024 08:59:05 [INFO|DP=4|PP=0|TP=6|ip-26-0-165-24]: Throughput logging complete
+12/03/2024 08:59:05 [INFO|DP=11|PP=0|TP=7|ip-26-0-171-21]: Throughput logging complete
+12/03/2024 08:59:05 [INFO|DP=11|PP=0|TP=1|ip-26-0-171-21]: Throughput logging complete
+12/03/2024 08:59:05 [INFO|DP=10|PP=0|TP=5|ip-26-0-170-31]: Throughput logging complete
+12/03/2024 08:59:05 [INFO|DP=10|PP=0|TP=4|ip-26-0-170-31]: Throughput logging complete
+12/03/2024 08:59:05 [INFO|DP=10|PP=0|TP=6|ip-26-0-170-31]: Throughput logging complete
+12/03/2024 08:59:05 [INFO|DP=15|PP=0|TP=2|ip-26-0-174-36]: Throughput logging complete
+12/03/2024 08:59:05 [INFO|DP=15|PP=0|TP=3|ip-26-0-174-36]: Throughput logging complete
+12/03/2024 08:59:05 [INFO|DP=3|PP=0|TP=1|ip-26-0-162-180]: Throughput logging complete
+12/03/2024 08:59:05 [INFO|DP=3|PP=0|TP=0|ip-26-0-162-180]: Throughput logging complete
+12/03/2024 08:59:05 [INFO|DP=1|PP=0|TP=1|ip-26-0-161-178]: Throughput logging complete
+12/03/2024 08:59:05 [INFO|DP=9|PP=0|TP=3|ip-26-0-168-95]: Throughput logging complete
+12/03/2024 08:59:05 [INFO|DP=9|PP=0|TP=4|ip-26-0-168-95]: Throughput logging complete
+12/03/2024 08:59:05 [INFO|DP=9|PP=0|TP=2|ip-26-0-168-95]: Throughput logging complete
+12/03/2024 08:59:05 [INFO|DP=14|PP=0|TP=1|ip-26-0-173-246]: Throughput logging complete
+12/03/2024 08:59:05 [INFO|DP=5|PP=0|TP=7|ip-26-0-165-59]: Throughput logging complete
+12/03/2024 08:59:05 [INFO|DP=2|PP=0|TP=6|ip-26-0-161-78]: Throughput logging complete
+12/03/2024 08:59:05 [INFO|DP=6|PP=0|TP=7|ip-26-0-166-125]: Throughput logging complete
+12/03/2024 08:59:05 [INFO|DP=7|PP=0|TP=0|ip-26-0-167-245]: Throughput logging complete
+12/03/2024 08:59:05 [INFO|DP=7|PP=0|TP=1|ip-26-0-167-245]: Throughput logging complete
+12/03/2024 08:59:05 [INFO|DP=13|PP=0|TP=5|ip-26-0-173-202]: Throughput logging complete
+12/03/2024 08:59:05 [INFO|DP=13|PP=0|TP=2|ip-26-0-173-202]: Throughput logging complete
+12/03/2024 08:59:05 [INFO|DP=13|PP=0|TP=4|ip-26-0-173-202]: Throughput logging complete
+12/03/2024 08:59:05 [INFO|DP=13|PP=0|TP=7|ip-26-0-173-202]: Throughput logging complete
+12/03/2024 08:59:05 [INFO|DP=4|PP=0|TP=7|ip-26-0-165-24]: Throughput logging complete
+12/03/2024 08:59:05 [INFO|DP=11|PP=0|TP=3|ip-26-0-171-21]: Throughput logging complete
+12/03/2024 08:59:05 [INFO|DP=11|PP=0|TP=5|ip-26-0-171-21]: Throughput logging complete
+12/03/2024 08:59:05 [INFO|DP=10|PP=0|TP=7|ip-26-0-170-31]: Throughput logging complete
+12/03/2024 08:59:05 [INFO|DP=8|PP=0|TP=1|ip-26-0-168-120]: Throughput logging complete
+12/03/2024 08:59:05 [INFO|DP=8|PP=0|TP=2|ip-26-0-168-120]: Throughput logging complete
+12/03/2024 08:59:05 [INFO|DP=15|PP=0|TP=4|ip-26-0-174-36]: Throughput logging complete
+12/03/2024 08:59:05 [INFO|DP=3|PP=0|TP=4|ip-26-0-162-180]: Throughput logging complete
+12/03/2024 08:59:05 [INFO|DP=3|PP=0|TP=7|ip-26-0-162-180]: Throughput logging complete
+12/03/2024 08:59:05 [INFO|DP=3|PP=0|TP=5|ip-26-0-162-180]: Throughput logging complete
+12/03/2024 08:59:05 [INFO|DP=3|PP=0|TP=6|ip-26-0-162-180]: Throughput logging complete
+12/03/2024 08:59:05 [INFO|DP=1|PP=0|TP=3|ip-26-0-161-178]: Throughput logging complete
+12/03/2024 08:59:05 [INFO|DP=5|PP=0|TP=4|ip-26-0-165-59]: Throughput logging complete
+12/03/2024 08:59:05 [INFO|DP=5|PP=0|TP=3|ip-26-0-165-59]: Throughput logging complete
+12/03/2024 08:59:05 [INFO|DP=5|PP=0|TP=5|ip-26-0-165-59]: Throughput logging complete
+12/03/2024 08:59:05 [INFO|DP=5|PP=0|TP=6|ip-26-0-165-59]: Throughput logging complete
+12/03/2024 08:59:05 [INFO|DP=5|PP=0|TP=0|ip-26-0-165-59]: Throughput logging complete
+12/03/2024 08:59:05 [INFO|DP=5|PP=0|TP=2|ip-26-0-165-59]: Throughput logging complete
+12/03/2024 08:59:05 [INFO|DP=7|PP=0|TP=2|ip-26-0-167-245]: Throughput logging complete
+12/03/2024 08:59:05 [INFO|DP=7|PP=0|TP=3|ip-26-0-167-245]: Throughput logging complete
+12/03/2024 08:59:05 [INFO|DP=7|PP=0|TP=4|ip-26-0-167-245]: Throughput logging complete
+12/03/2024 08:59:05 [INFO|DP=13|PP=0|TP=6|ip-26-0-173-202]: Throughput logging complete
+12/03/2024 08:59:05 [INFO|DP=11|PP=0|TP=6|ip-26-0-171-21]: Throughput logging complete
+12/03/2024 08:59:05 [INFO|DP=15|PP=0|TP=7|ip-26-0-174-36]: Throughput logging complete
+12/03/2024 08:59:05 [INFO|DP=15|PP=0|TP=6|ip-26-0-174-36]: Throughput logging complete
+12/03/2024 08:59:05 [INFO|DP=1|PP=0|TP=4|ip-26-0-161-178]: Throughput logging complete
+12/03/2024 08:59:05 [INFO|DP=9|PP=0|TP=6|ip-26-0-168-95]: Throughput logging complete
+12/03/2024 08:59:05 [INFO|DP=9|PP=0|TP=5|ip-26-0-168-95]: Throughput logging complete
+12/03/2024 08:59:05 [INFO|DP=9|PP=0|TP=7|ip-26-0-168-95]: Throughput logging complete
+12/03/2024 08:59:05 [INFO|DP=5|PP=0|TP=1|ip-26-0-165-59]: Throughput logging complete
+12/03/2024 08:59:05 [INFO|DP=7|PP=0|TP=7|ip-26-0-167-245]: Throughput logging complete
+12/03/2024 08:59:05 [INFO|DP=1|PP=0|TP=6|ip-26-0-161-178]: Throughput logging complete
+12/03/2024 08:59:05 [INFO|DP=7|PP=0|TP=6|ip-26-0-167-245]: Throughput logging complete
+12/03/2024 08:59:05 [INFO|DP=1|PP=0|TP=7|ip-26-0-161-178]: Throughput logging complete
+12/03/2024 08:59:05 [INFO|DP=6|PP=0|TP=4|ip-26-0-166-125]: Throughput logging complete
+12/03/2024 08:59:05 [INFO|DP=1|PP=0|TP=2|ip-26-0-161-178]: Throughput logging complete
+12/03/2024 08:59:05 [INFO|DP=10|PP=0|TP=0|ip-26-0-170-31]: Throughput logging complete
+12/03/2024 08:59:05 [INFO|DP=10|PP=0|TP=2|ip-26-0-170-31]: Throughput logging complete
+12/03/2024 08:59:05 [INFO|DP=6|PP=0|TP=2|ip-26-0-166-125]: Throughput logging complete
+12/03/2024 08:59:05 [INFO|DP=6|PP=0|TP=0|ip-26-0-166-125]: Throughput logging complete
+12/03/2024 08:59:05 [INFO|DP=9|PP=0|TP=0|ip-26-0-168-95]: Throughput logging complete
+12/03/2024 08:59:05 [INFO|DP=9|PP=0|TP=1|ip-26-0-168-95]: Throughput logging complete
+12/03/2024 08:59:05 [INFO|DP=11|PP=0|TP=2|ip-26-0-171-21]: Throughput logging complete
+12/03/2024 08:59:05 [INFO|DP=13|PP=0|TP=3|ip-26-0-173-202]: Throughput logging complete
+12/03/2024 08:59:05 [INFO|DP=13|PP=0|TP=1|ip-26-0-173-202]: Throughput logging complete
+slurmstepd: error: *** STEP 12344322.0 ON ip-26-0-161-78 CANCELLED AT 2024-12-03T08:59:05 ***
+[2024-12-03 08:59:05,371] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-03 08:59:05,372] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 4106766 closing signal SIGTERM
+[2024-12-03 08:59:05,372] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 4106767 closing signal SIGTERM
+[2024-12-03 08:59:05,372] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 4106768 closing signal SIGTERM
+[2024-12-03 08:59:05,372] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-03 08:59:05,372] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 4106769 closing signal SIGTERM
+[2024-12-03 08:59:05,373] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2091686 closing signal SIGTERM
+[2024-12-03 08:59:05,373] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2091687 closing signal SIGTERM
+[2024-12-03 08:59:05,372] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-03 08:59:05,373] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2091688 closing signal SIGTERM
+slurmstepd: error: *** JOB 12344322 ON ip-26-0-161-78 CANCELLED AT 2024-12-03T08:59:05 ***
+[2024-12-03 08:59:05,372] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 21069 closing signal SIGTERM
+[2024-12-03 08:59:05,372] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 21070 closing signal SIGTERM
+[2024-12-03 08:59:05,371] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-03 08:59:05,372] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-03 08:59:05,372] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1197886 closing signal SIGTERM
+[2024-12-03 08:59:05,372] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1197887 closing signal SIGTERM
+[2024-12-03 08:59:05,372] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1818035 closing signal SIGTERM
+[2024-12-03 08:59:05,372] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-03 08:59:05,372] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1197888 closing signal SIGTERM
+[2024-12-03 08:59:05,372] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1197889 closing signal SIGTERM
+[2024-12-03 08:59:05,373] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-03 08:59:05,373] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 451530 closing signal SIGTERM
+[2024-12-03 08:59:05,374] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 451531 closing signal SIGTERM
+[2024-12-03 08:59:05,372] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 365894 closing signal SIGTERM
+[2024-12-03 08:59:05,372] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 365895 closing signal SIGTERM
+[2024-12-03 08:59:05,372] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 365896 closing signal SIGTERM
+[2024-12-03 08:59:05,372] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 365897 closing signal SIGTERM
+[2024-12-03 08:59:05,374] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 451532 closing signal SIGTERM
+[2024-12-03 08:59:05,374] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 451533 closing signal SIGTERM
+srun: Job step aborted: Waiting up to 32 seconds for job step to finish.
+[2024-12-03 08:59:05,372] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 21071 closing signal SIGTERM
+[2024-12-03 08:59:05,372] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 21072 closing signal SIGTERM
+[2024-12-03 08:59:05,372] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1818036 closing signal SIGTERM
+[2024-12-03 08:59:05,372] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 4106770 closing signal SIGTERM
+[2024-12-03 08:59:05,374] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 451534 closing signal SIGTERM
+[2024-12-03 08:59:05,372] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-03 08:59:05,372] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2288858 closing signal SIGTERM
+[2024-12-03 08:59:05,372] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2288859 closing signal SIGTERM
+[2024-12-03 08:59:05,372] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2288860 closing signal SIGTERM
+[2024-12-03 08:59:05,374] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-03 08:59:05,374] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1433052 closing signal SIGTERM
+[2024-12-03 08:59:05,374] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1433053 closing signal SIGTERM
+[2024-12-03 08:59:05,374] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1433054 closing signal SIGTERM
+[2024-12-03 08:59:05,372] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-03 08:59:05,372] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-03 08:59:05,372] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-03 08:59:05,373] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-03 08:59:05,372] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 593321 closing signal SIGTERM
+[2024-12-03 08:59:05,372] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3452494 closing signal SIGTERM
+[2024-12-03 08:59:05,373] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 593322 closing signal SIGTERM
+[2024-12-03 08:59:05,372] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3452495 closing signal SIGTERM
+[2024-12-03 08:59:05,373] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 593323 closing signal SIGTERM
+[2024-12-03 08:59:05,373] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 567631 closing signal SIGTERM
+[2024-12-03 08:59:05,372] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3452496 closing signal SIGTERM
+[2024-12-03 08:59:05,374] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 21073 closing signal SIGTERM
+[2024-12-03 08:59:05,373] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 593324 closing signal SIGTERM
+[2024-12-03 08:59:05,374] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 567632 closing signal SIGTERM
+[2024-12-03 08:59:05,372] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3452497 closing signal SIGTERM
+[2024-12-03 08:59:05,374] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 21074 closing signal SIGTERM
+[2024-12-03 08:59:05,374] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 567633 closing signal SIGTERM
+[2024-12-03 08:59:05,374] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 21075 closing signal SIGTERM
+[2024-12-03 08:59:05,374] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 567634 closing signal SIGTERM
+[2024-12-03 08:59:05,374] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 21076 closing signal SIGTERM
+[2024-12-03 08:59:05,370] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-03 08:59:05,372] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 4059225 closing signal SIGTERM
+[2024-12-03 08:59:05,372] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 4059226 closing signal SIGTERM
+[2024-12-03 08:59:05,373] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-03 08:59:05,373] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1197890 closing signal SIGTERM
+[2024-12-03 08:59:05,374] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 567635 closing signal SIGTERM
+[2024-12-03 08:59:05,372] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 4059227 closing signal SIGTERM
+[2024-12-03 08:59:05,372] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 4059228 closing signal SIGTERM
+[2024-12-03 08:59:05,372] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 4059229 closing signal SIGTERM
+[2024-12-03 08:59:05,371] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1310410 closing signal SIGTERM
+[2024-12-03 08:59:05,373] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 4059230 closing signal SIGTERM
+[2024-12-03 08:59:05,371] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1310411 closing signal SIGTERM
+[2024-12-03 08:59:05,373] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 4059231 closing signal SIGTERM
+12/03/2024 08:59:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-142]: | 12344322 | 3.56G_dp16_tp8_pp1_acc1_mbs16_seq2048_zero0_l28_h3072_heads24 | 16    | 2048    | 16  | 1           | 256 | 237.94  | 237.94  | 11124.44  | 202.90           | 111.04           | 114.97               | 460.20               | 265.10               | 265.42               | 7.94           | 31.17        | 16 | 1  | 8  | <nanotron.parallel.pipeline_parallel.engine.OneForwardOneBackwardPipelineEngine object at 0x7f8256244df0> | TensorParallelLinearMode.REDUCE_SCATTER | True          | 3072        | silu       | 28         | 24        | 8            | 2048    | 128256     | True                | torch.bfloat16 | 0          | 25                | True                    | 3.21G        | 402M         |
+[2024-12-03 08:59:05,373] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 695159 closing signal SIGTERM
+12/03/2024 08:59:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-142]: Throughput logging complete
+[2024-12-03 08:59:05,374] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1197891 closing signal SIGTERM
+12/03/2024 08:59:05 [INFO|DP=0|PP=0|TP=4|ip-26-0-161-142]: Throughput logging complete
+[2024-12-03 08:59:05,371] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1310412 closing signal SIGTERM
+12/03/2024 08:59:05 [INFO|DP=0|PP=0|TP=1|ip-26-0-161-142]: Throughput logging complete
+12/03/2024 08:59:05 [INFO|DP=0|PP=0|TP=5|ip-26-0-161-142]: Throughput logging complete
+12/03/2024 08:59:05 [INFO|DP=0|PP=0|TP=2|ip-26-0-161-142]: Throughput logging complete
+12/03/2024 08:59:05 [INFO|DP=0|PP=0|TP=6|ip-26-0-161-142]: Throughput logging complete
+[2024-12-03 08:59:05,374] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 4059232 closing signal SIGTERM
+12/03/2024 08:59:05 [INFO|DP=0|PP=0|TP=7|ip-26-0-161-142]: Throughput logging complete
+12/03/2024 08:59:05 [INFO|DP=0|PP=0|TP=3|ip-26-0-161-142]: Throughput logging complete
+[2024-12-03 08:59:05,371] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1310413 closing signal SIGTERM
+[2024-12-03 08:59:05,371] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1310414 closing signal SIGTERM
+[2024-12-03 08:59:05,374] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1197892 closing signal SIGTERM
+[2024-12-03 08:59:05,374] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 4106771 closing signal SIGTERM
+[2024-12-03 08:59:05,372] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1818037 closing signal SIGTERM
+[2024-12-03 08:59:05,373] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 365898 closing signal SIGTERM
+[2024-12-03 08:59:05,374] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2091689 closing signal SIGTERM
+[2024-12-03 08:59:05,374] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2288861 closing signal SIGTERM
+[2024-12-03 08:59:05,372] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-03 08:59:05,372] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 345968 closing signal SIGTERM
+[2024-12-03 08:59:05,374] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2288862 closing signal SIGTERM
+[2024-12-03 08:59:05,372] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 345969 closing signal SIGTERM
+[2024-12-03 08:59:05,372] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 345970 closing signal SIGTERM
+[2024-12-03 08:59:05,372] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 345971 closing signal SIGTERM
+[2024-12-03 08:59:05,374] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2091690 closing signal SIGTERM
+[2024-12-03 08:59:05,374] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2091691 closing signal SIGTERM
+[2024-12-03 08:59:05,374] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2091692 closing signal SIGTERM
+[2024-12-03 08:59:05,375] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 4106772 closing signal SIGTERM
+[2024-12-03 08:59:05,375] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2091693 closing signal SIGTERM
+[2024-12-03 08:59:05,375] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 4106773 closing signal SIGTERM
+[2024-12-03 08:59:05,374] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2288863 closing signal SIGTERM
+[2024-12-03 08:59:05,374] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2288864 closing signal SIGTERM
+[2024-12-03 08:59:05,373] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 695160 closing signal SIGTERM
+[2024-12-03 08:59:05,375] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1433055 closing signal SIGTERM
+[2024-12-03 08:59:05,374] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1818038 closing signal SIGTERM
+[2024-12-03 08:59:05,374] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1818039 closing signal SIGTERM
+[2024-12-03 08:59:05,374] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1818040 closing signal SIGTERM
+[2024-12-03 08:59:05,374] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1818041 closing signal SIGTERM
+[2024-12-03 08:59:05,375] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 365899 closing signal SIGTERM
+[2024-12-03 08:59:05,375] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 593325 closing signal SIGTERM
+[2024-12-03 08:59:05,375] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 567636 closing signal SIGTERM
+[2024-12-03 08:59:05,375] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 593326 closing signal SIGTERM
+[2024-12-03 08:59:05,375] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 365900 closing signal SIGTERM
+[2024-12-03 08:59:05,375] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3452498 closing signal SIGTERM
+[2024-12-03 08:59:05,375] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 365901 closing signal SIGTERM
+[2024-12-03 08:59:05,375] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 593327 closing signal SIGTERM
+[2024-12-03 08:59:05,375] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3452499 closing signal SIGTERM
+[2024-12-03 08:59:05,375] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3452500 closing signal SIGTERM
+[2024-12-03 08:59:05,377] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 451535 closing signal SIGTERM
+[2024-12-03 08:59:05,377] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 451536 closing signal SIGTERM
+[2024-12-03 08:59:05,375] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1818042 closing signal SIGTERM
+[2024-12-03 08:59:05,375] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 593328 closing signal SIGTERM
+[2024-12-03 08:59:05,374] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2288865 closing signal SIGTERM
+[2024-12-03 08:59:05,375] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1433056 closing signal SIGTERM
+[2024-12-03 08:59:05,375] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1433057 closing signal SIGTERM
+[2024-12-03 08:59:05,377] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1433058 closing signal SIGTERM
+[2024-12-03 08:59:05,377] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1433059 closing signal SIGTERM
+[2024-12-03 08:59:05,375] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 567637 closing signal SIGTERM
+[2024-12-03 08:59:05,375] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 567638 closing signal SIGTERM
+[2024-12-03 08:59:05,376] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1197893 closing signal SIGTERM
+[2024-12-03 08:59:05,376] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3452501 closing signal SIGTERM
+[2024-12-03 08:59:05,374] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1310415 closing signal SIGTERM
+[2024-12-03 08:59:05,377] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 451537 closing signal SIGTERM
+[2024-12-03 08:59:05,374] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1310416 closing signal SIGTERM
+[2024-12-03 08:59:05,374] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1310417 closing signal SIGTERM
+[2024-12-03 08:59:05,373] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 695161 closing signal SIGTERM
+[2024-12-03 08:59:05,373] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 695162 closing signal SIGTERM
+[2024-12-03 08:59:05,376] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 695163 closing signal SIGTERM
+[2024-12-03 08:59:05,376] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 695164 closing signal SIGTERM
+[2024-12-03 08:59:05,376] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 695165 closing signal SIGTERM
+[2024-12-03 08:59:05,376] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 695166 closing signal SIGTERM
+[2024-12-03 08:59:05,374] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 345972 closing signal SIGTERM
+[2024-12-03 08:59:05,374] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 345973 closing signal SIGTERM
+[2024-12-03 08:59:05,374] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 345974 closing signal SIGTERM
+[2024-12-03 08:59:05,374] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 345975 closing signal SIGTERM
diff --git a/logs/12344441-bench_3.56G_dp64_tp1_pp1_acc1_mbs2_seq2048_zero1_l28_h3072_heads24.out b/logs/12344441-bench_3.56G_dp64_tp1_pp1_acc1_mbs2_seq2048_zero1_l28_h3072_heads24.out
new file mode 100644
index 0000000000000000000000000000000000000000..a9f7d287d672166637fd939ef628bdbc58e46a04
--- /dev/null
+++ b/logs/12344441-bench_3.56G_dp64_tp1_pp1_acc1_mbs2_seq2048_zero1_l28_h3072_heads24.out
@@ -0,0 +1,944 @@
++ source /etc/profile.d/modules.sh
+++ . /usr/share/modules/init/bash
++++ unset _mlshdbg
++++ '[' 0 = 1 ']'
++++ unset _mlre _mlIFS
++++ '[' -n x ']'
++++ _mlIFS=' 	
+'
++++ IFS=' '
++++ '[' -n '' ']'
+++++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash autoinit
++++ _mlcode='module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+MODULEPATH_modshare=/etc/environment-modules/modules:1:/usr/share/modules/\$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1; export MODULEPATH_modshare;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+LOADEDMODULES=; export LOADEDMODULES;
+MODULEPATH=/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/\$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles; export MODULEPATH;
+test 0;'
++++ _mlret=0
++++ '[' -n x ']'
++++ IFS=' 	
+'
++++ unset _mlIFS
++++ unset _mlre _mlv _mlrv
++++ '[' 0 -eq 0 ']'
++++ eval 'module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+MODULEPATH_modshare=/etc/environment-modules/modules:1:/usr/share/modules/\$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1; export MODULEPATH_modshare;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+LOADEDMODULES=; export LOADEDMODULES;
+MODULEPATH=/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/\$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles; export MODULEPATH;
+test 0;'
+++++ MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl
+++++ export MODULES_CMD
+++++ ENV=/usr/share/modules/init/profile.sh
+++++ export ENV
+++++ MODULEPATH_modshare='/etc/environment-modules/modules:1:/usr/share/modules/$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1'
+++++ export MODULEPATH_modshare
+++++ BASH_ENV=/usr/share/modules/init/bash
+++++ export BASH_ENV
+++++ MODULESHOME=/usr/share/modules
+++++ export MODULESHOME
+++++ LOADEDMODULES=
+++++ export LOADEDMODULES
+++++ MODULEPATH='/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles'
+++++ export MODULEPATH
+++++ test 0
++++ '[' 0 = 1 ']'
++++ '[' -t 2 ']'
++++ export -f module
++++ export -f switchml
++++ '[' 5 -ge 3 ']'
++++ [[ ehxB =~ i ]]
++++ [[ ! :/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin: =~ :/usr/bin: ]]
++++ '[' '!' -n '' ']'
++++ MANPATH=:
++++ export MANPATH
+++++ manpath
++++ [[ ! :/admin/home/nouamane/.local/share/man:/fsx/nouamane/miniconda/envs/2-1-cu121/man:/fsx/nouamane/miniconda/envs/2-1-cu121/share/man:/fsx/nouamane/miniconda/man:/fsx/nouamane/miniconda/share/man:/opt/amazon/openmpi/share/man:/opt/amazon/efa/share/man:/opt/slurm/share/man:/usr/local/man:/usr/local/share/man:/usr/share/man:/admin/home/nouamane/.fzf/man:: =~ :/usr/share/man: ]]
++++ unset _mlcode _mlret
++++ '[' -n '' ']'
++ module load cuda/12.1
++ unset _mlshdbg
++ '[' 0 = 1 ']'
++ unset _mlre _mlIFS
++ '[' -n x ']'
++ _mlIFS=' 	
+'
++ IFS=' '
++ '[' -n '' ']'
+++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash load cuda/12.1
++ eval 'CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include;' export 'CPATH;
+LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:;' export 'LD_LIBRARY_PATH;
+FI_EFA_FORK_SAFE=1;' export 'FI_EFA_FORK_SAFE;
+MANPATH=/usr/local/cuda-12.1/share/man::;' export 'MANPATH;
+LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64;' export 'LIBRARY_PATH;
+_LMFILES_=/usr/share/modules/modulefiles/cuda/12.1;' export '_LMFILES_;
+LOADEDMODULES=cuda/12.1;' export 'LOADEDMODULES;
+MPI_PATH=/opt/amazon/openmpi;' export 'MPI_PATH;
+NCCL_HOME_modshare=/opt/nccl/build:1;' export 'NCCL_HOME_modshare;
+NCCL_PROTO=simple;' export 'NCCL_PROTO;
+MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1;' export 'MANPATH_modshare;
+LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1;' export 'LIBRARY_PATH_modshare;
+NCCL_SOCKET_IFNAME=enp;' export 'NCCL_SOCKET_IFNAME;
+AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl;' export 'AWS_OFI_NCCL_HOME;
+NCCL_HOME=/opt/nccl/build;' export 'NCCL_HOME;
+FI_PROVIDER=efa;' export 'FI_PROVIDER;
+AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1;' export 'AWS_OFI_NCCL_HOME_modshare;
+CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1;' export 'CPATH_modshare;
+LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2::1:/opt/nccl/build/lib:1:/opt/aws-ofi-nccl/lib:2:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/efa/lib:1;' export 'LD_LIBRARY_PATH_modshare;
+FI_EFA_ENABLE_SHM_TRANSFER=1;' export 'FI_EFA_ENABLE_SHM_TRANSFER;
+_LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1;' export '_LMFILES__modshare;
+LOADEDMODULES_modshare=cuda/12.1:1;' export 'LOADEDMODULES_modshare;
+MPI_PATH_modshare=/opt/amazon/openmpi:1;' export 'MPI_PATH_modshare;
+PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin;' export 'PATH;
+CUDA_HOME=/usr/local/cuda-12.1;' export 'CUDA_HOME;
+PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/usr/local/cuda-12.1/include:1:/opt/amazon/efa/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/usr/sbin:1:/fsx/nouamane/miniconda/bin:1:/opt/slurm/bin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:2:/opt/amazon/openmpi/bin:1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1;' export 'PATH_modshare;
+test' '0;'
+++ CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include
+++ export CPATH
+++ LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:
+++ export LD_LIBRARY_PATH
+++ FI_EFA_FORK_SAFE=1
+++ export FI_EFA_FORK_SAFE
+++ MANPATH=/usr/local/cuda-12.1/share/man::
+++ export MANPATH
+++ LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64
+++ export LIBRARY_PATH
+++ _LMFILES_=/usr/share/modules/modulefiles/cuda/12.1
+++ export _LMFILES_
+++ LOADEDMODULES=cuda/12.1
+++ export LOADEDMODULES
+++ MPI_PATH=/opt/amazon/openmpi
+++ export MPI_PATH
+++ NCCL_HOME_modshare=/opt/nccl/build:1
+++ export NCCL_HOME_modshare
+++ NCCL_PROTO=simple
+++ export NCCL_PROTO
+++ MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1
+++ export MANPATH_modshare
+++ LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1
+++ export LIBRARY_PATH_modshare
+++ NCCL_SOCKET_IFNAME=enp
+++ export NCCL_SOCKET_IFNAME
+++ AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl
+++ export AWS_OFI_NCCL_HOME
+++ NCCL_HOME=/opt/nccl/build
+++ export NCCL_HOME
+++ FI_PROVIDER=efa
+++ export FI_PROVIDER
+++ AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1
+++ export AWS_OFI_NCCL_HOME_modshare
+++ CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1
+++ export CPATH_modshare
+++ LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2::1:/opt/nccl/build/lib:1:/opt/aws-ofi-nccl/lib:2:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/efa/lib:1
+++ export LD_LIBRARY_PATH_modshare
+++ FI_EFA_ENABLE_SHM_TRANSFER=1
+++ export FI_EFA_ENABLE_SHM_TRANSFER
+++ _LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1
+++ export _LMFILES__modshare
+++ LOADEDMODULES_modshare=cuda/12.1:1
+++ export LOADEDMODULES_modshare
+++ MPI_PATH_modshare=/opt/amazon/openmpi:1
+++ export MPI_PATH_modshare
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ export PATH
+++ CUDA_HOME=/usr/local/cuda-12.1
+++ export CUDA_HOME
+++ PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/usr/local/cuda-12.1/include:1:/opt/amazon/efa/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/usr/sbin:1:/fsx/nouamane/miniconda/bin:1:/opt/slurm/bin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:2:/opt/amazon/openmpi/bin:1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1
+++ export PATH_modshare
+++ test 0
++ _mlstatus=0
++ '[' -n x ']'
++ IFS=' 	
+'
++ unset _mlre _mlv _mlrv _mlIFS
++ '[' -n '' ']'
++ unset _mlshdbg
++ return 0
++ source /fsx/nouamane/miniconda/bin/activate
+++ _CONDA_ROOT=/fsx/nouamane/miniconda
+++ . /fsx/nouamane/miniconda/etc/profile.d/conda.sh
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ '[' -z x ']'
+++ conda activate
+++ local cmd=activate
+++ case "$cmd" in
+++ __conda_activate activate
+++ '[' -n '' ']'
+++ local ask_conda
++++ PS1=
++++ __conda_exe shell.posix activate
++++ /fsx/nouamane/miniconda/bin/conda shell.posix activate
+++ ask_conda='. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_2='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\'''
+++ eval '. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_2='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\'''
++++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh
+++++ test -n ''
+++++ unset XML_CATALOG_FILES
+++++ unset xml_catalog_files_libxml2
++++ PS1='(base) '
++++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++++ export CONDA_PREFIX=/fsx/nouamane/miniconda
++++ CONDA_PREFIX=/fsx/nouamane/miniconda
++++ export CONDA_SHLVL=3
++++ CONDA_SHLVL=3
++++ export CONDA_DEFAULT_ENV=base
++++ CONDA_DEFAULT_ENV=base
++++ export 'CONDA_PROMPT_MODIFIER=(base) '
++++ CONDA_PROMPT_MODIFIER='(base) '
++++ export CONDA_PREFIX_2=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ CONDA_PREFIX_2=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ __conda_hashr
+++ '[' -n '' ']'
+++ '[' -n '' ']'
+++ hash -r
++ conda activate 2-1-cu121
++ local cmd=activate
++ case "$cmd" in
++ __conda_activate activate 2-1-cu121
++ '[' -n '' ']'
++ local ask_conda
+++ PS1='(base) '
+++ __conda_exe shell.posix activate 2-1-cu121
+++ /fsx/nouamane/miniconda/bin/conda shell.posix activate 2-1-cu121
++ ask_conda='PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''4'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_3='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
++ eval 'PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''4'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_3='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
+++ PS1='(2-1-cu121) '
+++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ export CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ export CONDA_SHLVL=4
+++ CONDA_SHLVL=4
+++ export CONDA_DEFAULT_ENV=2-1-cu121
+++ CONDA_DEFAULT_ENV=2-1-cu121
+++ export 'CONDA_PROMPT_MODIFIER=(2-1-cu121) '
+++ CONDA_PROMPT_MODIFIER='(2-1-cu121) '
+++ export CONDA_PREFIX_3=/fsx/nouamane/miniconda
+++ CONDA_PREFIX_3=/fsx/nouamane/miniconda
+++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ export _CE_M=
+++ _CE_M=
+++ export _CE_CONDA=
+++ _CE_CONDA=
+++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh
++++ test -n ''
++++ xml_catalog_files_libxml2=
++++ XML_CATALOG_FILES=
++++ conda_catalog_files=
++++ ifs_libxml2=' 	
+'
++++ IFS=' '
++++ rem=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ for pre in ${rem}
++++ test '' = /fsx/nouamane/miniconda/envs/2-1-cu121
++++ conda_catalog_files=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ rem=
++++ IFS=' 	
+'
++++ conda_catalog_files='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ export 'XML_CATALOG_FILES=file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ XML_CATALOG_FILES='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ unset conda_catalog_files ifs_libxml2 rem
++ __conda_hashr
++ '[' -n '' ']'
++ '[' -n '' ']'
++ hash -r
++ export PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++ PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ scontrol show hostnames 'ip-26-0-161-178,ip-26-0-162-180,ip-26-0-164-[18,187],ip-26-0-165-[24,59],ip-26-0-166-125,ip-26-0-167-51'
++ export 'NODELIST=ip-26-0-161-178
+ip-26-0-162-180
+ip-26-0-164-18
+ip-26-0-164-187
+ip-26-0-165-24
+ip-26-0-165-59
+ip-26-0-166-125
+ip-26-0-167-51'
++ NODELIST='ip-26-0-161-178
+ip-26-0-162-180
+ip-26-0-164-18
+ip-26-0-164-187
+ip-26-0-165-24
+ip-26-0-165-59
+ip-26-0-166-125
+ip-26-0-167-51'
+++ head -n1
+++ scontrol show hostnames 'ip-26-0-161-178,ip-26-0-162-180,ip-26-0-164-[18,187],ip-26-0-165-[24,59],ip-26-0-166-125,ip-26-0-167-51'
++ export MASTER_NODE=ip-26-0-161-178
++ MASTER_NODE=ip-26-0-161-178
++ export MASTER_PORT=12356
++ MASTER_PORT=12356
++ export NNODES=8
++ NNODES=8
++ export GPUS_PER_NODE=8
++ GPUS_PER_NODE=8
++ export WORLD_SIZE=64
++ WORLD_SIZE=64
++ export CUDA_DEVICE_MAX_CONNECTIONS=1
++ CUDA_DEVICE_MAX_CONNECTIONS=1
++ export NANOTRON_BENCHMARK=1
++ NANOTRON_BENCHMARK=1
++ echo 'Master node: ip-26-0-161-178'
+Master node: ip-26-0-161-178
++ echo 'All nodes: ip-26-0-161-178
+ip-26-0-162-180
+ip-26-0-164-18
+ip-26-0-164-187
+ip-26-0-165-24
+ip-26-0-165-59
+ip-26-0-166-125
+ip-26-0-167-51'
+All nodes: ip-26-0-161-178
+ip-26-0-162-180
+ip-26-0-164-18
+ip-26-0-164-187
+ip-26-0-165-24
+ip-26-0-165-59
+ip-26-0-166-125
+ip-26-0-167-51
++ echo 'World size: 64'
+World size: 64
++ srun torchrun --nnodes=8 --nproc_per_node=8 --rdzv_id=12344441 --rdzv_backend=c10d --rdzv_endpoint=ip-26-0-161-178:12356 run_train.py --config-file benchmark/configs/config_3.56G_dp64_tp1_pp1_acc1_mbs2_seq2048_zero1_l28_h3072_heads24.yaml
+[2024-12-03 09:15:40,641] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-03 09:15:40,641] torch.distributed.run: [WARNING] 
+[2024-12-03 09:15:40,641] torch.distributed.run: [WARNING] *****************************************
+[2024-12-03 09:15:40,641] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-03 09:15:40,641] torch.distributed.run: [WARNING] *****************************************
+[2024-12-03 09:15:40,640] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-03 09:15:40,643] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-03 09:15:40,643] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-03 09:15:40,643] torch.distributed.run: [WARNING] 
+[2024-12-03 09:15:40,643] torch.distributed.run: [WARNING] *****************************************
+[2024-12-03 09:15:40,643] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-03 09:15:40,643] torch.distributed.run: [WARNING] *****************************************
+[2024-12-03 09:15:40,643] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-03 09:15:40,644] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-03 09:15:40,644] torch.distributed.run: [WARNING] 
+[2024-12-03 09:15:40,644] torch.distributed.run: [WARNING] *****************************************
+[2024-12-03 09:15:40,644] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-03 09:15:40,644] torch.distributed.run: [WARNING] *****************************************
+[2024-12-03 09:15:40,647] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-03 09:15:40,647] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-03 09:15:40,641] torch.distributed.run: [WARNING] 
+[2024-12-03 09:15:40,641] torch.distributed.run: [WARNING] *****************************************
+[2024-12-03 09:15:40,641] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-03 09:15:40,641] torch.distributed.run: [WARNING] *****************************************
+[2024-12-03 09:15:40,643] torch.distributed.run: [WARNING] 
+[2024-12-03 09:15:40,643] torch.distributed.run: [WARNING] *****************************************
+[2024-12-03 09:15:40,643] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-03 09:15:40,643] torch.distributed.run: [WARNING] *****************************************
+[2024-12-03 09:15:40,643] torch.distributed.run: [WARNING] 
+[2024-12-03 09:15:40,643] torch.distributed.run: [WARNING] *****************************************
+[2024-12-03 09:15:40,643] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-03 09:15:40,643] torch.distributed.run: [WARNING] *****************************************
+[2024-12-03 09:15:40,647] torch.distributed.run: [WARNING] 
+[2024-12-03 09:15:40,647] torch.distributed.run: [WARNING] *****************************************
+[2024-12-03 09:15:40,647] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-03 09:15:40,647] torch.distributed.run: [WARNING] *****************************************
+[2024-12-03 09:15:40,647] torch.distributed.run: [WARNING] 
+[2024-12-03 09:15:40,647] torch.distributed.run: [WARNING] *****************************************
+[2024-12-03 09:15:40,647] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-03 09:15:40,647] torch.distributed.run: [WARNING] *****************************************
+12/03/2024 09:16:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]: Config:
+12/03/2024 09:16:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]: Config(general=GeneralArgs(project='debug',
+12/03/2024 09:16:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]:                            run='3.56G_dp64_tp1_pp1_acc1_mbs2_seq2048_zero1_l28_h3072_heads24',
+12/03/2024 09:16:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]:                            seed=42,
+12/03/2024 09:16:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]:                            step=None,
+12/03/2024 09:16:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]:                            consumed_train_samples=None,
+12/03/2024 09:16:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]:                            benchmark_csv_path=PosixPath('bench_elie.csv'),
+12/03/2024 09:16:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]:                            ignore_sanity_checks=True),
+12/03/2024 09:16:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]:        parallelism=ParallelismArgs(dp=64,
+12/03/2024 09:16:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]:                                    pp=1,
+12/03/2024 09:16:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]:                                    tp=1,
+12/03/2024 09:16:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]:                                    pp_engine=<nanotron.parallel.pipeline_parallel.engine.OneForwardOneBackwardPipelineEngine object at 0x7f682b40ce50>,
+12/03/2024 09:16:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]:                                    tp_mode=<TensorParallelLinearMode.REDUCE_SCATTER: 2>,
+12/03/2024 09:16:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]:                                    tp_linear_async_communication=True,
+12/03/2024 09:16:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]:                                    recompute_layer=False,
+12/03/2024 09:16:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]:                                    tp_recompute_allgather=True,
+12/03/2024 09:16:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]:                                    expert_parallel_size=1),
+12/03/2024 09:16:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]:        model=ModelArgs(model_config=LlamaConfig(bos_token_id=0,
+12/03/2024 09:16:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]:                                                 eos_token_id=0,
+12/03/2024 09:16:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]:                                                 hidden_act='silu',
+12/03/2024 09:16:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]:                                                 hidden_size=3072,
+12/03/2024 09:16:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]:                                                 initializer_range=0.02,
+12/03/2024 09:16:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]:                                                 intermediate_size=8192,
+12/03/2024 09:16:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]:                                                 is_llama_config=True,
+12/03/2024 09:16:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]:                                                 max_position_embeddings=2048,
+12/03/2024 09:16:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]:                                                 num_attention_heads=24,
+12/03/2024 09:16:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]:                                                 num_hidden_layers=28,
+12/03/2024 09:16:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]:                                                 num_key_value_heads=8,
+12/03/2024 09:16:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]:                                                 pad_token_id=None,
+12/03/2024 09:16:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]:                                                 pretraining_tp=1,
+12/03/2024 09:16:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]:                                                 rms_norm_eps=1e-05,
+12/03/2024 09:16:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]:                                                 rope_scaling=None,
+12/03/2024 09:16:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]:                                                 rope_theta=10000.0,
+12/03/2024 09:16:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]:                                                 rope_interleaved=False,
+12/03/2024 09:16:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]:                                                 tie_word_embeddings=True,
+12/03/2024 09:16:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]:                                                 use_cache=True,
+12/03/2024 09:16:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]:                                                 vocab_size=128256),
+12/03/2024 09:16:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]:                        init_method=RandomInit(std=0.02),
+12/03/2024 09:16:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]:                        dtype=torch.bfloat16,
+12/03/2024 09:16:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]:                        make_vocab_size_divisible_by=1,
+12/03/2024 09:16:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]:                        ddp_bucket_cap_mb=25),
+12/03/2024 09:16:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]:        tokenizer=TokenizerArgs(tokenizer_name_or_path='robot-test/dummy-tokenizer-wordlevel',
+12/03/2024 09:16:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]:                                tokenizer_revision=None,
+12/03/2024 09:16:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]:                                tokenizer_max_length=None),
+12/03/2024 09:16:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]:        checkpoints=CheckpointsArgs(checkpoints_path=PosixPath('checkpoints'),
+12/03/2024 09:16:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]:                                    checkpoint_interval=10000,
+12/03/2024 09:16:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]:                                    save_initial_state=False,
+12/03/2024 09:16:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]:                                    save_final_state=False,
+12/03/2024 09:16:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]:                                    resume_checkpoint_path=None,
+12/03/2024 09:16:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]:                                    checkpoints_path_is_shared_file_system=False),
+12/03/2024 09:16:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]:        logging=LoggingArgs(log_level='info',
+12/03/2024 09:16:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]:                            log_level_replica='info',
+12/03/2024 09:16:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]:                            iteration_step_info_interval=1),
+12/03/2024 09:16:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]:        tokens=TokensArgs(sequence_length=2048,
+12/03/2024 09:16:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]:                          train_steps=100,
+12/03/2024 09:16:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]:                          micro_batch_size=2,
+12/03/2024 09:16:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]:                          batch_accumulation_per_replica=1,
+12/03/2024 09:16:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]:                          val_check_interval=100,
+12/03/2024 09:16:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]:                          limit_val_batches=0,
+12/03/2024 09:16:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]:                          limit_test_batches=0),
+12/03/2024 09:16:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]:        optimizer=OptimizerArgs(optimizer_factory=AdamWOptimizerArgs(adam_eps=1e-08,
+12/03/2024 09:16:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]:                                                                     adam_beta1=0.9,
+12/03/2024 09:16:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]:                                                                     adam_beta2=0.95,
+12/03/2024 09:16:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]:                                                                     torch_adam_is_fused=True,
+12/03/2024 09:16:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]:                                                                     name='adamW'),
+12/03/2024 09:16:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]:                                zero_stage=1,
+12/03/2024 09:16:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]:                                weight_decay=0.01,
+12/03/2024 09:16:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]:                                clip_grad=1.0,
+12/03/2024 09:16:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]:                                accumulate_grad_in_fp32=True,
+12/03/2024 09:16:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]:                                learning_rate_scheduler=LRSchedulerArgs(learning_rate=0.0003,
+12/03/2024 09:16:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]:                                                                        lr_warmup_steps=2,
+12/03/2024 09:16:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]:                                                                        lr_warmup_style='linear',
+12/03/2024 09:16:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]:                                                                        lr_decay_style='cosine',
+12/03/2024 09:16:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]:                                                                        lr_decay_steps=13,
+12/03/2024 09:16:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]:                                                                        lr_decay_starting_step=None,
+12/03/2024 09:16:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]:                                                                        min_decay_lr=1e-05)),
+12/03/2024 09:16:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]:        data_stages=[DatasetStageArgs(name='Stable Training Stage',
+12/03/2024 09:16:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]:                                      start_training_step=1,
+12/03/2024 09:16:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]:                                      data=DataArgs(dataset=None,
+12/03/2024 09:16:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]:                                                    seed=42,
+12/03/2024 09:16:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]:                                                    num_loading_workers=1))],
+12/03/2024 09:16:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]:        profiler=None,
+12/03/2024 09:16:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]:        lighteval=None,
+12/03/2024 09:16:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]:        s3_upload=None)
+12/03/2024 09:16:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]: Model Config:
+12/03/2024 09:16:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]: LlamaConfig(bos_token_id=0,
+12/03/2024 09:16:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]:             eos_token_id=0,
+12/03/2024 09:16:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]:             hidden_act='silu',
+12/03/2024 09:16:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]:             hidden_size=3072,
+12/03/2024 09:16:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]:             initializer_range=0.02,
+12/03/2024 09:16:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]:             intermediate_size=8192,
+12/03/2024 09:16:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]:             is_llama_config=True,
+12/03/2024 09:16:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]:             max_position_embeddings=2048,
+12/03/2024 09:16:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]:             num_attention_heads=24,
+12/03/2024 09:16:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]:             num_hidden_layers=28,
+12/03/2024 09:16:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]:             num_key_value_heads=8,
+12/03/2024 09:16:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]:             pad_token_id=None,
+12/03/2024 09:16:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]:             pretraining_tp=1,
+12/03/2024 09:16:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]:             rms_norm_eps=1e-05,
+12/03/2024 09:16:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]:             rope_scaling=None,
+12/03/2024 09:16:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]:             rope_theta=10000.0,
+12/03/2024 09:16:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]:             rope_interleaved=False,
+12/03/2024 09:16:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]:             tie_word_embeddings=True,
+12/03/2024 09:16:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]:             use_cache=True,
+12/03/2024 09:16:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]:             vocab_size=128256)
+12/03/2024 09:16:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]: Building model..
+12/03/2024 09:16:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]: Initialize RoPE Theta = 10000.0
+12/03/2024 09:16:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]: Setting PP block ranks...
+12/03/2024 09:16:20 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]: Total number of parameters: 3.21G (6127.83MiB)
+12/03/2024 09:16:20 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]: Local number of parameters: 3.21G (6127.83MiB)
+12/03/2024 09:16:20 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]: [After model building] Memory usage: 6127.87MiB. Peak allocated: 6879.37MiB Peak reserved: 35106.00MiB
+12/03/2024 09:16:20 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]: No checkpoint path provided.
+12/03/2024 09:16:20 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]: Parametrizing model parameters using StandardParametrizator
+12/03/2024 09:16:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]: [Optimizer Building] Using LearningRateForSP as learning rate
+12/03/2024 09:16:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]: [ZeRO sharding] Size of optimizer params per rank:
+12/03/2024 09:16:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]: [ZeRO sharding] DP Rank 0 has 50.2M out of 3.21G (1.56%) params' optimizer states
+12/03/2024 09:16:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]: [ZeRO sharding] DP Rank 1 has 50.2M out of 3.21G (1.56%) params' optimizer states
+12/03/2024 09:16:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]: [ZeRO sharding] DP Rank 2 has 50.2M out of 3.21G (1.56%) params' optimizer states
+12/03/2024 09:16:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]: [ZeRO sharding] DP Rank 3 has 50.2M out of 3.21G (1.56%) params' optimizer states
+12/03/2024 09:16:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]: [ZeRO sharding] DP Rank 4 has 50.2M out of 3.21G (1.56%) params' optimizer states
+12/03/2024 09:16:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]: [ZeRO sharding] DP Rank 5 has 50.2M out of 3.21G (1.56%) params' optimizer states
+12/03/2024 09:16:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]: [ZeRO sharding] DP Rank 6 has 50.2M out of 3.21G (1.56%) params' optimizer states
+12/03/2024 09:16:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]: [ZeRO sharding] DP Rank 7 has 50.2M out of 3.21G (1.56%) params' optimizer states
+12/03/2024 09:16:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]: [ZeRO sharding] DP Rank 8 has 50.2M out of 3.21G (1.56%) params' optimizer states
+12/03/2024 09:16:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]: [ZeRO sharding] DP Rank 9 has 50.2M out of 3.21G (1.56%) params' optimizer states
+12/03/2024 09:16:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]: [ZeRO sharding] DP Rank 10 has 50.2M out of 3.21G (1.56%) params' optimizer states
+12/03/2024 09:16:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]: [ZeRO sharding] DP Rank 11 has 50.2M out of 3.21G (1.56%) params' optimizer states
+12/03/2024 09:16:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]: [ZeRO sharding] DP Rank 12 has 50.2M out of 3.21G (1.56%) params' optimizer states
+12/03/2024 09:16:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]: [ZeRO sharding] DP Rank 13 has 50.2M out of 3.21G (1.56%) params' optimizer states
+12/03/2024 09:16:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]: [ZeRO sharding] DP Rank 14 has 50.2M out of 3.21G (1.56%) params' optimizer states
+12/03/2024 09:16:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]: [ZeRO sharding] DP Rank 15 has 50.2M out of 3.21G (1.56%) params' optimizer states
+12/03/2024 09:16:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]: [ZeRO sharding] DP Rank 16 has 50.2M out of 3.21G (1.56%) params' optimizer states
+12/03/2024 09:16:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]: [ZeRO sharding] DP Rank 17 has 50.2M out of 3.21G (1.56%) params' optimizer states
+12/03/2024 09:16:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]: [ZeRO sharding] DP Rank 18 has 50.2M out of 3.21G (1.56%) params' optimizer states
+12/03/2024 09:16:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]: [ZeRO sharding] DP Rank 19 has 50.2M out of 3.21G (1.56%) params' optimizer states
+12/03/2024 09:16:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]: [ZeRO sharding] DP Rank 20 has 50.2M out of 3.21G (1.56%) params' optimizer states
+12/03/2024 09:16:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]: [ZeRO sharding] DP Rank 21 has 50.2M out of 3.21G (1.56%) params' optimizer states
+12/03/2024 09:16:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]: [ZeRO sharding] DP Rank 22 has 50.2M out of 3.21G (1.56%) params' optimizer states
+12/03/2024 09:16:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]: [ZeRO sharding] DP Rank 23 has 50.2M out of 3.21G (1.56%) params' optimizer states
+12/03/2024 09:16:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]: [ZeRO sharding] DP Rank 24 has 50.2M out of 3.21G (1.56%) params' optimizer states
+12/03/2024 09:16:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]: [ZeRO sharding] DP Rank 25 has 50.2M out of 3.21G (1.56%) params' optimizer states
+12/03/2024 09:16:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]: [ZeRO sharding] DP Rank 26 has 50.2M out of 3.21G (1.56%) params' optimizer states
+12/03/2024 09:16:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]: [ZeRO sharding] DP Rank 27 has 50.2M out of 3.21G (1.56%) params' optimizer states
+12/03/2024 09:16:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]: [ZeRO sharding] DP Rank 28 has 50.2M out of 3.21G (1.56%) params' optimizer states
+12/03/2024 09:16:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]: [ZeRO sharding] DP Rank 29 has 50.2M out of 3.21G (1.56%) params' optimizer states
+12/03/2024 09:16:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]: [ZeRO sharding] DP Rank 30 has 50.2M out of 3.21G (1.56%) params' optimizer states
+12/03/2024 09:16:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]: [ZeRO sharding] DP Rank 31 has 50.2M out of 3.21G (1.56%) params' optimizer states
+12/03/2024 09:16:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]: [ZeRO sharding] DP Rank 32 has 50.2M out of 3.21G (1.56%) params' optimizer states
+12/03/2024 09:16:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]: [ZeRO sharding] DP Rank 33 has 50.2M out of 3.21G (1.56%) params' optimizer states
+12/03/2024 09:16:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]: [ZeRO sharding] DP Rank 34 has 50.2M out of 3.21G (1.56%) params' optimizer states
+12/03/2024 09:16:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]: [ZeRO sharding] DP Rank 35 has 50.2M out of 3.21G (1.56%) params' optimizer states
+12/03/2024 09:16:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]: [ZeRO sharding] DP Rank 36 has 50.2M out of 3.21G (1.56%) params' optimizer states
+12/03/2024 09:16:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]: [ZeRO sharding] DP Rank 37 has 50.2M out of 3.21G (1.56%) params' optimizer states
+12/03/2024 09:16:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]: [ZeRO sharding] DP Rank 38 has 50.2M out of 3.21G (1.56%) params' optimizer states
+12/03/2024 09:16:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]: [ZeRO sharding] DP Rank 39 has 50.2M out of 3.21G (1.56%) params' optimizer states
+12/03/2024 09:16:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]: [ZeRO sharding] DP Rank 40 has 50.2M out of 3.21G (1.56%) params' optimizer states
+12/03/2024 09:16:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]: [ZeRO sharding] DP Rank 41 has 50.2M out of 3.21G (1.56%) params' optimizer states
+12/03/2024 09:16:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]: [ZeRO sharding] DP Rank 42 has 50.2M out of 3.21G (1.56%) params' optimizer states
+12/03/2024 09:16:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]: [ZeRO sharding] DP Rank 43 has 50.2M out of 3.21G (1.56%) params' optimizer states
+12/03/2024 09:16:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]: [ZeRO sharding] DP Rank 44 has 50.2M out of 3.21G (1.56%) params' optimizer states
+12/03/2024 09:16:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]: [ZeRO sharding] DP Rank 45 has 50.2M out of 3.21G (1.56%) params' optimizer states
+12/03/2024 09:16:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]: [ZeRO sharding] DP Rank 46 has 50.2M out of 3.21G (1.56%) params' optimizer states
+12/03/2024 09:16:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]: [ZeRO sharding] DP Rank 47 has 50.2M out of 3.21G (1.56%) params' optimizer states
+12/03/2024 09:16:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]: [ZeRO sharding] DP Rank 48 has 50.2M out of 3.21G (1.56%) params' optimizer states
+12/03/2024 09:16:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]: [ZeRO sharding] DP Rank 49 has 50.2M out of 3.21G (1.56%) params' optimizer states
+12/03/2024 09:16:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]: [ZeRO sharding] DP Rank 50 has 50.2M out of 3.21G (1.56%) params' optimizer states
+12/03/2024 09:16:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]: [ZeRO sharding] DP Rank 51 has 50.2M out of 3.21G (1.56%) params' optimizer states
+12/03/2024 09:16:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]: [ZeRO sharding] DP Rank 52 has 50.2M out of 3.21G (1.56%) params' optimizer states
+12/03/2024 09:16:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]: [ZeRO sharding] DP Rank 53 has 50.2M out of 3.21G (1.56%) params' optimizer states
+12/03/2024 09:16:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]: [ZeRO sharding] DP Rank 54 has 50.2M out of 3.21G (1.56%) params' optimizer states
+12/03/2024 09:16:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]: [ZeRO sharding] DP Rank 55 has 50.2M out of 3.21G (1.56%) params' optimizer states
+12/03/2024 09:16:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]: [ZeRO sharding] DP Rank 56 has 50.2M out of 3.21G (1.56%) params' optimizer states
+12/03/2024 09:16:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]: [ZeRO sharding] DP Rank 57 has 50.2M out of 3.21G (1.56%) params' optimizer states
+12/03/2024 09:16:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]: [ZeRO sharding] DP Rank 58 has 50.2M out of 3.21G (1.56%) params' optimizer states
+12/03/2024 09:16:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]: [ZeRO sharding] DP Rank 59 has 50.2M out of 3.21G (1.56%) params' optimizer states
+12/03/2024 09:16:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]: [ZeRO sharding] DP Rank 60 has 50.2M out of 3.21G (1.56%) params' optimizer states
+12/03/2024 09:16:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]: [ZeRO sharding] DP Rank 61 has 50.2M out of 3.21G (1.56%) params' optimizer states
+12/03/2024 09:16:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]: [ZeRO sharding] DP Rank 62 has 50.2M out of 3.21G (1.56%) params' optimizer states
+12/03/2024 09:16:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]: [ZeRO sharding] DP Rank 63 has 50.2M out of 3.21G (1.56%) params' optimizer states
+12/03/2024 09:16:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]: [Training Plan] Stage Stable Training Stage has 99 remaining training steps and has consumed 0 samples
+12/03/2024 09:16:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]: Using dummy data generator
+12/03/2024 09:16:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]: [Training Plan] There are 1 training stages 
+12/03/2024 09:16:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]: [Stage Stable Training Stage] start from step 1 
+12/03/2024 09:16:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]: 
+12/03/2024 09:16:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]: [Start training] datetime: 2024-12-03 09:16:29.641045 | mbs: 2 | grad_accum: 1 | global_batch_size: 128 | sequence_length: 2048 | train_steps: 100 | start_iteration_step: 0 | consumed_train_samples: 0
+wandb: Tracking run with wandb version 0.16.0
+wandb: W&B syncing is set to `offline` in this directory.  
+wandb: Run `wandb online` or set WANDB_MODE=online to enable cloud syncing.
+12/03/2024 09:16:38 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]: Resuming training from stage Stable Training Stage, it has trained for 0 samples and has 99 remaining train steps
+12/03/2024 09:16:38 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]:  Memory usage: 18575.37MiB. Peak allocated 18575.37MiB. Peak reserved: 47362.00MiB
+12/03/2024 09:16:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]:  Memory usage: 18653.44MiB. Peak allocated 34337.55MiB. Peak reserved: 35472.00MiB
+12/03/2024 09:16:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]: iteration: 1 / 100 | consumed_tokens: 262K | elapsed_time_per_iteration_ms: 2.75K | tokens_per_sec: 95.4K | tokens_per_sec_per_gpu: 1.49K | global_batch_size: 128 | lm_loss: 12.4 | lr: 0.00015 | model_tflops_per_gpu: 31.9 | hardware_tflops_per_gpu: 31.9 | grad_norm: 5.34 | cuda_memory_allocated: 20G | cuda_max_memory_reserved: 43.7G | hd_total_memory_tb: 312G | hd_used_memory_tb: 190G | hd_free_memory_tb: 123G
+12/03/2024 09:16:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]:  Memory usage: 19043.58MiB. Peak allocated 31395.17MiB. Peak reserved: 41710.00MiB
+12/03/2024 09:16:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]:  Memory usage: 19043.58MiB. Peak allocated 34759.18MiB. Peak reserved: 41710.00MiB
+12/03/2024 09:16:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]: iteration: 2 / 100 | consumed_tokens: 524K | elapsed_time_per_iteration_ms: 443 | tokens_per_sec: 592K | tokens_per_sec_per_gpu: 9.25K | global_batch_size: 128 | lm_loss: 12.4 | lr: 0.0003 | model_tflops_per_gpu: 198 | hardware_tflops_per_gpu: 198 | grad_norm: 5.33 | cuda_memory_allocated: 20G | cuda_max_memory_reserved: 43.7G | hd_total_memory_tb: 312G | hd_used_memory_tb: 190G | hd_free_memory_tb: 123G
+12/03/2024 09:16:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]:  Memory usage: 19043.58MiB. Peak allocated 31395.17MiB. Peak reserved: 41710.00MiB
+12/03/2024 09:16:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]:  Memory usage: 19043.58MiB. Peak allocated 34759.18MiB. Peak reserved: 41710.00MiB
+num_params
+{'total': 3212749824, 'local': 3212749824}
+num_params
+{'total': 3212749824, 'local': 3212749824}
+num_params
+{'total': 3212749824, 'local': 3212749824}
+num_params
+{'total': 3212749824, 'local': 3212749824}
+num_params
+{'total': 3212749824, 'local': 3212749824}
+num_params
+{'total': 3212749824, 'local': 3212749824}
+num_params
+{'total': 3212749824, 'local': 3212749824}
+num_params
+num_params
+num_params
+{'total': 3212749824, 'local': 3212749824}
+{'total': 3212749824, 'local': 3212749824}
+num_params
+{'total': 3212749824, 'local': 3212749824}
+num_params
+{'total': 3212749824, 'local': 3212749824}
+num_params
+{'total': 3212749824, 'local': 3212749824}
+num_params
+{'total': 3212749824, 'local': 3212749824}
+{'total': 3212749824, 'local': 3212749824}
+num_params
+{'total': 3212749824, 'local': 3212749824}
+num_params
+{'total': 3212749824, 'local': 3212749824}
+num_params
+num_params
+{'total': 3212749824, 'local': 3212749824}
+{'total': 3212749824, 'local': 3212749824}
+num_params
+{'total': 3212749824, 'local': 3212749824}
+num_params
+{'total': 3212749824, 'local': 3212749824}
+num_params
+{'total': 3212749824, 'local': 3212749824}
+num_params
+num_params
+{'total': 3212749824, 'local': 3212749824}
+num_params
+num_params
+{'total': 3212749824, 'local': 3212749824}
+num_params
+{'total': 3212749824, 'local': 3212749824}
+num_params{'total': 3212749824, 'local': 3212749824}
+
+{'total': 3212749824, 'local': 3212749824}
+num_params
+{'total': 3212749824, 'local': 3212749824}
+num_params
+{'total': 3212749824, 'local': 3212749824}
+{'total': 3212749824, 'local': 3212749824}
+num_params
+{'total': 3212749824, 'local': 3212749824}
+num_params
+{'total': 3212749824, 'local': 3212749824}
+num_params
+{'total': 3212749824, 'local': 3212749824}
+num_params
+num_params
+{'total': 3212749824, 'local': 3212749824}
+num_params
+{'total': 3212749824, 'local': 3212749824}
+{'total': 3212749824, 'local': 3212749824}
+num_params
+{'total': 3212749824, 'local': 3212749824}
+num_params
+num_params
+{'total': 3212749824, 'local': 3212749824}
+num_params
+{'total': 3212749824, 'local': 3212749824}
+num_params
+{'total': 3212749824, 'local': 3212749824}
+num_params
+num_params
+{'total': 3212749824, 'local': 3212749824}
+{'total': 3212749824, 'local': 3212749824}
+{'total': 3212749824, 'local': 3212749824}
+num_params
+{'total': 3212749824, 'local': 3212749824}
+num_params
+{'total': 3212749824, 'local': 3212749824}
+12/03/2024 09:16:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]: iteration: 3 / 100 | consumed_tokens: 786K | elapsed_time_per_iteration_ms: 445 | tokens_per_sec: 588K | tokens_per_sec_per_gpu: 9.19K | global_batch_size: 128 | lm_loss: 12.4 | lr: 0.000296 | model_tflops_per_gpu: 197 | hardware_tflops_per_gpu: 197 | grad_norm: 5.17 | cuda_memory_allocated: 20G | cuda_max_memory_reserved: 43.7G | hd_total_memory_tb: 312G | hd_used_memory_tb: 190G | hd_free_memory_tb: 123G
+num_params
+{'total': 3212749824, 'local': 3212749824}
+num_params
+{'total': 3212749824, 'local': 3212749824}
+num_paramsnum_params
+
+num_params
+{'total': 3212749824, 'local': 3212749824}
+num_paramsnum_params
+
+{'total': 3212749824, 'local': 3212749824}
+{'total': 3212749824, 'local': 3212749824}
+num_params
+{'total': 3212749824, 'local': 3212749824}
+{'total': 3212749824, 'local': 3212749824}{'total': 3212749824, 'local': 3212749824}
+
+num_params
+{'total': 3212749824, 'local': 3212749824}
+num_params
+{'total': 3212749824, 'local': 3212749824}
+num_params
+{'total': 3212749824, 'local': 3212749824}
+num_params
+{'total': 3212749824, 'local': 3212749824}
+num_params
+{'total': 3212749824, 'local': 3212749824}
+num_params
+{'total': 3212749824, 'local': 3212749824}
+num_params
+{'total': 3212749824, 'local': 3212749824}
+num_params
+num_params
+{'total': 3212749824, 'local': 3212749824}num_params
+
+{'total': 3212749824, 'local': 3212749824}
+num_params
+{'total': 3212749824, 'local': 3212749824}
+num_params
+{'total': 3212749824, 'local': 3212749824}
+{'total': 3212749824, 'local': 3212749824}
+12/03/2024 09:16:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]: | job_id   | name                                                         | nodes | seq_len | mbs | batch_accum | gbs | mTFLOPs | hTFLOPs | tok/s/gpu | AllReduce (GB/s) | AllGather (GB/s) | ReduceScatter (GB/s) | AR Intra-node (GB/s) | AG Intra-node (GB/s) | RS Intra-node (GB/s) | Mem Alloc (GB) | Mem Res (GB) | dp | pp | tp | pp_engine                                                                                                 | tp_mode                                 | tp_async_comm | hidden_size | hidden_act | num_layers | num_heads | num_kv_heads | max_pos | vocab_size | tie_word_embeddings | dtype          | zero_stage | ddp_bucket_cap_mb | accumulate_grad_in_fp32 | Total Params | Local Params |
+12/03/2024 09:16:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]: | -------- | ------------------------------------------------------------ | ----- | ------- | --- | ----------- | --- | ------- | ------- | --------- | ---------------- | ---------------- | -------------------- | -------------------- | -------------------- | -------------------- | -------------- | ------------ | -- | -- | -- | --------------------------------------------------------------------------------------------------------- | --------------------------------------- | ------------- | ----------- | ---------- | ---------- | --------- | ------------ | ------- | ---------- | ------------------- | -------------- | ---------- | ----------------- | ----------------------- | ------------ | ------------ |
+12/03/2024 09:16:41 [INFO|DP=50|PP=0|TP=0|ip-26-0-166-125]: Throughput logging complete
+12/03/2024 09:16:41 [INFO|DP=51|PP=0|TP=0|ip-26-0-166-125]: Throughput logging complete
+12/03/2024 09:16:41 [INFO|DP=48|PP=0|TP=0|ip-26-0-166-125]: Throughput logging complete
+12/03/2024 09:16:41 [INFO|DP=35|PP=0|TP=0|ip-26-0-165-24]: Throughput logging complete
+12/03/2024 09:16:41 [INFO|DP=18|PP=0|TP=0|ip-26-0-164-18]: Throughput logging complete
+12/03/2024 09:16:41 [INFO|DP=17|PP=0|TP=0|ip-26-0-164-18]: Throughput logging complete
+12/03/2024 09:16:41 [INFO|DP=16|PP=0|TP=0|ip-26-0-164-18]: Throughput logging complete
+12/03/2024 09:16:41 [INFO|DP=49|PP=0|TP=0|ip-26-0-166-125]: Throughput logging complete
+12/03/2024 09:16:41 [INFO|DP=36|PP=0|TP=0|ip-26-0-165-24]: Throughput logging complete
+12/03/2024 09:16:41 [INFO|DP=19|PP=0|TP=0|ip-26-0-164-18]: Throughput logging complete
+12/03/2024 09:16:41 [INFO|DP=32|PP=0|TP=0|ip-26-0-165-24]: Throughput logging complete
+12/03/2024 09:16:41 [INFO|DP=33|PP=0|TP=0|ip-26-0-165-24]: Throughput logging complete
+12/03/2024 09:16:41 [INFO|DP=9|PP=0|TP=0|ip-26-0-162-180]: Throughput logging complete
+12/03/2024 09:16:41 [INFO|DP=25|PP=0|TP=0|ip-26-0-164-187]: Throughput logging complete
+12/03/2024 09:16:41 [INFO|DP=23|PP=0|TP=0|ip-26-0-164-18]: Throughput logging complete
+12/03/2024 09:16:41 [INFO|DP=20|PP=0|TP=0|ip-26-0-164-18]: Throughput logging complete
+12/03/2024 09:16:41 [INFO|DP=52|PP=0|TP=0|ip-26-0-166-125]: Throughput logging complete
+12/03/2024 09:16:41 [INFO|DP=53|PP=0|TP=0|ip-26-0-166-125]: Throughput logging complete
+12/03/2024 09:16:41 [INFO|DP=55|PP=0|TP=0|ip-26-0-166-125]: Throughput logging complete
+12/03/2024 09:16:41 [INFO|DP=43|PP=0|TP=0|ip-26-0-165-59]: Throughput logging complete
+12/03/2024 09:16:41 [INFO|DP=41|PP=0|TP=0|ip-26-0-165-59]: Throughput logging complete
+12/03/2024 09:16:41 [INFO|DP=12|PP=0|TP=0|ip-26-0-162-180]: Throughput logging complete
+12/03/2024 09:16:41 [INFO|DP=10|PP=0|TP=0|ip-26-0-162-180]: Throughput logging complete
+12/03/2024 09:16:41 [INFO|DP=58|PP=0|TP=0|ip-26-0-167-51]: Throughput logging complete
+12/03/2024 09:16:41 [INFO|DP=57|PP=0|TP=0|ip-26-0-167-51]: Throughput logging complete
+12/03/2024 09:16:41 [INFO|DP=62|PP=0|TP=0|ip-26-0-167-51]: Throughput logging complete
+12/03/2024 09:16:41 [INFO|DP=26|PP=0|TP=0|ip-26-0-164-187]: Throughput logging complete
+12/03/2024 09:16:41 [INFO|DP=24|PP=0|TP=0|ip-26-0-164-187]: Throughput logging complete
+12/03/2024 09:16:41 [INFO|DP=37|PP=0|TP=0|ip-26-0-165-24]: Throughput logging complete
+12/03/2024 09:16:41 [INFO|DP=22|PP=0|TP=0|ip-26-0-164-18]: Throughput logging complete
+12/03/2024 09:16:41 [INFO|DP=21|PP=0|TP=0|ip-26-0-164-18]: Throughput logging complete
+12/03/2024 09:16:41 [INFO|DP=15|PP=0|TP=0|ip-26-0-162-180]: Throughput logging complete
+12/03/2024 09:16:41 [INFO|DP=27|PP=0|TP=0|ip-26-0-164-187]: Throughput logging complete
+12/03/2024 09:16:41 [INFO|DP=28|PP=0|TP=0|ip-26-0-164-187]: Throughput logging complete
+12/03/2024 09:16:41 [INFO|DP=54|PP=0|TP=0|ip-26-0-166-125]: Throughput logging complete
+12/03/2024 09:16:41 [INFO|DP=39|PP=0|TP=0|ip-26-0-165-24]: Throughput logging complete
+12/03/2024 09:16:41 [INFO|DP=34|PP=0|TP=0|ip-26-0-165-24]: Throughput logging complete
+12/03/2024 09:16:41 [INFO|DP=38|PP=0|TP=0|ip-26-0-165-24]: Throughput logging complete
+12/03/2024 09:16:41 [INFO|DP=42|PP=0|TP=0|ip-26-0-165-59]: Throughput logging complete
+12/03/2024 09:16:41 [INFO|DP=44|PP=0|TP=0|ip-26-0-165-59]: Throughput logging complete
+12/03/2024 09:16:41 [INFO|DP=45|PP=0|TP=0|ip-26-0-165-59]: Throughput logging complete
+12/03/2024 09:16:41 [INFO|DP=13|PP=0|TP=0|ip-26-0-162-180]: Throughput logging complete
+12/03/2024 09:16:41 [INFO|DP=56|PP=0|TP=0|ip-26-0-167-51]: Throughput logging complete
+12/03/2024 09:16:41 [INFO|DP=60|PP=0|TP=0|ip-26-0-167-51]: Throughput logging complete
+12/03/2024 09:16:41 [INFO|DP=47|PP=0|TP=0|ip-26-0-165-59]: Throughput logging complete
+12/03/2024 09:16:41 [INFO|DP=46|PP=0|TP=0|ip-26-0-165-59]: Throughput logging complete
+12/03/2024 09:16:41 [INFO|DP=40|PP=0|TP=0|ip-26-0-165-59]: Throughput logging complete
+12/03/2024 09:16:41 [INFO|DP=11|PP=0|TP=0|ip-26-0-162-180]: Throughput logging complete
+12/03/2024 09:16:41 [INFO|DP=59|PP=0|TP=0|ip-26-0-167-51]: Throughput logging complete
+12/03/2024 09:16:41 [INFO|DP=29|PP=0|TP=0|ip-26-0-164-187]: Throughput logging complete
+12/03/2024 09:16:41 [INFO|DP=30|PP=0|TP=0|ip-26-0-164-187]: Throughput logging complete
+12/03/2024 09:16:41 [INFO|DP=31|PP=0|TP=0|ip-26-0-164-187]: Throughput logging complete
+12/03/2024 09:16:41 [INFO|DP=14|PP=0|TP=0|ip-26-0-162-180]: Throughput logging complete
+12/03/2024 09:16:41 [INFO|DP=61|PP=0|TP=0|ip-26-0-167-51]: Throughput logging complete
+12/03/2024 09:16:41 [INFO|DP=8|PP=0|TP=0|ip-26-0-162-180]: Throughput logging complete
+12/03/2024 09:16:41 [INFO|DP=63|PP=0|TP=0|ip-26-0-167-51]: Throughput logging complete
+12/03/2024 09:16:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]: | 12344441 | 3.56G_dp64_tp1_pp1_acc1_mbs2_seq2048_zero1_l28_h3072_heads24 | 8     | 2048    | 2   | 1           | 128 | 196.66  | 196.66  | 9194.35   | 213.10           | 162.48           | 164.26               | 459.18               | 264.89               | 265.19               | 30.66          | 40.73        | 64 | 1  | 1  | <nanotron.parallel.pipeline_parallel.engine.OneForwardOneBackwardPipelineEngine object at 0x7f682b40ce50> | TensorParallelLinearMode.REDUCE_SCATTER | True          | 3072        | silu       | 28         | 24        | 8            | 2048    | 128256     | True                | torch.bfloat16 | 1          | 25                | True                    | 3.21G        | 3.21G        |
+12/03/2024 09:16:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-178]: Throughput logging complete
+12/03/2024 09:16:41 [INFO|DP=3|PP=0|TP=0|ip-26-0-161-178]: Throughput logging complete
+12/03/2024 09:16:41 [INFO|DP=4|PP=0|TP=0|ip-26-0-161-178]: Throughput logging complete
+12/03/2024 09:16:41 [INFO|DP=1|PP=0|TP=0|ip-26-0-161-178]: Throughput logging complete
+12/03/2024 09:16:41 [INFO|DP=6|PP=0|TP=0|ip-26-0-161-178]: Throughput logging complete
+12/03/2024 09:16:41 [INFO|DP=2|PP=0|TP=0|ip-26-0-161-178]: Throughput logging complete
+12/03/2024 09:16:41 [INFO|DP=5|PP=0|TP=0|ip-26-0-161-178]: Throughput logging complete
+12/03/2024 09:16:41 [INFO|DP=7|PP=0|TP=0|ip-26-0-161-178]: Throughput logging complete
+srun: Job step aborted: Waiting up to 32 seconds for job step to finish.
+slurmstepd: error: *** JOB 12344441 ON ip-26-0-161-178 CANCELLED AT 2024-12-03T09:16:41 ***
+[2024-12-03 09:16:41,838] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-03 09:16:41,839] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 375531 closing signal SIGTERM
+[2024-12-03 09:16:41,839] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 375532 closing signal SIGTERM
+[2024-12-03 09:16:41,839] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 375533 closing signal SIGTERM
+[2024-12-03 09:16:41,839] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 375534 closing signal SIGTERM
+[2024-12-03 09:16:41,839] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-03 09:16:41,838] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-03 09:16:41,839] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 375535 closing signal SIGTERM
+[2024-12-03 09:16:41,839] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3461578 closing signal SIGTERM
+[2024-12-03 09:16:41,839] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1827641 closing signal SIGTERM
+[2024-12-03 09:16:41,839] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3461579 closing signal SIGTERM
+[2024-12-03 09:16:41,839] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3461580 closing signal SIGTERM
+[2024-12-03 09:16:41,839] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1827642 closing signal SIGTERM
+[2024-12-03 09:16:41,839] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-03 09:16:41,839] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-03 09:16:41,839] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1827643 closing signal SIGTERM
+[2024-12-03 09:16:41,839] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1827644 closing signal SIGTERM
+[2024-12-03 09:16:41,840] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 152755 closing signal SIGTERM
+[2024-12-03 09:16:41,839] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1714077 closing signal SIGTERM
+[2024-12-03 09:16:41,839] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1827645 closing signal SIGTERM
+[2024-12-03 09:16:41,840] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 152756 closing signal SIGTERM
+[2024-12-03 09:16:41,839] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-03 09:16:41,839] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1714078 closing signal SIGTERM
+[2024-12-03 09:16:41,839] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2472112 closing signal SIGTERM
+[2024-12-03 09:16:41,840] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 152757 closing signal SIGTERM
+[2024-12-03 09:16:41,839] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1827646 closing signal SIGTERM
+[2024-12-03 09:16:41,840] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 152758 closing signal SIGTERM
+[2024-12-03 09:16:41,839] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1714079 closing signal SIGTERM
+[2024-12-03 09:16:41,839] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1714080 closing signal SIGTERM
+[2024-12-03 09:16:41,839] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2472113 closing signal SIGTERM
+[2024-12-03 09:16:41,839] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2472114 closing signal SIGTERM
+[2024-12-03 09:16:41,839] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-03 09:16:41,839] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3461581 closing signal SIGTERM
+[2024-12-03 09:16:41,839] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 4068478 closing signal SIGTERM
+[2024-12-03 09:16:41,839] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 4068479 closing signal SIGTERM
+[2024-12-03 09:16:41,839] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 4068480 closing signal SIGTERM
+[2024-12-03 09:16:41,840] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 4068481 closing signal SIGTERM
+[2024-12-03 09:16:41,839] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1714081 closing signal SIGTERM
+[2024-12-03 09:16:41,841] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2472115 closing signal SIGTERM
+[2024-12-03 09:16:41,841] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2472116 closing signal SIGTERM
+[2024-12-03 09:16:41,841] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2472117 closing signal SIGTERM
+[2024-12-03 09:16:41,840] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1827647 closing signal SIGTERM
+[2024-12-03 09:16:41,841] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 4068482 closing signal SIGTERM
+[2024-12-03 09:16:41,841] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2472118 closing signal SIGTERM
+[2024-12-03 09:16:41,841] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 4068483 closing signal SIGTERM
+[2024-12-03 09:16:41,842] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2472119 closing signal SIGTERM
+[2024-12-03 09:16:41,841] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 4068484 closing signal SIGTERM
+[2024-12-03 09:16:41,842] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1714082 closing signal SIGTERM
+[2024-12-03 09:16:41,842] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3461582 closing signal SIGTERM
+[2024-12-03 09:16:41,843] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1714083 closing signal SIGTERM
+[2024-12-03 09:16:41,841] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 152759 closing signal SIGTERM
+[2024-12-03 09:16:41,842] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3461583 closing signal SIGTERM
+[2024-12-03 09:16:41,842] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3461584 closing signal SIGTERM
+[2024-12-03 09:16:41,843] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3461585 closing signal SIGTERM
+[2024-12-03 09:16:41,843] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 4068485 closing signal SIGTERM
+[2024-12-03 09:16:41,843] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1714084 closing signal SIGTERM
+[2024-12-03 09:16:41,843] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1827648 closing signal SIGTERM
diff --git a/logs/13212436-bench_1.34G_dp8_tp1_pp1_acc8_mbs32_seq2048_zero1_tpmodeRED_vocab131k.out b/logs/13212436-bench_1.34G_dp8_tp1_pp1_acc8_mbs32_seq2048_zero1_tpmodeRED_vocab131k.out
new file mode 100644
index 0000000000000000000000000000000000000000..c9d4b639e0be9cbc82fb107fa0417145491c1168
--- /dev/null
+++ b/logs/13212436-bench_1.34G_dp8_tp1_pp1_acc8_mbs32_seq2048_zero1_tpmodeRED_vocab131k.out
@@ -0,0 +1,1065 @@
++ source /etc/profile.d/modules.sh
+++ . /usr/share/modules/init/bash
++++ unset _mlshdbg
++++ '[' 0 = 1 ']'
++++ unset _mlre _mlIFS
++++ '[' -n x ']'
++++ _mlIFS=' 	
+'
++++ IFS=' '
++++ '[' -n '' ']'
+++++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash autoinit
++++ _mlcode='module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+MODULEPATH_modshare=/etc/environment-modules/modules:1:/usr/share/modules/\$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1; export MODULEPATH_modshare;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+LOADEDMODULES=; export LOADEDMODULES;
+MODULEPATH=/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/\$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles; export MODULEPATH;
+test 0;'
++++ _mlret=0
++++ '[' -n x ']'
++++ IFS=' 	
+'
++++ unset _mlIFS
++++ unset _mlre _mlv _mlrv
++++ '[' 0 -eq 0 ']'
++++ eval 'module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+MODULEPATH_modshare=/etc/environment-modules/modules:1:/usr/share/modules/\$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1; export MODULEPATH_modshare;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+LOADEDMODULES=; export LOADEDMODULES;
+MODULEPATH=/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/\$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles; export MODULEPATH;
+test 0;'
+++++ MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl
+++++ export MODULES_CMD
+++++ ENV=/usr/share/modules/init/profile.sh
+++++ export ENV
+++++ MODULEPATH_modshare='/etc/environment-modules/modules:1:/usr/share/modules/$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1'
+++++ export MODULEPATH_modshare
+++++ BASH_ENV=/usr/share/modules/init/bash
+++++ export BASH_ENV
+++++ MODULESHOME=/usr/share/modules
+++++ export MODULESHOME
+++++ LOADEDMODULES=
+++++ export LOADEDMODULES
+++++ MODULEPATH='/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles'
+++++ export MODULEPATH
+++++ test 0
++++ '[' 0 = 1 ']'
++++ '[' -t 2 ']'
++++ export -f module
++++ export -f switchml
++++ '[' 5 -ge 3 ']'
++++ [[ ehxB =~ i ]]
++++ [[ ! :/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin: =~ :/usr/bin: ]]
++++ '[' '!' -n '' ']'
++++ MANPATH=:
++++ export MANPATH
+++++ manpath
++++ [[ ! :/admin/home/nouamane/.local/share/man:/fsx/nouamane/miniconda/envs/2-1-cu121/man:/fsx/nouamane/miniconda/envs/2-1-cu121/share/man:/fsx/nouamane/miniconda/man:/fsx/nouamane/miniconda/share/man:/opt/amazon/openmpi/share/man:/opt/amazon/efa/share/man:/opt/slurm/share/man:/usr/local/man:/usr/local/share/man:/usr/share/man:/admin/home/nouamane/.fzf/man:: =~ :/usr/share/man: ]]
++++ unset _mlcode _mlret
++++ '[' -n '' ']'
++ module load cuda/12.1
++ unset _mlshdbg
++ '[' 0 = 1 ']'
++ unset _mlre _mlIFS
++ '[' -n x ']'
++ _mlIFS=' 	
+'
++ IFS=' '
++ '[' -n '' ']'
+++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash load cuda/12.1
++ eval 'CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include;' export 'CPATH;
+LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:;' export 'LD_LIBRARY_PATH;
+FI_EFA_FORK_SAFE=1;' export 'FI_EFA_FORK_SAFE;
+MANPATH=/usr/local/cuda-12.1/share/man::;' export 'MANPATH;
+LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64;' export 'LIBRARY_PATH;
+_LMFILES_=/usr/share/modules/modulefiles/cuda/12.1;' export '_LMFILES_;
+LOADEDMODULES=cuda/12.1;' export 'LOADEDMODULES;
+MPI_PATH=/opt/amazon/openmpi;' export 'MPI_PATH;
+NCCL_HOME_modshare=/opt/nccl/build:1;' export 'NCCL_HOME_modshare;
+NCCL_PROTO=simple;' export 'NCCL_PROTO;
+MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1;' export 'MANPATH_modshare;
+LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1;' export 'LIBRARY_PATH_modshare;
+NCCL_SOCKET_IFNAME=enp;' export 'NCCL_SOCKET_IFNAME;
+AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl;' export 'AWS_OFI_NCCL_HOME;
+NCCL_HOME=/opt/nccl/build;' export 'NCCL_HOME;
+FI_PROVIDER=efa;' export 'FI_PROVIDER;
+AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1;' export 'AWS_OFI_NCCL_HOME_modshare;
+CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1;' export 'CPATH_modshare;
+LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2::1:/opt/nccl/build/lib:1:/opt/aws-ofi-nccl/lib:2:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/efa/lib:1;' export 'LD_LIBRARY_PATH_modshare;
+FI_EFA_ENABLE_SHM_TRANSFER=1;' export 'FI_EFA_ENABLE_SHM_TRANSFER;
+_LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1;' export '_LMFILES__modshare;
+LOADEDMODULES_modshare=cuda/12.1:1;' export 'LOADEDMODULES_modshare;
+MPI_PATH_modshare=/opt/amazon/openmpi:1;' export 'MPI_PATH_modshare;
+PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin;' export 'PATH;
+CUDA_HOME=/usr/local/cuda-12.1;' export 'CUDA_HOME;
+PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/usr/local/cuda-12.1/include:1:/opt/amazon/efa/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/usr/sbin:1:/fsx/nouamane/miniconda/bin:1:/opt/slurm/bin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:2:/opt/amazon/openmpi/bin:1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1;' export 'PATH_modshare;
+test' '0;'
+++ CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include
+++ export CPATH
+++ LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:
+++ export LD_LIBRARY_PATH
+++ FI_EFA_FORK_SAFE=1
+++ export FI_EFA_FORK_SAFE
+++ MANPATH=/usr/local/cuda-12.1/share/man::
+++ export MANPATH
+++ LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64
+++ export LIBRARY_PATH
+++ _LMFILES_=/usr/share/modules/modulefiles/cuda/12.1
+++ export _LMFILES_
+++ LOADEDMODULES=cuda/12.1
+++ export LOADEDMODULES
+++ MPI_PATH=/opt/amazon/openmpi
+++ export MPI_PATH
+++ NCCL_HOME_modshare=/opt/nccl/build:1
+++ export NCCL_HOME_modshare
+++ NCCL_PROTO=simple
+++ export NCCL_PROTO
+++ MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1
+++ export MANPATH_modshare
+++ LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1
+++ export LIBRARY_PATH_modshare
+++ NCCL_SOCKET_IFNAME=enp
+++ export NCCL_SOCKET_IFNAME
+++ AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl
+++ export AWS_OFI_NCCL_HOME
+++ NCCL_HOME=/opt/nccl/build
+++ export NCCL_HOME
+++ FI_PROVIDER=efa
+++ export FI_PROVIDER
+++ AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1
+++ export AWS_OFI_NCCL_HOME_modshare
+++ CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1
+++ export CPATH_modshare
+++ LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2::1:/opt/nccl/build/lib:1:/opt/aws-ofi-nccl/lib:2:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/efa/lib:1
+++ export LD_LIBRARY_PATH_modshare
+++ FI_EFA_ENABLE_SHM_TRANSFER=1
+++ export FI_EFA_ENABLE_SHM_TRANSFER
+++ _LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1
+++ export _LMFILES__modshare
+++ LOADEDMODULES_modshare=cuda/12.1:1
+++ export LOADEDMODULES_modshare
+++ MPI_PATH_modshare=/opt/amazon/openmpi:1
+++ export MPI_PATH_modshare
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ export PATH
+++ CUDA_HOME=/usr/local/cuda-12.1
+++ export CUDA_HOME
+++ PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/usr/local/cuda-12.1/include:1:/opt/amazon/efa/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/usr/sbin:1:/fsx/nouamane/miniconda/bin:1:/opt/slurm/bin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:2:/opt/amazon/openmpi/bin:1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1
+++ export PATH_modshare
+++ test 0
++ _mlstatus=0
++ '[' -n x ']'
++ IFS=' 	
+'
++ unset _mlre _mlv _mlrv _mlIFS
++ '[' -n '' ']'
++ unset _mlshdbg
++ return 0
++ source /fsx/nouamane/miniconda/bin/activate
+++ _CONDA_ROOT=/fsx/nouamane/miniconda
+++ . /fsx/nouamane/miniconda/etc/profile.d/conda.sh
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ '[' -z x ']'
+++ conda activate
+++ local cmd=activate
+++ case "$cmd" in
+++ __conda_activate activate
+++ '[' -n '' ']'
+++ local ask_conda
++++ PS1=
++++ __conda_exe shell.posix activate
++++ /fsx/nouamane/miniconda/bin/conda shell.posix activate
+++ ask_conda='. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_2='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\'''
+++ eval '. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_2='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\'''
++++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh
+++++ test -n ''
+++++ unset XML_CATALOG_FILES
+++++ unset xml_catalog_files_libxml2
++++ PS1='(base) '
++++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++++ export CONDA_PREFIX=/fsx/nouamane/miniconda
++++ CONDA_PREFIX=/fsx/nouamane/miniconda
++++ export CONDA_SHLVL=3
++++ CONDA_SHLVL=3
++++ export CONDA_DEFAULT_ENV=base
++++ CONDA_DEFAULT_ENV=base
++++ export 'CONDA_PROMPT_MODIFIER=(base) '
++++ CONDA_PROMPT_MODIFIER='(base) '
++++ export CONDA_PREFIX_2=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ CONDA_PREFIX_2=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ __conda_hashr
+++ '[' -n '' ']'
+++ '[' -n '' ']'
+++ hash -r
++ conda activate 2-1-cu121
++ local cmd=activate
++ case "$cmd" in
++ __conda_activate activate 2-1-cu121
++ '[' -n '' ']'
++ local ask_conda
+++ PS1='(base) '
+++ __conda_exe shell.posix activate 2-1-cu121
+++ /fsx/nouamane/miniconda/bin/conda shell.posix activate 2-1-cu121
++ ask_conda='PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''4'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_3='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
++ eval 'PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''4'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_3='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
+++ PS1='(2-1-cu121) '
+++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ export CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ export CONDA_SHLVL=4
+++ CONDA_SHLVL=4
+++ export CONDA_DEFAULT_ENV=2-1-cu121
+++ CONDA_DEFAULT_ENV=2-1-cu121
+++ export 'CONDA_PROMPT_MODIFIER=(2-1-cu121) '
+++ CONDA_PROMPT_MODIFIER='(2-1-cu121) '
+++ export CONDA_PREFIX_3=/fsx/nouamane/miniconda
+++ CONDA_PREFIX_3=/fsx/nouamane/miniconda
+++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ export _CE_M=
+++ _CE_M=
+++ export _CE_CONDA=
+++ _CE_CONDA=
+++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh
++++ test -n ''
++++ xml_catalog_files_libxml2=
++++ XML_CATALOG_FILES=
++++ conda_catalog_files=
++++ ifs_libxml2=' 	
+'
++++ IFS=' '
++++ rem=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ for pre in ${rem}
++++ test '' = /fsx/nouamane/miniconda/envs/2-1-cu121
++++ conda_catalog_files=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ rem=
++++ IFS=' 	
+'
++++ conda_catalog_files='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ export 'XML_CATALOG_FILES=file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ XML_CATALOG_FILES='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ unset conda_catalog_files ifs_libxml2 rem
++ __conda_hashr
++ '[' -n '' ']'
++ '[' -n '' ']'
++ hash -r
++ export PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++ PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ scontrol show hostnames ip-26-0-169-207
++ export NODELIST=ip-26-0-169-207
++ NODELIST=ip-26-0-169-207
+++ scontrol show hostnames ip-26-0-169-207
+++ head -n1
++ export MASTER_NODE=ip-26-0-169-207
++ MASTER_NODE=ip-26-0-169-207
++ export MASTER_PORT=12356
++ MASTER_PORT=12356
++ export NNODES=1
++ NNODES=1
++ export GPUS_PER_NODE=8
++ GPUS_PER_NODE=8
++ export WORLD_SIZE=8
++ WORLD_SIZE=8
++ export CUDA_DEVICE_MAX_CONNECTIONS=1
++ CUDA_DEVICE_MAX_CONNECTIONS=1
++ export NANOTRON_BENCHMARK=1
++ NANOTRON_BENCHMARK=1
++ echo 'Master node: ip-26-0-169-207'
+Master node: ip-26-0-169-207
++ echo 'All nodes: ip-26-0-169-207'
+All nodes: ip-26-0-169-207
++ echo 'World size: 8'
+World size: 8
++ srun torchrun --nnodes=1 --nproc_per_node=8 --rdzv_id=13212436 --rdzv_backend=c10d --rdzv_endpoint=ip-26-0-169-207:12356 run_train.py --config-file benchmark/configs/config_1.34G_dp8_tp1_pp1_acc8_mbs32_seq2048_zero1_tpmodeRED_vocab131k.yaml
+[2024-12-09 19:08:27,143] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-09 19:08:27,144] torch.distributed.run: [WARNING] 
+[2024-12-09 19:08:27,144] torch.distributed.run: [WARNING] *****************************************
+[2024-12-09 19:08:27,144] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-09 19:08:27,144] torch.distributed.run: [WARNING] *****************************************
+12/09/2024 19:09:01 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]: Config:
+12/09/2024 19:09:01 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]: Config(general=GeneralArgs(project='debug',
+12/09/2024 19:09:01 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                            run='1.34G_dp8_tp1_pp1_acc8_mbs32_seq2048_zero1_tpmodeRED_vocab131k',
+12/09/2024 19:09:01 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                            seed=42,
+12/09/2024 19:09:01 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                            step=None,
+12/09/2024 19:09:01 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                            consumed_train_samples=None,
+12/09/2024 19:09:01 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                            benchmark_csv_path=PosixPath('benchmark/results/bench_final.csv'),
+12/09/2024 19:09:01 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                            ignore_sanity_checks=True),
+12/09/2024 19:09:01 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:        parallelism=ParallelismArgs(dp=8,
+12/09/2024 19:09:01 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                    pp=1,
+12/09/2024 19:09:01 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                    tp=1,
+12/09/2024 19:09:01 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                    pp_engine=<nanotron.parallel.pipeline_parallel.engine.OneForwardOneBackwardPipelineEngine object at 0x7f4b7ac8cc70>,
+12/09/2024 19:09:01 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                    tp_mode=<TensorParallelLinearMode.REDUCE_SCATTER: 2>,
+12/09/2024 19:09:01 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                    tp_linear_async_communication=True,
+12/09/2024 19:09:01 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                    recompute_layer=False,
+12/09/2024 19:09:01 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                    tp_recompute_allgather=True,
+12/09/2024 19:09:01 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                    expert_parallel_size=1),
+12/09/2024 19:09:01 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:        model=ModelArgs(model_config=LlamaConfig(bos_token_id=0,
+12/09/2024 19:09:01 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                                 eos_token_id=0,
+12/09/2024 19:09:01 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                                 hidden_act='silu',
+12/09/2024 19:09:01 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                                 hidden_size=2048,
+12/09/2024 19:09:01 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                                 initializer_range=0.02,
+12/09/2024 19:09:01 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                                 intermediate_size=8192,
+12/09/2024 19:09:01 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                                 is_llama_config=True,
+12/09/2024 19:09:01 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                                 max_position_embeddings=2048,
+12/09/2024 19:09:01 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                                 num_attention_heads=32,
+12/09/2024 19:09:01 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                                 num_hidden_layers=16,
+12/09/2024 19:09:01 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                                 num_key_value_heads=8,
+12/09/2024 19:09:01 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                                 pad_token_id=None,
+12/09/2024 19:09:01 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                                 pretraining_tp=1,
+12/09/2024 19:09:01 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                                 rms_norm_eps=1e-05,
+12/09/2024 19:09:01 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                                 rope_scaling=None,
+12/09/2024 19:09:01 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                                 rope_theta=10000.0,
+12/09/2024 19:09:01 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                                 rope_interleaved=False,
+12/09/2024 19:09:01 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                                 tie_word_embeddings=True,
+12/09/2024 19:09:01 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                                 use_cache=True,
+12/09/2024 19:09:01 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                                 vocab_size=131072),
+12/09/2024 19:09:01 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                        init_method=RandomInit(std=0.02),
+12/09/2024 19:09:01 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                        dtype=torch.bfloat16,
+12/09/2024 19:09:01 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                        make_vocab_size_divisible_by=1,
+12/09/2024 19:09:01 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                        ddp_bucket_cap_mb=25),
+12/09/2024 19:09:01 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:        tokenizer=TokenizerArgs(tokenizer_name_or_path='robot-test/dummy-tokenizer-wordlevel',
+12/09/2024 19:09:01 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                tokenizer_revision=None,
+12/09/2024 19:09:01 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                tokenizer_max_length=None),
+12/09/2024 19:09:01 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:        checkpoints=CheckpointsArgs(checkpoints_path=PosixPath('checkpoints'),
+12/09/2024 19:09:01 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                    checkpoint_interval=10000,
+12/09/2024 19:09:01 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                    save_initial_state=False,
+12/09/2024 19:09:01 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                    save_final_state=False,
+12/09/2024 19:09:01 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                    resume_checkpoint_path=None,
+12/09/2024 19:09:01 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                    checkpoints_path_is_shared_file_system=False),
+12/09/2024 19:09:01 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:        logging=LoggingArgs(log_level='info',
+12/09/2024 19:09:01 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                            log_level_replica='info',
+12/09/2024 19:09:01 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                            iteration_step_info_interval=1),
+12/09/2024 19:09:01 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:        tokens=TokensArgs(sequence_length=2048,
+12/09/2024 19:09:01 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                          train_steps=100,
+12/09/2024 19:09:01 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                          micro_batch_size=32,
+12/09/2024 19:09:01 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                          batch_accumulation_per_replica=8,
+12/09/2024 19:09:01 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                          val_check_interval=100,
+12/09/2024 19:09:01 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                          limit_val_batches=0,
+12/09/2024 19:09:01 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                          limit_test_batches=0),
+12/09/2024 19:09:01 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:        optimizer=OptimizerArgs(optimizer_factory=AdamWOptimizerArgs(adam_eps=1e-08,
+12/09/2024 19:09:01 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                                                     adam_beta1=0.9,
+12/09/2024 19:09:01 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                                                     adam_beta2=0.95,
+12/09/2024 19:09:01 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                                                     torch_adam_is_fused=True,
+12/09/2024 19:09:01 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                                                     name='adamW'),
+12/09/2024 19:09:01 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                zero_stage=1,
+12/09/2024 19:09:01 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                weight_decay=0.01,
+12/09/2024 19:09:01 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                clip_grad=1.0,
+12/09/2024 19:09:01 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                accumulate_grad_in_fp32=True,
+12/09/2024 19:09:01 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                learning_rate_scheduler=LRSchedulerArgs(learning_rate=0.0003,
+12/09/2024 19:09:01 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                                                        lr_warmup_steps=2,
+12/09/2024 19:09:01 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                                                        lr_warmup_style='linear',
+12/09/2024 19:09:01 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                                                        lr_decay_style='cosine',
+12/09/2024 19:09:01 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                                                        lr_decay_steps=13,
+12/09/2024 19:09:01 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                                                        lr_decay_starting_step=None,
+12/09/2024 19:09:01 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                                                        min_decay_lr=1e-05)),
+12/09/2024 19:09:01 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:        data_stages=[DatasetStageArgs(name='Stable Training Stage',
+12/09/2024 19:09:01 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                      start_training_step=1,
+12/09/2024 19:09:01 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                      data=DataArgs(dataset=None,
+12/09/2024 19:09:01 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                                    seed=42,
+12/09/2024 19:09:01 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                                    num_loading_workers=1))],
+12/09/2024 19:09:01 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:        profiler=None,
+12/09/2024 19:09:01 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:        lighteval=None,
+12/09/2024 19:09:01 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:        s3_upload=None)
+12/09/2024 19:09:01 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]: Model Config:
+12/09/2024 19:09:01 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]: LlamaConfig(bos_token_id=0,
+12/09/2024 19:09:01 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:             eos_token_id=0,
+12/09/2024 19:09:01 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:             hidden_act='silu',
+12/09/2024 19:09:01 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:             hidden_size=2048,
+12/09/2024 19:09:01 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:             initializer_range=0.02,
+12/09/2024 19:09:01 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:             intermediate_size=8192,
+12/09/2024 19:09:01 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:             is_llama_config=True,
+12/09/2024 19:09:01 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:             max_position_embeddings=2048,
+12/09/2024 19:09:01 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:             num_attention_heads=32,
+12/09/2024 19:09:01 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:             num_hidden_layers=16,
+12/09/2024 19:09:01 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:             num_key_value_heads=8,
+12/09/2024 19:09:01 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:             pad_token_id=None,
+12/09/2024 19:09:01 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:             pretraining_tp=1,
+12/09/2024 19:09:01 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:             rms_norm_eps=1e-05,
+12/09/2024 19:09:01 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:             rope_scaling=None,
+12/09/2024 19:09:01 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:             rope_theta=10000.0,
+12/09/2024 19:09:01 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:             rope_interleaved=False,
+12/09/2024 19:09:01 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:             tie_word_embeddings=True,
+12/09/2024 19:09:01 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:             use_cache=True,
+12/09/2024 19:09:01 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:             vocab_size=131072)
+12/09/2024 19:09:01 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]: Building model..
+12/09/2024 19:09:01 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]: Initialize RoPE Theta = 10000.0
+12/09/2024 19:09:01 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]: Setting PP block ranks...
+12/09/2024 19:09:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]: Total number of parameters: 1.24G (2368.13MiB)
+12/09/2024 19:09:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]: Local number of parameters: 1.24G (2368.13MiB)
+12/09/2024 19:09:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]: [After model building] Memory usage: 2368.15MiB. Peak allocated: 5632.00MiB Peak reserved: 18946.00MiB
+12/09/2024 19:09:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]: No checkpoint path provided.
+12/09/2024 19:09:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]: Parametrizing model parameters using StandardParametrizator
+12/09/2024 19:09:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]: [Optimizer Building] Using LearningRateForSP as learning rate
+12/09/2024 19:09:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]: [ZeRO sharding] Size of optimizer params per rank:
+12/09/2024 19:09:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]: [ZeRO sharding] DP Rank 0 has 155M out of 1.24G (12.50%) params' optimizer states
+12/09/2024 19:09:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]: [ZeRO sharding] DP Rank 1 has 155M out of 1.24G (12.50%) params' optimizer states
+12/09/2024 19:09:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]: [ZeRO sharding] DP Rank 2 has 155M out of 1.24G (12.50%) params' optimizer states
+12/09/2024 19:09:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]: [ZeRO sharding] DP Rank 3 has 155M out of 1.24G (12.50%) params' optimizer states
+12/09/2024 19:09:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]: [ZeRO sharding] DP Rank 4 has 155M out of 1.24G (12.50%) params' optimizer states
+12/09/2024 19:09:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]: [ZeRO sharding] DP Rank 5 has 155M out of 1.24G (12.50%) params' optimizer states
+12/09/2024 19:09:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]: [ZeRO sharding] DP Rank 6 has 155M out of 1.24G (12.50%) params' optimizer states
+12/09/2024 19:09:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]: [ZeRO sharding] DP Rank 7 has 155M out of 1.24G (12.50%) params' optimizer states
+12/09/2024 19:09:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]: [Training Plan] Stage Stable Training Stage has 99 remaining training steps and has consumed 0 samples
+12/09/2024 19:09:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]: Using dummy data generator
+12/09/2024 19:09:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]: [Training Plan] There are 1 training stages 
+12/09/2024 19:09:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]: [Stage Stable Training Stage] start from step 1 
+12/09/2024 19:09:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]: 
+12/09/2024 19:09:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]: [Start training] datetime: 2024-12-09 19:09:03.897743 | mbs: 32 | grad_accum: 8 | global_batch_size: 2048 | sequence_length: 2048 | train_steps: 100 | start_iteration_step: 0 | consumed_train_samples: 0
+wandb: WARNING Path /fsx/nouamane/.cache/wandb/wandb/ wasn't writable, using system temp directory.
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 725, in _core_forward
+    output = self.attn(hidden_states=hidden_states, sequence_mask=sequence_mask)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 679, in forward
+    attention_output = self.attention(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/utils.py", line 115, in wrapper
+    return func(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 286, in forward
+    attn_output = flash_attn_varlen_func(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/flash_attn/flash_attn_interface.py", line 1059, in flash_attn_varlen_func
+    return FlashAttnVarlenFunc.apply(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/flash_attn/flash_attn_interface.py", line 576, in forward
+    out, q, k, v, out_padded, softmax_lse, S_dmask, rng_state = _flash_attn_varlen_forward(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/flash_attn/flash_attn_interface.py", line 85, in _flash_attn_varlen_forward
+    out, q, k, v, out_padded, softmax_lse, S_dmask, rng_state = flash_attn_cuda.varlen_fwd(
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 256.00 MiB. GPU 3 has a total capacty of 79.33 GiB of which 117.94 MiB is free. Including non-PyTorch memory, this process has 79.20 GiB memory in use. Of the allocated memory 70.77 GiB is allocated by PyTorch, and 369.66 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 244, in forward
+    merged_states = self.gate_up_proj(hidden_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 89, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    return column_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 441, in column_linear
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    return _ColumnLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode, tp_recompute_allgather)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+return forward_call(*args, **kwargs)
+    trainer.train(dataloader)  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/utils.py", line 40, in wrapper
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+        outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)return func(*args, **kwargs)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 140, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return F.linear(gathered_tensor, weight, bias)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 2.00 GiB. GPU 7 has a total capacty of 79.33 GiB of which 69.94 MiB is free. Including non-PyTorch memory, this process has 79.25 GiB memory in use. Of the allocated memory 71.40 GiB is allocated by PyTorch, and 489.41 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+        return self._call_impl(*args, **kwargs)output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 725, in _core_forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = self.attn(hidden_states=hidden_states, sequence_mask=sequence_mask)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    sharded_logits = self.model(
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)output = self.pp_block(**new_kwargs)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 679, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    attention_output = self.attention(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 725, in _core_forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/utils.py", line 115, in wrapper
+        output = self.attn(hidden_states=hidden_states, sequence_mask=sequence_mask)return forward_call(*args, **kwargs)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    return func(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 286, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    attn_output = flash_attn_varlen_func(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/flash_attn/flash_attn_interface.py", line 1059, in flash_attn_varlen_func
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 679, in forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = model(**micro_batch)        
+return FlashAttnVarlenFunc.apply(
+return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    attention_output = self.attention(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/flash_attn/flash_attn_interface.py", line 576, in forward
+    return self._call_impl(*args, **kwargs)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/utils.py", line 115, in wrapper
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 725, in _core_forward
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.attn(hidden_states=hidden_states, sequence_mask=sequence_mask)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    out, q, k, v, out_padded, softmax_lse, S_dmask, rng_state = _flash_attn_varlen_forward(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/flash_attn/flash_attn_interface.py", line 85, in _flash_attn_varlen_forward
+    return func(*args, **kwargs)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 286, in forward
+sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    out, q, k, v, out_padded, softmax_lse, S_dmask, rng_state = flash_attn_cuda.varlen_fwd(
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 256.00 MiB. GPU 2 has a total capacty of 79.33 GiB of which 117.94 MiB is free. Including non-PyTorch memory, this process has 79.20 GiB memory in use. Of the allocated memory 70.77 GiB is allocated by PyTorch, and 369.66 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    attn_output = flash_attn_varlen_func(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/flash_attn/flash_attn_interface.py", line 1059, in flash_attn_varlen_func
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+        return FlashAttnVarlenFunc.apply(return forward_call(*args, **kwargs)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 679, in forward
+        return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]return self._call_impl(*args, **kwargs)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    attention_output = self.attention(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/flash_attn/flash_attn_interface.py", line 576, in forward
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/utils.py", line 115, in wrapper
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    out, q, k, v, out_padded, softmax_lse, S_dmask, rng_state = _flash_attn_varlen_forward(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/flash_attn/flash_attn_interface.py", line 85, in _flash_attn_varlen_forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+        return func(*args, **kwargs)hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)    
+
+out, q, k, v, out_padded, softmax_lse, S_dmask, rng_state = flash_attn_cuda.varlen_fwd(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 286, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 725, in _core_forward
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 256.00 MiB. GPU 5 has a total capacty of 79.33 GiB of which 117.94 MiB is free. Including non-PyTorch memory, this process has 79.20 GiB memory in use. Of the allocated memory 70.77 GiB is allocated by PyTorch, and 369.66 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF    
+output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    attn_output = flash_attn_varlen_func(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/flash_attn/flash_attn_interface.py", line 1059, in flash_attn_varlen_func
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = self.attn(hidden_states=hidden_states, sequence_mask=sequence_mask)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return FlashAttnVarlenFunc.apply(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/flash_attn/flash_attn_interface.py", line 576, in forward
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 725, in _core_forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 679, in forward
+    out, q, k, v, out_padded, softmax_lse, S_dmask, rng_state = _flash_attn_varlen_forward(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/flash_attn/flash_attn_interface.py", line 85, in _flash_attn_varlen_forward
+    output = self.attn(hidden_states=hidden_states, sequence_mask=sequence_mask)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    out, q, k, v, out_padded, softmax_lse, S_dmask, rng_state = flash_attn_cuda.varlen_fwd(
+    torch.cuda    .attention_output = self.attention(return self._call_impl(*args, **kwargs)
+OutOfMemoryError
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+:   File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+CUDA out of memory. Tried to allocate 256.00 MiB. GPU 4 has a total capacty of 79.33 GiB of which 117.94 MiB is free. Including non-PyTorch memory, this process has 79.20 GiB memory in use. Of the allocated memory 70.77 GiB is allocated by PyTorch, and 369.66 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 679, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/utils.py", line 115, in wrapper
+    attention_output = self.attention(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return func(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 286, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    attn_output = flash_attn_varlen_func(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/flash_attn/flash_attn_interface.py", line 1059, in flash_attn_varlen_func
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/utils.py", line 115, in wrapper
+    return func(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 286, in forward
+    attn_output = flash_attn_varlen_func(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/flash_attn/flash_attn_interface.py", line 1059, in flash_attn_varlen_func
+    return FlashAttnVarlenFunc.apply(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/flash_attn/flash_attn_interface.py", line 576, in forward
+    out, q, k, v, out_padded, softmax_lse, S_dmask, rng_state = _flash_attn_varlen_forward(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/flash_attn/flash_attn_interface.py", line 85, in _flash_attn_varlen_forward
+    out, q, k, v, out_padded, softmax_lse, S_dmask, rng_state = flash_attn_cuda.varlen_fwd(
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 256.00 MiB. GPU 6 has a total capacty of 79.33 GiB of which 117.94 MiB is free. Including non-PyTorch memory, this process has 79.20 GiB memory in use. Of the allocated memory 70.77 GiB is allocated by PyTorch, and 369.66 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return FlashAttnVarlenFunc.apply(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/flash_attn/flash_attn_interface.py", line 576, in forward
+    out, q, k, v, out_padded, softmax_lse, S_dmask, rng_state = _flash_attn_varlen_forward(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/flash_attn/flash_attn_interface.py", line 85, in _flash_attn_varlen_forward
+    out, q, k, v, out_padded, softmax_lse, S_dmask, rng_state = flash_attn_cuda.varlen_fwd(
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 256.00 MiB. GPU 1 has a total capacty of 79.33 GiB of which 117.94 MiB is free. Including non-PyTorch memory, this process has 79.20 GiB memory in use. Of the allocated memory 70.77 GiB is allocated by PyTorch, and 369.66 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+wandb: WARNING Path /fsx/nouamane/.cache/wandb/wandb/ wasn't writable, using system temp directory
+[2024-12-09 19:09:07,444] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1811525 closing signal SIGTERM
+[2024-12-09 19:09:07,959] torch.distributed.elastic.multiprocessing.api: [ERROR] failed (exitcode: 1) local_rank: 1 (pid: 1811526) of binary: /fsx/nouamane/miniconda/envs/2-1-cu121/bin/python
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 33, in <module>
+    sys.exit(load_entry_point('torch==2.1.1', 'console_scripts', 'torchrun')())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 264, in launch_agent
+    raise ChildFailedError(
+torch.distributed.elastic.multiprocessing.errors.ChildFailedError: 
+============================================================
+run_train.py FAILED
+------------------------------------------------------------
+Failures:
+[1]:
+  time      : 2024-12-09_19:09:07
+  host      : ip-26-0-169-207.ec2.internal
+  rank      : 2 (local_rank: 2)
+  exitcode  : 1 (pid: 1811527)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[2]:
+  time      : 2024-12-09_19:09:07
+  host      : ip-26-0-169-207.ec2.internal
+  rank      : 3 (local_rank: 3)
+  exitcode  : 1 (pid: 1811528)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[3]:
+  time      : 2024-12-09_19:09:07
+  host      : ip-26-0-169-207.ec2.internal
+  rank      : 4 (local_rank: 4)
+  exitcode  : 1 (pid: 1811529)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[4]:
+  time      : 2024-12-09_19:09:07
+  host      : ip-26-0-169-207.ec2.internal
+  rank      : 5 (local_rank: 5)
+  exitcode  : 1 (pid: 1811530)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[5]:
+  time      : 2024-12-09_19:09:07
+  host      : ip-26-0-169-207.ec2.internal
+  rank      : 6 (local_rank: 6)
+  exitcode  : 1 (pid: 1811531)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[6]:
+  time      : 2024-12-09_19:09:07
+  host      : ip-26-0-169-207.ec2.internal
+  rank      : 7 (local_rank: 7)
+  exitcode  : 1 (pid: 1811532)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+------------------------------------------------------------
+Root Cause (first observed failure):
+[0]:
+  time      : 2024-12-09_19:09:07
+  host      : ip-26-0-169-207.ec2.internal
+  rank      : 1 (local_rank: 1)
+  exitcode  : 1 (pid: 1811526)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+============================================================
+srun: error: ip-26-0-169-207: task 0: Exited with exit code 1
diff --git a/logs/13264474-bench_1.14G_dp32_tp1_pp1_acc32_mbs2_seq2048_zero1_tpmodeRED_vocab32k.out b/logs/13264474-bench_1.14G_dp32_tp1_pp1_acc32_mbs2_seq2048_zero1_tpmodeRED_vocab32k.out
new file mode 100644
index 0000000000000000000000000000000000000000..2315224b7ac5fb1b7c00919fb32af920421a6d77
--- /dev/null
+++ b/logs/13264474-bench_1.14G_dp32_tp1_pp1_acc32_mbs2_seq2048_zero1_tpmodeRED_vocab32k.out
@@ -0,0 +1,762 @@
++ source /etc/profile.d/modules.sh
+++ . /usr/share/modules/init/bash
++++ unset _mlshdbg
++++ '[' 0 = 1 ']'
++++ unset _mlre _mlIFS
++++ '[' -n x ']'
++++ _mlIFS=' 	
+'
++++ IFS=' '
++++ '[' -n '' ']'
+++++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash autoinit
++++ _mlcode='module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+MODULEPATH_modshare=/etc/environment-modules/modules:1:/usr/share/modules/\$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1; export MODULEPATH_modshare;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+LOADEDMODULES=; export LOADEDMODULES;
+MODULEPATH=/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/\$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles; export MODULEPATH;
+test 0;'
++++ _mlret=0
++++ '[' -n x ']'
++++ IFS=' 	
+'
++++ unset _mlIFS
++++ unset _mlre _mlv _mlrv
++++ '[' 0 -eq 0 ']'
++++ eval 'module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+MODULEPATH_modshare=/etc/environment-modules/modules:1:/usr/share/modules/\$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1; export MODULEPATH_modshare;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+LOADEDMODULES=; export LOADEDMODULES;
+MODULEPATH=/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/\$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles; export MODULEPATH;
+test 0;'
+++++ MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl
+++++ export MODULES_CMD
+++++ ENV=/usr/share/modules/init/profile.sh
+++++ export ENV
+++++ MODULEPATH_modshare='/etc/environment-modules/modules:1:/usr/share/modules/$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1'
+++++ export MODULEPATH_modshare
+++++ BASH_ENV=/usr/share/modules/init/bash
+++++ export BASH_ENV
+++++ MODULESHOME=/usr/share/modules
+++++ export MODULESHOME
+++++ LOADEDMODULES=
+++++ export LOADEDMODULES
+++++ MODULEPATH='/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles'
+++++ export MODULEPATH
+++++ test 0
++++ '[' 0 = 1 ']'
++++ '[' -t 2 ']'
++++ export -f module
++++ export -f switchml
++++ '[' 5 -ge 3 ']'
++++ [[ ehxB =~ i ]]
++++ [[ ! :/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin: =~ :/usr/bin: ]]
++++ '[' '!' -n '' ']'
++++ MANPATH=:
++++ export MANPATH
+++++ manpath
++++ [[ ! :/admin/home/nouamane/.local/share/man:/fsx/nouamane/miniconda/envs/2-1-cu121/man:/fsx/nouamane/miniconda/envs/2-1-cu121/share/man:/fsx/nouamane/miniconda/man:/fsx/nouamane/miniconda/share/man:/opt/amazon/openmpi/share/man:/opt/amazon/efa/share/man:/opt/slurm/share/man:/usr/local/man:/usr/local/share/man:/usr/share/man:/admin/home/nouamane/.fzf/man:: =~ :/usr/share/man: ]]
++++ unset _mlcode _mlret
++++ '[' -n '' ']'
++ module load cuda/12.1
++ unset _mlshdbg
++ '[' 0 = 1 ']'
++ unset _mlre _mlIFS
++ '[' -n x ']'
++ _mlIFS=' 	
+'
++ IFS=' '
++ '[' -n '' ']'
+++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash load cuda/12.1
++ eval 'CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include;' export 'CPATH;
+LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:;' export 'LD_LIBRARY_PATH;
+FI_EFA_FORK_SAFE=1;' export 'FI_EFA_FORK_SAFE;
+MANPATH=/usr/local/cuda-12.1/share/man::;' export 'MANPATH;
+LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64;' export 'LIBRARY_PATH;
+_LMFILES_=/usr/share/modules/modulefiles/cuda/12.1;' export '_LMFILES_;
+LOADEDMODULES=cuda/12.1;' export 'LOADEDMODULES;
+MPI_PATH=/opt/amazon/openmpi;' export 'MPI_PATH;
+NCCL_HOME_modshare=/opt/nccl/build:1;' export 'NCCL_HOME_modshare;
+NCCL_PROTO=simple;' export 'NCCL_PROTO;
+MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1;' export 'MANPATH_modshare;
+LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1;' export 'LIBRARY_PATH_modshare;
+NCCL_SOCKET_IFNAME=enp;' export 'NCCL_SOCKET_IFNAME;
+AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl;' export 'AWS_OFI_NCCL_HOME;
+NCCL_HOME=/opt/nccl/build;' export 'NCCL_HOME;
+FI_PROVIDER=efa;' export 'FI_PROVIDER;
+AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1;' export 'AWS_OFI_NCCL_HOME_modshare;
+CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1;' export 'CPATH_modshare;
+LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2::1:/opt/nccl/build/lib:1:/opt/aws-ofi-nccl/lib:2:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/efa/lib:1;' export 'LD_LIBRARY_PATH_modshare;
+FI_EFA_ENABLE_SHM_TRANSFER=1;' export 'FI_EFA_ENABLE_SHM_TRANSFER;
+_LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1;' export '_LMFILES__modshare;
+LOADEDMODULES_modshare=cuda/12.1:1;' export 'LOADEDMODULES_modshare;
+MPI_PATH_modshare=/opt/amazon/openmpi:1;' export 'MPI_PATH_modshare;
+PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin;' export 'PATH;
+CUDA_HOME=/usr/local/cuda-12.1;' export 'CUDA_HOME;
+PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/usr/local/cuda-12.1/include:1:/opt/amazon/efa/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/usr/sbin:1:/fsx/nouamane/miniconda/bin:1:/opt/slurm/bin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:2:/opt/amazon/openmpi/bin:1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1;' export 'PATH_modshare;
+test' '0;'
+++ CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include
+++ export CPATH
+++ LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:
+++ export LD_LIBRARY_PATH
+++ FI_EFA_FORK_SAFE=1
+++ export FI_EFA_FORK_SAFE
+++ MANPATH=/usr/local/cuda-12.1/share/man::
+++ export MANPATH
+++ LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64
+++ export LIBRARY_PATH
+++ _LMFILES_=/usr/share/modules/modulefiles/cuda/12.1
+++ export _LMFILES_
+++ LOADEDMODULES=cuda/12.1
+++ export LOADEDMODULES
+++ MPI_PATH=/opt/amazon/openmpi
+++ export MPI_PATH
+++ NCCL_HOME_modshare=/opt/nccl/build:1
+++ export NCCL_HOME_modshare
+++ NCCL_PROTO=simple
+++ export NCCL_PROTO
+++ MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1
+++ export MANPATH_modshare
+++ LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1
+++ export LIBRARY_PATH_modshare
+++ NCCL_SOCKET_IFNAME=enp
+++ export NCCL_SOCKET_IFNAME
+++ AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl
+++ export AWS_OFI_NCCL_HOME
+++ NCCL_HOME=/opt/nccl/build
+++ export NCCL_HOME
+++ FI_PROVIDER=efa
+++ export FI_PROVIDER
+++ AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1
+++ export AWS_OFI_NCCL_HOME_modshare
+++ CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1
+++ export CPATH_modshare
+++ LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2::1:/opt/nccl/build/lib:1:/opt/aws-ofi-nccl/lib:2:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/efa/lib:1
+++ export LD_LIBRARY_PATH_modshare
+++ FI_EFA_ENABLE_SHM_TRANSFER=1
+++ export FI_EFA_ENABLE_SHM_TRANSFER
+++ _LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1
+++ export _LMFILES__modshare
+++ LOADEDMODULES_modshare=cuda/12.1:1
+++ export LOADEDMODULES_modshare
+++ MPI_PATH_modshare=/opt/amazon/openmpi:1
+++ export MPI_PATH_modshare
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ export PATH
+++ CUDA_HOME=/usr/local/cuda-12.1
+++ export CUDA_HOME
+++ PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/usr/local/cuda-12.1/include:1:/opt/amazon/efa/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/usr/sbin:1:/fsx/nouamane/miniconda/bin:1:/opt/slurm/bin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:2:/opt/amazon/openmpi/bin:1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1
+++ export PATH_modshare
+++ test 0
++ _mlstatus=0
++ '[' -n x ']'
++ IFS=' 	
+'
++ unset _mlre _mlv _mlrv _mlIFS
++ '[' -n '' ']'
++ unset _mlshdbg
++ return 0
++ source /fsx/nouamane/miniconda/bin/activate
+++ _CONDA_ROOT=/fsx/nouamane/miniconda
+++ . /fsx/nouamane/miniconda/etc/profile.d/conda.sh
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ '[' -z x ']'
+++ conda activate
+++ local cmd=activate
+++ case "$cmd" in
+++ __conda_activate activate
+++ '[' -n '' ']'
+++ local ask_conda
++++ PS1=
++++ __conda_exe shell.posix activate
++++ /fsx/nouamane/miniconda/bin/conda shell.posix activate
+++ ask_conda='. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_2='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\'''
+++ eval '. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_2='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\'''
++++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh
+++++ test -n ''
+++++ unset XML_CATALOG_FILES
+++++ unset xml_catalog_files_libxml2
++++ PS1='(base) '
++++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++++ export CONDA_PREFIX=/fsx/nouamane/miniconda
++++ CONDA_PREFIX=/fsx/nouamane/miniconda
++++ export CONDA_SHLVL=3
++++ CONDA_SHLVL=3
++++ export CONDA_DEFAULT_ENV=base
++++ CONDA_DEFAULT_ENV=base
++++ export 'CONDA_PROMPT_MODIFIER=(base) '
++++ CONDA_PROMPT_MODIFIER='(base) '
++++ export CONDA_PREFIX_2=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ CONDA_PREFIX_2=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ __conda_hashr
+++ '[' -n '' ']'
+++ '[' -n '' ']'
+++ hash -r
++ conda activate 2-1-cu121
++ local cmd=activate
++ case "$cmd" in
++ __conda_activate activate 2-1-cu121
++ '[' -n '' ']'
++ local ask_conda
+++ PS1='(base) '
+++ __conda_exe shell.posix activate 2-1-cu121
+++ /fsx/nouamane/miniconda/bin/conda shell.posix activate 2-1-cu121
++ ask_conda='PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''4'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_3='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
++ eval 'PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''4'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_3='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
+++ PS1='(2-1-cu121) '
+++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ export CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ export CONDA_SHLVL=4
+++ CONDA_SHLVL=4
+++ export CONDA_DEFAULT_ENV=2-1-cu121
+++ CONDA_DEFAULT_ENV=2-1-cu121
+++ export 'CONDA_PROMPT_MODIFIER=(2-1-cu121) '
+++ CONDA_PROMPT_MODIFIER='(2-1-cu121) '
+++ export CONDA_PREFIX_3=/fsx/nouamane/miniconda
+++ CONDA_PREFIX_3=/fsx/nouamane/miniconda
+++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ export _CE_M=
+++ _CE_M=
+++ export _CE_CONDA=
+++ _CE_CONDA=
+++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh
++++ test -n ''
++++ xml_catalog_files_libxml2=
++++ XML_CATALOG_FILES=
++++ conda_catalog_files=
++++ ifs_libxml2=' 	
+'
++++ IFS=' '
++++ rem=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ for pre in ${rem}
++++ test '' = /fsx/nouamane/miniconda/envs/2-1-cu121
++++ conda_catalog_files=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ rem=
++++ IFS=' 	
+'
++++ conda_catalog_files='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ export 'XML_CATALOG_FILES=file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ XML_CATALOG_FILES='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ unset conda_catalog_files ifs_libxml2 rem
++ __conda_hashr
++ '[' -n '' ']'
++ '[' -n '' ']'
++ hash -r
++ export PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++ PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ scontrol show hostnames 'ip-26-0-166-[36,214],ip-26-0-169-[86,132]'
++ export 'NODELIST=ip-26-0-166-36
+ip-26-0-166-214
+ip-26-0-169-86
+ip-26-0-169-132'
++ NODELIST='ip-26-0-166-36
+ip-26-0-166-214
+ip-26-0-169-86
+ip-26-0-169-132'
+++ head -n1
+++ scontrol show hostnames 'ip-26-0-166-[36,214],ip-26-0-169-[86,132]'
++ export MASTER_NODE=ip-26-0-166-36
++ MASTER_NODE=ip-26-0-166-36
++ export MASTER_PORT=12356
++ MASTER_PORT=12356
++ export NNODES=4
++ NNODES=4
++ export GPUS_PER_NODE=8
++ GPUS_PER_NODE=8
++ export WORLD_SIZE=32
++ WORLD_SIZE=32
++ export CUDA_DEVICE_MAX_CONNECTIONS=1
++ CUDA_DEVICE_MAX_CONNECTIONS=1
++ export NANOTRON_BENCHMARK=1
++ NANOTRON_BENCHMARK=1
++ echo 'Master node: ip-26-0-166-36'
+Master node: ip-26-0-166-36
++ echo 'All nodes: ip-26-0-166-36
+ip-26-0-166-214
+ip-26-0-169-86
+ip-26-0-169-132'
+All nodes: ip-26-0-166-36
+ip-26-0-166-214
+ip-26-0-169-86
+ip-26-0-169-132
++ echo 'World size: 32'
+World size: 32
++ srun torchrun --nnodes=4 --nproc_per_node=8 --rdzv_id=13264474 --rdzv_backend=c10d --rdzv_endpoint=ip-26-0-166-36:12356 run_train.py --config-file benchmark/configs/config_1.14G_dp32_tp1_pp1_acc32_mbs2_seq2048_zero1_tpmodeRED_vocab32k.yaml
+[2024-12-09 21:36:49,573] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-09 21:36:49,581] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-09 21:36:49,591] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-09 21:36:49,573] torch.distributed.run: [WARNING] 
+[2024-12-09 21:36:49,573] torch.distributed.run: [WARNING] *****************************************
+[2024-12-09 21:36:49,573] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-09 21:36:49,573] torch.distributed.run: [WARNING] *****************************************
+[2024-12-09 21:36:49,581] torch.distributed.run: [WARNING] 
+[2024-12-09 21:36:49,581] torch.distributed.run: [WARNING] *****************************************
+[2024-12-09 21:36:49,581] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-09 21:36:49,581] torch.distributed.run: [WARNING] *****************************************
+[2024-12-09 21:36:49,591] torch.distributed.run: [WARNING] 
+[2024-12-09 21:36:49,591] torch.distributed.run: [WARNING] *****************************************
+[2024-12-09 21:36:49,591] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-09 21:36:49,591] torch.distributed.run: [WARNING] *****************************************
+[2024-12-09 21:36:53,828] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-09 21:36:53,828] torch.distributed.run: [WARNING] 
+[2024-12-09 21:36:53,828] torch.distributed.run: [WARNING] *****************************************
+[2024-12-09 21:36:53,828] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-09 21:36:53,828] torch.distributed.run: [WARNING] *****************************************
+12/09/2024 21:37:38 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]: Config:
+12/09/2024 21:37:38 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]: Config(general=GeneralArgs(project='debug',
+12/09/2024 21:37:38 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]:                            run='1.14G_dp32_tp1_pp1_acc32_mbs2_seq2048_zero1_tpmodeRED_vocab32k',
+12/09/2024 21:37:38 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]:                            seed=42,
+12/09/2024 21:37:38 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]:                            step=None,
+12/09/2024 21:37:38 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]:                            consumed_train_samples=None,
+12/09/2024 21:37:38 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]:                            benchmark_csv_path=PosixPath('benchmark/results/bench_final.csv'),
+12/09/2024 21:37:38 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]:                            ignore_sanity_checks=True),
+12/09/2024 21:37:38 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]:        parallelism=ParallelismArgs(dp=32,
+12/09/2024 21:37:38 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]:                                    pp=1,
+12/09/2024 21:37:38 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]:                                    tp=1,
+12/09/2024 21:37:38 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]:                                    pp_engine=<nanotron.parallel.pipeline_parallel.engine.OneForwardOneBackwardPipelineEngine object at 0x7f99b3ae4af0>,
+12/09/2024 21:37:38 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]:                                    tp_mode=<TensorParallelLinearMode.REDUCE_SCATTER: 2>,
+12/09/2024 21:37:38 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]:                                    tp_linear_async_communication=True,
+12/09/2024 21:37:38 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]:                                    recompute_layer=False,
+12/09/2024 21:37:38 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]:                                    tp_recompute_allgather=True,
+12/09/2024 21:37:38 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]:                                    expert_parallel_size=1),
+12/09/2024 21:37:38 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]:        model=ModelArgs(model_config=LlamaConfig(bos_token_id=0,
+12/09/2024 21:37:38 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]:                                                 eos_token_id=0,
+12/09/2024 21:37:38 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]:                                                 hidden_act='silu',
+12/09/2024 21:37:38 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]:                                                 hidden_size=2048,
+12/09/2024 21:37:38 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]:                                                 initializer_range=0.02,
+12/09/2024 21:37:38 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]:                                                 intermediate_size=8192,
+12/09/2024 21:37:38 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]:                                                 is_llama_config=True,
+12/09/2024 21:37:38 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]:                                                 max_position_embeddings=2048,
+12/09/2024 21:37:38 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]:                                                 num_attention_heads=32,
+12/09/2024 21:37:38 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]:                                                 num_hidden_layers=16,
+12/09/2024 21:37:38 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]:                                                 num_key_value_heads=8,
+12/09/2024 21:37:38 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]:                                                 pad_token_id=None,
+12/09/2024 21:37:38 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]:                                                 pretraining_tp=1,
+12/09/2024 21:37:38 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]:                                                 rms_norm_eps=1e-05,
+12/09/2024 21:37:38 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]:                                                 rope_scaling=None,
+12/09/2024 21:37:38 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]:                                                 rope_theta=10000.0,
+12/09/2024 21:37:38 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]:                                                 rope_interleaved=False,
+12/09/2024 21:37:38 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]:                                                 tie_word_embeddings=True,
+12/09/2024 21:37:38 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]:                                                 use_cache=True,
+12/09/2024 21:37:38 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]:                                                 vocab_size=32768),
+12/09/2024 21:37:38 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]:                        init_method=RandomInit(std=0.02),
+12/09/2024 21:37:38 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]:                        dtype=torch.bfloat16,
+12/09/2024 21:37:38 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]:                        make_vocab_size_divisible_by=1,
+12/09/2024 21:37:38 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]:                        ddp_bucket_cap_mb=25),
+12/09/2024 21:37:38 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]:        tokenizer=TokenizerArgs(tokenizer_name_or_path='robot-test/dummy-tokenizer-wordlevel',
+12/09/2024 21:37:38 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]:                                tokenizer_revision=None,
+12/09/2024 21:37:38 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]:                                tokenizer_max_length=None),
+12/09/2024 21:37:38 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]:        checkpoints=CheckpointsArgs(checkpoints_path=PosixPath('checkpoints'),
+12/09/2024 21:37:38 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]:                                    checkpoint_interval=10000,
+12/09/2024 21:37:38 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]:                                    save_initial_state=False,
+12/09/2024 21:37:38 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]:                                    save_final_state=False,
+12/09/2024 21:37:38 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]:                                    resume_checkpoint_path=None,
+12/09/2024 21:37:38 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]:                                    checkpoints_path_is_shared_file_system=False),
+12/09/2024 21:37:38 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]:        logging=LoggingArgs(log_level='info',
+12/09/2024 21:37:38 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]:                            log_level_replica='info',
+12/09/2024 21:37:38 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]:                            iteration_step_info_interval=1),
+12/09/2024 21:37:38 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]:        tokens=TokensArgs(sequence_length=2048,
+12/09/2024 21:37:38 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]:                          train_steps=100,
+12/09/2024 21:37:38 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]:                          micro_batch_size=2,
+12/09/2024 21:37:38 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]:                          batch_accumulation_per_replica=32,
+12/09/2024 21:37:38 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]:                          val_check_interval=100,
+12/09/2024 21:37:38 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]:                          limit_val_batches=0,
+12/09/2024 21:37:38 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]:                          limit_test_batches=0),
+12/09/2024 21:37:38 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]:        optimizer=OptimizerArgs(optimizer_factory=AdamWOptimizerArgs(adam_eps=1e-08,
+12/09/2024 21:37:38 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]:                                                                     adam_beta1=0.9,
+12/09/2024 21:37:38 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]:                                                                     adam_beta2=0.95,
+12/09/2024 21:37:38 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]:                                                                     torch_adam_is_fused=True,
+12/09/2024 21:37:38 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]:                                                                     name='adamW'),
+12/09/2024 21:37:38 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]:                                zero_stage=1,
+12/09/2024 21:37:38 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]:                                weight_decay=0.01,
+12/09/2024 21:37:38 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]:                                clip_grad=1.0,
+12/09/2024 21:37:38 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]:                                accumulate_grad_in_fp32=True,
+12/09/2024 21:37:38 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]:                                learning_rate_scheduler=LRSchedulerArgs(learning_rate=0.0003,
+12/09/2024 21:37:38 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]:                                                                        lr_warmup_steps=2,
+12/09/2024 21:37:38 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]:                                                                        lr_warmup_style='linear',
+12/09/2024 21:37:38 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]:                                                                        lr_decay_style='cosine',
+12/09/2024 21:37:38 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]:                                                                        lr_decay_steps=13,
+12/09/2024 21:37:38 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]:                                                                        lr_decay_starting_step=None,
+12/09/2024 21:37:38 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]:                                                                        min_decay_lr=1e-05)),
+12/09/2024 21:37:38 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]:        data_stages=[DatasetStageArgs(name='Stable Training Stage',
+12/09/2024 21:37:38 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]:                                      start_training_step=1,
+12/09/2024 21:37:38 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]:                                      data=DataArgs(dataset=None,
+12/09/2024 21:37:38 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]:                                                    seed=42,
+12/09/2024 21:37:38 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]:                                                    num_loading_workers=1))],
+12/09/2024 21:37:38 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]:        profiler=None,
+12/09/2024 21:37:38 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]:        lighteval=None,
+12/09/2024 21:37:38 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]:        s3_upload=None)
+12/09/2024 21:37:38 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]: Model Config:
+12/09/2024 21:37:38 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]: LlamaConfig(bos_token_id=0,
+12/09/2024 21:37:38 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]:             eos_token_id=0,
+12/09/2024 21:37:38 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]:             hidden_act='silu',
+12/09/2024 21:37:38 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]:             hidden_size=2048,
+12/09/2024 21:37:38 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]:             initializer_range=0.02,
+12/09/2024 21:37:38 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]:             intermediate_size=8192,
+12/09/2024 21:37:38 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]:             is_llama_config=True,
+12/09/2024 21:37:38 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]:             max_position_embeddings=2048,
+12/09/2024 21:37:38 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]:             num_attention_heads=32,
+12/09/2024 21:37:38 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]:             num_hidden_layers=16,
+12/09/2024 21:37:38 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]:             num_key_value_heads=8,
+12/09/2024 21:37:38 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]:             pad_token_id=None,
+12/09/2024 21:37:38 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]:             pretraining_tp=1,
+12/09/2024 21:37:38 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]:             rms_norm_eps=1e-05,
+12/09/2024 21:37:38 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]:             rope_scaling=None,
+12/09/2024 21:37:38 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]:             rope_theta=10000.0,
+12/09/2024 21:37:38 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]:             rope_interleaved=False,
+12/09/2024 21:37:38 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]:             tie_word_embeddings=True,
+12/09/2024 21:37:38 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]:             use_cache=True,
+12/09/2024 21:37:38 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]:             vocab_size=32768)
+12/09/2024 21:37:38 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]: Building model..
+12/09/2024 21:37:38 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]: Initialize RoPE Theta = 10000.0
+12/09/2024 21:37:38 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]: Setting PP block ranks...
+12/09/2024 21:37:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]: Total number of parameters: 1.04G (1984.13MiB)
+12/09/2024 21:37:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]: Local number of parameters: 1.04G (1984.13MiB)
+12/09/2024 21:37:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]: [After model building] Memory usage: 1984.15MiB. Peak allocated: 5440.00MiB Peak reserved: 22850.00MiB
+12/09/2024 21:37:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]: No checkpoint path provided.
+12/09/2024 21:37:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]: Parametrizing model parameters using StandardParametrizator
+12/09/2024 21:37:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]: [Optimizer Building] Using LearningRateForSP as learning rate
+12/09/2024 21:37:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]: [ZeRO sharding] Size of optimizer params per rank:
+12/09/2024 21:37:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]: [ZeRO sharding] DP Rank 0 has 32.5M out of 1.04G (3.12%) params' optimizer states
+12/09/2024 21:37:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]: [ZeRO sharding] DP Rank 1 has 32.5M out of 1.04G (3.12%) params' optimizer states
+12/09/2024 21:37:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]: [ZeRO sharding] DP Rank 2 has 32.5M out of 1.04G (3.12%) params' optimizer states
+12/09/2024 21:37:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]: [ZeRO sharding] DP Rank 3 has 32.5M out of 1.04G (3.12%) params' optimizer states
+12/09/2024 21:37:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]: [ZeRO sharding] DP Rank 4 has 32.5M out of 1.04G (3.12%) params' optimizer states
+12/09/2024 21:37:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]: [ZeRO sharding] DP Rank 5 has 32.5M out of 1.04G (3.12%) params' optimizer states
+12/09/2024 21:37:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]: [ZeRO sharding] DP Rank 6 has 32.5M out of 1.04G (3.12%) params' optimizer states
+12/09/2024 21:37:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]: [ZeRO sharding] DP Rank 7 has 32.5M out of 1.04G (3.12%) params' optimizer states
+12/09/2024 21:37:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]: [ZeRO sharding] DP Rank 8 has 32.5M out of 1.04G (3.12%) params' optimizer states
+12/09/2024 21:37:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]: [ZeRO sharding] DP Rank 9 has 32.5M out of 1.04G (3.12%) params' optimizer states
+12/09/2024 21:37:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]: [ZeRO sharding] DP Rank 10 has 32.5M out of 1.04G (3.12%) params' optimizer states
+12/09/2024 21:37:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]: [ZeRO sharding] DP Rank 11 has 32.5M out of 1.04G (3.12%) params' optimizer states
+12/09/2024 21:37:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]: [ZeRO sharding] DP Rank 12 has 32.5M out of 1.04G (3.12%) params' optimizer states
+12/09/2024 21:37:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]: [ZeRO sharding] DP Rank 13 has 32.5M out of 1.04G (3.12%) params' optimizer states
+12/09/2024 21:37:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]: [ZeRO sharding] DP Rank 14 has 32.5M out of 1.04G (3.12%) params' optimizer states
+12/09/2024 21:37:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]: [ZeRO sharding] DP Rank 15 has 32.5M out of 1.04G (3.12%) params' optimizer states
+12/09/2024 21:37:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]: [ZeRO sharding] DP Rank 16 has 32.5M out of 1.04G (3.12%) params' optimizer states
+12/09/2024 21:37:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]: [ZeRO sharding] DP Rank 17 has 32.5M out of 1.04G (3.12%) params' optimizer states
+12/09/2024 21:37:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]: [ZeRO sharding] DP Rank 18 has 32.5M out of 1.04G (3.12%) params' optimizer states
+12/09/2024 21:37:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]: [ZeRO sharding] DP Rank 19 has 32.5M out of 1.04G (3.12%) params' optimizer states
+12/09/2024 21:37:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]: [ZeRO sharding] DP Rank 20 has 32.5M out of 1.04G (3.12%) params' optimizer states
+12/09/2024 21:37:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]: [ZeRO sharding] DP Rank 21 has 32.5M out of 1.04G (3.12%) params' optimizer states
+12/09/2024 21:37:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]: [ZeRO sharding] DP Rank 22 has 32.5M out of 1.04G (3.12%) params' optimizer states
+12/09/2024 21:37:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]: [ZeRO sharding] DP Rank 23 has 32.5M out of 1.04G (3.12%) params' optimizer states
+12/09/2024 21:37:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]: [ZeRO sharding] DP Rank 24 has 32.5M out of 1.04G (3.12%) params' optimizer states
+12/09/2024 21:37:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]: [ZeRO sharding] DP Rank 25 has 32.5M out of 1.04G (3.12%) params' optimizer states
+12/09/2024 21:37:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]: [ZeRO sharding] DP Rank 26 has 32.5M out of 1.04G (3.12%) params' optimizer states
+12/09/2024 21:37:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]: [ZeRO sharding] DP Rank 27 has 32.5M out of 1.04G (3.12%) params' optimizer states
+12/09/2024 21:37:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]: [ZeRO sharding] DP Rank 28 has 32.5M out of 1.04G (3.12%) params' optimizer states
+12/09/2024 21:37:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]: [ZeRO sharding] DP Rank 29 has 32.5M out of 1.04G (3.12%) params' optimizer states
+12/09/2024 21:37:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]: [ZeRO sharding] DP Rank 30 has 32.5M out of 1.04G (3.12%) params' optimizer states
+12/09/2024 21:37:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]: [ZeRO sharding] DP Rank 31 has 32.5M out of 1.04G (3.12%) params' optimizer states
+12/09/2024 21:37:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]: [Training Plan] Stage Stable Training Stage has 99 remaining training steps and has consumed 0 samples
+12/09/2024 21:37:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]: Using dummy data generator
+12/09/2024 21:37:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]: [Training Plan] There are 1 training stages 
+12/09/2024 21:37:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]: [Stage Stable Training Stage] start from step 1 
+12/09/2024 21:37:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]: 
+12/09/2024 21:37:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]: [Start training] datetime: 2024-12-09 21:37:51.301121 | mbs: 2 | grad_accum: 32 | global_batch_size: 2048 | sequence_length: 2048 | train_steps: 100 | start_iteration_step: 0 | consumed_train_samples: 0
+wandb: WARNING Path /fsx/nouamane/.cache/wandb/wandb/ wasn't writable, using system temp directory.
+wandb: WARNING Path /fsx/nouamane/.cache/wandb/wandb/ wasn't writable, using system temp directory
+wandb: Tracking run with wandb version 0.16.0
+wandb: W&B syncing is set to `offline` in this directory.  
+wandb: Run `wandb online` or set WANDB_MODE=online to enable cloud syncing.
+12/09/2024 21:38:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]: Resuming training from stage Stable Training Stage, it has trained for 0 samples and has 99 remaining train steps
+12/09/2024 21:38:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]:  Memory usage: 6076.42MiB. Peak allocated 6076.42MiB. Peak reserved: 26820.00MiB
+12/09/2024 21:38:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]:  Memory usage: 6144.51MiB. Peak allocated 12969.22MiB. Peak reserved: 13766.00MiB
+12/09/2024 21:38:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]: iteration: 1 / 100 | consumed_tokens: 4.19M | elapsed_time_per_iteration_ms: 4.37K | tokens_per_sec: 959K | tokens_per_sec_per_gpu: 30K | global_batch_size: 2.05K | lm_loss: 10.8 | lr: 0.00015 | model_tflops_per_gpu: 211 | hardware_tflops_per_gpu: 211 | grad_norm: 0.581 | cuda_memory_allocated: 6.7G | cuda_max_memory_reserved: 16.6G | hd_total_memory_tb: 312G | hd_used_memory_tb: 69.7G | hd_free_memory_tb: 242G
+12/09/2024 21:38:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]:  Memory usage: 6392.59MiB. Peak allocated 10422.86MiB. Peak reserved: 15784.00MiB
+12/09/2024 21:38:09 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]:  Memory usage: 6392.60MiB. Peak allocated 13217.32MiB. Peak reserved: 15794.00MiB
+12/09/2024 21:38:09 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]: iteration: 2 / 100 | consumed_tokens: 8.39M | elapsed_time_per_iteration_ms: 2.45K | tokens_per_sec: 1.71M | tokens_per_sec_per_gpu: 53.4K | global_batch_size: 2.05K | lm_loss: 10.8 | lr: 0.0003 | model_tflops_per_gpu: 377 | hardware_tflops_per_gpu: 377 | grad_norm: 0.58 | cuda_memory_allocated: 6.7G | cuda_max_memory_reserved: 16.6G | hd_total_memory_tb: 312G | hd_used_memory_tb: 69.7G | hd_free_memory_tb: 242G
+12/09/2024 21:38:09 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]:  Memory usage: 6392.59MiB. Peak allocated 10422.88MiB. Peak reserved: 15794.00MiB
+12/09/2024 21:38:11 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]:  Memory usage: 6392.60MiB. Peak allocated 13217.32MiB. Peak reserved: 15794.00MiB
+num_paramsnum_paramsnum_paramsnum_params
+
+
+num_params
+{'total': 1040254976, 'local': 1040254976}{'total': 1040254976, 'local': 1040254976}{'total': 1040254976, 'local': 1040254976}
+
+
+
+{'total': 1040254976, 'local': 1040254976}
+num_paramsnum_paramsnum_params
+
+
+num_params{'total': 1040254976, 'local': 1040254976}{'total': 1040254976, 'local': 1040254976}
+
+{'total': 1040254976, 'local': 1040254976}num_params
+
+
+{'total': 1040254976, 'local': 1040254976}num_params
+num_params
+num_params{'total': 1040254976, 'local': 1040254976}
+
+{'total': 1040254976, 'local': 1040254976}num_params
+
+{'total': 1040254976, 'local': 1040254976}
+
+num_params{'total': 1040254976, 'local': 1040254976}
+
+{'total': 1040254976, 'local': 1040254976}
+{'total': 1040254976, 'local': 1040254976}
+num_paramsnum_params
+
+{'total': 1040254976, 'local': 1040254976}
+{'total': 1040254976, 'local': 1040254976}
+num_params
+{'total': 1040254976, 'local': 1040254976}
+{'total': 1040254976, 'local': 1040254976}num_params
+num_params
+num_params
+num_paramsnum_params
+{'total': 1040254976, 'local': 1040254976}
+
+num_params{'total': 1040254976, 'local': 1040254976}
+
+num_paramsnum_paramsnum_params{'total': 1040254976, 'local': 1040254976}
+
+
+
+{'total': 1040254976, 'local': 1040254976}
+
+num_params{'total': 1040254976, 'local': 1040254976}
+{'total': 1040254976, 'local': 1040254976}
+
+{'total': 1040254976, 'local': 1040254976}
+num_params
+{'total': 1040254976, 'local': 1040254976}
+{'total': 1040254976, 'local': 1040254976}{'total': 1040254976, 'local': 1040254976}num_params
+num_params{'total': 1040254976, 'local': 1040254976}
+
+
+
+{'total': 1040254976, 'local': 1040254976}{'total': 1040254976, 'local': 1040254976}
+
+12/09/2024 21:38:12 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]: iteration: 3 / 100 | consumed_tokens: 12.6M | elapsed_time_per_iteration_ms: 2.43K | tokens_per_sec: 1.72M | tokens_per_sec_per_gpu: 53.9K | global_batch_size: 2.05K | lm_loss: 10.8 | lr: 0.000296 | model_tflops_per_gpu: 380 | hardware_tflops_per_gpu: 380 | grad_norm: 0.571 | cuda_memory_allocated: 6.7G | cuda_max_memory_reserved: 16.6G | hd_total_memory_tb: 312G | hd_used_memory_tb: 69.7G | hd_free_memory_tb: 242G
+num_params
+{'total': 1040254976, 'local': 1040254976}
+12/09/2024 21:38:12 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]: | job_id   | name                                                           | nodes | seq_len | mbs | batch_accum | gbs  | mTFLOPs | hTFLOPs | tok/s/gpu | AllReduce (GB/s) | AllGather (GB/s) | ReduceScatter (GB/s) | AR Intra-node (GB/s) | AG Intra-node (GB/s) | RS Intra-node (GB/s) | Mem Alloc (GB) | Mem Res (GB) | dp | pp | tp | pp_engine                                                                                                 | tp_mode                                 | tp_async_comm | hidden_size | hidden_act | num_layers | num_heads | num_kv_heads | max_pos | vocab_size | tie_word_embeddings | dtype          | zero_stage | ddp_bucket_cap_mb | accumulate_grad_in_fp32 | Total Params | Local Params |
+12/09/2024 21:38:12 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]: | -------- | -------------------------------------------------------------- | ----- | ------- | --- | ----------- | ---- | ------- | ------- | --------- | ---------------- | ---------------- | -------------------- | -------------------- | -------------------- | -------------------- | -------------- | ------------ | -- | -- | -- | --------------------------------------------------------------------------------------------------------- | --------------------------------------- | ------------- | ----------- | ---------- | ---------- | --------- | ------------ | ------- | ---------- | ------------------- | -------------- | ---------- | ----------------- | ----------------------- | ------------ | ------------ |
+12/09/2024 21:38:12 [INFO|DP=16|PP=0|TP=0|ip-26-0-169-132]: Throughput logging complete
+12/09/2024 21:38:12 [INFO|DP=17|PP=0|TP=0|ip-26-0-169-132]: Throughput logging complete
+12/09/2024 21:38:12 [INFO|DP=24|PP=0|TP=0|ip-26-0-169-86]: Throughput logging complete
+12/09/2024 21:38:12 [INFO|DP=10|PP=0|TP=0|ip-26-0-166-36]: Throughput logging complete
+12/09/2024 21:38:12 [INFO|DP=9|PP=0|TP=0|ip-26-0-166-36]: Throughput logging complete
+12/09/2024 21:38:12 [INFO|DP=11|PP=0|TP=0|ip-26-0-166-36]: Throughput logging complete
+12/09/2024 21:38:12 [INFO|DP=20|PP=0|TP=0|ip-26-0-169-132]: Throughput logging complete
+12/09/2024 21:38:12 [INFO|DP=21|PP=0|TP=0|ip-26-0-169-132]: Throughput logging complete
+12/09/2024 21:38:12 [INFO|DP=19|PP=0|TP=0|ip-26-0-169-132]: Throughput logging complete
+12/09/2024 21:38:12 [INFO|DP=18|PP=0|TP=0|ip-26-0-169-132]: Throughput logging complete
+12/09/2024 21:38:12 [INFO|DP=25|PP=0|TP=0|ip-26-0-169-86]: Throughput logging complete
+12/09/2024 21:38:12 [INFO|DP=26|PP=0|TP=0|ip-26-0-169-86]: Throughput logging complete
+12/09/2024 21:38:12 [INFO|DP=27|PP=0|TP=0|ip-26-0-169-86]: Throughput logging complete
+12/09/2024 21:38:12 [INFO|DP=28|PP=0|TP=0|ip-26-0-169-86]: Throughput logging complete
+12/09/2024 21:38:12 [INFO|DP=23|PP=0|TP=0|ip-26-0-169-132]: Throughput logging complete
+12/09/2024 21:38:12 [INFO|DP=22|PP=0|TP=0|ip-26-0-169-132]: Throughput logging complete
+12/09/2024 21:38:12 [INFO|DP=29|PP=0|TP=0|ip-26-0-169-86]: Throughput logging complete
+12/09/2024 21:38:12 [INFO|DP=8|PP=0|TP=0|ip-26-0-166-36]: Throughput logging complete
+12/09/2024 21:38:12 [INFO|DP=12|PP=0|TP=0|ip-26-0-166-36]: Throughput logging complete
+12/09/2024 21:38:12 [INFO|DP=13|PP=0|TP=0|ip-26-0-166-36]: Throughput logging complete
+12/09/2024 21:38:12 [INFO|DP=30|PP=0|TP=0|ip-26-0-169-86]: Throughput logging complete
+12/09/2024 21:38:12 [INFO|DP=31|PP=0|TP=0|ip-26-0-169-86]: Throughput logging complete
+12/09/2024 21:38:12 [INFO|DP=15|PP=0|TP=0|ip-26-0-166-36]: Throughput logging complete
+12/09/2024 21:38:12 [INFO|DP=14|PP=0|TP=0|ip-26-0-166-36]: Throughput logging complete
+12/09/2024 21:38:12 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]: | 13264474 | 1.14G_dp32_tp1_pp1_acc32_mbs2_seq2048_zero1_tpmodeRED_vocab32k | 4     | 2048    | 2   | 32          | 2048 | 379.85  | 379.85  | 53906.05  | 245.37           | 213.04           | 205.72               | 460.42               | 265.19               | 265.31               | 10.18          | 15.42        | 32 | 1  | 1  | <nanotron.parallel.pipeline_parallel.engine.OneForwardOneBackwardPipelineEngine object at 0x7f99b3ae4af0> | TensorParallelLinearMode.REDUCE_SCATTER | True          | 2048        | silu       | 16         | 32        | 8            | 2048    | 32768      | True                | torch.bfloat16 | 1          | 25                | True                    | 1.04G        | 1.04G        |
+12/09/2024 21:38:12 [INFO|DP=0|PP=0|TP=0|ip-26-0-166-214]: Throughput logging complete
+12/09/2024 21:38:12 [INFO|DP=2|PP=0|TP=0|ip-26-0-166-214]: Throughput logging complete
+12/09/2024 21:38:12 [INFO|DP=1|PP=0|TP=0|ip-26-0-166-214]: Throughput logging complete
+12/09/2024 21:38:12 [INFO|DP=3|PP=0|TP=0|ip-26-0-166-214]: Throughput logging complete
+12/09/2024 21:38:12 [INFO|DP=4|PP=0|TP=0|ip-26-0-166-214]: Throughput logging complete
+12/09/2024 21:38:12 [INFO|DP=5|PP=0|TP=0|ip-26-0-166-214]: Throughput logging complete
+12/09/2024 21:38:12 [INFO|DP=7|PP=0|TP=0|ip-26-0-166-214]: Throughput logging complete
+12/09/2024 21:38:12 [INFO|DP=6|PP=0|TP=0|ip-26-0-166-214]: Throughput logging complete
+slurmstepd: error: *** STEP 13264474.0 ON ip-26-0-166-36 CANCELLED AT 2024-12-09T21:38:12 ***
+[2024-12-09 21:38:12,214] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-09 21:38:12,214] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1489790 closing signal SIGTERM
+[2024-12-09 21:38:12,214] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1489791 closing signal SIGTERM
+[2024-12-09 21:38:12,214] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-09 21:38:12,214] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 516901 closing signal SIGTERM
+[2024-12-09 21:38:12,214] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1489792 closing signal SIGTERM
+[2024-12-09 21:38:12,215] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1489793 closing signal SIGTERM
+[2024-12-09 21:38:12,214] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 516902 closing signal SIGTERM
+[2024-12-09 21:38:12,215] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1489794 closing signal SIGTERM
+[2024-12-09 21:38:12,216] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1489795 closing signal SIGTERM
+[2024-12-09 21:38:12,216] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1489796 closing signal SIGTERM
+[2024-12-09 21:38:12,216] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1489797 closing signal SIGTERM
+slurmstepd: error: *** JOB 13264474 ON ip-26-0-166-36 CANCELLED AT 2024-12-09T21:38:12 ***
+srun: Job step aborted: Waiting up to 32 seconds for job step to finish.
+[2024-12-09 21:38:12,216] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-09 21:38:12,216] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 289630 closing signal SIGTERM
+[2024-12-09 21:38:12,214] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 516903 closing signal SIGTERM
+[2024-12-09 21:38:12,214] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 516904 closing signal SIGTERM
+[2024-12-09 21:38:12,217] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 289632 closing signal SIGTERM
+[2024-12-09 21:38:12,217] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 289633 closing signal SIGTERM
+[2024-12-09 21:38:12,217] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 516905 closing signal SIGTERM
+[2024-12-09 21:38:12,217] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 516906 closing signal SIGTERM
+[2024-12-09 21:38:12,217] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 289634 closing signal SIGTERM
+[2024-12-09 21:38:12,217] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 516907 closing signal SIGTERM
+[2024-12-09 21:38:12,217] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 289635 closing signal SIGTERM
+[2024-12-09 21:38:12,217] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 516908 closing signal SIGTERM
+[2024-12-09 21:38:12,217] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-09 21:38:12,217] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 289636 closing signal SIGTERM
+[2024-12-09 21:38:12,217] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 289637 closing signal SIGTERM
+[2024-12-09 21:38:12,217] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 590391 closing signal SIGTERM
+[2024-12-09 21:38:12,217] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 590392 closing signal SIGTERM
+[2024-12-09 21:38:12,217] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 590393 closing signal SIGTERM
+[2024-12-09 21:38:12,217] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 590394 closing signal SIGTERM
+[2024-12-09 21:38:12,218] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 590395 closing signal SIGTERM
+[2024-12-09 21:38:12,219] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 590396 closing signal SIGTERM
+[2024-12-09 21:38:12,219] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 590397 closing signal SIGTERM
+[2024-12-09 21:38:12,220] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 289638 closing signal SIGTERM
+[2024-12-09 21:38:12,220] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 590398 closing signal SIGTERM
diff --git a/logs/13264577-bench_1.34G_dp128_tp1_pp1_acc1_mbs1_seq32768_zero1_tpmodeRED_vocab131k.out b/logs/13264577-bench_1.34G_dp128_tp1_pp1_acc1_mbs1_seq32768_zero1_tpmodeRED_vocab131k.out
new file mode 100644
index 0000000000000000000000000000000000000000..c28c7007d5d3244dd0bb3bf7d8deb3287bd16dcd
--- /dev/null
+++ b/logs/13264577-bench_1.34G_dp128_tp1_pp1_acc1_mbs1_seq32768_zero1_tpmodeRED_vocab131k.out
@@ -0,0 +1,6205 @@
++ source /etc/profile.d/modules.sh
+++ . /usr/share/modules/init/bash
++++ unset _mlshdbg
++++ '[' 0 = 1 ']'
++++ unset _mlre _mlIFS
++++ '[' -n x ']'
++++ _mlIFS=' 	
+'
++++ IFS=' '
++++ '[' -n '' ']'
+++++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash autoinit
++++ _mlcode='module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+MODULEPATH_modshare=/etc/environment-modules/modules:1:/usr/share/modules/\$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1; export MODULEPATH_modshare;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+LOADEDMODULES=; export LOADEDMODULES;
+MODULEPATH=/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/\$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles; export MODULEPATH;
+test 0;'
++++ _mlret=0
++++ '[' -n x ']'
++++ IFS=' 	
+'
++++ unset _mlIFS
++++ unset _mlre _mlv _mlrv
++++ '[' 0 -eq 0 ']'
++++ eval 'module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+MODULEPATH_modshare=/etc/environment-modules/modules:1:/usr/share/modules/\$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1; export MODULEPATH_modshare;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+LOADEDMODULES=; export LOADEDMODULES;
+MODULEPATH=/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/\$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles; export MODULEPATH;
+test 0;'
+++++ MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl
+++++ export MODULES_CMD
+++++ ENV=/usr/share/modules/init/profile.sh
+++++ export ENV
+++++ MODULEPATH_modshare='/etc/environment-modules/modules:1:/usr/share/modules/$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1'
+++++ export MODULEPATH_modshare
+++++ BASH_ENV=/usr/share/modules/init/bash
+++++ export BASH_ENV
+++++ MODULESHOME=/usr/share/modules
+++++ export MODULESHOME
+++++ LOADEDMODULES=
+++++ export LOADEDMODULES
+++++ MODULEPATH='/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles'
+++++ export MODULEPATH
+++++ test 0
++++ '[' 0 = 1 ']'
++++ '[' -t 2 ']'
++++ export -f module
++++ export -f switchml
++++ '[' 5 -ge 3 ']'
++++ [[ ehxB =~ i ]]
++++ [[ ! :/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin: =~ :/usr/bin: ]]
++++ '[' '!' -n '' ']'
++++ MANPATH=:
++++ export MANPATH
+++++ manpath
++++ [[ ! :/admin/home/nouamane/.local/share/man:/fsx/nouamane/miniconda/envs/2-1-cu121/man:/fsx/nouamane/miniconda/envs/2-1-cu121/share/man:/fsx/nouamane/miniconda/man:/fsx/nouamane/miniconda/share/man:/opt/amazon/openmpi/share/man:/opt/amazon/efa/share/man:/opt/slurm/share/man:/usr/local/man:/usr/local/share/man:/usr/share/man:/admin/home/nouamane/.fzf/man:: =~ :/usr/share/man: ]]
++++ unset _mlcode _mlret
++++ '[' -n '' ']'
++ module load cuda/12.1
++ unset _mlshdbg
++ '[' 0 = 1 ']'
++ unset _mlre _mlIFS
++ '[' -n x ']'
++ _mlIFS=' 	
+'
++ IFS=' '
++ '[' -n '' ']'
+++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash load cuda/12.1
++ eval 'CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include;' export 'CPATH;
+LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:;' export 'LD_LIBRARY_PATH;
+FI_EFA_FORK_SAFE=1;' export 'FI_EFA_FORK_SAFE;
+MANPATH=/usr/local/cuda-12.1/share/man::;' export 'MANPATH;
+LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64;' export 'LIBRARY_PATH;
+_LMFILES_=/usr/share/modules/modulefiles/cuda/12.1;' export '_LMFILES_;
+LOADEDMODULES=cuda/12.1;' export 'LOADEDMODULES;
+MPI_PATH=/opt/amazon/openmpi;' export 'MPI_PATH;
+NCCL_HOME_modshare=/opt/nccl/build:1;' export 'NCCL_HOME_modshare;
+NCCL_PROTO=simple;' export 'NCCL_PROTO;
+MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1;' export 'MANPATH_modshare;
+LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1;' export 'LIBRARY_PATH_modshare;
+NCCL_SOCKET_IFNAME=enp;' export 'NCCL_SOCKET_IFNAME;
+AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl;' export 'AWS_OFI_NCCL_HOME;
+NCCL_HOME=/opt/nccl/build;' export 'NCCL_HOME;
+FI_PROVIDER=efa;' export 'FI_PROVIDER;
+AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1;' export 'AWS_OFI_NCCL_HOME_modshare;
+CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1;' export 'CPATH_modshare;
+LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2::1:/opt/nccl/build/lib:1:/opt/aws-ofi-nccl/lib:2:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/efa/lib:1;' export 'LD_LIBRARY_PATH_modshare;
+FI_EFA_ENABLE_SHM_TRANSFER=1;' export 'FI_EFA_ENABLE_SHM_TRANSFER;
+_LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1;' export '_LMFILES__modshare;
+LOADEDMODULES_modshare=cuda/12.1:1;' export 'LOADEDMODULES_modshare;
+MPI_PATH_modshare=/opt/amazon/openmpi:1;' export 'MPI_PATH_modshare;
+PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin;' export 'PATH;
+CUDA_HOME=/usr/local/cuda-12.1;' export 'CUDA_HOME;
+PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/usr/local/cuda-12.1/include:1:/opt/amazon/efa/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/usr/sbin:1:/fsx/nouamane/miniconda/bin:1:/opt/slurm/bin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:2:/opt/amazon/openmpi/bin:1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1;' export 'PATH_modshare;
+test' '0;'
+++ CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include
+++ export CPATH
+++ LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:
+++ export LD_LIBRARY_PATH
+++ FI_EFA_FORK_SAFE=1
+++ export FI_EFA_FORK_SAFE
+++ MANPATH=/usr/local/cuda-12.1/share/man::
+++ export MANPATH
+++ LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64
+++ export LIBRARY_PATH
+++ _LMFILES_=/usr/share/modules/modulefiles/cuda/12.1
+++ export _LMFILES_
+++ LOADEDMODULES=cuda/12.1
+++ export LOADEDMODULES
+++ MPI_PATH=/opt/amazon/openmpi
+++ export MPI_PATH
+++ NCCL_HOME_modshare=/opt/nccl/build:1
+++ export NCCL_HOME_modshare
+++ NCCL_PROTO=simple
+++ export NCCL_PROTO
+++ MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1
+++ export MANPATH_modshare
+++ LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1
+++ export LIBRARY_PATH_modshare
+++ NCCL_SOCKET_IFNAME=enp
+++ export NCCL_SOCKET_IFNAME
+++ AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl
+++ export AWS_OFI_NCCL_HOME
+++ NCCL_HOME=/opt/nccl/build
+++ export NCCL_HOME
+++ FI_PROVIDER=efa
+++ export FI_PROVIDER
+++ AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1
+++ export AWS_OFI_NCCL_HOME_modshare
+++ CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1
+++ export CPATH_modshare
+++ LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2::1:/opt/nccl/build/lib:1:/opt/aws-ofi-nccl/lib:2:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/efa/lib:1
+++ export LD_LIBRARY_PATH_modshare
+++ FI_EFA_ENABLE_SHM_TRANSFER=1
+++ export FI_EFA_ENABLE_SHM_TRANSFER
+++ _LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1
+++ export _LMFILES__modshare
+++ LOADEDMODULES_modshare=cuda/12.1:1
+++ export LOADEDMODULES_modshare
+++ MPI_PATH_modshare=/opt/amazon/openmpi:1
+++ export MPI_PATH_modshare
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ export PATH
+++ CUDA_HOME=/usr/local/cuda-12.1
+++ export CUDA_HOME
+++ PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/usr/local/cuda-12.1/include:1:/opt/amazon/efa/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/usr/sbin:1:/fsx/nouamane/miniconda/bin:1:/opt/slurm/bin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:2:/opt/amazon/openmpi/bin:1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1
+++ export PATH_modshare
+++ test 0
++ _mlstatus=0
++ '[' -n x ']'
++ IFS=' 	
+'
++ unset _mlre _mlv _mlrv _mlIFS
++ '[' -n '' ']'
++ unset _mlshdbg
++ return 0
++ source /fsx/nouamane/miniconda/bin/activate
+++ _CONDA_ROOT=/fsx/nouamane/miniconda
+++ . /fsx/nouamane/miniconda/etc/profile.d/conda.sh
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ '[' -z x ']'
+++ conda activate
+++ local cmd=activate
+++ case "$cmd" in
+++ __conda_activate activate
+++ '[' -n '' ']'
+++ local ask_conda
++++ PS1=
++++ __conda_exe shell.posix activate
++++ /fsx/nouamane/miniconda/bin/conda shell.posix activate
+++ ask_conda='. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_2='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\'''
+++ eval '. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_2='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\'''
++++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh
+++++ test -n ''
+++++ unset XML_CATALOG_FILES
+++++ unset xml_catalog_files_libxml2
++++ PS1='(base) '
++++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++++ export CONDA_PREFIX=/fsx/nouamane/miniconda
++++ CONDA_PREFIX=/fsx/nouamane/miniconda
++++ export CONDA_SHLVL=3
++++ CONDA_SHLVL=3
++++ export CONDA_DEFAULT_ENV=base
++++ CONDA_DEFAULT_ENV=base
++++ export 'CONDA_PROMPT_MODIFIER=(base) '
++++ CONDA_PROMPT_MODIFIER='(base) '
++++ export CONDA_PREFIX_2=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ CONDA_PREFIX_2=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ __conda_hashr
+++ '[' -n '' ']'
+++ '[' -n '' ']'
+++ hash -r
++ conda activate 2-1-cu121
++ local cmd=activate
++ case "$cmd" in
++ __conda_activate activate 2-1-cu121
++ '[' -n '' ']'
++ local ask_conda
+++ PS1='(base) '
+++ __conda_exe shell.posix activate 2-1-cu121
+++ /fsx/nouamane/miniconda/bin/conda shell.posix activate 2-1-cu121
++ ask_conda='PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''4'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_3='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
++ eval 'PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''4'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_3='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
+++ PS1='(2-1-cu121) '
+++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ export CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ export CONDA_SHLVL=4
+++ CONDA_SHLVL=4
+++ export CONDA_DEFAULT_ENV=2-1-cu121
+++ CONDA_DEFAULT_ENV=2-1-cu121
+++ export 'CONDA_PROMPT_MODIFIER=(2-1-cu121) '
+++ CONDA_PROMPT_MODIFIER='(2-1-cu121) '
+++ export CONDA_PREFIX_3=/fsx/nouamane/miniconda
+++ CONDA_PREFIX_3=/fsx/nouamane/miniconda
+++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ export _CE_M=
+++ _CE_M=
+++ export _CE_CONDA=
+++ _CE_CONDA=
+++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh
++++ test -n ''
++++ xml_catalog_files_libxml2=
++++ XML_CATALOG_FILES=
++++ conda_catalog_files=
++++ ifs_libxml2=' 	
+'
++++ IFS=' '
++++ rem=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ for pre in ${rem}
++++ test '' = /fsx/nouamane/miniconda/envs/2-1-cu121
++++ conda_catalog_files=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ rem=
++++ IFS=' 	
+'
++++ conda_catalog_files='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ export 'XML_CATALOG_FILES=file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ XML_CATALOG_FILES='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ unset conda_catalog_files ifs_libxml2 rem
++ __conda_hashr
++ '[' -n '' ']'
++ '[' -n '' ']'
++ hash -r
++ export PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++ PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ scontrol show hostnames 'ip-26-0-164-[45,75,187,207,236],ip-26-0-166-[15,36,125,214,244],ip-26-0-167-[9,51],ip-26-0-171-[88,102,168,230]'
++ export 'NODELIST=ip-26-0-164-45
+ip-26-0-164-75
+ip-26-0-164-187
+ip-26-0-164-207
+ip-26-0-164-236
+ip-26-0-166-15
+ip-26-0-166-36
+ip-26-0-166-125
+ip-26-0-166-214
+ip-26-0-166-244
+ip-26-0-167-9
+ip-26-0-167-51
+ip-26-0-171-88
+ip-26-0-171-102
+ip-26-0-171-168
+ip-26-0-171-230'
++ NODELIST='ip-26-0-164-45
+ip-26-0-164-75
+ip-26-0-164-187
+ip-26-0-164-207
+ip-26-0-164-236
+ip-26-0-166-15
+ip-26-0-166-36
+ip-26-0-166-125
+ip-26-0-166-214
+ip-26-0-166-244
+ip-26-0-167-9
+ip-26-0-167-51
+ip-26-0-171-88
+ip-26-0-171-102
+ip-26-0-171-168
+ip-26-0-171-230'
+++ head -n1
+++ scontrol show hostnames 'ip-26-0-164-[45,75,187,207,236],ip-26-0-166-[15,36,125,214,244],ip-26-0-167-[9,51],ip-26-0-171-[88,102,168,230]'
++ export MASTER_NODE=ip-26-0-164-45
++ MASTER_NODE=ip-26-0-164-45
++ export MASTER_PORT=12356
++ MASTER_PORT=12356
++ export NNODES=16
++ NNODES=16
++ export GPUS_PER_NODE=8
++ GPUS_PER_NODE=8
++ export WORLD_SIZE=128
++ WORLD_SIZE=128
++ export CUDA_DEVICE_MAX_CONNECTIONS=1
++ CUDA_DEVICE_MAX_CONNECTIONS=1
++ export NANOTRON_BENCHMARK=1
++ NANOTRON_BENCHMARK=1
++ echo 'Master node: ip-26-0-164-45'
+Master node: ip-26-0-164-45
++ echo 'All nodes: ip-26-0-164-45
+ip-26-0-164-75
+ip-26-0-164-187
+ip-26-0-164-207
+ip-26-0-164-236
+ip-26-0-166-15
+ip-26-0-166-36
+ip-26-0-166-125
+ip-26-0-166-214
+ip-26-0-166-244
+ip-26-0-167-9
+ip-26-0-167-51
+ip-26-0-171-88
+ip-26-0-171-102
+ip-26-0-171-168
+ip-26-0-171-230'
+All nodes: ip-26-0-164-45
+ip-26-0-164-75
+ip-26-0-164-187
+ip-26-0-164-207
+ip-26-0-164-236
+ip-26-0-166-15
+ip-26-0-166-36
+ip-26-0-166-125
+ip-26-0-166-214
+ip-26-0-166-244
+ip-26-0-167-9
+ip-26-0-167-51
+ip-26-0-171-88
+ip-26-0-171-102
+ip-26-0-171-168
+ip-26-0-171-230
++ echo 'World size: 128'
+World size: 128
++ srun torchrun --nnodes=16 --nproc_per_node=8 --rdzv_id=13264577 --rdzv_backend=c10d --rdzv_endpoint=ip-26-0-164-45:12356 run_train.py --config-file benchmark/configs/config_1.34G_dp128_tp1_pp1_acc1_mbs1_seq32768_zero1_tpmodeRED_vocab131k.yaml
+[2024-12-09 23:29:12,955] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-09 23:29:12,960] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-09 23:29:12,959] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-09 23:29:12,959] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-09 23:29:12,960] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-09 23:29:12,961] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-09 23:29:12,970] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-09 23:29:12,970] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-09 23:29:12,982] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-09 23:29:12,984] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-09 23:29:12,984] torch.distributed.run: [WARNING] 
+[2024-12-09 23:29:12,984] torch.distributed.run: [WARNING] *****************************************
+[2024-12-09 23:29:12,984] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-09 23:29:12,984] torch.distributed.run: [WARNING] *****************************************
+[2024-12-09 23:29:12,987] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-09 23:29:12,988] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-09 23:29:12,961] torch.distributed.run: [WARNING] 
+[2024-12-09 23:29:12,961] torch.distributed.run: [WARNING] *****************************************
+[2024-12-09 23:29:12,961] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-09 23:29:12,961] torch.distributed.run: [WARNING] *****************************************
+[2024-12-09 23:29:12,960] torch.distributed.run: [WARNING] 
+[2024-12-09 23:29:12,960] torch.distributed.run: [WARNING] *****************************************
+[2024-12-09 23:29:12,960] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-09 23:29:12,960] torch.distributed.run: [WARNING] *****************************************
+[2024-12-09 23:29:12,960] torch.distributed.run: [WARNING] 
+[2024-12-09 23:29:12,960] torch.distributed.run: [WARNING] *****************************************
+[2024-12-09 23:29:12,960] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-09 23:29:12,960] torch.distributed.run: [WARNING] *****************************************
+[2024-12-09 23:29:12,959] torch.distributed.run: [WARNING] 
+[2024-12-09 23:29:12,959] torch.distributed.run: [WARNING] *****************************************
+[2024-12-09 23:29:12,959] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-09 23:29:12,959] torch.distributed.run: [WARNING] *****************************************
+[2024-12-09 23:29:12,955] torch.distributed.run: [WARNING] 
+[2024-12-09 23:29:12,955] torch.distributed.run: [WARNING] *****************************************
+[2024-12-09 23:29:12,955] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-09 23:29:12,955] torch.distributed.run: [WARNING] *****************************************
+[2024-12-09 23:29:12,960] torch.distributed.run: [WARNING] 
+[2024-12-09 23:29:12,960] torch.distributed.run: [WARNING] *****************************************
+[2024-12-09 23:29:12,960] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-09 23:29:12,960] torch.distributed.run: [WARNING] *****************************************
+[2024-12-09 23:29:13,003] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-09 23:29:12,971] torch.distributed.run: [WARNING] 
+[2024-12-09 23:29:12,971] torch.distributed.run: [WARNING] *****************************************
+[2024-12-09 23:29:12,971] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-09 23:29:12,971] torch.distributed.run: [WARNING] *****************************************
+[2024-12-09 23:29:12,970] torch.distributed.run: [WARNING] 
+[2024-12-09 23:29:12,970] torch.distributed.run: [WARNING] *****************************************
+[2024-12-09 23:29:12,970] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-09 23:29:12,970] torch.distributed.run: [WARNING] *****************************************
+[2024-12-09 23:29:13,024] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-09 23:29:12,982] torch.distributed.run: [WARNING] 
+[2024-12-09 23:29:12,982] torch.distributed.run: [WARNING] *****************************************
+[2024-12-09 23:29:12,982] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-09 23:29:12,982] torch.distributed.run: [WARNING] *****************************************
+[2024-12-09 23:29:12,988] torch.distributed.run: [WARNING] 
+[2024-12-09 23:29:12,988] torch.distributed.run: [WARNING] *****************************************
+[2024-12-09 23:29:12,988] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-09 23:29:12,988] torch.distributed.run: [WARNING] *****************************************
+[2024-12-09 23:29:12,987] torch.distributed.run: [WARNING] 
+[2024-12-09 23:29:12,987] torch.distributed.run: [WARNING] *****************************************
+[2024-12-09 23:29:12,987] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-09 23:29:12,987] torch.distributed.run: [WARNING] *****************************************
+[2024-12-09 23:29:13,003] torch.distributed.run: [WARNING] 
+[2024-12-09 23:29:13,003] torch.distributed.run: [WARNING] *****************************************
+[2024-12-09 23:29:13,003] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-09 23:29:13,003] torch.distributed.run: [WARNING] *****************************************
+[2024-12-09 23:29:13,024] torch.distributed.run: [WARNING] 
+[2024-12-09 23:29:13,024] torch.distributed.run: [WARNING] *****************************************
+[2024-12-09 23:29:13,024] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-09 23:29:13,024] torch.distributed.run: [WARNING] *****************************************
+[2024-12-09 23:29:13,085] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-09 23:29:13,104] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-09 23:29:13,085] torch.distributed.run: [WARNING] 
+[2024-12-09 23:29:13,085] torch.distributed.run: [WARNING] *****************************************
+[2024-12-09 23:29:13,085] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-09 23:29:13,085] torch.distributed.run: [WARNING] *****************************************
+[2024-12-09 23:29:13,105] torch.distributed.run: [WARNING] 
+[2024-12-09 23:29:13,105] torch.distributed.run: [WARNING] *****************************************
+[2024-12-09 23:29:13,105] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-09 23:29:13,105] torch.distributed.run: [WARNING] *****************************************
+12/09/2024 23:29:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: Config:
+12/09/2024 23:29:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: Config(general=GeneralArgs(project='debug',
+12/09/2024 23:29:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]:                            run='1.34G_dp128_tp1_pp1_acc1_mbs1_seq32768_zero1_tpmodeRED_vocab131k',
+12/09/2024 23:29:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]:                            seed=42,
+12/09/2024 23:29:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]:                            step=None,
+12/09/2024 23:29:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]:                            consumed_train_samples=None,
+12/09/2024 23:29:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]:                            benchmark_csv_path=PosixPath('benchmark/results/bench_final.csv'),
+12/09/2024 23:29:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]:                            ignore_sanity_checks=True),
+12/09/2024 23:29:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]:        parallelism=ParallelismArgs(dp=128,
+12/09/2024 23:29:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]:                                    pp=1,
+12/09/2024 23:29:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]:                                    tp=1,
+12/09/2024 23:29:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]:                                    pp_engine=<nanotron.parallel.pipeline_parallel.engine.OneForwardOneBackwardPipelineEngine object at 0x7fe955778af0>,
+12/09/2024 23:29:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]:                                    tp_mode=<TensorParallelLinearMode.REDUCE_SCATTER: 2>,
+12/09/2024 23:29:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]:                                    tp_linear_async_communication=True,
+12/09/2024 23:29:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]:                                    recompute_layer=False,
+12/09/2024 23:29:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]:                                    tp_recompute_allgather=True,
+12/09/2024 23:29:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]:                                    expert_parallel_size=1),
+12/09/2024 23:29:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]:        model=ModelArgs(model_config=LlamaConfig(bos_token_id=0,
+12/09/2024 23:29:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]:                                                 eos_token_id=0,
+12/09/2024 23:29:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]:                                                 hidden_act='silu',
+12/09/2024 23:29:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]:                                                 hidden_size=2048,
+12/09/2024 23:29:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]:                                                 initializer_range=0.02,
+12/09/2024 23:29:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]:                                                 intermediate_size=8192,
+12/09/2024 23:29:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]:                                                 is_llama_config=True,
+12/09/2024 23:29:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]:                                                 max_position_embeddings=32768,
+12/09/2024 23:29:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]:                                                 num_attention_heads=32,
+12/09/2024 23:29:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]:                                                 num_hidden_layers=16,
+12/09/2024 23:29:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]:                                                 num_key_value_heads=8,
+12/09/2024 23:29:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]:                                                 pad_token_id=None,
+12/09/2024 23:29:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]:                                                 pretraining_tp=1,
+12/09/2024 23:29:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]:                                                 rms_norm_eps=1e-05,
+12/09/2024 23:29:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]:                                                 rope_scaling=None,
+12/09/2024 23:29:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]:                                                 rope_theta=10000.0,
+12/09/2024 23:29:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]:                                                 rope_interleaved=False,
+12/09/2024 23:29:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]:                                                 tie_word_embeddings=True,
+12/09/2024 23:29:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]:                                                 use_cache=True,
+12/09/2024 23:29:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]:                                                 vocab_size=131072),
+12/09/2024 23:29:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]:                        init_method=RandomInit(std=0.02),
+12/09/2024 23:29:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]:                        dtype=torch.bfloat16,
+12/09/2024 23:29:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]:                        make_vocab_size_divisible_by=1,
+12/09/2024 23:29:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]:                        ddp_bucket_cap_mb=25),
+12/09/2024 23:29:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]:        tokenizer=TokenizerArgs(tokenizer_name_or_path='robot-test/dummy-tokenizer-wordlevel',
+12/09/2024 23:29:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]:                                tokenizer_revision=None,
+12/09/2024 23:29:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]:                                tokenizer_max_length=None),
+12/09/2024 23:29:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]:        checkpoints=CheckpointsArgs(checkpoints_path=PosixPath('checkpoints'),
+12/09/2024 23:29:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]:                                    checkpoint_interval=10000,
+12/09/2024 23:29:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]:                                    save_initial_state=False,
+12/09/2024 23:29:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]:                                    save_final_state=False,
+12/09/2024 23:29:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]:                                    resume_checkpoint_path=None,
+12/09/2024 23:29:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]:                                    checkpoints_path_is_shared_file_system=False),
+12/09/2024 23:29:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]:        logging=LoggingArgs(log_level='info',
+12/09/2024 23:29:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]:                            log_level_replica='info',
+12/09/2024 23:29:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]:                            iteration_step_info_interval=1),
+12/09/2024 23:29:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]:        tokens=TokensArgs(sequence_length=32768,
+12/09/2024 23:29:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]:                          train_steps=100,
+12/09/2024 23:29:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]:                          micro_batch_size=1,
+12/09/2024 23:29:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]:                          batch_accumulation_per_replica=1,
+12/09/2024 23:29:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]:                          val_check_interval=100,
+12/09/2024 23:29:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]:                          limit_val_batches=0,
+12/09/2024 23:29:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]:                          limit_test_batches=0),
+12/09/2024 23:29:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]:        optimizer=OptimizerArgs(optimizer_factory=AdamWOptimizerArgs(adam_eps=1e-08,
+12/09/2024 23:29:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]:                                                                     adam_beta1=0.9,
+12/09/2024 23:29:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]:                                                                     adam_beta2=0.95,
+12/09/2024 23:29:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]:                                                                     torch_adam_is_fused=True,
+12/09/2024 23:29:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]:                                                                     name='adamW'),
+12/09/2024 23:29:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]:                                zero_stage=1,
+12/09/2024 23:29:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]:                                weight_decay=0.01,
+12/09/2024 23:29:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]:                                clip_grad=1.0,
+12/09/2024 23:29:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]:                                accumulate_grad_in_fp32=True,
+12/09/2024 23:29:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]:                                learning_rate_scheduler=LRSchedulerArgs(learning_rate=0.0003,
+12/09/2024 23:29:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]:                                                                        lr_warmup_steps=2,
+12/09/2024 23:29:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]:                                                                        lr_warmup_style='linear',
+12/09/2024 23:29:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]:                                                                        lr_decay_style='cosine',
+12/09/2024 23:29:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]:                                                                        lr_decay_steps=13,
+12/09/2024 23:29:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]:                                                                        lr_decay_starting_step=None,
+12/09/2024 23:29:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]:                                                                        min_decay_lr=1e-05)),
+12/09/2024 23:29:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]:        data_stages=[DatasetStageArgs(name='Stable Training Stage',
+12/09/2024 23:29:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]:                                      start_training_step=1,
+12/09/2024 23:29:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]:                                      data=DataArgs(dataset=None,
+12/09/2024 23:29:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]:                                                    seed=42,
+12/09/2024 23:29:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]:                                                    num_loading_workers=1))],
+12/09/2024 23:29:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]:        profiler=None,
+12/09/2024 23:29:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]:        lighteval=None,
+12/09/2024 23:29:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]:        s3_upload=None)
+12/09/2024 23:29:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: Model Config:
+12/09/2024 23:29:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: LlamaConfig(bos_token_id=0,
+12/09/2024 23:29:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]:             eos_token_id=0,
+12/09/2024 23:29:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]:             hidden_act='silu',
+12/09/2024 23:29:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]:             hidden_size=2048,
+12/09/2024 23:29:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]:             initializer_range=0.02,
+12/09/2024 23:29:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]:             intermediate_size=8192,
+12/09/2024 23:29:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]:             is_llama_config=True,
+12/09/2024 23:29:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]:             max_position_embeddings=32768,
+12/09/2024 23:29:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]:             num_attention_heads=32,
+12/09/2024 23:29:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]:             num_hidden_layers=16,
+12/09/2024 23:29:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]:             num_key_value_heads=8,
+12/09/2024 23:29:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]:             pad_token_id=None,
+12/09/2024 23:29:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]:             pretraining_tp=1,
+12/09/2024 23:29:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]:             rms_norm_eps=1e-05,
+12/09/2024 23:29:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]:             rope_scaling=None,
+12/09/2024 23:29:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]:             rope_theta=10000.0,
+12/09/2024 23:29:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]:             rope_interleaved=False,
+12/09/2024 23:29:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]:             tie_word_embeddings=True,
+12/09/2024 23:29:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]:             use_cache=True,
+12/09/2024 23:29:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]:             vocab_size=131072)
+12/09/2024 23:29:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: Building model..
+12/09/2024 23:29:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: Initialize RoPE Theta = 10000.0
+12/09/2024 23:29:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: Setting PP block ranks...
+12/09/2024 23:29:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: Total number of parameters: 1.24G (2368.13MiB)
+12/09/2024 23:29:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: Local number of parameters: 1.24G (2368.13MiB)
+12/09/2024 23:29:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [After model building] Memory usage: 2368.15MiB. Peak allocated: 5392.00MiB Peak reserved: 29974.00MiB
+12/09/2024 23:29:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: No checkpoint path provided.
+12/09/2024 23:29:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: Parametrizing model parameters using StandardParametrizator
+12/09/2024 23:30:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [Optimizer Building] Using LearningRateForSP as learning rate
+12/09/2024 23:30:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [ZeRO sharding] Size of optimizer params per rank:
+12/09/2024 23:30:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [ZeRO sharding] DP Rank 0 has 9.7M out of 1.24G (0.78%) params' optimizer states
+12/09/2024 23:30:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [ZeRO sharding] DP Rank 1 has 9.7M out of 1.24G (0.78%) params' optimizer states
+12/09/2024 23:30:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [ZeRO sharding] DP Rank 2 has 9.7M out of 1.24G (0.78%) params' optimizer states
+12/09/2024 23:30:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [ZeRO sharding] DP Rank 3 has 9.7M out of 1.24G (0.78%) params' optimizer states
+12/09/2024 23:30:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [ZeRO sharding] DP Rank 4 has 9.7M out of 1.24G (0.78%) params' optimizer states
+12/09/2024 23:30:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [ZeRO sharding] DP Rank 5 has 9.7M out of 1.24G (0.78%) params' optimizer states
+12/09/2024 23:30:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [ZeRO sharding] DP Rank 6 has 9.7M out of 1.24G (0.78%) params' optimizer states
+12/09/2024 23:30:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [ZeRO sharding] DP Rank 7 has 9.7M out of 1.24G (0.78%) params' optimizer states
+12/09/2024 23:30:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [ZeRO sharding] DP Rank 8 has 9.7M out of 1.24G (0.78%) params' optimizer states
+12/09/2024 23:30:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [ZeRO sharding] DP Rank 9 has 9.7M out of 1.24G (0.78%) params' optimizer states
+12/09/2024 23:30:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [ZeRO sharding] DP Rank 10 has 9.7M out of 1.24G (0.78%) params' optimizer states
+12/09/2024 23:30:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [ZeRO sharding] DP Rank 11 has 9.7M out of 1.24G (0.78%) params' optimizer states
+12/09/2024 23:30:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [ZeRO sharding] DP Rank 12 has 9.7M out of 1.24G (0.78%) params' optimizer states
+12/09/2024 23:30:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [ZeRO sharding] DP Rank 13 has 9.7M out of 1.24G (0.78%) params' optimizer states
+12/09/2024 23:30:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [ZeRO sharding] DP Rank 14 has 9.7M out of 1.24G (0.78%) params' optimizer states
+12/09/2024 23:30:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [ZeRO sharding] DP Rank 15 has 9.7M out of 1.24G (0.78%) params' optimizer states
+12/09/2024 23:30:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [ZeRO sharding] DP Rank 16 has 9.7M out of 1.24G (0.78%) params' optimizer states
+12/09/2024 23:30:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [ZeRO sharding] DP Rank 17 has 9.7M out of 1.24G (0.78%) params' optimizer states
+12/09/2024 23:30:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [ZeRO sharding] DP Rank 18 has 9.7M out of 1.24G (0.78%) params' optimizer states
+12/09/2024 23:30:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [ZeRO sharding] DP Rank 19 has 9.7M out of 1.24G (0.78%) params' optimizer states
+12/09/2024 23:30:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [ZeRO sharding] DP Rank 20 has 9.7M out of 1.24G (0.78%) params' optimizer states
+12/09/2024 23:30:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [ZeRO sharding] DP Rank 21 has 9.7M out of 1.24G (0.78%) params' optimizer states
+12/09/2024 23:30:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [ZeRO sharding] DP Rank 22 has 9.7M out of 1.24G (0.78%) params' optimizer states
+12/09/2024 23:30:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [ZeRO sharding] DP Rank 23 has 9.7M out of 1.24G (0.78%) params' optimizer states
+12/09/2024 23:30:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [ZeRO sharding] DP Rank 24 has 9.7M out of 1.24G (0.78%) params' optimizer states
+12/09/2024 23:30:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [ZeRO sharding] DP Rank 25 has 9.7M out of 1.24G (0.78%) params' optimizer states
+12/09/2024 23:30:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [ZeRO sharding] DP Rank 26 has 9.7M out of 1.24G (0.78%) params' optimizer states
+12/09/2024 23:30:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [ZeRO sharding] DP Rank 27 has 9.7M out of 1.24G (0.78%) params' optimizer states
+12/09/2024 23:30:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [ZeRO sharding] DP Rank 28 has 9.7M out of 1.24G (0.78%) params' optimizer states
+12/09/2024 23:30:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [ZeRO sharding] DP Rank 29 has 9.7M out of 1.24G (0.78%) params' optimizer states
+12/09/2024 23:30:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [ZeRO sharding] DP Rank 30 has 9.7M out of 1.24G (0.78%) params' optimizer states
+12/09/2024 23:30:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [ZeRO sharding] DP Rank 31 has 9.7M out of 1.24G (0.78%) params' optimizer states
+12/09/2024 23:30:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [ZeRO sharding] DP Rank 32 has 9.7M out of 1.24G (0.78%) params' optimizer states
+12/09/2024 23:30:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [ZeRO sharding] DP Rank 33 has 9.7M out of 1.24G (0.78%) params' optimizer states
+12/09/2024 23:30:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [ZeRO sharding] DP Rank 34 has 9.7M out of 1.24G (0.78%) params' optimizer states
+12/09/2024 23:30:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [ZeRO sharding] DP Rank 35 has 9.7M out of 1.24G (0.78%) params' optimizer states
+12/09/2024 23:30:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [ZeRO sharding] DP Rank 36 has 9.7M out of 1.24G (0.78%) params' optimizer states
+12/09/2024 23:30:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [ZeRO sharding] DP Rank 37 has 9.7M out of 1.24G (0.78%) params' optimizer states
+12/09/2024 23:30:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [ZeRO sharding] DP Rank 38 has 9.7M out of 1.24G (0.78%) params' optimizer states
+12/09/2024 23:30:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [ZeRO sharding] DP Rank 39 has 9.7M out of 1.24G (0.78%) params' optimizer states
+12/09/2024 23:30:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [ZeRO sharding] DP Rank 40 has 9.7M out of 1.24G (0.78%) params' optimizer states
+12/09/2024 23:30:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [ZeRO sharding] DP Rank 41 has 9.7M out of 1.24G (0.78%) params' optimizer states
+12/09/2024 23:30:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [ZeRO sharding] DP Rank 42 has 9.7M out of 1.24G (0.78%) params' optimizer states
+12/09/2024 23:30:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [ZeRO sharding] DP Rank 43 has 9.7M out of 1.24G (0.78%) params' optimizer states
+12/09/2024 23:30:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [ZeRO sharding] DP Rank 44 has 9.7M out of 1.24G (0.78%) params' optimizer states
+12/09/2024 23:30:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [ZeRO sharding] DP Rank 45 has 9.7M out of 1.24G (0.78%) params' optimizer states
+12/09/2024 23:30:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [ZeRO sharding] DP Rank 46 has 9.7M out of 1.24G (0.78%) params' optimizer states
+12/09/2024 23:30:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [ZeRO sharding] DP Rank 47 has 9.7M out of 1.24G (0.78%) params' optimizer states
+12/09/2024 23:30:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [ZeRO sharding] DP Rank 48 has 9.7M out of 1.24G (0.78%) params' optimizer states
+12/09/2024 23:30:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [ZeRO sharding] DP Rank 49 has 9.7M out of 1.24G (0.78%) params' optimizer states
+12/09/2024 23:30:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [ZeRO sharding] DP Rank 50 has 9.7M out of 1.24G (0.78%) params' optimizer states
+12/09/2024 23:30:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [ZeRO sharding] DP Rank 51 has 9.7M out of 1.24G (0.78%) params' optimizer states
+12/09/2024 23:30:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [ZeRO sharding] DP Rank 52 has 9.7M out of 1.24G (0.78%) params' optimizer states
+12/09/2024 23:30:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [ZeRO sharding] DP Rank 53 has 9.7M out of 1.24G (0.78%) params' optimizer states
+12/09/2024 23:30:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [ZeRO sharding] DP Rank 54 has 9.7M out of 1.24G (0.78%) params' optimizer states
+12/09/2024 23:30:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [ZeRO sharding] DP Rank 55 has 9.7M out of 1.24G (0.78%) params' optimizer states
+12/09/2024 23:30:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [ZeRO sharding] DP Rank 56 has 9.7M out of 1.24G (0.78%) params' optimizer states
+12/09/2024 23:30:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [ZeRO sharding] DP Rank 57 has 9.7M out of 1.24G (0.78%) params' optimizer states
+12/09/2024 23:30:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [ZeRO sharding] DP Rank 58 has 9.7M out of 1.24G (0.78%) params' optimizer states
+12/09/2024 23:30:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [ZeRO sharding] DP Rank 59 has 9.7M out of 1.24G (0.78%) params' optimizer states
+12/09/2024 23:30:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [ZeRO sharding] DP Rank 60 has 9.7M out of 1.24G (0.78%) params' optimizer states
+12/09/2024 23:30:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [ZeRO sharding] DP Rank 61 has 9.7M out of 1.24G (0.78%) params' optimizer states
+12/09/2024 23:30:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [ZeRO sharding] DP Rank 62 has 9.7M out of 1.24G (0.78%) params' optimizer states
+12/09/2024 23:30:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [ZeRO sharding] DP Rank 63 has 9.7M out of 1.24G (0.78%) params' optimizer states
+12/09/2024 23:30:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [ZeRO sharding] DP Rank 64 has 9.7M out of 1.24G (0.78%) params' optimizer states
+12/09/2024 23:30:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [ZeRO sharding] DP Rank 65 has 9.7M out of 1.24G (0.78%) params' optimizer states
+12/09/2024 23:30:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [ZeRO sharding] DP Rank 66 has 9.7M out of 1.24G (0.78%) params' optimizer states
+12/09/2024 23:30:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [ZeRO sharding] DP Rank 67 has 9.7M out of 1.24G (0.78%) params' optimizer states
+12/09/2024 23:30:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [ZeRO sharding] DP Rank 68 has 9.7M out of 1.24G (0.78%) params' optimizer states
+12/09/2024 23:30:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [ZeRO sharding] DP Rank 69 has 9.7M out of 1.24G (0.78%) params' optimizer states
+12/09/2024 23:30:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [ZeRO sharding] DP Rank 70 has 9.7M out of 1.24G (0.78%) params' optimizer states
+12/09/2024 23:30:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [ZeRO sharding] DP Rank 71 has 9.7M out of 1.24G (0.78%) params' optimizer states
+12/09/2024 23:30:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [ZeRO sharding] DP Rank 72 has 9.7M out of 1.24G (0.78%) params' optimizer states
+12/09/2024 23:30:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [ZeRO sharding] DP Rank 73 has 9.7M out of 1.24G (0.78%) params' optimizer states
+12/09/2024 23:30:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [ZeRO sharding] DP Rank 74 has 9.7M out of 1.24G (0.78%) params' optimizer states
+12/09/2024 23:30:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [ZeRO sharding] DP Rank 75 has 9.7M out of 1.24G (0.78%) params' optimizer states
+12/09/2024 23:30:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [ZeRO sharding] DP Rank 76 has 9.7M out of 1.24G (0.78%) params' optimizer states
+12/09/2024 23:30:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [ZeRO sharding] DP Rank 77 has 9.7M out of 1.24G (0.78%) params' optimizer states
+12/09/2024 23:30:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [ZeRO sharding] DP Rank 78 has 9.7M out of 1.24G (0.78%) params' optimizer states
+12/09/2024 23:30:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [ZeRO sharding] DP Rank 79 has 9.7M out of 1.24G (0.78%) params' optimizer states
+12/09/2024 23:30:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [ZeRO sharding] DP Rank 80 has 9.7M out of 1.24G (0.78%) params' optimizer states
+12/09/2024 23:30:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [ZeRO sharding] DP Rank 81 has 9.7M out of 1.24G (0.78%) params' optimizer states
+12/09/2024 23:30:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [ZeRO sharding] DP Rank 82 has 9.7M out of 1.24G (0.78%) params' optimizer states
+12/09/2024 23:30:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [ZeRO sharding] DP Rank 83 has 9.7M out of 1.24G (0.78%) params' optimizer states
+12/09/2024 23:30:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [ZeRO sharding] DP Rank 84 has 9.7M out of 1.24G (0.78%) params' optimizer states
+12/09/2024 23:30:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [ZeRO sharding] DP Rank 85 has 9.7M out of 1.24G (0.78%) params' optimizer states
+12/09/2024 23:30:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [ZeRO sharding] DP Rank 86 has 9.7M out of 1.24G (0.78%) params' optimizer states
+12/09/2024 23:30:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [ZeRO sharding] DP Rank 87 has 9.7M out of 1.24G (0.78%) params' optimizer states
+12/09/2024 23:30:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [ZeRO sharding] DP Rank 88 has 9.7M out of 1.24G (0.78%) params' optimizer states
+12/09/2024 23:30:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [ZeRO sharding] DP Rank 89 has 9.7M out of 1.24G (0.78%) params' optimizer states
+12/09/2024 23:30:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [ZeRO sharding] DP Rank 90 has 9.7M out of 1.24G (0.78%) params' optimizer states
+12/09/2024 23:30:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [ZeRO sharding] DP Rank 91 has 9.7M out of 1.24G (0.78%) params' optimizer states
+12/09/2024 23:30:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [ZeRO sharding] DP Rank 92 has 9.7M out of 1.24G (0.78%) params' optimizer states
+12/09/2024 23:30:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [ZeRO sharding] DP Rank 93 has 9.7M out of 1.24G (0.78%) params' optimizer states
+12/09/2024 23:30:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [ZeRO sharding] DP Rank 94 has 9.7M out of 1.24G (0.78%) params' optimizer states
+12/09/2024 23:30:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [ZeRO sharding] DP Rank 95 has 9.7M out of 1.24G (0.78%) params' optimizer states
+12/09/2024 23:30:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [ZeRO sharding] DP Rank 96 has 9.7M out of 1.24G (0.78%) params' optimizer states
+12/09/2024 23:30:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [ZeRO sharding] DP Rank 97 has 9.7M out of 1.24G (0.78%) params' optimizer states
+12/09/2024 23:30:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [ZeRO sharding] DP Rank 98 has 9.7M out of 1.24G (0.78%) params' optimizer states
+12/09/2024 23:30:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [ZeRO sharding] DP Rank 99 has 9.7M out of 1.24G (0.78%) params' optimizer states
+12/09/2024 23:30:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [ZeRO sharding] DP Rank 100 has 9.7M out of 1.24G (0.78%) params' optimizer states
+12/09/2024 23:30:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [ZeRO sharding] DP Rank 101 has 9.7M out of 1.24G (0.78%) params' optimizer states
+12/09/2024 23:30:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [ZeRO sharding] DP Rank 102 has 9.7M out of 1.24G (0.78%) params' optimizer states
+12/09/2024 23:30:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [ZeRO sharding] DP Rank 103 has 9.7M out of 1.24G (0.78%) params' optimizer states
+12/09/2024 23:30:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [ZeRO sharding] DP Rank 104 has 9.7M out of 1.24G (0.78%) params' optimizer states
+12/09/2024 23:30:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [ZeRO sharding] DP Rank 105 has 9.7M out of 1.24G (0.78%) params' optimizer states
+12/09/2024 23:30:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [ZeRO sharding] DP Rank 106 has 9.7M out of 1.24G (0.78%) params' optimizer states
+12/09/2024 23:30:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [ZeRO sharding] DP Rank 107 has 9.7M out of 1.24G (0.78%) params' optimizer states
+12/09/2024 23:30:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [ZeRO sharding] DP Rank 108 has 9.7M out of 1.24G (0.78%) params' optimizer states
+12/09/2024 23:30:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [ZeRO sharding] DP Rank 109 has 9.7M out of 1.24G (0.78%) params' optimizer states
+12/09/2024 23:30:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [ZeRO sharding] DP Rank 110 has 9.7M out of 1.24G (0.78%) params' optimizer states
+12/09/2024 23:30:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [ZeRO sharding] DP Rank 111 has 9.7M out of 1.24G (0.78%) params' optimizer states
+12/09/2024 23:30:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [ZeRO sharding] DP Rank 112 has 9.7M out of 1.24G (0.78%) params' optimizer states
+12/09/2024 23:30:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [ZeRO sharding] DP Rank 113 has 9.7M out of 1.24G (0.78%) params' optimizer states
+12/09/2024 23:30:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [ZeRO sharding] DP Rank 114 has 9.7M out of 1.24G (0.78%) params' optimizer states
+12/09/2024 23:30:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [ZeRO sharding] DP Rank 115 has 9.7M out of 1.24G (0.78%) params' optimizer states
+12/09/2024 23:30:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [ZeRO sharding] DP Rank 116 has 9.7M out of 1.24G (0.78%) params' optimizer states
+12/09/2024 23:30:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [ZeRO sharding] DP Rank 117 has 9.7M out of 1.24G (0.78%) params' optimizer states
+12/09/2024 23:30:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [ZeRO sharding] DP Rank 118 has 9.7M out of 1.24G (0.78%) params' optimizer states
+12/09/2024 23:30:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [ZeRO sharding] DP Rank 119 has 9.7M out of 1.24G (0.78%) params' optimizer states
+12/09/2024 23:30:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [ZeRO sharding] DP Rank 120 has 9.7M out of 1.24G (0.78%) params' optimizer states
+12/09/2024 23:30:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [ZeRO sharding] DP Rank 121 has 9.7M out of 1.24G (0.78%) params' optimizer states
+12/09/2024 23:30:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [ZeRO sharding] DP Rank 122 has 9.7M out of 1.24G (0.78%) params' optimizer states
+12/09/2024 23:30:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [ZeRO sharding] DP Rank 123 has 9.7M out of 1.24G (0.78%) params' optimizer states
+12/09/2024 23:30:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [ZeRO sharding] DP Rank 124 has 9.7M out of 1.24G (0.78%) params' optimizer states
+12/09/2024 23:30:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [ZeRO sharding] DP Rank 125 has 9.7M out of 1.24G (0.78%) params' optimizer states
+12/09/2024 23:30:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [ZeRO sharding] DP Rank 126 has 9.7M out of 1.24G (0.78%) params' optimizer states
+12/09/2024 23:30:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [ZeRO sharding] DP Rank 127 has 9.7M out of 1.24G (0.78%) params' optimizer states
+12/09/2024 23:30:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [Training Plan] Stage Stable Training Stage has 99 remaining training steps and has consumed 0 samples
+12/09/2024 23:30:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: Using dummy data generator
+12/09/2024 23:30:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [Training Plan] There are 1 training stages 
+12/09/2024 23:30:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [Stage Stable Training Stage] start from step 1 
+12/09/2024 23:30:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: 
+12/09/2024 23:30:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-187]: [Start training] datetime: 2024-12-09 23:30:07.191178 | mbs: 1 | grad_accum: 1 | global_batch_size: 128 | sequence_length: 32768 | train_steps: 100 | start_iteration_step: 0 | consumed_train_samples: 0
+wandb: WARNING Path /fsx/nouamane/.cache/wandb/wandb/ wasn't writable, using system temp directory.
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    output = model(**micro_batch)
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    sharded_logits = self.model(
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    module_builder=lambda: lambda x: x.float(),
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 1 has a total capacty of 79.33 GiB of which 6.80 GiB is free. Including non-PyTorch memory, this process has 72.51 GiB memory in use. Of the allocated memory 60.39 GiB is allocated by PyTorch, and 208.88 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    module_builder=lambda: lambda x: x.float(),
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 7 has a total capacty of 79.33 GiB of which 7.10 GiB is free. Including non-PyTorch memory, this process has 72.22 GiB memory in use. Of the allocated memory 60.39 GiB is allocated by PyTorch, and 208.88 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+wandb: WARNING Path /fsx/nouamane/.cache/wandb/wandb/ wasn't writable, using system temp directory
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    trainer.train(dataloader)
+    output = model(**micro_batch)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    module_builder=lambda: lambda x: x.float(),
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 3 has a total capacty of 79.33 GiB of which 6.69 GiB is free. Including non-PyTorch memory, this process has 72.63 GiB memory in use. Of the allocated memory 60.39 GiB is allocated by PyTorch, and 208.88 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    module_builder=lambda: lambda x: x.float(),
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 3 has a total capacty of 79.33 GiB of which 7.68 GiB is free. Including non-PyTorch memory, this process has 71.63 GiB memory in use. Of the allocated memory 60.39 GiB is allocated by PyTorch, and 208.88 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    module_builder=lambda: lambda x: x.float(),
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 5 has a total capacty of 79.33 GiB of which 6.98 GiB is free. Including non-PyTorch memory, this process has 72.34 GiB memory in use. Of the allocated memory 60.39 GiB is allocated by PyTorch, and 208.88 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = model(**micro_batch)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    module_builder=lambda: lambda x: x.float(),
+    output = model(**micro_batch)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 4 has a total capacty of 79.33 GiB of which 6.84 GiB is free. Including non-PyTorch memory, this process has 72.48 GiB memory in use. Of the allocated memory 60.39 GiB is allocated by PyTorch, and 208.88 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+    sharded_logits = self.model(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    module_builder=lambda: lambda x: x.float(),
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 3 has a total capacty of 79.33 GiB of which 6.63 GiB is free. Including non-PyTorch memory, this process has 72.69 GiB memory in use. Of the allocated memory 60.39 GiB is allocated by PyTorch, and 208.88 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    module_builder=lambda: lambda x: x.float(),
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 1 has a total capacty of 79.33 GiB of which 6.80 GiB is free. Including non-PyTorch memory, this process has 72.51 GiB memory in use. Of the allocated memory 60.39 GiB is allocated by PyTorch, and 208.88 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    module_builder=lambda: lambda x: x.float(),
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 4 has a total capacty of 79.33 GiB of which 6.90 GiB is free. Including non-PyTorch memory, this process has 72.42 GiB memory in use. Of the allocated memory 60.39 GiB is allocated by PyTorch, and 208.88 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    module_builder=lambda: lambda x: x.float(),
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 4 has a total capacty of 79.33 GiB of which 6.96 GiB is free. Including non-PyTorch memory, this process has 72.36 GiB memory in use. Of the allocated memory 60.39 GiB is allocated by PyTorch, and 208.88 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    module_builder=lambda: lambda x: x.float(),
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 3 has a total capacty of 79.33 GiB of which 6.75 GiB is free. Including non-PyTorch memory, this process has 72.57 GiB memory in use. Of the allocated memory 60.39 GiB is allocated by PyTorch, and 208.88 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    module_builder=lambda: lambda x: x.float(),
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 1 has a total capacty of 79.33 GiB of which 6.98 GiB is free. Including non-PyTorch memory, this process has 72.34 GiB memory in use. Of the allocated memory 60.39 GiB is allocated by PyTorch, and 208.88 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    module_builder=lambda: lambda x: x.float(),
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 5 has a total capacty of 79.33 GiB of which 6.80 GiB is free. Including non-PyTorch memory, this process has 72.51 GiB memory in use. Of the allocated memory 60.39 GiB is allocated by PyTorch, and 208.88 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    module_builder=lambda: lambda x: x.float(),
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 2 has a total capacty of 79.33 GiB of which 6.98 GiB is free. Including non-PyTorch memory, this process has 72.34 GiB memory in use. Of the allocated memory 60.39 GiB is allocated by PyTorch, and 208.88 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    module_builder=lambda: lambda x: x.float(),
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 6 has a total capacty of 79.33 GiB of which 6.86 GiB is free. Including non-PyTorch memory, this process has 72.46 GiB memory in use. Of the allocated memory 60.39 GiB is allocated by PyTorch, and 208.88 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    module_builder=lambda: lambda x: x.float(),
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 4 has a total capacty of 79.33 GiB of which 6.96 GiB is free. Including non-PyTorch memory, this process has 72.36 GiB memory in use. Of the allocated memory 60.39 GiB is allocated by PyTorch, and 208.88 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    output = model(**micro_batch)
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    module_builder=lambda: lambda x: x.float(),
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 6 has a total capacty of 79.33 GiB of which 6.86 GiB is free. Including non-PyTorch memory, this process has 72.46 GiB memory in use. Of the allocated memory 60.39 GiB is allocated by PyTorch, and 208.88 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    trainer.train(dataloader)
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+Traceback (most recent call last):
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    return self._call_impl(*args, **kwargs)
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+        output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    module_builder=lambda: lambda x: x.float(),
+    return self._call_impl(*args, **kwargs)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 2 has a total capacty of 79.33 GiB of which 6.80 GiB is free. Including non-PyTorch memory, this process has 72.51 GiB memory in use. Of the allocated memory 60.39 GiB is allocated by PyTorch, and 208.88 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    module_builder=lambda: lambda x: x.float(),
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 7 has a total capacty of 79.33 GiB of which 8.04 GiB is free. Including non-PyTorch memory, this process has 71.28 GiB memory in use. Of the allocated memory 60.39 GiB is allocated by PyTorch, and 208.88 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+    return forward_call(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+    module_builder=lambda: lambda x: x.float(),
+    module_builder=lambda: lambda x: x.float(),
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 6 has a total capacty of 79.33 GiB of which 6.80 GiB is free. Including non-PyTorch memory, this process has 72.51 GiB memory in use. Of the allocated memory 60.39 GiB is allocated by PyTorch, and 208.88 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 0 has a total capacty of 79.33 GiB of which 7.00 GiB is free. Including non-PyTorch memory, this process has 72.31 GiB memory in use. Of the allocated memory 60.39 GiB is allocated by PyTorch, and 208.88 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    module_builder=lambda: lambda x: x.float(),
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 6 has a total capacty of 79.33 GiB of which 7.86 GiB is free. Including non-PyTorch memory, this process has 71.46 GiB memory in use. Of the allocated memory 60.39 GiB is allocated by PyTorch, and 208.88 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    module_builder=lambda: lambda x: x.float(),
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 6 has a total capacty of 79.33 GiB of which 6.98 GiB is free. Including non-PyTorch memory, this process has 72.34 GiB memory in use. Of the allocated memory 60.39 GiB is allocated by PyTorch, and 208.88 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    module_builder=lambda: lambda x: x.float(),
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 4 has a total capacty of 79.33 GiB of which 6.78 GiB is free. Including non-PyTorch memory, this process has 72.54 GiB memory in use. Of the allocated memory 60.39 GiB is allocated by PyTorch, and 208.88 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    output = model(**micro_batch)
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    return self._call_impl(*args, **kwargs)
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    return forward_call(*args, **kwargs)
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+    output = model(**micro_batch)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    return forward_call(*args, **kwargs)
+    output = model(**micro_batch)
+    sharded_logits = self.model(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    module_builder=lambda: lambda x: x.float(),
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 3 has a total capacty of 79.33 GiB of which 6.80 GiB is free. Including non-PyTorch memory, this process has 72.51 GiB memory in use. Of the allocated memory 60.39 GiB is allocated by PyTorch, and 208.88 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    sharded_logits = self.model(
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    module_builder=lambda: lambda x: x.float(),
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 5 has a total capacty of 79.33 GiB of which 6.92 GiB is free. Including non-PyTorch memory, this process has 72.40 GiB memory in use. Of the allocated memory 60.39 GiB is allocated by PyTorch, and 208.88 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    module_builder=lambda: lambda x: x.float(),
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 4 has a total capacty of 79.33 GiB of which 6.84 GiB is free. Including non-PyTorch memory, this process has 72.48 GiB memory in use. Of the allocated memory 60.39 GiB is allocated by PyTorch, and 208.88 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    module_builder=lambda: lambda x: x.float(),
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 5 has a total capacty of 79.33 GiB of which 6.86 GiB is free. Including non-PyTorch memory, this process has 72.46 GiB memory in use. Of the allocated memory 60.39 GiB is allocated by PyTorch, and 208.88 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    module_builder=lambda: lambda x: x.float(),
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 2 has a total capacty of 79.33 GiB of which 6.92 GiB is free. Including non-PyTorch memory, this process has 72.40 GiB memory in use. Of the allocated memory 60.39 GiB is allocated by PyTorch, and 208.88 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self._call_impl(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+    return forward_call(*args, **kwargs)
+    return forward_call(*args, **kwargs)
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+    module_builder=lambda: lambda x: x.float(),
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 2 has a total capacty of 79.33 GiB of which 6.86 GiB is free. Including non-PyTorch memory, this process has 72.46 GiB memory in use. Of the allocated memory 60.39 GiB is allocated by PyTorch, and 208.88 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    module_builder=lambda: lambda x: x.float(),
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 3 has a total capacty of 79.33 GiB of which 6.63 GiB is free. Including non-PyTorch memory, this process has 72.69 GiB memory in use. Of the allocated memory 60.39 GiB is allocated by PyTorch, and 208.88 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    module_builder=lambda: lambda x: x.float(),
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 7 has a total capacty of 79.33 GiB of which 7.04 GiB is free. Including non-PyTorch memory, this process has 72.28 GiB memory in use. Of the allocated memory 60.39 GiB is allocated by PyTorch, and 208.88 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    trainer.train(dataloader)
+    module_builder=lambda: lambda x: x.float(),
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 4 has a total capacty of 79.33 GiB of which 6.78 GiB is free. Including non-PyTorch memory, this process has 72.54 GiB memory in use. Of the allocated memory 60.39 GiB is allocated by PyTorch, and 208.88 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    module_builder=lambda: lambda x: x.float(),
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 3 has a total capacty of 79.33 GiB of which 6.63 GiB is free. Including non-PyTorch memory, this process has 72.69 GiB memory in use. Of the allocated memory 60.39 GiB is allocated by PyTorch, and 208.88 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    module_builder=lambda: lambda x: x.float(),
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 7 has a total capacty of 79.33 GiB of which 7.04 GiB is free. Including non-PyTorch memory, this process has 72.28 GiB memory in use. Of the allocated memory 60.39 GiB is allocated by PyTorch, and 208.88 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    module_builder=lambda: lambda x: x.float(),
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 2 has a total capacty of 79.33 GiB of which 6.92 GiB is free. Including non-PyTorch memory, this process has 72.40 GiB memory in use. Of the allocated memory 60.39 GiB is allocated by PyTorch, and 208.88 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    module_builder=lambda: lambda x: x.float(),
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 1 has a total capacty of 79.33 GiB of which 6.92 GiB is free. Including non-PyTorch memory, this process has 72.40 GiB memory in use. Of the allocated memory 60.39 GiB is allocated by PyTorch, and 208.88 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    module_builder=lambda: lambda x: x.float(),
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 3 has a total capacty of 79.33 GiB of which 6.69 GiB is free. Including non-PyTorch memory, this process has 72.63 GiB memory in use. Of the allocated memory 60.39 GiB is allocated by PyTorch, and 208.88 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    outputs = self.pipeline_engine.train_batch_iter(
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    return self._call_impl(*args, **kwargs)
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    sharded_logits = self.model(
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    trainer.train(dataloader)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    sharded_logits = self.model(
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self._call_impl(*args, **kwargs)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    outputs = self.pipeline_engine.train_batch_iter(
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    outputs = self.pipeline_engine.train_batch_iter(
+    return forward_call(*args, **kwargs)
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    outputs = self.pipeline_engine.train_batch_iter(
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+    module_builder=lambda: lambda x: x.float(),
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 4 has a total capacty of 79.33 GiB of which 7.84 GiB is free. Including non-PyTorch memory, this process has 71.48 GiB memory in use. Of the allocated memory 60.39 GiB is allocated by PyTorch, and 208.88 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = model(**micro_batch)
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+Traceback (most recent call last):
+      File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+return self._call_impl(*args, **kwargs)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    return forward_call(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    sharded_logits = self.model(
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    module_builder=lambda: lambda x: x.float(),
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 0 has a total capacty of 79.33 GiB of which 6.89 GiB is free. Including non-PyTorch memory, this process has 72.43 GiB memory in use. Of the allocated memory 60.39 GiB is allocated by PyTorch, and 208.88 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    module_builder=lambda: lambda x: x.float(),
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    sharded_logits = self.model(
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 5 has a total capacty of 79.33 GiB of which 6.86 GiB is free. Including non-PyTorch memory, this process has 72.46 GiB memory in use. Of the allocated memory 60.39 GiB is allocated by PyTorch, and 208.88 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    return self._call_impl(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = model(**micro_batch)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    output = self.pp_block(**new_kwargs)
+    sharded_logits = self.model(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    outputs = self.pipeline_engine.train_batch_iter(
+    return self._call_impl(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    module_builder=lambda: lambda x: x.float(),
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 6 has a total capacty of 79.33 GiB of which 6.92 GiB is free. Including non-PyTorch memory, this process has 72.40 GiB memory in use. Of the allocated memory 60.39 GiB is allocated by PyTorch, and 208.88 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    return forward_call(*args, **kwargs)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    module_builder=lambda: lambda x: x.float(),
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 5 has a total capacty of 79.33 GiB of which 7.86 GiB is free. Including non-PyTorch memory, this process has 71.46 GiB memory in use. Of the allocated memory 60.39 GiB is allocated by PyTorch, and 208.88 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    module_builder=lambda: lambda x: x.float(),
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 5 has a total capacty of 79.33 GiB of which 6.92 GiB is free. Including non-PyTorch memory, this process has 72.40 GiB memory in use. Of the allocated memory 60.39 GiB is allocated by PyTorch, and 208.88 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+    sharded_logits = self.model(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+    output = self.pp_block(**new_kwargs)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    module_builder=lambda: lambda x: x.float(),
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 6 has a total capacty of 79.33 GiB of which 7.86 GiB is free. Including non-PyTorch memory, this process has 71.46 GiB memory in use. Of the allocated memory 60.39 GiB is allocated by PyTorch, and 208.88 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    module_builder=lambda: lambda x: x.float(),
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 4 has a total capacty of 79.33 GiB of which 6.84 GiB is free. Including non-PyTorch memory, this process has 72.48 GiB memory in use. Of the allocated memory 60.39 GiB is allocated by PyTorch, and 208.88 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return forward_call(*args, **kwargs)
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    module_builder=lambda: lambda x: x.float(),
+    outputs = self.pipeline_engine.train_batch_iter(
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 2 has a total capacty of 79.33 GiB of which 6.80 GiB is free. Including non-PyTorch memory, this process has 72.51 GiB memory in use. Of the allocated memory 60.39 GiB is allocated by PyTorch, and 208.88 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    module_builder=lambda: lambda x: x.float(),
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 1 has a total capacty of 79.33 GiB of which 6.86 GiB is free. Including non-PyTorch memory, this process has 72.46 GiB memory in use. Of the allocated memory 60.39 GiB is allocated by PyTorch, and 208.88 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    module_builder=lambda: lambda x: x.float(),
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 3 has a total capacty of 79.33 GiB of which 6.80 GiB is free. Including non-PyTorch memory, this process has 72.51 GiB memory in use. Of the allocated memory 60.39 GiB is allocated by PyTorch, and 208.88 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    module_builder=lambda: lambda x: x.float(),
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 0 has a total capacty of 79.33 GiB of which 6.95 GiB is free. Including non-PyTorch memory, this process has 72.37 GiB memory in use. Of the allocated memory 60.39 GiB is allocated by PyTorch, and 208.88 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    trainer.train(dataloader)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+        outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)output = model(**micro_batch)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    trainer.train(dataloader)    return forward_call(*args, **kwargs)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    module_builder=lambda: lambda x: x.float(),
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 5 has a total capacty of 79.33 GiB of which 6.98 GiB is free. Including non-PyTorch memory, this process has 72.34 GiB memory in use. Of the allocated memory 60.39 GiB is allocated by PyTorch, and 208.88 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    trainer.train(dataloader)
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    module_builder=lambda: lambda x: x.float(),
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 2 has a total capacty of 79.33 GiB of which 6.92 GiB is free. Including non-PyTorch memory, this process has 72.40 GiB memory in use. Of the allocated memory 60.39 GiB is allocated by PyTorch, and 208.88 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    sharded_logits = self.model(
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+        return self._call_impl(*args, **kwargs)return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    module_builder=lambda: lambda x: x.float(),
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 4 has a total capacty of 79.33 GiB of which 6.90 GiB is free. Including non-PyTorch memory, this process has 72.42 GiB memory in use. Of the allocated memory 60.39 GiB is allocated by PyTorch, and 208.88 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return forward_call(*args, **kwargs)
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+    module_builder=lambda: lambda x: x.float(),
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 0 has a total capacty of 79.33 GiB of which 6.95 GiB is free. Including non-PyTorch memory, this process has 72.37 GiB memory in use. Of the allocated memory 60.39 GiB is allocated by PyTorch, and 208.88 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    sharded_logits = self.model(
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    module_builder=lambda: lambda x: x.float(),
+    output = self.pp_block(**new_kwargs)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 5 has a total capacty of 79.33 GiB of which 6.80 GiB is free. Including non-PyTorch memory, this process has 72.51 GiB memory in use. Of the allocated memory 60.39 GiB is allocated by PyTorch, and 208.88 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    module_builder=lambda: lambda x: x.float(),
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 1 has a total capacty of 79.33 GiB of which 6.98 GiB is free. Including non-PyTorch memory, this process has 72.34 GiB memory in use. Of the allocated memory 60.39 GiB is allocated by PyTorch, and 208.88 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    module_builder=lambda: lambda x: x.float(),
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 6 has a total capacty of 79.33 GiB of which 6.80 GiB is free. Including non-PyTorch memory, this process has 72.51 GiB memory in use. Of the allocated memory 60.39 GiB is allocated by PyTorch, and 208.88 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    return self._call_impl(*args, **kwargs)
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+Traceback (most recent call last):
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    module_builder=lambda: lambda x: x.float(),
+    return forward_call(*args, **kwargs)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 6 has a total capacty of 79.33 GiB of which 6.92 GiB is free. Including non-PyTorch memory, this process has 72.40 GiB memory in use. Of the allocated memory 60.39 GiB is allocated by PyTorch, and 208.88 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    output = model(**micro_batch)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    trainer.train(dataloader)
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+        fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    output = model(**micro_batch)
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = model(**micro_batch)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = self.pp_block(**new_kwargs)
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    module_builder=lambda: lambda x: x.float(),
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 7 has a total capacty of 79.33 GiB of which 7.16 GiB is free. Including non-PyTorch memory, this process has 72.16 GiB memory in use. Of the allocated memory 60.39 GiB is allocated by PyTorch, and 208.88 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    output = model(**micro_batch)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return forward_call(*args, **kwargs)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+    sharded_logits = self.model(
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    module_builder=lambda: lambda x: x.float(),
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 4 has a total capacty of 79.33 GiB of which 6.96 GiB is free. Including non-PyTorch memory, this process has 72.36 GiB memory in use. Of the allocated memory 60.39 GiB is allocated by PyTorch, and 208.88 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    output = model(**micro_batch)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return forward_call(*args, **kwargs)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    trainer.train(dataloader)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    return self._call_impl(*args, **kwargs)
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    module_builder=lambda: lambda x: x.float(),
+    module_builder=lambda: lambda x: x.float(),
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 0 has a total capacty of 79.33 GiB of which 6.89 GiB is free. Including non-PyTorch memory, this process has 72.43 GiB memory in use. Of the allocated memory 60.39 GiB is allocated by PyTorch, and 208.88 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 7 has a total capacty of 79.33 GiB of which 6.98 GiB is free. Including non-PyTorch memory, this process has 72.34 GiB memory in use. Of the allocated memory 60.39 GiB is allocated by PyTorch, and 208.88 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    module_builder=lambda: lambda x: x.float(),
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 0 has a total capacty of 79.33 GiB of which 6.83 GiB is free. Including non-PyTorch memory, this process has 72.49 GiB memory in use. Of the allocated memory 60.39 GiB is allocated by PyTorch, and 208.88 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    sharded_logits = self.model(
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+Traceback (most recent call last):
+    sharded_logits = self.model(
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    return self._call_impl(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    return self._call_impl(*args, **kwargs)
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    output = model(**micro_batch)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    module_builder=lambda: lambda x: x.float(),
+    return forward_call(*args, **kwargs)
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 0 has a total capacty of 79.33 GiB of which 7.00 GiB is free. Including non-PyTorch memory, this process has 72.31 GiB memory in use. Of the allocated memory 60.39 GiB is allocated by PyTorch, and 208.88 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    return self._call_impl(*args, **kwargs)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    module_builder=lambda: lambda x: x.float(),
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 1 has a total capacty of 79.33 GiB of which 6.92 GiB is free. Including non-PyTorch memory, this process has 72.40 GiB memory in use. Of the allocated memory 60.39 GiB is allocated by PyTorch, and 208.88 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return forward_call(*args, **kwargs)
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    trainer.train(dataloader)
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    return forward_call(*args, **kwargs)
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    module_builder=lambda: lambda x: x.float(),
+    module_builder=lambda: lambda x: x.float(),
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 7 has a total capacty of 79.33 GiB of which 6.98 GiB is free. Including non-PyTorch memory, this process has 72.34 GiB memory in use. Of the allocated memory 60.39 GiB is allocated by PyTorch, and 208.88 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 1 has a total capacty of 79.33 GiB of which 6.86 GiB is free. Including non-PyTorch memory, this process has 72.46 GiB memory in use. Of the allocated memory 60.39 GiB is allocated by PyTorch, and 208.88 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    module_builder=lambda: lambda x: x.float(),
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 3 has a total capacty of 79.33 GiB of which 6.75 GiB is free. Including non-PyTorch memory, this process has 72.57 GiB memory in use. Of the allocated memory 60.39 GiB is allocated by PyTorch, and 208.88 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return forward_call(*args, **kwargs)
+    module_builder=lambda: lambda x: x.float(),
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 1 has a total capacty of 79.33 GiB of which 6.98 GiB is free. Including non-PyTorch memory, this process has 72.34 GiB memory in use. Of the allocated memory 60.39 GiB is allocated by PyTorch, and 208.88 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    module_builder=lambda: lambda x: x.float(),
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 6 has a total capacty of 79.33 GiB of which 6.86 GiB is free. Including non-PyTorch memory, this process has 72.46 GiB memory in use. Of the allocated memory 60.39 GiB is allocated by PyTorch, and 208.88 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    module_builder=lambda: lambda x: x.float(),
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 0 has a total capacty of 79.33 GiB of which 6.95 GiB is free. Including non-PyTorch memory, this process has 72.37 GiB memory in use. Of the allocated memory 60.39 GiB is allocated by PyTorch, and 208.88 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    module_builder=lambda: lambda x: x.float(),
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 4 has a total capacty of 79.33 GiB of which 6.96 GiB is free. Including non-PyTorch memory, this process has 72.36 GiB memory in use. Of the allocated memory 60.39 GiB is allocated by PyTorch, and 208.88 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    module_builder=lambda: lambda x: x.float(),
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 2 has a total capacty of 79.33 GiB of which 6.86 GiB is free. Including non-PyTorch memory, this process has 72.46 GiB memory in use. Of the allocated memory 60.39 GiB is allocated by PyTorch, and 208.88 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    output = self.pp_block(**new_kwargs)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    module_builder=lambda: lambda x: x.float(),
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 1 has a total capacty of 79.33 GiB of which 6.80 GiB is free. Including non-PyTorch memory, this process has 72.51 GiB memory in use. Of the allocated memory 60.39 GiB is allocated by PyTorch, and 208.88 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    module_builder=lambda: lambda x: x.float(),
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 1 has a total capacty of 79.33 GiB of which 7.86 GiB is free. Including non-PyTorch memory, this process has 71.46 GiB memory in use. Of the allocated memory 60.39 GiB is allocated by PyTorch, and 208.88 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = model(**micro_batch)
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    return forward_call(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    trainer.train(dataloader)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    module_builder=lambda: lambda x: x.float(),
+    output = self.pp_block(**new_kwargs)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 6 has a total capacty of 79.33 GiB of which 6.98 GiB is free. Including non-PyTorch memory, this process has 72.34 GiB memory in use. Of the allocated memory 60.39 GiB is allocated by PyTorch, and 208.88 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    module_builder=lambda: lambda x: x.float(),
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 4 has a total capacty of 79.33 GiB of which 7.84 GiB is free. Including non-PyTorch memory, this process has 71.48 GiB memory in use. Of the allocated memory 60.39 GiB is allocated by PyTorch, and 208.88 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    module_builder=lambda: lambda x: x.float(),
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 1 has a total capacty of 79.33 GiB of which 6.98 GiB is free. Including non-PyTorch memory, this process has 72.34 GiB memory in use. Of the allocated memory 60.39 GiB is allocated by PyTorch, and 208.88 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    outputs = self.pipeline_engine.train_batch_iter(
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    return forward_call(*args, **kwargs)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    return self._call_impl(*args, **kwargs)
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    output = model(**micro_batch)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    sharded_logits = self.model(
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return self._call_impl(*args, **kwargs)
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    return forward_call(*args, **kwargs)
+    module_builder=lambda: lambda x: x.float(),
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 2 has a total capacty of 79.33 GiB of which 6.98 GiB is free. Including non-PyTorch memory, this process has 72.34 GiB memory in use. Of the allocated memory 60.39 GiB is allocated by PyTorch, and 208.88 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    output = model(**micro_batch)
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    output = self.pp_block(**new_kwargs)
+    module_builder=lambda: lambda x: x.float(),
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 7 has a total capacty of 79.33 GiB of which 7.04 GiB is free. Including non-PyTorch memory, this process has 72.28 GiB memory in use. Of the allocated memory 60.39 GiB is allocated by PyTorch, and 208.88 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    module_builder=lambda: lambda x: x.float(),
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 2 has a total capacty of 79.33 GiB of which 6.98 GiB is free. Including non-PyTorch memory, this process has 72.34 GiB memory in use. Of the allocated memory 60.39 GiB is allocated by PyTorch, and 208.88 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    module_builder=lambda: lambda x: x.float(),
+    return forward_call(*args, **kwargs)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 7 has a total capacty of 79.33 GiB of which 7.10 GiB is free. Including non-PyTorch memory, this process has 72.22 GiB memory in use. Of the allocated memory 60.39 GiB is allocated by PyTorch, and 208.88 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+    return self._call_impl(*args, **kwargs)
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    outputs = self.pipeline_engine.train_batch_iter(
+    sharded_logits = self.model(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    return forward_call(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    return self._call_impl(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    module_builder=lambda: lambda x: x.float(),
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 5 has a total capacty of 79.33 GiB of which 7.86 GiB is free. Including non-PyTorch memory, this process has 71.46 GiB memory in use. Of the allocated memory 60.39 GiB is allocated by PyTorch, and 208.88 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    return forward_call(*args, **kwargs)
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    output = self.pp_block(**new_kwargs)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+    module_builder=lambda: lambda x: x.float(),
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 7 has a total capacty of 79.33 GiB of which 7.16 GiB is free. Including non-PyTorch memory, this process has 72.16 GiB memory in use. Of the allocated memory 60.39 GiB is allocated by PyTorch, and 208.88 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+    return self._call_impl(*args, **kwargs)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    outputs = self.pipeline_engine.train_batch_iter(
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    output = self.pp_block(**new_kwargs)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    output = model(**micro_batch)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    module_builder=lambda: lambda x: x.float(),
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 3 has a total capacty of 79.33 GiB of which 7.68 GiB is free. Including non-PyTorch memory, this process has 71.63 GiB memory in use. Of the allocated memory 60.39 GiB is allocated by PyTorch, and 208.88 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = model(**micro_batch)
+    return self._call_impl(*args, **kwargs)
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    module_builder=lambda: lambda x: x.float(),
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 6 has a total capacty of 79.33 GiB of which 6.98 GiB is free. Including non-PyTorch memory, this process has 72.34 GiB memory in use. Of the allocated memory 60.39 GiB is allocated by PyTorch, and 208.88 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return self._call_impl(*args, **kwargs)
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    trainer.train(dataloader)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    module_builder=lambda: lambda x: x.float(),
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 2 has a total capacty of 79.33 GiB of which 6.86 GiB is free. Including non-PyTorch memory, this process has 72.46 GiB memory in use. Of the allocated memory 60.39 GiB is allocated by PyTorch, and 208.88 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    output = model(**micro_batch)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    outputs = self.pipeline_engine.train_batch_iter(
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    module_builder=lambda: lambda x: x.float(),
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 2 has a total capacty of 79.33 GiB of which 7.86 GiB is free. Including non-PyTorch memory, this process has 71.46 GiB memory in use. Of the allocated memory 60.39 GiB is allocated by PyTorch, and 208.88 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    return forward_call(*args, **kwargs)
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return self._call_impl(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.pp_block(**new_kwargs)
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    module_builder=lambda: lambda x: x.float(),
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 6 has a total capacty of 79.33 GiB of which 6.86 GiB is free. Including non-PyTorch memory, this process has 72.46 GiB memory in use. Of the allocated memory 60.39 GiB is allocated by PyTorch, and 208.88 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    module_builder=lambda: lambda x: x.float(),
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 5 has a total capacty of 79.33 GiB of which 6.92 GiB is free. Including non-PyTorch memory, this process has 72.40 GiB memory in use. Of the allocated memory 60.39 GiB is allocated by PyTorch, and 208.88 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    sharded_logits = self.model(
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self._call_impl(*args, **kwargs)
+    module_builder=lambda: lambda x: x.float(),
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 0 has a total capacty of 79.33 GiB of which 6.89 GiB is free. Including non-PyTorch memory, this process has 72.43 GiB memory in use. Of the allocated memory 60.39 GiB is allocated by PyTorch, and 208.88 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    sharded_logits = self.model(
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+    return self._call_impl(*args, **kwargs)
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    module_builder=lambda: lambda x: x.float(),
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 3 has a total capacty of 79.33 GiB of which 6.80 GiB is free. Including non-PyTorch memory, this process has 72.51 GiB memory in use. Of the allocated memory 60.39 GiB is allocated by PyTorch, and 208.88 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+Traceback (most recent call last):
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    module_builder=lambda: lambda x: x.float(),
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 7 has a total capacty of 79.33 GiB of which 6.98 GiB is free. Including non-PyTorch memory, this process has 72.34 GiB memory in use. Of the allocated memory 60.39 GiB is allocated by PyTorch, and 208.88 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    trainer.train(dataloader)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    module_builder=lambda: lambda x: x.float(),
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 0 has a total capacty of 79.33 GiB of which 6.83 GiB is free. Including non-PyTorch memory, this process has 72.49 GiB memory in use. Of the allocated memory 60.39 GiB is allocated by PyTorch, and 208.88 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    outputs = self.pipeline_engine.train_batch_iter(
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    return self._call_impl(*args, **kwargs)
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    output = model(**micro_batch)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = model(**micro_batch)
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    sharded_logits = self.model(
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return forward_call(*args, **kwargs)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+    trainer.train(dataloader)
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+    module_builder=lambda: lambda x: x.float(),
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 6 has a total capacty of 79.33 GiB of which 6.92 GiB is free. Including non-PyTorch memory, this process has 72.40 GiB memory in use. Of the allocated memory 60.39 GiB is allocated by PyTorch, and 208.88 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    module_builder=lambda: lambda x: x.float(),
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 5 has a total capacty of 79.33 GiB of which 6.80 GiB is free. Including non-PyTorch memory, this process has 72.51 GiB memory in use. Of the allocated memory 60.39 GiB is allocated by PyTorch, and 208.88 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self._call_impl(*args, **kwargs)
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+    return forward_call(*args, **kwargs)
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    module_builder=lambda: lambda x: x.float(),
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 7 has a total capacty of 79.33 GiB of which 7.16 GiB is free. Including non-PyTorch memory, this process has 72.16 GiB memory in use. Of the allocated memory 60.39 GiB is allocated by PyTorch, and 208.88 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    module_builder=lambda: lambda x: x.float(),
+    return self._call_impl(*args, **kwargs)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 4 has a total capacty of 79.33 GiB of which 6.90 GiB is free. Including non-PyTorch memory, this process has 72.42 GiB memory in use. Of the allocated memory 60.39 GiB is allocated by PyTorch, and 208.88 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    module_builder=lambda: lambda x: x.float(),
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 2 has a total capacty of 79.33 GiB of which 6.98 GiB is free. Including non-PyTorch memory, this process has 72.34 GiB memory in use. Of the allocated memory 60.39 GiB is allocated by PyTorch, and 208.88 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    module_builder=lambda: lambda x: x.float(),
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 1 has a total capacty of 79.33 GiB of which 6.86 GiB is free. Including non-PyTorch memory, this process has 72.46 GiB memory in use. Of the allocated memory 60.39 GiB is allocated by PyTorch, and 208.88 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    return self._call_impl(*args, **kwargs)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    module_builder=lambda: lambda x: x.float(),
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 0 has a total capacty of 79.33 GiB of which 6.89 GiB is free. Including non-PyTorch memory, this process has 72.43 GiB memory in use. Of the allocated memory 60.39 GiB is allocated by PyTorch, and 208.88 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    return self._call_impl(*args, **kwargs)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+module_builder=lambda: lambda x: x.float(),
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 2 has a total capacty of 79.33 GiB of which 7.86 GiB is free. Including non-PyTorch memory, this process has 71.46 GiB memory in use. Of the allocated memory 60.39 GiB is allocated by PyTorch, and 208.88 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    output = model(**micro_batch)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    return forward_call(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    return forward_call(*args, **kwargs)
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    sharded_logits = self.model(
+    module_builder=lambda: lambda x: x.float(),
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 7 has a total capacty of 79.33 GiB of which 8.04 GiB is free. Including non-PyTorch memory, this process has 71.28 GiB memory in use. Of the allocated memory 60.39 GiB is allocated by PyTorch, and 208.88 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    return self._call_impl(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    trainer.train(dataloader)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    module_builder=lambda: lambda x: x.float(),
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 7 has a total capacty of 79.33 GiB of which 7.04 GiB is free. Including non-PyTorch memory, this process has 72.28 GiB memory in use. Of the allocated memory 60.39 GiB is allocated by PyTorch, and 208.88 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    output = self.pp_block(**new_kwargs)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+Traceback (most recent call last):
+    module_builder=lambda: lambda x: x.float(),
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 5 has a total capacty of 79.33 GiB of which 6.86 GiB is free. Including non-PyTorch memory, this process has 72.46 GiB memory in use. Of the allocated memory 60.39 GiB is allocated by PyTorch, and 208.88 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    output = model(**micro_batch)
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+        return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]trainer.train(dataloader)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+        fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    return forward_call(*args, **kwargs)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+        return self._call_impl(*args, **kwargs)module_builder=lambda: lambda x: x.float(),
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 3 has a total capacty of 79.33 GiB of which 6.80 GiB is free. Including non-PyTorch memory, this process has 72.51 GiB memory in use. Of the allocated memory 60.39 GiB is allocated by PyTorch, and 208.88 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    trainer.train(dataloader)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+    trainer.train(dataloader)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    output = model(**micro_batch)
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return forward_call(*args, **kwargs)
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    module_builder=lambda: lambda x: x.float(),
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 1 has a total capacty of 79.33 GiB of which 7.86 GiB is free. Including non-PyTorch memory, this process has 71.46 GiB memory in use. Of the allocated memory 60.39 GiB is allocated by PyTorch, and 208.88 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    module_builder=lambda: lambda x: x.float(),
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 5 has a total capacty of 79.33 GiB of which 6.98 GiB is free. Including non-PyTorch memory, this process has 72.34 GiB memory in use. Of the allocated memory 60.39 GiB is allocated by PyTorch, and 208.88 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+        return self._call_impl(*args, **kwargs)return self._call_impl(*args, **kwargs)
+
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+        sharded_logits = self.model(fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    return self._call_impl(*args, **kwargs)
+    module_builder=lambda: lambda x: x.float(),
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 4 has a total capacty of 79.33 GiB of which 6.84 GiB is free. Including non-PyTorch memory, this process has 72.48 GiB memory in use. Of the allocated memory 60.39 GiB is allocated by PyTorch, and 208.88 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return self._call_impl(*args, **kwargs)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    module_builder=lambda: lambda x: x.float(),
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 5 has a total capacty of 79.33 GiB of which 6.86 GiB is free. Including non-PyTorch memory, this process has 72.46 GiB memory in use. Of the allocated memory 60.39 GiB is allocated by PyTorch, and 208.88 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    module_builder=lambda: lambda x: x.float(),
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 6 has a total capacty of 79.33 GiB of which 6.98 GiB is free. Including non-PyTorch memory, this process has 72.34 GiB memory in use. Of the allocated memory 60.39 GiB is allocated by PyTorch, and 208.88 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    return self._call_impl(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    module_builder=lambda: lambda x: x.float(),
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 0 has a total capacty of 79.33 GiB of which 7.00 GiB is free. Including non-PyTorch memory, this process has 72.31 GiB memory in use. Of the allocated memory 60.39 GiB is allocated by PyTorch, and 208.88 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return forward_call(*args, **kwargs)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    module_builder=lambda: lambda x: x.float(),
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 3 has a total capacty of 79.33 GiB of which 6.69 GiB is free. Including non-PyTorch memory, this process has 72.63 GiB memory in use. Of the allocated memory 60.39 GiB is allocated by PyTorch, and 208.88 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    module_builder=lambda: lambda x: x.float(),
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 1 has a total capacty of 79.33 GiB of which 6.86 GiB is free. Including non-PyTorch memory, this process has 72.46 GiB memory in use. Of the allocated memory 60.39 GiB is allocated by PyTorch, and 208.88 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    module_builder=lambda: lambda x: x.float(),
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 0 has a total capacty of 79.33 GiB of which 7.88 GiB is free. Including non-PyTorch memory, this process has 71.44 GiB memory in use. Of the allocated memory 60.39 GiB is allocated by PyTorch, and 208.88 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    sharded_logits = self.model(
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    module_builder=lambda: lambda x: x.float(),
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 7 has a total capacty of 79.33 GiB of which 7.16 GiB is free. Including non-PyTorch memory, this process has 72.16 GiB memory in use. Of the allocated memory 60.39 GiB is allocated by PyTorch, and 208.88 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    module_builder=lambda: lambda x: x.float(),
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 0 has a total capacty of 79.33 GiB of which 7.00 GiB is free. Including non-PyTorch memory, this process has 72.31 GiB memory in use. Of the allocated memory 60.39 GiB is allocated by PyTorch, and 208.88 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    module_builder=lambda: lambda x: x.float(),
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 3 has a total capacty of 79.33 GiB of which 6.69 GiB is free. Including non-PyTorch memory, this process has 72.63 GiB memory in use. Of the allocated memory 60.39 GiB is allocated by PyTorch, and 208.88 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    module_builder=lambda: lambda x: x.float(),
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 2 has a total capacty of 79.33 GiB of which 6.86 GiB is free. Including non-PyTorch memory, this process has 72.46 GiB memory in use. Of the allocated memory 60.39 GiB is allocated by PyTorch, and 208.88 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    module_builder=lambda: lambda x: x.float(),
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 5 has a total capacty of 79.33 GiB of which 6.98 GiB is free. Including non-PyTorch memory, this process has 72.34 GiB memory in use. Of the allocated memory 60.39 GiB is allocated by PyTorch, and 208.88 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    module_builder=lambda: lambda x: x.float(),
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 0 has a total capacty of 79.33 GiB of which 6.83 GiB is free. Including non-PyTorch memory, this process has 72.49 GiB memory in use. Of the allocated memory 60.39 GiB is allocated by PyTorch, and 208.88 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    module_builder=lambda: lambda x: x.float(),
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 7 has a total capacty of 79.33 GiB of which 7.10 GiB is free. Including non-PyTorch memory, this process has 72.22 GiB memory in use. Of the allocated memory 60.39 GiB is allocated by PyTorch, and 208.88 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    module_builder=lambda: lambda x: x.float(),
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 4 has a total capacty of 79.33 GiB of which 6.78 GiB is free. Including non-PyTorch memory, this process has 72.54 GiB memory in use. Of the allocated memory 60.39 GiB is allocated by PyTorch, and 208.88 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    module_builder=lambda: lambda x: x.float(),
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 3 has a total capacty of 79.33 GiB of which 6.75 GiB is free. Including non-PyTorch memory, this process has 72.57 GiB memory in use. Of the allocated memory 60.39 GiB is allocated by PyTorch, and 208.88 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    module_builder=lambda: lambda x: x.float(),
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 2 has a total capacty of 79.33 GiB of which 6.80 GiB is free. Including non-PyTorch memory, this process has 72.51 GiB memory in use. Of the allocated memory 60.39 GiB is allocated by PyTorch, and 208.88 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    module_builder=lambda: lambda x: x.float(),
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 6 has a total capacty of 79.33 GiB of which 6.80 GiB is free. Including non-PyTorch memory, this process has 72.51 GiB memory in use. Of the allocated memory 60.39 GiB is allocated by PyTorch, and 208.88 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    module_builder=lambda: lambda x: x.float(),
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 1 has a total capacty of 79.33 GiB of which 6.92 GiB is free. Including non-PyTorch memory, this process has 72.40 GiB memory in use. Of the allocated memory 60.39 GiB is allocated by PyTorch, and 208.88 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+[2024-12-09 23:30:15,367] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3767348 closing signal SIGTERM
+[2024-12-09 23:30:15,369] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1200152 closing signal SIGTERM
+[2024-12-09 23:30:15,683] torch.distributed.elastic.multiprocessing.api: [ERROR] failed (exitcode: 1) local_rank: 0 (pid: 1200150) of binary: /fsx/nouamane/miniconda/envs/2-1-cu121/bin/python
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 33, in <module>
+    sys.exit(load_entry_point('torch==2.1.1', 'console_scripts', 'torchrun')())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 264, in launch_agent
+    raise ChildFailedError(
+torch.distributed.elastic.multiprocessing.errors.ChildFailedError: 
+============================================================
+run_train.py FAILED
+------------------------------------------------------------
+Failures:
+[1]:
+  time      : 2024-12-09_23:30:15
+  host      : ip-26-0-171-88.ec2.internal
+  rank      : 121 (local_rank: 1)
+  exitcode  : 1 (pid: 1200151)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[2]:
+  time      : 2024-12-09_23:30:15
+  host      : ip-26-0-171-88.ec2.internal
+  rank      : 123 (local_rank: 3)
+  exitcode  : 1 (pid: 1200153)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[3]:
+  time      : 2024-12-09_23:30:15
+  host      : ip-26-0-171-88.ec2.internal
+  rank      : 124 (local_rank: 4)
+  exitcode  : 1 (pid: 1200154)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[4]:
+  time      : 2024-12-09_23:30:15
+  host      : ip-26-0-171-88.ec2.internal
+  rank      : 125 (local_rank: 5)
+  exitcode  : 1 (pid: 1200155)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[5]:
+  time      : 2024-12-09_23:30:15
+  host      : ip-26-0-171-88.ec2.internal
+  rank      : 126 (local_rank: 6)
+  exitcode  : 1 (pid: 1200156)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[6]:
+  time      : 2024-12-09_23:30:15
+  host      : ip-26-0-171-88.ec2.internal
+  rank      : 127 (local_rank: 7)
+  exitcode  : 1 (pid: 1200157)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+------------------------------------------------------------
+Root Cause (first observed failure):
+[0]:
+  time      : 2024-12-09_23:30:15
+  host      : ip-26-0-171-88.ec2.internal
+  rank      : 120 (local_rank: 0)
+  exitcode  : 1 (pid: 1200150)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+============================================================
+srun: error: ip-26-0-171-88: task 12: Exited with exit code 1
+[2024-12-09 23:30:18,386] torch.distributed.elastic.multiprocessing.api: [ERROR] failed (exitcode: 1) local_rank: 1 (pid: 3767349) of binary: /fsx/nouamane/miniconda/envs/2-1-cu121/bin/python
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 33, in <module>
+    sys.exit(load_entry_point('torch==2.1.1', 'console_scripts', 'torchrun')())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 264, in launch_agent
+    raise ChildFailedError(
+torch.distributed.elastic.multiprocessing.errors.ChildFailedError: 
+============================================================
+run_train.py FAILED
+------------------------------------------------------------
+Failures:
+[1]:
+  time      : 2024-12-09_23:30:15
+  host      : ip-26-0-164-187.ec2.internal
+  rank      : 2 (local_rank: 2)
+  exitcode  : 1 (pid: 3767350)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[2]:
+  time      : 2024-12-09_23:30:15
+  host      : ip-26-0-164-187.ec2.internal
+  rank      : 3 (local_rank: 3)
+  exitcode  : 1 (pid: 3767351)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[3]:
+  time      : 2024-12-09_23:30:15
+  host      : ip-26-0-164-187.ec2.internal
+  rank      : 4 (local_rank: 4)
+  exitcode  : 1 (pid: 3767352)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[4]:
+  time      : 2024-12-09_23:30:15
+  host      : ip-26-0-164-187.ec2.internal
+  rank      : 5 (local_rank: 5)
+  exitcode  : 1 (pid: 3767353)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[5]:
+  time      : 2024-12-09_23:30:15
+  host      : ip-26-0-164-187.ec2.internal
+  rank      : 6 (local_rank: 6)
+  exitcode  : 1 (pid: 3767354)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[6]:
+  time      : 2024-12-09_23:30:15
+  host      : ip-26-0-164-187.ec2.internal
+  rank      : 7 (local_rank: 7)
+  exitcode  : 1 (pid: 3767355)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+------------------------------------------------------------
+Root Cause (first observed failure):
+[0]:
+  time      : 2024-12-09_23:30:15
+  host      : ip-26-0-164-187.ec2.internal
+  rank      : 1 (local_rank: 1)
+  exitcode  : 1 (pid: 3767349)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+============================================================
+srun: error: ip-26-0-164-187: task 2: Exited with exit code 1
+[2024-12-09 23:30:20,362] torch.distributed.elastic.multiprocessing.api: [ERROR] failed (exitcode: 1) local_rank: 0 (pid: 59337) of binary: /fsx/nouamane/miniconda/envs/2-1-cu121/bin/python
+[2024-12-09 23:30:20,365] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 38262 closing signal SIGTERM
+[2024-12-09 23:30:20,365] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 38266 closing signal SIGTERM
+[2024-12-09 23:30:20,365] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 38268 closing signal SIGTERM
+[2024-12-09 23:30:20,366] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2698937 closing signal SIGTERM
+[2024-12-09 23:30:20,367] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2698938 closing signal SIGTERM
+[2024-12-09 23:30:20,367] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2698941 closing signal SIGTERM
+[2024-12-09 23:30:20,369] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2943353 closing signal SIGTERM
+[2024-12-09 23:30:20,370] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 98595 closing signal SIGTERM
+[2024-12-09 23:30:20,371] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2076231 closing signal SIGTERM
+[2024-12-09 23:30:20,371] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2076233 closing signal SIGTERM
+[2024-12-09 23:30:20,371] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2076235 closing signal SIGTERM
+[2024-12-09 23:30:20,371] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2076237 closing signal SIGTERM
+[2024-12-09 23:30:20,371] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2555935 closing signal SIGTERM
+[2024-12-09 23:30:20,371] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2555937 closing signal SIGTERM
+[2024-12-09 23:30:20,371] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2988336 closing signal SIGTERM
+[2024-12-09 23:30:20,372] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1241061 closing signal SIGTERM
+[2024-12-09 23:30:20,374] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2773673 closing signal SIGTERM
+[2024-12-09 23:30:20,375] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2773674 closing signal SIGTERM
+[2024-12-09 23:30:20,375] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2773675 closing signal SIGTERM
+[2024-12-09 23:30:20,375] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2773677 closing signal SIGTERM
+[2024-12-09 23:30:20,375] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2773679 closing signal SIGTERM
+[2024-12-09 23:30:20,370] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 40605 closing signal SIGTERM
+[2024-12-09 23:30:20,371] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 40606 closing signal SIGTERM
+[2024-12-09 23:30:20,377] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1780859 closing signal SIGTERM
+[2024-12-09 23:30:20,378] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1780861 closing signal SIGTERM
+[2024-12-09 23:30:20,379] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 896951 closing signal SIGTERM
+[2024-12-09 23:30:20,379] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 896955 closing signal SIGTERM
+[2024-12-09 23:30:20,379] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 896956 closing signal SIGTERM
+[2024-12-09 23:30:20,379] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 896957 closing signal SIGTERM
+[2024-12-09 23:30:20,378] torch.distributed.elastic.multiprocessing.api: [ERROR] failed (exitcode: 1) local_rank: 0 (pid: 1541235) of binary: /fsx/nouamane/miniconda/envs/2-1-cu121/bin/python
+[2024-12-09 23:30:20,379] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 896958 closing signal SIGTERM
+[2024-12-09 23:30:20,385] torch.distributed.elastic.multiprocessing.api: [ERROR] failed (exitcode: 1) local_rank: 0 (pid: 98592) of binary: /fsx/nouamane/miniconda/envs/2-1-cu121/bin/python
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 33, in <module>
+    sys.exit(load_entry_point('torch==2.1.1', 'console_scripts', 'torchrun')())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 264, in launch_agent
+    raise ChildFailedError(
+torch.distributed.elastic.multiprocessing.errors.ChildFailedError: 
+============================================================
+run_train.py FAILED
+------------------------------------------------------------
+Failures:
+[1]:
+  time      : 2024-12-09_23:30:20
+  host      : ip-26-0-166-244.ec2.internal
+  rank      : 65 (local_rank: 1)
+  exitcode  : 1 (pid: 59338)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[2]:
+  time      : 2024-12-09_23:30:20
+  host      : ip-26-0-166-244.ec2.internal
+  rank      : 66 (local_rank: 2)
+  exitcode  : 1 (pid: 59339)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[3]:
+  time      : 2024-12-09_23:30:20
+  host      : ip-26-0-166-244.ec2.internal
+  rank      : 67 (local_rank: 3)
+  exitcode  : 1 (pid: 59340)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[4]:
+  time      : 2024-12-09_23:30:20
+  host      : ip-26-0-166-244.ec2.internal
+  rank      : 68 (local_rank: 4)
+  exitcode  : 1 (pid: 59341)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[5]:
+  time      : 2024-12-09_23:30:20
+  host      : ip-26-0-166-244.ec2.internal
+  rank      : 69 (local_rank: 5)
+  exitcode  : 1 (pid: 59342)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[6]:
+  time      : 2024-12-09_23:30:20
+  host      : ip-26-0-166-244.ec2.internal
+  rank      : 70 (local_rank: 6)
+  exitcode  : 1 (pid: 59343)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[7]:
+  time      : 2024-12-09_23:30:20
+  host      : ip-26-0-166-244.ec2.internal
+  rank      : 71 (local_rank: 7)
+  exitcode  : 1 (pid: 59344)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+------------------------------------------------------------
+Root Cause (first observed failure):
+[0]:
+  time      : 2024-12-09_23:30:20
+  host      : ip-26-0-166-244.ec2.internal
+  rank      : 64 (local_rank: 0)
+  exitcode  : 1 (pid: 59337)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+============================================================
+[2024-12-09 23:30:20,436] torch.distributed.elastic.multiprocessing.api: [ERROR] failed (exitcode: 1) local_rank: 1 (pid: 1241062) of binary: /fsx/nouamane/miniconda/envs/2-1-cu121/bin/python
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 33, in <module>
+    sys.exit(load_entry_point('torch==2.1.1', 'console_scripts', 'torchrun')())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 33, in <module>
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    sys.exit(load_entry_point('torch==2.1.1', 'console_scripts', 'torchrun')())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 264, in launch_agent
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    raise ChildFailedError(
+torch.distributed.elastic.multiprocessing.errors.ChildFailedError: 
+============================================================
+run_train.py FAILED
+------------------------------------------------------------
+Failures:
+[1]:
+  time      : 2024-12-09_23:30:20
+  host      : ip-26-0-166-36.ec2.internal
+  rank      : 73 (local_rank: 1)
+  exitcode  : 1 (pid: 1541236)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[2]:
+  time      : 2024-12-09_23:30:20
+  host      : ip-26-0-166-36.ec2.internal
+  rank      : 74 (local_rank: 2)
+  exitcode  : 1 (pid: 1541237)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[3]:
+  time      : 2024-12-09_23:30:20
+  host      : ip-26-0-166-36.ec2.internal
+  rank      : 75 (local_rank: 3)
+  exitcode  : 1 (pid: 1541238)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[4]:
+  time      : 2024-12-09_23:30:20
+  host      : ip-26-0-166-36.ec2.internal
+  rank      : 76 (local_rank: 4)
+  exitcode  : 1 (pid: 1541239)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[5]:
+  time      : 2024-12-09_23:30:20
+  host      : ip-26-0-166-36.ec2.internal
+  rank      : 77 (local_rank: 5)
+  exitcode  : 1 (pid: 1541240)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[6]:
+  time      : 2024-12-09_23:30:20
+  host      : ip-26-0-166-36.ec2.internal
+  rank      : 78 (local_rank: 6)
+  exitcode  : 1 (pid: 1541241)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[7]:
+  time      : 2024-12-09_23:30:20
+  host      : ip-26-0-166-36.ec2.internal
+  rank      : 79 (local_rank: 7)
+  exitcode  : 1 (pid: 1541242)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+------------------------------------------------------------
+Root Cause (first observed failure):
+[0]:
+  time      : 2024-12-09_23:30:20
+  host      : ip-26-0-166-36.ec2.internal
+  rank      : 72 (local_rank: 0)
+  exitcode  : 1 (pid: 1541235)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+============================================================
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 264, in launch_agent
+    raise ChildFailedError(
+torch.distributed.elastic.multiprocessing.errors.ChildFailedError: 
+============================================================
+run_train.py FAILED
+------------------------------------------------------------
+Failures:
+[1]:
+  time      : 2024-12-09_23:30:20
+  host      : ip-26-0-167-51.ec2.internal
+  rank      : 81 (local_rank: 1)
+  exitcode  : 1 (pid: 98593)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[2]:
+  time      : 2024-12-09_23:30:20
+  host      : ip-26-0-167-51.ec2.internal
+  rank      : 82 (local_rank: 2)
+  exitcode  : 1 (pid: 98594)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[3]:
+  time      : 2024-12-09_23:30:20
+  host      : ip-26-0-167-51.ec2.internal
+  rank      : 84 (local_rank: 4)
+  exitcode  : 1 (pid: 98596)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[4]:
+  time      : 2024-12-09_23:30:20
+  host      : ip-26-0-167-51.ec2.internal
+  rank      : 85 (local_rank: 5)
+  exitcode  : 1 (pid: 98597)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[5]:
+  time      : 2024-12-09_23:30:20
+  host      : ip-26-0-167-51.ec2.internal
+  rank      : 86 (local_rank: 6)
+  exitcode  : 1 (pid: 98598)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[6]:
+  time      : 2024-12-09_23:30:20
+  host      : ip-26-0-167-51.ec2.internal
+  rank      : 87 (local_rank: 7)
+  exitcode  : 1 (pid: 98599)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+------------------------------------------------------------
+Root Cause (first observed failure):
+[0]:
+  time      : 2024-12-09_23:30:20
+  host      : ip-26-0-167-51.ec2.internal
+  rank      : 80 (local_rank: 0)
+  exitcode  : 1 (pid: 98592)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+============================================================
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 33, in <module>
+    sys.exit(load_entry_point('torch==2.1.1', 'console_scripts', 'torchrun')())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 264, in launch_agent
+    raise ChildFailedError(
+torch.distributed.elastic.multiprocessing.errors.ChildFailedError: 
+============================================================
+run_train.py FAILED
+------------------------------------------------------------
+Failures:
+[1]:
+  time      : 2024-12-09_23:30:20
+  host      : ip-26-0-166-125.ec2.internal
+  rank      : 42 (local_rank: 2)
+  exitcode  : 1 (pid: 1241063)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[2]:
+  time      : 2024-12-09_23:30:20
+  host      : ip-26-0-166-125.ec2.internal
+  rank      : 43 (local_rank: 3)
+  exitcode  : 1 (pid: 1241064)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[3]:
+  time      : 2024-12-09_23:30:20
+  host      : ip-26-0-166-125.ec2.internal
+  rank      : 44 (local_rank: 4)
+  exitcode  : 1 (pid: 1241065)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[4]:
+  time      : 2024-12-09_23:30:20
+  host      : ip-26-0-166-125.ec2.internal
+  rank      : 45 (local_rank: 5)
+  exitcode  : 1 (pid: 1241066)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[5]:
+  time      : 2024-12-09_23:30:20
+  host      : ip-26-0-166-125.ec2.internal
+  rank      : 46 (local_rank: 6)
+  exitcode  : 1 (pid: 1241067)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[6]:
+  time      : 2024-12-09_23:30:20
+  host      : ip-26-0-166-125.ec2.internal
+  rank      : 47 (local_rank: 7)
+  exitcode  : 1 (pid: 1241068)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+------------------------------------------------------------
+Root Cause (first observed failure):
+[0]:
+  time      : 2024-12-09_23:30:20
+  host      : ip-26-0-166-125.ec2.internal
+  rank      : 41 (local_rank: 1)
+  exitcode  : 1 (pid: 1241062)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+============================================================
+[2024-12-09 23:30:20,584] torch.distributed.elastic.multiprocessing.api: [ERROR] failed (exitcode: 1) local_rank: 0 (pid: 40602) of binary: /fsx/nouamane/miniconda/envs/2-1-cu121/bin/python
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 33, in <module>
+[2024-12-09 23:30:20,635] torch.distributed.elastic.multiprocessing.api: [ERROR] failed (exitcode: 1) local_rank: 0 (pid: 2988329) of binary: /fsx/nouamane/miniconda/envs/2-1-cu121/bin/python
+    sys.exit(load_entry_point('torch==2.1.1', 'console_scripts', 'torchrun')())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 264, in launch_agent
+    raise ChildFailedError(
+torch.distributed.elastic.multiprocessing.errors.ChildFailedError: 
+============================================================
+run_train.py FAILED
+------------------------------------------------------------
+Failures:
+[1]:
+  time      : 2024-12-09_23:30:20
+  host      : ip-26-0-166-214.ec2.internal
+  rank      : 57 (local_rank: 1)
+  exitcode  : 1 (pid: 40603)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[2]:
+  time      : 2024-12-09_23:30:20
+  host      : ip-26-0-166-214.ec2.internal
+  rank      : 58 (local_rank: 2)
+  exitcode  : 1 (pid: 40604)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[3]:
+  time      : 2024-12-09_23:30:20
+  host      : ip-26-0-166-214.ec2.internal
+  rank      : 61 (local_rank: 5)
+  exitcode  : 1 (pid: 40607)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[4]:
+  time      : 2024-12-09_23:30:20
+  host      : ip-26-0-166-214.ec2.internal
+  rank      : 62 (local_rank: 6)
+  exitcode  : 1 (pid: 40608)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[5]:
+  time      : 2024-12-09_23:30:20
+  host      : ip-26-0-166-214.ec2.internal
+  rank      : 63 (local_rank: 7)
+  exitcode  : 1 (pid: 40609)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+------------------------------------------------------------
+Root Cause (first observed failure):
+[0]:
+  time      : 2024-12-09_23:30:20
+  host      : ip-26-0-166-214.ec2.internal
+  rank      : 56 (local_rank: 0)
+  exitcode  : 1 (pid: 40602)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+============================================================
+[2024-12-09 23:30:20,683] torch.distributed.elastic.multiprocessing.api: [ERROR] failed (exitcode: 1) local_rank: 0 (pid: 2943352) of binary: /fsx/nouamane/miniconda/envs/2-1-cu121/bin/python
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 33, in <module>
+    sys.exit(load_entry_point('torch==2.1.1', 'console_scripts', 'torchrun')())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+[2024-12-09 23:30:20,692] torch.distributed.elastic.multiprocessing.api: [ERROR] failed (exitcode: 1) local_rank: 0 (pid: 1780856) of binary: /fsx/nouamane/miniconda/envs/2-1-cu121/bin/python
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 264, in launch_agent
+    raise ChildFailedError(
+torch.distributed.elastic.multiprocessing.errors.ChildFailedError: 
+============================================================
+run_train.py FAILED
+------------------------------------------------------------
+Failures:
+[1]:
+  time      : 2024-12-09_23:30:20
+  host      : ip-26-0-171-102.ec2.internal
+  rank      : 97 (local_rank: 1)
+  exitcode  : 1 (pid: 2988330)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[2]:
+  time      : 2024-12-09_23:30:20
+  host      : ip-26-0-171-102.ec2.internal
+  rank      : 98 (local_rank: 2)
+  exitcode  : 1 (pid: 2988331)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[3]:
+  time      : 2024-12-09_23:30:20
+  host      : ip-26-0-171-102.ec2.internal
+  rank      : 99 (local_rank: 3)
+  exitcode  : 1 (pid: 2988332)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[4]:
+  time      : 2024-12-09_23:30:20
+  host      : ip-26-0-171-102.ec2.internal
+  rank      : 100 (local_rank: 4)
+  exitcode  : 1 (pid: 2988333)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[5]:
+  time      : 2024-12-09_23:30:20
+  host      : ip-26-0-171-102.ec2.internal
+  rank      : 101 (local_rank: 5)
+  exitcode  : 1 (pid: 2988334)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[6]:
+  time      : 2024-12-09_23:30:20
+  host      : ip-26-0-171-102.ec2.internal
+  rank      : 102 (local_rank: 6)
+  exitcode  : 1 (pid: 2988335)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+------------------------------------------------------------
+Root Cause (first observed failure):
+[0]:
+  time      : 2024-12-09_23:30:20
+  host      : ip-26-0-171-102.ec2.internal
+  rank      : 96 (local_rank: 0)
+  exitcode  : 1 (pid: 2988329)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+============================================================
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 33, in <module>
+    sys.exit(load_entry_point('torch==2.1.1', 'console_scripts', 'torchrun')())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 264, in launch_agent
+    raise ChildFailedError(
+torch.distributed.elastic.multiprocessing.errors.ChildFailedError: 
+============================================================
+run_train.py FAILED
+------------------------------------------------------------
+Failures:
+[1]:
+  time      : 2024-12-09_23:30:20
+  host      : ip-26-0-171-230.ec2.internal
+  rank      : 114 (local_rank: 2)
+  exitcode  : 1 (pid: 2943354)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[2]:
+  time      : 2024-12-09_23:30:20
+  host      : ip-26-0-171-230.ec2.internal
+  rank      : 115 (local_rank: 3)
+  exitcode  : 1 (pid: 2943355)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[3]:
+  time      : 2024-12-09_23:30:20
+  host      : ip-26-0-171-230.ec2.internal
+  rank      : 116 (local_rank: 4)
+  exitcode  : 1 (pid: 2943356)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[4]:
+  time      : 2024-12-09_23:30:20
+  host      : ip-26-0-171-230.ec2.internal
+  rank      : 117 (local_rank: 5)
+  exitcode  : 1 (pid: 2943357)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[5]:
+  time      : 2024-12-09_23:30:20
+  host      : ip-26-0-171-230.ec2.internal
+  rank      : 118 (local_rank: 6)
+  exitcode  : 1 (pid: 2943358)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[6]:
+  time      : 2024-12-09_23:30:20
+  host      : ip-26-0-171-230.ec2.internal
+  rank      : 119 (local_rank: 7)
+  exitcode  : 1 (pid: 2943359)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+------------------------------------------------------------
+Root Cause (first observed failure):
+[0]:
+  time      : 2024-12-09_23:30:20
+  host      : ip-26-0-171-230.ec2.internal
+  rank      : 112 (local_rank: 0)
+  exitcode  : 1 (pid: 2943352)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+============================================================
+srun: error: ip-26-0-166-125: task 7: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 33, in <module>
+    sys.exit(load_entry_point('torch==2.1.1', 'console_scripts', 'torchrun')())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 264, in launch_agent
+    raise ChildFailedError(
+torch.distributed.elastic.multiprocessing.errors.ChildFailedError: 
+============================================================
+run_train.py FAILED
+------------------------------------------------------------
+Failures:
+[1]:
+  time      : 2024-12-09_23:30:20
+  host      : ip-26-0-166-15.ec2.internal
+  rank      : 49 (local_rank: 1)
+  exitcode  : 1 (pid: 1780857)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[2]:
+  time      : 2024-12-09_23:30:20
+  host      : ip-26-0-166-15.ec2.internal
+  rank      : 50 (local_rank: 2)
+  exitcode  : 1 (pid: 1780858)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[3]:
+  time      : 2024-12-09_23:30:20
+  host      : ip-26-0-166-15.ec2.internal
+  rank      : 52 (local_rank: 4)
+  exitcode  : 1 (pid: 1780860)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[4]:
+  time      : 2024-12-09_23:30:20
+  host      : ip-26-0-166-15.ec2.internal
+  rank      : 54 (local_rank: 6)
+  exitcode  : 1 (pid: 1780862)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[5]:
+  time      : 2024-12-09_23:30:20
+  host      : ip-26-0-166-15.ec2.internal
+  rank      : 55 (local_rank: 7)
+  exitcode  : 1 (pid: 1780863)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+------------------------------------------------------------
+Root Cause (first observed failure):
+[0]:
+  time      : 2024-12-09_23:30:20
+  host      : ip-26-0-166-15.ec2.internal
+  rank      : 48 (local_rank: 0)
+  exitcode  : 1 (pid: 1780856)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+============================================================
+srun: error: ip-26-0-167-51: task 11: Exited with exit code 1
+srun: error: ip-26-0-166-244: task 9: Exited with exit code 1
+srun: error: ip-26-0-166-36: task 6: Exited with exit code 1
+[2024-12-09 23:30:20,949] torch.distributed.elastic.multiprocessing.api: [ERROR] failed (exitcode: 1) local_rank: 1 (pid: 2555936) of binary: /fsx/nouamane/miniconda/envs/2-1-cu121/bin/python
+srun: error: ip-26-0-171-230: task 15: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 33, in <module>
+    sys.exit(load_entry_point('torch==2.1.1', 'console_scripts', 'torchrun')())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 264, in launch_agent
+    raise ChildFailedError(
+torch.distributed.elastic.multiprocessing.errors.ChildFailedError: 
+============================================================
+run_train.py FAILED
+------------------------------------------------------------
+Failures:
+[1]:
+  time      : 2024-12-09_23:30:20
+  host      : ip-26-0-164-207.ec2.internal
+  rank      : 11 (local_rank: 3)
+  exitcode  : 1 (pid: 2555938)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[2]:
+  time      : 2024-12-09_23:30:20
+  host      : ip-26-0-164-207.ec2.internal
+  rank      : 12 (local_rank: 4)
+  exitcode  : 1 (pid: 2555939)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[3]:
+  time      : 2024-12-09_23:30:20
+  host      : ip-26-0-164-207.ec2.internal
+  rank      : 13 (local_rank: 5)
+  exitcode  : 1 (pid: 2555940)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[4]:
+  time      : 2024-12-09_23:30:20
+  host      : ip-26-0-164-207.ec2.internal
+  rank      : 14 (local_rank: 6)
+  exitcode  : 1 (pid: 2555941)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[5]:
+  time      : 2024-12-09_23:30:20
+  host      : ip-26-0-164-207.ec2.internal
+  rank      : 15 (local_rank: 7)
+  exitcode  : 1 (pid: 2555942)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+------------------------------------------------------------
+Root Cause (first observed failure):
+[0]:
+  time      : 2024-12-09_23:30:20
+  host      : ip-26-0-164-207.ec2.internal
+  rank      : 9 (local_rank: 1)
+  exitcode  : 1 (pid: 2555936)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+============================================================
+srun: error: ip-26-0-166-214: task 8: Exited with exit code 1
+srun: error: ip-26-0-171-102: task 13: Exited with exit code 1
+srun: error: ip-26-0-166-15: task 5: Exited with exit code 1
+[2024-12-09 23:30:21,296] torch.distributed.elastic.multiprocessing.api: [ERROR] failed (exitcode: 1) local_rank: 0 (pid: 2698935) of binary: /fsx/nouamane/miniconda/envs/2-1-cu121/bin/python
+[2024-12-09 23:30:21,307] torch.distributed.elastic.multiprocessing.api: [ERROR] failed (exitcode: 1) local_rank: 0 (pid: 38261) of binary: /fsx/nouamane/miniconda/envs/2-1-cu121/bin/python
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 33, in <module>
+    sys.exit(load_entry_point('torch==2.1.1', 'console_scripts', 'torchrun')())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 264, in launch_agent
+    raise ChildFailedError(
+torch.distributed.elastic.multiprocessing.errors.ChildFailedError: 
+============================================================
+run_train.py FAILED
+------------------------------------------------------------
+Failures:
+[1]:
+  time      : 2024-12-09_23:30:20
+  host      : ip-26-0-164-236.ec2.internal
+  rank      : 17 (local_rank: 1)
+  exitcode  : 1 (pid: 2698936)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[2]:
+  time      : 2024-12-09_23:30:20
+  host      : ip-26-0-164-236.ec2.internal
+  rank      : 20 (local_rank: 4)
+  exitcode  : 1 (pid: 2698939)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[3]:
+  time      : 2024-12-09_23:30:20
+  host      : ip-26-0-164-236.ec2.internal
+  rank      : 21 (local_rank: 5)
+  exitcode  : 1 (pid: 2698940)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[4]:
+  time      : 2024-12-09_23:30:20
+  host      : ip-26-0-164-236.ec2.internal
+  rank      : 23 (local_rank: 7)
+  exitcode  : 1 (pid: 2698942)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+------------------------------------------------------------
+Root Cause (first observed failure):
+[0]:
+  time      : 2024-12-09_23:30:20
+  host      : ip-26-0-164-236.ec2.internal
+  rank      : 16 (local_rank: 0)
+  exitcode  : 1 (pid: 2698935)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+============================================================
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 33, in <module>
+    sys.exit(load_entry_point('torch==2.1.1', 'console_scripts', 'torchrun')())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 264, in launch_agent
+    raise ChildFailedError(
+torch.distributed.elastic.multiprocessing.errors.ChildFailedError: 
+============================================================
+run_train.py FAILED
+------------------------------------------------------------
+Failures:
+[1]:
+  time      : 2024-12-09_23:30:20
+  host      : ip-26-0-164-75.ec2.internal
+  rank      : 34 (local_rank: 2)
+  exitcode  : 1 (pid: 38263)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[2]:
+  time      : 2024-12-09_23:30:20
+  host      : ip-26-0-164-75.ec2.internal
+  rank      : 35 (local_rank: 3)
+  exitcode  : 1 (pid: 38264)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[3]:
+  time      : 2024-12-09_23:30:20
+  host      : ip-26-0-164-75.ec2.internal
+  rank      : 36 (local_rank: 4)
+  exitcode  : 1 (pid: 38265)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[4]:
+  time      : 2024-12-09_23:30:20
+  host      : ip-26-0-164-75.ec2.internal
+  rank      : 38 (local_rank: 6)
+  exitcode  : 1 (pid: 38267)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+------------------------------------------------------------
+Root Cause (first observed failure):
+[0]:
+  time      : 2024-12-09_23:30:20
+  host      : ip-26-0-164-75.ec2.internal
+  rank      : 32 (local_rank: 0)
+  exitcode  : 1 (pid: 38261)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+============================================================
+[2024-12-09 23:30:21,363] torch.distributed.elastic.multiprocessing.api: [ERROR] failed (exitcode: 1) local_rank: 1 (pid: 2076232) of binary: /fsx/nouamane/miniconda/envs/2-1-cu121/bin/python
+srun: error: ip-26-0-164-207: task 3: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 33, in <module>
+    sys.exit(load_entry_point('torch==2.1.1', 'console_scripts', 'torchrun')())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 264, in launch_agent
+    raise ChildFailedError(
+torch.distributed.elastic.multiprocessing.errors.ChildFailedError: 
+============================================================
+run_train.py FAILED
+------------------------------------------------------------
+Failures:
+[1]:
+  time      : 2024-12-09_23:30:20
+  host      : ip-26-0-171-168.ec2.internal
+  rank      : 107 (local_rank: 3)
+  exitcode  : 1 (pid: 2076234)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[2]:
+  time      : 2024-12-09_23:30:20
+  host      : ip-26-0-171-168.ec2.internal
+  rank      : 109 (local_rank: 5)
+  exitcode  : 1 (pid: 2076236)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[3]:
+  time      : 2024-12-09_23:30:20
+  host      : ip-26-0-171-168.ec2.internal
+  rank      : 111 (local_rank: 7)
+  exitcode  : 1 (pid: 2076238)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+------------------------------------------------------------
+Root Cause (first observed failure):
+[0]:
+  time      : 2024-12-09_23:30:20
+  host      : ip-26-0-171-168.ec2.internal
+  rank      : 105 (local_rank: 1)
+  exitcode  : 1 (pid: 2076232)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+============================================================
+[2024-12-09 23:30:21,545] torch.distributed.elastic.multiprocessing.api: [ERROR] failed (exitcode: 1) local_rank: 1 (pid: 896952) of binary: /fsx/nouamane/miniconda/envs/2-1-cu121/bin/python
+[2024-12-09 23:30:21,554] torch.distributed.elastic.multiprocessing.api: [ERROR] failed (exitcode: 1) local_rank: 3 (pid: 2773676) of binary: /fsx/nouamane/miniconda/envs/2-1-cu121/bin/python
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 33, in <module>
+    sys.exit(load_entry_point('torch==2.1.1', 'console_scripts', 'torchrun')())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 264, in launch_agent
+    raise ChildFailedError(
+torch.distributed.elastic.multiprocessing.errors.ChildFailedError: 
+============================================================
+run_train.py FAILED
+------------------------------------------------------------
+Failures:
+[1]:
+  time      : 2024-12-09_23:30:20
+  host      : ip-26-0-164-45.ec2.internal
+  rank      : 29 (local_rank: 5)
+  exitcode  : 1 (pid: 2773678)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[2]:
+  time      : 2024-12-09_23:30:20
+  host      : ip-26-0-164-45.ec2.internal
+  rank      : 31 (local_rank: 7)
+  exitcode  : 1 (pid: 2773680)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+------------------------------------------------------------
+Root Cause (first observed failure):
+[0]:
+  time      : 2024-12-09_23:30:20
+  host      : ip-26-0-164-45.ec2.internal
+  rank      : 27 (local_rank: 3)
+  exitcode  : 1 (pid: 2773676)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+============================================================
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 33, in <module>
+    sys.exit(load_entry_point('torch==2.1.1', 'console_scripts', 'torchrun')())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 264, in launch_agent
+    raise ChildFailedError(
+torch.distributed.elastic.multiprocessing.errors.ChildFailedError: 
+============================================================
+run_train.py FAILED
+------------------------------------------------------------
+Failures:
+[1]:
+  time      : 2024-12-09_23:30:20
+  host      : ip-26-0-167-9.ec2.internal
+  rank      : 90 (local_rank: 2)
+  exitcode  : 1 (pid: 896953)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[2]:
+  time      : 2024-12-09_23:30:20
+  host      : ip-26-0-167-9.ec2.internal
+  rank      : 91 (local_rank: 3)
+  exitcode  : 1 (pid: 896954)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+------------------------------------------------------------
+Root Cause (first observed failure):
+[0]:
+  time      : 2024-12-09_23:30:20
+  host      : ip-26-0-167-9.ec2.internal
+  rank      : 89 (local_rank: 1)
+  exitcode  : 1 (pid: 896952)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+============================================================
+srun: error: ip-26-0-164-75: task 1: Exited with exit code 1
+srun: error: ip-26-0-164-236: task 4: Exited with exit code 1
+srun: error: ip-26-0-171-168: task 14: Exited with exit code 1
+srun: error: ip-26-0-164-45: task 0: Exited with exit code 1
+srun: error: ip-26-0-167-9: task 10: Exited with exit code 1
diff --git a/logs/13264840-bench_1.14G_dp8_tp2_pp1_acc32_mbs2_seq8192_zero1_tpmodeRED_vocab32k.out b/logs/13264840-bench_1.14G_dp8_tp2_pp1_acc32_mbs2_seq8192_zero1_tpmodeRED_vocab32k.out
new file mode 100644
index 0000000000000000000000000000000000000000..4011616c6db01a2208fd01dd8061a6467bbf15fb
--- /dev/null
+++ b/logs/13264840-bench_1.14G_dp8_tp2_pp1_acc32_mbs2_seq8192_zero1_tpmodeRED_vocab32k.out
@@ -0,0 +1,657 @@
++ source /etc/profile.d/modules.sh
+++ . /usr/share/modules/init/bash
++++ unset _mlshdbg
++++ '[' 0 = 1 ']'
++++ unset _mlre _mlIFS
++++ '[' -n x ']'
++++ _mlIFS=' 	
+'
++++ IFS=' '
++++ '[' -n '' ']'
+++++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash autoinit
++++ _mlcode='module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+MODULEPATH_modshare=/etc/environment-modules/modules:1:/usr/share/modules/\$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1; export MODULEPATH_modshare;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+LOADEDMODULES=; export LOADEDMODULES;
+MODULEPATH=/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/\$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles; export MODULEPATH;
+test 0;'
++++ _mlret=0
++++ '[' -n x ']'
++++ IFS=' 	
+'
++++ unset _mlIFS
++++ unset _mlre _mlv _mlrv
++++ '[' 0 -eq 0 ']'
++++ eval 'module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+MODULEPATH_modshare=/etc/environment-modules/modules:1:/usr/share/modules/\$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1; export MODULEPATH_modshare;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+LOADEDMODULES=; export LOADEDMODULES;
+MODULEPATH=/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/\$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles; export MODULEPATH;
+test 0;'
+++++ MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl
+++++ export MODULES_CMD
+++++ ENV=/usr/share/modules/init/profile.sh
+++++ export ENV
+++++ MODULEPATH_modshare='/etc/environment-modules/modules:1:/usr/share/modules/$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1'
+++++ export MODULEPATH_modshare
+++++ BASH_ENV=/usr/share/modules/init/bash
+++++ export BASH_ENV
+++++ MODULESHOME=/usr/share/modules
+++++ export MODULESHOME
+++++ LOADEDMODULES=
+++++ export LOADEDMODULES
+++++ MODULEPATH='/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles'
+++++ export MODULEPATH
+++++ test 0
++++ '[' 0 = 1 ']'
++++ '[' -t 2 ']'
++++ export -f module
++++ export -f switchml
++++ '[' 5 -ge 3 ']'
++++ [[ ehxB =~ i ]]
++++ [[ ! :/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin: =~ :/usr/bin: ]]
++++ '[' '!' -n '' ']'
++++ MANPATH=:
++++ export MANPATH
+++++ manpath
++++ [[ ! :/admin/home/nouamane/.local/share/man:/fsx/nouamane/miniconda/envs/2-1-cu121/man:/fsx/nouamane/miniconda/envs/2-1-cu121/share/man:/fsx/nouamane/miniconda/man:/fsx/nouamane/miniconda/share/man:/opt/amazon/openmpi/share/man:/opt/amazon/efa/share/man:/opt/slurm/share/man:/usr/local/man:/usr/local/share/man:/usr/share/man:/admin/home/nouamane/.fzf/man:: =~ :/usr/share/man: ]]
++++ unset _mlcode _mlret
++++ '[' -n '' ']'
++ module load cuda/12.1
++ unset _mlshdbg
++ '[' 0 = 1 ']'
++ unset _mlre _mlIFS
++ '[' -n x ']'
++ _mlIFS=' 	
+'
++ IFS=' '
++ '[' -n '' ']'
+++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash load cuda/12.1
++ eval 'CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include;' export 'CPATH;
+LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:;' export 'LD_LIBRARY_PATH;
+FI_EFA_FORK_SAFE=1;' export 'FI_EFA_FORK_SAFE;
+MANPATH=/usr/local/cuda-12.1/share/man::;' export 'MANPATH;
+LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64;' export 'LIBRARY_PATH;
+_LMFILES_=/usr/share/modules/modulefiles/cuda/12.1;' export '_LMFILES_;
+LOADEDMODULES=cuda/12.1;' export 'LOADEDMODULES;
+MPI_PATH=/opt/amazon/openmpi;' export 'MPI_PATH;
+NCCL_HOME_modshare=/opt/nccl/build:1;' export 'NCCL_HOME_modshare;
+NCCL_PROTO=simple;' export 'NCCL_PROTO;
+MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1;' export 'MANPATH_modshare;
+LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1;' export 'LIBRARY_PATH_modshare;
+NCCL_SOCKET_IFNAME=enp;' export 'NCCL_SOCKET_IFNAME;
+AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl;' export 'AWS_OFI_NCCL_HOME;
+NCCL_HOME=/opt/nccl/build;' export 'NCCL_HOME;
+FI_PROVIDER=efa;' export 'FI_PROVIDER;
+AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1;' export 'AWS_OFI_NCCL_HOME_modshare;
+CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1;' export 'CPATH_modshare;
+LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2::1:/opt/nccl/build/lib:1:/opt/aws-ofi-nccl/lib:2:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/efa/lib:1;' export 'LD_LIBRARY_PATH_modshare;
+FI_EFA_ENABLE_SHM_TRANSFER=1;' export 'FI_EFA_ENABLE_SHM_TRANSFER;
+_LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1;' export '_LMFILES__modshare;
+LOADEDMODULES_modshare=cuda/12.1:1;' export 'LOADEDMODULES_modshare;
+MPI_PATH_modshare=/opt/amazon/openmpi:1;' export 'MPI_PATH_modshare;
+PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin;' export 'PATH;
+CUDA_HOME=/usr/local/cuda-12.1;' export 'CUDA_HOME;
+PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/usr/local/cuda-12.1/include:1:/opt/amazon/efa/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/usr/sbin:1:/fsx/nouamane/miniconda/bin:1:/opt/slurm/bin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:2:/opt/amazon/openmpi/bin:1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1;' export 'PATH_modshare;
+test' '0;'
+++ CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include
+++ export CPATH
+++ LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:
+++ export LD_LIBRARY_PATH
+++ FI_EFA_FORK_SAFE=1
+++ export FI_EFA_FORK_SAFE
+++ MANPATH=/usr/local/cuda-12.1/share/man::
+++ export MANPATH
+++ LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64
+++ export LIBRARY_PATH
+++ _LMFILES_=/usr/share/modules/modulefiles/cuda/12.1
+++ export _LMFILES_
+++ LOADEDMODULES=cuda/12.1
+++ export LOADEDMODULES
+++ MPI_PATH=/opt/amazon/openmpi
+++ export MPI_PATH
+++ NCCL_HOME_modshare=/opt/nccl/build:1
+++ export NCCL_HOME_modshare
+++ NCCL_PROTO=simple
+++ export NCCL_PROTO
+++ MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1
+++ export MANPATH_modshare
+++ LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1
+++ export LIBRARY_PATH_modshare
+++ NCCL_SOCKET_IFNAME=enp
+++ export NCCL_SOCKET_IFNAME
+++ AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl
+++ export AWS_OFI_NCCL_HOME
+++ NCCL_HOME=/opt/nccl/build
+++ export NCCL_HOME
+++ FI_PROVIDER=efa
+++ export FI_PROVIDER
+++ AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1
+++ export AWS_OFI_NCCL_HOME_modshare
+++ CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1
+++ export CPATH_modshare
+++ LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2::1:/opt/nccl/build/lib:1:/opt/aws-ofi-nccl/lib:2:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/efa/lib:1
+++ export LD_LIBRARY_PATH_modshare
+++ FI_EFA_ENABLE_SHM_TRANSFER=1
+++ export FI_EFA_ENABLE_SHM_TRANSFER
+++ _LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1
+++ export _LMFILES__modshare
+++ LOADEDMODULES_modshare=cuda/12.1:1
+++ export LOADEDMODULES_modshare
+++ MPI_PATH_modshare=/opt/amazon/openmpi:1
+++ export MPI_PATH_modshare
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ export PATH
+++ CUDA_HOME=/usr/local/cuda-12.1
+++ export CUDA_HOME
+++ PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/usr/local/cuda-12.1/include:1:/opt/amazon/efa/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/usr/sbin:1:/fsx/nouamane/miniconda/bin:1:/opt/slurm/bin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:2:/opt/amazon/openmpi/bin:1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1
+++ export PATH_modshare
+++ test 0
++ _mlstatus=0
++ '[' -n x ']'
++ IFS=' 	
+'
++ unset _mlre _mlv _mlrv _mlIFS
++ '[' -n '' ']'
++ unset _mlshdbg
++ return 0
++ source /fsx/nouamane/miniconda/bin/activate
+++ _CONDA_ROOT=/fsx/nouamane/miniconda
+++ . /fsx/nouamane/miniconda/etc/profile.d/conda.sh
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ '[' -z x ']'
+++ conda activate
+++ local cmd=activate
+++ case "$cmd" in
+++ __conda_activate activate
+++ '[' -n '' ']'
+++ local ask_conda
++++ PS1=
++++ __conda_exe shell.posix activate
++++ /fsx/nouamane/miniconda/bin/conda shell.posix activate
+++ ask_conda='. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_2='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\'''
+++ eval '. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_2='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\'''
++++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh
+++++ test -n ''
+++++ unset XML_CATALOG_FILES
+++++ unset xml_catalog_files_libxml2
++++ PS1='(base) '
++++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++++ export CONDA_PREFIX=/fsx/nouamane/miniconda
++++ CONDA_PREFIX=/fsx/nouamane/miniconda
++++ export CONDA_SHLVL=3
++++ CONDA_SHLVL=3
++++ export CONDA_DEFAULT_ENV=base
++++ CONDA_DEFAULT_ENV=base
++++ export 'CONDA_PROMPT_MODIFIER=(base) '
++++ CONDA_PROMPT_MODIFIER='(base) '
++++ export CONDA_PREFIX_2=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ CONDA_PREFIX_2=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ __conda_hashr
+++ '[' -n '' ']'
+++ '[' -n '' ']'
+++ hash -r
++ conda activate 2-1-cu121
++ local cmd=activate
++ case "$cmd" in
++ __conda_activate activate 2-1-cu121
++ '[' -n '' ']'
++ local ask_conda
+++ PS1='(base) '
+++ __conda_exe shell.posix activate 2-1-cu121
+++ /fsx/nouamane/miniconda/bin/conda shell.posix activate 2-1-cu121
++ ask_conda='PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''4'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_3='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
++ eval 'PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''4'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_3='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
+++ PS1='(2-1-cu121) '
+++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ export CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ export CONDA_SHLVL=4
+++ CONDA_SHLVL=4
+++ export CONDA_DEFAULT_ENV=2-1-cu121
+++ CONDA_DEFAULT_ENV=2-1-cu121
+++ export 'CONDA_PROMPT_MODIFIER=(2-1-cu121) '
+++ CONDA_PROMPT_MODIFIER='(2-1-cu121) '
+++ export CONDA_PREFIX_3=/fsx/nouamane/miniconda
+++ CONDA_PREFIX_3=/fsx/nouamane/miniconda
+++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ export _CE_M=
+++ _CE_M=
+++ export _CE_CONDA=
+++ _CE_CONDA=
+++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh
++++ test -n ''
++++ xml_catalog_files_libxml2=
++++ XML_CATALOG_FILES=
++++ conda_catalog_files=
++++ ifs_libxml2=' 	
+'
++++ IFS=' '
++++ rem=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ for pre in ${rem}
++++ test '' = /fsx/nouamane/miniconda/envs/2-1-cu121
++++ conda_catalog_files=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ rem=
++++ IFS=' 	
+'
++++ conda_catalog_files='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ export 'XML_CATALOG_FILES=file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ XML_CATALOG_FILES='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ unset conda_catalog_files ifs_libxml2 rem
++ __conda_hashr
++ '[' -n '' ']'
++ '[' -n '' ']'
++ hash -r
++ export PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++ PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ scontrol show hostnames 'ip-26-0-175-[170,241]'
++ export 'NODELIST=ip-26-0-175-170
+ip-26-0-175-241'
++ NODELIST='ip-26-0-175-170
+ip-26-0-175-241'
+++ head -n1
+++ scontrol show hostnames 'ip-26-0-175-[170,241]'
++ export MASTER_NODE=ip-26-0-175-170
++ MASTER_NODE=ip-26-0-175-170
++ export MASTER_PORT=12356
++ MASTER_PORT=12356
++ export NNODES=2
++ NNODES=2
++ export GPUS_PER_NODE=8
++ GPUS_PER_NODE=8
++ export WORLD_SIZE=16
++ WORLD_SIZE=16
++ export CUDA_DEVICE_MAX_CONNECTIONS=1
++ CUDA_DEVICE_MAX_CONNECTIONS=1
++ export NANOTRON_BENCHMARK=1
++ NANOTRON_BENCHMARK=1
++ echo 'Master node: ip-26-0-175-170'
+Master node: ip-26-0-175-170
++ echo 'All nodes: ip-26-0-175-170
+ip-26-0-175-241'
+All nodes: ip-26-0-175-170
+ip-26-0-175-241
++ echo 'World size: 16'
+World size: 16
++ srun torchrun --nnodes=2 --nproc_per_node=8 --rdzv_id=13264840 --rdzv_backend=c10d --rdzv_endpoint=ip-26-0-175-170:12356 run_train.py --config-file benchmark/configs/config_1.14G_dp8_tp2_pp1_acc32_mbs2_seq8192_zero1_tpmodeRED_vocab32k.yaml
+[2024-12-09 23:25:32,194] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-09 23:25:32,199] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-09 23:25:32,194] torch.distributed.run: [WARNING] 
+[2024-12-09 23:25:32,194] torch.distributed.run: [WARNING] *****************************************
+[2024-12-09 23:25:32,194] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-09 23:25:32,194] torch.distributed.run: [WARNING] *****************************************
+[2024-12-09 23:25:32,199] torch.distributed.run: [WARNING] 
+[2024-12-09 23:25:32,199] torch.distributed.run: [WARNING] *****************************************
+[2024-12-09 23:25:32,199] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-09 23:25:32,199] torch.distributed.run: [WARNING] *****************************************
+12/09/2024 23:26:10 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]: Config:
+12/09/2024 23:26:10 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]: Config(general=GeneralArgs(project='debug',
+12/09/2024 23:26:10 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                            run='1.14G_dp8_tp2_pp1_acc32_mbs2_seq8192_zero1_tpmodeRED_vocab32k',
+12/09/2024 23:26:10 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                            seed=42,
+12/09/2024 23:26:10 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                            step=None,
+12/09/2024 23:26:10 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                            consumed_train_samples=None,
+12/09/2024 23:26:10 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                            benchmark_csv_path=PosixPath('benchmark/results/bench_final.csv'),
+12/09/2024 23:26:10 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                            ignore_sanity_checks=True),
+12/09/2024 23:26:10 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:        parallelism=ParallelismArgs(dp=8,
+12/09/2024 23:26:10 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                                    pp=1,
+12/09/2024 23:26:10 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                                    tp=2,
+12/09/2024 23:26:10 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                                    pp_engine=<nanotron.parallel.pipeline_parallel.engine.OneForwardOneBackwardPipelineEngine object at 0x7fae9f420e80>,
+12/09/2024 23:26:10 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                                    tp_mode=<TensorParallelLinearMode.REDUCE_SCATTER: 2>,
+12/09/2024 23:26:10 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                                    tp_linear_async_communication=True,
+12/09/2024 23:26:10 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                                    recompute_layer=False,
+12/09/2024 23:26:10 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                                    tp_recompute_allgather=True,
+12/09/2024 23:26:10 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                                    expert_parallel_size=1),
+12/09/2024 23:26:10 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:        model=ModelArgs(model_config=LlamaConfig(bos_token_id=0,
+12/09/2024 23:26:10 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                                                 eos_token_id=0,
+12/09/2024 23:26:10 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                                                 hidden_act='silu',
+12/09/2024 23:26:10 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                                                 hidden_size=2048,
+12/09/2024 23:26:10 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                                                 initializer_range=0.02,
+12/09/2024 23:26:10 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                                                 intermediate_size=8192,
+12/09/2024 23:26:10 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                                                 is_llama_config=True,
+12/09/2024 23:26:10 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                                                 max_position_embeddings=8192,
+12/09/2024 23:26:10 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                                                 num_attention_heads=32,
+12/09/2024 23:26:10 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                                                 num_hidden_layers=16,
+12/09/2024 23:26:10 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                                                 num_key_value_heads=8,
+12/09/2024 23:26:10 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                                                 pad_token_id=None,
+12/09/2024 23:26:10 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                                                 pretraining_tp=1,
+12/09/2024 23:26:10 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                                                 rms_norm_eps=1e-05,
+12/09/2024 23:26:10 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                                                 rope_scaling=None,
+12/09/2024 23:26:10 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                                                 rope_theta=10000.0,
+12/09/2024 23:26:10 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                                                 rope_interleaved=False,
+12/09/2024 23:26:10 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                                                 tie_word_embeddings=True,
+12/09/2024 23:26:10 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                                                 use_cache=True,
+12/09/2024 23:26:10 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                                                 vocab_size=32768),
+12/09/2024 23:26:10 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                        init_method=RandomInit(std=0.02),
+12/09/2024 23:26:10 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                        dtype=torch.bfloat16,
+12/09/2024 23:26:10 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                        make_vocab_size_divisible_by=1,
+12/09/2024 23:26:10 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                        ddp_bucket_cap_mb=25),
+12/09/2024 23:26:10 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:        tokenizer=TokenizerArgs(tokenizer_name_or_path='robot-test/dummy-tokenizer-wordlevel',
+12/09/2024 23:26:10 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                                tokenizer_revision=None,
+12/09/2024 23:26:10 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                                tokenizer_max_length=None),
+12/09/2024 23:26:10 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:        checkpoints=CheckpointsArgs(checkpoints_path=PosixPath('checkpoints'),
+12/09/2024 23:26:10 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                                    checkpoint_interval=10000,
+12/09/2024 23:26:10 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                                    save_initial_state=False,
+12/09/2024 23:26:10 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                                    save_final_state=False,
+12/09/2024 23:26:10 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                                    resume_checkpoint_path=None,
+12/09/2024 23:26:10 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                                    checkpoints_path_is_shared_file_system=False),
+12/09/2024 23:26:10 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:        logging=LoggingArgs(log_level='info',
+12/09/2024 23:26:10 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                            log_level_replica='info',
+12/09/2024 23:26:10 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                            iteration_step_info_interval=1),
+12/09/2024 23:26:10 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:        tokens=TokensArgs(sequence_length=8192,
+12/09/2024 23:26:10 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                          train_steps=100,
+12/09/2024 23:26:10 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                          micro_batch_size=2,
+12/09/2024 23:26:10 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                          batch_accumulation_per_replica=32,
+12/09/2024 23:26:10 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                          val_check_interval=100,
+12/09/2024 23:26:10 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                          limit_val_batches=0,
+12/09/2024 23:26:10 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                          limit_test_batches=0),
+12/09/2024 23:26:10 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:        optimizer=OptimizerArgs(optimizer_factory=AdamWOptimizerArgs(adam_eps=1e-08,
+12/09/2024 23:26:10 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                                                                     adam_beta1=0.9,
+12/09/2024 23:26:10 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                                                                     adam_beta2=0.95,
+12/09/2024 23:26:10 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                                                                     torch_adam_is_fused=True,
+12/09/2024 23:26:10 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                                                                     name='adamW'),
+12/09/2024 23:26:10 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                                zero_stage=1,
+12/09/2024 23:26:10 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                                weight_decay=0.01,
+12/09/2024 23:26:10 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                                clip_grad=1.0,
+12/09/2024 23:26:10 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                                accumulate_grad_in_fp32=True,
+12/09/2024 23:26:10 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                                learning_rate_scheduler=LRSchedulerArgs(learning_rate=0.0003,
+12/09/2024 23:26:10 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                                                                        lr_warmup_steps=2,
+12/09/2024 23:26:10 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                                                                        lr_warmup_style='linear',
+12/09/2024 23:26:10 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                                                                        lr_decay_style='cosine',
+12/09/2024 23:26:10 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                                                                        lr_decay_steps=13,
+12/09/2024 23:26:10 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                                                                        lr_decay_starting_step=None,
+12/09/2024 23:26:10 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                                                                        min_decay_lr=1e-05)),
+12/09/2024 23:26:10 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:        data_stages=[DatasetStageArgs(name='Stable Training Stage',
+12/09/2024 23:26:10 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                                      start_training_step=1,
+12/09/2024 23:26:10 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                                      data=DataArgs(dataset=None,
+12/09/2024 23:26:10 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                                                    seed=42,
+12/09/2024 23:26:10 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                                                    num_loading_workers=1))],
+12/09/2024 23:26:10 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:        profiler=None,
+12/09/2024 23:26:10 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:        lighteval=None,
+12/09/2024 23:26:10 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:        s3_upload=None)
+12/09/2024 23:26:10 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]: Model Config:
+12/09/2024 23:26:10 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]: LlamaConfig(bos_token_id=0,
+12/09/2024 23:26:10 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:             eos_token_id=0,
+12/09/2024 23:26:10 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:             hidden_act='silu',
+12/09/2024 23:26:10 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:             hidden_size=2048,
+12/09/2024 23:26:10 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:             initializer_range=0.02,
+12/09/2024 23:26:10 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:             intermediate_size=8192,
+12/09/2024 23:26:10 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:             is_llama_config=True,
+12/09/2024 23:26:10 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:             max_position_embeddings=8192,
+12/09/2024 23:26:10 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:             num_attention_heads=32,
+12/09/2024 23:26:10 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:             num_hidden_layers=16,
+12/09/2024 23:26:10 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:             num_key_value_heads=8,
+12/09/2024 23:26:10 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:             pad_token_id=None,
+12/09/2024 23:26:10 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:             pretraining_tp=1,
+12/09/2024 23:26:10 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:             rms_norm_eps=1e-05,
+12/09/2024 23:26:10 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:             rope_scaling=None,
+12/09/2024 23:26:10 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:             rope_theta=10000.0,
+12/09/2024 23:26:10 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:             rope_interleaved=False,
+12/09/2024 23:26:10 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:             tie_word_embeddings=True,
+12/09/2024 23:26:10 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:             use_cache=True,
+12/09/2024 23:26:10 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:             vocab_size=32768)
+12/09/2024 23:26:10 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]: Building model..
+12/09/2024 23:26:10 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]: Initialize RoPE Theta = 10000.0
+12/09/2024 23:26:10 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]: Setting PP block ranks...
+12/09/2024 23:26:13 [INFO|DP=0|PP=0|TP=1|ip-26-0-175-170]: Local number of parameters: 520M (992.13MiB)
+12/09/2024 23:26:13 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]: Total number of parameters: 1.04G (1984.26MiB)
+12/09/2024 23:26:13 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]: Local number of parameters: 520M (992.13MiB)
+12/09/2024 23:26:13 [INFO|DP=0|PP=0|TP=1|ip-26-0-175-170]: [After model building] Memory usage: 992.15MiB. Peak allocated: 5504.00MiB Peak reserved: 25986.00MiB
+12/09/2024 23:26:13 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]: [After model building] Memory usage: 992.15MiB. Peak allocated: 5504.00MiB Peak reserved: 25986.00MiB
+12/09/2024 23:26:13 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]: No checkpoint path provided.
+12/09/2024 23:26:13 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]: Parametrizing model parameters using StandardParametrizator
+12/09/2024 23:26:20 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]: [Optimizer Building] Using LearningRateForSP as learning rate
+12/09/2024 23:26:20 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]: [ZeRO sharding] Size of optimizer params per rank:
+12/09/2024 23:26:20 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]: [ZeRO sharding] DP Rank 0 has 65M out of 520M (12.50%) params' optimizer states
+12/09/2024 23:26:20 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]: [ZeRO sharding] DP Rank 1 has 65M out of 520M (12.50%) params' optimizer states
+12/09/2024 23:26:20 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]: [ZeRO sharding] DP Rank 2 has 65M out of 520M (12.50%) params' optimizer states
+12/09/2024 23:26:20 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]: [ZeRO sharding] DP Rank 3 has 65M out of 520M (12.50%) params' optimizer states
+12/09/2024 23:26:20 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]: [ZeRO sharding] DP Rank 4 has 65M out of 520M (12.50%) params' optimizer states
+12/09/2024 23:26:20 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]: [ZeRO sharding] DP Rank 5 has 65M out of 520M (12.50%) params' optimizer states
+12/09/2024 23:26:20 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]: [ZeRO sharding] DP Rank 6 has 65M out of 520M (12.50%) params' optimizer states
+12/09/2024 23:26:20 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]: [ZeRO sharding] DP Rank 7 has 65M out of 520M (12.50%) params' optimizer states
+12/09/2024 23:26:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]: [Training Plan] Stage Stable Training Stage has 99 remaining training steps and has consumed 0 samples
+12/09/2024 23:26:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]: Using dummy data generator
+12/09/2024 23:26:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]: [Training Plan] There are 1 training stages 
+12/09/2024 23:26:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]: [Stage Stable Training Stage] start from step 1 
+12/09/2024 23:26:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]: 
+12/09/2024 23:26:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]: [Start training] datetime: 2024-12-09 23:26:21.285455 | mbs: 2 | grad_accum: 32 | global_batch_size: 512 | sequence_length: 8192 | train_steps: 100 | start_iteration_step: 0 | consumed_train_samples: 0
+wandb: WARNING Path /fsx/nouamane/.cache/wandb/wandb/ wasn't writable, using system temp directory.
+wandb: WARNING Path /fsx/nouamane/.cache/wandb/wandb/ wasn't writable, using system temp directory
+wandb: Tracking run with wandb version 0.16.0
+wandb: W&B syncing is set to `offline` in this directory.  
+wandb: Run `wandb online` or set WANDB_MODE=online to enable cloud syncing.
+12/09/2024 23:26:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]: Resuming training from stage Stable Training Stage, it has trained for 0 samples and has 99 remaining train steps
+12/09/2024 23:26:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:  Memory usage: 3224.44MiB. Peak allocated 5504.00MiB. Peak reserved: 27972.00MiB
+12/09/2024 23:26:37 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:  Memory usage: 3432.74MiB. Peak allocated 17082.64MiB. Peak reserved: 22012.00MiB
+12/09/2024 23:26:38 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]: iteration: 1 / 100 | consumed_tokens: 4.19M | elapsed_time_per_iteration_ms: 8.37K | tokens_per_sec: 501K | tokens_per_sec_per_gpu: 31.3K | global_batch_size: 512 | lm_loss: 10.8 | lr: 0.00015 | model_tflops_per_gpu: 296 | hardware_tflops_per_gpu: 296 | grad_norm: 0.415 | cuda_memory_allocated: 4.13G | cuda_max_memory_reserved: 23.1G | hd_total_memory_tb: 312G | hd_used_memory_tb: 69.8G | hd_free_memory_tb: 242G
+12/09/2024 23:26:38 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:  Memory usage: 3940.60MiB. Peak allocated 6048.87MiB. Peak reserved: 22026.00MiB
+12/09/2024 23:26:44 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:  Memory usage: 3940.61MiB. Peak allocated 17590.51MiB. Peak reserved: 23068.00MiB
+12/09/2024 23:26:44 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]: iteration: 2 / 100 | consumed_tokens: 8.39M | elapsed_time_per_iteration_ms: 6.43K | tokens_per_sec: 652K | tokens_per_sec_per_gpu: 40.8K | global_batch_size: 512 | lm_loss: 10.8 | lr: 0.0003 | model_tflops_per_gpu: 386 | hardware_tflops_per_gpu: 386 | grad_norm: 0.415 | cuda_memory_allocated: 4.13G | cuda_max_memory_reserved: 24.2G | hd_total_memory_tb: 312G | hd_used_memory_tb: 69.8G | hd_free_memory_tb: 242G
+12/09/2024 23:26:44 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:  Memory usage: 3940.60MiB. Peak allocated 6048.89MiB. Peak reserved: 23068.00MiB
+12/09/2024 23:26:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:  Memory usage: 3940.61MiB. Peak allocated 17590.51MiB. Peak reserved: 23068.00MiB
+num_params
+{'total': 1040322560, 'local': 520161280}
+num_paramsnum_params
+
+{'total': 1040322560, 'local': 520161280}
+{'total': 1040322560, 'local': 520161280}
+num_paramsnum_paramsnum_paramsnum_params
+
+
+
+{'total': 1040322560, 'local': 520161280}
+{'total': 1040322560, 'local': 520161280}{'total': 1040322560, 'local': 520161280}
+{'total': 1040322560, 'local': 520161280}
+
+num_paramsnum_params
+
+{'total': 1040322560, 'local': 520161280}num_params{'total': 1040322560, 'local': 520161280}
+
+
+{'total': 1040322560, 'local': 520161280}
+num_params
+{'total': 1040322560, 'local': 520161280}
+num_params
+{'total': 1040322560, 'local': 520161280}
+num_params
+{'total': 1040322560, 'local': 520161280}
+num_params
+num_params{'total': 1040322560, 'local': 520161280}
+
+{'total': 1040322560, 'local': 520161280}
+12/09/2024 23:26:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]: iteration: 3 / 100 | consumed_tokens: 12.6M | elapsed_time_per_iteration_ms: 6.45K | tokens_per_sec: 651K | tokens_per_sec_per_gpu: 40.7K | global_batch_size: 512 | lm_loss: 10.8 | lr: 0.000296 | model_tflops_per_gpu: 385 | hardware_tflops_per_gpu: 385 | grad_norm: 0.405 | cuda_memory_allocated: 4.13G | cuda_max_memory_reserved: 24.2G | hd_total_memory_tb: 312G | hd_used_memory_tb: 69.8G | hd_free_memory_tb: 242G
+num_params
+{'total': 1040322560, 'local': 520161280}
+12/09/2024 23:26:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]: | job_id   | name                                                          | nodes | seq_len | mbs | batch_accum | gbs | mTFLOPs | hTFLOPs | tok/s/gpu | AllReduce (GB/s) | AllGather (GB/s) | ReduceScatter (GB/s) | AR Intra-node (GB/s) | AG Intra-node (GB/s) | RS Intra-node (GB/s) | Mem Alloc (GB) | Mem Res (GB) | dp | pp | tp | pp_engine                                                                                                 | tp_mode                                 | tp_async_comm | hidden_size | hidden_act | num_layers | num_heads | num_kv_heads | max_pos | vocab_size | tie_word_embeddings | dtype          | zero_stage | ddp_bucket_cap_mb | accumulate_grad_in_fp32 | Total Params | Local Params |
+12/09/2024 23:26:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]: | -------- | ------------------------------------------------------------- | ----- | ------- | --- | ----------- | --- | ------- | ------- | --------- | ---------------- | ---------------- | -------------------- | -------------------- | -------------------- | -------------------- | -------------- | ------------ | -- | -- | -- | --------------------------------------------------------------------------------------------------------- | --------------------------------------- | ------------- | ----------- | ---------- | ---------- | --------- | ------------ | ------- | ---------- | ------------------- | -------------- | ---------- | ----------------- | ----------------------- | ------------ | ------------ |
+12/09/2024 23:26:50 [INFO|DP=4|PP=0|TP=0|ip-26-0-175-241]: Throughput logging complete
+12/09/2024 23:26:50 [INFO|DP=6|PP=0|TP=0|ip-26-0-175-241]: Throughput logging complete
+12/09/2024 23:26:50 [INFO|DP=6|PP=0|TP=1|ip-26-0-175-241]: Throughput logging complete
+12/09/2024 23:26:50 [INFO|DP=4|PP=0|TP=1|ip-26-0-175-241]: Throughput logging complete
+12/09/2024 23:26:50 [INFO|DP=7|PP=0|TP=0|ip-26-0-175-241]: Throughput logging complete
+12/09/2024 23:26:50 [INFO|DP=5|PP=0|TP=1|ip-26-0-175-241]: Throughput logging complete
+12/09/2024 23:26:50 [INFO|DP=7|PP=0|TP=1|ip-26-0-175-241]: Throughput logging complete
+12/09/2024 23:26:50 [INFO|DP=5|PP=0|TP=0|ip-26-0-175-241]: Throughput logging complete
+12/09/2024 23:26:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]: | 13264840 | 1.14G_dp8_tp2_pp1_acc32_mbs2_seq8192_zero1_tpmodeRED_vocab32k | 2     | 8192    | 2   | 32          | 512 | 384.71  | 384.71  | 40657.09  | 441.31           | 209.23           | 207.20               | 460.12               | 263.95               | 265.07               | 5.91           | 22.53        | 8  | 1  | 2  | <nanotron.parallel.pipeline_parallel.engine.OneForwardOneBackwardPipelineEngine object at 0x7fae9f420e80> | TensorParallelLinearMode.REDUCE_SCATTER | True          | 2048        | silu       | 16         | 32        | 8            | 8192    | 32768      | True                | torch.bfloat16 | 1          | 25                | True                    | 1.04G        | 520M         |
+12/09/2024 23:26:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]: Throughput logging complete
+12/09/2024 23:26:50 [INFO|DP=0|PP=0|TP=1|ip-26-0-175-170]: Throughput logging complete
+12/09/2024 23:26:50 [INFO|DP=1|PP=0|TP=0|ip-26-0-175-170]: Throughput logging complete
+12/09/2024 23:26:50 [INFO|DP=1|PP=0|TP=1|ip-26-0-175-170]: Throughput logging complete
+12/09/2024 23:26:50 [INFO|DP=2|PP=0|TP=0|ip-26-0-175-170]: Throughput logging complete
+12/09/2024 23:26:50 [INFO|DP=2|PP=0|TP=1|ip-26-0-175-170]: Throughput logging complete
+12/09/2024 23:26:50 [INFO|DP=3|PP=0|TP=0|ip-26-0-175-170]: Throughput logging complete
+12/09/2024 23:26:50 [INFO|DP=3|PP=0|TP=1|ip-26-0-175-170]: Throughput logging complete
+srun: Job step aborted: Waiting up to 32 seconds for job step to finish.
+slurmstepd: error: *** JOB 13264840 ON ip-26-0-175-170 CANCELLED AT 2024-12-09T23:26:51 ***
+[2024-12-09 23:26:51,025] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-09 23:26:51,025] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3206353 closing signal SIGTERM
+[2024-12-09 23:26:51,025] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3206354 closing signal SIGTERM
+[2024-12-09 23:26:51,025] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3206355 closing signal SIGTERM
+[2024-12-09 23:26:51,025] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3206356 closing signal SIGTERM
+[2024-12-09 23:26:51,028] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3206357 closing signal SIGTERM
+[2024-12-09 23:26:51,028] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3206358 closing signal SIGTERM
+[2024-12-09 23:26:51,028] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3206359 closing signal SIGTERM
+[2024-12-09 23:26:51,030] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3206360 closing signal SIGTERM
+12/09/2024 23:26:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:  Memory usage: 3940.60MiB. Peak allocated 6048.89MiB. Peak reserved: 23068.00MiB
+slurmstepd: error: *** STEP 13264840.0 ON ip-26-0-175-170 CANCELLED AT 2024-12-09T23:26:51 ***
+[2024-12-09 23:26:51,024] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-09 23:26:51,024] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1412749 closing signal SIGTERM
+[2024-12-09 23:26:51,024] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1412750 closing signal SIGTERM
+[2024-12-09 23:26:51,024] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1412751 closing signal SIGTERM
+[2024-12-09 23:26:51,024] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1412752 closing signal SIGTERM
+[2024-12-09 23:26:51,026] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1412753 closing signal SIGTERM
+[2024-12-09 23:26:51,026] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1412754 closing signal SIGTERM
+[2024-12-09 23:26:51,026] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1412755 closing signal SIGTERM
+[2024-12-09 23:26:51,026] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1412756 closing signal SIGTERM
diff --git a/logs/13265127-bench_1.34G_dp64_tp2_pp1_acc8_mbs1_seq2048_zero1_tpmodeALL_vocab131k.out b/logs/13265127-bench_1.34G_dp64_tp2_pp1_acc8_mbs1_seq2048_zero1_tpmodeALL_vocab131k.out
new file mode 100644
index 0000000000000000000000000000000000000000..3e2dcba5a873b84b5477ff7f3ee2accbaa270cf4
--- /dev/null
+++ b/logs/13265127-bench_1.34G_dp64_tp2_pp1_acc8_mbs1_seq2048_zero1_tpmodeALL_vocab131k.out
@@ -0,0 +1,1300 @@
++ source /etc/profile.d/modules.sh
+++ . /usr/share/modules/init/bash
++++ unset _mlshdbg
++++ '[' 0 = 1 ']'
++++ unset _mlre _mlIFS
++++ '[' -n x ']'
++++ _mlIFS=' 	
+'
++++ IFS=' '
++++ '[' -n '' ']'
+++++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash autoinit
++++ _mlcode='module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+MODULEPATH_modshare=/etc/environment-modules/modules:1:/usr/share/modules/\$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1; export MODULEPATH_modshare;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+LOADEDMODULES=; export LOADEDMODULES;
+MODULEPATH=/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/\$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles; export MODULEPATH;
+test 0;'
++++ _mlret=0
++++ '[' -n x ']'
++++ IFS=' 	
+'
++++ unset _mlIFS
++++ unset _mlre _mlv _mlrv
++++ '[' 0 -eq 0 ']'
++++ eval 'module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+MODULEPATH_modshare=/etc/environment-modules/modules:1:/usr/share/modules/\$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1; export MODULEPATH_modshare;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+LOADEDMODULES=; export LOADEDMODULES;
+MODULEPATH=/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/\$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles; export MODULEPATH;
+test 0;'
+++++ MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl
+++++ export MODULES_CMD
+++++ ENV=/usr/share/modules/init/profile.sh
+++++ export ENV
+++++ MODULEPATH_modshare='/etc/environment-modules/modules:1:/usr/share/modules/$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1'
+++++ export MODULEPATH_modshare
+++++ BASH_ENV=/usr/share/modules/init/bash
+++++ export BASH_ENV
+++++ MODULESHOME=/usr/share/modules
+++++ export MODULESHOME
+++++ LOADEDMODULES=
+++++ export LOADEDMODULES
+++++ MODULEPATH='/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles'
+++++ export MODULEPATH
+++++ test 0
++++ '[' 0 = 1 ']'
++++ '[' -t 2 ']'
++++ export -f module
++++ export -f switchml
++++ '[' 5 -ge 3 ']'
++++ [[ ehxB =~ i ]]
++++ [[ ! :/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin: =~ :/usr/bin: ]]
++++ '[' '!' -n '' ']'
++++ MANPATH=:
++++ export MANPATH
+++++ manpath
++++ [[ ! :/admin/home/nouamane/.local/share/man:/fsx/nouamane/miniconda/envs/2-1-cu121/man:/fsx/nouamane/miniconda/envs/2-1-cu121/share/man:/fsx/nouamane/miniconda/man:/fsx/nouamane/miniconda/share/man:/opt/amazon/openmpi/share/man:/opt/amazon/efa/share/man:/opt/slurm/share/man:/usr/local/man:/usr/local/share/man:/usr/share/man:/admin/home/nouamane/.fzf/man:: =~ :/usr/share/man: ]]
++++ unset _mlcode _mlret
++++ '[' -n '' ']'
++ module load cuda/12.1
++ unset _mlshdbg
++ '[' 0 = 1 ']'
++ unset _mlre _mlIFS
++ '[' -n x ']'
++ _mlIFS=' 	
+'
++ IFS=' '
++ '[' -n '' ']'
+++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash load cuda/12.1
++ eval 'CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include;' export 'CPATH;
+LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:;' export 'LD_LIBRARY_PATH;
+FI_EFA_FORK_SAFE=1;' export 'FI_EFA_FORK_SAFE;
+MANPATH=/usr/local/cuda-12.1/share/man::;' export 'MANPATH;
+LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64;' export 'LIBRARY_PATH;
+_LMFILES_=/usr/share/modules/modulefiles/cuda/12.1;' export '_LMFILES_;
+LOADEDMODULES=cuda/12.1;' export 'LOADEDMODULES;
+MPI_PATH=/opt/amazon/openmpi;' export 'MPI_PATH;
+NCCL_HOME_modshare=/opt/nccl/build:1;' export 'NCCL_HOME_modshare;
+NCCL_PROTO=simple;' export 'NCCL_PROTO;
+MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1;' export 'MANPATH_modshare;
+LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1;' export 'LIBRARY_PATH_modshare;
+NCCL_SOCKET_IFNAME=enp;' export 'NCCL_SOCKET_IFNAME;
+AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl;' export 'AWS_OFI_NCCL_HOME;
+NCCL_HOME=/opt/nccl/build;' export 'NCCL_HOME;
+FI_PROVIDER=efa;' export 'FI_PROVIDER;
+AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1;' export 'AWS_OFI_NCCL_HOME_modshare;
+CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1;' export 'CPATH_modshare;
+LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2::1:/opt/nccl/build/lib:1:/opt/aws-ofi-nccl/lib:2:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/efa/lib:1;' export 'LD_LIBRARY_PATH_modshare;
+FI_EFA_ENABLE_SHM_TRANSFER=1;' export 'FI_EFA_ENABLE_SHM_TRANSFER;
+_LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1;' export '_LMFILES__modshare;
+LOADEDMODULES_modshare=cuda/12.1:1;' export 'LOADEDMODULES_modshare;
+MPI_PATH_modshare=/opt/amazon/openmpi:1;' export 'MPI_PATH_modshare;
+PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin;' export 'PATH;
+CUDA_HOME=/usr/local/cuda-12.1;' export 'CUDA_HOME;
+PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/usr/local/cuda-12.1/include:1:/opt/amazon/efa/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/usr/sbin:1:/fsx/nouamane/miniconda/bin:1:/opt/slurm/bin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:2:/opt/amazon/openmpi/bin:1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1;' export 'PATH_modshare;
+test' '0;'
+++ CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include
+++ export CPATH
+++ LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:
+++ export LD_LIBRARY_PATH
+++ FI_EFA_FORK_SAFE=1
+++ export FI_EFA_FORK_SAFE
+++ MANPATH=/usr/local/cuda-12.1/share/man::
+++ export MANPATH
+++ LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64
+++ export LIBRARY_PATH
+++ _LMFILES_=/usr/share/modules/modulefiles/cuda/12.1
+++ export _LMFILES_
+++ LOADEDMODULES=cuda/12.1
+++ export LOADEDMODULES
+++ MPI_PATH=/opt/amazon/openmpi
+++ export MPI_PATH
+++ NCCL_HOME_modshare=/opt/nccl/build:1
+++ export NCCL_HOME_modshare
+++ NCCL_PROTO=simple
+++ export NCCL_PROTO
+++ MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1
+++ export MANPATH_modshare
+++ LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1
+++ export LIBRARY_PATH_modshare
+++ NCCL_SOCKET_IFNAME=enp
+++ export NCCL_SOCKET_IFNAME
+++ AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl
+++ export AWS_OFI_NCCL_HOME
+++ NCCL_HOME=/opt/nccl/build
+++ export NCCL_HOME
+++ FI_PROVIDER=efa
+++ export FI_PROVIDER
+++ AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1
+++ export AWS_OFI_NCCL_HOME_modshare
+++ CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1
+++ export CPATH_modshare
+++ LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2::1:/opt/nccl/build/lib:1:/opt/aws-ofi-nccl/lib:2:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/efa/lib:1
+++ export LD_LIBRARY_PATH_modshare
+++ FI_EFA_ENABLE_SHM_TRANSFER=1
+++ export FI_EFA_ENABLE_SHM_TRANSFER
+++ _LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1
+++ export _LMFILES__modshare
+++ LOADEDMODULES_modshare=cuda/12.1:1
+++ export LOADEDMODULES_modshare
+++ MPI_PATH_modshare=/opt/amazon/openmpi:1
+++ export MPI_PATH_modshare
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ export PATH
+++ CUDA_HOME=/usr/local/cuda-12.1
+++ export CUDA_HOME
+++ PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/usr/local/cuda-12.1/include:1:/opt/amazon/efa/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/usr/sbin:1:/fsx/nouamane/miniconda/bin:1:/opt/slurm/bin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:2:/opt/amazon/openmpi/bin:1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1
+++ export PATH_modshare
+++ test 0
++ _mlstatus=0
++ '[' -n x ']'
++ IFS=' 	
+'
++ unset _mlre _mlv _mlrv _mlIFS
++ '[' -n '' ']'
++ unset _mlshdbg
++ return 0
++ source /fsx/nouamane/miniconda/bin/activate
+++ _CONDA_ROOT=/fsx/nouamane/miniconda
+++ . /fsx/nouamane/miniconda/etc/profile.d/conda.sh
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ '[' -z x ']'
+++ conda activate
+++ local cmd=activate
+++ case "$cmd" in
+++ __conda_activate activate
+++ '[' -n '' ']'
+++ local ask_conda
++++ PS1=
++++ __conda_exe shell.posix activate
++++ /fsx/nouamane/miniconda/bin/conda shell.posix activate
+++ ask_conda='. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_2='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\'''
+++ eval '. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_2='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\'''
++++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh
+++++ test -n ''
+++++ unset XML_CATALOG_FILES
+++++ unset xml_catalog_files_libxml2
++++ PS1='(base) '
++++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++++ export CONDA_PREFIX=/fsx/nouamane/miniconda
++++ CONDA_PREFIX=/fsx/nouamane/miniconda
++++ export CONDA_SHLVL=3
++++ CONDA_SHLVL=3
++++ export CONDA_DEFAULT_ENV=base
++++ CONDA_DEFAULT_ENV=base
++++ export 'CONDA_PROMPT_MODIFIER=(base) '
++++ CONDA_PROMPT_MODIFIER='(base) '
++++ export CONDA_PREFIX_2=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ CONDA_PREFIX_2=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ __conda_hashr
+++ '[' -n '' ']'
+++ '[' -n '' ']'
+++ hash -r
++ conda activate 2-1-cu121
++ local cmd=activate
++ case "$cmd" in
++ __conda_activate activate 2-1-cu121
++ '[' -n '' ']'
++ local ask_conda
+++ PS1='(base) '
+++ __conda_exe shell.posix activate 2-1-cu121
+++ /fsx/nouamane/miniconda/bin/conda shell.posix activate 2-1-cu121
++ ask_conda='PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''4'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_3='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
++ eval 'PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''4'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_3='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
+++ PS1='(2-1-cu121) '
+++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ export CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ export CONDA_SHLVL=4
+++ CONDA_SHLVL=4
+++ export CONDA_DEFAULT_ENV=2-1-cu121
+++ CONDA_DEFAULT_ENV=2-1-cu121
+++ export 'CONDA_PROMPT_MODIFIER=(2-1-cu121) '
+++ CONDA_PROMPT_MODIFIER='(2-1-cu121) '
+++ export CONDA_PREFIX_3=/fsx/nouamane/miniconda
+++ CONDA_PREFIX_3=/fsx/nouamane/miniconda
+++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ export _CE_M=
+++ _CE_M=
+++ export _CE_CONDA=
+++ _CE_CONDA=
+++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh
++++ test -n ''
++++ xml_catalog_files_libxml2=
++++ XML_CATALOG_FILES=
++++ conda_catalog_files=
++++ ifs_libxml2=' 	
+'
++++ IFS=' '
++++ rem=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ for pre in ${rem}
++++ test '' = /fsx/nouamane/miniconda/envs/2-1-cu121
++++ conda_catalog_files=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ rem=
++++ IFS=' 	
+'
++++ conda_catalog_files='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ export 'XML_CATALOG_FILES=file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ XML_CATALOG_FILES='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ unset conda_catalog_files ifs_libxml2 rem
++ __conda_hashr
++ '[' -n '' ']'
++ '[' -n '' ']'
++ hash -r
++ export PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++ PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ scontrol show hostnames 'ip-26-0-160-225,ip-26-0-164-[45,75,187,207,236],ip-26-0-166-[125,214,244],ip-26-0-167-9,ip-26-0-168-[95,120],ip-26-0-171-[88,102,168,230]'
++ export 'NODELIST=ip-26-0-160-225
+ip-26-0-164-45
+ip-26-0-164-75
+ip-26-0-164-187
+ip-26-0-164-207
+ip-26-0-164-236
+ip-26-0-166-125
+ip-26-0-166-214
+ip-26-0-166-244
+ip-26-0-167-9
+ip-26-0-168-95
+ip-26-0-168-120
+ip-26-0-171-88
+ip-26-0-171-102
+ip-26-0-171-168
+ip-26-0-171-230'
++ NODELIST='ip-26-0-160-225
+ip-26-0-164-45
+ip-26-0-164-75
+ip-26-0-164-187
+ip-26-0-164-207
+ip-26-0-164-236
+ip-26-0-166-125
+ip-26-0-166-214
+ip-26-0-166-244
+ip-26-0-167-9
+ip-26-0-168-95
+ip-26-0-168-120
+ip-26-0-171-88
+ip-26-0-171-102
+ip-26-0-171-168
+ip-26-0-171-230'
+++ head -n1
+++ scontrol show hostnames 'ip-26-0-160-225,ip-26-0-164-[45,75,187,207,236],ip-26-0-166-[125,214,244],ip-26-0-167-9,ip-26-0-168-[95,120],ip-26-0-171-[88,102,168,230]'
++ export MASTER_NODE=ip-26-0-160-225
++ MASTER_NODE=ip-26-0-160-225
++ export MASTER_PORT=12356
++ MASTER_PORT=12356
++ export NNODES=16
++ NNODES=16
++ export GPUS_PER_NODE=8
++ GPUS_PER_NODE=8
++ export WORLD_SIZE=128
++ WORLD_SIZE=128
++ export CUDA_DEVICE_MAX_CONNECTIONS=1
++ CUDA_DEVICE_MAX_CONNECTIONS=1
++ export NANOTRON_BENCHMARK=1
++ NANOTRON_BENCHMARK=1
++ echo 'Master node: ip-26-0-160-225'
+Master node: ip-26-0-160-225
++ echo 'All nodes: ip-26-0-160-225
+ip-26-0-164-45
+ip-26-0-164-75
+ip-26-0-164-187
+ip-26-0-164-207
+ip-26-0-164-236
+ip-26-0-166-125
+ip-26-0-166-214
+ip-26-0-166-244
+ip-26-0-167-9
+ip-26-0-168-95
+ip-26-0-168-120
+ip-26-0-171-88
+ip-26-0-171-102
+ip-26-0-171-168
+ip-26-0-171-230'
+All nodes: ip-26-0-160-225
+ip-26-0-164-45
+ip-26-0-164-75
+ip-26-0-164-187
+ip-26-0-164-207
+ip-26-0-164-236
+ip-26-0-166-125
+ip-26-0-166-214
+ip-26-0-166-244
+ip-26-0-167-9
+ip-26-0-168-95
+ip-26-0-168-120
+ip-26-0-171-88
+ip-26-0-171-102
+ip-26-0-171-168
+ip-26-0-171-230
++ echo 'World size: 128'
+World size: 128
++ srun torchrun --nnodes=16 --nproc_per_node=8 --rdzv_id=13265127 --rdzv_backend=c10d --rdzv_endpoint=ip-26-0-160-225:12356 run_train.py --config-file benchmark/configs/config_1.34G_dp64_tp2_pp1_acc8_mbs1_seq2048_zero1_tpmodeALL_vocab131k.yaml
+[2024-12-10 02:39:19,973] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-10 02:39:19,973] torch.distributed.run: [WARNING] 
+[2024-12-10 02:39:19,973] torch.distributed.run: [WARNING] *****************************************
+[2024-12-10 02:39:19,973] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-10 02:39:19,973] torch.distributed.run: [WARNING] *****************************************
+[2024-12-10 02:39:19,974] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-10 02:39:19,974] torch.distributed.run: [WARNING] 
+[2024-12-10 02:39:19,974] torch.distributed.run: [WARNING] *****************************************
+[2024-12-10 02:39:19,974] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-10 02:39:19,974] torch.distributed.run: [WARNING] *****************************************
+[2024-12-10 02:39:19,974] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-10 02:39:19,974] torch.distributed.run: [WARNING] 
+[2024-12-10 02:39:19,974] torch.distributed.run: [WARNING] *****************************************
+[2024-12-10 02:39:19,974] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-10 02:39:19,974] torch.distributed.run: [WARNING] *****************************************
+[2024-12-10 02:39:19,970] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-10 02:39:19,970] torch.distributed.run: [WARNING] 
+[2024-12-10 02:39:19,970] torch.distributed.run: [WARNING] *****************************************
+[2024-12-10 02:39:19,970] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-10 02:39:19,970] torch.distributed.run: [WARNING] *****************************************
+[2024-12-10 02:39:19,976] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-10 02:39:19,976] torch.distributed.run: [WARNING] 
+[2024-12-10 02:39:19,976] torch.distributed.run: [WARNING] *****************************************
+[2024-12-10 02:39:19,976] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-10 02:39:19,976] torch.distributed.run: [WARNING] *****************************************
+[2024-12-10 02:39:19,977] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-10 02:39:19,977] torch.distributed.run: [WARNING] 
+[2024-12-10 02:39:19,977] torch.distributed.run: [WARNING] *****************************************
+[2024-12-10 02:39:19,977] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-10 02:39:19,977] torch.distributed.run: [WARNING] *****************************************
+[2024-12-10 02:39:19,979] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-10 02:39:19,979] torch.distributed.run: [WARNING] 
+[2024-12-10 02:39:19,979] torch.distributed.run: [WARNING] *****************************************
+[2024-12-10 02:39:19,979] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-10 02:39:19,979] torch.distributed.run: [WARNING] *****************************************
+[2024-12-10 02:39:19,982] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-10 02:39:19,982] torch.distributed.run: [WARNING] 
+[2024-12-10 02:39:19,982] torch.distributed.run: [WARNING] *****************************************
+[2024-12-10 02:39:19,982] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-10 02:39:19,982] torch.distributed.run: [WARNING] *****************************************
+[2024-12-10 02:39:19,983] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-10 02:39:19,983] torch.distributed.run: [WARNING] 
+[2024-12-10 02:39:19,983] torch.distributed.run: [WARNING] *****************************************
+[2024-12-10 02:39:19,983] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-10 02:39:19,983] torch.distributed.run: [WARNING] *****************************************
+[2024-12-10 02:39:19,983] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-10 02:39:19,983] torch.distributed.run: [WARNING] 
+[2024-12-10 02:39:19,983] torch.distributed.run: [WARNING] *****************************************
+[2024-12-10 02:39:19,983] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-10 02:39:19,983] torch.distributed.run: [WARNING] *****************************************
+[2024-12-10 02:39:19,985] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-10 02:39:19,986] torch.distributed.run: [WARNING] 
+[2024-12-10 02:39:19,986] torch.distributed.run: [WARNING] *****************************************
+[2024-12-10 02:39:19,986] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-10 02:39:19,986] torch.distributed.run: [WARNING] *****************************************
+[2024-12-10 02:39:20,011] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-10 02:39:20,011] torch.distributed.run: [WARNING] 
+[2024-12-10 02:39:20,011] torch.distributed.run: [WARNING] *****************************************
+[2024-12-10 02:39:20,011] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-10 02:39:20,011] torch.distributed.run: [WARNING] *****************************************
+[2024-12-10 02:39:20,034] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-10 02:39:20,034] torch.distributed.run: [WARNING] 
+[2024-12-10 02:39:20,034] torch.distributed.run: [WARNING] *****************************************
+[2024-12-10 02:39:20,034] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-10 02:39:20,034] torch.distributed.run: [WARNING] *****************************************
+[2024-12-10 02:39:20,042] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-10 02:39:20,043] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-10 02:39:20,044] torch.distributed.run: [WARNING] 
+[2024-12-10 02:39:20,044] torch.distributed.run: [WARNING] *****************************************
+[2024-12-10 02:39:20,044] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-10 02:39:20,044] torch.distributed.run: [WARNING] *****************************************
+[2024-12-10 02:39:20,042] torch.distributed.run: [WARNING] 
+[2024-12-10 02:39:20,042] torch.distributed.run: [WARNING] *****************************************
+[2024-12-10 02:39:20,042] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-10 02:39:20,042] torch.distributed.run: [WARNING] *****************************************
+[2024-12-10 02:39:20,106] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-10 02:39:20,106] torch.distributed.run: [WARNING] 
+[2024-12-10 02:39:20,106] torch.distributed.run: [WARNING] *****************************************
+[2024-12-10 02:39:20,106] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-10 02:39:20,106] torch.distributed.run: [WARNING] *****************************************
+12/10/2024 02:40:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: Config:
+12/10/2024 02:40:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: Config(general=GeneralArgs(project='debug',
+12/10/2024 02:40:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                            run='1.34G_dp64_tp2_pp1_acc8_mbs1_seq2048_zero1_tpmodeALL_vocab131k',
+12/10/2024 02:40:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                            seed=42,
+12/10/2024 02:40:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                            step=None,
+12/10/2024 02:40:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                            consumed_train_samples=None,
+12/10/2024 02:40:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                            benchmark_csv_path=PosixPath('benchmark/results/bench_final.csv'),
+12/10/2024 02:40:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                            ignore_sanity_checks=True),
+12/10/2024 02:40:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:        parallelism=ParallelismArgs(dp=64,
+12/10/2024 02:40:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                    pp=1,
+12/10/2024 02:40:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                    tp=2,
+12/10/2024 02:40:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                    pp_engine=<nanotron.parallel.pipeline_parallel.engine.OneForwardOneBackwardPipelineEngine object at 0x7f5636ee0c10>,
+12/10/2024 02:40:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                    tp_mode=<TensorParallelLinearMode.ALL_REDUCE: 1>,
+12/10/2024 02:40:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                    tp_linear_async_communication=False,
+12/10/2024 02:40:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                    recompute_layer=False,
+12/10/2024 02:40:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                    tp_recompute_allgather=True,
+12/10/2024 02:40:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                    expert_parallel_size=1),
+12/10/2024 02:40:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:        model=ModelArgs(model_config=LlamaConfig(bos_token_id=0,
+12/10/2024 02:40:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 eos_token_id=0,
+12/10/2024 02:40:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 hidden_act='silu',
+12/10/2024 02:40:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 hidden_size=2048,
+12/10/2024 02:40:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 initializer_range=0.02,
+12/10/2024 02:40:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 intermediate_size=8192,
+12/10/2024 02:40:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 is_llama_config=True,
+12/10/2024 02:40:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 max_position_embeddings=2048,
+12/10/2024 02:40:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 num_attention_heads=32,
+12/10/2024 02:40:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 num_hidden_layers=16,
+12/10/2024 02:40:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 num_key_value_heads=8,
+12/10/2024 02:40:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 pad_token_id=None,
+12/10/2024 02:40:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 pretraining_tp=1,
+12/10/2024 02:40:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 rms_norm_eps=1e-05,
+12/10/2024 02:40:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 rope_scaling=None,
+12/10/2024 02:40:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 rope_theta=10000.0,
+12/10/2024 02:40:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 rope_interleaved=False,
+12/10/2024 02:40:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 tie_word_embeddings=True,
+12/10/2024 02:40:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 use_cache=True,
+12/10/2024 02:40:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 vocab_size=131072),
+12/10/2024 02:40:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                        init_method=RandomInit(std=0.02),
+12/10/2024 02:40:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                        dtype=torch.bfloat16,
+12/10/2024 02:40:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                        make_vocab_size_divisible_by=1,
+12/10/2024 02:40:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                        ddp_bucket_cap_mb=25),
+12/10/2024 02:40:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:        tokenizer=TokenizerArgs(tokenizer_name_or_path='robot-test/dummy-tokenizer-wordlevel',
+12/10/2024 02:40:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                tokenizer_revision=None,
+12/10/2024 02:40:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                tokenizer_max_length=None),
+12/10/2024 02:40:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:        checkpoints=CheckpointsArgs(checkpoints_path=PosixPath('checkpoints'),
+12/10/2024 02:40:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                    checkpoint_interval=10000,
+12/10/2024 02:40:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                    save_initial_state=False,
+12/10/2024 02:40:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                    save_final_state=False,
+12/10/2024 02:40:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                    resume_checkpoint_path=None,
+12/10/2024 02:40:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                    checkpoints_path_is_shared_file_system=False),
+12/10/2024 02:40:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:        logging=LoggingArgs(log_level='info',
+12/10/2024 02:40:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                            log_level_replica='info',
+12/10/2024 02:40:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                            iteration_step_info_interval=1),
+12/10/2024 02:40:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:        tokens=TokensArgs(sequence_length=2048,
+12/10/2024 02:40:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                          train_steps=100,
+12/10/2024 02:40:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                          micro_batch_size=1,
+12/10/2024 02:40:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                          batch_accumulation_per_replica=8,
+12/10/2024 02:40:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                          val_check_interval=100,
+12/10/2024 02:40:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                          limit_val_batches=0,
+12/10/2024 02:40:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                          limit_test_batches=0),
+12/10/2024 02:40:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:        optimizer=OptimizerArgs(optimizer_factory=AdamWOptimizerArgs(adam_eps=1e-08,
+12/10/2024 02:40:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                                     adam_beta1=0.9,
+12/10/2024 02:40:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                                     adam_beta2=0.95,
+12/10/2024 02:40:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                                     torch_adam_is_fused=True,
+12/10/2024 02:40:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                                     name='adamW'),
+12/10/2024 02:40:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                zero_stage=1,
+12/10/2024 02:40:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                weight_decay=0.01,
+12/10/2024 02:40:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                clip_grad=1.0,
+12/10/2024 02:40:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                accumulate_grad_in_fp32=True,
+12/10/2024 02:40:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                learning_rate_scheduler=LRSchedulerArgs(learning_rate=0.0003,
+12/10/2024 02:40:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                                        lr_warmup_steps=2,
+12/10/2024 02:40:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                                        lr_warmup_style='linear',
+12/10/2024 02:40:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                                        lr_decay_style='cosine',
+12/10/2024 02:40:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                                        lr_decay_steps=13,
+12/10/2024 02:40:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                                        lr_decay_starting_step=None,
+12/10/2024 02:40:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                                        min_decay_lr=1e-05)),
+12/10/2024 02:40:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:        data_stages=[DatasetStageArgs(name='Stable Training Stage',
+12/10/2024 02:40:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                      start_training_step=1,
+12/10/2024 02:40:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                      data=DataArgs(dataset=None,
+12/10/2024 02:40:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                    seed=42,
+12/10/2024 02:40:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                    num_loading_workers=1))],
+12/10/2024 02:40:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:        profiler=None,
+12/10/2024 02:40:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:        lighteval=None,
+12/10/2024 02:40:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:        s3_upload=None)
+12/10/2024 02:40:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: Model Config:
+12/10/2024 02:40:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: LlamaConfig(bos_token_id=0,
+12/10/2024 02:40:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             eos_token_id=0,
+12/10/2024 02:40:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             hidden_act='silu',
+12/10/2024 02:40:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             hidden_size=2048,
+12/10/2024 02:40:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             initializer_range=0.02,
+12/10/2024 02:40:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             intermediate_size=8192,
+12/10/2024 02:40:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             is_llama_config=True,
+12/10/2024 02:40:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             max_position_embeddings=2048,
+12/10/2024 02:40:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             num_attention_heads=32,
+12/10/2024 02:40:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             num_hidden_layers=16,
+12/10/2024 02:40:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             num_key_value_heads=8,
+12/10/2024 02:40:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             pad_token_id=None,
+12/10/2024 02:40:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             pretraining_tp=1,
+12/10/2024 02:40:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             rms_norm_eps=1e-05,
+12/10/2024 02:40:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             rope_scaling=None,
+12/10/2024 02:40:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             rope_theta=10000.0,
+12/10/2024 02:40:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             rope_interleaved=False,
+12/10/2024 02:40:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             tie_word_embeddings=True,
+12/10/2024 02:40:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             use_cache=True,
+12/10/2024 02:40:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             vocab_size=131072)
+12/10/2024 02:40:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: Building model..
+12/10/2024 02:40:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: Initialize RoPE Theta = 10000.0
+12/10/2024 02:40:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: Setting PP block ranks...
+12/10/2024 02:40:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: Total number of parameters: 1.24G (2368.26MiB)
+12/10/2024 02:40:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: Local number of parameters: 621M (1184.13MiB)
+12/10/2024 02:40:08 [INFO|DP=0|PP=0|TP=1|ip-26-0-160-225]: Local number of parameters: 621M (1184.13MiB)
+12/10/2024 02:40:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [After model building] Memory usage: 1184.15MiB. Peak allocated: 5392.00MiB Peak reserved: 29974.00MiB
+12/10/2024 02:40:08 [INFO|DP=0|PP=0|TP=1|ip-26-0-160-225]: [After model building] Memory usage: 1184.15MiB. Peak allocated: 5392.00MiB Peak reserved: 29974.00MiB
+12/10/2024 02:40:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: No checkpoint path provided.
+12/10/2024 02:40:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: Parametrizing model parameters using StandardParametrizator
+12/10/2024 02:40:17 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [Optimizer Building] Using LearningRateForSP as learning rate
+12/10/2024 02:40:17 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [ZeRO sharding] Size of optimizer params per rank:
+12/10/2024 02:40:17 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [ZeRO sharding] DP Rank 0 has 9.7M out of 621M (1.56%) params' optimizer states
+12/10/2024 02:40:17 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [ZeRO sharding] DP Rank 1 has 9.7M out of 621M (1.56%) params' optimizer states
+12/10/2024 02:40:17 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [ZeRO sharding] DP Rank 2 has 9.7M out of 621M (1.56%) params' optimizer states
+12/10/2024 02:40:17 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [ZeRO sharding] DP Rank 3 has 9.7M out of 621M (1.56%) params' optimizer states
+12/10/2024 02:40:17 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [ZeRO sharding] DP Rank 4 has 9.7M out of 621M (1.56%) params' optimizer states
+12/10/2024 02:40:17 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [ZeRO sharding] DP Rank 5 has 9.7M out of 621M (1.56%) params' optimizer states
+12/10/2024 02:40:17 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [ZeRO sharding] DP Rank 6 has 9.7M out of 621M (1.56%) params' optimizer states
+12/10/2024 02:40:17 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [ZeRO sharding] DP Rank 7 has 9.7M out of 621M (1.56%) params' optimizer states
+12/10/2024 02:40:17 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [ZeRO sharding] DP Rank 8 has 9.7M out of 621M (1.56%) params' optimizer states
+12/10/2024 02:40:17 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [ZeRO sharding] DP Rank 9 has 9.7M out of 621M (1.56%) params' optimizer states
+12/10/2024 02:40:17 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [ZeRO sharding] DP Rank 10 has 9.7M out of 621M (1.56%) params' optimizer states
+12/10/2024 02:40:17 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [ZeRO sharding] DP Rank 11 has 9.7M out of 621M (1.56%) params' optimizer states
+12/10/2024 02:40:17 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [ZeRO sharding] DP Rank 12 has 9.7M out of 621M (1.56%) params' optimizer states
+12/10/2024 02:40:17 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [ZeRO sharding] DP Rank 13 has 9.7M out of 621M (1.56%) params' optimizer states
+12/10/2024 02:40:17 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [ZeRO sharding] DP Rank 14 has 9.7M out of 621M (1.56%) params' optimizer states
+12/10/2024 02:40:17 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [ZeRO sharding] DP Rank 15 has 9.7M out of 621M (1.56%) params' optimizer states
+12/10/2024 02:40:17 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [ZeRO sharding] DP Rank 16 has 9.7M out of 621M (1.56%) params' optimizer states
+12/10/2024 02:40:17 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [ZeRO sharding] DP Rank 17 has 9.7M out of 621M (1.56%) params' optimizer states
+12/10/2024 02:40:17 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [ZeRO sharding] DP Rank 18 has 9.7M out of 621M (1.56%) params' optimizer states
+12/10/2024 02:40:17 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [ZeRO sharding] DP Rank 19 has 9.7M out of 621M (1.56%) params' optimizer states
+12/10/2024 02:40:17 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [ZeRO sharding] DP Rank 20 has 9.7M out of 621M (1.56%) params' optimizer states
+12/10/2024 02:40:17 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [ZeRO sharding] DP Rank 21 has 9.7M out of 621M (1.56%) params' optimizer states
+12/10/2024 02:40:17 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [ZeRO sharding] DP Rank 22 has 9.7M out of 621M (1.56%) params' optimizer states
+12/10/2024 02:40:17 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [ZeRO sharding] DP Rank 23 has 9.7M out of 621M (1.56%) params' optimizer states
+12/10/2024 02:40:17 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [ZeRO sharding] DP Rank 24 has 9.7M out of 621M (1.56%) params' optimizer states
+12/10/2024 02:40:17 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [ZeRO sharding] DP Rank 25 has 9.7M out of 621M (1.56%) params' optimizer states
+12/10/2024 02:40:17 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [ZeRO sharding] DP Rank 26 has 9.7M out of 621M (1.56%) params' optimizer states
+12/10/2024 02:40:17 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [ZeRO sharding] DP Rank 27 has 9.7M out of 621M (1.56%) params' optimizer states
+12/10/2024 02:40:17 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [ZeRO sharding] DP Rank 28 has 9.7M out of 621M (1.56%) params' optimizer states
+12/10/2024 02:40:17 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [ZeRO sharding] DP Rank 29 has 9.7M out of 621M (1.56%) params' optimizer states
+12/10/2024 02:40:17 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [ZeRO sharding] DP Rank 30 has 9.7M out of 621M (1.56%) params' optimizer states
+12/10/2024 02:40:17 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [ZeRO sharding] DP Rank 31 has 9.7M out of 621M (1.56%) params' optimizer states
+12/10/2024 02:40:17 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [ZeRO sharding] DP Rank 32 has 9.7M out of 621M (1.56%) params' optimizer states
+12/10/2024 02:40:17 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [ZeRO sharding] DP Rank 33 has 9.7M out of 621M (1.56%) params' optimizer states
+12/10/2024 02:40:17 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [ZeRO sharding] DP Rank 34 has 9.7M out of 621M (1.56%) params' optimizer states
+12/10/2024 02:40:17 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [ZeRO sharding] DP Rank 35 has 9.7M out of 621M (1.56%) params' optimizer states
+12/10/2024 02:40:17 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [ZeRO sharding] DP Rank 36 has 9.7M out of 621M (1.56%) params' optimizer states
+12/10/2024 02:40:17 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [ZeRO sharding] DP Rank 37 has 9.7M out of 621M (1.56%) params' optimizer states
+12/10/2024 02:40:17 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [ZeRO sharding] DP Rank 38 has 9.7M out of 621M (1.56%) params' optimizer states
+12/10/2024 02:40:17 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [ZeRO sharding] DP Rank 39 has 9.7M out of 621M (1.56%) params' optimizer states
+12/10/2024 02:40:17 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [ZeRO sharding] DP Rank 40 has 9.7M out of 621M (1.56%) params' optimizer states
+12/10/2024 02:40:17 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [ZeRO sharding] DP Rank 41 has 9.7M out of 621M (1.56%) params' optimizer states
+12/10/2024 02:40:17 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [ZeRO sharding] DP Rank 42 has 9.7M out of 621M (1.56%) params' optimizer states
+12/10/2024 02:40:17 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [ZeRO sharding] DP Rank 43 has 9.7M out of 621M (1.56%) params' optimizer states
+12/10/2024 02:40:17 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [ZeRO sharding] DP Rank 44 has 9.7M out of 621M (1.56%) params' optimizer states
+12/10/2024 02:40:17 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [ZeRO sharding] DP Rank 45 has 9.7M out of 621M (1.56%) params' optimizer states
+12/10/2024 02:40:17 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [ZeRO sharding] DP Rank 46 has 9.7M out of 621M (1.56%) params' optimizer states
+12/10/2024 02:40:17 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [ZeRO sharding] DP Rank 47 has 9.7M out of 621M (1.56%) params' optimizer states
+12/10/2024 02:40:17 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [ZeRO sharding] DP Rank 48 has 9.7M out of 621M (1.56%) params' optimizer states
+12/10/2024 02:40:17 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [ZeRO sharding] DP Rank 49 has 9.7M out of 621M (1.56%) params' optimizer states
+12/10/2024 02:40:17 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [ZeRO sharding] DP Rank 50 has 9.7M out of 621M (1.56%) params' optimizer states
+12/10/2024 02:40:17 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [ZeRO sharding] DP Rank 51 has 9.7M out of 621M (1.56%) params' optimizer states
+12/10/2024 02:40:17 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [ZeRO sharding] DP Rank 52 has 9.7M out of 621M (1.56%) params' optimizer states
+12/10/2024 02:40:17 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [ZeRO sharding] DP Rank 53 has 9.7M out of 621M (1.56%) params' optimizer states
+12/10/2024 02:40:17 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [ZeRO sharding] DP Rank 54 has 9.7M out of 621M (1.56%) params' optimizer states
+12/10/2024 02:40:17 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [ZeRO sharding] DP Rank 55 has 9.7M out of 621M (1.56%) params' optimizer states
+12/10/2024 02:40:17 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [ZeRO sharding] DP Rank 56 has 9.7M out of 621M (1.56%) params' optimizer states
+12/10/2024 02:40:17 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [ZeRO sharding] DP Rank 57 has 9.7M out of 621M (1.56%) params' optimizer states
+12/10/2024 02:40:17 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [ZeRO sharding] DP Rank 58 has 9.7M out of 621M (1.56%) params' optimizer states
+12/10/2024 02:40:17 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [ZeRO sharding] DP Rank 59 has 9.7M out of 621M (1.56%) params' optimizer states
+12/10/2024 02:40:17 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [ZeRO sharding] DP Rank 60 has 9.7M out of 621M (1.56%) params' optimizer states
+12/10/2024 02:40:17 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [ZeRO sharding] DP Rank 61 has 9.7M out of 621M (1.56%) params' optimizer states
+12/10/2024 02:40:17 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [ZeRO sharding] DP Rank 62 has 9.7M out of 621M (1.56%) params' optimizer states
+12/10/2024 02:40:17 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [ZeRO sharding] DP Rank 63 has 9.7M out of 621M (1.56%) params' optimizer states
+12/10/2024 02:40:17 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [Training Plan] Stage Stable Training Stage has 99 remaining training steps and has consumed 0 samples
+12/10/2024 02:40:17 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: Using dummy data generator
+12/10/2024 02:40:17 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [Training Plan] There are 1 training stages 
+12/10/2024 02:40:17 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [Stage Stable Training Stage] start from step 1 
+12/10/2024 02:40:17 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: 
+12/10/2024 02:40:17 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [Start training] datetime: 2024-12-10 02:40:17.756794 | mbs: 1 | grad_accum: 8 | global_batch_size: 512 | sequence_length: 2048 | train_steps: 100 | start_iteration_step: 0 | consumed_train_samples: 0
+wandb: WARNING Path /fsx/nouamane/.cache/wandb/wandb/ wasn't writable, using system temp directory.
+wandb: WARNING Path /fsx/nouamane/.cache/wandb/wandb/ wasn't writable, using system temp directory
+wandb: Tracking run with wandb version 0.16.0
+wandb: W&B syncing is set to `offline` in this directory.  
+wandb: Run `wandb online` or set WANDB_MODE=online to enable cloud syncing.
+12/10/2024 02:40:26 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: Resuming training from stage Stable Training Stage, it has trained for 0 samples and has 99 remaining train steps
+12/10/2024 02:40:26 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:  Memory usage: 3589.42MiB. Peak allocated 5392.00MiB. Peak reserved: 32344.00MiB
+12/10/2024 02:40:28 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:  Memory usage: 3657.45MiB. Peak allocated 6395.83MiB. Peak reserved: 7178.00MiB
+12/10/2024 02:40:28 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: iteration: 1 / 100 | consumed_tokens: 1.05M | elapsed_time_per_iteration_ms: 2.79K | tokens_per_sec: 375K | tokens_per_sec_per_gpu: 2.93K | global_batch_size: 512 | lm_loss: 12.2 | lr: 0.00015 | model_tflops_per_gpu: 24.2 | hardware_tflops_per_gpu: 24.2 | grad_norm: 0.861 | cuda_memory_allocated: 3.91G | cuda_max_memory_reserved: 8.83G | hd_total_memory_tb: 312G | hd_used_memory_tb: 67.5G | hd_free_memory_tb: 245G
+12/10/2024 02:40:28 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:  Memory usage: 3731.54MiB. Peak allocated 6118.32MiB. Peak reserved: 8418.00MiB
+12/10/2024 02:40:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:  Memory usage: 3731.54MiB. Peak allocated 6469.91MiB. Peak reserved: 8422.00MiB
+12/10/2024 02:40:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: iteration: 2 / 100 | consumed_tokens: 2.1M | elapsed_time_per_iteration_ms: 602 | tokens_per_sec: 1.74M | tokens_per_sec_per_gpu: 13.6K | global_batch_size: 512 | lm_loss: 12.2 | lr: 0.0003 | model_tflops_per_gpu: 112 | hardware_tflops_per_gpu: 112 | grad_norm: 0.86 | cuda_memory_allocated: 3.91G | cuda_max_memory_reserved: 8.83G | hd_total_memory_tb: 312G | hd_used_memory_tb: 67.5G | hd_free_memory_tb: 245G
+12/10/2024 02:40:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:  Memory usage: 3731.54MiB. Peak allocated 6118.33MiB. Peak reserved: 8422.00MiB
+12/10/2024 02:40:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:  Memory usage: 3731.54MiB. Peak allocated 6469.91MiB. Peak reserved: 8422.00MiB
+num_paramsnum_params
+
+{'total': 1241649152, 'local': 620824576}num_paramsnum_params
+{'total': 1241649152, 'local': 620824576}num_paramsnum_params
+
+
+
+
+{'total': 1241649152, 'local': 620824576}num_params{'total': 1241649152, 'local': 620824576}
+{'total': 1241649152, 'local': 620824576}
+
+{'total': 1241649152, 'local': 620824576}
+
+{'total': 1241649152, 'local': 620824576}
+num_paramsnum_params
+
+{'total': 1241649152, 'local': 620824576}
+{'total': 1241649152, 'local': 620824576}
+num_paramsnum_params
+num_params
+{'total': 1241649152, 'local': 620824576}
+num_paramsnum_paramsnum_params
+
+num_params
+{'total': 1241649152, 'local': 620824576}{'total': 1241649152, 'local': 620824576}
+{'total': 1241649152, 'local': 620824576}
+
+
+num_params
+{'total': 1241649152, 'local': 620824576}
+num_params{'total': 1241649152, 'local': 620824576}num_params
+
+num_paramsnum_paramsnum_params
+
+
+{'total': 1241649152, 'local': 620824576}num_params
+{'total': 1241649152, 'local': 620824576}{'total': 1241649152, 'local': 620824576}num_params
+
+
+
+num_params{'total': 1241649152, 'local': 620824576}num_params
+num_paramsnum_params
+
+{'total': 1241649152, 'local': 620824576}{'total': 1241649152, 'local': 620824576}
+
+num_paramsnum_params
+
+num_params{'total': 1241649152, 'local': 620824576}
+{'total': 1241649152, 'local': 620824576}
+
+num_paramsnum_params
+
+num_params{'total': 1241649152, 'local': 620824576}
+{'total': 1241649152, 'local': 620824576}
+
+num_params{'total': 1241649152, 'local': 620824576}
+
+num_paramsnum_paramsnum_params{'total': 1241649152, 'local': 620824576}num_params
+
+
+num_paramsnum_params
+num_params
+{'total': 1241649152, 'local': 620824576}num_paramsnum_params
+
+{'total': 1241649152, 'local': 620824576}
+
+
+{'total': 1241649152, 'local': 620824576}
+{'total': 1241649152, 'local': 620824576}{'total': 1241649152, 'local': 620824576}
+
+num_params
+{'total': 1241649152, 'local': 620824576}
+num_params
+num_params{'total': 1241649152, 'local': 620824576}
+{'total': 1241649152, 'local': 620824576}
+
+
+{'total': 1241649152, 'local': 620824576}num_params
+{'total': 1241649152, 'local': 620824576}
+num_paramsnum_params
+
+{'total': 1241649152, 'local': 620824576}
+{'total': 1241649152, 'local': 620824576}
+num_paramsnum_params
+{'total': 1241649152, 'local': 620824576}
+
+num_params{'total': 1241649152, 'local': 620824576}
+
+{'total': 1241649152, 'local': 620824576}
+num_paramsnum_paramsnum_params
+
+
+{'total': 1241649152, 'local': 620824576}
+{'total': 1241649152, 'local': 620824576}{'total': 1241649152, 'local': 620824576}
+num_params
+
+{'total': 1241649152, 'local': 620824576}
+{'total': 1241649152, 'local': 620824576}
+
+
+{'total': 1241649152, 'local': 620824576}
+{'total': 1241649152, 'local': 620824576}
+{'total': 1241649152, 'local': 620824576}
+
+
+{'total': 1241649152, 'local': 620824576}
+{'total': 1241649152, 'local': 620824576}
+{'total': 1241649152, 'local': 620824576}{'total': 1241649152, 'local': 620824576}
+
+num_paramsnum_params
+
+{'total': 1241649152, 'local': 620824576}
+{'total': 1241649152, 'local': 620824576}num_params
+
+num_paramsnum_params
+num_params
+{'total': 1241649152, 'local': 620824576}
+
+num_paramsnum_params{'total': 1241649152, 'local': 620824576}{'total': 1241649152, 'local': 620824576}num_paramsnum_params
+
+
+num_params
+
+
+{'total': 1241649152, 'local': 620824576}{'total': 1241649152, 'local': 620824576}
+num_paramsnum_params
+num_params
+{'total': 1241649152, 'local': 620824576}
+
+{'total': 1241649152, 'local': 620824576}
+{'total': 1241649152, 'local': 620824576}
+num_paramsnum_paramsnum_paramsnum_params
+
+
+{'total': 1241649152, 'local': 620824576}
+
+{'total': 1241649152, 'local': 620824576}
+
+{'total': 1241649152, 'local': 620824576}
+num_paramsnum_params
+num_params
+num_params{'total': 1241649152, 'local': 620824576}
+
+{'total': 1241649152, 'local': 620824576}
+
+num_params{'total': 1241649152, 'local': 620824576}
+{'total': 1241649152, 'local': 620824576}
+
+{'total': 1241649152, 'local': 620824576}
+{'total': 1241649152, 'local': 620824576}
+
+{'total': 1241649152, 'local': 620824576}
+num_paramsnum_params
+
+num_params{'total': 1241649152, 'local': 620824576}
+
+{'total': 1241649152, 'local': 620824576}
+{'total': 1241649152, 'local': 620824576}
+num_paramsnum_params
+
+{'total': 1241649152, 'local': 620824576}num_params
+{'total': 1241649152, 'local': 620824576}
+
+num_paramsnum_params{'total': 1241649152, 'local': 620824576}num_paramsnum_params
+
+
+
+
+{'total': 1241649152, 'local': 620824576}num_params
+{'total': 1241649152, 'local': 620824576}{'total': 1241649152, 'local': 620824576}
+num_paramsnum_params
+num_params
+
+{'total': 1241649152, 'local': 620824576}
+{'total': 1241649152, 'local': 620824576}
+{'total': 1241649152, 'local': 620824576}
+num_paramsnum_params
+num_params
+num_paramsnum_params{'total': 1241649152, 'local': 620824576}
+{'total': 1241649152, 'local': 620824576}
+
+
+
+{'total': 1241649152, 'local': 620824576}
+{'total': 1241649152, 'local': 620824576}
+{'total': 1241649152, 'local': 620824576}
+num_params
+{'total': 1241649152, 'local': 620824576}num_params
+num_params
+{'total': 1241649152, 'local': 620824576}num_params
+
+num_params{'total': 1241649152, 'local': 620824576}
+
+{'total': 1241649152, 'local': 620824576}
+
+{'total': 1241649152, 'local': 620824576}
+{'total': 1241649152, 'local': 620824576}
+
+{'total': 1241649152, 'local': 620824576}
+num_params
+num_paramsnum_params{'total': 1241649152, 'local': 620824576}
+
+
+{'total': 1241649152, 'local': 620824576}num_params{'total': 1241649152, 'local': 620824576}
+{'total': 1241649152, 'local': 620824576}{'total': 1241649152, 'local': 620824576}
+
+num_paramsnum_paramsnum_paramsnum_params
+
+
+{'total': 1241649152, 'local': 620824576}
+{'total': 1241649152, 'local': 620824576}
+
+{'total': 1241649152, 'local': 620824576}
+num_paramsnum_params
+num_params
+{'total': 1241649152, 'local': 620824576}
+
+num_params
+{'total': 1241649152, 'local': 620824576}
+num_params
+{'total': 1241649152, 'local': 620824576}num_params
+
+{'total': 1241649152, 'local': 620824576}
+
+{'total': 1241649152, 'local': 620824576}
+
+{'total': 1241649152, 'local': 620824576}
+num_params
+{'total': 1241649152, 'local': 620824576}
+
+{'total': 1241649152, 'local': 620824576}num_params
+num_params
+{'total': 1241649152, 'local': 620824576}
+{'total': 1241649152, 'local': 620824576}
+
+
+{'total': 1241649152, 'local': 620824576}
+num_params
+num_params
+{'total': 1241649152, 'local': 620824576}
+{'total': 1241649152, 'local': 620824576}
+num_paramsnum_params
+
+{'total': 1241649152, 'local': 620824576}{'total': 1241649152, 'local': 620824576}
+
+{'total': 1241649152, 'local': 620824576}
+{'total': 1241649152, 'local': 620824576}
+
+{'total': 1241649152, 'local': 620824576}
+{'total': 1241649152, 'local': 620824576}
+num_params
+{'total': 1241649152, 'local': 620824576}
+num_params
+{'total': 1241649152, 'local': 620824576}
+12/10/2024 02:40:30 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: iteration: 3 / 100 | consumed_tokens: 3.15M | elapsed_time_per_iteration_ms: 600 | tokens_per_sec: 1.75M | tokens_per_sec_per_gpu: 13.7K | global_batch_size: 512 | lm_loss: 12.2 | lr: 0.000296 | model_tflops_per_gpu: 113 | hardware_tflops_per_gpu: 113 | grad_norm: 0.849 | cuda_memory_allocated: 3.91G | cuda_max_memory_reserved: 8.83G | hd_total_memory_tb: 312G | hd_used_memory_tb: 67.5G | hd_free_memory_tb: 245G
+num_params
+{'total': 1241649152, 'local': 620824576}
+12/10/2024 02:40:30 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: | job_id   | name                                                           | nodes | seq_len | mbs | batch_accum | gbs | mTFLOPs | hTFLOPs | tok/s/gpu | AllReduce (GB/s) | AllGather (GB/s) | ReduceScatter (GB/s) | AR Intra-node (GB/s) | AG Intra-node (GB/s) | RS Intra-node (GB/s) | Mem Alloc (GB) | Mem Res (GB) | dp | pp | tp | pp_engine                                                                                                 | tp_mode                             | tp_async_comm | hidden_size | hidden_act | num_layers | num_heads | num_kv_heads | max_pos | vocab_size | tie_word_embeddings | dtype          | zero_stage | ddp_bucket_cap_mb | accumulate_grad_in_fp32 | Total Params | Local Params |
+12/10/2024 02:40:30 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: | -------- | -------------------------------------------------------------- | ----- | ------- | --- | ----------- | --- | ------- | ------- | --------- | ---------------- | ---------------- | -------------------- | -------------------- | -------------------- | -------------------- | -------------- | ------------ | -- | -- | -- | --------------------------------------------------------------------------------------------------------- | ----------------------------------- | ------------- | ----------- | ---------- | ---------- | --------- | ------------ | ------- | ---------- | ------------------- | -------------- | ---------- | ----------------- | ----------------------- | ------------ | ------------ |
+12/10/2024 02:40:30 [INFO|DP=28|PP=0|TP=1|ip-26-0-166-214]: Throughput logging complete
+12/10/2024 02:40:30 [INFO|DP=16|PP=0|TP=1|ip-26-0-164-45]: Throughput logging complete
+12/10/2024 02:40:30 [INFO|DP=12|PP=0|TP=0|ip-26-0-164-236]: Throughput logging complete
+12/10/2024 02:40:30 [INFO|DP=12|PP=0|TP=1|ip-26-0-164-236]: Throughput logging complete
+12/10/2024 02:40:30 [INFO|DP=48|PP=0|TP=1|ip-26-0-171-102]: Throughput logging complete
+12/10/2024 02:40:30 [INFO|DP=33|PP=0|TP=0|ip-26-0-166-244]: Throughput logging complete
+12/10/2024 02:40:30 [INFO|DP=32|PP=0|TP=0|ip-26-0-166-244]: Throughput logging complete
+12/10/2024 02:40:30 [INFO|DP=32|PP=0|TP=1|ip-26-0-166-244]: Throughput logging complete
+12/10/2024 02:40:30 [INFO|DP=40|PP=0|TP=1|ip-26-0-168-120]: Throughput logging complete
+12/10/2024 02:40:30 [INFO|DP=40|PP=0|TP=0|ip-26-0-168-120]: Throughput logging complete
+12/10/2024 02:40:30 [INFO|DP=50|PP=0|TP=0|ip-26-0-171-102]: Throughput logging complete
+12/10/2024 02:40:30 [INFO|DP=49|PP=0|TP=1|ip-26-0-171-102]: Throughput logging complete
+12/10/2024 02:40:30 [INFO|DP=56|PP=0|TP=0|ip-26-0-171-230]: Throughput logging complete
+12/10/2024 02:40:30 [INFO|DP=48|PP=0|TP=0|ip-26-0-171-102]: Throughput logging complete
+12/10/2024 02:40:30 [INFO|DP=16|PP=0|TP=0|ip-26-0-164-45]: Throughput logging complete
+12/10/2024 02:40:30 [INFO|DP=34|PP=0|TP=0|ip-26-0-166-244]: Throughput logging complete
+12/10/2024 02:40:30 [INFO|DP=33|PP=0|TP=1|ip-26-0-166-244]: Throughput logging complete
+12/10/2024 02:40:30 [INFO|DP=49|PP=0|TP=0|ip-26-0-171-102]: Throughput logging complete
+12/10/2024 02:40:30 [INFO|DP=50|PP=0|TP=1|ip-26-0-171-102]: Throughput logging complete
+12/10/2024 02:40:30 [INFO|DP=24|PP=0|TP=1|ip-26-0-166-125]: Throughput logging complete
+12/10/2024 02:40:30 [INFO|DP=52|PP=0|TP=1|ip-26-0-171-168]: Throughput logging complete
+12/10/2024 02:40:30 [INFO|DP=17|PP=0|TP=1|ip-26-0-164-45]: Throughput logging complete
+12/10/2024 02:40:30 [INFO|DP=17|PP=0|TP=0|ip-26-0-164-45]: Throughput logging complete
+12/10/2024 02:40:30 [INFO|DP=56|PP=0|TP=1|ip-26-0-171-230]: Throughput logging complete
+12/10/2024 02:40:30 [INFO|DP=20|PP=0|TP=0|ip-26-0-164-75]: Throughput logging complete
+12/10/2024 02:40:30 [INFO|DP=18|PP=0|TP=0|ip-26-0-164-45]: Throughput logging complete
+12/10/2024 02:40:30 [INFO|DP=14|PP=0|TP=1|ip-26-0-164-236]: Throughput logging complete
+12/10/2024 02:40:30 [INFO|DP=13|PP=0|TP=1|ip-26-0-164-236]: Throughput logging complete
+12/10/2024 02:40:30 [INFO|DP=8|PP=0|TP=0|ip-26-0-164-207]: Throughput logging complete
+12/10/2024 02:40:30 [INFO|DP=8|PP=0|TP=1|ip-26-0-164-207]: Throughput logging complete
+12/10/2024 02:40:30 [INFO|DP=60|PP=0|TP=1|ip-26-0-171-88]: Throughput logging complete
+12/10/2024 02:40:30 [INFO|DP=60|PP=0|TP=0|ip-26-0-171-88]: Throughput logging complete
+12/10/2024 02:40:30 [INFO|DP=18|PP=0|TP=1|ip-26-0-164-45]: Throughput logging complete
+12/10/2024 02:40:30 [INFO|DP=24|PP=0|TP=0|ip-26-0-166-125]: Throughput logging complete
+12/10/2024 02:40:30 [INFO|DP=36|PP=0|TP=1|ip-26-0-167-9]: Throughput logging complete
+12/10/2024 02:40:30 [INFO|DP=34|PP=0|TP=1|ip-26-0-166-244]: Throughput logging complete
+12/10/2024 02:40:30 [INFO|DP=14|PP=0|TP=0|ip-26-0-164-236]: Throughput logging complete
+12/10/2024 02:40:30 [INFO|DP=29|PP=0|TP=0|ip-26-0-166-214]: Throughput logging complete
+12/10/2024 02:40:30 [INFO|DP=30|PP=0|TP=1|ip-26-0-166-214]: Throughput logging complete
+12/10/2024 02:40:30 [INFO|DP=30|PP=0|TP=0|ip-26-0-166-214]: Throughput logging complete
+12/10/2024 02:40:30 [INFO|DP=29|PP=0|TP=1|ip-26-0-166-214]: Throughput logging complete
+12/10/2024 02:40:30 [INFO|DP=13|PP=0|TP=0|ip-26-0-164-236]: Throughput logging complete
+12/10/2024 02:40:30 [INFO|DP=9|PP=0|TP=0|ip-26-0-164-207]: Throughput logging complete
+12/10/2024 02:40:30 [INFO|DP=52|PP=0|TP=0|ip-26-0-171-168]: Throughput logging complete
+12/10/2024 02:40:30 [INFO|DP=26|PP=0|TP=0|ip-26-0-166-125]: Throughput logging complete
+12/10/2024 02:40:30 [INFO|DP=44|PP=0|TP=1|ip-26-0-168-95]: Throughput logging complete
+12/10/2024 02:40:30 [INFO|DP=45|PP=0|TP=0|ip-26-0-168-95]: Throughput logging complete
+12/10/2024 02:40:30 [INFO|DP=45|PP=0|TP=1|ip-26-0-168-95]: Throughput logging complete
+12/10/2024 02:40:30 [INFO|DP=15|PP=0|TP=0|ip-26-0-164-236]: Throughput logging complete
+12/10/2024 02:40:30 [INFO|DP=28|PP=0|TP=0|ip-26-0-166-214]: Throughput logging complete
+12/10/2024 02:40:30 [INFO|DP=20|PP=0|TP=1|ip-26-0-164-75]: Throughput logging complete
+12/10/2024 02:40:30 [INFO|DP=36|PP=0|TP=0|ip-26-0-167-9]: Throughput logging complete
+12/10/2024 02:40:30 [INFO|DP=4|PP=0|TP=0|ip-26-0-164-187]: Throughput logging complete
+12/10/2024 02:40:30 [INFO|DP=4|PP=0|TP=1|ip-26-0-164-187]: Throughput logging complete
+12/10/2024 02:40:30 [INFO|DP=31|PP=0|TP=1|ip-26-0-166-214]: Throughput logging complete
+12/10/2024 02:40:30 [INFO|DP=46|PP=0|TP=0|ip-26-0-168-95]: Throughput logging complete
+12/10/2024 02:40:30 [INFO|DP=61|PP=0|TP=0|ip-26-0-171-88]: Throughput logging complete
+12/10/2024 02:40:30 [INFO|DP=15|PP=0|TP=1|ip-26-0-164-236]: Throughput logging complete
+12/10/2024 02:40:30 [INFO|DP=25|PP=0|TP=0|ip-26-0-166-125]: Throughput logging complete
+12/10/2024 02:40:30 [INFO|DP=47|PP=0|TP=0|ip-26-0-168-95]: Throughput logging complete
+12/10/2024 02:40:30 [INFO|DP=57|PP=0|TP=0|ip-26-0-171-230]: Throughput logging complete
+12/10/2024 02:40:30 [INFO|DP=31|PP=0|TP=0|ip-26-0-166-214]: Throughput logging complete
+12/10/2024 02:40:30 [INFO|DP=5|PP=0|TP=0|ip-26-0-164-187]: Throughput logging complete
+12/10/2024 02:40:30 [INFO|DP=41|PP=0|TP=0|ip-26-0-168-120]: Throughput logging complete
+12/10/2024 02:40:30 [INFO|DP=42|PP=0|TP=0|ip-26-0-168-120]: Throughput logging complete
+12/10/2024 02:40:30 [INFO|DP=54|PP=0|TP=0|ip-26-0-171-168]: Throughput logging complete
+12/10/2024 02:40:30 [INFO|DP=53|PP=0|TP=0|ip-26-0-171-168]: Throughput logging complete
+12/10/2024 02:40:30 [INFO|DP=58|PP=0|TP=0|ip-26-0-171-230]: Throughput logging complete
+12/10/2024 02:40:30 [INFO|DP=51|PP=0|TP=1|ip-26-0-171-102]: Throughput logging complete
+12/10/2024 02:40:30 [INFO|DP=51|PP=0|TP=0|ip-26-0-171-102]: Throughput logging complete
+12/10/2024 02:40:30 [INFO|DP=10|PP=0|TP=0|ip-26-0-164-207]: Throughput logging complete
+12/10/2024 02:40:30 [INFO|DP=61|PP=0|TP=1|ip-26-0-171-88]: Throughput logging complete
+12/10/2024 02:40:30 [INFO|DP=26|PP=0|TP=1|ip-26-0-166-125]: Throughput logging complete
+12/10/2024 02:40:30 [INFO|DP=25|PP=0|TP=1|ip-26-0-166-125]: Throughput logging complete
+12/10/2024 02:40:30 [INFO|DP=35|PP=0|TP=0|ip-26-0-166-244]: Throughput logging complete
+12/10/2024 02:40:30 [INFO|DP=58|PP=0|TP=1|ip-26-0-171-230]: Throughput logging complete
+12/10/2024 02:40:30 [INFO|DP=41|PP=0|TP=1|ip-26-0-168-120]: Throughput logging complete
+12/10/2024 02:40:30 [INFO|DP=21|PP=0|TP=0|ip-26-0-164-75]: Throughput logging complete
+12/10/2024 02:40:30 [INFO|DP=19|PP=0|TP=0|ip-26-0-164-45]: Throughput logging complete
+12/10/2024 02:40:30 [INFO|DP=19|PP=0|TP=1|ip-26-0-164-45]: Throughput logging complete
+12/10/2024 02:40:30 [INFO|DP=6|PP=0|TP=0|ip-26-0-164-187]: Throughput logging complete
+12/10/2024 02:40:30 [INFO|DP=53|PP=0|TP=1|ip-26-0-171-168]: Throughput logging complete
+12/10/2024 02:40:30 [INFO|DP=42|PP=0|TP=1|ip-26-0-168-120]: Throughput logging complete
+12/10/2024 02:40:30 [INFO|DP=37|PP=0|TP=0|ip-26-0-167-9]: Throughput logging complete
+12/10/2024 02:40:30 [INFO|DP=38|PP=0|TP=1|ip-26-0-167-9]: Throughput logging complete
+12/10/2024 02:40:30 [INFO|DP=38|PP=0|TP=0|ip-26-0-167-9]: Throughput logging complete
+12/10/2024 02:40:30 [INFO|DP=62|PP=0|TP=0|ip-26-0-171-88]: Throughput logging complete
+12/10/2024 02:40:30 [INFO|DP=47|PP=0|TP=1|ip-26-0-168-95]: Throughput logging complete
+12/10/2024 02:40:30 [INFO|DP=22|PP=0|TP=0|ip-26-0-164-75]: Throughput logging complete
+12/10/2024 02:40:30 [INFO|DP=62|PP=0|TP=1|ip-26-0-171-88]: Throughput logging complete
+12/10/2024 02:40:30 [INFO|DP=22|PP=0|TP=1|ip-26-0-164-75]: Throughput logging complete
+12/10/2024 02:40:30 [INFO|DP=27|PP=0|TP=0|ip-26-0-166-125]: Throughput logging complete
+12/10/2024 02:40:30 [INFO|DP=37|PP=0|TP=1|ip-26-0-167-9]: Throughput logging complete
+12/10/2024 02:40:30 [INFO|DP=6|PP=0|TP=1|ip-26-0-164-187]: Throughput logging complete
+12/10/2024 02:40:30 [INFO|DP=54|PP=0|TP=1|ip-26-0-171-168]: Throughput logging complete
+12/10/2024 02:40:30 [INFO|DP=10|PP=0|TP=1|ip-26-0-164-207]: Throughput logging complete
+12/10/2024 02:40:30 [INFO|DP=43|PP=0|TP=1|ip-26-0-168-120]: Throughput logging complete
+12/10/2024 02:40:30 [INFO|DP=55|PP=0|TP=0|ip-26-0-171-168]: Throughput logging complete
+12/10/2024 02:40:30 [INFO|DP=59|PP=0|TP=1|ip-26-0-171-230]: Throughput logging complete
+12/10/2024 02:40:30 [INFO|DP=59|PP=0|TP=0|ip-26-0-171-230]: Throughput logging complete
+12/10/2024 02:40:30 [INFO|DP=21|PP=0|TP=1|ip-26-0-164-75]: Throughput logging complete
+12/10/2024 02:40:30 [INFO|DP=43|PP=0|TP=0|ip-26-0-168-120]: Throughput logging complete
+12/10/2024 02:40:30 [INFO|DP=9|PP=0|TP=1|ip-26-0-164-207]: Throughput logging complete
+12/10/2024 02:40:30 [INFO|DP=11|PP=0|TP=0|ip-26-0-164-207]: Throughput logging complete
+12/10/2024 02:40:30 [INFO|DP=23|PP=0|TP=1|ip-26-0-164-75]: Throughput logging complete
+12/10/2024 02:40:30 [INFO|DP=55|PP=0|TP=1|ip-26-0-171-168]: Throughput logging complete
+12/10/2024 02:40:30 [INFO|DP=23|PP=0|TP=0|ip-26-0-164-75]: Throughput logging complete
+12/10/2024 02:40:30 [INFO|DP=63|PP=0|TP=1|ip-26-0-171-88]: Throughput logging complete
+12/10/2024 02:40:30 [INFO|DP=63|PP=0|TP=0|ip-26-0-171-88]: Throughput logging complete
+12/10/2024 02:40:30 [INFO|DP=27|PP=0|TP=1|ip-26-0-166-125]: Throughput logging complete
+12/10/2024 02:40:30 [INFO|DP=39|PP=0|TP=0|ip-26-0-167-9]: Throughput logging complete
+12/10/2024 02:40:30 [INFO|DP=39|PP=0|TP=1|ip-26-0-167-9]: Throughput logging complete
+12/10/2024 02:40:30 [INFO|DP=7|PP=0|TP=1|ip-26-0-164-187]: Throughput logging complete
+12/10/2024 02:40:30 [INFO|DP=7|PP=0|TP=0|ip-26-0-164-187]: Throughput logging complete
+12/10/2024 02:40:30 [INFO|DP=46|PP=0|TP=1|ip-26-0-168-95]: Throughput logging complete
+12/10/2024 02:40:30 [INFO|DP=11|PP=0|TP=1|ip-26-0-164-207]: Throughput logging complete
+12/10/2024 02:40:30 [INFO|DP=57|PP=0|TP=1|ip-26-0-171-230]: Throughput logging complete
+12/10/2024 02:40:30 [INFO|DP=5|PP=0|TP=1|ip-26-0-164-187]: Throughput logging complete
+12/10/2024 02:40:30 [INFO|DP=44|PP=0|TP=0|ip-26-0-168-95]: Throughput logging complete
+12/10/2024 02:40:30 [INFO|DP=35|PP=0|TP=1|ip-26-0-166-244]: Throughput logging complete
+[2024-12-10 02:40:30,196] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-10 02:40:30,196] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3060214 closing signal SIGTERM
+[2024-12-10 02:40:30,196] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3060215 closing signal SIGTERM
+[2024-12-10 02:40:30,197] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3060216 closing signal SIGTERM
+[2024-12-10 02:40:30,197] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3060217 closing signal SIGTERM
+[2024-12-10 02:40:30,196] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-10 02:40:30,197] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1269573 closing signal SIGTERM
+[2024-12-10 02:40:30,197] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1269574 closing signal SIGTERM
+[2024-12-10 02:40:30,196] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-10 02:40:30,197] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1269575 closing signal SIGTERM
+[2024-12-10 02:40:30,196] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-10 02:40:30,197] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1312410 closing signal SIGTERM
+[2024-12-10 02:40:30,197] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1269576 closing signal SIGTERM
+[2024-12-10 02:40:30,192] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-10 02:40:30,197] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1312411 closing signal SIGTERM
+[2024-12-10 02:40:30,197] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-10 02:40:30,192] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 104603 closing signal SIGTERM
+[2024-12-10 02:40:30,197] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1312412 closing signal SIGTERM
+[2024-12-10 02:40:30,196] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 56319 closing signal SIGTERM
+[2024-12-10 02:40:30,197] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-10 02:40:30,197] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2621459 closing signal SIGTERM
+[2024-12-10 02:40:30,192] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 104604 closing signal SIGTERM
+[2024-12-10 02:40:30,197] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1312413 closing signal SIGTERM
+[2024-12-10 02:40:30,197] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1312414 closing signal SIGTERM
+[2024-12-10 02:40:30,197] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2621460 closing signal SIGTERM
+[2024-12-10 02:40:30,198] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 961668 closing signal SIGTERM
+[2024-12-10 02:40:30,197] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 56320 closing signal SIGTERM
+[2024-12-10 02:40:30,196] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-10 02:40:30,197] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-10 02:40:30,197] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-10 02:40:30,198] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 961669 closing signal SIGTERM
+[2024-12-10 02:40:30,197] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2621461 closing signal SIGTERM
+[2024-12-10 02:40:30,197] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3886974 closing signal SIGTERM
+[2024-12-10 02:40:30,198] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2763430 closing signal SIGTERM
+[2024-12-10 02:40:30,197] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2621462 closing signal SIGTERM
+[2024-12-10 02:40:30,197] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 103833 closing signal SIGTERM
+[2024-12-10 02:40:30,198] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 961670 closing signal SIGTERM
+[2024-12-10 02:40:30,198] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-10 02:40:30,198] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 961671 closing signal SIGTERM
+[2024-12-10 02:40:30,198] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2763431 closing signal SIGTERM
+[2024-12-10 02:40:30,197] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3886975 closing signal SIGTERM
+[2024-12-10 02:40:30,197] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 103834 closing signal SIGTERM
+[2024-12-10 02:40:30,198] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2763432 closing signal SIGTERM
+[2024-12-10 02:40:30,198] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 124468 closing signal SIGTERM
+[2024-12-10 02:40:30,198] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 124469 closing signal SIGTERM
+[2024-12-10 02:40:30,197] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-10 02:40:30,197] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3886976 closing signal SIGTERM
+[2024-12-10 02:40:30,197] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 56321 closing signal SIGTERM
+[2024-12-10 02:40:30,197] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 103835 closing signal SIGTERM
+[2024-12-10 02:40:30,197] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 103836 closing signal SIGTERM
+[2024-12-10 02:40:30,198] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2576485 closing signal SIGTERM
+[2024-12-10 02:40:30,198] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2576486 closing signal SIGTERM
+[2024-12-10 02:40:30,198] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2763433 closing signal SIGTERM
+[2024-12-10 02:40:30,198] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2763434 closing signal SIGTERM
+[2024-12-10 02:40:30,198] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2576487 closing signal SIGTERM
+[2024-12-10 02:40:30,192] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 104605 closing signal SIGTERM
+[2024-12-10 02:40:30,192] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 104606 closing signal SIGTERM
+[2024-12-10 02:40:30,198] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2576488 closing signal SIGTERM
+[2024-12-10 02:40:30,198] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2576489 closing signal SIGTERM
+[2024-12-10 02:40:30,198] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2576490 closing signal SIGTERM
+[2024-12-10 02:40:30,197] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3060218 closing signal SIGTERM
+[2024-12-10 02:40:30,198] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3060219 closing signal SIGTERM
+[2024-12-10 02:40:30,199] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 961672 closing signal SIGTERM
+[2024-12-10 02:40:30,198] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 124470 closing signal SIGTERM
+12/10/2024 02:40:30 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: | 13265127 | 1.34G_dp64_tp2_pp1_acc8_mbs1_seq2048_zero1_tpmodeALL_vocab131k | 16    | 2048    | 1   | 8           | 512 | 112.76  | 112.76  | 13660.19  | 187.29           | 110.47           | 110.67               | 457.43               | 265.14               | 264.70               | 5.97           | 8.22         | 64 | 1  | 2  | <nanotron.parallel.pipeline_parallel.engine.OneForwardOneBackwardPipelineEngine object at 0x7f5636ee0c10> | TensorParallelLinearMode.ALL_REDUCE | False         | 2048        | silu       | 16         | 32        | 8            | 2048    | 131072     | True                | torch.bfloat16 | 1          | 25                | True                    | 1.24G        | 621M         |
+12/10/2024 02:40:30 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: Throughput logging complete
+12/10/2024 02:40:30 [INFO|DP=0|PP=0|TP=1|ip-26-0-160-225]: Throughput logging complete
+12/10/2024 02:40:30 [INFO|DP=1|PP=0|TP=0|ip-26-0-160-225]: Throughput logging complete
+12/10/2024 02:40:30 [INFO|DP=2|PP=0|TP=0|ip-26-0-160-225]: Throughput logging complete
+12/10/2024 02:40:30 [INFO|DP=1|PP=0|TP=1|ip-26-0-160-225]: Throughput logging complete
+12/10/2024 02:40:30 [INFO|DP=2|PP=0|TP=1|ip-26-0-160-225]: Throughput logging complete
+12/10/2024 02:40:30 [INFO|DP=3|PP=0|TP=1|ip-26-0-160-225]: Throughput logging complete
+12/10/2024 02:40:30 [INFO|DP=3|PP=0|TP=0|ip-26-0-160-225]: Throughput logging complete
+[2024-12-10 02:40:30,197] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2621463 closing signal SIGTERM
+[2024-12-10 02:40:30,197] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1269577 closing signal SIGTERM
+[2024-12-10 02:40:30,198] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1269578 closing signal SIGTERM
+[2024-12-10 02:40:30,198] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1269579 closing signal SIGTERM
+[2024-12-10 02:40:30,200] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+slurmstepd: error: *** JOB 13265127 ON ip-26-0-160-225 CANCELLED AT 2024-12-10T02:40:30 ***
+[2024-12-10 02:40:30,200] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2142894 closing signal SIGTERM
+[2024-12-10 02:40:30,200] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2142895 closing signal SIGTERM
+[2024-12-10 02:40:30,200] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2142896 closing signal SIGTERM
+[2024-12-10 02:40:30,200] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2142897 closing signal SIGTERM
+[2024-12-10 02:40:30,200] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 124471 closing signal SIGTERM
+[2024-12-10 02:40:30,199] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 56322 closing signal SIGTERM
+[2024-12-10 02:40:30,200] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 124472 closing signal SIGTERM
+[2024-12-10 02:40:30,199] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 56323 closing signal SIGTERM
+[2024-12-10 02:40:30,200] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 124473 closing signal SIGTERM
+[2024-12-10 02:40:30,199] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 56324 closing signal SIGTERM
+[2024-12-10 02:40:30,200] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2142898 closing signal SIGTERM
+[2024-12-10 02:40:30,197] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3886977 closing signal SIGTERM
+[2024-12-10 02:40:30,199] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3886978 closing signal SIGTERM
+[2024-12-10 02:40:30,200] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 961673 closing signal SIGTERM
+[2024-12-10 02:40:30,199] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3886979 closing signal SIGTERM
+[2024-12-10 02:40:30,200] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 961674 closing signal SIGTERM
+[2024-12-10 02:40:30,200] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 961675 closing signal SIGTERM
+[2024-12-10 02:40:30,200] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3886980 closing signal SIGTERM
+[2024-12-10 02:40:30,200] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1312415 closing signal SIGTERM
+[2024-12-10 02:40:30,200] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1269580 closing signal SIGTERM
+[2024-12-10 02:40:30,198] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 103837 closing signal SIGTERM
+[2024-12-10 02:40:30,199] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 56325 closing signal SIGTERM
+[2024-12-10 02:40:30,200] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 103838 closing signal SIGTERM
+[2024-12-10 02:40:30,200] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 103839 closing signal SIGTERM
+[2024-12-10 02:40:30,200] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-10 02:40:30,200] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2841505 closing signal SIGTERM
+[2024-12-10 02:40:30,201] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3060220 closing signal SIGTERM
+[2024-12-10 02:40:30,200] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 56326 closing signal SIGTERM
+[2024-12-10 02:40:30,200] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2763435 closing signal SIGTERM
+[2024-12-10 02:40:30,200] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3886981 closing signal SIGTERM
+[2024-12-10 02:40:30,200] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 103840 closing signal SIGTERM
+[2024-12-10 02:40:30,201] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2841506 closing signal SIGTERM
+[2024-12-10 02:40:30,199] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2621464 closing signal SIGTERM
+[2024-12-10 02:40:30,201] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3060221 closing signal SIGTERM
+[2024-12-10 02:40:30,201] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 124474 closing signal SIGTERM
+[2024-12-10 02:40:30,201] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2763436 closing signal SIGTERM
+[2024-12-10 02:40:30,201] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1312416 closing signal SIGTERM
+[2024-12-10 02:40:30,201] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1312417 closing signal SIGTERM
+[2024-12-10 02:40:30,202] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2142899 closing signal SIGTERM
+[2024-12-10 02:40:30,201] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 124475 closing signal SIGTERM
+[2024-12-10 02:40:30,200] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2621465 closing signal SIGTERM
+[2024-12-10 02:40:30,202] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2763437 closing signal SIGTERM
+[2024-12-10 02:40:30,201] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2841507 closing signal SIGTERM
+[2024-12-10 02:40:30,202] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2576491 closing signal SIGTERM
+[2024-12-10 02:40:30,201] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2841508 closing signal SIGTERM
+[2024-12-10 02:40:30,202] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2621466 closing signal SIGTERM
+[2024-12-10 02:40:30,197] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 104607 closing signal SIGTERM
+[2024-12-10 02:40:30,197] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 104608 closing signal SIGTERM
+[2024-12-10 02:40:30,197] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 104609 closing signal SIGTERM
+[2024-12-10 02:40:30,197] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 104610 closing signal SIGTERM
+srun: Job step aborted: Waiting up to 32 seconds for job step to finish.
+[2024-12-10 02:40:30,202] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2576492 closing signal SIGTERM
+[2024-12-10 02:40:30,203] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-10 02:40:30,202] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2142900 closing signal SIGTERM
+[2024-12-10 02:40:30,202] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2142901 closing signal SIGTERM
+[2024-12-10 02:40:30,203] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3835080 closing signal SIGTERM
+[2024-12-10 02:40:30,204] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3835081 closing signal SIGTERM
+[2024-12-10 02:40:30,204] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3835082 closing signal SIGTERM
+[2024-12-10 02:40:30,204] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3835083 closing signal SIGTERM
+[2024-12-10 02:40:30,201] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2841509 closing signal SIGTERM
+[2024-12-10 02:40:30,202] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2841510 closing signal SIGTERM
+[2024-12-10 02:40:30,202] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2841511 closing signal SIGTERM
+[2024-12-10 02:40:30,204] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2841512 closing signal SIGTERM
+[2024-12-10 02:40:30,206] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3835084 closing signal SIGTERM
+[2024-12-10 02:40:30,206] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3835085 closing signal SIGTERM
+[2024-12-10 02:40:30,206] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3835086 closing signal SIGTERM
+[2024-12-10 02:40:30,208] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3835087 closing signal SIGTERM
+slurmstepd: error: *** STEP 13265127.0 ON ip-26-0-160-225 CANCELLED AT 2024-12-10T02:40:30 ***
+[2024-12-10 02:40:30,201] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-10 02:40:30,201] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 83327 closing signal SIGTERM
+[2024-12-10 02:40:30,201] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 83328 closing signal SIGTERM
+[2024-12-10 02:40:30,201] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 83329 closing signal SIGTERM
+[2024-12-10 02:40:30,201] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 83330 closing signal SIGTERM
+[2024-12-10 02:40:30,202] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 83331 closing signal SIGTERM
+[2024-12-10 02:40:30,203] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 83332 closing signal SIGTERM
+[2024-12-10 02:40:30,203] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 83333 closing signal SIGTERM
+[2024-12-10 02:40:30,203] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 83334 closing signal SIGTERM
diff --git a/logs/13265423-bench_1.34G_dp2_tp4_pp1_acc1_mbs64_seq8192_zero1_tpmodeALL_vocab131k.out b/logs/13265423-bench_1.34G_dp2_tp4_pp1_acc1_mbs64_seq8192_zero1_tpmodeALL_vocab131k.out
new file mode 100644
index 0000000000000000000000000000000000000000..b6c70e1a81af13f82fb580e4e40a2862db2d3c6b
--- /dev/null
+++ b/logs/13265423-bench_1.34G_dp2_tp4_pp1_acc1_mbs64_seq8192_zero1_tpmodeALL_vocab131k.out
@@ -0,0 +1,814 @@
++ source /etc/profile.d/modules.sh
+++ . /usr/share/modules/init/bash
++++ unset _mlshdbg
++++ '[' 0 = 1 ']'
++++ unset _mlre _mlIFS
++++ '[' -n x ']'
++++ _mlIFS=' 	
+'
++++ IFS=' '
++++ '[' -n '' ']'
+++++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash autoinit
++++ _mlcode='module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+MODULEPATH_modshare=/etc/environment-modules/modules:1:/usr/share/modules/\$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1; export MODULEPATH_modshare;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+LOADEDMODULES=; export LOADEDMODULES;
+MODULEPATH=/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/\$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles; export MODULEPATH;
+test 0;'
++++ _mlret=0
++++ '[' -n x ']'
++++ IFS=' 	
+'
++++ unset _mlIFS
++++ unset _mlre _mlv _mlrv
++++ '[' 0 -eq 0 ']'
++++ eval 'module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+MODULEPATH_modshare=/etc/environment-modules/modules:1:/usr/share/modules/\$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1; export MODULEPATH_modshare;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+LOADEDMODULES=; export LOADEDMODULES;
+MODULEPATH=/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/\$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles; export MODULEPATH;
+test 0;'
+++++ MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl
+++++ export MODULES_CMD
+++++ ENV=/usr/share/modules/init/profile.sh
+++++ export ENV
+++++ MODULEPATH_modshare='/etc/environment-modules/modules:1:/usr/share/modules/$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1'
+++++ export MODULEPATH_modshare
+++++ BASH_ENV=/usr/share/modules/init/bash
+++++ export BASH_ENV
+++++ MODULESHOME=/usr/share/modules
+++++ export MODULESHOME
+++++ LOADEDMODULES=
+++++ export LOADEDMODULES
+++++ MODULEPATH='/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles'
+++++ export MODULEPATH
+++++ test 0
++++ '[' 0 = 1 ']'
++++ '[' -t 2 ']'
++++ export -f module
++++ export -f switchml
++++ '[' 5 -ge 3 ']'
++++ [[ ehxB =~ i ]]
++++ [[ ! :/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin: =~ :/usr/bin: ]]
++++ '[' '!' -n '' ']'
++++ MANPATH=:
++++ export MANPATH
+++++ manpath
++++ [[ ! :/admin/home/nouamane/.local/share/man:/fsx/nouamane/miniconda/envs/2-1-cu121/man:/fsx/nouamane/miniconda/envs/2-1-cu121/share/man:/fsx/nouamane/miniconda/man:/fsx/nouamane/miniconda/share/man:/opt/amazon/openmpi/share/man:/opt/amazon/efa/share/man:/opt/slurm/share/man:/usr/local/man:/usr/local/share/man:/usr/share/man:/admin/home/nouamane/.fzf/man:: =~ :/usr/share/man: ]]
++++ unset _mlcode _mlret
++++ '[' -n '' ']'
++ module load cuda/12.1
++ unset _mlshdbg
++ '[' 0 = 1 ']'
++ unset _mlre _mlIFS
++ '[' -n x ']'
++ _mlIFS=' 	
+'
++ IFS=' '
++ '[' -n '' ']'
+++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash load cuda/12.1
++ eval 'CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include;' export 'CPATH;
+LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:;' export 'LD_LIBRARY_PATH;
+FI_EFA_FORK_SAFE=1;' export 'FI_EFA_FORK_SAFE;
+MANPATH=/usr/local/cuda-12.1/share/man::;' export 'MANPATH;
+LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64;' export 'LIBRARY_PATH;
+_LMFILES_=/usr/share/modules/modulefiles/cuda/12.1;' export '_LMFILES_;
+LOADEDMODULES=cuda/12.1;' export 'LOADEDMODULES;
+MPI_PATH=/opt/amazon/openmpi;' export 'MPI_PATH;
+NCCL_HOME_modshare=/opt/nccl/build:1;' export 'NCCL_HOME_modshare;
+NCCL_PROTO=simple;' export 'NCCL_PROTO;
+MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1;' export 'MANPATH_modshare;
+LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1;' export 'LIBRARY_PATH_modshare;
+NCCL_SOCKET_IFNAME=enp;' export 'NCCL_SOCKET_IFNAME;
+AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl;' export 'AWS_OFI_NCCL_HOME;
+NCCL_HOME=/opt/nccl/build;' export 'NCCL_HOME;
+FI_PROVIDER=efa;' export 'FI_PROVIDER;
+AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1;' export 'AWS_OFI_NCCL_HOME_modshare;
+CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1;' export 'CPATH_modshare;
+LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2::1:/opt/nccl/build/lib:1:/opt/aws-ofi-nccl/lib:2:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/efa/lib:1;' export 'LD_LIBRARY_PATH_modshare;
+FI_EFA_ENABLE_SHM_TRANSFER=1;' export 'FI_EFA_ENABLE_SHM_TRANSFER;
+_LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1;' export '_LMFILES__modshare;
+LOADEDMODULES_modshare=cuda/12.1:1;' export 'LOADEDMODULES_modshare;
+MPI_PATH_modshare=/opt/amazon/openmpi:1;' export 'MPI_PATH_modshare;
+PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin;' export 'PATH;
+CUDA_HOME=/usr/local/cuda-12.1;' export 'CUDA_HOME;
+PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/usr/local/cuda-12.1/include:1:/opt/amazon/efa/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/usr/sbin:1:/fsx/nouamane/miniconda/bin:1:/opt/slurm/bin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:2:/opt/amazon/openmpi/bin:1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1;' export 'PATH_modshare;
+test' '0;'
+++ CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include
+++ export CPATH
+++ LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:
+++ export LD_LIBRARY_PATH
+++ FI_EFA_FORK_SAFE=1
+++ export FI_EFA_FORK_SAFE
+++ MANPATH=/usr/local/cuda-12.1/share/man::
+++ export MANPATH
+++ LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64
+++ export LIBRARY_PATH
+++ _LMFILES_=/usr/share/modules/modulefiles/cuda/12.1
+++ export _LMFILES_
+++ LOADEDMODULES=cuda/12.1
+++ export LOADEDMODULES
+++ MPI_PATH=/opt/amazon/openmpi
+++ export MPI_PATH
+++ NCCL_HOME_modshare=/opt/nccl/build:1
+++ export NCCL_HOME_modshare
+++ NCCL_PROTO=simple
+++ export NCCL_PROTO
+++ MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1
+++ export MANPATH_modshare
+++ LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1
+++ export LIBRARY_PATH_modshare
+++ NCCL_SOCKET_IFNAME=enp
+++ export NCCL_SOCKET_IFNAME
+++ AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl
+++ export AWS_OFI_NCCL_HOME
+++ NCCL_HOME=/opt/nccl/build
+++ export NCCL_HOME
+++ FI_PROVIDER=efa
+++ export FI_PROVIDER
+++ AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1
+++ export AWS_OFI_NCCL_HOME_modshare
+++ CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1
+++ export CPATH_modshare
+++ LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2::1:/opt/nccl/build/lib:1:/opt/aws-ofi-nccl/lib:2:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/efa/lib:1
+++ export LD_LIBRARY_PATH_modshare
+++ FI_EFA_ENABLE_SHM_TRANSFER=1
+++ export FI_EFA_ENABLE_SHM_TRANSFER
+++ _LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1
+++ export _LMFILES__modshare
+++ LOADEDMODULES_modshare=cuda/12.1:1
+++ export LOADEDMODULES_modshare
+++ MPI_PATH_modshare=/opt/amazon/openmpi:1
+++ export MPI_PATH_modshare
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ export PATH
+++ CUDA_HOME=/usr/local/cuda-12.1
+++ export CUDA_HOME
+++ PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/usr/local/cuda-12.1/include:1:/opt/amazon/efa/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/usr/sbin:1:/fsx/nouamane/miniconda/bin:1:/opt/slurm/bin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:2:/opt/amazon/openmpi/bin:1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1
+++ export PATH_modshare
+++ test 0
++ _mlstatus=0
++ '[' -n x ']'
++ IFS=' 	
+'
++ unset _mlre _mlv _mlrv _mlIFS
++ '[' -n '' ']'
++ unset _mlshdbg
++ return 0
++ source /fsx/nouamane/miniconda/bin/activate
+++ _CONDA_ROOT=/fsx/nouamane/miniconda
+++ . /fsx/nouamane/miniconda/etc/profile.d/conda.sh
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ '[' -z x ']'
+++ conda activate
+++ local cmd=activate
+++ case "$cmd" in
+++ __conda_activate activate
+++ '[' -n '' ']'
+++ local ask_conda
++++ PS1=
++++ __conda_exe shell.posix activate
++++ /fsx/nouamane/miniconda/bin/conda shell.posix activate
+++ ask_conda='. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_2='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\'''
+++ eval '. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_2='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\'''
++++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh
+++++ test -n ''
+++++ unset XML_CATALOG_FILES
+++++ unset xml_catalog_files_libxml2
++++ PS1='(base) '
++++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++++ export CONDA_PREFIX=/fsx/nouamane/miniconda
++++ CONDA_PREFIX=/fsx/nouamane/miniconda
++++ export CONDA_SHLVL=3
++++ CONDA_SHLVL=3
++++ export CONDA_DEFAULT_ENV=base
++++ CONDA_DEFAULT_ENV=base
++++ export 'CONDA_PROMPT_MODIFIER=(base) '
++++ CONDA_PROMPT_MODIFIER='(base) '
++++ export CONDA_PREFIX_2=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ CONDA_PREFIX_2=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ __conda_hashr
+++ '[' -n '' ']'
+++ '[' -n '' ']'
+++ hash -r
++ conda activate 2-1-cu121
++ local cmd=activate
++ case "$cmd" in
++ __conda_activate activate 2-1-cu121
++ '[' -n '' ']'
++ local ask_conda
+++ PS1='(base) '
+++ __conda_exe shell.posix activate 2-1-cu121
+++ /fsx/nouamane/miniconda/bin/conda shell.posix activate 2-1-cu121
++ ask_conda='PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''4'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_3='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
++ eval 'PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''4'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_3='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
+++ PS1='(2-1-cu121) '
+++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ export CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ export CONDA_SHLVL=4
+++ CONDA_SHLVL=4
+++ export CONDA_DEFAULT_ENV=2-1-cu121
+++ CONDA_DEFAULT_ENV=2-1-cu121
+++ export 'CONDA_PROMPT_MODIFIER=(2-1-cu121) '
+++ CONDA_PROMPT_MODIFIER='(2-1-cu121) '
+++ export CONDA_PREFIX_3=/fsx/nouamane/miniconda
+++ CONDA_PREFIX_3=/fsx/nouamane/miniconda
+++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ export _CE_M=
+++ _CE_M=
+++ export _CE_CONDA=
+++ _CE_CONDA=
+++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh
++++ test -n ''
++++ xml_catalog_files_libxml2=
++++ XML_CATALOG_FILES=
++++ conda_catalog_files=
++++ ifs_libxml2=' 	
+'
++++ IFS=' '
++++ rem=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ for pre in ${rem}
++++ test '' = /fsx/nouamane/miniconda/envs/2-1-cu121
++++ conda_catalog_files=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ rem=
++++ IFS=' 	
+'
++++ conda_catalog_files='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ export 'XML_CATALOG_FILES=file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ XML_CATALOG_FILES='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ unset conda_catalog_files ifs_libxml2 rem
++ __conda_hashr
++ '[' -n '' ']'
++ '[' -n '' ']'
++ hash -r
++ export PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++ PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ scontrol show hostnames ip-26-0-173-121
++ export NODELIST=ip-26-0-173-121
++ NODELIST=ip-26-0-173-121
+++ scontrol show hostnames ip-26-0-173-121
+++ head -n1
++ export MASTER_NODE=ip-26-0-173-121
++ MASTER_NODE=ip-26-0-173-121
++ export MASTER_PORT=12356
++ MASTER_PORT=12356
++ export NNODES=1
++ NNODES=1
++ export GPUS_PER_NODE=8
++ GPUS_PER_NODE=8
++ export WORLD_SIZE=8
++ WORLD_SIZE=8
++ export CUDA_DEVICE_MAX_CONNECTIONS=1
++ CUDA_DEVICE_MAX_CONNECTIONS=1
++ export NANOTRON_BENCHMARK=1
++ NANOTRON_BENCHMARK=1
++ echo 'Master node: ip-26-0-173-121'
+Master node: ip-26-0-173-121
++ echo 'All nodes: ip-26-0-173-121'
+All nodes: ip-26-0-173-121
++ echo 'World size: 8'
+World size: 8
++ srun torchrun --nnodes=1 --nproc_per_node=8 --rdzv_id=13265423 --rdzv_backend=c10d --rdzv_endpoint=ip-26-0-173-121:12356 run_train.py --config-file benchmark/configs/config_1.34G_dp2_tp4_pp1_acc1_mbs64_seq8192_zero1_tpmodeALL_vocab131k.yaml
+[2024-12-10 01:22:02,430] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-10 01:22:02,431] torch.distributed.run: [WARNING] 
+[2024-12-10 01:22:02,431] torch.distributed.run: [WARNING] *****************************************
+[2024-12-10 01:22:02,431] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-10 01:22:02,431] torch.distributed.run: [WARNING] *****************************************
+12/10/2024 01:22:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-173-121]: Config:
+12/10/2024 01:22:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-173-121]: Config(general=GeneralArgs(project='debug',
+12/10/2024 01:22:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-173-121]:                            run='1.34G_dp2_tp4_pp1_acc1_mbs64_seq8192_zero1_tpmodeALL_vocab131k',
+12/10/2024 01:22:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-173-121]:                            seed=42,
+12/10/2024 01:22:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-173-121]:                            step=None,
+12/10/2024 01:22:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-173-121]:                            consumed_train_samples=None,
+12/10/2024 01:22:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-173-121]:                            benchmark_csv_path=PosixPath('benchmark/results/bench_final.csv'),
+12/10/2024 01:22:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-173-121]:                            ignore_sanity_checks=True),
+12/10/2024 01:22:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-173-121]:        parallelism=ParallelismArgs(dp=2,
+12/10/2024 01:22:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-173-121]:                                    pp=1,
+12/10/2024 01:22:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-173-121]:                                    tp=4,
+12/10/2024 01:22:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-173-121]:                                    pp_engine=<nanotron.parallel.pipeline_parallel.engine.OneForwardOneBackwardPipelineEngine object at 0x7f8589000c70>,
+12/10/2024 01:22:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-173-121]:                                    tp_mode=<TensorParallelLinearMode.ALL_REDUCE: 1>,
+12/10/2024 01:22:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-173-121]:                                    tp_linear_async_communication=False,
+12/10/2024 01:22:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-173-121]:                                    recompute_layer=False,
+12/10/2024 01:22:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-173-121]:                                    tp_recompute_allgather=True,
+12/10/2024 01:22:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-173-121]:                                    expert_parallel_size=1),
+12/10/2024 01:22:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-173-121]:        model=ModelArgs(model_config=LlamaConfig(bos_token_id=0,
+12/10/2024 01:22:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-173-121]:                                                 eos_token_id=0,
+12/10/2024 01:22:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-173-121]:                                                 hidden_act='silu',
+12/10/2024 01:22:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-173-121]:                                                 hidden_size=2048,
+12/10/2024 01:22:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-173-121]:                                                 initializer_range=0.02,
+12/10/2024 01:22:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-173-121]:                                                 intermediate_size=8192,
+12/10/2024 01:22:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-173-121]:                                                 is_llama_config=True,
+12/10/2024 01:22:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-173-121]:                                                 max_position_embeddings=8192,
+12/10/2024 01:22:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-173-121]:                                                 num_attention_heads=32,
+12/10/2024 01:22:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-173-121]:                                                 num_hidden_layers=16,
+12/10/2024 01:22:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-173-121]:                                                 num_key_value_heads=8,
+12/10/2024 01:22:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-173-121]:                                                 pad_token_id=None,
+12/10/2024 01:22:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-173-121]:                                                 pretraining_tp=1,
+12/10/2024 01:22:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-173-121]:                                                 rms_norm_eps=1e-05,
+12/10/2024 01:22:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-173-121]:                                                 rope_scaling=None,
+12/10/2024 01:22:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-173-121]:                                                 rope_theta=10000.0,
+12/10/2024 01:22:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-173-121]:                                                 rope_interleaved=False,
+12/10/2024 01:22:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-173-121]:                                                 tie_word_embeddings=True,
+12/10/2024 01:22:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-173-121]:                                                 use_cache=True,
+12/10/2024 01:22:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-173-121]:                                                 vocab_size=131072),
+12/10/2024 01:22:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-173-121]:                        init_method=RandomInit(std=0.02),
+12/10/2024 01:22:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-173-121]:                        dtype=torch.bfloat16,
+12/10/2024 01:22:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-173-121]:                        make_vocab_size_divisible_by=1,
+12/10/2024 01:22:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-173-121]:                        ddp_bucket_cap_mb=25),
+12/10/2024 01:22:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-173-121]:        tokenizer=TokenizerArgs(tokenizer_name_or_path='robot-test/dummy-tokenizer-wordlevel',
+12/10/2024 01:22:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-173-121]:                                tokenizer_revision=None,
+12/10/2024 01:22:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-173-121]:                                tokenizer_max_length=None),
+12/10/2024 01:22:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-173-121]:        checkpoints=CheckpointsArgs(checkpoints_path=PosixPath('checkpoints'),
+12/10/2024 01:22:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-173-121]:                                    checkpoint_interval=10000,
+12/10/2024 01:22:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-173-121]:                                    save_initial_state=False,
+12/10/2024 01:22:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-173-121]:                                    save_final_state=False,
+12/10/2024 01:22:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-173-121]:                                    resume_checkpoint_path=None,
+12/10/2024 01:22:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-173-121]:                                    checkpoints_path_is_shared_file_system=False),
+12/10/2024 01:22:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-173-121]:        logging=LoggingArgs(log_level='info',
+12/10/2024 01:22:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-173-121]:                            log_level_replica='info',
+12/10/2024 01:22:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-173-121]:                            iteration_step_info_interval=1),
+12/10/2024 01:22:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-173-121]:        tokens=TokensArgs(sequence_length=8192,
+12/10/2024 01:22:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-173-121]:                          train_steps=100,
+12/10/2024 01:22:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-173-121]:                          micro_batch_size=64,
+12/10/2024 01:22:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-173-121]:                          batch_accumulation_per_replica=1,
+12/10/2024 01:22:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-173-121]:                          val_check_interval=100,
+12/10/2024 01:22:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-173-121]:                          limit_val_batches=0,
+12/10/2024 01:22:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-173-121]:                          limit_test_batches=0),
+12/10/2024 01:22:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-173-121]:        optimizer=OptimizerArgs(optimizer_factory=AdamWOptimizerArgs(adam_eps=1e-08,
+12/10/2024 01:22:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-173-121]:                                                                     adam_beta1=0.9,
+12/10/2024 01:22:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-173-121]:                                                                     adam_beta2=0.95,
+12/10/2024 01:22:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-173-121]:                                                                     torch_adam_is_fused=True,
+12/10/2024 01:22:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-173-121]:                                                                     name='adamW'),
+12/10/2024 01:22:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-173-121]:                                zero_stage=1,
+12/10/2024 01:22:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-173-121]:                                weight_decay=0.01,
+12/10/2024 01:22:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-173-121]:                                clip_grad=1.0,
+12/10/2024 01:22:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-173-121]:                                accumulate_grad_in_fp32=True,
+12/10/2024 01:22:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-173-121]:                                learning_rate_scheduler=LRSchedulerArgs(learning_rate=0.0003,
+12/10/2024 01:22:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-173-121]:                                                                        lr_warmup_steps=2,
+12/10/2024 01:22:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-173-121]:                                                                        lr_warmup_style='linear',
+12/10/2024 01:22:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-173-121]:                                                                        lr_decay_style='cosine',
+12/10/2024 01:22:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-173-121]:                                                                        lr_decay_steps=13,
+12/10/2024 01:22:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-173-121]:                                                                        lr_decay_starting_step=None,
+12/10/2024 01:22:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-173-121]:                                                                        min_decay_lr=1e-05)),
+12/10/2024 01:22:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-173-121]:        data_stages=[DatasetStageArgs(name='Stable Training Stage',
+12/10/2024 01:22:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-173-121]:                                      start_training_step=1,
+12/10/2024 01:22:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-173-121]:                                      data=DataArgs(dataset=None,
+12/10/2024 01:22:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-173-121]:                                                    seed=42,
+12/10/2024 01:22:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-173-121]:                                                    num_loading_workers=1))],
+12/10/2024 01:22:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-173-121]:        profiler=None,
+12/10/2024 01:22:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-173-121]:        lighteval=None,
+12/10/2024 01:22:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-173-121]:        s3_upload=None)
+12/10/2024 01:22:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-173-121]: Model Config:
+12/10/2024 01:22:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-173-121]: LlamaConfig(bos_token_id=0,
+12/10/2024 01:22:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-173-121]:             eos_token_id=0,
+12/10/2024 01:22:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-173-121]:             hidden_act='silu',
+12/10/2024 01:22:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-173-121]:             hidden_size=2048,
+12/10/2024 01:22:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-173-121]:             initializer_range=0.02,
+12/10/2024 01:22:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-173-121]:             intermediate_size=8192,
+12/10/2024 01:22:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-173-121]:             is_llama_config=True,
+12/10/2024 01:22:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-173-121]:             max_position_embeddings=8192,
+12/10/2024 01:22:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-173-121]:             num_attention_heads=32,
+12/10/2024 01:22:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-173-121]:             num_hidden_layers=16,
+12/10/2024 01:22:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-173-121]:             num_key_value_heads=8,
+12/10/2024 01:22:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-173-121]:             pad_token_id=None,
+12/10/2024 01:22:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-173-121]:             pretraining_tp=1,
+12/10/2024 01:22:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-173-121]:             rms_norm_eps=1e-05,
+12/10/2024 01:22:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-173-121]:             rope_scaling=None,
+12/10/2024 01:22:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-173-121]:             rope_theta=10000.0,
+12/10/2024 01:22:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-173-121]:             rope_interleaved=False,
+12/10/2024 01:22:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-173-121]:             tie_word_embeddings=True,
+12/10/2024 01:22:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-173-121]:             use_cache=True,
+12/10/2024 01:22:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-173-121]:             vocab_size=131072)
+12/10/2024 01:22:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-173-121]: Building model..
+12/10/2024 01:22:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-173-121]: Initialize RoPE Theta = 10000.0
+12/10/2024 01:22:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-173-121]: Setting PP block ranks...
+12/10/2024 01:22:42 [INFO|DP=0|PP=0|TP=0|ip-26-0-173-121]: Total number of parameters: 1.24G (2368.52MiB)
+12/10/2024 01:22:42 [INFO|DP=0|PP=0|TP=1|ip-26-0-173-121]: Local number of parameters: 310M (592.13MiB)
+12/10/2024 01:22:42 [INFO|DP=0|PP=0|TP=0|ip-26-0-173-121]: Local number of parameters: 310M (592.13MiB)
+12/10/2024 01:22:42 [INFO|DP=0|PP=0|TP=2|ip-26-0-173-121]: Local number of parameters: 310M (592.13MiB)
+12/10/2024 01:22:42 [INFO|DP=0|PP=0|TP=3|ip-26-0-173-121]: Local number of parameters: 310M (592.13MiB)
+12/10/2024 01:22:42 [INFO|DP=0|PP=0|TP=2|ip-26-0-173-121]: [After model building] Memory usage: 592.15MiB. Peak allocated: 5632.00MiB Peak reserved: 15874.00MiB
+12/10/2024 01:22:42 [INFO|DP=0|PP=0|TP=1|ip-26-0-173-121]: [After model building] Memory usage: 592.15MiB. Peak allocated: 5632.00MiB Peak reserved: 15874.00MiB
+12/10/2024 01:22:42 [INFO|DP=0|PP=0|TP=0|ip-26-0-173-121]: [After model building] Memory usage: 592.15MiB. Peak allocated: 5632.00MiB Peak reserved: 17922.00MiB
+12/10/2024 01:22:42 [INFO|DP=0|PP=0|TP=3|ip-26-0-173-121]: [After model building] Memory usage: 592.15MiB. Peak allocated: 5632.00MiB Peak reserved: 15874.00MiB
+12/10/2024 01:22:42 [INFO|DP=0|PP=0|TP=0|ip-26-0-173-121]: No checkpoint path provided.
+12/10/2024 01:22:42 [INFO|DP=0|PP=0|TP=0|ip-26-0-173-121]: Parametrizing model parameters using StandardParametrizator
+12/10/2024 01:22:44 [INFO|DP=0|PP=0|TP=0|ip-26-0-173-121]: [Optimizer Building] Using LearningRateForSP as learning rate
+12/10/2024 01:22:44 [INFO|DP=0|PP=0|TP=0|ip-26-0-173-121]: [ZeRO sharding] Size of optimizer params per rank:
+12/10/2024 01:22:44 [INFO|DP=0|PP=0|TP=0|ip-26-0-173-121]: [ZeRO sharding] DP Rank 0 has 155M out of 310M (50.00%) params' optimizer states
+12/10/2024 01:22:44 [INFO|DP=0|PP=0|TP=0|ip-26-0-173-121]: [ZeRO sharding] DP Rank 1 has 155M out of 310M (50.00%) params' optimizer states
+12/10/2024 01:22:44 [INFO|DP=0|PP=0|TP=0|ip-26-0-173-121]: [Training Plan] Stage Stable Training Stage has 99 remaining training steps and has consumed 0 samples
+12/10/2024 01:22:44 [INFO|DP=0|PP=0|TP=0|ip-26-0-173-121]: Using dummy data generator
+12/10/2024 01:22:44 [INFO|DP=0|PP=0|TP=0|ip-26-0-173-121]: [Training Plan] There are 1 training stages 
+12/10/2024 01:22:44 [INFO|DP=0|PP=0|TP=0|ip-26-0-173-121]: [Stage Stable Training Stage] start from step 1 
+12/10/2024 01:22:44 [INFO|DP=0|PP=0|TP=0|ip-26-0-173-121]: 
+12/10/2024 01:22:44 [INFO|DP=0|PP=0|TP=0|ip-26-0-173-121]: [Start training] datetime: 2024-12-10 01:22:44.946086 | mbs: 64 | grad_accum: 1 | global_batch_size: 128 | sequence_length: 8192 | train_steps: 100 | start_iteration_step: 0 | consumed_train_samples: 0
+wandb: WARNING Path /fsx/nouamane/.cache/wandb/wandb/ wasn't writable, using system temp directory.
+wandb: WARNING Path /fsx/nouamane/.cache/wandb/wandb/ wasn't writable, using system temp directory
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 244, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    merged_states = self.gate_up_proj(hidden_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 89, in forward
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    return column_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 445, in column_linear
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    return F.linear(input, weight, bias)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 4.00 GiB. GPU 6 has a total capacty of 79.33 GiB of which 1005.94 MiB is free. Including non-PyTorch memory, this process has 78.34 GiB memory in use. Of the allocated memory 67.44 GiB is allocated by PyTorch, and 487.95 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 244, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    merged_states = self.gate_up_proj(hidden_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 89, in forward
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return column_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 445, in column_linear
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    return F.linear(input, weight, bias)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 4.00 GiB. GPU 5 has a total capacty of 79.33 GiB of which 1005.94 MiB is free. Including non-PyTorch memory, this process has 78.34 GiB memory in use. Of the allocated memory 67.44 GiB is allocated by PyTorch, and 487.95 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return forward_call(*args, **kwargs)
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 244, in forward
+    merged_states = self.gate_up_proj(hidden_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 89, in forward
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    return column_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 445, in column_linear
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return F.linear(input, weight, bias)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 4.00 GiB. GPU 7 has a total capacty of 79.33 GiB of which 1.92 GiB is free. Including non-PyTorch memory, this process has 77.40 GiB memory in use. Of the allocated memory 67.44 GiB is allocated by PyTorch, and 487.95 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 244, in forward
+    merged_states = self.gate_up_proj(hidden_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 89, in forward
+    return column_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 445, in column_linear
+    return F.linear(input, weight, bias)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 4.00 GiB. GPU 4 has a total capacty of 79.33 GiB of which 1.03 GiB is free. Including non-PyTorch memory, this process has 78.29 GiB memory in use. Of the allocated memory 67.44 GiB is allocated by PyTorch, and 487.95 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+[2024-12-10 01:22:47,648] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 823385 closing signal SIGTERM
+[2024-12-10 01:22:47,648] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 823386 closing signal SIGTERM
+[2024-12-10 01:22:47,648] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 823387 closing signal SIGTERM
+[2024-12-10 01:22:47,649] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 823388 closing signal SIGTERM
+[2024-12-10 01:22:47,649] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 823389 closing signal SIGTERM
+[2024-12-10 01:22:47,650] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 823391 closing signal SIGTERM
+[2024-12-10 01:22:47,650] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 823392 closing signal SIGTERM
+[2024-12-10 01:22:48,893] torch.distributed.elastic.multiprocessing.api: [ERROR] failed (exitcode: 1) local_rank: 5 (pid: 823390) of binary: /fsx/nouamane/miniconda/envs/2-1-cu121/bin/python
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 33, in <module>
+    sys.exit(load_entry_point('torch==2.1.1', 'console_scripts', 'torchrun')())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 264, in launch_agent
+    raise ChildFailedError(
+torch.distributed.elastic.multiprocessing.errors.ChildFailedError: 
+============================================================
+run_train.py FAILED
+------------------------------------------------------------
+Failures:
+  <NO_OTHER_FAILURES>
+------------------------------------------------------------
+Root Cause (first observed failure):
+[0]:
+  time      : 2024-12-10_01:22:47
+  host      : ip-26-0-173-121.ec2.internal
+  rank      : 5 (local_rank: 5)
+  exitcode  : 1 (pid: 823390)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+============================================================
+srun: error: ip-26-0-173-121: task 0: Exited with exit code 1
diff --git a/logs/13265633-bench_1.14G_dp4_tp4_pp1_acc4_mbs8_seq32768_zero1_tpmodeALL_vocab32k.out b/logs/13265633-bench_1.14G_dp4_tp4_pp1_acc4_mbs8_seq32768_zero1_tpmodeALL_vocab32k.out
new file mode 100644
index 0000000000000000000000000000000000000000..a9832ef709e04b5461d3a441407956905690f977
--- /dev/null
+++ b/logs/13265633-bench_1.14G_dp4_tp4_pp1_acc4_mbs8_seq32768_zero1_tpmodeALL_vocab32k.out
@@ -0,0 +1,1175 @@
++ source /etc/profile.d/modules.sh
+++ . /usr/share/modules/init/bash
++++ unset _mlshdbg
++++ '[' 0 = 1 ']'
++++ unset _mlre _mlIFS
++++ '[' -n x ']'
++++ _mlIFS=' 	
+'
++++ IFS=' '
++++ '[' -n '' ']'
+++++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash autoinit
++++ _mlcode='module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+MODULEPATH_modshare=/etc/environment-modules/modules:1:/usr/share/modules/\$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1; export MODULEPATH_modshare;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+LOADEDMODULES=; export LOADEDMODULES;
+MODULEPATH=/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/\$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles; export MODULEPATH;
+test 0;'
++++ _mlret=0
++++ '[' -n x ']'
++++ IFS=' 	
+'
++++ unset _mlIFS
++++ unset _mlre _mlv _mlrv
++++ '[' 0 -eq 0 ']'
++++ eval 'module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+MODULEPATH_modshare=/etc/environment-modules/modules:1:/usr/share/modules/\$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1; export MODULEPATH_modshare;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+LOADEDMODULES=; export LOADEDMODULES;
+MODULEPATH=/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/\$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles; export MODULEPATH;
+test 0;'
+++++ MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl
+++++ export MODULES_CMD
+++++ ENV=/usr/share/modules/init/profile.sh
+++++ export ENV
+++++ MODULEPATH_modshare='/etc/environment-modules/modules:1:/usr/share/modules/$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1'
+++++ export MODULEPATH_modshare
+++++ BASH_ENV=/usr/share/modules/init/bash
+++++ export BASH_ENV
+++++ MODULESHOME=/usr/share/modules
+++++ export MODULESHOME
+++++ LOADEDMODULES=
+++++ export LOADEDMODULES
+++++ MODULEPATH='/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles'
+++++ export MODULEPATH
+++++ test 0
++++ '[' 0 = 1 ']'
++++ '[' -t 2 ']'
++++ export -f module
++++ export -f switchml
++++ '[' 5 -ge 3 ']'
++++ [[ ehxB =~ i ]]
++++ [[ ! :/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin: =~ :/usr/bin: ]]
++++ '[' '!' -n '' ']'
++++ MANPATH=:
++++ export MANPATH
+++++ manpath
++++ [[ ! :/admin/home/nouamane/.local/share/man:/fsx/nouamane/miniconda/envs/2-1-cu121/man:/fsx/nouamane/miniconda/envs/2-1-cu121/share/man:/fsx/nouamane/miniconda/man:/fsx/nouamane/miniconda/share/man:/opt/amazon/openmpi/share/man:/opt/amazon/efa/share/man:/opt/slurm/share/man:/usr/local/man:/usr/local/share/man:/usr/share/man:/admin/home/nouamane/.fzf/man:: =~ :/usr/share/man: ]]
++++ unset _mlcode _mlret
++++ '[' -n '' ']'
++ module load cuda/12.1
++ unset _mlshdbg
++ '[' 0 = 1 ']'
++ unset _mlre _mlIFS
++ '[' -n x ']'
++ _mlIFS=' 	
+'
++ IFS=' '
++ '[' -n '' ']'
+++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash load cuda/12.1
++ eval 'CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include;' export 'CPATH;
+LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:;' export 'LD_LIBRARY_PATH;
+FI_EFA_FORK_SAFE=1;' export 'FI_EFA_FORK_SAFE;
+MANPATH=/usr/local/cuda-12.1/share/man::;' export 'MANPATH;
+LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64;' export 'LIBRARY_PATH;
+_LMFILES_=/usr/share/modules/modulefiles/cuda/12.1;' export '_LMFILES_;
+LOADEDMODULES=cuda/12.1;' export 'LOADEDMODULES;
+MPI_PATH=/opt/amazon/openmpi;' export 'MPI_PATH;
+NCCL_HOME_modshare=/opt/nccl/build:1;' export 'NCCL_HOME_modshare;
+NCCL_PROTO=simple;' export 'NCCL_PROTO;
+MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1;' export 'MANPATH_modshare;
+LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1;' export 'LIBRARY_PATH_modshare;
+NCCL_SOCKET_IFNAME=enp;' export 'NCCL_SOCKET_IFNAME;
+AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl;' export 'AWS_OFI_NCCL_HOME;
+NCCL_HOME=/opt/nccl/build;' export 'NCCL_HOME;
+FI_PROVIDER=efa;' export 'FI_PROVIDER;
+AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1;' export 'AWS_OFI_NCCL_HOME_modshare;
+CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1;' export 'CPATH_modshare;
+LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2::1:/opt/nccl/build/lib:1:/opt/aws-ofi-nccl/lib:2:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/efa/lib:1;' export 'LD_LIBRARY_PATH_modshare;
+FI_EFA_ENABLE_SHM_TRANSFER=1;' export 'FI_EFA_ENABLE_SHM_TRANSFER;
+_LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1;' export '_LMFILES__modshare;
+LOADEDMODULES_modshare=cuda/12.1:1;' export 'LOADEDMODULES_modshare;
+MPI_PATH_modshare=/opt/amazon/openmpi:1;' export 'MPI_PATH_modshare;
+PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin;' export 'PATH;
+CUDA_HOME=/usr/local/cuda-12.1;' export 'CUDA_HOME;
+PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/usr/local/cuda-12.1/include:1:/opt/amazon/efa/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/usr/sbin:1:/fsx/nouamane/miniconda/bin:1:/opt/slurm/bin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:2:/opt/amazon/openmpi/bin:1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1;' export 'PATH_modshare;
+test' '0;'
+++ CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include
+++ export CPATH
+++ LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:
+++ export LD_LIBRARY_PATH
+++ FI_EFA_FORK_SAFE=1
+++ export FI_EFA_FORK_SAFE
+++ MANPATH=/usr/local/cuda-12.1/share/man::
+++ export MANPATH
+++ LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64
+++ export LIBRARY_PATH
+++ _LMFILES_=/usr/share/modules/modulefiles/cuda/12.1
+++ export _LMFILES_
+++ LOADEDMODULES=cuda/12.1
+++ export LOADEDMODULES
+++ MPI_PATH=/opt/amazon/openmpi
+++ export MPI_PATH
+++ NCCL_HOME_modshare=/opt/nccl/build:1
+++ export NCCL_HOME_modshare
+++ NCCL_PROTO=simple
+++ export NCCL_PROTO
+++ MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1
+++ export MANPATH_modshare
+++ LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1
+++ export LIBRARY_PATH_modshare
+++ NCCL_SOCKET_IFNAME=enp
+++ export NCCL_SOCKET_IFNAME
+++ AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl
+++ export AWS_OFI_NCCL_HOME
+++ NCCL_HOME=/opt/nccl/build
+++ export NCCL_HOME
+++ FI_PROVIDER=efa
+++ export FI_PROVIDER
+++ AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1
+++ export AWS_OFI_NCCL_HOME_modshare
+++ CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1
+++ export CPATH_modshare
+++ LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2::1:/opt/nccl/build/lib:1:/opt/aws-ofi-nccl/lib:2:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/efa/lib:1
+++ export LD_LIBRARY_PATH_modshare
+++ FI_EFA_ENABLE_SHM_TRANSFER=1
+++ export FI_EFA_ENABLE_SHM_TRANSFER
+++ _LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1
+++ export _LMFILES__modshare
+++ LOADEDMODULES_modshare=cuda/12.1:1
+++ export LOADEDMODULES_modshare
+++ MPI_PATH_modshare=/opt/amazon/openmpi:1
+++ export MPI_PATH_modshare
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ export PATH
+++ CUDA_HOME=/usr/local/cuda-12.1
+++ export CUDA_HOME
+++ PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/usr/local/cuda-12.1/include:1:/opt/amazon/efa/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/usr/sbin:1:/fsx/nouamane/miniconda/bin:1:/opt/slurm/bin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:2:/opt/amazon/openmpi/bin:1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1
+++ export PATH_modshare
+++ test 0
++ _mlstatus=0
++ '[' -n x ']'
++ IFS=' 	
+'
++ unset _mlre _mlv _mlrv _mlIFS
++ '[' -n '' ']'
++ unset _mlshdbg
++ return 0
++ source /fsx/nouamane/miniconda/bin/activate
+++ _CONDA_ROOT=/fsx/nouamane/miniconda
+++ . /fsx/nouamane/miniconda/etc/profile.d/conda.sh
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ '[' -z x ']'
+++ conda activate
+++ local cmd=activate
+++ case "$cmd" in
+++ __conda_activate activate
+++ '[' -n '' ']'
+++ local ask_conda
++++ PS1=
++++ __conda_exe shell.posix activate
++++ /fsx/nouamane/miniconda/bin/conda shell.posix activate
+++ ask_conda='. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_2='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\'''
+++ eval '. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_2='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\'''
++++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh
+++++ test -n ''
+++++ unset XML_CATALOG_FILES
+++++ unset xml_catalog_files_libxml2
++++ PS1='(base) '
++++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++++ export CONDA_PREFIX=/fsx/nouamane/miniconda
++++ CONDA_PREFIX=/fsx/nouamane/miniconda
++++ export CONDA_SHLVL=3
++++ CONDA_SHLVL=3
++++ export CONDA_DEFAULT_ENV=base
++++ CONDA_DEFAULT_ENV=base
++++ export 'CONDA_PROMPT_MODIFIER=(base) '
++++ CONDA_PROMPT_MODIFIER='(base) '
++++ export CONDA_PREFIX_2=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ CONDA_PREFIX_2=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ __conda_hashr
+++ '[' -n '' ']'
+++ '[' -n '' ']'
+++ hash -r
++ conda activate 2-1-cu121
++ local cmd=activate
++ case "$cmd" in
++ __conda_activate activate 2-1-cu121
++ '[' -n '' ']'
++ local ask_conda
+++ PS1='(base) '
+++ __conda_exe shell.posix activate 2-1-cu121
+++ /fsx/nouamane/miniconda/bin/conda shell.posix activate 2-1-cu121
++ ask_conda='PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''4'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_3='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
++ eval 'PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''4'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_3='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
+++ PS1='(2-1-cu121) '
+++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ export CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ export CONDA_SHLVL=4
+++ CONDA_SHLVL=4
+++ export CONDA_DEFAULT_ENV=2-1-cu121
+++ CONDA_DEFAULT_ENV=2-1-cu121
+++ export 'CONDA_PROMPT_MODIFIER=(2-1-cu121) '
+++ CONDA_PROMPT_MODIFIER='(2-1-cu121) '
+++ export CONDA_PREFIX_3=/fsx/nouamane/miniconda
+++ CONDA_PREFIX_3=/fsx/nouamane/miniconda
+++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ export _CE_M=
+++ _CE_M=
+++ export _CE_CONDA=
+++ _CE_CONDA=
+++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh
++++ test -n ''
++++ xml_catalog_files_libxml2=
++++ XML_CATALOG_FILES=
++++ conda_catalog_files=
++++ ifs_libxml2=' 	
+'
++++ IFS=' '
++++ rem=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ for pre in ${rem}
++++ test '' = /fsx/nouamane/miniconda/envs/2-1-cu121
++++ conda_catalog_files=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ rem=
++++ IFS=' 	
+'
++++ conda_catalog_files='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ export 'XML_CATALOG_FILES=file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ XML_CATALOG_FILES='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ unset conda_catalog_files ifs_libxml2 rem
++ __conda_hashr
++ '[' -n '' ']'
++ '[' -n '' ']'
++ hash -r
++ export PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++ PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ scontrol show hostnames ip-26-0-163-58,ip-26-0-164-45
++ export 'NODELIST=ip-26-0-163-58
+ip-26-0-164-45'
++ NODELIST='ip-26-0-163-58
+ip-26-0-164-45'
+++ scontrol show hostnames ip-26-0-163-58,ip-26-0-164-45
+++ head -n1
++ export MASTER_NODE=ip-26-0-163-58
++ MASTER_NODE=ip-26-0-163-58
++ export MASTER_PORT=12356
++ MASTER_PORT=12356
++ export NNODES=2
++ NNODES=2
++ export GPUS_PER_NODE=8
++ GPUS_PER_NODE=8
++ export WORLD_SIZE=16
++ WORLD_SIZE=16
++ export CUDA_DEVICE_MAX_CONNECTIONS=1
++ CUDA_DEVICE_MAX_CONNECTIONS=1
++ export NANOTRON_BENCHMARK=1
++ NANOTRON_BENCHMARK=1
++ echo 'Master node: ip-26-0-163-58'
+Master node: ip-26-0-163-58
++ echo 'All nodes: ip-26-0-163-58
+ip-26-0-164-45'
+All nodes: ip-26-0-163-58
+ip-26-0-164-45
++ echo 'World size: 16'
+World size: 16
++ srun torchrun --nnodes=2 --nproc_per_node=8 --rdzv_id=13265633 --rdzv_backend=c10d --rdzv_endpoint=ip-26-0-163-58:12356 run_train.py --config-file benchmark/configs/config_1.14G_dp4_tp4_pp1_acc4_mbs8_seq32768_zero1_tpmodeALL_vocab32k.yaml
+[2024-12-10 03:15:50,086] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-10 03:15:50,086] torch.distributed.run: [WARNING] 
+[2024-12-10 03:15:50,086] torch.distributed.run: [WARNING] *****************************************
+[2024-12-10 03:15:50,086] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-10 03:15:50,086] torch.distributed.run: [WARNING] *****************************************
+[2024-12-10 03:15:50,143] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-10 03:15:50,143] torch.distributed.run: [WARNING] 
+[2024-12-10 03:15:50,143] torch.distributed.run: [WARNING] *****************************************
+[2024-12-10 03:15:50,143] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-10 03:15:50,143] torch.distributed.run: [WARNING] *****************************************
+12/10/2024 03:16:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-58]: Config:
+12/10/2024 03:16:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-58]: Config(general=GeneralArgs(project='debug',
+12/10/2024 03:16:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-58]:                            run='1.14G_dp4_tp4_pp1_acc4_mbs8_seq32768_zero1_tpmodeALL_vocab32k',
+12/10/2024 03:16:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-58]:                            seed=42,
+12/10/2024 03:16:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-58]:                            step=None,
+12/10/2024 03:16:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-58]:                            consumed_train_samples=None,
+12/10/2024 03:16:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-58]:                            benchmark_csv_path=PosixPath('benchmark/results/bench_final.csv'),
+12/10/2024 03:16:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-58]:                            ignore_sanity_checks=True),
+12/10/2024 03:16:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-58]:        parallelism=ParallelismArgs(dp=4,
+12/10/2024 03:16:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-58]:                                    pp=1,
+12/10/2024 03:16:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-58]:                                    tp=4,
+12/10/2024 03:16:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-58]:                                    pp_engine=<nanotron.parallel.pipeline_parallel.engine.OneForwardOneBackwardPipelineEngine object at 0x7f15d2cd0d00>,
+12/10/2024 03:16:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-58]:                                    tp_mode=<TensorParallelLinearMode.ALL_REDUCE: 1>,
+12/10/2024 03:16:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-58]:                                    tp_linear_async_communication=False,
+12/10/2024 03:16:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-58]:                                    recompute_layer=False,
+12/10/2024 03:16:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-58]:                                    tp_recompute_allgather=True,
+12/10/2024 03:16:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-58]:                                    expert_parallel_size=1),
+12/10/2024 03:16:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-58]:        model=ModelArgs(model_config=LlamaConfig(bos_token_id=0,
+12/10/2024 03:16:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-58]:                                                 eos_token_id=0,
+12/10/2024 03:16:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-58]:                                                 hidden_act='silu',
+12/10/2024 03:16:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-58]:                                                 hidden_size=2048,
+12/10/2024 03:16:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-58]:                                                 initializer_range=0.02,
+12/10/2024 03:16:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-58]:                                                 intermediate_size=8192,
+12/10/2024 03:16:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-58]:                                                 is_llama_config=True,
+12/10/2024 03:16:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-58]:                                                 max_position_embeddings=32768,
+12/10/2024 03:16:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-58]:                                                 num_attention_heads=32,
+12/10/2024 03:16:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-58]:                                                 num_hidden_layers=16,
+12/10/2024 03:16:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-58]:                                                 num_key_value_heads=8,
+12/10/2024 03:16:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-58]:                                                 pad_token_id=None,
+12/10/2024 03:16:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-58]:                                                 pretraining_tp=1,
+12/10/2024 03:16:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-58]:                                                 rms_norm_eps=1e-05,
+12/10/2024 03:16:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-58]:                                                 rope_scaling=None,
+12/10/2024 03:16:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-58]:                                                 rope_theta=10000.0,
+12/10/2024 03:16:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-58]:                                                 rope_interleaved=False,
+12/10/2024 03:16:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-58]:                                                 tie_word_embeddings=True,
+12/10/2024 03:16:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-58]:                                                 use_cache=True,
+12/10/2024 03:16:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-58]:                                                 vocab_size=32768),
+12/10/2024 03:16:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-58]:                        init_method=RandomInit(std=0.02),
+12/10/2024 03:16:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-58]:                        dtype=torch.bfloat16,
+12/10/2024 03:16:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-58]:                        make_vocab_size_divisible_by=1,
+12/10/2024 03:16:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-58]:                        ddp_bucket_cap_mb=25),
+12/10/2024 03:16:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-58]:        tokenizer=TokenizerArgs(tokenizer_name_or_path='robot-test/dummy-tokenizer-wordlevel',
+12/10/2024 03:16:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-58]:                                tokenizer_revision=None,
+12/10/2024 03:16:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-58]:                                tokenizer_max_length=None),
+12/10/2024 03:16:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-58]:        checkpoints=CheckpointsArgs(checkpoints_path=PosixPath('checkpoints'),
+12/10/2024 03:16:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-58]:                                    checkpoint_interval=10000,
+12/10/2024 03:16:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-58]:                                    save_initial_state=False,
+12/10/2024 03:16:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-58]:                                    save_final_state=False,
+12/10/2024 03:16:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-58]:                                    resume_checkpoint_path=None,
+12/10/2024 03:16:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-58]:                                    checkpoints_path_is_shared_file_system=False),
+12/10/2024 03:16:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-58]:        logging=LoggingArgs(log_level='info',
+12/10/2024 03:16:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-58]:                            log_level_replica='info',
+12/10/2024 03:16:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-58]:                            iteration_step_info_interval=1),
+12/10/2024 03:16:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-58]:        tokens=TokensArgs(sequence_length=32768,
+12/10/2024 03:16:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-58]:                          train_steps=100,
+12/10/2024 03:16:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-58]:                          micro_batch_size=8,
+12/10/2024 03:16:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-58]:                          batch_accumulation_per_replica=4,
+12/10/2024 03:16:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-58]:                          val_check_interval=100,
+12/10/2024 03:16:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-58]:                          limit_val_batches=0,
+12/10/2024 03:16:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-58]:                          limit_test_batches=0),
+12/10/2024 03:16:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-58]:        optimizer=OptimizerArgs(optimizer_factory=AdamWOptimizerArgs(adam_eps=1e-08,
+12/10/2024 03:16:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-58]:                                                                     adam_beta1=0.9,
+12/10/2024 03:16:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-58]:                                                                     adam_beta2=0.95,
+12/10/2024 03:16:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-58]:                                                                     torch_adam_is_fused=True,
+12/10/2024 03:16:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-58]:                                                                     name='adamW'),
+12/10/2024 03:16:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-58]:                                zero_stage=1,
+12/10/2024 03:16:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-58]:                                weight_decay=0.01,
+12/10/2024 03:16:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-58]:                                clip_grad=1.0,
+12/10/2024 03:16:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-58]:                                accumulate_grad_in_fp32=True,
+12/10/2024 03:16:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-58]:                                learning_rate_scheduler=LRSchedulerArgs(learning_rate=0.0003,
+12/10/2024 03:16:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-58]:                                                                        lr_warmup_steps=2,
+12/10/2024 03:16:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-58]:                                                                        lr_warmup_style='linear',
+12/10/2024 03:16:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-58]:                                                                        lr_decay_style='cosine',
+12/10/2024 03:16:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-58]:                                                                        lr_decay_steps=13,
+12/10/2024 03:16:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-58]:                                                                        lr_decay_starting_step=None,
+12/10/2024 03:16:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-58]:                                                                        min_decay_lr=1e-05)),
+12/10/2024 03:16:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-58]:        data_stages=[DatasetStageArgs(name='Stable Training Stage',
+12/10/2024 03:16:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-58]:                                      start_training_step=1,
+12/10/2024 03:16:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-58]:                                      data=DataArgs(dataset=None,
+12/10/2024 03:16:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-58]:                                                    seed=42,
+12/10/2024 03:16:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-58]:                                                    num_loading_workers=1))],
+12/10/2024 03:16:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-58]:        profiler=None,
+12/10/2024 03:16:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-58]:        lighteval=None,
+12/10/2024 03:16:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-58]:        s3_upload=None)
+12/10/2024 03:16:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-58]: Model Config:
+12/10/2024 03:16:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-58]: LlamaConfig(bos_token_id=0,
+12/10/2024 03:16:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-58]:             eos_token_id=0,
+12/10/2024 03:16:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-58]:             hidden_act='silu',
+12/10/2024 03:16:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-58]:             hidden_size=2048,
+12/10/2024 03:16:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-58]:             initializer_range=0.02,
+12/10/2024 03:16:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-58]:             intermediate_size=8192,
+12/10/2024 03:16:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-58]:             is_llama_config=True,
+12/10/2024 03:16:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-58]:             max_position_embeddings=32768,
+12/10/2024 03:16:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-58]:             num_attention_heads=32,
+12/10/2024 03:16:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-58]:             num_hidden_layers=16,
+12/10/2024 03:16:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-58]:             num_key_value_heads=8,
+12/10/2024 03:16:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-58]:             pad_token_id=None,
+12/10/2024 03:16:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-58]:             pretraining_tp=1,
+12/10/2024 03:16:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-58]:             rms_norm_eps=1e-05,
+12/10/2024 03:16:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-58]:             rope_scaling=None,
+12/10/2024 03:16:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-58]:             rope_theta=10000.0,
+12/10/2024 03:16:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-58]:             rope_interleaved=False,
+12/10/2024 03:16:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-58]:             tie_word_embeddings=True,
+12/10/2024 03:16:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-58]:             use_cache=True,
+12/10/2024 03:16:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-58]:             vocab_size=32768)
+12/10/2024 03:16:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-58]: Building model..
+12/10/2024 03:16:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-58]: Initialize RoPE Theta = 10000.0
+12/10/2024 03:16:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-58]: Setting PP block ranks...
+12/10/2024 03:16:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-58]: Total number of parameters: 1.04G (1984.52MiB)
+12/10/2024 03:16:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-58]: Local number of parameters: 260M (496.13MiB)
+12/10/2024 03:16:33 [INFO|DP=0|PP=0|TP=2|ip-26-0-163-58]: Local number of parameters: 260M (496.13MiB)
+12/10/2024 03:16:33 [INFO|DP=0|PP=0|TP=3|ip-26-0-163-58]: Local number of parameters: 260M (496.13MiB)
+12/10/2024 03:16:33 [INFO|DP=0|PP=0|TP=1|ip-26-0-163-58]: Local number of parameters: 260M (496.13MiB)
+12/10/2024 03:16:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-58]: [After model building] Memory usage: 497.15MiB. Peak allocated: 5504.00MiB Peak reserved: 22914.00MiB
+12/10/2024 03:16:33 [INFO|DP=0|PP=0|TP=2|ip-26-0-163-58]: [After model building] Memory usage: 497.15MiB. Peak allocated: 5504.00MiB Peak reserved: 22914.00MiB
+12/10/2024 03:16:33 [INFO|DP=0|PP=0|TP=3|ip-26-0-163-58]: [After model building] Memory usage: 497.15MiB. Peak allocated: 5504.00MiB Peak reserved: 22914.00MiB
+12/10/2024 03:16:33 [INFO|DP=0|PP=0|TP=1|ip-26-0-163-58]: [After model building] Memory usage: 497.15MiB. Peak allocated: 5504.00MiB Peak reserved: 22914.00MiB
+12/10/2024 03:16:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-58]: No checkpoint path provided.
+12/10/2024 03:16:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-58]: Parametrizing model parameters using StandardParametrizator
+12/10/2024 03:16:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-58]: [Optimizer Building] Using LearningRateForSP as learning rate
+12/10/2024 03:16:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-58]: [ZeRO sharding] Size of optimizer params per rank:
+12/10/2024 03:16:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-58]: [ZeRO sharding] DP Rank 0 has 65M out of 260M (25.00%) params' optimizer states
+12/10/2024 03:16:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-58]: [ZeRO sharding] DP Rank 1 has 65M out of 260M (25.00%) params' optimizer states
+12/10/2024 03:16:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-58]: [ZeRO sharding] DP Rank 2 has 65M out of 260M (25.00%) params' optimizer states
+12/10/2024 03:16:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-58]: [ZeRO sharding] DP Rank 3 has 65M out of 260M (25.00%) params' optimizer states
+12/10/2024 03:16:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-58]: [Training Plan] Stage Stable Training Stage has 99 remaining training steps and has consumed 0 samples
+12/10/2024 03:16:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-58]: Using dummy data generator
+12/10/2024 03:16:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-58]: [Training Plan] There are 1 training stages 
+12/10/2024 03:16:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-58]: [Stage Stable Training Stage] start from step 1 
+12/10/2024 03:16:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-58]: 
+12/10/2024 03:16:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-58]: [Start training] datetime: 2024-12-10 03:16:36.682169 | mbs: 8 | grad_accum: 4 | global_batch_size: 128 | sequence_length: 32768 | train_steps: 100 | start_iteration_step: 0 | consumed_train_samples: 0
+wandb: WARNING Path /fsx/nouamane/.cache/wandb/wandb/ wasn't writable, using system temp directory.
+wandb: WARNING Path /fsx/nouamane/.cache/wandb/wandb/ wasn't writable, using system temp directory
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 727, in _core_forward
+    hidden_states = hidden_states + residual
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 1024.00 MiB. GPU 7 has a total capacty of 79.33 GiB of which 927.94 MiB is free. Including non-PyTorch memory, this process has 78.41 GiB memory in use. Of the allocated memory 67.84 GiB is allocated by PyTorch, and 559.83 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 727, in _core_forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states = hidden_states + residual
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 1024.00 MiB. GPU 5 has a total capacty of 79.33 GiB of which 359.94 MiB is free. Including non-PyTorch memory, this process has 78.97 GiB memory in use. Of the allocated memory 67.84 GiB is allocated by PyTorch, and 687.83 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 727, in _core_forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_states = hidden_states + residual
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 1024.00 MiB. GPU 6 has a total capacty of 79.33 GiB of which 359.94 MiB is free. Including non-PyTorch memory, this process has 78.97 GiB memory in use. Of the allocated memory 67.84 GiB is allocated by PyTorch, and 687.83 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 727, in _core_forward
+    hidden_states = hidden_states + residual
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 1024.00 MiB. GPU 4 has a total capacty of 79.33 GiB of which 263.94 MiB is free. Including non-PyTorch memory, this process has 79.06 GiB memory in use. Of the allocated memory 67.84 GiB is allocated by PyTorch, and 815.83 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 727, in _core_forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    hidden_states = hidden_states + residual
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 1024.00 MiB. GPU 1 has a total capacty of 79.33 GiB of which 359.94 MiB is free. Including non-PyTorch memory, this process has 78.97 GiB memory in use. Of the allocated memory 67.84 GiB is allocated by PyTorch, and 687.83 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 727, in _core_forward
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    hidden_states = hidden_states + residual
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 1024.00 MiB. GPU 0 has a total capacty of 79.33 GiB of which 439.94 MiB is free. Including non-PyTorch memory, this process has 78.89 GiB memory in use. Of the allocated memory 67.84 GiB is allocated by PyTorch, and 687.83 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return forward_call(*args, **kwargs)
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 727, in _core_forward
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_states = hidden_states + residual
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 1024.00 MiB. GPU 2 has a total capacty of 79.33 GiB of which 359.94 MiB is free. Including non-PyTorch memory, this process has 78.97 GiB memory in use. Of the allocated memory 67.84 GiB is allocated by PyTorch, and 687.83 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 727, in _core_forward
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 727, in _core_forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_states = hidden_states + residual
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    trainer.train(dataloader)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 1024.00 MiB. GPU 3 has a total capacty of 79.33 GiB of which 351.94 MiB is free. Including non-PyTorch memory, this process has 78.97 GiB memory in use. Of the allocated memory 67.84 GiB is allocated by PyTorch, and 815.83 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    hidden_states = hidden_states + residual
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 1024.00 MiB. GPU 7 has a total capacty of 79.33 GiB of which 671.94 MiB is free. Including non-PyTorch memory, this process has 78.66 GiB memory in use. Of the allocated memory 67.84 GiB is allocated by PyTorch, and 815.83 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)    
+hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 727, in _core_forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states = hidden_states + residual
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 1024.00 MiB. GPU 4 has a total capacty of 79.33 GiB of which 263.94 MiB is free. Including non-PyTorch memory, this process has 79.06 GiB memory in use. Of the allocated memory 67.84 GiB is allocated by PyTorch, and 815.83 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+        return self._call_impl(*args, **kwargs)output = self.pp_block(**new_kwargs)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 727, in _core_forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_states = hidden_states + residual
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 1024.00 MiB. GPU 6 has a total capacty of 79.33 GiB of which 231.94 MiB is free. Including non-PyTorch memory, this process has 79.09 GiB memory in use. Of the allocated memory 67.84 GiB is allocated by PyTorch, and 815.83 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 727, in _core_forward
+    hidden_states = hidden_states + residual
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 1024.00 MiB. GPU 5 has a total capacty of 79.33 GiB of which 359.94 MiB is free. Including non-PyTorch memory, this process has 78.97 GiB memory in use. Of the allocated memory 67.84 GiB is allocated by PyTorch, and 687.83 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+[2024-12-10 03:16:42,239] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 58635 closing signal SIGTERM
+[2024-12-10 03:16:42,240] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 58636 closing signal SIGTERM
+[2024-12-10 03:16:42,240] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 58637 closing signal SIGTERM
+[2024-12-10 03:16:42,240] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 58638 closing signal SIGTERM
+[2024-12-10 03:16:42,240] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 58641 closing signal SIGTERM
+[2024-12-10 03:16:42,240] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 58642 closing signal SIGTERM
+[2024-12-10 03:16:45,649] torch.distributed.elastic.multiprocessing.api: [ERROR] failed (exitcode: 1) local_rank: 4 (pid: 58639) of binary: /fsx/nouamane/miniconda/envs/2-1-cu121/bin/python
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 33, in <module>
+    sys.exit(load_entry_point('torch==2.1.1', 'console_scripts', 'torchrun')())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 264, in launch_agent
+    raise ChildFailedError(
+torch.distributed.elastic.multiprocessing.errors.ChildFailedError: 
+============================================================
+run_train.py FAILED
+------------------------------------------------------------
+Failures:
+[1]:
+  time      : 2024-12-10_03:16:42
+  host      : ip-26-0-163-58.ec2.internal
+  rank      : 5 (local_rank: 5)
+  exitcode  : 1 (pid: 58640)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+------------------------------------------------------------
+Root Cause (first observed failure):
+[0]:
+  time      : 2024-12-10_03:16:42
+  host      : ip-26-0-163-58.ec2.internal
+  rank      : 4 (local_rank: 4)
+  exitcode  : 1 (pid: 58639)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+============================================================
+[2024-12-10 03:16:46,337] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-164-45.ec2.internal_2855078_0' has failed to send a keep-alive heartbeat to the rendezvous '13265633' due to an error of type RendezvousConnectionError.
+srun: error: ip-26-0-163-58: task 0: Exited with exit code 1
+[2024-12-10 03:16:47,246] torch.distributed.elastic.multiprocessing.api: [ERROR] failed (exitcode: 1) local_rank: 0 (pid: 2855088) of binary: /fsx/nouamane/miniconda/envs/2-1-cu121/bin/python
+[2024-12-10 03:16:47,301] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-164-45.ec2.internal_2855078_0' has failed to shutdown the rendezvous '13265633' due to an error of type RendezvousConnectionError.
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 33, in <module>
+    sys.exit(load_entry_point('torch==2.1.1', 'console_scripts', 'torchrun')())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 264, in launch_agent
+    raise ChildFailedError(
+torch.distributed.elastic.multiprocessing.errors.ChildFailedError: 
+============================================================
+run_train.py FAILED
+------------------------------------------------------------
+Failures:
+[1]:
+  time      : 2024-12-10_03:16:47
+  host      : ip-26-0-164-45.ec2.internal
+  rank      : 9 (local_rank: 1)
+  exitcode  : 1 (pid: 2855089)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[2]:
+  time      : 2024-12-10_03:16:47
+  host      : ip-26-0-164-45.ec2.internal
+  rank      : 10 (local_rank: 2)
+  exitcode  : 1 (pid: 2855090)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[3]:
+  time      : 2024-12-10_03:16:47
+  host      : ip-26-0-164-45.ec2.internal
+  rank      : 11 (local_rank: 3)
+  exitcode  : 1 (pid: 2855091)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[4]:
+  time      : 2024-12-10_03:16:47
+  host      : ip-26-0-164-45.ec2.internal
+  rank      : 12 (local_rank: 4)
+  exitcode  : 1 (pid: 2855092)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[5]:
+  time      : 2024-12-10_03:16:47
+  host      : ip-26-0-164-45.ec2.internal
+  rank      : 13 (local_rank: 5)
+  exitcode  : 1 (pid: 2855093)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[6]:
+  time      : 2024-12-10_03:16:47
+  host      : ip-26-0-164-45.ec2.internal
+  rank      : 14 (local_rank: 6)
+  exitcode  : 1 (pid: 2855094)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[7]:
+  time      : 2024-12-10_03:16:47
+  host      : ip-26-0-164-45.ec2.internal
+  rank      : 15 (local_rank: 7)
+  exitcode  : 1 (pid: 2855095)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+------------------------------------------------------------
+Root Cause (first observed failure):
+[0]:
+  time      : 2024-12-10_03:16:47
+  host      : ip-26-0-164-45.ec2.internal
+  rank      : 8 (local_rank: 0)
+  exitcode  : 1 (pid: 2855088)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+============================================================
+srun: error: ip-26-0-164-45: task 1: Exited with exit code 1
diff --git a/logs/13265744-bench_1.14G_dp8_tp4_pp1_acc4_mbs16_seq8192_zero1_tpmodeRED_vocab32k.out b/logs/13265744-bench_1.14G_dp8_tp4_pp1_acc4_mbs16_seq8192_zero1_tpmodeRED_vocab32k.out
new file mode 100644
index 0000000000000000000000000000000000000000..b1d2de2f49d14f9bd7e83d53edf13733a66fbbfc
--- /dev/null
+++ b/logs/13265744-bench_1.14G_dp8_tp4_pp1_acc4_mbs16_seq8192_zero1_tpmodeRED_vocab32k.out
@@ -0,0 +1,952 @@
++ source /etc/profile.d/modules.sh
+++ . /usr/share/modules/init/bash
++++ unset _mlshdbg
++++ '[' 0 = 1 ']'
++++ unset _mlre _mlIFS
++++ '[' -n x ']'
++++ _mlIFS=' 	
+'
++++ IFS=' '
++++ '[' -n '' ']'
+++++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash autoinit
++++ _mlcode='module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+MODULEPATH_modshare=/etc/environment-modules/modules:1:/usr/share/modules/\$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1; export MODULEPATH_modshare;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+LOADEDMODULES=; export LOADEDMODULES;
+MODULEPATH=/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/\$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles; export MODULEPATH;
+test 0;'
++++ _mlret=0
++++ '[' -n x ']'
++++ IFS=' 	
+'
++++ unset _mlIFS
++++ unset _mlre _mlv _mlrv
++++ '[' 0 -eq 0 ']'
++++ eval 'module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+MODULEPATH_modshare=/etc/environment-modules/modules:1:/usr/share/modules/\$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1; export MODULEPATH_modshare;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+LOADEDMODULES=; export LOADEDMODULES;
+MODULEPATH=/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/\$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles; export MODULEPATH;
+test 0;'
+++++ MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl
+++++ export MODULES_CMD
+++++ ENV=/usr/share/modules/init/profile.sh
+++++ export ENV
+++++ MODULEPATH_modshare='/etc/environment-modules/modules:1:/usr/share/modules/$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1'
+++++ export MODULEPATH_modshare
+++++ BASH_ENV=/usr/share/modules/init/bash
+++++ export BASH_ENV
+++++ MODULESHOME=/usr/share/modules
+++++ export MODULESHOME
+++++ LOADEDMODULES=
+++++ export LOADEDMODULES
+++++ MODULEPATH='/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles'
+++++ export MODULEPATH
+++++ test 0
++++ '[' 0 = 1 ']'
++++ '[' -t 2 ']'
++++ export -f module
++++ export -f switchml
++++ '[' 5 -ge 3 ']'
++++ [[ ehxB =~ i ]]
++++ [[ ! :/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin: =~ :/usr/bin: ]]
++++ '[' '!' -n '' ']'
++++ MANPATH=:
++++ export MANPATH
+++++ manpath
++++ [[ ! :/admin/home/nouamane/.local/share/man:/fsx/nouamane/miniconda/envs/2-1-cu121/man:/fsx/nouamane/miniconda/envs/2-1-cu121/share/man:/fsx/nouamane/miniconda/man:/fsx/nouamane/miniconda/share/man:/opt/amazon/openmpi/share/man:/opt/amazon/efa/share/man:/opt/slurm/share/man:/usr/local/man:/usr/local/share/man:/usr/share/man:/admin/home/nouamane/.fzf/man:: =~ :/usr/share/man: ]]
++++ unset _mlcode _mlret
++++ '[' -n '' ']'
++ module load cuda/12.1
++ unset _mlshdbg
++ '[' 0 = 1 ']'
++ unset _mlre _mlIFS
++ '[' -n x ']'
++ _mlIFS=' 	
+'
++ IFS=' '
++ '[' -n '' ']'
+++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash load cuda/12.1
++ eval 'CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include;' export 'CPATH;
+LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:;' export 'LD_LIBRARY_PATH;
+FI_EFA_FORK_SAFE=1;' export 'FI_EFA_FORK_SAFE;
+MANPATH=/usr/local/cuda-12.1/share/man::;' export 'MANPATH;
+LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64;' export 'LIBRARY_PATH;
+_LMFILES_=/usr/share/modules/modulefiles/cuda/12.1;' export '_LMFILES_;
+LOADEDMODULES=cuda/12.1;' export 'LOADEDMODULES;
+MPI_PATH=/opt/amazon/openmpi;' export 'MPI_PATH;
+NCCL_HOME_modshare=/opt/nccl/build:1;' export 'NCCL_HOME_modshare;
+NCCL_PROTO=simple;' export 'NCCL_PROTO;
+MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1;' export 'MANPATH_modshare;
+LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1;' export 'LIBRARY_PATH_modshare;
+NCCL_SOCKET_IFNAME=enp;' export 'NCCL_SOCKET_IFNAME;
+AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl;' export 'AWS_OFI_NCCL_HOME;
+NCCL_HOME=/opt/nccl/build;' export 'NCCL_HOME;
+FI_PROVIDER=efa;' export 'FI_PROVIDER;
+AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1;' export 'AWS_OFI_NCCL_HOME_modshare;
+CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1;' export 'CPATH_modshare;
+LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2::1:/opt/nccl/build/lib:1:/opt/aws-ofi-nccl/lib:2:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/efa/lib:1;' export 'LD_LIBRARY_PATH_modshare;
+FI_EFA_ENABLE_SHM_TRANSFER=1;' export 'FI_EFA_ENABLE_SHM_TRANSFER;
+_LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1;' export '_LMFILES__modshare;
+LOADEDMODULES_modshare=cuda/12.1:1;' export 'LOADEDMODULES_modshare;
+MPI_PATH_modshare=/opt/amazon/openmpi:1;' export 'MPI_PATH_modshare;
+PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin;' export 'PATH;
+CUDA_HOME=/usr/local/cuda-12.1;' export 'CUDA_HOME;
+PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/usr/local/cuda-12.1/include:1:/opt/amazon/efa/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/usr/sbin:1:/fsx/nouamane/miniconda/bin:1:/opt/slurm/bin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:2:/opt/amazon/openmpi/bin:1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1;' export 'PATH_modshare;
+test' '0;'
+++ CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include
+++ export CPATH
+++ LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:
+++ export LD_LIBRARY_PATH
+++ FI_EFA_FORK_SAFE=1
+++ export FI_EFA_FORK_SAFE
+++ MANPATH=/usr/local/cuda-12.1/share/man::
+++ export MANPATH
+++ LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64
+++ export LIBRARY_PATH
+++ _LMFILES_=/usr/share/modules/modulefiles/cuda/12.1
+++ export _LMFILES_
+++ LOADEDMODULES=cuda/12.1
+++ export LOADEDMODULES
+++ MPI_PATH=/opt/amazon/openmpi
+++ export MPI_PATH
+++ NCCL_HOME_modshare=/opt/nccl/build:1
+++ export NCCL_HOME_modshare
+++ NCCL_PROTO=simple
+++ export NCCL_PROTO
+++ MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1
+++ export MANPATH_modshare
+++ LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1
+++ export LIBRARY_PATH_modshare
+++ NCCL_SOCKET_IFNAME=enp
+++ export NCCL_SOCKET_IFNAME
+++ AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl
+++ export AWS_OFI_NCCL_HOME
+++ NCCL_HOME=/opt/nccl/build
+++ export NCCL_HOME
+++ FI_PROVIDER=efa
+++ export FI_PROVIDER
+++ AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1
+++ export AWS_OFI_NCCL_HOME_modshare
+++ CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1
+++ export CPATH_modshare
+++ LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2::1:/opt/nccl/build/lib:1:/opt/aws-ofi-nccl/lib:2:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/efa/lib:1
+++ export LD_LIBRARY_PATH_modshare
+++ FI_EFA_ENABLE_SHM_TRANSFER=1
+++ export FI_EFA_ENABLE_SHM_TRANSFER
+++ _LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1
+++ export _LMFILES__modshare
+++ LOADEDMODULES_modshare=cuda/12.1:1
+++ export LOADEDMODULES_modshare
+++ MPI_PATH_modshare=/opt/amazon/openmpi:1
+++ export MPI_PATH_modshare
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ export PATH
+++ CUDA_HOME=/usr/local/cuda-12.1
+++ export CUDA_HOME
+++ PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/usr/local/cuda-12.1/include:1:/opt/amazon/efa/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/usr/sbin:1:/fsx/nouamane/miniconda/bin:1:/opt/slurm/bin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:2:/opt/amazon/openmpi/bin:1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1
+++ export PATH_modshare
+++ test 0
++ _mlstatus=0
++ '[' -n x ']'
++ IFS=' 	
+'
++ unset _mlre _mlv _mlrv _mlIFS
++ '[' -n '' ']'
++ unset _mlshdbg
++ return 0
++ source /fsx/nouamane/miniconda/bin/activate
+++ _CONDA_ROOT=/fsx/nouamane/miniconda
+++ . /fsx/nouamane/miniconda/etc/profile.d/conda.sh
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ '[' -z x ']'
+++ conda activate
+++ local cmd=activate
+++ case "$cmd" in
+++ __conda_activate activate
+++ '[' -n '' ']'
+++ local ask_conda
++++ PS1=
++++ __conda_exe shell.posix activate
++++ /fsx/nouamane/miniconda/bin/conda shell.posix activate
+++ ask_conda='. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_2='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\'''
+++ eval '. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_2='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\'''
++++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh
+++++ test -n ''
+++++ unset XML_CATALOG_FILES
+++++ unset xml_catalog_files_libxml2
++++ PS1='(base) '
++++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++++ export CONDA_PREFIX=/fsx/nouamane/miniconda
++++ CONDA_PREFIX=/fsx/nouamane/miniconda
++++ export CONDA_SHLVL=3
++++ CONDA_SHLVL=3
++++ export CONDA_DEFAULT_ENV=base
++++ CONDA_DEFAULT_ENV=base
++++ export 'CONDA_PROMPT_MODIFIER=(base) '
++++ CONDA_PROMPT_MODIFIER='(base) '
++++ export CONDA_PREFIX_2=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ CONDA_PREFIX_2=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ __conda_hashr
+++ '[' -n '' ']'
+++ '[' -n '' ']'
+++ hash -r
++ conda activate 2-1-cu121
++ local cmd=activate
++ case "$cmd" in
++ __conda_activate activate 2-1-cu121
++ '[' -n '' ']'
++ local ask_conda
+++ PS1='(base) '
+++ __conda_exe shell.posix activate 2-1-cu121
+++ /fsx/nouamane/miniconda/bin/conda shell.posix activate 2-1-cu121
++ ask_conda='PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''4'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_3='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
++ eval 'PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''4'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_3='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
+++ PS1='(2-1-cu121) '
+++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ export CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ export CONDA_SHLVL=4
+++ CONDA_SHLVL=4
+++ export CONDA_DEFAULT_ENV=2-1-cu121
+++ CONDA_DEFAULT_ENV=2-1-cu121
+++ export 'CONDA_PROMPT_MODIFIER=(2-1-cu121) '
+++ CONDA_PROMPT_MODIFIER='(2-1-cu121) '
+++ export CONDA_PREFIX_3=/fsx/nouamane/miniconda
+++ CONDA_PREFIX_3=/fsx/nouamane/miniconda
+++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ export _CE_M=
+++ _CE_M=
+++ export _CE_CONDA=
+++ _CE_CONDA=
+++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh
++++ test -n ''
++++ xml_catalog_files_libxml2=
++++ XML_CATALOG_FILES=
++++ conda_catalog_files=
++++ ifs_libxml2=' 	
+'
++++ IFS=' '
++++ rem=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ for pre in ${rem}
++++ test '' = /fsx/nouamane/miniconda/envs/2-1-cu121
++++ conda_catalog_files=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ rem=
++++ IFS=' 	
+'
++++ conda_catalog_files='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ export 'XML_CATALOG_FILES=file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ XML_CATALOG_FILES='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ unset conda_catalog_files ifs_libxml2 rem
++ __conda_hashr
++ '[' -n '' ']'
++ '[' -n '' ']'
++ hash -r
++ export PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++ PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ scontrol show hostnames 'ip-26-0-160-225,ip-26-0-164-236,ip-26-0-171-[88,102]'
++ export 'NODELIST=ip-26-0-160-225
+ip-26-0-164-236
+ip-26-0-171-88
+ip-26-0-171-102'
++ NODELIST='ip-26-0-160-225
+ip-26-0-164-236
+ip-26-0-171-88
+ip-26-0-171-102'
+++ head -n1
+++ scontrol show hostnames 'ip-26-0-160-225,ip-26-0-164-236,ip-26-0-171-[88,102]'
++ export MASTER_NODE=ip-26-0-160-225
++ MASTER_NODE=ip-26-0-160-225
++ export MASTER_PORT=12356
++ MASTER_PORT=12356
++ export NNODES=4
++ NNODES=4
++ export GPUS_PER_NODE=8
++ GPUS_PER_NODE=8
++ export WORLD_SIZE=32
++ WORLD_SIZE=32
++ export CUDA_DEVICE_MAX_CONNECTIONS=1
++ CUDA_DEVICE_MAX_CONNECTIONS=1
++ export NANOTRON_BENCHMARK=1
++ NANOTRON_BENCHMARK=1
++ echo 'Master node: ip-26-0-160-225'
+Master node: ip-26-0-160-225
++ echo 'All nodes: ip-26-0-160-225
+ip-26-0-164-236
+ip-26-0-171-88
+ip-26-0-171-102'
+All nodes: ip-26-0-160-225
+ip-26-0-164-236
+ip-26-0-171-88
+ip-26-0-171-102
++ echo 'World size: 32'
+World size: 32
++ srun torchrun --nnodes=4 --nproc_per_node=8 --rdzv_id=13265744 --rdzv_backend=c10d --rdzv_endpoint=ip-26-0-160-225:12356 run_train.py --config-file benchmark/configs/config_1.14G_dp8_tp4_pp1_acc4_mbs16_seq8192_zero1_tpmodeRED_vocab32k.yaml
+[2024-12-10 06:00:51,461] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-10 06:00:51,465] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-10 06:00:51,499] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-10 06:00:51,461] torch.distributed.run: [WARNING] 
+[2024-12-10 06:00:51,461] torch.distributed.run: [WARNING] *****************************************
+[2024-12-10 06:00:51,461] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-10 06:00:51,461] torch.distributed.run: [WARNING] *****************************************
+[2024-12-10 06:00:51,466] torch.distributed.run: [WARNING] 
+[2024-12-10 06:00:51,466] torch.distributed.run: [WARNING] *****************************************
+[2024-12-10 06:00:51,466] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-10 06:00:51,466] torch.distributed.run: [WARNING] *****************************************
+[2024-12-10 06:00:51,499] torch.distributed.run: [WARNING] 
+[2024-12-10 06:00:51,499] torch.distributed.run: [WARNING] *****************************************
+[2024-12-10 06:00:51,499] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-10 06:00:51,499] torch.distributed.run: [WARNING] *****************************************
+[2024-12-10 06:00:51,628] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-10 06:00:51,628] torch.distributed.run: [WARNING] 
+[2024-12-10 06:00:51,628] torch.distributed.run: [WARNING] *****************************************
+[2024-12-10 06:00:51,628] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-10 06:00:51,628] torch.distributed.run: [WARNING] *****************************************
+12/10/2024 06:01:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: Config:
+12/10/2024 06:01:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: Config(general=GeneralArgs(project='debug',
+12/10/2024 06:01:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                            run='1.14G_dp8_tp4_pp1_acc4_mbs16_seq8192_zero1_tpmodeRED_vocab32k',
+12/10/2024 06:01:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                            seed=42,
+12/10/2024 06:01:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                            step=None,
+12/10/2024 06:01:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                            consumed_train_samples=None,
+12/10/2024 06:01:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                            benchmark_csv_path=PosixPath('benchmark/results/bench_final.csv'),
+12/10/2024 06:01:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                            ignore_sanity_checks=True),
+12/10/2024 06:01:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:        parallelism=ParallelismArgs(dp=8,
+12/10/2024 06:01:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                    pp=1,
+12/10/2024 06:01:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                    tp=4,
+12/10/2024 06:01:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                    pp_engine=<nanotron.parallel.pipeline_parallel.engine.OneForwardOneBackwardPipelineEngine object at 0x7f43d5774e80>,
+12/10/2024 06:01:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                    tp_mode=<TensorParallelLinearMode.REDUCE_SCATTER: 2>,
+12/10/2024 06:01:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                    tp_linear_async_communication=True,
+12/10/2024 06:01:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                    recompute_layer=False,
+12/10/2024 06:01:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                    tp_recompute_allgather=True,
+12/10/2024 06:01:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                    expert_parallel_size=1),
+12/10/2024 06:01:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:        model=ModelArgs(model_config=LlamaConfig(bos_token_id=0,
+12/10/2024 06:01:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 eos_token_id=0,
+12/10/2024 06:01:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 hidden_act='silu',
+12/10/2024 06:01:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 hidden_size=2048,
+12/10/2024 06:01:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 initializer_range=0.02,
+12/10/2024 06:01:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 intermediate_size=8192,
+12/10/2024 06:01:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 is_llama_config=True,
+12/10/2024 06:01:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 max_position_embeddings=8192,
+12/10/2024 06:01:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 num_attention_heads=32,
+12/10/2024 06:01:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 num_hidden_layers=16,
+12/10/2024 06:01:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 num_key_value_heads=8,
+12/10/2024 06:01:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 pad_token_id=None,
+12/10/2024 06:01:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 pretraining_tp=1,
+12/10/2024 06:01:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 rms_norm_eps=1e-05,
+12/10/2024 06:01:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 rope_scaling=None,
+12/10/2024 06:01:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 rope_theta=10000.0,
+12/10/2024 06:01:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 rope_interleaved=False,
+12/10/2024 06:01:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 tie_word_embeddings=True,
+12/10/2024 06:01:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 use_cache=True,
+12/10/2024 06:01:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 vocab_size=32768),
+12/10/2024 06:01:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                        init_method=RandomInit(std=0.02),
+12/10/2024 06:01:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                        dtype=torch.bfloat16,
+12/10/2024 06:01:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                        make_vocab_size_divisible_by=1,
+12/10/2024 06:01:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                        ddp_bucket_cap_mb=25),
+12/10/2024 06:01:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:        tokenizer=TokenizerArgs(tokenizer_name_or_path='robot-test/dummy-tokenizer-wordlevel',
+12/10/2024 06:01:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                tokenizer_revision=None,
+12/10/2024 06:01:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                tokenizer_max_length=None),
+12/10/2024 06:01:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:        checkpoints=CheckpointsArgs(checkpoints_path=PosixPath('checkpoints'),
+12/10/2024 06:01:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                    checkpoint_interval=10000,
+12/10/2024 06:01:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                    save_initial_state=False,
+12/10/2024 06:01:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                    save_final_state=False,
+12/10/2024 06:01:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                    resume_checkpoint_path=None,
+12/10/2024 06:01:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                    checkpoints_path_is_shared_file_system=False),
+12/10/2024 06:01:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:        logging=LoggingArgs(log_level='info',
+12/10/2024 06:01:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                            log_level_replica='info',
+12/10/2024 06:01:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                            iteration_step_info_interval=1),
+12/10/2024 06:01:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:        tokens=TokensArgs(sequence_length=8192,
+12/10/2024 06:01:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                          train_steps=100,
+12/10/2024 06:01:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                          micro_batch_size=16,
+12/10/2024 06:01:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                          batch_accumulation_per_replica=4,
+12/10/2024 06:01:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                          val_check_interval=100,
+12/10/2024 06:01:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                          limit_val_batches=0,
+12/10/2024 06:01:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                          limit_test_batches=0),
+12/10/2024 06:01:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:        optimizer=OptimizerArgs(optimizer_factory=AdamWOptimizerArgs(adam_eps=1e-08,
+12/10/2024 06:01:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                                     adam_beta1=0.9,
+12/10/2024 06:01:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                                     adam_beta2=0.95,
+12/10/2024 06:01:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                                     torch_adam_is_fused=True,
+12/10/2024 06:01:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                                     name='adamW'),
+12/10/2024 06:01:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                zero_stage=1,
+12/10/2024 06:01:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                weight_decay=0.01,
+12/10/2024 06:01:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                clip_grad=1.0,
+12/10/2024 06:01:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                accumulate_grad_in_fp32=True,
+12/10/2024 06:01:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                learning_rate_scheduler=LRSchedulerArgs(learning_rate=0.0003,
+12/10/2024 06:01:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                                        lr_warmup_steps=2,
+12/10/2024 06:01:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                                        lr_warmup_style='linear',
+12/10/2024 06:01:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                                        lr_decay_style='cosine',
+12/10/2024 06:01:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                                        lr_decay_steps=13,
+12/10/2024 06:01:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                                        lr_decay_starting_step=None,
+12/10/2024 06:01:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                                        min_decay_lr=1e-05)),
+12/10/2024 06:01:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:        data_stages=[DatasetStageArgs(name='Stable Training Stage',
+12/10/2024 06:01:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                      start_training_step=1,
+12/10/2024 06:01:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                      data=DataArgs(dataset=None,
+12/10/2024 06:01:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                    seed=42,
+12/10/2024 06:01:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                    num_loading_workers=1))],
+12/10/2024 06:01:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:        profiler=None,
+12/10/2024 06:01:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:        lighteval=None,
+12/10/2024 06:01:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:        s3_upload=None)
+12/10/2024 06:01:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: Model Config:
+12/10/2024 06:01:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: LlamaConfig(bos_token_id=0,
+12/10/2024 06:01:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             eos_token_id=0,
+12/10/2024 06:01:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             hidden_act='silu',
+12/10/2024 06:01:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             hidden_size=2048,
+12/10/2024 06:01:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             initializer_range=0.02,
+12/10/2024 06:01:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             intermediate_size=8192,
+12/10/2024 06:01:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             is_llama_config=True,
+12/10/2024 06:01:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             max_position_embeddings=8192,
+12/10/2024 06:01:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             num_attention_heads=32,
+12/10/2024 06:01:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             num_hidden_layers=16,
+12/10/2024 06:01:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             num_key_value_heads=8,
+12/10/2024 06:01:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             pad_token_id=None,
+12/10/2024 06:01:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             pretraining_tp=1,
+12/10/2024 06:01:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             rms_norm_eps=1e-05,
+12/10/2024 06:01:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             rope_scaling=None,
+12/10/2024 06:01:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             rope_theta=10000.0,
+12/10/2024 06:01:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             rope_interleaved=False,
+12/10/2024 06:01:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             tie_word_embeddings=True,
+12/10/2024 06:01:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             use_cache=True,
+12/10/2024 06:01:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             vocab_size=32768)
+12/10/2024 06:01:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: Building model..
+12/10/2024 06:01:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: Initialize RoPE Theta = 10000.0
+12/10/2024 06:01:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: Setting PP block ranks...
+12/10/2024 06:01:37 [INFO|DP=0|PP=0|TP=2|ip-26-0-160-225]: Local number of parameters: 260M (496.13MiB)
+12/10/2024 06:01:37 [INFO|DP=0|PP=0|TP=1|ip-26-0-160-225]: Local number of parameters: 260M (496.13MiB)
+12/10/2024 06:01:37 [INFO|DP=0|PP=0|TP=3|ip-26-0-160-225]: Local number of parameters: 260M (496.13MiB)
+12/10/2024 06:01:37 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: Total number of parameters: 1.04G (1984.52MiB)
+12/10/2024 06:01:37 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: Local number of parameters: 260M (496.13MiB)
+12/10/2024 06:01:37 [INFO|DP=0|PP=0|TP=2|ip-26-0-160-225]: [After model building] Memory usage: 497.15MiB. Peak allocated: 5440.00MiB Peak reserved: 35138.00MiB
+12/10/2024 06:01:37 [INFO|DP=0|PP=0|TP=3|ip-26-0-160-225]: [After model building] Memory usage: 497.15MiB. Peak allocated: 5440.00MiB Peak reserved: 35138.00MiB
+12/10/2024 06:01:37 [INFO|DP=0|PP=0|TP=1|ip-26-0-160-225]: [After model building] Memory usage: 497.15MiB. Peak allocated: 5440.00MiB Peak reserved: 35138.00MiB
+12/10/2024 06:01:37 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [After model building] Memory usage: 497.15MiB. Peak allocated: 5440.00MiB Peak reserved: 35138.00MiB
+12/10/2024 06:01:37 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: No checkpoint path provided.
+12/10/2024 06:01:37 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: Parametrizing model parameters using StandardParametrizator
+12/10/2024 06:01:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [Optimizer Building] Using LearningRateForSP as learning rate
+12/10/2024 06:01:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [ZeRO sharding] Size of optimizer params per rank:
+12/10/2024 06:01:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [ZeRO sharding] DP Rank 0 has 32.5M out of 260M (12.50%) params' optimizer states
+12/10/2024 06:01:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [ZeRO sharding] DP Rank 1 has 32.5M out of 260M (12.50%) params' optimizer states
+12/10/2024 06:01:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [ZeRO sharding] DP Rank 2 has 32.5M out of 260M (12.50%) params' optimizer states
+12/10/2024 06:01:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [ZeRO sharding] DP Rank 3 has 32.5M out of 260M (12.50%) params' optimizer states
+12/10/2024 06:01:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [ZeRO sharding] DP Rank 4 has 32.5M out of 260M (12.50%) params' optimizer states
+12/10/2024 06:01:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [ZeRO sharding] DP Rank 5 has 32.5M out of 260M (12.50%) params' optimizer states
+12/10/2024 06:01:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [ZeRO sharding] DP Rank 6 has 32.5M out of 260M (12.50%) params' optimizer states
+12/10/2024 06:01:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [ZeRO sharding] DP Rank 7 has 32.5M out of 260M (12.50%) params' optimizer states
+12/10/2024 06:01:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [Training Plan] Stage Stable Training Stage has 99 remaining training steps and has consumed 0 samples
+12/10/2024 06:01:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: Using dummy data generator
+12/10/2024 06:01:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [Training Plan] There are 1 training stages 
+12/10/2024 06:01:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [Stage Stable Training Stage] start from step 1 
+12/10/2024 06:01:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: 
+12/10/2024 06:01:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [Start training] datetime: 2024-12-10 06:01:41.423585 | mbs: 16 | grad_accum: 4 | global_batch_size: 512 | sequence_length: 8192 | train_steps: 100 | start_iteration_step: 0 | consumed_train_samples: 0
+wandb: WARNING Path /fsx/nouamane/.cache/wandb/wandb/ wasn't writable, using system temp directory.
+wandb: WARNING Path /fsx/nouamane/.cache/wandb/wandb/ wasn't writable, using system temp directory
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1017, in forward
+    loss = self.loss(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 971, in forward
+    loss = sharded_cross_entropy(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 115, in sharded_cross_entropy
+    return _ShardedCrossEntropy.apply(sharded_logits, target, group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 43, in forward
+    sharded_logits = sharded_logits - logits_max.unsqueeze(dim=-1)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 4.00 GiB. GPU 5 has a total capacty of 79.33 GiB of which 3.74 GiB is free. Including non-PyTorch memory, this process has 75.58 GiB memory in use. Of the allocated memory 51.98 GiB is allocated by PyTorch, and 12.86 GiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1017, in forward
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    loss = self.loss(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1017, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 971, in forward
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    loss = self.loss(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    loss = sharded_cross_entropy(
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 115, in sharded_cross_entropy
+outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return _ShardedCrossEntropy.apply(sharded_logits, target, group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 43, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 971, in forward
+    sharded_logits = sharded_logits - logits_max.unsqueeze(dim=-1)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 4.00 GiB. GPU 1 has a total capacty of 79.33 GiB of which 2.12 GiB is free. Including non-PyTorch memory, this process has 77.20 GiB memory in use. Of the allocated memory 51.98 GiB is allocated by PyTorch, and 13.74 GiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1017, in forward
+    loss = sharded_cross_entropy(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 115, in sharded_cross_entropy
+    loss = self.loss(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return _ShardedCrossEntropy.apply(sharded_logits, target, group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 43, in forward
+    sharded_logits = sharded_logits - logits_max.unsqueeze(dim=-1)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 4.00 GiB. GPU 0 has a total capacty of 79.33 GiB of which 3.78 GiB is free. Including non-PyTorch memory, this process has 75.54 GiB memory in use. Of the allocated memory 51.98 GiB is allocated by PyTorch, and 12.24 GiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 971, in forward
+    loss = sharded_cross_entropy(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 115, in sharded_cross_entropy
+    return _ShardedCrossEntropy.apply(sharded_logits, target, group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 43, in forward
+    sharded_logits = sharded_logits - logits_max.unsqueeze(dim=-1)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 4.00 GiB. GPU 2 has a total capacty of 79.33 GiB of which 2.07 GiB is free. Including non-PyTorch memory, this process has 77.24 GiB memory in use. Of the allocated memory 51.98 GiB is allocated by PyTorch, and 13.86 GiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1017, in forward
+    loss = self.loss(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 971, in forward
+    loss = sharded_cross_entropy(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 115, in sharded_cross_entropy
+    return _ShardedCrossEntropy.apply(sharded_logits, target, group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 43, in forward
+    sharded_logits = sharded_logits - logits_max.unsqueeze(dim=-1)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 4.00 GiB. GPU 2 has a total capacty of 79.33 GiB of which 3.49 GiB is free. Including non-PyTorch memory, this process has 75.83 GiB memory in use. Of the allocated memory 51.98 GiB is allocated by PyTorch, and 13.11 GiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+wandb: Tracking run with wandb version 0.16.0
+wandb: W&B syncing is set to `offline` in this directory.  
+wandb: Run `wandb online` or set WANDB_MODE=online to enable cloud syncing.
+12/10/2024 06:01:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: Resuming training from stage Stable Training Stage, it has trained for 0 samples and has 99 remaining train steps
+12/10/2024 06:01:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:  Memory usage: 1613.44MiB. Peak allocated 5440.00MiB. Peak reserved: 35138.00MiB
+[2024-12-10 06:01:53,682] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3132537 closing signal SIGTERM
+[2024-12-10 06:01:53,682] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3132538 closing signal SIGTERM
+[2024-12-10 06:01:53,682] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3132539 closing signal SIGTERM
+[2024-12-10 06:01:53,682] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3132540 closing signal SIGTERM
+[2024-12-10 06:01:53,684] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3132541 closing signal SIGTERM
+[2024-12-10 06:01:53,691] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1337032 closing signal SIGTERM
+[2024-12-10 06:01:53,691] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1337033 closing signal SIGTERM
+[2024-12-10 06:01:53,691] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1337035 closing signal SIGTERM
+[2024-12-10 06:01:53,691] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1337036 closing signal SIGTERM
+[2024-12-10 06:01:53,692] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1337037 closing signal SIGTERM
+[2024-12-10 06:01:53,693] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1337038 closing signal SIGTERM
+[2024-12-10 06:01:53,694] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1337039 closing signal SIGTERM
+[2024-12-10 06:01:53,697] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 158061 closing signal SIGTERM
+[2024-12-10 06:01:53,697] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 158062 closing signal SIGTERM
+[2024-12-10 06:01:53,697] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 158063 closing signal SIGTERM
+[2024-12-10 06:01:53,698] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 158064 closing signal SIGTERM
+[2024-12-10 06:01:53,698] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 158065 closing signal SIGTERM
+[2024-12-10 06:01:53,699] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 158067 closing signal SIGTERM
+[2024-12-10 06:01:53,699] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 158068 closing signal SIGTERM
+[2024-12-10 06:01:59,173] torch.distributed.elastic.multiprocessing.api: [ERROR] failed (exitcode: 1) local_rank: 5 (pid: 158066) of binary: /fsx/nouamane/miniconda/envs/2-1-cu121/bin/python
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 33, in <module>
+    sys.exit(load_entry_point('torch==2.1.1', 'console_scripts', 'torchrun')())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 264, in launch_agent
+    raise ChildFailedError(
+torch.distributed.elastic.multiprocessing.errors.ChildFailedError: 
+============================================================
+run_train.py FAILED
+------------------------------------------------------------
+Failures:
+  <NO_OTHER_FAILURES>
+------------------------------------------------------------
+Root Cause (first observed failure):
+[0]:
+  time      : 2024-12-10_06:01:53
+  host      : ip-26-0-160-225.ec2.internal
+  rank      : 5 (local_rank: 5)
+  exitcode  : 1 (pid: 158066)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+============================================================
+[2024-12-10 06:01:59,883] torch.distributed.elastic.multiprocessing.api: [ERROR] failed (exitcode: 1) local_rank: 0 (pid: 3132534) of binary: /fsx/nouamane/miniconda/envs/2-1-cu121/bin/python
+[2024-12-10 06:01:59,915] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-171-102.ec2.internal_3132525_0' has failed to shutdown the rendezvous '13265744' due to an error of type RendezvousConnectionError.
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 33, in <module>
+    sys.exit(load_entry_point('torch==2.1.1', 'console_scripts', 'torchrun')())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 264, in launch_agent
+    raise ChildFailedError(
+torch.distributed.elastic.multiprocessing.errors.ChildFailedError: 
+============================================================
+run_train.py FAILED
+------------------------------------------------------------
+Failures:
+[1]:
+  time      : 2024-12-10_06:01:53
+  host      : ip-26-0-171-102.ec2.internal
+  rank      : 17 (local_rank: 1)
+  exitcode  : 1 (pid: 3132535)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[2]:
+  time      : 2024-12-10_06:01:53
+  host      : ip-26-0-171-102.ec2.internal
+  rank      : 18 (local_rank: 2)
+  exitcode  : 1 (pid: 3132536)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+------------------------------------------------------------
+Root Cause (first observed failure):
+[0]:
+  time      : 2024-12-10_06:01:53
+  host      : ip-26-0-171-102.ec2.internal
+  rank      : 16 (local_rank: 0)
+  exitcode  : 1 (pid: 3132534)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+============================================================
+[2024-12-10 06:02:00,082] torch.distributed.elastic.multiprocessing.api: [ERROR] failed (exitcode: 1) local_rank: 2 (pid: 1337034) of binary: /fsx/nouamane/miniconda/envs/2-1-cu121/bin/python
+[2024-12-10 06:02:00,110] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-171-88.ec2.internal_1337022_0' has failed to shutdown the rendezvous '13265744' due to an error of type RendezvousConnectionError.
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 33, in <module>
+    sys.exit(load_entry_point('torch==2.1.1', 'console_scripts', 'torchrun')())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 264, in launch_agent
+    raise ChildFailedError(
+torch.distributed.elastic.multiprocessing.errors.ChildFailedError: 
+============================================================
+run_train.py FAILED
+------------------------------------------------------------
+Failures:
+  <NO_OTHER_FAILURES>
+------------------------------------------------------------
+Root Cause (first observed failure):
+[0]:
+  time      : 2024-12-10_06:01:53
+  host      : ip-26-0-171-88.ec2.internal
+  rank      : 26 (local_rank: 2)
+  exitcode  : 1 (pid: 1337034)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+============================================================
+srun: error: ip-26-0-171-102: task 3: Exited with exit code 1
+srun: error: ip-26-0-171-88: task 2: Exited with exit code 1
+srun: error: ip-26-0-160-225: task 0: Exited with exit code 1
+[2024-12-10 06:02:03,610] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-164-236.ec2.internal_27335_0' has failed to send a keep-alive heartbeat to the rendezvous '13265744' due to an error of type RendezvousConnectionError.
+[2024-12-10 06:02:03,709] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 27343 closing signal SIGTERM
+[2024-12-10 06:02:03,710] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 27344 closing signal SIGTERM
+[2024-12-10 06:02:03,710] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 27345 closing signal SIGTERM
+[2024-12-10 06:02:03,710] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 27346 closing signal SIGTERM
+[2024-12-10 06:02:03,711] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 27347 closing signal SIGTERM
+[2024-12-10 06:02:03,711] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 27348 closing signal SIGTERM
+[2024-12-10 06:02:03,713] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 27349 closing signal SIGTERM
+[2024-12-10 06:02:03,713] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 27350 closing signal SIGTERM
+[2024-12-10 06:02:08,612] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-164-236.ec2.internal_27335_0' has failed to send a keep-alive heartbeat to the rendezvous '13265744' due to an error of type RendezvousConnectionError.
+[2024-12-10 06:02:12,962] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-164-236.ec2.internal_27335_0' has failed to shutdown the rendezvous '13265744' due to an error of type RendezvousConnectionError.
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/rendezvous/c10d_rendezvous_backend.py", line 113, in _call_store
+    return getattr(self._store, store_op)(*args, **kwargs)
+RuntimeError: Broken pipe
+
+The above exception was the direct cause of the following exception:
+
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 33, in <module>
+    sys.exit(load_entry_point('torch==2.1.1', 'console_scripts', 'torchrun')())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 909, in _invoke_run
+    num_nodes_waiting = rdzv_handler.num_nodes_waiting()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/rendezvous/dynamic_rendezvous.py", line 1083, in num_nodes_waiting
+    self._state_holder.sync()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/rendezvous/dynamic_rendezvous.py", line 409, in sync
+    get_response = self._backend.get_state()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/rendezvous/c10d_rendezvous_backend.py", line 73, in get_state
+    base64_state: bytes = self._call_store("get", self._key)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/rendezvous/c10d_rendezvous_backend.py", line 115, in _call_store
+    raise RendezvousConnectionError(
+torch.distributed.elastic.rendezvous.api.RendezvousConnectionError: The connection to the C10d store has failed. See inner exception for details.
+srun: error: ip-26-0-164-236: task 1: Exited with exit code 1
diff --git a/logs/13265746-bench_1.34G_dp8_tp4_pp1_acc4_mbs16_seq8192_zero1_tpmodeRED_vocab131k.out b/logs/13265746-bench_1.34G_dp8_tp4_pp1_acc4_mbs16_seq8192_zero1_tpmodeRED_vocab131k.out
new file mode 100644
index 0000000000000000000000000000000000000000..14a8af6b4fb0ad1b4f7da8a69ff2be2a64ef48fa
--- /dev/null
+++ b/logs/13265746-bench_1.34G_dp8_tp4_pp1_acc4_mbs16_seq8192_zero1_tpmodeRED_vocab131k.out
@@ -0,0 +1,1805 @@
++ source /etc/profile.d/modules.sh
+++ . /usr/share/modules/init/bash
++++ unset _mlshdbg
++++ '[' 0 = 1 ']'
++++ unset _mlre _mlIFS
++++ '[' -n x ']'
++++ _mlIFS=' 	
+'
++++ IFS=' '
++++ '[' -n '' ']'
+++++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash autoinit
++++ _mlcode='module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+MODULEPATH_modshare=/etc/environment-modules/modules:1:/usr/share/modules/\$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1; export MODULEPATH_modshare;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+LOADEDMODULES=; export LOADEDMODULES;
+MODULEPATH=/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/\$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles; export MODULEPATH;
+test 0;'
++++ _mlret=0
++++ '[' -n x ']'
++++ IFS=' 	
+'
++++ unset _mlIFS
++++ unset _mlre _mlv _mlrv
++++ '[' 0 -eq 0 ']'
++++ eval 'module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+MODULEPATH_modshare=/etc/environment-modules/modules:1:/usr/share/modules/\$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1; export MODULEPATH_modshare;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+LOADEDMODULES=; export LOADEDMODULES;
+MODULEPATH=/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/\$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles; export MODULEPATH;
+test 0;'
+++++ MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl
+++++ export MODULES_CMD
+++++ ENV=/usr/share/modules/init/profile.sh
+++++ export ENV
+++++ MODULEPATH_modshare='/etc/environment-modules/modules:1:/usr/share/modules/$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1'
+++++ export MODULEPATH_modshare
+++++ BASH_ENV=/usr/share/modules/init/bash
+++++ export BASH_ENV
+++++ MODULESHOME=/usr/share/modules
+++++ export MODULESHOME
+++++ LOADEDMODULES=
+++++ export LOADEDMODULES
+++++ MODULEPATH='/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles'
+++++ export MODULEPATH
+++++ test 0
++++ '[' 0 = 1 ']'
++++ '[' -t 2 ']'
++++ export -f module
++++ export -f switchml
++++ '[' 5 -ge 3 ']'
++++ [[ ehxB =~ i ]]
++++ [[ ! :/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin: =~ :/usr/bin: ]]
++++ '[' '!' -n '' ']'
++++ MANPATH=:
++++ export MANPATH
+++++ manpath
++++ [[ ! :/admin/home/nouamane/.local/share/man:/fsx/nouamane/miniconda/envs/2-1-cu121/man:/fsx/nouamane/miniconda/envs/2-1-cu121/share/man:/fsx/nouamane/miniconda/man:/fsx/nouamane/miniconda/share/man:/opt/amazon/openmpi/share/man:/opt/amazon/efa/share/man:/opt/slurm/share/man:/usr/local/man:/usr/local/share/man:/usr/share/man:/admin/home/nouamane/.fzf/man:: =~ :/usr/share/man: ]]
++++ unset _mlcode _mlret
++++ '[' -n '' ']'
++ module load cuda/12.1
++ unset _mlshdbg
++ '[' 0 = 1 ']'
++ unset _mlre _mlIFS
++ '[' -n x ']'
++ _mlIFS=' 	
+'
++ IFS=' '
++ '[' -n '' ']'
+++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash load cuda/12.1
++ eval 'CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include;' export 'CPATH;
+LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:;' export 'LD_LIBRARY_PATH;
+FI_EFA_FORK_SAFE=1;' export 'FI_EFA_FORK_SAFE;
+MANPATH=/usr/local/cuda-12.1/share/man::;' export 'MANPATH;
+LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64;' export 'LIBRARY_PATH;
+_LMFILES_=/usr/share/modules/modulefiles/cuda/12.1;' export '_LMFILES_;
+LOADEDMODULES=cuda/12.1;' export 'LOADEDMODULES;
+MPI_PATH=/opt/amazon/openmpi;' export 'MPI_PATH;
+NCCL_HOME_modshare=/opt/nccl/build:1;' export 'NCCL_HOME_modshare;
+NCCL_PROTO=simple;' export 'NCCL_PROTO;
+MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1;' export 'MANPATH_modshare;
+LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1;' export 'LIBRARY_PATH_modshare;
+NCCL_SOCKET_IFNAME=enp;' export 'NCCL_SOCKET_IFNAME;
+AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl;' export 'AWS_OFI_NCCL_HOME;
+NCCL_HOME=/opt/nccl/build;' export 'NCCL_HOME;
+FI_PROVIDER=efa;' export 'FI_PROVIDER;
+AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1;' export 'AWS_OFI_NCCL_HOME_modshare;
+CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1;' export 'CPATH_modshare;
+LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2::1:/opt/nccl/build/lib:1:/opt/aws-ofi-nccl/lib:2:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/efa/lib:1;' export 'LD_LIBRARY_PATH_modshare;
+FI_EFA_ENABLE_SHM_TRANSFER=1;' export 'FI_EFA_ENABLE_SHM_TRANSFER;
+_LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1;' export '_LMFILES__modshare;
+LOADEDMODULES_modshare=cuda/12.1:1;' export 'LOADEDMODULES_modshare;
+MPI_PATH_modshare=/opt/amazon/openmpi:1;' export 'MPI_PATH_modshare;
+PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin;' export 'PATH;
+CUDA_HOME=/usr/local/cuda-12.1;' export 'CUDA_HOME;
+PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/usr/local/cuda-12.1/include:1:/opt/amazon/efa/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/usr/sbin:1:/fsx/nouamane/miniconda/bin:1:/opt/slurm/bin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:2:/opt/amazon/openmpi/bin:1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1;' export 'PATH_modshare;
+test' '0;'
+++ CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include
+++ export CPATH
+++ LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:
+++ export LD_LIBRARY_PATH
+++ FI_EFA_FORK_SAFE=1
+++ export FI_EFA_FORK_SAFE
+++ MANPATH=/usr/local/cuda-12.1/share/man::
+++ export MANPATH
+++ LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64
+++ export LIBRARY_PATH
+++ _LMFILES_=/usr/share/modules/modulefiles/cuda/12.1
+++ export _LMFILES_
+++ LOADEDMODULES=cuda/12.1
+++ export LOADEDMODULES
+++ MPI_PATH=/opt/amazon/openmpi
+++ export MPI_PATH
+++ NCCL_HOME_modshare=/opt/nccl/build:1
+++ export NCCL_HOME_modshare
+++ NCCL_PROTO=simple
+++ export NCCL_PROTO
+++ MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1
+++ export MANPATH_modshare
+++ LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1
+++ export LIBRARY_PATH_modshare
+++ NCCL_SOCKET_IFNAME=enp
+++ export NCCL_SOCKET_IFNAME
+++ AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl
+++ export AWS_OFI_NCCL_HOME
+++ NCCL_HOME=/opt/nccl/build
+++ export NCCL_HOME
+++ FI_PROVIDER=efa
+++ export FI_PROVIDER
+++ AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1
+++ export AWS_OFI_NCCL_HOME_modshare
+++ CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1
+++ export CPATH_modshare
+++ LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2::1:/opt/nccl/build/lib:1:/opt/aws-ofi-nccl/lib:2:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/efa/lib:1
+++ export LD_LIBRARY_PATH_modshare
+++ FI_EFA_ENABLE_SHM_TRANSFER=1
+++ export FI_EFA_ENABLE_SHM_TRANSFER
+++ _LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1
+++ export _LMFILES__modshare
+++ LOADEDMODULES_modshare=cuda/12.1:1
+++ export LOADEDMODULES_modshare
+++ MPI_PATH_modshare=/opt/amazon/openmpi:1
+++ export MPI_PATH_modshare
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ export PATH
+++ CUDA_HOME=/usr/local/cuda-12.1
+++ export CUDA_HOME
+++ PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/usr/local/cuda-12.1/include:1:/opt/amazon/efa/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/usr/sbin:1:/fsx/nouamane/miniconda/bin:1:/opt/slurm/bin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:2:/opt/amazon/openmpi/bin:1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1
+++ export PATH_modshare
+++ test 0
++ _mlstatus=0
++ '[' -n x ']'
++ IFS=' 	
+'
++ unset _mlre _mlv _mlrv _mlIFS
++ '[' -n '' ']'
++ unset _mlshdbg
++ return 0
++ source /fsx/nouamane/miniconda/bin/activate
+++ _CONDA_ROOT=/fsx/nouamane/miniconda
+++ . /fsx/nouamane/miniconda/etc/profile.d/conda.sh
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ '[' -z x ']'
+++ conda activate
+++ local cmd=activate
+++ case "$cmd" in
+++ __conda_activate activate
+++ '[' -n '' ']'
+++ local ask_conda
++++ PS1=
++++ __conda_exe shell.posix activate
++++ /fsx/nouamane/miniconda/bin/conda shell.posix activate
+++ ask_conda='. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_2='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\'''
+++ eval '. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_2='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\'''
++++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh
+++++ test -n ''
+++++ unset XML_CATALOG_FILES
+++++ unset xml_catalog_files_libxml2
++++ PS1='(base) '
++++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++++ export CONDA_PREFIX=/fsx/nouamane/miniconda
++++ CONDA_PREFIX=/fsx/nouamane/miniconda
++++ export CONDA_SHLVL=3
++++ CONDA_SHLVL=3
++++ export CONDA_DEFAULT_ENV=base
++++ CONDA_DEFAULT_ENV=base
++++ export 'CONDA_PROMPT_MODIFIER=(base) '
++++ CONDA_PROMPT_MODIFIER='(base) '
++++ export CONDA_PREFIX_2=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ CONDA_PREFIX_2=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ __conda_hashr
+++ '[' -n '' ']'
+++ '[' -n '' ']'
+++ hash -r
++ conda activate 2-1-cu121
++ local cmd=activate
++ case "$cmd" in
++ __conda_activate activate 2-1-cu121
++ '[' -n '' ']'
++ local ask_conda
+++ PS1='(base) '
+++ __conda_exe shell.posix activate 2-1-cu121
+++ /fsx/nouamane/miniconda/bin/conda shell.posix activate 2-1-cu121
++ ask_conda='PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''4'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_3='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
++ eval 'PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''4'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_3='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
+++ PS1='(2-1-cu121) '
+++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ export CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ export CONDA_SHLVL=4
+++ CONDA_SHLVL=4
+++ export CONDA_DEFAULT_ENV=2-1-cu121
+++ CONDA_DEFAULT_ENV=2-1-cu121
+++ export 'CONDA_PROMPT_MODIFIER=(2-1-cu121) '
+++ CONDA_PROMPT_MODIFIER='(2-1-cu121) '
+++ export CONDA_PREFIX_3=/fsx/nouamane/miniconda
+++ CONDA_PREFIX_3=/fsx/nouamane/miniconda
+++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ export _CE_M=
+++ _CE_M=
+++ export _CE_CONDA=
+++ _CE_CONDA=
+++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh
++++ test -n ''
++++ xml_catalog_files_libxml2=
++++ XML_CATALOG_FILES=
++++ conda_catalog_files=
++++ ifs_libxml2=' 	
+'
++++ IFS=' '
++++ rem=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ for pre in ${rem}
++++ test '' = /fsx/nouamane/miniconda/envs/2-1-cu121
++++ conda_catalog_files=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ rem=
++++ IFS=' 	
+'
++++ conda_catalog_files='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ export 'XML_CATALOG_FILES=file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ XML_CATALOG_FILES='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ unset conda_catalog_files ifs_libxml2 rem
++ __conda_hashr
++ '[' -n '' ']'
++ '[' -n '' ']'
++ hash -r
++ export PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++ PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ scontrol show hostnames 'ip-26-0-161-[138,142,153,178]'
++ export 'NODELIST=ip-26-0-161-138
+ip-26-0-161-142
+ip-26-0-161-153
+ip-26-0-161-178'
++ NODELIST='ip-26-0-161-138
+ip-26-0-161-142
+ip-26-0-161-153
+ip-26-0-161-178'
+++ head -n1
+++ scontrol show hostnames 'ip-26-0-161-[138,142,153,178]'
++ export MASTER_NODE=ip-26-0-161-138
++ MASTER_NODE=ip-26-0-161-138
++ export MASTER_PORT=12356
++ MASTER_PORT=12356
++ export NNODES=4
++ NNODES=4
++ export GPUS_PER_NODE=8
++ GPUS_PER_NODE=8
++ export WORLD_SIZE=32
++ WORLD_SIZE=32
++ export CUDA_DEVICE_MAX_CONNECTIONS=1
++ CUDA_DEVICE_MAX_CONNECTIONS=1
++ export NANOTRON_BENCHMARK=1
++ NANOTRON_BENCHMARK=1
++ echo 'Master node: ip-26-0-161-138'
+Master node: ip-26-0-161-138
++ echo 'All nodes: ip-26-0-161-138
+ip-26-0-161-142
+ip-26-0-161-153
+ip-26-0-161-178'
+All nodes: ip-26-0-161-138
+ip-26-0-161-142
+ip-26-0-161-153
+ip-26-0-161-178
++ echo 'World size: 32'
+World size: 32
++ srun torchrun --nnodes=4 --nproc_per_node=8 --rdzv_id=13265746 --rdzv_backend=c10d --rdzv_endpoint=ip-26-0-161-138:12356 run_train.py --config-file benchmark/configs/config_1.34G_dp8_tp4_pp1_acc4_mbs16_seq8192_zero1_tpmodeRED_vocab131k.yaml
+[2024-12-10 06:03:14,500] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-10 06:03:14,500] torch.distributed.run: [WARNING] 
+[2024-12-10 06:03:14,500] torch.distributed.run: [WARNING] *****************************************
+[2024-12-10 06:03:14,500] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-10 06:03:14,500] torch.distributed.run: [WARNING] *****************************************
+[2024-12-10 06:03:14,508] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-10 06:03:14,516] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-10 06:03:14,516] torch.distributed.run: [WARNING] 
+[2024-12-10 06:03:14,516] torch.distributed.run: [WARNING] *****************************************
+[2024-12-10 06:03:14,516] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-10 06:03:14,516] torch.distributed.run: [WARNING] *****************************************
+[2024-12-10 06:03:14,508] torch.distributed.run: [WARNING] 
+[2024-12-10 06:03:14,508] torch.distributed.run: [WARNING] *****************************************
+[2024-12-10 06:03:14,508] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-10 06:03:14,508] torch.distributed.run: [WARNING] *****************************************
+[2024-12-10 06:03:14,614] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-10 06:03:14,614] torch.distributed.run: [WARNING] 
+[2024-12-10 06:03:14,614] torch.distributed.run: [WARNING] *****************************************
+[2024-12-10 06:03:14,614] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-10 06:03:14,614] torch.distributed.run: [WARNING] *****************************************
+12/10/2024 06:03:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]: Config:
+12/10/2024 06:03:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]: Config(general=GeneralArgs(project='debug',
+12/10/2024 06:03:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                            run='1.34G_dp8_tp4_pp1_acc4_mbs16_seq8192_zero1_tpmodeRED_vocab131k',
+12/10/2024 06:03:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                            seed=42,
+12/10/2024 06:03:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                            step=None,
+12/10/2024 06:03:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                            consumed_train_samples=None,
+12/10/2024 06:03:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                            benchmark_csv_path=PosixPath('benchmark/results/bench_final.csv'),
+12/10/2024 06:03:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                            ignore_sanity_checks=True),
+12/10/2024 06:03:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:        parallelism=ParallelismArgs(dp=8,
+12/10/2024 06:03:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                                    pp=1,
+12/10/2024 06:03:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                                    tp=4,
+12/10/2024 06:03:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                                    pp_engine=<nanotron.parallel.pipeline_parallel.engine.OneForwardOneBackwardPipelineEngine object at 0x7f0069b04af0>,
+12/10/2024 06:03:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                                    tp_mode=<TensorParallelLinearMode.REDUCE_SCATTER: 2>,
+12/10/2024 06:03:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                                    tp_linear_async_communication=True,
+12/10/2024 06:03:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                                    recompute_layer=False,
+12/10/2024 06:03:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                                    tp_recompute_allgather=True,
+12/10/2024 06:03:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                                    expert_parallel_size=1),
+12/10/2024 06:03:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:        model=ModelArgs(model_config=LlamaConfig(bos_token_id=0,
+12/10/2024 06:03:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                                                 eos_token_id=0,
+12/10/2024 06:03:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                                                 hidden_act='silu',
+12/10/2024 06:03:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                                                 hidden_size=2048,
+12/10/2024 06:03:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                                                 initializer_range=0.02,
+12/10/2024 06:03:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                                                 intermediate_size=8192,
+12/10/2024 06:03:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                                                 is_llama_config=True,
+12/10/2024 06:03:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                                                 max_position_embeddings=8192,
+12/10/2024 06:03:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                                                 num_attention_heads=32,
+12/10/2024 06:03:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                                                 num_hidden_layers=16,
+12/10/2024 06:03:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                                                 num_key_value_heads=8,
+12/10/2024 06:03:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                                                 pad_token_id=None,
+12/10/2024 06:03:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                                                 pretraining_tp=1,
+12/10/2024 06:03:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                                                 rms_norm_eps=1e-05,
+12/10/2024 06:03:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                                                 rope_scaling=None,
+12/10/2024 06:03:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                                                 rope_theta=10000.0,
+12/10/2024 06:03:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                                                 rope_interleaved=False,
+12/10/2024 06:03:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                                                 tie_word_embeddings=True,
+12/10/2024 06:03:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                                                 use_cache=True,
+12/10/2024 06:03:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                                                 vocab_size=131072),
+12/10/2024 06:03:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                        init_method=RandomInit(std=0.02),
+12/10/2024 06:03:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                        dtype=torch.bfloat16,
+12/10/2024 06:03:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                        make_vocab_size_divisible_by=1,
+12/10/2024 06:03:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                        ddp_bucket_cap_mb=25),
+12/10/2024 06:03:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:        tokenizer=TokenizerArgs(tokenizer_name_or_path='robot-test/dummy-tokenizer-wordlevel',
+12/10/2024 06:03:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                                tokenizer_revision=None,
+12/10/2024 06:03:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                                tokenizer_max_length=None),
+12/10/2024 06:03:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:        checkpoints=CheckpointsArgs(checkpoints_path=PosixPath('checkpoints'),
+12/10/2024 06:03:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                                    checkpoint_interval=10000,
+12/10/2024 06:03:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                                    save_initial_state=False,
+12/10/2024 06:03:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                                    save_final_state=False,
+12/10/2024 06:03:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                                    resume_checkpoint_path=None,
+12/10/2024 06:03:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                                    checkpoints_path_is_shared_file_system=False),
+12/10/2024 06:03:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:        logging=LoggingArgs(log_level='info',
+12/10/2024 06:03:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                            log_level_replica='info',
+12/10/2024 06:03:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                            iteration_step_info_interval=1),
+12/10/2024 06:03:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:        tokens=TokensArgs(sequence_length=8192,
+12/10/2024 06:03:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                          train_steps=100,
+12/10/2024 06:03:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                          micro_batch_size=16,
+12/10/2024 06:03:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                          batch_accumulation_per_replica=4,
+12/10/2024 06:03:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                          val_check_interval=100,
+12/10/2024 06:03:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                          limit_val_batches=0,
+12/10/2024 06:03:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                          limit_test_batches=0),
+12/10/2024 06:03:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:        optimizer=OptimizerArgs(optimizer_factory=AdamWOptimizerArgs(adam_eps=1e-08,
+12/10/2024 06:03:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                                                                     adam_beta1=0.9,
+12/10/2024 06:03:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                                                                     adam_beta2=0.95,
+12/10/2024 06:03:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                                                                     torch_adam_is_fused=True,
+12/10/2024 06:03:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                                                                     name='adamW'),
+12/10/2024 06:03:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                                zero_stage=1,
+12/10/2024 06:03:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                                weight_decay=0.01,
+12/10/2024 06:03:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                                clip_grad=1.0,
+12/10/2024 06:03:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                                accumulate_grad_in_fp32=True,
+12/10/2024 06:03:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                                learning_rate_scheduler=LRSchedulerArgs(learning_rate=0.0003,
+12/10/2024 06:03:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                                                                        lr_warmup_steps=2,
+12/10/2024 06:03:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                                                                        lr_warmup_style='linear',
+12/10/2024 06:03:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                                                                        lr_decay_style='cosine',
+12/10/2024 06:03:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                                                                        lr_decay_steps=13,
+12/10/2024 06:03:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                                                                        lr_decay_starting_step=None,
+12/10/2024 06:03:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                                                                        min_decay_lr=1e-05)),
+12/10/2024 06:03:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:        data_stages=[DatasetStageArgs(name='Stable Training Stage',
+12/10/2024 06:03:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                                      start_training_step=1,
+12/10/2024 06:03:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                                      data=DataArgs(dataset=None,
+12/10/2024 06:03:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                                                    seed=42,
+12/10/2024 06:03:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                                                    num_loading_workers=1))],
+12/10/2024 06:03:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:        profiler=None,
+12/10/2024 06:03:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:        lighteval=None,
+12/10/2024 06:03:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:        s3_upload=None)
+12/10/2024 06:03:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]: Model Config:
+12/10/2024 06:03:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]: LlamaConfig(bos_token_id=0,
+12/10/2024 06:03:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:             eos_token_id=0,
+12/10/2024 06:03:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:             hidden_act='silu',
+12/10/2024 06:03:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:             hidden_size=2048,
+12/10/2024 06:03:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:             initializer_range=0.02,
+12/10/2024 06:03:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:             intermediate_size=8192,
+12/10/2024 06:03:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:             is_llama_config=True,
+12/10/2024 06:03:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:             max_position_embeddings=8192,
+12/10/2024 06:03:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:             num_attention_heads=32,
+12/10/2024 06:03:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:             num_hidden_layers=16,
+12/10/2024 06:03:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:             num_key_value_heads=8,
+12/10/2024 06:03:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:             pad_token_id=None,
+12/10/2024 06:03:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:             pretraining_tp=1,
+12/10/2024 06:03:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:             rms_norm_eps=1e-05,
+12/10/2024 06:03:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:             rope_scaling=None,
+12/10/2024 06:03:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:             rope_theta=10000.0,
+12/10/2024 06:03:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:             rope_interleaved=False,
+12/10/2024 06:03:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:             tie_word_embeddings=True,
+12/10/2024 06:03:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:             use_cache=True,
+12/10/2024 06:03:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:             vocab_size=131072)
+12/10/2024 06:03:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]: Building model..
+12/10/2024 06:03:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]: Initialize RoPE Theta = 10000.0
+12/10/2024 06:03:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]: Setting PP block ranks...
+12/10/2024 06:04:01 [INFO|DP=0|PP=0|TP=1|ip-26-0-161-138]: Local number of parameters: 310M (592.13MiB)
+12/10/2024 06:04:01 [INFO|DP=0|PP=0|TP=3|ip-26-0-161-138]: Local number of parameters: 310M (592.13MiB)
+12/10/2024 06:04:01 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]: Total number of parameters: 1.24G (2368.52MiB)
+12/10/2024 06:04:01 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]: Local number of parameters: 310M (592.13MiB)
+12/10/2024 06:04:01 [INFO|DP=0|PP=0|TP=2|ip-26-0-161-138]: Local number of parameters: 310M (592.13MiB)
+12/10/2024 06:04:01 [INFO|DP=0|PP=0|TP=1|ip-26-0-161-138]: [After model building] Memory usage: 594.15MiB. Peak allocated: 5440.00MiB Peak reserved: 34114.00MiB
+12/10/2024 06:04:01 [INFO|DP=0|PP=0|TP=3|ip-26-0-161-138]: [After model building] Memory usage: 594.15MiB. Peak allocated: 5440.00MiB Peak reserved: 34114.00MiB
+12/10/2024 06:04:01 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]: [After model building] Memory usage: 594.15MiB. Peak allocated: 5440.00MiB Peak reserved: 34114.00MiB
+12/10/2024 06:04:01 [INFO|DP=0|PP=0|TP=2|ip-26-0-161-138]: [After model building] Memory usage: 594.15MiB. Peak allocated: 5440.00MiB Peak reserved: 35138.00MiB
+12/10/2024 06:04:01 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]: No checkpoint path provided.
+12/10/2024 06:04:01 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]: Parametrizing model parameters using StandardParametrizator
+12/10/2024 06:04:04 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]: [Optimizer Building] Using LearningRateForSP as learning rate
+12/10/2024 06:04:04 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]: [ZeRO sharding] Size of optimizer params per rank:
+12/10/2024 06:04:04 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]: [ZeRO sharding] DP Rank 0 has 38.8M out of 310M (12.50%) params' optimizer states
+12/10/2024 06:04:04 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]: [ZeRO sharding] DP Rank 1 has 38.8M out of 310M (12.50%) params' optimizer states
+12/10/2024 06:04:04 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]: [ZeRO sharding] DP Rank 2 has 38.8M out of 310M (12.50%) params' optimizer states
+12/10/2024 06:04:04 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]: [ZeRO sharding] DP Rank 3 has 38.8M out of 310M (12.50%) params' optimizer states
+12/10/2024 06:04:04 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]: [ZeRO sharding] DP Rank 4 has 38.8M out of 310M (12.50%) params' optimizer states
+12/10/2024 06:04:04 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]: [ZeRO sharding] DP Rank 5 has 38.8M out of 310M (12.50%) params' optimizer states
+12/10/2024 06:04:04 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]: [ZeRO sharding] DP Rank 6 has 38.8M out of 310M (12.50%) params' optimizer states
+12/10/2024 06:04:04 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]: [ZeRO sharding] DP Rank 7 has 38.8M out of 310M (12.50%) params' optimizer states
+12/10/2024 06:04:04 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]: [Training Plan] Stage Stable Training Stage has 99 remaining training steps and has consumed 0 samples
+12/10/2024 06:04:04 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]: Using dummy data generator
+12/10/2024 06:04:04 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]: [Training Plan] There are 1 training stages 
+12/10/2024 06:04:04 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]: [Stage Stable Training Stage] start from step 1 
+12/10/2024 06:04:04 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]: 
+12/10/2024 06:04:04 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]: [Start training] datetime: 2024-12-10 06:04:04.901719 | mbs: 16 | grad_accum: 4 | global_batch_size: 512 | sequence_length: 8192 | train_steps: 100 | start_iteration_step: 0 | consumed_train_samples: 0
+wandb: WARNING Path /fsx/nouamane/.cache/wandb/wandb/ wasn't writable, using system temp directory.
+wandb: WARNING Path /fsx/nouamane/.cache/wandb/wandb/ wasn't writable, using system temp directory
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    module_builder=lambda: lambda x: x.float(),
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 6 has a total capacty of 79.33 GiB of which 12.39 GiB is free. Including non-PyTorch memory, this process has 66.92 GiB memory in use. Of the allocated memory 55.75 GiB is allocated by PyTorch, and 702.03 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    module_builder=lambda: lambda x: x.float(),
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 7 has a total capacty of 79.33 GiB of which 12.76 GiB is free. Including non-PyTorch memory, this process has 66.56 GiB memory in use. Of the allocated memory 55.75 GiB is allocated by PyTorch, and 766.03 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    module_builder=lambda: lambda x: x.float(),
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 5 has a total capacty of 79.33 GiB of which 12.39 GiB is free. Including non-PyTorch memory, this process has 66.92 GiB memory in use. Of the allocated memory 55.75 GiB is allocated by PyTorch, and 702.03 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    module_builder=lambda: lambda x: x.float(),
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 4 has a total capacty of 79.33 GiB of which 12.55 GiB is free. Including non-PyTorch memory, this process has 66.77 GiB memory in use. Of the allocated memory 55.75 GiB is allocated by PyTorch, and 574.03 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    module_builder=lambda: lambda x: x.float(),
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 7 has a total capacty of 79.33 GiB of which 12.35 GiB is free. Including non-PyTorch memory, this process has 66.97 GiB memory in use. Of the allocated memory 55.75 GiB is allocated by PyTorch, and 510.03 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    module_builder=lambda: lambda x: x.float(),
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 4 has a total capacty of 79.33 GiB of which 11.75 GiB is free. Including non-PyTorch memory, this process has 67.57 GiB memory in use. Of the allocated memory 55.75 GiB is allocated by PyTorch, and 638.03 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    module_builder=lambda: lambda x: x.float(),
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 6 has a total capacty of 79.33 GiB of which 11.65 GiB is free. Including non-PyTorch memory, this process has 67.67 GiB memory in use. Of the allocated memory 55.75 GiB is allocated by PyTorch, and 702.03 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    module_builder=lambda: lambda x: x.float(),
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 2 has a total capacty of 79.33 GiB of which 11.40 GiB is free. Including non-PyTorch memory, this process has 67.92 GiB memory in use. Of the allocated memory 55.75 GiB is allocated by PyTorch, and 958.03 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+        outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)output = self.pp_block(**new_kwargs)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    module_builder=lambda: lambda x: x.float(),
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 0 has a total capacty of 79.33 GiB of which 11.98 GiB is free. Including non-PyTorch memory, this process has 67.34 GiB memory in use. Of the allocated memory 55.75 GiB is allocated by PyTorch, and 446.03 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    module_builder=lambda: lambda x: x.float(),
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 5 has a total capacty of 79.33 GiB of which 11.79 GiB is free. Including non-PyTorch memory, this process has 67.53 GiB memory in use. Of the allocated memory 55.75 GiB is allocated by PyTorch, and 638.03 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    module_builder=lambda: lambda x: x.float(),
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 1 has a total capacty of 79.33 GiB of which 11.79 GiB is free. Including non-PyTorch memory, this process has 67.53 GiB memory in use. Of the allocated memory 55.75 GiB is allocated by PyTorch, and 638.03 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    module_builder=lambda: lambda x: x.float(),
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 3 has a total capacty of 79.33 GiB of which 11.66 GiB is free. Including non-PyTorch memory, this process has 67.66 GiB memory in use. Of the allocated memory 55.75 GiB is allocated by PyTorch, and 894.03 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    trainer.train(dataloader)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    module_builder=lambda: lambda x: x.float(),
+    output = self.pp_block(**new_kwargs)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 7 has a total capacty of 79.33 GiB of which 12.70 GiB is free. Including non-PyTorch memory, this process has 66.62 GiB memory in use. Of the allocated memory 55.75 GiB is allocated by PyTorch, and 830.03 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    module_builder=lambda: lambda x: x.float(),
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 2 has a total capacty of 79.33 GiB of which 12.04 GiB is free. Including non-PyTorch memory, this process has 67.28 GiB memory in use. Of the allocated memory 55.75 GiB is allocated by PyTorch, and 382.03 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    module_builder=lambda: lambda x: x.float(),
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 0 has a total capacty of 79.33 GiB of which 12.91 GiB is free. Including non-PyTorch memory, this process has 66.41 GiB memory in use. Of the allocated memory 55.75 GiB is allocated by PyTorch, and 254.03 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    module_builder=lambda: lambda x: x.float(),
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 5 has a total capacty of 79.33 GiB of which 12.39 GiB is free. Including non-PyTorch memory, this process has 66.92 GiB memory in use. Of the allocated memory 55.75 GiB is allocated by PyTorch, and 702.03 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    module_builder=lambda: lambda x: x.float(),
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 6 has a total capacty of 79.33 GiB of which 12.46 GiB is free. Including non-PyTorch memory, this process has 66.86 GiB memory in use. Of the allocated memory 55.75 GiB is allocated by PyTorch, and 638.03 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    outputs = self.pipeline_engine.train_batch_iter(
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    module_builder=lambda: lambda x: x.float(),
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 2 has a total capacty of 79.33 GiB of which 12.39 GiB is free. Including non-PyTorch memory, this process has 66.92 GiB memory in use. Of the allocated memory 55.75 GiB is allocated by PyTorch, and 702.03 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    module_builder=lambda: lambda x: x.float(),
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 4 has a total capacty of 79.33 GiB of which 12.61 GiB is free. Including non-PyTorch memory, this process has 66.71 GiB memory in use. Of the allocated memory 55.75 GiB is allocated by PyTorch, and 510.03 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    module_builder=lambda: lambda x: x.float(),
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 3 has a total capacty of 79.33 GiB of which 12.26 GiB is free. Including non-PyTorch memory, this process has 67.06 GiB memory in use. Of the allocated memory 55.75 GiB is allocated by PyTorch, and 958.03 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    module_builder=lambda: lambda x: x.float(),
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 1 has a total capacty of 79.33 GiB of which 12.52 GiB is free. Including non-PyTorch memory, this process has 66.80 GiB memory in use. Of the allocated memory 55.75 GiB is allocated by PyTorch, and 574.03 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    module_builder=lambda: lambda x: x.float(),
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 1 has a total capacty of 79.33 GiB of which 11.96 GiB is free. Including non-PyTorch memory, this process has 67.35 GiB memory in use. Of the allocated memory 55.75 GiB is allocated by PyTorch, and 382.03 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    module_builder=lambda: lambda x: x.float(),
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 0 has a total capacty of 79.33 GiB of which 11.93 GiB is free. Including non-PyTorch memory, this process has 67.38 GiB memory in use. Of the allocated memory 55.75 GiB is allocated by PyTorch, and 574.03 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    module_builder=lambda: lambda x: x.float(),
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 3 has a total capacty of 79.33 GiB of which 11.77 GiB is free. Including non-PyTorch memory, this process has 67.55 GiB memory in use. Of the allocated memory 55.75 GiB is allocated by PyTorch, and 702.03 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    module_builder=lambda: lambda x: x.float(),
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 5 has a total capacty of 79.33 GiB of which 11.53 GiB is free. Including non-PyTorch memory, this process has 67.79 GiB memory in use. Of the allocated memory 55.75 GiB is allocated by PyTorch, and 830.03 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    module_builder=lambda: lambda x: x.float(),
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 4 has a total capacty of 79.33 GiB of which 11.39 GiB is free. Including non-PyTorch memory, this process has 67.93 GiB memory in use. Of the allocated memory 55.75 GiB is allocated by PyTorch, and 1.06 GiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 250, in <module>
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 456, in train
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 493, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 278, in train_batch_iter
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 44, in forward
+    module_builder=lambda: lambda x: x.float(),
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 7 has a total capacty of 79.33 GiB of which 12.27 GiB is free. Including non-PyTorch memory, this process has 67.05 GiB memory in use. Of the allocated memory 55.75 GiB is allocated by PyTorch, and 510.03 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 909, in forward_with_hidden_states
+    fp32_sharded_logits = self.cast_to_fp32(x=sharded_logits)["output"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 876, in <lambda>
+    module_builder=lambda: lambda x: x.float(),
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 GiB. GPU 6 has a total capacty of 79.33 GiB of which 11.86 GiB is free. Including non-PyTorch memory, this process has 67.46 GiB memory in use. Of the allocated memory 55.75 GiB is allocated by PyTorch, and 574.03 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+[2024-12-10 06:04:11,661] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 585246 closing signal SIGTERM
+[2024-12-10 06:04:11,662] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 585247 closing signal SIGTERM
+[2024-12-10 06:04:11,662] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 585248 closing signal SIGTERM
+[2024-12-10 06:04:11,663] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 585249 closing signal SIGTERM
+[2024-12-10 06:04:11,663] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 502277 closing signal SIGTERM
+[2024-12-10 06:04:11,663] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 502279 closing signal SIGTERM
+[2024-12-10 06:04:11,663] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 502280 closing signal SIGTERM
+[2024-12-10 06:04:11,663] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 502281 closing signal SIGTERM
+[2024-12-10 06:04:11,663] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 502282 closing signal SIGTERM
+[2024-12-10 06:04:11,663] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 502283 closing signal SIGTERM
+[2024-12-10 06:04:12,905] torch.distributed.elastic.multiprocessing.api: [ERROR] failed (exitcode: 1) local_rank: 1 (pid: 502278) of binary: /fsx/nouamane/miniconda/envs/2-1-cu121/bin/python
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 33, in <module>
+    sys.exit(load_entry_point('torch==2.1.1', 'console_scripts', 'torchrun')())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 264, in launch_agent
+    raise ChildFailedError(
+torch.distributed.elastic.multiprocessing.errors.ChildFailedError: 
+============================================================
+run_train.py FAILED
+------------------------------------------------------------
+Failures:
+[1]:
+  time      : 2024-12-10_06:04:11
+  host      : ip-26-0-161-178.ec2.internal
+  rank      : 31 (local_rank: 7)
+  exitcode  : 1 (pid: 502284)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+------------------------------------------------------------
+Root Cause (first observed failure):
+[0]:
+  time      : 2024-12-10_06:04:11
+  host      : ip-26-0-161-178.ec2.internal
+  rank      : 25 (local_rank: 1)
+  exitcode  : 1 (pid: 502278)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+============================================================
+srun: error: ip-26-0-161-178: task 3: Exited with exit code 1
+[2024-12-10 06:04:15,395] torch.distributed.elastic.multiprocessing.api: [ERROR] failed (exitcode: 1) local_rank: 4 (pid: 585250) of binary: /fsx/nouamane/miniconda/envs/2-1-cu121/bin/python
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 33, in <module>
+    sys.exit(load_entry_point('torch==2.1.1', 'console_scripts', 'torchrun')())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 264, in launch_agent
+    raise ChildFailedError(
+torch.distributed.elastic.multiprocessing.errors.ChildFailedError: 
+============================================================
+run_train.py FAILED
+------------------------------------------------------------
+Failures:
+[1]:
+  time      : 2024-12-10_06:04:11
+  host      : ip-26-0-161-138.ec2.internal
+  rank      : 5 (local_rank: 5)
+  exitcode  : 1 (pid: 585251)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[2]:
+  time      : 2024-12-10_06:04:11
+  host      : ip-26-0-161-138.ec2.internal
+  rank      : 6 (local_rank: 6)
+  exitcode  : 1 (pid: 585252)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[3]:
+  time      : 2024-12-10_06:04:11
+  host      : ip-26-0-161-138.ec2.internal
+  rank      : 7 (local_rank: 7)
+  exitcode  : 1 (pid: 585253)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+------------------------------------------------------------
+Root Cause (first observed failure):
+[0]:
+  time      : 2024-12-10_06:04:11
+  host      : ip-26-0-161-138.ec2.internal
+  rank      : 4 (local_rank: 4)
+  exitcode  : 1 (pid: 585250)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+============================================================
+[2024-12-10 06:04:15,735] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-161-153.ec2.internal_77017_0' has failed to send a keep-alive heartbeat to the rendezvous '13265746' due to an error of type RendezvousConnectionError.
+[2024-12-10 06:04:16,609] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-161-142.ec2.internal_90723_0' has failed to send a keep-alive heartbeat to the rendezvous '13265746' due to an error of type RendezvousConnectionError.
+[2024-12-10 06:04:16,667] torch.distributed.elastic.multiprocessing.api: [ERROR] failed (exitcode: 1) local_rank: 0 (pid: 77027) of binary: /fsx/nouamane/miniconda/envs/2-1-cu121/bin/python
+[2024-12-10 06:04:16,678] torch.distributed.elastic.multiprocessing.api: [ERROR] failed (exitcode: 1) local_rank: 0 (pid: 90732) of binary: /fsx/nouamane/miniconda/envs/2-1-cu121/bin/python
+[2024-12-10 06:04:16,728] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-161-142.ec2.internal_90723_0' has failed to shutdown the rendezvous '13265746' due to an error of type RendezvousConnectionError.
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 33, in <module>
+    sys.exit(load_entry_point('torch==2.1.1', 'console_scripts', 'torchrun')())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+[2024-12-10 06:04:16,729] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-161-153.ec2.internal_77017_0' has failed to shutdown the rendezvous '13265746' due to an error of type RendezvousConnectionError.
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 33, in <module>
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    sys.exit(load_entry_point('torch==2.1.1', 'console_scripts', 'torchrun')())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 264, in launch_agent
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    raise ChildFailedError(
+torch.distributed.elastic.multiprocessing.errors.ChildFailedError: 
+============================================================
+run_train.py FAILED
+------------------------------------------------------------
+Failures:
+[1]:
+  time      : 2024-12-10_06:04:16
+  host      : ip-26-0-161-142.ec2.internal
+  rank      : 9 (local_rank: 1)
+  exitcode  : 1 (pid: 90733)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[2]:
+  time      : 2024-12-10_06:04:16
+  host      : ip-26-0-161-142.ec2.internal
+  rank      : 10 (local_rank: 2)
+  exitcode  : 1 (pid: 90734)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[3]:
+  time      : 2024-12-10_06:04:16
+  host      : ip-26-0-161-142.ec2.internal
+  rank      : 11 (local_rank: 3)
+  exitcode  : 1 (pid: 90735)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[4]:
+  time      : 2024-12-10_06:04:16
+  host      : ip-26-0-161-142.ec2.internal
+  rank      : 12 (local_rank: 4)
+  exitcode  : 1 (pid: 90736)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[5]:
+  time      : 2024-12-10_06:04:16
+  host      : ip-26-0-161-142.ec2.internal
+  rank      : 13 (local_rank: 5)
+  exitcode  : 1 (pid: 90737)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[6]:
+  time      : 2024-12-10_06:04:16
+  host      : ip-26-0-161-142.ec2.internal
+  rank      : 14 (local_rank: 6)
+  exitcode  : 1 (pid: 90738)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[7]:
+  time      : 2024-12-10_06:04:16
+  host      : ip-26-0-161-142.ec2.internal
+  rank      : 15 (local_rank: 7)
+  exitcode  : 1 (pid: 90739)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+------------------------------------------------------------
+Root Cause (first observed failure):
+[0]:
+  time      : 2024-12-10_06:04:16
+  host      : ip-26-0-161-142.ec2.internal
+  rank      : 8 (local_rank: 0)
+  exitcode  : 1 (pid: 90732)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+============================================================
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 264, in launch_agent
+    raise ChildFailedError(
+torch.distributed.elastic.multiprocessing.errors.ChildFailedError: 
+============================================================
+run_train.py FAILED
+------------------------------------------------------------
+Failures:
+[1]:
+  time      : 2024-12-10_06:04:16
+  host      : ip-26-0-161-153.ec2.internal
+  rank      : 17 (local_rank: 1)
+  exitcode  : 1 (pid: 77028)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[2]:
+  time      : 2024-12-10_06:04:16
+  host      : ip-26-0-161-153.ec2.internal
+  rank      : 18 (local_rank: 2)
+  exitcode  : 1 (pid: 77029)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[3]:
+  time      : 2024-12-10_06:04:16
+  host      : ip-26-0-161-153.ec2.internal
+  rank      : 19 (local_rank: 3)
+  exitcode  : 1 (pid: 77030)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[4]:
+  time      : 2024-12-10_06:04:16
+  host      : ip-26-0-161-153.ec2.internal
+  rank      : 20 (local_rank: 4)
+  exitcode  : 1 (pid: 77031)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[5]:
+  time      : 2024-12-10_06:04:16
+  host      : ip-26-0-161-153.ec2.internal
+  rank      : 21 (local_rank: 5)
+  exitcode  : 1 (pid: 77032)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[6]:
+  time      : 2024-12-10_06:04:16
+  host      : ip-26-0-161-153.ec2.internal
+  rank      : 22 (local_rank: 6)
+  exitcode  : 1 (pid: 77033)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[7]:
+  time      : 2024-12-10_06:04:16
+  host      : ip-26-0-161-153.ec2.internal
+  rank      : 23 (local_rank: 7)
+  exitcode  : 1 (pid: 77034)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+------------------------------------------------------------
+Root Cause (first observed failure):
+[0]:
+  time      : 2024-12-10_06:04:16
+  host      : ip-26-0-161-153.ec2.internal
+  rank      : 16 (local_rank: 0)
+  exitcode  : 1 (pid: 77027)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+============================================================
+srun: error: ip-26-0-161-138: task 0: Exited with exit code 1
+srun: error: ip-26-0-161-153: task 2: Exited with exit code 1
+srun: error: ip-26-0-161-142: task 1: Exited with exit code 1
diff --git a/logs/13265919-bench_1.34G_dp32_tp4_pp1_acc8_mbs2_seq2048_zero1_tpmodeALL_vocab131k.out b/logs/13265919-bench_1.34G_dp32_tp4_pp1_acc8_mbs2_seq2048_zero1_tpmodeALL_vocab131k.out
new file mode 100644
index 0000000000000000000000000000000000000000..d823246b09a4e313a03281b2708fe9e728941bde
--- /dev/null
+++ b/logs/13265919-bench_1.34G_dp32_tp4_pp1_acc8_mbs2_seq2048_zero1_tpmodeALL_vocab131k.out
@@ -0,0 +1,1272 @@
++ source /etc/profile.d/modules.sh
+++ . /usr/share/modules/init/bash
++++ unset _mlshdbg
++++ '[' 0 = 1 ']'
++++ unset _mlre _mlIFS
++++ '[' -n x ']'
++++ _mlIFS=' 	
+'
++++ IFS=' '
++++ '[' -n '' ']'
+++++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash autoinit
++++ _mlcode='module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+MODULEPATH_modshare=/etc/environment-modules/modules:1:/usr/share/modules/\$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1; export MODULEPATH_modshare;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+LOADEDMODULES=; export LOADEDMODULES;
+MODULEPATH=/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/\$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles; export MODULEPATH;
+test 0;'
++++ _mlret=0
++++ '[' -n x ']'
++++ IFS=' 	
+'
++++ unset _mlIFS
++++ unset _mlre _mlv _mlrv
++++ '[' 0 -eq 0 ']'
++++ eval 'module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+MODULEPATH_modshare=/etc/environment-modules/modules:1:/usr/share/modules/\$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1; export MODULEPATH_modshare;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+LOADEDMODULES=; export LOADEDMODULES;
+MODULEPATH=/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/\$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles; export MODULEPATH;
+test 0;'
+++++ MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl
+++++ export MODULES_CMD
+++++ ENV=/usr/share/modules/init/profile.sh
+++++ export ENV
+++++ MODULEPATH_modshare='/etc/environment-modules/modules:1:/usr/share/modules/$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1'
+++++ export MODULEPATH_modshare
+++++ BASH_ENV=/usr/share/modules/init/bash
+++++ export BASH_ENV
+++++ MODULESHOME=/usr/share/modules
+++++ export MODULESHOME
+++++ LOADEDMODULES=
+++++ export LOADEDMODULES
+++++ MODULEPATH='/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles'
+++++ export MODULEPATH
+++++ test 0
++++ '[' 0 = 1 ']'
++++ '[' -t 2 ']'
++++ export -f module
++++ export -f switchml
++++ '[' 5 -ge 3 ']'
++++ [[ ehxB =~ i ]]
++++ [[ ! :/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin: =~ :/usr/bin: ]]
++++ '[' '!' -n '' ']'
++++ MANPATH=:
++++ export MANPATH
+++++ manpath
++++ [[ ! :/admin/home/nouamane/.local/share/man:/fsx/nouamane/miniconda/envs/2-1-cu121/man:/fsx/nouamane/miniconda/envs/2-1-cu121/share/man:/fsx/nouamane/miniconda/man:/fsx/nouamane/miniconda/share/man:/opt/amazon/openmpi/share/man:/opt/amazon/efa/share/man:/opt/slurm/share/man:/usr/local/man:/usr/local/share/man:/usr/share/man:/admin/home/nouamane/.fzf/man:: =~ :/usr/share/man: ]]
++++ unset _mlcode _mlret
++++ '[' -n '' ']'
++ module load cuda/12.1
++ unset _mlshdbg
++ '[' 0 = 1 ']'
++ unset _mlre _mlIFS
++ '[' -n x ']'
++ _mlIFS=' 	
+'
++ IFS=' '
++ '[' -n '' ']'
+++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash load cuda/12.1
++ eval 'CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include;' export 'CPATH;
+LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:;' export 'LD_LIBRARY_PATH;
+FI_EFA_FORK_SAFE=1;' export 'FI_EFA_FORK_SAFE;
+MANPATH=/usr/local/cuda-12.1/share/man::;' export 'MANPATH;
+LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64;' export 'LIBRARY_PATH;
+_LMFILES_=/usr/share/modules/modulefiles/cuda/12.1;' export '_LMFILES_;
+LOADEDMODULES=cuda/12.1;' export 'LOADEDMODULES;
+MPI_PATH=/opt/amazon/openmpi;' export 'MPI_PATH;
+NCCL_HOME_modshare=/opt/nccl/build:1;' export 'NCCL_HOME_modshare;
+NCCL_PROTO=simple;' export 'NCCL_PROTO;
+MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1;' export 'MANPATH_modshare;
+LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1;' export 'LIBRARY_PATH_modshare;
+NCCL_SOCKET_IFNAME=enp;' export 'NCCL_SOCKET_IFNAME;
+AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl;' export 'AWS_OFI_NCCL_HOME;
+NCCL_HOME=/opt/nccl/build;' export 'NCCL_HOME;
+FI_PROVIDER=efa;' export 'FI_PROVIDER;
+AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1;' export 'AWS_OFI_NCCL_HOME_modshare;
+CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1;' export 'CPATH_modshare;
+LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2::1:/opt/nccl/build/lib:1:/opt/aws-ofi-nccl/lib:2:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/efa/lib:1;' export 'LD_LIBRARY_PATH_modshare;
+FI_EFA_ENABLE_SHM_TRANSFER=1;' export 'FI_EFA_ENABLE_SHM_TRANSFER;
+_LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1;' export '_LMFILES__modshare;
+LOADEDMODULES_modshare=cuda/12.1:1;' export 'LOADEDMODULES_modshare;
+MPI_PATH_modshare=/opt/amazon/openmpi:1;' export 'MPI_PATH_modshare;
+PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin;' export 'PATH;
+CUDA_HOME=/usr/local/cuda-12.1;' export 'CUDA_HOME;
+PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/usr/local/cuda-12.1/include:1:/opt/amazon/efa/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/usr/sbin:1:/fsx/nouamane/miniconda/bin:1:/opt/slurm/bin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:2:/opt/amazon/openmpi/bin:1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1;' export 'PATH_modshare;
+test' '0;'
+++ CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include
+++ export CPATH
+++ LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:
+++ export LD_LIBRARY_PATH
+++ FI_EFA_FORK_SAFE=1
+++ export FI_EFA_FORK_SAFE
+++ MANPATH=/usr/local/cuda-12.1/share/man::
+++ export MANPATH
+++ LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64
+++ export LIBRARY_PATH
+++ _LMFILES_=/usr/share/modules/modulefiles/cuda/12.1
+++ export _LMFILES_
+++ LOADEDMODULES=cuda/12.1
+++ export LOADEDMODULES
+++ MPI_PATH=/opt/amazon/openmpi
+++ export MPI_PATH
+++ NCCL_HOME_modshare=/opt/nccl/build:1
+++ export NCCL_HOME_modshare
+++ NCCL_PROTO=simple
+++ export NCCL_PROTO
+++ MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1
+++ export MANPATH_modshare
+++ LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1
+++ export LIBRARY_PATH_modshare
+++ NCCL_SOCKET_IFNAME=enp
+++ export NCCL_SOCKET_IFNAME
+++ AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl
+++ export AWS_OFI_NCCL_HOME
+++ NCCL_HOME=/opt/nccl/build
+++ export NCCL_HOME
+++ FI_PROVIDER=efa
+++ export FI_PROVIDER
+++ AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1
+++ export AWS_OFI_NCCL_HOME_modshare
+++ CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1
+++ export CPATH_modshare
+++ LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2::1:/opt/nccl/build/lib:1:/opt/aws-ofi-nccl/lib:2:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/efa/lib:1
+++ export LD_LIBRARY_PATH_modshare
+++ FI_EFA_ENABLE_SHM_TRANSFER=1
+++ export FI_EFA_ENABLE_SHM_TRANSFER
+++ _LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1
+++ export _LMFILES__modshare
+++ LOADEDMODULES_modshare=cuda/12.1:1
+++ export LOADEDMODULES_modshare
+++ MPI_PATH_modshare=/opt/amazon/openmpi:1
+++ export MPI_PATH_modshare
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ export PATH
+++ CUDA_HOME=/usr/local/cuda-12.1
+++ export CUDA_HOME
+++ PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/usr/local/cuda-12.1/include:1:/opt/amazon/efa/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/usr/sbin:1:/fsx/nouamane/miniconda/bin:1:/opt/slurm/bin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:2:/opt/amazon/openmpi/bin:1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1
+++ export PATH_modshare
+++ test 0
++ _mlstatus=0
++ '[' -n x ']'
++ IFS=' 	
+'
++ unset _mlre _mlv _mlrv _mlIFS
++ '[' -n '' ']'
++ unset _mlshdbg
++ return 0
++ source /fsx/nouamane/miniconda/bin/activate
+++ _CONDA_ROOT=/fsx/nouamane/miniconda
+++ . /fsx/nouamane/miniconda/etc/profile.d/conda.sh
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ '[' -z x ']'
+++ conda activate
+++ local cmd=activate
+++ case "$cmd" in
+++ __conda_activate activate
+++ '[' -n '' ']'
+++ local ask_conda
++++ PS1=
++++ __conda_exe shell.posix activate
++++ /fsx/nouamane/miniconda/bin/conda shell.posix activate
+++ ask_conda='. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_2='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\'''
+++ eval '. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_2='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\'''
++++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh
+++++ test -n ''
+++++ unset XML_CATALOG_FILES
+++++ unset xml_catalog_files_libxml2
++++ PS1='(base) '
++++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++++ export CONDA_PREFIX=/fsx/nouamane/miniconda
++++ CONDA_PREFIX=/fsx/nouamane/miniconda
++++ export CONDA_SHLVL=3
++++ CONDA_SHLVL=3
++++ export CONDA_DEFAULT_ENV=base
++++ CONDA_DEFAULT_ENV=base
++++ export 'CONDA_PROMPT_MODIFIER=(base) '
++++ CONDA_PROMPT_MODIFIER='(base) '
++++ export CONDA_PREFIX_2=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ CONDA_PREFIX_2=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ __conda_hashr
+++ '[' -n '' ']'
+++ '[' -n '' ']'
+++ hash -r
++ conda activate 2-1-cu121
++ local cmd=activate
++ case "$cmd" in
++ __conda_activate activate 2-1-cu121
++ '[' -n '' ']'
++ local ask_conda
+++ PS1='(base) '
+++ __conda_exe shell.posix activate 2-1-cu121
+++ /fsx/nouamane/miniconda/bin/conda shell.posix activate 2-1-cu121
++ ask_conda='PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''4'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_3='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
++ eval 'PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''4'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_3='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
+++ PS1='(2-1-cu121) '
+++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ export CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ export CONDA_SHLVL=4
+++ CONDA_SHLVL=4
+++ export CONDA_DEFAULT_ENV=2-1-cu121
+++ CONDA_DEFAULT_ENV=2-1-cu121
+++ export 'CONDA_PROMPT_MODIFIER=(2-1-cu121) '
+++ CONDA_PROMPT_MODIFIER='(2-1-cu121) '
+++ export CONDA_PREFIX_3=/fsx/nouamane/miniconda
+++ CONDA_PREFIX_3=/fsx/nouamane/miniconda
+++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ export _CE_M=
+++ _CE_M=
+++ export _CE_CONDA=
+++ _CE_CONDA=
+++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh
++++ test -n ''
++++ xml_catalog_files_libxml2=
++++ XML_CATALOG_FILES=
++++ conda_catalog_files=
++++ ifs_libxml2=' 	
+'
++++ IFS=' '
++++ rem=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ for pre in ${rem}
++++ test '' = /fsx/nouamane/miniconda/envs/2-1-cu121
++++ conda_catalog_files=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ rem=
++++ IFS=' 	
+'
++++ conda_catalog_files='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ export 'XML_CATALOG_FILES=file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ XML_CATALOG_FILES='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ unset conda_catalog_files ifs_libxml2 rem
++ __conda_hashr
++ '[' -n '' ']'
++ '[' -n '' ']'
++ hash -r
++ export PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++ PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ scontrol show hostnames 'ip-26-0-160-225,ip-26-0-161-[138,142,153,178],ip-26-0-164-[45,75,187,207,236],ip-26-0-165-[24,38],ip-26-0-166-[214,244],ip-26-0-168-[95,120]'
++ export 'NODELIST=ip-26-0-160-225
+ip-26-0-161-138
+ip-26-0-161-142
+ip-26-0-161-153
+ip-26-0-161-178
+ip-26-0-164-45
+ip-26-0-164-75
+ip-26-0-164-187
+ip-26-0-164-207
+ip-26-0-164-236
+ip-26-0-165-24
+ip-26-0-165-38
+ip-26-0-166-214
+ip-26-0-166-244
+ip-26-0-168-95
+ip-26-0-168-120'
++ NODELIST='ip-26-0-160-225
+ip-26-0-161-138
+ip-26-0-161-142
+ip-26-0-161-153
+ip-26-0-161-178
+ip-26-0-164-45
+ip-26-0-164-75
+ip-26-0-164-187
+ip-26-0-164-207
+ip-26-0-164-236
+ip-26-0-165-24
+ip-26-0-165-38
+ip-26-0-166-214
+ip-26-0-166-244
+ip-26-0-168-95
+ip-26-0-168-120'
+++ head -n1
+++ scontrol show hostnames 'ip-26-0-160-225,ip-26-0-161-[138,142,153,178],ip-26-0-164-[45,75,187,207,236],ip-26-0-165-[24,38],ip-26-0-166-[214,244],ip-26-0-168-[95,120]'
++ export MASTER_NODE=ip-26-0-160-225
++ MASTER_NODE=ip-26-0-160-225
++ export MASTER_PORT=12356
++ MASTER_PORT=12356
++ export NNODES=16
++ NNODES=16
++ export GPUS_PER_NODE=8
++ GPUS_PER_NODE=8
++ export WORLD_SIZE=128
++ WORLD_SIZE=128
++ export CUDA_DEVICE_MAX_CONNECTIONS=1
++ CUDA_DEVICE_MAX_CONNECTIONS=1
++ export NANOTRON_BENCHMARK=1
++ NANOTRON_BENCHMARK=1
++ echo 'Master node: ip-26-0-160-225'
+Master node: ip-26-0-160-225
++ echo 'All nodes: ip-26-0-160-225
+ip-26-0-161-138
+ip-26-0-161-142
+ip-26-0-161-153
+ip-26-0-161-178
+ip-26-0-164-45
+ip-26-0-164-75
+ip-26-0-164-187
+ip-26-0-164-207
+ip-26-0-164-236
+ip-26-0-165-24
+ip-26-0-165-38
+ip-26-0-166-214
+ip-26-0-166-244
+ip-26-0-168-95
+ip-26-0-168-120'
+All nodes: ip-26-0-160-225
+ip-26-0-161-138
+ip-26-0-161-142
+ip-26-0-161-153
+ip-26-0-161-178
+ip-26-0-164-45
+ip-26-0-164-75
+ip-26-0-164-187
+ip-26-0-164-207
+ip-26-0-164-236
+ip-26-0-165-24
+ip-26-0-165-38
+ip-26-0-166-214
+ip-26-0-166-244
+ip-26-0-168-95
+ip-26-0-168-120
++ echo 'World size: 128'
+World size: 128
++ srun torchrun --nnodes=16 --nproc_per_node=8 --rdzv_id=13265919 --rdzv_backend=c10d --rdzv_endpoint=ip-26-0-160-225:12356 run_train.py --config-file benchmark/configs/config_1.34G_dp32_tp4_pp1_acc8_mbs2_seq2048_zero1_tpmodeALL_vocab131k.yaml
+[2024-12-10 08:36:15,280] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-10 08:36:15,281] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-10 08:36:15,281] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-10 08:36:15,281] torch.distributed.run: [WARNING] 
+[2024-12-10 08:36:15,281] torch.distributed.run: [WARNING] *****************************************
+[2024-12-10 08:36:15,281] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-10 08:36:15,281] torch.distributed.run: [WARNING] *****************************************
+[2024-12-10 08:36:15,281] torch.distributed.run: [WARNING] 
+[2024-12-10 08:36:15,281] torch.distributed.run: [WARNING] *****************************************
+[2024-12-10 08:36:15,281] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-10 08:36:15,281] torch.distributed.run: [WARNING] *****************************************
+[2024-12-10 08:36:15,281] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-10 08:36:15,281] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-10 08:36:15,281] torch.distributed.run: [WARNING] 
+[2024-12-10 08:36:15,281] torch.distributed.run: [WARNING] *****************************************
+[2024-12-10 08:36:15,281] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-10 08:36:15,281] torch.distributed.run: [WARNING] *****************************************
+[2024-12-10 08:36:15,282] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-10 08:36:15,282] torch.distributed.run: [WARNING] 
+[2024-12-10 08:36:15,282] torch.distributed.run: [WARNING] *****************************************
+[2024-12-10 08:36:15,282] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-10 08:36:15,282] torch.distributed.run: [WARNING] *****************************************
+[2024-12-10 08:36:15,288] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-10 08:36:15,289] torch.distributed.run: [WARNING] 
+[2024-12-10 08:36:15,289] torch.distributed.run: [WARNING] *****************************************
+[2024-12-10 08:36:15,289] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-10 08:36:15,289] torch.distributed.run: [WARNING] *****************************************
+[2024-12-10 08:36:15,290] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-10 08:36:15,290] torch.distributed.run: [WARNING] 
+[2024-12-10 08:36:15,290] torch.distributed.run: [WARNING] *****************************************
+[2024-12-10 08:36:15,290] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-10 08:36:15,290] torch.distributed.run: [WARNING] *****************************************
+[2024-12-10 08:36:15,292] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-10 08:36:15,293] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-10 08:36:15,293] torch.distributed.run: [WARNING] 
+[2024-12-10 08:36:15,293] torch.distributed.run: [WARNING] *****************************************
+[2024-12-10 08:36:15,293] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-10 08:36:15,293] torch.distributed.run: [WARNING] *****************************************
+[2024-12-10 08:36:15,295] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-10 08:36:15,295] torch.distributed.run: [WARNING] 
+[2024-12-10 08:36:15,295] torch.distributed.run: [WARNING] *****************************************
+[2024-12-10 08:36:15,295] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-10 08:36:15,295] torch.distributed.run: [WARNING] *****************************************
+[2024-12-10 08:36:15,295] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-10 08:36:15,295] torch.distributed.run: [WARNING] 
+[2024-12-10 08:36:15,295] torch.distributed.run: [WARNING] *****************************************
+[2024-12-10 08:36:15,295] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-10 08:36:15,295] torch.distributed.run: [WARNING] *****************************************
+[2024-12-10 08:36:15,281] torch.distributed.run: [WARNING] 
+[2024-12-10 08:36:15,281] torch.distributed.run: [WARNING] *****************************************
+[2024-12-10 08:36:15,281] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-10 08:36:15,281] torch.distributed.run: [WARNING] *****************************************
+[2024-12-10 08:36:15,281] torch.distributed.run: [WARNING] 
+[2024-12-10 08:36:15,281] torch.distributed.run: [WARNING] *****************************************
+[2024-12-10 08:36:15,281] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-10 08:36:15,281] torch.distributed.run: [WARNING] *****************************************
+[2024-12-10 08:36:15,292] torch.distributed.run: [WARNING] 
+[2024-12-10 08:36:15,292] torch.distributed.run: [WARNING] *****************************************
+[2024-12-10 08:36:15,292] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-10 08:36:15,292] torch.distributed.run: [WARNING] *****************************************
+[2024-12-10 08:36:15,342] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-10 08:36:15,362] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-10 08:36:15,366] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-10 08:36:15,343] torch.distributed.run: [WARNING] 
+[2024-12-10 08:36:15,343] torch.distributed.run: [WARNING] *****************************************
+[2024-12-10 08:36:15,343] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-10 08:36:15,343] torch.distributed.run: [WARNING] *****************************************
+[2024-12-10 08:36:15,362] torch.distributed.run: [WARNING] 
+[2024-12-10 08:36:15,362] torch.distributed.run: [WARNING] *****************************************
+[2024-12-10 08:36:15,362] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-10 08:36:15,362] torch.distributed.run: [WARNING] *****************************************
+[2024-12-10 08:36:15,366] torch.distributed.run: [WARNING] 
+[2024-12-10 08:36:15,366] torch.distributed.run: [WARNING] *****************************************
+[2024-12-10 08:36:15,366] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-10 08:36:15,366] torch.distributed.run: [WARNING] *****************************************
+[2024-12-10 08:36:15,481] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-10 08:36:15,481] torch.distributed.run: [WARNING] 
+[2024-12-10 08:36:15,481] torch.distributed.run: [WARNING] *****************************************
+[2024-12-10 08:36:15,481] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-10 08:36:15,481] torch.distributed.run: [WARNING] *****************************************
+12/10/2024 08:37:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: Config:
+12/10/2024 08:37:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: Config(general=GeneralArgs(project='debug',
+12/10/2024 08:37:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                            run='1.34G_dp32_tp4_pp1_acc8_mbs2_seq2048_zero1_tpmodeALL_vocab131k',
+12/10/2024 08:37:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                            seed=42,
+12/10/2024 08:37:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                            step=None,
+12/10/2024 08:37:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                            consumed_train_samples=None,
+12/10/2024 08:37:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                            benchmark_csv_path=PosixPath('benchmark/results/bench_final.csv'),
+12/10/2024 08:37:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                            ignore_sanity_checks=True),
+12/10/2024 08:37:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:        parallelism=ParallelismArgs(dp=32,
+12/10/2024 08:37:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                    pp=1,
+12/10/2024 08:37:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                    tp=4,
+12/10/2024 08:37:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                    pp_engine=<nanotron.parallel.pipeline_parallel.engine.OneForwardOneBackwardPipelineEngine object at 0x7f052b0a8c10>,
+12/10/2024 08:37:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                    tp_mode=<TensorParallelLinearMode.ALL_REDUCE: 1>,
+12/10/2024 08:37:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                    tp_linear_async_communication=False,
+12/10/2024 08:37:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                    recompute_layer=False,
+12/10/2024 08:37:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                    tp_recompute_allgather=True,
+12/10/2024 08:37:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                    expert_parallel_size=1),
+12/10/2024 08:37:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:        model=ModelArgs(model_config=LlamaConfig(bos_token_id=0,
+12/10/2024 08:37:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 eos_token_id=0,
+12/10/2024 08:37:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 hidden_act='silu',
+12/10/2024 08:37:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 hidden_size=2048,
+12/10/2024 08:37:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 initializer_range=0.02,
+12/10/2024 08:37:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 intermediate_size=8192,
+12/10/2024 08:37:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 is_llama_config=True,
+12/10/2024 08:37:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 max_position_embeddings=2048,
+12/10/2024 08:37:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 num_attention_heads=32,
+12/10/2024 08:37:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 num_hidden_layers=16,
+12/10/2024 08:37:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 num_key_value_heads=8,
+12/10/2024 08:37:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 pad_token_id=None,
+12/10/2024 08:37:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 pretraining_tp=1,
+12/10/2024 08:37:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 rms_norm_eps=1e-05,
+12/10/2024 08:37:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 rope_scaling=None,
+12/10/2024 08:37:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 rope_theta=10000.0,
+12/10/2024 08:37:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 rope_interleaved=False,
+12/10/2024 08:37:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 tie_word_embeddings=True,
+12/10/2024 08:37:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 use_cache=True,
+12/10/2024 08:37:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 vocab_size=131072),
+12/10/2024 08:37:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                        init_method=RandomInit(std=0.02),
+12/10/2024 08:37:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                        dtype=torch.bfloat16,
+12/10/2024 08:37:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                        make_vocab_size_divisible_by=1,
+12/10/2024 08:37:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                        ddp_bucket_cap_mb=25),
+12/10/2024 08:37:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:        tokenizer=TokenizerArgs(tokenizer_name_or_path='robot-test/dummy-tokenizer-wordlevel',
+12/10/2024 08:37:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                tokenizer_revision=None,
+12/10/2024 08:37:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                tokenizer_max_length=None),
+12/10/2024 08:37:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:        checkpoints=CheckpointsArgs(checkpoints_path=PosixPath('checkpoints'),
+12/10/2024 08:37:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                    checkpoint_interval=10000,
+12/10/2024 08:37:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                    save_initial_state=False,
+12/10/2024 08:37:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                    save_final_state=False,
+12/10/2024 08:37:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                    resume_checkpoint_path=None,
+12/10/2024 08:37:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                    checkpoints_path_is_shared_file_system=False),
+12/10/2024 08:37:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:        logging=LoggingArgs(log_level='info',
+12/10/2024 08:37:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                            log_level_replica='info',
+12/10/2024 08:37:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                            iteration_step_info_interval=1),
+12/10/2024 08:37:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:        tokens=TokensArgs(sequence_length=2048,
+12/10/2024 08:37:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                          train_steps=100,
+12/10/2024 08:37:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                          micro_batch_size=2,
+12/10/2024 08:37:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                          batch_accumulation_per_replica=8,
+12/10/2024 08:37:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                          val_check_interval=100,
+12/10/2024 08:37:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                          limit_val_batches=0,
+12/10/2024 08:37:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                          limit_test_batches=0),
+12/10/2024 08:37:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:        optimizer=OptimizerArgs(optimizer_factory=AdamWOptimizerArgs(adam_eps=1e-08,
+12/10/2024 08:37:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                                     adam_beta1=0.9,
+12/10/2024 08:37:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                                     adam_beta2=0.95,
+12/10/2024 08:37:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                                     torch_adam_is_fused=True,
+12/10/2024 08:37:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                                     name='adamW'),
+12/10/2024 08:37:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                zero_stage=1,
+12/10/2024 08:37:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                weight_decay=0.01,
+12/10/2024 08:37:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                clip_grad=1.0,
+12/10/2024 08:37:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                accumulate_grad_in_fp32=True,
+12/10/2024 08:37:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                learning_rate_scheduler=LRSchedulerArgs(learning_rate=0.0003,
+12/10/2024 08:37:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                                        lr_warmup_steps=2,
+12/10/2024 08:37:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                                        lr_warmup_style='linear',
+12/10/2024 08:37:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                                        lr_decay_style='cosine',
+12/10/2024 08:37:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                                        lr_decay_steps=13,
+12/10/2024 08:37:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                                        lr_decay_starting_step=None,
+12/10/2024 08:37:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                                        min_decay_lr=1e-05)),
+12/10/2024 08:37:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:        data_stages=[DatasetStageArgs(name='Stable Training Stage',
+12/10/2024 08:37:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                      start_training_step=1,
+12/10/2024 08:37:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                      data=DataArgs(dataset=None,
+12/10/2024 08:37:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                    seed=42,
+12/10/2024 08:37:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                    num_loading_workers=1))],
+12/10/2024 08:37:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:        profiler=None,
+12/10/2024 08:37:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:        lighteval=None,
+12/10/2024 08:37:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:        s3_upload=None)
+12/10/2024 08:37:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: Model Config:
+12/10/2024 08:37:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: LlamaConfig(bos_token_id=0,
+12/10/2024 08:37:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             eos_token_id=0,
+12/10/2024 08:37:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             hidden_act='silu',
+12/10/2024 08:37:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             hidden_size=2048,
+12/10/2024 08:37:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             initializer_range=0.02,
+12/10/2024 08:37:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             intermediate_size=8192,
+12/10/2024 08:37:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             is_llama_config=True,
+12/10/2024 08:37:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             max_position_embeddings=2048,
+12/10/2024 08:37:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             num_attention_heads=32,
+12/10/2024 08:37:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             num_hidden_layers=16,
+12/10/2024 08:37:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             num_key_value_heads=8,
+12/10/2024 08:37:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             pad_token_id=None,
+12/10/2024 08:37:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             pretraining_tp=1,
+12/10/2024 08:37:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             rms_norm_eps=1e-05,
+12/10/2024 08:37:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             rope_scaling=None,
+12/10/2024 08:37:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             rope_theta=10000.0,
+12/10/2024 08:37:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             rope_interleaved=False,
+12/10/2024 08:37:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             tie_word_embeddings=True,
+12/10/2024 08:37:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             use_cache=True,
+12/10/2024 08:37:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             vocab_size=131072)
+12/10/2024 08:37:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: Building model..
+12/10/2024 08:37:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: Initialize RoPE Theta = 10000.0
+12/10/2024 08:37:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: Setting PP block ranks...
+12/10/2024 08:37:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: Total number of parameters: 1.24G (2368.52MiB)
+12/10/2024 08:37:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: Local number of parameters: 310M (592.13MiB)
+12/10/2024 08:37:05 [INFO|DP=0|PP=0|TP=1|ip-26-0-160-225]: Local number of parameters: 310M (592.13MiB)
+12/10/2024 08:37:05 [INFO|DP=0|PP=0|TP=2|ip-26-0-160-225]: Local number of parameters: 310M (592.13MiB)
+12/10/2024 08:37:05 [INFO|DP=0|PP=0|TP=3|ip-26-0-160-225]: Local number of parameters: 310M (592.13MiB)
+12/10/2024 08:37:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [After model building] Memory usage: 592.15MiB. Peak allocated: 5392.00MiB Peak reserved: 29974.00MiB
+12/10/2024 08:37:05 [INFO|DP=0|PP=0|TP=1|ip-26-0-160-225]: [After model building] Memory usage: 592.15MiB. Peak allocated: 5392.00MiB Peak reserved: 29974.00MiB
+12/10/2024 08:37:05 [INFO|DP=0|PP=0|TP=2|ip-26-0-160-225]: [After model building] Memory usage: 592.15MiB. Peak allocated: 5392.00MiB Peak reserved: 29974.00MiB
+12/10/2024 08:37:05 [INFO|DP=0|PP=0|TP=3|ip-26-0-160-225]: [After model building] Memory usage: 592.15MiB. Peak allocated: 5392.00MiB Peak reserved: 29974.00MiB
+12/10/2024 08:37:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: No checkpoint path provided.
+12/10/2024 08:37:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: Parametrizing model parameters using StandardParametrizator
+12/10/2024 08:37:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [Optimizer Building] Using LearningRateForSP as learning rate
+12/10/2024 08:37:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [ZeRO sharding] Size of optimizer params per rank:
+12/10/2024 08:37:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [ZeRO sharding] DP Rank 0 has 9.7M out of 310M (3.12%) params' optimizer states
+12/10/2024 08:37:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [ZeRO sharding] DP Rank 1 has 9.7M out of 310M (3.12%) params' optimizer states
+12/10/2024 08:37:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [ZeRO sharding] DP Rank 2 has 9.7M out of 310M (3.12%) params' optimizer states
+12/10/2024 08:37:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [ZeRO sharding] DP Rank 3 has 9.7M out of 310M (3.12%) params' optimizer states
+12/10/2024 08:37:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [ZeRO sharding] DP Rank 4 has 9.7M out of 310M (3.12%) params' optimizer states
+12/10/2024 08:37:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [ZeRO sharding] DP Rank 5 has 9.7M out of 310M (3.12%) params' optimizer states
+12/10/2024 08:37:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [ZeRO sharding] DP Rank 6 has 9.7M out of 310M (3.12%) params' optimizer states
+12/10/2024 08:37:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [ZeRO sharding] DP Rank 7 has 9.7M out of 310M (3.12%) params' optimizer states
+12/10/2024 08:37:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [ZeRO sharding] DP Rank 8 has 9.7M out of 310M (3.12%) params' optimizer states
+12/10/2024 08:37:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [ZeRO sharding] DP Rank 9 has 9.7M out of 310M (3.12%) params' optimizer states
+12/10/2024 08:37:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [ZeRO sharding] DP Rank 10 has 9.7M out of 310M (3.12%) params' optimizer states
+12/10/2024 08:37:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [ZeRO sharding] DP Rank 11 has 9.7M out of 310M (3.12%) params' optimizer states
+12/10/2024 08:37:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [ZeRO sharding] DP Rank 12 has 9.7M out of 310M (3.12%) params' optimizer states
+12/10/2024 08:37:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [ZeRO sharding] DP Rank 13 has 9.7M out of 310M (3.12%) params' optimizer states
+12/10/2024 08:37:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [ZeRO sharding] DP Rank 14 has 9.7M out of 310M (3.12%) params' optimizer states
+12/10/2024 08:37:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [ZeRO sharding] DP Rank 15 has 9.7M out of 310M (3.12%) params' optimizer states
+12/10/2024 08:37:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [ZeRO sharding] DP Rank 16 has 9.7M out of 310M (3.12%) params' optimizer states
+12/10/2024 08:37:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [ZeRO sharding] DP Rank 17 has 9.7M out of 310M (3.12%) params' optimizer states
+12/10/2024 08:37:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [ZeRO sharding] DP Rank 18 has 9.7M out of 310M (3.12%) params' optimizer states
+12/10/2024 08:37:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [ZeRO sharding] DP Rank 19 has 9.7M out of 310M (3.12%) params' optimizer states
+12/10/2024 08:37:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [ZeRO sharding] DP Rank 20 has 9.7M out of 310M (3.12%) params' optimizer states
+12/10/2024 08:37:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [ZeRO sharding] DP Rank 21 has 9.7M out of 310M (3.12%) params' optimizer states
+12/10/2024 08:37:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [ZeRO sharding] DP Rank 22 has 9.7M out of 310M (3.12%) params' optimizer states
+12/10/2024 08:37:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [ZeRO sharding] DP Rank 23 has 9.7M out of 310M (3.12%) params' optimizer states
+12/10/2024 08:37:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [ZeRO sharding] DP Rank 24 has 9.7M out of 310M (3.12%) params' optimizer states
+12/10/2024 08:37:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [ZeRO sharding] DP Rank 25 has 9.7M out of 310M (3.12%) params' optimizer states
+12/10/2024 08:37:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [ZeRO sharding] DP Rank 26 has 9.7M out of 310M (3.12%) params' optimizer states
+12/10/2024 08:37:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [ZeRO sharding] DP Rank 27 has 9.7M out of 310M (3.12%) params' optimizer states
+12/10/2024 08:37:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [ZeRO sharding] DP Rank 28 has 9.7M out of 310M (3.12%) params' optimizer states
+12/10/2024 08:37:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [ZeRO sharding] DP Rank 29 has 9.7M out of 310M (3.12%) params' optimizer states
+12/10/2024 08:37:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [ZeRO sharding] DP Rank 30 has 9.7M out of 310M (3.12%) params' optimizer states
+12/10/2024 08:37:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [ZeRO sharding] DP Rank 31 has 9.7M out of 310M (3.12%) params' optimizer states
+12/10/2024 08:37:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [Training Plan] Stage Stable Training Stage has 99 remaining training steps and has consumed 0 samples
+12/10/2024 08:37:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: Using dummy data generator
+12/10/2024 08:37:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [Training Plan] There are 1 training stages 
+12/10/2024 08:37:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [Stage Stable Training Stage] start from step 1 
+12/10/2024 08:37:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: 
+12/10/2024 08:37:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [Start training] datetime: 2024-12-10 08:37:08.439226 | mbs: 2 | grad_accum: 8 | global_batch_size: 512 | sequence_length: 2048 | train_steps: 100 | start_iteration_step: 0 | consumed_train_samples: 0
+wandb: WARNING Path /fsx/nouamane/.cache/wandb/wandb/ wasn't writable, using system temp directory.
+wandb: WARNING Path /fsx/nouamane/.cache/wandb/wandb/ wasn't writable, using system temp directory
+wandb: Tracking run with wandb version 0.16.0
+wandb: W&B syncing is set to `offline` in this directory.  
+wandb: Run `wandb online` or set WANDB_MODE=online to enable cloud syncing.
+12/10/2024 08:37:16 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: Resuming training from stage Stable Training Stage, it has trained for 0 samples and has 99 remaining train steps
+12/10/2024 08:37:16 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:  Memory usage: 1813.42MiB. Peak allocated 5392.00MiB. Peak reserved: 31160.00MiB
+12/10/2024 08:37:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:  Memory usage: 1881.49MiB. Peak allocated 5212.24MiB. Peak reserved: 6004.00MiB
+12/10/2024 08:37:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: iteration: 1 / 100 | consumed_tokens: 1.05M | elapsed_time_per_iteration_ms: 2.63K | tokens_per_sec: 399K | tokens_per_sec_per_gpu: 3.12K | global_batch_size: 512 | lm_loss: 12.2 | lr: 0.00015 | model_tflops_per_gpu: 25.8 | hardware_tflops_per_gpu: 25.8 | grad_norm: 0.65 | cuda_memory_allocated: 2.05G | cuda_max_memory_reserved: 6.35G | hd_total_memory_tb: 312G | hd_used_memory_tb: 67.5G | hd_free_memory_tb: 245G
+12/10/2024 08:37:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:  Memory usage: 1955.57MiB. Peak allocated 3159.34MiB. Peak reserved: 6058.00MiB
+12/10/2024 08:37:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:  Memory usage: 1955.58MiB. Peak allocated 5286.33MiB. Peak reserved: 6062.00MiB
+12/10/2024 08:37:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: iteration: 2 / 100 | consumed_tokens: 2.1M | elapsed_time_per_iteration_ms: 552 | tokens_per_sec: 1.9M | tokens_per_sec_per_gpu: 14.8K | global_batch_size: 512 | lm_loss: 12.2 | lr: 0.0003 | model_tflops_per_gpu: 122 | hardware_tflops_per_gpu: 122 | grad_norm: 0.651 | cuda_memory_allocated: 2.05G | cuda_max_memory_reserved: 6.36G | hd_total_memory_tb: 312G | hd_used_memory_tb: 67.5G | hd_free_memory_tb: 245G
+12/10/2024 08:37:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:  Memory usage: 1955.57MiB. Peak allocated 3159.34MiB. Peak reserved: 6062.00MiB
+12/10/2024 08:37:20 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:  Memory usage: 1955.58MiB. Peak allocated 5286.33MiB. Peak reserved: 6062.00MiB
+num_paramsnum_params
+num_paramsnum_params
+num_params
+{'total': 1241784320, 'local': 310446080}
+{'total': 1241784320, 'local': 310446080}
+
+
+{'total': 1241784320, 'local': 310446080}{'total': 1241784320, 'local': 310446080}{'total': 1241784320, 'local': 310446080}
+
+
+num_params
+{'total': 1241784320, 'local': 310446080}
+num_params
+{'total': 1241784320, 'local': 310446080}
+num_params
+{'total': 1241784320, 'local': 310446080}
+num_paramsnum_paramsnum_params
+
+
+num_params{'total': 1241784320, 'local': 310446080}{'total': 1241784320, 'local': 310446080}
+{'total': 1241784320, 'local': 310446080}
+
+
+num_params{'total': 1241784320, 'local': 310446080}
+
+num_paramsnum_params
+
+{'total': 1241784320, 'local': 310446080}
+{'total': 1241784320, 'local': 310446080}
+num_params
+num_params{'total': 1241784320, 'local': 310446080}
+
+{'total': 1241784320, 'local': 310446080}
+num_params
+num_paramsnum_paramsnum_params
+
+
+{'total': 1241784320, 'local': 310446080}
+{'total': 1241784320, 'local': 310446080}{'total': 1241784320, 'local': 310446080}
+num_params
+
+{'total': 1241784320, 'local': 310446080}
+num_paramsnum_paramsnum_params
+
+num_params
+{'total': 1241784320, 'local': 310446080}
+
+{'total': 1241784320, 'local': 310446080}
+{'total': 1241784320, 'local': 310446080}num_params{'total': 1241784320, 'local': 310446080}
+
+
+{'total': 1241784320, 'local': 310446080}
+num_paramsnum_paramsnum_paramsnum_params
+num_paramsnum_params
+
+{'total': 1241784320, 'local': 310446080}
+
+
+
+num_params{'total': 1241784320, 'local': 310446080}{'total': 1241784320, 'local': 310446080}{'total': 1241784320, 'local': 310446080}{'total': 1241784320, 'local': 310446080}
+
+
+num_params
+{'total': 1241784320, 'local': 310446080}
+
+{'total': 1241784320, 'local': 310446080}
+
+num_params
+{'total': 1241784320, 'local': 310446080}
+num_params
+num_params{'total': 1241784320, 'local': 310446080}
+
+{'total': 1241784320, 'local': 310446080}
+num_params
+num_params{'total': 1241784320, 'local': 310446080}num_params
+
+
+num_params{'total': 1241784320, 'local': 310446080}
+{'total': 1241784320, 'local': 310446080}
+num_paramsnum_params
+
+
+num_paramsnum_paramsnum_params
+
+
+{'total': 1241784320, 'local': 310446080}{'total': 1241784320, 'local': 310446080}{'total': 1241784320, 'local': 310446080}num_params
+
+num_params
+
+num_paramsnum_params
+{'total': 1241784320, 'local': 310446080}num_params
+
+
+num_paramsnum_params
+num_params
+num_params{'total': 1241784320, 'local': 310446080}
+num_params
+{'total': 1241784320, 'local': 310446080}
+
+{'total': 1241784320, 'local': 310446080}
+
+num_params{'total': 1241784320, 'local': 310446080}num_params
+{'total': 1241784320, 'local': 310446080}
+
+
+num_params
+{'total': 1241784320, 'local': 310446080}
+num_params
+num_params{'total': 1241784320, 'local': 310446080}
+
+num_params{'total': 1241784320, 'local': 310446080}
+
+num_params
+{'total': 1241784320, 'local': 310446080}
+num_params
+num_params{'total': 1241784320, 'local': 310446080}
+
+num_paramsnum_paramsnum_params
+
+
+{'total': 1241784320, 'local': 310446080}{'total': 1241784320, 'local': 310446080}num_params
+{'total': 1241784320, 'local': 310446080}num_params
+
+
+
+num_params
+{'total': 1241784320, 'local': 310446080}{'total': 1241784320, 'local': 310446080}
+
+num_params
+num_paramsnum_params{'total': 1241784320, 'local': 310446080}
+
+
+{'total': 1241784320, 'local': 310446080}
+num_paramsnum_params
+
+{'total': 1241784320, 'local': 310446080}
+num_params{'total': 1241784320, 'local': 310446080}
+
+num_paramsnum_params
+num_paramsnum_params
+{'total': 1241784320, 'local': 310446080}
+
+num_params
+{'total': 1241784320, 'local': 310446080}{'total': 1241784320, 'local': 310446080}
+{'total': 1241784320, 'local': 310446080}
+{'total': 1241784320, 'local': 310446080}num_params
+num_params
+{'total': 1241784320, 'local': 310446080}
+
+num_paramsnum_params
+
+{'total': 1241784320, 'local': 310446080}{'total': 1241784320, 'local': 310446080}
+
+num_paramsnum_params
+num_paramsnum_params
+num_params{'total': 1241784320, 'local': 310446080}
+
+
+{'total': 1241784320, 'local': 310446080}
+{'total': 1241784320, 'local': 310446080}
+{'total': 1241784320, 'local': 310446080}
+num_paramsnum_paramsnum_params
+
+num_params{'total': 1241784320, 'local': 310446080}
+
+{'total': 1241784320, 'local': 310446080}
+
+{'total': 1241784320, 'local': 310446080}
+{'total': 1241784320, 'local': 310446080}num_params{'total': 1241784320, 'local': 310446080}
+
+
+{'total': 1241784320, 'local': 310446080}
+{'total': 1241784320, 'local': 310446080}
+
+{'total': 1241784320, 'local': 310446080}
+{'total': 1241784320, 'local': 310446080}
+{'total': 1241784320, 'local': 310446080}
+{'total': 1241784320, 'local': 310446080}
+num_params{'total': 1241784320, 'local': 310446080}
+
+num_params{'total': 1241784320, 'local': 310446080}
+
+num_params{'total': 1241784320, 'local': 310446080}
+
+num_params{'total': 1241784320, 'local': 310446080}
+
+num_params{'total': 1241784320, 'local': 310446080}
+
+num_paramsnum_params
+
+{'total': 1241784320, 'local': 310446080}num_params{'total': 1241784320, 'local': 310446080}num_paramsnum_params
+
+
+
+
+{'total': 1241784320, 'local': 310446080}{'total': 1241784320, 'local': 310446080}
+
+num_params{'total': 1241784320, 'local': 310446080}
+
+{'total': 1241784320, 'local': 310446080}
+num_params
+num_params{'total': 1241784320, 'local': 310446080}
+
+num_params{'total': 1241784320, 'local': 310446080}
+
+{'total': 1241784320, 'local': 310446080}
+num_params{'total': 1241784320, 'local': 310446080}{'total': 1241784320, 'local': 310446080}
+
+num_paramsnum_params
+
+
+num_params{'total': 1241784320, 'local': 310446080}{'total': 1241784320, 'local': 310446080}
+
+{'total': 1241784320, 'local': 310446080}
+
+{'total': 1241784320, 'local': 310446080}
+{'total': 1241784320, 'local': 310446080}
+
+
+num_params{'total': 1241784320, 'local': 310446080}
+
+{'total': 1241784320, 'local': 310446080}
+num_params{'total': 1241784320, 'local': 310446080}
+
+{'total': 1241784320, 'local': 310446080}
+{'total': 1241784320, 'local': 310446080}
+
+{'total': 1241784320, 'local': 310446080}
+{'total': 1241784320, 'local': 310446080}
+{'total': 1241784320, 'local': 310446080}
+num_params
+{'total': 1241784320, 'local': 310446080}
+{'total': 1241784320, 'local': 310446080}
+num_params
+{'total': 1241784320, 'local': 310446080}
+num_params
+{'total': 1241784320, 'local': 310446080}
+num_params
+{'total': 1241784320, 'local': 310446080}
+{'total': 1241784320, 'local': 310446080}
+{'total': 1241784320, 'local': 310446080}
+num_params
+{'total': 1241784320, 'local': 310446080}
+num_params
+num_params
+num_params
+{'total': 1241784320, 'local': 310446080}
+{'total': 1241784320, 'local': 310446080}
+{'total': 1241784320, 'local': 310446080}
+num_params{'total': 1241784320, 'local': 310446080}
+
+num_params
+{'total': 1241784320, 'local': 310446080}
+{'total': 1241784320, 'local': 310446080}
+12/10/2024 08:37:20 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: iteration: 3 / 100 | consumed_tokens: 3.15M | elapsed_time_per_iteration_ms: 551 | tokens_per_sec: 1.9M | tokens_per_sec_per_gpu: 14.9K | global_batch_size: 512 | lm_loss: 12.2 | lr: 0.000296 | model_tflops_per_gpu: 123 | hardware_tflops_per_gpu: 123 | grad_norm: 0.642 | cuda_memory_allocated: 2.05G | cuda_max_memory_reserved: 6.36G | hd_total_memory_tb: 312G | hd_used_memory_tb: 67.5G | hd_free_memory_tb: 245G
+num_params
+{'total': 1241784320, 'local': 310446080}
+12/10/2024 08:37:20 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: | job_id   | name                                                           | nodes | seq_len | mbs | batch_accum | gbs | mTFLOPs | hTFLOPs | tok/s/gpu | AllReduce (GB/s) | AllGather (GB/s) | ReduceScatter (GB/s) | AR Intra-node (GB/s) | AG Intra-node (GB/s) | RS Intra-node (GB/s) | Mem Alloc (GB) | Mem Res (GB) | dp | pp | tp | pp_engine                                                                                                 | tp_mode                             | tp_async_comm | hidden_size | hidden_act | num_layers | num_heads | num_kv_heads | max_pos | vocab_size | tie_word_embeddings | dtype          | zero_stage | ddp_bucket_cap_mb | accumulate_grad_in_fp32 | Total Params | Local Params |
+12/10/2024 08:37:20 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: | -------- | -------------------------------------------------------------- | ----- | ------- | --- | ----------- | --- | ------- | ------- | --------- | ---------------- | ---------------- | -------------------- | -------------------- | -------------------- | -------------------- | -------------- | ------------ | -- | -- | -- | --------------------------------------------------------------------------------------------------------- | ----------------------------------- | ------------- | ----------- | ---------- | ---------- | --------- | ------------ | ------- | ---------- | ------------------- | -------------- | ---------- | ----------------- | ----------------------- | ------------ | ------------ |
+12/10/2024 08:37:20 [INFO|DP=28|PP=0|TP=0|ip-26-0-168-120]: Throughput logging complete
+12/10/2024 08:37:20 [INFO|DP=12|PP=0|TP=1|ip-26-0-164-207]: Throughput logging complete
+12/10/2024 08:37:20 [INFO|DP=8|PP=0|TP=3|ip-26-0-161-178]: Throughput logging complete
+12/10/2024 08:37:20 [INFO|DP=9|PP=0|TP=0|ip-26-0-161-178]: Throughput logging complete
+12/10/2024 08:37:20 [INFO|DP=9|PP=0|TP=1|ip-26-0-161-178]: Throughput logging complete
+12/10/2024 08:37:20 [INFO|DP=9|PP=0|TP=2|ip-26-0-161-178]: Throughput logging complete
+12/10/2024 08:37:20 [INFO|DP=20|PP=0|TP=0|ip-26-0-165-24]: Throughput logging complete
+12/10/2024 08:37:20 [INFO|DP=2|PP=0|TP=1|ip-26-0-161-138]: Throughput logging complete
+12/10/2024 08:37:20 [INFO|DP=16|PP=0|TP=3|ip-26-0-164-45]: Throughput logging complete
+12/10/2024 08:37:20 [INFO|DP=16|PP=0|TP=2|ip-26-0-164-45]: Throughput logging complete
+12/10/2024 08:37:20 [INFO|DP=17|PP=0|TP=0|ip-26-0-164-45]: Throughput logging complete
+12/10/2024 08:37:20 [INFO|DP=16|PP=0|TP=0|ip-26-0-164-45]: Throughput logging complete
+12/10/2024 08:37:20 [INFO|DP=17|PP=0|TP=2|ip-26-0-164-45]: Throughput logging complete
+12/10/2024 08:37:20 [INFO|DP=17|PP=0|TP=1|ip-26-0-164-45]: Throughput logging complete
+12/10/2024 08:37:20 [INFO|DP=8|PP=0|TP=1|ip-26-0-161-178]: Throughput logging complete
+12/10/2024 08:37:20 [INFO|DP=26|PP=0|TP=1|ip-26-0-166-244]: Throughput logging complete
+12/10/2024 08:37:20 [INFO|DP=26|PP=0|TP=0|ip-26-0-166-244]: Throughput logging complete
+12/10/2024 08:37:20 [INFO|DP=24|PP=0|TP=2|ip-26-0-166-214]: Throughput logging complete
+12/10/2024 08:37:20 [INFO|DP=24|PP=0|TP=1|ip-26-0-166-214]: Throughput logging complete
+12/10/2024 08:37:20 [INFO|DP=14|PP=0|TP=3|ip-26-0-164-236]: Throughput logging complete
+12/10/2024 08:37:20 [INFO|DP=15|PP=0|TP=3|ip-26-0-164-236]: Throughput logging complete
+12/10/2024 08:37:20 [INFO|DP=8|PP=0|TP=2|ip-26-0-161-178]: Throughput logging complete
+12/10/2024 08:37:20 [INFO|DP=6|PP=0|TP=1|ip-26-0-161-153]: Throughput logging complete
+12/10/2024 08:37:20 [INFO|DP=16|PP=0|TP=1|ip-26-0-164-45]: Throughput logging complete
+12/10/2024 08:37:20 [INFO|DP=17|PP=0|TP=3|ip-26-0-164-45]: Throughput logging complete
+12/10/2024 08:37:20 [INFO|DP=3|PP=0|TP=0|ip-26-0-161-138]: Throughput logging complete
+12/10/2024 08:37:20 [INFO|DP=4|PP=0|TP=0|ip-26-0-161-142]: Throughput logging complete
+12/10/2024 08:37:20 [INFO|DP=4|PP=0|TP=1|ip-26-0-161-142]: Throughput logging complete
+12/10/2024 08:37:20 [INFO|DP=12|PP=0|TP=3|ip-26-0-164-207]: Throughput logging complete
+12/10/2024 08:37:20 [INFO|DP=12|PP=0|TP=2|ip-26-0-164-207]: Throughput logging complete
+12/10/2024 08:37:20 [INFO|DP=13|PP=0|TP=2|ip-26-0-164-207]: Throughput logging complete
+12/10/2024 08:37:20 [INFO|DP=12|PP=0|TP=0|ip-26-0-164-207]: Throughput logging complete
+12/10/2024 08:37:20 [INFO|DP=8|PP=0|TP=0|ip-26-0-161-178]: Throughput logging complete
+12/10/2024 08:37:20 [INFO|DP=21|PP=0|TP=3|ip-26-0-165-24]: Throughput logging complete
+12/10/2024 08:37:20 [INFO|DP=20|PP=0|TP=2|ip-26-0-165-24]: Throughput logging complete
+12/10/2024 08:37:20 [INFO|DP=20|PP=0|TP=1|ip-26-0-165-24]: Throughput logging complete
+12/10/2024 08:37:20 [INFO|DP=26|PP=0|TP=3|ip-26-0-166-244]: Throughput logging complete
+12/10/2024 08:37:20 [INFO|DP=2|PP=0|TP=0|ip-26-0-161-138]: Throughput logging complete
+12/10/2024 08:37:20 [INFO|DP=9|PP=0|TP=3|ip-26-0-161-178]: Throughput logging complete
+12/10/2024 08:37:20 [INFO|DP=23|PP=0|TP=0|ip-26-0-165-38]: Throughput logging complete
+12/10/2024 08:37:20 [INFO|DP=22|PP=0|TP=3|ip-26-0-165-38]: Throughput logging complete
+12/10/2024 08:37:20 [INFO|DP=22|PP=0|TP=1|ip-26-0-165-38]: Throughput logging complete
+12/10/2024 08:37:20 [INFO|DP=19|PP=0|TP=0|ip-26-0-164-75]: Throughput logging complete
+12/10/2024 08:37:20 [INFO|DP=18|PP=0|TP=2|ip-26-0-164-75]: Throughput logging complete
+12/10/2024 08:37:20 [INFO|DP=19|PP=0|TP=1|ip-26-0-164-75]: Throughput logging complete
+12/10/2024 08:37:20 [INFO|DP=18|PP=0|TP=3|ip-26-0-164-75]: Throughput logging complete
+12/10/2024 08:37:20 [INFO|DP=29|PP=0|TP=0|ip-26-0-168-120]: Throughput logging complete
+12/10/2024 08:37:20 [INFO|DP=29|PP=0|TP=2|ip-26-0-168-120]: Throughput logging complete
+12/10/2024 08:37:20 [INFO|DP=29|PP=0|TP=3|ip-26-0-168-120]: Throughput logging complete
+12/10/2024 08:37:20 [INFO|DP=28|PP=0|TP=2|ip-26-0-168-120]: Throughput logging complete
+12/10/2024 08:37:20 [INFO|DP=29|PP=0|TP=1|ip-26-0-168-120]: Throughput logging complete
+12/10/2024 08:37:20 [INFO|DP=28|PP=0|TP=1|ip-26-0-168-120]: Throughput logging complete
+12/10/2024 08:37:20 [INFO|DP=26|PP=0|TP=2|ip-26-0-166-244]: Throughput logging complete
+12/10/2024 08:37:20 [INFO|DP=2|PP=0|TP=3|ip-26-0-161-138]: Throughput logging complete
+12/10/2024 08:37:20 [INFO|DP=3|PP=0|TP=2|ip-26-0-161-138]: Throughput logging complete
+12/10/2024 08:37:20 [INFO|DP=3|PP=0|TP=3|ip-26-0-161-138]: Throughput logging complete
+12/10/2024 08:37:20 [INFO|DP=4|PP=0|TP=3|ip-26-0-161-142]: Throughput logging complete
+12/10/2024 08:37:20 [INFO|DP=5|PP=0|TP=0|ip-26-0-161-142]: Throughput logging complete
+12/10/2024 08:37:20 [INFO|DP=13|PP=0|TP=0|ip-26-0-164-207]: Throughput logging complete
+12/10/2024 08:37:20 [INFO|DP=11|PP=0|TP=3|ip-26-0-164-187]: Throughput logging complete
+12/10/2024 08:37:20 [INFO|DP=11|PP=0|TP=2|ip-26-0-164-187]: Throughput logging complete
+12/10/2024 08:37:20 [INFO|DP=10|PP=0|TP=3|ip-26-0-164-187]: Throughput logging complete
+12/10/2024 08:37:20 [INFO|DP=24|PP=0|TP=3|ip-26-0-166-214]: Throughput logging complete
+12/10/2024 08:37:20 [INFO|DP=21|PP=0|TP=0|ip-26-0-165-24]: Throughput logging complete
+12/10/2024 08:37:20 [INFO|DP=6|PP=0|TP=2|ip-26-0-161-153]: Throughput logging complete
+12/10/2024 08:37:20 [INFO|DP=28|PP=0|TP=3|ip-26-0-168-120]: Throughput logging complete
+12/10/2024 08:37:20 [INFO|DP=27|PP=0|TP=0|ip-26-0-166-244]: Throughput logging complete
+12/10/2024 08:37:20 [INFO|DP=27|PP=0|TP=1|ip-26-0-166-244]: Throughput logging complete
+12/10/2024 08:37:20 [INFO|DP=27|PP=0|TP=2|ip-26-0-166-244]: Throughput logging complete
+12/10/2024 08:37:20 [INFO|DP=27|PP=0|TP=3|ip-26-0-166-244]: Throughput logging complete
+12/10/2024 08:37:20 [INFO|DP=3|PP=0|TP=1|ip-26-0-161-138]: Throughput logging complete
+12/10/2024 08:37:20 [INFO|DP=5|PP=0|TP=2|ip-26-0-161-142]: Throughput logging complete
+12/10/2024 08:37:20 [INFO|DP=5|PP=0|TP=3|ip-26-0-161-142]: Throughput logging complete
+12/10/2024 08:37:20 [INFO|DP=5|PP=0|TP=1|ip-26-0-161-142]: Throughput logging complete
+12/10/2024 08:37:20 [INFO|DP=13|PP=0|TP=1|ip-26-0-164-207]: Throughput logging complete
+12/10/2024 08:37:20 [INFO|DP=13|PP=0|TP=3|ip-26-0-164-207]: Throughput logging complete
+12/10/2024 08:37:20 [INFO|DP=10|PP=0|TP=2|ip-26-0-164-187]: Throughput logging complete
+12/10/2024 08:37:20 [INFO|DP=11|PP=0|TP=0|ip-26-0-164-187]: Throughput logging complete
+12/10/2024 08:37:20 [INFO|DP=25|PP=0|TP=1|ip-26-0-166-214]: Throughput logging complete
+12/10/2024 08:37:20 [INFO|DP=25|PP=0|TP=3|ip-26-0-166-214]: Throughput logging complete
+12/10/2024 08:37:20 [INFO|DP=14|PP=0|TP=0|ip-26-0-164-236]: Throughput logging complete
+12/10/2024 08:37:20 [INFO|DP=14|PP=0|TP=2|ip-26-0-164-236]: Throughput logging complete
+12/10/2024 08:37:20 [INFO|DP=15|PP=0|TP=1|ip-26-0-164-236]: Throughput logging complete
+12/10/2024 08:37:20 [INFO|DP=15|PP=0|TP=0|ip-26-0-164-236]: Throughput logging complete
+12/10/2024 08:37:20 [INFO|DP=15|PP=0|TP=2|ip-26-0-164-236]: Throughput logging complete
+12/10/2024 08:37:20 [INFO|DP=22|PP=0|TP=2|ip-26-0-165-38]: Throughput logging complete
+12/10/2024 08:37:20 [INFO|DP=23|PP=0|TP=1|ip-26-0-165-38]: Throughput logging complete
+12/10/2024 08:37:20 [INFO|DP=23|PP=0|TP=3|ip-26-0-165-38]: Throughput logging complete
+12/10/2024 08:37:20 [INFO|DP=23|PP=0|TP=2|ip-26-0-165-38]: Throughput logging complete
+12/10/2024 08:37:20 [INFO|DP=21|PP=0|TP=2|ip-26-0-165-24]: Throughput logging complete
+12/10/2024 08:37:20 [INFO|DP=18|PP=0|TP=0|ip-26-0-164-75]: Throughput logging complete
+12/10/2024 08:37:20 [INFO|DP=19|PP=0|TP=2|ip-26-0-164-75]: Throughput logging complete
+12/10/2024 08:37:20 [INFO|DP=18|PP=0|TP=1|ip-26-0-164-75]: Throughput logging complete
+12/10/2024 08:37:20 [INFO|DP=19|PP=0|TP=3|ip-26-0-164-75]: Throughput logging complete
+12/10/2024 08:37:20 [INFO|DP=2|PP=0|TP=2|ip-26-0-161-138]: Throughput logging complete
+12/10/2024 08:37:20 [INFO|DP=11|PP=0|TP=1|ip-26-0-164-187]: Throughput logging complete
+12/10/2024 08:37:20 [INFO|DP=10|PP=0|TP=0|ip-26-0-164-187]: Throughput logging complete
+12/10/2024 08:37:20 [INFO|DP=10|PP=0|TP=1|ip-26-0-164-187]: Throughput logging complete
+12/10/2024 08:37:20 [INFO|DP=24|PP=0|TP=0|ip-26-0-166-214]: Throughput logging complete
+12/10/2024 08:37:20 [INFO|DP=21|PP=0|TP=1|ip-26-0-165-24]: Throughput logging complete
+12/10/2024 08:37:20 [INFO|DP=25|PP=0|TP=2|ip-26-0-166-214]: Throughput logging complete
+12/10/2024 08:37:20 [INFO|DP=14|PP=0|TP=1|ip-26-0-164-236]: Throughput logging complete
+12/10/2024 08:37:20 [INFO|DP=31|PP=0|TP=2|ip-26-0-168-95]: Throughput logging complete
+12/10/2024 08:37:20 [INFO|DP=31|PP=0|TP=3|ip-26-0-168-95]: Throughput logging complete
+12/10/2024 08:37:20 [INFO|DP=22|PP=0|TP=0|ip-26-0-165-38]: Throughput logging complete
+12/10/2024 08:37:20 [INFO|DP=20|PP=0|TP=3|ip-26-0-165-24]: Throughput logging complete
+12/10/2024 08:37:20 [INFO|DP=7|PP=0|TP=1|ip-26-0-161-153]: Throughput logging complete
+12/10/2024 08:37:20 [INFO|DP=6|PP=0|TP=0|ip-26-0-161-153]: Throughput logging complete
+12/10/2024 08:37:20 [INFO|DP=7|PP=0|TP=2|ip-26-0-161-153]: Throughput logging complete
+12/10/2024 08:37:20 [INFO|DP=7|PP=0|TP=0|ip-26-0-161-153]: Throughput logging complete
+12/10/2024 08:37:20 [INFO|DP=25|PP=0|TP=0|ip-26-0-166-214]: Throughput logging complete
+12/10/2024 08:37:20 [INFO|DP=7|PP=0|TP=3|ip-26-0-161-153]: Throughput logging complete
+12/10/2024 08:37:20 [INFO|DP=30|PP=0|TP=2|ip-26-0-168-95]: Throughput logging complete
+12/10/2024 08:37:20 [INFO|DP=30|PP=0|TP=3|ip-26-0-168-95]: Throughput logging complete
+12/10/2024 08:37:20 [INFO|DP=30|PP=0|TP=1|ip-26-0-168-95]: Throughput logging complete
+12/10/2024 08:37:20 [INFO|DP=31|PP=0|TP=0|ip-26-0-168-95]: Throughput logging complete
+12/10/2024 08:37:20 [INFO|DP=31|PP=0|TP=1|ip-26-0-168-95]: Throughput logging complete
+12/10/2024 08:37:20 [INFO|DP=4|PP=0|TP=2|ip-26-0-161-142]: Throughput logging complete
+12/10/2024 08:37:20 [INFO|DP=30|PP=0|TP=0|ip-26-0-168-95]: Throughput logging complete
+12/10/2024 08:37:20 [INFO|DP=6|PP=0|TP=3|ip-26-0-161-153]: Throughput logging complete
+12/10/2024 08:37:20 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: | 13265919 | 1.34G_dp32_tp4_pp1_acc8_mbs2_seq2048_zero1_tpmodeALL_vocab131k | 16    | 2048    | 2   | 8           | 512 | 122.66  | 122.66  | 14860.36  | 200.62           | 116.32           | 118.29               | 457.01               | 264.40               | 263.74               | 3.09           | 5.92         | 32 | 1  | 4  | <nanotron.parallel.pipeline_parallel.engine.OneForwardOneBackwardPipelineEngine object at 0x7f052b0a8c10> | TensorParallelLinearMode.ALL_REDUCE | False         | 2048        | silu       | 16         | 32        | 8            | 2048    | 131072     | True                | torch.bfloat16 | 1          | 25                | True                    | 1.24G        | 310M         |
+12/10/2024 08:37:20 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: Throughput logging complete
+12/10/2024 08:37:20 [INFO|DP=0|PP=0|TP=1|ip-26-0-160-225]: Throughput logging complete
+12/10/2024 08:37:20 [INFO|DP=0|PP=0|TP=2|ip-26-0-160-225]: Throughput logging complete
+12/10/2024 08:37:20 [INFO|DP=0|PP=0|TP=3|ip-26-0-160-225]: Throughput logging complete
+12/10/2024 08:37:20 [INFO|DP=1|PP=0|TP=0|ip-26-0-160-225]: Throughput logging complete
+12/10/2024 08:37:20 [INFO|DP=1|PP=0|TP=3|ip-26-0-160-225]: Throughput logging complete
+12/10/2024 08:37:20 [INFO|DP=1|PP=0|TP=2|ip-26-0-160-225]: Throughput logging complete
+12/10/2024 08:37:20 [INFO|DP=1|PP=0|TP=1|ip-26-0-160-225]: Throughput logging complete
+srun: Job step aborted: Waiting up to 32 seconds for job step to finish.
+slurmstepd: error: *** JOB 13265919 ON ip-26-0-160-225 CANCELLED AT 2024-12-10T08:37:20 ***
+[2024-12-10 08:37:20,553] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-10 08:37:20,554] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 647496 closing signal SIGTERM
+[2024-12-10 08:37:20,554] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 647497 closing signal SIGTERM
+[2024-12-10 08:37:20,554] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 647498 closing signal SIGTERM
+[2024-12-10 08:37:20,554] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 647499 closing signal SIGTERM
+[2024-12-10 08:37:20,554] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-10 08:37:20,554] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-10 08:37:20,554] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 29689 closing signal SIGTERM
+[2024-12-10 08:37:20,554] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 123952 closing signal SIGTERM
+[2024-12-10 08:37:20,554] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 29690 closing signal SIGTERM
+[2024-12-10 08:37:20,555] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 29691 closing signal SIGTERM
+[2024-12-10 08:37:20,554] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 123953 closing signal SIGTERM
+[2024-12-10 08:37:20,554] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-10 08:37:20,555] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 123954 closing signal SIGTERM
+[2024-12-10 08:37:20,554] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 560396 closing signal SIGTERM
+[2024-12-10 08:37:20,555] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 123955 closing signal SIGTERM
+[2024-12-10 08:37:20,555] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 560397 closing signal SIGTERM
+[2024-12-10 08:37:20,553] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-10 08:37:20,555] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 560398 closing signal SIGTERM
+[2024-12-10 08:37:20,554] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 4010932 closing signal SIGTERM
+[2024-12-10 08:37:20,554] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 4010933 closing signal SIGTERM
+[2024-12-10 08:37:20,554] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 4010934 closing signal SIGTERM
+[2024-12-10 08:37:20,555] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 560399 closing signal SIGTERM
+[2024-12-10 08:37:20,554] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-10 08:37:20,554] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 84292 closing signal SIGTERM
+[2024-12-10 08:37:20,555] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 29692 closing signal SIGTERM
+[2024-12-10 08:37:20,555] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 84293 closing signal SIGTERM
+[2024-12-10 08:37:20,554] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 4010935 closing signal SIGTERM
+[2024-12-10 08:37:20,554] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-10 08:37:20,555] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 84294 closing signal SIGTERM
+[2024-12-10 08:37:20,555] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 84295 closing signal SIGTERM
+[2024-12-10 08:37:20,555] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 102818 closing signal SIGTERM
+[2024-12-10 08:37:20,555] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-10 08:37:20,555] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 39317 closing signal SIGTERM
+[2024-12-10 08:37:20,555] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 102819 closing signal SIGTERM
+[2024-12-10 08:37:20,555] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 39318 closing signal SIGTERM
+[2024-12-10 08:37:20,555] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 102820 closing signal SIGTERM
+[2024-12-10 08:37:20,556] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-10 08:37:20,555] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 39319 closing signal SIGTERM
+[2024-12-10 08:37:20,555] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 102821 closing signal SIGTERM
+[2024-12-10 08:37:20,555] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 39320 closing signal SIGTERM
+[2024-12-10 08:37:20,556] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 81753 closing signal SIGTERM
+[2024-12-10 08:37:20,556] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 81754 closing signal SIGTERM
+[2024-12-10 08:37:20,555] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 39321 closing signal SIGTERM
+[2024-12-10 08:37:20,556] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 81755 closing signal SIGTERM
+[2024-12-10 08:37:20,557] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 647500 closing signal SIGTERM
+[2024-12-10 08:37:20,557] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 647501 closing signal SIGTERM
+[2024-12-10 08:37:20,557] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 647502 closing signal SIGTERM
+[2024-12-10 08:37:20,556] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-10 08:37:20,555] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-10 08:37:20,556] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 30082 closing signal SIGTERM
+[2024-12-10 08:37:20,556] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 225320 closing signal SIGTERM
+[2024-12-10 08:37:20,556] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 30083 closing signal SIGTERM
+[2024-12-10 08:37:20,555] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-10 08:37:20,556] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 225321 closing signal SIGTERM
+[2024-12-10 08:37:20,556] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 39322 closing signal SIGTERM
+[2024-12-10 08:37:20,557] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 81756 closing signal SIGTERM
+[2024-12-10 08:37:20,556] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 30084 closing signal SIGTERM
+[2024-12-10 08:37:20,556] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3962907 closing signal SIGTERM
+[2024-12-10 08:37:20,556] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 225322 closing signal SIGTERM
+[2024-12-10 08:37:20,556] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 4010936 closing signal SIGTERM
+[2024-12-10 08:37:20,556] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 225323 closing signal SIGTERM
+[2024-12-10 08:37:20,556] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3962908 closing signal SIGTERM
+[2024-12-10 08:37:20,556] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 30085 closing signal SIGTERM
+[2024-12-10 08:37:20,556] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 30086 closing signal SIGTERM
+[2024-12-10 08:37:20,557] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 647503 closing signal SIGTERM
+[2024-12-10 08:37:20,557] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 30087 closing signal SIGTERM
+[2024-12-10 08:37:20,556] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3962909 closing signal SIGTERM
+[2024-12-10 08:37:20,556] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3962910 closing signal SIGTERM
+[2024-12-10 08:37:20,556] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 102822 closing signal SIGTERM
+[2024-12-10 08:37:20,557] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 29693 closing signal SIGTERM
+[2024-12-10 08:37:20,556] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 102823 closing signal SIGTERM
+[2024-12-10 08:37:20,558] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-10 08:37:20,557] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-10 08:37:20,557] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 29694 closing signal SIGTERM
+[2024-12-10 08:37:20,556] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 102824 closing signal SIGTERM
+[2024-12-10 08:37:20,558] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 859509 closing signal SIGTERM
+[2024-12-10 08:37:20,557] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 67539 closing signal SIGTERM
+[2024-12-10 08:37:20,556] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 102825 closing signal SIGTERM
+[2024-12-10 08:37:20,556] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 84296 closing signal SIGTERM
+[2024-12-10 08:37:20,557] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 67540 closing signal SIGTERM
+[2024-12-10 08:37:20,556] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 84297 closing signal SIGTERM
+[2024-12-10 08:37:20,558] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 859510 closing signal SIGTERM
+[2024-12-10 08:37:20,557] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 67541 closing signal SIGTERM
+[2024-12-10 08:37:20,558] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 859511 closing signal SIGTERM
+[2024-12-10 08:37:20,557] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 123956 closing signal SIGTERM
+[2024-12-10 08:37:20,557] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 67542 closing signal SIGTERM
+[2024-12-10 08:37:20,556] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 4010937 closing signal SIGTERM
+[2024-12-10 08:37:20,556] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 4010938 closing signal SIGTERM
+[2024-12-10 08:37:20,557] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 123957 closing signal SIGTERM
+[2024-12-10 08:37:20,557] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 560400 closing signal SIGTERM
+[2024-12-10 08:37:20,557] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 560401 closing signal SIGTERM
+[2024-12-10 08:37:20,558] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 81757 closing signal SIGTERM
+[2024-12-10 08:37:20,557] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 29695 closing signal SIGTERM
+[2024-12-10 08:37:20,558] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 560402 closing signal SIGTERM
+[2024-12-10 08:37:20,557] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 39323 closing signal SIGTERM
+[2024-12-10 08:37:20,557] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3962911 closing signal SIGTERM
+[2024-12-10 08:37:20,558] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 560403 closing signal SIGTERM
+[2024-12-10 08:37:20,557] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-10 08:37:20,557] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 4010939 closing signal SIGTERM
+[2024-12-10 08:37:20,557] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 123958 closing signal SIGTERM
+[2024-12-10 08:37:20,558] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2698214 closing signal SIGTERM
+[2024-12-10 08:37:20,558] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2698215 closing signal SIGTERM
+[2024-12-10 08:37:20,558] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2698216 closing signal SIGTERM
+[2024-12-10 08:37:20,556] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 84298 closing signal SIGTERM
+[2024-12-10 08:37:20,557] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 84299 closing signal SIGTERM
+[2024-12-10 08:37:20,558] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2698217 closing signal SIGTERM
+[2024-12-10 08:37:20,557] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 67543 closing signal SIGTERM
+[2024-12-10 08:37:20,558] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 81758 closing signal SIGTERM
+[2024-12-10 08:37:20,558] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2698218 closing signal SIGTERM
+[2024-12-10 08:37:20,558] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2698219 closing signal SIGTERM
+[2024-12-10 08:37:20,558] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 29696 closing signal SIGTERM
+[2024-12-10 08:37:20,559] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 859512 closing signal SIGTERM
+[2024-12-10 08:37:20,560] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 859513 closing signal SIGTERM
+[2024-12-10 08:37:20,560] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 859514 closing signal SIGTERM
+[2024-12-10 08:37:20,559] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 123959 closing signal SIGTERM
+[2024-12-10 08:37:20,560] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 859515 closing signal SIGTERM
+[2024-12-10 08:37:20,557] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3962912 closing signal SIGTERM
+[2024-12-10 08:37:20,557] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3962913 closing signal SIGTERM
+[2024-12-10 08:37:20,557] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3962914 closing signal SIGTERM
+[2024-12-10 08:37:20,558] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 225324 closing signal SIGTERM
+[2024-12-10 08:37:20,559] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 39324 closing signal SIGTERM
+[2024-12-10 08:37:20,558] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 225325 closing signal SIGTERM
+[2024-12-10 08:37:20,558] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 225326 closing signal SIGTERM
+[2024-12-10 08:37:20,558] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 225327 closing signal SIGTERM
+[2024-12-10 08:37:20,560] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 81759 closing signal SIGTERM
+[2024-12-10 08:37:20,561] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 859516 closing signal SIGTERM
+[2024-12-10 08:37:20,560] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 81760 closing signal SIGTERM
+[2024-12-10 08:37:20,559] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 67544 closing signal SIGTERM
+[2024-12-10 08:37:20,560] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 67545 closing signal SIGTERM
+[2024-12-10 08:37:20,560] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 67546 closing signal SIGTERM
+[2024-12-10 08:37:20,560] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 30088 closing signal SIGTERM
+[2024-12-10 08:37:20,562] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 30089 closing signal SIGTERM
+[2024-12-10 08:37:20,562] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2698220 closing signal SIGTERM
+[2024-12-10 08:37:20,562] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2698221 closing signal SIGTERM
+slurmstepd: error: *** STEP 13265919.0 ON ip-26-0-160-225 CANCELLED AT 2024-12-10T08:37:20 ***
+[2024-12-10 08:37:20,551] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-10 08:37:20,551] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 217936 closing signal SIGTERM
+[2024-12-10 08:37:20,552] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 217937 closing signal SIGTERM
+[2024-12-10 08:37:20,552] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 217938 closing signal SIGTERM
+[2024-12-10 08:37:20,553] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 217939 closing signal SIGTERM
+[2024-12-10 08:37:20,553] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 217940 closing signal SIGTERM
+[2024-12-10 08:37:20,553] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 217941 closing signal SIGTERM
+[2024-12-10 08:37:20,555] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 217942 closing signal SIGTERM
+[2024-12-10 08:37:20,555] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 217943 closing signal SIGTERM
diff --git a/logs/13266579-bench_1.14G_dp2_tp8_pp1_acc16_mbs16_seq8192_zero1_tpmodeRED_vocab32k.out b/logs/13266579-bench_1.14G_dp2_tp8_pp1_acc16_mbs16_seq8192_zero1_tpmodeRED_vocab32k.out
new file mode 100644
index 0000000000000000000000000000000000000000..dff439b8c5a00f0f13e1611611596e2c5b21d8f0
--- /dev/null
+++ b/logs/13266579-bench_1.14G_dp2_tp8_pp1_acc16_mbs16_seq8192_zero1_tpmodeRED_vocab32k.out
@@ -0,0 +1,644 @@
++ source /etc/profile.d/modules.sh
+++ . /usr/share/modules/init/bash
++++ unset _mlshdbg
++++ '[' 0 = 1 ']'
++++ unset _mlre _mlIFS
++++ '[' -n x ']'
++++ _mlIFS=' 	
+'
++++ IFS=' '
++++ '[' -n '' ']'
+++++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash autoinit
++++ _mlcode='module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+MODULEPATH_modshare=/etc/environment-modules/modules:1:/usr/share/modules/\$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1; export MODULEPATH_modshare;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+LOADEDMODULES=; export LOADEDMODULES;
+MODULEPATH=/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/\$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles; export MODULEPATH;
+test 0;'
++++ _mlret=0
++++ '[' -n x ']'
++++ IFS=' 	
+'
++++ unset _mlIFS
++++ unset _mlre _mlv _mlrv
++++ '[' 0 -eq 0 ']'
++++ eval 'module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+MODULEPATH_modshare=/etc/environment-modules/modules:1:/usr/share/modules/\$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1; export MODULEPATH_modshare;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+LOADEDMODULES=; export LOADEDMODULES;
+MODULEPATH=/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/\$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles; export MODULEPATH;
+test 0;'
+++++ MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl
+++++ export MODULES_CMD
+++++ ENV=/usr/share/modules/init/profile.sh
+++++ export ENV
+++++ MODULEPATH_modshare='/etc/environment-modules/modules:1:/usr/share/modules/$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1'
+++++ export MODULEPATH_modshare
+++++ BASH_ENV=/usr/share/modules/init/bash
+++++ export BASH_ENV
+++++ MODULESHOME=/usr/share/modules
+++++ export MODULESHOME
+++++ LOADEDMODULES=
+++++ export LOADEDMODULES
+++++ MODULEPATH='/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles'
+++++ export MODULEPATH
+++++ test 0
++++ '[' 0 = 1 ']'
++++ '[' -t 2 ']'
++++ export -f module
++++ export -f switchml
++++ '[' 5 -ge 3 ']'
++++ [[ ehxB =~ i ]]
++++ [[ ! :/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin: =~ :/usr/bin: ]]
++++ '[' '!' -n '' ']'
++++ MANPATH=:
++++ export MANPATH
+++++ manpath
++++ [[ ! :/admin/home/nouamane/.local/share/man:/fsx/nouamane/miniconda/envs/2-1-cu121/man:/fsx/nouamane/miniconda/envs/2-1-cu121/share/man:/fsx/nouamane/miniconda/man:/fsx/nouamane/miniconda/share/man:/opt/amazon/openmpi/share/man:/opt/amazon/efa/share/man:/opt/slurm/share/man:/usr/local/man:/usr/local/share/man:/usr/share/man:/admin/home/nouamane/.fzf/man:: =~ :/usr/share/man: ]]
++++ unset _mlcode _mlret
++++ '[' -n '' ']'
++ module load cuda/12.1
++ unset _mlshdbg
++ '[' 0 = 1 ']'
++ unset _mlre _mlIFS
++ '[' -n x ']'
++ _mlIFS=' 	
+'
++ IFS=' '
++ '[' -n '' ']'
+++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash load cuda/12.1
++ eval 'CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include;' export 'CPATH;
+LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:;' export 'LD_LIBRARY_PATH;
+FI_EFA_FORK_SAFE=1;' export 'FI_EFA_FORK_SAFE;
+MANPATH=/usr/local/cuda-12.1/share/man::;' export 'MANPATH;
+LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64;' export 'LIBRARY_PATH;
+_LMFILES_=/usr/share/modules/modulefiles/cuda/12.1;' export '_LMFILES_;
+LOADEDMODULES=cuda/12.1;' export 'LOADEDMODULES;
+MPI_PATH=/opt/amazon/openmpi;' export 'MPI_PATH;
+NCCL_HOME_modshare=/opt/nccl/build:1;' export 'NCCL_HOME_modshare;
+NCCL_PROTO=simple;' export 'NCCL_PROTO;
+MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1;' export 'MANPATH_modshare;
+LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1;' export 'LIBRARY_PATH_modshare;
+NCCL_SOCKET_IFNAME=enp;' export 'NCCL_SOCKET_IFNAME;
+AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl;' export 'AWS_OFI_NCCL_HOME;
+NCCL_HOME=/opt/nccl/build;' export 'NCCL_HOME;
+FI_PROVIDER=efa;' export 'FI_PROVIDER;
+AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1;' export 'AWS_OFI_NCCL_HOME_modshare;
+CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1;' export 'CPATH_modshare;
+LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2::1:/opt/nccl/build/lib:1:/opt/aws-ofi-nccl/lib:2:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/efa/lib:1;' export 'LD_LIBRARY_PATH_modshare;
+FI_EFA_ENABLE_SHM_TRANSFER=1;' export 'FI_EFA_ENABLE_SHM_TRANSFER;
+_LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1;' export '_LMFILES__modshare;
+LOADEDMODULES_modshare=cuda/12.1:1;' export 'LOADEDMODULES_modshare;
+MPI_PATH_modshare=/opt/amazon/openmpi:1;' export 'MPI_PATH_modshare;
+PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin;' export 'PATH;
+CUDA_HOME=/usr/local/cuda-12.1;' export 'CUDA_HOME;
+PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/usr/local/cuda-12.1/include:1:/opt/amazon/efa/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/usr/sbin:1:/fsx/nouamane/miniconda/bin:1:/opt/slurm/bin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:2:/opt/amazon/openmpi/bin:1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1;' export 'PATH_modshare;
+test' '0;'
+++ CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include
+++ export CPATH
+++ LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:
+++ export LD_LIBRARY_PATH
+++ FI_EFA_FORK_SAFE=1
+++ export FI_EFA_FORK_SAFE
+++ MANPATH=/usr/local/cuda-12.1/share/man::
+++ export MANPATH
+++ LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64
+++ export LIBRARY_PATH
+++ _LMFILES_=/usr/share/modules/modulefiles/cuda/12.1
+++ export _LMFILES_
+++ LOADEDMODULES=cuda/12.1
+++ export LOADEDMODULES
+++ MPI_PATH=/opt/amazon/openmpi
+++ export MPI_PATH
+++ NCCL_HOME_modshare=/opt/nccl/build:1
+++ export NCCL_HOME_modshare
+++ NCCL_PROTO=simple
+++ export NCCL_PROTO
+++ MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1
+++ export MANPATH_modshare
+++ LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1
+++ export LIBRARY_PATH_modshare
+++ NCCL_SOCKET_IFNAME=enp
+++ export NCCL_SOCKET_IFNAME
+++ AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl
+++ export AWS_OFI_NCCL_HOME
+++ NCCL_HOME=/opt/nccl/build
+++ export NCCL_HOME
+++ FI_PROVIDER=efa
+++ export FI_PROVIDER
+++ AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1
+++ export AWS_OFI_NCCL_HOME_modshare
+++ CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1
+++ export CPATH_modshare
+++ LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2::1:/opt/nccl/build/lib:1:/opt/aws-ofi-nccl/lib:2:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/efa/lib:1
+++ export LD_LIBRARY_PATH_modshare
+++ FI_EFA_ENABLE_SHM_TRANSFER=1
+++ export FI_EFA_ENABLE_SHM_TRANSFER
+++ _LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1
+++ export _LMFILES__modshare
+++ LOADEDMODULES_modshare=cuda/12.1:1
+++ export LOADEDMODULES_modshare
+++ MPI_PATH_modshare=/opt/amazon/openmpi:1
+++ export MPI_PATH_modshare
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ export PATH
+++ CUDA_HOME=/usr/local/cuda-12.1
+++ export CUDA_HOME
+++ PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/usr/local/cuda-12.1/include:1:/opt/amazon/efa/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/usr/sbin:1:/fsx/nouamane/miniconda/bin:1:/opt/slurm/bin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:2:/opt/amazon/openmpi/bin:1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1
+++ export PATH_modshare
+++ test 0
++ _mlstatus=0
++ '[' -n x ']'
++ IFS=' 	
+'
++ unset _mlre _mlv _mlrv _mlIFS
++ '[' -n '' ']'
++ unset _mlshdbg
++ return 0
++ source /fsx/nouamane/miniconda/bin/activate
+++ _CONDA_ROOT=/fsx/nouamane/miniconda
+++ . /fsx/nouamane/miniconda/etc/profile.d/conda.sh
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ '[' -z x ']'
+++ conda activate
+++ local cmd=activate
+++ case "$cmd" in
+++ __conda_activate activate
+++ '[' -n '' ']'
+++ local ask_conda
++++ PS1=
++++ __conda_exe shell.posix activate
++++ /fsx/nouamane/miniconda/bin/conda shell.posix activate
+++ ask_conda='. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_2='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\'''
+++ eval '. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_2='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\'''
++++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh
+++++ test -n ''
+++++ unset XML_CATALOG_FILES
+++++ unset xml_catalog_files_libxml2
++++ PS1='(base) '
++++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++++ export CONDA_PREFIX=/fsx/nouamane/miniconda
++++ CONDA_PREFIX=/fsx/nouamane/miniconda
++++ export CONDA_SHLVL=3
++++ CONDA_SHLVL=3
++++ export CONDA_DEFAULT_ENV=base
++++ CONDA_DEFAULT_ENV=base
++++ export 'CONDA_PROMPT_MODIFIER=(base) '
++++ CONDA_PROMPT_MODIFIER='(base) '
++++ export CONDA_PREFIX_2=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ CONDA_PREFIX_2=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ __conda_hashr
+++ '[' -n '' ']'
+++ '[' -n '' ']'
+++ hash -r
++ conda activate 2-1-cu121
++ local cmd=activate
++ case "$cmd" in
++ __conda_activate activate 2-1-cu121
++ '[' -n '' ']'
++ local ask_conda
+++ PS1='(base) '
+++ __conda_exe shell.posix activate 2-1-cu121
+++ /fsx/nouamane/miniconda/bin/conda shell.posix activate 2-1-cu121
++ ask_conda='PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''4'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_3='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
++ eval 'PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''4'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_3='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
+++ PS1='(2-1-cu121) '
+++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ export CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ export CONDA_SHLVL=4
+++ CONDA_SHLVL=4
+++ export CONDA_DEFAULT_ENV=2-1-cu121
+++ CONDA_DEFAULT_ENV=2-1-cu121
+++ export 'CONDA_PROMPT_MODIFIER=(2-1-cu121) '
+++ CONDA_PROMPT_MODIFIER='(2-1-cu121) '
+++ export CONDA_PREFIX_3=/fsx/nouamane/miniconda
+++ CONDA_PREFIX_3=/fsx/nouamane/miniconda
+++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ export _CE_M=
+++ _CE_M=
+++ export _CE_CONDA=
+++ _CE_CONDA=
+++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh
++++ test -n ''
++++ xml_catalog_files_libxml2=
++++ XML_CATALOG_FILES=
++++ conda_catalog_files=
++++ ifs_libxml2=' 	
+'
++++ IFS=' '
++++ rem=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ for pre in ${rem}
++++ test '' = /fsx/nouamane/miniconda/envs/2-1-cu121
++++ conda_catalog_files=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ rem=
++++ IFS=' 	
+'
++++ conda_catalog_files='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ export 'XML_CATALOG_FILES=file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ XML_CATALOG_FILES='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ unset conda_catalog_files ifs_libxml2 rem
++ __conda_hashr
++ '[' -n '' ']'
++ '[' -n '' ']'
++ hash -r
++ export PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++ PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ scontrol show hostnames 'ip-26-0-171-[56,230]'
++ export 'NODELIST=ip-26-0-171-56
+ip-26-0-171-230'
++ NODELIST='ip-26-0-171-56
+ip-26-0-171-230'
+++ head -n1
+++ scontrol show hostnames 'ip-26-0-171-[56,230]'
++ export MASTER_NODE=ip-26-0-171-56
++ MASTER_NODE=ip-26-0-171-56
++ export MASTER_PORT=12356
++ MASTER_PORT=12356
++ export NNODES=2
++ NNODES=2
++ export GPUS_PER_NODE=8
++ GPUS_PER_NODE=8
++ export WORLD_SIZE=16
++ WORLD_SIZE=16
++ export CUDA_DEVICE_MAX_CONNECTIONS=1
++ CUDA_DEVICE_MAX_CONNECTIONS=1
++ export NANOTRON_BENCHMARK=1
++ NANOTRON_BENCHMARK=1
++ echo 'Master node: ip-26-0-171-56'
+Master node: ip-26-0-171-56
++ echo 'All nodes: ip-26-0-171-56
+ip-26-0-171-230'
+All nodes: ip-26-0-171-56
+ip-26-0-171-230
++ echo 'World size: 16'
+World size: 16
++ srun torchrun --nnodes=2 --nproc_per_node=8 --rdzv_id=13266579 --rdzv_backend=c10d --rdzv_endpoint=ip-26-0-171-56:12356 run_train.py --config-file benchmark/configs/config_1.14G_dp2_tp8_pp1_acc16_mbs16_seq8192_zero1_tpmodeRED_vocab32k.yaml
+[2024-12-10 08:20:08,798] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-10 08:20:08,798] torch.distributed.run: [WARNING] 
+[2024-12-10 08:20:08,798] torch.distributed.run: [WARNING] *****************************************
+[2024-12-10 08:20:08,798] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-10 08:20:08,798] torch.distributed.run: [WARNING] *****************************************
+[2024-12-10 08:20:09,027] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-10 08:20:09,027] torch.distributed.run: [WARNING] 
+[2024-12-10 08:20:09,027] torch.distributed.run: [WARNING] *****************************************
+[2024-12-10 08:20:09,027] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-10 08:20:09,027] torch.distributed.run: [WARNING] *****************************************
+12/10/2024 08:20:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-230]: Config:
+12/10/2024 08:20:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-230]: Config(general=GeneralArgs(project='debug',
+12/10/2024 08:20:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-230]:                            run='1.14G_dp2_tp8_pp1_acc16_mbs16_seq8192_zero1_tpmodeRED_vocab32k',
+12/10/2024 08:20:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-230]:                            seed=42,
+12/10/2024 08:20:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-230]:                            step=None,
+12/10/2024 08:20:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-230]:                            consumed_train_samples=None,
+12/10/2024 08:20:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-230]:                            benchmark_csv_path=PosixPath('benchmark/results/bench_final.csv'),
+12/10/2024 08:20:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-230]:                            ignore_sanity_checks=True),
+12/10/2024 08:20:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-230]:        parallelism=ParallelismArgs(dp=2,
+12/10/2024 08:20:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-230]:                                    pp=1,
+12/10/2024 08:20:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-230]:                                    tp=8,
+12/10/2024 08:20:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-230]:                                    pp_engine=<nanotron.parallel.pipeline_parallel.engine.OneForwardOneBackwardPipelineEngine object at 0x7f33918acbe0>,
+12/10/2024 08:20:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-230]:                                    tp_mode=<TensorParallelLinearMode.REDUCE_SCATTER: 2>,
+12/10/2024 08:20:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-230]:                                    tp_linear_async_communication=True,
+12/10/2024 08:20:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-230]:                                    recompute_layer=False,
+12/10/2024 08:20:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-230]:                                    tp_recompute_allgather=True,
+12/10/2024 08:20:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-230]:                                    expert_parallel_size=1),
+12/10/2024 08:20:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-230]:        model=ModelArgs(model_config=LlamaConfig(bos_token_id=0,
+12/10/2024 08:20:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-230]:                                                 eos_token_id=0,
+12/10/2024 08:20:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-230]:                                                 hidden_act='silu',
+12/10/2024 08:20:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-230]:                                                 hidden_size=2048,
+12/10/2024 08:20:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-230]:                                                 initializer_range=0.02,
+12/10/2024 08:20:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-230]:                                                 intermediate_size=8192,
+12/10/2024 08:20:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-230]:                                                 is_llama_config=True,
+12/10/2024 08:20:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-230]:                                                 max_position_embeddings=8192,
+12/10/2024 08:20:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-230]:                                                 num_attention_heads=32,
+12/10/2024 08:20:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-230]:                                                 num_hidden_layers=16,
+12/10/2024 08:20:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-230]:                                                 num_key_value_heads=8,
+12/10/2024 08:20:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-230]:                                                 pad_token_id=None,
+12/10/2024 08:20:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-230]:                                                 pretraining_tp=1,
+12/10/2024 08:20:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-230]:                                                 rms_norm_eps=1e-05,
+12/10/2024 08:20:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-230]:                                                 rope_scaling=None,
+12/10/2024 08:20:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-230]:                                                 rope_theta=10000.0,
+12/10/2024 08:20:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-230]:                                                 rope_interleaved=False,
+12/10/2024 08:20:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-230]:                                                 tie_word_embeddings=True,
+12/10/2024 08:20:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-230]:                                                 use_cache=True,
+12/10/2024 08:20:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-230]:                                                 vocab_size=32768),
+12/10/2024 08:20:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-230]:                        init_method=RandomInit(std=0.02),
+12/10/2024 08:20:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-230]:                        dtype=torch.bfloat16,
+12/10/2024 08:20:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-230]:                        make_vocab_size_divisible_by=1,
+12/10/2024 08:20:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-230]:                        ddp_bucket_cap_mb=25),
+12/10/2024 08:20:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-230]:        tokenizer=TokenizerArgs(tokenizer_name_or_path='robot-test/dummy-tokenizer-wordlevel',
+12/10/2024 08:20:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-230]:                                tokenizer_revision=None,
+12/10/2024 08:20:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-230]:                                tokenizer_max_length=None),
+12/10/2024 08:20:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-230]:        checkpoints=CheckpointsArgs(checkpoints_path=PosixPath('checkpoints'),
+12/10/2024 08:20:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-230]:                                    checkpoint_interval=10000,
+12/10/2024 08:20:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-230]:                                    save_initial_state=False,
+12/10/2024 08:20:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-230]:                                    save_final_state=False,
+12/10/2024 08:20:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-230]:                                    resume_checkpoint_path=None,
+12/10/2024 08:20:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-230]:                                    checkpoints_path_is_shared_file_system=False),
+12/10/2024 08:20:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-230]:        logging=LoggingArgs(log_level='info',
+12/10/2024 08:20:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-230]:                            log_level_replica='info',
+12/10/2024 08:20:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-230]:                            iteration_step_info_interval=1),
+12/10/2024 08:20:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-230]:        tokens=TokensArgs(sequence_length=8192,
+12/10/2024 08:20:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-230]:                          train_steps=100,
+12/10/2024 08:20:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-230]:                          micro_batch_size=16,
+12/10/2024 08:20:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-230]:                          batch_accumulation_per_replica=16,
+12/10/2024 08:20:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-230]:                          val_check_interval=100,
+12/10/2024 08:20:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-230]:                          limit_val_batches=0,
+12/10/2024 08:20:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-230]:                          limit_test_batches=0),
+12/10/2024 08:20:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-230]:        optimizer=OptimizerArgs(optimizer_factory=AdamWOptimizerArgs(adam_eps=1e-08,
+12/10/2024 08:20:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-230]:                                                                     adam_beta1=0.9,
+12/10/2024 08:20:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-230]:                                                                     adam_beta2=0.95,
+12/10/2024 08:20:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-230]:                                                                     torch_adam_is_fused=True,
+12/10/2024 08:20:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-230]:                                                                     name='adamW'),
+12/10/2024 08:20:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-230]:                                zero_stage=1,
+12/10/2024 08:20:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-230]:                                weight_decay=0.01,
+12/10/2024 08:20:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-230]:                                clip_grad=1.0,
+12/10/2024 08:20:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-230]:                                accumulate_grad_in_fp32=True,
+12/10/2024 08:20:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-230]:                                learning_rate_scheduler=LRSchedulerArgs(learning_rate=0.0003,
+12/10/2024 08:20:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-230]:                                                                        lr_warmup_steps=2,
+12/10/2024 08:20:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-230]:                                                                        lr_warmup_style='linear',
+12/10/2024 08:20:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-230]:                                                                        lr_decay_style='cosine',
+12/10/2024 08:20:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-230]:                                                                        lr_decay_steps=13,
+12/10/2024 08:20:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-230]:                                                                        lr_decay_starting_step=None,
+12/10/2024 08:20:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-230]:                                                                        min_decay_lr=1e-05)),
+12/10/2024 08:20:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-230]:        data_stages=[DatasetStageArgs(name='Stable Training Stage',
+12/10/2024 08:20:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-230]:                                      start_training_step=1,
+12/10/2024 08:20:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-230]:                                      data=DataArgs(dataset=None,
+12/10/2024 08:20:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-230]:                                                    seed=42,
+12/10/2024 08:20:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-230]:                                                    num_loading_workers=1))],
+12/10/2024 08:20:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-230]:        profiler=None,
+12/10/2024 08:20:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-230]:        lighteval=None,
+12/10/2024 08:20:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-230]:        s3_upload=None)
+12/10/2024 08:20:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-230]: Model Config:
+12/10/2024 08:20:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-230]: LlamaConfig(bos_token_id=0,
+12/10/2024 08:20:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-230]:             eos_token_id=0,
+12/10/2024 08:20:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-230]:             hidden_act='silu',
+12/10/2024 08:20:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-230]:             hidden_size=2048,
+12/10/2024 08:20:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-230]:             initializer_range=0.02,
+12/10/2024 08:20:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-230]:             intermediate_size=8192,
+12/10/2024 08:20:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-230]:             is_llama_config=True,
+12/10/2024 08:20:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-230]:             max_position_embeddings=8192,
+12/10/2024 08:20:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-230]:             num_attention_heads=32,
+12/10/2024 08:20:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-230]:             num_hidden_layers=16,
+12/10/2024 08:20:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-230]:             num_key_value_heads=8,
+12/10/2024 08:20:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-230]:             pad_token_id=None,
+12/10/2024 08:20:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-230]:             pretraining_tp=1,
+12/10/2024 08:20:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-230]:             rms_norm_eps=1e-05,
+12/10/2024 08:20:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-230]:             rope_scaling=None,
+12/10/2024 08:20:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-230]:             rope_theta=10000.0,
+12/10/2024 08:20:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-230]:             rope_interleaved=False,
+12/10/2024 08:20:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-230]:             tie_word_embeddings=True,
+12/10/2024 08:20:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-230]:             use_cache=True,
+12/10/2024 08:20:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-230]:             vocab_size=32768)
+12/10/2024 08:20:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-230]: Building model..
+12/10/2024 08:20:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-230]: Initialize RoPE Theta = 10000.0
+12/10/2024 08:20:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-230]: Setting PP block ranks...
+12/10/2024 08:20:52 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-230]: Total number of parameters: 1.04G (1985.03MiB)
+12/10/2024 08:20:52 [INFO|DP=0|PP=0|TP=1|ip-26-0-171-230]: Local number of parameters: 130M (248.13MiB)
+12/10/2024 08:20:52 [INFO|DP=0|PP=0|TP=6|ip-26-0-171-230]: Local number of parameters: 130M (248.13MiB)
+12/10/2024 08:20:52 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-230]: Local number of parameters: 130M (248.13MiB)
+12/10/2024 08:20:52 [INFO|DP=0|PP=0|TP=4|ip-26-0-171-230]: Local number of parameters: 130M (248.13MiB)
+12/10/2024 08:20:52 [INFO|DP=0|PP=0|TP=2|ip-26-0-171-230]: Local number of parameters: 130M (248.13MiB)
+12/10/2024 08:20:52 [INFO|DP=0|PP=0|TP=7|ip-26-0-171-230]: Local number of parameters: 130M (248.13MiB)
+12/10/2024 08:20:52 [INFO|DP=0|PP=0|TP=5|ip-26-0-171-230]: Local number of parameters: 130M (248.13MiB)
+12/10/2024 08:20:52 [INFO|DP=0|PP=0|TP=3|ip-26-0-171-230]: Local number of parameters: 130M (248.13MiB)
+12/10/2024 08:20:52 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-230]: [After model building] Memory usage: 249.15MiB. Peak allocated: 5504.00MiB Peak reserved: 35218.00MiB
+12/10/2024 08:20:52 [INFO|DP=0|PP=0|TP=1|ip-26-0-171-230]: [After model building] Memory usage: 249.15MiB. Peak allocated: 5504.00MiB Peak reserved: 35218.00MiB
+12/10/2024 08:20:52 [INFO|DP=0|PP=0|TP=6|ip-26-0-171-230]: [After model building] Memory usage: 249.15MiB. Peak allocated: 5504.00MiB Peak reserved: 35218.00MiB
+12/10/2024 08:20:52 [INFO|DP=0|PP=0|TP=2|ip-26-0-171-230]: [After model building] Memory usage: 249.15MiB. Peak allocated: 5504.00MiB Peak reserved: 35218.00MiB
+12/10/2024 08:20:52 [INFO|DP=0|PP=0|TP=4|ip-26-0-171-230]: [After model building] Memory usage: 249.15MiB. Peak allocated: 5504.00MiB Peak reserved: 35218.00MiB
+12/10/2024 08:20:52 [INFO|DP=0|PP=0|TP=7|ip-26-0-171-230]: [After model building] Memory usage: 249.15MiB. Peak allocated: 5504.00MiB Peak reserved: 35218.00MiB
+12/10/2024 08:20:52 [INFO|DP=0|PP=0|TP=5|ip-26-0-171-230]: [After model building] Memory usage: 249.15MiB. Peak allocated: 5504.00MiB Peak reserved: 35218.00MiB
+12/10/2024 08:20:52 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-230]: No checkpoint path provided.
+12/10/2024 08:20:52 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-230]: Parametrizing model parameters using StandardParametrizator
+12/10/2024 08:20:52 [INFO|DP=0|PP=0|TP=3|ip-26-0-171-230]: [After model building] Memory usage: 249.15MiB. Peak allocated: 5504.00MiB Peak reserved: 35218.00MiB
+12/10/2024 08:20:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-230]: [Optimizer Building] Using LearningRateForSP as learning rate
+12/10/2024 08:20:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-230]: [ZeRO sharding] Size of optimizer params per rank:
+12/10/2024 08:20:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-230]: [ZeRO sharding] DP Rank 0 has 65M out of 130M (50.00%) params' optimizer states
+12/10/2024 08:20:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-230]: [ZeRO sharding] DP Rank 1 has 65M out of 130M (50.00%) params' optimizer states
+12/10/2024 08:20:54 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-230]: [Training Plan] Stage Stable Training Stage has 99 remaining training steps and has consumed 0 samples
+12/10/2024 08:20:54 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-230]: Using dummy data generator
+12/10/2024 08:20:54 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-230]: [Training Plan] There are 1 training stages 
+12/10/2024 08:20:54 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-230]: [Stage Stable Training Stage] start from step 1 
+12/10/2024 08:20:54 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-230]: 
+12/10/2024 08:20:54 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-230]: [Start training] datetime: 2024-12-10 08:20:54.512028 | mbs: 16 | grad_accum: 16 | global_batch_size: 512 | sequence_length: 8192 | train_steps: 100 | start_iteration_step: 0 | consumed_train_samples: 0
+wandb: WARNING Path /fsx/nouamane/.cache/wandb/wandb/ wasn't writable, using system temp directory.
+wandb: WARNING Path /fsx/nouamane/.cache/wandb/wandb/ wasn't writable, using system temp directory
+wandb: Tracking run with wandb version 0.16.0
+wandb: W&B syncing is set to `offline` in this directory.  
+wandb: Run `wandb online` or set WANDB_MODE=online to enable cloud syncing.
+12/10/2024 08:21:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-230]: Resuming training from stage Stable Training Stage, it has trained for 0 samples and has 99 remaining train steps
+12/10/2024 08:21:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-230]:  Memory usage: 993.54MiB. Peak allocated 5504.00MiB. Peak reserved: 35218.00MiB
+12/10/2024 08:21:13 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-230]:  Memory usage: 2099.80MiB. Peak allocated 29405.63MiB. Peak reserved: 55038.00MiB
+12/10/2024 08:21:13 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-230]: iteration: 1 / 100 | consumed_tokens: 4.19M | elapsed_time_per_iteration_ms: 10.7K | tokens_per_sec: 392K | tokens_per_sec_per_gpu: 24.5K | global_batch_size: 512 | lm_loss: 10.8 | lr: 0.00015 | model_tflops_per_gpu: 232 | hardware_tflops_per_gpu: 232 | grad_norm: 0.242 | cuda_memory_allocated: 2.73G | cuda_max_memory_reserved: 57.7G | hd_total_memory_tb: 312G | hd_used_memory_tb: 69.8G | hd_free_memory_tb: 242G
+12/10/2024 08:21:13 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-230]:  Memory usage: 2600.22MiB. Peak allocated 3220.54MiB. Peak reserved: 55046.00MiB
+12/10/2024 08:21:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-230]:  Memory usage: 2600.23MiB. Peak allocated 29905.56MiB. Peak reserved: 55070.00MiB
+12/10/2024 08:21:22 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-230]: iteration: 2 / 100 | consumed_tokens: 8.39M | elapsed_time_per_iteration_ms: 8.59K | tokens_per_sec: 488K | tokens_per_sec_per_gpu: 30.5K | global_batch_size: 512 | lm_loss: 10.8 | lr: 0.0003 | model_tflops_per_gpu: 289 | hardware_tflops_per_gpu: 289 | grad_norm: 0.242 | cuda_memory_allocated: 2.73G | cuda_max_memory_reserved: 57.7G | hd_total_memory_tb: 312G | hd_used_memory_tb: 69.8G | hd_free_memory_tb: 242G
+12/10/2024 08:21:22 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-230]:  Memory usage: 2600.22MiB. Peak allocated 3220.55MiB. Peak reserved: 55070.00MiB
+12/10/2024 08:21:30 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-230]:  Memory usage: 2600.23MiB. Peak allocated 29905.56MiB. Peak reserved: 55070.00MiB
+num_params
+{'total': 1040728064, 'local': 130091008}
+num_params
+{'total': 1040728064, 'local': 130091008}
+num_paramsnum_paramsnum_params
+
+
+{'total': 1040728064, 'local': 130091008}
+{'total': 1040728064, 'local': 130091008}
+{'total': 1040728064, 'local': 130091008}
+num_params
+{'total': 1040728064, 'local': 130091008}
+num_params
+num_params{'total': 1040728064, 'local': 130091008}
+
+{'total': 1040728064, 'local': 130091008}
+num_paramsnum_params
+
+num_params
+{'total': 1040728064, 'local': 130091008}
+{'total': 1040728064, 'local': 130091008}{'total': 1040728064, 'local': 130091008}
+
+num_params
+{'total': 1040728064, 'local': 130091008}
+num_params
+{'total': 1040728064, 'local': 130091008}
+num_params
+{'total': 1040728064, 'local': 130091008}
+num_params
+{'total': 1040728064, 'local': 130091008}
+12/10/2024 08:21:30 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-230]: iteration: 3 / 100 | consumed_tokens: 12.6M | elapsed_time_per_iteration_ms: 8.59K | tokens_per_sec: 488K | tokens_per_sec_per_gpu: 30.5K | global_batch_size: 512 | lm_loss: 10.8 | lr: 0.000296 | model_tflops_per_gpu: 289 | hardware_tflops_per_gpu: 289 | grad_norm: 0.289 | cuda_memory_allocated: 2.73G | cuda_max_memory_reserved: 57.7G | hd_total_memory_tb: 312G | hd_used_memory_tb: 69.8G | hd_free_memory_tb: 242G
+num_params
+{'total': 1040728064, 'local': 130091008}
+12/10/2024 08:21:30 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-230]: | job_id   | name                                                           | nodes | seq_len | mbs | batch_accum | gbs | mTFLOPs | hTFLOPs | tok/s/gpu | AllReduce (GB/s) | AllGather (GB/s) | ReduceScatter (GB/s) | AR Intra-node (GB/s) | AG Intra-node (GB/s) | RS Intra-node (GB/s) | Mem Alloc (GB) | Mem Res (GB) | dp | pp | tp | pp_engine                                                                                                 | tp_mode                                 | tp_async_comm | hidden_size | hidden_act | num_layers | num_heads | num_kv_heads | max_pos | vocab_size | tie_word_embeddings | dtype          | zero_stage | ddp_bucket_cap_mb | accumulate_grad_in_fp32 | Total Params | Local Params |
+12/10/2024 08:21:30 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-230]: | -------- | -------------------------------------------------------------- | ----- | ------- | --- | ----------- | --- | ------- | ------- | --------- | ---------------- | ---------------- | -------------------- | -------------------- | -------------------- | -------------------- | -------------- | ------------ | -- | -- | -- | --------------------------------------------------------------------------------------------------------- | --------------------------------------- | ------------- | ----------- | ---------- | ---------- | --------- | ------------ | ------- | ---------- | ------------------- | -------------- | ---------- | ----------------- | ----------------------- | ------------ | ------------ |
+12/10/2024 08:21:30 [INFO|DP=1|PP=0|TP=0|ip-26-0-171-56]: Throughput logging complete
+12/10/2024 08:21:30 [INFO|DP=1|PP=0|TP=3|ip-26-0-171-56]: Throughput logging complete
+12/10/2024 08:21:30 [INFO|DP=1|PP=0|TP=4|ip-26-0-171-56]: Throughput logging complete
+12/10/2024 08:21:30 [INFO|DP=1|PP=0|TP=2|ip-26-0-171-56]: Throughput logging complete
+12/10/2024 08:21:30 [INFO|DP=1|PP=0|TP=5|ip-26-0-171-56]: Throughput logging complete
+12/10/2024 08:21:30 [INFO|DP=1|PP=0|TP=6|ip-26-0-171-56]: Throughput logging complete
+12/10/2024 08:21:30 [INFO|DP=1|PP=0|TP=7|ip-26-0-171-56]: Throughput logging complete
+12/10/2024 08:21:30 [INFO|DP=1|PP=0|TP=1|ip-26-0-171-56]: Throughput logging complete
+srun: Job step aborted: Waiting up to 32 seconds for job step to finish.
+slurmstepd: error: *** STEP 13266579.0 ON ip-26-0-171-56 CANCELLED AT 2024-12-10T08:21:30 ***
+slurmstepd: error: *** JOB 13266579 ON ip-26-0-171-56 CANCELLED AT 2024-12-10T08:21:30 ***
+[2024-12-10 08:21:30,703] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-10 08:21:30,703] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 149927 closing signal SIGTERM
+[2024-12-10 08:21:30,703] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 149928 closing signal SIGTERM
+[2024-12-10 08:21:30,703] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 149929 closing signal SIGTERM
+[2024-12-10 08:21:30,704] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 149930 closing signal SIGTERM
+[2024-12-10 08:21:30,705] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 149931 closing signal SIGTERM
+[2024-12-10 08:21:30,705] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 149932 closing signal SIGTERM
+[2024-12-10 08:21:30,706] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 149933 closing signal SIGTERM
+[2024-12-10 08:21:30,706] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 149934 closing signal SIGTERM
diff --git a/logs/13266667-bench_1.14G_dp4_tp8_pp1_acc32_mbs4_seq2048_zero1_tpmodeRED_vocab32k.out b/logs/13266667-bench_1.14G_dp4_tp8_pp1_acc32_mbs4_seq2048_zero1_tpmodeRED_vocab32k.out
new file mode 100644
index 0000000000000000000000000000000000000000..3bd23669cad4a4da2d4c920aaba1a002132fa3d2
--- /dev/null
+++ b/logs/13266667-bench_1.14G_dp4_tp8_pp1_acc32_mbs4_seq2048_zero1_tpmodeRED_vocab32k.out
@@ -0,0 +1,749 @@
++ source /etc/profile.d/modules.sh
+++ . /usr/share/modules/init/bash
++++ unset _mlshdbg
++++ '[' 0 = 1 ']'
++++ unset _mlre _mlIFS
++++ '[' -n x ']'
++++ _mlIFS=' 	
+'
++++ IFS=' '
++++ '[' -n '' ']'
+++++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash autoinit
++++ _mlcode='module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+MODULEPATH_modshare=/etc/environment-modules/modules:1:/usr/share/modules/\$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1; export MODULEPATH_modshare;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+LOADEDMODULES=; export LOADEDMODULES;
+MODULEPATH=/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/\$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles; export MODULEPATH;
+test 0;'
++++ _mlret=0
++++ '[' -n x ']'
++++ IFS=' 	
+'
++++ unset _mlIFS
++++ unset _mlre _mlv _mlrv
++++ '[' 0 -eq 0 ']'
++++ eval 'module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+MODULEPATH_modshare=/etc/environment-modules/modules:1:/usr/share/modules/\$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1; export MODULEPATH_modshare;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+LOADEDMODULES=; export LOADEDMODULES;
+MODULEPATH=/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/\$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles; export MODULEPATH;
+test 0;'
+++++ MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl
+++++ export MODULES_CMD
+++++ ENV=/usr/share/modules/init/profile.sh
+++++ export ENV
+++++ MODULEPATH_modshare='/etc/environment-modules/modules:1:/usr/share/modules/$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1'
+++++ export MODULEPATH_modshare
+++++ BASH_ENV=/usr/share/modules/init/bash
+++++ export BASH_ENV
+++++ MODULESHOME=/usr/share/modules
+++++ export MODULESHOME
+++++ LOADEDMODULES=
+++++ export LOADEDMODULES
+++++ MODULEPATH='/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles'
+++++ export MODULEPATH
+++++ test 0
++++ '[' 0 = 1 ']'
++++ '[' -t 2 ']'
++++ export -f module
++++ export -f switchml
++++ '[' 5 -ge 3 ']'
++++ [[ ehxB =~ i ]]
++++ [[ ! :/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin: =~ :/usr/bin: ]]
++++ '[' '!' -n '' ']'
++++ MANPATH=:
++++ export MANPATH
+++++ manpath
++++ [[ ! :/admin/home/nouamane/.local/share/man:/fsx/nouamane/miniconda/envs/2-1-cu121/man:/fsx/nouamane/miniconda/envs/2-1-cu121/share/man:/fsx/nouamane/miniconda/man:/fsx/nouamane/miniconda/share/man:/opt/amazon/openmpi/share/man:/opt/amazon/efa/share/man:/opt/slurm/share/man:/usr/local/man:/usr/local/share/man:/usr/share/man:/admin/home/nouamane/.fzf/man:: =~ :/usr/share/man: ]]
++++ unset _mlcode _mlret
++++ '[' -n '' ']'
++ module load cuda/12.1
++ unset _mlshdbg
++ '[' 0 = 1 ']'
++ unset _mlre _mlIFS
++ '[' -n x ']'
++ _mlIFS=' 	
+'
++ IFS=' '
++ '[' -n '' ']'
+++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash load cuda/12.1
++ eval 'CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include;' export 'CPATH;
+LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:;' export 'LD_LIBRARY_PATH;
+FI_EFA_FORK_SAFE=1;' export 'FI_EFA_FORK_SAFE;
+MANPATH=/usr/local/cuda-12.1/share/man::;' export 'MANPATH;
+LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64;' export 'LIBRARY_PATH;
+_LMFILES_=/usr/share/modules/modulefiles/cuda/12.1;' export '_LMFILES_;
+LOADEDMODULES=cuda/12.1;' export 'LOADEDMODULES;
+MPI_PATH=/opt/amazon/openmpi;' export 'MPI_PATH;
+NCCL_HOME_modshare=/opt/nccl/build:1;' export 'NCCL_HOME_modshare;
+NCCL_PROTO=simple;' export 'NCCL_PROTO;
+MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1;' export 'MANPATH_modshare;
+LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1;' export 'LIBRARY_PATH_modshare;
+NCCL_SOCKET_IFNAME=enp;' export 'NCCL_SOCKET_IFNAME;
+AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl;' export 'AWS_OFI_NCCL_HOME;
+NCCL_HOME=/opt/nccl/build;' export 'NCCL_HOME;
+FI_PROVIDER=efa;' export 'FI_PROVIDER;
+AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1;' export 'AWS_OFI_NCCL_HOME_modshare;
+CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1;' export 'CPATH_modshare;
+LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2::1:/opt/nccl/build/lib:1:/opt/aws-ofi-nccl/lib:2:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/efa/lib:1;' export 'LD_LIBRARY_PATH_modshare;
+FI_EFA_ENABLE_SHM_TRANSFER=1;' export 'FI_EFA_ENABLE_SHM_TRANSFER;
+_LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1;' export '_LMFILES__modshare;
+LOADEDMODULES_modshare=cuda/12.1:1;' export 'LOADEDMODULES_modshare;
+MPI_PATH_modshare=/opt/amazon/openmpi:1;' export 'MPI_PATH_modshare;
+PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin;' export 'PATH;
+CUDA_HOME=/usr/local/cuda-12.1;' export 'CUDA_HOME;
+PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/usr/local/cuda-12.1/include:1:/opt/amazon/efa/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/usr/sbin:1:/fsx/nouamane/miniconda/bin:1:/opt/slurm/bin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:2:/opt/amazon/openmpi/bin:1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1;' export 'PATH_modshare;
+test' '0;'
+++ CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include
+++ export CPATH
+++ LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:
+++ export LD_LIBRARY_PATH
+++ FI_EFA_FORK_SAFE=1
+++ export FI_EFA_FORK_SAFE
+++ MANPATH=/usr/local/cuda-12.1/share/man::
+++ export MANPATH
+++ LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64
+++ export LIBRARY_PATH
+++ _LMFILES_=/usr/share/modules/modulefiles/cuda/12.1
+++ export _LMFILES_
+++ LOADEDMODULES=cuda/12.1
+++ export LOADEDMODULES
+++ MPI_PATH=/opt/amazon/openmpi
+++ export MPI_PATH
+++ NCCL_HOME_modshare=/opt/nccl/build:1
+++ export NCCL_HOME_modshare
+++ NCCL_PROTO=simple
+++ export NCCL_PROTO
+++ MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1
+++ export MANPATH_modshare
+++ LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1
+++ export LIBRARY_PATH_modshare
+++ NCCL_SOCKET_IFNAME=enp
+++ export NCCL_SOCKET_IFNAME
+++ AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl
+++ export AWS_OFI_NCCL_HOME
+++ NCCL_HOME=/opt/nccl/build
+++ export NCCL_HOME
+++ FI_PROVIDER=efa
+++ export FI_PROVIDER
+++ AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1
+++ export AWS_OFI_NCCL_HOME_modshare
+++ CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1
+++ export CPATH_modshare
+++ LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2::1:/opt/nccl/build/lib:1:/opt/aws-ofi-nccl/lib:2:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/efa/lib:1
+++ export LD_LIBRARY_PATH_modshare
+++ FI_EFA_ENABLE_SHM_TRANSFER=1
+++ export FI_EFA_ENABLE_SHM_TRANSFER
+++ _LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1
+++ export _LMFILES__modshare
+++ LOADEDMODULES_modshare=cuda/12.1:1
+++ export LOADEDMODULES_modshare
+++ MPI_PATH_modshare=/opt/amazon/openmpi:1
+++ export MPI_PATH_modshare
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ export PATH
+++ CUDA_HOME=/usr/local/cuda-12.1
+++ export CUDA_HOME
+++ PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/usr/local/cuda-12.1/include:1:/opt/amazon/efa/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/usr/sbin:1:/fsx/nouamane/miniconda/bin:1:/opt/slurm/bin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:2:/opt/amazon/openmpi/bin:1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1
+++ export PATH_modshare
+++ test 0
++ _mlstatus=0
++ '[' -n x ']'
++ IFS=' 	
+'
++ unset _mlre _mlv _mlrv _mlIFS
++ '[' -n '' ']'
++ unset _mlshdbg
++ return 0
++ source /fsx/nouamane/miniconda/bin/activate
+++ _CONDA_ROOT=/fsx/nouamane/miniconda
+++ . /fsx/nouamane/miniconda/etc/profile.d/conda.sh
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ '[' -z x ']'
+++ conda activate
+++ local cmd=activate
+++ case "$cmd" in
+++ __conda_activate activate
+++ '[' -n '' ']'
+++ local ask_conda
++++ PS1=
++++ __conda_exe shell.posix activate
++++ /fsx/nouamane/miniconda/bin/conda shell.posix activate
+++ ask_conda='. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_2='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\'''
+++ eval '. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_2='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\'''
++++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh
+++++ test -n ''
+++++ unset XML_CATALOG_FILES
+++++ unset xml_catalog_files_libxml2
++++ PS1='(base) '
++++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++++ export CONDA_PREFIX=/fsx/nouamane/miniconda
++++ CONDA_PREFIX=/fsx/nouamane/miniconda
++++ export CONDA_SHLVL=3
++++ CONDA_SHLVL=3
++++ export CONDA_DEFAULT_ENV=base
++++ CONDA_DEFAULT_ENV=base
++++ export 'CONDA_PROMPT_MODIFIER=(base) '
++++ CONDA_PROMPT_MODIFIER='(base) '
++++ export CONDA_PREFIX_2=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ CONDA_PREFIX_2=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ __conda_hashr
+++ '[' -n '' ']'
+++ '[' -n '' ']'
+++ hash -r
++ conda activate 2-1-cu121
++ local cmd=activate
++ case "$cmd" in
++ __conda_activate activate 2-1-cu121
++ '[' -n '' ']'
++ local ask_conda
+++ PS1='(base) '
+++ __conda_exe shell.posix activate 2-1-cu121
+++ /fsx/nouamane/miniconda/bin/conda shell.posix activate 2-1-cu121
++ ask_conda='PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''4'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_3='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
++ eval 'PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''4'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_3='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
+++ PS1='(2-1-cu121) '
+++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ export CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ export CONDA_SHLVL=4
+++ CONDA_SHLVL=4
+++ export CONDA_DEFAULT_ENV=2-1-cu121
+++ CONDA_DEFAULT_ENV=2-1-cu121
+++ export 'CONDA_PROMPT_MODIFIER=(2-1-cu121) '
+++ CONDA_PROMPT_MODIFIER='(2-1-cu121) '
+++ export CONDA_PREFIX_3=/fsx/nouamane/miniconda
+++ CONDA_PREFIX_3=/fsx/nouamane/miniconda
+++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ export _CE_M=
+++ _CE_M=
+++ export _CE_CONDA=
+++ _CE_CONDA=
+++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh
++++ test -n ''
++++ xml_catalog_files_libxml2=
++++ XML_CATALOG_FILES=
++++ conda_catalog_files=
++++ ifs_libxml2=' 	
+'
++++ IFS=' '
++++ rem=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ for pre in ${rem}
++++ test '' = /fsx/nouamane/miniconda/envs/2-1-cu121
++++ conda_catalog_files=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ rem=
++++ IFS=' 	
+'
++++ conda_catalog_files='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ export 'XML_CATALOG_FILES=file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ XML_CATALOG_FILES='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ unset conda_catalog_files ifs_libxml2 rem
++ __conda_hashr
++ '[' -n '' ']'
++ '[' -n '' ']'
++ hash -r
++ export PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++ PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ scontrol show hostnames 'ip-26-0-170-[143,160],ip-26-0-171-[88,102]'
++ export 'NODELIST=ip-26-0-170-143
+ip-26-0-170-160
+ip-26-0-171-88
+ip-26-0-171-102'
++ NODELIST='ip-26-0-170-143
+ip-26-0-170-160
+ip-26-0-171-88
+ip-26-0-171-102'
+++ head -n1
+++ scontrol show hostnames 'ip-26-0-170-[143,160],ip-26-0-171-[88,102]'
++ export MASTER_NODE=ip-26-0-170-143
++ MASTER_NODE=ip-26-0-170-143
++ export MASTER_PORT=12356
++ MASTER_PORT=12356
++ export NNODES=4
++ NNODES=4
++ export GPUS_PER_NODE=8
++ GPUS_PER_NODE=8
++ export WORLD_SIZE=32
++ WORLD_SIZE=32
++ export CUDA_DEVICE_MAX_CONNECTIONS=1
++ CUDA_DEVICE_MAX_CONNECTIONS=1
++ export NANOTRON_BENCHMARK=1
++ NANOTRON_BENCHMARK=1
++ echo 'Master node: ip-26-0-170-143'
+Master node: ip-26-0-170-143
++ echo 'All nodes: ip-26-0-170-143
+ip-26-0-170-160
+ip-26-0-171-88
+ip-26-0-171-102'
+All nodes: ip-26-0-170-143
+ip-26-0-170-160
+ip-26-0-171-88
+ip-26-0-171-102
++ echo 'World size: 32'
+World size: 32
++ srun torchrun --nnodes=4 --nproc_per_node=8 --rdzv_id=13266667 --rdzv_backend=c10d --rdzv_endpoint=ip-26-0-170-143:12356 run_train.py --config-file benchmark/configs/config_1.14G_dp4_tp8_pp1_acc32_mbs4_seq2048_zero1_tpmodeRED_vocab32k.yaml
+[2024-12-10 09:15:14,225] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-10 09:15:14,225] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-10 09:15:14,236] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-10 09:15:14,225] torch.distributed.run: [WARNING] 
+[2024-12-10 09:15:14,225] torch.distributed.run: [WARNING] *****************************************
+[2024-12-10 09:15:14,225] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-10 09:15:14,225] torch.distributed.run: [WARNING] *****************************************
+[2024-12-10 09:15:14,225] torch.distributed.run: [WARNING] 
+[2024-12-10 09:15:14,225] torch.distributed.run: [WARNING] *****************************************
+[2024-12-10 09:15:14,225] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-10 09:15:14,225] torch.distributed.run: [WARNING] *****************************************
+[2024-12-10 09:15:14,236] torch.distributed.run: [WARNING] 
+[2024-12-10 09:15:14,236] torch.distributed.run: [WARNING] *****************************************
+[2024-12-10 09:15:14,236] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-10 09:15:14,236] torch.distributed.run: [WARNING] *****************************************
+[2024-12-10 09:15:14,333] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-10 09:15:14,333] torch.distributed.run: [WARNING] 
+[2024-12-10 09:15:14,333] torch.distributed.run: [WARNING] *****************************************
+[2024-12-10 09:15:14,333] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-10 09:15:14,333] torch.distributed.run: [WARNING] *****************************************
+12/10/2024 09:15:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]: Config:
+12/10/2024 09:15:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]: Config(general=GeneralArgs(project='debug',
+12/10/2024 09:15:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]:                            run='1.14G_dp4_tp8_pp1_acc32_mbs4_seq2048_zero1_tpmodeRED_vocab32k',
+12/10/2024 09:15:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]:                            seed=42,
+12/10/2024 09:15:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]:                            step=None,
+12/10/2024 09:15:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]:                            consumed_train_samples=None,
+12/10/2024 09:15:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]:                            benchmark_csv_path=PosixPath('benchmark/results/bench_final.csv'),
+12/10/2024 09:15:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]:                            ignore_sanity_checks=True),
+12/10/2024 09:15:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]:        parallelism=ParallelismArgs(dp=4,
+12/10/2024 09:15:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]:                                    pp=1,
+12/10/2024 09:15:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]:                                    tp=8,
+12/10/2024 09:15:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]:                                    pp_engine=<nanotron.parallel.pipeline_parallel.engine.OneForwardOneBackwardPipelineEngine object at 0x7fe9fb5b8d00>,
+12/10/2024 09:15:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]:                                    tp_mode=<TensorParallelLinearMode.REDUCE_SCATTER: 2>,
+12/10/2024 09:15:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]:                                    tp_linear_async_communication=True,
+12/10/2024 09:15:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]:                                    recompute_layer=False,
+12/10/2024 09:15:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]:                                    tp_recompute_allgather=True,
+12/10/2024 09:15:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]:                                    expert_parallel_size=1),
+12/10/2024 09:15:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]:        model=ModelArgs(model_config=LlamaConfig(bos_token_id=0,
+12/10/2024 09:15:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]:                                                 eos_token_id=0,
+12/10/2024 09:15:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]:                                                 hidden_act='silu',
+12/10/2024 09:15:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]:                                                 hidden_size=2048,
+12/10/2024 09:15:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]:                                                 initializer_range=0.02,
+12/10/2024 09:15:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]:                                                 intermediate_size=8192,
+12/10/2024 09:15:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]:                                                 is_llama_config=True,
+12/10/2024 09:15:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]:                                                 max_position_embeddings=2048,
+12/10/2024 09:15:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]:                                                 num_attention_heads=32,
+12/10/2024 09:15:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]:                                                 num_hidden_layers=16,
+12/10/2024 09:15:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]:                                                 num_key_value_heads=8,
+12/10/2024 09:15:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]:                                                 pad_token_id=None,
+12/10/2024 09:15:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]:                                                 pretraining_tp=1,
+12/10/2024 09:15:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]:                                                 rms_norm_eps=1e-05,
+12/10/2024 09:15:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]:                                                 rope_scaling=None,
+12/10/2024 09:15:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]:                                                 rope_theta=10000.0,
+12/10/2024 09:15:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]:                                                 rope_interleaved=False,
+12/10/2024 09:15:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]:                                                 tie_word_embeddings=True,
+12/10/2024 09:15:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]:                                                 use_cache=True,
+12/10/2024 09:15:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]:                                                 vocab_size=32768),
+12/10/2024 09:15:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]:                        init_method=RandomInit(std=0.02),
+12/10/2024 09:15:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]:                        dtype=torch.bfloat16,
+12/10/2024 09:15:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]:                        make_vocab_size_divisible_by=1,
+12/10/2024 09:15:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]:                        ddp_bucket_cap_mb=25),
+12/10/2024 09:15:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]:        tokenizer=TokenizerArgs(tokenizer_name_or_path='robot-test/dummy-tokenizer-wordlevel',
+12/10/2024 09:15:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]:                                tokenizer_revision=None,
+12/10/2024 09:15:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]:                                tokenizer_max_length=None),
+12/10/2024 09:15:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]:        checkpoints=CheckpointsArgs(checkpoints_path=PosixPath('checkpoints'),
+12/10/2024 09:15:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]:                                    checkpoint_interval=10000,
+12/10/2024 09:15:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]:                                    save_initial_state=False,
+12/10/2024 09:15:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]:                                    save_final_state=False,
+12/10/2024 09:15:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]:                                    resume_checkpoint_path=None,
+12/10/2024 09:15:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]:                                    checkpoints_path_is_shared_file_system=False),
+12/10/2024 09:15:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]:        logging=LoggingArgs(log_level='info',
+12/10/2024 09:15:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]:                            log_level_replica='info',
+12/10/2024 09:15:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]:                            iteration_step_info_interval=1),
+12/10/2024 09:15:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]:        tokens=TokensArgs(sequence_length=2048,
+12/10/2024 09:15:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]:                          train_steps=100,
+12/10/2024 09:15:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]:                          micro_batch_size=4,
+12/10/2024 09:15:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]:                          batch_accumulation_per_replica=32,
+12/10/2024 09:15:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]:                          val_check_interval=100,
+12/10/2024 09:15:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]:                          limit_val_batches=0,
+12/10/2024 09:15:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]:                          limit_test_batches=0),
+12/10/2024 09:15:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]:        optimizer=OptimizerArgs(optimizer_factory=AdamWOptimizerArgs(adam_eps=1e-08,
+12/10/2024 09:15:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]:                                                                     adam_beta1=0.9,
+12/10/2024 09:15:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]:                                                                     adam_beta2=0.95,
+12/10/2024 09:15:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]:                                                                     torch_adam_is_fused=True,
+12/10/2024 09:15:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]:                                                                     name='adamW'),
+12/10/2024 09:15:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]:                                zero_stage=1,
+12/10/2024 09:15:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]:                                weight_decay=0.01,
+12/10/2024 09:15:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]:                                clip_grad=1.0,
+12/10/2024 09:15:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]:                                accumulate_grad_in_fp32=True,
+12/10/2024 09:15:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]:                                learning_rate_scheduler=LRSchedulerArgs(learning_rate=0.0003,
+12/10/2024 09:15:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]:                                                                        lr_warmup_steps=2,
+12/10/2024 09:15:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]:                                                                        lr_warmup_style='linear',
+12/10/2024 09:15:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]:                                                                        lr_decay_style='cosine',
+12/10/2024 09:15:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]:                                                                        lr_decay_steps=13,
+12/10/2024 09:15:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]:                                                                        lr_decay_starting_step=None,
+12/10/2024 09:15:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]:                                                                        min_decay_lr=1e-05)),
+12/10/2024 09:15:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]:        data_stages=[DatasetStageArgs(name='Stable Training Stage',
+12/10/2024 09:15:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]:                                      start_training_step=1,
+12/10/2024 09:15:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]:                                      data=DataArgs(dataset=None,
+12/10/2024 09:15:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]:                                                    seed=42,
+12/10/2024 09:15:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]:                                                    num_loading_workers=1))],
+12/10/2024 09:15:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]:        profiler=None,
+12/10/2024 09:15:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]:        lighteval=None,
+12/10/2024 09:15:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]:        s3_upload=None)
+12/10/2024 09:15:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]: Model Config:
+12/10/2024 09:15:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]: LlamaConfig(bos_token_id=0,
+12/10/2024 09:15:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]:             eos_token_id=0,
+12/10/2024 09:15:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]:             hidden_act='silu',
+12/10/2024 09:15:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]:             hidden_size=2048,
+12/10/2024 09:15:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]:             initializer_range=0.02,
+12/10/2024 09:15:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]:             intermediate_size=8192,
+12/10/2024 09:15:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]:             is_llama_config=True,
+12/10/2024 09:15:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]:             max_position_embeddings=2048,
+12/10/2024 09:15:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]:             num_attention_heads=32,
+12/10/2024 09:15:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]:             num_hidden_layers=16,
+12/10/2024 09:15:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]:             num_key_value_heads=8,
+12/10/2024 09:15:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]:             pad_token_id=None,
+12/10/2024 09:15:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]:             pretraining_tp=1,
+12/10/2024 09:15:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]:             rms_norm_eps=1e-05,
+12/10/2024 09:15:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]:             rope_scaling=None,
+12/10/2024 09:15:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]:             rope_theta=10000.0,
+12/10/2024 09:15:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]:             rope_interleaved=False,
+12/10/2024 09:15:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]:             tie_word_embeddings=True,
+12/10/2024 09:15:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]:             use_cache=True,
+12/10/2024 09:15:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]:             vocab_size=32768)
+12/10/2024 09:15:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]: Building model..
+12/10/2024 09:15:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]: Initialize RoPE Theta = 10000.0
+12/10/2024 09:15:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]: Setting PP block ranks...
+12/10/2024 09:15:58 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]: Total number of parameters: 1.04G (1985.03MiB)
+12/10/2024 09:15:58 [INFO|DP=0|PP=0|TP=3|ip-26-0-170-143]: Local number of parameters: 130M (248.13MiB)
+12/10/2024 09:15:58 [INFO|DP=0|PP=0|TP=1|ip-26-0-170-143]: Local number of parameters: 130M (248.13MiB)
+12/10/2024 09:15:58 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]: Local number of parameters: 130M (248.13MiB)
+12/10/2024 09:15:58 [INFO|DP=0|PP=0|TP=5|ip-26-0-170-143]: Local number of parameters: 130M (248.13MiB)
+12/10/2024 09:15:58 [INFO|DP=0|PP=0|TP=4|ip-26-0-170-143]: Local number of parameters: 130M (248.13MiB)
+12/10/2024 09:15:58 [INFO|DP=0|PP=0|TP=6|ip-26-0-170-143]: Local number of parameters: 130M (248.13MiB)
+12/10/2024 09:15:58 [INFO|DP=0|PP=0|TP=7|ip-26-0-170-143]: Local number of parameters: 130M (248.13MiB)
+12/10/2024 09:15:58 [INFO|DP=0|PP=0|TP=2|ip-26-0-170-143]: Local number of parameters: 130M (248.13MiB)
+12/10/2024 09:15:58 [INFO|DP=0|PP=0|TP=3|ip-26-0-170-143]: [After model building] Memory usage: 251.15MiB. Peak allocated: 5440.00MiB Peak reserved: 22866.00MiB
+12/10/2024 09:15:58 [INFO|DP=0|PP=0|TP=5|ip-26-0-170-143]: [After model building] Memory usage: 251.15MiB. Peak allocated: 5440.00MiB Peak reserved: 24914.00MiB
+12/10/2024 09:15:58 [INFO|DP=0|PP=0|TP=6|ip-26-0-170-143]: [After model building] Memory usage: 251.15MiB. Peak allocated: 5440.00MiB Peak reserved: 22866.00MiB
+12/10/2024 09:15:58 [INFO|DP=0|PP=0|TP=1|ip-26-0-170-143]: [After model building] Memory usage: 251.15MiB. Peak allocated: 5440.00MiB Peak reserved: 21842.00MiB
+12/10/2024 09:15:58 [INFO|DP=0|PP=0|TP=7|ip-26-0-170-143]: [After model building] Memory usage: 251.15MiB. Peak allocated: 5440.00MiB Peak reserved: 22866.00MiB
+12/10/2024 09:15:58 [INFO|DP=0|PP=0|TP=4|ip-26-0-170-143]: [After model building] Memory usage: 251.15MiB. Peak allocated: 5440.00MiB Peak reserved: 22866.00MiB
+12/10/2024 09:15:58 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]: [After model building] Memory usage: 251.15MiB. Peak allocated: 5440.00MiB Peak reserved: 21842.00MiB
+12/10/2024 09:15:58 [INFO|DP=0|PP=0|TP=2|ip-26-0-170-143]: [After model building] Memory usage: 251.15MiB. Peak allocated: 5440.00MiB Peak reserved: 22866.00MiB
+12/10/2024 09:15:58 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]: No checkpoint path provided.
+12/10/2024 09:15:58 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]: Parametrizing model parameters using StandardParametrizator
+12/10/2024 09:15:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]: [Optimizer Building] Using LearningRateForSP as learning rate
+12/10/2024 09:15:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]: [ZeRO sharding] Size of optimizer params per rank:
+12/10/2024 09:15:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]: [ZeRO sharding] DP Rank 0 has 32.5M out of 130M (25.00%) params' optimizer states
+12/10/2024 09:15:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]: [ZeRO sharding] DP Rank 1 has 32.5M out of 130M (25.00%) params' optimizer states
+12/10/2024 09:15:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]: [ZeRO sharding] DP Rank 2 has 32.5M out of 130M (25.00%) params' optimizer states
+12/10/2024 09:15:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]: [ZeRO sharding] DP Rank 3 has 32.5M out of 130M (25.00%) params' optimizer states
+12/10/2024 09:16:00 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]: [Training Plan] Stage Stable Training Stage has 99 remaining training steps and has consumed 0 samples
+12/10/2024 09:16:00 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]: Using dummy data generator
+12/10/2024 09:16:00 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]: [Training Plan] There are 1 training stages 
+12/10/2024 09:16:00 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]: [Stage Stable Training Stage] start from step 1 
+12/10/2024 09:16:00 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]: 
+12/10/2024 09:16:00 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]: [Start training] datetime: 2024-12-10 09:16:00.529764 | mbs: 4 | grad_accum: 32 | global_batch_size: 512 | sequence_length: 2048 | train_steps: 100 | start_iteration_step: 0 | consumed_train_samples: 0
+wandb: WARNING Path /fsx/nouamane/.cache/wandb/wandb/ wasn't writable, using system temp directory.
+wandb: WARNING Path /fsx/nouamane/.cache/wandb/wandb/ wasn't writable, using system temp directory
+wandb: Tracking run with wandb version 0.16.0
+wandb: W&B syncing is set to `offline` in this directory.  
+wandb: Run `wandb online` or set WANDB_MODE=online to enable cloud syncing.
+12/10/2024 09:16:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]: Resuming training from stage Stable Training Stage, it has trained for 0 samples and has 99 remaining train steps
+12/10/2024 09:16:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]:  Memory usage: 871.48MiB. Peak allocated 5440.00MiB. Peak reserved: 21842.00MiB
+12/10/2024 09:16:13 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]:  Memory usage: 1003.63MiB. Peak allocated 2711.20MiB. Peak reserved: 4492.00MiB
+12/10/2024 09:16:13 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]: iteration: 1 / 100 | consumed_tokens: 1.05M | elapsed_time_per_iteration_ms: 4.62K | tokens_per_sec: 227K | tokens_per_sec_per_gpu: 7.09K | global_batch_size: 512 | lm_loss: 10.8 | lr: 0.00015 | model_tflops_per_gpu: 50 | hardware_tflops_per_gpu: 50 | grad_norm: 0.504 | cuda_memory_allocated: 1.31G | cuda_max_memory_reserved: 4.74G | hd_total_memory_tb: 312G | hd_used_memory_tb: 70.1G | hd_free_memory_tb: 242G
+12/10/2024 09:16:13 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]:  Memory usage: 1251.81MiB. Peak allocated 1810.10MiB. Peak reserved: 4516.00MiB
+12/10/2024 09:16:15 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]:  Memory usage: 1251.83MiB. Peak allocated 2959.40MiB. Peak reserved: 4662.00MiB
+12/10/2024 09:16:16 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]: iteration: 2 / 100 | consumed_tokens: 2.1M | elapsed_time_per_iteration_ms: 2.6K | tokens_per_sec: 403K | tokens_per_sec_per_gpu: 12.6K | global_batch_size: 512 | lm_loss: 10.8 | lr: 0.0003 | model_tflops_per_gpu: 88.8 | hardware_tflops_per_gpu: 88.8 | grad_norm: 0.504 | cuda_memory_allocated: 1.31G | cuda_max_memory_reserved: 4.89G | hd_total_memory_tb: 312G | hd_used_memory_tb: 70.1G | hd_free_memory_tb: 242G
+12/10/2024 09:16:16 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]:  Memory usage: 1251.81MiB. Peak allocated 1810.12MiB. Peak reserved: 4662.00MiB
+12/10/2024 09:16:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]:  Memory usage: 1251.83MiB. Peak allocated 2959.40MiB. Peak reserved: 4662.00MiB
+num_paramsnum_params
+
+{'total': 1040728064, 'local': 130091008}{'total': 1040728064, 'local': 130091008}num_params
+
+
+{'total': 1040728064, 'local': 130091008}
+num_paramsnum_paramsnum_params
+
+
+{'total': 1040728064, 'local': 130091008}{'total': 1040728064, 'local': 130091008}
+
+num_params{'total': 1040728064, 'local': 130091008}
+
+num_paramsnum_params
+{'total': 1040728064, 'local': 130091008}
+
+{'total': 1040728064, 'local': 130091008}
+num_params
+num_params{'total': 1040728064, 'local': 130091008}num_params
+
+
+{'total': 1040728064, 'local': 130091008}
+num_params{'total': 1040728064, 'local': 130091008}
+
+num_paramsnum_params
+num_params{'total': 1040728064, 'local': 130091008}
+num_params
+num_params
+{'total': 1040728064, 'local': 130091008}
+
+
+{'total': 1040728064, 'local': 130091008}
+{'total': 1040728064, 'local': 130091008}
+{'total': 1040728064, 'local': 130091008}
+num_params
+{'total': 1040728064, 'local': 130091008}
+num_params
+{'total': 1040728064, 'local': 130091008}
+num_params{'total': 1040728064, 'local': 130091008}
+
+{'total': 1040728064, 'local': 130091008}
+num_params
+{'total': 1040728064, 'local': 130091008}
+num_params{'total': 1040728064, 'local': 130091008}
+
+{'total': 1040728064, 'local': 130091008}
+num_params
+{'total': 1040728064, 'local': 130091008}
+num_params
+{'total': 1040728064, 'local': 130091008}
+num_params
+{'total': 1040728064, 'local': 130091008}
+num_params
+{'total': 1040728064, 'local': 130091008}
+num_paramsnum_params
+
+{'total': 1040728064, 'local': 130091008}
+{'total': 1040728064, 'local': 130091008}
+num_paramsnum_params
+
+{'total': 1040728064, 'local': 130091008}{'total': 1040728064, 'local': 130091008}
+
+12/10/2024 09:16:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]: iteration: 3 / 100 | consumed_tokens: 3.15M | elapsed_time_per_iteration_ms: 2.6K | tokens_per_sec: 403K | tokens_per_sec_per_gpu: 12.6K | global_batch_size: 512 | lm_loss: 10.8 | lr: 0.000296 | model_tflops_per_gpu: 88.8 | hardware_tflops_per_gpu: 88.8 | grad_norm: 0.489 | cuda_memory_allocated: 1.31G | cuda_max_memory_reserved: 4.89G | hd_total_memory_tb: 312G | hd_used_memory_tb: 70.1G | hd_free_memory_tb: 242G
+num_params
+{'total': 1040728064, 'local': 130091008}
+12/10/2024 09:16:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]: | job_id   | name                                                          | nodes | seq_len | mbs | batch_accum | gbs | mTFLOPs | hTFLOPs | tok/s/gpu | AllReduce (GB/s) | AllGather (GB/s) | ReduceScatter (GB/s) | AR Intra-node (GB/s) | AG Intra-node (GB/s) | RS Intra-node (GB/s) | Mem Alloc (GB) | Mem Res (GB) | dp | pp | tp | pp_engine                                                                                                 | tp_mode                                 | tp_async_comm | hidden_size | hidden_act | num_layers | num_heads | num_kv_heads | max_pos | vocab_size | tie_word_embeddings | dtype          | zero_stage | ddp_bucket_cap_mb | accumulate_grad_in_fp32 | Total Params | Local Params |
+12/10/2024 09:16:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]: | -------- | ------------------------------------------------------------- | ----- | ------- | --- | ----------- | --- | ------- | ------- | --------- | ---------------- | ---------------- | -------------------- | -------------------- | -------------------- | -------------------- | -------------- | ------------ | -- | -- | -- | --------------------------------------------------------------------------------------------------------- | --------------------------------------- | ------------- | ----------- | ---------- | ---------- | --------- | ------------ | ------- | ---------- | ------------------- | -------------- | ---------- | ----------------- | ----------------------- | ------------ | ------------ |
+12/10/2024 09:16:18 [INFO|DP=2|PP=0|TP=2|ip-26-0-171-102]: Throughput logging complete
+12/10/2024 09:16:18 [INFO|DP=1|PP=0|TP=0|ip-26-0-170-160]: Throughput logging complete
+12/10/2024 09:16:18 [INFO|DP=1|PP=0|TP=1|ip-26-0-170-160]: Throughput logging complete
+12/10/2024 09:16:18 [INFO|DP=1|PP=0|TP=6|ip-26-0-170-160]: Throughput logging complete
+12/10/2024 09:16:18 [INFO|DP=1|PP=0|TP=4|ip-26-0-170-160]: Throughput logging complete
+12/10/2024 09:16:18 [INFO|DP=1|PP=0|TP=2|ip-26-0-170-160]: Throughput logging complete
+12/10/2024 09:16:18 [INFO|DP=3|PP=0|TP=3|ip-26-0-171-88]: Throughput logging complete
+12/10/2024 09:16:18 [INFO|DP=2|PP=0|TP=1|ip-26-0-171-102]: Throughput logging complete
+12/10/2024 09:16:18 [INFO|DP=2|PP=0|TP=4|ip-26-0-171-102]: Throughput logging complete
+12/10/2024 09:16:18 [INFO|DP=2|PP=0|TP=3|ip-26-0-171-102]: Throughput logging complete
+12/10/2024 09:16:18 [INFO|DP=2|PP=0|TP=7|ip-26-0-171-102]: Throughput logging complete
+12/10/2024 09:16:18 [INFO|DP=2|PP=0|TP=5|ip-26-0-171-102]: Throughput logging complete
+12/10/2024 09:16:18 [INFO|DP=2|PP=0|TP=6|ip-26-0-171-102]: Throughput logging complete
+12/10/2024 09:16:18 [INFO|DP=3|PP=0|TP=1|ip-26-0-171-88]: Throughput logging complete
+12/10/2024 09:16:18 [INFO|DP=3|PP=0|TP=0|ip-26-0-171-88]: Throughput logging complete
+12/10/2024 09:16:18 [INFO|DP=3|PP=0|TP=4|ip-26-0-171-88]: Throughput logging complete
+12/10/2024 09:16:18 [INFO|DP=1|PP=0|TP=5|ip-26-0-170-160]: Throughput logging complete
+12/10/2024 09:16:18 [INFO|DP=3|PP=0|TP=6|ip-26-0-171-88]: Throughput logging complete
+12/10/2024 09:16:18 [INFO|DP=3|PP=0|TP=5|ip-26-0-171-88]: Throughput logging complete
+12/10/2024 09:16:18 [INFO|DP=3|PP=0|TP=2|ip-26-0-171-88]: Throughput logging complete
+12/10/2024 09:16:18 [INFO|DP=3|PP=0|TP=7|ip-26-0-171-88]: Throughput logging complete
+12/10/2024 09:16:18 [INFO|DP=1|PP=0|TP=7|ip-26-0-170-160]: Throughput logging complete
+12/10/2024 09:16:18 [INFO|DP=1|PP=0|TP=3|ip-26-0-170-160]: Throughput logging complete
+12/10/2024 09:16:18 [INFO|DP=2|PP=0|TP=0|ip-26-0-171-102]: Throughput logging complete
+12/10/2024 09:16:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]: | 13266667 | 1.14G_dp4_tp8_pp1_acc32_mbs4_seq2048_zero1_tpmodeRED_vocab32k | 4     | 2048    | 4   | 32          | 512 | 88.80   | 88.80   | 12601.94  | 235.70           | 214.91           | 208.85               | 460.21               | 265.16               | 264.93               | 1.77           | 4.55         | 4  | 1  | 8  | <nanotron.parallel.pipeline_parallel.engine.OneForwardOneBackwardPipelineEngine object at 0x7fe9fb5b8d00> | TensorParallelLinearMode.REDUCE_SCATTER | True          | 2048        | silu       | 16         | 32        | 8            | 2048    | 32768      | True                | torch.bfloat16 | 1          | 25                | True                    | 1.04G        | 130M         |
+12/10/2024 09:16:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]: Throughput logging complete
+12/10/2024 09:16:18 [INFO|DP=0|PP=0|TP=6|ip-26-0-170-143]: Throughput logging complete
+12/10/2024 09:16:18 [INFO|DP=0|PP=0|TP=5|ip-26-0-170-143]: Throughput logging complete
+12/10/2024 09:16:18 [INFO|DP=0|PP=0|TP=4|ip-26-0-170-143]: Throughput logging complete
+12/10/2024 09:16:18 [INFO|DP=0|PP=0|TP=3|ip-26-0-170-143]: Throughput logging complete
+12/10/2024 09:16:18 [INFO|DP=0|PP=0|TP=1|ip-26-0-170-143]: Throughput logging complete
+12/10/2024 09:16:18 [INFO|DP=0|PP=0|TP=7|ip-26-0-170-143]: Throughput logging complete
+12/10/2024 09:16:18 [INFO|DP=0|PP=0|TP=2|ip-26-0-170-143]: Throughput logging complete
+srun: Job step aborted: Waiting up to 32 seconds for job step to finish.
+[2024-12-10 09:16:19,483] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-10 09:16:19,483] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-10 09:16:19,484] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 96003 closing signal SIGTERM
+[2024-12-10 09:16:19,483] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1691392 closing signal SIGTERM
+[2024-12-10 09:16:19,483] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-10 09:16:19,483] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1410119 closing signal SIGTERM
+[2024-12-10 09:16:19,484] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 96004 closing signal SIGTERM
+[2024-12-10 09:16:19,484] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1691393 closing signal SIGTERM
+[2024-12-10 09:16:19,484] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1410120 closing signal SIGTERM
+[2024-12-10 09:16:19,484] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 96005 closing signal SIGTERM
+[2024-12-10 09:16:19,484] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1410121 closing signal SIGTERM
+[2024-12-10 09:16:19,484] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1691394 closing signal SIGTERM
+[2024-12-10 09:16:19,484] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1410122 closing signal SIGTERM
+[2024-12-10 09:16:19,484] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 96006 closing signal SIGTERM
+[2024-12-10 09:16:19,484] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 96007 closing signal SIGTERM
+[2024-12-10 09:16:19,484] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1691395 closing signal SIGTERM
+slurmstepd: error: *** STEP 13266667.0 ON ip-26-0-170-143 CANCELLED AT 2024-12-10T09:16:19 ***
+slurmstepd: error: *** JOB 13266667 ON ip-26-0-170-143 CANCELLED AT 2024-12-10T09:16:19 ***
+12/10/2024 09:16:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-170-143]:  Memory usage: 1251.81MiB. Peak allocated 1810.12MiB. Peak reserved: 4662.00MiB
+[2024-12-10 09:16:19,485] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-10 09:16:19,485] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 34792 closing signal SIGTERM
+[2024-12-10 09:16:19,485] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 34793 closing signal SIGTERM
+[2024-12-10 09:16:19,485] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 34794 closing signal SIGTERM
+[2024-12-10 09:16:19,485] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 34795 closing signal SIGTERM
+[2024-12-10 09:16:19,485] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 34796 closing signal SIGTERM
+[2024-12-10 09:16:19,485] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 34797 closing signal SIGTERM
+[2024-12-10 09:16:19,485] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1691396 closing signal SIGTERM
+[2024-12-10 09:16:19,486] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1410123 closing signal SIGTERM
+[2024-12-10 09:16:19,486] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 96008 closing signal SIGTERM
+[2024-12-10 09:16:19,486] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 34798 closing signal SIGTERM
+[2024-12-10 09:16:19,486] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1691397 closing signal SIGTERM
+[2024-12-10 09:16:19,486] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1410124 closing signal SIGTERM
+[2024-12-10 09:16:19,486] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 96009 closing signal SIGTERM
+[2024-12-10 09:16:19,487] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1691398 closing signal SIGTERM
+[2024-12-10 09:16:19,486] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 96010 closing signal SIGTERM
+[2024-12-10 09:16:19,488] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1691399 closing signal SIGTERM
+[2024-12-10 09:16:19,487] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1410125 closing signal SIGTERM
+[2024-12-10 09:16:19,487] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1410126 closing signal SIGTERM
+[2024-12-10 09:16:19,498] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 34799 closing signal SIGTERM
diff --git a/logs/13392392-bench_stress_test.out b/logs/13392392-bench_stress_test.out
new file mode 100644
index 0000000000000000000000000000000000000000..cae17b72f7f95cdb99099c06af84027a67bd4cc9
--- /dev/null
+++ b/logs/13392392-bench_stress_test.out
@@ -0,0 +1,3737 @@
++ source /etc/profile.d/modules.sh
+++ . /usr/share/modules/init/bash
++++ unset _mlshdbg
++++ '[' 0 = 1 ']'
++++ unset _mlre _mlIFS
++++ '[' -n x ']'
++++ _mlIFS=' 	
+'
++++ IFS=' '
++++ '[' -n '' ']'
+++++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash autoinit
++++ _mlcode='module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+MODULEPATH_modshare=/etc/environment-modules/modules:1:/usr/share/modules/\$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1; export MODULEPATH_modshare;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+LOADEDMODULES=; export LOADEDMODULES;
+MODULEPATH=/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/\$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles; export MODULEPATH;
+test 0;'
++++ _mlret=0
++++ '[' -n x ']'
++++ IFS=' 	
+'
++++ unset _mlIFS
++++ unset _mlre _mlv _mlrv
++++ '[' 0 -eq 0 ']'
++++ eval 'module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+MODULEPATH_modshare=/etc/environment-modules/modules:1:/usr/share/modules/\$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1; export MODULEPATH_modshare;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+LOADEDMODULES=; export LOADEDMODULES;
+MODULEPATH=/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/\$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles; export MODULEPATH;
+test 0;'
+++++ MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl
+++++ export MODULES_CMD
+++++ ENV=/usr/share/modules/init/profile.sh
+++++ export ENV
+++++ MODULEPATH_modshare='/etc/environment-modules/modules:1:/usr/share/modules/$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1'
+++++ export MODULEPATH_modshare
+++++ BASH_ENV=/usr/share/modules/init/bash
+++++ export BASH_ENV
+++++ MODULESHOME=/usr/share/modules
+++++ export MODULESHOME
+++++ LOADEDMODULES=
+++++ export LOADEDMODULES
+++++ MODULEPATH='/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles'
+++++ export MODULEPATH
+++++ test 0
++++ '[' 0 = 1 ']'
++++ '[' -t 2 ']'
++++ export -f module
++++ export -f switchml
++++ '[' 5 -ge 3 ']'
++++ [[ ehxB =~ i ]]
++++ [[ ! :/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin: =~ :/usr/bin: ]]
++++ '[' '!' -n '' ']'
++++ MANPATH=:
++++ export MANPATH
+++++ manpath
++++ [[ ! :/admin/home/nouamane/.local/share/man:/fsx/nouamane/miniconda/envs/2-1-cu121/man:/fsx/nouamane/miniconda/envs/2-1-cu121/share/man:/fsx/nouamane/miniconda/man:/fsx/nouamane/miniconda/share/man:/opt/amazon/openmpi/share/man:/opt/amazon/efa/share/man:/opt/slurm/share/man:/usr/local/man:/usr/local/share/man:/usr/share/man:/admin/home/nouamane/.fzf/man:: =~ :/usr/share/man: ]]
++++ unset _mlcode _mlret
++++ '[' -n '' ']'
++ module load cuda/12.1
++ unset _mlshdbg
++ '[' 0 = 1 ']'
++ unset _mlre _mlIFS
++ '[' -n x ']'
++ _mlIFS=' 	
+'
++ IFS=' '
++ '[' -n '' ']'
+++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash load cuda/12.1
++ eval 'CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include;' export 'CPATH;
+LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:;' export 'LD_LIBRARY_PATH;
+FI_EFA_FORK_SAFE=1;' export 'FI_EFA_FORK_SAFE;
+MANPATH=/usr/local/cuda-12.1/share/man::;' export 'MANPATH;
+LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64;' export 'LIBRARY_PATH;
+_LMFILES_=/usr/share/modules/modulefiles/cuda/12.1;' export '_LMFILES_;
+LOADEDMODULES=cuda/12.1;' export 'LOADEDMODULES;
+MPI_PATH=/opt/amazon/openmpi;' export 'MPI_PATH;
+NCCL_HOME_modshare=/opt/nccl/build:1;' export 'NCCL_HOME_modshare;
+NCCL_PROTO=simple;' export 'NCCL_PROTO;
+MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1;' export 'MANPATH_modshare;
+LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1;' export 'LIBRARY_PATH_modshare;
+NCCL_SOCKET_IFNAME=enp;' export 'NCCL_SOCKET_IFNAME;
+AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl;' export 'AWS_OFI_NCCL_HOME;
+NCCL_HOME=/opt/nccl/build;' export 'NCCL_HOME;
+FI_PROVIDER=efa;' export 'FI_PROVIDER;
+AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1;' export 'AWS_OFI_NCCL_HOME_modshare;
+CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1;' export 'CPATH_modshare;
+LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2::1:/opt/nccl/build/lib:1:/opt/aws-ofi-nccl/lib:2:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/efa/lib:1;' export 'LD_LIBRARY_PATH_modshare;
+FI_EFA_ENABLE_SHM_TRANSFER=1;' export 'FI_EFA_ENABLE_SHM_TRANSFER;
+_LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1;' export '_LMFILES__modshare;
+LOADEDMODULES_modshare=cuda/12.1:1;' export 'LOADEDMODULES_modshare;
+MPI_PATH_modshare=/opt/amazon/openmpi:1;' export 'MPI_PATH_modshare;
+PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin;' export 'PATH;
+CUDA_HOME=/usr/local/cuda-12.1;' export 'CUDA_HOME;
+PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/usr/local/cuda-12.1/include:1:/opt/amazon/efa/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/usr/sbin:1:/fsx/nouamane/miniconda/bin:1:/opt/slurm/bin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:2:/opt/amazon/openmpi/bin:1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1;' export 'PATH_modshare;
+test' '0;'
+++ CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include
+++ export CPATH
+++ LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:
+++ export LD_LIBRARY_PATH
+++ FI_EFA_FORK_SAFE=1
+++ export FI_EFA_FORK_SAFE
+++ MANPATH=/usr/local/cuda-12.1/share/man::
+++ export MANPATH
+++ LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64
+++ export LIBRARY_PATH
+++ _LMFILES_=/usr/share/modules/modulefiles/cuda/12.1
+++ export _LMFILES_
+++ LOADEDMODULES=cuda/12.1
+++ export LOADEDMODULES
+++ MPI_PATH=/opt/amazon/openmpi
+++ export MPI_PATH
+++ NCCL_HOME_modshare=/opt/nccl/build:1
+++ export NCCL_HOME_modshare
+++ NCCL_PROTO=simple
+++ export NCCL_PROTO
+++ MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1
+++ export MANPATH_modshare
+++ LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1
+++ export LIBRARY_PATH_modshare
+++ NCCL_SOCKET_IFNAME=enp
+++ export NCCL_SOCKET_IFNAME
+++ AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl
+++ export AWS_OFI_NCCL_HOME
+++ NCCL_HOME=/opt/nccl/build
+++ export NCCL_HOME
+++ FI_PROVIDER=efa
+++ export FI_PROVIDER
+++ AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1
+++ export AWS_OFI_NCCL_HOME_modshare
+++ CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1
+++ export CPATH_modshare
+++ LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2::1:/opt/nccl/build/lib:1:/opt/aws-ofi-nccl/lib:2:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/efa/lib:1
+++ export LD_LIBRARY_PATH_modshare
+++ FI_EFA_ENABLE_SHM_TRANSFER=1
+++ export FI_EFA_ENABLE_SHM_TRANSFER
+++ _LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1
+++ export _LMFILES__modshare
+++ LOADEDMODULES_modshare=cuda/12.1:1
+++ export LOADEDMODULES_modshare
+++ MPI_PATH_modshare=/opt/amazon/openmpi:1
+++ export MPI_PATH_modshare
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ export PATH
+++ CUDA_HOME=/usr/local/cuda-12.1
+++ export CUDA_HOME
+++ PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/usr/local/cuda-12.1/include:1:/opt/amazon/efa/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/usr/sbin:1:/fsx/nouamane/miniconda/bin:1:/opt/slurm/bin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:2:/opt/amazon/openmpi/bin:1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1
+++ export PATH_modshare
+++ test 0
++ _mlstatus=0
++ '[' -n x ']'
++ IFS=' 	
+'
++ unset _mlre _mlv _mlrv _mlIFS
++ '[' -n '' ']'
++ unset _mlshdbg
++ return 0
++ source /fsx/nouamane/miniconda/bin/activate
+++ _CONDA_ROOT=/fsx/nouamane/miniconda
+++ . /fsx/nouamane/miniconda/etc/profile.d/conda.sh
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ '[' -z x ']'
+++ conda activate
+++ local cmd=activate
+++ case "$cmd" in
+++ __conda_activate activate
+++ '[' -n '' ']'
+++ local ask_conda
++++ PS1=
++++ __conda_exe shell.posix activate
++++ /fsx/nouamane/miniconda/bin/conda shell.posix activate
+++ ask_conda='. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_2='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\'''
+++ eval '. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_2='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\'''
++++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh
+++++ test -n ''
+++++ unset XML_CATALOG_FILES
+++++ unset xml_catalog_files_libxml2
++++ PS1='(base) '
++++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++++ export CONDA_PREFIX=/fsx/nouamane/miniconda
++++ CONDA_PREFIX=/fsx/nouamane/miniconda
++++ export CONDA_SHLVL=3
++++ CONDA_SHLVL=3
++++ export CONDA_DEFAULT_ENV=base
++++ CONDA_DEFAULT_ENV=base
++++ export 'CONDA_PROMPT_MODIFIER=(base) '
++++ CONDA_PROMPT_MODIFIER='(base) '
++++ export CONDA_PREFIX_2=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ CONDA_PREFIX_2=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ __conda_hashr
+++ '[' -n '' ']'
+++ '[' -n '' ']'
+++ hash -r
++ conda activate 2-1-cu121
++ local cmd=activate
++ case "$cmd" in
++ __conda_activate activate 2-1-cu121
++ '[' -n '' ']'
++ local ask_conda
+++ PS1='(base) '
+++ __conda_exe shell.posix activate 2-1-cu121
+++ /fsx/nouamane/miniconda/bin/conda shell.posix activate 2-1-cu121
++ ask_conda='PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''4'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_3='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
++ eval 'PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''4'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_3='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
+++ PS1='(2-1-cu121) '
+++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ export CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ export CONDA_SHLVL=4
+++ CONDA_SHLVL=4
+++ export CONDA_DEFAULT_ENV=2-1-cu121
+++ CONDA_DEFAULT_ENV=2-1-cu121
+++ export 'CONDA_PROMPT_MODIFIER=(2-1-cu121) '
+++ CONDA_PROMPT_MODIFIER='(2-1-cu121) '
+++ export CONDA_PREFIX_3=/fsx/nouamane/miniconda
+++ CONDA_PREFIX_3=/fsx/nouamane/miniconda
+++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ export _CE_M=
+++ _CE_M=
+++ export _CE_CONDA=
+++ _CE_CONDA=
+++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh
++++ test -n ''
++++ xml_catalog_files_libxml2=
++++ XML_CATALOG_FILES=
++++ conda_catalog_files=
++++ ifs_libxml2=' 	
+'
++++ IFS=' '
++++ rem=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ for pre in ${rem}
++++ test '' = /fsx/nouamane/miniconda/envs/2-1-cu121
++++ conda_catalog_files=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ rem=
++++ IFS=' 	
+'
++++ conda_catalog_files='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ export 'XML_CATALOG_FILES=file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ XML_CATALOG_FILES='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ unset conda_catalog_files ifs_libxml2 rem
++ __conda_hashr
++ '[' -n '' ']'
++ '[' -n '' ']'
++ hash -r
++ export PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++ PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ scontrol show hostnames ip-26-0-169-207
++ export NODELIST=ip-26-0-169-207
++ NODELIST=ip-26-0-169-207
+++ scontrol show hostnames ip-26-0-169-207
+++ head -n1
++ export MASTER_NODE=ip-26-0-169-207
++ MASTER_NODE=ip-26-0-169-207
++ export MASTER_PORT=12356
++ MASTER_PORT=12356
++ export NNODES=1
++ NNODES=1
++ export GPUS_PER_NODE=8
++ GPUS_PER_NODE=8
++ export WORLD_SIZE=8
++ WORLD_SIZE=8
++ export CUDA_DEVICE_MAX_CONNECTIONS=1
++ CUDA_DEVICE_MAX_CONNECTIONS=1
++ export NCCL_DEBUG=INFO
++ NCCL_DEBUG=INFO
++ export NANOTRON_BENCHMARK=1
++ NANOTRON_BENCHMARK=1
++ echo 'Master node: ip-26-0-169-207'
+Master node: ip-26-0-169-207
++ echo 'All nodes: ip-26-0-169-207'
+All nodes: ip-26-0-169-207
++ echo 'World size: 8'
+World size: 8
++ srun torchrun --nnodes=1 --nproc_per_node=8 --rdzv_id=13392392 --rdzv_backend=c10d --rdzv_endpoint=ip-26-0-169-207:12356 run_train.py --config-file benchmark/configs/config_1.14G_dp4_tp2_pp1_acc256_mbs2_seq2048_zero1_tpmodeRED_vocab32k.yaml
+[2024-12-12 21:38:11,076] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-12 21:38:11,076] torch.distributed.run: [WARNING] 
+[2024-12-12 21:38:11,076] torch.distributed.run: [WARNING] *****************************************
+[2024-12-12 21:38:11,076] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-12 21:38:11,076] torch.distributed.run: [WARNING] *****************************************
+ip-26-0-169-207:162589:162589 [0] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-169-207:162589:162589 [0] NCCL INFO Bootstrap : Using enp71s0:26.0.169.207<0>
+ip-26-0-169-207:162589:162589 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-169-207:162589:162589 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-169-207:162589:162589 [0] NCCL INFO cudaDriverVersion 12020
+NCCL version 2.18.5+cuda12.2
+ip-26-0-169-207:162595:162595 [6] NCCL INFO cudaDriverVersion 12020
+ip-26-0-169-207:162591:162591 [2] NCCL INFO cudaDriverVersion 12020
+ip-26-0-169-207:162595:162595 [6] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-169-207:162591:162591 [2] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-169-207:162593:162593 [4] NCCL INFO cudaDriverVersion 12020
+ip-26-0-169-207:162593:162593 [4] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-169-207:162590:162590 [1] NCCL INFO cudaDriverVersion 12020
+ip-26-0-169-207:162592:162592 [3] NCCL INFO cudaDriverVersion 12020
+ip-26-0-169-207:162590:162590 [1] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-169-207:162592:162592 [3] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-169-207:162596:162596 [7] NCCL INFO cudaDriverVersion 12020
+ip-26-0-169-207:162596:162596 [7] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-169-207:162595:162595 [6] NCCL INFO Bootstrap : Using enp71s0:26.0.169.207<0>
+ip-26-0-169-207:162591:162591 [2] NCCL INFO Bootstrap : Using enp71s0:26.0.169.207<0>
+ip-26-0-169-207:162594:162594 [5] NCCL INFO cudaDriverVersion 12020
+ip-26-0-169-207:162595:162595 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-169-207:162595:162595 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-169-207:162594:162594 [5] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-169-207:162591:162591 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-169-207:162591:162591 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-169-207:162593:162593 [4] NCCL INFO Bootstrap : Using enp71s0:26.0.169.207<0>
+ip-26-0-169-207:162590:162590 [1] NCCL INFO Bootstrap : Using enp71s0:26.0.169.207<0>
+ip-26-0-169-207:162592:162592 [3] NCCL INFO Bootstrap : Using enp71s0:26.0.169.207<0>
+ip-26-0-169-207:162596:162596 [7] NCCL INFO Bootstrap : Using enp71s0:26.0.169.207<0>
+ip-26-0-169-207:162594:162594 [5] NCCL INFO Bootstrap : Using enp71s0:26.0.169.207<0>
+ip-26-0-169-207:162593:162593 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-169-207:162593:162593 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-169-207:162592:162592 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-169-207:162590:162590 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-169-207:162596:162596 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-169-207:162592:162592 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-169-207:162590:162590 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-169-207:162596:162596 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-169-207:162594:162594 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-169-207:162594:162594 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-169-207:162589:162711 [0] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-169-207:162589:162711 [0] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-169-207:162589:162711 [0] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-169-207:162589:162711 [0] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-169-207:162589:162711 [0] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-169-207:162589:162711 [0] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-169-207:162589:162711 [0] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-169-207:162589:162711 [0] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-169-207:162589:162711 [0] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-169-207:162594:162713 [5] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-169-207:162594:162713 [5] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-169-207:162594:162713 [5] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-169-207:162594:162713 [5] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-169-207:162594:162713 [5] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-169-207:162594:162713 [5] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-169-207:162594:162713 [5] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-169-207:162594:162713 [5] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-169-207:162594:162713 [5] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-169-207:162591:162715 [2] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-169-207:162591:162715 [2] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-169-207:162591:162715 [2] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-169-207:162591:162715 [2] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-169-207:162591:162715 [2] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-169-207:162591:162715 [2] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-169-207:162591:162715 [2] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-169-207:162591:162715 [2] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-169-207:162591:162715 [2] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-169-207:162595:162712 [6] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-169-207:162595:162712 [6] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-169-207:162592:162717 [3] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-169-207:162592:162717 [3] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-169-207:162596:162716 [7] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-169-207:162596:162716 [7] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-169-207:162593:162714 [4] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-169-207:162593:162714 [4] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-169-207:162590:162718 [1] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-169-207:162590:162718 [1] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-169-207:162595:162712 [6] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-169-207:162595:162712 [6] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-169-207:162595:162712 [6] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-169-207:162595:162712 [6] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-169-207:162595:162712 [6] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-169-207:162595:162712 [6] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-169-207:162595:162712 [6] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-169-207:162592:162717 [3] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-169-207:162592:162717 [3] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-169-207:162592:162717 [3] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-169-207:162592:162717 [3] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-169-207:162592:162717 [3] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-169-207:162592:162717 [3] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-169-207:162592:162717 [3] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-169-207:162596:162716 [7] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-169-207:162596:162716 [7] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-169-207:162596:162716 [7] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-169-207:162596:162716 [7] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-169-207:162596:162716 [7] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-169-207:162596:162716 [7] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-169-207:162596:162716 [7] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-169-207:162593:162714 [4] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-169-207:162593:162714 [4] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-169-207:162593:162714 [4] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-169-207:162593:162714 [4] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-169-207:162593:162714 [4] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-169-207:162593:162714 [4] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-169-207:162593:162714 [4] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-169-207:162590:162718 [1] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-169-207:162590:162718 [1] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-169-207:162590:162718 [1] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-169-207:162590:162718 [1] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-169-207:162590:162718 [1] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-169-207:162590:162718 [1] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-169-207:162590:162718 [1] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-169-207:162589:162711 [0] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-169-207:162594:162713 [5] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-169-207:162595:162712 [6] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-169-207:162591:162715 [2] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-169-207:162593:162714 [4] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-169-207:162596:162716 [7] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-169-207:162592:162717 [3] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-169-207:162590:162718 [1] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-169-207:162589:162711 [0] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-169-207:162589:162711 [0] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-169-207:162589:162711 [0] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-169-207:162589:162711 [0] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-169-207:162589:162711 [0] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-169-207:162589:162711 [0] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-169-207:162589:162711 [0] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-169-207:162589:162711 [0] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-169-207:162589:162711 [0] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-169-207:162589:162711 [0] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-169-207:162589:162711 [0] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-169-207:162589:162711 [0] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-169-207:162589:162711 [0] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-169-207:162589:162711 [0] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-169-207:162589:162711 [0] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-169-207:162589:162711 [0] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-169-207:162589:162711 [0] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-169-207:162589:162711 [0] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-169-207:162589:162711 [0] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-169-207:162589:162711 [0] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-169-207:162589:162711 [0] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-169-207:162589:162711 [0] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-169-207:162589:162711 [0] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-169-207:162589:162711 [0] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-169-207:162589:162711 [0] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-169-207:162589:162711 [0] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-169-207:162589:162711 [0] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-169-207:162589:162711 [0] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-169-207:162589:162711 [0] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-169-207:162589:162711 [0] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-169-207:162589:162711 [0] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-169-207:162589:162711 [0] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-169-207:162589:162711 [0] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/162
+ip-26-0-169-207:162589:162711 [0] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-169-207:162594:162713 [5] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-169-207:162594:162713 [5] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-169-207:162594:162713 [5] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-169-207:162594:162713 [5] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-169-207:162594:162713 [5] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-169-207:162594:162713 [5] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-169-207:162594:162713 [5] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-169-207:162594:162713 [5] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-169-207:162594:162713 [5] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-169-207:162594:162713 [5] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-169-207:162594:162713 [5] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-169-207:162594:162713 [5] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-169-207:162594:162713 [5] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-169-207:162594:162713 [5] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-169-207:162594:162713 [5] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-169-207:162594:162713 [5] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-169-207:162594:162713 [5] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-169-207:162594:162713 [5] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-169-207:162594:162713 [5] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-169-207:162594:162713 [5] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-169-207:162594:162713 [5] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-169-207:162594:162713 [5] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-169-207:162594:162713 [5] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-169-207:162594:162713 [5] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-169-207:162594:162713 [5] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-169-207:162594:162713 [5] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-169-207:162594:162713 [5] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-169-207:162594:162713 [5] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-169-207:162594:162713 [5] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-169-207:162594:162713 [5] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-169-207:162594:162713 [5] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-169-207:162594:162713 [5] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-169-207:162594:162713 [5] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/149
+ip-26-0-169-207:162594:162713 [5] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-169-207:162589:162711 [0] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-169-207:162589:162711 [0] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-169-207:162589:162711 [0] NCCL INFO Using network Libfabric
+ip-26-0-169-207:162596:162716 [7] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-169-207:162596:162716 [7] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-169-207:162596:162716 [7] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-169-207:162596:162716 [7] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-169-207:162596:162716 [7] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-169-207:162596:162716 [7] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-169-207:162596:162716 [7] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-169-207:162596:162716 [7] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-169-207:162596:162716 [7] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-169-207:162596:162716 [7] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-169-207:162596:162716 [7] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-169-207:162596:162716 [7] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-169-207:162596:162716 [7] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-169-207:162596:162716 [7] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-169-207:162596:162716 [7] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-169-207:162596:162716 [7] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-169-207:162596:162716 [7] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-169-207:162596:162716 [7] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-169-207:162596:162716 [7] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-169-207:162596:162716 [7] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-169-207:162596:162716 [7] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-169-207:162596:162716 [7] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-169-207:162596:162716 [7] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-169-207:162596:162716 [7] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-169-207:162596:162716 [7] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-169-207:162596:162716 [7] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-169-207:162596:162716 [7] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-169-207:162596:162716 [7] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-169-207:162596:162716 [7] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-169-207:162596:162716 [7] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-169-207:162596:162716 [7] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-169-207:162596:162716 [7] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-169-207:162596:162716 [7] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/149
+ip-26-0-169-207:162596:162716 [7] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-169-207:162592:162717 [3] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-169-207:162592:162717 [3] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-169-207:162592:162717 [3] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-169-207:162592:162717 [3] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-169-207:162592:162717 [3] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-169-207:162592:162717 [3] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-169-207:162592:162717 [3] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-169-207:162592:162717 [3] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-169-207:162592:162717 [3] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-169-207:162592:162717 [3] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-169-207:162592:162717 [3] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-169-207:162592:162717 [3] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-169-207:162592:162717 [3] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-169-207:162592:162717 [3] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-169-207:162592:162717 [3] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-169-207:162592:162717 [3] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-169-207:162592:162717 [3] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-169-207:162592:162717 [3] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-169-207:162592:162717 [3] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-169-207:162592:162717 [3] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-169-207:162592:162717 [3] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-169-207:162592:162717 [3] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-169-207:162592:162717 [3] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-169-207:162592:162717 [3] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-169-207:162592:162717 [3] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-169-207:162592:162717 [3] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-169-207:162592:162717 [3] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-169-207:162592:162717 [3] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-169-207:162592:162717 [3] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-169-207:162592:162717 [3] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-169-207:162592:162717 [3] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-169-207:162592:162717 [3] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-169-207:162592:162717 [3] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/149
+ip-26-0-169-207:162592:162717 [3] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-169-207:162591:162715 [2] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-169-207:162591:162715 [2] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-169-207:162591:162715 [2] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-169-207:162591:162715 [2] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-169-207:162591:162715 [2] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-169-207:162591:162715 [2] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-169-207:162591:162715 [2] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-169-207:162591:162715 [2] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-169-207:162591:162715 [2] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-169-207:162591:162715 [2] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-169-207:162591:162715 [2] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-169-207:162591:162715 [2] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-169-207:162591:162715 [2] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-169-207:162591:162715 [2] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-169-207:162591:162715 [2] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-169-207:162591:162715 [2] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-169-207:162591:162715 [2] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-169-207:162591:162715 [2] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-169-207:162591:162715 [2] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-169-207:162591:162715 [2] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-169-207:162591:162715 [2] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-169-207:162591:162715 [2] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-169-207:162591:162715 [2] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-169-207:162591:162715 [2] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-169-207:162591:162715 [2] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-169-207:162591:162715 [2] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-169-207:162591:162715 [2] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-169-207:162591:162715 [2] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-169-207:162591:162715 [2] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-169-207:162591:162715 [2] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-169-207:162591:162715 [2] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-169-207:162591:162715 [2] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-169-207:162591:162715 [2] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/149
+ip-26-0-169-207:162591:162715 [2] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-169-207:162593:162714 [4] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-169-207:162593:162714 [4] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-169-207:162593:162714 [4] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-169-207:162593:162714 [4] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-169-207:162593:162714 [4] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-169-207:162593:162714 [4] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-169-207:162593:162714 [4] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-169-207:162593:162714 [4] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-169-207:162593:162714 [4] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-169-207:162593:162714 [4] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-169-207:162593:162714 [4] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-169-207:162593:162714 [4] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-169-207:162593:162714 [4] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-169-207:162593:162714 [4] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-169-207:162593:162714 [4] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-169-207:162593:162714 [4] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-169-207:162593:162714 [4] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-169-207:162593:162714 [4] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-169-207:162593:162714 [4] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-169-207:162593:162714 [4] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-169-207:162593:162714 [4] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-169-207:162593:162714 [4] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-169-207:162593:162714 [4] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-169-207:162593:162714 [4] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-169-207:162593:162714 [4] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-169-207:162593:162714 [4] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-169-207:162593:162714 [4] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-169-207:162593:162714 [4] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-169-207:162593:162714 [4] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-169-207:162593:162714 [4] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-169-207:162593:162714 [4] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-169-207:162593:162714 [4] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-169-207:162593:162714 [4] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/149
+ip-26-0-169-207:162593:162714 [4] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-169-207:162590:162718 [1] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-169-207:162590:162718 [1] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-169-207:162590:162718 [1] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-169-207:162590:162718 [1] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-169-207:162590:162718 [1] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-169-207:162590:162718 [1] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-169-207:162590:162718 [1] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-169-207:162590:162718 [1] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-169-207:162590:162718 [1] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-169-207:162590:162718 [1] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-169-207:162590:162718 [1] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-169-207:162590:162718 [1] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-169-207:162590:162718 [1] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-169-207:162590:162718 [1] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-169-207:162590:162718 [1] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-169-207:162590:162718 [1] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-169-207:162590:162718 [1] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-169-207:162590:162718 [1] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-169-207:162590:162718 [1] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-169-207:162590:162718 [1] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-169-207:162590:162718 [1] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-169-207:162590:162718 [1] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-169-207:162590:162718 [1] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-169-207:162590:162718 [1] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-169-207:162590:162718 [1] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-169-207:162590:162718 [1] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-169-207:162590:162718 [1] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-169-207:162590:162718 [1] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-169-207:162590:162718 [1] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-169-207:162590:162718 [1] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-169-207:162590:162718 [1] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-169-207:162590:162718 [1] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-169-207:162595:162712 [6] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-169-207:162595:162712 [6] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-169-207:162595:162712 [6] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-169-207:162595:162712 [6] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-169-207:162595:162712 [6] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-169-207:162595:162712 [6] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-169-207:162595:162712 [6] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-169-207:162595:162712 [6] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-169-207:162595:162712 [6] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-169-207:162595:162712 [6] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-169-207:162595:162712 [6] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-169-207:162595:162712 [6] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-169-207:162595:162712 [6] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-169-207:162595:162712 [6] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-169-207:162595:162712 [6] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-169-207:162595:162712 [6] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-169-207:162595:162712 [6] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-169-207:162595:162712 [6] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-169-207:162595:162712 [6] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-169-207:162595:162712 [6] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-169-207:162595:162712 [6] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-169-207:162595:162712 [6] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-169-207:162595:162712 [6] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-169-207:162595:162712 [6] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-169-207:162595:162712 [6] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-169-207:162595:162712 [6] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-169-207:162595:162712 [6] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-169-207:162595:162712 [6] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-169-207:162595:162712 [6] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-169-207:162595:162712 [6] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-169-207:162595:162712 [6] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-169-207:162595:162712 [6] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-169-207:162590:162718 [1] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/149
+ip-26-0-169-207:162590:162718 [1] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-169-207:162595:162712 [6] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/149
+ip-26-0-169-207:162595:162712 [6] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-169-207:162594:162713 [5] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-169-207:162594:162713 [5] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-169-207:162594:162713 [5] NCCL INFO Using network Libfabric
+ip-26-0-169-207:162591:162715 [2] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-169-207:162591:162715 [2] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-169-207:162591:162715 [2] NCCL INFO Using network Libfabric
+ip-26-0-169-207:162595:162712 [6] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-169-207:162595:162712 [6] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-169-207:162595:162712 [6] NCCL INFO Using network Libfabric
+ip-26-0-169-207:162593:162714 [4] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-169-207:162593:162714 [4] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-169-207:162593:162714 [4] NCCL INFO Using network Libfabric
+ip-26-0-169-207:162596:162716 [7] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-169-207:162596:162716 [7] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-169-207:162596:162716 [7] NCCL INFO Using network Libfabric
+ip-26-0-169-207:162590:162718 [1] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-169-207:162590:162718 [1] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-169-207:162590:162718 [1] NCCL INFO Using network Libfabric
+ip-26-0-169-207:162592:162717 [3] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-169-207:162592:162717 [3] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-169-207:162592:162717 [3] NCCL INFO Using network Libfabric
+ip-26-0-169-207:162596:162716 [7] NCCL INFO comm 0x870afe0 rank 7 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0x7811e3c53c62fe4 - Init START
+ip-26-0-169-207:162590:162718 [1] NCCL INFO comm 0x8a47ad0 rank 1 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0x7811e3c53c62fe4 - Init START
+ip-26-0-169-207:162591:162715 [2] NCCL INFO comm 0x8888440 rank 2 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0x7811e3c53c62fe4 - Init START
+ip-26-0-169-207:162592:162717 [3] NCCL INFO comm 0xa423730 rank 3 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0x7811e3c53c62fe4 - Init START
+ip-26-0-169-207:162594:162713 [5] NCCL INFO comm 0x92b5980 rank 5 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0x7811e3c53c62fe4 - Init START
+ip-26-0-169-207:162595:162712 [6] NCCL INFO comm 0x90da360 rank 6 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0x7811e3c53c62fe4 - Init START
+ip-26-0-169-207:162593:162714 [4] NCCL INFO comm 0x9bdf740 rank 4 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0x7811e3c53c62fe4 - Init START
+ip-26-0-169-207:162589:162711 [0] NCCL INFO comm 0x930ed10 rank 0 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0x7811e3c53c62fe4 - Init START
+ip-26-0-169-207:162589:162711 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/162
+ip-26-0-169-207:162590:162718 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/149
+ip-26-0-169-207:162591:162715 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/149
+ip-26-0-169-207:162596:162716 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/149
+ip-26-0-169-207:162593:162714 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/149
+ip-26-0-169-207:162594:162713 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/149
+ip-26-0-169-207:162592:162717 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/149
+ip-26-0-169-207:162595:162712 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/149
+ip-26-0-169-207:162593:162714 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-169-207:162593:162714 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-169-207:162591:162715 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-169-207:162591:162715 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-169-207:162592:162717 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-169-207:162592:162717 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-169-207:162596:162716 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-169-207:162596:162716 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-169-207:162595:162712 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-169-207:162595:162712 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-169-207:162590:162718 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-169-207:162590:162718 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-169-207:162589:162711 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-169-207:162594:162713 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-169-207:162594:162713 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-169-207:162589:162711 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-169-207:162591:162715 [2] NCCL INFO Trees [0] 3/-1/-1->2->1 [1] 3/-1/-1->2->1 [2] 3/-1/-1->2->1 [3] 3/-1/-1->2->1 [4] 3/-1/-1->2->1 [5] 3/-1/-1->2->1 [6] 3/-1/-1->2->1 [7] 3/-1/-1->2->1 [8] 3/-1/-1->2->1 [9] 3/-1/-1->2->1 [10] 3/-1/-1->2->1 [11] 3/-1/-1->2->1 [12] 3/-1/-1->2->1 [13] 3/-1/-1->2->1 [14] 3/-1/-1->2->1 [15] 3/-1/-1->2->1 [16] 3/-1/-1->2->1 [17] 3/-1/-1->2->1 [18] 3/-1/-1->2->1 [19] 3/-1/-1->2->1 [20] 3/-1/-1->2->1 [21] 3/-1/-1->2->1 [22] 3/-1/-1->2->1 [23] 3/-1/-1->2->1
+ip-26-0-169-207:162595:162712 [6] NCCL INFO Trees [0] 7/-1/-1->6->5 [1] 7/-1/-1->6->5 [2] 7/-1/-1->6->5 [3] 7/-1/-1->6->5 [4] 7/-1/-1->6->5 [5] 7/-1/-1->6->5 [6] 7/-1/-1->6->5 [7] 7/-1/-1->6->5 [8] 7/-1/-1->6->5 [9] 7/-1/-1->6->5 [10] 7/-1/-1->6->5 [11] 7/-1/-1->6->5 [12] 7/-1/-1->6->5 [13] 7/-1/-1->6->5 [14] 7/-1/-1->6->5 [15] 7/-1/-1->6->5 [16] 7/-1/-1->6->5 [17] 7/-1/-1->6->5 [18] 7/-1/-1->6->5 [19] 7/-1/-1->6->5 [20] 7/-1/-1->6->5 [21] 7/-1/-1->6->5 [22] 7/-1/-1->6->5 [23] 7/-1/-1->6->5
+ip-26-0-169-207:162594:162713 [5] NCCL INFO Trees [0] 6/-1/-1->5->4 [1] 6/-1/-1->5->4 [2] 6/-1/-1->5->4 [3] 6/-1/-1->5->4 [4] 6/-1/-1->5->4 [5] 6/-1/-1->5->4 [6] 6/-1/-1->5->4 [7] 6/-1/-1->5->4 [8] 6/-1/-1->5->4 [9] 6/-1/-1->5->4 [10] 6/-1/-1->5->4 [11] 6/-1/-1->5->4 [12] 6/-1/-1->5->4 [13] 6/-1/-1->5->4 [14] 6/-1/-1->5->4 [15] 6/-1/-1->5->4 [16] 6/-1/-1->5->4 [17] 6/-1/-1->5->4 [18] 6/-1/-1->5->4 [19] 6/-1/-1->5->4 [20] 6/-1/-1->5->4 [21] 6/-1/-1->5->4 [22] 6/-1/-1->5->4 [23] 6/-1/-1->5->4
+ip-26-0-169-207:162591:162715 [2] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-169-207:162595:162712 [6] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-169-207:162594:162713 [5] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-169-207:162593:162714 [4] NCCL INFO Trees [0] 5/-1/-1->4->3 [1] 5/-1/-1->4->3 [2] 5/-1/-1->4->3 [3] 5/-1/-1->4->3 [4] 5/-1/-1->4->3 [5] 5/-1/-1->4->3 [6] 5/-1/-1->4->3 [7] 5/-1/-1->4->3 [8] 5/-1/-1->4->3 [9] 5/-1/-1->4->3 [10] 5/-1/-1->4->3 [11] 5/-1/-1->4->3 [12] 5/-1/-1->4->3 [13] 5/-1/-1->4->3 [14] 5/-1/-1->4->3 [15] 5/-1/-1->4->3 [16] 5/-1/-1->4->3 [17] 5/-1/-1->4->3 [18] 5/-1/-1->4->3 [19] 5/-1/-1->4->3 [20] 5/-1/-1->4->3 [21] 5/-1/-1->4->3 [22] 5/-1/-1->4->3 [23] 5/-1/-1->4->3
+ip-26-0-169-207:162589:162711 [0] NCCL INFO Channel 00/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:162589:162711 [0] NCCL INFO Channel 01/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:162593:162714 [4] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-169-207:162589:162711 [0] NCCL INFO Channel 02/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:162589:162711 [0] NCCL INFO Channel 03/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:162589:162711 [0] NCCL INFO Channel 04/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:162596:162716 [7] NCCL INFO Trees [0] -1/-1/-1->7->6 [1] -1/-1/-1->7->6 [2] -1/-1/-1->7->6 [3] -1/-1/-1->7->6 [4] -1/-1/-1->7->6 [5] -1/-1/-1->7->6 [6] -1/-1/-1->7->6 [7] -1/-1/-1->7->6 [8] -1/-1/-1->7->6 [9] -1/-1/-1->7->6 [10] -1/-1/-1->7->6 [11] -1/-1/-1->7->6 [12] -1/-1/-1->7->6 [13] -1/-1/-1->7->6 [14] -1/-1/-1->7->6 [15] -1/-1/-1->7->6 [16] -1/-1/-1->7->6 [17] -1/-1/-1->7->6 [18] -1/-1/-1->7->6 [19] -1/-1/-1->7->6 [20] -1/-1/-1->7->6 [21] -1/-1/-1->7->6 [22] -1/-1/-1->7->6 [23] -1/-1/-1->7->6
+ip-26-0-169-207:162589:162711 [0] NCCL INFO Channel 05/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:162589:162711 [0] NCCL INFO Channel 06/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:162596:162716 [7] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-169-207:162589:162711 [0] NCCL INFO Channel 07/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:162589:162711 [0] NCCL INFO Channel 08/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:162589:162711 [0] NCCL INFO Channel 09/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:162589:162711 [0] NCCL INFO Channel 10/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:162589:162711 [0] NCCL INFO Channel 11/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:162589:162711 [0] NCCL INFO Channel 12/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:162589:162711 [0] NCCL INFO Channel 13/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:162592:162717 [3] NCCL INFO Trees [0] 4/-1/-1->3->2 [1] 4/-1/-1->3->2 [2] 4/-1/-1->3->2 [3] 4/-1/-1->3->2 [4] 4/-1/-1->3->2 [5] 4/-1/-1->3->2 [6] 4/-1/-1->3->2 [7] 4/-1/-1->3->2 [8] 4/-1/-1->3->2 [9] 4/-1/-1->3->2 [10] 4/-1/-1->3->2 [11] 4/-1/-1->3->2 [12] 4/-1/-1->3->2 [13] 4/-1/-1->3->2 [14] 4/-1/-1->3->2 [15] 4/-1/-1->3->2 [16] 4/-1/-1->3->2 [17] 4/-1/-1->3->2 [18] 4/-1/-1->3->2 [19] 4/-1/-1->3->2 [20] 4/-1/-1->3->2 [21] 4/-1/-1->3->2 [22] 4/-1/-1->3->2 [23] 4/-1/-1->3->2
+ip-26-0-169-207:162590:162718 [1] NCCL INFO Trees [0] 2/-1/-1->1->0 [1] 2/-1/-1->1->0 [2] 2/-1/-1->1->0 [3] 2/-1/-1->1->0 [4] 2/-1/-1->1->0 [5] 2/-1/-1->1->0 [6] 2/-1/-1->1->0 [7] 2/-1/-1->1->0 [8] 2/-1/-1->1->0 [9] 2/-1/-1->1->0 [10] 2/-1/-1->1->0 [11] 2/-1/-1->1->0 [12] 2/-1/-1->1->0 [13] 2/-1/-1->1->0 [14] 2/-1/-1->1->0 [15] 2/-1/-1->1->0 [16] 2/-1/-1->1->0 [17] 2/-1/-1->1->0 [18] 2/-1/-1->1->0 [19] 2/-1/-1->1->0 [20] 2/-1/-1->1->0 [21] 2/-1/-1->1->0 [22] 2/-1/-1->1->0 [23] 2/-1/-1->1->0
+ip-26-0-169-207:162589:162711 [0] NCCL INFO Channel 14/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:162589:162711 [0] NCCL INFO Channel 15/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:162589:162711 [0] NCCL INFO Channel 16/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:162589:162711 [0] NCCL INFO Channel 17/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:162589:162711 [0] NCCL INFO Channel 18/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:162589:162711 [0] NCCL INFO Channel 19/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:162589:162711 [0] NCCL INFO Channel 20/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:162589:162711 [0] NCCL INFO Channel 21/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:162589:162711 [0] NCCL INFO Channel 22/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:162589:162711 [0] NCCL INFO Channel 23/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:162589:162711 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] 1/-1/-1->0->-1 [7] 1/-1/-1->0->-1 [8] 1/-1/-1->0->-1 [9] 1/-1/-1->0->-1 [10] 1/-1/-1->0->-1 [11] 1/-1/-1->0->-1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] 1/-1/-1->0->-1 [19] 1/-1/-1->0->-1 [20] 1/-1/-1->0->-1 [21] 1/-1/-1->0->-1 [22] 1/-1/-1->0->-1 [23] 1/-1/-1->0->-1
+ip-26-0-169-207:162589:162711 [0] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-169-207:162590:162718 [1] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-169-207:162592:162717 [3] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-169-207:162596:162716 [7] NCCL INFO Channel 00/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162596:162716 [7] NCCL INFO Channel 01/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162596:162716 [7] NCCL INFO Channel 02/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162590:162718 [1] NCCL INFO Channel 00/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162596:162716 [7] NCCL INFO Channel 03/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162593:162714 [4] NCCL INFO Channel 00/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162590:162718 [1] NCCL INFO Channel 01/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162596:162716 [7] NCCL INFO Channel 04/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162593:162714 [4] NCCL INFO Channel 01/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162592:162717 [3] NCCL INFO Channel 00/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162590:162718 [1] NCCL INFO Channel 02/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162593:162714 [4] NCCL INFO Channel 02/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162596:162716 [7] NCCL INFO Channel 05/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162592:162717 [3] NCCL INFO Channel 01/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162590:162718 [1] NCCL INFO Channel 03/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162596:162716 [7] NCCL INFO Channel 06/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162593:162714 [4] NCCL INFO Channel 03/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162592:162717 [3] NCCL INFO Channel 02/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162590:162718 [1] NCCL INFO Channel 04/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162596:162716 [7] NCCL INFO Channel 07/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162593:162714 [4] NCCL INFO Channel 04/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162590:162718 [1] NCCL INFO Channel 05/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162592:162717 [3] NCCL INFO Channel 03/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162596:162716 [7] NCCL INFO Channel 08/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162593:162714 [4] NCCL INFO Channel 05/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162590:162718 [1] NCCL INFO Channel 06/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162592:162717 [3] NCCL INFO Channel 04/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162596:162716 [7] NCCL INFO Channel 09/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162593:162714 [4] NCCL INFO Channel 06/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162590:162718 [1] NCCL INFO Channel 07/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162592:162717 [3] NCCL INFO Channel 05/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162596:162716 [7] NCCL INFO Channel 10/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162593:162714 [4] NCCL INFO Channel 07/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162592:162717 [3] NCCL INFO Channel 06/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162590:162718 [1] NCCL INFO Channel 08/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162596:162716 [7] NCCL INFO Channel 11/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162593:162714 [4] NCCL INFO Channel 08/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162591:162715 [2] NCCL INFO Channel 00/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162590:162718 [1] NCCL INFO Channel 09/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162592:162717 [3] NCCL INFO Channel 07/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162596:162716 [7] NCCL INFO Channel 12/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162593:162714 [4] NCCL INFO Channel 09/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162594:162713 [5] NCCL INFO Channel 00/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162591:162715 [2] NCCL INFO Channel 01/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162589:162711 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162592:162717 [3] NCCL INFO Channel 08/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162590:162718 [1] NCCL INFO Channel 10/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162596:162716 [7] NCCL INFO Channel 13/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162593:162714 [4] NCCL INFO Channel 10/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162594:162713 [5] NCCL INFO Channel 01/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162591:162715 [2] NCCL INFO Channel 02/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162589:162711 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162592:162717 [3] NCCL INFO Channel 09/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162590:162718 [1] NCCL INFO Channel 11/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162596:162716 [7] NCCL INFO Channel 14/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162593:162714 [4] NCCL INFO Channel 11/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162594:162713 [5] NCCL INFO Channel 02/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162591:162715 [2] NCCL INFO Channel 03/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162589:162711 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162592:162717 [3] NCCL INFO Channel 10/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162590:162718 [1] NCCL INFO Channel 12/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162596:162716 [7] NCCL INFO Channel 15/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162593:162714 [4] NCCL INFO Channel 12/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162594:162713 [5] NCCL INFO Channel 03/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162591:162715 [2] NCCL INFO Channel 04/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162595:162712 [6] NCCL INFO Channel 00/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:162592:162717 [3] NCCL INFO Channel 11/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162589:162711 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162590:162718 [1] NCCL INFO Channel 13/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162596:162716 [7] NCCL INFO Channel 16/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162593:162714 [4] NCCL INFO Channel 13/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162594:162713 [5] NCCL INFO Channel 04/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162591:162715 [2] NCCL INFO Channel 05/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162595:162712 [6] NCCL INFO Channel 01/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:162592:162717 [3] NCCL INFO Channel 12/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162589:162711 [0] NCCL INFO Channel 04/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162590:162718 [1] NCCL INFO Channel 14/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162596:162716 [7] NCCL INFO Channel 17/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162593:162714 [4] NCCL INFO Channel 14/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162594:162713 [5] NCCL INFO Channel 05/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162591:162715 [2] NCCL INFO Channel 06/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162595:162712 [6] NCCL INFO Channel 02/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:162592:162717 [3] NCCL INFO Channel 13/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162589:162711 [0] NCCL INFO Channel 05/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162590:162718 [1] NCCL INFO Channel 15/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162596:162716 [7] NCCL INFO Channel 18/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162593:162714 [4] NCCL INFO Channel 15/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162594:162713 [5] NCCL INFO Channel 06/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162591:162715 [2] NCCL INFO Channel 07/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162595:162712 [6] NCCL INFO Channel 03/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:162592:162717 [3] NCCL INFO Channel 14/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162589:162711 [0] NCCL INFO Channel 06/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162590:162718 [1] NCCL INFO Channel 16/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162596:162716 [7] NCCL INFO Channel 19/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162593:162714 [4] NCCL INFO Channel 16/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162594:162713 [5] NCCL INFO Channel 07/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162591:162715 [2] NCCL INFO Channel 08/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162595:162712 [6] NCCL INFO Channel 04/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:162592:162717 [3] NCCL INFO Channel 15/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162589:162711 [0] NCCL INFO Channel 07/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162596:162716 [7] NCCL INFO Channel 20/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162590:162718 [1] NCCL INFO Channel 17/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162593:162714 [4] NCCL INFO Channel 17/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162594:162713 [5] NCCL INFO Channel 08/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162591:162715 [2] NCCL INFO Channel 09/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162589:162711 [0] NCCL INFO Channel 08/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162593:162714 [4] NCCL INFO Channel 18/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162596:162716 [7] NCCL INFO Channel 21/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162589:162711 [0] NCCL INFO Channel 09/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162594:162713 [5] NCCL INFO Channel 09/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162593:162714 [4] NCCL INFO Channel 19/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162592:162717 [3] NCCL INFO Channel 16/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162596:162716 [7] NCCL INFO Channel 22/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162589:162711 [0] NCCL INFO Channel 10/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162594:162713 [5] NCCL INFO Channel 10/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162593:162714 [4] NCCL INFO Channel 20/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162592:162717 [3] NCCL INFO Channel 17/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162596:162716 [7] NCCL INFO Channel 23/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162589:162711 [0] NCCL INFO Channel 11/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162594:162713 [5] NCCL INFO Channel 11/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162593:162714 [4] NCCL INFO Channel 21/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162592:162717 [3] NCCL INFO Channel 18/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162589:162711 [0] NCCL INFO Channel 12/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162594:162713 [5] NCCL INFO Channel 12/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162593:162714 [4] NCCL INFO Channel 22/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162592:162717 [3] NCCL INFO Channel 19/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162590:162718 [1] NCCL INFO Channel 18/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162594:162713 [5] NCCL INFO Channel 13/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162593:162714 [4] NCCL INFO Channel 23/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162589:162711 [0] NCCL INFO Channel 13/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162592:162717 [3] NCCL INFO Channel 20/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162590:162718 [1] NCCL INFO Channel 19/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162594:162713 [5] NCCL INFO Channel 14/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162589:162711 [0] NCCL INFO Channel 14/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162592:162717 [3] NCCL INFO Channel 21/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162590:162718 [1] NCCL INFO Channel 20/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162594:162713 [5] NCCL INFO Channel 15/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162589:162711 [0] NCCL INFO Channel 15/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162592:162717 [3] NCCL INFO Channel 22/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162590:162718 [1] NCCL INFO Channel 21/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162594:162713 [5] NCCL INFO Channel 16/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162589:162711 [0] NCCL INFO Channel 16/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162592:162717 [3] NCCL INFO Channel 23/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162595:162712 [6] NCCL INFO Channel 05/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:162590:162718 [1] NCCL INFO Channel 22/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162594:162713 [5] NCCL INFO Channel 17/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162589:162711 [0] NCCL INFO Channel 17/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162595:162712 [6] NCCL INFO Channel 06/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:162590:162718 [1] NCCL INFO Channel 23/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162591:162715 [2] NCCL INFO Channel 10/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162594:162713 [5] NCCL INFO Channel 18/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162589:162711 [0] NCCL INFO Channel 18/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162595:162712 [6] NCCL INFO Channel 07/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:162591:162715 [2] NCCL INFO Channel 11/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162594:162713 [5] NCCL INFO Channel 19/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162589:162711 [0] NCCL INFO Channel 19/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162595:162712 [6] NCCL INFO Channel 08/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:162591:162715 [2] NCCL INFO Channel 12/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162594:162713 [5] NCCL INFO Channel 20/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162589:162711 [0] NCCL INFO Channel 20/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162595:162712 [6] NCCL INFO Channel 09/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:162591:162715 [2] NCCL INFO Channel 13/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162594:162713 [5] NCCL INFO Channel 21/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162589:162711 [0] NCCL INFO Channel 21/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162595:162712 [6] NCCL INFO Channel 10/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:162591:162715 [2] NCCL INFO Channel 14/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162594:162713 [5] NCCL INFO Channel 22/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162595:162712 [6] NCCL INFO Channel 11/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:162589:162711 [0] NCCL INFO Channel 22/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162591:162715 [2] NCCL INFO Channel 15/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162594:162713 [5] NCCL INFO Channel 23/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162595:162712 [6] NCCL INFO Channel 12/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:162591:162715 [2] NCCL INFO Channel 16/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162589:162711 [0] NCCL INFO Channel 23/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162595:162712 [6] NCCL INFO Channel 13/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:162591:162715 [2] NCCL INFO Channel 17/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162595:162712 [6] NCCL INFO Channel 14/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:162591:162715 [2] NCCL INFO Channel 18/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162591:162715 [2] NCCL INFO Channel 19/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162595:162712 [6] NCCL INFO Channel 15/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:162591:162715 [2] NCCL INFO Channel 20/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162595:162712 [6] NCCL INFO Channel 16/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:162591:162715 [2] NCCL INFO Channel 21/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162595:162712 [6] NCCL INFO Channel 17/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:162591:162715 [2] NCCL INFO Channel 22/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162595:162712 [6] NCCL INFO Channel 18/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:162591:162715 [2] NCCL INFO Channel 23/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162595:162712 [6] NCCL INFO Channel 19/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:162595:162712 [6] NCCL INFO Channel 20/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:162595:162712 [6] NCCL INFO Channel 21/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:162595:162712 [6] NCCL INFO Channel 22/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:162595:162712 [6] NCCL INFO Channel 23/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:162589:162711 [0] NCCL INFO Connected all rings
+ip-26-0-169-207:162593:162714 [4] NCCL INFO Connected all rings
+ip-26-0-169-207:162591:162715 [2] NCCL INFO Connected all rings
+ip-26-0-169-207:162592:162717 [3] NCCL INFO Connected all rings
+ip-26-0-169-207:162590:162718 [1] NCCL INFO Connected all rings
+ip-26-0-169-207:162595:162712 [6] NCCL INFO Connected all rings
+ip-26-0-169-207:162594:162713 [5] NCCL INFO Connected all rings
+ip-26-0-169-207:162596:162716 [7] NCCL INFO Connected all rings
+ip-26-0-169-207:162596:162716 [7] NCCL INFO Channel 00/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162596:162716 [7] NCCL INFO Channel 01/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162596:162716 [7] NCCL INFO Channel 02/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162596:162716 [7] NCCL INFO Channel 03/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162596:162716 [7] NCCL INFO Channel 04/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162596:162716 [7] NCCL INFO Channel 05/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162596:162716 [7] NCCL INFO Channel 06/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162596:162716 [7] NCCL INFO Channel 07/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162593:162714 [4] NCCL INFO Channel 00/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162596:162716 [7] NCCL INFO Channel 08/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162593:162714 [4] NCCL INFO Channel 01/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162596:162716 [7] NCCL INFO Channel 09/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162593:162714 [4] NCCL INFO Channel 02/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162596:162716 [7] NCCL INFO Channel 10/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162593:162714 [4] NCCL INFO Channel 03/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162596:162716 [7] NCCL INFO Channel 11/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162593:162714 [4] NCCL INFO Channel 04/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162596:162716 [7] NCCL INFO Channel 12/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162591:162715 [2] NCCL INFO Channel 00/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162593:162714 [4] NCCL INFO Channel 05/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162596:162716 [7] NCCL INFO Channel 13/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162591:162715 [2] NCCL INFO Channel 01/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162593:162714 [4] NCCL INFO Channel 06/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162596:162716 [7] NCCL INFO Channel 14/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162591:162715 [2] NCCL INFO Channel 02/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162593:162714 [4] NCCL INFO Channel 07/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162596:162716 [7] NCCL INFO Channel 15/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162591:162715 [2] NCCL INFO Channel 03/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162593:162714 [4] NCCL INFO Channel 08/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162596:162716 [7] NCCL INFO Channel 16/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162591:162715 [2] NCCL INFO Channel 04/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162593:162714 [4] NCCL INFO Channel 09/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162596:162716 [7] NCCL INFO Channel 17/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162591:162715 [2] NCCL INFO Channel 05/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162593:162714 [4] NCCL INFO Channel 10/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162596:162716 [7] NCCL INFO Channel 18/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162591:162715 [2] NCCL INFO Channel 06/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162593:162714 [4] NCCL INFO Channel 11/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162596:162716 [7] NCCL INFO Channel 19/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162592:162717 [3] NCCL INFO Channel 00/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162591:162715 [2] NCCL INFO Channel 07/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162593:162714 [4] NCCL INFO Channel 12/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162596:162716 [7] NCCL INFO Channel 20/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162590:162718 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162591:162715 [2] NCCL INFO Channel 08/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162592:162717 [3] NCCL INFO Channel 01/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162593:162714 [4] NCCL INFO Channel 13/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162596:162716 [7] NCCL INFO Channel 21/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162590:162718 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162591:162715 [2] NCCL INFO Channel 09/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162592:162717 [3] NCCL INFO Channel 02/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162595:162712 [6] NCCL INFO Channel 00/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162593:162714 [4] NCCL INFO Channel 14/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162596:162716 [7] NCCL INFO Channel 22/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162591:162715 [2] NCCL INFO Channel 10/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162592:162717 [3] NCCL INFO Channel 03/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162595:162712 [6] NCCL INFO Channel 01/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162590:162718 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162594:162713 [5] NCCL INFO Channel 00/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162593:162714 [4] NCCL INFO Channel 15/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162596:162716 [7] NCCL INFO Channel 23/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162591:162715 [2] NCCL INFO Channel 11/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162592:162717 [3] NCCL INFO Channel 04/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162595:162712 [6] NCCL INFO Channel 02/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162594:162713 [5] NCCL INFO Channel 01/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162590:162718 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162593:162714 [4] NCCL INFO Channel 16/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162591:162715 [2] NCCL INFO Channel 12/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162592:162717 [3] NCCL INFO Channel 05/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162595:162712 [6] NCCL INFO Channel 03/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162594:162713 [5] NCCL INFO Channel 02/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162590:162718 [1] NCCL INFO Channel 04/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162593:162714 [4] NCCL INFO Channel 17/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162591:162715 [2] NCCL INFO Channel 13/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162592:162717 [3] NCCL INFO Channel 06/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162595:162712 [6] NCCL INFO Channel 04/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162590:162718 [1] NCCL INFO Channel 05/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162593:162714 [4] NCCL INFO Channel 18/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162591:162715 [2] NCCL INFO Channel 14/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162592:162717 [3] NCCL INFO Channel 07/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162595:162712 [6] NCCL INFO Channel 05/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162593:162714 [4] NCCL INFO Channel 19/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162594:162713 [5] NCCL INFO Channel 03/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162590:162718 [1] NCCL INFO Channel 06/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162591:162715 [2] NCCL INFO Channel 15/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162592:162717 [3] NCCL INFO Channel 08/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162595:162712 [6] NCCL INFO Channel 06/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162593:162714 [4] NCCL INFO Channel 20/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162594:162713 [5] NCCL INFO Channel 04/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162590:162718 [1] NCCL INFO Channel 07/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162591:162715 [2] NCCL INFO Channel 16/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162592:162717 [3] NCCL INFO Channel 09/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162595:162712 [6] NCCL INFO Channel 07/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162593:162714 [4] NCCL INFO Channel 21/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162594:162713 [5] NCCL INFO Channel 05/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162590:162718 [1] NCCL INFO Channel 08/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162591:162715 [2] NCCL INFO Channel 17/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162592:162717 [3] NCCL INFO Channel 10/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162595:162712 [6] NCCL INFO Channel 08/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162593:162714 [4] NCCL INFO Channel 22/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162594:162713 [5] NCCL INFO Channel 06/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162590:162718 [1] NCCL INFO Channel 09/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162591:162715 [2] NCCL INFO Channel 18/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162592:162717 [3] NCCL INFO Channel 11/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162595:162712 [6] NCCL INFO Channel 09/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162593:162714 [4] NCCL INFO Channel 23/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162594:162713 [5] NCCL INFO Channel 07/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162590:162718 [1] NCCL INFO Channel 10/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162591:162715 [2] NCCL INFO Channel 19/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162592:162717 [3] NCCL INFO Channel 12/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162595:162712 [6] NCCL INFO Channel 10/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162594:162713 [5] NCCL INFO Channel 08/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162590:162718 [1] NCCL INFO Channel 11/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162591:162715 [2] NCCL INFO Channel 20/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162592:162717 [3] NCCL INFO Channel 13/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162595:162712 [6] NCCL INFO Channel 11/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162594:162713 [5] NCCL INFO Channel 09/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162590:162718 [1] NCCL INFO Channel 12/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162591:162715 [2] NCCL INFO Channel 21/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162592:162717 [3] NCCL INFO Channel 14/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162595:162712 [6] NCCL INFO Channel 12/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162594:162713 [5] NCCL INFO Channel 10/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162590:162718 [1] NCCL INFO Channel 13/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162591:162715 [2] NCCL INFO Channel 22/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162592:162717 [3] NCCL INFO Channel 15/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162595:162712 [6] NCCL INFO Channel 13/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162594:162713 [5] NCCL INFO Channel 11/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162590:162718 [1] NCCL INFO Channel 14/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162591:162715 [2] NCCL INFO Channel 23/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162592:162717 [3] NCCL INFO Channel 16/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162595:162712 [6] NCCL INFO Channel 14/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162594:162713 [5] NCCL INFO Channel 12/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162590:162718 [1] NCCL INFO Channel 15/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162592:162717 [3] NCCL INFO Channel 17/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162595:162712 [6] NCCL INFO Channel 15/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162594:162713 [5] NCCL INFO Channel 13/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162590:162718 [1] NCCL INFO Channel 16/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162592:162717 [3] NCCL INFO Channel 18/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162595:162712 [6] NCCL INFO Channel 16/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162594:162713 [5] NCCL INFO Channel 14/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162590:162718 [1] NCCL INFO Channel 17/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162592:162717 [3] NCCL INFO Channel 19/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162595:162712 [6] NCCL INFO Channel 17/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162594:162713 [5] NCCL INFO Channel 15/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162590:162718 [1] NCCL INFO Channel 18/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162592:162717 [3] NCCL INFO Channel 20/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162595:162712 [6] NCCL INFO Channel 18/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162594:162713 [5] NCCL INFO Channel 16/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162590:162718 [1] NCCL INFO Channel 19/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162592:162717 [3] NCCL INFO Channel 21/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162595:162712 [6] NCCL INFO Channel 19/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162594:162713 [5] NCCL INFO Channel 17/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162590:162718 [1] NCCL INFO Channel 20/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162592:162717 [3] NCCL INFO Channel 22/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162595:162712 [6] NCCL INFO Channel 20/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162594:162713 [5] NCCL INFO Channel 18/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162592:162717 [3] NCCL INFO Channel 23/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162590:162718 [1] NCCL INFO Channel 21/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162595:162712 [6] NCCL INFO Channel 21/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162594:162713 [5] NCCL INFO Channel 19/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162590:162718 [1] NCCL INFO Channel 22/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162595:162712 [6] NCCL INFO Channel 22/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162594:162713 [5] NCCL INFO Channel 20/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162590:162718 [1] NCCL INFO Channel 23/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162595:162712 [6] NCCL INFO Channel 23/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162594:162713 [5] NCCL INFO Channel 21/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162594:162713 [5] NCCL INFO Channel 22/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162594:162713 [5] NCCL INFO Channel 23/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162596:162716 [7] NCCL INFO Connected all trees
+ip-26-0-169-207:162596:162716 [7] NCCL INFO NVLS comm 0x870afe0 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-169-207:162589:162711 [0] NCCL INFO Connected all trees
+ip-26-0-169-207:162589:162711 [0] NCCL INFO NVLS comm 0x930ed10 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-169-207:162592:162717 [3] NCCL INFO Connected all trees
+ip-26-0-169-207:162592:162717 [3] NCCL INFO NVLS comm 0xa423730 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-169-207:162591:162715 [2] NCCL INFO Connected all trees
+ip-26-0-169-207:162591:162715 [2] NCCL INFO NVLS comm 0x8888440 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-169-207:162595:162712 [6] NCCL INFO Connected all trees
+ip-26-0-169-207:162595:162712 [6] NCCL INFO NVLS comm 0x90da360 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-169-207:162590:162718 [1] NCCL INFO Connected all trees
+ip-26-0-169-207:162590:162718 [1] NCCL INFO NVLS comm 0x8a47ad0 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-169-207:162594:162713 [5] NCCL INFO Connected all trees
+ip-26-0-169-207:162594:162713 [5] NCCL INFO NVLS comm 0x92b5980 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-169-207:162593:162714 [4] NCCL INFO Connected all trees
+ip-26-0-169-207:162593:162714 [4] NCCL INFO NVLS comm 0x9bdf740 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-169-207:162590:162718 [1] NCCL INFO Connected NVLS tree
+ip-26-0-169-207:162590:162718 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-169-207:162590:162718 [1] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-169-207:162590:162718 [1] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:162596:162716 [7] NCCL INFO Connected NVLS tree
+ip-26-0-169-207:162596:162716 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-169-207:162596:162716 [7] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-169-207:162596:162716 [7] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:162593:162714 [4] NCCL INFO Connected NVLS tree
+ip-26-0-169-207:162595:162712 [6] NCCL INFO Connected NVLS tree
+ip-26-0-169-207:162595:162712 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-169-207:162593:162714 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-169-207:162595:162712 [6] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-169-207:162594:162713 [5] NCCL INFO Connected NVLS tree
+ip-26-0-169-207:162593:162714 [4] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-169-207:162592:162717 [3] NCCL INFO Connected NVLS tree
+ip-26-0-169-207:162595:162712 [6] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:162593:162714 [4] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:162594:162713 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-169-207:162594:162713 [5] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-169-207:162594:162713 [5] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:162592:162717 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-169-207:162592:162717 [3] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-169-207:162592:162717 [3] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:162591:162715 [2] NCCL INFO Connected NVLS tree
+ip-26-0-169-207:162591:162715 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-169-207:162591:162715 [2] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-169-207:162591:162715 [2] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:162589:162711 [0] NCCL INFO Connected NVLS tree
+ip-26-0-169-207:162589:162711 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-169-207:162589:162711 [0] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-169-207:162589:162711 [0] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:162591:162715 [2] NCCL INFO comm 0x8888440 rank 2 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0x7811e3c53c62fe4 - Init COMPLETE
+ip-26-0-169-207:162595:162712 [6] NCCL INFO comm 0x90da360 rank 6 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0x7811e3c53c62fe4 - Init COMPLETE
+ip-26-0-169-207:162590:162718 [1] NCCL INFO comm 0x8a47ad0 rank 1 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0x7811e3c53c62fe4 - Init COMPLETE
+ip-26-0-169-207:162596:162716 [7] NCCL INFO comm 0x870afe0 rank 7 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0x7811e3c53c62fe4 - Init COMPLETE
+ip-26-0-169-207:162592:162717 [3] NCCL INFO comm 0xa423730 rank 3 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0x7811e3c53c62fe4 - Init COMPLETE
+ip-26-0-169-207:162594:162713 [5] NCCL INFO comm 0x92b5980 rank 5 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0x7811e3c53c62fe4 - Init COMPLETE
+ip-26-0-169-207:162589:162711 [0] NCCL INFO comm 0x930ed10 rank 0 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0x7811e3c53c62fe4 - Init COMPLETE
+ip-26-0-169-207:162593:162714 [4] NCCL INFO comm 0x9bdf740 rank 4 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0x7811e3c53c62fe4 - Init COMPLETE
+ip-26-0-169-207:162589:162782 [0] NCCL INFO Using network Libfabric
+ip-26-0-169-207:162593:162786 [4] NCCL INFO Using network Libfabric
+ip-26-0-169-207:162590:162785 [1] NCCL INFO Using network Libfabric
+ip-26-0-169-207:162596:162783 [7] NCCL INFO Using network Libfabric
+ip-26-0-169-207:162592:162784 [3] NCCL INFO Using network Libfabric
+ip-26-0-169-207:162595:162787 [6] NCCL INFO Using network Libfabric
+ip-26-0-169-207:162594:162788 [5] NCCL INFO Using network Libfabric
+ip-26-0-169-207:162591:162789 [2] NCCL INFO Using network Libfabric
+ip-26-0-169-207:162589:162782 [0] NCCL INFO comm 0x9538ce0 rank 0 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0xe3a983ad389222f8 - Init START
+ip-26-0-169-207:162590:162785 [1] NCCL INFO comm 0x8c71020 rank 1 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0xe3a983ad389222f8 - Init START
+ip-26-0-169-207:162596:162783 [7] NCCL INFO comm 0x8934de0 rank 7 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0xe3a983ad389222f8 - Init START
+ip-26-0-169-207:162591:162789 [2] NCCL INFO comm 0x8ab1f00 rank 2 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0xe3a983ad389222f8 - Init START
+ip-26-0-169-207:162594:162788 [5] NCCL INFO comm 0x94df540 rank 5 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0xe3a983ad389222f8 - Init START
+ip-26-0-169-207:162592:162784 [3] NCCL INFO comm 0xa64e300 rank 3 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0xe3a983ad389222f8 - Init START
+ip-26-0-169-207:162593:162786 [4] NCCL INFO comm 0x9e09520 rank 4 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0xe3a983ad389222f8 - Init START
+ip-26-0-169-207:162595:162787 [6] NCCL INFO comm 0x9303c90 rank 6 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0xe3a983ad389222f8 - Init START
+ip-26-0-169-207:162589:162782 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/162
+ip-26-0-169-207:162591:162789 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/149
+ip-26-0-169-207:162596:162783 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/149
+ip-26-0-169-207:162594:162788 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/149
+ip-26-0-169-207:162592:162784 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/149
+ip-26-0-169-207:162595:162787 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/149
+ip-26-0-169-207:162590:162785 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/149
+ip-26-0-169-207:162593:162786 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/149
+ip-26-0-169-207:162592:162784 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-169-207:162592:162784 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-169-207:162596:162783 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-169-207:162596:162783 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-169-207:162591:162789 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-169-207:162591:162789 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-169-207:162593:162786 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-169-207:162593:162786 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-169-207:162589:162782 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-169-207:162589:162782 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-169-207:162594:162788 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-169-207:162594:162788 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-169-207:162590:162785 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-169-207:162590:162785 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-169-207:162595:162787 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-169-207:162595:162787 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-169-207:162595:162787 [6] NCCL INFO Trees [0] 7/-1/-1->6->5 [1] 7/-1/-1->6->5 [2] 7/-1/-1->6->5 [3] 7/-1/-1->6->5 [4] 7/-1/-1->6->5 [5] 7/-1/-1->6->5 [6] 7/-1/-1->6->5 [7] 7/-1/-1->6->5 [8] 7/-1/-1->6->5 [9] 7/-1/-1->6->5 [10] 7/-1/-1->6->5 [11] 7/-1/-1->6->5 [12] 7/-1/-1->6->5 [13] 7/-1/-1->6->5 [14] 7/-1/-1->6->5 [15] 7/-1/-1->6->5 [16] 7/-1/-1->6->5 [17] 7/-1/-1->6->5 [18] 7/-1/-1->6->5 [19] 7/-1/-1->6->5 [20] 7/-1/-1->6->5 [21] 7/-1/-1->6->5 [22] 7/-1/-1->6->5 [23] 7/-1/-1->6->5
+ip-26-0-169-207:162596:162783 [7] NCCL INFO Trees [0] -1/-1/-1->7->6 [1] -1/-1/-1->7->6 [2] -1/-1/-1->7->6 [3] -1/-1/-1->7->6 [4] -1/-1/-1->7->6 [5] -1/-1/-1->7->6 [6] -1/-1/-1->7->6 [7] -1/-1/-1->7->6 [8] -1/-1/-1->7->6 [9] -1/-1/-1->7->6 [10] -1/-1/-1->7->6 [11] -1/-1/-1->7->6 [12] -1/-1/-1->7->6 [13] -1/-1/-1->7->6 [14] -1/-1/-1->7->6 [15] -1/-1/-1->7->6 [16] -1/-1/-1->7->6 [17] -1/-1/-1->7->6 [18] -1/-1/-1->7->6 [19] -1/-1/-1->7->6 [20] -1/-1/-1->7->6 [21] -1/-1/-1->7->6 [22] -1/-1/-1->7->6 [23] -1/-1/-1->7->6
+ip-26-0-169-207:162595:162787 [6] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-169-207:162596:162783 [7] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-169-207:162589:162782 [0] NCCL INFO Channel 00/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:162589:162782 [0] NCCL INFO Channel 01/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:162589:162782 [0] NCCL INFO Channel 02/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:162589:162782 [0] NCCL INFO Channel 03/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:162589:162782 [0] NCCL INFO Channel 04/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:162589:162782 [0] NCCL INFO Channel 05/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:162589:162782 [0] NCCL INFO Channel 06/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:162589:162782 [0] NCCL INFO Channel 07/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:162589:162782 [0] NCCL INFO Channel 08/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:162589:162782 [0] NCCL INFO Channel 09/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:162589:162782 [0] NCCL INFO Channel 10/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:162589:162782 [0] NCCL INFO Channel 11/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:162589:162782 [0] NCCL INFO Channel 12/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:162589:162782 [0] NCCL INFO Channel 13/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:162589:162782 [0] NCCL INFO Channel 14/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:162589:162782 [0] NCCL INFO Channel 15/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:162589:162782 [0] NCCL INFO Channel 16/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:162589:162782 [0] NCCL INFO Channel 17/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:162589:162782 [0] NCCL INFO Channel 18/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:162589:162782 [0] NCCL INFO Channel 19/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:162589:162782 [0] NCCL INFO Channel 20/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:162589:162782 [0] NCCL INFO Channel 21/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:162589:162782 [0] NCCL INFO Channel 22/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:162589:162782 [0] NCCL INFO Channel 23/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:162589:162782 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] 1/-1/-1->0->-1 [7] 1/-1/-1->0->-1 [8] 1/-1/-1->0->-1 [9] 1/-1/-1->0->-1 [10] 1/-1/-1->0->-1 [11] 1/-1/-1->0->-1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] 1/-1/-1->0->-1 [19] 1/-1/-1->0->-1 [20] 1/-1/-1->0->-1 [21] 1/-1/-1->0->-1 [22] 1/-1/-1->0->-1 [23] 1/-1/-1->0->-1
+ip-26-0-169-207:162589:162782 [0] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-169-207:162591:162789 [2] NCCL INFO Trees [0] 3/-1/-1->2->1 [1] 3/-1/-1->2->1 [2] 3/-1/-1->2->1 [3] 3/-1/-1->2->1 [4] 3/-1/-1->2->1 [5] 3/-1/-1->2->1 [6] 3/-1/-1->2->1 [7] 3/-1/-1->2->1 [8] 3/-1/-1->2->1 [9] 3/-1/-1->2->1 [10] 3/-1/-1->2->1 [11] 3/-1/-1->2->1 [12] 3/-1/-1->2->1 [13] 3/-1/-1->2->1 [14] 3/-1/-1->2->1 [15] 3/-1/-1->2->1 [16] 3/-1/-1->2->1 [17] 3/-1/-1->2->1 [18] 3/-1/-1->2->1 [19] 3/-1/-1->2->1 [20] 3/-1/-1->2->1 [21] 3/-1/-1->2->1 [22] 3/-1/-1->2->1 [23] 3/-1/-1->2->1
+ip-26-0-169-207:162591:162789 [2] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-169-207:162594:162788 [5] NCCL INFO Trees [0] 6/-1/-1->5->4 [1] 6/-1/-1->5->4 [2] 6/-1/-1->5->4 [3] 6/-1/-1->5->4 [4] 6/-1/-1->5->4 [5] 6/-1/-1->5->4 [6] 6/-1/-1->5->4 [7] 6/-1/-1->5->4 [8] 6/-1/-1->5->4 [9] 6/-1/-1->5->4 [10] 6/-1/-1->5->4 [11] 6/-1/-1->5->4 [12] 6/-1/-1->5->4 [13] 6/-1/-1->5->4 [14] 6/-1/-1->5->4 [15] 6/-1/-1->5->4 [16] 6/-1/-1->5->4 [17] 6/-1/-1->5->4 [18] 6/-1/-1->5->4 [19] 6/-1/-1->5->4 [20] 6/-1/-1->5->4 [21] 6/-1/-1->5->4 [22] 6/-1/-1->5->4 [23] 6/-1/-1->5->4
+ip-26-0-169-207:162594:162788 [5] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-169-207:162593:162786 [4] NCCL INFO Trees [0] 5/-1/-1->4->3 [1] 5/-1/-1->4->3 [2] 5/-1/-1->4->3 [3] 5/-1/-1->4->3 [4] 5/-1/-1->4->3 [5] 5/-1/-1->4->3 [6] 5/-1/-1->4->3 [7] 5/-1/-1->4->3 [8] 5/-1/-1->4->3 [9] 5/-1/-1->4->3 [10] 5/-1/-1->4->3 [11] 5/-1/-1->4->3 [12] 5/-1/-1->4->3 [13] 5/-1/-1->4->3 [14] 5/-1/-1->4->3 [15] 5/-1/-1->4->3 [16] 5/-1/-1->4->3 [17] 5/-1/-1->4->3 [18] 5/-1/-1->4->3 [19] 5/-1/-1->4->3 [20] 5/-1/-1->4->3 [21] 5/-1/-1->4->3 [22] 5/-1/-1->4->3 [23] 5/-1/-1->4->3
+ip-26-0-169-207:162590:162785 [1] NCCL INFO Trees [0] 2/-1/-1->1->0 [1] 2/-1/-1->1->0 [2] 2/-1/-1->1->0 [3] 2/-1/-1->1->0 [4] 2/-1/-1->1->0 [5] 2/-1/-1->1->0 [6] 2/-1/-1->1->0 [7] 2/-1/-1->1->0 [8] 2/-1/-1->1->0 [9] 2/-1/-1->1->0 [10] 2/-1/-1->1->0 [11] 2/-1/-1->1->0 [12] 2/-1/-1->1->0 [13] 2/-1/-1->1->0 [14] 2/-1/-1->1->0 [15] 2/-1/-1->1->0 [16] 2/-1/-1->1->0 [17] 2/-1/-1->1->0 [18] 2/-1/-1->1->0 [19] 2/-1/-1->1->0 [20] 2/-1/-1->1->0 [21] 2/-1/-1->1->0 [22] 2/-1/-1->1->0 [23] 2/-1/-1->1->0
+ip-26-0-169-207:162592:162784 [3] NCCL INFO Trees [0] 4/-1/-1->3->2 [1] 4/-1/-1->3->2 [2] 4/-1/-1->3->2 [3] 4/-1/-1->3->2 [4] 4/-1/-1->3->2 [5] 4/-1/-1->3->2 [6] 4/-1/-1->3->2 [7] 4/-1/-1->3->2 [8] 4/-1/-1->3->2 [9] 4/-1/-1->3->2 [10] 4/-1/-1->3->2 [11] 4/-1/-1->3->2 [12] 4/-1/-1->3->2 [13] 4/-1/-1->3->2 [14] 4/-1/-1->3->2 [15] 4/-1/-1->3->2 [16] 4/-1/-1->3->2 [17] 4/-1/-1->3->2 [18] 4/-1/-1->3->2 [19] 4/-1/-1->3->2 [20] 4/-1/-1->3->2 [21] 4/-1/-1->3->2 [22] 4/-1/-1->3->2 [23] 4/-1/-1->3->2
+ip-26-0-169-207:162593:162786 [4] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-169-207:162592:162784 [3] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-169-207:162590:162785 [1] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-169-207:162595:162787 [6] NCCL INFO Channel 00/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:162596:162783 [7] NCCL INFO Channel 00/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162595:162787 [6] NCCL INFO Channel 01/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:162596:162783 [7] NCCL INFO Channel 01/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162595:162787 [6] NCCL INFO Channel 02/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:162596:162783 [7] NCCL INFO Channel 02/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162595:162787 [6] NCCL INFO Channel 03/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:162596:162783 [7] NCCL INFO Channel 03/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162595:162787 [6] NCCL INFO Channel 04/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:162596:162783 [7] NCCL INFO Channel 04/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162595:162787 [6] NCCL INFO Channel 05/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:162596:162783 [7] NCCL INFO Channel 05/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162595:162787 [6] NCCL INFO Channel 06/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:162596:162783 [7] NCCL INFO Channel 06/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162595:162787 [6] NCCL INFO Channel 07/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:162596:162783 [7] NCCL INFO Channel 07/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162595:162787 [6] NCCL INFO Channel 08/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:162589:162782 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162595:162787 [6] NCCL INFO Channel 09/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:162596:162783 [7] NCCL INFO Channel 08/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162591:162789 [2] NCCL INFO Channel 00/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162595:162787 [6] NCCL INFO Channel 10/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:162596:162783 [7] NCCL INFO Channel 09/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162591:162789 [2] NCCL INFO Channel 01/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162595:162787 [6] NCCL INFO Channel 11/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:162592:162784 [3] NCCL INFO Channel 00/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162596:162783 [7] NCCL INFO Channel 10/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162591:162789 [2] NCCL INFO Channel 02/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162592:162784 [3] NCCL INFO Channel 01/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162589:162782 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162592:162784 [3] NCCL INFO Channel 02/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162589:162782 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162592:162784 [3] NCCL INFO Channel 03/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162589:162782 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162590:162785 [1] NCCL INFO Channel 00/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162592:162784 [3] NCCL INFO Channel 04/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162589:162782 [0] NCCL INFO Channel 04/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162590:162785 [1] NCCL INFO Channel 01/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162592:162784 [3] NCCL INFO Channel 05/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162593:162786 [4] NCCL INFO Channel 00/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162589:162782 [0] NCCL INFO Channel 05/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162590:162785 [1] NCCL INFO Channel 02/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162593:162786 [4] NCCL INFO Channel 01/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162592:162784 [3] NCCL INFO Channel 06/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162589:162782 [0] NCCL INFO Channel 06/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162590:162785 [1] NCCL INFO Channel 03/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162595:162787 [6] NCCL INFO Channel 12/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:162593:162786 [4] NCCL INFO Channel 02/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162595:162787 [6] NCCL INFO Channel 13/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:162596:162783 [7] NCCL INFO Channel 11/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162591:162789 [2] NCCL INFO Channel 03/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162595:162787 [6] NCCL INFO Channel 14/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:162596:162783 [7] NCCL INFO Channel 12/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162591:162789 [2] NCCL INFO Channel 04/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162595:162787 [6] NCCL INFO Channel 15/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:162596:162783 [7] NCCL INFO Channel 13/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162591:162789 [2] NCCL INFO Channel 05/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162594:162788 [5] NCCL INFO Channel 00/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162595:162787 [6] NCCL INFO Channel 16/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:162596:162783 [7] NCCL INFO Channel 14/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162591:162789 [2] NCCL INFO Channel 06/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162594:162788 [5] NCCL INFO Channel 01/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162595:162787 [6] NCCL INFO Channel 17/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:162596:162783 [7] NCCL INFO Channel 15/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162591:162789 [2] NCCL INFO Channel 07/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162589:162782 [0] NCCL INFO Channel 07/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162596:162783 [7] NCCL INFO Channel 16/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162594:162788 [5] NCCL INFO Channel 02/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162595:162787 [6] NCCL INFO Channel 18/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:162591:162789 [2] NCCL INFO Channel 08/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162589:162782 [0] NCCL INFO Channel 08/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162594:162788 [5] NCCL INFO Channel 03/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162596:162783 [7] NCCL INFO Channel 17/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162595:162787 [6] NCCL INFO Channel 19/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:162592:162784 [3] NCCL INFO Channel 07/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162589:162782 [0] NCCL INFO Channel 09/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162591:162789 [2] NCCL INFO Channel 09/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162590:162785 [1] NCCL INFO Channel 04/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162594:162788 [5] NCCL INFO Channel 04/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162596:162783 [7] NCCL INFO Channel 18/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162591:162789 [2] NCCL INFO Channel 10/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162589:162782 [0] NCCL INFO Channel 10/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162590:162785 [1] NCCL INFO Channel 05/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162596:162783 [7] NCCL INFO Channel 19/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162591:162789 [2] NCCL INFO Channel 11/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162589:162782 [0] NCCL INFO Channel 11/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162590:162785 [1] NCCL INFO Channel 06/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162596:162783 [7] NCCL INFO Channel 20/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162591:162789 [2] NCCL INFO Channel 12/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162589:162782 [0] NCCL INFO Channel 12/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162590:162785 [1] NCCL INFO Channel 07/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162596:162783 [7] NCCL INFO Channel 21/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162591:162789 [2] NCCL INFO Channel 13/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162589:162782 [0] NCCL INFO Channel 13/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162590:162785 [1] NCCL INFO Channel 08/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162592:162784 [3] NCCL INFO Channel 08/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162596:162783 [7] NCCL INFO Channel 22/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162591:162789 [2] NCCL INFO Channel 14/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162590:162785 [1] NCCL INFO Channel 09/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162589:162782 [0] NCCL INFO Channel 14/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162592:162784 [3] NCCL INFO Channel 09/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162596:162783 [7] NCCL INFO Channel 23/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162591:162789 [2] NCCL INFO Channel 15/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162590:162785 [1] NCCL INFO Channel 10/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162589:162782 [0] NCCL INFO Channel 15/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162592:162784 [3] NCCL INFO Channel 10/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162591:162789 [2] NCCL INFO Channel 16/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162590:162785 [1] NCCL INFO Channel 11/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162589:162782 [0] NCCL INFO Channel 16/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162592:162784 [3] NCCL INFO Channel 11/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162593:162786 [4] NCCL INFO Channel 03/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162591:162789 [2] NCCL INFO Channel 17/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162595:162787 [6] NCCL INFO Channel 20/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:162590:162785 [1] NCCL INFO Channel 12/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162592:162784 [3] NCCL INFO Channel 12/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162589:162782 [0] NCCL INFO Channel 17/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162593:162786 [4] NCCL INFO Channel 04/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162591:162789 [2] NCCL INFO Channel 18/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162594:162788 [5] NCCL INFO Channel 05/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162590:162785 [1] NCCL INFO Channel 13/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162595:162787 [6] NCCL INFO Channel 21/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:162589:162782 [0] NCCL INFO Channel 18/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162593:162786 [4] NCCL INFO Channel 05/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162591:162789 [2] NCCL INFO Channel 19/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162592:162784 [3] NCCL INFO Channel 13/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162594:162788 [5] NCCL INFO Channel 06/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162590:162785 [1] NCCL INFO Channel 14/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162595:162787 [6] NCCL INFO Channel 22/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:162589:162782 [0] NCCL INFO Channel 19/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162593:162786 [4] NCCL INFO Channel 06/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162591:162789 [2] NCCL INFO Channel 20/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162592:162784 [3] NCCL INFO Channel 14/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162594:162788 [5] NCCL INFO Channel 07/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162595:162787 [6] NCCL INFO Channel 23/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:162590:162785 [1] NCCL INFO Channel 15/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162589:162782 [0] NCCL INFO Channel 20/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162593:162786 [4] NCCL INFO Channel 07/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162591:162789 [2] NCCL INFO Channel 21/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162594:162788 [5] NCCL INFO Channel 08/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162592:162784 [3] NCCL INFO Channel 15/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162590:162785 [1] NCCL INFO Channel 16/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162589:162782 [0] NCCL INFO Channel 21/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162593:162786 [4] NCCL INFO Channel 08/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162591:162789 [2] NCCL INFO Channel 22/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162594:162788 [5] NCCL INFO Channel 09/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162592:162784 [3] NCCL INFO Channel 16/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162589:162782 [0] NCCL INFO Channel 22/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162590:162785 [1] NCCL INFO Channel 17/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162593:162786 [4] NCCL INFO Channel 09/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162591:162789 [2] NCCL INFO Channel 23/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162594:162788 [5] NCCL INFO Channel 10/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162592:162784 [3] NCCL INFO Channel 17/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162589:162782 [0] NCCL INFO Channel 23/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162590:162785 [1] NCCL INFO Channel 18/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162593:162786 [4] NCCL INFO Channel 10/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162594:162788 [5] NCCL INFO Channel 11/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162592:162784 [3] NCCL INFO Channel 18/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162590:162785 [1] NCCL INFO Channel 19/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162593:162786 [4] NCCL INFO Channel 11/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162594:162788 [5] NCCL INFO Channel 12/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162592:162784 [3] NCCL INFO Channel 19/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162590:162785 [1] NCCL INFO Channel 20/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162593:162786 [4] NCCL INFO Channel 12/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162594:162788 [5] NCCL INFO Channel 13/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162592:162784 [3] NCCL INFO Channel 20/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162590:162785 [1] NCCL INFO Channel 21/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162593:162786 [4] NCCL INFO Channel 13/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162592:162784 [3] NCCL INFO Channel 21/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162594:162788 [5] NCCL INFO Channel 14/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162590:162785 [1] NCCL INFO Channel 22/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162593:162786 [4] NCCL INFO Channel 14/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162592:162784 [3] NCCL INFO Channel 22/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162594:162788 [5] NCCL INFO Channel 15/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162590:162785 [1] NCCL INFO Channel 23/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162593:162786 [4] NCCL INFO Channel 15/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162592:162784 [3] NCCL INFO Channel 23/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162594:162788 [5] NCCL INFO Channel 16/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162593:162786 [4] NCCL INFO Channel 16/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162594:162788 [5] NCCL INFO Channel 17/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162593:162786 [4] NCCL INFO Channel 17/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162593:162786 [4] NCCL INFO Channel 18/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162594:162788 [5] NCCL INFO Channel 18/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162594:162788 [5] NCCL INFO Channel 19/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162593:162786 [4] NCCL INFO Channel 19/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162594:162788 [5] NCCL INFO Channel 20/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162593:162786 [4] NCCL INFO Channel 20/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162594:162788 [5] NCCL INFO Channel 21/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162593:162786 [4] NCCL INFO Channel 21/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162594:162788 [5] NCCL INFO Channel 22/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162593:162786 [4] NCCL INFO Channel 22/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162594:162788 [5] NCCL INFO Channel 23/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162593:162786 [4] NCCL INFO Channel 23/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162596:162783 [7] NCCL INFO Connected all rings
+ip-26-0-169-207:162596:162783 [7] NCCL INFO Channel 00/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162596:162783 [7] NCCL INFO Channel 01/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162596:162783 [7] NCCL INFO Channel 02/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162596:162783 [7] NCCL INFO Channel 03/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162596:162783 [7] NCCL INFO Channel 04/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162591:162789 [2] NCCL INFO Connected all rings
+ip-26-0-169-207:162596:162783 [7] NCCL INFO Channel 05/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162596:162783 [7] NCCL INFO Channel 06/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162590:162785 [1] NCCL INFO Connected all rings
+ip-26-0-169-207:162589:162782 [0] NCCL INFO Connected all rings
+ip-26-0-169-207:162596:162783 [7] NCCL INFO Channel 07/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162596:162783 [7] NCCL INFO Channel 08/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162596:162783 [7] NCCL INFO Channel 09/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162596:162783 [7] NCCL INFO Channel 10/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162596:162783 [7] NCCL INFO Channel 11/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162596:162783 [7] NCCL INFO Channel 12/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162596:162783 [7] NCCL INFO Channel 13/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162592:162784 [3] NCCL INFO Connected all rings
+ip-26-0-169-207:162596:162783 [7] NCCL INFO Channel 14/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162596:162783 [7] NCCL INFO Channel 15/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162595:162787 [6] NCCL INFO Connected all rings
+ip-26-0-169-207:162596:162783 [7] NCCL INFO Channel 16/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162596:162783 [7] NCCL INFO Channel 17/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162594:162788 [5] NCCL INFO Connected all rings
+ip-26-0-169-207:162593:162786 [4] NCCL INFO Connected all rings
+ip-26-0-169-207:162596:162783 [7] NCCL INFO Channel 18/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162596:162783 [7] NCCL INFO Channel 19/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162596:162783 [7] NCCL INFO Channel 20/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162596:162783 [7] NCCL INFO Channel 21/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162596:162783 [7] NCCL INFO Channel 22/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162596:162783 [7] NCCL INFO Channel 23/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162591:162789 [2] NCCL INFO Channel 00/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162591:162789 [2] NCCL INFO Channel 01/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162591:162789 [2] NCCL INFO Channel 02/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162590:162785 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162591:162789 [2] NCCL INFO Channel 03/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162590:162785 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162591:162789 [2] NCCL INFO Channel 04/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162590:162785 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162590:162785 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162591:162789 [2] NCCL INFO Channel 05/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162590:162785 [1] NCCL INFO Channel 04/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162591:162789 [2] NCCL INFO Channel 06/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162590:162785 [1] NCCL INFO Channel 05/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162591:162789 [2] NCCL INFO Channel 07/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162590:162785 [1] NCCL INFO Channel 06/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162591:162789 [2] NCCL INFO Channel 08/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162590:162785 [1] NCCL INFO Channel 07/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162590:162785 [1] NCCL INFO Channel 08/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162591:162789 [2] NCCL INFO Channel 09/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162590:162785 [1] NCCL INFO Channel 09/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162591:162789 [2] NCCL INFO Channel 10/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162592:162784 [3] NCCL INFO Channel 00/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162590:162785 [1] NCCL INFO Channel 10/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162591:162789 [2] NCCL INFO Channel 11/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162592:162784 [3] NCCL INFO Channel 01/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162594:162788 [5] NCCL INFO Channel 00/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162590:162785 [1] NCCL INFO Channel 11/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162591:162789 [2] NCCL INFO Channel 12/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162593:162786 [4] NCCL INFO Channel 00/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162592:162784 [3] NCCL INFO Channel 02/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162595:162787 [6] NCCL INFO Channel 00/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162594:162788 [5] NCCL INFO Channel 01/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162590:162785 [1] NCCL INFO Channel 12/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162591:162789 [2] NCCL INFO Channel 13/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162593:162786 [4] NCCL INFO Channel 01/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162595:162787 [6] NCCL INFO Channel 01/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162592:162784 [3] NCCL INFO Channel 03/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162594:162788 [5] NCCL INFO Channel 02/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162590:162785 [1] NCCL INFO Channel 13/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162591:162789 [2] NCCL INFO Channel 14/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162593:162786 [4] NCCL INFO Channel 02/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162595:162787 [6] NCCL INFO Channel 02/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162594:162788 [5] NCCL INFO Channel 03/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162592:162784 [3] NCCL INFO Channel 04/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162591:162789 [2] NCCL INFO Channel 15/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162590:162785 [1] NCCL INFO Channel 14/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162593:162786 [4] NCCL INFO Channel 03/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162595:162787 [6] NCCL INFO Channel 03/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162592:162784 [3] NCCL INFO Channel 05/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162594:162788 [5] NCCL INFO Channel 04/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162591:162789 [2] NCCL INFO Channel 16/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162590:162785 [1] NCCL INFO Channel 15/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162593:162786 [4] NCCL INFO Channel 04/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162595:162787 [6] NCCL INFO Channel 04/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162592:162784 [3] NCCL INFO Channel 06/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162594:162788 [5] NCCL INFO Channel 05/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162591:162789 [2] NCCL INFO Channel 17/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162590:162785 [1] NCCL INFO Channel 16/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162593:162786 [4] NCCL INFO Channel 05/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162595:162787 [6] NCCL INFO Channel 05/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162592:162784 [3] NCCL INFO Channel 07/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162591:162789 [2] NCCL INFO Channel 18/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162590:162785 [1] NCCL INFO Channel 17/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162594:162788 [5] NCCL INFO Channel 06/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162593:162786 [4] NCCL INFO Channel 06/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162595:162787 [6] NCCL INFO Channel 06/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162592:162784 [3] NCCL INFO Channel 08/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162591:162789 [2] NCCL INFO Channel 19/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162590:162785 [1] NCCL INFO Channel 18/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162593:162786 [4] NCCL INFO Channel 07/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162594:162788 [5] NCCL INFO Channel 07/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162595:162787 [6] NCCL INFO Channel 07/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162592:162784 [3] NCCL INFO Channel 09/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162591:162789 [2] NCCL INFO Channel 20/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162590:162785 [1] NCCL INFO Channel 19/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162593:162786 [4] NCCL INFO Channel 08/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162594:162788 [5] NCCL INFO Channel 08/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162595:162787 [6] NCCL INFO Channel 08/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162592:162784 [3] NCCL INFO Channel 10/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162590:162785 [1] NCCL INFO Channel 20/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162591:162789 [2] NCCL INFO Channel 21/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162593:162786 [4] NCCL INFO Channel 09/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162594:162788 [5] NCCL INFO Channel 09/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162595:162787 [6] NCCL INFO Channel 09/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162592:162784 [3] NCCL INFO Channel 11/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162591:162789 [2] NCCL INFO Channel 22/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162590:162785 [1] NCCL INFO Channel 21/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162593:162786 [4] NCCL INFO Channel 10/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162594:162788 [5] NCCL INFO Channel 10/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162595:162787 [6] NCCL INFO Channel 10/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162592:162784 [3] NCCL INFO Channel 12/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162591:162789 [2] NCCL INFO Channel 23/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162590:162785 [1] NCCL INFO Channel 22/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162593:162786 [4] NCCL INFO Channel 11/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162594:162788 [5] NCCL INFO Channel 11/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162595:162787 [6] NCCL INFO Channel 11/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162592:162784 [3] NCCL INFO Channel 13/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162593:162786 [4] NCCL INFO Channel 12/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162590:162785 [1] NCCL INFO Channel 23/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162594:162788 [5] NCCL INFO Channel 12/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162595:162787 [6] NCCL INFO Channel 12/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162592:162784 [3] NCCL INFO Channel 14/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162593:162786 [4] NCCL INFO Channel 13/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162594:162788 [5] NCCL INFO Channel 13/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162595:162787 [6] NCCL INFO Channel 13/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162592:162784 [3] NCCL INFO Channel 15/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162593:162786 [4] NCCL INFO Channel 14/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162594:162788 [5] NCCL INFO Channel 14/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162595:162787 [6] NCCL INFO Channel 14/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162592:162784 [3] NCCL INFO Channel 16/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162593:162786 [4] NCCL INFO Channel 15/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162594:162788 [5] NCCL INFO Channel 15/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162595:162787 [6] NCCL INFO Channel 15/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162592:162784 [3] NCCL INFO Channel 17/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162593:162786 [4] NCCL INFO Channel 16/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162594:162788 [5] NCCL INFO Channel 16/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162595:162787 [6] NCCL INFO Channel 16/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162592:162784 [3] NCCL INFO Channel 18/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162593:162786 [4] NCCL INFO Channel 17/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162595:162787 [6] NCCL INFO Channel 17/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162593:162786 [4] NCCL INFO Channel 18/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162594:162788 [5] NCCL INFO Channel 17/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162595:162787 [6] NCCL INFO Channel 18/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162592:162784 [3] NCCL INFO Channel 19/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162594:162788 [5] NCCL INFO Channel 18/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162593:162786 [4] NCCL INFO Channel 19/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162594:162788 [5] NCCL INFO Channel 19/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162595:162787 [6] NCCL INFO Channel 19/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162592:162784 [3] NCCL INFO Channel 20/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162593:162786 [4] NCCL INFO Channel 20/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162594:162788 [5] NCCL INFO Channel 20/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162595:162787 [6] NCCL INFO Channel 20/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162592:162784 [3] NCCL INFO Channel 21/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162593:162786 [4] NCCL INFO Channel 21/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162594:162788 [5] NCCL INFO Channel 21/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162595:162787 [6] NCCL INFO Channel 21/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162593:162786 [4] NCCL INFO Channel 22/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162592:162784 [3] NCCL INFO Channel 22/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162594:162788 [5] NCCL INFO Channel 22/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162595:162787 [6] NCCL INFO Channel 22/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162593:162786 [4] NCCL INFO Channel 23/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162592:162784 [3] NCCL INFO Channel 23/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162594:162788 [5] NCCL INFO Channel 23/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162595:162787 [6] NCCL INFO Channel 23/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162589:162782 [0] NCCL INFO Connected all trees
+ip-26-0-169-207:162589:162782 [0] NCCL INFO NVLS comm 0x9538ce0 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-169-207:162596:162783 [7] NCCL INFO Connected all trees
+ip-26-0-169-207:162596:162783 [7] NCCL INFO NVLS comm 0x8934de0 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-169-207:162590:162785 [1] NCCL INFO Connected all trees
+ip-26-0-169-207:162590:162785 [1] NCCL INFO NVLS comm 0x8c71020 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-169-207:162594:162788 [5] NCCL INFO Connected all trees
+ip-26-0-169-207:162594:162788 [5] NCCL INFO NVLS comm 0x94df540 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-169-207:162591:162789 [2] NCCL INFO Connected all trees
+ip-26-0-169-207:162591:162789 [2] NCCL INFO NVLS comm 0x8ab1f00 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-169-207:162595:162787 [6] NCCL INFO Connected all trees
+ip-26-0-169-207:162595:162787 [6] NCCL INFO NVLS comm 0x9303c90 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-169-207:162593:162786 [4] NCCL INFO Connected all trees
+ip-26-0-169-207:162593:162786 [4] NCCL INFO NVLS comm 0x9e09520 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-169-207:162592:162784 [3] NCCL INFO Connected all trees
+ip-26-0-169-207:162592:162784 [3] NCCL INFO NVLS comm 0xa64e300 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-169-207:162591:162789 [2] NCCL INFO Connected NVLS tree
+ip-26-0-169-207:162593:162786 [4] NCCL INFO Connected NVLS tree
+ip-26-0-169-207:162591:162789 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-169-207:162591:162789 [2] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-169-207:162593:162786 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-169-207:162591:162789 [2] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:162593:162786 [4] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-169-207:162593:162786 [4] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:162596:162783 [7] NCCL INFO Connected NVLS tree
+ip-26-0-169-207:162596:162783 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-169-207:162596:162783 [7] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-169-207:162596:162783 [7] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:162592:162784 [3] NCCL INFO Connected NVLS tree
+ip-26-0-169-207:162595:162787 [6] NCCL INFO Connected NVLS tree
+ip-26-0-169-207:162592:162784 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-169-207:162595:162787 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-169-207:162592:162784 [3] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-169-207:162592:162784 [3] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:162595:162787 [6] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-169-207:162595:162787 [6] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:162589:162782 [0] NCCL INFO Connected NVLS tree
+ip-26-0-169-207:162589:162782 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-169-207:162589:162782 [0] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-169-207:162589:162782 [0] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:162590:162785 [1] NCCL INFO Connected NVLS tree
+ip-26-0-169-207:162590:162785 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-169-207:162590:162785 [1] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-169-207:162590:162785 [1] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:162594:162788 [5] NCCL INFO Connected NVLS tree
+ip-26-0-169-207:162594:162788 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-169-207:162594:162788 [5] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-169-207:162594:162788 [5] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:162595:162787 [6] NCCL INFO comm 0x9303c90 rank 6 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0xe3a983ad389222f8 - Init COMPLETE
+ip-26-0-169-207:162591:162789 [2] NCCL INFO comm 0x8ab1f00 rank 2 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0xe3a983ad389222f8 - Init COMPLETE
+ip-26-0-169-207:162593:162786 [4] NCCL INFO comm 0x9e09520 rank 4 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0xe3a983ad389222f8 - Init COMPLETE
+ip-26-0-169-207:162589:162782 [0] NCCL INFO comm 0x9538ce0 rank 0 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0xe3a983ad389222f8 - Init COMPLETE
+ip-26-0-169-207:162596:162783 [7] NCCL INFO comm 0x8934de0 rank 7 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0xe3a983ad389222f8 - Init COMPLETE
+ip-26-0-169-207:162592:162784 [3] NCCL INFO comm 0xa64e300 rank 3 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0xe3a983ad389222f8 - Init COMPLETE
+ip-26-0-169-207:162590:162785 [1] NCCL INFO comm 0x8c71020 rank 1 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0xe3a983ad389222f8 - Init COMPLETE
+ip-26-0-169-207:162594:162788 [5] NCCL INFO comm 0x94df540 rank 5 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0xe3a983ad389222f8 - Init COMPLETE
+ip-26-0-169-207:162589:162815 [0] NCCL INFO Using network Libfabric
+ip-26-0-169-207:162590:162816 [1] NCCL INFO Using network Libfabric
+ip-26-0-169-207:162595:162819 [6] NCCL INFO Using network Libfabric
+ip-26-0-169-207:162596:162817 [7] NCCL INFO Using network Libfabric
+ip-26-0-169-207:162592:162820 [3] NCCL INFO Using network Libfabric
+ip-26-0-169-207:162593:162822 [4] NCCL INFO Using network Libfabric
+ip-26-0-169-207:162591:162821 [2] NCCL INFO Using network Libfabric
+ip-26-0-169-207:162594:162818 [5] NCCL INFO Using network Libfabric
+ip-26-0-169-207:162590:162816 [1] NCCL INFO comm 0x8c85770 rank 1 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0x87e3927732a5fde8 - Init START
+ip-26-0-169-207:162589:162815 [0] NCCL INFO comm 0x954d000 rank 0 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0x87e3927732a5fde8 - Init START
+ip-26-0-169-207:162591:162821 [2] NCCL INFO comm 0x8ac6650 rank 2 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0x87e3927732a5fde8 - Init START
+ip-26-0-169-207:162596:162817 [7] NCCL INFO comm 0x8949320 rank 7 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0x87e3927732a5fde8 - Init START
+ip-26-0-169-207:162593:162822 [4] NCCL INFO comm 0x9e1dcb0 rank 4 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0x87e3927732a5fde8 - Init START
+ip-26-0-169-207:162595:162819 [6] NCCL INFO comm 0x9318400 rank 6 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0x87e3927732a5fde8 - Init START
+ip-26-0-169-207:162592:162820 [3] NCCL INFO comm 0xa6628e0 rank 3 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0x87e3927732a5fde8 - Init START
+ip-26-0-169-207:162594:162818 [5] NCCL INFO comm 0x94f3ae0 rank 5 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0x87e3927732a5fde8 - Init START
+ip-26-0-169-207:162590:162816 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/149
+ip-26-0-169-207:162591:162821 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/149
+ip-26-0-169-207:162592:162820 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/149
+ip-26-0-169-207:162596:162817 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/149
+ip-26-0-169-207:162593:162822 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/149
+ip-26-0-169-207:162594:162818 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/149
+ip-26-0-169-207:162595:162819 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/149
+ip-26-0-169-207:162589:162815 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/162
+ip-26-0-169-207:162589:162815 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-169-207:162589:162815 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-169-207:162596:162817 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-169-207:162596:162817 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-169-207:162593:162822 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-169-207:162593:162822 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-169-207:162591:162821 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-169-207:162591:162821 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-169-207:162592:162820 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-169-207:162592:162820 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-169-207:162595:162819 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-169-207:162595:162819 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-169-207:162590:162816 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-169-207:162590:162816 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-169-207:162594:162818 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-169-207:162594:162818 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-169-207:162595:162819 [6] NCCL INFO Trees [0] 7/-1/-1->6->5 [1] 7/-1/-1->6->5 [2] 7/-1/-1->6->5 [3] 7/-1/-1->6->5 [4] 7/-1/-1->6->5 [5] 7/-1/-1->6->5 [6] 7/-1/-1->6->5 [7] 7/-1/-1->6->5 [8] 7/-1/-1->6->5 [9] 7/-1/-1->6->5 [10] 7/-1/-1->6->5 [11] 7/-1/-1->6->5 [12] 7/-1/-1->6->5 [13] 7/-1/-1->6->5 [14] 7/-1/-1->6->5 [15] 7/-1/-1->6->5 [16] 7/-1/-1->6->5 [17] 7/-1/-1->6->5 [18] 7/-1/-1->6->5 [19] 7/-1/-1->6->5 [20] 7/-1/-1->6->5 [21] 7/-1/-1->6->5 [22] 7/-1/-1->6->5 [23] 7/-1/-1->6->5
+ip-26-0-169-207:162594:162818 [5] NCCL INFO Trees [0] 6/-1/-1->5->4 [1] 6/-1/-1->5->4 [2] 6/-1/-1->5->4 [3] 6/-1/-1->5->4 [4] 6/-1/-1->5->4 [5] 6/-1/-1->5->4 [6] 6/-1/-1->5->4 [7] 6/-1/-1->5->4 [8] 6/-1/-1->5->4 [9] 6/-1/-1->5->4 [10] 6/-1/-1->5->4 [11] 6/-1/-1->5->4 [12] 6/-1/-1->5->4 [13] 6/-1/-1->5->4 [14] 6/-1/-1->5->4 [15] 6/-1/-1->5->4 [16] 6/-1/-1->5->4 [17] 6/-1/-1->5->4 [18] 6/-1/-1->5->4 [19] 6/-1/-1->5->4 [20] 6/-1/-1->5->4 [21] 6/-1/-1->5->4 [22] 6/-1/-1->5->4 [23] 6/-1/-1->5->4
+ip-26-0-169-207:162591:162821 [2] NCCL INFO Trees [0] 3/-1/-1->2->1 [1] 3/-1/-1->2->1 [2] 3/-1/-1->2->1 [3] 3/-1/-1->2->1 [4] 3/-1/-1->2->1 [5] 3/-1/-1->2->1 [6] 3/-1/-1->2->1 [7] 3/-1/-1->2->1 [8] 3/-1/-1->2->1 [9] 3/-1/-1->2->1 [10] 3/-1/-1->2->1 [11] 3/-1/-1->2->1 [12] 3/-1/-1->2->1 [13] 3/-1/-1->2->1 [14] 3/-1/-1->2->1 [15] 3/-1/-1->2->1 [16] 3/-1/-1->2->1 [17] 3/-1/-1->2->1 [18] 3/-1/-1->2->1 [19] 3/-1/-1->2->1 [20] 3/-1/-1->2->1 [21] 3/-1/-1->2->1 [22] 3/-1/-1->2->1 [23] 3/-1/-1->2->1
+ip-26-0-169-207:162593:162822 [4] NCCL INFO Trees [0] 5/-1/-1->4->3 [1] 5/-1/-1->4->3 [2] 5/-1/-1->4->3 [3] 5/-1/-1->4->3 [4] 5/-1/-1->4->3 [5] 5/-1/-1->4->3 [6] 5/-1/-1->4->3 [7] 5/-1/-1->4->3 [8] 5/-1/-1->4->3 [9] 5/-1/-1->4->3 [10] 5/-1/-1->4->3 [11] 5/-1/-1->4->3 [12] 5/-1/-1->4->3 [13] 5/-1/-1->4->3 [14] 5/-1/-1->4->3 [15] 5/-1/-1->4->3 [16] 5/-1/-1->4->3 [17] 5/-1/-1->4->3 [18] 5/-1/-1->4->3 [19] 5/-1/-1->4->3 [20] 5/-1/-1->4->3 [21] 5/-1/-1->4->3 [22] 5/-1/-1->4->3 [23] 5/-1/-1->4->3
+ip-26-0-169-207:162590:162816 [1] NCCL INFO Trees [0] 2/-1/-1->1->0 [1] 2/-1/-1->1->0 [2] 2/-1/-1->1->0 [3] 2/-1/-1->1->0 [4] 2/-1/-1->1->0 [5] 2/-1/-1->1->0 [6] 2/-1/-1->1->0 [7] 2/-1/-1->1->0 [8] 2/-1/-1->1->0 [9] 2/-1/-1->1->0 [10] 2/-1/-1->1->0 [11] 2/-1/-1->1->0 [12] 2/-1/-1->1->0 [13] 2/-1/-1->1->0 [14] 2/-1/-1->1->0 [15] 2/-1/-1->1->0 [16] 2/-1/-1->1->0 [17] 2/-1/-1->1->0 [18] 2/-1/-1->1->0 [19] 2/-1/-1->1->0 [20] 2/-1/-1->1->0 [21] 2/-1/-1->1->0 [22] 2/-1/-1->1->0 [23] 2/-1/-1->1->0
+ip-26-0-169-207:162595:162819 [6] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-169-207:162589:162815 [0] NCCL INFO Channel 00/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:162594:162818 [5] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-169-207:162593:162822 [4] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-169-207:162590:162816 [1] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-169-207:162592:162820 [3] NCCL INFO Trees [0] 4/-1/-1->3->2 [1] 4/-1/-1->3->2 [2] 4/-1/-1->3->2 [3] 4/-1/-1->3->2 [4] 4/-1/-1->3->2 [5] 4/-1/-1->3->2 [6] 4/-1/-1->3->2 [7] 4/-1/-1->3->2 [8] 4/-1/-1->3->2 [9] 4/-1/-1->3->2 [10] 4/-1/-1->3->2 [11] 4/-1/-1->3->2 [12] 4/-1/-1->3->2 [13] 4/-1/-1->3->2 [14] 4/-1/-1->3->2 [15] 4/-1/-1->3->2 [16] 4/-1/-1->3->2 [17] 4/-1/-1->3->2 [18] 4/-1/-1->3->2 [19] 4/-1/-1->3->2 [20] 4/-1/-1->3->2 [21] 4/-1/-1->3->2 [22] 4/-1/-1->3->2 [23] 4/-1/-1->3->2
+ip-26-0-169-207:162596:162817 [7] NCCL INFO Trees [0] -1/-1/-1->7->6 [1] -1/-1/-1->7->6 [2] -1/-1/-1->7->6 [3] -1/-1/-1->7->6 [4] -1/-1/-1->7->6 [5] -1/-1/-1->7->6 [6] -1/-1/-1->7->6 [7] -1/-1/-1->7->6 [8] -1/-1/-1->7->6 [9] -1/-1/-1->7->6 [10] -1/-1/-1->7->6 [11] -1/-1/-1->7->6 [12] -1/-1/-1->7->6 [13] -1/-1/-1->7->6 [14] -1/-1/-1->7->6 [15] -1/-1/-1->7->6 [16] -1/-1/-1->7->6 [17] -1/-1/-1->7->6 [18] -1/-1/-1->7->6 [19] -1/-1/-1->7->6 [20] -1/-1/-1->7->6 [21] -1/-1/-1->7->6 [22] -1/-1/-1->7->6 [23] -1/-1/-1->7->6
+ip-26-0-169-207:162591:162821 [2] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-169-207:162589:162815 [0] NCCL INFO Channel 01/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:162592:162820 [3] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-169-207:162589:162815 [0] NCCL INFO Channel 02/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:162589:162815 [0] NCCL INFO Channel 03/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:162596:162817 [7] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-169-207:162589:162815 [0] NCCL INFO Channel 04/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:162589:162815 [0] NCCL INFO Channel 05/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:162589:162815 [0] NCCL INFO Channel 06/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:162589:162815 [0] NCCL INFO Channel 07/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:162589:162815 [0] NCCL INFO Channel 08/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:162589:162815 [0] NCCL INFO Channel 09/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:162589:162815 [0] NCCL INFO Channel 10/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:162589:162815 [0] NCCL INFO Channel 11/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:162589:162815 [0] NCCL INFO Channel 12/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:162589:162815 [0] NCCL INFO Channel 13/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:162589:162815 [0] NCCL INFO Channel 14/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:162589:162815 [0] NCCL INFO Channel 15/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:162589:162815 [0] NCCL INFO Channel 16/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:162589:162815 [0] NCCL INFO Channel 17/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:162589:162815 [0] NCCL INFO Channel 18/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:162589:162815 [0] NCCL INFO Channel 19/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:162589:162815 [0] NCCL INFO Channel 20/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:162589:162815 [0] NCCL INFO Channel 21/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:162589:162815 [0] NCCL INFO Channel 22/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:162589:162815 [0] NCCL INFO Channel 23/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:162589:162815 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] 1/-1/-1->0->-1 [7] 1/-1/-1->0->-1 [8] 1/-1/-1->0->-1 [9] 1/-1/-1->0->-1 [10] 1/-1/-1->0->-1 [11] 1/-1/-1->0->-1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] 1/-1/-1->0->-1 [19] 1/-1/-1->0->-1 [20] 1/-1/-1->0->-1 [21] 1/-1/-1->0->-1 [22] 1/-1/-1->0->-1 [23] 1/-1/-1->0->-1
+ip-26-0-169-207:162589:162815 [0] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-169-207:162595:162819 [6] NCCL INFO Channel 00/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:162593:162822 [4] NCCL INFO Channel 00/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162596:162817 [7] NCCL INFO Channel 00/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162594:162818 [5] NCCL INFO Channel 00/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162595:162819 [6] NCCL INFO Channel 01/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:162593:162822 [4] NCCL INFO Channel 01/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162596:162817 [7] NCCL INFO Channel 01/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162594:162818 [5] NCCL INFO Channel 01/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162595:162819 [6] NCCL INFO Channel 02/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:162593:162822 [4] NCCL INFO Channel 02/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162596:162817 [7] NCCL INFO Channel 02/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162594:162818 [5] NCCL INFO Channel 02/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162590:162816 [1] NCCL INFO Channel 00/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162595:162819 [6] NCCL INFO Channel 03/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:162593:162822 [4] NCCL INFO Channel 03/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162596:162817 [7] NCCL INFO Channel 03/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162594:162818 [5] NCCL INFO Channel 03/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162590:162816 [1] NCCL INFO Channel 01/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162595:162819 [6] NCCL INFO Channel 04/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:162594:162818 [5] NCCL INFO Channel 04/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162594:162818 [5] NCCL INFO Channel 05/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162596:162817 [7] NCCL INFO Channel 04/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162589:162815 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162594:162818 [5] NCCL INFO Channel 06/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162596:162817 [7] NCCL INFO Channel 05/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162591:162821 [2] NCCL INFO Channel 00/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162593:162822 [4] NCCL INFO Channel 04/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162594:162818 [5] NCCL INFO Channel 07/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162596:162817 [7] NCCL INFO Channel 06/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162591:162821 [2] NCCL INFO Channel 01/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162593:162822 [4] NCCL INFO Channel 05/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162594:162818 [5] NCCL INFO Channel 08/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162590:162816 [1] NCCL INFO Channel 02/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162595:162819 [6] NCCL INFO Channel 05/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:162594:162818 [5] NCCL INFO Channel 09/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162590:162816 [1] NCCL INFO Channel 03/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162595:162819 [6] NCCL INFO Channel 06/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:162594:162818 [5] NCCL INFO Channel 10/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162590:162816 [1] NCCL INFO Channel 04/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162592:162820 [3] NCCL INFO Channel 00/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162595:162819 [6] NCCL INFO Channel 07/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:162594:162818 [5] NCCL INFO Channel 11/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162590:162816 [1] NCCL INFO Channel 05/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162592:162820 [3] NCCL INFO Channel 01/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162595:162819 [6] NCCL INFO Channel 08/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:162594:162818 [5] NCCL INFO Channel 12/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162589:162815 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162590:162816 [1] NCCL INFO Channel 06/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162592:162820 [3] NCCL INFO Channel 02/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162596:162817 [7] NCCL INFO Channel 07/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162591:162821 [2] NCCL INFO Channel 02/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162595:162819 [6] NCCL INFO Channel 09/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:162593:162822 [4] NCCL INFO Channel 06/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162594:162818 [5] NCCL INFO Channel 13/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162589:162815 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162590:162816 [1] NCCL INFO Channel 07/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162592:162820 [3] NCCL INFO Channel 03/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162596:162817 [7] NCCL INFO Channel 08/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162591:162821 [2] NCCL INFO Channel 03/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162595:162819 [6] NCCL INFO Channel 10/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:162593:162822 [4] NCCL INFO Channel 07/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162594:162818 [5] NCCL INFO Channel 14/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162589:162815 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162590:162816 [1] NCCL INFO Channel 08/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162592:162820 [3] NCCL INFO Channel 04/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162596:162817 [7] NCCL INFO Channel 09/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162591:162821 [2] NCCL INFO Channel 04/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162595:162819 [6] NCCL INFO Channel 11/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:162593:162822 [4] NCCL INFO Channel 08/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162594:162818 [5] NCCL INFO Channel 15/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162589:162815 [0] NCCL INFO Channel 04/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162590:162816 [1] NCCL INFO Channel 09/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162592:162820 [3] NCCL INFO Channel 05/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162596:162817 [7] NCCL INFO Channel 10/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162591:162821 [2] NCCL INFO Channel 05/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162595:162819 [6] NCCL INFO Channel 12/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:162593:162822 [4] NCCL INFO Channel 09/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162594:162818 [5] NCCL INFO Channel 16/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162589:162815 [0] NCCL INFO Channel 05/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162590:162816 [1] NCCL INFO Channel 10/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162592:162820 [3] NCCL INFO Channel 06/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162596:162817 [7] NCCL INFO Channel 11/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162591:162821 [2] NCCL INFO Channel 06/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162595:162819 [6] NCCL INFO Channel 13/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:162594:162818 [5] NCCL INFO Channel 17/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162593:162822 [4] NCCL INFO Channel 10/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162589:162815 [0] NCCL INFO Channel 06/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162590:162816 [1] NCCL INFO Channel 11/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162592:162820 [3] NCCL INFO Channel 07/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162596:162817 [7] NCCL INFO Channel 12/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162591:162821 [2] NCCL INFO Channel 07/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162595:162819 [6] NCCL INFO Channel 14/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:162594:162818 [5] NCCL INFO Channel 18/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162593:162822 [4] NCCL INFO Channel 11/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162589:162815 [0] NCCL INFO Channel 07/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162592:162820 [3] NCCL INFO Channel 08/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162590:162816 [1] NCCL INFO Channel 12/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162596:162817 [7] NCCL INFO Channel 13/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162591:162821 [2] NCCL INFO Channel 08/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162595:162819 [6] NCCL INFO Channel 15/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:162594:162818 [5] NCCL INFO Channel 19/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162593:162822 [4] NCCL INFO Channel 12/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162589:162815 [0] NCCL INFO Channel 08/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162592:162820 [3] NCCL INFO Channel 09/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162590:162816 [1] NCCL INFO Channel 13/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162596:162817 [7] NCCL INFO Channel 14/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162595:162819 [6] NCCL INFO Channel 16/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:162593:162822 [4] NCCL INFO Channel 13/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162592:162820 [3] NCCL INFO Channel 10/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162595:162819 [6] NCCL INFO Channel 17/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:162589:162815 [0] NCCL INFO Channel 09/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162593:162822 [4] NCCL INFO Channel 14/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162591:162821 [2] NCCL INFO Channel 09/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162594:162818 [5] NCCL INFO Channel 20/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162590:162816 [1] NCCL INFO Channel 14/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162591:162821 [2] NCCL INFO Channel 10/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162594:162818 [5] NCCL INFO Channel 21/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162596:162817 [7] NCCL INFO Channel 15/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162590:162816 [1] NCCL INFO Channel 15/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162591:162821 [2] NCCL INFO Channel 11/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162592:162820 [3] NCCL INFO Channel 11/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162594:162818 [5] NCCL INFO Channel 22/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162596:162817 [7] NCCL INFO Channel 16/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162590:162816 [1] NCCL INFO Channel 16/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162591:162821 [2] NCCL INFO Channel 12/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162592:162820 [3] NCCL INFO Channel 12/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162594:162818 [5] NCCL INFO Channel 23/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162596:162817 [7] NCCL INFO Channel 17/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162590:162816 [1] NCCL INFO Channel 17/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162591:162821 [2] NCCL INFO Channel 13/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162592:162820 [3] NCCL INFO Channel 13/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162593:162822 [4] NCCL INFO Channel 15/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162596:162817 [7] NCCL INFO Channel 18/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162590:162816 [1] NCCL INFO Channel 18/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162591:162821 [2] NCCL INFO Channel 14/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162592:162820 [3] NCCL INFO Channel 14/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162595:162819 [6] NCCL INFO Channel 18/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:162593:162822 [4] NCCL INFO Channel 16/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162596:162817 [7] NCCL INFO Channel 19/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162590:162816 [1] NCCL INFO Channel 19/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162591:162821 [2] NCCL INFO Channel 15/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162592:162820 [3] NCCL INFO Channel 15/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162589:162815 [0] NCCL INFO Channel 10/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162595:162819 [6] NCCL INFO Channel 19/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:162593:162822 [4] NCCL INFO Channel 17/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162596:162817 [7] NCCL INFO Channel 20/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162590:162816 [1] NCCL INFO Channel 20/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162591:162821 [2] NCCL INFO Channel 16/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162592:162820 [3] NCCL INFO Channel 16/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162589:162815 [0] NCCL INFO Channel 11/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162595:162819 [6] NCCL INFO Channel 20/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:162593:162822 [4] NCCL INFO Channel 18/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162596:162817 [7] NCCL INFO Channel 21/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162591:162821 [2] NCCL INFO Channel 17/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162590:162816 [1] NCCL INFO Channel 21/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162592:162820 [3] NCCL INFO Channel 17/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162589:162815 [0] NCCL INFO Channel 12/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162595:162819 [6] NCCL INFO Channel 21/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:162593:162822 [4] NCCL INFO Channel 19/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162596:162817 [7] NCCL INFO Channel 22/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162591:162821 [2] NCCL INFO Channel 18/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162592:162820 [3] NCCL INFO Channel 18/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162590:162816 [1] NCCL INFO Channel 22/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162589:162815 [0] NCCL INFO Channel 13/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162595:162819 [6] NCCL INFO Channel 22/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:162593:162822 [4] NCCL INFO Channel 20/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162596:162817 [7] NCCL INFO Channel 23/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162591:162821 [2] NCCL INFO Channel 19/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162592:162820 [3] NCCL INFO Channel 19/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162590:162816 [1] NCCL INFO Channel 23/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162589:162815 [0] NCCL INFO Channel 14/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162595:162819 [6] NCCL INFO Channel 23/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:162593:162822 [4] NCCL INFO Channel 21/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162591:162821 [2] NCCL INFO Channel 20/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162589:162815 [0] NCCL INFO Channel 15/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162592:162820 [3] NCCL INFO Channel 20/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162593:162822 [4] NCCL INFO Channel 22/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162591:162821 [2] NCCL INFO Channel 21/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162592:162820 [3] NCCL INFO Channel 21/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162589:162815 [0] NCCL INFO Channel 16/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162593:162822 [4] NCCL INFO Channel 23/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162591:162821 [2] NCCL INFO Channel 22/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162592:162820 [3] NCCL INFO Channel 22/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162589:162815 [0] NCCL INFO Channel 17/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162591:162821 [2] NCCL INFO Channel 23/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162592:162820 [3] NCCL INFO Channel 23/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162589:162815 [0] NCCL INFO Channel 18/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162589:162815 [0] NCCL INFO Channel 19/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162589:162815 [0] NCCL INFO Channel 20/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162589:162815 [0] NCCL INFO Channel 21/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162589:162815 [0] NCCL INFO Channel 22/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162589:162815 [0] NCCL INFO Channel 23/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162595:162819 [6] NCCL INFO Connected all rings
+ip-26-0-169-207:162594:162818 [5] NCCL INFO Connected all rings
+ip-26-0-169-207:162592:162820 [3] NCCL INFO Connected all rings
+ip-26-0-169-207:162593:162822 [4] NCCL INFO Connected all rings
+ip-26-0-169-207:162591:162821 [2] NCCL INFO Connected all rings
+ip-26-0-169-207:162589:162815 [0] NCCL INFO Connected all rings
+ip-26-0-169-207:162590:162816 [1] NCCL INFO Connected all rings
+ip-26-0-169-207:162596:162817 [7] NCCL INFO Connected all rings
+ip-26-0-169-207:162596:162817 [7] NCCL INFO Channel 00/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162596:162817 [7] NCCL INFO Channel 01/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162596:162817 [7] NCCL INFO Channel 02/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162596:162817 [7] NCCL INFO Channel 03/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162596:162817 [7] NCCL INFO Channel 04/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162593:162822 [4] NCCL INFO Channel 00/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162594:162818 [5] NCCL INFO Channel 00/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162593:162822 [4] NCCL INFO Channel 01/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162594:162818 [5] NCCL INFO Channel 01/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162593:162822 [4] NCCL INFO Channel 02/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162596:162817 [7] NCCL INFO Channel 05/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162594:162818 [5] NCCL INFO Channel 02/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162593:162822 [4] NCCL INFO Channel 03/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162596:162817 [7] NCCL INFO Channel 06/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162591:162821 [2] NCCL INFO Channel 00/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162594:162818 [5] NCCL INFO Channel 03/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162593:162822 [4] NCCL INFO Channel 04/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162596:162817 [7] NCCL INFO Channel 07/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162591:162821 [2] NCCL INFO Channel 01/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162594:162818 [5] NCCL INFO Channel 04/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162593:162822 [4] NCCL INFO Channel 05/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162595:162819 [6] NCCL INFO Channel 00/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162596:162817 [7] NCCL INFO Channel 08/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162591:162821 [2] NCCL INFO Channel 02/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162594:162818 [5] NCCL INFO Channel 05/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162593:162822 [4] NCCL INFO Channel 06/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162595:162819 [6] NCCL INFO Channel 01/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162596:162817 [7] NCCL INFO Channel 09/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162591:162821 [2] NCCL INFO Channel 03/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162594:162818 [5] NCCL INFO Channel 06/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162593:162822 [4] NCCL INFO Channel 07/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162595:162819 [6] NCCL INFO Channel 02/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162596:162817 [7] NCCL INFO Channel 10/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162591:162821 [2] NCCL INFO Channel 04/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162594:162818 [5] NCCL INFO Channel 07/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162593:162822 [4] NCCL INFO Channel 08/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162595:162819 [6] NCCL INFO Channel 03/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162591:162821 [2] NCCL INFO Channel 05/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162596:162817 [7] NCCL INFO Channel 11/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162594:162818 [5] NCCL INFO Channel 08/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162592:162820 [3] NCCL INFO Channel 00/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162593:162822 [4] NCCL INFO Channel 09/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162595:162819 [6] NCCL INFO Channel 04/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162591:162821 [2] NCCL INFO Channel 06/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162596:162817 [7] NCCL INFO Channel 12/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162594:162818 [5] NCCL INFO Channel 09/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162592:162820 [3] NCCL INFO Channel 01/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162593:162822 [4] NCCL INFO Channel 10/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162595:162819 [6] NCCL INFO Channel 05/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162591:162821 [2] NCCL INFO Channel 07/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162596:162817 [7] NCCL INFO Channel 13/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162594:162818 [5] NCCL INFO Channel 10/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162592:162820 [3] NCCL INFO Channel 02/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162593:162822 [4] NCCL INFO Channel 11/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162595:162819 [6] NCCL INFO Channel 06/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162591:162821 [2] NCCL INFO Channel 08/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162596:162817 [7] NCCL INFO Channel 14/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162594:162818 [5] NCCL INFO Channel 11/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162592:162820 [3] NCCL INFO Channel 03/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162593:162822 [4] NCCL INFO Channel 12/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162595:162819 [6] NCCL INFO Channel 07/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162591:162821 [2] NCCL INFO Channel 09/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162596:162817 [7] NCCL INFO Channel 15/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162594:162818 [5] NCCL INFO Channel 12/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162592:162820 [3] NCCL INFO Channel 04/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162593:162822 [4] NCCL INFO Channel 13/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162595:162819 [6] NCCL INFO Channel 08/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162591:162821 [2] NCCL INFO Channel 10/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162596:162817 [7] NCCL INFO Channel 16/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162594:162818 [5] NCCL INFO Channel 13/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162592:162820 [3] NCCL INFO Channel 05/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162593:162822 [4] NCCL INFO Channel 14/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162595:162819 [6] NCCL INFO Channel 09/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162591:162821 [2] NCCL INFO Channel 11/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162596:162817 [7] NCCL INFO Channel 17/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162594:162818 [5] NCCL INFO Channel 14/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162592:162820 [3] NCCL INFO Channel 06/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162593:162822 [4] NCCL INFO Channel 15/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162595:162819 [6] NCCL INFO Channel 10/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162591:162821 [2] NCCL INFO Channel 12/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162596:162817 [7] NCCL INFO Channel 18/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162594:162818 [5] NCCL INFO Channel 15/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162592:162820 [3] NCCL INFO Channel 07/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162593:162822 [4] NCCL INFO Channel 16/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162595:162819 [6] NCCL INFO Channel 11/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162591:162821 [2] NCCL INFO Channel 13/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162596:162817 [7] NCCL INFO Channel 19/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162592:162820 [3] NCCL INFO Channel 08/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162594:162818 [5] NCCL INFO Channel 16/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162590:162816 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162593:162822 [4] NCCL INFO Channel 17/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162595:162819 [6] NCCL INFO Channel 12/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162591:162821 [2] NCCL INFO Channel 14/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162596:162817 [7] NCCL INFO Channel 20/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162592:162820 [3] NCCL INFO Channel 09/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162594:162818 [5] NCCL INFO Channel 17/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162590:162816 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162593:162822 [4] NCCL INFO Channel 18/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162595:162819 [6] NCCL INFO Channel 13/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162591:162821 [2] NCCL INFO Channel 15/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162596:162817 [7] NCCL INFO Channel 21/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162592:162820 [3] NCCL INFO Channel 10/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162594:162818 [5] NCCL INFO Channel 18/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162590:162816 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162593:162822 [4] NCCL INFO Channel 19/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162595:162819 [6] NCCL INFO Channel 14/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162591:162821 [2] NCCL INFO Channel 16/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162596:162817 [7] NCCL INFO Channel 22/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162592:162820 [3] NCCL INFO Channel 11/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162594:162818 [5] NCCL INFO Channel 19/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162590:162816 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162593:162822 [4] NCCL INFO Channel 20/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162595:162819 [6] NCCL INFO Channel 15/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162591:162821 [2] NCCL INFO Channel 17/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162596:162817 [7] NCCL INFO Channel 23/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:162592:162820 [3] NCCL INFO Channel 12/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162594:162818 [5] NCCL INFO Channel 20/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162590:162816 [1] NCCL INFO Channel 04/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162593:162822 [4] NCCL INFO Channel 21/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162595:162819 [6] NCCL INFO Channel 16/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162591:162821 [2] NCCL INFO Channel 18/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162592:162820 [3] NCCL INFO Channel 13/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162594:162818 [5] NCCL INFO Channel 21/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162590:162816 [1] NCCL INFO Channel 05/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162593:162822 [4] NCCL INFO Channel 22/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162595:162819 [6] NCCL INFO Channel 17/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162591:162821 [2] NCCL INFO Channel 19/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162592:162820 [3] NCCL INFO Channel 14/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162594:162818 [5] NCCL INFO Channel 22/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162590:162816 [1] NCCL INFO Channel 06/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162593:162822 [4] NCCL INFO Channel 23/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:162595:162819 [6] NCCL INFO Channel 18/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162591:162821 [2] NCCL INFO Channel 20/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162592:162820 [3] NCCL INFO Channel 15/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162594:162818 [5] NCCL INFO Channel 23/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:162590:162816 [1] NCCL INFO Channel 07/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162595:162819 [6] NCCL INFO Channel 19/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162591:162821 [2] NCCL INFO Channel 21/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162592:162820 [3] NCCL INFO Channel 16/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162590:162816 [1] NCCL INFO Channel 08/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162595:162819 [6] NCCL INFO Channel 20/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162591:162821 [2] NCCL INFO Channel 22/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162592:162820 [3] NCCL INFO Channel 17/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162590:162816 [1] NCCL INFO Channel 09/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162595:162819 [6] NCCL INFO Channel 21/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162591:162821 [2] NCCL INFO Channel 23/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162592:162820 [3] NCCL INFO Channel 18/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162590:162816 [1] NCCL INFO Channel 10/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162595:162819 [6] NCCL INFO Channel 22/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162592:162820 [3] NCCL INFO Channel 19/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162590:162816 [1] NCCL INFO Channel 11/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162595:162819 [6] NCCL INFO Channel 23/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:162590:162816 [1] NCCL INFO Channel 12/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162592:162820 [3] NCCL INFO Channel 20/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162590:162816 [1] NCCL INFO Channel 13/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162592:162820 [3] NCCL INFO Channel 21/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162590:162816 [1] NCCL INFO Channel 14/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162592:162820 [3] NCCL INFO Channel 22/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162590:162816 [1] NCCL INFO Channel 15/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162592:162820 [3] NCCL INFO Channel 23/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:162590:162816 [1] NCCL INFO Channel 16/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162590:162816 [1] NCCL INFO Channel 17/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162590:162816 [1] NCCL INFO Channel 18/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162590:162816 [1] NCCL INFO Channel 19/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162590:162816 [1] NCCL INFO Channel 20/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162590:162816 [1] NCCL INFO Channel 21/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162590:162816 [1] NCCL INFO Channel 22/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162590:162816 [1] NCCL INFO Channel 23/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162596:162817 [7] NCCL INFO Connected all trees
+ip-26-0-169-207:162596:162817 [7] NCCL INFO NVLS comm 0x8949320 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-169-207:162589:162815 [0] NCCL INFO Connected all trees
+ip-26-0-169-207:162589:162815 [0] NCCL INFO NVLS comm 0x954d000 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-169-207:162594:162818 [5] NCCL INFO Connected all trees
+ip-26-0-169-207:162594:162818 [5] NCCL INFO NVLS comm 0x94f3ae0 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-169-207:162595:162819 [6] NCCL INFO Connected all trees
+ip-26-0-169-207:162595:162819 [6] NCCL INFO NVLS comm 0x9318400 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-169-207:162592:162820 [3] NCCL INFO Connected all trees
+ip-26-0-169-207:162592:162820 [3] NCCL INFO NVLS comm 0xa6628e0 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-169-207:162593:162822 [4] NCCL INFO Connected all trees
+ip-26-0-169-207:162593:162822 [4] NCCL INFO NVLS comm 0x9e1dcb0 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-169-207:162591:162821 [2] NCCL INFO Connected all trees
+ip-26-0-169-207:162591:162821 [2] NCCL INFO NVLS comm 0x8ac6650 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-169-207:162590:162816 [1] NCCL INFO Connected all trees
+ip-26-0-169-207:162590:162816 [1] NCCL INFO NVLS comm 0x8c85770 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-169-207:162593:162822 [4] NCCL INFO Connected NVLS tree
+ip-26-0-169-207:162593:162822 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-169-207:162593:162822 [4] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-169-207:162593:162822 [4] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:162590:162816 [1] NCCL INFO Connected NVLS tree
+ip-26-0-169-207:162590:162816 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-169-207:162594:162818 [5] NCCL INFO Connected NVLS tree
+ip-26-0-169-207:162591:162821 [2] NCCL INFO Connected NVLS tree
+ip-26-0-169-207:162590:162816 [1] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-169-207:162590:162816 [1] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:162591:162821 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-169-207:162594:162818 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-169-207:162591:162821 [2] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-169-207:162591:162821 [2] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:162594:162818 [5] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-169-207:162594:162818 [5] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:162589:162815 [0] NCCL INFO Connected NVLS tree
+ip-26-0-169-207:162595:162819 [6] NCCL INFO Connected NVLS tree
+ip-26-0-169-207:162596:162817 [7] NCCL INFO Connected NVLS tree
+ip-26-0-169-207:162595:162819 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-169-207:162589:162815 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-169-207:162595:162819 [6] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-169-207:162595:162819 [6] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:162596:162817 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-169-207:162596:162817 [7] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-169-207:162596:162817 [7] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:162589:162815 [0] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-169-207:162589:162815 [0] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:162592:162820 [3] NCCL INFO Connected NVLS tree
+ip-26-0-169-207:162592:162820 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-169-207:162592:162820 [3] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-169-207:162592:162820 [3] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:162596:162817 [7] NCCL INFO comm 0x8949320 rank 7 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0x87e3927732a5fde8 - Init COMPLETE
+ip-26-0-169-207:162592:162820 [3] NCCL INFO comm 0xa6628e0 rank 3 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0x87e3927732a5fde8 - Init COMPLETE
+ip-26-0-169-207:162590:162816 [1] NCCL INFO comm 0x8c85770 rank 1 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0x87e3927732a5fde8 - Init COMPLETE
+ip-26-0-169-207:162594:162818 [5] NCCL INFO comm 0x94f3ae0 rank 5 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0x87e3927732a5fde8 - Init COMPLETE
+ip-26-0-169-207:162595:162819 [6] NCCL INFO comm 0x9318400 rank 6 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0x87e3927732a5fde8 - Init COMPLETE
+ip-26-0-169-207:162593:162822 [4] NCCL INFO comm 0x9e1dcb0 rank 4 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0x87e3927732a5fde8 - Init COMPLETE
+ip-26-0-169-207:162591:162821 [2] NCCL INFO comm 0x8ac6650 rank 2 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0x87e3927732a5fde8 - Init COMPLETE
+ip-26-0-169-207:162589:162815 [0] NCCL INFO comm 0x954d000 rank 0 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0x87e3927732a5fde8 - Init COMPLETE
+12/12/2024 21:38:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]: Config:
+12/12/2024 21:38:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]: Config(general=GeneralArgs(project='debug',
+12/12/2024 21:38:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                            run='1.14G_dp4_tp2_pp1_acc256_mbs2_seq2048_zero1_tpmodeRED_vocab32k',
+12/12/2024 21:38:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                            seed=42,
+12/12/2024 21:38:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                            step=None,
+12/12/2024 21:38:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                            consumed_train_samples=None,
+12/12/2024 21:38:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                            benchmark_csv_path=PosixPath('benchmark/results/bench_final.csv'),
+12/12/2024 21:38:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                            ignore_sanity_checks=True),
+12/12/2024 21:38:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:        parallelism=ParallelismArgs(dp=4,
+12/12/2024 21:38:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                    pp=1,
+12/12/2024 21:38:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                    tp=2,
+12/12/2024 21:38:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                    pp_engine=<nanotron.parallel.pipeline_parallel.engine.OneForwardOneBackwardPipelineEngine object at 0x7fd0f341ce80>,
+12/12/2024 21:38:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                    tp_mode=<TensorParallelLinearMode.REDUCE_SCATTER: 2>,
+12/12/2024 21:38:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                    tp_linear_async_communication=True,
+12/12/2024 21:38:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                    recompute_layer=False,
+12/12/2024 21:38:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                    tp_recompute_allgather=True,
+12/12/2024 21:38:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                    expert_parallel_size=1),
+12/12/2024 21:38:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:        model=ModelArgs(model_config=LlamaConfig(bos_token_id=0,
+12/12/2024 21:38:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                                 eos_token_id=0,
+12/12/2024 21:38:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                                 hidden_act='silu',
+12/12/2024 21:38:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                                 hidden_size=2048,
+12/12/2024 21:38:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                                 initializer_range=0.02,
+12/12/2024 21:38:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                                 intermediate_size=8192,
+12/12/2024 21:38:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                                 is_llama_config=True,
+12/12/2024 21:38:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                                 max_position_embeddings=2048,
+12/12/2024 21:38:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                                 num_attention_heads=32,
+12/12/2024 21:38:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                                 num_hidden_layers=16,
+12/12/2024 21:38:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                                 num_key_value_heads=8,
+12/12/2024 21:38:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                                 pad_token_id=None,
+12/12/2024 21:38:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                                 pretraining_tp=1,
+12/12/2024 21:38:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                                 rms_norm_eps=1e-05,
+12/12/2024 21:38:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                                 rope_scaling=None,
+12/12/2024 21:38:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                                 rope_theta=10000.0,
+12/12/2024 21:38:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                                 rope_interleaved=False,
+12/12/2024 21:38:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                                 tie_word_embeddings=True,
+12/12/2024 21:38:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                                 use_cache=True,
+12/12/2024 21:38:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                                 vocab_size=32768),
+12/12/2024 21:38:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                        init_method=RandomInit(std=0.02),
+12/12/2024 21:38:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                        dtype=torch.bfloat16,
+12/12/2024 21:38:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                        make_vocab_size_divisible_by=1,
+12/12/2024 21:38:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                        ddp_bucket_cap_mb=25),
+12/12/2024 21:38:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:        tokenizer=TokenizerArgs(tokenizer_name_or_path='robot-test/dummy-tokenizer-wordlevel',
+12/12/2024 21:38:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                tokenizer_revision=None,
+12/12/2024 21:38:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                tokenizer_max_length=None),
+12/12/2024 21:38:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:        checkpoints=CheckpointsArgs(checkpoints_path=PosixPath('checkpoints'),
+12/12/2024 21:38:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                    checkpoint_interval=10000,
+12/12/2024 21:38:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                    save_initial_state=False,
+12/12/2024 21:38:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                    save_final_state=False,
+12/12/2024 21:38:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                    resume_checkpoint_path=None,
+12/12/2024 21:38:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                    checkpoints_path_is_shared_file_system=False),
+12/12/2024 21:38:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:        logging=LoggingArgs(log_level='info',
+12/12/2024 21:38:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                            log_level_replica='info',
+12/12/2024 21:38:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                            iteration_step_info_interval=1),
+12/12/2024 21:38:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:        tokens=TokensArgs(sequence_length=2048,
+12/12/2024 21:38:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                          train_steps=100,
+12/12/2024 21:38:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                          micro_batch_size=2,
+12/12/2024 21:38:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                          batch_accumulation_per_replica=256,
+12/12/2024 21:38:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                          val_check_interval=100,
+12/12/2024 21:38:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                          limit_val_batches=0,
+12/12/2024 21:38:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                          limit_test_batches=0),
+12/12/2024 21:38:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:        optimizer=OptimizerArgs(optimizer_factory=AdamWOptimizerArgs(adam_eps=1e-08,
+12/12/2024 21:38:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                                                     adam_beta1=0.9,
+12/12/2024 21:38:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                                                     adam_beta2=0.95,
+12/12/2024 21:38:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                                                     torch_adam_is_fused=True,
+12/12/2024 21:38:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                                                     name='adamW'),
+12/12/2024 21:38:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                zero_stage=1,
+12/12/2024 21:38:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                weight_decay=0.01,
+12/12/2024 21:38:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                clip_grad=1.0,
+12/12/2024 21:38:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                accumulate_grad_in_fp32=True,
+12/12/2024 21:38:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                learning_rate_scheduler=LRSchedulerArgs(learning_rate=0.0003,
+12/12/2024 21:38:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                                                        lr_warmup_steps=2,
+12/12/2024 21:38:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                                                        lr_warmup_style='linear',
+12/12/2024 21:38:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                                                        lr_decay_style='cosine',
+12/12/2024 21:38:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                                                        lr_decay_steps=13,
+12/12/2024 21:38:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                                                        lr_decay_starting_step=None,
+12/12/2024 21:38:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                                                        min_decay_lr=1e-05)),
+12/12/2024 21:38:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:        data_stages=[DatasetStageArgs(name='Stable Training Stage',
+12/12/2024 21:38:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                      start_training_step=1,
+12/12/2024 21:38:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                      data=DataArgs(dataset=None,
+12/12/2024 21:38:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                                    seed=42,
+12/12/2024 21:38:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                                    num_loading_workers=1))],
+12/12/2024 21:38:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:        profiler=None,
+12/12/2024 21:38:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:        lighteval=None,
+12/12/2024 21:38:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:        s3_upload=None)
+12/12/2024 21:38:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]: Model Config:
+12/12/2024 21:38:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]: LlamaConfig(bos_token_id=0,
+12/12/2024 21:38:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:             eos_token_id=0,
+12/12/2024 21:38:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:             hidden_act='silu',
+12/12/2024 21:38:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:             hidden_size=2048,
+12/12/2024 21:38:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:             initializer_range=0.02,
+12/12/2024 21:38:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:             intermediate_size=8192,
+12/12/2024 21:38:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:             is_llama_config=True,
+12/12/2024 21:38:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:             max_position_embeddings=2048,
+12/12/2024 21:38:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:             num_attention_heads=32,
+12/12/2024 21:38:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:             num_hidden_layers=16,
+12/12/2024 21:38:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:             num_key_value_heads=8,
+12/12/2024 21:38:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:             pad_token_id=None,
+12/12/2024 21:38:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:             pretraining_tp=1,
+12/12/2024 21:38:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:             rms_norm_eps=1e-05,
+12/12/2024 21:38:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:             rope_scaling=None,
+12/12/2024 21:38:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:             rope_theta=10000.0,
+12/12/2024 21:38:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:             rope_interleaved=False,
+12/12/2024 21:38:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:             tie_word_embeddings=True,
+12/12/2024 21:38:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:             use_cache=True,
+12/12/2024 21:38:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:             vocab_size=32768)
+12/12/2024 21:38:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]: Building model..
+12/12/2024 21:38:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]: Initialize RoPE Theta = 10000.0
+12/12/2024 21:38:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]: Setting PP block ranks...
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+ip-26-0-169-207:162589:162875 [0] NCCL INFO Using network Libfabric
+ip-26-0-169-207:162595:162877 [6] NCCL INFO Using network Libfabric
+ip-26-0-169-207:162593:162878 [4] NCCL INFO Using network Libfabric
+ip-26-0-169-207:162591:162876 [2] NCCL INFO Using network Libfabric
+ip-26-0-169-207:162596:162882 [7] NCCL INFO Using network Libfabric
+ip-26-0-169-207:162590:162880 [1] NCCL INFO Using network Libfabric
+ip-26-0-169-207:162592:162879 [3] NCCL INFO Using network Libfabric
+ip-26-0-169-207:162594:162881 [5] NCCL INFO Using network Libfabric
+ip-26-0-169-207:162595:162877 [6] NCCL INFO comm 0xb124570 rank 0 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0xa7eda90d6dd01434 - Init START
+ip-26-0-169-207:162596:162882 [7] NCCL INFO comm 0xa758030 rank 1 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0xa7eda90d6dd01434 - Init START
+ip-26-0-169-207:162589:162875 [0] NCCL INFO comm 0xb35adf0 rank 0 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0xb183c3b2a397968d - Init START
+ip-26-0-169-207:162590:162880 [1] NCCL INFO comm 0xaa91d70 rank 1 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0xb183c3b2a397968d - Init START
+ip-26-0-169-207:162591:162876 [2] NCCL INFO comm 0xa8d2f60 rank 0 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0x476139e31e2f3419 - Init START
+ip-26-0-169-207:162592:162879 [3] NCCL INFO comm 0xc46b0f0 rank 1 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0x476139e31e2f3419 - Init START
+ip-26-0-169-207:162595:162877 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/149
+ip-26-0-169-207:162589:162875 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/162
+ip-26-0-169-207:162591:162876 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/149
+ip-26-0-169-207:162593:162878 [4] NCCL INFO comm 0xbc298f0 rank 0 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0x9225bffb0391ac2f - Init START
+ip-26-0-169-207:162594:162881 [5] NCCL INFO comm 0xb303070 rank 1 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0x9225bffb0391ac2f - Init START
+ip-26-0-169-207:162593:162878 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/149
+ip-26-0-169-207:162590:162880 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/149
+ip-26-0-169-207:162592:162879 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/149
+ip-26-0-169-207:162596:162882 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/149
+ip-26-0-169-207:162594:162881 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/149
+ip-26-0-169-207:162593:162878 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-169-207:162592:162879 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-169-207:162591:162876 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-169-207:162591:162876 [2] NCCL INFO Channel 00/24 :    0   1
+ip-26-0-169-207:162591:162876 [2] NCCL INFO Channel 01/24 :    0   1
+ip-26-0-169-207:162591:162876 [2] NCCL INFO Channel 02/24 :    0   1
+ip-26-0-169-207:162591:162876 [2] NCCL INFO Channel 03/24 :    0   1
+ip-26-0-169-207:162591:162876 [2] NCCL INFO Channel 04/24 :    0   1
+ip-26-0-169-207:162591:162876 [2] NCCL INFO Channel 05/24 :    0   1
+ip-26-0-169-207:162591:162876 [2] NCCL INFO Channel 06/24 :    0   1
+ip-26-0-169-207:162591:162876 [2] NCCL INFO Channel 07/24 :    0   1
+ip-26-0-169-207:162591:162876 [2] NCCL INFO Channel 08/24 :    0   1
+ip-26-0-169-207:162591:162876 [2] NCCL INFO Channel 09/24 :    0   1
+ip-26-0-169-207:162591:162876 [2] NCCL INFO Channel 10/24 :    0   1
+ip-26-0-169-207:162591:162876 [2] NCCL INFO Channel 11/24 :    0   1
+ip-26-0-169-207:162591:162876 [2] NCCL INFO Channel 12/24 :    0   1
+ip-26-0-169-207:162591:162876 [2] NCCL INFO Channel 13/24 :    0   1
+ip-26-0-169-207:162591:162876 [2] NCCL INFO Channel 14/24 :    0   1
+ip-26-0-169-207:162591:162876 [2] NCCL INFO Channel 15/24 :    0   1
+ip-26-0-169-207:162591:162876 [2] NCCL INFO Channel 16/24 :    0   1
+ip-26-0-169-207:162591:162876 [2] NCCL INFO Channel 17/24 :    0   1
+ip-26-0-169-207:162591:162876 [2] NCCL INFO Channel 18/24 :    0   1
+ip-26-0-169-207:162591:162876 [2] NCCL INFO Channel 19/24 :    0   1
+ip-26-0-169-207:162591:162876 [2] NCCL INFO Channel 20/24 :    0   1
+ip-26-0-169-207:162591:162876 [2] NCCL INFO Channel 21/24 :    0   1
+ip-26-0-169-207:162591:162876 [2] NCCL INFO Channel 22/24 :    0   1
+ip-26-0-169-207:162591:162876 [2] NCCL INFO Channel 23/24 :    0   1
+ip-26-0-169-207:162592:162879 [3] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] -1/-1/-1->1->0 [3] -1/-1/-1->1->0 [4] -1/-1/-1->1->0 [5] -1/-1/-1->1->0 [6] 0/-1/-1->1->-1 [7] 0/-1/-1->1->-1 [8] 0/-1/-1->1->-1 [9] 0/-1/-1->1->-1 [10] 0/-1/-1->1->-1 [11] 0/-1/-1->1->-1 [12] -1/-1/-1->1->0 [13] -1/-1/-1->1->0 [14] -1/-1/-1->1->0 [15] -1/-1/-1->1->0 [16] -1/-1/-1->1->0 [17] -1/-1/-1->1->0 [18] 0/-1/-1->1->-1 [19] 0/-1/-1->1->-1 [20] 0/-1/-1->1->-1 [21] 0/-1/-1->1->-1 [22] 0/-1/-1->1->-1 [23] 0/-1/-1->1->-1
+ip-26-0-169-207:162591:162876 [2] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] -1/-1/-1->0->1 [7] -1/-1/-1->0->1 [8] -1/-1/-1->0->1 [9] -1/-1/-1->0->1 [10] -1/-1/-1->0->1 [11] -1/-1/-1->0->1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] -1/-1/-1->0->1 [19] -1/-1/-1->0->1 [20] -1/-1/-1->0->1 [21] -1/-1/-1->0->1 [22] -1/-1/-1->0->1 [23] -1/-1/-1->0->1
+ip-26-0-169-207:162592:162879 [3] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-169-207:162591:162876 [2] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-169-207:162594:162881 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-169-207:162593:162878 [4] NCCL INFO Channel 00/24 :    0   1
+ip-26-0-169-207:162593:162878 [4] NCCL INFO Channel 01/24 :    0   1
+ip-26-0-169-207:162593:162878 [4] NCCL INFO Channel 02/24 :    0   1
+ip-26-0-169-207:162593:162878 [4] NCCL INFO Channel 03/24 :    0   1
+ip-26-0-169-207:162593:162878 [4] NCCL INFO Channel 04/24 :    0   1
+ip-26-0-169-207:162593:162878 [4] NCCL INFO Channel 05/24 :    0   1
+ip-26-0-169-207:162593:162878 [4] NCCL INFO Channel 06/24 :    0   1
+ip-26-0-169-207:162593:162878 [4] NCCL INFO Channel 07/24 :    0   1
+ip-26-0-169-207:162593:162878 [4] NCCL INFO Channel 08/24 :    0   1
+ip-26-0-169-207:162593:162878 [4] NCCL INFO Channel 09/24 :    0   1
+ip-26-0-169-207:162593:162878 [4] NCCL INFO Channel 10/24 :    0   1
+ip-26-0-169-207:162593:162878 [4] NCCL INFO Channel 11/24 :    0   1
+ip-26-0-169-207:162593:162878 [4] NCCL INFO Channel 12/24 :    0   1
+ip-26-0-169-207:162593:162878 [4] NCCL INFO Channel 13/24 :    0   1
+ip-26-0-169-207:162593:162878 [4] NCCL INFO Channel 14/24 :    0   1
+ip-26-0-169-207:162593:162878 [4] NCCL INFO Channel 15/24 :    0   1
+ip-26-0-169-207:162593:162878 [4] NCCL INFO Channel 16/24 :    0   1
+ip-26-0-169-207:162593:162878 [4] NCCL INFO Channel 17/24 :    0   1
+ip-26-0-169-207:162593:162878 [4] NCCL INFO Channel 18/24 :    0   1
+ip-26-0-169-207:162593:162878 [4] NCCL INFO Channel 19/24 :    0   1
+ip-26-0-169-207:162593:162878 [4] NCCL INFO Channel 20/24 :    0   1
+ip-26-0-169-207:162593:162878 [4] NCCL INFO Channel 21/24 :    0   1
+ip-26-0-169-207:162593:162878 [4] NCCL INFO Channel 22/24 :    0   1
+ip-26-0-169-207:162593:162878 [4] NCCL INFO Channel 23/24 :    0   1
+ip-26-0-169-207:162594:162881 [5] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] -1/-1/-1->1->0 [3] -1/-1/-1->1->0 [4] -1/-1/-1->1->0 [5] -1/-1/-1->1->0 [6] 0/-1/-1->1->-1 [7] 0/-1/-1->1->-1 [8] 0/-1/-1->1->-1 [9] 0/-1/-1->1->-1 [10] 0/-1/-1->1->-1 [11] 0/-1/-1->1->-1 [12] -1/-1/-1->1->0 [13] -1/-1/-1->1->0 [14] -1/-1/-1->1->0 [15] -1/-1/-1->1->0 [16] -1/-1/-1->1->0 [17] -1/-1/-1->1->0 [18] 0/-1/-1->1->-1 [19] 0/-1/-1->1->-1 [20] 0/-1/-1->1->-1 [21] 0/-1/-1->1->-1 [22] 0/-1/-1->1->-1 [23] 0/-1/-1->1->-1
+ip-26-0-169-207:162594:162881 [5] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-169-207:162593:162878 [4] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] -1/-1/-1->0->1 [7] -1/-1/-1->0->1 [8] -1/-1/-1->0->1 [9] -1/-1/-1->0->1 [10] -1/-1/-1->0->1 [11] -1/-1/-1->0->1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] -1/-1/-1->0->1 [19] -1/-1/-1->0->1 [20] -1/-1/-1->0->1 [21] -1/-1/-1->0->1 [22] -1/-1/-1->0->1 [23] -1/-1/-1->0->1
+ip-26-0-169-207:162593:162878 [4] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-169-207:162590:162880 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-169-207:162596:162882 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-169-207:162595:162877 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-169-207:162595:162877 [6] NCCL INFO Channel 00/24 :    0   1
+ip-26-0-169-207:162595:162877 [6] NCCL INFO Channel 01/24 :    0   1
+ip-26-0-169-207:162595:162877 [6] NCCL INFO Channel 02/24 :    0   1
+ip-26-0-169-207:162595:162877 [6] NCCL INFO Channel 03/24 :    0   1
+ip-26-0-169-207:162595:162877 [6] NCCL INFO Channel 04/24 :    0   1
+ip-26-0-169-207:162595:162877 [6] NCCL INFO Channel 05/24 :    0   1
+ip-26-0-169-207:162595:162877 [6] NCCL INFO Channel 06/24 :    0   1
+ip-26-0-169-207:162596:162882 [7] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] -1/-1/-1->1->0 [3] -1/-1/-1->1->0 [4] -1/-1/-1->1->0 [5] -1/-1/-1->1->0 [6] 0/-1/-1->1->-1 [7] 0/-1/-1->1->-1 [8] 0/-1/-1->1->-1 [9] 0/-1/-1->1->-1 [10] 0/-1/-1->1->-1 [11] 0/-1/-1->1->-1 [12] -1/-1/-1->1->0 [13] -1/-1/-1->1->0 [14] -1/-1/-1->1->0 [15] -1/-1/-1->1->0 [16] -1/-1/-1->1->0 [17] -1/-1/-1->1->0 [18] 0/-1/-1->1->-1 [19] 0/-1/-1->1->-1 [20] 0/-1/-1->1->-1 [21] 0/-1/-1->1->-1 [22] 0/-1/-1->1->-1 [23] 0/-1/-1->1->-1
+ip-26-0-169-207:162595:162877 [6] NCCL INFO Channel 07/24 :    0   1
+ip-26-0-169-207:162595:162877 [6] NCCL INFO Channel 08/24 :    0   1
+ip-26-0-169-207:162596:162882 [7] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-169-207:162595:162877 [6] NCCL INFO Channel 09/24 :    0   1
+ip-26-0-169-207:162595:162877 [6] NCCL INFO Channel 10/24 :    0   1
+ip-26-0-169-207:162595:162877 [6] NCCL INFO Channel 11/24 :    0   1
+ip-26-0-169-207:162595:162877 [6] NCCL INFO Channel 12/24 :    0   1
+ip-26-0-169-207:162595:162877 [6] NCCL INFO Channel 13/24 :    0   1
+ip-26-0-169-207:162595:162877 [6] NCCL INFO Channel 14/24 :    0   1
+ip-26-0-169-207:162595:162877 [6] NCCL INFO Channel 15/24 :    0   1
+ip-26-0-169-207:162595:162877 [6] NCCL INFO Channel 16/24 :    0   1
+ip-26-0-169-207:162595:162877 [6] NCCL INFO Channel 17/24 :    0   1
+ip-26-0-169-207:162595:162877 [6] NCCL INFO Channel 18/24 :    0   1
+ip-26-0-169-207:162595:162877 [6] NCCL INFO Channel 19/24 :    0   1
+ip-26-0-169-207:162595:162877 [6] NCCL INFO Channel 20/24 :    0   1
+ip-26-0-169-207:162595:162877 [6] NCCL INFO Channel 21/24 :    0   1
+ip-26-0-169-207:162595:162877 [6] NCCL INFO Channel 22/24 :    0   1
+ip-26-0-169-207:162595:162877 [6] NCCL INFO Channel 23/24 :    0   1
+ip-26-0-169-207:162595:162877 [6] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] -1/-1/-1->0->1 [7] -1/-1/-1->0->1 [8] -1/-1/-1->0->1 [9] -1/-1/-1->0->1 [10] -1/-1/-1->0->1 [11] -1/-1/-1->0->1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] -1/-1/-1->0->1 [19] -1/-1/-1->0->1 [20] -1/-1/-1->0->1 [21] -1/-1/-1->0->1 [22] -1/-1/-1->0->1 [23] -1/-1/-1->0->1
+ip-26-0-169-207:162595:162877 [6] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-169-207:162589:162875 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-169-207:162589:162875 [0] NCCL INFO Channel 00/24 :    0   1
+ip-26-0-169-207:162589:162875 [0] NCCL INFO Channel 01/24 :    0   1
+ip-26-0-169-207:162589:162875 [0] NCCL INFO Channel 02/24 :    0   1
+ip-26-0-169-207:162589:162875 [0] NCCL INFO Channel 03/24 :    0   1
+ip-26-0-169-207:162590:162880 [1] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] -1/-1/-1->1->0 [3] -1/-1/-1->1->0 [4] -1/-1/-1->1->0 [5] -1/-1/-1->1->0 [6] 0/-1/-1->1->-1 [7] 0/-1/-1->1->-1 [8] 0/-1/-1->1->-1 [9] 0/-1/-1->1->-1 [10] 0/-1/-1->1->-1 [11] 0/-1/-1->1->-1 [12] -1/-1/-1->1->0 [13] -1/-1/-1->1->0 [14] -1/-1/-1->1->0 [15] -1/-1/-1->1->0 [16] -1/-1/-1->1->0 [17] -1/-1/-1->1->0 [18] 0/-1/-1->1->-1 [19] 0/-1/-1->1->-1 [20] 0/-1/-1->1->-1 [21] 0/-1/-1->1->-1 [22] 0/-1/-1->1->-1 [23] 0/-1/-1->1->-1
+ip-26-0-169-207:162589:162875 [0] NCCL INFO Channel 04/24 :    0   1
+ip-26-0-169-207:162589:162875 [0] NCCL INFO Channel 05/24 :    0   1
+ip-26-0-169-207:162590:162880 [1] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-169-207:162589:162875 [0] NCCL INFO Channel 06/24 :    0   1
+ip-26-0-169-207:162589:162875 [0] NCCL INFO Channel 07/24 :    0   1
+ip-26-0-169-207:162589:162875 [0] NCCL INFO Channel 08/24 :    0   1
+ip-26-0-169-207:162589:162875 [0] NCCL INFO Channel 09/24 :    0   1
+ip-26-0-169-207:162589:162875 [0] NCCL INFO Channel 10/24 :    0   1
+ip-26-0-169-207:162589:162875 [0] NCCL INFO Channel 11/24 :    0   1
+ip-26-0-169-207:162589:162875 [0] NCCL INFO Channel 12/24 :    0   1
+ip-26-0-169-207:162589:162875 [0] NCCL INFO Channel 13/24 :    0   1
+ip-26-0-169-207:162589:162875 [0] NCCL INFO Channel 14/24 :    0   1
+ip-26-0-169-207:162589:162875 [0] NCCL INFO Channel 15/24 :    0   1
+ip-26-0-169-207:162589:162875 [0] NCCL INFO Channel 16/24 :    0   1
+ip-26-0-169-207:162589:162875 [0] NCCL INFO Channel 17/24 :    0   1
+ip-26-0-169-207:162589:162875 [0] NCCL INFO Channel 18/24 :    0   1
+ip-26-0-169-207:162589:162875 [0] NCCL INFO Channel 19/24 :    0   1
+ip-26-0-169-207:162589:162875 [0] NCCL INFO Channel 20/24 :    0   1
+ip-26-0-169-207:162589:162875 [0] NCCL INFO Channel 21/24 :    0   1
+ip-26-0-169-207:162589:162875 [0] NCCL INFO Channel 22/24 :    0   1
+ip-26-0-169-207:162589:162875 [0] NCCL INFO Channel 23/24 :    0   1
+ip-26-0-169-207:162589:162875 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] -1/-1/-1->0->1 [7] -1/-1/-1->0->1 [8] -1/-1/-1->0->1 [9] -1/-1/-1->0->1 [10] -1/-1/-1->0->1 [11] -1/-1/-1->0->1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] -1/-1/-1->0->1 [19] -1/-1/-1->0->1 [20] -1/-1/-1->0->1 [21] -1/-1/-1->0->1 [22] -1/-1/-1->0->1 [23] -1/-1/-1->0->1
+ip-26-0-169-207:162589:162875 [0] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-169-207:162591:162876 [2] NCCL INFO Channel 00/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-169-207:162591:162876 [2] NCCL INFO Channel 01/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-169-207:162593:162878 [4] NCCL INFO Channel 00/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-169-207:162591:162876 [2] NCCL INFO Channel 02/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-169-207:162593:162878 [4] NCCL INFO Channel 01/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-169-207:162591:162876 [2] NCCL INFO Channel 03/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-169-207:162593:162878 [4] NCCL INFO Channel 02/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-169-207:162591:162876 [2] NCCL INFO Channel 04/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-169-207:162593:162878 [4] NCCL INFO Channel 03/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-169-207:162591:162876 [2] NCCL INFO Channel 05/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-169-207:162592:162879 [3] NCCL INFO Channel 00/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-169-207:162593:162878 [4] NCCL INFO Channel 04/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-169-207:162591:162876 [2] NCCL INFO Channel 06/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-169-207:162592:162879 [3] NCCL INFO Channel 01/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-169-207:162593:162878 [4] NCCL INFO Channel 05/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-169-207:162591:162876 [2] NCCL INFO Channel 07/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-169-207:162596:162882 [7] NCCL INFO Channel 00/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-169-207:162592:162879 [3] NCCL INFO Channel 02/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-169-207:162593:162878 [4] NCCL INFO Channel 06/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-169-207:162594:162881 [5] NCCL INFO Channel 00/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-169-207:162591:162876 [2] NCCL INFO Channel 08/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-169-207:162596:162882 [7] NCCL INFO Channel 01/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-169-207:162592:162879 [3] NCCL INFO Channel 03/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-169-207:162593:162878 [4] NCCL INFO Channel 07/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-169-207:162594:162881 [5] NCCL INFO Channel 01/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-169-207:162591:162876 [2] NCCL INFO Channel 09/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-169-207:162596:162882 [7] NCCL INFO Channel 02/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-169-207:162591:162876 [2] NCCL INFO Channel 10/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-169-207:162592:162879 [3] NCCL INFO Channel 04/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-169-207:162593:162878 [4] NCCL INFO Channel 08/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-169-207:162592:162879 [3] NCCL INFO Channel 05/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-169-207:162594:162881 [5] NCCL INFO Channel 02/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-169-207:162593:162878 [4] NCCL INFO Channel 09/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-169-207:162592:162879 [3] NCCL INFO Channel 06/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-169-207:162594:162881 [5] NCCL INFO Channel 03/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-169-207:162593:162878 [4] NCCL INFO Channel 10/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-169-207:162592:162879 [3] NCCL INFO Channel 07/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-169-207:162595:162877 [6] NCCL INFO Channel 00/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-169-207:162594:162881 [5] NCCL INFO Channel 04/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-169-207:162593:162878 [4] NCCL INFO Channel 11/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-169-207:162592:162879 [3] NCCL INFO Channel 08/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-169-207:162595:162877 [6] NCCL INFO Channel 01/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-169-207:162594:162881 [5] NCCL INFO Channel 05/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-169-207:162593:162878 [4] NCCL INFO Channel 12/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-169-207:162592:162879 [3] NCCL INFO Channel 09/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-169-207:162595:162877 [6] NCCL INFO Channel 02/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-169-207:162594:162881 [5] NCCL INFO Channel 06/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-169-207:162593:162878 [4] NCCL INFO Channel 13/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-169-207:162592:162879 [3] NCCL INFO Channel 10/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-169-207:162589:162875 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162595:162877 [6] NCCL INFO Channel 03/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-169-207:162594:162881 [5] NCCL INFO Channel 07/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-169-207:162593:162878 [4] NCCL INFO Channel 14/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-169-207:162596:162882 [7] NCCL INFO Channel 03/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-169-207:162592:162879 [3] NCCL INFO Channel 11/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-169-207:162589:162875 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162595:162877 [6] NCCL INFO Channel 04/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-169-207:162594:162881 [5] NCCL INFO Channel 08/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-169-207:162593:162878 [4] NCCL INFO Channel 15/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-169-207:162591:162876 [2] NCCL INFO Channel 11/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-169-207:162596:162882 [7] NCCL INFO Channel 04/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-169-207:162590:162880 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162592:162879 [3] NCCL INFO Channel 12/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-169-207:162589:162875 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162595:162877 [6] NCCL INFO Channel 05/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-169-207:162594:162881 [5] NCCL INFO Channel 09/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-169-207:162593:162878 [4] NCCL INFO Channel 16/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-169-207:162591:162876 [2] NCCL INFO Channel 12/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-169-207:162596:162882 [7] NCCL INFO Channel 05/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-169-207:162592:162879 [3] NCCL INFO Channel 13/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-169-207:162590:162880 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162589:162875 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162595:162877 [6] NCCL INFO Channel 06/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-169-207:162594:162881 [5] NCCL INFO Channel 10/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-169-207:162593:162878 [4] NCCL INFO Channel 17/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-169-207:162591:162876 [2] NCCL INFO Channel 13/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-169-207:162596:162882 [7] NCCL INFO Channel 06/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-169-207:162590:162880 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162592:162879 [3] NCCL INFO Channel 14/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-169-207:162589:162875 [0] NCCL INFO Channel 04/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162595:162877 [6] NCCL INFO Channel 07/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-169-207:162594:162881 [5] NCCL INFO Channel 11/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-169-207:162593:162878 [4] NCCL INFO Channel 18/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-169-207:162591:162876 [2] NCCL INFO Channel 14/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-169-207:162596:162882 [7] NCCL INFO Channel 07/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-169-207:162590:162880 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162592:162879 [3] NCCL INFO Channel 15/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-169-207:162589:162875 [0] NCCL INFO Channel 05/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162595:162877 [6] NCCL INFO Channel 08/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-169-207:162594:162881 [5] NCCL INFO Channel 12/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-169-207:162593:162878 [4] NCCL INFO Channel 19/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-169-207:162591:162876 [2] NCCL INFO Channel 15/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-169-207:162596:162882 [7] NCCL INFO Channel 08/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-169-207:162590:162880 [1] NCCL INFO Channel 04/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162592:162879 [3] NCCL INFO Channel 16/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-169-207:162589:162875 [0] NCCL INFO Channel 06/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162595:162877 [6] NCCL INFO Channel 09/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-169-207:162593:162878 [4] NCCL INFO Channel 20/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-169-207:162594:162881 [5] NCCL INFO Channel 13/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-169-207:162591:162876 [2] NCCL INFO Channel 16/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-169-207:162596:162882 [7] NCCL INFO Channel 09/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-169-207:162590:162880 [1] NCCL INFO Channel 05/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162592:162879 [3] NCCL INFO Channel 17/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-169-207:162589:162875 [0] NCCL INFO Channel 07/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162594:162881 [5] NCCL INFO Channel 14/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-169-207:162595:162877 [6] NCCL INFO Channel 10/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-169-207:162593:162878 [4] NCCL INFO Channel 21/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-169-207:162591:162876 [2] NCCL INFO Channel 17/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-169-207:162596:162882 [7] NCCL INFO Channel 10/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-169-207:162590:162880 [1] NCCL INFO Channel 06/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162592:162879 [3] NCCL INFO Channel 18/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-169-207:162589:162875 [0] NCCL INFO Channel 08/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162594:162881 [5] NCCL INFO Channel 15/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-169-207:162595:162877 [6] NCCL INFO Channel 11/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-169-207:162593:162878 [4] NCCL INFO Channel 22/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-169-207:162591:162876 [2] NCCL INFO Channel 18/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-169-207:162596:162882 [7] NCCL INFO Channel 11/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-169-207:162590:162880 [1] NCCL INFO Channel 07/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162592:162879 [3] NCCL INFO Channel 19/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-169-207:162589:162875 [0] NCCL INFO Channel 09/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162594:162881 [5] NCCL INFO Channel 16/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-169-207:162595:162877 [6] NCCL INFO Channel 12/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-169-207:162593:162878 [4] NCCL INFO Channel 23/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-169-207:162591:162876 [2] NCCL INFO Channel 19/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-169-207:162596:162882 [7] NCCL INFO Channel 12/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-169-207:162589:162875 [0] NCCL INFO Channel 10/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162595:162877 [6] NCCL INFO Channel 13/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-169-207:162596:162882 [7] NCCL INFO Channel 13/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-169-207:162590:162880 [1] NCCL INFO Channel 08/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162589:162875 [0] NCCL INFO Channel 11/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162595:162877 [6] NCCL INFO Channel 14/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-169-207:162596:162882 [7] NCCL INFO Channel 14/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-169-207:162590:162880 [1] NCCL INFO Channel 09/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162589:162875 [0] NCCL INFO Channel 12/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162595:162877 [6] NCCL INFO Channel 15/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-169-207:162596:162882 [7] NCCL INFO Channel 15/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-169-207:162590:162880 [1] NCCL INFO Channel 10/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162589:162875 [0] NCCL INFO Channel 13/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162595:162877 [6] NCCL INFO Channel 16/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-169-207:162596:162882 [7] NCCL INFO Channel 16/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-169-207:162590:162880 [1] NCCL INFO Channel 11/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162589:162875 [0] NCCL INFO Channel 14/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162595:162877 [6] NCCL INFO Channel 17/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-169-207:162596:162882 [7] NCCL INFO Channel 17/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-169-207:162590:162880 [1] NCCL INFO Channel 12/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162589:162875 [0] NCCL INFO Channel 15/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162595:162877 [6] NCCL INFO Channel 18/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-169-207:162596:162882 [7] NCCL INFO Channel 18/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-169-207:162590:162880 [1] NCCL INFO Channel 13/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162589:162875 [0] NCCL INFO Channel 16/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162595:162877 [6] NCCL INFO Channel 19/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-169-207:162596:162882 [7] NCCL INFO Channel 19/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-169-207:162592:162879 [3] NCCL INFO Channel 20/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-169-207:162590:162880 [1] NCCL INFO Channel 14/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162589:162875 [0] NCCL INFO Channel 17/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162594:162881 [5] NCCL INFO Channel 17/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-169-207:162595:162877 [6] NCCL INFO Channel 20/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-169-207:162591:162876 [2] NCCL INFO Channel 20/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-169-207:162596:162882 [7] NCCL INFO Channel 20/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-169-207:162592:162879 [3] NCCL INFO Channel 21/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-169-207:162590:162880 [1] NCCL INFO Channel 15/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162589:162875 [0] NCCL INFO Channel 18/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162594:162881 [5] NCCL INFO Channel 18/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-169-207:162595:162877 [6] NCCL INFO Channel 21/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-169-207:162591:162876 [2] NCCL INFO Channel 21/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-169-207:162596:162882 [7] NCCL INFO Channel 21/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-169-207:162592:162879 [3] NCCL INFO Channel 22/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-169-207:162590:162880 [1] NCCL INFO Channel 16/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162589:162875 [0] NCCL INFO Channel 19/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162594:162881 [5] NCCL INFO Channel 19/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-169-207:162595:162877 [6] NCCL INFO Channel 22/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-169-207:162591:162876 [2] NCCL INFO Channel 22/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-169-207:162596:162882 [7] NCCL INFO Channel 22/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-169-207:162592:162879 [3] NCCL INFO Channel 23/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-169-207:162589:162875 [0] NCCL INFO Channel 20/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162590:162880 [1] NCCL INFO Channel 17/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162594:162881 [5] NCCL INFO Channel 20/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-169-207:162595:162877 [6] NCCL INFO Channel 23/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-169-207:162591:162876 [2] NCCL INFO Channel 23/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-169-207:162596:162882 [7] NCCL INFO Channel 23/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-169-207:162590:162880 [1] NCCL INFO Channel 18/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162589:162875 [0] NCCL INFO Channel 21/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162594:162881 [5] NCCL INFO Channel 21/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-169-207:162589:162875 [0] NCCL INFO Channel 22/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162590:162880 [1] NCCL INFO Channel 19/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162594:162881 [5] NCCL INFO Channel 22/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-169-207:162589:162875 [0] NCCL INFO Channel 23/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:162590:162880 [1] NCCL INFO Channel 20/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162594:162881 [5] NCCL INFO Channel 23/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-169-207:162590:162880 [1] NCCL INFO Channel 21/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162590:162880 [1] NCCL INFO Channel 22/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162590:162880 [1] NCCL INFO Channel 23/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162592:162879 [3] NCCL INFO Connected all rings
+ip-26-0-169-207:162592:162879 [3] NCCL INFO Connected all trees
+ip-26-0-169-207:162592:162879 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-169-207:162592:162879 [3] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-169-207:162592:162879 [3] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:162596:162882 [7] NCCL INFO Connected all rings
+ip-26-0-169-207:162596:162882 [7] NCCL INFO Connected all trees
+ip-26-0-169-207:162596:162882 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-169-207:162596:162882 [7] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-169-207:162596:162882 [7] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:162595:162877 [6] NCCL INFO Connected all rings
+ip-26-0-169-207:162595:162877 [6] NCCL INFO Connected all trees
+ip-26-0-169-207:162595:162877 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-169-207:162595:162877 [6] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-169-207:162595:162877 [6] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:162591:162876 [2] NCCL INFO Connected all rings
+ip-26-0-169-207:162591:162876 [2] NCCL INFO Connected all trees
+ip-26-0-169-207:162591:162876 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-169-207:162591:162876 [2] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-169-207:162591:162876 [2] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:162594:162881 [5] NCCL INFO Connected all rings
+ip-26-0-169-207:162594:162881 [5] NCCL INFO Connected all trees
+ip-26-0-169-207:162594:162881 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-169-207:162594:162881 [5] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-169-207:162594:162881 [5] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:162593:162878 [4] NCCL INFO Connected all rings
+ip-26-0-169-207:162593:162878 [4] NCCL INFO Connected all trees
+ip-26-0-169-207:162593:162878 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-169-207:162593:162878 [4] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-169-207:162593:162878 [4] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:162592:162879 [3] NCCL INFO comm 0xc46b0f0 rank 1 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0x476139e31e2f3419 - Init COMPLETE
+ip-26-0-169-207:162591:162876 [2] NCCL INFO comm 0xa8d2f60 rank 0 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0x476139e31e2f3419 - Init COMPLETE
+NCCL version 2.18.5+cuda12.2
+ip-26-0-169-207:162591:162900 [2] NCCL INFO Using network Libfabric
+ip-26-0-169-207:162591:162900 [2] NCCL INFO comm 0xa8e6060 rank 0 nranks 1 cudaDev 2 nvmlDev 2 busId 75000 commId 0xd5e80c39810b46c5 - Init START
+ip-26-0-169-207:162591:162900 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/149
+ip-26-0-169-207:162592:162902 [3] NCCL INFO Using network Libfabric
+ip-26-0-169-207:162592:162902 [3] NCCL INFO comm 0xc47ddb0 rank 0 nranks 1 cudaDev 3 nvmlDev 3 busId 86000 commId 0x55552d6a1be0dbdd - Init START
+ip-26-0-169-207:162592:162902 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/149
+ip-26-0-169-207:162593:162878 [4] NCCL INFO comm 0xbc298f0 rank 0 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0x9225bffb0391ac2f - Init COMPLETE
+ip-26-0-169-207:162594:162881 [5] NCCL INFO comm 0xb303070 rank 1 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0x9225bffb0391ac2f - Init COMPLETE
+ip-26-0-169-207:162596:162882 [7] NCCL INFO comm 0xa758030 rank 1 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0xa7eda90d6dd01434 - Init COMPLETE
+ip-26-0-169-207:162595:162877 [6] NCCL INFO comm 0xb124570 rank 0 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0xa7eda90d6dd01434 - Init COMPLETE
+ip-26-0-169-207:162589:162875 [0] NCCL INFO Connected all rings
+ip-26-0-169-207:162589:162875 [0] NCCL INFO Connected all trees
+ip-26-0-169-207:162589:162875 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-169-207:162589:162875 [0] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-169-207:162589:162875 [0] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:162593:162904 [4] NCCL INFO Using network Libfabric
+ip-26-0-169-207:162593:162904 [4] NCCL INFO comm 0xbc3c8e0 rank 0 nranks 1 cudaDev 4 nvmlDev 4 busId 97000 commId 0x279428a0c5d2f3aa - Init START
+ip-26-0-169-207:162593:162904 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/149
+NCCL version 2.18.5+cuda12.2
+ip-26-0-169-207:162594:162907 [5] NCCL INFO Using network Libfabric
+ip-26-0-169-207:162594:162907 [5] NCCL INFO comm 0xb315d30 rank 0 nranks 1 cudaDev 5 nvmlDev 5 busId a8000 commId 0x70662a167e82cbc0 - Init START
+ip-26-0-169-207:162594:162907 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/149
+ip-26-0-169-207:162595:162909 [6] NCCL INFO Using network Libfabric
+ip-26-0-169-207:162595:162909 [6] NCCL INFO comm 0xb137670 rank 0 nranks 1 cudaDev 6 nvmlDev 6 busId b9000 commId 0x8d4317fcab0a9d58 - Init START
+ip-26-0-169-207:162595:162909 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/149
+NCCL version 2.18.5+cuda12.2
+ip-26-0-169-207:162596:162911 [7] NCCL INFO Using network Libfabric
+ip-26-0-169-207:162596:162911 [7] NCCL INFO comm 0xa76acf0 rank 0 nranks 1 cudaDev 7 nvmlDev 7 busId ca000 commId 0x6c5d2b4ad9a41899 - Init START
+ip-26-0-169-207:162596:162911 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/149
+ip-26-0-169-207:162590:162880 [1] NCCL INFO Connected all rings
+ip-26-0-169-207:162590:162880 [1] NCCL INFO Connected all trees
+ip-26-0-169-207:162590:162880 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-169-207:162590:162880 [1] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-169-207:162590:162880 [1] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:162590:162880 [1] NCCL INFO comm 0xaa91d70 rank 1 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0xb183c3b2a397968d - Init COMPLETE
+ip-26-0-169-207:162589:162875 [0] NCCL INFO comm 0xb35adf0 rank 0 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0xb183c3b2a397968d - Init COMPLETE
+NCCL version 2.18.5+cuda12.2
+ip-26-0-169-207:162590:162914 [1] NCCL INFO Using network Libfabric
+ip-26-0-169-207:162590:162914 [1] NCCL INFO comm 0xaaa4a30 rank 0 nranks 1 cudaDev 1 nvmlDev 1 busId 64000 commId 0x24918949dfdbf18e - Init START
+ip-26-0-169-207:162590:162914 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/149
+ip-26-0-169-207:162589:162916 [0] NCCL INFO Using network Libfabric
+ip-26-0-169-207:162589:162916 [0] NCCL INFO comm 0xb36dde0 rank 0 nranks 1 cudaDev 0 nvmlDev 0 busId 53000 commId 0x5f1072d1d3fa1a2b - Init START
+ip-26-0-169-207:162589:162916 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/162
+ip-26-0-169-207:162591:162900 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-169-207:162591:162900 [2] NCCL INFO Channel 00/32 :    0
+ip-26-0-169-207:162591:162900 [2] NCCL INFO Channel 01/32 :    0
+ip-26-0-169-207:162591:162900 [2] NCCL INFO Channel 02/32 :    0
+ip-26-0-169-207:162591:162900 [2] NCCL INFO Channel 03/32 :    0
+ip-26-0-169-207:162591:162900 [2] NCCL INFO Channel 04/32 :    0
+ip-26-0-169-207:162591:162900 [2] NCCL INFO Channel 05/32 :    0
+ip-26-0-169-207:162591:162900 [2] NCCL INFO Channel 06/32 :    0
+ip-26-0-169-207:162591:162900 [2] NCCL INFO Channel 07/32 :    0
+ip-26-0-169-207:162591:162900 [2] NCCL INFO Channel 08/32 :    0
+ip-26-0-169-207:162591:162900 [2] NCCL INFO Channel 09/32 :    0
+ip-26-0-169-207:162591:162900 [2] NCCL INFO Channel 10/32 :    0
+ip-26-0-169-207:162591:162900 [2] NCCL INFO Channel 11/32 :    0
+ip-26-0-169-207:162591:162900 [2] NCCL INFO Channel 12/32 :    0
+ip-26-0-169-207:162591:162900 [2] NCCL INFO Channel 13/32 :    0
+ip-26-0-169-207:162591:162900 [2] NCCL INFO Channel 14/32 :    0
+ip-26-0-169-207:162591:162900 [2] NCCL INFO Channel 15/32 :    0
+ip-26-0-169-207:162591:162900 [2] NCCL INFO Channel 16/32 :    0
+ip-26-0-169-207:162591:162900 [2] NCCL INFO Channel 17/32 :    0
+ip-26-0-169-207:162591:162900 [2] NCCL INFO Channel 18/32 :    0
+ip-26-0-169-207:162591:162900 [2] NCCL INFO Channel 19/32 :    0
+ip-26-0-169-207:162591:162900 [2] NCCL INFO Channel 20/32 :    0
+ip-26-0-169-207:162591:162900 [2] NCCL INFO Channel 21/32 :    0
+ip-26-0-169-207:162591:162900 [2] NCCL INFO Channel 22/32 :    0
+ip-26-0-169-207:162591:162900 [2] NCCL INFO Channel 23/32 :    0
+ip-26-0-169-207:162591:162900 [2] NCCL INFO Channel 24/32 :    0
+ip-26-0-169-207:162591:162900 [2] NCCL INFO Channel 25/32 :    0
+ip-26-0-169-207:162591:162900 [2] NCCL INFO Channel 26/32 :    0
+ip-26-0-169-207:162591:162900 [2] NCCL INFO Channel 27/32 :    0
+ip-26-0-169-207:162591:162900 [2] NCCL INFO Channel 28/32 :    0
+ip-26-0-169-207:162591:162900 [2] NCCL INFO Channel 29/32 :    0
+ip-26-0-169-207:162591:162900 [2] NCCL INFO Channel 30/32 :    0
+ip-26-0-169-207:162591:162900 [2] NCCL INFO Channel 31/32 :    0
+ip-26-0-169-207:162591:162900 [2] NCCL INFO Trees [0] -1/-1/-1->0->-1 [1] -1/-1/-1->0->-1 [2] -1/-1/-1->0->-1 [3] -1/-1/-1->0->-1 [4] -1/-1/-1->0->-1 [5] -1/-1/-1->0->-1 [6] -1/-1/-1->0->-1 [7] -1/-1/-1->0->-1 [8] -1/-1/-1->0->-1 [9] -1/-1/-1->0->-1 [10] -1/-1/-1->0->-1 [11] -1/-1/-1->0->-1 [12] -1/-1/-1->0->-1 [13] -1/-1/-1->0->-1 [14] -1/-1/-1->0->-1 [15] -1/-1/-1->0->-1 [16] -1/-1/-1->0->-1 [17] -1/-1/-1->0->-1 [18] -1/-1/-1->0->-1 [19] -1/-1/-1->0->-1 [20] -1/-1/-1->0->-1 [21] -1/-1/-1->0->-1 [22] -1/-1/-1->0->-1 [23] -1/-1/-1->0->-1 [24] -1/-1/-1->0->-1 [25] -1/-1/-1->0->-1 [26] -1/-1/-1->0->-1 [27] -1/-1/-1->0->-1 [28] -1/-1/-1->0->-1 [29] -1/-1/-1->0->-1 [30] -1/-1/-1->0->-1 [31] -1/-1/-1->0->-1
+ip-26-0-169-207:162591:162900 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-169-207:162591:162900 [2] NCCL INFO Connected all rings
+ip-26-0-169-207:162591:162900 [2] NCCL INFO Connected all trees
+ip-26-0-169-207:162591:162900 [2] NCCL INFO 32 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:162593:162904 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-169-207:162593:162904 [4] NCCL INFO Channel 00/32 :    0
+ip-26-0-169-207:162593:162904 [4] NCCL INFO Channel 01/32 :    0
+ip-26-0-169-207:162593:162904 [4] NCCL INFO Channel 02/32 :    0
+ip-26-0-169-207:162593:162904 [4] NCCL INFO Channel 03/32 :    0
+ip-26-0-169-207:162593:162904 [4] NCCL INFO Channel 04/32 :    0
+ip-26-0-169-207:162593:162904 [4] NCCL INFO Channel 05/32 :    0
+ip-26-0-169-207:162593:162904 [4] NCCL INFO Channel 06/32 :    0
+ip-26-0-169-207:162593:162904 [4] NCCL INFO Channel 07/32 :    0
+ip-26-0-169-207:162593:162904 [4] NCCL INFO Channel 08/32 :    0
+ip-26-0-169-207:162593:162904 [4] NCCL INFO Channel 09/32 :    0
+ip-26-0-169-207:162593:162904 [4] NCCL INFO Channel 10/32 :    0
+ip-26-0-169-207:162593:162904 [4] NCCL INFO Channel 11/32 :    0
+ip-26-0-169-207:162593:162904 [4] NCCL INFO Channel 12/32 :    0
+ip-26-0-169-207:162593:162904 [4] NCCL INFO Channel 13/32 :    0
+ip-26-0-169-207:162593:162904 [4] NCCL INFO Channel 14/32 :    0
+ip-26-0-169-207:162593:162904 [4] NCCL INFO Channel 15/32 :    0
+ip-26-0-169-207:162593:162904 [4] NCCL INFO Channel 16/32 :    0
+ip-26-0-169-207:162593:162904 [4] NCCL INFO Channel 17/32 :    0
+ip-26-0-169-207:162593:162904 [4] NCCL INFO Channel 18/32 :    0
+ip-26-0-169-207:162593:162904 [4] NCCL INFO Channel 19/32 :    0
+ip-26-0-169-207:162593:162904 [4] NCCL INFO Channel 20/32 :    0
+ip-26-0-169-207:162593:162904 [4] NCCL INFO Channel 21/32 :    0
+ip-26-0-169-207:162593:162904 [4] NCCL INFO Channel 22/32 :    0
+ip-26-0-169-207:162593:162904 [4] NCCL INFO Channel 23/32 :    0
+ip-26-0-169-207:162593:162904 [4] NCCL INFO Channel 24/32 :    0
+ip-26-0-169-207:162593:162904 [4] NCCL INFO Channel 25/32 :    0
+ip-26-0-169-207:162593:162904 [4] NCCL INFO Channel 26/32 :    0
+ip-26-0-169-207:162593:162904 [4] NCCL INFO Channel 27/32 :    0
+ip-26-0-169-207:162593:162904 [4] NCCL INFO Channel 28/32 :    0
+ip-26-0-169-207:162593:162904 [4] NCCL INFO Channel 29/32 :    0
+ip-26-0-169-207:162593:162904 [4] NCCL INFO Channel 30/32 :    0
+ip-26-0-169-207:162593:162904 [4] NCCL INFO Channel 31/32 :    0
+ip-26-0-169-207:162593:162904 [4] NCCL INFO Trees [0] -1/-1/-1->0->-1 [1] -1/-1/-1->0->-1 [2] -1/-1/-1->0->-1 [3] -1/-1/-1->0->-1 [4] -1/-1/-1->0->-1 [5] -1/-1/-1->0->-1 [6] -1/-1/-1->0->-1 [7] -1/-1/-1->0->-1 [8] -1/-1/-1->0->-1 [9] -1/-1/-1->0->-1 [10] -1/-1/-1->0->-1 [11] -1/-1/-1->0->-1 [12] -1/-1/-1->0->-1 [13] -1/-1/-1->0->-1 [14] -1/-1/-1->0->-1 [15] -1/-1/-1->0->-1 [16] -1/-1/-1->0->-1 [17] -1/-1/-1->0->-1 [18] -1/-1/-1->0->-1 [19] -1/-1/-1->0->-1 [20] -1/-1/-1->0->-1 [21] -1/-1/-1->0->-1 [22] -1/-1/-1->0->-1 [23] -1/-1/-1->0->-1 [24] -1/-1/-1->0->-1 [25] -1/-1/-1->0->-1 [26] -1/-1/-1->0->-1 [27] -1/-1/-1->0->-1 [28] -1/-1/-1->0->-1 [29] -1/-1/-1->0->-1 [30] -1/-1/-1->0->-1 [31] -1/-1/-1->0->-1
+ip-26-0-169-207:162593:162904 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-169-207:162593:162904 [4] NCCL INFO Connected all rings
+ip-26-0-169-207:162593:162904 [4] NCCL INFO Connected all trees
+ip-26-0-169-207:162593:162904 [4] NCCL INFO 32 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:162594:162907 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-169-207:162594:162907 [5] NCCL INFO Channel 00/32 :    0
+ip-26-0-169-207:162594:162907 [5] NCCL INFO Channel 01/32 :    0
+ip-26-0-169-207:162594:162907 [5] NCCL INFO Channel 02/32 :    0
+ip-26-0-169-207:162594:162907 [5] NCCL INFO Channel 03/32 :    0
+ip-26-0-169-207:162594:162907 [5] NCCL INFO Channel 04/32 :    0
+ip-26-0-169-207:162594:162907 [5] NCCL INFO Channel 05/32 :    0
+ip-26-0-169-207:162594:162907 [5] NCCL INFO Channel 06/32 :    0
+ip-26-0-169-207:162594:162907 [5] NCCL INFO Channel 07/32 :    0
+ip-26-0-169-207:162594:162907 [5] NCCL INFO Channel 08/32 :    0
+ip-26-0-169-207:162594:162907 [5] NCCL INFO Channel 09/32 :    0
+ip-26-0-169-207:162594:162907 [5] NCCL INFO Channel 10/32 :    0
+ip-26-0-169-207:162594:162907 [5] NCCL INFO Channel 11/32 :    0
+ip-26-0-169-207:162594:162907 [5] NCCL INFO Channel 12/32 :    0
+ip-26-0-169-207:162594:162907 [5] NCCL INFO Channel 13/32 :    0
+ip-26-0-169-207:162594:162907 [5] NCCL INFO Channel 14/32 :    0
+ip-26-0-169-207:162594:162907 [5] NCCL INFO Channel 15/32 :    0
+ip-26-0-169-207:162594:162907 [5] NCCL INFO Channel 16/32 :    0
+ip-26-0-169-207:162594:162907 [5] NCCL INFO Channel 17/32 :    0
+ip-26-0-169-207:162594:162907 [5] NCCL INFO Channel 18/32 :    0
+ip-26-0-169-207:162594:162907 [5] NCCL INFO Channel 19/32 :    0
+ip-26-0-169-207:162594:162907 [5] NCCL INFO Channel 20/32 :    0
+ip-26-0-169-207:162594:162907 [5] NCCL INFO Channel 21/32 :    0
+ip-26-0-169-207:162594:162907 [5] NCCL INFO Channel 22/32 :    0
+ip-26-0-169-207:162594:162907 [5] NCCL INFO Channel 23/32 :    0
+ip-26-0-169-207:162594:162907 [5] NCCL INFO Channel 24/32 :    0
+ip-26-0-169-207:162594:162907 [5] NCCL INFO Channel 25/32 :    0
+ip-26-0-169-207:162594:162907 [5] NCCL INFO Channel 26/32 :    0
+ip-26-0-169-207:162594:162907 [5] NCCL INFO Channel 27/32 :    0
+ip-26-0-169-207:162594:162907 [5] NCCL INFO Channel 28/32 :    0
+ip-26-0-169-207:162594:162907 [5] NCCL INFO Channel 29/32 :    0
+ip-26-0-169-207:162594:162907 [5] NCCL INFO Channel 30/32 :    0
+ip-26-0-169-207:162594:162907 [5] NCCL INFO Channel 31/32 :    0
+ip-26-0-169-207:162594:162907 [5] NCCL INFO Trees [0] -1/-1/-1->0->-1 [1] -1/-1/-1->0->-1 [2] -1/-1/-1->0->-1 [3] -1/-1/-1->0->-1 [4] -1/-1/-1->0->-1 [5] -1/-1/-1->0->-1 [6] -1/-1/-1->0->-1 [7] -1/-1/-1->0->-1 [8] -1/-1/-1->0->-1 [9] -1/-1/-1->0->-1 [10] -1/-1/-1->0->-1 [11] -1/-1/-1->0->-1 [12] -1/-1/-1->0->-1 [13] -1/-1/-1->0->-1 [14] -1/-1/-1->0->-1 [15] -1/-1/-1->0->-1 [16] -1/-1/-1->0->-1 [17] -1/-1/-1->0->-1 [18] -1/-1/-1->0->-1 [19] -1/-1/-1->0->-1 [20] -1/-1/-1->0->-1 [21] -1/-1/-1->0->-1 [22] -1/-1/-1->0->-1 [23] -1/-1/-1->0->-1 [24] -1/-1/-1->0->-1 [25] -1/-1/-1->0->-1 [26] -1/-1/-1->0->-1 [27] -1/-1/-1->0->-1 [28] -1/-1/-1->0->-1 [29] -1/-1/-1->0->-1 [30] -1/-1/-1->0->-1 [31] -1/-1/-1->0->-1
+ip-26-0-169-207:162594:162907 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-169-207:162594:162907 [5] NCCL INFO Connected all rings
+ip-26-0-169-207:162594:162907 [5] NCCL INFO Connected all trees
+ip-26-0-169-207:162594:162907 [5] NCCL INFO 32 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:162592:162902 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-169-207:162592:162902 [3] NCCL INFO Channel 00/32 :    0
+ip-26-0-169-207:162592:162902 [3] NCCL INFO Channel 01/32 :    0
+ip-26-0-169-207:162592:162902 [3] NCCL INFO Channel 02/32 :    0
+ip-26-0-169-207:162592:162902 [3] NCCL INFO Channel 03/32 :    0
+ip-26-0-169-207:162592:162902 [3] NCCL INFO Channel 04/32 :    0
+ip-26-0-169-207:162592:162902 [3] NCCL INFO Channel 05/32 :    0
+ip-26-0-169-207:162592:162902 [3] NCCL INFO Channel 06/32 :    0
+ip-26-0-169-207:162592:162902 [3] NCCL INFO Channel 07/32 :    0
+ip-26-0-169-207:162592:162902 [3] NCCL INFO Channel 08/32 :    0
+ip-26-0-169-207:162592:162902 [3] NCCL INFO Channel 09/32 :    0
+ip-26-0-169-207:162592:162902 [3] NCCL INFO Channel 10/32 :    0
+ip-26-0-169-207:162592:162902 [3] NCCL INFO Channel 11/32 :    0
+ip-26-0-169-207:162592:162902 [3] NCCL INFO Channel 12/32 :    0
+ip-26-0-169-207:162592:162902 [3] NCCL INFO Channel 13/32 :    0
+ip-26-0-169-207:162592:162902 [3] NCCL INFO Channel 14/32 :    0
+ip-26-0-169-207:162592:162902 [3] NCCL INFO Channel 15/32 :    0
+ip-26-0-169-207:162592:162902 [3] NCCL INFO Channel 16/32 :    0
+ip-26-0-169-207:162592:162902 [3] NCCL INFO Channel 17/32 :    0
+ip-26-0-169-207:162592:162902 [3] NCCL INFO Channel 18/32 :    0
+ip-26-0-169-207:162592:162902 [3] NCCL INFO Channel 19/32 :    0
+ip-26-0-169-207:162592:162902 [3] NCCL INFO Channel 20/32 :    0
+ip-26-0-169-207:162592:162902 [3] NCCL INFO Channel 21/32 :    0
+ip-26-0-169-207:162592:162902 [3] NCCL INFO Channel 22/32 :    0
+ip-26-0-169-207:162592:162902 [3] NCCL INFO Channel 23/32 :    0
+ip-26-0-169-207:162592:162902 [3] NCCL INFO Channel 24/32 :    0
+ip-26-0-169-207:162592:162902 [3] NCCL INFO Channel 25/32 :    0
+ip-26-0-169-207:162592:162902 [3] NCCL INFO Channel 26/32 :    0
+ip-26-0-169-207:162592:162902 [3] NCCL INFO Channel 27/32 :    0
+ip-26-0-169-207:162592:162902 [3] NCCL INFO Channel 28/32 :    0
+ip-26-0-169-207:162592:162902 [3] NCCL INFO Channel 29/32 :    0
+ip-26-0-169-207:162592:162902 [3] NCCL INFO Channel 30/32 :    0
+ip-26-0-169-207:162592:162902 [3] NCCL INFO Channel 31/32 :    0
+ip-26-0-169-207:162592:162902 [3] NCCL INFO Trees [0] -1/-1/-1->0->-1 [1] -1/-1/-1->0->-1 [2] -1/-1/-1->0->-1 [3] -1/-1/-1->0->-1 [4] -1/-1/-1->0->-1 [5] -1/-1/-1->0->-1 [6] -1/-1/-1->0->-1 [7] -1/-1/-1->0->-1 [8] -1/-1/-1->0->-1 [9] -1/-1/-1->0->-1 [10] -1/-1/-1->0->-1 [11] -1/-1/-1->0->-1 [12] -1/-1/-1->0->-1 [13] -1/-1/-1->0->-1 [14] -1/-1/-1->0->-1 [15] -1/-1/-1->0->-1 [16] -1/-1/-1->0->-1 [17] -1/-1/-1->0->-1 [18] -1/-1/-1->0->-1 [19] -1/-1/-1->0->-1 [20] -1/-1/-1->0->-1 [21] -1/-1/-1->0->-1 [22] -1/-1/-1->0->-1 [23] -1/-1/-1->0->-1 [24] -1/-1/-1->0->-1 [25] -1/-1/-1->0->-1 [26] -1/-1/-1->0->-1 [27] -1/-1/-1->0->-1 [28] -1/-1/-1->0->-1 [29] -1/-1/-1->0->-1 [30] -1/-1/-1->0->-1 [31] -1/-1/-1->0->-1
+ip-26-0-169-207:162592:162902 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-169-207:162592:162902 [3] NCCL INFO Connected all rings
+ip-26-0-169-207:162592:162902 [3] NCCL INFO Connected all trees
+ip-26-0-169-207:162592:162902 [3] NCCL INFO 32 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:162591:162900 [2] NCCL INFO comm 0xa8e6060 rank 0 nranks 1 cudaDev 2 nvmlDev 2 busId 75000 commId 0xd5e80c39810b46c5 - Init COMPLETE
+ip-26-0-169-207:162596:162911 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-169-207:162596:162911 [7] NCCL INFO Channel 00/32 :    0
+ip-26-0-169-207:162596:162911 [7] NCCL INFO Channel 01/32 :    0
+ip-26-0-169-207:162596:162911 [7] NCCL INFO Channel 02/32 :    0
+ip-26-0-169-207:162596:162911 [7] NCCL INFO Channel 03/32 :    0
+ip-26-0-169-207:162596:162911 [7] NCCL INFO Channel 04/32 :    0
+ip-26-0-169-207:162596:162911 [7] NCCL INFO Channel 05/32 :    0
+ip-26-0-169-207:162596:162911 [7] NCCL INFO Channel 06/32 :    0
+ip-26-0-169-207:162596:162911 [7] NCCL INFO Channel 07/32 :    0
+ip-26-0-169-207:162596:162911 [7] NCCL INFO Channel 08/32 :    0
+ip-26-0-169-207:162596:162911 [7] NCCL INFO Channel 09/32 :    0
+ip-26-0-169-207:162596:162911 [7] NCCL INFO Channel 10/32 :    0
+ip-26-0-169-207:162596:162911 [7] NCCL INFO Channel 11/32 :    0
+ip-26-0-169-207:162596:162911 [7] NCCL INFO Channel 12/32 :    0
+ip-26-0-169-207:162596:162911 [7] NCCL INFO Channel 13/32 :    0
+ip-26-0-169-207:162596:162911 [7] NCCL INFO Channel 14/32 :    0
+ip-26-0-169-207:162596:162911 [7] NCCL INFO Channel 15/32 :    0
+ip-26-0-169-207:162596:162911 [7] NCCL INFO Channel 16/32 :    0
+ip-26-0-169-207:162596:162911 [7] NCCL INFO Channel 17/32 :    0
+ip-26-0-169-207:162596:162911 [7] NCCL INFO Channel 18/32 :    0
+ip-26-0-169-207:162596:162911 [7] NCCL INFO Channel 19/32 :    0
+ip-26-0-169-207:162596:162911 [7] NCCL INFO Channel 20/32 :    0
+ip-26-0-169-207:162596:162911 [7] NCCL INFO Channel 21/32 :    0
+ip-26-0-169-207:162596:162911 [7] NCCL INFO Channel 22/32 :    0
+ip-26-0-169-207:162596:162911 [7] NCCL INFO Channel 23/32 :    0
+ip-26-0-169-207:162596:162911 [7] NCCL INFO Channel 24/32 :    0
+ip-26-0-169-207:162596:162911 [7] NCCL INFO Channel 25/32 :    0
+ip-26-0-169-207:162596:162911 [7] NCCL INFO Channel 26/32 :    0
+ip-26-0-169-207:162596:162911 [7] NCCL INFO Channel 27/32 :    0
+ip-26-0-169-207:162596:162911 [7] NCCL INFO Channel 28/32 :    0
+ip-26-0-169-207:162596:162911 [7] NCCL INFO Channel 29/32 :    0
+ip-26-0-169-207:162596:162911 [7] NCCL INFO Channel 30/32 :    0
+ip-26-0-169-207:162596:162911 [7] NCCL INFO Channel 31/32 :    0
+ip-26-0-169-207:162596:162911 [7] NCCL INFO Trees [0] -1/-1/-1->0->-1 [1] -1/-1/-1->0->-1 [2] -1/-1/-1->0->-1 [3] -1/-1/-1->0->-1 [4] -1/-1/-1->0->-1 [5] -1/-1/-1->0->-1 [6] -1/-1/-1->0->-1 [7] -1/-1/-1->0->-1 [8] -1/-1/-1->0->-1 [9] -1/-1/-1->0->-1 [10] -1/-1/-1->0->-1 [11] -1/-1/-1->0->-1 [12] -1/-1/-1->0->-1 [13] -1/-1/-1->0->-1 [14] -1/-1/-1->0->-1 [15] -1/-1/-1->0->-1 [16] -1/-1/-1->0->-1 [17] -1/-1/-1->0->-1 [18] -1/-1/-1->0->-1 [19] -1/-1/-1->0->-1 [20] -1/-1/-1->0->-1 [21] -1/-1/-1->0->-1 [22] -1/-1/-1->0->-1 [23] -1/-1/-1->0->-1 [24] -1/-1/-1->0->-1 [25] -1/-1/-1->0->-1 [26] -1/-1/-1->0->-1 [27] -1/-1/-1->0->-1 [28] -1/-1/-1->0->-1 [29] -1/-1/-1->0->-1 [30] -1/-1/-1->0->-1 [31] -1/-1/-1->0->-1
+ip-26-0-169-207:162596:162911 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-169-207:162596:162911 [7] NCCL INFO Connected all rings
+ip-26-0-169-207:162596:162911 [7] NCCL INFO Connected all trees
+ip-26-0-169-207:162596:162911 [7] NCCL INFO 32 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:162595:162909 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-169-207:162595:162909 [6] NCCL INFO Channel 00/32 :    0
+ip-26-0-169-207:162595:162909 [6] NCCL INFO Channel 01/32 :    0
+ip-26-0-169-207:162595:162909 [6] NCCL INFO Channel 02/32 :    0
+ip-26-0-169-207:162595:162909 [6] NCCL INFO Channel 03/32 :    0
+ip-26-0-169-207:162595:162909 [6] NCCL INFO Channel 04/32 :    0
+ip-26-0-169-207:162595:162909 [6] NCCL INFO Channel 05/32 :    0
+ip-26-0-169-207:162595:162909 [6] NCCL INFO Channel 06/32 :    0
+ip-26-0-169-207:162595:162909 [6] NCCL INFO Channel 07/32 :    0
+ip-26-0-169-207:162595:162909 [6] NCCL INFO Channel 08/32 :    0
+ip-26-0-169-207:162595:162909 [6] NCCL INFO Channel 09/32 :    0
+ip-26-0-169-207:162595:162909 [6] NCCL INFO Channel 10/32 :    0
+ip-26-0-169-207:162595:162909 [6] NCCL INFO Channel 11/32 :    0
+ip-26-0-169-207:162595:162909 [6] NCCL INFO Channel 12/32 :    0
+ip-26-0-169-207:162595:162909 [6] NCCL INFO Channel 13/32 :    0
+ip-26-0-169-207:162595:162909 [6] NCCL INFO Channel 14/32 :    0
+ip-26-0-169-207:162595:162909 [6] NCCL INFO Channel 15/32 :    0
+ip-26-0-169-207:162595:162909 [6] NCCL INFO Channel 16/32 :    0
+ip-26-0-169-207:162595:162909 [6] NCCL INFO Channel 17/32 :    0
+ip-26-0-169-207:162595:162909 [6] NCCL INFO Channel 18/32 :    0
+ip-26-0-169-207:162595:162909 [6] NCCL INFO Channel 19/32 :    0
+ip-26-0-169-207:162595:162909 [6] NCCL INFO Channel 20/32 :    0
+ip-26-0-169-207:162595:162909 [6] NCCL INFO Channel 21/32 :    0
+ip-26-0-169-207:162595:162909 [6] NCCL INFO Channel 22/32 :    0
+ip-26-0-169-207:162595:162909 [6] NCCL INFO Channel 23/32 :    0
+ip-26-0-169-207:162595:162909 [6] NCCL INFO Channel 24/32 :    0
+ip-26-0-169-207:162595:162909 [6] NCCL INFO Channel 25/32 :    0
+ip-26-0-169-207:162595:162909 [6] NCCL INFO Channel 26/32 :    0
+ip-26-0-169-207:162595:162909 [6] NCCL INFO Channel 27/32 :    0
+ip-26-0-169-207:162595:162909 [6] NCCL INFO Channel 28/32 :    0
+ip-26-0-169-207:162595:162909 [6] NCCL INFO Channel 29/32 :    0
+ip-26-0-169-207:162595:162909 [6] NCCL INFO Channel 30/32 :    0
+ip-26-0-169-207:162595:162909 [6] NCCL INFO Channel 31/32 :    0
+ip-26-0-169-207:162595:162909 [6] NCCL INFO Trees [0] -1/-1/-1->0->-1 [1] -1/-1/-1->0->-1 [2] -1/-1/-1->0->-1 [3] -1/-1/-1->0->-1 [4] -1/-1/-1->0->-1 [5] -1/-1/-1->0->-1 [6] -1/-1/-1->0->-1 [7] -1/-1/-1->0->-1 [8] -1/-1/-1->0->-1 [9] -1/-1/-1->0->-1 [10] -1/-1/-1->0->-1 [11] -1/-1/-1->0->-1 [12] -1/-1/-1->0->-1 [13] -1/-1/-1->0->-1 [14] -1/-1/-1->0->-1 [15] -1/-1/-1->0->-1 [16] -1/-1/-1->0->-1 [17] -1/-1/-1->0->-1 [18] -1/-1/-1->0->-1 [19] -1/-1/-1->0->-1 [20] -1/-1/-1->0->-1 [21] -1/-1/-1->0->-1 [22] -1/-1/-1->0->-1 [23] -1/-1/-1->0->-1 [24] -1/-1/-1->0->-1 [25] -1/-1/-1->0->-1 [26] -1/-1/-1->0->-1 [27] -1/-1/-1->0->-1 [28] -1/-1/-1->0->-1 [29] -1/-1/-1->0->-1 [30] -1/-1/-1->0->-1 [31] -1/-1/-1->0->-1
+ip-26-0-169-207:162595:162909 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-169-207:162595:162909 [6] NCCL INFO Connected all rings
+ip-26-0-169-207:162595:162909 [6] NCCL INFO Connected all trees
+ip-26-0-169-207:162595:162909 [6] NCCL INFO 32 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:162593:162904 [4] NCCL INFO comm 0xbc3c8e0 rank 0 nranks 1 cudaDev 4 nvmlDev 4 busId 97000 commId 0x279428a0c5d2f3aa - Init COMPLETE
+ip-26-0-169-207:162592:162902 [3] NCCL INFO comm 0xc47ddb0 rank 0 nranks 1 cudaDev 3 nvmlDev 3 busId 86000 commId 0x55552d6a1be0dbdd - Init COMPLETE
+ip-26-0-169-207:162594:162907 [5] NCCL INFO comm 0xb315d30 rank 0 nranks 1 cudaDev 5 nvmlDev 5 busId a8000 commId 0x70662a167e82cbc0 - Init COMPLETE
+ip-26-0-169-207:162590:162914 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-169-207:162590:162914 [1] NCCL INFO Channel 00/32 :    0
+ip-26-0-169-207:162590:162914 [1] NCCL INFO Channel 01/32 :    0
+ip-26-0-169-207:162590:162914 [1] NCCL INFO Channel 02/32 :    0
+ip-26-0-169-207:162590:162914 [1] NCCL INFO Channel 03/32 :    0
+ip-26-0-169-207:162590:162914 [1] NCCL INFO Channel 04/32 :    0
+ip-26-0-169-207:162590:162914 [1] NCCL INFO Channel 05/32 :    0
+ip-26-0-169-207:162590:162914 [1] NCCL INFO Channel 06/32 :    0
+ip-26-0-169-207:162590:162914 [1] NCCL INFO Channel 07/32 :    0
+ip-26-0-169-207:162590:162914 [1] NCCL INFO Channel 08/32 :    0
+ip-26-0-169-207:162590:162914 [1] NCCL INFO Channel 09/32 :    0
+ip-26-0-169-207:162590:162914 [1] NCCL INFO Channel 10/32 :    0
+ip-26-0-169-207:162590:162914 [1] NCCL INFO Channel 11/32 :    0
+ip-26-0-169-207:162590:162914 [1] NCCL INFO Channel 12/32 :    0
+ip-26-0-169-207:162590:162914 [1] NCCL INFO Channel 13/32 :    0
+ip-26-0-169-207:162590:162914 [1] NCCL INFO Channel 14/32 :    0
+ip-26-0-169-207:162590:162914 [1] NCCL INFO Channel 15/32 :    0
+ip-26-0-169-207:162590:162914 [1] NCCL INFO Channel 16/32 :    0
+ip-26-0-169-207:162590:162914 [1] NCCL INFO Channel 17/32 :    0
+ip-26-0-169-207:162590:162914 [1] NCCL INFO Channel 18/32 :    0
+ip-26-0-169-207:162590:162914 [1] NCCL INFO Channel 19/32 :    0
+ip-26-0-169-207:162590:162914 [1] NCCL INFO Channel 20/32 :    0
+ip-26-0-169-207:162590:162914 [1] NCCL INFO Channel 21/32 :    0
+ip-26-0-169-207:162590:162914 [1] NCCL INFO Channel 22/32 :    0
+ip-26-0-169-207:162590:162914 [1] NCCL INFO Channel 23/32 :    0
+ip-26-0-169-207:162590:162914 [1] NCCL INFO Channel 24/32 :    0
+ip-26-0-169-207:162590:162914 [1] NCCL INFO Channel 25/32 :    0
+ip-26-0-169-207:162590:162914 [1] NCCL INFO Channel 26/32 :    0
+ip-26-0-169-207:162590:162914 [1] NCCL INFO Channel 27/32 :    0
+ip-26-0-169-207:162590:162914 [1] NCCL INFO Channel 28/32 :    0
+ip-26-0-169-207:162590:162914 [1] NCCL INFO Channel 29/32 :    0
+ip-26-0-169-207:162590:162914 [1] NCCL INFO Channel 30/32 :    0
+ip-26-0-169-207:162590:162914 [1] NCCL INFO Channel 31/32 :    0
+ip-26-0-169-207:162590:162914 [1] NCCL INFO Trees [0] -1/-1/-1->0->-1 [1] -1/-1/-1->0->-1 [2] -1/-1/-1->0->-1 [3] -1/-1/-1->0->-1 [4] -1/-1/-1->0->-1 [5] -1/-1/-1->0->-1 [6] -1/-1/-1->0->-1 [7] -1/-1/-1->0->-1 [8] -1/-1/-1->0->-1 [9] -1/-1/-1->0->-1 [10] -1/-1/-1->0->-1 [11] -1/-1/-1->0->-1 [12] -1/-1/-1->0->-1 [13] -1/-1/-1->0->-1 [14] -1/-1/-1->0->-1 [15] -1/-1/-1->0->-1 [16] -1/-1/-1->0->-1 [17] -1/-1/-1->0->-1 [18] -1/-1/-1->0->-1 [19] -1/-1/-1->0->-1 [20] -1/-1/-1->0->-1 [21] -1/-1/-1->0->-1 [22] -1/-1/-1->0->-1 [23] -1/-1/-1->0->-1 [24] -1/-1/-1->0->-1 [25] -1/-1/-1->0->-1 [26] -1/-1/-1->0->-1 [27] -1/-1/-1->0->-1 [28] -1/-1/-1->0->-1 [29] -1/-1/-1->0->-1 [30] -1/-1/-1->0->-1 [31] -1/-1/-1->0->-1
+ip-26-0-169-207:162590:162914 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-169-207:162590:162914 [1] NCCL INFO Connected all rings
+ip-26-0-169-207:162590:162914 [1] NCCL INFO Connected all trees
+ip-26-0-169-207:162590:162914 [1] NCCL INFO 32 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:162589:162916 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-169-207:162589:162916 [0] NCCL INFO Channel 00/32 :    0
+ip-26-0-169-207:162589:162916 [0] NCCL INFO Channel 01/32 :    0
+ip-26-0-169-207:162589:162916 [0] NCCL INFO Channel 02/32 :    0
+ip-26-0-169-207:162589:162916 [0] NCCL INFO Channel 03/32 :    0
+ip-26-0-169-207:162589:162916 [0] NCCL INFO Channel 04/32 :    0
+ip-26-0-169-207:162589:162916 [0] NCCL INFO Channel 05/32 :    0
+ip-26-0-169-207:162589:162916 [0] NCCL INFO Channel 06/32 :    0
+ip-26-0-169-207:162589:162916 [0] NCCL INFO Channel 07/32 :    0
+ip-26-0-169-207:162589:162916 [0] NCCL INFO Channel 08/32 :    0
+ip-26-0-169-207:162589:162916 [0] NCCL INFO Channel 09/32 :    0
+ip-26-0-169-207:162589:162916 [0] NCCL INFO Channel 10/32 :    0
+ip-26-0-169-207:162589:162916 [0] NCCL INFO Channel 11/32 :    0
+ip-26-0-169-207:162589:162916 [0] NCCL INFO Channel 12/32 :    0
+ip-26-0-169-207:162589:162916 [0] NCCL INFO Channel 13/32 :    0
+ip-26-0-169-207:162589:162916 [0] NCCL INFO Channel 14/32 :    0
+ip-26-0-169-207:162589:162916 [0] NCCL INFO Channel 15/32 :    0
+ip-26-0-169-207:162589:162916 [0] NCCL INFO Channel 16/32 :    0
+ip-26-0-169-207:162589:162916 [0] NCCL INFO Channel 17/32 :    0
+ip-26-0-169-207:162589:162916 [0] NCCL INFO Channel 18/32 :    0
+ip-26-0-169-207:162589:162916 [0] NCCL INFO Channel 19/32 :    0
+ip-26-0-169-207:162589:162916 [0] NCCL INFO Channel 20/32 :    0
+ip-26-0-169-207:162589:162916 [0] NCCL INFO Channel 21/32 :    0
+ip-26-0-169-207:162589:162916 [0] NCCL INFO Channel 22/32 :    0
+ip-26-0-169-207:162589:162916 [0] NCCL INFO Channel 23/32 :    0
+ip-26-0-169-207:162589:162916 [0] NCCL INFO Channel 24/32 :    0
+ip-26-0-169-207:162589:162916 [0] NCCL INFO Channel 25/32 :    0
+ip-26-0-169-207:162589:162916 [0] NCCL INFO Channel 26/32 :    0
+ip-26-0-169-207:162589:162916 [0] NCCL INFO Channel 27/32 :    0
+ip-26-0-169-207:162589:162916 [0] NCCL INFO Channel 28/32 :    0
+ip-26-0-169-207:162589:162916 [0] NCCL INFO Channel 29/32 :    0
+ip-26-0-169-207:162589:162916 [0] NCCL INFO Channel 30/32 :    0
+ip-26-0-169-207:162589:162916 [0] NCCL INFO Channel 31/32 :    0
+ip-26-0-169-207:162589:162916 [0] NCCL INFO Trees [0] -1/-1/-1->0->-1 [1] -1/-1/-1->0->-1 [2] -1/-1/-1->0->-1 [3] -1/-1/-1->0->-1 [4] -1/-1/-1->0->-1 [5] -1/-1/-1->0->-1 [6] -1/-1/-1->0->-1 [7] -1/-1/-1->0->-1 [8] -1/-1/-1->0->-1 [9] -1/-1/-1->0->-1 [10] -1/-1/-1->0->-1 [11] -1/-1/-1->0->-1 [12] -1/-1/-1->0->-1 [13] -1/-1/-1->0->-1 [14] -1/-1/-1->0->-1 [15] -1/-1/-1->0->-1 [16] -1/-1/-1->0->-1 [17] -1/-1/-1->0->-1 [18] -1/-1/-1->0->-1 [19] -1/-1/-1->0->-1 [20] -1/-1/-1->0->-1 [21] -1/-1/-1->0->-1 [22] -1/-1/-1->0->-1 [23] -1/-1/-1->0->-1 [24] -1/-1/-1->0->-1 [25] -1/-1/-1->0->-1 [26] -1/-1/-1->0->-1 [27] -1/-1/-1->0->-1 [28] -1/-1/-1->0->-1 [29] -1/-1/-1->0->-1 [30] -1/-1/-1->0->-1 [31] -1/-1/-1->0->-1
+ip-26-0-169-207:162589:162916 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-169-207:162589:162916 [0] NCCL INFO Connected all rings
+ip-26-0-169-207:162589:162916 [0] NCCL INFO Connected all trees
+ip-26-0-169-207:162589:162916 [0] NCCL INFO 32 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:162596:162911 [7] NCCL INFO comm 0xa76acf0 rank 0 nranks 1 cudaDev 7 nvmlDev 7 busId ca000 commId 0x6c5d2b4ad9a41899 - Init COMPLETE
+ip-26-0-169-207:162595:162909 [6] NCCL INFO comm 0xb137670 rank 0 nranks 1 cudaDev 6 nvmlDev 6 busId b9000 commId 0x8d4317fcab0a9d58 - Init COMPLETE
+ip-26-0-169-207:162590:162914 [1] NCCL INFO comm 0xaaa4a30 rank 0 nranks 1 cudaDev 1 nvmlDev 1 busId 64000 commId 0x24918949dfdbf18e - Init COMPLETE
+ip-26-0-169-207:162589:162916 [0] NCCL INFO comm 0xb36dde0 rank 0 nranks 1 cudaDev 0 nvmlDev 0 busId 53000 commId 0x5f1072d1d3fa1a2b - Init COMPLETE
+12/12/2024 21:38:49 [INFO|DP=0|PP=0|TP=1|ip-26-0-169-207]: Local number of parameters: 520M (992.13MiB)
+12/12/2024 21:38:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]: Total number of parameters: 1.04G (1984.26MiB)
+12/12/2024 21:38:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]: Local number of parameters: 520M (992.13MiB)
+12/12/2024 21:38:49 [INFO|DP=0|PP=0|TP=1|ip-26-0-169-207]: [After model building] Memory usage: 992.15MiB. Peak allocated: 5632.00MiB Peak reserved: 16898.00MiB
+12/12/2024 21:38:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]: [After model building] Memory usage: 992.15MiB. Peak allocated: 5632.00MiB Peak reserved: 16898.00MiB
+12/12/2024 21:38:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]: No checkpoint path provided.
+12/12/2024 21:38:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]: Parametrizing model parameters using StandardParametrizator
+ip-26-0-169-207:162595:162938 [6] NCCL INFO Using network Libfabric
+ip-26-0-169-207:162591:162940 [2] NCCL INFO Using network Libfabric
+ip-26-0-169-207:162593:162939 [4] NCCL INFO Using network Libfabric
+ip-26-0-169-207:162589:162937 [0] NCCL INFO Using network Libfabric
+ip-26-0-169-207:162590:162936 [1] NCCL INFO Using network Libfabric
+ip-26-0-169-207:162594:162943 [5] NCCL INFO Using network Libfabric
+ip-26-0-169-207:162596:162941 [7] NCCL INFO Using network Libfabric
+ip-26-0-169-207:162592:162942 [3] NCCL INFO Using network Libfabric
+ip-26-0-169-207:162595:162938 [6] NCCL INFO comm 0xb13a320 rank 3 nranks 4 cudaDev 6 nvmlDev 6 busId b9000 commId 0x8d184ed925e1cb1 - Init START
+ip-26-0-169-207:162589:162937 [0] NCCL INFO comm 0xb370a90 rank 0 nranks 4 cudaDev 0 nvmlDev 0 busId 53000 commId 0x8d184ed925e1cb1 - Init START
+ip-26-0-169-207:162591:162940 [2] NCCL INFO comm 0xa8e8d10 rank 1 nranks 4 cudaDev 2 nvmlDev 2 busId 75000 commId 0x8d184ed925e1cb1 - Init START
+ip-26-0-169-207:162593:162939 [4] NCCL INFO comm 0xbc3f590 rank 2 nranks 4 cudaDev 4 nvmlDev 4 busId 97000 commId 0x8d184ed925e1cb1 - Init START
+ip-26-0-169-207:162592:162942 [3] NCCL INFO comm 0xc480a60 rank 1 nranks 4 cudaDev 3 nvmlDev 3 busId 86000 commId 0x7d5c542bb93ccd3a - Init START
+ip-26-0-169-207:162590:162936 [1] NCCL INFO comm 0xaaa76e0 rank 0 nranks 4 cudaDev 1 nvmlDev 1 busId 64000 commId 0x7d5c542bb93ccd3a - Init START
+ip-26-0-169-207:162596:162941 [7] NCCL INFO comm 0xa76d9a0 rank 3 nranks 4 cudaDev 7 nvmlDev 7 busId ca000 commId 0x7d5c542bb93ccd3a - Init START
+ip-26-0-169-207:162594:162943 [5] NCCL INFO comm 0xb3189e0 rank 2 nranks 4 cudaDev 5 nvmlDev 5 busId a8000 commId 0x7d5c542bb93ccd3a - Init START
+ip-26-0-169-207:162595:162938 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/149
+ip-26-0-169-207:162591:162940 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/149
+ip-26-0-169-207:162593:162939 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/149
+ip-26-0-169-207:162589:162937 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/162
+ip-26-0-169-207:162592:162942 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/149
+ip-26-0-169-207:162594:162943 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/149
+ip-26-0-169-207:162596:162941 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/149
+ip-26-0-169-207:162590:162936 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/149
+ip-26-0-169-207:162589:162937 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-169-207:162589:162937 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-169-207:162595:162938 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-169-207:162595:162938 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-169-207:162590:162936 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-169-207:162590:162936 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-169-207:162591:162940 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-169-207:162591:162940 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-169-207:162593:162939 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-169-207:162593:162939 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-169-207:162596:162941 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-169-207:162596:162941 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-169-207:162593:162939 [4] NCCL INFO Trees [0] 3/-1/-1->2->1 [1] 3/-1/-1->2->1 [2] 3/-1/-1->2->1 [3] 3/-1/-1->2->1 [4] 3/-1/-1->2->1 [5] 3/-1/-1->2->1 [6] 3/-1/-1->2->1 [7] 3/-1/-1->2->1 [8] 3/-1/-1->2->1 [9] 3/-1/-1->2->1 [10] 3/-1/-1->2->1 [11] 3/-1/-1->2->1 [12] 3/-1/-1->2->1 [13] 3/-1/-1->2->1 [14] 3/-1/-1->2->1 [15] 3/-1/-1->2->1 [16] 3/-1/-1->2->1 [17] 3/-1/-1->2->1 [18] 3/-1/-1->2->1 [19] 3/-1/-1->2->1 [20] 3/-1/-1->2->1 [21] 3/-1/-1->2->1 [22] 3/-1/-1->2->1 [23] 3/-1/-1->2->1
+ip-26-0-169-207:162593:162939 [4] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-169-207:162591:162940 [2] NCCL INFO Trees [0] 2/-1/-1->1->0 [1] 2/-1/-1->1->0 [2] 2/-1/-1->1->0 [3] 2/-1/-1->1->0 [4] 2/-1/-1->1->0 [5] 2/-1/-1->1->0 [6] 2/-1/-1->1->0 [7] 2/-1/-1->1->0 [8] 2/-1/-1->1->0 [9] 2/-1/-1->1->0 [10] 2/-1/-1->1->0 [11] 2/-1/-1->1->0 [12] 2/-1/-1->1->0 [13] 2/-1/-1->1->0 [14] 2/-1/-1->1->0 [15] 2/-1/-1->1->0 [16] 2/-1/-1->1->0 [17] 2/-1/-1->1->0 [18] 2/-1/-1->1->0 [19] 2/-1/-1->1->0 [20] 2/-1/-1->1->0 [21] 2/-1/-1->1->0 [22] 2/-1/-1->1->0 [23] 2/-1/-1->1->0
+ip-26-0-169-207:162591:162940 [2] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-169-207:162589:162937 [0] NCCL INFO Channel 00/24 :    0   1   2   3
+ip-26-0-169-207:162589:162937 [0] NCCL INFO Channel 01/24 :    0   1   2   3
+ip-26-0-169-207:162589:162937 [0] NCCL INFO Channel 02/24 :    0   1   2   3
+ip-26-0-169-207:162595:162938 [6] NCCL INFO Trees [0] -1/-1/-1->3->2 [1] -1/-1/-1->3->2 [2] -1/-1/-1->3->2 [3] -1/-1/-1->3->2 [4] -1/-1/-1->3->2 [5] -1/-1/-1->3->2 [6] -1/-1/-1->3->2 [7] -1/-1/-1->3->2 [8] -1/-1/-1->3->2 [9] -1/-1/-1->3->2 [10] -1/-1/-1->3->2 [11] -1/-1/-1->3->2 [12] -1/-1/-1->3->2 [13] -1/-1/-1->3->2 [14] -1/-1/-1->3->2 [15] -1/-1/-1->3->2 [16] -1/-1/-1->3->2 [17] -1/-1/-1->3->2 [18] -1/-1/-1->3->2 [19] -1/-1/-1->3->2 [20] -1/-1/-1->3->2 [21] -1/-1/-1->3->2 [22] -1/-1/-1->3->2 [23] -1/-1/-1->3->2
+ip-26-0-169-207:162589:162937 [0] NCCL INFO Channel 03/24 :    0   1   2   3
+ip-26-0-169-207:162589:162937 [0] NCCL INFO Channel 04/24 :    0   1   2   3
+ip-26-0-169-207:162595:162938 [6] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-169-207:162589:162937 [0] NCCL INFO Channel 05/24 :    0   1   2   3
+ip-26-0-169-207:162589:162937 [0] NCCL INFO Channel 06/24 :    0   1   2   3
+ip-26-0-169-207:162589:162937 [0] NCCL INFO Channel 07/24 :    0   1   2   3
+ip-26-0-169-207:162589:162937 [0] NCCL INFO Channel 08/24 :    0   1   2   3
+ip-26-0-169-207:162589:162937 [0] NCCL INFO Channel 09/24 :    0   1   2   3
+ip-26-0-169-207:162589:162937 [0] NCCL INFO Channel 10/24 :    0   1   2   3
+ip-26-0-169-207:162589:162937 [0] NCCL INFO Channel 11/24 :    0   1   2   3
+ip-26-0-169-207:162589:162937 [0] NCCL INFO Channel 12/24 :    0   1   2   3
+ip-26-0-169-207:162589:162937 [0] NCCL INFO Channel 13/24 :    0   1   2   3
+ip-26-0-169-207:162589:162937 [0] NCCL INFO Channel 14/24 :    0   1   2   3
+ip-26-0-169-207:162589:162937 [0] NCCL INFO Channel 15/24 :    0   1   2   3
+ip-26-0-169-207:162589:162937 [0] NCCL INFO Channel 16/24 :    0   1   2   3
+ip-26-0-169-207:162589:162937 [0] NCCL INFO Channel 17/24 :    0   1   2   3
+ip-26-0-169-207:162589:162937 [0] NCCL INFO Channel 18/24 :    0   1   2   3
+ip-26-0-169-207:162589:162937 [0] NCCL INFO Channel 19/24 :    0   1   2   3
+ip-26-0-169-207:162589:162937 [0] NCCL INFO Channel 20/24 :    0   1   2   3
+ip-26-0-169-207:162589:162937 [0] NCCL INFO Channel 21/24 :    0   1   2   3
+ip-26-0-169-207:162589:162937 [0] NCCL INFO Channel 22/24 :    0   1   2   3
+ip-26-0-169-207:162589:162937 [0] NCCL INFO Channel 23/24 :    0   1   2   3
+ip-26-0-169-207:162589:162937 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] 1/-1/-1->0->-1 [7] 1/-1/-1->0->-1 [8] 1/-1/-1->0->-1 [9] 1/-1/-1->0->-1 [10] 1/-1/-1->0->-1 [11] 1/-1/-1->0->-1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] 1/-1/-1->0->-1 [19] 1/-1/-1->0->-1 [20] 1/-1/-1->0->-1 [21] 1/-1/-1->0->-1 [22] 1/-1/-1->0->-1 [23] 1/-1/-1->0->-1
+ip-26-0-169-207:162589:162937 [0] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-169-207:162594:162943 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-169-207:162594:162943 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-169-207:162592:162942 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-169-207:162592:162942 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-169-207:162591:162940 [2] NCCL INFO Channel 00/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-169-207:162593:162939 [4] NCCL INFO Channel 00/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-169-207:162591:162940 [2] NCCL INFO Channel 01/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-169-207:162593:162939 [4] NCCL INFO Channel 01/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-169-207:162595:162938 [6] NCCL INFO Channel 00/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162591:162940 [2] NCCL INFO Channel 02/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-169-207:162595:162938 [6] NCCL INFO Channel 01/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162593:162939 [4] NCCL INFO Channel 02/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-169-207:162589:162937 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-169-207:162591:162940 [2] NCCL INFO Channel 03/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-169-207:162592:162942 [3] NCCL INFO Trees [0] 2/-1/-1->1->0 [1] 2/-1/-1->1->0 [2] 2/-1/-1->1->0 [3] 2/-1/-1->1->0 [4] 2/-1/-1->1->0 [5] 2/-1/-1->1->0 [6] 2/-1/-1->1->0 [7] 2/-1/-1->1->0 [8] 2/-1/-1->1->0 [9] 2/-1/-1->1->0 [10] 2/-1/-1->1->0 [11] 2/-1/-1->1->0 [12] 2/-1/-1->1->0 [13] 2/-1/-1->1->0 [14] 2/-1/-1->1->0 [15] 2/-1/-1->1->0 [16] 2/-1/-1->1->0 [17] 2/-1/-1->1->0 [18] 2/-1/-1->1->0 [19] 2/-1/-1->1->0 [20] 2/-1/-1->1->0 [21] 2/-1/-1->1->0 [22] 2/-1/-1->1->0 [23] 2/-1/-1->1->0
+ip-26-0-169-207:162592:162942 [3] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-169-207:162590:162936 [1] NCCL INFO Channel 00/24 :    0   1   2   3
+ip-26-0-169-207:162590:162936 [1] NCCL INFO Channel 01/24 :    0   1   2   3
+ip-26-0-169-207:162590:162936 [1] NCCL INFO Channel 02/24 :    0   1   2   3
+ip-26-0-169-207:162590:162936 [1] NCCL INFO Channel 03/24 :    0   1   2   3
+ip-26-0-169-207:162590:162936 [1] NCCL INFO Channel 04/24 :    0   1   2   3
+ip-26-0-169-207:162590:162936 [1] NCCL INFO Channel 05/24 :    0   1   2   3
+ip-26-0-169-207:162590:162936 [1] NCCL INFO Channel 06/24 :    0   1   2   3
+ip-26-0-169-207:162590:162936 [1] NCCL INFO Channel 07/24 :    0   1   2   3
+ip-26-0-169-207:162590:162936 [1] NCCL INFO Channel 08/24 :    0   1   2   3
+ip-26-0-169-207:162590:162936 [1] NCCL INFO Channel 09/24 :    0   1   2   3
+ip-26-0-169-207:162596:162941 [7] NCCL INFO Trees [0] -1/-1/-1->3->2 [1] -1/-1/-1->3->2 [2] -1/-1/-1->3->2 [3] -1/-1/-1->3->2 [4] -1/-1/-1->3->2 [5] -1/-1/-1->3->2 [6] -1/-1/-1->3->2 [7] -1/-1/-1->3->2 [8] -1/-1/-1->3->2 [9] -1/-1/-1->3->2 [10] -1/-1/-1->3->2 [11] -1/-1/-1->3->2 [12] -1/-1/-1->3->2 [13] -1/-1/-1->3->2 [14] -1/-1/-1->3->2 [15] -1/-1/-1->3->2 [16] -1/-1/-1->3->2 [17] -1/-1/-1->3->2 [18] -1/-1/-1->3->2 [19] -1/-1/-1->3->2 [20] -1/-1/-1->3->2 [21] -1/-1/-1->3->2 [22] -1/-1/-1->3->2 [23] -1/-1/-1->3->2
+ip-26-0-169-207:162590:162936 [1] NCCL INFO Channel 10/24 :    0   1   2   3
+ip-26-0-169-207:162590:162936 [1] NCCL INFO Channel 11/24 :    0   1   2   3
+ip-26-0-169-207:162596:162941 [7] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-169-207:162590:162936 [1] NCCL INFO Channel 12/24 :    0   1   2   3
+ip-26-0-169-207:162590:162936 [1] NCCL INFO Channel 13/24 :    0   1   2   3
+ip-26-0-169-207:162590:162936 [1] NCCL INFO Channel 14/24 :    0   1   2   3
+ip-26-0-169-207:162590:162936 [1] NCCL INFO Channel 15/24 :    0   1   2   3
+ip-26-0-169-207:162590:162936 [1] NCCL INFO Channel 16/24 :    0   1   2   3
+ip-26-0-169-207:162590:162936 [1] NCCL INFO Channel 17/24 :    0   1   2   3
+ip-26-0-169-207:162594:162943 [5] NCCL INFO Trees [0] 3/-1/-1->2->1 [1] 3/-1/-1->2->1 [2] 3/-1/-1->2->1 [3] 3/-1/-1->2->1 [4] 3/-1/-1->2->1 [5] 3/-1/-1->2->1 [6] 3/-1/-1->2->1 [7] 3/-1/-1->2->1 [8] 3/-1/-1->2->1 [9] 3/-1/-1->2->1 [10] 3/-1/-1->2->1 [11] 3/-1/-1->2->1 [12] 3/-1/-1->2->1 [13] 3/-1/-1->2->1 [14] 3/-1/-1->2->1 [15] 3/-1/-1->2->1 [16] 3/-1/-1->2->1 [17] 3/-1/-1->2->1 [18] 3/-1/-1->2->1 [19] 3/-1/-1->2->1 [20] 3/-1/-1->2->1 [21] 3/-1/-1->2->1 [22] 3/-1/-1->2->1 [23] 3/-1/-1->2->1
+ip-26-0-169-207:162590:162936 [1] NCCL INFO Channel 18/24 :    0   1   2   3
+ip-26-0-169-207:162590:162936 [1] NCCL INFO Channel 19/24 :    0   1   2   3
+ip-26-0-169-207:162594:162943 [5] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-169-207:162590:162936 [1] NCCL INFO Channel 20/24 :    0   1   2   3
+ip-26-0-169-207:162590:162936 [1] NCCL INFO Channel 21/24 :    0   1   2   3
+ip-26-0-169-207:162590:162936 [1] NCCL INFO Channel 22/24 :    0   1   2   3
+ip-26-0-169-207:162590:162936 [1] NCCL INFO Channel 23/24 :    0   1   2   3
+ip-26-0-169-207:162590:162936 [1] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] 1/-1/-1->0->-1 [7] 1/-1/-1->0->-1 [8] 1/-1/-1->0->-1 [9] 1/-1/-1->0->-1 [10] 1/-1/-1->0->-1 [11] 1/-1/-1->0->-1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] 1/-1/-1->0->-1 [19] 1/-1/-1->0->-1 [20] 1/-1/-1->0->-1 [21] 1/-1/-1->0->-1 [22] 1/-1/-1->0->-1 [23] 1/-1/-1->0->-1
+ip-26-0-169-207:162590:162936 [1] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-169-207:162595:162938 [6] NCCL INFO Channel 02/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162589:162937 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-169-207:162593:162939 [4] NCCL INFO Channel 03/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-169-207:162591:162940 [2] NCCL INFO Channel 04/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-169-207:162595:162938 [6] NCCL INFO Channel 03/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162593:162939 [4] NCCL INFO Channel 04/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-169-207:162589:162937 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-169-207:162591:162940 [2] NCCL INFO Channel 05/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-169-207:162591:162940 [2] NCCL INFO Channel 06/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-169-207:162595:162938 [6] NCCL INFO Channel 04/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162589:162937 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-169-207:162593:162939 [4] NCCL INFO Channel 05/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-169-207:162591:162940 [2] NCCL INFO Channel 07/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-169-207:162595:162938 [6] NCCL INFO Channel 05/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162589:162937 [0] NCCL INFO Channel 04/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-169-207:162593:162939 [4] NCCL INFO Channel 06/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-169-207:162591:162940 [2] NCCL INFO Channel 08/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-169-207:162595:162938 [6] NCCL INFO Channel 06/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162589:162937 [0] NCCL INFO Channel 05/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-169-207:162593:162939 [4] NCCL INFO Channel 07/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-169-207:162591:162940 [2] NCCL INFO Channel 09/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-169-207:162595:162938 [6] NCCL INFO Channel 07/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162589:162937 [0] NCCL INFO Channel 06/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-169-207:162593:162939 [4] NCCL INFO Channel 08/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-169-207:162591:162940 [2] NCCL INFO Channel 10/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-169-207:162595:162938 [6] NCCL INFO Channel 08/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162589:162937 [0] NCCL INFO Channel 07/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-169-207:162593:162939 [4] NCCL INFO Channel 09/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-169-207:162591:162940 [2] NCCL INFO Channel 11/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-169-207:162595:162938 [6] NCCL INFO Channel 09/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162589:162937 [0] NCCL INFO Channel 08/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-169-207:162593:162939 [4] NCCL INFO Channel 10/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-169-207:162591:162940 [2] NCCL INFO Channel 12/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-169-207:162595:162938 [6] NCCL INFO Channel 10/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162593:162939 [4] NCCL INFO Channel 11/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-169-207:162589:162937 [0] NCCL INFO Channel 09/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-169-207:162591:162940 [2] NCCL INFO Channel 13/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-169-207:162595:162938 [6] NCCL INFO Channel 11/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162593:162939 [4] NCCL INFO Channel 12/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-169-207:162589:162937 [0] NCCL INFO Channel 10/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-169-207:162591:162940 [2] NCCL INFO Channel 14/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-169-207:162595:162938 [6] NCCL INFO Channel 12/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162593:162939 [4] NCCL INFO Channel 13/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-169-207:162589:162937 [0] NCCL INFO Channel 11/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-169-207:162589:162937 [0] NCCL INFO Channel 12/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-169-207:162591:162940 [2] NCCL INFO Channel 15/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-169-207:162595:162938 [6] NCCL INFO Channel 13/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162593:162939 [4] NCCL INFO Channel 14/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-169-207:162591:162940 [2] NCCL INFO Channel 16/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-169-207:162589:162937 [0] NCCL INFO Channel 13/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-169-207:162593:162939 [4] NCCL INFO Channel 15/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-169-207:162595:162938 [6] NCCL INFO Channel 14/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162591:162940 [2] NCCL INFO Channel 17/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-169-207:162589:162937 [0] NCCL INFO Channel 14/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-169-207:162593:162939 [4] NCCL INFO Channel 16/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-169-207:162595:162938 [6] NCCL INFO Channel 15/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162591:162940 [2] NCCL INFO Channel 18/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-169-207:162591:162940 [2] NCCL INFO Channel 19/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-169-207:162589:162937 [0] NCCL INFO Channel 15/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-169-207:162591:162940 [2] NCCL INFO Channel 20/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-169-207:162593:162939 [4] NCCL INFO Channel 17/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-169-207:162589:162937 [0] NCCL INFO Channel 16/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-169-207:162591:162940 [2] NCCL INFO Channel 21/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-169-207:162595:162938 [6] NCCL INFO Channel 16/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162593:162939 [4] NCCL INFO Channel 18/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-169-207:162591:162940 [2] NCCL INFO Channel 22/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-169-207:162589:162937 [0] NCCL INFO Channel 17/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-169-207:162593:162939 [4] NCCL INFO Channel 19/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-169-207:162595:162938 [6] NCCL INFO Channel 17/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162591:162940 [2] NCCL INFO Channel 23/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-169-207:162589:162937 [0] NCCL INFO Channel 18/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-169-207:162593:162939 [4] NCCL INFO Channel 20/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-169-207:162595:162938 [6] NCCL INFO Channel 18/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162589:162937 [0] NCCL INFO Channel 19/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-169-207:162593:162939 [4] NCCL INFO Channel 21/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-169-207:162595:162938 [6] NCCL INFO Channel 19/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162589:162937 [0] NCCL INFO Channel 20/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-169-207:162593:162939 [4] NCCL INFO Channel 22/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-169-207:162595:162938 [6] NCCL INFO Channel 20/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162589:162937 [0] NCCL INFO Channel 21/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-169-207:162593:162939 [4] NCCL INFO Channel 23/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-169-207:162595:162938 [6] NCCL INFO Channel 21/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162589:162937 [0] NCCL INFO Channel 22/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-169-207:162595:162938 [6] NCCL INFO Channel 22/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162589:162937 [0] NCCL INFO Channel 23/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-169-207:162595:162938 [6] NCCL INFO Channel 23/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162596:162941 [7] NCCL INFO Channel 00/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-169-207:162596:162941 [7] NCCL INFO Channel 01/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-169-207:162596:162941 [7] NCCL INFO Channel 02/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-169-207:162592:162942 [3] NCCL INFO Channel 00/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-169-207:162590:162936 [1] NCCL INFO Channel 00/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-169-207:162592:162942 [3] NCCL INFO Channel 01/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-169-207:162596:162941 [7] NCCL INFO Channel 03/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-169-207:162590:162936 [1] NCCL INFO Channel 01/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-169-207:162594:162943 [5] NCCL INFO Channel 00/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-169-207:162596:162941 [7] NCCL INFO Channel 04/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-169-207:162592:162942 [3] NCCL INFO Channel 02/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-169-207:162592:162942 [3] NCCL INFO Channel 03/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-169-207:162590:162936 [1] NCCL INFO Channel 02/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-169-207:162594:162943 [5] NCCL INFO Channel 01/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-169-207:162596:162941 [7] NCCL INFO Channel 05/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-169-207:162592:162942 [3] NCCL INFO Channel 04/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-169-207:162590:162936 [1] NCCL INFO Channel 03/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-169-207:162594:162943 [5] NCCL INFO Channel 02/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-169-207:162596:162941 [7] NCCL INFO Channel 06/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-169-207:162592:162942 [3] NCCL INFO Channel 05/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-169-207:162590:162936 [1] NCCL INFO Channel 04/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-169-207:162594:162943 [5] NCCL INFO Channel 03/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-169-207:162596:162941 [7] NCCL INFO Channel 07/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-169-207:162594:162943 [5] NCCL INFO Channel 04/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-169-207:162592:162942 [3] NCCL INFO Channel 06/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-169-207:162596:162941 [7] NCCL INFO Channel 08/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-169-207:162596:162941 [7] NCCL INFO Channel 09/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-169-207:162596:162941 [7] NCCL INFO Channel 10/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-169-207:162596:162941 [7] NCCL INFO Channel 11/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-169-207:162596:162941 [7] NCCL INFO Channel 12/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-169-207:162590:162936 [1] NCCL INFO Channel 05/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-169-207:162594:162943 [5] NCCL INFO Channel 05/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-169-207:162592:162942 [3] NCCL INFO Channel 07/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-169-207:162596:162941 [7] NCCL INFO Channel 13/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-169-207:162590:162936 [1] NCCL INFO Channel 06/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-169-207:162594:162943 [5] NCCL INFO Channel 06/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-169-207:162592:162942 [3] NCCL INFO Channel 08/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-169-207:162596:162941 [7] NCCL INFO Channel 14/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-169-207:162590:162936 [1] NCCL INFO Channel 07/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-169-207:162594:162943 [5] NCCL INFO Channel 07/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-169-207:162592:162942 [3] NCCL INFO Channel 09/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-169-207:162590:162936 [1] NCCL INFO Channel 08/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-169-207:162592:162942 [3] NCCL INFO Channel 10/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-169-207:162590:162936 [1] NCCL INFO Channel 09/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-169-207:162592:162942 [3] NCCL INFO Channel 11/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-169-207:162594:162943 [5] NCCL INFO Channel 08/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-169-207:162590:162936 [1] NCCL INFO Channel 10/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-169-207:162596:162941 [7] NCCL INFO Channel 15/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-169-207:162594:162943 [5] NCCL INFO Channel 09/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-169-207:162590:162936 [1] NCCL INFO Channel 11/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-169-207:162592:162942 [3] NCCL INFO Channel 12/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-169-207:162594:162943 [5] NCCL INFO Channel 10/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-169-207:162594:162943 [5] NCCL INFO Channel 11/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-169-207:162596:162941 [7] NCCL INFO Channel 16/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-169-207:162590:162936 [1] NCCL INFO Channel 12/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-169-207:162592:162942 [3] NCCL INFO Channel 13/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-169-207:162594:162943 [5] NCCL INFO Channel 12/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-169-207:162596:162941 [7] NCCL INFO Channel 17/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-169-207:162590:162936 [1] NCCL INFO Channel 13/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-169-207:162592:162942 [3] NCCL INFO Channel 14/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-169-207:162594:162943 [5] NCCL INFO Channel 13/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-169-207:162596:162941 [7] NCCL INFO Channel 18/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-169-207:162590:162936 [1] NCCL INFO Channel 14/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-169-207:162592:162942 [3] NCCL INFO Channel 15/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-169-207:162594:162943 [5] NCCL INFO Channel 14/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-169-207:162596:162941 [7] NCCL INFO Channel 19/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-169-207:162590:162936 [1] NCCL INFO Channel 15/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-169-207:162594:162943 [5] NCCL INFO Channel 15/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-169-207:162592:162942 [3] NCCL INFO Channel 16/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-169-207:162590:162936 [1] NCCL INFO Channel 16/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-169-207:162594:162943 [5] NCCL INFO Channel 16/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-169-207:162592:162942 [3] NCCL INFO Channel 17/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-169-207:162594:162943 [5] NCCL INFO Channel 17/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-169-207:162590:162936 [1] NCCL INFO Channel 17/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-169-207:162596:162941 [7] NCCL INFO Channel 20/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-169-207:162592:162942 [3] NCCL INFO Channel 18/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-169-207:162594:162943 [5] NCCL INFO Channel 18/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-169-207:162590:162936 [1] NCCL INFO Channel 18/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-169-207:162596:162941 [7] NCCL INFO Channel 21/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-169-207:162592:162942 [3] NCCL INFO Channel 19/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-169-207:162594:162943 [5] NCCL INFO Channel 19/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-169-207:162590:162936 [1] NCCL INFO Channel 19/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-169-207:162594:162943 [5] NCCL INFO Channel 20/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-169-207:162592:162942 [3] NCCL INFO Channel 20/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-169-207:162596:162941 [7] NCCL INFO Channel 22/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-169-207:162590:162936 [1] NCCL INFO Channel 20/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-169-207:162592:162942 [3] NCCL INFO Channel 21/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-169-207:162594:162943 [5] NCCL INFO Channel 21/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-169-207:162596:162941 [7] NCCL INFO Channel 23/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-169-207:162590:162936 [1] NCCL INFO Channel 21/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-169-207:162592:162942 [3] NCCL INFO Channel 22/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-169-207:162594:162943 [5] NCCL INFO Channel 22/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-169-207:162590:162936 [1] NCCL INFO Channel 22/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-169-207:162592:162942 [3] NCCL INFO Channel 23/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-169-207:162594:162943 [5] NCCL INFO Channel 23/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-169-207:162590:162936 [1] NCCL INFO Channel 23/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-169-207:162591:162940 [2] NCCL INFO Connected all rings
+ip-26-0-169-207:162595:162938 [6] NCCL INFO Connected all rings
+ip-26-0-169-207:162595:162938 [6] NCCL INFO Channel 00/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-169-207:162589:162937 [0] NCCL INFO Connected all rings
+ip-26-0-169-207:162595:162938 [6] NCCL INFO Channel 01/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-169-207:162595:162938 [6] NCCL INFO Channel 02/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-169-207:162595:162938 [6] NCCL INFO Channel 03/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-169-207:162593:162939 [4] NCCL INFO Connected all rings
+ip-26-0-169-207:162595:162938 [6] NCCL INFO Channel 04/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-169-207:162595:162938 [6] NCCL INFO Channel 05/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-169-207:162595:162938 [6] NCCL INFO Channel 06/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-169-207:162595:162938 [6] NCCL INFO Channel 07/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-169-207:162595:162938 [6] NCCL INFO Channel 08/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-169-207:162595:162938 [6] NCCL INFO Channel 09/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-169-207:162595:162938 [6] NCCL INFO Channel 10/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-169-207:162595:162938 [6] NCCL INFO Channel 11/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-169-207:162595:162938 [6] NCCL INFO Channel 12/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-169-207:162595:162938 [6] NCCL INFO Channel 13/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-169-207:162595:162938 [6] NCCL INFO Channel 14/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-169-207:162595:162938 [6] NCCL INFO Channel 15/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-169-207:162595:162938 [6] NCCL INFO Channel 16/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-169-207:162595:162938 [6] NCCL INFO Channel 17/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-169-207:162595:162938 [6] NCCL INFO Channel 18/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-169-207:162595:162938 [6] NCCL INFO Channel 19/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-169-207:162595:162938 [6] NCCL INFO Channel 20/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-169-207:162595:162938 [6] NCCL INFO Channel 21/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-169-207:162595:162938 [6] NCCL INFO Channel 22/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-169-207:162595:162938 [6] NCCL INFO Channel 23/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-169-207:162590:162936 [1] NCCL INFO Connected all rings
+ip-26-0-169-207:162591:162940 [2] NCCL INFO Channel 00/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162594:162943 [5] NCCL INFO Connected all rings
+ip-26-0-169-207:162591:162940 [2] NCCL INFO Channel 01/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162591:162940 [2] NCCL INFO Channel 02/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162596:162941 [7] NCCL INFO Connected all rings
+ip-26-0-169-207:162596:162941 [7] NCCL INFO Channel 00/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-169-207:162591:162940 [2] NCCL INFO Channel 03/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162596:162941 [7] NCCL INFO Channel 01/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-169-207:162593:162939 [4] NCCL INFO Channel 00/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-169-207:162591:162940 [2] NCCL INFO Channel 04/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162592:162942 [3] NCCL INFO Connected all rings
+ip-26-0-169-207:162596:162941 [7] NCCL INFO Channel 02/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-169-207:162593:162939 [4] NCCL INFO Channel 01/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-169-207:162591:162940 [2] NCCL INFO Channel 05/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162596:162941 [7] NCCL INFO Channel 03/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-169-207:162593:162939 [4] NCCL INFO Channel 02/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-169-207:162591:162940 [2] NCCL INFO Channel 06/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162596:162941 [7] NCCL INFO Channel 04/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-169-207:162593:162939 [4] NCCL INFO Channel 03/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-169-207:162591:162940 [2] NCCL INFO Channel 07/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162596:162941 [7] NCCL INFO Channel 05/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-169-207:162593:162939 [4] NCCL INFO Channel 04/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-169-207:162591:162940 [2] NCCL INFO Channel 08/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162596:162941 [7] NCCL INFO Channel 06/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-169-207:162593:162939 [4] NCCL INFO Channel 05/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-169-207:162591:162940 [2] NCCL INFO Channel 09/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162596:162941 [7] NCCL INFO Channel 07/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-169-207:162593:162939 [4] NCCL INFO Channel 06/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-169-207:162591:162940 [2] NCCL INFO Channel 10/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162596:162941 [7] NCCL INFO Channel 08/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-169-207:162593:162939 [4] NCCL INFO Channel 07/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-169-207:162591:162940 [2] NCCL INFO Channel 11/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162596:162941 [7] NCCL INFO Channel 09/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-169-207:162593:162939 [4] NCCL INFO Channel 08/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-169-207:162591:162940 [2] NCCL INFO Channel 12/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162596:162941 [7] NCCL INFO Channel 10/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-169-207:162593:162939 [4] NCCL INFO Channel 09/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-169-207:162591:162940 [2] NCCL INFO Channel 13/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162596:162941 [7] NCCL INFO Channel 11/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-169-207:162593:162939 [4] NCCL INFO Channel 10/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-169-207:162591:162940 [2] NCCL INFO Channel 14/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162596:162941 [7] NCCL INFO Channel 12/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-169-207:162593:162939 [4] NCCL INFO Channel 11/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-169-207:162591:162940 [2] NCCL INFO Channel 15/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162596:162941 [7] NCCL INFO Channel 13/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-169-207:162593:162939 [4] NCCL INFO Channel 12/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-169-207:162591:162940 [2] NCCL INFO Channel 16/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162596:162941 [7] NCCL INFO Channel 14/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-169-207:162593:162939 [4] NCCL INFO Channel 13/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-169-207:162591:162940 [2] NCCL INFO Channel 17/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162596:162941 [7] NCCL INFO Channel 15/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-169-207:162593:162939 [4] NCCL INFO Channel 14/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-169-207:162591:162940 [2] NCCL INFO Channel 18/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162596:162941 [7] NCCL INFO Channel 16/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-169-207:162593:162939 [4] NCCL INFO Channel 15/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-169-207:162591:162940 [2] NCCL INFO Channel 19/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162596:162941 [7] NCCL INFO Channel 17/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-169-207:162593:162939 [4] NCCL INFO Channel 16/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-169-207:162591:162940 [2] NCCL INFO Channel 20/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162596:162941 [7] NCCL INFO Channel 18/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-169-207:162593:162939 [4] NCCL INFO Channel 17/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-169-207:162591:162940 [2] NCCL INFO Channel 21/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162596:162941 [7] NCCL INFO Channel 19/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-169-207:162593:162939 [4] NCCL INFO Channel 18/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-169-207:162591:162940 [2] NCCL INFO Channel 22/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162596:162941 [7] NCCL INFO Channel 20/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-169-207:162593:162939 [4] NCCL INFO Channel 19/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-169-207:162591:162940 [2] NCCL INFO Channel 23/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-169-207:162596:162941 [7] NCCL INFO Channel 21/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-169-207:162593:162939 [4] NCCL INFO Channel 20/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-169-207:162594:162943 [5] NCCL INFO Channel 00/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-169-207:162596:162941 [7] NCCL INFO Channel 22/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-169-207:162593:162939 [4] NCCL INFO Channel 21/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-169-207:162594:162943 [5] NCCL INFO Channel 01/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-169-207:162596:162941 [7] NCCL INFO Channel 23/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-169-207:162593:162939 [4] NCCL INFO Channel 22/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-169-207:162594:162943 [5] NCCL INFO Channel 02/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-169-207:162593:162939 [4] NCCL INFO Channel 23/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-169-207:162594:162943 [5] NCCL INFO Channel 03/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-169-207:162592:162942 [3] NCCL INFO Channel 00/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-169-207:162594:162943 [5] NCCL INFO Channel 04/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-169-207:162592:162942 [3] NCCL INFO Channel 01/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-169-207:162594:162943 [5] NCCL INFO Channel 05/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-169-207:162592:162942 [3] NCCL INFO Channel 02/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-169-207:162594:162943 [5] NCCL INFO Channel 06/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-169-207:162592:162942 [3] NCCL INFO Channel 03/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-169-207:162594:162943 [5] NCCL INFO Channel 07/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-169-207:162592:162942 [3] NCCL INFO Channel 04/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-169-207:162594:162943 [5] NCCL INFO Channel 08/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-169-207:162592:162942 [3] NCCL INFO Channel 05/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-169-207:162594:162943 [5] NCCL INFO Channel 09/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-169-207:162592:162942 [3] NCCL INFO Channel 06/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-169-207:162594:162943 [5] NCCL INFO Channel 10/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-169-207:162592:162942 [3] NCCL INFO Channel 07/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-169-207:162594:162943 [5] NCCL INFO Channel 11/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-169-207:162592:162942 [3] NCCL INFO Channel 08/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-169-207:162594:162943 [5] NCCL INFO Channel 12/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-169-207:162592:162942 [3] NCCL INFO Channel 09/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-169-207:162594:162943 [5] NCCL INFO Channel 13/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-169-207:162592:162942 [3] NCCL INFO Channel 10/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-169-207:162594:162943 [5] NCCL INFO Channel 14/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-169-207:162592:162942 [3] NCCL INFO Channel 11/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-169-207:162594:162943 [5] NCCL INFO Channel 15/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-169-207:162589:162937 [0] NCCL INFO Connected all trees
+ip-26-0-169-207:162589:162937 [0] NCCL INFO NVLS comm 0xb370a90 headRank 0 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-169-207:162592:162942 [3] NCCL INFO Channel 12/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-169-207:162595:162938 [6] NCCL INFO Connected all trees
+ip-26-0-169-207:162594:162943 [5] NCCL INFO Channel 16/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-169-207:162595:162938 [6] NCCL INFO NVLS comm 0xb13a320 headRank 3 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-169-207:162592:162942 [3] NCCL INFO Channel 13/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-169-207:162594:162943 [5] NCCL INFO Channel 17/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-169-207:162592:162942 [3] NCCL INFO Channel 14/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-169-207:162594:162943 [5] NCCL INFO Channel 18/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-169-207:162592:162942 [3] NCCL INFO Channel 15/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-169-207:162592:162942 [3] NCCL INFO Channel 16/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-169-207:162594:162943 [5] NCCL INFO Channel 19/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-169-207:162592:162942 [3] NCCL INFO Channel 17/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-169-207:162594:162943 [5] NCCL INFO Channel 20/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-169-207:162592:162942 [3] NCCL INFO Channel 18/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-169-207:162594:162943 [5] NCCL INFO Channel 21/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-169-207:162592:162942 [3] NCCL INFO Channel 19/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-169-207:162594:162943 [5] NCCL INFO Channel 22/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-169-207:162592:162942 [3] NCCL INFO Channel 20/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-169-207:162594:162943 [5] NCCL INFO Channel 23/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-169-207:162592:162942 [3] NCCL INFO Channel 21/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-169-207:162592:162942 [3] NCCL INFO Channel 22/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-169-207:162592:162942 [3] NCCL INFO Channel 23/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-169-207:162591:162940 [2] NCCL INFO Connected all trees
+ip-26-0-169-207:162591:162940 [2] NCCL INFO NVLS comm 0xa8e8d10 headRank 1 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-169-207:162593:162939 [4] NCCL INFO Connected all trees
+ip-26-0-169-207:162593:162939 [4] NCCL INFO NVLS comm 0xbc3f590 headRank 2 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-169-207:162596:162941 [7] NCCL INFO Connected all trees
+ip-26-0-169-207:162596:162941 [7] NCCL INFO NVLS comm 0xa76d9a0 headRank 3 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-169-207:162593:162939 [4] NCCL INFO Connected NVLS tree
+ip-26-0-169-207:162593:162939 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-169-207:162593:162939 [4] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-169-207:162593:162939 [4] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:162595:162938 [6] NCCL INFO Connected NVLS tree
+ip-26-0-169-207:162595:162938 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-169-207:162595:162938 [6] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-169-207:162595:162938 [6] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:162589:162937 [0] NCCL INFO Connected NVLS tree
+ip-26-0-169-207:162591:162940 [2] NCCL INFO Connected NVLS tree
+ip-26-0-169-207:162591:162940 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-169-207:162589:162937 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-169-207:162591:162940 [2] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-169-207:162591:162940 [2] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:162589:162937 [0] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-169-207:162589:162937 [0] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:162590:162936 [1] NCCL INFO Connected all trees
+ip-26-0-169-207:162590:162936 [1] NCCL INFO NVLS comm 0xaaa76e0 headRank 0 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-169-207:162593:162939 [4] NCCL INFO comm 0xbc3f590 rank 2 nranks 4 cudaDev 4 nvmlDev 4 busId 97000 commId 0x8d184ed925e1cb1 - Init COMPLETE
+ip-26-0-169-207:162589:162937 [0] NCCL INFO comm 0xb370a90 rank 0 nranks 4 cudaDev 0 nvmlDev 0 busId 53000 commId 0x8d184ed925e1cb1 - Init COMPLETE
+ip-26-0-169-207:162591:162940 [2] NCCL INFO comm 0xa8e8d10 rank 1 nranks 4 cudaDev 2 nvmlDev 2 busId 75000 commId 0x8d184ed925e1cb1 - Init COMPLETE
+ip-26-0-169-207:162595:162938 [6] NCCL INFO comm 0xb13a320 rank 3 nranks 4 cudaDev 6 nvmlDev 6 busId b9000 commId 0x8d184ed925e1cb1 - Init COMPLETE
+ip-26-0-169-207:162594:162943 [5] NCCL INFO Connected all trees
+ip-26-0-169-207:162594:162943 [5] NCCL INFO NVLS comm 0xb3189e0 headRank 2 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-169-207:162592:162942 [3] NCCL INFO Connected all trees
+ip-26-0-169-207:162592:162942 [3] NCCL INFO NVLS comm 0xc480a60 headRank 1 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-169-207:162596:162941 [7] NCCL INFO Connected NVLS tree
+ip-26-0-169-207:162592:162942 [3] NCCL INFO Connected NVLS tree
+ip-26-0-169-207:162596:162941 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-169-207:162596:162941 [7] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-169-207:162596:162941 [7] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:162592:162942 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-169-207:162592:162942 [3] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-169-207:162592:162942 [3] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:162594:162943 [5] NCCL INFO Connected NVLS tree
+ip-26-0-169-207:162594:162943 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-169-207:162594:162943 [5] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-169-207:162594:162943 [5] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:162590:162936 [1] NCCL INFO Connected NVLS tree
+ip-26-0-169-207:162590:162936 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-169-207:162590:162936 [1] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-169-207:162590:162936 [1] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:162596:162941 [7] NCCL INFO comm 0xa76d9a0 rank 3 nranks 4 cudaDev 7 nvmlDev 7 busId ca000 commId 0x7d5c542bb93ccd3a - Init COMPLETE
+ip-26-0-169-207:162592:162942 [3] NCCL INFO comm 0xc480a60 rank 1 nranks 4 cudaDev 3 nvmlDev 3 busId 86000 commId 0x7d5c542bb93ccd3a - Init COMPLETE
+ip-26-0-169-207:162594:162943 [5] NCCL INFO comm 0xb3189e0 rank 2 nranks 4 cudaDev 5 nvmlDev 5 busId a8000 commId 0x7d5c542bb93ccd3a - Init COMPLETE
+ip-26-0-169-207:162590:162936 [1] NCCL INFO comm 0xaaa76e0 rank 0 nranks 4 cudaDev 1 nvmlDev 1 busId 64000 commId 0x7d5c542bb93ccd3a - Init COMPLETE
+12/12/2024 21:38:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]: [Optimizer Building] Using LearningRateForSP as learning rate
+12/12/2024 21:38:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]: [ZeRO sharding] Size of optimizer params per rank:
+12/12/2024 21:38:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]: [ZeRO sharding] DP Rank 0 has 130M out of 520M (25.00%) params' optimizer states
+12/12/2024 21:38:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]: [ZeRO sharding] DP Rank 1 has 130M out of 520M (25.00%) params' optimizer states
+12/12/2024 21:38:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]: [ZeRO sharding] DP Rank 2 has 130M out of 520M (25.00%) params' optimizer states
+12/12/2024 21:38:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]: [ZeRO sharding] DP Rank 3 has 130M out of 520M (25.00%) params' optimizer states
+12/12/2024 21:38:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]: [Training Plan] Stage Stable Training Stage has 99 remaining training steps and has consumed 0 samples
+12/12/2024 21:38:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]: Using dummy data generator
+12/12/2024 21:38:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]: [Training Plan] There are 1 training stages 
+12/12/2024 21:38:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]: [Stage Stable Training Stage] start from step 1 
+12/12/2024 21:38:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]: 
+12/12/2024 21:38:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]: [Start training] datetime: 2024-12-12 21:38:53.938763 | mbs: 2 | grad_accum: 256 | global_batch_size: 2048 | sequence_length: 2048 | train_steps: 100 | start_iteration_step: 0 | consumed_train_samples: 0
+wandb: WARNING Path /fsx/nouamane/.cache/wandb/wandb/ wasn't writable, using system temp directory.
+wandb: WARNING Path /fsx/nouamane/.cache/wandb/wandb/ wasn't writable, using system temp directory
+wandb: Tracking run with wandb version 0.16.0
+wandb: W&B syncing is set to `offline` in this directory.  
+wandb: Run `wandb online` or set WANDB_MODE=online to enable cloud syncing.
+12/12/2024 21:39:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]: Resuming training from stage Stable Training Stage, it has trained for 0 samples and has 99 remaining train steps
+12/12/2024 21:39:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:  Memory usage: 3472.48MiB. Peak allocated 5632.00MiB. Peak reserved: 18884.00MiB
+slurmstepd: error: *** JOB 13392392 ON ip-26-0-169-207 CANCELLED AT 2024-12-12T21:39:07 DUE TO TIME LIMIT ***
+slurmstepd: error: *** STEP 13392392.0 ON ip-26-0-169-207 CANCELLED AT 2024-12-12T21:39:07 DUE TO TIME LIMIT ***
+[2024-12-12 21:39:07,606] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-12 21:39:07,607] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 162589 closing signal SIGTERM
+[2024-12-12 21:39:07,607] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 162590 closing signal SIGTERM
+[2024-12-12 21:39:07,607] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 162591 closing signal SIGTERM
+[2024-12-12 21:39:07,608] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 162592 closing signal SIGTERM
+[2024-12-12 21:39:07,608] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 162593 closing signal SIGTERM
+[2024-12-12 21:39:07,608] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 162594 closing signal SIGTERM
+[2024-12-12 21:39:07,608] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 162595 closing signal SIGTERM
+[2024-12-12 21:39:07,609] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 162596 closing signal SIGTERM
+srun: Job step aborted: Waiting up to 32 seconds for job step to finish.
+[2024-12-12 21:39:07,746] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 162589 closing signal SIGTERM
+[2024-12-12 21:39:07,746] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 162590 closing signal SIGTERM
+[2024-12-12 21:39:07,746] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 162591 closing signal SIGTERM
+[2024-12-12 21:39:07,746] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 162592 closing signal SIGTERM
+[2024-12-12 21:39:07,746] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 162593 closing signal SIGTERM
+[2024-12-12 21:39:07,747] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 162594 closing signal SIGTERM
+[2024-12-12 21:39:07,747] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 162595 closing signal SIGTERM
+[2024-12-12 21:39:07,747] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 162596 closing signal SIGTERM
diff --git a/logs/13395441-bench_stress_test.out b/logs/13395441-bench_stress_test.out
new file mode 100644
index 0000000000000000000000000000000000000000..171549444f9b8627379222d4996551a59b261ed3
--- /dev/null
+++ b/logs/13395441-bench_stress_test.out
@@ -0,0 +1,4118 @@
++ source /etc/profile.d/modules.sh
+++ . /usr/share/modules/init/bash
++++ unset _mlshdbg
++++ '[' 0 = 1 ']'
++++ unset _mlre _mlIFS
++++ '[' -n x ']'
++++ _mlIFS=' 	
+'
++++ IFS=' '
++++ '[' -n '' ']'
+++++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash autoinit
++++ _mlcode='module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+MODULEPATH_modshare=/etc/environment-modules/modules:1:/usr/share/modules/\$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1; export MODULEPATH_modshare;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+LOADEDMODULES=; export LOADEDMODULES;
+MODULEPATH=/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/\$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles; export MODULEPATH;
+test 0;'
++++ _mlret=0
++++ '[' -n x ']'
++++ IFS=' 	
+'
++++ unset _mlIFS
++++ unset _mlre _mlv _mlrv
++++ '[' 0 -eq 0 ']'
++++ eval 'module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+MODULEPATH_modshare=/etc/environment-modules/modules:1:/usr/share/modules/\$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1; export MODULEPATH_modshare;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+LOADEDMODULES=; export LOADEDMODULES;
+MODULEPATH=/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/\$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles; export MODULEPATH;
+test 0;'
+++++ MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl
+++++ export MODULES_CMD
+++++ ENV=/usr/share/modules/init/profile.sh
+++++ export ENV
+++++ MODULEPATH_modshare='/etc/environment-modules/modules:1:/usr/share/modules/$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1'
+++++ export MODULEPATH_modshare
+++++ BASH_ENV=/usr/share/modules/init/bash
+++++ export BASH_ENV
+++++ MODULESHOME=/usr/share/modules
+++++ export MODULESHOME
+++++ LOADEDMODULES=
+++++ export LOADEDMODULES
+++++ MODULEPATH='/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles'
+++++ export MODULEPATH
+++++ test 0
++++ '[' 0 = 1 ']'
++++ '[' -t 2 ']'
++++ export -f module
++++ export -f switchml
++++ '[' 5 -ge 3 ']'
++++ [[ ehxB =~ i ]]
++++ [[ ! :/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin: =~ :/usr/bin: ]]
++++ '[' '!' -n '' ']'
++++ MANPATH=:
++++ export MANPATH
+++++ manpath
++++ [[ ! :/admin/home/nouamane/.local/share/man:/fsx/nouamane/miniconda/envs/2-1-cu121/man:/fsx/nouamane/miniconda/envs/2-1-cu121/share/man:/fsx/nouamane/miniconda/man:/fsx/nouamane/miniconda/share/man:/opt/amazon/openmpi/share/man:/opt/amazon/efa/share/man:/opt/slurm/share/man:/usr/local/man:/usr/local/share/man:/usr/share/man:/admin/home/nouamane/.fzf/man:: =~ :/usr/share/man: ]]
++++ unset _mlcode _mlret
++++ '[' -n '' ']'
++ module load cuda/12.1
++ unset _mlshdbg
++ '[' 0 = 1 ']'
++ unset _mlre _mlIFS
++ '[' -n x ']'
++ _mlIFS=' 	
+'
++ IFS=' '
++ '[' -n '' ']'
+++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash load cuda/12.1
++ eval 'CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include;' export 'CPATH;
+LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:;' export 'LD_LIBRARY_PATH;
+FI_EFA_FORK_SAFE=1;' export 'FI_EFA_FORK_SAFE;
+MANPATH=/usr/local/cuda-12.1/share/man::;' export 'MANPATH;
+LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64;' export 'LIBRARY_PATH;
+_LMFILES_=/usr/share/modules/modulefiles/cuda/12.1;' export '_LMFILES_;
+LOADEDMODULES=cuda/12.1;' export 'LOADEDMODULES;
+MPI_PATH=/opt/amazon/openmpi;' export 'MPI_PATH;
+NCCL_HOME_modshare=/opt/nccl/build:1;' export 'NCCL_HOME_modshare;
+NCCL_PROTO=simple;' export 'NCCL_PROTO;
+MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1;' export 'MANPATH_modshare;
+LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1;' export 'LIBRARY_PATH_modshare;
+NCCL_SOCKET_IFNAME=enp;' export 'NCCL_SOCKET_IFNAME;
+AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl;' export 'AWS_OFI_NCCL_HOME;
+NCCL_HOME=/opt/nccl/build;' export 'NCCL_HOME;
+FI_PROVIDER=efa;' export 'FI_PROVIDER;
+AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1;' export 'AWS_OFI_NCCL_HOME_modshare;
+CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1;' export 'CPATH_modshare;
+LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2::1:/opt/nccl/build/lib:1:/opt/aws-ofi-nccl/lib:2:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/efa/lib:1;' export 'LD_LIBRARY_PATH_modshare;
+FI_EFA_ENABLE_SHM_TRANSFER=1;' export 'FI_EFA_ENABLE_SHM_TRANSFER;
+_LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1;' export '_LMFILES__modshare;
+LOADEDMODULES_modshare=cuda/12.1:1;' export 'LOADEDMODULES_modshare;
+MPI_PATH_modshare=/opt/amazon/openmpi:1;' export 'MPI_PATH_modshare;
+PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin;' export 'PATH;
+CUDA_HOME=/usr/local/cuda-12.1;' export 'CUDA_HOME;
+PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/usr/local/cuda-12.1/include:1:/opt/amazon/efa/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/usr/sbin:1:/fsx/nouamane/miniconda/bin:1:/opt/slurm/bin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:2:/opt/amazon/openmpi/bin:1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1;' export 'PATH_modshare;
+test' '0;'
+++ CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include
+++ export CPATH
+++ LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:
+++ export LD_LIBRARY_PATH
+++ FI_EFA_FORK_SAFE=1
+++ export FI_EFA_FORK_SAFE
+++ MANPATH=/usr/local/cuda-12.1/share/man::
+++ export MANPATH
+++ LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64
+++ export LIBRARY_PATH
+++ _LMFILES_=/usr/share/modules/modulefiles/cuda/12.1
+++ export _LMFILES_
+++ LOADEDMODULES=cuda/12.1
+++ export LOADEDMODULES
+++ MPI_PATH=/opt/amazon/openmpi
+++ export MPI_PATH
+++ NCCL_HOME_modshare=/opt/nccl/build:1
+++ export NCCL_HOME_modshare
+++ NCCL_PROTO=simple
+++ export NCCL_PROTO
+++ MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1
+++ export MANPATH_modshare
+++ LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1
+++ export LIBRARY_PATH_modshare
+++ NCCL_SOCKET_IFNAME=enp
+++ export NCCL_SOCKET_IFNAME
+++ AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl
+++ export AWS_OFI_NCCL_HOME
+++ NCCL_HOME=/opt/nccl/build
+++ export NCCL_HOME
+++ FI_PROVIDER=efa
+++ export FI_PROVIDER
+++ AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1
+++ export AWS_OFI_NCCL_HOME_modshare
+++ CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1
+++ export CPATH_modshare
+++ LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2::1:/opt/nccl/build/lib:1:/opt/aws-ofi-nccl/lib:2:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/efa/lib:1
+++ export LD_LIBRARY_PATH_modshare
+++ FI_EFA_ENABLE_SHM_TRANSFER=1
+++ export FI_EFA_ENABLE_SHM_TRANSFER
+++ _LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1
+++ export _LMFILES__modshare
+++ LOADEDMODULES_modshare=cuda/12.1:1
+++ export LOADEDMODULES_modshare
+++ MPI_PATH_modshare=/opt/amazon/openmpi:1
+++ export MPI_PATH_modshare
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ export PATH
+++ CUDA_HOME=/usr/local/cuda-12.1
+++ export CUDA_HOME
+++ PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/usr/local/cuda-12.1/include:1:/opt/amazon/efa/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/usr/sbin:1:/fsx/nouamane/miniconda/bin:1:/opt/slurm/bin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:2:/opt/amazon/openmpi/bin:1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1
+++ export PATH_modshare
+++ test 0
++ _mlstatus=0
++ '[' -n x ']'
++ IFS=' 	
+'
++ unset _mlre _mlv _mlrv _mlIFS
++ '[' -n '' ']'
++ unset _mlshdbg
++ return 0
++ source /fsx/nouamane/miniconda/bin/activate
+++ _CONDA_ROOT=/fsx/nouamane/miniconda
+++ . /fsx/nouamane/miniconda/etc/profile.d/conda.sh
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ '[' -z x ']'
+++ conda activate
+++ local cmd=activate
+++ case "$cmd" in
+++ __conda_activate activate
+++ '[' -n '' ']'
+++ local ask_conda
++++ PS1=
++++ __conda_exe shell.posix activate
++++ /fsx/nouamane/miniconda/bin/conda shell.posix activate
+++ ask_conda='. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_2='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\'''
+++ eval '. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_2='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\'''
++++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh
+++++ test -n ''
+++++ unset XML_CATALOG_FILES
+++++ unset xml_catalog_files_libxml2
++++ PS1='(base) '
++++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++++ export CONDA_PREFIX=/fsx/nouamane/miniconda
++++ CONDA_PREFIX=/fsx/nouamane/miniconda
++++ export CONDA_SHLVL=3
++++ CONDA_SHLVL=3
++++ export CONDA_DEFAULT_ENV=base
++++ CONDA_DEFAULT_ENV=base
++++ export 'CONDA_PROMPT_MODIFIER=(base) '
++++ CONDA_PROMPT_MODIFIER='(base) '
++++ export CONDA_PREFIX_2=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ CONDA_PREFIX_2=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ __conda_hashr
+++ '[' -n '' ']'
+++ '[' -n '' ']'
+++ hash -r
++ conda activate 2-1-cu121
++ local cmd=activate
++ case "$cmd" in
++ __conda_activate activate 2-1-cu121
++ '[' -n '' ']'
++ local ask_conda
+++ PS1='(base) '
+++ __conda_exe shell.posix activate 2-1-cu121
+++ /fsx/nouamane/miniconda/bin/conda shell.posix activate 2-1-cu121
++ ask_conda='PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''4'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_3='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
++ eval 'PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''4'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_3='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
+++ PS1='(2-1-cu121) '
+++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ export CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ export CONDA_SHLVL=4
+++ CONDA_SHLVL=4
+++ export CONDA_DEFAULT_ENV=2-1-cu121
+++ CONDA_DEFAULT_ENV=2-1-cu121
+++ export 'CONDA_PROMPT_MODIFIER=(2-1-cu121) '
+++ CONDA_PROMPT_MODIFIER='(2-1-cu121) '
+++ export CONDA_PREFIX_3=/fsx/nouamane/miniconda
+++ CONDA_PREFIX_3=/fsx/nouamane/miniconda
+++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ export _CE_M=
+++ _CE_M=
+++ export _CE_CONDA=
+++ _CE_CONDA=
+++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh
++++ test -n ''
++++ xml_catalog_files_libxml2=
++++ XML_CATALOG_FILES=
++++ conda_catalog_files=
++++ ifs_libxml2=' 	
+'
++++ IFS=' '
++++ rem=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ for pre in ${rem}
++++ test '' = /fsx/nouamane/miniconda/envs/2-1-cu121
++++ conda_catalog_files=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ rem=
++++ IFS=' 	
+'
++++ conda_catalog_files='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ export 'XML_CATALOG_FILES=file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ XML_CATALOG_FILES='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ unset conda_catalog_files ifs_libxml2 rem
++ __conda_hashr
++ '[' -n '' ']'
++ '[' -n '' ']'
++ hash -r
++ export PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++ PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ scontrol show hostnames ip-26-0-160-103
++ export NODELIST=ip-26-0-160-103
++ NODELIST=ip-26-0-160-103
+++ scontrol show hostnames ip-26-0-160-103
+++ head -n1
++ export MASTER_NODE=ip-26-0-160-103
++ MASTER_NODE=ip-26-0-160-103
++ export MASTER_PORT=12356
++ MASTER_PORT=12356
++ export NNODES=1
++ NNODES=1
++ export GPUS_PER_NODE=8
++ GPUS_PER_NODE=8
++ export WORLD_SIZE=8
++ WORLD_SIZE=8
++ export CUDA_DEVICE_MAX_CONNECTIONS=1
++ CUDA_DEVICE_MAX_CONNECTIONS=1
++ export NCCL_DEBUG=INFO
++ NCCL_DEBUG=INFO
++ export NANOTRON_BENCHMARK=1
++ NANOTRON_BENCHMARK=1
++ echo 'Master node: ip-26-0-160-103'
+Master node: ip-26-0-160-103
++ echo 'All nodes: ip-26-0-160-103'
+All nodes: ip-26-0-160-103
++ echo 'World size: 8'
+World size: 8
++ srun torchrun --nnodes=1 --nproc_per_node=8 --rdzv_id=13395441 --rdzv_backend=c10d --rdzv_endpoint=ip-26-0-160-103:12356 run_train.py --config-file benchmark/configs/config_1.14G_dp4_tp2_pp1_acc256_mbs2_seq2048_zero1_tpmodeRED_vocab32k.yaml
+[2024-12-12 22:02:10,495] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-12 22:02:10,495] torch.distributed.run: [WARNING] 
+[2024-12-12 22:02:10,495] torch.distributed.run: [WARNING] *****************************************
+[2024-12-12 22:02:10,495] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-12 22:02:10,495] torch.distributed.run: [WARNING] *****************************************
+ip-26-0-160-103:2068009:2068009 [0] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-160-103:2068009:2068009 [0] NCCL INFO Bootstrap : Using enp71s0:26.0.160.103<0>
+ip-26-0-160-103:2068009:2068009 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-160-103:2068009:2068009 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-160-103:2068009:2068009 [0] NCCL INFO cudaDriverVersion 12020
+NCCL version 2.18.5+cuda12.2
+ip-26-0-160-103:2068012:2068012 [3] NCCL INFO cudaDriverVersion 12020
+ip-26-0-160-103:2068012:2068012 [3] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-160-103:2068014:2068014 [5] NCCL INFO cudaDriverVersion 12020
+ip-26-0-160-103:2068014:2068014 [5] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-160-103:2068012:2068012 [3] NCCL INFO Bootstrap : Using enp71s0:26.0.160.103<0>
+ip-26-0-160-103:2068013:2068013 [4] NCCL INFO cudaDriverVersion 12020
+ip-26-0-160-103:2068014:2068014 [5] NCCL INFO Bootstrap : Using enp71s0:26.0.160.103<0>
+ip-26-0-160-103:2068016:2068016 [7] NCCL INFO cudaDriverVersion 12020
+ip-26-0-160-103:2068013:2068013 [4] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-160-103:2068016:2068016 [7] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-160-103:2068012:2068012 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-160-103:2068012:2068012 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-160-103:2068014:2068014 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-160-103:2068014:2068014 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-160-103:2068011:2068011 [2] NCCL INFO cudaDriverVersion 12020
+ip-26-0-160-103:2068010:2068010 [1] NCCL INFO cudaDriverVersion 12020
+ip-26-0-160-103:2068015:2068015 [6] NCCL INFO cudaDriverVersion 12020
+ip-26-0-160-103:2068013:2068013 [4] NCCL INFO Bootstrap : Using enp71s0:26.0.160.103<0>
+ip-26-0-160-103:2068016:2068016 [7] NCCL INFO Bootstrap : Using enp71s0:26.0.160.103<0>
+ip-26-0-160-103:2068011:2068011 [2] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-160-103:2068010:2068010 [1] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-160-103:2068015:2068015 [6] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-160-103:2068013:2068013 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-160-103:2068013:2068013 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-160-103:2068016:2068016 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-160-103:2068016:2068016 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-160-103:2068011:2068011 [2] NCCL INFO Bootstrap : Using enp71s0:26.0.160.103<0>
+ip-26-0-160-103:2068010:2068010 [1] NCCL INFO Bootstrap : Using enp71s0:26.0.160.103<0>
+ip-26-0-160-103:2068015:2068015 [6] NCCL INFO Bootstrap : Using enp71s0:26.0.160.103<0>
+ip-26-0-160-103:2068011:2068011 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-160-103:2068011:2068011 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-160-103:2068010:2068010 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-160-103:2068010:2068010 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-160-103:2068015:2068015 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-160-103:2068015:2068015 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-160-103:2068009:2068132 [0] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.9.1-aws
+ip-26-0-160-103:2068009:2068132 [0] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.9.1-aws
+ip-26-0-160-103:2068009:2068132 [0] NCCL INFO NET/OFI Using Libfabric version 1.21
+ip-26-0-160-103:2068009:2068132 [0] NCCL INFO NET/OFI Using CUDA driver version 12020
+ip-26-0-160-103:2068009:2068132 [0] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-160-103:2068009:2068132 [0] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-160-103:2068009:2068132 [0] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-160-103:2068009:2068132 [0] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-160-103:2068009:2068132 [0] NCCL INFO NET/OFI Running on p5.48xlarge platform, Setting NCCL_TOPO_FILE environment variable to /opt/aws-ofi-nccl/share/aws-ofi-nccl/xml/p5.48xl-topo.xml
+ip-26-0-160-103:2068009:2068132 [0] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-160-103:2068009:2068132 [0] NCCL INFO NET/OFI Using transport protocol RDMA
+ip-26-0-160-103:2068010:2068136 [1] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.9.1-aws
+ip-26-0-160-103:2068010:2068136 [1] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.9.1-aws
+ip-26-0-160-103:2068010:2068136 [1] NCCL INFO NET/OFI Using Libfabric version 1.21
+ip-26-0-160-103:2068010:2068136 [1] NCCL INFO NET/OFI Using CUDA driver version 12020
+ip-26-0-160-103:2068010:2068136 [1] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-160-103:2068010:2068136 [1] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-160-103:2068010:2068136 [1] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-160-103:2068010:2068136 [1] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-160-103:2068010:2068136 [1] NCCL INFO NET/OFI Running on p5.48xlarge platform, Setting NCCL_TOPO_FILE environment variable to /opt/aws-ofi-nccl/share/aws-ofi-nccl/xml/p5.48xl-topo.xml
+ip-26-0-160-103:2068010:2068136 [1] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-160-103:2068010:2068136 [1] NCCL INFO NET/OFI Using transport protocol RDMA
+ip-26-0-160-103:2068016:2068133 [7] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.9.1-aws
+ip-26-0-160-103:2068016:2068133 [7] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.9.1-aws
+ip-26-0-160-103:2068016:2068133 [7] NCCL INFO NET/OFI Using Libfabric version 1.21
+ip-26-0-160-103:2068016:2068133 [7] NCCL INFO NET/OFI Using CUDA driver version 12020
+ip-26-0-160-103:2068016:2068133 [7] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-160-103:2068016:2068133 [7] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-160-103:2068016:2068133 [7] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-160-103:2068016:2068133 [7] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-160-103:2068016:2068133 [7] NCCL INFO NET/OFI Running on p5.48xlarge platform, Setting NCCL_TOPO_FILE environment variable to /opt/aws-ofi-nccl/share/aws-ofi-nccl/xml/p5.48xl-topo.xml
+ip-26-0-160-103:2068016:2068133 [7] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-160-103:2068016:2068133 [7] NCCL INFO NET/OFI Using transport protocol RDMA
+ip-26-0-160-103:2068011:2068137 [2] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.9.1-aws
+ip-26-0-160-103:2068011:2068137 [2] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.9.1-aws
+ip-26-0-160-103:2068011:2068137 [2] NCCL INFO NET/OFI Using Libfabric version 1.21
+ip-26-0-160-103:2068011:2068137 [2] NCCL INFO NET/OFI Using CUDA driver version 12020
+ip-26-0-160-103:2068011:2068137 [2] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-160-103:2068011:2068137 [2] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-160-103:2068011:2068137 [2] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-160-103:2068011:2068137 [2] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-160-103:2068011:2068137 [2] NCCL INFO NET/OFI Running on p5.48xlarge platform, Setting NCCL_TOPO_FILE environment variable to /opt/aws-ofi-nccl/share/aws-ofi-nccl/xml/p5.48xl-topo.xml
+ip-26-0-160-103:2068011:2068137 [2] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-160-103:2068011:2068137 [2] NCCL INFO NET/OFI Using transport protocol RDMA
+ip-26-0-160-103:2068014:2068135 [5] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.9.1-aws
+ip-26-0-160-103:2068014:2068135 [5] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.9.1-aws
+ip-26-0-160-103:2068014:2068135 [5] NCCL INFO NET/OFI Using Libfabric version 1.21
+ip-26-0-160-103:2068014:2068135 [5] NCCL INFO NET/OFI Using CUDA driver version 12020
+ip-26-0-160-103:2068014:2068135 [5] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-160-103:2068014:2068135 [5] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-160-103:2068014:2068135 [5] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-160-103:2068014:2068135 [5] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-160-103:2068014:2068135 [5] NCCL INFO NET/OFI Running on p5.48xlarge platform, Setting NCCL_TOPO_FILE environment variable to /opt/aws-ofi-nccl/share/aws-ofi-nccl/xml/p5.48xl-topo.xml
+ip-26-0-160-103:2068014:2068135 [5] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-160-103:2068014:2068135 [5] NCCL INFO NET/OFI Using transport protocol RDMA
+ip-26-0-160-103:2068012:2068134 [3] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.9.1-aws
+ip-26-0-160-103:2068012:2068134 [3] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.9.1-aws
+ip-26-0-160-103:2068012:2068134 [3] NCCL INFO NET/OFI Using Libfabric version 1.21
+ip-26-0-160-103:2068012:2068134 [3] NCCL INFO NET/OFI Using CUDA driver version 12020
+ip-26-0-160-103:2068012:2068134 [3] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-160-103:2068012:2068134 [3] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-160-103:2068012:2068134 [3] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-160-103:2068012:2068134 [3] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-160-103:2068012:2068134 [3] NCCL INFO NET/OFI Running on p5.48xlarge platform, Setting NCCL_TOPO_FILE environment variable to /opt/aws-ofi-nccl/share/aws-ofi-nccl/xml/p5.48xl-topo.xml
+ip-26-0-160-103:2068012:2068134 [3] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-160-103:2068012:2068134 [3] NCCL INFO NET/OFI Using transport protocol RDMA
+ip-26-0-160-103:2068013:2068138 [4] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.9.1-aws
+ip-26-0-160-103:2068013:2068138 [4] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.9.1-aws
+ip-26-0-160-103:2068013:2068138 [4] NCCL INFO NET/OFI Using Libfabric version 1.21
+ip-26-0-160-103:2068013:2068138 [4] NCCL INFO NET/OFI Using CUDA driver version 12020
+ip-26-0-160-103:2068013:2068138 [4] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-160-103:2068013:2068138 [4] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-160-103:2068013:2068138 [4] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-160-103:2068013:2068138 [4] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-160-103:2068013:2068138 [4] NCCL INFO NET/OFI Running on p5.48xlarge platform, Setting NCCL_TOPO_FILE environment variable to /opt/aws-ofi-nccl/share/aws-ofi-nccl/xml/p5.48xl-topo.xml
+ip-26-0-160-103:2068013:2068138 [4] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-160-103:2068013:2068138 [4] NCCL INFO NET/OFI Using transport protocol RDMA
+ip-26-0-160-103:2068015:2068139 [6] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.9.1-aws
+ip-26-0-160-103:2068015:2068139 [6] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.9.1-aws
+ip-26-0-160-103:2068015:2068139 [6] NCCL INFO NET/OFI Using Libfabric version 1.21
+ip-26-0-160-103:2068015:2068139 [6] NCCL INFO NET/OFI Using CUDA driver version 12020
+ip-26-0-160-103:2068015:2068139 [6] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-160-103:2068015:2068139 [6] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-160-103:2068015:2068139 [6] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-160-103:2068015:2068139 [6] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-160-103:2068015:2068139 [6] NCCL INFO NET/OFI Running on p5.48xlarge platform, Setting NCCL_TOPO_FILE environment variable to /opt/aws-ofi-nccl/share/aws-ofi-nccl/xml/p5.48xl-topo.xml
+ip-26-0-160-103:2068015:2068139 [6] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-160-103:2068015:2068139 [6] NCCL INFO NET/OFI Using transport protocol RDMA
+ip-26-0-160-103:2068010:2068136 [1] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-160-103:2068009:2068132 [0] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-160-103:2068011:2068137 [2] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-160-103:2068016:2068133 [7] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-160-103:2068013:2068138 [4] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-160-103:2068014:2068135 [5] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-160-103:2068012:2068134 [3] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-160-103:2068015:2068139 [6] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-160-103:2068009:2068132 [0] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-160-103:2068009:2068132 [0] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-160-103:2068009:2068132 [0] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-160-103:2068009:2068132 [0] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-160-103:2068009:2068132 [0] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-160-103:2068009:2068132 [0] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-160-103:2068009:2068132 [0] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-160-103:2068009:2068132 [0] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-160-103:2068009:2068132 [0] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-160-103:2068009:2068132 [0] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-160-103:2068009:2068132 [0] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-160-103:2068009:2068132 [0] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-160-103:2068009:2068132 [0] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-160-103:2068009:2068132 [0] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-160-103:2068009:2068132 [0] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-160-103:2068009:2068132 [0] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-160-103:2068009:2068132 [0] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-160-103:2068009:2068132 [0] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-160-103:2068009:2068132 [0] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-160-103:2068009:2068132 [0] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-160-103:2068009:2068132 [0] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-160-103:2068009:2068132 [0] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-160-103:2068009:2068132 [0] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-160-103:2068009:2068132 [0] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-160-103:2068009:2068132 [0] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-160-103:2068009:2068132 [0] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-160-103:2068009:2068132 [0] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-160-103:2068009:2068132 [0] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-160-103:2068009:2068132 [0] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-160-103:2068009:2068132 [0] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-160-103:2068009:2068132 [0] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-160-103:2068009:2068132 [0] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-160-103:2068009:2068132 [0] NCCL INFO Using network AWS Libfabric
+ip-26-0-160-103:2068016:2068133 [7] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-160-103:2068016:2068133 [7] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-160-103:2068016:2068133 [7] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-160-103:2068016:2068133 [7] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-160-103:2068016:2068133 [7] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-160-103:2068016:2068133 [7] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-160-103:2068016:2068133 [7] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-160-103:2068016:2068133 [7] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-160-103:2068016:2068133 [7] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-160-103:2068016:2068133 [7] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-160-103:2068016:2068133 [7] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-160-103:2068016:2068133 [7] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-160-103:2068016:2068133 [7] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-160-103:2068016:2068133 [7] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-160-103:2068016:2068133 [7] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-160-103:2068016:2068133 [7] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-160-103:2068016:2068133 [7] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-160-103:2068016:2068133 [7] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-160-103:2068016:2068133 [7] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-160-103:2068016:2068133 [7] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-160-103:2068016:2068133 [7] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-160-103:2068016:2068133 [7] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-160-103:2068016:2068133 [7] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-160-103:2068016:2068133 [7] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-160-103:2068016:2068133 [7] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-160-103:2068016:2068133 [7] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-160-103:2068016:2068133 [7] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-160-103:2068016:2068133 [7] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-160-103:2068016:2068133 [7] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-160-103:2068016:2068133 [7] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-160-103:2068016:2068133 [7] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-160-103:2068016:2068133 [7] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-160-103:2068010:2068136 [1] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-160-103:2068010:2068136 [1] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-160-103:2068010:2068136 [1] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-160-103:2068010:2068136 [1] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-160-103:2068010:2068136 [1] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-160-103:2068010:2068136 [1] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-160-103:2068010:2068136 [1] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-160-103:2068010:2068136 [1] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-160-103:2068010:2068136 [1] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-160-103:2068010:2068136 [1] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-160-103:2068010:2068136 [1] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-160-103:2068010:2068136 [1] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-160-103:2068010:2068136 [1] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-160-103:2068010:2068136 [1] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-160-103:2068010:2068136 [1] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-160-103:2068010:2068136 [1] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-160-103:2068010:2068136 [1] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-160-103:2068010:2068136 [1] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-160-103:2068010:2068136 [1] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-160-103:2068010:2068136 [1] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-160-103:2068010:2068136 [1] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-160-103:2068010:2068136 [1] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-160-103:2068010:2068136 [1] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-160-103:2068010:2068136 [1] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-160-103:2068010:2068136 [1] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-160-103:2068010:2068136 [1] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-160-103:2068010:2068136 [1] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-160-103:2068010:2068136 [1] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-160-103:2068010:2068136 [1] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-160-103:2068010:2068136 [1] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-160-103:2068010:2068136 [1] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-160-103:2068010:2068136 [1] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-160-103:2068011:2068137 [2] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-160-103:2068011:2068137 [2] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-160-103:2068011:2068137 [2] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-160-103:2068011:2068137 [2] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-160-103:2068011:2068137 [2] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-160-103:2068011:2068137 [2] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-160-103:2068011:2068137 [2] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-160-103:2068011:2068137 [2] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-160-103:2068011:2068137 [2] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-160-103:2068011:2068137 [2] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-160-103:2068011:2068137 [2] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-160-103:2068011:2068137 [2] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-160-103:2068011:2068137 [2] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-160-103:2068011:2068137 [2] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-160-103:2068011:2068137 [2] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-160-103:2068011:2068137 [2] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-160-103:2068011:2068137 [2] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-160-103:2068011:2068137 [2] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-160-103:2068011:2068137 [2] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-160-103:2068011:2068137 [2] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-160-103:2068011:2068137 [2] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-160-103:2068011:2068137 [2] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-160-103:2068011:2068137 [2] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-160-103:2068011:2068137 [2] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-160-103:2068011:2068137 [2] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-160-103:2068011:2068137 [2] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-160-103:2068011:2068137 [2] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-160-103:2068011:2068137 [2] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-160-103:2068011:2068137 [2] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-160-103:2068011:2068137 [2] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-160-103:2068011:2068137 [2] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-160-103:2068011:2068137 [2] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-160-103:2068013:2068138 [4] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-160-103:2068013:2068138 [4] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-160-103:2068013:2068138 [4] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-160-103:2068013:2068138 [4] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-160-103:2068013:2068138 [4] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-160-103:2068013:2068138 [4] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-160-103:2068013:2068138 [4] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-160-103:2068013:2068138 [4] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-160-103:2068013:2068138 [4] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-160-103:2068013:2068138 [4] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-160-103:2068013:2068138 [4] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-160-103:2068013:2068138 [4] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-160-103:2068013:2068138 [4] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-160-103:2068013:2068138 [4] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-160-103:2068013:2068138 [4] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-160-103:2068013:2068138 [4] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-160-103:2068013:2068138 [4] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-160-103:2068013:2068138 [4] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-160-103:2068013:2068138 [4] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-160-103:2068013:2068138 [4] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-160-103:2068013:2068138 [4] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-160-103:2068013:2068138 [4] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-160-103:2068013:2068138 [4] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-160-103:2068013:2068138 [4] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-160-103:2068013:2068138 [4] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-160-103:2068013:2068138 [4] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-160-103:2068013:2068138 [4] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-160-103:2068013:2068138 [4] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-160-103:2068013:2068138 [4] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-160-103:2068013:2068138 [4] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-160-103:2068013:2068138 [4] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-160-103:2068013:2068138 [4] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-160-103:2068014:2068135 [5] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-160-103:2068014:2068135 [5] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-160-103:2068014:2068135 [5] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-160-103:2068014:2068135 [5] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-160-103:2068014:2068135 [5] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-160-103:2068014:2068135 [5] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-160-103:2068014:2068135 [5] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-160-103:2068014:2068135 [5] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-160-103:2068014:2068135 [5] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-160-103:2068014:2068135 [5] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-160-103:2068014:2068135 [5] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-160-103:2068014:2068135 [5] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-160-103:2068014:2068135 [5] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-160-103:2068014:2068135 [5] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-160-103:2068014:2068135 [5] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-160-103:2068014:2068135 [5] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-160-103:2068014:2068135 [5] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-160-103:2068014:2068135 [5] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-160-103:2068014:2068135 [5] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-160-103:2068014:2068135 [5] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-160-103:2068014:2068135 [5] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-160-103:2068014:2068135 [5] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-160-103:2068014:2068135 [5] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-160-103:2068014:2068135 [5] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-160-103:2068014:2068135 [5] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-160-103:2068014:2068135 [5] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-160-103:2068014:2068135 [5] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-160-103:2068014:2068135 [5] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-160-103:2068014:2068135 [5] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-160-103:2068014:2068135 [5] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-160-103:2068014:2068135 [5] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-160-103:2068014:2068135 [5] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-160-103:2068012:2068134 [3] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-160-103:2068012:2068134 [3] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-160-103:2068012:2068134 [3] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-160-103:2068012:2068134 [3] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-160-103:2068012:2068134 [3] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-160-103:2068012:2068134 [3] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-160-103:2068012:2068134 [3] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-160-103:2068012:2068134 [3] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-160-103:2068012:2068134 [3] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-160-103:2068012:2068134 [3] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-160-103:2068012:2068134 [3] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-160-103:2068012:2068134 [3] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-160-103:2068012:2068134 [3] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-160-103:2068012:2068134 [3] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-160-103:2068012:2068134 [3] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-160-103:2068012:2068134 [3] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-160-103:2068012:2068134 [3] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-160-103:2068012:2068134 [3] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-160-103:2068012:2068134 [3] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-160-103:2068012:2068134 [3] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-160-103:2068012:2068134 [3] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-160-103:2068012:2068134 [3] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-160-103:2068012:2068134 [3] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-160-103:2068012:2068134 [3] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-160-103:2068012:2068134 [3] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-160-103:2068012:2068134 [3] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-160-103:2068012:2068134 [3] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-160-103:2068012:2068134 [3] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-160-103:2068012:2068134 [3] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-160-103:2068012:2068134 [3] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-160-103:2068012:2068134 [3] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-160-103:2068012:2068134 [3] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-160-103:2068015:2068139 [6] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-160-103:2068015:2068139 [6] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-160-103:2068015:2068139 [6] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-160-103:2068015:2068139 [6] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-160-103:2068015:2068139 [6] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-160-103:2068015:2068139 [6] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-160-103:2068015:2068139 [6] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-160-103:2068015:2068139 [6] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-160-103:2068015:2068139 [6] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-160-103:2068015:2068139 [6] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-160-103:2068015:2068139 [6] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-160-103:2068015:2068139 [6] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-160-103:2068015:2068139 [6] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-160-103:2068015:2068139 [6] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-160-103:2068015:2068139 [6] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-160-103:2068015:2068139 [6] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-160-103:2068015:2068139 [6] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-160-103:2068015:2068139 [6] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-160-103:2068015:2068139 [6] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-160-103:2068015:2068139 [6] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-160-103:2068015:2068139 [6] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-160-103:2068015:2068139 [6] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-160-103:2068015:2068139 [6] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-160-103:2068015:2068139 [6] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-160-103:2068015:2068139 [6] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-160-103:2068015:2068139 [6] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-160-103:2068015:2068139 [6] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-160-103:2068015:2068139 [6] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-160-103:2068015:2068139 [6] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-160-103:2068015:2068139 [6] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-160-103:2068015:2068139 [6] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-160-103:2068015:2068139 [6] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-160-103:2068016:2068133 [7] NCCL INFO Using network AWS Libfabric
+ip-26-0-160-103:2068010:2068136 [1] NCCL INFO Using network AWS Libfabric
+ip-26-0-160-103:2068013:2068138 [4] NCCL INFO Using network AWS Libfabric
+ip-26-0-160-103:2068012:2068134 [3] NCCL INFO Using network AWS Libfabric
+ip-26-0-160-103:2068014:2068135 [5] NCCL INFO Using network AWS Libfabric
+ip-26-0-160-103:2068011:2068137 [2] NCCL INFO Using network AWS Libfabric
+ip-26-0-160-103:2068015:2068139 [6] NCCL INFO Using network AWS Libfabric
+ip-26-0-160-103:2068014:2068135 [5] NCCL INFO comm 0x93b5970 rank 5 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0x5e761070a40b28a4 - Init START
+ip-26-0-160-103:2068010:2068136 [1] NCCL INFO comm 0x9feabe0 rank 1 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0x5e761070a40b28a4 - Init START
+ip-26-0-160-103:2068011:2068137 [2] NCCL INFO comm 0x9dc7840 rank 2 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0x5e761070a40b28a4 - Init START
+ip-26-0-160-103:2068015:2068139 [6] NCCL INFO comm 0x90b34a0 rank 6 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0x5e761070a40b28a4 - Init START
+ip-26-0-160-103:2068012:2068134 [3] NCCL INFO comm 0xa0e9200 rank 3 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0x5e761070a40b28a4 - Init START
+ip-26-0-160-103:2068016:2068133 [7] NCCL INFO comm 0x93ff880 rank 7 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0x5e761070a40b28a4 - Init START
+ip-26-0-160-103:2068013:2068138 [4] NCCL INFO comm 0x9ad4b40 rank 4 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0x5e761070a40b28a4 - Init START
+ip-26-0-160-103:2068009:2068132 [0] NCCL INFO comm 0x966e410 rank 0 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0x5e761070a40b28a4 - Init START
+ip-26-0-160-103:2068014:2068135 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /opt/aws-ofi-nccl/share/aws-ofi-nccl/xml/p5.48xl-topo.xml
+ip-26-0-160-103:2068012:2068134 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /opt/aws-ofi-nccl/share/aws-ofi-nccl/xml/p5.48xl-topo.xml
+ip-26-0-160-103:2068011:2068137 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /opt/aws-ofi-nccl/share/aws-ofi-nccl/xml/p5.48xl-topo.xml
+ip-26-0-160-103:2068009:2068132 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /opt/aws-ofi-nccl/share/aws-ofi-nccl/xml/p5.48xl-topo.xml
+ip-26-0-160-103:2068013:2068138 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /opt/aws-ofi-nccl/share/aws-ofi-nccl/xml/p5.48xl-topo.xml
+ip-26-0-160-103:2068016:2068133 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /opt/aws-ofi-nccl/share/aws-ofi-nccl/xml/p5.48xl-topo.xml
+ip-26-0-160-103:2068010:2068136 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /opt/aws-ofi-nccl/share/aws-ofi-nccl/xml/p5.48xl-topo.xml
+ip-26-0-160-103:2068015:2068139 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /opt/aws-ofi-nccl/share/aws-ofi-nccl/xml/p5.48xl-topo.xml
+ip-26-0-160-103:2068013:2068138 [4] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068013:2068138 [4] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068013:2068138 [4] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068013:2068138 [4] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068013:2068138 [4] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068013:2068138 [4] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068013:2068138 [4] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068013:2068138 [4] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068013:2068138 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-160-103:2068013:2068138 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-160-103:2068009:2068132 [0] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068009:2068132 [0] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068009:2068132 [0] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068009:2068132 [0] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068009:2068132 [0] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068009:2068132 [0] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068009:2068132 [0] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068009:2068132 [0] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068009:2068132 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-160-103:2068009:2068132 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-160-103:2068011:2068137 [2] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068011:2068137 [2] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068011:2068137 [2] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068011:2068137 [2] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068011:2068137 [2] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068011:2068137 [2] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068011:2068137 [2] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068011:2068137 [2] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068011:2068137 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-160-103:2068011:2068137 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-160-103:2068010:2068136 [1] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068010:2068136 [1] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068010:2068136 [1] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068010:2068136 [1] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068010:2068136 [1] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068010:2068136 [1] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068010:2068136 [1] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068010:2068136 [1] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068010:2068136 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-160-103:2068010:2068136 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-160-103:2068015:2068139 [6] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068015:2068139 [6] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068015:2068139 [6] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068015:2068139 [6] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068015:2068139 [6] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068015:2068139 [6] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068015:2068139 [6] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068015:2068139 [6] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068016:2068133 [7] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068016:2068133 [7] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068016:2068133 [7] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068016:2068133 [7] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068016:2068133 [7] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068016:2068133 [7] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068016:2068133 [7] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068016:2068133 [7] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068015:2068139 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-160-103:2068015:2068139 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-160-103:2068012:2068134 [3] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068012:2068134 [3] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068012:2068134 [3] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068012:2068134 [3] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068012:2068134 [3] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068012:2068134 [3] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068012:2068134 [3] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068012:2068134 [3] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068016:2068133 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-160-103:2068016:2068133 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-160-103:2068014:2068135 [5] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068014:2068135 [5] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068014:2068135 [5] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068012:2068134 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-160-103:2068012:2068134 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-160-103:2068014:2068135 [5] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068014:2068135 [5] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068014:2068135 [5] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068014:2068135 [5] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068014:2068135 [5] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068014:2068135 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-160-103:2068014:2068135 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-160-103:2068014:2068135 [5] NCCL INFO Trees [0] 6/-1/-1->5->4 [1] 6/-1/-1->5->4 [2] 6/-1/-1->5->4 [3] 6/-1/-1->5->4 [4] 6/-1/-1->5->4 [5] 6/-1/-1->5->4 [6] 6/-1/-1->5->4 [7] 6/-1/-1->5->4 [8] 6/-1/-1->5->4 [9] 6/-1/-1->5->4 [10] 6/-1/-1->5->4 [11] 6/-1/-1->5->4 [12] 6/-1/-1->5->4 [13] 6/-1/-1->5->4 [14] 6/-1/-1->5->4 [15] 6/-1/-1->5->4 [16] 6/-1/-1->5->4 [17] 6/-1/-1->5->4 [18] 6/-1/-1->5->4 [19] 6/-1/-1->5->4 [20] 6/-1/-1->5->4 [21] 6/-1/-1->5->4 [22] 6/-1/-1->5->4 [23] 6/-1/-1->5->4
+ip-26-0-160-103:2068014:2068135 [5] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-160-103:2068016:2068133 [7] NCCL INFO Trees [0] -1/-1/-1->7->6 [1] -1/-1/-1->7->6 [2] -1/-1/-1->7->6 [3] -1/-1/-1->7->6 [4] -1/-1/-1->7->6 [5] -1/-1/-1->7->6 [6] -1/-1/-1->7->6 [7] -1/-1/-1->7->6 [8] -1/-1/-1->7->6 [9] -1/-1/-1->7->6 [10] -1/-1/-1->7->6 [11] -1/-1/-1->7->6 [12] -1/-1/-1->7->6 [13] -1/-1/-1->7->6 [14] -1/-1/-1->7->6 [15] -1/-1/-1->7->6 [16] -1/-1/-1->7->6 [17] -1/-1/-1->7->6 [18] -1/-1/-1->7->6 [19] -1/-1/-1->7->6 [20] -1/-1/-1->7->6 [21] -1/-1/-1->7->6 [22] -1/-1/-1->7->6 [23] -1/-1/-1->7->6
+ip-26-0-160-103:2068016:2068133 [7] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-160-103:2068011:2068137 [2] NCCL INFO Trees [0] 3/-1/-1->2->1 [1] 3/-1/-1->2->1 [2] 3/-1/-1->2->1 [3] 3/-1/-1->2->1 [4] 3/-1/-1->2->1 [5] 3/-1/-1->2->1 [6] 3/-1/-1->2->1 [7] 3/-1/-1->2->1 [8] 3/-1/-1->2->1 [9] 3/-1/-1->2->1 [10] 3/-1/-1->2->1 [11] 3/-1/-1->2->1 [12] 3/-1/-1->2->1 [13] 3/-1/-1->2->1 [14] 3/-1/-1->2->1 [15] 3/-1/-1->2->1 [16] 3/-1/-1->2->1 [17] 3/-1/-1->2->1 [18] 3/-1/-1->2->1 [19] 3/-1/-1->2->1 [20] 3/-1/-1->2->1 [21] 3/-1/-1->2->1 [22] 3/-1/-1->2->1 [23] 3/-1/-1->2->1
+ip-26-0-160-103:2068012:2068134 [3] NCCL INFO Trees [0] 4/-1/-1->3->2 [1] 4/-1/-1->3->2 [2] 4/-1/-1->3->2 [3] 4/-1/-1->3->2 [4] 4/-1/-1->3->2 [5] 4/-1/-1->3->2 [6] 4/-1/-1->3->2 [7] 4/-1/-1->3->2 [8] 4/-1/-1->3->2 [9] 4/-1/-1->3->2 [10] 4/-1/-1->3->2 [11] 4/-1/-1->3->2 [12] 4/-1/-1->3->2 [13] 4/-1/-1->3->2 [14] 4/-1/-1->3->2 [15] 4/-1/-1->3->2 [16] 4/-1/-1->3->2 [17] 4/-1/-1->3->2 [18] 4/-1/-1->3->2 [19] 4/-1/-1->3->2 [20] 4/-1/-1->3->2 [21] 4/-1/-1->3->2 [22] 4/-1/-1->3->2 [23] 4/-1/-1->3->2
+ip-26-0-160-103:2068015:2068139 [6] NCCL INFO Trees [0] 7/-1/-1->6->5 [1] 7/-1/-1->6->5 [2] 7/-1/-1->6->5 [3] 7/-1/-1->6->5 [4] 7/-1/-1->6->5 [5] 7/-1/-1->6->5 [6] 7/-1/-1->6->5 [7] 7/-1/-1->6->5 [8] 7/-1/-1->6->5 [9] 7/-1/-1->6->5 [10] 7/-1/-1->6->5 [11] 7/-1/-1->6->5 [12] 7/-1/-1->6->5 [13] 7/-1/-1->6->5 [14] 7/-1/-1->6->5 [15] 7/-1/-1->6->5 [16] 7/-1/-1->6->5 [17] 7/-1/-1->6->5 [18] 7/-1/-1->6->5 [19] 7/-1/-1->6->5 [20] 7/-1/-1->6->5 [21] 7/-1/-1->6->5 [22] 7/-1/-1->6->5 [23] 7/-1/-1->6->5
+ip-26-0-160-103:2068011:2068137 [2] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-160-103:2068010:2068136 [1] NCCL INFO Trees [0] 2/-1/-1->1->0 [1] 2/-1/-1->1->0 [2] 2/-1/-1->1->0 [3] 2/-1/-1->1->0 [4] 2/-1/-1->1->0 [5] 2/-1/-1->1->0 [6] 2/-1/-1->1->0 [7] 2/-1/-1->1->0 [8] 2/-1/-1->1->0 [9] 2/-1/-1->1->0 [10] 2/-1/-1->1->0 [11] 2/-1/-1->1->0 [12] 2/-1/-1->1->0 [13] 2/-1/-1->1->0 [14] 2/-1/-1->1->0 [15] 2/-1/-1->1->0 [16] 2/-1/-1->1->0 [17] 2/-1/-1->1->0 [18] 2/-1/-1->1->0 [19] 2/-1/-1->1->0 [20] 2/-1/-1->1->0 [21] 2/-1/-1->1->0 [22] 2/-1/-1->1->0 [23] 2/-1/-1->1->0
+ip-26-0-160-103:2068012:2068134 [3] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-160-103:2068013:2068138 [4] NCCL INFO Trees [0] 5/-1/-1->4->3 [1] 5/-1/-1->4->3 [2] 5/-1/-1->4->3 [3] 5/-1/-1->4->3 [4] 5/-1/-1->4->3 [5] 5/-1/-1->4->3 [6] 5/-1/-1->4->3 [7] 5/-1/-1->4->3 [8] 5/-1/-1->4->3 [9] 5/-1/-1->4->3 [10] 5/-1/-1->4->3 [11] 5/-1/-1->4->3 [12] 5/-1/-1->4->3 [13] 5/-1/-1->4->3 [14] 5/-1/-1->4->3 [15] 5/-1/-1->4->3 [16] 5/-1/-1->4->3 [17] 5/-1/-1->4->3 [18] 5/-1/-1->4->3 [19] 5/-1/-1->4->3 [20] 5/-1/-1->4->3 [21] 5/-1/-1->4->3 [22] 5/-1/-1->4->3 [23] 5/-1/-1->4->3
+ip-26-0-160-103:2068010:2068136 [1] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-160-103:2068015:2068139 [6] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-160-103:2068013:2068138 [4] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-160-103:2068009:2068132 [0] NCCL INFO Channel 00/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2068009:2068132 [0] NCCL INFO Channel 01/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2068009:2068132 [0] NCCL INFO Channel 02/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2068009:2068132 [0] NCCL INFO Channel 03/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2068009:2068132 [0] NCCL INFO Channel 04/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2068009:2068132 [0] NCCL INFO Channel 05/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2068009:2068132 [0] NCCL INFO Channel 06/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2068009:2068132 [0] NCCL INFO Channel 07/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2068009:2068132 [0] NCCL INFO Channel 08/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2068009:2068132 [0] NCCL INFO Channel 09/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2068009:2068132 [0] NCCL INFO Channel 10/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2068009:2068132 [0] NCCL INFO Channel 11/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2068009:2068132 [0] NCCL INFO Channel 12/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2068009:2068132 [0] NCCL INFO Channel 13/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2068009:2068132 [0] NCCL INFO Channel 14/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2068009:2068132 [0] NCCL INFO Channel 15/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2068009:2068132 [0] NCCL INFO Channel 16/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2068009:2068132 [0] NCCL INFO Channel 17/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2068009:2068132 [0] NCCL INFO Channel 18/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2068009:2068132 [0] NCCL INFO Channel 19/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2068009:2068132 [0] NCCL INFO Channel 20/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2068009:2068132 [0] NCCL INFO Channel 21/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2068009:2068132 [0] NCCL INFO Channel 22/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2068009:2068132 [0] NCCL INFO Channel 23/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2068009:2068132 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] 1/-1/-1->0->-1 [7] 1/-1/-1->0->-1 [8] 1/-1/-1->0->-1 [9] 1/-1/-1->0->-1 [10] 1/-1/-1->0->-1 [11] 1/-1/-1->0->-1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] 1/-1/-1->0->-1 [19] 1/-1/-1->0->-1 [20] 1/-1/-1->0->-1 [21] 1/-1/-1->0->-1 [22] 1/-1/-1->0->-1 [23] 1/-1/-1->0->-1
+ip-26-0-160-103:2068009:2068132 [0] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-160-103:2068014:2068135 [5] NCCL INFO Channel 00/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068016:2068133 [7] NCCL INFO Channel 00/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068012:2068134 [3] NCCL INFO Channel 00/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068011:2068137 [2] NCCL INFO Channel 00/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068014:2068135 [5] NCCL INFO Channel 01/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068016:2068133 [7] NCCL INFO Channel 01/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068012:2068134 [3] NCCL INFO Channel 01/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068014:2068135 [5] NCCL INFO Channel 02/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068014:2068135 [5] NCCL INFO Channel 03/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068014:2068135 [5] NCCL INFO Channel 04/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068014:2068135 [5] NCCL INFO Channel 05/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068016:2068133 [7] NCCL INFO Channel 02/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068014:2068135 [5] NCCL INFO Channel 06/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068016:2068133 [7] NCCL INFO Channel 03/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068014:2068135 [5] NCCL INFO Channel 07/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068016:2068133 [7] NCCL INFO Channel 04/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068014:2068135 [5] NCCL INFO Channel 08/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068011:2068137 [2] NCCL INFO Channel 01/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068016:2068133 [7] NCCL INFO Channel 05/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068012:2068134 [3] NCCL INFO Channel 02/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068014:2068135 [5] NCCL INFO Channel 09/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068011:2068137 [2] NCCL INFO Channel 02/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068016:2068133 [7] NCCL INFO Channel 06/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068012:2068134 [3] NCCL INFO Channel 03/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068014:2068135 [5] NCCL INFO Channel 10/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068013:2068138 [4] NCCL INFO Channel 00/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068016:2068133 [7] NCCL INFO Channel 07/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068010:2068136 [1] NCCL INFO Channel 00/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068011:2068137 [2] NCCL INFO Channel 03/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068012:2068134 [3] NCCL INFO Channel 04/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068014:2068135 [5] NCCL INFO Channel 11/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068013:2068138 [4] NCCL INFO Channel 01/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068015:2068139 [6] NCCL INFO Channel 00/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2068016:2068133 [7] NCCL INFO Channel 08/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068010:2068136 [1] NCCL INFO Channel 01/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068011:2068137 [2] NCCL INFO Channel 04/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068012:2068134 [3] NCCL INFO Channel 05/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068014:2068135 [5] NCCL INFO Channel 12/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068013:2068138 [4] NCCL INFO Channel 02/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068015:2068139 [6] NCCL INFO Channel 01/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2068016:2068133 [7] NCCL INFO Channel 09/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068013:2068138 [4] NCCL INFO Channel 03/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068010:2068136 [1] NCCL INFO Channel 02/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068012:2068134 [3] NCCL INFO Channel 06/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068014:2068135 [5] NCCL INFO Channel 13/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068016:2068133 [7] NCCL INFO Channel 10/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068013:2068138 [4] NCCL INFO Channel 04/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068012:2068134 [3] NCCL INFO Channel 07/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068010:2068136 [1] NCCL INFO Channel 03/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068014:2068135 [5] NCCL INFO Channel 14/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068016:2068133 [7] NCCL INFO Channel 11/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068013:2068138 [4] NCCL INFO Channel 05/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068012:2068134 [3] NCCL INFO Channel 08/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068010:2068136 [1] NCCL INFO Channel 04/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068014:2068135 [5] NCCL INFO Channel 15/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068016:2068133 [7] NCCL INFO Channel 12/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068013:2068138 [4] NCCL INFO Channel 06/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068012:2068134 [3] NCCL INFO Channel 09/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068010:2068136 [1] NCCL INFO Channel 05/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068014:2068135 [5] NCCL INFO Channel 16/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068016:2068133 [7] NCCL INFO Channel 13/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068013:2068138 [4] NCCL INFO Channel 07/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068012:2068134 [3] NCCL INFO Channel 10/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068010:2068136 [1] NCCL INFO Channel 06/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068014:2068135 [5] NCCL INFO Channel 17/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068016:2068133 [7] NCCL INFO Channel 14/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068013:2068138 [4] NCCL INFO Channel 08/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068010:2068136 [1] NCCL INFO Channel 07/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068012:2068134 [3] NCCL INFO Channel 11/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068014:2068135 [5] NCCL INFO Channel 18/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068016:2068133 [7] NCCL INFO Channel 15/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068013:2068138 [4] NCCL INFO Channel 09/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068012:2068134 [3] NCCL INFO Channel 12/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068010:2068136 [1] NCCL INFO Channel 08/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068014:2068135 [5] NCCL INFO Channel 19/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068016:2068133 [7] NCCL INFO Channel 16/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068012:2068134 [3] NCCL INFO Channel 13/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068013:2068138 [4] NCCL INFO Channel 10/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068010:2068136 [1] NCCL INFO Channel 09/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068014:2068135 [5] NCCL INFO Channel 20/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068016:2068133 [7] NCCL INFO Channel 17/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068012:2068134 [3] NCCL INFO Channel 14/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068013:2068138 [4] NCCL INFO Channel 11/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068010:2068136 [1] NCCL INFO Channel 10/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068011:2068137 [2] NCCL INFO Channel 05/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068014:2068135 [5] NCCL INFO Channel 21/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068016:2068133 [7] NCCL INFO Channel 18/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068012:2068134 [3] NCCL INFO Channel 15/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068015:2068139 [6] NCCL INFO Channel 02/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2068013:2068138 [4] NCCL INFO Channel 12/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068010:2068136 [1] NCCL INFO Channel 11/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068011:2068137 [2] NCCL INFO Channel 06/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068014:2068135 [5] NCCL INFO Channel 22/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068016:2068133 [7] NCCL INFO Channel 19/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068015:2068139 [6] NCCL INFO Channel 03/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2068010:2068136 [1] NCCL INFO Channel 12/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068009:2068132 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068011:2068137 [2] NCCL INFO Channel 07/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068015:2068139 [6] NCCL INFO Channel 04/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2068010:2068136 [1] NCCL INFO Channel 13/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068009:2068132 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068011:2068137 [2] NCCL INFO Channel 08/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068015:2068139 [6] NCCL INFO Channel 05/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2068010:2068136 [1] NCCL INFO Channel 14/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068012:2068134 [3] NCCL INFO Channel 16/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068009:2068132 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068011:2068137 [2] NCCL INFO Channel 09/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068015:2068139 [6] NCCL INFO Channel 06/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2068010:2068136 [1] NCCL INFO Channel 15/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068012:2068134 [3] NCCL INFO Channel 17/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068011:2068137 [2] NCCL INFO Channel 10/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068014:2068135 [5] NCCL INFO Channel 23/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068016:2068133 [7] NCCL INFO Channel 20/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068011:2068137 [2] NCCL INFO Channel 11/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068013:2068138 [4] NCCL INFO Channel 13/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068011:2068137 [2] NCCL INFO Channel 12/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068010:2068136 [1] NCCL INFO Channel 16/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068009:2068132 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068015:2068139 [6] NCCL INFO Channel 07/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2068010:2068136 [1] NCCL INFO Channel 17/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068016:2068133 [7] NCCL INFO Channel 21/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068011:2068137 [2] NCCL INFO Channel 13/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068009:2068132 [0] NCCL INFO Channel 04/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068013:2068138 [4] NCCL INFO Channel 14/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068010:2068136 [1] NCCL INFO Channel 18/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068016:2068133 [7] NCCL INFO Channel 22/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068011:2068137 [2] NCCL INFO Channel 14/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068009:2068132 [0] NCCL INFO Channel 05/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068013:2068138 [4] NCCL INFO Channel 15/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068016:2068133 [7] NCCL INFO Channel 23/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068010:2068136 [1] NCCL INFO Channel 19/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068011:2068137 [2] NCCL INFO Channel 15/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068015:2068139 [6] NCCL INFO Channel 08/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2068009:2068132 [0] NCCL INFO Channel 06/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068013:2068138 [4] NCCL INFO Channel 16/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068011:2068137 [2] NCCL INFO Channel 16/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068012:2068134 [3] NCCL INFO Channel 18/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068010:2068136 [1] NCCL INFO Channel 20/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068013:2068138 [4] NCCL INFO Channel 17/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068011:2068137 [2] NCCL INFO Channel 17/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068012:2068134 [3] NCCL INFO Channel 19/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068010:2068136 [1] NCCL INFO Channel 21/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068013:2068138 [4] NCCL INFO Channel 18/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068011:2068137 [2] NCCL INFO Channel 18/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068012:2068134 [3] NCCL INFO Channel 20/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068010:2068136 [1] NCCL INFO Channel 22/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068013:2068138 [4] NCCL INFO Channel 19/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068011:2068137 [2] NCCL INFO Channel 19/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068012:2068134 [3] NCCL INFO Channel 21/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068010:2068136 [1] NCCL INFO Channel 23/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068013:2068138 [4] NCCL INFO Channel 20/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068011:2068137 [2] NCCL INFO Channel 20/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068012:2068134 [3] NCCL INFO Channel 22/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068013:2068138 [4] NCCL INFO Channel 21/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068011:2068137 [2] NCCL INFO Channel 21/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068012:2068134 [3] NCCL INFO Channel 23/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068013:2068138 [4] NCCL INFO Channel 22/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068015:2068139 [6] NCCL INFO Channel 09/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2068011:2068137 [2] NCCL INFO Channel 22/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068013:2068138 [4] NCCL INFO Channel 23/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068015:2068139 [6] NCCL INFO Channel 10/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2068011:2068137 [2] NCCL INFO Channel 23/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068009:2068132 [0] NCCL INFO Channel 07/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068015:2068139 [6] NCCL INFO Channel 11/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2068009:2068132 [0] NCCL INFO Channel 08/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068015:2068139 [6] NCCL INFO Channel 12/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2068009:2068132 [0] NCCL INFO Channel 09/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068015:2068139 [6] NCCL INFO Channel 13/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2068009:2068132 [0] NCCL INFO Channel 10/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068015:2068139 [6] NCCL INFO Channel 14/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2068009:2068132 [0] NCCL INFO Channel 11/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068015:2068139 [6] NCCL INFO Channel 15/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2068009:2068132 [0] NCCL INFO Channel 12/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068015:2068139 [6] NCCL INFO Channel 16/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2068009:2068132 [0] NCCL INFO Channel 13/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068015:2068139 [6] NCCL INFO Channel 17/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2068009:2068132 [0] NCCL INFO Channel 14/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068015:2068139 [6] NCCL INFO Channel 18/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2068009:2068132 [0] NCCL INFO Channel 15/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068015:2068139 [6] NCCL INFO Channel 19/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2068009:2068132 [0] NCCL INFO Channel 16/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068015:2068139 [6] NCCL INFO Channel 20/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2068009:2068132 [0] NCCL INFO Channel 17/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068015:2068139 [6] NCCL INFO Channel 21/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2068009:2068132 [0] NCCL INFO Channel 18/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068015:2068139 [6] NCCL INFO Channel 22/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2068009:2068132 [0] NCCL INFO Channel 19/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068015:2068139 [6] NCCL INFO Channel 23/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2068009:2068132 [0] NCCL INFO Channel 20/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068009:2068132 [0] NCCL INFO Channel 21/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068009:2068132 [0] NCCL INFO Channel 22/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068009:2068132 [0] NCCL INFO Channel 23/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068011:2068137 [2] NCCL INFO Connected all rings
+ip-26-0-160-103:2068013:2068138 [4] NCCL INFO Connected all rings
+ip-26-0-160-103:2068012:2068134 [3] NCCL INFO Connected all rings
+ip-26-0-160-103:2068010:2068136 [1] NCCL INFO Connected all rings
+ip-26-0-160-103:2068015:2068139 [6] NCCL INFO Connected all rings
+ip-26-0-160-103:2068014:2068135 [5] NCCL INFO Connected all rings
+ip-26-0-160-103:2068009:2068132 [0] NCCL INFO Connected all rings
+ip-26-0-160-103:2068016:2068133 [7] NCCL INFO Connected all rings
+ip-26-0-160-103:2068016:2068133 [7] NCCL INFO Channel 00/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068016:2068133 [7] NCCL INFO Channel 01/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068016:2068133 [7] NCCL INFO Channel 02/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068016:2068133 [7] NCCL INFO Channel 03/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068016:2068133 [7] NCCL INFO Channel 04/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068011:2068137 [2] NCCL INFO Channel 00/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068016:2068133 [7] NCCL INFO Channel 05/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068011:2068137 [2] NCCL INFO Channel 01/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068016:2068133 [7] NCCL INFO Channel 06/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068011:2068137 [2] NCCL INFO Channel 02/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068016:2068133 [7] NCCL INFO Channel 07/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068011:2068137 [2] NCCL INFO Channel 03/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068012:2068134 [3] NCCL INFO Channel 00/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068016:2068133 [7] NCCL INFO Channel 08/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068011:2068137 [2] NCCL INFO Channel 04/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068012:2068134 [3] NCCL INFO Channel 01/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068016:2068133 [7] NCCL INFO Channel 09/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068011:2068137 [2] NCCL INFO Channel 05/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068013:2068138 [4] NCCL INFO Channel 00/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068012:2068134 [3] NCCL INFO Channel 02/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068016:2068133 [7] NCCL INFO Channel 10/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068011:2068137 [2] NCCL INFO Channel 06/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068013:2068138 [4] NCCL INFO Channel 01/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068012:2068134 [3] NCCL INFO Channel 03/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068016:2068133 [7] NCCL INFO Channel 11/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068011:2068137 [2] NCCL INFO Channel 07/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068013:2068138 [4] NCCL INFO Channel 02/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068012:2068134 [3] NCCL INFO Channel 04/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068016:2068133 [7] NCCL INFO Channel 12/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068011:2068137 [2] NCCL INFO Channel 08/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068013:2068138 [4] NCCL INFO Channel 03/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068012:2068134 [3] NCCL INFO Channel 05/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068016:2068133 [7] NCCL INFO Channel 13/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068011:2068137 [2] NCCL INFO Channel 09/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068013:2068138 [4] NCCL INFO Channel 04/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068012:2068134 [3] NCCL INFO Channel 06/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068016:2068133 [7] NCCL INFO Channel 14/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068011:2068137 [2] NCCL INFO Channel 10/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068013:2068138 [4] NCCL INFO Channel 05/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068012:2068134 [3] NCCL INFO Channel 07/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068010:2068136 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068016:2068133 [7] NCCL INFO Channel 15/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068011:2068137 [2] NCCL INFO Channel 11/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068013:2068138 [4] NCCL INFO Channel 06/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068015:2068139 [6] NCCL INFO Channel 00/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068012:2068134 [3] NCCL INFO Channel 08/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068010:2068136 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068016:2068133 [7] NCCL INFO Channel 16/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068011:2068137 [2] NCCL INFO Channel 12/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068013:2068138 [4] NCCL INFO Channel 07/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068015:2068139 [6] NCCL INFO Channel 01/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068012:2068134 [3] NCCL INFO Channel 09/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068010:2068136 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068011:2068137 [2] NCCL INFO Channel 13/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068016:2068133 [7] NCCL INFO Channel 17/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068013:2068138 [4] NCCL INFO Channel 08/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068015:2068139 [6] NCCL INFO Channel 02/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068012:2068134 [3] NCCL INFO Channel 10/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068010:2068136 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068011:2068137 [2] NCCL INFO Channel 14/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068016:2068133 [7] NCCL INFO Channel 18/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068013:2068138 [4] NCCL INFO Channel 09/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068015:2068139 [6] NCCL INFO Channel 03/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068012:2068134 [3] NCCL INFO Channel 11/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068010:2068136 [1] NCCL INFO Channel 04/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068016:2068133 [7] NCCL INFO Channel 19/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068011:2068137 [2] NCCL INFO Channel 15/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068013:2068138 [4] NCCL INFO Channel 10/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068015:2068139 [6] NCCL INFO Channel 04/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068012:2068134 [3] NCCL INFO Channel 12/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068010:2068136 [1] NCCL INFO Channel 05/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068016:2068133 [7] NCCL INFO Channel 20/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068011:2068137 [2] NCCL INFO Channel 16/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068013:2068138 [4] NCCL INFO Channel 11/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068015:2068139 [6] NCCL INFO Channel 05/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068012:2068134 [3] NCCL INFO Channel 13/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068010:2068136 [1] NCCL INFO Channel 06/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068016:2068133 [7] NCCL INFO Channel 21/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068011:2068137 [2] NCCL INFO Channel 17/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068014:2068135 [5] NCCL INFO Channel 00/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068013:2068138 [4] NCCL INFO Channel 12/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068015:2068139 [6] NCCL INFO Channel 06/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068012:2068134 [3] NCCL INFO Channel 14/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068010:2068136 [1] NCCL INFO Channel 07/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068016:2068133 [7] NCCL INFO Channel 22/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068011:2068137 [2] NCCL INFO Channel 18/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068014:2068135 [5] NCCL INFO Channel 01/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068013:2068138 [4] NCCL INFO Channel 13/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068015:2068139 [6] NCCL INFO Channel 07/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068012:2068134 [3] NCCL INFO Channel 15/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068010:2068136 [1] NCCL INFO Channel 08/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068016:2068133 [7] NCCL INFO Channel 23/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068011:2068137 [2] NCCL INFO Channel 19/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068014:2068135 [5] NCCL INFO Channel 02/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068013:2068138 [4] NCCL INFO Channel 14/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068015:2068139 [6] NCCL INFO Channel 08/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068012:2068134 [3] NCCL INFO Channel 16/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068010:2068136 [1] NCCL INFO Channel 09/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068011:2068137 [2] NCCL INFO Channel 20/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068014:2068135 [5] NCCL INFO Channel 03/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068013:2068138 [4] NCCL INFO Channel 15/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068015:2068139 [6] NCCL INFO Channel 09/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068012:2068134 [3] NCCL INFO Channel 17/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068010:2068136 [1] NCCL INFO Channel 10/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068011:2068137 [2] NCCL INFO Channel 21/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068014:2068135 [5] NCCL INFO Channel 04/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068013:2068138 [4] NCCL INFO Channel 16/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068015:2068139 [6] NCCL INFO Channel 10/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068012:2068134 [3] NCCL INFO Channel 18/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068010:2068136 [1] NCCL INFO Channel 11/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068011:2068137 [2] NCCL INFO Channel 22/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068014:2068135 [5] NCCL INFO Channel 05/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068013:2068138 [4] NCCL INFO Channel 17/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068015:2068139 [6] NCCL INFO Channel 11/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068012:2068134 [3] NCCL INFO Channel 19/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068010:2068136 [1] NCCL INFO Channel 12/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068011:2068137 [2] NCCL INFO Channel 23/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068013:2068138 [4] NCCL INFO Channel 18/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068014:2068135 [5] NCCL INFO Channel 06/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068015:2068139 [6] NCCL INFO Channel 12/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068012:2068134 [3] NCCL INFO Channel 20/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068010:2068136 [1] NCCL INFO Channel 13/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068013:2068138 [4] NCCL INFO Channel 19/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068014:2068135 [5] NCCL INFO Channel 07/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068015:2068139 [6] NCCL INFO Channel 13/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068012:2068134 [3] NCCL INFO Channel 21/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068010:2068136 [1] NCCL INFO Channel 14/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068013:2068138 [4] NCCL INFO Channel 20/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068014:2068135 [5] NCCL INFO Channel 08/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068015:2068139 [6] NCCL INFO Channel 14/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068010:2068136 [1] NCCL INFO Channel 15/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068014:2068135 [5] NCCL INFO Channel 09/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068015:2068139 [6] NCCL INFO Channel 15/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068012:2068134 [3] NCCL INFO Channel 22/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068010:2068136 [1] NCCL INFO Channel 16/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068014:2068135 [5] NCCL INFO Channel 10/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068015:2068139 [6] NCCL INFO Channel 16/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068012:2068134 [3] NCCL INFO Channel 23/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068010:2068136 [1] NCCL INFO Channel 17/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068014:2068135 [5] NCCL INFO Channel 11/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068015:2068139 [6] NCCL INFO Channel 17/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068010:2068136 [1] NCCL INFO Channel 18/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068014:2068135 [5] NCCL INFO Channel 12/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068015:2068139 [6] NCCL INFO Channel 18/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068013:2068138 [4] NCCL INFO Channel 21/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068010:2068136 [1] NCCL INFO Channel 19/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068014:2068135 [5] NCCL INFO Channel 13/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068015:2068139 [6] NCCL INFO Channel 19/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068013:2068138 [4] NCCL INFO Channel 22/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068010:2068136 [1] NCCL INFO Channel 20/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068014:2068135 [5] NCCL INFO Channel 14/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068015:2068139 [6] NCCL INFO Channel 20/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068013:2068138 [4] NCCL INFO Channel 23/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068010:2068136 [1] NCCL INFO Channel 21/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068014:2068135 [5] NCCL INFO Channel 15/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068015:2068139 [6] NCCL INFO Channel 21/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068010:2068136 [1] NCCL INFO Channel 22/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068014:2068135 [5] NCCL INFO Channel 16/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068015:2068139 [6] NCCL INFO Channel 22/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068010:2068136 [1] NCCL INFO Channel 23/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068014:2068135 [5] NCCL INFO Channel 17/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068015:2068139 [6] NCCL INFO Channel 23/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068014:2068135 [5] NCCL INFO Channel 18/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068014:2068135 [5] NCCL INFO Channel 19/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068014:2068135 [5] NCCL INFO Channel 20/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068014:2068135 [5] NCCL INFO Channel 21/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068014:2068135 [5] NCCL INFO Channel 22/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068014:2068135 [5] NCCL INFO Channel 23/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068016:2068133 [7] NCCL INFO Connected all trees
+ip-26-0-160-103:2068016:2068133 [7] NCCL INFO NVLS comm 0x93ff880 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-103:2068009:2068132 [0] NCCL INFO Connected all trees
+ip-26-0-160-103:2068009:2068132 [0] NCCL INFO NVLS comm 0x966e410 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-103:2068012:2068134 [3] NCCL INFO Connected all trees
+ip-26-0-160-103:2068012:2068134 [3] NCCL INFO NVLS comm 0xa0e9200 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-103:2068010:2068136 [1] NCCL INFO Connected all trees
+ip-26-0-160-103:2068010:2068136 [1] NCCL INFO NVLS comm 0x9feabe0 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-103:2068011:2068137 [2] NCCL INFO Connected all trees
+ip-26-0-160-103:2068011:2068137 [2] NCCL INFO NVLS comm 0x9dc7840 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-103:2068013:2068138 [4] NCCL INFO Connected all trees
+ip-26-0-160-103:2068013:2068138 [4] NCCL INFO NVLS comm 0x9ad4b40 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-103:2068015:2068139 [6] NCCL INFO Connected all trees
+ip-26-0-160-103:2068015:2068139 [6] NCCL INFO NVLS comm 0x90b34a0 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-103:2068014:2068135 [5] NCCL INFO Connected all trees
+ip-26-0-160-103:2068014:2068135 [5] NCCL INFO NVLS comm 0x93b5970 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-103:2068010:2068136 [1] NCCL INFO Connected NVLS tree
+ip-26-0-160-103:2068010:2068136 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-103:2068010:2068136 [1] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-160-103:2068010:2068136 [1] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-160-103:2068014:2068135 [5] NCCL INFO Connected NVLS tree
+ip-26-0-160-103:2068014:2068135 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-103:2068014:2068135 [5] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-160-103:2068014:2068135 [5] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-160-103:2068016:2068133 [7] NCCL INFO Connected NVLS tree
+ip-26-0-160-103:2068016:2068133 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-103:2068016:2068133 [7] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-160-103:2068016:2068133 [7] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-160-103:2068012:2068134 [3] NCCL INFO Connected NVLS tree
+ip-26-0-160-103:2068012:2068134 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-103:2068012:2068134 [3] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-160-103:2068012:2068134 [3] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-160-103:2068013:2068138 [4] NCCL INFO Connected NVLS tree
+ip-26-0-160-103:2068013:2068138 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-103:2068013:2068138 [4] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-160-103:2068013:2068138 [4] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-160-103:2068011:2068137 [2] NCCL INFO Connected NVLS tree
+ip-26-0-160-103:2068011:2068137 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-103:2068011:2068137 [2] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-160-103:2068011:2068137 [2] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-160-103:2068009:2068132 [0] NCCL INFO Connected NVLS tree
+ip-26-0-160-103:2068009:2068132 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-103:2068015:2068139 [6] NCCL INFO Connected NVLS tree
+ip-26-0-160-103:2068009:2068132 [0] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-160-103:2068009:2068132 [0] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-160-103:2068015:2068139 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-103:2068015:2068139 [6] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-160-103:2068015:2068139 [6] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-160-103:2068010:2068136 [1] NCCL INFO comm 0x9feabe0 rank 1 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0x5e761070a40b28a4 - Init COMPLETE
+ip-26-0-160-103:2068014:2068135 [5] NCCL INFO comm 0x93b5970 rank 5 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0x5e761070a40b28a4 - Init COMPLETE
+ip-26-0-160-103:2068016:2068133 [7] NCCL INFO comm 0x93ff880 rank 7 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0x5e761070a40b28a4 - Init COMPLETE
+ip-26-0-160-103:2068013:2068138 [4] NCCL INFO comm 0x9ad4b40 rank 4 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0x5e761070a40b28a4 - Init COMPLETE
+ip-26-0-160-103:2068012:2068134 [3] NCCL INFO comm 0xa0e9200 rank 3 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0x5e761070a40b28a4 - Init COMPLETE
+ip-26-0-160-103:2068011:2068137 [2] NCCL INFO comm 0x9dc7840 rank 2 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0x5e761070a40b28a4 - Init COMPLETE
+ip-26-0-160-103:2068009:2068132 [0] NCCL INFO comm 0x966e410 rank 0 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0x5e761070a40b28a4 - Init COMPLETE
+ip-26-0-160-103:2068015:2068139 [6] NCCL INFO comm 0x90b34a0 rank 6 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0x5e761070a40b28a4 - Init COMPLETE
+ip-26-0-160-103:2068009:2068202 [0] NCCL INFO Using network AWS Libfabric
+ip-26-0-160-103:2068014:2068205 [5] NCCL INFO Using network AWS Libfabric
+ip-26-0-160-103:2068010:2068204 [1] NCCL INFO Using network AWS Libfabric
+ip-26-0-160-103:2068015:2068208 [6] NCCL INFO Using network AWS Libfabric
+ip-26-0-160-103:2068011:2068209 [2] NCCL INFO Using network AWS Libfabric
+ip-26-0-160-103:2068016:2068206 [7] NCCL INFO Using network AWS Libfabric
+ip-26-0-160-103:2068012:2068207 [3] NCCL INFO Using network AWS Libfabric
+ip-26-0-160-103:2068013:2068203 [4] NCCL INFO Using network AWS Libfabric
+ip-26-0-160-103:2068010:2068204 [1] NCCL INFO comm 0xa214810 rank 1 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0xcc4f0ed1a7bd4c20 - Init START
+ip-26-0-160-103:2068011:2068209 [2] NCCL INFO comm 0x9ff19a0 rank 2 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0xcc4f0ed1a7bd4c20 - Init START
+ip-26-0-160-103:2068013:2068203 [4] NCCL INFO comm 0x9cfe500 rank 4 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0xcc4f0ed1a7bd4c20 - Init START
+ip-26-0-160-103:2068016:2068206 [7] NCCL INFO comm 0x96292f0 rank 7 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0xcc4f0ed1a7bd4c20 - Init START
+ip-26-0-160-103:2068014:2068205 [5] NCCL INFO comm 0x95df8e0 rank 5 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0xcc4f0ed1a7bd4c20 - Init START
+ip-26-0-160-103:2068015:2068208 [6] NCCL INFO comm 0x92dd050 rank 6 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0xcc4f0ed1a7bd4c20 - Init START
+ip-26-0-160-103:2068009:2068202 [0] NCCL INFO comm 0x9897c70 rank 0 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0xcc4f0ed1a7bd4c20 - Init START
+ip-26-0-160-103:2068012:2068207 [3] NCCL INFO comm 0xa312d40 rank 3 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0xcc4f0ed1a7bd4c20 - Init START
+ip-26-0-160-103:2068010:2068204 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /opt/aws-ofi-nccl/share/aws-ofi-nccl/xml/p5.48xl-topo.xml
+ip-26-0-160-103:2068011:2068209 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /opt/aws-ofi-nccl/share/aws-ofi-nccl/xml/p5.48xl-topo.xml
+ip-26-0-160-103:2068016:2068206 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /opt/aws-ofi-nccl/share/aws-ofi-nccl/xml/p5.48xl-topo.xml
+ip-26-0-160-103:2068014:2068205 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /opt/aws-ofi-nccl/share/aws-ofi-nccl/xml/p5.48xl-topo.xml
+ip-26-0-160-103:2068015:2068208 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /opt/aws-ofi-nccl/share/aws-ofi-nccl/xml/p5.48xl-topo.xml
+ip-26-0-160-103:2068009:2068202 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /opt/aws-ofi-nccl/share/aws-ofi-nccl/xml/p5.48xl-topo.xml
+ip-26-0-160-103:2068013:2068203 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /opt/aws-ofi-nccl/share/aws-ofi-nccl/xml/p5.48xl-topo.xml
+ip-26-0-160-103:2068012:2068207 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /opt/aws-ofi-nccl/share/aws-ofi-nccl/xml/p5.48xl-topo.xml
+ip-26-0-160-103:2068015:2068208 [6] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068015:2068208 [6] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068015:2068208 [6] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068015:2068208 [6] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068015:2068208 [6] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068015:2068208 [6] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068015:2068208 [6] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068015:2068208 [6] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068015:2068208 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-160-103:2068015:2068208 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-160-103:2068014:2068205 [5] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068014:2068205 [5] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068014:2068205 [5] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068014:2068205 [5] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068014:2068205 [5] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068014:2068205 [5] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068014:2068205 [5] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068014:2068205 [5] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068014:2068205 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-160-103:2068014:2068205 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-160-103:2068012:2068207 [3] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068012:2068207 [3] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068012:2068207 [3] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068012:2068207 [3] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068012:2068207 [3] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068012:2068207 [3] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068012:2068207 [3] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068012:2068207 [3] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068012:2068207 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-160-103:2068012:2068207 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-160-103:2068013:2068203 [4] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068013:2068203 [4] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068013:2068203 [4] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068013:2068203 [4] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068013:2068203 [4] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068013:2068203 [4] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068013:2068203 [4] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068013:2068203 [4] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068013:2068203 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-160-103:2068013:2068203 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-160-103:2068009:2068202 [0] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068009:2068202 [0] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068009:2068202 [0] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068009:2068202 [0] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068009:2068202 [0] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068009:2068202 [0] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068009:2068202 [0] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068009:2068202 [0] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068011:2068209 [2] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068011:2068209 [2] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068011:2068209 [2] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068011:2068209 [2] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068011:2068209 [2] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068011:2068209 [2] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068011:2068209 [2] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068009:2068202 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-160-103:2068011:2068209 [2] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068009:2068202 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-160-103:2068011:2068209 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-160-103:2068011:2068209 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-160-103:2068016:2068206 [7] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068016:2068206 [7] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068016:2068206 [7] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068016:2068206 [7] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068016:2068206 [7] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068016:2068206 [7] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068016:2068206 [7] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068016:2068206 [7] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068010:2068204 [1] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068010:2068204 [1] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068010:2068204 [1] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068010:2068204 [1] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068010:2068204 [1] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068010:2068204 [1] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068010:2068204 [1] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068010:2068204 [1] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068016:2068206 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-160-103:2068016:2068206 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-160-103:2068010:2068204 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-160-103:2068010:2068204 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-160-103:2068014:2068205 [5] NCCL INFO Trees [0] 6/-1/-1->5->4 [1] 6/-1/-1->5->4 [2] 6/-1/-1->5->4 [3] 6/-1/-1->5->4 [4] 6/-1/-1->5->4 [5] 6/-1/-1->5->4 [6] 6/-1/-1->5->4 [7] 6/-1/-1->5->4 [8] 6/-1/-1->5->4 [9] 6/-1/-1->5->4 [10] 6/-1/-1->5->4 [11] 6/-1/-1->5->4 [12] 6/-1/-1->5->4 [13] 6/-1/-1->5->4 [14] 6/-1/-1->5->4 [15] 6/-1/-1->5->4 [16] 6/-1/-1->5->4 [17] 6/-1/-1->5->4 [18] 6/-1/-1->5->4 [19] 6/-1/-1->5->4 [20] 6/-1/-1->5->4 [21] 6/-1/-1->5->4 [22] 6/-1/-1->5->4 [23] 6/-1/-1->5->4
+ip-26-0-160-103:2068014:2068205 [5] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-160-103:2068013:2068203 [4] NCCL INFO Trees [0] 5/-1/-1->4->3 [1] 5/-1/-1->4->3 [2] 5/-1/-1->4->3 [3] 5/-1/-1->4->3 [4] 5/-1/-1->4->3 [5] 5/-1/-1->4->3 [6] 5/-1/-1->4->3 [7] 5/-1/-1->4->3 [8] 5/-1/-1->4->3 [9] 5/-1/-1->4->3 [10] 5/-1/-1->4->3 [11] 5/-1/-1->4->3 [12] 5/-1/-1->4->3 [13] 5/-1/-1->4->3 [14] 5/-1/-1->4->3 [15] 5/-1/-1->4->3 [16] 5/-1/-1->4->3 [17] 5/-1/-1->4->3 [18] 5/-1/-1->4->3 [19] 5/-1/-1->4->3 [20] 5/-1/-1->4->3 [21] 5/-1/-1->4->3 [22] 5/-1/-1->4->3 [23] 5/-1/-1->4->3
+ip-26-0-160-103:2068013:2068203 [4] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-160-103:2068012:2068207 [3] NCCL INFO Trees [0] 4/-1/-1->3->2 [1] 4/-1/-1->3->2 [2] 4/-1/-1->3->2 [3] 4/-1/-1->3->2 [4] 4/-1/-1->3->2 [5] 4/-1/-1->3->2 [6] 4/-1/-1->3->2 [7] 4/-1/-1->3->2 [8] 4/-1/-1->3->2 [9] 4/-1/-1->3->2 [10] 4/-1/-1->3->2 [11] 4/-1/-1->3->2 [12] 4/-1/-1->3->2 [13] 4/-1/-1->3->2 [14] 4/-1/-1->3->2 [15] 4/-1/-1->3->2 [16] 4/-1/-1->3->2 [17] 4/-1/-1->3->2 [18] 4/-1/-1->3->2 [19] 4/-1/-1->3->2 [20] 4/-1/-1->3->2 [21] 4/-1/-1->3->2 [22] 4/-1/-1->3->2 [23] 4/-1/-1->3->2
+ip-26-0-160-103:2068012:2068207 [3] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-160-103:2068010:2068204 [1] NCCL INFO Trees [0] 2/-1/-1->1->0 [1] 2/-1/-1->1->0 [2] 2/-1/-1->1->0 [3] 2/-1/-1->1->0 [4] 2/-1/-1->1->0 [5] 2/-1/-1->1->0 [6] 2/-1/-1->1->0 [7] 2/-1/-1->1->0 [8] 2/-1/-1->1->0 [9] 2/-1/-1->1->0 [10] 2/-1/-1->1->0 [11] 2/-1/-1->1->0 [12] 2/-1/-1->1->0 [13] 2/-1/-1->1->0 [14] 2/-1/-1->1->0 [15] 2/-1/-1->1->0 [16] 2/-1/-1->1->0 [17] 2/-1/-1->1->0 [18] 2/-1/-1->1->0 [19] 2/-1/-1->1->0 [20] 2/-1/-1->1->0 [21] 2/-1/-1->1->0 [22] 2/-1/-1->1->0 [23] 2/-1/-1->1->0
+ip-26-0-160-103:2068010:2068204 [1] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-160-103:2068011:2068209 [2] NCCL INFO Trees [0] 3/-1/-1->2->1 [1] 3/-1/-1->2->1 [2] 3/-1/-1->2->1 [3] 3/-1/-1->2->1 [4] 3/-1/-1->2->1 [5] 3/-1/-1->2->1 [6] 3/-1/-1->2->1 [7] 3/-1/-1->2->1 [8] 3/-1/-1->2->1 [9] 3/-1/-1->2->1 [10] 3/-1/-1->2->1 [11] 3/-1/-1->2->1 [12] 3/-1/-1->2->1 [13] 3/-1/-1->2->1 [14] 3/-1/-1->2->1 [15] 3/-1/-1->2->1 [16] 3/-1/-1->2->1 [17] 3/-1/-1->2->1 [18] 3/-1/-1->2->1 [19] 3/-1/-1->2->1 [20] 3/-1/-1->2->1 [21] 3/-1/-1->2->1 [22] 3/-1/-1->2->1 [23] 3/-1/-1->2->1
+ip-26-0-160-103:2068011:2068209 [2] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-160-103:2068009:2068202 [0] NCCL INFO Channel 00/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2068009:2068202 [0] NCCL INFO Channel 01/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2068009:2068202 [0] NCCL INFO Channel 02/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2068009:2068202 [0] NCCL INFO Channel 03/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2068009:2068202 [0] NCCL INFO Channel 04/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2068009:2068202 [0] NCCL INFO Channel 05/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2068016:2068206 [7] NCCL INFO Trees [0] -1/-1/-1->7->6 [1] -1/-1/-1->7->6 [2] -1/-1/-1->7->6 [3] -1/-1/-1->7->6 [4] -1/-1/-1->7->6 [5] -1/-1/-1->7->6 [6] -1/-1/-1->7->6 [7] -1/-1/-1->7->6 [8] -1/-1/-1->7->6 [9] -1/-1/-1->7->6 [10] -1/-1/-1->7->6 [11] -1/-1/-1->7->6 [12] -1/-1/-1->7->6 [13] -1/-1/-1->7->6 [14] -1/-1/-1->7->6 [15] -1/-1/-1->7->6 [16] -1/-1/-1->7->6 [17] -1/-1/-1->7->6 [18] -1/-1/-1->7->6 [19] -1/-1/-1->7->6 [20] -1/-1/-1->7->6 [21] -1/-1/-1->7->6 [22] -1/-1/-1->7->6 [23] -1/-1/-1->7->6
+ip-26-0-160-103:2068009:2068202 [0] NCCL INFO Channel 06/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2068016:2068206 [7] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-160-103:2068009:2068202 [0] NCCL INFO Channel 07/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2068009:2068202 [0] NCCL INFO Channel 08/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2068009:2068202 [0] NCCL INFO Channel 09/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2068009:2068202 [0] NCCL INFO Channel 10/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2068009:2068202 [0] NCCL INFO Channel 11/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2068009:2068202 [0] NCCL INFO Channel 12/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2068009:2068202 [0] NCCL INFO Channel 13/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2068009:2068202 [0] NCCL INFO Channel 14/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2068009:2068202 [0] NCCL INFO Channel 15/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2068009:2068202 [0] NCCL INFO Channel 16/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2068009:2068202 [0] NCCL INFO Channel 17/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2068009:2068202 [0] NCCL INFO Channel 18/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2068009:2068202 [0] NCCL INFO Channel 19/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2068009:2068202 [0] NCCL INFO Channel 20/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2068015:2068208 [6] NCCL INFO Trees [0] 7/-1/-1->6->5 [1] 7/-1/-1->6->5 [2] 7/-1/-1->6->5 [3] 7/-1/-1->6->5 [4] 7/-1/-1->6->5 [5] 7/-1/-1->6->5 [6] 7/-1/-1->6->5 [7] 7/-1/-1->6->5 [8] 7/-1/-1->6->5 [9] 7/-1/-1->6->5 [10] 7/-1/-1->6->5 [11] 7/-1/-1->6->5 [12] 7/-1/-1->6->5 [13] 7/-1/-1->6->5 [14] 7/-1/-1->6->5 [15] 7/-1/-1->6->5 [16] 7/-1/-1->6->5 [17] 7/-1/-1->6->5 [18] 7/-1/-1->6->5 [19] 7/-1/-1->6->5 [20] 7/-1/-1->6->5 [21] 7/-1/-1->6->5 [22] 7/-1/-1->6->5 [23] 7/-1/-1->6->5
+ip-26-0-160-103:2068009:2068202 [0] NCCL INFO Channel 21/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2068009:2068202 [0] NCCL INFO Channel 22/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2068009:2068202 [0] NCCL INFO Channel 23/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2068015:2068208 [6] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-160-103:2068009:2068202 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] 1/-1/-1->0->-1 [7] 1/-1/-1->0->-1 [8] 1/-1/-1->0->-1 [9] 1/-1/-1->0->-1 [10] 1/-1/-1->0->-1 [11] 1/-1/-1->0->-1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] 1/-1/-1->0->-1 [19] 1/-1/-1->0->-1 [20] 1/-1/-1->0->-1 [21] 1/-1/-1->0->-1 [22] 1/-1/-1->0->-1 [23] 1/-1/-1->0->-1
+ip-26-0-160-103:2068009:2068202 [0] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-160-103:2068011:2068209 [2] NCCL INFO Channel 00/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068011:2068209 [2] NCCL INFO Channel 01/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068013:2068203 [4] NCCL INFO Channel 00/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068011:2068209 [2] NCCL INFO Channel 02/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068013:2068203 [4] NCCL INFO Channel 01/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068010:2068204 [1] NCCL INFO Channel 00/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068013:2068203 [4] NCCL INFO Channel 02/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068010:2068204 [1] NCCL INFO Channel 01/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068011:2068209 [2] NCCL INFO Channel 03/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068010:2068204 [1] NCCL INFO Channel 02/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068011:2068209 [2] NCCL INFO Channel 04/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068009:2068202 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068010:2068204 [1] NCCL INFO Channel 03/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068011:2068209 [2] NCCL INFO Channel 05/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068012:2068207 [3] NCCL INFO Channel 00/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068009:2068202 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068010:2068204 [1] NCCL INFO Channel 04/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068011:2068209 [2] NCCL INFO Channel 06/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068012:2068207 [3] NCCL INFO Channel 01/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068015:2068208 [6] NCCL INFO Channel 00/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2068009:2068202 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068010:2068204 [1] NCCL INFO Channel 05/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068011:2068209 [2] NCCL INFO Channel 07/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068012:2068207 [3] NCCL INFO Channel 02/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068015:2068208 [6] NCCL INFO Channel 01/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2068009:2068202 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068010:2068204 [1] NCCL INFO Channel 06/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068011:2068209 [2] NCCL INFO Channel 08/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068012:2068207 [3] NCCL INFO Channel 03/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068015:2068208 [6] NCCL INFO Channel 02/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2068009:2068202 [0] NCCL INFO Channel 04/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068010:2068204 [1] NCCL INFO Channel 07/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068011:2068209 [2] NCCL INFO Channel 09/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068012:2068207 [3] NCCL INFO Channel 04/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068013:2068203 [4] NCCL INFO Channel 03/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068015:2068208 [6] NCCL INFO Channel 03/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2068009:2068202 [0] NCCL INFO Channel 05/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068014:2068205 [5] NCCL INFO Channel 00/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068010:2068204 [1] NCCL INFO Channel 08/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068011:2068209 [2] NCCL INFO Channel 10/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068012:2068207 [3] NCCL INFO Channel 05/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068013:2068203 [4] NCCL INFO Channel 04/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068015:2068208 [6] NCCL INFO Channel 04/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2068009:2068202 [0] NCCL INFO Channel 06/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068014:2068205 [5] NCCL INFO Channel 01/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068011:2068209 [2] NCCL INFO Channel 11/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068010:2068204 [1] NCCL INFO Channel 09/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068012:2068207 [3] NCCL INFO Channel 06/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068013:2068203 [4] NCCL INFO Channel 05/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068015:2068208 [6] NCCL INFO Channel 05/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2068011:2068209 [2] NCCL INFO Channel 12/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068009:2068202 [0] NCCL INFO Channel 07/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068016:2068206 [7] NCCL INFO Channel 00/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068015:2068208 [6] NCCL INFO Channel 06/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2068011:2068209 [2] NCCL INFO Channel 13/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068009:2068202 [0] NCCL INFO Channel 08/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068016:2068206 [7] NCCL INFO Channel 01/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068015:2068208 [6] NCCL INFO Channel 07/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2068011:2068209 [2] NCCL INFO Channel 14/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068009:2068202 [0] NCCL INFO Channel 09/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068016:2068206 [7] NCCL INFO Channel 02/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068015:2068208 [6] NCCL INFO Channel 08/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2068011:2068209 [2] NCCL INFO Channel 15/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068009:2068202 [0] NCCL INFO Channel 10/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068016:2068206 [7] NCCL INFO Channel 03/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068011:2068209 [2] NCCL INFO Channel 16/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068015:2068208 [6] NCCL INFO Channel 09/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2068009:2068202 [0] NCCL INFO Channel 11/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068016:2068206 [7] NCCL INFO Channel 04/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068014:2068205 [5] NCCL INFO Channel 02/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068010:2068204 [1] NCCL INFO Channel 10/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068012:2068207 [3] NCCL INFO Channel 07/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068011:2068209 [2] NCCL INFO Channel 17/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068015:2068208 [6] NCCL INFO Channel 10/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2068013:2068203 [4] NCCL INFO Channel 06/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068009:2068202 [0] NCCL INFO Channel 12/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068016:2068206 [7] NCCL INFO Channel 05/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068014:2068205 [5] NCCL INFO Channel 03/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068010:2068204 [1] NCCL INFO Channel 11/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068012:2068207 [3] NCCL INFO Channel 08/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068011:2068209 [2] NCCL INFO Channel 18/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068015:2068208 [6] NCCL INFO Channel 11/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2068013:2068203 [4] NCCL INFO Channel 07/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068009:2068202 [0] NCCL INFO Channel 13/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068016:2068206 [7] NCCL INFO Channel 06/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068014:2068205 [5] NCCL INFO Channel 04/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068010:2068204 [1] NCCL INFO Channel 12/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068012:2068207 [3] NCCL INFO Channel 09/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068015:2068208 [6] NCCL INFO Channel 12/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2068011:2068209 [2] NCCL INFO Channel 19/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068016:2068206 [7] NCCL INFO Channel 07/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068013:2068203 [4] NCCL INFO Channel 08/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068009:2068202 [0] NCCL INFO Channel 14/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068014:2068205 [5] NCCL INFO Channel 05/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068010:2068204 [1] NCCL INFO Channel 13/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068012:2068207 [3] NCCL INFO Channel 10/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068011:2068209 [2] NCCL INFO Channel 20/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068015:2068208 [6] NCCL INFO Channel 13/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2068016:2068206 [7] NCCL INFO Channel 08/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068013:2068203 [4] NCCL INFO Channel 09/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068009:2068202 [0] NCCL INFO Channel 15/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068014:2068205 [5] NCCL INFO Channel 06/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068010:2068204 [1] NCCL INFO Channel 14/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068012:2068207 [3] NCCL INFO Channel 11/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068011:2068209 [2] NCCL INFO Channel 21/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068015:2068208 [6] NCCL INFO Channel 14/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2068016:2068206 [7] NCCL INFO Channel 09/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068013:2068203 [4] NCCL INFO Channel 10/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068009:2068202 [0] NCCL INFO Channel 16/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068014:2068205 [5] NCCL INFO Channel 07/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068010:2068204 [1] NCCL INFO Channel 15/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068012:2068207 [3] NCCL INFO Channel 12/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068015:2068208 [6] NCCL INFO Channel 15/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2068011:2068209 [2] NCCL INFO Channel 22/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068016:2068206 [7] NCCL INFO Channel 10/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068013:2068203 [4] NCCL INFO Channel 11/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068014:2068205 [5] NCCL INFO Channel 08/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068009:2068202 [0] NCCL INFO Channel 17/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068010:2068204 [1] NCCL INFO Channel 16/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068012:2068207 [3] NCCL INFO Channel 13/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068015:2068208 [6] NCCL INFO Channel 16/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2068011:2068209 [2] NCCL INFO Channel 23/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068016:2068206 [7] NCCL INFO Channel 11/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068013:2068203 [4] NCCL INFO Channel 12/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068014:2068205 [5] NCCL INFO Channel 09/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068009:2068202 [0] NCCL INFO Channel 18/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068010:2068204 [1] NCCL INFO Channel 17/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068012:2068207 [3] NCCL INFO Channel 14/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068015:2068208 [6] NCCL INFO Channel 17/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2068016:2068206 [7] NCCL INFO Channel 12/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068014:2068205 [5] NCCL INFO Channel 10/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068013:2068203 [4] NCCL INFO Channel 13/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068009:2068202 [0] NCCL INFO Channel 19/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068010:2068204 [1] NCCL INFO Channel 18/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068012:2068207 [3] NCCL INFO Channel 15/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068015:2068208 [6] NCCL INFO Channel 18/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2068016:2068206 [7] NCCL INFO Channel 13/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068009:2068202 [0] NCCL INFO Channel 20/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068014:2068205 [5] NCCL INFO Channel 11/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068013:2068203 [4] NCCL INFO Channel 14/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068012:2068207 [3] NCCL INFO Channel 16/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068010:2068204 [1] NCCL INFO Channel 19/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068015:2068208 [6] NCCL INFO Channel 19/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2068016:2068206 [7] NCCL INFO Channel 14/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068014:2068205 [5] NCCL INFO Channel 12/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068009:2068202 [0] NCCL INFO Channel 21/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068013:2068203 [4] NCCL INFO Channel 15/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068012:2068207 [3] NCCL INFO Channel 17/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068010:2068204 [1] NCCL INFO Channel 20/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068015:2068208 [6] NCCL INFO Channel 20/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2068016:2068206 [7] NCCL INFO Channel 15/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068014:2068205 [5] NCCL INFO Channel 13/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068013:2068203 [4] NCCL INFO Channel 16/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068009:2068202 [0] NCCL INFO Channel 22/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068012:2068207 [3] NCCL INFO Channel 18/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068010:2068204 [1] NCCL INFO Channel 21/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068015:2068208 [6] NCCL INFO Channel 21/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2068016:2068206 [7] NCCL INFO Channel 16/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068014:2068205 [5] NCCL INFO Channel 14/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068013:2068203 [4] NCCL INFO Channel 17/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068009:2068202 [0] NCCL INFO Channel 23/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068012:2068207 [3] NCCL INFO Channel 19/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068010:2068204 [1] NCCL INFO Channel 22/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068015:2068208 [6] NCCL INFO Channel 22/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2068016:2068206 [7] NCCL INFO Channel 17/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068014:2068205 [5] NCCL INFO Channel 15/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068012:2068207 [3] NCCL INFO Channel 20/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068013:2068203 [4] NCCL INFO Channel 18/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068010:2068204 [1] NCCL INFO Channel 23/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068015:2068208 [6] NCCL INFO Channel 23/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2068016:2068206 [7] NCCL INFO Channel 18/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068014:2068205 [5] NCCL INFO Channel 16/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068012:2068207 [3] NCCL INFO Channel 21/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068013:2068203 [4] NCCL INFO Channel 19/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068016:2068206 [7] NCCL INFO Channel 19/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068014:2068205 [5] NCCL INFO Channel 17/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068012:2068207 [3] NCCL INFO Channel 22/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068013:2068203 [4] NCCL INFO Channel 20/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068016:2068206 [7] NCCL INFO Channel 20/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068014:2068205 [5] NCCL INFO Channel 18/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068012:2068207 [3] NCCL INFO Channel 23/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068013:2068203 [4] NCCL INFO Channel 21/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068016:2068206 [7] NCCL INFO Channel 21/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068014:2068205 [5] NCCL INFO Channel 19/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068013:2068203 [4] NCCL INFO Channel 22/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068016:2068206 [7] NCCL INFO Channel 22/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068014:2068205 [5] NCCL INFO Channel 20/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068013:2068203 [4] NCCL INFO Channel 23/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068016:2068206 [7] NCCL INFO Channel 23/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068014:2068205 [5] NCCL INFO Channel 21/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068014:2068205 [5] NCCL INFO Channel 22/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068014:2068205 [5] NCCL INFO Channel 23/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068010:2068204 [1] NCCL INFO Connected all rings
+ip-26-0-160-103:2068012:2068207 [3] NCCL INFO Connected all rings
+ip-26-0-160-103:2068009:2068202 [0] NCCL INFO Connected all rings
+ip-26-0-160-103:2068013:2068203 [4] NCCL INFO Connected all rings
+ip-26-0-160-103:2068014:2068205 [5] NCCL INFO Connected all rings
+ip-26-0-160-103:2068015:2068208 [6] NCCL INFO Connected all rings
+ip-26-0-160-103:2068011:2068209 [2] NCCL INFO Connected all rings
+ip-26-0-160-103:2068016:2068206 [7] NCCL INFO Connected all rings
+ip-26-0-160-103:2068016:2068206 [7] NCCL INFO Channel 00/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068016:2068206 [7] NCCL INFO Channel 01/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068016:2068206 [7] NCCL INFO Channel 02/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068016:2068206 [7] NCCL INFO Channel 03/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068016:2068206 [7] NCCL INFO Channel 04/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068016:2068206 [7] NCCL INFO Channel 05/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068016:2068206 [7] NCCL INFO Channel 06/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068016:2068206 [7] NCCL INFO Channel 07/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068016:2068206 [7] NCCL INFO Channel 08/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068016:2068206 [7] NCCL INFO Channel 09/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068016:2068206 [7] NCCL INFO Channel 10/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068016:2068206 [7] NCCL INFO Channel 11/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068016:2068206 [7] NCCL INFO Channel 12/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068010:2068204 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068016:2068206 [7] NCCL INFO Channel 13/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068010:2068204 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068016:2068206 [7] NCCL INFO Channel 14/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068010:2068204 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068016:2068206 [7] NCCL INFO Channel 15/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068010:2068204 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068016:2068206 [7] NCCL INFO Channel 16/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068010:2068204 [1] NCCL INFO Channel 04/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068016:2068206 [7] NCCL INFO Channel 17/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068010:2068204 [1] NCCL INFO Channel 05/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068016:2068206 [7] NCCL INFO Channel 18/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068010:2068204 [1] NCCL INFO Channel 06/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068012:2068207 [3] NCCL INFO Channel 00/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068016:2068206 [7] NCCL INFO Channel 19/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068010:2068204 [1] NCCL INFO Channel 07/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068012:2068207 [3] NCCL INFO Channel 01/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068016:2068206 [7] NCCL INFO Channel 20/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068013:2068203 [4] NCCL INFO Channel 00/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068010:2068204 [1] NCCL INFO Channel 08/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068016:2068206 [7] NCCL INFO Channel 21/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068012:2068207 [3] NCCL INFO Channel 02/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068013:2068203 [4] NCCL INFO Channel 01/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068010:2068204 [1] NCCL INFO Channel 09/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068014:2068205 [5] NCCL INFO Channel 00/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068012:2068207 [3] NCCL INFO Channel 03/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068016:2068206 [7] NCCL INFO Channel 22/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068013:2068203 [4] NCCL INFO Channel 02/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068010:2068204 [1] NCCL INFO Channel 10/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068015:2068208 [6] NCCL INFO Channel 00/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068014:2068205 [5] NCCL INFO Channel 01/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068011:2068209 [2] NCCL INFO Channel 00/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068012:2068207 [3] NCCL INFO Channel 04/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068013:2068203 [4] NCCL INFO Channel 03/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068016:2068206 [7] NCCL INFO Channel 23/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068010:2068204 [1] NCCL INFO Channel 11/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068015:2068208 [6] NCCL INFO Channel 01/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068011:2068209 [2] NCCL INFO Channel 01/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068013:2068203 [4] NCCL INFO Channel 04/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068010:2068204 [1] NCCL INFO Channel 12/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068014:2068205 [5] NCCL INFO Channel 02/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068011:2068209 [2] NCCL INFO Channel 02/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068013:2068203 [4] NCCL INFO Channel 05/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068012:2068207 [3] NCCL INFO Channel 05/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068013:2068203 [4] NCCL INFO Channel 06/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068015:2068208 [6] NCCL INFO Channel 02/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068014:2068205 [5] NCCL INFO Channel 03/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068011:2068209 [2] NCCL INFO Channel 03/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068013:2068203 [4] NCCL INFO Channel 07/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068015:2068208 [6] NCCL INFO Channel 03/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068014:2068205 [5] NCCL INFO Channel 04/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068011:2068209 [2] NCCL INFO Channel 04/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068013:2068203 [4] NCCL INFO Channel 08/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068015:2068208 [6] NCCL INFO Channel 04/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068014:2068205 [5] NCCL INFO Channel 05/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068011:2068209 [2] NCCL INFO Channel 05/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068013:2068203 [4] NCCL INFO Channel 09/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068015:2068208 [6] NCCL INFO Channel 05/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068014:2068205 [5] NCCL INFO Channel 06/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068011:2068209 [2] NCCL INFO Channel 06/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068010:2068204 [1] NCCL INFO Channel 13/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068013:2068203 [4] NCCL INFO Channel 10/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068015:2068208 [6] NCCL INFO Channel 06/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068014:2068205 [5] NCCL INFO Channel 07/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068011:2068209 [2] NCCL INFO Channel 07/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068010:2068204 [1] NCCL INFO Channel 14/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068013:2068203 [4] NCCL INFO Channel 11/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068015:2068208 [6] NCCL INFO Channel 07/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068014:2068205 [5] NCCL INFO Channel 08/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068011:2068209 [2] NCCL INFO Channel 08/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068010:2068204 [1] NCCL INFO Channel 15/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068013:2068203 [4] NCCL INFO Channel 12/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068015:2068208 [6] NCCL INFO Channel 08/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068014:2068205 [5] NCCL INFO Channel 09/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068011:2068209 [2] NCCL INFO Channel 09/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068010:2068204 [1] NCCL INFO Channel 16/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068013:2068203 [4] NCCL INFO Channel 13/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068015:2068208 [6] NCCL INFO Channel 09/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068012:2068207 [3] NCCL INFO Channel 06/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068014:2068205 [5] NCCL INFO Channel 10/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068011:2068209 [2] NCCL INFO Channel 10/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068010:2068204 [1] NCCL INFO Channel 17/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068013:2068203 [4] NCCL INFO Channel 14/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068015:2068208 [6] NCCL INFO Channel 10/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068012:2068207 [3] NCCL INFO Channel 07/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068014:2068205 [5] NCCL INFO Channel 11/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068013:2068203 [4] NCCL INFO Channel 15/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068015:2068208 [6] NCCL INFO Channel 11/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068012:2068207 [3] NCCL INFO Channel 08/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068014:2068205 [5] NCCL INFO Channel 12/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068013:2068203 [4] NCCL INFO Channel 16/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068015:2068208 [6] NCCL INFO Channel 12/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068012:2068207 [3] NCCL INFO Channel 09/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068014:2068205 [5] NCCL INFO Channel 13/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068013:2068203 [4] NCCL INFO Channel 17/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068015:2068208 [6] NCCL INFO Channel 13/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068012:2068207 [3] NCCL INFO Channel 10/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068013:2068203 [4] NCCL INFO Channel 18/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068014:2068205 [5] NCCL INFO Channel 14/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068015:2068208 [6] NCCL INFO Channel 14/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068012:2068207 [3] NCCL INFO Channel 11/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068013:2068203 [4] NCCL INFO Channel 19/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068014:2068205 [5] NCCL INFO Channel 15/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068015:2068208 [6] NCCL INFO Channel 15/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068012:2068207 [3] NCCL INFO Channel 12/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068013:2068203 [4] NCCL INFO Channel 20/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068014:2068205 [5] NCCL INFO Channel 16/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068015:2068208 [6] NCCL INFO Channel 16/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068012:2068207 [3] NCCL INFO Channel 13/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068014:2068205 [5] NCCL INFO Channel 17/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068011:2068209 [2] NCCL INFO Channel 11/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068010:2068204 [1] NCCL INFO Channel 18/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068014:2068205 [5] NCCL INFO Channel 18/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068013:2068203 [4] NCCL INFO Channel 21/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068011:2068209 [2] NCCL INFO Channel 12/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068012:2068207 [3] NCCL INFO Channel 14/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068010:2068204 [1] NCCL INFO Channel 19/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068014:2068205 [5] NCCL INFO Channel 19/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068013:2068203 [4] NCCL INFO Channel 22/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068015:2068208 [6] NCCL INFO Channel 17/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068011:2068209 [2] NCCL INFO Channel 13/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068012:2068207 [3] NCCL INFO Channel 15/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068010:2068204 [1] NCCL INFO Channel 20/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068014:2068205 [5] NCCL INFO Channel 20/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068013:2068203 [4] NCCL INFO Channel 23/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068015:2068208 [6] NCCL INFO Channel 18/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068011:2068209 [2] NCCL INFO Channel 14/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068012:2068207 [3] NCCL INFO Channel 16/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068010:2068204 [1] NCCL INFO Channel 21/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068014:2068205 [5] NCCL INFO Channel 21/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068015:2068208 [6] NCCL INFO Channel 19/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068011:2068209 [2] NCCL INFO Channel 15/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068012:2068207 [3] NCCL INFO Channel 17/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068010:2068204 [1] NCCL INFO Channel 22/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068014:2068205 [5] NCCL INFO Channel 22/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068015:2068208 [6] NCCL INFO Channel 20/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068012:2068207 [3] NCCL INFO Channel 18/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068011:2068209 [2] NCCL INFO Channel 16/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068010:2068204 [1] NCCL INFO Channel 23/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068012:2068207 [3] NCCL INFO Channel 19/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068014:2068205 [5] NCCL INFO Channel 23/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068015:2068208 [6] NCCL INFO Channel 21/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068012:2068207 [3] NCCL INFO Channel 20/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068015:2068208 [6] NCCL INFO Channel 22/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068012:2068207 [3] NCCL INFO Channel 21/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068011:2068209 [2] NCCL INFO Channel 17/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068015:2068208 [6] NCCL INFO Channel 23/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068012:2068207 [3] NCCL INFO Channel 22/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068012:2068207 [3] NCCL INFO Channel 23/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068011:2068209 [2] NCCL INFO Channel 18/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068011:2068209 [2] NCCL INFO Channel 19/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068011:2068209 [2] NCCL INFO Channel 20/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068011:2068209 [2] NCCL INFO Channel 21/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068011:2068209 [2] NCCL INFO Channel 22/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068011:2068209 [2] NCCL INFO Channel 23/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068009:2068202 [0] NCCL INFO Connected all trees
+ip-26-0-160-103:2068009:2068202 [0] NCCL INFO NVLS comm 0x9897c70 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-103:2068016:2068206 [7] NCCL INFO Connected all trees
+ip-26-0-160-103:2068016:2068206 [7] NCCL INFO NVLS comm 0x96292f0 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-103:2068015:2068208 [6] NCCL INFO Connected all trees
+ip-26-0-160-103:2068015:2068208 [6] NCCL INFO NVLS comm 0x92dd050 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-103:2068014:2068205 [5] NCCL INFO Connected all trees
+ip-26-0-160-103:2068014:2068205 [5] NCCL INFO NVLS comm 0x95df8e0 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-103:2068013:2068203 [4] NCCL INFO Connected all trees
+ip-26-0-160-103:2068013:2068203 [4] NCCL INFO NVLS comm 0x9cfe500 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-103:2068011:2068209 [2] NCCL INFO Connected all trees
+ip-26-0-160-103:2068011:2068209 [2] NCCL INFO NVLS comm 0x9ff19a0 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-103:2068010:2068204 [1] NCCL INFO Connected all trees
+ip-26-0-160-103:2068010:2068204 [1] NCCL INFO NVLS comm 0xa214810 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-103:2068012:2068207 [3] NCCL INFO Connected all trees
+ip-26-0-160-103:2068012:2068207 [3] NCCL INFO NVLS comm 0xa312d40 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-103:2068012:2068207 [3] NCCL INFO Connected NVLS tree
+ip-26-0-160-103:2068012:2068207 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-103:2068012:2068207 [3] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-160-103:2068012:2068207 [3] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-160-103:2068011:2068209 [2] NCCL INFO Connected NVLS tree
+ip-26-0-160-103:2068011:2068209 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-103:2068011:2068209 [2] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-160-103:2068011:2068209 [2] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-160-103:2068013:2068203 [4] NCCL INFO Connected NVLS tree
+ip-26-0-160-103:2068013:2068203 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-103:2068013:2068203 [4] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-160-103:2068013:2068203 [4] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-160-103:2068014:2068205 [5] NCCL INFO Connected NVLS tree
+ip-26-0-160-103:2068014:2068205 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-103:2068014:2068205 [5] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-160-103:2068014:2068205 [5] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-160-103:2068015:2068208 [6] NCCL INFO Connected NVLS tree
+ip-26-0-160-103:2068015:2068208 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-103:2068015:2068208 [6] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-160-103:2068015:2068208 [6] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-160-103:2068010:2068204 [1] NCCL INFO Connected NVLS tree
+ip-26-0-160-103:2068010:2068204 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-103:2068010:2068204 [1] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-160-103:2068010:2068204 [1] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-160-103:2068009:2068202 [0] NCCL INFO Connected NVLS tree
+ip-26-0-160-103:2068009:2068202 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-103:2068009:2068202 [0] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-160-103:2068009:2068202 [0] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-160-103:2068016:2068206 [7] NCCL INFO Connected NVLS tree
+ip-26-0-160-103:2068016:2068206 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-103:2068016:2068206 [7] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-160-103:2068016:2068206 [7] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-160-103:2068009:2068202 [0] NCCL INFO comm 0x9897c70 rank 0 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0xcc4f0ed1a7bd4c20 - Init COMPLETE
+ip-26-0-160-103:2068013:2068203 [4] NCCL INFO comm 0x9cfe500 rank 4 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0xcc4f0ed1a7bd4c20 - Init COMPLETE
+ip-26-0-160-103:2068015:2068208 [6] NCCL INFO comm 0x92dd050 rank 6 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0xcc4f0ed1a7bd4c20 - Init COMPLETE
+ip-26-0-160-103:2068011:2068209 [2] NCCL INFO comm 0x9ff19a0 rank 2 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0xcc4f0ed1a7bd4c20 - Init COMPLETE
+ip-26-0-160-103:2068014:2068205 [5] NCCL INFO comm 0x95df8e0 rank 5 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0xcc4f0ed1a7bd4c20 - Init COMPLETE
+ip-26-0-160-103:2068012:2068207 [3] NCCL INFO comm 0xa312d40 rank 3 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0xcc4f0ed1a7bd4c20 - Init COMPLETE
+ip-26-0-160-103:2068010:2068204 [1] NCCL INFO comm 0xa214810 rank 1 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0xcc4f0ed1a7bd4c20 - Init COMPLETE
+ip-26-0-160-103:2068016:2068206 [7] NCCL INFO comm 0x96292f0 rank 7 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0xcc4f0ed1a7bd4c20 - Init COMPLETE
+ip-26-0-160-103:2068013:2068242 [4] NCCL INFO Using network AWS Libfabric
+ip-26-0-160-103:2068012:2068237 [3] NCCL INFO Using network AWS Libfabric
+ip-26-0-160-103:2068016:2068241 [7] NCCL INFO Using network AWS Libfabric
+ip-26-0-160-103:2068015:2068240 [6] NCCL INFO Using network AWS Libfabric
+ip-26-0-160-103:2068009:2068236 [0] NCCL INFO Using network AWS Libfabric
+ip-26-0-160-103:2068014:2068239 [5] NCCL INFO Using network AWS Libfabric
+ip-26-0-160-103:2068010:2068238 [1] NCCL INFO Using network AWS Libfabric
+ip-26-0-160-103:2068011:2068243 [2] NCCL INFO Using network AWS Libfabric
+ip-26-0-160-103:2068010:2068238 [1] NCCL INFO comm 0xa228dd0 rank 1 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0x653eb5b6ce3cef7f - Init START
+ip-26-0-160-103:2068014:2068239 [5] NCCL INFO comm 0x95f4040 rank 5 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0x653eb5b6ce3cef7f - Init START
+ip-26-0-160-103:2068009:2068236 [0] NCCL INFO comm 0x98ac2d0 rank 0 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0x653eb5b6ce3cef7f - Init START
+ip-26-0-160-103:2068012:2068237 [3] NCCL INFO comm 0xa327490 rank 3 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0x653eb5b6ce3cef7f - Init START
+ip-26-0-160-103:2068011:2068243 [2] NCCL INFO comm 0xa005f70 rank 2 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0x653eb5b6ce3cef7f - Init START
+ip-26-0-160-103:2068016:2068241 [7] NCCL INFO comm 0x963d790 rank 7 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0x653eb5b6ce3cef7f - Init START
+ip-26-0-160-103:2068015:2068240 [6] NCCL INFO comm 0x92f1620 rank 6 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0x653eb5b6ce3cef7f - Init START
+ip-26-0-160-103:2068013:2068242 [4] NCCL INFO comm 0x9d12c60 rank 4 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0x653eb5b6ce3cef7f - Init START
+ip-26-0-160-103:2068011:2068243 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /opt/aws-ofi-nccl/share/aws-ofi-nccl/xml/p5.48xl-topo.xml
+ip-26-0-160-103:2068010:2068238 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /opt/aws-ofi-nccl/share/aws-ofi-nccl/xml/p5.48xl-topo.xml
+ip-26-0-160-103:2068013:2068242 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /opt/aws-ofi-nccl/share/aws-ofi-nccl/xml/p5.48xl-topo.xml
+ip-26-0-160-103:2068012:2068237 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /opt/aws-ofi-nccl/share/aws-ofi-nccl/xml/p5.48xl-topo.xml
+ip-26-0-160-103:2068014:2068239 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /opt/aws-ofi-nccl/share/aws-ofi-nccl/xml/p5.48xl-topo.xml
+ip-26-0-160-103:2068016:2068241 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /opt/aws-ofi-nccl/share/aws-ofi-nccl/xml/p5.48xl-topo.xml
+ip-26-0-160-103:2068015:2068240 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /opt/aws-ofi-nccl/share/aws-ofi-nccl/xml/p5.48xl-topo.xml
+ip-26-0-160-103:2068009:2068236 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /opt/aws-ofi-nccl/share/aws-ofi-nccl/xml/p5.48xl-topo.xml
+ip-26-0-160-103:2068011:2068243 [2] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068011:2068243 [2] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068011:2068243 [2] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068011:2068243 [2] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068011:2068243 [2] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068011:2068243 [2] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068011:2068243 [2] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068011:2068243 [2] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068011:2068243 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-160-103:2068011:2068243 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-160-103:2068012:2068237 [3] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068012:2068237 [3] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068012:2068237 [3] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068012:2068237 [3] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068012:2068237 [3] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068012:2068237 [3] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068012:2068237 [3] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068012:2068237 [3] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068012:2068237 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-160-103:2068012:2068237 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-160-103:2068014:2068239 [5] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068014:2068239 [5] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068014:2068239 [5] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068014:2068239 [5] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068014:2068239 [5] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068014:2068239 [5] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068014:2068239 [5] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068014:2068239 [5] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068014:2068239 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-160-103:2068014:2068239 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-160-103:2068016:2068241 [7] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068016:2068241 [7] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068016:2068241 [7] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068016:2068241 [7] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068016:2068241 [7] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068016:2068241 [7] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068016:2068241 [7] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068016:2068241 [7] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068009:2068236 [0] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068009:2068236 [0] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068009:2068236 [0] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068009:2068236 [0] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068009:2068236 [0] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068009:2068236 [0] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068009:2068236 [0] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068009:2068236 [0] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068009:2068236 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-160-103:2068016:2068241 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-160-103:2068016:2068241 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-160-103:2068009:2068236 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-160-103:2068010:2068238 [1] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068010:2068238 [1] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068010:2068238 [1] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068010:2068238 [1] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068010:2068238 [1] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068010:2068238 [1] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068010:2068238 [1] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068010:2068238 [1] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068010:2068238 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-160-103:2068010:2068238 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-160-103:2068015:2068240 [6] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068015:2068240 [6] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068015:2068240 [6] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068015:2068240 [6] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068015:2068240 [6] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068015:2068240 [6] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068015:2068240 [6] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068015:2068240 [6] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068015:2068240 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-160-103:2068015:2068240 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-160-103:2068013:2068242 [4] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068013:2068242 [4] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068013:2068242 [4] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068013:2068242 [4] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068013:2068242 [4] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068013:2068242 [4] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068013:2068242 [4] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068013:2068242 [4] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068013:2068242 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-160-103:2068013:2068242 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-160-103:2068013:2068242 [4] NCCL INFO Trees [0] 5/-1/-1->4->3 [1] 5/-1/-1->4->3 [2] 5/-1/-1->4->3 [3] 5/-1/-1->4->3 [4] 5/-1/-1->4->3 [5] 5/-1/-1->4->3 [6] 5/-1/-1->4->3 [7] 5/-1/-1->4->3 [8] 5/-1/-1->4->3 [9] 5/-1/-1->4->3 [10] 5/-1/-1->4->3 [11] 5/-1/-1->4->3 [12] 5/-1/-1->4->3 [13] 5/-1/-1->4->3 [14] 5/-1/-1->4->3 [15] 5/-1/-1->4->3 [16] 5/-1/-1->4->3 [17] 5/-1/-1->4->3 [18] 5/-1/-1->4->3 [19] 5/-1/-1->4->3 [20] 5/-1/-1->4->3 [21] 5/-1/-1->4->3 [22] 5/-1/-1->4->3 [23] 5/-1/-1->4->3
+ip-26-0-160-103:2068013:2068242 [4] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-160-103:2068014:2068239 [5] NCCL INFO Trees [0] 6/-1/-1->5->4 [1] 6/-1/-1->5->4 [2] 6/-1/-1->5->4 [3] 6/-1/-1->5->4 [4] 6/-1/-1->5->4 [5] 6/-1/-1->5->4 [6] 6/-1/-1->5->4 [7] 6/-1/-1->5->4 [8] 6/-1/-1->5->4 [9] 6/-1/-1->5->4 [10] 6/-1/-1->5->4 [11] 6/-1/-1->5->4 [12] 6/-1/-1->5->4 [13] 6/-1/-1->5->4 [14] 6/-1/-1->5->4 [15] 6/-1/-1->5->4 [16] 6/-1/-1->5->4 [17] 6/-1/-1->5->4 [18] 6/-1/-1->5->4 [19] 6/-1/-1->5->4 [20] 6/-1/-1->5->4 [21] 6/-1/-1->5->4 [22] 6/-1/-1->5->4 [23] 6/-1/-1->5->4
+ip-26-0-160-103:2068014:2068239 [5] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-160-103:2068012:2068237 [3] NCCL INFO Trees [0] 4/-1/-1->3->2 [1] 4/-1/-1->3->2 [2] 4/-1/-1->3->2 [3] 4/-1/-1->3->2 [4] 4/-1/-1->3->2 [5] 4/-1/-1->3->2 [6] 4/-1/-1->3->2 [7] 4/-1/-1->3->2 [8] 4/-1/-1->3->2 [9] 4/-1/-1->3->2 [10] 4/-1/-1->3->2 [11] 4/-1/-1->3->2 [12] 4/-1/-1->3->2 [13] 4/-1/-1->3->2 [14] 4/-1/-1->3->2 [15] 4/-1/-1->3->2 [16] 4/-1/-1->3->2 [17] 4/-1/-1->3->2 [18] 4/-1/-1->3->2 [19] 4/-1/-1->3->2 [20] 4/-1/-1->3->2 [21] 4/-1/-1->3->2 [22] 4/-1/-1->3->2 [23] 4/-1/-1->3->2
+ip-26-0-160-103:2068010:2068238 [1] NCCL INFO Trees [0] 2/-1/-1->1->0 [1] 2/-1/-1->1->0 [2] 2/-1/-1->1->0 [3] 2/-1/-1->1->0 [4] 2/-1/-1->1->0 [5] 2/-1/-1->1->0 [6] 2/-1/-1->1->0 [7] 2/-1/-1->1->0 [8] 2/-1/-1->1->0 [9] 2/-1/-1->1->0 [10] 2/-1/-1->1->0 [11] 2/-1/-1->1->0 [12] 2/-1/-1->1->0 [13] 2/-1/-1->1->0 [14] 2/-1/-1->1->0 [15] 2/-1/-1->1->0 [16] 2/-1/-1->1->0 [17] 2/-1/-1->1->0 [18] 2/-1/-1->1->0 [19] 2/-1/-1->1->0 [20] 2/-1/-1->1->0 [21] 2/-1/-1->1->0 [22] 2/-1/-1->1->0 [23] 2/-1/-1->1->0
+ip-26-0-160-103:2068011:2068243 [2] NCCL INFO Trees [0] 3/-1/-1->2->1 [1] 3/-1/-1->2->1 [2] 3/-1/-1->2->1 [3] 3/-1/-1->2->1 [4] 3/-1/-1->2->1 [5] 3/-1/-1->2->1 [6] 3/-1/-1->2->1 [7] 3/-1/-1->2->1 [8] 3/-1/-1->2->1 [9] 3/-1/-1->2->1 [10] 3/-1/-1->2->1 [11] 3/-1/-1->2->1 [12] 3/-1/-1->2->1 [13] 3/-1/-1->2->1 [14] 3/-1/-1->2->1 [15] 3/-1/-1->2->1 [16] 3/-1/-1->2->1 [17] 3/-1/-1->2->1 [18] 3/-1/-1->2->1 [19] 3/-1/-1->2->1 [20] 3/-1/-1->2->1 [21] 3/-1/-1->2->1 [22] 3/-1/-1->2->1 [23] 3/-1/-1->2->1
+ip-26-0-160-103:2068012:2068237 [3] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-160-103:2068011:2068243 [2] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-160-103:2068009:2068236 [0] NCCL INFO Channel 00/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2068015:2068240 [6] NCCL INFO Trees [0] 7/-1/-1->6->5 [1] 7/-1/-1->6->5 [2] 7/-1/-1->6->5 [3] 7/-1/-1->6->5 [4] 7/-1/-1->6->5 [5] 7/-1/-1->6->5 [6] 7/-1/-1->6->5 [7] 7/-1/-1->6->5 [8] 7/-1/-1->6->5 [9] 7/-1/-1->6->5 [10] 7/-1/-1->6->5 [11] 7/-1/-1->6->5 [12] 7/-1/-1->6->5 [13] 7/-1/-1->6->5 [14] 7/-1/-1->6->5 [15] 7/-1/-1->6->5 [16] 7/-1/-1->6->5 [17] 7/-1/-1->6->5 [18] 7/-1/-1->6->5 [19] 7/-1/-1->6->5 [20] 7/-1/-1->6->5 [21] 7/-1/-1->6->5 [22] 7/-1/-1->6->5 [23] 7/-1/-1->6->5
+ip-26-0-160-103:2068010:2068238 [1] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-160-103:2068009:2068236 [0] NCCL INFO Channel 01/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2068015:2068240 [6] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-160-103:2068016:2068241 [7] NCCL INFO Trees [0] -1/-1/-1->7->6 [1] -1/-1/-1->7->6 [2] -1/-1/-1->7->6 [3] -1/-1/-1->7->6 [4] -1/-1/-1->7->6 [5] -1/-1/-1->7->6 [6] -1/-1/-1->7->6 [7] -1/-1/-1->7->6 [8] -1/-1/-1->7->6 [9] -1/-1/-1->7->6 [10] -1/-1/-1->7->6 [11] -1/-1/-1->7->6 [12] -1/-1/-1->7->6 [13] -1/-1/-1->7->6 [14] -1/-1/-1->7->6 [15] -1/-1/-1->7->6 [16] -1/-1/-1->7->6 [17] -1/-1/-1->7->6 [18] -1/-1/-1->7->6 [19] -1/-1/-1->7->6 [20] -1/-1/-1->7->6 [21] -1/-1/-1->7->6 [22] -1/-1/-1->7->6 [23] -1/-1/-1->7->6
+ip-26-0-160-103:2068009:2068236 [0] NCCL INFO Channel 02/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2068016:2068241 [7] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-160-103:2068009:2068236 [0] NCCL INFO Channel 03/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2068009:2068236 [0] NCCL INFO Channel 04/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2068009:2068236 [0] NCCL INFO Channel 05/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2068009:2068236 [0] NCCL INFO Channel 06/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2068009:2068236 [0] NCCL INFO Channel 07/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2068009:2068236 [0] NCCL INFO Channel 08/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2068009:2068236 [0] NCCL INFO Channel 09/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2068009:2068236 [0] NCCL INFO Channel 10/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2068009:2068236 [0] NCCL INFO Channel 11/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2068009:2068236 [0] NCCL INFO Channel 12/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2068009:2068236 [0] NCCL INFO Channel 13/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2068009:2068236 [0] NCCL INFO Channel 14/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2068009:2068236 [0] NCCL INFO Channel 15/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2068009:2068236 [0] NCCL INFO Channel 16/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2068009:2068236 [0] NCCL INFO Channel 17/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2068009:2068236 [0] NCCL INFO Channel 18/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2068009:2068236 [0] NCCL INFO Channel 19/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2068009:2068236 [0] NCCL INFO Channel 20/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2068009:2068236 [0] NCCL INFO Channel 21/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2068009:2068236 [0] NCCL INFO Channel 22/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2068009:2068236 [0] NCCL INFO Channel 23/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2068009:2068236 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] 1/-1/-1->0->-1 [7] 1/-1/-1->0->-1 [8] 1/-1/-1->0->-1 [9] 1/-1/-1->0->-1 [10] 1/-1/-1->0->-1 [11] 1/-1/-1->0->-1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] 1/-1/-1->0->-1 [19] 1/-1/-1->0->-1 [20] 1/-1/-1->0->-1 [21] 1/-1/-1->0->-1 [22] 1/-1/-1->0->-1 [23] 1/-1/-1->0->-1
+ip-26-0-160-103:2068009:2068236 [0] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-160-103:2068016:2068241 [7] NCCL INFO Channel 00/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068013:2068242 [4] NCCL INFO Channel 00/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068012:2068237 [3] NCCL INFO Channel 00/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068011:2068243 [2] NCCL INFO Channel 00/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068009:2068236 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068015:2068240 [6] NCCL INFO Channel 00/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2068016:2068241 [7] NCCL INFO Channel 01/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068011:2068243 [2] NCCL INFO Channel 01/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068009:2068236 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068016:2068241 [7] NCCL INFO Channel 02/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068015:2068240 [6] NCCL INFO Channel 01/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2068011:2068243 [2] NCCL INFO Channel 02/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068009:2068236 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068015:2068240 [6] NCCL INFO Channel 02/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2068016:2068241 [7] NCCL INFO Channel 03/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068014:2068239 [5] NCCL INFO Channel 00/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068010:2068238 [1] NCCL INFO Channel 00/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068011:2068243 [2] NCCL INFO Channel 03/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068009:2068236 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068012:2068237 [3] NCCL INFO Channel 01/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068014:2068239 [5] NCCL INFO Channel 01/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068010:2068238 [1] NCCL INFO Channel 01/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068011:2068243 [2] NCCL INFO Channel 04/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068009:2068236 [0] NCCL INFO Channel 04/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068012:2068237 [3] NCCL INFO Channel 02/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068011:2068243 [2] NCCL INFO Channel 05/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068014:2068239 [5] NCCL INFO Channel 02/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068010:2068238 [1] NCCL INFO Channel 02/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068013:2068242 [4] NCCL INFO Channel 01/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068014:2068239 [5] NCCL INFO Channel 03/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068015:2068240 [6] NCCL INFO Channel 03/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2068016:2068241 [7] NCCL INFO Channel 04/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068013:2068242 [4] NCCL INFO Channel 02/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068014:2068239 [5] NCCL INFO Channel 04/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068015:2068240 [6] NCCL INFO Channel 04/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2068016:2068241 [7] NCCL INFO Channel 05/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068012:2068237 [3] NCCL INFO Channel 03/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068014:2068239 [5] NCCL INFO Channel 05/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068009:2068236 [0] NCCL INFO Channel 05/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068010:2068238 [1] NCCL INFO Channel 03/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068011:2068243 [2] NCCL INFO Channel 06/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068015:2068240 [6] NCCL INFO Channel 05/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2068016:2068241 [7] NCCL INFO Channel 06/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068011:2068243 [2] NCCL INFO Channel 07/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068012:2068237 [3] NCCL INFO Channel 04/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068015:2068240 [6] NCCL INFO Channel 06/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2068016:2068241 [7] NCCL INFO Channel 07/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068011:2068243 [2] NCCL INFO Channel 08/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068009:2068236 [0] NCCL INFO Channel 06/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068012:2068237 [3] NCCL INFO Channel 05/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068015:2068240 [6] NCCL INFO Channel 07/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2068016:2068241 [7] NCCL INFO Channel 08/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068011:2068243 [2] NCCL INFO Channel 09/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068009:2068236 [0] NCCL INFO Channel 07/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068012:2068237 [3] NCCL INFO Channel 06/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068016:2068241 [7] NCCL INFO Channel 09/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068015:2068240 [6] NCCL INFO Channel 08/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2068011:2068243 [2] NCCL INFO Channel 10/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068009:2068236 [0] NCCL INFO Channel 08/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068012:2068237 [3] NCCL INFO Channel 07/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068016:2068241 [7] NCCL INFO Channel 10/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068015:2068240 [6] NCCL INFO Channel 09/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2068011:2068243 [2] NCCL INFO Channel 11/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068009:2068236 [0] NCCL INFO Channel 09/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068012:2068237 [3] NCCL INFO Channel 08/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068016:2068241 [7] NCCL INFO Channel 11/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068015:2068240 [6] NCCL INFO Channel 10/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2068011:2068243 [2] NCCL INFO Channel 12/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068009:2068236 [0] NCCL INFO Channel 10/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068012:2068237 [3] NCCL INFO Channel 09/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068016:2068241 [7] NCCL INFO Channel 12/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068015:2068240 [6] NCCL INFO Channel 11/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2068011:2068243 [2] NCCL INFO Channel 13/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068009:2068236 [0] NCCL INFO Channel 11/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068013:2068242 [4] NCCL INFO Channel 03/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068012:2068237 [3] NCCL INFO Channel 10/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068016:2068241 [7] NCCL INFO Channel 13/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068014:2068239 [5] NCCL INFO Channel 06/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068015:2068240 [6] NCCL INFO Channel 12/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2068010:2068238 [1] NCCL INFO Channel 04/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068011:2068243 [2] NCCL INFO Channel 14/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068009:2068236 [0] NCCL INFO Channel 12/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068013:2068242 [4] NCCL INFO Channel 04/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068012:2068237 [3] NCCL INFO Channel 11/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068016:2068241 [7] NCCL INFO Channel 14/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068014:2068239 [5] NCCL INFO Channel 07/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068010:2068238 [1] NCCL INFO Channel 05/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068015:2068240 [6] NCCL INFO Channel 13/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2068011:2068243 [2] NCCL INFO Channel 15/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068009:2068236 [0] NCCL INFO Channel 13/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068013:2068242 [4] NCCL INFO Channel 05/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068012:2068237 [3] NCCL INFO Channel 12/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068014:2068239 [5] NCCL INFO Channel 08/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068016:2068241 [7] NCCL INFO Channel 15/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068010:2068238 [1] NCCL INFO Channel 06/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068011:2068243 [2] NCCL INFO Channel 16/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068015:2068240 [6] NCCL INFO Channel 14/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2068009:2068236 [0] NCCL INFO Channel 14/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068013:2068242 [4] NCCL INFO Channel 06/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068012:2068237 [3] NCCL INFO Channel 13/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068014:2068239 [5] NCCL INFO Channel 09/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068010:2068238 [1] NCCL INFO Channel 07/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068011:2068243 [2] NCCL INFO Channel 17/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068009:2068236 [0] NCCL INFO Channel 15/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068016:2068241 [7] NCCL INFO Channel 16/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068015:2068240 [6] NCCL INFO Channel 15/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2068013:2068242 [4] NCCL INFO Channel 07/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068011:2068243 [2] NCCL INFO Channel 18/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068009:2068236 [0] NCCL INFO Channel 16/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068013:2068242 [4] NCCL INFO Channel 08/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068012:2068237 [3] NCCL INFO Channel 14/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068010:2068238 [1] NCCL INFO Channel 08/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068012:2068237 [3] NCCL INFO Channel 15/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068014:2068239 [5] NCCL INFO Channel 10/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068016:2068241 [7] NCCL INFO Channel 17/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068010:2068238 [1] NCCL INFO Channel 09/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068012:2068237 [3] NCCL INFO Channel 16/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068014:2068239 [5] NCCL INFO Channel 11/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068016:2068241 [7] NCCL INFO Channel 18/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068012:2068237 [3] NCCL INFO Channel 17/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068010:2068238 [1] NCCL INFO Channel 10/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068014:2068239 [5] NCCL INFO Channel 12/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068016:2068241 [7] NCCL INFO Channel 19/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068012:2068237 [3] NCCL INFO Channel 18/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068010:2068238 [1] NCCL INFO Channel 11/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068014:2068239 [5] NCCL INFO Channel 13/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068016:2068241 [7] NCCL INFO Channel 20/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068012:2068237 [3] NCCL INFO Channel 19/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068010:2068238 [1] NCCL INFO Channel 12/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068014:2068239 [5] NCCL INFO Channel 14/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068015:2068240 [6] NCCL INFO Channel 16/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2068016:2068241 [7] NCCL INFO Channel 21/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068010:2068238 [1] NCCL INFO Channel 13/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068012:2068237 [3] NCCL INFO Channel 20/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068014:2068239 [5] NCCL INFO Channel 15/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068015:2068240 [6] NCCL INFO Channel 17/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2068016:2068241 [7] NCCL INFO Channel 22/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068012:2068237 [3] NCCL INFO Channel 21/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068010:2068238 [1] NCCL INFO Channel 14/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068014:2068239 [5] NCCL INFO Channel 16/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068015:2068240 [6] NCCL INFO Channel 18/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2068016:2068241 [7] NCCL INFO Channel 23/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068012:2068237 [3] NCCL INFO Channel 22/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068010:2068238 [1] NCCL INFO Channel 15/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068015:2068240 [6] NCCL INFO Channel 19/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2068014:2068239 [5] NCCL INFO Channel 17/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068012:2068237 [3] NCCL INFO Channel 23/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068010:2068238 [1] NCCL INFO Channel 16/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068015:2068240 [6] NCCL INFO Channel 20/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2068014:2068239 [5] NCCL INFO Channel 18/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068011:2068243 [2] NCCL INFO Channel 19/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068009:2068236 [0] NCCL INFO Channel 17/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068010:2068238 [1] NCCL INFO Channel 17/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068013:2068242 [4] NCCL INFO Channel 09/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068015:2068240 [6] NCCL INFO Channel 21/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2068014:2068239 [5] NCCL INFO Channel 19/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068011:2068243 [2] NCCL INFO Channel 20/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068009:2068236 [0] NCCL INFO Channel 18/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068010:2068238 [1] NCCL INFO Channel 18/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068013:2068242 [4] NCCL INFO Channel 10/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068015:2068240 [6] NCCL INFO Channel 22/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2068014:2068239 [5] NCCL INFO Channel 20/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068011:2068243 [2] NCCL INFO Channel 21/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068009:2068236 [0] NCCL INFO Channel 19/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068010:2068238 [1] NCCL INFO Channel 19/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068013:2068242 [4] NCCL INFO Channel 11/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068015:2068240 [6] NCCL INFO Channel 23/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2068014:2068239 [5] NCCL INFO Channel 21/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068009:2068236 [0] NCCL INFO Channel 20/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068011:2068243 [2] NCCL INFO Channel 22/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068010:2068238 [1] NCCL INFO Channel 20/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068013:2068242 [4] NCCL INFO Channel 12/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068014:2068239 [5] NCCL INFO Channel 22/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068009:2068236 [0] NCCL INFO Channel 21/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068010:2068238 [1] NCCL INFO Channel 21/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068014:2068239 [5] NCCL INFO Channel 23/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068009:2068236 [0] NCCL INFO Channel 22/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068010:2068238 [1] NCCL INFO Channel 22/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068009:2068236 [0] NCCL INFO Channel 23/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068013:2068242 [4] NCCL INFO Channel 13/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068010:2068238 [1] NCCL INFO Channel 23/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068013:2068242 [4] NCCL INFO Channel 14/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068011:2068243 [2] NCCL INFO Channel 23/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068013:2068242 [4] NCCL INFO Channel 15/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068013:2068242 [4] NCCL INFO Channel 16/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068013:2068242 [4] NCCL INFO Channel 17/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068013:2068242 [4] NCCL INFO Channel 18/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068013:2068242 [4] NCCL INFO Channel 19/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068013:2068242 [4] NCCL INFO Channel 20/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068013:2068242 [4] NCCL INFO Channel 21/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068013:2068242 [4] NCCL INFO Channel 22/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068013:2068242 [4] NCCL INFO Channel 23/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068011:2068243 [2] NCCL INFO Connected all rings
+ip-26-0-160-103:2068016:2068241 [7] NCCL INFO Connected all rings
+ip-26-0-160-103:2068016:2068241 [7] NCCL INFO Channel 00/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068016:2068241 [7] NCCL INFO Channel 01/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068015:2068240 [6] NCCL INFO Connected all rings
+ip-26-0-160-103:2068009:2068236 [0] NCCL INFO Connected all rings
+ip-26-0-160-103:2068016:2068241 [7] NCCL INFO Channel 02/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068010:2068238 [1] NCCL INFO Connected all rings
+ip-26-0-160-103:2068016:2068241 [7] NCCL INFO Channel 03/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068016:2068241 [7] NCCL INFO Channel 04/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068013:2068242 [4] NCCL INFO Connected all rings
+ip-26-0-160-103:2068016:2068241 [7] NCCL INFO Channel 05/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068016:2068241 [7] NCCL INFO Channel 06/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068016:2068241 [7] NCCL INFO Channel 07/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068012:2068237 [3] NCCL INFO Connected all rings
+ip-26-0-160-103:2068016:2068241 [7] NCCL INFO Channel 08/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068014:2068239 [5] NCCL INFO Connected all rings
+ip-26-0-160-103:2068016:2068241 [7] NCCL INFO Channel 09/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068016:2068241 [7] NCCL INFO Channel 10/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068016:2068241 [7] NCCL INFO Channel 11/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068016:2068241 [7] NCCL INFO Channel 12/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068016:2068241 [7] NCCL INFO Channel 13/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068016:2068241 [7] NCCL INFO Channel 14/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068016:2068241 [7] NCCL INFO Channel 15/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068016:2068241 [7] NCCL INFO Channel 16/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068016:2068241 [7] NCCL INFO Channel 17/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068016:2068241 [7] NCCL INFO Channel 18/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068016:2068241 [7] NCCL INFO Channel 19/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068016:2068241 [7] NCCL INFO Channel 20/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068016:2068241 [7] NCCL INFO Channel 21/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068016:2068241 [7] NCCL INFO Channel 22/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068011:2068243 [2] NCCL INFO Channel 00/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068016:2068241 [7] NCCL INFO Channel 23/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2068011:2068243 [2] NCCL INFO Channel 01/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068015:2068240 [6] NCCL INFO Channel 00/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068011:2068243 [2] NCCL INFO Channel 02/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068015:2068240 [6] NCCL INFO Channel 01/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068011:2068243 [2] NCCL INFO Channel 03/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068015:2068240 [6] NCCL INFO Channel 02/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068011:2068243 [2] NCCL INFO Channel 04/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068015:2068240 [6] NCCL INFO Channel 03/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068011:2068243 [2] NCCL INFO Channel 05/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068011:2068243 [2] NCCL INFO Channel 06/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068015:2068240 [6] NCCL INFO Channel 04/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068011:2068243 [2] NCCL INFO Channel 07/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068015:2068240 [6] NCCL INFO Channel 05/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068011:2068243 [2] NCCL INFO Channel 08/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068015:2068240 [6] NCCL INFO Channel 06/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068011:2068243 [2] NCCL INFO Channel 09/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068012:2068237 [3] NCCL INFO Channel 00/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068010:2068238 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068015:2068240 [6] NCCL INFO Channel 07/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068013:2068242 [4] NCCL INFO Channel 00/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068011:2068243 [2] NCCL INFO Channel 10/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068012:2068237 [3] NCCL INFO Channel 01/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068010:2068238 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068015:2068240 [6] NCCL INFO Channel 08/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068013:2068242 [4] NCCL INFO Channel 01/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068011:2068243 [2] NCCL INFO Channel 11/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068012:2068237 [3] NCCL INFO Channel 02/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068010:2068238 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068015:2068240 [6] NCCL INFO Channel 09/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068013:2068242 [4] NCCL INFO Channel 02/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068011:2068243 [2] NCCL INFO Channel 12/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068012:2068237 [3] NCCL INFO Channel 03/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068010:2068238 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068015:2068240 [6] NCCL INFO Channel 10/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068013:2068242 [4] NCCL INFO Channel 03/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068011:2068243 [2] NCCL INFO Channel 13/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068012:2068237 [3] NCCL INFO Channel 04/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068010:2068238 [1] NCCL INFO Channel 04/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068015:2068240 [6] NCCL INFO Channel 11/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068013:2068242 [4] NCCL INFO Channel 04/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068011:2068243 [2] NCCL INFO Channel 14/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068012:2068237 [3] NCCL INFO Channel 05/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068010:2068238 [1] NCCL INFO Channel 05/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068015:2068240 [6] NCCL INFO Channel 12/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068013:2068242 [4] NCCL INFO Channel 05/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068011:2068243 [2] NCCL INFO Channel 15/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068012:2068237 [3] NCCL INFO Channel 06/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068010:2068238 [1] NCCL INFO Channel 06/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068015:2068240 [6] NCCL INFO Channel 13/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068014:2068239 [5] NCCL INFO Channel 00/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068013:2068242 [4] NCCL INFO Channel 06/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068011:2068243 [2] NCCL INFO Channel 16/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068012:2068237 [3] NCCL INFO Channel 07/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068010:2068238 [1] NCCL INFO Channel 07/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068015:2068240 [6] NCCL INFO Channel 14/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068014:2068239 [5] NCCL INFO Channel 01/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068013:2068242 [4] NCCL INFO Channel 07/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068011:2068243 [2] NCCL INFO Channel 17/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068012:2068237 [3] NCCL INFO Channel 08/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068010:2068238 [1] NCCL INFO Channel 08/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068015:2068240 [6] NCCL INFO Channel 15/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068013:2068242 [4] NCCL INFO Channel 08/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068012:2068237 [3] NCCL INFO Channel 09/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068011:2068243 [2] NCCL INFO Channel 18/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068010:2068238 [1] NCCL INFO Channel 09/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068015:2068240 [6] NCCL INFO Channel 16/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068013:2068242 [4] NCCL INFO Channel 09/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068014:2068239 [5] NCCL INFO Channel 02/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068012:2068237 [3] NCCL INFO Channel 10/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068011:2068243 [2] NCCL INFO Channel 19/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068010:2068238 [1] NCCL INFO Channel 10/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068013:2068242 [4] NCCL INFO Channel 10/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068014:2068239 [5] NCCL INFO Channel 03/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068011:2068243 [2] NCCL INFO Channel 20/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068012:2068237 [3] NCCL INFO Channel 11/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068010:2068238 [1] NCCL INFO Channel 11/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068013:2068242 [4] NCCL INFO Channel 11/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068014:2068239 [5] NCCL INFO Channel 04/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068011:2068243 [2] NCCL INFO Channel 21/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068012:2068237 [3] NCCL INFO Channel 12/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068010:2068238 [1] NCCL INFO Channel 12/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068013:2068242 [4] NCCL INFO Channel 12/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068015:2068240 [6] NCCL INFO Channel 17/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068012:2068237 [3] NCCL INFO Channel 13/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068013:2068242 [4] NCCL INFO Channel 13/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068015:2068240 [6] NCCL INFO Channel 18/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068014:2068239 [5] NCCL INFO Channel 05/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068010:2068238 [1] NCCL INFO Channel 13/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068012:2068237 [3] NCCL INFO Channel 14/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068013:2068242 [4] NCCL INFO Channel 14/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068014:2068239 [5] NCCL INFO Channel 06/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068015:2068240 [6] NCCL INFO Channel 19/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068010:2068238 [1] NCCL INFO Channel 14/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068012:2068237 [3] NCCL INFO Channel 15/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068013:2068242 [4] NCCL INFO Channel 15/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068015:2068240 [6] NCCL INFO Channel 20/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068014:2068239 [5] NCCL INFO Channel 07/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068011:2068243 [2] NCCL INFO Channel 22/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068010:2068238 [1] NCCL INFO Channel 15/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068013:2068242 [4] NCCL INFO Channel 16/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068011:2068243 [2] NCCL INFO Channel 23/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068015:2068240 [6] NCCL INFO Channel 21/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068010:2068238 [1] NCCL INFO Channel 16/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068013:2068242 [4] NCCL INFO Channel 17/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068012:2068237 [3] NCCL INFO Channel 16/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068014:2068239 [5] NCCL INFO Channel 08/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068010:2068238 [1] NCCL INFO Channel 17/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068015:2068240 [6] NCCL INFO Channel 22/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068013:2068242 [4] NCCL INFO Channel 18/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068012:2068237 [3] NCCL INFO Channel 17/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068014:2068239 [5] NCCL INFO Channel 09/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068015:2068240 [6] NCCL INFO Channel 23/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2068010:2068238 [1] NCCL INFO Channel 18/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068013:2068242 [4] NCCL INFO Channel 19/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068012:2068237 [3] NCCL INFO Channel 18/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068014:2068239 [5] NCCL INFO Channel 10/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068013:2068242 [4] NCCL INFO Channel 20/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068010:2068238 [1] NCCL INFO Channel 19/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068012:2068237 [3] NCCL INFO Channel 19/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068014:2068239 [5] NCCL INFO Channel 11/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068013:2068242 [4] NCCL INFO Channel 21/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068010:2068238 [1] NCCL INFO Channel 20/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068012:2068237 [3] NCCL INFO Channel 20/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068014:2068239 [5] NCCL INFO Channel 12/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068013:2068242 [4] NCCL INFO Channel 22/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068010:2068238 [1] NCCL INFO Channel 21/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068012:2068237 [3] NCCL INFO Channel 21/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068014:2068239 [5] NCCL INFO Channel 13/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068010:2068238 [1] NCCL INFO Channel 22/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068012:2068237 [3] NCCL INFO Channel 22/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068013:2068242 [4] NCCL INFO Channel 23/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2068012:2068237 [3] NCCL INFO Channel 23/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2068010:2068238 [1] NCCL INFO Channel 23/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068014:2068239 [5] NCCL INFO Channel 14/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068014:2068239 [5] NCCL INFO Channel 15/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068014:2068239 [5] NCCL INFO Channel 16/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068014:2068239 [5] NCCL INFO Channel 17/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068014:2068239 [5] NCCL INFO Channel 18/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068014:2068239 [5] NCCL INFO Channel 19/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068016:2068241 [7] NCCL INFO Connected all trees
+ip-26-0-160-103:2068016:2068241 [7] NCCL INFO NVLS comm 0x963d790 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-103:2068014:2068239 [5] NCCL INFO Channel 20/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068014:2068239 [5] NCCL INFO Channel 21/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068014:2068239 [5] NCCL INFO Channel 22/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068014:2068239 [5] NCCL INFO Channel 23/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2068009:2068236 [0] NCCL INFO Connected all trees
+ip-26-0-160-103:2068009:2068236 [0] NCCL INFO NVLS comm 0x98ac2d0 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-103:2068012:2068237 [3] NCCL INFO Connected all trees
+ip-26-0-160-103:2068012:2068237 [3] NCCL INFO NVLS comm 0xa327490 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-103:2068010:2068238 [1] NCCL INFO Connected all trees
+ip-26-0-160-103:2068010:2068238 [1] NCCL INFO NVLS comm 0xa228dd0 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-103:2068011:2068243 [2] NCCL INFO Connected all trees
+ip-26-0-160-103:2068011:2068243 [2] NCCL INFO NVLS comm 0xa005f70 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-103:2068014:2068239 [5] NCCL INFO Connected all trees
+ip-26-0-160-103:2068014:2068239 [5] NCCL INFO NVLS comm 0x95f4040 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-103:2068013:2068242 [4] NCCL INFO Connected all trees
+ip-26-0-160-103:2068013:2068242 [4] NCCL INFO NVLS comm 0x9d12c60 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-103:2068015:2068240 [6] NCCL INFO Connected all trees
+ip-26-0-160-103:2068015:2068240 [6] NCCL INFO NVLS comm 0x92f1620 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-103:2068012:2068237 [3] NCCL INFO Connected NVLS tree
+ip-26-0-160-103:2068012:2068237 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-103:2068012:2068237 [3] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-160-103:2068012:2068237 [3] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-160-103:2068014:2068239 [5] NCCL INFO Connected NVLS tree
+ip-26-0-160-103:2068014:2068239 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-103:2068014:2068239 [5] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-160-103:2068014:2068239 [5] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-160-103:2068013:2068242 [4] NCCL INFO Connected NVLS tree
+ip-26-0-160-103:2068013:2068242 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-103:2068013:2068242 [4] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-160-103:2068013:2068242 [4] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-160-103:2068011:2068243 [2] NCCL INFO Connected NVLS tree
+ip-26-0-160-103:2068011:2068243 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-103:2068011:2068243 [2] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-160-103:2068011:2068243 [2] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-160-103:2068015:2068240 [6] NCCL INFO Connected NVLS tree
+ip-26-0-160-103:2068015:2068240 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-103:2068015:2068240 [6] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-160-103:2068015:2068240 [6] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-160-103:2068016:2068241 [7] NCCL INFO Connected NVLS tree
+ip-26-0-160-103:2068016:2068241 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-103:2068016:2068241 [7] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-160-103:2068016:2068241 [7] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-160-103:2068009:2068236 [0] NCCL INFO Connected NVLS tree
+ip-26-0-160-103:2068009:2068236 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-103:2068010:2068238 [1] NCCL INFO Connected NVLS tree
+ip-26-0-160-103:2068009:2068236 [0] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-160-103:2068009:2068236 [0] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-160-103:2068010:2068238 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-103:2068010:2068238 [1] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-160-103:2068010:2068238 [1] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-160-103:2068011:2068243 [2] NCCL INFO comm 0xa005f70 rank 2 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0x653eb5b6ce3cef7f - Init COMPLETE
+ip-26-0-160-103:2068015:2068240 [6] NCCL INFO comm 0x92f1620 rank 6 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0x653eb5b6ce3cef7f - Init COMPLETE
+ip-26-0-160-103:2068009:2068236 [0] NCCL INFO comm 0x98ac2d0 rank 0 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0x653eb5b6ce3cef7f - Init COMPLETE
+ip-26-0-160-103:2068013:2068242 [4] NCCL INFO comm 0x9d12c60 rank 4 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0x653eb5b6ce3cef7f - Init COMPLETE
+ip-26-0-160-103:2068014:2068239 [5] NCCL INFO comm 0x95f4040 rank 5 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0x653eb5b6ce3cef7f - Init COMPLETE
+ip-26-0-160-103:2068010:2068238 [1] NCCL INFO comm 0xa228dd0 rank 1 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0x653eb5b6ce3cef7f - Init COMPLETE
+ip-26-0-160-103:2068012:2068237 [3] NCCL INFO comm 0xa327490 rank 3 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0x653eb5b6ce3cef7f - Init COMPLETE
+ip-26-0-160-103:2068016:2068241 [7] NCCL INFO comm 0x963d790 rank 7 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0x653eb5b6ce3cef7f - Init COMPLETE
+12/12/2024 22:02:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]: Config:
+12/12/2024 22:02:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]: Config(general=GeneralArgs(project='debug',
+12/12/2024 22:02:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                            run='1.14G_dp4_tp2_pp1_acc256_mbs2_seq2048_zero1_tpmodeRED_vocab32k',
+12/12/2024 22:02:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                            seed=42,
+12/12/2024 22:02:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                            step=None,
+12/12/2024 22:02:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                            consumed_train_samples=None,
+12/12/2024 22:02:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                            benchmark_csv_path=PosixPath('benchmark/results/bench_final.csv'),
+12/12/2024 22:02:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                            ignore_sanity_checks=True),
+12/12/2024 22:02:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:        parallelism=ParallelismArgs(dp=4,
+12/12/2024 22:02:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                    pp=1,
+12/12/2024 22:02:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                    tp=2,
+12/12/2024 22:02:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                    pp_engine=<nanotron.parallel.pipeline_parallel.engine.OneForwardOneBackwardPipelineEngine object at 0x7fc0dae7ce80>,
+12/12/2024 22:02:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                    tp_mode=<TensorParallelLinearMode.REDUCE_SCATTER: 2>,
+12/12/2024 22:02:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                    tp_linear_async_communication=True,
+12/12/2024 22:02:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                    recompute_layer=False,
+12/12/2024 22:02:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                    tp_recompute_allgather=True,
+12/12/2024 22:02:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                    expert_parallel_size=1),
+12/12/2024 22:02:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:        model=ModelArgs(model_config=LlamaConfig(bos_token_id=0,
+12/12/2024 22:02:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                                 eos_token_id=0,
+12/12/2024 22:02:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                                 hidden_act='silu',
+12/12/2024 22:02:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                                 hidden_size=2048,
+12/12/2024 22:02:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                                 initializer_range=0.02,
+12/12/2024 22:02:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                                 intermediate_size=8192,
+12/12/2024 22:02:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                                 is_llama_config=True,
+12/12/2024 22:02:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                                 max_position_embeddings=2048,
+12/12/2024 22:02:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                                 num_attention_heads=32,
+12/12/2024 22:02:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                                 num_hidden_layers=16,
+12/12/2024 22:02:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                                 num_key_value_heads=8,
+12/12/2024 22:02:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                                 pad_token_id=None,
+12/12/2024 22:02:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                                 pretraining_tp=1,
+12/12/2024 22:02:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                                 rms_norm_eps=1e-05,
+12/12/2024 22:02:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                                 rope_scaling=None,
+12/12/2024 22:02:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                                 rope_theta=10000.0,
+12/12/2024 22:02:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                                 rope_interleaved=False,
+12/12/2024 22:02:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                                 tie_word_embeddings=True,
+12/12/2024 22:02:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                                 use_cache=True,
+12/12/2024 22:02:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                                 vocab_size=32768),
+12/12/2024 22:02:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                        init_method=RandomInit(std=0.02),
+12/12/2024 22:02:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                        dtype=torch.bfloat16,
+12/12/2024 22:02:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                        make_vocab_size_divisible_by=1,
+12/12/2024 22:02:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                        ddp_bucket_cap_mb=25),
+12/12/2024 22:02:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:        tokenizer=TokenizerArgs(tokenizer_name_or_path='robot-test/dummy-tokenizer-wordlevel',
+12/12/2024 22:02:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                tokenizer_revision=None,
+12/12/2024 22:02:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                tokenizer_max_length=None),
+12/12/2024 22:02:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:        checkpoints=CheckpointsArgs(checkpoints_path=PosixPath('checkpoints'),
+12/12/2024 22:02:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                    checkpoint_interval=10000,
+12/12/2024 22:02:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                    save_initial_state=False,
+12/12/2024 22:02:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                    save_final_state=False,
+12/12/2024 22:02:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                    resume_checkpoint_path=None,
+12/12/2024 22:02:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                    checkpoints_path_is_shared_file_system=False),
+12/12/2024 22:02:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:        logging=LoggingArgs(log_level='info',
+12/12/2024 22:02:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                            log_level_replica='info',
+12/12/2024 22:02:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                            iteration_step_info_interval=1),
+12/12/2024 22:02:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:        tokens=TokensArgs(sequence_length=2048,
+12/12/2024 22:02:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                          train_steps=100,
+12/12/2024 22:02:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                          micro_batch_size=2,
+12/12/2024 22:02:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                          batch_accumulation_per_replica=256,
+12/12/2024 22:02:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                          val_check_interval=100,
+12/12/2024 22:02:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                          limit_val_batches=0,
+12/12/2024 22:02:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                          limit_test_batches=0),
+12/12/2024 22:02:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:        optimizer=OptimizerArgs(optimizer_factory=AdamWOptimizerArgs(adam_eps=1e-08,
+12/12/2024 22:02:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                                                     adam_beta1=0.9,
+12/12/2024 22:02:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                                                     adam_beta2=0.95,
+12/12/2024 22:02:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                                                     torch_adam_is_fused=True,
+12/12/2024 22:02:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                                                     name='adamW'),
+12/12/2024 22:02:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                zero_stage=1,
+12/12/2024 22:02:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                weight_decay=0.01,
+12/12/2024 22:02:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                clip_grad=1.0,
+12/12/2024 22:02:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                accumulate_grad_in_fp32=True,
+12/12/2024 22:02:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                learning_rate_scheduler=LRSchedulerArgs(learning_rate=0.0003,
+12/12/2024 22:02:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                                                        lr_warmup_steps=2,
+12/12/2024 22:02:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                                                        lr_warmup_style='linear',
+12/12/2024 22:02:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                                                        lr_decay_style='cosine',
+12/12/2024 22:02:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                                                        lr_decay_steps=13,
+12/12/2024 22:02:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                                                        lr_decay_starting_step=None,
+12/12/2024 22:02:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                                                        min_decay_lr=1e-05)),
+12/12/2024 22:02:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:        data_stages=[DatasetStageArgs(name='Stable Training Stage',
+12/12/2024 22:02:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                      start_training_step=1,
+12/12/2024 22:02:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                      data=DataArgs(dataset=None,
+12/12/2024 22:02:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                                    seed=42,
+12/12/2024 22:02:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                                    num_loading_workers=1))],
+12/12/2024 22:02:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:        profiler=None,
+12/12/2024 22:02:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:        lighteval=None,
+12/12/2024 22:02:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:        s3_upload=None)
+12/12/2024 22:02:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]: Model Config:
+12/12/2024 22:02:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]: LlamaConfig(bos_token_id=0,
+12/12/2024 22:02:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:             eos_token_id=0,
+12/12/2024 22:02:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:             hidden_act='silu',
+12/12/2024 22:02:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:             hidden_size=2048,
+12/12/2024 22:02:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:             initializer_range=0.02,
+12/12/2024 22:02:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:             intermediate_size=8192,
+12/12/2024 22:02:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:             is_llama_config=True,
+12/12/2024 22:02:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:             max_position_embeddings=2048,
+12/12/2024 22:02:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:             num_attention_heads=32,
+12/12/2024 22:02:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:             num_hidden_layers=16,
+12/12/2024 22:02:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:             num_key_value_heads=8,
+12/12/2024 22:02:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:             pad_token_id=None,
+12/12/2024 22:02:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:             pretraining_tp=1,
+12/12/2024 22:02:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:             rms_norm_eps=1e-05,
+12/12/2024 22:02:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:             rope_scaling=None,
+12/12/2024 22:02:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:             rope_theta=10000.0,
+12/12/2024 22:02:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:             rope_interleaved=False,
+12/12/2024 22:02:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:             tie_word_embeddings=True,
+12/12/2024 22:02:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:             use_cache=True,
+12/12/2024 22:02:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:             vocab_size=32768)
+12/12/2024 22:02:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]: Building model..
+12/12/2024 22:02:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]: Initialize RoPE Theta = 10000.0
+12/12/2024 22:02:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]: Setting PP block ranks...
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+ip-26-0-160-103:2068009:2068291 [0] NCCL INFO Using network AWS Libfabric
+ip-26-0-160-103:2068010:2068295 [1] NCCL INFO Using network AWS Libfabric
+ip-26-0-160-103:2068012:2068293 [3] NCCL INFO Using network AWS Libfabric
+ip-26-0-160-103:2068015:2068294 [6] NCCL INFO Using network AWS Libfabric
+ip-26-0-160-103:2068011:2068289 [2] NCCL INFO Using network AWS Libfabric
+ip-26-0-160-103:2068013:2068296 [4] NCCL INFO Using network AWS Libfabric
+ip-26-0-160-103:2068014:2068298 [5] NCCL INFO Using network AWS Libfabric
+ip-26-0-160-103:2068016:2068297 [7] NCCL INFO Using network AWS Libfabric
+ip-26-0-160-103:2068012:2068293 [3] NCCL INFO comm 0xc126b60 rank 1 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0xfc4c75a801d21d42 - Init START
+ip-26-0-160-103:2068011:2068289 [2] NCCL INFO comm 0xbe06620 rank 0 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0xfc4c75a801d21d42 - Init START
+ip-26-0-160-103:2068009:2068291 [0] NCCL INFO comm 0xb6b2df0 rank 0 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0xa5b203c08395a96f - Init START
+ip-26-0-160-103:2068010:2068295 [1] NCCL INFO comm 0xc029060 rank 1 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0xa5b203c08395a96f - Init START
+ip-26-0-160-103:2068011:2068289 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /opt/aws-ofi-nccl/share/aws-ofi-nccl/xml/p5.48xl-topo.xml
+ip-26-0-160-103:2068009:2068291 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /opt/aws-ofi-nccl/share/aws-ofi-nccl/xml/p5.48xl-topo.xml
+ip-26-0-160-103:2068013:2068296 [4] NCCL INFO comm 0xbb13820 rank 0 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0x30d5d301e52d5dde - Init START
+ip-26-0-160-103:2068014:2068298 [5] NCCL INFO comm 0xb3f52f0 rank 1 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0x30d5d301e52d5dde - Init START
+ip-26-0-160-103:2068015:2068294 [6] NCCL INFO comm 0xb0f09f0 rank 0 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0xd096124edf8257a6 - Init START
+ip-26-0-160-103:2068016:2068297 [7] NCCL INFO comm 0xb441df0 rank 1 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0xd096124edf8257a6 - Init START
+ip-26-0-160-103:2068013:2068296 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /opt/aws-ofi-nccl/share/aws-ofi-nccl/xml/p5.48xl-topo.xml
+ip-26-0-160-103:2068015:2068294 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /opt/aws-ofi-nccl/share/aws-ofi-nccl/xml/p5.48xl-topo.xml
+ip-26-0-160-103:2068012:2068293 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /opt/aws-ofi-nccl/share/aws-ofi-nccl/xml/p5.48xl-topo.xml
+ip-26-0-160-103:2068010:2068295 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /opt/aws-ofi-nccl/share/aws-ofi-nccl/xml/p5.48xl-topo.xml
+ip-26-0-160-103:2068014:2068298 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /opt/aws-ofi-nccl/share/aws-ofi-nccl/xml/p5.48xl-topo.xml
+ip-26-0-160-103:2068016:2068297 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /opt/aws-ofi-nccl/share/aws-ofi-nccl/xml/p5.48xl-topo.xml
+ip-26-0-160-103:2068009:2068291 [0] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068009:2068291 [0] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068009:2068291 [0] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068009:2068291 [0] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068009:2068291 [0] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068009:2068291 [0] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068009:2068291 [0] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068009:2068291 [0] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068009:2068291 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-160-103:2068010:2068295 [1] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068010:2068295 [1] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068010:2068295 [1] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068010:2068295 [1] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068010:2068295 [1] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068010:2068295 [1] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068010:2068295 [1] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068010:2068295 [1] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068010:2068295 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-160-103:2068009:2068291 [0] NCCL INFO Channel 00/24 :    0   1
+ip-26-0-160-103:2068009:2068291 [0] NCCL INFO Channel 01/24 :    0   1
+ip-26-0-160-103:2068009:2068291 [0] NCCL INFO Channel 02/24 :    0   1
+ip-26-0-160-103:2068009:2068291 [0] NCCL INFO Channel 03/24 :    0   1
+ip-26-0-160-103:2068009:2068291 [0] NCCL INFO Channel 04/24 :    0   1
+ip-26-0-160-103:2068009:2068291 [0] NCCL INFO Channel 05/24 :    0   1
+ip-26-0-160-103:2068009:2068291 [0] NCCL INFO Channel 06/24 :    0   1
+ip-26-0-160-103:2068009:2068291 [0] NCCL INFO Channel 07/24 :    0   1
+ip-26-0-160-103:2068010:2068295 [1] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] -1/-1/-1->1->0 [3] -1/-1/-1->1->0 [4] -1/-1/-1->1->0 [5] -1/-1/-1->1->0 [6] 0/-1/-1->1->-1 [7] 0/-1/-1->1->-1 [8] 0/-1/-1->1->-1 [9] 0/-1/-1->1->-1 [10] 0/-1/-1->1->-1 [11] 0/-1/-1->1->-1 [12] -1/-1/-1->1->0 [13] -1/-1/-1->1->0 [14] -1/-1/-1->1->0 [15] -1/-1/-1->1->0 [16] -1/-1/-1->1->0 [17] -1/-1/-1->1->0 [18] 0/-1/-1->1->-1 [19] 0/-1/-1->1->-1 [20] 0/-1/-1->1->-1 [21] 0/-1/-1->1->-1 [22] 0/-1/-1->1->-1 [23] 0/-1/-1->1->-1
+ip-26-0-160-103:2068009:2068291 [0] NCCL INFO Channel 08/24 :    0   1
+ip-26-0-160-103:2068009:2068291 [0] NCCL INFO Channel 09/24 :    0   1
+ip-26-0-160-103:2068010:2068295 [1] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-160-103:2068009:2068291 [0] NCCL INFO Channel 10/24 :    0   1
+ip-26-0-160-103:2068009:2068291 [0] NCCL INFO Channel 11/24 :    0   1
+ip-26-0-160-103:2068009:2068291 [0] NCCL INFO Channel 12/24 :    0   1
+ip-26-0-160-103:2068009:2068291 [0] NCCL INFO Channel 13/24 :    0   1
+ip-26-0-160-103:2068009:2068291 [0] NCCL INFO Channel 14/24 :    0   1
+ip-26-0-160-103:2068009:2068291 [0] NCCL INFO Channel 15/24 :    0   1
+ip-26-0-160-103:2068009:2068291 [0] NCCL INFO Channel 16/24 :    0   1
+ip-26-0-160-103:2068009:2068291 [0] NCCL INFO Channel 17/24 :    0   1
+ip-26-0-160-103:2068009:2068291 [0] NCCL INFO Channel 18/24 :    0   1
+ip-26-0-160-103:2068009:2068291 [0] NCCL INFO Channel 19/24 :    0   1
+ip-26-0-160-103:2068009:2068291 [0] NCCL INFO Channel 20/24 :    0   1
+ip-26-0-160-103:2068009:2068291 [0] NCCL INFO Channel 21/24 :    0   1
+ip-26-0-160-103:2068009:2068291 [0] NCCL INFO Channel 22/24 :    0   1
+ip-26-0-160-103:2068009:2068291 [0] NCCL INFO Channel 23/24 :    0   1
+ip-26-0-160-103:2068009:2068291 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] -1/-1/-1->0->1 [7] -1/-1/-1->0->1 [8] -1/-1/-1->0->1 [9] -1/-1/-1->0->1 [10] -1/-1/-1->0->1 [11] -1/-1/-1->0->1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] -1/-1/-1->0->1 [19] -1/-1/-1->0->1 [20] -1/-1/-1->0->1 [21] -1/-1/-1->0->1 [22] -1/-1/-1->0->1 [23] -1/-1/-1->0->1
+ip-26-0-160-103:2068009:2068291 [0] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-160-103:2068011:2068289 [2] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068011:2068289 [2] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068011:2068289 [2] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068011:2068289 [2] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068011:2068289 [2] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068011:2068289 [2] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068011:2068289 [2] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068011:2068289 [2] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068011:2068289 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-160-103:2068013:2068296 [4] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068013:2068296 [4] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068013:2068296 [4] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068013:2068296 [4] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068013:2068296 [4] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068013:2068296 [4] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068013:2068296 [4] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068013:2068296 [4] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068013:2068296 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-160-103:2068012:2068293 [3] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068012:2068293 [3] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068012:2068293 [3] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068012:2068293 [3] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068012:2068293 [3] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068012:2068293 [3] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068012:2068293 [3] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068015:2068294 [6] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068012:2068293 [3] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068015:2068294 [6] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068015:2068294 [6] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068015:2068294 [6] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068015:2068294 [6] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068015:2068294 [6] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068015:2068294 [6] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068015:2068294 [6] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068012:2068293 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-160-103:2068011:2068289 [2] NCCL INFO Channel 00/24 :    0   1
+ip-26-0-160-103:2068011:2068289 [2] NCCL INFO Channel 01/24 :    0   1
+ip-26-0-160-103:2068011:2068289 [2] NCCL INFO Channel 02/24 :    0   1
+ip-26-0-160-103:2068011:2068289 [2] NCCL INFO Channel 03/24 :    0   1
+ip-26-0-160-103:2068011:2068289 [2] NCCL INFO Channel 04/24 :    0   1
+ip-26-0-160-103:2068011:2068289 [2] NCCL INFO Channel 05/24 :    0   1
+ip-26-0-160-103:2068011:2068289 [2] NCCL INFO Channel 06/24 :    0   1
+ip-26-0-160-103:2068012:2068293 [3] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] -1/-1/-1->1->0 [3] -1/-1/-1->1->0 [4] -1/-1/-1->1->0 [5] -1/-1/-1->1->0 [6] 0/-1/-1->1->-1 [7] 0/-1/-1->1->-1 [8] 0/-1/-1->1->-1 [9] 0/-1/-1->1->-1 [10] 0/-1/-1->1->-1 [11] 0/-1/-1->1->-1 [12] -1/-1/-1->1->0 [13] -1/-1/-1->1->0 [14] -1/-1/-1->1->0 [15] -1/-1/-1->1->0 [16] -1/-1/-1->1->0 [17] -1/-1/-1->1->0 [18] 0/-1/-1->1->-1 [19] 0/-1/-1->1->-1 [20] 0/-1/-1->1->-1 [21] 0/-1/-1->1->-1 [22] 0/-1/-1->1->-1 [23] 0/-1/-1->1->-1
+ip-26-0-160-103:2068011:2068289 [2] NCCL INFO Channel 07/24 :    0   1
+ip-26-0-160-103:2068011:2068289 [2] NCCL INFO Channel 08/24 :    0   1
+ip-26-0-160-103:2068012:2068293 [3] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-160-103:2068011:2068289 [2] NCCL INFO Channel 09/24 :    0   1
+ip-26-0-160-103:2068011:2068289 [2] NCCL INFO Channel 10/24 :    0   1
+ip-26-0-160-103:2068011:2068289 [2] NCCL INFO Channel 11/24 :    0   1
+ip-26-0-160-103:2068011:2068289 [2] NCCL INFO Channel 12/24 :    0   1
+ip-26-0-160-103:2068011:2068289 [2] NCCL INFO Channel 13/24 :    0   1
+ip-26-0-160-103:2068011:2068289 [2] NCCL INFO Channel 14/24 :    0   1
+ip-26-0-160-103:2068011:2068289 [2] NCCL INFO Channel 15/24 :    0   1
+ip-26-0-160-103:2068011:2068289 [2] NCCL INFO Channel 16/24 :    0   1
+ip-26-0-160-103:2068011:2068289 [2] NCCL INFO Channel 17/24 :    0   1
+ip-26-0-160-103:2068011:2068289 [2] NCCL INFO Channel 18/24 :    0   1
+ip-26-0-160-103:2068011:2068289 [2] NCCL INFO Channel 19/24 :    0   1
+ip-26-0-160-103:2068011:2068289 [2] NCCL INFO Channel 20/24 :    0   1
+ip-26-0-160-103:2068011:2068289 [2] NCCL INFO Channel 21/24 :    0   1
+ip-26-0-160-103:2068011:2068289 [2] NCCL INFO Channel 22/24 :    0   1
+ip-26-0-160-103:2068011:2068289 [2] NCCL INFO Channel 23/24 :    0   1
+ip-26-0-160-103:2068011:2068289 [2] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] -1/-1/-1->0->1 [7] -1/-1/-1->0->1 [8] -1/-1/-1->0->1 [9] -1/-1/-1->0->1 [10] -1/-1/-1->0->1 [11] -1/-1/-1->0->1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] -1/-1/-1->0->1 [19] -1/-1/-1->0->1 [20] -1/-1/-1->0->1 [21] -1/-1/-1->0->1 [22] -1/-1/-1->0->1 [23] -1/-1/-1->0->1
+ip-26-0-160-103:2068011:2068289 [2] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-160-103:2068016:2068297 [7] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068016:2068297 [7] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068016:2068297 [7] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068016:2068297 [7] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068016:2068297 [7] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068016:2068297 [7] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068015:2068294 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-160-103:2068016:2068297 [7] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068014:2068298 [5] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068016:2068297 [7] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068014:2068298 [5] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068014:2068298 [5] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068014:2068298 [5] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068016:2068297 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-160-103:2068014:2068298 [5] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068015:2068294 [6] NCCL INFO Channel 00/24 :    0   1
+ip-26-0-160-103:2068015:2068294 [6] NCCL INFO Channel 01/24 :    0   1
+ip-26-0-160-103:2068016:2068297 [7] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] -1/-1/-1->1->0 [3] -1/-1/-1->1->0 [4] -1/-1/-1->1->0 [5] -1/-1/-1->1->0 [6] 0/-1/-1->1->-1 [7] 0/-1/-1->1->-1 [8] 0/-1/-1->1->-1 [9] 0/-1/-1->1->-1 [10] 0/-1/-1->1->-1 [11] 0/-1/-1->1->-1 [12] -1/-1/-1->1->0 [13] -1/-1/-1->1->0 [14] -1/-1/-1->1->0 [15] -1/-1/-1->1->0 [16] -1/-1/-1->1->0 [17] -1/-1/-1->1->0 [18] 0/-1/-1->1->-1 [19] 0/-1/-1->1->-1 [20] 0/-1/-1->1->-1 [21] 0/-1/-1->1->-1 [22] 0/-1/-1->1->-1 [23] 0/-1/-1->1->-1
+ip-26-0-160-103:2068015:2068294 [6] NCCL INFO Channel 02/24 :    0   1
+ip-26-0-160-103:2068015:2068294 [6] NCCL INFO Channel 03/24 :    0   1
+ip-26-0-160-103:2068016:2068297 [7] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-160-103:2068015:2068294 [6] NCCL INFO Channel 04/24 :    0   1
+ip-26-0-160-103:2068015:2068294 [6] NCCL INFO Channel 05/24 :    0   1
+ip-26-0-160-103:2068015:2068294 [6] NCCL INFO Channel 06/24 :    0   1
+ip-26-0-160-103:2068015:2068294 [6] NCCL INFO Channel 07/24 :    0   1
+ip-26-0-160-103:2068015:2068294 [6] NCCL INFO Channel 08/24 :    0   1
+ip-26-0-160-103:2068015:2068294 [6] NCCL INFO Channel 09/24 :    0   1
+ip-26-0-160-103:2068015:2068294 [6] NCCL INFO Channel 10/24 :    0   1
+ip-26-0-160-103:2068015:2068294 [6] NCCL INFO Channel 11/24 :    0   1
+ip-26-0-160-103:2068015:2068294 [6] NCCL INFO Channel 12/24 :    0   1
+ip-26-0-160-103:2068015:2068294 [6] NCCL INFO Channel 13/24 :    0   1
+ip-26-0-160-103:2068015:2068294 [6] NCCL INFO Channel 14/24 :    0   1
+ip-26-0-160-103:2068015:2068294 [6] NCCL INFO Channel 15/24 :    0   1
+ip-26-0-160-103:2068015:2068294 [6] NCCL INFO Channel 16/24 :    0   1
+ip-26-0-160-103:2068015:2068294 [6] NCCL INFO Channel 17/24 :    0   1
+ip-26-0-160-103:2068015:2068294 [6] NCCL INFO Channel 18/24 :    0   1
+ip-26-0-160-103:2068015:2068294 [6] NCCL INFO Channel 19/24 :    0   1
+ip-26-0-160-103:2068015:2068294 [6] NCCL INFO Channel 20/24 :    0   1
+ip-26-0-160-103:2068015:2068294 [6] NCCL INFO Channel 21/24 :    0   1
+ip-26-0-160-103:2068015:2068294 [6] NCCL INFO Channel 22/24 :    0   1
+ip-26-0-160-103:2068015:2068294 [6] NCCL INFO Channel 23/24 :    0   1
+ip-26-0-160-103:2068015:2068294 [6] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] -1/-1/-1->0->1 [7] -1/-1/-1->0->1 [8] -1/-1/-1->0->1 [9] -1/-1/-1->0->1 [10] -1/-1/-1->0->1 [11] -1/-1/-1->0->1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] -1/-1/-1->0->1 [19] -1/-1/-1->0->1 [20] -1/-1/-1->0->1 [21] -1/-1/-1->0->1 [22] -1/-1/-1->0->1 [23] -1/-1/-1->0->1
+ip-26-0-160-103:2068015:2068294 [6] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-160-103:2068014:2068298 [5] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068014:2068298 [5] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068014:2068298 [5] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068014:2068298 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-160-103:2068013:2068296 [4] NCCL INFO Channel 00/24 :    0   1
+ip-26-0-160-103:2068013:2068296 [4] NCCL INFO Channel 01/24 :    0   1
+ip-26-0-160-103:2068013:2068296 [4] NCCL INFO Channel 02/24 :    0   1
+ip-26-0-160-103:2068013:2068296 [4] NCCL INFO Channel 03/24 :    0   1
+ip-26-0-160-103:2068013:2068296 [4] NCCL INFO Channel 04/24 :    0   1
+ip-26-0-160-103:2068013:2068296 [4] NCCL INFO Channel 05/24 :    0   1
+ip-26-0-160-103:2068013:2068296 [4] NCCL INFO Channel 06/24 :    0   1
+ip-26-0-160-103:2068013:2068296 [4] NCCL INFO Channel 07/24 :    0   1
+ip-26-0-160-103:2068013:2068296 [4] NCCL INFO Channel 08/24 :    0   1
+ip-26-0-160-103:2068013:2068296 [4] NCCL INFO Channel 09/24 :    0   1
+ip-26-0-160-103:2068013:2068296 [4] NCCL INFO Channel 10/24 :    0   1
+ip-26-0-160-103:2068014:2068298 [5] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] -1/-1/-1->1->0 [3] -1/-1/-1->1->0 [4] -1/-1/-1->1->0 [5] -1/-1/-1->1->0 [6] 0/-1/-1->1->-1 [7] 0/-1/-1->1->-1 [8] 0/-1/-1->1->-1 [9] 0/-1/-1->1->-1 [10] 0/-1/-1->1->-1 [11] 0/-1/-1->1->-1 [12] -1/-1/-1->1->0 [13] -1/-1/-1->1->0 [14] -1/-1/-1->1->0 [15] -1/-1/-1->1->0 [16] -1/-1/-1->1->0 [17] -1/-1/-1->1->0 [18] 0/-1/-1->1->-1 [19] 0/-1/-1->1->-1 [20] 0/-1/-1->1->-1 [21] 0/-1/-1->1->-1 [22] 0/-1/-1->1->-1 [23] 0/-1/-1->1->-1
+ip-26-0-160-103:2068013:2068296 [4] NCCL INFO Channel 11/24 :    0   1
+ip-26-0-160-103:2068013:2068296 [4] NCCL INFO Channel 12/24 :    0   1
+ip-26-0-160-103:2068014:2068298 [5] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-160-103:2068013:2068296 [4] NCCL INFO Channel 13/24 :    0   1
+ip-26-0-160-103:2068013:2068296 [4] NCCL INFO Channel 14/24 :    0   1
+ip-26-0-160-103:2068013:2068296 [4] NCCL INFO Channel 15/24 :    0   1
+ip-26-0-160-103:2068013:2068296 [4] NCCL INFO Channel 16/24 :    0   1
+ip-26-0-160-103:2068013:2068296 [4] NCCL INFO Channel 17/24 :    0   1
+ip-26-0-160-103:2068013:2068296 [4] NCCL INFO Channel 18/24 :    0   1
+ip-26-0-160-103:2068013:2068296 [4] NCCL INFO Channel 19/24 :    0   1
+ip-26-0-160-103:2068013:2068296 [4] NCCL INFO Channel 20/24 :    0   1
+ip-26-0-160-103:2068013:2068296 [4] NCCL INFO Channel 21/24 :    0   1
+ip-26-0-160-103:2068013:2068296 [4] NCCL INFO Channel 22/24 :    0   1
+ip-26-0-160-103:2068013:2068296 [4] NCCL INFO Channel 23/24 :    0   1
+ip-26-0-160-103:2068013:2068296 [4] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] -1/-1/-1->0->1 [7] -1/-1/-1->0->1 [8] -1/-1/-1->0->1 [9] -1/-1/-1->0->1 [10] -1/-1/-1->0->1 [11] -1/-1/-1->0->1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] -1/-1/-1->0->1 [19] -1/-1/-1->0->1 [20] -1/-1/-1->0->1 [21] -1/-1/-1->0->1 [22] -1/-1/-1->0->1 [23] -1/-1/-1->0->1
+ip-26-0-160-103:2068013:2068296 [4] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-160-103:2068009:2068291 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068011:2068289 [2] NCCL INFO Channel 00/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2068009:2068291 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068009:2068291 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068009:2068291 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068009:2068291 [0] NCCL INFO Channel 04/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068015:2068294 [6] NCCL INFO Channel 00/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-160-103:2068009:2068291 [0] NCCL INFO Channel 05/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068015:2068294 [6] NCCL INFO Channel 01/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-160-103:2068009:2068291 [0] NCCL INFO Channel 06/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068015:2068294 [6] NCCL INFO Channel 02/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-160-103:2068011:2068289 [2] NCCL INFO Channel 01/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2068009:2068291 [0] NCCL INFO Channel 07/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068015:2068294 [6] NCCL INFO Channel 03/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-160-103:2068011:2068289 [2] NCCL INFO Channel 02/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2068009:2068291 [0] NCCL INFO Channel 08/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068012:2068293 [3] NCCL INFO Channel 00/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-160-103:2068015:2068294 [6] NCCL INFO Channel 04/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-160-103:2068011:2068289 [2] NCCL INFO Channel 03/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2068009:2068291 [0] NCCL INFO Channel 09/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068015:2068294 [6] NCCL INFO Channel 05/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-160-103:2068012:2068293 [3] NCCL INFO Channel 01/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-160-103:2068011:2068289 [2] NCCL INFO Channel 04/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2068009:2068291 [0] NCCL INFO Channel 10/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068010:2068295 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068011:2068289 [2] NCCL INFO Channel 05/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2068009:2068291 [0] NCCL INFO Channel 11/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068015:2068294 [6] NCCL INFO Channel 06/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-160-103:2068015:2068294 [6] NCCL INFO Channel 07/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-160-103:2068015:2068294 [6] NCCL INFO Channel 08/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-160-103:2068015:2068294 [6] NCCL INFO Channel 09/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-160-103:2068015:2068294 [6] NCCL INFO Channel 10/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-160-103:2068015:2068294 [6] NCCL INFO Channel 11/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-160-103:2068015:2068294 [6] NCCL INFO Channel 12/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-160-103:2068014:2068298 [5] NCCL INFO Channel 00/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-160-103:2068015:2068294 [6] NCCL INFO Channel 13/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-160-103:2068014:2068298 [5] NCCL INFO Channel 01/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-160-103:2068015:2068294 [6] NCCL INFO Channel 14/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-160-103:2068014:2068298 [5] NCCL INFO Channel 02/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-160-103:2068015:2068294 [6] NCCL INFO Channel 15/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-160-103:2068014:2068298 [5] NCCL INFO Channel 03/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-160-103:2068016:2068297 [7] NCCL INFO Channel 00/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-160-103:2068015:2068294 [6] NCCL INFO Channel 16/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-160-103:2068009:2068291 [0] NCCL INFO Channel 12/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068016:2068297 [7] NCCL INFO Channel 01/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-160-103:2068012:2068293 [3] NCCL INFO Channel 02/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-160-103:2068015:2068294 [6] NCCL INFO Channel 17/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-160-103:2068009:2068291 [0] NCCL INFO Channel 13/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068016:2068297 [7] NCCL INFO Channel 02/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-160-103:2068015:2068294 [6] NCCL INFO Channel 18/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-160-103:2068012:2068293 [3] NCCL INFO Channel 03/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-160-103:2068009:2068291 [0] NCCL INFO Channel 14/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068016:2068297 [7] NCCL INFO Channel 03/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-160-103:2068015:2068294 [6] NCCL INFO Channel 19/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-160-103:2068012:2068293 [3] NCCL INFO Channel 04/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-160-103:2068009:2068291 [0] NCCL INFO Channel 15/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068016:2068297 [7] NCCL INFO Channel 04/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-160-103:2068015:2068294 [6] NCCL INFO Channel 20/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-160-103:2068012:2068293 [3] NCCL INFO Channel 05/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-160-103:2068009:2068291 [0] NCCL INFO Channel 16/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068016:2068297 [7] NCCL INFO Channel 05/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-160-103:2068015:2068294 [6] NCCL INFO Channel 21/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-160-103:2068012:2068293 [3] NCCL INFO Channel 06/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-160-103:2068009:2068291 [0] NCCL INFO Channel 17/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068016:2068297 [7] NCCL INFO Channel 06/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-160-103:2068015:2068294 [6] NCCL INFO Channel 22/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-160-103:2068012:2068293 [3] NCCL INFO Channel 07/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-160-103:2068009:2068291 [0] NCCL INFO Channel 18/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068016:2068297 [7] NCCL INFO Channel 07/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-160-103:2068015:2068294 [6] NCCL INFO Channel 23/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-160-103:2068012:2068293 [3] NCCL INFO Channel 08/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-160-103:2068009:2068291 [0] NCCL INFO Channel 19/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068016:2068297 [7] NCCL INFO Channel 08/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-160-103:2068012:2068293 [3] NCCL INFO Channel 09/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-160-103:2068009:2068291 [0] NCCL INFO Channel 20/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068010:2068295 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068016:2068297 [7] NCCL INFO Channel 09/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-160-103:2068012:2068293 [3] NCCL INFO Channel 10/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-160-103:2068009:2068291 [0] NCCL INFO Channel 21/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068010:2068295 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068016:2068297 [7] NCCL INFO Channel 10/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-160-103:2068012:2068293 [3] NCCL INFO Channel 11/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-160-103:2068009:2068291 [0] NCCL INFO Channel 22/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068016:2068297 [7] NCCL INFO Channel 11/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-160-103:2068010:2068295 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068012:2068293 [3] NCCL INFO Channel 12/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-160-103:2068011:2068289 [2] NCCL INFO Channel 06/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2068014:2068298 [5] NCCL INFO Channel 04/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-160-103:2068009:2068291 [0] NCCL INFO Channel 23/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2068016:2068297 [7] NCCL INFO Channel 12/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-160-103:2068010:2068295 [1] NCCL INFO Channel 04/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068012:2068293 [3] NCCL INFO Channel 13/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-160-103:2068011:2068289 [2] NCCL INFO Channel 07/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2068014:2068298 [5] NCCL INFO Channel 05/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-160-103:2068012:2068293 [3] NCCL INFO Channel 14/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-160-103:2068016:2068297 [7] NCCL INFO Channel 13/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-160-103:2068010:2068295 [1] NCCL INFO Channel 05/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068011:2068289 [2] NCCL INFO Channel 08/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2068014:2068298 [5] NCCL INFO Channel 06/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-160-103:2068012:2068293 [3] NCCL INFO Channel 15/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-160-103:2068016:2068297 [7] NCCL INFO Channel 14/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-160-103:2068010:2068295 [1] NCCL INFO Channel 06/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068011:2068289 [2] NCCL INFO Channel 09/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2068014:2068298 [5] NCCL INFO Channel 07/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-160-103:2068012:2068293 [3] NCCL INFO Channel 16/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-160-103:2068016:2068297 [7] NCCL INFO Channel 15/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-160-103:2068010:2068295 [1] NCCL INFO Channel 07/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068011:2068289 [2] NCCL INFO Channel 10/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2068014:2068298 [5] NCCL INFO Channel 08/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-160-103:2068012:2068293 [3] NCCL INFO Channel 17/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-160-103:2068010:2068295 [1] NCCL INFO Channel 08/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068016:2068297 [7] NCCL INFO Channel 16/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-160-103:2068011:2068289 [2] NCCL INFO Channel 11/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2068014:2068298 [5] NCCL INFO Channel 09/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-160-103:2068012:2068293 [3] NCCL INFO Channel 18/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-160-103:2068010:2068295 [1] NCCL INFO Channel 09/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068016:2068297 [7] NCCL INFO Channel 17/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-160-103:2068011:2068289 [2] NCCL INFO Channel 12/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2068014:2068298 [5] NCCL INFO Channel 10/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-160-103:2068013:2068296 [4] NCCL INFO Channel 00/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-160-103:2068012:2068293 [3] NCCL INFO Channel 19/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-160-103:2068010:2068295 [1] NCCL INFO Channel 10/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068016:2068297 [7] NCCL INFO Channel 18/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-160-103:2068011:2068289 [2] NCCL INFO Channel 13/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2068014:2068298 [5] NCCL INFO Channel 11/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-160-103:2068013:2068296 [4] NCCL INFO Channel 01/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-160-103:2068010:2068295 [1] NCCL INFO Channel 11/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068012:2068293 [3] NCCL INFO Channel 20/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-160-103:2068016:2068297 [7] NCCL INFO Channel 19/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-160-103:2068011:2068289 [2] NCCL INFO Channel 14/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2068014:2068298 [5] NCCL INFO Channel 12/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-160-103:2068013:2068296 [4] NCCL INFO Channel 02/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-160-103:2068010:2068295 [1] NCCL INFO Channel 12/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068016:2068297 [7] NCCL INFO Channel 20/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-160-103:2068012:2068293 [3] NCCL INFO Channel 21/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-160-103:2068011:2068289 [2] NCCL INFO Channel 15/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2068014:2068298 [5] NCCL INFO Channel 13/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-160-103:2068013:2068296 [4] NCCL INFO Channel 03/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-160-103:2068010:2068295 [1] NCCL INFO Channel 13/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068016:2068297 [7] NCCL INFO Channel 21/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-160-103:2068011:2068289 [2] NCCL INFO Channel 16/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2068012:2068293 [3] NCCL INFO Channel 22/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-160-103:2068014:2068298 [5] NCCL INFO Channel 14/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-160-103:2068013:2068296 [4] NCCL INFO Channel 04/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-160-103:2068010:2068295 [1] NCCL INFO Channel 14/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068016:2068297 [7] NCCL INFO Channel 22/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-160-103:2068011:2068289 [2] NCCL INFO Channel 17/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2068014:2068298 [5] NCCL INFO Channel 15/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-160-103:2068012:2068293 [3] NCCL INFO Channel 23/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-160-103:2068013:2068296 [4] NCCL INFO Channel 05/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-160-103:2068010:2068295 [1] NCCL INFO Channel 15/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068016:2068297 [7] NCCL INFO Channel 23/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-160-103:2068011:2068289 [2] NCCL INFO Channel 18/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2068014:2068298 [5] NCCL INFO Channel 16/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-160-103:2068010:2068295 [1] NCCL INFO Channel 16/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068013:2068296 [4] NCCL INFO Channel 06/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-160-103:2068011:2068289 [2] NCCL INFO Channel 19/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2068014:2068298 [5] NCCL INFO Channel 17/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-160-103:2068010:2068295 [1] NCCL INFO Channel 17/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068013:2068296 [4] NCCL INFO Channel 07/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-160-103:2068011:2068289 [2] NCCL INFO Channel 20/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2068014:2068298 [5] NCCL INFO Channel 18/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-160-103:2068010:2068295 [1] NCCL INFO Channel 18/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068013:2068296 [4] NCCL INFO Channel 08/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-160-103:2068011:2068289 [2] NCCL INFO Channel 21/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2068014:2068298 [5] NCCL INFO Channel 19/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-160-103:2068013:2068296 [4] NCCL INFO Channel 09/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-160-103:2068010:2068295 [1] NCCL INFO Channel 19/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068011:2068289 [2] NCCL INFO Channel 22/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2068014:2068298 [5] NCCL INFO Channel 20/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-160-103:2068010:2068295 [1] NCCL INFO Channel 20/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068013:2068296 [4] NCCL INFO Channel 10/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-160-103:2068011:2068289 [2] NCCL INFO Channel 23/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2068014:2068298 [5] NCCL INFO Channel 21/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-160-103:2068010:2068295 [1] NCCL INFO Channel 21/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068013:2068296 [4] NCCL INFO Channel 11/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-160-103:2068014:2068298 [5] NCCL INFO Channel 22/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-160-103:2068014:2068298 [5] NCCL INFO Channel 23/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-160-103:2068010:2068295 [1] NCCL INFO Channel 22/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068013:2068296 [4] NCCL INFO Channel 12/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-160-103:2068010:2068295 [1] NCCL INFO Channel 23/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068013:2068296 [4] NCCL INFO Channel 13/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-160-103:2068013:2068296 [4] NCCL INFO Channel 14/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-160-103:2068013:2068296 [4] NCCL INFO Channel 15/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-160-103:2068013:2068296 [4] NCCL INFO Channel 16/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-160-103:2068013:2068296 [4] NCCL INFO Channel 17/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-160-103:2068013:2068296 [4] NCCL INFO Channel 18/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-160-103:2068013:2068296 [4] NCCL INFO Channel 19/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-160-103:2068013:2068296 [4] NCCL INFO Channel 20/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-160-103:2068013:2068296 [4] NCCL INFO Channel 21/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-160-103:2068013:2068296 [4] NCCL INFO Channel 22/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-160-103:2068013:2068296 [4] NCCL INFO Channel 23/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-160-103:2068015:2068294 [6] NCCL INFO Connected all rings
+ip-26-0-160-103:2068015:2068294 [6] NCCL INFO Connected all trees
+ip-26-0-160-103:2068015:2068294 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-103:2068015:2068294 [6] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-160-103:2068015:2068294 [6] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-160-103:2068016:2068297 [7] NCCL INFO Connected all rings
+ip-26-0-160-103:2068016:2068297 [7] NCCL INFO Connected all trees
+ip-26-0-160-103:2068016:2068297 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-103:2068016:2068297 [7] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-160-103:2068016:2068297 [7] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-160-103:2068012:2068293 [3] NCCL INFO Connected all rings
+ip-26-0-160-103:2068012:2068293 [3] NCCL INFO Connected all trees
+ip-26-0-160-103:2068012:2068293 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-103:2068012:2068293 [3] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-160-103:2068012:2068293 [3] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-160-103:2068016:2068297 [7] NCCL INFO comm 0xb441df0 rank 1 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0xd096124edf8257a6 - Init COMPLETE
+ip-26-0-160-103:2068015:2068294 [6] NCCL INFO comm 0xb0f09f0 rank 0 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0xd096124edf8257a6 - Init COMPLETE
+ip-26-0-160-103:2068009:2068291 [0] NCCL INFO Connected all rings
+ip-26-0-160-103:2068009:2068291 [0] NCCL INFO Connected all trees
+ip-26-0-160-103:2068009:2068291 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-103:2068009:2068291 [0] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-160-103:2068009:2068291 [0] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-160-103:2068011:2068289 [2] NCCL INFO Connected all rings
+ip-26-0-160-103:2068011:2068289 [2] NCCL INFO Connected all trees
+ip-26-0-160-103:2068011:2068289 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-103:2068011:2068289 [2] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-160-103:2068011:2068289 [2] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+NCCL version 2.18.5+cuda12.2
+ip-26-0-160-103:2068015:2068316 [6] NCCL INFO Using network AWS Libfabric
+ip-26-0-160-103:2068016:2068314 [7] NCCL INFO Using network AWS Libfabric
+ip-26-0-160-103:2068015:2068316 [6] NCCL INFO comm 0xb104d70 rank 0 nranks 1 cudaDev 6 nvmlDev 6 busId b9000 commId 0xbe3e3bb987a9de04 - Init START
+ip-26-0-160-103:2068016:2068314 [7] NCCL INFO comm 0xb4557a0 rank 0 nranks 1 cudaDev 7 nvmlDev 7 busId ca000 commId 0xf95d4c4d4f14b109 - Init START
+ip-26-0-160-103:2068015:2068316 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /opt/aws-ofi-nccl/share/aws-ofi-nccl/xml/p5.48xl-topo.xml
+ip-26-0-160-103:2068016:2068314 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /opt/aws-ofi-nccl/share/aws-ofi-nccl/xml/p5.48xl-topo.xml
+ip-26-0-160-103:2068010:2068295 [1] NCCL INFO Connected all rings
+ip-26-0-160-103:2068010:2068295 [1] NCCL INFO Connected all trees
+ip-26-0-160-103:2068010:2068295 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-103:2068010:2068295 [1] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-160-103:2068010:2068295 [1] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-160-103:2068011:2068289 [2] NCCL INFO comm 0xbe06620 rank 0 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0xfc4c75a801d21d42 - Init COMPLETE
+ip-26-0-160-103:2068012:2068293 [3] NCCL INFO comm 0xc126b60 rank 1 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0xfc4c75a801d21d42 - Init COMPLETE
+NCCL version 2.18.5+cuda12.2
+ip-26-0-160-103:2068012:2068320 [3] NCCL INFO Using network AWS Libfabric
+ip-26-0-160-103:2068012:2068320 [3] NCCL INFO comm 0xc13a440 rank 0 nranks 1 cudaDev 3 nvmlDev 3 busId 86000 commId 0xc2cbd04264fa9e0c - Init START
+ip-26-0-160-103:2068012:2068320 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /opt/aws-ofi-nccl/share/aws-ofi-nccl/xml/p5.48xl-topo.xml
+ip-26-0-160-103:2068010:2068295 [1] NCCL INFO comm 0xc029060 rank 1 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0xa5b203c08395a96f - Init COMPLETE
+ip-26-0-160-103:2068009:2068291 [0] NCCL INFO comm 0xb6b2df0 rank 0 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0xa5b203c08395a96f - Init COMPLETE
+ip-26-0-160-103:2068011:2068322 [2] NCCL INFO Using network AWS Libfabric
+ip-26-0-160-103:2068011:2068322 [2] NCCL INFO comm 0xbe1a250 rank 0 nranks 1 cudaDev 2 nvmlDev 2 busId 75000 commId 0x59b3bcc53f35f7e9 - Init START
+ip-26-0-160-103:2068011:2068322 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /opt/aws-ofi-nccl/share/aws-ofi-nccl/xml/p5.48xl-topo.xml
+NCCL version 2.18.5+cuda12.2
+ip-26-0-160-103:2068010:2068324 [1] NCCL INFO Using network AWS Libfabric
+ip-26-0-160-103:2068010:2068324 [1] NCCL INFO comm 0xc03cea0 rank 0 nranks 1 cudaDev 1 nvmlDev 1 busId 64000 commId 0x3a509b9a635d2b6b - Init START
+ip-26-0-160-103:2068010:2068324 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /opt/aws-ofi-nccl/share/aws-ofi-nccl/xml/p5.48xl-topo.xml
+ip-26-0-160-103:2068009:2068326 [0] NCCL INFO Using network AWS Libfabric
+ip-26-0-160-103:2068009:2068326 [0] NCCL INFO comm 0xb6c6c80 rank 0 nranks 1 cudaDev 0 nvmlDev 0 busId 53000 commId 0x47ad8297bfe6e876 - Init START
+ip-26-0-160-103:2068009:2068326 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /opt/aws-ofi-nccl/share/aws-ofi-nccl/xml/p5.48xl-topo.xml
+ip-26-0-160-103:2068014:2068298 [5] NCCL INFO Connected all rings
+ip-26-0-160-103:2068014:2068298 [5] NCCL INFO Connected all trees
+ip-26-0-160-103:2068014:2068298 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-103:2068014:2068298 [5] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-160-103:2068014:2068298 [5] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-160-103:2068013:2068296 [4] NCCL INFO Connected all rings
+ip-26-0-160-103:2068013:2068296 [4] NCCL INFO Connected all trees
+ip-26-0-160-103:2068013:2068296 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-103:2068013:2068296 [4] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-160-103:2068013:2068296 [4] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-160-103:2068013:2068296 [4] NCCL INFO comm 0xbb13820 rank 0 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0x30d5d301e52d5dde - Init COMPLETE
+ip-26-0-160-103:2068014:2068298 [5] NCCL INFO comm 0xb3f52f0 rank 1 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0x30d5d301e52d5dde - Init COMPLETE
+NCCL version 2.18.5+cuda12.2
+ip-26-0-160-103:2068013:2068330 [4] NCCL INFO Using network AWS Libfabric
+ip-26-0-160-103:2068013:2068330 [4] NCCL INFO comm 0xbb27720 rank 0 nranks 1 cudaDev 4 nvmlDev 4 busId 97000 commId 0xc0dc7d61e0e5d53e - Init START
+ip-26-0-160-103:2068013:2068330 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /opt/aws-ofi-nccl/share/aws-ofi-nccl/xml/p5.48xl-topo.xml
+ip-26-0-160-103:2068014:2068332 [5] NCCL INFO Using network AWS Libfabric
+ip-26-0-160-103:2068014:2068332 [5] NCCL INFO comm 0xb408f20 rank 0 nranks 1 cudaDev 5 nvmlDev 5 busId a8000 commId 0xf554a31af9bccf28 - Init START
+ip-26-0-160-103:2068014:2068332 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /opt/aws-ofi-nccl/share/aws-ofi-nccl/xml/p5.48xl-topo.xml
+ip-26-0-160-103:2068009:2068326 [0] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068009:2068326 [0] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068009:2068326 [0] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068009:2068326 [0] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068009:2068326 [0] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068009:2068326 [0] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068012:2068320 [3] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068009:2068326 [0] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068012:2068320 [3] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068009:2068326 [0] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068012:2068320 [3] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068012:2068320 [3] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068012:2068320 [3] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068012:2068320 [3] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068012:2068320 [3] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068012:2068320 [3] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068009:2068326 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-160-103:2068009:2068326 [0] NCCL INFO Channel 00/32 :    0
+ip-26-0-160-103:2068009:2068326 [0] NCCL INFO Channel 01/32 :    0
+ip-26-0-160-103:2068009:2068326 [0] NCCL INFO Channel 02/32 :    0
+ip-26-0-160-103:2068009:2068326 [0] NCCL INFO Channel 03/32 :    0
+ip-26-0-160-103:2068009:2068326 [0] NCCL INFO Channel 04/32 :    0
+ip-26-0-160-103:2068009:2068326 [0] NCCL INFO Channel 05/32 :    0
+ip-26-0-160-103:2068009:2068326 [0] NCCL INFO Channel 06/32 :    0
+ip-26-0-160-103:2068009:2068326 [0] NCCL INFO Channel 07/32 :    0
+ip-26-0-160-103:2068009:2068326 [0] NCCL INFO Channel 08/32 :    0
+ip-26-0-160-103:2068009:2068326 [0] NCCL INFO Channel 09/32 :    0
+ip-26-0-160-103:2068009:2068326 [0] NCCL INFO Channel 10/32 :    0
+ip-26-0-160-103:2068009:2068326 [0] NCCL INFO Channel 11/32 :    0
+ip-26-0-160-103:2068009:2068326 [0] NCCL INFO Channel 12/32 :    0
+ip-26-0-160-103:2068009:2068326 [0] NCCL INFO Channel 13/32 :    0
+ip-26-0-160-103:2068009:2068326 [0] NCCL INFO Channel 14/32 :    0
+ip-26-0-160-103:2068009:2068326 [0] NCCL INFO Channel 15/32 :    0
+ip-26-0-160-103:2068009:2068326 [0] NCCL INFO Channel 16/32 :    0
+ip-26-0-160-103:2068009:2068326 [0] NCCL INFO Channel 17/32 :    0
+ip-26-0-160-103:2068009:2068326 [0] NCCL INFO Channel 18/32 :    0
+ip-26-0-160-103:2068009:2068326 [0] NCCL INFO Channel 19/32 :    0
+ip-26-0-160-103:2068009:2068326 [0] NCCL INFO Channel 20/32 :    0
+ip-26-0-160-103:2068009:2068326 [0] NCCL INFO Channel 21/32 :    0
+ip-26-0-160-103:2068009:2068326 [0] NCCL INFO Channel 22/32 :    0
+ip-26-0-160-103:2068009:2068326 [0] NCCL INFO Channel 23/32 :    0
+ip-26-0-160-103:2068009:2068326 [0] NCCL INFO Channel 24/32 :    0
+ip-26-0-160-103:2068009:2068326 [0] NCCL INFO Channel 25/32 :    0
+ip-26-0-160-103:2068009:2068326 [0] NCCL INFO Channel 26/32 :    0
+ip-26-0-160-103:2068009:2068326 [0] NCCL INFO Channel 27/32 :    0
+ip-26-0-160-103:2068009:2068326 [0] NCCL INFO Channel 28/32 :    0
+ip-26-0-160-103:2068009:2068326 [0] NCCL INFO Channel 29/32 :    0
+ip-26-0-160-103:2068009:2068326 [0] NCCL INFO Channel 30/32 :    0
+ip-26-0-160-103:2068009:2068326 [0] NCCL INFO Channel 31/32 :    0
+ip-26-0-160-103:2068009:2068326 [0] NCCL INFO Trees [0] -1/-1/-1->0->-1 [1] -1/-1/-1->0->-1 [2] -1/-1/-1->0->-1 [3] -1/-1/-1->0->-1 [4] -1/-1/-1->0->-1 [5] -1/-1/-1->0->-1 [6] -1/-1/-1->0->-1 [7] -1/-1/-1->0->-1 [8] -1/-1/-1->0->-1 [9] -1/-1/-1->0->-1 [10] -1/-1/-1->0->-1 [11] -1/-1/-1->0->-1 [12] -1/-1/-1->0->-1 [13] -1/-1/-1->0->-1 [14] -1/-1/-1->0->-1 [15] -1/-1/-1->0->-1 [16] -1/-1/-1->0->-1 [17] -1/-1/-1->0->-1 [18] -1/-1/-1->0->-1 [19] -1/-1/-1->0->-1 [20] -1/-1/-1->0->-1 [21] -1/-1/-1->0->-1 [22] -1/-1/-1->0->-1 [23] -1/-1/-1->0->-1 [24] -1/-1/-1->0->-1 [25] -1/-1/-1->0->-1 [26] -1/-1/-1->0->-1 [27] -1/-1/-1->0->-1 [28] -1/-1/-1->0->-1 [29] -1/-1/-1->0->-1 [30] -1/-1/-1->0->-1 [31] -1/-1/-1->0->-1
+ip-26-0-160-103:2068009:2068326 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-103:2068012:2068320 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-160-103:2068012:2068320 [3] NCCL INFO Channel 00/32 :    0
+ip-26-0-160-103:2068012:2068320 [3] NCCL INFO Channel 01/32 :    0
+ip-26-0-160-103:2068012:2068320 [3] NCCL INFO Channel 02/32 :    0
+ip-26-0-160-103:2068012:2068320 [3] NCCL INFO Channel 03/32 :    0
+ip-26-0-160-103:2068012:2068320 [3] NCCL INFO Channel 04/32 :    0
+ip-26-0-160-103:2068012:2068320 [3] NCCL INFO Channel 05/32 :    0
+ip-26-0-160-103:2068012:2068320 [3] NCCL INFO Channel 06/32 :    0
+ip-26-0-160-103:2068012:2068320 [3] NCCL INFO Channel 07/32 :    0
+ip-26-0-160-103:2068012:2068320 [3] NCCL INFO Channel 08/32 :    0
+ip-26-0-160-103:2068012:2068320 [3] NCCL INFO Channel 09/32 :    0
+ip-26-0-160-103:2068012:2068320 [3] NCCL INFO Channel 10/32 :    0
+ip-26-0-160-103:2068012:2068320 [3] NCCL INFO Channel 11/32 :    0
+ip-26-0-160-103:2068012:2068320 [3] NCCL INFO Channel 12/32 :    0
+ip-26-0-160-103:2068012:2068320 [3] NCCL INFO Channel 13/32 :    0
+ip-26-0-160-103:2068012:2068320 [3] NCCL INFO Channel 14/32 :    0
+ip-26-0-160-103:2068012:2068320 [3] NCCL INFO Channel 15/32 :    0
+ip-26-0-160-103:2068012:2068320 [3] NCCL INFO Channel 16/32 :    0
+ip-26-0-160-103:2068012:2068320 [3] NCCL INFO Channel 17/32 :    0
+ip-26-0-160-103:2068012:2068320 [3] NCCL INFO Channel 18/32 :    0
+ip-26-0-160-103:2068012:2068320 [3] NCCL INFO Channel 19/32 :    0
+ip-26-0-160-103:2068012:2068320 [3] NCCL INFO Channel 20/32 :    0
+ip-26-0-160-103:2068012:2068320 [3] NCCL INFO Channel 21/32 :    0
+ip-26-0-160-103:2068012:2068320 [3] NCCL INFO Channel 22/32 :    0
+ip-26-0-160-103:2068012:2068320 [3] NCCL INFO Channel 23/32 :    0
+ip-26-0-160-103:2068012:2068320 [3] NCCL INFO Channel 24/32 :    0
+ip-26-0-160-103:2068012:2068320 [3] NCCL INFO Channel 25/32 :    0
+ip-26-0-160-103:2068012:2068320 [3] NCCL INFO Channel 26/32 :    0
+ip-26-0-160-103:2068012:2068320 [3] NCCL INFO Channel 27/32 :    0
+ip-26-0-160-103:2068012:2068320 [3] NCCL INFO Channel 28/32 :    0
+ip-26-0-160-103:2068012:2068320 [3] NCCL INFO Channel 29/32 :    0
+ip-26-0-160-103:2068012:2068320 [3] NCCL INFO Channel 30/32 :    0
+ip-26-0-160-103:2068012:2068320 [3] NCCL INFO Channel 31/32 :    0
+ip-26-0-160-103:2068012:2068320 [3] NCCL INFO Trees [0] -1/-1/-1->0->-1 [1] -1/-1/-1->0->-1 [2] -1/-1/-1->0->-1 [3] -1/-1/-1->0->-1 [4] -1/-1/-1->0->-1 [5] -1/-1/-1->0->-1 [6] -1/-1/-1->0->-1 [7] -1/-1/-1->0->-1 [8] -1/-1/-1->0->-1 [9] -1/-1/-1->0->-1 [10] -1/-1/-1->0->-1 [11] -1/-1/-1->0->-1 [12] -1/-1/-1->0->-1 [13] -1/-1/-1->0->-1 [14] -1/-1/-1->0->-1 [15] -1/-1/-1->0->-1 [16] -1/-1/-1->0->-1 [17] -1/-1/-1->0->-1 [18] -1/-1/-1->0->-1 [19] -1/-1/-1->0->-1 [20] -1/-1/-1->0->-1 [21] -1/-1/-1->0->-1 [22] -1/-1/-1->0->-1 [23] -1/-1/-1->0->-1 [24] -1/-1/-1->0->-1 [25] -1/-1/-1->0->-1 [26] -1/-1/-1->0->-1 [27] -1/-1/-1->0->-1 [28] -1/-1/-1->0->-1 [29] -1/-1/-1->0->-1 [30] -1/-1/-1->0->-1 [31] -1/-1/-1->0->-1
+ip-26-0-160-103:2068012:2068320 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-103:2068012:2068320 [3] NCCL INFO Connected all rings
+ip-26-0-160-103:2068012:2068320 [3] NCCL INFO Connected all trees
+ip-26-0-160-103:2068012:2068320 [3] NCCL INFO 32 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-160-103:2068009:2068326 [0] NCCL INFO Connected all rings
+ip-26-0-160-103:2068009:2068326 [0] NCCL INFO Connected all trees
+ip-26-0-160-103:2068009:2068326 [0] NCCL INFO 32 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-160-103:2068016:2068314 [7] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068016:2068314 [7] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068016:2068314 [7] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068016:2068314 [7] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068016:2068314 [7] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068016:2068314 [7] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068016:2068314 [7] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068016:2068314 [7] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068016:2068314 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-160-103:2068016:2068314 [7] NCCL INFO Channel 00/32 :    0
+ip-26-0-160-103:2068016:2068314 [7] NCCL INFO Channel 01/32 :    0
+ip-26-0-160-103:2068016:2068314 [7] NCCL INFO Channel 02/32 :    0
+ip-26-0-160-103:2068016:2068314 [7] NCCL INFO Channel 03/32 :    0
+ip-26-0-160-103:2068016:2068314 [7] NCCL INFO Channel 04/32 :    0
+ip-26-0-160-103:2068016:2068314 [7] NCCL INFO Channel 05/32 :    0
+ip-26-0-160-103:2068016:2068314 [7] NCCL INFO Channel 06/32 :    0
+ip-26-0-160-103:2068016:2068314 [7] NCCL INFO Channel 07/32 :    0
+ip-26-0-160-103:2068016:2068314 [7] NCCL INFO Channel 08/32 :    0
+ip-26-0-160-103:2068016:2068314 [7] NCCL INFO Channel 09/32 :    0
+ip-26-0-160-103:2068016:2068314 [7] NCCL INFO Channel 10/32 :    0
+ip-26-0-160-103:2068016:2068314 [7] NCCL INFO Channel 11/32 :    0
+ip-26-0-160-103:2068016:2068314 [7] NCCL INFO Channel 12/32 :    0
+ip-26-0-160-103:2068016:2068314 [7] NCCL INFO Channel 13/32 :    0
+ip-26-0-160-103:2068016:2068314 [7] NCCL INFO Channel 14/32 :    0
+ip-26-0-160-103:2068016:2068314 [7] NCCL INFO Channel 15/32 :    0
+ip-26-0-160-103:2068016:2068314 [7] NCCL INFO Channel 16/32 :    0
+ip-26-0-160-103:2068016:2068314 [7] NCCL INFO Channel 17/32 :    0
+ip-26-0-160-103:2068016:2068314 [7] NCCL INFO Channel 18/32 :    0
+ip-26-0-160-103:2068016:2068314 [7] NCCL INFO Channel 19/32 :    0
+ip-26-0-160-103:2068016:2068314 [7] NCCL INFO Channel 20/32 :    0
+ip-26-0-160-103:2068016:2068314 [7] NCCL INFO Channel 21/32 :    0
+ip-26-0-160-103:2068016:2068314 [7] NCCL INFO Channel 22/32 :    0
+ip-26-0-160-103:2068016:2068314 [7] NCCL INFO Channel 23/32 :    0
+ip-26-0-160-103:2068016:2068314 [7] NCCL INFO Channel 24/32 :    0
+ip-26-0-160-103:2068016:2068314 [7] NCCL INFO Channel 25/32 :    0
+ip-26-0-160-103:2068016:2068314 [7] NCCL INFO Channel 26/32 :    0
+ip-26-0-160-103:2068016:2068314 [7] NCCL INFO Channel 27/32 :    0
+ip-26-0-160-103:2068016:2068314 [7] NCCL INFO Channel 28/32 :    0
+ip-26-0-160-103:2068016:2068314 [7] NCCL INFO Channel 29/32 :    0
+ip-26-0-160-103:2068016:2068314 [7] NCCL INFO Channel 30/32 :    0
+ip-26-0-160-103:2068016:2068314 [7] NCCL INFO Channel 31/32 :    0
+ip-26-0-160-103:2068016:2068314 [7] NCCL INFO Trees [0] -1/-1/-1->0->-1 [1] -1/-1/-1->0->-1 [2] -1/-1/-1->0->-1 [3] -1/-1/-1->0->-1 [4] -1/-1/-1->0->-1 [5] -1/-1/-1->0->-1 [6] -1/-1/-1->0->-1 [7] -1/-1/-1->0->-1 [8] -1/-1/-1->0->-1 [9] -1/-1/-1->0->-1 [10] -1/-1/-1->0->-1 [11] -1/-1/-1->0->-1 [12] -1/-1/-1->0->-1 [13] -1/-1/-1->0->-1 [14] -1/-1/-1->0->-1 [15] -1/-1/-1->0->-1 [16] -1/-1/-1->0->-1 [17] -1/-1/-1->0->-1 [18] -1/-1/-1->0->-1 [19] -1/-1/-1->0->-1 [20] -1/-1/-1->0->-1 [21] -1/-1/-1->0->-1 [22] -1/-1/-1->0->-1 [23] -1/-1/-1->0->-1 [24] -1/-1/-1->0->-1 [25] -1/-1/-1->0->-1 [26] -1/-1/-1->0->-1 [27] -1/-1/-1->0->-1 [28] -1/-1/-1->0->-1 [29] -1/-1/-1->0->-1 [30] -1/-1/-1->0->-1 [31] -1/-1/-1->0->-1
+ip-26-0-160-103:2068016:2068314 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-103:2068016:2068314 [7] NCCL INFO Connected all rings
+ip-26-0-160-103:2068016:2068314 [7] NCCL INFO Connected all trees
+ip-26-0-160-103:2068016:2068314 [7] NCCL INFO 32 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-160-103:2068015:2068316 [6] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068015:2068316 [6] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068015:2068316 [6] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068015:2068316 [6] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068015:2068316 [6] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068015:2068316 [6] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068015:2068316 [6] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068015:2068316 [6] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068015:2068316 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-160-103:2068015:2068316 [6] NCCL INFO Channel 00/32 :    0
+ip-26-0-160-103:2068015:2068316 [6] NCCL INFO Channel 01/32 :    0
+ip-26-0-160-103:2068015:2068316 [6] NCCL INFO Channel 02/32 :    0
+ip-26-0-160-103:2068015:2068316 [6] NCCL INFO Channel 03/32 :    0
+ip-26-0-160-103:2068015:2068316 [6] NCCL INFO Channel 04/32 :    0
+ip-26-0-160-103:2068015:2068316 [6] NCCL INFO Channel 05/32 :    0
+ip-26-0-160-103:2068015:2068316 [6] NCCL INFO Channel 06/32 :    0
+ip-26-0-160-103:2068015:2068316 [6] NCCL INFO Channel 07/32 :    0
+ip-26-0-160-103:2068015:2068316 [6] NCCL INFO Channel 08/32 :    0
+ip-26-0-160-103:2068015:2068316 [6] NCCL INFO Channel 09/32 :    0
+ip-26-0-160-103:2068015:2068316 [6] NCCL INFO Channel 10/32 :    0
+ip-26-0-160-103:2068015:2068316 [6] NCCL INFO Channel 11/32 :    0
+ip-26-0-160-103:2068015:2068316 [6] NCCL INFO Channel 12/32 :    0
+ip-26-0-160-103:2068015:2068316 [6] NCCL INFO Channel 13/32 :    0
+ip-26-0-160-103:2068015:2068316 [6] NCCL INFO Channel 14/32 :    0
+ip-26-0-160-103:2068015:2068316 [6] NCCL INFO Channel 15/32 :    0
+ip-26-0-160-103:2068015:2068316 [6] NCCL INFO Channel 16/32 :    0
+ip-26-0-160-103:2068015:2068316 [6] NCCL INFO Channel 17/32 :    0
+ip-26-0-160-103:2068015:2068316 [6] NCCL INFO Channel 18/32 :    0
+ip-26-0-160-103:2068015:2068316 [6] NCCL INFO Channel 19/32 :    0
+ip-26-0-160-103:2068015:2068316 [6] NCCL INFO Channel 20/32 :    0
+ip-26-0-160-103:2068015:2068316 [6] NCCL INFO Channel 21/32 :    0
+ip-26-0-160-103:2068015:2068316 [6] NCCL INFO Channel 22/32 :    0
+ip-26-0-160-103:2068015:2068316 [6] NCCL INFO Channel 23/32 :    0
+ip-26-0-160-103:2068015:2068316 [6] NCCL INFO Channel 24/32 :    0
+ip-26-0-160-103:2068015:2068316 [6] NCCL INFO Channel 25/32 :    0
+ip-26-0-160-103:2068015:2068316 [6] NCCL INFO Channel 26/32 :    0
+ip-26-0-160-103:2068015:2068316 [6] NCCL INFO Channel 27/32 :    0
+ip-26-0-160-103:2068015:2068316 [6] NCCL INFO Channel 28/32 :    0
+ip-26-0-160-103:2068015:2068316 [6] NCCL INFO Channel 29/32 :    0
+ip-26-0-160-103:2068015:2068316 [6] NCCL INFO Channel 30/32 :    0
+ip-26-0-160-103:2068015:2068316 [6] NCCL INFO Channel 31/32 :    0
+ip-26-0-160-103:2068015:2068316 [6] NCCL INFO Trees [0] -1/-1/-1->0->-1 [1] -1/-1/-1->0->-1 [2] -1/-1/-1->0->-1 [3] -1/-1/-1->0->-1 [4] -1/-1/-1->0->-1 [5] -1/-1/-1->0->-1 [6] -1/-1/-1->0->-1 [7] -1/-1/-1->0->-1 [8] -1/-1/-1->0->-1 [9] -1/-1/-1->0->-1 [10] -1/-1/-1->0->-1 [11] -1/-1/-1->0->-1 [12] -1/-1/-1->0->-1 [13] -1/-1/-1->0->-1 [14] -1/-1/-1->0->-1 [15] -1/-1/-1->0->-1 [16] -1/-1/-1->0->-1 [17] -1/-1/-1->0->-1 [18] -1/-1/-1->0->-1 [19] -1/-1/-1->0->-1 [20] -1/-1/-1->0->-1 [21] -1/-1/-1->0->-1 [22] -1/-1/-1->0->-1 [23] -1/-1/-1->0->-1 [24] -1/-1/-1->0->-1 [25] -1/-1/-1->0->-1 [26] -1/-1/-1->0->-1 [27] -1/-1/-1->0->-1 [28] -1/-1/-1->0->-1 [29] -1/-1/-1->0->-1 [30] -1/-1/-1->0->-1 [31] -1/-1/-1->0->-1
+ip-26-0-160-103:2068015:2068316 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-103:2068015:2068316 [6] NCCL INFO Connected all rings
+ip-26-0-160-103:2068015:2068316 [6] NCCL INFO Connected all trees
+ip-26-0-160-103:2068015:2068316 [6] NCCL INFO 32 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-160-103:2068009:2068326 [0] NCCL INFO comm 0xb6c6c80 rank 0 nranks 1 cudaDev 0 nvmlDev 0 busId 53000 commId 0x47ad8297bfe6e876 - Init COMPLETE
+ip-26-0-160-103:2068012:2068320 [3] NCCL INFO comm 0xc13a440 rank 0 nranks 1 cudaDev 3 nvmlDev 3 busId 86000 commId 0xc2cbd04264fa9e0c - Init COMPLETE
+ip-26-0-160-103:2068015:2068316 [6] NCCL INFO comm 0xb104d70 rank 0 nranks 1 cudaDev 6 nvmlDev 6 busId b9000 commId 0xbe3e3bb987a9de04 - Init COMPLETE
+ip-26-0-160-103:2068016:2068314 [7] NCCL INFO comm 0xb4557a0 rank 0 nranks 1 cudaDev 7 nvmlDev 7 busId ca000 commId 0xf95d4c4d4f14b109 - Init COMPLETE
+ip-26-0-160-103:2068010:2068324 [1] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068010:2068324 [1] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068010:2068324 [1] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068010:2068324 [1] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068010:2068324 [1] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068010:2068324 [1] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068010:2068324 [1] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068010:2068324 [1] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068010:2068324 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-160-103:2068010:2068324 [1] NCCL INFO Channel 00/32 :    0
+ip-26-0-160-103:2068010:2068324 [1] NCCL INFO Channel 01/32 :    0
+ip-26-0-160-103:2068010:2068324 [1] NCCL INFO Channel 02/32 :    0
+ip-26-0-160-103:2068010:2068324 [1] NCCL INFO Channel 03/32 :    0
+ip-26-0-160-103:2068010:2068324 [1] NCCL INFO Channel 04/32 :    0
+ip-26-0-160-103:2068010:2068324 [1] NCCL INFO Channel 05/32 :    0
+ip-26-0-160-103:2068010:2068324 [1] NCCL INFO Channel 06/32 :    0
+ip-26-0-160-103:2068010:2068324 [1] NCCL INFO Channel 07/32 :    0
+ip-26-0-160-103:2068010:2068324 [1] NCCL INFO Channel 08/32 :    0
+ip-26-0-160-103:2068010:2068324 [1] NCCL INFO Channel 09/32 :    0
+ip-26-0-160-103:2068010:2068324 [1] NCCL INFO Channel 10/32 :    0
+ip-26-0-160-103:2068010:2068324 [1] NCCL INFO Channel 11/32 :    0
+ip-26-0-160-103:2068010:2068324 [1] NCCL INFO Channel 12/32 :    0
+ip-26-0-160-103:2068010:2068324 [1] NCCL INFO Channel 13/32 :    0
+ip-26-0-160-103:2068010:2068324 [1] NCCL INFO Channel 14/32 :    0
+ip-26-0-160-103:2068010:2068324 [1] NCCL INFO Channel 15/32 :    0
+ip-26-0-160-103:2068010:2068324 [1] NCCL INFO Channel 16/32 :    0
+ip-26-0-160-103:2068010:2068324 [1] NCCL INFO Channel 17/32 :    0
+ip-26-0-160-103:2068010:2068324 [1] NCCL INFO Channel 18/32 :    0
+ip-26-0-160-103:2068010:2068324 [1] NCCL INFO Channel 19/32 :    0
+ip-26-0-160-103:2068010:2068324 [1] NCCL INFO Channel 20/32 :    0
+ip-26-0-160-103:2068010:2068324 [1] NCCL INFO Channel 21/32 :    0
+ip-26-0-160-103:2068010:2068324 [1] NCCL INFO Channel 22/32 :    0
+ip-26-0-160-103:2068010:2068324 [1] NCCL INFO Channel 23/32 :    0
+ip-26-0-160-103:2068010:2068324 [1] NCCL INFO Channel 24/32 :    0
+ip-26-0-160-103:2068010:2068324 [1] NCCL INFO Channel 25/32 :    0
+ip-26-0-160-103:2068010:2068324 [1] NCCL INFO Channel 26/32 :    0
+ip-26-0-160-103:2068010:2068324 [1] NCCL INFO Channel 27/32 :    0
+ip-26-0-160-103:2068010:2068324 [1] NCCL INFO Channel 28/32 :    0
+ip-26-0-160-103:2068010:2068324 [1] NCCL INFO Channel 29/32 :    0
+ip-26-0-160-103:2068010:2068324 [1] NCCL INFO Channel 30/32 :    0
+ip-26-0-160-103:2068010:2068324 [1] NCCL INFO Channel 31/32 :    0
+ip-26-0-160-103:2068010:2068324 [1] NCCL INFO Trees [0] -1/-1/-1->0->-1 [1] -1/-1/-1->0->-1 [2] -1/-1/-1->0->-1 [3] -1/-1/-1->0->-1 [4] -1/-1/-1->0->-1 [5] -1/-1/-1->0->-1 [6] -1/-1/-1->0->-1 [7] -1/-1/-1->0->-1 [8] -1/-1/-1->0->-1 [9] -1/-1/-1->0->-1 [10] -1/-1/-1->0->-1 [11] -1/-1/-1->0->-1 [12] -1/-1/-1->0->-1 [13] -1/-1/-1->0->-1 [14] -1/-1/-1->0->-1 [15] -1/-1/-1->0->-1 [16] -1/-1/-1->0->-1 [17] -1/-1/-1->0->-1 [18] -1/-1/-1->0->-1 [19] -1/-1/-1->0->-1 [20] -1/-1/-1->0->-1 [21] -1/-1/-1->0->-1 [22] -1/-1/-1->0->-1 [23] -1/-1/-1->0->-1 [24] -1/-1/-1->0->-1 [25] -1/-1/-1->0->-1 [26] -1/-1/-1->0->-1 [27] -1/-1/-1->0->-1 [28] -1/-1/-1->0->-1 [29] -1/-1/-1->0->-1 [30] -1/-1/-1->0->-1 [31] -1/-1/-1->0->-1
+ip-26-0-160-103:2068010:2068324 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-103:2068011:2068322 [2] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068011:2068322 [2] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068011:2068322 [2] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068011:2068322 [2] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068011:2068322 [2] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068011:2068322 [2] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068011:2068322 [2] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068011:2068322 [2] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068010:2068324 [1] NCCL INFO Connected all rings
+ip-26-0-160-103:2068010:2068324 [1] NCCL INFO Connected all trees
+ip-26-0-160-103:2068010:2068324 [1] NCCL INFO 32 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-160-103:2068011:2068322 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-160-103:2068011:2068322 [2] NCCL INFO Channel 00/32 :    0
+ip-26-0-160-103:2068011:2068322 [2] NCCL INFO Channel 01/32 :    0
+ip-26-0-160-103:2068011:2068322 [2] NCCL INFO Channel 02/32 :    0
+ip-26-0-160-103:2068011:2068322 [2] NCCL INFO Channel 03/32 :    0
+ip-26-0-160-103:2068011:2068322 [2] NCCL INFO Channel 04/32 :    0
+ip-26-0-160-103:2068011:2068322 [2] NCCL INFO Channel 05/32 :    0
+ip-26-0-160-103:2068011:2068322 [2] NCCL INFO Channel 06/32 :    0
+ip-26-0-160-103:2068011:2068322 [2] NCCL INFO Channel 07/32 :    0
+ip-26-0-160-103:2068011:2068322 [2] NCCL INFO Channel 08/32 :    0
+ip-26-0-160-103:2068011:2068322 [2] NCCL INFO Channel 09/32 :    0
+ip-26-0-160-103:2068011:2068322 [2] NCCL INFO Channel 10/32 :    0
+ip-26-0-160-103:2068011:2068322 [2] NCCL INFO Channel 11/32 :    0
+ip-26-0-160-103:2068011:2068322 [2] NCCL INFO Channel 12/32 :    0
+ip-26-0-160-103:2068011:2068322 [2] NCCL INFO Channel 13/32 :    0
+ip-26-0-160-103:2068011:2068322 [2] NCCL INFO Channel 14/32 :    0
+ip-26-0-160-103:2068011:2068322 [2] NCCL INFO Channel 15/32 :    0
+ip-26-0-160-103:2068011:2068322 [2] NCCL INFO Channel 16/32 :    0
+ip-26-0-160-103:2068011:2068322 [2] NCCL INFO Channel 17/32 :    0
+ip-26-0-160-103:2068011:2068322 [2] NCCL INFO Channel 18/32 :    0
+ip-26-0-160-103:2068011:2068322 [2] NCCL INFO Channel 19/32 :    0
+ip-26-0-160-103:2068011:2068322 [2] NCCL INFO Channel 20/32 :    0
+ip-26-0-160-103:2068011:2068322 [2] NCCL INFO Channel 21/32 :    0
+ip-26-0-160-103:2068011:2068322 [2] NCCL INFO Channel 22/32 :    0
+ip-26-0-160-103:2068011:2068322 [2] NCCL INFO Channel 23/32 :    0
+ip-26-0-160-103:2068011:2068322 [2] NCCL INFO Channel 24/32 :    0
+ip-26-0-160-103:2068011:2068322 [2] NCCL INFO Channel 25/32 :    0
+ip-26-0-160-103:2068011:2068322 [2] NCCL INFO Channel 26/32 :    0
+ip-26-0-160-103:2068011:2068322 [2] NCCL INFO Channel 27/32 :    0
+ip-26-0-160-103:2068011:2068322 [2] NCCL INFO Channel 28/32 :    0
+ip-26-0-160-103:2068011:2068322 [2] NCCL INFO Channel 29/32 :    0
+ip-26-0-160-103:2068011:2068322 [2] NCCL INFO Channel 30/32 :    0
+ip-26-0-160-103:2068011:2068322 [2] NCCL INFO Channel 31/32 :    0
+ip-26-0-160-103:2068011:2068322 [2] NCCL INFO Trees [0] -1/-1/-1->0->-1 [1] -1/-1/-1->0->-1 [2] -1/-1/-1->0->-1 [3] -1/-1/-1->0->-1 [4] -1/-1/-1->0->-1 [5] -1/-1/-1->0->-1 [6] -1/-1/-1->0->-1 [7] -1/-1/-1->0->-1 [8] -1/-1/-1->0->-1 [9] -1/-1/-1->0->-1 [10] -1/-1/-1->0->-1 [11] -1/-1/-1->0->-1 [12] -1/-1/-1->0->-1 [13] -1/-1/-1->0->-1 [14] -1/-1/-1->0->-1 [15] -1/-1/-1->0->-1 [16] -1/-1/-1->0->-1 [17] -1/-1/-1->0->-1 [18] -1/-1/-1->0->-1 [19] -1/-1/-1->0->-1 [20] -1/-1/-1->0->-1 [21] -1/-1/-1->0->-1 [22] -1/-1/-1->0->-1 [23] -1/-1/-1->0->-1 [24] -1/-1/-1->0->-1 [25] -1/-1/-1->0->-1 [26] -1/-1/-1->0->-1 [27] -1/-1/-1->0->-1 [28] -1/-1/-1->0->-1 [29] -1/-1/-1->0->-1 [30] -1/-1/-1->0->-1 [31] -1/-1/-1->0->-1
+ip-26-0-160-103:2068011:2068322 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-103:2068011:2068322 [2] NCCL INFO Connected all rings
+ip-26-0-160-103:2068011:2068322 [2] NCCL INFO Connected all trees
+ip-26-0-160-103:2068011:2068322 [2] NCCL INFO 32 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-160-103:2068010:2068324 [1] NCCL INFO comm 0xc03cea0 rank 0 nranks 1 cudaDev 1 nvmlDev 1 busId 64000 commId 0x3a509b9a635d2b6b - Init COMPLETE
+12/12/2024 22:02:48 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]: Total number of parameters: 1.04G (1984.26MiB)
+12/12/2024 22:02:48 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]: Local number of parameters: 520M (992.13MiB)
+12/12/2024 22:02:48 [INFO|DP=0|PP=0|TP=1|ip-26-0-160-103]: Local number of parameters: 520M (992.13MiB)
+12/12/2024 22:02:48 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]: [After model building] Memory usage: 992.15MiB. Peak allocated: 5632.00MiB Peak reserved: 17922.00MiB
+12/12/2024 22:02:48 [INFO|DP=0|PP=0|TP=1|ip-26-0-160-103]: [After model building] Memory usage: 992.15MiB. Peak allocated: 5632.00MiB Peak reserved: 16898.00MiB
+12/12/2024 22:02:48 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]: No checkpoint path provided.
+12/12/2024 22:02:48 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]: Parametrizing model parameters using StandardParametrizator
+ip-26-0-160-103:2068011:2068322 [2] NCCL INFO comm 0xbe1a250 rank 0 nranks 1 cudaDev 2 nvmlDev 2 busId 75000 commId 0x59b3bcc53f35f7e9 - Init COMPLETE
+ip-26-0-160-103:2068010:2068349 [1] NCCL INFO Using network AWS Libfabric
+ip-26-0-160-103:2068016:2068351 [7] NCCL INFO Using network AWS Libfabric
+ip-26-0-160-103:2068009:2068348 [0] NCCL INFO Using network AWS Libfabric
+ip-26-0-160-103:2068015:2068350 [6] NCCL INFO Using network AWS Libfabric
+ip-26-0-160-103:2068014:2068332 [5] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068014:2068332 [5] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068014:2068332 [5] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068014:2068332 [5] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068014:2068332 [5] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068014:2068332 [5] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068014:2068332 [5] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068014:2068332 [5] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068011:2068352 [2] NCCL INFO Using network AWS Libfabric
+ip-26-0-160-103:2068014:2068332 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-160-103:2068014:2068332 [5] NCCL INFO Channel 00/32 :    0
+ip-26-0-160-103:2068014:2068332 [5] NCCL INFO Channel 01/32 :    0
+ip-26-0-160-103:2068014:2068332 [5] NCCL INFO Channel 02/32 :    0
+ip-26-0-160-103:2068014:2068332 [5] NCCL INFO Channel 03/32 :    0
+ip-26-0-160-103:2068014:2068332 [5] NCCL INFO Channel 04/32 :    0
+ip-26-0-160-103:2068014:2068332 [5] NCCL INFO Channel 05/32 :    0
+ip-26-0-160-103:2068014:2068332 [5] NCCL INFO Channel 06/32 :    0
+ip-26-0-160-103:2068014:2068332 [5] NCCL INFO Channel 07/32 :    0
+ip-26-0-160-103:2068014:2068332 [5] NCCL INFO Channel 08/32 :    0
+ip-26-0-160-103:2068014:2068332 [5] NCCL INFO Channel 09/32 :    0
+ip-26-0-160-103:2068014:2068332 [5] NCCL INFO Channel 10/32 :    0
+ip-26-0-160-103:2068014:2068332 [5] NCCL INFO Channel 11/32 :    0
+ip-26-0-160-103:2068014:2068332 [5] NCCL INFO Channel 12/32 :    0
+ip-26-0-160-103:2068014:2068332 [5] NCCL INFO Channel 13/32 :    0
+ip-26-0-160-103:2068014:2068332 [5] NCCL INFO Channel 14/32 :    0
+ip-26-0-160-103:2068014:2068332 [5] NCCL INFO Channel 15/32 :    0
+ip-26-0-160-103:2068014:2068332 [5] NCCL INFO Channel 16/32 :    0
+ip-26-0-160-103:2068014:2068332 [5] NCCL INFO Channel 17/32 :    0
+ip-26-0-160-103:2068014:2068332 [5] NCCL INFO Channel 18/32 :    0
+ip-26-0-160-103:2068014:2068332 [5] NCCL INFO Channel 19/32 :    0
+ip-26-0-160-103:2068014:2068332 [5] NCCL INFO Channel 20/32 :    0
+ip-26-0-160-103:2068014:2068332 [5] NCCL INFO Channel 21/32 :    0
+ip-26-0-160-103:2068014:2068332 [5] NCCL INFO Channel 22/32 :    0
+ip-26-0-160-103:2068014:2068332 [5] NCCL INFO Channel 23/32 :    0
+ip-26-0-160-103:2068014:2068332 [5] NCCL INFO Channel 24/32 :    0
+ip-26-0-160-103:2068014:2068332 [5] NCCL INFO Channel 25/32 :    0
+ip-26-0-160-103:2068014:2068332 [5] NCCL INFO Channel 26/32 :    0
+ip-26-0-160-103:2068014:2068332 [5] NCCL INFO Channel 27/32 :    0
+ip-26-0-160-103:2068014:2068332 [5] NCCL INFO Channel 28/32 :    0
+ip-26-0-160-103:2068014:2068332 [5] NCCL INFO Channel 29/32 :    0
+ip-26-0-160-103:2068014:2068332 [5] NCCL INFO Channel 30/32 :    0
+ip-26-0-160-103:2068014:2068332 [5] NCCL INFO Channel 31/32 :    0
+ip-26-0-160-103:2068014:2068332 [5] NCCL INFO Trees [0] -1/-1/-1->0->-1 [1] -1/-1/-1->0->-1 [2] -1/-1/-1->0->-1 [3] -1/-1/-1->0->-1 [4] -1/-1/-1->0->-1 [5] -1/-1/-1->0->-1 [6] -1/-1/-1->0->-1 [7] -1/-1/-1->0->-1 [8] -1/-1/-1->0->-1 [9] -1/-1/-1->0->-1 [10] -1/-1/-1->0->-1 [11] -1/-1/-1->0->-1 [12] -1/-1/-1->0->-1 [13] -1/-1/-1->0->-1 [14] -1/-1/-1->0->-1 [15] -1/-1/-1->0->-1 [16] -1/-1/-1->0->-1 [17] -1/-1/-1->0->-1 [18] -1/-1/-1->0->-1 [19] -1/-1/-1->0->-1 [20] -1/-1/-1->0->-1 [21] -1/-1/-1->0->-1 [22] -1/-1/-1->0->-1 [23] -1/-1/-1->0->-1 [24] -1/-1/-1->0->-1 [25] -1/-1/-1->0->-1 [26] -1/-1/-1->0->-1 [27] -1/-1/-1->0->-1 [28] -1/-1/-1->0->-1 [29] -1/-1/-1->0->-1 [30] -1/-1/-1->0->-1 [31] -1/-1/-1->0->-1
+ip-26-0-160-103:2068014:2068332 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-103:2068012:2068353 [3] NCCL INFO Using network AWS Libfabric
+ip-26-0-160-103:2068014:2068332 [5] NCCL INFO Connected all rings
+ip-26-0-160-103:2068014:2068332 [5] NCCL INFO Connected all trees
+ip-26-0-160-103:2068014:2068332 [5] NCCL INFO 32 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-160-103:2068013:2068330 [4] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068013:2068330 [4] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068013:2068330 [4] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068013:2068330 [4] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068013:2068330 [4] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068013:2068330 [4] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068013:2068330 [4] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068013:2068330 [4] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068013:2068330 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-160-103:2068013:2068330 [4] NCCL INFO Channel 00/32 :    0
+ip-26-0-160-103:2068013:2068330 [4] NCCL INFO Channel 01/32 :    0
+ip-26-0-160-103:2068013:2068330 [4] NCCL INFO Channel 02/32 :    0
+ip-26-0-160-103:2068013:2068330 [4] NCCL INFO Channel 03/32 :    0
+ip-26-0-160-103:2068013:2068330 [4] NCCL INFO Channel 04/32 :    0
+ip-26-0-160-103:2068013:2068330 [4] NCCL INFO Channel 05/32 :    0
+ip-26-0-160-103:2068013:2068330 [4] NCCL INFO Channel 06/32 :    0
+ip-26-0-160-103:2068013:2068330 [4] NCCL INFO Channel 07/32 :    0
+ip-26-0-160-103:2068013:2068330 [4] NCCL INFO Channel 08/32 :    0
+ip-26-0-160-103:2068013:2068330 [4] NCCL INFO Channel 09/32 :    0
+ip-26-0-160-103:2068013:2068330 [4] NCCL INFO Channel 10/32 :    0
+ip-26-0-160-103:2068013:2068330 [4] NCCL INFO Channel 11/32 :    0
+ip-26-0-160-103:2068013:2068330 [4] NCCL INFO Channel 12/32 :    0
+ip-26-0-160-103:2068013:2068330 [4] NCCL INFO Channel 13/32 :    0
+ip-26-0-160-103:2068013:2068330 [4] NCCL INFO Channel 14/32 :    0
+ip-26-0-160-103:2068013:2068330 [4] NCCL INFO Channel 15/32 :    0
+ip-26-0-160-103:2068013:2068330 [4] NCCL INFO Channel 16/32 :    0
+ip-26-0-160-103:2068013:2068330 [4] NCCL INFO Channel 17/32 :    0
+ip-26-0-160-103:2068013:2068330 [4] NCCL INFO Channel 18/32 :    0
+ip-26-0-160-103:2068013:2068330 [4] NCCL INFO Channel 19/32 :    0
+ip-26-0-160-103:2068013:2068330 [4] NCCL INFO Channel 20/32 :    0
+ip-26-0-160-103:2068013:2068330 [4] NCCL INFO Channel 21/32 :    0
+ip-26-0-160-103:2068013:2068330 [4] NCCL INFO Channel 22/32 :    0
+ip-26-0-160-103:2068013:2068330 [4] NCCL INFO Channel 23/32 :    0
+ip-26-0-160-103:2068013:2068330 [4] NCCL INFO Channel 24/32 :    0
+ip-26-0-160-103:2068013:2068330 [4] NCCL INFO Channel 25/32 :    0
+ip-26-0-160-103:2068013:2068330 [4] NCCL INFO Channel 26/32 :    0
+ip-26-0-160-103:2068013:2068330 [4] NCCL INFO Channel 27/32 :    0
+ip-26-0-160-103:2068013:2068330 [4] NCCL INFO Channel 28/32 :    0
+ip-26-0-160-103:2068013:2068330 [4] NCCL INFO Channel 29/32 :    0
+ip-26-0-160-103:2068013:2068330 [4] NCCL INFO Channel 30/32 :    0
+ip-26-0-160-103:2068013:2068330 [4] NCCL INFO Channel 31/32 :    0
+ip-26-0-160-103:2068013:2068330 [4] NCCL INFO Trees [0] -1/-1/-1->0->-1 [1] -1/-1/-1->0->-1 [2] -1/-1/-1->0->-1 [3] -1/-1/-1->0->-1 [4] -1/-1/-1->0->-1 [5] -1/-1/-1->0->-1 [6] -1/-1/-1->0->-1 [7] -1/-1/-1->0->-1 [8] -1/-1/-1->0->-1 [9] -1/-1/-1->0->-1 [10] -1/-1/-1->0->-1 [11] -1/-1/-1->0->-1 [12] -1/-1/-1->0->-1 [13] -1/-1/-1->0->-1 [14] -1/-1/-1->0->-1 [15] -1/-1/-1->0->-1 [16] -1/-1/-1->0->-1 [17] -1/-1/-1->0->-1 [18] -1/-1/-1->0->-1 [19] -1/-1/-1->0->-1 [20] -1/-1/-1->0->-1 [21] -1/-1/-1->0->-1 [22] -1/-1/-1->0->-1 [23] -1/-1/-1->0->-1 [24] -1/-1/-1->0->-1 [25] -1/-1/-1->0->-1 [26] -1/-1/-1->0->-1 [27] -1/-1/-1->0->-1 [28] -1/-1/-1->0->-1 [29] -1/-1/-1->0->-1 [30] -1/-1/-1->0->-1 [31] -1/-1/-1->0->-1
+ip-26-0-160-103:2068013:2068330 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-103:2068013:2068330 [4] NCCL INFO Connected all rings
+ip-26-0-160-103:2068013:2068330 [4] NCCL INFO Connected all trees
+ip-26-0-160-103:2068013:2068330 [4] NCCL INFO 32 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-160-103:2068014:2068332 [5] NCCL INFO comm 0xb408f20 rank 0 nranks 1 cudaDev 5 nvmlDev 5 busId a8000 commId 0xf554a31af9bccf28 - Init COMPLETE
+ip-26-0-160-103:2068013:2068330 [4] NCCL INFO comm 0xbb27720 rank 0 nranks 1 cudaDev 4 nvmlDev 4 busId 97000 commId 0xc0dc7d61e0e5d53e - Init COMPLETE
+ip-26-0-160-103:2068014:2068359 [5] NCCL INFO Using network AWS Libfabric
+ip-26-0-160-103:2068014:2068359 [5] NCCL INFO comm 0xb40dcb0 rank 2 nranks 4 cudaDev 5 nvmlDev 5 busId a8000 commId 0x2e5e3292a10478a0 - Init START
+ip-26-0-160-103:2068016:2068351 [7] NCCL INFO comm 0xb45a580 rank 3 nranks 4 cudaDev 7 nvmlDev 7 busId ca000 commId 0x2e5e3292a10478a0 - Init START
+ip-26-0-160-103:2068010:2068349 [1] NCCL INFO comm 0xc042060 rank 0 nranks 4 cudaDev 1 nvmlDev 1 busId 64000 commId 0x2e5e3292a10478a0 - Init START
+ip-26-0-160-103:2068012:2068353 [3] NCCL INFO comm 0xc13fae0 rank 1 nranks 4 cudaDev 3 nvmlDev 3 busId 86000 commId 0x2e5e3292a10478a0 - Init START
+ip-26-0-160-103:2068012:2068353 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /opt/aws-ofi-nccl/share/aws-ofi-nccl/xml/p5.48xl-topo.xml
+ip-26-0-160-103:2068014:2068359 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /opt/aws-ofi-nccl/share/aws-ofi-nccl/xml/p5.48xl-topo.xml
+ip-26-0-160-103:2068016:2068351 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /opt/aws-ofi-nccl/share/aws-ofi-nccl/xml/p5.48xl-topo.xml
+ip-26-0-160-103:2068010:2068349 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /opt/aws-ofi-nccl/share/aws-ofi-nccl/xml/p5.48xl-topo.xml
+ip-26-0-160-103:2068013:2068360 [4] NCCL INFO Using network AWS Libfabric
+ip-26-0-160-103:2068009:2068348 [0] NCCL INFO comm 0xb6cbb80 rank 0 nranks 4 cudaDev 0 nvmlDev 0 busId 53000 commId 0x91ce3d4cd722004e - Init START
+ip-26-0-160-103:2068013:2068360 [4] NCCL INFO comm 0xbb2c450 rank 2 nranks 4 cudaDev 4 nvmlDev 4 busId 97000 commId 0x91ce3d4cd722004e - Init START
+ip-26-0-160-103:2068011:2068352 [2] NCCL INFO comm 0xbe1f620 rank 1 nranks 4 cudaDev 2 nvmlDev 2 busId 75000 commId 0x91ce3d4cd722004e - Init START
+ip-26-0-160-103:2068015:2068350 [6] NCCL INFO comm 0xb109a50 rank 3 nranks 4 cudaDev 6 nvmlDev 6 busId b9000 commId 0x91ce3d4cd722004e - Init START
+ip-26-0-160-103:2068011:2068352 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /opt/aws-ofi-nccl/share/aws-ofi-nccl/xml/p5.48xl-topo.xml
+ip-26-0-160-103:2068015:2068350 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /opt/aws-ofi-nccl/share/aws-ofi-nccl/xml/p5.48xl-topo.xml
+ip-26-0-160-103:2068013:2068360 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /opt/aws-ofi-nccl/share/aws-ofi-nccl/xml/p5.48xl-topo.xml
+ip-26-0-160-103:2068009:2068348 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /opt/aws-ofi-nccl/share/aws-ofi-nccl/xml/p5.48xl-topo.xml
+ip-26-0-160-103:2068009:2068348 [0] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068009:2068348 [0] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068009:2068348 [0] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068009:2068348 [0] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068009:2068348 [0] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068009:2068348 [0] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068009:2068348 [0] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068009:2068348 [0] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068012:2068353 [3] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068012:2068353 [3] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068012:2068353 [3] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068009:2068348 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-160-103:2068009:2068348 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-160-103:2068012:2068353 [3] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068012:2068353 [3] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068012:2068353 [3] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068012:2068353 [3] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068012:2068353 [3] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068012:2068353 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-160-103:2068012:2068353 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-160-103:2068015:2068350 [6] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068015:2068350 [6] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068015:2068350 [6] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068015:2068350 [6] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068015:2068350 [6] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068015:2068350 [6] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068015:2068350 [6] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068015:2068350 [6] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068015:2068350 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-160-103:2068015:2068350 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-160-103:2068014:2068359 [5] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068014:2068359 [5] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068014:2068359 [5] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068014:2068359 [5] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068014:2068359 [5] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068014:2068359 [5] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068014:2068359 [5] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068014:2068359 [5] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068014:2068359 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-160-103:2068014:2068359 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-160-103:2068010:2068349 [1] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068010:2068349 [1] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068010:2068349 [1] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068010:2068349 [1] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068010:2068349 [1] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068010:2068349 [1] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068010:2068349 [1] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068010:2068349 [1] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068010:2068349 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-160-103:2068010:2068349 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-160-103:2068011:2068352 [2] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068011:2068352 [2] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068011:2068352 [2] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068011:2068352 [2] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068011:2068352 [2] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068011:2068352 [2] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068011:2068352 [2] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068011:2068352 [2] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068011:2068352 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-160-103:2068011:2068352 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-160-103:2068013:2068360 [4] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068013:2068360 [4] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068013:2068360 [4] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068013:2068360 [4] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068013:2068360 [4] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068013:2068360 [4] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068013:2068360 [4] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068013:2068360 [4] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068013:2068360 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-160-103:2068013:2068360 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-160-103:2068016:2068351 [7] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068016:2068351 [7] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068016:2068351 [7] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068016:2068351 [7] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068016:2068351 [7] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068016:2068351 [7] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068016:2068351 [7] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068016:2068351 [7] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2068016:2068351 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-160-103:2068016:2068351 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-160-103:2068013:2068360 [4] NCCL INFO Trees [0] 3/-1/-1->2->1 [1] 3/-1/-1->2->1 [2] 3/-1/-1->2->1 [3] 3/-1/-1->2->1 [4] 3/-1/-1->2->1 [5] 3/-1/-1->2->1 [6] 3/-1/-1->2->1 [7] 3/-1/-1->2->1 [8] 3/-1/-1->2->1 [9] 3/-1/-1->2->1 [10] 3/-1/-1->2->1 [11] 3/-1/-1->2->1 [12] 3/-1/-1->2->1 [13] 3/-1/-1->2->1 [14] 3/-1/-1->2->1 [15] 3/-1/-1->2->1 [16] 3/-1/-1->2->1 [17] 3/-1/-1->2->1 [18] 3/-1/-1->2->1 [19] 3/-1/-1->2->1 [20] 3/-1/-1->2->1 [21] 3/-1/-1->2->1 [22] 3/-1/-1->2->1 [23] 3/-1/-1->2->1
+ip-26-0-160-103:2068013:2068360 [4] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-160-103:2068009:2068348 [0] NCCL INFO Channel 00/24 :    0   1   2   3
+ip-26-0-160-103:2068009:2068348 [0] NCCL INFO Channel 01/24 :    0   1   2   3
+ip-26-0-160-103:2068009:2068348 [0] NCCL INFO Channel 02/24 :    0   1   2   3
+ip-26-0-160-103:2068011:2068352 [2] NCCL INFO Trees [0] 2/-1/-1->1->0 [1] 2/-1/-1->1->0 [2] 2/-1/-1->1->0 [3] 2/-1/-1->1->0 [4] 2/-1/-1->1->0 [5] 2/-1/-1->1->0 [6] 2/-1/-1->1->0 [7] 2/-1/-1->1->0 [8] 2/-1/-1->1->0 [9] 2/-1/-1->1->0 [10] 2/-1/-1->1->0 [11] 2/-1/-1->1->0 [12] 2/-1/-1->1->0 [13] 2/-1/-1->1->0 [14] 2/-1/-1->1->0 [15] 2/-1/-1->1->0 [16] 2/-1/-1->1->0 [17] 2/-1/-1->1->0 [18] 2/-1/-1->1->0 [19] 2/-1/-1->1->0 [20] 2/-1/-1->1->0 [21] 2/-1/-1->1->0 [22] 2/-1/-1->1->0 [23] 2/-1/-1->1->0
+ip-26-0-160-103:2068009:2068348 [0] NCCL INFO Channel 03/24 :    0   1   2   3
+ip-26-0-160-103:2068009:2068348 [0] NCCL INFO Channel 04/24 :    0   1   2   3
+ip-26-0-160-103:2068011:2068352 [2] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-160-103:2068009:2068348 [0] NCCL INFO Channel 05/24 :    0   1   2   3
+ip-26-0-160-103:2068009:2068348 [0] NCCL INFO Channel 06/24 :    0   1   2   3
+ip-26-0-160-103:2068009:2068348 [0] NCCL INFO Channel 07/24 :    0   1   2   3
+ip-26-0-160-103:2068009:2068348 [0] NCCL INFO Channel 08/24 :    0   1   2   3
+ip-26-0-160-103:2068009:2068348 [0] NCCL INFO Channel 09/24 :    0   1   2   3
+ip-26-0-160-103:2068009:2068348 [0] NCCL INFO Channel 10/24 :    0   1   2   3
+ip-26-0-160-103:2068009:2068348 [0] NCCL INFO Channel 11/24 :    0   1   2   3
+ip-26-0-160-103:2068009:2068348 [0] NCCL INFO Channel 12/24 :    0   1   2   3
+ip-26-0-160-103:2068009:2068348 [0] NCCL INFO Channel 13/24 :    0   1   2   3
+ip-26-0-160-103:2068015:2068350 [6] NCCL INFO Trees [0] -1/-1/-1->3->2 [1] -1/-1/-1->3->2 [2] -1/-1/-1->3->2 [3] -1/-1/-1->3->2 [4] -1/-1/-1->3->2 [5] -1/-1/-1->3->2 [6] -1/-1/-1->3->2 [7] -1/-1/-1->3->2 [8] -1/-1/-1->3->2 [9] -1/-1/-1->3->2 [10] -1/-1/-1->3->2 [11] -1/-1/-1->3->2 [12] -1/-1/-1->3->2 [13] -1/-1/-1->3->2 [14] -1/-1/-1->3->2 [15] -1/-1/-1->3->2 [16] -1/-1/-1->3->2 [17] -1/-1/-1->3->2 [18] -1/-1/-1->3->2 [19] -1/-1/-1->3->2 [20] -1/-1/-1->3->2 [21] -1/-1/-1->3->2 [22] -1/-1/-1->3->2 [23] -1/-1/-1->3->2
+ip-26-0-160-103:2068009:2068348 [0] NCCL INFO Channel 14/24 :    0   1   2   3
+ip-26-0-160-103:2068009:2068348 [0] NCCL INFO Channel 15/24 :    0   1   2   3
+ip-26-0-160-103:2068015:2068350 [6] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-160-103:2068009:2068348 [0] NCCL INFO Channel 16/24 :    0   1   2   3
+ip-26-0-160-103:2068009:2068348 [0] NCCL INFO Channel 17/24 :    0   1   2   3
+ip-26-0-160-103:2068009:2068348 [0] NCCL INFO Channel 18/24 :    0   1   2   3
+ip-26-0-160-103:2068009:2068348 [0] NCCL INFO Channel 19/24 :    0   1   2   3
+ip-26-0-160-103:2068009:2068348 [0] NCCL INFO Channel 20/24 :    0   1   2   3
+ip-26-0-160-103:2068009:2068348 [0] NCCL INFO Channel 21/24 :    0   1   2   3
+ip-26-0-160-103:2068009:2068348 [0] NCCL INFO Channel 22/24 :    0   1   2   3
+ip-26-0-160-103:2068009:2068348 [0] NCCL INFO Channel 23/24 :    0   1   2   3
+ip-26-0-160-103:2068009:2068348 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] 1/-1/-1->0->-1 [7] 1/-1/-1->0->-1 [8] 1/-1/-1->0->-1 [9] 1/-1/-1->0->-1 [10] 1/-1/-1->0->-1 [11] 1/-1/-1->0->-1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] 1/-1/-1->0->-1 [19] 1/-1/-1->0->-1 [20] 1/-1/-1->0->-1 [21] 1/-1/-1->0->-1 [22] 1/-1/-1->0->-1 [23] 1/-1/-1->0->-1
+ip-26-0-160-103:2068009:2068348 [0] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-160-103:2068016:2068351 [7] NCCL INFO Trees [0] -1/-1/-1->3->2 [1] -1/-1/-1->3->2 [2] -1/-1/-1->3->2 [3] -1/-1/-1->3->2 [4] -1/-1/-1->3->2 [5] -1/-1/-1->3->2 [6] -1/-1/-1->3->2 [7] -1/-1/-1->3->2 [8] -1/-1/-1->3->2 [9] -1/-1/-1->3->2 [10] -1/-1/-1->3->2 [11] -1/-1/-1->3->2 [12] -1/-1/-1->3->2 [13] -1/-1/-1->3->2 [14] -1/-1/-1->3->2 [15] -1/-1/-1->3->2 [16] -1/-1/-1->3->2 [17] -1/-1/-1->3->2 [18] -1/-1/-1->3->2 [19] -1/-1/-1->3->2 [20] -1/-1/-1->3->2 [21] -1/-1/-1->3->2 [22] -1/-1/-1->3->2 [23] -1/-1/-1->3->2
+ip-26-0-160-103:2068016:2068351 [7] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-160-103:2068014:2068359 [5] NCCL INFO Trees [0] 3/-1/-1->2->1 [1] 3/-1/-1->2->1 [2] 3/-1/-1->2->1 [3] 3/-1/-1->2->1 [4] 3/-1/-1->2->1 [5] 3/-1/-1->2->1 [6] 3/-1/-1->2->1 [7] 3/-1/-1->2->1 [8] 3/-1/-1->2->1 [9] 3/-1/-1->2->1 [10] 3/-1/-1->2->1 [11] 3/-1/-1->2->1 [12] 3/-1/-1->2->1 [13] 3/-1/-1->2->1 [14] 3/-1/-1->2->1 [15] 3/-1/-1->2->1 [16] 3/-1/-1->2->1 [17] 3/-1/-1->2->1 [18] 3/-1/-1->2->1 [19] 3/-1/-1->2->1 [20] 3/-1/-1->2->1 [21] 3/-1/-1->2->1 [22] 3/-1/-1->2->1 [23] 3/-1/-1->2->1
+ip-26-0-160-103:2068010:2068349 [1] NCCL INFO Channel 00/24 :    0   1   2   3
+ip-26-0-160-103:2068014:2068359 [5] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-160-103:2068010:2068349 [1] NCCL INFO Channel 01/24 :    0   1   2   3
+ip-26-0-160-103:2068010:2068349 [1] NCCL INFO Channel 02/24 :    0   1   2   3
+ip-26-0-160-103:2068010:2068349 [1] NCCL INFO Channel 03/24 :    0   1   2   3
+ip-26-0-160-103:2068012:2068353 [3] NCCL INFO Trees [0] 2/-1/-1->1->0 [1] 2/-1/-1->1->0 [2] 2/-1/-1->1->0 [3] 2/-1/-1->1->0 [4] 2/-1/-1->1->0 [5] 2/-1/-1->1->0 [6] 2/-1/-1->1->0 [7] 2/-1/-1->1->0 [8] 2/-1/-1->1->0 [9] 2/-1/-1->1->0 [10] 2/-1/-1->1->0 [11] 2/-1/-1->1->0 [12] 2/-1/-1->1->0 [13] 2/-1/-1->1->0 [14] 2/-1/-1->1->0 [15] 2/-1/-1->1->0 [16] 2/-1/-1->1->0 [17] 2/-1/-1->1->0 [18] 2/-1/-1->1->0 [19] 2/-1/-1->1->0 [20] 2/-1/-1->1->0 [21] 2/-1/-1->1->0 [22] 2/-1/-1->1->0 [23] 2/-1/-1->1->0
+ip-26-0-160-103:2068010:2068349 [1] NCCL INFO Channel 04/24 :    0   1   2   3
+ip-26-0-160-103:2068010:2068349 [1] NCCL INFO Channel 05/24 :    0   1   2   3
+ip-26-0-160-103:2068012:2068353 [3] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-160-103:2068010:2068349 [1] NCCL INFO Channel 06/24 :    0   1   2   3
+ip-26-0-160-103:2068010:2068349 [1] NCCL INFO Channel 07/24 :    0   1   2   3
+ip-26-0-160-103:2068010:2068349 [1] NCCL INFO Channel 08/24 :    0   1   2   3
+ip-26-0-160-103:2068010:2068349 [1] NCCL INFO Channel 09/24 :    0   1   2   3
+ip-26-0-160-103:2068010:2068349 [1] NCCL INFO Channel 10/24 :    0   1   2   3
+ip-26-0-160-103:2068010:2068349 [1] NCCL INFO Channel 11/24 :    0   1   2   3
+ip-26-0-160-103:2068010:2068349 [1] NCCL INFO Channel 12/24 :    0   1   2   3
+ip-26-0-160-103:2068010:2068349 [1] NCCL INFO Channel 13/24 :    0   1   2   3
+ip-26-0-160-103:2068010:2068349 [1] NCCL INFO Channel 14/24 :    0   1   2   3
+ip-26-0-160-103:2068010:2068349 [1] NCCL INFO Channel 15/24 :    0   1   2   3
+ip-26-0-160-103:2068010:2068349 [1] NCCL INFO Channel 16/24 :    0   1   2   3
+ip-26-0-160-103:2068010:2068349 [1] NCCL INFO Channel 17/24 :    0   1   2   3
+ip-26-0-160-103:2068010:2068349 [1] NCCL INFO Channel 18/24 :    0   1   2   3
+ip-26-0-160-103:2068010:2068349 [1] NCCL INFO Channel 19/24 :    0   1   2   3
+ip-26-0-160-103:2068010:2068349 [1] NCCL INFO Channel 20/24 :    0   1   2   3
+ip-26-0-160-103:2068010:2068349 [1] NCCL INFO Channel 21/24 :    0   1   2   3
+ip-26-0-160-103:2068010:2068349 [1] NCCL INFO Channel 22/24 :    0   1   2   3
+ip-26-0-160-103:2068010:2068349 [1] NCCL INFO Channel 23/24 :    0   1   2   3
+ip-26-0-160-103:2068010:2068349 [1] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] 1/-1/-1->0->-1 [7] 1/-1/-1->0->-1 [8] 1/-1/-1->0->-1 [9] 1/-1/-1->0->-1 [10] 1/-1/-1->0->-1 [11] 1/-1/-1->0->-1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] 1/-1/-1->0->-1 [19] 1/-1/-1->0->-1 [20] 1/-1/-1->0->-1 [21] 1/-1/-1->0->-1 [22] 1/-1/-1->0->-1 [23] 1/-1/-1->0->-1
+ip-26-0-160-103:2068010:2068349 [1] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-160-103:2068013:2068360 [4] NCCL INFO Channel 00/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-160-103:2068013:2068360 [4] NCCL INFO Channel 01/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-160-103:2068015:2068350 [6] NCCL INFO Channel 00/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068013:2068360 [4] NCCL INFO Channel 02/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-160-103:2068016:2068351 [7] NCCL INFO Channel 00/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-160-103:2068015:2068350 [6] NCCL INFO Channel 01/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068016:2068351 [7] NCCL INFO Channel 01/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-160-103:2068013:2068360 [4] NCCL INFO Channel 03/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-160-103:2068013:2068360 [4] NCCL INFO Channel 04/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-160-103:2068016:2068351 [7] NCCL INFO Channel 02/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-160-103:2068013:2068360 [4] NCCL INFO Channel 05/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-160-103:2068016:2068351 [7] NCCL INFO Channel 03/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-160-103:2068013:2068360 [4] NCCL INFO Channel 06/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-160-103:2068016:2068351 [7] NCCL INFO Channel 04/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-160-103:2068010:2068349 [1] NCCL INFO Channel 00/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2068013:2068360 [4] NCCL INFO Channel 07/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-160-103:2068016:2068351 [7] NCCL INFO Channel 05/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-160-103:2068011:2068352 [2] NCCL INFO Channel 00/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-160-103:2068010:2068349 [1] NCCL INFO Channel 01/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2068009:2068348 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-160-103:2068013:2068360 [4] NCCL INFO Channel 08/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-160-103:2068016:2068351 [7] NCCL INFO Channel 06/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-160-103:2068010:2068349 [1] NCCL INFO Channel 02/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2068015:2068350 [6] NCCL INFO Channel 02/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068011:2068352 [2] NCCL INFO Channel 01/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-160-103:2068009:2068348 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-160-103:2068013:2068360 [4] NCCL INFO Channel 09/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-160-103:2068016:2068351 [7] NCCL INFO Channel 07/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-160-103:2068010:2068349 [1] NCCL INFO Channel 03/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2068011:2068352 [2] NCCL INFO Channel 02/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-160-103:2068015:2068350 [6] NCCL INFO Channel 03/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068009:2068348 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-160-103:2068013:2068360 [4] NCCL INFO Channel 10/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-160-103:2068016:2068351 [7] NCCL INFO Channel 08/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-160-103:2068010:2068349 [1] NCCL INFO Channel 04/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2068011:2068352 [2] NCCL INFO Channel 03/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-160-103:2068015:2068350 [6] NCCL INFO Channel 04/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068009:2068348 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-160-103:2068013:2068360 [4] NCCL INFO Channel 11/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-160-103:2068016:2068351 [7] NCCL INFO Channel 09/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-160-103:2068011:2068352 [2] NCCL INFO Channel 04/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-160-103:2068010:2068349 [1] NCCL INFO Channel 05/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2068015:2068350 [6] NCCL INFO Channel 05/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068013:2068360 [4] NCCL INFO Channel 12/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-160-103:2068016:2068351 [7] NCCL INFO Channel 10/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-160-103:2068010:2068349 [1] NCCL INFO Channel 06/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2068013:2068360 [4] NCCL INFO Channel 13/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-160-103:2068016:2068351 [7] NCCL INFO Channel 11/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-160-103:2068010:2068349 [1] NCCL INFO Channel 07/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2068013:2068360 [4] NCCL INFO Channel 14/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-160-103:2068016:2068351 [7] NCCL INFO Channel 12/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-160-103:2068012:2068353 [3] NCCL INFO Channel 00/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-160-103:2068010:2068349 [1] NCCL INFO Channel 08/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2068013:2068360 [4] NCCL INFO Channel 15/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-160-103:2068016:2068351 [7] NCCL INFO Channel 13/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-160-103:2068012:2068353 [3] NCCL INFO Channel 01/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-160-103:2068010:2068349 [1] NCCL INFO Channel 09/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2068013:2068360 [4] NCCL INFO Channel 16/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-160-103:2068016:2068351 [7] NCCL INFO Channel 14/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-160-103:2068010:2068349 [1] NCCL INFO Channel 10/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2068009:2068348 [0] NCCL INFO Channel 04/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-160-103:2068012:2068353 [3] NCCL INFO Channel 02/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-160-103:2068013:2068360 [4] NCCL INFO Channel 17/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-160-103:2068016:2068351 [7] NCCL INFO Channel 15/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-160-103:2068011:2068352 [2] NCCL INFO Channel 05/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-160-103:2068015:2068350 [6] NCCL INFO Channel 06/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068010:2068349 [1] NCCL INFO Channel 11/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2068012:2068353 [3] NCCL INFO Channel 03/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-160-103:2068009:2068348 [0] NCCL INFO Channel 05/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-160-103:2068014:2068359 [5] NCCL INFO Channel 00/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-160-103:2068013:2068360 [4] NCCL INFO Channel 18/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-160-103:2068016:2068351 [7] NCCL INFO Channel 16/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-160-103:2068011:2068352 [2] NCCL INFO Channel 06/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-160-103:2068015:2068350 [6] NCCL INFO Channel 07/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068012:2068353 [3] NCCL INFO Channel 04/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-160-103:2068010:2068349 [1] NCCL INFO Channel 12/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2068009:2068348 [0] NCCL INFO Channel 06/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-160-103:2068014:2068359 [5] NCCL INFO Channel 01/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-160-103:2068013:2068360 [4] NCCL INFO Channel 19/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-160-103:2068016:2068351 [7] NCCL INFO Channel 17/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-160-103:2068011:2068352 [2] NCCL INFO Channel 07/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-160-103:2068015:2068350 [6] NCCL INFO Channel 08/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068012:2068353 [3] NCCL INFO Channel 05/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-160-103:2068009:2068348 [0] NCCL INFO Channel 07/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-160-103:2068010:2068349 [1] NCCL INFO Channel 13/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2068014:2068359 [5] NCCL INFO Channel 02/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-160-103:2068013:2068360 [4] NCCL INFO Channel 20/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-160-103:2068016:2068351 [7] NCCL INFO Channel 18/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-160-103:2068011:2068352 [2] NCCL INFO Channel 08/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-160-103:2068012:2068353 [3] NCCL INFO Channel 06/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-160-103:2068009:2068348 [0] NCCL INFO Channel 08/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-160-103:2068010:2068349 [1] NCCL INFO Channel 14/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2068015:2068350 [6] NCCL INFO Channel 09/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068014:2068359 [5] NCCL INFO Channel 03/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-160-103:2068013:2068360 [4] NCCL INFO Channel 21/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-160-103:2068016:2068351 [7] NCCL INFO Channel 19/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-160-103:2068011:2068352 [2] NCCL INFO Channel 09/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-160-103:2068010:2068349 [1] NCCL INFO Channel 15/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2068012:2068353 [3] NCCL INFO Channel 07/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-160-103:2068009:2068348 [0] NCCL INFO Channel 09/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-160-103:2068015:2068350 [6] NCCL INFO Channel 10/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068014:2068359 [5] NCCL INFO Channel 04/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-160-103:2068013:2068360 [4] NCCL INFO Channel 22/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-160-103:2068016:2068351 [7] NCCL INFO Channel 20/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-160-103:2068011:2068352 [2] NCCL INFO Channel 10/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-160-103:2068010:2068349 [1] NCCL INFO Channel 16/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2068012:2068353 [3] NCCL INFO Channel 08/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-160-103:2068009:2068348 [0] NCCL INFO Channel 10/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-160-103:2068014:2068359 [5] NCCL INFO Channel 05/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-160-103:2068015:2068350 [6] NCCL INFO Channel 11/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068013:2068360 [4] NCCL INFO Channel 23/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-160-103:2068016:2068351 [7] NCCL INFO Channel 21/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-160-103:2068010:2068349 [1] NCCL INFO Channel 17/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2068011:2068352 [2] NCCL INFO Channel 11/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-160-103:2068012:2068353 [3] NCCL INFO Channel 09/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-160-103:2068009:2068348 [0] NCCL INFO Channel 11/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-160-103:2068014:2068359 [5] NCCL INFO Channel 06/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-160-103:2068015:2068350 [6] NCCL INFO Channel 12/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068016:2068351 [7] NCCL INFO Channel 22/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-160-103:2068010:2068349 [1] NCCL INFO Channel 18/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2068012:2068353 [3] NCCL INFO Channel 10/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-160-103:2068011:2068352 [2] NCCL INFO Channel 12/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-160-103:2068009:2068348 [0] NCCL INFO Channel 12/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-160-103:2068014:2068359 [5] NCCL INFO Channel 07/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-160-103:2068015:2068350 [6] NCCL INFO Channel 13/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068016:2068351 [7] NCCL INFO Channel 23/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-160-103:2068010:2068349 [1] NCCL INFO Channel 19/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2068012:2068353 [3] NCCL INFO Channel 11/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-160-103:2068011:2068352 [2] NCCL INFO Channel 13/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-160-103:2068009:2068348 [0] NCCL INFO Channel 13/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-160-103:2068014:2068359 [5] NCCL INFO Channel 08/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-160-103:2068015:2068350 [6] NCCL INFO Channel 14/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068010:2068349 [1] NCCL INFO Channel 20/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2068015:2068350 [6] NCCL INFO Channel 15/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068012:2068353 [3] NCCL INFO Channel 12/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-160-103:2068011:2068352 [2] NCCL INFO Channel 14/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-160-103:2068010:2068349 [1] NCCL INFO Channel 21/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2068015:2068350 [6] NCCL INFO Channel 16/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068012:2068353 [3] NCCL INFO Channel 13/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-160-103:2068011:2068352 [2] NCCL INFO Channel 15/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-160-103:2068010:2068349 [1] NCCL INFO Channel 22/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2068015:2068350 [6] NCCL INFO Channel 17/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068012:2068353 [3] NCCL INFO Channel 14/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-160-103:2068011:2068352 [2] NCCL INFO Channel 16/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-160-103:2068010:2068349 [1] NCCL INFO Channel 23/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2068015:2068350 [6] NCCL INFO Channel 18/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068011:2068352 [2] NCCL INFO Channel 17/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-160-103:2068012:2068353 [3] NCCL INFO Channel 15/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-160-103:2068015:2068350 [6] NCCL INFO Channel 19/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068011:2068352 [2] NCCL INFO Channel 18/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-160-103:2068012:2068353 [3] NCCL INFO Channel 16/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-160-103:2068009:2068348 [0] NCCL INFO Channel 14/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-160-103:2068015:2068350 [6] NCCL INFO Channel 20/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068011:2068352 [2] NCCL INFO Channel 19/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-160-103:2068012:2068353 [3] NCCL INFO Channel 17/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-160-103:2068009:2068348 [0] NCCL INFO Channel 15/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-160-103:2068014:2068359 [5] NCCL INFO Channel 09/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-160-103:2068015:2068350 [6] NCCL INFO Channel 21/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068011:2068352 [2] NCCL INFO Channel 20/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-160-103:2068009:2068348 [0] NCCL INFO Channel 16/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-160-103:2068012:2068353 [3] NCCL INFO Channel 18/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-160-103:2068014:2068359 [5] NCCL INFO Channel 10/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-160-103:2068015:2068350 [6] NCCL INFO Channel 22/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068011:2068352 [2] NCCL INFO Channel 21/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-160-103:2068012:2068353 [3] NCCL INFO Channel 19/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-160-103:2068009:2068348 [0] NCCL INFO Channel 17/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-160-103:2068014:2068359 [5] NCCL INFO Channel 11/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-160-103:2068015:2068350 [6] NCCL INFO Channel 23/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068011:2068352 [2] NCCL INFO Channel 22/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-160-103:2068012:2068353 [3] NCCL INFO Channel 20/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-160-103:2068009:2068348 [0] NCCL INFO Channel 18/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-160-103:2068014:2068359 [5] NCCL INFO Channel 12/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-160-103:2068011:2068352 [2] NCCL INFO Channel 23/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-160-103:2068009:2068348 [0] NCCL INFO Channel 19/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-160-103:2068012:2068353 [3] NCCL INFO Channel 21/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-160-103:2068014:2068359 [5] NCCL INFO Channel 13/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-160-103:2068009:2068348 [0] NCCL INFO Channel 20/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-160-103:2068012:2068353 [3] NCCL INFO Channel 22/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-160-103:2068014:2068359 [5] NCCL INFO Channel 14/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-160-103:2068012:2068353 [3] NCCL INFO Channel 23/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-160-103:2068009:2068348 [0] NCCL INFO Channel 21/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-160-103:2068014:2068359 [5] NCCL INFO Channel 15/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-160-103:2068009:2068348 [0] NCCL INFO Channel 22/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-160-103:2068014:2068359 [5] NCCL INFO Channel 16/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-160-103:2068014:2068359 [5] NCCL INFO Channel 17/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-160-103:2068009:2068348 [0] NCCL INFO Channel 23/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-160-103:2068014:2068359 [5] NCCL INFO Channel 18/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-160-103:2068014:2068359 [5] NCCL INFO Channel 19/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-160-103:2068014:2068359 [5] NCCL INFO Channel 20/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-160-103:2068014:2068359 [5] NCCL INFO Channel 21/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-160-103:2068014:2068359 [5] NCCL INFO Channel 22/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-160-103:2068014:2068359 [5] NCCL INFO Channel 23/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-160-103:2068013:2068360 [4] NCCL INFO Connected all rings
+ip-26-0-160-103:2068010:2068349 [1] NCCL INFO Connected all rings
+ip-26-0-160-103:2068009:2068348 [0] NCCL INFO Connected all rings
+ip-26-0-160-103:2068015:2068350 [6] NCCL INFO Connected all rings
+ip-26-0-160-103:2068015:2068350 [6] NCCL INFO Channel 00/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-160-103:2068011:2068352 [2] NCCL INFO Connected all rings
+ip-26-0-160-103:2068014:2068359 [5] NCCL INFO Connected all rings
+ip-26-0-160-103:2068015:2068350 [6] NCCL INFO Channel 01/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-160-103:2068015:2068350 [6] NCCL INFO Channel 02/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-160-103:2068015:2068350 [6] NCCL INFO Channel 03/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-160-103:2068016:2068351 [7] NCCL INFO Connected all rings
+ip-26-0-160-103:2068016:2068351 [7] NCCL INFO Channel 00/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-160-103:2068015:2068350 [6] NCCL INFO Channel 04/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-160-103:2068012:2068353 [3] NCCL INFO Connected all rings
+ip-26-0-160-103:2068016:2068351 [7] NCCL INFO Channel 01/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-160-103:2068015:2068350 [6] NCCL INFO Channel 05/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-160-103:2068016:2068351 [7] NCCL INFO Channel 02/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-160-103:2068015:2068350 [6] NCCL INFO Channel 06/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-160-103:2068016:2068351 [7] NCCL INFO Channel 03/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-160-103:2068015:2068350 [6] NCCL INFO Channel 07/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-160-103:2068016:2068351 [7] NCCL INFO Channel 04/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-160-103:2068015:2068350 [6] NCCL INFO Channel 08/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-160-103:2068016:2068351 [7] NCCL INFO Channel 05/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-160-103:2068015:2068350 [6] NCCL INFO Channel 09/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-160-103:2068016:2068351 [7] NCCL INFO Channel 06/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-160-103:2068015:2068350 [6] NCCL INFO Channel 10/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-160-103:2068016:2068351 [7] NCCL INFO Channel 07/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-160-103:2068015:2068350 [6] NCCL INFO Channel 11/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-160-103:2068016:2068351 [7] NCCL INFO Channel 08/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-160-103:2068015:2068350 [6] NCCL INFO Channel 12/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-160-103:2068016:2068351 [7] NCCL INFO Channel 09/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-160-103:2068015:2068350 [6] NCCL INFO Channel 13/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-160-103:2068016:2068351 [7] NCCL INFO Channel 10/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-160-103:2068015:2068350 [6] NCCL INFO Channel 14/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-160-103:2068016:2068351 [7] NCCL INFO Channel 11/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-160-103:2068015:2068350 [6] NCCL INFO Channel 15/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-160-103:2068016:2068351 [7] NCCL INFO Channel 12/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-160-103:2068015:2068350 [6] NCCL INFO Channel 16/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-160-103:2068016:2068351 [7] NCCL INFO Channel 13/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-160-103:2068015:2068350 [6] NCCL INFO Channel 17/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-160-103:2068016:2068351 [7] NCCL INFO Channel 14/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-160-103:2068015:2068350 [6] NCCL INFO Channel 18/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-160-103:2068016:2068351 [7] NCCL INFO Channel 15/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-160-103:2068015:2068350 [6] NCCL INFO Channel 19/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-160-103:2068013:2068360 [4] NCCL INFO Channel 00/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-160-103:2068016:2068351 [7] NCCL INFO Channel 16/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-160-103:2068015:2068350 [6] NCCL INFO Channel 20/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-160-103:2068013:2068360 [4] NCCL INFO Channel 01/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-160-103:2068016:2068351 [7] NCCL INFO Channel 17/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-160-103:2068015:2068350 [6] NCCL INFO Channel 21/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-160-103:2068013:2068360 [4] NCCL INFO Channel 02/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-160-103:2068016:2068351 [7] NCCL INFO Channel 18/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-160-103:2068015:2068350 [6] NCCL INFO Channel 22/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-160-103:2068013:2068360 [4] NCCL INFO Channel 03/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-160-103:2068016:2068351 [7] NCCL INFO Channel 19/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-160-103:2068015:2068350 [6] NCCL INFO Channel 23/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-160-103:2068013:2068360 [4] NCCL INFO Channel 04/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-160-103:2068016:2068351 [7] NCCL INFO Channel 20/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-160-103:2068011:2068352 [2] NCCL INFO Channel 00/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068013:2068360 [4] NCCL INFO Channel 05/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-160-103:2068016:2068351 [7] NCCL INFO Channel 21/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-160-103:2068011:2068352 [2] NCCL INFO Channel 01/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068013:2068360 [4] NCCL INFO Channel 06/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-160-103:2068016:2068351 [7] NCCL INFO Channel 22/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-160-103:2068011:2068352 [2] NCCL INFO Channel 02/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068013:2068360 [4] NCCL INFO Channel 07/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-160-103:2068016:2068351 [7] NCCL INFO Channel 23/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-160-103:2068014:2068359 [5] NCCL INFO Channel 00/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2068011:2068352 [2] NCCL INFO Channel 03/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068013:2068360 [4] NCCL INFO Channel 08/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-160-103:2068014:2068359 [5] NCCL INFO Channel 01/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2068011:2068352 [2] NCCL INFO Channel 04/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068013:2068360 [4] NCCL INFO Channel 09/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-160-103:2068012:2068353 [3] NCCL INFO Channel 00/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-160-103:2068014:2068359 [5] NCCL INFO Channel 02/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2068011:2068352 [2] NCCL INFO Channel 05/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068013:2068360 [4] NCCL INFO Channel 10/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-160-103:2068012:2068353 [3] NCCL INFO Channel 01/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-160-103:2068011:2068352 [2] NCCL INFO Channel 06/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068014:2068359 [5] NCCL INFO Channel 03/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2068013:2068360 [4] NCCL INFO Channel 11/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-160-103:2068012:2068353 [3] NCCL INFO Channel 02/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-160-103:2068011:2068352 [2] NCCL INFO Channel 07/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068014:2068359 [5] NCCL INFO Channel 04/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2068013:2068360 [4] NCCL INFO Channel 12/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-160-103:2068012:2068353 [3] NCCL INFO Channel 03/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-160-103:2068011:2068352 [2] NCCL INFO Channel 08/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068014:2068359 [5] NCCL INFO Channel 05/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2068013:2068360 [4] NCCL INFO Channel 13/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-160-103:2068012:2068353 [3] NCCL INFO Channel 04/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-160-103:2068014:2068359 [5] NCCL INFO Channel 06/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2068011:2068352 [2] NCCL INFO Channel 09/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068013:2068360 [4] NCCL INFO Channel 14/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-160-103:2068012:2068353 [3] NCCL INFO Channel 05/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-160-103:2068014:2068359 [5] NCCL INFO Channel 07/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2068011:2068352 [2] NCCL INFO Channel 10/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068013:2068360 [4] NCCL INFO Channel 15/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-160-103:2068011:2068352 [2] NCCL INFO Channel 11/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068012:2068353 [3] NCCL INFO Channel 06/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-160-103:2068014:2068359 [5] NCCL INFO Channel 08/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2068013:2068360 [4] NCCL INFO Channel 16/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-160-103:2068012:2068353 [3] NCCL INFO Channel 07/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-160-103:2068011:2068352 [2] NCCL INFO Channel 12/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068014:2068359 [5] NCCL INFO Channel 09/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2068013:2068360 [4] NCCL INFO Channel 17/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-160-103:2068012:2068353 [3] NCCL INFO Channel 08/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-160-103:2068011:2068352 [2] NCCL INFO Channel 13/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068013:2068360 [4] NCCL INFO Channel 18/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-160-103:2068014:2068359 [5] NCCL INFO Channel 10/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2068012:2068353 [3] NCCL INFO Channel 09/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-160-103:2068011:2068352 [2] NCCL INFO Channel 14/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068013:2068360 [4] NCCL INFO Channel 19/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-160-103:2068014:2068359 [5] NCCL INFO Channel 11/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2068012:2068353 [3] NCCL INFO Channel 10/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-160-103:2068011:2068352 [2] NCCL INFO Channel 15/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068013:2068360 [4] NCCL INFO Channel 20/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-160-103:2068014:2068359 [5] NCCL INFO Channel 12/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2068012:2068353 [3] NCCL INFO Channel 11/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-160-103:2068013:2068360 [4] NCCL INFO Channel 21/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-160-103:2068011:2068352 [2] NCCL INFO Channel 16/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068014:2068359 [5] NCCL INFO Channel 13/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2068012:2068353 [3] NCCL INFO Channel 12/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-160-103:2068013:2068360 [4] NCCL INFO Channel 22/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-160-103:2068011:2068352 [2] NCCL INFO Channel 17/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068014:2068359 [5] NCCL INFO Channel 14/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2068012:2068353 [3] NCCL INFO Channel 13/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-160-103:2068013:2068360 [4] NCCL INFO Channel 23/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-160-103:2068011:2068352 [2] NCCL INFO Channel 18/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068014:2068359 [5] NCCL INFO Channel 15/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2068012:2068353 [3] NCCL INFO Channel 14/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-160-103:2068011:2068352 [2] NCCL INFO Channel 19/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068014:2068359 [5] NCCL INFO Channel 16/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2068012:2068353 [3] NCCL INFO Channel 15/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-160-103:2068011:2068352 [2] NCCL INFO Channel 20/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068014:2068359 [5] NCCL INFO Channel 17/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2068012:2068353 [3] NCCL INFO Channel 16/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-160-103:2068011:2068352 [2] NCCL INFO Channel 21/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068014:2068359 [5] NCCL INFO Channel 18/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2068012:2068353 [3] NCCL INFO Channel 17/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-160-103:2068012:2068353 [3] NCCL INFO Channel 18/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-160-103:2068014:2068359 [5] NCCL INFO Channel 19/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2068012:2068353 [3] NCCL INFO Channel 19/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-160-103:2068014:2068359 [5] NCCL INFO Channel 20/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2068012:2068353 [3] NCCL INFO Channel 20/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-160-103:2068011:2068352 [2] NCCL INFO Channel 22/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068014:2068359 [5] NCCL INFO Channel 21/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2068012:2068353 [3] NCCL INFO Channel 21/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-160-103:2068011:2068352 [2] NCCL INFO Channel 23/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2068014:2068359 [5] NCCL INFO Channel 22/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2068012:2068353 [3] NCCL INFO Channel 22/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-160-103:2068014:2068359 [5] NCCL INFO Channel 23/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2068012:2068353 [3] NCCL INFO Channel 23/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-160-103:2068015:2068350 [6] NCCL INFO Connected all trees
+ip-26-0-160-103:2068015:2068350 [6] NCCL INFO NVLS comm 0xb109a50 headRank 3 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-160-103:2068016:2068351 [7] NCCL INFO Connected all trees
+ip-26-0-160-103:2068016:2068351 [7] NCCL INFO NVLS comm 0xb45a580 headRank 3 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-160-103:2068009:2068348 [0] NCCL INFO Connected all trees
+ip-26-0-160-103:2068009:2068348 [0] NCCL INFO NVLS comm 0xb6cbb80 headRank 0 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-160-103:2068010:2068349 [1] NCCL INFO Connected all trees
+ip-26-0-160-103:2068010:2068349 [1] NCCL INFO NVLS comm 0xc042060 headRank 0 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-160-103:2068013:2068360 [4] NCCL INFO Connected all trees
+ip-26-0-160-103:2068013:2068360 [4] NCCL INFO NVLS comm 0xbb2c450 headRank 2 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-160-103:2068014:2068359 [5] NCCL INFO Connected all trees
+ip-26-0-160-103:2068014:2068359 [5] NCCL INFO NVLS comm 0xb40dcb0 headRank 2 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-160-103:2068011:2068352 [2] NCCL INFO Connected all trees
+ip-26-0-160-103:2068011:2068352 [2] NCCL INFO NVLS comm 0xbe1f620 headRank 1 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-160-103:2068012:2068353 [3] NCCL INFO Connected all trees
+ip-26-0-160-103:2068012:2068353 [3] NCCL INFO NVLS comm 0xc13fae0 headRank 1 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-160-103:2068013:2068360 [4] NCCL INFO Connected NVLS tree
+ip-26-0-160-103:2068013:2068360 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-103:2068013:2068360 [4] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-160-103:2068013:2068360 [4] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-160-103:2068011:2068352 [2] NCCL INFO Connected NVLS tree
+ip-26-0-160-103:2068011:2068352 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-103:2068011:2068352 [2] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-160-103:2068011:2068352 [2] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-160-103:2068009:2068348 [0] NCCL INFO Connected NVLS tree
+ip-26-0-160-103:2068009:2068348 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-103:2068009:2068348 [0] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-160-103:2068009:2068348 [0] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-160-103:2068015:2068350 [6] NCCL INFO Connected NVLS tree
+ip-26-0-160-103:2068015:2068350 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-103:2068015:2068350 [6] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-160-103:2068015:2068350 [6] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-160-103:2068013:2068360 [4] NCCL INFO comm 0xbb2c450 rank 2 nranks 4 cudaDev 4 nvmlDev 4 busId 97000 commId 0x91ce3d4cd722004e - Init COMPLETE
+ip-26-0-160-103:2068011:2068352 [2] NCCL INFO comm 0xbe1f620 rank 1 nranks 4 cudaDev 2 nvmlDev 2 busId 75000 commId 0x91ce3d4cd722004e - Init COMPLETE
+ip-26-0-160-103:2068009:2068348 [0] NCCL INFO comm 0xb6cbb80 rank 0 nranks 4 cudaDev 0 nvmlDev 0 busId 53000 commId 0x91ce3d4cd722004e - Init COMPLETE
+ip-26-0-160-103:2068015:2068350 [6] NCCL INFO comm 0xb109a50 rank 3 nranks 4 cudaDev 6 nvmlDev 6 busId b9000 commId 0x91ce3d4cd722004e - Init COMPLETE
+ip-26-0-160-103:2068010:2068349 [1] NCCL INFO Connected NVLS tree
+ip-26-0-160-103:2068014:2068359 [5] NCCL INFO Connected NVLS tree
+ip-26-0-160-103:2068014:2068359 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-103:2068014:2068359 [5] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-160-103:2068014:2068359 [5] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-160-103:2068010:2068349 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-103:2068016:2068351 [7] NCCL INFO Connected NVLS tree
+ip-26-0-160-103:2068012:2068353 [3] NCCL INFO Connected NVLS tree
+ip-26-0-160-103:2068016:2068351 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-103:2068012:2068353 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-103:2068016:2068351 [7] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-160-103:2068016:2068351 [7] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-160-103:2068012:2068353 [3] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-160-103:2068012:2068353 [3] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-160-103:2068010:2068349 [1] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-160-103:2068010:2068349 [1] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-160-103:2068012:2068353 [3] NCCL INFO comm 0xc13fae0 rank 1 nranks 4 cudaDev 3 nvmlDev 3 busId 86000 commId 0x2e5e3292a10478a0 - Init COMPLETE
+ip-26-0-160-103:2068016:2068351 [7] NCCL INFO comm 0xb45a580 rank 3 nranks 4 cudaDev 7 nvmlDev 7 busId ca000 commId 0x2e5e3292a10478a0 - Init COMPLETE
+ip-26-0-160-103:2068014:2068359 [5] NCCL INFO comm 0xb40dcb0 rank 2 nranks 4 cudaDev 5 nvmlDev 5 busId a8000 commId 0x2e5e3292a10478a0 - Init COMPLETE
+ip-26-0-160-103:2068010:2068349 [1] NCCL INFO comm 0xc042060 rank 0 nranks 4 cudaDev 1 nvmlDev 1 busId 64000 commId 0x2e5e3292a10478a0 - Init COMPLETE
+12/12/2024 22:02:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]: [Optimizer Building] Using LearningRateForSP as learning rate
+12/12/2024 22:02:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]: [ZeRO sharding] Size of optimizer params per rank:
+12/12/2024 22:02:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]: [ZeRO sharding] DP Rank 0 has 130M out of 520M (25.00%) params' optimizer states
+12/12/2024 22:02:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]: [ZeRO sharding] DP Rank 1 has 130M out of 520M (25.00%) params' optimizer states
+12/12/2024 22:02:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]: [ZeRO sharding] DP Rank 2 has 130M out of 520M (25.00%) params' optimizer states
+12/12/2024 22:02:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]: [ZeRO sharding] DP Rank 3 has 130M out of 520M (25.00%) params' optimizer states
+12/12/2024 22:02:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]: [Training Plan] Stage Stable Training Stage has 99 remaining training steps and has consumed 0 samples
+12/12/2024 22:02:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]: Using dummy data generator
+12/12/2024 22:02:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]: [Training Plan] There are 1 training stages 
+12/12/2024 22:02:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]: [Stage Stable Training Stage] start from step 1 
+12/12/2024 22:02:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]: 
+12/12/2024 22:02:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]: [Start training] datetime: 2024-12-12 22:02:53.752192 | mbs: 2 | grad_accum: 256 | global_batch_size: 2048 | sequence_length: 2048 | train_steps: 100 | start_iteration_step: 0 | consumed_train_samples: 0
+wandb: WARNING Path /fsx/nouamane/.cache/wandb/wandb/ wasn't writable, using system temp directory.
+wandb: WARNING Path /fsx/nouamane/.cache/wandb/wandb/ wasn't writable, using system temp directory
+wandb: Tracking run with wandb version 0.16.0
+wandb: W&B syncing is set to `offline` in this directory.  
+wandb: Run `wandb online` or set WANDB_MODE=online to enable cloud syncing.
+12/12/2024 22:03:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]: Resuming training from stage Stable Training Stage, it has trained for 0 samples and has 99 remaining train steps
+12/12/2024 22:03:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:  Memory usage: 3472.48MiB. Peak allocated 5632.00MiB. Peak reserved: 19908.00MiB
+12/12/2024 22:03:23 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:  Memory usage: 3572.67MiB. Peak allocated 6985.16MiB. Peak reserved: 7434.00MiB
+12/12/2024 22:03:23 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]: iteration: 1 / 100 | consumed_tokens: 4.19M | elapsed_time_per_iteration_ms: 21.2K | tokens_per_sec: 198K | tokens_per_sec_per_gpu: 24.7K | global_batch_size: 2.05K | lm_loss: 10.8 | lr: 0.00015 | model_tflops_per_gpu: 174 | hardware_tflops_per_gpu: 174 | grad_norm: 0.431 | cuda_memory_allocated: 4.81G | cuda_max_memory_reserved: 9.62G | hd_total_memory_tb: 312G | hd_used_memory_tb: 69.4G | hd_free_memory_tb: 243G
+12/12/2024 22:03:23 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:  Memory usage: 4587.85MiB. Peak allocated 6820.14MiB. Peak reserved: 9178.00MiB
+12/12/2024 22:03:42 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:  Memory usage: 4587.97MiB. Peak allocated 8000.47MiB. Peak reserved: 9178.00MiB
+12/12/2024 22:03:42 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]: iteration: 2 / 100 | consumed_tokens: 8.39M | elapsed_time_per_iteration_ms: 19.2K | tokens_per_sec: 219K | tokens_per_sec_per_gpu: 27.4K | global_batch_size: 2.05K | lm_loss: 10.8 | lr: 0.0003 | model_tflops_per_gpu: 193 | hardware_tflops_per_gpu: 193 | grad_norm: 0.431 | cuda_memory_allocated: 4.81G | cuda_max_memory_reserved: 9.62G | hd_total_memory_tb: 312G | hd_used_memory_tb: 69.4G | hd_free_memory_tb: 243G
+12/12/2024 22:03:42 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:  Memory usage: 4587.85MiB. Peak allocated 6820.26MiB. Peak reserved: 9178.00MiB
+12/12/2024 22:04:01 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:  Memory usage: 4587.97MiB. Peak allocated 8000.47MiB. Peak reserved: 9178.00MiB
+12/12/2024 22:04:01 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]: iteration: 3 / 100 | consumed_tokens: 12.6M | elapsed_time_per_iteration_ms: 19K | tokens_per_sec: 221K | tokens_per_sec_per_gpu: 27.7K | global_batch_size: 2.05K | lm_loss: 10.8 | lr: 0.000296 | model_tflops_per_gpu: 195 | hardware_tflops_per_gpu: 195 | grad_norm: 0.424 | cuda_memory_allocated: 4.81G | cuda_max_memory_reserved: 9.62G | hd_total_memory_tb: 312G | hd_used_memory_tb: 69.4G | hd_free_memory_tb: 243G
+12/12/2024 22:04:01 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]: | job_id   | name                                                           | nodes | seq_len | mbs | batch_accum | gbs  | mTFLOPs | hTFLOPs | tok/s/gpu | AllReduce (GB/s) | AllGather (GB/s) | ReduceScatter (GB/s) | AR Intra-node (GB/s) | AG Intra-node (GB/s) | RS Intra-node (GB/s) | Mem Alloc (GB) | Mem Res (GB) | dp | pp | tp | pp_engine                                                                                                 | tp_mode                                 | tp_async_comm | hidden_size | hidden_act | num_layers | num_heads | num_kv_heads | max_pos | vocab_size | tie_word_embeddings | dtype          | zero_stage | ddp_bucket_cap_mb | accumulate_grad_in_fp32 | Total Params | Local Params |
+12/12/2024 22:04:01 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]: | -------- | -------------------------------------------------------------- | ----- | ------- | --- | ----------- | ---- | ------- | ------- | --------- | ---------------- | ---------------- | -------------------- | -------------------- | -------------------- | -------------------- | -------------- | ------------ | -- | -- | -- | --------------------------------------------------------------------------------------------------------- | --------------------------------------- | ------------- | ----------- | ---------- | ---------- | --------- | ------------ | ------- | ---------- | ------------------- | -------------- | ---------- | ----------------- | ----------------------- | ------------ | ------------ |
+12/12/2024 22:04:01 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]: | 13395441 | 1.14G_dp4_tp2_pp1_acc256_mbs2_seq2048_zero1_tpmodeRED_vocab32k | 1     | 2048    | 2   | 256         | 2048 | 194.91  | 194.91  | 27661.40  | 459.84           | 264.29           | 262.43               | 457.01               | 265.18               | 264.12               | 6.66           | 8.96         | 4  | 1  | 2  | <nanotron.parallel.pipeline_parallel.engine.OneForwardOneBackwardPipelineEngine object at 0x7fc0dae7ce80> | TensorParallelLinearMode.REDUCE_SCATTER | True          | 2048        | silu       | 16         | 32        | 8            | 2048    | 32768      | True                | torch.bfloat16 | 1          | 25                | True                    | 1.04G        | 520M         |
+12/12/2024 22:04:01 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]: Attempting to write benchmark results to CSV file: benchmark/results/bench_final.csv
+12/12/2024 22:04:01 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]: Acquired lock for CSV file: benchmark/results/bench_final.csv
+slurmstepd: error: *** JOB 13395441 ON ip-26-0-160-103 CANCELLED AT 2024-12-12T22:04:18 DUE TO TIME LIMIT ***
+slurmstepd: error: *** STEP 13395441.0 ON ip-26-0-160-103 CANCELLED AT 2024-12-12T22:04:18 DUE TO TIME LIMIT ***
+[2024-12-12 22:04:18,047] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-12 22:04:18,048] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2068009 closing signal SIGTERM
+[2024-12-12 22:04:18,048] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2068010 closing signal SIGTERM
+[2024-12-12 22:04:18,048] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2068011 closing signal SIGTERM
+[2024-12-12 22:04:18,048] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2068012 closing signal SIGTERM
+[2024-12-12 22:04:18,050] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2068013 closing signal SIGTERM
+[2024-12-12 22:04:18,050] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2068014 closing signal SIGTERM
+[2024-12-12 22:04:18,050] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2068015 closing signal SIGTERM
+[2024-12-12 22:04:18,050] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2068016 closing signal SIGTERM
+srun: Job step aborted: Waiting up to 32 seconds for job step to finish.
+[2024-12-12 22:04:18,166] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2068009 closing signal SIGTERM
+[2024-12-12 22:04:18,166] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2068010 closing signal SIGTERM
+[2024-12-12 22:04:18,166] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2068011 closing signal SIGTERM
+[2024-12-12 22:04:18,166] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2068012 closing signal SIGTERM
+[2024-12-12 22:04:18,166] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2068013 closing signal SIGTERM
+[2024-12-12 22:04:18,166] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2068014 closing signal SIGTERM
+[2024-12-12 22:04:18,166] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2068015 closing signal SIGTERM
+[2024-12-12 22:04:18,166] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2068016 closing signal SIGTERM
diff --git a/logs/13401706-bench_stress_test.out b/logs/13401706-bench_stress_test.out
new file mode 100644
index 0000000000000000000000000000000000000000..ffbd428f4136cf477db6d76f35ca08e6d090ea11
--- /dev/null
+++ b/logs/13401706-bench_stress_test.out
@@ -0,0 +1,518 @@
++ source /etc/profile.d/modules.sh
+++ . /usr/share/modules/init/bash
++++ unset _mlshdbg
++++ '[' 0 = 1 ']'
++++ unset _mlre _mlIFS
++++ '[' -n x ']'
++++ _mlIFS=' 	
+'
++++ IFS=' '
++++ '[' -n '' ']'
+++++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash autoinit
++++ _mlcode='module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+MODULEPATH_modshare=/etc/environment-modules/modules:1:/usr/share/modules/\$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1; export MODULEPATH_modshare;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+LOADEDMODULES=; export LOADEDMODULES;
+MODULEPATH=/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/\$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles; export MODULEPATH;
+test 0;'
++++ _mlret=0
++++ '[' -n x ']'
++++ IFS=' 	
+'
++++ unset _mlIFS
++++ unset _mlre _mlv _mlrv
++++ '[' 0 -eq 0 ']'
++++ eval 'module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+MODULEPATH_modshare=/etc/environment-modules/modules:1:/usr/share/modules/\$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1; export MODULEPATH_modshare;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+LOADEDMODULES=; export LOADEDMODULES;
+MODULEPATH=/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/\$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles; export MODULEPATH;
+test 0;'
+++++ MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl
+++++ export MODULES_CMD
+++++ ENV=/usr/share/modules/init/profile.sh
+++++ export ENV
+++++ MODULEPATH_modshare='/etc/environment-modules/modules:1:/usr/share/modules/$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1'
+++++ export MODULEPATH_modshare
+++++ BASH_ENV=/usr/share/modules/init/bash
+++++ export BASH_ENV
+++++ MODULESHOME=/usr/share/modules
+++++ export MODULESHOME
+++++ LOADEDMODULES=
+++++ export LOADEDMODULES
+++++ MODULEPATH='/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles'
+++++ export MODULEPATH
+++++ test 0
++++ '[' 0 = 1 ']'
++++ '[' -t 2 ']'
++++ export -f module
++++ export -f switchml
++++ '[' 5 -ge 3 ']'
++++ [[ ehxB =~ i ]]
++++ [[ ! :/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin: =~ :/usr/bin: ]]
++++ '[' '!' -n '' ']'
++++ MANPATH=:
++++ export MANPATH
+++++ manpath
++++ [[ ! :/admin/home/nouamane/.local/share/man:/fsx/nouamane/miniconda/envs/2-1-cu121/man:/fsx/nouamane/miniconda/envs/2-1-cu121/share/man:/fsx/nouamane/miniconda/man:/fsx/nouamane/miniconda/share/man:/opt/amazon/openmpi/share/man:/opt/amazon/efa/share/man:/opt/slurm/share/man:/usr/local/man:/usr/local/share/man:/usr/share/man:/admin/home/nouamane/.fzf/man:: =~ :/usr/share/man: ]]
++++ unset _mlcode _mlret
++++ '[' -n '' ']'
++ module load cuda/12.1
++ unset _mlshdbg
++ '[' 0 = 1 ']'
++ unset _mlre _mlIFS
++ '[' -n x ']'
++ _mlIFS=' 	
+'
++ IFS=' '
++ '[' -n '' ']'
+++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash load cuda/12.1
++ eval 'CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include;' export 'CPATH;
+LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:;' export 'LD_LIBRARY_PATH;
+FI_EFA_FORK_SAFE=1;' export 'FI_EFA_FORK_SAFE;
+MANPATH=/usr/local/cuda-12.1/share/man::;' export 'MANPATH;
+LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64;' export 'LIBRARY_PATH;
+_LMFILES_=/usr/share/modules/modulefiles/cuda/12.1;' export '_LMFILES_;
+LOADEDMODULES=cuda/12.1;' export 'LOADEDMODULES;
+MPI_PATH=/opt/amazon/openmpi;' export 'MPI_PATH;
+NCCL_HOME_modshare=/opt/nccl/build:1;' export 'NCCL_HOME_modshare;
+NCCL_PROTO=simple;' export 'NCCL_PROTO;
+MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1;' export 'MANPATH_modshare;
+LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1;' export 'LIBRARY_PATH_modshare;
+NCCL_SOCKET_IFNAME=enp;' export 'NCCL_SOCKET_IFNAME;
+AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl;' export 'AWS_OFI_NCCL_HOME;
+NCCL_HOME=/opt/nccl/build;' export 'NCCL_HOME;
+FI_PROVIDER=efa;' export 'FI_PROVIDER;
+AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1;' export 'AWS_OFI_NCCL_HOME_modshare;
+CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1;' export 'CPATH_modshare;
+LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2::1:/opt/nccl/build/lib:1:/opt/aws-ofi-nccl/lib:2:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/efa/lib:1;' export 'LD_LIBRARY_PATH_modshare;
+FI_EFA_ENABLE_SHM_TRANSFER=1;' export 'FI_EFA_ENABLE_SHM_TRANSFER;
+_LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1;' export '_LMFILES__modshare;
+LOADEDMODULES_modshare=cuda/12.1:1;' export 'LOADEDMODULES_modshare;
+MPI_PATH_modshare=/opt/amazon/openmpi:1;' export 'MPI_PATH_modshare;
+PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin;' export 'PATH;
+CUDA_HOME=/usr/local/cuda-12.1;' export 'CUDA_HOME;
+PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/usr/local/cuda-12.1/include:1:/opt/amazon/efa/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/usr/sbin:1:/fsx/nouamane/miniconda/bin:1:/opt/slurm/bin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:2:/opt/amazon/openmpi/bin:1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1;' export 'PATH_modshare;
+test' '0;'
+++ CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include
+++ export CPATH
+++ LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:
+++ export LD_LIBRARY_PATH
+++ FI_EFA_FORK_SAFE=1
+++ export FI_EFA_FORK_SAFE
+++ MANPATH=/usr/local/cuda-12.1/share/man::
+++ export MANPATH
+++ LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64
+++ export LIBRARY_PATH
+++ _LMFILES_=/usr/share/modules/modulefiles/cuda/12.1
+++ export _LMFILES_
+++ LOADEDMODULES=cuda/12.1
+++ export LOADEDMODULES
+++ MPI_PATH=/opt/amazon/openmpi
+++ export MPI_PATH
+++ NCCL_HOME_modshare=/opt/nccl/build:1
+++ export NCCL_HOME_modshare
+++ NCCL_PROTO=simple
+++ export NCCL_PROTO
+++ MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1
+++ export MANPATH_modshare
+++ LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1
+++ export LIBRARY_PATH_modshare
+++ NCCL_SOCKET_IFNAME=enp
+++ export NCCL_SOCKET_IFNAME
+++ AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl
+++ export AWS_OFI_NCCL_HOME
+++ NCCL_HOME=/opt/nccl/build
+++ export NCCL_HOME
+++ FI_PROVIDER=efa
+++ export FI_PROVIDER
+++ AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1
+++ export AWS_OFI_NCCL_HOME_modshare
+++ CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1
+++ export CPATH_modshare
+++ LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2::1:/opt/nccl/build/lib:1:/opt/aws-ofi-nccl/lib:2:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/efa/lib:1
+++ export LD_LIBRARY_PATH_modshare
+++ FI_EFA_ENABLE_SHM_TRANSFER=1
+++ export FI_EFA_ENABLE_SHM_TRANSFER
+++ _LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1
+++ export _LMFILES__modshare
+++ LOADEDMODULES_modshare=cuda/12.1:1
+++ export LOADEDMODULES_modshare
+++ MPI_PATH_modshare=/opt/amazon/openmpi:1
+++ export MPI_PATH_modshare
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ export PATH
+++ CUDA_HOME=/usr/local/cuda-12.1
+++ export CUDA_HOME
+++ PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/usr/local/cuda-12.1/include:1:/opt/amazon/efa/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/usr/sbin:1:/fsx/nouamane/miniconda/bin:1:/opt/slurm/bin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:2:/opt/amazon/openmpi/bin:1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1
+++ export PATH_modshare
+++ test 0
++ _mlstatus=0
++ '[' -n x ']'
++ IFS=' 	
+'
++ unset _mlre _mlv _mlrv _mlIFS
++ '[' -n '' ']'
++ unset _mlshdbg
++ return 0
++ source /fsx/nouamane/miniconda/bin/activate
+++ _CONDA_ROOT=/fsx/nouamane/miniconda
+++ . /fsx/nouamane/miniconda/etc/profile.d/conda.sh
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ '[' -z x ']'
+++ conda activate
+++ local cmd=activate
+++ case "$cmd" in
+++ __conda_activate activate
+++ '[' -n '' ']'
+++ local ask_conda
++++ PS1=
++++ __conda_exe shell.posix activate
++++ /fsx/nouamane/miniconda/bin/conda shell.posix activate
+++ ask_conda='. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_2='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\'''
+++ eval '. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_2='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\'''
++++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh
+++++ test -n ''
+++++ unset XML_CATALOG_FILES
+++++ unset xml_catalog_files_libxml2
++++ PS1='(base) '
++++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++++ export CONDA_PREFIX=/fsx/nouamane/miniconda
++++ CONDA_PREFIX=/fsx/nouamane/miniconda
++++ export CONDA_SHLVL=3
++++ CONDA_SHLVL=3
++++ export CONDA_DEFAULT_ENV=base
++++ CONDA_DEFAULT_ENV=base
++++ export 'CONDA_PROMPT_MODIFIER=(base) '
++++ CONDA_PROMPT_MODIFIER='(base) '
++++ export CONDA_PREFIX_2=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ CONDA_PREFIX_2=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ __conda_hashr
+++ '[' -n '' ']'
+++ '[' -n '' ']'
+++ hash -r
++ conda activate 2-1-cu121
++ local cmd=activate
++ case "$cmd" in
++ __conda_activate activate 2-1-cu121
++ '[' -n '' ']'
++ local ask_conda
+++ PS1='(base) '
+++ __conda_exe shell.posix activate 2-1-cu121
+++ /fsx/nouamane/miniconda/bin/conda shell.posix activate 2-1-cu121
++ ask_conda='PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''4'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_3='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
++ eval 'PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''4'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_3='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
+++ PS1='(2-1-cu121) '
+++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ export CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ export CONDA_SHLVL=4
+++ CONDA_SHLVL=4
+++ export CONDA_DEFAULT_ENV=2-1-cu121
+++ CONDA_DEFAULT_ENV=2-1-cu121
+++ export 'CONDA_PROMPT_MODIFIER=(2-1-cu121) '
+++ CONDA_PROMPT_MODIFIER='(2-1-cu121) '
+++ export CONDA_PREFIX_3=/fsx/nouamane/miniconda
+++ CONDA_PREFIX_3=/fsx/nouamane/miniconda
+++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ export _CE_M=
+++ _CE_M=
+++ export _CE_CONDA=
+++ _CE_CONDA=
+++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh
++++ test -n ''
++++ xml_catalog_files_libxml2=
++++ XML_CATALOG_FILES=
++++ conda_catalog_files=
++++ ifs_libxml2=' 	
+'
++++ IFS=' '
++++ rem=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ for pre in ${rem}
++++ test '' = /fsx/nouamane/miniconda/envs/2-1-cu121
++++ conda_catalog_files=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ rem=
++++ IFS=' 	
+'
++++ conda_catalog_files='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ export 'XML_CATALOG_FILES=file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ XML_CATALOG_FILES='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ unset conda_catalog_files ifs_libxml2 rem
++ __conda_hashr
++ '[' -n '' ']'
++ '[' -n '' ']'
++ hash -r
++ export PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++ PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ scontrol show hostnames ip-26-0-160-103
++ export NODELIST=ip-26-0-160-103
++ NODELIST=ip-26-0-160-103
+++ scontrol show hostnames ip-26-0-160-103
+++ head -n1
++ export MASTER_NODE=ip-26-0-160-103
++ MASTER_NODE=ip-26-0-160-103
++ export MASTER_PORT=12356
++ MASTER_PORT=12356
++ export NNODES=1
++ NNODES=1
++ export GPUS_PER_NODE=8
++ GPUS_PER_NODE=8
++ export WORLD_SIZE=8
++ WORLD_SIZE=8
++ export CUDA_DEVICE_MAX_CONNECTIONS=1
++ CUDA_DEVICE_MAX_CONNECTIONS=1
++ export NCCL_DEBUG=INFO
++ NCCL_DEBUG=INFO
++ export NANOTRON_BENCHMARK=1
++ NANOTRON_BENCHMARK=1
++ echo 'Master node: ip-26-0-160-103'
+Master node: ip-26-0-160-103
++ echo 'All nodes: ip-26-0-160-103'
+All nodes: ip-26-0-160-103
++ echo 'World size: 8'
+World size: 8
++ srun torchrun --nnodes=1 --nproc_per_node=8 --rdzv_id=13401706 --rdzv_backend=c10d --rdzv_endpoint=ip-26-0-160-103:12356 run_train.py --config-file benchmark/configs/config_1.14G_dp4_tp2_pp1_acc256_mbs2_seq2048_zero1_tpmodeRED_vocab32k.yaml
+[2024-12-13 00:21:20,910] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-13 00:21:20,911] torch.distributed.run: [WARNING] 
+[2024-12-13 00:21:20,911] torch.distributed.run: [WARNING] *****************************************
+[2024-12-13 00:21:20,911] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-13 00:21:20,911] torch.distributed.run: [WARNING] *****************************************
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+[2024-12-13 00:21:26,253] torch.distributed.elastic.multiprocessing.api: [ERROR] failed (exitcode: 2) local_rank: 0 (pid: 2134806) of binary: /fsx/nouamane/miniconda/envs/2-1-cu121/bin/python
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 33, in <module>
+    sys.exit(load_entry_point('torch==2.1.1', 'console_scripts', 'torchrun')())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 264, in launch_agent
+    raise ChildFailedError(
+torch.distributed.elastic.multiprocessing.errors.ChildFailedError: 
+============================================================
+run_train.py FAILED
+------------------------------------------------------------
+Failures:
+[1]:
+  time      : 2024-12-13_00:21:26
+  host      : ip-26-0-160-103.ec2.internal
+  rank      : 1 (local_rank: 1)
+  exitcode  : 2 (pid: 2134807)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[2]:
+  time      : 2024-12-13_00:21:26
+  host      : ip-26-0-160-103.ec2.internal
+  rank      : 2 (local_rank: 2)
+  exitcode  : 2 (pid: 2134808)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[3]:
+  time      : 2024-12-13_00:21:26
+  host      : ip-26-0-160-103.ec2.internal
+  rank      : 3 (local_rank: 3)
+  exitcode  : 2 (pid: 2134809)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[4]:
+  time      : 2024-12-13_00:21:26
+  host      : ip-26-0-160-103.ec2.internal
+  rank      : 4 (local_rank: 4)
+  exitcode  : 2 (pid: 2134810)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[5]:
+  time      : 2024-12-13_00:21:26
+  host      : ip-26-0-160-103.ec2.internal
+  rank      : 5 (local_rank: 5)
+  exitcode  : 2 (pid: 2134811)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[6]:
+  time      : 2024-12-13_00:21:26
+  host      : ip-26-0-160-103.ec2.internal
+  rank      : 6 (local_rank: 6)
+  exitcode  : 2 (pid: 2134812)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[7]:
+  time      : 2024-12-13_00:21:26
+  host      : ip-26-0-160-103.ec2.internal
+  rank      : 7 (local_rank: 7)
+  exitcode  : 2 (pid: 2134813)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+------------------------------------------------------------
+Root Cause (first observed failure):
+[0]:
+  time      : 2024-12-13_00:21:26
+  host      : ip-26-0-160-103.ec2.internal
+  rank      : 0 (local_rank: 0)
+  exitcode  : 2 (pid: 2134806)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+============================================================
+srun: error: ip-26-0-160-103: task 0: Exited with exit code 1
diff --git a/logs/13402229-bench_stress_test.out b/logs/13402229-bench_stress_test.out
new file mode 100644
index 0000000000000000000000000000000000000000..28290eff0ee215a7b172b9e98cb7bb3229df2722
--- /dev/null
+++ b/logs/13402229-bench_stress_test.out
@@ -0,0 +1,518 @@
++ source /etc/profile.d/modules.sh
+++ . /usr/share/modules/init/bash
++++ unset _mlshdbg
++++ '[' 0 = 1 ']'
++++ unset _mlre _mlIFS
++++ '[' -n x ']'
++++ _mlIFS=' 	
+'
++++ IFS=' '
++++ '[' -n '' ']'
+++++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash autoinit
++++ _mlcode='module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+MODULEPATH_modshare=/etc/environment-modules/modules:1:/usr/share/modules/\$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1; export MODULEPATH_modshare;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+LOADEDMODULES=; export LOADEDMODULES;
+MODULEPATH=/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/\$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles; export MODULEPATH;
+test 0;'
++++ _mlret=0
++++ '[' -n x ']'
++++ IFS=' 	
+'
++++ unset _mlIFS
++++ unset _mlre _mlv _mlrv
++++ '[' 0 -eq 0 ']'
++++ eval 'module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+MODULEPATH_modshare=/etc/environment-modules/modules:1:/usr/share/modules/\$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1; export MODULEPATH_modshare;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+LOADEDMODULES=; export LOADEDMODULES;
+MODULEPATH=/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/\$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles; export MODULEPATH;
+test 0;'
+++++ MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl
+++++ export MODULES_CMD
+++++ ENV=/usr/share/modules/init/profile.sh
+++++ export ENV
+++++ MODULEPATH_modshare='/etc/environment-modules/modules:1:/usr/share/modules/$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1'
+++++ export MODULEPATH_modshare
+++++ BASH_ENV=/usr/share/modules/init/bash
+++++ export BASH_ENV
+++++ MODULESHOME=/usr/share/modules
+++++ export MODULESHOME
+++++ LOADEDMODULES=
+++++ export LOADEDMODULES
+++++ MODULEPATH='/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles'
+++++ export MODULEPATH
+++++ test 0
++++ '[' 0 = 1 ']'
++++ '[' -t 2 ']'
++++ export -f module
++++ export -f switchml
++++ '[' 5 -ge 3 ']'
++++ [[ ehxB =~ i ]]
++++ [[ ! :/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin: =~ :/usr/bin: ]]
++++ '[' '!' -n '' ']'
++++ MANPATH=:
++++ export MANPATH
+++++ manpath
++++ [[ ! :/admin/home/nouamane/.local/share/man:/fsx/nouamane/miniconda/envs/2-1-cu121/man:/fsx/nouamane/miniconda/envs/2-1-cu121/share/man:/fsx/nouamane/miniconda/man:/fsx/nouamane/miniconda/share/man:/opt/amazon/openmpi/share/man:/opt/amazon/efa/share/man:/opt/slurm/share/man:/usr/local/man:/usr/local/share/man:/usr/share/man:/admin/home/nouamane/.fzf/man:: =~ :/usr/share/man: ]]
++++ unset _mlcode _mlret
++++ '[' -n '' ']'
++ module load cuda/12.1
++ unset _mlshdbg
++ '[' 0 = 1 ']'
++ unset _mlre _mlIFS
++ '[' -n x ']'
++ _mlIFS=' 	
+'
++ IFS=' '
++ '[' -n '' ']'
+++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash load cuda/12.1
++ eval 'CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include;' export 'CPATH;
+LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:;' export 'LD_LIBRARY_PATH;
+FI_EFA_FORK_SAFE=1;' export 'FI_EFA_FORK_SAFE;
+MANPATH=/usr/local/cuda-12.1/share/man::;' export 'MANPATH;
+LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64;' export 'LIBRARY_PATH;
+_LMFILES_=/usr/share/modules/modulefiles/cuda/12.1;' export '_LMFILES_;
+LOADEDMODULES=cuda/12.1;' export 'LOADEDMODULES;
+MPI_PATH=/opt/amazon/openmpi;' export 'MPI_PATH;
+NCCL_HOME_modshare=/opt/nccl/build:1;' export 'NCCL_HOME_modshare;
+NCCL_PROTO=simple;' export 'NCCL_PROTO;
+MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1;' export 'MANPATH_modshare;
+LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1;' export 'LIBRARY_PATH_modshare;
+NCCL_SOCKET_IFNAME=enp;' export 'NCCL_SOCKET_IFNAME;
+AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl;' export 'AWS_OFI_NCCL_HOME;
+NCCL_HOME=/opt/nccl/build;' export 'NCCL_HOME;
+FI_PROVIDER=efa;' export 'FI_PROVIDER;
+AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1;' export 'AWS_OFI_NCCL_HOME_modshare;
+CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1;' export 'CPATH_modshare;
+LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2::1:/opt/nccl/build/lib:1:/opt/aws-ofi-nccl/lib:2:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/efa/lib:1;' export 'LD_LIBRARY_PATH_modshare;
+FI_EFA_ENABLE_SHM_TRANSFER=1;' export 'FI_EFA_ENABLE_SHM_TRANSFER;
+_LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1;' export '_LMFILES__modshare;
+LOADEDMODULES_modshare=cuda/12.1:1;' export 'LOADEDMODULES_modshare;
+MPI_PATH_modshare=/opt/amazon/openmpi:1;' export 'MPI_PATH_modshare;
+PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin;' export 'PATH;
+CUDA_HOME=/usr/local/cuda-12.1;' export 'CUDA_HOME;
+PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/usr/local/cuda-12.1/include:1:/opt/amazon/efa/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/usr/sbin:1:/fsx/nouamane/miniconda/bin:1:/opt/slurm/bin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:2:/opt/amazon/openmpi/bin:1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1;' export 'PATH_modshare;
+test' '0;'
+++ CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include
+++ export CPATH
+++ LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:
+++ export LD_LIBRARY_PATH
+++ FI_EFA_FORK_SAFE=1
+++ export FI_EFA_FORK_SAFE
+++ MANPATH=/usr/local/cuda-12.1/share/man::
+++ export MANPATH
+++ LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64
+++ export LIBRARY_PATH
+++ _LMFILES_=/usr/share/modules/modulefiles/cuda/12.1
+++ export _LMFILES_
+++ LOADEDMODULES=cuda/12.1
+++ export LOADEDMODULES
+++ MPI_PATH=/opt/amazon/openmpi
+++ export MPI_PATH
+++ NCCL_HOME_modshare=/opt/nccl/build:1
+++ export NCCL_HOME_modshare
+++ NCCL_PROTO=simple
+++ export NCCL_PROTO
+++ MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1
+++ export MANPATH_modshare
+++ LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1
+++ export LIBRARY_PATH_modshare
+++ NCCL_SOCKET_IFNAME=enp
+++ export NCCL_SOCKET_IFNAME
+++ AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl
+++ export AWS_OFI_NCCL_HOME
+++ NCCL_HOME=/opt/nccl/build
+++ export NCCL_HOME
+++ FI_PROVIDER=efa
+++ export FI_PROVIDER
+++ AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1
+++ export AWS_OFI_NCCL_HOME_modshare
+++ CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1
+++ export CPATH_modshare
+++ LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2::1:/opt/nccl/build/lib:1:/opt/aws-ofi-nccl/lib:2:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/efa/lib:1
+++ export LD_LIBRARY_PATH_modshare
+++ FI_EFA_ENABLE_SHM_TRANSFER=1
+++ export FI_EFA_ENABLE_SHM_TRANSFER
+++ _LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1
+++ export _LMFILES__modshare
+++ LOADEDMODULES_modshare=cuda/12.1:1
+++ export LOADEDMODULES_modshare
+++ MPI_PATH_modshare=/opt/amazon/openmpi:1
+++ export MPI_PATH_modshare
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ export PATH
+++ CUDA_HOME=/usr/local/cuda-12.1
+++ export CUDA_HOME
+++ PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/usr/local/cuda-12.1/include:1:/opt/amazon/efa/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/usr/sbin:1:/fsx/nouamane/miniconda/bin:1:/opt/slurm/bin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:2:/opt/amazon/openmpi/bin:1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1
+++ export PATH_modshare
+++ test 0
++ _mlstatus=0
++ '[' -n x ']'
++ IFS=' 	
+'
++ unset _mlre _mlv _mlrv _mlIFS
++ '[' -n '' ']'
++ unset _mlshdbg
++ return 0
++ source /fsx/nouamane/miniconda/bin/activate
+++ _CONDA_ROOT=/fsx/nouamane/miniconda
+++ . /fsx/nouamane/miniconda/etc/profile.d/conda.sh
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ '[' -z x ']'
+++ conda activate
+++ local cmd=activate
+++ case "$cmd" in
+++ __conda_activate activate
+++ '[' -n '' ']'
+++ local ask_conda
++++ PS1=
++++ __conda_exe shell.posix activate
++++ /fsx/nouamane/miniconda/bin/conda shell.posix activate
+++ ask_conda='. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_2='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\'''
+++ eval '. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_2='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\'''
++++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh
+++++ test -n ''
+++++ unset XML_CATALOG_FILES
+++++ unset xml_catalog_files_libxml2
++++ PS1='(base) '
++++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++++ export CONDA_PREFIX=/fsx/nouamane/miniconda
++++ CONDA_PREFIX=/fsx/nouamane/miniconda
++++ export CONDA_SHLVL=3
++++ CONDA_SHLVL=3
++++ export CONDA_DEFAULT_ENV=base
++++ CONDA_DEFAULT_ENV=base
++++ export 'CONDA_PROMPT_MODIFIER=(base) '
++++ CONDA_PROMPT_MODIFIER='(base) '
++++ export CONDA_PREFIX_2=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ CONDA_PREFIX_2=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ __conda_hashr
+++ '[' -n '' ']'
+++ '[' -n '' ']'
+++ hash -r
++ conda activate 2-1-cu121
++ local cmd=activate
++ case "$cmd" in
++ __conda_activate activate 2-1-cu121
++ '[' -n '' ']'
++ local ask_conda
+++ PS1='(base) '
+++ __conda_exe shell.posix activate 2-1-cu121
+++ /fsx/nouamane/miniconda/bin/conda shell.posix activate 2-1-cu121
++ ask_conda='PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''4'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_3='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
++ eval 'PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''4'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_3='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
+++ PS1='(2-1-cu121) '
+++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ export CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ export CONDA_SHLVL=4
+++ CONDA_SHLVL=4
+++ export CONDA_DEFAULT_ENV=2-1-cu121
+++ CONDA_DEFAULT_ENV=2-1-cu121
+++ export 'CONDA_PROMPT_MODIFIER=(2-1-cu121) '
+++ CONDA_PROMPT_MODIFIER='(2-1-cu121) '
+++ export CONDA_PREFIX_3=/fsx/nouamane/miniconda
+++ CONDA_PREFIX_3=/fsx/nouamane/miniconda
+++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ export _CE_M=
+++ _CE_M=
+++ export _CE_CONDA=
+++ _CE_CONDA=
+++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh
++++ test -n ''
++++ xml_catalog_files_libxml2=
++++ XML_CATALOG_FILES=
++++ conda_catalog_files=
++++ ifs_libxml2=' 	
+'
++++ IFS=' '
++++ rem=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ for pre in ${rem}
++++ test '' = /fsx/nouamane/miniconda/envs/2-1-cu121
++++ conda_catalog_files=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ rem=
++++ IFS=' 	
+'
++++ conda_catalog_files='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ export 'XML_CATALOG_FILES=file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ XML_CATALOG_FILES='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ unset conda_catalog_files ifs_libxml2 rem
++ __conda_hashr
++ '[' -n '' ']'
++ '[' -n '' ']'
++ hash -r
++ export PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++ PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ scontrol show hostnames ip-26-0-169-207
++ export NODELIST=ip-26-0-169-207
++ NODELIST=ip-26-0-169-207
+++ scontrol show hostnames ip-26-0-169-207
+++ head -n1
++ export MASTER_NODE=ip-26-0-169-207
++ MASTER_NODE=ip-26-0-169-207
++ export MASTER_PORT=12356
++ MASTER_PORT=12356
++ export NNODES=1
++ NNODES=1
++ export GPUS_PER_NODE=8
++ GPUS_PER_NODE=8
++ export WORLD_SIZE=8
++ WORLD_SIZE=8
++ export CUDA_DEVICE_MAX_CONNECTIONS=1
++ CUDA_DEVICE_MAX_CONNECTIONS=1
++ export NCCL_DEBUG=INFO
++ NCCL_DEBUG=INFO
++ export NANOTRON_BENCHMARK=1
++ NANOTRON_BENCHMARK=1
++ echo 'Master node: ip-26-0-169-207'
+Master node: ip-26-0-169-207
++ echo 'All nodes: ip-26-0-169-207'
+All nodes: ip-26-0-169-207
++ echo 'World size: 8'
+World size: 8
++ srun torchrun --nnodes=1 --nproc_per_node=8 --rdzv_id=13402229 --rdzv_backend=c10d --rdzv_endpoint=ip-26-0-169-207:12356 run_train.py --config-file benchmark/configs/config_1.14G_dp4_tp2_pp1_acc256_mbs2_seq2048_zero1_tpmodeRED_vocab32k.yaml
+[2024-12-13 01:49:19,737] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-13 01:49:19,737] torch.distributed.run: [WARNING] 
+[2024-12-13 01:49:19,737] torch.distributed.run: [WARNING] *****************************************
+[2024-12-13 01:49:19,737] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-13 01:49:19,737] torch.distributed.run: [WARNING] *****************************************
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+[2024-12-13 01:49:24,883] torch.distributed.elastic.multiprocessing.api: [ERROR] failed (exitcode: 2) local_rank: 0 (pid: 325848) of binary: /fsx/nouamane/miniconda/envs/2-1-cu121/bin/python
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 33, in <module>
+    sys.exit(load_entry_point('torch==2.1.1', 'console_scripts', 'torchrun')())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 264, in launch_agent
+    raise ChildFailedError(
+torch.distributed.elastic.multiprocessing.errors.ChildFailedError: 
+============================================================
+run_train.py FAILED
+------------------------------------------------------------
+Failures:
+[1]:
+  time      : 2024-12-13_01:49:24
+  host      : ip-26-0-169-207.ec2.internal
+  rank      : 1 (local_rank: 1)
+  exitcode  : 2 (pid: 325849)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[2]:
+  time      : 2024-12-13_01:49:24
+  host      : ip-26-0-169-207.ec2.internal
+  rank      : 2 (local_rank: 2)
+  exitcode  : 2 (pid: 325850)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[3]:
+  time      : 2024-12-13_01:49:24
+  host      : ip-26-0-169-207.ec2.internal
+  rank      : 3 (local_rank: 3)
+  exitcode  : 2 (pid: 325851)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[4]:
+  time      : 2024-12-13_01:49:24
+  host      : ip-26-0-169-207.ec2.internal
+  rank      : 4 (local_rank: 4)
+  exitcode  : 2 (pid: 325852)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[5]:
+  time      : 2024-12-13_01:49:24
+  host      : ip-26-0-169-207.ec2.internal
+  rank      : 5 (local_rank: 5)
+  exitcode  : 2 (pid: 325853)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[6]:
+  time      : 2024-12-13_01:49:24
+  host      : ip-26-0-169-207.ec2.internal
+  rank      : 6 (local_rank: 6)
+  exitcode  : 2 (pid: 325854)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[7]:
+  time      : 2024-12-13_01:49:24
+  host      : ip-26-0-169-207.ec2.internal
+  rank      : 7 (local_rank: 7)
+  exitcode  : 2 (pid: 325855)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+------------------------------------------------------------
+Root Cause (first observed failure):
+[0]:
+  time      : 2024-12-13_01:49:24
+  host      : ip-26-0-169-207.ec2.internal
+  rank      : 0 (local_rank: 0)
+  exitcode  : 2 (pid: 325848)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+============================================================
+srun: error: ip-26-0-169-207: task 0: Exited with exit code 1
diff --git a/logs/13402337-bench_stress_test.out b/logs/13402337-bench_stress_test.out
new file mode 100644
index 0000000000000000000000000000000000000000..4ddbdf26e2aca213e557cfc859a1d9dbaeeb73d4
--- /dev/null
+++ b/logs/13402337-bench_stress_test.out
@@ -0,0 +1,518 @@
++ source /etc/profile.d/modules.sh
+++ . /usr/share/modules/init/bash
++++ unset _mlshdbg
++++ '[' 0 = 1 ']'
++++ unset _mlre _mlIFS
++++ '[' -n x ']'
++++ _mlIFS=' 	
+'
++++ IFS=' '
++++ '[' -n '' ']'
+++++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash autoinit
++++ _mlcode='module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+MODULEPATH_modshare=/etc/environment-modules/modules:1:/usr/share/modules/\$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1; export MODULEPATH_modshare;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+LOADEDMODULES=; export LOADEDMODULES;
+MODULEPATH=/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/\$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles; export MODULEPATH;
+test 0;'
++++ _mlret=0
++++ '[' -n x ']'
++++ IFS=' 	
+'
++++ unset _mlIFS
++++ unset _mlre _mlv _mlrv
++++ '[' 0 -eq 0 ']'
++++ eval 'module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+MODULEPATH_modshare=/etc/environment-modules/modules:1:/usr/share/modules/\$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1; export MODULEPATH_modshare;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+LOADEDMODULES=; export LOADEDMODULES;
+MODULEPATH=/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/\$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles; export MODULEPATH;
+test 0;'
+++++ MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl
+++++ export MODULES_CMD
+++++ ENV=/usr/share/modules/init/profile.sh
+++++ export ENV
+++++ MODULEPATH_modshare='/etc/environment-modules/modules:1:/usr/share/modules/$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1'
+++++ export MODULEPATH_modshare
+++++ BASH_ENV=/usr/share/modules/init/bash
+++++ export BASH_ENV
+++++ MODULESHOME=/usr/share/modules
+++++ export MODULESHOME
+++++ LOADEDMODULES=
+++++ export LOADEDMODULES
+++++ MODULEPATH='/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles'
+++++ export MODULEPATH
+++++ test 0
++++ '[' 0 = 1 ']'
++++ '[' -t 2 ']'
++++ export -f module
++++ export -f switchml
++++ '[' 5 -ge 3 ']'
++++ [[ ehxB =~ i ]]
++++ [[ ! :/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin: =~ :/usr/bin: ]]
++++ '[' '!' -n '' ']'
++++ MANPATH=:
++++ export MANPATH
+++++ manpath
++++ [[ ! :/admin/home/nouamane/.local/share/man:/fsx/nouamane/miniconda/envs/2-1-cu121/man:/fsx/nouamane/miniconda/envs/2-1-cu121/share/man:/fsx/nouamane/miniconda/man:/fsx/nouamane/miniconda/share/man:/opt/amazon/openmpi/share/man:/opt/amazon/efa/share/man:/opt/slurm/share/man:/usr/local/man:/usr/local/share/man:/usr/share/man:/admin/home/nouamane/.fzf/man:: =~ :/usr/share/man: ]]
++++ unset _mlcode _mlret
++++ '[' -n '' ']'
++ module load cuda/12.1
++ unset _mlshdbg
++ '[' 0 = 1 ']'
++ unset _mlre _mlIFS
++ '[' -n x ']'
++ _mlIFS=' 	
+'
++ IFS=' '
++ '[' -n '' ']'
+++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash load cuda/12.1
++ eval 'CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include;' export 'CPATH;
+LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:;' export 'LD_LIBRARY_PATH;
+FI_EFA_FORK_SAFE=1;' export 'FI_EFA_FORK_SAFE;
+MANPATH=/usr/local/cuda-12.1/share/man::;' export 'MANPATH;
+LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64;' export 'LIBRARY_PATH;
+_LMFILES_=/usr/share/modules/modulefiles/cuda/12.1;' export '_LMFILES_;
+LOADEDMODULES=cuda/12.1;' export 'LOADEDMODULES;
+MPI_PATH=/opt/amazon/openmpi;' export 'MPI_PATH;
+NCCL_HOME_modshare=/opt/nccl/build:1;' export 'NCCL_HOME_modshare;
+NCCL_PROTO=simple;' export 'NCCL_PROTO;
+MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1;' export 'MANPATH_modshare;
+LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1;' export 'LIBRARY_PATH_modshare;
+NCCL_SOCKET_IFNAME=enp;' export 'NCCL_SOCKET_IFNAME;
+AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl;' export 'AWS_OFI_NCCL_HOME;
+NCCL_HOME=/opt/nccl/build;' export 'NCCL_HOME;
+FI_PROVIDER=efa;' export 'FI_PROVIDER;
+AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1;' export 'AWS_OFI_NCCL_HOME_modshare;
+CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1;' export 'CPATH_modshare;
+LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2::1:/opt/nccl/build/lib:1:/opt/aws-ofi-nccl/lib:2:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/efa/lib:1;' export 'LD_LIBRARY_PATH_modshare;
+FI_EFA_ENABLE_SHM_TRANSFER=1;' export 'FI_EFA_ENABLE_SHM_TRANSFER;
+_LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1;' export '_LMFILES__modshare;
+LOADEDMODULES_modshare=cuda/12.1:1;' export 'LOADEDMODULES_modshare;
+MPI_PATH_modshare=/opt/amazon/openmpi:1;' export 'MPI_PATH_modshare;
+PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin;' export 'PATH;
+CUDA_HOME=/usr/local/cuda-12.1;' export 'CUDA_HOME;
+PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/usr/local/cuda-12.1/include:1:/opt/amazon/efa/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/usr/sbin:1:/fsx/nouamane/miniconda/bin:1:/opt/slurm/bin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:2:/opt/amazon/openmpi/bin:1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1;' export 'PATH_modshare;
+test' '0;'
+++ CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include
+++ export CPATH
+++ LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:
+++ export LD_LIBRARY_PATH
+++ FI_EFA_FORK_SAFE=1
+++ export FI_EFA_FORK_SAFE
+++ MANPATH=/usr/local/cuda-12.1/share/man::
+++ export MANPATH
+++ LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64
+++ export LIBRARY_PATH
+++ _LMFILES_=/usr/share/modules/modulefiles/cuda/12.1
+++ export _LMFILES_
+++ LOADEDMODULES=cuda/12.1
+++ export LOADEDMODULES
+++ MPI_PATH=/opt/amazon/openmpi
+++ export MPI_PATH
+++ NCCL_HOME_modshare=/opt/nccl/build:1
+++ export NCCL_HOME_modshare
+++ NCCL_PROTO=simple
+++ export NCCL_PROTO
+++ MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1
+++ export MANPATH_modshare
+++ LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1
+++ export LIBRARY_PATH_modshare
+++ NCCL_SOCKET_IFNAME=enp
+++ export NCCL_SOCKET_IFNAME
+++ AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl
+++ export AWS_OFI_NCCL_HOME
+++ NCCL_HOME=/opt/nccl/build
+++ export NCCL_HOME
+++ FI_PROVIDER=efa
+++ export FI_PROVIDER
+++ AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1
+++ export AWS_OFI_NCCL_HOME_modshare
+++ CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1
+++ export CPATH_modshare
+++ LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2::1:/opt/nccl/build/lib:1:/opt/aws-ofi-nccl/lib:2:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/efa/lib:1
+++ export LD_LIBRARY_PATH_modshare
+++ FI_EFA_ENABLE_SHM_TRANSFER=1
+++ export FI_EFA_ENABLE_SHM_TRANSFER
+++ _LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1
+++ export _LMFILES__modshare
+++ LOADEDMODULES_modshare=cuda/12.1:1
+++ export LOADEDMODULES_modshare
+++ MPI_PATH_modshare=/opt/amazon/openmpi:1
+++ export MPI_PATH_modshare
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ export PATH
+++ CUDA_HOME=/usr/local/cuda-12.1
+++ export CUDA_HOME
+++ PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/usr/local/cuda-12.1/include:1:/opt/amazon/efa/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/usr/sbin:1:/fsx/nouamane/miniconda/bin:1:/opt/slurm/bin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:2:/opt/amazon/openmpi/bin:1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1
+++ export PATH_modshare
+++ test 0
++ _mlstatus=0
++ '[' -n x ']'
++ IFS=' 	
+'
++ unset _mlre _mlv _mlrv _mlIFS
++ '[' -n '' ']'
++ unset _mlshdbg
++ return 0
++ source /fsx/nouamane/miniconda/bin/activate
+++ _CONDA_ROOT=/fsx/nouamane/miniconda
+++ . /fsx/nouamane/miniconda/etc/profile.d/conda.sh
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ '[' -z x ']'
+++ conda activate
+++ local cmd=activate
+++ case "$cmd" in
+++ __conda_activate activate
+++ '[' -n '' ']'
+++ local ask_conda
++++ PS1=
++++ __conda_exe shell.posix activate
++++ /fsx/nouamane/miniconda/bin/conda shell.posix activate
+++ ask_conda='. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_2='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\'''
+++ eval '. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_2='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\'''
++++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh
+++++ test -n ''
+++++ unset XML_CATALOG_FILES
+++++ unset xml_catalog_files_libxml2
++++ PS1='(base) '
++++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++++ export CONDA_PREFIX=/fsx/nouamane/miniconda
++++ CONDA_PREFIX=/fsx/nouamane/miniconda
++++ export CONDA_SHLVL=3
++++ CONDA_SHLVL=3
++++ export CONDA_DEFAULT_ENV=base
++++ CONDA_DEFAULT_ENV=base
++++ export 'CONDA_PROMPT_MODIFIER=(base) '
++++ CONDA_PROMPT_MODIFIER='(base) '
++++ export CONDA_PREFIX_2=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ CONDA_PREFIX_2=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ __conda_hashr
+++ '[' -n '' ']'
+++ '[' -n '' ']'
+++ hash -r
++ conda activate 2-1-cu121
++ local cmd=activate
++ case "$cmd" in
++ __conda_activate activate 2-1-cu121
++ '[' -n '' ']'
++ local ask_conda
+++ PS1='(base) '
+++ __conda_exe shell.posix activate 2-1-cu121
+++ /fsx/nouamane/miniconda/bin/conda shell.posix activate 2-1-cu121
++ ask_conda='PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''4'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_3='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
++ eval 'PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''4'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_3='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
+++ PS1='(2-1-cu121) '
+++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ export CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ export CONDA_SHLVL=4
+++ CONDA_SHLVL=4
+++ export CONDA_DEFAULT_ENV=2-1-cu121
+++ CONDA_DEFAULT_ENV=2-1-cu121
+++ export 'CONDA_PROMPT_MODIFIER=(2-1-cu121) '
+++ CONDA_PROMPT_MODIFIER='(2-1-cu121) '
+++ export CONDA_PREFIX_3=/fsx/nouamane/miniconda
+++ CONDA_PREFIX_3=/fsx/nouamane/miniconda
+++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ export _CE_M=
+++ _CE_M=
+++ export _CE_CONDA=
+++ _CE_CONDA=
+++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh
++++ test -n ''
++++ xml_catalog_files_libxml2=
++++ XML_CATALOG_FILES=
++++ conda_catalog_files=
++++ ifs_libxml2=' 	
+'
++++ IFS=' '
++++ rem=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ for pre in ${rem}
++++ test '' = /fsx/nouamane/miniconda/envs/2-1-cu121
++++ conda_catalog_files=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ rem=
++++ IFS=' 	
+'
++++ conda_catalog_files='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ export 'XML_CATALOG_FILES=file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ XML_CATALOG_FILES='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ unset conda_catalog_files ifs_libxml2 rem
++ __conda_hashr
++ '[' -n '' ']'
++ '[' -n '' ']'
++ hash -r
++ export PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++ PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ scontrol show hostnames ip-26-0-169-207
++ export NODELIST=ip-26-0-169-207
++ NODELIST=ip-26-0-169-207
+++ scontrol show hostnames ip-26-0-169-207
+++ head -n1
++ export MASTER_NODE=ip-26-0-169-207
++ MASTER_NODE=ip-26-0-169-207
++ export MASTER_PORT=12356
++ MASTER_PORT=12356
++ export NNODES=1
++ NNODES=1
++ export GPUS_PER_NODE=8
++ GPUS_PER_NODE=8
++ export WORLD_SIZE=8
++ WORLD_SIZE=8
++ export CUDA_DEVICE_MAX_CONNECTIONS=1
++ CUDA_DEVICE_MAX_CONNECTIONS=1
++ export NCCL_DEBUG=INFO
++ NCCL_DEBUG=INFO
++ export NANOTRON_BENCHMARK=1
++ NANOTRON_BENCHMARK=1
++ echo 'Master node: ip-26-0-169-207'
+Master node: ip-26-0-169-207
++ echo 'All nodes: ip-26-0-169-207'
+All nodes: ip-26-0-169-207
++ echo 'World size: 8'
+World size: 8
++ srun torchrun --nnodes=1 --nproc_per_node=8 --rdzv_id=13402337 --rdzv_backend=c10d --rdzv_endpoint=ip-26-0-169-207:12356 run_train.py --config-file benchmark/configs/config_1.14G_dp4_tp2_pp1_acc256_mbs2_seq2048_zero1_tpmodeRED_vocab32k.yaml
+[2024-12-13 02:07:27,475] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-13 02:07:27,475] torch.distributed.run: [WARNING] 
+[2024-12-13 02:07:27,475] torch.distributed.run: [WARNING] *****************************************
+[2024-12-13 02:07:27,475] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-13 02:07:27,475] torch.distributed.run: [WARNING] *****************************************
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+[2024-12-13 02:07:32,774] torch.distributed.elastic.multiprocessing.api: [ERROR] failed (exitcode: 2) local_rank: 0 (pid: 342940) of binary: /fsx/nouamane/miniconda/envs/2-1-cu121/bin/python
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 33, in <module>
+    sys.exit(load_entry_point('torch==2.1.1', 'console_scripts', 'torchrun')())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 264, in launch_agent
+    raise ChildFailedError(
+torch.distributed.elastic.multiprocessing.errors.ChildFailedError: 
+============================================================
+run_train.py FAILED
+------------------------------------------------------------
+Failures:
+[1]:
+  time      : 2024-12-13_02:07:32
+  host      : ip-26-0-169-207.ec2.internal
+  rank      : 1 (local_rank: 1)
+  exitcode  : 2 (pid: 342941)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[2]:
+  time      : 2024-12-13_02:07:32
+  host      : ip-26-0-169-207.ec2.internal
+  rank      : 2 (local_rank: 2)
+  exitcode  : 2 (pid: 342942)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[3]:
+  time      : 2024-12-13_02:07:32
+  host      : ip-26-0-169-207.ec2.internal
+  rank      : 3 (local_rank: 3)
+  exitcode  : 2 (pid: 342943)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[4]:
+  time      : 2024-12-13_02:07:32
+  host      : ip-26-0-169-207.ec2.internal
+  rank      : 4 (local_rank: 4)
+  exitcode  : 2 (pid: 342944)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[5]:
+  time      : 2024-12-13_02:07:32
+  host      : ip-26-0-169-207.ec2.internal
+  rank      : 5 (local_rank: 5)
+  exitcode  : 2 (pid: 342945)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[6]:
+  time      : 2024-12-13_02:07:32
+  host      : ip-26-0-169-207.ec2.internal
+  rank      : 6 (local_rank: 6)
+  exitcode  : 2 (pid: 342946)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[7]:
+  time      : 2024-12-13_02:07:32
+  host      : ip-26-0-169-207.ec2.internal
+  rank      : 7 (local_rank: 7)
+  exitcode  : 2 (pid: 342947)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+------------------------------------------------------------
+Root Cause (first observed failure):
+[0]:
+  time      : 2024-12-13_02:07:32
+  host      : ip-26-0-169-207.ec2.internal
+  rank      : 0 (local_rank: 0)
+  exitcode  : 2 (pid: 342940)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+============================================================
+srun: error: ip-26-0-169-207: task 0: Exited with exit code 1
diff --git a/logs/13402368-bench_stress_test.out b/logs/13402368-bench_stress_test.out
new file mode 100644
index 0000000000000000000000000000000000000000..a5aca51b6cb33f6dafeeb70febff2d7a32b57709
--- /dev/null
+++ b/logs/13402368-bench_stress_test.out
@@ -0,0 +1,518 @@
++ source /etc/profile.d/modules.sh
+++ . /usr/share/modules/init/bash
++++ unset _mlshdbg
++++ '[' 0 = 1 ']'
++++ unset _mlre _mlIFS
++++ '[' -n x ']'
++++ _mlIFS=' 	
+'
++++ IFS=' '
++++ '[' -n '' ']'
+++++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash autoinit
++++ _mlcode='module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+MODULEPATH_modshare=/etc/environment-modules/modules:1:/usr/share/modules/\$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1; export MODULEPATH_modshare;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+LOADEDMODULES=; export LOADEDMODULES;
+MODULEPATH=/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/\$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles; export MODULEPATH;
+test 0;'
++++ _mlret=0
++++ '[' -n x ']'
++++ IFS=' 	
+'
++++ unset _mlIFS
++++ unset _mlre _mlv _mlrv
++++ '[' 0 -eq 0 ']'
++++ eval 'module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+MODULEPATH_modshare=/etc/environment-modules/modules:1:/usr/share/modules/\$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1; export MODULEPATH_modshare;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+LOADEDMODULES=; export LOADEDMODULES;
+MODULEPATH=/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/\$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles; export MODULEPATH;
+test 0;'
+++++ MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl
+++++ export MODULES_CMD
+++++ ENV=/usr/share/modules/init/profile.sh
+++++ export ENV
+++++ MODULEPATH_modshare='/etc/environment-modules/modules:1:/usr/share/modules/$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1'
+++++ export MODULEPATH_modshare
+++++ BASH_ENV=/usr/share/modules/init/bash
+++++ export BASH_ENV
+++++ MODULESHOME=/usr/share/modules
+++++ export MODULESHOME
+++++ LOADEDMODULES=
+++++ export LOADEDMODULES
+++++ MODULEPATH='/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles'
+++++ export MODULEPATH
+++++ test 0
++++ '[' 0 = 1 ']'
++++ '[' -t 2 ']'
++++ export -f module
++++ export -f switchml
++++ '[' 5 -ge 3 ']'
++++ [[ ehxB =~ i ]]
++++ [[ ! :/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin: =~ :/usr/bin: ]]
++++ '[' '!' -n '' ']'
++++ MANPATH=:
++++ export MANPATH
+++++ manpath
++++ [[ ! :/admin/home/nouamane/.local/share/man:/fsx/nouamane/miniconda/envs/2-1-cu121/man:/fsx/nouamane/miniconda/envs/2-1-cu121/share/man:/fsx/nouamane/miniconda/man:/fsx/nouamane/miniconda/share/man:/opt/amazon/openmpi/share/man:/opt/amazon/efa/share/man:/opt/slurm/share/man:/usr/local/man:/usr/local/share/man:/usr/share/man:/admin/home/nouamane/.fzf/man:: =~ :/usr/share/man: ]]
++++ unset _mlcode _mlret
++++ '[' -n '' ']'
++ module load cuda/12.1
++ unset _mlshdbg
++ '[' 0 = 1 ']'
++ unset _mlre _mlIFS
++ '[' -n x ']'
++ _mlIFS=' 	
+'
++ IFS=' '
++ '[' -n '' ']'
+++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash load cuda/12.1
++ eval 'CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include;' export 'CPATH;
+LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:;' export 'LD_LIBRARY_PATH;
+FI_EFA_FORK_SAFE=1;' export 'FI_EFA_FORK_SAFE;
+MANPATH=/usr/local/cuda-12.1/share/man::;' export 'MANPATH;
+LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64;' export 'LIBRARY_PATH;
+_LMFILES_=/usr/share/modules/modulefiles/cuda/12.1;' export '_LMFILES_;
+LOADEDMODULES=cuda/12.1;' export 'LOADEDMODULES;
+MPI_PATH=/opt/amazon/openmpi;' export 'MPI_PATH;
+NCCL_HOME_modshare=/opt/nccl/build:1;' export 'NCCL_HOME_modshare;
+NCCL_PROTO=simple;' export 'NCCL_PROTO;
+MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1;' export 'MANPATH_modshare;
+LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1;' export 'LIBRARY_PATH_modshare;
+NCCL_SOCKET_IFNAME=enp;' export 'NCCL_SOCKET_IFNAME;
+AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl;' export 'AWS_OFI_NCCL_HOME;
+NCCL_HOME=/opt/nccl/build;' export 'NCCL_HOME;
+FI_PROVIDER=efa;' export 'FI_PROVIDER;
+AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1;' export 'AWS_OFI_NCCL_HOME_modshare;
+CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1;' export 'CPATH_modshare;
+LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2::1:/opt/nccl/build/lib:1:/opt/aws-ofi-nccl/lib:2:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/efa/lib:1;' export 'LD_LIBRARY_PATH_modshare;
+FI_EFA_ENABLE_SHM_TRANSFER=1;' export 'FI_EFA_ENABLE_SHM_TRANSFER;
+_LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1;' export '_LMFILES__modshare;
+LOADEDMODULES_modshare=cuda/12.1:1;' export 'LOADEDMODULES_modshare;
+MPI_PATH_modshare=/opt/amazon/openmpi:1;' export 'MPI_PATH_modshare;
+PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin;' export 'PATH;
+CUDA_HOME=/usr/local/cuda-12.1;' export 'CUDA_HOME;
+PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/usr/local/cuda-12.1/include:1:/opt/amazon/efa/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/usr/sbin:1:/fsx/nouamane/miniconda/bin:1:/opt/slurm/bin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:2:/opt/amazon/openmpi/bin:1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1;' export 'PATH_modshare;
+test' '0;'
+++ CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include
+++ export CPATH
+++ LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:
+++ export LD_LIBRARY_PATH
+++ FI_EFA_FORK_SAFE=1
+++ export FI_EFA_FORK_SAFE
+++ MANPATH=/usr/local/cuda-12.1/share/man::
+++ export MANPATH
+++ LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64
+++ export LIBRARY_PATH
+++ _LMFILES_=/usr/share/modules/modulefiles/cuda/12.1
+++ export _LMFILES_
+++ LOADEDMODULES=cuda/12.1
+++ export LOADEDMODULES
+++ MPI_PATH=/opt/amazon/openmpi
+++ export MPI_PATH
+++ NCCL_HOME_modshare=/opt/nccl/build:1
+++ export NCCL_HOME_modshare
+++ NCCL_PROTO=simple
+++ export NCCL_PROTO
+++ MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1
+++ export MANPATH_modshare
+++ LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1
+++ export LIBRARY_PATH_modshare
+++ NCCL_SOCKET_IFNAME=enp
+++ export NCCL_SOCKET_IFNAME
+++ AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl
+++ export AWS_OFI_NCCL_HOME
+++ NCCL_HOME=/opt/nccl/build
+++ export NCCL_HOME
+++ FI_PROVIDER=efa
+++ export FI_PROVIDER
+++ AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1
+++ export AWS_OFI_NCCL_HOME_modshare
+++ CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1
+++ export CPATH_modshare
+++ LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2::1:/opt/nccl/build/lib:1:/opt/aws-ofi-nccl/lib:2:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/efa/lib:1
+++ export LD_LIBRARY_PATH_modshare
+++ FI_EFA_ENABLE_SHM_TRANSFER=1
+++ export FI_EFA_ENABLE_SHM_TRANSFER
+++ _LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1
+++ export _LMFILES__modshare
+++ LOADEDMODULES_modshare=cuda/12.1:1
+++ export LOADEDMODULES_modshare
+++ MPI_PATH_modshare=/opt/amazon/openmpi:1
+++ export MPI_PATH_modshare
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ export PATH
+++ CUDA_HOME=/usr/local/cuda-12.1
+++ export CUDA_HOME
+++ PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/usr/local/cuda-12.1/include:1:/opt/amazon/efa/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/usr/sbin:1:/fsx/nouamane/miniconda/bin:1:/opt/slurm/bin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:2:/opt/amazon/openmpi/bin:1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1
+++ export PATH_modshare
+++ test 0
++ _mlstatus=0
++ '[' -n x ']'
++ IFS=' 	
+'
++ unset _mlre _mlv _mlrv _mlIFS
++ '[' -n '' ']'
++ unset _mlshdbg
++ return 0
++ source /fsx/nouamane/miniconda/bin/activate
+++ _CONDA_ROOT=/fsx/nouamane/miniconda
+++ . /fsx/nouamane/miniconda/etc/profile.d/conda.sh
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ '[' -z x ']'
+++ conda activate
+++ local cmd=activate
+++ case "$cmd" in
+++ __conda_activate activate
+++ '[' -n '' ']'
+++ local ask_conda
++++ PS1=
++++ __conda_exe shell.posix activate
++++ /fsx/nouamane/miniconda/bin/conda shell.posix activate
+++ ask_conda='. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_2='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\'''
+++ eval '. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_2='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\'''
++++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh
+++++ test -n ''
+++++ unset XML_CATALOG_FILES
+++++ unset xml_catalog_files_libxml2
++++ PS1='(base) '
++++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++++ export CONDA_PREFIX=/fsx/nouamane/miniconda
++++ CONDA_PREFIX=/fsx/nouamane/miniconda
++++ export CONDA_SHLVL=3
++++ CONDA_SHLVL=3
++++ export CONDA_DEFAULT_ENV=base
++++ CONDA_DEFAULT_ENV=base
++++ export 'CONDA_PROMPT_MODIFIER=(base) '
++++ CONDA_PROMPT_MODIFIER='(base) '
++++ export CONDA_PREFIX_2=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ CONDA_PREFIX_2=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ __conda_hashr
+++ '[' -n '' ']'
+++ '[' -n '' ']'
+++ hash -r
++ conda activate 2-1-cu121
++ local cmd=activate
++ case "$cmd" in
++ __conda_activate activate 2-1-cu121
++ '[' -n '' ']'
++ local ask_conda
+++ PS1='(base) '
+++ __conda_exe shell.posix activate 2-1-cu121
+++ /fsx/nouamane/miniconda/bin/conda shell.posix activate 2-1-cu121
++ ask_conda='PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''4'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_3='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
++ eval 'PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''4'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_3='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
+++ PS1='(2-1-cu121) '
+++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ export CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ export CONDA_SHLVL=4
+++ CONDA_SHLVL=4
+++ export CONDA_DEFAULT_ENV=2-1-cu121
+++ CONDA_DEFAULT_ENV=2-1-cu121
+++ export 'CONDA_PROMPT_MODIFIER=(2-1-cu121) '
+++ CONDA_PROMPT_MODIFIER='(2-1-cu121) '
+++ export CONDA_PREFIX_3=/fsx/nouamane/miniconda
+++ CONDA_PREFIX_3=/fsx/nouamane/miniconda
+++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ export _CE_M=
+++ _CE_M=
+++ export _CE_CONDA=
+++ _CE_CONDA=
+++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh
++++ test -n ''
++++ xml_catalog_files_libxml2=
++++ XML_CATALOG_FILES=
++++ conda_catalog_files=
++++ ifs_libxml2=' 	
+'
++++ IFS=' '
++++ rem=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ for pre in ${rem}
++++ test '' = /fsx/nouamane/miniconda/envs/2-1-cu121
++++ conda_catalog_files=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ rem=
++++ IFS=' 	
+'
++++ conda_catalog_files='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ export 'XML_CATALOG_FILES=file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ XML_CATALOG_FILES='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ unset conda_catalog_files ifs_libxml2 rem
++ __conda_hashr
++ '[' -n '' ']'
++ '[' -n '' ']'
++ hash -r
++ export PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++ PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ scontrol show hostnames ip-26-0-160-103
++ export NODELIST=ip-26-0-160-103
++ NODELIST=ip-26-0-160-103
+++ scontrol show hostnames ip-26-0-160-103
+++ head -n1
++ export MASTER_NODE=ip-26-0-160-103
++ MASTER_NODE=ip-26-0-160-103
++ export MASTER_PORT=12356
++ MASTER_PORT=12356
++ export NNODES=1
++ NNODES=1
++ export GPUS_PER_NODE=8
++ GPUS_PER_NODE=8
++ export WORLD_SIZE=8
++ WORLD_SIZE=8
++ export CUDA_DEVICE_MAX_CONNECTIONS=1
++ CUDA_DEVICE_MAX_CONNECTIONS=1
++ export NCCL_DEBUG=INFO
++ NCCL_DEBUG=INFO
++ export NANOTRON_BENCHMARK=1
++ NANOTRON_BENCHMARK=1
++ echo 'Master node: ip-26-0-160-103'
+Master node: ip-26-0-160-103
++ echo 'All nodes: ip-26-0-160-103'
+All nodes: ip-26-0-160-103
++ echo 'World size: 8'
+World size: 8
++ srun torchrun --nnodes=1 --nproc_per_node=8 --rdzv_id=13402368 --rdzv_backend=c10d --rdzv_endpoint=ip-26-0-160-103:12356 run_train.py --config-file benchmark/configs/config_1.14G_dp4_tp2_pp1_acc256_mbs2_seq2048_zero1_tpmodeRED_vocab32k.yaml
+[2024-12-13 02:12:33,266] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-13 02:12:33,266] torch.distributed.run: [WARNING] 
+[2024-12-13 02:12:33,266] torch.distributed.run: [WARNING] *****************************************
+[2024-12-13 02:12:33,266] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-13 02:12:33,266] torch.distributed.run: [WARNING] *****************************************
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+[2024-12-13 02:12:38,420] torch.distributed.elastic.multiprocessing.api: [ERROR] failed (exitcode: 2) local_rank: 0 (pid: 2238309) of binary: /fsx/nouamane/miniconda/envs/2-1-cu121/bin/python
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 33, in <module>
+    sys.exit(load_entry_point('torch==2.1.1', 'console_scripts', 'torchrun')())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 264, in launch_agent
+    raise ChildFailedError(
+torch.distributed.elastic.multiprocessing.errors.ChildFailedError: 
+============================================================
+run_train.py FAILED
+------------------------------------------------------------
+Failures:
+[1]:
+  time      : 2024-12-13_02:12:38
+  host      : ip-26-0-160-103.ec2.internal
+  rank      : 1 (local_rank: 1)
+  exitcode  : 2 (pid: 2238310)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[2]:
+  time      : 2024-12-13_02:12:38
+  host      : ip-26-0-160-103.ec2.internal
+  rank      : 2 (local_rank: 2)
+  exitcode  : 2 (pid: 2238311)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[3]:
+  time      : 2024-12-13_02:12:38
+  host      : ip-26-0-160-103.ec2.internal
+  rank      : 3 (local_rank: 3)
+  exitcode  : 2 (pid: 2238312)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[4]:
+  time      : 2024-12-13_02:12:38
+  host      : ip-26-0-160-103.ec2.internal
+  rank      : 4 (local_rank: 4)
+  exitcode  : 2 (pid: 2238313)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[5]:
+  time      : 2024-12-13_02:12:38
+  host      : ip-26-0-160-103.ec2.internal
+  rank      : 5 (local_rank: 5)
+  exitcode  : 2 (pid: 2238314)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[6]:
+  time      : 2024-12-13_02:12:38
+  host      : ip-26-0-160-103.ec2.internal
+  rank      : 6 (local_rank: 6)
+  exitcode  : 2 (pid: 2238315)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[7]:
+  time      : 2024-12-13_02:12:38
+  host      : ip-26-0-160-103.ec2.internal
+  rank      : 7 (local_rank: 7)
+  exitcode  : 2 (pid: 2238316)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+------------------------------------------------------------
+Root Cause (first observed failure):
+[0]:
+  time      : 2024-12-13_02:12:38
+  host      : ip-26-0-160-103.ec2.internal
+  rank      : 0 (local_rank: 0)
+  exitcode  : 2 (pid: 2238309)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+============================================================
+srun: error: ip-26-0-160-103: task 0: Exited with exit code 1
diff --git a/logs/13402387-bench_stress_test.out b/logs/13402387-bench_stress_test.out
new file mode 100644
index 0000000000000000000000000000000000000000..9552fcc164d17dceb451a8f1b200eb40e8157661
--- /dev/null
+++ b/logs/13402387-bench_stress_test.out
@@ -0,0 +1,518 @@
++ source /etc/profile.d/modules.sh
+++ . /usr/share/modules/init/bash
++++ unset _mlshdbg
++++ '[' 0 = 1 ']'
++++ unset _mlre _mlIFS
++++ '[' -n x ']'
++++ _mlIFS=' 	
+'
++++ IFS=' '
++++ '[' -n '' ']'
+++++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash autoinit
++++ _mlcode='module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+MODULEPATH_modshare=/etc/environment-modules/modules:1:/usr/share/modules/\$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1; export MODULEPATH_modshare;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+LOADEDMODULES=; export LOADEDMODULES;
+MODULEPATH=/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/\$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles; export MODULEPATH;
+test 0;'
++++ _mlret=0
++++ '[' -n x ']'
++++ IFS=' 	
+'
++++ unset _mlIFS
++++ unset _mlre _mlv _mlrv
++++ '[' 0 -eq 0 ']'
++++ eval 'module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+MODULEPATH_modshare=/etc/environment-modules/modules:1:/usr/share/modules/\$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1; export MODULEPATH_modshare;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+LOADEDMODULES=; export LOADEDMODULES;
+MODULEPATH=/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/\$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles; export MODULEPATH;
+test 0;'
+++++ MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl
+++++ export MODULES_CMD
+++++ ENV=/usr/share/modules/init/profile.sh
+++++ export ENV
+++++ MODULEPATH_modshare='/etc/environment-modules/modules:1:/usr/share/modules/$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1'
+++++ export MODULEPATH_modshare
+++++ BASH_ENV=/usr/share/modules/init/bash
+++++ export BASH_ENV
+++++ MODULESHOME=/usr/share/modules
+++++ export MODULESHOME
+++++ LOADEDMODULES=
+++++ export LOADEDMODULES
+++++ MODULEPATH='/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles'
+++++ export MODULEPATH
+++++ test 0
++++ '[' 0 = 1 ']'
++++ '[' -t 2 ']'
++++ export -f module
++++ export -f switchml
++++ '[' 5 -ge 3 ']'
++++ [[ ehxB =~ i ]]
++++ [[ ! :/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin: =~ :/usr/bin: ]]
++++ '[' '!' -n '' ']'
++++ MANPATH=:
++++ export MANPATH
+++++ manpath
++++ [[ ! :/admin/home/nouamane/.local/share/man:/fsx/nouamane/miniconda/envs/2-1-cu121/man:/fsx/nouamane/miniconda/envs/2-1-cu121/share/man:/fsx/nouamane/miniconda/man:/fsx/nouamane/miniconda/share/man:/opt/amazon/openmpi/share/man:/opt/amazon/efa/share/man:/opt/slurm/share/man:/usr/local/man:/usr/local/share/man:/usr/share/man:/admin/home/nouamane/.fzf/man:: =~ :/usr/share/man: ]]
++++ unset _mlcode _mlret
++++ '[' -n '' ']'
++ module load cuda/12.1
++ unset _mlshdbg
++ '[' 0 = 1 ']'
++ unset _mlre _mlIFS
++ '[' -n x ']'
++ _mlIFS=' 	
+'
++ IFS=' '
++ '[' -n '' ']'
+++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash load cuda/12.1
++ eval 'CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include;' export 'CPATH;
+LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:;' export 'LD_LIBRARY_PATH;
+FI_EFA_FORK_SAFE=1;' export 'FI_EFA_FORK_SAFE;
+MANPATH=/usr/local/cuda-12.1/share/man::;' export 'MANPATH;
+LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64;' export 'LIBRARY_PATH;
+_LMFILES_=/usr/share/modules/modulefiles/cuda/12.1;' export '_LMFILES_;
+LOADEDMODULES=cuda/12.1;' export 'LOADEDMODULES;
+MPI_PATH=/opt/amazon/openmpi;' export 'MPI_PATH;
+NCCL_HOME_modshare=/opt/nccl/build:1;' export 'NCCL_HOME_modshare;
+NCCL_PROTO=simple;' export 'NCCL_PROTO;
+MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1;' export 'MANPATH_modshare;
+LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1;' export 'LIBRARY_PATH_modshare;
+NCCL_SOCKET_IFNAME=enp;' export 'NCCL_SOCKET_IFNAME;
+AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl;' export 'AWS_OFI_NCCL_HOME;
+NCCL_HOME=/opt/nccl/build;' export 'NCCL_HOME;
+FI_PROVIDER=efa;' export 'FI_PROVIDER;
+AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1;' export 'AWS_OFI_NCCL_HOME_modshare;
+CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1;' export 'CPATH_modshare;
+LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2::1:/opt/nccl/build/lib:1:/opt/aws-ofi-nccl/lib:2:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/efa/lib:1;' export 'LD_LIBRARY_PATH_modshare;
+FI_EFA_ENABLE_SHM_TRANSFER=1;' export 'FI_EFA_ENABLE_SHM_TRANSFER;
+_LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1;' export '_LMFILES__modshare;
+LOADEDMODULES_modshare=cuda/12.1:1;' export 'LOADEDMODULES_modshare;
+MPI_PATH_modshare=/opt/amazon/openmpi:1;' export 'MPI_PATH_modshare;
+PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin;' export 'PATH;
+CUDA_HOME=/usr/local/cuda-12.1;' export 'CUDA_HOME;
+PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/usr/local/cuda-12.1/include:1:/opt/amazon/efa/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/usr/sbin:1:/fsx/nouamane/miniconda/bin:1:/opt/slurm/bin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:2:/opt/amazon/openmpi/bin:1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1;' export 'PATH_modshare;
+test' '0;'
+++ CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include
+++ export CPATH
+++ LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:
+++ export LD_LIBRARY_PATH
+++ FI_EFA_FORK_SAFE=1
+++ export FI_EFA_FORK_SAFE
+++ MANPATH=/usr/local/cuda-12.1/share/man::
+++ export MANPATH
+++ LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64
+++ export LIBRARY_PATH
+++ _LMFILES_=/usr/share/modules/modulefiles/cuda/12.1
+++ export _LMFILES_
+++ LOADEDMODULES=cuda/12.1
+++ export LOADEDMODULES
+++ MPI_PATH=/opt/amazon/openmpi
+++ export MPI_PATH
+++ NCCL_HOME_modshare=/opt/nccl/build:1
+++ export NCCL_HOME_modshare
+++ NCCL_PROTO=simple
+++ export NCCL_PROTO
+++ MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1
+++ export MANPATH_modshare
+++ LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1
+++ export LIBRARY_PATH_modshare
+++ NCCL_SOCKET_IFNAME=enp
+++ export NCCL_SOCKET_IFNAME
+++ AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl
+++ export AWS_OFI_NCCL_HOME
+++ NCCL_HOME=/opt/nccl/build
+++ export NCCL_HOME
+++ FI_PROVIDER=efa
+++ export FI_PROVIDER
+++ AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1
+++ export AWS_OFI_NCCL_HOME_modshare
+++ CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1
+++ export CPATH_modshare
+++ LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2::1:/opt/nccl/build/lib:1:/opt/aws-ofi-nccl/lib:2:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/efa/lib:1
+++ export LD_LIBRARY_PATH_modshare
+++ FI_EFA_ENABLE_SHM_TRANSFER=1
+++ export FI_EFA_ENABLE_SHM_TRANSFER
+++ _LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1
+++ export _LMFILES__modshare
+++ LOADEDMODULES_modshare=cuda/12.1:1
+++ export LOADEDMODULES_modshare
+++ MPI_PATH_modshare=/opt/amazon/openmpi:1
+++ export MPI_PATH_modshare
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ export PATH
+++ CUDA_HOME=/usr/local/cuda-12.1
+++ export CUDA_HOME
+++ PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/usr/local/cuda-12.1/include:1:/opt/amazon/efa/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/usr/sbin:1:/fsx/nouamane/miniconda/bin:1:/opt/slurm/bin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:2:/opt/amazon/openmpi/bin:1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1
+++ export PATH_modshare
+++ test 0
++ _mlstatus=0
++ '[' -n x ']'
++ IFS=' 	
+'
++ unset _mlre _mlv _mlrv _mlIFS
++ '[' -n '' ']'
++ unset _mlshdbg
++ return 0
++ source /fsx/nouamane/miniconda/bin/activate
+++ _CONDA_ROOT=/fsx/nouamane/miniconda
+++ . /fsx/nouamane/miniconda/etc/profile.d/conda.sh
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ '[' -z x ']'
+++ conda activate
+++ local cmd=activate
+++ case "$cmd" in
+++ __conda_activate activate
+++ '[' -n '' ']'
+++ local ask_conda
++++ PS1=
++++ __conda_exe shell.posix activate
++++ /fsx/nouamane/miniconda/bin/conda shell.posix activate
+++ ask_conda='. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_2='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\'''
+++ eval '. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_2='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\'''
++++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh
+++++ test -n ''
+++++ unset XML_CATALOG_FILES
+++++ unset xml_catalog_files_libxml2
++++ PS1='(base) '
++++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++++ export CONDA_PREFIX=/fsx/nouamane/miniconda
++++ CONDA_PREFIX=/fsx/nouamane/miniconda
++++ export CONDA_SHLVL=3
++++ CONDA_SHLVL=3
++++ export CONDA_DEFAULT_ENV=base
++++ CONDA_DEFAULT_ENV=base
++++ export 'CONDA_PROMPT_MODIFIER=(base) '
++++ CONDA_PROMPT_MODIFIER='(base) '
++++ export CONDA_PREFIX_2=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ CONDA_PREFIX_2=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ __conda_hashr
+++ '[' -n '' ']'
+++ '[' -n '' ']'
+++ hash -r
++ conda activate 2-1-cu121
++ local cmd=activate
++ case "$cmd" in
++ __conda_activate activate 2-1-cu121
++ '[' -n '' ']'
++ local ask_conda
+++ PS1='(base) '
+++ __conda_exe shell.posix activate 2-1-cu121
+++ /fsx/nouamane/miniconda/bin/conda shell.posix activate 2-1-cu121
++ ask_conda='PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''4'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_3='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
++ eval 'PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''4'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_3='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
+++ PS1='(2-1-cu121) '
+++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ export CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ export CONDA_SHLVL=4
+++ CONDA_SHLVL=4
+++ export CONDA_DEFAULT_ENV=2-1-cu121
+++ CONDA_DEFAULT_ENV=2-1-cu121
+++ export 'CONDA_PROMPT_MODIFIER=(2-1-cu121) '
+++ CONDA_PROMPT_MODIFIER='(2-1-cu121) '
+++ export CONDA_PREFIX_3=/fsx/nouamane/miniconda
+++ CONDA_PREFIX_3=/fsx/nouamane/miniconda
+++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ export _CE_M=
+++ _CE_M=
+++ export _CE_CONDA=
+++ _CE_CONDA=
+++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh
++++ test -n ''
++++ xml_catalog_files_libxml2=
++++ XML_CATALOG_FILES=
++++ conda_catalog_files=
++++ ifs_libxml2=' 	
+'
++++ IFS=' '
++++ rem=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ for pre in ${rem}
++++ test '' = /fsx/nouamane/miniconda/envs/2-1-cu121
++++ conda_catalog_files=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ rem=
++++ IFS=' 	
+'
++++ conda_catalog_files='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ export 'XML_CATALOG_FILES=file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ XML_CATALOG_FILES='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ unset conda_catalog_files ifs_libxml2 rem
++ __conda_hashr
++ '[' -n '' ']'
++ '[' -n '' ']'
++ hash -r
++ export PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++ PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ scontrol show hostnames ip-26-0-169-207
++ export NODELIST=ip-26-0-169-207
++ NODELIST=ip-26-0-169-207
+++ scontrol show hostnames ip-26-0-169-207
+++ head -n1
++ export MASTER_NODE=ip-26-0-169-207
++ MASTER_NODE=ip-26-0-169-207
++ export MASTER_PORT=12356
++ MASTER_PORT=12356
++ export NNODES=1
++ NNODES=1
++ export GPUS_PER_NODE=8
++ GPUS_PER_NODE=8
++ export WORLD_SIZE=8
++ WORLD_SIZE=8
++ export CUDA_DEVICE_MAX_CONNECTIONS=1
++ CUDA_DEVICE_MAX_CONNECTIONS=1
++ export NCCL_DEBUG=INFO
++ NCCL_DEBUG=INFO
++ export NANOTRON_BENCHMARK=1
++ NANOTRON_BENCHMARK=1
++ echo 'Master node: ip-26-0-169-207'
+Master node: ip-26-0-169-207
++ echo 'All nodes: ip-26-0-169-207'
+All nodes: ip-26-0-169-207
++ echo 'World size: 8'
+World size: 8
++ srun torchrun --nnodes=1 --nproc_per_node=8 --rdzv_id=13402387 --rdzv_backend=c10d --rdzv_endpoint=ip-26-0-169-207:12356 run_train.py --config-file benchmark/configs/config_1.14G_dp4_tp2_pp1_acc256_mbs2_seq2048_zero1_tpmodeRED_vocab32k.yaml
+[2024-12-13 02:15:46,057] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-13 02:15:46,057] torch.distributed.run: [WARNING] 
+[2024-12-13 02:15:46,057] torch.distributed.run: [WARNING] *****************************************
+[2024-12-13 02:15:46,057] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-13 02:15:46,057] torch.distributed.run: [WARNING] *****************************************
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+[2024-12-13 02:15:51,395] torch.distributed.elastic.multiprocessing.api: [ERROR] failed (exitcode: 2) local_rank: 0 (pid: 350721) of binary: /fsx/nouamane/miniconda/envs/2-1-cu121/bin/python
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 33, in <module>
+    sys.exit(load_entry_point('torch==2.1.1', 'console_scripts', 'torchrun')())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 264, in launch_agent
+    raise ChildFailedError(
+torch.distributed.elastic.multiprocessing.errors.ChildFailedError: 
+============================================================
+run_train.py FAILED
+------------------------------------------------------------
+Failures:
+[1]:
+  time      : 2024-12-13_02:15:51
+  host      : ip-26-0-169-207.ec2.internal
+  rank      : 1 (local_rank: 1)
+  exitcode  : 2 (pid: 350722)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[2]:
+  time      : 2024-12-13_02:15:51
+  host      : ip-26-0-169-207.ec2.internal
+  rank      : 2 (local_rank: 2)
+  exitcode  : 2 (pid: 350723)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[3]:
+  time      : 2024-12-13_02:15:51
+  host      : ip-26-0-169-207.ec2.internal
+  rank      : 3 (local_rank: 3)
+  exitcode  : 2 (pid: 350724)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[4]:
+  time      : 2024-12-13_02:15:51
+  host      : ip-26-0-169-207.ec2.internal
+  rank      : 4 (local_rank: 4)
+  exitcode  : 2 (pid: 350725)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[5]:
+  time      : 2024-12-13_02:15:51
+  host      : ip-26-0-169-207.ec2.internal
+  rank      : 5 (local_rank: 5)
+  exitcode  : 2 (pid: 350726)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[6]:
+  time      : 2024-12-13_02:15:51
+  host      : ip-26-0-169-207.ec2.internal
+  rank      : 6 (local_rank: 6)
+  exitcode  : 2 (pid: 350727)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[7]:
+  time      : 2024-12-13_02:15:51
+  host      : ip-26-0-169-207.ec2.internal
+  rank      : 7 (local_rank: 7)
+  exitcode  : 2 (pid: 350728)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+------------------------------------------------------------
+Root Cause (first observed failure):
+[0]:
+  time      : 2024-12-13_02:15:51
+  host      : ip-26-0-169-207.ec2.internal
+  rank      : 0 (local_rank: 0)
+  exitcode  : 2 (pid: 350721)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+============================================================
+srun: error: ip-26-0-169-207: task 0: Exited with exit code 1
diff --git a/logs/13402490-bench_stress_test.out b/logs/13402490-bench_stress_test.out
new file mode 100644
index 0000000000000000000000000000000000000000..d78b58c72fd74a74eb9732b0eec249ac83a5ea35
--- /dev/null
+++ b/logs/13402490-bench_stress_test.out
@@ -0,0 +1,518 @@
++ source /etc/profile.d/modules.sh
+++ . /usr/share/modules/init/bash
++++ unset _mlshdbg
++++ '[' 0 = 1 ']'
++++ unset _mlre _mlIFS
++++ '[' -n x ']'
++++ _mlIFS=' 	
+'
++++ IFS=' '
++++ '[' -n '' ']'
+++++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash autoinit
++++ _mlcode='module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+MODULEPATH_modshare=/etc/environment-modules/modules:1:/usr/share/modules/\$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1; export MODULEPATH_modshare;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+LOADEDMODULES=; export LOADEDMODULES;
+MODULEPATH=/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/\$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles; export MODULEPATH;
+test 0;'
++++ _mlret=0
++++ '[' -n x ']'
++++ IFS=' 	
+'
++++ unset _mlIFS
++++ unset _mlre _mlv _mlrv
++++ '[' 0 -eq 0 ']'
++++ eval 'module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+MODULEPATH_modshare=/etc/environment-modules/modules:1:/usr/share/modules/\$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1; export MODULEPATH_modshare;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+LOADEDMODULES=; export LOADEDMODULES;
+MODULEPATH=/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/\$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles; export MODULEPATH;
+test 0;'
+++++ MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl
+++++ export MODULES_CMD
+++++ ENV=/usr/share/modules/init/profile.sh
+++++ export ENV
+++++ MODULEPATH_modshare='/etc/environment-modules/modules:1:/usr/share/modules/$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1'
+++++ export MODULEPATH_modshare
+++++ BASH_ENV=/usr/share/modules/init/bash
+++++ export BASH_ENV
+++++ MODULESHOME=/usr/share/modules
+++++ export MODULESHOME
+++++ LOADEDMODULES=
+++++ export LOADEDMODULES
+++++ MODULEPATH='/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles'
+++++ export MODULEPATH
+++++ test 0
++++ '[' 0 = 1 ']'
++++ '[' -t 2 ']'
++++ export -f module
++++ export -f switchml
++++ '[' 5 -ge 3 ']'
++++ [[ ehxB =~ i ]]
++++ [[ ! :/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin: =~ :/usr/bin: ]]
++++ '[' '!' -n '' ']'
++++ MANPATH=:
++++ export MANPATH
+++++ manpath
++++ [[ ! :/admin/home/nouamane/.local/share/man:/fsx/nouamane/miniconda/envs/2-1-cu121/man:/fsx/nouamane/miniconda/envs/2-1-cu121/share/man:/fsx/nouamane/miniconda/man:/fsx/nouamane/miniconda/share/man:/opt/amazon/openmpi/share/man:/opt/amazon/efa/share/man:/opt/slurm/share/man:/usr/local/man:/usr/local/share/man:/usr/share/man:/admin/home/nouamane/.fzf/man:: =~ :/usr/share/man: ]]
++++ unset _mlcode _mlret
++++ '[' -n '' ']'
++ module load cuda/12.1
++ unset _mlshdbg
++ '[' 0 = 1 ']'
++ unset _mlre _mlIFS
++ '[' -n x ']'
++ _mlIFS=' 	
+'
++ IFS=' '
++ '[' -n '' ']'
+++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash load cuda/12.1
++ eval 'CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include;' export 'CPATH;
+LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:;' export 'LD_LIBRARY_PATH;
+FI_EFA_FORK_SAFE=1;' export 'FI_EFA_FORK_SAFE;
+MANPATH=/usr/local/cuda-12.1/share/man::;' export 'MANPATH;
+LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64;' export 'LIBRARY_PATH;
+_LMFILES_=/usr/share/modules/modulefiles/cuda/12.1;' export '_LMFILES_;
+LOADEDMODULES=cuda/12.1;' export 'LOADEDMODULES;
+MPI_PATH=/opt/amazon/openmpi;' export 'MPI_PATH;
+NCCL_HOME_modshare=/opt/nccl/build:1;' export 'NCCL_HOME_modshare;
+NCCL_PROTO=simple;' export 'NCCL_PROTO;
+MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1;' export 'MANPATH_modshare;
+LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1;' export 'LIBRARY_PATH_modshare;
+NCCL_SOCKET_IFNAME=enp;' export 'NCCL_SOCKET_IFNAME;
+AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl;' export 'AWS_OFI_NCCL_HOME;
+NCCL_HOME=/opt/nccl/build;' export 'NCCL_HOME;
+FI_PROVIDER=efa;' export 'FI_PROVIDER;
+AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1;' export 'AWS_OFI_NCCL_HOME_modshare;
+CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1;' export 'CPATH_modshare;
+LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2::1:/opt/nccl/build/lib:1:/opt/aws-ofi-nccl/lib:2:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/efa/lib:1;' export 'LD_LIBRARY_PATH_modshare;
+FI_EFA_ENABLE_SHM_TRANSFER=1;' export 'FI_EFA_ENABLE_SHM_TRANSFER;
+_LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1;' export '_LMFILES__modshare;
+LOADEDMODULES_modshare=cuda/12.1:1;' export 'LOADEDMODULES_modshare;
+MPI_PATH_modshare=/opt/amazon/openmpi:1;' export 'MPI_PATH_modshare;
+PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin;' export 'PATH;
+CUDA_HOME=/usr/local/cuda-12.1;' export 'CUDA_HOME;
+PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/usr/local/cuda-12.1/include:1:/opt/amazon/efa/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/usr/sbin:1:/fsx/nouamane/miniconda/bin:1:/opt/slurm/bin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:2:/opt/amazon/openmpi/bin:1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1;' export 'PATH_modshare;
+test' '0;'
+++ CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include
+++ export CPATH
+++ LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:
+++ export LD_LIBRARY_PATH
+++ FI_EFA_FORK_SAFE=1
+++ export FI_EFA_FORK_SAFE
+++ MANPATH=/usr/local/cuda-12.1/share/man::
+++ export MANPATH
+++ LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64
+++ export LIBRARY_PATH
+++ _LMFILES_=/usr/share/modules/modulefiles/cuda/12.1
+++ export _LMFILES_
+++ LOADEDMODULES=cuda/12.1
+++ export LOADEDMODULES
+++ MPI_PATH=/opt/amazon/openmpi
+++ export MPI_PATH
+++ NCCL_HOME_modshare=/opt/nccl/build:1
+++ export NCCL_HOME_modshare
+++ NCCL_PROTO=simple
+++ export NCCL_PROTO
+++ MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1
+++ export MANPATH_modshare
+++ LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1
+++ export LIBRARY_PATH_modshare
+++ NCCL_SOCKET_IFNAME=enp
+++ export NCCL_SOCKET_IFNAME
+++ AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl
+++ export AWS_OFI_NCCL_HOME
+++ NCCL_HOME=/opt/nccl/build
+++ export NCCL_HOME
+++ FI_PROVIDER=efa
+++ export FI_PROVIDER
+++ AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1
+++ export AWS_OFI_NCCL_HOME_modshare
+++ CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1
+++ export CPATH_modshare
+++ LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2::1:/opt/nccl/build/lib:1:/opt/aws-ofi-nccl/lib:2:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/efa/lib:1
+++ export LD_LIBRARY_PATH_modshare
+++ FI_EFA_ENABLE_SHM_TRANSFER=1
+++ export FI_EFA_ENABLE_SHM_TRANSFER
+++ _LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1
+++ export _LMFILES__modshare
+++ LOADEDMODULES_modshare=cuda/12.1:1
+++ export LOADEDMODULES_modshare
+++ MPI_PATH_modshare=/opt/amazon/openmpi:1
+++ export MPI_PATH_modshare
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ export PATH
+++ CUDA_HOME=/usr/local/cuda-12.1
+++ export CUDA_HOME
+++ PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/usr/local/cuda-12.1/include:1:/opt/amazon/efa/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/usr/sbin:1:/fsx/nouamane/miniconda/bin:1:/opt/slurm/bin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:2:/opt/amazon/openmpi/bin:1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1
+++ export PATH_modshare
+++ test 0
++ _mlstatus=0
++ '[' -n x ']'
++ IFS=' 	
+'
++ unset _mlre _mlv _mlrv _mlIFS
++ '[' -n '' ']'
++ unset _mlshdbg
++ return 0
++ source /fsx/nouamane/miniconda/bin/activate
+++ _CONDA_ROOT=/fsx/nouamane/miniconda
+++ . /fsx/nouamane/miniconda/etc/profile.d/conda.sh
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ '[' -z x ']'
+++ conda activate
+++ local cmd=activate
+++ case "$cmd" in
+++ __conda_activate activate
+++ '[' -n '' ']'
+++ local ask_conda
++++ PS1=
++++ __conda_exe shell.posix activate
++++ /fsx/nouamane/miniconda/bin/conda shell.posix activate
+++ ask_conda='. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_2='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\'''
+++ eval '. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_2='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\'''
++++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh
+++++ test -n ''
+++++ unset XML_CATALOG_FILES
+++++ unset xml_catalog_files_libxml2
++++ PS1='(base) '
++++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++++ export CONDA_PREFIX=/fsx/nouamane/miniconda
++++ CONDA_PREFIX=/fsx/nouamane/miniconda
++++ export CONDA_SHLVL=3
++++ CONDA_SHLVL=3
++++ export CONDA_DEFAULT_ENV=base
++++ CONDA_DEFAULT_ENV=base
++++ export 'CONDA_PROMPT_MODIFIER=(base) '
++++ CONDA_PROMPT_MODIFIER='(base) '
++++ export CONDA_PREFIX_2=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ CONDA_PREFIX_2=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ __conda_hashr
+++ '[' -n '' ']'
+++ '[' -n '' ']'
+++ hash -r
++ conda activate 2-1-cu121
++ local cmd=activate
++ case "$cmd" in
++ __conda_activate activate 2-1-cu121
++ '[' -n '' ']'
++ local ask_conda
+++ PS1='(base) '
+++ __conda_exe shell.posix activate 2-1-cu121
+++ /fsx/nouamane/miniconda/bin/conda shell.posix activate 2-1-cu121
++ ask_conda='PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''4'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_3='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
++ eval 'PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''4'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_3='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
+++ PS1='(2-1-cu121) '
+++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ export CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ export CONDA_SHLVL=4
+++ CONDA_SHLVL=4
+++ export CONDA_DEFAULT_ENV=2-1-cu121
+++ CONDA_DEFAULT_ENV=2-1-cu121
+++ export 'CONDA_PROMPT_MODIFIER=(2-1-cu121) '
+++ CONDA_PROMPT_MODIFIER='(2-1-cu121) '
+++ export CONDA_PREFIX_3=/fsx/nouamane/miniconda
+++ CONDA_PREFIX_3=/fsx/nouamane/miniconda
+++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ export _CE_M=
+++ _CE_M=
+++ export _CE_CONDA=
+++ _CE_CONDA=
+++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh
++++ test -n ''
++++ xml_catalog_files_libxml2=
++++ XML_CATALOG_FILES=
++++ conda_catalog_files=
++++ ifs_libxml2=' 	
+'
++++ IFS=' '
++++ rem=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ for pre in ${rem}
++++ test '' = /fsx/nouamane/miniconda/envs/2-1-cu121
++++ conda_catalog_files=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ rem=
++++ IFS=' 	
+'
++++ conda_catalog_files='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ export 'XML_CATALOG_FILES=file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ XML_CATALOG_FILES='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ unset conda_catalog_files ifs_libxml2 rem
++ __conda_hashr
++ '[' -n '' ']'
++ '[' -n '' ']'
++ hash -r
++ export PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++ PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ scontrol show hostnames ip-26-0-169-207
++ export NODELIST=ip-26-0-169-207
++ NODELIST=ip-26-0-169-207
+++ scontrol show hostnames ip-26-0-169-207
+++ head -n1
++ export MASTER_NODE=ip-26-0-169-207
++ MASTER_NODE=ip-26-0-169-207
++ export MASTER_PORT=12356
++ MASTER_PORT=12356
++ export NNODES=1
++ NNODES=1
++ export GPUS_PER_NODE=8
++ GPUS_PER_NODE=8
++ export WORLD_SIZE=8
++ WORLD_SIZE=8
++ export CUDA_DEVICE_MAX_CONNECTIONS=1
++ CUDA_DEVICE_MAX_CONNECTIONS=1
++ export NCCL_DEBUG=INFO
++ NCCL_DEBUG=INFO
++ export NANOTRON_BENCHMARK=1
++ NANOTRON_BENCHMARK=1
++ echo 'Master node: ip-26-0-169-207'
+Master node: ip-26-0-169-207
++ echo 'All nodes: ip-26-0-169-207'
+All nodes: ip-26-0-169-207
++ echo 'World size: 8'
+World size: 8
++ srun torchrun --nnodes=1 --nproc_per_node=8 --rdzv_id=13402490 --rdzv_backend=c10d --rdzv_endpoint=ip-26-0-169-207:12356 run_train.py --config-file benchmark/configs/config_1.14G_dp4_tp2_pp1_acc256_mbs2_seq2048_zero1_tpmodeRED_vocab32k.yaml
+[2024-12-13 02:32:24,943] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-13 02:32:24,943] torch.distributed.run: [WARNING] 
+[2024-12-13 02:32:24,943] torch.distributed.run: [WARNING] *****************************************
+[2024-12-13 02:32:24,943] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-13 02:32:24,943] torch.distributed.run: [WARNING] *****************************************
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+[2024-12-13 02:32:29,997] torch.distributed.elastic.multiprocessing.api: [ERROR] failed (exitcode: 2) local_rank: 0 (pid: 366986) of binary: /fsx/nouamane/miniconda/envs/2-1-cu121/bin/python
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 33, in <module>
+    sys.exit(load_entry_point('torch==2.1.1', 'console_scripts', 'torchrun')())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 264, in launch_agent
+    raise ChildFailedError(
+torch.distributed.elastic.multiprocessing.errors.ChildFailedError: 
+============================================================
+run_train.py FAILED
+------------------------------------------------------------
+Failures:
+[1]:
+  time      : 2024-12-13_02:32:29
+  host      : ip-26-0-169-207.ec2.internal
+  rank      : 1 (local_rank: 1)
+  exitcode  : 2 (pid: 366987)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[2]:
+  time      : 2024-12-13_02:32:29
+  host      : ip-26-0-169-207.ec2.internal
+  rank      : 2 (local_rank: 2)
+  exitcode  : 2 (pid: 366988)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[3]:
+  time      : 2024-12-13_02:32:29
+  host      : ip-26-0-169-207.ec2.internal
+  rank      : 3 (local_rank: 3)
+  exitcode  : 2 (pid: 366989)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[4]:
+  time      : 2024-12-13_02:32:29
+  host      : ip-26-0-169-207.ec2.internal
+  rank      : 4 (local_rank: 4)
+  exitcode  : 2 (pid: 366990)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[5]:
+  time      : 2024-12-13_02:32:29
+  host      : ip-26-0-169-207.ec2.internal
+  rank      : 5 (local_rank: 5)
+  exitcode  : 2 (pid: 366991)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[6]:
+  time      : 2024-12-13_02:32:29
+  host      : ip-26-0-169-207.ec2.internal
+  rank      : 6 (local_rank: 6)
+  exitcode  : 2 (pid: 366992)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[7]:
+  time      : 2024-12-13_02:32:29
+  host      : ip-26-0-169-207.ec2.internal
+  rank      : 7 (local_rank: 7)
+  exitcode  : 2 (pid: 366993)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+------------------------------------------------------------
+Root Cause (first observed failure):
+[0]:
+  time      : 2024-12-13_02:32:29
+  host      : ip-26-0-169-207.ec2.internal
+  rank      : 0 (local_rank: 0)
+  exitcode  : 2 (pid: 366986)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+============================================================
+srun: error: ip-26-0-169-207: task 0: Exited with exit code 1
diff --git a/logs/13402561-bench_stress_test.out b/logs/13402561-bench_stress_test.out
new file mode 100644
index 0000000000000000000000000000000000000000..4bc8e1b49f0bf403adc46d924fd71a71773e618f
--- /dev/null
+++ b/logs/13402561-bench_stress_test.out
@@ -0,0 +1,518 @@
++ source /etc/profile.d/modules.sh
+++ . /usr/share/modules/init/bash
++++ unset _mlshdbg
++++ '[' 0 = 1 ']'
++++ unset _mlre _mlIFS
++++ '[' -n x ']'
++++ _mlIFS=' 	
+'
++++ IFS=' '
++++ '[' -n '' ']'
+++++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash autoinit
++++ _mlcode='module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+MODULEPATH_modshare=/etc/environment-modules/modules:1:/usr/share/modules/\$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1; export MODULEPATH_modshare;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+LOADEDMODULES=; export LOADEDMODULES;
+MODULEPATH=/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/\$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles; export MODULEPATH;
+test 0;'
++++ _mlret=0
++++ '[' -n x ']'
++++ IFS=' 	
+'
++++ unset _mlIFS
++++ unset _mlre _mlv _mlrv
++++ '[' 0 -eq 0 ']'
++++ eval 'module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+MODULEPATH_modshare=/etc/environment-modules/modules:1:/usr/share/modules/\$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1; export MODULEPATH_modshare;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+LOADEDMODULES=; export LOADEDMODULES;
+MODULEPATH=/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/\$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles; export MODULEPATH;
+test 0;'
+++++ MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl
+++++ export MODULES_CMD
+++++ ENV=/usr/share/modules/init/profile.sh
+++++ export ENV
+++++ MODULEPATH_modshare='/etc/environment-modules/modules:1:/usr/share/modules/$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1'
+++++ export MODULEPATH_modshare
+++++ BASH_ENV=/usr/share/modules/init/bash
+++++ export BASH_ENV
+++++ MODULESHOME=/usr/share/modules
+++++ export MODULESHOME
+++++ LOADEDMODULES=
+++++ export LOADEDMODULES
+++++ MODULEPATH='/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles'
+++++ export MODULEPATH
+++++ test 0
++++ '[' 0 = 1 ']'
++++ '[' -t 2 ']'
++++ export -f module
++++ export -f switchml
++++ '[' 5 -ge 3 ']'
++++ [[ ehxB =~ i ]]
++++ [[ ! :/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin: =~ :/usr/bin: ]]
++++ '[' '!' -n '' ']'
++++ MANPATH=:
++++ export MANPATH
+++++ manpath
++++ [[ ! :/admin/home/nouamane/.local/share/man:/fsx/nouamane/miniconda/envs/2-1-cu121/man:/fsx/nouamane/miniconda/envs/2-1-cu121/share/man:/fsx/nouamane/miniconda/man:/fsx/nouamane/miniconda/share/man:/opt/amazon/openmpi/share/man:/opt/amazon/efa/share/man:/opt/slurm/share/man:/usr/local/man:/usr/local/share/man:/usr/share/man:/admin/home/nouamane/.fzf/man:: =~ :/usr/share/man: ]]
++++ unset _mlcode _mlret
++++ '[' -n '' ']'
++ module load cuda/12.1
++ unset _mlshdbg
++ '[' 0 = 1 ']'
++ unset _mlre _mlIFS
++ '[' -n x ']'
++ _mlIFS=' 	
+'
++ IFS=' '
++ '[' -n '' ']'
+++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash load cuda/12.1
++ eval 'CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include;' export 'CPATH;
+LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:;' export 'LD_LIBRARY_PATH;
+FI_EFA_FORK_SAFE=1;' export 'FI_EFA_FORK_SAFE;
+MANPATH=/usr/local/cuda-12.1/share/man::;' export 'MANPATH;
+LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64;' export 'LIBRARY_PATH;
+_LMFILES_=/usr/share/modules/modulefiles/cuda/12.1;' export '_LMFILES_;
+LOADEDMODULES=cuda/12.1;' export 'LOADEDMODULES;
+MPI_PATH=/opt/amazon/openmpi;' export 'MPI_PATH;
+NCCL_HOME_modshare=/opt/nccl/build:1;' export 'NCCL_HOME_modshare;
+NCCL_PROTO=simple;' export 'NCCL_PROTO;
+MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1;' export 'MANPATH_modshare;
+LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1;' export 'LIBRARY_PATH_modshare;
+NCCL_SOCKET_IFNAME=enp;' export 'NCCL_SOCKET_IFNAME;
+AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl;' export 'AWS_OFI_NCCL_HOME;
+NCCL_HOME=/opt/nccl/build;' export 'NCCL_HOME;
+FI_PROVIDER=efa;' export 'FI_PROVIDER;
+AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1;' export 'AWS_OFI_NCCL_HOME_modshare;
+CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1;' export 'CPATH_modshare;
+LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2::1:/opt/nccl/build/lib:1:/opt/aws-ofi-nccl/lib:2:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/efa/lib:1;' export 'LD_LIBRARY_PATH_modshare;
+FI_EFA_ENABLE_SHM_TRANSFER=1;' export 'FI_EFA_ENABLE_SHM_TRANSFER;
+_LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1;' export '_LMFILES__modshare;
+LOADEDMODULES_modshare=cuda/12.1:1;' export 'LOADEDMODULES_modshare;
+MPI_PATH_modshare=/opt/amazon/openmpi:1;' export 'MPI_PATH_modshare;
+PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin;' export 'PATH;
+CUDA_HOME=/usr/local/cuda-12.1;' export 'CUDA_HOME;
+PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/usr/local/cuda-12.1/include:1:/opt/amazon/efa/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/usr/sbin:1:/fsx/nouamane/miniconda/bin:1:/opt/slurm/bin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:2:/opt/amazon/openmpi/bin:1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1;' export 'PATH_modshare;
+test' '0;'
+++ CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include
+++ export CPATH
+++ LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:
+++ export LD_LIBRARY_PATH
+++ FI_EFA_FORK_SAFE=1
+++ export FI_EFA_FORK_SAFE
+++ MANPATH=/usr/local/cuda-12.1/share/man::
+++ export MANPATH
+++ LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64
+++ export LIBRARY_PATH
+++ _LMFILES_=/usr/share/modules/modulefiles/cuda/12.1
+++ export _LMFILES_
+++ LOADEDMODULES=cuda/12.1
+++ export LOADEDMODULES
+++ MPI_PATH=/opt/amazon/openmpi
+++ export MPI_PATH
+++ NCCL_HOME_modshare=/opt/nccl/build:1
+++ export NCCL_HOME_modshare
+++ NCCL_PROTO=simple
+++ export NCCL_PROTO
+++ MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1
+++ export MANPATH_modshare
+++ LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1
+++ export LIBRARY_PATH_modshare
+++ NCCL_SOCKET_IFNAME=enp
+++ export NCCL_SOCKET_IFNAME
+++ AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl
+++ export AWS_OFI_NCCL_HOME
+++ NCCL_HOME=/opt/nccl/build
+++ export NCCL_HOME
+++ FI_PROVIDER=efa
+++ export FI_PROVIDER
+++ AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1
+++ export AWS_OFI_NCCL_HOME_modshare
+++ CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1
+++ export CPATH_modshare
+++ LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2::1:/opt/nccl/build/lib:1:/opt/aws-ofi-nccl/lib:2:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/efa/lib:1
+++ export LD_LIBRARY_PATH_modshare
+++ FI_EFA_ENABLE_SHM_TRANSFER=1
+++ export FI_EFA_ENABLE_SHM_TRANSFER
+++ _LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1
+++ export _LMFILES__modshare
+++ LOADEDMODULES_modshare=cuda/12.1:1
+++ export LOADEDMODULES_modshare
+++ MPI_PATH_modshare=/opt/amazon/openmpi:1
+++ export MPI_PATH_modshare
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ export PATH
+++ CUDA_HOME=/usr/local/cuda-12.1
+++ export CUDA_HOME
+++ PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/usr/local/cuda-12.1/include:1:/opt/amazon/efa/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/usr/sbin:1:/fsx/nouamane/miniconda/bin:1:/opt/slurm/bin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:2:/opt/amazon/openmpi/bin:1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1
+++ export PATH_modshare
+++ test 0
++ _mlstatus=0
++ '[' -n x ']'
++ IFS=' 	
+'
++ unset _mlre _mlv _mlrv _mlIFS
++ '[' -n '' ']'
++ unset _mlshdbg
++ return 0
++ source /fsx/nouamane/miniconda/bin/activate
+++ _CONDA_ROOT=/fsx/nouamane/miniconda
+++ . /fsx/nouamane/miniconda/etc/profile.d/conda.sh
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ '[' -z x ']'
+++ conda activate
+++ local cmd=activate
+++ case "$cmd" in
+++ __conda_activate activate
+++ '[' -n '' ']'
+++ local ask_conda
++++ PS1=
++++ __conda_exe shell.posix activate
++++ /fsx/nouamane/miniconda/bin/conda shell.posix activate
+++ ask_conda='. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_2='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\'''
+++ eval '. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_2='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\'''
++++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh
+++++ test -n ''
+++++ unset XML_CATALOG_FILES
+++++ unset xml_catalog_files_libxml2
++++ PS1='(base) '
++++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++++ export CONDA_PREFIX=/fsx/nouamane/miniconda
++++ CONDA_PREFIX=/fsx/nouamane/miniconda
++++ export CONDA_SHLVL=3
++++ CONDA_SHLVL=3
++++ export CONDA_DEFAULT_ENV=base
++++ CONDA_DEFAULT_ENV=base
++++ export 'CONDA_PROMPT_MODIFIER=(base) '
++++ CONDA_PROMPT_MODIFIER='(base) '
++++ export CONDA_PREFIX_2=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ CONDA_PREFIX_2=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ __conda_hashr
+++ '[' -n '' ']'
+++ '[' -n '' ']'
+++ hash -r
++ conda activate 2-1-cu121
++ local cmd=activate
++ case "$cmd" in
++ __conda_activate activate 2-1-cu121
++ '[' -n '' ']'
++ local ask_conda
+++ PS1='(base) '
+++ __conda_exe shell.posix activate 2-1-cu121
+++ /fsx/nouamane/miniconda/bin/conda shell.posix activate 2-1-cu121
++ ask_conda='PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''4'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_3='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
++ eval 'PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''4'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_3='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
+++ PS1='(2-1-cu121) '
+++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ export CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ export CONDA_SHLVL=4
+++ CONDA_SHLVL=4
+++ export CONDA_DEFAULT_ENV=2-1-cu121
+++ CONDA_DEFAULT_ENV=2-1-cu121
+++ export 'CONDA_PROMPT_MODIFIER=(2-1-cu121) '
+++ CONDA_PROMPT_MODIFIER='(2-1-cu121) '
+++ export CONDA_PREFIX_3=/fsx/nouamane/miniconda
+++ CONDA_PREFIX_3=/fsx/nouamane/miniconda
+++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ export _CE_M=
+++ _CE_M=
+++ export _CE_CONDA=
+++ _CE_CONDA=
+++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh
++++ test -n ''
++++ xml_catalog_files_libxml2=
++++ XML_CATALOG_FILES=
++++ conda_catalog_files=
++++ ifs_libxml2=' 	
+'
++++ IFS=' '
++++ rem=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ for pre in ${rem}
++++ test '' = /fsx/nouamane/miniconda/envs/2-1-cu121
++++ conda_catalog_files=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ rem=
++++ IFS=' 	
+'
++++ conda_catalog_files='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ export 'XML_CATALOG_FILES=file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ XML_CATALOG_FILES='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ unset conda_catalog_files ifs_libxml2 rem
++ __conda_hashr
++ '[' -n '' ']'
++ '[' -n '' ']'
++ hash -r
++ export PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++ PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ scontrol show hostnames ip-26-0-160-103
++ export NODELIST=ip-26-0-160-103
++ NODELIST=ip-26-0-160-103
+++ scontrol show hostnames ip-26-0-160-103
+++ head -n1
++ export MASTER_NODE=ip-26-0-160-103
++ MASTER_NODE=ip-26-0-160-103
++ export MASTER_PORT=12356
++ MASTER_PORT=12356
++ export NNODES=1
++ NNODES=1
++ export GPUS_PER_NODE=8
++ GPUS_PER_NODE=8
++ export WORLD_SIZE=8
++ WORLD_SIZE=8
++ export CUDA_DEVICE_MAX_CONNECTIONS=1
++ CUDA_DEVICE_MAX_CONNECTIONS=1
++ export NCCL_DEBUG=INFO
++ NCCL_DEBUG=INFO
++ export NANOTRON_BENCHMARK=1
++ NANOTRON_BENCHMARK=1
++ echo 'Master node: ip-26-0-160-103'
+Master node: ip-26-0-160-103
++ echo 'All nodes: ip-26-0-160-103'
+All nodes: ip-26-0-160-103
++ echo 'World size: 8'
+World size: 8
++ srun torchrun --nnodes=1 --nproc_per_node=8 --rdzv_id=13402561 --rdzv_backend=c10d --rdzv_endpoint=ip-26-0-160-103:12356 run_train.py --config-file benchmark/configs/config_1.14G_dp4_tp2_pp1_acc256_mbs2_seq2048_zero1_tpmodeRED_vocab32k.yaml
+[2024-12-13 02:43:56,374] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-13 02:43:56,374] torch.distributed.run: [WARNING] 
+[2024-12-13 02:43:56,374] torch.distributed.run: [WARNING] *****************************************
+[2024-12-13 02:43:56,374] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-13 02:43:56,374] torch.distributed.run: [WARNING] *****************************************
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+[2024-12-13 02:44:01,500] torch.distributed.elastic.multiprocessing.api: [ERROR] failed (exitcode: 2) local_rank: 0 (pid: 2268278) of binary: /fsx/nouamane/miniconda/envs/2-1-cu121/bin/python
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 33, in <module>
+    sys.exit(load_entry_point('torch==2.1.1', 'console_scripts', 'torchrun')())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 264, in launch_agent
+    raise ChildFailedError(
+torch.distributed.elastic.multiprocessing.errors.ChildFailedError: 
+============================================================
+run_train.py FAILED
+------------------------------------------------------------
+Failures:
+[1]:
+  time      : 2024-12-13_02:44:01
+  host      : ip-26-0-160-103.ec2.internal
+  rank      : 1 (local_rank: 1)
+  exitcode  : 2 (pid: 2268279)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[2]:
+  time      : 2024-12-13_02:44:01
+  host      : ip-26-0-160-103.ec2.internal
+  rank      : 2 (local_rank: 2)
+  exitcode  : 2 (pid: 2268280)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[3]:
+  time      : 2024-12-13_02:44:01
+  host      : ip-26-0-160-103.ec2.internal
+  rank      : 3 (local_rank: 3)
+  exitcode  : 2 (pid: 2268281)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[4]:
+  time      : 2024-12-13_02:44:01
+  host      : ip-26-0-160-103.ec2.internal
+  rank      : 4 (local_rank: 4)
+  exitcode  : 2 (pid: 2268282)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[5]:
+  time      : 2024-12-13_02:44:01
+  host      : ip-26-0-160-103.ec2.internal
+  rank      : 5 (local_rank: 5)
+  exitcode  : 2 (pid: 2268283)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[6]:
+  time      : 2024-12-13_02:44:01
+  host      : ip-26-0-160-103.ec2.internal
+  rank      : 6 (local_rank: 6)
+  exitcode  : 2 (pid: 2268284)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[7]:
+  time      : 2024-12-13_02:44:01
+  host      : ip-26-0-160-103.ec2.internal
+  rank      : 7 (local_rank: 7)
+  exitcode  : 2 (pid: 2268285)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+------------------------------------------------------------
+Root Cause (first observed failure):
+[0]:
+  time      : 2024-12-13_02:44:01
+  host      : ip-26-0-160-103.ec2.internal
+  rank      : 0 (local_rank: 0)
+  exitcode  : 2 (pid: 2268278)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+============================================================
+srun: error: ip-26-0-160-103: task 0: Exited with exit code 1
diff --git a/logs/13404086-bench_stress_test.out b/logs/13404086-bench_stress_test.out
new file mode 100644
index 0000000000000000000000000000000000000000..510e1d5f0bc7394116c31a3b0de7e52cee788363
--- /dev/null
+++ b/logs/13404086-bench_stress_test.out
@@ -0,0 +1,518 @@
++ source /etc/profile.d/modules.sh
+++ . /usr/share/modules/init/bash
++++ unset _mlshdbg
++++ '[' 0 = 1 ']'
++++ unset _mlre _mlIFS
++++ '[' -n x ']'
++++ _mlIFS=' 	
+'
++++ IFS=' '
++++ '[' -n '' ']'
+++++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash autoinit
++++ _mlcode='module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+MODULEPATH_modshare=/etc/environment-modules/modules:1:/usr/share/modules/\$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1; export MODULEPATH_modshare;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+LOADEDMODULES=; export LOADEDMODULES;
+MODULEPATH=/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/\$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles; export MODULEPATH;
+test 0;'
++++ _mlret=0
++++ '[' -n x ']'
++++ IFS=' 	
+'
++++ unset _mlIFS
++++ unset _mlre _mlv _mlrv
++++ '[' 0 -eq 0 ']'
++++ eval 'module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+MODULEPATH_modshare=/etc/environment-modules/modules:1:/usr/share/modules/\$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1; export MODULEPATH_modshare;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+LOADEDMODULES=; export LOADEDMODULES;
+MODULEPATH=/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/\$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles; export MODULEPATH;
+test 0;'
+++++ MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl
+++++ export MODULES_CMD
+++++ ENV=/usr/share/modules/init/profile.sh
+++++ export ENV
+++++ MODULEPATH_modshare='/etc/environment-modules/modules:1:/usr/share/modules/$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1'
+++++ export MODULEPATH_modshare
+++++ BASH_ENV=/usr/share/modules/init/bash
+++++ export BASH_ENV
+++++ MODULESHOME=/usr/share/modules
+++++ export MODULESHOME
+++++ LOADEDMODULES=
+++++ export LOADEDMODULES
+++++ MODULEPATH='/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles'
+++++ export MODULEPATH
+++++ test 0
++++ '[' 0 = 1 ']'
++++ '[' -t 2 ']'
++++ export -f module
++++ export -f switchml
++++ '[' 5 -ge 3 ']'
++++ [[ ehxB =~ i ]]
++++ [[ ! :/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin: =~ :/usr/bin: ]]
++++ '[' '!' -n '' ']'
++++ MANPATH=:
++++ export MANPATH
+++++ manpath
++++ [[ ! :/admin/home/nouamane/.local/share/man:/fsx/nouamane/miniconda/envs/2-1-cu121/man:/fsx/nouamane/miniconda/envs/2-1-cu121/share/man:/fsx/nouamane/miniconda/man:/fsx/nouamane/miniconda/share/man:/opt/amazon/openmpi/share/man:/opt/amazon/efa/share/man:/opt/slurm/share/man:/usr/local/man:/usr/local/share/man:/usr/share/man:/admin/home/nouamane/.fzf/man:: =~ :/usr/share/man: ]]
++++ unset _mlcode _mlret
++++ '[' -n '' ']'
++ module load cuda/12.1
++ unset _mlshdbg
++ '[' 0 = 1 ']'
++ unset _mlre _mlIFS
++ '[' -n x ']'
++ _mlIFS=' 	
+'
++ IFS=' '
++ '[' -n '' ']'
+++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash load cuda/12.1
++ eval 'CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include;' export 'CPATH;
+LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:;' export 'LD_LIBRARY_PATH;
+FI_EFA_FORK_SAFE=1;' export 'FI_EFA_FORK_SAFE;
+MANPATH=/usr/local/cuda-12.1/share/man::;' export 'MANPATH;
+LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64;' export 'LIBRARY_PATH;
+_LMFILES_=/usr/share/modules/modulefiles/cuda/12.1;' export '_LMFILES_;
+LOADEDMODULES=cuda/12.1;' export 'LOADEDMODULES;
+MPI_PATH=/opt/amazon/openmpi;' export 'MPI_PATH;
+NCCL_HOME_modshare=/opt/nccl/build:1;' export 'NCCL_HOME_modshare;
+NCCL_PROTO=simple;' export 'NCCL_PROTO;
+MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1;' export 'MANPATH_modshare;
+LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1;' export 'LIBRARY_PATH_modshare;
+NCCL_SOCKET_IFNAME=enp;' export 'NCCL_SOCKET_IFNAME;
+AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl;' export 'AWS_OFI_NCCL_HOME;
+NCCL_HOME=/opt/nccl/build;' export 'NCCL_HOME;
+FI_PROVIDER=efa;' export 'FI_PROVIDER;
+AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1;' export 'AWS_OFI_NCCL_HOME_modshare;
+CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1;' export 'CPATH_modshare;
+LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2::1:/opt/nccl/build/lib:1:/opt/aws-ofi-nccl/lib:2:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/efa/lib:1;' export 'LD_LIBRARY_PATH_modshare;
+FI_EFA_ENABLE_SHM_TRANSFER=1;' export 'FI_EFA_ENABLE_SHM_TRANSFER;
+_LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1;' export '_LMFILES__modshare;
+LOADEDMODULES_modshare=cuda/12.1:1;' export 'LOADEDMODULES_modshare;
+MPI_PATH_modshare=/opt/amazon/openmpi:1;' export 'MPI_PATH_modshare;
+PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin;' export 'PATH;
+CUDA_HOME=/usr/local/cuda-12.1;' export 'CUDA_HOME;
+PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/usr/local/cuda-12.1/include:1:/opt/amazon/efa/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/usr/sbin:1:/fsx/nouamane/miniconda/bin:1:/opt/slurm/bin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:2:/opt/amazon/openmpi/bin:1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1;' export 'PATH_modshare;
+test' '0;'
+++ CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include
+++ export CPATH
+++ LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:
+++ export LD_LIBRARY_PATH
+++ FI_EFA_FORK_SAFE=1
+++ export FI_EFA_FORK_SAFE
+++ MANPATH=/usr/local/cuda-12.1/share/man::
+++ export MANPATH
+++ LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64
+++ export LIBRARY_PATH
+++ _LMFILES_=/usr/share/modules/modulefiles/cuda/12.1
+++ export _LMFILES_
+++ LOADEDMODULES=cuda/12.1
+++ export LOADEDMODULES
+++ MPI_PATH=/opt/amazon/openmpi
+++ export MPI_PATH
+++ NCCL_HOME_modshare=/opt/nccl/build:1
+++ export NCCL_HOME_modshare
+++ NCCL_PROTO=simple
+++ export NCCL_PROTO
+++ MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1
+++ export MANPATH_modshare
+++ LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1
+++ export LIBRARY_PATH_modshare
+++ NCCL_SOCKET_IFNAME=enp
+++ export NCCL_SOCKET_IFNAME
+++ AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl
+++ export AWS_OFI_NCCL_HOME
+++ NCCL_HOME=/opt/nccl/build
+++ export NCCL_HOME
+++ FI_PROVIDER=efa
+++ export FI_PROVIDER
+++ AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1
+++ export AWS_OFI_NCCL_HOME_modshare
+++ CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1
+++ export CPATH_modshare
+++ LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2::1:/opt/nccl/build/lib:1:/opt/aws-ofi-nccl/lib:2:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/efa/lib:1
+++ export LD_LIBRARY_PATH_modshare
+++ FI_EFA_ENABLE_SHM_TRANSFER=1
+++ export FI_EFA_ENABLE_SHM_TRANSFER
+++ _LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1
+++ export _LMFILES__modshare
+++ LOADEDMODULES_modshare=cuda/12.1:1
+++ export LOADEDMODULES_modshare
+++ MPI_PATH_modshare=/opt/amazon/openmpi:1
+++ export MPI_PATH_modshare
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ export PATH
+++ CUDA_HOME=/usr/local/cuda-12.1
+++ export CUDA_HOME
+++ PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/usr/local/cuda-12.1/include:1:/opt/amazon/efa/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/usr/sbin:1:/fsx/nouamane/miniconda/bin:1:/opt/slurm/bin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:2:/opt/amazon/openmpi/bin:1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1
+++ export PATH_modshare
+++ test 0
++ _mlstatus=0
++ '[' -n x ']'
++ IFS=' 	
+'
++ unset _mlre _mlv _mlrv _mlIFS
++ '[' -n '' ']'
++ unset _mlshdbg
++ return 0
++ source /fsx/nouamane/miniconda/bin/activate
+++ _CONDA_ROOT=/fsx/nouamane/miniconda
+++ . /fsx/nouamane/miniconda/etc/profile.d/conda.sh
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ '[' -z x ']'
+++ conda activate
+++ local cmd=activate
+++ case "$cmd" in
+++ __conda_activate activate
+++ '[' -n '' ']'
+++ local ask_conda
++++ PS1=
++++ __conda_exe shell.posix activate
++++ /fsx/nouamane/miniconda/bin/conda shell.posix activate
+++ ask_conda='. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_2='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\'''
+++ eval '. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_2='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\'''
++++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh
+++++ test -n ''
+++++ unset XML_CATALOG_FILES
+++++ unset xml_catalog_files_libxml2
++++ PS1='(base) '
++++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++++ export CONDA_PREFIX=/fsx/nouamane/miniconda
++++ CONDA_PREFIX=/fsx/nouamane/miniconda
++++ export CONDA_SHLVL=3
++++ CONDA_SHLVL=3
++++ export CONDA_DEFAULT_ENV=base
++++ CONDA_DEFAULT_ENV=base
++++ export 'CONDA_PROMPT_MODIFIER=(base) '
++++ CONDA_PROMPT_MODIFIER='(base) '
++++ export CONDA_PREFIX_2=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ CONDA_PREFIX_2=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ __conda_hashr
+++ '[' -n '' ']'
+++ '[' -n '' ']'
+++ hash -r
++ conda activate 2-1-cu121
++ local cmd=activate
++ case "$cmd" in
++ __conda_activate activate 2-1-cu121
++ '[' -n '' ']'
++ local ask_conda
+++ PS1='(base) '
+++ __conda_exe shell.posix activate 2-1-cu121
+++ /fsx/nouamane/miniconda/bin/conda shell.posix activate 2-1-cu121
++ ask_conda='PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''4'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_3='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
++ eval 'PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''4'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_3='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
+++ PS1='(2-1-cu121) '
+++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ export CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ export CONDA_SHLVL=4
+++ CONDA_SHLVL=4
+++ export CONDA_DEFAULT_ENV=2-1-cu121
+++ CONDA_DEFAULT_ENV=2-1-cu121
+++ export 'CONDA_PROMPT_MODIFIER=(2-1-cu121) '
+++ CONDA_PROMPT_MODIFIER='(2-1-cu121) '
+++ export CONDA_PREFIX_3=/fsx/nouamane/miniconda
+++ CONDA_PREFIX_3=/fsx/nouamane/miniconda
+++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ export _CE_M=
+++ _CE_M=
+++ export _CE_CONDA=
+++ _CE_CONDA=
+++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh
++++ test -n ''
++++ xml_catalog_files_libxml2=
++++ XML_CATALOG_FILES=
++++ conda_catalog_files=
++++ ifs_libxml2=' 	
+'
++++ IFS=' '
++++ rem=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ for pre in ${rem}
++++ test '' = /fsx/nouamane/miniconda/envs/2-1-cu121
++++ conda_catalog_files=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ rem=
++++ IFS=' 	
+'
++++ conda_catalog_files='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ export 'XML_CATALOG_FILES=file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ XML_CATALOG_FILES='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ unset conda_catalog_files ifs_libxml2 rem
++ __conda_hashr
++ '[' -n '' ']'
++ '[' -n '' ']'
++ hash -r
++ export PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++ PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ scontrol show hostnames ip-26-0-169-207
++ export NODELIST=ip-26-0-169-207
++ NODELIST=ip-26-0-169-207
+++ scontrol show hostnames ip-26-0-169-207
+++ head -n1
++ export MASTER_NODE=ip-26-0-169-207
++ MASTER_NODE=ip-26-0-169-207
++ export MASTER_PORT=12356
++ MASTER_PORT=12356
++ export NNODES=1
++ NNODES=1
++ export GPUS_PER_NODE=8
++ GPUS_PER_NODE=8
++ export WORLD_SIZE=8
++ WORLD_SIZE=8
++ export CUDA_DEVICE_MAX_CONNECTIONS=1
++ CUDA_DEVICE_MAX_CONNECTIONS=1
++ export NCCL_DEBUG=INFO
++ NCCL_DEBUG=INFO
++ export NANOTRON_BENCHMARK=1
++ NANOTRON_BENCHMARK=1
++ echo 'Master node: ip-26-0-169-207'
+Master node: ip-26-0-169-207
++ echo 'All nodes: ip-26-0-169-207'
+All nodes: ip-26-0-169-207
++ echo 'World size: 8'
+World size: 8
++ srun torchrun --nnodes=1 --nproc_per_node=8 --rdzv_id=13404086 --rdzv_backend=c10d --rdzv_endpoint=ip-26-0-169-207:12356 run_train.py --config-file benchmark/configs/config_1.14G_dp4_tp2_pp1_acc256_mbs2_seq2048_zero1_tpmodeRED_vocab32k.yaml
+[2024-12-13 02:51:06,053] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-13 02:51:06,053] torch.distributed.run: [WARNING] 
+[2024-12-13 02:51:06,053] torch.distributed.run: [WARNING] *****************************************
+[2024-12-13 02:51:06,053] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-13 02:51:06,053] torch.distributed.run: [WARNING] *****************************************
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+[2024-12-13 02:51:11,150] torch.distributed.elastic.multiprocessing.api: [ERROR] failed (exitcode: 2) local_rank: 0 (pid: 385269) of binary: /fsx/nouamane/miniconda/envs/2-1-cu121/bin/python
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 33, in <module>
+    sys.exit(load_entry_point('torch==2.1.1', 'console_scripts', 'torchrun')())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 264, in launch_agent
+    raise ChildFailedError(
+torch.distributed.elastic.multiprocessing.errors.ChildFailedError: 
+============================================================
+run_train.py FAILED
+------------------------------------------------------------
+Failures:
+[1]:
+  time      : 2024-12-13_02:51:11
+  host      : ip-26-0-169-207.ec2.internal
+  rank      : 1 (local_rank: 1)
+  exitcode  : 2 (pid: 385270)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[2]:
+  time      : 2024-12-13_02:51:11
+  host      : ip-26-0-169-207.ec2.internal
+  rank      : 2 (local_rank: 2)
+  exitcode  : 2 (pid: 385271)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[3]:
+  time      : 2024-12-13_02:51:11
+  host      : ip-26-0-169-207.ec2.internal
+  rank      : 3 (local_rank: 3)
+  exitcode  : 2 (pid: 385272)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[4]:
+  time      : 2024-12-13_02:51:11
+  host      : ip-26-0-169-207.ec2.internal
+  rank      : 4 (local_rank: 4)
+  exitcode  : 2 (pid: 385273)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[5]:
+  time      : 2024-12-13_02:51:11
+  host      : ip-26-0-169-207.ec2.internal
+  rank      : 5 (local_rank: 5)
+  exitcode  : 2 (pid: 385274)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[6]:
+  time      : 2024-12-13_02:51:11
+  host      : ip-26-0-169-207.ec2.internal
+  rank      : 6 (local_rank: 6)
+  exitcode  : 2 (pid: 385275)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[7]:
+  time      : 2024-12-13_02:51:11
+  host      : ip-26-0-169-207.ec2.internal
+  rank      : 7 (local_rank: 7)
+  exitcode  : 2 (pid: 385276)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+------------------------------------------------------------
+Root Cause (first observed failure):
+[0]:
+  time      : 2024-12-13_02:51:11
+  host      : ip-26-0-169-207.ec2.internal
+  rank      : 0 (local_rank: 0)
+  exitcode  : 2 (pid: 385269)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+============================================================
+srun: error: ip-26-0-169-207: task 0: Exited with exit code 1
diff --git a/logs/13404128-bench_stress_test.out b/logs/13404128-bench_stress_test.out
new file mode 100644
index 0000000000000000000000000000000000000000..f2057cb18818778806727d3e93f5718b0ee9da72
--- /dev/null
+++ b/logs/13404128-bench_stress_test.out
@@ -0,0 +1,518 @@
++ source /etc/profile.d/modules.sh
+++ . /usr/share/modules/init/bash
++++ unset _mlshdbg
++++ '[' 0 = 1 ']'
++++ unset _mlre _mlIFS
++++ '[' -n x ']'
++++ _mlIFS=' 	
+'
++++ IFS=' '
++++ '[' -n '' ']'
+++++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash autoinit
++++ _mlcode='module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+MODULEPATH_modshare=/etc/environment-modules/modules:1:/usr/share/modules/\$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1; export MODULEPATH_modshare;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+LOADEDMODULES=; export LOADEDMODULES;
+MODULEPATH=/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/\$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles; export MODULEPATH;
+test 0;'
++++ _mlret=0
++++ '[' -n x ']'
++++ IFS=' 	
+'
++++ unset _mlIFS
++++ unset _mlre _mlv _mlrv
++++ '[' 0 -eq 0 ']'
++++ eval 'module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+MODULEPATH_modshare=/etc/environment-modules/modules:1:/usr/share/modules/\$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1; export MODULEPATH_modshare;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+LOADEDMODULES=; export LOADEDMODULES;
+MODULEPATH=/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/\$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles; export MODULEPATH;
+test 0;'
+++++ MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl
+++++ export MODULES_CMD
+++++ ENV=/usr/share/modules/init/profile.sh
+++++ export ENV
+++++ MODULEPATH_modshare='/etc/environment-modules/modules:1:/usr/share/modules/$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1'
+++++ export MODULEPATH_modshare
+++++ BASH_ENV=/usr/share/modules/init/bash
+++++ export BASH_ENV
+++++ MODULESHOME=/usr/share/modules
+++++ export MODULESHOME
+++++ LOADEDMODULES=
+++++ export LOADEDMODULES
+++++ MODULEPATH='/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles'
+++++ export MODULEPATH
+++++ test 0
++++ '[' 0 = 1 ']'
++++ '[' -t 2 ']'
++++ export -f module
++++ export -f switchml
++++ '[' 5 -ge 3 ']'
++++ [[ ehxB =~ i ]]
++++ [[ ! :/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin: =~ :/usr/bin: ]]
++++ '[' '!' -n '' ']'
++++ MANPATH=:
++++ export MANPATH
+++++ manpath
++++ [[ ! :/admin/home/nouamane/.local/share/man:/fsx/nouamane/miniconda/envs/2-1-cu121/man:/fsx/nouamane/miniconda/envs/2-1-cu121/share/man:/fsx/nouamane/miniconda/man:/fsx/nouamane/miniconda/share/man:/opt/amazon/openmpi/share/man:/opt/amazon/efa/share/man:/opt/slurm/share/man:/usr/local/man:/usr/local/share/man:/usr/share/man:/admin/home/nouamane/.fzf/man:: =~ :/usr/share/man: ]]
++++ unset _mlcode _mlret
++++ '[' -n '' ']'
++ module load cuda/12.1
++ unset _mlshdbg
++ '[' 0 = 1 ']'
++ unset _mlre _mlIFS
++ '[' -n x ']'
++ _mlIFS=' 	
+'
++ IFS=' '
++ '[' -n '' ']'
+++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash load cuda/12.1
++ eval 'CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include;' export 'CPATH;
+LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:;' export 'LD_LIBRARY_PATH;
+FI_EFA_FORK_SAFE=1;' export 'FI_EFA_FORK_SAFE;
+MANPATH=/usr/local/cuda-12.1/share/man::;' export 'MANPATH;
+LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64;' export 'LIBRARY_PATH;
+_LMFILES_=/usr/share/modules/modulefiles/cuda/12.1;' export '_LMFILES_;
+LOADEDMODULES=cuda/12.1;' export 'LOADEDMODULES;
+MPI_PATH=/opt/amazon/openmpi;' export 'MPI_PATH;
+NCCL_HOME_modshare=/opt/nccl/build:1;' export 'NCCL_HOME_modshare;
+NCCL_PROTO=simple;' export 'NCCL_PROTO;
+MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1;' export 'MANPATH_modshare;
+LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1;' export 'LIBRARY_PATH_modshare;
+NCCL_SOCKET_IFNAME=enp;' export 'NCCL_SOCKET_IFNAME;
+AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl;' export 'AWS_OFI_NCCL_HOME;
+NCCL_HOME=/opt/nccl/build;' export 'NCCL_HOME;
+FI_PROVIDER=efa;' export 'FI_PROVIDER;
+AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1;' export 'AWS_OFI_NCCL_HOME_modshare;
+CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1;' export 'CPATH_modshare;
+LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2::1:/opt/nccl/build/lib:1:/opt/aws-ofi-nccl/lib:2:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/efa/lib:1;' export 'LD_LIBRARY_PATH_modshare;
+FI_EFA_ENABLE_SHM_TRANSFER=1;' export 'FI_EFA_ENABLE_SHM_TRANSFER;
+_LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1;' export '_LMFILES__modshare;
+LOADEDMODULES_modshare=cuda/12.1:1;' export 'LOADEDMODULES_modshare;
+MPI_PATH_modshare=/opt/amazon/openmpi:1;' export 'MPI_PATH_modshare;
+PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin;' export 'PATH;
+CUDA_HOME=/usr/local/cuda-12.1;' export 'CUDA_HOME;
+PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/usr/local/cuda-12.1/include:1:/opt/amazon/efa/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/usr/sbin:1:/fsx/nouamane/miniconda/bin:1:/opt/slurm/bin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:2:/opt/amazon/openmpi/bin:1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1;' export 'PATH_modshare;
+test' '0;'
+++ CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include
+++ export CPATH
+++ LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:
+++ export LD_LIBRARY_PATH
+++ FI_EFA_FORK_SAFE=1
+++ export FI_EFA_FORK_SAFE
+++ MANPATH=/usr/local/cuda-12.1/share/man::
+++ export MANPATH
+++ LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64
+++ export LIBRARY_PATH
+++ _LMFILES_=/usr/share/modules/modulefiles/cuda/12.1
+++ export _LMFILES_
+++ LOADEDMODULES=cuda/12.1
+++ export LOADEDMODULES
+++ MPI_PATH=/opt/amazon/openmpi
+++ export MPI_PATH
+++ NCCL_HOME_modshare=/opt/nccl/build:1
+++ export NCCL_HOME_modshare
+++ NCCL_PROTO=simple
+++ export NCCL_PROTO
+++ MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1
+++ export MANPATH_modshare
+++ LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1
+++ export LIBRARY_PATH_modshare
+++ NCCL_SOCKET_IFNAME=enp
+++ export NCCL_SOCKET_IFNAME
+++ AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl
+++ export AWS_OFI_NCCL_HOME
+++ NCCL_HOME=/opt/nccl/build
+++ export NCCL_HOME
+++ FI_PROVIDER=efa
+++ export FI_PROVIDER
+++ AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1
+++ export AWS_OFI_NCCL_HOME_modshare
+++ CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1
+++ export CPATH_modshare
+++ LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2::1:/opt/nccl/build/lib:1:/opt/aws-ofi-nccl/lib:2:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/efa/lib:1
+++ export LD_LIBRARY_PATH_modshare
+++ FI_EFA_ENABLE_SHM_TRANSFER=1
+++ export FI_EFA_ENABLE_SHM_TRANSFER
+++ _LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1
+++ export _LMFILES__modshare
+++ LOADEDMODULES_modshare=cuda/12.1:1
+++ export LOADEDMODULES_modshare
+++ MPI_PATH_modshare=/opt/amazon/openmpi:1
+++ export MPI_PATH_modshare
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ export PATH
+++ CUDA_HOME=/usr/local/cuda-12.1
+++ export CUDA_HOME
+++ PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/usr/local/cuda-12.1/include:1:/opt/amazon/efa/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/usr/sbin:1:/fsx/nouamane/miniconda/bin:1:/opt/slurm/bin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:2:/opt/amazon/openmpi/bin:1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1
+++ export PATH_modshare
+++ test 0
++ _mlstatus=0
++ '[' -n x ']'
++ IFS=' 	
+'
++ unset _mlre _mlv _mlrv _mlIFS
++ '[' -n '' ']'
++ unset _mlshdbg
++ return 0
++ source /fsx/nouamane/miniconda/bin/activate
+++ _CONDA_ROOT=/fsx/nouamane/miniconda
+++ . /fsx/nouamane/miniconda/etc/profile.d/conda.sh
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ '[' -z x ']'
+++ conda activate
+++ local cmd=activate
+++ case "$cmd" in
+++ __conda_activate activate
+++ '[' -n '' ']'
+++ local ask_conda
++++ PS1=
++++ __conda_exe shell.posix activate
++++ /fsx/nouamane/miniconda/bin/conda shell.posix activate
+++ ask_conda='. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_2='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\'''
+++ eval '. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_2='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\'''
++++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh
+++++ test -n ''
+++++ unset XML_CATALOG_FILES
+++++ unset xml_catalog_files_libxml2
++++ PS1='(base) '
++++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++++ export CONDA_PREFIX=/fsx/nouamane/miniconda
++++ CONDA_PREFIX=/fsx/nouamane/miniconda
++++ export CONDA_SHLVL=3
++++ CONDA_SHLVL=3
++++ export CONDA_DEFAULT_ENV=base
++++ CONDA_DEFAULT_ENV=base
++++ export 'CONDA_PROMPT_MODIFIER=(base) '
++++ CONDA_PROMPT_MODIFIER='(base) '
++++ export CONDA_PREFIX_2=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ CONDA_PREFIX_2=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ __conda_hashr
+++ '[' -n '' ']'
+++ '[' -n '' ']'
+++ hash -r
++ conda activate 2-1-cu121
++ local cmd=activate
++ case "$cmd" in
++ __conda_activate activate 2-1-cu121
++ '[' -n '' ']'
++ local ask_conda
+++ PS1='(base) '
+++ __conda_exe shell.posix activate 2-1-cu121
+++ /fsx/nouamane/miniconda/bin/conda shell.posix activate 2-1-cu121
++ ask_conda='PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''4'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_3='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
++ eval 'PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''4'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_3='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
+++ PS1='(2-1-cu121) '
+++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ export CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ export CONDA_SHLVL=4
+++ CONDA_SHLVL=4
+++ export CONDA_DEFAULT_ENV=2-1-cu121
+++ CONDA_DEFAULT_ENV=2-1-cu121
+++ export 'CONDA_PROMPT_MODIFIER=(2-1-cu121) '
+++ CONDA_PROMPT_MODIFIER='(2-1-cu121) '
+++ export CONDA_PREFIX_3=/fsx/nouamane/miniconda
+++ CONDA_PREFIX_3=/fsx/nouamane/miniconda
+++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ export _CE_M=
+++ _CE_M=
+++ export _CE_CONDA=
+++ _CE_CONDA=
+++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh
++++ test -n ''
++++ xml_catalog_files_libxml2=
++++ XML_CATALOG_FILES=
++++ conda_catalog_files=
++++ ifs_libxml2=' 	
+'
++++ IFS=' '
++++ rem=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ for pre in ${rem}
++++ test '' = /fsx/nouamane/miniconda/envs/2-1-cu121
++++ conda_catalog_files=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ rem=
++++ IFS=' 	
+'
++++ conda_catalog_files='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ export 'XML_CATALOG_FILES=file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ XML_CATALOG_FILES='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ unset conda_catalog_files ifs_libxml2 rem
++ __conda_hashr
++ '[' -n '' ']'
++ '[' -n '' ']'
++ hash -r
++ export PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++ PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ scontrol show hostnames ip-26-0-160-103
++ export NODELIST=ip-26-0-160-103
++ NODELIST=ip-26-0-160-103
+++ scontrol show hostnames ip-26-0-160-103
+++ head -n1
++ export MASTER_NODE=ip-26-0-160-103
++ MASTER_NODE=ip-26-0-160-103
++ export MASTER_PORT=12356
++ MASTER_PORT=12356
++ export NNODES=1
++ NNODES=1
++ export GPUS_PER_NODE=8
++ GPUS_PER_NODE=8
++ export WORLD_SIZE=8
++ WORLD_SIZE=8
++ export CUDA_DEVICE_MAX_CONNECTIONS=1
++ CUDA_DEVICE_MAX_CONNECTIONS=1
++ export NCCL_DEBUG=INFO
++ NCCL_DEBUG=INFO
++ export NANOTRON_BENCHMARK=1
++ NANOTRON_BENCHMARK=1
++ echo 'Master node: ip-26-0-160-103'
+Master node: ip-26-0-160-103
++ echo 'All nodes: ip-26-0-160-103'
+All nodes: ip-26-0-160-103
++ echo 'World size: 8'
+World size: 8
++ srun torchrun --nnodes=1 --nproc_per_node=8 --rdzv_id=13404128 --rdzv_backend=c10d --rdzv_endpoint=ip-26-0-160-103:12356 run_train.py --config-file benchmark/configs/config_1.14G_dp4_tp2_pp1_acc256_mbs2_seq2048_zero1_tpmodeRED_vocab32k.yaml
+[2024-12-13 02:57:44,074] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-13 02:57:44,074] torch.distributed.run: [WARNING] 
+[2024-12-13 02:57:44,074] torch.distributed.run: [WARNING] *****************************************
+[2024-12-13 02:57:44,074] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-13 02:57:44,074] torch.distributed.run: [WARNING] *****************************************
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+[2024-12-13 02:57:49,322] torch.distributed.elastic.multiprocessing.api: [ERROR] failed (exitcode: 2) local_rank: 0 (pid: 2281907) of binary: /fsx/nouamane/miniconda/envs/2-1-cu121/bin/python
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 33, in <module>
+    sys.exit(load_entry_point('torch==2.1.1', 'console_scripts', 'torchrun')())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 264, in launch_agent
+    raise ChildFailedError(
+torch.distributed.elastic.multiprocessing.errors.ChildFailedError: 
+============================================================
+run_train.py FAILED
+------------------------------------------------------------
+Failures:
+[1]:
+  time      : 2024-12-13_02:57:49
+  host      : ip-26-0-160-103.ec2.internal
+  rank      : 1 (local_rank: 1)
+  exitcode  : 2 (pid: 2281908)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[2]:
+  time      : 2024-12-13_02:57:49
+  host      : ip-26-0-160-103.ec2.internal
+  rank      : 2 (local_rank: 2)
+  exitcode  : 2 (pid: 2281909)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[3]:
+  time      : 2024-12-13_02:57:49
+  host      : ip-26-0-160-103.ec2.internal
+  rank      : 3 (local_rank: 3)
+  exitcode  : 2 (pid: 2281910)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[4]:
+  time      : 2024-12-13_02:57:49
+  host      : ip-26-0-160-103.ec2.internal
+  rank      : 4 (local_rank: 4)
+  exitcode  : 2 (pid: 2281911)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[5]:
+  time      : 2024-12-13_02:57:49
+  host      : ip-26-0-160-103.ec2.internal
+  rank      : 5 (local_rank: 5)
+  exitcode  : 2 (pid: 2281912)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[6]:
+  time      : 2024-12-13_02:57:49
+  host      : ip-26-0-160-103.ec2.internal
+  rank      : 6 (local_rank: 6)
+  exitcode  : 2 (pid: 2281913)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[7]:
+  time      : 2024-12-13_02:57:49
+  host      : ip-26-0-160-103.ec2.internal
+  rank      : 7 (local_rank: 7)
+  exitcode  : 2 (pid: 2281914)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+------------------------------------------------------------
+Root Cause (first observed failure):
+[0]:
+  time      : 2024-12-13_02:57:49
+  host      : ip-26-0-160-103.ec2.internal
+  rank      : 0 (local_rank: 0)
+  exitcode  : 2 (pid: 2281907)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+============================================================
+srun: error: ip-26-0-160-103: task 0: Exited with exit code 1
diff --git a/logs/13404177-bench_stress_test.out b/logs/13404177-bench_stress_test.out
new file mode 100644
index 0000000000000000000000000000000000000000..d871d939121d41b186c53fbb776e24562d1df8e8
--- /dev/null
+++ b/logs/13404177-bench_stress_test.out
@@ -0,0 +1,518 @@
++ source /etc/profile.d/modules.sh
+++ . /usr/share/modules/init/bash
++++ unset _mlshdbg
++++ '[' 0 = 1 ']'
++++ unset _mlre _mlIFS
++++ '[' -n x ']'
++++ _mlIFS=' 	
+'
++++ IFS=' '
++++ '[' -n '' ']'
+++++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash autoinit
++++ _mlcode='module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+MODULEPATH_modshare=/etc/environment-modules/modules:1:/usr/share/modules/\$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1; export MODULEPATH_modshare;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+LOADEDMODULES=; export LOADEDMODULES;
+MODULEPATH=/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/\$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles; export MODULEPATH;
+test 0;'
++++ _mlret=0
++++ '[' -n x ']'
++++ IFS=' 	
+'
++++ unset _mlIFS
++++ unset _mlre _mlv _mlrv
++++ '[' 0 -eq 0 ']'
++++ eval 'module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+MODULEPATH_modshare=/etc/environment-modules/modules:1:/usr/share/modules/\$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1; export MODULEPATH_modshare;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+LOADEDMODULES=; export LOADEDMODULES;
+MODULEPATH=/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/\$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles; export MODULEPATH;
+test 0;'
+++++ MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl
+++++ export MODULES_CMD
+++++ ENV=/usr/share/modules/init/profile.sh
+++++ export ENV
+++++ MODULEPATH_modshare='/etc/environment-modules/modules:1:/usr/share/modules/$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1'
+++++ export MODULEPATH_modshare
+++++ BASH_ENV=/usr/share/modules/init/bash
+++++ export BASH_ENV
+++++ MODULESHOME=/usr/share/modules
+++++ export MODULESHOME
+++++ LOADEDMODULES=
+++++ export LOADEDMODULES
+++++ MODULEPATH='/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles'
+++++ export MODULEPATH
+++++ test 0
++++ '[' 0 = 1 ']'
++++ '[' -t 2 ']'
++++ export -f module
++++ export -f switchml
++++ '[' 5 -ge 3 ']'
++++ [[ ehxB =~ i ]]
++++ [[ ! :/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin: =~ :/usr/bin: ]]
++++ '[' '!' -n '' ']'
++++ MANPATH=:
++++ export MANPATH
+++++ manpath
++++ [[ ! :/admin/home/nouamane/.local/share/man:/fsx/nouamane/miniconda/envs/2-1-cu121/man:/fsx/nouamane/miniconda/envs/2-1-cu121/share/man:/fsx/nouamane/miniconda/man:/fsx/nouamane/miniconda/share/man:/opt/amazon/openmpi/share/man:/opt/amazon/efa/share/man:/opt/slurm/share/man:/usr/local/man:/usr/local/share/man:/usr/share/man:/admin/home/nouamane/.fzf/man:: =~ :/usr/share/man: ]]
++++ unset _mlcode _mlret
++++ '[' -n '' ']'
++ module load cuda/12.1
++ unset _mlshdbg
++ '[' 0 = 1 ']'
++ unset _mlre _mlIFS
++ '[' -n x ']'
++ _mlIFS=' 	
+'
++ IFS=' '
++ '[' -n '' ']'
+++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash load cuda/12.1
++ eval 'CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include;' export 'CPATH;
+LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:;' export 'LD_LIBRARY_PATH;
+FI_EFA_FORK_SAFE=1;' export 'FI_EFA_FORK_SAFE;
+MANPATH=/usr/local/cuda-12.1/share/man::;' export 'MANPATH;
+LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64;' export 'LIBRARY_PATH;
+_LMFILES_=/usr/share/modules/modulefiles/cuda/12.1;' export '_LMFILES_;
+LOADEDMODULES=cuda/12.1;' export 'LOADEDMODULES;
+MPI_PATH=/opt/amazon/openmpi;' export 'MPI_PATH;
+NCCL_HOME_modshare=/opt/nccl/build:1;' export 'NCCL_HOME_modshare;
+NCCL_PROTO=simple;' export 'NCCL_PROTO;
+MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1;' export 'MANPATH_modshare;
+LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1;' export 'LIBRARY_PATH_modshare;
+NCCL_SOCKET_IFNAME=enp;' export 'NCCL_SOCKET_IFNAME;
+AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl;' export 'AWS_OFI_NCCL_HOME;
+NCCL_HOME=/opt/nccl/build;' export 'NCCL_HOME;
+FI_PROVIDER=efa;' export 'FI_PROVIDER;
+AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1;' export 'AWS_OFI_NCCL_HOME_modshare;
+CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1;' export 'CPATH_modshare;
+LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2::1:/opt/nccl/build/lib:1:/opt/aws-ofi-nccl/lib:2:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/efa/lib:1;' export 'LD_LIBRARY_PATH_modshare;
+FI_EFA_ENABLE_SHM_TRANSFER=1;' export 'FI_EFA_ENABLE_SHM_TRANSFER;
+_LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1;' export '_LMFILES__modshare;
+LOADEDMODULES_modshare=cuda/12.1:1;' export 'LOADEDMODULES_modshare;
+MPI_PATH_modshare=/opt/amazon/openmpi:1;' export 'MPI_PATH_modshare;
+PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin;' export 'PATH;
+CUDA_HOME=/usr/local/cuda-12.1;' export 'CUDA_HOME;
+PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/usr/local/cuda-12.1/include:1:/opt/amazon/efa/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/usr/sbin:1:/fsx/nouamane/miniconda/bin:1:/opt/slurm/bin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:2:/opt/amazon/openmpi/bin:1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1;' export 'PATH_modshare;
+test' '0;'
+++ CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include
+++ export CPATH
+++ LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:
+++ export LD_LIBRARY_PATH
+++ FI_EFA_FORK_SAFE=1
+++ export FI_EFA_FORK_SAFE
+++ MANPATH=/usr/local/cuda-12.1/share/man::
+++ export MANPATH
+++ LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64
+++ export LIBRARY_PATH
+++ _LMFILES_=/usr/share/modules/modulefiles/cuda/12.1
+++ export _LMFILES_
+++ LOADEDMODULES=cuda/12.1
+++ export LOADEDMODULES
+++ MPI_PATH=/opt/amazon/openmpi
+++ export MPI_PATH
+++ NCCL_HOME_modshare=/opt/nccl/build:1
+++ export NCCL_HOME_modshare
+++ NCCL_PROTO=simple
+++ export NCCL_PROTO
+++ MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1
+++ export MANPATH_modshare
+++ LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1
+++ export LIBRARY_PATH_modshare
+++ NCCL_SOCKET_IFNAME=enp
+++ export NCCL_SOCKET_IFNAME
+++ AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl
+++ export AWS_OFI_NCCL_HOME
+++ NCCL_HOME=/opt/nccl/build
+++ export NCCL_HOME
+++ FI_PROVIDER=efa
+++ export FI_PROVIDER
+++ AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1
+++ export AWS_OFI_NCCL_HOME_modshare
+++ CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1
+++ export CPATH_modshare
+++ LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2::1:/opt/nccl/build/lib:1:/opt/aws-ofi-nccl/lib:2:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/efa/lib:1
+++ export LD_LIBRARY_PATH_modshare
+++ FI_EFA_ENABLE_SHM_TRANSFER=1
+++ export FI_EFA_ENABLE_SHM_TRANSFER
+++ _LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1
+++ export _LMFILES__modshare
+++ LOADEDMODULES_modshare=cuda/12.1:1
+++ export LOADEDMODULES_modshare
+++ MPI_PATH_modshare=/opt/amazon/openmpi:1
+++ export MPI_PATH_modshare
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ export PATH
+++ CUDA_HOME=/usr/local/cuda-12.1
+++ export CUDA_HOME
+++ PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/usr/local/cuda-12.1/include:1:/opt/amazon/efa/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/usr/sbin:1:/fsx/nouamane/miniconda/bin:1:/opt/slurm/bin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:2:/opt/amazon/openmpi/bin:1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1
+++ export PATH_modshare
+++ test 0
++ _mlstatus=0
++ '[' -n x ']'
++ IFS=' 	
+'
++ unset _mlre _mlv _mlrv _mlIFS
++ '[' -n '' ']'
++ unset _mlshdbg
++ return 0
++ source /fsx/nouamane/miniconda/bin/activate
+++ _CONDA_ROOT=/fsx/nouamane/miniconda
+++ . /fsx/nouamane/miniconda/etc/profile.d/conda.sh
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ '[' -z x ']'
+++ conda activate
+++ local cmd=activate
+++ case "$cmd" in
+++ __conda_activate activate
+++ '[' -n '' ']'
+++ local ask_conda
++++ PS1=
++++ __conda_exe shell.posix activate
++++ /fsx/nouamane/miniconda/bin/conda shell.posix activate
+++ ask_conda='. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_2='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\'''
+++ eval '. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_2='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\'''
++++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh
+++++ test -n ''
+++++ unset XML_CATALOG_FILES
+++++ unset xml_catalog_files_libxml2
++++ PS1='(base) '
++++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++++ export CONDA_PREFIX=/fsx/nouamane/miniconda
++++ CONDA_PREFIX=/fsx/nouamane/miniconda
++++ export CONDA_SHLVL=3
++++ CONDA_SHLVL=3
++++ export CONDA_DEFAULT_ENV=base
++++ CONDA_DEFAULT_ENV=base
++++ export 'CONDA_PROMPT_MODIFIER=(base) '
++++ CONDA_PROMPT_MODIFIER='(base) '
++++ export CONDA_PREFIX_2=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ CONDA_PREFIX_2=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ __conda_hashr
+++ '[' -n '' ']'
+++ '[' -n '' ']'
+++ hash -r
++ conda activate 2-1-cu121
++ local cmd=activate
++ case "$cmd" in
++ __conda_activate activate 2-1-cu121
++ '[' -n '' ']'
++ local ask_conda
+++ PS1='(base) '
+++ __conda_exe shell.posix activate 2-1-cu121
+++ /fsx/nouamane/miniconda/bin/conda shell.posix activate 2-1-cu121
++ ask_conda='PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''4'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_3='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
++ eval 'PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''4'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_3='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
+++ PS1='(2-1-cu121) '
+++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ export CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ export CONDA_SHLVL=4
+++ CONDA_SHLVL=4
+++ export CONDA_DEFAULT_ENV=2-1-cu121
+++ CONDA_DEFAULT_ENV=2-1-cu121
+++ export 'CONDA_PROMPT_MODIFIER=(2-1-cu121) '
+++ CONDA_PROMPT_MODIFIER='(2-1-cu121) '
+++ export CONDA_PREFIX_3=/fsx/nouamane/miniconda
+++ CONDA_PREFIX_3=/fsx/nouamane/miniconda
+++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ export _CE_M=
+++ _CE_M=
+++ export _CE_CONDA=
+++ _CE_CONDA=
+++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh
++++ test -n ''
++++ xml_catalog_files_libxml2=
++++ XML_CATALOG_FILES=
++++ conda_catalog_files=
++++ ifs_libxml2=' 	
+'
++++ IFS=' '
++++ rem=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ for pre in ${rem}
++++ test '' = /fsx/nouamane/miniconda/envs/2-1-cu121
++++ conda_catalog_files=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ rem=
++++ IFS=' 	
+'
++++ conda_catalog_files='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ export 'XML_CATALOG_FILES=file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ XML_CATALOG_FILES='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ unset conda_catalog_files ifs_libxml2 rem
++ __conda_hashr
++ '[' -n '' ']'
++ '[' -n '' ']'
++ hash -r
++ export PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++ PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ scontrol show hostnames ip-26-0-160-103
++ export NODELIST=ip-26-0-160-103
++ NODELIST=ip-26-0-160-103
+++ scontrol show hostnames ip-26-0-160-103
+++ head -n1
++ export MASTER_NODE=ip-26-0-160-103
++ MASTER_NODE=ip-26-0-160-103
++ export MASTER_PORT=12356
++ MASTER_PORT=12356
++ export NNODES=1
++ NNODES=1
++ export GPUS_PER_NODE=8
++ GPUS_PER_NODE=8
++ export WORLD_SIZE=8
++ WORLD_SIZE=8
++ export CUDA_DEVICE_MAX_CONNECTIONS=1
++ CUDA_DEVICE_MAX_CONNECTIONS=1
++ export NCCL_DEBUG=INFO
++ NCCL_DEBUG=INFO
++ export NANOTRON_BENCHMARK=1
++ NANOTRON_BENCHMARK=1
++ echo 'Master node: ip-26-0-160-103'
+Master node: ip-26-0-160-103
++ echo 'All nodes: ip-26-0-160-103'
+All nodes: ip-26-0-160-103
++ echo 'World size: 8'
+World size: 8
++ srun torchrun --nnodes=1 --nproc_per_node=8 --rdzv_id=13404177 --rdzv_backend=c10d --rdzv_endpoint=ip-26-0-160-103:12356 run_train.py --config-file benchmark/configs/config_1.14G_dp4_tp2_pp1_acc256_mbs2_seq2048_zero1_tpmodeRED_vocab32k.yaml
+[2024-12-13 03:05:38,857] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-13 03:05:38,857] torch.distributed.run: [WARNING] 
+[2024-12-13 03:05:38,857] torch.distributed.run: [WARNING] *****************************************
+[2024-12-13 03:05:38,857] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-13 03:05:38,857] torch.distributed.run: [WARNING] *****************************************
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+[2024-12-13 03:05:44,103] torch.distributed.elastic.multiprocessing.api: [ERROR] failed (exitcode: 2) local_rank: 0 (pid: 2289649) of binary: /fsx/nouamane/miniconda/envs/2-1-cu121/bin/python
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 33, in <module>
+    sys.exit(load_entry_point('torch==2.1.1', 'console_scripts', 'torchrun')())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 264, in launch_agent
+    raise ChildFailedError(
+torch.distributed.elastic.multiprocessing.errors.ChildFailedError: 
+============================================================
+run_train.py FAILED
+------------------------------------------------------------
+Failures:
+[1]:
+  time      : 2024-12-13_03:05:44
+  host      : ip-26-0-160-103.ec2.internal
+  rank      : 1 (local_rank: 1)
+  exitcode  : 2 (pid: 2289650)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[2]:
+  time      : 2024-12-13_03:05:44
+  host      : ip-26-0-160-103.ec2.internal
+  rank      : 2 (local_rank: 2)
+  exitcode  : 2 (pid: 2289651)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[3]:
+  time      : 2024-12-13_03:05:44
+  host      : ip-26-0-160-103.ec2.internal
+  rank      : 3 (local_rank: 3)
+  exitcode  : 2 (pid: 2289652)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[4]:
+  time      : 2024-12-13_03:05:44
+  host      : ip-26-0-160-103.ec2.internal
+  rank      : 4 (local_rank: 4)
+  exitcode  : 2 (pid: 2289653)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[5]:
+  time      : 2024-12-13_03:05:44
+  host      : ip-26-0-160-103.ec2.internal
+  rank      : 5 (local_rank: 5)
+  exitcode  : 2 (pid: 2289654)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[6]:
+  time      : 2024-12-13_03:05:44
+  host      : ip-26-0-160-103.ec2.internal
+  rank      : 6 (local_rank: 6)
+  exitcode  : 2 (pid: 2289655)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[7]:
+  time      : 2024-12-13_03:05:44
+  host      : ip-26-0-160-103.ec2.internal
+  rank      : 7 (local_rank: 7)
+  exitcode  : 2 (pid: 2289656)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+------------------------------------------------------------
+Root Cause (first observed failure):
+[0]:
+  time      : 2024-12-13_03:05:44
+  host      : ip-26-0-160-103.ec2.internal
+  rank      : 0 (local_rank: 0)
+  exitcode  : 2 (pid: 2289649)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+============================================================
+srun: error: ip-26-0-160-103: task 0: Exited with exit code 1
diff --git a/logs/13404198-bench_stress_test.out b/logs/13404198-bench_stress_test.out
new file mode 100644
index 0000000000000000000000000000000000000000..baf3a9a7fee7bf399c6ed0ccffb2f8c8068594aa
--- /dev/null
+++ b/logs/13404198-bench_stress_test.out
@@ -0,0 +1,518 @@
++ source /etc/profile.d/modules.sh
+++ . /usr/share/modules/init/bash
++++ unset _mlshdbg
++++ '[' 0 = 1 ']'
++++ unset _mlre _mlIFS
++++ '[' -n x ']'
++++ _mlIFS=' 	
+'
++++ IFS=' '
++++ '[' -n '' ']'
+++++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash autoinit
++++ _mlcode='module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+MODULEPATH_modshare=/etc/environment-modules/modules:1:/usr/share/modules/\$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1; export MODULEPATH_modshare;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+LOADEDMODULES=; export LOADEDMODULES;
+MODULEPATH=/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/\$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles; export MODULEPATH;
+test 0;'
++++ _mlret=0
++++ '[' -n x ']'
++++ IFS=' 	
+'
++++ unset _mlIFS
++++ unset _mlre _mlv _mlrv
++++ '[' 0 -eq 0 ']'
++++ eval 'module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+MODULEPATH_modshare=/etc/environment-modules/modules:1:/usr/share/modules/\$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1; export MODULEPATH_modshare;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+LOADEDMODULES=; export LOADEDMODULES;
+MODULEPATH=/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/\$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles; export MODULEPATH;
+test 0;'
+++++ MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl
+++++ export MODULES_CMD
+++++ ENV=/usr/share/modules/init/profile.sh
+++++ export ENV
+++++ MODULEPATH_modshare='/etc/environment-modules/modules:1:/usr/share/modules/$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1'
+++++ export MODULEPATH_modshare
+++++ BASH_ENV=/usr/share/modules/init/bash
+++++ export BASH_ENV
+++++ MODULESHOME=/usr/share/modules
+++++ export MODULESHOME
+++++ LOADEDMODULES=
+++++ export LOADEDMODULES
+++++ MODULEPATH='/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles'
+++++ export MODULEPATH
+++++ test 0
++++ '[' 0 = 1 ']'
++++ '[' -t 2 ']'
++++ export -f module
++++ export -f switchml
++++ '[' 5 -ge 3 ']'
++++ [[ ehxB =~ i ]]
++++ [[ ! :/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin: =~ :/usr/bin: ]]
++++ '[' '!' -n '' ']'
++++ MANPATH=:
++++ export MANPATH
+++++ manpath
++++ [[ ! :/admin/home/nouamane/.local/share/man:/fsx/nouamane/miniconda/envs/2-1-cu121/man:/fsx/nouamane/miniconda/envs/2-1-cu121/share/man:/fsx/nouamane/miniconda/man:/fsx/nouamane/miniconda/share/man:/opt/amazon/openmpi/share/man:/opt/amazon/efa/share/man:/opt/slurm/share/man:/usr/local/man:/usr/local/share/man:/usr/share/man:/admin/home/nouamane/.fzf/man:: =~ :/usr/share/man: ]]
++++ unset _mlcode _mlret
++++ '[' -n '' ']'
++ module load cuda/12.1
++ unset _mlshdbg
++ '[' 0 = 1 ']'
++ unset _mlre _mlIFS
++ '[' -n x ']'
++ _mlIFS=' 	
+'
++ IFS=' '
++ '[' -n '' ']'
+++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash load cuda/12.1
++ eval 'CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include;' export 'CPATH;
+LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:;' export 'LD_LIBRARY_PATH;
+FI_EFA_FORK_SAFE=1;' export 'FI_EFA_FORK_SAFE;
+MANPATH=/usr/local/cuda-12.1/share/man::;' export 'MANPATH;
+LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64;' export 'LIBRARY_PATH;
+_LMFILES_=/usr/share/modules/modulefiles/cuda/12.1;' export '_LMFILES_;
+LOADEDMODULES=cuda/12.1;' export 'LOADEDMODULES;
+MPI_PATH=/opt/amazon/openmpi;' export 'MPI_PATH;
+NCCL_HOME_modshare=/opt/nccl/build:1;' export 'NCCL_HOME_modshare;
+NCCL_PROTO=simple;' export 'NCCL_PROTO;
+MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1;' export 'MANPATH_modshare;
+LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1;' export 'LIBRARY_PATH_modshare;
+NCCL_SOCKET_IFNAME=enp;' export 'NCCL_SOCKET_IFNAME;
+AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl;' export 'AWS_OFI_NCCL_HOME;
+NCCL_HOME=/opt/nccl/build;' export 'NCCL_HOME;
+FI_PROVIDER=efa;' export 'FI_PROVIDER;
+AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1;' export 'AWS_OFI_NCCL_HOME_modshare;
+CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1;' export 'CPATH_modshare;
+LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2::1:/opt/nccl/build/lib:1:/opt/aws-ofi-nccl/lib:2:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/efa/lib:1;' export 'LD_LIBRARY_PATH_modshare;
+FI_EFA_ENABLE_SHM_TRANSFER=1;' export 'FI_EFA_ENABLE_SHM_TRANSFER;
+_LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1;' export '_LMFILES__modshare;
+LOADEDMODULES_modshare=cuda/12.1:1;' export 'LOADEDMODULES_modshare;
+MPI_PATH_modshare=/opt/amazon/openmpi:1;' export 'MPI_PATH_modshare;
+PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin;' export 'PATH;
+CUDA_HOME=/usr/local/cuda-12.1;' export 'CUDA_HOME;
+PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/usr/local/cuda-12.1/include:1:/opt/amazon/efa/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/usr/sbin:1:/fsx/nouamane/miniconda/bin:1:/opt/slurm/bin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:2:/opt/amazon/openmpi/bin:1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1;' export 'PATH_modshare;
+test' '0;'
+++ CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include
+++ export CPATH
+++ LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:
+++ export LD_LIBRARY_PATH
+++ FI_EFA_FORK_SAFE=1
+++ export FI_EFA_FORK_SAFE
+++ MANPATH=/usr/local/cuda-12.1/share/man::
+++ export MANPATH
+++ LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64
+++ export LIBRARY_PATH
+++ _LMFILES_=/usr/share/modules/modulefiles/cuda/12.1
+++ export _LMFILES_
+++ LOADEDMODULES=cuda/12.1
+++ export LOADEDMODULES
+++ MPI_PATH=/opt/amazon/openmpi
+++ export MPI_PATH
+++ NCCL_HOME_modshare=/opt/nccl/build:1
+++ export NCCL_HOME_modshare
+++ NCCL_PROTO=simple
+++ export NCCL_PROTO
+++ MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1
+++ export MANPATH_modshare
+++ LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1
+++ export LIBRARY_PATH_modshare
+++ NCCL_SOCKET_IFNAME=enp
+++ export NCCL_SOCKET_IFNAME
+++ AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl
+++ export AWS_OFI_NCCL_HOME
+++ NCCL_HOME=/opt/nccl/build
+++ export NCCL_HOME
+++ FI_PROVIDER=efa
+++ export FI_PROVIDER
+++ AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1
+++ export AWS_OFI_NCCL_HOME_modshare
+++ CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1
+++ export CPATH_modshare
+++ LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2::1:/opt/nccl/build/lib:1:/opt/aws-ofi-nccl/lib:2:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/efa/lib:1
+++ export LD_LIBRARY_PATH_modshare
+++ FI_EFA_ENABLE_SHM_TRANSFER=1
+++ export FI_EFA_ENABLE_SHM_TRANSFER
+++ _LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1
+++ export _LMFILES__modshare
+++ LOADEDMODULES_modshare=cuda/12.1:1
+++ export LOADEDMODULES_modshare
+++ MPI_PATH_modshare=/opt/amazon/openmpi:1
+++ export MPI_PATH_modshare
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ export PATH
+++ CUDA_HOME=/usr/local/cuda-12.1
+++ export CUDA_HOME
+++ PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/usr/local/cuda-12.1/include:1:/opt/amazon/efa/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/usr/sbin:1:/fsx/nouamane/miniconda/bin:1:/opt/slurm/bin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:2:/opt/amazon/openmpi/bin:1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1
+++ export PATH_modshare
+++ test 0
++ _mlstatus=0
++ '[' -n x ']'
++ IFS=' 	
+'
++ unset _mlre _mlv _mlrv _mlIFS
++ '[' -n '' ']'
++ unset _mlshdbg
++ return 0
++ source /fsx/nouamane/miniconda/bin/activate
+++ _CONDA_ROOT=/fsx/nouamane/miniconda
+++ . /fsx/nouamane/miniconda/etc/profile.d/conda.sh
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ '[' -z x ']'
+++ conda activate
+++ local cmd=activate
+++ case "$cmd" in
+++ __conda_activate activate
+++ '[' -n '' ']'
+++ local ask_conda
++++ PS1=
++++ __conda_exe shell.posix activate
++++ /fsx/nouamane/miniconda/bin/conda shell.posix activate
+++ ask_conda='. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_2='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\'''
+++ eval '. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_2='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\'''
++++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh
+++++ test -n ''
+++++ unset XML_CATALOG_FILES
+++++ unset xml_catalog_files_libxml2
++++ PS1='(base) '
++++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++++ export CONDA_PREFIX=/fsx/nouamane/miniconda
++++ CONDA_PREFIX=/fsx/nouamane/miniconda
++++ export CONDA_SHLVL=3
++++ CONDA_SHLVL=3
++++ export CONDA_DEFAULT_ENV=base
++++ CONDA_DEFAULT_ENV=base
++++ export 'CONDA_PROMPT_MODIFIER=(base) '
++++ CONDA_PROMPT_MODIFIER='(base) '
++++ export CONDA_PREFIX_2=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ CONDA_PREFIX_2=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ __conda_hashr
+++ '[' -n '' ']'
+++ '[' -n '' ']'
+++ hash -r
++ conda activate 2-1-cu121
++ local cmd=activate
++ case "$cmd" in
++ __conda_activate activate 2-1-cu121
++ '[' -n '' ']'
++ local ask_conda
+++ PS1='(base) '
+++ __conda_exe shell.posix activate 2-1-cu121
+++ /fsx/nouamane/miniconda/bin/conda shell.posix activate 2-1-cu121
++ ask_conda='PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''4'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_3='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
++ eval 'PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''4'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_3='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
+++ PS1='(2-1-cu121) '
+++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ export CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ export CONDA_SHLVL=4
+++ CONDA_SHLVL=4
+++ export CONDA_DEFAULT_ENV=2-1-cu121
+++ CONDA_DEFAULT_ENV=2-1-cu121
+++ export 'CONDA_PROMPT_MODIFIER=(2-1-cu121) '
+++ CONDA_PROMPT_MODIFIER='(2-1-cu121) '
+++ export CONDA_PREFIX_3=/fsx/nouamane/miniconda
+++ CONDA_PREFIX_3=/fsx/nouamane/miniconda
+++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ export _CE_M=
+++ _CE_M=
+++ export _CE_CONDA=
+++ _CE_CONDA=
+++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh
++++ test -n ''
++++ xml_catalog_files_libxml2=
++++ XML_CATALOG_FILES=
++++ conda_catalog_files=
++++ ifs_libxml2=' 	
+'
++++ IFS=' '
++++ rem=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ for pre in ${rem}
++++ test '' = /fsx/nouamane/miniconda/envs/2-1-cu121
++++ conda_catalog_files=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ rem=
++++ IFS=' 	
+'
++++ conda_catalog_files='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ export 'XML_CATALOG_FILES=file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ XML_CATALOG_FILES='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ unset conda_catalog_files ifs_libxml2 rem
++ __conda_hashr
++ '[' -n '' ']'
++ '[' -n '' ']'
++ hash -r
++ export PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++ PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ scontrol show hostnames ip-26-0-169-207
++ export NODELIST=ip-26-0-169-207
++ NODELIST=ip-26-0-169-207
+++ scontrol show hostnames ip-26-0-169-207
+++ head -n1
++ export MASTER_NODE=ip-26-0-169-207
++ MASTER_NODE=ip-26-0-169-207
++ export MASTER_PORT=12356
++ MASTER_PORT=12356
++ export NNODES=1
++ NNODES=1
++ export GPUS_PER_NODE=8
++ GPUS_PER_NODE=8
++ export WORLD_SIZE=8
++ WORLD_SIZE=8
++ export CUDA_DEVICE_MAX_CONNECTIONS=1
++ CUDA_DEVICE_MAX_CONNECTIONS=1
++ export NCCL_DEBUG=INFO
++ NCCL_DEBUG=INFO
++ export NANOTRON_BENCHMARK=1
++ NANOTRON_BENCHMARK=1
++ echo 'Master node: ip-26-0-169-207'
+Master node: ip-26-0-169-207
++ echo 'All nodes: ip-26-0-169-207'
+All nodes: ip-26-0-169-207
++ echo 'World size: 8'
+World size: 8
++ srun torchrun --nnodes=1 --nproc_per_node=8 --rdzv_id=13404198 --rdzv_backend=c10d --rdzv_endpoint=ip-26-0-169-207:12356 run_train.py --config-file benchmark/configs/config_1.14G_dp4_tp2_pp1_acc256_mbs2_seq2048_zero1_tpmodeRED_vocab32k.yaml
+[2024-12-13 03:08:39,258] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-13 03:08:39,258] torch.distributed.run: [WARNING] 
+[2024-12-13 03:08:39,258] torch.distributed.run: [WARNING] *****************************************
+[2024-12-13 03:08:39,258] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-13 03:08:39,258] torch.distributed.run: [WARNING] *****************************************
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+[2024-12-13 03:08:44,567] torch.distributed.elastic.multiprocessing.api: [ERROR] failed (exitcode: 2) local_rank: 0 (pid: 402832) of binary: /fsx/nouamane/miniconda/envs/2-1-cu121/bin/python
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 33, in <module>
+    sys.exit(load_entry_point('torch==2.1.1', 'console_scripts', 'torchrun')())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 264, in launch_agent
+    raise ChildFailedError(
+torch.distributed.elastic.multiprocessing.errors.ChildFailedError: 
+============================================================
+run_train.py FAILED
+------------------------------------------------------------
+Failures:
+[1]:
+  time      : 2024-12-13_03:08:44
+  host      : ip-26-0-169-207.ec2.internal
+  rank      : 1 (local_rank: 1)
+  exitcode  : 2 (pid: 402833)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[2]:
+  time      : 2024-12-13_03:08:44
+  host      : ip-26-0-169-207.ec2.internal
+  rank      : 2 (local_rank: 2)
+  exitcode  : 2 (pid: 402834)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[3]:
+  time      : 2024-12-13_03:08:44
+  host      : ip-26-0-169-207.ec2.internal
+  rank      : 3 (local_rank: 3)
+  exitcode  : 2 (pid: 402835)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[4]:
+  time      : 2024-12-13_03:08:44
+  host      : ip-26-0-169-207.ec2.internal
+  rank      : 4 (local_rank: 4)
+  exitcode  : 2 (pid: 402836)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[5]:
+  time      : 2024-12-13_03:08:44
+  host      : ip-26-0-169-207.ec2.internal
+  rank      : 5 (local_rank: 5)
+  exitcode  : 2 (pid: 402837)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[6]:
+  time      : 2024-12-13_03:08:44
+  host      : ip-26-0-169-207.ec2.internal
+  rank      : 6 (local_rank: 6)
+  exitcode  : 2 (pid: 402838)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[7]:
+  time      : 2024-12-13_03:08:44
+  host      : ip-26-0-169-207.ec2.internal
+  rank      : 7 (local_rank: 7)
+  exitcode  : 2 (pid: 402839)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+------------------------------------------------------------
+Root Cause (first observed failure):
+[0]:
+  time      : 2024-12-13_03:08:44
+  host      : ip-26-0-169-207.ec2.internal
+  rank      : 0 (local_rank: 0)
+  exitcode  : 2 (pid: 402832)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+============================================================
+srun: error: ip-26-0-169-207: task 0: Exited with exit code 1
diff --git a/logs/13404345-bench_stress_test.out b/logs/13404345-bench_stress_test.out
new file mode 100644
index 0000000000000000000000000000000000000000..a6eccb411ba33ef7026382c994bf329b92825391
--- /dev/null
+++ b/logs/13404345-bench_stress_test.out
@@ -0,0 +1,518 @@
++ source /etc/profile.d/modules.sh
+++ . /usr/share/modules/init/bash
++++ unset _mlshdbg
++++ '[' 0 = 1 ']'
++++ unset _mlre _mlIFS
++++ '[' -n x ']'
++++ _mlIFS=' 	
+'
++++ IFS=' '
++++ '[' -n '' ']'
+++++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash autoinit
++++ _mlcode='module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+MODULEPATH_modshare=/etc/environment-modules/modules:1:/usr/share/modules/\$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1; export MODULEPATH_modshare;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+LOADEDMODULES=; export LOADEDMODULES;
+MODULEPATH=/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/\$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles; export MODULEPATH;
+test 0;'
++++ _mlret=0
++++ '[' -n x ']'
++++ IFS=' 	
+'
++++ unset _mlIFS
++++ unset _mlre _mlv _mlrv
++++ '[' 0 -eq 0 ']'
++++ eval 'module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+MODULEPATH_modshare=/etc/environment-modules/modules:1:/usr/share/modules/\$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1; export MODULEPATH_modshare;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+LOADEDMODULES=; export LOADEDMODULES;
+MODULEPATH=/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/\$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles; export MODULEPATH;
+test 0;'
+++++ MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl
+++++ export MODULES_CMD
+++++ ENV=/usr/share/modules/init/profile.sh
+++++ export ENV
+++++ MODULEPATH_modshare='/etc/environment-modules/modules:1:/usr/share/modules/$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1'
+++++ export MODULEPATH_modshare
+++++ BASH_ENV=/usr/share/modules/init/bash
+++++ export BASH_ENV
+++++ MODULESHOME=/usr/share/modules
+++++ export MODULESHOME
+++++ LOADEDMODULES=
+++++ export LOADEDMODULES
+++++ MODULEPATH='/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles'
+++++ export MODULEPATH
+++++ test 0
++++ '[' 0 = 1 ']'
++++ '[' -t 2 ']'
++++ export -f module
++++ export -f switchml
++++ '[' 5 -ge 3 ']'
++++ [[ ehxB =~ i ]]
++++ [[ ! :/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin: =~ :/usr/bin: ]]
++++ '[' '!' -n '' ']'
++++ MANPATH=:
++++ export MANPATH
+++++ manpath
++++ [[ ! :/admin/home/nouamane/.local/share/man:/fsx/nouamane/miniconda/envs/2-1-cu121/man:/fsx/nouamane/miniconda/envs/2-1-cu121/share/man:/fsx/nouamane/miniconda/man:/fsx/nouamane/miniconda/share/man:/opt/amazon/openmpi/share/man:/opt/amazon/efa/share/man:/opt/slurm/share/man:/usr/local/man:/usr/local/share/man:/usr/share/man:/admin/home/nouamane/.fzf/man:: =~ :/usr/share/man: ]]
++++ unset _mlcode _mlret
++++ '[' -n '' ']'
++ module load cuda/12.1
++ unset _mlshdbg
++ '[' 0 = 1 ']'
++ unset _mlre _mlIFS
++ '[' -n x ']'
++ _mlIFS=' 	
+'
++ IFS=' '
++ '[' -n '' ']'
+++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash load cuda/12.1
++ eval 'CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include;' export 'CPATH;
+LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:;' export 'LD_LIBRARY_PATH;
+FI_EFA_FORK_SAFE=1;' export 'FI_EFA_FORK_SAFE;
+MANPATH=/usr/local/cuda-12.1/share/man::;' export 'MANPATH;
+LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64;' export 'LIBRARY_PATH;
+_LMFILES_=/usr/share/modules/modulefiles/cuda/12.1;' export '_LMFILES_;
+LOADEDMODULES=cuda/12.1;' export 'LOADEDMODULES;
+MPI_PATH=/opt/amazon/openmpi;' export 'MPI_PATH;
+NCCL_HOME_modshare=/opt/nccl/build:1;' export 'NCCL_HOME_modshare;
+NCCL_PROTO=simple;' export 'NCCL_PROTO;
+MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1;' export 'MANPATH_modshare;
+LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1;' export 'LIBRARY_PATH_modshare;
+NCCL_SOCKET_IFNAME=enp;' export 'NCCL_SOCKET_IFNAME;
+AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl;' export 'AWS_OFI_NCCL_HOME;
+NCCL_HOME=/opt/nccl/build;' export 'NCCL_HOME;
+FI_PROVIDER=efa;' export 'FI_PROVIDER;
+AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1;' export 'AWS_OFI_NCCL_HOME_modshare;
+CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1;' export 'CPATH_modshare;
+LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2::1:/opt/nccl/build/lib:1:/opt/aws-ofi-nccl/lib:2:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/efa/lib:1;' export 'LD_LIBRARY_PATH_modshare;
+FI_EFA_ENABLE_SHM_TRANSFER=1;' export 'FI_EFA_ENABLE_SHM_TRANSFER;
+_LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1;' export '_LMFILES__modshare;
+LOADEDMODULES_modshare=cuda/12.1:1;' export 'LOADEDMODULES_modshare;
+MPI_PATH_modshare=/opt/amazon/openmpi:1;' export 'MPI_PATH_modshare;
+PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin;' export 'PATH;
+CUDA_HOME=/usr/local/cuda-12.1;' export 'CUDA_HOME;
+PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/usr/local/cuda-12.1/include:1:/opt/amazon/efa/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/usr/sbin:1:/fsx/nouamane/miniconda/bin:1:/opt/slurm/bin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:2:/opt/amazon/openmpi/bin:1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1;' export 'PATH_modshare;
+test' '0;'
+++ CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include
+++ export CPATH
+++ LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:
+++ export LD_LIBRARY_PATH
+++ FI_EFA_FORK_SAFE=1
+++ export FI_EFA_FORK_SAFE
+++ MANPATH=/usr/local/cuda-12.1/share/man::
+++ export MANPATH
+++ LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64
+++ export LIBRARY_PATH
+++ _LMFILES_=/usr/share/modules/modulefiles/cuda/12.1
+++ export _LMFILES_
+++ LOADEDMODULES=cuda/12.1
+++ export LOADEDMODULES
+++ MPI_PATH=/opt/amazon/openmpi
+++ export MPI_PATH
+++ NCCL_HOME_modshare=/opt/nccl/build:1
+++ export NCCL_HOME_modshare
+++ NCCL_PROTO=simple
+++ export NCCL_PROTO
+++ MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1
+++ export MANPATH_modshare
+++ LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1
+++ export LIBRARY_PATH_modshare
+++ NCCL_SOCKET_IFNAME=enp
+++ export NCCL_SOCKET_IFNAME
+++ AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl
+++ export AWS_OFI_NCCL_HOME
+++ NCCL_HOME=/opt/nccl/build
+++ export NCCL_HOME
+++ FI_PROVIDER=efa
+++ export FI_PROVIDER
+++ AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1
+++ export AWS_OFI_NCCL_HOME_modshare
+++ CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1
+++ export CPATH_modshare
+++ LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2::1:/opt/nccl/build/lib:1:/opt/aws-ofi-nccl/lib:2:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/efa/lib:1
+++ export LD_LIBRARY_PATH_modshare
+++ FI_EFA_ENABLE_SHM_TRANSFER=1
+++ export FI_EFA_ENABLE_SHM_TRANSFER
+++ _LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1
+++ export _LMFILES__modshare
+++ LOADEDMODULES_modshare=cuda/12.1:1
+++ export LOADEDMODULES_modshare
+++ MPI_PATH_modshare=/opt/amazon/openmpi:1
+++ export MPI_PATH_modshare
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ export PATH
+++ CUDA_HOME=/usr/local/cuda-12.1
+++ export CUDA_HOME
+++ PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/usr/local/cuda-12.1/include:1:/opt/amazon/efa/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/usr/sbin:1:/fsx/nouamane/miniconda/bin:1:/opt/slurm/bin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:2:/opt/amazon/openmpi/bin:1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1
+++ export PATH_modshare
+++ test 0
++ _mlstatus=0
++ '[' -n x ']'
++ IFS=' 	
+'
++ unset _mlre _mlv _mlrv _mlIFS
++ '[' -n '' ']'
++ unset _mlshdbg
++ return 0
++ source /fsx/nouamane/miniconda/bin/activate
+++ _CONDA_ROOT=/fsx/nouamane/miniconda
+++ . /fsx/nouamane/miniconda/etc/profile.d/conda.sh
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ '[' -z x ']'
+++ conda activate
+++ local cmd=activate
+++ case "$cmd" in
+++ __conda_activate activate
+++ '[' -n '' ']'
+++ local ask_conda
++++ PS1=
++++ __conda_exe shell.posix activate
++++ /fsx/nouamane/miniconda/bin/conda shell.posix activate
+++ ask_conda='. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_2='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\'''
+++ eval '. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_2='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\'''
++++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh
+++++ test -n ''
+++++ unset XML_CATALOG_FILES
+++++ unset xml_catalog_files_libxml2
++++ PS1='(base) '
++++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++++ export CONDA_PREFIX=/fsx/nouamane/miniconda
++++ CONDA_PREFIX=/fsx/nouamane/miniconda
++++ export CONDA_SHLVL=3
++++ CONDA_SHLVL=3
++++ export CONDA_DEFAULT_ENV=base
++++ CONDA_DEFAULT_ENV=base
++++ export 'CONDA_PROMPT_MODIFIER=(base) '
++++ CONDA_PROMPT_MODIFIER='(base) '
++++ export CONDA_PREFIX_2=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ CONDA_PREFIX_2=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ __conda_hashr
+++ '[' -n '' ']'
+++ '[' -n '' ']'
+++ hash -r
++ conda activate 2-1-cu121
++ local cmd=activate
++ case "$cmd" in
++ __conda_activate activate 2-1-cu121
++ '[' -n '' ']'
++ local ask_conda
+++ PS1='(base) '
+++ __conda_exe shell.posix activate 2-1-cu121
+++ /fsx/nouamane/miniconda/bin/conda shell.posix activate 2-1-cu121
++ ask_conda='PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''4'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_3='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
++ eval 'PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''4'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_3='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
+++ PS1='(2-1-cu121) '
+++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ export CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ export CONDA_SHLVL=4
+++ CONDA_SHLVL=4
+++ export CONDA_DEFAULT_ENV=2-1-cu121
+++ CONDA_DEFAULT_ENV=2-1-cu121
+++ export 'CONDA_PROMPT_MODIFIER=(2-1-cu121) '
+++ CONDA_PROMPT_MODIFIER='(2-1-cu121) '
+++ export CONDA_PREFIX_3=/fsx/nouamane/miniconda
+++ CONDA_PREFIX_3=/fsx/nouamane/miniconda
+++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ export _CE_M=
+++ _CE_M=
+++ export _CE_CONDA=
+++ _CE_CONDA=
+++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh
++++ test -n ''
++++ xml_catalog_files_libxml2=
++++ XML_CATALOG_FILES=
++++ conda_catalog_files=
++++ ifs_libxml2=' 	
+'
++++ IFS=' '
++++ rem=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ for pre in ${rem}
++++ test '' = /fsx/nouamane/miniconda/envs/2-1-cu121
++++ conda_catalog_files=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ rem=
++++ IFS=' 	
+'
++++ conda_catalog_files='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ export 'XML_CATALOG_FILES=file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ XML_CATALOG_FILES='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ unset conda_catalog_files ifs_libxml2 rem
++ __conda_hashr
++ '[' -n '' ']'
++ '[' -n '' ']'
++ hash -r
++ export PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++ PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ scontrol show hostnames ip-26-0-160-103
++ export NODELIST=ip-26-0-160-103
++ NODELIST=ip-26-0-160-103
+++ scontrol show hostnames ip-26-0-160-103
+++ head -n1
++ export MASTER_NODE=ip-26-0-160-103
++ MASTER_NODE=ip-26-0-160-103
++ export MASTER_PORT=12356
++ MASTER_PORT=12356
++ export NNODES=1
++ NNODES=1
++ export GPUS_PER_NODE=8
++ GPUS_PER_NODE=8
++ export WORLD_SIZE=8
++ WORLD_SIZE=8
++ export CUDA_DEVICE_MAX_CONNECTIONS=1
++ CUDA_DEVICE_MAX_CONNECTIONS=1
++ export NCCL_DEBUG=INFO
++ NCCL_DEBUG=INFO
++ export NANOTRON_BENCHMARK=1
++ NANOTRON_BENCHMARK=1
++ echo 'Master node: ip-26-0-160-103'
+Master node: ip-26-0-160-103
++ echo 'All nodes: ip-26-0-160-103'
+All nodes: ip-26-0-160-103
++ echo 'World size: 8'
+World size: 8
++ srun torchrun --nnodes=1 --nproc_per_node=8 --rdzv_id=13404345 --rdzv_backend=c10d --rdzv_endpoint=ip-26-0-160-103:12356 run_train.py --config-file benchmark/configs/config_1.14G_dp4_tp2_pp1_acc256_mbs2_seq2048_zero1_tpmodeRED_vocab32k.yaml
+[2024-12-13 03:31:41,908] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-13 03:31:41,908] torch.distributed.run: [WARNING] 
+[2024-12-13 03:31:41,908] torch.distributed.run: [WARNING] *****************************************
+[2024-12-13 03:31:41,908] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-13 03:31:41,908] torch.distributed.run: [WARNING] *****************************************
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+[2024-12-13 03:31:46,984] torch.distributed.elastic.multiprocessing.api: [ERROR] failed (exitcode: 2) local_rank: 0 (pid: 2315819) of binary: /fsx/nouamane/miniconda/envs/2-1-cu121/bin/python
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 33, in <module>
+    sys.exit(load_entry_point('torch==2.1.1', 'console_scripts', 'torchrun')())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 264, in launch_agent
+    raise ChildFailedError(
+torch.distributed.elastic.multiprocessing.errors.ChildFailedError: 
+============================================================
+run_train.py FAILED
+------------------------------------------------------------
+Failures:
+[1]:
+  time      : 2024-12-13_03:31:46
+  host      : ip-26-0-160-103.ec2.internal
+  rank      : 1 (local_rank: 1)
+  exitcode  : 2 (pid: 2315820)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[2]:
+  time      : 2024-12-13_03:31:46
+  host      : ip-26-0-160-103.ec2.internal
+  rank      : 2 (local_rank: 2)
+  exitcode  : 2 (pid: 2315821)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[3]:
+  time      : 2024-12-13_03:31:46
+  host      : ip-26-0-160-103.ec2.internal
+  rank      : 3 (local_rank: 3)
+  exitcode  : 2 (pid: 2315822)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[4]:
+  time      : 2024-12-13_03:31:46
+  host      : ip-26-0-160-103.ec2.internal
+  rank      : 4 (local_rank: 4)
+  exitcode  : 2 (pid: 2315823)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[5]:
+  time      : 2024-12-13_03:31:46
+  host      : ip-26-0-160-103.ec2.internal
+  rank      : 5 (local_rank: 5)
+  exitcode  : 2 (pid: 2315824)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[6]:
+  time      : 2024-12-13_03:31:46
+  host      : ip-26-0-160-103.ec2.internal
+  rank      : 6 (local_rank: 6)
+  exitcode  : 2 (pid: 2315825)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[7]:
+  time      : 2024-12-13_03:31:46
+  host      : ip-26-0-160-103.ec2.internal
+  rank      : 7 (local_rank: 7)
+  exitcode  : 2 (pid: 2315826)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+------------------------------------------------------------
+Root Cause (first observed failure):
+[0]:
+  time      : 2024-12-13_03:31:46
+  host      : ip-26-0-160-103.ec2.internal
+  rank      : 0 (local_rank: 0)
+  exitcode  : 2 (pid: 2315819)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+============================================================
+srun: error: ip-26-0-160-103: task 0: Exited with exit code 1
diff --git a/logs/13404452-bench_stress_test.out b/logs/13404452-bench_stress_test.out
new file mode 100644
index 0000000000000000000000000000000000000000..1944107a0b5ee052da46450d9b3938d1dbabd934
--- /dev/null
+++ b/logs/13404452-bench_stress_test.out
@@ -0,0 +1,518 @@
++ source /etc/profile.d/modules.sh
+++ . /usr/share/modules/init/bash
++++ unset _mlshdbg
++++ '[' 0 = 1 ']'
++++ unset _mlre _mlIFS
++++ '[' -n x ']'
++++ _mlIFS=' 	
+'
++++ IFS=' '
++++ '[' -n '' ']'
+++++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash autoinit
++++ _mlcode='module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+MODULEPATH_modshare=/etc/environment-modules/modules:1:/usr/share/modules/\$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1; export MODULEPATH_modshare;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+LOADEDMODULES=; export LOADEDMODULES;
+MODULEPATH=/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/\$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles; export MODULEPATH;
+test 0;'
++++ _mlret=0
++++ '[' -n x ']'
++++ IFS=' 	
+'
++++ unset _mlIFS
++++ unset _mlre _mlv _mlrv
++++ '[' 0 -eq 0 ']'
++++ eval 'module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+MODULEPATH_modshare=/etc/environment-modules/modules:1:/usr/share/modules/\$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1; export MODULEPATH_modshare;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+LOADEDMODULES=; export LOADEDMODULES;
+MODULEPATH=/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/\$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles; export MODULEPATH;
+test 0;'
+++++ MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl
+++++ export MODULES_CMD
+++++ ENV=/usr/share/modules/init/profile.sh
+++++ export ENV
+++++ MODULEPATH_modshare='/etc/environment-modules/modules:1:/usr/share/modules/$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1'
+++++ export MODULEPATH_modshare
+++++ BASH_ENV=/usr/share/modules/init/bash
+++++ export BASH_ENV
+++++ MODULESHOME=/usr/share/modules
+++++ export MODULESHOME
+++++ LOADEDMODULES=
+++++ export LOADEDMODULES
+++++ MODULEPATH='/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles'
+++++ export MODULEPATH
+++++ test 0
++++ '[' 0 = 1 ']'
++++ '[' -t 2 ']'
++++ export -f module
++++ export -f switchml
++++ '[' 5 -ge 3 ']'
++++ [[ ehxB =~ i ]]
++++ [[ ! :/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin: =~ :/usr/bin: ]]
++++ '[' '!' -n '' ']'
++++ MANPATH=:
++++ export MANPATH
+++++ manpath
++++ [[ ! :/admin/home/nouamane/.local/share/man:/fsx/nouamane/miniconda/envs/2-1-cu121/man:/fsx/nouamane/miniconda/envs/2-1-cu121/share/man:/fsx/nouamane/miniconda/man:/fsx/nouamane/miniconda/share/man:/opt/amazon/openmpi/share/man:/opt/amazon/efa/share/man:/opt/slurm/share/man:/usr/local/man:/usr/local/share/man:/usr/share/man:/admin/home/nouamane/.fzf/man:: =~ :/usr/share/man: ]]
++++ unset _mlcode _mlret
++++ '[' -n '' ']'
++ module load cuda/12.1
++ unset _mlshdbg
++ '[' 0 = 1 ']'
++ unset _mlre _mlIFS
++ '[' -n x ']'
++ _mlIFS=' 	
+'
++ IFS=' '
++ '[' -n '' ']'
+++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash load cuda/12.1
++ eval 'CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include;' export 'CPATH;
+LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:;' export 'LD_LIBRARY_PATH;
+FI_EFA_FORK_SAFE=1;' export 'FI_EFA_FORK_SAFE;
+MANPATH=/usr/local/cuda-12.1/share/man::;' export 'MANPATH;
+LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64;' export 'LIBRARY_PATH;
+_LMFILES_=/usr/share/modules/modulefiles/cuda/12.1;' export '_LMFILES_;
+LOADEDMODULES=cuda/12.1;' export 'LOADEDMODULES;
+MPI_PATH=/opt/amazon/openmpi;' export 'MPI_PATH;
+NCCL_HOME_modshare=/opt/nccl/build:1;' export 'NCCL_HOME_modshare;
+NCCL_PROTO=simple;' export 'NCCL_PROTO;
+MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1;' export 'MANPATH_modshare;
+LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1;' export 'LIBRARY_PATH_modshare;
+NCCL_SOCKET_IFNAME=enp;' export 'NCCL_SOCKET_IFNAME;
+AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl;' export 'AWS_OFI_NCCL_HOME;
+NCCL_HOME=/opt/nccl/build;' export 'NCCL_HOME;
+FI_PROVIDER=efa;' export 'FI_PROVIDER;
+AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1;' export 'AWS_OFI_NCCL_HOME_modshare;
+CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1;' export 'CPATH_modshare;
+LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2::1:/opt/nccl/build/lib:1:/opt/aws-ofi-nccl/lib:2:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/efa/lib:1;' export 'LD_LIBRARY_PATH_modshare;
+FI_EFA_ENABLE_SHM_TRANSFER=1;' export 'FI_EFA_ENABLE_SHM_TRANSFER;
+_LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1;' export '_LMFILES__modshare;
+LOADEDMODULES_modshare=cuda/12.1:1;' export 'LOADEDMODULES_modshare;
+MPI_PATH_modshare=/opt/amazon/openmpi:1;' export 'MPI_PATH_modshare;
+PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin;' export 'PATH;
+CUDA_HOME=/usr/local/cuda-12.1;' export 'CUDA_HOME;
+PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/usr/local/cuda-12.1/include:1:/opt/amazon/efa/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/usr/sbin:1:/fsx/nouamane/miniconda/bin:1:/opt/slurm/bin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:2:/opt/amazon/openmpi/bin:1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1;' export 'PATH_modshare;
+test' '0;'
+++ CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include
+++ export CPATH
+++ LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:
+++ export LD_LIBRARY_PATH
+++ FI_EFA_FORK_SAFE=1
+++ export FI_EFA_FORK_SAFE
+++ MANPATH=/usr/local/cuda-12.1/share/man::
+++ export MANPATH
+++ LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64
+++ export LIBRARY_PATH
+++ _LMFILES_=/usr/share/modules/modulefiles/cuda/12.1
+++ export _LMFILES_
+++ LOADEDMODULES=cuda/12.1
+++ export LOADEDMODULES
+++ MPI_PATH=/opt/amazon/openmpi
+++ export MPI_PATH
+++ NCCL_HOME_modshare=/opt/nccl/build:1
+++ export NCCL_HOME_modshare
+++ NCCL_PROTO=simple
+++ export NCCL_PROTO
+++ MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1
+++ export MANPATH_modshare
+++ LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1
+++ export LIBRARY_PATH_modshare
+++ NCCL_SOCKET_IFNAME=enp
+++ export NCCL_SOCKET_IFNAME
+++ AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl
+++ export AWS_OFI_NCCL_HOME
+++ NCCL_HOME=/opt/nccl/build
+++ export NCCL_HOME
+++ FI_PROVIDER=efa
+++ export FI_PROVIDER
+++ AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1
+++ export AWS_OFI_NCCL_HOME_modshare
+++ CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1
+++ export CPATH_modshare
+++ LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2::1:/opt/nccl/build/lib:1:/opt/aws-ofi-nccl/lib:2:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/efa/lib:1
+++ export LD_LIBRARY_PATH_modshare
+++ FI_EFA_ENABLE_SHM_TRANSFER=1
+++ export FI_EFA_ENABLE_SHM_TRANSFER
+++ _LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1
+++ export _LMFILES__modshare
+++ LOADEDMODULES_modshare=cuda/12.1:1
+++ export LOADEDMODULES_modshare
+++ MPI_PATH_modshare=/opt/amazon/openmpi:1
+++ export MPI_PATH_modshare
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ export PATH
+++ CUDA_HOME=/usr/local/cuda-12.1
+++ export CUDA_HOME
+++ PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/usr/local/cuda-12.1/include:1:/opt/amazon/efa/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/usr/sbin:1:/fsx/nouamane/miniconda/bin:1:/opt/slurm/bin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:2:/opt/amazon/openmpi/bin:1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1
+++ export PATH_modshare
+++ test 0
++ _mlstatus=0
++ '[' -n x ']'
++ IFS=' 	
+'
++ unset _mlre _mlv _mlrv _mlIFS
++ '[' -n '' ']'
++ unset _mlshdbg
++ return 0
++ source /fsx/nouamane/miniconda/bin/activate
+++ _CONDA_ROOT=/fsx/nouamane/miniconda
+++ . /fsx/nouamane/miniconda/etc/profile.d/conda.sh
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ '[' -z x ']'
+++ conda activate
+++ local cmd=activate
+++ case "$cmd" in
+++ __conda_activate activate
+++ '[' -n '' ']'
+++ local ask_conda
++++ PS1=
++++ __conda_exe shell.posix activate
++++ /fsx/nouamane/miniconda/bin/conda shell.posix activate
+++ ask_conda='. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_2='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\'''
+++ eval '. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_2='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\'''
++++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh
+++++ test -n ''
+++++ unset XML_CATALOG_FILES
+++++ unset xml_catalog_files_libxml2
++++ PS1='(base) '
++++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++++ export CONDA_PREFIX=/fsx/nouamane/miniconda
++++ CONDA_PREFIX=/fsx/nouamane/miniconda
++++ export CONDA_SHLVL=3
++++ CONDA_SHLVL=3
++++ export CONDA_DEFAULT_ENV=base
++++ CONDA_DEFAULT_ENV=base
++++ export 'CONDA_PROMPT_MODIFIER=(base) '
++++ CONDA_PROMPT_MODIFIER='(base) '
++++ export CONDA_PREFIX_2=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ CONDA_PREFIX_2=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ __conda_hashr
+++ '[' -n '' ']'
+++ '[' -n '' ']'
+++ hash -r
++ conda activate 2-1-cu121
++ local cmd=activate
++ case "$cmd" in
++ __conda_activate activate 2-1-cu121
++ '[' -n '' ']'
++ local ask_conda
+++ PS1='(base) '
+++ __conda_exe shell.posix activate 2-1-cu121
+++ /fsx/nouamane/miniconda/bin/conda shell.posix activate 2-1-cu121
++ ask_conda='PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''4'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_3='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
++ eval 'PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''4'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_3='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
+++ PS1='(2-1-cu121) '
+++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ export CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ export CONDA_SHLVL=4
+++ CONDA_SHLVL=4
+++ export CONDA_DEFAULT_ENV=2-1-cu121
+++ CONDA_DEFAULT_ENV=2-1-cu121
+++ export 'CONDA_PROMPT_MODIFIER=(2-1-cu121) '
+++ CONDA_PROMPT_MODIFIER='(2-1-cu121) '
+++ export CONDA_PREFIX_3=/fsx/nouamane/miniconda
+++ CONDA_PREFIX_3=/fsx/nouamane/miniconda
+++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ export _CE_M=
+++ _CE_M=
+++ export _CE_CONDA=
+++ _CE_CONDA=
+++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh
++++ test -n ''
++++ xml_catalog_files_libxml2=
++++ XML_CATALOG_FILES=
++++ conda_catalog_files=
++++ ifs_libxml2=' 	
+'
++++ IFS=' '
++++ rem=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ for pre in ${rem}
++++ test '' = /fsx/nouamane/miniconda/envs/2-1-cu121
++++ conda_catalog_files=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ rem=
++++ IFS=' 	
+'
++++ conda_catalog_files='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ export 'XML_CATALOG_FILES=file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ XML_CATALOG_FILES='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ unset conda_catalog_files ifs_libxml2 rem
++ __conda_hashr
++ '[' -n '' ']'
++ '[' -n '' ']'
++ hash -r
++ export PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++ PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ scontrol show hostnames ip-26-0-160-103
++ export NODELIST=ip-26-0-160-103
++ NODELIST=ip-26-0-160-103
+++ scontrol show hostnames ip-26-0-160-103
+++ head -n1
++ export MASTER_NODE=ip-26-0-160-103
++ MASTER_NODE=ip-26-0-160-103
++ export MASTER_PORT=12356
++ MASTER_PORT=12356
++ export NNODES=1
++ NNODES=1
++ export GPUS_PER_NODE=8
++ GPUS_PER_NODE=8
++ export WORLD_SIZE=8
++ WORLD_SIZE=8
++ export CUDA_DEVICE_MAX_CONNECTIONS=1
++ CUDA_DEVICE_MAX_CONNECTIONS=1
++ export NCCL_DEBUG=INFO
++ NCCL_DEBUG=INFO
++ export NANOTRON_BENCHMARK=1
++ NANOTRON_BENCHMARK=1
++ echo 'Master node: ip-26-0-160-103'
+Master node: ip-26-0-160-103
++ echo 'All nodes: ip-26-0-160-103'
+All nodes: ip-26-0-160-103
++ echo 'World size: 8'
+World size: 8
++ srun torchrun --nnodes=1 --nproc_per_node=8 --rdzv_id=13404452 --rdzv_backend=c10d --rdzv_endpoint=ip-26-0-160-103:12356 run_train.py --config-file benchmark/configs/config_1.14G_dp4_tp2_pp1_acc256_mbs2_seq2048_zero1_tpmodeRED_vocab32k.yaml
+[2024-12-13 03:48:41,510] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-13 03:48:41,510] torch.distributed.run: [WARNING] 
+[2024-12-13 03:48:41,510] torch.distributed.run: [WARNING] *****************************************
+[2024-12-13 03:48:41,510] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-13 03:48:41,510] torch.distributed.run: [WARNING] *****************************************
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+[2024-12-13 03:48:46,607] torch.distributed.elastic.multiprocessing.api: [ERROR] failed (exitcode: 2) local_rank: 0 (pid: 2332236) of binary: /fsx/nouamane/miniconda/envs/2-1-cu121/bin/python
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 33, in <module>
+    sys.exit(load_entry_point('torch==2.1.1', 'console_scripts', 'torchrun')())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 264, in launch_agent
+    raise ChildFailedError(
+torch.distributed.elastic.multiprocessing.errors.ChildFailedError: 
+============================================================
+run_train.py FAILED
+------------------------------------------------------------
+Failures:
+[1]:
+  time      : 2024-12-13_03:48:46
+  host      : ip-26-0-160-103.ec2.internal
+  rank      : 1 (local_rank: 1)
+  exitcode  : 2 (pid: 2332237)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[2]:
+  time      : 2024-12-13_03:48:46
+  host      : ip-26-0-160-103.ec2.internal
+  rank      : 2 (local_rank: 2)
+  exitcode  : 2 (pid: 2332238)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[3]:
+  time      : 2024-12-13_03:48:46
+  host      : ip-26-0-160-103.ec2.internal
+  rank      : 3 (local_rank: 3)
+  exitcode  : 2 (pid: 2332239)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[4]:
+  time      : 2024-12-13_03:48:46
+  host      : ip-26-0-160-103.ec2.internal
+  rank      : 4 (local_rank: 4)
+  exitcode  : 2 (pid: 2332240)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[5]:
+  time      : 2024-12-13_03:48:46
+  host      : ip-26-0-160-103.ec2.internal
+  rank      : 5 (local_rank: 5)
+  exitcode  : 2 (pid: 2332241)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[6]:
+  time      : 2024-12-13_03:48:46
+  host      : ip-26-0-160-103.ec2.internal
+  rank      : 6 (local_rank: 6)
+  exitcode  : 2 (pid: 2332242)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[7]:
+  time      : 2024-12-13_03:48:46
+  host      : ip-26-0-160-103.ec2.internal
+  rank      : 7 (local_rank: 7)
+  exitcode  : 2 (pid: 2332243)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+------------------------------------------------------------
+Root Cause (first observed failure):
+[0]:
+  time      : 2024-12-13_03:48:46
+  host      : ip-26-0-160-103.ec2.internal
+  rank      : 0 (local_rank: 0)
+  exitcode  : 2 (pid: 2332236)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+============================================================
+srun: error: ip-26-0-160-103: task 0: Exited with exit code 1
diff --git a/logs/13404513-bench_stress_test.out b/logs/13404513-bench_stress_test.out
new file mode 100644
index 0000000000000000000000000000000000000000..91818e1c5af726b8897779a2c79e9153021e5453
--- /dev/null
+++ b/logs/13404513-bench_stress_test.out
@@ -0,0 +1,518 @@
++ source /etc/profile.d/modules.sh
+++ . /usr/share/modules/init/bash
++++ unset _mlshdbg
++++ '[' 0 = 1 ']'
++++ unset _mlre _mlIFS
++++ '[' -n x ']'
++++ _mlIFS=' 	
+'
++++ IFS=' '
++++ '[' -n '' ']'
+++++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash autoinit
++++ _mlcode='module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+MODULEPATH_modshare=/etc/environment-modules/modules:1:/usr/share/modules/\$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1; export MODULEPATH_modshare;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+LOADEDMODULES=; export LOADEDMODULES;
+MODULEPATH=/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/\$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles; export MODULEPATH;
+test 0;'
++++ _mlret=0
++++ '[' -n x ']'
++++ IFS=' 	
+'
++++ unset _mlIFS
++++ unset _mlre _mlv _mlrv
++++ '[' 0 -eq 0 ']'
++++ eval 'module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+MODULEPATH_modshare=/etc/environment-modules/modules:1:/usr/share/modules/\$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1; export MODULEPATH_modshare;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+LOADEDMODULES=; export LOADEDMODULES;
+MODULEPATH=/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/\$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles; export MODULEPATH;
+test 0;'
+++++ MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl
+++++ export MODULES_CMD
+++++ ENV=/usr/share/modules/init/profile.sh
+++++ export ENV
+++++ MODULEPATH_modshare='/etc/environment-modules/modules:1:/usr/share/modules/$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1'
+++++ export MODULEPATH_modshare
+++++ BASH_ENV=/usr/share/modules/init/bash
+++++ export BASH_ENV
+++++ MODULESHOME=/usr/share/modules
+++++ export MODULESHOME
+++++ LOADEDMODULES=
+++++ export LOADEDMODULES
+++++ MODULEPATH='/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles'
+++++ export MODULEPATH
+++++ test 0
++++ '[' 0 = 1 ']'
++++ '[' -t 2 ']'
++++ export -f module
++++ export -f switchml
++++ '[' 5 -ge 3 ']'
++++ [[ ehxB =~ i ]]
++++ [[ ! :/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin: =~ :/usr/bin: ]]
++++ '[' '!' -n '' ']'
++++ MANPATH=:
++++ export MANPATH
+++++ manpath
++++ [[ ! :/admin/home/nouamane/.local/share/man:/fsx/nouamane/miniconda/envs/2-1-cu121/man:/fsx/nouamane/miniconda/envs/2-1-cu121/share/man:/fsx/nouamane/miniconda/man:/fsx/nouamane/miniconda/share/man:/opt/amazon/openmpi/share/man:/opt/amazon/efa/share/man:/opt/slurm/share/man:/usr/local/man:/usr/local/share/man:/usr/share/man:/admin/home/nouamane/.fzf/man:: =~ :/usr/share/man: ]]
++++ unset _mlcode _mlret
++++ '[' -n '' ']'
++ module load cuda/12.1
++ unset _mlshdbg
++ '[' 0 = 1 ']'
++ unset _mlre _mlIFS
++ '[' -n x ']'
++ _mlIFS=' 	
+'
++ IFS=' '
++ '[' -n '' ']'
+++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash load cuda/12.1
++ eval 'CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include;' export 'CPATH;
+LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:;' export 'LD_LIBRARY_PATH;
+FI_EFA_FORK_SAFE=1;' export 'FI_EFA_FORK_SAFE;
+MANPATH=/usr/local/cuda-12.1/share/man::;' export 'MANPATH;
+LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64;' export 'LIBRARY_PATH;
+_LMFILES_=/usr/share/modules/modulefiles/cuda/12.1;' export '_LMFILES_;
+LOADEDMODULES=cuda/12.1;' export 'LOADEDMODULES;
+MPI_PATH=/opt/amazon/openmpi;' export 'MPI_PATH;
+NCCL_HOME_modshare=/opt/nccl/build:1;' export 'NCCL_HOME_modshare;
+NCCL_PROTO=simple;' export 'NCCL_PROTO;
+MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1;' export 'MANPATH_modshare;
+LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1;' export 'LIBRARY_PATH_modshare;
+NCCL_SOCKET_IFNAME=enp;' export 'NCCL_SOCKET_IFNAME;
+AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl;' export 'AWS_OFI_NCCL_HOME;
+NCCL_HOME=/opt/nccl/build;' export 'NCCL_HOME;
+FI_PROVIDER=efa;' export 'FI_PROVIDER;
+AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1;' export 'AWS_OFI_NCCL_HOME_modshare;
+CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1;' export 'CPATH_modshare;
+LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2::1:/opt/nccl/build/lib:1:/opt/aws-ofi-nccl/lib:2:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/efa/lib:1;' export 'LD_LIBRARY_PATH_modshare;
+FI_EFA_ENABLE_SHM_TRANSFER=1;' export 'FI_EFA_ENABLE_SHM_TRANSFER;
+_LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1;' export '_LMFILES__modshare;
+LOADEDMODULES_modshare=cuda/12.1:1;' export 'LOADEDMODULES_modshare;
+MPI_PATH_modshare=/opt/amazon/openmpi:1;' export 'MPI_PATH_modshare;
+PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin;' export 'PATH;
+CUDA_HOME=/usr/local/cuda-12.1;' export 'CUDA_HOME;
+PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/usr/local/cuda-12.1/include:1:/opt/amazon/efa/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/usr/sbin:1:/fsx/nouamane/miniconda/bin:1:/opt/slurm/bin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:2:/opt/amazon/openmpi/bin:1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1;' export 'PATH_modshare;
+test' '0;'
+++ CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include
+++ export CPATH
+++ LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:
+++ export LD_LIBRARY_PATH
+++ FI_EFA_FORK_SAFE=1
+++ export FI_EFA_FORK_SAFE
+++ MANPATH=/usr/local/cuda-12.1/share/man::
+++ export MANPATH
+++ LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64
+++ export LIBRARY_PATH
+++ _LMFILES_=/usr/share/modules/modulefiles/cuda/12.1
+++ export _LMFILES_
+++ LOADEDMODULES=cuda/12.1
+++ export LOADEDMODULES
+++ MPI_PATH=/opt/amazon/openmpi
+++ export MPI_PATH
+++ NCCL_HOME_modshare=/opt/nccl/build:1
+++ export NCCL_HOME_modshare
+++ NCCL_PROTO=simple
+++ export NCCL_PROTO
+++ MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1
+++ export MANPATH_modshare
+++ LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1
+++ export LIBRARY_PATH_modshare
+++ NCCL_SOCKET_IFNAME=enp
+++ export NCCL_SOCKET_IFNAME
+++ AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl
+++ export AWS_OFI_NCCL_HOME
+++ NCCL_HOME=/opt/nccl/build
+++ export NCCL_HOME
+++ FI_PROVIDER=efa
+++ export FI_PROVIDER
+++ AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1
+++ export AWS_OFI_NCCL_HOME_modshare
+++ CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1
+++ export CPATH_modshare
+++ LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2::1:/opt/nccl/build/lib:1:/opt/aws-ofi-nccl/lib:2:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/efa/lib:1
+++ export LD_LIBRARY_PATH_modshare
+++ FI_EFA_ENABLE_SHM_TRANSFER=1
+++ export FI_EFA_ENABLE_SHM_TRANSFER
+++ _LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1
+++ export _LMFILES__modshare
+++ LOADEDMODULES_modshare=cuda/12.1:1
+++ export LOADEDMODULES_modshare
+++ MPI_PATH_modshare=/opt/amazon/openmpi:1
+++ export MPI_PATH_modshare
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ export PATH
+++ CUDA_HOME=/usr/local/cuda-12.1
+++ export CUDA_HOME
+++ PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/usr/local/cuda-12.1/include:1:/opt/amazon/efa/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/usr/sbin:1:/fsx/nouamane/miniconda/bin:1:/opt/slurm/bin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:2:/opt/amazon/openmpi/bin:1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1
+++ export PATH_modshare
+++ test 0
++ _mlstatus=0
++ '[' -n x ']'
++ IFS=' 	
+'
++ unset _mlre _mlv _mlrv _mlIFS
++ '[' -n '' ']'
++ unset _mlshdbg
++ return 0
++ source /fsx/nouamane/miniconda/bin/activate
+++ _CONDA_ROOT=/fsx/nouamane/miniconda
+++ . /fsx/nouamane/miniconda/etc/profile.d/conda.sh
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ '[' -z x ']'
+++ conda activate
+++ local cmd=activate
+++ case "$cmd" in
+++ __conda_activate activate
+++ '[' -n '' ']'
+++ local ask_conda
++++ PS1=
++++ __conda_exe shell.posix activate
++++ /fsx/nouamane/miniconda/bin/conda shell.posix activate
+++ ask_conda='. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_2='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\'''
+++ eval '. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_2='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\'''
++++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh
+++++ test -n ''
+++++ unset XML_CATALOG_FILES
+++++ unset xml_catalog_files_libxml2
++++ PS1='(base) '
++++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++++ export CONDA_PREFIX=/fsx/nouamane/miniconda
++++ CONDA_PREFIX=/fsx/nouamane/miniconda
++++ export CONDA_SHLVL=3
++++ CONDA_SHLVL=3
++++ export CONDA_DEFAULT_ENV=base
++++ CONDA_DEFAULT_ENV=base
++++ export 'CONDA_PROMPT_MODIFIER=(base) '
++++ CONDA_PROMPT_MODIFIER='(base) '
++++ export CONDA_PREFIX_2=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ CONDA_PREFIX_2=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ __conda_hashr
+++ '[' -n '' ']'
+++ '[' -n '' ']'
+++ hash -r
++ conda activate 2-1-cu121
++ local cmd=activate
++ case "$cmd" in
++ __conda_activate activate 2-1-cu121
++ '[' -n '' ']'
++ local ask_conda
+++ PS1='(base) '
+++ __conda_exe shell.posix activate 2-1-cu121
+++ /fsx/nouamane/miniconda/bin/conda shell.posix activate 2-1-cu121
++ ask_conda='PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''4'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_3='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
++ eval 'PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''4'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_3='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
+++ PS1='(2-1-cu121) '
+++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ export CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ export CONDA_SHLVL=4
+++ CONDA_SHLVL=4
+++ export CONDA_DEFAULT_ENV=2-1-cu121
+++ CONDA_DEFAULT_ENV=2-1-cu121
+++ export 'CONDA_PROMPT_MODIFIER=(2-1-cu121) '
+++ CONDA_PROMPT_MODIFIER='(2-1-cu121) '
+++ export CONDA_PREFIX_3=/fsx/nouamane/miniconda
+++ CONDA_PREFIX_3=/fsx/nouamane/miniconda
+++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ export _CE_M=
+++ _CE_M=
+++ export _CE_CONDA=
+++ _CE_CONDA=
+++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh
++++ test -n ''
++++ xml_catalog_files_libxml2=
++++ XML_CATALOG_FILES=
++++ conda_catalog_files=
++++ ifs_libxml2=' 	
+'
++++ IFS=' '
++++ rem=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ for pre in ${rem}
++++ test '' = /fsx/nouamane/miniconda/envs/2-1-cu121
++++ conda_catalog_files=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ rem=
++++ IFS=' 	
+'
++++ conda_catalog_files='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ export 'XML_CATALOG_FILES=file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ XML_CATALOG_FILES='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ unset conda_catalog_files ifs_libxml2 rem
++ __conda_hashr
++ '[' -n '' ']'
++ '[' -n '' ']'
++ hash -r
++ export PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++ PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ scontrol show hostnames ip-26-0-169-207
++ export NODELIST=ip-26-0-169-207
++ NODELIST=ip-26-0-169-207
+++ scontrol show hostnames ip-26-0-169-207
+++ head -n1
++ export MASTER_NODE=ip-26-0-169-207
++ MASTER_NODE=ip-26-0-169-207
++ export MASTER_PORT=12356
++ MASTER_PORT=12356
++ export NNODES=1
++ NNODES=1
++ export GPUS_PER_NODE=8
++ GPUS_PER_NODE=8
++ export WORLD_SIZE=8
++ WORLD_SIZE=8
++ export CUDA_DEVICE_MAX_CONNECTIONS=1
++ CUDA_DEVICE_MAX_CONNECTIONS=1
++ export NCCL_DEBUG=INFO
++ NCCL_DEBUG=INFO
++ export NANOTRON_BENCHMARK=1
++ NANOTRON_BENCHMARK=1
++ echo 'Master node: ip-26-0-169-207'
+Master node: ip-26-0-169-207
++ echo 'All nodes: ip-26-0-169-207'
+All nodes: ip-26-0-169-207
++ echo 'World size: 8'
+World size: 8
++ srun torchrun --nnodes=1 --nproc_per_node=8 --rdzv_id=13404513 --rdzv_backend=c10d --rdzv_endpoint=ip-26-0-169-207:12356 run_train.py --config-file benchmark/configs/config_1.14G_dp4_tp2_pp1_acc256_mbs2_seq2048_zero1_tpmodeRED_vocab32k.yaml
+[2024-12-13 03:58:02,517] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-13 03:58:02,517] torch.distributed.run: [WARNING] 
+[2024-12-13 03:58:02,517] torch.distributed.run: [WARNING] *****************************************
+[2024-12-13 03:58:02,517] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-13 03:58:02,517] torch.distributed.run: [WARNING] *****************************************
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+[2024-12-13 03:58:07,668] torch.distributed.elastic.multiprocessing.api: [ERROR] failed (exitcode: 2) local_rank: 0 (pid: 451921) of binary: /fsx/nouamane/miniconda/envs/2-1-cu121/bin/python
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 33, in <module>
+    sys.exit(load_entry_point('torch==2.1.1', 'console_scripts', 'torchrun')())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 264, in launch_agent
+    raise ChildFailedError(
+torch.distributed.elastic.multiprocessing.errors.ChildFailedError: 
+============================================================
+run_train.py FAILED
+------------------------------------------------------------
+Failures:
+[1]:
+  time      : 2024-12-13_03:58:07
+  host      : ip-26-0-169-207.ec2.internal
+  rank      : 1 (local_rank: 1)
+  exitcode  : 2 (pid: 451922)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[2]:
+  time      : 2024-12-13_03:58:07
+  host      : ip-26-0-169-207.ec2.internal
+  rank      : 2 (local_rank: 2)
+  exitcode  : 2 (pid: 451923)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[3]:
+  time      : 2024-12-13_03:58:07
+  host      : ip-26-0-169-207.ec2.internal
+  rank      : 3 (local_rank: 3)
+  exitcode  : 2 (pid: 451924)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[4]:
+  time      : 2024-12-13_03:58:07
+  host      : ip-26-0-169-207.ec2.internal
+  rank      : 4 (local_rank: 4)
+  exitcode  : 2 (pid: 451925)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[5]:
+  time      : 2024-12-13_03:58:07
+  host      : ip-26-0-169-207.ec2.internal
+  rank      : 5 (local_rank: 5)
+  exitcode  : 2 (pid: 451926)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[6]:
+  time      : 2024-12-13_03:58:07
+  host      : ip-26-0-169-207.ec2.internal
+  rank      : 6 (local_rank: 6)
+  exitcode  : 2 (pid: 451927)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[7]:
+  time      : 2024-12-13_03:58:07
+  host      : ip-26-0-169-207.ec2.internal
+  rank      : 7 (local_rank: 7)
+  exitcode  : 2 (pid: 451928)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+------------------------------------------------------------
+Root Cause (first observed failure):
+[0]:
+  time      : 2024-12-13_03:58:07
+  host      : ip-26-0-169-207.ec2.internal
+  rank      : 0 (local_rank: 0)
+  exitcode  : 2 (pid: 451921)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+============================================================
+srun: error: ip-26-0-169-207: task 0: Exited with exit code 1
diff --git a/logs/13404660-bench_stress_test.out b/logs/13404660-bench_stress_test.out
new file mode 100644
index 0000000000000000000000000000000000000000..109d3f5dec912d8712aa0b813903daaa5c0c7231
--- /dev/null
+++ b/logs/13404660-bench_stress_test.out
@@ -0,0 +1,518 @@
++ source /etc/profile.d/modules.sh
+++ . /usr/share/modules/init/bash
++++ unset _mlshdbg
++++ '[' 0 = 1 ']'
++++ unset _mlre _mlIFS
++++ '[' -n x ']'
++++ _mlIFS=' 	
+'
++++ IFS=' '
++++ '[' -n '' ']'
+++++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash autoinit
++++ _mlcode='module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+MODULEPATH_modshare=/etc/environment-modules/modules:1:/usr/share/modules/\$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1; export MODULEPATH_modshare;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+LOADEDMODULES=; export LOADEDMODULES;
+MODULEPATH=/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/\$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles; export MODULEPATH;
+test 0;'
++++ _mlret=0
++++ '[' -n x ']'
++++ IFS=' 	
+'
++++ unset _mlIFS
++++ unset _mlre _mlv _mlrv
++++ '[' 0 -eq 0 ']'
++++ eval 'module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+MODULEPATH_modshare=/etc/environment-modules/modules:1:/usr/share/modules/\$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1; export MODULEPATH_modshare;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+LOADEDMODULES=; export LOADEDMODULES;
+MODULEPATH=/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/\$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles; export MODULEPATH;
+test 0;'
+++++ MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl
+++++ export MODULES_CMD
+++++ ENV=/usr/share/modules/init/profile.sh
+++++ export ENV
+++++ MODULEPATH_modshare='/etc/environment-modules/modules:1:/usr/share/modules/$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1'
+++++ export MODULEPATH_modshare
+++++ BASH_ENV=/usr/share/modules/init/bash
+++++ export BASH_ENV
+++++ MODULESHOME=/usr/share/modules
+++++ export MODULESHOME
+++++ LOADEDMODULES=
+++++ export LOADEDMODULES
+++++ MODULEPATH='/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles'
+++++ export MODULEPATH
+++++ test 0
++++ '[' 0 = 1 ']'
++++ '[' -t 2 ']'
++++ export -f module
++++ export -f switchml
++++ '[' 5 -ge 3 ']'
++++ [[ ehxB =~ i ]]
++++ [[ ! :/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin: =~ :/usr/bin: ]]
++++ '[' '!' -n '' ']'
++++ MANPATH=:
++++ export MANPATH
+++++ manpath
++++ [[ ! :/admin/home/nouamane/.local/share/man:/fsx/nouamane/miniconda/envs/2-1-cu121/man:/fsx/nouamane/miniconda/envs/2-1-cu121/share/man:/fsx/nouamane/miniconda/man:/fsx/nouamane/miniconda/share/man:/opt/amazon/openmpi/share/man:/opt/amazon/efa/share/man:/opt/slurm/share/man:/usr/local/man:/usr/local/share/man:/usr/share/man:/admin/home/nouamane/.fzf/man:: =~ :/usr/share/man: ]]
++++ unset _mlcode _mlret
++++ '[' -n '' ']'
++ module load cuda/12.1
++ unset _mlshdbg
++ '[' 0 = 1 ']'
++ unset _mlre _mlIFS
++ '[' -n x ']'
++ _mlIFS=' 	
+'
++ IFS=' '
++ '[' -n '' ']'
+++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash load cuda/12.1
++ eval 'CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include;' export 'CPATH;
+LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:;' export 'LD_LIBRARY_PATH;
+FI_EFA_FORK_SAFE=1;' export 'FI_EFA_FORK_SAFE;
+MANPATH=/usr/local/cuda-12.1/share/man::;' export 'MANPATH;
+LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64;' export 'LIBRARY_PATH;
+_LMFILES_=/usr/share/modules/modulefiles/cuda/12.1;' export '_LMFILES_;
+LOADEDMODULES=cuda/12.1;' export 'LOADEDMODULES;
+MPI_PATH=/opt/amazon/openmpi;' export 'MPI_PATH;
+NCCL_HOME_modshare=/opt/nccl/build:1;' export 'NCCL_HOME_modshare;
+NCCL_PROTO=simple;' export 'NCCL_PROTO;
+MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1;' export 'MANPATH_modshare;
+LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1;' export 'LIBRARY_PATH_modshare;
+NCCL_SOCKET_IFNAME=enp;' export 'NCCL_SOCKET_IFNAME;
+AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl;' export 'AWS_OFI_NCCL_HOME;
+NCCL_HOME=/opt/nccl/build;' export 'NCCL_HOME;
+FI_PROVIDER=efa;' export 'FI_PROVIDER;
+AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1;' export 'AWS_OFI_NCCL_HOME_modshare;
+CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1;' export 'CPATH_modshare;
+LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2::1:/opt/nccl/build/lib:1:/opt/aws-ofi-nccl/lib:2:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/efa/lib:1;' export 'LD_LIBRARY_PATH_modshare;
+FI_EFA_ENABLE_SHM_TRANSFER=1;' export 'FI_EFA_ENABLE_SHM_TRANSFER;
+_LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1;' export '_LMFILES__modshare;
+LOADEDMODULES_modshare=cuda/12.1:1;' export 'LOADEDMODULES_modshare;
+MPI_PATH_modshare=/opt/amazon/openmpi:1;' export 'MPI_PATH_modshare;
+PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin;' export 'PATH;
+CUDA_HOME=/usr/local/cuda-12.1;' export 'CUDA_HOME;
+PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/usr/local/cuda-12.1/include:1:/opt/amazon/efa/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/usr/sbin:1:/fsx/nouamane/miniconda/bin:1:/opt/slurm/bin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:2:/opt/amazon/openmpi/bin:1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1;' export 'PATH_modshare;
+test' '0;'
+++ CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include
+++ export CPATH
+++ LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:
+++ export LD_LIBRARY_PATH
+++ FI_EFA_FORK_SAFE=1
+++ export FI_EFA_FORK_SAFE
+++ MANPATH=/usr/local/cuda-12.1/share/man::
+++ export MANPATH
+++ LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64
+++ export LIBRARY_PATH
+++ _LMFILES_=/usr/share/modules/modulefiles/cuda/12.1
+++ export _LMFILES_
+++ LOADEDMODULES=cuda/12.1
+++ export LOADEDMODULES
+++ MPI_PATH=/opt/amazon/openmpi
+++ export MPI_PATH
+++ NCCL_HOME_modshare=/opt/nccl/build:1
+++ export NCCL_HOME_modshare
+++ NCCL_PROTO=simple
+++ export NCCL_PROTO
+++ MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1
+++ export MANPATH_modshare
+++ LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1
+++ export LIBRARY_PATH_modshare
+++ NCCL_SOCKET_IFNAME=enp
+++ export NCCL_SOCKET_IFNAME
+++ AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl
+++ export AWS_OFI_NCCL_HOME
+++ NCCL_HOME=/opt/nccl/build
+++ export NCCL_HOME
+++ FI_PROVIDER=efa
+++ export FI_PROVIDER
+++ AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1
+++ export AWS_OFI_NCCL_HOME_modshare
+++ CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1
+++ export CPATH_modshare
+++ LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2::1:/opt/nccl/build/lib:1:/opt/aws-ofi-nccl/lib:2:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/efa/lib:1
+++ export LD_LIBRARY_PATH_modshare
+++ FI_EFA_ENABLE_SHM_TRANSFER=1
+++ export FI_EFA_ENABLE_SHM_TRANSFER
+++ _LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1
+++ export _LMFILES__modshare
+++ LOADEDMODULES_modshare=cuda/12.1:1
+++ export LOADEDMODULES_modshare
+++ MPI_PATH_modshare=/opt/amazon/openmpi:1
+++ export MPI_PATH_modshare
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ export PATH
+++ CUDA_HOME=/usr/local/cuda-12.1
+++ export CUDA_HOME
+++ PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/usr/local/cuda-12.1/include:1:/opt/amazon/efa/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/usr/sbin:1:/fsx/nouamane/miniconda/bin:1:/opt/slurm/bin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:2:/opt/amazon/openmpi/bin:1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1
+++ export PATH_modshare
+++ test 0
++ _mlstatus=0
++ '[' -n x ']'
++ IFS=' 	
+'
++ unset _mlre _mlv _mlrv _mlIFS
++ '[' -n '' ']'
++ unset _mlshdbg
++ return 0
++ source /fsx/nouamane/miniconda/bin/activate
+++ _CONDA_ROOT=/fsx/nouamane/miniconda
+++ . /fsx/nouamane/miniconda/etc/profile.d/conda.sh
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ '[' -z x ']'
+++ conda activate
+++ local cmd=activate
+++ case "$cmd" in
+++ __conda_activate activate
+++ '[' -n '' ']'
+++ local ask_conda
++++ PS1=
++++ __conda_exe shell.posix activate
++++ /fsx/nouamane/miniconda/bin/conda shell.posix activate
+++ ask_conda='. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_2='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\'''
+++ eval '. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_2='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\'''
++++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh
+++++ test -n ''
+++++ unset XML_CATALOG_FILES
+++++ unset xml_catalog_files_libxml2
++++ PS1='(base) '
++++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++++ export CONDA_PREFIX=/fsx/nouamane/miniconda
++++ CONDA_PREFIX=/fsx/nouamane/miniconda
++++ export CONDA_SHLVL=3
++++ CONDA_SHLVL=3
++++ export CONDA_DEFAULT_ENV=base
++++ CONDA_DEFAULT_ENV=base
++++ export 'CONDA_PROMPT_MODIFIER=(base) '
++++ CONDA_PROMPT_MODIFIER='(base) '
++++ export CONDA_PREFIX_2=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ CONDA_PREFIX_2=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ __conda_hashr
+++ '[' -n '' ']'
+++ '[' -n '' ']'
+++ hash -r
++ conda activate 2-1-cu121
++ local cmd=activate
++ case "$cmd" in
++ __conda_activate activate 2-1-cu121
++ '[' -n '' ']'
++ local ask_conda
+++ PS1='(base) '
+++ __conda_exe shell.posix activate 2-1-cu121
+++ /fsx/nouamane/miniconda/bin/conda shell.posix activate 2-1-cu121
++ ask_conda='PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''4'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_3='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
++ eval 'PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''4'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_3='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
+++ PS1='(2-1-cu121) '
+++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ export CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ export CONDA_SHLVL=4
+++ CONDA_SHLVL=4
+++ export CONDA_DEFAULT_ENV=2-1-cu121
+++ CONDA_DEFAULT_ENV=2-1-cu121
+++ export 'CONDA_PROMPT_MODIFIER=(2-1-cu121) '
+++ CONDA_PROMPT_MODIFIER='(2-1-cu121) '
+++ export CONDA_PREFIX_3=/fsx/nouamane/miniconda
+++ CONDA_PREFIX_3=/fsx/nouamane/miniconda
+++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ export _CE_M=
+++ _CE_M=
+++ export _CE_CONDA=
+++ _CE_CONDA=
+++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh
++++ test -n ''
++++ xml_catalog_files_libxml2=
++++ XML_CATALOG_FILES=
++++ conda_catalog_files=
++++ ifs_libxml2=' 	
+'
++++ IFS=' '
++++ rem=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ for pre in ${rem}
++++ test '' = /fsx/nouamane/miniconda/envs/2-1-cu121
++++ conda_catalog_files=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ rem=
++++ IFS=' 	
+'
++++ conda_catalog_files='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ export 'XML_CATALOG_FILES=file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ XML_CATALOG_FILES='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ unset conda_catalog_files ifs_libxml2 rem
++ __conda_hashr
++ '[' -n '' ']'
++ '[' -n '' ']'
++ hash -r
++ export PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++ PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ scontrol show hostnames ip-26-0-169-207
++ export NODELIST=ip-26-0-169-207
++ NODELIST=ip-26-0-169-207
+++ scontrol show hostnames ip-26-0-169-207
+++ head -n1
++ export MASTER_NODE=ip-26-0-169-207
++ MASTER_NODE=ip-26-0-169-207
++ export MASTER_PORT=12356
++ MASTER_PORT=12356
++ export NNODES=1
++ NNODES=1
++ export GPUS_PER_NODE=8
++ GPUS_PER_NODE=8
++ export WORLD_SIZE=8
++ WORLD_SIZE=8
++ export CUDA_DEVICE_MAX_CONNECTIONS=1
++ CUDA_DEVICE_MAX_CONNECTIONS=1
++ export NCCL_DEBUG=INFO
++ NCCL_DEBUG=INFO
++ export NANOTRON_BENCHMARK=1
++ NANOTRON_BENCHMARK=1
++ echo 'Master node: ip-26-0-169-207'
+Master node: ip-26-0-169-207
++ echo 'All nodes: ip-26-0-169-207'
+All nodes: ip-26-0-169-207
++ echo 'World size: 8'
+World size: 8
++ srun torchrun --nnodes=1 --nproc_per_node=8 --rdzv_id=13404660 --rdzv_backend=c10d --rdzv_endpoint=ip-26-0-169-207:12356 run_train.py --config-file benchmark/configs/config_1.14G_dp4_tp2_pp1_acc256_mbs2_seq2048_zero1_tpmodeRED_vocab32k.yaml
+[2024-12-13 04:21:02,830] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-13 04:21:02,830] torch.distributed.run: [WARNING] 
+[2024-12-13 04:21:02,830] torch.distributed.run: [WARNING] *****************************************
+[2024-12-13 04:21:02,830] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-13 04:21:02,830] torch.distributed.run: [WARNING] *****************************************
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+[2024-12-13 04:21:07,942] torch.distributed.elastic.multiprocessing.api: [ERROR] failed (exitcode: 2) local_rank: 0 (pid: 475390) of binary: /fsx/nouamane/miniconda/envs/2-1-cu121/bin/python
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 33, in <module>
+    sys.exit(load_entry_point('torch==2.1.1', 'console_scripts', 'torchrun')())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 264, in launch_agent
+    raise ChildFailedError(
+torch.distributed.elastic.multiprocessing.errors.ChildFailedError: 
+============================================================
+run_train.py FAILED
+------------------------------------------------------------
+Failures:
+[1]:
+  time      : 2024-12-13_04:21:07
+  host      : ip-26-0-169-207.ec2.internal
+  rank      : 1 (local_rank: 1)
+  exitcode  : 2 (pid: 475391)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[2]:
+  time      : 2024-12-13_04:21:07
+  host      : ip-26-0-169-207.ec2.internal
+  rank      : 2 (local_rank: 2)
+  exitcode  : 2 (pid: 475392)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[3]:
+  time      : 2024-12-13_04:21:07
+  host      : ip-26-0-169-207.ec2.internal
+  rank      : 3 (local_rank: 3)
+  exitcode  : 2 (pid: 475393)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[4]:
+  time      : 2024-12-13_04:21:07
+  host      : ip-26-0-169-207.ec2.internal
+  rank      : 4 (local_rank: 4)
+  exitcode  : 2 (pid: 475394)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[5]:
+  time      : 2024-12-13_04:21:07
+  host      : ip-26-0-169-207.ec2.internal
+  rank      : 5 (local_rank: 5)
+  exitcode  : 2 (pid: 475395)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[6]:
+  time      : 2024-12-13_04:21:07
+  host      : ip-26-0-169-207.ec2.internal
+  rank      : 6 (local_rank: 6)
+  exitcode  : 2 (pid: 475396)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[7]:
+  time      : 2024-12-13_04:21:07
+  host      : ip-26-0-169-207.ec2.internal
+  rank      : 7 (local_rank: 7)
+  exitcode  : 2 (pid: 475397)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+------------------------------------------------------------
+Root Cause (first observed failure):
+[0]:
+  time      : 2024-12-13_04:21:07
+  host      : ip-26-0-169-207.ec2.internal
+  rank      : 0 (local_rank: 0)
+  exitcode  : 2 (pid: 475390)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+============================================================
+srun: error: ip-26-0-169-207: task 0: Exited with exit code 1
diff --git a/logs/13404721-bench_stress_test.out b/logs/13404721-bench_stress_test.out
new file mode 100644
index 0000000000000000000000000000000000000000..5a5609485134af6a78f23b4e14b161381d673bdf
--- /dev/null
+++ b/logs/13404721-bench_stress_test.out
@@ -0,0 +1,518 @@
++ source /etc/profile.d/modules.sh
+++ . /usr/share/modules/init/bash
++++ unset _mlshdbg
++++ '[' 0 = 1 ']'
++++ unset _mlre _mlIFS
++++ '[' -n x ']'
++++ _mlIFS=' 	
+'
++++ IFS=' '
++++ '[' -n '' ']'
+++++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash autoinit
++++ _mlcode='module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+MODULEPATH_modshare=/etc/environment-modules/modules:1:/usr/share/modules/\$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1; export MODULEPATH_modshare;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+LOADEDMODULES=; export LOADEDMODULES;
+MODULEPATH=/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/\$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles; export MODULEPATH;
+test 0;'
++++ _mlret=0
++++ '[' -n x ']'
++++ IFS=' 	
+'
++++ unset _mlIFS
++++ unset _mlre _mlv _mlrv
++++ '[' 0 -eq 0 ']'
++++ eval 'module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+MODULEPATH_modshare=/etc/environment-modules/modules:1:/usr/share/modules/\$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1; export MODULEPATH_modshare;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+LOADEDMODULES=; export LOADEDMODULES;
+MODULEPATH=/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/\$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles; export MODULEPATH;
+test 0;'
+++++ MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl
+++++ export MODULES_CMD
+++++ ENV=/usr/share/modules/init/profile.sh
+++++ export ENV
+++++ MODULEPATH_modshare='/etc/environment-modules/modules:1:/usr/share/modules/$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1'
+++++ export MODULEPATH_modshare
+++++ BASH_ENV=/usr/share/modules/init/bash
+++++ export BASH_ENV
+++++ MODULESHOME=/usr/share/modules
+++++ export MODULESHOME
+++++ LOADEDMODULES=
+++++ export LOADEDMODULES
+++++ MODULEPATH='/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles'
+++++ export MODULEPATH
+++++ test 0
++++ '[' 0 = 1 ']'
++++ '[' -t 2 ']'
++++ export -f module
++++ export -f switchml
++++ '[' 5 -ge 3 ']'
++++ [[ ehxB =~ i ]]
++++ [[ ! :/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin: =~ :/usr/bin: ]]
++++ '[' '!' -n '' ']'
++++ MANPATH=:
++++ export MANPATH
+++++ manpath
++++ [[ ! :/admin/home/nouamane/.local/share/man:/fsx/nouamane/miniconda/envs/2-1-cu121/man:/fsx/nouamane/miniconda/envs/2-1-cu121/share/man:/fsx/nouamane/miniconda/man:/fsx/nouamane/miniconda/share/man:/opt/amazon/openmpi/share/man:/opt/amazon/efa/share/man:/opt/slurm/share/man:/usr/local/man:/usr/local/share/man:/usr/share/man:/admin/home/nouamane/.fzf/man:: =~ :/usr/share/man: ]]
++++ unset _mlcode _mlret
++++ '[' -n '' ']'
++ module load cuda/12.1
++ unset _mlshdbg
++ '[' 0 = 1 ']'
++ unset _mlre _mlIFS
++ '[' -n x ']'
++ _mlIFS=' 	
+'
++ IFS=' '
++ '[' -n '' ']'
+++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash load cuda/12.1
++ eval 'CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include;' export 'CPATH;
+LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:;' export 'LD_LIBRARY_PATH;
+FI_EFA_FORK_SAFE=1;' export 'FI_EFA_FORK_SAFE;
+MANPATH=/usr/local/cuda-12.1/share/man::;' export 'MANPATH;
+LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64;' export 'LIBRARY_PATH;
+_LMFILES_=/usr/share/modules/modulefiles/cuda/12.1;' export '_LMFILES_;
+LOADEDMODULES=cuda/12.1;' export 'LOADEDMODULES;
+MPI_PATH=/opt/amazon/openmpi;' export 'MPI_PATH;
+NCCL_HOME_modshare=/opt/nccl/build:1;' export 'NCCL_HOME_modshare;
+NCCL_PROTO=simple;' export 'NCCL_PROTO;
+MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1;' export 'MANPATH_modshare;
+LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1;' export 'LIBRARY_PATH_modshare;
+NCCL_SOCKET_IFNAME=enp;' export 'NCCL_SOCKET_IFNAME;
+AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl;' export 'AWS_OFI_NCCL_HOME;
+NCCL_HOME=/opt/nccl/build;' export 'NCCL_HOME;
+FI_PROVIDER=efa;' export 'FI_PROVIDER;
+AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1;' export 'AWS_OFI_NCCL_HOME_modshare;
+CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1;' export 'CPATH_modshare;
+LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2::1:/opt/nccl/build/lib:1:/opt/aws-ofi-nccl/lib:2:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/efa/lib:1;' export 'LD_LIBRARY_PATH_modshare;
+FI_EFA_ENABLE_SHM_TRANSFER=1;' export 'FI_EFA_ENABLE_SHM_TRANSFER;
+_LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1;' export '_LMFILES__modshare;
+LOADEDMODULES_modshare=cuda/12.1:1;' export 'LOADEDMODULES_modshare;
+MPI_PATH_modshare=/opt/amazon/openmpi:1;' export 'MPI_PATH_modshare;
+PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin;' export 'PATH;
+CUDA_HOME=/usr/local/cuda-12.1;' export 'CUDA_HOME;
+PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/usr/local/cuda-12.1/include:1:/opt/amazon/efa/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/usr/sbin:1:/fsx/nouamane/miniconda/bin:1:/opt/slurm/bin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:2:/opt/amazon/openmpi/bin:1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1;' export 'PATH_modshare;
+test' '0;'
+++ CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include
+++ export CPATH
+++ LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:
+++ export LD_LIBRARY_PATH
+++ FI_EFA_FORK_SAFE=1
+++ export FI_EFA_FORK_SAFE
+++ MANPATH=/usr/local/cuda-12.1/share/man::
+++ export MANPATH
+++ LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64
+++ export LIBRARY_PATH
+++ _LMFILES_=/usr/share/modules/modulefiles/cuda/12.1
+++ export _LMFILES_
+++ LOADEDMODULES=cuda/12.1
+++ export LOADEDMODULES
+++ MPI_PATH=/opt/amazon/openmpi
+++ export MPI_PATH
+++ NCCL_HOME_modshare=/opt/nccl/build:1
+++ export NCCL_HOME_modshare
+++ NCCL_PROTO=simple
+++ export NCCL_PROTO
+++ MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1
+++ export MANPATH_modshare
+++ LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1
+++ export LIBRARY_PATH_modshare
+++ NCCL_SOCKET_IFNAME=enp
+++ export NCCL_SOCKET_IFNAME
+++ AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl
+++ export AWS_OFI_NCCL_HOME
+++ NCCL_HOME=/opt/nccl/build
+++ export NCCL_HOME
+++ FI_PROVIDER=efa
+++ export FI_PROVIDER
+++ AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1
+++ export AWS_OFI_NCCL_HOME_modshare
+++ CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1
+++ export CPATH_modshare
+++ LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2::1:/opt/nccl/build/lib:1:/opt/aws-ofi-nccl/lib:2:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/efa/lib:1
+++ export LD_LIBRARY_PATH_modshare
+++ FI_EFA_ENABLE_SHM_TRANSFER=1
+++ export FI_EFA_ENABLE_SHM_TRANSFER
+++ _LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1
+++ export _LMFILES__modshare
+++ LOADEDMODULES_modshare=cuda/12.1:1
+++ export LOADEDMODULES_modshare
+++ MPI_PATH_modshare=/opt/amazon/openmpi:1
+++ export MPI_PATH_modshare
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ export PATH
+++ CUDA_HOME=/usr/local/cuda-12.1
+++ export CUDA_HOME
+++ PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/usr/local/cuda-12.1/include:1:/opt/amazon/efa/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/usr/sbin:1:/fsx/nouamane/miniconda/bin:1:/opt/slurm/bin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:2:/opt/amazon/openmpi/bin:1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1
+++ export PATH_modshare
+++ test 0
++ _mlstatus=0
++ '[' -n x ']'
++ IFS=' 	
+'
++ unset _mlre _mlv _mlrv _mlIFS
++ '[' -n '' ']'
++ unset _mlshdbg
++ return 0
++ source /fsx/nouamane/miniconda/bin/activate
+++ _CONDA_ROOT=/fsx/nouamane/miniconda
+++ . /fsx/nouamane/miniconda/etc/profile.d/conda.sh
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ '[' -z x ']'
+++ conda activate
+++ local cmd=activate
+++ case "$cmd" in
+++ __conda_activate activate
+++ '[' -n '' ']'
+++ local ask_conda
++++ PS1=
++++ __conda_exe shell.posix activate
++++ /fsx/nouamane/miniconda/bin/conda shell.posix activate
+++ ask_conda='. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_2='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\'''
+++ eval '. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_2='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\'''
++++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh
+++++ test -n ''
+++++ unset XML_CATALOG_FILES
+++++ unset xml_catalog_files_libxml2
++++ PS1='(base) '
++++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++++ export CONDA_PREFIX=/fsx/nouamane/miniconda
++++ CONDA_PREFIX=/fsx/nouamane/miniconda
++++ export CONDA_SHLVL=3
++++ CONDA_SHLVL=3
++++ export CONDA_DEFAULT_ENV=base
++++ CONDA_DEFAULT_ENV=base
++++ export 'CONDA_PROMPT_MODIFIER=(base) '
++++ CONDA_PROMPT_MODIFIER='(base) '
++++ export CONDA_PREFIX_2=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ CONDA_PREFIX_2=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ __conda_hashr
+++ '[' -n '' ']'
+++ '[' -n '' ']'
+++ hash -r
++ conda activate 2-1-cu121
++ local cmd=activate
++ case "$cmd" in
++ __conda_activate activate 2-1-cu121
++ '[' -n '' ']'
++ local ask_conda
+++ PS1='(base) '
+++ __conda_exe shell.posix activate 2-1-cu121
+++ /fsx/nouamane/miniconda/bin/conda shell.posix activate 2-1-cu121
++ ask_conda='PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''4'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_3='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
++ eval 'PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''4'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_3='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
+++ PS1='(2-1-cu121) '
+++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ export CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ export CONDA_SHLVL=4
+++ CONDA_SHLVL=4
+++ export CONDA_DEFAULT_ENV=2-1-cu121
+++ CONDA_DEFAULT_ENV=2-1-cu121
+++ export 'CONDA_PROMPT_MODIFIER=(2-1-cu121) '
+++ CONDA_PROMPT_MODIFIER='(2-1-cu121) '
+++ export CONDA_PREFIX_3=/fsx/nouamane/miniconda
+++ CONDA_PREFIX_3=/fsx/nouamane/miniconda
+++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ export _CE_M=
+++ _CE_M=
+++ export _CE_CONDA=
+++ _CE_CONDA=
+++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh
++++ test -n ''
++++ xml_catalog_files_libxml2=
++++ XML_CATALOG_FILES=
++++ conda_catalog_files=
++++ ifs_libxml2=' 	
+'
++++ IFS=' '
++++ rem=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ for pre in ${rem}
++++ test '' = /fsx/nouamane/miniconda/envs/2-1-cu121
++++ conda_catalog_files=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ rem=
++++ IFS=' 	
+'
++++ conda_catalog_files='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ export 'XML_CATALOG_FILES=file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ XML_CATALOG_FILES='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ unset conda_catalog_files ifs_libxml2 rem
++ __conda_hashr
++ '[' -n '' ']'
++ '[' -n '' ']'
++ hash -r
++ export PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++ PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ scontrol show hostnames ip-26-0-169-207
++ export NODELIST=ip-26-0-169-207
++ NODELIST=ip-26-0-169-207
+++ scontrol show hostnames ip-26-0-169-207
+++ head -n1
++ export MASTER_NODE=ip-26-0-169-207
++ MASTER_NODE=ip-26-0-169-207
++ export MASTER_PORT=12356
++ MASTER_PORT=12356
++ export NNODES=1
++ NNODES=1
++ export GPUS_PER_NODE=8
++ GPUS_PER_NODE=8
++ export WORLD_SIZE=8
++ WORLD_SIZE=8
++ export CUDA_DEVICE_MAX_CONNECTIONS=1
++ CUDA_DEVICE_MAX_CONNECTIONS=1
++ export NCCL_DEBUG=INFO
++ NCCL_DEBUG=INFO
++ export NANOTRON_BENCHMARK=1
++ NANOTRON_BENCHMARK=1
++ echo 'Master node: ip-26-0-169-207'
+Master node: ip-26-0-169-207
++ echo 'All nodes: ip-26-0-169-207'
+All nodes: ip-26-0-169-207
++ echo 'World size: 8'
+World size: 8
++ srun torchrun --nnodes=1 --nproc_per_node=8 --rdzv_id=13404721 --rdzv_backend=c10d --rdzv_endpoint=ip-26-0-169-207:12356 run_train.py --config-file benchmark/configs/config_1.14G_dp4_tp2_pp1_acc256_mbs2_seq2048_zero1_tpmodeRED_vocab32k.yaml
+[2024-12-13 04:30:52,910] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-13 04:30:52,910] torch.distributed.run: [WARNING] 
+[2024-12-13 04:30:52,910] torch.distributed.run: [WARNING] *****************************************
+[2024-12-13 04:30:52,910] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-13 04:30:52,910] torch.distributed.run: [WARNING] *****************************************
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+[2024-12-13 04:30:58,189] torch.distributed.elastic.multiprocessing.api: [ERROR] failed (exitcode: 2) local_rank: 0 (pid: 484882) of binary: /fsx/nouamane/miniconda/envs/2-1-cu121/bin/python
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 33, in <module>
+    sys.exit(load_entry_point('torch==2.1.1', 'console_scripts', 'torchrun')())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 264, in launch_agent
+    raise ChildFailedError(
+torch.distributed.elastic.multiprocessing.errors.ChildFailedError: 
+============================================================
+run_train.py FAILED
+------------------------------------------------------------
+Failures:
+[1]:
+  time      : 2024-12-13_04:30:58
+  host      : ip-26-0-169-207.ec2.internal
+  rank      : 1 (local_rank: 1)
+  exitcode  : 2 (pid: 484883)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[2]:
+  time      : 2024-12-13_04:30:58
+  host      : ip-26-0-169-207.ec2.internal
+  rank      : 2 (local_rank: 2)
+  exitcode  : 2 (pid: 484884)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[3]:
+  time      : 2024-12-13_04:30:58
+  host      : ip-26-0-169-207.ec2.internal
+  rank      : 3 (local_rank: 3)
+  exitcode  : 2 (pid: 484885)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[4]:
+  time      : 2024-12-13_04:30:58
+  host      : ip-26-0-169-207.ec2.internal
+  rank      : 4 (local_rank: 4)
+  exitcode  : 2 (pid: 484886)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[5]:
+  time      : 2024-12-13_04:30:58
+  host      : ip-26-0-169-207.ec2.internal
+  rank      : 5 (local_rank: 5)
+  exitcode  : 2 (pid: 484887)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[6]:
+  time      : 2024-12-13_04:30:58
+  host      : ip-26-0-169-207.ec2.internal
+  rank      : 6 (local_rank: 6)
+  exitcode  : 2 (pid: 484888)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[7]:
+  time      : 2024-12-13_04:30:58
+  host      : ip-26-0-169-207.ec2.internal
+  rank      : 7 (local_rank: 7)
+  exitcode  : 2 (pid: 484889)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+------------------------------------------------------------
+Root Cause (first observed failure):
+[0]:
+  time      : 2024-12-13_04:30:58
+  host      : ip-26-0-169-207.ec2.internal
+  rank      : 0 (local_rank: 0)
+  exitcode  : 2 (pid: 484882)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+============================================================
+srun: error: ip-26-0-169-207: task 0: Exited with exit code 1
diff --git a/logs/13404811-bench_stress_test.out b/logs/13404811-bench_stress_test.out
new file mode 100644
index 0000000000000000000000000000000000000000..cd1bd338a39142a6e02d672fc98850b663208716
--- /dev/null
+++ b/logs/13404811-bench_stress_test.out
@@ -0,0 +1,518 @@
++ source /etc/profile.d/modules.sh
+++ . /usr/share/modules/init/bash
++++ unset _mlshdbg
++++ '[' 0 = 1 ']'
++++ unset _mlre _mlIFS
++++ '[' -n x ']'
++++ _mlIFS=' 	
+'
++++ IFS=' '
++++ '[' -n '' ']'
+++++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash autoinit
++++ _mlcode='module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+MODULEPATH_modshare=/etc/environment-modules/modules:1:/usr/share/modules/\$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1; export MODULEPATH_modshare;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+LOADEDMODULES=; export LOADEDMODULES;
+MODULEPATH=/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/\$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles; export MODULEPATH;
+test 0;'
++++ _mlret=0
++++ '[' -n x ']'
++++ IFS=' 	
+'
++++ unset _mlIFS
++++ unset _mlre _mlv _mlrv
++++ '[' 0 -eq 0 ']'
++++ eval 'module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+MODULEPATH_modshare=/etc/environment-modules/modules:1:/usr/share/modules/\$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1; export MODULEPATH_modshare;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+LOADEDMODULES=; export LOADEDMODULES;
+MODULEPATH=/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/\$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles; export MODULEPATH;
+test 0;'
+++++ MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl
+++++ export MODULES_CMD
+++++ ENV=/usr/share/modules/init/profile.sh
+++++ export ENV
+++++ MODULEPATH_modshare='/etc/environment-modules/modules:1:/usr/share/modules/$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1'
+++++ export MODULEPATH_modshare
+++++ BASH_ENV=/usr/share/modules/init/bash
+++++ export BASH_ENV
+++++ MODULESHOME=/usr/share/modules
+++++ export MODULESHOME
+++++ LOADEDMODULES=
+++++ export LOADEDMODULES
+++++ MODULEPATH='/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles'
+++++ export MODULEPATH
+++++ test 0
++++ '[' 0 = 1 ']'
++++ '[' -t 2 ']'
++++ export -f module
++++ export -f switchml
++++ '[' 5 -ge 3 ']'
++++ [[ ehxB =~ i ]]
++++ [[ ! :/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin: =~ :/usr/bin: ]]
++++ '[' '!' -n '' ']'
++++ MANPATH=:
++++ export MANPATH
+++++ manpath
++++ [[ ! :/admin/home/nouamane/.local/share/man:/fsx/nouamane/miniconda/envs/2-1-cu121/man:/fsx/nouamane/miniconda/envs/2-1-cu121/share/man:/fsx/nouamane/miniconda/man:/fsx/nouamane/miniconda/share/man:/opt/amazon/openmpi/share/man:/opt/amazon/efa/share/man:/opt/slurm/share/man:/usr/local/man:/usr/local/share/man:/usr/share/man:/admin/home/nouamane/.fzf/man:: =~ :/usr/share/man: ]]
++++ unset _mlcode _mlret
++++ '[' -n '' ']'
++ module load cuda/12.1
++ unset _mlshdbg
++ '[' 0 = 1 ']'
++ unset _mlre _mlIFS
++ '[' -n x ']'
++ _mlIFS=' 	
+'
++ IFS=' '
++ '[' -n '' ']'
+++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash load cuda/12.1
++ eval 'CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include;' export 'CPATH;
+LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:;' export 'LD_LIBRARY_PATH;
+FI_EFA_FORK_SAFE=1;' export 'FI_EFA_FORK_SAFE;
+MANPATH=/usr/local/cuda-12.1/share/man::;' export 'MANPATH;
+LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64;' export 'LIBRARY_PATH;
+_LMFILES_=/usr/share/modules/modulefiles/cuda/12.1;' export '_LMFILES_;
+LOADEDMODULES=cuda/12.1;' export 'LOADEDMODULES;
+MPI_PATH=/opt/amazon/openmpi;' export 'MPI_PATH;
+NCCL_HOME_modshare=/opt/nccl/build:1;' export 'NCCL_HOME_modshare;
+NCCL_PROTO=simple;' export 'NCCL_PROTO;
+MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1;' export 'MANPATH_modshare;
+LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1;' export 'LIBRARY_PATH_modshare;
+NCCL_SOCKET_IFNAME=enp;' export 'NCCL_SOCKET_IFNAME;
+AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl;' export 'AWS_OFI_NCCL_HOME;
+NCCL_HOME=/opt/nccl/build;' export 'NCCL_HOME;
+FI_PROVIDER=efa;' export 'FI_PROVIDER;
+AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1;' export 'AWS_OFI_NCCL_HOME_modshare;
+CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1;' export 'CPATH_modshare;
+LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2::1:/opt/nccl/build/lib:1:/opt/aws-ofi-nccl/lib:2:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/efa/lib:1;' export 'LD_LIBRARY_PATH_modshare;
+FI_EFA_ENABLE_SHM_TRANSFER=1;' export 'FI_EFA_ENABLE_SHM_TRANSFER;
+_LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1;' export '_LMFILES__modshare;
+LOADEDMODULES_modshare=cuda/12.1:1;' export 'LOADEDMODULES_modshare;
+MPI_PATH_modshare=/opt/amazon/openmpi:1;' export 'MPI_PATH_modshare;
+PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin;' export 'PATH;
+CUDA_HOME=/usr/local/cuda-12.1;' export 'CUDA_HOME;
+PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/usr/local/cuda-12.1/include:1:/opt/amazon/efa/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/usr/sbin:1:/fsx/nouamane/miniconda/bin:1:/opt/slurm/bin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:2:/opt/amazon/openmpi/bin:1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1;' export 'PATH_modshare;
+test' '0;'
+++ CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include
+++ export CPATH
+++ LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:
+++ export LD_LIBRARY_PATH
+++ FI_EFA_FORK_SAFE=1
+++ export FI_EFA_FORK_SAFE
+++ MANPATH=/usr/local/cuda-12.1/share/man::
+++ export MANPATH
+++ LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64
+++ export LIBRARY_PATH
+++ _LMFILES_=/usr/share/modules/modulefiles/cuda/12.1
+++ export _LMFILES_
+++ LOADEDMODULES=cuda/12.1
+++ export LOADEDMODULES
+++ MPI_PATH=/opt/amazon/openmpi
+++ export MPI_PATH
+++ NCCL_HOME_modshare=/opt/nccl/build:1
+++ export NCCL_HOME_modshare
+++ NCCL_PROTO=simple
+++ export NCCL_PROTO
+++ MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1
+++ export MANPATH_modshare
+++ LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1
+++ export LIBRARY_PATH_modshare
+++ NCCL_SOCKET_IFNAME=enp
+++ export NCCL_SOCKET_IFNAME
+++ AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl
+++ export AWS_OFI_NCCL_HOME
+++ NCCL_HOME=/opt/nccl/build
+++ export NCCL_HOME
+++ FI_PROVIDER=efa
+++ export FI_PROVIDER
+++ AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1
+++ export AWS_OFI_NCCL_HOME_modshare
+++ CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1
+++ export CPATH_modshare
+++ LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2::1:/opt/nccl/build/lib:1:/opt/aws-ofi-nccl/lib:2:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/efa/lib:1
+++ export LD_LIBRARY_PATH_modshare
+++ FI_EFA_ENABLE_SHM_TRANSFER=1
+++ export FI_EFA_ENABLE_SHM_TRANSFER
+++ _LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1
+++ export _LMFILES__modshare
+++ LOADEDMODULES_modshare=cuda/12.1:1
+++ export LOADEDMODULES_modshare
+++ MPI_PATH_modshare=/opt/amazon/openmpi:1
+++ export MPI_PATH_modshare
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ export PATH
+++ CUDA_HOME=/usr/local/cuda-12.1
+++ export CUDA_HOME
+++ PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/usr/local/cuda-12.1/include:1:/opt/amazon/efa/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/usr/sbin:1:/fsx/nouamane/miniconda/bin:1:/opt/slurm/bin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:2:/opt/amazon/openmpi/bin:1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1
+++ export PATH_modshare
+++ test 0
++ _mlstatus=0
++ '[' -n x ']'
++ IFS=' 	
+'
++ unset _mlre _mlv _mlrv _mlIFS
++ '[' -n '' ']'
++ unset _mlshdbg
++ return 0
++ source /fsx/nouamane/miniconda/bin/activate
+++ _CONDA_ROOT=/fsx/nouamane/miniconda
+++ . /fsx/nouamane/miniconda/etc/profile.d/conda.sh
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ '[' -z x ']'
+++ conda activate
+++ local cmd=activate
+++ case "$cmd" in
+++ __conda_activate activate
+++ '[' -n '' ']'
+++ local ask_conda
++++ PS1=
++++ __conda_exe shell.posix activate
++++ /fsx/nouamane/miniconda/bin/conda shell.posix activate
+++ ask_conda='. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_2='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\'''
+++ eval '. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_2='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\'''
++++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh
+++++ test -n ''
+++++ unset XML_CATALOG_FILES
+++++ unset xml_catalog_files_libxml2
++++ PS1='(base) '
++++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++++ export CONDA_PREFIX=/fsx/nouamane/miniconda
++++ CONDA_PREFIX=/fsx/nouamane/miniconda
++++ export CONDA_SHLVL=3
++++ CONDA_SHLVL=3
++++ export CONDA_DEFAULT_ENV=base
++++ CONDA_DEFAULT_ENV=base
++++ export 'CONDA_PROMPT_MODIFIER=(base) '
++++ CONDA_PROMPT_MODIFIER='(base) '
++++ export CONDA_PREFIX_2=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ CONDA_PREFIX_2=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ __conda_hashr
+++ '[' -n '' ']'
+++ '[' -n '' ']'
+++ hash -r
++ conda activate 2-1-cu121
++ local cmd=activate
++ case "$cmd" in
++ __conda_activate activate 2-1-cu121
++ '[' -n '' ']'
++ local ask_conda
+++ PS1='(base) '
+++ __conda_exe shell.posix activate 2-1-cu121
+++ /fsx/nouamane/miniconda/bin/conda shell.posix activate 2-1-cu121
++ ask_conda='PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''4'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_3='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
++ eval 'PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''4'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_3='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
+++ PS1='(2-1-cu121) '
+++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ export CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ export CONDA_SHLVL=4
+++ CONDA_SHLVL=4
+++ export CONDA_DEFAULT_ENV=2-1-cu121
+++ CONDA_DEFAULT_ENV=2-1-cu121
+++ export 'CONDA_PROMPT_MODIFIER=(2-1-cu121) '
+++ CONDA_PROMPT_MODIFIER='(2-1-cu121) '
+++ export CONDA_PREFIX_3=/fsx/nouamane/miniconda
+++ CONDA_PREFIX_3=/fsx/nouamane/miniconda
+++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ export _CE_M=
+++ _CE_M=
+++ export _CE_CONDA=
+++ _CE_CONDA=
+++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh
++++ test -n ''
++++ xml_catalog_files_libxml2=
++++ XML_CATALOG_FILES=
++++ conda_catalog_files=
++++ ifs_libxml2=' 	
+'
++++ IFS=' '
++++ rem=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ for pre in ${rem}
++++ test '' = /fsx/nouamane/miniconda/envs/2-1-cu121
++++ conda_catalog_files=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ rem=
++++ IFS=' 	
+'
++++ conda_catalog_files='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ export 'XML_CATALOG_FILES=file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ XML_CATALOG_FILES='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ unset conda_catalog_files ifs_libxml2 rem
++ __conda_hashr
++ '[' -n '' ']'
++ '[' -n '' ']'
++ hash -r
++ export PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++ PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ scontrol show hostnames ip-26-0-169-207
++ export NODELIST=ip-26-0-169-207
++ NODELIST=ip-26-0-169-207
+++ scontrol show hostnames ip-26-0-169-207
+++ head -n1
++ export MASTER_NODE=ip-26-0-169-207
++ MASTER_NODE=ip-26-0-169-207
++ export MASTER_PORT=12356
++ MASTER_PORT=12356
++ export NNODES=1
++ NNODES=1
++ export GPUS_PER_NODE=8
++ GPUS_PER_NODE=8
++ export WORLD_SIZE=8
++ WORLD_SIZE=8
++ export CUDA_DEVICE_MAX_CONNECTIONS=1
++ CUDA_DEVICE_MAX_CONNECTIONS=1
++ export NCCL_DEBUG=INFO
++ NCCL_DEBUG=INFO
++ export NANOTRON_BENCHMARK=1
++ NANOTRON_BENCHMARK=1
++ echo 'Master node: ip-26-0-169-207'
+Master node: ip-26-0-169-207
++ echo 'All nodes: ip-26-0-169-207'
+All nodes: ip-26-0-169-207
++ echo 'World size: 8'
+World size: 8
++ srun torchrun --nnodes=1 --nproc_per_node=8 --rdzv_id=13404811 --rdzv_backend=c10d --rdzv_endpoint=ip-26-0-169-207:12356 run_train.py --config-file benchmark/configs/config_1.14G_dp4_tp2_pp1_acc256_mbs2_seq2048_zero1_tpmodeRED_vocab32k.yaml
+[2024-12-13 04:44:44,960] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-13 04:44:44,960] torch.distributed.run: [WARNING] 
+[2024-12-13 04:44:44,960] torch.distributed.run: [WARNING] *****************************************
+[2024-12-13 04:44:44,960] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-13 04:44:44,960] torch.distributed.run: [WARNING] *****************************************
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+[2024-12-13 04:44:50,086] torch.distributed.elastic.multiprocessing.api: [ERROR] failed (exitcode: 2) local_rank: 0 (pid: 498779) of binary: /fsx/nouamane/miniconda/envs/2-1-cu121/bin/python
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 33, in <module>
+    sys.exit(load_entry_point('torch==2.1.1', 'console_scripts', 'torchrun')())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 264, in launch_agent
+    raise ChildFailedError(
+torch.distributed.elastic.multiprocessing.errors.ChildFailedError: 
+============================================================
+run_train.py FAILED
+------------------------------------------------------------
+Failures:
+[1]:
+  time      : 2024-12-13_04:44:50
+  host      : ip-26-0-169-207.ec2.internal
+  rank      : 1 (local_rank: 1)
+  exitcode  : 2 (pid: 498780)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[2]:
+  time      : 2024-12-13_04:44:50
+  host      : ip-26-0-169-207.ec2.internal
+  rank      : 2 (local_rank: 2)
+  exitcode  : 2 (pid: 498781)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[3]:
+  time      : 2024-12-13_04:44:50
+  host      : ip-26-0-169-207.ec2.internal
+  rank      : 3 (local_rank: 3)
+  exitcode  : 2 (pid: 498782)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[4]:
+  time      : 2024-12-13_04:44:50
+  host      : ip-26-0-169-207.ec2.internal
+  rank      : 4 (local_rank: 4)
+  exitcode  : 2 (pid: 498783)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[5]:
+  time      : 2024-12-13_04:44:50
+  host      : ip-26-0-169-207.ec2.internal
+  rank      : 5 (local_rank: 5)
+  exitcode  : 2 (pid: 498784)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[6]:
+  time      : 2024-12-13_04:44:50
+  host      : ip-26-0-169-207.ec2.internal
+  rank      : 6 (local_rank: 6)
+  exitcode  : 2 (pid: 498785)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[7]:
+  time      : 2024-12-13_04:44:50
+  host      : ip-26-0-169-207.ec2.internal
+  rank      : 7 (local_rank: 7)
+  exitcode  : 2 (pid: 498786)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+------------------------------------------------------------
+Root Cause (first observed failure):
+[0]:
+  time      : 2024-12-13_04:44:50
+  host      : ip-26-0-169-207.ec2.internal
+  rank      : 0 (local_rank: 0)
+  exitcode  : 2 (pid: 498779)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+============================================================
+srun: error: ip-26-0-169-207: task 0: Exited with exit code 1
diff --git a/logs/13404950-bench_stress_test.out b/logs/13404950-bench_stress_test.out
new file mode 100644
index 0000000000000000000000000000000000000000..217172167ac1d7227b5d84b72c51e578af388963
--- /dev/null
+++ b/logs/13404950-bench_stress_test.out
@@ -0,0 +1,518 @@
++ source /etc/profile.d/modules.sh
+++ . /usr/share/modules/init/bash
++++ unset _mlshdbg
++++ '[' 0 = 1 ']'
++++ unset _mlre _mlIFS
++++ '[' -n x ']'
++++ _mlIFS=' 	
+'
++++ IFS=' '
++++ '[' -n '' ']'
+++++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash autoinit
++++ _mlcode='module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+MODULEPATH_modshare=/etc/environment-modules/modules:1:/usr/share/modules/\$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1; export MODULEPATH_modshare;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+LOADEDMODULES=; export LOADEDMODULES;
+MODULEPATH=/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/\$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles; export MODULEPATH;
+test 0;'
++++ _mlret=0
++++ '[' -n x ']'
++++ IFS=' 	
+'
++++ unset _mlIFS
++++ unset _mlre _mlv _mlrv
++++ '[' 0 -eq 0 ']'
++++ eval 'module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+MODULEPATH_modshare=/etc/environment-modules/modules:1:/usr/share/modules/\$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1; export MODULEPATH_modshare;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+LOADEDMODULES=; export LOADEDMODULES;
+MODULEPATH=/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/\$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles; export MODULEPATH;
+test 0;'
+++++ MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl
+++++ export MODULES_CMD
+++++ ENV=/usr/share/modules/init/profile.sh
+++++ export ENV
+++++ MODULEPATH_modshare='/etc/environment-modules/modules:1:/usr/share/modules/$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1'
+++++ export MODULEPATH_modshare
+++++ BASH_ENV=/usr/share/modules/init/bash
+++++ export BASH_ENV
+++++ MODULESHOME=/usr/share/modules
+++++ export MODULESHOME
+++++ LOADEDMODULES=
+++++ export LOADEDMODULES
+++++ MODULEPATH='/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles'
+++++ export MODULEPATH
+++++ test 0
++++ '[' 0 = 1 ']'
++++ '[' -t 2 ']'
++++ export -f module
++++ export -f switchml
++++ '[' 5 -ge 3 ']'
++++ [[ ehxB =~ i ]]
++++ [[ ! :/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin: =~ :/usr/bin: ]]
++++ '[' '!' -n '' ']'
++++ MANPATH=:
++++ export MANPATH
+++++ manpath
++++ [[ ! :/admin/home/nouamane/.local/share/man:/fsx/nouamane/miniconda/envs/2-1-cu121/man:/fsx/nouamane/miniconda/envs/2-1-cu121/share/man:/fsx/nouamane/miniconda/man:/fsx/nouamane/miniconda/share/man:/opt/amazon/openmpi/share/man:/opt/amazon/efa/share/man:/opt/slurm/share/man:/usr/local/man:/usr/local/share/man:/usr/share/man:/admin/home/nouamane/.fzf/man:: =~ :/usr/share/man: ]]
++++ unset _mlcode _mlret
++++ '[' -n '' ']'
++ module load cuda/12.1
++ unset _mlshdbg
++ '[' 0 = 1 ']'
++ unset _mlre _mlIFS
++ '[' -n x ']'
++ _mlIFS=' 	
+'
++ IFS=' '
++ '[' -n '' ']'
+++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash load cuda/12.1
++ eval 'CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include;' export 'CPATH;
+LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:;' export 'LD_LIBRARY_PATH;
+FI_EFA_FORK_SAFE=1;' export 'FI_EFA_FORK_SAFE;
+MANPATH=/usr/local/cuda-12.1/share/man::;' export 'MANPATH;
+LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64;' export 'LIBRARY_PATH;
+_LMFILES_=/usr/share/modules/modulefiles/cuda/12.1;' export '_LMFILES_;
+LOADEDMODULES=cuda/12.1;' export 'LOADEDMODULES;
+MPI_PATH=/opt/amazon/openmpi;' export 'MPI_PATH;
+NCCL_HOME_modshare=/opt/nccl/build:1;' export 'NCCL_HOME_modshare;
+NCCL_PROTO=simple;' export 'NCCL_PROTO;
+MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1;' export 'MANPATH_modshare;
+LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1;' export 'LIBRARY_PATH_modshare;
+NCCL_SOCKET_IFNAME=enp;' export 'NCCL_SOCKET_IFNAME;
+AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl;' export 'AWS_OFI_NCCL_HOME;
+NCCL_HOME=/opt/nccl/build;' export 'NCCL_HOME;
+FI_PROVIDER=efa;' export 'FI_PROVIDER;
+AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1;' export 'AWS_OFI_NCCL_HOME_modshare;
+CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1;' export 'CPATH_modshare;
+LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2::1:/opt/nccl/build/lib:1:/opt/aws-ofi-nccl/lib:2:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/efa/lib:1;' export 'LD_LIBRARY_PATH_modshare;
+FI_EFA_ENABLE_SHM_TRANSFER=1;' export 'FI_EFA_ENABLE_SHM_TRANSFER;
+_LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1;' export '_LMFILES__modshare;
+LOADEDMODULES_modshare=cuda/12.1:1;' export 'LOADEDMODULES_modshare;
+MPI_PATH_modshare=/opt/amazon/openmpi:1;' export 'MPI_PATH_modshare;
+PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin;' export 'PATH;
+CUDA_HOME=/usr/local/cuda-12.1;' export 'CUDA_HOME;
+PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/usr/local/cuda-12.1/include:1:/opt/amazon/efa/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/usr/sbin:1:/fsx/nouamane/miniconda/bin:1:/opt/slurm/bin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:2:/opt/amazon/openmpi/bin:1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1;' export 'PATH_modshare;
+test' '0;'
+++ CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include
+++ export CPATH
+++ LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:
+++ export LD_LIBRARY_PATH
+++ FI_EFA_FORK_SAFE=1
+++ export FI_EFA_FORK_SAFE
+++ MANPATH=/usr/local/cuda-12.1/share/man::
+++ export MANPATH
+++ LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64
+++ export LIBRARY_PATH
+++ _LMFILES_=/usr/share/modules/modulefiles/cuda/12.1
+++ export _LMFILES_
+++ LOADEDMODULES=cuda/12.1
+++ export LOADEDMODULES
+++ MPI_PATH=/opt/amazon/openmpi
+++ export MPI_PATH
+++ NCCL_HOME_modshare=/opt/nccl/build:1
+++ export NCCL_HOME_modshare
+++ NCCL_PROTO=simple
+++ export NCCL_PROTO
+++ MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1
+++ export MANPATH_modshare
+++ LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1
+++ export LIBRARY_PATH_modshare
+++ NCCL_SOCKET_IFNAME=enp
+++ export NCCL_SOCKET_IFNAME
+++ AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl
+++ export AWS_OFI_NCCL_HOME
+++ NCCL_HOME=/opt/nccl/build
+++ export NCCL_HOME
+++ FI_PROVIDER=efa
+++ export FI_PROVIDER
+++ AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1
+++ export AWS_OFI_NCCL_HOME_modshare
+++ CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1
+++ export CPATH_modshare
+++ LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2::1:/opt/nccl/build/lib:1:/opt/aws-ofi-nccl/lib:2:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/efa/lib:1
+++ export LD_LIBRARY_PATH_modshare
+++ FI_EFA_ENABLE_SHM_TRANSFER=1
+++ export FI_EFA_ENABLE_SHM_TRANSFER
+++ _LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1
+++ export _LMFILES__modshare
+++ LOADEDMODULES_modshare=cuda/12.1:1
+++ export LOADEDMODULES_modshare
+++ MPI_PATH_modshare=/opt/amazon/openmpi:1
+++ export MPI_PATH_modshare
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ export PATH
+++ CUDA_HOME=/usr/local/cuda-12.1
+++ export CUDA_HOME
+++ PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/usr/local/cuda-12.1/include:1:/opt/amazon/efa/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/usr/sbin:1:/fsx/nouamane/miniconda/bin:1:/opt/slurm/bin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:2:/opt/amazon/openmpi/bin:1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1
+++ export PATH_modshare
+++ test 0
++ _mlstatus=0
++ '[' -n x ']'
++ IFS=' 	
+'
++ unset _mlre _mlv _mlrv _mlIFS
++ '[' -n '' ']'
++ unset _mlshdbg
++ return 0
++ source /fsx/nouamane/miniconda/bin/activate
+++ _CONDA_ROOT=/fsx/nouamane/miniconda
+++ . /fsx/nouamane/miniconda/etc/profile.d/conda.sh
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ '[' -z x ']'
+++ conda activate
+++ local cmd=activate
+++ case "$cmd" in
+++ __conda_activate activate
+++ '[' -n '' ']'
+++ local ask_conda
++++ PS1=
++++ __conda_exe shell.posix activate
++++ /fsx/nouamane/miniconda/bin/conda shell.posix activate
+++ ask_conda='. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_2='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\'''
+++ eval '. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_2='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\'''
++++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh
+++++ test -n ''
+++++ unset XML_CATALOG_FILES
+++++ unset xml_catalog_files_libxml2
++++ PS1='(base) '
++++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++++ export CONDA_PREFIX=/fsx/nouamane/miniconda
++++ CONDA_PREFIX=/fsx/nouamane/miniconda
++++ export CONDA_SHLVL=3
++++ CONDA_SHLVL=3
++++ export CONDA_DEFAULT_ENV=base
++++ CONDA_DEFAULT_ENV=base
++++ export 'CONDA_PROMPT_MODIFIER=(base) '
++++ CONDA_PROMPT_MODIFIER='(base) '
++++ export CONDA_PREFIX_2=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ CONDA_PREFIX_2=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ __conda_hashr
+++ '[' -n '' ']'
+++ '[' -n '' ']'
+++ hash -r
++ conda activate 2-1-cu121
++ local cmd=activate
++ case "$cmd" in
++ __conda_activate activate 2-1-cu121
++ '[' -n '' ']'
++ local ask_conda
+++ PS1='(base) '
+++ __conda_exe shell.posix activate 2-1-cu121
+++ /fsx/nouamane/miniconda/bin/conda shell.posix activate 2-1-cu121
++ ask_conda='PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''4'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_3='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
++ eval 'PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''4'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_3='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
+++ PS1='(2-1-cu121) '
+++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ export CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ export CONDA_SHLVL=4
+++ CONDA_SHLVL=4
+++ export CONDA_DEFAULT_ENV=2-1-cu121
+++ CONDA_DEFAULT_ENV=2-1-cu121
+++ export 'CONDA_PROMPT_MODIFIER=(2-1-cu121) '
+++ CONDA_PROMPT_MODIFIER='(2-1-cu121) '
+++ export CONDA_PREFIX_3=/fsx/nouamane/miniconda
+++ CONDA_PREFIX_3=/fsx/nouamane/miniconda
+++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ export _CE_M=
+++ _CE_M=
+++ export _CE_CONDA=
+++ _CE_CONDA=
+++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh
++++ test -n ''
++++ xml_catalog_files_libxml2=
++++ XML_CATALOG_FILES=
++++ conda_catalog_files=
++++ ifs_libxml2=' 	
+'
++++ IFS=' '
++++ rem=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ for pre in ${rem}
++++ test '' = /fsx/nouamane/miniconda/envs/2-1-cu121
++++ conda_catalog_files=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ rem=
++++ IFS=' 	
+'
++++ conda_catalog_files='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ export 'XML_CATALOG_FILES=file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ XML_CATALOG_FILES='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ unset conda_catalog_files ifs_libxml2 rem
++ __conda_hashr
++ '[' -n '' ']'
++ '[' -n '' ']'
++ hash -r
++ export PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++ PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ scontrol show hostnames ip-26-0-169-207
++ export NODELIST=ip-26-0-169-207
++ NODELIST=ip-26-0-169-207
+++ scontrol show hostnames ip-26-0-169-207
+++ head -n1
++ export MASTER_NODE=ip-26-0-169-207
++ MASTER_NODE=ip-26-0-169-207
++ export MASTER_PORT=12356
++ MASTER_PORT=12356
++ export NNODES=1
++ NNODES=1
++ export GPUS_PER_NODE=8
++ GPUS_PER_NODE=8
++ export WORLD_SIZE=8
++ WORLD_SIZE=8
++ export CUDA_DEVICE_MAX_CONNECTIONS=1
++ CUDA_DEVICE_MAX_CONNECTIONS=1
++ export NCCL_DEBUG=INFO
++ NCCL_DEBUG=INFO
++ export NANOTRON_BENCHMARK=1
++ NANOTRON_BENCHMARK=1
++ echo 'Master node: ip-26-0-169-207'
+Master node: ip-26-0-169-207
++ echo 'All nodes: ip-26-0-169-207'
+All nodes: ip-26-0-169-207
++ echo 'World size: 8'
+World size: 8
++ srun torchrun --nnodes=1 --nproc_per_node=8 --rdzv_id=13404950 --rdzv_backend=c10d --rdzv_endpoint=ip-26-0-169-207:12356 run_train.py --config-file benchmark/configs/config_1.14G_dp4_tp2_pp1_acc256_mbs2_seq2048_zero1_tpmodeRED_vocab32k.yaml
+[2024-12-13 04:58:52,213] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-13 04:58:52,213] torch.distributed.run: [WARNING] 
+[2024-12-13 04:58:52,213] torch.distributed.run: [WARNING] *****************************************
+[2024-12-13 04:58:52,213] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-13 04:58:52,213] torch.distributed.run: [WARNING] *****************************************
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+[2024-12-13 04:58:57,540] torch.distributed.elastic.multiprocessing.api: [ERROR] failed (exitcode: 2) local_rank: 0 (pid: 515847) of binary: /fsx/nouamane/miniconda/envs/2-1-cu121/bin/python
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 33, in <module>
+    sys.exit(load_entry_point('torch==2.1.1', 'console_scripts', 'torchrun')())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 264, in launch_agent
+    raise ChildFailedError(
+torch.distributed.elastic.multiprocessing.errors.ChildFailedError: 
+============================================================
+run_train.py FAILED
+------------------------------------------------------------
+Failures:
+[1]:
+  time      : 2024-12-13_04:58:57
+  host      : ip-26-0-169-207.ec2.internal
+  rank      : 1 (local_rank: 1)
+  exitcode  : 2 (pid: 515848)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[2]:
+  time      : 2024-12-13_04:58:57
+  host      : ip-26-0-169-207.ec2.internal
+  rank      : 2 (local_rank: 2)
+  exitcode  : 2 (pid: 515849)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[3]:
+  time      : 2024-12-13_04:58:57
+  host      : ip-26-0-169-207.ec2.internal
+  rank      : 3 (local_rank: 3)
+  exitcode  : 2 (pid: 515850)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[4]:
+  time      : 2024-12-13_04:58:57
+  host      : ip-26-0-169-207.ec2.internal
+  rank      : 4 (local_rank: 4)
+  exitcode  : 2 (pid: 515851)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[5]:
+  time      : 2024-12-13_04:58:57
+  host      : ip-26-0-169-207.ec2.internal
+  rank      : 5 (local_rank: 5)
+  exitcode  : 2 (pid: 515852)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[6]:
+  time      : 2024-12-13_04:58:57
+  host      : ip-26-0-169-207.ec2.internal
+  rank      : 6 (local_rank: 6)
+  exitcode  : 2 (pid: 515853)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[7]:
+  time      : 2024-12-13_04:58:57
+  host      : ip-26-0-169-207.ec2.internal
+  rank      : 7 (local_rank: 7)
+  exitcode  : 2 (pid: 515854)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+------------------------------------------------------------
+Root Cause (first observed failure):
+[0]:
+  time      : 2024-12-13_04:58:57
+  host      : ip-26-0-169-207.ec2.internal
+  rank      : 0 (local_rank: 0)
+  exitcode  : 2 (pid: 515847)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+============================================================
+srun: error: ip-26-0-169-207: task 0: Exited with exit code 1
diff --git a/logs/13405030-bench_stress_test.out b/logs/13405030-bench_stress_test.out
new file mode 100644
index 0000000000000000000000000000000000000000..8bf3e566d1ebd7f9216d3e6536d9a221621292fa
--- /dev/null
+++ b/logs/13405030-bench_stress_test.out
@@ -0,0 +1,518 @@
++ source /etc/profile.d/modules.sh
+++ . /usr/share/modules/init/bash
++++ unset _mlshdbg
++++ '[' 0 = 1 ']'
++++ unset _mlre _mlIFS
++++ '[' -n x ']'
++++ _mlIFS=' 	
+'
++++ IFS=' '
++++ '[' -n '' ']'
+++++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash autoinit
++++ _mlcode='module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+MODULEPATH_modshare=/etc/environment-modules/modules:1:/usr/share/modules/\$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1; export MODULEPATH_modshare;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+LOADEDMODULES=; export LOADEDMODULES;
+MODULEPATH=/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/\$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles; export MODULEPATH;
+test 0;'
++++ _mlret=0
++++ '[' -n x ']'
++++ IFS=' 	
+'
++++ unset _mlIFS
++++ unset _mlre _mlv _mlrv
++++ '[' 0 -eq 0 ']'
++++ eval 'module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+MODULEPATH_modshare=/etc/environment-modules/modules:1:/usr/share/modules/\$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1; export MODULEPATH_modshare;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+LOADEDMODULES=; export LOADEDMODULES;
+MODULEPATH=/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/\$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles; export MODULEPATH;
+test 0;'
+++++ MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl
+++++ export MODULES_CMD
+++++ ENV=/usr/share/modules/init/profile.sh
+++++ export ENV
+++++ MODULEPATH_modshare='/etc/environment-modules/modules:1:/usr/share/modules/$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1'
+++++ export MODULEPATH_modshare
+++++ BASH_ENV=/usr/share/modules/init/bash
+++++ export BASH_ENV
+++++ MODULESHOME=/usr/share/modules
+++++ export MODULESHOME
+++++ LOADEDMODULES=
+++++ export LOADEDMODULES
+++++ MODULEPATH='/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles'
+++++ export MODULEPATH
+++++ test 0
++++ '[' 0 = 1 ']'
++++ '[' -t 2 ']'
++++ export -f module
++++ export -f switchml
++++ '[' 5 -ge 3 ']'
++++ [[ ehxB =~ i ]]
++++ [[ ! :/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin: =~ :/usr/bin: ]]
++++ '[' '!' -n '' ']'
++++ MANPATH=:
++++ export MANPATH
+++++ manpath
++++ [[ ! :/admin/home/nouamane/.local/share/man:/fsx/nouamane/miniconda/envs/2-1-cu121/man:/fsx/nouamane/miniconda/envs/2-1-cu121/share/man:/fsx/nouamane/miniconda/man:/fsx/nouamane/miniconda/share/man:/opt/amazon/openmpi/share/man:/opt/amazon/efa/share/man:/opt/slurm/share/man:/usr/local/man:/usr/local/share/man:/usr/share/man:/admin/home/nouamane/.fzf/man:: =~ :/usr/share/man: ]]
++++ unset _mlcode _mlret
++++ '[' -n '' ']'
++ module load cuda/12.1
++ unset _mlshdbg
++ '[' 0 = 1 ']'
++ unset _mlre _mlIFS
++ '[' -n x ']'
++ _mlIFS=' 	
+'
++ IFS=' '
++ '[' -n '' ']'
+++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash load cuda/12.1
++ eval 'CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include;' export 'CPATH;
+LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:;' export 'LD_LIBRARY_PATH;
+FI_EFA_FORK_SAFE=1;' export 'FI_EFA_FORK_SAFE;
+MANPATH=/usr/local/cuda-12.1/share/man::;' export 'MANPATH;
+LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64;' export 'LIBRARY_PATH;
+_LMFILES_=/usr/share/modules/modulefiles/cuda/12.1;' export '_LMFILES_;
+LOADEDMODULES=cuda/12.1;' export 'LOADEDMODULES;
+MPI_PATH=/opt/amazon/openmpi;' export 'MPI_PATH;
+NCCL_HOME_modshare=/opt/nccl/build:1;' export 'NCCL_HOME_modshare;
+NCCL_PROTO=simple;' export 'NCCL_PROTO;
+MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1;' export 'MANPATH_modshare;
+LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1;' export 'LIBRARY_PATH_modshare;
+NCCL_SOCKET_IFNAME=enp;' export 'NCCL_SOCKET_IFNAME;
+AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl;' export 'AWS_OFI_NCCL_HOME;
+NCCL_HOME=/opt/nccl/build;' export 'NCCL_HOME;
+FI_PROVIDER=efa;' export 'FI_PROVIDER;
+AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1;' export 'AWS_OFI_NCCL_HOME_modshare;
+CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1;' export 'CPATH_modshare;
+LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2::1:/opt/nccl/build/lib:1:/opt/aws-ofi-nccl/lib:2:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/efa/lib:1;' export 'LD_LIBRARY_PATH_modshare;
+FI_EFA_ENABLE_SHM_TRANSFER=1;' export 'FI_EFA_ENABLE_SHM_TRANSFER;
+_LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1;' export '_LMFILES__modshare;
+LOADEDMODULES_modshare=cuda/12.1:1;' export 'LOADEDMODULES_modshare;
+MPI_PATH_modshare=/opt/amazon/openmpi:1;' export 'MPI_PATH_modshare;
+PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin;' export 'PATH;
+CUDA_HOME=/usr/local/cuda-12.1;' export 'CUDA_HOME;
+PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/usr/local/cuda-12.1/include:1:/opt/amazon/efa/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/usr/sbin:1:/fsx/nouamane/miniconda/bin:1:/opt/slurm/bin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:2:/opt/amazon/openmpi/bin:1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1;' export 'PATH_modshare;
+test' '0;'
+++ CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include
+++ export CPATH
+++ LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:
+++ export LD_LIBRARY_PATH
+++ FI_EFA_FORK_SAFE=1
+++ export FI_EFA_FORK_SAFE
+++ MANPATH=/usr/local/cuda-12.1/share/man::
+++ export MANPATH
+++ LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64
+++ export LIBRARY_PATH
+++ _LMFILES_=/usr/share/modules/modulefiles/cuda/12.1
+++ export _LMFILES_
+++ LOADEDMODULES=cuda/12.1
+++ export LOADEDMODULES
+++ MPI_PATH=/opt/amazon/openmpi
+++ export MPI_PATH
+++ NCCL_HOME_modshare=/opt/nccl/build:1
+++ export NCCL_HOME_modshare
+++ NCCL_PROTO=simple
+++ export NCCL_PROTO
+++ MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1
+++ export MANPATH_modshare
+++ LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1
+++ export LIBRARY_PATH_modshare
+++ NCCL_SOCKET_IFNAME=enp
+++ export NCCL_SOCKET_IFNAME
+++ AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl
+++ export AWS_OFI_NCCL_HOME
+++ NCCL_HOME=/opt/nccl/build
+++ export NCCL_HOME
+++ FI_PROVIDER=efa
+++ export FI_PROVIDER
+++ AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1
+++ export AWS_OFI_NCCL_HOME_modshare
+++ CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1
+++ export CPATH_modshare
+++ LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2::1:/opt/nccl/build/lib:1:/opt/aws-ofi-nccl/lib:2:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/efa/lib:1
+++ export LD_LIBRARY_PATH_modshare
+++ FI_EFA_ENABLE_SHM_TRANSFER=1
+++ export FI_EFA_ENABLE_SHM_TRANSFER
+++ _LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1
+++ export _LMFILES__modshare
+++ LOADEDMODULES_modshare=cuda/12.1:1
+++ export LOADEDMODULES_modshare
+++ MPI_PATH_modshare=/opt/amazon/openmpi:1
+++ export MPI_PATH_modshare
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ export PATH
+++ CUDA_HOME=/usr/local/cuda-12.1
+++ export CUDA_HOME
+++ PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/usr/local/cuda-12.1/include:1:/opt/amazon/efa/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/usr/sbin:1:/fsx/nouamane/miniconda/bin:1:/opt/slurm/bin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:2:/opt/amazon/openmpi/bin:1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1
+++ export PATH_modshare
+++ test 0
++ _mlstatus=0
++ '[' -n x ']'
++ IFS=' 	
+'
++ unset _mlre _mlv _mlrv _mlIFS
++ '[' -n '' ']'
++ unset _mlshdbg
++ return 0
++ source /fsx/nouamane/miniconda/bin/activate
+++ _CONDA_ROOT=/fsx/nouamane/miniconda
+++ . /fsx/nouamane/miniconda/etc/profile.d/conda.sh
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ '[' -z x ']'
+++ conda activate
+++ local cmd=activate
+++ case "$cmd" in
+++ __conda_activate activate
+++ '[' -n '' ']'
+++ local ask_conda
++++ PS1=
++++ __conda_exe shell.posix activate
++++ /fsx/nouamane/miniconda/bin/conda shell.posix activate
+++ ask_conda='. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_2='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\'''
+++ eval '. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_2='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\'''
++++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh
+++++ test -n ''
+++++ unset XML_CATALOG_FILES
+++++ unset xml_catalog_files_libxml2
++++ PS1='(base) '
++++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++++ export CONDA_PREFIX=/fsx/nouamane/miniconda
++++ CONDA_PREFIX=/fsx/nouamane/miniconda
++++ export CONDA_SHLVL=3
++++ CONDA_SHLVL=3
++++ export CONDA_DEFAULT_ENV=base
++++ CONDA_DEFAULT_ENV=base
++++ export 'CONDA_PROMPT_MODIFIER=(base) '
++++ CONDA_PROMPT_MODIFIER='(base) '
++++ export CONDA_PREFIX_2=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ CONDA_PREFIX_2=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ __conda_hashr
+++ '[' -n '' ']'
+++ '[' -n '' ']'
+++ hash -r
++ conda activate 2-1-cu121
++ local cmd=activate
++ case "$cmd" in
++ __conda_activate activate 2-1-cu121
++ '[' -n '' ']'
++ local ask_conda
+++ PS1='(base) '
+++ __conda_exe shell.posix activate 2-1-cu121
+++ /fsx/nouamane/miniconda/bin/conda shell.posix activate 2-1-cu121
++ ask_conda='PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''4'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_3='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
++ eval 'PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''4'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_3='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
+++ PS1='(2-1-cu121) '
+++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ export CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ export CONDA_SHLVL=4
+++ CONDA_SHLVL=4
+++ export CONDA_DEFAULT_ENV=2-1-cu121
+++ CONDA_DEFAULT_ENV=2-1-cu121
+++ export 'CONDA_PROMPT_MODIFIER=(2-1-cu121) '
+++ CONDA_PROMPT_MODIFIER='(2-1-cu121) '
+++ export CONDA_PREFIX_3=/fsx/nouamane/miniconda
+++ CONDA_PREFIX_3=/fsx/nouamane/miniconda
+++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ export _CE_M=
+++ _CE_M=
+++ export _CE_CONDA=
+++ _CE_CONDA=
+++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh
++++ test -n ''
++++ xml_catalog_files_libxml2=
++++ XML_CATALOG_FILES=
++++ conda_catalog_files=
++++ ifs_libxml2=' 	
+'
++++ IFS=' '
++++ rem=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ for pre in ${rem}
++++ test '' = /fsx/nouamane/miniconda/envs/2-1-cu121
++++ conda_catalog_files=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ rem=
++++ IFS=' 	
+'
++++ conda_catalog_files='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ export 'XML_CATALOG_FILES=file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ XML_CATALOG_FILES='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ unset conda_catalog_files ifs_libxml2 rem
++ __conda_hashr
++ '[' -n '' ']'
++ '[' -n '' ']'
++ hash -r
++ export PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++ PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ scontrol show hostnames ip-26-0-160-103
++ export NODELIST=ip-26-0-160-103
++ NODELIST=ip-26-0-160-103
+++ scontrol show hostnames ip-26-0-160-103
+++ head -n1
++ export MASTER_NODE=ip-26-0-160-103
++ MASTER_NODE=ip-26-0-160-103
++ export MASTER_PORT=12356
++ MASTER_PORT=12356
++ export NNODES=1
++ NNODES=1
++ export GPUS_PER_NODE=8
++ GPUS_PER_NODE=8
++ export WORLD_SIZE=8
++ WORLD_SIZE=8
++ export CUDA_DEVICE_MAX_CONNECTIONS=1
++ CUDA_DEVICE_MAX_CONNECTIONS=1
++ export NCCL_DEBUG=INFO
++ NCCL_DEBUG=INFO
++ export NANOTRON_BENCHMARK=1
++ NANOTRON_BENCHMARK=1
++ echo 'Master node: ip-26-0-160-103'
+Master node: ip-26-0-160-103
++ echo 'All nodes: ip-26-0-160-103'
+All nodes: ip-26-0-160-103
++ echo 'World size: 8'
+World size: 8
++ srun torchrun --nnodes=1 --nproc_per_node=8 --rdzv_id=13405030 --rdzv_backend=c10d --rdzv_endpoint=ip-26-0-160-103:12356 run_train.py --config-file benchmark/configs/config_1.14G_dp4_tp2_pp1_acc256_mbs2_seq2048_zero1_tpmodeRED_vocab32k.yaml
+[2024-12-13 05:08:59,026] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-13 05:08:59,026] torch.distributed.run: [WARNING] 
+[2024-12-13 05:08:59,026] torch.distributed.run: [WARNING] *****************************************
+[2024-12-13 05:08:59,026] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-13 05:08:59,026] torch.distributed.run: [WARNING] *****************************************
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+/fsx/nouamane/miniconda/envs/2-1-cu121/bin/python: can't open file '/fsx/nouamane/projects/run_train.py': [Errno 2] No such file or directory
+[2024-12-13 05:09:04,380] torch.distributed.elastic.multiprocessing.api: [ERROR] failed (exitcode: 2) local_rank: 0 (pid: 2416745) of binary: /fsx/nouamane/miniconda/envs/2-1-cu121/bin/python
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 33, in <module>
+    sys.exit(load_entry_point('torch==2.1.1', 'console_scripts', 'torchrun')())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 264, in launch_agent
+    raise ChildFailedError(
+torch.distributed.elastic.multiprocessing.errors.ChildFailedError: 
+============================================================
+run_train.py FAILED
+------------------------------------------------------------
+Failures:
+[1]:
+  time      : 2024-12-13_05:09:04
+  host      : ip-26-0-160-103.ec2.internal
+  rank      : 1 (local_rank: 1)
+  exitcode  : 2 (pid: 2416746)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[2]:
+  time      : 2024-12-13_05:09:04
+  host      : ip-26-0-160-103.ec2.internal
+  rank      : 2 (local_rank: 2)
+  exitcode  : 2 (pid: 2416747)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[3]:
+  time      : 2024-12-13_05:09:04
+  host      : ip-26-0-160-103.ec2.internal
+  rank      : 3 (local_rank: 3)
+  exitcode  : 2 (pid: 2416748)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[4]:
+  time      : 2024-12-13_05:09:04
+  host      : ip-26-0-160-103.ec2.internal
+  rank      : 4 (local_rank: 4)
+  exitcode  : 2 (pid: 2416749)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[5]:
+  time      : 2024-12-13_05:09:04
+  host      : ip-26-0-160-103.ec2.internal
+  rank      : 5 (local_rank: 5)
+  exitcode  : 2 (pid: 2416750)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[6]:
+  time      : 2024-12-13_05:09:04
+  host      : ip-26-0-160-103.ec2.internal
+  rank      : 6 (local_rank: 6)
+  exitcode  : 2 (pid: 2416751)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[7]:
+  time      : 2024-12-13_05:09:04
+  host      : ip-26-0-160-103.ec2.internal
+  rank      : 7 (local_rank: 7)
+  exitcode  : 2 (pid: 2416752)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+------------------------------------------------------------
+Root Cause (first observed failure):
+[0]:
+  time      : 2024-12-13_05:09:04
+  host      : ip-26-0-160-103.ec2.internal
+  rank      : 0 (local_rank: 0)
+  exitcode  : 2 (pid: 2416745)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+============================================================
+srun: error: ip-26-0-160-103: task 0: Exited with exit code 1
diff --git a/logs/13417747-bench_stress_test.out b/logs/13417747-bench_stress_test.out
new file mode 100644
index 0000000000000000000000000000000000000000..3e0dbb7805e72386f13f688ba5c5155d434d2fac
--- /dev/null
+++ b/logs/13417747-bench_stress_test.out
@@ -0,0 +1,3768 @@
++ source /etc/profile.d/modules.sh
+++ . /usr/share/modules/init/bash
++++ unset _mlshdbg
++++ '[' 0 = 1 ']'
++++ unset _mlre _mlIFS
++++ '[' -n x ']'
++++ _mlIFS=' 	
+'
++++ IFS=' '
++++ '[' -n '' ']'
+++++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash autoinit
++++ _mlcode='module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+MODULEPATH_modshare=/etc/environment-modules/modules:1:/usr/share/modules/\$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1; export MODULEPATH_modshare;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+LOADEDMODULES=; export LOADEDMODULES;
+MODULEPATH=/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/\$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles; export MODULEPATH;
+test 0;'
++++ _mlret=0
++++ '[' -n x ']'
++++ IFS=' 	
+'
++++ unset _mlIFS
++++ unset _mlre _mlv _mlrv
++++ '[' 0 -eq 0 ']'
++++ eval 'module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+MODULEPATH_modshare=/etc/environment-modules/modules:1:/usr/share/modules/\$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1; export MODULEPATH_modshare;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+LOADEDMODULES=; export LOADEDMODULES;
+MODULEPATH=/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/\$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles; export MODULEPATH;
+test 0;'
+++++ MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl
+++++ export MODULES_CMD
+++++ ENV=/usr/share/modules/init/profile.sh
+++++ export ENV
+++++ MODULEPATH_modshare='/etc/environment-modules/modules:1:/usr/share/modules/$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1'
+++++ export MODULEPATH_modshare
+++++ BASH_ENV=/usr/share/modules/init/bash
+++++ export BASH_ENV
+++++ MODULESHOME=/usr/share/modules
+++++ export MODULESHOME
+++++ LOADEDMODULES=
+++++ export LOADEDMODULES
+++++ MODULEPATH='/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles'
+++++ export MODULEPATH
+++++ test 0
++++ '[' 0 = 1 ']'
++++ '[' -t 2 ']'
++++ export -f module
++++ export -f switchml
++++ '[' 5 -ge 3 ']'
++++ [[ ehxB =~ i ]]
++++ [[ ! :/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin: =~ :/usr/bin: ]]
++++ '[' '!' -n '' ']'
++++ MANPATH=:
++++ export MANPATH
+++++ manpath
++++ [[ ! :/admin/home/nouamane/.local/share/man:/fsx/nouamane/miniconda/envs/2-1-cu121/man:/fsx/nouamane/miniconda/envs/2-1-cu121/share/man:/fsx/nouamane/miniconda/man:/fsx/nouamane/miniconda/share/man:/opt/amazon/openmpi/share/man:/opt/amazon/efa/share/man:/opt/slurm/share/man:/usr/local/man:/usr/local/share/man:/usr/share/man:/admin/home/nouamane/.fzf/man:: =~ :/usr/share/man: ]]
++++ unset _mlcode _mlret
++++ '[' -n '' ']'
++ module load cuda/12.1
++ unset _mlshdbg
++ '[' 0 = 1 ']'
++ unset _mlre _mlIFS
++ '[' -n x ']'
++ _mlIFS=' 	
+'
++ IFS=' '
++ '[' -n '' ']'
+++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash load cuda/12.1
++ eval 'CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include;' export 'CPATH;
+LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:;' export 'LD_LIBRARY_PATH;
+FI_EFA_FORK_SAFE=1;' export 'FI_EFA_FORK_SAFE;
+MANPATH=/usr/local/cuda-12.1/share/man::;' export 'MANPATH;
+LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64;' export 'LIBRARY_PATH;
+_LMFILES_=/usr/share/modules/modulefiles/cuda/12.1;' export '_LMFILES_;
+LOADEDMODULES=cuda/12.1;' export 'LOADEDMODULES;
+MPI_PATH=/opt/amazon/openmpi;' export 'MPI_PATH;
+NCCL_HOME_modshare=/opt/nccl/build:1;' export 'NCCL_HOME_modshare;
+NCCL_PROTO=simple;' export 'NCCL_PROTO;
+MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1;' export 'MANPATH_modshare;
+LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1;' export 'LIBRARY_PATH_modshare;
+NCCL_SOCKET_IFNAME=enp;' export 'NCCL_SOCKET_IFNAME;
+AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl;' export 'AWS_OFI_NCCL_HOME;
+NCCL_HOME=/opt/nccl/build;' export 'NCCL_HOME;
+FI_PROVIDER=efa;' export 'FI_PROVIDER;
+AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1;' export 'AWS_OFI_NCCL_HOME_modshare;
+CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1;' export 'CPATH_modshare;
+LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2::1:/opt/nccl/build/lib:1:/opt/aws-ofi-nccl/lib:2:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/efa/lib:1;' export 'LD_LIBRARY_PATH_modshare;
+FI_EFA_ENABLE_SHM_TRANSFER=1;' export 'FI_EFA_ENABLE_SHM_TRANSFER;
+_LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1;' export '_LMFILES__modshare;
+LOADEDMODULES_modshare=cuda/12.1:1;' export 'LOADEDMODULES_modshare;
+MPI_PATH_modshare=/opt/amazon/openmpi:1;' export 'MPI_PATH_modshare;
+PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin;' export 'PATH;
+CUDA_HOME=/usr/local/cuda-12.1;' export 'CUDA_HOME;
+PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/usr/local/cuda-12.1/include:1:/opt/amazon/efa/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/usr/sbin:1:/fsx/nouamane/miniconda/bin:1:/opt/slurm/bin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:2:/opt/amazon/openmpi/bin:1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1;' export 'PATH_modshare;
+test' '0;'
+++ CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include
+++ export CPATH
+++ LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:
+++ export LD_LIBRARY_PATH
+++ FI_EFA_FORK_SAFE=1
+++ export FI_EFA_FORK_SAFE
+++ MANPATH=/usr/local/cuda-12.1/share/man::
+++ export MANPATH
+++ LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64
+++ export LIBRARY_PATH
+++ _LMFILES_=/usr/share/modules/modulefiles/cuda/12.1
+++ export _LMFILES_
+++ LOADEDMODULES=cuda/12.1
+++ export LOADEDMODULES
+++ MPI_PATH=/opt/amazon/openmpi
+++ export MPI_PATH
+++ NCCL_HOME_modshare=/opt/nccl/build:1
+++ export NCCL_HOME_modshare
+++ NCCL_PROTO=simple
+++ export NCCL_PROTO
+++ MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1
+++ export MANPATH_modshare
+++ LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1
+++ export LIBRARY_PATH_modshare
+++ NCCL_SOCKET_IFNAME=enp
+++ export NCCL_SOCKET_IFNAME
+++ AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl
+++ export AWS_OFI_NCCL_HOME
+++ NCCL_HOME=/opt/nccl/build
+++ export NCCL_HOME
+++ FI_PROVIDER=efa
+++ export FI_PROVIDER
+++ AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1
+++ export AWS_OFI_NCCL_HOME_modshare
+++ CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1
+++ export CPATH_modshare
+++ LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2::1:/opt/nccl/build/lib:1:/opt/aws-ofi-nccl/lib:2:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/efa/lib:1
+++ export LD_LIBRARY_PATH_modshare
+++ FI_EFA_ENABLE_SHM_TRANSFER=1
+++ export FI_EFA_ENABLE_SHM_TRANSFER
+++ _LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1
+++ export _LMFILES__modshare
+++ LOADEDMODULES_modshare=cuda/12.1:1
+++ export LOADEDMODULES_modshare
+++ MPI_PATH_modshare=/opt/amazon/openmpi:1
+++ export MPI_PATH_modshare
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ export PATH
+++ CUDA_HOME=/usr/local/cuda-12.1
+++ export CUDA_HOME
+++ PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/usr/local/cuda-12.1/include:1:/opt/amazon/efa/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/usr/sbin:1:/fsx/nouamane/miniconda/bin:1:/opt/slurm/bin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:2:/opt/amazon/openmpi/bin:1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1
+++ export PATH_modshare
+++ test 0
++ _mlstatus=0
++ '[' -n x ']'
++ IFS=' 	
+'
++ unset _mlre _mlv _mlrv _mlIFS
++ '[' -n '' ']'
++ unset _mlshdbg
++ return 0
++ source /fsx/nouamane/miniconda/bin/activate
+++ _CONDA_ROOT=/fsx/nouamane/miniconda
+++ . /fsx/nouamane/miniconda/etc/profile.d/conda.sh
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ '[' -z x ']'
+++ conda activate
+++ local cmd=activate
+++ case "$cmd" in
+++ __conda_activate activate
+++ '[' -n '' ']'
+++ local ask_conda
++++ PS1=
++++ __conda_exe shell.posix activate
++++ /fsx/nouamane/miniconda/bin/conda shell.posix activate
+++ ask_conda='. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_2='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\'''
+++ eval '. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_2='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\'''
++++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh
+++++ test -n ''
+++++ unset XML_CATALOG_FILES
+++++ unset xml_catalog_files_libxml2
++++ PS1='(base) '
++++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++++ export CONDA_PREFIX=/fsx/nouamane/miniconda
++++ CONDA_PREFIX=/fsx/nouamane/miniconda
++++ export CONDA_SHLVL=3
++++ CONDA_SHLVL=3
++++ export CONDA_DEFAULT_ENV=base
++++ CONDA_DEFAULT_ENV=base
++++ export 'CONDA_PROMPT_MODIFIER=(base) '
++++ CONDA_PROMPT_MODIFIER='(base) '
++++ export CONDA_PREFIX_2=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ CONDA_PREFIX_2=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ __conda_hashr
+++ '[' -n '' ']'
+++ '[' -n '' ']'
+++ hash -r
++ conda activate 2-1-cu121
++ local cmd=activate
++ case "$cmd" in
++ __conda_activate activate 2-1-cu121
++ '[' -n '' ']'
++ local ask_conda
+++ PS1='(base) '
+++ __conda_exe shell.posix activate 2-1-cu121
+++ /fsx/nouamane/miniconda/bin/conda shell.posix activate 2-1-cu121
++ ask_conda='PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''4'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_3='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
++ eval 'PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''4'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_3='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
+++ PS1='(2-1-cu121) '
+++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ export CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ export CONDA_SHLVL=4
+++ CONDA_SHLVL=4
+++ export CONDA_DEFAULT_ENV=2-1-cu121
+++ CONDA_DEFAULT_ENV=2-1-cu121
+++ export 'CONDA_PROMPT_MODIFIER=(2-1-cu121) '
+++ CONDA_PROMPT_MODIFIER='(2-1-cu121) '
+++ export CONDA_PREFIX_3=/fsx/nouamane/miniconda
+++ CONDA_PREFIX_3=/fsx/nouamane/miniconda
+++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ export _CE_M=
+++ _CE_M=
+++ export _CE_CONDA=
+++ _CE_CONDA=
+++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh
++++ test -n ''
++++ xml_catalog_files_libxml2=
++++ XML_CATALOG_FILES=
++++ conda_catalog_files=
++++ ifs_libxml2=' 	
+'
++++ IFS=' '
++++ rem=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ for pre in ${rem}
++++ test '' = /fsx/nouamane/miniconda/envs/2-1-cu121
++++ conda_catalog_files=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ rem=
++++ IFS=' 	
+'
++++ conda_catalog_files='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ export 'XML_CATALOG_FILES=file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ XML_CATALOG_FILES='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ unset conda_catalog_files ifs_libxml2 rem
++ __conda_hashr
++ '[' -n '' ']'
++ '[' -n '' ']'
++ hash -r
++ export PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++ PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ scontrol show hostnames ip-26-0-169-207
++ export NODELIST=ip-26-0-169-207
++ NODELIST=ip-26-0-169-207
+++ scontrol show hostnames ip-26-0-169-207
+++ head -n1
++ export MASTER_NODE=ip-26-0-169-207
++ MASTER_NODE=ip-26-0-169-207
++ export MASTER_PORT=12356
++ MASTER_PORT=12356
++ export NNODES=1
++ NNODES=1
++ export GPUS_PER_NODE=8
++ GPUS_PER_NODE=8
++ export WORLD_SIZE=8
++ WORLD_SIZE=8
++ export CUDA_DEVICE_MAX_CONNECTIONS=1
++ CUDA_DEVICE_MAX_CONNECTIONS=1
++ export NCCL_DEBUG=INFO
++ NCCL_DEBUG=INFO
++ export NANOTRON_BENCHMARK=1
++ NANOTRON_BENCHMARK=1
++ echo '=== GPU Topology ==='
+=== GPU Topology ===
++ nvidia-smi topo -m
+	GPU0	GPU1	GPU2	GPU3	GPU4	GPU5	GPU6	GPU7	CPU Affinity	NUMA Affinity	GPU NUMA ID
+GPU0	 X 	NV18	NV18	NV18	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU1	NV18	 X 	NV18	NV18	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU2	NV18	NV18	 X 	NV18	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU3	NV18	NV18	NV18	 X 	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU4	NV18	NV18	NV18	NV18	 X 	NV18	NV18	NV18	48-95	1		N/A
+GPU5	NV18	NV18	NV18	NV18	NV18	 X 	NV18	NV18	48-95	1		N/A
+GPU6	NV18	NV18	NV18	NV18	NV18	NV18	 X 	NV18	48-95	1		N/A
+GPU7	NV18	NV18	NV18	NV18	NV18	NV18	NV18	 X 	48-95	1		N/A
+
+Legend:
+
+  X    = Self
+  SYS  = Connection traversing PCIe as well as the SMP interconnect between NUMA nodes (e.g., QPI/UPI)
+  NODE = Connection traversing PCIe as well as the interconnect between PCIe Host Bridges within a NUMA node
+  PHB  = Connection traversing PCIe as well as a PCIe Host Bridge (typically the CPU)
+  PXB  = Connection traversing multiple PCIe bridges (without traversing the PCIe Host Bridge)
+  PIX  = Connection traversing at most a single PCIe bridge
+  NV#  = Connection traversing a bonded set of # NVLinks
++ echo ==================
+==================
++ echo 'Master node: ip-26-0-169-207'
+Master node: ip-26-0-169-207
++ echo 'All nodes: ip-26-0-169-207'
+All nodes: ip-26-0-169-207
++ echo 'World size: 8'
+World size: 8
++ srun torchrun --nnodes=1 --nproc_per_node=8 --rdzv_id=13417747 --rdzv_backend=c10d --rdzv_endpoint=ip-26-0-169-207:12356 /fsx/nouamane/projects/nanotron/run_train.py --config-file benchmark/configs/config_1.14G_dp4_tp2_pp1_acc256_mbs2_seq2048_zero1_tpmodeRED_vocab32k.yaml
+[2024-12-13 12:22:02,893] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-13 12:22:02,893] torch.distributed.run: [WARNING] 
+[2024-12-13 12:22:02,893] torch.distributed.run: [WARNING] *****************************************
+[2024-12-13 12:22:02,893] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-13 12:22:02,893] torch.distributed.run: [WARNING] *****************************************
+ip-26-0-169-207:675551:675551 [0] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-169-207:675551:675551 [0] NCCL INFO Bootstrap : Using enp71s0:26.0.169.207<0>
+ip-26-0-169-207:675551:675551 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-169-207:675551:675551 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-169-207:675551:675551 [0] NCCL INFO cudaDriverVersion 12020
+NCCL version 2.18.5+cuda12.2
+ip-26-0-169-207:675555:675555 [4] NCCL INFO cudaDriverVersion 12020
+ip-26-0-169-207:675553:675553 [2] NCCL INFO cudaDriverVersion 12020
+ip-26-0-169-207:675555:675555 [4] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-169-207:675557:675557 [6] NCCL INFO cudaDriverVersion 12020
+ip-26-0-169-207:675553:675553 [2] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-169-207:675557:675557 [6] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-169-207:675555:675555 [4] NCCL INFO Bootstrap : Using enp71s0:26.0.169.207<0>
+ip-26-0-169-207:675557:675557 [6] NCCL INFO Bootstrap : Using enp71s0:26.0.169.207<0>
+ip-26-0-169-207:675553:675553 [2] NCCL INFO Bootstrap : Using enp71s0:26.0.169.207<0>
+ip-26-0-169-207:675555:675555 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-169-207:675555:675555 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-169-207:675557:675557 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-169-207:675553:675553 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-169-207:675557:675557 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-169-207:675553:675553 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-169-207:675552:675552 [1] NCCL INFO cudaDriverVersion 12020
+ip-26-0-169-207:675552:675552 [1] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-169-207:675552:675552 [1] NCCL INFO Bootstrap : Using enp71s0:26.0.169.207<0>
+ip-26-0-169-207:675558:675558 [7] NCCL INFO cudaDriverVersion 12020
+ip-26-0-169-207:675552:675552 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-169-207:675552:675552 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-169-207:675556:675556 [5] NCCL INFO cudaDriverVersion 12020
+ip-26-0-169-207:675558:675558 [7] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-169-207:675556:675556 [5] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-169-207:675558:675558 [7] NCCL INFO Bootstrap : Using enp71s0:26.0.169.207<0>
+ip-26-0-169-207:675556:675556 [5] NCCL INFO Bootstrap : Using enp71s0:26.0.169.207<0>
+ip-26-0-169-207:675558:675558 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-169-207:675558:675558 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-169-207:675556:675556 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-169-207:675556:675556 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-169-207:675554:675554 [3] NCCL INFO cudaDriverVersion 12020
+ip-26-0-169-207:675554:675554 [3] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-169-207:675554:675554 [3] NCCL INFO Bootstrap : Using enp71s0:26.0.169.207<0>
+ip-26-0-169-207:675554:675554 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-169-207:675554:675554 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-169-207:675551:675673 [0] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-169-207:675551:675673 [0] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-169-207:675551:675673 [0] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-169-207:675551:675673 [0] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-169-207:675551:675673 [0] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-169-207:675551:675673 [0] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-169-207:675551:675673 [0] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-169-207:675551:675673 [0] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-169-207:675551:675673 [0] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-169-207:675555:675675 [4] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-169-207:675555:675675 [4] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-169-207:675555:675675 [4] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-169-207:675555:675675 [4] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-169-207:675555:675675 [4] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-169-207:675555:675675 [4] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-169-207:675555:675675 [4] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-169-207:675555:675675 [4] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-169-207:675555:675675 [4] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-169-207:675552:675677 [1] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-169-207:675552:675677 [1] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-169-207:675553:675674 [2] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-169-207:675553:675674 [2] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-169-207:675558:675676 [7] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-169-207:675558:675676 [7] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-169-207:675554:675680 [3] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-169-207:675554:675680 [3] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-169-207:675556:675679 [5] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-169-207:675556:675679 [5] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-169-207:675552:675677 [1] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-169-207:675552:675677 [1] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-169-207:675552:675677 [1] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-169-207:675552:675677 [1] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-169-207:675552:675677 [1] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-169-207:675552:675677 [1] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-169-207:675552:675677 [1] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-169-207:675553:675674 [2] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-169-207:675553:675674 [2] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-169-207:675553:675674 [2] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-169-207:675553:675674 [2] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-169-207:675553:675674 [2] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-169-207:675553:675674 [2] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-169-207:675553:675674 [2] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-169-207:675558:675676 [7] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-169-207:675558:675676 [7] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-169-207:675558:675676 [7] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-169-207:675558:675676 [7] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-169-207:675558:675676 [7] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-169-207:675558:675676 [7] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-169-207:675558:675676 [7] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-169-207:675556:675679 [5] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-169-207:675556:675679 [5] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-169-207:675556:675679 [5] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-169-207:675556:675679 [5] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-169-207:675556:675679 [5] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-169-207:675556:675679 [5] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-169-207:675556:675679 [5] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-169-207:675554:675680 [3] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-169-207:675554:675680 [3] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-169-207:675554:675680 [3] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-169-207:675554:675680 [3] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-169-207:675554:675680 [3] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-169-207:675554:675680 [3] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-169-207:675554:675680 [3] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-169-207:675557:675678 [6] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-169-207:675557:675678 [6] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-169-207:675557:675678 [6] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-169-207:675557:675678 [6] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-169-207:675557:675678 [6] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-169-207:675557:675678 [6] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-169-207:675557:675678 [6] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-169-207:675557:675678 [6] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-169-207:675557:675678 [6] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-169-207:675551:675673 [0] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-169-207:675555:675675 [4] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-169-207:675554:675680 [3] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-169-207:675553:675674 [2] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-169-207:675558:675676 [7] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-169-207:675552:675677 [1] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-169-207:675556:675679 [5] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-169-207:675557:675678 [6] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-169-207:675551:675673 [0] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-169-207:675551:675673 [0] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-169-207:675551:675673 [0] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-169-207:675551:675673 [0] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-169-207:675551:675673 [0] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-169-207:675551:675673 [0] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-169-207:675551:675673 [0] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-169-207:675551:675673 [0] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-169-207:675551:675673 [0] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-169-207:675551:675673 [0] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-169-207:675551:675673 [0] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-169-207:675551:675673 [0] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-169-207:675551:675673 [0] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-169-207:675551:675673 [0] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-169-207:675551:675673 [0] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-169-207:675551:675673 [0] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-169-207:675551:675673 [0] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-169-207:675551:675673 [0] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-169-207:675551:675673 [0] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-169-207:675551:675673 [0] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-169-207:675551:675673 [0] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-169-207:675551:675673 [0] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-169-207:675551:675673 [0] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-169-207:675551:675673 [0] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-169-207:675551:675673 [0] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-169-207:675551:675673 [0] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-169-207:675551:675673 [0] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-169-207:675551:675673 [0] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-169-207:675551:675673 [0] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-169-207:675551:675673 [0] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-169-207:675551:675673 [0] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-169-207:675551:675673 [0] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-169-207:675551:675673 [0] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/162
+ip-26-0-169-207:675551:675673 [0] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-169-207:675551:675673 [0] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-169-207:675551:675673 [0] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-169-207:675551:675673 [0] NCCL INFO Using network Libfabric
+ip-26-0-169-207:675555:675675 [4] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-169-207:675555:675675 [4] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-169-207:675555:675675 [4] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-169-207:675555:675675 [4] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-169-207:675555:675675 [4] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-169-207:675555:675675 [4] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-169-207:675555:675675 [4] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-169-207:675555:675675 [4] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-169-207:675555:675675 [4] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-169-207:675555:675675 [4] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-169-207:675555:675675 [4] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-169-207:675555:675675 [4] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-169-207:675555:675675 [4] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-169-207:675555:675675 [4] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-169-207:675555:675675 [4] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-169-207:675555:675675 [4] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-169-207:675555:675675 [4] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-169-207:675555:675675 [4] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-169-207:675555:675675 [4] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-169-207:675555:675675 [4] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-169-207:675555:675675 [4] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-169-207:675555:675675 [4] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-169-207:675555:675675 [4] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-169-207:675555:675675 [4] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-169-207:675555:675675 [4] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-169-207:675555:675675 [4] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-169-207:675555:675675 [4] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-169-207:675555:675675 [4] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-169-207:675555:675675 [4] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-169-207:675555:675675 [4] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-169-207:675555:675675 [4] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-169-207:675555:675675 [4] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-169-207:675555:675675 [4] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/149
+ip-26-0-169-207:675555:675675 [4] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-169-207:675556:675679 [5] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-169-207:675556:675679 [5] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-169-207:675556:675679 [5] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-169-207:675556:675679 [5] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-169-207:675556:675679 [5] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-169-207:675556:675679 [5] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-169-207:675556:675679 [5] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-169-207:675556:675679 [5] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-169-207:675556:675679 [5] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-169-207:675556:675679 [5] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-169-207:675556:675679 [5] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-169-207:675556:675679 [5] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-169-207:675556:675679 [5] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-169-207:675556:675679 [5] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-169-207:675556:675679 [5] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-169-207:675556:675679 [5] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-169-207:675556:675679 [5] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-169-207:675556:675679 [5] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-169-207:675556:675679 [5] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-169-207:675556:675679 [5] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-169-207:675556:675679 [5] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-169-207:675556:675679 [5] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-169-207:675556:675679 [5] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-169-207:675556:675679 [5] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-169-207:675556:675679 [5] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-169-207:675556:675679 [5] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-169-207:675556:675679 [5] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-169-207:675556:675679 [5] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-169-207:675556:675679 [5] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-169-207:675556:675679 [5] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-169-207:675556:675679 [5] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-169-207:675556:675679 [5] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-169-207:675556:675679 [5] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/149
+ip-26-0-169-207:675556:675679 [5] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-169-207:675557:675678 [6] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-169-207:675557:675678 [6] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-169-207:675557:675678 [6] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-169-207:675557:675678 [6] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-169-207:675557:675678 [6] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-169-207:675557:675678 [6] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-169-207:675557:675678 [6] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-169-207:675557:675678 [6] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-169-207:675557:675678 [6] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-169-207:675557:675678 [6] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-169-207:675557:675678 [6] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-169-207:675557:675678 [6] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-169-207:675557:675678 [6] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-169-207:675557:675678 [6] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-169-207:675557:675678 [6] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-169-207:675557:675678 [6] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-169-207:675557:675678 [6] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-169-207:675557:675678 [6] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-169-207:675557:675678 [6] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-169-207:675557:675678 [6] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-169-207:675557:675678 [6] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-169-207:675557:675678 [6] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-169-207:675557:675678 [6] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-169-207:675557:675678 [6] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-169-207:675557:675678 [6] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-169-207:675557:675678 [6] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-169-207:675557:675678 [6] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-169-207:675557:675678 [6] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-169-207:675557:675678 [6] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-169-207:675557:675678 [6] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-169-207:675557:675678 [6] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-169-207:675557:675678 [6] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-169-207:675557:675678 [6] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/149
+ip-26-0-169-207:675557:675678 [6] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-169-207:675558:675676 [7] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-169-207:675558:675676 [7] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-169-207:675558:675676 [7] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-169-207:675558:675676 [7] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-169-207:675558:675676 [7] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-169-207:675558:675676 [7] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-169-207:675558:675676 [7] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-169-207:675558:675676 [7] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-169-207:675558:675676 [7] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-169-207:675558:675676 [7] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-169-207:675558:675676 [7] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-169-207:675558:675676 [7] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-169-207:675558:675676 [7] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-169-207:675558:675676 [7] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-169-207:675558:675676 [7] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-169-207:675558:675676 [7] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-169-207:675558:675676 [7] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-169-207:675558:675676 [7] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-169-207:675558:675676 [7] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-169-207:675558:675676 [7] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-169-207:675558:675676 [7] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-169-207:675558:675676 [7] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-169-207:675558:675676 [7] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-169-207:675558:675676 [7] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-169-207:675558:675676 [7] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-169-207:675558:675676 [7] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-169-207:675558:675676 [7] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-169-207:675558:675676 [7] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-169-207:675558:675676 [7] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-169-207:675558:675676 [7] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-169-207:675558:675676 [7] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-169-207:675558:675676 [7] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-169-207:675558:675676 [7] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/149
+ip-26-0-169-207:675558:675676 [7] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-169-207:675553:675674 [2] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-169-207:675553:675674 [2] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-169-207:675553:675674 [2] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-169-207:675553:675674 [2] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-169-207:675553:675674 [2] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-169-207:675553:675674 [2] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-169-207:675553:675674 [2] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-169-207:675553:675674 [2] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-169-207:675553:675674 [2] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-169-207:675553:675674 [2] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-169-207:675553:675674 [2] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-169-207:675553:675674 [2] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-169-207:675553:675674 [2] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-169-207:675553:675674 [2] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-169-207:675553:675674 [2] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-169-207:675553:675674 [2] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-169-207:675553:675674 [2] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-169-207:675553:675674 [2] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-169-207:675553:675674 [2] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-169-207:675553:675674 [2] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-169-207:675553:675674 [2] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-169-207:675553:675674 [2] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-169-207:675553:675674 [2] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-169-207:675553:675674 [2] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-169-207:675553:675674 [2] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-169-207:675553:675674 [2] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-169-207:675553:675674 [2] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-169-207:675553:675674 [2] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-169-207:675553:675674 [2] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-169-207:675553:675674 [2] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-169-207:675553:675674 [2] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-169-207:675553:675674 [2] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-169-207:675553:675674 [2] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/149
+ip-26-0-169-207:675553:675674 [2] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-169-207:675554:675680 [3] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-169-207:675554:675680 [3] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-169-207:675554:675680 [3] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-169-207:675554:675680 [3] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-169-207:675554:675680 [3] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-169-207:675554:675680 [3] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-169-207:675554:675680 [3] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-169-207:675554:675680 [3] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-169-207:675554:675680 [3] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-169-207:675554:675680 [3] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-169-207:675554:675680 [3] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-169-207:675554:675680 [3] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-169-207:675554:675680 [3] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-169-207:675554:675680 [3] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-169-207:675554:675680 [3] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-169-207:675554:675680 [3] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-169-207:675554:675680 [3] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-169-207:675554:675680 [3] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-169-207:675554:675680 [3] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-169-207:675554:675680 [3] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-169-207:675554:675680 [3] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-169-207:675554:675680 [3] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-169-207:675554:675680 [3] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-169-207:675554:675680 [3] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-169-207:675554:675680 [3] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-169-207:675554:675680 [3] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-169-207:675554:675680 [3] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-169-207:675554:675680 [3] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-169-207:675554:675680 [3] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-169-207:675554:675680 [3] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-169-207:675554:675680 [3] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-169-207:675554:675680 [3] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-169-207:675554:675680 [3] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/149
+ip-26-0-169-207:675554:675680 [3] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-169-207:675552:675677 [1] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-169-207:675552:675677 [1] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-169-207:675552:675677 [1] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-169-207:675552:675677 [1] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-169-207:675552:675677 [1] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-169-207:675552:675677 [1] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-169-207:675552:675677 [1] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-169-207:675552:675677 [1] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-169-207:675552:675677 [1] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-169-207:675552:675677 [1] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-169-207:675552:675677 [1] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-169-207:675552:675677 [1] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-169-207:675552:675677 [1] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-169-207:675552:675677 [1] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-169-207:675552:675677 [1] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-169-207:675552:675677 [1] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-169-207:675552:675677 [1] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-169-207:675552:675677 [1] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-169-207:675552:675677 [1] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-169-207:675552:675677 [1] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-169-207:675552:675677 [1] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-169-207:675552:675677 [1] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-169-207:675552:675677 [1] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-169-207:675552:675677 [1] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-169-207:675552:675677 [1] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-169-207:675552:675677 [1] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-169-207:675552:675677 [1] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-169-207:675552:675677 [1] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-169-207:675552:675677 [1] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-169-207:675552:675677 [1] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-169-207:675552:675677 [1] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-169-207:675552:675677 [1] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-169-207:675552:675677 [1] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/149
+ip-26-0-169-207:675552:675677 [1] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-169-207:675555:675675 [4] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-169-207:675555:675675 [4] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-169-207:675555:675675 [4] NCCL INFO Using network Libfabric
+ip-26-0-169-207:675556:675679 [5] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-169-207:675556:675679 [5] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-169-207:675556:675679 [5] NCCL INFO Using network Libfabric
+ip-26-0-169-207:675557:675678 [6] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-169-207:675557:675678 [6] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-169-207:675557:675678 [6] NCCL INFO Using network Libfabric
+ip-26-0-169-207:675558:675676 [7] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-169-207:675558:675676 [7] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-169-207:675558:675676 [7] NCCL INFO Using network Libfabric
+ip-26-0-169-207:675554:675680 [3] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-169-207:675554:675680 [3] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-169-207:675554:675680 [3] NCCL INFO Using network Libfabric
+ip-26-0-169-207:675552:675677 [1] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-169-207:675552:675677 [1] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-169-207:675552:675677 [1] NCCL INFO Using network Libfabric
+ip-26-0-169-207:675553:675674 [2] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-169-207:675553:675674 [2] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-169-207:675553:675674 [2] NCCL INFO Using network Libfabric
+ip-26-0-169-207:675554:675680 [3] NCCL INFO comm 0xa3c4230 rank 3 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0x2fd95b402c5b9b62 - Init START
+ip-26-0-169-207:675555:675675 [4] NCCL INFO comm 0x931d470 rank 4 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0x2fd95b402c5b9b62 - Init START
+ip-26-0-169-207:675558:675676 [7] NCCL INFO comm 0x95795e0 rank 7 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0x2fd95b402c5b9b62 - Init START
+ip-26-0-169-207:675557:675678 [6] NCCL INFO comm 0x9ba1fd0 rank 6 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0x2fd95b402c5b9b62 - Init START
+ip-26-0-169-207:675552:675677 [1] NCCL INFO comm 0x9423860 rank 1 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0x2fd95b402c5b9b62 - Init START
+ip-26-0-169-207:675553:675674 [2] NCCL INFO comm 0x856b2e0 rank 2 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0x2fd95b402c5b9b62 - Init START
+ip-26-0-169-207:675556:675679 [5] NCCL INFO comm 0xa17ac90 rank 5 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0x2fd95b402c5b9b62 - Init START
+ip-26-0-169-207:675551:675673 [0] NCCL INFO comm 0x8d1d5d0 rank 0 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0x2fd95b402c5b9b62 - Init START
+ip-26-0-169-207:675552:675677 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/149
+ip-26-0-169-207:675554:675680 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/149
+ip-26-0-169-207:675551:675673 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/162
+ip-26-0-169-207:675553:675674 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/149
+ip-26-0-169-207:675555:675675 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/149
+ip-26-0-169-207:675557:675678 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/149
+ip-26-0-169-207:675558:675676 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/149
+ip-26-0-169-207:675556:675679 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/149
+ip-26-0-169-207:675551:675673 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-169-207:675551:675673 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-169-207:675558:675676 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-169-207:675558:675676 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-169-207:675554:675680 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-169-207:675554:675680 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-169-207:675552:675677 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-169-207:675552:675677 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-169-207:675557:675678 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-169-207:675557:675678 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-169-207:675556:675679 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-169-207:675556:675679 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-169-207:675553:675674 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-169-207:675555:675675 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-169-207:675555:675675 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-169-207:675553:675674 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-169-207:675553:675674 [2] NCCL INFO Trees [0] 3/-1/-1->2->1 [1] 3/-1/-1->2->1 [2] 3/-1/-1->2->1 [3] 3/-1/-1->2->1 [4] 3/-1/-1->2->1 [5] 3/-1/-1->2->1 [6] 3/-1/-1->2->1 [7] 3/-1/-1->2->1 [8] 3/-1/-1->2->1 [9] 3/-1/-1->2->1 [10] 3/-1/-1->2->1 [11] 3/-1/-1->2->1 [12] 3/-1/-1->2->1 [13] 3/-1/-1->2->1 [14] 3/-1/-1->2->1 [15] 3/-1/-1->2->1 [16] 3/-1/-1->2->1 [17] 3/-1/-1->2->1 [18] 3/-1/-1->2->1 [19] 3/-1/-1->2->1 [20] 3/-1/-1->2->1 [21] 3/-1/-1->2->1 [22] 3/-1/-1->2->1 [23] 3/-1/-1->2->1
+ip-26-0-169-207:675553:675674 [2] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-169-207:675552:675677 [1] NCCL INFO Trees [0] 2/-1/-1->1->0 [1] 2/-1/-1->1->0 [2] 2/-1/-1->1->0 [3] 2/-1/-1->1->0 [4] 2/-1/-1->1->0 [5] 2/-1/-1->1->0 [6] 2/-1/-1->1->0 [7] 2/-1/-1->1->0 [8] 2/-1/-1->1->0 [9] 2/-1/-1->1->0 [10] 2/-1/-1->1->0 [11] 2/-1/-1->1->0 [12] 2/-1/-1->1->0 [13] 2/-1/-1->1->0 [14] 2/-1/-1->1->0 [15] 2/-1/-1->1->0 [16] 2/-1/-1->1->0 [17] 2/-1/-1->1->0 [18] 2/-1/-1->1->0 [19] 2/-1/-1->1->0 [20] 2/-1/-1->1->0 [21] 2/-1/-1->1->0 [22] 2/-1/-1->1->0 [23] 2/-1/-1->1->0
+ip-26-0-169-207:675552:675677 [1] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-169-207:675558:675676 [7] NCCL INFO Trees [0] -1/-1/-1->7->6 [1] -1/-1/-1->7->6 [2] -1/-1/-1->7->6 [3] -1/-1/-1->7->6 [4] -1/-1/-1->7->6 [5] -1/-1/-1->7->6 [6] -1/-1/-1->7->6 [7] -1/-1/-1->7->6 [8] -1/-1/-1->7->6 [9] -1/-1/-1->7->6 [10] -1/-1/-1->7->6 [11] -1/-1/-1->7->6 [12] -1/-1/-1->7->6 [13] -1/-1/-1->7->6 [14] -1/-1/-1->7->6 [15] -1/-1/-1->7->6 [16] -1/-1/-1->7->6 [17] -1/-1/-1->7->6 [18] -1/-1/-1->7->6 [19] -1/-1/-1->7->6 [20] -1/-1/-1->7->6 [21] -1/-1/-1->7->6 [22] -1/-1/-1->7->6 [23] -1/-1/-1->7->6
+ip-26-0-169-207:675555:675675 [4] NCCL INFO Trees [0] 5/-1/-1->4->3 [1] 5/-1/-1->4->3 [2] 5/-1/-1->4->3 [3] 5/-1/-1->4->3 [4] 5/-1/-1->4->3 [5] 5/-1/-1->4->3 [6] 5/-1/-1->4->3 [7] 5/-1/-1->4->3 [8] 5/-1/-1->4->3 [9] 5/-1/-1->4->3 [10] 5/-1/-1->4->3 [11] 5/-1/-1->4->3 [12] 5/-1/-1->4->3 [13] 5/-1/-1->4->3 [14] 5/-1/-1->4->3 [15] 5/-1/-1->4->3 [16] 5/-1/-1->4->3 [17] 5/-1/-1->4->3 [18] 5/-1/-1->4->3 [19] 5/-1/-1->4->3 [20] 5/-1/-1->4->3 [21] 5/-1/-1->4->3 [22] 5/-1/-1->4->3 [23] 5/-1/-1->4->3
+ip-26-0-169-207:675557:675678 [6] NCCL INFO Trees [0] 7/-1/-1->6->5 [1] 7/-1/-1->6->5 [2] 7/-1/-1->6->5 [3] 7/-1/-1->6->5 [4] 7/-1/-1->6->5 [5] 7/-1/-1->6->5 [6] 7/-1/-1->6->5 [7] 7/-1/-1->6->5 [8] 7/-1/-1->6->5 [9] 7/-1/-1->6->5 [10] 7/-1/-1->6->5 [11] 7/-1/-1->6->5 [12] 7/-1/-1->6->5 [13] 7/-1/-1->6->5 [14] 7/-1/-1->6->5 [15] 7/-1/-1->6->5 [16] 7/-1/-1->6->5 [17] 7/-1/-1->6->5 [18] 7/-1/-1->6->5 [19] 7/-1/-1->6->5 [20] 7/-1/-1->6->5 [21] 7/-1/-1->6->5 [22] 7/-1/-1->6->5 [23] 7/-1/-1->6->5
+ip-26-0-169-207:675558:675676 [7] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-169-207:675555:675675 [4] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-169-207:675557:675678 [6] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-169-207:675554:675680 [3] NCCL INFO Trees [0] 4/-1/-1->3->2 [1] 4/-1/-1->3->2 [2] 4/-1/-1->3->2 [3] 4/-1/-1->3->2 [4] 4/-1/-1->3->2 [5] 4/-1/-1->3->2 [6] 4/-1/-1->3->2 [7] 4/-1/-1->3->2 [8] 4/-1/-1->3->2 [9] 4/-1/-1->3->2 [10] 4/-1/-1->3->2 [11] 4/-1/-1->3->2 [12] 4/-1/-1->3->2 [13] 4/-1/-1->3->2 [14] 4/-1/-1->3->2 [15] 4/-1/-1->3->2 [16] 4/-1/-1->3->2 [17] 4/-1/-1->3->2 [18] 4/-1/-1->3->2 [19] 4/-1/-1->3->2 [20] 4/-1/-1->3->2 [21] 4/-1/-1->3->2 [22] 4/-1/-1->3->2 [23] 4/-1/-1->3->2
+ip-26-0-169-207:675551:675673 [0] NCCL INFO Channel 00/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:675556:675679 [5] NCCL INFO Trees [0] 6/-1/-1->5->4 [1] 6/-1/-1->5->4 [2] 6/-1/-1->5->4 [3] 6/-1/-1->5->4 [4] 6/-1/-1->5->4 [5] 6/-1/-1->5->4 [6] 6/-1/-1->5->4 [7] 6/-1/-1->5->4 [8] 6/-1/-1->5->4 [9] 6/-1/-1->5->4 [10] 6/-1/-1->5->4 [11] 6/-1/-1->5->4 [12] 6/-1/-1->5->4 [13] 6/-1/-1->5->4 [14] 6/-1/-1->5->4 [15] 6/-1/-1->5->4 [16] 6/-1/-1->5->4 [17] 6/-1/-1->5->4 [18] 6/-1/-1->5->4 [19] 6/-1/-1->5->4 [20] 6/-1/-1->5->4 [21] 6/-1/-1->5->4 [22] 6/-1/-1->5->4 [23] 6/-1/-1->5->4
+ip-26-0-169-207:675551:675673 [0] NCCL INFO Channel 01/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:675556:675679 [5] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-169-207:675551:675673 [0] NCCL INFO Channel 02/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:675551:675673 [0] NCCL INFO Channel 03/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:675551:675673 [0] NCCL INFO Channel 04/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:675551:675673 [0] NCCL INFO Channel 05/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:675551:675673 [0] NCCL INFO Channel 06/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:675551:675673 [0] NCCL INFO Channel 07/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:675551:675673 [0] NCCL INFO Channel 08/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:675551:675673 [0] NCCL INFO Channel 09/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:675551:675673 [0] NCCL INFO Channel 10/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:675551:675673 [0] NCCL INFO Channel 11/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:675551:675673 [0] NCCL INFO Channel 12/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:675551:675673 [0] NCCL INFO Channel 13/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:675551:675673 [0] NCCL INFO Channel 14/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:675551:675673 [0] NCCL INFO Channel 15/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:675554:675680 [3] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-169-207:675551:675673 [0] NCCL INFO Channel 16/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:675551:675673 [0] NCCL INFO Channel 17/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:675551:675673 [0] NCCL INFO Channel 18/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:675551:675673 [0] NCCL INFO Channel 19/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:675551:675673 [0] NCCL INFO Channel 20/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:675551:675673 [0] NCCL INFO Channel 21/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:675551:675673 [0] NCCL INFO Channel 22/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:675551:675673 [0] NCCL INFO Channel 23/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:675551:675673 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] 1/-1/-1->0->-1 [7] 1/-1/-1->0->-1 [8] 1/-1/-1->0->-1 [9] 1/-1/-1->0->-1 [10] 1/-1/-1->0->-1 [11] 1/-1/-1->0->-1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] 1/-1/-1->0->-1 [19] 1/-1/-1->0->-1 [20] 1/-1/-1->0->-1 [21] 1/-1/-1->0->-1 [22] 1/-1/-1->0->-1 [23] 1/-1/-1->0->-1
+ip-26-0-169-207:675551:675673 [0] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-169-207:675552:675677 [1] NCCL INFO Channel 00/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675555:675675 [4] NCCL INFO Channel 00/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675554:675680 [3] NCCL INFO Channel 00/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675552:675677 [1] NCCL INFO Channel 01/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675551:675673 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675555:675675 [4] NCCL INFO Channel 01/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675553:675674 [2] NCCL INFO Channel 00/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675554:675680 [3] NCCL INFO Channel 01/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675552:675677 [1] NCCL INFO Channel 02/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675551:675673 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675555:675675 [4] NCCL INFO Channel 02/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675553:675674 [2] NCCL INFO Channel 01/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675554:675680 [3] NCCL INFO Channel 02/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675551:675673 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675553:675674 [2] NCCL INFO Channel 02/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675555:675675 [4] NCCL INFO Channel 03/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675552:675677 [1] NCCL INFO Channel 03/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675554:675680 [3] NCCL INFO Channel 03/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675551:675673 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675555:675675 [4] NCCL INFO Channel 04/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675553:675674 [2] NCCL INFO Channel 03/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675552:675677 [1] NCCL INFO Channel 04/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675551:675673 [0] NCCL INFO Channel 04/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675554:675680 [3] NCCL INFO Channel 04/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675555:675675 [4] NCCL INFO Channel 05/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675553:675674 [2] NCCL INFO Channel 04/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675552:675677 [1] NCCL INFO Channel 05/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675551:675673 [0] NCCL INFO Channel 05/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675554:675680 [3] NCCL INFO Channel 05/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675555:675675 [4] NCCL INFO Channel 06/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675553:675674 [2] NCCL INFO Channel 05/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675551:675673 [0] NCCL INFO Channel 06/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675556:675679 [5] NCCL INFO Channel 00/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675553:675674 [2] NCCL INFO Channel 06/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675551:675673 [0] NCCL INFO Channel 07/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675555:675675 [4] NCCL INFO Channel 07/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675553:675674 [2] NCCL INFO Channel 07/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675556:675679 [5] NCCL INFO Channel 01/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675551:675673 [0] NCCL INFO Channel 08/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675554:675680 [3] NCCL INFO Channel 06/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675555:675675 [4] NCCL INFO Channel 08/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675553:675674 [2] NCCL INFO Channel 08/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675556:675679 [5] NCCL INFO Channel 02/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675551:675673 [0] NCCL INFO Channel 09/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675554:675680 [3] NCCL INFO Channel 07/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675555:675675 [4] NCCL INFO Channel 09/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675553:675674 [2] NCCL INFO Channel 09/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675551:675673 [0] NCCL INFO Channel 10/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675556:675679 [5] NCCL INFO Channel 03/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675554:675680 [3] NCCL INFO Channel 08/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675555:675675 [4] NCCL INFO Channel 10/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675553:675674 [2] NCCL INFO Channel 10/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675551:675673 [0] NCCL INFO Channel 11/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675554:675680 [3] NCCL INFO Channel 09/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675556:675679 [5] NCCL INFO Channel 04/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675555:675675 [4] NCCL INFO Channel 11/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675553:675674 [2] NCCL INFO Channel 11/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675552:675677 [1] NCCL INFO Channel 06/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675554:675680 [3] NCCL INFO Channel 10/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675555:675675 [4] NCCL INFO Channel 12/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675553:675674 [2] NCCL INFO Channel 12/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675558:675676 [7] NCCL INFO Channel 00/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675557:675678 [6] NCCL INFO Channel 00/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:675551:675673 [0] NCCL INFO Channel 12/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675552:675677 [1] NCCL INFO Channel 07/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675554:675680 [3] NCCL INFO Channel 11/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675555:675675 [4] NCCL INFO Channel 13/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675553:675674 [2] NCCL INFO Channel 13/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675558:675676 [7] NCCL INFO Channel 01/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675557:675678 [6] NCCL INFO Channel 01/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:675551:675673 [0] NCCL INFO Channel 13/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675552:675677 [1] NCCL INFO Channel 08/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675554:675680 [3] NCCL INFO Channel 12/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675555:675675 [4] NCCL INFO Channel 14/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675553:675674 [2] NCCL INFO Channel 14/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675558:675676 [7] NCCL INFO Channel 02/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675557:675678 [6] NCCL INFO Channel 02/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:675552:675677 [1] NCCL INFO Channel 09/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675551:675673 [0] NCCL INFO Channel 14/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675554:675680 [3] NCCL INFO Channel 13/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675555:675675 [4] NCCL INFO Channel 15/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675553:675674 [2] NCCL INFO Channel 15/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675558:675676 [7] NCCL INFO Channel 03/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675557:675678 [6] NCCL INFO Channel 03/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:675552:675677 [1] NCCL INFO Channel 10/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675551:675673 [0] NCCL INFO Channel 15/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675554:675680 [3] NCCL INFO Channel 14/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675555:675675 [4] NCCL INFO Channel 16/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675553:675674 [2] NCCL INFO Channel 16/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675558:675676 [7] NCCL INFO Channel 04/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675557:675678 [6] NCCL INFO Channel 04/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:675552:675677 [1] NCCL INFO Channel 11/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675551:675673 [0] NCCL INFO Channel 16/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675554:675680 [3] NCCL INFO Channel 15/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675555:675675 [4] NCCL INFO Channel 17/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675553:675674 [2] NCCL INFO Channel 17/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675558:675676 [7] NCCL INFO Channel 05/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675557:675678 [6] NCCL INFO Channel 05/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:675551:675673 [0] NCCL INFO Channel 17/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675556:675679 [5] NCCL INFO Channel 05/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675554:675680 [3] NCCL INFO Channel 16/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675555:675675 [4] NCCL INFO Channel 18/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675553:675674 [2] NCCL INFO Channel 18/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675552:675677 [1] NCCL INFO Channel 12/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675551:675673 [0] NCCL INFO Channel 18/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675558:675676 [7] NCCL INFO Channel 06/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675557:675678 [6] NCCL INFO Channel 06/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:675552:675677 [1] NCCL INFO Channel 13/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675551:675673 [0] NCCL INFO Channel 19/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675558:675676 [7] NCCL INFO Channel 07/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675553:675674 [2] NCCL INFO Channel 19/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675557:675678 [6] NCCL INFO Channel 07/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:675552:675677 [1] NCCL INFO Channel 14/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675551:675673 [0] NCCL INFO Channel 20/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675558:675676 [7] NCCL INFO Channel 08/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675553:675674 [2] NCCL INFO Channel 20/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675557:675678 [6] NCCL INFO Channel 08/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:675552:675677 [1] NCCL INFO Channel 15/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675551:675673 [0] NCCL INFO Channel 21/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675558:675676 [7] NCCL INFO Channel 09/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675553:675674 [2] NCCL INFO Channel 21/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675557:675678 [6] NCCL INFO Channel 09/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:675552:675677 [1] NCCL INFO Channel 16/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675551:675673 [0] NCCL INFO Channel 22/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675558:675676 [7] NCCL INFO Channel 10/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675553:675674 [2] NCCL INFO Channel 22/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675557:675678 [6] NCCL INFO Channel 10/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:675552:675677 [1] NCCL INFO Channel 17/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675551:675673 [0] NCCL INFO Channel 23/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675558:675676 [7] NCCL INFO Channel 11/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675556:675679 [5] NCCL INFO Channel 06/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675553:675674 [2] NCCL INFO Channel 23/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675554:675680 [3] NCCL INFO Channel 17/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675557:675678 [6] NCCL INFO Channel 11/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:675552:675677 [1] NCCL INFO Channel 18/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675555:675675 [4] NCCL INFO Channel 19/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675558:675676 [7] NCCL INFO Channel 12/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675556:675679 [5] NCCL INFO Channel 07/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675554:675680 [3] NCCL INFO Channel 18/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675557:675678 [6] NCCL INFO Channel 12/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:675552:675677 [1] NCCL INFO Channel 19/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675555:675675 [4] NCCL INFO Channel 20/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675558:675676 [7] NCCL INFO Channel 13/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675556:675679 [5] NCCL INFO Channel 08/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675554:675680 [3] NCCL INFO Channel 19/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675557:675678 [6] NCCL INFO Channel 13/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:675552:675677 [1] NCCL INFO Channel 20/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675555:675675 [4] NCCL INFO Channel 21/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675558:675676 [7] NCCL INFO Channel 14/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675556:675679 [5] NCCL INFO Channel 09/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675554:675680 [3] NCCL INFO Channel 20/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675557:675678 [6] NCCL INFO Channel 14/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:675552:675677 [1] NCCL INFO Channel 21/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675555:675675 [4] NCCL INFO Channel 22/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675558:675676 [7] NCCL INFO Channel 15/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675556:675679 [5] NCCL INFO Channel 10/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675554:675680 [3] NCCL INFO Channel 21/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675557:675678 [6] NCCL INFO Channel 15/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:675552:675677 [1] NCCL INFO Channel 22/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675555:675675 [4] NCCL INFO Channel 23/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675558:675676 [7] NCCL INFO Channel 16/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675556:675679 [5] NCCL INFO Channel 11/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675554:675680 [3] NCCL INFO Channel 22/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675557:675678 [6] NCCL INFO Channel 16/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:675552:675677 [1] NCCL INFO Channel 23/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675558:675676 [7] NCCL INFO Channel 17/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675556:675679 [5] NCCL INFO Channel 12/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675554:675680 [3] NCCL INFO Channel 23/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675557:675678 [6] NCCL INFO Channel 17/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:675558:675676 [7] NCCL INFO Channel 18/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675556:675679 [5] NCCL INFO Channel 13/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675557:675678 [6] NCCL INFO Channel 18/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:675558:675676 [7] NCCL INFO Channel 19/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675556:675679 [5] NCCL INFO Channel 14/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675557:675678 [6] NCCL INFO Channel 19/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:675558:675676 [7] NCCL INFO Channel 20/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675556:675679 [5] NCCL INFO Channel 15/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675557:675678 [6] NCCL INFO Channel 20/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:675558:675676 [7] NCCL INFO Channel 21/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675556:675679 [5] NCCL INFO Channel 16/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675557:675678 [6] NCCL INFO Channel 21/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:675558:675676 [7] NCCL INFO Channel 22/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675556:675679 [5] NCCL INFO Channel 17/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675558:675676 [7] NCCL INFO Channel 23/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675557:675678 [6] NCCL INFO Channel 22/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:675556:675679 [5] NCCL INFO Channel 18/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675557:675678 [6] NCCL INFO Channel 23/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:675556:675679 [5] NCCL INFO Channel 19/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675556:675679 [5] NCCL INFO Channel 20/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675556:675679 [5] NCCL INFO Channel 21/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675556:675679 [5] NCCL INFO Channel 22/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675556:675679 [5] NCCL INFO Channel 23/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675552:675677 [1] NCCL INFO Connected all rings
+ip-26-0-169-207:675553:675674 [2] NCCL INFO Connected all rings
+ip-26-0-169-207:675554:675680 [3] NCCL INFO Connected all rings
+ip-26-0-169-207:675551:675673 [0] NCCL INFO Connected all rings
+ip-26-0-169-207:675558:675676 [7] NCCL INFO Connected all rings
+ip-26-0-169-207:675558:675676 [7] NCCL INFO Channel 00/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675558:675676 [7] NCCL INFO Channel 01/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675558:675676 [7] NCCL INFO Channel 02/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675558:675676 [7] NCCL INFO Channel 03/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675558:675676 [7] NCCL INFO Channel 04/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675558:675676 [7] NCCL INFO Channel 05/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675558:675676 [7] NCCL INFO Channel 06/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675558:675676 [7] NCCL INFO Channel 07/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675558:675676 [7] NCCL INFO Channel 08/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675558:675676 [7] NCCL INFO Channel 09/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675558:675676 [7] NCCL INFO Channel 10/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675558:675676 [7] NCCL INFO Channel 11/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675555:675675 [4] NCCL INFO Connected all rings
+ip-26-0-169-207:675558:675676 [7] NCCL INFO Channel 12/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675558:675676 [7] NCCL INFO Channel 13/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675558:675676 [7] NCCL INFO Channel 14/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675556:675679 [5] NCCL INFO Connected all rings
+ip-26-0-169-207:675557:675678 [6] NCCL INFO Connected all rings
+ip-26-0-169-207:675558:675676 [7] NCCL INFO Channel 15/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675558:675676 [7] NCCL INFO Channel 16/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675558:675676 [7] NCCL INFO Channel 17/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675558:675676 [7] NCCL INFO Channel 18/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675558:675676 [7] NCCL INFO Channel 19/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675558:675676 [7] NCCL INFO Channel 20/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675558:675676 [7] NCCL INFO Channel 21/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675558:675676 [7] NCCL INFO Channel 22/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675558:675676 [7] NCCL INFO Channel 23/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675553:675674 [2] NCCL INFO Channel 00/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675553:675674 [2] NCCL INFO Channel 01/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675553:675674 [2] NCCL INFO Channel 02/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675552:675677 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675553:675674 [2] NCCL INFO Channel 03/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675552:675677 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675553:675674 [2] NCCL INFO Channel 04/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675552:675677 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675554:675680 [3] NCCL INFO Channel 00/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675553:675674 [2] NCCL INFO Channel 05/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675552:675677 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675554:675680 [3] NCCL INFO Channel 01/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675553:675674 [2] NCCL INFO Channel 06/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675552:675677 [1] NCCL INFO Channel 04/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675554:675680 [3] NCCL INFO Channel 02/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675553:675674 [2] NCCL INFO Channel 07/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675552:675677 [1] NCCL INFO Channel 05/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675554:675680 [3] NCCL INFO Channel 03/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675553:675674 [2] NCCL INFO Channel 08/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675552:675677 [1] NCCL INFO Channel 06/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675554:675680 [3] NCCL INFO Channel 04/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675553:675674 [2] NCCL INFO Channel 09/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675552:675677 [1] NCCL INFO Channel 07/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675555:675675 [4] NCCL INFO Channel 00/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675554:675680 [3] NCCL INFO Channel 05/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675553:675674 [2] NCCL INFO Channel 10/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675552:675677 [1] NCCL INFO Channel 08/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675555:675675 [4] NCCL INFO Channel 01/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675554:675680 [3] NCCL INFO Channel 06/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675553:675674 [2] NCCL INFO Channel 11/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675552:675677 [1] NCCL INFO Channel 09/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675555:675675 [4] NCCL INFO Channel 02/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675554:675680 [3] NCCL INFO Channel 07/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675553:675674 [2] NCCL INFO Channel 12/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675552:675677 [1] NCCL INFO Channel 10/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675555:675675 [4] NCCL INFO Channel 03/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675554:675680 [3] NCCL INFO Channel 08/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675553:675674 [2] NCCL INFO Channel 13/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675556:675679 [5] NCCL INFO Channel 00/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675552:675677 [1] NCCL INFO Channel 11/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675555:675675 [4] NCCL INFO Channel 04/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675554:675680 [3] NCCL INFO Channel 09/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675553:675674 [2] NCCL INFO Channel 14/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675556:675679 [5] NCCL INFO Channel 01/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675552:675677 [1] NCCL INFO Channel 12/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675555:675675 [4] NCCL INFO Channel 05/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675554:675680 [3] NCCL INFO Channel 10/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675553:675674 [2] NCCL INFO Channel 15/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675556:675679 [5] NCCL INFO Channel 02/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675552:675677 [1] NCCL INFO Channel 13/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675555:675675 [4] NCCL INFO Channel 06/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675557:675678 [6] NCCL INFO Channel 00/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675554:675680 [3] NCCL INFO Channel 11/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675553:675674 [2] NCCL INFO Channel 16/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675556:675679 [5] NCCL INFO Channel 03/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675552:675677 [1] NCCL INFO Channel 14/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675555:675675 [4] NCCL INFO Channel 07/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675557:675678 [6] NCCL INFO Channel 01/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675554:675680 [3] NCCL INFO Channel 12/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675553:675674 [2] NCCL INFO Channel 17/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675552:675677 [1] NCCL INFO Channel 15/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675556:675679 [5] NCCL INFO Channel 04/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675555:675675 [4] NCCL INFO Channel 08/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675557:675678 [6] NCCL INFO Channel 02/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675554:675680 [3] NCCL INFO Channel 13/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675553:675674 [2] NCCL INFO Channel 18/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675552:675677 [1] NCCL INFO Channel 16/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675556:675679 [5] NCCL INFO Channel 05/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675555:675675 [4] NCCL INFO Channel 09/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675557:675678 [6] NCCL INFO Channel 03/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675553:675674 [2] NCCL INFO Channel 19/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675554:675680 [3] NCCL INFO Channel 14/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675552:675677 [1] NCCL INFO Channel 17/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675556:675679 [5] NCCL INFO Channel 06/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675555:675675 [4] NCCL INFO Channel 10/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675557:675678 [6] NCCL INFO Channel 04/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675553:675674 [2] NCCL INFO Channel 20/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675554:675680 [3] NCCL INFO Channel 15/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675552:675677 [1] NCCL INFO Channel 18/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675556:675679 [5] NCCL INFO Channel 07/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675555:675675 [4] NCCL INFO Channel 11/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675557:675678 [6] NCCL INFO Channel 05/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675553:675674 [2] NCCL INFO Channel 21/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675554:675680 [3] NCCL INFO Channel 16/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675552:675677 [1] NCCL INFO Channel 19/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675556:675679 [5] NCCL INFO Channel 08/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675555:675675 [4] NCCL INFO Channel 12/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675557:675678 [6] NCCL INFO Channel 06/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675553:675674 [2] NCCL INFO Channel 22/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675554:675680 [3] NCCL INFO Channel 17/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675552:675677 [1] NCCL INFO Channel 20/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675556:675679 [5] NCCL INFO Channel 09/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675555:675675 [4] NCCL INFO Channel 13/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675557:675678 [6] NCCL INFO Channel 07/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675553:675674 [2] NCCL INFO Channel 23/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675554:675680 [3] NCCL INFO Channel 18/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675552:675677 [1] NCCL INFO Channel 21/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675556:675679 [5] NCCL INFO Channel 10/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675555:675675 [4] NCCL INFO Channel 14/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675557:675678 [6] NCCL INFO Channel 08/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675556:675679 [5] NCCL INFO Channel 11/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675555:675675 [4] NCCL INFO Channel 15/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675552:675677 [1] NCCL INFO Channel 22/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675557:675678 [6] NCCL INFO Channel 09/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675556:675679 [5] NCCL INFO Channel 12/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675555:675675 [4] NCCL INFO Channel 16/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675552:675677 [1] NCCL INFO Channel 23/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675557:675678 [6] NCCL INFO Channel 10/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675554:675680 [3] NCCL INFO Channel 19/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675556:675679 [5] NCCL INFO Channel 13/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675555:675675 [4] NCCL INFO Channel 17/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675557:675678 [6] NCCL INFO Channel 11/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675554:675680 [3] NCCL INFO Channel 20/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675556:675679 [5] NCCL INFO Channel 14/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675555:675675 [4] NCCL INFO Channel 18/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675557:675678 [6] NCCL INFO Channel 12/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675557:675678 [6] NCCL INFO Channel 13/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675557:675678 [6] NCCL INFO Channel 14/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675557:675678 [6] NCCL INFO Channel 15/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675557:675678 [6] NCCL INFO Channel 16/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675557:675678 [6] NCCL INFO Channel 17/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675557:675678 [6] NCCL INFO Channel 18/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675557:675678 [6] NCCL INFO Channel 19/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675557:675678 [6] NCCL INFO Channel 20/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675557:675678 [6] NCCL INFO Channel 21/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675557:675678 [6] NCCL INFO Channel 22/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675557:675678 [6] NCCL INFO Channel 23/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675554:675680 [3] NCCL INFO Channel 21/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675556:675679 [5] NCCL INFO Channel 15/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675555:675675 [4] NCCL INFO Channel 19/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675554:675680 [3] NCCL INFO Channel 22/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675556:675679 [5] NCCL INFO Channel 16/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675556:675679 [5] NCCL INFO Channel 17/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675555:675675 [4] NCCL INFO Channel 20/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675554:675680 [3] NCCL INFO Channel 23/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675556:675679 [5] NCCL INFO Channel 18/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675556:675679 [5] NCCL INFO Channel 19/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675556:675679 [5] NCCL INFO Channel 20/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675555:675675 [4] NCCL INFO Channel 21/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675556:675679 [5] NCCL INFO Channel 21/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675555:675675 [4] NCCL INFO Channel 22/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675556:675679 [5] NCCL INFO Channel 22/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675556:675679 [5] NCCL INFO Channel 23/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675555:675675 [4] NCCL INFO Channel 23/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675551:675673 [0] NCCL INFO Connected all trees
+ip-26-0-169-207:675551:675673 [0] NCCL INFO NVLS comm 0x8d1d5d0 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-169-207:675558:675676 [7] NCCL INFO Connected all trees
+ip-26-0-169-207:675558:675676 [7] NCCL INFO NVLS comm 0x95795e0 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-169-207:675553:675674 [2] NCCL INFO Connected all trees
+ip-26-0-169-207:675553:675674 [2] NCCL INFO NVLS comm 0x856b2e0 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-169-207:675552:675677 [1] NCCL INFO Connected all trees
+ip-26-0-169-207:675552:675677 [1] NCCL INFO NVLS comm 0x9423860 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-169-207:675556:675679 [5] NCCL INFO Connected all trees
+ip-26-0-169-207:675556:675679 [5] NCCL INFO NVLS comm 0xa17ac90 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-169-207:675557:675678 [6] NCCL INFO Connected all trees
+ip-26-0-169-207:675557:675678 [6] NCCL INFO NVLS comm 0x9ba1fd0 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-169-207:675555:675675 [4] NCCL INFO Connected all trees
+ip-26-0-169-207:675555:675675 [4] NCCL INFO NVLS comm 0x931d470 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-169-207:675554:675680 [3] NCCL INFO Connected all trees
+ip-26-0-169-207:675554:675680 [3] NCCL INFO NVLS comm 0xa3c4230 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-169-207:675557:675678 [6] NCCL INFO Connected NVLS tree
+ip-26-0-169-207:675557:675678 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-169-207:675557:675678 [6] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-169-207:675555:675675 [4] NCCL INFO Connected NVLS tree
+ip-26-0-169-207:675557:675678 [6] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:675555:675675 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-169-207:675555:675675 [4] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-169-207:675555:675675 [4] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:675551:675673 [0] NCCL INFO Connected NVLS tree
+ip-26-0-169-207:675552:675677 [1] NCCL INFO Connected NVLS tree
+ip-26-0-169-207:675552:675677 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-169-207:675551:675673 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-169-207:675552:675677 [1] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-169-207:675552:675677 [1] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:675553:675674 [2] NCCL INFO Connected NVLS tree
+ip-26-0-169-207:675551:675673 [0] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-169-207:675551:675673 [0] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:675556:675679 [5] NCCL INFO Connected NVLS tree
+ip-26-0-169-207:675553:675674 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-169-207:675556:675679 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-169-207:675556:675679 [5] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-169-207:675556:675679 [5] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:675553:675674 [2] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-169-207:675553:675674 [2] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:675558:675676 [7] NCCL INFO Connected NVLS tree
+ip-26-0-169-207:675558:675676 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-169-207:675558:675676 [7] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-169-207:675558:675676 [7] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:675554:675680 [3] NCCL INFO Connected NVLS tree
+ip-26-0-169-207:675554:675680 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-169-207:675554:675680 [3] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-169-207:675554:675680 [3] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:675552:675677 [1] NCCL INFO comm 0x9423860 rank 1 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0x2fd95b402c5b9b62 - Init COMPLETE
+ip-26-0-169-207:675558:675676 [7] NCCL INFO comm 0x95795e0 rank 7 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0x2fd95b402c5b9b62 - Init COMPLETE
+ip-26-0-169-207:675556:675679 [5] NCCL INFO comm 0xa17ac90 rank 5 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0x2fd95b402c5b9b62 - Init COMPLETE
+ip-26-0-169-207:675554:675680 [3] NCCL INFO comm 0xa3c4230 rank 3 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0x2fd95b402c5b9b62 - Init COMPLETE
+ip-26-0-169-207:675555:675675 [4] NCCL INFO comm 0x931d470 rank 4 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0x2fd95b402c5b9b62 - Init COMPLETE
+ip-26-0-169-207:675557:675678 [6] NCCL INFO comm 0x9ba1fd0 rank 6 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0x2fd95b402c5b9b62 - Init COMPLETE
+ip-26-0-169-207:675553:675674 [2] NCCL INFO comm 0x856b2e0 rank 2 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0x2fd95b402c5b9b62 - Init COMPLETE
+ip-26-0-169-207:675551:675673 [0] NCCL INFO comm 0x8d1d5d0 rank 0 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0x2fd95b402c5b9b62 - Init COMPLETE
+ip-26-0-169-207:675551:675745 [0] NCCL INFO Using network Libfabric
+ip-26-0-169-207:675552:675747 [1] NCCL INFO Using network Libfabric
+ip-26-0-169-207:675556:675746 [5] NCCL INFO Using network Libfabric
+ip-26-0-169-207:675555:675748 [4] NCCL INFO Using network Libfabric
+ip-26-0-169-207:675557:675749 [6] NCCL INFO Using network Libfabric
+ip-26-0-169-207:675558:675750 [7] NCCL INFO Using network Libfabric
+ip-26-0-169-207:675554:675751 [3] NCCL INFO Using network Libfabric
+ip-26-0-169-207:675553:675752 [2] NCCL INFO Using network Libfabric
+ip-26-0-169-207:675553:675752 [2] NCCL INFO comm 0x87954f0 rank 2 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0xb47817b68ab66608 - Init START
+ip-26-0-169-207:675554:675751 [3] NCCL INFO comm 0xa5edc10 rank 3 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0xb47817b68ab66608 - Init START
+ip-26-0-169-207:675551:675745 [0] NCCL INFO comm 0x8f470a0 rank 0 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0xb47817b68ab66608 - Init START
+ip-26-0-169-207:675555:675748 [4] NCCL INFO comm 0x9547310 rank 4 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0xb47817b68ab66608 - Init START
+ip-26-0-169-207:675557:675749 [6] NCCL INFO comm 0x9dcb560 rank 6 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0xb47817b68ab66608 - Init START
+ip-26-0-169-207:675558:675750 [7] NCCL INFO comm 0x97a3390 rank 7 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0xb47817b68ab66608 - Init START
+ip-26-0-169-207:675552:675747 [1] NCCL INFO comm 0x964dd80 rank 1 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0xb47817b68ab66608 - Init START
+ip-26-0-169-207:675556:675746 [5] NCCL INFO comm 0xa3a4de0 rank 5 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0xb47817b68ab66608 - Init START
+ip-26-0-169-207:675555:675748 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/149
+ip-26-0-169-207:675557:675749 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/149
+ip-26-0-169-207:675556:675746 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/149
+ip-26-0-169-207:675558:675750 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/149
+ip-26-0-169-207:675554:675751 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/149
+ip-26-0-169-207:675551:675745 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/162
+ip-26-0-169-207:675552:675747 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/149
+ip-26-0-169-207:675553:675752 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/149
+ip-26-0-169-207:675551:675745 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-169-207:675551:675745 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-169-207:675554:675751 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-169-207:675554:675751 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-169-207:675552:675747 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-169-207:675552:675747 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-169-207:675553:675752 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-169-207:675553:675752 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-169-207:675555:675748 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-169-207:675555:675748 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-169-207:675557:675749 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-169-207:675557:675749 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-169-207:675558:675750 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-169-207:675558:675750 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-169-207:675556:675746 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-169-207:675556:675746 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-169-207:675556:675746 [5] NCCL INFO Trees [0] 6/-1/-1->5->4 [1] 6/-1/-1->5->4 [2] 6/-1/-1->5->4 [3] 6/-1/-1->5->4 [4] 6/-1/-1->5->4 [5] 6/-1/-1->5->4 [6] 6/-1/-1->5->4 [7] 6/-1/-1->5->4 [8] 6/-1/-1->5->4 [9] 6/-1/-1->5->4 [10] 6/-1/-1->5->4 [11] 6/-1/-1->5->4 [12] 6/-1/-1->5->4 [13] 6/-1/-1->5->4 [14] 6/-1/-1->5->4 [15] 6/-1/-1->5->4 [16] 6/-1/-1->5->4 [17] 6/-1/-1->5->4 [18] 6/-1/-1->5->4 [19] 6/-1/-1->5->4 [20] 6/-1/-1->5->4 [21] 6/-1/-1->5->4 [22] 6/-1/-1->5->4 [23] 6/-1/-1->5->4
+ip-26-0-169-207:675556:675746 [5] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-169-207:675557:675749 [6] NCCL INFO Trees [0] 7/-1/-1->6->5 [1] 7/-1/-1->6->5 [2] 7/-1/-1->6->5 [3] 7/-1/-1->6->5 [4] 7/-1/-1->6->5 [5] 7/-1/-1->6->5 [6] 7/-1/-1->6->5 [7] 7/-1/-1->6->5 [8] 7/-1/-1->6->5 [9] 7/-1/-1->6->5 [10] 7/-1/-1->6->5 [11] 7/-1/-1->6->5 [12] 7/-1/-1->6->5 [13] 7/-1/-1->6->5 [14] 7/-1/-1->6->5 [15] 7/-1/-1->6->5 [16] 7/-1/-1->6->5 [17] 7/-1/-1->6->5 [18] 7/-1/-1->6->5 [19] 7/-1/-1->6->5 [20] 7/-1/-1->6->5 [21] 7/-1/-1->6->5 [22] 7/-1/-1->6->5 [23] 7/-1/-1->6->5
+ip-26-0-169-207:675555:675748 [4] NCCL INFO Trees [0] 5/-1/-1->4->3 [1] 5/-1/-1->4->3 [2] 5/-1/-1->4->3 [3] 5/-1/-1->4->3 [4] 5/-1/-1->4->3 [5] 5/-1/-1->4->3 [6] 5/-1/-1->4->3 [7] 5/-1/-1->4->3 [8] 5/-1/-1->4->3 [9] 5/-1/-1->4->3 [10] 5/-1/-1->4->3 [11] 5/-1/-1->4->3 [12] 5/-1/-1->4->3 [13] 5/-1/-1->4->3 [14] 5/-1/-1->4->3 [15] 5/-1/-1->4->3 [16] 5/-1/-1->4->3 [17] 5/-1/-1->4->3 [18] 5/-1/-1->4->3 [19] 5/-1/-1->4->3 [20] 5/-1/-1->4->3 [21] 5/-1/-1->4->3 [22] 5/-1/-1->4->3 [23] 5/-1/-1->4->3
+ip-26-0-169-207:675551:675745 [0] NCCL INFO Channel 00/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:675554:675751 [3] NCCL INFO Trees [0] 4/-1/-1->3->2 [1] 4/-1/-1->3->2 [2] 4/-1/-1->3->2 [3] 4/-1/-1->3->2 [4] 4/-1/-1->3->2 [5] 4/-1/-1->3->2 [6] 4/-1/-1->3->2 [7] 4/-1/-1->3->2 [8] 4/-1/-1->3->2 [9] 4/-1/-1->3->2 [10] 4/-1/-1->3->2 [11] 4/-1/-1->3->2 [12] 4/-1/-1->3->2 [13] 4/-1/-1->3->2 [14] 4/-1/-1->3->2 [15] 4/-1/-1->3->2 [16] 4/-1/-1->3->2 [17] 4/-1/-1->3->2 [18] 4/-1/-1->3->2 [19] 4/-1/-1->3->2 [20] 4/-1/-1->3->2 [21] 4/-1/-1->3->2 [22] 4/-1/-1->3->2 [23] 4/-1/-1->3->2
+ip-26-0-169-207:675555:675748 [4] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-169-207:675557:675749 [6] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-169-207:675558:675750 [7] NCCL INFO Trees [0] -1/-1/-1->7->6 [1] -1/-1/-1->7->6 [2] -1/-1/-1->7->6 [3] -1/-1/-1->7->6 [4] -1/-1/-1->7->6 [5] -1/-1/-1->7->6 [6] -1/-1/-1->7->6 [7] -1/-1/-1->7->6 [8] -1/-1/-1->7->6 [9] -1/-1/-1->7->6 [10] -1/-1/-1->7->6 [11] -1/-1/-1->7->6 [12] -1/-1/-1->7->6 [13] -1/-1/-1->7->6 [14] -1/-1/-1->7->6 [15] -1/-1/-1->7->6 [16] -1/-1/-1->7->6 [17] -1/-1/-1->7->6 [18] -1/-1/-1->7->6 [19] -1/-1/-1->7->6 [20] -1/-1/-1->7->6 [21] -1/-1/-1->7->6 [22] -1/-1/-1->7->6 [23] -1/-1/-1->7->6
+ip-26-0-169-207:675554:675751 [3] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-169-207:675553:675752 [2] NCCL INFO Trees [0] 3/-1/-1->2->1 [1] 3/-1/-1->2->1 [2] 3/-1/-1->2->1 [3] 3/-1/-1->2->1 [4] 3/-1/-1->2->1 [5] 3/-1/-1->2->1 [6] 3/-1/-1->2->1 [7] 3/-1/-1->2->1 [8] 3/-1/-1->2->1 [9] 3/-1/-1->2->1 [10] 3/-1/-1->2->1 [11] 3/-1/-1->2->1 [12] 3/-1/-1->2->1 [13] 3/-1/-1->2->1 [14] 3/-1/-1->2->1 [15] 3/-1/-1->2->1 [16] 3/-1/-1->2->1 [17] 3/-1/-1->2->1 [18] 3/-1/-1->2->1 [19] 3/-1/-1->2->1 [20] 3/-1/-1->2->1 [21] 3/-1/-1->2->1 [22] 3/-1/-1->2->1 [23] 3/-1/-1->2->1
+ip-26-0-169-207:675551:675745 [0] NCCL INFO Channel 01/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:675558:675750 [7] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-169-207:675552:675747 [1] NCCL INFO Trees [0] 2/-1/-1->1->0 [1] 2/-1/-1->1->0 [2] 2/-1/-1->1->0 [3] 2/-1/-1->1->0 [4] 2/-1/-1->1->0 [5] 2/-1/-1->1->0 [6] 2/-1/-1->1->0 [7] 2/-1/-1->1->0 [8] 2/-1/-1->1->0 [9] 2/-1/-1->1->0 [10] 2/-1/-1->1->0 [11] 2/-1/-1->1->0 [12] 2/-1/-1->1->0 [13] 2/-1/-1->1->0 [14] 2/-1/-1->1->0 [15] 2/-1/-1->1->0 [16] 2/-1/-1->1->0 [17] 2/-1/-1->1->0 [18] 2/-1/-1->1->0 [19] 2/-1/-1->1->0 [20] 2/-1/-1->1->0 [21] 2/-1/-1->1->0 [22] 2/-1/-1->1->0 [23] 2/-1/-1->1->0
+ip-26-0-169-207:675551:675745 [0] NCCL INFO Channel 02/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:675553:675752 [2] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-169-207:675551:675745 [0] NCCL INFO Channel 03/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:675552:675747 [1] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-169-207:675551:675745 [0] NCCL INFO Channel 04/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:675551:675745 [0] NCCL INFO Channel 05/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:675551:675745 [0] NCCL INFO Channel 06/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:675551:675745 [0] NCCL INFO Channel 07/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:675551:675745 [0] NCCL INFO Channel 08/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:675551:675745 [0] NCCL INFO Channel 09/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:675551:675745 [0] NCCL INFO Channel 10/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:675551:675745 [0] NCCL INFO Channel 11/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:675551:675745 [0] NCCL INFO Channel 12/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:675551:675745 [0] NCCL INFO Channel 13/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:675551:675745 [0] NCCL INFO Channel 14/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:675551:675745 [0] NCCL INFO Channel 15/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:675551:675745 [0] NCCL INFO Channel 16/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:675551:675745 [0] NCCL INFO Channel 17/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:675551:675745 [0] NCCL INFO Channel 18/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:675551:675745 [0] NCCL INFO Channel 19/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:675551:675745 [0] NCCL INFO Channel 20/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:675551:675745 [0] NCCL INFO Channel 21/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:675551:675745 [0] NCCL INFO Channel 22/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:675551:675745 [0] NCCL INFO Channel 23/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:675551:675745 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] 1/-1/-1->0->-1 [7] 1/-1/-1->0->-1 [8] 1/-1/-1->0->-1 [9] 1/-1/-1->0->-1 [10] 1/-1/-1->0->-1 [11] 1/-1/-1->0->-1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] 1/-1/-1->0->-1 [19] 1/-1/-1->0->-1 [20] 1/-1/-1->0->-1 [21] 1/-1/-1->0->-1 [22] 1/-1/-1->0->-1 [23] 1/-1/-1->0->-1
+ip-26-0-169-207:675551:675745 [0] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-169-207:675557:675749 [6] NCCL INFO Channel 00/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:675552:675747 [1] NCCL INFO Channel 00/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675558:675750 [7] NCCL INFO Channel 00/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675557:675749 [6] NCCL INFO Channel 01/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:675552:675747 [1] NCCL INFO Channel 01/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675558:675750 [7] NCCL INFO Channel 01/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675557:675749 [6] NCCL INFO Channel 02/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:675554:675751 [3] NCCL INFO Channel 00/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675552:675747 [1] NCCL INFO Channel 02/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675558:675750 [7] NCCL INFO Channel 02/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675557:675749 [6] NCCL INFO Channel 03/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:675554:675751 [3] NCCL INFO Channel 01/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675552:675747 [1] NCCL INFO Channel 03/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675558:675750 [7] NCCL INFO Channel 03/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675557:675749 [6] NCCL INFO Channel 04/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:675554:675751 [3] NCCL INFO Channel 02/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675552:675747 [1] NCCL INFO Channel 04/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675551:675745 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675557:675749 [6] NCCL INFO Channel 05/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:675558:675750 [7] NCCL INFO Channel 04/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675555:675748 [4] NCCL INFO Channel 00/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675554:675751 [3] NCCL INFO Channel 03/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675552:675747 [1] NCCL INFO Channel 05/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675551:675745 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675553:675752 [2] NCCL INFO Channel 00/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675556:675746 [5] NCCL INFO Channel 00/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675557:675749 [6] NCCL INFO Channel 06/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:675558:675750 [7] NCCL INFO Channel 05/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675555:675748 [4] NCCL INFO Channel 01/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675551:675745 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675554:675751 [3] NCCL INFO Channel 04/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675552:675747 [1] NCCL INFO Channel 06/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675553:675752 [2] NCCL INFO Channel 01/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675556:675746 [5] NCCL INFO Channel 01/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675557:675749 [6] NCCL INFO Channel 07/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:675558:675750 [7] NCCL INFO Channel 06/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675555:675748 [4] NCCL INFO Channel 02/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675551:675745 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675552:675747 [1] NCCL INFO Channel 07/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675553:675752 [2] NCCL INFO Channel 02/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675556:675746 [5] NCCL INFO Channel 02/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675557:675749 [6] NCCL INFO Channel 08/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:675558:675750 [7] NCCL INFO Channel 07/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675555:675748 [4] NCCL INFO Channel 03/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675551:675745 [0] NCCL INFO Channel 04/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675552:675747 [1] NCCL INFO Channel 08/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675553:675752 [2] NCCL INFO Channel 03/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675554:675751 [3] NCCL INFO Channel 05/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675556:675746 [5] NCCL INFO Channel 03/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675557:675749 [6] NCCL INFO Channel 09/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:675558:675750 [7] NCCL INFO Channel 08/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675555:675748 [4] NCCL INFO Channel 04/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675551:675745 [0] NCCL INFO Channel 05/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675552:675747 [1] NCCL INFO Channel 09/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675553:675752 [2] NCCL INFO Channel 04/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675557:675749 [6] NCCL INFO Channel 10/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:675556:675746 [5] NCCL INFO Channel 04/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675558:675750 [7] NCCL INFO Channel 09/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675555:675748 [4] NCCL INFO Channel 05/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675552:675747 [1] NCCL INFO Channel 10/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675554:675751 [3] NCCL INFO Channel 06/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675553:675752 [2] NCCL INFO Channel 05/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675557:675749 [6] NCCL INFO Channel 11/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:675556:675746 [5] NCCL INFO Channel 05/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675558:675750 [7] NCCL INFO Channel 10/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675555:675748 [4] NCCL INFO Channel 06/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675554:675751 [3] NCCL INFO Channel 07/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675552:675747 [1] NCCL INFO Channel 11/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675553:675752 [2] NCCL INFO Channel 06/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675551:675745 [0] NCCL INFO Channel 06/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675557:675749 [6] NCCL INFO Channel 12/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:675556:675746 [5] NCCL INFO Channel 06/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675558:675750 [7] NCCL INFO Channel 11/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675555:675748 [4] NCCL INFO Channel 07/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675554:675751 [3] NCCL INFO Channel 08/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675553:675752 [2] NCCL INFO Channel 07/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675552:675747 [1] NCCL INFO Channel 12/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675557:675749 [6] NCCL INFO Channel 13/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:675556:675746 [5] NCCL INFO Channel 07/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675558:675750 [7] NCCL INFO Channel 12/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675555:675748 [4] NCCL INFO Channel 08/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675553:675752 [2] NCCL INFO Channel 08/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675552:675747 [1] NCCL INFO Channel 13/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675554:675751 [3] NCCL INFO Channel 09/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675557:675749 [6] NCCL INFO Channel 14/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:675556:675746 [5] NCCL INFO Channel 08/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675551:675745 [0] NCCL INFO Channel 07/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675558:675750 [7] NCCL INFO Channel 13/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675555:675748 [4] NCCL INFO Channel 09/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675552:675747 [1] NCCL INFO Channel 14/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675554:675751 [3] NCCL INFO Channel 10/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675557:675749 [6] NCCL INFO Channel 15/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:675556:675746 [5] NCCL INFO Channel 09/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675552:675747 [1] NCCL INFO Channel 15/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675554:675751 [3] NCCL INFO Channel 11/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675557:675749 [6] NCCL INFO Channel 16/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:675556:675746 [5] NCCL INFO Channel 10/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675552:675747 [1] NCCL INFO Channel 16/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675554:675751 [3] NCCL INFO Channel 12/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675557:675749 [6] NCCL INFO Channel 17/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:675554:675751 [3] NCCL INFO Channel 13/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675558:675750 [7] NCCL INFO Channel 14/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675553:675752 [2] NCCL INFO Channel 09/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675555:675748 [4] NCCL INFO Channel 10/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675558:675750 [7] NCCL INFO Channel 15/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675554:675751 [3] NCCL INFO Channel 14/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675553:675752 [2] NCCL INFO Channel 10/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675551:675745 [0] NCCL INFO Channel 08/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675552:675747 [1] NCCL INFO Channel 17/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675556:675746 [5] NCCL INFO Channel 11/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675551:675745 [0] NCCL INFO Channel 09/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675557:675749 [6] NCCL INFO Channel 18/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:675555:675748 [4] NCCL INFO Channel 11/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675556:675746 [5] NCCL INFO Channel 12/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675551:675745 [0] NCCL INFO Channel 10/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675553:675752 [2] NCCL INFO Channel 11/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675557:675749 [6] NCCL INFO Channel 19/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:675556:675746 [5] NCCL INFO Channel 13/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675555:675748 [4] NCCL INFO Channel 12/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675558:675750 [7] NCCL INFO Channel 16/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675551:675745 [0] NCCL INFO Channel 11/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675553:675752 [2] NCCL INFO Channel 12/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675557:675749 [6] NCCL INFO Channel 20/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:675556:675746 [5] NCCL INFO Channel 14/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675555:675748 [4] NCCL INFO Channel 13/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675558:675750 [7] NCCL INFO Channel 17/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675551:675745 [0] NCCL INFO Channel 12/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675553:675752 [2] NCCL INFO Channel 13/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675558:675750 [7] NCCL INFO Channel 18/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675554:675751 [3] NCCL INFO Channel 15/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675552:675747 [1] NCCL INFO Channel 18/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675558:675750 [7] NCCL INFO Channel 19/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675553:675752 [2] NCCL INFO Channel 14/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675554:675751 [3] NCCL INFO Channel 16/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675555:675748 [4] NCCL INFO Channel 14/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675557:675749 [6] NCCL INFO Channel 21/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:675556:675746 [5] NCCL INFO Channel 15/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675551:675745 [0] NCCL INFO Channel 13/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675555:675748 [4] NCCL INFO Channel 15/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675557:675749 [6] NCCL INFO Channel 22/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:675556:675746 [5] NCCL INFO Channel 16/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675551:675745 [0] NCCL INFO Channel 14/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675554:675751 [3] NCCL INFO Channel 17/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675555:675748 [4] NCCL INFO Channel 16/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675556:675746 [5] NCCL INFO Channel 17/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675557:675749 [6] NCCL INFO Channel 23/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:675551:675745 [0] NCCL INFO Channel 15/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675553:675752 [2] NCCL INFO Channel 15/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675554:675751 [3] NCCL INFO Channel 18/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675555:675748 [4] NCCL INFO Channel 17/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675556:675746 [5] NCCL INFO Channel 18/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675553:675752 [2] NCCL INFO Channel 16/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675554:675751 [3] NCCL INFO Channel 19/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675555:675748 [4] NCCL INFO Channel 18/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675556:675746 [5] NCCL INFO Channel 19/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675552:675747 [1] NCCL INFO Channel 19/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675553:675752 [2] NCCL INFO Channel 17/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675554:675751 [3] NCCL INFO Channel 20/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675555:675748 [4] NCCL INFO Channel 19/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675556:675746 [5] NCCL INFO Channel 20/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675552:675747 [1] NCCL INFO Channel 20/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675553:675752 [2] NCCL INFO Channel 18/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675554:675751 [3] NCCL INFO Channel 21/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675555:675748 [4] NCCL INFO Channel 20/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675556:675746 [5] NCCL INFO Channel 21/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675552:675747 [1] NCCL INFO Channel 21/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675558:675750 [7] NCCL INFO Channel 20/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675555:675748 [4] NCCL INFO Channel 21/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675553:675752 [2] NCCL INFO Channel 19/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675554:675751 [3] NCCL INFO Channel 22/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675556:675746 [5] NCCL INFO Channel 22/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675552:675747 [1] NCCL INFO Channel 22/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675558:675750 [7] NCCL INFO Channel 21/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675555:675748 [4] NCCL INFO Channel 22/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675553:675752 [2] NCCL INFO Channel 20/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675556:675746 [5] NCCL INFO Channel 23/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675554:675751 [3] NCCL INFO Channel 23/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675552:675747 [1] NCCL INFO Channel 23/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675551:675745 [0] NCCL INFO Channel 16/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675558:675750 [7] NCCL INFO Channel 22/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675555:675748 [4] NCCL INFO Channel 23/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675553:675752 [2] NCCL INFO Channel 21/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675551:675745 [0] NCCL INFO Channel 17/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675558:675750 [7] NCCL INFO Channel 23/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675553:675752 [2] NCCL INFO Channel 22/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675551:675745 [0] NCCL INFO Channel 18/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675553:675752 [2] NCCL INFO Channel 23/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675551:675745 [0] NCCL INFO Channel 19/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675551:675745 [0] NCCL INFO Channel 20/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675551:675745 [0] NCCL INFO Channel 21/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675551:675745 [0] NCCL INFO Channel 22/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675551:675745 [0] NCCL INFO Channel 23/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675555:675748 [4] NCCL INFO Connected all rings
+ip-26-0-169-207:675557:675749 [6] NCCL INFO Connected all rings
+ip-26-0-169-207:675554:675751 [3] NCCL INFO Connected all rings
+ip-26-0-169-207:675553:675752 [2] NCCL INFO Connected all rings
+ip-26-0-169-207:675556:675746 [5] NCCL INFO Connected all rings
+ip-26-0-169-207:675552:675747 [1] NCCL INFO Connected all rings
+ip-26-0-169-207:675558:675750 [7] NCCL INFO Connected all rings
+ip-26-0-169-207:675558:675750 [7] NCCL INFO Channel 00/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675558:675750 [7] NCCL INFO Channel 01/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675551:675745 [0] NCCL INFO Connected all rings
+ip-26-0-169-207:675558:675750 [7] NCCL INFO Channel 02/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675558:675750 [7] NCCL INFO Channel 03/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675558:675750 [7] NCCL INFO Channel 04/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675558:675750 [7] NCCL INFO Channel 05/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675558:675750 [7] NCCL INFO Channel 06/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675555:675748 [4] NCCL INFO Channel 00/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675558:675750 [7] NCCL INFO Channel 07/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675555:675748 [4] NCCL INFO Channel 01/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675558:675750 [7] NCCL INFO Channel 08/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675557:675749 [6] NCCL INFO Channel 00/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675555:675748 [4] NCCL INFO Channel 02/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675558:675750 [7] NCCL INFO Channel 09/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675557:675749 [6] NCCL INFO Channel 01/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675555:675748 [4] NCCL INFO Channel 03/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675558:675750 [7] NCCL INFO Channel 10/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675557:675749 [6] NCCL INFO Channel 02/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675555:675748 [4] NCCL INFO Channel 04/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675558:675750 [7] NCCL INFO Channel 11/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675557:675749 [6] NCCL INFO Channel 03/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675556:675746 [5] NCCL INFO Channel 00/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675555:675748 [4] NCCL INFO Channel 05/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675558:675750 [7] NCCL INFO Channel 12/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675557:675749 [6] NCCL INFO Channel 04/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675552:675747 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675556:675746 [5] NCCL INFO Channel 01/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675558:675750 [7] NCCL INFO Channel 13/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675555:675748 [4] NCCL INFO Channel 06/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675557:675749 [6] NCCL INFO Channel 05/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675556:675746 [5] NCCL INFO Channel 02/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675552:675747 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675558:675750 [7] NCCL INFO Channel 14/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675555:675748 [4] NCCL INFO Channel 07/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675557:675749 [6] NCCL INFO Channel 06/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675556:675746 [5] NCCL INFO Channel 03/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675552:675747 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675555:675748 [4] NCCL INFO Channel 08/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675558:675750 [7] NCCL INFO Channel 15/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675557:675749 [6] NCCL INFO Channel 07/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675553:675752 [2] NCCL INFO Channel 00/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675556:675746 [5] NCCL INFO Channel 04/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675552:675747 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675555:675748 [4] NCCL INFO Channel 09/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675558:675750 [7] NCCL INFO Channel 16/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675557:675749 [6] NCCL INFO Channel 08/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675553:675752 [2] NCCL INFO Channel 01/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675556:675746 [5] NCCL INFO Channel 05/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675552:675747 [1] NCCL INFO Channel 04/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675555:675748 [4] NCCL INFO Channel 10/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675557:675749 [6] NCCL INFO Channel 09/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675553:675752 [2] NCCL INFO Channel 02/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675558:675750 [7] NCCL INFO Channel 17/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675556:675746 [5] NCCL INFO Channel 06/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675552:675747 [1] NCCL INFO Channel 05/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675555:675748 [4] NCCL INFO Channel 11/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675554:675751 [3] NCCL INFO Channel 00/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675557:675749 [6] NCCL INFO Channel 10/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675553:675752 [2] NCCL INFO Channel 03/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675558:675750 [7] NCCL INFO Channel 18/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675556:675746 [5] NCCL INFO Channel 07/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675552:675747 [1] NCCL INFO Channel 06/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675555:675748 [4] NCCL INFO Channel 12/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675557:675749 [6] NCCL INFO Channel 11/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675554:675751 [3] NCCL INFO Channel 01/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675553:675752 [2] NCCL INFO Channel 04/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675558:675750 [7] NCCL INFO Channel 19/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675556:675746 [5] NCCL INFO Channel 08/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675552:675747 [1] NCCL INFO Channel 07/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675555:675748 [4] NCCL INFO Channel 13/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675557:675749 [6] NCCL INFO Channel 12/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675553:675752 [2] NCCL INFO Channel 05/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675554:675751 [3] NCCL INFO Channel 02/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675558:675750 [7] NCCL INFO Channel 20/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675556:675746 [5] NCCL INFO Channel 09/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675552:675747 [1] NCCL INFO Channel 08/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675555:675748 [4] NCCL INFO Channel 14/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675557:675749 [6] NCCL INFO Channel 13/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675554:675751 [3] NCCL INFO Channel 03/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675553:675752 [2] NCCL INFO Channel 06/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675558:675750 [7] NCCL INFO Channel 21/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675556:675746 [5] NCCL INFO Channel 10/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675552:675747 [1] NCCL INFO Channel 09/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675555:675748 [4] NCCL INFO Channel 15/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675557:675749 [6] NCCL INFO Channel 14/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675554:675751 [3] NCCL INFO Channel 04/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675553:675752 [2] NCCL INFO Channel 07/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675558:675750 [7] NCCL INFO Channel 22/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675556:675746 [5] NCCL INFO Channel 11/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675552:675747 [1] NCCL INFO Channel 10/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675555:675748 [4] NCCL INFO Channel 16/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675554:675751 [3] NCCL INFO Channel 05/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675557:675749 [6] NCCL INFO Channel 15/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675553:675752 [2] NCCL INFO Channel 08/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675558:675750 [7] NCCL INFO Channel 23/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675556:675746 [5] NCCL INFO Channel 12/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675552:675747 [1] NCCL INFO Channel 11/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675553:675752 [2] NCCL INFO Channel 09/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675552:675747 [1] NCCL INFO Channel 12/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675553:675752 [2] NCCL INFO Channel 10/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675552:675747 [1] NCCL INFO Channel 13/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675555:675748 [4] NCCL INFO Channel 17/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675553:675752 [2] NCCL INFO Channel 11/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675557:675749 [6] NCCL INFO Channel 16/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675554:675751 [3] NCCL INFO Channel 06/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675552:675747 [1] NCCL INFO Channel 14/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675555:675748 [4] NCCL INFO Channel 18/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675553:675752 [2] NCCL INFO Channel 12/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675557:675749 [6] NCCL INFO Channel 17/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675554:675751 [3] NCCL INFO Channel 07/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675552:675747 [1] NCCL INFO Channel 15/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675555:675748 [4] NCCL INFO Channel 19/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675553:675752 [2] NCCL INFO Channel 13/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675557:675749 [6] NCCL INFO Channel 18/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675554:675751 [3] NCCL INFO Channel 08/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675555:675748 [4] NCCL INFO Channel 20/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675552:675747 [1] NCCL INFO Channel 16/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675553:675752 [2] NCCL INFO Channel 14/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675557:675749 [6] NCCL INFO Channel 19/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675554:675751 [3] NCCL INFO Channel 09/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675552:675747 [1] NCCL INFO Channel 17/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675555:675748 [4] NCCL INFO Channel 21/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675553:675752 [2] NCCL INFO Channel 15/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675557:675749 [6] NCCL INFO Channel 20/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675554:675751 [3] NCCL INFO Channel 10/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675556:675746 [5] NCCL INFO Channel 13/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675552:675747 [1] NCCL INFO Channel 18/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675555:675748 [4] NCCL INFO Channel 22/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675553:675752 [2] NCCL INFO Channel 16/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675557:675749 [6] NCCL INFO Channel 21/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675554:675751 [3] NCCL INFO Channel 11/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675556:675746 [5] NCCL INFO Channel 14/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675555:675748 [4] NCCL INFO Channel 23/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675552:675747 [1] NCCL INFO Channel 19/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675553:675752 [2] NCCL INFO Channel 17/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675557:675749 [6] NCCL INFO Channel 22/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675554:675751 [3] NCCL INFO Channel 12/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675556:675746 [5] NCCL INFO Channel 15/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675552:675747 [1] NCCL INFO Channel 20/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675553:675752 [2] NCCL INFO Channel 18/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675557:675749 [6] NCCL INFO Channel 23/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675554:675751 [3] NCCL INFO Channel 13/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675556:675746 [5] NCCL INFO Channel 16/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675553:675752 [2] NCCL INFO Channel 19/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675552:675747 [1] NCCL INFO Channel 21/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675554:675751 [3] NCCL INFO Channel 14/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675556:675746 [5] NCCL INFO Channel 17/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675553:675752 [2] NCCL INFO Channel 20/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675552:675747 [1] NCCL INFO Channel 22/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675554:675751 [3] NCCL INFO Channel 15/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675556:675746 [5] NCCL INFO Channel 18/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675553:675752 [2] NCCL INFO Channel 21/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675552:675747 [1] NCCL INFO Channel 23/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675554:675751 [3] NCCL INFO Channel 16/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675556:675746 [5] NCCL INFO Channel 19/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675553:675752 [2] NCCL INFO Channel 22/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675554:675751 [3] NCCL INFO Channel 17/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675556:675746 [5] NCCL INFO Channel 20/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675554:675751 [3] NCCL INFO Channel 18/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675553:675752 [2] NCCL INFO Channel 23/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675556:675746 [5] NCCL INFO Channel 21/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675554:675751 [3] NCCL INFO Channel 19/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675556:675746 [5] NCCL INFO Channel 22/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675554:675751 [3] NCCL INFO Channel 20/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675556:675746 [5] NCCL INFO Channel 23/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675554:675751 [3] NCCL INFO Channel 21/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675554:675751 [3] NCCL INFO Channel 22/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675554:675751 [3] NCCL INFO Channel 23/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675558:675750 [7] NCCL INFO Connected all trees
+ip-26-0-169-207:675558:675750 [7] NCCL INFO NVLS comm 0x97a3390 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-169-207:675551:675745 [0] NCCL INFO Connected all trees
+ip-26-0-169-207:675551:675745 [0] NCCL INFO NVLS comm 0x8f470a0 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-169-207:675552:675747 [1] NCCL INFO Connected all trees
+ip-26-0-169-207:675552:675747 [1] NCCL INFO NVLS comm 0x964dd80 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-169-207:675557:675749 [6] NCCL INFO Connected all trees
+ip-26-0-169-207:675557:675749 [6] NCCL INFO NVLS comm 0x9dcb560 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-169-207:675554:675751 [3] NCCL INFO Connected all trees
+ip-26-0-169-207:675554:675751 [3] NCCL INFO NVLS comm 0xa5edc10 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-169-207:675553:675752 [2] NCCL INFO Connected all trees
+ip-26-0-169-207:675553:675752 [2] NCCL INFO NVLS comm 0x87954f0 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-169-207:675556:675746 [5] NCCL INFO Connected all trees
+ip-26-0-169-207:675556:675746 [5] NCCL INFO NVLS comm 0xa3a4de0 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-169-207:675555:675748 [4] NCCL INFO Connected all trees
+ip-26-0-169-207:675555:675748 [4] NCCL INFO NVLS comm 0x9547310 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-169-207:675558:675750 [7] NCCL INFO Connected NVLS tree
+ip-26-0-169-207:675558:675750 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-169-207:675558:675750 [7] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-169-207:675558:675750 [7] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:675552:675747 [1] NCCL INFO Connected NVLS tree
+ip-26-0-169-207:675552:675747 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-169-207:675552:675747 [1] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-169-207:675552:675747 [1] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:675556:675746 [5] NCCL INFO Connected NVLS tree
+ip-26-0-169-207:675554:675751 [3] NCCL INFO Connected NVLS tree
+ip-26-0-169-207:675556:675746 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-169-207:675554:675751 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-169-207:675556:675746 [5] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-169-207:675556:675746 [5] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:675554:675751 [3] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-169-207:675554:675751 [3] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:675555:675748 [4] NCCL INFO Connected NVLS tree
+ip-26-0-169-207:675555:675748 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-169-207:675555:675748 [4] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-169-207:675555:675748 [4] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:675557:675749 [6] NCCL INFO Connected NVLS tree
+ip-26-0-169-207:675557:675749 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-169-207:675557:675749 [6] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-169-207:675557:675749 [6] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:675553:675752 [2] NCCL INFO Connected NVLS tree
+ip-26-0-169-207:675553:675752 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-169-207:675553:675752 [2] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-169-207:675553:675752 [2] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:675551:675745 [0] NCCL INFO Connected NVLS tree
+ip-26-0-169-207:675551:675745 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-169-207:675551:675745 [0] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-169-207:675551:675745 [0] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:675551:675745 [0] NCCL INFO comm 0x8f470a0 rank 0 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0xb47817b68ab66608 - Init COMPLETE
+ip-26-0-169-207:675557:675749 [6] NCCL INFO comm 0x9dcb560 rank 6 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0xb47817b68ab66608 - Init COMPLETE
+ip-26-0-169-207:675558:675750 [7] NCCL INFO comm 0x97a3390 rank 7 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0xb47817b68ab66608 - Init COMPLETE
+ip-26-0-169-207:675555:675748 [4] NCCL INFO comm 0x9547310 rank 4 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0xb47817b68ab66608 - Init COMPLETE
+ip-26-0-169-207:675553:675752 [2] NCCL INFO comm 0x87954f0 rank 2 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0xb47817b68ab66608 - Init COMPLETE
+ip-26-0-169-207:675556:675746 [5] NCCL INFO comm 0xa3a4de0 rank 5 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0xb47817b68ab66608 - Init COMPLETE
+ip-26-0-169-207:675552:675747 [1] NCCL INFO comm 0x964dd80 rank 1 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0xb47817b68ab66608 - Init COMPLETE
+ip-26-0-169-207:675554:675751 [3] NCCL INFO comm 0xa5edc10 rank 3 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0xb47817b68ab66608 - Init COMPLETE
+ip-26-0-169-207:675553:675781 [2] NCCL INFO Using network Libfabric
+ip-26-0-169-207:675558:675786 [7] NCCL INFO Using network Libfabric
+ip-26-0-169-207:675554:675785 [3] NCCL INFO Using network Libfabric
+ip-26-0-169-207:675557:675782 [6] NCCL INFO Using network Libfabric
+ip-26-0-169-207:675552:675780 [1] NCCL INFO Using network Libfabric
+ip-26-0-169-207:675556:675784 [5] NCCL INFO Using network Libfabric
+ip-26-0-169-207:675551:675779 [0] NCCL INFO Using network Libfabric
+ip-26-0-169-207:675555:675783 [4] NCCL INFO Using network Libfabric
+ip-26-0-169-207:675551:675779 [0] NCCL INFO comm 0x8f5b450 rank 0 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0x288ecad0b3f80051 - Init START
+ip-26-0-169-207:675558:675786 [7] NCCL INFO comm 0x97b7870 rank 7 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0x288ecad0b3f80051 - Init START
+ip-26-0-169-207:675557:675782 [6] NCCL INFO comm 0x9ddfca0 rank 6 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0x288ecad0b3f80051 - Init START
+ip-26-0-169-207:675555:675783 [4] NCCL INFO comm 0x955b8e0 rank 4 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0x288ecad0b3f80051 - Init START
+ip-26-0-169-207:675556:675784 [5] NCCL INFO comm 0xa3b9420 rank 5 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0x288ecad0b3f80051 - Init START
+ip-26-0-169-207:675552:675780 [1] NCCL INFO comm 0x9662230 rank 1 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0x288ecad0b3f80051 - Init START
+ip-26-0-169-207:675554:675785 [3] NCCL INFO comm 0xa6023a0 rank 3 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0x288ecad0b3f80051 - Init START
+ip-26-0-169-207:675553:675781 [2] NCCL INFO comm 0x87a9c80 rank 2 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0x288ecad0b3f80051 - Init START
+ip-26-0-169-207:675554:675785 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/149
+ip-26-0-169-207:675555:675783 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/149
+ip-26-0-169-207:675552:675780 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/149
+ip-26-0-169-207:675553:675781 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/149
+ip-26-0-169-207:675558:675786 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/149
+ip-26-0-169-207:675557:675782 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/149
+ip-26-0-169-207:675556:675784 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/149
+ip-26-0-169-207:675551:675779 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/162
+ip-26-0-169-207:675552:675780 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-169-207:675552:675780 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-169-207:675558:675786 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-169-207:675558:675786 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-169-207:675557:675782 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-169-207:675557:675782 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-169-207:675553:675781 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-169-207:675553:675781 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-169-207:675554:675785 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-169-207:675554:675785 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-169-207:675555:675783 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-169-207:675555:675783 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-169-207:675556:675784 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-169-207:675556:675784 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-169-207:675551:675779 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-169-207:675551:675779 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-169-207:675551:675779 [0] NCCL INFO Channel 00/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:675551:675779 [0] NCCL INFO Channel 01/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:675551:675779 [0] NCCL INFO Channel 02/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:675551:675779 [0] NCCL INFO Channel 03/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:675551:675779 [0] NCCL INFO Channel 04/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:675551:675779 [0] NCCL INFO Channel 05/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:675551:675779 [0] NCCL INFO Channel 06/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:675551:675779 [0] NCCL INFO Channel 07/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:675551:675779 [0] NCCL INFO Channel 08/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:675551:675779 [0] NCCL INFO Channel 09/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:675551:675779 [0] NCCL INFO Channel 10/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:675551:675779 [0] NCCL INFO Channel 11/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:675551:675779 [0] NCCL INFO Channel 12/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:675551:675779 [0] NCCL INFO Channel 13/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:675551:675779 [0] NCCL INFO Channel 14/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:675551:675779 [0] NCCL INFO Channel 15/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:675551:675779 [0] NCCL INFO Channel 16/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:675551:675779 [0] NCCL INFO Channel 17/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:675551:675779 [0] NCCL INFO Channel 18/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:675551:675779 [0] NCCL INFO Channel 19/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:675551:675779 [0] NCCL INFO Channel 20/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:675552:675780 [1] NCCL INFO Trees [0] 2/-1/-1->1->0 [1] 2/-1/-1->1->0 [2] 2/-1/-1->1->0 [3] 2/-1/-1->1->0 [4] 2/-1/-1->1->0 [5] 2/-1/-1->1->0 [6] 2/-1/-1->1->0 [7] 2/-1/-1->1->0 [8] 2/-1/-1->1->0 [9] 2/-1/-1->1->0 [10] 2/-1/-1->1->0 [11] 2/-1/-1->1->0 [12] 2/-1/-1->1->0 [13] 2/-1/-1->1->0 [14] 2/-1/-1->1->0 [15] 2/-1/-1->1->0 [16] 2/-1/-1->1->0 [17] 2/-1/-1->1->0 [18] 2/-1/-1->1->0 [19] 2/-1/-1->1->0 [20] 2/-1/-1->1->0 [21] 2/-1/-1->1->0 [22] 2/-1/-1->1->0 [23] 2/-1/-1->1->0
+ip-26-0-169-207:675551:675779 [0] NCCL INFO Channel 21/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:675553:675781 [2] NCCL INFO Trees [0] 3/-1/-1->2->1 [1] 3/-1/-1->2->1 [2] 3/-1/-1->2->1 [3] 3/-1/-1->2->1 [4] 3/-1/-1->2->1 [5] 3/-1/-1->2->1 [6] 3/-1/-1->2->1 [7] 3/-1/-1->2->1 [8] 3/-1/-1->2->1 [9] 3/-1/-1->2->1 [10] 3/-1/-1->2->1 [11] 3/-1/-1->2->1 [12] 3/-1/-1->2->1 [13] 3/-1/-1->2->1 [14] 3/-1/-1->2->1 [15] 3/-1/-1->2->1 [16] 3/-1/-1->2->1 [17] 3/-1/-1->2->1 [18] 3/-1/-1->2->1 [19] 3/-1/-1->2->1 [20] 3/-1/-1->2->1 [21] 3/-1/-1->2->1 [22] 3/-1/-1->2->1 [23] 3/-1/-1->2->1
+ip-26-0-169-207:675552:675780 [1] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-169-207:675551:675779 [0] NCCL INFO Channel 22/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:675553:675781 [2] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-169-207:675554:675785 [3] NCCL INFO Trees [0] 4/-1/-1->3->2 [1] 4/-1/-1->3->2 [2] 4/-1/-1->3->2 [3] 4/-1/-1->3->2 [4] 4/-1/-1->3->2 [5] 4/-1/-1->3->2 [6] 4/-1/-1->3->2 [7] 4/-1/-1->3->2 [8] 4/-1/-1->3->2 [9] 4/-1/-1->3->2 [10] 4/-1/-1->3->2 [11] 4/-1/-1->3->2 [12] 4/-1/-1->3->2 [13] 4/-1/-1->3->2 [14] 4/-1/-1->3->2 [15] 4/-1/-1->3->2 [16] 4/-1/-1->3->2 [17] 4/-1/-1->3->2 [18] 4/-1/-1->3->2 [19] 4/-1/-1->3->2 [20] 4/-1/-1->3->2 [21] 4/-1/-1->3->2 [22] 4/-1/-1->3->2 [23] 4/-1/-1->3->2
+ip-26-0-169-207:675551:675779 [0] NCCL INFO Channel 23/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:675554:675785 [3] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-169-207:675551:675779 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] 1/-1/-1->0->-1 [7] 1/-1/-1->0->-1 [8] 1/-1/-1->0->-1 [9] 1/-1/-1->0->-1 [10] 1/-1/-1->0->-1 [11] 1/-1/-1->0->-1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] 1/-1/-1->0->-1 [19] 1/-1/-1->0->-1 [20] 1/-1/-1->0->-1 [21] 1/-1/-1->0->-1 [22] 1/-1/-1->0->-1 [23] 1/-1/-1->0->-1
+ip-26-0-169-207:675551:675779 [0] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-169-207:675555:675783 [4] NCCL INFO Trees [0] 5/-1/-1->4->3 [1] 5/-1/-1->4->3 [2] 5/-1/-1->4->3 [3] 5/-1/-1->4->3 [4] 5/-1/-1->4->3 [5] 5/-1/-1->4->3 [6] 5/-1/-1->4->3 [7] 5/-1/-1->4->3 [8] 5/-1/-1->4->3 [9] 5/-1/-1->4->3 [10] 5/-1/-1->4->3 [11] 5/-1/-1->4->3 [12] 5/-1/-1->4->3 [13] 5/-1/-1->4->3 [14] 5/-1/-1->4->3 [15] 5/-1/-1->4->3 [16] 5/-1/-1->4->3 [17] 5/-1/-1->4->3 [18] 5/-1/-1->4->3 [19] 5/-1/-1->4->3 [20] 5/-1/-1->4->3 [21] 5/-1/-1->4->3 [22] 5/-1/-1->4->3 [23] 5/-1/-1->4->3
+ip-26-0-169-207:675555:675783 [4] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-169-207:675556:675784 [5] NCCL INFO Trees [0] 6/-1/-1->5->4 [1] 6/-1/-1->5->4 [2] 6/-1/-1->5->4 [3] 6/-1/-1->5->4 [4] 6/-1/-1->5->4 [5] 6/-1/-1->5->4 [6] 6/-1/-1->5->4 [7] 6/-1/-1->5->4 [8] 6/-1/-1->5->4 [9] 6/-1/-1->5->4 [10] 6/-1/-1->5->4 [11] 6/-1/-1->5->4 [12] 6/-1/-1->5->4 [13] 6/-1/-1->5->4 [14] 6/-1/-1->5->4 [15] 6/-1/-1->5->4 [16] 6/-1/-1->5->4 [17] 6/-1/-1->5->4 [18] 6/-1/-1->5->4 [19] 6/-1/-1->5->4 [20] 6/-1/-1->5->4 [21] 6/-1/-1->5->4 [22] 6/-1/-1->5->4 [23] 6/-1/-1->5->4
+ip-26-0-169-207:675556:675784 [5] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-169-207:675557:675782 [6] NCCL INFO Trees [0] 7/-1/-1->6->5 [1] 7/-1/-1->6->5 [2] 7/-1/-1->6->5 [3] 7/-1/-1->6->5 [4] 7/-1/-1->6->5 [5] 7/-1/-1->6->5 [6] 7/-1/-1->6->5 [7] 7/-1/-1->6->5 [8] 7/-1/-1->6->5 [9] 7/-1/-1->6->5 [10] 7/-1/-1->6->5 [11] 7/-1/-1->6->5 [12] 7/-1/-1->6->5 [13] 7/-1/-1->6->5 [14] 7/-1/-1->6->5 [15] 7/-1/-1->6->5 [16] 7/-1/-1->6->5 [17] 7/-1/-1->6->5 [18] 7/-1/-1->6->5 [19] 7/-1/-1->6->5 [20] 7/-1/-1->6->5 [21] 7/-1/-1->6->5 [22] 7/-1/-1->6->5 [23] 7/-1/-1->6->5
+ip-26-0-169-207:675557:675782 [6] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-169-207:675558:675786 [7] NCCL INFO Trees [0] -1/-1/-1->7->6 [1] -1/-1/-1->7->6 [2] -1/-1/-1->7->6 [3] -1/-1/-1->7->6 [4] -1/-1/-1->7->6 [5] -1/-1/-1->7->6 [6] -1/-1/-1->7->6 [7] -1/-1/-1->7->6 [8] -1/-1/-1->7->6 [9] -1/-1/-1->7->6 [10] -1/-1/-1->7->6 [11] -1/-1/-1->7->6 [12] -1/-1/-1->7->6 [13] -1/-1/-1->7->6 [14] -1/-1/-1->7->6 [15] -1/-1/-1->7->6 [16] -1/-1/-1->7->6 [17] -1/-1/-1->7->6 [18] -1/-1/-1->7->6 [19] -1/-1/-1->7->6 [20] -1/-1/-1->7->6 [21] -1/-1/-1->7->6 [22] -1/-1/-1->7->6 [23] -1/-1/-1->7->6
+ip-26-0-169-207:675558:675786 [7] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-169-207:675556:675784 [5] NCCL INFO Channel 00/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675557:675782 [6] NCCL INFO Channel 00/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:675556:675784 [5] NCCL INFO Channel 01/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675557:675782 [6] NCCL INFO Channel 01/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:675551:675779 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675556:675784 [5] NCCL INFO Channel 02/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675557:675782 [6] NCCL INFO Channel 02/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:675551:675779 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675556:675784 [5] NCCL INFO Channel 03/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675557:675782 [6] NCCL INFO Channel 03/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:675555:675783 [4] NCCL INFO Channel 00/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675551:675779 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675556:675784 [5] NCCL INFO Channel 04/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675557:675782 [6] NCCL INFO Channel 04/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:675554:675785 [3] NCCL INFO Channel 00/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675555:675783 [4] NCCL INFO Channel 01/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675558:675786 [7] NCCL INFO Channel 00/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675551:675779 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675556:675784 [5] NCCL INFO Channel 05/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675557:675782 [6] NCCL INFO Channel 05/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:675554:675785 [3] NCCL INFO Channel 01/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675555:675783 [4] NCCL INFO Channel 02/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675551:675779 [0] NCCL INFO Channel 04/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675558:675786 [7] NCCL INFO Channel 01/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675556:675784 [5] NCCL INFO Channel 06/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675557:675782 [6] NCCL INFO Channel 06/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:675554:675785 [3] NCCL INFO Channel 02/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675551:675779 [0] NCCL INFO Channel 05/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675555:675783 [4] NCCL INFO Channel 03/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675558:675786 [7] NCCL INFO Channel 02/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675556:675784 [5] NCCL INFO Channel 07/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675557:675782 [6] NCCL INFO Channel 07/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:675552:675780 [1] NCCL INFO Channel 00/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675551:675779 [0] NCCL INFO Channel 06/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675554:675785 [3] NCCL INFO Channel 03/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675555:675783 [4] NCCL INFO Channel 04/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675558:675786 [7] NCCL INFO Channel 03/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675553:675781 [2] NCCL INFO Channel 00/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675556:675784 [5] NCCL INFO Channel 08/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675557:675782 [6] NCCL INFO Channel 08/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:675552:675780 [1] NCCL INFO Channel 01/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675551:675779 [0] NCCL INFO Channel 07/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675554:675785 [3] NCCL INFO Channel 04/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675555:675783 [4] NCCL INFO Channel 05/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675558:675786 [7] NCCL INFO Channel 04/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675556:675784 [5] NCCL INFO Channel 09/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675553:675781 [2] NCCL INFO Channel 01/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675557:675782 [6] NCCL INFO Channel 09/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:675552:675780 [1] NCCL INFO Channel 02/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675551:675779 [0] NCCL INFO Channel 08/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675554:675785 [3] NCCL INFO Channel 05/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675555:675783 [4] NCCL INFO Channel 06/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675558:675786 [7] NCCL INFO Channel 05/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675556:675784 [5] NCCL INFO Channel 10/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675557:675782 [6] NCCL INFO Channel 10/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:675553:675781 [2] NCCL INFO Channel 02/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675552:675780 [1] NCCL INFO Channel 03/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675551:675779 [0] NCCL INFO Channel 09/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675554:675785 [3] NCCL INFO Channel 06/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675555:675783 [4] NCCL INFO Channel 07/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675558:675786 [7] NCCL INFO Channel 06/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675556:675784 [5] NCCL INFO Channel 11/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675557:675782 [6] NCCL INFO Channel 11/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:675553:675781 [2] NCCL INFO Channel 03/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675552:675780 [1] NCCL INFO Channel 04/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675551:675779 [0] NCCL INFO Channel 10/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675555:675783 [4] NCCL INFO Channel 08/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675554:675785 [3] NCCL INFO Channel 07/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675557:675782 [6] NCCL INFO Channel 12/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:675558:675786 [7] NCCL INFO Channel 07/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675556:675784 [5] NCCL INFO Channel 12/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675553:675781 [2] NCCL INFO Channel 04/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675552:675780 [1] NCCL INFO Channel 05/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675555:675783 [4] NCCL INFO Channel 09/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675551:675779 [0] NCCL INFO Channel 11/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675554:675785 [3] NCCL INFO Channel 08/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675558:675786 [7] NCCL INFO Channel 08/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675557:675782 [6] NCCL INFO Channel 13/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:675556:675784 [5] NCCL INFO Channel 13/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675553:675781 [2] NCCL INFO Channel 05/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675552:675780 [1] NCCL INFO Channel 06/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675555:675783 [4] NCCL INFO Channel 10/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675551:675779 [0] NCCL INFO Channel 12/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675554:675785 [3] NCCL INFO Channel 09/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675558:675786 [7] NCCL INFO Channel 09/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675553:675781 [2] NCCL INFO Channel 06/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675557:675782 [6] NCCL INFO Channel 14/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:675556:675784 [5] NCCL INFO Channel 14/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675552:675780 [1] NCCL INFO Channel 07/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675555:675783 [4] NCCL INFO Channel 11/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675551:675779 [0] NCCL INFO Channel 13/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675558:675786 [7] NCCL INFO Channel 10/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675554:675785 [3] NCCL INFO Channel 10/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675553:675781 [2] NCCL INFO Channel 07/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675557:675782 [6] NCCL INFO Channel 15/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:675556:675784 [5] NCCL INFO Channel 15/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675552:675780 [1] NCCL INFO Channel 08/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675555:675783 [4] NCCL INFO Channel 12/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675551:675779 [0] NCCL INFO Channel 14/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675558:675786 [7] NCCL INFO Channel 11/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675553:675781 [2] NCCL INFO Channel 08/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675554:675785 [3] NCCL INFO Channel 11/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675557:675782 [6] NCCL INFO Channel 16/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:675556:675784 [5] NCCL INFO Channel 16/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675552:675780 [1] NCCL INFO Channel 09/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675555:675783 [4] NCCL INFO Channel 13/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675551:675779 [0] NCCL INFO Channel 15/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675553:675781 [2] NCCL INFO Channel 09/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675558:675786 [7] NCCL INFO Channel 12/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675557:675782 [6] NCCL INFO Channel 17/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:675554:675785 [3] NCCL INFO Channel 12/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675556:675784 [5] NCCL INFO Channel 17/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675552:675780 [1] NCCL INFO Channel 10/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675555:675783 [4] NCCL INFO Channel 14/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675553:675781 [2] NCCL INFO Channel 10/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675551:675779 [0] NCCL INFO Channel 16/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675558:675786 [7] NCCL INFO Channel 13/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675557:675782 [6] NCCL INFO Channel 18/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:675556:675784 [5] NCCL INFO Channel 18/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675552:675780 [1] NCCL INFO Channel 11/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675554:675785 [3] NCCL INFO Channel 13/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675555:675783 [4] NCCL INFO Channel 15/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675551:675779 [0] NCCL INFO Channel 17/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675553:675781 [2] NCCL INFO Channel 11/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675558:675786 [7] NCCL INFO Channel 14/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675552:675780 [1] NCCL INFO Channel 12/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675557:675782 [6] NCCL INFO Channel 19/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:675556:675784 [5] NCCL INFO Channel 19/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675554:675785 [3] NCCL INFO Channel 14/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675555:675783 [4] NCCL INFO Channel 16/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675551:675779 [0] NCCL INFO Channel 18/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675553:675781 [2] NCCL INFO Channel 12/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675558:675786 [7] NCCL INFO Channel 15/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675552:675780 [1] NCCL INFO Channel 13/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675557:675782 [6] NCCL INFO Channel 20/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:675556:675784 [5] NCCL INFO Channel 20/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675554:675785 [3] NCCL INFO Channel 15/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675555:675783 [4] NCCL INFO Channel 17/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675551:675779 [0] NCCL INFO Channel 19/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675553:675781 [2] NCCL INFO Channel 13/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675558:675786 [7] NCCL INFO Channel 16/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675552:675780 [1] NCCL INFO Channel 14/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675557:675782 [6] NCCL INFO Channel 21/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:675556:675784 [5] NCCL INFO Channel 21/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675554:675785 [3] NCCL INFO Channel 16/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675551:675779 [0] NCCL INFO Channel 20/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675555:675783 [4] NCCL INFO Channel 18/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675558:675786 [7] NCCL INFO Channel 17/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675553:675781 [2] NCCL INFO Channel 14/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675557:675782 [6] NCCL INFO Channel 22/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:675552:675780 [1] NCCL INFO Channel 15/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675556:675784 [5] NCCL INFO Channel 22/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675554:675785 [3] NCCL INFO Channel 17/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675551:675779 [0] NCCL INFO Channel 21/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675558:675786 [7] NCCL INFO Channel 18/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675555:675783 [4] NCCL INFO Channel 19/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675552:675780 [1] NCCL INFO Channel 16/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675553:675781 [2] NCCL INFO Channel 15/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675557:675782 [6] NCCL INFO Channel 23/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:675554:675785 [3] NCCL INFO Channel 18/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675556:675784 [5] NCCL INFO Channel 23/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675551:675779 [0] NCCL INFO Channel 22/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675555:675783 [4] NCCL INFO Channel 20/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675552:675780 [1] NCCL INFO Channel 17/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675558:675786 [7] NCCL INFO Channel 19/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675553:675781 [2] NCCL INFO Channel 16/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675554:675785 [3] NCCL INFO Channel 19/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675551:675779 [0] NCCL INFO Channel 23/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675555:675783 [4] NCCL INFO Channel 21/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675552:675780 [1] NCCL INFO Channel 18/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675558:675786 [7] NCCL INFO Channel 20/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675553:675781 [2] NCCL INFO Channel 17/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675554:675785 [3] NCCL INFO Channel 20/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675555:675783 [4] NCCL INFO Channel 22/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675552:675780 [1] NCCL INFO Channel 19/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675553:675781 [2] NCCL INFO Channel 18/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675554:675785 [3] NCCL INFO Channel 21/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675558:675786 [7] NCCL INFO Channel 21/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675552:675780 [1] NCCL INFO Channel 20/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675553:675781 [2] NCCL INFO Channel 19/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675554:675785 [3] NCCL INFO Channel 22/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675555:675783 [4] NCCL INFO Channel 23/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675558:675786 [7] NCCL INFO Channel 22/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675552:675780 [1] NCCL INFO Channel 21/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675553:675781 [2] NCCL INFO Channel 20/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675554:675785 [3] NCCL INFO Channel 23/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675558:675786 [7] NCCL INFO Channel 23/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675553:675781 [2] NCCL INFO Channel 21/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675552:675780 [1] NCCL INFO Channel 22/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675553:675781 [2] NCCL INFO Channel 22/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675552:675780 [1] NCCL INFO Channel 23/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675553:675781 [2] NCCL INFO Channel 23/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675555:675783 [4] NCCL INFO Connected all rings
+ip-26-0-169-207:675551:675779 [0] NCCL INFO Connected all rings
+ip-26-0-169-207:675558:675786 [7] NCCL INFO Connected all rings
+ip-26-0-169-207:675558:675786 [7] NCCL INFO Channel 00/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675558:675786 [7] NCCL INFO Channel 01/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675558:675786 [7] NCCL INFO Channel 02/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675558:675786 [7] NCCL INFO Channel 03/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675556:675784 [5] NCCL INFO Connected all rings
+ip-26-0-169-207:675558:675786 [7] NCCL INFO Channel 04/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675553:675781 [2] NCCL INFO Connected all rings
+ip-26-0-169-207:675552:675780 [1] NCCL INFO Connected all rings
+ip-26-0-169-207:675558:675786 [7] NCCL INFO Channel 05/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675557:675782 [6] NCCL INFO Connected all rings
+ip-26-0-169-207:675558:675786 [7] NCCL INFO Channel 06/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675554:675785 [3] NCCL INFO Connected all rings
+ip-26-0-169-207:675558:675786 [7] NCCL INFO Channel 07/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675558:675786 [7] NCCL INFO Channel 08/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675558:675786 [7] NCCL INFO Channel 09/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675558:675786 [7] NCCL INFO Channel 10/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675558:675786 [7] NCCL INFO Channel 11/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675558:675786 [7] NCCL INFO Channel 12/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675558:675786 [7] NCCL INFO Channel 13/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675558:675786 [7] NCCL INFO Channel 14/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675558:675786 [7] NCCL INFO Channel 15/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675558:675786 [7] NCCL INFO Channel 16/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675558:675786 [7] NCCL INFO Channel 17/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675558:675786 [7] NCCL INFO Channel 18/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675558:675786 [7] NCCL INFO Channel 19/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675558:675786 [7] NCCL INFO Channel 20/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675558:675786 [7] NCCL INFO Channel 21/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675558:675786 [7] NCCL INFO Channel 22/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675558:675786 [7] NCCL INFO Channel 23/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:675555:675783 [4] NCCL INFO Channel 00/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675555:675783 [4] NCCL INFO Channel 01/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675555:675783 [4] NCCL INFO Channel 02/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675555:675783 [4] NCCL INFO Channel 03/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675556:675784 [5] NCCL INFO Channel 00/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675556:675784 [5] NCCL INFO Channel 01/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675555:675783 [4] NCCL INFO Channel 04/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675552:675780 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675555:675783 [4] NCCL INFO Channel 05/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675557:675782 [6] NCCL INFO Channel 00/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675552:675780 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675555:675783 [4] NCCL INFO Channel 06/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675557:675782 [6] NCCL INFO Channel 01/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675554:675785 [3] NCCL INFO Channel 00/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675552:675780 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675555:675783 [4] NCCL INFO Channel 07/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675553:675781 [2] NCCL INFO Channel 00/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675557:675782 [6] NCCL INFO Channel 02/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675554:675785 [3] NCCL INFO Channel 01/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675552:675780 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675555:675783 [4] NCCL INFO Channel 08/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675553:675781 [2] NCCL INFO Channel 01/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675557:675782 [6] NCCL INFO Channel 03/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675554:675785 [3] NCCL INFO Channel 02/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675552:675780 [1] NCCL INFO Channel 04/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675555:675783 [4] NCCL INFO Channel 09/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675553:675781 [2] NCCL INFO Channel 02/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675557:675782 [6] NCCL INFO Channel 04/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675554:675785 [3] NCCL INFO Channel 03/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675552:675780 [1] NCCL INFO Channel 05/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675555:675783 [4] NCCL INFO Channel 10/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675553:675781 [2] NCCL INFO Channel 03/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675557:675782 [6] NCCL INFO Channel 05/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675554:675785 [3] NCCL INFO Channel 04/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675552:675780 [1] NCCL INFO Channel 06/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675557:675782 [6] NCCL INFO Channel 06/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675555:675783 [4] NCCL INFO Channel 11/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675554:675785 [3] NCCL INFO Channel 05/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675556:675784 [5] NCCL INFO Channel 02/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675557:675782 [6] NCCL INFO Channel 07/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675555:675783 [4] NCCL INFO Channel 12/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675554:675785 [3] NCCL INFO Channel 06/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675556:675784 [5] NCCL INFO Channel 03/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675557:675782 [6] NCCL INFO Channel 08/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675555:675783 [4] NCCL INFO Channel 13/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675554:675785 [3] NCCL INFO Channel 07/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675556:675784 [5] NCCL INFO Channel 04/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675553:675781 [2] NCCL INFO Channel 04/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675557:675782 [6] NCCL INFO Channel 09/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675556:675784 [5] NCCL INFO Channel 05/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675555:675783 [4] NCCL INFO Channel 14/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675552:675780 [1] NCCL INFO Channel 07/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675553:675781 [2] NCCL INFO Channel 05/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675556:675784 [5] NCCL INFO Channel 06/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675552:675780 [1] NCCL INFO Channel 08/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675555:675783 [4] NCCL INFO Channel 15/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675554:675785 [3] NCCL INFO Channel 08/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675553:675781 [2] NCCL INFO Channel 06/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675556:675784 [5] NCCL INFO Channel 07/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675552:675780 [1] NCCL INFO Channel 09/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675555:675783 [4] NCCL INFO Channel 16/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675553:675781 [2] NCCL INFO Channel 07/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675554:675785 [3] NCCL INFO Channel 09/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675556:675784 [5] NCCL INFO Channel 08/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675555:675783 [4] NCCL INFO Channel 17/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675552:675780 [1] NCCL INFO Channel 10/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675554:675785 [3] NCCL INFO Channel 10/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675556:675784 [5] NCCL INFO Channel 09/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675555:675783 [4] NCCL INFO Channel 18/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675552:675780 [1] NCCL INFO Channel 11/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675554:675785 [3] NCCL INFO Channel 11/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675556:675784 [5] NCCL INFO Channel 10/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675553:675781 [2] NCCL INFO Channel 08/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675552:675780 [1] NCCL INFO Channel 12/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675557:675782 [6] NCCL INFO Channel 10/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675555:675783 [4] NCCL INFO Channel 19/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675553:675781 [2] NCCL INFO Channel 09/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675557:675782 [6] NCCL INFO Channel 11/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675552:675780 [1] NCCL INFO Channel 13/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675555:675783 [4] NCCL INFO Channel 20/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675553:675781 [2] NCCL INFO Channel 10/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675554:675785 [3] NCCL INFO Channel 12/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675552:675780 [1] NCCL INFO Channel 14/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675557:675782 [6] NCCL INFO Channel 12/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675555:675783 [4] NCCL INFO Channel 21/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675556:675784 [5] NCCL INFO Channel 11/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675554:675785 [3] NCCL INFO Channel 13/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675553:675781 [2] NCCL INFO Channel 11/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675552:675780 [1] NCCL INFO Channel 15/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675557:675782 [6] NCCL INFO Channel 13/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675555:675783 [4] NCCL INFO Channel 22/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675556:675784 [5] NCCL INFO Channel 12/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675554:675785 [3] NCCL INFO Channel 14/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675553:675781 [2] NCCL INFO Channel 12/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675552:675780 [1] NCCL INFO Channel 16/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675557:675782 [6] NCCL INFO Channel 14/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675555:675783 [4] NCCL INFO Channel 23/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:675556:675784 [5] NCCL INFO Channel 13/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675553:675781 [2] NCCL INFO Channel 13/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675552:675780 [1] NCCL INFO Channel 17/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675554:675785 [3] NCCL INFO Channel 15/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675557:675782 [6] NCCL INFO Channel 15/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675556:675784 [5] NCCL INFO Channel 14/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675554:675785 [3] NCCL INFO Channel 16/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675552:675780 [1] NCCL INFO Channel 18/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675553:675781 [2] NCCL INFO Channel 14/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675557:675782 [6] NCCL INFO Channel 16/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675554:675785 [3] NCCL INFO Channel 17/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675553:675781 [2] NCCL INFO Channel 15/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675556:675784 [5] NCCL INFO Channel 15/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675552:675780 [1] NCCL INFO Channel 19/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675557:675782 [6] NCCL INFO Channel 17/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675553:675781 [2] NCCL INFO Channel 16/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675554:675785 [3] NCCL INFO Channel 18/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675557:675782 [6] NCCL INFO Channel 18/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675556:675784 [5] NCCL INFO Channel 16/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675553:675781 [2] NCCL INFO Channel 17/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675554:675785 [3] NCCL INFO Channel 19/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675557:675782 [6] NCCL INFO Channel 19/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675556:675784 [5] NCCL INFO Channel 17/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675553:675781 [2] NCCL INFO Channel 18/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675554:675785 [3] NCCL INFO Channel 20/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675557:675782 [6] NCCL INFO Channel 20/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675556:675784 [5] NCCL INFO Channel 18/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675553:675781 [2] NCCL INFO Channel 19/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675554:675785 [3] NCCL INFO Channel 21/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675557:675782 [6] NCCL INFO Channel 21/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675556:675784 [5] NCCL INFO Channel 19/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675553:675781 [2] NCCL INFO Channel 20/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675554:675785 [3] NCCL INFO Channel 22/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675557:675782 [6] NCCL INFO Channel 22/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675556:675784 [5] NCCL INFO Channel 20/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675553:675781 [2] NCCL INFO Channel 21/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675554:675785 [3] NCCL INFO Channel 23/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:675552:675780 [1] NCCL INFO Channel 20/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675557:675782 [6] NCCL INFO Channel 23/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:675556:675784 [5] NCCL INFO Channel 21/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675553:675781 [2] NCCL INFO Channel 22/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675552:675780 [1] NCCL INFO Channel 21/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675556:675784 [5] NCCL INFO Channel 22/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675553:675781 [2] NCCL INFO Channel 23/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675552:675780 [1] NCCL INFO Channel 22/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675556:675784 [5] NCCL INFO Channel 23/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:675552:675780 [1] NCCL INFO Channel 23/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675558:675786 [7] NCCL INFO Connected all trees
+ip-26-0-169-207:675558:675786 [7] NCCL INFO NVLS comm 0x97b7870 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-169-207:675551:675779 [0] NCCL INFO Connected all trees
+ip-26-0-169-207:675551:675779 [0] NCCL INFO NVLS comm 0x8f5b450 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-169-207:675555:675783 [4] NCCL INFO Connected all trees
+ip-26-0-169-207:675555:675783 [4] NCCL INFO NVLS comm 0x955b8e0 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-169-207:675552:675780 [1] NCCL INFO Connected all trees
+ip-26-0-169-207:675552:675780 [1] NCCL INFO NVLS comm 0x9662230 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-169-207:675554:675785 [3] NCCL INFO Connected all trees
+ip-26-0-169-207:675554:675785 [3] NCCL INFO NVLS comm 0xa6023a0 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-169-207:675556:675784 [5] NCCL INFO Connected all trees
+ip-26-0-169-207:675556:675784 [5] NCCL INFO NVLS comm 0xa3b9420 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-169-207:675553:675781 [2] NCCL INFO Connected all trees
+ip-26-0-169-207:675553:675781 [2] NCCL INFO NVLS comm 0x87a9c80 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-169-207:675557:675782 [6] NCCL INFO Connected all trees
+ip-26-0-169-207:675557:675782 [6] NCCL INFO NVLS comm 0x9ddfca0 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-169-207:675552:675780 [1] NCCL INFO Connected NVLS tree
+ip-26-0-169-207:675552:675780 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-169-207:675552:675780 [1] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-169-207:675552:675780 [1] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:675556:675784 [5] NCCL INFO Connected NVLS tree
+ip-26-0-169-207:675556:675784 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-169-207:675556:675784 [5] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-169-207:675556:675784 [5] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:675554:675785 [3] NCCL INFO Connected NVLS tree
+ip-26-0-169-207:675554:675785 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-169-207:675554:675785 [3] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-169-207:675554:675785 [3] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:675555:675783 [4] NCCL INFO Connected NVLS tree
+ip-26-0-169-207:675555:675783 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-169-207:675555:675783 [4] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-169-207:675555:675783 [4] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:675557:675782 [6] NCCL INFO Connected NVLS tree
+ip-26-0-169-207:675557:675782 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-169-207:675557:675782 [6] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-169-207:675557:675782 [6] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:675551:675779 [0] NCCL INFO Connected NVLS tree
+ip-26-0-169-207:675551:675779 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-169-207:675553:675781 [2] NCCL INFO Connected NVLS tree
+ip-26-0-169-207:675551:675779 [0] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-169-207:675551:675779 [0] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:675553:675781 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-169-207:675553:675781 [2] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-169-207:675553:675781 [2] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:675558:675786 [7] NCCL INFO Connected NVLS tree
+ip-26-0-169-207:675558:675786 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-169-207:675558:675786 [7] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-169-207:675558:675786 [7] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:675558:675786 [7] NCCL INFO comm 0x97b7870 rank 7 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0x288ecad0b3f80051 - Init COMPLETE
+ip-26-0-169-207:675552:675780 [1] NCCL INFO comm 0x9662230 rank 1 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0x288ecad0b3f80051 - Init COMPLETE
+ip-26-0-169-207:675556:675784 [5] NCCL INFO comm 0xa3b9420 rank 5 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0x288ecad0b3f80051 - Init COMPLETE
+ip-26-0-169-207:675554:675785 [3] NCCL INFO comm 0xa6023a0 rank 3 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0x288ecad0b3f80051 - Init COMPLETE
+ip-26-0-169-207:675553:675781 [2] NCCL INFO comm 0x87a9c80 rank 2 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0x288ecad0b3f80051 - Init COMPLETE
+ip-26-0-169-207:675557:675782 [6] NCCL INFO comm 0x9ddfca0 rank 6 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0x288ecad0b3f80051 - Init COMPLETE
+ip-26-0-169-207:675555:675783 [4] NCCL INFO comm 0x955b8e0 rank 4 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0x288ecad0b3f80051 - Init COMPLETE
+ip-26-0-169-207:675551:675779 [0] NCCL INFO comm 0x8f5b450 rank 0 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0x288ecad0b3f80051 - Init COMPLETE
+12/13/2024 12:22:37 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]: Config:
+12/13/2024 12:22:37 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]: Config(general=GeneralArgs(project='debug',
+12/13/2024 12:22:37 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                            run='1.14G_dp4_tp2_pp1_acc256_mbs2_seq2048_zero1_tpmodeRED_vocab32k',
+12/13/2024 12:22:37 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                            seed=42,
+12/13/2024 12:22:37 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                            step=None,
+12/13/2024 12:22:37 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                            consumed_train_samples=None,
+12/13/2024 12:22:37 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                            benchmark_csv_path=PosixPath('benchmark/results/bench_final.csv'),
+12/13/2024 12:22:37 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                            ignore_sanity_checks=True),
+12/13/2024 12:22:37 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:        parallelism=ParallelismArgs(dp=4,
+12/13/2024 12:22:37 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                    pp=1,
+12/13/2024 12:22:37 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                    tp=2,
+12/13/2024 12:22:37 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                    pp_engine=<nanotron.parallel.pipeline_parallel.engine.OneForwardOneBackwardPipelineEngine object at 0x7fc454fa0d00>,
+12/13/2024 12:22:37 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                    tp_mode=<TensorParallelLinearMode.REDUCE_SCATTER: 2>,
+12/13/2024 12:22:37 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                    tp_linear_async_communication=True,
+12/13/2024 12:22:37 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                    recompute_layer=False,
+12/13/2024 12:22:37 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                    tp_recompute_allgather=True,
+12/13/2024 12:22:37 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                    expert_parallel_size=1),
+12/13/2024 12:22:37 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:        model=ModelArgs(model_config=LlamaConfig(bos_token_id=0,
+12/13/2024 12:22:37 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                                 eos_token_id=0,
+12/13/2024 12:22:37 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                                 hidden_act='silu',
+12/13/2024 12:22:37 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                                 hidden_size=2048,
+12/13/2024 12:22:37 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                                 initializer_range=0.02,
+12/13/2024 12:22:37 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                                 intermediate_size=8192,
+12/13/2024 12:22:37 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                                 is_llama_config=True,
+12/13/2024 12:22:37 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                                 max_position_embeddings=2048,
+12/13/2024 12:22:37 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                                 num_attention_heads=32,
+12/13/2024 12:22:37 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                                 num_hidden_layers=16,
+12/13/2024 12:22:37 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                                 num_key_value_heads=8,
+12/13/2024 12:22:37 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                                 pad_token_id=None,
+12/13/2024 12:22:37 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                                 pretraining_tp=1,
+12/13/2024 12:22:37 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                                 rms_norm_eps=1e-05,
+12/13/2024 12:22:37 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                                 rope_scaling=None,
+12/13/2024 12:22:37 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                                 rope_theta=10000.0,
+12/13/2024 12:22:37 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                                 rope_interleaved=False,
+12/13/2024 12:22:37 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                                 tie_word_embeddings=True,
+12/13/2024 12:22:37 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                                 use_cache=True,
+12/13/2024 12:22:37 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                                 vocab_size=32768),
+12/13/2024 12:22:37 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                        init_method=RandomInit(std=0.02),
+12/13/2024 12:22:37 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                        dtype=torch.bfloat16,
+12/13/2024 12:22:37 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                        make_vocab_size_divisible_by=1,
+12/13/2024 12:22:37 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                        ddp_bucket_cap_mb=25),
+12/13/2024 12:22:37 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:        tokenizer=TokenizerArgs(tokenizer_name_or_path='robot-test/dummy-tokenizer-wordlevel',
+12/13/2024 12:22:37 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                tokenizer_revision=None,
+12/13/2024 12:22:37 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                tokenizer_max_length=None),
+12/13/2024 12:22:37 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:        checkpoints=CheckpointsArgs(checkpoints_path=PosixPath('checkpoints'),
+12/13/2024 12:22:37 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                    checkpoint_interval=10000,
+12/13/2024 12:22:37 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                    save_initial_state=False,
+12/13/2024 12:22:37 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                    save_final_state=False,
+12/13/2024 12:22:37 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                    resume_checkpoint_path=None,
+12/13/2024 12:22:37 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                    checkpoints_path_is_shared_file_system=False),
+12/13/2024 12:22:37 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:        logging=LoggingArgs(log_level='info',
+12/13/2024 12:22:37 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                            log_level_replica='info',
+12/13/2024 12:22:37 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                            iteration_step_info_interval=1),
+12/13/2024 12:22:37 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:        tokens=TokensArgs(sequence_length=2048,
+12/13/2024 12:22:37 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                          train_steps=100,
+12/13/2024 12:22:37 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                          micro_batch_size=2,
+12/13/2024 12:22:37 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                          batch_accumulation_per_replica=256,
+12/13/2024 12:22:37 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                          val_check_interval=100,
+12/13/2024 12:22:37 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                          limit_val_batches=0,
+12/13/2024 12:22:37 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                          limit_test_batches=0),
+12/13/2024 12:22:37 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:        optimizer=OptimizerArgs(optimizer_factory=AdamWOptimizerArgs(adam_eps=1e-08,
+12/13/2024 12:22:37 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                                                     adam_beta1=0.9,
+12/13/2024 12:22:37 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                                                     adam_beta2=0.95,
+12/13/2024 12:22:37 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                                                     torch_adam_is_fused=True,
+12/13/2024 12:22:37 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                                                     name='adamW'),
+12/13/2024 12:22:37 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                zero_stage=1,
+12/13/2024 12:22:37 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                weight_decay=0.01,
+12/13/2024 12:22:37 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                clip_grad=1.0,
+12/13/2024 12:22:37 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                accumulate_grad_in_fp32=True,
+12/13/2024 12:22:37 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                learning_rate_scheduler=LRSchedulerArgs(learning_rate=0.0003,
+12/13/2024 12:22:37 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                                                        lr_warmup_steps=2,
+12/13/2024 12:22:37 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                                                        lr_warmup_style='linear',
+12/13/2024 12:22:37 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                                                        lr_decay_style='cosine',
+12/13/2024 12:22:37 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                                                        lr_decay_steps=13,
+12/13/2024 12:22:37 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                                                        lr_decay_starting_step=None,
+12/13/2024 12:22:37 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                                                        min_decay_lr=1e-05)),
+12/13/2024 12:22:37 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:        data_stages=[DatasetStageArgs(name='Stable Training Stage',
+12/13/2024 12:22:37 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                      start_training_step=1,
+12/13/2024 12:22:37 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                      data=DataArgs(dataset=None,
+12/13/2024 12:22:37 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                                    seed=42,
+12/13/2024 12:22:37 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                                    num_loading_workers=1))],
+12/13/2024 12:22:37 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:        profiler=None,
+12/13/2024 12:22:37 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:        lighteval=None,
+12/13/2024 12:22:37 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:        s3_upload=None)
+12/13/2024 12:22:37 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]: Model Config:
+12/13/2024 12:22:37 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]: LlamaConfig(bos_token_id=0,
+12/13/2024 12:22:37 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:             eos_token_id=0,
+12/13/2024 12:22:37 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:             hidden_act='silu',
+12/13/2024 12:22:37 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:             hidden_size=2048,
+12/13/2024 12:22:37 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:             initializer_range=0.02,
+12/13/2024 12:22:37 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:             intermediate_size=8192,
+12/13/2024 12:22:37 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:             is_llama_config=True,
+12/13/2024 12:22:37 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:             max_position_embeddings=2048,
+12/13/2024 12:22:37 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:             num_attention_heads=32,
+12/13/2024 12:22:37 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:             num_hidden_layers=16,
+12/13/2024 12:22:37 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:             num_key_value_heads=8,
+12/13/2024 12:22:37 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:             pad_token_id=None,
+12/13/2024 12:22:37 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:             pretraining_tp=1,
+12/13/2024 12:22:37 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:             rms_norm_eps=1e-05,
+12/13/2024 12:22:37 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:             rope_scaling=None,
+12/13/2024 12:22:37 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:             rope_theta=10000.0,
+12/13/2024 12:22:37 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:             rope_interleaved=False,
+12/13/2024 12:22:37 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:             tie_word_embeddings=True,
+12/13/2024 12:22:37 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:             use_cache=True,
+12/13/2024 12:22:37 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:             vocab_size=32768)
+12/13/2024 12:22:37 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]: Building model..
+12/13/2024 12:22:37 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]: Initialize RoPE Theta = 10000.0
+12/13/2024 12:22:37 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]: Setting PP block ranks...
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+ip-26-0-169-207:675555:675833 [4] NCCL INFO Using network Libfabric
+ip-26-0-169-207:675551:675832 [0] NCCL INFO Using network Libfabric
+ip-26-0-169-207:675557:675834 [6] NCCL INFO Using network Libfabric
+ip-26-0-169-207:675553:675835 [2] NCCL INFO Using network Libfabric
+ip-26-0-169-207:675552:675837 [1] NCCL INFO Using network Libfabric
+ip-26-0-169-207:675556:675836 [5] NCCL INFO Using network Libfabric
+ip-26-0-169-207:675554:675839 [3] NCCL INFO Using network Libfabric
+ip-26-0-169-207:675558:675838 [7] NCCL INFO Using network Libfabric
+ip-26-0-169-207:675551:675832 [0] NCCL INFO comm 0xad5c630 rank 0 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0x43c20e17bdec1db - Init START
+ip-26-0-169-207:675552:675837 [1] NCCL INFO comm 0xb464bf0 rank 1 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0x43c20e17bdec1db - Init START
+ip-26-0-169-207:675555:675833 [4] NCCL INFO comm 0xb35dff0 rank 0 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0xcdaefaa4cc5cec90 - Init START
+ip-26-0-169-207:675556:675836 [5] NCCL INFO comm 0xc1bc3b0 rank 1 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0xcdaefaa4cc5cec90 - Init START
+ip-26-0-169-207:675553:675835 [2] NCCL INFO comm 0xa5aafe0 rank 0 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0xa80432eabeb5875c - Init START
+ip-26-0-169-207:675554:675839 [3] NCCL INFO comm 0xc404770 rank 1 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0xa80432eabeb5875c - Init START
+ip-26-0-169-207:675551:675832 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/162
+ip-26-0-169-207:675555:675833 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/149
+ip-26-0-169-207:675553:675835 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/149
+ip-26-0-169-207:675557:675834 [6] NCCL INFO comm 0xbbdf3e0 rank 0 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0x2d4ea52ae5c54be - Init START
+ip-26-0-169-207:675558:675838 [7] NCCL INFO comm 0xb5babb0 rank 1 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0x2d4ea52ae5c54be - Init START
+ip-26-0-169-207:675557:675834 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/149
+ip-26-0-169-207:675552:675837 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/149
+ip-26-0-169-207:675556:675836 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/149
+ip-26-0-169-207:675558:675838 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/149
+ip-26-0-169-207:675554:675839 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/149
+ip-26-0-169-207:675556:675836 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-169-207:675552:675837 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-169-207:675558:675838 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-169-207:675553:675835 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-169-207:675554:675839 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-169-207:675553:675835 [2] NCCL INFO Channel 00/24 :    0   1
+ip-26-0-169-207:675553:675835 [2] NCCL INFO Channel 01/24 :    0   1
+ip-26-0-169-207:675553:675835 [2] NCCL INFO Channel 02/24 :    0   1
+ip-26-0-169-207:675553:675835 [2] NCCL INFO Channel 03/24 :    0   1
+ip-26-0-169-207:675553:675835 [2] NCCL INFO Channel 04/24 :    0   1
+ip-26-0-169-207:675553:675835 [2] NCCL INFO Channel 05/24 :    0   1
+ip-26-0-169-207:675553:675835 [2] NCCL INFO Channel 06/24 :    0   1
+ip-26-0-169-207:675553:675835 [2] NCCL INFO Channel 07/24 :    0   1
+ip-26-0-169-207:675553:675835 [2] NCCL INFO Channel 08/24 :    0   1
+ip-26-0-169-207:675553:675835 [2] NCCL INFO Channel 09/24 :    0   1
+ip-26-0-169-207:675553:675835 [2] NCCL INFO Channel 10/24 :    0   1
+ip-26-0-169-207:675554:675839 [3] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] -1/-1/-1->1->0 [3] -1/-1/-1->1->0 [4] -1/-1/-1->1->0 [5] -1/-1/-1->1->0 [6] 0/-1/-1->1->-1 [7] 0/-1/-1->1->-1 [8] 0/-1/-1->1->-1 [9] 0/-1/-1->1->-1 [10] 0/-1/-1->1->-1 [11] 0/-1/-1->1->-1 [12] -1/-1/-1->1->0 [13] -1/-1/-1->1->0 [14] -1/-1/-1->1->0 [15] -1/-1/-1->1->0 [16] -1/-1/-1->1->0 [17] -1/-1/-1->1->0 [18] 0/-1/-1->1->-1 [19] 0/-1/-1->1->-1 [20] 0/-1/-1->1->-1 [21] 0/-1/-1->1->-1 [22] 0/-1/-1->1->-1 [23] 0/-1/-1->1->-1
+ip-26-0-169-207:675553:675835 [2] NCCL INFO Channel 11/24 :    0   1
+ip-26-0-169-207:675553:675835 [2] NCCL INFO Channel 12/24 :    0   1
+ip-26-0-169-207:675553:675835 [2] NCCL INFO Channel 13/24 :    0   1
+ip-26-0-169-207:675554:675839 [3] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-169-207:675553:675835 [2] NCCL INFO Channel 14/24 :    0   1
+ip-26-0-169-207:675553:675835 [2] NCCL INFO Channel 15/24 :    0   1
+ip-26-0-169-207:675553:675835 [2] NCCL INFO Channel 16/24 :    0   1
+ip-26-0-169-207:675553:675835 [2] NCCL INFO Channel 17/24 :    0   1
+ip-26-0-169-207:675553:675835 [2] NCCL INFO Channel 18/24 :    0   1
+ip-26-0-169-207:675553:675835 [2] NCCL INFO Channel 19/24 :    0   1
+ip-26-0-169-207:675553:675835 [2] NCCL INFO Channel 20/24 :    0   1
+ip-26-0-169-207:675553:675835 [2] NCCL INFO Channel 21/24 :    0   1
+ip-26-0-169-207:675553:675835 [2] NCCL INFO Channel 22/24 :    0   1
+ip-26-0-169-207:675553:675835 [2] NCCL INFO Channel 23/24 :    0   1
+ip-26-0-169-207:675553:675835 [2] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] -1/-1/-1->0->1 [7] -1/-1/-1->0->1 [8] -1/-1/-1->0->1 [9] -1/-1/-1->0->1 [10] -1/-1/-1->0->1 [11] -1/-1/-1->0->1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] -1/-1/-1->0->1 [19] -1/-1/-1->0->1 [20] -1/-1/-1->0->1 [21] -1/-1/-1->0->1 [22] -1/-1/-1->0->1 [23] -1/-1/-1->0->1
+ip-26-0-169-207:675553:675835 [2] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-169-207:675551:675832 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-169-207:675551:675832 [0] NCCL INFO Channel 00/24 :    0   1
+ip-26-0-169-207:675551:675832 [0] NCCL INFO Channel 01/24 :    0   1
+ip-26-0-169-207:675551:675832 [0] NCCL INFO Channel 02/24 :    0   1
+ip-26-0-169-207:675552:675837 [1] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] -1/-1/-1->1->0 [3] -1/-1/-1->1->0 [4] -1/-1/-1->1->0 [5] -1/-1/-1->1->0 [6] 0/-1/-1->1->-1 [7] 0/-1/-1->1->-1 [8] 0/-1/-1->1->-1 [9] 0/-1/-1->1->-1 [10] 0/-1/-1->1->-1 [11] 0/-1/-1->1->-1 [12] -1/-1/-1->1->0 [13] -1/-1/-1->1->0 [14] -1/-1/-1->1->0 [15] -1/-1/-1->1->0 [16] -1/-1/-1->1->0 [17] -1/-1/-1->1->0 [18] 0/-1/-1->1->-1 [19] 0/-1/-1->1->-1 [20] 0/-1/-1->1->-1 [21] 0/-1/-1->1->-1 [22] 0/-1/-1->1->-1 [23] 0/-1/-1->1->-1
+ip-26-0-169-207:675551:675832 [0] NCCL INFO Channel 03/24 :    0   1
+ip-26-0-169-207:675551:675832 [0] NCCL INFO Channel 04/24 :    0   1
+ip-26-0-169-207:675552:675837 [1] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-169-207:675551:675832 [0] NCCL INFO Channel 05/24 :    0   1
+ip-26-0-169-207:675551:675832 [0] NCCL INFO Channel 06/24 :    0   1
+ip-26-0-169-207:675551:675832 [0] NCCL INFO Channel 07/24 :    0   1
+ip-26-0-169-207:675551:675832 [0] NCCL INFO Channel 08/24 :    0   1
+ip-26-0-169-207:675551:675832 [0] NCCL INFO Channel 09/24 :    0   1
+ip-26-0-169-207:675551:675832 [0] NCCL INFO Channel 10/24 :    0   1
+ip-26-0-169-207:675551:675832 [0] NCCL INFO Channel 11/24 :    0   1
+ip-26-0-169-207:675551:675832 [0] NCCL INFO Channel 12/24 :    0   1
+ip-26-0-169-207:675551:675832 [0] NCCL INFO Channel 13/24 :    0   1
+ip-26-0-169-207:675551:675832 [0] NCCL INFO Channel 14/24 :    0   1
+ip-26-0-169-207:675551:675832 [0] NCCL INFO Channel 15/24 :    0   1
+ip-26-0-169-207:675551:675832 [0] NCCL INFO Channel 16/24 :    0   1
+ip-26-0-169-207:675551:675832 [0] NCCL INFO Channel 17/24 :    0   1
+ip-26-0-169-207:675551:675832 [0] NCCL INFO Channel 18/24 :    0   1
+ip-26-0-169-207:675551:675832 [0] NCCL INFO Channel 19/24 :    0   1
+ip-26-0-169-207:675551:675832 [0] NCCL INFO Channel 20/24 :    0   1
+ip-26-0-169-207:675551:675832 [0] NCCL INFO Channel 21/24 :    0   1
+ip-26-0-169-207:675551:675832 [0] NCCL INFO Channel 22/24 :    0   1
+ip-26-0-169-207:675551:675832 [0] NCCL INFO Channel 23/24 :    0   1
+ip-26-0-169-207:675551:675832 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] -1/-1/-1->0->1 [7] -1/-1/-1->0->1 [8] -1/-1/-1->0->1 [9] -1/-1/-1->0->1 [10] -1/-1/-1->0->1 [11] -1/-1/-1->0->1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] -1/-1/-1->0->1 [19] -1/-1/-1->0->1 [20] -1/-1/-1->0->1 [21] -1/-1/-1->0->1 [22] -1/-1/-1->0->1 [23] -1/-1/-1->0->1
+ip-26-0-169-207:675551:675832 [0] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-169-207:675557:675834 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-169-207:675557:675834 [6] NCCL INFO Channel 00/24 :    0   1
+ip-26-0-169-207:675557:675834 [6] NCCL INFO Channel 01/24 :    0   1
+ip-26-0-169-207:675557:675834 [6] NCCL INFO Channel 02/24 :    0   1
+ip-26-0-169-207:675557:675834 [6] NCCL INFO Channel 03/24 :    0   1
+ip-26-0-169-207:675557:675834 [6] NCCL INFO Channel 04/24 :    0   1
+ip-26-0-169-207:675557:675834 [6] NCCL INFO Channel 05/24 :    0   1
+ip-26-0-169-207:675557:675834 [6] NCCL INFO Channel 06/24 :    0   1
+ip-26-0-169-207:675557:675834 [6] NCCL INFO Channel 07/24 :    0   1
+ip-26-0-169-207:675557:675834 [6] NCCL INFO Channel 08/24 :    0   1
+ip-26-0-169-207:675557:675834 [6] NCCL INFO Channel 09/24 :    0   1
+ip-26-0-169-207:675557:675834 [6] NCCL INFO Channel 10/24 :    0   1
+ip-26-0-169-207:675557:675834 [6] NCCL INFO Channel 11/24 :    0   1
+ip-26-0-169-207:675557:675834 [6] NCCL INFO Channel 12/24 :    0   1
+ip-26-0-169-207:675557:675834 [6] NCCL INFO Channel 13/24 :    0   1
+ip-26-0-169-207:675557:675834 [6] NCCL INFO Channel 14/24 :    0   1
+ip-26-0-169-207:675557:675834 [6] NCCL INFO Channel 15/24 :    0   1
+ip-26-0-169-207:675557:675834 [6] NCCL INFO Channel 16/24 :    0   1
+ip-26-0-169-207:675557:675834 [6] NCCL INFO Channel 17/24 :    0   1
+ip-26-0-169-207:675557:675834 [6] NCCL INFO Channel 18/24 :    0   1
+ip-26-0-169-207:675557:675834 [6] NCCL INFO Channel 19/24 :    0   1
+ip-26-0-169-207:675557:675834 [6] NCCL INFO Channel 20/24 :    0   1
+ip-26-0-169-207:675557:675834 [6] NCCL INFO Channel 21/24 :    0   1
+ip-26-0-169-207:675557:675834 [6] NCCL INFO Channel 22/24 :    0   1
+ip-26-0-169-207:675557:675834 [6] NCCL INFO Channel 23/24 :    0   1
+ip-26-0-169-207:675558:675838 [7] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] -1/-1/-1->1->0 [3] -1/-1/-1->1->0 [4] -1/-1/-1->1->0 [5] -1/-1/-1->1->0 [6] 0/-1/-1->1->-1 [7] 0/-1/-1->1->-1 [8] 0/-1/-1->1->-1 [9] 0/-1/-1->1->-1 [10] 0/-1/-1->1->-1 [11] 0/-1/-1->1->-1 [12] -1/-1/-1->1->0 [13] -1/-1/-1->1->0 [14] -1/-1/-1->1->0 [15] -1/-1/-1->1->0 [16] -1/-1/-1->1->0 [17] -1/-1/-1->1->0 [18] 0/-1/-1->1->-1 [19] 0/-1/-1->1->-1 [20] 0/-1/-1->1->-1 [21] 0/-1/-1->1->-1 [22] 0/-1/-1->1->-1 [23] 0/-1/-1->1->-1
+ip-26-0-169-207:675558:675838 [7] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-169-207:675557:675834 [6] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] -1/-1/-1->0->1 [7] -1/-1/-1->0->1 [8] -1/-1/-1->0->1 [9] -1/-1/-1->0->1 [10] -1/-1/-1->0->1 [11] -1/-1/-1->0->1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] -1/-1/-1->0->1 [19] -1/-1/-1->0->1 [20] -1/-1/-1->0->1 [21] -1/-1/-1->0->1 [22] -1/-1/-1->0->1 [23] -1/-1/-1->0->1
+ip-26-0-169-207:675557:675834 [6] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-169-207:675555:675833 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-169-207:675555:675833 [4] NCCL INFO Channel 00/24 :    0   1
+ip-26-0-169-207:675555:675833 [4] NCCL INFO Channel 01/24 :    0   1
+ip-26-0-169-207:675555:675833 [4] NCCL INFO Channel 02/24 :    0   1
+ip-26-0-169-207:675555:675833 [4] NCCL INFO Channel 03/24 :    0   1
+ip-26-0-169-207:675555:675833 [4] NCCL INFO Channel 04/24 :    0   1
+ip-26-0-169-207:675555:675833 [4] NCCL INFO Channel 05/24 :    0   1
+ip-26-0-169-207:675555:675833 [4] NCCL INFO Channel 06/24 :    0   1
+ip-26-0-169-207:675555:675833 [4] NCCL INFO Channel 07/24 :    0   1
+ip-26-0-169-207:675555:675833 [4] NCCL INFO Channel 08/24 :    0   1
+ip-26-0-169-207:675555:675833 [4] NCCL INFO Channel 09/24 :    0   1
+ip-26-0-169-207:675555:675833 [4] NCCL INFO Channel 10/24 :    0   1
+ip-26-0-169-207:675555:675833 [4] NCCL INFO Channel 11/24 :    0   1
+ip-26-0-169-207:675555:675833 [4] NCCL INFO Channel 12/24 :    0   1
+ip-26-0-169-207:675555:675833 [4] NCCL INFO Channel 13/24 :    0   1
+ip-26-0-169-207:675555:675833 [4] NCCL INFO Channel 14/24 :    0   1
+ip-26-0-169-207:675555:675833 [4] NCCL INFO Channel 15/24 :    0   1
+ip-26-0-169-207:675555:675833 [4] NCCL INFO Channel 16/24 :    0   1
+ip-26-0-169-207:675555:675833 [4] NCCL INFO Channel 17/24 :    0   1
+ip-26-0-169-207:675555:675833 [4] NCCL INFO Channel 18/24 :    0   1
+ip-26-0-169-207:675555:675833 [4] NCCL INFO Channel 19/24 :    0   1
+ip-26-0-169-207:675555:675833 [4] NCCL INFO Channel 20/24 :    0   1
+ip-26-0-169-207:675555:675833 [4] NCCL INFO Channel 21/24 :    0   1
+ip-26-0-169-207:675555:675833 [4] NCCL INFO Channel 22/24 :    0   1
+ip-26-0-169-207:675555:675833 [4] NCCL INFO Channel 23/24 :    0   1
+ip-26-0-169-207:675555:675833 [4] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] -1/-1/-1->0->1 [7] -1/-1/-1->0->1 [8] -1/-1/-1->0->1 [9] -1/-1/-1->0->1 [10] -1/-1/-1->0->1 [11] -1/-1/-1->0->1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] -1/-1/-1->0->1 [19] -1/-1/-1->0->1 [20] -1/-1/-1->0->1 [21] -1/-1/-1->0->1 [22] -1/-1/-1->0->1 [23] -1/-1/-1->0->1
+ip-26-0-169-207:675555:675833 [4] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-169-207:675556:675836 [5] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] -1/-1/-1->1->0 [3] -1/-1/-1->1->0 [4] -1/-1/-1->1->0 [5] -1/-1/-1->1->0 [6] 0/-1/-1->1->-1 [7] 0/-1/-1->1->-1 [8] 0/-1/-1->1->-1 [9] 0/-1/-1->1->-1 [10] 0/-1/-1->1->-1 [11] 0/-1/-1->1->-1 [12] -1/-1/-1->1->0 [13] -1/-1/-1->1->0 [14] -1/-1/-1->1->0 [15] -1/-1/-1->1->0 [16] -1/-1/-1->1->0 [17] -1/-1/-1->1->0 [18] 0/-1/-1->1->-1 [19] 0/-1/-1->1->-1 [20] 0/-1/-1->1->-1 [21] 0/-1/-1->1->-1 [22] 0/-1/-1->1->-1 [23] 0/-1/-1->1->-1
+ip-26-0-169-207:675556:675836 [5] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-169-207:675554:675839 [3] NCCL INFO Channel 00/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-169-207:675554:675839 [3] NCCL INFO Channel 01/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-169-207:675552:675837 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675554:675839 [3] NCCL INFO Channel 02/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-169-207:675551:675832 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675552:675837 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675554:675839 [3] NCCL INFO Channel 03/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-169-207:675551:675832 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675552:675837 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675554:675839 [3] NCCL INFO Channel 04/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-169-207:675551:675832 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675552:675837 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675554:675839 [3] NCCL INFO Channel 05/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-169-207:675551:675832 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675557:675834 [6] NCCL INFO Channel 00/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-169-207:675552:675837 [1] NCCL INFO Channel 04/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675554:675839 [3] NCCL INFO Channel 06/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-169-207:675551:675832 [0] NCCL INFO Channel 04/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675552:675837 [1] NCCL INFO Channel 05/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675554:675839 [3] NCCL INFO Channel 07/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-169-207:675551:675832 [0] NCCL INFO Channel 05/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675552:675837 [1] NCCL INFO Channel 06/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675554:675839 [3] NCCL INFO Channel 08/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-169-207:675551:675832 [0] NCCL INFO Channel 06/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675552:675837 [1] NCCL INFO Channel 07/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675554:675839 [3] NCCL INFO Channel 09/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-169-207:675551:675832 [0] NCCL INFO Channel 07/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675552:675837 [1] NCCL INFO Channel 08/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675554:675839 [3] NCCL INFO Channel 10/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-169-207:675558:675838 [7] NCCL INFO Channel 00/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-169-207:675551:675832 [0] NCCL INFO Channel 08/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675555:675833 [4] NCCL INFO Channel 00/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-169-207:675556:675836 [5] NCCL INFO Channel 00/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-169-207:675552:675837 [1] NCCL INFO Channel 09/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675557:675834 [6] NCCL INFO Channel 01/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-169-207:675554:675839 [3] NCCL INFO Channel 11/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-169-207:675558:675838 [7] NCCL INFO Channel 01/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-169-207:675551:675832 [0] NCCL INFO Channel 09/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675553:675835 [2] NCCL INFO Channel 00/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-169-207:675555:675833 [4] NCCL INFO Channel 01/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-169-207:675556:675836 [5] NCCL INFO Channel 01/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-169-207:675557:675834 [6] NCCL INFO Channel 02/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-169-207:675552:675837 [1] NCCL INFO Channel 10/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675558:675838 [7] NCCL INFO Channel 02/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-169-207:675554:675839 [3] NCCL INFO Channel 12/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-169-207:675551:675832 [0] NCCL INFO Channel 10/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675553:675835 [2] NCCL INFO Channel 01/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-169-207:675555:675833 [4] NCCL INFO Channel 02/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-169-207:675556:675836 [5] NCCL INFO Channel 02/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-169-207:675557:675834 [6] NCCL INFO Channel 03/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-169-207:675552:675837 [1] NCCL INFO Channel 11/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675558:675838 [7] NCCL INFO Channel 03/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-169-207:675554:675839 [3] NCCL INFO Channel 13/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-169-207:675551:675832 [0] NCCL INFO Channel 11/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675553:675835 [2] NCCL INFO Channel 02/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-169-207:675555:675833 [4] NCCL INFO Channel 03/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-169-207:675556:675836 [5] NCCL INFO Channel 03/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-169-207:675557:675834 [6] NCCL INFO Channel 04/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-169-207:675552:675837 [1] NCCL INFO Channel 12/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675558:675838 [7] NCCL INFO Channel 04/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-169-207:675554:675839 [3] NCCL INFO Channel 14/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-169-207:675551:675832 [0] NCCL INFO Channel 12/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675553:675835 [2] NCCL INFO Channel 03/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-169-207:675555:675833 [4] NCCL INFO Channel 04/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-169-207:675556:675836 [5] NCCL INFO Channel 04/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-169-207:675552:675837 [1] NCCL INFO Channel 13/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675557:675834 [6] NCCL INFO Channel 05/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-169-207:675558:675838 [7] NCCL INFO Channel 05/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-169-207:675554:675839 [3] NCCL INFO Channel 15/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-169-207:675551:675832 [0] NCCL INFO Channel 13/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675553:675835 [2] NCCL INFO Channel 04/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-169-207:675555:675833 [4] NCCL INFO Channel 05/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-169-207:675556:675836 [5] NCCL INFO Channel 05/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-169-207:675552:675837 [1] NCCL INFO Channel 14/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675557:675834 [6] NCCL INFO Channel 06/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-169-207:675558:675838 [7] NCCL INFO Channel 06/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-169-207:675554:675839 [3] NCCL INFO Channel 16/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-169-207:675551:675832 [0] NCCL INFO Channel 14/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675553:675835 [2] NCCL INFO Channel 05/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-169-207:675555:675833 [4] NCCL INFO Channel 06/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-169-207:675556:675836 [5] NCCL INFO Channel 06/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-169-207:675552:675837 [1] NCCL INFO Channel 15/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675557:675834 [6] NCCL INFO Channel 07/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-169-207:675558:675838 [7] NCCL INFO Channel 07/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-169-207:675554:675839 [3] NCCL INFO Channel 17/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-169-207:675551:675832 [0] NCCL INFO Channel 15/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675553:675835 [2] NCCL INFO Channel 06/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-169-207:675555:675833 [4] NCCL INFO Channel 07/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-169-207:675556:675836 [5] NCCL INFO Channel 07/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-169-207:675552:675837 [1] NCCL INFO Channel 16/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675557:675834 [6] NCCL INFO Channel 08/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-169-207:675558:675838 [7] NCCL INFO Channel 08/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-169-207:675554:675839 [3] NCCL INFO Channel 18/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-169-207:675551:675832 [0] NCCL INFO Channel 16/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675553:675835 [2] NCCL INFO Channel 07/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-169-207:675555:675833 [4] NCCL INFO Channel 08/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-169-207:675556:675836 [5] NCCL INFO Channel 08/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-169-207:675552:675837 [1] NCCL INFO Channel 17/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675557:675834 [6] NCCL INFO Channel 09/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-169-207:675558:675838 [7] NCCL INFO Channel 09/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-169-207:675554:675839 [3] NCCL INFO Channel 19/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-169-207:675551:675832 [0] NCCL INFO Channel 17/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675555:675833 [4] NCCL INFO Channel 09/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-169-207:675556:675836 [5] NCCL INFO Channel 09/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-169-207:675552:675837 [1] NCCL INFO Channel 18/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675553:675835 [2] NCCL INFO Channel 08/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-169-207:675557:675834 [6] NCCL INFO Channel 10/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-169-207:675558:675838 [7] NCCL INFO Channel 10/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-169-207:675554:675839 [3] NCCL INFO Channel 20/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-169-207:675551:675832 [0] NCCL INFO Channel 18/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675555:675833 [4] NCCL INFO Channel 10/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-169-207:675552:675837 [1] NCCL INFO Channel 19/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675556:675836 [5] NCCL INFO Channel 10/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-169-207:675553:675835 [2] NCCL INFO Channel 09/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-169-207:675557:675834 [6] NCCL INFO Channel 11/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-169-207:675558:675838 [7] NCCL INFO Channel 11/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-169-207:675554:675839 [3] NCCL INFO Channel 21/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-169-207:675551:675832 [0] NCCL INFO Channel 19/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675555:675833 [4] NCCL INFO Channel 11/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-169-207:675556:675836 [5] NCCL INFO Channel 11/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-169-207:675552:675837 [1] NCCL INFO Channel 20/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675553:675835 [2] NCCL INFO Channel 10/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-169-207:675557:675834 [6] NCCL INFO Channel 12/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-169-207:675558:675838 [7] NCCL INFO Channel 12/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-169-207:675554:675839 [3] NCCL INFO Channel 22/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-169-207:675551:675832 [0] NCCL INFO Channel 20/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675555:675833 [4] NCCL INFO Channel 12/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-169-207:675556:675836 [5] NCCL INFO Channel 12/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-169-207:675552:675837 [1] NCCL INFO Channel 21/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675553:675835 [2] NCCL INFO Channel 11/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-169-207:675557:675834 [6] NCCL INFO Channel 13/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-169-207:675558:675838 [7] NCCL INFO Channel 13/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-169-207:675556:675836 [5] NCCL INFO Channel 13/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-169-207:675553:675835 [2] NCCL INFO Channel 12/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-169-207:675551:675832 [0] NCCL INFO Channel 21/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675552:675837 [1] NCCL INFO Channel 22/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675556:675836 [5] NCCL INFO Channel 14/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-169-207:675553:675835 [2] NCCL INFO Channel 13/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-169-207:675551:675832 [0] NCCL INFO Channel 22/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675554:675839 [3] NCCL INFO Channel 23/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-169-207:675552:675837 [1] NCCL INFO Channel 23/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675551:675832 [0] NCCL INFO Channel 23/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:675555:675833 [4] NCCL INFO Channel 13/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-169-207:675557:675834 [6] NCCL INFO Channel 14/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-169-207:675555:675833 [4] NCCL INFO Channel 14/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-169-207:675558:675838 [7] NCCL INFO Channel 14/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-169-207:675557:675834 [6] NCCL INFO Channel 15/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-169-207:675555:675833 [4] NCCL INFO Channel 15/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-169-207:675558:675838 [7] NCCL INFO Channel 15/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-169-207:675557:675834 [6] NCCL INFO Channel 16/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-169-207:675556:675836 [5] NCCL INFO Channel 15/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-169-207:675553:675835 [2] NCCL INFO Channel 14/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-169-207:675555:675833 [4] NCCL INFO Channel 16/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-169-207:675558:675838 [7] NCCL INFO Channel 16/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-169-207:675557:675834 [6] NCCL INFO Channel 17/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-169-207:675556:675836 [5] NCCL INFO Channel 16/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-169-207:675553:675835 [2] NCCL INFO Channel 15/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-169-207:675555:675833 [4] NCCL INFO Channel 17/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-169-207:675558:675838 [7] NCCL INFO Channel 17/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-169-207:675557:675834 [6] NCCL INFO Channel 18/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-169-207:675556:675836 [5] NCCL INFO Channel 17/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-169-207:675553:675835 [2] NCCL INFO Channel 16/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-169-207:675558:675838 [7] NCCL INFO Channel 18/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-169-207:675557:675834 [6] NCCL INFO Channel 19/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-169-207:675556:675836 [5] NCCL INFO Channel 18/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-169-207:675553:675835 [2] NCCL INFO Channel 17/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-169-207:675558:675838 [7] NCCL INFO Channel 19/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-169-207:675555:675833 [4] NCCL INFO Channel 18/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-169-207:675557:675834 [6] NCCL INFO Channel 20/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-169-207:675557:675834 [6] NCCL INFO Channel 21/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-169-207:675557:675834 [6] NCCL INFO Channel 22/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-169-207:675556:675836 [5] NCCL INFO Channel 19/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-169-207:675553:675835 [2] NCCL INFO Channel 18/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-169-207:675557:675834 [6] NCCL INFO Channel 23/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-169-207:675555:675833 [4] NCCL INFO Channel 19/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-169-207:675558:675838 [7] NCCL INFO Channel 20/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-169-207:675556:675836 [5] NCCL INFO Channel 20/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-169-207:675553:675835 [2] NCCL INFO Channel 19/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-169-207:675555:675833 [4] NCCL INFO Channel 20/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-169-207:675558:675838 [7] NCCL INFO Channel 21/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-169-207:675556:675836 [5] NCCL INFO Channel 21/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-169-207:675553:675835 [2] NCCL INFO Channel 20/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-169-207:675555:675833 [4] NCCL INFO Channel 21/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-169-207:675558:675838 [7] NCCL INFO Channel 22/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-169-207:675556:675836 [5] NCCL INFO Channel 22/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-169-207:675553:675835 [2] NCCL INFO Channel 21/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-169-207:675555:675833 [4] NCCL INFO Channel 22/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-169-207:675558:675838 [7] NCCL INFO Channel 23/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-169-207:675553:675835 [2] NCCL INFO Channel 22/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-169-207:675556:675836 [5] NCCL INFO Channel 23/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-169-207:675555:675833 [4] NCCL INFO Channel 23/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-169-207:675553:675835 [2] NCCL INFO Channel 23/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-169-207:675551:675832 [0] NCCL INFO Connected all rings
+ip-26-0-169-207:675551:675832 [0] NCCL INFO Connected all trees
+ip-26-0-169-207:675551:675832 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-169-207:675551:675832 [0] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-169-207:675551:675832 [0] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:675552:675837 [1] NCCL INFO Connected all rings
+ip-26-0-169-207:675552:675837 [1] NCCL INFO Connected all trees
+ip-26-0-169-207:675552:675837 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-169-207:675552:675837 [1] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-169-207:675552:675837 [1] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:675551:675832 [0] NCCL INFO comm 0xad5c630 rank 0 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0x43c20e17bdec1db - Init COMPLETE
+ip-26-0-169-207:675552:675837 [1] NCCL INFO comm 0xb464bf0 rank 1 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0x43c20e17bdec1db - Init COMPLETE
+ip-26-0-169-207:675551:675854 [0] NCCL INFO Using network Libfabric
+ip-26-0-169-207:675551:675854 [0] NCCL INFO comm 0xad70c80 rank 0 nranks 1 cudaDev 0 nvmlDev 0 busId 53000 commId 0x71b2c42d6b537046 - Init START
+ip-26-0-169-207:675551:675854 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/162
+NCCL version 2.18.5+cuda12.2
+ip-26-0-169-207:675552:675856 [1] NCCL INFO Using network Libfabric
+ip-26-0-169-207:675552:675856 [1] NCCL INFO comm 0xb478820 rank 0 nranks 1 cudaDev 1 nvmlDev 1 busId 64000 commId 0xf3db9c78f1849ebb - Init START
+ip-26-0-169-207:675552:675856 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/149
+ip-26-0-169-207:675557:675834 [6] NCCL INFO Connected all rings
+ip-26-0-169-207:675557:675834 [6] NCCL INFO Connected all trees
+ip-26-0-169-207:675557:675834 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-169-207:675557:675834 [6] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-169-207:675557:675834 [6] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:675558:675838 [7] NCCL INFO Connected all rings
+ip-26-0-169-207:675558:675838 [7] NCCL INFO Connected all trees
+ip-26-0-169-207:675558:675838 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-169-207:675558:675838 [7] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-169-207:675558:675838 [7] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:675555:675833 [4] NCCL INFO Connected all rings
+ip-26-0-169-207:675555:675833 [4] NCCL INFO Connected all trees
+ip-26-0-169-207:675555:675833 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-169-207:675555:675833 [4] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-169-207:675555:675833 [4] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:675553:675835 [2] NCCL INFO Connected all rings
+ip-26-0-169-207:675553:675835 [2] NCCL INFO Connected all trees
+ip-26-0-169-207:675553:675835 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-169-207:675553:675835 [2] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-169-207:675553:675835 [2] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:675557:675834 [6] NCCL INFO comm 0xbbdf3e0 rank 0 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0x2d4ea52ae5c54be - Init COMPLETE
+ip-26-0-169-207:675558:675838 [7] NCCL INFO comm 0xb5babb0 rank 1 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0x2d4ea52ae5c54be - Init COMPLETE
+ip-26-0-169-207:675556:675836 [5] NCCL INFO Connected all rings
+ip-26-0-169-207:675556:675836 [5] NCCL INFO Connected all trees
+ip-26-0-169-207:675556:675836 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-169-207:675556:675836 [5] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-169-207:675556:675836 [5] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:675554:675839 [3] NCCL INFO Connected all rings
+ip-26-0-169-207:675554:675839 [3] NCCL INFO Connected all trees
+ip-26-0-169-207:675554:675839 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-169-207:675554:675839 [3] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-169-207:675554:675839 [3] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:675557:675864 [6] NCCL INFO Using network Libfabric
+NCCL version 2.18.5+cuda12.2
+ip-26-0-169-207:675557:675864 [6] NCCL INFO comm 0xbbf3600 rank 0 nranks 1 cudaDev 6 nvmlDev 6 busId b9000 commId 0xd7c97b1be77a8035 - Init START
+ip-26-0-169-207:675557:675864 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/149
+ip-26-0-169-207:675558:675866 [7] NCCL INFO Using network Libfabric
+ip-26-0-169-207:675558:675866 [7] NCCL INFO comm 0xb5ce780 rank 0 nranks 1 cudaDev 7 nvmlDev 7 busId ca000 commId 0xb998602d9ce2bcdd - Init START
+ip-26-0-169-207:675558:675866 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/149
+ip-26-0-169-207:675555:675833 [4] NCCL INFO comm 0xb35dff0 rank 0 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0xcdaefaa4cc5cec90 - Init COMPLETE
+ip-26-0-169-207:675556:675836 [5] NCCL INFO comm 0xc1bc3b0 rank 1 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0xcdaefaa4cc5cec90 - Init COMPLETE
+ip-26-0-169-207:675553:675835 [2] NCCL INFO comm 0xa5aafe0 rank 0 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0xa80432eabeb5875c - Init COMPLETE
+ip-26-0-169-207:675554:675839 [3] NCCL INFO comm 0xc404770 rank 1 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0xa80432eabeb5875c - Init COMPLETE
+NCCL version 2.18.5+cuda12.2
+ip-26-0-169-207:675555:675868 [4] NCCL INFO Using network Libfabric
+ip-26-0-169-207:675555:675868 [4] NCCL INFO comm 0xb372270 rank 0 nranks 1 cudaDev 4 nvmlDev 4 busId 97000 commId 0x58429d7c03467c60 - Init START
+ip-26-0-169-207:675555:675868 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/149
+ip-26-0-169-207:675556:675870 [5] NCCL INFO Using network Libfabric
+ip-26-0-169-207:675556:675870 [5] NCCL INFO comm 0xc1cffe0 rank 0 nranks 1 cudaDev 5 nvmlDev 5 busId a8000 commId 0x9d9af66d97ffef4 - Init START
+ip-26-0-169-207:675556:675870 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/149
+NCCL version 2.18.5+cuda12.2
+ip-26-0-169-207:675554:675872 [3] NCCL INFO Using network Libfabric
+ip-26-0-169-207:675554:675872 [3] NCCL INFO comm 0xc4183a0 rank 0 nranks 1 cudaDev 3 nvmlDev 3 busId 86000 commId 0x54f5f31ca8d351e8 - Init START
+ip-26-0-169-207:675554:675872 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/149
+ip-26-0-169-207:675553:675874 [2] NCCL INFO Using network Libfabric
+ip-26-0-169-207:675553:675874 [2] NCCL INFO comm 0xa5bf230 rank 0 nranks 1 cudaDev 2 nvmlDev 2 busId 75000 commId 0x9a3cece7c464fc26 - Init START
+ip-26-0-169-207:675553:675874 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/149
+ip-26-0-169-207:675551:675854 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-169-207:675551:675854 [0] NCCL INFO Channel 00/32 :    0
+ip-26-0-169-207:675551:675854 [0] NCCL INFO Channel 01/32 :    0
+ip-26-0-169-207:675551:675854 [0] NCCL INFO Channel 02/32 :    0
+ip-26-0-169-207:675551:675854 [0] NCCL INFO Channel 03/32 :    0
+ip-26-0-169-207:675551:675854 [0] NCCL INFO Channel 04/32 :    0
+ip-26-0-169-207:675551:675854 [0] NCCL INFO Channel 05/32 :    0
+ip-26-0-169-207:675551:675854 [0] NCCL INFO Channel 06/32 :    0
+ip-26-0-169-207:675551:675854 [0] NCCL INFO Channel 07/32 :    0
+ip-26-0-169-207:675551:675854 [0] NCCL INFO Channel 08/32 :    0
+ip-26-0-169-207:675551:675854 [0] NCCL INFO Channel 09/32 :    0
+ip-26-0-169-207:675551:675854 [0] NCCL INFO Channel 10/32 :    0
+ip-26-0-169-207:675551:675854 [0] NCCL INFO Channel 11/32 :    0
+ip-26-0-169-207:675551:675854 [0] NCCL INFO Channel 12/32 :    0
+ip-26-0-169-207:675551:675854 [0] NCCL INFO Channel 13/32 :    0
+ip-26-0-169-207:675551:675854 [0] NCCL INFO Channel 14/32 :    0
+ip-26-0-169-207:675551:675854 [0] NCCL INFO Channel 15/32 :    0
+ip-26-0-169-207:675551:675854 [0] NCCL INFO Channel 16/32 :    0
+ip-26-0-169-207:675551:675854 [0] NCCL INFO Channel 17/32 :    0
+ip-26-0-169-207:675551:675854 [0] NCCL INFO Channel 18/32 :    0
+ip-26-0-169-207:675551:675854 [0] NCCL INFO Channel 19/32 :    0
+ip-26-0-169-207:675551:675854 [0] NCCL INFO Channel 20/32 :    0
+ip-26-0-169-207:675551:675854 [0] NCCL INFO Channel 21/32 :    0
+ip-26-0-169-207:675551:675854 [0] NCCL INFO Channel 22/32 :    0
+ip-26-0-169-207:675551:675854 [0] NCCL INFO Channel 23/32 :    0
+ip-26-0-169-207:675551:675854 [0] NCCL INFO Channel 24/32 :    0
+ip-26-0-169-207:675551:675854 [0] NCCL INFO Channel 25/32 :    0
+ip-26-0-169-207:675551:675854 [0] NCCL INFO Channel 26/32 :    0
+ip-26-0-169-207:675551:675854 [0] NCCL INFO Channel 27/32 :    0
+ip-26-0-169-207:675551:675854 [0] NCCL INFO Channel 28/32 :    0
+ip-26-0-169-207:675551:675854 [0] NCCL INFO Channel 29/32 :    0
+ip-26-0-169-207:675551:675854 [0] NCCL INFO Channel 30/32 :    0
+ip-26-0-169-207:675551:675854 [0] NCCL INFO Channel 31/32 :    0
+ip-26-0-169-207:675551:675854 [0] NCCL INFO Trees [0] -1/-1/-1->0->-1 [1] -1/-1/-1->0->-1 [2] -1/-1/-1->0->-1 [3] -1/-1/-1->0->-1 [4] -1/-1/-1->0->-1 [5] -1/-1/-1->0->-1 [6] -1/-1/-1->0->-1 [7] -1/-1/-1->0->-1 [8] -1/-1/-1->0->-1 [9] -1/-1/-1->0->-1 [10] -1/-1/-1->0->-1 [11] -1/-1/-1->0->-1 [12] -1/-1/-1->0->-1 [13] -1/-1/-1->0->-1 [14] -1/-1/-1->0->-1 [15] -1/-1/-1->0->-1 [16] -1/-1/-1->0->-1 [17] -1/-1/-1->0->-1 [18] -1/-1/-1->0->-1 [19] -1/-1/-1->0->-1 [20] -1/-1/-1->0->-1 [21] -1/-1/-1->0->-1 [22] -1/-1/-1->0->-1 [23] -1/-1/-1->0->-1 [24] -1/-1/-1->0->-1 [25] -1/-1/-1->0->-1 [26] -1/-1/-1->0->-1 [27] -1/-1/-1->0->-1 [28] -1/-1/-1->0->-1 [29] -1/-1/-1->0->-1 [30] -1/-1/-1->0->-1 [31] -1/-1/-1->0->-1
+ip-26-0-169-207:675551:675854 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-169-207:675551:675854 [0] NCCL INFO Connected all rings
+ip-26-0-169-207:675551:675854 [0] NCCL INFO Connected all trees
+ip-26-0-169-207:675551:675854 [0] NCCL INFO 32 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:675552:675856 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-169-207:675552:675856 [1] NCCL INFO Channel 00/32 :    0
+ip-26-0-169-207:675552:675856 [1] NCCL INFO Channel 01/32 :    0
+ip-26-0-169-207:675552:675856 [1] NCCL INFO Channel 02/32 :    0
+ip-26-0-169-207:675552:675856 [1] NCCL INFO Channel 03/32 :    0
+ip-26-0-169-207:675552:675856 [1] NCCL INFO Channel 04/32 :    0
+ip-26-0-169-207:675552:675856 [1] NCCL INFO Channel 05/32 :    0
+ip-26-0-169-207:675552:675856 [1] NCCL INFO Channel 06/32 :    0
+ip-26-0-169-207:675552:675856 [1] NCCL INFO Channel 07/32 :    0
+ip-26-0-169-207:675552:675856 [1] NCCL INFO Channel 08/32 :    0
+ip-26-0-169-207:675552:675856 [1] NCCL INFO Channel 09/32 :    0
+ip-26-0-169-207:675552:675856 [1] NCCL INFO Channel 10/32 :    0
+ip-26-0-169-207:675552:675856 [1] NCCL INFO Channel 11/32 :    0
+ip-26-0-169-207:675552:675856 [1] NCCL INFO Channel 12/32 :    0
+ip-26-0-169-207:675552:675856 [1] NCCL INFO Channel 13/32 :    0
+ip-26-0-169-207:675552:675856 [1] NCCL INFO Channel 14/32 :    0
+ip-26-0-169-207:675552:675856 [1] NCCL INFO Channel 15/32 :    0
+ip-26-0-169-207:675552:675856 [1] NCCL INFO Channel 16/32 :    0
+ip-26-0-169-207:675552:675856 [1] NCCL INFO Channel 17/32 :    0
+ip-26-0-169-207:675552:675856 [1] NCCL INFO Channel 18/32 :    0
+ip-26-0-169-207:675552:675856 [1] NCCL INFO Channel 19/32 :    0
+ip-26-0-169-207:675552:675856 [1] NCCL INFO Channel 20/32 :    0
+ip-26-0-169-207:675552:675856 [1] NCCL INFO Channel 21/32 :    0
+ip-26-0-169-207:675552:675856 [1] NCCL INFO Channel 22/32 :    0
+ip-26-0-169-207:675552:675856 [1] NCCL INFO Channel 23/32 :    0
+ip-26-0-169-207:675552:675856 [1] NCCL INFO Channel 24/32 :    0
+ip-26-0-169-207:675552:675856 [1] NCCL INFO Channel 25/32 :    0
+ip-26-0-169-207:675552:675856 [1] NCCL INFO Channel 26/32 :    0
+ip-26-0-169-207:675552:675856 [1] NCCL INFO Channel 27/32 :    0
+ip-26-0-169-207:675552:675856 [1] NCCL INFO Channel 28/32 :    0
+ip-26-0-169-207:675552:675856 [1] NCCL INFO Channel 29/32 :    0
+ip-26-0-169-207:675552:675856 [1] NCCL INFO Channel 30/32 :    0
+ip-26-0-169-207:675552:675856 [1] NCCL INFO Channel 31/32 :    0
+ip-26-0-169-207:675552:675856 [1] NCCL INFO Trees [0] -1/-1/-1->0->-1 [1] -1/-1/-1->0->-1 [2] -1/-1/-1->0->-1 [3] -1/-1/-1->0->-1 [4] -1/-1/-1->0->-1 [5] -1/-1/-1->0->-1 [6] -1/-1/-1->0->-1 [7] -1/-1/-1->0->-1 [8] -1/-1/-1->0->-1 [9] -1/-1/-1->0->-1 [10] -1/-1/-1->0->-1 [11] -1/-1/-1->0->-1 [12] -1/-1/-1->0->-1 [13] -1/-1/-1->0->-1 [14] -1/-1/-1->0->-1 [15] -1/-1/-1->0->-1 [16] -1/-1/-1->0->-1 [17] -1/-1/-1->0->-1 [18] -1/-1/-1->0->-1 [19] -1/-1/-1->0->-1 [20] -1/-1/-1->0->-1 [21] -1/-1/-1->0->-1 [22] -1/-1/-1->0->-1 [23] -1/-1/-1->0->-1 [24] -1/-1/-1->0->-1 [25] -1/-1/-1->0->-1 [26] -1/-1/-1->0->-1 [27] -1/-1/-1->0->-1 [28] -1/-1/-1->0->-1 [29] -1/-1/-1->0->-1 [30] -1/-1/-1->0->-1 [31] -1/-1/-1->0->-1
+ip-26-0-169-207:675552:675856 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-169-207:675552:675856 [1] NCCL INFO Connected all rings
+ip-26-0-169-207:675552:675856 [1] NCCL INFO Connected all trees
+ip-26-0-169-207:675552:675856 [1] NCCL INFO 32 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:675552:675856 [1] NCCL INFO comm 0xb478820 rank 0 nranks 1 cudaDev 1 nvmlDev 1 busId 64000 commId 0xf3db9c78f1849ebb - Init COMPLETE
+ip-26-0-169-207:675551:675854 [0] NCCL INFO comm 0xad70c80 rank 0 nranks 1 cudaDev 0 nvmlDev 0 busId 53000 commId 0x71b2c42d6b537046 - Init COMPLETE
+12/13/2024 12:22:40 [INFO|DP=0|PP=0|TP=1|ip-26-0-169-207]: Local number of parameters: 520M (992.13MiB)
+12/13/2024 12:22:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]: Total number of parameters: 1.04G (1984.26MiB)
+12/13/2024 12:22:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]: Local number of parameters: 520M (992.13MiB)
+12/13/2024 12:22:40 [INFO|DP=0|PP=0|TP=1|ip-26-0-169-207]: [After model building] Memory usage: 992.15MiB. Peak allocated: 5632.00MiB Peak reserved: 16898.00MiB
+12/13/2024 12:22:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]: [After model building] Memory usage: 992.15MiB. Peak allocated: 5632.00MiB Peak reserved: 16898.00MiB
+12/13/2024 12:22:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]: No checkpoint path provided.
+12/13/2024 12:22:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]: Parametrizing model parameters using StandardParametrizator
+ip-26-0-169-207:675551:675881 [0] NCCL INFO Using network Libfabric
+ip-26-0-169-207:675552:675883 [1] NCCL INFO Using network Libfabric
+ip-26-0-169-207:675557:675864 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-169-207:675557:675864 [6] NCCL INFO Channel 00/32 :    0
+ip-26-0-169-207:675557:675864 [6] NCCL INFO Channel 01/32 :    0
+ip-26-0-169-207:675557:675864 [6] NCCL INFO Channel 02/32 :    0
+ip-26-0-169-207:675557:675864 [6] NCCL INFO Channel 03/32 :    0
+ip-26-0-169-207:675557:675864 [6] NCCL INFO Channel 04/32 :    0
+ip-26-0-169-207:675557:675864 [6] NCCL INFO Channel 05/32 :    0
+ip-26-0-169-207:675557:675864 [6] NCCL INFO Channel 06/32 :    0
+ip-26-0-169-207:675557:675864 [6] NCCL INFO Channel 07/32 :    0
+ip-26-0-169-207:675557:675864 [6] NCCL INFO Channel 08/32 :    0
+ip-26-0-169-207:675557:675864 [6] NCCL INFO Channel 09/32 :    0
+ip-26-0-169-207:675557:675864 [6] NCCL INFO Channel 10/32 :    0
+ip-26-0-169-207:675557:675864 [6] NCCL INFO Channel 11/32 :    0
+ip-26-0-169-207:675557:675864 [6] NCCL INFO Channel 12/32 :    0
+ip-26-0-169-207:675557:675864 [6] NCCL INFO Channel 13/32 :    0
+ip-26-0-169-207:675557:675864 [6] NCCL INFO Channel 14/32 :    0
+ip-26-0-169-207:675557:675864 [6] NCCL INFO Channel 15/32 :    0
+ip-26-0-169-207:675557:675864 [6] NCCL INFO Channel 16/32 :    0
+ip-26-0-169-207:675557:675864 [6] NCCL INFO Channel 17/32 :    0
+ip-26-0-169-207:675557:675864 [6] NCCL INFO Channel 18/32 :    0
+ip-26-0-169-207:675557:675864 [6] NCCL INFO Channel 19/32 :    0
+ip-26-0-169-207:675557:675864 [6] NCCL INFO Channel 20/32 :    0
+ip-26-0-169-207:675557:675864 [6] NCCL INFO Channel 21/32 :    0
+ip-26-0-169-207:675557:675864 [6] NCCL INFO Channel 22/32 :    0
+ip-26-0-169-207:675557:675864 [6] NCCL INFO Channel 23/32 :    0
+ip-26-0-169-207:675557:675864 [6] NCCL INFO Channel 24/32 :    0
+ip-26-0-169-207:675557:675864 [6] NCCL INFO Channel 25/32 :    0
+ip-26-0-169-207:675557:675864 [6] NCCL INFO Channel 26/32 :    0
+ip-26-0-169-207:675557:675864 [6] NCCL INFO Channel 27/32 :    0
+ip-26-0-169-207:675557:675864 [6] NCCL INFO Channel 28/32 :    0
+ip-26-0-169-207:675557:675864 [6] NCCL INFO Channel 29/32 :    0
+ip-26-0-169-207:675557:675864 [6] NCCL INFO Channel 30/32 :    0
+ip-26-0-169-207:675557:675864 [6] NCCL INFO Channel 31/32 :    0
+ip-26-0-169-207:675557:675864 [6] NCCL INFO Trees [0] -1/-1/-1->0->-1 [1] -1/-1/-1->0->-1 [2] -1/-1/-1->0->-1 [3] -1/-1/-1->0->-1 [4] -1/-1/-1->0->-1 [5] -1/-1/-1->0->-1 [6] -1/-1/-1->0->-1 [7] -1/-1/-1->0->-1 [8] -1/-1/-1->0->-1 [9] -1/-1/-1->0->-1 [10] -1/-1/-1->0->-1 [11] -1/-1/-1->0->-1 [12] -1/-1/-1->0->-1 [13] -1/-1/-1->0->-1 [14] -1/-1/-1->0->-1 [15] -1/-1/-1->0->-1 [16] -1/-1/-1->0->-1 [17] -1/-1/-1->0->-1 [18] -1/-1/-1->0->-1 [19] -1/-1/-1->0->-1 [20] -1/-1/-1->0->-1 [21] -1/-1/-1->0->-1 [22] -1/-1/-1->0->-1 [23] -1/-1/-1->0->-1 [24] -1/-1/-1->0->-1 [25] -1/-1/-1->0->-1 [26] -1/-1/-1->0->-1 [27] -1/-1/-1->0->-1 [28] -1/-1/-1->0->-1 [29] -1/-1/-1->0->-1 [30] -1/-1/-1->0->-1 [31] -1/-1/-1->0->-1
+ip-26-0-169-207:675557:675864 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-169-207:675557:675864 [6] NCCL INFO Connected all rings
+ip-26-0-169-207:675557:675864 [6] NCCL INFO Connected all trees
+ip-26-0-169-207:675557:675864 [6] NCCL INFO 32 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:675555:675868 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-169-207:675555:675868 [4] NCCL INFO Channel 00/32 :    0
+ip-26-0-169-207:675555:675868 [4] NCCL INFO Channel 01/32 :    0
+ip-26-0-169-207:675555:675868 [4] NCCL INFO Channel 02/32 :    0
+ip-26-0-169-207:675555:675868 [4] NCCL INFO Channel 03/32 :    0
+ip-26-0-169-207:675555:675868 [4] NCCL INFO Channel 04/32 :    0
+ip-26-0-169-207:675555:675868 [4] NCCL INFO Channel 05/32 :    0
+ip-26-0-169-207:675555:675868 [4] NCCL INFO Channel 06/32 :    0
+ip-26-0-169-207:675555:675868 [4] NCCL INFO Channel 07/32 :    0
+ip-26-0-169-207:675555:675868 [4] NCCL INFO Channel 08/32 :    0
+ip-26-0-169-207:675555:675868 [4] NCCL INFO Channel 09/32 :    0
+ip-26-0-169-207:675555:675868 [4] NCCL INFO Channel 10/32 :    0
+ip-26-0-169-207:675555:675868 [4] NCCL INFO Channel 11/32 :    0
+ip-26-0-169-207:675555:675868 [4] NCCL INFO Channel 12/32 :    0
+ip-26-0-169-207:675555:675868 [4] NCCL INFO Channel 13/32 :    0
+ip-26-0-169-207:675555:675868 [4] NCCL INFO Channel 14/32 :    0
+ip-26-0-169-207:675555:675868 [4] NCCL INFO Channel 15/32 :    0
+ip-26-0-169-207:675555:675868 [4] NCCL INFO Channel 16/32 :    0
+ip-26-0-169-207:675555:675868 [4] NCCL INFO Channel 17/32 :    0
+ip-26-0-169-207:675555:675868 [4] NCCL INFO Channel 18/32 :    0
+ip-26-0-169-207:675555:675868 [4] NCCL INFO Channel 19/32 :    0
+ip-26-0-169-207:675555:675868 [4] NCCL INFO Channel 20/32 :    0
+ip-26-0-169-207:675555:675868 [4] NCCL INFO Channel 21/32 :    0
+ip-26-0-169-207:675555:675868 [4] NCCL INFO Channel 22/32 :    0
+ip-26-0-169-207:675555:675868 [4] NCCL INFO Channel 23/32 :    0
+ip-26-0-169-207:675555:675868 [4] NCCL INFO Channel 24/32 :    0
+ip-26-0-169-207:675555:675868 [4] NCCL INFO Channel 25/32 :    0
+ip-26-0-169-207:675555:675868 [4] NCCL INFO Channel 26/32 :    0
+ip-26-0-169-207:675555:675868 [4] NCCL INFO Channel 27/32 :    0
+ip-26-0-169-207:675555:675868 [4] NCCL INFO Channel 28/32 :    0
+ip-26-0-169-207:675555:675868 [4] NCCL INFO Channel 29/32 :    0
+ip-26-0-169-207:675555:675868 [4] NCCL INFO Channel 30/32 :    0
+ip-26-0-169-207:675555:675868 [4] NCCL INFO Channel 31/32 :    0
+ip-26-0-169-207:675555:675868 [4] NCCL INFO Trees [0] -1/-1/-1->0->-1 [1] -1/-1/-1->0->-1 [2] -1/-1/-1->0->-1 [3] -1/-1/-1->0->-1 [4] -1/-1/-1->0->-1 [5] -1/-1/-1->0->-1 [6] -1/-1/-1->0->-1 [7] -1/-1/-1->0->-1 [8] -1/-1/-1->0->-1 [9] -1/-1/-1->0->-1 [10] -1/-1/-1->0->-1 [11] -1/-1/-1->0->-1 [12] -1/-1/-1->0->-1 [13] -1/-1/-1->0->-1 [14] -1/-1/-1->0->-1 [15] -1/-1/-1->0->-1 [16] -1/-1/-1->0->-1 [17] -1/-1/-1->0->-1 [18] -1/-1/-1->0->-1 [19] -1/-1/-1->0->-1 [20] -1/-1/-1->0->-1 [21] -1/-1/-1->0->-1 [22] -1/-1/-1->0->-1 [23] -1/-1/-1->0->-1 [24] -1/-1/-1->0->-1 [25] -1/-1/-1->0->-1 [26] -1/-1/-1->0->-1 [27] -1/-1/-1->0->-1 [28] -1/-1/-1->0->-1 [29] -1/-1/-1->0->-1 [30] -1/-1/-1->0->-1 [31] -1/-1/-1->0->-1
+ip-26-0-169-207:675555:675868 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-169-207:675555:675868 [4] NCCL INFO Connected all rings
+ip-26-0-169-207:675555:675868 [4] NCCL INFO Connected all trees
+ip-26-0-169-207:675555:675868 [4] NCCL INFO 32 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:675553:675874 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-169-207:675553:675874 [2] NCCL INFO Channel 00/32 :    0
+ip-26-0-169-207:675553:675874 [2] NCCL INFO Channel 01/32 :    0
+ip-26-0-169-207:675553:675874 [2] NCCL INFO Channel 02/32 :    0
+ip-26-0-169-207:675553:675874 [2] NCCL INFO Channel 03/32 :    0
+ip-26-0-169-207:675553:675874 [2] NCCL INFO Channel 04/32 :    0
+ip-26-0-169-207:675553:675874 [2] NCCL INFO Channel 05/32 :    0
+ip-26-0-169-207:675553:675874 [2] NCCL INFO Channel 06/32 :    0
+ip-26-0-169-207:675553:675874 [2] NCCL INFO Channel 07/32 :    0
+ip-26-0-169-207:675553:675874 [2] NCCL INFO Channel 08/32 :    0
+ip-26-0-169-207:675553:675874 [2] NCCL INFO Channel 09/32 :    0
+ip-26-0-169-207:675553:675874 [2] NCCL INFO Channel 10/32 :    0
+ip-26-0-169-207:675553:675874 [2] NCCL INFO Channel 11/32 :    0
+ip-26-0-169-207:675553:675874 [2] NCCL INFO Channel 12/32 :    0
+ip-26-0-169-207:675553:675874 [2] NCCL INFO Channel 13/32 :    0
+ip-26-0-169-207:675553:675874 [2] NCCL INFO Channel 14/32 :    0
+ip-26-0-169-207:675553:675874 [2] NCCL INFO Channel 15/32 :    0
+ip-26-0-169-207:675553:675874 [2] NCCL INFO Channel 16/32 :    0
+ip-26-0-169-207:675553:675874 [2] NCCL INFO Channel 17/32 :    0
+ip-26-0-169-207:675553:675874 [2] NCCL INFO Channel 18/32 :    0
+ip-26-0-169-207:675553:675874 [2] NCCL INFO Channel 19/32 :    0
+ip-26-0-169-207:675553:675874 [2] NCCL INFO Channel 20/32 :    0
+ip-26-0-169-207:675553:675874 [2] NCCL INFO Channel 21/32 :    0
+ip-26-0-169-207:675553:675874 [2] NCCL INFO Channel 22/32 :    0
+ip-26-0-169-207:675553:675874 [2] NCCL INFO Channel 23/32 :    0
+ip-26-0-169-207:675553:675874 [2] NCCL INFO Channel 24/32 :    0
+ip-26-0-169-207:675553:675874 [2] NCCL INFO Channel 25/32 :    0
+ip-26-0-169-207:675553:675874 [2] NCCL INFO Channel 26/32 :    0
+ip-26-0-169-207:675553:675874 [2] NCCL INFO Channel 27/32 :    0
+ip-26-0-169-207:675553:675874 [2] NCCL INFO Channel 28/32 :    0
+ip-26-0-169-207:675553:675874 [2] NCCL INFO Channel 29/32 :    0
+ip-26-0-169-207:675553:675874 [2] NCCL INFO Channel 30/32 :    0
+ip-26-0-169-207:675553:675874 [2] NCCL INFO Channel 31/32 :    0
+ip-26-0-169-207:675553:675874 [2] NCCL INFO Trees [0] -1/-1/-1->0->-1 [1] -1/-1/-1->0->-1 [2] -1/-1/-1->0->-1 [3] -1/-1/-1->0->-1 [4] -1/-1/-1->0->-1 [5] -1/-1/-1->0->-1 [6] -1/-1/-1->0->-1 [7] -1/-1/-1->0->-1 [8] -1/-1/-1->0->-1 [9] -1/-1/-1->0->-1 [10] -1/-1/-1->0->-1 [11] -1/-1/-1->0->-1 [12] -1/-1/-1->0->-1 [13] -1/-1/-1->0->-1 [14] -1/-1/-1->0->-1 [15] -1/-1/-1->0->-1 [16] -1/-1/-1->0->-1 [17] -1/-1/-1->0->-1 [18] -1/-1/-1->0->-1 [19] -1/-1/-1->0->-1 [20] -1/-1/-1->0->-1 [21] -1/-1/-1->0->-1 [22] -1/-1/-1->0->-1 [23] -1/-1/-1->0->-1 [24] -1/-1/-1->0->-1 [25] -1/-1/-1->0->-1 [26] -1/-1/-1->0->-1 [27] -1/-1/-1->0->-1 [28] -1/-1/-1->0->-1 [29] -1/-1/-1->0->-1 [30] -1/-1/-1->0->-1 [31] -1/-1/-1->0->-1
+ip-26-0-169-207:675553:675874 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-169-207:675553:675874 [2] NCCL INFO Connected all rings
+ip-26-0-169-207:675553:675874 [2] NCCL INFO Connected all trees
+ip-26-0-169-207:675553:675874 [2] NCCL INFO 32 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:675558:675866 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-169-207:675558:675866 [7] NCCL INFO Channel 00/32 :    0
+ip-26-0-169-207:675558:675866 [7] NCCL INFO Channel 01/32 :    0
+ip-26-0-169-207:675558:675866 [7] NCCL INFO Channel 02/32 :    0
+ip-26-0-169-207:675558:675866 [7] NCCL INFO Channel 03/32 :    0
+ip-26-0-169-207:675558:675866 [7] NCCL INFO Channel 04/32 :    0
+ip-26-0-169-207:675558:675866 [7] NCCL INFO Channel 05/32 :    0
+ip-26-0-169-207:675558:675866 [7] NCCL INFO Channel 06/32 :    0
+ip-26-0-169-207:675558:675866 [7] NCCL INFO Channel 07/32 :    0
+ip-26-0-169-207:675558:675866 [7] NCCL INFO Channel 08/32 :    0
+ip-26-0-169-207:675558:675866 [7] NCCL INFO Channel 09/32 :    0
+ip-26-0-169-207:675558:675866 [7] NCCL INFO Channel 10/32 :    0
+ip-26-0-169-207:675558:675866 [7] NCCL INFO Channel 11/32 :    0
+ip-26-0-169-207:675558:675866 [7] NCCL INFO Channel 12/32 :    0
+ip-26-0-169-207:675558:675866 [7] NCCL INFO Channel 13/32 :    0
+ip-26-0-169-207:675558:675866 [7] NCCL INFO Channel 14/32 :    0
+ip-26-0-169-207:675558:675866 [7] NCCL INFO Channel 15/32 :    0
+ip-26-0-169-207:675558:675866 [7] NCCL INFO Channel 16/32 :    0
+ip-26-0-169-207:675558:675866 [7] NCCL INFO Channel 17/32 :    0
+ip-26-0-169-207:675558:675866 [7] NCCL INFO Channel 18/32 :    0
+ip-26-0-169-207:675558:675866 [7] NCCL INFO Channel 19/32 :    0
+ip-26-0-169-207:675558:675866 [7] NCCL INFO Channel 20/32 :    0
+ip-26-0-169-207:675558:675866 [7] NCCL INFO Channel 21/32 :    0
+ip-26-0-169-207:675558:675866 [7] NCCL INFO Channel 22/32 :    0
+ip-26-0-169-207:675558:675866 [7] NCCL INFO Channel 23/32 :    0
+ip-26-0-169-207:675558:675866 [7] NCCL INFO Channel 24/32 :    0
+ip-26-0-169-207:675558:675866 [7] NCCL INFO Channel 25/32 :    0
+ip-26-0-169-207:675558:675866 [7] NCCL INFO Channel 26/32 :    0
+ip-26-0-169-207:675558:675866 [7] NCCL INFO Channel 27/32 :    0
+ip-26-0-169-207:675558:675866 [7] NCCL INFO Channel 28/32 :    0
+ip-26-0-169-207:675558:675866 [7] NCCL INFO Channel 29/32 :    0
+ip-26-0-169-207:675558:675866 [7] NCCL INFO Channel 30/32 :    0
+ip-26-0-169-207:675558:675866 [7] NCCL INFO Channel 31/32 :    0
+ip-26-0-169-207:675558:675866 [7] NCCL INFO Trees [0] -1/-1/-1->0->-1 [1] -1/-1/-1->0->-1 [2] -1/-1/-1->0->-1 [3] -1/-1/-1->0->-1 [4] -1/-1/-1->0->-1 [5] -1/-1/-1->0->-1 [6] -1/-1/-1->0->-1 [7] -1/-1/-1->0->-1 [8] -1/-1/-1->0->-1 [9] -1/-1/-1->0->-1 [10] -1/-1/-1->0->-1 [11] -1/-1/-1->0->-1 [12] -1/-1/-1->0->-1 [13] -1/-1/-1->0->-1 [14] -1/-1/-1->0->-1 [15] -1/-1/-1->0->-1 [16] -1/-1/-1->0->-1 [17] -1/-1/-1->0->-1 [18] -1/-1/-1->0->-1 [19] -1/-1/-1->0->-1 [20] -1/-1/-1->0->-1 [21] -1/-1/-1->0->-1 [22] -1/-1/-1->0->-1 [23] -1/-1/-1->0->-1 [24] -1/-1/-1->0->-1 [25] -1/-1/-1->0->-1 [26] -1/-1/-1->0->-1 [27] -1/-1/-1->0->-1 [28] -1/-1/-1->0->-1 [29] -1/-1/-1->0->-1 [30] -1/-1/-1->0->-1 [31] -1/-1/-1->0->-1
+ip-26-0-169-207:675558:675866 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-169-207:675558:675866 [7] NCCL INFO Connected all rings
+ip-26-0-169-207:675558:675866 [7] NCCL INFO Connected all trees
+ip-26-0-169-207:675558:675866 [7] NCCL INFO 32 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:675557:675864 [6] NCCL INFO comm 0xbbf3600 rank 0 nranks 1 cudaDev 6 nvmlDev 6 busId b9000 commId 0xd7c97b1be77a8035 - Init COMPLETE
+ip-26-0-169-207:675554:675872 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-169-207:675554:675872 [3] NCCL INFO Channel 00/32 :    0
+ip-26-0-169-207:675554:675872 [3] NCCL INFO Channel 01/32 :    0
+ip-26-0-169-207:675554:675872 [3] NCCL INFO Channel 02/32 :    0
+ip-26-0-169-207:675554:675872 [3] NCCL INFO Channel 03/32 :    0
+ip-26-0-169-207:675554:675872 [3] NCCL INFO Channel 04/32 :    0
+ip-26-0-169-207:675554:675872 [3] NCCL INFO Channel 05/32 :    0
+ip-26-0-169-207:675554:675872 [3] NCCL INFO Channel 06/32 :    0
+ip-26-0-169-207:675554:675872 [3] NCCL INFO Channel 07/32 :    0
+ip-26-0-169-207:675554:675872 [3] NCCL INFO Channel 08/32 :    0
+ip-26-0-169-207:675554:675872 [3] NCCL INFO Channel 09/32 :    0
+ip-26-0-169-207:675554:675872 [3] NCCL INFO Channel 10/32 :    0
+ip-26-0-169-207:675554:675872 [3] NCCL INFO Channel 11/32 :    0
+ip-26-0-169-207:675554:675872 [3] NCCL INFO Channel 12/32 :    0
+ip-26-0-169-207:675554:675872 [3] NCCL INFO Channel 13/32 :    0
+ip-26-0-169-207:675554:675872 [3] NCCL INFO Channel 14/32 :    0
+ip-26-0-169-207:675554:675872 [3] NCCL INFO Channel 15/32 :    0
+ip-26-0-169-207:675554:675872 [3] NCCL INFO Channel 16/32 :    0
+ip-26-0-169-207:675554:675872 [3] NCCL INFO Channel 17/32 :    0
+ip-26-0-169-207:675554:675872 [3] NCCL INFO Channel 18/32 :    0
+ip-26-0-169-207:675554:675872 [3] NCCL INFO Channel 19/32 :    0
+ip-26-0-169-207:675554:675872 [3] NCCL INFO Channel 20/32 :    0
+ip-26-0-169-207:675554:675872 [3] NCCL INFO Channel 21/32 :    0
+ip-26-0-169-207:675554:675872 [3] NCCL INFO Channel 22/32 :    0
+ip-26-0-169-207:675554:675872 [3] NCCL INFO Channel 23/32 :    0
+ip-26-0-169-207:675554:675872 [3] NCCL INFO Channel 24/32 :    0
+ip-26-0-169-207:675554:675872 [3] NCCL INFO Channel 25/32 :    0
+ip-26-0-169-207:675554:675872 [3] NCCL INFO Channel 26/32 :    0
+ip-26-0-169-207:675554:675872 [3] NCCL INFO Channel 27/32 :    0
+ip-26-0-169-207:675554:675872 [3] NCCL INFO Channel 28/32 :    0
+ip-26-0-169-207:675554:675872 [3] NCCL INFO Channel 29/32 :    0
+ip-26-0-169-207:675554:675872 [3] NCCL INFO Channel 30/32 :    0
+ip-26-0-169-207:675554:675872 [3] NCCL INFO Channel 31/32 :    0
+ip-26-0-169-207:675554:675872 [3] NCCL INFO Trees [0] -1/-1/-1->0->-1 [1] -1/-1/-1->0->-1 [2] -1/-1/-1->0->-1 [3] -1/-1/-1->0->-1 [4] -1/-1/-1->0->-1 [5] -1/-1/-1->0->-1 [6] -1/-1/-1->0->-1 [7] -1/-1/-1->0->-1 [8] -1/-1/-1->0->-1 [9] -1/-1/-1->0->-1 [10] -1/-1/-1->0->-1 [11] -1/-1/-1->0->-1 [12] -1/-1/-1->0->-1 [13] -1/-1/-1->0->-1 [14] -1/-1/-1->0->-1 [15] -1/-1/-1->0->-1 [16] -1/-1/-1->0->-1 [17] -1/-1/-1->0->-1 [18] -1/-1/-1->0->-1 [19] -1/-1/-1->0->-1 [20] -1/-1/-1->0->-1 [21] -1/-1/-1->0->-1 [22] -1/-1/-1->0->-1 [23] -1/-1/-1->0->-1 [24] -1/-1/-1->0->-1 [25] -1/-1/-1->0->-1 [26] -1/-1/-1->0->-1 [27] -1/-1/-1->0->-1 [28] -1/-1/-1->0->-1 [29] -1/-1/-1->0->-1 [30] -1/-1/-1->0->-1 [31] -1/-1/-1->0->-1
+ip-26-0-169-207:675554:675872 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-169-207:675554:675872 [3] NCCL INFO Connected all rings
+ip-26-0-169-207:675554:675872 [3] NCCL INFO Connected all trees
+ip-26-0-169-207:675554:675872 [3] NCCL INFO 32 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:675556:675870 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-169-207:675556:675870 [5] NCCL INFO Channel 00/32 :    0
+ip-26-0-169-207:675556:675870 [5] NCCL INFO Channel 01/32 :    0
+ip-26-0-169-207:675556:675870 [5] NCCL INFO Channel 02/32 :    0
+ip-26-0-169-207:675556:675870 [5] NCCL INFO Channel 03/32 :    0
+ip-26-0-169-207:675556:675870 [5] NCCL INFO Channel 04/32 :    0
+ip-26-0-169-207:675556:675870 [5] NCCL INFO Channel 05/32 :    0
+ip-26-0-169-207:675556:675870 [5] NCCL INFO Channel 06/32 :    0
+ip-26-0-169-207:675556:675870 [5] NCCL INFO Channel 07/32 :    0
+ip-26-0-169-207:675556:675870 [5] NCCL INFO Channel 08/32 :    0
+ip-26-0-169-207:675556:675870 [5] NCCL INFO Channel 09/32 :    0
+ip-26-0-169-207:675556:675870 [5] NCCL INFO Channel 10/32 :    0
+ip-26-0-169-207:675556:675870 [5] NCCL INFO Channel 11/32 :    0
+ip-26-0-169-207:675556:675870 [5] NCCL INFO Channel 12/32 :    0
+ip-26-0-169-207:675556:675870 [5] NCCL INFO Channel 13/32 :    0
+ip-26-0-169-207:675556:675870 [5] NCCL INFO Channel 14/32 :    0
+ip-26-0-169-207:675556:675870 [5] NCCL INFO Channel 15/32 :    0
+ip-26-0-169-207:675556:675870 [5] NCCL INFO Channel 16/32 :    0
+ip-26-0-169-207:675556:675870 [5] NCCL INFO Channel 17/32 :    0
+ip-26-0-169-207:675556:675870 [5] NCCL INFO Channel 18/32 :    0
+ip-26-0-169-207:675556:675870 [5] NCCL INFO Channel 19/32 :    0
+ip-26-0-169-207:675556:675870 [5] NCCL INFO Channel 20/32 :    0
+ip-26-0-169-207:675556:675870 [5] NCCL INFO Channel 21/32 :    0
+ip-26-0-169-207:675556:675870 [5] NCCL INFO Channel 22/32 :    0
+ip-26-0-169-207:675556:675870 [5] NCCL INFO Channel 23/32 :    0
+ip-26-0-169-207:675556:675870 [5] NCCL INFO Channel 24/32 :    0
+ip-26-0-169-207:675556:675870 [5] NCCL INFO Channel 25/32 :    0
+ip-26-0-169-207:675556:675870 [5] NCCL INFO Channel 26/32 :    0
+ip-26-0-169-207:675556:675870 [5] NCCL INFO Channel 27/32 :    0
+ip-26-0-169-207:675556:675870 [5] NCCL INFO Channel 28/32 :    0
+ip-26-0-169-207:675556:675870 [5] NCCL INFO Channel 29/32 :    0
+ip-26-0-169-207:675556:675870 [5] NCCL INFO Channel 30/32 :    0
+ip-26-0-169-207:675556:675870 [5] NCCL INFO Channel 31/32 :    0
+ip-26-0-169-207:675556:675870 [5] NCCL INFO Trees [0] -1/-1/-1->0->-1 [1] -1/-1/-1->0->-1 [2] -1/-1/-1->0->-1 [3] -1/-1/-1->0->-1 [4] -1/-1/-1->0->-1 [5] -1/-1/-1->0->-1 [6] -1/-1/-1->0->-1 [7] -1/-1/-1->0->-1 [8] -1/-1/-1->0->-1 [9] -1/-1/-1->0->-1 [10] -1/-1/-1->0->-1 [11] -1/-1/-1->0->-1 [12] -1/-1/-1->0->-1 [13] -1/-1/-1->0->-1 [14] -1/-1/-1->0->-1 [15] -1/-1/-1->0->-1 [16] -1/-1/-1->0->-1 [17] -1/-1/-1->0->-1 [18] -1/-1/-1->0->-1 [19] -1/-1/-1->0->-1 [20] -1/-1/-1->0->-1 [21] -1/-1/-1->0->-1 [22] -1/-1/-1->0->-1 [23] -1/-1/-1->0->-1 [24] -1/-1/-1->0->-1 [25] -1/-1/-1->0->-1 [26] -1/-1/-1->0->-1 [27] -1/-1/-1->0->-1 [28] -1/-1/-1->0->-1 [29] -1/-1/-1->0->-1 [30] -1/-1/-1->0->-1 [31] -1/-1/-1->0->-1
+ip-26-0-169-207:675556:675870 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-169-207:675555:675868 [4] NCCL INFO comm 0xb372270 rank 0 nranks 1 cudaDev 4 nvmlDev 4 busId 97000 commId 0x58429d7c03467c60 - Init COMPLETE
+ip-26-0-169-207:675556:675870 [5] NCCL INFO Connected all rings
+ip-26-0-169-207:675556:675870 [5] NCCL INFO Connected all trees
+ip-26-0-169-207:675556:675870 [5] NCCL INFO 32 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:675553:675874 [2] NCCL INFO comm 0xa5bf230 rank 0 nranks 1 cudaDev 2 nvmlDev 2 busId 75000 commId 0x9a3cece7c464fc26 - Init COMPLETE
+ip-26-0-169-207:675558:675866 [7] NCCL INFO comm 0xb5ce780 rank 0 nranks 1 cudaDev 7 nvmlDev 7 busId ca000 commId 0xb998602d9ce2bcdd - Init COMPLETE
+ip-26-0-169-207:675554:675872 [3] NCCL INFO comm 0xc4183a0 rank 0 nranks 1 cudaDev 3 nvmlDev 3 busId 86000 commId 0x54f5f31ca8d351e8 - Init COMPLETE
+ip-26-0-169-207:675556:675870 [5] NCCL INFO comm 0xc1cffe0 rank 0 nranks 1 cudaDev 5 nvmlDev 5 busId a8000 commId 0x9d9af66d97ffef4 - Init COMPLETE
+ip-26-0-169-207:675557:675896 [6] NCCL INFO Using network Libfabric
+ip-26-0-169-207:675558:675897 [7] NCCL INFO Using network Libfabric
+ip-26-0-169-207:675554:675898 [3] NCCL INFO Using network Libfabric
+ip-26-0-169-207:675553:675899 [2] NCCL INFO Using network Libfabric
+ip-26-0-169-207:675555:675900 [4] NCCL INFO Using network Libfabric
+ip-26-0-169-207:675555:675900 [4] NCCL INFO comm 0xb376d10 rank 2 nranks 4 cudaDev 4 nvmlDev 4 busId 97000 commId 0xed6922ce1d2e3935 - Init START
+ip-26-0-169-207:675553:675899 [2] NCCL INFO comm 0xa5c4150 rank 1 nranks 4 cudaDev 2 nvmlDev 2 busId 75000 commId 0xed6922ce1d2e3935 - Init START
+ip-26-0-169-207:675557:675896 [6] NCCL INFO comm 0xbbf8710 rank 3 nranks 4 cudaDev 6 nvmlDev 6 busId b9000 commId 0xed6922ce1d2e3935 - Init START
+ip-26-0-169-207:675551:675881 [0] NCCL INFO comm 0xad76160 rank 0 nranks 4 cudaDev 0 nvmlDev 0 busId 53000 commId 0xed6922ce1d2e3935 - Init START
+ip-26-0-169-207:675557:675896 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/149
+ip-26-0-169-207:675553:675899 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/149
+ip-26-0-169-207:675551:675881 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/162
+ip-26-0-169-207:675556:675901 [5] NCCL INFO Using network Libfabric
+ip-26-0-169-207:675555:675900 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/149
+ip-26-0-169-207:675554:675898 [3] NCCL INFO comm 0xc41d030 rank 1 nranks 4 cudaDev 3 nvmlDev 3 busId 86000 commId 0x8e3037642664602d - Init START
+ip-26-0-169-207:675556:675901 [5] NCCL INFO comm 0xc1d51c0 rank 2 nranks 4 cudaDev 5 nvmlDev 5 busId a8000 commId 0x8e3037642664602d - Init START
+ip-26-0-169-207:675558:675897 [7] NCCL INFO comm 0xb5d3930 rank 3 nranks 4 cudaDev 7 nvmlDev 7 busId ca000 commId 0x8e3037642664602d - Init START
+ip-26-0-169-207:675552:675883 [1] NCCL INFO comm 0xb47d9b0 rank 0 nranks 4 cudaDev 1 nvmlDev 1 busId 64000 commId 0x8e3037642664602d - Init START
+ip-26-0-169-207:675558:675897 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/149
+ip-26-0-169-207:675556:675901 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/149
+ip-26-0-169-207:675554:675898 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/149
+ip-26-0-169-207:675552:675883 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/149
+ip-26-0-169-207:675557:675896 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-169-207:675557:675896 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-169-207:675555:675900 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-169-207:675555:675900 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-169-207:675558:675897 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-169-207:675558:675897 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-169-207:675551:675881 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-169-207:675551:675881 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-169-207:675553:675899 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-169-207:675553:675899 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-169-207:675551:675881 [0] NCCL INFO Channel 00/24 :    0   1   2   3
+ip-26-0-169-207:675553:675899 [2] NCCL INFO Trees [0] 2/-1/-1->1->0 [1] 2/-1/-1->1->0 [2] 2/-1/-1->1->0 [3] 2/-1/-1->1->0 [4] 2/-1/-1->1->0 [5] 2/-1/-1->1->0 [6] 2/-1/-1->1->0 [7] 2/-1/-1->1->0 [8] 2/-1/-1->1->0 [9] 2/-1/-1->1->0 [10] 2/-1/-1->1->0 [11] 2/-1/-1->1->0 [12] 2/-1/-1->1->0 [13] 2/-1/-1->1->0 [14] 2/-1/-1->1->0 [15] 2/-1/-1->1->0 [16] 2/-1/-1->1->0 [17] 2/-1/-1->1->0 [18] 2/-1/-1->1->0 [19] 2/-1/-1->1->0 [20] 2/-1/-1->1->0 [21] 2/-1/-1->1->0 [22] 2/-1/-1->1->0 [23] 2/-1/-1->1->0
+ip-26-0-169-207:675555:675900 [4] NCCL INFO Trees [0] 3/-1/-1->2->1 [1] 3/-1/-1->2->1 [2] 3/-1/-1->2->1 [3] 3/-1/-1->2->1 [4] 3/-1/-1->2->1 [5] 3/-1/-1->2->1 [6] 3/-1/-1->2->1 [7] 3/-1/-1->2->1 [8] 3/-1/-1->2->1 [9] 3/-1/-1->2->1 [10] 3/-1/-1->2->1 [11] 3/-1/-1->2->1 [12] 3/-1/-1->2->1 [13] 3/-1/-1->2->1 [14] 3/-1/-1->2->1 [15] 3/-1/-1->2->1 [16] 3/-1/-1->2->1 [17] 3/-1/-1->2->1 [18] 3/-1/-1->2->1 [19] 3/-1/-1->2->1 [20] 3/-1/-1->2->1 [21] 3/-1/-1->2->1 [22] 3/-1/-1->2->1 [23] 3/-1/-1->2->1
+ip-26-0-169-207:675557:675896 [6] NCCL INFO Trees [0] -1/-1/-1->3->2 [1] -1/-1/-1->3->2 [2] -1/-1/-1->3->2 [3] -1/-1/-1->3->2 [4] -1/-1/-1->3->2 [5] -1/-1/-1->3->2 [6] -1/-1/-1->3->2 [7] -1/-1/-1->3->2 [8] -1/-1/-1->3->2 [9] -1/-1/-1->3->2 [10] -1/-1/-1->3->2 [11] -1/-1/-1->3->2 [12] -1/-1/-1->3->2 [13] -1/-1/-1->3->2 [14] -1/-1/-1->3->2 [15] -1/-1/-1->3->2 [16] -1/-1/-1->3->2 [17] -1/-1/-1->3->2 [18] -1/-1/-1->3->2 [19] -1/-1/-1->3->2 [20] -1/-1/-1->3->2 [21] -1/-1/-1->3->2 [22] -1/-1/-1->3->2 [23] -1/-1/-1->3->2
+ip-26-0-169-207:675551:675881 [0] NCCL INFO Channel 01/24 :    0   1   2   3
+ip-26-0-169-207:675553:675899 [2] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-169-207:675557:675896 [6] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-169-207:675555:675900 [4] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-169-207:675551:675881 [0] NCCL INFO Channel 02/24 :    0   1   2   3
+ip-26-0-169-207:675551:675881 [0] NCCL INFO Channel 03/24 :    0   1   2   3
+ip-26-0-169-207:675551:675881 [0] NCCL INFO Channel 04/24 :    0   1   2   3
+ip-26-0-169-207:675551:675881 [0] NCCL INFO Channel 05/24 :    0   1   2   3
+ip-26-0-169-207:675551:675881 [0] NCCL INFO Channel 06/24 :    0   1   2   3
+ip-26-0-169-207:675551:675881 [0] NCCL INFO Channel 07/24 :    0   1   2   3
+ip-26-0-169-207:675551:675881 [0] NCCL INFO Channel 08/24 :    0   1   2   3
+ip-26-0-169-207:675551:675881 [0] NCCL INFO Channel 09/24 :    0   1   2   3
+ip-26-0-169-207:675551:675881 [0] NCCL INFO Channel 10/24 :    0   1   2   3
+ip-26-0-169-207:675551:675881 [0] NCCL INFO Channel 11/24 :    0   1   2   3
+ip-26-0-169-207:675551:675881 [0] NCCL INFO Channel 12/24 :    0   1   2   3
+ip-26-0-169-207:675551:675881 [0] NCCL INFO Channel 13/24 :    0   1   2   3
+ip-26-0-169-207:675551:675881 [0] NCCL INFO Channel 14/24 :    0   1   2   3
+ip-26-0-169-207:675551:675881 [0] NCCL INFO Channel 15/24 :    0   1   2   3
+ip-26-0-169-207:675551:675881 [0] NCCL INFO Channel 16/24 :    0   1   2   3
+ip-26-0-169-207:675551:675881 [0] NCCL INFO Channel 17/24 :    0   1   2   3
+ip-26-0-169-207:675551:675881 [0] NCCL INFO Channel 18/24 :    0   1   2   3
+ip-26-0-169-207:675551:675881 [0] NCCL INFO Channel 19/24 :    0   1   2   3
+ip-26-0-169-207:675551:675881 [0] NCCL INFO Channel 20/24 :    0   1   2   3
+ip-26-0-169-207:675551:675881 [0] NCCL INFO Channel 21/24 :    0   1   2   3
+ip-26-0-169-207:675551:675881 [0] NCCL INFO Channel 22/24 :    0   1   2   3
+ip-26-0-169-207:675551:675881 [0] NCCL INFO Channel 23/24 :    0   1   2   3
+ip-26-0-169-207:675551:675881 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] 1/-1/-1->0->-1 [7] 1/-1/-1->0->-1 [8] 1/-1/-1->0->-1 [9] 1/-1/-1->0->-1 [10] 1/-1/-1->0->-1 [11] 1/-1/-1->0->-1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] 1/-1/-1->0->-1 [19] 1/-1/-1->0->-1 [20] 1/-1/-1->0->-1 [21] 1/-1/-1->0->-1 [22] 1/-1/-1->0->-1 [23] 1/-1/-1->0->-1
+ip-26-0-169-207:675551:675881 [0] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-169-207:675554:675898 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-169-207:675554:675898 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-169-207:675556:675901 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-169-207:675556:675901 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-169-207:675552:675883 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-169-207:675552:675883 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-169-207:675553:675899 [2] NCCL INFO Channel 00/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-169-207:675551:675881 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-169-207:675553:675899 [2] NCCL INFO Channel 01/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-169-207:675555:675900 [4] NCCL INFO Channel 00/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-169-207:675555:675900 [4] NCCL INFO Channel 01/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-169-207:675553:675899 [2] NCCL INFO Channel 02/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-169-207:675551:675881 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-169-207:675555:675900 [4] NCCL INFO Channel 02/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-169-207:675553:675899 [2] NCCL INFO Channel 03/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-169-207:675551:675881 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-169-207:675555:675900 [4] NCCL INFO Channel 03/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-169-207:675553:675899 [2] NCCL INFO Channel 04/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-169-207:675551:675881 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-169-207:675557:675896 [6] NCCL INFO Channel 00/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675555:675900 [4] NCCL INFO Channel 04/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-169-207:675553:675899 [2] NCCL INFO Channel 05/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-169-207:675551:675881 [0] NCCL INFO Channel 04/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-169-207:675557:675896 [6] NCCL INFO Channel 01/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675555:675900 [4] NCCL INFO Channel 05/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-169-207:675553:675899 [2] NCCL INFO Channel 06/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-169-207:675551:675881 [0] NCCL INFO Channel 05/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-169-207:675557:675896 [6] NCCL INFO Channel 02/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675555:675900 [4] NCCL INFO Channel 06/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-169-207:675553:675899 [2] NCCL INFO Channel 07/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-169-207:675551:675881 [0] NCCL INFO Channel 06/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-169-207:675555:675900 [4] NCCL INFO Channel 07/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-169-207:675553:675899 [2] NCCL INFO Channel 08/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-169-207:675557:675896 [6] NCCL INFO Channel 03/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675551:675881 [0] NCCL INFO Channel 07/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-169-207:675555:675900 [4] NCCL INFO Channel 08/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-169-207:675553:675899 [2] NCCL INFO Channel 09/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-169-207:675557:675896 [6] NCCL INFO Channel 04/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675551:675881 [0] NCCL INFO Channel 08/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-169-207:675552:675883 [1] NCCL INFO Channel 00/24 :    0   1   2   3
+ip-26-0-169-207:675552:675883 [1] NCCL INFO Channel 01/24 :    0   1   2   3
+ip-26-0-169-207:675552:675883 [1] NCCL INFO Channel 02/24 :    0   1   2   3
+ip-26-0-169-207:675552:675883 [1] NCCL INFO Channel 03/24 :    0   1   2   3
+ip-26-0-169-207:675552:675883 [1] NCCL INFO Channel 04/24 :    0   1   2   3
+ip-26-0-169-207:675552:675883 [1] NCCL INFO Channel 05/24 :    0   1   2   3
+ip-26-0-169-207:675552:675883 [1] NCCL INFO Channel 06/24 :    0   1   2   3
+ip-26-0-169-207:675552:675883 [1] NCCL INFO Channel 07/24 :    0   1   2   3
+ip-26-0-169-207:675552:675883 [1] NCCL INFO Channel 08/24 :    0   1   2   3
+ip-26-0-169-207:675552:675883 [1] NCCL INFO Channel 09/24 :    0   1   2   3
+ip-26-0-169-207:675558:675897 [7] NCCL INFO Trees [0] -1/-1/-1->3->2 [1] -1/-1/-1->3->2 [2] -1/-1/-1->3->2 [3] -1/-1/-1->3->2 [4] -1/-1/-1->3->2 [5] -1/-1/-1->3->2 [6] -1/-1/-1->3->2 [7] -1/-1/-1->3->2 [8] -1/-1/-1->3->2 [9] -1/-1/-1->3->2 [10] -1/-1/-1->3->2 [11] -1/-1/-1->3->2 [12] -1/-1/-1->3->2 [13] -1/-1/-1->3->2 [14] -1/-1/-1->3->2 [15] -1/-1/-1->3->2 [16] -1/-1/-1->3->2 [17] -1/-1/-1->3->2 [18] -1/-1/-1->3->2 [19] -1/-1/-1->3->2 [20] -1/-1/-1->3->2 [21] -1/-1/-1->3->2 [22] -1/-1/-1->3->2 [23] -1/-1/-1->3->2
+ip-26-0-169-207:675552:675883 [1] NCCL INFO Channel 10/24 :    0   1   2   3
+ip-26-0-169-207:675552:675883 [1] NCCL INFO Channel 11/24 :    0   1   2   3
+ip-26-0-169-207:675558:675897 [7] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-169-207:675552:675883 [1] NCCL INFO Channel 12/24 :    0   1   2   3
+ip-26-0-169-207:675552:675883 [1] NCCL INFO Channel 13/24 :    0   1   2   3
+ip-26-0-169-207:675552:675883 [1] NCCL INFO Channel 14/24 :    0   1   2   3
+ip-26-0-169-207:675552:675883 [1] NCCL INFO Channel 15/24 :    0   1   2   3
+ip-26-0-169-207:675552:675883 [1] NCCL INFO Channel 16/24 :    0   1   2   3
+ip-26-0-169-207:675552:675883 [1] NCCL INFO Channel 17/24 :    0   1   2   3
+ip-26-0-169-207:675552:675883 [1] NCCL INFO Channel 18/24 :    0   1   2   3
+ip-26-0-169-207:675552:675883 [1] NCCL INFO Channel 19/24 :    0   1   2   3
+ip-26-0-169-207:675552:675883 [1] NCCL INFO Channel 20/24 :    0   1   2   3
+ip-26-0-169-207:675556:675901 [5] NCCL INFO Trees [0] 3/-1/-1->2->1 [1] 3/-1/-1->2->1 [2] 3/-1/-1->2->1 [3] 3/-1/-1->2->1 [4] 3/-1/-1->2->1 [5] 3/-1/-1->2->1 [6] 3/-1/-1->2->1 [7] 3/-1/-1->2->1 [8] 3/-1/-1->2->1 [9] 3/-1/-1->2->1 [10] 3/-1/-1->2->1 [11] 3/-1/-1->2->1 [12] 3/-1/-1->2->1 [13] 3/-1/-1->2->1 [14] 3/-1/-1->2->1 [15] 3/-1/-1->2->1 [16] 3/-1/-1->2->1 [17] 3/-1/-1->2->1 [18] 3/-1/-1->2->1 [19] 3/-1/-1->2->1 [20] 3/-1/-1->2->1 [21] 3/-1/-1->2->1 [22] 3/-1/-1->2->1 [23] 3/-1/-1->2->1
+ip-26-0-169-207:675552:675883 [1] NCCL INFO Channel 21/24 :    0   1   2   3
+ip-26-0-169-207:675554:675898 [3] NCCL INFO Trees [0] 2/-1/-1->1->0 [1] 2/-1/-1->1->0 [2] 2/-1/-1->1->0 [3] 2/-1/-1->1->0 [4] 2/-1/-1->1->0 [5] 2/-1/-1->1->0 [6] 2/-1/-1->1->0 [7] 2/-1/-1->1->0 [8] 2/-1/-1->1->0 [9] 2/-1/-1->1->0 [10] 2/-1/-1->1->0 [11] 2/-1/-1->1->0 [12] 2/-1/-1->1->0 [13] 2/-1/-1->1->0 [14] 2/-1/-1->1->0 [15] 2/-1/-1->1->0 [16] 2/-1/-1->1->0 [17] 2/-1/-1->1->0 [18] 2/-1/-1->1->0 [19] 2/-1/-1->1->0 [20] 2/-1/-1->1->0 [21] 2/-1/-1->1->0 [22] 2/-1/-1->1->0 [23] 2/-1/-1->1->0
+ip-26-0-169-207:675552:675883 [1] NCCL INFO Channel 22/24 :    0   1   2   3
+ip-26-0-169-207:675556:675901 [5] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-169-207:675552:675883 [1] NCCL INFO Channel 23/24 :    0   1   2   3
+ip-26-0-169-207:675554:675898 [3] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-169-207:675552:675883 [1] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] 1/-1/-1->0->-1 [7] 1/-1/-1->0->-1 [8] 1/-1/-1->0->-1 [9] 1/-1/-1->0->-1 [10] 1/-1/-1->0->-1 [11] 1/-1/-1->0->-1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] 1/-1/-1->0->-1 [19] 1/-1/-1->0->-1 [20] 1/-1/-1->0->-1 [21] 1/-1/-1->0->-1 [22] 1/-1/-1->0->-1 [23] 1/-1/-1->0->-1
+ip-26-0-169-207:675552:675883 [1] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-169-207:675555:675900 [4] NCCL INFO Channel 09/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-169-207:675553:675899 [2] NCCL INFO Channel 10/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-169-207:675557:675896 [6] NCCL INFO Channel 05/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675551:675881 [0] NCCL INFO Channel 09/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-169-207:675555:675900 [4] NCCL INFO Channel 10/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-169-207:675553:675899 [2] NCCL INFO Channel 11/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-169-207:675557:675896 [6] NCCL INFO Channel 06/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675551:675881 [0] NCCL INFO Channel 10/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-169-207:675553:675899 [2] NCCL INFO Channel 12/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-169-207:675555:675900 [4] NCCL INFO Channel 11/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-169-207:675551:675881 [0] NCCL INFO Channel 11/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-169-207:675557:675896 [6] NCCL INFO Channel 07/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675551:675881 [0] NCCL INFO Channel 12/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-169-207:675555:675900 [4] NCCL INFO Channel 12/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-169-207:675557:675896 [6] NCCL INFO Channel 08/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675551:675881 [0] NCCL INFO Channel 13/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-169-207:675555:675900 [4] NCCL INFO Channel 13/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-169-207:675557:675896 [6] NCCL INFO Channel 09/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675555:675900 [4] NCCL INFO Channel 14/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-169-207:675551:675881 [0] NCCL INFO Channel 14/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-169-207:675557:675896 [6] NCCL INFO Channel 10/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675555:675900 [4] NCCL INFO Channel 15/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-169-207:675551:675881 [0] NCCL INFO Channel 15/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-169-207:675557:675896 [6] NCCL INFO Channel 11/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675551:675881 [0] NCCL INFO Channel 16/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-169-207:675555:675900 [4] NCCL INFO Channel 16/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-169-207:675557:675896 [6] NCCL INFO Channel 12/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675551:675881 [0] NCCL INFO Channel 17/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-169-207:675555:675900 [4] NCCL INFO Channel 17/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-169-207:675557:675896 [6] NCCL INFO Channel 13/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675555:675900 [4] NCCL INFO Channel 18/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-169-207:675551:675881 [0] NCCL INFO Channel 18/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-169-207:675553:675899 [2] NCCL INFO Channel 13/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-169-207:675557:675896 [6] NCCL INFO Channel 14/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675555:675900 [4] NCCL INFO Channel 19/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-169-207:675551:675881 [0] NCCL INFO Channel 19/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-169-207:675553:675899 [2] NCCL INFO Channel 14/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-169-207:675557:675896 [6] NCCL INFO Channel 15/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675555:675900 [4] NCCL INFO Channel 20/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-169-207:675553:675899 [2] NCCL INFO Channel 15/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-169-207:675551:675881 [0] NCCL INFO Channel 20/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-169-207:675557:675896 [6] NCCL INFO Channel 16/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675555:675900 [4] NCCL INFO Channel 21/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-169-207:675553:675899 [2] NCCL INFO Channel 16/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-169-207:675557:675896 [6] NCCL INFO Channel 17/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675551:675881 [0] NCCL INFO Channel 21/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-169-207:675555:675900 [4] NCCL INFO Channel 22/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-169-207:675553:675899 [2] NCCL INFO Channel 17/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-169-207:675557:675896 [6] NCCL INFO Channel 18/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675551:675881 [0] NCCL INFO Channel 22/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-169-207:675555:675900 [4] NCCL INFO Channel 23/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-169-207:675553:675899 [2] NCCL INFO Channel 18/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-169-207:675557:675896 [6] NCCL INFO Channel 19/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675551:675881 [0] NCCL INFO Channel 23/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-169-207:675553:675899 [2] NCCL INFO Channel 19/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-169-207:675557:675896 [6] NCCL INFO Channel 20/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675553:675899 [2] NCCL INFO Channel 20/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-169-207:675557:675896 [6] NCCL INFO Channel 21/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675557:675896 [6] NCCL INFO Channel 22/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675553:675899 [2] NCCL INFO Channel 21/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-169-207:675557:675896 [6] NCCL INFO Channel 23/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675553:675899 [2] NCCL INFO Channel 22/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-169-207:675553:675899 [2] NCCL INFO Channel 23/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-169-207:675558:675897 [7] NCCL INFO Channel 00/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-169-207:675552:675883 [1] NCCL INFO Channel 00/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-169-207:675552:675883 [1] NCCL INFO Channel 01/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-169-207:675558:675897 [7] NCCL INFO Channel 01/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-169-207:675552:675883 [1] NCCL INFO Channel 02/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-169-207:675558:675897 [7] NCCL INFO Channel 02/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-169-207:675552:675883 [1] NCCL INFO Channel 03/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-169-207:675556:675901 [5] NCCL INFO Channel 00/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-169-207:675558:675897 [7] NCCL INFO Channel 03/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-169-207:675552:675883 [1] NCCL INFO Channel 04/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-169-207:675556:675901 [5] NCCL INFO Channel 01/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-169-207:675558:675897 [7] NCCL INFO Channel 04/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-169-207:675552:675883 [1] NCCL INFO Channel 05/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-169-207:675552:675883 [1] NCCL INFO Channel 06/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-169-207:675556:675901 [5] NCCL INFO Channel 02/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-169-207:675556:675901 [5] NCCL INFO Channel 03/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-169-207:675552:675883 [1] NCCL INFO Channel 07/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-169-207:675558:675897 [7] NCCL INFO Channel 05/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-169-207:675556:675901 [5] NCCL INFO Channel 04/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-169-207:675556:675901 [5] NCCL INFO Channel 05/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-169-207:675552:675883 [1] NCCL INFO Channel 08/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-169-207:675558:675897 [7] NCCL INFO Channel 06/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-169-207:675556:675901 [5] NCCL INFO Channel 06/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-169-207:675552:675883 [1] NCCL INFO Channel 09/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-169-207:675558:675897 [7] NCCL INFO Channel 07/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-169-207:675554:675898 [3] NCCL INFO Channel 00/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-169-207:675552:675883 [1] NCCL INFO Channel 10/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-169-207:675556:675901 [5] NCCL INFO Channel 07/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-169-207:675558:675897 [7] NCCL INFO Channel 08/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-169-207:675554:675898 [3] NCCL INFO Channel 01/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-169-207:675552:675883 [1] NCCL INFO Channel 11/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-169-207:675556:675901 [5] NCCL INFO Channel 08/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-169-207:675554:675898 [3] NCCL INFO Channel 02/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-169-207:675558:675897 [7] NCCL INFO Channel 09/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-169-207:675552:675883 [1] NCCL INFO Channel 12/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-169-207:675556:675901 [5] NCCL INFO Channel 09/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-169-207:675554:675898 [3] NCCL INFO Channel 03/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-169-207:675558:675897 [7] NCCL INFO Channel 10/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-169-207:675552:675883 [1] NCCL INFO Channel 13/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-169-207:675556:675901 [5] NCCL INFO Channel 10/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-169-207:675554:675898 [3] NCCL INFO Channel 04/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-169-207:675558:675897 [7] NCCL INFO Channel 11/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-169-207:675552:675883 [1] NCCL INFO Channel 14/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-169-207:675556:675901 [5] NCCL INFO Channel 11/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-169-207:675552:675883 [1] NCCL INFO Channel 15/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-169-207:675558:675897 [7] NCCL INFO Channel 12/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-169-207:675552:675883 [1] NCCL INFO Channel 16/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-169-207:675554:675898 [3] NCCL INFO Channel 05/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-169-207:675558:675897 [7] NCCL INFO Channel 13/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-169-207:675552:675883 [1] NCCL INFO Channel 17/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-169-207:675556:675901 [5] NCCL INFO Channel 12/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-169-207:675552:675883 [1] NCCL INFO Channel 18/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-169-207:675556:675901 [5] NCCL INFO Channel 13/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-169-207:675552:675883 [1] NCCL INFO Channel 19/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-169-207:675556:675901 [5] NCCL INFO Channel 14/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-169-207:675554:675898 [3] NCCL INFO Channel 06/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-169-207:675554:675898 [3] NCCL INFO Channel 07/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-169-207:675552:675883 [1] NCCL INFO Channel 20/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-169-207:675558:675897 [7] NCCL INFO Channel 14/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-169-207:675554:675898 [3] NCCL INFO Channel 08/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-169-207:675558:675897 [7] NCCL INFO Channel 15/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-169-207:675554:675898 [3] NCCL INFO Channel 09/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-169-207:675558:675897 [7] NCCL INFO Channel 16/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-169-207:675556:675901 [5] NCCL INFO Channel 15/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-169-207:675553:675899 [2] NCCL INFO Connected all rings
+ip-26-0-169-207:675554:675898 [3] NCCL INFO Channel 10/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-169-207:675552:675883 [1] NCCL INFO Channel 21/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-169-207:675556:675901 [5] NCCL INFO Channel 16/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-169-207:675558:675897 [7] NCCL INFO Channel 17/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-169-207:675556:675901 [5] NCCL INFO Channel 17/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-169-207:675554:675898 [3] NCCL INFO Channel 11/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-169-207:675556:675901 [5] NCCL INFO Channel 18/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-169-207:675552:675883 [1] NCCL INFO Channel 22/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-169-207:675558:675897 [7] NCCL INFO Channel 18/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-169-207:675554:675898 [3] NCCL INFO Channel 12/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-169-207:675557:675896 [6] NCCL INFO Connected all rings
+ip-26-0-169-207:675557:675896 [6] NCCL INFO Channel 00/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-169-207:675554:675898 [3] NCCL INFO Channel 13/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-169-207:675556:675901 [5] NCCL INFO Channel 19/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-169-207:675552:675883 [1] NCCL INFO Channel 23/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-169-207:675557:675896 [6] NCCL INFO Channel 01/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-169-207:675558:675897 [7] NCCL INFO Channel 19/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-169-207:675554:675898 [3] NCCL INFO Channel 14/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-169-207:675557:675896 [6] NCCL INFO Channel 02/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-169-207:675558:675897 [7] NCCL INFO Channel 20/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-169-207:675554:675898 [3] NCCL INFO Channel 15/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-169-207:675557:675896 [6] NCCL INFO Channel 03/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-169-207:675556:675901 [5] NCCL INFO Channel 20/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-169-207:675558:675897 [7] NCCL INFO Channel 21/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-169-207:675557:675896 [6] NCCL INFO Channel 04/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-169-207:675554:675898 [3] NCCL INFO Channel 16/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-169-207:675555:675900 [4] NCCL INFO Connected all rings
+ip-26-0-169-207:675557:675896 [6] NCCL INFO Channel 05/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-169-207:675556:675901 [5] NCCL INFO Channel 21/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-169-207:675558:675897 [7] NCCL INFO Channel 22/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-169-207:675554:675898 [3] NCCL INFO Channel 17/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-169-207:675557:675896 [6] NCCL INFO Channel 06/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-169-207:675556:675901 [5] NCCL INFO Channel 22/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-169-207:675557:675896 [6] NCCL INFO Channel 07/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-169-207:675556:675901 [5] NCCL INFO Channel 23/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-169-207:675551:675881 [0] NCCL INFO Connected all rings
+ip-26-0-169-207:675557:675896 [6] NCCL INFO Channel 08/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-169-207:675557:675896 [6] NCCL INFO Channel 09/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-169-207:675557:675896 [6] NCCL INFO Channel 10/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-169-207:675557:675896 [6] NCCL INFO Channel 11/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-169-207:675558:675897 [7] NCCL INFO Channel 23/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-169-207:675554:675898 [3] NCCL INFO Channel 18/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-169-207:675557:675896 [6] NCCL INFO Channel 12/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-169-207:675554:675898 [3] NCCL INFO Channel 19/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-169-207:675554:675898 [3] NCCL INFO Channel 20/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-169-207:675554:675898 [3] NCCL INFO Channel 21/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-169-207:675557:675896 [6] NCCL INFO Channel 13/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-169-207:675554:675898 [3] NCCL INFO Channel 22/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-169-207:675557:675896 [6] NCCL INFO Channel 14/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-169-207:675554:675898 [3] NCCL INFO Channel 23/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-169-207:675557:675896 [6] NCCL INFO Channel 15/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-169-207:675557:675896 [6] NCCL INFO Channel 16/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-169-207:675557:675896 [6] NCCL INFO Channel 17/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-169-207:675557:675896 [6] NCCL INFO Channel 18/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-169-207:675557:675896 [6] NCCL INFO Channel 19/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-169-207:675557:675896 [6] NCCL INFO Channel 20/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-169-207:675557:675896 [6] NCCL INFO Channel 21/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-169-207:675553:675899 [2] NCCL INFO Channel 00/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675557:675896 [6] NCCL INFO Channel 22/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-169-207:675557:675896 [6] NCCL INFO Channel 23/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-169-207:675553:675899 [2] NCCL INFO Channel 01/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675553:675899 [2] NCCL INFO Channel 02/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675553:675899 [2] NCCL INFO Channel 03/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675553:675899 [2] NCCL INFO Channel 04/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675555:675900 [4] NCCL INFO Channel 00/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-169-207:675553:675899 [2] NCCL INFO Channel 05/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675555:675900 [4] NCCL INFO Channel 01/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-169-207:675553:675899 [2] NCCL INFO Channel 06/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675555:675900 [4] NCCL INFO Channel 02/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-169-207:675553:675899 [2] NCCL INFO Channel 07/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675553:675899 [2] NCCL INFO Channel 08/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675555:675900 [4] NCCL INFO Channel 03/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-169-207:675553:675899 [2] NCCL INFO Channel 09/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675555:675900 [4] NCCL INFO Channel 04/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-169-207:675553:675899 [2] NCCL INFO Channel 10/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675555:675900 [4] NCCL INFO Channel 05/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-169-207:675553:675899 [2] NCCL INFO Channel 11/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675555:675900 [4] NCCL INFO Channel 06/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-169-207:675553:675899 [2] NCCL INFO Channel 12/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675555:675900 [4] NCCL INFO Channel 07/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-169-207:675553:675899 [2] NCCL INFO Channel 13/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675555:675900 [4] NCCL INFO Channel 08/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-169-207:675553:675899 [2] NCCL INFO Channel 14/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675555:675900 [4] NCCL INFO Channel 09/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-169-207:675553:675899 [2] NCCL INFO Channel 15/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675555:675900 [4] NCCL INFO Channel 10/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-169-207:675553:675899 [2] NCCL INFO Channel 16/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675552:675883 [1] NCCL INFO Connected all rings
+ip-26-0-169-207:675555:675900 [4] NCCL INFO Channel 11/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-169-207:675553:675899 [2] NCCL INFO Channel 17/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675555:675900 [4] NCCL INFO Channel 12/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-169-207:675555:675900 [4] NCCL INFO Channel 13/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-169-207:675555:675900 [4] NCCL INFO Channel 14/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-169-207:675553:675899 [2] NCCL INFO Channel 18/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675554:675898 [3] NCCL INFO Connected all rings
+ip-26-0-169-207:675555:675900 [4] NCCL INFO Channel 15/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-169-207:675553:675899 [2] NCCL INFO Channel 19/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675555:675900 [4] NCCL INFO Channel 16/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-169-207:675556:675901 [5] NCCL INFO Connected all rings
+ip-26-0-169-207:675558:675897 [7] NCCL INFO Connected all rings
+ip-26-0-169-207:675558:675897 [7] NCCL INFO Channel 00/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-169-207:675553:675899 [2] NCCL INFO Channel 20/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675558:675897 [7] NCCL INFO Channel 01/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-169-207:675553:675899 [2] NCCL INFO Channel 21/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675558:675897 [7] NCCL INFO Channel 02/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-169-207:675555:675900 [4] NCCL INFO Channel 17/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-169-207:675558:675897 [7] NCCL INFO Channel 03/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-169-207:675555:675900 [4] NCCL INFO Channel 18/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-169-207:675558:675897 [7] NCCL INFO Channel 04/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-169-207:675555:675900 [4] NCCL INFO Channel 19/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-169-207:675558:675897 [7] NCCL INFO Channel 05/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-169-207:675555:675900 [4] NCCL INFO Channel 20/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-169-207:675553:675899 [2] NCCL INFO Channel 22/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675558:675897 [7] NCCL INFO Channel 06/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-169-207:675555:675900 [4] NCCL INFO Channel 21/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-169-207:675558:675897 [7] NCCL INFO Channel 07/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-169-207:675555:675900 [4] NCCL INFO Channel 22/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-169-207:675558:675897 [7] NCCL INFO Channel 08/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-169-207:675555:675900 [4] NCCL INFO Channel 23/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-169-207:675553:675899 [2] NCCL INFO Channel 23/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-169-207:675558:675897 [7] NCCL INFO Channel 09/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-169-207:675558:675897 [7] NCCL INFO Channel 10/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-169-207:675558:675897 [7] NCCL INFO Channel 11/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-169-207:675558:675897 [7] NCCL INFO Channel 12/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-169-207:675558:675897 [7] NCCL INFO Channel 13/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-169-207:675558:675897 [7] NCCL INFO Channel 14/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-169-207:675558:675897 [7] NCCL INFO Channel 15/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-169-207:675558:675897 [7] NCCL INFO Channel 16/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-169-207:675558:675897 [7] NCCL INFO Channel 17/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-169-207:675558:675897 [7] NCCL INFO Channel 18/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-169-207:675558:675897 [7] NCCL INFO Channel 19/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-169-207:675557:675896 [6] NCCL INFO Connected all trees
+ip-26-0-169-207:675557:675896 [6] NCCL INFO NVLS comm 0xbbf8710 headRank 3 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-169-207:675558:675897 [7] NCCL INFO Channel 20/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-169-207:675551:675881 [0] NCCL INFO Connected all trees
+ip-26-0-169-207:675558:675897 [7] NCCL INFO Channel 21/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-169-207:675551:675881 [0] NCCL INFO NVLS comm 0xad76160 headRank 0 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-169-207:675558:675897 [7] NCCL INFO Channel 22/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-169-207:675554:675898 [3] NCCL INFO Channel 00/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-169-207:675558:675897 [7] NCCL INFO Channel 23/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-169-207:675556:675901 [5] NCCL INFO Channel 00/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-169-207:675554:675898 [3] NCCL INFO Channel 01/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-169-207:675556:675901 [5] NCCL INFO Channel 01/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-169-207:675554:675898 [3] NCCL INFO Channel 02/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-169-207:675556:675901 [5] NCCL INFO Channel 02/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-169-207:675554:675898 [3] NCCL INFO Channel 03/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-169-207:675556:675901 [5] NCCL INFO Channel 03/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-169-207:675554:675898 [3] NCCL INFO Channel 04/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-169-207:675556:675901 [5] NCCL INFO Channel 04/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-169-207:675554:675898 [3] NCCL INFO Channel 05/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-169-207:675556:675901 [5] NCCL INFO Channel 05/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-169-207:675554:675898 [3] NCCL INFO Channel 06/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-169-207:675556:675901 [5] NCCL INFO Channel 06/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-169-207:675554:675898 [3] NCCL INFO Channel 07/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-169-207:675554:675898 [3] NCCL INFO Channel 08/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-169-207:675556:675901 [5] NCCL INFO Channel 07/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-169-207:675554:675898 [3] NCCL INFO Channel 09/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-169-207:675556:675901 [5] NCCL INFO Channel 08/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-169-207:675554:675898 [3] NCCL INFO Channel 10/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-169-207:675556:675901 [5] NCCL INFO Channel 09/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-169-207:675554:675898 [3] NCCL INFO Channel 11/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-169-207:675556:675901 [5] NCCL INFO Channel 10/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-169-207:675554:675898 [3] NCCL INFO Channel 12/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-169-207:675556:675901 [5] NCCL INFO Channel 11/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-169-207:675554:675898 [3] NCCL INFO Channel 13/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-169-207:675556:675901 [5] NCCL INFO Channel 12/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-169-207:675554:675898 [3] NCCL INFO Channel 14/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-169-207:675555:675900 [4] NCCL INFO Connected all trees
+ip-26-0-169-207:675556:675901 [5] NCCL INFO Channel 13/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-169-207:675555:675900 [4] NCCL INFO NVLS comm 0xb376d10 headRank 2 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-169-207:675554:675898 [3] NCCL INFO Channel 15/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-169-207:675556:675901 [5] NCCL INFO Channel 14/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-169-207:675553:675899 [2] NCCL INFO Connected all trees
+ip-26-0-169-207:675554:675898 [3] NCCL INFO Channel 16/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-169-207:675556:675901 [5] NCCL INFO Channel 15/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-169-207:675553:675899 [2] NCCL INFO NVLS comm 0xa5c4150 headRank 1 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-169-207:675554:675898 [3] NCCL INFO Channel 17/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-169-207:675556:675901 [5] NCCL INFO Channel 16/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-169-207:675554:675898 [3] NCCL INFO Channel 18/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-169-207:675556:675901 [5] NCCL INFO Channel 17/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-169-207:675554:675898 [3] NCCL INFO Channel 19/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-169-207:675556:675901 [5] NCCL INFO Channel 18/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-169-207:675554:675898 [3] NCCL INFO Channel 20/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-169-207:675556:675901 [5] NCCL INFO Channel 19/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-169-207:675554:675898 [3] NCCL INFO Channel 21/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-169-207:675556:675901 [5] NCCL INFO Channel 20/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-169-207:675554:675898 [3] NCCL INFO Channel 22/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-169-207:675556:675901 [5] NCCL INFO Channel 21/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-169-207:675554:675898 [3] NCCL INFO Channel 23/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-169-207:675556:675901 [5] NCCL INFO Channel 22/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-169-207:675556:675901 [5] NCCL INFO Channel 23/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-169-207:675555:675900 [4] NCCL INFO Connected NVLS tree
+ip-26-0-169-207:675555:675900 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-169-207:675555:675900 [4] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-169-207:675555:675900 [4] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:675557:675896 [6] NCCL INFO Connected NVLS tree
+ip-26-0-169-207:675557:675896 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-169-207:675551:675881 [0] NCCL INFO Connected NVLS tree
+ip-26-0-169-207:675557:675896 [6] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-169-207:675557:675896 [6] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:675551:675881 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-169-207:675551:675881 [0] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-169-207:675551:675881 [0] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:675553:675899 [2] NCCL INFO Connected NVLS tree
+ip-26-0-169-207:675553:675899 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-169-207:675553:675899 [2] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-169-207:675553:675899 [2] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:675551:675881 [0] NCCL INFO comm 0xad76160 rank 0 nranks 4 cudaDev 0 nvmlDev 0 busId 53000 commId 0xed6922ce1d2e3935 - Init COMPLETE
+ip-26-0-169-207:675555:675900 [4] NCCL INFO comm 0xb376d10 rank 2 nranks 4 cudaDev 4 nvmlDev 4 busId 97000 commId 0xed6922ce1d2e3935 - Init COMPLETE
+ip-26-0-169-207:675557:675896 [6] NCCL INFO comm 0xbbf8710 rank 3 nranks 4 cudaDev 6 nvmlDev 6 busId b9000 commId 0xed6922ce1d2e3935 - Init COMPLETE
+ip-26-0-169-207:675553:675899 [2] NCCL INFO comm 0xa5c4150 rank 1 nranks 4 cudaDev 2 nvmlDev 2 busId 75000 commId 0xed6922ce1d2e3935 - Init COMPLETE
+ip-26-0-169-207:675552:675883 [1] NCCL INFO Connected all trees
+ip-26-0-169-207:675552:675883 [1] NCCL INFO NVLS comm 0xb47d9b0 headRank 0 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-169-207:675558:675897 [7] NCCL INFO Connected all trees
+ip-26-0-169-207:675558:675897 [7] NCCL INFO NVLS comm 0xb5d3930 headRank 3 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-169-207:675556:675901 [5] NCCL INFO Connected all trees
+ip-26-0-169-207:675556:675901 [5] NCCL INFO NVLS comm 0xc1d51c0 headRank 2 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-169-207:675554:675898 [3] NCCL INFO Connected all trees
+ip-26-0-169-207:675554:675898 [3] NCCL INFO NVLS comm 0xc41d030 headRank 1 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-169-207:675558:675897 [7] NCCL INFO Connected NVLS tree
+ip-26-0-169-207:675558:675897 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-169-207:675558:675897 [7] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-169-207:675558:675897 [7] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:675552:675883 [1] NCCL INFO Connected NVLS tree
+ip-26-0-169-207:675554:675898 [3] NCCL INFO Connected NVLS tree
+ip-26-0-169-207:675552:675883 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-169-207:675554:675898 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-169-207:675554:675898 [3] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-169-207:675552:675883 [1] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-169-207:675554:675898 [3] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:675552:675883 [1] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:675556:675901 [5] NCCL INFO Connected NVLS tree
+ip-26-0-169-207:675556:675901 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-169-207:675556:675901 [5] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-169-207:675556:675901 [5] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:675558:675897 [7] NCCL INFO comm 0xb5d3930 rank 3 nranks 4 cudaDev 7 nvmlDev 7 busId ca000 commId 0x8e3037642664602d - Init COMPLETE
+ip-26-0-169-207:675554:675898 [3] NCCL INFO comm 0xc41d030 rank 1 nranks 4 cudaDev 3 nvmlDev 3 busId 86000 commId 0x8e3037642664602d - Init COMPLETE
+ip-26-0-169-207:675552:675883 [1] NCCL INFO comm 0xb47d9b0 rank 0 nranks 4 cudaDev 1 nvmlDev 1 busId 64000 commId 0x8e3037642664602d - Init COMPLETE
+ip-26-0-169-207:675556:675901 [5] NCCL INFO comm 0xc1d51c0 rank 2 nranks 4 cudaDev 5 nvmlDev 5 busId a8000 commId 0x8e3037642664602d - Init COMPLETE
+12/13/2024 12:22:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]: [Optimizer Building] Using LearningRateForSP as learning rate
+12/13/2024 12:22:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]: [ZeRO sharding] Size of optimizer params per rank:
+12/13/2024 12:22:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]: [ZeRO sharding] DP Rank 0 has 130M out of 520M (25.00%) params' optimizer states
+12/13/2024 12:22:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]: [ZeRO sharding] DP Rank 1 has 130M out of 520M (25.00%) params' optimizer states
+12/13/2024 12:22:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]: [ZeRO sharding] DP Rank 2 has 130M out of 520M (25.00%) params' optimizer states
+12/13/2024 12:22:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]: [ZeRO sharding] DP Rank 3 has 130M out of 520M (25.00%) params' optimizer states
+12/13/2024 12:22:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]: [Training Plan] Stage Stable Training Stage has 99 remaining training steps and has consumed 0 samples
+12/13/2024 12:22:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]: Using dummy data generator
+12/13/2024 12:22:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]: [Training Plan] There are 1 training stages 
+12/13/2024 12:22:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]: [Stage Stable Training Stage] start from step 1 
+12/13/2024 12:22:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]: 
+12/13/2024 12:22:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]: [Start training] datetime: 2024-12-13 12:22:45.793053 | mbs: 2 | grad_accum: 256 | global_batch_size: 2048 | sequence_length: 2048 | train_steps: 100 | start_iteration_step: 0 | consumed_train_samples: 0
+wandb: WARNING Path /fsx/nouamane/.cache/wandb/wandb/ wasn't writable, using system temp directory.
+wandb: WARNING Path /fsx/nouamane/.cache/wandb/wandb/ wasn't writable, using system temp directory
+wandb: Tracking run with wandb version 0.16.0
+wandb: W&B syncing is set to `offline` in this directory.  
+wandb: Run `wandb online` or set WANDB_MODE=online to enable cloud syncing.
+12/13/2024 12:22:54 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]: Waiting for 16 seconds
+12/13/2024 12:23:10 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]: Resuming training from stage Stable Training Stage, it has trained for 0 samples and has 99 remaining train steps
+12/13/2024 12:23:10 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:  Memory usage: 3472.48MiB. Peak allocated 5632.00MiB. Peak reserved: 18884.00MiB
+12/13/2024 12:23:31 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:  Memory usage: 3572.67MiB. Peak allocated 6985.16MiB. Peak reserved: 7434.00MiB
+12/13/2024 12:23:31 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]: iteration: 1 / 100 | consumed_tokens: 4.19M | elapsed_time_per_iteration_ms: 20.9K | tokens_per_sec: 200K | tokens_per_sec_per_gpu: 25K | global_batch_size: 2.05K | lm_loss: 10.8 | lr: 0.00015 | model_tflops_per_gpu: 176 | hardware_tflops_per_gpu: 176 | grad_norm: 0.431 | cuda_memory_allocated: 4.81G | cuda_max_memory_reserved: 9.62G | hd_total_memory_tb: 312G | hd_used_memory_tb: 69.7G | hd_free_memory_tb: 242G
+12/13/2024 12:23:31 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:  Memory usage: 4587.85MiB. Peak allocated 6820.14MiB. Peak reserved: 9178.00MiB
+12/13/2024 12:23:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:  Memory usage: 4587.97MiB. Peak allocated 8000.47MiB. Peak reserved: 9178.00MiB
+12/13/2024 12:23:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]: iteration: 2 / 100 | consumed_tokens: 8.39M | elapsed_time_per_iteration_ms: 18.9K | tokens_per_sec: 222K | tokens_per_sec_per_gpu: 27.7K | global_batch_size: 2.05K | lm_loss: 10.8 | lr: 0.0003 | model_tflops_per_gpu: 196 | hardware_tflops_per_gpu: 196 | grad_norm: 0.431 | cuda_memory_allocated: 4.81G | cuda_max_memory_reserved: 9.62G | hd_total_memory_tb: 312G | hd_used_memory_tb: 69.7G | hd_free_memory_tb: 242G
+12/13/2024 12:23:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:  Memory usage: 4587.85MiB. Peak allocated 6820.26MiB. Peak reserved: 9178.00MiB
+slurmstepd: error: *** STEP 13417747.0 ON ip-26-0-169-207 CANCELLED AT 2024-12-13T12:23:58 DUE TO TIME LIMIT ***
+srun: Job step aborted: Waiting up to 32 seconds for job step to finish.
+slurmstepd: error: *** JOB 13417747 ON ip-26-0-169-207 CANCELLED AT 2024-12-13T12:23:58 DUE TO TIME LIMIT ***
+[2024-12-13 12:23:58,739] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-13 12:23:58,739] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 675551 closing signal SIGTERM
+[2024-12-13 12:23:58,739] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 675552 closing signal SIGTERM
+[2024-12-13 12:23:58,739] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 675553 closing signal SIGTERM
+[2024-12-13 12:23:58,739] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 675554 closing signal SIGTERM
+[2024-12-13 12:23:58,741] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 675555 closing signal SIGTERM
+[2024-12-13 12:23:58,741] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 675556 closing signal SIGTERM
+[2024-12-13 12:23:58,741] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 675557 closing signal SIGTERM
+[2024-12-13 12:23:58,742] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 675558 closing signal SIGTERM
+[2024-12-13 12:23:58,828] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 675551 closing signal SIGTERM
+[2024-12-13 12:23:58,828] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 675552 closing signal SIGTERM
+[2024-12-13 12:23:58,828] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 675553 closing signal SIGTERM
+[2024-12-13 12:23:58,828] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 675554 closing signal SIGTERM
+[2024-12-13 12:23:58,829] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 675555 closing signal SIGTERM
+[2024-12-13 12:23:58,829] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 675556 closing signal SIGTERM
+[2024-12-13 12:23:58,829] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 675557 closing signal SIGTERM
+[2024-12-13 12:23:58,829] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 675558 closing signal SIGTERM
diff --git a/logs/13417877-bench_stress_test.out b/logs/13417877-bench_stress_test.out
new file mode 100644
index 0000000000000000000000000000000000000000..9eb7478e130b7caedfbddd5fd5f39d88ca8f3ce7
--- /dev/null
+++ b/logs/13417877-bench_stress_test.out
@@ -0,0 +1,4143 @@
++ source /etc/profile.d/modules.sh
+++ . /usr/share/modules/init/bash
++++ unset _mlshdbg
++++ '[' 0 = 1 ']'
++++ unset _mlre _mlIFS
++++ '[' -n x ']'
++++ _mlIFS=' 	
+'
++++ IFS=' '
++++ '[' -n '' ']'
+++++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash autoinit
++++ _mlcode='module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+MODULEPATH_modshare=/etc/environment-modules/modules:1:/usr/share/modules/\$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1; export MODULEPATH_modshare;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+LOADEDMODULES=; export LOADEDMODULES;
+MODULEPATH=/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/\$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles; export MODULEPATH;
+test 0;'
++++ _mlret=0
++++ '[' -n x ']'
++++ IFS=' 	
+'
++++ unset _mlIFS
++++ unset _mlre _mlv _mlrv
++++ '[' 0 -eq 0 ']'
++++ eval 'module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+MODULEPATH_modshare=/etc/environment-modules/modules:1:/usr/share/modules/\$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1; export MODULEPATH_modshare;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+LOADEDMODULES=; export LOADEDMODULES;
+MODULEPATH=/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/\$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles; export MODULEPATH;
+test 0;'
+++++ MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl
+++++ export MODULES_CMD
+++++ ENV=/usr/share/modules/init/profile.sh
+++++ export ENV
+++++ MODULEPATH_modshare='/etc/environment-modules/modules:1:/usr/share/modules/$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1'
+++++ export MODULEPATH_modshare
+++++ BASH_ENV=/usr/share/modules/init/bash
+++++ export BASH_ENV
+++++ MODULESHOME=/usr/share/modules
+++++ export MODULESHOME
+++++ LOADEDMODULES=
+++++ export LOADEDMODULES
+++++ MODULEPATH='/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles'
+++++ export MODULEPATH
+++++ test 0
++++ '[' 0 = 1 ']'
++++ '[' -t 2 ']'
++++ export -f module
++++ export -f switchml
++++ '[' 5 -ge 3 ']'
++++ [[ ehxB =~ i ]]
++++ [[ ! :/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin: =~ :/usr/bin: ]]
++++ '[' '!' -n '' ']'
++++ MANPATH=:
++++ export MANPATH
+++++ manpath
++++ [[ ! :/admin/home/nouamane/.local/share/man:/fsx/nouamane/miniconda/envs/2-1-cu121/man:/fsx/nouamane/miniconda/envs/2-1-cu121/share/man:/fsx/nouamane/miniconda/man:/fsx/nouamane/miniconda/share/man:/opt/amazon/openmpi/share/man:/opt/amazon/efa/share/man:/opt/slurm/share/man:/usr/local/man:/usr/local/share/man:/usr/share/man:/admin/home/nouamane/.fzf/man:: =~ :/usr/share/man: ]]
++++ unset _mlcode _mlret
++++ '[' -n '' ']'
++ module load cuda/12.1
++ unset _mlshdbg
++ '[' 0 = 1 ']'
++ unset _mlre _mlIFS
++ '[' -n x ']'
++ _mlIFS=' 	
+'
++ IFS=' '
++ '[' -n '' ']'
+++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash load cuda/12.1
++ eval 'CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include;' export 'CPATH;
+LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:;' export 'LD_LIBRARY_PATH;
+FI_EFA_FORK_SAFE=1;' export 'FI_EFA_FORK_SAFE;
+MANPATH=/usr/local/cuda-12.1/share/man::;' export 'MANPATH;
+LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64;' export 'LIBRARY_PATH;
+_LMFILES_=/usr/share/modules/modulefiles/cuda/12.1;' export '_LMFILES_;
+LOADEDMODULES=cuda/12.1;' export 'LOADEDMODULES;
+MPI_PATH=/opt/amazon/openmpi;' export 'MPI_PATH;
+NCCL_HOME_modshare=/opt/nccl/build:1;' export 'NCCL_HOME_modshare;
+NCCL_PROTO=simple;' export 'NCCL_PROTO;
+MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1;' export 'MANPATH_modshare;
+LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1;' export 'LIBRARY_PATH_modshare;
+NCCL_SOCKET_IFNAME=enp;' export 'NCCL_SOCKET_IFNAME;
+AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl;' export 'AWS_OFI_NCCL_HOME;
+NCCL_HOME=/opt/nccl/build;' export 'NCCL_HOME;
+FI_PROVIDER=efa;' export 'FI_PROVIDER;
+AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1;' export 'AWS_OFI_NCCL_HOME_modshare;
+CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1;' export 'CPATH_modshare;
+LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2::1:/opt/nccl/build/lib:1:/opt/aws-ofi-nccl/lib:2:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/efa/lib:1;' export 'LD_LIBRARY_PATH_modshare;
+FI_EFA_ENABLE_SHM_TRANSFER=1;' export 'FI_EFA_ENABLE_SHM_TRANSFER;
+_LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1;' export '_LMFILES__modshare;
+LOADEDMODULES_modshare=cuda/12.1:1;' export 'LOADEDMODULES_modshare;
+MPI_PATH_modshare=/opt/amazon/openmpi:1;' export 'MPI_PATH_modshare;
+PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin;' export 'PATH;
+CUDA_HOME=/usr/local/cuda-12.1;' export 'CUDA_HOME;
+PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/usr/local/cuda-12.1/include:1:/opt/amazon/efa/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/usr/sbin:1:/fsx/nouamane/miniconda/bin:1:/opt/slurm/bin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:2:/opt/amazon/openmpi/bin:1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1;' export 'PATH_modshare;
+test' '0;'
+++ CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include
+++ export CPATH
+++ LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:
+++ export LD_LIBRARY_PATH
+++ FI_EFA_FORK_SAFE=1
+++ export FI_EFA_FORK_SAFE
+++ MANPATH=/usr/local/cuda-12.1/share/man::
+++ export MANPATH
+++ LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64
+++ export LIBRARY_PATH
+++ _LMFILES_=/usr/share/modules/modulefiles/cuda/12.1
+++ export _LMFILES_
+++ LOADEDMODULES=cuda/12.1
+++ export LOADEDMODULES
+++ MPI_PATH=/opt/amazon/openmpi
+++ export MPI_PATH
+++ NCCL_HOME_modshare=/opt/nccl/build:1
+++ export NCCL_HOME_modshare
+++ NCCL_PROTO=simple
+++ export NCCL_PROTO
+++ MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1
+++ export MANPATH_modshare
+++ LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1
+++ export LIBRARY_PATH_modshare
+++ NCCL_SOCKET_IFNAME=enp
+++ export NCCL_SOCKET_IFNAME
+++ AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl
+++ export AWS_OFI_NCCL_HOME
+++ NCCL_HOME=/opt/nccl/build
+++ export NCCL_HOME
+++ FI_PROVIDER=efa
+++ export FI_PROVIDER
+++ AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1
+++ export AWS_OFI_NCCL_HOME_modshare
+++ CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1
+++ export CPATH_modshare
+++ LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2::1:/opt/nccl/build/lib:1:/opt/aws-ofi-nccl/lib:2:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/efa/lib:1
+++ export LD_LIBRARY_PATH_modshare
+++ FI_EFA_ENABLE_SHM_TRANSFER=1
+++ export FI_EFA_ENABLE_SHM_TRANSFER
+++ _LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1
+++ export _LMFILES__modshare
+++ LOADEDMODULES_modshare=cuda/12.1:1
+++ export LOADEDMODULES_modshare
+++ MPI_PATH_modshare=/opt/amazon/openmpi:1
+++ export MPI_PATH_modshare
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ export PATH
+++ CUDA_HOME=/usr/local/cuda-12.1
+++ export CUDA_HOME
+++ PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/usr/local/cuda-12.1/include:1:/opt/amazon/efa/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/usr/sbin:1:/fsx/nouamane/miniconda/bin:1:/opt/slurm/bin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:2:/opt/amazon/openmpi/bin:1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1
+++ export PATH_modshare
+++ test 0
++ _mlstatus=0
++ '[' -n x ']'
++ IFS=' 	
+'
++ unset _mlre _mlv _mlrv _mlIFS
++ '[' -n '' ']'
++ unset _mlshdbg
++ return 0
++ source /fsx/nouamane/miniconda/bin/activate
+++ _CONDA_ROOT=/fsx/nouamane/miniconda
+++ . /fsx/nouamane/miniconda/etc/profile.d/conda.sh
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ '[' -z x ']'
+++ conda activate
+++ local cmd=activate
+++ case "$cmd" in
+++ __conda_activate activate
+++ '[' -n '' ']'
+++ local ask_conda
++++ PS1=
++++ __conda_exe shell.posix activate
++++ /fsx/nouamane/miniconda/bin/conda shell.posix activate
+++ ask_conda='. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_2='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\'''
+++ eval '. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_2='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\'''
++++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh
+++++ test -n ''
+++++ unset XML_CATALOG_FILES
+++++ unset xml_catalog_files_libxml2
++++ PS1='(base) '
++++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++++ export CONDA_PREFIX=/fsx/nouamane/miniconda
++++ CONDA_PREFIX=/fsx/nouamane/miniconda
++++ export CONDA_SHLVL=3
++++ CONDA_SHLVL=3
++++ export CONDA_DEFAULT_ENV=base
++++ CONDA_DEFAULT_ENV=base
++++ export 'CONDA_PROMPT_MODIFIER=(base) '
++++ CONDA_PROMPT_MODIFIER='(base) '
++++ export CONDA_PREFIX_2=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ CONDA_PREFIX_2=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ __conda_hashr
+++ '[' -n '' ']'
+++ '[' -n '' ']'
+++ hash -r
++ conda activate 2-1-cu121
++ local cmd=activate
++ case "$cmd" in
++ __conda_activate activate 2-1-cu121
++ '[' -n '' ']'
++ local ask_conda
+++ PS1='(base) '
+++ __conda_exe shell.posix activate 2-1-cu121
+++ /fsx/nouamane/miniconda/bin/conda shell.posix activate 2-1-cu121
++ ask_conda='PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''4'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_3='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
++ eval 'PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''4'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_3='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
+++ PS1='(2-1-cu121) '
+++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ export CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ export CONDA_SHLVL=4
+++ CONDA_SHLVL=4
+++ export CONDA_DEFAULT_ENV=2-1-cu121
+++ CONDA_DEFAULT_ENV=2-1-cu121
+++ export 'CONDA_PROMPT_MODIFIER=(2-1-cu121) '
+++ CONDA_PROMPT_MODIFIER='(2-1-cu121) '
+++ export CONDA_PREFIX_3=/fsx/nouamane/miniconda
+++ CONDA_PREFIX_3=/fsx/nouamane/miniconda
+++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ export _CE_M=
+++ _CE_M=
+++ export _CE_CONDA=
+++ _CE_CONDA=
+++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh
++++ test -n ''
++++ xml_catalog_files_libxml2=
++++ XML_CATALOG_FILES=
++++ conda_catalog_files=
++++ ifs_libxml2=' 	
+'
++++ IFS=' '
++++ rem=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ for pre in ${rem}
++++ test '' = /fsx/nouamane/miniconda/envs/2-1-cu121
++++ conda_catalog_files=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ rem=
++++ IFS=' 	
+'
++++ conda_catalog_files='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ export 'XML_CATALOG_FILES=file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ XML_CATALOG_FILES='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ unset conda_catalog_files ifs_libxml2 rem
++ __conda_hashr
++ '[' -n '' ']'
++ '[' -n '' ']'
++ hash -r
++ export PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++ PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ scontrol show hostnames ip-26-0-160-103
++ export NODELIST=ip-26-0-160-103
++ NODELIST=ip-26-0-160-103
+++ scontrol show hostnames ip-26-0-160-103
+++ head -n1
++ export MASTER_NODE=ip-26-0-160-103
++ MASTER_NODE=ip-26-0-160-103
++ export MASTER_PORT=12356
++ MASTER_PORT=12356
++ export NNODES=1
++ NNODES=1
++ export GPUS_PER_NODE=8
++ GPUS_PER_NODE=8
++ export WORLD_SIZE=8
++ WORLD_SIZE=8
++ export CUDA_DEVICE_MAX_CONNECTIONS=1
++ CUDA_DEVICE_MAX_CONNECTIONS=1
++ export NCCL_DEBUG=INFO
++ NCCL_DEBUG=INFO
++ export NANOTRON_BENCHMARK=1
++ NANOTRON_BENCHMARK=1
++ echo '=== GPU Topology ==='
+=== GPU Topology ===
++ nvidia-smi topo -m
+	GPU0	GPU1	GPU2	GPU3	GPU4	GPU5	GPU6	GPU7	CPU Affinity	NUMA Affinity	GPU NUMA ID
+GPU0	 X 	NV18	NV18	NV18	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU1	NV18	 X 	NV18	NV18	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU2	NV18	NV18	 X 	NV18	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU3	NV18	NV18	NV18	 X 	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU4	NV18	NV18	NV18	NV18	 X 	NV18	NV18	NV18	48-95	1		N/A
+GPU5	NV18	NV18	NV18	NV18	NV18	 X 	NV18	NV18	48-95	1		N/A
+GPU6	NV18	NV18	NV18	NV18	NV18	NV18	 X 	NV18	48-95	1		N/A
+GPU7	NV18	NV18	NV18	NV18	NV18	NV18	NV18	 X 	48-95	1		N/A
+
+Legend:
+
+  X    = Self
+  SYS  = Connection traversing PCIe as well as the SMP interconnect between NUMA nodes (e.g., QPI/UPI)
+  NODE = Connection traversing PCIe as well as the interconnect between PCIe Host Bridges within a NUMA node
+  PHB  = Connection traversing PCIe as well as a PCIe Host Bridge (typically the CPU)
+  PXB  = Connection traversing multiple PCIe bridges (without traversing the PCIe Host Bridge)
+  PIX  = Connection traversing at most a single PCIe bridge
+  NV#  = Connection traversing a bonded set of # NVLinks
++ echo ==================
+==================
++ echo 'Master node: ip-26-0-160-103'
+Master node: ip-26-0-160-103
++ echo 'All nodes: ip-26-0-160-103'
+All nodes: ip-26-0-160-103
++ echo 'World size: 8'
+World size: 8
++ srun torchrun --nnodes=1 --nproc_per_node=8 --rdzv_id=13417877 --rdzv_backend=c10d --rdzv_endpoint=ip-26-0-160-103:12356 /fsx/nouamane/projects/nanotron/run_train.py --config-file benchmark/configs/config_1.14G_dp4_tp2_pp1_acc256_mbs2_seq2048_zero1_tpmodeRED_vocab32k.yaml
+[2024-12-13 14:16:46,352] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-13 14:16:46,353] torch.distributed.run: [WARNING] 
+[2024-12-13 14:16:46,353] torch.distributed.run: [WARNING] *****************************************
+[2024-12-13 14:16:46,353] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-13 14:16:46,353] torch.distributed.run: [WARNING] *****************************************
+ip-26-0-160-103:2609916:2609916 [0] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-160-103:2609916:2609916 [0] NCCL INFO Bootstrap : Using enp71s0:26.0.160.103<0>
+ip-26-0-160-103:2609916:2609916 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-160-103:2609916:2609916 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-160-103:2609916:2609916 [0] NCCL INFO cudaDriverVersion 12020
+NCCL version 2.18.5+cuda12.2
+ip-26-0-160-103:2609919:2609919 [3] NCCL INFO cudaDriverVersion 12020
+ip-26-0-160-103:2609923:2609923 [7] NCCL INFO cudaDriverVersion 12020
+ip-26-0-160-103:2609921:2609921 [5] NCCL INFO cudaDriverVersion 12020
+ip-26-0-160-103:2609919:2609919 [3] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-160-103:2609923:2609923 [7] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-160-103:2609921:2609921 [5] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-160-103:2609919:2609919 [3] NCCL INFO Bootstrap : Using enp71s0:26.0.160.103<0>
+ip-26-0-160-103:2609923:2609923 [7] NCCL INFO Bootstrap : Using enp71s0:26.0.160.103<0>
+ip-26-0-160-103:2609921:2609921 [5] NCCL INFO Bootstrap : Using enp71s0:26.0.160.103<0>
+ip-26-0-160-103:2609923:2609923 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-160-103:2609921:2609921 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-160-103:2609923:2609923 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-160-103:2609919:2609919 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-160-103:2609921:2609921 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-160-103:2609919:2609919 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-160-103:2609917:2609917 [1] NCCL INFO cudaDriverVersion 12020
+ip-26-0-160-103:2609917:2609917 [1] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-160-103:2609917:2609917 [1] NCCL INFO Bootstrap : Using enp71s0:26.0.160.103<0>
+ip-26-0-160-103:2609917:2609917 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-160-103:2609917:2609917 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-160-103:2609918:2609918 [2] NCCL INFO cudaDriverVersion 12020
+ip-26-0-160-103:2609922:2609922 [6] NCCL INFO cudaDriverVersion 12020
+ip-26-0-160-103:2609920:2609920 [4] NCCL INFO cudaDriverVersion 12020
+ip-26-0-160-103:2609922:2609922 [6] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-160-103:2609918:2609918 [2] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-160-103:2609920:2609920 [4] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-160-103:2609922:2609922 [6] NCCL INFO Bootstrap : Using enp71s0:26.0.160.103<0>
+ip-26-0-160-103:2609918:2609918 [2] NCCL INFO Bootstrap : Using enp71s0:26.0.160.103<0>
+ip-26-0-160-103:2609920:2609920 [4] NCCL INFO Bootstrap : Using enp71s0:26.0.160.103<0>
+ip-26-0-160-103:2609922:2609922 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-160-103:2609922:2609922 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-160-103:2609918:2609918 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-160-103:2609920:2609920 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-160-103:2609918:2609918 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-160-103:2609920:2609920 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-160-103:2609923:2610039 [7] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.9.1-aws
+ip-26-0-160-103:2609923:2610039 [7] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.9.1-aws
+ip-26-0-160-103:2609923:2610039 [7] NCCL INFO NET/OFI Using Libfabric version 1.21
+ip-26-0-160-103:2609923:2610039 [7] NCCL INFO NET/OFI Using CUDA driver version 12020
+ip-26-0-160-103:2609923:2610039 [7] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-160-103:2609923:2610039 [7] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-160-103:2609923:2610039 [7] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-160-103:2609923:2610039 [7] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-160-103:2609923:2610039 [7] NCCL INFO NET/OFI Running on p5.48xlarge platform, Setting NCCL_TOPO_FILE environment variable to /opt/aws-ofi-nccl/share/aws-ofi-nccl/xml/p5.48xl-topo.xml
+ip-26-0-160-103:2609923:2610039 [7] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-160-103:2609923:2610039 [7] NCCL INFO NET/OFI Using transport protocol RDMA
+ip-26-0-160-103:2609916:2610036 [0] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.9.1-aws
+ip-26-0-160-103:2609916:2610036 [0] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.9.1-aws
+ip-26-0-160-103:2609916:2610036 [0] NCCL INFO NET/OFI Using Libfabric version 1.21
+ip-26-0-160-103:2609916:2610036 [0] NCCL INFO NET/OFI Using CUDA driver version 12020
+ip-26-0-160-103:2609916:2610036 [0] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-160-103:2609916:2610036 [0] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-160-103:2609916:2610036 [0] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-160-103:2609916:2610036 [0] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-160-103:2609916:2610036 [0] NCCL INFO NET/OFI Running on p5.48xlarge platform, Setting NCCL_TOPO_FILE environment variable to /opt/aws-ofi-nccl/share/aws-ofi-nccl/xml/p5.48xl-topo.xml
+ip-26-0-160-103:2609916:2610036 [0] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-160-103:2609916:2610036 [0] NCCL INFO NET/OFI Using transport protocol RDMA
+ip-26-0-160-103:2609921:2610037 [5] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.9.1-aws
+ip-26-0-160-103:2609921:2610037 [5] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.9.1-aws
+ip-26-0-160-103:2609921:2610037 [5] NCCL INFO NET/OFI Using Libfabric version 1.21
+ip-26-0-160-103:2609921:2610037 [5] NCCL INFO NET/OFI Using CUDA driver version 12020
+ip-26-0-160-103:2609921:2610037 [5] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-160-103:2609921:2610037 [5] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-160-103:2609921:2610037 [5] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-160-103:2609921:2610037 [5] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-160-103:2609921:2610037 [5] NCCL INFO NET/OFI Running on p5.48xlarge platform, Setting NCCL_TOPO_FILE environment variable to /opt/aws-ofi-nccl/share/aws-ofi-nccl/xml/p5.48xl-topo.xml
+ip-26-0-160-103:2609921:2610037 [5] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-160-103:2609921:2610037 [5] NCCL INFO NET/OFI Using transport protocol RDMA
+ip-26-0-160-103:2609919:2610038 [3] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.9.1-aws
+ip-26-0-160-103:2609919:2610038 [3] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.9.1-aws
+ip-26-0-160-103:2609919:2610038 [3] NCCL INFO NET/OFI Using Libfabric version 1.21
+ip-26-0-160-103:2609919:2610038 [3] NCCL INFO NET/OFI Using CUDA driver version 12020
+ip-26-0-160-103:2609919:2610038 [3] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-160-103:2609919:2610038 [3] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-160-103:2609919:2610038 [3] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-160-103:2609919:2610038 [3] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-160-103:2609919:2610038 [3] NCCL INFO NET/OFI Running on p5.48xlarge platform, Setting NCCL_TOPO_FILE environment variable to /opt/aws-ofi-nccl/share/aws-ofi-nccl/xml/p5.48xl-topo.xml
+ip-26-0-160-103:2609919:2610038 [3] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-160-103:2609919:2610038 [3] NCCL INFO NET/OFI Using transport protocol RDMA
+ip-26-0-160-103:2609922:2610041 [6] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.9.1-aws
+ip-26-0-160-103:2609922:2610041 [6] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.9.1-aws
+ip-26-0-160-103:2609922:2610041 [6] NCCL INFO NET/OFI Using Libfabric version 1.21
+ip-26-0-160-103:2609922:2610041 [6] NCCL INFO NET/OFI Using CUDA driver version 12020
+ip-26-0-160-103:2609922:2610041 [6] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-160-103:2609922:2610041 [6] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-160-103:2609922:2610041 [6] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-160-103:2609922:2610041 [6] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-160-103:2609922:2610041 [6] NCCL INFO NET/OFI Running on p5.48xlarge platform, Setting NCCL_TOPO_FILE environment variable to /opt/aws-ofi-nccl/share/aws-ofi-nccl/xml/p5.48xl-topo.xml
+ip-26-0-160-103:2609922:2610041 [6] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-160-103:2609922:2610041 [6] NCCL INFO NET/OFI Using transport protocol RDMA
+ip-26-0-160-103:2609917:2610040 [1] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.9.1-aws
+ip-26-0-160-103:2609917:2610040 [1] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.9.1-aws
+ip-26-0-160-103:2609917:2610040 [1] NCCL INFO NET/OFI Using Libfabric version 1.21
+ip-26-0-160-103:2609917:2610040 [1] NCCL INFO NET/OFI Using CUDA driver version 12020
+ip-26-0-160-103:2609917:2610040 [1] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-160-103:2609917:2610040 [1] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-160-103:2609917:2610040 [1] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-160-103:2609917:2610040 [1] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-160-103:2609917:2610040 [1] NCCL INFO NET/OFI Running on p5.48xlarge platform, Setting NCCL_TOPO_FILE environment variable to /opt/aws-ofi-nccl/share/aws-ofi-nccl/xml/p5.48xl-topo.xml
+ip-26-0-160-103:2609917:2610040 [1] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-160-103:2609917:2610040 [1] NCCL INFO NET/OFI Using transport protocol RDMA
+ip-26-0-160-103:2609920:2610043 [4] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.9.1-aws
+ip-26-0-160-103:2609920:2610043 [4] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.9.1-aws
+ip-26-0-160-103:2609920:2610043 [4] NCCL INFO NET/OFI Using Libfabric version 1.21
+ip-26-0-160-103:2609920:2610043 [4] NCCL INFO NET/OFI Using CUDA driver version 12020
+ip-26-0-160-103:2609920:2610043 [4] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-160-103:2609920:2610043 [4] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-160-103:2609920:2610043 [4] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-160-103:2609920:2610043 [4] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-160-103:2609920:2610043 [4] NCCL INFO NET/OFI Running on p5.48xlarge platform, Setting NCCL_TOPO_FILE environment variable to /opt/aws-ofi-nccl/share/aws-ofi-nccl/xml/p5.48xl-topo.xml
+ip-26-0-160-103:2609920:2610043 [4] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-160-103:2609920:2610043 [4] NCCL INFO NET/OFI Using transport protocol RDMA
+ip-26-0-160-103:2609918:2610042 [2] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.9.1-aws
+ip-26-0-160-103:2609918:2610042 [2] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.9.1-aws
+ip-26-0-160-103:2609918:2610042 [2] NCCL INFO NET/OFI Using Libfabric version 1.21
+ip-26-0-160-103:2609918:2610042 [2] NCCL INFO NET/OFI Using CUDA driver version 12020
+ip-26-0-160-103:2609918:2610042 [2] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-160-103:2609918:2610042 [2] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-160-103:2609918:2610042 [2] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-160-103:2609918:2610042 [2] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-160-103:2609918:2610042 [2] NCCL INFO NET/OFI Running on p5.48xlarge platform, Setting NCCL_TOPO_FILE environment variable to /opt/aws-ofi-nccl/share/aws-ofi-nccl/xml/p5.48xl-topo.xml
+ip-26-0-160-103:2609918:2610042 [2] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-160-103:2609918:2610042 [2] NCCL INFO NET/OFI Using transport protocol RDMA
+ip-26-0-160-103:2609923:2610039 [7] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-160-103:2609916:2610036 [0] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-160-103:2609921:2610037 [5] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-160-103:2609919:2610038 [3] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-160-103:2609922:2610041 [6] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-160-103:2609917:2610040 [1] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-160-103:2609920:2610043 [4] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-160-103:2609918:2610042 [2] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-160-103:2609916:2610036 [0] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-160-103:2609916:2610036 [0] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-160-103:2609916:2610036 [0] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-160-103:2609916:2610036 [0] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-160-103:2609916:2610036 [0] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-160-103:2609916:2610036 [0] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-160-103:2609916:2610036 [0] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-160-103:2609916:2610036 [0] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-160-103:2609916:2610036 [0] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-160-103:2609916:2610036 [0] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-160-103:2609916:2610036 [0] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-160-103:2609916:2610036 [0] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-160-103:2609916:2610036 [0] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-160-103:2609916:2610036 [0] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-160-103:2609916:2610036 [0] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-160-103:2609916:2610036 [0] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-160-103:2609916:2610036 [0] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-160-103:2609916:2610036 [0] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-160-103:2609916:2610036 [0] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-160-103:2609916:2610036 [0] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-160-103:2609916:2610036 [0] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-160-103:2609916:2610036 [0] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-160-103:2609916:2610036 [0] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-160-103:2609916:2610036 [0] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-160-103:2609916:2610036 [0] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-160-103:2609916:2610036 [0] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-160-103:2609916:2610036 [0] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-160-103:2609916:2610036 [0] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-160-103:2609916:2610036 [0] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-160-103:2609916:2610036 [0] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-160-103:2609916:2610036 [0] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-160-103:2609916:2610036 [0] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-160-103:2609916:2610036 [0] NCCL INFO Using network AWS Libfabric
+ip-26-0-160-103:2609921:2610037 [5] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-160-103:2609921:2610037 [5] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-160-103:2609921:2610037 [5] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-160-103:2609921:2610037 [5] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-160-103:2609921:2610037 [5] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-160-103:2609921:2610037 [5] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-160-103:2609921:2610037 [5] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-160-103:2609921:2610037 [5] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-160-103:2609921:2610037 [5] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-160-103:2609921:2610037 [5] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-160-103:2609921:2610037 [5] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-160-103:2609921:2610037 [5] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-160-103:2609921:2610037 [5] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-160-103:2609921:2610037 [5] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-160-103:2609921:2610037 [5] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-160-103:2609921:2610037 [5] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-160-103:2609921:2610037 [5] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-160-103:2609921:2610037 [5] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-160-103:2609921:2610037 [5] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-160-103:2609921:2610037 [5] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-160-103:2609921:2610037 [5] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-160-103:2609921:2610037 [5] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-160-103:2609921:2610037 [5] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-160-103:2609921:2610037 [5] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-160-103:2609921:2610037 [5] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-160-103:2609921:2610037 [5] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-160-103:2609921:2610037 [5] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-160-103:2609921:2610037 [5] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-160-103:2609921:2610037 [5] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-160-103:2609921:2610037 [5] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-160-103:2609921:2610037 [5] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-160-103:2609921:2610037 [5] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-160-103:2609923:2610039 [7] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-160-103:2609923:2610039 [7] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-160-103:2609923:2610039 [7] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-160-103:2609923:2610039 [7] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-160-103:2609923:2610039 [7] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-160-103:2609923:2610039 [7] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-160-103:2609923:2610039 [7] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-160-103:2609923:2610039 [7] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-160-103:2609923:2610039 [7] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-160-103:2609923:2610039 [7] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-160-103:2609923:2610039 [7] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-160-103:2609923:2610039 [7] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-160-103:2609923:2610039 [7] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-160-103:2609923:2610039 [7] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-160-103:2609923:2610039 [7] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-160-103:2609923:2610039 [7] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-160-103:2609923:2610039 [7] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-160-103:2609923:2610039 [7] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-160-103:2609923:2610039 [7] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-160-103:2609923:2610039 [7] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-160-103:2609923:2610039 [7] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-160-103:2609923:2610039 [7] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-160-103:2609923:2610039 [7] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-160-103:2609923:2610039 [7] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-160-103:2609923:2610039 [7] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-160-103:2609923:2610039 [7] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-160-103:2609923:2610039 [7] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-160-103:2609923:2610039 [7] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-160-103:2609923:2610039 [7] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-160-103:2609923:2610039 [7] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-160-103:2609923:2610039 [7] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-160-103:2609923:2610039 [7] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-160-103:2609919:2610038 [3] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-160-103:2609919:2610038 [3] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-160-103:2609919:2610038 [3] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-160-103:2609919:2610038 [3] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-160-103:2609919:2610038 [3] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-160-103:2609919:2610038 [3] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-160-103:2609919:2610038 [3] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-160-103:2609919:2610038 [3] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-160-103:2609919:2610038 [3] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-160-103:2609919:2610038 [3] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-160-103:2609919:2610038 [3] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-160-103:2609919:2610038 [3] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-160-103:2609919:2610038 [3] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-160-103:2609919:2610038 [3] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-160-103:2609919:2610038 [3] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-160-103:2609919:2610038 [3] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-160-103:2609919:2610038 [3] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-160-103:2609919:2610038 [3] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-160-103:2609919:2610038 [3] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-160-103:2609919:2610038 [3] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-160-103:2609919:2610038 [3] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-160-103:2609919:2610038 [3] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-160-103:2609919:2610038 [3] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-160-103:2609919:2610038 [3] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-160-103:2609919:2610038 [3] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-160-103:2609919:2610038 [3] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-160-103:2609919:2610038 [3] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-160-103:2609919:2610038 [3] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-160-103:2609919:2610038 [3] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-160-103:2609919:2610038 [3] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-160-103:2609919:2610038 [3] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-160-103:2609919:2610038 [3] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-160-103:2609920:2610043 [4] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-160-103:2609920:2610043 [4] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-160-103:2609920:2610043 [4] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-160-103:2609920:2610043 [4] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-160-103:2609920:2610043 [4] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-160-103:2609920:2610043 [4] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-160-103:2609920:2610043 [4] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-160-103:2609920:2610043 [4] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-160-103:2609920:2610043 [4] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-160-103:2609920:2610043 [4] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-160-103:2609920:2610043 [4] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-160-103:2609920:2610043 [4] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-160-103:2609920:2610043 [4] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-160-103:2609920:2610043 [4] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-160-103:2609920:2610043 [4] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-160-103:2609920:2610043 [4] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-160-103:2609920:2610043 [4] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-160-103:2609920:2610043 [4] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-160-103:2609920:2610043 [4] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-160-103:2609920:2610043 [4] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-160-103:2609920:2610043 [4] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-160-103:2609920:2610043 [4] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-160-103:2609920:2610043 [4] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-160-103:2609920:2610043 [4] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-160-103:2609920:2610043 [4] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-160-103:2609920:2610043 [4] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-160-103:2609920:2610043 [4] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-160-103:2609920:2610043 [4] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-160-103:2609920:2610043 [4] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-160-103:2609920:2610043 [4] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-160-103:2609920:2610043 [4] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-160-103:2609920:2610043 [4] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-160-103:2609922:2610041 [6] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-160-103:2609922:2610041 [6] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-160-103:2609922:2610041 [6] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-160-103:2609922:2610041 [6] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-160-103:2609922:2610041 [6] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-160-103:2609922:2610041 [6] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-160-103:2609922:2610041 [6] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-160-103:2609922:2610041 [6] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-160-103:2609922:2610041 [6] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-160-103:2609922:2610041 [6] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-160-103:2609922:2610041 [6] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-160-103:2609922:2610041 [6] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-160-103:2609922:2610041 [6] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-160-103:2609922:2610041 [6] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-160-103:2609922:2610041 [6] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-160-103:2609922:2610041 [6] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-160-103:2609922:2610041 [6] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-160-103:2609922:2610041 [6] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-160-103:2609922:2610041 [6] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-160-103:2609922:2610041 [6] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-160-103:2609922:2610041 [6] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-160-103:2609922:2610041 [6] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-160-103:2609922:2610041 [6] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-160-103:2609922:2610041 [6] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-160-103:2609922:2610041 [6] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-160-103:2609922:2610041 [6] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-160-103:2609922:2610041 [6] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-160-103:2609922:2610041 [6] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-160-103:2609922:2610041 [6] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-160-103:2609922:2610041 [6] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-160-103:2609922:2610041 [6] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-160-103:2609922:2610041 [6] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-160-103:2609917:2610040 [1] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-160-103:2609917:2610040 [1] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-160-103:2609917:2610040 [1] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-160-103:2609917:2610040 [1] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-160-103:2609917:2610040 [1] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-160-103:2609917:2610040 [1] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-160-103:2609917:2610040 [1] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-160-103:2609917:2610040 [1] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-160-103:2609917:2610040 [1] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-160-103:2609917:2610040 [1] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-160-103:2609917:2610040 [1] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-160-103:2609917:2610040 [1] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-160-103:2609917:2610040 [1] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-160-103:2609917:2610040 [1] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-160-103:2609917:2610040 [1] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-160-103:2609917:2610040 [1] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-160-103:2609917:2610040 [1] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-160-103:2609917:2610040 [1] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-160-103:2609917:2610040 [1] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-160-103:2609917:2610040 [1] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-160-103:2609917:2610040 [1] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-160-103:2609917:2610040 [1] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-160-103:2609917:2610040 [1] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-160-103:2609917:2610040 [1] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-160-103:2609917:2610040 [1] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-160-103:2609917:2610040 [1] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-160-103:2609917:2610040 [1] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-160-103:2609917:2610040 [1] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-160-103:2609917:2610040 [1] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-160-103:2609917:2610040 [1] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-160-103:2609917:2610040 [1] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-160-103:2609917:2610040 [1] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-160-103:2609918:2610042 [2] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-160-103:2609918:2610042 [2] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-160-103:2609918:2610042 [2] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-160-103:2609918:2610042 [2] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-160-103:2609918:2610042 [2] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-160-103:2609918:2610042 [2] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-160-103:2609918:2610042 [2] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-160-103:2609918:2610042 [2] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-160-103:2609918:2610042 [2] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-160-103:2609918:2610042 [2] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-160-103:2609918:2610042 [2] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-160-103:2609918:2610042 [2] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-160-103:2609918:2610042 [2] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-160-103:2609918:2610042 [2] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-160-103:2609918:2610042 [2] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-160-103:2609918:2610042 [2] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-160-103:2609918:2610042 [2] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-160-103:2609918:2610042 [2] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-160-103:2609918:2610042 [2] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-160-103:2609918:2610042 [2] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-160-103:2609918:2610042 [2] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-160-103:2609918:2610042 [2] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-160-103:2609918:2610042 [2] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-160-103:2609918:2610042 [2] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-160-103:2609918:2610042 [2] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-160-103:2609918:2610042 [2] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-160-103:2609918:2610042 [2] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-160-103:2609918:2610042 [2] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-160-103:2609918:2610042 [2] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-160-103:2609918:2610042 [2] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-160-103:2609918:2610042 [2] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-160-103:2609918:2610042 [2] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-160-103:2609921:2610037 [5] NCCL INFO Using network AWS Libfabric
+ip-26-0-160-103:2609919:2610038 [3] NCCL INFO Using network AWS Libfabric
+ip-26-0-160-103:2609923:2610039 [7] NCCL INFO Using network AWS Libfabric
+ip-26-0-160-103:2609920:2610043 [4] NCCL INFO Using network AWS Libfabric
+ip-26-0-160-103:2609918:2610042 [2] NCCL INFO Using network AWS Libfabric
+ip-26-0-160-103:2609922:2610041 [6] NCCL INFO Using network AWS Libfabric
+ip-26-0-160-103:2609917:2610040 [1] NCCL INFO Using network AWS Libfabric
+ip-26-0-160-103:2609918:2610042 [2] NCCL INFO comm 0xa137660 rank 2 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0xb6efe550dff8b73b - Init START
+ip-26-0-160-103:2609917:2610040 [1] NCCL INFO comm 0x8f63d60 rank 1 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0xb6efe550dff8b73b - Init START
+ip-26-0-160-103:2609922:2610041 [6] NCCL INFO comm 0x965c530 rank 6 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0xb6efe550dff8b73b - Init START
+ip-26-0-160-103:2609921:2610037 [5] NCCL INFO comm 0x8d76730 rank 5 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0xb6efe550dff8b73b - Init START
+ip-26-0-160-103:2609923:2610039 [7] NCCL INFO comm 0xa228250 rank 7 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0xb6efe550dff8b73b - Init START
+ip-26-0-160-103:2609920:2610043 [4] NCCL INFO comm 0x910c300 rank 4 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0xb6efe550dff8b73b - Init START
+ip-26-0-160-103:2609919:2610038 [3] NCCL INFO comm 0x84d9a10 rank 3 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0xb6efe550dff8b73b - Init START
+ip-26-0-160-103:2609916:2610036 [0] NCCL INFO comm 0x9971740 rank 0 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0xb6efe550dff8b73b - Init START
+ip-26-0-160-103:2609917:2610040 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /opt/aws-ofi-nccl/share/aws-ofi-nccl/xml/p5.48xl-topo.xml
+ip-26-0-160-103:2609923:2610039 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /opt/aws-ofi-nccl/share/aws-ofi-nccl/xml/p5.48xl-topo.xml
+ip-26-0-160-103:2609918:2610042 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /opt/aws-ofi-nccl/share/aws-ofi-nccl/xml/p5.48xl-topo.xml
+ip-26-0-160-103:2609916:2610036 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /opt/aws-ofi-nccl/share/aws-ofi-nccl/xml/p5.48xl-topo.xml
+ip-26-0-160-103:2609922:2610041 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /opt/aws-ofi-nccl/share/aws-ofi-nccl/xml/p5.48xl-topo.xml
+ip-26-0-160-103:2609919:2610038 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /opt/aws-ofi-nccl/share/aws-ofi-nccl/xml/p5.48xl-topo.xml
+ip-26-0-160-103:2609921:2610037 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /opt/aws-ofi-nccl/share/aws-ofi-nccl/xml/p5.48xl-topo.xml
+ip-26-0-160-103:2609920:2610043 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /opt/aws-ofi-nccl/share/aws-ofi-nccl/xml/p5.48xl-topo.xml
+ip-26-0-160-103:2609916:2610036 [0] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609916:2610036 [0] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609916:2610036 [0] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609916:2610036 [0] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609916:2610036 [0] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609916:2610036 [0] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609916:2610036 [0] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609916:2610036 [0] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609916:2610036 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-160-103:2609916:2610036 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-160-103:2609917:2610040 [1] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609917:2610040 [1] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609917:2610040 [1] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609917:2610040 [1] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609917:2610040 [1] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609917:2610040 [1] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609917:2610040 [1] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609917:2610040 [1] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609918:2610042 [2] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609918:2610042 [2] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609918:2610042 [2] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609918:2610042 [2] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609918:2610042 [2] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609918:2610042 [2] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609918:2610042 [2] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609918:2610042 [2] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609920:2610043 [4] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609920:2610043 [4] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609920:2610043 [4] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609920:2610043 [4] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609920:2610043 [4] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609920:2610043 [4] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609920:2610043 [4] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609920:2610043 [4] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609921:2610037 [5] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609921:2610037 [5] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609921:2610037 [5] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609921:2610037 [5] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609921:2610037 [5] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609921:2610037 [5] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609921:2610037 [5] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609921:2610037 [5] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609917:2610040 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-160-103:2609917:2610040 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-160-103:2609918:2610042 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-160-103:2609918:2610042 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-160-103:2609920:2610043 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-160-103:2609920:2610043 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-160-103:2609921:2610037 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-160-103:2609921:2610037 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-160-103:2609919:2610038 [3] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609919:2610038 [3] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609919:2610038 [3] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609919:2610038 [3] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609919:2610038 [3] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609919:2610038 [3] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609919:2610038 [3] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609919:2610038 [3] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609923:2610039 [7] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609923:2610039 [7] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609923:2610039 [7] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609923:2610039 [7] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609923:2610039 [7] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609923:2610039 [7] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609923:2610039 [7] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609923:2610039 [7] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609919:2610038 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-160-103:2609922:2610041 [6] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609919:2610038 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-160-103:2609922:2610041 [6] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609922:2610041 [6] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609922:2610041 [6] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609922:2610041 [6] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609922:2610041 [6] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609922:2610041 [6] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609922:2610041 [6] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609923:2610039 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-160-103:2609923:2610039 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-160-103:2609922:2610041 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-160-103:2609922:2610041 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-160-103:2609923:2610039 [7] NCCL INFO Trees [0] -1/-1/-1->7->6 [1] -1/-1/-1->7->6 [2] -1/-1/-1->7->6 [3] -1/-1/-1->7->6 [4] -1/-1/-1->7->6 [5] -1/-1/-1->7->6 [6] -1/-1/-1->7->6 [7] -1/-1/-1->7->6 [8] -1/-1/-1->7->6 [9] -1/-1/-1->7->6 [10] -1/-1/-1->7->6 [11] -1/-1/-1->7->6 [12] -1/-1/-1->7->6 [13] -1/-1/-1->7->6 [14] -1/-1/-1->7->6 [15] -1/-1/-1->7->6 [16] -1/-1/-1->7->6 [17] -1/-1/-1->7->6 [18] -1/-1/-1->7->6 [19] -1/-1/-1->7->6 [20] -1/-1/-1->7->6 [21] -1/-1/-1->7->6 [22] -1/-1/-1->7->6 [23] -1/-1/-1->7->6
+ip-26-0-160-103:2609923:2610039 [7] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-160-103:2609916:2610036 [0] NCCL INFO Channel 00/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2609916:2610036 [0] NCCL INFO Channel 01/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2609916:2610036 [0] NCCL INFO Channel 02/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2609916:2610036 [0] NCCL INFO Channel 03/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2609916:2610036 [0] NCCL INFO Channel 04/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2609916:2610036 [0] NCCL INFO Channel 05/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2609918:2610042 [2] NCCL INFO Trees [0] 3/-1/-1->2->1 [1] 3/-1/-1->2->1 [2] 3/-1/-1->2->1 [3] 3/-1/-1->2->1 [4] 3/-1/-1->2->1 [5] 3/-1/-1->2->1 [6] 3/-1/-1->2->1 [7] 3/-1/-1->2->1 [8] 3/-1/-1->2->1 [9] 3/-1/-1->2->1 [10] 3/-1/-1->2->1 [11] 3/-1/-1->2->1 [12] 3/-1/-1->2->1 [13] 3/-1/-1->2->1 [14] 3/-1/-1->2->1 [15] 3/-1/-1->2->1 [16] 3/-1/-1->2->1 [17] 3/-1/-1->2->1 [18] 3/-1/-1->2->1 [19] 3/-1/-1->2->1 [20] 3/-1/-1->2->1 [21] 3/-1/-1->2->1 [22] 3/-1/-1->2->1 [23] 3/-1/-1->2->1
+ip-26-0-160-103:2609916:2610036 [0] NCCL INFO Channel 06/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2609916:2610036 [0] NCCL INFO Channel 07/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2609918:2610042 [2] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-160-103:2609916:2610036 [0] NCCL INFO Channel 08/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2609916:2610036 [0] NCCL INFO Channel 09/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2609916:2610036 [0] NCCL INFO Channel 10/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2609916:2610036 [0] NCCL INFO Channel 11/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2609916:2610036 [0] NCCL INFO Channel 12/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2609916:2610036 [0] NCCL INFO Channel 13/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2609916:2610036 [0] NCCL INFO Channel 14/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2609916:2610036 [0] NCCL INFO Channel 15/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2609916:2610036 [0] NCCL INFO Channel 16/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2609917:2610040 [1] NCCL INFO Trees [0] 2/-1/-1->1->0 [1] 2/-1/-1->1->0 [2] 2/-1/-1->1->0 [3] 2/-1/-1->1->0 [4] 2/-1/-1->1->0 [5] 2/-1/-1->1->0 [6] 2/-1/-1->1->0 [7] 2/-1/-1->1->0 [8] 2/-1/-1->1->0 [9] 2/-1/-1->1->0 [10] 2/-1/-1->1->0 [11] 2/-1/-1->1->0 [12] 2/-1/-1->1->0 [13] 2/-1/-1->1->0 [14] 2/-1/-1->1->0 [15] 2/-1/-1->1->0 [16] 2/-1/-1->1->0 [17] 2/-1/-1->1->0 [18] 2/-1/-1->1->0 [19] 2/-1/-1->1->0 [20] 2/-1/-1->1->0 [21] 2/-1/-1->1->0 [22] 2/-1/-1->1->0 [23] 2/-1/-1->1->0
+ip-26-0-160-103:2609916:2610036 [0] NCCL INFO Channel 17/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2609917:2610040 [1] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-160-103:2609916:2610036 [0] NCCL INFO Channel 18/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2609916:2610036 [0] NCCL INFO Channel 19/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2609916:2610036 [0] NCCL INFO Channel 20/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2609916:2610036 [0] NCCL INFO Channel 21/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2609916:2610036 [0] NCCL INFO Channel 22/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2609916:2610036 [0] NCCL INFO Channel 23/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2609916:2610036 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] 1/-1/-1->0->-1 [7] 1/-1/-1->0->-1 [8] 1/-1/-1->0->-1 [9] 1/-1/-1->0->-1 [10] 1/-1/-1->0->-1 [11] 1/-1/-1->0->-1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] 1/-1/-1->0->-1 [19] 1/-1/-1->0->-1 [20] 1/-1/-1->0->-1 [21] 1/-1/-1->0->-1 [22] 1/-1/-1->0->-1 [23] 1/-1/-1->0->-1
+ip-26-0-160-103:2609916:2610036 [0] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-160-103:2609919:2610038 [3] NCCL INFO Trees [0] 4/-1/-1->3->2 [1] 4/-1/-1->3->2 [2] 4/-1/-1->3->2 [3] 4/-1/-1->3->2 [4] 4/-1/-1->3->2 [5] 4/-1/-1->3->2 [6] 4/-1/-1->3->2 [7] 4/-1/-1->3->2 [8] 4/-1/-1->3->2 [9] 4/-1/-1->3->2 [10] 4/-1/-1->3->2 [11] 4/-1/-1->3->2 [12] 4/-1/-1->3->2 [13] 4/-1/-1->3->2 [14] 4/-1/-1->3->2 [15] 4/-1/-1->3->2 [16] 4/-1/-1->3->2 [17] 4/-1/-1->3->2 [18] 4/-1/-1->3->2 [19] 4/-1/-1->3->2 [20] 4/-1/-1->3->2 [21] 4/-1/-1->3->2 [22] 4/-1/-1->3->2 [23] 4/-1/-1->3->2
+ip-26-0-160-103:2609919:2610038 [3] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-160-103:2609922:2610041 [6] NCCL INFO Trees [0] 7/-1/-1->6->5 [1] 7/-1/-1->6->5 [2] 7/-1/-1->6->5 [3] 7/-1/-1->6->5 [4] 7/-1/-1->6->5 [5] 7/-1/-1->6->5 [6] 7/-1/-1->6->5 [7] 7/-1/-1->6->5 [8] 7/-1/-1->6->5 [9] 7/-1/-1->6->5 [10] 7/-1/-1->6->5 [11] 7/-1/-1->6->5 [12] 7/-1/-1->6->5 [13] 7/-1/-1->6->5 [14] 7/-1/-1->6->5 [15] 7/-1/-1->6->5 [16] 7/-1/-1->6->5 [17] 7/-1/-1->6->5 [18] 7/-1/-1->6->5 [19] 7/-1/-1->6->5 [20] 7/-1/-1->6->5 [21] 7/-1/-1->6->5 [22] 7/-1/-1->6->5 [23] 7/-1/-1->6->5
+ip-26-0-160-103:2609921:2610037 [5] NCCL INFO Trees [0] 6/-1/-1->5->4 [1] 6/-1/-1->5->4 [2] 6/-1/-1->5->4 [3] 6/-1/-1->5->4 [4] 6/-1/-1->5->4 [5] 6/-1/-1->5->4 [6] 6/-1/-1->5->4 [7] 6/-1/-1->5->4 [8] 6/-1/-1->5->4 [9] 6/-1/-1->5->4 [10] 6/-1/-1->5->4 [11] 6/-1/-1->5->4 [12] 6/-1/-1->5->4 [13] 6/-1/-1->5->4 [14] 6/-1/-1->5->4 [15] 6/-1/-1->5->4 [16] 6/-1/-1->5->4 [17] 6/-1/-1->5->4 [18] 6/-1/-1->5->4 [19] 6/-1/-1->5->4 [20] 6/-1/-1->5->4 [21] 6/-1/-1->5->4 [22] 6/-1/-1->5->4 [23] 6/-1/-1->5->4
+ip-26-0-160-103:2609922:2610041 [6] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-160-103:2609921:2610037 [5] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-160-103:2609920:2610043 [4] NCCL INFO Trees [0] 5/-1/-1->4->3 [1] 5/-1/-1->4->3 [2] 5/-1/-1->4->3 [3] 5/-1/-1->4->3 [4] 5/-1/-1->4->3 [5] 5/-1/-1->4->3 [6] 5/-1/-1->4->3 [7] 5/-1/-1->4->3 [8] 5/-1/-1->4->3 [9] 5/-1/-1->4->3 [10] 5/-1/-1->4->3 [11] 5/-1/-1->4->3 [12] 5/-1/-1->4->3 [13] 5/-1/-1->4->3 [14] 5/-1/-1->4->3 [15] 5/-1/-1->4->3 [16] 5/-1/-1->4->3 [17] 5/-1/-1->4->3 [18] 5/-1/-1->4->3 [19] 5/-1/-1->4->3 [20] 5/-1/-1->4->3 [21] 5/-1/-1->4->3 [22] 5/-1/-1->4->3 [23] 5/-1/-1->4->3
+ip-26-0-160-103:2609920:2610043 [4] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-160-103:2609918:2610042 [2] NCCL INFO Channel 00/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609918:2610042 [2] NCCL INFO Channel 01/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609916:2610036 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609918:2610042 [2] NCCL INFO Channel 02/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609916:2610036 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609918:2610042 [2] NCCL INFO Channel 03/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609919:2610038 [3] NCCL INFO Channel 00/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609921:2610037 [5] NCCL INFO Channel 00/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609916:2610036 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609918:2610042 [2] NCCL INFO Channel 04/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609919:2610038 [3] NCCL INFO Channel 01/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609920:2610043 [4] NCCL INFO Channel 00/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609921:2610037 [5] NCCL INFO Channel 01/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609919:2610038 [3] NCCL INFO Channel 02/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609918:2610042 [2] NCCL INFO Channel 05/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609916:2610036 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609920:2610043 [4] NCCL INFO Channel 01/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609921:2610037 [5] NCCL INFO Channel 02/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609918:2610042 [2] NCCL INFO Channel 06/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609919:2610038 [3] NCCL INFO Channel 03/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609916:2610036 [0] NCCL INFO Channel 04/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609917:2610040 [1] NCCL INFO Channel 00/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609920:2610043 [4] NCCL INFO Channel 02/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609921:2610037 [5] NCCL INFO Channel 03/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609918:2610042 [2] NCCL INFO Channel 07/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609922:2610041 [6] NCCL INFO Channel 00/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2609919:2610038 [3] NCCL INFO Channel 04/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609916:2610036 [0] NCCL INFO Channel 05/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609917:2610040 [1] NCCL INFO Channel 01/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609921:2610037 [5] NCCL INFO Channel 04/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609920:2610043 [4] NCCL INFO Channel 03/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609918:2610042 [2] NCCL INFO Channel 08/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609922:2610041 [6] NCCL INFO Channel 01/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2609916:2610036 [0] NCCL INFO Channel 06/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609919:2610038 [3] NCCL INFO Channel 05/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609917:2610040 [1] NCCL INFO Channel 02/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609923:2610039 [7] NCCL INFO Channel 00/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609921:2610037 [5] NCCL INFO Channel 05/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609920:2610043 [4] NCCL INFO Channel 04/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609918:2610042 [2] NCCL INFO Channel 09/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609922:2610041 [6] NCCL INFO Channel 02/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2609916:2610036 [0] NCCL INFO Channel 07/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609919:2610038 [3] NCCL INFO Channel 06/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609917:2610040 [1] NCCL INFO Channel 03/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609921:2610037 [5] NCCL INFO Channel 06/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609923:2610039 [7] NCCL INFO Channel 01/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609920:2610043 [4] NCCL INFO Channel 05/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609918:2610042 [2] NCCL INFO Channel 10/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609916:2610036 [0] NCCL INFO Channel 08/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609922:2610041 [6] NCCL INFO Channel 03/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2609919:2610038 [3] NCCL INFO Channel 07/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609917:2610040 [1] NCCL INFO Channel 04/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609921:2610037 [5] NCCL INFO Channel 07/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609920:2610043 [4] NCCL INFO Channel 06/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609918:2610042 [2] NCCL INFO Channel 11/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609916:2610036 [0] NCCL INFO Channel 09/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609922:2610041 [6] NCCL INFO Channel 04/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2609919:2610038 [3] NCCL INFO Channel 08/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609917:2610040 [1] NCCL INFO Channel 05/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609921:2610037 [5] NCCL INFO Channel 08/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609920:2610043 [4] NCCL INFO Channel 07/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609918:2610042 [2] NCCL INFO Channel 12/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609916:2610036 [0] NCCL INFO Channel 10/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609922:2610041 [6] NCCL INFO Channel 05/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2609919:2610038 [3] NCCL INFO Channel 09/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609917:2610040 [1] NCCL INFO Channel 06/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609921:2610037 [5] NCCL INFO Channel 09/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609920:2610043 [4] NCCL INFO Channel 08/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609918:2610042 [2] NCCL INFO Channel 13/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609916:2610036 [0] NCCL INFO Channel 11/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609922:2610041 [6] NCCL INFO Channel 06/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2609919:2610038 [3] NCCL INFO Channel 10/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609917:2610040 [1] NCCL INFO Channel 07/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609920:2610043 [4] NCCL INFO Channel 09/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609921:2610037 [5] NCCL INFO Channel 10/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609918:2610042 [2] NCCL INFO Channel 14/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609916:2610036 [0] NCCL INFO Channel 12/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609919:2610038 [3] NCCL INFO Channel 11/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609922:2610041 [6] NCCL INFO Channel 07/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2609917:2610040 [1] NCCL INFO Channel 08/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609920:2610043 [4] NCCL INFO Channel 10/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609921:2610037 [5] NCCL INFO Channel 11/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609918:2610042 [2] NCCL INFO Channel 15/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609916:2610036 [0] NCCL INFO Channel 13/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609919:2610038 [3] NCCL INFO Channel 12/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609922:2610041 [6] NCCL INFO Channel 08/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2609917:2610040 [1] NCCL INFO Channel 09/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609920:2610043 [4] NCCL INFO Channel 11/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609921:2610037 [5] NCCL INFO Channel 12/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609918:2610042 [2] NCCL INFO Channel 16/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609916:2610036 [0] NCCL INFO Channel 14/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609919:2610038 [3] NCCL INFO Channel 13/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609922:2610041 [6] NCCL INFO Channel 09/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2609917:2610040 [1] NCCL INFO Channel 10/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609920:2610043 [4] NCCL INFO Channel 12/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609921:2610037 [5] NCCL INFO Channel 13/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609918:2610042 [2] NCCL INFO Channel 17/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609916:2610036 [0] NCCL INFO Channel 15/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609919:2610038 [3] NCCL INFO Channel 14/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609922:2610041 [6] NCCL INFO Channel 10/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2609917:2610040 [1] NCCL INFO Channel 11/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609920:2610043 [4] NCCL INFO Channel 13/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609921:2610037 [5] NCCL INFO Channel 14/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609918:2610042 [2] NCCL INFO Channel 18/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609916:2610036 [0] NCCL INFO Channel 16/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609919:2610038 [3] NCCL INFO Channel 15/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609922:2610041 [6] NCCL INFO Channel 11/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2609917:2610040 [1] NCCL INFO Channel 12/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609920:2610043 [4] NCCL INFO Channel 14/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609921:2610037 [5] NCCL INFO Channel 15/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609923:2610039 [7] NCCL INFO Channel 02/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609919:2610038 [3] NCCL INFO Channel 16/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609920:2610043 [4] NCCL INFO Channel 15/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609921:2610037 [5] NCCL INFO Channel 16/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609923:2610039 [7] NCCL INFO Channel 03/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609918:2610042 [2] NCCL INFO Channel 19/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609916:2610036 [0] NCCL INFO Channel 17/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609922:2610041 [6] NCCL INFO Channel 12/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2609917:2610040 [1] NCCL INFO Channel 13/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609918:2610042 [2] NCCL INFO Channel 20/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609916:2610036 [0] NCCL INFO Channel 18/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609921:2610037 [5] NCCL INFO Channel 17/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609922:2610041 [6] NCCL INFO Channel 13/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2609917:2610040 [1] NCCL INFO Channel 14/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609918:2610042 [2] NCCL INFO Channel 21/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609916:2610036 [0] NCCL INFO Channel 19/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609921:2610037 [5] NCCL INFO Channel 18/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609917:2610040 [1] NCCL INFO Channel 15/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609922:2610041 [6] NCCL INFO Channel 14/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2609918:2610042 [2] NCCL INFO Channel 22/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609916:2610036 [0] NCCL INFO Channel 20/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609921:2610037 [5] NCCL INFO Channel 19/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609917:2610040 [1] NCCL INFO Channel 16/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609922:2610041 [6] NCCL INFO Channel 15/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2609918:2610042 [2] NCCL INFO Channel 23/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609916:2610036 [0] NCCL INFO Channel 21/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609921:2610037 [5] NCCL INFO Channel 20/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609917:2610040 [1] NCCL INFO Channel 17/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609922:2610041 [6] NCCL INFO Channel 16/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2609916:2610036 [0] NCCL INFO Channel 22/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609921:2610037 [5] NCCL INFO Channel 21/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609917:2610040 [1] NCCL INFO Channel 18/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609922:2610041 [6] NCCL INFO Channel 17/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2609916:2610036 [0] NCCL INFO Channel 23/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609921:2610037 [5] NCCL INFO Channel 22/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609917:2610040 [1] NCCL INFO Channel 19/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609922:2610041 [6] NCCL INFO Channel 18/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2609919:2610038 [3] NCCL INFO Channel 17/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609920:2610043 [4] NCCL INFO Channel 16/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609921:2610037 [5] NCCL INFO Channel 23/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609917:2610040 [1] NCCL INFO Channel 20/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609923:2610039 [7] NCCL INFO Channel 04/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609922:2610041 [6] NCCL INFO Channel 19/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2609919:2610038 [3] NCCL INFO Channel 18/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609920:2610043 [4] NCCL INFO Channel 17/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609917:2610040 [1] NCCL INFO Channel 21/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609923:2610039 [7] NCCL INFO Channel 05/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609922:2610041 [6] NCCL INFO Channel 20/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2609919:2610038 [3] NCCL INFO Channel 19/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609920:2610043 [4] NCCL INFO Channel 18/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609917:2610040 [1] NCCL INFO Channel 22/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609923:2610039 [7] NCCL INFO Channel 06/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609922:2610041 [6] NCCL INFO Channel 21/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2609919:2610038 [3] NCCL INFO Channel 20/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609920:2610043 [4] NCCL INFO Channel 19/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609923:2610039 [7] NCCL INFO Channel 07/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609917:2610040 [1] NCCL INFO Channel 23/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609922:2610041 [6] NCCL INFO Channel 22/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2609919:2610038 [3] NCCL INFO Channel 21/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609920:2610043 [4] NCCL INFO Channel 20/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609923:2610039 [7] NCCL INFO Channel 08/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609922:2610041 [6] NCCL INFO Channel 23/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2609919:2610038 [3] NCCL INFO Channel 22/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609920:2610043 [4] NCCL INFO Channel 21/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609923:2610039 [7] NCCL INFO Channel 09/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609919:2610038 [3] NCCL INFO Channel 23/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609920:2610043 [4] NCCL INFO Channel 22/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609923:2610039 [7] NCCL INFO Channel 10/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609920:2610043 [4] NCCL INFO Channel 23/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609923:2610039 [7] NCCL INFO Channel 11/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609923:2610039 [7] NCCL INFO Channel 12/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609923:2610039 [7] NCCL INFO Channel 13/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609923:2610039 [7] NCCL INFO Channel 14/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609923:2610039 [7] NCCL INFO Channel 15/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609923:2610039 [7] NCCL INFO Channel 16/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609923:2610039 [7] NCCL INFO Channel 17/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609923:2610039 [7] NCCL INFO Channel 18/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609923:2610039 [7] NCCL INFO Channel 19/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609923:2610039 [7] NCCL INFO Channel 20/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609923:2610039 [7] NCCL INFO Channel 21/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609923:2610039 [7] NCCL INFO Channel 22/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609923:2610039 [7] NCCL INFO Channel 23/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609920:2610043 [4] NCCL INFO Connected all rings
+ip-26-0-160-103:2609921:2610037 [5] NCCL INFO Connected all rings
+ip-26-0-160-103:2609917:2610040 [1] NCCL INFO Connected all rings
+ip-26-0-160-103:2609918:2610042 [2] NCCL INFO Connected all rings
+ip-26-0-160-103:2609919:2610038 [3] NCCL INFO Connected all rings
+ip-26-0-160-103:2609923:2610039 [7] NCCL INFO Connected all rings
+ip-26-0-160-103:2609923:2610039 [7] NCCL INFO Channel 00/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609923:2610039 [7] NCCL INFO Channel 01/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609923:2610039 [7] NCCL INFO Channel 02/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609916:2610036 [0] NCCL INFO Connected all rings
+ip-26-0-160-103:2609923:2610039 [7] NCCL INFO Channel 03/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609923:2610039 [7] NCCL INFO Channel 04/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609923:2610039 [7] NCCL INFO Channel 05/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609923:2610039 [7] NCCL INFO Channel 06/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609923:2610039 [7] NCCL INFO Channel 07/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609922:2610041 [6] NCCL INFO Connected all rings
+ip-26-0-160-103:2609923:2610039 [7] NCCL INFO Channel 08/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609923:2610039 [7] NCCL INFO Channel 09/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609923:2610039 [7] NCCL INFO Channel 10/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609917:2610040 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609923:2610039 [7] NCCL INFO Channel 11/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609920:2610043 [4] NCCL INFO Channel 00/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609918:2610042 [2] NCCL INFO Channel 00/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609917:2610040 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609923:2610039 [7] NCCL INFO Channel 12/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609920:2610043 [4] NCCL INFO Channel 01/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609918:2610042 [2] NCCL INFO Channel 01/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609917:2610040 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609920:2610043 [4] NCCL INFO Channel 02/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609918:2610042 [2] NCCL INFO Channel 02/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609923:2610039 [7] NCCL INFO Channel 13/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609917:2610040 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609920:2610043 [4] NCCL INFO Channel 03/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609918:2610042 [2] NCCL INFO Channel 03/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609923:2610039 [7] NCCL INFO Channel 14/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609917:2610040 [1] NCCL INFO Channel 04/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609920:2610043 [4] NCCL INFO Channel 04/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609918:2610042 [2] NCCL INFO Channel 04/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609923:2610039 [7] NCCL INFO Channel 15/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609917:2610040 [1] NCCL INFO Channel 05/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609920:2610043 [4] NCCL INFO Channel 05/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609923:2610039 [7] NCCL INFO Channel 16/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609918:2610042 [2] NCCL INFO Channel 05/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609917:2610040 [1] NCCL INFO Channel 06/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609920:2610043 [4] NCCL INFO Channel 06/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609918:2610042 [2] NCCL INFO Channel 06/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609923:2610039 [7] NCCL INFO Channel 17/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609917:2610040 [1] NCCL INFO Channel 07/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609920:2610043 [4] NCCL INFO Channel 07/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609923:2610039 [7] NCCL INFO Channel 18/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609918:2610042 [2] NCCL INFO Channel 07/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609919:2610038 [3] NCCL INFO Channel 00/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609917:2610040 [1] NCCL INFO Channel 08/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609920:2610043 [4] NCCL INFO Channel 08/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609923:2610039 [7] NCCL INFO Channel 19/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609918:2610042 [2] NCCL INFO Channel 08/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609919:2610038 [3] NCCL INFO Channel 01/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609920:2610043 [4] NCCL INFO Channel 09/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609917:2610040 [1] NCCL INFO Channel 09/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609923:2610039 [7] NCCL INFO Channel 20/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609918:2610042 [2] NCCL INFO Channel 09/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609919:2610038 [3] NCCL INFO Channel 02/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609920:2610043 [4] NCCL INFO Channel 10/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609917:2610040 [1] NCCL INFO Channel 10/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609923:2610039 [7] NCCL INFO Channel 21/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609918:2610042 [2] NCCL INFO Channel 10/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609919:2610038 [3] NCCL INFO Channel 03/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609920:2610043 [4] NCCL INFO Channel 11/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609917:2610040 [1] NCCL INFO Channel 11/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609921:2610037 [5] NCCL INFO Channel 00/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609923:2610039 [7] NCCL INFO Channel 22/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609918:2610042 [2] NCCL INFO Channel 11/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609919:2610038 [3] NCCL INFO Channel 04/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609920:2610043 [4] NCCL INFO Channel 12/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609917:2610040 [1] NCCL INFO Channel 12/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609921:2610037 [5] NCCL INFO Channel 01/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609923:2610039 [7] NCCL INFO Channel 23/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609918:2610042 [2] NCCL INFO Channel 12/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609919:2610038 [3] NCCL INFO Channel 05/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609920:2610043 [4] NCCL INFO Channel 13/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609917:2610040 [1] NCCL INFO Channel 13/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609921:2610037 [5] NCCL INFO Channel 02/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609918:2610042 [2] NCCL INFO Channel 13/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609919:2610038 [3] NCCL INFO Channel 06/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609920:2610043 [4] NCCL INFO Channel 14/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609917:2610040 [1] NCCL INFO Channel 14/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609921:2610037 [5] NCCL INFO Channel 03/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609918:2610042 [2] NCCL INFO Channel 14/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609920:2610043 [4] NCCL INFO Channel 15/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609919:2610038 [3] NCCL INFO Channel 07/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609917:2610040 [1] NCCL INFO Channel 15/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609921:2610037 [5] NCCL INFO Channel 04/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609918:2610042 [2] NCCL INFO Channel 15/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609920:2610043 [4] NCCL INFO Channel 16/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609919:2610038 [3] NCCL INFO Channel 08/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609917:2610040 [1] NCCL INFO Channel 16/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609918:2610042 [2] NCCL INFO Channel 16/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609918:2610042 [2] NCCL INFO Channel 17/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609919:2610038 [3] NCCL INFO Channel 09/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609918:2610042 [2] NCCL INFO Channel 18/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609919:2610038 [3] NCCL INFO Channel 10/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609918:2610042 [2] NCCL INFO Channel 19/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609919:2610038 [3] NCCL INFO Channel 11/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609918:2610042 [2] NCCL INFO Channel 20/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609919:2610038 [3] NCCL INFO Channel 12/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609922:2610041 [6] NCCL INFO Channel 00/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609918:2610042 [2] NCCL INFO Channel 21/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609920:2610043 [4] NCCL INFO Channel 17/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609917:2610040 [1] NCCL INFO Channel 17/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609919:2610038 [3] NCCL INFO Channel 13/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609922:2610041 [6] NCCL INFO Channel 01/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609918:2610042 [2] NCCL INFO Channel 22/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609920:2610043 [4] NCCL INFO Channel 18/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609917:2610040 [1] NCCL INFO Channel 18/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609919:2610038 [3] NCCL INFO Channel 14/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609922:2610041 [6] NCCL INFO Channel 02/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609918:2610042 [2] NCCL INFO Channel 23/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609920:2610043 [4] NCCL INFO Channel 19/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609917:2610040 [1] NCCL INFO Channel 19/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609919:2610038 [3] NCCL INFO Channel 15/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609922:2610041 [6] NCCL INFO Channel 03/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609920:2610043 [4] NCCL INFO Channel 20/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609921:2610037 [5] NCCL INFO Channel 05/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609917:2610040 [1] NCCL INFO Channel 20/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609919:2610038 [3] NCCL INFO Channel 16/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609922:2610041 [6] NCCL INFO Channel 04/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609921:2610037 [5] NCCL INFO Channel 06/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609917:2610040 [1] NCCL INFO Channel 21/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609919:2610038 [3] NCCL INFO Channel 17/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609922:2610041 [6] NCCL INFO Channel 05/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609917:2610040 [1] NCCL INFO Channel 22/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609919:2610038 [3] NCCL INFO Channel 18/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609920:2610043 [4] NCCL INFO Channel 21/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609917:2610040 [1] NCCL INFO Channel 23/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609919:2610038 [3] NCCL INFO Channel 19/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609920:2610043 [4] NCCL INFO Channel 22/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609921:2610037 [5] NCCL INFO Channel 07/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609919:2610038 [3] NCCL INFO Channel 20/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609920:2610043 [4] NCCL INFO Channel 23/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609921:2610037 [5] NCCL INFO Channel 08/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609919:2610038 [3] NCCL INFO Channel 21/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609921:2610037 [5] NCCL INFO Channel 09/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609922:2610041 [6] NCCL INFO Channel 06/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609919:2610038 [3] NCCL INFO Channel 22/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609921:2610037 [5] NCCL INFO Channel 10/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609919:2610038 [3] NCCL INFO Channel 23/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609922:2610041 [6] NCCL INFO Channel 07/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609921:2610037 [5] NCCL INFO Channel 11/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609922:2610041 [6] NCCL INFO Channel 08/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609921:2610037 [5] NCCL INFO Channel 12/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609921:2610037 [5] NCCL INFO Channel 13/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609922:2610041 [6] NCCL INFO Channel 09/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609921:2610037 [5] NCCL INFO Channel 14/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609922:2610041 [6] NCCL INFO Channel 10/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609921:2610037 [5] NCCL INFO Channel 15/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609922:2610041 [6] NCCL INFO Channel 11/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609921:2610037 [5] NCCL INFO Channel 16/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609922:2610041 [6] NCCL INFO Channel 12/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609921:2610037 [5] NCCL INFO Channel 17/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609921:2610037 [5] NCCL INFO Channel 18/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609922:2610041 [6] NCCL INFO Channel 13/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609921:2610037 [5] NCCL INFO Channel 19/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609922:2610041 [6] NCCL INFO Channel 14/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609921:2610037 [5] NCCL INFO Channel 20/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609922:2610041 [6] NCCL INFO Channel 15/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609921:2610037 [5] NCCL INFO Channel 21/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609922:2610041 [6] NCCL INFO Channel 16/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609921:2610037 [5] NCCL INFO Channel 22/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609922:2610041 [6] NCCL INFO Channel 17/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609921:2610037 [5] NCCL INFO Channel 23/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609922:2610041 [6] NCCL INFO Channel 18/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609916:2610036 [0] NCCL INFO Connected all trees
+ip-26-0-160-103:2609916:2610036 [0] NCCL INFO NVLS comm 0x9971740 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-103:2609922:2610041 [6] NCCL INFO Channel 19/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609922:2610041 [6] NCCL INFO Channel 20/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609922:2610041 [6] NCCL INFO Channel 21/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609922:2610041 [6] NCCL INFO Channel 22/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609922:2610041 [6] NCCL INFO Channel 23/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609919:2610038 [3] NCCL INFO Connected all trees
+ip-26-0-160-103:2609919:2610038 [3] NCCL INFO NVLS comm 0x84d9a10 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-103:2609917:2610040 [1] NCCL INFO Connected all trees
+ip-26-0-160-103:2609917:2610040 [1] NCCL INFO NVLS comm 0x8f63d60 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-103:2609923:2610039 [7] NCCL INFO Connected all trees
+ip-26-0-160-103:2609923:2610039 [7] NCCL INFO NVLS comm 0xa228250 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-103:2609918:2610042 [2] NCCL INFO Connected all trees
+ip-26-0-160-103:2609918:2610042 [2] NCCL INFO NVLS comm 0xa137660 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-103:2609920:2610043 [4] NCCL INFO Connected all trees
+ip-26-0-160-103:2609920:2610043 [4] NCCL INFO NVLS comm 0x910c300 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-103:2609921:2610037 [5] NCCL INFO Connected all trees
+ip-26-0-160-103:2609921:2610037 [5] NCCL INFO NVLS comm 0x8d76730 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-103:2609922:2610041 [6] NCCL INFO Connected all trees
+ip-26-0-160-103:2609922:2610041 [6] NCCL INFO NVLS comm 0x965c530 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-103:2609917:2610040 [1] NCCL INFO Connected NVLS tree
+ip-26-0-160-103:2609917:2610040 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-103:2609923:2610039 [7] NCCL INFO Connected NVLS tree
+ip-26-0-160-103:2609917:2610040 [1] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-160-103:2609917:2610040 [1] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-160-103:2609923:2610039 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-103:2609923:2610039 [7] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-160-103:2609923:2610039 [7] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-160-103:2609921:2610037 [5] NCCL INFO Connected NVLS tree
+ip-26-0-160-103:2609921:2610037 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-103:2609921:2610037 [5] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-160-103:2609921:2610037 [5] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-160-103:2609922:2610041 [6] NCCL INFO Connected NVLS tree
+ip-26-0-160-103:2609922:2610041 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-103:2609922:2610041 [6] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-160-103:2609922:2610041 [6] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-160-103:2609919:2610038 [3] NCCL INFO Connected NVLS tree
+ip-26-0-160-103:2609919:2610038 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-103:2609919:2610038 [3] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-160-103:2609919:2610038 [3] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-160-103:2609918:2610042 [2] NCCL INFO Connected NVLS tree
+ip-26-0-160-103:2609918:2610042 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-103:2609918:2610042 [2] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-160-103:2609918:2610042 [2] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-160-103:2609920:2610043 [4] NCCL INFO Connected NVLS tree
+ip-26-0-160-103:2609916:2610036 [0] NCCL INFO Connected NVLS tree
+ip-26-0-160-103:2609916:2610036 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-103:2609920:2610043 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-103:2609920:2610043 [4] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-160-103:2609920:2610043 [4] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-160-103:2609916:2610036 [0] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-160-103:2609916:2610036 [0] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-160-103:2609921:2610037 [5] NCCL INFO comm 0x8d76730 rank 5 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0xb6efe550dff8b73b - Init COMPLETE
+ip-26-0-160-103:2609923:2610039 [7] NCCL INFO comm 0xa228250 rank 7 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0xb6efe550dff8b73b - Init COMPLETE
+ip-26-0-160-103:2609922:2610041 [6] NCCL INFO comm 0x965c530 rank 6 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0xb6efe550dff8b73b - Init COMPLETE
+ip-26-0-160-103:2609918:2610042 [2] NCCL INFO comm 0xa137660 rank 2 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0xb6efe550dff8b73b - Init COMPLETE
+ip-26-0-160-103:2609919:2610038 [3] NCCL INFO comm 0x84d9a10 rank 3 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0xb6efe550dff8b73b - Init COMPLETE
+ip-26-0-160-103:2609917:2610040 [1] NCCL INFO comm 0x8f63d60 rank 1 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0xb6efe550dff8b73b - Init COMPLETE
+ip-26-0-160-103:2609920:2610043 [4] NCCL INFO comm 0x910c300 rank 4 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0xb6efe550dff8b73b - Init COMPLETE
+ip-26-0-160-103:2609916:2610036 [0] NCCL INFO comm 0x9971740 rank 0 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0xb6efe550dff8b73b - Init COMPLETE
+ip-26-0-160-103:2609916:2610111 [0] NCCL INFO Using network AWS Libfabric
+ip-26-0-160-103:2609922:2610112 [6] NCCL INFO Using network AWS Libfabric
+ip-26-0-160-103:2609920:2610113 [4] NCCL INFO Using network AWS Libfabric
+ip-26-0-160-103:2609918:2610114 [2] NCCL INFO Using network AWS Libfabric
+ip-26-0-160-103:2609919:2610116 [3] NCCL INFO Using network AWS Libfabric
+ip-26-0-160-103:2609921:2610115 [5] NCCL INFO Using network AWS Libfabric
+ip-26-0-160-103:2609923:2610117 [7] NCCL INFO Using network AWS Libfabric
+ip-26-0-160-103:2609917:2610118 [1] NCCL INFO Using network AWS Libfabric
+ip-26-0-160-103:2609917:2610118 [1] NCCL INFO comm 0x918d5d0 rank 1 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0xfa772de19d2f0956 - Init START
+ip-26-0-160-103:2609916:2610111 [0] NCCL INFO comm 0x9b9b3e0 rank 0 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0xfa772de19d2f0956 - Init START
+ip-26-0-160-103:2609923:2610117 [7] NCCL INFO comm 0xa451a50 rank 7 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0xfa772de19d2f0956 - Init START
+ip-26-0-160-103:2609922:2610112 [6] NCCL INFO comm 0x98863c0 rank 6 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0xfa772de19d2f0956 - Init START
+ip-26-0-160-103:2609918:2610114 [2] NCCL INFO comm 0xa360d80 rank 2 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0xfa772de19d2f0956 - Init START
+ip-26-0-160-103:2609921:2610115 [5] NCCL INFO comm 0x8f9ffa0 rank 5 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0xfa772de19d2f0956 - Init START
+ip-26-0-160-103:2609919:2610116 [3] NCCL INFO comm 0x8703130 rank 3 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0xfa772de19d2f0956 - Init START
+ip-26-0-160-103:2609920:2610113 [4] NCCL INFO comm 0x9335e20 rank 4 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0xfa772de19d2f0956 - Init START
+ip-26-0-160-103:2609916:2610111 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /opt/aws-ofi-nccl/share/aws-ofi-nccl/xml/p5.48xl-topo.xml
+ip-26-0-160-103:2609922:2610112 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /opt/aws-ofi-nccl/share/aws-ofi-nccl/xml/p5.48xl-topo.xml
+ip-26-0-160-103:2609917:2610118 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /opt/aws-ofi-nccl/share/aws-ofi-nccl/xml/p5.48xl-topo.xml
+ip-26-0-160-103:2609920:2610113 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /opt/aws-ofi-nccl/share/aws-ofi-nccl/xml/p5.48xl-topo.xml
+ip-26-0-160-103:2609919:2610116 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /opt/aws-ofi-nccl/share/aws-ofi-nccl/xml/p5.48xl-topo.xml
+ip-26-0-160-103:2609918:2610114 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /opt/aws-ofi-nccl/share/aws-ofi-nccl/xml/p5.48xl-topo.xml
+ip-26-0-160-103:2609923:2610117 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /opt/aws-ofi-nccl/share/aws-ofi-nccl/xml/p5.48xl-topo.xml
+ip-26-0-160-103:2609921:2610115 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /opt/aws-ofi-nccl/share/aws-ofi-nccl/xml/p5.48xl-topo.xml
+ip-26-0-160-103:2609919:2610116 [3] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609919:2610116 [3] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609919:2610116 [3] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609919:2610116 [3] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609919:2610116 [3] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609919:2610116 [3] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609919:2610116 [3] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609919:2610116 [3] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609919:2610116 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-160-103:2609919:2610116 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-160-103:2609921:2610115 [5] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609921:2610115 [5] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609921:2610115 [5] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609921:2610115 [5] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609921:2610115 [5] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609921:2610115 [5] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609921:2610115 [5] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609921:2610115 [5] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609921:2610115 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-160-103:2609921:2610115 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-160-103:2609917:2610118 [1] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609917:2610118 [1] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609917:2610118 [1] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609917:2610118 [1] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609917:2610118 [1] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609917:2610118 [1] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609917:2610118 [1] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609917:2610118 [1] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609917:2610118 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-160-103:2609917:2610118 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-160-103:2609920:2610113 [4] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609920:2610113 [4] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609920:2610113 [4] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609920:2610113 [4] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609920:2610113 [4] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609920:2610113 [4] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609920:2610113 [4] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609920:2610113 [4] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609920:2610113 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-160-103:2609920:2610113 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-160-103:2609923:2610117 [7] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609923:2610117 [7] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609923:2610117 [7] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609923:2610117 [7] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609923:2610117 [7] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609923:2610117 [7] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609923:2610117 [7] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609923:2610117 [7] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609923:2610117 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-160-103:2609923:2610117 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-160-103:2609922:2610112 [6] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609922:2610112 [6] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609922:2610112 [6] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609922:2610112 [6] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609922:2610112 [6] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609922:2610112 [6] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609922:2610112 [6] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609922:2610112 [6] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609922:2610112 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-160-103:2609922:2610112 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-160-103:2609918:2610114 [2] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609918:2610114 [2] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609918:2610114 [2] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609918:2610114 [2] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609918:2610114 [2] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609918:2610114 [2] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609918:2610114 [2] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609918:2610114 [2] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609918:2610114 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-160-103:2609918:2610114 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-160-103:2609916:2610111 [0] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609916:2610111 [0] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609916:2610111 [0] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609916:2610111 [0] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609916:2610111 [0] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609916:2610111 [0] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609916:2610111 [0] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609916:2610111 [0] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609916:2610111 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-160-103:2609916:2610111 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-160-103:2609916:2610111 [0] NCCL INFO Channel 00/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2609916:2610111 [0] NCCL INFO Channel 01/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2609916:2610111 [0] NCCL INFO Channel 02/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2609916:2610111 [0] NCCL INFO Channel 03/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2609916:2610111 [0] NCCL INFO Channel 04/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2609916:2610111 [0] NCCL INFO Channel 05/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2609916:2610111 [0] NCCL INFO Channel 06/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2609917:2610118 [1] NCCL INFO Trees [0] 2/-1/-1->1->0 [1] 2/-1/-1->1->0 [2] 2/-1/-1->1->0 [3] 2/-1/-1->1->0 [4] 2/-1/-1->1->0 [5] 2/-1/-1->1->0 [6] 2/-1/-1->1->0 [7] 2/-1/-1->1->0 [8] 2/-1/-1->1->0 [9] 2/-1/-1->1->0 [10] 2/-1/-1->1->0 [11] 2/-1/-1->1->0 [12] 2/-1/-1->1->0 [13] 2/-1/-1->1->0 [14] 2/-1/-1->1->0 [15] 2/-1/-1->1->0 [16] 2/-1/-1->1->0 [17] 2/-1/-1->1->0 [18] 2/-1/-1->1->0 [19] 2/-1/-1->1->0 [20] 2/-1/-1->1->0 [21] 2/-1/-1->1->0 [22] 2/-1/-1->1->0 [23] 2/-1/-1->1->0
+ip-26-0-160-103:2609916:2610111 [0] NCCL INFO Channel 07/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2609916:2610111 [0] NCCL INFO Channel 08/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2609917:2610118 [1] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-160-103:2609916:2610111 [0] NCCL INFO Channel 09/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2609916:2610111 [0] NCCL INFO Channel 10/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2609918:2610114 [2] NCCL INFO Trees [0] 3/-1/-1->2->1 [1] 3/-1/-1->2->1 [2] 3/-1/-1->2->1 [3] 3/-1/-1->2->1 [4] 3/-1/-1->2->1 [5] 3/-1/-1->2->1 [6] 3/-1/-1->2->1 [7] 3/-1/-1->2->1 [8] 3/-1/-1->2->1 [9] 3/-1/-1->2->1 [10] 3/-1/-1->2->1 [11] 3/-1/-1->2->1 [12] 3/-1/-1->2->1 [13] 3/-1/-1->2->1 [14] 3/-1/-1->2->1 [15] 3/-1/-1->2->1 [16] 3/-1/-1->2->1 [17] 3/-1/-1->2->1 [18] 3/-1/-1->2->1 [19] 3/-1/-1->2->1 [20] 3/-1/-1->2->1 [21] 3/-1/-1->2->1 [22] 3/-1/-1->2->1 [23] 3/-1/-1->2->1
+ip-26-0-160-103:2609916:2610111 [0] NCCL INFO Channel 11/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2609918:2610114 [2] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-160-103:2609923:2610117 [7] NCCL INFO Trees [0] -1/-1/-1->7->6 [1] -1/-1/-1->7->6 [2] -1/-1/-1->7->6 [3] -1/-1/-1->7->6 [4] -1/-1/-1->7->6 [5] -1/-1/-1->7->6 [6] -1/-1/-1->7->6 [7] -1/-1/-1->7->6 [8] -1/-1/-1->7->6 [9] -1/-1/-1->7->6 [10] -1/-1/-1->7->6 [11] -1/-1/-1->7->6 [12] -1/-1/-1->7->6 [13] -1/-1/-1->7->6 [14] -1/-1/-1->7->6 [15] -1/-1/-1->7->6 [16] -1/-1/-1->7->6 [17] -1/-1/-1->7->6 [18] -1/-1/-1->7->6 [19] -1/-1/-1->7->6 [20] -1/-1/-1->7->6 [21] -1/-1/-1->7->6 [22] -1/-1/-1->7->6 [23] -1/-1/-1->7->6
+ip-26-0-160-103:2609916:2610111 [0] NCCL INFO Channel 12/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2609923:2610117 [7] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-160-103:2609916:2610111 [0] NCCL INFO Channel 13/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2609922:2610112 [6] NCCL INFO Trees [0] 7/-1/-1->6->5 [1] 7/-1/-1->6->5 [2] 7/-1/-1->6->5 [3] 7/-1/-1->6->5 [4] 7/-1/-1->6->5 [5] 7/-1/-1->6->5 [6] 7/-1/-1->6->5 [7] 7/-1/-1->6->5 [8] 7/-1/-1->6->5 [9] 7/-1/-1->6->5 [10] 7/-1/-1->6->5 [11] 7/-1/-1->6->5 [12] 7/-1/-1->6->5 [13] 7/-1/-1->6->5 [14] 7/-1/-1->6->5 [15] 7/-1/-1->6->5 [16] 7/-1/-1->6->5 [17] 7/-1/-1->6->5 [18] 7/-1/-1->6->5 [19] 7/-1/-1->6->5 [20] 7/-1/-1->6->5 [21] 7/-1/-1->6->5 [22] 7/-1/-1->6->5 [23] 7/-1/-1->6->5
+ip-26-0-160-103:2609916:2610111 [0] NCCL INFO Channel 14/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2609922:2610112 [6] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-160-103:2609921:2610115 [5] NCCL INFO Trees [0] 6/-1/-1->5->4 [1] 6/-1/-1->5->4 [2] 6/-1/-1->5->4 [3] 6/-1/-1->5->4 [4] 6/-1/-1->5->4 [5] 6/-1/-1->5->4 [6] 6/-1/-1->5->4 [7] 6/-1/-1->5->4 [8] 6/-1/-1->5->4 [9] 6/-1/-1->5->4 [10] 6/-1/-1->5->4 [11] 6/-1/-1->5->4 [12] 6/-1/-1->5->4 [13] 6/-1/-1->5->4 [14] 6/-1/-1->5->4 [15] 6/-1/-1->5->4 [16] 6/-1/-1->5->4 [17] 6/-1/-1->5->4 [18] 6/-1/-1->5->4 [19] 6/-1/-1->5->4 [20] 6/-1/-1->5->4 [21] 6/-1/-1->5->4 [22] 6/-1/-1->5->4 [23] 6/-1/-1->5->4
+ip-26-0-160-103:2609916:2610111 [0] NCCL INFO Channel 15/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2609921:2610115 [5] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-160-103:2609916:2610111 [0] NCCL INFO Channel 16/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2609920:2610113 [4] NCCL INFO Trees [0] 5/-1/-1->4->3 [1] 5/-1/-1->4->3 [2] 5/-1/-1->4->3 [3] 5/-1/-1->4->3 [4] 5/-1/-1->4->3 [5] 5/-1/-1->4->3 [6] 5/-1/-1->4->3 [7] 5/-1/-1->4->3 [8] 5/-1/-1->4->3 [9] 5/-1/-1->4->3 [10] 5/-1/-1->4->3 [11] 5/-1/-1->4->3 [12] 5/-1/-1->4->3 [13] 5/-1/-1->4->3 [14] 5/-1/-1->4->3 [15] 5/-1/-1->4->3 [16] 5/-1/-1->4->3 [17] 5/-1/-1->4->3 [18] 5/-1/-1->4->3 [19] 5/-1/-1->4->3 [20] 5/-1/-1->4->3 [21] 5/-1/-1->4->3 [22] 5/-1/-1->4->3 [23] 5/-1/-1->4->3
+ip-26-0-160-103:2609916:2610111 [0] NCCL INFO Channel 17/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2609920:2610113 [4] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-160-103:2609916:2610111 [0] NCCL INFO Channel 18/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2609916:2610111 [0] NCCL INFO Channel 19/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2609916:2610111 [0] NCCL INFO Channel 20/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2609916:2610111 [0] NCCL INFO Channel 21/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2609916:2610111 [0] NCCL INFO Channel 22/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2609916:2610111 [0] NCCL INFO Channel 23/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2609916:2610111 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] 1/-1/-1->0->-1 [7] 1/-1/-1->0->-1 [8] 1/-1/-1->0->-1 [9] 1/-1/-1->0->-1 [10] 1/-1/-1->0->-1 [11] 1/-1/-1->0->-1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] 1/-1/-1->0->-1 [19] 1/-1/-1->0->-1 [20] 1/-1/-1->0->-1 [21] 1/-1/-1->0->-1 [22] 1/-1/-1->0->-1 [23] 1/-1/-1->0->-1
+ip-26-0-160-103:2609916:2610111 [0] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-160-103:2609919:2610116 [3] NCCL INFO Trees [0] 4/-1/-1->3->2 [1] 4/-1/-1->3->2 [2] 4/-1/-1->3->2 [3] 4/-1/-1->3->2 [4] 4/-1/-1->3->2 [5] 4/-1/-1->3->2 [6] 4/-1/-1->3->2 [7] 4/-1/-1->3->2 [8] 4/-1/-1->3->2 [9] 4/-1/-1->3->2 [10] 4/-1/-1->3->2 [11] 4/-1/-1->3->2 [12] 4/-1/-1->3->2 [13] 4/-1/-1->3->2 [14] 4/-1/-1->3->2 [15] 4/-1/-1->3->2 [16] 4/-1/-1->3->2 [17] 4/-1/-1->3->2 [18] 4/-1/-1->3->2 [19] 4/-1/-1->3->2 [20] 4/-1/-1->3->2 [21] 4/-1/-1->3->2 [22] 4/-1/-1->3->2 [23] 4/-1/-1->3->2
+ip-26-0-160-103:2609919:2610116 [3] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-160-103:2609918:2610114 [2] NCCL INFO Channel 00/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609918:2610114 [2] NCCL INFO Channel 01/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609918:2610114 [2] NCCL INFO Channel 02/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609918:2610114 [2] NCCL INFO Channel 03/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609918:2610114 [2] NCCL INFO Channel 04/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609918:2610114 [2] NCCL INFO Channel 05/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609918:2610114 [2] NCCL INFO Channel 06/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609920:2610113 [4] NCCL INFO Channel 00/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609918:2610114 [2] NCCL INFO Channel 07/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609920:2610113 [4] NCCL INFO Channel 01/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609918:2610114 [2] NCCL INFO Channel 08/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609920:2610113 [4] NCCL INFO Channel 02/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609918:2610114 [2] NCCL INFO Channel 09/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609920:2610113 [4] NCCL INFO Channel 03/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609921:2610115 [5] NCCL INFO Channel 00/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609918:2610114 [2] NCCL INFO Channel 10/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609916:2610111 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609917:2610118 [1] NCCL INFO Channel 00/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609920:2610113 [4] NCCL INFO Channel 04/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609921:2610115 [5] NCCL INFO Channel 01/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609918:2610114 [2] NCCL INFO Channel 11/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609916:2610111 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609922:2610112 [6] NCCL INFO Channel 00/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2609917:2610118 [1] NCCL INFO Channel 01/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609920:2610113 [4] NCCL INFO Channel 05/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609921:2610115 [5] NCCL INFO Channel 02/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609918:2610114 [2] NCCL INFO Channel 12/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609916:2610111 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609922:2610112 [6] NCCL INFO Channel 01/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2609917:2610118 [1] NCCL INFO Channel 02/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609920:2610113 [4] NCCL INFO Channel 06/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609921:2610115 [5] NCCL INFO Channel 03/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609918:2610114 [2] NCCL INFO Channel 13/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609916:2610111 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609922:2610112 [6] NCCL INFO Channel 02/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2609917:2610118 [1] NCCL INFO Channel 03/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609920:2610113 [4] NCCL INFO Channel 07/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609921:2610115 [5] NCCL INFO Channel 04/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609918:2610114 [2] NCCL INFO Channel 14/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609922:2610112 [6] NCCL INFO Channel 03/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2609916:2610111 [0] NCCL INFO Channel 04/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609917:2610118 [1] NCCL INFO Channel 04/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609920:2610113 [4] NCCL INFO Channel 08/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609921:2610115 [5] NCCL INFO Channel 05/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609918:2610114 [2] NCCL INFO Channel 15/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609922:2610112 [6] NCCL INFO Channel 04/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2609916:2610111 [0] NCCL INFO Channel 05/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609917:2610118 [1] NCCL INFO Channel 05/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609920:2610113 [4] NCCL INFO Channel 09/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609921:2610115 [5] NCCL INFO Channel 06/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609918:2610114 [2] NCCL INFO Channel 16/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609917:2610118 [1] NCCL INFO Channel 06/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609916:2610111 [0] NCCL INFO Channel 06/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609922:2610112 [6] NCCL INFO Channel 05/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2609920:2610113 [4] NCCL INFO Channel 10/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609921:2610115 [5] NCCL INFO Channel 07/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609918:2610114 [2] NCCL INFO Channel 17/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609917:2610118 [1] NCCL INFO Channel 07/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609916:2610111 [0] NCCL INFO Channel 07/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609922:2610112 [6] NCCL INFO Channel 06/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2609920:2610113 [4] NCCL INFO Channel 11/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609921:2610115 [5] NCCL INFO Channel 08/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609918:2610114 [2] NCCL INFO Channel 18/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609917:2610118 [1] NCCL INFO Channel 08/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609916:2610111 [0] NCCL INFO Channel 08/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609922:2610112 [6] NCCL INFO Channel 07/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2609920:2610113 [4] NCCL INFO Channel 12/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609921:2610115 [5] NCCL INFO Channel 09/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609918:2610114 [2] NCCL INFO Channel 19/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609917:2610118 [1] NCCL INFO Channel 09/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609916:2610111 [0] NCCL INFO Channel 09/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609922:2610112 [6] NCCL INFO Channel 08/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2609920:2610113 [4] NCCL INFO Channel 13/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609921:2610115 [5] NCCL INFO Channel 10/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609918:2610114 [2] NCCL INFO Channel 20/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609917:2610118 [1] NCCL INFO Channel 10/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609916:2610111 [0] NCCL INFO Channel 10/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609922:2610112 [6] NCCL INFO Channel 09/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2609920:2610113 [4] NCCL INFO Channel 14/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609921:2610115 [5] NCCL INFO Channel 11/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609918:2610114 [2] NCCL INFO Channel 21/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609916:2610111 [0] NCCL INFO Channel 11/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609917:2610118 [1] NCCL INFO Channel 11/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609922:2610112 [6] NCCL INFO Channel 10/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2609923:2610117 [7] NCCL INFO Channel 00/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609920:2610113 [4] NCCL INFO Channel 15/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609921:2610115 [5] NCCL INFO Channel 12/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609918:2610114 [2] NCCL INFO Channel 22/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609916:2610111 [0] NCCL INFO Channel 12/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609917:2610118 [1] NCCL INFO Channel 12/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609922:2610112 [6] NCCL INFO Channel 11/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2609923:2610117 [7] NCCL INFO Channel 01/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609920:2610113 [4] NCCL INFO Channel 16/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609921:2610115 [5] NCCL INFO Channel 13/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609918:2610114 [2] NCCL INFO Channel 23/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609916:2610111 [0] NCCL INFO Channel 13/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609917:2610118 [1] NCCL INFO Channel 13/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609922:2610112 [6] NCCL INFO Channel 12/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2609923:2610117 [7] NCCL INFO Channel 02/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609920:2610113 [4] NCCL INFO Channel 17/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609921:2610115 [5] NCCL INFO Channel 14/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609916:2610111 [0] NCCL INFO Channel 14/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609917:2610118 [1] NCCL INFO Channel 14/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609922:2610112 [6] NCCL INFO Channel 13/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2609923:2610117 [7] NCCL INFO Channel 03/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609920:2610113 [4] NCCL INFO Channel 18/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609921:2610115 [5] NCCL INFO Channel 15/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609916:2610111 [0] NCCL INFO Channel 15/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609917:2610118 [1] NCCL INFO Channel 15/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609922:2610112 [6] NCCL INFO Channel 14/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2609923:2610117 [7] NCCL INFO Channel 04/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609919:2610116 [3] NCCL INFO Channel 00/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609920:2610113 [4] NCCL INFO Channel 19/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609921:2610115 [5] NCCL INFO Channel 16/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609917:2610118 [1] NCCL INFO Channel 16/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609916:2610111 [0] NCCL INFO Channel 16/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609922:2610112 [6] NCCL INFO Channel 15/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2609923:2610117 [7] NCCL INFO Channel 05/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609919:2610116 [3] NCCL INFO Channel 01/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609920:2610113 [4] NCCL INFO Channel 20/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609921:2610115 [5] NCCL INFO Channel 17/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609917:2610118 [1] NCCL INFO Channel 17/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609922:2610112 [6] NCCL INFO Channel 16/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2609916:2610111 [0] NCCL INFO Channel 17/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609923:2610117 [7] NCCL INFO Channel 06/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609919:2610116 [3] NCCL INFO Channel 02/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609920:2610113 [4] NCCL INFO Channel 21/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609921:2610115 [5] NCCL INFO Channel 18/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609917:2610118 [1] NCCL INFO Channel 18/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609916:2610111 [0] NCCL INFO Channel 18/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609922:2610112 [6] NCCL INFO Channel 17/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2609923:2610117 [7] NCCL INFO Channel 07/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609919:2610116 [3] NCCL INFO Channel 03/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609920:2610113 [4] NCCL INFO Channel 22/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609917:2610118 [1] NCCL INFO Channel 19/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609921:2610115 [5] NCCL INFO Channel 19/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609922:2610112 [6] NCCL INFO Channel 18/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2609916:2610111 [0] NCCL INFO Channel 19/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609923:2610117 [7] NCCL INFO Channel 08/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609919:2610116 [3] NCCL INFO Channel 04/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609917:2610118 [1] NCCL INFO Channel 20/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609920:2610113 [4] NCCL INFO Channel 23/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609921:2610115 [5] NCCL INFO Channel 20/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609916:2610111 [0] NCCL INFO Channel 20/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609922:2610112 [6] NCCL INFO Channel 19/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2609923:2610117 [7] NCCL INFO Channel 09/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609919:2610116 [3] NCCL INFO Channel 05/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609917:2610118 [1] NCCL INFO Channel 21/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609921:2610115 [5] NCCL INFO Channel 21/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609916:2610111 [0] NCCL INFO Channel 21/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609922:2610112 [6] NCCL INFO Channel 20/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2609923:2610117 [7] NCCL INFO Channel 10/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609917:2610118 [1] NCCL INFO Channel 22/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609919:2610116 [3] NCCL INFO Channel 06/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609921:2610115 [5] NCCL INFO Channel 22/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609916:2610111 [0] NCCL INFO Channel 22/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609922:2610112 [6] NCCL INFO Channel 21/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2609923:2610117 [7] NCCL INFO Channel 11/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609917:2610118 [1] NCCL INFO Channel 23/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609919:2610116 [3] NCCL INFO Channel 07/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609921:2610115 [5] NCCL INFO Channel 23/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609916:2610111 [0] NCCL INFO Channel 23/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609922:2610112 [6] NCCL INFO Channel 22/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2609923:2610117 [7] NCCL INFO Channel 12/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609919:2610116 [3] NCCL INFO Channel 08/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609923:2610117 [7] NCCL INFO Channel 13/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609922:2610112 [6] NCCL INFO Channel 23/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2609919:2610116 [3] NCCL INFO Channel 09/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609923:2610117 [7] NCCL INFO Channel 14/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609919:2610116 [3] NCCL INFO Channel 10/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609923:2610117 [7] NCCL INFO Channel 15/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609919:2610116 [3] NCCL INFO Channel 11/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609923:2610117 [7] NCCL INFO Channel 16/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609919:2610116 [3] NCCL INFO Channel 12/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609923:2610117 [7] NCCL INFO Channel 17/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609919:2610116 [3] NCCL INFO Channel 13/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609923:2610117 [7] NCCL INFO Channel 18/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609919:2610116 [3] NCCL INFO Channel 14/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609923:2610117 [7] NCCL INFO Channel 19/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609919:2610116 [3] NCCL INFO Channel 15/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609923:2610117 [7] NCCL INFO Channel 20/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609919:2610116 [3] NCCL INFO Channel 16/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609919:2610116 [3] NCCL INFO Channel 17/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609923:2610117 [7] NCCL INFO Channel 21/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609919:2610116 [3] NCCL INFO Channel 18/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609923:2610117 [7] NCCL INFO Channel 22/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609919:2610116 [3] NCCL INFO Channel 19/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609923:2610117 [7] NCCL INFO Channel 23/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609919:2610116 [3] NCCL INFO Channel 20/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609919:2610116 [3] NCCL INFO Channel 21/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609919:2610116 [3] NCCL INFO Channel 22/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609919:2610116 [3] NCCL INFO Channel 23/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609917:2610118 [1] NCCL INFO Connected all rings
+ip-26-0-160-103:2609921:2610115 [5] NCCL INFO Connected all rings
+ip-26-0-160-103:2609922:2610112 [6] NCCL INFO Connected all rings
+ip-26-0-160-103:2609920:2610113 [4] NCCL INFO Connected all rings
+ip-26-0-160-103:2609916:2610111 [0] NCCL INFO Connected all rings
+ip-26-0-160-103:2609918:2610114 [2] NCCL INFO Connected all rings
+ip-26-0-160-103:2609923:2610117 [7] NCCL INFO Connected all rings
+ip-26-0-160-103:2609923:2610117 [7] NCCL INFO Channel 00/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609923:2610117 [7] NCCL INFO Channel 01/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609919:2610116 [3] NCCL INFO Connected all rings
+ip-26-0-160-103:2609923:2610117 [7] NCCL INFO Channel 02/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609917:2610118 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609923:2610117 [7] NCCL INFO Channel 03/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609917:2610118 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609917:2610118 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609917:2610118 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609923:2610117 [7] NCCL INFO Channel 04/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609917:2610118 [1] NCCL INFO Channel 04/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609923:2610117 [7] NCCL INFO Channel 05/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609923:2610117 [7] NCCL INFO Channel 06/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609921:2610115 [5] NCCL INFO Channel 00/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609917:2610118 [1] NCCL INFO Channel 05/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609917:2610118 [1] NCCL INFO Channel 06/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609921:2610115 [5] NCCL INFO Channel 01/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609921:2610115 [5] NCCL INFO Channel 02/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609923:2610117 [7] NCCL INFO Channel 07/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609921:2610115 [5] NCCL INFO Channel 03/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609917:2610118 [1] NCCL INFO Channel 07/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609923:2610117 [7] NCCL INFO Channel 08/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609921:2610115 [5] NCCL INFO Channel 04/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609917:2610118 [1] NCCL INFO Channel 08/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609923:2610117 [7] NCCL INFO Channel 09/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609921:2610115 [5] NCCL INFO Channel 05/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609917:2610118 [1] NCCL INFO Channel 09/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609923:2610117 [7] NCCL INFO Channel 10/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609921:2610115 [5] NCCL INFO Channel 06/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609917:2610118 [1] NCCL INFO Channel 10/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609923:2610117 [7] NCCL INFO Channel 11/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609921:2610115 [5] NCCL INFO Channel 07/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609917:2610118 [1] NCCL INFO Channel 11/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609923:2610117 [7] NCCL INFO Channel 12/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609921:2610115 [5] NCCL INFO Channel 08/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609917:2610118 [1] NCCL INFO Channel 12/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609923:2610117 [7] NCCL INFO Channel 13/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609921:2610115 [5] NCCL INFO Channel 09/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609917:2610118 [1] NCCL INFO Channel 13/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609918:2610114 [2] NCCL INFO Channel 00/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609923:2610117 [7] NCCL INFO Channel 14/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609922:2610112 [6] NCCL INFO Channel 00/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609921:2610115 [5] NCCL INFO Channel 10/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609917:2610118 [1] NCCL INFO Channel 14/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609918:2610114 [2] NCCL INFO Channel 01/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609923:2610117 [7] NCCL INFO Channel 15/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609922:2610112 [6] NCCL INFO Channel 01/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609921:2610115 [5] NCCL INFO Channel 11/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609917:2610118 [1] NCCL INFO Channel 15/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609918:2610114 [2] NCCL INFO Channel 02/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609923:2610117 [7] NCCL INFO Channel 16/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609922:2610112 [6] NCCL INFO Channel 02/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609921:2610115 [5] NCCL INFO Channel 12/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609917:2610118 [1] NCCL INFO Channel 16/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609918:2610114 [2] NCCL INFO Channel 03/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609923:2610117 [7] NCCL INFO Channel 17/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609922:2610112 [6] NCCL INFO Channel 03/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609917:2610118 [1] NCCL INFO Channel 17/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609918:2610114 [2] NCCL INFO Channel 04/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609923:2610117 [7] NCCL INFO Channel 18/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609918:2610114 [2] NCCL INFO Channel 05/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609917:2610118 [1] NCCL INFO Channel 18/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609923:2610117 [7] NCCL INFO Channel 19/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609917:2610118 [1] NCCL INFO Channel 19/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609918:2610114 [2] NCCL INFO Channel 06/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609919:2610116 [3] NCCL INFO Channel 00/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609923:2610117 [7] NCCL INFO Channel 20/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609917:2610118 [1] NCCL INFO Channel 20/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609918:2610114 [2] NCCL INFO Channel 07/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609919:2610116 [3] NCCL INFO Channel 01/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609923:2610117 [7] NCCL INFO Channel 21/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609918:2610114 [2] NCCL INFO Channel 08/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609917:2610118 [1] NCCL INFO Channel 21/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609919:2610116 [3] NCCL INFO Channel 02/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609923:2610117 [7] NCCL INFO Channel 22/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609918:2610114 [2] NCCL INFO Channel 09/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609917:2610118 [1] NCCL INFO Channel 22/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609919:2610116 [3] NCCL INFO Channel 03/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609923:2610117 [7] NCCL INFO Channel 23/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609918:2610114 [2] NCCL INFO Channel 10/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609917:2610118 [1] NCCL INFO Channel 23/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609919:2610116 [3] NCCL INFO Channel 04/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609918:2610114 [2] NCCL INFO Channel 11/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609921:2610115 [5] NCCL INFO Channel 13/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609919:2610116 [3] NCCL INFO Channel 05/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609918:2610114 [2] NCCL INFO Channel 12/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609921:2610115 [5] NCCL INFO Channel 14/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609919:2610116 [3] NCCL INFO Channel 06/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609922:2610112 [6] NCCL INFO Channel 04/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609918:2610114 [2] NCCL INFO Channel 13/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609921:2610115 [5] NCCL INFO Channel 15/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609919:2610116 [3] NCCL INFO Channel 07/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609922:2610112 [6] NCCL INFO Channel 05/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609918:2610114 [2] NCCL INFO Channel 14/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609921:2610115 [5] NCCL INFO Channel 16/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609919:2610116 [3] NCCL INFO Channel 08/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609918:2610114 [2] NCCL INFO Channel 15/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609922:2610112 [6] NCCL INFO Channel 06/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609921:2610115 [5] NCCL INFO Channel 17/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609919:2610116 [3] NCCL INFO Channel 09/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609918:2610114 [2] NCCL INFO Channel 16/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609920:2610113 [4] NCCL INFO Channel 00/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609922:2610112 [6] NCCL INFO Channel 07/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609921:2610115 [5] NCCL INFO Channel 18/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609920:2610113 [4] NCCL INFO Channel 01/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609922:2610112 [6] NCCL INFO Channel 08/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609921:2610115 [5] NCCL INFO Channel 19/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609919:2610116 [3] NCCL INFO Channel 10/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609920:2610113 [4] NCCL INFO Channel 02/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609918:2610114 [2] NCCL INFO Channel 17/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609922:2610112 [6] NCCL INFO Channel 09/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609921:2610115 [5] NCCL INFO Channel 20/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609919:2610116 [3] NCCL INFO Channel 11/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609920:2610113 [4] NCCL INFO Channel 03/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609918:2610114 [2] NCCL INFO Channel 18/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609922:2610112 [6] NCCL INFO Channel 10/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609921:2610115 [5] NCCL INFO Channel 21/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609919:2610116 [3] NCCL INFO Channel 12/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609920:2610113 [4] NCCL INFO Channel 04/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609918:2610114 [2] NCCL INFO Channel 19/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609922:2610112 [6] NCCL INFO Channel 11/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609921:2610115 [5] NCCL INFO Channel 22/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609919:2610116 [3] NCCL INFO Channel 13/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609920:2610113 [4] NCCL INFO Channel 05/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609919:2610116 [3] NCCL INFO Channel 14/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609919:2610116 [3] NCCL INFO Channel 15/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609920:2610113 [4] NCCL INFO Channel 06/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609916:2610111 [0] NCCL INFO Connected all trees
+ip-26-0-160-103:2609919:2610116 [3] NCCL INFO Channel 16/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609918:2610114 [2] NCCL INFO Channel 20/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609920:2610113 [4] NCCL INFO Channel 07/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609916:2610111 [0] NCCL INFO NVLS comm 0x9b9b3e0 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-103:2609922:2610112 [6] NCCL INFO Channel 12/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609921:2610115 [5] NCCL INFO Channel 23/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609919:2610116 [3] NCCL INFO Channel 17/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609918:2610114 [2] NCCL INFO Channel 21/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609922:2610112 [6] NCCL INFO Channel 13/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609920:2610113 [4] NCCL INFO Channel 08/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609919:2610116 [3] NCCL INFO Channel 18/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609918:2610114 [2] NCCL INFO Channel 22/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609922:2610112 [6] NCCL INFO Channel 14/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609920:2610113 [4] NCCL INFO Channel 09/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609919:2610116 [3] NCCL INFO Channel 19/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609918:2610114 [2] NCCL INFO Channel 23/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609922:2610112 [6] NCCL INFO Channel 15/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609919:2610116 [3] NCCL INFO Channel 20/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609920:2610113 [4] NCCL INFO Channel 10/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609922:2610112 [6] NCCL INFO Channel 16/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609919:2610116 [3] NCCL INFO Channel 21/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609920:2610113 [4] NCCL INFO Channel 11/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609922:2610112 [6] NCCL INFO Channel 17/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609919:2610116 [3] NCCL INFO Channel 22/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609920:2610113 [4] NCCL INFO Channel 12/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609922:2610112 [6] NCCL INFO Channel 18/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609922:2610112 [6] NCCL INFO Channel 19/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609920:2610113 [4] NCCL INFO Channel 13/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609919:2610116 [3] NCCL INFO Channel 23/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609922:2610112 [6] NCCL INFO Channel 20/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609920:2610113 [4] NCCL INFO Channel 14/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609922:2610112 [6] NCCL INFO Channel 21/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609920:2610113 [4] NCCL INFO Channel 15/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609922:2610112 [6] NCCL INFO Channel 22/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609920:2610113 [4] NCCL INFO Channel 16/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609922:2610112 [6] NCCL INFO Channel 23/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609920:2610113 [4] NCCL INFO Channel 17/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609920:2610113 [4] NCCL INFO Channel 18/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609920:2610113 [4] NCCL INFO Channel 19/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609920:2610113 [4] NCCL INFO Channel 20/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609920:2610113 [4] NCCL INFO Channel 21/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609920:2610113 [4] NCCL INFO Channel 22/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609920:2610113 [4] NCCL INFO Channel 23/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609923:2610117 [7] NCCL INFO Connected all trees
+ip-26-0-160-103:2609923:2610117 [7] NCCL INFO NVLS comm 0xa451a50 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-103:2609917:2610118 [1] NCCL INFO Connected all trees
+ip-26-0-160-103:2609917:2610118 [1] NCCL INFO NVLS comm 0x918d5d0 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-103:2609918:2610114 [2] NCCL INFO Connected all trees
+ip-26-0-160-103:2609918:2610114 [2] NCCL INFO NVLS comm 0xa360d80 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-103:2609922:2610112 [6] NCCL INFO Connected all trees
+ip-26-0-160-103:2609922:2610112 [6] NCCL INFO NVLS comm 0x98863c0 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-103:2609920:2610113 [4] NCCL INFO Connected all trees
+ip-26-0-160-103:2609920:2610113 [4] NCCL INFO NVLS comm 0x9335e20 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-103:2609919:2610116 [3] NCCL INFO Connected all trees
+ip-26-0-160-103:2609919:2610116 [3] NCCL INFO NVLS comm 0x8703130 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-103:2609921:2610115 [5] NCCL INFO Connected all trees
+ip-26-0-160-103:2609921:2610115 [5] NCCL INFO NVLS comm 0x8f9ffa0 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-103:2609921:2610115 [5] NCCL INFO Connected NVLS tree
+ip-26-0-160-103:2609921:2610115 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-103:2609921:2610115 [5] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-160-103:2609921:2610115 [5] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-160-103:2609917:2610118 [1] NCCL INFO Connected NVLS tree
+ip-26-0-160-103:2609917:2610118 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-103:2609917:2610118 [1] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-160-103:2609917:2610118 [1] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-160-103:2609919:2610116 [3] NCCL INFO Connected NVLS tree
+ip-26-0-160-103:2609919:2610116 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-103:2609922:2610112 [6] NCCL INFO Connected NVLS tree
+ip-26-0-160-103:2609919:2610116 [3] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-160-103:2609919:2610116 [3] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-160-103:2609922:2610112 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-103:2609922:2610112 [6] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-160-103:2609922:2610112 [6] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-160-103:2609918:2610114 [2] NCCL INFO Connected NVLS tree
+ip-26-0-160-103:2609918:2610114 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-103:2609920:2610113 [4] NCCL INFO Connected NVLS tree
+ip-26-0-160-103:2609918:2610114 [2] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-160-103:2609918:2610114 [2] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-160-103:2609920:2610113 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-103:2609920:2610113 [4] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-160-103:2609920:2610113 [4] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-160-103:2609923:2610117 [7] NCCL INFO Connected NVLS tree
+ip-26-0-160-103:2609923:2610117 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-103:2609923:2610117 [7] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-160-103:2609923:2610117 [7] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-160-103:2609916:2610111 [0] NCCL INFO Connected NVLS tree
+ip-26-0-160-103:2609916:2610111 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-103:2609916:2610111 [0] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-160-103:2609916:2610111 [0] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-160-103:2609921:2610115 [5] NCCL INFO comm 0x8f9ffa0 rank 5 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0xfa772de19d2f0956 - Init COMPLETE
+ip-26-0-160-103:2609923:2610117 [7] NCCL INFO comm 0xa451a50 rank 7 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0xfa772de19d2f0956 - Init COMPLETE
+ip-26-0-160-103:2609918:2610114 [2] NCCL INFO comm 0xa360d80 rank 2 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0xfa772de19d2f0956 - Init COMPLETE
+ip-26-0-160-103:2609919:2610116 [3] NCCL INFO comm 0x8703130 rank 3 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0xfa772de19d2f0956 - Init COMPLETE
+ip-26-0-160-103:2609917:2610118 [1] NCCL INFO comm 0x918d5d0 rank 1 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0xfa772de19d2f0956 - Init COMPLETE
+ip-26-0-160-103:2609922:2610112 [6] NCCL INFO comm 0x98863c0 rank 6 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0xfa772de19d2f0956 - Init COMPLETE
+ip-26-0-160-103:2609916:2610111 [0] NCCL INFO comm 0x9b9b3e0 rank 0 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0xfa772de19d2f0956 - Init COMPLETE
+ip-26-0-160-103:2609920:2610113 [4] NCCL INFO comm 0x9335e20 rank 4 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0xfa772de19d2f0956 - Init COMPLETE
+ip-26-0-160-103:2609916:2610144 [0] NCCL INFO Using network AWS Libfabric
+ip-26-0-160-103:2609920:2610149 [4] NCCL INFO Using network AWS Libfabric
+ip-26-0-160-103:2609922:2610150 [6] NCCL INFO Using network AWS Libfabric
+ip-26-0-160-103:2609917:2610147 [1] NCCL INFO Using network AWS Libfabric
+ip-26-0-160-103:2609919:2610146 [3] NCCL INFO Using network AWS Libfabric
+ip-26-0-160-103:2609921:2610148 [5] NCCL INFO Using network AWS Libfabric
+ip-26-0-160-103:2609918:2610145 [2] NCCL INFO Using network AWS Libfabric
+ip-26-0-160-103:2609923:2610151 [7] NCCL INFO Using network AWS Libfabric
+ip-26-0-160-103:2609918:2610145 [2] NCCL INFO comm 0xa3753a0 rank 2 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0x5fae3ab7c1a3f336 - Init START
+ip-26-0-160-103:2609916:2610144 [0] NCCL INFO comm 0x9baf880 rank 0 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0x5fae3ab7c1a3f336 - Init START
+ip-26-0-160-103:2609917:2610147 [1] NCCL INFO comm 0x91a1d50 rank 1 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0x5fae3ab7c1a3f336 - Init START
+ip-26-0-160-103:2609923:2610151 [7] NCCL INFO comm 0xa465e80 rank 7 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0x5fae3ab7c1a3f336 - Init START
+ip-26-0-160-103:2609921:2610148 [5] NCCL INFO comm 0x8fb46f0 rank 5 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0x5fae3ab7c1a3f336 - Init START
+ip-26-0-160-103:2609922:2610150 [6] NCCL INFO comm 0x989a860 rank 6 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0x5fae3ab7c1a3f336 - Init START
+ip-26-0-160-103:2609920:2610149 [4] NCCL INFO comm 0x934a5a0 rank 4 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0x5fae3ab7c1a3f336 - Init START
+ip-26-0-160-103:2609919:2610146 [3] NCCL INFO comm 0x87178b0 rank 3 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0x5fae3ab7c1a3f336 - Init START
+ip-26-0-160-103:2609917:2610147 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /opt/aws-ofi-nccl/share/aws-ofi-nccl/xml/p5.48xl-topo.xml
+ip-26-0-160-103:2609918:2610145 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /opt/aws-ofi-nccl/share/aws-ofi-nccl/xml/p5.48xl-topo.xml
+ip-26-0-160-103:2609923:2610151 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /opt/aws-ofi-nccl/share/aws-ofi-nccl/xml/p5.48xl-topo.xml
+ip-26-0-160-103:2609919:2610146 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /opt/aws-ofi-nccl/share/aws-ofi-nccl/xml/p5.48xl-topo.xml
+ip-26-0-160-103:2609922:2610150 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /opt/aws-ofi-nccl/share/aws-ofi-nccl/xml/p5.48xl-topo.xml
+ip-26-0-160-103:2609920:2610149 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /opt/aws-ofi-nccl/share/aws-ofi-nccl/xml/p5.48xl-topo.xml
+ip-26-0-160-103:2609921:2610148 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /opt/aws-ofi-nccl/share/aws-ofi-nccl/xml/p5.48xl-topo.xml
+ip-26-0-160-103:2609916:2610144 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /opt/aws-ofi-nccl/share/aws-ofi-nccl/xml/p5.48xl-topo.xml
+ip-26-0-160-103:2609922:2610150 [6] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609922:2610150 [6] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609922:2610150 [6] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609922:2610150 [6] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609922:2610150 [6] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609922:2610150 [6] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609922:2610150 [6] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609922:2610150 [6] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609922:2610150 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-160-103:2609922:2610150 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-160-103:2609923:2610151 [7] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609923:2610151 [7] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609923:2610151 [7] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609923:2610151 [7] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609923:2610151 [7] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609923:2610151 [7] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609923:2610151 [7] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609923:2610151 [7] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609918:2610145 [2] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609918:2610145 [2] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609918:2610145 [2] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609918:2610145 [2] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609918:2610145 [2] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609918:2610145 [2] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609918:2610145 [2] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609918:2610145 [2] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609923:2610151 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-160-103:2609918:2610145 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-160-103:2609918:2610145 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-160-103:2609923:2610151 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-160-103:2609920:2610149 [4] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609920:2610149 [4] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609920:2610149 [4] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609920:2610149 [4] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609920:2610149 [4] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609920:2610149 [4] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609920:2610149 [4] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609920:2610149 [4] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609920:2610149 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-160-103:2609920:2610149 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-160-103:2609916:2610144 [0] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609916:2610144 [0] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609916:2610144 [0] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609916:2610144 [0] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609916:2610144 [0] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609916:2610144 [0] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609916:2610144 [0] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609916:2610144 [0] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609916:2610144 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-160-103:2609916:2610144 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-160-103:2609919:2610146 [3] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609919:2610146 [3] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609919:2610146 [3] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609919:2610146 [3] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609919:2610146 [3] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609919:2610146 [3] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609919:2610146 [3] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609919:2610146 [3] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609921:2610148 [5] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609921:2610148 [5] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609921:2610148 [5] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609921:2610148 [5] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609921:2610148 [5] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609921:2610148 [5] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609921:2610148 [5] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609921:2610148 [5] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609919:2610146 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-160-103:2609921:2610148 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-160-103:2609921:2610148 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-160-103:2609919:2610146 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-160-103:2609917:2610147 [1] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609917:2610147 [1] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609917:2610147 [1] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609917:2610147 [1] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609917:2610147 [1] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609917:2610147 [1] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609917:2610147 [1] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609917:2610147 [1] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609917:2610147 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-160-103:2609917:2610147 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-160-103:2609917:2610147 [1] NCCL INFO Trees [0] 2/-1/-1->1->0 [1] 2/-1/-1->1->0 [2] 2/-1/-1->1->0 [3] 2/-1/-1->1->0 [4] 2/-1/-1->1->0 [5] 2/-1/-1->1->0 [6] 2/-1/-1->1->0 [7] 2/-1/-1->1->0 [8] 2/-1/-1->1->0 [9] 2/-1/-1->1->0 [10] 2/-1/-1->1->0 [11] 2/-1/-1->1->0 [12] 2/-1/-1->1->0 [13] 2/-1/-1->1->0 [14] 2/-1/-1->1->0 [15] 2/-1/-1->1->0 [16] 2/-1/-1->1->0 [17] 2/-1/-1->1->0 [18] 2/-1/-1->1->0 [19] 2/-1/-1->1->0 [20] 2/-1/-1->1->0 [21] 2/-1/-1->1->0 [22] 2/-1/-1->1->0 [23] 2/-1/-1->1->0
+ip-26-0-160-103:2609917:2610147 [1] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-160-103:2609918:2610145 [2] NCCL INFO Trees [0] 3/-1/-1->2->1 [1] 3/-1/-1->2->1 [2] 3/-1/-1->2->1 [3] 3/-1/-1->2->1 [4] 3/-1/-1->2->1 [5] 3/-1/-1->2->1 [6] 3/-1/-1->2->1 [7] 3/-1/-1->2->1 [8] 3/-1/-1->2->1 [9] 3/-1/-1->2->1 [10] 3/-1/-1->2->1 [11] 3/-1/-1->2->1 [12] 3/-1/-1->2->1 [13] 3/-1/-1->2->1 [14] 3/-1/-1->2->1 [15] 3/-1/-1->2->1 [16] 3/-1/-1->2->1 [17] 3/-1/-1->2->1 [18] 3/-1/-1->2->1 [19] 3/-1/-1->2->1 [20] 3/-1/-1->2->1 [21] 3/-1/-1->2->1 [22] 3/-1/-1->2->1 [23] 3/-1/-1->2->1
+ip-26-0-160-103:2609918:2610145 [2] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-160-103:2609919:2610146 [3] NCCL INFO Trees [0] 4/-1/-1->3->2 [1] 4/-1/-1->3->2 [2] 4/-1/-1->3->2 [3] 4/-1/-1->3->2 [4] 4/-1/-1->3->2 [5] 4/-1/-1->3->2 [6] 4/-1/-1->3->2 [7] 4/-1/-1->3->2 [8] 4/-1/-1->3->2 [9] 4/-1/-1->3->2 [10] 4/-1/-1->3->2 [11] 4/-1/-1->3->2 [12] 4/-1/-1->3->2 [13] 4/-1/-1->3->2 [14] 4/-1/-1->3->2 [15] 4/-1/-1->3->2 [16] 4/-1/-1->3->2 [17] 4/-1/-1->3->2 [18] 4/-1/-1->3->2 [19] 4/-1/-1->3->2 [20] 4/-1/-1->3->2 [21] 4/-1/-1->3->2 [22] 4/-1/-1->3->2 [23] 4/-1/-1->3->2
+ip-26-0-160-103:2609919:2610146 [3] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-160-103:2609921:2610148 [5] NCCL INFO Trees [0] 6/-1/-1->5->4 [1] 6/-1/-1->5->4 [2] 6/-1/-1->5->4 [3] 6/-1/-1->5->4 [4] 6/-1/-1->5->4 [5] 6/-1/-1->5->4 [6] 6/-1/-1->5->4 [7] 6/-1/-1->5->4 [8] 6/-1/-1->5->4 [9] 6/-1/-1->5->4 [10] 6/-1/-1->5->4 [11] 6/-1/-1->5->4 [12] 6/-1/-1->5->4 [13] 6/-1/-1->5->4 [14] 6/-1/-1->5->4 [15] 6/-1/-1->5->4 [16] 6/-1/-1->5->4 [17] 6/-1/-1->5->4 [18] 6/-1/-1->5->4 [19] 6/-1/-1->5->4 [20] 6/-1/-1->5->4 [21] 6/-1/-1->5->4 [22] 6/-1/-1->5->4 [23] 6/-1/-1->5->4
+ip-26-0-160-103:2609920:2610149 [4] NCCL INFO Trees [0] 5/-1/-1->4->3 [1] 5/-1/-1->4->3 [2] 5/-1/-1->4->3 [3] 5/-1/-1->4->3 [4] 5/-1/-1->4->3 [5] 5/-1/-1->4->3 [6] 5/-1/-1->4->3 [7] 5/-1/-1->4->3 [8] 5/-1/-1->4->3 [9] 5/-1/-1->4->3 [10] 5/-1/-1->4->3 [11] 5/-1/-1->4->3 [12] 5/-1/-1->4->3 [13] 5/-1/-1->4->3 [14] 5/-1/-1->4->3 [15] 5/-1/-1->4->3 [16] 5/-1/-1->4->3 [17] 5/-1/-1->4->3 [18] 5/-1/-1->4->3 [19] 5/-1/-1->4->3 [20] 5/-1/-1->4->3 [21] 5/-1/-1->4->3 [22] 5/-1/-1->4->3 [23] 5/-1/-1->4->3
+ip-26-0-160-103:2609921:2610148 [5] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-160-103:2609920:2610149 [4] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-160-103:2609923:2610151 [7] NCCL INFO Trees [0] -1/-1/-1->7->6 [1] -1/-1/-1->7->6 [2] -1/-1/-1->7->6 [3] -1/-1/-1->7->6 [4] -1/-1/-1->7->6 [5] -1/-1/-1->7->6 [6] -1/-1/-1->7->6 [7] -1/-1/-1->7->6 [8] -1/-1/-1->7->6 [9] -1/-1/-1->7->6 [10] -1/-1/-1->7->6 [11] -1/-1/-1->7->6 [12] -1/-1/-1->7->6 [13] -1/-1/-1->7->6 [14] -1/-1/-1->7->6 [15] -1/-1/-1->7->6 [16] -1/-1/-1->7->6 [17] -1/-1/-1->7->6 [18] -1/-1/-1->7->6 [19] -1/-1/-1->7->6 [20] -1/-1/-1->7->6 [21] -1/-1/-1->7->6 [22] -1/-1/-1->7->6 [23] -1/-1/-1->7->6
+ip-26-0-160-103:2609923:2610151 [7] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-160-103:2609922:2610150 [6] NCCL INFO Trees [0] 7/-1/-1->6->5 [1] 7/-1/-1->6->5 [2] 7/-1/-1->6->5 [3] 7/-1/-1->6->5 [4] 7/-1/-1->6->5 [5] 7/-1/-1->6->5 [6] 7/-1/-1->6->5 [7] 7/-1/-1->6->5 [8] 7/-1/-1->6->5 [9] 7/-1/-1->6->5 [10] 7/-1/-1->6->5 [11] 7/-1/-1->6->5 [12] 7/-1/-1->6->5 [13] 7/-1/-1->6->5 [14] 7/-1/-1->6->5 [15] 7/-1/-1->6->5 [16] 7/-1/-1->6->5 [17] 7/-1/-1->6->5 [18] 7/-1/-1->6->5 [19] 7/-1/-1->6->5 [20] 7/-1/-1->6->5 [21] 7/-1/-1->6->5 [22] 7/-1/-1->6->5 [23] 7/-1/-1->6->5
+ip-26-0-160-103:2609922:2610150 [6] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-160-103:2609916:2610144 [0] NCCL INFO Channel 00/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2609916:2610144 [0] NCCL INFO Channel 01/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2609916:2610144 [0] NCCL INFO Channel 02/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2609916:2610144 [0] NCCL INFO Channel 03/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2609916:2610144 [0] NCCL INFO Channel 04/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2609916:2610144 [0] NCCL INFO Channel 05/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2609916:2610144 [0] NCCL INFO Channel 06/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2609916:2610144 [0] NCCL INFO Channel 07/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2609916:2610144 [0] NCCL INFO Channel 08/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2609916:2610144 [0] NCCL INFO Channel 09/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2609916:2610144 [0] NCCL INFO Channel 10/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2609916:2610144 [0] NCCL INFO Channel 11/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2609916:2610144 [0] NCCL INFO Channel 12/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2609916:2610144 [0] NCCL INFO Channel 13/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2609916:2610144 [0] NCCL INFO Channel 14/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2609916:2610144 [0] NCCL INFO Channel 15/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2609916:2610144 [0] NCCL INFO Channel 16/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2609916:2610144 [0] NCCL INFO Channel 17/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2609916:2610144 [0] NCCL INFO Channel 18/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2609916:2610144 [0] NCCL INFO Channel 19/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2609916:2610144 [0] NCCL INFO Channel 20/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2609916:2610144 [0] NCCL INFO Channel 21/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2609916:2610144 [0] NCCL INFO Channel 22/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2609916:2610144 [0] NCCL INFO Channel 23/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-103:2609916:2610144 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] 1/-1/-1->0->-1 [7] 1/-1/-1->0->-1 [8] 1/-1/-1->0->-1 [9] 1/-1/-1->0->-1 [10] 1/-1/-1->0->-1 [11] 1/-1/-1->0->-1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] 1/-1/-1->0->-1 [19] 1/-1/-1->0->-1 [20] 1/-1/-1->0->-1 [21] 1/-1/-1->0->-1 [22] 1/-1/-1->0->-1 [23] 1/-1/-1->0->-1
+ip-26-0-160-103:2609916:2610144 [0] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-160-103:2609922:2610150 [6] NCCL INFO Channel 00/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2609917:2610147 [1] NCCL INFO Channel 00/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609916:2610144 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609921:2610148 [5] NCCL INFO Channel 00/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609922:2610150 [6] NCCL INFO Channel 01/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2609916:2610144 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609921:2610148 [5] NCCL INFO Channel 01/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609922:2610150 [6] NCCL INFO Channel 02/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2609916:2610144 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609921:2610148 [5] NCCL INFO Channel 02/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609922:2610150 [6] NCCL INFO Channel 03/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2609916:2610144 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609921:2610148 [5] NCCL INFO Channel 03/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609922:2610150 [6] NCCL INFO Channel 04/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2609923:2610151 [7] NCCL INFO Channel 00/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609916:2610144 [0] NCCL INFO Channel 04/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609920:2610149 [4] NCCL INFO Channel 00/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609917:2610147 [1] NCCL INFO Channel 01/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609923:2610151 [7] NCCL INFO Channel 01/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609916:2610144 [0] NCCL INFO Channel 05/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609920:2610149 [4] NCCL INFO Channel 01/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609917:2610147 [1] NCCL INFO Channel 02/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609923:2610151 [7] NCCL INFO Channel 02/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609916:2610144 [0] NCCL INFO Channel 06/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609920:2610149 [4] NCCL INFO Channel 02/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609917:2610147 [1] NCCL INFO Channel 03/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609921:2610148 [5] NCCL INFO Channel 04/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609918:2610145 [2] NCCL INFO Channel 00/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609923:2610151 [7] NCCL INFO Channel 03/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609920:2610149 [4] NCCL INFO Channel 03/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609917:2610147 [1] NCCL INFO Channel 04/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609916:2610144 [0] NCCL INFO Channel 07/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609921:2610148 [5] NCCL INFO Channel 05/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609918:2610145 [2] NCCL INFO Channel 01/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609923:2610151 [7] NCCL INFO Channel 04/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609920:2610149 [4] NCCL INFO Channel 04/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609917:2610147 [1] NCCL INFO Channel 05/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609916:2610144 [0] NCCL INFO Channel 08/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609921:2610148 [5] NCCL INFO Channel 06/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609918:2610145 [2] NCCL INFO Channel 02/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609922:2610150 [6] NCCL INFO Channel 05/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2609923:2610151 [7] NCCL INFO Channel 05/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609920:2610149 [4] NCCL INFO Channel 05/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609917:2610147 [1] NCCL INFO Channel 06/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609916:2610144 [0] NCCL INFO Channel 09/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609921:2610148 [5] NCCL INFO Channel 07/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609918:2610145 [2] NCCL INFO Channel 03/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609922:2610150 [6] NCCL INFO Channel 06/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2609923:2610151 [7] NCCL INFO Channel 06/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609920:2610149 [4] NCCL INFO Channel 06/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609917:2610147 [1] NCCL INFO Channel 07/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609916:2610144 [0] NCCL INFO Channel 10/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609921:2610148 [5] NCCL INFO Channel 08/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609918:2610145 [2] NCCL INFO Channel 04/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609922:2610150 [6] NCCL INFO Channel 07/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2609919:2610146 [3] NCCL INFO Channel 00/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609923:2610151 [7] NCCL INFO Channel 07/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609920:2610149 [4] NCCL INFO Channel 07/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609917:2610147 [1] NCCL INFO Channel 08/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609916:2610144 [0] NCCL INFO Channel 11/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609921:2610148 [5] NCCL INFO Channel 09/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609918:2610145 [2] NCCL INFO Channel 05/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609922:2610150 [6] NCCL INFO Channel 08/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2609923:2610151 [7] NCCL INFO Channel 08/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609919:2610146 [3] NCCL INFO Channel 01/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609917:2610147 [1] NCCL INFO Channel 09/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609920:2610149 [4] NCCL INFO Channel 08/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609916:2610144 [0] NCCL INFO Channel 12/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609918:2610145 [2] NCCL INFO Channel 06/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609921:2610148 [5] NCCL INFO Channel 10/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609922:2610150 [6] NCCL INFO Channel 09/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2609923:2610151 [7] NCCL INFO Channel 09/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609919:2610146 [3] NCCL INFO Channel 02/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609917:2610147 [1] NCCL INFO Channel 10/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609920:2610149 [4] NCCL INFO Channel 09/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609916:2610144 [0] NCCL INFO Channel 13/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609921:2610148 [5] NCCL INFO Channel 11/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609918:2610145 [2] NCCL INFO Channel 07/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609922:2610150 [6] NCCL INFO Channel 10/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2609923:2610151 [7] NCCL INFO Channel 10/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609919:2610146 [3] NCCL INFO Channel 03/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609917:2610147 [1] NCCL INFO Channel 11/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609920:2610149 [4] NCCL INFO Channel 10/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609916:2610144 [0] NCCL INFO Channel 14/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609921:2610148 [5] NCCL INFO Channel 12/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609918:2610145 [2] NCCL INFO Channel 08/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609923:2610151 [7] NCCL INFO Channel 11/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609922:2610150 [6] NCCL INFO Channel 11/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2609919:2610146 [3] NCCL INFO Channel 04/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609917:2610147 [1] NCCL INFO Channel 12/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609916:2610144 [0] NCCL INFO Channel 15/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609920:2610149 [4] NCCL INFO Channel 11/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609921:2610148 [5] NCCL INFO Channel 13/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609918:2610145 [2] NCCL INFO Channel 09/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609923:2610151 [7] NCCL INFO Channel 12/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609922:2610150 [6] NCCL INFO Channel 12/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2609919:2610146 [3] NCCL INFO Channel 05/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609917:2610147 [1] NCCL INFO Channel 13/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609916:2610144 [0] NCCL INFO Channel 16/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609920:2610149 [4] NCCL INFO Channel 12/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609921:2610148 [5] NCCL INFO Channel 14/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609918:2610145 [2] NCCL INFO Channel 10/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609923:2610151 [7] NCCL INFO Channel 13/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609922:2610150 [6] NCCL INFO Channel 13/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2609919:2610146 [3] NCCL INFO Channel 06/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609917:2610147 [1] NCCL INFO Channel 14/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609916:2610144 [0] NCCL INFO Channel 17/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609920:2610149 [4] NCCL INFO Channel 13/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609921:2610148 [5] NCCL INFO Channel 15/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609923:2610151 [7] NCCL INFO Channel 14/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609918:2610145 [2] NCCL INFO Channel 11/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609922:2610150 [6] NCCL INFO Channel 14/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2609919:2610146 [3] NCCL INFO Channel 07/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609917:2610147 [1] NCCL INFO Channel 15/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609916:2610144 [0] NCCL INFO Channel 18/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609920:2610149 [4] NCCL INFO Channel 14/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609921:2610148 [5] NCCL INFO Channel 16/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609923:2610151 [7] NCCL INFO Channel 15/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609918:2610145 [2] NCCL INFO Channel 12/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609922:2610150 [6] NCCL INFO Channel 15/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2609919:2610146 [3] NCCL INFO Channel 08/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609917:2610147 [1] NCCL INFO Channel 16/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609916:2610144 [0] NCCL INFO Channel 19/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609920:2610149 [4] NCCL INFO Channel 15/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609921:2610148 [5] NCCL INFO Channel 17/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609923:2610151 [7] NCCL INFO Channel 16/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609918:2610145 [2] NCCL INFO Channel 13/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609922:2610150 [6] NCCL INFO Channel 16/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2609919:2610146 [3] NCCL INFO Channel 09/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609917:2610147 [1] NCCL INFO Channel 17/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609916:2610144 [0] NCCL INFO Channel 20/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609920:2610149 [4] NCCL INFO Channel 16/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609921:2610148 [5] NCCL INFO Channel 18/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609923:2610151 [7] NCCL INFO Channel 17/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609918:2610145 [2] NCCL INFO Channel 14/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609922:2610150 [6] NCCL INFO Channel 17/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2609919:2610146 [3] NCCL INFO Channel 10/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609917:2610147 [1] NCCL INFO Channel 18/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609916:2610144 [0] NCCL INFO Channel 21/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609923:2610151 [7] NCCL INFO Channel 18/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609921:2610148 [5] NCCL INFO Channel 19/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609918:2610145 [2] NCCL INFO Channel 15/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609922:2610150 [6] NCCL INFO Channel 18/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2609919:2610146 [3] NCCL INFO Channel 11/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609917:2610147 [1] NCCL INFO Channel 19/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609916:2610144 [0] NCCL INFO Channel 22/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609923:2610151 [7] NCCL INFO Channel 19/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609921:2610148 [5] NCCL INFO Channel 20/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609918:2610145 [2] NCCL INFO Channel 16/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609920:2610149 [4] NCCL INFO Channel 17/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609922:2610150 [6] NCCL INFO Channel 19/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2609919:2610146 [3] NCCL INFO Channel 12/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609917:2610147 [1] NCCL INFO Channel 20/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609916:2610144 [0] NCCL INFO Channel 23/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609923:2610151 [7] NCCL INFO Channel 20/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609921:2610148 [5] NCCL INFO Channel 21/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609918:2610145 [2] NCCL INFO Channel 17/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609920:2610149 [4] NCCL INFO Channel 18/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609922:2610150 [6] NCCL INFO Channel 20/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2609919:2610146 [3] NCCL INFO Channel 13/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609917:2610147 [1] NCCL INFO Channel 21/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609923:2610151 [7] NCCL INFO Channel 21/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609918:2610145 [2] NCCL INFO Channel 18/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609921:2610148 [5] NCCL INFO Channel 22/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609920:2610149 [4] NCCL INFO Channel 19/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609922:2610150 [6] NCCL INFO Channel 21/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2609919:2610146 [3] NCCL INFO Channel 14/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609917:2610147 [1] NCCL INFO Channel 22/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609918:2610145 [2] NCCL INFO Channel 19/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609923:2610151 [7] NCCL INFO Channel 22/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609920:2610149 [4] NCCL INFO Channel 20/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609922:2610150 [6] NCCL INFO Channel 22/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2609919:2610146 [3] NCCL INFO Channel 15/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609921:2610148 [5] NCCL INFO Channel 23/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609917:2610147 [1] NCCL INFO Channel 23/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609918:2610145 [2] NCCL INFO Channel 20/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609920:2610149 [4] NCCL INFO Channel 21/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609923:2610151 [7] NCCL INFO Channel 23/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609922:2610150 [6] NCCL INFO Channel 23/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-103:2609919:2610146 [3] NCCL INFO Channel 16/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609918:2610145 [2] NCCL INFO Channel 21/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609920:2610149 [4] NCCL INFO Channel 22/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609919:2610146 [3] NCCL INFO Channel 17/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609918:2610145 [2] NCCL INFO Channel 22/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609919:2610146 [3] NCCL INFO Channel 18/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609920:2610149 [4] NCCL INFO Channel 23/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609919:2610146 [3] NCCL INFO Channel 19/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609918:2610145 [2] NCCL INFO Channel 23/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609919:2610146 [3] NCCL INFO Channel 20/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609919:2610146 [3] NCCL INFO Channel 21/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609919:2610146 [3] NCCL INFO Channel 22/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609919:2610146 [3] NCCL INFO Channel 23/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609923:2610151 [7] NCCL INFO Connected all rings
+ip-26-0-160-103:2609923:2610151 [7] NCCL INFO Channel 00/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609916:2610144 [0] NCCL INFO Connected all rings
+ip-26-0-160-103:2609923:2610151 [7] NCCL INFO Channel 01/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609922:2610150 [6] NCCL INFO Connected all rings
+ip-26-0-160-103:2609923:2610151 [7] NCCL INFO Channel 02/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609923:2610151 [7] NCCL INFO Channel 03/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609923:2610151 [7] NCCL INFO Channel 04/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609923:2610151 [7] NCCL INFO Channel 05/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609923:2610151 [7] NCCL INFO Channel 06/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609921:2610148 [5] NCCL INFO Connected all rings
+ip-26-0-160-103:2609923:2610151 [7] NCCL INFO Channel 07/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609917:2610147 [1] NCCL INFO Connected all rings
+ip-26-0-160-103:2609923:2610151 [7] NCCL INFO Channel 08/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609918:2610145 [2] NCCL INFO Connected all rings
+ip-26-0-160-103:2609923:2610151 [7] NCCL INFO Channel 09/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609920:2610149 [4] NCCL INFO Connected all rings
+ip-26-0-160-103:2609919:2610146 [3] NCCL INFO Connected all rings
+ip-26-0-160-103:2609923:2610151 [7] NCCL INFO Channel 10/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609923:2610151 [7] NCCL INFO Channel 11/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609923:2610151 [7] NCCL INFO Channel 12/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609923:2610151 [7] NCCL INFO Channel 13/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609923:2610151 [7] NCCL INFO Channel 14/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609923:2610151 [7] NCCL INFO Channel 15/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609923:2610151 [7] NCCL INFO Channel 16/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609923:2610151 [7] NCCL INFO Channel 17/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609923:2610151 [7] NCCL INFO Channel 18/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609923:2610151 [7] NCCL INFO Channel 19/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609923:2610151 [7] NCCL INFO Channel 20/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609923:2610151 [7] NCCL INFO Channel 21/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609921:2610148 [5] NCCL INFO Channel 00/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609923:2610151 [7] NCCL INFO Channel 22/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609921:2610148 [5] NCCL INFO Channel 01/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609923:2610151 [7] NCCL INFO Channel 23/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-103:2609917:2610147 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609921:2610148 [5] NCCL INFO Channel 02/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609917:2610147 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609921:2610148 [5] NCCL INFO Channel 03/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609917:2610147 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609921:2610148 [5] NCCL INFO Channel 04/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609922:2610150 [6] NCCL INFO Channel 00/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609918:2610145 [2] NCCL INFO Channel 00/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609917:2610147 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609921:2610148 [5] NCCL INFO Channel 05/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609922:2610150 [6] NCCL INFO Channel 01/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609918:2610145 [2] NCCL INFO Channel 01/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609919:2610146 [3] NCCL INFO Channel 00/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609917:2610147 [1] NCCL INFO Channel 04/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609921:2610148 [5] NCCL INFO Channel 06/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609922:2610150 [6] NCCL INFO Channel 02/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609918:2610145 [2] NCCL INFO Channel 02/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609919:2610146 [3] NCCL INFO Channel 01/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609917:2610147 [1] NCCL INFO Channel 05/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609921:2610148 [5] NCCL INFO Channel 07/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609922:2610150 [6] NCCL INFO Channel 03/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609918:2610145 [2] NCCL INFO Channel 03/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609919:2610146 [3] NCCL INFO Channel 02/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609917:2610147 [1] NCCL INFO Channel 06/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609921:2610148 [5] NCCL INFO Channel 08/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609922:2610150 [6] NCCL INFO Channel 04/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609918:2610145 [2] NCCL INFO Channel 04/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609919:2610146 [3] NCCL INFO Channel 03/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609917:2610147 [1] NCCL INFO Channel 07/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609921:2610148 [5] NCCL INFO Channel 09/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609922:2610150 [6] NCCL INFO Channel 05/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609918:2610145 [2] NCCL INFO Channel 05/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609919:2610146 [3] NCCL INFO Channel 04/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609917:2610147 [1] NCCL INFO Channel 08/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609921:2610148 [5] NCCL INFO Channel 10/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609922:2610150 [6] NCCL INFO Channel 06/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609918:2610145 [2] NCCL INFO Channel 06/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609919:2610146 [3] NCCL INFO Channel 05/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609917:2610147 [1] NCCL INFO Channel 09/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609921:2610148 [5] NCCL INFO Channel 11/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609922:2610150 [6] NCCL INFO Channel 07/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609918:2610145 [2] NCCL INFO Channel 07/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609919:2610146 [3] NCCL INFO Channel 06/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609917:2610147 [1] NCCL INFO Channel 10/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609921:2610148 [5] NCCL INFO Channel 12/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609922:2610150 [6] NCCL INFO Channel 08/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609918:2610145 [2] NCCL INFO Channel 08/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609919:2610146 [3] NCCL INFO Channel 07/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609920:2610149 [4] NCCL INFO Channel 00/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609917:2610147 [1] NCCL INFO Channel 11/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609921:2610148 [5] NCCL INFO Channel 13/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609922:2610150 [6] NCCL INFO Channel 09/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609918:2610145 [2] NCCL INFO Channel 09/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609919:2610146 [3] NCCL INFO Channel 08/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609920:2610149 [4] NCCL INFO Channel 01/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609917:2610147 [1] NCCL INFO Channel 12/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609921:2610148 [5] NCCL INFO Channel 14/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609922:2610150 [6] NCCL INFO Channel 10/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609918:2610145 [2] NCCL INFO Channel 10/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609919:2610146 [3] NCCL INFO Channel 09/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609920:2610149 [4] NCCL INFO Channel 02/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609917:2610147 [1] NCCL INFO Channel 13/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609921:2610148 [5] NCCL INFO Channel 15/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609922:2610150 [6] NCCL INFO Channel 11/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609918:2610145 [2] NCCL INFO Channel 11/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609920:2610149 [4] NCCL INFO Channel 03/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609919:2610146 [3] NCCL INFO Channel 10/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609917:2610147 [1] NCCL INFO Channel 14/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609921:2610148 [5] NCCL INFO Channel 16/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609922:2610150 [6] NCCL INFO Channel 12/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609918:2610145 [2] NCCL INFO Channel 12/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609920:2610149 [4] NCCL INFO Channel 04/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609919:2610146 [3] NCCL INFO Channel 11/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609917:2610147 [1] NCCL INFO Channel 15/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609921:2610148 [5] NCCL INFO Channel 17/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609922:2610150 [6] NCCL INFO Channel 13/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609918:2610145 [2] NCCL INFO Channel 13/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609920:2610149 [4] NCCL INFO Channel 05/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609917:2610147 [1] NCCL INFO Channel 16/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609919:2610146 [3] NCCL INFO Channel 12/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609921:2610148 [5] NCCL INFO Channel 18/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609922:2610150 [6] NCCL INFO Channel 14/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609918:2610145 [2] NCCL INFO Channel 14/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609920:2610149 [4] NCCL INFO Channel 06/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609919:2610146 [3] NCCL INFO Channel 13/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609917:2610147 [1] NCCL INFO Channel 17/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609921:2610148 [5] NCCL INFO Channel 19/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609922:2610150 [6] NCCL INFO Channel 15/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609918:2610145 [2] NCCL INFO Channel 15/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609920:2610149 [4] NCCL INFO Channel 07/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609919:2610146 [3] NCCL INFO Channel 14/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609921:2610148 [5] NCCL INFO Channel 20/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609917:2610147 [1] NCCL INFO Channel 18/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609922:2610150 [6] NCCL INFO Channel 16/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609918:2610145 [2] NCCL INFO Channel 16/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609920:2610149 [4] NCCL INFO Channel 08/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609919:2610146 [3] NCCL INFO Channel 15/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609921:2610148 [5] NCCL INFO Channel 21/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609922:2610150 [6] NCCL INFO Channel 17/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609920:2610149 [4] NCCL INFO Channel 09/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609919:2610146 [3] NCCL INFO Channel 16/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609921:2610148 [5] NCCL INFO Channel 22/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609922:2610150 [6] NCCL INFO Channel 18/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609920:2610149 [4] NCCL INFO Channel 10/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609919:2610146 [3] NCCL INFO Channel 17/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609921:2610148 [5] NCCL INFO Channel 23/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-103:2609922:2610150 [6] NCCL INFO Channel 19/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609920:2610149 [4] NCCL INFO Channel 11/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609919:2610146 [3] NCCL INFO Channel 18/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609917:2610147 [1] NCCL INFO Channel 19/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609918:2610145 [2] NCCL INFO Channel 17/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609922:2610150 [6] NCCL INFO Channel 20/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609920:2610149 [4] NCCL INFO Channel 12/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609919:2610146 [3] NCCL INFO Channel 19/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609917:2610147 [1] NCCL INFO Channel 20/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609918:2610145 [2] NCCL INFO Channel 18/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609922:2610150 [6] NCCL INFO Channel 21/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609920:2610149 [4] NCCL INFO Channel 13/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609919:2610146 [3] NCCL INFO Channel 20/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609918:2610145 [2] NCCL INFO Channel 19/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609917:2610147 [1] NCCL INFO Channel 21/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609922:2610150 [6] NCCL INFO Channel 22/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609918:2610145 [2] NCCL INFO Channel 20/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609920:2610149 [4] NCCL INFO Channel 14/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609917:2610147 [1] NCCL INFO Channel 22/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609922:2610150 [6] NCCL INFO Channel 23/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-103:2609918:2610145 [2] NCCL INFO Channel 21/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609920:2610149 [4] NCCL INFO Channel 15/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609917:2610147 [1] NCCL INFO Channel 23/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609920:2610149 [4] NCCL INFO Channel 16/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609918:2610145 [2] NCCL INFO Channel 22/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609920:2610149 [4] NCCL INFO Channel 17/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609920:2610149 [4] NCCL INFO Channel 18/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609920:2610149 [4] NCCL INFO Channel 19/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609920:2610149 [4] NCCL INFO Channel 20/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609920:2610149 [4] NCCL INFO Channel 21/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609920:2610149 [4] NCCL INFO Channel 22/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609920:2610149 [4] NCCL INFO Channel 23/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-103:2609919:2610146 [3] NCCL INFO Channel 21/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609918:2610145 [2] NCCL INFO Channel 23/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609919:2610146 [3] NCCL INFO Channel 22/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609919:2610146 [3] NCCL INFO Channel 23/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-103:2609923:2610151 [7] NCCL INFO Connected all trees
+ip-26-0-160-103:2609923:2610151 [7] NCCL INFO NVLS comm 0xa465e80 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-103:2609916:2610144 [0] NCCL INFO Connected all trees
+ip-26-0-160-103:2609916:2610144 [0] NCCL INFO NVLS comm 0x9baf880 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-103:2609922:2610150 [6] NCCL INFO Connected all trees
+ip-26-0-160-103:2609922:2610150 [6] NCCL INFO NVLS comm 0x989a860 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-103:2609921:2610148 [5] NCCL INFO Connected all trees
+ip-26-0-160-103:2609921:2610148 [5] NCCL INFO NVLS comm 0x8fb46f0 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-103:2609918:2610145 [2] NCCL INFO Connected all trees
+ip-26-0-160-103:2609918:2610145 [2] NCCL INFO NVLS comm 0xa3753a0 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-103:2609917:2610147 [1] NCCL INFO Connected all trees
+ip-26-0-160-103:2609917:2610147 [1] NCCL INFO NVLS comm 0x91a1d50 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-103:2609919:2610146 [3] NCCL INFO Connected all trees
+ip-26-0-160-103:2609919:2610146 [3] NCCL INFO NVLS comm 0x87178b0 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-103:2609920:2610149 [4] NCCL INFO Connected all trees
+ip-26-0-160-103:2609920:2610149 [4] NCCL INFO NVLS comm 0x934a5a0 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-103:2609919:2610146 [3] NCCL INFO Connected NVLS tree
+ip-26-0-160-103:2609919:2610146 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-103:2609919:2610146 [3] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-160-103:2609919:2610146 [3] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-160-103:2609922:2610150 [6] NCCL INFO Connected NVLS tree
+ip-26-0-160-103:2609922:2610150 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-103:2609922:2610150 [6] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-160-103:2609922:2610150 [6] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-160-103:2609920:2610149 [4] NCCL INFO Connected NVLS tree
+ip-26-0-160-103:2609920:2610149 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-103:2609916:2610144 [0] NCCL INFO Connected NVLS tree
+ip-26-0-160-103:2609920:2610149 [4] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-160-103:2609920:2610149 [4] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-160-103:2609916:2610144 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-103:2609918:2610145 [2] NCCL INFO Connected NVLS tree
+ip-26-0-160-103:2609918:2610145 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-103:2609918:2610145 [2] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-160-103:2609918:2610145 [2] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-160-103:2609916:2610144 [0] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-160-103:2609916:2610144 [0] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-160-103:2609917:2610147 [1] NCCL INFO Connected NVLS tree
+ip-26-0-160-103:2609917:2610147 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-103:2609917:2610147 [1] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-160-103:2609917:2610147 [1] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-160-103:2609923:2610151 [7] NCCL INFO Connected NVLS tree
+ip-26-0-160-103:2609923:2610151 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-103:2609923:2610151 [7] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-160-103:2609923:2610151 [7] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-160-103:2609921:2610148 [5] NCCL INFO Connected NVLS tree
+ip-26-0-160-103:2609921:2610148 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-103:2609921:2610148 [5] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-160-103:2609921:2610148 [5] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-160-103:2609918:2610145 [2] NCCL INFO comm 0xa3753a0 rank 2 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0x5fae3ab7c1a3f336 - Init COMPLETE
+ip-26-0-160-103:2609922:2610150 [6] NCCL INFO comm 0x989a860 rank 6 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0x5fae3ab7c1a3f336 - Init COMPLETE
+ip-26-0-160-103:2609920:2610149 [4] NCCL INFO comm 0x934a5a0 rank 4 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0x5fae3ab7c1a3f336 - Init COMPLETE
+ip-26-0-160-103:2609916:2610144 [0] NCCL INFO comm 0x9baf880 rank 0 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0x5fae3ab7c1a3f336 - Init COMPLETE
+ip-26-0-160-103:2609921:2610148 [5] NCCL INFO comm 0x8fb46f0 rank 5 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0x5fae3ab7c1a3f336 - Init COMPLETE
+ip-26-0-160-103:2609917:2610147 [1] NCCL INFO comm 0x91a1d50 rank 1 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0x5fae3ab7c1a3f336 - Init COMPLETE
+ip-26-0-160-103:2609919:2610146 [3] NCCL INFO comm 0x87178b0 rank 3 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0x5fae3ab7c1a3f336 - Init COMPLETE
+ip-26-0-160-103:2609923:2610151 [7] NCCL INFO comm 0xa465e80 rank 7 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0x5fae3ab7c1a3f336 - Init COMPLETE
+12/13/2024 14:17:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]: Config:
+12/13/2024 14:17:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]: Config(general=GeneralArgs(project='debug',
+12/13/2024 14:17:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                            run='1.14G_dp4_tp2_pp1_acc256_mbs2_seq2048_zero1_tpmodeRED_vocab32k',
+12/13/2024 14:17:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                            seed=42,
+12/13/2024 14:17:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                            step=None,
+12/13/2024 14:17:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                            consumed_train_samples=None,
+12/13/2024 14:17:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                            benchmark_csv_path=PosixPath('benchmark/results/bench_final.csv'),
+12/13/2024 14:17:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                            ignore_sanity_checks=True),
+12/13/2024 14:17:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:        parallelism=ParallelismArgs(dp=4,
+12/13/2024 14:17:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                    pp=1,
+12/13/2024 14:17:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                    tp=2,
+12/13/2024 14:17:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                    pp_engine=<nanotron.parallel.pipeline_parallel.engine.OneForwardOneBackwardPipelineEngine object at 0x7facd0d48d00>,
+12/13/2024 14:17:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                    tp_mode=<TensorParallelLinearMode.REDUCE_SCATTER: 2>,
+12/13/2024 14:17:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                    tp_linear_async_communication=True,
+12/13/2024 14:17:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                    recompute_layer=False,
+12/13/2024 14:17:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                    tp_recompute_allgather=True,
+12/13/2024 14:17:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                    expert_parallel_size=1),
+12/13/2024 14:17:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:        model=ModelArgs(model_config=LlamaConfig(bos_token_id=0,
+12/13/2024 14:17:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                                 eos_token_id=0,
+12/13/2024 14:17:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                                 hidden_act='silu',
+12/13/2024 14:17:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                                 hidden_size=2048,
+12/13/2024 14:17:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                                 initializer_range=0.02,
+12/13/2024 14:17:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                                 intermediate_size=8192,
+12/13/2024 14:17:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                                 is_llama_config=True,
+12/13/2024 14:17:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                                 max_position_embeddings=2048,
+12/13/2024 14:17:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                                 num_attention_heads=32,
+12/13/2024 14:17:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                                 num_hidden_layers=16,
+12/13/2024 14:17:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                                 num_key_value_heads=8,
+12/13/2024 14:17:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                                 pad_token_id=None,
+12/13/2024 14:17:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                                 pretraining_tp=1,
+12/13/2024 14:17:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                                 rms_norm_eps=1e-05,
+12/13/2024 14:17:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                                 rope_scaling=None,
+12/13/2024 14:17:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                                 rope_theta=10000.0,
+12/13/2024 14:17:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                                 rope_interleaved=False,
+12/13/2024 14:17:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                                 tie_word_embeddings=True,
+12/13/2024 14:17:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                                 use_cache=True,
+12/13/2024 14:17:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                                 vocab_size=32768),
+12/13/2024 14:17:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                        init_method=RandomInit(std=0.02),
+12/13/2024 14:17:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                        dtype=torch.bfloat16,
+12/13/2024 14:17:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                        make_vocab_size_divisible_by=1,
+12/13/2024 14:17:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                        ddp_bucket_cap_mb=25),
+12/13/2024 14:17:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:        tokenizer=TokenizerArgs(tokenizer_name_or_path='robot-test/dummy-tokenizer-wordlevel',
+12/13/2024 14:17:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                tokenizer_revision=None,
+12/13/2024 14:17:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                tokenizer_max_length=None),
+12/13/2024 14:17:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:        checkpoints=CheckpointsArgs(checkpoints_path=PosixPath('checkpoints'),
+12/13/2024 14:17:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                    checkpoint_interval=10000,
+12/13/2024 14:17:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                    save_initial_state=False,
+12/13/2024 14:17:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                    save_final_state=False,
+12/13/2024 14:17:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                    resume_checkpoint_path=None,
+12/13/2024 14:17:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                    checkpoints_path_is_shared_file_system=False),
+12/13/2024 14:17:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:        logging=LoggingArgs(log_level='info',
+12/13/2024 14:17:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                            log_level_replica='info',
+12/13/2024 14:17:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                            iteration_step_info_interval=1),
+12/13/2024 14:17:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:        tokens=TokensArgs(sequence_length=2048,
+12/13/2024 14:17:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                          train_steps=100,
+12/13/2024 14:17:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                          micro_batch_size=2,
+12/13/2024 14:17:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                          batch_accumulation_per_replica=256,
+12/13/2024 14:17:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                          val_check_interval=100,
+12/13/2024 14:17:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                          limit_val_batches=0,
+12/13/2024 14:17:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                          limit_test_batches=0),
+12/13/2024 14:17:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:        optimizer=OptimizerArgs(optimizer_factory=AdamWOptimizerArgs(adam_eps=1e-08,
+12/13/2024 14:17:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                                                     adam_beta1=0.9,
+12/13/2024 14:17:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                                                     adam_beta2=0.95,
+12/13/2024 14:17:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                                                     torch_adam_is_fused=True,
+12/13/2024 14:17:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                                                     name='adamW'),
+12/13/2024 14:17:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                zero_stage=1,
+12/13/2024 14:17:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                weight_decay=0.01,
+12/13/2024 14:17:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                clip_grad=1.0,
+12/13/2024 14:17:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                accumulate_grad_in_fp32=True,
+12/13/2024 14:17:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                learning_rate_scheduler=LRSchedulerArgs(learning_rate=0.0003,
+12/13/2024 14:17:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                                                        lr_warmup_steps=2,
+12/13/2024 14:17:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                                                        lr_warmup_style='linear',
+12/13/2024 14:17:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                                                        lr_decay_style='cosine',
+12/13/2024 14:17:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                                                        lr_decay_steps=13,
+12/13/2024 14:17:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                                                        lr_decay_starting_step=None,
+12/13/2024 14:17:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                                                        min_decay_lr=1e-05)),
+12/13/2024 14:17:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:        data_stages=[DatasetStageArgs(name='Stable Training Stage',
+12/13/2024 14:17:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                      start_training_step=1,
+12/13/2024 14:17:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                      data=DataArgs(dataset=None,
+12/13/2024 14:17:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                                    seed=42,
+12/13/2024 14:17:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                                    num_loading_workers=1))],
+12/13/2024 14:17:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:        profiler=None,
+12/13/2024 14:17:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:        lighteval=None,
+12/13/2024 14:17:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:        s3_upload=None)
+12/13/2024 14:17:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]: Model Config:
+12/13/2024 14:17:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]: LlamaConfig(bos_token_id=0,
+12/13/2024 14:17:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:             eos_token_id=0,
+12/13/2024 14:17:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:             hidden_act='silu',
+12/13/2024 14:17:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:             hidden_size=2048,
+12/13/2024 14:17:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:             initializer_range=0.02,
+12/13/2024 14:17:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:             intermediate_size=8192,
+12/13/2024 14:17:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:             is_llama_config=True,
+12/13/2024 14:17:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:             max_position_embeddings=2048,
+12/13/2024 14:17:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:             num_attention_heads=32,
+12/13/2024 14:17:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:             num_hidden_layers=16,
+12/13/2024 14:17:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:             num_key_value_heads=8,
+12/13/2024 14:17:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:             pad_token_id=None,
+12/13/2024 14:17:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:             pretraining_tp=1,
+12/13/2024 14:17:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:             rms_norm_eps=1e-05,
+12/13/2024 14:17:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:             rope_scaling=None,
+12/13/2024 14:17:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:             rope_theta=10000.0,
+12/13/2024 14:17:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:             rope_interleaved=False,
+12/13/2024 14:17:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:             tie_word_embeddings=True,
+12/13/2024 14:17:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:             use_cache=True,
+12/13/2024 14:17:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:             vocab_size=32768)
+12/13/2024 14:17:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]: Building model..
+12/13/2024 14:17:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]: Initialize RoPE Theta = 10000.0
+12/13/2024 14:17:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]: Setting PP block ranks...
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+ip-26-0-160-103:2609916:2610199 [0] NCCL INFO Using network AWS Libfabric
+ip-26-0-160-103:2609922:2610200 [6] NCCL INFO Using network AWS Libfabric
+ip-26-0-160-103:2609919:2610201 [3] NCCL INFO Using network AWS Libfabric
+ip-26-0-160-103:2609918:2610198 [2] NCCL INFO Using network AWS Libfabric
+ip-26-0-160-103:2609917:2610202 [1] NCCL INFO Using network AWS Libfabric
+ip-26-0-160-103:2609923:2610203 [7] NCCL INFO Using network AWS Libfabric
+ip-26-0-160-103:2609918:2610198 [2] NCCL INFO comm 0xc175cb0 rank 0 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0xaa4d6a470f5df587 - Init START
+ip-26-0-160-103:2609919:2610201 [3] NCCL INFO comm 0xa51d7e0 rank 1 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0xaa4d6a470f5df587 - Init START
+ip-26-0-160-103:2609920:2610205 [4] NCCL INFO Using network AWS Libfabric
+ip-26-0-160-103:2609916:2610199 [0] NCCL INFO comm 0xb9b4270 rank 0 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0x2c3909f8817a8fb9 - Init START
+ip-26-0-160-103:2609917:2610202 [1] NCCL INFO comm 0xafa7870 rank 1 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0x2c3909f8817a8fb9 - Init START
+ip-26-0-160-103:2609921:2610206 [5] NCCL INFO Using network AWS Libfabric
+ip-26-0-160-103:2609918:2610198 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /opt/aws-ofi-nccl/share/aws-ofi-nccl/xml/p5.48xl-topo.xml
+ip-26-0-160-103:2609916:2610199 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /opt/aws-ofi-nccl/share/aws-ofi-nccl/xml/p5.48xl-topo.xml
+ip-26-0-160-103:2609922:2610200 [6] NCCL INFO comm 0xb69a470 rank 0 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0x6bb75a1aa2d88ceb - Init START
+ip-26-0-160-103:2609923:2610203 [7] NCCL INFO comm 0xc26b3e0 rank 1 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0x6bb75a1aa2d88ceb - Init START
+ip-26-0-160-103:2609922:2610200 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /opt/aws-ofi-nccl/share/aws-ofi-nccl/xml/p5.48xl-topo.xml
+ip-26-0-160-103:2609921:2610206 [5] NCCL INFO comm 0xadb73f0 rank 1 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0x8ad574a31761b571 - Init START
+ip-26-0-160-103:2609920:2610205 [4] NCCL INFO comm 0xb150330 rank 0 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0x8ad574a31761b571 - Init START
+ip-26-0-160-103:2609920:2610205 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /opt/aws-ofi-nccl/share/aws-ofi-nccl/xml/p5.48xl-topo.xml
+ip-26-0-160-103:2609919:2610201 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /opt/aws-ofi-nccl/share/aws-ofi-nccl/xml/p5.48xl-topo.xml
+ip-26-0-160-103:2609917:2610202 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /opt/aws-ofi-nccl/share/aws-ofi-nccl/xml/p5.48xl-topo.xml
+ip-26-0-160-103:2609921:2610206 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /opt/aws-ofi-nccl/share/aws-ofi-nccl/xml/p5.48xl-topo.xml
+ip-26-0-160-103:2609923:2610203 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /opt/aws-ofi-nccl/share/aws-ofi-nccl/xml/p5.48xl-topo.xml
+ip-26-0-160-103:2609921:2610206 [5] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609921:2610206 [5] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609921:2610206 [5] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609921:2610206 [5] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609921:2610206 [5] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609921:2610206 [5] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609921:2610206 [5] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609921:2610206 [5] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609921:2610206 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-160-103:2609917:2610202 [1] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609917:2610202 [1] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609917:2610202 [1] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609917:2610202 [1] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609917:2610202 [1] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609917:2610202 [1] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609917:2610202 [1] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609916:2610199 [0] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609917:2610202 [1] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609916:2610199 [0] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609916:2610199 [0] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609916:2610199 [0] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609916:2610199 [0] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609917:2610202 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-160-103:2609916:2610199 [0] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609916:2610199 [0] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609916:2610199 [0] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609916:2610199 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-160-103:2609920:2610205 [4] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609920:2610205 [4] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609920:2610205 [4] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609920:2610205 [4] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609916:2610199 [0] NCCL INFO Channel 00/24 :    0   1
+ip-26-0-160-103:2609916:2610199 [0] NCCL INFO Channel 01/24 :    0   1
+ip-26-0-160-103:2609916:2610199 [0] NCCL INFO Channel 02/24 :    0   1
+ip-26-0-160-103:2609916:2610199 [0] NCCL INFO Channel 03/24 :    0   1
+ip-26-0-160-103:2609916:2610199 [0] NCCL INFO Channel 04/24 :    0   1
+ip-26-0-160-103:2609916:2610199 [0] NCCL INFO Channel 05/24 :    0   1
+ip-26-0-160-103:2609916:2610199 [0] NCCL INFO Channel 06/24 :    0   1
+ip-26-0-160-103:2609916:2610199 [0] NCCL INFO Channel 07/24 :    0   1
+ip-26-0-160-103:2609916:2610199 [0] NCCL INFO Channel 08/24 :    0   1
+ip-26-0-160-103:2609916:2610199 [0] NCCL INFO Channel 09/24 :    0   1
+ip-26-0-160-103:2609916:2610199 [0] NCCL INFO Channel 10/24 :    0   1
+ip-26-0-160-103:2609916:2610199 [0] NCCL INFO Channel 11/24 :    0   1
+ip-26-0-160-103:2609916:2610199 [0] NCCL INFO Channel 12/24 :    0   1
+ip-26-0-160-103:2609916:2610199 [0] NCCL INFO Channel 13/24 :    0   1
+ip-26-0-160-103:2609916:2610199 [0] NCCL INFO Channel 14/24 :    0   1
+ip-26-0-160-103:2609916:2610199 [0] NCCL INFO Channel 15/24 :    0   1
+ip-26-0-160-103:2609916:2610199 [0] NCCL INFO Channel 16/24 :    0   1
+ip-26-0-160-103:2609916:2610199 [0] NCCL INFO Channel 17/24 :    0   1
+ip-26-0-160-103:2609916:2610199 [0] NCCL INFO Channel 18/24 :    0   1
+ip-26-0-160-103:2609916:2610199 [0] NCCL INFO Channel 19/24 :    0   1
+ip-26-0-160-103:2609916:2610199 [0] NCCL INFO Channel 20/24 :    0   1
+ip-26-0-160-103:2609916:2610199 [0] NCCL INFO Channel 21/24 :    0   1
+ip-26-0-160-103:2609916:2610199 [0] NCCL INFO Channel 22/24 :    0   1
+ip-26-0-160-103:2609916:2610199 [0] NCCL INFO Channel 23/24 :    0   1
+ip-26-0-160-103:2609916:2610199 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] -1/-1/-1->0->1 [7] -1/-1/-1->0->1 [8] -1/-1/-1->0->1 [9] -1/-1/-1->0->1 [10] -1/-1/-1->0->1 [11] -1/-1/-1->0->1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] -1/-1/-1->0->1 [19] -1/-1/-1->0->1 [20] -1/-1/-1->0->1 [21] -1/-1/-1->0->1 [22] -1/-1/-1->0->1 [23] -1/-1/-1->0->1
+ip-26-0-160-103:2609916:2610199 [0] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-160-103:2609920:2610205 [4] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609920:2610205 [4] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609920:2610205 [4] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609920:2610205 [4] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609917:2610202 [1] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] -1/-1/-1->1->0 [3] -1/-1/-1->1->0 [4] -1/-1/-1->1->0 [5] -1/-1/-1->1->0 [6] 0/-1/-1->1->-1 [7] 0/-1/-1->1->-1 [8] 0/-1/-1->1->-1 [9] 0/-1/-1->1->-1 [10] 0/-1/-1->1->-1 [11] 0/-1/-1->1->-1 [12] -1/-1/-1->1->0 [13] -1/-1/-1->1->0 [14] -1/-1/-1->1->0 [15] -1/-1/-1->1->0 [16] -1/-1/-1->1->0 [17] -1/-1/-1->1->0 [18] 0/-1/-1->1->-1 [19] 0/-1/-1->1->-1 [20] 0/-1/-1->1->-1 [21] 0/-1/-1->1->-1 [22] 0/-1/-1->1->-1 [23] 0/-1/-1->1->-1
+ip-26-0-160-103:2609917:2610202 [1] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-160-103:2609920:2610205 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-160-103:2609920:2610205 [4] NCCL INFO Channel 00/24 :    0   1
+ip-26-0-160-103:2609921:2610206 [5] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] -1/-1/-1->1->0 [3] -1/-1/-1->1->0 [4] -1/-1/-1->1->0 [5] -1/-1/-1->1->0 [6] 0/-1/-1->1->-1 [7] 0/-1/-1->1->-1 [8] 0/-1/-1->1->-1 [9] 0/-1/-1->1->-1 [10] 0/-1/-1->1->-1 [11] 0/-1/-1->1->-1 [12] -1/-1/-1->1->0 [13] -1/-1/-1->1->0 [14] -1/-1/-1->1->0 [15] -1/-1/-1->1->0 [16] -1/-1/-1->1->0 [17] -1/-1/-1->1->0 [18] 0/-1/-1->1->-1 [19] 0/-1/-1->1->-1 [20] 0/-1/-1->1->-1 [21] 0/-1/-1->1->-1 [22] 0/-1/-1->1->-1 [23] 0/-1/-1->1->-1
+ip-26-0-160-103:2609920:2610205 [4] NCCL INFO Channel 01/24 :    0   1
+ip-26-0-160-103:2609920:2610205 [4] NCCL INFO Channel 02/24 :    0   1
+ip-26-0-160-103:2609921:2610206 [5] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-160-103:2609920:2610205 [4] NCCL INFO Channel 03/24 :    0   1
+ip-26-0-160-103:2609920:2610205 [4] NCCL INFO Channel 04/24 :    0   1
+ip-26-0-160-103:2609920:2610205 [4] NCCL INFO Channel 05/24 :    0   1
+ip-26-0-160-103:2609920:2610205 [4] NCCL INFO Channel 06/24 :    0   1
+ip-26-0-160-103:2609920:2610205 [4] NCCL INFO Channel 07/24 :    0   1
+ip-26-0-160-103:2609920:2610205 [4] NCCL INFO Channel 08/24 :    0   1
+ip-26-0-160-103:2609920:2610205 [4] NCCL INFO Channel 09/24 :    0   1
+ip-26-0-160-103:2609920:2610205 [4] NCCL INFO Channel 10/24 :    0   1
+ip-26-0-160-103:2609920:2610205 [4] NCCL INFO Channel 11/24 :    0   1
+ip-26-0-160-103:2609920:2610205 [4] NCCL INFO Channel 12/24 :    0   1
+ip-26-0-160-103:2609920:2610205 [4] NCCL INFO Channel 13/24 :    0   1
+ip-26-0-160-103:2609920:2610205 [4] NCCL INFO Channel 14/24 :    0   1
+ip-26-0-160-103:2609920:2610205 [4] NCCL INFO Channel 15/24 :    0   1
+ip-26-0-160-103:2609920:2610205 [4] NCCL INFO Channel 16/24 :    0   1
+ip-26-0-160-103:2609920:2610205 [4] NCCL INFO Channel 17/24 :    0   1
+ip-26-0-160-103:2609920:2610205 [4] NCCL INFO Channel 18/24 :    0   1
+ip-26-0-160-103:2609920:2610205 [4] NCCL INFO Channel 19/24 :    0   1
+ip-26-0-160-103:2609920:2610205 [4] NCCL INFO Channel 20/24 :    0   1
+ip-26-0-160-103:2609920:2610205 [4] NCCL INFO Channel 21/24 :    0   1
+ip-26-0-160-103:2609920:2610205 [4] NCCL INFO Channel 22/24 :    0   1
+ip-26-0-160-103:2609920:2610205 [4] NCCL INFO Channel 23/24 :    0   1
+ip-26-0-160-103:2609920:2610205 [4] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] -1/-1/-1->0->1 [7] -1/-1/-1->0->1 [8] -1/-1/-1->0->1 [9] -1/-1/-1->0->1 [10] -1/-1/-1->0->1 [11] -1/-1/-1->0->1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] -1/-1/-1->0->1 [19] -1/-1/-1->0->1 [20] -1/-1/-1->0->1 [21] -1/-1/-1->0->1 [22] -1/-1/-1->0->1 [23] -1/-1/-1->0->1
+ip-26-0-160-103:2609920:2610205 [4] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-160-103:2609919:2610201 [3] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609919:2610201 [3] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609919:2610201 [3] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609919:2610201 [3] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609919:2610201 [3] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609919:2610201 [3] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609919:2610201 [3] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609919:2610201 [3] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609919:2610201 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-160-103:2609923:2610203 [7] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609923:2610203 [7] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609923:2610203 [7] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609923:2610203 [7] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609923:2610203 [7] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609923:2610203 [7] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609923:2610203 [7] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609923:2610203 [7] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609923:2610203 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-160-103:2609918:2610198 [2] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609918:2610198 [2] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609918:2610198 [2] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609918:2610198 [2] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609918:2610198 [2] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609918:2610198 [2] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609918:2610198 [2] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609918:2610198 [2] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609918:2610198 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-160-103:2609918:2610198 [2] NCCL INFO Channel 00/24 :    0   1
+ip-26-0-160-103:2609918:2610198 [2] NCCL INFO Channel 01/24 :    0   1
+ip-26-0-160-103:2609918:2610198 [2] NCCL INFO Channel 02/24 :    0   1
+ip-26-0-160-103:2609918:2610198 [2] NCCL INFO Channel 03/24 :    0   1
+ip-26-0-160-103:2609918:2610198 [2] NCCL INFO Channel 04/24 :    0   1
+ip-26-0-160-103:2609918:2610198 [2] NCCL INFO Channel 05/24 :    0   1
+ip-26-0-160-103:2609918:2610198 [2] NCCL INFO Channel 06/24 :    0   1
+ip-26-0-160-103:2609918:2610198 [2] NCCL INFO Channel 07/24 :    0   1
+ip-26-0-160-103:2609918:2610198 [2] NCCL INFO Channel 08/24 :    0   1
+ip-26-0-160-103:2609918:2610198 [2] NCCL INFO Channel 09/24 :    0   1
+ip-26-0-160-103:2609918:2610198 [2] NCCL INFO Channel 10/24 :    0   1
+ip-26-0-160-103:2609918:2610198 [2] NCCL INFO Channel 11/24 :    0   1
+ip-26-0-160-103:2609918:2610198 [2] NCCL INFO Channel 12/24 :    0   1
+ip-26-0-160-103:2609918:2610198 [2] NCCL INFO Channel 13/24 :    0   1
+ip-26-0-160-103:2609918:2610198 [2] NCCL INFO Channel 14/24 :    0   1
+ip-26-0-160-103:2609918:2610198 [2] NCCL INFO Channel 15/24 :    0   1
+ip-26-0-160-103:2609918:2610198 [2] NCCL INFO Channel 16/24 :    0   1
+ip-26-0-160-103:2609919:2610201 [3] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] -1/-1/-1->1->0 [3] -1/-1/-1->1->0 [4] -1/-1/-1->1->0 [5] -1/-1/-1->1->0 [6] 0/-1/-1->1->-1 [7] 0/-1/-1->1->-1 [8] 0/-1/-1->1->-1 [9] 0/-1/-1->1->-1 [10] 0/-1/-1->1->-1 [11] 0/-1/-1->1->-1 [12] -1/-1/-1->1->0 [13] -1/-1/-1->1->0 [14] -1/-1/-1->1->0 [15] -1/-1/-1->1->0 [16] -1/-1/-1->1->0 [17] -1/-1/-1->1->0 [18] 0/-1/-1->1->-1 [19] 0/-1/-1->1->-1 [20] 0/-1/-1->1->-1 [21] 0/-1/-1->1->-1 [22] 0/-1/-1->1->-1 [23] 0/-1/-1->1->-1
+ip-26-0-160-103:2609918:2610198 [2] NCCL INFO Channel 17/24 :    0   1
+ip-26-0-160-103:2609918:2610198 [2] NCCL INFO Channel 18/24 :    0   1
+ip-26-0-160-103:2609919:2610201 [3] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-160-103:2609918:2610198 [2] NCCL INFO Channel 19/24 :    0   1
+ip-26-0-160-103:2609918:2610198 [2] NCCL INFO Channel 20/24 :    0   1
+ip-26-0-160-103:2609918:2610198 [2] NCCL INFO Channel 21/24 :    0   1
+ip-26-0-160-103:2609918:2610198 [2] NCCL INFO Channel 22/24 :    0   1
+ip-26-0-160-103:2609918:2610198 [2] NCCL INFO Channel 23/24 :    0   1
+ip-26-0-160-103:2609918:2610198 [2] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] -1/-1/-1->0->1 [7] -1/-1/-1->0->1 [8] -1/-1/-1->0->1 [9] -1/-1/-1->0->1 [10] -1/-1/-1->0->1 [11] -1/-1/-1->0->1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] -1/-1/-1->0->1 [19] -1/-1/-1->0->1 [20] -1/-1/-1->0->1 [21] -1/-1/-1->0->1 [22] -1/-1/-1->0->1 [23] -1/-1/-1->0->1
+ip-26-0-160-103:2609918:2610198 [2] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-160-103:2609922:2610200 [6] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609922:2610200 [6] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609922:2610200 [6] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609922:2610200 [6] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609922:2610200 [6] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609922:2610200 [6] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609922:2610200 [6] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609922:2610200 [6] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609922:2610200 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-160-103:2609922:2610200 [6] NCCL INFO Channel 00/24 :    0   1
+ip-26-0-160-103:2609922:2610200 [6] NCCL INFO Channel 01/24 :    0   1
+ip-26-0-160-103:2609922:2610200 [6] NCCL INFO Channel 02/24 :    0   1
+ip-26-0-160-103:2609922:2610200 [6] NCCL INFO Channel 03/24 :    0   1
+ip-26-0-160-103:2609922:2610200 [6] NCCL INFO Channel 04/24 :    0   1
+ip-26-0-160-103:2609922:2610200 [6] NCCL INFO Channel 05/24 :    0   1
+ip-26-0-160-103:2609922:2610200 [6] NCCL INFO Channel 06/24 :    0   1
+ip-26-0-160-103:2609922:2610200 [6] NCCL INFO Channel 07/24 :    0   1
+ip-26-0-160-103:2609922:2610200 [6] NCCL INFO Channel 08/24 :    0   1
+ip-26-0-160-103:2609922:2610200 [6] NCCL INFO Channel 09/24 :    0   1
+ip-26-0-160-103:2609922:2610200 [6] NCCL INFO Channel 10/24 :    0   1
+ip-26-0-160-103:2609922:2610200 [6] NCCL INFO Channel 11/24 :    0   1
+ip-26-0-160-103:2609922:2610200 [6] NCCL INFO Channel 12/24 :    0   1
+ip-26-0-160-103:2609922:2610200 [6] NCCL INFO Channel 13/24 :    0   1
+ip-26-0-160-103:2609922:2610200 [6] NCCL INFO Channel 14/24 :    0   1
+ip-26-0-160-103:2609922:2610200 [6] NCCL INFO Channel 15/24 :    0   1
+ip-26-0-160-103:2609922:2610200 [6] NCCL INFO Channel 16/24 :    0   1
+ip-26-0-160-103:2609922:2610200 [6] NCCL INFO Channel 17/24 :    0   1
+ip-26-0-160-103:2609922:2610200 [6] NCCL INFO Channel 18/24 :    0   1
+ip-26-0-160-103:2609922:2610200 [6] NCCL INFO Channel 19/24 :    0   1
+ip-26-0-160-103:2609922:2610200 [6] NCCL INFO Channel 20/24 :    0   1
+ip-26-0-160-103:2609922:2610200 [6] NCCL INFO Channel 21/24 :    0   1
+ip-26-0-160-103:2609922:2610200 [6] NCCL INFO Channel 22/24 :    0   1
+ip-26-0-160-103:2609922:2610200 [6] NCCL INFO Channel 23/24 :    0   1
+ip-26-0-160-103:2609922:2610200 [6] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] -1/-1/-1->0->1 [7] -1/-1/-1->0->1 [8] -1/-1/-1->0->1 [9] -1/-1/-1->0->1 [10] -1/-1/-1->0->1 [11] -1/-1/-1->0->1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] -1/-1/-1->0->1 [19] -1/-1/-1->0->1 [20] -1/-1/-1->0->1 [21] -1/-1/-1->0->1 [22] -1/-1/-1->0->1 [23] -1/-1/-1->0->1
+ip-26-0-160-103:2609922:2610200 [6] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-160-103:2609923:2610203 [7] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] -1/-1/-1->1->0 [3] -1/-1/-1->1->0 [4] -1/-1/-1->1->0 [5] -1/-1/-1->1->0 [6] 0/-1/-1->1->-1 [7] 0/-1/-1->1->-1 [8] 0/-1/-1->1->-1 [9] 0/-1/-1->1->-1 [10] 0/-1/-1->1->-1 [11] 0/-1/-1->1->-1 [12] -1/-1/-1->1->0 [13] -1/-1/-1->1->0 [14] -1/-1/-1->1->0 [15] -1/-1/-1->1->0 [16] -1/-1/-1->1->0 [17] -1/-1/-1->1->0 [18] 0/-1/-1->1->-1 [19] 0/-1/-1->1->-1 [20] 0/-1/-1->1->-1 [21] 0/-1/-1->1->-1 [22] 0/-1/-1->1->-1 [23] 0/-1/-1->1->-1
+ip-26-0-160-103:2609923:2610203 [7] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-160-103:2609917:2610202 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609917:2610202 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609917:2610202 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609919:2610201 [3] NCCL INFO Channel 00/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-160-103:2609917:2610202 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609919:2610201 [3] NCCL INFO Channel 01/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-160-103:2609917:2610202 [1] NCCL INFO Channel 04/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609919:2610201 [3] NCCL INFO Channel 02/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-160-103:2609917:2610202 [1] NCCL INFO Channel 05/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609919:2610201 [3] NCCL INFO Channel 03/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-160-103:2609917:2610202 [1] NCCL INFO Channel 06/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609916:2610199 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609919:2610201 [3] NCCL INFO Channel 04/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-160-103:2609917:2610202 [1] NCCL INFO Channel 07/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609916:2610199 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609919:2610201 [3] NCCL INFO Channel 05/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-160-103:2609917:2610202 [1] NCCL INFO Channel 08/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609918:2610198 [2] NCCL INFO Channel 00/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2609916:2610199 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609919:2610201 [3] NCCL INFO Channel 06/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-160-103:2609917:2610202 [1] NCCL INFO Channel 09/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609921:2610206 [5] NCCL INFO Channel 00/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-160-103:2609918:2610198 [2] NCCL INFO Channel 01/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2609916:2610199 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609919:2610201 [3] NCCL INFO Channel 07/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-160-103:2609917:2610202 [1] NCCL INFO Channel 10/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609921:2610206 [5] NCCL INFO Channel 01/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-160-103:2609918:2610198 [2] NCCL INFO Channel 02/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2609916:2610199 [0] NCCL INFO Channel 04/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609919:2610201 [3] NCCL INFO Channel 08/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-160-103:2609917:2610202 [1] NCCL INFO Channel 11/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609921:2610206 [5] NCCL INFO Channel 02/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-160-103:2609916:2610199 [0] NCCL INFO Channel 05/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609918:2610198 [2] NCCL INFO Channel 03/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2609919:2610201 [3] NCCL INFO Channel 09/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-160-103:2609917:2610202 [1] NCCL INFO Channel 12/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609921:2610206 [5] NCCL INFO Channel 03/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-160-103:2609916:2610199 [0] NCCL INFO Channel 06/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609918:2610198 [2] NCCL INFO Channel 04/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2609919:2610201 [3] NCCL INFO Channel 10/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-160-103:2609922:2610200 [6] NCCL INFO Channel 00/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-160-103:2609923:2610203 [7] NCCL INFO Channel 00/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-160-103:2609917:2610202 [1] NCCL INFO Channel 13/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609921:2610206 [5] NCCL INFO Channel 04/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-160-103:2609916:2610199 [0] NCCL INFO Channel 07/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609918:2610198 [2] NCCL INFO Channel 05/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2609919:2610201 [3] NCCL INFO Channel 11/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-160-103:2609922:2610200 [6] NCCL INFO Channel 01/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-160-103:2609923:2610203 [7] NCCL INFO Channel 01/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-160-103:2609921:2610206 [5] NCCL INFO Channel 05/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-160-103:2609917:2610202 [1] NCCL INFO Channel 14/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609916:2610199 [0] NCCL INFO Channel 08/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609918:2610198 [2] NCCL INFO Channel 06/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2609919:2610201 [3] NCCL INFO Channel 12/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-160-103:2609922:2610200 [6] NCCL INFO Channel 02/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-160-103:2609923:2610203 [7] NCCL INFO Channel 02/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-160-103:2609920:2610205 [4] NCCL INFO Channel 00/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-160-103:2609921:2610206 [5] NCCL INFO Channel 06/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-160-103:2609917:2610202 [1] NCCL INFO Channel 15/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609916:2610199 [0] NCCL INFO Channel 09/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609918:2610198 [2] NCCL INFO Channel 07/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2609922:2610200 [6] NCCL INFO Channel 03/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-160-103:2609919:2610201 [3] NCCL INFO Channel 13/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-160-103:2609923:2610203 [7] NCCL INFO Channel 03/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-160-103:2609920:2610205 [4] NCCL INFO Channel 01/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-160-103:2609921:2610206 [5] NCCL INFO Channel 07/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-160-103:2609917:2610202 [1] NCCL INFO Channel 16/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609918:2610198 [2] NCCL INFO Channel 08/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2609916:2610199 [0] NCCL INFO Channel 10/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609922:2610200 [6] NCCL INFO Channel 04/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-160-103:2609919:2610201 [3] NCCL INFO Channel 14/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-160-103:2609923:2610203 [7] NCCL INFO Channel 04/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-160-103:2609921:2610206 [5] NCCL INFO Channel 08/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-160-103:2609920:2610205 [4] NCCL INFO Channel 02/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-160-103:2609917:2610202 [1] NCCL INFO Channel 17/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609918:2610198 [2] NCCL INFO Channel 09/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2609916:2610199 [0] NCCL INFO Channel 11/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609922:2610200 [6] NCCL INFO Channel 05/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-160-103:2609919:2610201 [3] NCCL INFO Channel 15/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-160-103:2609923:2610203 [7] NCCL INFO Channel 05/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-160-103:2609921:2610206 [5] NCCL INFO Channel 09/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-160-103:2609917:2610202 [1] NCCL INFO Channel 18/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609920:2610205 [4] NCCL INFO Channel 03/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-160-103:2609918:2610198 [2] NCCL INFO Channel 10/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2609916:2610199 [0] NCCL INFO Channel 12/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609922:2610200 [6] NCCL INFO Channel 06/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-160-103:2609919:2610201 [3] NCCL INFO Channel 16/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-160-103:2609923:2610203 [7] NCCL INFO Channel 06/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-160-103:2609921:2610206 [5] NCCL INFO Channel 10/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-160-103:2609920:2610205 [4] NCCL INFO Channel 04/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-160-103:2609917:2610202 [1] NCCL INFO Channel 19/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609918:2610198 [2] NCCL INFO Channel 11/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2609923:2610203 [7] NCCL INFO Channel 07/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-160-103:2609920:2610205 [4] NCCL INFO Channel 05/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-160-103:2609918:2610198 [2] NCCL INFO Channel 12/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2609916:2610199 [0] NCCL INFO Channel 13/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609923:2610203 [7] NCCL INFO Channel 08/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-160-103:2609920:2610205 [4] NCCL INFO Channel 06/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-160-103:2609918:2610198 [2] NCCL INFO Channel 13/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2609916:2610199 [0] NCCL INFO Channel 14/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609923:2610203 [7] NCCL INFO Channel 09/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-160-103:2609920:2610205 [4] NCCL INFO Channel 07/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-160-103:2609918:2610198 [2] NCCL INFO Channel 14/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2609916:2610199 [0] NCCL INFO Channel 15/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609923:2610203 [7] NCCL INFO Channel 10/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-160-103:2609920:2610205 [4] NCCL INFO Channel 08/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-160-103:2609918:2610198 [2] NCCL INFO Channel 15/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2609916:2610199 [0] NCCL INFO Channel 16/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609923:2610203 [7] NCCL INFO Channel 11/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-160-103:2609920:2610205 [4] NCCL INFO Channel 09/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-160-103:2609918:2610198 [2] NCCL INFO Channel 16/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2609917:2610202 [1] NCCL INFO Channel 20/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609916:2610199 [0] NCCL INFO Channel 17/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609923:2610203 [7] NCCL INFO Channel 12/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-160-103:2609920:2610205 [4] NCCL INFO Channel 10/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-160-103:2609918:2610198 [2] NCCL INFO Channel 17/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2609917:2610202 [1] NCCL INFO Channel 21/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609916:2610199 [0] NCCL INFO Channel 18/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609923:2610203 [7] NCCL INFO Channel 13/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-160-103:2609920:2610205 [4] NCCL INFO Channel 11/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-160-103:2609918:2610198 [2] NCCL INFO Channel 18/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2609916:2610199 [0] NCCL INFO Channel 19/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609917:2610202 [1] NCCL INFO Channel 22/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609923:2610203 [7] NCCL INFO Channel 14/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-160-103:2609920:2610205 [4] NCCL INFO Channel 12/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-160-103:2609918:2610198 [2] NCCL INFO Channel 19/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2609916:2610199 [0] NCCL INFO Channel 20/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609917:2610202 [1] NCCL INFO Channel 23/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609923:2610203 [7] NCCL INFO Channel 15/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-160-103:2609922:2610200 [6] NCCL INFO Channel 07/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-160-103:2609920:2610205 [4] NCCL INFO Channel 13/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-160-103:2609919:2610201 [3] NCCL INFO Channel 17/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-160-103:2609918:2610198 [2] NCCL INFO Channel 20/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2609916:2610199 [0] NCCL INFO Channel 21/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609921:2610206 [5] NCCL INFO Channel 11/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-160-103:2609923:2610203 [7] NCCL INFO Channel 16/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-160-103:2609918:2610198 [2] NCCL INFO Channel 21/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2609922:2610200 [6] NCCL INFO Channel 08/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-160-103:2609920:2610205 [4] NCCL INFO Channel 14/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-160-103:2609923:2610203 [7] NCCL INFO Channel 17/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-160-103:2609918:2610198 [2] NCCL INFO Channel 22/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2609922:2610200 [6] NCCL INFO Channel 09/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-160-103:2609920:2610205 [4] NCCL INFO Channel 15/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-160-103:2609923:2610203 [7] NCCL INFO Channel 18/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-160-103:2609918:2610198 [2] NCCL INFO Channel 23/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2609920:2610205 [4] NCCL INFO Channel 16/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-160-103:2609922:2610200 [6] NCCL INFO Channel 10/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-160-103:2609923:2610203 [7] NCCL INFO Channel 19/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-160-103:2609922:2610200 [6] NCCL INFO Channel 11/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-160-103:2609920:2610205 [4] NCCL INFO Channel 17/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-160-103:2609923:2610203 [7] NCCL INFO Channel 20/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-160-103:2609919:2610201 [3] NCCL INFO Channel 18/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-160-103:2609922:2610200 [6] NCCL INFO Channel 12/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-160-103:2609920:2610205 [4] NCCL INFO Channel 18/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-160-103:2609923:2610203 [7] NCCL INFO Channel 21/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-160-103:2609919:2610201 [3] NCCL INFO Channel 19/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-160-103:2609922:2610200 [6] NCCL INFO Channel 13/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-160-103:2609920:2610205 [4] NCCL INFO Channel 19/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-160-103:2609916:2610199 [0] NCCL INFO Channel 22/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609919:2610201 [3] NCCL INFO Channel 20/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-160-103:2609923:2610203 [7] NCCL INFO Channel 22/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-160-103:2609921:2610206 [5] NCCL INFO Channel 12/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-160-103:2609919:2610201 [3] NCCL INFO Channel 21/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-160-103:2609922:2610200 [6] NCCL INFO Channel 14/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-160-103:2609920:2610205 [4] NCCL INFO Channel 20/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-160-103:2609923:2610203 [7] NCCL INFO Channel 23/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-160-103:2609916:2610199 [0] NCCL INFO Channel 23/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-103:2609921:2610206 [5] NCCL INFO Channel 13/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-160-103:2609919:2610201 [3] NCCL INFO Channel 22/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-160-103:2609920:2610205 [4] NCCL INFO Channel 21/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-160-103:2609922:2610200 [6] NCCL INFO Channel 15/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-160-103:2609921:2610206 [5] NCCL INFO Channel 14/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-160-103:2609919:2610201 [3] NCCL INFO Channel 23/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-160-103:2609922:2610200 [6] NCCL INFO Channel 16/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-160-103:2609920:2610205 [4] NCCL INFO Channel 22/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-160-103:2609921:2610206 [5] NCCL INFO Channel 15/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-160-103:2609922:2610200 [6] NCCL INFO Channel 17/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-160-103:2609920:2610205 [4] NCCL INFO Channel 23/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-160-103:2609921:2610206 [5] NCCL INFO Channel 16/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-160-103:2609922:2610200 [6] NCCL INFO Channel 18/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-160-103:2609921:2610206 [5] NCCL INFO Channel 17/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-160-103:2609922:2610200 [6] NCCL INFO Channel 19/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-160-103:2609922:2610200 [6] NCCL INFO Channel 20/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-160-103:2609921:2610206 [5] NCCL INFO Channel 18/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-160-103:2609922:2610200 [6] NCCL INFO Channel 21/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-160-103:2609921:2610206 [5] NCCL INFO Channel 19/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-160-103:2609922:2610200 [6] NCCL INFO Channel 22/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-160-103:2609921:2610206 [5] NCCL INFO Channel 20/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-160-103:2609922:2610200 [6] NCCL INFO Channel 23/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-160-103:2609921:2610206 [5] NCCL INFO Channel 21/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-160-103:2609921:2610206 [5] NCCL INFO Channel 22/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-160-103:2609921:2610206 [5] NCCL INFO Channel 23/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-160-103:2609918:2610198 [2] NCCL INFO Connected all rings
+ip-26-0-160-103:2609918:2610198 [2] NCCL INFO Connected all trees
+ip-26-0-160-103:2609918:2610198 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-103:2609918:2610198 [2] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-160-103:2609918:2610198 [2] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-160-103:2609917:2610202 [1] NCCL INFO Connected all rings
+ip-26-0-160-103:2609917:2610202 [1] NCCL INFO Connected all trees
+ip-26-0-160-103:2609917:2610202 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-103:2609917:2610202 [1] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-160-103:2609917:2610202 [1] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-160-103:2609916:2610199 [0] NCCL INFO Connected all rings
+ip-26-0-160-103:2609916:2610199 [0] NCCL INFO Connected all trees
+ip-26-0-160-103:2609916:2610199 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-103:2609916:2610199 [0] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-160-103:2609916:2610199 [0] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-160-103:2609919:2610201 [3] NCCL INFO Connected all rings
+ip-26-0-160-103:2609919:2610201 [3] NCCL INFO Connected all trees
+ip-26-0-160-103:2609919:2610201 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-103:2609919:2610201 [3] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-160-103:2609919:2610201 [3] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-160-103:2609916:2610199 [0] NCCL INFO comm 0xb9b4270 rank 0 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0x2c3909f8817a8fb9 - Init COMPLETE
+ip-26-0-160-103:2609917:2610202 [1] NCCL INFO comm 0xafa7870 rank 1 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0x2c3909f8817a8fb9 - Init COMPLETE
+ip-26-0-160-103:2609916:2610223 [0] NCCL INFO Using network AWS Libfabric
+ip-26-0-160-103:2609916:2610223 [0] NCCL INFO comm 0xb9c7f00 rank 0 nranks 1 cudaDev 0 nvmlDev 0 busId 53000 commId 0x8c109fe6848d4bb9 - Init START
+NCCL version 2.18.5+cuda12.2
+ip-26-0-160-103:2609916:2610223 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /opt/aws-ofi-nccl/share/aws-ofi-nccl/xml/p5.48xl-topo.xml
+ip-26-0-160-103:2609917:2610225 [1] NCCL INFO Using network AWS Libfabric
+ip-26-0-160-103:2609917:2610225 [1] NCCL INFO comm 0xafbb190 rank 0 nranks 1 cudaDev 1 nvmlDev 1 busId 64000 commId 0x5addba7505c507dd - Init START
+ip-26-0-160-103:2609917:2610225 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /opt/aws-ofi-nccl/share/aws-ofi-nccl/xml/p5.48xl-topo.xml
+ip-26-0-160-103:2609922:2610200 [6] NCCL INFO Connected all rings
+ip-26-0-160-103:2609922:2610200 [6] NCCL INFO Connected all trees
+ip-26-0-160-103:2609922:2610200 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-103:2609922:2610200 [6] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-160-103:2609922:2610200 [6] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-160-103:2609919:2610201 [3] NCCL INFO comm 0xa51d7e0 rank 1 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0xaa4d6a470f5df587 - Init COMPLETE
+ip-26-0-160-103:2609918:2610198 [2] NCCL INFO comm 0xc175cb0 rank 0 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0xaa4d6a470f5df587 - Init COMPLETE
+ip-26-0-160-103:2609918:2610228 [2] NCCL INFO Using network AWS Libfabric
+ip-26-0-160-103:2609918:2610228 [2] NCCL INFO comm 0xc189dd0 rank 0 nranks 1 cudaDev 2 nvmlDev 2 busId 75000 commId 0x2ec0434ac0ed87c4 - Init START
+ip-26-0-160-103:2609918:2610228 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /opt/aws-ofi-nccl/share/aws-ofi-nccl/xml/p5.48xl-topo.xml
+NCCL version 2.18.5+cuda12.2
+ip-26-0-160-103:2609919:2610230 [3] NCCL INFO Using network AWS Libfabric
+ip-26-0-160-103:2609919:2610230 [3] NCCL INFO comm 0xa530ec0 rank 0 nranks 1 cudaDev 3 nvmlDev 3 busId 86000 commId 0x2e436d5a15a93408 - Init START
+ip-26-0-160-103:2609919:2610230 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /opt/aws-ofi-nccl/share/aws-ofi-nccl/xml/p5.48xl-topo.xml
+ip-26-0-160-103:2609920:2610205 [4] NCCL INFO Connected all rings
+ip-26-0-160-103:2609920:2610205 [4] NCCL INFO Connected all trees
+ip-26-0-160-103:2609920:2610205 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-103:2609920:2610205 [4] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-160-103:2609920:2610205 [4] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-160-103:2609923:2610203 [7] NCCL INFO Connected all rings
+ip-26-0-160-103:2609923:2610203 [7] NCCL INFO Connected all trees
+ip-26-0-160-103:2609923:2610203 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-103:2609923:2610203 [7] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-160-103:2609923:2610203 [7] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-160-103:2609923:2610203 [7] NCCL INFO comm 0xc26b3e0 rank 1 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0x6bb75a1aa2d88ceb - Init COMPLETE
+ip-26-0-160-103:2609922:2610200 [6] NCCL INFO comm 0xb69a470 rank 0 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0x6bb75a1aa2d88ceb - Init COMPLETE
+NCCL version 2.18.5+cuda12.2
+ip-26-0-160-103:2609921:2610206 [5] NCCL INFO Connected all rings
+ip-26-0-160-103:2609921:2610206 [5] NCCL INFO Connected all trees
+ip-26-0-160-103:2609921:2610206 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-103:2609921:2610206 [5] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-160-103:2609921:2610206 [5] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-160-103:2609923:2610234 [7] NCCL INFO Using network AWS Libfabric
+ip-26-0-160-103:2609923:2610234 [7] NCCL INFO comm 0xc27eaf0 rank 0 nranks 1 cudaDev 7 nvmlDev 7 busId ca000 commId 0xde4709c46e1dc844 - Init START
+ip-26-0-160-103:2609923:2610234 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /opt/aws-ofi-nccl/share/aws-ofi-nccl/xml/p5.48xl-topo.xml
+ip-26-0-160-103:2609922:2610237 [6] NCCL INFO Using network AWS Libfabric
+ip-26-0-160-103:2609922:2610237 [6] NCCL INFO comm 0xb6ae0d0 rank 0 nranks 1 cudaDev 6 nvmlDev 6 busId b9000 commId 0x33f609ebe74c0f2d - Init START
+ip-26-0-160-103:2609922:2610237 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /opt/aws-ofi-nccl/share/aws-ofi-nccl/xml/p5.48xl-topo.xml
+ip-26-0-160-103:2609921:2610206 [5] NCCL INFO comm 0xadb73f0 rank 1 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0x8ad574a31761b571 - Init COMPLETE
+ip-26-0-160-103:2609920:2610205 [4] NCCL INFO comm 0xb150330 rank 0 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0x8ad574a31761b571 - Init COMPLETE
+NCCL version 2.18.5+cuda12.2
+ip-26-0-160-103:2609921:2610239 [5] NCCL INFO Using network AWS Libfabric
+ip-26-0-160-103:2609921:2610239 [5] NCCL INFO comm 0xadcb040 rank 0 nranks 1 cudaDev 5 nvmlDev 5 busId a8000 commId 0x80c61819814e44a6 - Init START
+ip-26-0-160-103:2609921:2610239 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /opt/aws-ofi-nccl/share/aws-ofi-nccl/xml/p5.48xl-topo.xml
+ip-26-0-160-103:2609920:2610241 [4] NCCL INFO Using network AWS Libfabric
+ip-26-0-160-103:2609920:2610241 [4] NCCL INFO comm 0xb164240 rank 0 nranks 1 cudaDev 4 nvmlDev 4 busId 97000 commId 0x31e3be134e0000d5 - Init START
+ip-26-0-160-103:2609920:2610241 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /opt/aws-ofi-nccl/share/aws-ofi-nccl/xml/p5.48xl-topo.xml
+ip-26-0-160-103:2609917:2610225 [1] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609917:2610225 [1] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609917:2610225 [1] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609917:2610225 [1] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609917:2610225 [1] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609917:2610225 [1] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609917:2610225 [1] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609917:2610225 [1] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609917:2610225 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-160-103:2609917:2610225 [1] NCCL INFO Channel 00/32 :    0
+ip-26-0-160-103:2609917:2610225 [1] NCCL INFO Channel 01/32 :    0
+ip-26-0-160-103:2609917:2610225 [1] NCCL INFO Channel 02/32 :    0
+ip-26-0-160-103:2609917:2610225 [1] NCCL INFO Channel 03/32 :    0
+ip-26-0-160-103:2609917:2610225 [1] NCCL INFO Channel 04/32 :    0
+ip-26-0-160-103:2609917:2610225 [1] NCCL INFO Channel 05/32 :    0
+ip-26-0-160-103:2609917:2610225 [1] NCCL INFO Channel 06/32 :    0
+ip-26-0-160-103:2609917:2610225 [1] NCCL INFO Channel 07/32 :    0
+ip-26-0-160-103:2609917:2610225 [1] NCCL INFO Channel 08/32 :    0
+ip-26-0-160-103:2609917:2610225 [1] NCCL INFO Channel 09/32 :    0
+ip-26-0-160-103:2609917:2610225 [1] NCCL INFO Channel 10/32 :    0
+ip-26-0-160-103:2609917:2610225 [1] NCCL INFO Channel 11/32 :    0
+ip-26-0-160-103:2609917:2610225 [1] NCCL INFO Channel 12/32 :    0
+ip-26-0-160-103:2609917:2610225 [1] NCCL INFO Channel 13/32 :    0
+ip-26-0-160-103:2609917:2610225 [1] NCCL INFO Channel 14/32 :    0
+ip-26-0-160-103:2609917:2610225 [1] NCCL INFO Channel 15/32 :    0
+ip-26-0-160-103:2609917:2610225 [1] NCCL INFO Channel 16/32 :    0
+ip-26-0-160-103:2609917:2610225 [1] NCCL INFO Channel 17/32 :    0
+ip-26-0-160-103:2609917:2610225 [1] NCCL INFO Channel 18/32 :    0
+ip-26-0-160-103:2609917:2610225 [1] NCCL INFO Channel 19/32 :    0
+ip-26-0-160-103:2609917:2610225 [1] NCCL INFO Channel 20/32 :    0
+ip-26-0-160-103:2609917:2610225 [1] NCCL INFO Channel 21/32 :    0
+ip-26-0-160-103:2609917:2610225 [1] NCCL INFO Channel 22/32 :    0
+ip-26-0-160-103:2609917:2610225 [1] NCCL INFO Channel 23/32 :    0
+ip-26-0-160-103:2609917:2610225 [1] NCCL INFO Channel 24/32 :    0
+ip-26-0-160-103:2609917:2610225 [1] NCCL INFO Channel 25/32 :    0
+ip-26-0-160-103:2609917:2610225 [1] NCCL INFO Channel 26/32 :    0
+ip-26-0-160-103:2609917:2610225 [1] NCCL INFO Channel 27/32 :    0
+ip-26-0-160-103:2609917:2610225 [1] NCCL INFO Channel 28/32 :    0
+ip-26-0-160-103:2609917:2610225 [1] NCCL INFO Channel 29/32 :    0
+ip-26-0-160-103:2609917:2610225 [1] NCCL INFO Channel 30/32 :    0
+ip-26-0-160-103:2609917:2610225 [1] NCCL INFO Channel 31/32 :    0
+ip-26-0-160-103:2609917:2610225 [1] NCCL INFO Trees [0] -1/-1/-1->0->-1 [1] -1/-1/-1->0->-1 [2] -1/-1/-1->0->-1 [3] -1/-1/-1->0->-1 [4] -1/-1/-1->0->-1 [5] -1/-1/-1->0->-1 [6] -1/-1/-1->0->-1 [7] -1/-1/-1->0->-1 [8] -1/-1/-1->0->-1 [9] -1/-1/-1->0->-1 [10] -1/-1/-1->0->-1 [11] -1/-1/-1->0->-1 [12] -1/-1/-1->0->-1 [13] -1/-1/-1->0->-1 [14] -1/-1/-1->0->-1 [15] -1/-1/-1->0->-1 [16] -1/-1/-1->0->-1 [17] -1/-1/-1->0->-1 [18] -1/-1/-1->0->-1 [19] -1/-1/-1->0->-1 [20] -1/-1/-1->0->-1 [21] -1/-1/-1->0->-1 [22] -1/-1/-1->0->-1 [23] -1/-1/-1->0->-1 [24] -1/-1/-1->0->-1 [25] -1/-1/-1->0->-1 [26] -1/-1/-1->0->-1 [27] -1/-1/-1->0->-1 [28] -1/-1/-1->0->-1 [29] -1/-1/-1->0->-1 [30] -1/-1/-1->0->-1 [31] -1/-1/-1->0->-1
+ip-26-0-160-103:2609917:2610225 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-103:2609917:2610225 [1] NCCL INFO Connected all rings
+ip-26-0-160-103:2609917:2610225 [1] NCCL INFO Connected all trees
+ip-26-0-160-103:2609917:2610225 [1] NCCL INFO 32 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-160-103:2609916:2610223 [0] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609916:2610223 [0] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609916:2610223 [0] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609916:2610223 [0] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609916:2610223 [0] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609916:2610223 [0] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609916:2610223 [0] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609916:2610223 [0] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609917:2610225 [1] NCCL INFO comm 0xafbb190 rank 0 nranks 1 cudaDev 1 nvmlDev 1 busId 64000 commId 0x5addba7505c507dd - Init COMPLETE
+ip-26-0-160-103:2609916:2610223 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-160-103:2609916:2610223 [0] NCCL INFO Channel 00/32 :    0
+ip-26-0-160-103:2609916:2610223 [0] NCCL INFO Channel 01/32 :    0
+ip-26-0-160-103:2609916:2610223 [0] NCCL INFO Channel 02/32 :    0
+ip-26-0-160-103:2609916:2610223 [0] NCCL INFO Channel 03/32 :    0
+ip-26-0-160-103:2609916:2610223 [0] NCCL INFO Channel 04/32 :    0
+ip-26-0-160-103:2609916:2610223 [0] NCCL INFO Channel 05/32 :    0
+ip-26-0-160-103:2609916:2610223 [0] NCCL INFO Channel 06/32 :    0
+ip-26-0-160-103:2609916:2610223 [0] NCCL INFO Channel 07/32 :    0
+ip-26-0-160-103:2609916:2610223 [0] NCCL INFO Channel 08/32 :    0
+ip-26-0-160-103:2609916:2610223 [0] NCCL INFO Channel 09/32 :    0
+ip-26-0-160-103:2609916:2610223 [0] NCCL INFO Channel 10/32 :    0
+ip-26-0-160-103:2609916:2610223 [0] NCCL INFO Channel 11/32 :    0
+ip-26-0-160-103:2609916:2610223 [0] NCCL INFO Channel 12/32 :    0
+ip-26-0-160-103:2609916:2610223 [0] NCCL INFO Channel 13/32 :    0
+ip-26-0-160-103:2609916:2610223 [0] NCCL INFO Channel 14/32 :    0
+ip-26-0-160-103:2609916:2610223 [0] NCCL INFO Channel 15/32 :    0
+ip-26-0-160-103:2609916:2610223 [0] NCCL INFO Channel 16/32 :    0
+ip-26-0-160-103:2609916:2610223 [0] NCCL INFO Channel 17/32 :    0
+ip-26-0-160-103:2609916:2610223 [0] NCCL INFO Channel 18/32 :    0
+ip-26-0-160-103:2609916:2610223 [0] NCCL INFO Channel 19/32 :    0
+ip-26-0-160-103:2609916:2610223 [0] NCCL INFO Channel 20/32 :    0
+ip-26-0-160-103:2609916:2610223 [0] NCCL INFO Channel 21/32 :    0
+ip-26-0-160-103:2609916:2610223 [0] NCCL INFO Channel 22/32 :    0
+ip-26-0-160-103:2609916:2610223 [0] NCCL INFO Channel 23/32 :    0
+ip-26-0-160-103:2609916:2610223 [0] NCCL INFO Channel 24/32 :    0
+ip-26-0-160-103:2609916:2610223 [0] NCCL INFO Channel 25/32 :    0
+ip-26-0-160-103:2609916:2610223 [0] NCCL INFO Channel 26/32 :    0
+ip-26-0-160-103:2609916:2610223 [0] NCCL INFO Channel 27/32 :    0
+ip-26-0-160-103:2609916:2610223 [0] NCCL INFO Channel 28/32 :    0
+ip-26-0-160-103:2609916:2610223 [0] NCCL INFO Channel 29/32 :    0
+ip-26-0-160-103:2609916:2610223 [0] NCCL INFO Channel 30/32 :    0
+ip-26-0-160-103:2609916:2610223 [0] NCCL INFO Channel 31/32 :    0
+ip-26-0-160-103:2609916:2610223 [0] NCCL INFO Trees [0] -1/-1/-1->0->-1 [1] -1/-1/-1->0->-1 [2] -1/-1/-1->0->-1 [3] -1/-1/-1->0->-1 [4] -1/-1/-1->0->-1 [5] -1/-1/-1->0->-1 [6] -1/-1/-1->0->-1 [7] -1/-1/-1->0->-1 [8] -1/-1/-1->0->-1 [9] -1/-1/-1->0->-1 [10] -1/-1/-1->0->-1 [11] -1/-1/-1->0->-1 [12] -1/-1/-1->0->-1 [13] -1/-1/-1->0->-1 [14] -1/-1/-1->0->-1 [15] -1/-1/-1->0->-1 [16] -1/-1/-1->0->-1 [17] -1/-1/-1->0->-1 [18] -1/-1/-1->0->-1 [19] -1/-1/-1->0->-1 [20] -1/-1/-1->0->-1 [21] -1/-1/-1->0->-1 [22] -1/-1/-1->0->-1 [23] -1/-1/-1->0->-1 [24] -1/-1/-1->0->-1 [25] -1/-1/-1->0->-1 [26] -1/-1/-1->0->-1 [27] -1/-1/-1->0->-1 [28] -1/-1/-1->0->-1 [29] -1/-1/-1->0->-1 [30] -1/-1/-1->0->-1 [31] -1/-1/-1->0->-1
+ip-26-0-160-103:2609916:2610223 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-103:2609916:2610223 [0] NCCL INFO Connected all rings
+ip-26-0-160-103:2609916:2610223 [0] NCCL INFO Connected all trees
+ip-26-0-160-103:2609916:2610223 [0] NCCL INFO 32 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-160-103:2609916:2610223 [0] NCCL INFO comm 0xb9c7f00 rank 0 nranks 1 cudaDev 0 nvmlDev 0 busId 53000 commId 0x8c109fe6848d4bb9 - Init COMPLETE
+12/13/2024 14:17:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]: Total number of parameters: 1.04G (1984.26MiB)
+12/13/2024 14:17:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]: Local number of parameters: 520M (992.13MiB)
+12/13/2024 14:17:24 [INFO|DP=0|PP=0|TP=1|ip-26-0-160-103]: Local number of parameters: 520M (992.13MiB)
+12/13/2024 14:17:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]: [After model building] Memory usage: 992.15MiB. Peak allocated: 5632.00MiB Peak reserved: 18946.00MiB
+12/13/2024 14:17:24 [INFO|DP=0|PP=0|TP=1|ip-26-0-160-103]: [After model building] Memory usage: 992.15MiB. Peak allocated: 5632.00MiB Peak reserved: 19970.00MiB
+12/13/2024 14:17:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]: No checkpoint path provided.
+12/13/2024 14:17:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]: Parametrizing model parameters using StandardParametrizator
+ip-26-0-160-103:2609918:2610228 [2] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609918:2610228 [2] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609918:2610228 [2] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609918:2610228 [2] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609918:2610228 [2] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609918:2610228 [2] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609918:2610228 [2] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609918:2610228 [2] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609918:2610228 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-160-103:2609918:2610228 [2] NCCL INFO Channel 00/32 :    0
+ip-26-0-160-103:2609918:2610228 [2] NCCL INFO Channel 01/32 :    0
+ip-26-0-160-103:2609918:2610228 [2] NCCL INFO Channel 02/32 :    0
+ip-26-0-160-103:2609918:2610228 [2] NCCL INFO Channel 03/32 :    0
+ip-26-0-160-103:2609918:2610228 [2] NCCL INFO Channel 04/32 :    0
+ip-26-0-160-103:2609918:2610228 [2] NCCL INFO Channel 05/32 :    0
+ip-26-0-160-103:2609918:2610228 [2] NCCL INFO Channel 06/32 :    0
+ip-26-0-160-103:2609918:2610228 [2] NCCL INFO Channel 07/32 :    0
+ip-26-0-160-103:2609918:2610228 [2] NCCL INFO Channel 08/32 :    0
+ip-26-0-160-103:2609918:2610228 [2] NCCL INFO Channel 09/32 :    0
+ip-26-0-160-103:2609918:2610228 [2] NCCL INFO Channel 10/32 :    0
+ip-26-0-160-103:2609918:2610228 [2] NCCL INFO Channel 11/32 :    0
+ip-26-0-160-103:2609918:2610228 [2] NCCL INFO Channel 12/32 :    0
+ip-26-0-160-103:2609918:2610228 [2] NCCL INFO Channel 13/32 :    0
+ip-26-0-160-103:2609918:2610228 [2] NCCL INFO Channel 14/32 :    0
+ip-26-0-160-103:2609918:2610228 [2] NCCL INFO Channel 15/32 :    0
+ip-26-0-160-103:2609918:2610228 [2] NCCL INFO Channel 16/32 :    0
+ip-26-0-160-103:2609918:2610228 [2] NCCL INFO Channel 17/32 :    0
+ip-26-0-160-103:2609918:2610228 [2] NCCL INFO Channel 18/32 :    0
+ip-26-0-160-103:2609918:2610228 [2] NCCL INFO Channel 19/32 :    0
+ip-26-0-160-103:2609918:2610228 [2] NCCL INFO Channel 20/32 :    0
+ip-26-0-160-103:2609918:2610228 [2] NCCL INFO Channel 21/32 :    0
+ip-26-0-160-103:2609918:2610228 [2] NCCL INFO Channel 22/32 :    0
+ip-26-0-160-103:2609918:2610228 [2] NCCL INFO Channel 23/32 :    0
+ip-26-0-160-103:2609918:2610228 [2] NCCL INFO Channel 24/32 :    0
+ip-26-0-160-103:2609918:2610228 [2] NCCL INFO Channel 25/32 :    0
+ip-26-0-160-103:2609918:2610228 [2] NCCL INFO Channel 26/32 :    0
+ip-26-0-160-103:2609918:2610228 [2] NCCL INFO Channel 27/32 :    0
+ip-26-0-160-103:2609918:2610228 [2] NCCL INFO Channel 28/32 :    0
+ip-26-0-160-103:2609918:2610228 [2] NCCL INFO Channel 29/32 :    0
+ip-26-0-160-103:2609918:2610228 [2] NCCL INFO Channel 30/32 :    0
+ip-26-0-160-103:2609918:2610228 [2] NCCL INFO Channel 31/32 :    0
+ip-26-0-160-103:2609918:2610228 [2] NCCL INFO Trees [0] -1/-1/-1->0->-1 [1] -1/-1/-1->0->-1 [2] -1/-1/-1->0->-1 [3] -1/-1/-1->0->-1 [4] -1/-1/-1->0->-1 [5] -1/-1/-1->0->-1 [6] -1/-1/-1->0->-1 [7] -1/-1/-1->0->-1 [8] -1/-1/-1->0->-1 [9] -1/-1/-1->0->-1 [10] -1/-1/-1->0->-1 [11] -1/-1/-1->0->-1 [12] -1/-1/-1->0->-1 [13] -1/-1/-1->0->-1 [14] -1/-1/-1->0->-1 [15] -1/-1/-1->0->-1 [16] -1/-1/-1->0->-1 [17] -1/-1/-1->0->-1 [18] -1/-1/-1->0->-1 [19] -1/-1/-1->0->-1 [20] -1/-1/-1->0->-1 [21] -1/-1/-1->0->-1 [22] -1/-1/-1->0->-1 [23] -1/-1/-1->0->-1 [24] -1/-1/-1->0->-1 [25] -1/-1/-1->0->-1 [26] -1/-1/-1->0->-1 [27] -1/-1/-1->0->-1 [28] -1/-1/-1->0->-1 [29] -1/-1/-1->0->-1 [30] -1/-1/-1->0->-1 [31] -1/-1/-1->0->-1
+ip-26-0-160-103:2609918:2610228 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-103:2609917:2610249 [1] NCCL INFO Using network AWS Libfabric
+ip-26-0-160-103:2609916:2610247 [0] NCCL INFO Using network AWS Libfabric
+ip-26-0-160-103:2609918:2610228 [2] NCCL INFO Connected all rings
+ip-26-0-160-103:2609918:2610228 [2] NCCL INFO Connected all trees
+ip-26-0-160-103:2609918:2610228 [2] NCCL INFO 32 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-160-103:2609919:2610230 [3] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609919:2610230 [3] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609919:2610230 [3] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609919:2610230 [3] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609919:2610230 [3] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609919:2610230 [3] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609919:2610230 [3] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609919:2610230 [3] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609919:2610230 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-160-103:2609919:2610230 [3] NCCL INFO Channel 00/32 :    0
+ip-26-0-160-103:2609919:2610230 [3] NCCL INFO Channel 01/32 :    0
+ip-26-0-160-103:2609919:2610230 [3] NCCL INFO Channel 02/32 :    0
+ip-26-0-160-103:2609919:2610230 [3] NCCL INFO Channel 03/32 :    0
+ip-26-0-160-103:2609919:2610230 [3] NCCL INFO Channel 04/32 :    0
+ip-26-0-160-103:2609919:2610230 [3] NCCL INFO Channel 05/32 :    0
+ip-26-0-160-103:2609919:2610230 [3] NCCL INFO Channel 06/32 :    0
+ip-26-0-160-103:2609919:2610230 [3] NCCL INFO Channel 07/32 :    0
+ip-26-0-160-103:2609919:2610230 [3] NCCL INFO Channel 08/32 :    0
+ip-26-0-160-103:2609919:2610230 [3] NCCL INFO Channel 09/32 :    0
+ip-26-0-160-103:2609919:2610230 [3] NCCL INFO Channel 10/32 :    0
+ip-26-0-160-103:2609919:2610230 [3] NCCL INFO Channel 11/32 :    0
+ip-26-0-160-103:2609919:2610230 [3] NCCL INFO Channel 12/32 :    0
+ip-26-0-160-103:2609919:2610230 [3] NCCL INFO Channel 13/32 :    0
+ip-26-0-160-103:2609919:2610230 [3] NCCL INFO Channel 14/32 :    0
+ip-26-0-160-103:2609919:2610230 [3] NCCL INFO Channel 15/32 :    0
+ip-26-0-160-103:2609919:2610230 [3] NCCL INFO Channel 16/32 :    0
+ip-26-0-160-103:2609919:2610230 [3] NCCL INFO Channel 17/32 :    0
+ip-26-0-160-103:2609919:2610230 [3] NCCL INFO Channel 18/32 :    0
+ip-26-0-160-103:2609919:2610230 [3] NCCL INFO Channel 19/32 :    0
+ip-26-0-160-103:2609919:2610230 [3] NCCL INFO Channel 20/32 :    0
+ip-26-0-160-103:2609919:2610230 [3] NCCL INFO Channel 21/32 :    0
+ip-26-0-160-103:2609919:2610230 [3] NCCL INFO Channel 22/32 :    0
+ip-26-0-160-103:2609919:2610230 [3] NCCL INFO Channel 23/32 :    0
+ip-26-0-160-103:2609919:2610230 [3] NCCL INFO Channel 24/32 :    0
+ip-26-0-160-103:2609919:2610230 [3] NCCL INFO Channel 25/32 :    0
+ip-26-0-160-103:2609919:2610230 [3] NCCL INFO Channel 26/32 :    0
+ip-26-0-160-103:2609919:2610230 [3] NCCL INFO Channel 27/32 :    0
+ip-26-0-160-103:2609919:2610230 [3] NCCL INFO Channel 28/32 :    0
+ip-26-0-160-103:2609919:2610230 [3] NCCL INFO Channel 29/32 :    0
+ip-26-0-160-103:2609919:2610230 [3] NCCL INFO Channel 30/32 :    0
+ip-26-0-160-103:2609919:2610230 [3] NCCL INFO Channel 31/32 :    0
+ip-26-0-160-103:2609919:2610230 [3] NCCL INFO Trees [0] -1/-1/-1->0->-1 [1] -1/-1/-1->0->-1 [2] -1/-1/-1->0->-1 [3] -1/-1/-1->0->-1 [4] -1/-1/-1->0->-1 [5] -1/-1/-1->0->-1 [6] -1/-1/-1->0->-1 [7] -1/-1/-1->0->-1 [8] -1/-1/-1->0->-1 [9] -1/-1/-1->0->-1 [10] -1/-1/-1->0->-1 [11] -1/-1/-1->0->-1 [12] -1/-1/-1->0->-1 [13] -1/-1/-1->0->-1 [14] -1/-1/-1->0->-1 [15] -1/-1/-1->0->-1 [16] -1/-1/-1->0->-1 [17] -1/-1/-1->0->-1 [18] -1/-1/-1->0->-1 [19] -1/-1/-1->0->-1 [20] -1/-1/-1->0->-1 [21] -1/-1/-1->0->-1 [22] -1/-1/-1->0->-1 [23] -1/-1/-1->0->-1 [24] -1/-1/-1->0->-1 [25] -1/-1/-1->0->-1 [26] -1/-1/-1->0->-1 [27] -1/-1/-1->0->-1 [28] -1/-1/-1->0->-1 [29] -1/-1/-1->0->-1 [30] -1/-1/-1->0->-1 [31] -1/-1/-1->0->-1
+ip-26-0-160-103:2609919:2610230 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-103:2609923:2610234 [7] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609923:2610234 [7] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609919:2610230 [3] NCCL INFO Connected all rings
+ip-26-0-160-103:2609919:2610230 [3] NCCL INFO Connected all trees
+ip-26-0-160-103:2609919:2610230 [3] NCCL INFO 32 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-160-103:2609923:2610234 [7] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609923:2610234 [7] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609923:2610234 [7] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609923:2610234 [7] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609923:2610234 [7] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609923:2610234 [7] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609923:2610234 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-160-103:2609923:2610234 [7] NCCL INFO Channel 00/32 :    0
+ip-26-0-160-103:2609923:2610234 [7] NCCL INFO Channel 01/32 :    0
+ip-26-0-160-103:2609923:2610234 [7] NCCL INFO Channel 02/32 :    0
+ip-26-0-160-103:2609923:2610234 [7] NCCL INFO Channel 03/32 :    0
+ip-26-0-160-103:2609923:2610234 [7] NCCL INFO Channel 04/32 :    0
+ip-26-0-160-103:2609923:2610234 [7] NCCL INFO Channel 05/32 :    0
+ip-26-0-160-103:2609923:2610234 [7] NCCL INFO Channel 06/32 :    0
+ip-26-0-160-103:2609923:2610234 [7] NCCL INFO Channel 07/32 :    0
+ip-26-0-160-103:2609923:2610234 [7] NCCL INFO Channel 08/32 :    0
+ip-26-0-160-103:2609923:2610234 [7] NCCL INFO Channel 09/32 :    0
+ip-26-0-160-103:2609923:2610234 [7] NCCL INFO Channel 10/32 :    0
+ip-26-0-160-103:2609923:2610234 [7] NCCL INFO Channel 11/32 :    0
+ip-26-0-160-103:2609923:2610234 [7] NCCL INFO Channel 12/32 :    0
+ip-26-0-160-103:2609923:2610234 [7] NCCL INFO Channel 13/32 :    0
+ip-26-0-160-103:2609923:2610234 [7] NCCL INFO Channel 14/32 :    0
+ip-26-0-160-103:2609923:2610234 [7] NCCL INFO Channel 15/32 :    0
+ip-26-0-160-103:2609923:2610234 [7] NCCL INFO Channel 16/32 :    0
+ip-26-0-160-103:2609923:2610234 [7] NCCL INFO Channel 17/32 :    0
+ip-26-0-160-103:2609923:2610234 [7] NCCL INFO Channel 18/32 :    0
+ip-26-0-160-103:2609923:2610234 [7] NCCL INFO Channel 19/32 :    0
+ip-26-0-160-103:2609923:2610234 [7] NCCL INFO Channel 20/32 :    0
+ip-26-0-160-103:2609923:2610234 [7] NCCL INFO Channel 21/32 :    0
+ip-26-0-160-103:2609923:2610234 [7] NCCL INFO Channel 22/32 :    0
+ip-26-0-160-103:2609923:2610234 [7] NCCL INFO Channel 23/32 :    0
+ip-26-0-160-103:2609923:2610234 [7] NCCL INFO Channel 24/32 :    0
+ip-26-0-160-103:2609923:2610234 [7] NCCL INFO Channel 25/32 :    0
+ip-26-0-160-103:2609923:2610234 [7] NCCL INFO Channel 26/32 :    0
+ip-26-0-160-103:2609923:2610234 [7] NCCL INFO Channel 27/32 :    0
+ip-26-0-160-103:2609923:2610234 [7] NCCL INFO Channel 28/32 :    0
+ip-26-0-160-103:2609923:2610234 [7] NCCL INFO Channel 29/32 :    0
+ip-26-0-160-103:2609923:2610234 [7] NCCL INFO Channel 30/32 :    0
+ip-26-0-160-103:2609923:2610234 [7] NCCL INFO Channel 31/32 :    0
+ip-26-0-160-103:2609923:2610234 [7] NCCL INFO Trees [0] -1/-1/-1->0->-1 [1] -1/-1/-1->0->-1 [2] -1/-1/-1->0->-1 [3] -1/-1/-1->0->-1 [4] -1/-1/-1->0->-1 [5] -1/-1/-1->0->-1 [6] -1/-1/-1->0->-1 [7] -1/-1/-1->0->-1 [8] -1/-1/-1->0->-1 [9] -1/-1/-1->0->-1 [10] -1/-1/-1->0->-1 [11] -1/-1/-1->0->-1 [12] -1/-1/-1->0->-1 [13] -1/-1/-1->0->-1 [14] -1/-1/-1->0->-1 [15] -1/-1/-1->0->-1 [16] -1/-1/-1->0->-1 [17] -1/-1/-1->0->-1 [18] -1/-1/-1->0->-1 [19] -1/-1/-1->0->-1 [20] -1/-1/-1->0->-1 [21] -1/-1/-1->0->-1 [22] -1/-1/-1->0->-1 [23] -1/-1/-1->0->-1 [24] -1/-1/-1->0->-1 [25] -1/-1/-1->0->-1 [26] -1/-1/-1->0->-1 [27] -1/-1/-1->0->-1 [28] -1/-1/-1->0->-1 [29] -1/-1/-1->0->-1 [30] -1/-1/-1->0->-1 [31] -1/-1/-1->0->-1
+ip-26-0-160-103:2609923:2610234 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-103:2609923:2610234 [7] NCCL INFO Connected all rings
+ip-26-0-160-103:2609923:2610234 [7] NCCL INFO Connected all trees
+ip-26-0-160-103:2609923:2610234 [7] NCCL INFO 32 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-160-103:2609918:2610228 [2] NCCL INFO comm 0xc189dd0 rank 0 nranks 1 cudaDev 2 nvmlDev 2 busId 75000 commId 0x2ec0434ac0ed87c4 - Init COMPLETE
+ip-26-0-160-103:2609919:2610230 [3] NCCL INFO comm 0xa530ec0 rank 0 nranks 1 cudaDev 3 nvmlDev 3 busId 86000 commId 0x2e436d5a15a93408 - Init COMPLETE
+ip-26-0-160-103:2609918:2610256 [2] NCCL INFO Using network AWS Libfabric
+ip-26-0-160-103:2609919:2610257 [3] NCCL INFO Using network AWS Libfabric
+ip-26-0-160-103:2609923:2610234 [7] NCCL INFO comm 0xc27eaf0 rank 0 nranks 1 cudaDev 7 nvmlDev 7 busId ca000 commId 0xde4709c46e1dc844 - Init COMPLETE
+ip-26-0-160-103:2609922:2610237 [6] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609922:2610237 [6] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609922:2610237 [6] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609922:2610237 [6] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609922:2610237 [6] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609922:2610237 [6] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609922:2610237 [6] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609922:2610237 [6] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609922:2610237 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-160-103:2609922:2610237 [6] NCCL INFO Channel 00/32 :    0
+ip-26-0-160-103:2609922:2610237 [6] NCCL INFO Channel 01/32 :    0
+ip-26-0-160-103:2609922:2610237 [6] NCCL INFO Channel 02/32 :    0
+ip-26-0-160-103:2609922:2610237 [6] NCCL INFO Channel 03/32 :    0
+ip-26-0-160-103:2609922:2610237 [6] NCCL INFO Channel 04/32 :    0
+ip-26-0-160-103:2609922:2610237 [6] NCCL INFO Channel 05/32 :    0
+ip-26-0-160-103:2609922:2610237 [6] NCCL INFO Channel 06/32 :    0
+ip-26-0-160-103:2609922:2610237 [6] NCCL INFO Channel 07/32 :    0
+ip-26-0-160-103:2609922:2610237 [6] NCCL INFO Channel 08/32 :    0
+ip-26-0-160-103:2609922:2610237 [6] NCCL INFO Channel 09/32 :    0
+ip-26-0-160-103:2609922:2610237 [6] NCCL INFO Channel 10/32 :    0
+ip-26-0-160-103:2609922:2610237 [6] NCCL INFO Channel 11/32 :    0
+ip-26-0-160-103:2609922:2610237 [6] NCCL INFO Channel 12/32 :    0
+ip-26-0-160-103:2609922:2610237 [6] NCCL INFO Channel 13/32 :    0
+ip-26-0-160-103:2609922:2610237 [6] NCCL INFO Channel 14/32 :    0
+ip-26-0-160-103:2609922:2610237 [6] NCCL INFO Channel 15/32 :    0
+ip-26-0-160-103:2609922:2610237 [6] NCCL INFO Channel 16/32 :    0
+ip-26-0-160-103:2609922:2610237 [6] NCCL INFO Channel 17/32 :    0
+ip-26-0-160-103:2609922:2610237 [6] NCCL INFO Channel 18/32 :    0
+ip-26-0-160-103:2609922:2610237 [6] NCCL INFO Channel 19/32 :    0
+ip-26-0-160-103:2609922:2610237 [6] NCCL INFO Channel 20/32 :    0
+ip-26-0-160-103:2609922:2610237 [6] NCCL INFO Channel 21/32 :    0
+ip-26-0-160-103:2609922:2610237 [6] NCCL INFO Channel 22/32 :    0
+ip-26-0-160-103:2609922:2610237 [6] NCCL INFO Channel 23/32 :    0
+ip-26-0-160-103:2609922:2610237 [6] NCCL INFO Channel 24/32 :    0
+ip-26-0-160-103:2609922:2610237 [6] NCCL INFO Channel 25/32 :    0
+ip-26-0-160-103:2609922:2610237 [6] NCCL INFO Channel 26/32 :    0
+ip-26-0-160-103:2609922:2610237 [6] NCCL INFO Channel 27/32 :    0
+ip-26-0-160-103:2609922:2610237 [6] NCCL INFO Channel 28/32 :    0
+ip-26-0-160-103:2609922:2610237 [6] NCCL INFO Channel 29/32 :    0
+ip-26-0-160-103:2609922:2610237 [6] NCCL INFO Channel 30/32 :    0
+ip-26-0-160-103:2609922:2610237 [6] NCCL INFO Channel 31/32 :    0
+ip-26-0-160-103:2609922:2610237 [6] NCCL INFO Trees [0] -1/-1/-1->0->-1 [1] -1/-1/-1->0->-1 [2] -1/-1/-1->0->-1 [3] -1/-1/-1->0->-1 [4] -1/-1/-1->0->-1 [5] -1/-1/-1->0->-1 [6] -1/-1/-1->0->-1 [7] -1/-1/-1->0->-1 [8] -1/-1/-1->0->-1 [9] -1/-1/-1->0->-1 [10] -1/-1/-1->0->-1 [11] -1/-1/-1->0->-1 [12] -1/-1/-1->0->-1 [13] -1/-1/-1->0->-1 [14] -1/-1/-1->0->-1 [15] -1/-1/-1->0->-1 [16] -1/-1/-1->0->-1 [17] -1/-1/-1->0->-1 [18] -1/-1/-1->0->-1 [19] -1/-1/-1->0->-1 [20] -1/-1/-1->0->-1 [21] -1/-1/-1->0->-1 [22] -1/-1/-1->0->-1 [23] -1/-1/-1->0->-1 [24] -1/-1/-1->0->-1 [25] -1/-1/-1->0->-1 [26] -1/-1/-1->0->-1 [27] -1/-1/-1->0->-1 [28] -1/-1/-1->0->-1 [29] -1/-1/-1->0->-1 [30] -1/-1/-1->0->-1 [31] -1/-1/-1->0->-1
+ip-26-0-160-103:2609922:2610237 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-103:2609922:2610237 [6] NCCL INFO Connected all rings
+ip-26-0-160-103:2609922:2610237 [6] NCCL INFO Connected all trees
+ip-26-0-160-103:2609922:2610237 [6] NCCL INFO 32 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-160-103:2609920:2610241 [4] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609920:2610241 [4] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609920:2610241 [4] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609920:2610241 [4] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609920:2610241 [4] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609920:2610241 [4] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609920:2610241 [4] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609920:2610241 [4] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609921:2610239 [5] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609921:2610239 [5] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609921:2610239 [5] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609921:2610239 [5] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609920:2610241 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-160-103:2609921:2610239 [5] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609920:2610241 [4] NCCL INFO Channel 00/32 :    0
+ip-26-0-160-103:2609920:2610241 [4] NCCL INFO Channel 01/32 :    0
+ip-26-0-160-103:2609920:2610241 [4] NCCL INFO Channel 02/32 :    0
+ip-26-0-160-103:2609920:2610241 [4] NCCL INFO Channel 03/32 :    0
+ip-26-0-160-103:2609920:2610241 [4] NCCL INFO Channel 04/32 :    0
+ip-26-0-160-103:2609920:2610241 [4] NCCL INFO Channel 05/32 :    0
+ip-26-0-160-103:2609920:2610241 [4] NCCL INFO Channel 06/32 :    0
+ip-26-0-160-103:2609920:2610241 [4] NCCL INFO Channel 07/32 :    0
+ip-26-0-160-103:2609920:2610241 [4] NCCL INFO Channel 08/32 :    0
+ip-26-0-160-103:2609920:2610241 [4] NCCL INFO Channel 09/32 :    0
+ip-26-0-160-103:2609920:2610241 [4] NCCL INFO Channel 10/32 :    0
+ip-26-0-160-103:2609920:2610241 [4] NCCL INFO Channel 11/32 :    0
+ip-26-0-160-103:2609920:2610241 [4] NCCL INFO Channel 12/32 :    0
+ip-26-0-160-103:2609920:2610241 [4] NCCL INFO Channel 13/32 :    0
+ip-26-0-160-103:2609920:2610241 [4] NCCL INFO Channel 14/32 :    0
+ip-26-0-160-103:2609920:2610241 [4] NCCL INFO Channel 15/32 :    0
+ip-26-0-160-103:2609920:2610241 [4] NCCL INFO Channel 16/32 :    0
+ip-26-0-160-103:2609920:2610241 [4] NCCL INFO Channel 17/32 :    0
+ip-26-0-160-103:2609920:2610241 [4] NCCL INFO Channel 18/32 :    0
+ip-26-0-160-103:2609920:2610241 [4] NCCL INFO Channel 19/32 :    0
+ip-26-0-160-103:2609920:2610241 [4] NCCL INFO Channel 20/32 :    0
+ip-26-0-160-103:2609920:2610241 [4] NCCL INFO Channel 21/32 :    0
+ip-26-0-160-103:2609920:2610241 [4] NCCL INFO Channel 22/32 :    0
+ip-26-0-160-103:2609920:2610241 [4] NCCL INFO Channel 23/32 :    0
+ip-26-0-160-103:2609920:2610241 [4] NCCL INFO Channel 24/32 :    0
+ip-26-0-160-103:2609920:2610241 [4] NCCL INFO Channel 25/32 :    0
+ip-26-0-160-103:2609920:2610241 [4] NCCL INFO Channel 26/32 :    0
+ip-26-0-160-103:2609920:2610241 [4] NCCL INFO Channel 27/32 :    0
+ip-26-0-160-103:2609920:2610241 [4] NCCL INFO Channel 28/32 :    0
+ip-26-0-160-103:2609920:2610241 [4] NCCL INFO Channel 29/32 :    0
+ip-26-0-160-103:2609920:2610241 [4] NCCL INFO Channel 30/32 :    0
+ip-26-0-160-103:2609920:2610241 [4] NCCL INFO Channel 31/32 :    0
+ip-26-0-160-103:2609920:2610241 [4] NCCL INFO Trees [0] -1/-1/-1->0->-1 [1] -1/-1/-1->0->-1 [2] -1/-1/-1->0->-1 [3] -1/-1/-1->0->-1 [4] -1/-1/-1->0->-1 [5] -1/-1/-1->0->-1 [6] -1/-1/-1->0->-1 [7] -1/-1/-1->0->-1 [8] -1/-1/-1->0->-1 [9] -1/-1/-1->0->-1 [10] -1/-1/-1->0->-1 [11] -1/-1/-1->0->-1 [12] -1/-1/-1->0->-1 [13] -1/-1/-1->0->-1 [14] -1/-1/-1->0->-1 [15] -1/-1/-1->0->-1 [16] -1/-1/-1->0->-1 [17] -1/-1/-1->0->-1 [18] -1/-1/-1->0->-1 [19] -1/-1/-1->0->-1 [20] -1/-1/-1->0->-1 [21] -1/-1/-1->0->-1 [22] -1/-1/-1->0->-1 [23] -1/-1/-1->0->-1 [24] -1/-1/-1->0->-1 [25] -1/-1/-1->0->-1 [26] -1/-1/-1->0->-1 [27] -1/-1/-1->0->-1 [28] -1/-1/-1->0->-1 [29] -1/-1/-1->0->-1 [30] -1/-1/-1->0->-1 [31] -1/-1/-1->0->-1
+ip-26-0-160-103:2609920:2610241 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-103:2609921:2610239 [5] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609921:2610239 [5] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609921:2610239 [5] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609920:2610241 [4] NCCL INFO Connected all rings
+ip-26-0-160-103:2609920:2610241 [4] NCCL INFO Connected all trees
+ip-26-0-160-103:2609920:2610241 [4] NCCL INFO 32 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-160-103:2609921:2610239 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-160-103:2609921:2610239 [5] NCCL INFO Channel 00/32 :    0
+ip-26-0-160-103:2609921:2610239 [5] NCCL INFO Channel 01/32 :    0
+ip-26-0-160-103:2609921:2610239 [5] NCCL INFO Channel 02/32 :    0
+ip-26-0-160-103:2609921:2610239 [5] NCCL INFO Channel 03/32 :    0
+ip-26-0-160-103:2609921:2610239 [5] NCCL INFO Channel 04/32 :    0
+ip-26-0-160-103:2609921:2610239 [5] NCCL INFO Channel 05/32 :    0
+ip-26-0-160-103:2609921:2610239 [5] NCCL INFO Channel 06/32 :    0
+ip-26-0-160-103:2609921:2610239 [5] NCCL INFO Channel 07/32 :    0
+ip-26-0-160-103:2609921:2610239 [5] NCCL INFO Channel 08/32 :    0
+ip-26-0-160-103:2609921:2610239 [5] NCCL INFO Channel 09/32 :    0
+ip-26-0-160-103:2609921:2610239 [5] NCCL INFO Channel 10/32 :    0
+ip-26-0-160-103:2609921:2610239 [5] NCCL INFO Channel 11/32 :    0
+ip-26-0-160-103:2609921:2610239 [5] NCCL INFO Channel 12/32 :    0
+ip-26-0-160-103:2609921:2610239 [5] NCCL INFO Channel 13/32 :    0
+ip-26-0-160-103:2609921:2610239 [5] NCCL INFO Channel 14/32 :    0
+ip-26-0-160-103:2609921:2610239 [5] NCCL INFO Channel 15/32 :    0
+ip-26-0-160-103:2609921:2610239 [5] NCCL INFO Channel 16/32 :    0
+ip-26-0-160-103:2609921:2610239 [5] NCCL INFO Channel 17/32 :    0
+ip-26-0-160-103:2609921:2610239 [5] NCCL INFO Channel 18/32 :    0
+ip-26-0-160-103:2609921:2610239 [5] NCCL INFO Channel 19/32 :    0
+ip-26-0-160-103:2609921:2610239 [5] NCCL INFO Channel 20/32 :    0
+ip-26-0-160-103:2609921:2610239 [5] NCCL INFO Channel 21/32 :    0
+ip-26-0-160-103:2609921:2610239 [5] NCCL INFO Channel 22/32 :    0
+ip-26-0-160-103:2609921:2610239 [5] NCCL INFO Channel 23/32 :    0
+ip-26-0-160-103:2609921:2610239 [5] NCCL INFO Channel 24/32 :    0
+ip-26-0-160-103:2609921:2610239 [5] NCCL INFO Channel 25/32 :    0
+ip-26-0-160-103:2609921:2610239 [5] NCCL INFO Channel 26/32 :    0
+ip-26-0-160-103:2609921:2610239 [5] NCCL INFO Channel 27/32 :    0
+ip-26-0-160-103:2609921:2610239 [5] NCCL INFO Channel 28/32 :    0
+ip-26-0-160-103:2609921:2610239 [5] NCCL INFO Channel 29/32 :    0
+ip-26-0-160-103:2609921:2610239 [5] NCCL INFO Channel 30/32 :    0
+ip-26-0-160-103:2609921:2610239 [5] NCCL INFO Channel 31/32 :    0
+ip-26-0-160-103:2609921:2610239 [5] NCCL INFO Trees [0] -1/-1/-1->0->-1 [1] -1/-1/-1->0->-1 [2] -1/-1/-1->0->-1 [3] -1/-1/-1->0->-1 [4] -1/-1/-1->0->-1 [5] -1/-1/-1->0->-1 [6] -1/-1/-1->0->-1 [7] -1/-1/-1->0->-1 [8] -1/-1/-1->0->-1 [9] -1/-1/-1->0->-1 [10] -1/-1/-1->0->-1 [11] -1/-1/-1->0->-1 [12] -1/-1/-1->0->-1 [13] -1/-1/-1->0->-1 [14] -1/-1/-1->0->-1 [15] -1/-1/-1->0->-1 [16] -1/-1/-1->0->-1 [17] -1/-1/-1->0->-1 [18] -1/-1/-1->0->-1 [19] -1/-1/-1->0->-1 [20] -1/-1/-1->0->-1 [21] -1/-1/-1->0->-1 [22] -1/-1/-1->0->-1 [23] -1/-1/-1->0->-1 [24] -1/-1/-1->0->-1 [25] -1/-1/-1->0->-1 [26] -1/-1/-1->0->-1 [27] -1/-1/-1->0->-1 [28] -1/-1/-1->0->-1 [29] -1/-1/-1->0->-1 [30] -1/-1/-1->0->-1 [31] -1/-1/-1->0->-1
+ip-26-0-160-103:2609921:2610239 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-103:2609921:2610239 [5] NCCL INFO Connected all rings
+ip-26-0-160-103:2609921:2610239 [5] NCCL INFO Connected all trees
+ip-26-0-160-103:2609921:2610239 [5] NCCL INFO 32 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-160-103:2609922:2610237 [6] NCCL INFO comm 0xb6ae0d0 rank 0 nranks 1 cudaDev 6 nvmlDev 6 busId b9000 commId 0x33f609ebe74c0f2d - Init COMPLETE
+ip-26-0-160-103:2609920:2610241 [4] NCCL INFO comm 0xb164240 rank 0 nranks 1 cudaDev 4 nvmlDev 4 busId 97000 commId 0x31e3be134e0000d5 - Init COMPLETE
+ip-26-0-160-103:2609922:2610264 [6] NCCL INFO Using network AWS Libfabric
+ip-26-0-160-103:2609921:2610239 [5] NCCL INFO comm 0xadcb040 rank 0 nranks 1 cudaDev 5 nvmlDev 5 busId a8000 commId 0x80c61819814e44a6 - Init COMPLETE
+ip-26-0-160-103:2609923:2610265 [7] NCCL INFO Using network AWS Libfabric
+ip-26-0-160-103:2609921:2610267 [5] NCCL INFO Using network AWS Libfabric
+ip-26-0-160-103:2609920:2610266 [4] NCCL INFO Using network AWS Libfabric
+ip-26-0-160-103:2609921:2610267 [5] NCCL INFO comm 0xadd0310 rank 2 nranks 4 cudaDev 5 nvmlDev 5 busId a8000 commId 0x35779fcbea0678c9 - Init START
+ip-26-0-160-103:2609919:2610257 [3] NCCL INFO comm 0xa535e00 rank 1 nranks 4 cudaDev 3 nvmlDev 3 busId 86000 commId 0x35779fcbea0678c9 - Init START
+ip-26-0-160-103:2609923:2610265 [7] NCCL INFO comm 0xc284020 rank 3 nranks 4 cudaDev 7 nvmlDev 7 busId ca000 commId 0x35779fcbea0678c9 - Init START
+ip-26-0-160-103:2609917:2610249 [1] NCCL INFO comm 0xafc0a30 rank 0 nranks 4 cudaDev 1 nvmlDev 1 busId 64000 commId 0x35779fcbea0678c9 - Init START
+ip-26-0-160-103:2609923:2610265 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /opt/aws-ofi-nccl/share/aws-ofi-nccl/xml/p5.48xl-topo.xml
+ip-26-0-160-103:2609919:2610257 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /opt/aws-ofi-nccl/share/aws-ofi-nccl/xml/p5.48xl-topo.xml
+ip-26-0-160-103:2609921:2610267 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /opt/aws-ofi-nccl/share/aws-ofi-nccl/xml/p5.48xl-topo.xml
+ip-26-0-160-103:2609917:2610249 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /opt/aws-ofi-nccl/share/aws-ofi-nccl/xml/p5.48xl-topo.xml
+ip-26-0-160-103:2609920:2610266 [4] NCCL INFO comm 0xb168a60 rank 2 nranks 4 cudaDev 4 nvmlDev 4 busId 97000 commId 0xe076a754049e1216 - Init START
+ip-26-0-160-103:2609922:2610264 [6] NCCL INFO comm 0xb6b3550 rank 3 nranks 4 cudaDev 6 nvmlDev 6 busId b9000 commId 0xe076a754049e1216 - Init START
+ip-26-0-160-103:2609918:2610256 [2] NCCL INFO comm 0xc18ede0 rank 1 nranks 4 cudaDev 2 nvmlDev 2 busId 75000 commId 0xe076a754049e1216 - Init START
+ip-26-0-160-103:2609916:2610247 [0] NCCL INFO comm 0xb9cd580 rank 0 nranks 4 cudaDev 0 nvmlDev 0 busId 53000 commId 0xe076a754049e1216 - Init START
+ip-26-0-160-103:2609920:2610266 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /opt/aws-ofi-nccl/share/aws-ofi-nccl/xml/p5.48xl-topo.xml
+ip-26-0-160-103:2609922:2610264 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /opt/aws-ofi-nccl/share/aws-ofi-nccl/xml/p5.48xl-topo.xml
+ip-26-0-160-103:2609918:2610256 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /opt/aws-ofi-nccl/share/aws-ofi-nccl/xml/p5.48xl-topo.xml
+ip-26-0-160-103:2609916:2610247 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /opt/aws-ofi-nccl/share/aws-ofi-nccl/xml/p5.48xl-topo.xml
+ip-26-0-160-103:2609916:2610247 [0] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609916:2610247 [0] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609916:2610247 [0] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609916:2610247 [0] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609916:2610247 [0] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609916:2610247 [0] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609916:2610247 [0] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609916:2610247 [0] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609916:2610247 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-160-103:2609916:2610247 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-160-103:2609923:2610265 [7] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609923:2610265 [7] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609923:2610265 [7] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609923:2610265 [7] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609923:2610265 [7] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609923:2610265 [7] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609923:2610265 [7] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609919:2610257 [3] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609923:2610265 [7] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609919:2610257 [3] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609919:2610257 [3] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609919:2610257 [3] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609919:2610257 [3] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609919:2610257 [3] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609919:2610257 [3] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609919:2610257 [3] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609923:2610265 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-160-103:2609923:2610265 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-160-103:2609919:2610257 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-160-103:2609919:2610257 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-160-103:2609921:2610267 [5] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609921:2610267 [5] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609921:2610267 [5] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609921:2610267 [5] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609921:2610267 [5] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609921:2610267 [5] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609921:2610267 [5] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609921:2610267 [5] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609921:2610267 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-160-103:2609921:2610267 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-160-103:2609920:2610266 [4] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609920:2610266 [4] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609920:2610266 [4] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609920:2610266 [4] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609920:2610266 [4] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609920:2610266 [4] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609920:2610266 [4] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609920:2610266 [4] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609920:2610266 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-160-103:2609920:2610266 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-160-103:2609917:2610249 [1] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609917:2610249 [1] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609917:2610249 [1] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609917:2610249 [1] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609917:2610249 [1] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609917:2610249 [1] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609917:2610249 [1] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609917:2610249 [1] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609922:2610264 [6] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609922:2610264 [6] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609922:2610264 [6] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609922:2610264 [6] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609922:2610264 [6] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609917:2610249 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-160-103:2609917:2610249 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-160-103:2609922:2610264 [6] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609922:2610264 [6] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609922:2610264 [6] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609922:2610264 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-160-103:2609922:2610264 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-160-103:2609918:2610256 [2] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609918:2610256 [2] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609918:2610256 [2] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609918:2610256 [2] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609918:2610256 [2] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609918:2610256 [2] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609918:2610256 [2] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609918:2610256 [2] NCCL INFO NET/OFI Libfabric provider associates MRs with domains
+ip-26-0-160-103:2609918:2610256 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-160-103:2609918:2610256 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-160-103:2609917:2610249 [1] NCCL INFO Channel 00/24 :    0   1   2   3
+ip-26-0-160-103:2609917:2610249 [1] NCCL INFO Channel 01/24 :    0   1   2   3
+ip-26-0-160-103:2609917:2610249 [1] NCCL INFO Channel 02/24 :    0   1   2   3
+ip-26-0-160-103:2609917:2610249 [1] NCCL INFO Channel 03/24 :    0   1   2   3
+ip-26-0-160-103:2609917:2610249 [1] NCCL INFO Channel 04/24 :    0   1   2   3
+ip-26-0-160-103:2609917:2610249 [1] NCCL INFO Channel 05/24 :    0   1   2   3
+ip-26-0-160-103:2609917:2610249 [1] NCCL INFO Channel 06/24 :    0   1   2   3
+ip-26-0-160-103:2609917:2610249 [1] NCCL INFO Channel 07/24 :    0   1   2   3
+ip-26-0-160-103:2609919:2610257 [3] NCCL INFO Trees [0] 2/-1/-1->1->0 [1] 2/-1/-1->1->0 [2] 2/-1/-1->1->0 [3] 2/-1/-1->1->0 [4] 2/-1/-1->1->0 [5] 2/-1/-1->1->0 [6] 2/-1/-1->1->0 [7] 2/-1/-1->1->0 [8] 2/-1/-1->1->0 [9] 2/-1/-1->1->0 [10] 2/-1/-1->1->0 [11] 2/-1/-1->1->0 [12] 2/-1/-1->1->0 [13] 2/-1/-1->1->0 [14] 2/-1/-1->1->0 [15] 2/-1/-1->1->0 [16] 2/-1/-1->1->0 [17] 2/-1/-1->1->0 [18] 2/-1/-1->1->0 [19] 2/-1/-1->1->0 [20] 2/-1/-1->1->0 [21] 2/-1/-1->1->0 [22] 2/-1/-1->1->0 [23] 2/-1/-1->1->0
+ip-26-0-160-103:2609923:2610265 [7] NCCL INFO Trees [0] -1/-1/-1->3->2 [1] -1/-1/-1->3->2 [2] -1/-1/-1->3->2 [3] -1/-1/-1->3->2 [4] -1/-1/-1->3->2 [5] -1/-1/-1->3->2 [6] -1/-1/-1->3->2 [7] -1/-1/-1->3->2 [8] -1/-1/-1->3->2 [9] -1/-1/-1->3->2 [10] -1/-1/-1->3->2 [11] -1/-1/-1->3->2 [12] -1/-1/-1->3->2 [13] -1/-1/-1->3->2 [14] -1/-1/-1->3->2 [15] -1/-1/-1->3->2 [16] -1/-1/-1->3->2 [17] -1/-1/-1->3->2 [18] -1/-1/-1->3->2 [19] -1/-1/-1->3->2 [20] -1/-1/-1->3->2 [21] -1/-1/-1->3->2 [22] -1/-1/-1->3->2 [23] -1/-1/-1->3->2
+ip-26-0-160-103:2609919:2610257 [3] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-160-103:2609923:2610265 [7] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-160-103:2609917:2610249 [1] NCCL INFO Channel 08/24 :    0   1   2   3
+ip-26-0-160-103:2609921:2610267 [5] NCCL INFO Trees [0] 3/-1/-1->2->1 [1] 3/-1/-1->2->1 [2] 3/-1/-1->2->1 [3] 3/-1/-1->2->1 [4] 3/-1/-1->2->1 [5] 3/-1/-1->2->1 [6] 3/-1/-1->2->1 [7] 3/-1/-1->2->1 [8] 3/-1/-1->2->1 [9] 3/-1/-1->2->1 [10] 3/-1/-1->2->1 [11] 3/-1/-1->2->1 [12] 3/-1/-1->2->1 [13] 3/-1/-1->2->1 [14] 3/-1/-1->2->1 [15] 3/-1/-1->2->1 [16] 3/-1/-1->2->1 [17] 3/-1/-1->2->1 [18] 3/-1/-1->2->1 [19] 3/-1/-1->2->1 [20] 3/-1/-1->2->1 [21] 3/-1/-1->2->1 [22] 3/-1/-1->2->1 [23] 3/-1/-1->2->1
+ip-26-0-160-103:2609917:2610249 [1] NCCL INFO Channel 09/24 :    0   1   2   3
+ip-26-0-160-103:2609921:2610267 [5] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-160-103:2609917:2610249 [1] NCCL INFO Channel 10/24 :    0   1   2   3
+ip-26-0-160-103:2609917:2610249 [1] NCCL INFO Channel 11/24 :    0   1   2   3
+ip-26-0-160-103:2609917:2610249 [1] NCCL INFO Channel 12/24 :    0   1   2   3
+ip-26-0-160-103:2609917:2610249 [1] NCCL INFO Channel 13/24 :    0   1   2   3
+ip-26-0-160-103:2609917:2610249 [1] NCCL INFO Channel 14/24 :    0   1   2   3
+ip-26-0-160-103:2609917:2610249 [1] NCCL INFO Channel 15/24 :    0   1   2   3
+ip-26-0-160-103:2609917:2610249 [1] NCCL INFO Channel 16/24 :    0   1   2   3
+ip-26-0-160-103:2609917:2610249 [1] NCCL INFO Channel 17/24 :    0   1   2   3
+ip-26-0-160-103:2609917:2610249 [1] NCCL INFO Channel 18/24 :    0   1   2   3
+ip-26-0-160-103:2609917:2610249 [1] NCCL INFO Channel 19/24 :    0   1   2   3
+ip-26-0-160-103:2609917:2610249 [1] NCCL INFO Channel 20/24 :    0   1   2   3
+ip-26-0-160-103:2609917:2610249 [1] NCCL INFO Channel 21/24 :    0   1   2   3
+ip-26-0-160-103:2609917:2610249 [1] NCCL INFO Channel 22/24 :    0   1   2   3
+ip-26-0-160-103:2609917:2610249 [1] NCCL INFO Channel 23/24 :    0   1   2   3
+ip-26-0-160-103:2609917:2610249 [1] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] 1/-1/-1->0->-1 [7] 1/-1/-1->0->-1 [8] 1/-1/-1->0->-1 [9] 1/-1/-1->0->-1 [10] 1/-1/-1->0->-1 [11] 1/-1/-1->0->-1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] 1/-1/-1->0->-1 [19] 1/-1/-1->0->-1 [20] 1/-1/-1->0->-1 [21] 1/-1/-1->0->-1 [22] 1/-1/-1->0->-1 [23] 1/-1/-1->0->-1
+ip-26-0-160-103:2609917:2610249 [1] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-160-103:2609916:2610247 [0] NCCL INFO Channel 00/24 :    0   1   2   3
+ip-26-0-160-103:2609918:2610256 [2] NCCL INFO Trees [0] 2/-1/-1->1->0 [1] 2/-1/-1->1->0 [2] 2/-1/-1->1->0 [3] 2/-1/-1->1->0 [4] 2/-1/-1->1->0 [5] 2/-1/-1->1->0 [6] 2/-1/-1->1->0 [7] 2/-1/-1->1->0 [8] 2/-1/-1->1->0 [9] 2/-1/-1->1->0 [10] 2/-1/-1->1->0 [11] 2/-1/-1->1->0 [12] 2/-1/-1->1->0 [13] 2/-1/-1->1->0 [14] 2/-1/-1->1->0 [15] 2/-1/-1->1->0 [16] 2/-1/-1->1->0 [17] 2/-1/-1->1->0 [18] 2/-1/-1->1->0 [19] 2/-1/-1->1->0 [20] 2/-1/-1->1->0 [21] 2/-1/-1->1->0 [22] 2/-1/-1->1->0 [23] 2/-1/-1->1->0
+ip-26-0-160-103:2609916:2610247 [0] NCCL INFO Channel 01/24 :    0   1   2   3
+ip-26-0-160-103:2609918:2610256 [2] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-160-103:2609916:2610247 [0] NCCL INFO Channel 02/24 :    0   1   2   3
+ip-26-0-160-103:2609916:2610247 [0] NCCL INFO Channel 03/24 :    0   1   2   3
+ip-26-0-160-103:2609916:2610247 [0] NCCL INFO Channel 04/24 :    0   1   2   3
+ip-26-0-160-103:2609916:2610247 [0] NCCL INFO Channel 05/24 :    0   1   2   3
+ip-26-0-160-103:2609916:2610247 [0] NCCL INFO Channel 06/24 :    0   1   2   3
+ip-26-0-160-103:2609916:2610247 [0] NCCL INFO Channel 07/24 :    0   1   2   3
+ip-26-0-160-103:2609916:2610247 [0] NCCL INFO Channel 08/24 :    0   1   2   3
+ip-26-0-160-103:2609922:2610264 [6] NCCL INFO Trees [0] -1/-1/-1->3->2 [1] -1/-1/-1->3->2 [2] -1/-1/-1->3->2 [3] -1/-1/-1->3->2 [4] -1/-1/-1->3->2 [5] -1/-1/-1->3->2 [6] -1/-1/-1->3->2 [7] -1/-1/-1->3->2 [8] -1/-1/-1->3->2 [9] -1/-1/-1->3->2 [10] -1/-1/-1->3->2 [11] -1/-1/-1->3->2 [12] -1/-1/-1->3->2 [13] -1/-1/-1->3->2 [14] -1/-1/-1->3->2 [15] -1/-1/-1->3->2 [16] -1/-1/-1->3->2 [17] -1/-1/-1->3->2 [18] -1/-1/-1->3->2 [19] -1/-1/-1->3->2 [20] -1/-1/-1->3->2 [21] -1/-1/-1->3->2 [22] -1/-1/-1->3->2 [23] -1/-1/-1->3->2
+ip-26-0-160-103:2609916:2610247 [0] NCCL INFO Channel 09/24 :    0   1   2   3
+ip-26-0-160-103:2609920:2610266 [4] NCCL INFO Trees [0] 3/-1/-1->2->1 [1] 3/-1/-1->2->1 [2] 3/-1/-1->2->1 [3] 3/-1/-1->2->1 [4] 3/-1/-1->2->1 [5] 3/-1/-1->2->1 [6] 3/-1/-1->2->1 [7] 3/-1/-1->2->1 [8] 3/-1/-1->2->1 [9] 3/-1/-1->2->1 [10] 3/-1/-1->2->1 [11] 3/-1/-1->2->1 [12] 3/-1/-1->2->1 [13] 3/-1/-1->2->1 [14] 3/-1/-1->2->1 [15] 3/-1/-1->2->1 [16] 3/-1/-1->2->1 [17] 3/-1/-1->2->1 [18] 3/-1/-1->2->1 [19] 3/-1/-1->2->1 [20] 3/-1/-1->2->1 [21] 3/-1/-1->2->1 [22] 3/-1/-1->2->1 [23] 3/-1/-1->2->1
+ip-26-0-160-103:2609916:2610247 [0] NCCL INFO Channel 10/24 :    0   1   2   3
+ip-26-0-160-103:2609922:2610264 [6] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-160-103:2609920:2610266 [4] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-160-103:2609916:2610247 [0] NCCL INFO Channel 11/24 :    0   1   2   3
+ip-26-0-160-103:2609916:2610247 [0] NCCL INFO Channel 12/24 :    0   1   2   3
+ip-26-0-160-103:2609916:2610247 [0] NCCL INFO Channel 13/24 :    0   1   2   3
+ip-26-0-160-103:2609916:2610247 [0] NCCL INFO Channel 14/24 :    0   1   2   3
+ip-26-0-160-103:2609916:2610247 [0] NCCL INFO Channel 15/24 :    0   1   2   3
+ip-26-0-160-103:2609916:2610247 [0] NCCL INFO Channel 16/24 :    0   1   2   3
+ip-26-0-160-103:2609916:2610247 [0] NCCL INFO Channel 17/24 :    0   1   2   3
+ip-26-0-160-103:2609916:2610247 [0] NCCL INFO Channel 18/24 :    0   1   2   3
+ip-26-0-160-103:2609916:2610247 [0] NCCL INFO Channel 19/24 :    0   1   2   3
+ip-26-0-160-103:2609916:2610247 [0] NCCL INFO Channel 20/24 :    0   1   2   3
+ip-26-0-160-103:2609916:2610247 [0] NCCL INFO Channel 21/24 :    0   1   2   3
+ip-26-0-160-103:2609916:2610247 [0] NCCL INFO Channel 22/24 :    0   1   2   3
+ip-26-0-160-103:2609916:2610247 [0] NCCL INFO Channel 23/24 :    0   1   2   3
+ip-26-0-160-103:2609916:2610247 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] 1/-1/-1->0->-1 [7] 1/-1/-1->0->-1 [8] 1/-1/-1->0->-1 [9] 1/-1/-1->0->-1 [10] 1/-1/-1->0->-1 [11] 1/-1/-1->0->-1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] 1/-1/-1->0->-1 [19] 1/-1/-1->0->-1 [20] 1/-1/-1->0->-1 [21] 1/-1/-1->0->-1 [22] 1/-1/-1->0->-1 [23] 1/-1/-1->0->-1
+ip-26-0-160-103:2609916:2610247 [0] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-160-103:2609919:2610257 [3] NCCL INFO Channel 00/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-160-103:2609921:2610267 [5] NCCL INFO Channel 00/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-160-103:2609923:2610265 [7] NCCL INFO Channel 00/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-160-103:2609919:2610257 [3] NCCL INFO Channel 01/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-160-103:2609917:2610249 [1] NCCL INFO Channel 00/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2609921:2610267 [5] NCCL INFO Channel 01/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-160-103:2609923:2610265 [7] NCCL INFO Channel 01/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-160-103:2609919:2610257 [3] NCCL INFO Channel 02/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-160-103:2609919:2610257 [3] NCCL INFO Channel 03/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-160-103:2609921:2610267 [5] NCCL INFO Channel 02/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-160-103:2609923:2610265 [7] NCCL INFO Channel 02/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-160-103:2609919:2610257 [3] NCCL INFO Channel 04/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-160-103:2609921:2610267 [5] NCCL INFO Channel 03/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-160-103:2609923:2610265 [7] NCCL INFO Channel 03/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-160-103:2609917:2610249 [1] NCCL INFO Channel 01/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2609923:2610265 [7] NCCL INFO Channel 04/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-160-103:2609917:2610249 [1] NCCL INFO Channel 02/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2609923:2610265 [7] NCCL INFO Channel 05/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-160-103:2609917:2610249 [1] NCCL INFO Channel 03/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2609923:2610265 [7] NCCL INFO Channel 06/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-160-103:2609917:2610249 [1] NCCL INFO Channel 04/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2609923:2610265 [7] NCCL INFO Channel 07/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-160-103:2609917:2610249 [1] NCCL INFO Channel 05/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2609917:2610249 [1] NCCL INFO Channel 06/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2609923:2610265 [7] NCCL INFO Channel 08/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-160-103:2609921:2610267 [5] NCCL INFO Channel 04/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-160-103:2609921:2610267 [5] NCCL INFO Channel 05/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-160-103:2609919:2610257 [3] NCCL INFO Channel 05/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-160-103:2609919:2610257 [3] NCCL INFO Channel 06/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-160-103:2609923:2610265 [7] NCCL INFO Channel 09/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-160-103:2609919:2610257 [3] NCCL INFO Channel 07/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-160-103:2609923:2610265 [7] NCCL INFO Channel 10/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-160-103:2609917:2610249 [1] NCCL INFO Channel 07/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2609923:2610265 [7] NCCL INFO Channel 11/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-160-103:2609919:2610257 [3] NCCL INFO Channel 08/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-160-103:2609917:2610249 [1] NCCL INFO Channel 08/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2609923:2610265 [7] NCCL INFO Channel 12/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-160-103:2609919:2610257 [3] NCCL INFO Channel 09/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-160-103:2609917:2610249 [1] NCCL INFO Channel 09/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2609923:2610265 [7] NCCL INFO Channel 13/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-160-103:2609919:2610257 [3] NCCL INFO Channel 10/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-160-103:2609917:2610249 [1] NCCL INFO Channel 10/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2609923:2610265 [7] NCCL INFO Channel 14/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-160-103:2609919:2610257 [3] NCCL INFO Channel 11/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-160-103:2609917:2610249 [1] NCCL INFO Channel 11/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2609923:2610265 [7] NCCL INFO Channel 15/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-160-103:2609919:2610257 [3] NCCL INFO Channel 12/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-160-103:2609921:2610267 [5] NCCL INFO Channel 06/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-160-103:2609920:2610266 [4] NCCL INFO Channel 00/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-160-103:2609917:2610249 [1] NCCL INFO Channel 12/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2609918:2610256 [2] NCCL INFO Channel 00/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-160-103:2609923:2610265 [7] NCCL INFO Channel 16/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-160-103:2609919:2610257 [3] NCCL INFO Channel 13/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-160-103:2609921:2610267 [5] NCCL INFO Channel 07/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-160-103:2609920:2610266 [4] NCCL INFO Channel 01/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-160-103:2609917:2610249 [1] NCCL INFO Channel 13/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2609918:2610256 [2] NCCL INFO Channel 01/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-160-103:2609919:2610257 [3] NCCL INFO Channel 14/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-160-103:2609923:2610265 [7] NCCL INFO Channel 17/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-160-103:2609921:2610267 [5] NCCL INFO Channel 08/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-160-103:2609920:2610266 [4] NCCL INFO Channel 02/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-160-103:2609917:2610249 [1] NCCL INFO Channel 14/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2609918:2610256 [2] NCCL INFO Channel 02/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-160-103:2609922:2610264 [6] NCCL INFO Channel 00/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609923:2610265 [7] NCCL INFO Channel 18/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-160-103:2609919:2610257 [3] NCCL INFO Channel 15/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-160-103:2609921:2610267 [5] NCCL INFO Channel 09/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-160-103:2609920:2610266 [4] NCCL INFO Channel 03/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-160-103:2609917:2610249 [1] NCCL INFO Channel 15/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2609918:2610256 [2] NCCL INFO Channel 03/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-160-103:2609923:2610265 [7] NCCL INFO Channel 19/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-160-103:2609919:2610257 [3] NCCL INFO Channel 16/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-160-103:2609922:2610264 [6] NCCL INFO Channel 01/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609921:2610267 [5] NCCL INFO Channel 10/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-160-103:2609920:2610266 [4] NCCL INFO Channel 04/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-160-103:2609917:2610249 [1] NCCL INFO Channel 16/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2609918:2610256 [2] NCCL INFO Channel 04/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-160-103:2609923:2610265 [7] NCCL INFO Channel 20/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-160-103:2609919:2610257 [3] NCCL INFO Channel 17/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-160-103:2609922:2610264 [6] NCCL INFO Channel 02/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609921:2610267 [5] NCCL INFO Channel 11/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-160-103:2609920:2610266 [4] NCCL INFO Channel 05/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-160-103:2609917:2610249 [1] NCCL INFO Channel 17/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2609918:2610256 [2] NCCL INFO Channel 05/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-160-103:2609923:2610265 [7] NCCL INFO Channel 21/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-160-103:2609919:2610257 [3] NCCL INFO Channel 18/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-160-103:2609922:2610264 [6] NCCL INFO Channel 03/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609921:2610267 [5] NCCL INFO Channel 12/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-160-103:2609920:2610266 [4] NCCL INFO Channel 06/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-160-103:2609916:2610247 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-160-103:2609917:2610249 [1] NCCL INFO Channel 18/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2609918:2610256 [2] NCCL INFO Channel 06/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-160-103:2609923:2610265 [7] NCCL INFO Channel 22/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-160-103:2609919:2610257 [3] NCCL INFO Channel 19/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-160-103:2609922:2610264 [6] NCCL INFO Channel 04/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609921:2610267 [5] NCCL INFO Channel 13/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-160-103:2609920:2610266 [4] NCCL INFO Channel 07/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-160-103:2609916:2610247 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-160-103:2609917:2610249 [1] NCCL INFO Channel 19/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2609918:2610256 [2] NCCL INFO Channel 07/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-160-103:2609923:2610265 [7] NCCL INFO Channel 23/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-160-103:2609919:2610257 [3] NCCL INFO Channel 20/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-160-103:2609922:2610264 [6] NCCL INFO Channel 05/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609921:2610267 [5] NCCL INFO Channel 14/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-160-103:2609920:2610266 [4] NCCL INFO Channel 08/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-160-103:2609916:2610247 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-160-103:2609917:2610249 [1] NCCL INFO Channel 20/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2609918:2610256 [2] NCCL INFO Channel 08/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-160-103:2609919:2610257 [3] NCCL INFO Channel 21/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-160-103:2609922:2610264 [6] NCCL INFO Channel 06/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609921:2610267 [5] NCCL INFO Channel 15/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-160-103:2609916:2610247 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-160-103:2609917:2610249 [1] NCCL INFO Channel 21/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2609920:2610266 [4] NCCL INFO Channel 09/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-160-103:2609918:2610256 [2] NCCL INFO Channel 09/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-160-103:2609922:2610264 [6] NCCL INFO Channel 07/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609919:2610257 [3] NCCL INFO Channel 22/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-160-103:2609916:2610247 [0] NCCL INFO Channel 04/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-160-103:2609917:2610249 [1] NCCL INFO Channel 22/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2609921:2610267 [5] NCCL INFO Channel 16/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-160-103:2609920:2610266 [4] NCCL INFO Channel 10/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-160-103:2609918:2610256 [2] NCCL INFO Channel 10/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-160-103:2609919:2610257 [3] NCCL INFO Channel 23/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-160-103:2609922:2610264 [6] NCCL INFO Channel 08/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609916:2610247 [0] NCCL INFO Channel 05/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-160-103:2609917:2610249 [1] NCCL INFO Channel 23/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2609921:2610267 [5] NCCL INFO Channel 17/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-160-103:2609920:2610266 [4] NCCL INFO Channel 11/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-160-103:2609918:2610256 [2] NCCL INFO Channel 11/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-160-103:2609922:2610264 [6] NCCL INFO Channel 09/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609916:2610247 [0] NCCL INFO Channel 06/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-160-103:2609920:2610266 [4] NCCL INFO Channel 12/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-160-103:2609921:2610267 [5] NCCL INFO Channel 18/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-160-103:2609918:2610256 [2] NCCL INFO Channel 12/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-160-103:2609922:2610264 [6] NCCL INFO Channel 10/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609916:2610247 [0] NCCL INFO Channel 07/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-160-103:2609920:2610266 [4] NCCL INFO Channel 13/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-160-103:2609921:2610267 [5] NCCL INFO Channel 19/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-160-103:2609918:2610256 [2] NCCL INFO Channel 13/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-160-103:2609922:2610264 [6] NCCL INFO Channel 11/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609916:2610247 [0] NCCL INFO Channel 08/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-160-103:2609920:2610266 [4] NCCL INFO Channel 14/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-160-103:2609921:2610267 [5] NCCL INFO Channel 20/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-160-103:2609918:2610256 [2] NCCL INFO Channel 14/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-160-103:2609922:2610264 [6] NCCL INFO Channel 12/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609916:2610247 [0] NCCL INFO Channel 09/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-160-103:2609920:2610266 [4] NCCL INFO Channel 15/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-160-103:2609921:2610267 [5] NCCL INFO Channel 21/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-160-103:2609920:2610266 [4] NCCL INFO Channel 16/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-160-103:2609922:2610264 [6] NCCL INFO Channel 13/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609921:2610267 [5] NCCL INFO Channel 22/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-160-103:2609918:2610256 [2] NCCL INFO Channel 15/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-160-103:2609922:2610264 [6] NCCL INFO Channel 14/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609916:2610247 [0] NCCL INFO Channel 10/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-160-103:2609921:2610267 [5] NCCL INFO Channel 23/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-160-103:2609918:2610256 [2] NCCL INFO Channel 16/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-160-103:2609922:2610264 [6] NCCL INFO Channel 15/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609916:2610247 [0] NCCL INFO Channel 11/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-160-103:2609920:2610266 [4] NCCL INFO Channel 17/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-160-103:2609916:2610247 [0] NCCL INFO Channel 12/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-160-103:2609916:2610247 [0] NCCL INFO Channel 13/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-160-103:2609916:2610247 [0] NCCL INFO Channel 14/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-160-103:2609916:2610247 [0] NCCL INFO Channel 15/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-160-103:2609916:2610247 [0] NCCL INFO Channel 16/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-160-103:2609916:2610247 [0] NCCL INFO Channel 17/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-160-103:2609918:2610256 [2] NCCL INFO Channel 17/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-160-103:2609922:2610264 [6] NCCL INFO Channel 16/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609920:2610266 [4] NCCL INFO Channel 18/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-160-103:2609916:2610247 [0] NCCL INFO Channel 18/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-160-103:2609918:2610256 [2] NCCL INFO Channel 18/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-160-103:2609922:2610264 [6] NCCL INFO Channel 17/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609922:2610264 [6] NCCL INFO Channel 18/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609920:2610266 [4] NCCL INFO Channel 19/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-160-103:2609916:2610247 [0] NCCL INFO Channel 19/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-160-103:2609918:2610256 [2] NCCL INFO Channel 19/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-160-103:2609922:2610264 [6] NCCL INFO Channel 19/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609920:2610266 [4] NCCL INFO Channel 20/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-160-103:2609922:2610264 [6] NCCL INFO Channel 20/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609916:2610247 [0] NCCL INFO Channel 20/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-160-103:2609918:2610256 [2] NCCL INFO Channel 20/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-160-103:2609920:2610266 [4] NCCL INFO Channel 21/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-160-103:2609922:2610264 [6] NCCL INFO Channel 21/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609916:2610247 [0] NCCL INFO Channel 21/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-160-103:2609918:2610256 [2] NCCL INFO Channel 21/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-160-103:2609920:2610266 [4] NCCL INFO Channel 22/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-160-103:2609922:2610264 [6] NCCL INFO Channel 22/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609916:2610247 [0] NCCL INFO Channel 22/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-160-103:2609918:2610256 [2] NCCL INFO Channel 22/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-160-103:2609920:2610266 [4] NCCL INFO Channel 23/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-160-103:2609918:2610256 [2] NCCL INFO Channel 23/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-160-103:2609916:2610247 [0] NCCL INFO Channel 23/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-160-103:2609922:2610264 [6] NCCL INFO Channel 23/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609917:2610249 [1] NCCL INFO Connected all rings
+ip-26-0-160-103:2609921:2610267 [5] NCCL INFO Connected all rings
+ip-26-0-160-103:2609919:2610257 [3] NCCL INFO Connected all rings
+ip-26-0-160-103:2609923:2610265 [7] NCCL INFO Connected all rings
+ip-26-0-160-103:2609923:2610265 [7] NCCL INFO Channel 00/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-160-103:2609923:2610265 [7] NCCL INFO Channel 01/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-160-103:2609923:2610265 [7] NCCL INFO Channel 02/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-160-103:2609923:2610265 [7] NCCL INFO Channel 03/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-160-103:2609923:2610265 [7] NCCL INFO Channel 04/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-160-103:2609922:2610264 [6] NCCL INFO Connected all rings
+ip-26-0-160-103:2609922:2610264 [6] NCCL INFO Channel 00/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-160-103:2609923:2610265 [7] NCCL INFO Channel 05/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-160-103:2609922:2610264 [6] NCCL INFO Channel 01/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-160-103:2609918:2610256 [2] NCCL INFO Connected all rings
+ip-26-0-160-103:2609923:2610265 [7] NCCL INFO Channel 06/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-160-103:2609922:2610264 [6] NCCL INFO Channel 02/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-160-103:2609922:2610264 [6] NCCL INFO Channel 03/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-160-103:2609922:2610264 [6] NCCL INFO Channel 04/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-160-103:2609922:2610264 [6] NCCL INFO Channel 05/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-160-103:2609916:2610247 [0] NCCL INFO Connected all rings
+ip-26-0-160-103:2609923:2610265 [7] NCCL INFO Channel 07/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-160-103:2609923:2610265 [7] NCCL INFO Channel 08/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-160-103:2609923:2610265 [7] NCCL INFO Channel 09/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-160-103:2609920:2610266 [4] NCCL INFO Connected all rings
+ip-26-0-160-103:2609922:2610264 [6] NCCL INFO Channel 06/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-160-103:2609923:2610265 [7] NCCL INFO Channel 10/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-160-103:2609922:2610264 [6] NCCL INFO Channel 07/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-160-103:2609923:2610265 [7] NCCL INFO Channel 11/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-160-103:2609922:2610264 [6] NCCL INFO Channel 08/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-160-103:2609923:2610265 [7] NCCL INFO Channel 12/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-160-103:2609922:2610264 [6] NCCL INFO Channel 09/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-160-103:2609923:2610265 [7] NCCL INFO Channel 13/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-160-103:2609922:2610264 [6] NCCL INFO Channel 10/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-160-103:2609923:2610265 [7] NCCL INFO Channel 14/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-160-103:2609922:2610264 [6] NCCL INFO Channel 11/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-160-103:2609923:2610265 [7] NCCL INFO Channel 15/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-160-103:2609922:2610264 [6] NCCL INFO Channel 12/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-160-103:2609923:2610265 [7] NCCL INFO Channel 16/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-160-103:2609923:2610265 [7] NCCL INFO Channel 17/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-160-103:2609923:2610265 [7] NCCL INFO Channel 18/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-160-103:2609923:2610265 [7] NCCL INFO Channel 19/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-160-103:2609922:2610264 [6] NCCL INFO Channel 13/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-160-103:2609923:2610265 [7] NCCL INFO Channel 20/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-160-103:2609922:2610264 [6] NCCL INFO Channel 14/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-160-103:2609923:2610265 [7] NCCL INFO Channel 21/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-160-103:2609922:2610264 [6] NCCL INFO Channel 15/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-160-103:2609923:2610265 [7] NCCL INFO Channel 22/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-160-103:2609922:2610264 [6] NCCL INFO Channel 16/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-160-103:2609923:2610265 [7] NCCL INFO Channel 23/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-160-103:2609922:2610264 [6] NCCL INFO Channel 17/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-160-103:2609919:2610257 [3] NCCL INFO Channel 00/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-160-103:2609922:2610264 [6] NCCL INFO Channel 18/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-160-103:2609919:2610257 [3] NCCL INFO Channel 01/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-160-103:2609922:2610264 [6] NCCL INFO Channel 19/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-160-103:2609919:2610257 [3] NCCL INFO Channel 02/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-160-103:2609922:2610264 [6] NCCL INFO Channel 20/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-160-103:2609921:2610267 [5] NCCL INFO Channel 00/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2609919:2610257 [3] NCCL INFO Channel 03/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-160-103:2609922:2610264 [6] NCCL INFO Channel 21/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-160-103:2609921:2610267 [5] NCCL INFO Channel 01/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2609919:2610257 [3] NCCL INFO Channel 04/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-160-103:2609922:2610264 [6] NCCL INFO Channel 22/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-160-103:2609921:2610267 [5] NCCL INFO Channel 02/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2609919:2610257 [3] NCCL INFO Channel 05/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-160-103:2609922:2610264 [6] NCCL INFO Channel 23/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-160-103:2609921:2610267 [5] NCCL INFO Channel 03/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2609919:2610257 [3] NCCL INFO Channel 06/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-160-103:2609921:2610267 [5] NCCL INFO Channel 04/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2609919:2610257 [3] NCCL INFO Channel 07/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-160-103:2609921:2610267 [5] NCCL INFO Channel 05/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2609919:2610257 [3] NCCL INFO Channel 08/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-160-103:2609921:2610267 [5] NCCL INFO Channel 06/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2609919:2610257 [3] NCCL INFO Channel 09/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-160-103:2609921:2610267 [5] NCCL INFO Channel 07/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2609920:2610266 [4] NCCL INFO Channel 00/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-160-103:2609919:2610257 [3] NCCL INFO Channel 10/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-160-103:2609921:2610267 [5] NCCL INFO Channel 08/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2609920:2610266 [4] NCCL INFO Channel 01/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-160-103:2609918:2610256 [2] NCCL INFO Channel 00/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609919:2610257 [3] NCCL INFO Channel 11/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-160-103:2609921:2610267 [5] NCCL INFO Channel 09/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2609920:2610266 [4] NCCL INFO Channel 02/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-160-103:2609918:2610256 [2] NCCL INFO Channel 01/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609919:2610257 [3] NCCL INFO Channel 12/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-160-103:2609921:2610267 [5] NCCL INFO Channel 10/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2609918:2610256 [2] NCCL INFO Channel 02/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609920:2610266 [4] NCCL INFO Channel 03/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-160-103:2609919:2610257 [3] NCCL INFO Channel 13/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-160-103:2609921:2610267 [5] NCCL INFO Channel 11/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2609918:2610256 [2] NCCL INFO Channel 03/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609920:2610266 [4] NCCL INFO Channel 04/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-160-103:2609919:2610257 [3] NCCL INFO Channel 14/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-160-103:2609921:2610267 [5] NCCL INFO Channel 12/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2609918:2610256 [2] NCCL INFO Channel 04/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609920:2610266 [4] NCCL INFO Channel 05/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-160-103:2609919:2610257 [3] NCCL INFO Channel 15/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-160-103:2609921:2610267 [5] NCCL INFO Channel 13/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2609918:2610256 [2] NCCL INFO Channel 05/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609920:2610266 [4] NCCL INFO Channel 06/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-160-103:2609919:2610257 [3] NCCL INFO Channel 16/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-160-103:2609921:2610267 [5] NCCL INFO Channel 14/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2609918:2610256 [2] NCCL INFO Channel 06/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609920:2610266 [4] NCCL INFO Channel 07/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-160-103:2609919:2610257 [3] NCCL INFO Channel 17/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-160-103:2609921:2610267 [5] NCCL INFO Channel 15/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2609918:2610256 [2] NCCL INFO Channel 07/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609920:2610266 [4] NCCL INFO Channel 08/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-160-103:2609919:2610257 [3] NCCL INFO Channel 18/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-160-103:2609921:2610267 [5] NCCL INFO Channel 16/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2609918:2610256 [2] NCCL INFO Channel 08/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609919:2610257 [3] NCCL INFO Channel 19/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-160-103:2609920:2610266 [4] NCCL INFO Channel 09/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-160-103:2609921:2610267 [5] NCCL INFO Channel 17/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2609919:2610257 [3] NCCL INFO Channel 20/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-160-103:2609920:2610266 [4] NCCL INFO Channel 10/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-160-103:2609918:2610256 [2] NCCL INFO Channel 09/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609921:2610267 [5] NCCL INFO Channel 18/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2609918:2610256 [2] NCCL INFO Channel 10/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609920:2610266 [4] NCCL INFO Channel 11/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-160-103:2609919:2610257 [3] NCCL INFO Channel 21/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-160-103:2609921:2610267 [5] NCCL INFO Channel 19/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2609918:2610256 [2] NCCL INFO Channel 11/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609919:2610257 [3] NCCL INFO Channel 22/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-160-103:2609920:2610266 [4] NCCL INFO Channel 12/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-160-103:2609921:2610267 [5] NCCL INFO Channel 20/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2609918:2610256 [2] NCCL INFO Channel 12/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609920:2610266 [4] NCCL INFO Channel 13/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-160-103:2609919:2610257 [3] NCCL INFO Channel 23/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-160-103:2609921:2610267 [5] NCCL INFO Channel 21/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2609918:2610256 [2] NCCL INFO Channel 13/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609920:2610266 [4] NCCL INFO Channel 14/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-160-103:2609921:2610267 [5] NCCL INFO Channel 22/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2609918:2610256 [2] NCCL INFO Channel 14/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609920:2610266 [4] NCCL INFO Channel 15/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-160-103:2609921:2610267 [5] NCCL INFO Channel 23/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-160-103:2609918:2610256 [2] NCCL INFO Channel 15/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609920:2610266 [4] NCCL INFO Channel 16/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-160-103:2609918:2610256 [2] NCCL INFO Channel 16/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609920:2610266 [4] NCCL INFO Channel 17/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-160-103:2609920:2610266 [4] NCCL INFO Channel 18/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-160-103:2609918:2610256 [2] NCCL INFO Channel 17/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609920:2610266 [4] NCCL INFO Channel 19/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-160-103:2609918:2610256 [2] NCCL INFO Channel 18/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609920:2610266 [4] NCCL INFO Channel 20/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-160-103:2609918:2610256 [2] NCCL INFO Channel 19/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609920:2610266 [4] NCCL INFO Channel 21/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-160-103:2609918:2610256 [2] NCCL INFO Channel 20/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609920:2610266 [4] NCCL INFO Channel 22/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-160-103:2609918:2610256 [2] NCCL INFO Channel 21/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609920:2610266 [4] NCCL INFO Channel 23/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-160-103:2609918:2610256 [2] NCCL INFO Channel 22/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609918:2610256 [2] NCCL INFO Channel 23/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-160-103:2609923:2610265 [7] NCCL INFO Connected all trees
+ip-26-0-160-103:2609923:2610265 [7] NCCL INFO NVLS comm 0xc284020 headRank 3 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-160-103:2609917:2610249 [1] NCCL INFO Connected all trees
+ip-26-0-160-103:2609917:2610249 [1] NCCL INFO NVLS comm 0xafc0a30 headRank 0 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-160-103:2609922:2610264 [6] NCCL INFO Connected all trees
+ip-26-0-160-103:2609922:2610264 [6] NCCL INFO NVLS comm 0xb6b3550 headRank 3 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-160-103:2609916:2610247 [0] NCCL INFO Connected all trees
+ip-26-0-160-103:2609916:2610247 [0] NCCL INFO NVLS comm 0xb9cd580 headRank 0 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-160-103:2609919:2610257 [3] NCCL INFO Connected all trees
+ip-26-0-160-103:2609919:2610257 [3] NCCL INFO NVLS comm 0xa535e00 headRank 1 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-160-103:2609921:2610267 [5] NCCL INFO Connected all trees
+ip-26-0-160-103:2609921:2610267 [5] NCCL INFO NVLS comm 0xadd0310 headRank 2 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-160-103:2609918:2610256 [2] NCCL INFO Connected all trees
+ip-26-0-160-103:2609918:2610256 [2] NCCL INFO NVLS comm 0xc18ede0 headRank 1 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-160-103:2609923:2610265 [7] NCCL INFO Connected NVLS tree
+ip-26-0-160-103:2609923:2610265 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-103:2609923:2610265 [7] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-160-103:2609921:2610267 [5] NCCL INFO Connected NVLS tree
+ip-26-0-160-103:2609923:2610265 [7] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-160-103:2609921:2610267 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-103:2609921:2610267 [5] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-160-103:2609921:2610267 [5] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-160-103:2609919:2610257 [3] NCCL INFO Connected NVLS tree
+ip-26-0-160-103:2609919:2610257 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-103:2609919:2610257 [3] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-160-103:2609919:2610257 [3] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-160-103:2609917:2610249 [1] NCCL INFO Connected NVLS tree
+ip-26-0-160-103:2609917:2610249 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-103:2609917:2610249 [1] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-160-103:2609917:2610249 [1] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-160-103:2609920:2610266 [4] NCCL INFO Connected all trees
+ip-26-0-160-103:2609920:2610266 [4] NCCL INFO NVLS comm 0xb168a60 headRank 2 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-160-103:2609919:2610257 [3] NCCL INFO comm 0xa535e00 rank 1 nranks 4 cudaDev 3 nvmlDev 3 busId 86000 commId 0x35779fcbea0678c9 - Init COMPLETE
+ip-26-0-160-103:2609917:2610249 [1] NCCL INFO comm 0xafc0a30 rank 0 nranks 4 cudaDev 1 nvmlDev 1 busId 64000 commId 0x35779fcbea0678c9 - Init COMPLETE
+ip-26-0-160-103:2609921:2610267 [5] NCCL INFO comm 0xadd0310 rank 2 nranks 4 cudaDev 5 nvmlDev 5 busId a8000 commId 0x35779fcbea0678c9 - Init COMPLETE
+ip-26-0-160-103:2609923:2610265 [7] NCCL INFO comm 0xc284020 rank 3 nranks 4 cudaDev 7 nvmlDev 7 busId ca000 commId 0x35779fcbea0678c9 - Init COMPLETE
+ip-26-0-160-103:2609916:2610247 [0] NCCL INFO Connected NVLS tree
+ip-26-0-160-103:2609916:2610247 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-103:2609916:2610247 [0] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-160-103:2609916:2610247 [0] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-160-103:2609918:2610256 [2] NCCL INFO Connected NVLS tree
+ip-26-0-160-103:2609918:2610256 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-103:2609918:2610256 [2] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-160-103:2609918:2610256 [2] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-160-103:2609922:2610264 [6] NCCL INFO Connected NVLS tree
+ip-26-0-160-103:2609922:2610264 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-103:2609922:2610264 [6] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-160-103:2609922:2610264 [6] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-160-103:2609920:2610266 [4] NCCL INFO Connected NVLS tree
+ip-26-0-160-103:2609920:2610266 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-103:2609920:2610266 [4] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-160-103:2609920:2610266 [4] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-160-103:2609922:2610264 [6] NCCL INFO comm 0xb6b3550 rank 3 nranks 4 cudaDev 6 nvmlDev 6 busId b9000 commId 0xe076a754049e1216 - Init COMPLETE
+ip-26-0-160-103:2609918:2610256 [2] NCCL INFO comm 0xc18ede0 rank 1 nranks 4 cudaDev 2 nvmlDev 2 busId 75000 commId 0xe076a754049e1216 - Init COMPLETE
+ip-26-0-160-103:2609920:2610266 [4] NCCL INFO comm 0xb168a60 rank 2 nranks 4 cudaDev 4 nvmlDev 4 busId 97000 commId 0xe076a754049e1216 - Init COMPLETE
+ip-26-0-160-103:2609916:2610247 [0] NCCL INFO comm 0xb9cd580 rank 0 nranks 4 cudaDev 0 nvmlDev 0 busId 53000 commId 0xe076a754049e1216 - Init COMPLETE
+12/13/2024 14:17:28 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]: [Optimizer Building] Using LearningRateForSP as learning rate
+12/13/2024 14:17:28 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]: [ZeRO sharding] Size of optimizer params per rank:
+12/13/2024 14:17:28 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]: [ZeRO sharding] DP Rank 0 has 130M out of 520M (25.00%) params' optimizer states
+12/13/2024 14:17:28 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]: [ZeRO sharding] DP Rank 1 has 130M out of 520M (25.00%) params' optimizer states
+12/13/2024 14:17:28 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]: [ZeRO sharding] DP Rank 2 has 130M out of 520M (25.00%) params' optimizer states
+12/13/2024 14:17:28 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]: [ZeRO sharding] DP Rank 3 has 130M out of 520M (25.00%) params' optimizer states
+12/13/2024 14:17:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]: [Training Plan] Stage Stable Training Stage has 99 remaining training steps and has consumed 0 samples
+12/13/2024 14:17:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]: Using dummy data generator
+12/13/2024 14:17:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]: [Training Plan] There are 1 training stages 
+12/13/2024 14:17:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]: [Stage Stable Training Stage] start from step 1 
+12/13/2024 14:17:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]: 
+12/13/2024 14:17:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]: [Start training] datetime: 2024-12-13 14:17:29.176225 | mbs: 2 | grad_accum: 256 | global_batch_size: 2048 | sequence_length: 2048 | train_steps: 100 | start_iteration_step: 0 | consumed_train_samples: 0
+wandb: WARNING Path /fsx/nouamane/.cache/wandb/wandb/ wasn't writable, using system temp directory.
+wandb: WARNING Path /fsx/nouamane/.cache/wandb/wandb/ wasn't writable, using system temp directory
+wandb: Tracking run with wandb version 0.16.0
+wandb: W&B syncing is set to `offline` in this directory.  
+wandb: Run `wandb online` or set WANDB_MODE=online to enable cloud syncing.
+12/13/2024 14:17:37 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]: Waiting for 13 seconds
+12/13/2024 14:17:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]: Resuming training from stage Stable Training Stage, it has trained for 0 samples and has 99 remaining train steps
+12/13/2024 14:17:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:  Memory usage: 3472.48MiB. Peak allocated 5632.00MiB. Peak reserved: 20932.00MiB
+12/13/2024 14:18:11 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:  Memory usage: 3572.67MiB. Peak allocated 6985.16MiB. Peak reserved: 7434.00MiB
+12/13/2024 14:18:12 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]: iteration: 1 / 100 | consumed_tokens: 4.19M | elapsed_time_per_iteration_ms: 21.3K | tokens_per_sec: 197K | tokens_per_sec_per_gpu: 24.7K | global_batch_size: 2.05K | lm_loss: 10.8 | lr: 0.00015 | model_tflops_per_gpu: 174 | hardware_tflops_per_gpu: 174 | grad_norm: 0.431 | cuda_memory_allocated: 4.81G | cuda_max_memory_reserved: 9.62G | hd_total_memory_tb: 312G | hd_used_memory_tb: 69.5G | hd_free_memory_tb: 243G
+12/13/2024 14:18:12 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:  Memory usage: 4587.85MiB. Peak allocated 6820.14MiB. Peak reserved: 9178.00MiB
+12/13/2024 14:18:30 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:  Memory usage: 4587.97MiB. Peak allocated 8000.47MiB. Peak reserved: 9434.00MiB
+12/13/2024 14:18:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]: iteration: 2 / 100 | consumed_tokens: 8.39M | elapsed_time_per_iteration_ms: 20K | tokens_per_sec: 210K | tokens_per_sec_per_gpu: 26.3K | global_batch_size: 2.05K | lm_loss: 10.8 | lr: 0.0003 | model_tflops_per_gpu: 185 | hardware_tflops_per_gpu: 185 | grad_norm: 0.431 | cuda_memory_allocated: 4.81G | cuda_max_memory_reserved: 10.9G | hd_total_memory_tb: 312G | hd_used_memory_tb: 69.5G | hd_free_memory_tb: 243G
+12/13/2024 14:18:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:  Memory usage: 4587.85MiB. Peak allocated 6820.26MiB. Peak reserved: 10428.00MiB
+12/13/2024 14:18:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:  Memory usage: 4587.97MiB. Peak allocated 8000.47MiB. Peak reserved: 10428.00MiB
+12/13/2024 14:18:52 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]: iteration: 3 / 100 | consumed_tokens: 12.6M | elapsed_time_per_iteration_ms: 20.2K | tokens_per_sec: 208K | tokens_per_sec_per_gpu: 25.9K | global_batch_size: 2.05K | lm_loss: 10.8 | lr: 0.000296 | model_tflops_per_gpu: 183 | hardware_tflops_per_gpu: 183 | grad_norm: 0.424 | cuda_memory_allocated: 4.81G | cuda_max_memory_reserved: 10.9G | hd_total_memory_tb: 312G | hd_used_memory_tb: 69.5G | hd_free_memory_tb: 243G
+12/13/2024 14:18:52 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]: | job_id   | name                                                           | nodes | seq_len | mbs | batch_accum | gbs  | mTFLOPs | hTFLOPs | tok/s/gpu | AllReduce (GB/s) | AllGather (GB/s) | ReduceScatter (GB/s) | AR Intra-node (GB/s) | AG Intra-node (GB/s) | RS Intra-node (GB/s) | Mem Alloc (GB) | Mem Res (GB) | dp | pp | tp | pp_engine                                                                                                 | tp_mode                                 | tp_async_comm | hidden_size | hidden_act | num_layers | num_heads | num_kv_heads | max_pos | vocab_size | tie_word_embeddings | dtype          | zero_stage | ddp_bucket_cap_mb | accumulate_grad_in_fp32 | Total Params | Local Params |
+12/13/2024 14:18:52 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]: | -------- | -------------------------------------------------------------- | ----- | ------- | --- | ----------- | ---- | ------- | ------- | --------- | ---------------- | ---------------- | -------------------- | -------------------- | -------------------- | -------------------- | -------------- | ------------ | -- | -- | -- | --------------------------------------------------------------------------------------------------------- | --------------------------------------- | ------------- | ----------- | ---------- | ---------- | --------- | ------------ | ------- | ---------- | ------------------- | -------------- | ---------- | ----------------- | ----------------------- | ------------ | ------------ |
+12/13/2024 14:18:52 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]: | 13417877 | 1.14G_dp4_tp2_pp1_acc256_mbs2_seq2048_zero1_tpmodeRED_vocab32k | 1     | 2048    | 2   | 256         | 2048 | 182.78  | 182.78  | 25939.22  | 460.39           | 263.53           | 262.89               | 455.96               | 264.98               | 264.38               | 6.66           | 10.18        | 4  | 1  | 2  | <nanotron.parallel.pipeline_parallel.engine.OneForwardOneBackwardPipelineEngine object at 0x7facd0d48d00> | TensorParallelLinearMode.REDUCE_SCATTER | True          | 2048        | silu       | 16         | 32        | 8            | 2048    | 32768      | True                | torch.bfloat16 | 1          | 25                | True                    | 1.04G        | 520M         |
+12/13/2024 14:18:52 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]: Attempting to write benchmark results to CSV file: benchmark/results/bench_final.csv
+12/13/2024 14:18:52 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]: Acquired lock for CSV file: benchmark/results/bench_final.csv
+slurmstepd: error: *** JOB 13417877 ON ip-26-0-160-103 CANCELLED AT 2024-12-13T14:18:52 DUE TO TIME LIMIT ***
+srun: Job step aborted: Waiting up to 32 seconds for job step to finish.
+slurmstepd: error: *** STEP 13417877.0 ON ip-26-0-160-103 CANCELLED AT 2024-12-13T14:18:52 DUE TO TIME LIMIT ***
+[2024-12-13 14:18:52,745] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-13 14:18:52,746] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2609916 closing signal SIGTERM
+[2024-12-13 14:18:52,746] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2609917 closing signal SIGTERM
+[2024-12-13 14:18:52,746] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2609918 closing signal SIGTERM
+[2024-12-13 14:18:52,746] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2609919 closing signal SIGTERM
+[2024-12-13 14:18:52,746] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2609920 closing signal SIGTERM
+[2024-12-13 14:18:52,749] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2609921 closing signal SIGTERM
+[2024-12-13 14:18:52,749] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2609922 closing signal SIGTERM
+[2024-12-13 14:18:52,749] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2609923 closing signal SIGTERM
+[2024-12-13 14:18:52,864] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2609916 closing signal SIGTERM
+[2024-12-13 14:18:52,865] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2609917 closing signal SIGTERM
+[2024-12-13 14:18:52,865] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2609918 closing signal SIGTERM
+[2024-12-13 14:18:52,865] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2609919 closing signal SIGTERM
+[2024-12-13 14:18:52,865] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2609920 closing signal SIGTERM
+[2024-12-13 14:18:52,865] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2609921 closing signal SIGTERM
+[2024-12-13 14:18:52,865] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2609922 closing signal SIGTERM
+[2024-12-13 14:18:52,865] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2609923 closing signal SIGTERM
diff --git a/logs/13417898-bench_stress_test.out b/logs/13417898-bench_stress_test.out
new file mode 100644
index 0000000000000000000000000000000000000000..996566c82119e21ca596522443fe5e7c0b5d5cec
--- /dev/null
+++ b/logs/13417898-bench_stress_test.out
@@ -0,0 +1,3775 @@
++ source /etc/profile.d/modules.sh
+++ . /usr/share/modules/init/bash
++++ unset _mlshdbg
++++ '[' 0 = 1 ']'
++++ unset _mlre _mlIFS
++++ '[' -n x ']'
++++ _mlIFS=' 	
+'
++++ IFS=' '
++++ '[' -n '' ']'
+++++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash autoinit
++++ _mlcode='module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+MODULEPATH_modshare=/etc/environment-modules/modules:1:/usr/share/modules/\$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1; export MODULEPATH_modshare;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+LOADEDMODULES=; export LOADEDMODULES;
+MODULEPATH=/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/\$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles; export MODULEPATH;
+test 0;'
++++ _mlret=0
++++ '[' -n x ']'
++++ IFS=' 	
+'
++++ unset _mlIFS
++++ unset _mlre _mlv _mlrv
++++ '[' 0 -eq 0 ']'
++++ eval 'module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+MODULEPATH_modshare=/etc/environment-modules/modules:1:/usr/share/modules/\$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1; export MODULEPATH_modshare;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+LOADEDMODULES=; export LOADEDMODULES;
+MODULEPATH=/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/\$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles; export MODULEPATH;
+test 0;'
+++++ MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl
+++++ export MODULES_CMD
+++++ ENV=/usr/share/modules/init/profile.sh
+++++ export ENV
+++++ MODULEPATH_modshare='/etc/environment-modules/modules:1:/usr/share/modules/$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1'
+++++ export MODULEPATH_modshare
+++++ BASH_ENV=/usr/share/modules/init/bash
+++++ export BASH_ENV
+++++ MODULESHOME=/usr/share/modules
+++++ export MODULESHOME
+++++ LOADEDMODULES=
+++++ export LOADEDMODULES
+++++ MODULEPATH='/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles'
+++++ export MODULEPATH
+++++ test 0
++++ '[' 0 = 1 ']'
++++ '[' -t 2 ']'
++++ export -f module
++++ export -f switchml
++++ '[' 5 -ge 3 ']'
++++ [[ ehxB =~ i ]]
++++ [[ ! :/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin: =~ :/usr/bin: ]]
++++ '[' '!' -n '' ']'
++++ MANPATH=:
++++ export MANPATH
+++++ manpath
++++ [[ ! :/admin/home/nouamane/.local/share/man:/fsx/nouamane/miniconda/envs/2-1-cu121/man:/fsx/nouamane/miniconda/envs/2-1-cu121/share/man:/fsx/nouamane/miniconda/man:/fsx/nouamane/miniconda/share/man:/opt/amazon/openmpi/share/man:/opt/amazon/efa/share/man:/opt/slurm/share/man:/usr/local/man:/usr/local/share/man:/usr/share/man:/admin/home/nouamane/.fzf/man:: =~ :/usr/share/man: ]]
++++ unset _mlcode _mlret
++++ '[' -n '' ']'
++ module load cuda/12.1
++ unset _mlshdbg
++ '[' 0 = 1 ']'
++ unset _mlre _mlIFS
++ '[' -n x ']'
++ _mlIFS=' 	
+'
++ IFS=' '
++ '[' -n '' ']'
+++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash load cuda/12.1
++ eval 'CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include;' export 'CPATH;
+LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:;' export 'LD_LIBRARY_PATH;
+FI_EFA_FORK_SAFE=1;' export 'FI_EFA_FORK_SAFE;
+MANPATH=/usr/local/cuda-12.1/share/man::;' export 'MANPATH;
+LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64;' export 'LIBRARY_PATH;
+_LMFILES_=/usr/share/modules/modulefiles/cuda/12.1;' export '_LMFILES_;
+LOADEDMODULES=cuda/12.1;' export 'LOADEDMODULES;
+MPI_PATH=/opt/amazon/openmpi;' export 'MPI_PATH;
+NCCL_HOME_modshare=/opt/nccl/build:1;' export 'NCCL_HOME_modshare;
+NCCL_PROTO=simple;' export 'NCCL_PROTO;
+MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1;' export 'MANPATH_modshare;
+LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1;' export 'LIBRARY_PATH_modshare;
+NCCL_SOCKET_IFNAME=enp;' export 'NCCL_SOCKET_IFNAME;
+AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl;' export 'AWS_OFI_NCCL_HOME;
+NCCL_HOME=/opt/nccl/build;' export 'NCCL_HOME;
+FI_PROVIDER=efa;' export 'FI_PROVIDER;
+AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1;' export 'AWS_OFI_NCCL_HOME_modshare;
+CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1;' export 'CPATH_modshare;
+LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2::1:/opt/nccl/build/lib:1:/opt/aws-ofi-nccl/lib:2:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/efa/lib:1;' export 'LD_LIBRARY_PATH_modshare;
+FI_EFA_ENABLE_SHM_TRANSFER=1;' export 'FI_EFA_ENABLE_SHM_TRANSFER;
+_LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1;' export '_LMFILES__modshare;
+LOADEDMODULES_modshare=cuda/12.1:1;' export 'LOADEDMODULES_modshare;
+MPI_PATH_modshare=/opt/amazon/openmpi:1;' export 'MPI_PATH_modshare;
+PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin;' export 'PATH;
+CUDA_HOME=/usr/local/cuda-12.1;' export 'CUDA_HOME;
+PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/usr/local/cuda-12.1/include:1:/opt/amazon/efa/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/usr/sbin:1:/fsx/nouamane/miniconda/bin:1:/opt/slurm/bin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:2:/opt/amazon/openmpi/bin:1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1;' export 'PATH_modshare;
+test' '0;'
+++ CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include
+++ export CPATH
+++ LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:
+++ export LD_LIBRARY_PATH
+++ FI_EFA_FORK_SAFE=1
+++ export FI_EFA_FORK_SAFE
+++ MANPATH=/usr/local/cuda-12.1/share/man::
+++ export MANPATH
+++ LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64
+++ export LIBRARY_PATH
+++ _LMFILES_=/usr/share/modules/modulefiles/cuda/12.1
+++ export _LMFILES_
+++ LOADEDMODULES=cuda/12.1
+++ export LOADEDMODULES
+++ MPI_PATH=/opt/amazon/openmpi
+++ export MPI_PATH
+++ NCCL_HOME_modshare=/opt/nccl/build:1
+++ export NCCL_HOME_modshare
+++ NCCL_PROTO=simple
+++ export NCCL_PROTO
+++ MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1
+++ export MANPATH_modshare
+++ LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1
+++ export LIBRARY_PATH_modshare
+++ NCCL_SOCKET_IFNAME=enp
+++ export NCCL_SOCKET_IFNAME
+++ AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl
+++ export AWS_OFI_NCCL_HOME
+++ NCCL_HOME=/opt/nccl/build
+++ export NCCL_HOME
+++ FI_PROVIDER=efa
+++ export FI_PROVIDER
+++ AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1
+++ export AWS_OFI_NCCL_HOME_modshare
+++ CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1
+++ export CPATH_modshare
+++ LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2::1:/opt/nccl/build/lib:1:/opt/aws-ofi-nccl/lib:2:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/efa/lib:1
+++ export LD_LIBRARY_PATH_modshare
+++ FI_EFA_ENABLE_SHM_TRANSFER=1
+++ export FI_EFA_ENABLE_SHM_TRANSFER
+++ _LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1
+++ export _LMFILES__modshare
+++ LOADEDMODULES_modshare=cuda/12.1:1
+++ export LOADEDMODULES_modshare
+++ MPI_PATH_modshare=/opt/amazon/openmpi:1
+++ export MPI_PATH_modshare
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ export PATH
+++ CUDA_HOME=/usr/local/cuda-12.1
+++ export CUDA_HOME
+++ PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/usr/local/cuda-12.1/include:1:/opt/amazon/efa/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/usr/sbin:1:/fsx/nouamane/miniconda/bin:1:/opt/slurm/bin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:2:/opt/amazon/openmpi/bin:1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1
+++ export PATH_modshare
+++ test 0
++ _mlstatus=0
++ '[' -n x ']'
++ IFS=' 	
+'
++ unset _mlre _mlv _mlrv _mlIFS
++ '[' -n '' ']'
++ unset _mlshdbg
++ return 0
++ source /fsx/nouamane/miniconda/bin/activate
+++ _CONDA_ROOT=/fsx/nouamane/miniconda
+++ . /fsx/nouamane/miniconda/etc/profile.d/conda.sh
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ '[' -z x ']'
+++ conda activate
+++ local cmd=activate
+++ case "$cmd" in
+++ __conda_activate activate
+++ '[' -n '' ']'
+++ local ask_conda
++++ PS1=
++++ __conda_exe shell.posix activate
++++ /fsx/nouamane/miniconda/bin/conda shell.posix activate
+++ ask_conda='. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_2='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\'''
+++ eval '. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_2='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\'''
++++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh
+++++ test -n ''
+++++ unset XML_CATALOG_FILES
+++++ unset xml_catalog_files_libxml2
++++ PS1='(base) '
++++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++++ export CONDA_PREFIX=/fsx/nouamane/miniconda
++++ CONDA_PREFIX=/fsx/nouamane/miniconda
++++ export CONDA_SHLVL=3
++++ CONDA_SHLVL=3
++++ export CONDA_DEFAULT_ENV=base
++++ CONDA_DEFAULT_ENV=base
++++ export 'CONDA_PROMPT_MODIFIER=(base) '
++++ CONDA_PROMPT_MODIFIER='(base) '
++++ export CONDA_PREFIX_2=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ CONDA_PREFIX_2=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ __conda_hashr
+++ '[' -n '' ']'
+++ '[' -n '' ']'
+++ hash -r
++ conda activate 2-1-cu121
++ local cmd=activate
++ case "$cmd" in
++ __conda_activate activate 2-1-cu121
++ '[' -n '' ']'
++ local ask_conda
+++ PS1='(base) '
+++ __conda_exe shell.posix activate 2-1-cu121
+++ /fsx/nouamane/miniconda/bin/conda shell.posix activate 2-1-cu121
++ ask_conda='PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''4'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_3='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
++ eval 'PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''4'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_3='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
+++ PS1='(2-1-cu121) '
+++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ export CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ export CONDA_SHLVL=4
+++ CONDA_SHLVL=4
+++ export CONDA_DEFAULT_ENV=2-1-cu121
+++ CONDA_DEFAULT_ENV=2-1-cu121
+++ export 'CONDA_PROMPT_MODIFIER=(2-1-cu121) '
+++ CONDA_PROMPT_MODIFIER='(2-1-cu121) '
+++ export CONDA_PREFIX_3=/fsx/nouamane/miniconda
+++ CONDA_PREFIX_3=/fsx/nouamane/miniconda
+++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ export _CE_M=
+++ _CE_M=
+++ export _CE_CONDA=
+++ _CE_CONDA=
+++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh
++++ test -n ''
++++ xml_catalog_files_libxml2=
++++ XML_CATALOG_FILES=
++++ conda_catalog_files=
++++ ifs_libxml2=' 	
+'
++++ IFS=' '
++++ rem=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ for pre in ${rem}
++++ test '' = /fsx/nouamane/miniconda/envs/2-1-cu121
++++ conda_catalog_files=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ rem=
++++ IFS=' 	
+'
++++ conda_catalog_files='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ export 'XML_CATALOG_FILES=file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ XML_CATALOG_FILES='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ unset conda_catalog_files ifs_libxml2 rem
++ __conda_hashr
++ '[' -n '' ']'
++ '[' -n '' ']'
++ hash -r
++ export PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++ PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ scontrol show hostnames ip-26-0-169-207
++ export NODELIST=ip-26-0-169-207
++ NODELIST=ip-26-0-169-207
+++ scontrol show hostnames ip-26-0-169-207
+++ head -n1
++ export MASTER_NODE=ip-26-0-169-207
++ MASTER_NODE=ip-26-0-169-207
++ export MASTER_PORT=12356
++ MASTER_PORT=12356
++ export NNODES=1
++ NNODES=1
++ export GPUS_PER_NODE=8
++ GPUS_PER_NODE=8
++ export WORLD_SIZE=8
++ WORLD_SIZE=8
++ export CUDA_DEVICE_MAX_CONNECTIONS=1
++ CUDA_DEVICE_MAX_CONNECTIONS=1
++ export NCCL_DEBUG=INFO
++ NCCL_DEBUG=INFO
++ export NANOTRON_BENCHMARK=1
++ NANOTRON_BENCHMARK=1
++ echo '=== GPU Topology ==='
+=== GPU Topology ===
++ nvidia-smi topo -m
+	GPU0	GPU1	GPU2	GPU3	GPU4	GPU5	GPU6	GPU7	CPU Affinity	NUMA Affinity	GPU NUMA ID
+GPU0	 X 	NV18	NV18	NV18	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU1	NV18	 X 	NV18	NV18	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU2	NV18	NV18	 X 	NV18	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU3	NV18	NV18	NV18	 X 	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU4	NV18	NV18	NV18	NV18	 X 	NV18	NV18	NV18	48-95	1		N/A
+GPU5	NV18	NV18	NV18	NV18	NV18	 X 	NV18	NV18	48-95	1		N/A
+GPU6	NV18	NV18	NV18	NV18	NV18	NV18	 X 	NV18	48-95	1		N/A
+GPU7	NV18	NV18	NV18	NV18	NV18	NV18	NV18	 X 	48-95	1		N/A
+
+Legend:
+
+  X    = Self
+  SYS  = Connection traversing PCIe as well as the SMP interconnect between NUMA nodes (e.g., QPI/UPI)
+  NODE = Connection traversing PCIe as well as the interconnect between PCIe Host Bridges within a NUMA node
+  PHB  = Connection traversing PCIe as well as a PCIe Host Bridge (typically the CPU)
+  PXB  = Connection traversing multiple PCIe bridges (without traversing the PCIe Host Bridge)
+  PIX  = Connection traversing at most a single PCIe bridge
+  NV#  = Connection traversing a bonded set of # NVLinks
++ echo ==================
+==================
++ echo 'Master node: ip-26-0-169-207'
+Master node: ip-26-0-169-207
++ echo 'All nodes: ip-26-0-169-207'
+All nodes: ip-26-0-169-207
++ echo 'World size: 8'
+World size: 8
++ srun torchrun --nnodes=1 --nproc_per_node=8 --rdzv_id=13417898 --rdzv_backend=c10d --rdzv_endpoint=ip-26-0-169-207:12356 /fsx/nouamane/projects/nanotron/run_train.py --config-file benchmark/configs/config_1.14G_dp4_tp2_pp1_acc256_mbs2_seq2048_zero1_tpmodeRED_vocab32k.yaml
+[2024-12-13 14:41:56,149] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-13 14:41:56,149] torch.distributed.run: [WARNING] 
+[2024-12-13 14:41:56,149] torch.distributed.run: [WARNING] *****************************************
+[2024-12-13 14:41:56,149] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-13 14:41:56,149] torch.distributed.run: [WARNING] *****************************************
+ip-26-0-169-207:734075:734075 [0] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-169-207:734075:734075 [0] NCCL INFO Bootstrap : Using enp71s0:26.0.169.207<0>
+ip-26-0-169-207:734075:734075 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-169-207:734075:734075 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-169-207:734075:734075 [0] NCCL INFO cudaDriverVersion 12020
+NCCL version 2.18.5+cuda12.2
+ip-26-0-169-207:734081:734081 [6] NCCL INFO cudaDriverVersion 12020
+ip-26-0-169-207:734076:734076 [1] NCCL INFO cudaDriverVersion 12020
+ip-26-0-169-207:734081:734081 [6] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-169-207:734078:734078 [3] NCCL INFO cudaDriverVersion 12020
+ip-26-0-169-207:734077:734077 [2] NCCL INFO cudaDriverVersion 12020
+ip-26-0-169-207:734079:734079 [4] NCCL INFO cudaDriverVersion 12020
+ip-26-0-169-207:734076:734076 [1] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-169-207:734078:734078 [3] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-169-207:734077:734077 [2] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-169-207:734079:734079 [4] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-169-207:734082:734082 [7] NCCL INFO cudaDriverVersion 12020
+ip-26-0-169-207:734082:734082 [7] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-169-207:734080:734080 [5] NCCL INFO cudaDriverVersion 12020
+ip-26-0-169-207:734080:734080 [5] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-169-207:734081:734081 [6] NCCL INFO Bootstrap : Using enp71s0:26.0.169.207<0>
+ip-26-0-169-207:734081:734081 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-169-207:734081:734081 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-169-207:734076:734076 [1] NCCL INFO Bootstrap : Using enp71s0:26.0.169.207<0>
+ip-26-0-169-207:734078:734078 [3] NCCL INFO Bootstrap : Using enp71s0:26.0.169.207<0>
+ip-26-0-169-207:734077:734077 [2] NCCL INFO Bootstrap : Using enp71s0:26.0.169.207<0>
+ip-26-0-169-207:734079:734079 [4] NCCL INFO Bootstrap : Using enp71s0:26.0.169.207<0>
+ip-26-0-169-207:734082:734082 [7] NCCL INFO Bootstrap : Using enp71s0:26.0.169.207<0>
+ip-26-0-169-207:734080:734080 [5] NCCL INFO Bootstrap : Using enp71s0:26.0.169.207<0>
+ip-26-0-169-207:734076:734076 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-169-207:734076:734076 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-169-207:734079:734079 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-169-207:734079:734079 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-169-207:734080:734080 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-169-207:734080:734080 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-169-207:734078:734078 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-169-207:734077:734077 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-169-207:734082:734082 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-169-207:734078:734078 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-169-207:734082:734082 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-169-207:734077:734077 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-169-207:734075:734197 [0] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-169-207:734075:734197 [0] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-169-207:734075:734197 [0] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-169-207:734075:734197 [0] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-169-207:734075:734197 [0] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-169-207:734075:734197 [0] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-169-207:734075:734197 [0] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-169-207:734075:734197 [0] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-169-207:734075:734197 [0] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-169-207:734080:734204 [5] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-169-207:734080:734204 [5] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-169-207:734080:734204 [5] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-169-207:734080:734204 [5] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-169-207:734080:734204 [5] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-169-207:734080:734204 [5] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-169-207:734080:734204 [5] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-169-207:734080:734204 [5] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-169-207:734080:734204 [5] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-169-207:734081:734198 [6] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-169-207:734081:734198 [6] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-169-207:734081:734198 [6] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-169-207:734081:734198 [6] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-169-207:734081:734198 [6] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-169-207:734081:734198 [6] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-169-207:734081:734198 [6] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-169-207:734081:734198 [6] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-169-207:734081:734198 [6] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-169-207:734079:734201 [4] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-169-207:734079:734201 [4] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-169-207:734079:734201 [4] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-169-207:734079:734201 [4] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-169-207:734079:734201 [4] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-169-207:734079:734201 [4] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-169-207:734079:734201 [4] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-169-207:734082:734203 [7] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-169-207:734079:734201 [4] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-169-207:734079:734201 [4] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-169-207:734082:734203 [7] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-169-207:734077:734200 [2] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-169-207:734077:734200 [2] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-169-207:734078:734202 [3] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-169-207:734078:734202 [3] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-169-207:734075:734197 [0] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-169-207:734077:734200 [2] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-169-207:734077:734200 [2] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-169-207:734076:734199 [1] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-169-207:734076:734199 [1] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-169-207:734077:734200 [2] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-169-207:734077:734200 [2] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-169-207:734077:734200 [2] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-169-207:734077:734200 [2] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-169-207:734077:734200 [2] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-169-207:734082:734203 [7] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-169-207:734082:734203 [7] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-169-207:734082:734203 [7] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-169-207:734082:734203 [7] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-169-207:734082:734203 [7] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-169-207:734082:734203 [7] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-169-207:734082:734203 [7] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-169-207:734078:734202 [3] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-169-207:734078:734202 [3] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-169-207:734078:734202 [3] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-169-207:734078:734202 [3] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-169-207:734078:734202 [3] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-169-207:734078:734202 [3] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-169-207:734078:734202 [3] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-169-207:734076:734199 [1] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-169-207:734076:734199 [1] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-169-207:734076:734199 [1] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-169-207:734076:734199 [1] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-169-207:734076:734199 [1] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-169-207:734076:734199 [1] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-169-207:734076:734199 [1] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-169-207:734081:734198 [6] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-169-207:734080:734204 [5] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-169-207:734079:734201 [4] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-169-207:734075:734197 [0] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-169-207:734075:734197 [0] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-169-207:734075:734197 [0] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-169-207:734075:734197 [0] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-169-207:734075:734197 [0] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-169-207:734075:734197 [0] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-169-207:734075:734197 [0] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-169-207:734075:734197 [0] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-169-207:734075:734197 [0] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-169-207:734075:734197 [0] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-169-207:734075:734197 [0] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-169-207:734075:734197 [0] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-169-207:734075:734197 [0] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-169-207:734075:734197 [0] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-169-207:734075:734197 [0] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-169-207:734075:734197 [0] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-169-207:734075:734197 [0] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-169-207:734075:734197 [0] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-169-207:734075:734197 [0] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-169-207:734075:734197 [0] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-169-207:734075:734197 [0] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-169-207:734075:734197 [0] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-169-207:734075:734197 [0] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-169-207:734075:734197 [0] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-169-207:734075:734197 [0] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-169-207:734075:734197 [0] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-169-207:734075:734197 [0] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-169-207:734075:734197 [0] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-169-207:734075:734197 [0] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-169-207:734075:734197 [0] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-169-207:734075:734197 [0] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-169-207:734075:734197 [0] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-169-207:734076:734199 [1] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-169-207:734078:734202 [3] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-169-207:734077:734200 [2] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-169-207:734082:734203 [7] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-169-207:734075:734197 [0] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/162
+ip-26-0-169-207:734075:734197 [0] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-169-207:734079:734201 [4] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-169-207:734079:734201 [4] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-169-207:734079:734201 [4] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-169-207:734079:734201 [4] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-169-207:734079:734201 [4] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-169-207:734079:734201 [4] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-169-207:734079:734201 [4] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-169-207:734079:734201 [4] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-169-207:734079:734201 [4] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-169-207:734079:734201 [4] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-169-207:734079:734201 [4] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-169-207:734079:734201 [4] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-169-207:734079:734201 [4] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-169-207:734079:734201 [4] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-169-207:734079:734201 [4] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-169-207:734079:734201 [4] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-169-207:734079:734201 [4] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-169-207:734079:734201 [4] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-169-207:734079:734201 [4] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-169-207:734079:734201 [4] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-169-207:734079:734201 [4] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-169-207:734079:734201 [4] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-169-207:734079:734201 [4] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-169-207:734079:734201 [4] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-169-207:734079:734201 [4] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-169-207:734079:734201 [4] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-169-207:734079:734201 [4] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-169-207:734079:734201 [4] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-169-207:734079:734201 [4] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-169-207:734079:734201 [4] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-169-207:734079:734201 [4] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-169-207:734079:734201 [4] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-169-207:734079:734201 [4] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/149
+ip-26-0-169-207:734079:734201 [4] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-169-207:734075:734197 [0] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-169-207:734075:734197 [0] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-169-207:734075:734197 [0] NCCL INFO Using network Libfabric
+ip-26-0-169-207:734081:734198 [6] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-169-207:734081:734198 [6] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-169-207:734081:734198 [6] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-169-207:734081:734198 [6] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-169-207:734081:734198 [6] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-169-207:734081:734198 [6] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-169-207:734081:734198 [6] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-169-207:734081:734198 [6] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-169-207:734081:734198 [6] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-169-207:734081:734198 [6] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-169-207:734081:734198 [6] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-169-207:734081:734198 [6] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-169-207:734081:734198 [6] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-169-207:734081:734198 [6] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-169-207:734081:734198 [6] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-169-207:734081:734198 [6] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-169-207:734081:734198 [6] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-169-207:734081:734198 [6] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-169-207:734081:734198 [6] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-169-207:734081:734198 [6] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-169-207:734081:734198 [6] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-169-207:734081:734198 [6] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-169-207:734081:734198 [6] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-169-207:734081:734198 [6] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-169-207:734081:734198 [6] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-169-207:734081:734198 [6] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-169-207:734081:734198 [6] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-169-207:734081:734198 [6] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-169-207:734081:734198 [6] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-169-207:734081:734198 [6] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-169-207:734081:734198 [6] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-169-207:734081:734198 [6] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-169-207:734081:734198 [6] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/149
+ip-26-0-169-207:734081:734198 [6] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-169-207:734080:734204 [5] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-169-207:734080:734204 [5] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-169-207:734080:734204 [5] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-169-207:734080:734204 [5] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-169-207:734080:734204 [5] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-169-207:734080:734204 [5] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-169-207:734080:734204 [5] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-169-207:734080:734204 [5] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-169-207:734080:734204 [5] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-169-207:734080:734204 [5] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-169-207:734080:734204 [5] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-169-207:734080:734204 [5] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-169-207:734080:734204 [5] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-169-207:734080:734204 [5] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-169-207:734080:734204 [5] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-169-207:734080:734204 [5] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-169-207:734080:734204 [5] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-169-207:734080:734204 [5] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-169-207:734080:734204 [5] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-169-207:734080:734204 [5] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-169-207:734080:734204 [5] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-169-207:734080:734204 [5] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-169-207:734080:734204 [5] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-169-207:734080:734204 [5] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-169-207:734080:734204 [5] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-169-207:734080:734204 [5] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-169-207:734080:734204 [5] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-169-207:734080:734204 [5] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-169-207:734080:734204 [5] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-169-207:734080:734204 [5] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-169-207:734080:734204 [5] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-169-207:734080:734204 [5] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-169-207:734080:734204 [5] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/149
+ip-26-0-169-207:734080:734204 [5] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-169-207:734078:734202 [3] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-169-207:734078:734202 [3] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-169-207:734078:734202 [3] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-169-207:734078:734202 [3] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-169-207:734078:734202 [3] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-169-207:734078:734202 [3] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-169-207:734078:734202 [3] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-169-207:734078:734202 [3] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-169-207:734078:734202 [3] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-169-207:734078:734202 [3] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-169-207:734078:734202 [3] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-169-207:734078:734202 [3] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-169-207:734078:734202 [3] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-169-207:734078:734202 [3] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-169-207:734078:734202 [3] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-169-207:734078:734202 [3] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-169-207:734078:734202 [3] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-169-207:734078:734202 [3] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-169-207:734078:734202 [3] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-169-207:734078:734202 [3] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-169-207:734078:734202 [3] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-169-207:734078:734202 [3] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-169-207:734078:734202 [3] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-169-207:734078:734202 [3] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-169-207:734078:734202 [3] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-169-207:734078:734202 [3] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-169-207:734078:734202 [3] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-169-207:734078:734202 [3] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-169-207:734078:734202 [3] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-169-207:734078:734202 [3] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-169-207:734078:734202 [3] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-169-207:734078:734202 [3] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-169-207:734078:734202 [3] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/149
+ip-26-0-169-207:734078:734202 [3] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-169-207:734082:734203 [7] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-169-207:734082:734203 [7] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-169-207:734082:734203 [7] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-169-207:734082:734203 [7] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-169-207:734082:734203 [7] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-169-207:734082:734203 [7] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-169-207:734082:734203 [7] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-169-207:734082:734203 [7] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-169-207:734082:734203 [7] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-169-207:734082:734203 [7] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-169-207:734082:734203 [7] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-169-207:734082:734203 [7] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-169-207:734082:734203 [7] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-169-207:734082:734203 [7] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-169-207:734082:734203 [7] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-169-207:734082:734203 [7] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-169-207:734082:734203 [7] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-169-207:734082:734203 [7] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-169-207:734082:734203 [7] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-169-207:734082:734203 [7] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-169-207:734082:734203 [7] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-169-207:734082:734203 [7] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-169-207:734082:734203 [7] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-169-207:734082:734203 [7] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-169-207:734082:734203 [7] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-169-207:734082:734203 [7] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-169-207:734082:734203 [7] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-169-207:734082:734203 [7] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-169-207:734082:734203 [7] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-169-207:734082:734203 [7] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-169-207:734082:734203 [7] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-169-207:734082:734203 [7] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-169-207:734082:734203 [7] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/149
+ip-26-0-169-207:734082:734203 [7] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-169-207:734076:734199 [1] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-169-207:734076:734199 [1] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-169-207:734076:734199 [1] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-169-207:734076:734199 [1] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-169-207:734076:734199 [1] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-169-207:734076:734199 [1] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-169-207:734076:734199 [1] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-169-207:734076:734199 [1] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-169-207:734076:734199 [1] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-169-207:734076:734199 [1] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-169-207:734076:734199 [1] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-169-207:734076:734199 [1] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-169-207:734076:734199 [1] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-169-207:734076:734199 [1] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-169-207:734076:734199 [1] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-169-207:734076:734199 [1] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-169-207:734076:734199 [1] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-169-207:734076:734199 [1] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-169-207:734076:734199 [1] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-169-207:734076:734199 [1] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-169-207:734076:734199 [1] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-169-207:734076:734199 [1] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-169-207:734076:734199 [1] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-169-207:734076:734199 [1] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-169-207:734076:734199 [1] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-169-207:734076:734199 [1] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-169-207:734076:734199 [1] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-169-207:734076:734199 [1] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-169-207:734076:734199 [1] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-169-207:734076:734199 [1] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-169-207:734076:734199 [1] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-169-207:734076:734199 [1] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-169-207:734077:734200 [2] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-169-207:734077:734200 [2] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-169-207:734077:734200 [2] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-169-207:734077:734200 [2] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-169-207:734077:734200 [2] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-169-207:734077:734200 [2] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-169-207:734077:734200 [2] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-169-207:734077:734200 [2] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-169-207:734077:734200 [2] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-169-207:734077:734200 [2] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-169-207:734077:734200 [2] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-169-207:734077:734200 [2] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-169-207:734077:734200 [2] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-169-207:734077:734200 [2] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-169-207:734077:734200 [2] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-169-207:734077:734200 [2] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-169-207:734077:734200 [2] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-169-207:734077:734200 [2] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-169-207:734077:734200 [2] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-169-207:734077:734200 [2] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-169-207:734077:734200 [2] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-169-207:734077:734200 [2] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-169-207:734077:734200 [2] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-169-207:734077:734200 [2] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-169-207:734077:734200 [2] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-169-207:734077:734200 [2] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-169-207:734077:734200 [2] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-169-207:734077:734200 [2] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-169-207:734077:734200 [2] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-169-207:734077:734200 [2] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-169-207:734077:734200 [2] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-169-207:734077:734200 [2] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-169-207:734076:734199 [1] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/149
+ip-26-0-169-207:734076:734199 [1] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-169-207:734077:734200 [2] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/149
+ip-26-0-169-207:734077:734200 [2] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-169-207:734079:734201 [4] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-169-207:734079:734201 [4] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-169-207:734079:734201 [4] NCCL INFO Using network Libfabric
+ip-26-0-169-207:734081:734198 [6] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-169-207:734081:734198 [6] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-169-207:734081:734198 [6] NCCL INFO Using network Libfabric
+ip-26-0-169-207:734080:734204 [5] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-169-207:734080:734204 [5] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-169-207:734080:734204 [5] NCCL INFO Using network Libfabric
+ip-26-0-169-207:734078:734202 [3] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-169-207:734078:734202 [3] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-169-207:734078:734202 [3] NCCL INFO Using network Libfabric
+ip-26-0-169-207:734082:734203 [7] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-169-207:734082:734203 [7] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-169-207:734082:734203 [7] NCCL INFO Using network Libfabric
+ip-26-0-169-207:734076:734199 [1] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-169-207:734076:734199 [1] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-169-207:734076:734199 [1] NCCL INFO Using network Libfabric
+ip-26-0-169-207:734077:734200 [2] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-169-207:734077:734200 [2] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-169-207:734077:734200 [2] NCCL INFO Using network Libfabric
+ip-26-0-169-207:734078:734202 [3] NCCL INFO comm 0x9e458a0 rank 3 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0x6715c27ec3e7aad8 - Init START
+ip-26-0-169-207:734080:734204 [5] NCCL INFO comm 0xa05d1f0 rank 5 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0x6715c27ec3e7aad8 - Init START
+ip-26-0-169-207:734076:734199 [1] NCCL INFO comm 0x97e86b0 rank 1 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0x6715c27ec3e7aad8 - Init START
+ip-26-0-169-207:734082:734203 [7] NCCL INFO comm 0x8ee7f10 rank 7 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0x6715c27ec3e7aad8 - Init START
+ip-26-0-169-207:734077:734200 [2] NCCL INFO comm 0x8f43610 rank 2 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0x6715c27ec3e7aad8 - Init START
+ip-26-0-169-207:734079:734201 [4] NCCL INFO comm 0x94ca7c0 rank 4 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0x6715c27ec3e7aad8 - Init START
+ip-26-0-169-207:734075:734197 [0] NCCL INFO comm 0x91b26c0 rank 0 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0x6715c27ec3e7aad8 - Init START
+ip-26-0-169-207:734081:734198 [6] NCCL INFO comm 0x844c2c0 rank 6 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0x6715c27ec3e7aad8 - Init START
+ip-26-0-169-207:734081:734198 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/149
+ip-26-0-169-207:734082:734203 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/149
+ip-26-0-169-207:734076:734199 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/149
+ip-26-0-169-207:734078:734202 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/149
+ip-26-0-169-207:734080:734204 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/149
+ip-26-0-169-207:734075:734197 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/162
+ip-26-0-169-207:734079:734201 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/149
+ip-26-0-169-207:734077:734200 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/149
+ip-26-0-169-207:734082:734203 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-169-207:734082:734203 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-169-207:734080:734204 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-169-207:734080:734204 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-169-207:734081:734198 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-169-207:734081:734198 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-169-207:734079:734201 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-169-207:734079:734201 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-169-207:734075:734197 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-169-207:734075:734197 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-169-207:734077:734200 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-169-207:734077:734200 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-169-207:734078:734202 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-169-207:734078:734202 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-169-207:734076:734199 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-169-207:734076:734199 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-169-207:734076:734199 [1] NCCL INFO Trees [0] 2/-1/-1->1->0 [1] 2/-1/-1->1->0 [2] 2/-1/-1->1->0 [3] 2/-1/-1->1->0 [4] 2/-1/-1->1->0 [5] 2/-1/-1->1->0 [6] 2/-1/-1->1->0 [7] 2/-1/-1->1->0 [8] 2/-1/-1->1->0 [9] 2/-1/-1->1->0 [10] 2/-1/-1->1->0 [11] 2/-1/-1->1->0 [12] 2/-1/-1->1->0 [13] 2/-1/-1->1->0 [14] 2/-1/-1->1->0 [15] 2/-1/-1->1->0 [16] 2/-1/-1->1->0 [17] 2/-1/-1->1->0 [18] 2/-1/-1->1->0 [19] 2/-1/-1->1->0 [20] 2/-1/-1->1->0 [21] 2/-1/-1->1->0 [22] 2/-1/-1->1->0 [23] 2/-1/-1->1->0
+ip-26-0-169-207:734076:734199 [1] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-169-207:734075:734197 [0] NCCL INFO Channel 00/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:734075:734197 [0] NCCL INFO Channel 01/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:734075:734197 [0] NCCL INFO Channel 02/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:734075:734197 [0] NCCL INFO Channel 03/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:734075:734197 [0] NCCL INFO Channel 04/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:734075:734197 [0] NCCL INFO Channel 05/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:734075:734197 [0] NCCL INFO Channel 06/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:734075:734197 [0] NCCL INFO Channel 07/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:734075:734197 [0] NCCL INFO Channel 08/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:734075:734197 [0] NCCL INFO Channel 09/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:734075:734197 [0] NCCL INFO Channel 10/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:734077:734200 [2] NCCL INFO Trees [0] 3/-1/-1->2->1 [1] 3/-1/-1->2->1 [2] 3/-1/-1->2->1 [3] 3/-1/-1->2->1 [4] 3/-1/-1->2->1 [5] 3/-1/-1->2->1 [6] 3/-1/-1->2->1 [7] 3/-1/-1->2->1 [8] 3/-1/-1->2->1 [9] 3/-1/-1->2->1 [10] 3/-1/-1->2->1 [11] 3/-1/-1->2->1 [12] 3/-1/-1->2->1 [13] 3/-1/-1->2->1 [14] 3/-1/-1->2->1 [15] 3/-1/-1->2->1 [16] 3/-1/-1->2->1 [17] 3/-1/-1->2->1 [18] 3/-1/-1->2->1 [19] 3/-1/-1->2->1 [20] 3/-1/-1->2->1 [21] 3/-1/-1->2->1 [22] 3/-1/-1->2->1 [23] 3/-1/-1->2->1
+ip-26-0-169-207:734075:734197 [0] NCCL INFO Channel 11/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:734075:734197 [0] NCCL INFO Channel 12/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:734077:734200 [2] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-169-207:734075:734197 [0] NCCL INFO Channel 13/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:734075:734197 [0] NCCL INFO Channel 14/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:734075:734197 [0] NCCL INFO Channel 15/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:734075:734197 [0] NCCL INFO Channel 16/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:734075:734197 [0] NCCL INFO Channel 17/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:734075:734197 [0] NCCL INFO Channel 18/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:734075:734197 [0] NCCL INFO Channel 19/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:734075:734197 [0] NCCL INFO Channel 20/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:734075:734197 [0] NCCL INFO Channel 21/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:734082:734203 [7] NCCL INFO Trees [0] -1/-1/-1->7->6 [1] -1/-1/-1->7->6 [2] -1/-1/-1->7->6 [3] -1/-1/-1->7->6 [4] -1/-1/-1->7->6 [5] -1/-1/-1->7->6 [6] -1/-1/-1->7->6 [7] -1/-1/-1->7->6 [8] -1/-1/-1->7->6 [9] -1/-1/-1->7->6 [10] -1/-1/-1->7->6 [11] -1/-1/-1->7->6 [12] -1/-1/-1->7->6 [13] -1/-1/-1->7->6 [14] -1/-1/-1->7->6 [15] -1/-1/-1->7->6 [16] -1/-1/-1->7->6 [17] -1/-1/-1->7->6 [18] -1/-1/-1->7->6 [19] -1/-1/-1->7->6 [20] -1/-1/-1->7->6 [21] -1/-1/-1->7->6 [22] -1/-1/-1->7->6 [23] -1/-1/-1->7->6
+ip-26-0-169-207:734075:734197 [0] NCCL INFO Channel 22/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:734075:734197 [0] NCCL INFO Channel 23/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:734081:734198 [6] NCCL INFO Trees [0] 7/-1/-1->6->5 [1] 7/-1/-1->6->5 [2] 7/-1/-1->6->5 [3] 7/-1/-1->6->5 [4] 7/-1/-1->6->5 [5] 7/-1/-1->6->5 [6] 7/-1/-1->6->5 [7] 7/-1/-1->6->5 [8] 7/-1/-1->6->5 [9] 7/-1/-1->6->5 [10] 7/-1/-1->6->5 [11] 7/-1/-1->6->5 [12] 7/-1/-1->6->5 [13] 7/-1/-1->6->5 [14] 7/-1/-1->6->5 [15] 7/-1/-1->6->5 [16] 7/-1/-1->6->5 [17] 7/-1/-1->6->5 [18] 7/-1/-1->6->5 [19] 7/-1/-1->6->5 [20] 7/-1/-1->6->5 [21] 7/-1/-1->6->5 [22] 7/-1/-1->6->5 [23] 7/-1/-1->6->5
+ip-26-0-169-207:734078:734202 [3] NCCL INFO Trees [0] 4/-1/-1->3->2 [1] 4/-1/-1->3->2 [2] 4/-1/-1->3->2 [3] 4/-1/-1->3->2 [4] 4/-1/-1->3->2 [5] 4/-1/-1->3->2 [6] 4/-1/-1->3->2 [7] 4/-1/-1->3->2 [8] 4/-1/-1->3->2 [9] 4/-1/-1->3->2 [10] 4/-1/-1->3->2 [11] 4/-1/-1->3->2 [12] 4/-1/-1->3->2 [13] 4/-1/-1->3->2 [14] 4/-1/-1->3->2 [15] 4/-1/-1->3->2 [16] 4/-1/-1->3->2 [17] 4/-1/-1->3->2 [18] 4/-1/-1->3->2 [19] 4/-1/-1->3->2 [20] 4/-1/-1->3->2 [21] 4/-1/-1->3->2 [22] 4/-1/-1->3->2 [23] 4/-1/-1->3->2
+ip-26-0-169-207:734079:734201 [4] NCCL INFO Trees [0] 5/-1/-1->4->3 [1] 5/-1/-1->4->3 [2] 5/-1/-1->4->3 [3] 5/-1/-1->4->3 [4] 5/-1/-1->4->3 [5] 5/-1/-1->4->3 [6] 5/-1/-1->4->3 [7] 5/-1/-1->4->3 [8] 5/-1/-1->4->3 [9] 5/-1/-1->4->3 [10] 5/-1/-1->4->3 [11] 5/-1/-1->4->3 [12] 5/-1/-1->4->3 [13] 5/-1/-1->4->3 [14] 5/-1/-1->4->3 [15] 5/-1/-1->4->3 [16] 5/-1/-1->4->3 [17] 5/-1/-1->4->3 [18] 5/-1/-1->4->3 [19] 5/-1/-1->4->3 [20] 5/-1/-1->4->3 [21] 5/-1/-1->4->3 [22] 5/-1/-1->4->3 [23] 5/-1/-1->4->3
+ip-26-0-169-207:734078:734202 [3] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-169-207:734081:734198 [6] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-169-207:734079:734201 [4] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-169-207:734075:734197 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] 1/-1/-1->0->-1 [7] 1/-1/-1->0->-1 [8] 1/-1/-1->0->-1 [9] 1/-1/-1->0->-1 [10] 1/-1/-1->0->-1 [11] 1/-1/-1->0->-1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] 1/-1/-1->0->-1 [19] 1/-1/-1->0->-1 [20] 1/-1/-1->0->-1 [21] 1/-1/-1->0->-1 [22] 1/-1/-1->0->-1 [23] 1/-1/-1->0->-1
+ip-26-0-169-207:734075:734197 [0] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-169-207:734080:734204 [5] NCCL INFO Trees [0] 6/-1/-1->5->4 [1] 6/-1/-1->5->4 [2] 6/-1/-1->5->4 [3] 6/-1/-1->5->4 [4] 6/-1/-1->5->4 [5] 6/-1/-1->5->4 [6] 6/-1/-1->5->4 [7] 6/-1/-1->5->4 [8] 6/-1/-1->5->4 [9] 6/-1/-1->5->4 [10] 6/-1/-1->5->4 [11] 6/-1/-1->5->4 [12] 6/-1/-1->5->4 [13] 6/-1/-1->5->4 [14] 6/-1/-1->5->4 [15] 6/-1/-1->5->4 [16] 6/-1/-1->5->4 [17] 6/-1/-1->5->4 [18] 6/-1/-1->5->4 [19] 6/-1/-1->5->4 [20] 6/-1/-1->5->4 [21] 6/-1/-1->5->4 [22] 6/-1/-1->5->4 [23] 6/-1/-1->5->4
+ip-26-0-169-207:734080:734204 [5] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-169-207:734082:734203 [7] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-169-207:734076:734199 [1] NCCL INFO Channel 00/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734076:734199 [1] NCCL INFO Channel 01/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734075:734197 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734076:734199 [1] NCCL INFO Channel 02/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734075:734197 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734076:734199 [1] NCCL INFO Channel 03/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734077:734200 [2] NCCL INFO Channel 00/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734079:734201 [4] NCCL INFO Channel 00/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734081:734198 [6] NCCL INFO Channel 00/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:734080:734204 [5] NCCL INFO Channel 00/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734078:734202 [3] NCCL INFO Channel 00/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734075:734197 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734076:734199 [1] NCCL INFO Channel 04/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734077:734200 [2] NCCL INFO Channel 01/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734081:734198 [6] NCCL INFO Channel 01/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:734080:734204 [5] NCCL INFO Channel 01/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734078:734202 [3] NCCL INFO Channel 01/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734076:734199 [1] NCCL INFO Channel 05/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734075:734197 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734077:734200 [2] NCCL INFO Channel 02/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734080:734204 [5] NCCL INFO Channel 02/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734078:734202 [3] NCCL INFO Channel 02/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734081:734198 [6] NCCL INFO Channel 02/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:734076:734199 [1] NCCL INFO Channel 06/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734075:734197 [0] NCCL INFO Channel 04/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734077:734200 [2] NCCL INFO Channel 03/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734080:734204 [5] NCCL INFO Channel 03/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734078:734202 [3] NCCL INFO Channel 03/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734081:734198 [6] NCCL INFO Channel 03/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:734076:734199 [1] NCCL INFO Channel 07/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734075:734197 [0] NCCL INFO Channel 05/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734077:734200 [2] NCCL INFO Channel 04/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734079:734201 [4] NCCL INFO Channel 01/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734080:734204 [5] NCCL INFO Channel 04/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734078:734202 [3] NCCL INFO Channel 04/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734081:734198 [6] NCCL INFO Channel 04/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:734076:734199 [1] NCCL INFO Channel 08/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734075:734197 [0] NCCL INFO Channel 06/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734077:734200 [2] NCCL INFO Channel 05/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734079:734201 [4] NCCL INFO Channel 02/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734078:734202 [3] NCCL INFO Channel 05/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734080:734204 [5] NCCL INFO Channel 05/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734081:734198 [6] NCCL INFO Channel 05/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:734076:734199 [1] NCCL INFO Channel 09/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734075:734197 [0] NCCL INFO Channel 07/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734077:734200 [2] NCCL INFO Channel 06/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734079:734201 [4] NCCL INFO Channel 03/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734078:734202 [3] NCCL INFO Channel 06/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734080:734204 [5] NCCL INFO Channel 06/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734081:734198 [6] NCCL INFO Channel 06/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:734076:734199 [1] NCCL INFO Channel 10/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734075:734197 [0] NCCL INFO Channel 08/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734077:734200 [2] NCCL INFO Channel 07/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734079:734201 [4] NCCL INFO Channel 04/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734078:734202 [3] NCCL INFO Channel 07/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734080:734204 [5] NCCL INFO Channel 07/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734081:734198 [6] NCCL INFO Channel 07/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:734076:734199 [1] NCCL INFO Channel 11/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734075:734197 [0] NCCL INFO Channel 09/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734077:734200 [2] NCCL INFO Channel 08/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734079:734201 [4] NCCL INFO Channel 05/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734078:734202 [3] NCCL INFO Channel 08/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734080:734204 [5] NCCL INFO Channel 08/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734081:734198 [6] NCCL INFO Channel 08/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:734082:734203 [7] NCCL INFO Channel 00/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734076:734199 [1] NCCL INFO Channel 12/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734075:734197 [0] NCCL INFO Channel 10/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734077:734200 [2] NCCL INFO Channel 09/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734079:734201 [4] NCCL INFO Channel 06/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734078:734202 [3] NCCL INFO Channel 09/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734080:734204 [5] NCCL INFO Channel 09/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734081:734198 [6] NCCL INFO Channel 09/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:734082:734203 [7] NCCL INFO Channel 01/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734076:734199 [1] NCCL INFO Channel 13/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734075:734197 [0] NCCL INFO Channel 11/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734077:734200 [2] NCCL INFO Channel 10/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734079:734201 [4] NCCL INFO Channel 07/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734078:734202 [3] NCCL INFO Channel 10/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734080:734204 [5] NCCL INFO Channel 10/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734081:734198 [6] NCCL INFO Channel 10/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:734082:734203 [7] NCCL INFO Channel 02/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734076:734199 [1] NCCL INFO Channel 14/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734075:734197 [0] NCCL INFO Channel 12/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734077:734200 [2] NCCL INFO Channel 11/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734079:734201 [4] NCCL INFO Channel 08/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734078:734202 [3] NCCL INFO Channel 11/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734080:734204 [5] NCCL INFO Channel 11/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734081:734198 [6] NCCL INFO Channel 11/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:734082:734203 [7] NCCL INFO Channel 03/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734076:734199 [1] NCCL INFO Channel 15/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734075:734197 [0] NCCL INFO Channel 13/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734079:734201 [4] NCCL INFO Channel 09/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734077:734200 [2] NCCL INFO Channel 12/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734081:734198 [6] NCCL INFO Channel 12/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:734078:734202 [3] NCCL INFO Channel 12/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734080:734204 [5] NCCL INFO Channel 12/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734082:734203 [7] NCCL INFO Channel 04/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734076:734199 [1] NCCL INFO Channel 16/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734079:734201 [4] NCCL INFO Channel 10/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734075:734197 [0] NCCL INFO Channel 14/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734077:734200 [2] NCCL INFO Channel 13/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734081:734198 [6] NCCL INFO Channel 13/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:734078:734202 [3] NCCL INFO Channel 13/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734080:734204 [5] NCCL INFO Channel 13/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734082:734203 [7] NCCL INFO Channel 05/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734076:734199 [1] NCCL INFO Channel 17/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734079:734201 [4] NCCL INFO Channel 11/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734075:734197 [0] NCCL INFO Channel 15/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734077:734200 [2] NCCL INFO Channel 14/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734078:734202 [3] NCCL INFO Channel 14/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734081:734198 [6] NCCL INFO Channel 14/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:734080:734204 [5] NCCL INFO Channel 14/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734082:734203 [7] NCCL INFO Channel 06/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734076:734199 [1] NCCL INFO Channel 18/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734078:734202 [3] NCCL INFO Channel 15/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734081:734198 [6] NCCL INFO Channel 15/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:734080:734204 [5] NCCL INFO Channel 15/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734082:734203 [7] NCCL INFO Channel 07/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734078:734202 [3] NCCL INFO Channel 16/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734079:734201 [4] NCCL INFO Channel 12/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734081:734198 [6] NCCL INFO Channel 16/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:734080:734204 [5] NCCL INFO Channel 16/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734075:734197 [0] NCCL INFO Channel 16/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734082:734203 [7] NCCL INFO Channel 08/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734077:734200 [2] NCCL INFO Channel 15/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734076:734199 [1] NCCL INFO Channel 19/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734081:734198 [6] NCCL INFO Channel 17/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:734075:734197 [0] NCCL INFO Channel 17/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734082:734203 [7] NCCL INFO Channel 09/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734077:734200 [2] NCCL INFO Channel 16/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734076:734199 [1] NCCL INFO Channel 20/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734081:734198 [6] NCCL INFO Channel 18/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:734075:734197 [0] NCCL INFO Channel 18/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734082:734203 [7] NCCL INFO Channel 10/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734077:734200 [2] NCCL INFO Channel 17/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734076:734199 [1] NCCL INFO Channel 21/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734081:734198 [6] NCCL INFO Channel 19/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:734075:734197 [0] NCCL INFO Channel 19/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734082:734203 [7] NCCL INFO Channel 11/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734077:734200 [2] NCCL INFO Channel 18/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734076:734199 [1] NCCL INFO Channel 22/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734081:734198 [6] NCCL INFO Channel 20/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:734079:734201 [4] NCCL INFO Channel 13/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734082:734203 [7] NCCL INFO Channel 12/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734075:734197 [0] NCCL INFO Channel 20/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734077:734200 [2] NCCL INFO Channel 19/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734076:734199 [1] NCCL INFO Channel 23/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734078:734202 [3] NCCL INFO Channel 17/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734079:734201 [4] NCCL INFO Channel 14/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734080:734204 [5] NCCL INFO Channel 17/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734075:734197 [0] NCCL INFO Channel 21/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734078:734202 [3] NCCL INFO Channel 18/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734079:734201 [4] NCCL INFO Channel 15/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734080:734204 [5] NCCL INFO Channel 18/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734075:734197 [0] NCCL INFO Channel 22/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734078:734202 [3] NCCL INFO Channel 19/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734079:734201 [4] NCCL INFO Channel 16/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734080:734204 [5] NCCL INFO Channel 19/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734075:734197 [0] NCCL INFO Channel 23/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734078:734202 [3] NCCL INFO Channel 20/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734079:734201 [4] NCCL INFO Channel 17/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734080:734204 [5] NCCL INFO Channel 20/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734082:734203 [7] NCCL INFO Channel 13/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734078:734202 [3] NCCL INFO Channel 21/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734079:734201 [4] NCCL INFO Channel 18/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734080:734204 [5] NCCL INFO Channel 21/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734082:734203 [7] NCCL INFO Channel 14/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734078:734202 [3] NCCL INFO Channel 22/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734079:734201 [4] NCCL INFO Channel 19/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734080:734204 [5] NCCL INFO Channel 22/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734082:734203 [7] NCCL INFO Channel 15/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734078:734202 [3] NCCL INFO Channel 23/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734081:734198 [6] NCCL INFO Channel 21/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:734079:734201 [4] NCCL INFO Channel 20/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734080:734204 [5] NCCL INFO Channel 23/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734077:734200 [2] NCCL INFO Channel 20/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734082:734203 [7] NCCL INFO Channel 16/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734081:734198 [6] NCCL INFO Channel 22/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:734079:734201 [4] NCCL INFO Channel 21/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734077:734200 [2] NCCL INFO Channel 21/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734082:734203 [7] NCCL INFO Channel 17/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734081:734198 [6] NCCL INFO Channel 23/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:734079:734201 [4] NCCL INFO Channel 22/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734077:734200 [2] NCCL INFO Channel 22/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734082:734203 [7] NCCL INFO Channel 18/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734079:734201 [4] NCCL INFO Channel 23/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734077:734200 [2] NCCL INFO Channel 23/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734082:734203 [7] NCCL INFO Channel 19/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734082:734203 [7] NCCL INFO Channel 20/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734082:734203 [7] NCCL INFO Channel 21/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734082:734203 [7] NCCL INFO Channel 22/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734082:734203 [7] NCCL INFO Channel 23/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734080:734204 [5] NCCL INFO Connected all rings
+ip-26-0-169-207:734077:734200 [2] NCCL INFO Connected all rings
+ip-26-0-169-207:734076:734199 [1] NCCL INFO Connected all rings
+ip-26-0-169-207:734079:734201 [4] NCCL INFO Connected all rings
+ip-26-0-169-207:734081:734198 [6] NCCL INFO Connected all rings
+ip-26-0-169-207:734078:734202 [3] NCCL INFO Connected all rings
+ip-26-0-169-207:734082:734203 [7] NCCL INFO Connected all rings
+ip-26-0-169-207:734082:734203 [7] NCCL INFO Channel 00/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734075:734197 [0] NCCL INFO Connected all rings
+ip-26-0-169-207:734082:734203 [7] NCCL INFO Channel 01/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734082:734203 [7] NCCL INFO Channel 02/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734082:734203 [7] NCCL INFO Channel 03/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734082:734203 [7] NCCL INFO Channel 04/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734082:734203 [7] NCCL INFO Channel 05/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734082:734203 [7] NCCL INFO Channel 06/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734082:734203 [7] NCCL INFO Channel 07/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734082:734203 [7] NCCL INFO Channel 08/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734082:734203 [7] NCCL INFO Channel 09/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734082:734203 [7] NCCL INFO Channel 10/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734082:734203 [7] NCCL INFO Channel 11/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734082:734203 [7] NCCL INFO Channel 12/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734082:734203 [7] NCCL INFO Channel 13/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734082:734203 [7] NCCL INFO Channel 14/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734082:734203 [7] NCCL INFO Channel 15/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734082:734203 [7] NCCL INFO Channel 16/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734082:734203 [7] NCCL INFO Channel 17/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734080:734204 [5] NCCL INFO Channel 00/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734077:734200 [2] NCCL INFO Channel 00/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734082:734203 [7] NCCL INFO Channel 18/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734080:734204 [5] NCCL INFO Channel 01/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734076:734199 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734077:734200 [2] NCCL INFO Channel 01/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734082:734203 [7] NCCL INFO Channel 19/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734080:734204 [5] NCCL INFO Channel 02/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734076:734199 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734077:734200 [2] NCCL INFO Channel 02/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734082:734203 [7] NCCL INFO Channel 20/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734080:734204 [5] NCCL INFO Channel 03/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734076:734199 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734077:734200 [2] NCCL INFO Channel 03/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734082:734203 [7] NCCL INFO Channel 21/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734080:734204 [5] NCCL INFO Channel 04/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734076:734199 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734077:734200 [2] NCCL INFO Channel 04/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734079:734201 [4] NCCL INFO Channel 00/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734082:734203 [7] NCCL INFO Channel 22/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734080:734204 [5] NCCL INFO Channel 05/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734076:734199 [1] NCCL INFO Channel 04/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734077:734200 [2] NCCL INFO Channel 05/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734079:734201 [4] NCCL INFO Channel 01/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734082:734203 [7] NCCL INFO Channel 23/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734080:734204 [5] NCCL INFO Channel 06/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734078:734202 [3] NCCL INFO Channel 00/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734081:734198 [6] NCCL INFO Channel 00/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734076:734199 [1] NCCL INFO Channel 05/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734077:734200 [2] NCCL INFO Channel 06/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734079:734201 [4] NCCL INFO Channel 02/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734080:734204 [5] NCCL INFO Channel 07/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734078:734202 [3] NCCL INFO Channel 01/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734081:734198 [6] NCCL INFO Channel 01/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734076:734199 [1] NCCL INFO Channel 06/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734077:734200 [2] NCCL INFO Channel 07/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734079:734201 [4] NCCL INFO Channel 03/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734080:734204 [5] NCCL INFO Channel 08/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734078:734202 [3] NCCL INFO Channel 02/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734077:734200 [2] NCCL INFO Channel 08/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734081:734198 [6] NCCL INFO Channel 02/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734076:734199 [1] NCCL INFO Channel 07/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734079:734201 [4] NCCL INFO Channel 04/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734080:734204 [5] NCCL INFO Channel 09/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734078:734202 [3] NCCL INFO Channel 03/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734081:734198 [6] NCCL INFO Channel 03/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734076:734199 [1] NCCL INFO Channel 08/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734077:734200 [2] NCCL INFO Channel 09/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734079:734201 [4] NCCL INFO Channel 05/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734080:734204 [5] NCCL INFO Channel 10/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734078:734202 [3] NCCL INFO Channel 04/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734081:734198 [6] NCCL INFO Channel 04/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734077:734200 [2] NCCL INFO Channel 10/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734076:734199 [1] NCCL INFO Channel 09/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734079:734201 [4] NCCL INFO Channel 06/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734080:734204 [5] NCCL INFO Channel 11/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734078:734202 [3] NCCL INFO Channel 05/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734081:734198 [6] NCCL INFO Channel 05/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734077:734200 [2] NCCL INFO Channel 11/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734076:734199 [1] NCCL INFO Channel 10/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734079:734201 [4] NCCL INFO Channel 07/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734080:734204 [5] NCCL INFO Channel 12/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734078:734202 [3] NCCL INFO Channel 06/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734081:734198 [6] NCCL INFO Channel 06/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734077:734200 [2] NCCL INFO Channel 12/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734076:734199 [1] NCCL INFO Channel 11/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734079:734201 [4] NCCL INFO Channel 08/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734080:734204 [5] NCCL INFO Channel 13/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734081:734198 [6] NCCL INFO Channel 07/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734078:734202 [3] NCCL INFO Channel 07/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734077:734200 [2] NCCL INFO Channel 13/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734076:734199 [1] NCCL INFO Channel 12/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734079:734201 [4] NCCL INFO Channel 09/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734080:734204 [5] NCCL INFO Channel 14/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734081:734198 [6] NCCL INFO Channel 08/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734078:734202 [3] NCCL INFO Channel 08/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734077:734200 [2] NCCL INFO Channel 14/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734076:734199 [1] NCCL INFO Channel 13/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734079:734201 [4] NCCL INFO Channel 10/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734080:734204 [5] NCCL INFO Channel 15/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734081:734198 [6] NCCL INFO Channel 09/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734078:734202 [3] NCCL INFO Channel 09/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734077:734200 [2] NCCL INFO Channel 15/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734076:734199 [1] NCCL INFO Channel 14/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734079:734201 [4] NCCL INFO Channel 11/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734080:734204 [5] NCCL INFO Channel 16/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734081:734198 [6] NCCL INFO Channel 10/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734078:734202 [3] NCCL INFO Channel 10/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734077:734200 [2] NCCL INFO Channel 16/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734076:734199 [1] NCCL INFO Channel 15/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734079:734201 [4] NCCL INFO Channel 12/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734080:734204 [5] NCCL INFO Channel 17/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734081:734198 [6] NCCL INFO Channel 11/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734078:734202 [3] NCCL INFO Channel 11/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734077:734200 [2] NCCL INFO Channel 17/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734076:734199 [1] NCCL INFO Channel 16/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734079:734201 [4] NCCL INFO Channel 13/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734080:734204 [5] NCCL INFO Channel 18/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734081:734198 [6] NCCL INFO Channel 12/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734078:734202 [3] NCCL INFO Channel 12/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734077:734200 [2] NCCL INFO Channel 18/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734076:734199 [1] NCCL INFO Channel 17/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734079:734201 [4] NCCL INFO Channel 14/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734080:734204 [5] NCCL INFO Channel 19/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734081:734198 [6] NCCL INFO Channel 13/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734078:734202 [3] NCCL INFO Channel 13/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734077:734200 [2] NCCL INFO Channel 19/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734076:734199 [1] NCCL INFO Channel 18/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734079:734201 [4] NCCL INFO Channel 15/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734080:734204 [5] NCCL INFO Channel 20/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734081:734198 [6] NCCL INFO Channel 14/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734078:734202 [3] NCCL INFO Channel 14/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734077:734200 [2] NCCL INFO Channel 20/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734076:734199 [1] NCCL INFO Channel 19/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734079:734201 [4] NCCL INFO Channel 16/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734080:734204 [5] NCCL INFO Channel 21/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734081:734198 [6] NCCL INFO Channel 15/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734078:734202 [3] NCCL INFO Channel 15/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734077:734200 [2] NCCL INFO Channel 21/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734076:734199 [1] NCCL INFO Channel 20/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734079:734201 [4] NCCL INFO Channel 17/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734080:734204 [5] NCCL INFO Channel 22/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734081:734198 [6] NCCL INFO Channel 16/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734078:734202 [3] NCCL INFO Channel 16/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734077:734200 [2] NCCL INFO Channel 22/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734076:734199 [1] NCCL INFO Channel 21/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734079:734201 [4] NCCL INFO Channel 18/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734080:734204 [5] NCCL INFO Channel 23/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734081:734198 [6] NCCL INFO Channel 17/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734078:734202 [3] NCCL INFO Channel 17/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734077:734200 [2] NCCL INFO Channel 23/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734076:734199 [1] NCCL INFO Channel 22/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734079:734201 [4] NCCL INFO Channel 19/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734081:734198 [6] NCCL INFO Channel 18/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734078:734202 [3] NCCL INFO Channel 18/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734076:734199 [1] NCCL INFO Channel 23/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734079:734201 [4] NCCL INFO Channel 20/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734081:734198 [6] NCCL INFO Channel 19/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734078:734202 [3] NCCL INFO Channel 19/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734079:734201 [4] NCCL INFO Channel 21/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734081:734198 [6] NCCL INFO Channel 20/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734078:734202 [3] NCCL INFO Channel 20/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734079:734201 [4] NCCL INFO Channel 22/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734081:734198 [6] NCCL INFO Channel 21/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734078:734202 [3] NCCL INFO Channel 21/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734079:734201 [4] NCCL INFO Channel 23/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734081:734198 [6] NCCL INFO Channel 22/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734078:734202 [3] NCCL INFO Channel 22/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734081:734198 [6] NCCL INFO Channel 23/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734078:734202 [3] NCCL INFO Channel 23/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734075:734197 [0] NCCL INFO Connected all trees
+ip-26-0-169-207:734075:734197 [0] NCCL INFO NVLS comm 0x91b26c0 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-169-207:734082:734203 [7] NCCL INFO Connected all trees
+ip-26-0-169-207:734082:734203 [7] NCCL INFO NVLS comm 0x8ee7f10 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-169-207:734080:734204 [5] NCCL INFO Connected all trees
+ip-26-0-169-207:734080:734204 [5] NCCL INFO NVLS comm 0xa05d1f0 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-169-207:734076:734199 [1] NCCL INFO Connected all trees
+ip-26-0-169-207:734076:734199 [1] NCCL INFO NVLS comm 0x97e86b0 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-169-207:734078:734202 [3] NCCL INFO Connected all trees
+ip-26-0-169-207:734078:734202 [3] NCCL INFO NVLS comm 0x9e458a0 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-169-207:734077:734200 [2] NCCL INFO Connected all trees
+ip-26-0-169-207:734077:734200 [2] NCCL INFO NVLS comm 0x8f43610 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-169-207:734079:734201 [4] NCCL INFO Connected all trees
+ip-26-0-169-207:734079:734201 [4] NCCL INFO NVLS comm 0x94ca7c0 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-169-207:734081:734198 [6] NCCL INFO Connected all trees
+ip-26-0-169-207:734081:734198 [6] NCCL INFO NVLS comm 0x844c2c0 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-169-207:734076:734199 [1] NCCL INFO Connected NVLS tree
+ip-26-0-169-207:734076:734199 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-169-207:734076:734199 [1] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-169-207:734076:734199 [1] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:734082:734203 [7] NCCL INFO Connected NVLS tree
+ip-26-0-169-207:734082:734203 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-169-207:734082:734203 [7] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-169-207:734082:734203 [7] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:734075:734197 [0] NCCL INFO Connected NVLS tree
+ip-26-0-169-207:734075:734197 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-169-207:734075:734197 [0] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-169-207:734075:734197 [0] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:734077:734200 [2] NCCL INFO Connected NVLS tree
+ip-26-0-169-207:734077:734200 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-169-207:734077:734200 [2] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-169-207:734077:734200 [2] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:734081:734198 [6] NCCL INFO Connected NVLS tree
+ip-26-0-169-207:734079:734201 [4] NCCL INFO Connected NVLS tree
+ip-26-0-169-207:734078:734202 [3] NCCL INFO Connected NVLS tree
+ip-26-0-169-207:734079:734201 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-169-207:734081:734198 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-169-207:734079:734201 [4] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-169-207:734079:734201 [4] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:734081:734198 [6] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-169-207:734081:734198 [6] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:734078:734202 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-169-207:734078:734202 [3] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-169-207:734078:734202 [3] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:734080:734204 [5] NCCL INFO Connected NVLS tree
+ip-26-0-169-207:734080:734204 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-169-207:734080:734204 [5] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-169-207:734080:734204 [5] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:734078:734202 [3] NCCL INFO comm 0x9e458a0 rank 3 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0x6715c27ec3e7aad8 - Init COMPLETE
+ip-26-0-169-207:734076:734199 [1] NCCL INFO comm 0x97e86b0 rank 1 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0x6715c27ec3e7aad8 - Init COMPLETE
+ip-26-0-169-207:734082:734203 [7] NCCL INFO comm 0x8ee7f10 rank 7 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0x6715c27ec3e7aad8 - Init COMPLETE
+ip-26-0-169-207:734080:734204 [5] NCCL INFO comm 0xa05d1f0 rank 5 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0x6715c27ec3e7aad8 - Init COMPLETE
+ip-26-0-169-207:734075:734197 [0] NCCL INFO comm 0x91b26c0 rank 0 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0x6715c27ec3e7aad8 - Init COMPLETE
+ip-26-0-169-207:734079:734201 [4] NCCL INFO comm 0x94ca7c0 rank 4 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0x6715c27ec3e7aad8 - Init COMPLETE
+ip-26-0-169-207:734077:734200 [2] NCCL INFO comm 0x8f43610 rank 2 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0x6715c27ec3e7aad8 - Init COMPLETE
+ip-26-0-169-207:734081:734198 [6] NCCL INFO comm 0x844c2c0 rank 6 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0x6715c27ec3e7aad8 - Init COMPLETE
+ip-26-0-169-207:734075:734267 [0] NCCL INFO Using network Libfabric
+ip-26-0-169-207:734081:734269 [6] NCCL INFO Using network Libfabric
+ip-26-0-169-207:734076:734268 [1] NCCL INFO Using network Libfabric
+ip-26-0-169-207:734078:734270 [3] NCCL INFO Using network Libfabric
+ip-26-0-169-207:734082:734272 [7] NCCL INFO Using network Libfabric
+ip-26-0-169-207:734077:734271 [2] NCCL INFO Using network Libfabric
+ip-26-0-169-207:734079:734273 [4] NCCL INFO Using network Libfabric
+ip-26-0-169-207:734080:734274 [5] NCCL INFO Using network Libfabric
+ip-26-0-169-207:734080:734274 [5] NCCL INFO comm 0xa286f90 rank 5 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0xf812bb8dd01e9d71 - Init START
+ip-26-0-169-207:734077:734271 [2] NCCL INFO comm 0x916d2e0 rank 2 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0xf812bb8dd01e9d71 - Init START
+ip-26-0-169-207:734078:734270 [3] NCCL INFO comm 0xa071700 rank 3 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0xf812bb8dd01e9d71 - Init START
+ip-26-0-169-207:734081:734269 [6] NCCL INFO comm 0x86769c0 rank 6 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0xf812bb8dd01e9d71 - Init START
+ip-26-0-169-207:734075:734267 [0] NCCL INFO comm 0x93dc430 rank 0 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0xf812bb8dd01e9d71 - Init START
+ip-26-0-169-207:734076:734268 [1] NCCL INFO comm 0x9a12330 rank 1 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0xf812bb8dd01e9d71 - Init START
+ip-26-0-169-207:734079:734273 [4] NCCL INFO comm 0x96f4cd0 rank 4 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0xf812bb8dd01e9d71 - Init START
+ip-26-0-169-207:734082:734272 [7] NCCL INFO comm 0x9111820 rank 7 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0xf812bb8dd01e9d71 - Init START
+ip-26-0-169-207:734082:734272 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/149
+ip-26-0-169-207:734080:734274 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/149
+ip-26-0-169-207:734079:734273 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/149
+ip-26-0-169-207:734076:734268 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/149
+ip-26-0-169-207:734075:734267 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/162
+ip-26-0-169-207:734081:734269 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/149
+ip-26-0-169-207:734078:734270 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/149
+ip-26-0-169-207:734077:734271 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/149
+ip-26-0-169-207:734076:734268 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-169-207:734076:734268 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-169-207:734078:734270 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-169-207:734078:734270 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-169-207:734075:734267 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-169-207:734075:734267 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-169-207:734080:734274 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-169-207:734080:734274 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-169-207:734079:734273 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-169-207:734079:734273 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-169-207:734077:734271 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-169-207:734077:734271 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-169-207:734081:734269 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-169-207:734081:734269 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-169-207:734082:734272 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-169-207:734082:734272 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-169-207:734082:734272 [7] NCCL INFO Trees [0] -1/-1/-1->7->6 [1] -1/-1/-1->7->6 [2] -1/-1/-1->7->6 [3] -1/-1/-1->7->6 [4] -1/-1/-1->7->6 [5] -1/-1/-1->7->6 [6] -1/-1/-1->7->6 [7] -1/-1/-1->7->6 [8] -1/-1/-1->7->6 [9] -1/-1/-1->7->6 [10] -1/-1/-1->7->6 [11] -1/-1/-1->7->6 [12] -1/-1/-1->7->6 [13] -1/-1/-1->7->6 [14] -1/-1/-1->7->6 [15] -1/-1/-1->7->6 [16] -1/-1/-1->7->6 [17] -1/-1/-1->7->6 [18] -1/-1/-1->7->6 [19] -1/-1/-1->7->6 [20] -1/-1/-1->7->6 [21] -1/-1/-1->7->6 [22] -1/-1/-1->7->6 [23] -1/-1/-1->7->6
+ip-26-0-169-207:734082:734272 [7] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-169-207:734077:734271 [2] NCCL INFO Trees [0] 3/-1/-1->2->1 [1] 3/-1/-1->2->1 [2] 3/-1/-1->2->1 [3] 3/-1/-1->2->1 [4] 3/-1/-1->2->1 [5] 3/-1/-1->2->1 [6] 3/-1/-1->2->1 [7] 3/-1/-1->2->1 [8] 3/-1/-1->2->1 [9] 3/-1/-1->2->1 [10] 3/-1/-1->2->1 [11] 3/-1/-1->2->1 [12] 3/-1/-1->2->1 [13] 3/-1/-1->2->1 [14] 3/-1/-1->2->1 [15] 3/-1/-1->2->1 [16] 3/-1/-1->2->1 [17] 3/-1/-1->2->1 [18] 3/-1/-1->2->1 [19] 3/-1/-1->2->1 [20] 3/-1/-1->2->1 [21] 3/-1/-1->2->1 [22] 3/-1/-1->2->1 [23] 3/-1/-1->2->1
+ip-26-0-169-207:734075:734267 [0] NCCL INFO Channel 00/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:734076:734268 [1] NCCL INFO Trees [0] 2/-1/-1->1->0 [1] 2/-1/-1->1->0 [2] 2/-1/-1->1->0 [3] 2/-1/-1->1->0 [4] 2/-1/-1->1->0 [5] 2/-1/-1->1->0 [6] 2/-1/-1->1->0 [7] 2/-1/-1->1->0 [8] 2/-1/-1->1->0 [9] 2/-1/-1->1->0 [10] 2/-1/-1->1->0 [11] 2/-1/-1->1->0 [12] 2/-1/-1->1->0 [13] 2/-1/-1->1->0 [14] 2/-1/-1->1->0 [15] 2/-1/-1->1->0 [16] 2/-1/-1->1->0 [17] 2/-1/-1->1->0 [18] 2/-1/-1->1->0 [19] 2/-1/-1->1->0 [20] 2/-1/-1->1->0 [21] 2/-1/-1->1->0 [22] 2/-1/-1->1->0 [23] 2/-1/-1->1->0
+ip-26-0-169-207:734077:734271 [2] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-169-207:734075:734267 [0] NCCL INFO Channel 01/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:734076:734268 [1] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-169-207:734075:734267 [0] NCCL INFO Channel 02/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:734075:734267 [0] NCCL INFO Channel 03/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:734081:734269 [6] NCCL INFO Trees [0] 7/-1/-1->6->5 [1] 7/-1/-1->6->5 [2] 7/-1/-1->6->5 [3] 7/-1/-1->6->5 [4] 7/-1/-1->6->5 [5] 7/-1/-1->6->5 [6] 7/-1/-1->6->5 [7] 7/-1/-1->6->5 [8] 7/-1/-1->6->5 [9] 7/-1/-1->6->5 [10] 7/-1/-1->6->5 [11] 7/-1/-1->6->5 [12] 7/-1/-1->6->5 [13] 7/-1/-1->6->5 [14] 7/-1/-1->6->5 [15] 7/-1/-1->6->5 [16] 7/-1/-1->6->5 [17] 7/-1/-1->6->5 [18] 7/-1/-1->6->5 [19] 7/-1/-1->6->5 [20] 7/-1/-1->6->5 [21] 7/-1/-1->6->5 [22] 7/-1/-1->6->5 [23] 7/-1/-1->6->5
+ip-26-0-169-207:734080:734274 [5] NCCL INFO Trees [0] 6/-1/-1->5->4 [1] 6/-1/-1->5->4 [2] 6/-1/-1->5->4 [3] 6/-1/-1->5->4 [4] 6/-1/-1->5->4 [5] 6/-1/-1->5->4 [6] 6/-1/-1->5->4 [7] 6/-1/-1->5->4 [8] 6/-1/-1->5->4 [9] 6/-1/-1->5->4 [10] 6/-1/-1->5->4 [11] 6/-1/-1->5->4 [12] 6/-1/-1->5->4 [13] 6/-1/-1->5->4 [14] 6/-1/-1->5->4 [15] 6/-1/-1->5->4 [16] 6/-1/-1->5->4 [17] 6/-1/-1->5->4 [18] 6/-1/-1->5->4 [19] 6/-1/-1->5->4 [20] 6/-1/-1->5->4 [21] 6/-1/-1->5->4 [22] 6/-1/-1->5->4 [23] 6/-1/-1->5->4
+ip-26-0-169-207:734075:734267 [0] NCCL INFO Channel 04/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:734079:734273 [4] NCCL INFO Trees [0] 5/-1/-1->4->3 [1] 5/-1/-1->4->3 [2] 5/-1/-1->4->3 [3] 5/-1/-1->4->3 [4] 5/-1/-1->4->3 [5] 5/-1/-1->4->3 [6] 5/-1/-1->4->3 [7] 5/-1/-1->4->3 [8] 5/-1/-1->4->3 [9] 5/-1/-1->4->3 [10] 5/-1/-1->4->3 [11] 5/-1/-1->4->3 [12] 5/-1/-1->4->3 [13] 5/-1/-1->4->3 [14] 5/-1/-1->4->3 [15] 5/-1/-1->4->3 [16] 5/-1/-1->4->3 [17] 5/-1/-1->4->3 [18] 5/-1/-1->4->3 [19] 5/-1/-1->4->3 [20] 5/-1/-1->4->3 [21] 5/-1/-1->4->3 [22] 5/-1/-1->4->3 [23] 5/-1/-1->4->3
+ip-26-0-169-207:734081:734269 [6] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-169-207:734075:734267 [0] NCCL INFO Channel 05/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:734080:734274 [5] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-169-207:734079:734273 [4] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-169-207:734075:734267 [0] NCCL INFO Channel 06/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:734075:734267 [0] NCCL INFO Channel 07/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:734075:734267 [0] NCCL INFO Channel 08/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:734075:734267 [0] NCCL INFO Channel 09/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:734075:734267 [0] NCCL INFO Channel 10/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:734075:734267 [0] NCCL INFO Channel 11/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:734078:734270 [3] NCCL INFO Trees [0] 4/-1/-1->3->2 [1] 4/-1/-1->3->2 [2] 4/-1/-1->3->2 [3] 4/-1/-1->3->2 [4] 4/-1/-1->3->2 [5] 4/-1/-1->3->2 [6] 4/-1/-1->3->2 [7] 4/-1/-1->3->2 [8] 4/-1/-1->3->2 [9] 4/-1/-1->3->2 [10] 4/-1/-1->3->2 [11] 4/-1/-1->3->2 [12] 4/-1/-1->3->2 [13] 4/-1/-1->3->2 [14] 4/-1/-1->3->2 [15] 4/-1/-1->3->2 [16] 4/-1/-1->3->2 [17] 4/-1/-1->3->2 [18] 4/-1/-1->3->2 [19] 4/-1/-1->3->2 [20] 4/-1/-1->3->2 [21] 4/-1/-1->3->2 [22] 4/-1/-1->3->2 [23] 4/-1/-1->3->2
+ip-26-0-169-207:734075:734267 [0] NCCL INFO Channel 12/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:734075:734267 [0] NCCL INFO Channel 13/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:734075:734267 [0] NCCL INFO Channel 14/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:734075:734267 [0] NCCL INFO Channel 15/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:734075:734267 [0] NCCL INFO Channel 16/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:734075:734267 [0] NCCL INFO Channel 17/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:734075:734267 [0] NCCL INFO Channel 18/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:734075:734267 [0] NCCL INFO Channel 19/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:734075:734267 [0] NCCL INFO Channel 20/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:734075:734267 [0] NCCL INFO Channel 21/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:734075:734267 [0] NCCL INFO Channel 22/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:734075:734267 [0] NCCL INFO Channel 23/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:734075:734267 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] 1/-1/-1->0->-1 [7] 1/-1/-1->0->-1 [8] 1/-1/-1->0->-1 [9] 1/-1/-1->0->-1 [10] 1/-1/-1->0->-1 [11] 1/-1/-1->0->-1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] 1/-1/-1->0->-1 [19] 1/-1/-1->0->-1 [20] 1/-1/-1->0->-1 [21] 1/-1/-1->0->-1 [22] 1/-1/-1->0->-1 [23] 1/-1/-1->0->-1
+ip-26-0-169-207:734078:734270 [3] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-169-207:734075:734267 [0] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-169-207:734076:734268 [1] NCCL INFO Channel 00/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734082:734272 [7] NCCL INFO Channel 00/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734075:734267 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734077:734271 [2] NCCL INFO Channel 00/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734076:734268 [1] NCCL INFO Channel 01/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734080:734274 [5] NCCL INFO Channel 00/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734082:734272 [7] NCCL INFO Channel 01/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734079:734273 [4] NCCL INFO Channel 00/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734075:734267 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734077:734271 [2] NCCL INFO Channel 01/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734076:734268 [1] NCCL INFO Channel 02/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734080:734274 [5] NCCL INFO Channel 01/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734082:734272 [7] NCCL INFO Channel 02/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734079:734273 [4] NCCL INFO Channel 01/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734075:734267 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734077:734271 [2] NCCL INFO Channel 02/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734076:734268 [1] NCCL INFO Channel 03/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734082:734272 [7] NCCL INFO Channel 03/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734080:734274 [5] NCCL INFO Channel 02/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734079:734273 [4] NCCL INFO Channel 02/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734078:734270 [3] NCCL INFO Channel 00/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734075:734267 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734077:734271 [2] NCCL INFO Channel 03/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734076:734268 [1] NCCL INFO Channel 04/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734082:734272 [7] NCCL INFO Channel 04/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734080:734274 [5] NCCL INFO Channel 03/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734079:734273 [4] NCCL INFO Channel 03/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734078:734270 [3] NCCL INFO Channel 01/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734075:734267 [0] NCCL INFO Channel 04/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734077:734271 [2] NCCL INFO Channel 04/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734076:734268 [1] NCCL INFO Channel 05/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734082:734272 [7] NCCL INFO Channel 05/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734080:734274 [5] NCCL INFO Channel 04/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734079:734273 [4] NCCL INFO Channel 04/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734078:734270 [3] NCCL INFO Channel 02/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734075:734267 [0] NCCL INFO Channel 05/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734077:734271 [2] NCCL INFO Channel 05/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734076:734268 [1] NCCL INFO Channel 06/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734082:734272 [7] NCCL INFO Channel 06/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734080:734274 [5] NCCL INFO Channel 05/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734079:734273 [4] NCCL INFO Channel 05/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734078:734270 [3] NCCL INFO Channel 03/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734075:734267 [0] NCCL INFO Channel 06/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734077:734271 [2] NCCL INFO Channel 06/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734076:734268 [1] NCCL INFO Channel 07/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734081:734269 [6] NCCL INFO Channel 00/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:734082:734272 [7] NCCL INFO Channel 07/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734080:734274 [5] NCCL INFO Channel 06/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734079:734273 [4] NCCL INFO Channel 06/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734078:734270 [3] NCCL INFO Channel 04/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734075:734267 [0] NCCL INFO Channel 07/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734077:734271 [2] NCCL INFO Channel 07/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734076:734268 [1] NCCL INFO Channel 08/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734081:734269 [6] NCCL INFO Channel 01/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:734082:734272 [7] NCCL INFO Channel 08/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734080:734274 [5] NCCL INFO Channel 07/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734079:734273 [4] NCCL INFO Channel 07/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734078:734270 [3] NCCL INFO Channel 05/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734075:734267 [0] NCCL INFO Channel 08/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734077:734271 [2] NCCL INFO Channel 08/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734076:734268 [1] NCCL INFO Channel 09/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734081:734269 [6] NCCL INFO Channel 02/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:734082:734272 [7] NCCL INFO Channel 09/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734080:734274 [5] NCCL INFO Channel 08/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734079:734273 [4] NCCL INFO Channel 08/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734078:734270 [3] NCCL INFO Channel 06/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734075:734267 [0] NCCL INFO Channel 09/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734077:734271 [2] NCCL INFO Channel 09/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734076:734268 [1] NCCL INFO Channel 10/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734082:734272 [7] NCCL INFO Channel 10/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734080:734274 [5] NCCL INFO Channel 09/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734079:734273 [4] NCCL INFO Channel 09/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734081:734269 [6] NCCL INFO Channel 03/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:734078:734270 [3] NCCL INFO Channel 07/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734075:734267 [0] NCCL INFO Channel 10/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734076:734268 [1] NCCL INFO Channel 11/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734077:734271 [2] NCCL INFO Channel 10/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734082:734272 [7] NCCL INFO Channel 11/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734080:734274 [5] NCCL INFO Channel 10/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734079:734273 [4] NCCL INFO Channel 10/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734081:734269 [6] NCCL INFO Channel 04/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:734078:734270 [3] NCCL INFO Channel 08/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734075:734267 [0] NCCL INFO Channel 11/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734076:734268 [1] NCCL INFO Channel 12/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734077:734271 [2] NCCL INFO Channel 11/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734082:734272 [7] NCCL INFO Channel 12/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734080:734274 [5] NCCL INFO Channel 11/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734079:734273 [4] NCCL INFO Channel 11/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734081:734269 [6] NCCL INFO Channel 05/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:734075:734267 [0] NCCL INFO Channel 12/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734078:734270 [3] NCCL INFO Channel 09/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734076:734268 [1] NCCL INFO Channel 13/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734077:734271 [2] NCCL INFO Channel 12/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734082:734272 [7] NCCL INFO Channel 13/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734080:734274 [5] NCCL INFO Channel 12/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734079:734273 [4] NCCL INFO Channel 12/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734081:734269 [6] NCCL INFO Channel 06/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:734075:734267 [0] NCCL INFO Channel 13/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734076:734268 [1] NCCL INFO Channel 14/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734078:734270 [3] NCCL INFO Channel 10/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734077:734271 [2] NCCL INFO Channel 13/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734082:734272 [7] NCCL INFO Channel 14/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734080:734274 [5] NCCL INFO Channel 13/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734079:734273 [4] NCCL INFO Channel 13/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734081:734269 [6] NCCL INFO Channel 07/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:734075:734267 [0] NCCL INFO Channel 14/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734076:734268 [1] NCCL INFO Channel 15/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734078:734270 [3] NCCL INFO Channel 11/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734077:734271 [2] NCCL INFO Channel 14/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734082:734272 [7] NCCL INFO Channel 15/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734080:734274 [5] NCCL INFO Channel 14/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734079:734273 [4] NCCL INFO Channel 14/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734081:734269 [6] NCCL INFO Channel 08/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:734075:734267 [0] NCCL INFO Channel 15/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734076:734268 [1] NCCL INFO Channel 16/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734078:734270 [3] NCCL INFO Channel 12/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734077:734271 [2] NCCL INFO Channel 15/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734082:734272 [7] NCCL INFO Channel 16/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734080:734274 [5] NCCL INFO Channel 15/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734079:734273 [4] NCCL INFO Channel 15/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734081:734269 [6] NCCL INFO Channel 09/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:734075:734267 [0] NCCL INFO Channel 16/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734076:734268 [1] NCCL INFO Channel 17/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734078:734270 [3] NCCL INFO Channel 13/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734077:734271 [2] NCCL INFO Channel 16/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734082:734272 [7] NCCL INFO Channel 17/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734080:734274 [5] NCCL INFO Channel 16/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734079:734273 [4] NCCL INFO Channel 16/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734081:734269 [6] NCCL INFO Channel 10/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:734075:734267 [0] NCCL INFO Channel 17/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734076:734268 [1] NCCL INFO Channel 18/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734077:734271 [2] NCCL INFO Channel 17/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734078:734270 [3] NCCL INFO Channel 14/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734082:734272 [7] NCCL INFO Channel 18/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734080:734274 [5] NCCL INFO Channel 17/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734079:734273 [4] NCCL INFO Channel 17/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734081:734269 [6] NCCL INFO Channel 11/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:734075:734267 [0] NCCL INFO Channel 18/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734076:734268 [1] NCCL INFO Channel 19/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734078:734270 [3] NCCL INFO Channel 15/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734077:734271 [2] NCCL INFO Channel 18/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734082:734272 [7] NCCL INFO Channel 19/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734079:734273 [4] NCCL INFO Channel 18/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734080:734274 [5] NCCL INFO Channel 18/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734081:734269 [6] NCCL INFO Channel 12/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:734075:734267 [0] NCCL INFO Channel 19/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734076:734268 [1] NCCL INFO Channel 20/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734078:734270 [3] NCCL INFO Channel 16/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734077:734271 [2] NCCL INFO Channel 19/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734082:734272 [7] NCCL INFO Channel 20/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734079:734273 [4] NCCL INFO Channel 19/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734080:734274 [5] NCCL INFO Channel 19/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734081:734269 [6] NCCL INFO Channel 13/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:734075:734267 [0] NCCL INFO Channel 20/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734077:734271 [2] NCCL INFO Channel 20/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734076:734268 [1] NCCL INFO Channel 21/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734079:734273 [4] NCCL INFO Channel 20/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734080:734274 [5] NCCL INFO Channel 20/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734075:734267 [0] NCCL INFO Channel 21/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734077:734271 [2] NCCL INFO Channel 21/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734076:734268 [1] NCCL INFO Channel 22/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734079:734273 [4] NCCL INFO Channel 21/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734080:734274 [5] NCCL INFO Channel 21/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734075:734267 [0] NCCL INFO Channel 22/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734077:734271 [2] NCCL INFO Channel 22/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734076:734268 [1] NCCL INFO Channel 23/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734079:734273 [4] NCCL INFO Channel 22/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734080:734274 [5] NCCL INFO Channel 22/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734075:734267 [0] NCCL INFO Channel 23/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734077:734271 [2] NCCL INFO Channel 23/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734079:734273 [4] NCCL INFO Channel 23/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734080:734274 [5] NCCL INFO Channel 23/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734078:734270 [3] NCCL INFO Channel 17/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734082:734272 [7] NCCL INFO Channel 21/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734081:734269 [6] NCCL INFO Channel 14/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:734078:734270 [3] NCCL INFO Channel 18/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734081:734269 [6] NCCL INFO Channel 15/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:734078:734270 [3] NCCL INFO Channel 19/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734082:734272 [7] NCCL INFO Channel 22/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734078:734270 [3] NCCL INFO Channel 20/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734081:734269 [6] NCCL INFO Channel 16/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:734082:734272 [7] NCCL INFO Channel 23/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734081:734269 [6] NCCL INFO Channel 17/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:734081:734269 [6] NCCL INFO Channel 18/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:734078:734270 [3] NCCL INFO Channel 21/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734081:734269 [6] NCCL INFO Channel 19/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:734078:734270 [3] NCCL INFO Channel 22/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734081:734269 [6] NCCL INFO Channel 20/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:734078:734270 [3] NCCL INFO Channel 23/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734081:734269 [6] NCCL INFO Channel 21/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:734081:734269 [6] NCCL INFO Channel 22/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:734081:734269 [6] NCCL INFO Channel 23/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:734076:734268 [1] NCCL INFO Connected all rings
+ip-26-0-169-207:734075:734267 [0] NCCL INFO Connected all rings
+ip-26-0-169-207:734081:734269 [6] NCCL INFO Connected all rings
+ip-26-0-169-207:734078:734270 [3] NCCL INFO Connected all rings
+ip-26-0-169-207:734077:734271 [2] NCCL INFO Connected all rings
+ip-26-0-169-207:734079:734273 [4] NCCL INFO Connected all rings
+ip-26-0-169-207:734080:734274 [5] NCCL INFO Connected all rings
+ip-26-0-169-207:734082:734272 [7] NCCL INFO Connected all rings
+ip-26-0-169-207:734082:734272 [7] NCCL INFO Channel 00/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734082:734272 [7] NCCL INFO Channel 01/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734082:734272 [7] NCCL INFO Channel 02/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734082:734272 [7] NCCL INFO Channel 03/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734082:734272 [7] NCCL INFO Channel 04/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734076:734268 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734076:734268 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734076:734268 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734076:734268 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734076:734268 [1] NCCL INFO Channel 04/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734082:734272 [7] NCCL INFO Channel 05/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734076:734268 [1] NCCL INFO Channel 05/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734082:734272 [7] NCCL INFO Channel 06/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734076:734268 [1] NCCL INFO Channel 06/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734082:734272 [7] NCCL INFO Channel 07/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734076:734268 [1] NCCL INFO Channel 07/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734082:734272 [7] NCCL INFO Channel 08/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734076:734268 [1] NCCL INFO Channel 08/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734082:734272 [7] NCCL INFO Channel 09/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734076:734268 [1] NCCL INFO Channel 09/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734082:734272 [7] NCCL INFO Channel 10/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734076:734268 [1] NCCL INFO Channel 10/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734082:734272 [7] NCCL INFO Channel 11/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734076:734268 [1] NCCL INFO Channel 11/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734082:734272 [7] NCCL INFO Channel 12/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734076:734268 [1] NCCL INFO Channel 12/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734082:734272 [7] NCCL INFO Channel 13/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734079:734273 [4] NCCL INFO Channel 00/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734077:734271 [2] NCCL INFO Channel 00/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734078:734270 [3] NCCL INFO Channel 00/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734076:734268 [1] NCCL INFO Channel 13/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734082:734272 [7] NCCL INFO Channel 14/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734079:734273 [4] NCCL INFO Channel 01/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734077:734271 [2] NCCL INFO Channel 01/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734078:734270 [3] NCCL INFO Channel 01/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734076:734268 [1] NCCL INFO Channel 14/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734082:734272 [7] NCCL INFO Channel 15/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734079:734273 [4] NCCL INFO Channel 02/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734077:734271 [2] NCCL INFO Channel 02/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734078:734270 [3] NCCL INFO Channel 02/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734076:734268 [1] NCCL INFO Channel 15/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734082:734272 [7] NCCL INFO Channel 16/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734079:734273 [4] NCCL INFO Channel 03/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734077:734271 [2] NCCL INFO Channel 03/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734076:734268 [1] NCCL INFO Channel 16/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734078:734270 [3] NCCL INFO Channel 03/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734082:734272 [7] NCCL INFO Channel 17/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734079:734273 [4] NCCL INFO Channel 04/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734081:734269 [6] NCCL INFO Channel 00/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734076:734268 [1] NCCL INFO Channel 17/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734077:734271 [2] NCCL INFO Channel 04/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734078:734270 [3] NCCL INFO Channel 04/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734082:734272 [7] NCCL INFO Channel 18/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734079:734273 [4] NCCL INFO Channel 05/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734081:734269 [6] NCCL INFO Channel 01/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734077:734271 [2] NCCL INFO Channel 05/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734076:734268 [1] NCCL INFO Channel 18/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734078:734270 [3] NCCL INFO Channel 05/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734082:734272 [7] NCCL INFO Channel 19/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734081:734269 [6] NCCL INFO Channel 02/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734079:734273 [4] NCCL INFO Channel 06/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734077:734271 [2] NCCL INFO Channel 06/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734076:734268 [1] NCCL INFO Channel 19/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734078:734270 [3] NCCL INFO Channel 06/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734082:734272 [7] NCCL INFO Channel 20/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734081:734269 [6] NCCL INFO Channel 03/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734079:734273 [4] NCCL INFO Channel 07/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734077:734271 [2] NCCL INFO Channel 07/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734076:734268 [1] NCCL INFO Channel 20/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734078:734270 [3] NCCL INFO Channel 07/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734082:734272 [7] NCCL INFO Channel 21/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734081:734269 [6] NCCL INFO Channel 04/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734079:734273 [4] NCCL INFO Channel 08/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734077:734271 [2] NCCL INFO Channel 08/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734076:734268 [1] NCCL INFO Channel 21/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734078:734270 [3] NCCL INFO Channel 08/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734079:734273 [4] NCCL INFO Channel 09/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734082:734272 [7] NCCL INFO Channel 22/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734077:734271 [2] NCCL INFO Channel 09/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734079:734273 [4] NCCL INFO Channel 10/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734082:734272 [7] NCCL INFO Channel 23/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734077:734271 [2] NCCL INFO Channel 10/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734079:734273 [4] NCCL INFO Channel 11/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734077:734271 [2] NCCL INFO Channel 11/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734079:734273 [4] NCCL INFO Channel 12/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734081:734269 [6] NCCL INFO Channel 05/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734077:734271 [2] NCCL INFO Channel 12/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734079:734273 [4] NCCL INFO Channel 13/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734081:734269 [6] NCCL INFO Channel 06/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734080:734274 [5] NCCL INFO Channel 00/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734077:734271 [2] NCCL INFO Channel 13/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734079:734273 [4] NCCL INFO Channel 14/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734081:734269 [6] NCCL INFO Channel 07/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734077:734271 [2] NCCL INFO Channel 14/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734080:734274 [5] NCCL INFO Channel 01/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734079:734273 [4] NCCL INFO Channel 15/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734081:734269 [6] NCCL INFO Channel 08/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734077:734271 [2] NCCL INFO Channel 15/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734080:734274 [5] NCCL INFO Channel 02/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734079:734273 [4] NCCL INFO Channel 16/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734081:734269 [6] NCCL INFO Channel 09/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734077:734271 [2] NCCL INFO Channel 16/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734080:734274 [5] NCCL INFO Channel 03/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734079:734273 [4] NCCL INFO Channel 17/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734081:734269 [6] NCCL INFO Channel 10/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734076:734268 [1] NCCL INFO Channel 22/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734077:734271 [2] NCCL INFO Channel 17/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734080:734274 [5] NCCL INFO Channel 04/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734079:734273 [4] NCCL INFO Channel 18/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734081:734269 [6] NCCL INFO Channel 11/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734078:734270 [3] NCCL INFO Channel 09/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734076:734268 [1] NCCL INFO Channel 23/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734077:734271 [2] NCCL INFO Channel 18/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734080:734274 [5] NCCL INFO Channel 05/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734079:734273 [4] NCCL INFO Channel 19/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734081:734269 [6] NCCL INFO Channel 12/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734078:734270 [3] NCCL INFO Channel 10/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734077:734271 [2] NCCL INFO Channel 19/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734080:734274 [5] NCCL INFO Channel 06/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734079:734273 [4] NCCL INFO Channel 20/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734081:734269 [6] NCCL INFO Channel 13/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734078:734270 [3] NCCL INFO Channel 11/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734077:734271 [2] NCCL INFO Channel 20/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734080:734274 [5] NCCL INFO Channel 07/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734079:734273 [4] NCCL INFO Channel 21/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734081:734269 [6] NCCL INFO Channel 14/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734078:734270 [3] NCCL INFO Channel 12/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734077:734271 [2] NCCL INFO Channel 21/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734080:734274 [5] NCCL INFO Channel 08/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734081:734269 [6] NCCL INFO Channel 15/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734078:734270 [3] NCCL INFO Channel 13/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734077:734271 [2] NCCL INFO Channel 22/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734080:734274 [5] NCCL INFO Channel 09/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734081:734269 [6] NCCL INFO Channel 16/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734078:734270 [3] NCCL INFO Channel 14/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734077:734271 [2] NCCL INFO Channel 23/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734080:734274 [5] NCCL INFO Channel 10/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734081:734269 [6] NCCL INFO Channel 17/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734078:734270 [3] NCCL INFO Channel 15/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734080:734274 [5] NCCL INFO Channel 11/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734079:734273 [4] NCCL INFO Channel 22/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734081:734269 [6] NCCL INFO Channel 18/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734080:734274 [5] NCCL INFO Channel 12/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734078:734270 [3] NCCL INFO Channel 16/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734079:734273 [4] NCCL INFO Channel 23/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734081:734269 [6] NCCL INFO Channel 19/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734080:734274 [5] NCCL INFO Channel 13/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734078:734270 [3] NCCL INFO Channel 17/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734078:734270 [3] NCCL INFO Channel 18/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734078:734270 [3] NCCL INFO Channel 19/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734078:734270 [3] NCCL INFO Channel 20/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734078:734270 [3] NCCL INFO Channel 21/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734078:734270 [3] NCCL INFO Channel 22/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734078:734270 [3] NCCL INFO Channel 23/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734081:734269 [6] NCCL INFO Channel 20/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734080:734274 [5] NCCL INFO Channel 14/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734081:734269 [6] NCCL INFO Channel 21/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734080:734274 [5] NCCL INFO Channel 15/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734081:734269 [6] NCCL INFO Channel 22/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734080:734274 [5] NCCL INFO Channel 16/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734081:734269 [6] NCCL INFO Channel 23/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734080:734274 [5] NCCL INFO Channel 17/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734080:734274 [5] NCCL INFO Channel 18/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734080:734274 [5] NCCL INFO Channel 19/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734080:734274 [5] NCCL INFO Channel 20/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734080:734274 [5] NCCL INFO Channel 21/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734080:734274 [5] NCCL INFO Channel 22/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734075:734267 [0] NCCL INFO Connected all trees
+ip-26-0-169-207:734075:734267 [0] NCCL INFO NVLS comm 0x93dc430 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-169-207:734080:734274 [5] NCCL INFO Channel 23/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734082:734272 [7] NCCL INFO Connected all trees
+ip-26-0-169-207:734082:734272 [7] NCCL INFO NVLS comm 0x9111820 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-169-207:734076:734268 [1] NCCL INFO Connected all trees
+ip-26-0-169-207:734076:734268 [1] NCCL INFO NVLS comm 0x9a12330 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-169-207:734078:734270 [3] NCCL INFO Connected all trees
+ip-26-0-169-207:734078:734270 [3] NCCL INFO NVLS comm 0xa071700 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-169-207:734077:734271 [2] NCCL INFO Connected all trees
+ip-26-0-169-207:734077:734271 [2] NCCL INFO NVLS comm 0x916d2e0 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-169-207:734081:734269 [6] NCCL INFO Connected all trees
+ip-26-0-169-207:734081:734269 [6] NCCL INFO NVLS comm 0x86769c0 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-169-207:734079:734273 [4] NCCL INFO Connected all trees
+ip-26-0-169-207:734079:734273 [4] NCCL INFO NVLS comm 0x96f4cd0 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-169-207:734080:734274 [5] NCCL INFO Connected all trees
+ip-26-0-169-207:734080:734274 [5] NCCL INFO NVLS comm 0xa286f90 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-169-207:734076:734268 [1] NCCL INFO Connected NVLS tree
+ip-26-0-169-207:734076:734268 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-169-207:734076:734268 [1] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-169-207:734076:734268 [1] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:734082:734272 [7] NCCL INFO Connected NVLS tree
+ip-26-0-169-207:734080:734274 [5] NCCL INFO Connected NVLS tree
+ip-26-0-169-207:734082:734272 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-169-207:734080:734274 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-169-207:734082:734272 [7] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-169-207:734082:734272 [7] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:734080:734274 [5] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-169-207:734080:734274 [5] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:734081:734269 [6] NCCL INFO Connected NVLS tree
+ip-26-0-169-207:734081:734269 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-169-207:734081:734269 [6] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-169-207:734081:734269 [6] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:734079:734273 [4] NCCL INFO Connected NVLS tree
+ip-26-0-169-207:734079:734273 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-169-207:734079:734273 [4] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-169-207:734079:734273 [4] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:734078:734270 [3] NCCL INFO Connected NVLS tree
+ip-26-0-169-207:734078:734270 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-169-207:734075:734267 [0] NCCL INFO Connected NVLS tree
+ip-26-0-169-207:734075:734267 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-169-207:734078:734270 [3] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-169-207:734078:734270 [3] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:734077:734271 [2] NCCL INFO Connected NVLS tree
+ip-26-0-169-207:734075:734267 [0] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-169-207:734075:734267 [0] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:734077:734271 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-169-207:734077:734271 [2] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-169-207:734077:734271 [2] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:734076:734268 [1] NCCL INFO comm 0x9a12330 rank 1 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0xf812bb8dd01e9d71 - Init COMPLETE
+ip-26-0-169-207:734075:734267 [0] NCCL INFO comm 0x93dc430 rank 0 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0xf812bb8dd01e9d71 - Init COMPLETE
+ip-26-0-169-207:734080:734274 [5] NCCL INFO comm 0xa286f90 rank 5 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0xf812bb8dd01e9d71 - Init COMPLETE
+ip-26-0-169-207:734081:734269 [6] NCCL INFO comm 0x86769c0 rank 6 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0xf812bb8dd01e9d71 - Init COMPLETE
+ip-26-0-169-207:734079:734273 [4] NCCL INFO comm 0x96f4cd0 rank 4 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0xf812bb8dd01e9d71 - Init COMPLETE
+ip-26-0-169-207:734082:734272 [7] NCCL INFO comm 0x9111820 rank 7 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0xf812bb8dd01e9d71 - Init COMPLETE
+ip-26-0-169-207:734078:734270 [3] NCCL INFO comm 0xa071700 rank 3 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0xf812bb8dd01e9d71 - Init COMPLETE
+ip-26-0-169-207:734077:734271 [2] NCCL INFO comm 0x916d2e0 rank 2 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0xf812bb8dd01e9d71 - Init COMPLETE
+ip-26-0-169-207:734079:734303 [4] NCCL INFO Using network Libfabric
+ip-26-0-169-207:734078:734302 [3] NCCL INFO Using network Libfabric
+ip-26-0-169-207:734077:734301 [2] NCCL INFO Using network Libfabric
+ip-26-0-169-207:734081:734304 [6] NCCL INFO Using network Libfabric
+ip-26-0-169-207:734075:734300 [0] NCCL INFO Using network Libfabric
+ip-26-0-169-207:734080:734306 [5] NCCL INFO Using network Libfabric
+ip-26-0-169-207:734082:734307 [7] NCCL INFO Using network Libfabric
+ip-26-0-169-207:734076:734305 [1] NCCL INFO Using network Libfabric
+ip-26-0-169-207:734076:734305 [1] NCCL INFO comm 0x9a266c0 rank 1 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0xb22d5335ed94c7fa - Init START
+ip-26-0-169-207:734081:734304 [6] NCCL INFO comm 0x868ae70 rank 6 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0xb22d5335ed94c7fa - Init START
+ip-26-0-169-207:734080:734306 [5] NCCL INFO comm 0xa29b6b0 rank 5 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0xb22d5335ed94c7fa - Init START
+ip-26-0-169-207:734077:734301 [2] NCCL INFO comm 0x91816e0 rank 2 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0xb22d5335ed94c7fa - Init START
+ip-26-0-169-207:734075:734300 [0] NCCL INFO comm 0x93f0940 rank 0 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0xb22d5335ed94c7fa - Init START
+ip-26-0-169-207:734079:734303 [4] NCCL INFO comm 0x9709220 rank 4 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0xb22d5335ed94c7fa - Init START
+ip-26-0-169-207:734082:734307 [7] NCCL INFO comm 0x9125d50 rank 7 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0xb22d5335ed94c7fa - Init START
+ip-26-0-169-207:734078:734302 [3] NCCL INFO comm 0xa084bd0 rank 3 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0xb22d5335ed94c7fa - Init START
+ip-26-0-169-207:734077:734301 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/149
+ip-26-0-169-207:734076:734305 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/149
+ip-26-0-169-207:734081:734304 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/149
+ip-26-0-169-207:734082:734307 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/149
+ip-26-0-169-207:734080:734306 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/149
+ip-26-0-169-207:734078:734302 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/149
+ip-26-0-169-207:734079:734303 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/149
+ip-26-0-169-207:734075:734300 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/162
+ip-26-0-169-207:734076:734305 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-169-207:734076:734305 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-169-207:734079:734303 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-169-207:734079:734303 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-169-207:734075:734300 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-169-207:734075:734300 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-169-207:734080:734306 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-169-207:734080:734306 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-169-207:734082:734307 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-169-207:734082:734307 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-169-207:734078:734302 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-169-207:734078:734302 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-169-207:734077:734301 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-169-207:734077:734301 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-169-207:734081:734304 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-169-207:734081:734304 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-169-207:734081:734304 [6] NCCL INFO Trees [0] 7/-1/-1->6->5 [1] 7/-1/-1->6->5 [2] 7/-1/-1->6->5 [3] 7/-1/-1->6->5 [4] 7/-1/-1->6->5 [5] 7/-1/-1->6->5 [6] 7/-1/-1->6->5 [7] 7/-1/-1->6->5 [8] 7/-1/-1->6->5 [9] 7/-1/-1->6->5 [10] 7/-1/-1->6->5 [11] 7/-1/-1->6->5 [12] 7/-1/-1->6->5 [13] 7/-1/-1->6->5 [14] 7/-1/-1->6->5 [15] 7/-1/-1->6->5 [16] 7/-1/-1->6->5 [17] 7/-1/-1->6->5 [18] 7/-1/-1->6->5 [19] 7/-1/-1->6->5 [20] 7/-1/-1->6->5 [21] 7/-1/-1->6->5 [22] 7/-1/-1->6->5 [23] 7/-1/-1->6->5
+ip-26-0-169-207:734081:734304 [6] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-169-207:734078:734302 [3] NCCL INFO Trees [0] 4/-1/-1->3->2 [1] 4/-1/-1->3->2 [2] 4/-1/-1->3->2 [3] 4/-1/-1->3->2 [4] 4/-1/-1->3->2 [5] 4/-1/-1->3->2 [6] 4/-1/-1->3->2 [7] 4/-1/-1->3->2 [8] 4/-1/-1->3->2 [9] 4/-1/-1->3->2 [10] 4/-1/-1->3->2 [11] 4/-1/-1->3->2 [12] 4/-1/-1->3->2 [13] 4/-1/-1->3->2 [14] 4/-1/-1->3->2 [15] 4/-1/-1->3->2 [16] 4/-1/-1->3->2 [17] 4/-1/-1->3->2 [18] 4/-1/-1->3->2 [19] 4/-1/-1->3->2 [20] 4/-1/-1->3->2 [21] 4/-1/-1->3->2 [22] 4/-1/-1->3->2 [23] 4/-1/-1->3->2
+ip-26-0-169-207:734075:734300 [0] NCCL INFO Channel 00/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:734077:734301 [2] NCCL INFO Trees [0] 3/-1/-1->2->1 [1] 3/-1/-1->2->1 [2] 3/-1/-1->2->1 [3] 3/-1/-1->2->1 [4] 3/-1/-1->2->1 [5] 3/-1/-1->2->1 [6] 3/-1/-1->2->1 [7] 3/-1/-1->2->1 [8] 3/-1/-1->2->1 [9] 3/-1/-1->2->1 [10] 3/-1/-1->2->1 [11] 3/-1/-1->2->1 [12] 3/-1/-1->2->1 [13] 3/-1/-1->2->1 [14] 3/-1/-1->2->1 [15] 3/-1/-1->2->1 [16] 3/-1/-1->2->1 [17] 3/-1/-1->2->1 [18] 3/-1/-1->2->1 [19] 3/-1/-1->2->1 [20] 3/-1/-1->2->1 [21] 3/-1/-1->2->1 [22] 3/-1/-1->2->1 [23] 3/-1/-1->2->1
+ip-26-0-169-207:734078:734302 [3] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-169-207:734075:734300 [0] NCCL INFO Channel 01/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:734077:734301 [2] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-169-207:734075:734300 [0] NCCL INFO Channel 02/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:734082:734307 [7] NCCL INFO Trees [0] -1/-1/-1->7->6 [1] -1/-1/-1->7->6 [2] -1/-1/-1->7->6 [3] -1/-1/-1->7->6 [4] -1/-1/-1->7->6 [5] -1/-1/-1->7->6 [6] -1/-1/-1->7->6 [7] -1/-1/-1->7->6 [8] -1/-1/-1->7->6 [9] -1/-1/-1->7->6 [10] -1/-1/-1->7->6 [11] -1/-1/-1->7->6 [12] -1/-1/-1->7->6 [13] -1/-1/-1->7->6 [14] -1/-1/-1->7->6 [15] -1/-1/-1->7->6 [16] -1/-1/-1->7->6 [17] -1/-1/-1->7->6 [18] -1/-1/-1->7->6 [19] -1/-1/-1->7->6 [20] -1/-1/-1->7->6 [21] -1/-1/-1->7->6 [22] -1/-1/-1->7->6 [23] -1/-1/-1->7->6
+ip-26-0-169-207:734075:734300 [0] NCCL INFO Channel 03/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:734082:734307 [7] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-169-207:734075:734300 [0] NCCL INFO Channel 04/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:734075:734300 [0] NCCL INFO Channel 05/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:734076:734305 [1] NCCL INFO Trees [0] 2/-1/-1->1->0 [1] 2/-1/-1->1->0 [2] 2/-1/-1->1->0 [3] 2/-1/-1->1->0 [4] 2/-1/-1->1->0 [5] 2/-1/-1->1->0 [6] 2/-1/-1->1->0 [7] 2/-1/-1->1->0 [8] 2/-1/-1->1->0 [9] 2/-1/-1->1->0 [10] 2/-1/-1->1->0 [11] 2/-1/-1->1->0 [12] 2/-1/-1->1->0 [13] 2/-1/-1->1->0 [14] 2/-1/-1->1->0 [15] 2/-1/-1->1->0 [16] 2/-1/-1->1->0 [17] 2/-1/-1->1->0 [18] 2/-1/-1->1->0 [19] 2/-1/-1->1->0 [20] 2/-1/-1->1->0 [21] 2/-1/-1->1->0 [22] 2/-1/-1->1->0 [23] 2/-1/-1->1->0
+ip-26-0-169-207:734075:734300 [0] NCCL INFO Channel 06/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:734076:734305 [1] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-169-207:734075:734300 [0] NCCL INFO Channel 07/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:734075:734300 [0] NCCL INFO Channel 08/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:734075:734300 [0] NCCL INFO Channel 09/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:734075:734300 [0] NCCL INFO Channel 10/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:734075:734300 [0] NCCL INFO Channel 11/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:734075:734300 [0] NCCL INFO Channel 12/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:734075:734300 [0] NCCL INFO Channel 13/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:734075:734300 [0] NCCL INFO Channel 14/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:734075:734300 [0] NCCL INFO Channel 15/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:734075:734300 [0] NCCL INFO Channel 16/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:734075:734300 [0] NCCL INFO Channel 17/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:734075:734300 [0] NCCL INFO Channel 18/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:734075:734300 [0] NCCL INFO Channel 19/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:734075:734300 [0] NCCL INFO Channel 20/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:734075:734300 [0] NCCL INFO Channel 21/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:734075:734300 [0] NCCL INFO Channel 22/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:734075:734300 [0] NCCL INFO Channel 23/24 :    0   1   2   3   4   5   6   7
+ip-26-0-169-207:734079:734303 [4] NCCL INFO Trees [0] 5/-1/-1->4->3 [1] 5/-1/-1->4->3 [2] 5/-1/-1->4->3 [3] 5/-1/-1->4->3 [4] 5/-1/-1->4->3 [5] 5/-1/-1->4->3 [6] 5/-1/-1->4->3 [7] 5/-1/-1->4->3 [8] 5/-1/-1->4->3 [9] 5/-1/-1->4->3 [10] 5/-1/-1->4->3 [11] 5/-1/-1->4->3 [12] 5/-1/-1->4->3 [13] 5/-1/-1->4->3 [14] 5/-1/-1->4->3 [15] 5/-1/-1->4->3 [16] 5/-1/-1->4->3 [17] 5/-1/-1->4->3 [18] 5/-1/-1->4->3 [19] 5/-1/-1->4->3 [20] 5/-1/-1->4->3 [21] 5/-1/-1->4->3 [22] 5/-1/-1->4->3 [23] 5/-1/-1->4->3
+ip-26-0-169-207:734079:734303 [4] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-169-207:734075:734300 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] 1/-1/-1->0->-1 [7] 1/-1/-1->0->-1 [8] 1/-1/-1->0->-1 [9] 1/-1/-1->0->-1 [10] 1/-1/-1->0->-1 [11] 1/-1/-1->0->-1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] 1/-1/-1->0->-1 [19] 1/-1/-1->0->-1 [20] 1/-1/-1->0->-1 [21] 1/-1/-1->0->-1 [22] 1/-1/-1->0->-1 [23] 1/-1/-1->0->-1
+ip-26-0-169-207:734075:734300 [0] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-169-207:734080:734306 [5] NCCL INFO Trees [0] 6/-1/-1->5->4 [1] 6/-1/-1->5->4 [2] 6/-1/-1->5->4 [3] 6/-1/-1->5->4 [4] 6/-1/-1->5->4 [5] 6/-1/-1->5->4 [6] 6/-1/-1->5->4 [7] 6/-1/-1->5->4 [8] 6/-1/-1->5->4 [9] 6/-1/-1->5->4 [10] 6/-1/-1->5->4 [11] 6/-1/-1->5->4 [12] 6/-1/-1->5->4 [13] 6/-1/-1->5->4 [14] 6/-1/-1->5->4 [15] 6/-1/-1->5->4 [16] 6/-1/-1->5->4 [17] 6/-1/-1->5->4 [18] 6/-1/-1->5->4 [19] 6/-1/-1->5->4 [20] 6/-1/-1->5->4 [21] 6/-1/-1->5->4 [22] 6/-1/-1->5->4 [23] 6/-1/-1->5->4
+ip-26-0-169-207:734080:734306 [5] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-169-207:734078:734302 [3] NCCL INFO Channel 00/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734082:734307 [7] NCCL INFO Channel 00/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734078:734302 [3] NCCL INFO Channel 01/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734082:734307 [7] NCCL INFO Channel 01/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734078:734302 [3] NCCL INFO Channel 02/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734082:734307 [7] NCCL INFO Channel 02/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734078:734302 [3] NCCL INFO Channel 03/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734080:734306 [5] NCCL INFO Channel 00/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734082:734307 [7] NCCL INFO Channel 03/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734080:734306 [5] NCCL INFO Channel 01/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734078:734302 [3] NCCL INFO Channel 04/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734082:734307 [7] NCCL INFO Channel 04/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734080:734306 [5] NCCL INFO Channel 02/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734078:734302 [3] NCCL INFO Channel 05/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734082:734307 [7] NCCL INFO Channel 05/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734079:734303 [4] NCCL INFO Channel 00/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734080:734306 [5] NCCL INFO Channel 03/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734082:734307 [7] NCCL INFO Channel 06/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734078:734302 [3] NCCL INFO Channel 06/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734079:734303 [4] NCCL INFO Channel 01/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734080:734306 [5] NCCL INFO Channel 04/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734082:734307 [7] NCCL INFO Channel 07/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734078:734302 [3] NCCL INFO Channel 07/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734079:734303 [4] NCCL INFO Channel 02/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734080:734306 [5] NCCL INFO Channel 05/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734082:734307 [7] NCCL INFO Channel 08/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734078:734302 [3] NCCL INFO Channel 08/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734079:734303 [4] NCCL INFO Channel 03/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734076:734305 [1] NCCL INFO Channel 00/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734080:734306 [5] NCCL INFO Channel 06/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734082:734307 [7] NCCL INFO Channel 09/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734079:734303 [4] NCCL INFO Channel 04/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734076:734305 [1] NCCL INFO Channel 01/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734080:734306 [5] NCCL INFO Channel 07/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734082:734307 [7] NCCL INFO Channel 10/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734079:734303 [4] NCCL INFO Channel 05/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734081:734304 [6] NCCL INFO Channel 00/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:734076:734305 [1] NCCL INFO Channel 02/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734080:734306 [5] NCCL INFO Channel 08/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734082:734307 [7] NCCL INFO Channel 11/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734077:734301 [2] NCCL INFO Channel 00/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734079:734303 [4] NCCL INFO Channel 06/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734081:734304 [6] NCCL INFO Channel 01/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:734076:734305 [1] NCCL INFO Channel 03/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734080:734306 [5] NCCL INFO Channel 09/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734082:734307 [7] NCCL INFO Channel 12/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734077:734301 [2] NCCL INFO Channel 01/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734079:734303 [4] NCCL INFO Channel 07/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734081:734304 [6] NCCL INFO Channel 02/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:734076:734305 [1] NCCL INFO Channel 04/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734080:734306 [5] NCCL INFO Channel 10/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734078:734302 [3] NCCL INFO Channel 09/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734082:734307 [7] NCCL INFO Channel 13/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734077:734301 [2] NCCL INFO Channel 02/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734079:734303 [4] NCCL INFO Channel 08/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734081:734304 [6] NCCL INFO Channel 03/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:734076:734305 [1] NCCL INFO Channel 05/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734080:734306 [5] NCCL INFO Channel 11/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734078:734302 [3] NCCL INFO Channel 10/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734082:734307 [7] NCCL INFO Channel 14/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734077:734301 [2] NCCL INFO Channel 03/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734079:734303 [4] NCCL INFO Channel 09/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734081:734304 [6] NCCL INFO Channel 04/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:734076:734305 [1] NCCL INFO Channel 06/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734080:734306 [5] NCCL INFO Channel 12/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734078:734302 [3] NCCL INFO Channel 11/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734082:734307 [7] NCCL INFO Channel 15/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734075:734300 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734077:734301 [2] NCCL INFO Channel 04/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734079:734303 [4] NCCL INFO Channel 10/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734081:734304 [6] NCCL INFO Channel 05/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:734080:734306 [5] NCCL INFO Channel 13/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734078:734302 [3] NCCL INFO Channel 12/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734082:734307 [7] NCCL INFO Channel 16/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734075:734300 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734076:734305 [1] NCCL INFO Channel 07/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734077:734301 [2] NCCL INFO Channel 05/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734079:734303 [4] NCCL INFO Channel 11/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734081:734304 [6] NCCL INFO Channel 06/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:734080:734306 [5] NCCL INFO Channel 14/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734078:734302 [3] NCCL INFO Channel 13/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734082:734307 [7] NCCL INFO Channel 17/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734075:734300 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734077:734301 [2] NCCL INFO Channel 06/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734079:734303 [4] NCCL INFO Channel 12/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734081:734304 [6] NCCL INFO Channel 07/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:734080:734306 [5] NCCL INFO Channel 15/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734082:734307 [7] NCCL INFO Channel 18/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734075:734300 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734078:734302 [3] NCCL INFO Channel 14/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734077:734301 [2] NCCL INFO Channel 07/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734079:734303 [4] NCCL INFO Channel 13/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734081:734304 [6] NCCL INFO Channel 08/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:734080:734306 [5] NCCL INFO Channel 16/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734082:734307 [7] NCCL INFO Channel 19/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734077:734301 [2] NCCL INFO Channel 08/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734075:734300 [0] NCCL INFO Channel 04/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734078:734302 [3] NCCL INFO Channel 15/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734079:734303 [4] NCCL INFO Channel 14/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734081:734304 [6] NCCL INFO Channel 09/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:734080:734306 [5] NCCL INFO Channel 17/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734082:734307 [7] NCCL INFO Channel 20/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734078:734302 [3] NCCL INFO Channel 16/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734079:734303 [4] NCCL INFO Channel 15/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734081:734304 [6] NCCL INFO Channel 10/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:734080:734306 [5] NCCL INFO Channel 18/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734082:734307 [7] NCCL INFO Channel 21/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734078:734302 [3] NCCL INFO Channel 17/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734079:734303 [4] NCCL INFO Channel 16/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734081:734304 [6] NCCL INFO Channel 11/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:734080:734306 [5] NCCL INFO Channel 19/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734082:734307 [7] NCCL INFO Channel 22/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734078:734302 [3] NCCL INFO Channel 18/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734079:734303 [4] NCCL INFO Channel 17/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734081:734304 [6] NCCL INFO Channel 12/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:734076:734305 [1] NCCL INFO Channel 08/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734080:734306 [5] NCCL INFO Channel 20/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734082:734307 [7] NCCL INFO Channel 23/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734078:734302 [3] NCCL INFO Channel 19/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734079:734303 [4] NCCL INFO Channel 18/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734081:734304 [6] NCCL INFO Channel 13/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:734076:734305 [1] NCCL INFO Channel 09/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734080:734306 [5] NCCL INFO Channel 21/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734078:734302 [3] NCCL INFO Channel 20/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734079:734303 [4] NCCL INFO Channel 19/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734081:734304 [6] NCCL INFO Channel 14/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:734076:734305 [1] NCCL INFO Channel 10/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734080:734306 [5] NCCL INFO Channel 22/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734078:734302 [3] NCCL INFO Channel 21/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734079:734303 [4] NCCL INFO Channel 20/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734076:734305 [1] NCCL INFO Channel 11/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734081:734304 [6] NCCL INFO Channel 15/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:734080:734306 [5] NCCL INFO Channel 23/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734078:734302 [3] NCCL INFO Channel 22/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734079:734303 [4] NCCL INFO Channel 21/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734081:734304 [6] NCCL INFO Channel 16/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:734076:734305 [1] NCCL INFO Channel 12/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734078:734302 [3] NCCL INFO Channel 23/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734077:734301 [2] NCCL INFO Channel 09/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734079:734303 [4] NCCL INFO Channel 22/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734075:734300 [0] NCCL INFO Channel 05/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734081:734304 [6] NCCL INFO Channel 17/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:734076:734305 [1] NCCL INFO Channel 13/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734079:734303 [4] NCCL INFO Channel 23/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734077:734301 [2] NCCL INFO Channel 10/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734075:734300 [0] NCCL INFO Channel 06/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734081:734304 [6] NCCL INFO Channel 18/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:734076:734305 [1] NCCL INFO Channel 14/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734077:734301 [2] NCCL INFO Channel 11/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734075:734300 [0] NCCL INFO Channel 07/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734081:734304 [6] NCCL INFO Channel 19/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:734076:734305 [1] NCCL INFO Channel 15/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734075:734300 [0] NCCL INFO Channel 08/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734081:734304 [6] NCCL INFO Channel 20/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:734077:734301 [2] NCCL INFO Channel 12/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734076:734305 [1] NCCL INFO Channel 16/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734081:734304 [6] NCCL INFO Channel 21/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:734077:734301 [2] NCCL INFO Channel 13/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734076:734305 [1] NCCL INFO Channel 17/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734075:734300 [0] NCCL INFO Channel 09/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734081:734304 [6] NCCL INFO Channel 22/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:734077:734301 [2] NCCL INFO Channel 14/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734075:734300 [0] NCCL INFO Channel 10/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734076:734305 [1] NCCL INFO Channel 18/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734081:734304 [6] NCCL INFO Channel 23/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-169-207:734075:734300 [0] NCCL INFO Channel 11/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734077:734301 [2] NCCL INFO Channel 15/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734076:734305 [1] NCCL INFO Channel 19/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734075:734300 [0] NCCL INFO Channel 12/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734077:734301 [2] NCCL INFO Channel 16/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734076:734305 [1] NCCL INFO Channel 20/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734075:734300 [0] NCCL INFO Channel 13/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734077:734301 [2] NCCL INFO Channel 17/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734076:734305 [1] NCCL INFO Channel 21/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734076:734305 [1] NCCL INFO Channel 22/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734075:734300 [0] NCCL INFO Channel 14/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734076:734305 [1] NCCL INFO Channel 23/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734077:734301 [2] NCCL INFO Channel 18/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734075:734300 [0] NCCL INFO Channel 15/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734077:734301 [2] NCCL INFO Channel 19/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734075:734300 [0] NCCL INFO Channel 16/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734077:734301 [2] NCCL INFO Channel 20/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734075:734300 [0] NCCL INFO Channel 17/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734077:734301 [2] NCCL INFO Channel 21/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734075:734300 [0] NCCL INFO Channel 18/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734077:734301 [2] NCCL INFO Channel 22/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734075:734300 [0] NCCL INFO Channel 19/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734077:734301 [2] NCCL INFO Channel 23/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734075:734300 [0] NCCL INFO Channel 20/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734075:734300 [0] NCCL INFO Channel 21/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734075:734300 [0] NCCL INFO Channel 22/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734075:734300 [0] NCCL INFO Channel 23/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734079:734303 [4] NCCL INFO Connected all rings
+ip-26-0-169-207:734080:734306 [5] NCCL INFO Connected all rings
+ip-26-0-169-207:734081:734304 [6] NCCL INFO Connected all rings
+ip-26-0-169-207:734076:734305 [1] NCCL INFO Connected all rings
+ip-26-0-169-207:734077:734301 [2] NCCL INFO Connected all rings
+ip-26-0-169-207:734078:734302 [3] NCCL INFO Connected all rings
+ip-26-0-169-207:734075:734300 [0] NCCL INFO Connected all rings
+ip-26-0-169-207:734082:734307 [7] NCCL INFO Connected all rings
+ip-26-0-169-207:734082:734307 [7] NCCL INFO Channel 00/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734082:734307 [7] NCCL INFO Channel 01/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734082:734307 [7] NCCL INFO Channel 02/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734082:734307 [7] NCCL INFO Channel 03/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734079:734303 [4] NCCL INFO Channel 00/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734082:734307 [7] NCCL INFO Channel 04/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734079:734303 [4] NCCL INFO Channel 01/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734082:734307 [7] NCCL INFO Channel 05/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734079:734303 [4] NCCL INFO Channel 02/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734079:734303 [4] NCCL INFO Channel 03/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734082:734307 [7] NCCL INFO Channel 06/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734080:734306 [5] NCCL INFO Channel 00/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734079:734303 [4] NCCL INFO Channel 04/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734082:734307 [7] NCCL INFO Channel 07/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734080:734306 [5] NCCL INFO Channel 01/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734081:734304 [6] NCCL INFO Channel 00/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734079:734303 [4] NCCL INFO Channel 05/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734082:734307 [7] NCCL INFO Channel 08/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734080:734306 [5] NCCL INFO Channel 02/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734081:734304 [6] NCCL INFO Channel 01/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734079:734303 [4] NCCL INFO Channel 06/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734082:734307 [7] NCCL INFO Channel 09/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734080:734306 [5] NCCL INFO Channel 03/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734081:734304 [6] NCCL INFO Channel 02/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734079:734303 [4] NCCL INFO Channel 07/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734082:734307 [7] NCCL INFO Channel 10/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734080:734306 [5] NCCL INFO Channel 04/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734081:734304 [6] NCCL INFO Channel 03/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734079:734303 [4] NCCL INFO Channel 08/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734082:734307 [7] NCCL INFO Channel 11/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734080:734306 [5] NCCL INFO Channel 05/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734079:734303 [4] NCCL INFO Channel 09/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734081:734304 [6] NCCL INFO Channel 04/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734082:734307 [7] NCCL INFO Channel 12/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734079:734303 [4] NCCL INFO Channel 10/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734082:734307 [7] NCCL INFO Channel 13/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734081:734304 [6] NCCL INFO Channel 05/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734080:734306 [5] NCCL INFO Channel 06/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734081:734304 [6] NCCL INFO Channel 06/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734082:734307 [7] NCCL INFO Channel 14/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734079:734303 [4] NCCL INFO Channel 11/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734080:734306 [5] NCCL INFO Channel 07/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734081:734304 [6] NCCL INFO Channel 07/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734082:734307 [7] NCCL INFO Channel 15/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734079:734303 [4] NCCL INFO Channel 12/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734080:734306 [5] NCCL INFO Channel 08/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734081:734304 [6] NCCL INFO Channel 08/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734082:734307 [7] NCCL INFO Channel 16/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734079:734303 [4] NCCL INFO Channel 13/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734080:734306 [5] NCCL INFO Channel 09/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734081:734304 [6] NCCL INFO Channel 09/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734082:734307 [7] NCCL INFO Channel 17/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734079:734303 [4] NCCL INFO Channel 14/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734080:734306 [5] NCCL INFO Channel 10/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734081:734304 [6] NCCL INFO Channel 10/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734082:734307 [7] NCCL INFO Channel 18/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734080:734306 [5] NCCL INFO Channel 11/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734079:734303 [4] NCCL INFO Channel 15/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734078:734302 [3] NCCL INFO Channel 00/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734076:734305 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734077:734301 [2] NCCL INFO Channel 00/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734081:734304 [6] NCCL INFO Channel 11/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734079:734303 [4] NCCL INFO Channel 16/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734082:734307 [7] NCCL INFO Channel 19/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734078:734302 [3] NCCL INFO Channel 01/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734076:734305 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734077:734301 [2] NCCL INFO Channel 01/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734080:734306 [5] NCCL INFO Channel 12/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734081:734304 [6] NCCL INFO Channel 12/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734079:734303 [4] NCCL INFO Channel 17/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734078:734302 [3] NCCL INFO Channel 02/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734082:734307 [7] NCCL INFO Channel 20/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734076:734305 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734077:734301 [2] NCCL INFO Channel 02/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734080:734306 [5] NCCL INFO Channel 13/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734081:734304 [6] NCCL INFO Channel 13/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734079:734303 [4] NCCL INFO Channel 18/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734076:734305 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734078:734302 [3] NCCL INFO Channel 03/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734077:734301 [2] NCCL INFO Channel 03/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734082:734307 [7] NCCL INFO Channel 21/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734080:734306 [5] NCCL INFO Channel 14/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734081:734304 [6] NCCL INFO Channel 14/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734076:734305 [1] NCCL INFO Channel 04/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734078:734302 [3] NCCL INFO Channel 04/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734080:734306 [5] NCCL INFO Channel 15/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734081:734304 [6] NCCL INFO Channel 15/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734076:734305 [1] NCCL INFO Channel 05/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734078:734302 [3] NCCL INFO Channel 05/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734079:734303 [4] NCCL INFO Channel 19/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734082:734307 [7] NCCL INFO Channel 22/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734078:734302 [3] NCCL INFO Channel 06/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734077:734301 [2] NCCL INFO Channel 04/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734082:734307 [7] NCCL INFO Channel 23/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-169-207:734079:734303 [4] NCCL INFO Channel 20/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734078:734302 [3] NCCL INFO Channel 07/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734077:734301 [2] NCCL INFO Channel 05/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734079:734303 [4] NCCL INFO Channel 21/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734077:734301 [2] NCCL INFO Channel 06/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734078:734302 [3] NCCL INFO Channel 08/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734076:734305 [1] NCCL INFO Channel 06/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734079:734303 [4] NCCL INFO Channel 22/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734077:734301 [2] NCCL INFO Channel 07/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734078:734302 [3] NCCL INFO Channel 09/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734080:734306 [5] NCCL INFO Channel 16/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734076:734305 [1] NCCL INFO Channel 07/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734077:734301 [2] NCCL INFO Channel 08/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734079:734303 [4] NCCL INFO Channel 23/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-169-207:734078:734302 [3] NCCL INFO Channel 10/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734080:734306 [5] NCCL INFO Channel 17/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734076:734305 [1] NCCL INFO Channel 08/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734077:734301 [2] NCCL INFO Channel 09/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734078:734302 [3] NCCL INFO Channel 11/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734080:734306 [5] NCCL INFO Channel 18/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734076:734305 [1] NCCL INFO Channel 09/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734081:734304 [6] NCCL INFO Channel 16/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734077:734301 [2] NCCL INFO Channel 10/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734078:734302 [3] NCCL INFO Channel 12/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734080:734306 [5] NCCL INFO Channel 19/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734076:734305 [1] NCCL INFO Channel 10/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734081:734304 [6] NCCL INFO Channel 17/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734077:734301 [2] NCCL INFO Channel 11/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734078:734302 [3] NCCL INFO Channel 13/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734080:734306 [5] NCCL INFO Channel 20/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734076:734305 [1] NCCL INFO Channel 11/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734081:734304 [6] NCCL INFO Channel 18/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734077:734301 [2] NCCL INFO Channel 12/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734078:734302 [3] NCCL INFO Channel 14/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734080:734306 [5] NCCL INFO Channel 21/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734076:734305 [1] NCCL INFO Channel 12/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734081:734304 [6] NCCL INFO Channel 19/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734077:734301 [2] NCCL INFO Channel 13/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734078:734302 [3] NCCL INFO Channel 15/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734080:734306 [5] NCCL INFO Channel 22/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734076:734305 [1] NCCL INFO Channel 13/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734081:734304 [6] NCCL INFO Channel 20/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734078:734302 [3] NCCL INFO Channel 16/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734077:734301 [2] NCCL INFO Channel 14/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734080:734306 [5] NCCL INFO Channel 23/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-169-207:734076:734305 [1] NCCL INFO Channel 14/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734081:734304 [6] NCCL INFO Channel 21/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734078:734302 [3] NCCL INFO Channel 17/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734077:734301 [2] NCCL INFO Channel 15/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734076:734305 [1] NCCL INFO Channel 15/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734081:734304 [6] NCCL INFO Channel 22/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734078:734302 [3] NCCL INFO Channel 18/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734077:734301 [2] NCCL INFO Channel 16/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734076:734305 [1] NCCL INFO Channel 16/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734081:734304 [6] NCCL INFO Channel 23/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-169-207:734078:734302 [3] NCCL INFO Channel 19/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734077:734301 [2] NCCL INFO Channel 17/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734076:734305 [1] NCCL INFO Channel 17/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734078:734302 [3] NCCL INFO Channel 20/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734077:734301 [2] NCCL INFO Channel 18/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734076:734305 [1] NCCL INFO Channel 18/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734077:734301 [2] NCCL INFO Channel 19/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734078:734302 [3] NCCL INFO Channel 21/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734076:734305 [1] NCCL INFO Channel 19/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734077:734301 [2] NCCL INFO Channel 20/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734078:734302 [3] NCCL INFO Channel 22/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734076:734305 [1] NCCL INFO Channel 20/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734076:734305 [1] NCCL INFO Channel 21/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734077:734301 [2] NCCL INFO Channel 21/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734078:734302 [3] NCCL INFO Channel 23/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-169-207:734076:734305 [1] NCCL INFO Channel 22/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734077:734301 [2] NCCL INFO Channel 22/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734076:734305 [1] NCCL INFO Channel 23/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734077:734301 [2] NCCL INFO Channel 23/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734082:734307 [7] NCCL INFO Connected all trees
+ip-26-0-169-207:734082:734307 [7] NCCL INFO NVLS comm 0x9125d50 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-169-207:734075:734300 [0] NCCL INFO Connected all trees
+ip-26-0-169-207:734075:734300 [0] NCCL INFO NVLS comm 0x93f0940 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-169-207:734081:734304 [6] NCCL INFO Connected all trees
+ip-26-0-169-207:734081:734304 [6] NCCL INFO NVLS comm 0x868ae70 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-169-207:734079:734303 [4] NCCL INFO Connected all trees
+ip-26-0-169-207:734079:734303 [4] NCCL INFO NVLS comm 0x9709220 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-169-207:734080:734306 [5] NCCL INFO Connected all trees
+ip-26-0-169-207:734080:734306 [5] NCCL INFO NVLS comm 0xa29b6b0 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-169-207:734077:734301 [2] NCCL INFO Connected all trees
+ip-26-0-169-207:734077:734301 [2] NCCL INFO NVLS comm 0x91816e0 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-169-207:734078:734302 [3] NCCL INFO Connected all trees
+ip-26-0-169-207:734078:734302 [3] NCCL INFO NVLS comm 0xa084bd0 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-169-207:734076:734305 [1] NCCL INFO Connected all trees
+ip-26-0-169-207:734076:734305 [1] NCCL INFO NVLS comm 0x9a266c0 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-169-207:734082:734307 [7] NCCL INFO Connected NVLS tree
+ip-26-0-169-207:734082:734307 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-169-207:734082:734307 [7] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-169-207:734082:734307 [7] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:734076:734305 [1] NCCL INFO Connected NVLS tree
+ip-26-0-169-207:734076:734305 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-169-207:734076:734305 [1] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-169-207:734076:734305 [1] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:734078:734302 [3] NCCL INFO Connected NVLS tree
+ip-26-0-169-207:734078:734302 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-169-207:734075:734300 [0] NCCL INFO Connected NVLS tree
+ip-26-0-169-207:734078:734302 [3] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-169-207:734078:734302 [3] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:734080:734306 [5] NCCL INFO Connected NVLS tree
+ip-26-0-169-207:734075:734300 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-169-207:734080:734306 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-169-207:734080:734306 [5] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-169-207:734080:734306 [5] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:734075:734300 [0] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-169-207:734075:734300 [0] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:734079:734303 [4] NCCL INFO Connected NVLS tree
+ip-26-0-169-207:734079:734303 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-169-207:734079:734303 [4] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-169-207:734077:734301 [2] NCCL INFO Connected NVLS tree
+ip-26-0-169-207:734079:734303 [4] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:734081:734304 [6] NCCL INFO Connected NVLS tree
+ip-26-0-169-207:734077:734301 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-169-207:734081:734304 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-169-207:734077:734301 [2] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-169-207:734077:734301 [2] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:734081:734304 [6] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-169-207:734081:734304 [6] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:734082:734307 [7] NCCL INFO comm 0x9125d50 rank 7 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0xb22d5335ed94c7fa - Init COMPLETE
+ip-26-0-169-207:734078:734302 [3] NCCL INFO comm 0xa084bd0 rank 3 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0xb22d5335ed94c7fa - Init COMPLETE
+ip-26-0-169-207:734080:734306 [5] NCCL INFO comm 0xa29b6b0 rank 5 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0xb22d5335ed94c7fa - Init COMPLETE
+ip-26-0-169-207:734076:734305 [1] NCCL INFO comm 0x9a266c0 rank 1 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0xb22d5335ed94c7fa - Init COMPLETE
+ip-26-0-169-207:734075:734300 [0] NCCL INFO comm 0x93f0940 rank 0 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0xb22d5335ed94c7fa - Init COMPLETE
+ip-26-0-169-207:734077:734301 [2] NCCL INFO comm 0x91816e0 rank 2 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0xb22d5335ed94c7fa - Init COMPLETE
+ip-26-0-169-207:734081:734304 [6] NCCL INFO comm 0x868ae70 rank 6 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0xb22d5335ed94c7fa - Init COMPLETE
+ip-26-0-169-207:734079:734303 [4] NCCL INFO comm 0x9709220 rank 4 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0xb22d5335ed94c7fa - Init COMPLETE
+12/13/2024 14:42:30 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]: Config:
+12/13/2024 14:42:30 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]: Config(general=GeneralArgs(project='debug',
+12/13/2024 14:42:30 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                            run='1.14G_dp4_tp2_pp1_acc256_mbs2_seq2048_zero1_tpmodeRED_vocab32k',
+12/13/2024 14:42:30 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                            seed=42,
+12/13/2024 14:42:30 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                            step=None,
+12/13/2024 14:42:30 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                            consumed_train_samples=None,
+12/13/2024 14:42:30 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                            benchmark_csv_path=PosixPath('benchmark/results/bench_final.csv'),
+12/13/2024 14:42:30 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                            ignore_sanity_checks=True),
+12/13/2024 14:42:30 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:        parallelism=ParallelismArgs(dp=4,
+12/13/2024 14:42:30 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                    pp=1,
+12/13/2024 14:42:30 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                    tp=2,
+12/13/2024 14:42:30 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                    pp_engine=<nanotron.parallel.pipeline_parallel.engine.OneForwardOneBackwardPipelineEngine object at 0x7fecaa99cd00>,
+12/13/2024 14:42:30 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                    tp_mode=<TensorParallelLinearMode.REDUCE_SCATTER: 2>,
+12/13/2024 14:42:30 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                    tp_linear_async_communication=True,
+12/13/2024 14:42:30 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                    recompute_layer=False,
+12/13/2024 14:42:30 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                    tp_recompute_allgather=True,
+12/13/2024 14:42:30 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                    expert_parallel_size=1),
+12/13/2024 14:42:30 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:        model=ModelArgs(model_config=LlamaConfig(bos_token_id=0,
+12/13/2024 14:42:30 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                                 eos_token_id=0,
+12/13/2024 14:42:30 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                                 hidden_act='silu',
+12/13/2024 14:42:30 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                                 hidden_size=2048,
+12/13/2024 14:42:30 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                                 initializer_range=0.02,
+12/13/2024 14:42:30 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                                 intermediate_size=8192,
+12/13/2024 14:42:30 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                                 is_llama_config=True,
+12/13/2024 14:42:30 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                                 max_position_embeddings=2048,
+12/13/2024 14:42:30 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                                 num_attention_heads=32,
+12/13/2024 14:42:30 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                                 num_hidden_layers=16,
+12/13/2024 14:42:30 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                                 num_key_value_heads=8,
+12/13/2024 14:42:30 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                                 pad_token_id=None,
+12/13/2024 14:42:30 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                                 pretraining_tp=1,
+12/13/2024 14:42:30 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                                 rms_norm_eps=1e-05,
+12/13/2024 14:42:30 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                                 rope_scaling=None,
+12/13/2024 14:42:30 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                                 rope_theta=10000.0,
+12/13/2024 14:42:30 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                                 rope_interleaved=False,
+12/13/2024 14:42:30 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                                 tie_word_embeddings=True,
+12/13/2024 14:42:30 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                                 use_cache=True,
+12/13/2024 14:42:30 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                                 vocab_size=32768),
+12/13/2024 14:42:30 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                        init_method=RandomInit(std=0.02),
+12/13/2024 14:42:30 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                        dtype=torch.bfloat16,
+12/13/2024 14:42:30 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                        make_vocab_size_divisible_by=1,
+12/13/2024 14:42:30 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                        ddp_bucket_cap_mb=25),
+12/13/2024 14:42:30 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:        tokenizer=TokenizerArgs(tokenizer_name_or_path='robot-test/dummy-tokenizer-wordlevel',
+12/13/2024 14:42:30 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                tokenizer_revision=None,
+12/13/2024 14:42:30 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                tokenizer_max_length=None),
+12/13/2024 14:42:30 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:        checkpoints=CheckpointsArgs(checkpoints_path=PosixPath('checkpoints'),
+12/13/2024 14:42:30 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                    checkpoint_interval=10000,
+12/13/2024 14:42:30 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                    save_initial_state=False,
+12/13/2024 14:42:30 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                    save_final_state=False,
+12/13/2024 14:42:30 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                    resume_checkpoint_path=None,
+12/13/2024 14:42:30 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                    checkpoints_path_is_shared_file_system=False),
+12/13/2024 14:42:30 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:        logging=LoggingArgs(log_level='info',
+12/13/2024 14:42:30 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                            log_level_replica='info',
+12/13/2024 14:42:30 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                            iteration_step_info_interval=1),
+12/13/2024 14:42:30 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:        tokens=TokensArgs(sequence_length=2048,
+12/13/2024 14:42:30 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                          train_steps=100,
+12/13/2024 14:42:30 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                          micro_batch_size=2,
+12/13/2024 14:42:30 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                          batch_accumulation_per_replica=256,
+12/13/2024 14:42:30 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                          val_check_interval=100,
+12/13/2024 14:42:30 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                          limit_val_batches=0,
+12/13/2024 14:42:30 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                          limit_test_batches=0),
+12/13/2024 14:42:30 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:        optimizer=OptimizerArgs(optimizer_factory=AdamWOptimizerArgs(adam_eps=1e-08,
+12/13/2024 14:42:30 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                                                     adam_beta1=0.9,
+12/13/2024 14:42:30 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                                                     adam_beta2=0.95,
+12/13/2024 14:42:30 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                                                     torch_adam_is_fused=True,
+12/13/2024 14:42:30 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                                                     name='adamW'),
+12/13/2024 14:42:30 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                zero_stage=1,
+12/13/2024 14:42:30 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                weight_decay=0.01,
+12/13/2024 14:42:30 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                clip_grad=1.0,
+12/13/2024 14:42:30 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                accumulate_grad_in_fp32=True,
+12/13/2024 14:42:30 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                learning_rate_scheduler=LRSchedulerArgs(learning_rate=0.0003,
+12/13/2024 14:42:30 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                                                        lr_warmup_steps=2,
+12/13/2024 14:42:30 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                                                        lr_warmup_style='linear',
+12/13/2024 14:42:30 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                                                        lr_decay_style='cosine',
+12/13/2024 14:42:30 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                                                        lr_decay_steps=13,
+12/13/2024 14:42:30 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                                                        lr_decay_starting_step=None,
+12/13/2024 14:42:30 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                                                        min_decay_lr=1e-05)),
+12/13/2024 14:42:30 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:        data_stages=[DatasetStageArgs(name='Stable Training Stage',
+12/13/2024 14:42:30 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                      start_training_step=1,
+12/13/2024 14:42:30 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                      data=DataArgs(dataset=None,
+12/13/2024 14:42:30 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                                    seed=42,
+12/13/2024 14:42:30 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:                                                    num_loading_workers=1))],
+12/13/2024 14:42:30 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:        profiler=None,
+12/13/2024 14:42:30 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:        lighteval=None,
+12/13/2024 14:42:30 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:        s3_upload=None)
+12/13/2024 14:42:30 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]: Model Config:
+12/13/2024 14:42:30 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]: LlamaConfig(bos_token_id=0,
+12/13/2024 14:42:30 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:             eos_token_id=0,
+12/13/2024 14:42:30 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:             hidden_act='silu',
+12/13/2024 14:42:30 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:             hidden_size=2048,
+12/13/2024 14:42:30 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:             initializer_range=0.02,
+12/13/2024 14:42:30 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:             intermediate_size=8192,
+12/13/2024 14:42:30 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:             is_llama_config=True,
+12/13/2024 14:42:30 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:             max_position_embeddings=2048,
+12/13/2024 14:42:30 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:             num_attention_heads=32,
+12/13/2024 14:42:30 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:             num_hidden_layers=16,
+12/13/2024 14:42:30 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:             num_key_value_heads=8,
+12/13/2024 14:42:30 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:             pad_token_id=None,
+12/13/2024 14:42:30 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:             pretraining_tp=1,
+12/13/2024 14:42:30 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:             rms_norm_eps=1e-05,
+12/13/2024 14:42:30 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:             rope_scaling=None,
+12/13/2024 14:42:30 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:             rope_theta=10000.0,
+12/13/2024 14:42:30 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:             rope_interleaved=False,
+12/13/2024 14:42:30 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:             tie_word_embeddings=True,
+12/13/2024 14:42:30 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:             use_cache=True,
+12/13/2024 14:42:30 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:             vocab_size=32768)
+12/13/2024 14:42:30 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]: Building model..
+12/13/2024 14:42:30 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]: Initialize RoPE Theta = 10000.0
+12/13/2024 14:42:30 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]: Setting PP block ranks...
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+ip-26-0-169-207:734075:734359 [0] NCCL INFO Using network Libfabric
+ip-26-0-169-207:734081:734358 [6] NCCL INFO Using network Libfabric
+ip-26-0-169-207:734080:734362 [5] NCCL INFO Using network Libfabric
+ip-26-0-169-207:734079:734357 [4] NCCL INFO Using network Libfabric
+ip-26-0-169-207:734076:734363 [1] NCCL INFO Using network Libfabric
+ip-26-0-169-207:734078:734361 [3] NCCL INFO Using network Libfabric
+ip-26-0-169-207:734077:734356 [2] NCCL INFO Using network Libfabric
+ip-26-0-169-207:734082:734360 [7] NCCL INFO Using network Libfabric
+ip-26-0-169-207:734075:734359 [0] NCCL INFO comm 0xb1f48b0 rank 0 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0x6ae6917772618449 - Init START
+ip-26-0-169-207:734079:734357 [4] NCCL INFO comm 0xb50c4f0 rank 0 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0xad56e344ff6669eb - Init START
+ip-26-0-169-207:734076:734363 [1] NCCL INFO comm 0xb8287b0 rank 1 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0x6ae6917772618449 - Init START
+ip-26-0-169-207:734080:734362 [5] NCCL INFO comm 0xc09d230 rank 1 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0xad56e344ff6669eb - Init START
+ip-26-0-169-207:734079:734357 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/149
+ip-26-0-169-207:734075:734359 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/162
+ip-26-0-169-207:734077:734356 [2] NCCL INFO comm 0xaf86ab0 rank 0 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0xb886be90a6e220ef - Init START
+ip-26-0-169-207:734078:734361 [3] NCCL INFO comm 0xbe867b0 rank 1 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0xb886be90a6e220ef - Init START
+ip-26-0-169-207:734077:734356 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/149
+ip-26-0-169-207:734082:734360 [7] NCCL INFO comm 0xaf2c530 rank 1 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0x2c6bc1e8d21325a0 - Init START
+ip-26-0-169-207:734081:734358 [6] NCCL INFO comm 0xa48dcf0 rank 0 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0x2c6bc1e8d21325a0 - Init START
+ip-26-0-169-207:734081:734358 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/149
+ip-26-0-169-207:734076:734363 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/149
+ip-26-0-169-207:734082:734360 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/149
+ip-26-0-169-207:734080:734362 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/149
+ip-26-0-169-207:734078:734361 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/149
+ip-26-0-169-207:734075:734359 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-169-207:734078:734361 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-169-207:734077:734356 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-169-207:734077:734356 [2] NCCL INFO Channel 00/24 :    0   1
+ip-26-0-169-207:734077:734356 [2] NCCL INFO Channel 01/24 :    0   1
+ip-26-0-169-207:734077:734356 [2] NCCL INFO Channel 02/24 :    0   1
+ip-26-0-169-207:734077:734356 [2] NCCL INFO Channel 03/24 :    0   1
+ip-26-0-169-207:734077:734356 [2] NCCL INFO Channel 04/24 :    0   1
+ip-26-0-169-207:734077:734356 [2] NCCL INFO Channel 05/24 :    0   1
+ip-26-0-169-207:734077:734356 [2] NCCL INFO Channel 06/24 :    0   1
+ip-26-0-169-207:734077:734356 [2] NCCL INFO Channel 07/24 :    0   1
+ip-26-0-169-207:734077:734356 [2] NCCL INFO Channel 08/24 :    0   1
+ip-26-0-169-207:734077:734356 [2] NCCL INFO Channel 09/24 :    0   1
+ip-26-0-169-207:734077:734356 [2] NCCL INFO Channel 10/24 :    0   1
+ip-26-0-169-207:734077:734356 [2] NCCL INFO Channel 11/24 :    0   1
+ip-26-0-169-207:734077:734356 [2] NCCL INFO Channel 12/24 :    0   1
+ip-26-0-169-207:734078:734361 [3] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] -1/-1/-1->1->0 [3] -1/-1/-1->1->0 [4] -1/-1/-1->1->0 [5] -1/-1/-1->1->0 [6] 0/-1/-1->1->-1 [7] 0/-1/-1->1->-1 [8] 0/-1/-1->1->-1 [9] 0/-1/-1->1->-1 [10] 0/-1/-1->1->-1 [11] 0/-1/-1->1->-1 [12] -1/-1/-1->1->0 [13] -1/-1/-1->1->0 [14] -1/-1/-1->1->0 [15] -1/-1/-1->1->0 [16] -1/-1/-1->1->0 [17] -1/-1/-1->1->0 [18] 0/-1/-1->1->-1 [19] 0/-1/-1->1->-1 [20] 0/-1/-1->1->-1 [21] 0/-1/-1->1->-1 [22] 0/-1/-1->1->-1 [23] 0/-1/-1->1->-1
+ip-26-0-169-207:734077:734356 [2] NCCL INFO Channel 13/24 :    0   1
+ip-26-0-169-207:734077:734356 [2] NCCL INFO Channel 14/24 :    0   1
+ip-26-0-169-207:734078:734361 [3] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-169-207:734077:734356 [2] NCCL INFO Channel 15/24 :    0   1
+ip-26-0-169-207:734077:734356 [2] NCCL INFO Channel 16/24 :    0   1
+ip-26-0-169-207:734077:734356 [2] NCCL INFO Channel 17/24 :    0   1
+ip-26-0-169-207:734077:734356 [2] NCCL INFO Channel 18/24 :    0   1
+ip-26-0-169-207:734077:734356 [2] NCCL INFO Channel 19/24 :    0   1
+ip-26-0-169-207:734077:734356 [2] NCCL INFO Channel 20/24 :    0   1
+ip-26-0-169-207:734077:734356 [2] NCCL INFO Channel 21/24 :    0   1
+ip-26-0-169-207:734077:734356 [2] NCCL INFO Channel 22/24 :    0   1
+ip-26-0-169-207:734077:734356 [2] NCCL INFO Channel 23/24 :    0   1
+ip-26-0-169-207:734077:734356 [2] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] -1/-1/-1->0->1 [7] -1/-1/-1->0->1 [8] -1/-1/-1->0->1 [9] -1/-1/-1->0->1 [10] -1/-1/-1->0->1 [11] -1/-1/-1->0->1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] -1/-1/-1->0->1 [19] -1/-1/-1->0->1 [20] -1/-1/-1->0->1 [21] -1/-1/-1->0->1 [22] -1/-1/-1->0->1 [23] -1/-1/-1->0->1
+ip-26-0-169-207:734077:734356 [2] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-169-207:734081:734358 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-169-207:734080:734362 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-169-207:734076:734363 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-169-207:734075:734359 [0] NCCL INFO Channel 00/24 :    0   1
+ip-26-0-169-207:734075:734359 [0] NCCL INFO Channel 01/24 :    0   1
+ip-26-0-169-207:734075:734359 [0] NCCL INFO Channel 02/24 :    0   1
+ip-26-0-169-207:734075:734359 [0] NCCL INFO Channel 03/24 :    0   1
+ip-26-0-169-207:734075:734359 [0] NCCL INFO Channel 04/24 :    0   1
+ip-26-0-169-207:734075:734359 [0] NCCL INFO Channel 05/24 :    0   1
+ip-26-0-169-207:734076:734363 [1] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] -1/-1/-1->1->0 [3] -1/-1/-1->1->0 [4] -1/-1/-1->1->0 [5] -1/-1/-1->1->0 [6] 0/-1/-1->1->-1 [7] 0/-1/-1->1->-1 [8] 0/-1/-1->1->-1 [9] 0/-1/-1->1->-1 [10] 0/-1/-1->1->-1 [11] 0/-1/-1->1->-1 [12] -1/-1/-1->1->0 [13] -1/-1/-1->1->0 [14] -1/-1/-1->1->0 [15] -1/-1/-1->1->0 [16] -1/-1/-1->1->0 [17] -1/-1/-1->1->0 [18] 0/-1/-1->1->-1 [19] 0/-1/-1->1->-1 [20] 0/-1/-1->1->-1 [21] 0/-1/-1->1->-1 [22] 0/-1/-1->1->-1 [23] 0/-1/-1->1->-1
+ip-26-0-169-207:734075:734359 [0] NCCL INFO Channel 06/24 :    0   1
+ip-26-0-169-207:734076:734363 [1] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-169-207:734075:734359 [0] NCCL INFO Channel 07/24 :    0   1
+ip-26-0-169-207:734075:734359 [0] NCCL INFO Channel 08/24 :    0   1
+ip-26-0-169-207:734075:734359 [0] NCCL INFO Channel 09/24 :    0   1
+ip-26-0-169-207:734075:734359 [0] NCCL INFO Channel 10/24 :    0   1
+ip-26-0-169-207:734075:734359 [0] NCCL INFO Channel 11/24 :    0   1
+ip-26-0-169-207:734075:734359 [0] NCCL INFO Channel 12/24 :    0   1
+ip-26-0-169-207:734075:734359 [0] NCCL INFO Channel 13/24 :    0   1
+ip-26-0-169-207:734075:734359 [0] NCCL INFO Channel 14/24 :    0   1
+ip-26-0-169-207:734075:734359 [0] NCCL INFO Channel 15/24 :    0   1
+ip-26-0-169-207:734075:734359 [0] NCCL INFO Channel 16/24 :    0   1
+ip-26-0-169-207:734075:734359 [0] NCCL INFO Channel 17/24 :    0   1
+ip-26-0-169-207:734075:734359 [0] NCCL INFO Channel 18/24 :    0   1
+ip-26-0-169-207:734075:734359 [0] NCCL INFO Channel 19/24 :    0   1
+ip-26-0-169-207:734075:734359 [0] NCCL INFO Channel 20/24 :    0   1
+ip-26-0-169-207:734075:734359 [0] NCCL INFO Channel 21/24 :    0   1
+ip-26-0-169-207:734075:734359 [0] NCCL INFO Channel 22/24 :    0   1
+ip-26-0-169-207:734075:734359 [0] NCCL INFO Channel 23/24 :    0   1
+ip-26-0-169-207:734075:734359 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] -1/-1/-1->0->1 [7] -1/-1/-1->0->1 [8] -1/-1/-1->0->1 [9] -1/-1/-1->0->1 [10] -1/-1/-1->0->1 [11] -1/-1/-1->0->1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] -1/-1/-1->0->1 [19] -1/-1/-1->0->1 [20] -1/-1/-1->0->1 [21] -1/-1/-1->0->1 [22] -1/-1/-1->0->1 [23] -1/-1/-1->0->1
+ip-26-0-169-207:734075:734359 [0] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-169-207:734082:734360 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-169-207:734082:734360 [7] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] -1/-1/-1->1->0 [3] -1/-1/-1->1->0 [4] -1/-1/-1->1->0 [5] -1/-1/-1->1->0 [6] 0/-1/-1->1->-1 [7] 0/-1/-1->1->-1 [8] 0/-1/-1->1->-1 [9] 0/-1/-1->1->-1 [10] 0/-1/-1->1->-1 [11] 0/-1/-1->1->-1 [12] -1/-1/-1->1->0 [13] -1/-1/-1->1->0 [14] -1/-1/-1->1->0 [15] -1/-1/-1->1->0 [16] -1/-1/-1->1->0 [17] -1/-1/-1->1->0 [18] 0/-1/-1->1->-1 [19] 0/-1/-1->1->-1 [20] 0/-1/-1->1->-1 [21] 0/-1/-1->1->-1 [22] 0/-1/-1->1->-1 [23] 0/-1/-1->1->-1
+ip-26-0-169-207:734082:734360 [7] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-169-207:734081:734358 [6] NCCL INFO Channel 00/24 :    0   1
+ip-26-0-169-207:734081:734358 [6] NCCL INFO Channel 01/24 :    0   1
+ip-26-0-169-207:734081:734358 [6] NCCL INFO Channel 02/24 :    0   1
+ip-26-0-169-207:734081:734358 [6] NCCL INFO Channel 03/24 :    0   1
+ip-26-0-169-207:734081:734358 [6] NCCL INFO Channel 04/24 :    0   1
+ip-26-0-169-207:734081:734358 [6] NCCL INFO Channel 05/24 :    0   1
+ip-26-0-169-207:734081:734358 [6] NCCL INFO Channel 06/24 :    0   1
+ip-26-0-169-207:734081:734358 [6] NCCL INFO Channel 07/24 :    0   1
+ip-26-0-169-207:734081:734358 [6] NCCL INFO Channel 08/24 :    0   1
+ip-26-0-169-207:734081:734358 [6] NCCL INFO Channel 09/24 :    0   1
+ip-26-0-169-207:734081:734358 [6] NCCL INFO Channel 10/24 :    0   1
+ip-26-0-169-207:734081:734358 [6] NCCL INFO Channel 11/24 :    0   1
+ip-26-0-169-207:734081:734358 [6] NCCL INFO Channel 12/24 :    0   1
+ip-26-0-169-207:734081:734358 [6] NCCL INFO Channel 13/24 :    0   1
+ip-26-0-169-207:734081:734358 [6] NCCL INFO Channel 14/24 :    0   1
+ip-26-0-169-207:734081:734358 [6] NCCL INFO Channel 15/24 :    0   1
+ip-26-0-169-207:734081:734358 [6] NCCL INFO Channel 16/24 :    0   1
+ip-26-0-169-207:734081:734358 [6] NCCL INFO Channel 17/24 :    0   1
+ip-26-0-169-207:734081:734358 [6] NCCL INFO Channel 18/24 :    0   1
+ip-26-0-169-207:734081:734358 [6] NCCL INFO Channel 19/24 :    0   1
+ip-26-0-169-207:734081:734358 [6] NCCL INFO Channel 20/24 :    0   1
+ip-26-0-169-207:734081:734358 [6] NCCL INFO Channel 21/24 :    0   1
+ip-26-0-169-207:734081:734358 [6] NCCL INFO Channel 22/24 :    0   1
+ip-26-0-169-207:734081:734358 [6] NCCL INFO Channel 23/24 :    0   1
+ip-26-0-169-207:734081:734358 [6] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] -1/-1/-1->0->1 [7] -1/-1/-1->0->1 [8] -1/-1/-1->0->1 [9] -1/-1/-1->0->1 [10] -1/-1/-1->0->1 [11] -1/-1/-1->0->1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] -1/-1/-1->0->1 [19] -1/-1/-1->0->1 [20] -1/-1/-1->0->1 [21] -1/-1/-1->0->1 [22] -1/-1/-1->0->1 [23] -1/-1/-1->0->1
+ip-26-0-169-207:734081:734358 [6] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-169-207:734079:734357 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-169-207:734079:734357 [4] NCCL INFO Channel 00/24 :    0   1
+ip-26-0-169-207:734079:734357 [4] NCCL INFO Channel 01/24 :    0   1
+ip-26-0-169-207:734079:734357 [4] NCCL INFO Channel 02/24 :    0   1
+ip-26-0-169-207:734079:734357 [4] NCCL INFO Channel 03/24 :    0   1
+ip-26-0-169-207:734079:734357 [4] NCCL INFO Channel 04/24 :    0   1
+ip-26-0-169-207:734079:734357 [4] NCCL INFO Channel 05/24 :    0   1
+ip-26-0-169-207:734079:734357 [4] NCCL INFO Channel 06/24 :    0   1
+ip-26-0-169-207:734079:734357 [4] NCCL INFO Channel 07/24 :    0   1
+ip-26-0-169-207:734079:734357 [4] NCCL INFO Channel 08/24 :    0   1
+ip-26-0-169-207:734079:734357 [4] NCCL INFO Channel 09/24 :    0   1
+ip-26-0-169-207:734079:734357 [4] NCCL INFO Channel 10/24 :    0   1
+ip-26-0-169-207:734079:734357 [4] NCCL INFO Channel 11/24 :    0   1
+ip-26-0-169-207:734079:734357 [4] NCCL INFO Channel 12/24 :    0   1
+ip-26-0-169-207:734079:734357 [4] NCCL INFO Channel 13/24 :    0   1
+ip-26-0-169-207:734079:734357 [4] NCCL INFO Channel 14/24 :    0   1
+ip-26-0-169-207:734079:734357 [4] NCCL INFO Channel 15/24 :    0   1
+ip-26-0-169-207:734079:734357 [4] NCCL INFO Channel 16/24 :    0   1
+ip-26-0-169-207:734079:734357 [4] NCCL INFO Channel 17/24 :    0   1
+ip-26-0-169-207:734079:734357 [4] NCCL INFO Channel 18/24 :    0   1
+ip-26-0-169-207:734079:734357 [4] NCCL INFO Channel 19/24 :    0   1
+ip-26-0-169-207:734079:734357 [4] NCCL INFO Channel 20/24 :    0   1
+ip-26-0-169-207:734079:734357 [4] NCCL INFO Channel 21/24 :    0   1
+ip-26-0-169-207:734079:734357 [4] NCCL INFO Channel 22/24 :    0   1
+ip-26-0-169-207:734079:734357 [4] NCCL INFO Channel 23/24 :    0   1
+ip-26-0-169-207:734080:734362 [5] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] -1/-1/-1->1->0 [3] -1/-1/-1->1->0 [4] -1/-1/-1->1->0 [5] -1/-1/-1->1->0 [6] 0/-1/-1->1->-1 [7] 0/-1/-1->1->-1 [8] 0/-1/-1->1->-1 [9] 0/-1/-1->1->-1 [10] 0/-1/-1->1->-1 [11] 0/-1/-1->1->-1 [12] -1/-1/-1->1->0 [13] -1/-1/-1->1->0 [14] -1/-1/-1->1->0 [15] -1/-1/-1->1->0 [16] -1/-1/-1->1->0 [17] -1/-1/-1->1->0 [18] 0/-1/-1->1->-1 [19] 0/-1/-1->1->-1 [20] 0/-1/-1->1->-1 [21] 0/-1/-1->1->-1 [22] 0/-1/-1->1->-1 [23] 0/-1/-1->1->-1
+ip-26-0-169-207:734080:734362 [5] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-169-207:734079:734357 [4] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] -1/-1/-1->0->1 [7] -1/-1/-1->0->1 [8] -1/-1/-1->0->1 [9] -1/-1/-1->0->1 [10] -1/-1/-1->0->1 [11] -1/-1/-1->0->1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] -1/-1/-1->0->1 [19] -1/-1/-1->0->1 [20] -1/-1/-1->0->1 [21] -1/-1/-1->0->1 [22] -1/-1/-1->0->1 [23] -1/-1/-1->0->1
+ip-26-0-169-207:734079:734357 [4] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-169-207:734077:734356 [2] NCCL INFO Channel 00/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-169-207:734077:734356 [2] NCCL INFO Channel 01/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-169-207:734078:734361 [3] NCCL INFO Channel 00/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-169-207:734077:734356 [2] NCCL INFO Channel 02/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-169-207:734078:734361 [3] NCCL INFO Channel 01/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-169-207:734077:734356 [2] NCCL INFO Channel 03/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-169-207:734078:734361 [3] NCCL INFO Channel 02/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-169-207:734077:734356 [2] NCCL INFO Channel 04/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-169-207:734078:734361 [3] NCCL INFO Channel 03/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-169-207:734076:734363 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734077:734356 [2] NCCL INFO Channel 05/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-169-207:734078:734361 [3] NCCL INFO Channel 04/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-169-207:734076:734363 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734077:734356 [2] NCCL INFO Channel 06/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-169-207:734078:734361 [3] NCCL INFO Channel 05/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-169-207:734076:734363 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734077:734356 [2] NCCL INFO Channel 07/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-169-207:734075:734359 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734078:734361 [3] NCCL INFO Channel 06/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-169-207:734076:734363 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734077:734356 [2] NCCL INFO Channel 08/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-169-207:734075:734359 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734078:734361 [3] NCCL INFO Channel 07/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-169-207:734076:734363 [1] NCCL INFO Channel 04/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734077:734356 [2] NCCL INFO Channel 09/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-169-207:734075:734359 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734078:734361 [3] NCCL INFO Channel 08/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-169-207:734076:734363 [1] NCCL INFO Channel 05/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734077:734356 [2] NCCL INFO Channel 10/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-169-207:734075:734359 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734082:734360 [7] NCCL INFO Channel 00/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-169-207:734078:734361 [3] NCCL INFO Channel 09/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-169-207:734076:734363 [1] NCCL INFO Channel 06/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734077:734356 [2] NCCL INFO Channel 11/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-169-207:734075:734359 [0] NCCL INFO Channel 04/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734082:734360 [7] NCCL INFO Channel 01/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-169-207:734078:734361 [3] NCCL INFO Channel 10/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-169-207:734076:734363 [1] NCCL INFO Channel 07/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734077:734356 [2] NCCL INFO Channel 12/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-169-207:734075:734359 [0] NCCL INFO Channel 05/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734082:734360 [7] NCCL INFO Channel 02/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-169-207:734078:734361 [3] NCCL INFO Channel 11/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-169-207:734076:734363 [1] NCCL INFO Channel 08/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734077:734356 [2] NCCL INFO Channel 13/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-169-207:734075:734359 [0] NCCL INFO Channel 06/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734081:734358 [6] NCCL INFO Channel 00/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-169-207:734082:734360 [7] NCCL INFO Channel 03/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-169-207:734078:734361 [3] NCCL INFO Channel 12/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-169-207:734076:734363 [1] NCCL INFO Channel 09/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734077:734356 [2] NCCL INFO Channel 14/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-169-207:734079:734357 [4] NCCL INFO Channel 00/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-169-207:734075:734359 [0] NCCL INFO Channel 07/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734081:734358 [6] NCCL INFO Channel 01/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-169-207:734082:734360 [7] NCCL INFO Channel 04/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-169-207:734078:734361 [3] NCCL INFO Channel 13/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-169-207:734076:734363 [1] NCCL INFO Channel 10/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734077:734356 [2] NCCL INFO Channel 15/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-169-207:734079:734357 [4] NCCL INFO Channel 01/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-169-207:734075:734359 [0] NCCL INFO Channel 08/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734081:734358 [6] NCCL INFO Channel 02/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-169-207:734082:734360 [7] NCCL INFO Channel 05/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-169-207:734078:734361 [3] NCCL INFO Channel 14/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-169-207:734076:734363 [1] NCCL INFO Channel 11/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734077:734356 [2] NCCL INFO Channel 16/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-169-207:734080:734362 [5] NCCL INFO Channel 00/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-169-207:734079:734357 [4] NCCL INFO Channel 02/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-169-207:734075:734359 [0] NCCL INFO Channel 09/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734081:734358 [6] NCCL INFO Channel 03/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-169-207:734082:734360 [7] NCCL INFO Channel 06/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-169-207:734078:734361 [3] NCCL INFO Channel 15/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-169-207:734076:734363 [1] NCCL INFO Channel 12/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734077:734356 [2] NCCL INFO Channel 17/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-169-207:734080:734362 [5] NCCL INFO Channel 01/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-169-207:734079:734357 [4] NCCL INFO Channel 03/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-169-207:734075:734359 [0] NCCL INFO Channel 10/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734081:734358 [6] NCCL INFO Channel 04/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-169-207:734082:734360 [7] NCCL INFO Channel 07/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-169-207:734078:734361 [3] NCCL INFO Channel 16/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-169-207:734076:734363 [1] NCCL INFO Channel 13/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734077:734356 [2] NCCL INFO Channel 18/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-169-207:734079:734357 [4] NCCL INFO Channel 04/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-169-207:734080:734362 [5] NCCL INFO Channel 02/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-169-207:734075:734359 [0] NCCL INFO Channel 11/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734081:734358 [6] NCCL INFO Channel 05/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-169-207:734082:734360 [7] NCCL INFO Channel 08/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-169-207:734078:734361 [3] NCCL INFO Channel 17/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-169-207:734076:734363 [1] NCCL INFO Channel 14/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734079:734357 [4] NCCL INFO Channel 05/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-169-207:734077:734356 [2] NCCL INFO Channel 19/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-169-207:734080:734362 [5] NCCL INFO Channel 03/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-169-207:734075:734359 [0] NCCL INFO Channel 12/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734082:734360 [7] NCCL INFO Channel 09/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-169-207:734077:734356 [2] NCCL INFO Channel 20/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-169-207:734079:734357 [4] NCCL INFO Channel 06/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-169-207:734082:734360 [7] NCCL INFO Channel 10/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-169-207:734076:734363 [1] NCCL INFO Channel 15/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734081:734358 [6] NCCL INFO Channel 06/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-169-207:734077:734356 [2] NCCL INFO Channel 21/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-169-207:734079:734357 [4] NCCL INFO Channel 07/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-169-207:734082:734360 [7] NCCL INFO Channel 11/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-169-207:734076:734363 [1] NCCL INFO Channel 16/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734081:734358 [6] NCCL INFO Channel 07/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-169-207:734079:734357 [4] NCCL INFO Channel 08/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-169-207:734077:734356 [2] NCCL INFO Channel 22/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-169-207:734082:734360 [7] NCCL INFO Channel 12/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-169-207:734076:734363 [1] NCCL INFO Channel 17/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734081:734358 [6] NCCL INFO Channel 08/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-169-207:734079:734357 [4] NCCL INFO Channel 09/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-169-207:734077:734356 [2] NCCL INFO Channel 23/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-169-207:734082:734360 [7] NCCL INFO Channel 13/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-169-207:734076:734363 [1] NCCL INFO Channel 18/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734081:734358 [6] NCCL INFO Channel 09/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-169-207:734079:734357 [4] NCCL INFO Channel 10/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-169-207:734078:734361 [3] NCCL INFO Channel 18/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-169-207:734082:734360 [7] NCCL INFO Channel 14/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-169-207:734076:734363 [1] NCCL INFO Channel 19/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734081:734358 [6] NCCL INFO Channel 10/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-169-207:734080:734362 [5] NCCL INFO Channel 04/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-169-207:734075:734359 [0] NCCL INFO Channel 13/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734079:734357 [4] NCCL INFO Channel 11/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-169-207:734078:734361 [3] NCCL INFO Channel 19/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-169-207:734082:734360 [7] NCCL INFO Channel 15/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-169-207:734076:734363 [1] NCCL INFO Channel 20/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734081:734358 [6] NCCL INFO Channel 11/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-169-207:734080:734362 [5] NCCL INFO Channel 05/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-169-207:734075:734359 [0] NCCL INFO Channel 14/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734079:734357 [4] NCCL INFO Channel 12/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-169-207:734078:734361 [3] NCCL INFO Channel 20/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-169-207:734082:734360 [7] NCCL INFO Channel 16/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-169-207:734076:734363 [1] NCCL INFO Channel 21/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734081:734358 [6] NCCL INFO Channel 12/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-169-207:734080:734362 [5] NCCL INFO Channel 06/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-169-207:734075:734359 [0] NCCL INFO Channel 15/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734079:734357 [4] NCCL INFO Channel 13/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-169-207:734078:734361 [3] NCCL INFO Channel 21/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-169-207:734082:734360 [7] NCCL INFO Channel 17/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-169-207:734076:734363 [1] NCCL INFO Channel 22/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734081:734358 [6] NCCL INFO Channel 13/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-169-207:734080:734362 [5] NCCL INFO Channel 07/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-169-207:734075:734359 [0] NCCL INFO Channel 16/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734079:734357 [4] NCCL INFO Channel 14/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-169-207:734078:734361 [3] NCCL INFO Channel 22/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-169-207:734082:734360 [7] NCCL INFO Channel 18/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-169-207:734076:734363 [1] NCCL INFO Channel 23/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734081:734358 [6] NCCL INFO Channel 14/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-169-207:734080:734362 [5] NCCL INFO Channel 08/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-169-207:734075:734359 [0] NCCL INFO Channel 17/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734079:734357 [4] NCCL INFO Channel 15/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-169-207:734078:734361 [3] NCCL INFO Channel 23/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-169-207:734082:734360 [7] NCCL INFO Channel 19/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-169-207:734081:734358 [6] NCCL INFO Channel 15/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-169-207:734080:734362 [5] NCCL INFO Channel 09/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-169-207:734075:734359 [0] NCCL INFO Channel 18/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734079:734357 [4] NCCL INFO Channel 16/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-169-207:734082:734360 [7] NCCL INFO Channel 20/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-169-207:734081:734358 [6] NCCL INFO Channel 16/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-169-207:734080:734362 [5] NCCL INFO Channel 10/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-169-207:734075:734359 [0] NCCL INFO Channel 19/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734082:734360 [7] NCCL INFO Channel 21/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-169-207:734079:734357 [4] NCCL INFO Channel 17/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-169-207:734081:734358 [6] NCCL INFO Channel 17/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-169-207:734080:734362 [5] NCCL INFO Channel 11/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-169-207:734075:734359 [0] NCCL INFO Channel 20/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734080:734362 [5] NCCL INFO Channel 12/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-169-207:734079:734357 [4] NCCL INFO Channel 18/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-169-207:734082:734360 [7] NCCL INFO Channel 22/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-169-207:734075:734359 [0] NCCL INFO Channel 21/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734080:734362 [5] NCCL INFO Channel 13/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-169-207:734081:734358 [6] NCCL INFO Channel 18/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-169-207:734079:734357 [4] NCCL INFO Channel 19/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-169-207:734082:734360 [7] NCCL INFO Channel 23/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-169-207:734075:734359 [0] NCCL INFO Channel 22/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734080:734362 [5] NCCL INFO Channel 14/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-169-207:734081:734358 [6] NCCL INFO Channel 19/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-169-207:734079:734357 [4] NCCL INFO Channel 20/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-169-207:734075:734359 [0] NCCL INFO Channel 23/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-169-207:734080:734362 [5] NCCL INFO Channel 15/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-169-207:734081:734358 [6] NCCL INFO Channel 20/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-169-207:734079:734357 [4] NCCL INFO Channel 21/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-169-207:734080:734362 [5] NCCL INFO Channel 16/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-169-207:734081:734358 [6] NCCL INFO Channel 21/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-169-207:734079:734357 [4] NCCL INFO Channel 22/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-169-207:734080:734362 [5] NCCL INFO Channel 17/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-169-207:734081:734358 [6] NCCL INFO Channel 22/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-169-207:734079:734357 [4] NCCL INFO Channel 23/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-169-207:734080:734362 [5] NCCL INFO Channel 18/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-169-207:734081:734358 [6] NCCL INFO Channel 23/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-169-207:734080:734362 [5] NCCL INFO Channel 19/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-169-207:734080:734362 [5] NCCL INFO Channel 20/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-169-207:734080:734362 [5] NCCL INFO Channel 21/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-169-207:734080:734362 [5] NCCL INFO Channel 22/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-169-207:734080:734362 [5] NCCL INFO Channel 23/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-169-207:734075:734359 [0] NCCL INFO Connected all rings
+ip-26-0-169-207:734075:734359 [0] NCCL INFO Connected all trees
+ip-26-0-169-207:734075:734359 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-169-207:734075:734359 [0] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-169-207:734075:734359 [0] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:734077:734356 [2] NCCL INFO Connected all rings
+ip-26-0-169-207:734077:734356 [2] NCCL INFO Connected all trees
+ip-26-0-169-207:734077:734356 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-169-207:734077:734356 [2] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-169-207:734077:734356 [2] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:734076:734363 [1] NCCL INFO Connected all rings
+ip-26-0-169-207:734076:734363 [1] NCCL INFO Connected all trees
+ip-26-0-169-207:734076:734363 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-169-207:734076:734363 [1] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-169-207:734076:734363 [1] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:734081:734358 [6] NCCL INFO Connected all rings
+ip-26-0-169-207:734081:734358 [6] NCCL INFO Connected all trees
+ip-26-0-169-207:734081:734358 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-169-207:734081:734358 [6] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-169-207:734081:734358 [6] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:734078:734361 [3] NCCL INFO Connected all rings
+ip-26-0-169-207:734078:734361 [3] NCCL INFO Connected all trees
+ip-26-0-169-207:734078:734361 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-169-207:734078:734361 [3] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-169-207:734078:734361 [3] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:734082:734360 [7] NCCL INFO Connected all rings
+ip-26-0-169-207:734082:734360 [7] NCCL INFO Connected all trees
+ip-26-0-169-207:734082:734360 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-169-207:734082:734360 [7] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-169-207:734082:734360 [7] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:734075:734359 [0] NCCL INFO comm 0xb1f48b0 rank 0 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0x6ae6917772618449 - Init COMPLETE
+ip-26-0-169-207:734076:734363 [1] NCCL INFO comm 0xb8287b0 rank 1 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0x6ae6917772618449 - Init COMPLETE
+ip-26-0-169-207:734077:734356 [2] NCCL INFO comm 0xaf86ab0 rank 0 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0xb886be90a6e220ef - Init COMPLETE
+ip-26-0-169-207:734078:734361 [3] NCCL INFO comm 0xbe867b0 rank 1 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0xb886be90a6e220ef - Init COMPLETE
+NCCL version 2.18.5+cuda12.2
+ip-26-0-169-207:734078:734382 [3] NCCL INFO Using network Libfabric
+ip-26-0-169-207:734077:734383 [2] NCCL INFO Using network Libfabric
+ip-26-0-169-207:734078:734382 [3] NCCL INFO comm 0xbe9a3e0 rank 0 nranks 1 cudaDev 3 nvmlDev 3 busId 86000 commId 0xa0b36d7b31df2412 - Init START
+ip-26-0-169-207:734077:734383 [2] NCCL INFO comm 0xaf9a9b0 rank 0 nranks 1 cudaDev 2 nvmlDev 2 busId 75000 commId 0x358127bf1c808339 - Init START
+ip-26-0-169-207:734078:734382 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/149
+ip-26-0-169-207:734077:734383 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/149
+NCCL version 2.18.5+cuda12.2
+ip-26-0-169-207:734075:734385 [0] NCCL INFO Using network Libfabric
+ip-26-0-169-207:734075:734385 [0] NCCL INFO comm 0xb208750 rank 0 nranks 1 cudaDev 0 nvmlDev 0 busId 53000 commId 0xd03f5c8e35e0e5f2 - Init START
+ip-26-0-169-207:734075:734385 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/162
+ip-26-0-169-207:734076:734387 [1] NCCL INFO Using network Libfabric
+ip-26-0-169-207:734076:734387 [1] NCCL INFO comm 0xb83be50 rank 0 nranks 1 cudaDev 1 nvmlDev 1 busId 64000 commId 0x2a4952bfd31ae6d7 - Init START
+ip-26-0-169-207:734076:734387 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/149
+ip-26-0-169-207:734082:734360 [7] NCCL INFO comm 0xaf2c530 rank 1 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0x2c6bc1e8d21325a0 - Init COMPLETE
+ip-26-0-169-207:734081:734358 [6] NCCL INFO comm 0xa48dcf0 rank 0 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0x2c6bc1e8d21325a0 - Init COMPLETE
+ip-26-0-169-207:734079:734357 [4] NCCL INFO Connected all rings
+ip-26-0-169-207:734079:734357 [4] NCCL INFO Connected all trees
+ip-26-0-169-207:734079:734357 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-169-207:734079:734357 [4] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-169-207:734079:734357 [4] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+NCCL version 2.18.5+cuda12.2
+ip-26-0-169-207:734082:734390 [7] NCCL INFO Using network Libfabric
+ip-26-0-169-207:734082:734390 [7] NCCL INFO comm 0xaf3fb50 rank 0 nranks 1 cudaDev 7 nvmlDev 7 busId ca000 commId 0xd9379f321c367163 - Init START
+ip-26-0-169-207:734082:734390 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/149
+ip-26-0-169-207:734081:734392 [6] NCCL INFO Using network Libfabric
+ip-26-0-169-207:734081:734392 [6] NCCL INFO comm 0xa4a1e40 rank 0 nranks 1 cudaDev 6 nvmlDev 6 busId b9000 commId 0xb71df4298ed52be4 - Init START
+ip-26-0-169-207:734081:734392 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/149
+ip-26-0-169-207:734080:734362 [5] NCCL INFO Connected all rings
+ip-26-0-169-207:734080:734362 [5] NCCL INFO Connected all trees
+ip-26-0-169-207:734080:734362 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-169-207:734080:734362 [5] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-169-207:734080:734362 [5] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:734080:734362 [5] NCCL INFO comm 0xc09d230 rank 1 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0xad56e344ff6669eb - Init COMPLETE
+ip-26-0-169-207:734079:734357 [4] NCCL INFO comm 0xb50c4f0 rank 0 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0xad56e344ff6669eb - Init COMPLETE
+NCCL version 2.18.5+cuda12.2
+ip-26-0-169-207:734079:734395 [4] NCCL INFO Using network Libfabric
+ip-26-0-169-207:734080:734397 [5] NCCL INFO Using network Libfabric
+ip-26-0-169-207:734079:734395 [4] NCCL INFO comm 0xb520640 rank 0 nranks 1 cudaDev 4 nvmlDev 4 busId 97000 commId 0xe22e5eed5cd53a4e - Init START
+ip-26-0-169-207:734079:734395 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/149
+ip-26-0-169-207:734080:734397 [5] NCCL INFO comm 0xc0b0b30 rank 0 nranks 1 cudaDev 5 nvmlDev 5 busId a8000 commId 0x4b10886e911024bc - Init START
+ip-26-0-169-207:734080:734397 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/149
+ip-26-0-169-207:734078:734382 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-169-207:734078:734382 [3] NCCL INFO Channel 00/32 :    0
+ip-26-0-169-207:734078:734382 [3] NCCL INFO Channel 01/32 :    0
+ip-26-0-169-207:734078:734382 [3] NCCL INFO Channel 02/32 :    0
+ip-26-0-169-207:734078:734382 [3] NCCL INFO Channel 03/32 :    0
+ip-26-0-169-207:734078:734382 [3] NCCL INFO Channel 04/32 :    0
+ip-26-0-169-207:734078:734382 [3] NCCL INFO Channel 05/32 :    0
+ip-26-0-169-207:734078:734382 [3] NCCL INFO Channel 06/32 :    0
+ip-26-0-169-207:734078:734382 [3] NCCL INFO Channel 07/32 :    0
+ip-26-0-169-207:734078:734382 [3] NCCL INFO Channel 08/32 :    0
+ip-26-0-169-207:734078:734382 [3] NCCL INFO Channel 09/32 :    0
+ip-26-0-169-207:734078:734382 [3] NCCL INFO Channel 10/32 :    0
+ip-26-0-169-207:734078:734382 [3] NCCL INFO Channel 11/32 :    0
+ip-26-0-169-207:734078:734382 [3] NCCL INFO Channel 12/32 :    0
+ip-26-0-169-207:734078:734382 [3] NCCL INFO Channel 13/32 :    0
+ip-26-0-169-207:734078:734382 [3] NCCL INFO Channel 14/32 :    0
+ip-26-0-169-207:734078:734382 [3] NCCL INFO Channel 15/32 :    0
+ip-26-0-169-207:734078:734382 [3] NCCL INFO Channel 16/32 :    0
+ip-26-0-169-207:734078:734382 [3] NCCL INFO Channel 17/32 :    0
+ip-26-0-169-207:734078:734382 [3] NCCL INFO Channel 18/32 :    0
+ip-26-0-169-207:734078:734382 [3] NCCL INFO Channel 19/32 :    0
+ip-26-0-169-207:734078:734382 [3] NCCL INFO Channel 20/32 :    0
+ip-26-0-169-207:734078:734382 [3] NCCL INFO Channel 21/32 :    0
+ip-26-0-169-207:734078:734382 [3] NCCL INFO Channel 22/32 :    0
+ip-26-0-169-207:734078:734382 [3] NCCL INFO Channel 23/32 :    0
+ip-26-0-169-207:734078:734382 [3] NCCL INFO Channel 24/32 :    0
+ip-26-0-169-207:734078:734382 [3] NCCL INFO Channel 25/32 :    0
+ip-26-0-169-207:734078:734382 [3] NCCL INFO Channel 26/32 :    0
+ip-26-0-169-207:734078:734382 [3] NCCL INFO Channel 27/32 :    0
+ip-26-0-169-207:734078:734382 [3] NCCL INFO Channel 28/32 :    0
+ip-26-0-169-207:734078:734382 [3] NCCL INFO Channel 29/32 :    0
+ip-26-0-169-207:734078:734382 [3] NCCL INFO Channel 30/32 :    0
+ip-26-0-169-207:734078:734382 [3] NCCL INFO Channel 31/32 :    0
+ip-26-0-169-207:734078:734382 [3] NCCL INFO Trees [0] -1/-1/-1->0->-1 [1] -1/-1/-1->0->-1 [2] -1/-1/-1->0->-1 [3] -1/-1/-1->0->-1 [4] -1/-1/-1->0->-1 [5] -1/-1/-1->0->-1 [6] -1/-1/-1->0->-1 [7] -1/-1/-1->0->-1 [8] -1/-1/-1->0->-1 [9] -1/-1/-1->0->-1 [10] -1/-1/-1->0->-1 [11] -1/-1/-1->0->-1 [12] -1/-1/-1->0->-1 [13] -1/-1/-1->0->-1 [14] -1/-1/-1->0->-1 [15] -1/-1/-1->0->-1 [16] -1/-1/-1->0->-1 [17] -1/-1/-1->0->-1 [18] -1/-1/-1->0->-1 [19] -1/-1/-1->0->-1 [20] -1/-1/-1->0->-1 [21] -1/-1/-1->0->-1 [22] -1/-1/-1->0->-1 [23] -1/-1/-1->0->-1 [24] -1/-1/-1->0->-1 [25] -1/-1/-1->0->-1 [26] -1/-1/-1->0->-1 [27] -1/-1/-1->0->-1 [28] -1/-1/-1->0->-1 [29] -1/-1/-1->0->-1 [30] -1/-1/-1->0->-1 [31] -1/-1/-1->0->-1
+ip-26-0-169-207:734078:734382 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-169-207:734078:734382 [3] NCCL INFO Connected all rings
+ip-26-0-169-207:734078:734382 [3] NCCL INFO Connected all trees
+ip-26-0-169-207:734078:734382 [3] NCCL INFO 32 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:734081:734392 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-169-207:734076:734387 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-169-207:734076:734387 [1] NCCL INFO Channel 00/32 :    0
+ip-26-0-169-207:734076:734387 [1] NCCL INFO Channel 01/32 :    0
+ip-26-0-169-207:734076:734387 [1] NCCL INFO Channel 02/32 :    0
+ip-26-0-169-207:734076:734387 [1] NCCL INFO Channel 03/32 :    0
+ip-26-0-169-207:734076:734387 [1] NCCL INFO Channel 04/32 :    0
+ip-26-0-169-207:734076:734387 [1] NCCL INFO Channel 05/32 :    0
+ip-26-0-169-207:734076:734387 [1] NCCL INFO Channel 06/32 :    0
+ip-26-0-169-207:734076:734387 [1] NCCL INFO Channel 07/32 :    0
+ip-26-0-169-207:734076:734387 [1] NCCL INFO Channel 08/32 :    0
+ip-26-0-169-207:734076:734387 [1] NCCL INFO Channel 09/32 :    0
+ip-26-0-169-207:734076:734387 [1] NCCL INFO Channel 10/32 :    0
+ip-26-0-169-207:734076:734387 [1] NCCL INFO Channel 11/32 :    0
+ip-26-0-169-207:734076:734387 [1] NCCL INFO Channel 12/32 :    0
+ip-26-0-169-207:734076:734387 [1] NCCL INFO Channel 13/32 :    0
+ip-26-0-169-207:734076:734387 [1] NCCL INFO Channel 14/32 :    0
+ip-26-0-169-207:734076:734387 [1] NCCL INFO Channel 15/32 :    0
+ip-26-0-169-207:734076:734387 [1] NCCL INFO Channel 16/32 :    0
+ip-26-0-169-207:734076:734387 [1] NCCL INFO Channel 17/32 :    0
+ip-26-0-169-207:734076:734387 [1] NCCL INFO Channel 18/32 :    0
+ip-26-0-169-207:734076:734387 [1] NCCL INFO Channel 19/32 :    0
+ip-26-0-169-207:734076:734387 [1] NCCL INFO Channel 20/32 :    0
+ip-26-0-169-207:734076:734387 [1] NCCL INFO Channel 21/32 :    0
+ip-26-0-169-207:734076:734387 [1] NCCL INFO Channel 22/32 :    0
+ip-26-0-169-207:734076:734387 [1] NCCL INFO Channel 23/32 :    0
+ip-26-0-169-207:734076:734387 [1] NCCL INFO Channel 24/32 :    0
+ip-26-0-169-207:734076:734387 [1] NCCL INFO Channel 25/32 :    0
+ip-26-0-169-207:734076:734387 [1] NCCL INFO Channel 26/32 :    0
+ip-26-0-169-207:734076:734387 [1] NCCL INFO Channel 27/32 :    0
+ip-26-0-169-207:734076:734387 [1] NCCL INFO Channel 28/32 :    0
+ip-26-0-169-207:734076:734387 [1] NCCL INFO Channel 29/32 :    0
+ip-26-0-169-207:734076:734387 [1] NCCL INFO Channel 30/32 :    0
+ip-26-0-169-207:734076:734387 [1] NCCL INFO Channel 31/32 :    0
+ip-26-0-169-207:734076:734387 [1] NCCL INFO Trees [0] -1/-1/-1->0->-1 [1] -1/-1/-1->0->-1 [2] -1/-1/-1->0->-1 [3] -1/-1/-1->0->-1 [4] -1/-1/-1->0->-1 [5] -1/-1/-1->0->-1 [6] -1/-1/-1->0->-1 [7] -1/-1/-1->0->-1 [8] -1/-1/-1->0->-1 [9] -1/-1/-1->0->-1 [10] -1/-1/-1->0->-1 [11] -1/-1/-1->0->-1 [12] -1/-1/-1->0->-1 [13] -1/-1/-1->0->-1 [14] -1/-1/-1->0->-1 [15] -1/-1/-1->0->-1 [16] -1/-1/-1->0->-1 [17] -1/-1/-1->0->-1 [18] -1/-1/-1->0->-1 [19] -1/-1/-1->0->-1 [20] -1/-1/-1->0->-1 [21] -1/-1/-1->0->-1 [22] -1/-1/-1->0->-1 [23] -1/-1/-1->0->-1 [24] -1/-1/-1->0->-1 [25] -1/-1/-1->0->-1 [26] -1/-1/-1->0->-1 [27] -1/-1/-1->0->-1 [28] -1/-1/-1->0->-1 [29] -1/-1/-1->0->-1 [30] -1/-1/-1->0->-1 [31] -1/-1/-1->0->-1
+ip-26-0-169-207:734076:734387 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-169-207:734076:734387 [1] NCCL INFO Connected all rings
+ip-26-0-169-207:734076:734387 [1] NCCL INFO Connected all trees
+ip-26-0-169-207:734076:734387 [1] NCCL INFO 32 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:734081:734392 [6] NCCL INFO Channel 00/32 :    0
+ip-26-0-169-207:734081:734392 [6] NCCL INFO Channel 01/32 :    0
+ip-26-0-169-207:734081:734392 [6] NCCL INFO Channel 02/32 :    0
+ip-26-0-169-207:734081:734392 [6] NCCL INFO Channel 03/32 :    0
+ip-26-0-169-207:734081:734392 [6] NCCL INFO Channel 04/32 :    0
+ip-26-0-169-207:734081:734392 [6] NCCL INFO Channel 05/32 :    0
+ip-26-0-169-207:734081:734392 [6] NCCL INFO Channel 06/32 :    0
+ip-26-0-169-207:734081:734392 [6] NCCL INFO Channel 07/32 :    0
+ip-26-0-169-207:734081:734392 [6] NCCL INFO Channel 08/32 :    0
+ip-26-0-169-207:734081:734392 [6] NCCL INFO Channel 09/32 :    0
+ip-26-0-169-207:734081:734392 [6] NCCL INFO Channel 10/32 :    0
+ip-26-0-169-207:734081:734392 [6] NCCL INFO Channel 11/32 :    0
+ip-26-0-169-207:734081:734392 [6] NCCL INFO Channel 12/32 :    0
+ip-26-0-169-207:734081:734392 [6] NCCL INFO Channel 13/32 :    0
+ip-26-0-169-207:734081:734392 [6] NCCL INFO Channel 14/32 :    0
+ip-26-0-169-207:734081:734392 [6] NCCL INFO Channel 15/32 :    0
+ip-26-0-169-207:734081:734392 [6] NCCL INFO Channel 16/32 :    0
+ip-26-0-169-207:734081:734392 [6] NCCL INFO Channel 17/32 :    0
+ip-26-0-169-207:734081:734392 [6] NCCL INFO Channel 18/32 :    0
+ip-26-0-169-207:734081:734392 [6] NCCL INFO Channel 19/32 :    0
+ip-26-0-169-207:734081:734392 [6] NCCL INFO Channel 20/32 :    0
+ip-26-0-169-207:734081:734392 [6] NCCL INFO Channel 21/32 :    0
+ip-26-0-169-207:734081:734392 [6] NCCL INFO Channel 22/32 :    0
+ip-26-0-169-207:734081:734392 [6] NCCL INFO Channel 23/32 :    0
+ip-26-0-169-207:734081:734392 [6] NCCL INFO Channel 24/32 :    0
+ip-26-0-169-207:734081:734392 [6] NCCL INFO Channel 25/32 :    0
+ip-26-0-169-207:734081:734392 [6] NCCL INFO Channel 26/32 :    0
+ip-26-0-169-207:734081:734392 [6] NCCL INFO Channel 27/32 :    0
+ip-26-0-169-207:734081:734392 [6] NCCL INFO Channel 28/32 :    0
+ip-26-0-169-207:734081:734392 [6] NCCL INFO Channel 29/32 :    0
+ip-26-0-169-207:734081:734392 [6] NCCL INFO Channel 30/32 :    0
+ip-26-0-169-207:734081:734392 [6] NCCL INFO Channel 31/32 :    0
+ip-26-0-169-207:734081:734392 [6] NCCL INFO Trees [0] -1/-1/-1->0->-1 [1] -1/-1/-1->0->-1 [2] -1/-1/-1->0->-1 [3] -1/-1/-1->0->-1 [4] -1/-1/-1->0->-1 [5] -1/-1/-1->0->-1 [6] -1/-1/-1->0->-1 [7] -1/-1/-1->0->-1 [8] -1/-1/-1->0->-1 [9] -1/-1/-1->0->-1 [10] -1/-1/-1->0->-1 [11] -1/-1/-1->0->-1 [12] -1/-1/-1->0->-1 [13] -1/-1/-1->0->-1 [14] -1/-1/-1->0->-1 [15] -1/-1/-1->0->-1 [16] -1/-1/-1->0->-1 [17] -1/-1/-1->0->-1 [18] -1/-1/-1->0->-1 [19] -1/-1/-1->0->-1 [20] -1/-1/-1->0->-1 [21] -1/-1/-1->0->-1 [22] -1/-1/-1->0->-1 [23] -1/-1/-1->0->-1 [24] -1/-1/-1->0->-1 [25] -1/-1/-1->0->-1 [26] -1/-1/-1->0->-1 [27] -1/-1/-1->0->-1 [28] -1/-1/-1->0->-1 [29] -1/-1/-1->0->-1 [30] -1/-1/-1->0->-1 [31] -1/-1/-1->0->-1
+ip-26-0-169-207:734081:734392 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-169-207:734081:734392 [6] NCCL INFO Connected all rings
+ip-26-0-169-207:734081:734392 [6] NCCL INFO Connected all trees
+ip-26-0-169-207:734081:734392 [6] NCCL INFO 32 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:734077:734383 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-169-207:734077:734383 [2] NCCL INFO Channel 00/32 :    0
+ip-26-0-169-207:734077:734383 [2] NCCL INFO Channel 01/32 :    0
+ip-26-0-169-207:734077:734383 [2] NCCL INFO Channel 02/32 :    0
+ip-26-0-169-207:734077:734383 [2] NCCL INFO Channel 03/32 :    0
+ip-26-0-169-207:734077:734383 [2] NCCL INFO Channel 04/32 :    0
+ip-26-0-169-207:734077:734383 [2] NCCL INFO Channel 05/32 :    0
+ip-26-0-169-207:734077:734383 [2] NCCL INFO Channel 06/32 :    0
+ip-26-0-169-207:734077:734383 [2] NCCL INFO Channel 07/32 :    0
+ip-26-0-169-207:734077:734383 [2] NCCL INFO Channel 08/32 :    0
+ip-26-0-169-207:734077:734383 [2] NCCL INFO Channel 09/32 :    0
+ip-26-0-169-207:734077:734383 [2] NCCL INFO Channel 10/32 :    0
+ip-26-0-169-207:734077:734383 [2] NCCL INFO Channel 11/32 :    0
+ip-26-0-169-207:734077:734383 [2] NCCL INFO Channel 12/32 :    0
+ip-26-0-169-207:734077:734383 [2] NCCL INFO Channel 13/32 :    0
+ip-26-0-169-207:734077:734383 [2] NCCL INFO Channel 14/32 :    0
+ip-26-0-169-207:734077:734383 [2] NCCL INFO Channel 15/32 :    0
+ip-26-0-169-207:734077:734383 [2] NCCL INFO Channel 16/32 :    0
+ip-26-0-169-207:734077:734383 [2] NCCL INFO Channel 17/32 :    0
+ip-26-0-169-207:734077:734383 [2] NCCL INFO Channel 18/32 :    0
+ip-26-0-169-207:734077:734383 [2] NCCL INFO Channel 19/32 :    0
+ip-26-0-169-207:734077:734383 [2] NCCL INFO Channel 20/32 :    0
+ip-26-0-169-207:734077:734383 [2] NCCL INFO Channel 21/32 :    0
+ip-26-0-169-207:734077:734383 [2] NCCL INFO Channel 22/32 :    0
+ip-26-0-169-207:734077:734383 [2] NCCL INFO Channel 23/32 :    0
+ip-26-0-169-207:734077:734383 [2] NCCL INFO Channel 24/32 :    0
+ip-26-0-169-207:734077:734383 [2] NCCL INFO Channel 25/32 :    0
+ip-26-0-169-207:734077:734383 [2] NCCL INFO Channel 26/32 :    0
+ip-26-0-169-207:734077:734383 [2] NCCL INFO Channel 27/32 :    0
+ip-26-0-169-207:734077:734383 [2] NCCL INFO Channel 28/32 :    0
+ip-26-0-169-207:734077:734383 [2] NCCL INFO Channel 29/32 :    0
+ip-26-0-169-207:734077:734383 [2] NCCL INFO Channel 30/32 :    0
+ip-26-0-169-207:734077:734383 [2] NCCL INFO Channel 31/32 :    0
+ip-26-0-169-207:734077:734383 [2] NCCL INFO Trees [0] -1/-1/-1->0->-1 [1] -1/-1/-1->0->-1 [2] -1/-1/-1->0->-1 [3] -1/-1/-1->0->-1 [4] -1/-1/-1->0->-1 [5] -1/-1/-1->0->-1 [6] -1/-1/-1->0->-1 [7] -1/-1/-1->0->-1 [8] -1/-1/-1->0->-1 [9] -1/-1/-1->0->-1 [10] -1/-1/-1->0->-1 [11] -1/-1/-1->0->-1 [12] -1/-1/-1->0->-1 [13] -1/-1/-1->0->-1 [14] -1/-1/-1->0->-1 [15] -1/-1/-1->0->-1 [16] -1/-1/-1->0->-1 [17] -1/-1/-1->0->-1 [18] -1/-1/-1->0->-1 [19] -1/-1/-1->0->-1 [20] -1/-1/-1->0->-1 [21] -1/-1/-1->0->-1 [22] -1/-1/-1->0->-1 [23] -1/-1/-1->0->-1 [24] -1/-1/-1->0->-1 [25] -1/-1/-1->0->-1 [26] -1/-1/-1->0->-1 [27] -1/-1/-1->0->-1 [28] -1/-1/-1->0->-1 [29] -1/-1/-1->0->-1 [30] -1/-1/-1->0->-1 [31] -1/-1/-1->0->-1
+ip-26-0-169-207:734077:734383 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-169-207:734077:734383 [2] NCCL INFO Connected all rings
+ip-26-0-169-207:734077:734383 [2] NCCL INFO Connected all trees
+ip-26-0-169-207:734077:734383 [2] NCCL INFO 32 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:734075:734385 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-169-207:734075:734385 [0] NCCL INFO Channel 00/32 :    0
+ip-26-0-169-207:734075:734385 [0] NCCL INFO Channel 01/32 :    0
+ip-26-0-169-207:734075:734385 [0] NCCL INFO Channel 02/32 :    0
+ip-26-0-169-207:734075:734385 [0] NCCL INFO Channel 03/32 :    0
+ip-26-0-169-207:734075:734385 [0] NCCL INFO Channel 04/32 :    0
+ip-26-0-169-207:734075:734385 [0] NCCL INFO Channel 05/32 :    0
+ip-26-0-169-207:734075:734385 [0] NCCL INFO Channel 06/32 :    0
+ip-26-0-169-207:734075:734385 [0] NCCL INFO Channel 07/32 :    0
+ip-26-0-169-207:734075:734385 [0] NCCL INFO Channel 08/32 :    0
+ip-26-0-169-207:734075:734385 [0] NCCL INFO Channel 09/32 :    0
+ip-26-0-169-207:734075:734385 [0] NCCL INFO Channel 10/32 :    0
+ip-26-0-169-207:734075:734385 [0] NCCL INFO Channel 11/32 :    0
+ip-26-0-169-207:734075:734385 [0] NCCL INFO Channel 12/32 :    0
+ip-26-0-169-207:734075:734385 [0] NCCL INFO Channel 13/32 :    0
+ip-26-0-169-207:734075:734385 [0] NCCL INFO Channel 14/32 :    0
+ip-26-0-169-207:734075:734385 [0] NCCL INFO Channel 15/32 :    0
+ip-26-0-169-207:734075:734385 [0] NCCL INFO Channel 16/32 :    0
+ip-26-0-169-207:734075:734385 [0] NCCL INFO Channel 17/32 :    0
+ip-26-0-169-207:734075:734385 [0] NCCL INFO Channel 18/32 :    0
+ip-26-0-169-207:734075:734385 [0] NCCL INFO Channel 19/32 :    0
+ip-26-0-169-207:734075:734385 [0] NCCL INFO Channel 20/32 :    0
+ip-26-0-169-207:734075:734385 [0] NCCL INFO Channel 21/32 :    0
+ip-26-0-169-207:734075:734385 [0] NCCL INFO Channel 22/32 :    0
+ip-26-0-169-207:734075:734385 [0] NCCL INFO Channel 23/32 :    0
+ip-26-0-169-207:734075:734385 [0] NCCL INFO Channel 24/32 :    0
+ip-26-0-169-207:734075:734385 [0] NCCL INFO Channel 25/32 :    0
+ip-26-0-169-207:734075:734385 [0] NCCL INFO Channel 26/32 :    0
+ip-26-0-169-207:734075:734385 [0] NCCL INFO Channel 27/32 :    0
+ip-26-0-169-207:734075:734385 [0] NCCL INFO Channel 28/32 :    0
+ip-26-0-169-207:734075:734385 [0] NCCL INFO Channel 29/32 :    0
+ip-26-0-169-207:734075:734385 [0] NCCL INFO Channel 30/32 :    0
+ip-26-0-169-207:734075:734385 [0] NCCL INFO Channel 31/32 :    0
+ip-26-0-169-207:734075:734385 [0] NCCL INFO Trees [0] -1/-1/-1->0->-1 [1] -1/-1/-1->0->-1 [2] -1/-1/-1->0->-1 [3] -1/-1/-1->0->-1 [4] -1/-1/-1->0->-1 [5] -1/-1/-1->0->-1 [6] -1/-1/-1->0->-1 [7] -1/-1/-1->0->-1 [8] -1/-1/-1->0->-1 [9] -1/-1/-1->0->-1 [10] -1/-1/-1->0->-1 [11] -1/-1/-1->0->-1 [12] -1/-1/-1->0->-1 [13] -1/-1/-1->0->-1 [14] -1/-1/-1->0->-1 [15] -1/-1/-1->0->-1 [16] -1/-1/-1->0->-1 [17] -1/-1/-1->0->-1 [18] -1/-1/-1->0->-1 [19] -1/-1/-1->0->-1 [20] -1/-1/-1->0->-1 [21] -1/-1/-1->0->-1 [22] -1/-1/-1->0->-1 [23] -1/-1/-1->0->-1 [24] -1/-1/-1->0->-1 [25] -1/-1/-1->0->-1 [26] -1/-1/-1->0->-1 [27] -1/-1/-1->0->-1 [28] -1/-1/-1->0->-1 [29] -1/-1/-1->0->-1 [30] -1/-1/-1->0->-1 [31] -1/-1/-1->0->-1
+ip-26-0-169-207:734075:734385 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-169-207:734075:734385 [0] NCCL INFO Connected all rings
+ip-26-0-169-207:734075:734385 [0] NCCL INFO Connected all trees
+ip-26-0-169-207:734075:734385 [0] NCCL INFO 32 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:734081:734392 [6] NCCL INFO comm 0xa4a1e40 rank 0 nranks 1 cudaDev 6 nvmlDev 6 busId b9000 commId 0xb71df4298ed52be4 - Init COMPLETE
+ip-26-0-169-207:734078:734382 [3] NCCL INFO comm 0xbe9a3e0 rank 0 nranks 1 cudaDev 3 nvmlDev 3 busId 86000 commId 0xa0b36d7b31df2412 - Init COMPLETE
+ip-26-0-169-207:734082:734390 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-169-207:734082:734390 [7] NCCL INFO Channel 00/32 :    0
+ip-26-0-169-207:734082:734390 [7] NCCL INFO Channel 01/32 :    0
+ip-26-0-169-207:734082:734390 [7] NCCL INFO Channel 02/32 :    0
+ip-26-0-169-207:734082:734390 [7] NCCL INFO Channel 03/32 :    0
+ip-26-0-169-207:734082:734390 [7] NCCL INFO Channel 04/32 :    0
+ip-26-0-169-207:734082:734390 [7] NCCL INFO Channel 05/32 :    0
+ip-26-0-169-207:734082:734390 [7] NCCL INFO Channel 06/32 :    0
+ip-26-0-169-207:734082:734390 [7] NCCL INFO Channel 07/32 :    0
+ip-26-0-169-207:734082:734390 [7] NCCL INFO Channel 08/32 :    0
+ip-26-0-169-207:734082:734390 [7] NCCL INFO Channel 09/32 :    0
+ip-26-0-169-207:734082:734390 [7] NCCL INFO Channel 10/32 :    0
+ip-26-0-169-207:734082:734390 [7] NCCL INFO Channel 11/32 :    0
+ip-26-0-169-207:734082:734390 [7] NCCL INFO Channel 12/32 :    0
+ip-26-0-169-207:734082:734390 [7] NCCL INFO Channel 13/32 :    0
+ip-26-0-169-207:734082:734390 [7] NCCL INFO Channel 14/32 :    0
+ip-26-0-169-207:734082:734390 [7] NCCL INFO Channel 15/32 :    0
+ip-26-0-169-207:734082:734390 [7] NCCL INFO Channel 16/32 :    0
+ip-26-0-169-207:734082:734390 [7] NCCL INFO Channel 17/32 :    0
+ip-26-0-169-207:734082:734390 [7] NCCL INFO Channel 18/32 :    0
+ip-26-0-169-207:734082:734390 [7] NCCL INFO Channel 19/32 :    0
+ip-26-0-169-207:734082:734390 [7] NCCL INFO Channel 20/32 :    0
+ip-26-0-169-207:734082:734390 [7] NCCL INFO Channel 21/32 :    0
+ip-26-0-169-207:734082:734390 [7] NCCL INFO Channel 22/32 :    0
+ip-26-0-169-207:734082:734390 [7] NCCL INFO Channel 23/32 :    0
+ip-26-0-169-207:734082:734390 [7] NCCL INFO Channel 24/32 :    0
+ip-26-0-169-207:734082:734390 [7] NCCL INFO Channel 25/32 :    0
+ip-26-0-169-207:734082:734390 [7] NCCL INFO Channel 26/32 :    0
+ip-26-0-169-207:734082:734390 [7] NCCL INFO Channel 27/32 :    0
+ip-26-0-169-207:734082:734390 [7] NCCL INFO Channel 28/32 :    0
+ip-26-0-169-207:734082:734390 [7] NCCL INFO Channel 29/32 :    0
+ip-26-0-169-207:734082:734390 [7] NCCL INFO Channel 30/32 :    0
+ip-26-0-169-207:734082:734390 [7] NCCL INFO Channel 31/32 :    0
+ip-26-0-169-207:734082:734390 [7] NCCL INFO Trees [0] -1/-1/-1->0->-1 [1] -1/-1/-1->0->-1 [2] -1/-1/-1->0->-1 [3] -1/-1/-1->0->-1 [4] -1/-1/-1->0->-1 [5] -1/-1/-1->0->-1 [6] -1/-1/-1->0->-1 [7] -1/-1/-1->0->-1 [8] -1/-1/-1->0->-1 [9] -1/-1/-1->0->-1 [10] -1/-1/-1->0->-1 [11] -1/-1/-1->0->-1 [12] -1/-1/-1->0->-1 [13] -1/-1/-1->0->-1 [14] -1/-1/-1->0->-1 [15] -1/-1/-1->0->-1 [16] -1/-1/-1->0->-1 [17] -1/-1/-1->0->-1 [18] -1/-1/-1->0->-1 [19] -1/-1/-1->0->-1 [20] -1/-1/-1->0->-1 [21] -1/-1/-1->0->-1 [22] -1/-1/-1->0->-1 [23] -1/-1/-1->0->-1 [24] -1/-1/-1->0->-1 [25] -1/-1/-1->0->-1 [26] -1/-1/-1->0->-1 [27] -1/-1/-1->0->-1 [28] -1/-1/-1->0->-1 [29] -1/-1/-1->0->-1 [30] -1/-1/-1->0->-1 [31] -1/-1/-1->0->-1
+ip-26-0-169-207:734082:734390 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-169-207:734082:734390 [7] NCCL INFO Connected all rings
+ip-26-0-169-207:734082:734390 [7] NCCL INFO Connected all trees
+ip-26-0-169-207:734082:734390 [7] NCCL INFO 32 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:734076:734387 [1] NCCL INFO comm 0xb83be50 rank 0 nranks 1 cudaDev 1 nvmlDev 1 busId 64000 commId 0x2a4952bfd31ae6d7 - Init COMPLETE
+ip-26-0-169-207:734077:734383 [2] NCCL INFO comm 0xaf9a9b0 rank 0 nranks 1 cudaDev 2 nvmlDev 2 busId 75000 commId 0x358127bf1c808339 - Init COMPLETE
+ip-26-0-169-207:734075:734385 [0] NCCL INFO comm 0xb208750 rank 0 nranks 1 cudaDev 0 nvmlDev 0 busId 53000 commId 0xd03f5c8e35e0e5f2 - Init COMPLETE
+12/13/2024 14:42:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]: Total number of parameters: 1.04G (1984.26MiB)
+12/13/2024 14:42:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]: Local number of parameters: 520M (992.13MiB)
+12/13/2024 14:42:33 [INFO|DP=0|PP=0|TP=1|ip-26-0-169-207]: Local number of parameters: 520M (992.13MiB)
+12/13/2024 14:42:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]: [After model building] Memory usage: 992.15MiB. Peak allocated: 5632.00MiB Peak reserved: 15874.00MiB
+12/13/2024 14:42:33 [INFO|DP=0|PP=0|TP=1|ip-26-0-169-207]: [After model building] Memory usage: 992.15MiB. Peak allocated: 5632.00MiB Peak reserved: 15874.00MiB
+12/13/2024 14:42:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]: No checkpoint path provided.
+12/13/2024 14:42:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]: Parametrizing model parameters using StandardParametrizator
+ip-26-0-169-207:734077:734413 [2] NCCL INFO Using network Libfabric
+ip-26-0-169-207:734075:734412 [0] NCCL INFO Using network Libfabric
+ip-26-0-169-207:734076:734416 [1] NCCL INFO Using network Libfabric
+ip-26-0-169-207:734078:734417 [3] NCCL INFO Using network Libfabric
+ip-26-0-169-207:734079:734395 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-169-207:734079:734395 [4] NCCL INFO Channel 00/32 :    0
+ip-26-0-169-207:734079:734395 [4] NCCL INFO Channel 01/32 :    0
+ip-26-0-169-207:734079:734395 [4] NCCL INFO Channel 02/32 :    0
+ip-26-0-169-207:734079:734395 [4] NCCL INFO Channel 03/32 :    0
+ip-26-0-169-207:734079:734395 [4] NCCL INFO Channel 04/32 :    0
+ip-26-0-169-207:734079:734395 [4] NCCL INFO Channel 05/32 :    0
+ip-26-0-169-207:734079:734395 [4] NCCL INFO Channel 06/32 :    0
+ip-26-0-169-207:734079:734395 [4] NCCL INFO Channel 07/32 :    0
+ip-26-0-169-207:734079:734395 [4] NCCL INFO Channel 08/32 :    0
+ip-26-0-169-207:734079:734395 [4] NCCL INFO Channel 09/32 :    0
+ip-26-0-169-207:734079:734395 [4] NCCL INFO Channel 10/32 :    0
+ip-26-0-169-207:734079:734395 [4] NCCL INFO Channel 11/32 :    0
+ip-26-0-169-207:734079:734395 [4] NCCL INFO Channel 12/32 :    0
+ip-26-0-169-207:734079:734395 [4] NCCL INFO Channel 13/32 :    0
+ip-26-0-169-207:734079:734395 [4] NCCL INFO Channel 14/32 :    0
+ip-26-0-169-207:734079:734395 [4] NCCL INFO Channel 15/32 :    0
+ip-26-0-169-207:734079:734395 [4] NCCL INFO Channel 16/32 :    0
+ip-26-0-169-207:734079:734395 [4] NCCL INFO Channel 17/32 :    0
+ip-26-0-169-207:734079:734395 [4] NCCL INFO Channel 18/32 :    0
+ip-26-0-169-207:734079:734395 [4] NCCL INFO Channel 19/32 :    0
+ip-26-0-169-207:734079:734395 [4] NCCL INFO Channel 20/32 :    0
+ip-26-0-169-207:734079:734395 [4] NCCL INFO Channel 21/32 :    0
+ip-26-0-169-207:734079:734395 [4] NCCL INFO Channel 22/32 :    0
+ip-26-0-169-207:734079:734395 [4] NCCL INFO Channel 23/32 :    0
+ip-26-0-169-207:734079:734395 [4] NCCL INFO Channel 24/32 :    0
+ip-26-0-169-207:734079:734395 [4] NCCL INFO Channel 25/32 :    0
+ip-26-0-169-207:734079:734395 [4] NCCL INFO Channel 26/32 :    0
+ip-26-0-169-207:734079:734395 [4] NCCL INFO Channel 27/32 :    0
+ip-26-0-169-207:734079:734395 [4] NCCL INFO Channel 28/32 :    0
+ip-26-0-169-207:734079:734395 [4] NCCL INFO Channel 29/32 :    0
+ip-26-0-169-207:734079:734395 [4] NCCL INFO Channel 30/32 :    0
+ip-26-0-169-207:734079:734395 [4] NCCL INFO Channel 31/32 :    0
+ip-26-0-169-207:734079:734395 [4] NCCL INFO Trees [0] -1/-1/-1->0->-1 [1] -1/-1/-1->0->-1 [2] -1/-1/-1->0->-1 [3] -1/-1/-1->0->-1 [4] -1/-1/-1->0->-1 [5] -1/-1/-1->0->-1 [6] -1/-1/-1->0->-1 [7] -1/-1/-1->0->-1 [8] -1/-1/-1->0->-1 [9] -1/-1/-1->0->-1 [10] -1/-1/-1->0->-1 [11] -1/-1/-1->0->-1 [12] -1/-1/-1->0->-1 [13] -1/-1/-1->0->-1 [14] -1/-1/-1->0->-1 [15] -1/-1/-1->0->-1 [16] -1/-1/-1->0->-1 [17] -1/-1/-1->0->-1 [18] -1/-1/-1->0->-1 [19] -1/-1/-1->0->-1 [20] -1/-1/-1->0->-1 [21] -1/-1/-1->0->-1 [22] -1/-1/-1->0->-1 [23] -1/-1/-1->0->-1 [24] -1/-1/-1->0->-1 [25] -1/-1/-1->0->-1 [26] -1/-1/-1->0->-1 [27] -1/-1/-1->0->-1 [28] -1/-1/-1->0->-1 [29] -1/-1/-1->0->-1 [30] -1/-1/-1->0->-1 [31] -1/-1/-1->0->-1
+ip-26-0-169-207:734079:734395 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-169-207:734082:734390 [7] NCCL INFO comm 0xaf3fb50 rank 0 nranks 1 cudaDev 7 nvmlDev 7 busId ca000 commId 0xd9379f321c367163 - Init COMPLETE
+ip-26-0-169-207:734079:734395 [4] NCCL INFO Connected all rings
+ip-26-0-169-207:734079:734395 [4] NCCL INFO Connected all trees
+ip-26-0-169-207:734079:734395 [4] NCCL INFO 32 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:734082:734420 [7] NCCL INFO Using network Libfabric
+ip-26-0-169-207:734081:734421 [6] NCCL INFO Using network Libfabric
+ip-26-0-169-207:734080:734397 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-169-207:734080:734397 [5] NCCL INFO Channel 00/32 :    0
+ip-26-0-169-207:734080:734397 [5] NCCL INFO Channel 01/32 :    0
+ip-26-0-169-207:734080:734397 [5] NCCL INFO Channel 02/32 :    0
+ip-26-0-169-207:734080:734397 [5] NCCL INFO Channel 03/32 :    0
+ip-26-0-169-207:734080:734397 [5] NCCL INFO Channel 04/32 :    0
+ip-26-0-169-207:734080:734397 [5] NCCL INFO Channel 05/32 :    0
+ip-26-0-169-207:734080:734397 [5] NCCL INFO Channel 06/32 :    0
+ip-26-0-169-207:734080:734397 [5] NCCL INFO Channel 07/32 :    0
+ip-26-0-169-207:734080:734397 [5] NCCL INFO Channel 08/32 :    0
+ip-26-0-169-207:734080:734397 [5] NCCL INFO Channel 09/32 :    0
+ip-26-0-169-207:734080:734397 [5] NCCL INFO Channel 10/32 :    0
+ip-26-0-169-207:734080:734397 [5] NCCL INFO Channel 11/32 :    0
+ip-26-0-169-207:734080:734397 [5] NCCL INFO Channel 12/32 :    0
+ip-26-0-169-207:734080:734397 [5] NCCL INFO Channel 13/32 :    0
+ip-26-0-169-207:734080:734397 [5] NCCL INFO Channel 14/32 :    0
+ip-26-0-169-207:734080:734397 [5] NCCL INFO Channel 15/32 :    0
+ip-26-0-169-207:734080:734397 [5] NCCL INFO Channel 16/32 :    0
+ip-26-0-169-207:734080:734397 [5] NCCL INFO Channel 17/32 :    0
+ip-26-0-169-207:734080:734397 [5] NCCL INFO Channel 18/32 :    0
+ip-26-0-169-207:734080:734397 [5] NCCL INFO Channel 19/32 :    0
+ip-26-0-169-207:734080:734397 [5] NCCL INFO Channel 20/32 :    0
+ip-26-0-169-207:734080:734397 [5] NCCL INFO Channel 21/32 :    0
+ip-26-0-169-207:734080:734397 [5] NCCL INFO Channel 22/32 :    0
+ip-26-0-169-207:734080:734397 [5] NCCL INFO Channel 23/32 :    0
+ip-26-0-169-207:734080:734397 [5] NCCL INFO Channel 24/32 :    0
+ip-26-0-169-207:734080:734397 [5] NCCL INFO Channel 25/32 :    0
+ip-26-0-169-207:734080:734397 [5] NCCL INFO Channel 26/32 :    0
+ip-26-0-169-207:734080:734397 [5] NCCL INFO Channel 27/32 :    0
+ip-26-0-169-207:734080:734397 [5] NCCL INFO Channel 28/32 :    0
+ip-26-0-169-207:734080:734397 [5] NCCL INFO Channel 29/32 :    0
+ip-26-0-169-207:734080:734397 [5] NCCL INFO Channel 30/32 :    0
+ip-26-0-169-207:734080:734397 [5] NCCL INFO Channel 31/32 :    0
+ip-26-0-169-207:734080:734397 [5] NCCL INFO Trees [0] -1/-1/-1->0->-1 [1] -1/-1/-1->0->-1 [2] -1/-1/-1->0->-1 [3] -1/-1/-1->0->-1 [4] -1/-1/-1->0->-1 [5] -1/-1/-1->0->-1 [6] -1/-1/-1->0->-1 [7] -1/-1/-1->0->-1 [8] -1/-1/-1->0->-1 [9] -1/-1/-1->0->-1 [10] -1/-1/-1->0->-1 [11] -1/-1/-1->0->-1 [12] -1/-1/-1->0->-1 [13] -1/-1/-1->0->-1 [14] -1/-1/-1->0->-1 [15] -1/-1/-1->0->-1 [16] -1/-1/-1->0->-1 [17] -1/-1/-1->0->-1 [18] -1/-1/-1->0->-1 [19] -1/-1/-1->0->-1 [20] -1/-1/-1->0->-1 [21] -1/-1/-1->0->-1 [22] -1/-1/-1->0->-1 [23] -1/-1/-1->0->-1 [24] -1/-1/-1->0->-1 [25] -1/-1/-1->0->-1 [26] -1/-1/-1->0->-1 [27] -1/-1/-1->0->-1 [28] -1/-1/-1->0->-1 [29] -1/-1/-1->0->-1 [30] -1/-1/-1->0->-1 [31] -1/-1/-1->0->-1
+ip-26-0-169-207:734080:734397 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-169-207:734080:734397 [5] NCCL INFO Connected all rings
+ip-26-0-169-207:734080:734397 [5] NCCL INFO Connected all trees
+ip-26-0-169-207:734080:734397 [5] NCCL INFO 32 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:734079:734395 [4] NCCL INFO comm 0xb520640 rank 0 nranks 1 cudaDev 4 nvmlDev 4 busId 97000 commId 0xe22e5eed5cd53a4e - Init COMPLETE
+ip-26-0-169-207:734080:734397 [5] NCCL INFO comm 0xc0b0b30 rank 0 nranks 1 cudaDev 5 nvmlDev 5 busId a8000 commId 0x4b10886e911024bc - Init COMPLETE
+ip-26-0-169-207:734079:734424 [4] NCCL INFO Using network Libfabric
+ip-26-0-169-207:734080:734425 [5] NCCL INFO Using network Libfabric
+ip-26-0-169-207:734079:734424 [4] NCCL INFO comm 0xb525340 rank 2 nranks 4 cudaDev 4 nvmlDev 4 busId 97000 commId 0xbce7c056c060eb65 - Init START
+ip-26-0-169-207:734077:734413 [2] NCCL INFO comm 0xaf9fb00 rank 1 nranks 4 cudaDev 2 nvmlDev 2 busId 75000 commId 0xbce7c056c060eb65 - Init START
+ip-26-0-169-207:734081:734421 [6] NCCL INFO comm 0xa4a6b40 rank 3 nranks 4 cudaDev 6 nvmlDev 6 busId b9000 commId 0xbce7c056c060eb65 - Init START
+ip-26-0-169-207:734075:734412 [0] NCCL INFO comm 0xb20d930 rank 0 nranks 4 cudaDev 0 nvmlDev 0 busId 53000 commId 0xbce7c056c060eb65 - Init START
+ip-26-0-169-207:734082:734420 [7] NCCL INFO comm 0xaf44b50 rank 3 nranks 4 cudaDev 7 nvmlDev 7 busId ca000 commId 0x5c65f063e6977ca4 - Init START
+ip-26-0-169-207:734081:734421 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/149
+ip-26-0-169-207:734076:734416 [1] NCCL INFO comm 0xb840c40 rank 0 nranks 4 cudaDev 1 nvmlDev 1 busId 64000 commId 0x5c65f063e6977ca4 - Init START
+ip-26-0-169-207:734079:734424 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/149
+ip-26-0-169-207:734080:734425 [5] NCCL INFO comm 0xc0b5b40 rank 2 nranks 4 cudaDev 5 nvmlDev 5 busId a8000 commId 0x5c65f063e6977ca4 - Init START
+ip-26-0-169-207:734077:734413 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/149
+ip-26-0-169-207:734078:734417 [3] NCCL INFO comm 0xbe9f1a0 rank 1 nranks 4 cudaDev 3 nvmlDev 3 busId 86000 commId 0x5c65f063e6977ca4 - Init START
+ip-26-0-169-207:734075:734412 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/162
+ip-26-0-169-207:734080:734425 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/149
+ip-26-0-169-207:734078:734417 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/149
+ip-26-0-169-207:734082:734420 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/149
+ip-26-0-169-207:734076:734416 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/149
+ip-26-0-169-207:734079:734424 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-169-207:734079:734424 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-169-207:734076:734416 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-169-207:734076:734416 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-169-207:734082:734420 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-169-207:734082:734420 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-169-207:734077:734413 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-169-207:734077:734413 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-169-207:734078:734417 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-169-207:734078:734417 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-169-207:734081:734421 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-169-207:734081:734421 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-169-207:734075:734412 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-169-207:734075:734412 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-169-207:734080:734425 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-169-207:734080:734425 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-169-207:734075:734412 [0] NCCL INFO Channel 00/24 :    0   1   2   3
+ip-26-0-169-207:734075:734412 [0] NCCL INFO Channel 01/24 :    0   1   2   3
+ip-26-0-169-207:734075:734412 [0] NCCL INFO Channel 02/24 :    0   1   2   3
+ip-26-0-169-207:734075:734412 [0] NCCL INFO Channel 03/24 :    0   1   2   3
+ip-26-0-169-207:734075:734412 [0] NCCL INFO Channel 04/24 :    0   1   2   3
+ip-26-0-169-207:734075:734412 [0] NCCL INFO Channel 05/24 :    0   1   2   3
+ip-26-0-169-207:734075:734412 [0] NCCL INFO Channel 06/24 :    0   1   2   3
+ip-26-0-169-207:734075:734412 [0] NCCL INFO Channel 07/24 :    0   1   2   3
+ip-26-0-169-207:734075:734412 [0] NCCL INFO Channel 08/24 :    0   1   2   3
+ip-26-0-169-207:734075:734412 [0] NCCL INFO Channel 09/24 :    0   1   2   3
+ip-26-0-169-207:734075:734412 [0] NCCL INFO Channel 10/24 :    0   1   2   3
+ip-26-0-169-207:734081:734421 [6] NCCL INFO Trees [0] -1/-1/-1->3->2 [1] -1/-1/-1->3->2 [2] -1/-1/-1->3->2 [3] -1/-1/-1->3->2 [4] -1/-1/-1->3->2 [5] -1/-1/-1->3->2 [6] -1/-1/-1->3->2 [7] -1/-1/-1->3->2 [8] -1/-1/-1->3->2 [9] -1/-1/-1->3->2 [10] -1/-1/-1->3->2 [11] -1/-1/-1->3->2 [12] -1/-1/-1->3->2 [13] -1/-1/-1->3->2 [14] -1/-1/-1->3->2 [15] -1/-1/-1->3->2 [16] -1/-1/-1->3->2 [17] -1/-1/-1->3->2 [18] -1/-1/-1->3->2 [19] -1/-1/-1->3->2 [20] -1/-1/-1->3->2 [21] -1/-1/-1->3->2 [22] -1/-1/-1->3->2 [23] -1/-1/-1->3->2
+ip-26-0-169-207:734075:734412 [0] NCCL INFO Channel 11/24 :    0   1   2   3
+ip-26-0-169-207:734081:734421 [6] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-169-207:734075:734412 [0] NCCL INFO Channel 12/24 :    0   1   2   3
+ip-26-0-169-207:734075:734412 [0] NCCL INFO Channel 13/24 :    0   1   2   3
+ip-26-0-169-207:734075:734412 [0] NCCL INFO Channel 14/24 :    0   1   2   3
+ip-26-0-169-207:734075:734412 [0] NCCL INFO Channel 15/24 :    0   1   2   3
+ip-26-0-169-207:734077:734413 [2] NCCL INFO Trees [0] 2/-1/-1->1->0 [1] 2/-1/-1->1->0 [2] 2/-1/-1->1->0 [3] 2/-1/-1->1->0 [4] 2/-1/-1->1->0 [5] 2/-1/-1->1->0 [6] 2/-1/-1->1->0 [7] 2/-1/-1->1->0 [8] 2/-1/-1->1->0 [9] 2/-1/-1->1->0 [10] 2/-1/-1->1->0 [11] 2/-1/-1->1->0 [12] 2/-1/-1->1->0 [13] 2/-1/-1->1->0 [14] 2/-1/-1->1->0 [15] 2/-1/-1->1->0 [16] 2/-1/-1->1->0 [17] 2/-1/-1->1->0 [18] 2/-1/-1->1->0 [19] 2/-1/-1->1->0 [20] 2/-1/-1->1->0 [21] 2/-1/-1->1->0 [22] 2/-1/-1->1->0 [23] 2/-1/-1->1->0
+ip-26-0-169-207:734075:734412 [0] NCCL INFO Channel 16/24 :    0   1   2   3
+ip-26-0-169-207:734075:734412 [0] NCCL INFO Channel 17/24 :    0   1   2   3
+ip-26-0-169-207:734077:734413 [2] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-169-207:734075:734412 [0] NCCL INFO Channel 18/24 :    0   1   2   3
+ip-26-0-169-207:734075:734412 [0] NCCL INFO Channel 19/24 :    0   1   2   3
+ip-26-0-169-207:734075:734412 [0] NCCL INFO Channel 20/24 :    0   1   2   3
+ip-26-0-169-207:734075:734412 [0] NCCL INFO Channel 21/24 :    0   1   2   3
+ip-26-0-169-207:734075:734412 [0] NCCL INFO Channel 22/24 :    0   1   2   3
+ip-26-0-169-207:734075:734412 [0] NCCL INFO Channel 23/24 :    0   1   2   3
+ip-26-0-169-207:734079:734424 [4] NCCL INFO Trees [0] 3/-1/-1->2->1 [1] 3/-1/-1->2->1 [2] 3/-1/-1->2->1 [3] 3/-1/-1->2->1 [4] 3/-1/-1->2->1 [5] 3/-1/-1->2->1 [6] 3/-1/-1->2->1 [7] 3/-1/-1->2->1 [8] 3/-1/-1->2->1 [9] 3/-1/-1->2->1 [10] 3/-1/-1->2->1 [11] 3/-1/-1->2->1 [12] 3/-1/-1->2->1 [13] 3/-1/-1->2->1 [14] 3/-1/-1->2->1 [15] 3/-1/-1->2->1 [16] 3/-1/-1->2->1 [17] 3/-1/-1->2->1 [18] 3/-1/-1->2->1 [19] 3/-1/-1->2->1 [20] 3/-1/-1->2->1 [21] 3/-1/-1->2->1 [22] 3/-1/-1->2->1 [23] 3/-1/-1->2->1
+ip-26-0-169-207:734079:734424 [4] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-169-207:734075:734412 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] 1/-1/-1->0->-1 [7] 1/-1/-1->0->-1 [8] 1/-1/-1->0->-1 [9] 1/-1/-1->0->-1 [10] 1/-1/-1->0->-1 [11] 1/-1/-1->0->-1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] 1/-1/-1->0->-1 [19] 1/-1/-1->0->-1 [20] 1/-1/-1->0->-1 [21] 1/-1/-1->0->-1 [22] 1/-1/-1->0->-1 [23] 1/-1/-1->0->-1
+ip-26-0-169-207:734075:734412 [0] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-169-207:734076:734416 [1] NCCL INFO Channel 00/24 :    0   1   2   3
+ip-26-0-169-207:734080:734425 [5] NCCL INFO Trees [0] 3/-1/-1->2->1 [1] 3/-1/-1->2->1 [2] 3/-1/-1->2->1 [3] 3/-1/-1->2->1 [4] 3/-1/-1->2->1 [5] 3/-1/-1->2->1 [6] 3/-1/-1->2->1 [7] 3/-1/-1->2->1 [8] 3/-1/-1->2->1 [9] 3/-1/-1->2->1 [10] 3/-1/-1->2->1 [11] 3/-1/-1->2->1 [12] 3/-1/-1->2->1 [13] 3/-1/-1->2->1 [14] 3/-1/-1->2->1 [15] 3/-1/-1->2->1 [16] 3/-1/-1->2->1 [17] 3/-1/-1->2->1 [18] 3/-1/-1->2->1 [19] 3/-1/-1->2->1 [20] 3/-1/-1->2->1 [21] 3/-1/-1->2->1 [22] 3/-1/-1->2->1 [23] 3/-1/-1->2->1
+ip-26-0-169-207:734076:734416 [1] NCCL INFO Channel 01/24 :    0   1   2   3
+ip-26-0-169-207:734082:734420 [7] NCCL INFO Trees [0] -1/-1/-1->3->2 [1] -1/-1/-1->3->2 [2] -1/-1/-1->3->2 [3] -1/-1/-1->3->2 [4] -1/-1/-1->3->2 [5] -1/-1/-1->3->2 [6] -1/-1/-1->3->2 [7] -1/-1/-1->3->2 [8] -1/-1/-1->3->2 [9] -1/-1/-1->3->2 [10] -1/-1/-1->3->2 [11] -1/-1/-1->3->2 [12] -1/-1/-1->3->2 [13] -1/-1/-1->3->2 [14] -1/-1/-1->3->2 [15] -1/-1/-1->3->2 [16] -1/-1/-1->3->2 [17] -1/-1/-1->3->2 [18] -1/-1/-1->3->2 [19] -1/-1/-1->3->2 [20] -1/-1/-1->3->2 [21] -1/-1/-1->3->2 [22] -1/-1/-1->3->2 [23] -1/-1/-1->3->2
+ip-26-0-169-207:734076:734416 [1] NCCL INFO Channel 02/24 :    0   1   2   3
+ip-26-0-169-207:734080:734425 [5] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-169-207:734082:734420 [7] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-169-207:734076:734416 [1] NCCL INFO Channel 03/24 :    0   1   2   3
+ip-26-0-169-207:734076:734416 [1] NCCL INFO Channel 04/24 :    0   1   2   3
+ip-26-0-169-207:734076:734416 [1] NCCL INFO Channel 05/24 :    0   1   2   3
+ip-26-0-169-207:734078:734417 [3] NCCL INFO Trees [0] 2/-1/-1->1->0 [1] 2/-1/-1->1->0 [2] 2/-1/-1->1->0 [3] 2/-1/-1->1->0 [4] 2/-1/-1->1->0 [5] 2/-1/-1->1->0 [6] 2/-1/-1->1->0 [7] 2/-1/-1->1->0 [8] 2/-1/-1->1->0 [9] 2/-1/-1->1->0 [10] 2/-1/-1->1->0 [11] 2/-1/-1->1->0 [12] 2/-1/-1->1->0 [13] 2/-1/-1->1->0 [14] 2/-1/-1->1->0 [15] 2/-1/-1->1->0 [16] 2/-1/-1->1->0 [17] 2/-1/-1->1->0 [18] 2/-1/-1->1->0 [19] 2/-1/-1->1->0 [20] 2/-1/-1->1->0 [21] 2/-1/-1->1->0 [22] 2/-1/-1->1->0 [23] 2/-1/-1->1->0
+ip-26-0-169-207:734076:734416 [1] NCCL INFO Channel 06/24 :    0   1   2   3
+ip-26-0-169-207:734076:734416 [1] NCCL INFO Channel 07/24 :    0   1   2   3
+ip-26-0-169-207:734078:734417 [3] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-169-207:734076:734416 [1] NCCL INFO Channel 08/24 :    0   1   2   3
+ip-26-0-169-207:734076:734416 [1] NCCL INFO Channel 09/24 :    0   1   2   3
+ip-26-0-169-207:734076:734416 [1] NCCL INFO Channel 10/24 :    0   1   2   3
+ip-26-0-169-207:734076:734416 [1] NCCL INFO Channel 11/24 :    0   1   2   3
+ip-26-0-169-207:734076:734416 [1] NCCL INFO Channel 12/24 :    0   1   2   3
+ip-26-0-169-207:734076:734416 [1] NCCL INFO Channel 13/24 :    0   1   2   3
+ip-26-0-169-207:734076:734416 [1] NCCL INFO Channel 14/24 :    0   1   2   3
+ip-26-0-169-207:734076:734416 [1] NCCL INFO Channel 15/24 :    0   1   2   3
+ip-26-0-169-207:734076:734416 [1] NCCL INFO Channel 16/24 :    0   1   2   3
+ip-26-0-169-207:734076:734416 [1] NCCL INFO Channel 17/24 :    0   1   2   3
+ip-26-0-169-207:734076:734416 [1] NCCL INFO Channel 18/24 :    0   1   2   3
+ip-26-0-169-207:734076:734416 [1] NCCL INFO Channel 19/24 :    0   1   2   3
+ip-26-0-169-207:734076:734416 [1] NCCL INFO Channel 20/24 :    0   1   2   3
+ip-26-0-169-207:734076:734416 [1] NCCL INFO Channel 21/24 :    0   1   2   3
+ip-26-0-169-207:734076:734416 [1] NCCL INFO Channel 22/24 :    0   1   2   3
+ip-26-0-169-207:734076:734416 [1] NCCL INFO Channel 23/24 :    0   1   2   3
+ip-26-0-169-207:734076:734416 [1] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] 1/-1/-1->0->-1 [7] 1/-1/-1->0->-1 [8] 1/-1/-1->0->-1 [9] 1/-1/-1->0->-1 [10] 1/-1/-1->0->-1 [11] 1/-1/-1->0->-1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] 1/-1/-1->0->-1 [19] 1/-1/-1->0->-1 [20] 1/-1/-1->0->-1 [21] 1/-1/-1->0->-1 [22] 1/-1/-1->0->-1 [23] 1/-1/-1->0->-1
+ip-26-0-169-207:734076:734416 [1] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-169-207:734081:734421 [6] NCCL INFO Channel 00/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734079:734424 [4] NCCL INFO Channel 00/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-169-207:734081:734421 [6] NCCL INFO Channel 01/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734079:734424 [4] NCCL INFO Channel 01/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-169-207:734081:734421 [6] NCCL INFO Channel 02/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734079:734424 [4] NCCL INFO Channel 02/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-169-207:734081:734421 [6] NCCL INFO Channel 03/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734079:734424 [4] NCCL INFO Channel 03/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-169-207:734081:734421 [6] NCCL INFO Channel 04/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734079:734424 [4] NCCL INFO Channel 04/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-169-207:734081:734421 [6] NCCL INFO Channel 05/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734075:734412 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-169-207:734079:734424 [4] NCCL INFO Channel 05/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-169-207:734081:734421 [6] NCCL INFO Channel 06/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734075:734412 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-169-207:734081:734421 [6] NCCL INFO Channel 07/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734079:734424 [4] NCCL INFO Channel 06/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-169-207:734076:734416 [1] NCCL INFO Channel 00/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-169-207:734079:734424 [4] NCCL INFO Channel 07/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-169-207:734076:734416 [1] NCCL INFO Channel 01/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-169-207:734075:734412 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-169-207:734079:734424 [4] NCCL INFO Channel 08/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-169-207:734076:734416 [1] NCCL INFO Channel 02/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-169-207:734075:734412 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-169-207:734082:734420 [7] NCCL INFO Channel 00/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-169-207:734077:734413 [2] NCCL INFO Channel 00/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-169-207:734076:734416 [1] NCCL INFO Channel 03/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-169-207:734075:734412 [0] NCCL INFO Channel 04/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-169-207:734079:734424 [4] NCCL INFO Channel 09/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-169-207:734082:734420 [7] NCCL INFO Channel 01/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-169-207:734075:734412 [0] NCCL INFO Channel 05/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-169-207:734076:734416 [1] NCCL INFO Channel 04/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-169-207:734077:734413 [2] NCCL INFO Channel 01/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-169-207:734079:734424 [4] NCCL INFO Channel 10/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-169-207:734082:734420 [7] NCCL INFO Channel 02/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-169-207:734081:734421 [6] NCCL INFO Channel 08/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734076:734416 [1] NCCL INFO Channel 05/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-169-207:734075:734412 [0] NCCL INFO Channel 06/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-169-207:734077:734413 [2] NCCL INFO Channel 02/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-169-207:734081:734421 [6] NCCL INFO Channel 09/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734079:734424 [4] NCCL INFO Channel 11/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-169-207:734082:734420 [7] NCCL INFO Channel 03/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-169-207:734075:734412 [0] NCCL INFO Channel 07/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-169-207:734077:734413 [2] NCCL INFO Channel 03/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-169-207:734081:734421 [6] NCCL INFO Channel 10/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734079:734424 [4] NCCL INFO Channel 12/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-169-207:734082:734420 [7] NCCL INFO Channel 04/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-169-207:734075:734412 [0] NCCL INFO Channel 08/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-169-207:734077:734413 [2] NCCL INFO Channel 04/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-169-207:734081:734421 [6] NCCL INFO Channel 11/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734079:734424 [4] NCCL INFO Channel 13/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-169-207:734082:734420 [7] NCCL INFO Channel 05/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-169-207:734075:734412 [0] NCCL INFO Channel 09/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-169-207:734076:734416 [1] NCCL INFO Channel 06/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-169-207:734075:734412 [0] NCCL INFO Channel 10/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-169-207:734076:734416 [1] NCCL INFO Channel 07/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-169-207:734077:734413 [2] NCCL INFO Channel 05/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-169-207:734075:734412 [0] NCCL INFO Channel 11/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-169-207:734076:734416 [1] NCCL INFO Channel 08/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-169-207:734077:734413 [2] NCCL INFO Channel 06/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-169-207:734075:734412 [0] NCCL INFO Channel 12/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-169-207:734076:734416 [1] NCCL INFO Channel 09/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-169-207:734077:734413 [2] NCCL INFO Channel 07/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-169-207:734075:734412 [0] NCCL INFO Channel 13/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-169-207:734076:734416 [1] NCCL INFO Channel 10/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-169-207:734080:734425 [5] NCCL INFO Channel 00/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-169-207:734077:734413 [2] NCCL INFO Channel 08/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-169-207:734079:734424 [4] NCCL INFO Channel 14/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-169-207:734081:734421 [6] NCCL INFO Channel 12/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734082:734420 [7] NCCL INFO Channel 06/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-169-207:734075:734412 [0] NCCL INFO Channel 14/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-169-207:734076:734416 [1] NCCL INFO Channel 11/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-169-207:734080:734425 [5] NCCL INFO Channel 01/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-169-207:734079:734424 [4] NCCL INFO Channel 15/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-169-207:734077:734413 [2] NCCL INFO Channel 09/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-169-207:734075:734412 [0] NCCL INFO Channel 15/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-169-207:734082:734420 [7] NCCL INFO Channel 07/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-169-207:734076:734416 [1] NCCL INFO Channel 12/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-169-207:734081:734421 [6] NCCL INFO Channel 13/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734075:734412 [0] NCCL INFO Channel 16/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-169-207:734080:734425 [5] NCCL INFO Channel 02/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-169-207:734081:734421 [6] NCCL INFO Channel 14/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734075:734412 [0] NCCL INFO Channel 17/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-169-207:734080:734425 [5] NCCL INFO Channel 03/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-169-207:734081:734421 [6] NCCL INFO Channel 15/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734075:734412 [0] NCCL INFO Channel 18/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-169-207:734080:734425 [5] NCCL INFO Channel 04/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-169-207:734081:734421 [6] NCCL INFO Channel 16/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734075:734412 [0] NCCL INFO Channel 19/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-169-207:734080:734425 [5] NCCL INFO Channel 05/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-169-207:734081:734421 [6] NCCL INFO Channel 17/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734079:734424 [4] NCCL INFO Channel 16/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-169-207:734078:734417 [3] NCCL INFO Channel 00/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-169-207:734077:734413 [2] NCCL INFO Channel 10/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-169-207:734082:734420 [7] NCCL INFO Channel 08/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-169-207:734076:734416 [1] NCCL INFO Channel 13/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-169-207:734075:734412 [0] NCCL INFO Channel 20/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-169-207:734080:734425 [5] NCCL INFO Channel 06/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-169-207:734081:734421 [6] NCCL INFO Channel 18/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734078:734417 [3] NCCL INFO Channel 01/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-169-207:734079:734424 [4] NCCL INFO Channel 17/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-169-207:734077:734413 [2] NCCL INFO Channel 11/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-169-207:734082:734420 [7] NCCL INFO Channel 09/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-169-207:734076:734416 [1] NCCL INFO Channel 14/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-169-207:734075:734412 [0] NCCL INFO Channel 21/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-169-207:734082:734420 [7] NCCL INFO Channel 10/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-169-207:734075:734412 [0] NCCL INFO Channel 22/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-169-207:734080:734425 [5] NCCL INFO Channel 07/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-169-207:734076:734416 [1] NCCL INFO Channel 15/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-169-207:734081:734421 [6] NCCL INFO Channel 19/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734082:734420 [7] NCCL INFO Channel 11/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-169-207:734075:734412 [0] NCCL INFO Channel 23/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-169-207:734080:734425 [5] NCCL INFO Channel 08/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-169-207:734076:734416 [1] NCCL INFO Channel 16/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-169-207:734081:734421 [6] NCCL INFO Channel 20/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734082:734420 [7] NCCL INFO Channel 12/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-169-207:734080:734425 [5] NCCL INFO Channel 09/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-169-207:734076:734416 [1] NCCL INFO Channel 17/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-169-207:734081:734421 [6] NCCL INFO Channel 21/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734082:734420 [7] NCCL INFO Channel 13/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-169-207:734080:734425 [5] NCCL INFO Channel 10/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-169-207:734079:734424 [4] NCCL INFO Channel 18/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-169-207:734076:734416 [1] NCCL INFO Channel 18/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-169-207:734081:734421 [6] NCCL INFO Channel 22/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734082:734420 [7] NCCL INFO Channel 14/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-169-207:734080:734425 [5] NCCL INFO Channel 11/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-169-207:734079:734424 [4] NCCL INFO Channel 19/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-169-207:734076:734416 [1] NCCL INFO Channel 19/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-169-207:734081:734421 [6] NCCL INFO Channel 23/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734082:734420 [7] NCCL INFO Channel 15/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-169-207:734080:734425 [5] NCCL INFO Channel 12/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-169-207:734079:734424 [4] NCCL INFO Channel 20/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-169-207:734078:734417 [3] NCCL INFO Channel 02/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-169-207:734076:734416 [1] NCCL INFO Channel 20/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-169-207:734080:734425 [5] NCCL INFO Channel 13/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-169-207:734082:734420 [7] NCCL INFO Channel 16/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-169-207:734077:734413 [2] NCCL INFO Channel 12/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-169-207:734079:734424 [4] NCCL INFO Channel 21/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-169-207:734078:734417 [3] NCCL INFO Channel 03/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-169-207:734076:734416 [1] NCCL INFO Channel 21/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-169-207:734080:734425 [5] NCCL INFO Channel 14/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-169-207:734082:734420 [7] NCCL INFO Channel 17/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-169-207:734077:734413 [2] NCCL INFO Channel 13/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-169-207:734079:734424 [4] NCCL INFO Channel 22/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-169-207:734078:734417 [3] NCCL INFO Channel 04/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-169-207:734076:734416 [1] NCCL INFO Channel 22/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-169-207:734080:734425 [5] NCCL INFO Channel 15/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-169-207:734077:734413 [2] NCCL INFO Channel 14/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-169-207:734082:734420 [7] NCCL INFO Channel 18/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-169-207:734079:734424 [4] NCCL INFO Channel 23/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-169-207:734078:734417 [3] NCCL INFO Channel 05/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-169-207:734080:734425 [5] NCCL INFO Channel 16/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-169-207:734077:734413 [2] NCCL INFO Channel 15/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-169-207:734076:734416 [1] NCCL INFO Channel 23/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-169-207:734082:734420 [7] NCCL INFO Channel 19/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-169-207:734078:734417 [3] NCCL INFO Channel 06/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-169-207:734080:734425 [5] NCCL INFO Channel 17/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-169-207:734077:734413 [2] NCCL INFO Channel 16/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-169-207:734082:734420 [7] NCCL INFO Channel 20/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-169-207:734078:734417 [3] NCCL INFO Channel 07/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-169-207:734080:734425 [5] NCCL INFO Channel 18/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-169-207:734077:734413 [2] NCCL INFO Channel 17/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-169-207:734082:734420 [7] NCCL INFO Channel 21/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-169-207:734078:734417 [3] NCCL INFO Channel 08/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-169-207:734082:734420 [7] NCCL INFO Channel 22/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-169-207:734077:734413 [2] NCCL INFO Channel 18/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-169-207:734080:734425 [5] NCCL INFO Channel 19/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-169-207:734078:734417 [3] NCCL INFO Channel 09/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-169-207:734082:734420 [7] NCCL INFO Channel 23/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-169-207:734077:734413 [2] NCCL INFO Channel 19/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-169-207:734080:734425 [5] NCCL INFO Channel 20/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-169-207:734078:734417 [3] NCCL INFO Channel 10/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-169-207:734077:734413 [2] NCCL INFO Channel 20/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-169-207:734080:734425 [5] NCCL INFO Channel 21/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-169-207:734078:734417 [3] NCCL INFO Channel 11/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-169-207:734080:734425 [5] NCCL INFO Channel 22/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-169-207:734077:734413 [2] NCCL INFO Channel 21/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-169-207:734078:734417 [3] NCCL INFO Channel 12/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-169-207:734080:734425 [5] NCCL INFO Channel 23/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-169-207:734077:734413 [2] NCCL INFO Channel 22/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-169-207:734078:734417 [3] NCCL INFO Channel 13/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-169-207:734077:734413 [2] NCCL INFO Channel 23/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-169-207:734078:734417 [3] NCCL INFO Channel 14/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-169-207:734078:734417 [3] NCCL INFO Channel 15/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-169-207:734078:734417 [3] NCCL INFO Channel 16/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-169-207:734078:734417 [3] NCCL INFO Channel 17/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-169-207:734078:734417 [3] NCCL INFO Channel 18/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-169-207:734078:734417 [3] NCCL INFO Channel 19/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-169-207:734078:734417 [3] NCCL INFO Channel 20/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-169-207:734078:734417 [3] NCCL INFO Channel 21/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-169-207:734078:734417 [3] NCCL INFO Channel 22/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-169-207:734078:734417 [3] NCCL INFO Channel 23/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-169-207:734081:734421 [6] NCCL INFO Connected all rings
+ip-26-0-169-207:734081:734421 [6] NCCL INFO Channel 00/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-169-207:734081:734421 [6] NCCL INFO Channel 01/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-169-207:734081:734421 [6] NCCL INFO Channel 02/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-169-207:734081:734421 [6] NCCL INFO Channel 03/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-169-207:734081:734421 [6] NCCL INFO Channel 04/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-169-207:734081:734421 [6] NCCL INFO Channel 05/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-169-207:734081:734421 [6] NCCL INFO Channel 06/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-169-207:734081:734421 [6] NCCL INFO Channel 07/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-169-207:734081:734421 [6] NCCL INFO Channel 08/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-169-207:734081:734421 [6] NCCL INFO Channel 09/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-169-207:734081:734421 [6] NCCL INFO Channel 10/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-169-207:734081:734421 [6] NCCL INFO Channel 11/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-169-207:734077:734413 [2] NCCL INFO Connected all rings
+ip-26-0-169-207:734075:734412 [0] NCCL INFO Connected all rings
+ip-26-0-169-207:734081:734421 [6] NCCL INFO Channel 12/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-169-207:734081:734421 [6] NCCL INFO Channel 13/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-169-207:734082:734420 [7] NCCL INFO Connected all rings
+ip-26-0-169-207:734082:734420 [7] NCCL INFO Channel 00/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-169-207:734079:734424 [4] NCCL INFO Connected all rings
+ip-26-0-169-207:734082:734420 [7] NCCL INFO Channel 01/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-169-207:734082:734420 [7] NCCL INFO Channel 02/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-169-207:734082:734420 [7] NCCL INFO Channel 03/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-169-207:734081:734421 [6] NCCL INFO Channel 14/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-169-207:734081:734421 [6] NCCL INFO Channel 15/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-169-207:734082:734420 [7] NCCL INFO Channel 04/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-169-207:734082:734420 [7] NCCL INFO Channel 05/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-169-207:734078:734417 [3] NCCL INFO Connected all rings
+ip-26-0-169-207:734081:734421 [6] NCCL INFO Channel 16/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-169-207:734081:734421 [6] NCCL INFO Channel 17/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-169-207:734082:734420 [7] NCCL INFO Channel 06/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-169-207:734081:734421 [6] NCCL INFO Channel 18/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-169-207:734082:734420 [7] NCCL INFO Channel 07/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-169-207:734076:734416 [1] NCCL INFO Connected all rings
+ip-26-0-169-207:734081:734421 [6] NCCL INFO Channel 19/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-169-207:734081:734421 [6] NCCL INFO Channel 20/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-169-207:734082:734420 [7] NCCL INFO Channel 08/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-169-207:734082:734420 [7] NCCL INFO Channel 09/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-169-207:734082:734420 [7] NCCL INFO Channel 10/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-169-207:734082:734420 [7] NCCL INFO Channel 11/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-169-207:734081:734421 [6] NCCL INFO Channel 21/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-169-207:734082:734420 [7] NCCL INFO Channel 12/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-169-207:734081:734421 [6] NCCL INFO Channel 22/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-169-207:734082:734420 [7] NCCL INFO Channel 13/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-169-207:734080:734425 [5] NCCL INFO Connected all rings
+ip-26-0-169-207:734082:734420 [7] NCCL INFO Channel 14/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-169-207:734081:734421 [6] NCCL INFO Channel 23/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-169-207:734082:734420 [7] NCCL INFO Channel 15/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-169-207:734082:734420 [7] NCCL INFO Channel 16/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-169-207:734082:734420 [7] NCCL INFO Channel 17/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-169-207:734082:734420 [7] NCCL INFO Channel 18/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-169-207:734082:734420 [7] NCCL INFO Channel 19/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-169-207:734077:734413 [2] NCCL INFO Channel 00/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734082:734420 [7] NCCL INFO Channel 20/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-169-207:734077:734413 [2] NCCL INFO Channel 01/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734082:734420 [7] NCCL INFO Channel 21/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-169-207:734077:734413 [2] NCCL INFO Channel 02/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734082:734420 [7] NCCL INFO Channel 22/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-169-207:734077:734413 [2] NCCL INFO Channel 03/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734082:734420 [7] NCCL INFO Channel 23/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-169-207:734077:734413 [2] NCCL INFO Channel 04/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734077:734413 [2] NCCL INFO Channel 05/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734077:734413 [2] NCCL INFO Channel 06/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734077:734413 [2] NCCL INFO Channel 07/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734077:734413 [2] NCCL INFO Channel 08/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734079:734424 [4] NCCL INFO Channel 00/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-169-207:734078:734417 [3] NCCL INFO Channel 00/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-169-207:734079:734424 [4] NCCL INFO Channel 01/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-169-207:734077:734413 [2] NCCL INFO Channel 09/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734079:734424 [4] NCCL INFO Channel 02/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-169-207:734078:734417 [3] NCCL INFO Channel 01/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-169-207:734077:734413 [2] NCCL INFO Channel 10/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734078:734417 [3] NCCL INFO Channel 02/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-169-207:734079:734424 [4] NCCL INFO Channel 03/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-169-207:734077:734413 [2] NCCL INFO Channel 11/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734079:734424 [4] NCCL INFO Channel 04/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-169-207:734078:734417 [3] NCCL INFO Channel 03/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-169-207:734077:734413 [2] NCCL INFO Channel 12/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734079:734424 [4] NCCL INFO Channel 05/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-169-207:734078:734417 [3] NCCL INFO Channel 04/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-169-207:734077:734413 [2] NCCL INFO Channel 13/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734079:734424 [4] NCCL INFO Channel 06/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-169-207:734078:734417 [3] NCCL INFO Channel 05/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-169-207:734077:734413 [2] NCCL INFO Channel 14/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734079:734424 [4] NCCL INFO Channel 07/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-169-207:734078:734417 [3] NCCL INFO Channel 06/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-169-207:734077:734413 [2] NCCL INFO Channel 15/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734079:734424 [4] NCCL INFO Channel 08/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-169-207:734078:734417 [3] NCCL INFO Channel 07/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-169-207:734077:734413 [2] NCCL INFO Channel 16/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734079:734424 [4] NCCL INFO Channel 09/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-169-207:734078:734417 [3] NCCL INFO Channel 08/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-169-207:734077:734413 [2] NCCL INFO Channel 17/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734080:734425 [5] NCCL INFO Channel 00/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-169-207:734079:734424 [4] NCCL INFO Channel 10/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-169-207:734078:734417 [3] NCCL INFO Channel 09/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-169-207:734077:734413 [2] NCCL INFO Channel 18/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734080:734425 [5] NCCL INFO Channel 01/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-169-207:734079:734424 [4] NCCL INFO Channel 11/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-169-207:734078:734417 [3] NCCL INFO Channel 10/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-169-207:734077:734413 [2] NCCL INFO Channel 19/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734080:734425 [5] NCCL INFO Channel 02/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-169-207:734079:734424 [4] NCCL INFO Channel 12/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-169-207:734078:734417 [3] NCCL INFO Channel 11/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-169-207:734077:734413 [2] NCCL INFO Channel 20/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734080:734425 [5] NCCL INFO Channel 03/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-169-207:734079:734424 [4] NCCL INFO Channel 13/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-169-207:734078:734417 [3] NCCL INFO Channel 12/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-169-207:734077:734413 [2] NCCL INFO Channel 21/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734080:734425 [5] NCCL INFO Channel 04/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-169-207:734079:734424 [4] NCCL INFO Channel 14/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-169-207:734078:734417 [3] NCCL INFO Channel 13/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-169-207:734077:734413 [2] NCCL INFO Channel 22/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734080:734425 [5] NCCL INFO Channel 05/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-169-207:734079:734424 [4] NCCL INFO Channel 15/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-169-207:734078:734417 [3] NCCL INFO Channel 14/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-169-207:734077:734413 [2] NCCL INFO Channel 23/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-169-207:734080:734425 [5] NCCL INFO Channel 06/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-169-207:734079:734424 [4] NCCL INFO Channel 16/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-169-207:734078:734417 [3] NCCL INFO Channel 15/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-169-207:734080:734425 [5] NCCL INFO Channel 07/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-169-207:734078:734417 [3] NCCL INFO Channel 16/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-169-207:734079:734424 [4] NCCL INFO Channel 17/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-169-207:734080:734425 [5] NCCL INFO Channel 08/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-169-207:734079:734424 [4] NCCL INFO Channel 18/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-169-207:734078:734417 [3] NCCL INFO Channel 17/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-169-207:734080:734425 [5] NCCL INFO Channel 09/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-169-207:734079:734424 [4] NCCL INFO Channel 19/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-169-207:734078:734417 [3] NCCL INFO Channel 18/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-169-207:734080:734425 [5] NCCL INFO Channel 10/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-169-207:734078:734417 [3] NCCL INFO Channel 19/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-169-207:734079:734424 [4] NCCL INFO Channel 20/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-169-207:734080:734425 [5] NCCL INFO Channel 11/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-169-207:734078:734417 [3] NCCL INFO Channel 20/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-169-207:734079:734424 [4] NCCL INFO Channel 21/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-169-207:734080:734425 [5] NCCL INFO Channel 12/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-169-207:734078:734417 [3] NCCL INFO Channel 21/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-169-207:734079:734424 [4] NCCL INFO Channel 22/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-169-207:734080:734425 [5] NCCL INFO Channel 13/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-169-207:734078:734417 [3] NCCL INFO Channel 22/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-169-207:734078:734417 [3] NCCL INFO Channel 23/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-169-207:734080:734425 [5] NCCL INFO Channel 14/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-169-207:734079:734424 [4] NCCL INFO Channel 23/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-169-207:734080:734425 [5] NCCL INFO Channel 15/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-169-207:734080:734425 [5] NCCL INFO Channel 16/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-169-207:734080:734425 [5] NCCL INFO Channel 17/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-169-207:734080:734425 [5] NCCL INFO Channel 18/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-169-207:734080:734425 [5] NCCL INFO Channel 19/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-169-207:734080:734425 [5] NCCL INFO Channel 20/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-169-207:734080:734425 [5] NCCL INFO Channel 21/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-169-207:734080:734425 [5] NCCL INFO Channel 22/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-169-207:734080:734425 [5] NCCL INFO Channel 23/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-169-207:734075:734412 [0] NCCL INFO Connected all trees
+ip-26-0-169-207:734075:734412 [0] NCCL INFO NVLS comm 0xb20d930 headRank 0 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-169-207:734081:734421 [6] NCCL INFO Connected all trees
+ip-26-0-169-207:734081:734421 [6] NCCL INFO NVLS comm 0xa4a6b40 headRank 3 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-169-207:734076:734416 [1] NCCL INFO Connected all trees
+ip-26-0-169-207:734076:734416 [1] NCCL INFO NVLS comm 0xb840c40 headRank 0 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-169-207:734079:734424 [4] NCCL INFO Connected all trees
+ip-26-0-169-207:734079:734424 [4] NCCL INFO NVLS comm 0xb525340 headRank 2 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-169-207:734082:734420 [7] NCCL INFO Connected all trees
+ip-26-0-169-207:734082:734420 [7] NCCL INFO NVLS comm 0xaf44b50 headRank 3 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-169-207:734077:734413 [2] NCCL INFO Connected all trees
+ip-26-0-169-207:734077:734413 [2] NCCL INFO NVLS comm 0xaf9fb00 headRank 1 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-169-207:734078:734417 [3] NCCL INFO Connected all trees
+ip-26-0-169-207:734078:734417 [3] NCCL INFO NVLS comm 0xbe9f1a0 headRank 1 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-169-207:734075:734412 [0] NCCL INFO Connected NVLS tree
+ip-26-0-169-207:734075:734412 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-169-207:734075:734412 [0] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-169-207:734075:734412 [0] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:734079:734424 [4] NCCL INFO Connected NVLS tree
+ip-26-0-169-207:734077:734413 [2] NCCL INFO Connected NVLS tree
+ip-26-0-169-207:734077:734413 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-169-207:734079:734424 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-169-207:734077:734413 [2] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-169-207:734077:734413 [2] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:734079:734424 [4] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-169-207:734079:734424 [4] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:734081:734421 [6] NCCL INFO Connected NVLS tree
+ip-26-0-169-207:734081:734421 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-169-207:734081:734421 [6] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-169-207:734081:734421 [6] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:734080:734425 [5] NCCL INFO Connected all trees
+ip-26-0-169-207:734080:734425 [5] NCCL INFO NVLS comm 0xc0b5b40 headRank 2 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-169-207:734077:734413 [2] NCCL INFO comm 0xaf9fb00 rank 1 nranks 4 cudaDev 2 nvmlDev 2 busId 75000 commId 0xbce7c056c060eb65 - Init COMPLETE
+ip-26-0-169-207:734075:734412 [0] NCCL INFO comm 0xb20d930 rank 0 nranks 4 cudaDev 0 nvmlDev 0 busId 53000 commId 0xbce7c056c060eb65 - Init COMPLETE
+ip-26-0-169-207:734081:734421 [6] NCCL INFO comm 0xa4a6b40 rank 3 nranks 4 cudaDev 6 nvmlDev 6 busId b9000 commId 0xbce7c056c060eb65 - Init COMPLETE
+ip-26-0-169-207:734079:734424 [4] NCCL INFO comm 0xb525340 rank 2 nranks 4 cudaDev 4 nvmlDev 4 busId 97000 commId 0xbce7c056c060eb65 - Init COMPLETE
+ip-26-0-169-207:734082:734420 [7] NCCL INFO Connected NVLS tree
+ip-26-0-169-207:734082:734420 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-169-207:734082:734420 [7] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-169-207:734082:734420 [7] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:734078:734417 [3] NCCL INFO Connected NVLS tree
+ip-26-0-169-207:734078:734417 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-169-207:734078:734417 [3] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-169-207:734078:734417 [3] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:734076:734416 [1] NCCL INFO Connected NVLS tree
+ip-26-0-169-207:734076:734416 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-169-207:734080:734425 [5] NCCL INFO Connected NVLS tree
+ip-26-0-169-207:734080:734425 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-169-207:734080:734425 [5] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-169-207:734080:734425 [5] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:734076:734416 [1] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-169-207:734076:734416 [1] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-169-207:734076:734416 [1] NCCL INFO comm 0xb840c40 rank 0 nranks 4 cudaDev 1 nvmlDev 1 busId 64000 commId 0x5c65f063e6977ca4 - Init COMPLETE
+ip-26-0-169-207:734082:734420 [7] NCCL INFO comm 0xaf44b50 rank 3 nranks 4 cudaDev 7 nvmlDev 7 busId ca000 commId 0x5c65f063e6977ca4 - Init COMPLETE
+ip-26-0-169-207:734078:734417 [3] NCCL INFO comm 0xbe9f1a0 rank 1 nranks 4 cudaDev 3 nvmlDev 3 busId 86000 commId 0x5c65f063e6977ca4 - Init COMPLETE
+ip-26-0-169-207:734080:734425 [5] NCCL INFO comm 0xc0b5b40 rank 2 nranks 4 cudaDev 5 nvmlDev 5 busId a8000 commId 0x5c65f063e6977ca4 - Init COMPLETE
+12/13/2024 14:42:37 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]: [Optimizer Building] Using LearningRateForSP as learning rate
+12/13/2024 14:42:37 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]: [ZeRO sharding] Size of optimizer params per rank:
+12/13/2024 14:42:37 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]: [ZeRO sharding] DP Rank 0 has 130M out of 520M (25.00%) params' optimizer states
+12/13/2024 14:42:37 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]: [ZeRO sharding] DP Rank 1 has 130M out of 520M (25.00%) params' optimizer states
+12/13/2024 14:42:37 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]: [ZeRO sharding] DP Rank 2 has 130M out of 520M (25.00%) params' optimizer states
+12/13/2024 14:42:37 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]: [ZeRO sharding] DP Rank 3 has 130M out of 520M (25.00%) params' optimizer states
+12/13/2024 14:42:38 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]: [Training Plan] Stage Stable Training Stage has 99 remaining training steps and has consumed 0 samples
+12/13/2024 14:42:38 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]: Using dummy data generator
+12/13/2024 14:42:38 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]: [Training Plan] There are 1 training stages 
+12/13/2024 14:42:38 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]: [Stage Stable Training Stage] start from step 1 
+12/13/2024 14:42:38 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]: 
+12/13/2024 14:42:38 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]: [Start training] datetime: 2024-12-13 14:42:38.341051 | mbs: 2 | grad_accum: 256 | global_batch_size: 2048 | sequence_length: 2048 | train_steps: 100 | start_iteration_step: 0 | consumed_train_samples: 0
+wandb: WARNING Path /fsx/nouamane/.cache/wandb/wandb/ wasn't writable, using system temp directory.
+wandb: WARNING Path /fsx/nouamane/.cache/wandb/wandb/ wasn't writable, using system temp directory
+wandb: Tracking run with wandb version 0.16.0
+wandb: W&B syncing is set to `offline` in this directory.  
+wandb: Run `wandb online` or set WANDB_MODE=online to enable cloud syncing.
+12/13/2024 14:42:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]: Waiting for 14 seconds
+12/13/2024 14:43:00 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]: Resuming training from stage Stable Training Stage, it has trained for 0 samples and has 99 remaining train steps
+12/13/2024 14:43:00 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:  Memory usage: 3472.48MiB. Peak allocated 5632.00MiB. Peak reserved: 17860.00MiB
+12/13/2024 14:43:20 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:  Memory usage: 3572.67MiB. Peak allocated 6985.16MiB. Peak reserved: 7434.00MiB
+12/13/2024 14:43:20 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]: iteration: 1 / 100 | consumed_tokens: 4.19M | elapsed_time_per_iteration_ms: 19.9K | tokens_per_sec: 211K | tokens_per_sec_per_gpu: 26.4K | global_batch_size: 2.05K | lm_loss: 10.8 | lr: 0.00015 | model_tflops_per_gpu: 186 | hardware_tflops_per_gpu: 186 | grad_norm: 0.431 | cuda_memory_allocated: 4.81G | cuda_max_memory_reserved: 9.62G | hd_total_memory_tb: 312G | hd_used_memory_tb: 69.7G | hd_free_memory_tb: 242G
+12/13/2024 14:43:20 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:  Memory usage: 4587.85MiB. Peak allocated 6820.14MiB. Peak reserved: 9178.00MiB
+12/13/2024 14:43:38 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:  Memory usage: 4587.97MiB. Peak allocated 8000.47MiB. Peak reserved: 9178.00MiB
+12/13/2024 14:43:38 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]: iteration: 2 / 100 | consumed_tokens: 8.39M | elapsed_time_per_iteration_ms: 18.1K | tokens_per_sec: 231K | tokens_per_sec_per_gpu: 28.9K | global_batch_size: 2.05K | lm_loss: 10.8 | lr: 0.0003 | model_tflops_per_gpu: 204 | hardware_tflops_per_gpu: 204 | grad_norm: 0.431 | cuda_memory_allocated: 4.81G | cuda_max_memory_reserved: 9.62G | hd_total_memory_tb: 312G | hd_used_memory_tb: 69.7G | hd_free_memory_tb: 242G
+12/13/2024 14:43:38 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:  Memory usage: 4587.85MiB. Peak allocated 6820.26MiB. Peak reserved: 9178.00MiB
+12/13/2024 14:43:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]:  Memory usage: 4587.97MiB. Peak allocated 8000.47MiB. Peak reserved: 9434.00MiB
+12/13/2024 14:43:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]: iteration: 3 / 100 | consumed_tokens: 12.6M | elapsed_time_per_iteration_ms: 18.4K | tokens_per_sec: 228K | tokens_per_sec_per_gpu: 28.6K | global_batch_size: 2.05K | lm_loss: 10.8 | lr: 0.000296 | model_tflops_per_gpu: 201 | hardware_tflops_per_gpu: 201 | grad_norm: 0.424 | cuda_memory_allocated: 4.81G | cuda_max_memory_reserved: 10.9G | hd_total_memory_tb: 312G | hd_used_memory_tb: 69.7G | hd_free_memory_tb: 242G
+12/13/2024 14:43:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]: | job_id   | name                                                           | nodes | seq_len | mbs | batch_accum | gbs  | mTFLOPs | hTFLOPs | tok/s/gpu | AllReduce (GB/s) | AllGather (GB/s) | ReduceScatter (GB/s) | AR Intra-node (GB/s) | AG Intra-node (GB/s) | RS Intra-node (GB/s) | Mem Alloc (GB) | Mem Res (GB) | dp | pp | tp | pp_engine                                                                                                 | tp_mode                                 | tp_async_comm | hidden_size | hidden_act | num_layers | num_heads | num_kv_heads | max_pos | vocab_size | tie_word_embeddings | dtype          | zero_stage | ddp_bucket_cap_mb | accumulate_grad_in_fp32 | Total Params | Local Params |
+12/13/2024 14:43:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]: | -------- | -------------------------------------------------------------- | ----- | ------- | --- | ----------- | ---- | ------- | ------- | --------- | ---------------- | ---------------- | -------------------- | -------------------- | -------------------- | -------------------- | -------------- | ------------ | -- | -- | -- | --------------------------------------------------------------------------------------------------------- | --------------------------------------- | ------------- | ----------- | ---------- | ---------- | --------- | ------------ | ------- | ---------- | ------------------- | -------------- | ---------- | ----------------- | ----------------------- | ------------ | ------------ |
+12/13/2024 14:43:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]: | 13417898 | 1.14G_dp4_tp2_pp1_acc256_mbs2_seq2048_zero1_tpmodeRED_vocab32k | 1     | 2048    | 2   | 256         | 2048 | 201.22  | 201.22  | 28556.67  | 459.03           | 263.37           | 263.93               | 460.49               | 264.17               | 264.29               | 6.66           | 10.18        | 4  | 1  | 2  | <nanotron.parallel.pipeline_parallel.engine.OneForwardOneBackwardPipelineEngine object at 0x7fecaa99cd00> | TensorParallelLinearMode.REDUCE_SCATTER | True          | 2048        | silu       | 16         | 32        | 8            | 2048    | 32768      | True                | torch.bfloat16 | 1          | 25                | True                    | 1.04G        | 520M         |
+12/13/2024 14:43:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]: Attempting to write benchmark results to CSV file: benchmark/results/bench_final.csv
+12/13/2024 14:43:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-207]: Acquired lock for CSV file: benchmark/results/bench_final.csv
+slurmstepd: error: *** STEP 13417898.0 ON ip-26-0-169-207 CANCELLED AT 2024-12-13T14:43:58 DUE TO TIME LIMIT ***
+slurmstepd: error: *** JOB 13417898 ON ip-26-0-169-207 CANCELLED AT 2024-12-13T14:43:58 DUE TO TIME LIMIT ***
+srun: Job step aborted: Waiting up to 32 seconds for job step to finish.
+[2024-12-13 14:43:58,670] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-13 14:43:58,670] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 734075 closing signal SIGTERM
+[2024-12-13 14:43:58,670] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 734076 closing signal SIGTERM
+[2024-12-13 14:43:58,670] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 734077 closing signal SIGTERM
+[2024-12-13 14:43:58,670] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 734078 closing signal SIGTERM
+[2024-12-13 14:43:58,672] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 734079 closing signal SIGTERM
+[2024-12-13 14:43:58,672] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 734080 closing signal SIGTERM
+[2024-12-13 14:43:58,674] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 734081 closing signal SIGTERM
+[2024-12-13 14:43:58,674] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 734082 closing signal SIGTERM
+[2024-12-13 14:43:58,833] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 734075 closing signal SIGTERM
+[2024-12-13 14:43:58,834] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 734076 closing signal SIGTERM
+[2024-12-13 14:43:58,834] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 734077 closing signal SIGTERM
+[2024-12-13 14:43:58,834] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 734078 closing signal SIGTERM
+[2024-12-13 14:43:58,834] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 734079 closing signal SIGTERM
+[2024-12-13 14:43:58,834] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 734080 closing signal SIGTERM
+[2024-12-13 14:43:58,834] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 734081 closing signal SIGTERM
+[2024-12-13 14:43:58,834] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 734082 closing signal SIGTERM
diff --git a/logs/13438454-bench_3.57G_dp16_tp2_pp1_acc16_mbs1_seq4096_zero1_tpmodeRED_vocab131k.out b/logs/13438454-bench_3.57G_dp16_tp2_pp1_acc16_mbs1_seq4096_zero1_tpmodeRED_vocab131k.out
new file mode 100644
index 0000000000000000000000000000000000000000..42001cdc43009c644db91b75fcdf51a1ec68aece
--- /dev/null
+++ b/logs/13438454-bench_3.57G_dp16_tp2_pp1_acc16_mbs1_seq4096_zero1_tpmodeRED_vocab131k.out
@@ -0,0 +1,642 @@
++ source /etc/profile.d/modules.sh
+++ . /usr/share/modules/init/bash
++++ unset _mlshdbg
++++ '[' 0 = 1 ']'
++++ unset _mlre _mlIFS
++++ '[' -n x ']'
++++ _mlIFS=' 	
+'
++++ IFS=' '
++++ '[' -n '' ']'
+++++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash autoinit
++++ _mlcode='module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+MODULEPATH_modshare=/etc/environment-modules/modules:1:/usr/share/modules/\$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1; export MODULEPATH_modshare;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+LOADEDMODULES=; export LOADEDMODULES;
+MODULEPATH=/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/\$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles; export MODULEPATH;
+test 0;'
++++ _mlret=0
++++ '[' -n x ']'
++++ IFS=' 	
+'
++++ unset _mlIFS
++++ unset _mlre _mlv _mlrv
++++ '[' 0 -eq 0 ']'
++++ eval 'module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+MODULEPATH_modshare=/etc/environment-modules/modules:1:/usr/share/modules/\$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1; export MODULEPATH_modshare;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+LOADEDMODULES=; export LOADEDMODULES;
+MODULEPATH=/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/\$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles; export MODULEPATH;
+test 0;'
+++++ MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl
+++++ export MODULES_CMD
+++++ ENV=/usr/share/modules/init/profile.sh
+++++ export ENV
+++++ MODULEPATH_modshare='/etc/environment-modules/modules:1:/usr/share/modules/$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1'
+++++ export MODULEPATH_modshare
+++++ BASH_ENV=/usr/share/modules/init/bash
+++++ export BASH_ENV
+++++ MODULESHOME=/usr/share/modules
+++++ export MODULESHOME
+++++ LOADEDMODULES=
+++++ export LOADEDMODULES
+++++ MODULEPATH='/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles'
+++++ export MODULEPATH
+++++ test 0
++++ '[' 0 = 1 ']'
++++ '[' -t 2 ']'
++++ export -f module
++++ export -f switchml
++++ '[' 5 -ge 3 ']'
++++ [[ ehxB =~ i ]]
++++ [[ ! :/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin: =~ :/usr/bin: ]]
++++ '[' '!' -n '' ']'
++++ MANPATH=:
++++ export MANPATH
+++++ manpath
++++ [[ ! :/admin/home/nouamane/.local/share/man:/fsx/nouamane/miniconda/envs/2-1-cu121/man:/fsx/nouamane/miniconda/envs/2-1-cu121/share/man:/fsx/nouamane/miniconda/man:/fsx/nouamane/miniconda/share/man:/opt/amazon/openmpi/share/man:/opt/amazon/efa/share/man:/opt/slurm/share/man:/usr/local/man:/usr/local/share/man:/usr/share/man:/admin/home/nouamane/.fzf/man:: =~ :/usr/share/man: ]]
++++ unset _mlcode _mlret
++++ '[' -n '' ']'
++ module load cuda/12.1
++ unset _mlshdbg
++ '[' 0 = 1 ']'
++ unset _mlre _mlIFS
++ '[' -n x ']'
++ _mlIFS=' 	
+'
++ IFS=' '
++ '[' -n '' ']'
+++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash load cuda/12.1
++ eval 'CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include;' export 'CPATH;
+LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:;' export 'LD_LIBRARY_PATH;
+FI_EFA_FORK_SAFE=1;' export 'FI_EFA_FORK_SAFE;
+MANPATH=/usr/local/cuda-12.1/share/man::;' export 'MANPATH;
+LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64;' export 'LIBRARY_PATH;
+_LMFILES_=/usr/share/modules/modulefiles/cuda/12.1;' export '_LMFILES_;
+LOADEDMODULES=cuda/12.1;' export 'LOADEDMODULES;
+MPI_PATH=/opt/amazon/openmpi;' export 'MPI_PATH;
+NCCL_HOME_modshare=/opt/nccl/build:1;' export 'NCCL_HOME_modshare;
+NCCL_PROTO=simple;' export 'NCCL_PROTO;
+MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1;' export 'MANPATH_modshare;
+LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1;' export 'LIBRARY_PATH_modshare;
+NCCL_SOCKET_IFNAME=enp;' export 'NCCL_SOCKET_IFNAME;
+AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl;' export 'AWS_OFI_NCCL_HOME;
+NCCL_HOME=/opt/nccl/build;' export 'NCCL_HOME;
+FI_PROVIDER=efa;' export 'FI_PROVIDER;
+AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1;' export 'AWS_OFI_NCCL_HOME_modshare;
+CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1;' export 'CPATH_modshare;
+LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2::1:/opt/nccl/build/lib:1:/opt/aws-ofi-nccl/lib:2:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/efa/lib:1;' export 'LD_LIBRARY_PATH_modshare;
+FI_EFA_ENABLE_SHM_TRANSFER=1;' export 'FI_EFA_ENABLE_SHM_TRANSFER;
+_LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1;' export '_LMFILES__modshare;
+LOADEDMODULES_modshare=cuda/12.1:1;' export 'LOADEDMODULES_modshare;
+MPI_PATH_modshare=/opt/amazon/openmpi:1;' export 'MPI_PATH_modshare;
+PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin;' export 'PATH;
+CUDA_HOME=/usr/local/cuda-12.1;' export 'CUDA_HOME;
+PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/usr/local/cuda-12.1/include:1:/opt/amazon/efa/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/usr/sbin:1:/fsx/nouamane/miniconda/bin:1:/opt/slurm/bin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:2:/opt/amazon/openmpi/bin:1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1;' export 'PATH_modshare;
+test' '0;'
+++ CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include
+++ export CPATH
+++ LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:
+++ export LD_LIBRARY_PATH
+++ FI_EFA_FORK_SAFE=1
+++ export FI_EFA_FORK_SAFE
+++ MANPATH=/usr/local/cuda-12.1/share/man::
+++ export MANPATH
+++ LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64
+++ export LIBRARY_PATH
+++ _LMFILES_=/usr/share/modules/modulefiles/cuda/12.1
+++ export _LMFILES_
+++ LOADEDMODULES=cuda/12.1
+++ export LOADEDMODULES
+++ MPI_PATH=/opt/amazon/openmpi
+++ export MPI_PATH
+++ NCCL_HOME_modshare=/opt/nccl/build:1
+++ export NCCL_HOME_modshare
+++ NCCL_PROTO=simple
+++ export NCCL_PROTO
+++ MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1
+++ export MANPATH_modshare
+++ LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1
+++ export LIBRARY_PATH_modshare
+++ NCCL_SOCKET_IFNAME=enp
+++ export NCCL_SOCKET_IFNAME
+++ AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl
+++ export AWS_OFI_NCCL_HOME
+++ NCCL_HOME=/opt/nccl/build
+++ export NCCL_HOME
+++ FI_PROVIDER=efa
+++ export FI_PROVIDER
+++ AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1
+++ export AWS_OFI_NCCL_HOME_modshare
+++ CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1
+++ export CPATH_modshare
+++ LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2::1:/opt/nccl/build/lib:1:/opt/aws-ofi-nccl/lib:2:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/efa/lib:1
+++ export LD_LIBRARY_PATH_modshare
+++ FI_EFA_ENABLE_SHM_TRANSFER=1
+++ export FI_EFA_ENABLE_SHM_TRANSFER
+++ _LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1
+++ export _LMFILES__modshare
+++ LOADEDMODULES_modshare=cuda/12.1:1
+++ export LOADEDMODULES_modshare
+++ MPI_PATH_modshare=/opt/amazon/openmpi:1
+++ export MPI_PATH_modshare
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ export PATH
+++ CUDA_HOME=/usr/local/cuda-12.1
+++ export CUDA_HOME
+++ PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/usr/local/cuda-12.1/include:1:/opt/amazon/efa/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/usr/sbin:1:/fsx/nouamane/miniconda/bin:1:/opt/slurm/bin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:2:/opt/amazon/openmpi/bin:1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1
+++ export PATH_modshare
+++ test 0
++ _mlstatus=0
++ '[' -n x ']'
++ IFS=' 	
+'
++ unset _mlre _mlv _mlrv _mlIFS
++ '[' -n '' ']'
++ unset _mlshdbg
++ return 0
++ source /fsx/nouamane/miniconda/bin/activate
+++ _CONDA_ROOT=/fsx/nouamane/miniconda
+++ . /fsx/nouamane/miniconda/etc/profile.d/conda.sh
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ '[' -z x ']'
+++ conda activate
+++ local cmd=activate
+++ case "$cmd" in
+++ __conda_activate activate
+++ '[' -n '' ']'
+++ local ask_conda
++++ PS1=
++++ __conda_exe shell.posix activate
++++ /fsx/nouamane/miniconda/bin/conda shell.posix activate
+++ ask_conda='. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_2='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\'''
+++ eval '. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_2='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\'''
++++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh
+++++ test -n ''
+++++ unset XML_CATALOG_FILES
+++++ unset xml_catalog_files_libxml2
++++ PS1='(base) '
++++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++++ export CONDA_PREFIX=/fsx/nouamane/miniconda
++++ CONDA_PREFIX=/fsx/nouamane/miniconda
++++ export CONDA_SHLVL=3
++++ CONDA_SHLVL=3
++++ export CONDA_DEFAULT_ENV=base
++++ CONDA_DEFAULT_ENV=base
++++ export 'CONDA_PROMPT_MODIFIER=(base) '
++++ CONDA_PROMPT_MODIFIER='(base) '
++++ export CONDA_PREFIX_2=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ CONDA_PREFIX_2=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ __conda_hashr
+++ '[' -n '' ']'
+++ '[' -n '' ']'
+++ hash -r
++ conda activate 2-1-cu121
++ local cmd=activate
++ case "$cmd" in
++ __conda_activate activate 2-1-cu121
++ '[' -n '' ']'
++ local ask_conda
+++ PS1='(base) '
+++ __conda_exe shell.posix activate 2-1-cu121
+++ /fsx/nouamane/miniconda/bin/conda shell.posix activate 2-1-cu121
++ ask_conda='PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''4'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_3='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
++ eval 'PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''4'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_3='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
+++ PS1='(2-1-cu121) '
+++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ export CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ export CONDA_SHLVL=4
+++ CONDA_SHLVL=4
+++ export CONDA_DEFAULT_ENV=2-1-cu121
+++ CONDA_DEFAULT_ENV=2-1-cu121
+++ export 'CONDA_PROMPT_MODIFIER=(2-1-cu121) '
+++ CONDA_PROMPT_MODIFIER='(2-1-cu121) '
+++ export CONDA_PREFIX_3=/fsx/nouamane/miniconda
+++ CONDA_PREFIX_3=/fsx/nouamane/miniconda
+++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ export _CE_M=
+++ _CE_M=
+++ export _CE_CONDA=
+++ _CE_CONDA=
+++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh
++++ test -n ''
++++ xml_catalog_files_libxml2=
++++ XML_CATALOG_FILES=
++++ conda_catalog_files=
++++ ifs_libxml2=' 	
+'
++++ IFS=' '
++++ rem=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ for pre in ${rem}
++++ test '' = /fsx/nouamane/miniconda/envs/2-1-cu121
++++ conda_catalog_files=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ rem=
++++ IFS=' 	
+'
++++ conda_catalog_files='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ export 'XML_CATALOG_FILES=file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ XML_CATALOG_FILES='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ unset conda_catalog_files ifs_libxml2 rem
++ __conda_hashr
++ '[' -n '' ']'
++ '[' -n '' ']'
++ hash -r
++ export PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++ PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ scontrol show hostnames 'ip-26-0-160-[100,103],ip-26-0-165-[202,213]'
++ export 'NODELIST=ip-26-0-160-100
+ip-26-0-160-103
+ip-26-0-165-202
+ip-26-0-165-213'
++ NODELIST='ip-26-0-160-100
+ip-26-0-160-103
+ip-26-0-165-202
+ip-26-0-165-213'
+++ head -n1
+++ scontrol show hostnames 'ip-26-0-160-[100,103],ip-26-0-165-[202,213]'
++ export MASTER_NODE=ip-26-0-160-100
++ MASTER_NODE=ip-26-0-160-100
++ export MASTER_PORT=12356
++ MASTER_PORT=12356
++ export NNODES=4
++ NNODES=4
++ export GPUS_PER_NODE=8
++ GPUS_PER_NODE=8
++ export WORLD_SIZE=32
++ WORLD_SIZE=32
++ export CUDA_DEVICE_MAX_CONNECTIONS=1
++ CUDA_DEVICE_MAX_CONNECTIONS=1
++ export NCCL_DEBUG=WARN
++ NCCL_DEBUG=WARN
++ export NANOTRON_BENCHMARK=1
++ NANOTRON_BENCHMARK=1
++ export WANDB_MODE=disabled
++ WANDB_MODE=disabled
++ export TORCH_NCCL_ASYNC_ERROR_HANDLING=1
++ TORCH_NCCL_ASYNC_ERROR_HANDLING=1
++ echo '=== GPU Topology ==='
+=== GPU Topology ===
++ nvidia-smi topo -m
+	GPU0	GPU1	GPU2	GPU3	GPU4	GPU5	GPU6	GPU7	CPU Affinity	NUMA Affinity	GPU NUMA ID
+GPU0	 X 	NV18	NV18	NV18	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU1	NV18	 X 	NV18	NV18	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU2	NV18	NV18	 X 	NV18	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU3	NV18	NV18	NV18	 X 	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU4	NV18	NV18	NV18	NV18	 X 	NV18	NV18	NV18	48-95	1		N/A
+GPU5	NV18	NV18	NV18	NV18	NV18	 X 	NV18	NV18	48-95	1		N/A
+GPU6	NV18	NV18	NV18	NV18	NV18	NV18	 X 	NV18	48-95	1		N/A
+GPU7	NV18	NV18	NV18	NV18	NV18	NV18	NV18	 X 	48-95	1		N/A
+
+Legend:
+
+  X    = Self
+  SYS  = Connection traversing PCIe as well as the SMP interconnect between NUMA nodes (e.g., QPI/UPI)
+  NODE = Connection traversing PCIe as well as the interconnect between PCIe Host Bridges within a NUMA node
+  PHB  = Connection traversing PCIe as well as a PCIe Host Bridge (typically the CPU)
+  PXB  = Connection traversing multiple PCIe bridges (without traversing the PCIe Host Bridge)
+  PIX  = Connection traversing at most a single PCIe bridge
+  NV#  = Connection traversing a bonded set of # NVLinks
++ echo ==================
+==================
++ echo 'Master node: ip-26-0-160-100'
+Master node: ip-26-0-160-100
++ echo 'All nodes: ip-26-0-160-100
+ip-26-0-160-103
+ip-26-0-165-202
+ip-26-0-165-213'
+All nodes: ip-26-0-160-100
+ip-26-0-160-103
+ip-26-0-165-202
+ip-26-0-165-213
++ echo 'World size: 32'
+World size: 32
++ srun --wait=0 --kill-on-bad-exit=1 torchrun --nnodes=4 --nproc_per_node=8 --rdzv_id=13438454 --rdzv_backend=c10d --rdzv_endpoint=ip-26-0-160-100:12356 --max_restarts 0 --rdzv_conf timeout=60 /fsx/nouamane/projects/nanotron/run_train.py --config-file benchmark/configs/config_3.57G_dp16_tp2_pp1_acc16_mbs1_seq4096_zero1_tpmodeRED_vocab131k.yaml
+[2024-12-14 01:19:45,000] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-14 01:19:45,000] torch.distributed.run: [WARNING] 
+[2024-12-14 01:19:45,000] torch.distributed.run: [WARNING] *****************************************
+[2024-12-14 01:19:45,000] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-14 01:19:45,000] torch.distributed.run: [WARNING] *****************************************
+[2024-12-14 01:19:45,047] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-14 01:19:45,048] torch.distributed.run: [WARNING] 
+[2024-12-14 01:19:45,048] torch.distributed.run: [WARNING] *****************************************
+[2024-12-14 01:19:45,048] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-14 01:19:45,048] torch.distributed.run: [WARNING] *****************************************
+[2024-12-14 01:19:45,081] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-14 01:19:45,081] torch.distributed.run: [WARNING] 
+[2024-12-14 01:19:45,081] torch.distributed.run: [WARNING] *****************************************
+[2024-12-14 01:19:45,081] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-14 01:19:45,081] torch.distributed.run: [WARNING] *****************************************
+[2024-12-14 01:19:45,121] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-14 01:19:45,121] torch.distributed.run: [WARNING] 
+[2024-12-14 01:19:45,121] torch.distributed.run: [WARNING] *****************************************
+[2024-12-14 01:19:45,121] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-14 01:19:45,121] torch.distributed.run: [WARNING] *****************************************
+[W socket.cpp:436] [c10d] The server socket has failed to bind to [::]:48149 (errno: 98 - Address already in use).
+[W socket.cpp:436] [c10d] The server socket has failed to bind to ?UNKNOWN? (errno: 98 - Address already in use).
+[E socket.cpp:472] [c10d] The server socket has failed to listen on any local network address.
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 248, in <module>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 147, in __init__
+    self.parallel_context = ParallelContext(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/context.py", line 56, in __init__
+    dist.initialize_torch_distributed()
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/distributed.py", line 274, in initialize_torch_distributed
+    dist.init_process_group(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/c10d_logger.py", line 74, in wrapper
+    func_return = func(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/distributed_c10d.py", line 1141, in init_process_group
+    store, rank, world_size = next(rendezvous_iterator)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/rendezvous.py", line 241, in _env_rendezvous_handler
+    store = _create_c10d_store(master_addr, master_port, rank, world_size, timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/rendezvous.py", line 172, in _create_c10d_store
+    return TCPStore(
+RuntimeError: The server socket has failed to listen on any local network address. The server socket has failed to bind to [::]:48149 (errno: 98 - Address already in use). The server socket has failed to bind to ?UNKNOWN? (errno: 98 - Address already in use).
+[2024-12-14 01:20:02,403] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 48550 closing signal SIGTERM
+[2024-12-14 01:20:02,404] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 48551 closing signal SIGTERM
+[2024-12-14 01:20:02,404] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 48552 closing signal SIGTERM
+[2024-12-14 01:20:02,404] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 48553 closing signal SIGTERM
+[2024-12-14 01:20:02,405] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 48554 closing signal SIGTERM
+[2024-12-14 01:20:02,405] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 48555 closing signal SIGTERM
+[2024-12-14 01:20:02,405] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 48556 closing signal SIGTERM
+[2024-12-14 01:20:02,846] torch.distributed.elastic.multiprocessing.api: [ERROR] failed (exitcode: 1) local_rank: 0 (pid: 48549) of binary: /fsx/nouamane/miniconda/envs/2-1-cu121/bin/python
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 33, in <module>
+    sys.exit(load_entry_point('torch==2.1.1', 'console_scripts', 'torchrun')())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 264, in launch_agent
+    raise ChildFailedError(
+torch.distributed.elastic.multiprocessing.errors.ChildFailedError: 
+============================================================
+/fsx/nouamane/projects/nanotron/run_train.py FAILED
+------------------------------------------------------------
+Failures:
+  <NO_OTHER_FAILURES>
+------------------------------------------------------------
+Root Cause (first observed failure):
+[0]:
+  time      : 2024-12-14_01:20:02
+  host      : ip-26-0-160-100.ec2.internal
+  rank      : 0 (local_rank: 0)
+  exitcode  : 1 (pid: 48549)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+============================================================
+srun: error: ip-26-0-160-100: task 0: Exited with exit code 1
+srun: launch/slurm: _step_signal: Terminating StepId=13438454.0
+[2024-12-14 01:20:04,224] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-14 01:20:04,224] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 47810 closing signal SIGTERM
+[2024-12-14 01:20:04,224] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 47811 closing signal SIGTERM
+[2024-12-14 01:20:04,223] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-14 01:20:04,223] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 49350 closing signal SIGTERM
+[2024-12-14 01:20:04,225] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 47812 closing signal SIGTERM
+[2024-12-14 01:20:04,223] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 49351 closing signal SIGTERM
+[2024-12-14 01:20:04,224] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-14 01:20:04,224] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 23122 closing signal SIGTERM
+[2024-12-14 01:20:04,224] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 23123 closing signal SIGTERM
+[2024-12-14 01:20:04,225] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 47813 closing signal SIGTERM
+[2024-12-14 01:20:04,224] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 49352 closing signal SIGTERM
+[2024-12-14 01:20:04,225] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 47814 closing signal SIGTERM
+[2024-12-14 01:20:04,224] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 49353 closing signal SIGTERM
+[2024-12-14 01:20:04,225] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 47815 closing signal SIGTERM
+[2024-12-14 01:20:04,224] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 49354 closing signal SIGTERM
+[2024-12-14 01:20:04,225] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 47816 closing signal SIGTERM
+[2024-12-14 01:20:04,224] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 49355 closing signal SIGTERM
+[2024-12-14 01:20:04,225] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 47817 closing signal SIGTERM
+[2024-12-14 01:20:04,224] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 49356 closing signal SIGTERM
+[2024-12-14 01:20:04,224] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 23124 closing signal SIGTERM
+[2024-12-14 01:20:04,224] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 49357 closing signal SIGTERM
+[2024-12-14 01:20:04,225] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 23125 closing signal SIGTERM
+[2024-12-14 01:20:04,225] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 23126 closing signal SIGTERM
+[2024-12-14 01:20:04,225] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 23127 closing signal SIGTERM
+[2024-12-14 01:20:04,225] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 23128 closing signal SIGTERM
+[2024-12-14 01:20:04,225] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 23129 closing signal SIGTERM
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 33, in <module>
+    sys.exit(load_entry_point('torch==2.1.1', 'console_scripts', 'torchrun')())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 23048 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 33, in <module>
+    sys.exit(load_entry_point('torch==2.1.1', 'console_scripts', 'torchrun')())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 49279 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 33, in <module>
+    sys.exit(load_entry_point('torch==2.1.1', 'console_scripts', 'torchrun')())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 47738 got signal: 15
+srun: error: ip-26-0-165-213: task 3: Exited with exit code 1
+srun: error: ip-26-0-160-103: task 1: Exited with exit code 1
+srun: error: ip-26-0-165-202: task 2: Exited with exit code 1
diff --git a/logs/13442265-bench_3.57G_dp1_tp2_pp4_acc1_mbs256_seq4096_zero0_tpmodeRED_vocab131k.out b/logs/13442265-bench_3.57G_dp1_tp2_pp4_acc1_mbs256_seq4096_zero0_tpmodeRED_vocab131k.out
new file mode 100644
index 0000000000000000000000000000000000000000..e3cb2363ca12ebc2b80089e79998fc5e8fdbd10d
--- /dev/null
+++ b/logs/13442265-bench_3.57G_dp1_tp2_pp4_acc1_mbs256_seq4096_zero0_tpmodeRED_vocab131k.out
@@ -0,0 +1,765 @@
++ source /etc/profile.d/modules.sh
+++ . /usr/share/modules/init/bash
++++ unset _mlshdbg
++++ '[' 0 = 1 ']'
++++ unset _mlre _mlIFS
++++ '[' -n x ']'
++++ _mlIFS=' 	
+'
++++ IFS=' '
++++ '[' -n '' ']'
+++++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash autoinit
++++ _mlcode='module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+MODULEPATH_modshare=/etc/environment-modules/modules:1:/usr/share/modules/\$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1; export MODULEPATH_modshare;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+LOADEDMODULES=; export LOADEDMODULES;
+MODULEPATH=/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/\$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles; export MODULEPATH;
+test 0;'
++++ _mlret=0
++++ '[' -n x ']'
++++ IFS=' 	
+'
++++ unset _mlIFS
++++ unset _mlre _mlv _mlrv
++++ '[' 0 -eq 0 ']'
++++ eval 'module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+MODULEPATH_modshare=/etc/environment-modules/modules:1:/usr/share/modules/\$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1; export MODULEPATH_modshare;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+LOADEDMODULES=; export LOADEDMODULES;
+MODULEPATH=/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/\$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles; export MODULEPATH;
+test 0;'
+++++ MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl
+++++ export MODULES_CMD
+++++ ENV=/usr/share/modules/init/profile.sh
+++++ export ENV
+++++ MODULEPATH_modshare='/etc/environment-modules/modules:1:/usr/share/modules/$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1'
+++++ export MODULEPATH_modshare
+++++ BASH_ENV=/usr/share/modules/init/bash
+++++ export BASH_ENV
+++++ MODULESHOME=/usr/share/modules
+++++ export MODULESHOME
+++++ LOADEDMODULES=
+++++ export LOADEDMODULES
+++++ MODULEPATH='/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles'
+++++ export MODULEPATH
+++++ test 0
++++ '[' 0 = 1 ']'
++++ '[' -t 2 ']'
++++ export -f module
++++ export -f switchml
++++ '[' 5 -ge 3 ']'
++++ [[ ehxB =~ i ]]
++++ [[ ! :/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin: =~ :/usr/bin: ]]
++++ '[' '!' -n '' ']'
++++ MANPATH=:
++++ export MANPATH
+++++ manpath
++++ [[ ! :/admin/home/nouamane/.local/share/man:/fsx/nouamane/miniconda/envs/2-1-cu121/man:/fsx/nouamane/miniconda/envs/2-1-cu121/share/man:/fsx/nouamane/miniconda/man:/fsx/nouamane/miniconda/share/man:/opt/amazon/openmpi/share/man:/opt/amazon/efa/share/man:/opt/slurm/share/man:/usr/local/man:/usr/local/share/man:/usr/share/man:/admin/home/nouamane/.fzf/man:: =~ :/usr/share/man: ]]
++++ unset _mlcode _mlret
++++ '[' -n '' ']'
++ module load cuda/12.1
++ unset _mlshdbg
++ '[' 0 = 1 ']'
++ unset _mlre _mlIFS
++ '[' -n x ']'
++ _mlIFS=' 	
+'
++ IFS=' '
++ '[' -n '' ']'
+++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash load cuda/12.1
++ eval 'CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include;' export 'CPATH;
+LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:;' export 'LD_LIBRARY_PATH;
+FI_EFA_FORK_SAFE=1;' export 'FI_EFA_FORK_SAFE;
+MANPATH=/usr/local/cuda-12.1/share/man::;' export 'MANPATH;
+LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64;' export 'LIBRARY_PATH;
+_LMFILES_=/usr/share/modules/modulefiles/cuda/12.1;' export '_LMFILES_;
+LOADEDMODULES=cuda/12.1;' export 'LOADEDMODULES;
+MPI_PATH=/opt/amazon/openmpi;' export 'MPI_PATH;
+NCCL_HOME_modshare=/opt/nccl/build:1;' export 'NCCL_HOME_modshare;
+NCCL_PROTO=simple;' export 'NCCL_PROTO;
+MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1;' export 'MANPATH_modshare;
+LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1;' export 'LIBRARY_PATH_modshare;
+NCCL_SOCKET_IFNAME=enp;' export 'NCCL_SOCKET_IFNAME;
+AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl;' export 'AWS_OFI_NCCL_HOME;
+NCCL_HOME=/opt/nccl/build;' export 'NCCL_HOME;
+FI_PROVIDER=efa;' export 'FI_PROVIDER;
+AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1;' export 'AWS_OFI_NCCL_HOME_modshare;
+CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1;' export 'CPATH_modshare;
+LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2::1:/opt/nccl/build/lib:1:/opt/aws-ofi-nccl/lib:2:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/efa/lib:1;' export 'LD_LIBRARY_PATH_modshare;
+FI_EFA_ENABLE_SHM_TRANSFER=1;' export 'FI_EFA_ENABLE_SHM_TRANSFER;
+_LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1;' export '_LMFILES__modshare;
+LOADEDMODULES_modshare=cuda/12.1:1;' export 'LOADEDMODULES_modshare;
+MPI_PATH_modshare=/opt/amazon/openmpi:1;' export 'MPI_PATH_modshare;
+PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin;' export 'PATH;
+CUDA_HOME=/usr/local/cuda-12.1;' export 'CUDA_HOME;
+PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/usr/local/cuda-12.1/include:1:/opt/amazon/efa/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/usr/sbin:1:/fsx/nouamane/miniconda/bin:1:/opt/slurm/bin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:2:/opt/amazon/openmpi/bin:1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1;' export 'PATH_modshare;
+test' '0;'
+++ CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include
+++ export CPATH
+++ LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:
+++ export LD_LIBRARY_PATH
+++ FI_EFA_FORK_SAFE=1
+++ export FI_EFA_FORK_SAFE
+++ MANPATH=/usr/local/cuda-12.1/share/man::
+++ export MANPATH
+++ LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64
+++ export LIBRARY_PATH
+++ _LMFILES_=/usr/share/modules/modulefiles/cuda/12.1
+++ export _LMFILES_
+++ LOADEDMODULES=cuda/12.1
+++ export LOADEDMODULES
+++ MPI_PATH=/opt/amazon/openmpi
+++ export MPI_PATH
+++ NCCL_HOME_modshare=/opt/nccl/build:1
+++ export NCCL_HOME_modshare
+++ NCCL_PROTO=simple
+++ export NCCL_PROTO
+++ MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1
+++ export MANPATH_modshare
+++ LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1
+++ export LIBRARY_PATH_modshare
+++ NCCL_SOCKET_IFNAME=enp
+++ export NCCL_SOCKET_IFNAME
+++ AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl
+++ export AWS_OFI_NCCL_HOME
+++ NCCL_HOME=/opt/nccl/build
+++ export NCCL_HOME
+++ FI_PROVIDER=efa
+++ export FI_PROVIDER
+++ AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1
+++ export AWS_OFI_NCCL_HOME_modshare
+++ CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1
+++ export CPATH_modshare
+++ LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2::1:/opt/nccl/build/lib:1:/opt/aws-ofi-nccl/lib:2:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/efa/lib:1
+++ export LD_LIBRARY_PATH_modshare
+++ FI_EFA_ENABLE_SHM_TRANSFER=1
+++ export FI_EFA_ENABLE_SHM_TRANSFER
+++ _LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1
+++ export _LMFILES__modshare
+++ LOADEDMODULES_modshare=cuda/12.1:1
+++ export LOADEDMODULES_modshare
+++ MPI_PATH_modshare=/opt/amazon/openmpi:1
+++ export MPI_PATH_modshare
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ export PATH
+++ CUDA_HOME=/usr/local/cuda-12.1
+++ export CUDA_HOME
+++ PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/usr/local/cuda-12.1/include:1:/opt/amazon/efa/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/usr/sbin:1:/fsx/nouamane/miniconda/bin:1:/opt/slurm/bin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:2:/opt/amazon/openmpi/bin:1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1
+++ export PATH_modshare
+++ test 0
++ _mlstatus=0
++ '[' -n x ']'
++ IFS=' 	
+'
++ unset _mlre _mlv _mlrv _mlIFS
++ '[' -n '' ']'
++ unset _mlshdbg
++ return 0
++ source /fsx/nouamane/miniconda/bin/activate
+++ _CONDA_ROOT=/fsx/nouamane/miniconda
+++ . /fsx/nouamane/miniconda/etc/profile.d/conda.sh
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ '[' -z x ']'
+++ conda activate
+++ local cmd=activate
+++ case "$cmd" in
+++ __conda_activate activate
+++ '[' -n '' ']'
+++ local ask_conda
++++ PS1=
++++ __conda_exe shell.posix activate
++++ /fsx/nouamane/miniconda/bin/conda shell.posix activate
+++ ask_conda='. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_2='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\'''
+++ eval '. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_2='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\'''
++++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh
+++++ test -n ''
+++++ unset XML_CATALOG_FILES
+++++ unset xml_catalog_files_libxml2
++++ PS1='(base) '
++++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++++ export CONDA_PREFIX=/fsx/nouamane/miniconda
++++ CONDA_PREFIX=/fsx/nouamane/miniconda
++++ export CONDA_SHLVL=3
++++ CONDA_SHLVL=3
++++ export CONDA_DEFAULT_ENV=base
++++ CONDA_DEFAULT_ENV=base
++++ export 'CONDA_PROMPT_MODIFIER=(base) '
++++ CONDA_PROMPT_MODIFIER='(base) '
++++ export CONDA_PREFIX_2=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ CONDA_PREFIX_2=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ __conda_hashr
+++ '[' -n '' ']'
+++ '[' -n '' ']'
+++ hash -r
++ conda activate 2-1-cu121
++ local cmd=activate
++ case "$cmd" in
++ __conda_activate activate 2-1-cu121
++ '[' -n '' ']'
++ local ask_conda
+++ PS1='(base) '
+++ __conda_exe shell.posix activate 2-1-cu121
+++ /fsx/nouamane/miniconda/bin/conda shell.posix activate 2-1-cu121
++ ask_conda='PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''4'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_3='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
++ eval 'PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''4'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_3='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
+++ PS1='(2-1-cu121) '
+++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ export CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ export CONDA_SHLVL=4
+++ CONDA_SHLVL=4
+++ export CONDA_DEFAULT_ENV=2-1-cu121
+++ CONDA_DEFAULT_ENV=2-1-cu121
+++ export 'CONDA_PROMPT_MODIFIER=(2-1-cu121) '
+++ CONDA_PROMPT_MODIFIER='(2-1-cu121) '
+++ export CONDA_PREFIX_3=/fsx/nouamane/miniconda
+++ CONDA_PREFIX_3=/fsx/nouamane/miniconda
+++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ export _CE_M=
+++ _CE_M=
+++ export _CE_CONDA=
+++ _CE_CONDA=
+++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh
++++ test -n ''
++++ xml_catalog_files_libxml2=
++++ XML_CATALOG_FILES=
++++ conda_catalog_files=
++++ ifs_libxml2=' 	
+'
++++ IFS=' '
++++ rem=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ for pre in ${rem}
++++ test '' = /fsx/nouamane/miniconda/envs/2-1-cu121
++++ conda_catalog_files=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ rem=
++++ IFS=' 	
+'
++++ conda_catalog_files='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ export 'XML_CATALOG_FILES=file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ XML_CATALOG_FILES='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ unset conda_catalog_files ifs_libxml2 rem
++ __conda_hashr
++ '[' -n '' ']'
++ '[' -n '' ']'
++ hash -r
++ export PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++ PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ scontrol show hostnames ip-26-0-171-88
++ export NODELIST=ip-26-0-171-88
++ NODELIST=ip-26-0-171-88
+++ scontrol show hostnames ip-26-0-171-88
+++ head -n1
++ export MASTER_NODE=ip-26-0-171-88
++ MASTER_NODE=ip-26-0-171-88
++ export MASTER_PORT=12356
++ MASTER_PORT=12356
++ export NNODES=1
++ NNODES=1
++ export GPUS_PER_NODE=8
++ GPUS_PER_NODE=8
++ export WORLD_SIZE=8
++ WORLD_SIZE=8
++ export CUDA_DEVICE_MAX_CONNECTIONS=1
++ CUDA_DEVICE_MAX_CONNECTIONS=1
++ export NCCL_DEBUG=WARN
++ NCCL_DEBUG=WARN
++ export NANOTRON_BENCHMARK=1
++ NANOTRON_BENCHMARK=1
++ export WANDB_MODE=disabled
++ WANDB_MODE=disabled
++ export TORCH_NCCL_ASYNC_ERROR_HANDLING=1
++ TORCH_NCCL_ASYNC_ERROR_HANDLING=1
++ echo '=== GPU Topology ==='
+=== GPU Topology ===
++ nvidia-smi topo -m
+	GPU0	GPU1	GPU2	GPU3	GPU4	GPU5	GPU6	GPU7	CPU Affinity	NUMA Affinity	GPU NUMA ID
+GPU0	 X 	NV18	NV18	NV18	NV18	NV18	NV18	NV18	0-43	0		N/A
+GPU1	NV18	 X 	NV18	NV18	NV18	NV18	NV18	NV18	0-43	0		N/A
+GPU2	NV18	NV18	 X 	NV18	NV18	NV18	NV18	NV18	0-43	0		N/A
+GPU3	NV18	NV18	NV18	 X 	NV18	NV18	NV18	NV18	0-43	0		N/A
+GPU4	NV18	NV18	NV18	NV18	 X 	NV18	NV18	NV18	52-95	1		N/A
+GPU5	NV18	NV18	NV18	NV18	NV18	 X 	NV18	NV18	52-95	1		N/A
+GPU6	NV18	NV18	NV18	NV18	NV18	NV18	 X 	NV18	52-95	1		N/A
+GPU7	NV18	NV18	NV18	NV18	NV18	NV18	NV18	 X 	52-95	1		N/A
+
+Legend:
+
+  X    = Self
+  SYS  = Connection traversing PCIe as well as the SMP interconnect between NUMA nodes (e.g., QPI/UPI)
+  NODE = Connection traversing PCIe as well as the interconnect between PCIe Host Bridges within a NUMA node
+  PHB  = Connection traversing PCIe as well as a PCIe Host Bridge (typically the CPU)
+  PXB  = Connection traversing multiple PCIe bridges (without traversing the PCIe Host Bridge)
+  PIX  = Connection traversing at most a single PCIe bridge
+  NV#  = Connection traversing a bonded set of # NVLinks
++ echo ==================
+==================
++ echo 'Master node: ip-26-0-171-88'
+Master node: ip-26-0-171-88
++ echo 'All nodes: ip-26-0-171-88'
+All nodes: ip-26-0-171-88
++ echo 'World size: 8'
+World size: 8
++ srun --wait=0 --kill-on-bad-exit=1 torchrun --nnodes=1 --nproc_per_node=8 --rdzv_id=13442265 --rdzv_backend=c10d --rdzv_endpoint=ip-26-0-171-88:12356 --max_restarts 0 --rdzv_conf timeout=60 /fsx/nouamane/projects/nanotron/run_train.py --config-file benchmark/configs/config_3.57G_dp1_tp2_pp4_acc1_mbs256_seq4096_zero0_tpmodeRED_vocab131k.yaml
+[2024-12-14 03:49:28,311] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-14 03:49:28,311] torch.distributed.run: [WARNING] 
+[2024-12-14 03:49:28,311] torch.distributed.run: [WARNING] *****************************************
+[2024-12-14 03:49:28,311] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-14 03:49:28,311] torch.distributed.run: [WARNING] *****************************************
+NCCL version 2.18.5+cuda12.2
+12/14/2024 03:49:47 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-88]: Measuring inter-GPU and intra-node bandwidth...
+12/14/2024 03:50:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-88]: Bandwidth measurement complete. Time taken: 14.25 seconds
+12/14/2024 03:50:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-88]: Config:
+12/14/2024 03:50:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-88]: Config(general=GeneralArgs(project='debug',
+12/14/2024 03:50:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-88]:                            run='3.57G_dp1_tp2_pp4_acc1_mbs256_seq4096_zero0_tpmodeRED_vocab131k',
+12/14/2024 03:50:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-88]:                            seed=42,
+12/14/2024 03:50:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-88]:                            step=None,
+12/14/2024 03:50:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-88]:                            consumed_train_samples=None,
+12/14/2024 03:50:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-88]:                            benchmark_csv_path=PosixPath('benchmark/results/bench_final2.csv'),
+12/14/2024 03:50:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-88]:                            ignore_sanity_checks=True),
+12/14/2024 03:50:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-88]:        parallelism=ParallelismArgs(dp=1,
+12/14/2024 03:50:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-88]:                                    pp=4,
+12/14/2024 03:50:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-88]:                                    tp=2,
+12/14/2024 03:50:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-88]:                                    pp_engine=<nanotron.parallel.pipeline_parallel.engine.OneForwardOneBackwardPipelineEngine object at 0x7f2fe686ce20>,
+12/14/2024 03:50:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-88]:                                    tp_mode=<TensorParallelLinearMode.REDUCE_SCATTER: 2>,
+12/14/2024 03:50:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-88]:                                    tp_linear_async_communication=True,
+12/14/2024 03:50:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-88]:                                    recompute_layer=False,
+12/14/2024 03:50:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-88]:                                    tp_recompute_allgather=True,
+12/14/2024 03:50:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-88]:                                    expert_parallel_size=1),
+12/14/2024 03:50:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-88]:        model=ModelArgs(model_config=LlamaConfig(bos_token_id=0,
+12/14/2024 03:50:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-88]:                                                 eos_token_id=0,
+12/14/2024 03:50:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-88]:                                                 hidden_act='silu',
+12/14/2024 03:50:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-88]:                                                 hidden_size=3072,
+12/14/2024 03:50:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-88]:                                                 initializer_range=0.02,
+12/14/2024 03:50:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-88]:                                                 intermediate_size=8192,
+12/14/2024 03:50:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-88]:                                                 is_llama_config=True,
+12/14/2024 03:50:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-88]:                                                 max_position_embeddings=4096,
+12/14/2024 03:50:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-88]:                                                 num_attention_heads=32,
+12/14/2024 03:50:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-88]:                                                 num_hidden_layers=28,
+12/14/2024 03:50:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-88]:                                                 num_key_value_heads=32,
+12/14/2024 03:50:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-88]:                                                 pad_token_id=None,
+12/14/2024 03:50:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-88]:                                                 pretraining_tp=1,
+12/14/2024 03:50:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-88]:                                                 rms_norm_eps=1e-05,
+12/14/2024 03:50:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-88]:                                                 rope_scaling=None,
+12/14/2024 03:50:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-88]:                                                 rope_theta=10000.0,
+12/14/2024 03:50:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-88]:                                                 rope_interleaved=False,
+12/14/2024 03:50:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-88]:                                                 tie_word_embeddings=True,
+12/14/2024 03:50:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-88]:                                                 use_cache=True,
+12/14/2024 03:50:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-88]:                                                 vocab_size=131072),
+12/14/2024 03:50:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-88]:                        init_method=RandomInit(std=0.02),
+12/14/2024 03:50:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-88]:                        dtype=torch.bfloat16,
+12/14/2024 03:50:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-88]:                        make_vocab_size_divisible_by=1,
+12/14/2024 03:50:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-88]:                        ddp_bucket_cap_mb=25),
+12/14/2024 03:50:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-88]:        tokenizer=TokenizerArgs(tokenizer_name_or_path='robot-test/dummy-tokenizer-wordlevel',
+12/14/2024 03:50:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-88]:                                tokenizer_revision=None,
+12/14/2024 03:50:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-88]:                                tokenizer_max_length=None),
+12/14/2024 03:50:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-88]:        checkpoints=CheckpointsArgs(checkpoints_path=PosixPath('checkpoints'),
+12/14/2024 03:50:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-88]:                                    checkpoint_interval=10000,
+12/14/2024 03:50:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-88]:                                    save_initial_state=False,
+12/14/2024 03:50:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-88]:                                    save_final_state=False,
+12/14/2024 03:50:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-88]:                                    resume_checkpoint_path=None,
+12/14/2024 03:50:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-88]:                                    checkpoints_path_is_shared_file_system=False),
+12/14/2024 03:50:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-88]:        logging=LoggingArgs(log_level='info',
+12/14/2024 03:50:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-88]:                            log_level_replica='info',
+12/14/2024 03:50:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-88]:                            iteration_step_info_interval=1),
+12/14/2024 03:50:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-88]:        tokens=TokensArgs(sequence_length=4096,
+12/14/2024 03:50:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-88]:                          train_steps=100,
+12/14/2024 03:50:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-88]:                          micro_batch_size=256,
+12/14/2024 03:50:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-88]:                          batch_accumulation_per_replica=1,
+12/14/2024 03:50:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-88]:                          val_check_interval=100,
+12/14/2024 03:50:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-88]:                          limit_val_batches=0,
+12/14/2024 03:50:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-88]:                          limit_test_batches=0),
+12/14/2024 03:50:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-88]:        optimizer=OptimizerArgs(optimizer_factory=AdamWOptimizerArgs(adam_eps=1e-08,
+12/14/2024 03:50:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-88]:                                                                     adam_beta1=0.9,
+12/14/2024 03:50:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-88]:                                                                     adam_beta2=0.95,
+12/14/2024 03:50:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-88]:                                                                     torch_adam_is_fused=True,
+12/14/2024 03:50:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-88]:                                                                     name='adamW'),
+12/14/2024 03:50:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-88]:                                zero_stage=0,
+12/14/2024 03:50:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-88]:                                weight_decay=0.01,
+12/14/2024 03:50:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-88]:                                clip_grad=1.0,
+12/14/2024 03:50:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-88]:                                accumulate_grad_in_fp32=True,
+12/14/2024 03:50:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-88]:                                learning_rate_scheduler=LRSchedulerArgs(learning_rate=0.0003,
+12/14/2024 03:50:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-88]:                                                                        lr_warmup_steps=2,
+12/14/2024 03:50:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-88]:                                                                        lr_warmup_style='linear',
+12/14/2024 03:50:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-88]:                                                                        lr_decay_style='cosine',
+12/14/2024 03:50:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-88]:                                                                        lr_decay_steps=13,
+12/14/2024 03:50:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-88]:                                                                        lr_decay_starting_step=None,
+12/14/2024 03:50:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-88]:                                                                        min_decay_lr=1e-05)),
+12/14/2024 03:50:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-88]:        data_stages=[DatasetStageArgs(name='Stable Training Stage',
+12/14/2024 03:50:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-88]:                                      start_training_step=1,
+12/14/2024 03:50:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-88]:                                      data=DataArgs(dataset=None,
+12/14/2024 03:50:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-88]:                                                    seed=42,
+12/14/2024 03:50:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-88]:                                                    num_loading_workers=1))],
+12/14/2024 03:50:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-88]:        profiler=None,
+12/14/2024 03:50:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-88]:        lighteval=None,
+12/14/2024 03:50:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-88]:        s3_upload=None)
+12/14/2024 03:50:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-88]: Model Config:
+12/14/2024 03:50:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-88]: LlamaConfig(bos_token_id=0,
+12/14/2024 03:50:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-88]:             eos_token_id=0,
+12/14/2024 03:50:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-88]:             hidden_act='silu',
+12/14/2024 03:50:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-88]:             hidden_size=3072,
+12/14/2024 03:50:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-88]:             initializer_range=0.02,
+12/14/2024 03:50:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-88]:             intermediate_size=8192,
+12/14/2024 03:50:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-88]:             is_llama_config=True,
+12/14/2024 03:50:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-88]:             max_position_embeddings=4096,
+12/14/2024 03:50:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-88]:             num_attention_heads=32,
+12/14/2024 03:50:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-88]:             num_hidden_layers=28,
+12/14/2024 03:50:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-88]:             num_key_value_heads=32,
+12/14/2024 03:50:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-88]:             pad_token_id=None,
+12/14/2024 03:50:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-88]:             pretraining_tp=1,
+12/14/2024 03:50:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-88]:             rms_norm_eps=1e-05,
+12/14/2024 03:50:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-88]:             rope_scaling=None,
+12/14/2024 03:50:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-88]:             rope_theta=10000.0,
+12/14/2024 03:50:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-88]:             rope_interleaved=False,
+12/14/2024 03:50:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-88]:             tie_word_embeddings=True,
+12/14/2024 03:50:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-88]:             use_cache=True,
+12/14/2024 03:50:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-88]:             vocab_size=131072)
+12/14/2024 03:50:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-88]: Building model..
+12/14/2024 03:50:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-88]: Initialize RoPE Theta = 10000.0
+12/14/2024 03:50:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-88]: Setting PP block ranks...
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+12/14/2024 03:50:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-88]: Total number of parameters: 3.98G (7584.67MiB)
+12/14/2024 03:50:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-88]: Local number of parameters: 654M (1248.09MiB)
+12/14/2024 03:50:08 [INFO|DP=0|PP=2|TP=0|ip-26-0-171-88]: Local number of parameters: 453M (864.09MiB)
+12/14/2024 03:50:08 [INFO|DP=0|PP=1|TP=0|ip-26-0-171-88]: Local number of parameters: 453M (864.09MiB)
+12/14/2024 03:50:08 [INFO|DP=0|PP=0|TP=1|ip-26-0-171-88]: Local number of parameters: 654M (1248.09MiB)
+12/14/2024 03:50:08 [INFO|DP=0|PP=2|TP=1|ip-26-0-171-88]: Local number of parameters: 453M (864.09MiB)
+12/14/2024 03:50:08 [INFO|DP=0|PP=1|TP=1|ip-26-0-171-88]: Local number of parameters: 453M (864.09MiB)
+12/14/2024 03:50:08 [INFO|DP=0|PP=3|TP=0|ip-26-0-171-88]: Local number of parameters: 428M (816.05MiB)
+12/14/2024 03:50:08 [INFO|DP=0|PP=3|TP=1|ip-26-0-171-88]: Local number of parameters: 428M (816.05MiB)
+12/14/2024 03:50:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-88]: [After model building] Memory usage: 1248.11MiB. Peak allocated: 5632.00MiB Peak reserved: 12802.00MiB
+12/14/2024 03:50:08 [INFO|DP=0|PP=2|TP=0|ip-26-0-171-88]: [After model building] Memory usage: 864.11MiB. Peak allocated: 5632.00MiB Peak reserved: 11778.00MiB
+12/14/2024 03:50:08 [INFO|DP=0|PP=1|TP=0|ip-26-0-171-88]: [After model building] Memory usage: 864.11MiB. Peak allocated: 5632.00MiB Peak reserved: 12802.00MiB
+12/14/2024 03:50:08 [INFO|DP=0|PP=0|TP=1|ip-26-0-171-88]: [After model building] Memory usage: 1248.11MiB. Peak allocated: 5632.00MiB Peak reserved: 12802.00MiB
+12/14/2024 03:50:08 [INFO|DP=0|PP=1|TP=1|ip-26-0-171-88]: [After model building] Memory usage: 864.11MiB. Peak allocated: 5632.00MiB Peak reserved: 12802.00MiB
+12/14/2024 03:50:08 [INFO|DP=0|PP=2|TP=1|ip-26-0-171-88]: [After model building] Memory usage: 864.11MiB. Peak allocated: 5632.00MiB Peak reserved: 13826.00MiB
+12/14/2024 03:50:08 [INFO|DP=0|PP=3|TP=0|ip-26-0-171-88]: [After model building] Memory usage: 816.07MiB. Peak allocated: 5632.00MiB Peak reserved: 11778.00MiB
+12/14/2024 03:50:08 [INFO|DP=0|PP=3|TP=1|ip-26-0-171-88]: [After model building] Memory usage: 816.07MiB. Peak allocated: 5632.00MiB Peak reserved: 12802.00MiB
+12/14/2024 03:50:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-88]: No checkpoint path provided.
+12/14/2024 03:50:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-88]: Parametrizing model parameters using StandardParametrizator
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 252, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 461, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 498, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 243, in train_batch_iter
+    self.nb_microbatches >= pg.size() - 1
+AssertionError: Number of microbatches (1) must be at least PP_SIZE-1=3 when using the OneForwardOneBackwardPipelineEngine
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 252, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 461, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 498, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 243, in train_batch_iter
+    self.nb_microbatches >= pg.size() - 1
+AssertionError: Number of microbatches (1) must be at least PP_SIZE-1=3 when using the OneForwardOneBackwardPipelineEngine
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 252, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 461, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 498, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 243, in train_batch_iter
+    self.nb_microbatches >= pg.size() - 1
+AssertionError: Number of microbatches (1) must be at least PP_SIZE-1=3 when using the OneForwardOneBackwardPipelineEngine
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 252, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 461, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 498, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 243, in train_batch_iter
+    self.nb_microbatches >= pg.size() - 1
+AssertionError: Number of microbatches (1) must be at least PP_SIZE-1=3 when using the OneForwardOneBackwardPipelineEngine
+12/14/2024 03:50:10 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-88]: [Optimizer Building] Using LearningRateForSP as learning rate
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 252, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 461, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 498, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 243, in train_batch_iter
+    self.nb_microbatches >= pg.size() - 1
+AssertionError: Number of microbatches (1) must be at least PP_SIZE-1=3 when using the OneForwardOneBackwardPipelineEngine
+12/14/2024 03:50:10 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-88]: [Training Plan] Stage Stable Training Stage has 99 remaining training steps and has consumed 0 samples
+12/14/2024 03:50:10 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-88]: Using dummy data generator
+12/14/2024 03:50:10 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-88]: [Training Plan] There are 1 training stages 
+12/14/2024 03:50:10 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-88]: [Stage Stable Training Stage] start from step 1 
+12/14/2024 03:50:10 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-88]: 
+12/14/2024 03:50:10 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-88]: [Start training] datetime: 2024-12-14 03:50:10.711370 | mbs: 256 | grad_accum: 1 | global_batch_size: 256 | sequence_length: 4096 | train_steps: 100 | start_iteration_step: 0 | consumed_train_samples: 0
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 252, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 461, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 498, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 243, in train_batch_iter
+    self.nb_microbatches >= pg.size() - 1
+AssertionError: Number of microbatches (1) must be at least PP_SIZE-1=3 when using the OneForwardOneBackwardPipelineEngine
+12/14/2024 03:50:10 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-88]: Resuming training from stage Stable Training Stage, it has trained for 0 samples and has 99 remaining train steps
+12/14/2024 03:50:10 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-88]:  Memory usage: 6240.49MiB. Peak allocated 6240.49MiB. Peak reserved: 17798.00MiB
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 252, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 461, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 498, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 243, in train_batch_iter
+    self.nb_microbatches >= pg.size() - 1
+AssertionError: Number of microbatches (1) must be at least PP_SIZE-1=3 when using the OneForwardOneBackwardPipelineEngine
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 252, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 461, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 498, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 243, in train_batch_iter
+    self.nb_microbatches >= pg.size() - 1
+AssertionError: Number of microbatches (1) must be at least PP_SIZE-1=3 when using the OneForwardOneBackwardPipelineEngine
+[2024-12-14 03:50:23,505] torch.distributed.elastic.multiprocessing.api: [ERROR] failed (exitcode: 1) local_rank: 0 (pid: 97936) of binary: /fsx/nouamane/miniconda/envs/2-1-cu121/bin/python
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 33, in <module>
+    sys.exit(load_entry_point('torch==2.1.1', 'console_scripts', 'torchrun')())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 264, in launch_agent
+    raise ChildFailedError(
+torch.distributed.elastic.multiprocessing.errors.ChildFailedError: 
+============================================================
+/fsx/nouamane/projects/nanotron/run_train.py FAILED
+------------------------------------------------------------
+Failures:
+[1]:
+  time      : 2024-12-14_03:50:23
+  host      : ip-26-0-171-88.ec2.internal
+  rank      : 1 (local_rank: 1)
+  exitcode  : 1 (pid: 97937)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[2]:
+  time      : 2024-12-14_03:50:23
+  host      : ip-26-0-171-88.ec2.internal
+  rank      : 2 (local_rank: 2)
+  exitcode  : 1 (pid: 97938)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[3]:
+  time      : 2024-12-14_03:50:23
+  host      : ip-26-0-171-88.ec2.internal
+  rank      : 3 (local_rank: 3)
+  exitcode  : 1 (pid: 97939)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[4]:
+  time      : 2024-12-14_03:50:23
+  host      : ip-26-0-171-88.ec2.internal
+  rank      : 4 (local_rank: 4)
+  exitcode  : 1 (pid: 97940)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[5]:
+  time      : 2024-12-14_03:50:23
+  host      : ip-26-0-171-88.ec2.internal
+  rank      : 5 (local_rank: 5)
+  exitcode  : 1 (pid: 97941)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[6]:
+  time      : 2024-12-14_03:50:23
+  host      : ip-26-0-171-88.ec2.internal
+  rank      : 6 (local_rank: 6)
+  exitcode  : 1 (pid: 97942)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[7]:
+  time      : 2024-12-14_03:50:23
+  host      : ip-26-0-171-88.ec2.internal
+  rank      : 7 (local_rank: 7)
+  exitcode  : 1 (pid: 97943)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+------------------------------------------------------------
+Root Cause (first observed failure):
+[0]:
+  time      : 2024-12-14_03:50:23
+  host      : ip-26-0-171-88.ec2.internal
+  rank      : 0 (local_rank: 0)
+  exitcode  : 1 (pid: 97936)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+============================================================
+srun: error: ip-26-0-171-88: task 0: Exited with exit code 1
+srun: launch/slurm: _step_signal: Terminating StepId=13442265.0
diff --git a/logs/13442739-bench_3.57G_dp1_tp2_pp8_acc2_mbs128_seq4096_zero0_tpmodeRED_vocab131k.out b/logs/13442739-bench_3.57G_dp1_tp2_pp8_acc2_mbs128_seq4096_zero0_tpmodeRED_vocab131k.out
new file mode 100644
index 0000000000000000000000000000000000000000..96e5eab4e1d2a84b5761478ea740212116c09713
--- /dev/null
+++ b/logs/13442739-bench_3.57G_dp1_tp2_pp8_acc2_mbs128_seq4096_zero0_tpmodeRED_vocab131k.out
@@ -0,0 +1,876 @@
++ source /etc/profile.d/modules.sh
+++ . /usr/share/modules/init/bash
++++ unset _mlshdbg
++++ '[' 0 = 1 ']'
++++ unset _mlre _mlIFS
++++ '[' -n x ']'
++++ _mlIFS=' 	
+'
++++ IFS=' '
++++ '[' -n '' ']'
+++++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash autoinit
++++ _mlcode='module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+test 0;'
++++ _mlret=0
++++ '[' -n x ']'
++++ IFS=' 	
+'
++++ unset _mlIFS
++++ unset _mlre _mlv _mlrv
++++ '[' 0 -eq 0 ']'
++++ eval 'module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+test 0;'
+++++ MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl
+++++ export MODULES_CMD
+++++ ENV=/usr/share/modules/init/profile.sh
+++++ export ENV
+++++ BASH_ENV=/usr/share/modules/init/bash
+++++ export BASH_ENV
+++++ MODULESHOME=/usr/share/modules
+++++ export MODULESHOME
+++++ test 0
++++ '[' 0 = 1 ']'
++++ '[' -t 2 ']'
++++ export -f module
++++ export -f switchml
++++ '[' 5 -ge 3 ']'
++++ [[ ehxB =~ i ]]
++++ [[ ! :/admin/home/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin: =~ :/usr/bin: ]]
++++ '[' '!' -n x ']'
+++++ manpath
++++ [[ ! :/admin/home/nouamane/miniconda/envs/2-1-cu121/man:/admin/home/nouamane/miniconda/envs/2-1-cu121/share/man:/admin/home/nouamane/.local/share/man:/fsx/nouamane/miniconda/envs/2-1-cu121/man:/fsx/nouamane/miniconda/envs/2-1-cu121/share/man:/fsx/nouamane/miniconda/man:/fsx/nouamane/miniconda/share/man:/opt/amazon/openmpi/share/man:/opt/amazon/efa/share/man:/opt/slurm/share/man:/usr/local/man:/usr/local/share/man:/usr/share/man:/admin/home/nouamane/.fzf/man:: =~ :/usr/share/man: ]]
++++ unset _mlcode _mlret
++++ '[' -n '' ']'
++ module load cuda/12.1
++ unset _mlshdbg
++ '[' 0 = 1 ']'
++ unset _mlre _mlIFS
++ '[' -n x ']'
++ _mlIFS=' 	
+'
++ IFS=' '
++ '[' -n '' ']'
+++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash load cuda/12.1
++ eval 'CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include;' export 'CPATH;
+LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/opt/aws-ofi-nccl/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/targets/x86_64-linux/lib/:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/lib:/usr/lib::/opt/amazon/openmpi/lib:/opt/amazon/efa/lib;' export 'LD_LIBRARY_PATH;
+FI_EFA_FORK_SAFE=1;' export 'FI_EFA_FORK_SAFE;
+MANPATH=/usr/local/cuda-12.1/share/man::;' export 'MANPATH;
+LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64;' export 'LIBRARY_PATH;
+_LMFILES_=/usr/share/modules/modulefiles/cuda/12.1;' export '_LMFILES_;
+LOADEDMODULES=cuda/12.1;' export 'LOADEDMODULES;
+MPI_PATH=/opt/amazon/openmpi;' export 'MPI_PATH;
+NCCL_HOME_modshare=/opt/nccl/build:1;' export 'NCCL_HOME_modshare;
+NCCL_PROTO=simple;' export 'NCCL_PROTO;
+MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1;' export 'MANPATH_modshare;
+LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1;' export 'LIBRARY_PATH_modshare;
+NCCL_SOCKET_IFNAME=enp;' export 'NCCL_SOCKET_IFNAME;
+AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl;' export 'AWS_OFI_NCCL_HOME;
+NCCL_HOME=/opt/nccl/build;' export 'NCCL_HOME;
+FI_PROVIDER=efa;' export 'FI_PROVIDER;
+AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1;' export 'AWS_OFI_NCCL_HOME_modshare;
+CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1;' export 'CPATH_modshare;
+LD_LIBRARY_PATH_modshare=:1:/opt/amazon/efa/lib:3:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib/:1:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2:/opt/aws-ofi-nccl/lib:2:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/efa/lib:1:/usr/lib:1;' export 'LD_LIBRARY_PATH_modshare;
+FI_EFA_ENABLE_SHM_TRANSFER=1;' export 'FI_EFA_ENABLE_SHM_TRANSFER;
+_LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1;' export '_LMFILES__modshare;
+LOADEDMODULES_modshare=cuda/12.1:1;' export 'LOADEDMODULES_modshare;
+MPI_PATH_modshare=/opt/amazon/openmpi:1;' export 'MPI_PATH_modshare;
+PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin;' export 'PATH;
+CUDA_HOME=/usr/local/cuda-12.1;' export 'CUDA_HOME;
+PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/usr/local/cuda-12.1/include:1:/opt/amazon/efa/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/usr/sbin:1:/opt/slurm/bin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:2:/opt/amazon/openmpi/bin:1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:1:/admin/home/nouamane/.fzf/bin:1:/admin/home/nouamane/miniconda/envs/2-1-cu121/bin:1:/usr/local/games:1;' export 'PATH_modshare;
+test' '0;'
+++ CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include
+++ export CPATH
+++ LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/opt/aws-ofi-nccl/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/targets/x86_64-linux/lib/:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/lib:/usr/lib::/opt/amazon/openmpi/lib:/opt/amazon/efa/lib
+++ export LD_LIBRARY_PATH
+++ FI_EFA_FORK_SAFE=1
+++ export FI_EFA_FORK_SAFE
+++ MANPATH=/usr/local/cuda-12.1/share/man::
+++ export MANPATH
+++ LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64
+++ export LIBRARY_PATH
+++ _LMFILES_=/usr/share/modules/modulefiles/cuda/12.1
+++ export _LMFILES_
+++ LOADEDMODULES=cuda/12.1
+++ export LOADEDMODULES
+++ MPI_PATH=/opt/amazon/openmpi
+++ export MPI_PATH
+++ NCCL_HOME_modshare=/opt/nccl/build:1
+++ export NCCL_HOME_modshare
+++ NCCL_PROTO=simple
+++ export NCCL_PROTO
+++ MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1
+++ export MANPATH_modshare
+++ LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1
+++ export LIBRARY_PATH_modshare
+++ NCCL_SOCKET_IFNAME=enp
+++ export NCCL_SOCKET_IFNAME
+++ AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl
+++ export AWS_OFI_NCCL_HOME
+++ NCCL_HOME=/opt/nccl/build
+++ export NCCL_HOME
+++ FI_PROVIDER=efa
+++ export FI_PROVIDER
+++ AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1
+++ export AWS_OFI_NCCL_HOME_modshare
+++ CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1
+++ export CPATH_modshare
+++ LD_LIBRARY_PATH_modshare=:1:/opt/amazon/efa/lib:3:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib/:1:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2:/opt/aws-ofi-nccl/lib:2:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/efa/lib:1:/usr/lib:1
+++ export LD_LIBRARY_PATH_modshare
+++ FI_EFA_ENABLE_SHM_TRANSFER=1
+++ export FI_EFA_ENABLE_SHM_TRANSFER
+++ _LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1
+++ export _LMFILES__modshare
+++ LOADEDMODULES_modshare=cuda/12.1:1
+++ export LOADEDMODULES_modshare
+++ MPI_PATH_modshare=/opt/amazon/openmpi:1
+++ export MPI_PATH_modshare
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin
+++ export PATH
+++ CUDA_HOME=/usr/local/cuda-12.1
+++ export CUDA_HOME
+++ PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/usr/local/cuda-12.1/include:1:/opt/amazon/efa/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/usr/sbin:1:/opt/slurm/bin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:2:/opt/amazon/openmpi/bin:1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:1:/admin/home/nouamane/.fzf/bin:1:/admin/home/nouamane/miniconda/envs/2-1-cu121/bin:1:/usr/local/games:1
+++ export PATH_modshare
+++ test 0
++ _mlstatus=0
++ '[' -n x ']'
++ IFS=' 	
+'
++ unset _mlre _mlv _mlrv _mlIFS
++ '[' -n '' ']'
++ unset _mlshdbg
++ return 0
++ source /fsx/nouamane/miniconda/bin/activate
+++ _CONDA_ROOT=/fsx/nouamane/miniconda
+++ . /fsx/nouamane/miniconda/etc/profile.d/conda.sh
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ '[' -z x ']'
+++ conda activate
+++ local cmd=activate
+++ case "$cmd" in
+++ __conda_activate activate
+++ '[' -n '' ']'
+++ local ask_conda
++++ PS1=
++++ __conda_exe shell.posix activate
++++ /fsx/nouamane/miniconda/bin/conda shell.posix activate
+++ ask_conda='. "/admin/home/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/fsx/nouamane/miniconda/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_2='\''/admin/home/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\'''
+++ eval '. "/admin/home/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/fsx/nouamane/miniconda/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_2='\''/admin/home/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\'''
++++ . /admin/home/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh
+++++ test -n ''
+++++ unset XML_CATALOG_FILES
+++++ unset xml_catalog_files_libxml2
++++ PS1='(base) '
++++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/fsx/nouamane/miniconda/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin
++++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/fsx/nouamane/miniconda/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin
++++ export CONDA_PREFIX=/fsx/nouamane/miniconda
++++ CONDA_PREFIX=/fsx/nouamane/miniconda
++++ export CONDA_SHLVL=3
++++ CONDA_SHLVL=3
++++ export CONDA_DEFAULT_ENV=base
++++ CONDA_DEFAULT_ENV=base
++++ export 'CONDA_PROMPT_MODIFIER=(base) '
++++ CONDA_PROMPT_MODIFIER='(base) '
++++ export CONDA_PREFIX_2=/admin/home/nouamane/miniconda/envs/2-1-cu121
++++ CONDA_PREFIX_2=/admin/home/nouamane/miniconda/envs/2-1-cu121
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ __conda_hashr
+++ '[' -n '' ']'
+++ '[' -n '' ']'
+++ hash -r
++ conda activate 2-1-cu121
++ local cmd=activate
++ case "$cmd" in
++ __conda_activate activate 2-1-cu121
++ '[' -n '' ']'
++ local ask_conda
+++ PS1='(base) '
+++ __conda_exe shell.posix activate 2-1-cu121
+++ /fsx/nouamane/miniconda/bin/conda shell.posix activate 2-1-cu121
++ ask_conda='PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''4'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_3='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
++ eval 'PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''4'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_3='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
+++ PS1='(2-1-cu121) '
+++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin
+++ export CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ export CONDA_SHLVL=4
+++ CONDA_SHLVL=4
+++ export CONDA_DEFAULT_ENV=2-1-cu121
+++ CONDA_DEFAULT_ENV=2-1-cu121
+++ export 'CONDA_PROMPT_MODIFIER=(2-1-cu121) '
+++ CONDA_PROMPT_MODIFIER='(2-1-cu121) '
+++ export CONDA_PREFIX_3=/fsx/nouamane/miniconda
+++ CONDA_PREFIX_3=/fsx/nouamane/miniconda
+++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ export _CE_M=
+++ _CE_M=
+++ export _CE_CONDA=
+++ _CE_CONDA=
+++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh
++++ test -n ''
++++ xml_catalog_files_libxml2=
++++ XML_CATALOG_FILES=
++++ conda_catalog_files=
++++ ifs_libxml2=' 	
+'
++++ IFS=' '
++++ rem=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ for pre in ${rem}
++++ test '' = /fsx/nouamane/miniconda/envs/2-1-cu121
++++ conda_catalog_files=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ rem=
++++ IFS=' 	
+'
++++ conda_catalog_files='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ export 'XML_CATALOG_FILES=file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ XML_CATALOG_FILES='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ unset conda_catalog_files ifs_libxml2 rem
++ __conda_hashr
++ '[' -n '' ']'
++ '[' -n '' ']'
++ hash -r
++ export PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin
++ PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin
+++ scontrol show hostnames 'ip-26-0-165-[59,131]'
++ export 'NODELIST=ip-26-0-165-59
+ip-26-0-165-131'
++ NODELIST='ip-26-0-165-59
+ip-26-0-165-131'
+++ head -n1
+++ scontrol show hostnames 'ip-26-0-165-[59,131]'
++ export MASTER_NODE=ip-26-0-165-59
++ MASTER_NODE=ip-26-0-165-59
++ export MASTER_PORT=12356
++ MASTER_PORT=12356
++ export NNODES=2
++ NNODES=2
++ export GPUS_PER_NODE=8
++ GPUS_PER_NODE=8
++ export WORLD_SIZE=16
++ WORLD_SIZE=16
++ export CUDA_DEVICE_MAX_CONNECTIONS=1
++ CUDA_DEVICE_MAX_CONNECTIONS=1
++ export NCCL_DEBUG=WARN
++ NCCL_DEBUG=WARN
++ export NANOTRON_BENCHMARK=1
++ NANOTRON_BENCHMARK=1
++ export WANDB_MODE=disabled
++ WANDB_MODE=disabled
++ export TORCH_NCCL_ASYNC_ERROR_HANDLING=1
++ TORCH_NCCL_ASYNC_ERROR_HANDLING=1
++ echo '=== GPU Topology ==='
+=== GPU Topology ===
++ nvidia-smi topo -m
+	GPU0	GPU1	GPU2	GPU3	GPU4	GPU5	GPU6	GPU7	CPU Affinity	NUMA Affinity	GPU NUMA ID
+GPU0	 X 	NV18	NV18	NV18	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU1	NV18	 X 	NV18	NV18	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU2	NV18	NV18	 X 	NV18	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU3	NV18	NV18	NV18	 X 	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU4	NV18	NV18	NV18	NV18	 X 	NV18	NV18	NV18	48-95	1		N/A
+GPU5	NV18	NV18	NV18	NV18	NV18	 X 	NV18	NV18	48-95	1		N/A
+GPU6	NV18	NV18	NV18	NV18	NV18	NV18	 X 	NV18	48-95	1		N/A
+GPU7	NV18	NV18	NV18	NV18	NV18	NV18	NV18	 X 	48-95	1		N/A
+
+Legend:
+
+  X    = Self
+  SYS  = Connection traversing PCIe as well as the SMP interconnect between NUMA nodes (e.g., QPI/UPI)
+  NODE = Connection traversing PCIe as well as the interconnect between PCIe Host Bridges within a NUMA node
+  PHB  = Connection traversing PCIe as well as a PCIe Host Bridge (typically the CPU)
+  PXB  = Connection traversing multiple PCIe bridges (without traversing the PCIe Host Bridge)
+  PIX  = Connection traversing at most a single PCIe bridge
+  NV#  = Connection traversing a bonded set of # NVLinks
++ echo ==================
+==================
++ echo 'Master node: ip-26-0-165-59'
+Master node: ip-26-0-165-59
++ echo 'All nodes: ip-26-0-165-59
+ip-26-0-165-131'
+All nodes: ip-26-0-165-59
+ip-26-0-165-131
++ echo 'World size: 16'
+World size: 16
++ srun --wait=0 --kill-on-bad-exit=1 torchrun --nnodes=2 --nproc_per_node=8 --rdzv_id=13442739 --rdzv_backend=c10d --rdzv_endpoint=ip-26-0-165-59:12356 --max_restarts 0 --rdzv_conf timeout=60 /fsx/nouamane/projects/nanotron/run_train.py --config-file benchmark/configs/config_3.57G_dp1_tp2_pp8_acc2_mbs128_seq4096_zero0_tpmodeRED_vocab131k.yaml
+[2024-12-14 08:29:04,333] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-14 08:29:04,333] torch.distributed.run: [WARNING] 
+[2024-12-14 08:29:04,333] torch.distributed.run: [WARNING] *****************************************
+[2024-12-14 08:29:04,333] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-14 08:29:04,333] torch.distributed.run: [WARNING] *****************************************
+[2024-12-14 08:29:04,415] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-14 08:29:04,415] torch.distributed.run: [WARNING] 
+[2024-12-14 08:29:04,415] torch.distributed.run: [WARNING] *****************************************
+[2024-12-14 08:29:04,415] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-14 08:29:04,415] torch.distributed.run: [WARNING] *****************************************
+NCCL version 2.18.6+cuda12.1
+12/14/2024 08:29:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]: Measuring inter-GPU and intra-node bandwidth...
+NCCL version 2.18.6+cuda12.1
+12/14/2024 08:29:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]: Bandwidth measurement complete. Time taken: 15.82 seconds
+12/14/2024 08:29:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]: Config:
+12/14/2024 08:29:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]: Config(general=GeneralArgs(project='debug',
+12/14/2024 08:29:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                            run='3.57G_dp1_tp2_pp8_acc2_mbs128_seq4096_zero0_tpmodeRED_vocab131k',
+12/14/2024 08:29:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                            seed=42,
+12/14/2024 08:29:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                            step=None,
+12/14/2024 08:29:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                            consumed_train_samples=None,
+12/14/2024 08:29:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                            benchmark_csv_path=PosixPath('benchmark/results/bench_final2.csv'),
+12/14/2024 08:29:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                            ignore_sanity_checks=True),
+12/14/2024 08:29:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:        parallelism=ParallelismArgs(dp=1,
+12/14/2024 08:29:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                                    pp=8,
+12/14/2024 08:29:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                                    tp=2,
+12/14/2024 08:29:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                                    pp_engine=<nanotron.parallel.pipeline_parallel.engine.OneForwardOneBackwardPipelineEngine object at 0x7f923a5852a0>,
+12/14/2024 08:29:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                                    tp_mode=<TensorParallelLinearMode.REDUCE_SCATTER: 2>,
+12/14/2024 08:29:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                                    tp_linear_async_communication=True,
+12/14/2024 08:29:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                                    recompute_layer=False,
+12/14/2024 08:29:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                                    tp_recompute_allgather=True,
+12/14/2024 08:29:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                                    expert_parallel_size=1),
+12/14/2024 08:29:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:        model=ModelArgs(model_config=LlamaConfig(bos_token_id=0,
+12/14/2024 08:29:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                                                 eos_token_id=0,
+12/14/2024 08:29:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                                                 hidden_act='silu',
+12/14/2024 08:29:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                                                 hidden_size=3072,
+12/14/2024 08:29:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                                                 initializer_range=0.02,
+12/14/2024 08:29:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                                                 intermediate_size=8192,
+12/14/2024 08:29:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                                                 is_llama_config=True,
+12/14/2024 08:29:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                                                 max_position_embeddings=4096,
+12/14/2024 08:29:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                                                 num_attention_heads=32,
+12/14/2024 08:29:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                                                 num_hidden_layers=28,
+12/14/2024 08:29:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                                                 num_key_value_heads=32,
+12/14/2024 08:29:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                                                 pad_token_id=None,
+12/14/2024 08:29:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                                                 pretraining_tp=1,
+12/14/2024 08:29:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                                                 rms_norm_eps=1e-05,
+12/14/2024 08:29:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                                                 rope_scaling=None,
+12/14/2024 08:29:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                                                 rope_theta=10000.0,
+12/14/2024 08:29:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                                                 rope_interleaved=False,
+12/14/2024 08:29:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                                                 tie_word_embeddings=True,
+12/14/2024 08:29:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                                                 use_cache=True,
+12/14/2024 08:29:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                                                 vocab_size=131072),
+12/14/2024 08:29:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                        init_method=RandomInit(std=0.02),
+12/14/2024 08:29:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                        dtype=torch.bfloat16,
+12/14/2024 08:29:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                        make_vocab_size_divisible_by=1,
+12/14/2024 08:29:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                        ddp_bucket_cap_mb=25),
+12/14/2024 08:29:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:        tokenizer=TokenizerArgs(tokenizer_name_or_path='robot-test/dummy-tokenizer-wordlevel',
+12/14/2024 08:29:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                                tokenizer_revision=None,
+12/14/2024 08:29:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                                tokenizer_max_length=None),
+12/14/2024 08:29:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:        checkpoints=CheckpointsArgs(checkpoints_path=PosixPath('checkpoints'),
+12/14/2024 08:29:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                                    checkpoint_interval=10000,
+12/14/2024 08:29:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                                    save_initial_state=False,
+12/14/2024 08:29:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                                    save_final_state=False,
+12/14/2024 08:29:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                                    resume_checkpoint_path=None,
+12/14/2024 08:29:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                                    checkpoints_path_is_shared_file_system=False),
+12/14/2024 08:29:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:        logging=LoggingArgs(log_level='info',
+12/14/2024 08:29:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                            log_level_replica='info',
+12/14/2024 08:29:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                            iteration_step_info_interval=1),
+12/14/2024 08:29:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:        tokens=TokensArgs(sequence_length=4096,
+12/14/2024 08:29:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                          train_steps=100,
+12/14/2024 08:29:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                          micro_batch_size=128,
+12/14/2024 08:29:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                          batch_accumulation_per_replica=2,
+12/14/2024 08:29:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                          val_check_interval=100,
+12/14/2024 08:29:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                          limit_val_batches=0,
+12/14/2024 08:29:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                          limit_test_batches=0),
+12/14/2024 08:29:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:        optimizer=OptimizerArgs(optimizer_factory=AdamWOptimizerArgs(adam_eps=1e-08,
+12/14/2024 08:29:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                                                                     adam_beta1=0.9,
+12/14/2024 08:29:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                                                                     adam_beta2=0.95,
+12/14/2024 08:29:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                                                                     torch_adam_is_fused=True,
+12/14/2024 08:29:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                                                                     name='adamW'),
+12/14/2024 08:29:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                                zero_stage=0,
+12/14/2024 08:29:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                                weight_decay=0.01,
+12/14/2024 08:29:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                                clip_grad=1.0,
+12/14/2024 08:29:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                                accumulate_grad_in_fp32=True,
+12/14/2024 08:29:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                                learning_rate_scheduler=LRSchedulerArgs(learning_rate=0.0003,
+12/14/2024 08:29:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                                                                        lr_warmup_steps=2,
+12/14/2024 08:29:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                                                                        lr_warmup_style='linear',
+12/14/2024 08:29:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                                                                        lr_decay_style='cosine',
+12/14/2024 08:29:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                                                                        lr_decay_steps=13,
+12/14/2024 08:29:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                                                                        lr_decay_starting_step=None,
+12/14/2024 08:29:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                                                                        min_decay_lr=1e-05)),
+12/14/2024 08:29:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:        data_stages=[DatasetStageArgs(name='Stable Training Stage',
+12/14/2024 08:29:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                                      start_training_step=1,
+12/14/2024 08:29:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                                      data=DataArgs(dataset=None,
+12/14/2024 08:29:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                                                    seed=42,
+12/14/2024 08:29:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                                                    num_loading_workers=1))],
+12/14/2024 08:29:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:        profiler=None,
+12/14/2024 08:29:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:        lighteval=None,
+12/14/2024 08:29:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:        s3_upload=None)
+12/14/2024 08:29:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]: Model Config:
+12/14/2024 08:29:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]: LlamaConfig(bos_token_id=0,
+12/14/2024 08:29:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:             eos_token_id=0,
+12/14/2024 08:29:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:             hidden_act='silu',
+12/14/2024 08:29:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:             hidden_size=3072,
+12/14/2024 08:29:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:             initializer_range=0.02,
+12/14/2024 08:29:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:             intermediate_size=8192,
+12/14/2024 08:29:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:             is_llama_config=True,
+12/14/2024 08:29:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:             max_position_embeddings=4096,
+12/14/2024 08:29:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:             num_attention_heads=32,
+12/14/2024 08:29:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:             num_hidden_layers=28,
+12/14/2024 08:29:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:             num_key_value_heads=32,
+12/14/2024 08:29:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:             pad_token_id=None,
+12/14/2024 08:29:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:             pretraining_tp=1,
+12/14/2024 08:29:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:             rms_norm_eps=1e-05,
+12/14/2024 08:29:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:             rope_scaling=None,
+12/14/2024 08:29:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:             rope_theta=10000.0,
+12/14/2024 08:29:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:             rope_interleaved=False,
+12/14/2024 08:29:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:             tie_word_embeddings=True,
+12/14/2024 08:29:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:             use_cache=True,
+12/14/2024 08:29:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:             vocab_size=131072)
+12/14/2024 08:29:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]: Building model..
+12/14/2024 08:29:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]: Initialize RoPE Theta = 10000.0
+12/14/2024 08:29:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]: Setting PP block ranks...
+NCCL version 2.18.6+cuda12.1
+NCCL version 2.18.6+cuda12.1
+NCCL version 2.18.6+cuda12.1
+NCCL version 2.18.6+cuda12.1
+NCCL version 2.18.6+cuda12.1
+NCCL version 2.18.6+cuda12.1
+NCCL version 2.18.6+cuda12.1
+12/14/2024 08:29:51 [INFO|DP=0|PP=2|TP=0|ip-26-0-165-131]: Local number of parameters: 227M (432.05MiB)
+12/14/2024 08:29:51 [INFO|DP=0|PP=2|TP=1|ip-26-0-165-131]: Local number of parameters: 227M (432.05MiB)
+12/14/2024 08:29:51 [INFO|DP=0|PP=7|TP=0|ip-26-0-165-59]: Local number of parameters: 201M (384.01MiB)
+12/14/2024 08:29:51 [INFO|DP=0|PP=7|TP=1|ip-26-0-165-59]: Local number of parameters: 201M (384.01MiB)
+12/14/2024 08:29:51 [INFO|DP=0|PP=6|TP=0|ip-26-0-165-59]: Local number of parameters: 227M (432.05MiB)
+12/14/2024 08:29:51 [INFO|DP=0|PP=5|TP=0|ip-26-0-165-59]: Local number of parameters: 227M (432.05MiB)
+12/14/2024 08:29:51 [INFO|DP=0|PP=4|TP=0|ip-26-0-165-59]: Local number of parameters: 227M (432.05MiB)
+12/14/2024 08:29:51 [INFO|DP=0|PP=6|TP=1|ip-26-0-165-59]: Local number of parameters: 227M (432.05MiB)
+12/14/2024 08:29:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]: Total number of parameters: 3.98G (7584.67MiB)
+12/14/2024 08:29:51 [INFO|DP=0|PP=3|TP=1|ip-26-0-165-131]: Local number of parameters: 227M (432.05MiB)
+12/14/2024 08:29:51 [INFO|DP=0|PP=3|TP=0|ip-26-0-165-131]: Local number of parameters: 227M (432.05MiB)
+12/14/2024 08:29:51 [INFO|DP=0|PP=1|TP=0|ip-26-0-165-131]: Local number of parameters: 227M (432.05MiB)
+12/14/2024 08:29:51 [INFO|DP=0|PP=0|TP=1|ip-26-0-165-131]: Local number of parameters: 428M (816.05MiB)
+12/14/2024 08:29:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]: Local number of parameters: 428M (816.05MiB)
+12/14/2024 08:29:51 [INFO|DP=0|PP=1|TP=1|ip-26-0-165-131]: Local number of parameters: 227M (432.05MiB)
+12/14/2024 08:29:51 [INFO|DP=0|PP=2|TP=0|ip-26-0-165-131]: [After model building] Memory usage: 432.06MiB. Peak allocated: 5504.00MiB Peak reserved: 22914.00MiB
+12/14/2024 08:29:51 [INFO|DP=0|PP=5|TP=1|ip-26-0-165-59]: Local number of parameters: 227M (432.05MiB)
+12/14/2024 08:29:51 [INFO|DP=0|PP=4|TP=1|ip-26-0-165-59]: Local number of parameters: 227M (432.05MiB)
+12/14/2024 08:29:51 [INFO|DP=0|PP=7|TP=0|ip-26-0-165-59]: [After model building] Memory usage: 384.02MiB. Peak allocated: 5504.00MiB Peak reserved: 17794.00MiB
+12/14/2024 08:29:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]: [After model building] Memory usage: 816.06MiB. Peak allocated: 5504.00MiB Peak reserved: 22914.00MiB
+12/14/2024 08:29:51 [INFO|DP=0|PP=7|TP=1|ip-26-0-165-59]: [After model building] Memory usage: 384.02MiB. Peak allocated: 5504.00MiB Peak reserved: 17794.00MiB
+12/14/2024 08:29:51 [INFO|DP=0|PP=6|TP=0|ip-26-0-165-59]: [After model building] Memory usage: 432.06MiB. Peak allocated: 5504.00MiB Peak reserved: 17794.00MiB
+12/14/2024 08:29:51 [INFO|DP=0|PP=2|TP=1|ip-26-0-165-131]: [After model building] Memory usage: 432.06MiB. Peak allocated: 5504.00MiB Peak reserved: 22914.00MiB
+12/14/2024 08:29:51 [INFO|DP=0|PP=1|TP=0|ip-26-0-165-131]: [After model building] Memory usage: 432.06MiB. Peak allocated: 5504.00MiB Peak reserved: 22914.00MiB
+12/14/2024 08:29:51 [INFO|DP=0|PP=4|TP=0|ip-26-0-165-59]: [After model building] Memory usage: 432.06MiB. Peak allocated: 5504.00MiB Peak reserved: 19842.00MiB
+12/14/2024 08:29:51 [INFO|DP=0|PP=3|TP=0|ip-26-0-165-131]: [After model building] Memory usage: 432.06MiB. Peak allocated: 5504.00MiB Peak reserved: 22914.00MiB
+12/14/2024 08:29:51 [INFO|DP=0|PP=3|TP=1|ip-26-0-165-131]: [After model building] Memory usage: 432.06MiB. Peak allocated: 5504.00MiB Peak reserved: 22914.00MiB
+12/14/2024 08:29:51 [INFO|DP=0|PP=5|TP=0|ip-26-0-165-59]: [After model building] Memory usage: 432.06MiB. Peak allocated: 5504.00MiB Peak reserved: 18818.00MiB
+12/14/2024 08:29:51 [INFO|DP=0|PP=0|TP=1|ip-26-0-165-131]: [After model building] Memory usage: 816.06MiB. Peak allocated: 5504.00MiB Peak reserved: 22914.00MiB
+12/14/2024 08:29:51 [INFO|DP=0|PP=6|TP=1|ip-26-0-165-59]: [After model building] Memory usage: 432.06MiB. Peak allocated: 5504.00MiB Peak reserved: 18818.00MiB
+12/14/2024 08:29:51 [INFO|DP=0|PP=1|TP=1|ip-26-0-165-131]: [After model building] Memory usage: 432.06MiB. Peak allocated: 5504.00MiB Peak reserved: 22914.00MiB
+12/14/2024 08:29:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]: No checkpoint path provided.
+12/14/2024 08:29:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]: Parametrizing model parameters using StandardParametrizator
+12/14/2024 08:29:51 [INFO|DP=0|PP=4|TP=1|ip-26-0-165-59]: [After model building] Memory usage: 432.06MiB. Peak allocated: 5504.00MiB Peak reserved: 19842.00MiB
+12/14/2024 08:29:51 [INFO|DP=0|PP=5|TP=1|ip-26-0-165-59]: [After model building] Memory usage: 432.06MiB. Peak allocated: 5504.00MiB Peak reserved: 17794.00MiB
+NCCL version 2.18.6+cuda12.1
+NCCL version 2.18.6+cuda12.1
+NCCL version 2.18.6+cuda12.1
+NCCL version 2.18.6+cuda12.1
+NCCL version 2.18.6+cuda12.1
+NCCL version 2.18.6+cuda12.1
+NCCL version 2.18.6+cuda12.1
+12/14/2024 08:29:52 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]: [Optimizer Building] Using LearningRateForSP as learning rate
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 252, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 461, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 498, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 243, in train_batch_iter
+    self.nb_microbatches >= pg.size() - 1
+AssertionError: Number of microbatches (2) must be at least PP_SIZE-1=7 when using the OneForwardOneBackwardPipelineEngine
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 252, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 252, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 461, in train
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 461, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 498, in training_step
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 498, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 243, in train_batch_iter
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 243, in train_batch_iter
+    self.nb_microbatches >= pg.size() - 1
+AssertionError: Number of microbatches (2) must be at least PP_SIZE-1=7 when using the OneForwardOneBackwardPipelineEngine
+    self.nb_microbatches >= pg.size() - 1
+AssertionError: Number of microbatches (2) must be at least PP_SIZE-1=7 when using the OneForwardOneBackwardPipelineEngine
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 252, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 461, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 498, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 243, in train_batch_iter
+    self.nb_microbatches >= pg.size() - 1
+AssertionError: Number of microbatches (2) must be at least PP_SIZE-1=7 when using the OneForwardOneBackwardPipelineEngine
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 252, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 252, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 461, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 461, in train
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 498, in training_step
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 252, in <module>
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 243, in train_batch_iter
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 498, in training_step
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 461, in train
+    self.nb_microbatches >= pg.size() - 1
+AssertionError: Number of microbatches (2) must be at least PP_SIZE-1=7 when using the OneForwardOneBackwardPipelineEngine
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 243, in train_batch_iter
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 498, in training_step
+    self.nb_microbatches >= pg.size() - 1
+AssertionError: Number of microbatches (2) must be at least PP_SIZE-1=7 when using the OneForwardOneBackwardPipelineEngine
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 243, in train_batch_iter
+    self.nb_microbatches >= pg.size() - 1
+AssertionError: Number of microbatches (2) must be at least PP_SIZE-1=7 when using the OneForwardOneBackwardPipelineEngine
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 252, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 461, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 498, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 243, in train_batch_iter
+    self.nb_microbatches >= pg.size() - 1
+AssertionError: Number of microbatches (2) must be at least PP_SIZE-1=7 when using the OneForwardOneBackwardPipelineEngine
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 252, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 461, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 498, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 243, in train_batch_iter
+    self.nb_microbatches >= pg.size() - 1
+AssertionError: Number of microbatches (2) must be at least PP_SIZE-1=7 when using the OneForwardOneBackwardPipelineEngine
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 252, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 461, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 498, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 243, in train_batch_iter
+    self.nb_microbatches >= pg.size() - 1
+AssertionError: Number of microbatches (2) must be at least PP_SIZE-1=7 when using the OneForwardOneBackwardPipelineEngine
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 252, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 461, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 498, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 243, in train_batch_iter
+    self.nb_microbatches >= pg.size() - 1
+AssertionError: Number of microbatches (2) must be at least PP_SIZE-1=7 when using the OneForwardOneBackwardPipelineEngine
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 252, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 461, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 498, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 243, in train_batch_iter
+    self.nb_microbatches >= pg.size() - 1
+AssertionError: Number of microbatches (2) must be at least PP_SIZE-1=7 when using the OneForwardOneBackwardPipelineEngine
+12/14/2024 08:29:52 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]: [Training Plan] Stage Stable Training Stage has 99 remaining training steps and has consumed 0 samples
+12/14/2024 08:29:52 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]: Using dummy data generator
+12/14/2024 08:29:52 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]: [Training Plan] There are 1 training stages 
+12/14/2024 08:29:52 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]: [Stage Stable Training Stage] start from step 1 
+12/14/2024 08:29:52 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]: 
+12/14/2024 08:29:52 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]: [Start training] datetime: 2024-12-14 08:29:52.596855 | mbs: 128 | grad_accum: 2 | global_batch_size: 256 | sequence_length: 4096 | train_steps: 100 | start_iteration_step: 0 | consumed_train_samples: 0
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 252, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 461, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 498, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 243, in train_batch_iter
+    self.nb_microbatches >= pg.size() - 1
+AssertionError: Number of microbatches (2) must be at least PP_SIZE-1=7 when using the OneForwardOneBackwardPipelineEngine
+12/14/2024 08:29:52 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]: Resuming training from stage Stable Training Stage, it has trained for 0 samples and has 99 remaining train steps
+12/14/2024 08:29:52 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:  Memory usage: 4080.25MiB. Peak allocated 5504.00MiB. Peak reserved: 26182.00MiB
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 252, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 461, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 498, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 243, in train_batch_iter
+    self.nb_microbatches >= pg.size() - 1
+AssertionError: Number of microbatches (2) must be at least PP_SIZE-1=7 when using the OneForwardOneBackwardPipelineEngine
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 252, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 461, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 498, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 243, in train_batch_iter
+    self.nb_microbatches >= pg.size() - 1
+AssertionError: Number of microbatches (2) must be at least PP_SIZE-1=7 when using the OneForwardOneBackwardPipelineEngine
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 252, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 461, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 498, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 243, in train_batch_iter
+    self.nb_microbatches >= pg.size() - 1
+AssertionError: Number of microbatches (2) must be at least PP_SIZE-1=7 when using the OneForwardOneBackwardPipelineEngine
+[2024-12-14 08:30:10,532] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 85807 closing signal SIGTERM
+[2024-12-14 08:30:10,532] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 85808 closing signal SIGTERM
+[2024-12-14 08:30:10,532] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 85809 closing signal SIGTERM
+[2024-12-14 08:30:10,532] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 85810 closing signal SIGTERM
+[2024-12-14 08:30:10,532] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 85811 closing signal SIGTERM
+[2024-12-14 08:30:10,532] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 85812 closing signal SIGTERM
+[2024-12-14 08:30:10,545] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 190093 closing signal SIGTERM
+[2024-12-14 08:30:10,545] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 190095 closing signal SIGTERM
+[2024-12-14 08:30:10,545] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 190096 closing signal SIGTERM
+[2024-12-14 08:30:10,545] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 190098 closing signal SIGTERM
+[2024-12-14 08:30:10,545] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 190099 closing signal SIGTERM
+[2024-12-14 08:30:13,213] torch.distributed.elastic.multiprocessing.api: [ERROR] failed (exitcode: 1) local_rank: 6 (pid: 85813) of binary: /fsx/nouamane/miniconda/envs/2-1-cu121/bin/python
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 33, in <module>
+    sys.exit(load_entry_point('torch==2.1.1', 'console_scripts', 'torchrun')())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 264, in launch_agent
+    raise ChildFailedError(
+torch.distributed.elastic.multiprocessing.errors.ChildFailedError: 
+============================================================
+/fsx/nouamane/projects/nanotron/run_train.py FAILED
+------------------------------------------------------------
+Failures:
+[1]:
+  time      : 2024-12-14_08:30:10
+  host      : ip-26-0-165-131.ec2.internal
+  rank      : 7 (local_rank: 7)
+  exitcode  : 1 (pid: 85814)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+------------------------------------------------------------
+Root Cause (first observed failure):
+[0]:
+  time      : 2024-12-14_08:30:10
+  host      : ip-26-0-165-131.ec2.internal
+  rank      : 6 (local_rank: 6)
+  exitcode  : 1 (pid: 85813)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+============================================================
+srun: error: ip-26-0-165-131: task 1: Exited with exit code 1
+srun: launch/slurm: _step_signal: Terminating StepId=13442739.0
+slurmstepd: error: *** STEP 13442739.0 ON ip-26-0-165-59 CANCELLED AT 2024-12-14T08:30:13 ***
+[2024-12-14 08:30:13,526] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-14 08:30:13,526] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 190098 closing signal SIGTERM
+[2024-12-14 08:30:13,526] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 190099 closing signal SIGTERM
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 33, in <module>
+    sys.exit(load_entry_point('torch==2.1.1', 'console_scripts', 'torchrun')())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 878, in _invoke_run
+    run_result = self._monitor_workers(self._worker_group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/local_elastic_agent.py", line 307, in _monitor_workers
+    result = self._pcontext.wait(0)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 288, in wait
+    return self._poll()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 668, in _poll
+    self.close()  # terminate all running procs
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 331, in close
+    self._close(death_sig=death_sig, timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 713, in _close
+    handler.proc.wait(time_to_wait)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1209, in wait
+    return self._wait(timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1953, in _wait
+    time.sleep(delay)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 190020 got signal: 15
+srun: error: ip-26-0-165-59: task 0: Exited with exit code 1
diff --git a/logs/13442742-bench_8.86G_dp1_tp2_pp8_acc1_mbs256_seq4096_zero0_tpmodeRED_vocab131k.out b/logs/13442742-bench_8.86G_dp1_tp2_pp8_acc1_mbs256_seq4096_zero0_tpmodeRED_vocab131k.out
new file mode 100644
index 0000000000000000000000000000000000000000..ebd4685dfb3b2371a7cf84c764b96df7e26ae55e
--- /dev/null
+++ b/logs/13442742-bench_8.86G_dp1_tp2_pp8_acc1_mbs256_seq4096_zero0_tpmodeRED_vocab131k.out
@@ -0,0 +1,881 @@
++ source /etc/profile.d/modules.sh
+++ . /usr/share/modules/init/bash
++++ unset _mlshdbg
++++ '[' 0 = 1 ']'
++++ unset _mlre _mlIFS
++++ '[' -n x ']'
++++ _mlIFS=' 	
+'
++++ IFS=' '
++++ '[' -n '' ']'
+++++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash autoinit
++++ _mlcode='module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+test 0;'
++++ _mlret=0
++++ '[' -n x ']'
++++ IFS=' 	
+'
++++ unset _mlIFS
++++ unset _mlre _mlv _mlrv
++++ '[' 0 -eq 0 ']'
++++ eval 'module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+test 0;'
+++++ MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl
+++++ export MODULES_CMD
+++++ ENV=/usr/share/modules/init/profile.sh
+++++ export ENV
+++++ BASH_ENV=/usr/share/modules/init/bash
+++++ export BASH_ENV
+++++ MODULESHOME=/usr/share/modules
+++++ export MODULESHOME
+++++ test 0
++++ '[' 0 = 1 ']'
++++ '[' -t 2 ']'
++++ export -f module
++++ export -f switchml
++++ '[' 5 -ge 3 ']'
++++ [[ ehxB =~ i ]]
++++ [[ ! :/admin/home/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin: =~ :/usr/bin: ]]
++++ '[' '!' -n x ']'
+++++ manpath
++++ [[ ! :/admin/home/nouamane/miniconda/envs/2-1-cu121/man:/admin/home/nouamane/miniconda/envs/2-1-cu121/share/man:/admin/home/nouamane/.local/share/man:/fsx/nouamane/miniconda/envs/2-1-cu121/man:/fsx/nouamane/miniconda/envs/2-1-cu121/share/man:/fsx/nouamane/miniconda/man:/fsx/nouamane/miniconda/share/man:/opt/amazon/openmpi/share/man:/opt/amazon/efa/share/man:/opt/slurm/share/man:/usr/local/man:/usr/local/share/man:/usr/share/man:/admin/home/nouamane/.fzf/man:: =~ :/usr/share/man: ]]
++++ unset _mlcode _mlret
++++ '[' -n '' ']'
++ module load cuda/12.1
++ unset _mlshdbg
++ '[' 0 = 1 ']'
++ unset _mlre _mlIFS
++ '[' -n x ']'
++ _mlIFS=' 	
+'
++ IFS=' '
++ '[' -n '' ']'
+++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash load cuda/12.1
++ eval 'CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include;' export 'CPATH;
+LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/opt/aws-ofi-nccl/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/targets/x86_64-linux/lib/:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/lib:/usr/lib::/opt/amazon/openmpi/lib:/opt/amazon/efa/lib;' export 'LD_LIBRARY_PATH;
+FI_EFA_FORK_SAFE=1;' export 'FI_EFA_FORK_SAFE;
+MANPATH=/usr/local/cuda-12.1/share/man::;' export 'MANPATH;
+LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64;' export 'LIBRARY_PATH;
+_LMFILES_=/usr/share/modules/modulefiles/cuda/12.1;' export '_LMFILES_;
+LOADEDMODULES=cuda/12.1;' export 'LOADEDMODULES;
+MPI_PATH=/opt/amazon/openmpi;' export 'MPI_PATH;
+NCCL_HOME_modshare=/opt/nccl/build:1;' export 'NCCL_HOME_modshare;
+NCCL_PROTO=simple;' export 'NCCL_PROTO;
+MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1;' export 'MANPATH_modshare;
+LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1;' export 'LIBRARY_PATH_modshare;
+NCCL_SOCKET_IFNAME=enp;' export 'NCCL_SOCKET_IFNAME;
+AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl;' export 'AWS_OFI_NCCL_HOME;
+NCCL_HOME=/opt/nccl/build;' export 'NCCL_HOME;
+FI_PROVIDER=efa;' export 'FI_PROVIDER;
+AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1;' export 'AWS_OFI_NCCL_HOME_modshare;
+CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1;' export 'CPATH_modshare;
+LD_LIBRARY_PATH_modshare=:1:/opt/amazon/efa/lib:3:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib/:1:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2:/opt/aws-ofi-nccl/lib:2:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/efa/lib:1:/usr/lib:1;' export 'LD_LIBRARY_PATH_modshare;
+FI_EFA_ENABLE_SHM_TRANSFER=1;' export 'FI_EFA_ENABLE_SHM_TRANSFER;
+_LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1;' export '_LMFILES__modshare;
+LOADEDMODULES_modshare=cuda/12.1:1;' export 'LOADEDMODULES_modshare;
+MPI_PATH_modshare=/opt/amazon/openmpi:1;' export 'MPI_PATH_modshare;
+PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin;' export 'PATH;
+CUDA_HOME=/usr/local/cuda-12.1;' export 'CUDA_HOME;
+PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/usr/local/cuda-12.1/include:1:/opt/amazon/efa/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/usr/sbin:1:/opt/slurm/bin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:2:/opt/amazon/openmpi/bin:1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:1:/admin/home/nouamane/.fzf/bin:1:/admin/home/nouamane/miniconda/envs/2-1-cu121/bin:1:/usr/local/games:1;' export 'PATH_modshare;
+test' '0;'
+++ CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include
+++ export CPATH
+++ LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/opt/aws-ofi-nccl/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/targets/x86_64-linux/lib/:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/lib:/usr/lib::/opt/amazon/openmpi/lib:/opt/amazon/efa/lib
+++ export LD_LIBRARY_PATH
+++ FI_EFA_FORK_SAFE=1
+++ export FI_EFA_FORK_SAFE
+++ MANPATH=/usr/local/cuda-12.1/share/man::
+++ export MANPATH
+++ LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64
+++ export LIBRARY_PATH
+++ _LMFILES_=/usr/share/modules/modulefiles/cuda/12.1
+++ export _LMFILES_
+++ LOADEDMODULES=cuda/12.1
+++ export LOADEDMODULES
+++ MPI_PATH=/opt/amazon/openmpi
+++ export MPI_PATH
+++ NCCL_HOME_modshare=/opt/nccl/build:1
+++ export NCCL_HOME_modshare
+++ NCCL_PROTO=simple
+++ export NCCL_PROTO
+++ MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1
+++ export MANPATH_modshare
+++ LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1
+++ export LIBRARY_PATH_modshare
+++ NCCL_SOCKET_IFNAME=enp
+++ export NCCL_SOCKET_IFNAME
+++ AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl
+++ export AWS_OFI_NCCL_HOME
+++ NCCL_HOME=/opt/nccl/build
+++ export NCCL_HOME
+++ FI_PROVIDER=efa
+++ export FI_PROVIDER
+++ AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1
+++ export AWS_OFI_NCCL_HOME_modshare
+++ CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1
+++ export CPATH_modshare
+++ LD_LIBRARY_PATH_modshare=:1:/opt/amazon/efa/lib:3:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib/:1:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2:/opt/aws-ofi-nccl/lib:2:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/efa/lib:1:/usr/lib:1
+++ export LD_LIBRARY_PATH_modshare
+++ FI_EFA_ENABLE_SHM_TRANSFER=1
+++ export FI_EFA_ENABLE_SHM_TRANSFER
+++ _LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1
+++ export _LMFILES__modshare
+++ LOADEDMODULES_modshare=cuda/12.1:1
+++ export LOADEDMODULES_modshare
+++ MPI_PATH_modshare=/opt/amazon/openmpi:1
+++ export MPI_PATH_modshare
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin
+++ export PATH
+++ CUDA_HOME=/usr/local/cuda-12.1
+++ export CUDA_HOME
+++ PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/usr/local/cuda-12.1/include:1:/opt/amazon/efa/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/usr/sbin:1:/opt/slurm/bin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:2:/opt/amazon/openmpi/bin:1:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:1:/admin/home/nouamane/.fzf/bin:1:/admin/home/nouamane/miniconda/envs/2-1-cu121/bin:1:/usr/local/games:1
+++ export PATH_modshare
+++ test 0
++ _mlstatus=0
++ '[' -n x ']'
++ IFS=' 	
+'
++ unset _mlre _mlv _mlrv _mlIFS
++ '[' -n '' ']'
++ unset _mlshdbg
++ return 0
++ source /fsx/nouamane/miniconda/bin/activate
+++ _CONDA_ROOT=/fsx/nouamane/miniconda
+++ . /fsx/nouamane/miniconda/etc/profile.d/conda.sh
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ '[' -z x ']'
+++ conda activate
+++ local cmd=activate
+++ case "$cmd" in
+++ __conda_activate activate
+++ '[' -n '' ']'
+++ local ask_conda
++++ PS1=
++++ __conda_exe shell.posix activate
++++ /fsx/nouamane/miniconda/bin/conda shell.posix activate
+++ ask_conda='. "/admin/home/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/fsx/nouamane/miniconda/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_2='\''/admin/home/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\'''
+++ eval '. "/admin/home/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/fsx/nouamane/miniconda/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_2='\''/admin/home/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\'''
++++ . /admin/home/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh
+++++ test -n ''
+++++ unset XML_CATALOG_FILES
+++++ unset xml_catalog_files_libxml2
++++ PS1='(base) '
++++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/fsx/nouamane/miniconda/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin
++++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/fsx/nouamane/miniconda/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin
++++ export CONDA_PREFIX=/fsx/nouamane/miniconda
++++ CONDA_PREFIX=/fsx/nouamane/miniconda
++++ export CONDA_SHLVL=3
++++ CONDA_SHLVL=3
++++ export CONDA_DEFAULT_ENV=base
++++ CONDA_DEFAULT_ENV=base
++++ export 'CONDA_PROMPT_MODIFIER=(base) '
++++ CONDA_PROMPT_MODIFIER='(base) '
++++ export CONDA_PREFIX_2=/admin/home/nouamane/miniconda/envs/2-1-cu121
++++ CONDA_PREFIX_2=/admin/home/nouamane/miniconda/envs/2-1-cu121
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ __conda_hashr
+++ '[' -n '' ']'
+++ '[' -n '' ']'
+++ hash -r
++ conda activate 2-1-cu121
++ local cmd=activate
++ case "$cmd" in
++ __conda_activate activate 2-1-cu121
++ '[' -n '' ']'
++ local ask_conda
+++ PS1='(base) '
+++ __conda_exe shell.posix activate 2-1-cu121
+++ /fsx/nouamane/miniconda/bin/conda shell.posix activate 2-1-cu121
++ ask_conda='PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''4'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_3='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
++ eval 'PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''4'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_3='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
+++ PS1='(2-1-cu121) '
+++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin
+++ export CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ export CONDA_SHLVL=4
+++ CONDA_SHLVL=4
+++ export CONDA_DEFAULT_ENV=2-1-cu121
+++ CONDA_DEFAULT_ENV=2-1-cu121
+++ export 'CONDA_PROMPT_MODIFIER=(2-1-cu121) '
+++ CONDA_PROMPT_MODIFIER='(2-1-cu121) '
+++ export CONDA_PREFIX_3=/fsx/nouamane/miniconda
+++ CONDA_PREFIX_3=/fsx/nouamane/miniconda
+++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ export _CE_M=
+++ _CE_M=
+++ export _CE_CONDA=
+++ _CE_CONDA=
+++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh
++++ test -n ''
++++ xml_catalog_files_libxml2=
++++ XML_CATALOG_FILES=
++++ conda_catalog_files=
++++ ifs_libxml2=' 	
+'
++++ IFS=' '
++++ rem=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ for pre in ${rem}
++++ test '' = /fsx/nouamane/miniconda/envs/2-1-cu121
++++ conda_catalog_files=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ rem=
++++ IFS=' 	
+'
++++ conda_catalog_files='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ export 'XML_CATALOG_FILES=file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ XML_CATALOG_FILES='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ unset conda_catalog_files ifs_libxml2 rem
++ __conda_hashr
++ '[' -n '' ']'
++ '[' -n '' ']'
++ hash -r
++ export PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin
++ PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/001668006cc714afd397f4ef0d52862f5a095530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin
+++ scontrol show hostnames ip-26-0-169-132,ip-26-0-171-21
++ export 'NODELIST=ip-26-0-169-132
+ip-26-0-171-21'
++ NODELIST='ip-26-0-169-132
+ip-26-0-171-21'
+++ scontrol show hostnames ip-26-0-169-132,ip-26-0-171-21
+++ head -n1
++ export MASTER_NODE=ip-26-0-169-132
++ MASTER_NODE=ip-26-0-169-132
++ export MASTER_PORT=12356
++ MASTER_PORT=12356
++ export NNODES=2
++ NNODES=2
++ export GPUS_PER_NODE=8
++ GPUS_PER_NODE=8
++ export WORLD_SIZE=16
++ WORLD_SIZE=16
++ export CUDA_DEVICE_MAX_CONNECTIONS=1
++ CUDA_DEVICE_MAX_CONNECTIONS=1
++ export NCCL_DEBUG=WARN
++ NCCL_DEBUG=WARN
++ export NANOTRON_BENCHMARK=1
++ NANOTRON_BENCHMARK=1
++ export WANDB_MODE=disabled
++ WANDB_MODE=disabled
++ export TORCH_NCCL_ASYNC_ERROR_HANDLING=1
++ TORCH_NCCL_ASYNC_ERROR_HANDLING=1
++ echo '=== GPU Topology ==='
+=== GPU Topology ===
++ nvidia-smi topo -m
+	GPU0	GPU1	GPU2	GPU3	GPU4	GPU5	GPU6	GPU7	CPU Affinity	NUMA Affinity	GPU NUMA ID
+GPU0	 X 	NV18	NV18	NV18	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU1	NV18	 X 	NV18	NV18	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU2	NV18	NV18	 X 	NV18	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU3	NV18	NV18	NV18	 X 	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU4	NV18	NV18	NV18	NV18	 X 	NV18	NV18	NV18	48-95	1		N/A
+GPU5	NV18	NV18	NV18	NV18	NV18	 X 	NV18	NV18	48-95	1		N/A
+GPU6	NV18	NV18	NV18	NV18	NV18	NV18	 X 	NV18	48-95	1		N/A
+GPU7	NV18	NV18	NV18	NV18	NV18	NV18	NV18	 X 	48-95	1		N/A
+
+Legend:
+
+  X    = Self
+  SYS  = Connection traversing PCIe as well as the SMP interconnect between NUMA nodes (e.g., QPI/UPI)
+  NODE = Connection traversing PCIe as well as the interconnect between PCIe Host Bridges within a NUMA node
+  PHB  = Connection traversing PCIe as well as a PCIe Host Bridge (typically the CPU)
+  PXB  = Connection traversing multiple PCIe bridges (without traversing the PCIe Host Bridge)
+  PIX  = Connection traversing at most a single PCIe bridge
+  NV#  = Connection traversing a bonded set of # NVLinks
++ echo ==================
+==================
++ echo 'Master node: ip-26-0-169-132'
+Master node: ip-26-0-169-132
++ echo 'All nodes: ip-26-0-169-132
+ip-26-0-171-21'
+All nodes: ip-26-0-169-132
+ip-26-0-171-21
++ echo 'World size: 16'
+World size: 16
++ srun --wait=0 --kill-on-bad-exit=1 torchrun --nnodes=2 --nproc_per_node=8 --rdzv_id=13442742 --rdzv_backend=c10d --rdzv_endpoint=ip-26-0-169-132:12356 --max_restarts 0 --rdzv_conf timeout=60 /fsx/nouamane/projects/nanotron/run_train.py --config-file benchmark/configs/config_8.86G_dp1_tp2_pp8_acc1_mbs256_seq4096_zero0_tpmodeRED_vocab131k.yaml
+[2024-12-14 08:29:03,959] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-14 08:29:03,994] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-14 08:29:03,959] torch.distributed.run: [WARNING] 
+[2024-12-14 08:29:03,959] torch.distributed.run: [WARNING] *****************************************
+[2024-12-14 08:29:03,959] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-14 08:29:03,959] torch.distributed.run: [WARNING] *****************************************
+[2024-12-14 08:29:03,995] torch.distributed.run: [WARNING] 
+[2024-12-14 08:29:03,995] torch.distributed.run: [WARNING] *****************************************
+[2024-12-14 08:29:03,995] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-14 08:29:03,995] torch.distributed.run: [WARNING] *****************************************
+NCCL version 2.18.6+cuda12.1
+12/14/2024 08:29:26 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-132]: Measuring inter-GPU and intra-node bandwidth...
+NCCL version 2.18.6+cuda12.1
+12/14/2024 08:29:42 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-132]: Bandwidth measurement complete. Time taken: 16.21 seconds
+12/14/2024 08:29:42 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-132]: Config:
+12/14/2024 08:29:42 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-132]: Config(general=GeneralArgs(project='debug',
+12/14/2024 08:29:42 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-132]:                            run='8.86G_dp1_tp2_pp8_acc1_mbs256_seq4096_zero0_tpmodeRED_vocab131k',
+12/14/2024 08:29:42 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-132]:                            seed=42,
+12/14/2024 08:29:42 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-132]:                            step=None,
+12/14/2024 08:29:42 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-132]:                            consumed_train_samples=None,
+12/14/2024 08:29:42 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-132]:                            benchmark_csv_path=PosixPath('benchmark/results/bench_final2.csv'),
+12/14/2024 08:29:42 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-132]:                            ignore_sanity_checks=True),
+12/14/2024 08:29:42 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-132]:        parallelism=ParallelismArgs(dp=1,
+12/14/2024 08:29:42 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-132]:                                    pp=8,
+12/14/2024 08:29:42 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-132]:                                    tp=2,
+12/14/2024 08:29:42 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-132]:                                    pp_engine=<nanotron.parallel.pipeline_parallel.engine.OneForwardOneBackwardPipelineEngine object at 0x7fcd10120e80>,
+12/14/2024 08:29:42 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-132]:                                    tp_mode=<TensorParallelLinearMode.REDUCE_SCATTER: 2>,
+12/14/2024 08:29:42 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-132]:                                    tp_linear_async_communication=True,
+12/14/2024 08:29:42 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-132]:                                    recompute_layer=False,
+12/14/2024 08:29:42 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-132]:                                    tp_recompute_allgather=True,
+12/14/2024 08:29:42 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-132]:                                    expert_parallel_size=1),
+12/14/2024 08:29:42 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-132]:        model=ModelArgs(model_config=LlamaConfig(bos_token_id=0,
+12/14/2024 08:29:42 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-132]:                                                 eos_token_id=0,
+12/14/2024 08:29:42 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-132]:                                                 hidden_act='silu',
+12/14/2024 08:29:42 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-132]:                                                 hidden_size=4096,
+12/14/2024 08:29:42 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-132]:                                                 initializer_range=0.02,
+12/14/2024 08:29:42 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-132]:                                                 intermediate_size=14336,
+12/14/2024 08:29:42 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-132]:                                                 is_llama_config=True,
+12/14/2024 08:29:42 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-132]:                                                 max_position_embeddings=4096,
+12/14/2024 08:29:42 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-132]:                                                 num_attention_heads=32,
+12/14/2024 08:29:42 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-132]:                                                 num_hidden_layers=32,
+12/14/2024 08:29:42 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-132]:                                                 num_key_value_heads=32,
+12/14/2024 08:29:42 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-132]:                                                 pad_token_id=None,
+12/14/2024 08:29:42 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-132]:                                                 pretraining_tp=1,
+12/14/2024 08:29:42 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-132]:                                                 rms_norm_eps=1e-05,
+12/14/2024 08:29:42 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-132]:                                                 rope_scaling=None,
+12/14/2024 08:29:42 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-132]:                                                 rope_theta=10000.0,
+12/14/2024 08:29:42 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-132]:                                                 rope_interleaved=False,
+12/14/2024 08:29:42 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-132]:                                                 tie_word_embeddings=False,
+12/14/2024 08:29:42 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-132]:                                                 use_cache=True,
+12/14/2024 08:29:42 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-132]:                                                 vocab_size=131072),
+12/14/2024 08:29:42 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-132]:                        init_method=RandomInit(std=0.02),
+12/14/2024 08:29:42 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-132]:                        dtype=torch.bfloat16,
+12/14/2024 08:29:42 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-132]:                        make_vocab_size_divisible_by=1,
+12/14/2024 08:29:42 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-132]:                        ddp_bucket_cap_mb=25),
+12/14/2024 08:29:42 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-132]:        tokenizer=TokenizerArgs(tokenizer_name_or_path='robot-test/dummy-tokenizer-wordlevel',
+12/14/2024 08:29:42 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-132]:                                tokenizer_revision=None,
+12/14/2024 08:29:42 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-132]:                                tokenizer_max_length=None),
+12/14/2024 08:29:42 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-132]:        checkpoints=CheckpointsArgs(checkpoints_path=PosixPath('checkpoints'),
+12/14/2024 08:29:42 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-132]:                                    checkpoint_interval=10000,
+12/14/2024 08:29:42 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-132]:                                    save_initial_state=False,
+12/14/2024 08:29:42 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-132]:                                    save_final_state=False,
+12/14/2024 08:29:42 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-132]:                                    resume_checkpoint_path=None,
+12/14/2024 08:29:42 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-132]:                                    checkpoints_path_is_shared_file_system=False),
+12/14/2024 08:29:42 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-132]:        logging=LoggingArgs(log_level='info',
+12/14/2024 08:29:42 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-132]:                            log_level_replica='info',
+12/14/2024 08:29:42 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-132]:                            iteration_step_info_interval=1),
+12/14/2024 08:29:42 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-132]:        tokens=TokensArgs(sequence_length=4096,
+12/14/2024 08:29:42 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-132]:                          train_steps=100,
+12/14/2024 08:29:42 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-132]:                          micro_batch_size=256,
+12/14/2024 08:29:42 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-132]:                          batch_accumulation_per_replica=1,
+12/14/2024 08:29:42 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-132]:                          val_check_interval=100,
+12/14/2024 08:29:42 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-132]:                          limit_val_batches=0,
+12/14/2024 08:29:42 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-132]:                          limit_test_batches=0),
+12/14/2024 08:29:42 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-132]:        optimizer=OptimizerArgs(optimizer_factory=AdamWOptimizerArgs(adam_eps=1e-08,
+12/14/2024 08:29:42 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-132]:                                                                     adam_beta1=0.9,
+12/14/2024 08:29:42 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-132]:                                                                     adam_beta2=0.95,
+12/14/2024 08:29:42 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-132]:                                                                     torch_adam_is_fused=True,
+12/14/2024 08:29:42 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-132]:                                                                     name='adamW'),
+12/14/2024 08:29:42 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-132]:                                zero_stage=0,
+12/14/2024 08:29:42 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-132]:                                weight_decay=0.01,
+12/14/2024 08:29:42 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-132]:                                clip_grad=1.0,
+12/14/2024 08:29:42 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-132]:                                accumulate_grad_in_fp32=True,
+12/14/2024 08:29:42 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-132]:                                learning_rate_scheduler=LRSchedulerArgs(learning_rate=0.0003,
+12/14/2024 08:29:42 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-132]:                                                                        lr_warmup_steps=2,
+12/14/2024 08:29:42 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-132]:                                                                        lr_warmup_style='linear',
+12/14/2024 08:29:42 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-132]:                                                                        lr_decay_style='cosine',
+12/14/2024 08:29:42 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-132]:                                                                        lr_decay_steps=13,
+12/14/2024 08:29:42 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-132]:                                                                        lr_decay_starting_step=None,
+12/14/2024 08:29:42 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-132]:                                                                        min_decay_lr=1e-05)),
+12/14/2024 08:29:42 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-132]:        data_stages=[DatasetStageArgs(name='Stable Training Stage',
+12/14/2024 08:29:42 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-132]:                                      start_training_step=1,
+12/14/2024 08:29:42 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-132]:                                      data=DataArgs(dataset=None,
+12/14/2024 08:29:42 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-132]:                                                    seed=42,
+12/14/2024 08:29:42 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-132]:                                                    num_loading_workers=1))],
+12/14/2024 08:29:42 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-132]:        profiler=None,
+12/14/2024 08:29:42 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-132]:        lighteval=None,
+12/14/2024 08:29:42 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-132]:        s3_upload=None)
+12/14/2024 08:29:42 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-132]: Model Config:
+12/14/2024 08:29:42 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-132]: LlamaConfig(bos_token_id=0,
+12/14/2024 08:29:42 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-132]:             eos_token_id=0,
+12/14/2024 08:29:42 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-132]:             hidden_act='silu',
+12/14/2024 08:29:42 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-132]:             hidden_size=4096,
+12/14/2024 08:29:42 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-132]:             initializer_range=0.02,
+12/14/2024 08:29:42 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-132]:             intermediate_size=14336,
+12/14/2024 08:29:42 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-132]:             is_llama_config=True,
+12/14/2024 08:29:42 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-132]:             max_position_embeddings=4096,
+12/14/2024 08:29:42 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-132]:             num_attention_heads=32,
+12/14/2024 08:29:42 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-132]:             num_hidden_layers=32,
+12/14/2024 08:29:42 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-132]:             num_key_value_heads=32,
+12/14/2024 08:29:42 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-132]:             pad_token_id=None,
+12/14/2024 08:29:42 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-132]:             pretraining_tp=1,
+12/14/2024 08:29:42 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-132]:             rms_norm_eps=1e-05,
+12/14/2024 08:29:42 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-132]:             rope_scaling=None,
+12/14/2024 08:29:42 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-132]:             rope_theta=10000.0,
+12/14/2024 08:29:42 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-132]:             rope_interleaved=False,
+12/14/2024 08:29:42 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-132]:             tie_word_embeddings=False,
+12/14/2024 08:29:42 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-132]:             use_cache=True,
+12/14/2024 08:29:42 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-132]:             vocab_size=131072)
+12/14/2024 08:29:42 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-132]: Building model..
+12/14/2024 08:29:42 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-132]: Initialize RoPE Theta = 10000.0
+12/14/2024 08:29:42 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-132]: Setting PP block ranks...
+NCCL version 2.18.6+cuda12.1
+NCCL version 2.18.6+cuda12.1
+NCCL version 2.18.6+cuda12.1
+NCCL version 2.18.6+cuda12.1
+NCCL version 2.18.6+cuda12.1
+NCCL version 2.18.6+cuda12.1
+NCCL version 2.18.6+cuda12.1
+12/14/2024 08:29:52 [INFO|DP=0|PP=2|TP=0|ip-26-0-169-132]: Local number of parameters: 487M (928.06MiB)
+12/14/2024 08:29:52 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-132]: Total number of parameters: 8.86G (16897.02MiB)
+12/14/2024 08:29:52 [INFO|DP=0|PP=5|TP=0|ip-26-0-171-21]: Local number of parameters: 487M (928.06MiB)
+12/14/2024 08:29:52 [INFO|DP=0|PP=7|TP=0|ip-26-0-171-21]: Local number of parameters: 512M (976.04MiB)
+12/14/2024 08:29:52 [INFO|DP=0|PP=1|TP=0|ip-26-0-169-132]: Local number of parameters: 487M (928.06MiB)
+12/14/2024 08:29:52 [INFO|DP=0|PP=3|TP=0|ip-26-0-169-132]: Local number of parameters: 608M (1160.08MiB)
+12/14/2024 08:29:52 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-132]: Local number of parameters: 877M (1672.08MiB)
+12/14/2024 08:29:52 [INFO|DP=0|PP=4|TP=1|ip-26-0-171-21]: Local number of parameters: 487M (928.06MiB)
+12/14/2024 08:29:52 [INFO|DP=0|PP=4|TP=0|ip-26-0-171-21]: Local number of parameters: 487M (928.06MiB)
+12/14/2024 08:29:52 [INFO|DP=0|PP=6|TP=1|ip-26-0-171-21]: Local number of parameters: 487M (928.06MiB)
+12/14/2024 08:29:52 [INFO|DP=0|PP=7|TP=1|ip-26-0-171-21]: Local number of parameters: 512M (976.04MiB)
+12/14/2024 08:29:52 [INFO|DP=0|PP=5|TP=1|ip-26-0-171-21]: Local number of parameters: 487M (928.06MiB)
+12/14/2024 08:29:52 [INFO|DP=0|PP=6|TP=0|ip-26-0-171-21]: Local number of parameters: 487M (928.06MiB)
+12/14/2024 08:29:52 [INFO|DP=0|PP=2|TP=0|ip-26-0-169-132]: [After model building] Memory usage: 928.08MiB. Peak allocated: 5504.00MiB Peak reserved: 22914.00MiB
+12/14/2024 08:29:52 [INFO|DP=0|PP=7|TP=0|ip-26-0-171-21]: [After model building] Memory usage: 976.05MiB. Peak allocated: 5504.00MiB Peak reserved: 19842.00MiB
+12/14/2024 08:29:52 [INFO|DP=0|PP=5|TP=0|ip-26-0-171-21]: [After model building] Memory usage: 928.08MiB. Peak allocated: 5504.00MiB Peak reserved: 17794.00MiB
+12/14/2024 08:29:52 [INFO|DP=0|PP=3|TP=0|ip-26-0-169-132]: [After model building] Memory usage: 1160.09MiB. Peak allocated: 5504.00MiB Peak reserved: 22914.00MiB
+12/14/2024 08:29:52 [INFO|DP=0|PP=1|TP=0|ip-26-0-169-132]: [After model building] Memory usage: 928.08MiB. Peak allocated: 5504.00MiB Peak reserved: 22914.00MiB
+12/14/2024 08:29:52 [INFO|DP=0|PP=7|TP=1|ip-26-0-171-21]: [After model building] Memory usage: 976.05MiB. Peak allocated: 5504.00MiB Peak reserved: 19842.00MiB
+12/14/2024 08:29:52 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-132]: [After model building] Memory usage: 1672.09MiB. Peak allocated: 5504.00MiB Peak reserved: 22914.00MiB
+12/14/2024 08:29:52 [INFO|DP=0|PP=4|TP=1|ip-26-0-171-21]: [After model building] Memory usage: 928.08MiB. Peak allocated: 5504.00MiB Peak reserved: 18818.00MiB
+12/14/2024 08:29:52 [INFO|DP=0|PP=6|TP=1|ip-26-0-171-21]: [After model building] Memory usage: 928.08MiB. Peak allocated: 5504.00MiB Peak reserved: 19842.00MiB
+12/14/2024 08:29:52 [INFO|DP=0|PP=4|TP=0|ip-26-0-171-21]: [After model building] Memory usage: 928.08MiB. Peak allocated: 5504.00MiB Peak reserved: 18818.00MiB
+12/14/2024 08:29:52 [INFO|DP=0|PP=6|TP=0|ip-26-0-171-21]: [After model building] Memory usage: 928.08MiB. Peak allocated: 5504.00MiB Peak reserved: 18818.00MiB
+12/14/2024 08:29:52 [INFO|DP=0|PP=5|TP=1|ip-26-0-171-21]: [After model building] Memory usage: 928.08MiB. Peak allocated: 5504.00MiB Peak reserved: 19842.00MiB
+12/14/2024 08:29:52 [INFO|DP=0|PP=0|TP=1|ip-26-0-169-132]: Local number of parameters: 877M (1672.08MiB)
+12/14/2024 08:29:52 [INFO|DP=0|PP=2|TP=1|ip-26-0-169-132]: Local number of parameters: 487M (928.06MiB)
+12/14/2024 08:29:52 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-132]: No checkpoint path provided.
+12/14/2024 08:29:52 [INFO|DP=0|PP=3|TP=1|ip-26-0-169-132]: Local number of parameters: 608M (1160.08MiB)
+12/14/2024 08:29:52 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-132]: Parametrizing model parameters using StandardParametrizator
+12/14/2024 08:29:52 [INFO|DP=0|PP=0|TP=1|ip-26-0-169-132]: [After model building] Memory usage: 1672.09MiB. Peak allocated: 5504.00MiB Peak reserved: 22914.00MiB
+12/14/2024 08:29:52 [INFO|DP=0|PP=3|TP=1|ip-26-0-169-132]: [After model building] Memory usage: 1160.09MiB. Peak allocated: 5504.00MiB Peak reserved: 22914.00MiB
+12/14/2024 08:29:52 [INFO|DP=0|PP=2|TP=1|ip-26-0-169-132]: [After model building] Memory usage: 928.08MiB. Peak allocated: 5504.00MiB Peak reserved: 22914.00MiB
+12/14/2024 08:29:52 [INFO|DP=0|PP=1|TP=1|ip-26-0-169-132]: Local number of parameters: 487M (928.06MiB)
+12/14/2024 08:29:52 [INFO|DP=0|PP=1|TP=1|ip-26-0-169-132]: [After model building] Memory usage: 928.08MiB. Peak allocated: 5504.00MiB Peak reserved: 22914.00MiB
+NCCL version 2.18.6+cuda12.1
+NCCL version 2.18.6+cuda12.1
+NCCL version 2.18.6+cuda12.1
+NCCL version 2.18.6+cuda12.1
+NCCL version 2.18.6+cuda12.1
+NCCL version 2.18.6+cuda12.1
+NCCL version 2.18.6+cuda12.1
+12/14/2024 08:29:52 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-132]: [Optimizer Building] Using LearningRateForSP as learning rate
+12/14/2024 08:29:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-132]: [Training Plan] Stage Stable Training Stage has 99 remaining training steps and has consumed 0 samples
+12/14/2024 08:29:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-132]: Using dummy data generator
+12/14/2024 08:29:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-132]: [Training Plan] There are 1 training stages 
+12/14/2024 08:29:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-132]: [Stage Stable Training Stage] start from step 1 
+12/14/2024 08:29:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-132]: 
+12/14/2024 08:29:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-132]: [Start training] datetime: 2024-12-14 08:29:53.332100 | mbs: 256 | grad_accum: 1 | global_batch_size: 256 | sequence_length: 4096 | train_steps: 100 | start_iteration_step: 0 | consumed_train_samples: 0
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 252, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 461, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 498, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 243, in train_batch_iter
+    self.nb_microbatches >= pg.size() - 1
+AssertionError: Number of microbatches (1) must be at least PP_SIZE-1=7 when using the OneForwardOneBackwardPipelineEngine
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 252, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 461, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 498, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 243, in train_batch_iter
+    self.nb_microbatches >= pg.size() - 1
+AssertionError: Number of microbatches (1) must be at least PP_SIZE-1=7 when using the OneForwardOneBackwardPipelineEngine
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 252, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 461, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 498, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 243, in train_batch_iter
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 252, in <module>
+    self.nb_microbatches >= pg.size() - 1
+AssertionError: Number of microbatches (1) must be at least PP_SIZE-1=7 when using the OneForwardOneBackwardPipelineEngine
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 461, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 498, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 243, in train_batch_iter
+    self.nb_microbatches >= pg.size() - 1
+AssertionError: Number of microbatches (1) must be at least PP_SIZE-1=7 when using the OneForwardOneBackwardPipelineEngine
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 252, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 461, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 498, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 243, in train_batch_iter
+    self.nb_microbatches >= pg.size() - 1
+AssertionError: Number of microbatches (1) must be at least PP_SIZE-1=7 when using the OneForwardOneBackwardPipelineEngine
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 252, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 461, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 498, in training_step
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 252, in <module>
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 243, in train_batch_iter
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 461, in train
+    self.nb_microbatches >= pg.size() - 1
+AssertionError: Number of microbatches (1) must be at least PP_SIZE-1=7 when using the OneForwardOneBackwardPipelineEngine
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 498, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 243, in train_batch_iter
+    self.nb_microbatches >= pg.size() - 1
+AssertionError: Number of microbatches (1) must be at least PP_SIZE-1=7 when using the OneForwardOneBackwardPipelineEngine
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 252, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 461, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 498, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 243, in train_batch_iter
+    self.nb_microbatches >= pg.size() - 1
+AssertionError: Number of microbatches (1) must be at least PP_SIZE-1=7 when using the OneForwardOneBackwardPipelineEngine
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 252, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 461, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 498, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 243, in train_batch_iter
+    self.nb_microbatches >= pg.size() - 1
+AssertionError: Number of microbatches (1) must be at least PP_SIZE-1=7 when using the OneForwardOneBackwardPipelineEngine
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 252, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 461, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 498, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 243, in train_batch_iter
+    self.nb_microbatches >= pg.size() - 1
+AssertionError: Number of microbatches (1) must be at least PP_SIZE-1=7 when using the OneForwardOneBackwardPipelineEngine
+12/14/2024 08:29:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-132]: Resuming training from stage Stable Training Stage, it has trained for 0 samples and has 99 remaining train steps
+12/14/2024 08:29:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-169-132]:  Memory usage: 8360.40MiB. Peak allocated 8360.40MiB. Peak reserved: 29606.00MiB
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 252, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 461, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 498, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 243, in train_batch_iter
+    self.nb_microbatches >= pg.size() - 1
+AssertionError: Number of microbatches (1) must be at least PP_SIZE-1=7 when using the OneForwardOneBackwardPipelineEngine
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 252, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 461, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 498, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 243, in train_batch_iter
+    self.nb_microbatches >= pg.size() - 1
+AssertionError: Number of microbatches (1) must be at least PP_SIZE-1=7 when using the OneForwardOneBackwardPipelineEngine
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 252, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 461, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 498, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 243, in train_batch_iter
+    self.nb_microbatches >= pg.size() - 1
+AssertionError: Number of microbatches (1) must be at least PP_SIZE-1=7 when using the OneForwardOneBackwardPipelineEngine
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 252, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 252, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 461, in train
+    trainer.train(dataloader)    
+outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 461, in train
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 498, in training_step
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 498, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 243, in train_batch_iter
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 243, in train_batch_iter
+    self.nb_microbatches >= pg.size() - 1
+AssertionError: Number of microbatches (1) must be at least PP_SIZE-1=7 when using the OneForwardOneBackwardPipelineEngine
+    self.nb_microbatches >= pg.size() - 1
+AssertionError: Number of microbatches (1) must be at least PP_SIZE-1=7 when using the OneForwardOneBackwardPipelineEngine
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 252, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 461, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 498, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 243, in train_batch_iter
+    self.nb_microbatches >= pg.size() - 1
+AssertionError: Number of microbatches (1) must be at least PP_SIZE-1=7 when using the OneForwardOneBackwardPipelineEngine
+[2024-12-14 08:30:11,127] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 121508 closing signal SIGTERM
+[2024-12-14 08:30:11,127] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 121510 closing signal SIGTERM
+[2024-12-14 08:30:11,127] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 121511 closing signal SIGTERM
+[2024-12-14 08:30:11,127] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 121512 closing signal SIGTERM
+[2024-12-14 08:30:11,127] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 121513 closing signal SIGTERM
+[2024-12-14 08:30:11,127] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 32105 closing signal SIGTERM
+[2024-12-14 08:30:11,128] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 32106 closing signal SIGTERM
+[2024-12-14 08:30:11,128] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 32107 closing signal SIGTERM
+[2024-12-14 08:30:11,128] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 32108 closing signal SIGTERM
+[2024-12-14 08:30:11,128] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 32111 closing signal SIGTERM
+[2024-12-14 08:30:13,207] torch.distributed.elastic.multiprocessing.api: [ERROR] failed (exitcode: 1) local_rank: 0 (pid: 121507) of binary: /fsx/nouamane/miniconda/envs/2-1-cu121/bin/python
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 33, in <module>
+    sys.exit(load_entry_point('torch==2.1.1', 'console_scripts', 'torchrun')())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 264, in launch_agent
+    raise ChildFailedError(
+torch.distributed.elastic.multiprocessing.errors.ChildFailedError: 
+============================================================
+/fsx/nouamane/projects/nanotron/run_train.py FAILED
+------------------------------------------------------------
+Failures:
+[1]:
+  time      : 2024-12-14_08:30:11
+  host      : ip-26-0-171-21.ec2.internal
+  rank      : 10 (local_rank: 2)
+  exitcode  : 1 (pid: 121509)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[2]:
+  time      : 2024-12-14_08:30:11
+  host      : ip-26-0-171-21.ec2.internal
+  rank      : 15 (local_rank: 7)
+  exitcode  : 1 (pid: 121514)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+------------------------------------------------------------
+Root Cause (first observed failure):
+[0]:
+  time      : 2024-12-14_08:30:11
+  host      : ip-26-0-171-21.ec2.internal
+  rank      : 8 (local_rank: 0)
+  exitcode  : 1 (pid: 121507)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+============================================================
+srun: error: ip-26-0-171-21: task 1: Exited with exit code 1
+srun: launch/slurm: _step_signal: Terminating StepId=13442742.0
+slurmstepd: error: *** STEP 13442742.0 ON ip-26-0-169-132 CANCELLED AT 2024-12-14T08:30:13 ***
+[2024-12-14 08:30:13,613] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-14 08:30:13,613] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 32106 closing signal SIGTERM
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 33, in <module>
+    sys.exit(load_entry_point('torch==2.1.1', 'console_scripts', 'torchrun')())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 878, in _invoke_run
+    run_result = self._monitor_workers(self._worker_group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/local_elastic_agent.py", line 307, in _monitor_workers
+    result = self._pcontext.wait(0)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 288, in wait
+    return self._poll()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 668, in _poll
+    self.close()  # terminate all running procs
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 331, in close
+    self._close(death_sig=death_sig, timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 713, in _close
+    handler.proc.wait(time_to_wait)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1209, in wait
+    return self._wait(timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1953, in _wait
+    time.sleep(delay)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 32032 got signal: 15
+srun: error: ip-26-0-169-132: task 0: Exited with exit code 1
diff --git a/logs/13458892-bench_3.57G_dp8_tp8_pp2_acc4_mbs8_seq4096_zero1_tpmodeRED_vocab131k.out b/logs/13458892-bench_3.57G_dp8_tp8_pp2_acc4_mbs8_seq4096_zero1_tpmodeRED_vocab131k.out
new file mode 100644
index 0000000000000000000000000000000000000000..d6d7f165ba51da46d30a9f69a3cc70a651ab0320
--- /dev/null
+++ b/logs/13458892-bench_3.57G_dp8_tp8_pp2_acc4_mbs8_seq4096_zero1_tpmodeRED_vocab131k.out
@@ -0,0 +1,1222 @@
++ source /etc/profile.d/modules.sh
+++ . /usr/share/modules/init/bash
++++ unset _mlshdbg
++++ '[' 0 = 1 ']'
++++ unset _mlre _mlIFS
++++ '[' -n x ']'
++++ _mlIFS=' 	
+'
++++ IFS=' '
++++ '[' -n '' ']'
+++++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash autoinit
++++ _mlcode='module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+test 0;'
++++ _mlret=0
++++ '[' -n x ']'
++++ IFS=' 	
+'
++++ unset _mlIFS
++++ unset _mlre _mlv _mlrv
++++ '[' 0 -eq 0 ']'
++++ eval 'module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+test 0;'
+++++ MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl
+++++ export MODULES_CMD
+++++ ENV=/usr/share/modules/init/profile.sh
+++++ export ENV
+++++ BASH_ENV=/usr/share/modules/init/bash
+++++ export BASH_ENV
+++++ MODULESHOME=/usr/share/modules
+++++ export MODULESHOME
+++++ test 0
++++ '[' 0 = 1 ']'
++++ '[' -t 2 ']'
++++ export -f module
++++ export -f switchml
++++ '[' 5 -ge 3 ']'
++++ [[ ehxB =~ i ]]
++++ [[ ! :/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin: =~ :/usr/bin: ]]
++++ '[' '!' -n x ']'
+++++ manpath
++++ [[ ! :/admin/home/nouamane/.local/share/man:/fsx/nouamane/miniconda/envs/2-1-cu121/man:/fsx/nouamane/miniconda/envs/2-1-cu121/share/man:/fsx/nouamane/miniconda/man:/fsx/nouamane/miniconda/share/man:/opt/slurm/share/man:/opt/amazon/openmpi/share/man:/opt/amazon/efa/share/man:/usr/local/man:/usr/local/share/man:/usr/share/man:/admin/home/nouamane/.fzf/man:: =~ :/usr/share/man: ]]
++++ unset _mlcode _mlret
++++ '[' -n '' ']'
++ module load cuda/12.1
++ unset _mlshdbg
++ '[' 0 = 1 ']'
++ unset _mlre _mlIFS
++ '[' -n x ']'
++ _mlIFS=' 	
+'
++ IFS=' '
++ '[' -n '' ']'
+++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash load cuda/12.1
++ eval 'CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include;' export 'CPATH;
+LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/opt/aws-ofi-nccl/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/targets/x86_64-linux/lib/:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/lib:/usr/lib;' export 'LD_LIBRARY_PATH;
+FI_EFA_FORK_SAFE=1;' export 'FI_EFA_FORK_SAFE;
+MANPATH=/usr/local/cuda-12.1/share/man::;' export 'MANPATH;
+LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64;' export 'LIBRARY_PATH;
+_LMFILES_=/usr/share/modules/modulefiles/cuda/12.1;' export '_LMFILES_;
+LOADEDMODULES=cuda/12.1;' export 'LOADEDMODULES;
+MPI_PATH=/opt/amazon/openmpi;' export 'MPI_PATH;
+NCCL_HOME_modshare=/opt/nccl/build:1;' export 'NCCL_HOME_modshare;
+NCCL_PROTO=simple;' export 'NCCL_PROTO;
+MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1;' export 'MANPATH_modshare;
+LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1;' export 'LIBRARY_PATH_modshare;
+NCCL_SOCKET_IFNAME=enp;' export 'NCCL_SOCKET_IFNAME;
+AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl;' export 'AWS_OFI_NCCL_HOME;
+NCCL_HOME=/opt/nccl/build;' export 'NCCL_HOME;
+FI_PROVIDER=efa;' export 'FI_PROVIDER;
+AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1;' export 'AWS_OFI_NCCL_HOME_modshare;
+CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1;' export 'CPATH_modshare;
+LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib/:1:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2:/opt/aws-ofi-nccl/lib:2:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/efa/lib:1:/usr/lib:1;' export 'LD_LIBRARY_PATH_modshare;
+FI_EFA_ENABLE_SHM_TRANSFER=1;' export 'FI_EFA_ENABLE_SHM_TRANSFER;
+_LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1;' export '_LMFILES__modshare;
+LOADEDMODULES_modshare=cuda/12.1:1;' export 'LOADEDMODULES_modshare;
+MPI_PATH_modshare=/opt/amazon/openmpi:1;' export 'MPI_PATH_modshare;
+PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin;' export 'PATH;
+CUDA_HOME=/usr/local/cuda-12.1;' export 'CUDA_HOME;
+PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/usr/local/cuda-12.1/include:1:/opt/amazon/efa/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/usr/sbin:1:/fsx/nouamane/miniconda/bin:1:/opt/slurm/bin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:2:/opt/amazon/openmpi/bin:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1;' export 'PATH_modshare;
+test' '0;'
+++ CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include
+++ export CPATH
+++ LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/opt/aws-ofi-nccl/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/targets/x86_64-linux/lib/:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/lib:/usr/lib
+++ export LD_LIBRARY_PATH
+++ FI_EFA_FORK_SAFE=1
+++ export FI_EFA_FORK_SAFE
+++ MANPATH=/usr/local/cuda-12.1/share/man::
+++ export MANPATH
+++ LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64
+++ export LIBRARY_PATH
+++ _LMFILES_=/usr/share/modules/modulefiles/cuda/12.1
+++ export _LMFILES_
+++ LOADEDMODULES=cuda/12.1
+++ export LOADEDMODULES
+++ MPI_PATH=/opt/amazon/openmpi
+++ export MPI_PATH
+++ NCCL_HOME_modshare=/opt/nccl/build:1
+++ export NCCL_HOME_modshare
+++ NCCL_PROTO=simple
+++ export NCCL_PROTO
+++ MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1
+++ export MANPATH_modshare
+++ LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1
+++ export LIBRARY_PATH_modshare
+++ NCCL_SOCKET_IFNAME=enp
+++ export NCCL_SOCKET_IFNAME
+++ AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl
+++ export AWS_OFI_NCCL_HOME
+++ NCCL_HOME=/opt/nccl/build
+++ export NCCL_HOME
+++ FI_PROVIDER=efa
+++ export FI_PROVIDER
+++ AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1
+++ export AWS_OFI_NCCL_HOME_modshare
+++ CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1
+++ export CPATH_modshare
+++ LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib/:1:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2:/opt/aws-ofi-nccl/lib:2:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/efa/lib:1:/usr/lib:1
+++ export LD_LIBRARY_PATH_modshare
+++ FI_EFA_ENABLE_SHM_TRANSFER=1
+++ export FI_EFA_ENABLE_SHM_TRANSFER
+++ _LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1
+++ export _LMFILES__modshare
+++ LOADEDMODULES_modshare=cuda/12.1:1
+++ export LOADEDMODULES_modshare
+++ MPI_PATH_modshare=/opt/amazon/openmpi:1
+++ export MPI_PATH_modshare
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin
+++ export PATH
+++ CUDA_HOME=/usr/local/cuda-12.1
+++ export CUDA_HOME
+++ PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/usr/local/cuda-12.1/include:1:/opt/amazon/efa/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/usr/sbin:1:/fsx/nouamane/miniconda/bin:1:/opt/slurm/bin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:2:/opt/amazon/openmpi/bin:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1
+++ export PATH_modshare
+++ test 0
++ _mlstatus=0
++ '[' -n x ']'
++ IFS=' 	
+'
++ unset _mlre _mlv _mlrv _mlIFS
++ '[' -n '' ']'
++ unset _mlshdbg
++ return 0
++ source /fsx/nouamane/miniconda/bin/activate
+++ _CONDA_ROOT=/fsx/nouamane/miniconda
+++ . /fsx/nouamane/miniconda/etc/profile.d/conda.sh
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ '[' -z x ']'
+++ conda activate
+++ local cmd=activate
+++ case "$cmd" in
+++ __conda_activate activate
+++ '[' -n '' ']'
+++ local ask_conda
++++ PS1=
++++ __conda_exe shell.posix activate
++++ /fsx/nouamane/miniconda/bin/conda shell.posix activate
+++ ask_conda='PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin'\''
+export CONDA_SHLVL='\''1'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\'''
+++ eval 'PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin'\''
+export CONDA_SHLVL='\''1'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\'''
++++ PS1='(base) '
++++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin
++++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin
++++ export CONDA_SHLVL=1
++++ CONDA_SHLVL=1
++++ export 'CONDA_PROMPT_MODIFIER=(base) '
++++ CONDA_PROMPT_MODIFIER='(base) '
+++ __conda_hashr
+++ '[' -n '' ']'
+++ '[' -n '' ']'
+++ hash -r
++ conda activate 2-1-cu121
++ local cmd=activate
++ case "$cmd" in
++ __conda_activate activate 2-1-cu121
++ '[' -n '' ']'
++ local ask_conda
+++ PS1='(base) '
+++ __conda_exe shell.posix activate 2-1-cu121
+++ /fsx/nouamane/miniconda/bin/conda shell.posix activate 2-1-cu121
++ ask_conda='PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''2'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_1='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
++ eval 'PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''2'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_1='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
+++ PS1='(2-1-cu121) '
+++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin
+++ export CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ export CONDA_SHLVL=2
+++ CONDA_SHLVL=2
+++ export CONDA_DEFAULT_ENV=2-1-cu121
+++ CONDA_DEFAULT_ENV=2-1-cu121
+++ export 'CONDA_PROMPT_MODIFIER=(2-1-cu121) '
+++ CONDA_PROMPT_MODIFIER='(2-1-cu121) '
+++ export CONDA_PREFIX_1=/fsx/nouamane/miniconda
+++ CONDA_PREFIX_1=/fsx/nouamane/miniconda
+++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ export _CE_M=
+++ _CE_M=
+++ export _CE_CONDA=
+++ _CE_CONDA=
+++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh
++++ test -n ''
++++ xml_catalog_files_libxml2=
++++ XML_CATALOG_FILES=
++++ conda_catalog_files=
++++ ifs_libxml2=' 	
+'
++++ IFS=' '
++++ rem=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ for pre in ${rem}
++++ test '' = /fsx/nouamane/miniconda/envs/2-1-cu121
++++ conda_catalog_files=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ rem=
++++ IFS=' 	
+'
++++ conda_catalog_files='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ export 'XML_CATALOG_FILES=file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ XML_CATALOG_FILES='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ unset conda_catalog_files ifs_libxml2 rem
++ __conda_hashr
++ '[' -n '' ']'
++ '[' -n '' ']'
++ hash -r
++ export PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin
++ PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin
+++ scontrol show hostnames 'ip-26-0-160-[225,242],ip-26-0-161-[78,103,123],ip-26-0-162-14,ip-26-0-165-[38,59,131,164,202,213],ip-26-0-172-[116,142,147,252]'
++ export 'NODELIST=ip-26-0-160-225
+ip-26-0-160-242
+ip-26-0-161-78
+ip-26-0-161-103
+ip-26-0-161-123
+ip-26-0-162-14
+ip-26-0-165-38
+ip-26-0-165-59
+ip-26-0-165-131
+ip-26-0-165-164
+ip-26-0-165-202
+ip-26-0-165-213
+ip-26-0-172-116
+ip-26-0-172-142
+ip-26-0-172-147
+ip-26-0-172-252'
++ NODELIST='ip-26-0-160-225
+ip-26-0-160-242
+ip-26-0-161-78
+ip-26-0-161-103
+ip-26-0-161-123
+ip-26-0-162-14
+ip-26-0-165-38
+ip-26-0-165-59
+ip-26-0-165-131
+ip-26-0-165-164
+ip-26-0-165-202
+ip-26-0-165-213
+ip-26-0-172-116
+ip-26-0-172-142
+ip-26-0-172-147
+ip-26-0-172-252'
+++ head -n1
+++ scontrol show hostnames 'ip-26-0-160-[225,242],ip-26-0-161-[78,103,123],ip-26-0-162-14,ip-26-0-165-[38,59,131,164,202,213],ip-26-0-172-[116,142,147,252]'
++ export MASTER_NODE=ip-26-0-160-225
++ MASTER_NODE=ip-26-0-160-225
++ export MASTER_PORT=12356
++ MASTER_PORT=12356
++ export NNODES=16
++ NNODES=16
++ export GPUS_PER_NODE=8
++ GPUS_PER_NODE=8
++ export WORLD_SIZE=128
++ WORLD_SIZE=128
++ export CUDA_DEVICE_MAX_CONNECTIONS=1
++ CUDA_DEVICE_MAX_CONNECTIONS=1
++ export NCCL_DEBUG=WARN
++ NCCL_DEBUG=WARN
++ export NANOTRON_BENCHMARK=1
++ NANOTRON_BENCHMARK=1
++ export WANDB_MODE=disabled
++ WANDB_MODE=disabled
++ export TORCH_NCCL_ASYNC_ERROR_HANDLING=1
++ TORCH_NCCL_ASYNC_ERROR_HANDLING=1
++ echo '=== GPU Topology ==='
+=== GPU Topology ===
++ nvidia-smi topo -m
+	GPU0	GPU1	GPU2	GPU3	GPU4	GPU5	GPU6	GPU7	CPU Affinity	NUMA Affinity	GPU NUMA ID
+GPU0	 X 	NV18	NV18	NV18	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU1	NV18	 X 	NV18	NV18	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU2	NV18	NV18	 X 	NV18	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU3	NV18	NV18	NV18	 X 	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU4	NV18	NV18	NV18	NV18	 X 	NV18	NV18	NV18	48-95	1		N/A
+GPU5	NV18	NV18	NV18	NV18	NV18	 X 	NV18	NV18	48-95	1		N/A
+GPU6	NV18	NV18	NV18	NV18	NV18	NV18	 X 	NV18	48-95	1		N/A
+GPU7	NV18	NV18	NV18	NV18	NV18	NV18	NV18	 X 	48-95	1		N/A
+
+Legend:
+
+  X    = Self
+  SYS  = Connection traversing PCIe as well as the SMP interconnect between NUMA nodes (e.g., QPI/UPI)
+  NODE = Connection traversing PCIe as well as the interconnect between PCIe Host Bridges within a NUMA node
+  PHB  = Connection traversing PCIe as well as a PCIe Host Bridge (typically the CPU)
+  PXB  = Connection traversing multiple PCIe bridges (without traversing the PCIe Host Bridge)
+  PIX  = Connection traversing at most a single PCIe bridge
+  NV#  = Connection traversing a bonded set of # NVLinks
++ echo ==================
+==================
++ echo 'Master node: ip-26-0-160-225'
+Master node: ip-26-0-160-225
++ echo 'All nodes: ip-26-0-160-225
+ip-26-0-160-242
+ip-26-0-161-78
+ip-26-0-161-103
+ip-26-0-161-123
+ip-26-0-162-14
+ip-26-0-165-38
+ip-26-0-165-59
+ip-26-0-165-131
+ip-26-0-165-164
+ip-26-0-165-202
+ip-26-0-165-213
+ip-26-0-172-116
+ip-26-0-172-142
+ip-26-0-172-147
+ip-26-0-172-252'
+All nodes: ip-26-0-160-225
+ip-26-0-160-242
+ip-26-0-161-78
+ip-26-0-161-103
+ip-26-0-161-123
+ip-26-0-162-14
+ip-26-0-165-38
+ip-26-0-165-59
+ip-26-0-165-131
+ip-26-0-165-164
+ip-26-0-165-202
+ip-26-0-165-213
+ip-26-0-172-116
+ip-26-0-172-142
+ip-26-0-172-147
+ip-26-0-172-252
++ echo 'World size: 128'
+World size: 128
++ srun --wait=0 --kill-on-bad-exit=1 torchrun --nnodes=16 --nproc_per_node=8 --rdzv_id=13458892 --rdzv_backend=c10d --rdzv_endpoint=ip-26-0-160-225:12356 --max_restarts 0 --rdzv_conf timeout=60 /fsx/nouamane/projects/nanotron/run_train.py --config-file benchmark/configs/config_3.57G_dp8_tp8_pp2_acc4_mbs8_seq4096_zero1_tpmodeRED_vocab131k.yaml
+[2024-12-15 09:07:19,923] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-15 09:07:19,923] torch.distributed.run: [WARNING] 
+[2024-12-15 09:07:19,923] torch.distributed.run: [WARNING] *****************************************
+[2024-12-15 09:07:19,923] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-15 09:07:19,923] torch.distributed.run: [WARNING] *****************************************
+[2024-12-15 09:07:20,074] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-15 09:07:20,074] torch.distributed.run: [WARNING] 
+[2024-12-15 09:07:20,074] torch.distributed.run: [WARNING] *****************************************
+[2024-12-15 09:07:20,074] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-15 09:07:20,074] torch.distributed.run: [WARNING] *****************************************
+[2024-12-15 09:07:20,222] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-15 09:07:20,243] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-15 09:07:20,222] torch.distributed.run: [WARNING] 
+[2024-12-15 09:07:20,222] torch.distributed.run: [WARNING] *****************************************
+[2024-12-15 09:07:20,222] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-15 09:07:20,222] torch.distributed.run: [WARNING] *****************************************
+[2024-12-15 09:07:20,265] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-15 09:07:20,265] torch.distributed.run: [WARNING] 
+[2024-12-15 09:07:20,265] torch.distributed.run: [WARNING] *****************************************
+[2024-12-15 09:07:20,265] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-15 09:07:20,265] torch.distributed.run: [WARNING] *****************************************
+[2024-12-15 09:07:20,266] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-15 09:07:20,268] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-15 09:07:20,268] torch.distributed.run: [WARNING] 
+[2024-12-15 09:07:20,268] torch.distributed.run: [WARNING] *****************************************
+[2024-12-15 09:07:20,268] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-15 09:07:20,268] torch.distributed.run: [WARNING] *****************************************
+[2024-12-15 09:07:20,269] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-15 09:07:20,243] torch.distributed.run: [WARNING] 
+[2024-12-15 09:07:20,243] torch.distributed.run: [WARNING] *****************************************
+[2024-12-15 09:07:20,243] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-15 09:07:20,243] torch.distributed.run: [WARNING] *****************************************
+[2024-12-15 09:07:20,267] torch.distributed.run: [WARNING] 
+[2024-12-15 09:07:20,267] torch.distributed.run: [WARNING] *****************************************
+[2024-12-15 09:07:20,267] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-15 09:07:20,267] torch.distributed.run: [WARNING] *****************************************
+[2024-12-15 09:07:20,269] torch.distributed.run: [WARNING] 
+[2024-12-15 09:07:20,269] torch.distributed.run: [WARNING] *****************************************
+[2024-12-15 09:07:20,269] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-15 09:07:20,269] torch.distributed.run: [WARNING] *****************************************
+[2024-12-15 09:07:20,327] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-15 09:07:20,343] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-15 09:07:20,366] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-15 09:07:20,328] torch.distributed.run: [WARNING] 
+[2024-12-15 09:07:20,328] torch.distributed.run: [WARNING] *****************************************
+[2024-12-15 09:07:20,328] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-15 09:07:20,328] torch.distributed.run: [WARNING] *****************************************
+[2024-12-15 09:07:20,343] torch.distributed.run: [WARNING] 
+[2024-12-15 09:07:20,343] torch.distributed.run: [WARNING] *****************************************
+[2024-12-15 09:07:20,343] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-15 09:07:20,343] torch.distributed.run: [WARNING] *****************************************
+[2024-12-15 09:07:20,366] torch.distributed.run: [WARNING] 
+[2024-12-15 09:07:20,366] torch.distributed.run: [WARNING] *****************************************
+[2024-12-15 09:07:20,366] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-15 09:07:20,366] torch.distributed.run: [WARNING] *****************************************
+[2024-12-15 09:07:20,414] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-15 09:07:20,414] torch.distributed.run: [WARNING] 
+[2024-12-15 09:07:20,414] torch.distributed.run: [WARNING] *****************************************
+[2024-12-15 09:07:20,414] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-15 09:07:20,414] torch.distributed.run: [WARNING] *****************************************
+[2024-12-15 09:07:20,419] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-15 09:07:20,419] torch.distributed.run: [WARNING] 
+[2024-12-15 09:07:20,419] torch.distributed.run: [WARNING] *****************************************
+[2024-12-15 09:07:20,419] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-15 09:07:20,419] torch.distributed.run: [WARNING] *****************************************
+[2024-12-15 09:07:20,493] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-15 09:07:20,524] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-15 09:07:20,493] torch.distributed.run: [WARNING] 
+[2024-12-15 09:07:20,493] torch.distributed.run: [WARNING] *****************************************
+[2024-12-15 09:07:20,493] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-15 09:07:20,493] torch.distributed.run: [WARNING] *****************************************
+[2024-12-15 09:07:20,558] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-15 09:07:20,524] torch.distributed.run: [WARNING] 
+[2024-12-15 09:07:20,524] torch.distributed.run: [WARNING] *****************************************
+[2024-12-15 09:07:20,524] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-15 09:07:20,524] torch.distributed.run: [WARNING] *****************************************
+[2024-12-15 09:07:20,558] torch.distributed.run: [WARNING] 
+[2024-12-15 09:07:20,558] torch.distributed.run: [WARNING] *****************************************
+[2024-12-15 09:07:20,558] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-15 09:07:20,558] torch.distributed.run: [WARNING] *****************************************
+NCCL version 2.18.5+cuda12.2
+12/15/2024 09:07:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: Measuring inter-GPU and intra-node bandwidth...
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+12/15/2024 09:08:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: Bandwidth measurement complete. Time taken: 18.55 seconds
+12/15/2024 09:08:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: Config:
+12/15/2024 09:08:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: Config(general=GeneralArgs(project='debug',
+12/15/2024 09:08:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                            run='3.57G_dp8_tp8_pp2_acc4_mbs8_seq4096_zero1_tpmodeRED_vocab131k',
+12/15/2024 09:08:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                            seed=42,
+12/15/2024 09:08:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                            step=None,
+12/15/2024 09:08:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                            consumed_train_samples=None,
+12/15/2024 09:08:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                            benchmark_csv_path=PosixPath('benchmark/results/bench_final2.csv'),
+12/15/2024 09:08:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                            ignore_sanity_checks=True),
+12/15/2024 09:08:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:        parallelism=ParallelismArgs(dp=8,
+12/15/2024 09:08:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                    pp=2,
+12/15/2024 09:08:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                    tp=8,
+12/15/2024 09:08:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                    pp_engine=<nanotron.parallel.pipeline_parallel.engine.OneForwardOneBackwardPipelineEngine object at 0x7fd5a1c40c10>,
+12/15/2024 09:08:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                    tp_mode=<TensorParallelLinearMode.REDUCE_SCATTER: 2>,
+12/15/2024 09:08:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                    tp_linear_async_communication=True,
+12/15/2024 09:08:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                    recompute_layer=False,
+12/15/2024 09:08:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                    tp_recompute_allgather=True,
+12/15/2024 09:08:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                    expert_parallel_size=1),
+12/15/2024 09:08:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:        model=ModelArgs(model_config=LlamaConfig(bos_token_id=0,
+12/15/2024 09:08:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 eos_token_id=0,
+12/15/2024 09:08:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 hidden_act='silu',
+12/15/2024 09:08:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 hidden_size=3072,
+12/15/2024 09:08:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 initializer_range=0.02,
+12/15/2024 09:08:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 intermediate_size=8192,
+12/15/2024 09:08:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 is_llama_config=True,
+12/15/2024 09:08:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 max_position_embeddings=4096,
+12/15/2024 09:08:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 num_attention_heads=32,
+12/15/2024 09:08:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 num_hidden_layers=28,
+12/15/2024 09:08:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 num_key_value_heads=32,
+12/15/2024 09:08:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 pad_token_id=None,
+12/15/2024 09:08:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 pretraining_tp=1,
+12/15/2024 09:08:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 rms_norm_eps=1e-05,
+12/15/2024 09:08:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 rope_scaling=None,
+12/15/2024 09:08:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 rope_theta=10000.0,
+12/15/2024 09:08:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 rope_interleaved=False,
+12/15/2024 09:08:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 tie_word_embeddings=True,
+12/15/2024 09:08:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 use_cache=True,
+12/15/2024 09:08:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 vocab_size=131072),
+12/15/2024 09:08:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                        init_method=RandomInit(std=0.02),
+12/15/2024 09:08:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                        dtype=torch.bfloat16,
+12/15/2024 09:08:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                        make_vocab_size_divisible_by=1,
+12/15/2024 09:08:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                        ddp_bucket_cap_mb=25),
+12/15/2024 09:08:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:        tokenizer=TokenizerArgs(tokenizer_name_or_path='robot-test/dummy-tokenizer-wordlevel',
+12/15/2024 09:08:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                tokenizer_revision=None,
+12/15/2024 09:08:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                tokenizer_max_length=None),
+12/15/2024 09:08:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:        checkpoints=CheckpointsArgs(checkpoints_path=PosixPath('checkpoints'),
+12/15/2024 09:08:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                    checkpoint_interval=10000,
+12/15/2024 09:08:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                    save_initial_state=False,
+12/15/2024 09:08:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                    save_final_state=False,
+12/15/2024 09:08:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                    resume_checkpoint_path=None,
+12/15/2024 09:08:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                    checkpoints_path_is_shared_file_system=False),
+12/15/2024 09:08:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:        logging=LoggingArgs(log_level='info',
+12/15/2024 09:08:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                            log_level_replica='info',
+12/15/2024 09:08:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                            iteration_step_info_interval=1),
+12/15/2024 09:08:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:        tokens=TokensArgs(sequence_length=4096,
+12/15/2024 09:08:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                          train_steps=100,
+12/15/2024 09:08:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                          micro_batch_size=8,
+12/15/2024 09:08:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                          batch_accumulation_per_replica=4,
+12/15/2024 09:08:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                          val_check_interval=100,
+12/15/2024 09:08:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                          limit_val_batches=0,
+12/15/2024 09:08:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                          limit_test_batches=0),
+12/15/2024 09:08:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:        optimizer=OptimizerArgs(optimizer_factory=AdamWOptimizerArgs(adam_eps=1e-08,
+12/15/2024 09:08:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                                     adam_beta1=0.9,
+12/15/2024 09:08:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                                     adam_beta2=0.95,
+12/15/2024 09:08:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                                     torch_adam_is_fused=True,
+12/15/2024 09:08:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                                     name='adamW'),
+12/15/2024 09:08:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                zero_stage=1,
+12/15/2024 09:08:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                weight_decay=0.01,
+12/15/2024 09:08:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                clip_grad=1.0,
+12/15/2024 09:08:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                accumulate_grad_in_fp32=True,
+12/15/2024 09:08:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                learning_rate_scheduler=LRSchedulerArgs(learning_rate=0.0003,
+12/15/2024 09:08:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                                        lr_warmup_steps=2,
+12/15/2024 09:08:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                                        lr_warmup_style='linear',
+12/15/2024 09:08:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                                        lr_decay_style='cosine',
+12/15/2024 09:08:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                                        lr_decay_steps=13,
+12/15/2024 09:08:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                                        lr_decay_starting_step=None,
+12/15/2024 09:08:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                                        min_decay_lr=1e-05)),
+12/15/2024 09:08:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:        data_stages=[DatasetStageArgs(name='Stable Training Stage',
+12/15/2024 09:08:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                      start_training_step=1,
+12/15/2024 09:08:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                      data=DataArgs(dataset=None,
+12/15/2024 09:08:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                    seed=42,
+12/15/2024 09:08:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                    num_loading_workers=1))],
+12/15/2024 09:08:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:        profiler=None,
+12/15/2024 09:08:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:        lighteval=None,
+12/15/2024 09:08:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:        s3_upload=None)
+12/15/2024 09:08:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: Model Config:
+12/15/2024 09:08:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: LlamaConfig(bos_token_id=0,
+12/15/2024 09:08:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             eos_token_id=0,
+12/15/2024 09:08:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             hidden_act='silu',
+12/15/2024 09:08:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             hidden_size=3072,
+12/15/2024 09:08:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             initializer_range=0.02,
+12/15/2024 09:08:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             intermediate_size=8192,
+12/15/2024 09:08:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             is_llama_config=True,
+12/15/2024 09:08:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             max_position_embeddings=4096,
+12/15/2024 09:08:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             num_attention_heads=32,
+12/15/2024 09:08:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             num_hidden_layers=28,
+12/15/2024 09:08:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             num_key_value_heads=32,
+12/15/2024 09:08:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             pad_token_id=None,
+12/15/2024 09:08:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             pretraining_tp=1,
+12/15/2024 09:08:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             rms_norm_eps=1e-05,
+12/15/2024 09:08:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             rope_scaling=None,
+12/15/2024 09:08:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             rope_theta=10000.0,
+12/15/2024 09:08:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             rope_interleaved=False,
+12/15/2024 09:08:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             tie_word_embeddings=True,
+12/15/2024 09:08:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             use_cache=True,
+12/15/2024 09:08:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             vocab_size=131072)
+12/15/2024 09:08:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: Building model..
+12/15/2024 09:08:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: Initialize RoPE Theta = 10000.0
+12/15/2024 09:08:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: Setting PP block ranks...
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+12/15/2024 09:08:10 [INFO|DP=0|PP=0|TP=3|ip-26-0-160-225]: Local number of parameters: 277M (528.19MiB)
+12/15/2024 09:08:10 [INFO|DP=0|PP=1|TP=5|ip-26-0-165-202]: Local number of parameters: 220M (420.15MiB)
+12/15/2024 09:08:10 [INFO|DP=0|PP=1|TP=1|ip-26-0-165-202]: Local number of parameters: 220M (420.15MiB)
+12/15/2024 09:08:10 [INFO|DP=0|PP=0|TP=1|ip-26-0-160-225]: Local number of parameters: 277M (528.19MiB)
+12/15/2024 09:08:10 [INFO|DP=0|PP=0|TP=5|ip-26-0-160-225]: Local number of parameters: 277M (528.19MiB)
+12/15/2024 09:08:10 [INFO|DP=0|PP=1|TP=0|ip-26-0-165-202]: Local number of parameters: 220M (420.15MiB)
+12/15/2024 09:08:10 [INFO|DP=0|PP=0|TP=3|ip-26-0-160-225]: [After model building] Memory usage: 528.71MiB. Peak allocated: 5392.00MiB Peak reserved: 22806.00MiB
+12/15/2024 09:08:10 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: Total number of parameters: 3.98G (7586.67MiB)
+12/15/2024 09:08:10 [INFO|DP=0|PP=1|TP=3|ip-26-0-165-202]: Local number of parameters: 220M (420.15MiB)
+12/15/2024 09:08:10 [INFO|DP=0|PP=1|TP=5|ip-26-0-165-202]: [After model building] Memory usage: 420.67MiB. Peak allocated: 5392.00MiB Peak reserved: 22806.00MiB
+12/15/2024 09:08:10 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: Local number of parameters: 277M (528.19MiB)
+12/15/2024 09:08:10 [INFO|DP=0|PP=1|TP=1|ip-26-0-165-202]: [After model building] Memory usage: 420.67MiB. Peak allocated: 5392.00MiB Peak reserved: 22806.00MiB
+12/15/2024 09:08:10 [INFO|DP=0|PP=0|TP=1|ip-26-0-160-225]: [After model building] Memory usage: 528.71MiB. Peak allocated: 5392.00MiB Peak reserved: 22806.00MiB
+12/15/2024 09:08:10 [INFO|DP=0|PP=1|TP=0|ip-26-0-165-202]: [After model building] Memory usage: 420.67MiB. Peak allocated: 5392.00MiB Peak reserved: 22806.00MiB
+12/15/2024 09:08:10 [INFO|DP=0|PP=1|TP=7|ip-26-0-165-202]: Local number of parameters: 220M (420.15MiB)
+12/15/2024 09:08:10 [INFO|DP=0|PP=1|TP=3|ip-26-0-165-202]: [After model building] Memory usage: 420.67MiB. Peak allocated: 5392.00MiB Peak reserved: 22806.00MiB
+12/15/2024 09:08:10 [INFO|DP=0|PP=0|TP=5|ip-26-0-160-225]: [After model building] Memory usage: 528.71MiB. Peak allocated: 5392.00MiB Peak reserved: 22806.00MiB
+12/15/2024 09:08:10 [INFO|DP=0|PP=1|TP=2|ip-26-0-165-202]: Local number of parameters: 220M (420.15MiB)
+12/15/2024 09:08:10 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [After model building] Memory usage: 528.71MiB. Peak allocated: 5392.00MiB Peak reserved: 22806.00MiB
+12/15/2024 09:08:10 [INFO|DP=0|PP=0|TP=7|ip-26-0-160-225]: Local number of parameters: 277M (528.19MiB)
+12/15/2024 09:08:10 [INFO|DP=0|PP=0|TP=2|ip-26-0-160-225]: Local number of parameters: 277M (528.19MiB)
+12/15/2024 09:08:10 [INFO|DP=0|PP=1|TP=7|ip-26-0-165-202]: [After model building] Memory usage: 420.67MiB. Peak allocated: 5392.00MiB Peak reserved: 22806.00MiB
+12/15/2024 09:08:10 [INFO|DP=0|PP=1|TP=2|ip-26-0-165-202]: [After model building] Memory usage: 420.67MiB. Peak allocated: 5392.00MiB Peak reserved: 22806.00MiB
+12/15/2024 09:08:10 [INFO|DP=0|PP=0|TP=4|ip-26-0-160-225]: Local number of parameters: 277M (528.19MiB)
+12/15/2024 09:08:10 [INFO|DP=0|PP=1|TP=4|ip-26-0-165-202]: Local number of parameters: 220M (420.15MiB)
+12/15/2024 09:08:10 [INFO|DP=0|PP=0|TP=6|ip-26-0-160-225]: Local number of parameters: 277M (528.19MiB)
+12/15/2024 09:08:10 [INFO|DP=0|PP=0|TP=7|ip-26-0-160-225]: [After model building] Memory usage: 528.71MiB. Peak allocated: 5392.00MiB Peak reserved: 22806.00MiB
+12/15/2024 09:08:10 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: No checkpoint path provided.
+12/15/2024 09:08:10 [INFO|DP=0|PP=0|TP=2|ip-26-0-160-225]: [After model building] Memory usage: 528.71MiB. Peak allocated: 5392.00MiB Peak reserved: 22806.00MiB
+12/15/2024 09:08:10 [INFO|DP=0|PP=1|TP=6|ip-26-0-165-202]: Local number of parameters: 220M (420.15MiB)
+12/15/2024 09:08:10 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: Parametrizing model parameters using StandardParametrizator
+12/15/2024 09:08:10 [INFO|DP=0|PP=0|TP=4|ip-26-0-160-225]: [After model building] Memory usage: 528.71MiB. Peak allocated: 5392.00MiB Peak reserved: 22806.00MiB
+12/15/2024 09:08:10 [INFO|DP=0|PP=1|TP=4|ip-26-0-165-202]: [After model building] Memory usage: 420.67MiB. Peak allocated: 5392.00MiB Peak reserved: 22806.00MiB
+12/15/2024 09:08:10 [INFO|DP=0|PP=0|TP=6|ip-26-0-160-225]: [After model building] Memory usage: 528.71MiB. Peak allocated: 5392.00MiB Peak reserved: 22806.00MiB
+12/15/2024 09:08:10 [INFO|DP=0|PP=1|TP=6|ip-26-0-165-202]: [After model building] Memory usage: 420.67MiB. Peak allocated: 5392.00MiB Peak reserved: 22806.00MiB
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+12/15/2024 09:08:12 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [Optimizer Building] Using LearningRateForSP as learning rate
+12/15/2024 09:08:12 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [ZeRO sharding] Size of optimizer params per rank:
+12/15/2024 09:08:12 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [ZeRO sharding] DP Rank 0 has 34.6M out of 277M (12.50%) params' optimizer states
+12/15/2024 09:08:12 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [ZeRO sharding] DP Rank 1 has 34.6M out of 277M (12.50%) params' optimizer states
+12/15/2024 09:08:12 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [ZeRO sharding] DP Rank 2 has 34.6M out of 277M (12.50%) params' optimizer states
+12/15/2024 09:08:12 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [ZeRO sharding] DP Rank 3 has 34.6M out of 277M (12.50%) params' optimizer states
+12/15/2024 09:08:12 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [ZeRO sharding] DP Rank 4 has 34.6M out of 277M (12.50%) params' optimizer states
+12/15/2024 09:08:12 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [ZeRO sharding] DP Rank 5 has 34.6M out of 277M (12.50%) params' optimizer states
+12/15/2024 09:08:12 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [ZeRO sharding] DP Rank 6 has 34.6M out of 277M (12.50%) params' optimizer states
+12/15/2024 09:08:12 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [ZeRO sharding] DP Rank 7 has 34.6M out of 277M (12.50%) params' optimizer states
+12/15/2024 09:08:12 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [Training Plan] Stage Stable Training Stage has 99 remaining training steps and has consumed 0 samples
+12/15/2024 09:08:12 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: Using dummy data generator
+12/15/2024 09:08:12 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [Training Plan] There are 1 training stages 
+12/15/2024 09:08:12 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [Stage Stable Training Stage] start from step 1 
+12/15/2024 09:08:12 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: 
+12/15/2024 09:08:12 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [Start training] datetime: 2024-12-15 09:08:12.574199 | mbs: 8 | grad_accum: 4 | global_batch_size: 256 | sequence_length: 4096 | train_steps: 100 | start_iteration_step: 0 | consumed_train_samples: 0
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+12/15/2024 09:08:12 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: Resuming training from stage Stable Training Stage, it has trained for 0 samples and has 99 remaining train steps
+12/15/2024 09:08:12 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:  Memory usage: 1717.13MiB. Peak allocated 5392.00MiB. Peak reserved: 23864.00MiB
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+slurmstepd: error: *** STEP 13458892.0 ON ip-26-0-160-225 CANCELLED AT 2024-12-15T09:17:38 DUE TO TIME LIMIT ***
+slurmstepd: error: *** JOB 13458892 ON ip-26-0-160-225 CANCELLED AT 2024-12-15T09:17:38 DUE TO TIME LIMIT ***
+srun: Job step aborted: Waiting up to 32 seconds for job step to finish.
+[2024-12-15 09:17:38,568] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-15 09:17:38,568] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 211067 closing signal SIGTERM
+[2024-12-15 09:17:38,569] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 211068 closing signal SIGTERM
+[2024-12-15 09:17:38,569] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-15 09:17:38,570] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-15 09:17:38,570] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-15 09:17:38,570] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 45157 closing signal SIGTERM
+[2024-12-15 09:17:38,570] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 191897 closing signal SIGTERM
+[2024-12-15 09:17:38,570] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 400038 closing signal SIGTERM
+[2024-12-15 09:17:38,569] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-15 09:17:38,570] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 191898 closing signal SIGTERM
+[2024-12-15 09:17:38,570] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 45158 closing signal SIGTERM
+[2024-12-15 09:17:38,570] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 191899 closing signal SIGTERM
+[2024-12-15 09:17:38,570] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 400039 closing signal SIGTERM
+[2024-12-15 09:17:38,570] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 197010 closing signal SIGTERM
+[2024-12-15 09:17:38,570] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 45159 closing signal SIGTERM
+[2024-12-15 09:17:38,570] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 45160 closing signal SIGTERM
+[2024-12-15 09:17:38,570] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 191900 closing signal SIGTERM
+[2024-12-15 09:17:38,571] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 400040 closing signal SIGTERM
+[2024-12-15 09:17:38,570] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 197011 closing signal SIGTERM
+[2024-12-15 09:17:38,570] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 197012 closing signal SIGTERM
+[2024-12-15 09:17:38,570] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 197013 closing signal SIGTERM
+[2024-12-15 09:17:38,570] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-15 09:17:38,571] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 230368 closing signal SIGTERM
+[2024-12-15 09:17:38,569] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-15 09:17:38,570] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-15 09:17:38,571] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 230369 closing signal SIGTERM
+[2024-12-15 09:17:38,571] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 230370 closing signal SIGTERM
+[2024-12-15 09:17:38,569] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 90811 closing signal SIGTERM
+[2024-12-15 09:17:38,571] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 25285 closing signal SIGTERM
+[2024-12-15 09:17:38,571] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 25286 closing signal SIGTERM
+[2024-12-15 09:17:38,570] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 90812 closing signal SIGTERM
+[2024-12-15 09:17:38,571] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 25287 closing signal SIGTERM
+[2024-12-15 09:17:38,570] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-15 09:17:38,571] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 25288 closing signal SIGTERM
+[2024-12-15 09:17:38,571] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 276012 closing signal SIGTERM
+[2024-12-15 09:17:38,571] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 25289 closing signal SIGTERM
+[2024-12-15 09:17:38,571] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 276013 closing signal SIGTERM
+[2024-12-15 09:17:38,571] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 276014 closing signal SIGTERM
+[2024-12-15 09:17:38,570] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-15 09:17:38,571] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 230371 closing signal SIGTERM
+[2024-12-15 09:17:38,571] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 276015 closing signal SIGTERM
+[2024-12-15 09:17:38,571] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 230372 closing signal SIGTERM
+[2024-12-15 09:17:38,571] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 76991 closing signal SIGTERM
+[2024-12-15 09:17:38,571] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-15 09:17:38,571] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 25290 closing signal SIGTERM
+[2024-12-15 09:17:38,571] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 76992 closing signal SIGTERM
+[2024-12-15 09:17:38,571] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 25291 closing signal SIGTERM
+[2024-12-15 09:17:38,572] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 84222 closing signal SIGTERM
+[2024-12-15 09:17:38,571] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 76993 closing signal SIGTERM
+[2024-12-15 09:17:38,571] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 25292 closing signal SIGTERM
+[2024-12-15 09:17:38,572] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 84223 closing signal SIGTERM
+[2024-12-15 09:17:38,572] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 84224 closing signal SIGTERM
+[2024-12-15 09:17:38,571] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 76994 closing signal SIGTERM
+[2024-12-15 09:17:38,572] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 84225 closing signal SIGTERM
+[2024-12-15 09:17:38,571] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 400041 closing signal SIGTERM
+[2024-12-15 09:17:38,569] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 211069 closing signal SIGTERM
+[2024-12-15 09:17:38,569] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 211070 closing signal SIGTERM
+[2024-12-15 09:17:38,571] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 276016 closing signal SIGTERM
+[2024-12-15 09:17:38,571] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 276017 closing signal SIGTERM
+[2024-12-15 09:17:38,570] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 90813 closing signal SIGTERM
+[2024-12-15 09:17:38,570] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 90814 closing signal SIGTERM
+[2024-12-15 09:17:38,571] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 90815 closing signal SIGTERM
+[2024-12-15 09:17:38,572] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 400042 closing signal SIGTERM
+[2024-12-15 09:17:38,573] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 400043 closing signal SIGTERM
+[2024-12-15 09:17:38,573] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 400044 closing signal SIGTERM
+[2024-12-15 09:17:38,573] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 400045 closing signal SIGTERM
+[2024-12-15 09:17:38,573] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 230373 closing signal SIGTERM
+[2024-12-15 09:17:38,572] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 191901 closing signal SIGTERM
+[2024-12-15 09:17:38,571] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 90816 closing signal SIGTERM
+[2024-12-15 09:17:38,571] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 90817 closing signal SIGTERM
+[2024-12-15 09:17:38,571] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 90818 closing signal SIGTERM
+[2024-12-15 09:17:38,572] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 191902 closing signal SIGTERM
+[2024-12-15 09:17:38,572] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 197014 closing signal SIGTERM
+[2024-12-15 09:17:38,572] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 191903 closing signal SIGTERM
+[2024-12-15 09:17:38,571] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 211071 closing signal SIGTERM
+[2024-12-15 09:17:38,571] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 211072 closing signal SIGTERM
+[2024-12-15 09:17:38,571] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 76995 closing signal SIGTERM
+[2024-12-15 09:17:38,571] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 76996 closing signal SIGTERM
+[2024-12-15 09:17:38,572] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 76997 closing signal SIGTERM
+[2024-12-15 09:17:38,572] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 211073 closing signal SIGTERM
+[2024-12-15 09:17:38,572] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 197015 closing signal SIGTERM
+[2024-12-15 09:17:38,572] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 197016 closing signal SIGTERM
+[2024-12-15 09:17:38,573] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 230374 closing signal SIGTERM
+[2024-12-15 09:17:38,573] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 230375 closing signal SIGTERM
+[2024-12-15 09:17:38,574] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 84226 closing signal SIGTERM
+[2024-12-15 09:17:38,570] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 45161 closing signal SIGTERM
+[2024-12-15 09:17:38,574] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 84227 closing signal SIGTERM
+[2024-12-15 09:17:38,574] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 84228 closing signal SIGTERM
+[2024-12-15 09:17:38,572] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 197017 closing signal SIGTERM
+[2024-12-15 09:17:38,574] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 84229 closing signal SIGTERM
+[2024-12-15 09:17:38,573] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 45162 closing signal SIGTERM
+[2024-12-15 09:17:38,572] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 276018 closing signal SIGTERM
+[2024-12-15 09:17:38,572] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 276019 closing signal SIGTERM
+[2024-12-15 09:17:38,572] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 211074 closing signal SIGTERM
+[2024-12-15 09:17:38,574] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 76998 closing signal SIGTERM
+[2024-12-15 09:17:38,574] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 45163 closing signal SIGTERM
+[2024-12-15 09:17:38,572] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 211067 closing signal SIGTERM
+[2024-12-15 09:17:38,572] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 211068 closing signal SIGTERM
+[2024-12-15 09:17:38,575] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 45157 closing signal SIGTERM
+[2024-12-15 09:17:38,575] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 45158 closing signal SIGTERM
+[2024-12-15 09:17:38,574] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 211069 closing signal SIGTERM
+[2024-12-15 09:17:38,574] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 191904 closing signal SIGTERM
+[2024-12-15 09:17:38,575] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 45159 closing signal SIGTERM
+[2024-12-15 09:17:38,574] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 211070 closing signal SIGTERM
+[2024-12-15 09:17:38,574] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 211071 closing signal SIGTERM
+[2024-12-15 09:17:38,575] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 45160 closing signal SIGTERM
+[2024-12-15 09:17:38,575] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 45161 closing signal SIGTERM
+[2024-12-15 09:17:38,575] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 45162 closing signal SIGTERM
+[2024-12-15 09:17:38,574] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 211072 closing signal SIGTERM
+[2024-12-15 09:17:38,574] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 211073 closing signal SIGTERM
+[2024-12-15 09:17:38,575] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 45163 closing signal SIGTERM
+[2024-12-15 09:17:38,575] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 45164 closing signal SIGTERM
+[2024-12-15 09:17:38,574] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 211074 closing signal SIGTERM
+[2024-12-15 09:17:38,575] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 197010 closing signal SIGTERM
+[2024-12-15 09:17:38,575] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 90811 closing signal SIGTERM
+[2024-12-15 09:17:38,571] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-15 09:17:38,575] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 197011 closing signal SIGTERM
+[2024-12-15 09:17:38,575] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 90812 closing signal SIGTERM
+[2024-12-15 09:17:38,575] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 197012 closing signal SIGTERM
+[2024-12-15 09:17:38,575] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 90813 closing signal SIGTERM
+[2024-12-15 09:17:38,575] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 90814 closing signal SIGTERM
+[2024-12-15 09:17:38,575] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 90815 closing signal SIGTERM
+[2024-12-15 09:17:38,575] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 197013 closing signal SIGTERM
+[2024-12-15 09:17:38,575] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 90816 closing signal SIGTERM
+[2024-12-15 09:17:38,576] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 197014 closing signal SIGTERM
+[2024-12-15 09:17:38,576] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 197015 closing signal SIGTERM
+[2024-12-15 09:17:38,575] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 20890 closing signal SIGTERM
+[2024-12-15 09:17:38,575] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 90817 closing signal SIGTERM
+[2024-12-15 09:17:38,575] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 90818 closing signal SIGTERM
+[2024-12-15 09:17:38,576] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 197016 closing signal SIGTERM
+[2024-12-15 09:17:38,576] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 197017 closing signal SIGTERM
+[2024-12-15 09:17:38,575] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 20891 closing signal SIGTERM
+[2024-12-15 09:17:38,576] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 20892 closing signal SIGTERM
+[2024-12-15 09:17:38,576] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 20893 closing signal SIGTERM
+[2024-12-15 09:17:38,576] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 20894 closing signal SIGTERM
+[2024-12-15 09:17:38,576] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 20895 closing signal SIGTERM
+[2024-12-15 09:17:38,576] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 20896 closing signal SIGTERM
+[2024-12-15 09:17:38,576] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 20897 closing signal SIGTERM
+[2024-12-15 09:17:38,573] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-15 09:17:38,572] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-15 09:17:38,578] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 79589 closing signal SIGTERM
+[2024-12-15 09:17:38,577] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 36520 closing signal SIGTERM
+[2024-12-15 09:17:38,578] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 79590 closing signal SIGTERM
+[2024-12-15 09:17:38,578] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 36521 closing signal SIGTERM
+[2024-12-15 09:17:38,578] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 79591 closing signal SIGTERM
+[2024-12-15 09:17:38,578] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 36522 closing signal SIGTERM
+[2024-12-15 09:17:38,578] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 79592 closing signal SIGTERM
+[2024-12-15 09:17:38,578] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 79593 closing signal SIGTERM
+[2024-12-15 09:17:38,578] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 79594 closing signal SIGTERM
+[2024-12-15 09:17:38,578] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 36523 closing signal SIGTERM
+[2024-12-15 09:17:38,578] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 36524 closing signal SIGTERM
+[2024-12-15 09:17:38,578] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 36525 closing signal SIGTERM
+[2024-12-15 09:17:38,578] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 36526 closing signal SIGTERM
+[2024-12-15 09:17:38,578] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 79595 closing signal SIGTERM
+[2024-12-15 09:17:38,579] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 79596 closing signal SIGTERM
+[2024-12-15 09:17:38,578] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 36527 closing signal SIGTERM
+[2024-12-15 09:17:38,576] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-15 09:17:38,585] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 99279 closing signal SIGTERM
+[2024-12-15 09:17:38,585] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 99280 closing signal SIGTERM
+[2024-12-15 09:17:38,586] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 99281 closing signal SIGTERM
+[2024-12-15 09:17:38,586] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 99282 closing signal SIGTERM
+[2024-12-15 09:17:38,586] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 99283 closing signal SIGTERM
+[2024-12-15 09:17:38,586] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 99284 closing signal SIGTERM
+[2024-12-15 09:17:38,586] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 99285 closing signal SIGTERM
+[2024-12-15 09:17:38,586] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 99286 closing signal SIGTERM
+[2024-12-15 09:17:38,588] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-15 09:17:38,590] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 122705 closing signal SIGTERM
+[2024-12-15 09:17:38,590] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 122706 closing signal SIGTERM
+[2024-12-15 09:17:38,590] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 122707 closing signal SIGTERM
+[2024-12-15 09:17:38,590] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 122708 closing signal SIGTERM
+[2024-12-15 09:17:38,590] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 122709 closing signal SIGTERM
+[2024-12-15 09:17:38,590] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 122710 closing signal SIGTERM
+[2024-12-15 09:17:38,590] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 122711 closing signal SIGTERM
+[2024-12-15 09:17:38,590] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 122712 closing signal SIGTERM
+[2024-12-15 09:17:38,595] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 79589 closing signal SIGTERM
+[2024-12-15 09:17:38,595] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 79590 closing signal SIGTERM
+[2024-12-15 09:17:38,595] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 79591 closing signal SIGTERM
+[2024-12-15 09:17:38,595] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 79592 closing signal SIGTERM
+[2024-12-15 09:17:38,595] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 79593 closing signal SIGTERM
+[2024-12-15 09:17:38,595] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 79594 closing signal SIGTERM
+[2024-12-15 09:17:38,595] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 79595 closing signal SIGTERM
+[2024-12-15 09:17:38,595] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 79596 closing signal SIGTERM
+[2024-12-15 09:17:38,612] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 25285 closing signal SIGTERM
+[2024-12-15 09:17:38,612] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 25286 closing signal SIGTERM
+[2024-12-15 09:17:38,612] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 25287 closing signal SIGTERM
+[2024-12-15 09:17:38,612] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 25288 closing signal SIGTERM
+[2024-12-15 09:17:38,612] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 25289 closing signal SIGTERM
+[2024-12-15 09:17:38,612] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 25290 closing signal SIGTERM
+[2024-12-15 09:17:38,612] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 25291 closing signal SIGTERM
+[2024-12-15 09:17:38,612] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 25292 closing signal SIGTERM
+[2024-12-15 09:17:38,619] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 36520 closing signal SIGTERM
+[2024-12-15 09:17:38,619] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 36521 closing signal SIGTERM
+[2024-12-15 09:17:38,620] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 36522 closing signal SIGTERM
+[2024-12-15 09:17:38,620] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 36523 closing signal SIGTERM
+[2024-12-15 09:17:38,620] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 36524 closing signal SIGTERM
+[2024-12-15 09:17:38,620] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 36525 closing signal SIGTERM
+[2024-12-15 09:17:38,620] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 36526 closing signal SIGTERM
+[2024-12-15 09:17:38,620] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 36527 closing signal SIGTERM
+[2024-12-15 09:17:38,650] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 76991 closing signal SIGTERM
+[2024-12-15 09:17:38,650] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 76992 closing signal SIGTERM
+[2024-12-15 09:17:38,650] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 76993 closing signal SIGTERM
+[2024-12-15 09:17:38,650] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 76994 closing signal SIGTERM
+[2024-12-15 09:17:38,650] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 76995 closing signal SIGTERM
+[2024-12-15 09:17:38,650] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 76996 closing signal SIGTERM
+[2024-12-15 09:17:38,650] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 76997 closing signal SIGTERM
+[2024-12-15 09:17:38,650] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 76998 closing signal SIGTERM
+[2024-12-15 09:17:38,694] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 230368 closing signal SIGTERM
+[2024-12-15 09:17:38,694] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 230369 closing signal SIGTERM
+[2024-12-15 09:17:38,694] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 230370 closing signal SIGTERM
+[2024-12-15 09:17:38,694] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 230371 closing signal SIGTERM
+[2024-12-15 09:17:38,694] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 230372 closing signal SIGTERM
+[2024-12-15 09:17:38,695] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 230373 closing signal SIGTERM
+[2024-12-15 09:17:38,695] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 230374 closing signal SIGTERM
+[2024-12-15 09:17:38,695] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 230375 closing signal SIGTERM
+[2024-12-15 09:17:38,713] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 84222 closing signal SIGTERM
+[2024-12-15 09:17:38,713] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 84223 closing signal SIGTERM
+[2024-12-15 09:17:38,713] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 84224 closing signal SIGTERM
+[2024-12-15 09:17:38,713] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 84225 closing signal SIGTERM
+[2024-12-15 09:17:38,713] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 84226 closing signal SIGTERM
+[2024-12-15 09:17:38,713] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 84227 closing signal SIGTERM
+[2024-12-15 09:17:38,713] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 84228 closing signal SIGTERM
+[2024-12-15 09:17:38,713] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 84229 closing signal SIGTERM
+[2024-12-15 09:17:38,733] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 400038 closing signal SIGTERM
+[2024-12-15 09:17:38,733] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 400039 closing signal SIGTERM
+[2024-12-15 09:17:38,733] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 400040 closing signal SIGTERM
+[2024-12-15 09:17:38,733] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 400041 closing signal SIGTERM
+[2024-12-15 09:17:38,733] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 400042 closing signal SIGTERM
+[2024-12-15 09:17:38,733] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 400043 closing signal SIGTERM
+[2024-12-15 09:17:38,733] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 400044 closing signal SIGTERM
+[2024-12-15 09:17:38,733] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 400045 closing signal SIGTERM
+[2024-12-15 09:17:38,804] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 191897 closing signal SIGTERM
+[2024-12-15 09:17:38,804] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 191898 closing signal SIGTERM
+[2024-12-15 09:17:38,804] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 191899 closing signal SIGTERM
+[2024-12-15 09:17:38,804] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 191900 closing signal SIGTERM
+[2024-12-15 09:17:38,804] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 191901 closing signal SIGTERM
+[2024-12-15 09:17:38,804] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 191902 closing signal SIGTERM
+[2024-12-15 09:17:38,804] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 191903 closing signal SIGTERM
+[2024-12-15 09:17:38,804] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 191904 closing signal SIGTERM
diff --git a/logs/13458915-bench_1.34G_dp32_tp2_pp2_acc1_mbs8_seq4096_zero0_tpmodeRED_vocab131k.out b/logs/13458915-bench_1.34G_dp32_tp2_pp2_acc1_mbs8_seq4096_zero0_tpmodeRED_vocab131k.out
new file mode 100644
index 0000000000000000000000000000000000000000..25db596635c99f105174c10e327d7a1780e737bd
--- /dev/null
+++ b/logs/13458915-bench_1.34G_dp32_tp2_pp2_acc1_mbs8_seq4096_zero0_tpmodeRED_vocab131k.out
@@ -0,0 +1,1967 @@
++ source /etc/profile.d/modules.sh
+++ . /usr/share/modules/init/bash
++++ unset _mlshdbg
++++ '[' 0 = 1 ']'
++++ unset _mlre _mlIFS
++++ '[' -n x ']'
++++ _mlIFS=' 	
+'
++++ IFS=' '
++++ '[' -n '' ']'
+++++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash autoinit
++++ _mlcode='module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+test 0;'
++++ _mlret=0
++++ '[' -n x ']'
++++ IFS=' 	
+'
++++ unset _mlIFS
++++ unset _mlre _mlv _mlrv
++++ '[' 0 -eq 0 ']'
++++ eval 'module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+test 0;'
+++++ MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl
+++++ export MODULES_CMD
+++++ ENV=/usr/share/modules/init/profile.sh
+++++ export ENV
+++++ BASH_ENV=/usr/share/modules/init/bash
+++++ export BASH_ENV
+++++ MODULESHOME=/usr/share/modules
+++++ export MODULESHOME
+++++ test 0
++++ '[' 0 = 1 ']'
++++ '[' -t 2 ']'
++++ export -f module
++++ export -f switchml
++++ '[' 5 -ge 3 ']'
++++ [[ ehxB =~ i ]]
++++ [[ ! :/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin: =~ :/usr/bin: ]]
++++ '[' '!' -n x ']'
+++++ manpath
++++ [[ ! :/admin/home/nouamane/.local/share/man:/fsx/nouamane/miniconda/envs/2-1-cu121/man:/fsx/nouamane/miniconda/envs/2-1-cu121/share/man:/fsx/nouamane/miniconda/man:/fsx/nouamane/miniconda/share/man:/opt/slurm/share/man:/opt/amazon/openmpi/share/man:/opt/amazon/efa/share/man:/usr/local/man:/usr/local/share/man:/usr/share/man:/admin/home/nouamane/.fzf/man:: =~ :/usr/share/man: ]]
++++ unset _mlcode _mlret
++++ '[' -n '' ']'
++ module load cuda/12.1
++ unset _mlshdbg
++ '[' 0 = 1 ']'
++ unset _mlre _mlIFS
++ '[' -n x ']'
++ _mlIFS=' 	
+'
++ IFS=' '
++ '[' -n '' ']'
+++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash load cuda/12.1
++ eval 'CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include;' export 'CPATH;
+LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/opt/aws-ofi-nccl/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/targets/x86_64-linux/lib/:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/lib:/usr/lib;' export 'LD_LIBRARY_PATH;
+FI_EFA_FORK_SAFE=1;' export 'FI_EFA_FORK_SAFE;
+MANPATH=/usr/local/cuda-12.1/share/man::;' export 'MANPATH;
+LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64;' export 'LIBRARY_PATH;
+_LMFILES_=/usr/share/modules/modulefiles/cuda/12.1;' export '_LMFILES_;
+LOADEDMODULES=cuda/12.1;' export 'LOADEDMODULES;
+MPI_PATH=/opt/amazon/openmpi;' export 'MPI_PATH;
+NCCL_HOME_modshare=/opt/nccl/build:1;' export 'NCCL_HOME_modshare;
+NCCL_PROTO=simple;' export 'NCCL_PROTO;
+MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1;' export 'MANPATH_modshare;
+LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1;' export 'LIBRARY_PATH_modshare;
+NCCL_SOCKET_IFNAME=enp;' export 'NCCL_SOCKET_IFNAME;
+AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl;' export 'AWS_OFI_NCCL_HOME;
+NCCL_HOME=/opt/nccl/build;' export 'NCCL_HOME;
+FI_PROVIDER=efa;' export 'FI_PROVIDER;
+AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1;' export 'AWS_OFI_NCCL_HOME_modshare;
+CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1;' export 'CPATH_modshare;
+LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib/:1:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2:/opt/aws-ofi-nccl/lib:2:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/efa/lib:1:/usr/lib:1;' export 'LD_LIBRARY_PATH_modshare;
+FI_EFA_ENABLE_SHM_TRANSFER=1;' export 'FI_EFA_ENABLE_SHM_TRANSFER;
+_LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1;' export '_LMFILES__modshare;
+LOADEDMODULES_modshare=cuda/12.1:1;' export 'LOADEDMODULES_modshare;
+MPI_PATH_modshare=/opt/amazon/openmpi:1;' export 'MPI_PATH_modshare;
+PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin;' export 'PATH;
+CUDA_HOME=/usr/local/cuda-12.1;' export 'CUDA_HOME;
+PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/usr/local/cuda-12.1/include:1:/opt/amazon/efa/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/usr/sbin:1:/fsx/nouamane/miniconda/bin:1:/opt/slurm/bin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:2:/opt/amazon/openmpi/bin:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1;' export 'PATH_modshare;
+test' '0;'
+++ CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include
+++ export CPATH
+++ LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/opt/aws-ofi-nccl/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/targets/x86_64-linux/lib/:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/lib:/usr/lib
+++ export LD_LIBRARY_PATH
+++ FI_EFA_FORK_SAFE=1
+++ export FI_EFA_FORK_SAFE
+++ MANPATH=/usr/local/cuda-12.1/share/man::
+++ export MANPATH
+++ LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64
+++ export LIBRARY_PATH
+++ _LMFILES_=/usr/share/modules/modulefiles/cuda/12.1
+++ export _LMFILES_
+++ LOADEDMODULES=cuda/12.1
+++ export LOADEDMODULES
+++ MPI_PATH=/opt/amazon/openmpi
+++ export MPI_PATH
+++ NCCL_HOME_modshare=/opt/nccl/build:1
+++ export NCCL_HOME_modshare
+++ NCCL_PROTO=simple
+++ export NCCL_PROTO
+++ MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1
+++ export MANPATH_modshare
+++ LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1
+++ export LIBRARY_PATH_modshare
+++ NCCL_SOCKET_IFNAME=enp
+++ export NCCL_SOCKET_IFNAME
+++ AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl
+++ export AWS_OFI_NCCL_HOME
+++ NCCL_HOME=/opt/nccl/build
+++ export NCCL_HOME
+++ FI_PROVIDER=efa
+++ export FI_PROVIDER
+++ AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1
+++ export AWS_OFI_NCCL_HOME_modshare
+++ CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1
+++ export CPATH_modshare
+++ LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib/:1:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2:/opt/aws-ofi-nccl/lib:2:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/efa/lib:1:/usr/lib:1
+++ export LD_LIBRARY_PATH_modshare
+++ FI_EFA_ENABLE_SHM_TRANSFER=1
+++ export FI_EFA_ENABLE_SHM_TRANSFER
+++ _LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1
+++ export _LMFILES__modshare
+++ LOADEDMODULES_modshare=cuda/12.1:1
+++ export LOADEDMODULES_modshare
+++ MPI_PATH_modshare=/opt/amazon/openmpi:1
+++ export MPI_PATH_modshare
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin
+++ export PATH
+++ CUDA_HOME=/usr/local/cuda-12.1
+++ export CUDA_HOME
+++ PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/usr/local/cuda-12.1/include:1:/opt/amazon/efa/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/usr/sbin:1:/fsx/nouamane/miniconda/bin:1:/opt/slurm/bin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:2:/opt/amazon/openmpi/bin:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1
+++ export PATH_modshare
+++ test 0
++ _mlstatus=0
++ '[' -n x ']'
++ IFS=' 	
+'
++ unset _mlre _mlv _mlrv _mlIFS
++ '[' -n '' ']'
++ unset _mlshdbg
++ return 0
++ source /fsx/nouamane/miniconda/bin/activate
+++ _CONDA_ROOT=/fsx/nouamane/miniconda
+++ . /fsx/nouamane/miniconda/etc/profile.d/conda.sh
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ '[' -z x ']'
+++ conda activate
+++ local cmd=activate
+++ case "$cmd" in
+++ __conda_activate activate
+++ '[' -n '' ']'
+++ local ask_conda
++++ PS1=
++++ __conda_exe shell.posix activate
++++ /fsx/nouamane/miniconda/bin/conda shell.posix activate
+++ ask_conda='PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin'\''
+export CONDA_SHLVL='\''1'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\'''
+++ eval 'PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin'\''
+export CONDA_SHLVL='\''1'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\'''
++++ PS1='(base) '
++++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin
++++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin
++++ export CONDA_SHLVL=1
++++ CONDA_SHLVL=1
++++ export 'CONDA_PROMPT_MODIFIER=(base) '
++++ CONDA_PROMPT_MODIFIER='(base) '
+++ __conda_hashr
+++ '[' -n '' ']'
+++ '[' -n '' ']'
+++ hash -r
++ conda activate 2-1-cu121
++ local cmd=activate
++ case "$cmd" in
++ __conda_activate activate 2-1-cu121
++ '[' -n '' ']'
++ local ask_conda
+++ PS1='(base) '
+++ __conda_exe shell.posix activate 2-1-cu121
+++ /fsx/nouamane/miniconda/bin/conda shell.posix activate 2-1-cu121
++ ask_conda='PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''2'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_1='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
++ eval 'PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''2'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_1='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
+++ PS1='(2-1-cu121) '
+++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin
+++ export CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ export CONDA_SHLVL=2
+++ CONDA_SHLVL=2
+++ export CONDA_DEFAULT_ENV=2-1-cu121
+++ CONDA_DEFAULT_ENV=2-1-cu121
+++ export 'CONDA_PROMPT_MODIFIER=(2-1-cu121) '
+++ CONDA_PROMPT_MODIFIER='(2-1-cu121) '
+++ export CONDA_PREFIX_1=/fsx/nouamane/miniconda
+++ CONDA_PREFIX_1=/fsx/nouamane/miniconda
+++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ export _CE_M=
+++ _CE_M=
+++ export _CE_CONDA=
+++ _CE_CONDA=
+++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh
++++ test -n ''
++++ xml_catalog_files_libxml2=
++++ XML_CATALOG_FILES=
++++ conda_catalog_files=
++++ ifs_libxml2=' 	
+'
++++ IFS=' '
++++ rem=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ for pre in ${rem}
++++ test '' = /fsx/nouamane/miniconda/envs/2-1-cu121
++++ conda_catalog_files=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ rem=
++++ IFS=' 	
+'
++++ conda_catalog_files='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ export 'XML_CATALOG_FILES=file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ XML_CATALOG_FILES='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ unset conda_catalog_files ifs_libxml2 rem
++ __conda_hashr
++ '[' -n '' ']'
++ '[' -n '' ']'
++ hash -r
++ export PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin
++ PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin
+++ scontrol show hostnames 'ip-26-0-160-[100,103],ip-26-0-162-14,ip-26-0-165-[38,59,131,164,202,213],ip-26-0-172-[116,142,147,252],ip-26-0-173-[202,246],ip-26-0-174-36'
++ export 'NODELIST=ip-26-0-160-100
+ip-26-0-160-103
+ip-26-0-162-14
+ip-26-0-165-38
+ip-26-0-165-59
+ip-26-0-165-131
+ip-26-0-165-164
+ip-26-0-165-202
+ip-26-0-165-213
+ip-26-0-172-116
+ip-26-0-172-142
+ip-26-0-172-147
+ip-26-0-172-252
+ip-26-0-173-202
+ip-26-0-173-246
+ip-26-0-174-36'
++ NODELIST='ip-26-0-160-100
+ip-26-0-160-103
+ip-26-0-162-14
+ip-26-0-165-38
+ip-26-0-165-59
+ip-26-0-165-131
+ip-26-0-165-164
+ip-26-0-165-202
+ip-26-0-165-213
+ip-26-0-172-116
+ip-26-0-172-142
+ip-26-0-172-147
+ip-26-0-172-252
+ip-26-0-173-202
+ip-26-0-173-246
+ip-26-0-174-36'
+++ head -n1
+++ scontrol show hostnames 'ip-26-0-160-[100,103],ip-26-0-162-14,ip-26-0-165-[38,59,131,164,202,213],ip-26-0-172-[116,142,147,252],ip-26-0-173-[202,246],ip-26-0-174-36'
++ export MASTER_NODE=ip-26-0-160-100
++ MASTER_NODE=ip-26-0-160-100
++ export MASTER_PORT=12356
++ MASTER_PORT=12356
++ export NNODES=16
++ NNODES=16
++ export GPUS_PER_NODE=8
++ GPUS_PER_NODE=8
++ export WORLD_SIZE=128
++ WORLD_SIZE=128
++ export CUDA_DEVICE_MAX_CONNECTIONS=1
++ CUDA_DEVICE_MAX_CONNECTIONS=1
++ export NCCL_DEBUG=WARN
++ NCCL_DEBUG=WARN
++ export NANOTRON_BENCHMARK=1
++ NANOTRON_BENCHMARK=1
++ export WANDB_MODE=disabled
++ WANDB_MODE=disabled
++ export TORCH_NCCL_ASYNC_ERROR_HANDLING=1
++ TORCH_NCCL_ASYNC_ERROR_HANDLING=1
++ echo '=== GPU Topology ==='
+=== GPU Topology ===
++ nvidia-smi topo -m
+	GPU0	GPU1	GPU2	GPU3	GPU4	GPU5	GPU6	GPU7	CPU Affinity	NUMA Affinity	GPU NUMA ID
+GPU0	 X 	NV18	NV18	NV18	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU1	NV18	 X 	NV18	NV18	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU2	NV18	NV18	 X 	NV18	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU3	NV18	NV18	NV18	 X 	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU4	NV18	NV18	NV18	NV18	 X 	NV18	NV18	NV18	48-95	1		N/A
+GPU5	NV18	NV18	NV18	NV18	NV18	 X 	NV18	NV18	48-95	1		N/A
+GPU6	NV18	NV18	NV18	NV18	NV18	NV18	 X 	NV18	48-95	1		N/A
+GPU7	NV18	NV18	NV18	NV18	NV18	NV18	NV18	 X 	48-95	1		N/A
+
+Legend:
+
+  X    = Self
+  SYS  = Connection traversing PCIe as well as the SMP interconnect between NUMA nodes (e.g., QPI/UPI)
+  NODE = Connection traversing PCIe as well as the interconnect between PCIe Host Bridges within a NUMA node
+  PHB  = Connection traversing PCIe as well as a PCIe Host Bridge (typically the CPU)
+  PXB  = Connection traversing multiple PCIe bridges (without traversing the PCIe Host Bridge)
+  PIX  = Connection traversing at most a single PCIe bridge
+  NV#  = Connection traversing a bonded set of # NVLinks
++ echo ==================
+==================
++ echo 'Master node: ip-26-0-160-100'
+Master node: ip-26-0-160-100
++ echo 'All nodes: ip-26-0-160-100
+ip-26-0-160-103
+ip-26-0-162-14
+ip-26-0-165-38
+ip-26-0-165-59
+ip-26-0-165-131
+ip-26-0-165-164
+ip-26-0-165-202
+ip-26-0-165-213
+ip-26-0-172-116
+ip-26-0-172-142
+ip-26-0-172-147
+ip-26-0-172-252
+ip-26-0-173-202
+ip-26-0-173-246
+ip-26-0-174-36'
+All nodes: ip-26-0-160-100
+ip-26-0-160-103
+ip-26-0-162-14
+ip-26-0-165-38
+ip-26-0-165-59
+ip-26-0-165-131
+ip-26-0-165-164
+ip-26-0-165-202
+ip-26-0-165-213
+ip-26-0-172-116
+ip-26-0-172-142
+ip-26-0-172-147
+ip-26-0-172-252
+ip-26-0-173-202
+ip-26-0-173-246
+ip-26-0-174-36
++ echo 'World size: 128'
+World size: 128
++ srun --wait=0 --kill-on-bad-exit=1 torchrun --nnodes=16 --nproc_per_node=8 --rdzv_id=13458915 --rdzv_backend=c10d --rdzv_endpoint=ip-26-0-160-100:12356 --max_restarts 0 --rdzv_conf timeout=60 /fsx/nouamane/projects/nanotron/run_train.py --config-file benchmark/configs/config_1.34G_dp32_tp2_pp2_acc1_mbs8_seq4096_zero0_tpmodeRED_vocab131k.yaml
+[2024-12-15 10:52:43,582] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-15 10:52:43,582] torch.distributed.run: [WARNING] 
+[2024-12-15 10:52:43,582] torch.distributed.run: [WARNING] *****************************************
+[2024-12-15 10:52:43,582] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-15 10:52:43,582] torch.distributed.run: [WARNING] *****************************************
+[2024-12-15 10:52:43,582] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-15 10:52:43,582] torch.distributed.run: [WARNING] 
+[2024-12-15 10:52:43,582] torch.distributed.run: [WARNING] *****************************************
+[2024-12-15 10:52:43,582] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-15 10:52:43,582] torch.distributed.run: [WARNING] *****************************************
+[2024-12-15 10:52:43,583] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-15 10:52:43,583] torch.distributed.run: [WARNING] 
+[2024-12-15 10:52:43,583] torch.distributed.run: [WARNING] *****************************************
+[2024-12-15 10:52:43,583] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-15 10:52:43,583] torch.distributed.run: [WARNING] *****************************************
+[2024-12-15 10:52:43,583] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-15 10:52:43,583] torch.distributed.run: [WARNING] 
+[2024-12-15 10:52:43,583] torch.distributed.run: [WARNING] *****************************************
+[2024-12-15 10:52:43,583] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-15 10:52:43,583] torch.distributed.run: [WARNING] *****************************************
+[2024-12-15 10:52:43,583] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-15 10:52:43,583] torch.distributed.run: [WARNING] 
+[2024-12-15 10:52:43,583] torch.distributed.run: [WARNING] *****************************************
+[2024-12-15 10:52:43,583] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-15 10:52:43,583] torch.distributed.run: [WARNING] *****************************************
+[2024-12-15 10:52:43,585] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-15 10:52:43,585] torch.distributed.run: [WARNING] 
+[2024-12-15 10:52:43,585] torch.distributed.run: [WARNING] *****************************************
+[2024-12-15 10:52:43,585] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-15 10:52:43,585] torch.distributed.run: [WARNING] *****************************************
+[2024-12-15 10:52:43,585] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-15 10:52:43,585] torch.distributed.run: [WARNING] 
+[2024-12-15 10:52:43,585] torch.distributed.run: [WARNING] *****************************************
+[2024-12-15 10:52:43,585] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-15 10:52:43,585] torch.distributed.run: [WARNING] *****************************************
+[2024-12-15 10:52:43,584] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-15 10:52:43,584] torch.distributed.run: [WARNING] 
+[2024-12-15 10:52:43,584] torch.distributed.run: [WARNING] *****************************************
+[2024-12-15 10:52:43,584] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-15 10:52:43,584] torch.distributed.run: [WARNING] *****************************************
+[2024-12-15 10:52:43,588] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-15 10:52:43,588] torch.distributed.run: [WARNING] 
+[2024-12-15 10:52:43,588] torch.distributed.run: [WARNING] *****************************************
+[2024-12-15 10:52:43,588] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-15 10:52:43,588] torch.distributed.run: [WARNING] *****************************************
+[2024-12-15 10:52:43,590] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-15 10:52:43,590] torch.distributed.run: [WARNING] 
+[2024-12-15 10:52:43,590] torch.distributed.run: [WARNING] *****************************************
+[2024-12-15 10:52:43,590] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-15 10:52:43,590] torch.distributed.run: [WARNING] *****************************************
+[2024-12-15 10:52:43,611] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-15 10:52:43,611] torch.distributed.run: [WARNING] 
+[2024-12-15 10:52:43,611] torch.distributed.run: [WARNING] *****************************************
+[2024-12-15 10:52:43,611] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-15 10:52:43,611] torch.distributed.run: [WARNING] *****************************************
+[2024-12-15 10:52:47,406] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-15 10:52:47,406] torch.distributed.run: [WARNING] 
+[2024-12-15 10:52:47,406] torch.distributed.run: [WARNING] *****************************************
+[2024-12-15 10:52:47,406] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-15 10:52:47,406] torch.distributed.run: [WARNING] *****************************************
+[2024-12-15 10:52:47,406] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-15 10:52:47,406] torch.distributed.run: [WARNING] 
+[2024-12-15 10:52:47,406] torch.distributed.run: [WARNING] *****************************************
+[2024-12-15 10:52:47,406] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-15 10:52:47,406] torch.distributed.run: [WARNING] *****************************************
+[2024-12-15 10:52:47,407] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-15 10:52:47,411] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-15 10:52:47,411] torch.distributed.run: [WARNING] 
+[2024-12-15 10:52:47,411] torch.distributed.run: [WARNING] *****************************************
+[2024-12-15 10:52:47,411] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-15 10:52:47,411] torch.distributed.run: [WARNING] *****************************************
+[2024-12-15 10:52:47,437] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-15 10:52:47,407] torch.distributed.run: [WARNING] 
+[2024-12-15 10:52:47,407] torch.distributed.run: [WARNING] *****************************************
+[2024-12-15 10:52:47,407] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-15 10:52:47,407] torch.distributed.run: [WARNING] *****************************************
+[2024-12-15 10:52:47,438] torch.distributed.run: [WARNING] 
+[2024-12-15 10:52:47,438] torch.distributed.run: [WARNING] *****************************************
+[2024-12-15 10:52:47,438] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-15 10:52:47,438] torch.distributed.run: [WARNING] *****************************************
+NCCL version 2.18.5+cuda12.2
+12/15/2024 10:53:17 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]: Measuring inter-GPU and intra-node bandwidth...
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+12/15/2024 10:53:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]: Bandwidth measurement complete. Time taken: 18.54 seconds
+12/15/2024 10:53:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]: Config:
+12/15/2024 10:53:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]: Config(general=GeneralArgs(project='debug',
+12/15/2024 10:53:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                            run='1.34G_dp32_tp2_pp2_acc1_mbs8_seq4096_zero0_tpmodeRED_vocab131k',
+12/15/2024 10:53:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                            seed=42,
+12/15/2024 10:53:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                            step=None,
+12/15/2024 10:53:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                            consumed_train_samples=None,
+12/15/2024 10:53:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                            benchmark_csv_path=PosixPath('benchmark/results/bench_final2.csv'),
+12/15/2024 10:53:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                            ignore_sanity_checks=True),
+12/15/2024 10:53:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:        parallelism=ParallelismArgs(dp=32,
+12/15/2024 10:53:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                    pp=2,
+12/15/2024 10:53:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                    tp=2,
+12/15/2024 10:53:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                    pp_engine=<nanotron.parallel.pipeline_parallel.engine.OneForwardOneBackwardPipelineEngine object at 0x7f924af78e80>,
+12/15/2024 10:53:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                    tp_mode=<TensorParallelLinearMode.REDUCE_SCATTER: 2>,
+12/15/2024 10:53:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                    tp_linear_async_communication=True,
+12/15/2024 10:53:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                    recompute_layer=False,
+12/15/2024 10:53:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                    tp_recompute_allgather=True,
+12/15/2024 10:53:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                    expert_parallel_size=1),
+12/15/2024 10:53:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:        model=ModelArgs(model_config=LlamaConfig(bos_token_id=0,
+12/15/2024 10:53:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                                 eos_token_id=0,
+12/15/2024 10:53:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                                 hidden_act='silu',
+12/15/2024 10:53:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                                 hidden_size=2048,
+12/15/2024 10:53:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                                 initializer_range=0.02,
+12/15/2024 10:53:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                                 intermediate_size=8192,
+12/15/2024 10:53:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                                 is_llama_config=True,
+12/15/2024 10:53:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                                 max_position_embeddings=4096,
+12/15/2024 10:53:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                                 num_attention_heads=32,
+12/15/2024 10:53:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                                 num_hidden_layers=16,
+12/15/2024 10:53:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                                 num_key_value_heads=32,
+12/15/2024 10:53:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                                 pad_token_id=None,
+12/15/2024 10:53:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                                 pretraining_tp=1,
+12/15/2024 10:53:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                                 rms_norm_eps=1e-05,
+12/15/2024 10:53:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                                 rope_scaling=None,
+12/15/2024 10:53:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                                 rope_theta=10000.0,
+12/15/2024 10:53:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                                 rope_interleaved=False,
+12/15/2024 10:53:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                                 tie_word_embeddings=True,
+12/15/2024 10:53:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                                 use_cache=True,
+12/15/2024 10:53:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                                 vocab_size=131072),
+12/15/2024 10:53:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                        init_method=RandomInit(std=0.02),
+12/15/2024 10:53:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                        dtype=torch.bfloat16,
+12/15/2024 10:53:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                        make_vocab_size_divisible_by=1,
+12/15/2024 10:53:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                        ddp_bucket_cap_mb=25),
+12/15/2024 10:53:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:        tokenizer=TokenizerArgs(tokenizer_name_or_path='robot-test/dummy-tokenizer-wordlevel',
+12/15/2024 10:53:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                tokenizer_revision=None,
+12/15/2024 10:53:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                tokenizer_max_length=None),
+12/15/2024 10:53:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:        checkpoints=CheckpointsArgs(checkpoints_path=PosixPath('checkpoints'),
+12/15/2024 10:53:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                    checkpoint_interval=10000,
+12/15/2024 10:53:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                    save_initial_state=False,
+12/15/2024 10:53:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                    save_final_state=False,
+12/15/2024 10:53:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                    resume_checkpoint_path=None,
+12/15/2024 10:53:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                    checkpoints_path_is_shared_file_system=False),
+12/15/2024 10:53:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:        logging=LoggingArgs(log_level='info',
+12/15/2024 10:53:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                            log_level_replica='info',
+12/15/2024 10:53:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                            iteration_step_info_interval=1),
+12/15/2024 10:53:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:        tokens=TokensArgs(sequence_length=4096,
+12/15/2024 10:53:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                          train_steps=100,
+12/15/2024 10:53:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                          micro_batch_size=8,
+12/15/2024 10:53:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                          batch_accumulation_per_replica=1,
+12/15/2024 10:53:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                          val_check_interval=100,
+12/15/2024 10:53:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                          limit_val_batches=0,
+12/15/2024 10:53:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                          limit_test_batches=0),
+12/15/2024 10:53:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:        optimizer=OptimizerArgs(optimizer_factory=AdamWOptimizerArgs(adam_eps=1e-08,
+12/15/2024 10:53:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                                                     adam_beta1=0.9,
+12/15/2024 10:53:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                                                     adam_beta2=0.95,
+12/15/2024 10:53:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                                                     torch_adam_is_fused=True,
+12/15/2024 10:53:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                                                     name='adamW'),
+12/15/2024 10:53:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                zero_stage=0,
+12/15/2024 10:53:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                weight_decay=0.01,
+12/15/2024 10:53:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                clip_grad=1.0,
+12/15/2024 10:53:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                accumulate_grad_in_fp32=True,
+12/15/2024 10:53:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                learning_rate_scheduler=LRSchedulerArgs(learning_rate=0.0003,
+12/15/2024 10:53:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                                                        lr_warmup_steps=2,
+12/15/2024 10:53:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                                                        lr_warmup_style='linear',
+12/15/2024 10:53:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                                                        lr_decay_style='cosine',
+12/15/2024 10:53:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                                                        lr_decay_steps=13,
+12/15/2024 10:53:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                                                        lr_decay_starting_step=None,
+12/15/2024 10:53:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                                                        min_decay_lr=1e-05)),
+12/15/2024 10:53:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:        data_stages=[DatasetStageArgs(name='Stable Training Stage',
+12/15/2024 10:53:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                      start_training_step=1,
+12/15/2024 10:53:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                      data=DataArgs(dataset=None,
+12/15/2024 10:53:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                                    seed=42,
+12/15/2024 10:53:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                                    num_loading_workers=1))],
+12/15/2024 10:53:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:        profiler=None,
+12/15/2024 10:53:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:        lighteval=None,
+12/15/2024 10:53:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:        s3_upload=None)
+12/15/2024 10:53:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]: Model Config:
+12/15/2024 10:53:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]: LlamaConfig(bos_token_id=0,
+12/15/2024 10:53:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:             eos_token_id=0,
+12/15/2024 10:53:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:             hidden_act='silu',
+12/15/2024 10:53:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:             hidden_size=2048,
+12/15/2024 10:53:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:             initializer_range=0.02,
+12/15/2024 10:53:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:             intermediate_size=8192,
+12/15/2024 10:53:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:             is_llama_config=True,
+12/15/2024 10:53:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:             max_position_embeddings=4096,
+12/15/2024 10:53:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:             num_attention_heads=32,
+12/15/2024 10:53:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:             num_hidden_layers=16,
+12/15/2024 10:53:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:             num_key_value_heads=32,
+12/15/2024 10:53:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:             pad_token_id=None,
+12/15/2024 10:53:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:             pretraining_tp=1,
+12/15/2024 10:53:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:             rms_norm_eps=1e-05,
+12/15/2024 10:53:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:             rope_scaling=None,
+12/15/2024 10:53:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:             rope_theta=10000.0,
+12/15/2024 10:53:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:             rope_interleaved=False,
+12/15/2024 10:53:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:             tie_word_embeddings=True,
+12/15/2024 10:53:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:             use_cache=True,
+12/15/2024 10:53:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:             vocab_size=131072)
+12/15/2024 10:53:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]: Building model..
+12/15/2024 10:53:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]: Initialize RoPE Theta = 10000.0
+12/15/2024 10:53:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]: Setting PP block ranks...
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+12/15/2024 10:53:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]: Total number of parameters: 1.61G (3072.26MiB)
+12/15/2024 10:53:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]: Local number of parameters: 503M (960.09MiB)
+12/15/2024 10:53:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]: [After model building] Memory usage: 960.11MiB. Peak allocated: 5392.00MiB Peak reserved: 22806.00MiB
+12/15/2024 10:53:41 [INFO|DP=0|PP=1|TP=1|ip-26-0-165-59]: Local number of parameters: 302M (576.04MiB)
+12/15/2024 10:53:41 [INFO|DP=0|PP=0|TP=1|ip-26-0-160-100]: Local number of parameters: 503M (960.09MiB)
+12/15/2024 10:53:41 [INFO|DP=0|PP=1|TP=1|ip-26-0-165-59]: [After model building] Memory usage: 576.06MiB. Peak allocated: 5392.00MiB Peak reserved: 22806.00MiB
+12/15/2024 10:53:41 [INFO|DP=0|PP=0|TP=1|ip-26-0-160-100]: [After model building] Memory usage: 960.11MiB. Peak allocated: 5392.00MiB Peak reserved: 22806.00MiB
+12/15/2024 10:53:41 [INFO|DP=0|PP=1|TP=0|ip-26-0-165-59]: Local number of parameters: 302M (576.04MiB)
+12/15/2024 10:53:41 [INFO|DP=0|PP=1|TP=0|ip-26-0-165-59]: [After model building] Memory usage: 576.06MiB. Peak allocated: 5392.00MiB Peak reserved: 22806.00MiB
+NCCL version 2.18.5+cuda12.2
+12/15/2024 10:53:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]: No checkpoint path provided.
+12/15/2024 10:53:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]: Parametrizing model parameters using StandardParametrizator
+12/15/2024 10:53:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]: [Optimizer Building] Using LearningRateForSP as learning rate
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+12/15/2024 10:53:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]: [Training Plan] Stage Stable Training Stage has 99 remaining training steps and has consumed 0 samples
+12/15/2024 10:53:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]: Using dummy data generator
+12/15/2024 10:53:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]: [Training Plan] There are 1 training stages 
+12/15/2024 10:53:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]: [Stage Stable Training Stage] start from step 1 
+12/15/2024 10:53:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]: 
+12/15/2024 10:53:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]: [Start training] datetime: 2024-12-15 10:53:50.842438 | mbs: 8 | grad_accum: 1 | global_batch_size: 256 | sequence_length: 4096 | train_steps: 100 | start_iteration_step: 0 | consumed_train_samples: 0
+12/15/2024 10:53:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]: Resuming training from stage Stable Training Stage, it has trained for 0 samples and has 99 remaining train steps
+12/15/2024 10:53:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:  Memory usage: 5760.54MiB. Peak allocated 5760.54MiB. Peak reserved: 26650.00MiB
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at /opt/conda/conda-bld/pytorch_1699449201336/work/torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at /opt/conda/conda-bld/pytorch_1699449201336/work/torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at /opt/conda/conda-bld/pytorch_1699449201336/work/torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at /opt/conda/conda-bld/pytorch_1699449201336/work/torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at /opt/conda/conda-bld/pytorch_1699449201336/work/torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at /opt/conda/conda-bld/pytorch_1699449201336/work/torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at /opt/conda/conda-bld/pytorch_1699449201336/work/torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at /opt/conda/conda-bld/pytorch_1699449201336/work/torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at /opt/conda/conda-bld/pytorch_1699449201336/work/torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at /opt/conda/conda-bld/pytorch_1699449201336/work/torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at /opt/conda/conda-bld/pytorch_1699449201336/work/torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at /opt/conda/conda-bld/pytorch_1699449201336/work/torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at /opt/conda/conda-bld/pytorch_1699449201336/work/torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at /opt/conda/conda-bld/pytorch_1699449201336/work/torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at /opt/conda/conda-bld/pytorch_1699449201336/work/torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at /opt/conda/conda-bld/pytorch_1699449201336/work/torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at /opt/conda/conda-bld/pytorch_1699449201336/work/torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at /opt/conda/conda-bld/pytorch_1699449201336/work/torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at /opt/conda/conda-bld/pytorch_1699449201336/work/torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at /opt/conda/conda-bld/pytorch_1699449201336/work/torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at /opt/conda/conda-bld/pytorch_1699449201336/work/torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at /opt/conda/conda-bld/pytorch_1699449201336/work/torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at /opt/conda/conda-bld/pytorch_1699449201336/work/torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at /opt/conda/conda-bld/pytorch_1699449201336/work/torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at /opt/conda/conda-bld/pytorch_1699449201336/work/torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at /opt/conda/conda-bld/pytorch_1699449201336/work/torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at /opt/conda/conda-bld/pytorch_1699449201336/work/torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at /opt/conda/conda-bld/pytorch_1699449201336/work/torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at /opt/conda/conda-bld/pytorch_1699449201336/work/torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at /opt/conda/conda-bld/pytorch_1699449201336/work/torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at /opt/conda/conda-bld/pytorch_1699449201336/work/torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at /opt/conda/conda-bld/pytorch_1699449201336/work/torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at /opt/conda/conda-bld/pytorch_1699449201336/work/torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at /opt/conda/conda-bld/pytorch_1699449201336/work/torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at /opt/conda/conda-bld/pytorch_1699449201336/work/torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at /opt/conda/conda-bld/pytorch_1699449201336/work/torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at /opt/conda/conda-bld/pytorch_1699449201336/work/torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at /opt/conda/conda-bld/pytorch_1699449201336/work/torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at /opt/conda/conda-bld/pytorch_1699449201336/work/torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at /opt/conda/conda-bld/pytorch_1699449201336/work/torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at /opt/conda/conda-bld/pytorch_1699449201336/work/torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at /opt/conda/conda-bld/pytorch_1699449201336/work/torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at /opt/conda/conda-bld/pytorch_1699449201336/work/torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at /opt/conda/conda-bld/pytorch_1699449201336/work/torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at /opt/conda/conda-bld/pytorch_1699449201336/work/torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at /opt/conda/conda-bld/pytorch_1699449201336/work/torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at /opt/conda/conda-bld/pytorch_1699449201336/work/torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at /opt/conda/conda-bld/pytorch_1699449201336/work/torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at /opt/conda/conda-bld/pytorch_1699449201336/work/torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at /opt/conda/conda-bld/pytorch_1699449201336/work/torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at /opt/conda/conda-bld/pytorch_1699449201336/work/torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at /opt/conda/conda-bld/pytorch_1699449201336/work/torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at /opt/conda/conda-bld/pytorch_1699449201336/work/torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at /opt/conda/conda-bld/pytorch_1699449201336/work/torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at /opt/conda/conda-bld/pytorch_1699449201336/work/torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at /opt/conda/conda-bld/pytorch_1699449201336/work/torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+12/15/2024 10:53:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:  Memory usage: 6086.32MiB. Peak allocated 25005.99MiB. Peak reserved: 25176.00MiB
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at /opt/conda/conda-bld/pytorch_1699449201336/work/torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at /opt/conda/conda-bld/pytorch_1699449201336/work/torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at /opt/conda/conda-bld/pytorch_1699449201336/work/torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at /opt/conda/conda-bld/pytorch_1699449201336/work/torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at /opt/conda/conda-bld/pytorch_1699449201336/work/torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at /opt/conda/conda-bld/pytorch_1699449201336/work/torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at /opt/conda/conda-bld/pytorch_1699449201336/work/torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at /opt/conda/conda-bld/pytorch_1699449201336/work/torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+12/15/2024 10:53:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:  Memory usage: 9926.69MiB. Peak allocated 9926.69MiB. Peak reserved: 25176.00MiB
+12/15/2024 10:53:59 [INFO|DP=0|PP=1|TP=0|ip-26-0-165-59]: iteration: 1 / 100 | consumed_tokens: 1.05M | elapsed_time_per_iteration_ms: 7.3K | tokens_per_sec: 144K | tokens_per_sec_per_gpu: 1.12K | global_batch_size: 256 | lm_loss: 12 | lr: 0.00015 | model_tflops_per_gpu: 10.8 | hardware_tflops_per_gpu: 10.8 | grad_norm: 0.634 | cuda_memory_allocated: 6.38G | cuda_max_memory_reserved: 34.6G | hd_total_memory_tb: 312G | hd_used_memory_tb: 68.2G | hd_free_memory_tb: 244G
+12/15/2024 10:53:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:  Memory usage: 9926.69MiB. Peak allocated 28846.45MiB. Peak reserved: 30104.00MiB
+12/15/2024 10:54:00 [INFO|DP=0|PP=1|TP=0|ip-26-0-165-59]: iteration: 2 / 100 | consumed_tokens: 2.1M | elapsed_time_per_iteration_ms: 535 | tokens_per_sec: 1.96M | tokens_per_sec_per_gpu: 15.3K | global_batch_size: 256 | lm_loss: 12 | lr: 0.0003 | model_tflops_per_gpu: 148 | hardware_tflops_per_gpu: 148 | grad_norm: 0.634 | cuda_memory_allocated: 6.38G | cuda_max_memory_reserved: 34.6G | hd_total_memory_tb: 312G | hd_used_memory_tb: 68.2G | hd_free_memory_tb: 244G
+12/15/2024 10:54:00 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:  Memory usage: 9926.69MiB. Peak allocated 9926.73MiB. Peak reserved: 30104.00MiB
+12/15/2024 10:54:00 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:  Memory usage: 9926.69MiB. Peak allocated 28846.45MiB. Peak reserved: 31512.00MiB
+12/15/2024 10:54:00 [INFO|DP=0|PP=1|TP=0|ip-26-0-165-59]: iteration: 3 / 100 | consumed_tokens: 3.15M | elapsed_time_per_iteration_ms: 526 | tokens_per_sec: 1.99M | tokens_per_sec_per_gpu: 15.6K | global_batch_size: 256 | lm_loss: 12 | lr: 0.000296 | model_tflops_per_gpu: 150 | hardware_tflops_per_gpu: 150 | grad_norm: 0.629 | cuda_memory_allocated: 6.38G | cuda_max_memory_reserved: 34.6G | hd_total_memory_tb: 312G | hd_used_memory_tb: 68.2G | hd_free_memory_tb: 244G
+12/15/2024 10:54:00 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]: | job_id   | name                                                           | nodes | seq_len | mbs | batch_accum | gbs | mTFLOPs | hTFLOPs | tok/s/gpu | AllReduce (GB/s) | AllGather (GB/s) | ReduceScatter (GB/s) | AR Intra-node (GB/s) | AG Intra-node (GB/s) | RS Intra-node (GB/s) | Mem Alloc (GB) | Mem Res (GB) | dp | pp | tp | pp_engine                           | tp_mode        | tp_async_comm | hidden_size | hidden_act | num_layers | num_heads | num_kv_heads | max_pos | vocab_size | tie_word_embeddings | dtype          | zero_stage | ddp_bucket_cap_mb | accumulate_grad_in_fp32 | Total Params | Local Params |
+12/15/2024 10:54:00 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]: | -------- | -------------------------------------------------------------- | ----- | ------- | --- | ----------- | --- | ------- | ------- | --------- | ---------------- | ---------------- | -------------------- | -------------------- | -------------------- | -------------------- | -------------- | ------------ | -- | -- | -- | ----------------------------------- | -------------- | ------------- | ----------- | ---------- | ---------- | --------- | ------------ | ------- | ---------- | ------------------- | -------------- | ---------- | ----------------- | ----------------------- | ------------ | ------------ |
+12/15/2024 10:54:00 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]: | 13458915 | 1.34G_dp32_tp2_pp2_acc1_mbs8_seq4096_zero0_tpmodeRED_vocab131k | 16    | 4096    | 8   | 1           | 256 | 150.25  | 150.25  | 15548.29  | 178.13           | 125.27           | 113.63               | 456.66               | 263.75               | 263.26               | 9.69           | 30.77        | 32 | 2  | 2  | OneForwardOneBackwardPipelineEngine | REDUCE_SCATTER | True          | 2048        | silu       | 16         | 32        | 32           | 4096    | 131072     | True                | torch.bfloat16 | 0          | 25                | True                    | 1.61G        | 503M         |
+12/15/2024 10:54:00 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]: Attempting to write benchmark results to CSV file: benchmark/results/bench_final2.csv
+12/15/2024 10:54:00 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]: Acquired lock for CSV file: benchmark/results/bench_final2.csv
+12/15/2024 10:54:00 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]: Successfully wrote to CSV file: benchmark/results/bench_final2.csv. Releasing lock...
+12/15/2024 10:54:00 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]: Throughput logging complete
+[2024-12-15 10:54:24,852] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 130869 closing signal SIGTERM
+[2024-12-15 10:54:24,853] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 130870 closing signal SIGTERM
+[2024-12-15 10:54:24,853] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 130871 closing signal SIGTERM
+[2024-12-15 10:54:24,853] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 130872 closing signal SIGTERM
+[2024-12-15 10:54:24,853] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 130873 closing signal SIGTERM
+[2024-12-15 10:54:24,853] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 130875 closing signal SIGTERM
+[2024-12-15 10:54:24,853] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 130876 closing signal SIGTERM
+[2024-12-15 10:54:29,464] torch.distributed.elastic.multiprocessing.api: [ERROR] failed (exitcode: 1) local_rank: 5 (pid: 130874) of binary: /fsx/nouamane/miniconda/envs/2-1-cu121/bin/python
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 33, in <module>
+    sys.exit(load_entry_point('torch==2.1.1', 'console_scripts', 'torchrun')())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 264, in launch_agent
+    raise ChildFailedError(
+torch.distributed.elastic.multiprocessing.errors.ChildFailedError: 
+============================================================
+/fsx/nouamane/projects/nanotron/run_train.py FAILED
+------------------------------------------------------------
+Failures:
+  <NO_OTHER_FAILURES>
+------------------------------------------------------------
+Root Cause (first observed failure):
+[0]:
+  time      : 2024-12-15_10:54:24
+  host      : ip-26-0-160-100.ec2.internal
+  rank      : 5 (local_rank: 5)
+  exitcode  : 1 (pid: 130874)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+============================================================
+[2024-12-15 10:54:29,623] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-173-246.ec2.internal_96848_0' has failed to send a keep-alive heartbeat to the rendezvous '13458915' due to an error of type RendezvousConnectionError.
+[2024-12-15 10:54:29,734] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-165-202.ec2.internal_70428_0' has failed to send a keep-alive heartbeat to the rendezvous '13458915' due to an error of type RendezvousConnectionError.
+[2024-12-15 10:54:29,740] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-165-131.ec2.internal_304428_0' has failed to send a keep-alive heartbeat to the rendezvous '13458915' due to an error of type RendezvousConnectionError.
+[2024-12-15 10:54:29,746] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-172-252.ec2.internal_109100_0' has failed to send a keep-alive heartbeat to the rendezvous '13458915' due to an error of type RendezvousConnectionError.
+[2024-12-15 10:54:29,762] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-160-103.ec2.internal_311287_0' has failed to send a keep-alive heartbeat to the rendezvous '13458915' due to an error of type RendezvousConnectionError.
+[2024-12-15 10:54:29,808] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-174-36.ec2.internal_54905_0' has failed to send a keep-alive heartbeat to the rendezvous '13458915' due to an error of type RendezvousConnectionError.
+[2024-12-15 10:54:29,842] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 96930 closing signal SIGTERM
+[2024-12-15 10:54:29,842] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 96931 closing signal SIGTERM
+[2024-12-15 10:54:29,842] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 96932 closing signal SIGTERM
+[2024-12-15 10:54:29,842] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 96933 closing signal SIGTERM
+[2024-12-15 10:54:29,842] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 96934 closing signal SIGTERM
+[2024-12-15 10:54:29,842] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 96935 closing signal SIGTERM
+[2024-12-15 10:54:29,842] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 96936 closing signal SIGTERM
+[2024-12-15 10:54:29,842] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 96937 closing signal SIGTERM
+[2024-12-15 10:54:29,847] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 216907 closing signal SIGTERM
+[2024-12-15 10:54:29,847] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 216909 closing signal SIGTERM
+[2024-12-15 10:54:29,847] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 216910 closing signal SIGTERM
+[2024-12-15 10:54:29,847] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 216911 closing signal SIGTERM
+[2024-12-15 10:54:29,847] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 216912 closing signal SIGTERM
+[2024-12-15 10:54:29,847] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 216913 closing signal SIGTERM
+[2024-12-15 10:54:29,848] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 304507 closing signal SIGTERM
+[2024-12-15 10:54:29,848] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 304509 closing signal SIGTERM
+[2024-12-15 10:54:29,849] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 304510 closing signal SIGTERM
+[2024-12-15 10:54:29,849] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 304511 closing signal SIGTERM
+[2024-12-15 10:54:29,849] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 304512 closing signal SIGTERM
+[2024-12-15 10:54:29,849] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 304513 closing signal SIGTERM
+[2024-12-15 10:54:29,849] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 304514 closing signal SIGTERM
+[2024-12-15 10:54:29,852] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 50260 closing signal SIGTERM
+[2024-12-15 10:54:29,852] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 50261 closing signal SIGTERM
+[2024-12-15 10:54:29,852] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 50264 closing signal SIGTERM
+[2024-12-15 10:54:29,852] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 50265 closing signal SIGTERM
+[2024-12-15 10:54:29,852] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 50266 closing signal SIGTERM
+[2024-12-15 10:54:29,852] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 123553 closing signal SIGTERM
+[2024-12-15 10:54:29,852] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 123554 closing signal SIGTERM
+[2024-12-15 10:54:29,852] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 123555 closing signal SIGTERM
+[2024-12-15 10:54:29,853] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 123556 closing signal SIGTERM
+[2024-12-15 10:54:29,853] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 123558 closing signal SIGTERM
+[2024-12-15 10:54:29,853] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 123559 closing signal SIGTERM
+[2024-12-15 10:54:29,853] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 123560 closing signal SIGTERM
+[2024-12-15 10:54:29,853] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 311365 closing signal SIGTERM
+[2024-12-15 10:54:29,854] torch.distributed.elastic.multiprocessing.api: [ERROR] failed (exitcode: 1) local_rank: 0 (pid: 54983) of binary: /fsx/nouamane/miniconda/envs/2-1-cu121/bin/python
+[2024-12-15 10:54:29,853] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 311366 closing signal SIGTERM
+[2024-12-15 10:54:29,853] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 311367 closing signal SIGTERM
+[2024-12-15 10:54:29,853] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 311368 closing signal SIGTERM
+[2024-12-15 10:54:29,853] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 311369 closing signal SIGTERM
+[2024-12-15 10:54:29,853] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 311370 closing signal SIGTERM
+[2024-12-15 10:54:29,855] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 61510 closing signal SIGTERM
+[2024-12-15 10:54:29,853] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 311371 closing signal SIGTERM
+[2024-12-15 10:54:29,855] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 61512 closing signal SIGTERM
+[2024-12-15 10:54:29,855] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 61513 closing signal SIGTERM
+[2024-12-15 10:54:29,855] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 61514 closing signal SIGTERM
+[2024-12-15 10:54:29,855] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 61515 closing signal SIGTERM
+[2024-12-15 10:54:29,855] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 61516 closing signal SIGTERM
+[2024-12-15 10:54:29,856] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 86821 closing signal SIGTERM
+[2024-12-15 10:54:29,856] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 86822 closing signal SIGTERM
+[2024-12-15 10:54:29,856] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 86823 closing signal SIGTERM
+[2024-12-15 10:54:29,856] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 86824 closing signal SIGTERM
+[2024-12-15 10:54:29,856] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 86825 closing signal SIGTERM
+[2024-12-15 10:54:29,856] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 86826 closing signal SIGTERM
+[2024-12-15 10:54:29,856] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 70507 closing signal SIGTERM
+[2024-12-15 10:54:29,857] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 70508 closing signal SIGTERM
+[2024-12-15 10:54:29,856] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 101334 closing signal SIGTERM
+[2024-12-15 10:54:29,857] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 70510 closing signal SIGTERM
+[2024-12-15 10:54:29,857] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 101335 closing signal SIGTERM
+[2024-12-15 10:54:29,857] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 101336 closing signal SIGTERM
+[2024-12-15 10:54:29,857] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 70511 closing signal SIGTERM
+[2024-12-15 10:54:29,857] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 101337 closing signal SIGTERM
+[2024-12-15 10:54:29,857] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 101338 closing signal SIGTERM
+[2024-12-15 10:54:29,857] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 70512 closing signal SIGTERM
+[2024-12-15 10:54:29,857] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 70513 closing signal SIGTERM
+[2024-12-15 10:54:29,857] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 101339 closing signal SIGTERM
+[2024-12-15 10:54:29,857] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 255475 closing signal SIGTERM
+[2024-12-15 10:54:29,857] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 255476 closing signal SIGTERM
+[2024-12-15 10:54:29,858] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 255477 closing signal SIGTERM
+[2024-12-15 10:54:29,858] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 255478 closing signal SIGTERM
+[2024-12-15 10:54:29,858] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 255479 closing signal SIGTERM
+[2024-12-15 10:54:29,858] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 255480 closing signal SIGTERM
+[2024-12-15 10:54:29,858] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 255481 closing signal SIGTERM
+[2024-12-15 10:54:29,857] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 115767 closing signal SIGTERM
+[2024-12-15 10:54:29,858] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 115768 closing signal SIGTERM
+[2024-12-15 10:54:29,858] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 115769 closing signal SIGTERM
+[2024-12-15 10:54:29,858] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 115770 closing signal SIGTERM
+[2024-12-15 10:54:29,858] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 115771 closing signal SIGTERM
+[2024-12-15 10:54:29,858] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 115772 closing signal SIGTERM
+[2024-12-15 10:54:29,863] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 109176 closing signal SIGTERM
+[2024-12-15 10:54:29,864] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 109177 closing signal SIGTERM
+[2024-12-15 10:54:29,864] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 109179 closing signal SIGTERM
+[2024-12-15 10:54:29,864] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 109180 closing signal SIGTERM
+[2024-12-15 10:54:29,864] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 109181 closing signal SIGTERM
+[2024-12-15 10:54:29,864] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 109182 closing signal SIGTERM
+[2024-12-15 10:54:29,864] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 109183 closing signal SIGTERM
+[2024-12-15 10:54:29,910] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-174-36.ec2.internal_54905_0' has failed to shutdown the rendezvous '13458915' due to an error of type RendezvousConnectionError.
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 33, in <module>
+    sys.exit(load_entry_point('torch==2.1.1', 'console_scripts', 'torchrun')())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 264, in launch_agent
+    raise ChildFailedError(
+torch.distributed.elastic.multiprocessing.errors.ChildFailedError: 
+============================================================
+/fsx/nouamane/projects/nanotron/run_train.py FAILED
+------------------------------------------------------------
+Failures:
+[1]:
+  time      : 2024-12-15_10:54:29
+  host      : ip-26-0-174-36.ec2.internal
+  rank      : 121 (local_rank: 1)
+  exitcode  : 1 (pid: 54984)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[2]:
+  time      : 2024-12-15_10:54:29
+  host      : ip-26-0-174-36.ec2.internal
+  rank      : 122 (local_rank: 2)
+  exitcode  : 1 (pid: 54985)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[3]:
+  time      : 2024-12-15_10:54:29
+  host      : ip-26-0-174-36.ec2.internal
+  rank      : 123 (local_rank: 3)
+  exitcode  : 1 (pid: 54986)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[4]:
+  time      : 2024-12-15_10:54:29
+  host      : ip-26-0-174-36.ec2.internal
+  rank      : 124 (local_rank: 4)
+  exitcode  : 1 (pid: 54987)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[5]:
+  time      : 2024-12-15_10:54:29
+  host      : ip-26-0-174-36.ec2.internal
+  rank      : 125 (local_rank: 5)
+  exitcode  : 1 (pid: 54988)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[6]:
+  time      : 2024-12-15_10:54:29
+  host      : ip-26-0-174-36.ec2.internal
+  rank      : 126 (local_rank: 6)
+  exitcode  : 1 (pid: 54989)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[7]:
+  time      : 2024-12-15_10:54:29
+  host      : ip-26-0-174-36.ec2.internal
+  rank      : 127 (local_rank: 7)
+  exitcode  : 1 (pid: 54990)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+------------------------------------------------------------
+Root Cause (first observed failure):
+[0]:
+  time      : 2024-12-15_10:54:29
+  host      : ip-26-0-174-36.ec2.internal
+  rank      : 120 (local_rank: 0)
+  exitcode  : 1 (pid: 54983)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+============================================================
+srun: error: ip-26-0-160-100: task 0: Exited with exit code 1
+srun: launch/slurm: _step_signal: Terminating StepId=13458915.0
+[2024-12-15 10:54:29,936] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-15 10:54:29,936] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 311365 closing signal SIGTERM
+[2024-12-15 10:54:29,936] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 311366 closing signal SIGTERM
+[2024-12-15 10:54:29,936] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 311367 closing signal SIGTERM
+[2024-12-15 10:54:29,936] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 311368 closing signal SIGTERM
+[2024-12-15 10:54:29,936] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 311369 closing signal SIGTERM
+[2024-12-15 10:54:29,936] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 311370 closing signal SIGTERM
+[2024-12-15 10:54:29,938] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-15 10:54:29,936] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 311371 closing signal SIGTERM
+[2024-12-15 10:54:29,938] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 255475 closing signal SIGTERM
+[2024-12-15 10:54:29,938] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-15 10:54:29,938] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 255476 closing signal SIGTERM
+[2024-12-15 10:54:29,937] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-15 10:54:29,937] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-15 10:54:29,937] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-15 10:54:29,938] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 105394 closing signal SIGTERM
+[2024-12-15 10:54:29,938] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-15 10:54:29,938] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-15 10:54:29,938] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 304507 closing signal SIGTERM
+[2024-12-15 10:54:29,937] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 115767 closing signal SIGTERM
+[2024-12-15 10:54:29,937] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 216907 closing signal SIGTERM
+[2024-12-15 10:54:29,938] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 105395 closing signal SIGTERM
+[2024-12-15 10:54:29,938] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 50260 closing signal SIGTERM
+[2024-12-15 10:54:29,938] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 70507 closing signal SIGTERM
+[2024-12-15 10:54:29,937] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-15 10:54:29,937] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 115768 closing signal SIGTERM
+[2024-12-15 10:54:29,938] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-15 10:54:29,938] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 255477 closing signal SIGTERM
+[2024-12-15 10:54:29,938] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 304509 closing signal SIGTERM
+[2024-12-15 10:54:29,937] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 115769 closing signal SIGTERM
+[2024-12-15 10:54:29,938] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 105396 closing signal SIGTERM
+[2024-12-15 10:54:29,938] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 50261 closing signal SIGTERM
+[2024-12-15 10:54:29,938] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 70508 closing signal SIGTERM
+[2024-12-15 10:54:29,938] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 255478 closing signal SIGTERM
+[2024-12-15 10:54:29,938] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 304510 closing signal SIGTERM
+[2024-12-15 10:54:29,938] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 101334 closing signal SIGTERM
+[2024-12-15 10:54:29,937] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 115770 closing signal SIGTERM
+[2024-12-15 10:54:29,937] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 216909 closing signal SIGTERM
+[2024-12-15 10:54:29,938] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 105397 closing signal SIGTERM
+[2024-12-15 10:54:29,938] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 50264 closing signal SIGTERM
+[2024-12-15 10:54:29,938] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 61510 closing signal SIGTERM
+[2024-12-15 10:54:29,938] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 255479 closing signal SIGTERM
+[2024-12-15 10:54:29,938] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 304511 closing signal SIGTERM
+[2024-12-15 10:54:29,937] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 115771 closing signal SIGTERM
+[2024-12-15 10:54:29,937] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 216910 closing signal SIGTERM
+[2024-12-15 10:54:29,938] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 105398 closing signal SIGTERM
+[2024-12-15 10:54:29,938] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 50265 closing signal SIGTERM
+[2024-12-15 10:54:29,938] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 61512 closing signal SIGTERM
+[2024-12-15 10:54:29,938] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 304512 closing signal SIGTERM
+[2024-12-15 10:54:29,938] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 101335 closing signal SIGTERM
+[2024-12-15 10:54:29,937] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 115772 closing signal SIGTERM
+[2024-12-15 10:54:29,938] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 105399 closing signal SIGTERM
+[2024-12-15 10:54:29,938] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 50266 closing signal SIGTERM
+[2024-12-15 10:54:29,938] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 70510 closing signal SIGTERM
+[2024-12-15 10:54:29,938] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 304513 closing signal SIGTERM
+[2024-12-15 10:54:29,938] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 101336 closing signal SIGTERM
+[2024-12-15 10:54:29,937] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 216911 closing signal SIGTERM
+[2024-12-15 10:54:29,938] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 105400 closing signal SIGTERM
+[2024-12-15 10:54:29,938] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 61513 closing signal SIGTERM
+[2024-12-15 10:54:29,938] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 70511 closing signal SIGTERM
+[2024-12-15 10:54:29,938] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 255480 closing signal SIGTERM
+[2024-12-15 10:54:29,938] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 304514 closing signal SIGTERM
+[2024-12-15 10:54:29,938] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 101337 closing signal SIGTERM
+[2024-12-15 10:54:29,938] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 216912 closing signal SIGTERM
+[2024-12-15 10:54:29,938] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 105401 closing signal SIGTERM
+[2024-12-15 10:54:29,938] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 61514 closing signal SIGTERM
+[2024-12-15 10:54:29,938] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 255481 closing signal SIGTERM
+[2024-12-15 10:54:29,938] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 101338 closing signal SIGTERM
+[2024-12-15 10:54:29,938] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 61515 closing signal SIGTERM
+[2024-12-15 10:54:29,938] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 101339 closing signal SIGTERM
+[2024-12-15 10:54:29,938] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 70512 closing signal SIGTERM
+[2024-12-15 10:54:29,938] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 61516 closing signal SIGTERM
+[2024-12-15 10:54:29,938] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 70513 closing signal SIGTERM
+[2024-12-15 10:54:29,938] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 216913 closing signal SIGTERM
+[2024-12-15 10:54:29,938] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-15 10:54:29,938] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-15 10:54:29,938] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 123553 closing signal SIGTERM
+[2024-12-15 10:54:29,938] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 123554 closing signal SIGTERM
+[2024-12-15 10:54:29,938] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 109176 closing signal SIGTERM
+[2024-12-15 10:54:29,938] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 123555 closing signal SIGTERM
+[2024-12-15 10:54:29,938] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 123556 closing signal SIGTERM
+[2024-12-15 10:54:29,938] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 109177 closing signal SIGTERM
+[2024-12-15 10:54:29,938] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 123558 closing signal SIGTERM
+[2024-12-15 10:54:29,938] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 123559 closing signal SIGTERM
+[2024-12-15 10:54:29,938] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 123560 closing signal SIGTERM
+[2024-12-15 10:54:29,938] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 109179 closing signal SIGTERM
+[2024-12-15 10:54:29,938] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-15 10:54:29,939] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 109180 closing signal SIGTERM
+[2024-12-15 10:54:29,938] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 86821 closing signal SIGTERM
+[2024-12-15 10:54:29,939] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 109181 closing signal SIGTERM
+[2024-12-15 10:54:29,938] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 86822 closing signal SIGTERM
+[2024-12-15 10:54:29,939] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 109182 closing signal SIGTERM
+[2024-12-15 10:54:29,938] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 86823 closing signal SIGTERM
+[2024-12-15 10:54:29,938] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 86824 closing signal SIGTERM
+[2024-12-15 10:54:29,939] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 86825 closing signal SIGTERM
+[2024-12-15 10:54:29,939] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 109183 closing signal SIGTERM
+[2024-12-15 10:54:29,939] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 86826 closing signal SIGTERM
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/rendezvous/c10d_rendezvous_backend.py", line 113, in _call_store
+    return getattr(self._store, store_op)(*args, **kwargs)
+RuntimeError: Broken pipe
+
+The above exception was the direct cause of the following exception:
+
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 909, in _invoke_run
+    num_nodes_waiting = rdzv_handler.num_nodes_waiting()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/rendezvous/dynamic_rendezvous.py", line 1083, in num_nodes_waiting
+    self._state_holder.sync()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/rendezvous/dynamic_rendezvous.py", line 409, in sync
+    get_response = self._backend.get_state()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/rendezvous/c10d_rendezvous_backend.py", line 73, in get_state
+    base64_state: bytes = self._call_store("get", self._key)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/rendezvous/c10d_rendezvous_backend.py", line 115, in _call_store
+    raise RendezvousConnectionError(
+torch.distributed.elastic.rendezvous.api.RendezvousConnectionError: The connection to the C10d store has failed. See inner exception for details.
+
+During handling of the above exception, another exception occurred:
+
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 33, in <module>
+    sys.exit(load_entry_point('torch==2.1.1', 'console_scripts', 'torchrun')())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 748, in run
+    self._shutdown()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/local_elastic_agent.py", line 289, in _shutdown
+    self._pcontext.close(death_sig)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 331, in close
+    self._close(death_sig=death_sig, timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 713, in _close
+    handler.proc.wait(time_to_wait)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1209, in wait
+    return self._wait(timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1953, in _wait
+    time.sleep(delay)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 96848 got signal: 15
+srun: error: ip-26-0-174-36: task 15: Terminated
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 33, in <module>
+    sys.exit(load_entry_point('torch==2.1.1', 'console_scripts', 'torchrun')())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 878, in _invoke_run
+    run_result = self._monitor_workers(self._worker_group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/local_elastic_agent.py", line 307, in _monitor_workers
+    result = self._pcontext.wait(0)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 288, in wait
+    return self._poll()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 668, in _poll
+    self.close()  # terminate all running procs
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 331, in close
+    self._close(death_sig=death_sig, timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 713, in _close
+    handler.proc.wait(time_to_wait)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1209, in wait
+    return self._wait(timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1953, in _wait
+    time.sleep(delay)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 86740 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 33, in <module>
+    sys.exit(load_entry_point('torch==2.1.1', 'console_scripts', 'torchrun')())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 878, in _invoke_run
+    run_result = self._monitor_workers(self._worker_group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/local_elastic_agent.py", line 307, in _monitor_workers
+    result = self._pcontext.wait(0)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 288, in wait
+    return self._poll()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 668, in _poll
+    self.close()  # terminate all running procs
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 331, in close
+    self._close(death_sig=death_sig, timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 713, in _close
+    handler.proc.wait(time_to_wait)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1209, in wait
+    return self._wait(timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1953, in _wait
+    time.sleep(delay)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 61432 got signal: 15
+srun: error: ip-26-0-173-202: task 13: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 33, in <module>
+    sys.exit(load_entry_point('torch==2.1.1', 'console_scripts', 'torchrun')())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 878, in _invoke_run
+    run_result = self._monitor_workers(self._worker_group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/local_elastic_agent.py", line 307, in _monitor_workers
+    result = self._pcontext.wait(0)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 288, in wait
+    return self._poll()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 668, in _poll
+    self.close()  # terminate all running procs
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 331, in close
+    self._close(death_sig=death_sig, timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 713, in _close
+    handler.proc.wait(time_to_wait)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1209, in wait
+    return self._wait(timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1953, in _wait
+    time.sleep(delay)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 216828 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 33, in <module>
+    sys.exit(load_entry_point('torch==2.1.1', 'console_scripts', 'torchrun')())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 878, in _invoke_run
+    run_result = self._monitor_workers(self._worker_group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/local_elastic_agent.py", line 307, in _monitor_workers
+    result = self._pcontext.wait(0)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 288, in wait
+    return self._poll()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 668, in _poll
+    self.close()  # terminate all running procs
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 331, in close
+    self._close(death_sig=death_sig, timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 713, in _close
+srun: error: ip-26-0-165-213: task 8: Exited with exit code 1
+    handler.proc.wait(time_to_wait)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1209, in wait
+    return self._wait(timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1953, in _wait
+    time.sleep(delay)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 50182 got signal: 15
+srun: error: ip-26-0-165-59: task 4: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 33, in <module>
+    sys.exit(load_entry_point('torch==2.1.1', 'console_scripts', 'torchrun')())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 878, in _invoke_run
+    run_result = self._monitor_workers(self._worker_group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/local_elastic_agent.py", line 307, in _monitor_workers
+    result = self._pcontext.wait(0)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 288, in wait
+    return self._poll()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 668, in _poll
+    self.close()  # terminate all running procs
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 331, in close
+    self._close(death_sig=death_sig, timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 713, in _close
+    handler.proc.wait(time_to_wait)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1209, in wait
+    return self._wait(timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1953, in _wait
+    time.sleep(delay)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 115690 got signal: 15
+srun: error: ip-26-0-172-142: task 10: Exited with exit code 1
+[2024-12-15 10:54:33,870] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-162-14.ec2.internal_105316_0' has failed to send a keep-alive heartbeat to the rendezvous '13458915' due to an error of type RendezvousConnectionError.
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 33, in <module>
+    sys.exit(load_entry_point('torch==2.1.1', 'console_scripts', 'torchrun')())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 878, in _invoke_run
+    run_result = self._monitor_workers(self._worker_group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/local_elastic_agent.py", line 307, in _monitor_workers
+    result = self._pcontext.wait(0)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 288, in wait
+    return self._poll()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 668, in _poll
+    self.close()  # terminate all running procs
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 331, in close
+    self._close(death_sig=death_sig, timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 713, in _close
+    handler.proc.wait(time_to_wait)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1209, in wait
+    return self._wait(timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1953, in _wait
+    time.sleep(delay)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 101256 got signal: 15
+[2024-12-15 10:54:33,983] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-165-164.ec2.internal_255396_0' has failed to send a keep-alive heartbeat to the rendezvous '13458915' due to an error of type RendezvousConnectionError.
+[2024-12-15 10:54:34,011] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-172-116.ec2.internal_123476_0' has failed to send a keep-alive heartbeat to the rendezvous '13458915' due to an error of type RendezvousConnectionError.
+srun: error: ip-26-0-172-147: task 11: Exited with exit code 1
+srun: error: ip-26-0-165-38: task 3: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 33, in <module>
+    sys.exit(load_entry_point('torch==2.1.1', 'console_scripts', 'torchrun')())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 878, in _invoke_run
+    run_result = self._monitor_workers(self._worker_group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/local_elastic_agent.py", line 307, in _monitor_workers
+    result = self._pcontext.wait(0)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 288, in wait
+    return self._poll()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 668, in _poll
+    self.close()  # terminate all running procs
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 331, in close
+    self._close(death_sig=death_sig, timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 713, in _close
+    handler.proc.wait(time_to_wait)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1209, in wait
+    return self._wait(timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1953, in _wait
+    time.sleep(delay)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 70428 got signal: 15
+[2024-12-15 10:54:34,741] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-165-131.ec2.internal_304428_0' has failed to send a keep-alive heartbeat to the rendezvous '13458915' due to an error of type RendezvousConnectionError.
+[2024-12-15 10:54:34,748] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-172-252.ec2.internal_109100_0' has failed to send a keep-alive heartbeat to the rendezvous '13458915' due to an error of type RendezvousConnectionError.
+[2024-12-15 10:54:34,763] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-160-103.ec2.internal_311287_0' has failed to send a keep-alive heartbeat to the rendezvous '13458915' due to an error of type RendezvousConnectionError.
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 33, in <module>
+    sys.exit(load_entry_point('torch==2.1.1', 'console_scripts', 'torchrun')())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 33, in <module>
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    sys.exit(load_entry_point('torch==2.1.1', 'console_scripts', 'torchrun')())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    return launch_agent(self._config, self._entrypoint, list(args))
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 878, in _invoke_run
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    run_result = self._monitor_workers(self._worker_group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 878, in _invoke_run
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/local_elastic_agent.py", line 307, in _monitor_workers
+    run_result = self._monitor_workers(self._worker_group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = self._pcontext.wait(0)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 288, in wait
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/local_elastic_agent.py", line 307, in _monitor_workers
+    return self._poll()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 668, in _poll
+    result = self._pcontext.wait(0)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 288, in wait
+    self.close()  # terminate all running procs
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 331, in close
+    return self._poll()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 668, in _poll
+    self._close(death_sig=death_sig, timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 713, in _close
+    handler.proc.wait(time_to_wait)
+    self.close()  # terminate all running procs
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 331, in close
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1209, in wait
+    self._close(death_sig=death_sig, timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 713, in _close
+    return self._wait(timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1953, in _wait
+    handler.proc.wait(time_to_wait)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1209, in wait
+    time.sleep(delay)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    return self._wait(timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1953, in _wait
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 255396 got signal: 15
+    time.sleep(delay)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 109100 got signal: 15
+srun: error: ip-26-0-165-202: task 7: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 33, in <module>
+    sys.exit(load_entry_point('torch==2.1.1', 'console_scripts', 'torchrun')())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 878, in _invoke_run
+    run_result = self._monitor_workers(self._worker_group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/local_elastic_agent.py", line 307, in _monitor_workers
+    result = self._pcontext.wait(0)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 288, in wait
+    return self._poll()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 668, in _poll
+    self.close()  # terminate all running procs
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 331, in close
+    self._close(death_sig=death_sig, timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 713, in _close
+    handler.proc.wait(time_to_wait)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1209, in wait
+    return self._wait(timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1953, in _wait
+    time.sleep(delay)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 304428 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 33, in <module>
+    sys.exit(load_entry_point('torch==2.1.1', 'console_scripts', 'torchrun')())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 878, in _invoke_run
+    run_result = self._monitor_workers(self._worker_group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/local_elastic_agent.py", line 307, in _monitor_workers
+    result = self._pcontext.wait(0)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 288, in wait
+    return self._poll()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 668, in _poll
+    self.close()  # terminate all running procs
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 331, in close
+    self._close(death_sig=death_sig, timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 713, in _close
+    handler.proc.wait(time_to_wait)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1209, in wait
+    return self._wait(timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1953, in _wait
+    time.sleep(delay)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 123476 got signal: 15
+srun: error: ip-26-0-172-252: task 12: Exited with exit code 1
+srun: error: ip-26-0-165-164: task 6: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 33, in <module>
+    sys.exit(load_entry_point('torch==2.1.1', 'console_scripts', 'torchrun')())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 878, in _invoke_run
+    run_result = self._monitor_workers(self._worker_group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/local_elastic_agent.py", line 307, in _monitor_workers
+    result = self._pcontext.wait(0)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 288, in wait
+    return self._poll()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 668, in _poll
+    self.close()  # terminate all running procs
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 331, in close
+    self._close(death_sig=death_sig, timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 713, in _close
+    handler.proc.wait(time_to_wait)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1209, in wait
+    return self._wait(timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1953, in _wait
+    time.sleep(delay)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 311287 got signal: 15
+srun: error: ip-26-0-172-116: task 9: Exited with exit code 1
+srun: error: ip-26-0-165-131: task 5: Exited with exit code 1
+srun: error: ip-26-0-160-103: task 1: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 33, in <module>
+    sys.exit(load_entry_point('torch==2.1.1', 'console_scripts', 'torchrun')())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 105316 got signal: 15
+srun: error: ip-26-0-162-14: task 2: Exited with exit code 1
+srun: error: ip-26-0-173-246: task 14: Exited with exit code 1
+srun: Force Terminated StepId=13458915.0
diff --git a/logs/13459846-bench_1.34G_dp4_tp1_pp2_acc64_mbs1_seq4096_zero1_tpmodeRED_vocab131k.out b/logs/13459846-bench_1.34G_dp4_tp1_pp2_acc64_mbs1_seq4096_zero1_tpmodeRED_vocab131k.out
new file mode 100644
index 0000000000000000000000000000000000000000..4cfb76a0bff618e2ed1a7b123d35c3f66354d815
--- /dev/null
+++ b/logs/13459846-bench_1.34G_dp4_tp1_pp2_acc64_mbs1_seq4096_zero1_tpmodeRED_vocab131k.out
@@ -0,0 +1,624 @@
++ source /etc/profile.d/modules.sh
+++ . /usr/share/modules/init/bash
++++ unset _mlshdbg
++++ '[' 0 = 1 ']'
++++ unset _mlre _mlIFS
++++ '[' -n x ']'
++++ _mlIFS=' 	
+'
++++ IFS=' '
++++ '[' -n '' ']'
+++++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash autoinit
++++ _mlcode='module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+test 0;'
++++ _mlret=0
++++ '[' -n x ']'
++++ IFS=' 	
+'
++++ unset _mlIFS
++++ unset _mlre _mlv _mlrv
++++ '[' 0 -eq 0 ']'
++++ eval 'module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+test 0;'
+++++ MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl
+++++ export MODULES_CMD
+++++ ENV=/usr/share/modules/init/profile.sh
+++++ export ENV
+++++ BASH_ENV=/usr/share/modules/init/bash
+++++ export BASH_ENV
+++++ MODULESHOME=/usr/share/modules
+++++ export MODULESHOME
+++++ test 0
++++ '[' 0 = 1 ']'
++++ '[' -t 2 ']'
++++ export -f module
++++ export -f switchml
++++ '[' 5 -ge 3 ']'
++++ [[ ehxB =~ i ]]
++++ [[ ! :/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin: =~ :/usr/bin: ]]
++++ '[' '!' -n x ']'
+++++ manpath
++++ [[ ! :/admin/home/nouamane/.local/share/man:/fsx/nouamane/miniconda/envs/2-1-cu121/man:/fsx/nouamane/miniconda/envs/2-1-cu121/share/man:/fsx/nouamane/miniconda/man:/fsx/nouamane/miniconda/share/man:/opt/slurm/share/man:/opt/amazon/openmpi/share/man:/opt/amazon/efa/share/man:/usr/local/man:/usr/local/share/man:/usr/share/man:/admin/home/nouamane/.fzf/man:: =~ :/usr/share/man: ]]
++++ unset _mlcode _mlret
++++ '[' -n '' ']'
++ module load cuda/12.1
++ unset _mlshdbg
++ '[' 0 = 1 ']'
++ unset _mlre _mlIFS
++ '[' -n x ']'
++ _mlIFS=' 	
+'
++ IFS=' '
++ '[' -n '' ']'
+++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash load cuda/12.1
++ eval 'CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include;' export 'CPATH;
+LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/opt/aws-ofi-nccl/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/targets/x86_64-linux/lib/:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/lib:/usr/lib;' export 'LD_LIBRARY_PATH;
+FI_EFA_FORK_SAFE=1;' export 'FI_EFA_FORK_SAFE;
+MANPATH=/usr/local/cuda-12.1/share/man::;' export 'MANPATH;
+LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64;' export 'LIBRARY_PATH;
+_LMFILES_=/usr/share/modules/modulefiles/cuda/12.1;' export '_LMFILES_;
+LOADEDMODULES=cuda/12.1;' export 'LOADEDMODULES;
+MPI_PATH=/opt/amazon/openmpi;' export 'MPI_PATH;
+NCCL_HOME_modshare=/opt/nccl/build:1;' export 'NCCL_HOME_modshare;
+NCCL_PROTO=simple;' export 'NCCL_PROTO;
+MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1;' export 'MANPATH_modshare;
+LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1;' export 'LIBRARY_PATH_modshare;
+NCCL_SOCKET_IFNAME=enp;' export 'NCCL_SOCKET_IFNAME;
+AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl;' export 'AWS_OFI_NCCL_HOME;
+NCCL_HOME=/opt/nccl/build;' export 'NCCL_HOME;
+FI_PROVIDER=efa;' export 'FI_PROVIDER;
+AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1;' export 'AWS_OFI_NCCL_HOME_modshare;
+CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1;' export 'CPATH_modshare;
+LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib/:1:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2:/opt/aws-ofi-nccl/lib:2:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/efa/lib:1:/usr/lib:1;' export 'LD_LIBRARY_PATH_modshare;
+FI_EFA_ENABLE_SHM_TRANSFER=1;' export 'FI_EFA_ENABLE_SHM_TRANSFER;
+_LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1;' export '_LMFILES__modshare;
+LOADEDMODULES_modshare=cuda/12.1:1;' export 'LOADEDMODULES_modshare;
+MPI_PATH_modshare=/opt/amazon/openmpi:1;' export 'MPI_PATH_modshare;
+PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin;' export 'PATH;
+CUDA_HOME=/usr/local/cuda-12.1;' export 'CUDA_HOME;
+PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/usr/local/cuda-12.1/include:1:/opt/amazon/efa/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/usr/sbin:1:/fsx/nouamane/miniconda/bin:1:/opt/slurm/bin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:2:/opt/amazon/openmpi/bin:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1;' export 'PATH_modshare;
+test' '0;'
+++ CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include
+++ export CPATH
+++ LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/opt/aws-ofi-nccl/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/targets/x86_64-linux/lib/:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/lib:/usr/lib
+++ export LD_LIBRARY_PATH
+++ FI_EFA_FORK_SAFE=1
+++ export FI_EFA_FORK_SAFE
+++ MANPATH=/usr/local/cuda-12.1/share/man::
+++ export MANPATH
+++ LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64
+++ export LIBRARY_PATH
+++ _LMFILES_=/usr/share/modules/modulefiles/cuda/12.1
+++ export _LMFILES_
+++ LOADEDMODULES=cuda/12.1
+++ export LOADEDMODULES
+++ MPI_PATH=/opt/amazon/openmpi
+++ export MPI_PATH
+++ NCCL_HOME_modshare=/opt/nccl/build:1
+++ export NCCL_HOME_modshare
+++ NCCL_PROTO=simple
+++ export NCCL_PROTO
+++ MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1
+++ export MANPATH_modshare
+++ LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1
+++ export LIBRARY_PATH_modshare
+++ NCCL_SOCKET_IFNAME=enp
+++ export NCCL_SOCKET_IFNAME
+++ AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl
+++ export AWS_OFI_NCCL_HOME
+++ NCCL_HOME=/opt/nccl/build
+++ export NCCL_HOME
+++ FI_PROVIDER=efa
+++ export FI_PROVIDER
+++ AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1
+++ export AWS_OFI_NCCL_HOME_modshare
+++ CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1
+++ export CPATH_modshare
+++ LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib/:1:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2:/opt/aws-ofi-nccl/lib:2:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/efa/lib:1:/usr/lib:1
+++ export LD_LIBRARY_PATH_modshare
+++ FI_EFA_ENABLE_SHM_TRANSFER=1
+++ export FI_EFA_ENABLE_SHM_TRANSFER
+++ _LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1
+++ export _LMFILES__modshare
+++ LOADEDMODULES_modshare=cuda/12.1:1
+++ export LOADEDMODULES_modshare
+++ MPI_PATH_modshare=/opt/amazon/openmpi:1
+++ export MPI_PATH_modshare
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin
+++ export PATH
+++ CUDA_HOME=/usr/local/cuda-12.1
+++ export CUDA_HOME
+++ PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/usr/local/cuda-12.1/include:1:/opt/amazon/efa/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/usr/sbin:1:/fsx/nouamane/miniconda/bin:1:/opt/slurm/bin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:2:/opt/amazon/openmpi/bin:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1
+++ export PATH_modshare
+++ test 0
++ _mlstatus=0
++ '[' -n x ']'
++ IFS=' 	
+'
++ unset _mlre _mlv _mlrv _mlIFS
++ '[' -n '' ']'
++ unset _mlshdbg
++ return 0
++ source /fsx/nouamane/miniconda/bin/activate
+++ _CONDA_ROOT=/fsx/nouamane/miniconda
+++ . /fsx/nouamane/miniconda/etc/profile.d/conda.sh
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ '[' -z x ']'
+++ conda activate
+++ local cmd=activate
+++ case "$cmd" in
+++ __conda_activate activate
+++ '[' -n '' ']'
+++ local ask_conda
++++ PS1=
++++ __conda_exe shell.posix activate
++++ /fsx/nouamane/miniconda/bin/conda shell.posix activate
+++ ask_conda='PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin'\''
+export CONDA_SHLVL='\''1'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\'''
+++ eval 'PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin'\''
+export CONDA_SHLVL='\''1'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\'''
++++ PS1='(base) '
++++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin
++++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin
++++ export CONDA_SHLVL=1
++++ CONDA_SHLVL=1
++++ export 'CONDA_PROMPT_MODIFIER=(base) '
++++ CONDA_PROMPT_MODIFIER='(base) '
+++ __conda_hashr
+++ '[' -n '' ']'
+++ '[' -n '' ']'
+++ hash -r
++ conda activate 2-1-cu121
++ local cmd=activate
++ case "$cmd" in
++ __conda_activate activate 2-1-cu121
++ '[' -n '' ']'
++ local ask_conda
+++ PS1='(base) '
+++ __conda_exe shell.posix activate 2-1-cu121
+++ /fsx/nouamane/miniconda/bin/conda shell.posix activate 2-1-cu121
++ ask_conda='PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''2'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_1='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
++ eval 'PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''2'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_1='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
+++ PS1='(2-1-cu121) '
+++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin
+++ export CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ export CONDA_SHLVL=2
+++ CONDA_SHLVL=2
+++ export CONDA_DEFAULT_ENV=2-1-cu121
+++ CONDA_DEFAULT_ENV=2-1-cu121
+++ export 'CONDA_PROMPT_MODIFIER=(2-1-cu121) '
+++ CONDA_PROMPT_MODIFIER='(2-1-cu121) '
+++ export CONDA_PREFIX_1=/fsx/nouamane/miniconda
+++ CONDA_PREFIX_1=/fsx/nouamane/miniconda
+++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ export _CE_M=
+++ _CE_M=
+++ export _CE_CONDA=
+++ _CE_CONDA=
+++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh
++++ test -n ''
++++ xml_catalog_files_libxml2=
++++ XML_CATALOG_FILES=
++++ conda_catalog_files=
++++ ifs_libxml2=' 	
+'
++++ IFS=' '
++++ rem=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ for pre in ${rem}
++++ test '' = /fsx/nouamane/miniconda/envs/2-1-cu121
++++ conda_catalog_files=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ rem=
++++ IFS=' 	
+'
++++ conda_catalog_files='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ export 'XML_CATALOG_FILES=file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ XML_CATALOG_FILES='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ unset conda_catalog_files ifs_libxml2 rem
++ __conda_hashr
++ '[' -n '' ']'
++ '[' -n '' ']'
++ hash -r
++ export PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin
++ PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin
+++ scontrol show hostnames ip-26-0-161-123
++ export NODELIST=ip-26-0-161-123
++ NODELIST=ip-26-0-161-123
+++ scontrol show hostnames ip-26-0-161-123
+++ head -n1
++ export MASTER_NODE=ip-26-0-161-123
++ MASTER_NODE=ip-26-0-161-123
++ export MASTER_PORT=12356
++ MASTER_PORT=12356
++ export NNODES=1
++ NNODES=1
++ export GPUS_PER_NODE=8
++ GPUS_PER_NODE=8
++ export WORLD_SIZE=8
++ WORLD_SIZE=8
++ export CUDA_DEVICE_MAX_CONNECTIONS=1
++ CUDA_DEVICE_MAX_CONNECTIONS=1
++ export NCCL_DEBUG=WARN
++ NCCL_DEBUG=WARN
++ export NANOTRON_BENCHMARK=1
++ NANOTRON_BENCHMARK=1
++ export WANDB_MODE=disabled
++ WANDB_MODE=disabled
++ export TORCH_NCCL_ASYNC_ERROR_HANDLING=1
++ TORCH_NCCL_ASYNC_ERROR_HANDLING=1
++ echo '=== GPU Topology ==='
+=== GPU Topology ===
++ nvidia-smi topo -m
+	GPU0	GPU1	GPU2	GPU3	GPU4	GPU5	GPU6	GPU7	CPU Affinity	NUMA Affinity	GPU NUMA ID
+GPU0	 X 	NV18	NV18	NV18	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU1	NV18	 X 	NV18	NV18	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU2	NV18	NV18	 X 	NV18	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU3	NV18	NV18	NV18	 X 	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU4	NV18	NV18	NV18	NV18	 X 	NV18	NV18	NV18	48-95	1		N/A
+GPU5	NV18	NV18	NV18	NV18	NV18	 X 	NV18	NV18	48-95	1		N/A
+GPU6	NV18	NV18	NV18	NV18	NV18	NV18	 X 	NV18	48-95	1		N/A
+GPU7	NV18	NV18	NV18	NV18	NV18	NV18	NV18	 X 	48-95	1		N/A
+
+Legend:
+
+  X    = Self
+  SYS  = Connection traversing PCIe as well as the SMP interconnect between NUMA nodes (e.g., QPI/UPI)
+  NODE = Connection traversing PCIe as well as the interconnect between PCIe Host Bridges within a NUMA node
+  PHB  = Connection traversing PCIe as well as a PCIe Host Bridge (typically the CPU)
+  PXB  = Connection traversing multiple PCIe bridges (without traversing the PCIe Host Bridge)
+  PIX  = Connection traversing at most a single PCIe bridge
+  NV#  = Connection traversing a bonded set of # NVLinks
++ echo ==================
+==================
++ echo 'Master node: ip-26-0-161-123'
+Master node: ip-26-0-161-123
++ echo 'All nodes: ip-26-0-161-123'
+All nodes: ip-26-0-161-123
++ echo 'World size: 8'
+World size: 8
++ srun --wait=0 --kill-on-bad-exit=1 torchrun --nnodes=1 --nproc_per_node=8 --rdzv_id=13459846 --rdzv_backend=c10d --rdzv_endpoint=ip-26-0-161-123:12356 --max_restarts 0 --rdzv_conf timeout=60 /fsx/nouamane/projects/nanotron/run_train.py --config-file benchmark/configs/config_1.34G_dp4_tp1_pp2_acc64_mbs1_seq4096_zero1_tpmodeRED_vocab131k.yaml
+[2024-12-15 12:21:44,557] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-15 12:21:44,557] torch.distributed.run: [WARNING] 
+[2024-12-15 12:21:44,557] torch.distributed.run: [WARNING] *****************************************
+[2024-12-15 12:21:44,557] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-15 12:21:44,557] torch.distributed.run: [WARNING] *****************************************
+NCCL version 2.18.5+cuda12.2
+12/15/2024 12:22:04 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]: Measuring inter-GPU and intra-node bandwidth...
+12/15/2024 12:22:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]: Bandwidth measurement complete. Time taken: 14.44 seconds
+12/15/2024 12:22:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]: Config:
+12/15/2024 12:22:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]: Config(general=GeneralArgs(project='debug',
+12/15/2024 12:22:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]:                            run='1.34G_dp4_tp1_pp2_acc64_mbs1_seq4096_zero1_tpmodeRED_vocab131k',
+12/15/2024 12:22:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]:                            seed=42,
+12/15/2024 12:22:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]:                            step=None,
+12/15/2024 12:22:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]:                            consumed_train_samples=None,
+12/15/2024 12:22:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]:                            benchmark_csv_path=PosixPath('benchmark/results/bench_final2.csv'),
+12/15/2024 12:22:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]:                            ignore_sanity_checks=True),
+12/15/2024 12:22:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]:        parallelism=ParallelismArgs(dp=4,
+12/15/2024 12:22:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]:                                    pp=2,
+12/15/2024 12:22:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]:                                    tp=1,
+12/15/2024 12:22:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]:                                    pp_engine=<nanotron.parallel.pipeline_parallel.engine.OneForwardOneBackwardPipelineEngine object at 0x7f553084cbe0>,
+12/15/2024 12:22:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]:                                    tp_mode=<TensorParallelLinearMode.REDUCE_SCATTER: 2>,
+12/15/2024 12:22:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]:                                    tp_linear_async_communication=True,
+12/15/2024 12:22:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]:                                    recompute_layer=False,
+12/15/2024 12:22:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]:                                    tp_recompute_allgather=True,
+12/15/2024 12:22:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]:                                    expert_parallel_size=1),
+12/15/2024 12:22:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]:        model=ModelArgs(model_config=LlamaConfig(bos_token_id=0,
+12/15/2024 12:22:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]:                                                 eos_token_id=0,
+12/15/2024 12:22:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]:                                                 hidden_act='silu',
+12/15/2024 12:22:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]:                                                 hidden_size=2048,
+12/15/2024 12:22:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]:                                                 initializer_range=0.02,
+12/15/2024 12:22:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]:                                                 intermediate_size=8192,
+12/15/2024 12:22:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]:                                                 is_llama_config=True,
+12/15/2024 12:22:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]:                                                 max_position_embeddings=4096,
+12/15/2024 12:22:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]:                                                 num_attention_heads=32,
+12/15/2024 12:22:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]:                                                 num_hidden_layers=16,
+12/15/2024 12:22:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]:                                                 num_key_value_heads=32,
+12/15/2024 12:22:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]:                                                 pad_token_id=None,
+12/15/2024 12:22:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]:                                                 pretraining_tp=1,
+12/15/2024 12:22:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]:                                                 rms_norm_eps=1e-05,
+12/15/2024 12:22:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]:                                                 rope_scaling=None,
+12/15/2024 12:22:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]:                                                 rope_theta=10000.0,
+12/15/2024 12:22:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]:                                                 rope_interleaved=False,
+12/15/2024 12:22:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]:                                                 tie_word_embeddings=True,
+12/15/2024 12:22:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]:                                                 use_cache=True,
+12/15/2024 12:22:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]:                                                 vocab_size=131072),
+12/15/2024 12:22:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]:                        init_method=RandomInit(std=0.02),
+12/15/2024 12:22:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]:                        dtype=torch.bfloat16,
+12/15/2024 12:22:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]:                        make_vocab_size_divisible_by=1,
+12/15/2024 12:22:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]:                        ddp_bucket_cap_mb=25),
+12/15/2024 12:22:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]:        tokenizer=TokenizerArgs(tokenizer_name_or_path='robot-test/dummy-tokenizer-wordlevel',
+12/15/2024 12:22:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]:                                tokenizer_revision=None,
+12/15/2024 12:22:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]:                                tokenizer_max_length=None),
+12/15/2024 12:22:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]:        checkpoints=CheckpointsArgs(checkpoints_path=PosixPath('checkpoints'),
+12/15/2024 12:22:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]:                                    checkpoint_interval=10000,
+12/15/2024 12:22:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]:                                    save_initial_state=False,
+12/15/2024 12:22:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]:                                    save_final_state=False,
+12/15/2024 12:22:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]:                                    resume_checkpoint_path=None,
+12/15/2024 12:22:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]:                                    checkpoints_path_is_shared_file_system=False),
+12/15/2024 12:22:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]:        logging=LoggingArgs(log_level='info',
+12/15/2024 12:22:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]:                            log_level_replica='info',
+12/15/2024 12:22:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]:                            iteration_step_info_interval=1),
+12/15/2024 12:22:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]:        tokens=TokensArgs(sequence_length=4096,
+12/15/2024 12:22:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]:                          train_steps=100,
+12/15/2024 12:22:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]:                          micro_batch_size=1,
+12/15/2024 12:22:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]:                          batch_accumulation_per_replica=64,
+12/15/2024 12:22:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]:                          val_check_interval=100,
+12/15/2024 12:22:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]:                          limit_val_batches=0,
+12/15/2024 12:22:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]:                          limit_test_batches=0),
+12/15/2024 12:22:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]:        optimizer=OptimizerArgs(optimizer_factory=AdamWOptimizerArgs(adam_eps=1e-08,
+12/15/2024 12:22:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]:                                                                     adam_beta1=0.9,
+12/15/2024 12:22:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]:                                                                     adam_beta2=0.95,
+12/15/2024 12:22:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]:                                                                     torch_adam_is_fused=True,
+12/15/2024 12:22:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]:                                                                     name='adamW'),
+12/15/2024 12:22:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]:                                zero_stage=1,
+12/15/2024 12:22:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]:                                weight_decay=0.01,
+12/15/2024 12:22:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]:                                clip_grad=1.0,
+12/15/2024 12:22:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]:                                accumulate_grad_in_fp32=True,
+12/15/2024 12:22:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]:                                learning_rate_scheduler=LRSchedulerArgs(learning_rate=0.0003,
+12/15/2024 12:22:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]:                                                                        lr_warmup_steps=2,
+12/15/2024 12:22:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]:                                                                        lr_warmup_style='linear',
+12/15/2024 12:22:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]:                                                                        lr_decay_style='cosine',
+12/15/2024 12:22:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]:                                                                        lr_decay_steps=13,
+12/15/2024 12:22:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]:                                                                        lr_decay_starting_step=None,
+12/15/2024 12:22:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]:                                                                        min_decay_lr=1e-05)),
+12/15/2024 12:22:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]:        data_stages=[DatasetStageArgs(name='Stable Training Stage',
+12/15/2024 12:22:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]:                                      start_training_step=1,
+12/15/2024 12:22:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]:                                      data=DataArgs(dataset=None,
+12/15/2024 12:22:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]:                                                    seed=42,
+12/15/2024 12:22:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]:                                                    num_loading_workers=1))],
+12/15/2024 12:22:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]:        profiler=None,
+12/15/2024 12:22:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]:        lighteval=None,
+12/15/2024 12:22:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]:        s3_upload=None)
+12/15/2024 12:22:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]: Model Config:
+12/15/2024 12:22:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]: LlamaConfig(bos_token_id=0,
+12/15/2024 12:22:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]:             eos_token_id=0,
+12/15/2024 12:22:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]:             hidden_act='silu',
+12/15/2024 12:22:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]:             hidden_size=2048,
+12/15/2024 12:22:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]:             initializer_range=0.02,
+12/15/2024 12:22:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]:             intermediate_size=8192,
+12/15/2024 12:22:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]:             is_llama_config=True,
+12/15/2024 12:22:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]:             max_position_embeddings=4096,
+12/15/2024 12:22:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]:             num_attention_heads=32,
+12/15/2024 12:22:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]:             num_hidden_layers=16,
+12/15/2024 12:22:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]:             num_key_value_heads=32,
+12/15/2024 12:22:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]:             pad_token_id=None,
+12/15/2024 12:22:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]:             pretraining_tp=1,
+12/15/2024 12:22:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]:             rms_norm_eps=1e-05,
+12/15/2024 12:22:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]:             rope_scaling=None,
+12/15/2024 12:22:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]:             rope_theta=10000.0,
+12/15/2024 12:22:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]:             rope_interleaved=False,
+12/15/2024 12:22:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]:             tie_word_embeddings=True,
+12/15/2024 12:22:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]:             use_cache=True,
+12/15/2024 12:22:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]:             vocab_size=131072)
+12/15/2024 12:22:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]: Building model..
+12/15/2024 12:22:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]: Initialize RoPE Theta = 10000.0
+12/15/2024 12:22:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]: Setting PP block ranks...
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+12/15/2024 12:22:21 [INFO|DP=0|PP=1|TP=0|ip-26-0-161-123]: Local number of parameters: 604M (1152.04MiB)
+12/15/2024 12:22:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]: Total number of parameters: 1.61G (3072.13MiB)
+12/15/2024 12:22:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]: Local number of parameters: 1.01G (1920.09MiB)
+12/15/2024 12:22:21 [INFO|DP=0|PP=1|TP=0|ip-26-0-161-123]: [After model building] Memory usage: 1152.06MiB. Peak allocated: 5632.00MiB Peak reserved: 11778.00MiB
+12/15/2024 12:22:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]: [After model building] Memory usage: 1920.11MiB. Peak allocated: 5632.00MiB Peak reserved: 12802.00MiB
+12/15/2024 12:22:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]: No checkpoint path provided.
+12/15/2024 12:22:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]: Parametrizing model parameters using StandardParametrizator
+12/15/2024 12:22:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]: [Optimizer Building] Using LearningRateForSP as learning rate
+12/15/2024 12:22:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]: [ZeRO sharding] Size of optimizer params per rank:
+12/15/2024 12:22:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]: [ZeRO sharding] DP Rank 0 has 252M out of 1.01G (25.00%) params' optimizer states
+12/15/2024 12:22:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]: [ZeRO sharding] DP Rank 1 has 252M out of 1.01G (25.00%) params' optimizer states
+12/15/2024 12:22:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]: [ZeRO sharding] DP Rank 2 has 252M out of 1.01G (25.00%) params' optimizer states
+12/15/2024 12:22:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]: [ZeRO sharding] DP Rank 3 has 252M out of 1.01G (25.00%) params' optimizer states
+12/15/2024 12:22:26 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]: [Training Plan] Stage Stable Training Stage has 99 remaining training steps and has consumed 0 samples
+12/15/2024 12:22:26 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]: Using dummy data generator
+12/15/2024 12:22:26 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]: [Training Plan] There are 1 training stages 
+12/15/2024 12:22:26 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]: [Stage Stable Training Stage] start from step 1 
+12/15/2024 12:22:26 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]: 
+12/15/2024 12:22:26 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]: [Start training] datetime: 2024-12-15 12:22:26.380997 | mbs: 1 | grad_accum: 64 | global_batch_size: 256 | sequence_length: 4096 | train_steps: 100 | start_iteration_step: 0 | consumed_train_samples: 0
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+12/15/2024 12:22:26 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]: Resuming training from stage Stable Training Stage, it has trained for 0 samples and has 99 remaining train steps
+12/15/2024 12:22:26 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]:  Memory usage: 6720.32MiB. Peak allocated 6720.32MiB. Peak reserved: 16644.00MiB
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at /opt/conda/conda-bld/pytorch_1699449201336/work/torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at /opt/conda/conda-bld/pytorch_1699449201336/work/torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at /opt/conda/conda-bld/pytorch_1699449201336/work/torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at /opt/conda/conda-bld/pytorch_1699449201336/work/torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+12/15/2024 12:22:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]:  Memory usage: 6789.86MiB. Peak allocated 16129.61MiB. Peak reserved: 16908.00MiB
+12/15/2024 12:22:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]:  Memory usage: 8709.97MiB. Peak allocated 13030.17MiB. Peak reserved: 20758.00MiB
+12/15/2024 12:22:36 [INFO|DP=0|PP=1|TP=0|ip-26-0-161-123]: iteration: 1 / 100 | consumed_tokens: 1.05M | elapsed_time_per_iteration_ms: 8.71K | tokens_per_sec: 120K | tokens_per_sec_per_gpu: 15.1K | global_batch_size: 256 | lm_loss: 12 | lr: 0.00015 | model_tflops_per_gpu: 145 | hardware_tflops_per_gpu: 145 | grad_norm: 0.847 | cuda_memory_allocated: 5.51G | cuda_max_memory_reserved: 14G | hd_total_memory_tb: 312G | hd_used_memory_tb: 68.1G | hd_free_memory_tb: 244G
+12/15/2024 12:22:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]:  Memory usage: 8709.97MiB. Peak allocated 18049.73MiB. Peak reserved: 20758.00MiB
+12/15/2024 12:22:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]:  Memory usage: 8709.97MiB. Peak allocated 13030.17MiB. Peak reserved: 20758.00MiB
+12/15/2024 12:22:40 [INFO|DP=0|PP=1|TP=0|ip-26-0-161-123]: iteration: 2 / 100 | consumed_tokens: 2.1M | elapsed_time_per_iteration_ms: 3.84K | tokens_per_sec: 273K | tokens_per_sec_per_gpu: 34.1K | global_batch_size: 256 | lm_loss: 12 | lr: 0.0003 | model_tflops_per_gpu: 330 | hardware_tflops_per_gpu: 330 | grad_norm: 0.847 | cuda_memory_allocated: 5.51G | cuda_max_memory_reserved: 14G | hd_total_memory_tb: 312G | hd_used_memory_tb: 68.1G | hd_free_memory_tb: 244G
+12/15/2024 12:22:44 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]:  Memory usage: 8709.97MiB. Peak allocated 18049.73MiB. Peak reserved: 20758.00MiB
+12/15/2024 12:22:44 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]: | job_id   | name                                                           | nodes | seq_len | mbs | batch_accum | gbs | mTFLOPs | hTFLOPs | tok/s/gpu | AllReduce (GB/s) | AllGather (GB/s) | ReduceScatter (GB/s) | AR Intra-node (GB/s) | AG Intra-node (GB/s) | RS Intra-node (GB/s) | Mem Alloc (GB) | Mem Res (GB) | dp | pp | tp | pp_engine                           | tp_mode        | tp_async_comm | hidden_size | hidden_act | num_layers | num_heads | num_kv_heads | max_pos | vocab_size | tie_word_embeddings | dtype          | zero_stage | ddp_bucket_cap_mb | accumulate_grad_in_fp32 | Total Params | Local Params |
+12/15/2024 12:22:44 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]: | -------- | -------------------------------------------------------------- | ----- | ------- | --- | ----------- | --- | ------- | ------- | --------- | ---------------- | ---------------- | -------------------- | -------------------- | -------------------- | -------------------- | -------------- | ------------ | -- | -- | -- | ----------------------------------- | -------------- | ------------- | ----------- | ---------- | ---------- | --------- | ------------ | ------- | ---------- | ------------------- | -------------- | ---------- | ----------------- | ----------------------- | ------------ | ------------ |
+12/15/2024 12:22:44 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]: | 13459846 | 1.34G_dp4_tp1_pp2_acc64_mbs1_seq4096_zero1_tpmodeRED_vocab131k | 1     | 4096    | 1   | 64          | 256 | 329.75  | 329.75  | 34122.33  | 457.10           | 262.89           | 260.80               | 455.98               | 263.18               | 260.70               | 12.72          | 20.27        | 4  | 2  | 1  | OneForwardOneBackwardPipelineEngine | REDUCE_SCATTER | True          | 2048        | silu       | 16         | 32        | 32           | 4096    | 131072     | True                | torch.bfloat16 | 1          | 25                | True                    | 1.61G        | 1.01G        |
+12/15/2024 12:22:44 [INFO|DP=0|PP=1|TP=0|ip-26-0-161-123]: iteration: 3 / 100 | consumed_tokens: 3.15M | elapsed_time_per_iteration_ms: 3.84K | tokens_per_sec: 273K | tokens_per_sec_per_gpu: 34.1K | global_batch_size: 256 | lm_loss: 12 | lr: 0.000296 | model_tflops_per_gpu: 330 | hardware_tflops_per_gpu: 330 | grad_norm: 0.838 | cuda_memory_allocated: 5.51G | cuda_max_memory_reserved: 14G | hd_total_memory_tb: 312G | hd_used_memory_tb: 68.1G | hd_free_memory_tb: 244G
+12/15/2024 12:22:44 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]: Attempting to write benchmark results to CSV file: benchmark/results/bench_final2.csv
+12/15/2024 12:22:44 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]: Acquired lock for CSV file: benchmark/results/bench_final2.csv
+12/15/2024 12:22:44 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]: Successfully wrote to CSV file: benchmark/results/bench_final2.csv. Releasing lock...
+12/15/2024 12:22:44 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-123]: Throughput logging complete
+[2024-12-15 12:22:54,847] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 236712 closing signal SIGTERM
+[2024-12-15 12:22:54,847] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 236713 closing signal SIGTERM
+[2024-12-15 12:22:54,847] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 236714 closing signal SIGTERM
+[2024-12-15 12:22:54,847] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 236715 closing signal SIGTERM
+[2024-12-15 12:22:54,848] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 236716 closing signal SIGTERM
+[2024-12-15 12:22:54,848] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 236717 closing signal SIGTERM
+[2024-12-15 12:22:54,848] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 236718 closing signal SIGTERM
+[2024-12-15 12:22:57,342] torch.distributed.elastic.multiprocessing.api: [ERROR] failed (exitcode: 1) local_rank: 0 (pid: 236711) of binary: /fsx/nouamane/miniconda/envs/2-1-cu121/bin/python
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 33, in <module>
+    sys.exit(load_entry_point('torch==2.1.1', 'console_scripts', 'torchrun')())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 264, in launch_agent
+    raise ChildFailedError(
+torch.distributed.elastic.multiprocessing.errors.ChildFailedError: 
+============================================================
+/fsx/nouamane/projects/nanotron/run_train.py FAILED
+------------------------------------------------------------
+Failures:
+  <NO_OTHER_FAILURES>
+------------------------------------------------------------
+Root Cause (first observed failure):
+[0]:
+  time      : 2024-12-15_12:22:54
+  host      : ip-26-0-161-123.ec2.internal
+  rank      : 0 (local_rank: 0)
+  exitcode  : 1 (pid: 236711)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+============================================================
+srun: error: ip-26-0-161-123: task 0: Exited with exit code 1
+srun: launch/slurm: _step_signal: Terminating StepId=13459846.0
diff --git a/logs/13459892-bench_80G_dp2_tp2_pp4_acc128_mbs1_seq4096_zero1_tpmodeRED_vocab131k.out b/logs/13459892-bench_80G_dp2_tp2_pp4_acc128_mbs1_seq4096_zero1_tpmodeRED_vocab131k.out
new file mode 100644
index 0000000000000000000000000000000000000000..6d3500f742398f6760ddaf39451884f586be4f0d
--- /dev/null
+++ b/logs/13459892-bench_80G_dp2_tp2_pp4_acc128_mbs1_seq4096_zero1_tpmodeRED_vocab131k.out
@@ -0,0 +1,933 @@
++ source /etc/profile.d/modules.sh
+++ . /usr/share/modules/init/bash
++++ unset _mlshdbg
++++ '[' 0 = 1 ']'
++++ unset _mlre _mlIFS
++++ '[' -n x ']'
++++ _mlIFS=' 	
+'
++++ IFS=' '
++++ '[' -n '' ']'
+++++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash autoinit
++++ _mlcode='module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+test 0;'
++++ _mlret=0
++++ '[' -n x ']'
++++ IFS=' 	
+'
++++ unset _mlIFS
++++ unset _mlre _mlv _mlrv
++++ '[' 0 -eq 0 ']'
++++ eval 'module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+test 0;'
+++++ MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl
+++++ export MODULES_CMD
+++++ ENV=/usr/share/modules/init/profile.sh
+++++ export ENV
+++++ BASH_ENV=/usr/share/modules/init/bash
+++++ export BASH_ENV
+++++ MODULESHOME=/usr/share/modules
+++++ export MODULESHOME
+++++ test 0
++++ '[' 0 = 1 ']'
++++ '[' -t 2 ']'
++++ export -f module
++++ export -f switchml
++++ '[' 5 -ge 3 ']'
++++ [[ ehxB =~ i ]]
++++ [[ ! :/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin: =~ :/usr/bin: ]]
++++ '[' '!' -n x ']'
+++++ manpath
++++ [[ ! :/admin/home/nouamane/.local/share/man:/fsx/nouamane/miniconda/envs/2-1-cu121/man:/fsx/nouamane/miniconda/envs/2-1-cu121/share/man:/fsx/nouamane/miniconda/man:/fsx/nouamane/miniconda/share/man:/opt/slurm/share/man:/opt/amazon/openmpi/share/man:/opt/amazon/efa/share/man:/usr/local/man:/usr/local/share/man:/usr/share/man:/admin/home/nouamane/.fzf/man:: =~ :/usr/share/man: ]]
++++ unset _mlcode _mlret
++++ '[' -n '' ']'
++ module load cuda/12.1
++ unset _mlshdbg
++ '[' 0 = 1 ']'
++ unset _mlre _mlIFS
++ '[' -n x ']'
++ _mlIFS=' 	
+'
++ IFS=' '
++ '[' -n '' ']'
+++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash load cuda/12.1
++ eval 'CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include;' export 'CPATH;
+LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/opt/aws-ofi-nccl/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/targets/x86_64-linux/lib/:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/lib:/usr/lib;' export 'LD_LIBRARY_PATH;
+FI_EFA_FORK_SAFE=1;' export 'FI_EFA_FORK_SAFE;
+MANPATH=/usr/local/cuda-12.1/share/man::;' export 'MANPATH;
+LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64;' export 'LIBRARY_PATH;
+_LMFILES_=/usr/share/modules/modulefiles/cuda/12.1;' export '_LMFILES_;
+LOADEDMODULES=cuda/12.1;' export 'LOADEDMODULES;
+MPI_PATH=/opt/amazon/openmpi;' export 'MPI_PATH;
+NCCL_HOME_modshare=/opt/nccl/build:1;' export 'NCCL_HOME_modshare;
+NCCL_PROTO=simple;' export 'NCCL_PROTO;
+MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1;' export 'MANPATH_modshare;
+LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1;' export 'LIBRARY_PATH_modshare;
+NCCL_SOCKET_IFNAME=enp;' export 'NCCL_SOCKET_IFNAME;
+AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl;' export 'AWS_OFI_NCCL_HOME;
+NCCL_HOME=/opt/nccl/build;' export 'NCCL_HOME;
+FI_PROVIDER=efa;' export 'FI_PROVIDER;
+AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1;' export 'AWS_OFI_NCCL_HOME_modshare;
+CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1;' export 'CPATH_modshare;
+LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib/:1:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2:/opt/aws-ofi-nccl/lib:2:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/efa/lib:1:/usr/lib:1;' export 'LD_LIBRARY_PATH_modshare;
+FI_EFA_ENABLE_SHM_TRANSFER=1;' export 'FI_EFA_ENABLE_SHM_TRANSFER;
+_LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1;' export '_LMFILES__modshare;
+LOADEDMODULES_modshare=cuda/12.1:1;' export 'LOADEDMODULES_modshare;
+MPI_PATH_modshare=/opt/amazon/openmpi:1;' export 'MPI_PATH_modshare;
+PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin;' export 'PATH;
+CUDA_HOME=/usr/local/cuda-12.1;' export 'CUDA_HOME;
+PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/usr/local/cuda-12.1/include:1:/opt/amazon/efa/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/usr/sbin:1:/fsx/nouamane/miniconda/bin:1:/opt/slurm/bin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:2:/opt/amazon/openmpi/bin:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1;' export 'PATH_modshare;
+test' '0;'
+++ CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include
+++ export CPATH
+++ LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/opt/aws-ofi-nccl/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/targets/x86_64-linux/lib/:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/lib:/usr/lib
+++ export LD_LIBRARY_PATH
+++ FI_EFA_FORK_SAFE=1
+++ export FI_EFA_FORK_SAFE
+++ MANPATH=/usr/local/cuda-12.1/share/man::
+++ export MANPATH
+++ LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64
+++ export LIBRARY_PATH
+++ _LMFILES_=/usr/share/modules/modulefiles/cuda/12.1
+++ export _LMFILES_
+++ LOADEDMODULES=cuda/12.1
+++ export LOADEDMODULES
+++ MPI_PATH=/opt/amazon/openmpi
+++ export MPI_PATH
+++ NCCL_HOME_modshare=/opt/nccl/build:1
+++ export NCCL_HOME_modshare
+++ NCCL_PROTO=simple
+++ export NCCL_PROTO
+++ MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1
+++ export MANPATH_modshare
+++ LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1
+++ export LIBRARY_PATH_modshare
+++ NCCL_SOCKET_IFNAME=enp
+++ export NCCL_SOCKET_IFNAME
+++ AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl
+++ export AWS_OFI_NCCL_HOME
+++ NCCL_HOME=/opt/nccl/build
+++ export NCCL_HOME
+++ FI_PROVIDER=efa
+++ export FI_PROVIDER
+++ AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1
+++ export AWS_OFI_NCCL_HOME_modshare
+++ CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1
+++ export CPATH_modshare
+++ LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib/:1:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2:/opt/aws-ofi-nccl/lib:2:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/efa/lib:1:/usr/lib:1
+++ export LD_LIBRARY_PATH_modshare
+++ FI_EFA_ENABLE_SHM_TRANSFER=1
+++ export FI_EFA_ENABLE_SHM_TRANSFER
+++ _LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1
+++ export _LMFILES__modshare
+++ LOADEDMODULES_modshare=cuda/12.1:1
+++ export LOADEDMODULES_modshare
+++ MPI_PATH_modshare=/opt/amazon/openmpi:1
+++ export MPI_PATH_modshare
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin
+++ export PATH
+++ CUDA_HOME=/usr/local/cuda-12.1
+++ export CUDA_HOME
+++ PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/usr/local/cuda-12.1/include:1:/opt/amazon/efa/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/usr/sbin:1:/fsx/nouamane/miniconda/bin:1:/opt/slurm/bin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:2:/opt/amazon/openmpi/bin:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1
+++ export PATH_modshare
+++ test 0
++ _mlstatus=0
++ '[' -n x ']'
++ IFS=' 	
+'
++ unset _mlre _mlv _mlrv _mlIFS
++ '[' -n '' ']'
++ unset _mlshdbg
++ return 0
++ source /fsx/nouamane/miniconda/bin/activate
+++ _CONDA_ROOT=/fsx/nouamane/miniconda
+++ . /fsx/nouamane/miniconda/etc/profile.d/conda.sh
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ '[' -z x ']'
+++ conda activate
+++ local cmd=activate
+++ case "$cmd" in
+++ __conda_activate activate
+++ '[' -n '' ']'
+++ local ask_conda
++++ PS1=
++++ __conda_exe shell.posix activate
++++ /fsx/nouamane/miniconda/bin/conda shell.posix activate
+++ ask_conda='PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin'\''
+export CONDA_SHLVL='\''1'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\'''
+++ eval 'PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin'\''
+export CONDA_SHLVL='\''1'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\'''
++++ PS1='(base) '
++++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin
++++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin
++++ export CONDA_SHLVL=1
++++ CONDA_SHLVL=1
++++ export 'CONDA_PROMPT_MODIFIER=(base) '
++++ CONDA_PROMPT_MODIFIER='(base) '
+++ __conda_hashr
+++ '[' -n '' ']'
+++ '[' -n '' ']'
+++ hash -r
++ conda activate 2-1-cu121
++ local cmd=activate
++ case "$cmd" in
++ __conda_activate activate 2-1-cu121
++ '[' -n '' ']'
++ local ask_conda
+++ PS1='(base) '
+++ __conda_exe shell.posix activate 2-1-cu121
+++ /fsx/nouamane/miniconda/bin/conda shell.posix activate 2-1-cu121
++ ask_conda='PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''2'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_1='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
++ eval 'PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''2'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_1='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
+++ PS1='(2-1-cu121) '
+++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin
+++ export CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ export CONDA_SHLVL=2
+++ CONDA_SHLVL=2
+++ export CONDA_DEFAULT_ENV=2-1-cu121
+++ CONDA_DEFAULT_ENV=2-1-cu121
+++ export 'CONDA_PROMPT_MODIFIER=(2-1-cu121) '
+++ CONDA_PROMPT_MODIFIER='(2-1-cu121) '
+++ export CONDA_PREFIX_1=/fsx/nouamane/miniconda
+++ CONDA_PREFIX_1=/fsx/nouamane/miniconda
+++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ export _CE_M=
+++ _CE_M=
+++ export _CE_CONDA=
+++ _CE_CONDA=
+++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh
++++ test -n ''
++++ xml_catalog_files_libxml2=
++++ XML_CATALOG_FILES=
++++ conda_catalog_files=
++++ ifs_libxml2=' 	
+'
++++ IFS=' '
++++ rem=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ for pre in ${rem}
++++ test '' = /fsx/nouamane/miniconda/envs/2-1-cu121
++++ conda_catalog_files=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ rem=
++++ IFS=' 	
+'
++++ conda_catalog_files='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ export 'XML_CATALOG_FILES=file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ XML_CATALOG_FILES='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ unset conda_catalog_files ifs_libxml2 rem
++ __conda_hashr
++ '[' -n '' ']'
++ '[' -n '' ']'
++ hash -r
++ export PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin
++ PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin
+++ scontrol show hostnames 'ip-26-0-172-[116,142]'
++ export 'NODELIST=ip-26-0-172-116
+ip-26-0-172-142'
++ NODELIST='ip-26-0-172-116
+ip-26-0-172-142'
+++ head -n1
+++ scontrol show hostnames 'ip-26-0-172-[116,142]'
++ export MASTER_NODE=ip-26-0-172-116
++ MASTER_NODE=ip-26-0-172-116
++ export MASTER_PORT=12356
++ MASTER_PORT=12356
++ export NNODES=2
++ NNODES=2
++ export GPUS_PER_NODE=8
++ GPUS_PER_NODE=8
++ export WORLD_SIZE=16
++ WORLD_SIZE=16
++ export CUDA_DEVICE_MAX_CONNECTIONS=1
++ CUDA_DEVICE_MAX_CONNECTIONS=1
++ export NCCL_DEBUG=WARN
++ NCCL_DEBUG=WARN
++ export NANOTRON_BENCHMARK=1
++ NANOTRON_BENCHMARK=1
++ export WANDB_MODE=disabled
++ WANDB_MODE=disabled
++ export TORCH_NCCL_ASYNC_ERROR_HANDLING=1
++ TORCH_NCCL_ASYNC_ERROR_HANDLING=1
++ echo '=== GPU Topology ==='
+=== GPU Topology ===
++ nvidia-smi topo -m
+	GPU0	GPU1	GPU2	GPU3	GPU4	GPU5	GPU6	GPU7	CPU Affinity	NUMA Affinity	GPU NUMA ID
+GPU0	 X 	NV18	NV18	NV18	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU1	NV18	 X 	NV18	NV18	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU2	NV18	NV18	 X 	NV18	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU3	NV18	NV18	NV18	 X 	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU4	NV18	NV18	NV18	NV18	 X 	NV18	NV18	NV18	48-95	1		N/A
+GPU5	NV18	NV18	NV18	NV18	NV18	 X 	NV18	NV18	48-95	1		N/A
+GPU6	NV18	NV18	NV18	NV18	NV18	NV18	 X 	NV18	48-95	1		N/A
+GPU7	NV18	NV18	NV18	NV18	NV18	NV18	NV18	 X 	48-95	1		N/A
+
+Legend:
+
+  X    = Self
+  SYS  = Connection traversing PCIe as well as the SMP interconnect between NUMA nodes (e.g., QPI/UPI)
+  NODE = Connection traversing PCIe as well as the interconnect between PCIe Host Bridges within a NUMA node
+  PHB  = Connection traversing PCIe as well as a PCIe Host Bridge (typically the CPU)
+  PXB  = Connection traversing multiple PCIe bridges (without traversing the PCIe Host Bridge)
+  PIX  = Connection traversing at most a single PCIe bridge
+  NV#  = Connection traversing a bonded set of # NVLinks
++ echo ==================
+==================
++ echo 'Master node: ip-26-0-172-116'
+Master node: ip-26-0-172-116
++ echo 'All nodes: ip-26-0-172-116
+ip-26-0-172-142'
+All nodes: ip-26-0-172-116
+ip-26-0-172-142
++ echo 'World size: 16'
+World size: 16
++ srun --wait=0 --kill-on-bad-exit=1 torchrun --nnodes=2 --nproc_per_node=8 --rdzv_id=13459892 --rdzv_backend=c10d --rdzv_endpoint=ip-26-0-172-116:12356 --max_restarts 0 --rdzv_conf timeout=60 /fsx/nouamane/projects/nanotron/run_train.py --config-file benchmark/configs/config_80G_dp2_tp2_pp4_acc128_mbs1_seq4096_zero1_tpmodeRED_vocab131k.yaml
+[2024-12-15 12:35:32,622] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-15 12:35:32,622] torch.distributed.run: [WARNING] 
+[2024-12-15 12:35:32,622] torch.distributed.run: [WARNING] *****************************************
+[2024-12-15 12:35:32,622] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-15 12:35:32,622] torch.distributed.run: [WARNING] *****************************************
+[2024-12-15 12:35:32,746] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-15 12:35:32,746] torch.distributed.run: [WARNING] 
+[2024-12-15 12:35:32,746] torch.distributed.run: [WARNING] *****************************************
+[2024-12-15 12:35:32,746] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-15 12:35:32,746] torch.distributed.run: [WARNING] *****************************************
+NCCL version 2.18.5+cuda12.2
+12/15/2024 12:35:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-116]: Measuring inter-GPU and intra-node bandwidth...
+NCCL version 2.18.5+cuda12.2
+12/15/2024 12:36:11 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-116]: Bandwidth measurement complete. Time taken: 16.30 seconds
+12/15/2024 12:36:11 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-116]: Config:
+12/15/2024 12:36:11 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-116]: Config(general=GeneralArgs(project='debug',
+12/15/2024 12:36:11 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-116]:                            run='80G_dp2_tp2_pp4_acc128_mbs1_seq4096_zero1_tpmodeRED_vocab131k',
+12/15/2024 12:36:11 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-116]:                            seed=42,
+12/15/2024 12:36:11 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-116]:                            step=None,
+12/15/2024 12:36:11 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-116]:                            consumed_train_samples=None,
+12/15/2024 12:36:11 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-116]:                            benchmark_csv_path=PosixPath('benchmark/results/bench_final2.csv'),
+12/15/2024 12:36:11 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-116]:                            ignore_sanity_checks=True),
+12/15/2024 12:36:11 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-116]:        parallelism=ParallelismArgs(dp=2,
+12/15/2024 12:36:11 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-116]:                                    pp=4,
+12/15/2024 12:36:11 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-116]:                                    tp=2,
+12/15/2024 12:36:11 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-116]:                                    pp_engine=<nanotron.parallel.pipeline_parallel.engine.OneForwardOneBackwardPipelineEngine object at 0x7fcc3ef8ce20>,
+12/15/2024 12:36:11 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-116]:                                    tp_mode=<TensorParallelLinearMode.REDUCE_SCATTER: 2>,
+12/15/2024 12:36:11 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-116]:                                    tp_linear_async_communication=True,
+12/15/2024 12:36:11 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-116]:                                    recompute_layer=False,
+12/15/2024 12:36:11 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-116]:                                    tp_recompute_allgather=True,
+12/15/2024 12:36:11 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-116]:                                    expert_parallel_size=1),
+12/15/2024 12:36:11 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-116]:        model=ModelArgs(model_config=LlamaConfig(bos_token_id=0,
+12/15/2024 12:36:11 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-116]:                                                 eos_token_id=0,
+12/15/2024 12:36:11 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-116]:                                                 hidden_act='silu',
+12/15/2024 12:36:11 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-116]:                                                 hidden_size=8192,
+12/15/2024 12:36:11 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-116]:                                                 initializer_range=0.02,
+12/15/2024 12:36:11 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-116]:                                                 intermediate_size=28672,
+12/15/2024 12:36:11 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-116]:                                                 is_llama_config=True,
+12/15/2024 12:36:11 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-116]:                                                 max_position_embeddings=4096,
+12/15/2024 12:36:11 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-116]:                                                 num_attention_heads=64,
+12/15/2024 12:36:11 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-116]:                                                 num_hidden_layers=80,
+12/15/2024 12:36:11 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-116]:                                                 num_key_value_heads=64,
+12/15/2024 12:36:11 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-116]:                                                 pad_token_id=None,
+12/15/2024 12:36:11 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-116]:                                                 pretraining_tp=1,
+12/15/2024 12:36:11 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-116]:                                                 rms_norm_eps=1e-05,
+12/15/2024 12:36:11 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-116]:                                                 rope_scaling=None,
+12/15/2024 12:36:11 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-116]:                                                 rope_theta=10000.0,
+12/15/2024 12:36:11 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-116]:                                                 rope_interleaved=False,
+12/15/2024 12:36:11 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-116]:                                                 tie_word_embeddings=False,
+12/15/2024 12:36:11 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-116]:                                                 use_cache=True,
+12/15/2024 12:36:11 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-116]:                                                 vocab_size=131072),
+12/15/2024 12:36:11 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-116]:                        init_method=RandomInit(std=0.02),
+12/15/2024 12:36:11 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-116]:                        dtype=torch.bfloat16,
+12/15/2024 12:36:11 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-116]:                        make_vocab_size_divisible_by=1,
+12/15/2024 12:36:11 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-116]:                        ddp_bucket_cap_mb=25),
+12/15/2024 12:36:11 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-116]:        tokenizer=TokenizerArgs(tokenizer_name_or_path='robot-test/dummy-tokenizer-wordlevel',
+12/15/2024 12:36:11 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-116]:                                tokenizer_revision=None,
+12/15/2024 12:36:11 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-116]:                                tokenizer_max_length=None),
+12/15/2024 12:36:11 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-116]:        checkpoints=CheckpointsArgs(checkpoints_path=PosixPath('checkpoints'),
+12/15/2024 12:36:11 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-116]:                                    checkpoint_interval=10000,
+12/15/2024 12:36:11 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-116]:                                    save_initial_state=False,
+12/15/2024 12:36:11 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-116]:                                    save_final_state=False,
+12/15/2024 12:36:11 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-116]:                                    resume_checkpoint_path=None,
+12/15/2024 12:36:11 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-116]:                                    checkpoints_path_is_shared_file_system=False),
+12/15/2024 12:36:11 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-116]:        logging=LoggingArgs(log_level='info',
+12/15/2024 12:36:11 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-116]:                            log_level_replica='info',
+12/15/2024 12:36:11 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-116]:                            iteration_step_info_interval=1),
+12/15/2024 12:36:11 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-116]:        tokens=TokensArgs(sequence_length=4096,
+12/15/2024 12:36:11 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-116]:                          train_steps=100,
+12/15/2024 12:36:11 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-116]:                          micro_batch_size=1,
+12/15/2024 12:36:11 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-116]:                          batch_accumulation_per_replica=128,
+12/15/2024 12:36:11 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-116]:                          val_check_interval=100,
+12/15/2024 12:36:11 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-116]:                          limit_val_batches=0,
+12/15/2024 12:36:11 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-116]:                          limit_test_batches=0),
+12/15/2024 12:36:11 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-116]:        optimizer=OptimizerArgs(optimizer_factory=AdamWOptimizerArgs(adam_eps=1e-08,
+12/15/2024 12:36:11 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-116]:                                                                     adam_beta1=0.9,
+12/15/2024 12:36:11 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-116]:                                                                     adam_beta2=0.95,
+12/15/2024 12:36:11 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-116]:                                                                     torch_adam_is_fused=True,
+12/15/2024 12:36:11 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-116]:                                                                     name='adamW'),
+12/15/2024 12:36:11 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-116]:                                zero_stage=1,
+12/15/2024 12:36:11 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-116]:                                weight_decay=0.01,
+12/15/2024 12:36:11 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-116]:                                clip_grad=1.0,
+12/15/2024 12:36:11 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-116]:                                accumulate_grad_in_fp32=True,
+12/15/2024 12:36:11 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-116]:                                learning_rate_scheduler=LRSchedulerArgs(learning_rate=0.0003,
+12/15/2024 12:36:11 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-116]:                                                                        lr_warmup_steps=2,
+12/15/2024 12:36:11 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-116]:                                                                        lr_warmup_style='linear',
+12/15/2024 12:36:11 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-116]:                                                                        lr_decay_style='cosine',
+12/15/2024 12:36:11 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-116]:                                                                        lr_decay_steps=13,
+12/15/2024 12:36:11 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-116]:                                                                        lr_decay_starting_step=None,
+12/15/2024 12:36:11 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-116]:                                                                        min_decay_lr=1e-05)),
+12/15/2024 12:36:11 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-116]:        data_stages=[DatasetStageArgs(name='Stable Training Stage',
+12/15/2024 12:36:11 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-116]:                                      start_training_step=1,
+12/15/2024 12:36:11 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-116]:                                      data=DataArgs(dataset=None,
+12/15/2024 12:36:11 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-116]:                                                    seed=42,
+12/15/2024 12:36:11 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-116]:                                                    num_loading_workers=1))],
+12/15/2024 12:36:11 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-116]:        profiler=None,
+12/15/2024 12:36:11 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-116]:        lighteval=None,
+12/15/2024 12:36:11 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-116]:        s3_upload=None)
+12/15/2024 12:36:11 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-116]: Model Config:
+12/15/2024 12:36:11 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-116]: LlamaConfig(bos_token_id=0,
+12/15/2024 12:36:11 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-116]:             eos_token_id=0,
+12/15/2024 12:36:11 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-116]:             hidden_act='silu',
+12/15/2024 12:36:11 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-116]:             hidden_size=8192,
+12/15/2024 12:36:11 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-116]:             initializer_range=0.02,
+12/15/2024 12:36:11 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-116]:             intermediate_size=28672,
+12/15/2024 12:36:11 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-116]:             is_llama_config=True,
+12/15/2024 12:36:11 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-116]:             max_position_embeddings=4096,
+12/15/2024 12:36:11 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-116]:             num_attention_heads=64,
+12/15/2024 12:36:11 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-116]:             num_hidden_layers=80,
+12/15/2024 12:36:11 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-116]:             num_key_value_heads=64,
+12/15/2024 12:36:11 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-116]:             pad_token_id=None,
+12/15/2024 12:36:11 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-116]:             pretraining_tp=1,
+12/15/2024 12:36:11 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-116]:             rms_norm_eps=1e-05,
+12/15/2024 12:36:11 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-116]:             rope_scaling=None,
+12/15/2024 12:36:11 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-116]:             rope_theta=10000.0,
+12/15/2024 12:36:11 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-116]:             rope_interleaved=False,
+12/15/2024 12:36:11 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-116]:             tie_word_embeddings=False,
+12/15/2024 12:36:11 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-116]:             use_cache=True,
+12/15/2024 12:36:11 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-116]:             vocab_size=131072)
+12/15/2024 12:36:11 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-116]: Building model..
+12/15/2024 12:36:11 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-116]: Initialize RoPE Theta = 10000.0
+12/15/2024 12:36:11 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-116]: Setting PP block ranks...
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+12/15/2024 12:36:16 [INFO|DP=0|PP=3|TP=0|ip-26-0-172-142]: Local number of parameters: 9.78G (18656.61MiB)
+12/15/2024 12:36:16 [INFO|DP=0|PP=1|TP=0|ip-26-0-172-116]: Local number of parameters: 9.73G (18560.62MiB)
+12/15/2024 12:36:16 [INFO|DP=0|PP=2|TP=0|ip-26-0-172-142]: Local number of parameters: 9.73G (18560.62MiB)
+12/15/2024 12:36:16 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-116]: Total number of parameters: 80G (152581.03MiB)
+12/15/2024 12:36:16 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-116]: Local number of parameters: 10.8G (20512.66MiB)
+12/15/2024 12:36:16 [INFO|DP=0|PP=3|TP=1|ip-26-0-172-142]: Local number of parameters: 9.78G (18656.61MiB)
+12/15/2024 12:36:16 [INFO|DP=0|PP=1|TP=1|ip-26-0-172-116]: Local number of parameters: 9.73G (18560.62MiB)
+12/15/2024 12:36:16 [INFO|DP=0|PP=2|TP=1|ip-26-0-172-142]: Local number of parameters: 9.73G (18560.62MiB)
+12/15/2024 12:36:16 [INFO|DP=0|PP=3|TP=0|ip-26-0-172-142]: [After model building] Memory usage: 18656.64MiB. Peak allocated: 18656.64MiB Peak reserved: 22914.00MiB
+12/15/2024 12:36:16 [INFO|DP=0|PP=0|TP=1|ip-26-0-172-116]: Local number of parameters: 10.8G (20512.66MiB)
+12/15/2024 12:36:16 [INFO|DP=0|PP=1|TP=0|ip-26-0-172-116]: [After model building] Memory usage: 18560.65MiB. Peak allocated: 18560.66MiB Peak reserved: 20386.00MiB
+12/15/2024 12:36:16 [INFO|DP=0|PP=2|TP=0|ip-26-0-172-142]: [After model building] Memory usage: 18560.65MiB. Peak allocated: 18560.66MiB Peak reserved: 22914.00MiB
+12/15/2024 12:36:16 [INFO|DP=0|PP=3|TP=1|ip-26-0-172-142]: [After model building] Memory usage: 18656.64MiB. Peak allocated: 18656.64MiB Peak reserved: 22914.00MiB
+12/15/2024 12:36:16 [INFO|DP=0|PP=2|TP=1|ip-26-0-172-142]: [After model building] Memory usage: 18560.65MiB. Peak allocated: 18560.66MiB Peak reserved: 22914.00MiB
+12/15/2024 12:36:16 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-116]: [After model building] Memory usage: 20512.69MiB. Peak allocated: 20512.69MiB Peak reserved: 21954.00MiB
+12/15/2024 12:36:16 [INFO|DP=0|PP=1|TP=1|ip-26-0-172-116]: [After model building] Memory usage: 18560.65MiB. Peak allocated: 18560.66MiB Peak reserved: 20386.00MiB
+12/15/2024 12:36:16 [INFO|DP=0|PP=0|TP=1|ip-26-0-172-116]: [After model building] Memory usage: 20512.69MiB. Peak allocated: 20512.69MiB Peak reserved: 22274.00MiB
+12/15/2024 12:36:16 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-116]: No checkpoint path provided.
+12/15/2024 12:36:16 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-116]: Parametrizing model parameters using StandardParametrizator
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+12/15/2024 12:36:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-116]: [Optimizer Building] Using LearningRateForSP as learning rate
+12/15/2024 12:36:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-116]: [ZeRO sharding] Size of optimizer params per rank:
+12/15/2024 12:36:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-116]: [ZeRO sharding] DP Rank 0 has 5.38G out of 10.8G (50.00%) params' optimizer states
+12/15/2024 12:36:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-116]: [ZeRO sharding] DP Rank 1 has 5.38G out of 10.8G (50.00%) params' optimizer states
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 248, in <module>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 183, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 383, in init_optimizer_and_grad_accumulator
+    optimizer = ZeroDistributedOptimizer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/zero.py", line 91, in __init__
+    optimizer = optimizer_builder(param_groups_in_rank)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 248, in <module>
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 183, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 95, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 383, in init_optimizer_and_grad_accumulator
+    big_flat_buffer = torch.empty(length, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 18.13 GiB. GPU 5 has a total capacty of 79.33 GiB of which 13.77 GiB is free. Including non-PyTorch memory, this process has 65.54 GiB memory in use. Of the allocated memory 54.38 GiB is allocated by PyTorch, and 1.78 GiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    optimizer = ZeroDistributedOptimizer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/zero.py", line 91, in __init__
+    optimizer = optimizer_builder(param_groups_in_rank)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 95, in __init__
+    big_flat_buffer = torch.empty(length, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 20.03 GiB. GPU 0 has a total capacty of 79.33 GiB of which 8.46 GiB is free. Including non-PyTorch memory, this process has 70.86 GiB memory in use. Of the allocated memory 60.10 GiB is allocated by PyTorch, and 1.41 GiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 248, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 248, in <module>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 183, in __init__
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 183, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 383, in init_optimizer_and_grad_accumulator
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 383, in init_optimizer_and_grad_accumulator
+    optimizer = ZeroDistributedOptimizer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/zero.py", line 91, in __init__
+    optimizer = ZeroDistributedOptimizer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/zero.py", line 91, in __init__
+    optimizer = optimizer_builder(param_groups_in_rank)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    optimizer = optimizer_builder(param_groups_in_rank)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 95, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 95, in __init__
+    big_flat_buffer = torch.empty(length, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 20.03 GiB. GPU 2 has a total capacty of 79.33 GiB of which 8.12 GiB is free. Including non-PyTorch memory, this process has 71.20 GiB memory in use. Of the allocated memory 60.10 GiB is allocated by PyTorch, and 1.72 GiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    big_flat_buffer = torch.empty(length, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 18.13 GiB. GPU 7 has a total capacty of 79.33 GiB of which 13.97 GiB is free. Including non-PyTorch memory, this process has 65.35 GiB memory in use. Of the allocated memory 54.38 GiB is allocated by PyTorch, and 1.78 GiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 248, in <module>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 183, in __init__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 248, in <module>
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 383, in init_optimizer_and_grad_accumulator
+    optimizer = ZeroDistributedOptimizer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/zero.py", line 91, in __init__
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 183, in __init__
+    optimizer = optimizer_builder(param_groups_in_rank)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 383, in init_optimizer_and_grad_accumulator
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    optimizer = ZeroDistributedOptimizer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/zero.py", line 91, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    optimizer = optimizer_builder(param_groups_in_rank)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 95, in __init__
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    big_flat_buffer = torch.empty(length, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 18.13 GiB. GPU 6 has a total capacty of 79.33 GiB of which 13.46 GiB is free. Including non-PyTorch memory, this process has 65.86 GiB memory in use. Of the allocated memory 54.38 GiB is allocated by PyTorch, and 2.10 GiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 95, in __init__
+    big_flat_buffer = torch.empty(length, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 18.13 GiB. GPU 4 has a total capacty of 79.33 GiB of which 13.76 GiB is free. Including non-PyTorch memory, this process has 65.56 GiB memory in use. Of the allocated memory 54.38 GiB is allocated by PyTorch, and 1.78 GiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 248, in <module>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 183, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 383, in init_optimizer_and_grad_accumulator
+    optimizer = ZeroDistributedOptimizer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/zero.py", line 91, in __init__
+    optimizer = optimizer_builder(param_groups_in_rank)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 95, in __init__
+    big_flat_buffer = torch.empty(length, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 18.22 GiB. GPU 5 has a total capacty of 79.33 GiB of which 13.12 GiB is free. Including non-PyTorch memory, this process has 66.20 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 2.16 GiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 248, in <module>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 183, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 383, in init_optimizer_and_grad_accumulator
+    optimizer = ZeroDistributedOptimizer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/zero.py", line 91, in __init__
+    optimizer = optimizer_builder(param_groups_in_rank)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 95, in __init__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 248, in <module>
+    big_flat_buffer = torch.empty(length, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 18.13 GiB. GPU 1 has a total capacty of 79.33 GiB of which 13.30 GiB is free. Including non-PyTorch memory, this process has 66.01 GiB memory in use. Of the allocated memory 54.38 GiB is allocated by PyTorch, and 2.25 GiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 183, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 383, in init_optimizer_and_grad_accumulator
+    optimizer = ZeroDistributedOptimizer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/zero.py", line 91, in __init__
+    optimizer = optimizer_builder(param_groups_in_rank)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 95, in __init__
+    big_flat_buffer = torch.empty(length, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 18.13 GiB. GPU 3 has a total capacty of 79.33 GiB of which 13.19 GiB is free. Including non-PyTorch memory, this process has 66.13 GiB memory in use. Of the allocated memory 54.38 GiB is allocated by PyTorch, and 2.25 GiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 248, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 248, in <module>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 183, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 383, in init_optimizer_and_grad_accumulator
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 183, in __init__
+    optimizer = ZeroDistributedOptimizer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/zero.py", line 91, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 383, in init_optimizer_and_grad_accumulator
+    optimizer = optimizer_builder(param_groups_in_rank)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    optimizer = ZeroDistributedOptimizer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/zero.py", line 91, in __init__
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    optimizer = optimizer_builder(param_groups_in_rank)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 95, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 248, in <module>
+    big_flat_buffer = torch.empty(length, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 18.22 GiB. GPU 7 has a total capacty of 79.33 GiB of which 13.31 GiB is free. Including non-PyTorch memory, this process has 66.01 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 2.16 GiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 95, in __init__
+    big_flat_buffer = torch.empty(length, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 18.13 GiB. GPU 0 has a total capacty of 79.33 GiB of which 13.34 GiB is free. Including non-PyTorch memory, this process has 65.98 GiB memory in use. Of the allocated memory 54.38 GiB is allocated by PyTorch, and 2.25 GiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 183, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 383, in init_optimizer_and_grad_accumulator
+    optimizer = ZeroDistributedOptimizer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/zero.py", line 91, in __init__
+    optimizer = optimizer_builder(param_groups_in_rank)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 95, in __init__
+    big_flat_buffer = torch.empty(length, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 18.22 GiB. GPU 4 has a total capacty of 79.33 GiB of which 13.10 GiB is free. Including non-PyTorch memory, this process has 66.22 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 2.16 GiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 248, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 248, in <module>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 183, in __init__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 248, in <module>
+    trainer = DistributedTrainer(config_file)
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 183, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 383, in init_optimizer_and_grad_accumulator
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 383, in init_optimizer_and_grad_accumulator
+    optimizer = ZeroDistributedOptimizer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/zero.py", line 91, in __init__
+    optimizer = optimizer_builder(param_groups_in_rank)
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 183, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    optimizer = ZeroDistributedOptimizer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/zero.py", line 91, in __init__
+        result = OptimizerFromGradientAccumulator(optimizer = optimizer_builder(param_groups_in_rank)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 383, in init_optimizer_and_grad_accumulator
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    optimizer = ZeroDistributedOptimizer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/zero.py", line 91, in __init__
+        gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 95, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    optimizer = optimizer_builder(param_groups_in_rank)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    big_flat_buffer = torch.empty(length, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 18.22 GiB. GPU 6 has a total capacty of 79.33 GiB of which 13.12 GiB is free. Including non-PyTorch memory, this process has 66.20 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 2.16 GiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 95, in __init__
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    big_flat_buffer = torch.empty(length, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 18.13 GiB. GPU 2 has a total capacty of 79.33 GiB of which 13.30 GiB is free. Including non-PyTorch memory, this process has 66.01 GiB memory in use. Of the allocated memory 54.38 GiB is allocated by PyTorch, and 2.25 GiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 95, in __init__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 248, in <module>
+    big_flat_buffer = torch.empty(length, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 20.03 GiB. GPU 3 has a total capacty of 79.33 GiB of which 8.00 GiB is free. Including non-PyTorch memory, this process has 71.32 GiB memory in use. Of the allocated memory 60.10 GiB is allocated by PyTorch, and 1.72 GiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 183, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 383, in init_optimizer_and_grad_accumulator
+    optimizer = ZeroDistributedOptimizer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/zero.py", line 91, in __init__
+    optimizer = optimizer_builder(param_groups_in_rank)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 95, in __init__
+    big_flat_buffer = torch.empty(length, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 20.03 GiB. GPU 1 has a total capacty of 79.33 GiB of which 8.12 GiB is free. Including non-PyTorch memory, this process has 71.20 GiB memory in use. Of the allocated memory 60.10 GiB is allocated by PyTorch, and 1.72 GiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+[2024-12-15 12:36:34,880] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 143060 closing signal SIGTERM
+[2024-12-15 12:36:34,880] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 143061 closing signal SIGTERM
+[2024-12-15 12:36:34,880] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 143063 closing signal SIGTERM
+[2024-12-15 12:36:34,880] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 143064 closing signal SIGTERM
+[2024-12-15 12:36:34,880] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 143065 closing signal SIGTERM
+[2024-12-15 12:36:34,880] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 143066 closing signal SIGTERM
+[2024-12-15 12:36:34,880] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 143067 closing signal SIGTERM
+[2024-12-15 12:36:34,886] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 69624 closing signal SIGTERM
+[2024-12-15 12:36:34,886] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 69625 closing signal SIGTERM
+[2024-12-15 12:36:34,886] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 69626 closing signal SIGTERM
+[2024-12-15 12:36:34,886] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 69627 closing signal SIGTERM
+[2024-12-15 12:36:34,886] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 69628 closing signal SIGTERM
+[2024-12-15 12:36:34,886] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 69630 closing signal SIGTERM
+[2024-12-15 12:36:34,886] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 69631 closing signal SIGTERM
+[2024-12-15 12:36:38,026] torch.distributed.elastic.multiprocessing.api: [ERROR] failed (exitcode: 1) local_rank: 2 (pid: 143062) of binary: /fsx/nouamane/miniconda/envs/2-1-cu121/bin/python
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 33, in <module>
+    sys.exit(load_entry_point('torch==2.1.1', 'console_scripts', 'torchrun')())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 264, in launch_agent
+    raise ChildFailedError(
+torch.distributed.elastic.multiprocessing.errors.ChildFailedError: 
+============================================================
+/fsx/nouamane/projects/nanotron/run_train.py FAILED
+------------------------------------------------------------
+Failures:
+  <NO_OTHER_FAILURES>
+------------------------------------------------------------
+Root Cause (first observed failure):
+[0]:
+  time      : 2024-12-15_12:36:34
+  host      : ip-26-0-172-116.ec2.internal
+  rank      : 2 (local_rank: 2)
+  exitcode  : 1 (pid: 143062)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+============================================================
+[2024-12-15 12:36:38,433] torch.distributed.elastic.multiprocessing.api: [ERROR] failed (exitcode: 1) local_rank: 5 (pid: 69629) of binary: /fsx/nouamane/miniconda/envs/2-1-cu121/bin/python
+[2024-12-15 12:36:38,477] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-172-142.ec2.internal_69552_0' has failed to shutdown the rendezvous '13459892' due to an error of type RendezvousConnectionError.
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 33, in <module>
+    sys.exit(load_entry_point('torch==2.1.1', 'console_scripts', 'torchrun')())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 264, in launch_agent
+    raise ChildFailedError(
+torch.distributed.elastic.multiprocessing.errors.ChildFailedError: 
+============================================================
+/fsx/nouamane/projects/nanotron/run_train.py FAILED
+------------------------------------------------------------
+Failures:
+  <NO_OTHER_FAILURES>
+------------------------------------------------------------
+Root Cause (first observed failure):
+[0]:
+  time      : 2024-12-15_12:36:34
+  host      : ip-26-0-172-142.ec2.internal
+  rank      : 13 (local_rank: 5)
+  exitcode  : 1 (pid: 69629)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+============================================================
+srun: error: ip-26-0-172-116: task 0: Exited with exit code 1
+srun: launch/slurm: _step_signal: Terminating StepId=13459892.0
+srun: error: ip-26-0-172-142: task 1: Terminated
+srun: Force Terminated StepId=13459892.0
diff --git a/logs/13460026-bench_80G_dp4_tp4_pp2_acc64_mbs1_seq4096_zero0_tpmodeRED_vocab131k.out b/logs/13460026-bench_80G_dp4_tp4_pp2_acc64_mbs1_seq4096_zero0_tpmodeRED_vocab131k.out
new file mode 100644
index 0000000000000000000000000000000000000000..412f2479de951ad5eabaef7f51cec0f84b6439f1
--- /dev/null
+++ b/logs/13460026-bench_80G_dp4_tp4_pp2_acc64_mbs1_seq4096_zero0_tpmodeRED_vocab131k.out
@@ -0,0 +1,1324 @@
++ source /etc/profile.d/modules.sh
+++ . /usr/share/modules/init/bash
++++ unset _mlshdbg
++++ '[' 0 = 1 ']'
++++ unset _mlre _mlIFS
++++ '[' -n x ']'
++++ _mlIFS=' 	
+'
++++ IFS=' '
++++ '[' -n '' ']'
+++++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash autoinit
++++ _mlcode='module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+test 0;'
++++ _mlret=0
++++ '[' -n x ']'
++++ IFS=' 	
+'
++++ unset _mlIFS
++++ unset _mlre _mlv _mlrv
++++ '[' 0 -eq 0 ']'
++++ eval 'module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+test 0;'
+++++ MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl
+++++ export MODULES_CMD
+++++ ENV=/usr/share/modules/init/profile.sh
+++++ export ENV
+++++ BASH_ENV=/usr/share/modules/init/bash
+++++ export BASH_ENV
+++++ MODULESHOME=/usr/share/modules
+++++ export MODULESHOME
+++++ test 0
++++ '[' 0 = 1 ']'
++++ '[' -t 2 ']'
++++ export -f module
++++ export -f switchml
++++ '[' 5 -ge 3 ']'
++++ [[ ehxB =~ i ]]
++++ [[ ! :/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin: =~ :/usr/bin: ]]
++++ '[' '!' -n x ']'
+++++ manpath
++++ [[ ! :/admin/home/nouamane/.local/share/man:/fsx/nouamane/miniconda/envs/2-1-cu121/man:/fsx/nouamane/miniconda/envs/2-1-cu121/share/man:/fsx/nouamane/miniconda/man:/fsx/nouamane/miniconda/share/man:/opt/slurm/share/man:/opt/amazon/openmpi/share/man:/opt/amazon/efa/share/man:/usr/local/man:/usr/local/share/man:/usr/share/man:/admin/home/nouamane/.fzf/man:: =~ :/usr/share/man: ]]
++++ unset _mlcode _mlret
++++ '[' -n '' ']'
++ module load cuda/12.1
++ unset _mlshdbg
++ '[' 0 = 1 ']'
++ unset _mlre _mlIFS
++ '[' -n x ']'
++ _mlIFS=' 	
+'
++ IFS=' '
++ '[' -n '' ']'
+++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash load cuda/12.1
++ eval 'CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include;' export 'CPATH;
+LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/opt/aws-ofi-nccl/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/targets/x86_64-linux/lib/:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/lib:/usr/lib;' export 'LD_LIBRARY_PATH;
+FI_EFA_FORK_SAFE=1;' export 'FI_EFA_FORK_SAFE;
+MANPATH=/usr/local/cuda-12.1/share/man::;' export 'MANPATH;
+LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64;' export 'LIBRARY_PATH;
+_LMFILES_=/usr/share/modules/modulefiles/cuda/12.1;' export '_LMFILES_;
+LOADEDMODULES=cuda/12.1;' export 'LOADEDMODULES;
+MPI_PATH=/opt/amazon/openmpi;' export 'MPI_PATH;
+NCCL_HOME_modshare=/opt/nccl/build:1;' export 'NCCL_HOME_modshare;
+NCCL_PROTO=simple;' export 'NCCL_PROTO;
+MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1;' export 'MANPATH_modshare;
+LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1;' export 'LIBRARY_PATH_modshare;
+NCCL_SOCKET_IFNAME=enp;' export 'NCCL_SOCKET_IFNAME;
+AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl;' export 'AWS_OFI_NCCL_HOME;
+NCCL_HOME=/opt/nccl/build;' export 'NCCL_HOME;
+FI_PROVIDER=efa;' export 'FI_PROVIDER;
+AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1;' export 'AWS_OFI_NCCL_HOME_modshare;
+CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1;' export 'CPATH_modshare;
+LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib/:1:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2:/opt/aws-ofi-nccl/lib:2:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/efa/lib:1:/usr/lib:1;' export 'LD_LIBRARY_PATH_modshare;
+FI_EFA_ENABLE_SHM_TRANSFER=1;' export 'FI_EFA_ENABLE_SHM_TRANSFER;
+_LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1;' export '_LMFILES__modshare;
+LOADEDMODULES_modshare=cuda/12.1:1;' export 'LOADEDMODULES_modshare;
+MPI_PATH_modshare=/opt/amazon/openmpi:1;' export 'MPI_PATH_modshare;
+PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin;' export 'PATH;
+CUDA_HOME=/usr/local/cuda-12.1;' export 'CUDA_HOME;
+PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/usr/local/cuda-12.1/include:1:/opt/amazon/efa/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/usr/sbin:1:/fsx/nouamane/miniconda/bin:1:/opt/slurm/bin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:2:/opt/amazon/openmpi/bin:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1;' export 'PATH_modshare;
+test' '0;'
+++ CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include
+++ export CPATH
+++ LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/opt/aws-ofi-nccl/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/targets/x86_64-linux/lib/:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/lib:/usr/lib
+++ export LD_LIBRARY_PATH
+++ FI_EFA_FORK_SAFE=1
+++ export FI_EFA_FORK_SAFE
+++ MANPATH=/usr/local/cuda-12.1/share/man::
+++ export MANPATH
+++ LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64
+++ export LIBRARY_PATH
+++ _LMFILES_=/usr/share/modules/modulefiles/cuda/12.1
+++ export _LMFILES_
+++ LOADEDMODULES=cuda/12.1
+++ export LOADEDMODULES
+++ MPI_PATH=/opt/amazon/openmpi
+++ export MPI_PATH
+++ NCCL_HOME_modshare=/opt/nccl/build:1
+++ export NCCL_HOME_modshare
+++ NCCL_PROTO=simple
+++ export NCCL_PROTO
+++ MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1
+++ export MANPATH_modshare
+++ LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1
+++ export LIBRARY_PATH_modshare
+++ NCCL_SOCKET_IFNAME=enp
+++ export NCCL_SOCKET_IFNAME
+++ AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl
+++ export AWS_OFI_NCCL_HOME
+++ NCCL_HOME=/opt/nccl/build
+++ export NCCL_HOME
+++ FI_PROVIDER=efa
+++ export FI_PROVIDER
+++ AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1
+++ export AWS_OFI_NCCL_HOME_modshare
+++ CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1
+++ export CPATH_modshare
+++ LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib/:1:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2:/opt/aws-ofi-nccl/lib:2:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/efa/lib:1:/usr/lib:1
+++ export LD_LIBRARY_PATH_modshare
+++ FI_EFA_ENABLE_SHM_TRANSFER=1
+++ export FI_EFA_ENABLE_SHM_TRANSFER
+++ _LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1
+++ export _LMFILES__modshare
+++ LOADEDMODULES_modshare=cuda/12.1:1
+++ export LOADEDMODULES_modshare
+++ MPI_PATH_modshare=/opt/amazon/openmpi:1
+++ export MPI_PATH_modshare
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin
+++ export PATH
+++ CUDA_HOME=/usr/local/cuda-12.1
+++ export CUDA_HOME
+++ PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/usr/local/cuda-12.1/include:1:/opt/amazon/efa/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/usr/sbin:1:/fsx/nouamane/miniconda/bin:1:/opt/slurm/bin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:2:/opt/amazon/openmpi/bin:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1
+++ export PATH_modshare
+++ test 0
++ _mlstatus=0
++ '[' -n x ']'
++ IFS=' 	
+'
++ unset _mlre _mlv _mlrv _mlIFS
++ '[' -n '' ']'
++ unset _mlshdbg
++ return 0
++ source /fsx/nouamane/miniconda/bin/activate
+++ _CONDA_ROOT=/fsx/nouamane/miniconda
+++ . /fsx/nouamane/miniconda/etc/profile.d/conda.sh
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ '[' -z x ']'
+++ conda activate
+++ local cmd=activate
+++ case "$cmd" in
+++ __conda_activate activate
+++ '[' -n '' ']'
+++ local ask_conda
++++ PS1=
++++ __conda_exe shell.posix activate
++++ /fsx/nouamane/miniconda/bin/conda shell.posix activate
+++ ask_conda='PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin'\''
+export CONDA_SHLVL='\''1'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\'''
+++ eval 'PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin'\''
+export CONDA_SHLVL='\''1'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\'''
++++ PS1='(base) '
++++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin
++++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin
++++ export CONDA_SHLVL=1
++++ CONDA_SHLVL=1
++++ export 'CONDA_PROMPT_MODIFIER=(base) '
++++ CONDA_PROMPT_MODIFIER='(base) '
+++ __conda_hashr
+++ '[' -n '' ']'
+++ '[' -n '' ']'
+++ hash -r
++ conda activate 2-1-cu121
++ local cmd=activate
++ case "$cmd" in
++ __conda_activate activate 2-1-cu121
++ '[' -n '' ']'
++ local ask_conda
+++ PS1='(base) '
+++ __conda_exe shell.posix activate 2-1-cu121
+++ /fsx/nouamane/miniconda/bin/conda shell.posix activate 2-1-cu121
++ ask_conda='PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''2'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_1='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
++ eval 'PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''2'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_1='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
+++ PS1='(2-1-cu121) '
+++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin
+++ export CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ export CONDA_SHLVL=2
+++ CONDA_SHLVL=2
+++ export CONDA_DEFAULT_ENV=2-1-cu121
+++ CONDA_DEFAULT_ENV=2-1-cu121
+++ export 'CONDA_PROMPT_MODIFIER=(2-1-cu121) '
+++ CONDA_PROMPT_MODIFIER='(2-1-cu121) '
+++ export CONDA_PREFIX_1=/fsx/nouamane/miniconda
+++ CONDA_PREFIX_1=/fsx/nouamane/miniconda
+++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ export _CE_M=
+++ _CE_M=
+++ export _CE_CONDA=
+++ _CE_CONDA=
+++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh
++++ test -n ''
++++ xml_catalog_files_libxml2=
++++ XML_CATALOG_FILES=
++++ conda_catalog_files=
++++ ifs_libxml2=' 	
+'
++++ IFS=' '
++++ rem=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ for pre in ${rem}
++++ test '' = /fsx/nouamane/miniconda/envs/2-1-cu121
++++ conda_catalog_files=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ rem=
++++ IFS=' 	
+'
++++ conda_catalog_files='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ export 'XML_CATALOG_FILES=file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ XML_CATALOG_FILES='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ unset conda_catalog_files ifs_libxml2 rem
++ __conda_hashr
++ '[' -n '' ']'
++ '[' -n '' ']'
++ hash -r
++ export PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin
++ PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin
+++ scontrol show hostnames 'ip-26-0-165-[59,131,164,202]'
++ export 'NODELIST=ip-26-0-165-59
+ip-26-0-165-131
+ip-26-0-165-164
+ip-26-0-165-202'
++ NODELIST='ip-26-0-165-59
+ip-26-0-165-131
+ip-26-0-165-164
+ip-26-0-165-202'
+++ head -n1
+++ scontrol show hostnames 'ip-26-0-165-[59,131,164,202]'
++ export MASTER_NODE=ip-26-0-165-59
++ MASTER_NODE=ip-26-0-165-59
++ export MASTER_PORT=12356
++ MASTER_PORT=12356
++ export NNODES=4
++ NNODES=4
++ export GPUS_PER_NODE=8
++ GPUS_PER_NODE=8
++ export WORLD_SIZE=32
++ WORLD_SIZE=32
++ export CUDA_DEVICE_MAX_CONNECTIONS=1
++ CUDA_DEVICE_MAX_CONNECTIONS=1
++ export NCCL_DEBUG=WARN
++ NCCL_DEBUG=WARN
++ export NANOTRON_BENCHMARK=1
++ NANOTRON_BENCHMARK=1
++ export WANDB_MODE=disabled
++ WANDB_MODE=disabled
++ export TORCH_NCCL_ASYNC_ERROR_HANDLING=1
++ TORCH_NCCL_ASYNC_ERROR_HANDLING=1
++ echo '=== GPU Topology ==='
+=== GPU Topology ===
++ nvidia-smi topo -m
+	GPU0	GPU1	GPU2	GPU3	GPU4	GPU5	GPU6	GPU7	CPU Affinity	NUMA Affinity	GPU NUMA ID
+GPU0	 X 	NV18	NV18	NV18	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU1	NV18	 X 	NV18	NV18	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU2	NV18	NV18	 X 	NV18	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU3	NV18	NV18	NV18	 X 	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU4	NV18	NV18	NV18	NV18	 X 	NV18	NV18	NV18	48-95	1		N/A
+GPU5	NV18	NV18	NV18	NV18	NV18	 X 	NV18	NV18	48-95	1		N/A
+GPU6	NV18	NV18	NV18	NV18	NV18	NV18	 X 	NV18	48-95	1		N/A
+GPU7	NV18	NV18	NV18	NV18	NV18	NV18	NV18	 X 	48-95	1		N/A
+
+Legend:
+
+  X    = Self
+  SYS  = Connection traversing PCIe as well as the SMP interconnect between NUMA nodes (e.g., QPI/UPI)
+  NODE = Connection traversing PCIe as well as the interconnect between PCIe Host Bridges within a NUMA node
+  PHB  = Connection traversing PCIe as well as a PCIe Host Bridge (typically the CPU)
+  PXB  = Connection traversing multiple PCIe bridges (without traversing the PCIe Host Bridge)
+  PIX  = Connection traversing at most a single PCIe bridge
+  NV#  = Connection traversing a bonded set of # NVLinks
++ echo ==================
+==================
++ echo 'Master node: ip-26-0-165-59'
+Master node: ip-26-0-165-59
++ echo 'All nodes: ip-26-0-165-59
+ip-26-0-165-131
+ip-26-0-165-164
+ip-26-0-165-202'
+All nodes: ip-26-0-165-59
+ip-26-0-165-131
+ip-26-0-165-164
+ip-26-0-165-202
++ echo 'World size: 32'
+World size: 32
++ srun --wait=0 --kill-on-bad-exit=1 torchrun --nnodes=4 --nproc_per_node=8 --rdzv_id=13460026 --rdzv_backend=c10d --rdzv_endpoint=ip-26-0-165-59:12356 --max_restarts 0 --rdzv_conf timeout=60 /fsx/nouamane/projects/nanotron/run_train.py --config-file benchmark/configs/config_80G_dp4_tp4_pp2_acc64_mbs1_seq4096_zero0_tpmodeRED_vocab131k.yaml
+[2024-12-15 14:53:30,953] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-15 14:53:30,953] torch.distributed.run: [WARNING] 
+[2024-12-15 14:53:30,953] torch.distributed.run: [WARNING] *****************************************
+[2024-12-15 14:53:30,953] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-15 14:53:30,953] torch.distributed.run: [WARNING] *****************************************
+[2024-12-15 14:53:32,033] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-15 14:53:32,034] torch.distributed.run: [WARNING] 
+[2024-12-15 14:53:32,034] torch.distributed.run: [WARNING] *****************************************
+[2024-12-15 14:53:32,034] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-15 14:53:32,034] torch.distributed.run: [WARNING] *****************************************
+[2024-12-15 14:53:32,222] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-15 14:53:32,222] torch.distributed.run: [WARNING] 
+[2024-12-15 14:53:32,222] torch.distributed.run: [WARNING] *****************************************
+[2024-12-15 14:53:32,222] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-15 14:53:32,222] torch.distributed.run: [WARNING] *****************************************
+[2024-12-15 14:53:32,461] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-15 14:53:32,461] torch.distributed.run: [WARNING] 
+[2024-12-15 14:53:32,461] torch.distributed.run: [WARNING] *****************************************
+[2024-12-15 14:53:32,461] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-15 14:53:32,461] torch.distributed.run: [WARNING] *****************************************
+NCCL version 2.18.5+cuda12.2
+12/15/2024 14:53:58 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]: Measuring inter-GPU and intra-node bandwidth...
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+12/15/2024 14:54:15 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]: Bandwidth measurement complete. Time taken: 16.97 seconds
+12/15/2024 14:54:15 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]: Config:
+12/15/2024 14:54:15 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]: Config(general=GeneralArgs(project='debug',
+12/15/2024 14:54:15 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                            run='80G_dp4_tp4_pp2_acc64_mbs1_seq4096_zero0_tpmodeRED_vocab131k',
+12/15/2024 14:54:15 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                            seed=42,
+12/15/2024 14:54:15 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                            step=None,
+12/15/2024 14:54:15 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                            consumed_train_samples=None,
+12/15/2024 14:54:15 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                            benchmark_csv_path=PosixPath('benchmark/results/bench_final2.csv'),
+12/15/2024 14:54:15 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                            ignore_sanity_checks=True),
+12/15/2024 14:54:15 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:        parallelism=ParallelismArgs(dp=4,
+12/15/2024 14:54:15 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                                    pp=2,
+12/15/2024 14:54:15 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                                    tp=4,
+12/15/2024 14:54:15 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                                    pp_engine=<nanotron.parallel.pipeline_parallel.engine.OneForwardOneBackwardPipelineEngine object at 0x7f7244bc4be0>,
+12/15/2024 14:54:15 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                                    tp_mode=<TensorParallelLinearMode.REDUCE_SCATTER: 2>,
+12/15/2024 14:54:15 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                                    tp_linear_async_communication=True,
+12/15/2024 14:54:15 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                                    recompute_layer=False,
+12/15/2024 14:54:15 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                                    tp_recompute_allgather=True,
+12/15/2024 14:54:15 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                                    expert_parallel_size=1),
+12/15/2024 14:54:15 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:        model=ModelArgs(model_config=LlamaConfig(bos_token_id=0,
+12/15/2024 14:54:15 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                                                 eos_token_id=0,
+12/15/2024 14:54:15 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                                                 hidden_act='silu',
+12/15/2024 14:54:15 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                                                 hidden_size=8192,
+12/15/2024 14:54:15 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                                                 initializer_range=0.02,
+12/15/2024 14:54:15 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                                                 intermediate_size=28672,
+12/15/2024 14:54:15 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                                                 is_llama_config=True,
+12/15/2024 14:54:15 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                                                 max_position_embeddings=4096,
+12/15/2024 14:54:15 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                                                 num_attention_heads=64,
+12/15/2024 14:54:15 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                                                 num_hidden_layers=80,
+12/15/2024 14:54:15 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                                                 num_key_value_heads=64,
+12/15/2024 14:54:15 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                                                 pad_token_id=None,
+12/15/2024 14:54:15 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                                                 pretraining_tp=1,
+12/15/2024 14:54:15 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                                                 rms_norm_eps=1e-05,
+12/15/2024 14:54:15 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                                                 rope_scaling=None,
+12/15/2024 14:54:15 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                                                 rope_theta=10000.0,
+12/15/2024 14:54:15 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                                                 rope_interleaved=False,
+12/15/2024 14:54:15 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                                                 tie_word_embeddings=False,
+12/15/2024 14:54:15 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                                                 use_cache=True,
+12/15/2024 14:54:15 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                                                 vocab_size=131072),
+12/15/2024 14:54:15 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                        init_method=RandomInit(std=0.02),
+12/15/2024 14:54:15 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                        dtype=torch.bfloat16,
+12/15/2024 14:54:15 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                        make_vocab_size_divisible_by=1,
+12/15/2024 14:54:15 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                        ddp_bucket_cap_mb=25),
+12/15/2024 14:54:15 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:        tokenizer=TokenizerArgs(tokenizer_name_or_path='robot-test/dummy-tokenizer-wordlevel',
+12/15/2024 14:54:15 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                                tokenizer_revision=None,
+12/15/2024 14:54:15 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                                tokenizer_max_length=None),
+12/15/2024 14:54:15 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:        checkpoints=CheckpointsArgs(checkpoints_path=PosixPath('checkpoints'),
+12/15/2024 14:54:15 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                                    checkpoint_interval=10000,
+12/15/2024 14:54:15 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                                    save_initial_state=False,
+12/15/2024 14:54:15 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                                    save_final_state=False,
+12/15/2024 14:54:15 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                                    resume_checkpoint_path=None,
+12/15/2024 14:54:15 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                                    checkpoints_path_is_shared_file_system=False),
+12/15/2024 14:54:15 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:        logging=LoggingArgs(log_level='info',
+12/15/2024 14:54:15 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                            log_level_replica='info',
+12/15/2024 14:54:15 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                            iteration_step_info_interval=1),
+12/15/2024 14:54:15 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:        tokens=TokensArgs(sequence_length=4096,
+12/15/2024 14:54:15 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                          train_steps=100,
+12/15/2024 14:54:15 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                          micro_batch_size=1,
+12/15/2024 14:54:15 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                          batch_accumulation_per_replica=64,
+12/15/2024 14:54:15 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                          val_check_interval=100,
+12/15/2024 14:54:15 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                          limit_val_batches=0,
+12/15/2024 14:54:15 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                          limit_test_batches=0),
+12/15/2024 14:54:15 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:        optimizer=OptimizerArgs(optimizer_factory=AdamWOptimizerArgs(adam_eps=1e-08,
+12/15/2024 14:54:15 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                                                                     adam_beta1=0.9,
+12/15/2024 14:54:15 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                                                                     adam_beta2=0.95,
+12/15/2024 14:54:15 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                                                                     torch_adam_is_fused=True,
+12/15/2024 14:54:15 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                                                                     name='adamW'),
+12/15/2024 14:54:15 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                                zero_stage=0,
+12/15/2024 14:54:15 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                                weight_decay=0.01,
+12/15/2024 14:54:15 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                                clip_grad=1.0,
+12/15/2024 14:54:15 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                                accumulate_grad_in_fp32=True,
+12/15/2024 14:54:15 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                                learning_rate_scheduler=LRSchedulerArgs(learning_rate=0.0003,
+12/15/2024 14:54:15 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                                                                        lr_warmup_steps=2,
+12/15/2024 14:54:15 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                                                                        lr_warmup_style='linear',
+12/15/2024 14:54:15 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                                                                        lr_decay_style='cosine',
+12/15/2024 14:54:15 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                                                                        lr_decay_steps=13,
+12/15/2024 14:54:15 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                                                                        lr_decay_starting_step=None,
+12/15/2024 14:54:15 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                                                                        min_decay_lr=1e-05)),
+12/15/2024 14:54:15 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:        data_stages=[DatasetStageArgs(name='Stable Training Stage',
+12/15/2024 14:54:15 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                                      start_training_step=1,
+12/15/2024 14:54:15 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                                      data=DataArgs(dataset=None,
+12/15/2024 14:54:15 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                                                    seed=42,
+12/15/2024 14:54:15 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:                                                    num_loading_workers=1))],
+12/15/2024 14:54:15 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:        profiler=None,
+12/15/2024 14:54:15 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:        lighteval=None,
+12/15/2024 14:54:15 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:        s3_upload=None)
+12/15/2024 14:54:15 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]: Model Config:
+12/15/2024 14:54:15 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]: LlamaConfig(bos_token_id=0,
+12/15/2024 14:54:15 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:             eos_token_id=0,
+12/15/2024 14:54:15 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:             hidden_act='silu',
+12/15/2024 14:54:15 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:             hidden_size=8192,
+12/15/2024 14:54:15 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:             initializer_range=0.02,
+12/15/2024 14:54:15 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:             intermediate_size=28672,
+12/15/2024 14:54:15 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:             is_llama_config=True,
+12/15/2024 14:54:15 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:             max_position_embeddings=4096,
+12/15/2024 14:54:15 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:             num_attention_heads=64,
+12/15/2024 14:54:15 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:             num_hidden_layers=80,
+12/15/2024 14:54:15 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:             num_key_value_heads=64,
+12/15/2024 14:54:15 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:             pad_token_id=None,
+12/15/2024 14:54:15 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:             pretraining_tp=1,
+12/15/2024 14:54:15 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:             rms_norm_eps=1e-05,
+12/15/2024 14:54:15 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:             rope_scaling=None,
+12/15/2024 14:54:15 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:             rope_theta=10000.0,
+12/15/2024 14:54:15 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:             rope_interleaved=False,
+12/15/2024 14:54:15 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:             tie_word_embeddings=False,
+12/15/2024 14:54:15 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:             use_cache=True,
+12/15/2024 14:54:15 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]:             vocab_size=131072)
+12/15/2024 14:54:15 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]: Building model..
+12/15/2024 14:54:15 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]: Initialize RoPE Theta = 10000.0
+12/15/2024 14:54:15 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]: Setting PP block ranks...
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+12/15/2024 14:54:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]: Total number of parameters: 80G (152586.06MiB)
+12/15/2024 14:54:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]: Local number of parameters: 10.2G (19537.28MiB)
+12/15/2024 14:54:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]: [After model building] Memory usage: 19537.33MiB. Peak allocated: 19537.34MiB Peak reserved: 22850.00MiB
+12/15/2024 14:54:21 [INFO|DP=0|PP=1|TP=0|ip-26-0-165-202]: Local number of parameters: 9.76G (18609.23MiB)
+12/15/2024 14:54:21 [INFO|DP=0|PP=1|TP=0|ip-26-0-165-202]: [After model building] Memory usage: 18609.28MiB. Peak allocated: 18609.29MiB Peak reserved: 22850.00MiB
+12/15/2024 14:54:21 [INFO|DP=0|PP=1|TP=3|ip-26-0-165-202]: Local number of parameters: 9.76G (18609.23MiB)
+12/15/2024 14:54:21 [INFO|DP=0|PP=0|TP=1|ip-26-0-165-131]: Local number of parameters: 10.2G (19537.28MiB)
+12/15/2024 14:54:21 [INFO|DP=0|PP=0|TP=3|ip-26-0-165-131]: Local number of parameters: 10.2G (19537.28MiB)
+12/15/2024 14:54:21 [INFO|DP=0|PP=1|TP=2|ip-26-0-165-202]: Local number of parameters: 9.76G (18609.23MiB)
+12/15/2024 14:54:21 [INFO|DP=0|PP=1|TP=1|ip-26-0-165-202]: Local number of parameters: 9.76G (18609.23MiB)
+12/15/2024 14:54:21 [INFO|DP=0|PP=1|TP=3|ip-26-0-165-202]: [After model building] Memory usage: 18609.28MiB. Peak allocated: 18609.29MiB Peak reserved: 22850.00MiB
+12/15/2024 14:54:21 [INFO|DP=0|PP=0|TP=2|ip-26-0-165-131]: Local number of parameters: 10.2G (19537.28MiB)
+12/15/2024 14:54:21 [INFO|DP=0|PP=0|TP=1|ip-26-0-165-131]: [After model building] Memory usage: 19537.33MiB. Peak allocated: 19537.34MiB Peak reserved: 22850.00MiB
+12/15/2024 14:54:21 [INFO|DP=0|PP=0|TP=3|ip-26-0-165-131]: [After model building] Memory usage: 19537.33MiB. Peak allocated: 19537.34MiB Peak reserved: 22850.00MiB
+12/15/2024 14:54:21 [INFO|DP=0|PP=1|TP=2|ip-26-0-165-202]: [After model building] Memory usage: 18609.28MiB. Peak allocated: 18609.29MiB Peak reserved: 22850.00MiB
+12/15/2024 14:54:21 [INFO|DP=0|PP=1|TP=1|ip-26-0-165-202]: [After model building] Memory usage: 18609.28MiB. Peak allocated: 18609.29MiB Peak reserved: 22850.00MiB
+12/15/2024 14:54:21 [INFO|DP=0|PP=0|TP=2|ip-26-0-165-131]: [After model building] Memory usage: 19537.33MiB. Peak allocated: 19537.34MiB Peak reserved: 22850.00MiB
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+12/15/2024 14:54:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]: No checkpoint path provided.
+12/15/2024 14:54:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]: Parametrizing model parameters using StandardParametrizator
+12/15/2024 14:54:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-131]: [Optimizer Building] Using LearningRateForSP as learning rate
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 248, in <module>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 183, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 36.35 GiB. GPU 5 has a total capacty of 79.33 GiB of which 30.91 GiB is free. Including non-PyTorch memory, this process has 48.41 GiB memory in use. Of the allocated memory 36.35 GiB is allocated by PyTorch, and 1.14 GiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 248, in <module>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 183, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 248, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 248, in <module>
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 248, in <module>
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 36.35 GiB. GPU 7 has a total capacty of 79.33 GiB of which 31.34 GiB is free. Including non-PyTorch memory, this process has 47.98 GiB memory in use. Of the allocated memory 36.35 GiB is allocated by PyTorch, and 1.14 GiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+        trainer = DistributedTrainer(config_file)trainer = DistributedTrainer(config_file)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 183, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 183, in __init__
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 183, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    optimizer = optimizer_builder(named_param_groups)    
+result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+        self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 36.35 GiB. GPU 5 has a total capacty of 79.33 GiB of which 31.57 GiB is free. Including non-PyTorch memory, this process has 47.75 GiB memory in use. Of the allocated memory 36.35 GiB is allocated by PyTorch, and 1.14 GiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 36.35 GiB. GPU 4 has a total capacty of 79.33 GiB of which 31.60 GiB is free. Including non-PyTorch memory, this process has 47.72 GiB memory in use. Of the allocated memory 36.35 GiB is allocated by PyTorch, and 1.14 GiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 36.35 GiB. GPU 2 has a total capacty of 79.33 GiB of which 31.57 GiB is free. Including non-PyTorch memory, this process has 47.75 GiB memory in use. Of the allocated memory 36.35 GiB is allocated by PyTorch, and 1.14 GiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 248, in <module>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 183, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 36.35 GiB. GPU 2 has a total capacty of 79.33 GiB of which 30.98 GiB is free. Including non-PyTorch memory, this process has 48.33 GiB memory in use. Of the allocated memory 36.35 GiB is allocated by PyTorch, and 1.14 GiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 248, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 248, in <module>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 183, in __init__
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 183, in __init__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 248, in <module>
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 183, in __init__
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 36.35 GiB. GPU 0 has a total capacty of 79.33 GiB of which 31.65 GiB is free. Including non-PyTorch memory, this process has 47.67 GiB memory in use. Of the allocated memory 36.35 GiB is allocated by PyTorch, and 1.14 GiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 36.35 GiB. GPU 7 has a total capacty of 79.33 GiB of which 32.00 GiB is free. Including non-PyTorch memory, this process has 47.32 GiB memory in use. Of the allocated memory 36.35 GiB is allocated by PyTorch, and 1.14 GiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 248, in <module>
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 38.16 GiB. GPU 1 has a total capacty of 79.33 GiB of which 30.08 GiB is free. Including non-PyTorch memory, this process has 49.24 GiB memory in use. Of the allocated memory 38.16 GiB is allocated by PyTorch, and 240.67 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 183, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 36.35 GiB. GPU 1 has a total capacty of 79.33 GiB of which 31.57 GiB is free. Including non-PyTorch memory, this process has 47.75 GiB memory in use. Of the allocated memory 36.35 GiB is allocated by PyTorch, and 1.14 GiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 248, in <module>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 183, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 36.35 GiB. GPU 0 has a total capacty of 79.33 GiB of which 31.06 GiB is free. Including non-PyTorch memory, this process has 48.26 GiB memory in use. Of the allocated memory 36.35 GiB is allocated by PyTorch, and 1.14 GiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 248, in <module>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 183, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 248, in <module>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 183, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 36.35 GiB. GPU 6 has a total capacty of 79.33 GiB of which 31.57 GiB is free. Including non-PyTorch memory, this process has 47.75 GiB memory in use. Of the allocated memory 36.35 GiB is allocated by PyTorch, and 1.14 GiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 36.35 GiB. GPU 3 has a total capacty of 79.33 GiB of which 31.69 GiB is free. Including non-PyTorch memory, this process has 47.63 GiB memory in use. Of the allocated memory 36.35 GiB is allocated by PyTorch, and 1.14 GiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 248, in <module>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 183, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 248, in <module>
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 183, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 38.16 GiB. GPU 0 has a total capacty of 79.33 GiB of which 30.08 GiB is free. Including non-PyTorch memory, this process has 49.24 GiB memory in use. Of the allocated memory 38.16 GiB is allocated by PyTorch, and 240.67 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 248, in <module>
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 36.35 GiB. GPU 4 has a total capacty of 79.33 GiB of which 31.02 GiB is free. Including non-PyTorch memory, this process has 48.30 GiB memory in use. Of the allocated memory 36.35 GiB is allocated by PyTorch, and 1.14 GiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 183, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 248, in <module>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 183, in __init__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 248, in <module>
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 38.16 GiB. GPU 4 has a total capacty of 79.33 GiB of which 30.03 GiB is free. Including non-PyTorch memory, this process has 49.29 GiB memory in use. Of the allocated memory 38.16 GiB is allocated by PyTorch, and 240.67 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 183, in __init__
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    result = OptimizerFromGradientAccumulator(
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 248, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 248, in <module>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 183, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 36.35 GiB. GPU 3 has a total capacty of 79.33 GiB of which 31.02 GiB is free. Including non-PyTorch memory, this process has 48.29 GiB memory in use. Of the allocated memory 36.35 GiB is allocated by PyTorch, and 1.14 GiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    trainer = DistributedTrainer(config_file)
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 183, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 36.35 GiB. GPU 6 has a total capacty of 79.33 GiB of which 30.98 GiB is free. Including non-PyTorch memory, this process has 48.33 GiB memory in use. Of the allocated memory 36.35 GiB is allocated by PyTorch, and 1.14 GiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 248, in <module>
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 183, in __init__
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 248, in <module>
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    trainer = DistributedTrainer(config_file)
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 183, in __init__
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 38.16 GiB. GPU 1 has a total capacty of 79.33 GiB of which 30.66 GiB is free. Including non-PyTorch memory, this process has 48.65 GiB memory in use. Of the allocated memory 38.16 GiB is allocated by PyTorch, and 240.67 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 38.16 GiB. GPU 6 has a total capacty of 79.33 GiB of which 30.66 GiB is free. Including non-PyTorch memory, this process has 48.65 GiB memory in use. Of the allocated memory 38.16 GiB is allocated by PyTorch, and 240.67 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 36.35 GiB. GPU 1 has a total capacty of 79.33 GiB of which 30.91 GiB is free. Including non-PyTorch memory, this process has 48.41 GiB memory in use. Of the allocated memory 36.35 GiB is allocated by PyTorch, and 1.14 GiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 38.16 GiB. GPU 5 has a total capacty of 79.33 GiB of which 30.08 GiB is free. Including non-PyTorch memory, this process has 49.24 GiB memory in use. Of the allocated memory 38.16 GiB is allocated by PyTorch, and 240.67 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 248, in <module>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 183, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 38.16 GiB. GPU 6 has a total capacty of 79.33 GiB of which 30.00 GiB is free. Including non-PyTorch memory, this process has 49.32 GiB memory in use. Of the allocated memory 38.16 GiB is allocated by PyTorch, and 240.67 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 248, in <module>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 183, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 38.16 GiB. GPU 4 has a total capacty of 79.33 GiB of which 30.70 GiB is free. Including non-PyTorch memory, this process has 48.62 GiB memory in use. Of the allocated memory 38.16 GiB is allocated by PyTorch, and 240.67 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 248, in <module>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 183, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 248, in <module>
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 38.16 GiB. GPU 0 has a total capacty of 79.33 GiB of which 30.74 GiB is free. Including non-PyTorch memory, this process has 48.58 GiB memory in use. Of the allocated memory 38.16 GiB is allocated by PyTorch, and 240.67 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 183, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 38.16 GiB. GPU 2 has a total capacty of 79.33 GiB of which 30.66 GiB is free. Including non-PyTorch memory, this process has 48.65 GiB memory in use. Of the allocated memory 38.16 GiB is allocated by PyTorch, and 240.67 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 248, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 248, in <module>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 183, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 183, in __init__
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 38.16 GiB. GPU 3 has a total capacty of 79.33 GiB of which 30.20 GiB is free. Including non-PyTorch memory, this process has 49.12 GiB memory in use. Of the allocated memory 38.16 GiB is allocated by PyTorch, and 240.67 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 38.16 GiB. GPU 2 has a total capacty of 79.33 GiB of which 30.00 GiB is free. Including non-PyTorch memory, this process has 49.32 GiB memory in use. Of the allocated memory 38.16 GiB is allocated by PyTorch, and 240.67 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 248, in <module>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 183, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 38.16 GiB. GPU 7 has a total capacty of 79.33 GiB of which 30.51 GiB is free. Including non-PyTorch memory, this process has 48.81 GiB memory in use. Of the allocated memory 38.16 GiB is allocated by PyTorch, and 240.67 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 248, in <module>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 183, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 38.16 GiB. GPU 5 has a total capacty of 79.33 GiB of which 30.66 GiB is free. Including non-PyTorch memory, this process has 48.65 GiB memory in use. Of the allocated memory 38.16 GiB is allocated by PyTorch, and 240.67 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 248, in <module>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 183, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 38.16 GiB. GPU 3 has a total capacty of 79.33 GiB of which 30.78 GiB is free. Including non-PyTorch memory, this process has 48.54 GiB memory in use. Of the allocated memory 38.16 GiB is allocated by PyTorch, and 240.67 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 248, in <module>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 183, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 38.16 GiB. GPU 7 has a total capacty of 79.33 GiB of which 31.09 GiB is free. Including non-PyTorch memory, this process has 48.22 GiB memory in use. Of the allocated memory 38.16 GiB is allocated by PyTorch, and 240.67 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+[2024-12-15 14:54:44,188] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 261089 closing signal SIGTERM
+[2024-12-15 14:54:44,188] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 261090 closing signal SIGTERM
+[2024-12-15 14:54:44,188] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 261091 closing signal SIGTERM
+[2024-12-15 14:54:44,188] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 261092 closing signal SIGTERM
+[2024-12-15 14:54:44,188] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 261094 closing signal SIGTERM
+[2024-12-15 14:54:44,188] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 261095 closing signal SIGTERM
+[2024-12-15 14:54:44,188] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 261096 closing signal SIGTERM
+[2024-12-15 14:54:44,189] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 358143 closing signal SIGTERM
+[2024-12-15 14:54:44,189] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 358145 closing signal SIGTERM
+[2024-12-15 14:54:44,189] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 358146 closing signal SIGTERM
+[2024-12-15 14:54:44,189] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 358147 closing signal SIGTERM
+[2024-12-15 14:54:44,189] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 358148 closing signal SIGTERM
+[2024-12-15 14:54:44,190] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 358149 closing signal SIGTERM
+[2024-12-15 14:54:47,021] torch.distributed.elastic.multiprocessing.api: [ERROR] failed (exitcode: 1) local_rank: 0 (pid: 358142) of binary: /fsx/nouamane/miniconda/envs/2-1-cu121/bin/python
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 33, in <module>
+    sys.exit(load_entry_point('torch==2.1.1', 'console_scripts', 'torchrun')())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 264, in launch_agent
+    raise ChildFailedError(
+torch.distributed.elastic.multiprocessing.errors.ChildFailedError: 
+============================================================
+/fsx/nouamane/projects/nanotron/run_train.py FAILED
+------------------------------------------------------------
+Failures:
+[1]:
+  time      : 2024-12-15_14:54:44
+  host      : ip-26-0-165-131.ec2.internal
+  rank      : 2 (local_rank: 2)
+  exitcode  : 1 (pid: 358144)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+------------------------------------------------------------
+Root Cause (first observed failure):
+[0]:
+  time      : 2024-12-15_14:54:44
+  host      : ip-26-0-165-131.ec2.internal
+  rank      : 0 (local_rank: 0)
+  exitcode  : 1 (pid: 358142)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+============================================================
+srun: error: ip-26-0-165-131: task 1: Exited with exit code 1
+srun: launch/slurm: _step_signal: Terminating StepId=13460026.0
+slurmstepd: error: *** STEP 13460026.0 ON ip-26-0-165-59 CANCELLED AT 2024-12-15T14:54:47 ***
+[2024-12-15 14:54:47,370] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-15 14:54:47,370] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 261090 closing signal SIGTERM
+[2024-12-15 14:54:47,370] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 261091 closing signal SIGTERM
+[2024-12-15 14:54:47,370] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 261096 closing signal SIGTERM
+[2024-12-15 14:54:47,371] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-15 14:54:47,371] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 300487 closing signal SIGTERM
+[2024-12-15 14:54:47,371] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 300488 closing signal SIGTERM
+[2024-12-15 14:54:47,371] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-15 14:54:47,371] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 300490 closing signal SIGTERM
+[2024-12-15 14:54:47,371] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 300491 closing signal SIGTERM
+[2024-12-15 14:54:47,371] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 115191 closing signal SIGTERM
+[2024-12-15 14:54:47,371] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 300492 closing signal SIGTERM
+[2024-12-15 14:54:47,371] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 300493 closing signal SIGTERM
+[2024-12-15 14:54:47,371] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 115192 closing signal SIGTERM
+[2024-12-15 14:54:47,371] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 115193 closing signal SIGTERM
+[2024-12-15 14:54:47,371] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 115194 closing signal SIGTERM
+[2024-12-15 14:54:47,371] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 115196 closing signal SIGTERM
+[2024-12-15 14:54:47,371] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 115197 closing signal SIGTERM
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 33, in <module>
+    sys.exit(load_entry_point('torch==2.1.1', 'console_scripts', 'torchrun')())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 878, in _invoke_run
+    run_result = self._monitor_workers(self._worker_group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/local_elastic_agent.py", line 307, in _monitor_workers
+    result = self._pcontext.wait(0)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 288, in wait
+    return self._poll()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 668, in _poll
+    self.close()  # terminate all running procs
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 331, in close
+    self._close(death_sig=death_sig, timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 713, in _close
+    handler.proc.wait(time_to_wait)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1209, in wait
+    return self._wait(timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1953, in _wait
+    time.sleep(delay)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 261012 got signal: 15
+srun: error: ip-26-0-165-59: task 0: Exited with exit code 1
+[2024-12-15 14:54:48,319] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-165-202.ec2.internal_115113_0' has failed to send a keep-alive heartbeat to the rendezvous '13460026' due to an error of type RendezvousConnectionError.
+[2024-12-15 14:54:49,168] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-165-164.ec2.internal_300411_0' has failed to send a keep-alive heartbeat to the rendezvous '13460026' due to an error of type RendezvousConnectionError.
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 33, in <module>
+    sys.exit(load_entry_point('torch==2.1.1', 'console_scripts', 'torchrun')())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 115113 got signal: 15
+srun: error: ip-26-0-165-202: task 3: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 33, in <module>
+    sys.exit(load_entry_point('torch==2.1.1', 'console_scripts', 'torchrun')())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 300411 got signal: 15
+srun: error: ip-26-0-165-164: task 2: Exited with exit code 1
diff --git a/logs/13460155-bench_1.34G_dp4_tp4_pp4_acc64_mbs1_seq4096_zero0_tpmodeRED_vocab131k.out b/logs/13460155-bench_1.34G_dp4_tp4_pp4_acc64_mbs1_seq4096_zero0_tpmodeRED_vocab131k.out
new file mode 100644
index 0000000000000000000000000000000000000000..db88044658aebfbbffd016c8d96ea7913c0772e6
--- /dev/null
+++ b/logs/13460155-bench_1.34G_dp4_tp4_pp4_acc64_mbs1_seq4096_zero0_tpmodeRED_vocab131k.out
@@ -0,0 +1,926 @@
++ source /etc/profile.d/modules.sh
+++ . /usr/share/modules/init/bash
++++ unset _mlshdbg
++++ '[' 0 = 1 ']'
++++ unset _mlre _mlIFS
++++ '[' -n x ']'
++++ _mlIFS=' 	
+'
++++ IFS=' '
++++ '[' -n '' ']'
+++++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash autoinit
++++ _mlcode='module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+test 0;'
++++ _mlret=0
++++ '[' -n x ']'
++++ IFS=' 	
+'
++++ unset _mlIFS
++++ unset _mlre _mlv _mlrv
++++ '[' 0 -eq 0 ']'
++++ eval 'module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+test 0;'
+++++ MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl
+++++ export MODULES_CMD
+++++ ENV=/usr/share/modules/init/profile.sh
+++++ export ENV
+++++ BASH_ENV=/usr/share/modules/init/bash
+++++ export BASH_ENV
+++++ MODULESHOME=/usr/share/modules
+++++ export MODULESHOME
+++++ test 0
++++ '[' 0 = 1 ']'
++++ '[' -t 2 ']'
++++ export -f module
++++ export -f switchml
++++ '[' 5 -ge 3 ']'
++++ [[ ehxB =~ i ]]
++++ [[ ! :/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin: =~ :/usr/bin: ]]
++++ '[' '!' -n x ']'
+++++ manpath
++++ [[ ! :/admin/home/nouamane/.local/share/man:/fsx/nouamane/miniconda/envs/2-1-cu121/man:/fsx/nouamane/miniconda/envs/2-1-cu121/share/man:/fsx/nouamane/miniconda/man:/fsx/nouamane/miniconda/share/man:/opt/slurm/share/man:/opt/amazon/openmpi/share/man:/opt/amazon/efa/share/man:/usr/local/man:/usr/local/share/man:/usr/share/man:/admin/home/nouamane/.fzf/man:: =~ :/usr/share/man: ]]
++++ unset _mlcode _mlret
++++ '[' -n '' ']'
++ module load cuda/12.1
++ unset _mlshdbg
++ '[' 0 = 1 ']'
++ unset _mlre _mlIFS
++ '[' -n x ']'
++ _mlIFS=' 	
+'
++ IFS=' '
++ '[' -n '' ']'
+++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash load cuda/12.1
++ eval 'CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include;' export 'CPATH;
+LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/opt/aws-ofi-nccl/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/targets/x86_64-linux/lib/:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/lib:/usr/lib;' export 'LD_LIBRARY_PATH;
+FI_EFA_FORK_SAFE=1;' export 'FI_EFA_FORK_SAFE;
+MANPATH=/usr/local/cuda-12.1/share/man::;' export 'MANPATH;
+LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64;' export 'LIBRARY_PATH;
+_LMFILES_=/usr/share/modules/modulefiles/cuda/12.1;' export '_LMFILES_;
+LOADEDMODULES=cuda/12.1;' export 'LOADEDMODULES;
+MPI_PATH=/opt/amazon/openmpi;' export 'MPI_PATH;
+NCCL_HOME_modshare=/opt/nccl/build:1;' export 'NCCL_HOME_modshare;
+NCCL_PROTO=simple;' export 'NCCL_PROTO;
+MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1;' export 'MANPATH_modshare;
+LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1;' export 'LIBRARY_PATH_modshare;
+NCCL_SOCKET_IFNAME=enp;' export 'NCCL_SOCKET_IFNAME;
+AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl;' export 'AWS_OFI_NCCL_HOME;
+NCCL_HOME=/opt/nccl/build;' export 'NCCL_HOME;
+FI_PROVIDER=efa;' export 'FI_PROVIDER;
+AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1;' export 'AWS_OFI_NCCL_HOME_modshare;
+CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1;' export 'CPATH_modshare;
+LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib/:1:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2:/opt/aws-ofi-nccl/lib:2:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/efa/lib:1:/usr/lib:1;' export 'LD_LIBRARY_PATH_modshare;
+FI_EFA_ENABLE_SHM_TRANSFER=1;' export 'FI_EFA_ENABLE_SHM_TRANSFER;
+_LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1;' export '_LMFILES__modshare;
+LOADEDMODULES_modshare=cuda/12.1:1;' export 'LOADEDMODULES_modshare;
+MPI_PATH_modshare=/opt/amazon/openmpi:1;' export 'MPI_PATH_modshare;
+PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin;' export 'PATH;
+CUDA_HOME=/usr/local/cuda-12.1;' export 'CUDA_HOME;
+PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/usr/local/cuda-12.1/include:1:/opt/amazon/efa/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/usr/sbin:1:/fsx/nouamane/miniconda/bin:1:/opt/slurm/bin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:2:/opt/amazon/openmpi/bin:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1;' export 'PATH_modshare;
+test' '0;'
+++ CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include
+++ export CPATH
+++ LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/opt/aws-ofi-nccl/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/targets/x86_64-linux/lib/:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/lib:/usr/lib
+++ export LD_LIBRARY_PATH
+++ FI_EFA_FORK_SAFE=1
+++ export FI_EFA_FORK_SAFE
+++ MANPATH=/usr/local/cuda-12.1/share/man::
+++ export MANPATH
+++ LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64
+++ export LIBRARY_PATH
+++ _LMFILES_=/usr/share/modules/modulefiles/cuda/12.1
+++ export _LMFILES_
+++ LOADEDMODULES=cuda/12.1
+++ export LOADEDMODULES
+++ MPI_PATH=/opt/amazon/openmpi
+++ export MPI_PATH
+++ NCCL_HOME_modshare=/opt/nccl/build:1
+++ export NCCL_HOME_modshare
+++ NCCL_PROTO=simple
+++ export NCCL_PROTO
+++ MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1
+++ export MANPATH_modshare
+++ LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1
+++ export LIBRARY_PATH_modshare
+++ NCCL_SOCKET_IFNAME=enp
+++ export NCCL_SOCKET_IFNAME
+++ AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl
+++ export AWS_OFI_NCCL_HOME
+++ NCCL_HOME=/opt/nccl/build
+++ export NCCL_HOME
+++ FI_PROVIDER=efa
+++ export FI_PROVIDER
+++ AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1
+++ export AWS_OFI_NCCL_HOME_modshare
+++ CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1
+++ export CPATH_modshare
+++ LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib/:1:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2:/opt/aws-ofi-nccl/lib:2:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/efa/lib:1:/usr/lib:1
+++ export LD_LIBRARY_PATH_modshare
+++ FI_EFA_ENABLE_SHM_TRANSFER=1
+++ export FI_EFA_ENABLE_SHM_TRANSFER
+++ _LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1
+++ export _LMFILES__modshare
+++ LOADEDMODULES_modshare=cuda/12.1:1
+++ export LOADEDMODULES_modshare
+++ MPI_PATH_modshare=/opt/amazon/openmpi:1
+++ export MPI_PATH_modshare
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin
+++ export PATH
+++ CUDA_HOME=/usr/local/cuda-12.1
+++ export CUDA_HOME
+++ PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/usr/local/cuda-12.1/include:1:/opt/amazon/efa/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/usr/sbin:1:/fsx/nouamane/miniconda/bin:1:/opt/slurm/bin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:2:/opt/amazon/openmpi/bin:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1
+++ export PATH_modshare
+++ test 0
++ _mlstatus=0
++ '[' -n x ']'
++ IFS=' 	
+'
++ unset _mlre _mlv _mlrv _mlIFS
++ '[' -n '' ']'
++ unset _mlshdbg
++ return 0
++ source /fsx/nouamane/miniconda/bin/activate
+++ _CONDA_ROOT=/fsx/nouamane/miniconda
+++ . /fsx/nouamane/miniconda/etc/profile.d/conda.sh
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ '[' -z x ']'
+++ conda activate
+++ local cmd=activate
+++ case "$cmd" in
+++ __conda_activate activate
+++ '[' -n '' ']'
+++ local ask_conda
++++ PS1=
++++ __conda_exe shell.posix activate
++++ /fsx/nouamane/miniconda/bin/conda shell.posix activate
+++ ask_conda='PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin'\''
+export CONDA_SHLVL='\''1'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\'''
+++ eval 'PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin'\''
+export CONDA_SHLVL='\''1'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\'''
++++ PS1='(base) '
++++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin
++++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin
++++ export CONDA_SHLVL=1
++++ CONDA_SHLVL=1
++++ export 'CONDA_PROMPT_MODIFIER=(base) '
++++ CONDA_PROMPT_MODIFIER='(base) '
+++ __conda_hashr
+++ '[' -n '' ']'
+++ '[' -n '' ']'
+++ hash -r
++ conda activate 2-1-cu121
++ local cmd=activate
++ case "$cmd" in
++ __conda_activate activate 2-1-cu121
++ '[' -n '' ']'
++ local ask_conda
+++ PS1='(base) '
+++ __conda_exe shell.posix activate 2-1-cu121
+++ /fsx/nouamane/miniconda/bin/conda shell.posix activate 2-1-cu121
++ ask_conda='PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''2'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_1='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
++ eval 'PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''2'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_1='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
+++ PS1='(2-1-cu121) '
+++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin
+++ export CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ export CONDA_SHLVL=2
+++ CONDA_SHLVL=2
+++ export CONDA_DEFAULT_ENV=2-1-cu121
+++ CONDA_DEFAULT_ENV=2-1-cu121
+++ export 'CONDA_PROMPT_MODIFIER=(2-1-cu121) '
+++ CONDA_PROMPT_MODIFIER='(2-1-cu121) '
+++ export CONDA_PREFIX_1=/fsx/nouamane/miniconda
+++ CONDA_PREFIX_1=/fsx/nouamane/miniconda
+++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ export _CE_M=
+++ _CE_M=
+++ export _CE_CONDA=
+++ _CE_CONDA=
+++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh
++++ test -n ''
++++ xml_catalog_files_libxml2=
++++ XML_CATALOG_FILES=
++++ conda_catalog_files=
++++ ifs_libxml2=' 	
+'
++++ IFS=' '
++++ rem=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ for pre in ${rem}
++++ test '' = /fsx/nouamane/miniconda/envs/2-1-cu121
++++ conda_catalog_files=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ rem=
++++ IFS=' 	
+'
++++ conda_catalog_files='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ export 'XML_CATALOG_FILES=file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ XML_CATALOG_FILES='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ unset conda_catalog_files ifs_libxml2 rem
++ __conda_hashr
++ '[' -n '' ']'
++ '[' -n '' ']'
++ hash -r
++ export PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin
++ PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin
+++ scontrol show hostnames 'ip-26-0-160-[103,242],ip-26-0-165-213,ip-26-0-168-238,ip-26-0-169-[132,139],ip-26-0-172-252,ip-26-0-173-202'
++ export 'NODELIST=ip-26-0-160-103
+ip-26-0-160-242
+ip-26-0-165-213
+ip-26-0-168-238
+ip-26-0-169-132
+ip-26-0-169-139
+ip-26-0-172-252
+ip-26-0-173-202'
++ NODELIST='ip-26-0-160-103
+ip-26-0-160-242
+ip-26-0-165-213
+ip-26-0-168-238
+ip-26-0-169-132
+ip-26-0-169-139
+ip-26-0-172-252
+ip-26-0-173-202'
+++ head -n1
+++ scontrol show hostnames 'ip-26-0-160-[103,242],ip-26-0-165-213,ip-26-0-168-238,ip-26-0-169-[132,139],ip-26-0-172-252,ip-26-0-173-202'
++ export MASTER_NODE=ip-26-0-160-103
++ MASTER_NODE=ip-26-0-160-103
++ export MASTER_PORT=12356
++ MASTER_PORT=12356
++ export NNODES=8
++ NNODES=8
++ export GPUS_PER_NODE=8
++ GPUS_PER_NODE=8
++ export WORLD_SIZE=64
++ WORLD_SIZE=64
++ export CUDA_DEVICE_MAX_CONNECTIONS=1
++ CUDA_DEVICE_MAX_CONNECTIONS=1
++ export NCCL_DEBUG=WARN
++ NCCL_DEBUG=WARN
++ export NANOTRON_BENCHMARK=1
++ NANOTRON_BENCHMARK=1
++ export WANDB_MODE=disabled
++ WANDB_MODE=disabled
++ export TORCH_NCCL_ASYNC_ERROR_HANDLING=1
++ TORCH_NCCL_ASYNC_ERROR_HANDLING=1
++ echo '=== GPU Topology ==='
+=== GPU Topology ===
++ nvidia-smi topo -m
+	GPU0	GPU1	GPU2	GPU3	GPU4	GPU5	GPU6	GPU7	CPU Affinity	NUMA Affinity	GPU NUMA ID
+GPU0	 X 	NV18	NV18	NV18	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU1	NV18	 X 	NV18	NV18	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU2	NV18	NV18	 X 	NV18	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU3	NV18	NV18	NV18	 X 	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU4	NV18	NV18	NV18	NV18	 X 	NV18	NV18	NV18	48-95	1		N/A
+GPU5	NV18	NV18	NV18	NV18	NV18	 X 	NV18	NV18	48-95	1		N/A
+GPU6	NV18	NV18	NV18	NV18	NV18	NV18	 X 	NV18	48-95	1		N/A
+GPU7	NV18	NV18	NV18	NV18	NV18	NV18	NV18	 X 	48-95	1		N/A
+
+Legend:
+
+  X    = Self
+  SYS  = Connection traversing PCIe as well as the SMP interconnect between NUMA nodes (e.g., QPI/UPI)
+  NODE = Connection traversing PCIe as well as the interconnect between PCIe Host Bridges within a NUMA node
+  PHB  = Connection traversing PCIe as well as a PCIe Host Bridge (typically the CPU)
+  PXB  = Connection traversing multiple PCIe bridges (without traversing the PCIe Host Bridge)
+  PIX  = Connection traversing at most a single PCIe bridge
+  NV#  = Connection traversing a bonded set of # NVLinks
++ echo ==================
+==================
++ echo 'Master node: ip-26-0-160-103'
+Master node: ip-26-0-160-103
++ echo 'All nodes: ip-26-0-160-103
+ip-26-0-160-242
+ip-26-0-165-213
+ip-26-0-168-238
+ip-26-0-169-132
+ip-26-0-169-139
+ip-26-0-172-252
+ip-26-0-173-202'
+All nodes: ip-26-0-160-103
+ip-26-0-160-242
+ip-26-0-165-213
+ip-26-0-168-238
+ip-26-0-169-132
+ip-26-0-169-139
+ip-26-0-172-252
+ip-26-0-173-202
++ echo 'World size: 64'
+World size: 64
++ srun --wait=0 --kill-on-bad-exit=1 torchrun --nnodes=8 --nproc_per_node=8 --rdzv_id=13460155 --rdzv_backend=c10d --rdzv_endpoint=ip-26-0-160-103:12356 --max_restarts 0 --rdzv_conf timeout=60 /fsx/nouamane/projects/nanotron/run_train.py --config-file benchmark/configs/config_1.34G_dp4_tp4_pp4_acc64_mbs1_seq4096_zero0_tpmodeRED_vocab131k.yaml
+[2024-12-15 19:34:07,145] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-15 19:34:07,143] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-15 19:34:07,147] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-15 19:34:07,148] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-15 19:34:07,148] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-15 19:34:07,148] torch.distributed.run: [WARNING] 
+[2024-12-15 19:34:07,148] torch.distributed.run: [WARNING] *****************************************
+[2024-12-15 19:34:07,148] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-15 19:34:07,148] torch.distributed.run: [WARNING] *****************************************
+[2024-12-15 19:34:07,154] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-15 19:34:07,155] torch.distributed.run: [WARNING] 
+[2024-12-15 19:34:07,155] torch.distributed.run: [WARNING] *****************************************
+[2024-12-15 19:34:07,155] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-15 19:34:07,155] torch.distributed.run: [WARNING] *****************************************
+[2024-12-15 19:34:07,156] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-15 19:34:07,164] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-15 19:34:07,143] torch.distributed.run: [WARNING] 
+[2024-12-15 19:34:07,143] torch.distributed.run: [WARNING] *****************************************
+[2024-12-15 19:34:07,143] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-15 19:34:07,143] torch.distributed.run: [WARNING] *****************************************
+[2024-12-15 19:34:07,147] torch.distributed.run: [WARNING] 
+[2024-12-15 19:34:07,147] torch.distributed.run: [WARNING] *****************************************
+[2024-12-15 19:34:07,147] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-15 19:34:07,147] torch.distributed.run: [WARNING] *****************************************
+[2024-12-15 19:34:07,148] torch.distributed.run: [WARNING] 
+[2024-12-15 19:34:07,148] torch.distributed.run: [WARNING] *****************************************
+[2024-12-15 19:34:07,148] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-15 19:34:07,148] torch.distributed.run: [WARNING] *****************************************
+[2024-12-15 19:34:07,145] torch.distributed.run: [WARNING] 
+[2024-12-15 19:34:07,145] torch.distributed.run: [WARNING] *****************************************
+[2024-12-15 19:34:07,145] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-15 19:34:07,145] torch.distributed.run: [WARNING] *****************************************
+[2024-12-15 19:34:07,156] torch.distributed.run: [WARNING] 
+[2024-12-15 19:34:07,156] torch.distributed.run: [WARNING] *****************************************
+[2024-12-15 19:34:07,156] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-15 19:34:07,156] torch.distributed.run: [WARNING] *****************************************
+[2024-12-15 19:34:07,164] torch.distributed.run: [WARNING] 
+[2024-12-15 19:34:07,164] torch.distributed.run: [WARNING] *****************************************
+[2024-12-15 19:34:07,164] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-15 19:34:07,164] torch.distributed.run: [WARNING] *****************************************
+NCCL version 2.18.5+cuda12.2
+12/15/2024 19:34:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]: Measuring inter-GPU and intra-node bandwidth...
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+12/15/2024 19:34:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]: Bandwidth measurement complete. Time taken: 16.92 seconds
+12/15/2024 19:34:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]: Config:
+12/15/2024 19:34:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]: Config(general=GeneralArgs(project='debug',
+12/15/2024 19:34:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                            run='1.34G_dp4_tp4_pp4_acc64_mbs1_seq4096_zero0_tpmodeRED_vocab131k',
+12/15/2024 19:34:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                            seed=42,
+12/15/2024 19:34:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                            step=None,
+12/15/2024 19:34:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                            consumed_train_samples=None,
+12/15/2024 19:34:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                            benchmark_csv_path=PosixPath('benchmark/results/bench_final2.csv'),
+12/15/2024 19:34:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                            ignore_sanity_checks=True),
+12/15/2024 19:34:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:        parallelism=ParallelismArgs(dp=4,
+12/15/2024 19:34:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                    pp=4,
+12/15/2024 19:34:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                    tp=4,
+12/15/2024 19:34:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                    pp_engine=<nanotron.parallel.pipeline_parallel.engine.OneForwardOneBackwardPipelineEngine object at 0x7f34e0fbce20>,
+12/15/2024 19:34:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                    tp_mode=<TensorParallelLinearMode.REDUCE_SCATTER: 2>,
+12/15/2024 19:34:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                    tp_linear_async_communication=True,
+12/15/2024 19:34:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                    recompute_layer=False,
+12/15/2024 19:34:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                    tp_recompute_allgather=True,
+12/15/2024 19:34:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                    expert_parallel_size=1),
+12/15/2024 19:34:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:        model=ModelArgs(model_config=LlamaConfig(bos_token_id=0,
+12/15/2024 19:34:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                                 eos_token_id=0,
+12/15/2024 19:34:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                                 hidden_act='silu',
+12/15/2024 19:34:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                                 hidden_size=2048,
+12/15/2024 19:34:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                                 initializer_range=0.02,
+12/15/2024 19:34:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                                 intermediate_size=8192,
+12/15/2024 19:34:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                                 is_llama_config=True,
+12/15/2024 19:34:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                                 max_position_embeddings=4096,
+12/15/2024 19:34:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                                 num_attention_heads=32,
+12/15/2024 19:34:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                                 num_hidden_layers=16,
+12/15/2024 19:34:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                                 num_key_value_heads=32,
+12/15/2024 19:34:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                                 pad_token_id=None,
+12/15/2024 19:34:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                                 pretraining_tp=1,
+12/15/2024 19:34:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                                 rms_norm_eps=1e-05,
+12/15/2024 19:34:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                                 rope_scaling=None,
+12/15/2024 19:34:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                                 rope_theta=10000.0,
+12/15/2024 19:34:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                                 rope_interleaved=False,
+12/15/2024 19:34:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                                 tie_word_embeddings=True,
+12/15/2024 19:34:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                                 use_cache=True,
+12/15/2024 19:34:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                                 vocab_size=131072),
+12/15/2024 19:34:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                        init_method=RandomInit(std=0.02),
+12/15/2024 19:34:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                        dtype=torch.bfloat16,
+12/15/2024 19:34:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                        make_vocab_size_divisible_by=1,
+12/15/2024 19:34:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                        ddp_bucket_cap_mb=25),
+12/15/2024 19:34:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:        tokenizer=TokenizerArgs(tokenizer_name_or_path='robot-test/dummy-tokenizer-wordlevel',
+12/15/2024 19:34:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                tokenizer_revision=None,
+12/15/2024 19:34:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                tokenizer_max_length=None),
+12/15/2024 19:34:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:        checkpoints=CheckpointsArgs(checkpoints_path=PosixPath('checkpoints'),
+12/15/2024 19:34:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                    checkpoint_interval=10000,
+12/15/2024 19:34:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                    save_initial_state=False,
+12/15/2024 19:34:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                    save_final_state=False,
+12/15/2024 19:34:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                    resume_checkpoint_path=None,
+12/15/2024 19:34:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                    checkpoints_path_is_shared_file_system=False),
+12/15/2024 19:34:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:        logging=LoggingArgs(log_level='info',
+12/15/2024 19:34:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                            log_level_replica='info',
+12/15/2024 19:34:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                            iteration_step_info_interval=1),
+12/15/2024 19:34:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:        tokens=TokensArgs(sequence_length=4096,
+12/15/2024 19:34:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                          train_steps=100,
+12/15/2024 19:34:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                          micro_batch_size=1,
+12/15/2024 19:34:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                          batch_accumulation_per_replica=64,
+12/15/2024 19:34:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                          val_check_interval=100,
+12/15/2024 19:34:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                          limit_val_batches=0,
+12/15/2024 19:34:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                          limit_test_batches=0),
+12/15/2024 19:34:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:        optimizer=OptimizerArgs(optimizer_factory=AdamWOptimizerArgs(adam_eps=1e-08,
+12/15/2024 19:34:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                                                     adam_beta1=0.9,
+12/15/2024 19:34:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                                                     adam_beta2=0.95,
+12/15/2024 19:34:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                                                     torch_adam_is_fused=True,
+12/15/2024 19:34:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                                                     name='adamW'),
+12/15/2024 19:34:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                zero_stage=0,
+12/15/2024 19:34:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                weight_decay=0.01,
+12/15/2024 19:34:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                clip_grad=1.0,
+12/15/2024 19:34:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                accumulate_grad_in_fp32=True,
+12/15/2024 19:34:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                learning_rate_scheduler=LRSchedulerArgs(learning_rate=0.0003,
+12/15/2024 19:34:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                                                        lr_warmup_steps=2,
+12/15/2024 19:34:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                                                        lr_warmup_style='linear',
+12/15/2024 19:34:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                                                        lr_decay_style='cosine',
+12/15/2024 19:34:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                                                        lr_decay_steps=13,
+12/15/2024 19:34:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                                                        lr_decay_starting_step=None,
+12/15/2024 19:34:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                                                        min_decay_lr=1e-05)),
+12/15/2024 19:34:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:        data_stages=[DatasetStageArgs(name='Stable Training Stage',
+12/15/2024 19:34:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                      start_training_step=1,
+12/15/2024 19:34:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                      data=DataArgs(dataset=None,
+12/15/2024 19:34:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                                    seed=42,
+12/15/2024 19:34:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:                                                    num_loading_workers=1))],
+12/15/2024 19:34:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:        profiler=None,
+12/15/2024 19:34:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:        lighteval=None,
+12/15/2024 19:34:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:        s3_upload=None)
+12/15/2024 19:34:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]: Model Config:
+12/15/2024 19:34:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]: LlamaConfig(bos_token_id=0,
+12/15/2024 19:34:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:             eos_token_id=0,
+12/15/2024 19:34:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:             hidden_act='silu',
+12/15/2024 19:34:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:             hidden_size=2048,
+12/15/2024 19:34:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:             initializer_range=0.02,
+12/15/2024 19:34:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:             intermediate_size=8192,
+12/15/2024 19:34:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:             is_llama_config=True,
+12/15/2024 19:34:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:             max_position_embeddings=4096,
+12/15/2024 19:34:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:             num_attention_heads=32,
+12/15/2024 19:34:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:             num_hidden_layers=16,
+12/15/2024 19:34:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:             num_key_value_heads=32,
+12/15/2024 19:34:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:             pad_token_id=None,
+12/15/2024 19:34:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:             pretraining_tp=1,
+12/15/2024 19:34:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:             rms_norm_eps=1e-05,
+12/15/2024 19:34:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:             rope_scaling=None,
+12/15/2024 19:34:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:             rope_theta=10000.0,
+12/15/2024 19:34:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:             rope_interleaved=False,
+12/15/2024 19:34:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:             tie_word_embeddings=True,
+12/15/2024 19:34:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:             use_cache=True,
+12/15/2024 19:34:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:             vocab_size=131072)
+12/15/2024 19:34:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]: Building model..
+12/15/2024 19:34:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]: Initialize RoPE Theta = 10000.0
+12/15/2024 19:34:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]: Setting PP block ranks...
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+12/15/2024 19:34:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]: Total number of parameters: 1.61G (3072.52MiB)
+12/15/2024 19:34:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]: Local number of parameters: 168M (320.05MiB)
+12/15/2024 19:34:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]: [After model building] Memory usage: 320.06MiB. Peak allocated: 5408.00MiB Peak reserved: 22818.00MiB
+12/15/2024 19:34:55 [INFO|DP=0|PP=0|TP=3|ip-26-0-160-103]: Local number of parameters: 168M (320.05MiB)
+12/15/2024 19:34:55 [INFO|DP=0|PP=0|TP=2|ip-26-0-160-103]: Local number of parameters: 168M (320.05MiB)
+12/15/2024 19:34:55 [INFO|DP=0|PP=0|TP=3|ip-26-0-160-103]: [After model building] Memory usage: 320.06MiB. Peak allocated: 5408.00MiB Peak reserved: 22818.00MiB
+12/15/2024 19:34:55 [INFO|DP=0|PP=0|TP=2|ip-26-0-160-103]: [After model building] Memory usage: 320.06MiB. Peak allocated: 5408.00MiB Peak reserved: 22818.00MiB
+12/15/2024 19:34:55 [INFO|DP=0|PP=2|TP=0|ip-26-0-169-132]: Local number of parameters: 83.9M (160.04MiB)
+12/15/2024 19:34:55 [INFO|DP=0|PP=1|TP=0|ip-26-0-165-213]: Local number of parameters: 83.9M (160.04MiB)
+12/15/2024 19:34:55 [INFO|DP=0|PP=2|TP=0|ip-26-0-169-132]: [After model building] Memory usage: 160.05MiB. Peak allocated: 5408.00MiB Peak reserved: 22818.00MiB
+12/15/2024 19:34:55 [INFO|DP=0|PP=1|TP=0|ip-26-0-165-213]: [After model building] Memory usage: 160.05MiB. Peak allocated: 5408.00MiB Peak reserved: 22818.00MiB
+12/15/2024 19:34:55 [INFO|DP=0|PP=2|TP=2|ip-26-0-169-132]: Local number of parameters: 83.9M (160.04MiB)
+12/15/2024 19:34:55 [INFO|DP=0|PP=0|TP=1|ip-26-0-160-103]: Local number of parameters: 168M (320.05MiB)
+12/15/2024 19:34:55 [INFO|DP=0|PP=1|TP=2|ip-26-0-165-213]: Local number of parameters: 83.9M (160.04MiB)
+12/15/2024 19:34:55 [INFO|DP=0|PP=3|TP=2|ip-26-0-172-252]: Local number of parameters: 67.1M (128.00MiB)
+12/15/2024 19:34:55 [INFO|DP=0|PP=2|TP=2|ip-26-0-169-132]: [After model building] Memory usage: 160.05MiB. Peak allocated: 5408.00MiB Peak reserved: 22818.00MiB
+12/15/2024 19:34:55 [INFO|DP=0|PP=1|TP=2|ip-26-0-165-213]: [After model building] Memory usage: 160.05MiB. Peak allocated: 5408.00MiB Peak reserved: 22818.00MiB
+12/15/2024 19:34:55 [INFO|DP=0|PP=3|TP=2|ip-26-0-172-252]: [After model building] Memory usage: 128.01MiB. Peak allocated: 5408.00MiB Peak reserved: 22818.00MiB
+12/15/2024 19:34:55 [INFO|DP=0|PP=0|TP=1|ip-26-0-160-103]: [After model building] Memory usage: 320.06MiB. Peak allocated: 5408.00MiB Peak reserved: 22818.00MiB
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+12/15/2024 19:34:55 [INFO|DP=0|PP=1|TP=3|ip-26-0-165-213]: Local number of parameters: 83.9M (160.04MiB)
+12/15/2024 19:34:55 [INFO|DP=0|PP=3|TP=3|ip-26-0-172-252]: Local number of parameters: 67.1M (128.00MiB)
+12/15/2024 19:34:55 [INFO|DP=0|PP=2|TP=3|ip-26-0-169-132]: Local number of parameters: 83.9M (160.04MiB)
+12/15/2024 19:34:55 [INFO|DP=0|PP=3|TP=3|ip-26-0-172-252]: [After model building] Memory usage: 128.01MiB. Peak allocated: 5408.00MiB Peak reserved: 22818.00MiB
+12/15/2024 19:34:55 [INFO|DP=0|PP=1|TP=3|ip-26-0-165-213]: [After model building] Memory usage: 160.05MiB. Peak allocated: 5408.00MiB Peak reserved: 22818.00MiB
+12/15/2024 19:34:55 [INFO|DP=0|PP=2|TP=3|ip-26-0-169-132]: [After model building] Memory usage: 160.05MiB. Peak allocated: 5408.00MiB Peak reserved: 22818.00MiB
+12/15/2024 19:34:55 [INFO|DP=0|PP=2|TP=1|ip-26-0-169-132]: Local number of parameters: 83.9M (160.04MiB)
+12/15/2024 19:34:55 [INFO|DP=0|PP=3|TP=0|ip-26-0-172-252]: Local number of parameters: 67.1M (128.00MiB)
+12/15/2024 19:34:55 [INFO|DP=0|PP=3|TP=1|ip-26-0-172-252]: Local number of parameters: 67.1M (128.00MiB)
+12/15/2024 19:34:55 [INFO|DP=0|PP=1|TP=1|ip-26-0-165-213]: Local number of parameters: 83.9M (160.04MiB)
+12/15/2024 19:34:55 [INFO|DP=0|PP=2|TP=1|ip-26-0-169-132]: [After model building] Memory usage: 160.05MiB. Peak allocated: 5408.00MiB Peak reserved: 22818.00MiB
+12/15/2024 19:34:55 [INFO|DP=0|PP=3|TP=0|ip-26-0-172-252]: [After model building] Memory usage: 128.01MiB. Peak allocated: 5408.00MiB Peak reserved: 22818.00MiB
+NCCL version 2.18.5+cuda12.2
+12/15/2024 19:34:55 [INFO|DP=0|PP=1|TP=1|ip-26-0-165-213]: [After model building] Memory usage: 160.05MiB. Peak allocated: 5408.00MiB Peak reserved: 22818.00MiB
+12/15/2024 19:34:55 [INFO|DP=0|PP=3|TP=1|ip-26-0-172-252]: [After model building] Memory usage: 128.01MiB. Peak allocated: 5408.00MiB Peak reserved: 22818.00MiB
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+12/15/2024 19:34:58 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]: No checkpoint path provided.
+12/15/2024 19:34:58 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]: Parametrizing model parameters using StandardParametrizator
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+12/15/2024 19:34:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]: [Optimizer Building] Using LearningRateForSP as learning rate
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+12/15/2024 19:35:00 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]: [Training Plan] Stage Stable Training Stage has 99 remaining training steps and has consumed 0 samples
+12/15/2024 19:35:00 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]: Using dummy data generator
+12/15/2024 19:35:00 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]: [Training Plan] There are 1 training stages 
+12/15/2024 19:35:00 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]: [Stage Stable Training Stage] start from step 1 
+12/15/2024 19:35:00 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]: 
+12/15/2024 19:35:00 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]: [Start training] datetime: 2024-12-15 19:35:00.218789 | mbs: 1 | grad_accum: 64 | global_batch_size: 256 | sequence_length: 4096 | train_steps: 100 | start_iteration_step: 0 | consumed_train_samples: 0
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+12/15/2024 19:35:00 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]: Resuming training from stage Stable Training Stage, it has trained for 0 samples and has 99 remaining train steps
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+12/15/2024 19:35:00 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-103]:  Memory usage: 1920.30MiB. Peak allocated 5408.00MiB. Peak reserved: 22818.00MiB
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+NCCL version 2.18.5+cuda12.2
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+NCCL version 2.18.5+cuda12.2
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+NCCL version 2.18.5+cuda12.2
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+NCCL version 2.18.5+cuda12.2
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+NCCL version 2.18.5+cuda12.2
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+NCCL version 2.18.5+cuda12.2
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+NCCL version 2.18.5+cuda12.2
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+slurmstepd: error: *** JOB 13460155 ON ip-26-0-160-103 CANCELLED AT 2024-12-15T19:44:15 DUE TO TIME LIMIT ***
+[2024-12-15 19:44:15,873] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-15 19:44:15,872] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-15 19:44:15,873] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 194933 closing signal SIGTERM
+[2024-12-15 19:44:15,873] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 194934 closing signal SIGTERM
+[2024-12-15 19:44:15,874] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-15 19:44:15,874] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-15 19:44:15,872] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 140961 closing signal SIGTERM
+slurmstepd: error: *** STEP 13460155.0 ON ip-26-0-160-103 CANCELLED AT 2024-12-15T19:44:15 DUE TO TIME LIMIT ***
+[2024-12-15 19:44:15,874] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 194935 closing signal SIGTERM
+[2024-12-15 19:44:15,874] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 163444 closing signal SIGTERM
+[2024-12-15 19:44:15,872] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 140962 closing signal SIGTERM
+[2024-12-15 19:44:15,874] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 36986 closing signal SIGTERM
+[2024-12-15 19:44:15,874] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 163445 closing signal SIGTERM
+[2024-12-15 19:44:15,872] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 140963 closing signal SIGTERM
+[2024-12-15 19:44:15,874] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 194936 closing signal SIGTERM
+[2024-12-15 19:44:15,874] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 36987 closing signal SIGTERM
+[2024-12-15 19:44:15,874] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 36988 closing signal SIGTERM
+[2024-12-15 19:44:15,874] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 36989 closing signal SIGTERM
+[2024-12-15 19:44:15,874] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-15 19:44:15,874] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 56226 closing signal SIGTERM
+[2024-12-15 19:44:15,873] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-15 19:44:15,874] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 56227 closing signal SIGTERM
+[2024-12-15 19:44:15,874] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 58805 closing signal SIGTERM
+[2024-12-15 19:44:15,874] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 56228 closing signal SIGTERM
+[2024-12-15 19:44:15,874] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 58806 closing signal SIGTERM
+[2024-12-15 19:44:15,872] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 140964 closing signal SIGTERM
+[2024-12-15 19:44:15,874] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 194937 closing signal SIGTERM
+[2024-12-15 19:44:15,874] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 58807 closing signal SIGTERM
+[2024-12-15 19:44:15,874] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 58808 closing signal SIGTERM
+[2024-12-15 19:44:15,875] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-15 19:44:15,874] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 56229 closing signal SIGTERM
+srun: Job step aborted: Waiting up to 32 seconds for job step to finish.
+[2024-12-15 19:44:15,875] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 173097 closing signal SIGTERM
+[2024-12-15 19:44:15,875] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 173098 closing signal SIGTERM
+[2024-12-15 19:44:15,875] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 173099 closing signal SIGTERM
+[2024-12-15 19:44:15,875] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 173100 closing signal SIGTERM
+[2024-12-15 19:44:15,874] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 58809 closing signal SIGTERM
+[2024-12-15 19:44:15,874] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 163446 closing signal SIGTERM
+[2024-12-15 19:44:15,874] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 163447 closing signal SIGTERM
+[2024-12-15 19:44:15,873] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-15 19:44:15,873] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 408477 closing signal SIGTERM
+[2024-12-15 19:44:15,873] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 408478 closing signal SIGTERM
+[2024-12-15 19:44:15,873] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 408479 closing signal SIGTERM
+[2024-12-15 19:44:15,873] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 408480 closing signal SIGTERM
+[2024-12-15 19:44:15,875] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 173101 closing signal SIGTERM
+[2024-12-15 19:44:15,875] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 58810 closing signal SIGTERM
+[2024-12-15 19:44:15,873] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 408481 closing signal SIGTERM
+[2024-12-15 19:44:15,874] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 140965 closing signal SIGTERM
+[2024-12-15 19:44:15,874] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 36990 closing signal SIGTERM
+[2024-12-15 19:44:15,874] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 140966 closing signal SIGTERM
+[2024-12-15 19:44:15,874] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 140967 closing signal SIGTERM
+[2024-12-15 19:44:15,875] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 194938 closing signal SIGTERM
+[2024-12-15 19:44:15,876] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 56230 closing signal SIGTERM
+[2024-12-15 19:44:15,874] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 408482 closing signal SIGTERM
+[2024-12-15 19:44:15,877] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 56231 closing signal SIGTERM
+[2024-12-15 19:44:15,877] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 56232 closing signal SIGTERM
+[2024-12-15 19:44:15,877] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 56233 closing signal SIGTERM
+[2024-12-15 19:44:15,876] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 173102 closing signal SIGTERM
+[2024-12-15 19:44:15,877] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 173103 closing signal SIGTERM
+[2024-12-15 19:44:15,877] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 173104 closing signal SIGTERM
+[2024-12-15 19:44:15,877] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 36991 closing signal SIGTERM
+[2024-12-15 19:44:15,876] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 408483 closing signal SIGTERM
+[2024-12-15 19:44:15,878] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 36992 closing signal SIGTERM
+[2024-12-15 19:44:15,878] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 36993 closing signal SIGTERM
+[2024-12-15 19:44:15,877] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 194939 closing signal SIGTERM
+[2024-12-15 19:44:15,875] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 163448 closing signal SIGTERM
+[2024-12-15 19:44:15,875] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 140968 closing signal SIGTERM
+[2024-12-15 19:44:15,876] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 58811 closing signal SIGTERM
+[2024-12-15 19:44:15,876] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 408484 closing signal SIGTERM
+[2024-12-15 19:44:15,877] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 194940 closing signal SIGTERM
+[2024-12-15 19:44:15,877] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 163449 closing signal SIGTERM
+[2024-12-15 19:44:15,876] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 58812 closing signal SIGTERM
+[2024-12-15 19:44:15,877] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 163450 closing signal SIGTERM
+[2024-12-15 19:44:15,877] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 163451 closing signal SIGTERM
+[2024-12-15 19:44:15,878] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 140961 closing signal SIGTERM
+[2024-12-15 19:44:15,878] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 140962 closing signal SIGTERM
+[2024-12-15 19:44:15,878] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 140963 closing signal SIGTERM
+[2024-12-15 19:44:15,878] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 140964 closing signal SIGTERM
+[2024-12-15 19:44:15,878] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 140965 closing signal SIGTERM
+[2024-12-15 19:44:15,878] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 140966 closing signal SIGTERM
+[2024-12-15 19:44:15,878] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 140967 closing signal SIGTERM
+[2024-12-15 19:44:15,879] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 140968 closing signal SIGTERM
+[2024-12-15 19:44:15,923] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 173097 closing signal SIGTERM
+[2024-12-15 19:44:15,923] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 173098 closing signal SIGTERM
+[2024-12-15 19:44:15,923] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 173099 closing signal SIGTERM
+[2024-12-15 19:44:15,923] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 173100 closing signal SIGTERM
+[2024-12-15 19:44:15,923] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 173101 closing signal SIGTERM
+[2024-12-15 19:44:15,923] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 173102 closing signal SIGTERM
+[2024-12-15 19:44:15,923] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 173103 closing signal SIGTERM
+[2024-12-15 19:44:15,923] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 173104 closing signal SIGTERM
+[2024-12-15 19:44:16,005] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 58805 closing signal SIGTERM
+[2024-12-15 19:44:16,005] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 58806 closing signal SIGTERM
+[2024-12-15 19:44:16,005] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 58807 closing signal SIGTERM
+[2024-12-15 19:44:16,005] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 58808 closing signal SIGTERM
+[2024-12-15 19:44:16,005] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 58809 closing signal SIGTERM
+[2024-12-15 19:44:16,005] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 58810 closing signal SIGTERM
+[2024-12-15 19:44:16,005] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 58811 closing signal SIGTERM
+[2024-12-15 19:44:16,005] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 58812 closing signal SIGTERM
+[2024-12-15 19:44:16,033] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 36986 closing signal SIGTERM
+[2024-12-15 19:44:16,033] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 36987 closing signal SIGTERM
+[2024-12-15 19:44:16,033] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 36988 closing signal SIGTERM
+[2024-12-15 19:44:16,033] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 36989 closing signal SIGTERM
+[2024-12-15 19:44:16,033] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 36990 closing signal SIGTERM
+[2024-12-15 19:44:16,033] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 36991 closing signal SIGTERM
+[2024-12-15 19:44:16,033] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 36992 closing signal SIGTERM
+[2024-12-15 19:44:16,034] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 36993 closing signal SIGTERM
+[2024-12-15 19:44:16,087] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 56226 closing signal SIGTERM
+[2024-12-15 19:44:16,088] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 56227 closing signal SIGTERM
+[2024-12-15 19:44:16,088] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 56228 closing signal SIGTERM
+[2024-12-15 19:44:16,088] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 56229 closing signal SIGTERM
+[2024-12-15 19:44:16,088] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 56230 closing signal SIGTERM
+[2024-12-15 19:44:16,088] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 56231 closing signal SIGTERM
+[2024-12-15 19:44:16,088] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 56232 closing signal SIGTERM
+[2024-12-15 19:44:16,088] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 56233 closing signal SIGTERM
+[2024-12-15 19:44:16,096] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 408477 closing signal SIGTERM
+[2024-12-15 19:44:16,096] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 408478 closing signal SIGTERM
+[2024-12-15 19:44:16,096] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 408479 closing signal SIGTERM
+[2024-12-15 19:44:16,096] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 408480 closing signal SIGTERM
+[2024-12-15 19:44:16,096] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 408481 closing signal SIGTERM
+[2024-12-15 19:44:16,096] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 408482 closing signal SIGTERM
+[2024-12-15 19:44:16,096] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 408483 closing signal SIGTERM
+[2024-12-15 19:44:16,096] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 408484 closing signal SIGTERM
+[2024-12-15 19:44:16,099] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 194933 closing signal SIGTERM
+[2024-12-15 19:44:16,099] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 194934 closing signal SIGTERM
+[2024-12-15 19:44:16,099] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 194935 closing signal SIGTERM
+[2024-12-15 19:44:16,099] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 194936 closing signal SIGTERM
+[2024-12-15 19:44:16,099] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 194937 closing signal SIGTERM
+[2024-12-15 19:44:16,099] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 194938 closing signal SIGTERM
+[2024-12-15 19:44:16,099] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 194939 closing signal SIGTERM
+[2024-12-15 19:44:16,099] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 194940 closing signal SIGTERM
diff --git a/logs/13502020-bench_8.86G_dp128_tp2_pp1_acc1_mbs2_seq4096_zero1_tpmodeRED_vocab131k.out b/logs/13502020-bench_8.86G_dp128_tp2_pp1_acc1_mbs2_seq4096_zero1_tpmodeRED_vocab131k.out
new file mode 100644
index 0000000000000000000000000000000000000000..747703b1785947758f0d4d0d7afdb1ae5ec323c3
--- /dev/null
+++ b/logs/13502020-bench_8.86G_dp128_tp2_pp1_acc1_mbs2_seq4096_zero1_tpmodeRED_vocab131k.out
@@ -0,0 +1,1437 @@
++ '[' -z 13502020 ']'
++ source /etc/profile.d/modules.sh
+++ . /usr/share/modules/init/bash
++++ unset _mlshdbg
++++ '[' 0 = 1 ']'
++++ unset _mlre _mlIFS
++++ '[' -n x ']'
++++ _mlIFS=' 	
+'
++++ IFS=' '
++++ '[' -n '' ']'
+++++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash autoinit
++++ _mlcode='module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+MODULEPATH_modshare=/etc/environment-modules/modules:1:/usr/share/modules/\$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1; export MODULEPATH_modshare;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+LOADEDMODULES=; export LOADEDMODULES;
+MODULEPATH=/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/\$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles; export MODULEPATH;
+test 0;'
++++ _mlret=0
++++ '[' -n x ']'
++++ IFS=' 	
+'
++++ unset _mlIFS
++++ unset _mlre _mlv _mlrv
++++ '[' 0 -eq 0 ']'
++++ eval 'module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+MODULEPATH_modshare=/etc/environment-modules/modules:1:/usr/share/modules/\$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1; export MODULEPATH_modshare;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+LOADEDMODULES=; export LOADEDMODULES;
+MODULEPATH=/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/\$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles; export MODULEPATH;
+test 0;'
+++++ MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl
+++++ export MODULES_CMD
+++++ ENV=/usr/share/modules/init/profile.sh
+++++ export ENV
+++++ MODULEPATH_modshare='/etc/environment-modules/modules:1:/usr/share/modules/$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1'
+++++ export MODULEPATH_modshare
+++++ BASH_ENV=/usr/share/modules/init/bash
+++++ export BASH_ENV
+++++ MODULESHOME=/usr/share/modules
+++++ export MODULESHOME
+++++ LOADEDMODULES=
+++++ export LOADEDMODULES
+++++ MODULEPATH='/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles'
+++++ export MODULEPATH
+++++ test 0
++++ '[' 0 = 1 ']'
++++ '[' -t 2 ']'
++++ export -f module
++++ export -f switchml
++++ '[' 5 -ge 3 ']'
++++ [[ ehxB =~ i ]]
++++ [[ ! :/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin: =~ :/usr/bin: ]]
++++ '[' '!' -n '' ']'
++++ MANPATH=:
++++ export MANPATH
+++++ manpath
++++ [[ ! :/admin/home/nouamane/.local/share/man:/fsx/nouamane/miniconda/envs/2-1-cu121/man:/fsx/nouamane/miniconda/envs/2-1-cu121/share/man:/fsx/nouamane/miniconda/man:/fsx/nouamane/miniconda/share/man:/opt/slurm/share/man:/opt/amazon/openmpi/share/man:/opt/amazon/efa/share/man:/usr/local/man:/usr/local/share/man:/usr/share/man:/admin/home/nouamane/.fzf/man:: =~ :/usr/share/man: ]]
++++ unset _mlcode _mlret
++++ '[' -n '' ']'
++ module load cuda/12.1
++ unset _mlshdbg
++ '[' 0 = 1 ']'
++ unset _mlre _mlIFS
++ '[' -n x ']'
++ _mlIFS=' 	
+'
++ IFS=' '
++ '[' -n '' ']'
+++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash load cuda/12.1
++ eval 'CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include;' export 'CPATH;
+LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:;' export 'LD_LIBRARY_PATH;
+FI_EFA_FORK_SAFE=1;' export 'FI_EFA_FORK_SAFE;
+MANPATH=/usr/local/cuda-12.1/share/man::;' export 'MANPATH;
+LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64;' export 'LIBRARY_PATH;
+_LMFILES_=/usr/share/modules/modulefiles/cuda/12.1;' export '_LMFILES_;
+LOADEDMODULES=cuda/12.1;' export 'LOADEDMODULES;
+MPI_PATH=/opt/amazon/openmpi;' export 'MPI_PATH;
+NCCL_HOME_modshare=/opt/nccl/build:1;' export 'NCCL_HOME_modshare;
+NCCL_PROTO=simple;' export 'NCCL_PROTO;
+MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1;' export 'MANPATH_modshare;
+LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1;' export 'LIBRARY_PATH_modshare;
+NCCL_SOCKET_IFNAME=enp;' export 'NCCL_SOCKET_IFNAME;
+AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl;' export 'AWS_OFI_NCCL_HOME;
+NCCL_HOME=/opt/nccl/build;' export 'NCCL_HOME;
+FI_PROVIDER=efa;' export 'FI_PROVIDER;
+AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1;' export 'AWS_OFI_NCCL_HOME_modshare;
+CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1;' export 'CPATH_modshare;
+LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2::1:/opt/nccl/build/lib:1:/opt/aws-ofi-nccl/lib:2:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/efa/lib:1;' export 'LD_LIBRARY_PATH_modshare;
+FI_EFA_ENABLE_SHM_TRANSFER=1;' export 'FI_EFA_ENABLE_SHM_TRANSFER;
+_LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1;' export '_LMFILES__modshare;
+LOADEDMODULES_modshare=cuda/12.1:1;' export 'LOADEDMODULES_modshare;
+MPI_PATH_modshare=/opt/amazon/openmpi:1;' export 'MPI_PATH_modshare;
+PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin;' export 'PATH;
+CUDA_HOME=/usr/local/cuda-12.1;' export 'CUDA_HOME;
+PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/usr/local/cuda-12.1/include:1:/opt/amazon/efa/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/usr/sbin:1:/opt/slurm/bin:1:/fsx/nouamane/miniconda/bin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:2:/opt/amazon/openmpi/bin:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1;' export 'PATH_modshare;
+test' '0;'
+++ CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include
+++ export CPATH
+++ LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:
+++ export LD_LIBRARY_PATH
+++ FI_EFA_FORK_SAFE=1
+++ export FI_EFA_FORK_SAFE
+++ MANPATH=/usr/local/cuda-12.1/share/man::
+++ export MANPATH
+++ LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64
+++ export LIBRARY_PATH
+++ _LMFILES_=/usr/share/modules/modulefiles/cuda/12.1
+++ export _LMFILES_
+++ LOADEDMODULES=cuda/12.1
+++ export LOADEDMODULES
+++ MPI_PATH=/opt/amazon/openmpi
+++ export MPI_PATH
+++ NCCL_HOME_modshare=/opt/nccl/build:1
+++ export NCCL_HOME_modshare
+++ NCCL_PROTO=simple
+++ export NCCL_PROTO
+++ MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1
+++ export MANPATH_modshare
+++ LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1
+++ export LIBRARY_PATH_modshare
+++ NCCL_SOCKET_IFNAME=enp
+++ export NCCL_SOCKET_IFNAME
+++ AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl
+++ export AWS_OFI_NCCL_HOME
+++ NCCL_HOME=/opt/nccl/build
+++ export NCCL_HOME
+++ FI_PROVIDER=efa
+++ export FI_PROVIDER
+++ AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1
+++ export AWS_OFI_NCCL_HOME_modshare
+++ CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1
+++ export CPATH_modshare
+++ LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2::1:/opt/nccl/build/lib:1:/opt/aws-ofi-nccl/lib:2:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/efa/lib:1
+++ export LD_LIBRARY_PATH_modshare
+++ FI_EFA_ENABLE_SHM_TRANSFER=1
+++ export FI_EFA_ENABLE_SHM_TRANSFER
+++ _LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1
+++ export _LMFILES__modshare
+++ LOADEDMODULES_modshare=cuda/12.1:1
+++ export LOADEDMODULES_modshare
+++ MPI_PATH_modshare=/opt/amazon/openmpi:1
+++ export MPI_PATH_modshare
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ export PATH
+++ CUDA_HOME=/usr/local/cuda-12.1
+++ export CUDA_HOME
+++ PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/usr/local/cuda-12.1/include:1:/opt/amazon/efa/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/usr/sbin:1:/opt/slurm/bin:1:/fsx/nouamane/miniconda/bin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:2:/opt/amazon/openmpi/bin:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1
+++ export PATH_modshare
+++ test 0
++ _mlstatus=0
++ '[' -n x ']'
++ IFS=' 	
+'
++ unset _mlre _mlv _mlrv _mlIFS
++ '[' -n '' ']'
++ unset _mlshdbg
++ return 0
++ source /fsx/nouamane/miniconda/bin/activate
+++ _CONDA_ROOT=/fsx/nouamane/miniconda
+++ . /fsx/nouamane/miniconda/etc/profile.d/conda.sh
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ '[' -z x ']'
+++ conda activate
+++ local cmd=activate
+++ case "$cmd" in
+++ __conda_activate activate
+++ '[' -n '' ']'
+++ local ask_conda
++++ PS1=
++++ __conda_exe shell.posix activate
++++ /fsx/nouamane/miniconda/bin/conda shell.posix activate
+Error while loading conda entry point: conda-libmamba-solver (module 'libmambapy' has no attribute 'QueryFormat')
+++ ask_conda='. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_2='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/etc/conda/activate.d/libxml2_activate.sh"'
+++ eval '. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_2='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/etc/conda/activate.d/libxml2_activate.sh"'
++++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh
+++++ test -n ''
+++++ unset XML_CATALOG_FILES
+++++ unset xml_catalog_files_libxml2
++++ PS1='(base) '
++++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++++ export CONDA_PREFIX=/fsx/nouamane/miniconda
++++ CONDA_PREFIX=/fsx/nouamane/miniconda
++++ export CONDA_SHLVL=3
++++ CONDA_SHLVL=3
++++ export CONDA_DEFAULT_ENV=base
++++ CONDA_DEFAULT_ENV=base
++++ export 'CONDA_PROMPT_MODIFIER=(base) '
++++ CONDA_PROMPT_MODIFIER='(base) '
++++ export CONDA_PREFIX_2=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ CONDA_PREFIX_2=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ . /fsx/nouamane/miniconda/etc/conda/activate.d/libxml2_activate.sh
+++++ test -n ''
+++++ xml_catalog_files_libxml2=
+++++ XML_CATALOG_FILES=
+++++ conda_catalog_files=
+++++ ifs_libxml2=' 	
+'
+++++ IFS=' '
+++++ rem=/fsx/nouamane/miniconda
+++++ for pre in ${rem}
+++++ test '' = /fsx/nouamane/miniconda
+++++ conda_catalog_files=/fsx/nouamane/miniconda
+++++ rem=
+++++ IFS=' 	
+'
+++++ conda_catalog_files='file:///fsx/nouamane/miniconda/etc/xml/catalog file:///etc/xml/catalog'
+++++ export 'XML_CATALOG_FILES=file:///fsx/nouamane/miniconda/etc/xml/catalog file:///etc/xml/catalog'
+++++ XML_CATALOG_FILES='file:///fsx/nouamane/miniconda/etc/xml/catalog file:///etc/xml/catalog'
+++++ unset conda_catalog_files ifs_libxml2 rem
+++ __conda_hashr
+++ '[' -n '' ']'
+++ '[' -n '' ']'
+++ hash -r
++ conda activate 2-1-cu121
++ local cmd=activate
++ case "$cmd" in
++ __conda_activate activate 2-1-cu121
++ '[' -n '' ']'
++ local ask_conda
+++ PS1='(base) '
+++ __conda_exe shell.posix activate 2-1-cu121
+++ /fsx/nouamane/miniconda/bin/conda shell.posix activate 2-1-cu121
+Error while loading conda entry point: conda-libmamba-solver (module 'libmambapy' has no attribute 'QueryFormat')
++ ask_conda='. "/fsx/nouamane/miniconda/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''4'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_3='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
++ eval '. "/fsx/nouamane/miniconda/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''4'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_3='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
+++ . /fsx/nouamane/miniconda/etc/conda/deactivate.d/libxml2_deactivate.sh
++++ test -n ''
++++ unset XML_CATALOG_FILES
++++ unset xml_catalog_files_libxml2
+++ PS1='(2-1-cu121) '
+++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ export CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ export CONDA_SHLVL=4
+++ CONDA_SHLVL=4
+++ export CONDA_DEFAULT_ENV=2-1-cu121
+++ CONDA_DEFAULT_ENV=2-1-cu121
+++ export 'CONDA_PROMPT_MODIFIER=(2-1-cu121) '
+++ CONDA_PROMPT_MODIFIER='(2-1-cu121) '
+++ export CONDA_PREFIX_3=/fsx/nouamane/miniconda
+++ CONDA_PREFIX_3=/fsx/nouamane/miniconda
+++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ export _CE_M=
+++ _CE_M=
+++ export _CE_CONDA=
+++ _CE_CONDA=
+++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh
++++ test -n ''
++++ xml_catalog_files_libxml2=
++++ XML_CATALOG_FILES=
++++ conda_catalog_files=
++++ ifs_libxml2=' 	
+'
++++ IFS=' '
++++ rem=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ for pre in ${rem}
++++ test '' = /fsx/nouamane/miniconda/envs/2-1-cu121
++++ conda_catalog_files=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ rem=
++++ IFS=' 	
+'
++++ conda_catalog_files='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ export 'XML_CATALOG_FILES=file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ XML_CATALOG_FILES='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ unset conda_catalog_files ifs_libxml2 rem
++ __conda_hashr
++ '[' -n '' ']'
++ '[' -n '' ']'
++ hash -r
++ export PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++ PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++ '[' -z '' ']'
+++ scontrol show hostnames 'ip-26-0-161-[138,142,153,178,221],ip-26-0-166-[15,36,125,214,244],ip-26-0-167-[9,51,175],ip-26-0-168-238,ip-26-0-169-[86,132,139,207,239,247],ip-26-0-173-[7,121,202,246],ip-26-0-174-[36,100,186,196],ip-26-0-175-[34,132,165,170]'
++ export 'NODELIST=ip-26-0-161-138
+ip-26-0-161-142
+ip-26-0-161-153
+ip-26-0-161-178
+ip-26-0-161-221
+ip-26-0-166-15
+ip-26-0-166-36
+ip-26-0-166-125
+ip-26-0-166-214
+ip-26-0-166-244
+ip-26-0-167-9
+ip-26-0-167-51
+ip-26-0-167-175
+ip-26-0-168-238
+ip-26-0-169-86
+ip-26-0-169-132
+ip-26-0-169-139
+ip-26-0-169-207
+ip-26-0-169-239
+ip-26-0-169-247
+ip-26-0-173-7
+ip-26-0-173-121
+ip-26-0-173-202
+ip-26-0-173-246
+ip-26-0-174-36
+ip-26-0-174-100
+ip-26-0-174-186
+ip-26-0-174-196
+ip-26-0-175-34
+ip-26-0-175-132
+ip-26-0-175-165
+ip-26-0-175-170'
++ NODELIST='ip-26-0-161-138
+ip-26-0-161-142
+ip-26-0-161-153
+ip-26-0-161-178
+ip-26-0-161-221
+ip-26-0-166-15
+ip-26-0-166-36
+ip-26-0-166-125
+ip-26-0-166-214
+ip-26-0-166-244
+ip-26-0-167-9
+ip-26-0-167-51
+ip-26-0-167-175
+ip-26-0-168-238
+ip-26-0-169-86
+ip-26-0-169-132
+ip-26-0-169-139
+ip-26-0-169-207
+ip-26-0-169-239
+ip-26-0-169-247
+ip-26-0-173-7
+ip-26-0-173-121
+ip-26-0-173-202
+ip-26-0-173-246
+ip-26-0-174-36
+ip-26-0-174-100
+ip-26-0-174-186
+ip-26-0-174-196
+ip-26-0-175-34
+ip-26-0-175-132
+ip-26-0-175-165
+ip-26-0-175-170'
+++ head -n1
+++ scontrol show hostnames 'ip-26-0-161-[138,142,153,178,221],ip-26-0-166-[15,36,125,214,244],ip-26-0-167-[9,51,175],ip-26-0-168-238,ip-26-0-169-[86,132,139,207,239,247],ip-26-0-173-[7,121,202,246],ip-26-0-174-[36,100,186,196],ip-26-0-175-[34,132,165,170]'
++ export MASTER_NODE=ip-26-0-161-138
++ MASTER_NODE=ip-26-0-161-138
++ export MASTER_PORT=12356
++ MASTER_PORT=12356
++ export NNODES=32
++ NNODES=32
++ export GPUS_PER_NODE=8
++ GPUS_PER_NODE=8
++ export WORLD_SIZE=256
++ WORLD_SIZE=256
++ export CUDA_DEVICE_MAX_CONNECTIONS=1
++ CUDA_DEVICE_MAX_CONNECTIONS=1
++ export NCCL_DEBUG=WARN
++ NCCL_DEBUG=WARN
++ export NANOTRON_BENCHMARK=1
++ NANOTRON_BENCHMARK=1
++ export WANDB_MODE=disabled
++ WANDB_MODE=disabled
++ export TORCH_NCCL_ASYNC_ERROR_HANDLING=1
++ TORCH_NCCL_ASYNC_ERROR_HANDLING=1
++ '[' -z '' ']'
++ echo '=== GPU Topology ==='
+=== GPU Topology ===
++ nvidia-smi topo -m
+	GPU0	GPU1	GPU2	GPU3	GPU4	GPU5	GPU6	GPU7	CPU Affinity	NUMA Affinity	GPU NUMA ID
+GPU0	 X 	NV18	NV18	NV18	NV18	NV18	NV18	NV18	0-43	0		N/A
+GPU1	NV18	 X 	NV18	NV18	NV18	NV18	NV18	NV18	0-43	0		N/A
+GPU2	NV18	NV18	 X 	NV18	NV18	NV18	NV18	NV18	0-43	0		N/A
+GPU3	NV18	NV18	NV18	 X 	NV18	NV18	NV18	NV18	0-43	0		N/A
+GPU4	NV18	NV18	NV18	NV18	 X 	NV18	NV18	NV18	52-95	1		N/A
+GPU5	NV18	NV18	NV18	NV18	NV18	 X 	NV18	NV18	52-95	1		N/A
+GPU6	NV18	NV18	NV18	NV18	NV18	NV18	 X 	NV18	52-95	1		N/A
+GPU7	NV18	NV18	NV18	NV18	NV18	NV18	NV18	 X 	52-95	1		N/A
+
+Legend:
+
+  X    = Self
+  SYS  = Connection traversing PCIe as well as the SMP interconnect between NUMA nodes (e.g., QPI/UPI)
+  NODE = Connection traversing PCIe as well as the interconnect between PCIe Host Bridges within a NUMA node
+  PHB  = Connection traversing PCIe as well as a PCIe Host Bridge (typically the CPU)
+  PXB  = Connection traversing multiple PCIe bridges (without traversing the PCIe Host Bridge)
+  PIX  = Connection traversing at most a single PCIe bridge
+  NV#  = Connection traversing a bonded set of # NVLinks
++ echo ==================
+==================
++ export SRUN_ALLOC_ARGS=
++ SRUN_ALLOC_ARGS=
++ echo 'Master node: ip-26-0-161-138'
+Master node: ip-26-0-161-138
++ echo 'All nodes: ip-26-0-161-138
+ip-26-0-161-142
+ip-26-0-161-153
+ip-26-0-161-178
+ip-26-0-161-221
+ip-26-0-166-15
+ip-26-0-166-36
+ip-26-0-166-125
+ip-26-0-166-214
+ip-26-0-166-244
+ip-26-0-167-9
+ip-26-0-167-51
+ip-26-0-167-175
+ip-26-0-168-238
+ip-26-0-169-86
+ip-26-0-169-132
+ip-26-0-169-139
+ip-26-0-169-207
+ip-26-0-169-239
+ip-26-0-169-247
+ip-26-0-173-7
+ip-26-0-173-121
+ip-26-0-173-202
+ip-26-0-173-246
+ip-26-0-174-36
+ip-26-0-174-100
+ip-26-0-174-186
+ip-26-0-174-196
+ip-26-0-175-34
+ip-26-0-175-132
+ip-26-0-175-165
+ip-26-0-175-170'
+All nodes: ip-26-0-161-138
+ip-26-0-161-142
+ip-26-0-161-153
+ip-26-0-161-178
+ip-26-0-161-221
+ip-26-0-166-15
+ip-26-0-166-36
+ip-26-0-166-125
+ip-26-0-166-214
+ip-26-0-166-244
+ip-26-0-167-9
+ip-26-0-167-51
+ip-26-0-167-175
+ip-26-0-168-238
+ip-26-0-169-86
+ip-26-0-169-132
+ip-26-0-169-139
+ip-26-0-169-207
+ip-26-0-169-239
+ip-26-0-169-247
+ip-26-0-173-7
+ip-26-0-173-121
+ip-26-0-173-202
+ip-26-0-173-246
+ip-26-0-174-36
+ip-26-0-174-100
+ip-26-0-174-186
+ip-26-0-174-196
+ip-26-0-175-34
+ip-26-0-175-132
+ip-26-0-175-165
+ip-26-0-175-170
++ echo 'World size: 256'
+World size: 256
++ '[' -n '' ']'
++ srun --wait=0 --kill-on-bad-exit=1 torchrun --nnodes=32 --nproc_per_node=8 --rdzv_id=13502020 --rdzv_backend=c10d --rdzv_endpoint=ip-26-0-161-138:12356 --max_restarts 0 --rdzv_conf timeout=60 /fsx/nouamane/projects/nanotron/run_train.py --config-file benchmark/configs/config_8.86G_dp128_tp2_pp1_acc1_mbs2_seq4096_zero1_tpmodeRED_vocab131k.yaml
+[2024-12-19 00:18:56,036] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-19 00:18:56,036] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-19 00:18:56,036] torch.distributed.run: [WARNING] 
+[2024-12-19 00:18:56,036] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 00:18:56,036] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-19 00:18:56,036] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 00:18:56,038] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-19 00:18:56,037] torch.distributed.run: [WARNING] 
+[2024-12-19 00:18:56,037] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 00:18:56,037] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-19 00:18:56,037] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 00:18:56,038] torch.distributed.run: [WARNING] 
+[2024-12-19 00:18:56,038] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 00:18:56,038] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-19 00:18:56,038] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 00:18:56,040] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-19 00:18:56,040] torch.distributed.run: [WARNING] 
+[2024-12-19 00:18:56,040] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 00:18:56,040] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-19 00:18:56,040] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 00:18:56,038] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-19 00:18:56,038] torch.distributed.run: [WARNING] 
+[2024-12-19 00:18:56,038] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 00:18:56,038] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-19 00:18:56,038] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 00:18:56,038] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-19 00:18:56,038] torch.distributed.run: [WARNING] 
+[2024-12-19 00:18:56,038] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 00:18:56,038] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-19 00:18:56,038] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 00:18:56,042] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-19 00:18:56,042] torch.distributed.run: [WARNING] 
+[2024-12-19 00:18:56,042] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 00:18:56,042] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-19 00:18:56,042] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 00:18:56,042] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-19 00:18:56,042] torch.distributed.run: [WARNING] 
+[2024-12-19 00:18:56,042] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 00:18:56,042] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-19 00:18:56,042] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 00:18:56,042] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-19 00:18:56,043] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-19 00:18:56,042] torch.distributed.run: [WARNING] 
+[2024-12-19 00:18:56,042] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 00:18:56,042] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-19 00:18:56,042] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 00:18:56,043] torch.distributed.run: [WARNING] 
+[2024-12-19 00:18:56,043] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 00:18:56,043] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-19 00:18:56,043] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 00:18:56,043] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-19 00:18:56,043] torch.distributed.run: [WARNING] 
+[2024-12-19 00:18:56,043] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 00:18:56,043] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-19 00:18:56,043] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 00:18:56,044] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-19 00:18:56,044] torch.distributed.run: [WARNING] 
+[2024-12-19 00:18:56,044] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 00:18:56,044] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-19 00:18:56,044] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 00:18:56,044] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-19 00:18:56,044] torch.distributed.run: [WARNING] 
+[2024-12-19 00:18:56,044] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 00:18:56,044] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-19 00:18:56,044] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 00:18:56,044] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-19 00:18:56,044] torch.distributed.run: [WARNING] 
+[2024-12-19 00:18:56,044] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 00:18:56,044] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-19 00:18:56,044] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 00:18:56,041] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-19 00:18:56,042] torch.distributed.run: [WARNING] 
+[2024-12-19 00:18:56,042] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 00:18:56,042] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-19 00:18:56,042] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 00:18:56,045] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-19 00:18:56,045] torch.distributed.run: [WARNING] 
+[2024-12-19 00:18:56,045] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 00:18:56,045] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-19 00:18:56,045] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 00:18:56,045] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-19 00:18:56,046] torch.distributed.run: [WARNING] 
+[2024-12-19 00:18:56,046] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 00:18:56,046] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-19 00:18:56,046] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 00:18:56,045] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-19 00:18:56,045] torch.distributed.run: [WARNING] 
+[2024-12-19 00:18:56,045] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 00:18:56,045] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-19 00:18:56,045] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 00:18:56,047] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-19 00:18:56,046] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-19 00:18:56,046] torch.distributed.run: [WARNING] 
+[2024-12-19 00:18:56,046] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 00:18:56,046] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-19 00:18:56,046] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 00:18:56,047] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-19 00:18:56,047] torch.distributed.run: [WARNING] 
+[2024-12-19 00:18:56,047] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 00:18:56,047] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-19 00:18:56,047] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 00:18:56,049] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-19 00:18:56,050] torch.distributed.run: [WARNING] 
+[2024-12-19 00:18:56,050] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 00:18:56,050] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-19 00:18:56,050] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 00:18:56,053] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-19 00:18:56,053] torch.distributed.run: [WARNING] 
+[2024-12-19 00:18:56,053] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 00:18:56,053] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-19 00:18:56,053] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 00:18:56,060] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-19 00:18:56,047] torch.distributed.run: [WARNING] 
+[2024-12-19 00:18:56,047] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 00:18:56,047] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-19 00:18:56,047] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 00:18:56,060] torch.distributed.run: [WARNING] 
+[2024-12-19 00:18:56,060] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 00:18:56,060] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-19 00:18:56,060] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 00:18:59,527] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-19 00:18:59,526] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-19 00:18:59,527] torch.distributed.run: [WARNING] 
+[2024-12-19 00:18:59,527] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 00:18:59,527] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-19 00:18:59,527] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 00:18:59,526] torch.distributed.run: [WARNING] 
+[2024-12-19 00:18:59,526] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 00:18:59,526] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-19 00:18:59,526] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 00:18:59,526] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-19 00:18:59,527] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-19 00:18:59,527] torch.distributed.run: [WARNING] 
+[2024-12-19 00:18:59,527] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 00:18:59,527] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-19 00:18:59,527] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 00:18:59,527] torch.distributed.run: [WARNING] 
+[2024-12-19 00:18:59,527] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 00:18:59,527] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-19 00:18:59,527] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 00:18:59,536] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-19 00:18:59,536] torch.distributed.run: [WARNING] 
+[2024-12-19 00:18:59,536] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 00:18:59,536] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-19 00:18:59,536] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 00:18:59,734] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-19 00:18:59,734] torch.distributed.run: [WARNING] 
+[2024-12-19 00:18:59,734] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 00:18:59,734] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-19 00:18:59,734] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 00:18:59,903] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-19 00:18:59,903] torch.distributed.run: [WARNING] 
+[2024-12-19 00:18:59,903] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 00:18:59,903] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-19 00:18:59,903] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 00:19:00,032] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-19 00:19:00,032] torch.distributed.run: [WARNING] 
+[2024-12-19 00:19:00,032] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 00:19:00,032] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-19 00:19:00,032] torch.distributed.run: [WARNING] *****************************************
+NCCL version 2.18.5+cuda12.2
+12/19/2024 00:19:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]: Measuring inter-GPU and intra-node bandwidth...
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+12/19/2024 00:19:48 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]: Bandwidth measurement complete. Time taken: 18.93 seconds
+12/19/2024 00:19:48 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]: Config:
+12/19/2024 00:19:48 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]: Config(general=GeneralArgs(project='debug',
+12/19/2024 00:19:48 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                            run='8.86G_dp128_tp2_pp1_acc1_mbs2_seq4096_zero1_tpmodeRED_vocab131k',
+12/19/2024 00:19:48 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                            seed=42,
+12/19/2024 00:19:48 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                            step=None,
+12/19/2024 00:19:48 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                            consumed_train_samples=None,
+12/19/2024 00:19:48 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                            benchmark_csv_path=PosixPath('benchmark/results/bench_final2.csv'),
+12/19/2024 00:19:48 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                            ignore_sanity_checks=True),
+12/19/2024 00:19:48 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:        parallelism=ParallelismArgs(dp=128,
+12/19/2024 00:19:48 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                                    pp=1,
+12/19/2024 00:19:48 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                                    tp=2,
+12/19/2024 00:19:48 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                                    pp_engine=<nanotron.parallel.pipeline_parallel.engine.OneForwardOneBackwardPipelineEngine object at 0x7fc5f37b8dc0>,
+12/19/2024 00:19:48 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                                    tp_mode=<TensorParallelLinearMode.REDUCE_SCATTER: 2>,
+12/19/2024 00:19:48 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                                    tp_linear_async_communication=True,
+12/19/2024 00:19:48 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                                    recompute_layer=False,
+12/19/2024 00:19:48 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                                    tp_recompute_allgather=True,
+12/19/2024 00:19:48 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                                    expert_parallel_size=1),
+12/19/2024 00:19:48 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:        model=ModelArgs(model_config=LlamaConfig(bos_token_id=0,
+12/19/2024 00:19:48 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                                                 eos_token_id=0,
+12/19/2024 00:19:48 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                                                 hidden_act='silu',
+12/19/2024 00:19:48 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                                                 hidden_size=4096,
+12/19/2024 00:19:48 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                                                 initializer_range=0.02,
+12/19/2024 00:19:48 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                                                 intermediate_size=14336,
+12/19/2024 00:19:48 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                                                 is_llama_config=True,
+12/19/2024 00:19:48 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                                                 max_position_embeddings=4096,
+12/19/2024 00:19:48 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                                                 num_attention_heads=32,
+12/19/2024 00:19:48 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                                                 num_hidden_layers=32,
+12/19/2024 00:19:48 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                                                 num_key_value_heads=32,
+12/19/2024 00:19:48 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                                                 pad_token_id=None,
+12/19/2024 00:19:48 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                                                 pretraining_tp=1,
+12/19/2024 00:19:48 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                                                 rms_norm_eps=1e-05,
+12/19/2024 00:19:48 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                                                 rope_scaling=None,
+12/19/2024 00:19:48 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                                                 rope_theta=10000.0,
+12/19/2024 00:19:48 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                                                 rope_interleaved=False,
+12/19/2024 00:19:48 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                                                 tie_word_embeddings=False,
+12/19/2024 00:19:48 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                                                 use_cache=True,
+12/19/2024 00:19:48 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                                                 vocab_size=131072),
+12/19/2024 00:19:48 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                        init_method=RandomInit(std=0.02),
+12/19/2024 00:19:48 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                        dtype=torch.bfloat16,
+12/19/2024 00:19:48 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                        make_vocab_size_divisible_by=1,
+12/19/2024 00:19:48 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                        ddp_bucket_cap_mb=25),
+12/19/2024 00:19:48 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:        tokenizer=TokenizerArgs(tokenizer_name_or_path='robot-test/dummy-tokenizer-wordlevel',
+12/19/2024 00:19:48 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                                tokenizer_revision=None,
+12/19/2024 00:19:48 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                                tokenizer_max_length=None),
+12/19/2024 00:19:48 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:        checkpoints=CheckpointsArgs(checkpoints_path=PosixPath('checkpoints'),
+12/19/2024 00:19:48 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                                    checkpoint_interval=10000,
+12/19/2024 00:19:48 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                                    save_initial_state=False,
+12/19/2024 00:19:48 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                                    save_final_state=False,
+12/19/2024 00:19:48 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                                    resume_checkpoint_path=None,
+12/19/2024 00:19:48 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                                    checkpoints_path_is_shared_file_system=False),
+12/19/2024 00:19:48 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:        logging=LoggingArgs(log_level='info',
+12/19/2024 00:19:48 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                            log_level_replica='info',
+12/19/2024 00:19:48 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                            iteration_step_info_interval=1),
+12/19/2024 00:19:48 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:        tokens=TokensArgs(sequence_length=4096,
+12/19/2024 00:19:48 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                          train_steps=100,
+12/19/2024 00:19:48 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                          micro_batch_size=2,
+12/19/2024 00:19:48 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                          batch_accumulation_per_replica=1,
+12/19/2024 00:19:48 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                          val_check_interval=100,
+12/19/2024 00:19:48 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                          limit_val_batches=0,
+12/19/2024 00:19:48 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                          limit_test_batches=0),
+12/19/2024 00:19:48 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:        optimizer=OptimizerArgs(optimizer_factory=AdamWOptimizerArgs(adam_eps=1e-08,
+12/19/2024 00:19:48 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                                                                     adam_beta1=0.9,
+12/19/2024 00:19:48 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                                                                     adam_beta2=0.95,
+12/19/2024 00:19:48 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                                                                     torch_adam_is_fused=True,
+12/19/2024 00:19:48 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                                                                     name='adamW'),
+12/19/2024 00:19:48 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                                zero_stage=1,
+12/19/2024 00:19:48 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                                weight_decay=0.01,
+12/19/2024 00:19:48 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                                clip_grad=1.0,
+12/19/2024 00:19:48 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                                accumulate_grad_in_fp32=True,
+12/19/2024 00:19:48 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                                learning_rate_scheduler=LRSchedulerArgs(learning_rate=0.0003,
+12/19/2024 00:19:48 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                                                                        lr_warmup_steps=2,
+12/19/2024 00:19:48 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                                                                        lr_warmup_style='linear',
+12/19/2024 00:19:48 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                                                                        lr_decay_style='cosine',
+12/19/2024 00:19:48 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                                                                        lr_decay_steps=13,
+12/19/2024 00:19:48 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                                                                        lr_decay_starting_step=None,
+12/19/2024 00:19:48 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                                                                        min_decay_lr=1e-05)),
+12/19/2024 00:19:48 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:        data_stages=[DatasetStageArgs(name='Stable Training Stage',
+12/19/2024 00:19:48 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                                      start_training_step=1,
+12/19/2024 00:19:48 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                                      data=DataArgs(dataset=None,
+12/19/2024 00:19:48 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                                                    seed=42,
+12/19/2024 00:19:48 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:                                                    num_loading_workers=1))],
+12/19/2024 00:19:48 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:        profiler=None,
+12/19/2024 00:19:48 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:        lighteval=None,
+12/19/2024 00:19:48 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:        s3_upload=None)
+12/19/2024 00:19:48 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]: Model Config:
+12/19/2024 00:19:48 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]: LlamaConfig(bos_token_id=0,
+12/19/2024 00:19:48 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:             eos_token_id=0,
+12/19/2024 00:19:48 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:             hidden_act='silu',
+12/19/2024 00:19:48 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:             hidden_size=4096,
+12/19/2024 00:19:48 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:             initializer_range=0.02,
+12/19/2024 00:19:48 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:             intermediate_size=14336,
+12/19/2024 00:19:48 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:             is_llama_config=True,
+12/19/2024 00:19:48 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:             max_position_embeddings=4096,
+12/19/2024 00:19:48 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:             num_attention_heads=32,
+12/19/2024 00:19:48 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:             num_hidden_layers=32,
+12/19/2024 00:19:48 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:             num_key_value_heads=32,
+12/19/2024 00:19:48 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:             pad_token_id=None,
+12/19/2024 00:19:48 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:             pretraining_tp=1,
+12/19/2024 00:19:48 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:             rms_norm_eps=1e-05,
+12/19/2024 00:19:48 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:             rope_scaling=None,
+12/19/2024 00:19:48 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:             rope_theta=10000.0,
+12/19/2024 00:19:48 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:             rope_interleaved=False,
+12/19/2024 00:19:48 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:             tie_word_embeddings=False,
+12/19/2024 00:19:48 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:             use_cache=True,
+12/19/2024 00:19:48 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]:             vocab_size=131072)
+12/19/2024 00:19:48 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]: Building model..
+12/19/2024 00:19:48 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]: Initialize RoPE Theta = 10000.0
+12/19/2024 00:19:48 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]: Setting PP block ranks...
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+12/19/2024 00:19:52 [INFO|DP=0|PP=0|TP=1|ip-26-0-161-138]: Local number of parameters: 4.43G (8448.51MiB)
+12/19/2024 00:19:52 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]: Total number of parameters: 8.86G (16897.02MiB)
+12/19/2024 00:19:52 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]: Local number of parameters: 4.43G (8448.51MiB)
+12/19/2024 00:19:52 [INFO|DP=0|PP=0|TP=1|ip-26-0-161-138]: [After model building] Memory usage: 8448.55MiB. Peak allocated: 8448.62MiB Peak reserved: 17686.00MiB
+12/19/2024 00:19:52 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]: [After model building] Memory usage: 8448.55MiB. Peak allocated: 8448.62MiB Peak reserved: 17686.00MiB
+12/19/2024 00:19:52 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]: No checkpoint path provided.
+12/19/2024 00:19:52 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-138]: Parametrizing model parameters using StandardParametrizator
+slurmstepd: error: *** STEP 13502020.0 ON ip-26-0-161-138 CANCELLED AT 2024-12-19T01:29:12 DUE TO TIME LIMIT ***
+slurmstepd: error: *** JOB 13502020 ON ip-26-0-161-138 CANCELLED AT 2024-12-19T01:29:12 DUE TO TIME LIMIT ***
+srun: Job step aborted: Waiting up to 32 seconds for job step to finish.
+[2024-12-19 01:29:12,721] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-19 01:29:12,723] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-19 01:29:12,727] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 427354 closing signal SIGTERM
+[2024-12-19 01:29:12,723] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-19 01:29:12,728] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 349371 closing signal SIGTERM
+[2024-12-19 01:29:12,722] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-19 01:29:12,727] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 427355 closing signal SIGTERM
+[2024-12-19 01:29:12,729] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 21471 closing signal SIGTERM
+[2024-12-19 01:29:12,728] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 427356 closing signal SIGTERM
+[2024-12-19 01:29:12,729] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 349372 closing signal SIGTERM
+[2024-12-19 01:29:12,728] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 427357 closing signal SIGTERM
+[2024-12-19 01:29:12,728] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 427358 closing signal SIGTERM
+[2024-12-19 01:29:12,728] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 380175 closing signal SIGTERM
+[2024-12-19 01:29:12,729] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 349373 closing signal SIGTERM
+[2024-12-19 01:29:12,728] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 427359 closing signal SIGTERM
+[2024-12-19 01:29:12,729] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 349374 closing signal SIGTERM
+[2024-12-19 01:29:12,728] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 427360 closing signal SIGTERM
+[2024-12-19 01:29:12,728] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 427361 closing signal SIGTERM
+[2024-12-19 01:29:12,723] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-19 01:29:12,729] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 349375 closing signal SIGTERM
+[2024-12-19 01:29:12,729] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 349376 closing signal SIGTERM
+[2024-12-19 01:29:12,729] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 21472 closing signal SIGTERM
+[2024-12-19 01:29:12,729] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 349377 closing signal SIGTERM
+[2024-12-19 01:29:12,729] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 349378 closing signal SIGTERM
+[2024-12-19 01:29:12,729] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 21473 closing signal SIGTERM
+[2024-12-19 01:29:12,723] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-19 01:29:12,728] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 380176 closing signal SIGTERM
+[2024-12-19 01:29:12,729] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 21474 closing signal SIGTERM
+[2024-12-19 01:29:12,729] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 243904 closing signal SIGTERM
+[2024-12-19 01:29:12,728] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 380177 closing signal SIGTERM
+[2024-12-19 01:29:12,725] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-19 01:29:12,729] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 21475 closing signal SIGTERM
+[2024-12-19 01:29:12,729] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 21476 closing signal SIGTERM
+[2024-12-19 01:29:12,728] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 380178 closing signal SIGTERM
+[2024-12-19 01:29:12,724] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-19 01:29:12,723] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-19 01:29:12,729] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 21477 closing signal SIGTERM
+[2024-12-19 01:29:12,728] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 380179 closing signal SIGTERM
+[2024-12-19 01:29:12,729] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 21478 closing signal SIGTERM
+[2024-12-19 01:29:12,724] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-19 01:29:12,730] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 243372 closing signal SIGTERM
+[2024-12-19 01:29:12,728] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 380180 closing signal SIGTERM
+[2024-12-19 01:29:12,728] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 380181 closing signal SIGTERM
+[2024-12-19 01:29:12,728] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 380182 closing signal SIGTERM
+[2024-12-19 01:29:12,729] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 243905 closing signal SIGTERM
+[2024-12-19 01:29:12,729] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 384120 closing signal SIGTERM
+[2024-12-19 01:29:12,723] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-19 01:29:12,729] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 243906 closing signal SIGTERM
+[2024-12-19 01:29:12,729] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 243907 closing signal SIGTERM
+[2024-12-19 01:29:12,731] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 243373 closing signal SIGTERM
+[2024-12-19 01:29:12,730] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 373264 closing signal SIGTERM
+[2024-12-19 01:29:12,729] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 243908 closing signal SIGTERM
+[2024-12-19 01:29:12,729] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 243909 closing signal SIGTERM
+[2024-12-19 01:29:12,731] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 243374 closing signal SIGTERM
+[2024-12-19 01:29:12,729] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 243910 closing signal SIGTERM
+[2024-12-19 01:29:12,729] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 460766 closing signal SIGTERM
+[2024-12-19 01:29:12,723] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-19 01:29:12,731] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 243375 closing signal SIGTERM
+[2024-12-19 01:29:12,729] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 243911 closing signal SIGTERM
+[2024-12-19 01:29:12,725] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-19 01:29:12,723] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-19 01:29:12,724] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-19 01:29:12,730] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 527077 closing signal SIGTERM
+[2024-12-19 01:29:12,729] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 384121 closing signal SIGTERM
+[2024-12-19 01:29:12,731] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 243376 closing signal SIGTERM
+[2024-12-19 01:29:12,731] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 243377 closing signal SIGTERM
+[2024-12-19 01:29:12,729] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 384122 closing signal SIGTERM
+[2024-12-19 01:29:12,731] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 243378 closing signal SIGTERM
+[2024-12-19 01:29:12,731] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 243379 closing signal SIGTERM
+[2024-12-19 01:29:12,728] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 535228 closing signal SIGTERM
+[2024-12-19 01:29:12,729] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 384123 closing signal SIGTERM
+[2024-12-19 01:29:12,729] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 384124 closing signal SIGTERM
+[2024-12-19 01:29:12,730] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 373265 closing signal SIGTERM
+[2024-12-19 01:29:12,724] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-19 01:29:12,723] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-19 01:29:12,729] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 384125 closing signal SIGTERM
+[2024-12-19 01:29:12,729] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 384126 closing signal SIGTERM
+[2024-12-19 01:29:12,731] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 373266 closing signal SIGTERM
+[2024-12-19 01:29:12,729] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 460767 closing signal SIGTERM
+[2024-12-19 01:29:12,730] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 373468 closing signal SIGTERM
+[2024-12-19 01:29:12,731] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 373267 closing signal SIGTERM
+[2024-12-19 01:29:12,729] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 384127 closing signal SIGTERM
+[2024-12-19 01:29:12,730] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 527078 closing signal SIGTERM
+[2024-12-19 01:29:12,729] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 460768 closing signal SIGTERM
+[2024-12-19 01:29:12,731] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 373268 closing signal SIGTERM
+[2024-12-19 01:29:12,731] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 373269 closing signal SIGTERM
+[2024-12-19 01:29:12,730] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 527079 closing signal SIGTERM
+[2024-12-19 01:29:12,724] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-19 01:29:12,731] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 243022 closing signal SIGTERM
+[2024-12-19 01:29:12,729] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 19553 closing signal SIGTERM
+[2024-12-19 01:29:12,730] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 460769 closing signal SIGTERM
+[2024-12-19 01:29:12,728] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 388752 closing signal SIGTERM
+[2024-12-19 01:29:12,730] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 460770 closing signal SIGTERM
+[2024-12-19 01:29:12,729] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 39824 closing signal SIGTERM
+[2024-12-19 01:29:12,731] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 373270 closing signal SIGTERM
+[2024-12-19 01:29:12,731] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 373271 closing signal SIGTERM
+[2024-12-19 01:29:12,730] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 527080 closing signal SIGTERM
+[2024-12-19 01:29:12,730] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 527081 closing signal SIGTERM
+[2024-12-19 01:29:12,730] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 460771 closing signal SIGTERM
+[2024-12-19 01:29:12,730] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 460772 closing signal SIGTERM
+[2024-12-19 01:29:12,729] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 535229 closing signal SIGTERM
+[2024-12-19 01:29:12,723] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-19 01:29:12,730] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 527082 closing signal SIGTERM
+[2024-12-19 01:29:12,723] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-19 01:29:12,729] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 535230 closing signal SIGTERM
+[2024-12-19 01:29:12,730] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 460773 closing signal SIGTERM
+[2024-12-19 01:29:12,730] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 527083 closing signal SIGTERM
+[2024-12-19 01:29:12,730] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 373469 closing signal SIGTERM
+[2024-12-19 01:29:12,721] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-19 01:29:12,729] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 535231 closing signal SIGTERM
+[2024-12-19 01:29:12,724] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-19 01:29:12,729] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 535232 closing signal SIGTERM
+[2024-12-19 01:29:12,724] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-19 01:29:12,730] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 527084 closing signal SIGTERM
+[2024-12-19 01:29:12,729] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 119005 closing signal SIGTERM
+[2024-12-19 01:29:12,731] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 243023 closing signal SIGTERM
+[2024-12-19 01:29:12,730] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 373470 closing signal SIGTERM
+[2024-12-19 01:29:12,729] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 535233 closing signal SIGTERM
+[2024-12-19 01:29:12,729] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 19554 closing signal SIGTERM
+[2024-12-19 01:29:12,729] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 348696 closing signal SIGTERM
+[2024-12-19 01:29:12,729] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 39825 closing signal SIGTERM
+[2024-12-19 01:29:12,723] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-19 01:29:12,730] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 73543 closing signal SIGTERM
+[2024-12-19 01:29:12,731] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 243024 closing signal SIGTERM
+[2024-12-19 01:29:12,724] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-19 01:29:12,730] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 373471 closing signal SIGTERM
+[2024-12-19 01:29:12,730] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 373472 closing signal SIGTERM
+[2024-12-19 01:29:12,729] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 388753 closing signal SIGTERM
+[2024-12-19 01:29:12,729] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 535234 closing signal SIGTERM
+[2024-12-19 01:29:12,729] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 19555 closing signal SIGTERM
+[2024-12-19 01:29:12,729] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 535235 closing signal SIGTERM
+[2024-12-19 01:29:12,729] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 39826 closing signal SIGTERM
+[2024-12-19 01:29:12,731] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 243025 closing signal SIGTERM
+[2024-12-19 01:29:12,731] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 61811 closing signal SIGTERM
+[2024-12-19 01:29:12,730] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 373473 closing signal SIGTERM
+[2024-12-19 01:29:12,729] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 19556 closing signal SIGTERM
+[2024-12-19 01:29:12,729] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 388754 closing signal SIGTERM
+[2024-12-19 01:29:12,731] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 243026 closing signal SIGTERM
+[2024-12-19 01:29:12,729] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 19557 closing signal SIGTERM
+[2024-12-19 01:29:12,731] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 243027 closing signal SIGTERM
+[2024-12-19 01:29:12,730] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 22956 closing signal SIGTERM
+[2024-12-19 01:29:12,729] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 39827 closing signal SIGTERM
+[2024-12-19 01:29:12,722] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-19 01:29:12,729] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 39828 closing signal SIGTERM
+[2024-12-19 01:29:12,729] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 19558 closing signal SIGTERM
+[2024-12-19 01:29:12,727] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 464605 closing signal SIGTERM
+[2024-12-19 01:29:12,729] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 119006 closing signal SIGTERM
+[2024-12-19 01:29:12,730] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 373474 closing signal SIGTERM
+[2024-12-19 01:29:12,729] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 388755 closing signal SIGTERM
+[2024-12-19 01:29:12,731] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 243028 closing signal SIGTERM
+[2024-12-19 01:29:12,729] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 296777 closing signal SIGTERM
+[2024-12-19 01:29:12,730] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 373475 closing signal SIGTERM
+[2024-12-19 01:29:12,729] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 388756 closing signal SIGTERM
+[2024-12-19 01:29:12,729] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 39829 closing signal SIGTERM
+[2024-12-19 01:29:12,729] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 19559 closing signal SIGTERM
+[2024-12-19 01:29:12,729] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 119007 closing signal SIGTERM
+[2024-12-19 01:29:12,729] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 39830 closing signal SIGTERM
+[2024-12-19 01:29:12,729] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 19560 closing signal SIGTERM
+[2024-12-19 01:29:12,729] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 388757 closing signal SIGTERM
+[2024-12-19 01:29:12,723] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-19 01:29:12,731] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 243029 closing signal SIGTERM
+[2024-12-19 01:29:12,729] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 388758 closing signal SIGTERM
+[2024-12-19 01:29:12,729] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 348697 closing signal SIGTERM
+[2024-12-19 01:29:12,730] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 119008 closing signal SIGTERM
+[2024-12-19 01:29:12,730] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 39831 closing signal SIGTERM
+[2024-12-19 01:29:12,730] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 73544 closing signal SIGTERM
+[2024-12-19 01:29:12,730] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 119009 closing signal SIGTERM
+[2024-12-19 01:29:12,729] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 388759 closing signal SIGTERM
+[2024-12-19 01:29:12,729] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 348698 closing signal SIGTERM
+[2024-12-19 01:29:12,730] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 119010 closing signal SIGTERM
+[2024-12-19 01:29:12,730] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 73545 closing signal SIGTERM
+[2024-12-19 01:29:12,731] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 61812 closing signal SIGTERM
+[2024-12-19 01:29:12,730] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 119011 closing signal SIGTERM
+[2024-12-19 01:29:12,730] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 22957 closing signal SIGTERM
+[2024-12-19 01:29:12,730] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 484084 closing signal SIGTERM
+[2024-12-19 01:29:12,729] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 348699 closing signal SIGTERM
+[2024-12-19 01:29:12,729] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 348700 closing signal SIGTERM
+[2024-12-19 01:29:12,727] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 464606 closing signal SIGTERM
+[2024-12-19 01:29:12,730] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 119012 closing signal SIGTERM
+[2024-12-19 01:29:12,730] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 73546 closing signal SIGTERM
+[2024-12-19 01:29:12,731] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 61813 closing signal SIGTERM
+[2024-12-19 01:29:12,729] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 348701 closing signal SIGTERM
+[2024-12-19 01:29:12,730] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 73547 closing signal SIGTERM
+[2024-12-19 01:29:12,730] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 22958 closing signal SIGTERM
+[2024-12-19 01:29:12,729] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 348702 closing signal SIGTERM
+[2024-12-19 01:29:12,727] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 464607 closing signal SIGTERM
+[2024-12-19 01:29:12,730] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 73548 closing signal SIGTERM
+[2024-12-19 01:29:12,725] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-19 01:29:12,730] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 73549 closing signal SIGTERM
+[2024-12-19 01:29:12,729] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 296778 closing signal SIGTERM
+[2024-12-19 01:29:12,730] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 22959 closing signal SIGTERM
+[2024-12-19 01:29:12,731] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 61814 closing signal SIGTERM
+[2024-12-19 01:29:12,725] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-19 01:29:12,730] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 22960 closing signal SIGTERM
+[2024-12-19 01:29:12,731] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 61815 closing signal SIGTERM
+[2024-12-19 01:29:12,727] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 464608 closing signal SIGTERM
+[2024-12-19 01:29:12,729] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 348703 closing signal SIGTERM
+[2024-12-19 01:29:12,728] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 392229 closing signal SIGTERM
+[2024-12-19 01:29:12,731] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 61816 closing signal SIGTERM
+[2024-12-19 01:29:12,727] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 464609 closing signal SIGTERM
+[2024-12-19 01:29:12,730] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 73550 closing signal SIGTERM
+[2024-12-19 01:29:12,729] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 392230 closing signal SIGTERM
+[2024-12-19 01:29:12,731] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 61817 closing signal SIGTERM
+[2024-12-19 01:29:12,729] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 296779 closing signal SIGTERM
+[2024-12-19 01:29:12,729] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 392231 closing signal SIGTERM
+[2024-12-19 01:29:12,730] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 22961 closing signal SIGTERM
+[2024-12-19 01:29:12,727] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 464610 closing signal SIGTERM
+[2024-12-19 01:29:12,729] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 296780 closing signal SIGTERM
+[2024-12-19 01:29:12,729] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 392232 closing signal SIGTERM
+[2024-12-19 01:29:12,729] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 392233 closing signal SIGTERM
+[2024-12-19 01:29:12,729] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 392234 closing signal SIGTERM
+[2024-12-19 01:29:12,731] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 61818 closing signal SIGTERM
+[2024-12-19 01:29:12,731] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 386480 closing signal SIGTERM
+[2024-12-19 01:29:12,729] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 296781 closing signal SIGTERM
+[2024-12-19 01:29:12,729] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 392235 closing signal SIGTERM
+[2024-12-19 01:29:12,729] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 349013 closing signal SIGTERM
+[2024-12-19 01:29:12,729] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 296782 closing signal SIGTERM
+[2024-12-19 01:29:12,729] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 392236 closing signal SIGTERM
+[2024-12-19 01:29:12,730] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 22962 closing signal SIGTERM
+[2024-12-19 01:29:12,730] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 484085 closing signal SIGTERM
+[2024-12-19 01:29:12,727] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 464611 closing signal SIGTERM
+[2024-12-19 01:29:12,730] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 22963 closing signal SIGTERM
+[2024-12-19 01:29:12,727] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 464612 closing signal SIGTERM
+[2024-12-19 01:29:12,729] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 206511 closing signal SIGTERM
+[2024-12-19 01:29:12,729] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 296783 closing signal SIGTERM
+[2024-12-19 01:29:12,730] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 576216 closing signal SIGTERM
+[2024-12-19 01:29:12,729] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 296784 closing signal SIGTERM
+[2024-12-19 01:29:12,730] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 484086 closing signal SIGTERM
+[2024-12-19 01:29:12,731] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 484087 closing signal SIGTERM
+[2024-12-19 01:29:12,731] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 484088 closing signal SIGTERM
+[2024-12-19 01:29:12,731] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 484089 closing signal SIGTERM
+[2024-12-19 01:29:12,731] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 484090 closing signal SIGTERM
+[2024-12-19 01:29:12,731] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 386481 closing signal SIGTERM
+[2024-12-19 01:29:12,731] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 484091 closing signal SIGTERM
+[2024-12-19 01:29:12,723] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-19 01:29:12,731] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 576217 closing signal SIGTERM
+[2024-12-19 01:29:12,731] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 386482 closing signal SIGTERM
+[2024-12-19 01:29:12,729] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 349014 closing signal SIGTERM
+[2024-12-19 01:29:12,729] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 206512 closing signal SIGTERM
+[2024-12-19 01:29:12,731] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 386483 closing signal SIGTERM
+[2024-12-19 01:29:12,731] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 576218 closing signal SIGTERM
+[2024-12-19 01:29:12,731] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 386484 closing signal SIGTERM
+[2024-12-19 01:29:12,729] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 349015 closing signal SIGTERM
+[2024-12-19 01:29:12,729] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 206513 closing signal SIGTERM
+[2024-12-19 01:29:12,730] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 131382 closing signal SIGTERM
+[2024-12-19 01:29:12,731] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 386485 closing signal SIGTERM
+[2024-12-19 01:29:12,731] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 576219 closing signal SIGTERM
+[2024-12-19 01:29:12,731] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 576220 closing signal SIGTERM
+[2024-12-19 01:29:12,729] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 206514 closing signal SIGTERM
+[2024-12-19 01:29:12,729] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 349016 closing signal SIGTERM
+[2024-12-19 01:29:12,731] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 576221 closing signal SIGTERM
+[2024-12-19 01:29:12,731] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 386486 closing signal SIGTERM
+[2024-12-19 01:29:12,729] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 206515 closing signal SIGTERM
+[2024-12-19 01:29:12,730] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 349017 closing signal SIGTERM
+[2024-12-19 01:29:12,729] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 206516 closing signal SIGTERM
+[2024-12-19 01:29:12,730] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 349018 closing signal SIGTERM
+[2024-12-19 01:29:12,731] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 576222 closing signal SIGTERM
+[2024-12-19 01:29:12,731] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 386487 closing signal SIGTERM
+[2024-12-19 01:29:12,731] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 576223 closing signal SIGTERM
+[2024-12-19 01:29:12,729] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 206517 closing signal SIGTERM
+[2024-12-19 01:29:12,730] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 349019 closing signal SIGTERM
+[2024-12-19 01:29:12,729] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 206518 closing signal SIGTERM
+[2024-12-19 01:29:12,730] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 349020 closing signal SIGTERM
+[2024-12-19 01:29:12,731] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 131383 closing signal SIGTERM
+[2024-12-19 01:29:12,731] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 131384 closing signal SIGTERM
+[2024-12-19 01:29:12,731] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 131385 closing signal SIGTERM
+[2024-12-19 01:29:12,731] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 131386 closing signal SIGTERM
+[2024-12-19 01:29:12,724] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-19 01:29:12,731] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 131387 closing signal SIGTERM
+[2024-12-19 01:29:12,731] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 131388 closing signal SIGTERM
+[2024-12-19 01:29:12,731] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 131389 closing signal SIGTERM
+[2024-12-19 01:29:12,725] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-19 01:29:12,731] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 397623 closing signal SIGTERM
+[2024-12-19 01:29:12,731] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 59133 closing signal SIGTERM
+[2024-12-19 01:29:12,731] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 397624 closing signal SIGTERM
+[2024-12-19 01:29:12,731] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 397625 closing signal SIGTERM
+[2024-12-19 01:29:12,731] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 59134 closing signal SIGTERM
+[2024-12-19 01:29:12,731] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 397626 closing signal SIGTERM
+[2024-12-19 01:29:12,731] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 59135 closing signal SIGTERM
+[2024-12-19 01:29:12,731] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 59136 closing signal SIGTERM
+[2024-12-19 01:29:12,731] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 397627 closing signal SIGTERM
+[2024-12-19 01:29:12,731] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 397628 closing signal SIGTERM
+[2024-12-19 01:29:12,731] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 59137 closing signal SIGTERM
+[2024-12-19 01:29:12,731] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 397629 closing signal SIGTERM
+[2024-12-19 01:29:12,732] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 59138 closing signal SIGTERM
+[2024-12-19 01:29:12,732] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 59139 closing signal SIGTERM
+[2024-12-19 01:29:12,731] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 397630 closing signal SIGTERM
+[2024-12-19 01:29:12,732] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 59140 closing signal SIGTERM
diff --git a/logs/13502056-bench_80G_dp32_tp8_pp1_acc1_mbs8_seq4096_zero1_tpmodeRED_vocab131k.out b/logs/13502056-bench_80G_dp32_tp8_pp1_acc1_mbs8_seq4096_zero1_tpmodeRED_vocab131k.out
new file mode 100644
index 0000000000000000000000000000000000000000..54c1774dddba9bcf0ec65b9a3250ee679560dab4
--- /dev/null
+++ b/logs/13502056-bench_80G_dp32_tp8_pp1_acc1_mbs8_seq4096_zero1_tpmodeRED_vocab131k.out
@@ -0,0 +1,17473 @@
++ '[' -z 13502056 ']'
++ source /etc/profile.d/modules.sh
+++ . /usr/share/modules/init/bash
++++ unset _mlshdbg
++++ '[' 0 = 1 ']'
++++ unset _mlre _mlIFS
++++ '[' -n x ']'
++++ _mlIFS=' 	
+'
++++ IFS=' '
++++ '[' -n '' ']'
+++++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash autoinit
++++ _mlcode='module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+MODULEPATH_modshare=/etc/environment-modules/modules:1:/usr/share/modules/\$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1; export MODULEPATH_modshare;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+LOADEDMODULES=; export LOADEDMODULES;
+MODULEPATH=/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/\$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles; export MODULEPATH;
+test 0;'
++++ _mlret=0
++++ '[' -n x ']'
++++ IFS=' 	
+'
++++ unset _mlIFS
++++ unset _mlre _mlv _mlrv
++++ '[' 0 -eq 0 ']'
++++ eval 'module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+MODULEPATH_modshare=/etc/environment-modules/modules:1:/usr/share/modules/\$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1; export MODULEPATH_modshare;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+LOADEDMODULES=; export LOADEDMODULES;
+MODULEPATH=/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/\$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles; export MODULEPATH;
+test 0;'
+++++ MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl
+++++ export MODULES_CMD
+++++ ENV=/usr/share/modules/init/profile.sh
+++++ export ENV
+++++ MODULEPATH_modshare='/etc/environment-modules/modules:1:/usr/share/modules/$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1'
+++++ export MODULEPATH_modshare
+++++ BASH_ENV=/usr/share/modules/init/bash
+++++ export BASH_ENV
+++++ MODULESHOME=/usr/share/modules
+++++ export MODULESHOME
+++++ LOADEDMODULES=
+++++ export LOADEDMODULES
+++++ MODULEPATH='/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles'
+++++ export MODULEPATH
+++++ test 0
++++ '[' 0 = 1 ']'
++++ '[' -t 2 ']'
++++ export -f module
++++ export -f switchml
++++ '[' 5 -ge 3 ']'
++++ [[ ehxB =~ i ]]
++++ [[ ! :/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin: =~ :/usr/bin: ]]
++++ '[' '!' -n '' ']'
++++ MANPATH=:
++++ export MANPATH
+++++ manpath
++++ [[ ! :/admin/home/nouamane/.local/share/man:/fsx/nouamane/miniconda/envs/2-1-cu121/man:/fsx/nouamane/miniconda/envs/2-1-cu121/share/man:/fsx/nouamane/miniconda/man:/fsx/nouamane/miniconda/share/man:/opt/slurm/share/man:/opt/amazon/openmpi/share/man:/opt/amazon/efa/share/man:/usr/local/man:/usr/local/share/man:/usr/share/man:/admin/home/nouamane/.fzf/man:: =~ :/usr/share/man: ]]
++++ unset _mlcode _mlret
++++ '[' -n '' ']'
++ module load cuda/12.1
++ unset _mlshdbg
++ '[' 0 = 1 ']'
++ unset _mlre _mlIFS
++ '[' -n x ']'
++ _mlIFS=' 	
+'
++ IFS=' '
++ '[' -n '' ']'
+++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash load cuda/12.1
++ eval 'CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include;' export 'CPATH;
+LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:;' export 'LD_LIBRARY_PATH;
+FI_EFA_FORK_SAFE=1;' export 'FI_EFA_FORK_SAFE;
+MANPATH=/usr/local/cuda-12.1/share/man::;' export 'MANPATH;
+LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64;' export 'LIBRARY_PATH;
+_LMFILES_=/usr/share/modules/modulefiles/cuda/12.1;' export '_LMFILES_;
+LOADEDMODULES=cuda/12.1;' export 'LOADEDMODULES;
+MPI_PATH=/opt/amazon/openmpi;' export 'MPI_PATH;
+NCCL_HOME_modshare=/opt/nccl/build:1;' export 'NCCL_HOME_modshare;
+NCCL_PROTO=simple;' export 'NCCL_PROTO;
+MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1;' export 'MANPATH_modshare;
+LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1;' export 'LIBRARY_PATH_modshare;
+NCCL_SOCKET_IFNAME=enp;' export 'NCCL_SOCKET_IFNAME;
+AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl;' export 'AWS_OFI_NCCL_HOME;
+NCCL_HOME=/opt/nccl/build;' export 'NCCL_HOME;
+FI_PROVIDER=efa;' export 'FI_PROVIDER;
+AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1;' export 'AWS_OFI_NCCL_HOME_modshare;
+CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1;' export 'CPATH_modshare;
+LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2::1:/opt/nccl/build/lib:1:/opt/aws-ofi-nccl/lib:2:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/efa/lib:1;' export 'LD_LIBRARY_PATH_modshare;
+FI_EFA_ENABLE_SHM_TRANSFER=1;' export 'FI_EFA_ENABLE_SHM_TRANSFER;
+_LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1;' export '_LMFILES__modshare;
+LOADEDMODULES_modshare=cuda/12.1:1;' export 'LOADEDMODULES_modshare;
+MPI_PATH_modshare=/opt/amazon/openmpi:1;' export 'MPI_PATH_modshare;
+PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin;' export 'PATH;
+CUDA_HOME=/usr/local/cuda-12.1;' export 'CUDA_HOME;
+PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/usr/local/cuda-12.1/include:1:/opt/amazon/efa/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/usr/sbin:1:/opt/slurm/bin:1:/fsx/nouamane/miniconda/bin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:2:/opt/amazon/openmpi/bin:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1;' export 'PATH_modshare;
+test' '0;'
+++ CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include
+++ export CPATH
+++ LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:
+++ export LD_LIBRARY_PATH
+++ FI_EFA_FORK_SAFE=1
+++ export FI_EFA_FORK_SAFE
+++ MANPATH=/usr/local/cuda-12.1/share/man::
+++ export MANPATH
+++ LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64
+++ export LIBRARY_PATH
+++ _LMFILES_=/usr/share/modules/modulefiles/cuda/12.1
+++ export _LMFILES_
+++ LOADEDMODULES=cuda/12.1
+++ export LOADEDMODULES
+++ MPI_PATH=/opt/amazon/openmpi
+++ export MPI_PATH
+++ NCCL_HOME_modshare=/opt/nccl/build:1
+++ export NCCL_HOME_modshare
+++ NCCL_PROTO=simple
+++ export NCCL_PROTO
+++ MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1
+++ export MANPATH_modshare
+++ LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1
+++ export LIBRARY_PATH_modshare
+++ NCCL_SOCKET_IFNAME=enp
+++ export NCCL_SOCKET_IFNAME
+++ AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl
+++ export AWS_OFI_NCCL_HOME
+++ NCCL_HOME=/opt/nccl/build
+++ export NCCL_HOME
+++ FI_PROVIDER=efa
+++ export FI_PROVIDER
+++ AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1
+++ export AWS_OFI_NCCL_HOME_modshare
+++ CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1
+++ export CPATH_modshare
+++ LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2::1:/opt/nccl/build/lib:1:/opt/aws-ofi-nccl/lib:2:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/efa/lib:1
+++ export LD_LIBRARY_PATH_modshare
+++ FI_EFA_ENABLE_SHM_TRANSFER=1
+++ export FI_EFA_ENABLE_SHM_TRANSFER
+++ _LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1
+++ export _LMFILES__modshare
+++ LOADEDMODULES_modshare=cuda/12.1:1
+++ export LOADEDMODULES_modshare
+++ MPI_PATH_modshare=/opt/amazon/openmpi:1
+++ export MPI_PATH_modshare
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ export PATH
+++ CUDA_HOME=/usr/local/cuda-12.1
+++ export CUDA_HOME
+++ PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/usr/local/cuda-12.1/include:1:/opt/amazon/efa/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/usr/sbin:1:/opt/slurm/bin:1:/fsx/nouamane/miniconda/bin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:2:/opt/amazon/openmpi/bin:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1
+++ export PATH_modshare
+++ test 0
++ _mlstatus=0
++ '[' -n x ']'
++ IFS=' 	
+'
++ unset _mlre _mlv _mlrv _mlIFS
++ '[' -n '' ']'
++ unset _mlshdbg
++ return 0
++ source /fsx/nouamane/miniconda/bin/activate
+++ _CONDA_ROOT=/fsx/nouamane/miniconda
+++ . /fsx/nouamane/miniconda/etc/profile.d/conda.sh
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ '[' -z x ']'
+++ conda activate
+++ local cmd=activate
+++ case "$cmd" in
+++ __conda_activate activate
+++ '[' -n '' ']'
+++ local ask_conda
++++ PS1=
++++ __conda_exe shell.posix activate
++++ /fsx/nouamane/miniconda/bin/conda shell.posix activate
+Error while loading conda entry point: conda-libmamba-solver (module 'libmambapy' has no attribute 'QueryFormat')
+++ ask_conda='. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_2='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/etc/conda/activate.d/libxml2_activate.sh"'
+++ eval '. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_2='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/etc/conda/activate.d/libxml2_activate.sh"'
++++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh
+++++ test -n ''
+++++ unset XML_CATALOG_FILES
+++++ unset xml_catalog_files_libxml2
++++ PS1='(base) '
++++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++++ export CONDA_PREFIX=/fsx/nouamane/miniconda
++++ CONDA_PREFIX=/fsx/nouamane/miniconda
++++ export CONDA_SHLVL=3
++++ CONDA_SHLVL=3
++++ export CONDA_DEFAULT_ENV=base
++++ CONDA_DEFAULT_ENV=base
++++ export 'CONDA_PROMPT_MODIFIER=(base) '
++++ CONDA_PROMPT_MODIFIER='(base) '
++++ export CONDA_PREFIX_2=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ CONDA_PREFIX_2=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ . /fsx/nouamane/miniconda/etc/conda/activate.d/libxml2_activate.sh
+++++ test -n ''
+++++ xml_catalog_files_libxml2=
+++++ XML_CATALOG_FILES=
+++++ conda_catalog_files=
+++++ ifs_libxml2=' 	
+'
+++++ IFS=' '
+++++ rem=/fsx/nouamane/miniconda
+++++ for pre in ${rem}
+++++ test '' = /fsx/nouamane/miniconda
+++++ conda_catalog_files=/fsx/nouamane/miniconda
+++++ rem=
+++++ IFS=' 	
+'
+++++ conda_catalog_files='file:///fsx/nouamane/miniconda/etc/xml/catalog file:///etc/xml/catalog'
+++++ export 'XML_CATALOG_FILES=file:///fsx/nouamane/miniconda/etc/xml/catalog file:///etc/xml/catalog'
+++++ XML_CATALOG_FILES='file:///fsx/nouamane/miniconda/etc/xml/catalog file:///etc/xml/catalog'
+++++ unset conda_catalog_files ifs_libxml2 rem
+++ __conda_hashr
+++ '[' -n '' ']'
+++ '[' -n '' ']'
+++ hash -r
++ conda activate 2-1-cu121
++ local cmd=activate
++ case "$cmd" in
++ __conda_activate activate 2-1-cu121
++ '[' -n '' ']'
++ local ask_conda
+++ PS1='(base) '
+++ __conda_exe shell.posix activate 2-1-cu121
+++ /fsx/nouamane/miniconda/bin/conda shell.posix activate 2-1-cu121
+Error while loading conda entry point: conda-libmamba-solver (module 'libmambapy' has no attribute 'QueryFormat')
++ ask_conda='. "/fsx/nouamane/miniconda/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''4'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_3='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
++ eval '. "/fsx/nouamane/miniconda/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''4'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_3='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
+++ . /fsx/nouamane/miniconda/etc/conda/deactivate.d/libxml2_deactivate.sh
++++ test -n ''
++++ unset XML_CATALOG_FILES
++++ unset xml_catalog_files_libxml2
+++ PS1='(2-1-cu121) '
+++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ export CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ export CONDA_SHLVL=4
+++ CONDA_SHLVL=4
+++ export CONDA_DEFAULT_ENV=2-1-cu121
+++ CONDA_DEFAULT_ENV=2-1-cu121
+++ export 'CONDA_PROMPT_MODIFIER=(2-1-cu121) '
+++ CONDA_PROMPT_MODIFIER='(2-1-cu121) '
+++ export CONDA_PREFIX_3=/fsx/nouamane/miniconda
+++ CONDA_PREFIX_3=/fsx/nouamane/miniconda
+++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ export _CE_M=
+++ _CE_M=
+++ export _CE_CONDA=
+++ _CE_CONDA=
+++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh
++++ test -n ''
++++ xml_catalog_files_libxml2=
++++ XML_CATALOG_FILES=
++++ conda_catalog_files=
++++ ifs_libxml2=' 	
+'
++++ IFS=' '
++++ rem=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ for pre in ${rem}
++++ test '' = /fsx/nouamane/miniconda/envs/2-1-cu121
++++ conda_catalog_files=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ rem=
++++ IFS=' 	
+'
++++ conda_catalog_files='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ export 'XML_CATALOG_FILES=file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ XML_CATALOG_FILES='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ unset conda_catalog_files ifs_libxml2 rem
++ __conda_hashr
++ '[' -n '' ']'
++ '[' -n '' ']'
++ hash -r
++ export PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++ PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++ '[' -z '' ']'
+++ scontrol show hostnames 'ip-26-0-161-[78,103,123,138,142,153,178,221],ip-26-0-162-14,ip-26-0-163-[134,147,158,220,226],ip-26-0-164-236,ip-26-0-165-[24,38],ip-26-0-167-[51,175,177,217,245],ip-26-0-168-[30,120,238],ip-26-0-169-[86,132,139,207,239,247],ip-26-0-170-31'
++ export 'NODELIST=ip-26-0-161-78
+ip-26-0-161-103
+ip-26-0-161-123
+ip-26-0-161-138
+ip-26-0-161-142
+ip-26-0-161-153
+ip-26-0-161-178
+ip-26-0-161-221
+ip-26-0-162-14
+ip-26-0-163-134
+ip-26-0-163-147
+ip-26-0-163-158
+ip-26-0-163-220
+ip-26-0-163-226
+ip-26-0-164-236
+ip-26-0-165-24
+ip-26-0-165-38
+ip-26-0-167-51
+ip-26-0-167-175
+ip-26-0-167-177
+ip-26-0-167-217
+ip-26-0-167-245
+ip-26-0-168-30
+ip-26-0-168-120
+ip-26-0-168-238
+ip-26-0-169-86
+ip-26-0-169-132
+ip-26-0-169-139
+ip-26-0-169-207
+ip-26-0-169-239
+ip-26-0-169-247
+ip-26-0-170-31'
++ NODELIST='ip-26-0-161-78
+ip-26-0-161-103
+ip-26-0-161-123
+ip-26-0-161-138
+ip-26-0-161-142
+ip-26-0-161-153
+ip-26-0-161-178
+ip-26-0-161-221
+ip-26-0-162-14
+ip-26-0-163-134
+ip-26-0-163-147
+ip-26-0-163-158
+ip-26-0-163-220
+ip-26-0-163-226
+ip-26-0-164-236
+ip-26-0-165-24
+ip-26-0-165-38
+ip-26-0-167-51
+ip-26-0-167-175
+ip-26-0-167-177
+ip-26-0-167-217
+ip-26-0-167-245
+ip-26-0-168-30
+ip-26-0-168-120
+ip-26-0-168-238
+ip-26-0-169-86
+ip-26-0-169-132
+ip-26-0-169-139
+ip-26-0-169-207
+ip-26-0-169-239
+ip-26-0-169-247
+ip-26-0-170-31'
+++ head -n1
+++ scontrol show hostnames 'ip-26-0-161-[78,103,123,138,142,153,178,221],ip-26-0-162-14,ip-26-0-163-[134,147,158,220,226],ip-26-0-164-236,ip-26-0-165-[24,38],ip-26-0-167-[51,175,177,217,245],ip-26-0-168-[30,120,238],ip-26-0-169-[86,132,139,207,239,247],ip-26-0-170-31'
++ export MASTER_NODE=ip-26-0-161-78
++ MASTER_NODE=ip-26-0-161-78
++ export MASTER_PORT=12356
++ MASTER_PORT=12356
++ export NNODES=32
++ NNODES=32
++ export GPUS_PER_NODE=8
++ GPUS_PER_NODE=8
++ export WORLD_SIZE=256
++ WORLD_SIZE=256
++ export CUDA_DEVICE_MAX_CONNECTIONS=1
++ CUDA_DEVICE_MAX_CONNECTIONS=1
++ export NCCL_DEBUG=WARN
++ NCCL_DEBUG=WARN
++ export NANOTRON_BENCHMARK=1
++ NANOTRON_BENCHMARK=1
++ export WANDB_MODE=disabled
++ WANDB_MODE=disabled
++ export TORCH_NCCL_ASYNC_ERROR_HANDLING=1
++ TORCH_NCCL_ASYNC_ERROR_HANDLING=1
++ '[' -z '' ']'
++ echo '=== GPU Topology ==='
+=== GPU Topology ===
++ nvidia-smi topo -m
+	GPU0	GPU1	GPU2	GPU3	GPU4	GPU5	GPU6	GPU7	CPU Affinity	NUMA Affinity	GPU NUMA ID
+GPU0	 X 	NV18	NV18	NV18	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU1	NV18	 X 	NV18	NV18	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU2	NV18	NV18	 X 	NV18	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU3	NV18	NV18	NV18	 X 	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU4	NV18	NV18	NV18	NV18	 X 	NV18	NV18	NV18	48-95	1		N/A
+GPU5	NV18	NV18	NV18	NV18	NV18	 X 	NV18	NV18	48-95	1		N/A
+GPU6	NV18	NV18	NV18	NV18	NV18	NV18	 X 	NV18	48-95	1		N/A
+GPU7	NV18	NV18	NV18	NV18	NV18	NV18	NV18	 X 	48-95	1		N/A
+
+Legend:
+
+  X    = Self
+  SYS  = Connection traversing PCIe as well as the SMP interconnect between NUMA nodes (e.g., QPI/UPI)
+  NODE = Connection traversing PCIe as well as the interconnect between PCIe Host Bridges within a NUMA node
+  PHB  = Connection traversing PCIe as well as a PCIe Host Bridge (typically the CPU)
+  PXB  = Connection traversing multiple PCIe bridges (without traversing the PCIe Host Bridge)
+  PIX  = Connection traversing at most a single PCIe bridge
+  NV#  = Connection traversing a bonded set of # NVLinks
++ echo ==================
+==================
++ export SRUN_ALLOC_ARGS=
++ SRUN_ALLOC_ARGS=
++ echo 'Master node: ip-26-0-161-78'
+Master node: ip-26-0-161-78
++ echo 'All nodes: ip-26-0-161-78
+ip-26-0-161-103
+ip-26-0-161-123
+ip-26-0-161-138
+ip-26-0-161-142
+ip-26-0-161-153
+ip-26-0-161-178
+ip-26-0-161-221
+ip-26-0-162-14
+ip-26-0-163-134
+ip-26-0-163-147
+ip-26-0-163-158
+ip-26-0-163-220
+ip-26-0-163-226
+ip-26-0-164-236
+ip-26-0-165-24
+ip-26-0-165-38
+ip-26-0-167-51
+ip-26-0-167-175
+ip-26-0-167-177
+ip-26-0-167-217
+ip-26-0-167-245
+ip-26-0-168-30
+ip-26-0-168-120
+ip-26-0-168-238
+ip-26-0-169-86
+ip-26-0-169-132
+ip-26-0-169-139
+ip-26-0-169-207
+ip-26-0-169-239
+ip-26-0-169-247
+ip-26-0-170-31'
+All nodes: ip-26-0-161-78
+ip-26-0-161-103
+ip-26-0-161-123
+ip-26-0-161-138
+ip-26-0-161-142
+ip-26-0-161-153
+ip-26-0-161-178
+ip-26-0-161-221
+ip-26-0-162-14
+ip-26-0-163-134
+ip-26-0-163-147
+ip-26-0-163-158
+ip-26-0-163-220
+ip-26-0-163-226
+ip-26-0-164-236
+ip-26-0-165-24
+ip-26-0-165-38
+ip-26-0-167-51
+ip-26-0-167-175
+ip-26-0-167-177
+ip-26-0-167-217
+ip-26-0-167-245
+ip-26-0-168-30
+ip-26-0-168-120
+ip-26-0-168-238
+ip-26-0-169-86
+ip-26-0-169-132
+ip-26-0-169-139
+ip-26-0-169-207
+ip-26-0-169-239
+ip-26-0-169-247
+ip-26-0-170-31
++ echo 'World size: 256'
+World size: 256
++ '[' -n '' ']'
++ srun --wait=0 --kill-on-bad-exit=1 torchrun --nnodes=32 --nproc_per_node=8 --rdzv_id=13502056 --rdzv_backend=c10d --rdzv_endpoint=ip-26-0-161-78:12356 --max_restarts 0 --rdzv_conf timeout=60 /fsx/nouamane/projects/nanotron/run_train.py --config-file benchmark/configs/config_80G_dp32_tp8_pp1_acc1_mbs8_seq4096_zero1_tpmodeRED_vocab131k.yaml
+[2024-12-19 05:41:43,247] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-19 05:41:43,247] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-19 05:41:43,248] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-19 05:41:43,248] torch.distributed.run: [WARNING] 
+[2024-12-19 05:41:43,248] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 05:41:43,248] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-19 05:41:43,248] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 05:41:43,249] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-19 05:41:43,249] torch.distributed.run: [WARNING] 
+[2024-12-19 05:41:43,249] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 05:41:43,249] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-19 05:41:43,249] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 05:41:43,249] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-19 05:41:43,249] torch.distributed.run: [WARNING] 
+[2024-12-19 05:41:43,249] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 05:41:43,249] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-19 05:41:43,249] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 05:41:43,251] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-19 05:41:43,251] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-19 05:41:43,252] torch.distributed.run: [WARNING] 
+[2024-12-19 05:41:43,252] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 05:41:43,252] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-19 05:41:43,252] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 05:41:43,252] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-19 05:41:43,250] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-19 05:41:43,257] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-19 05:41:43,259] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-19 05:41:43,259] torch.distributed.run: [WARNING] 
+[2024-12-19 05:41:43,259] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 05:41:43,259] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-19 05:41:43,259] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 05:41:43,259] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-19 05:41:43,259] torch.distributed.run: [WARNING] 
+[2024-12-19 05:41:43,259] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 05:41:43,259] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-19 05:41:43,259] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 05:41:43,262] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-19 05:41:43,264] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-19 05:41:43,263] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-19 05:41:43,263] torch.distributed.run: [WARNING] 
+[2024-12-19 05:41:43,263] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 05:41:43,263] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-19 05:41:43,263] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 05:41:43,265] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-19 05:41:43,265] torch.distributed.run: [WARNING] 
+[2024-12-19 05:41:43,265] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 05:41:43,265] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-19 05:41:43,265] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 05:41:43,265] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-19 05:41:43,270] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-19 05:41:43,270] torch.distributed.run: [WARNING] 
+[2024-12-19 05:41:43,270] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 05:41:43,270] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-19 05:41:43,270] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 05:41:43,269] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-19 05:41:43,267] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-19 05:41:43,267] torch.distributed.run: [WARNING] 
+[2024-12-19 05:41:43,267] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 05:41:43,267] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-19 05:41:43,267] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 05:41:43,272] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-19 05:41:43,272] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-19 05:41:43,273] torch.distributed.run: [WARNING] 
+[2024-12-19 05:41:43,273] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 05:41:43,273] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-19 05:41:43,273] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 05:41:43,276] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-19 05:41:43,276] torch.distributed.run: [WARNING] 
+[2024-12-19 05:41:43,276] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 05:41:43,276] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-19 05:41:43,276] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 05:41:43,281] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-19 05:41:43,247] torch.distributed.run: [WARNING] 
+[2024-12-19 05:41:43,247] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 05:41:43,247] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-19 05:41:43,247] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 05:41:43,287] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-19 05:41:43,287] torch.distributed.run: [WARNING] 
+[2024-12-19 05:41:43,287] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 05:41:43,287] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-19 05:41:43,287] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 05:41:43,252] torch.distributed.run: [WARNING] 
+[2024-12-19 05:41:43,252] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 05:41:43,252] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-19 05:41:43,252] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 05:41:43,247] torch.distributed.run: [WARNING] 
+[2024-12-19 05:41:43,247] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 05:41:43,247] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-19 05:41:43,247] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 05:41:43,250] torch.distributed.run: [WARNING] 
+[2024-12-19 05:41:43,250] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 05:41:43,250] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-19 05:41:43,250] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 05:41:43,251] torch.distributed.run: [WARNING] 
+[2024-12-19 05:41:43,251] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 05:41:43,251] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-19 05:41:43,251] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 05:41:43,296] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-19 05:41:43,296] torch.distributed.run: [WARNING] 
+[2024-12-19 05:41:43,296] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 05:41:43,296] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-19 05:41:43,296] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 05:41:43,297] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-19 05:41:43,297] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-19 05:41:43,257] torch.distributed.run: [WARNING] 
+[2024-12-19 05:41:43,257] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 05:41:43,257] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-19 05:41:43,257] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 05:41:43,303] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-19 05:41:43,262] torch.distributed.run: [WARNING] 
+[2024-12-19 05:41:43,262] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 05:41:43,262] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-19 05:41:43,262] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 05:41:43,265] torch.distributed.run: [WARNING] 
+[2024-12-19 05:41:43,265] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 05:41:43,265] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-19 05:41:43,265] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 05:41:43,264] torch.distributed.run: [WARNING] 
+[2024-12-19 05:41:43,264] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 05:41:43,264] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-19 05:41:43,264] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 05:41:43,310] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-19 05:41:43,310] torch.distributed.run: [WARNING] 
+[2024-12-19 05:41:43,310] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 05:41:43,310] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-19 05:41:43,310] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 05:41:43,269] torch.distributed.run: [WARNING] 
+[2024-12-19 05:41:43,269] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 05:41:43,269] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-19 05:41:43,269] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 05:41:43,272] torch.distributed.run: [WARNING] 
+[2024-12-19 05:41:43,272] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 05:41:43,272] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-19 05:41:43,272] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 05:41:43,320] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-19 05:41:43,281] torch.distributed.run: [WARNING] 
+[2024-12-19 05:41:43,281] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 05:41:43,281] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-19 05:41:43,281] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 05:41:43,328] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-19 05:41:43,297] torch.distributed.run: [WARNING] 
+[2024-12-19 05:41:43,297] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 05:41:43,297] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-19 05:41:43,297] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 05:41:43,297] torch.distributed.run: [WARNING] 
+[2024-12-19 05:41:43,297] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 05:41:43,297] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-19 05:41:43,297] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 05:41:43,303] torch.distributed.run: [WARNING] 
+[2024-12-19 05:41:43,303] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 05:41:43,303] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-19 05:41:43,303] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 05:41:43,320] torch.distributed.run: [WARNING] 
+[2024-12-19 05:41:43,320] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 05:41:43,320] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-19 05:41:43,320] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 05:41:43,328] torch.distributed.run: [WARNING] 
+[2024-12-19 05:41:43,328] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 05:41:43,328] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-19 05:41:43,328] torch.distributed.run: [WARNING] *****************************************
+NCCL version 2.18.5+cuda12.2
+12/19/2024 05:42:10 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]: Measuring inter-GPU and intra-node bandwidth...
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+12/19/2024 05:42:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]: Bandwidth measurement complete. Time taken: 18.57 seconds
+12/19/2024 05:42:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]: Config:
+12/19/2024 05:42:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]: Config(general=GeneralArgs(project='debug',
+12/19/2024 05:42:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                            run='80G_dp32_tp8_pp1_acc1_mbs8_seq4096_zero1_tpmodeRED_vocab131k',
+12/19/2024 05:42:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                            seed=42,
+12/19/2024 05:42:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                            step=None,
+12/19/2024 05:42:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                            consumed_train_samples=None,
+12/19/2024 05:42:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                            benchmark_csv_path=PosixPath('benchmark/results/bench_final2.csv'),
+12/19/2024 05:42:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                            ignore_sanity_checks=True),
+12/19/2024 05:42:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:        parallelism=ParallelismArgs(dp=32,
+12/19/2024 05:42:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                                    pp=1,
+12/19/2024 05:42:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                                    tp=8,
+12/19/2024 05:42:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                                    pp_engine=<nanotron.parallel.pipeline_parallel.engine.OneForwardOneBackwardPipelineEngine object at 0x7fd76bc4cb80>,
+12/19/2024 05:42:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                                    tp_mode=<TensorParallelLinearMode.REDUCE_SCATTER: 2>,
+12/19/2024 05:42:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                                    tp_linear_async_communication=True,
+12/19/2024 05:42:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                                    recompute_layer=False,
+12/19/2024 05:42:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                                    tp_recompute_allgather=True,
+12/19/2024 05:42:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                                    expert_parallel_size=1),
+12/19/2024 05:42:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:        model=ModelArgs(model_config=LlamaConfig(bos_token_id=0,
+12/19/2024 05:42:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                                                 eos_token_id=0,
+12/19/2024 05:42:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                                                 hidden_act='silu',
+12/19/2024 05:42:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                                                 hidden_size=8192,
+12/19/2024 05:42:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                                                 initializer_range=0.02,
+12/19/2024 05:42:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                                                 intermediate_size=28672,
+12/19/2024 05:42:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                                                 is_llama_config=True,
+12/19/2024 05:42:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                                                 max_position_embeddings=4096,
+12/19/2024 05:42:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                                                 num_attention_heads=64,
+12/19/2024 05:42:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                                                 num_hidden_layers=80,
+12/19/2024 05:42:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                                                 num_key_value_heads=64,
+12/19/2024 05:42:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                                                 pad_token_id=None,
+12/19/2024 05:42:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                                                 pretraining_tp=1,
+12/19/2024 05:42:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                                                 rms_norm_eps=1e-05,
+12/19/2024 05:42:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                                                 rope_scaling=None,
+12/19/2024 05:42:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                                                 rope_theta=10000.0,
+12/19/2024 05:42:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                                                 rope_interleaved=False,
+12/19/2024 05:42:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                                                 tie_word_embeddings=False,
+12/19/2024 05:42:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                                                 use_cache=True,
+12/19/2024 05:42:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                                                 vocab_size=131072),
+12/19/2024 05:42:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                        init_method=RandomInit(std=0.02),
+12/19/2024 05:42:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                        dtype=torch.bfloat16,
+12/19/2024 05:42:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                        make_vocab_size_divisible_by=1,
+12/19/2024 05:42:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                        ddp_bucket_cap_mb=25),
+12/19/2024 05:42:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:        tokenizer=TokenizerArgs(tokenizer_name_or_path='robot-test/dummy-tokenizer-wordlevel',
+12/19/2024 05:42:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                                tokenizer_revision=None,
+12/19/2024 05:42:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                                tokenizer_max_length=None),
+12/19/2024 05:42:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:        checkpoints=CheckpointsArgs(checkpoints_path=PosixPath('checkpoints'),
+12/19/2024 05:42:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                                    checkpoint_interval=10000,
+12/19/2024 05:42:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                                    save_initial_state=False,
+12/19/2024 05:42:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                                    save_final_state=False,
+12/19/2024 05:42:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                                    resume_checkpoint_path=None,
+12/19/2024 05:42:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                                    checkpoints_path_is_shared_file_system=False),
+12/19/2024 05:42:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:        logging=LoggingArgs(log_level='info',
+12/19/2024 05:42:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                            log_level_replica='info',
+12/19/2024 05:42:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                            iteration_step_info_interval=1),
+12/19/2024 05:42:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:        tokens=TokensArgs(sequence_length=4096,
+12/19/2024 05:42:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                          train_steps=100,
+12/19/2024 05:42:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                          micro_batch_size=8,
+12/19/2024 05:42:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                          batch_accumulation_per_replica=1,
+12/19/2024 05:42:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                          val_check_interval=100,
+12/19/2024 05:42:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                          limit_val_batches=0,
+12/19/2024 05:42:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                          limit_test_batches=0),
+12/19/2024 05:42:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:        optimizer=OptimizerArgs(optimizer_factory=AdamWOptimizerArgs(adam_eps=1e-08,
+12/19/2024 05:42:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                                                                     adam_beta1=0.9,
+12/19/2024 05:42:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                                                                     adam_beta2=0.95,
+12/19/2024 05:42:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                                                                     torch_adam_is_fused=True,
+12/19/2024 05:42:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                                                                     name='adamW'),
+12/19/2024 05:42:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                                zero_stage=1,
+12/19/2024 05:42:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                                weight_decay=0.01,
+12/19/2024 05:42:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                                clip_grad=1.0,
+12/19/2024 05:42:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                                accumulate_grad_in_fp32=True,
+12/19/2024 05:42:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                                learning_rate_scheduler=LRSchedulerArgs(learning_rate=0.0003,
+12/19/2024 05:42:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                                                                        lr_warmup_steps=2,
+12/19/2024 05:42:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                                                                        lr_warmup_style='linear',
+12/19/2024 05:42:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                                                                        lr_decay_style='cosine',
+12/19/2024 05:42:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                                                                        lr_decay_steps=13,
+12/19/2024 05:42:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                                                                        lr_decay_starting_step=None,
+12/19/2024 05:42:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                                                                        min_decay_lr=1e-05)),
+12/19/2024 05:42:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:        data_stages=[DatasetStageArgs(name='Stable Training Stage',
+12/19/2024 05:42:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                                      start_training_step=1,
+12/19/2024 05:42:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                                      data=DataArgs(dataset=None,
+12/19/2024 05:42:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                                                    seed=42,
+12/19/2024 05:42:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                                                    num_loading_workers=1))],
+12/19/2024 05:42:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:        profiler=None,
+12/19/2024 05:42:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:        lighteval=None,
+12/19/2024 05:42:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:        s3_upload=None)
+12/19/2024 05:42:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]: Model Config:
+12/19/2024 05:42:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]: LlamaConfig(bos_token_id=0,
+12/19/2024 05:42:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:             eos_token_id=0,
+12/19/2024 05:42:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:             hidden_act='silu',
+12/19/2024 05:42:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:             hidden_size=8192,
+12/19/2024 05:42:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:             initializer_range=0.02,
+12/19/2024 05:42:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:             intermediate_size=28672,
+12/19/2024 05:42:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:             is_llama_config=True,
+12/19/2024 05:42:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:             max_position_embeddings=4096,
+12/19/2024 05:42:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:             num_attention_heads=64,
+12/19/2024 05:42:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:             num_hidden_layers=80,
+12/19/2024 05:42:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:             num_key_value_heads=64,
+12/19/2024 05:42:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:             pad_token_id=None,
+12/19/2024 05:42:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:             pretraining_tp=1,
+12/19/2024 05:42:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:             rms_norm_eps=1e-05,
+12/19/2024 05:42:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:             rope_scaling=None,
+12/19/2024 05:42:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:             rope_theta=10000.0,
+12/19/2024 05:42:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:             rope_interleaved=False,
+12/19/2024 05:42:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:             tie_word_embeddings=False,
+12/19/2024 05:42:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:             use_cache=True,
+12/19/2024 05:42:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:             vocab_size=131072)
+12/19/2024 05:42:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]: Building model..
+12/19/2024 05:42:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]: Initialize RoPE Theta = 10000.0
+12/19/2024 05:42:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]: Setting PP block ranks...
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+12/19/2024 05:42:30 [INFO|DP=0|PP=0|TP=3|ip-26-0-161-103]: Local number of parameters: 10G (19074.52MiB)
+12/19/2024 05:42:30 [INFO|DP=0|PP=0|TP=7|ip-26-0-161-103]: Local number of parameters: 10G (19074.52MiB)
+12/19/2024 05:42:30 [INFO|DP=0|PP=0|TP=4|ip-26-0-161-103]: Local number of parameters: 10G (19074.52MiB)
+12/19/2024 05:42:30 [INFO|DP=0|PP=0|TP=2|ip-26-0-161-103]: Local number of parameters: 10G (19074.52MiB)
+12/19/2024 05:42:30 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]: Total number of parameters: 80G (152596.12MiB)
+12/19/2024 05:42:30 [INFO|DP=0|PP=0|TP=5|ip-26-0-161-103]: Local number of parameters: 10G (19074.52MiB)
+12/19/2024 05:42:30 [INFO|DP=0|PP=0|TP=6|ip-26-0-161-103]: Local number of parameters: 10G (19074.52MiB)
+12/19/2024 05:42:30 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]: Local number of parameters: 10G (19074.52MiB)
+12/19/2024 05:42:30 [INFO|DP=0|PP=0|TP=1|ip-26-0-161-103]: Local number of parameters: 10G (19074.52MiB)
+12/19/2024 05:42:30 [INFO|DP=0|PP=0|TP=3|ip-26-0-161-103]: [After model building] Memory usage: 19074.60MiB. Peak allocated: 19074.73MiB Peak reserved: 19144.00MiB
+12/19/2024 05:42:30 [INFO|DP=0|PP=0|TP=4|ip-26-0-161-103]: [After model building] Memory usage: 19074.60MiB. Peak allocated: 19074.73MiB Peak reserved: 19144.00MiB
+12/19/2024 05:42:30 [INFO|DP=0|PP=0|TP=2|ip-26-0-161-103]: [After model building] Memory usage: 19074.60MiB. Peak allocated: 19074.73MiB Peak reserved: 19144.00MiB
+12/19/2024 05:42:30 [INFO|DP=0|PP=0|TP=7|ip-26-0-161-103]: [After model building] Memory usage: 19074.60MiB. Peak allocated: 19074.73MiB Peak reserved: 19144.00MiB
+12/19/2024 05:42:30 [INFO|DP=0|PP=0|TP=5|ip-26-0-161-103]: [After model building] Memory usage: 19074.60MiB. Peak allocated: 19074.73MiB Peak reserved: 19144.00MiB
+12/19/2024 05:42:30 [INFO|DP=0|PP=0|TP=6|ip-26-0-161-103]: [After model building] Memory usage: 19074.60MiB. Peak allocated: 19074.73MiB Peak reserved: 19144.00MiB
+12/19/2024 05:42:30 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]: [After model building] Memory usage: 19074.60MiB. Peak allocated: 19074.73MiB Peak reserved: 19144.00MiB
+12/19/2024 05:42:30 [INFO|DP=0|PP=0|TP=1|ip-26-0-161-103]: [After model building] Memory usage: 19074.60MiB. Peak allocated: 19074.73MiB Peak reserved: 19144.00MiB
+12/19/2024 05:42:30 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]: No checkpoint path provided.
+12/19/2024 05:42:30 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]: Parametrizing model parameters using StandardParametrizator
+12/19/2024 05:42:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]: [Optimizer Building] Using LearningRateForSP as learning rate
+12/19/2024 05:42:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]: [ZeRO sharding] Size of optimizer params per rank:
+12/19/2024 05:42:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]: [ZeRO sharding] DP Rank 0 has 313M out of 10G (3.12%) params' optimizer states
+12/19/2024 05:42:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]: [ZeRO sharding] DP Rank 1 has 313M out of 10G (3.12%) params' optimizer states
+12/19/2024 05:42:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]: [ZeRO sharding] DP Rank 2 has 313M out of 10G (3.12%) params' optimizer states
+12/19/2024 05:42:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]: [ZeRO sharding] DP Rank 3 has 313M out of 10G (3.12%) params' optimizer states
+12/19/2024 05:42:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]: [ZeRO sharding] DP Rank 4 has 313M out of 10G (3.12%) params' optimizer states
+12/19/2024 05:42:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]: [ZeRO sharding] DP Rank 5 has 313M out of 10G (3.12%) params' optimizer states
+12/19/2024 05:42:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]: [ZeRO sharding] DP Rank 6 has 313M out of 10G (3.12%) params' optimizer states
+12/19/2024 05:42:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]: [ZeRO sharding] DP Rank 7 has 313M out of 10G (3.12%) params' optimizer states
+12/19/2024 05:42:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]: [ZeRO sharding] DP Rank 8 has 313M out of 10G (3.12%) params' optimizer states
+12/19/2024 05:42:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]: [ZeRO sharding] DP Rank 9 has 313M out of 10G (3.12%) params' optimizer states
+12/19/2024 05:42:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]: [ZeRO sharding] DP Rank 10 has 313M out of 10G (3.12%) params' optimizer states
+12/19/2024 05:42:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]: [ZeRO sharding] DP Rank 11 has 313M out of 10G (3.12%) params' optimizer states
+12/19/2024 05:42:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]: [ZeRO sharding] DP Rank 12 has 313M out of 10G (3.12%) params' optimizer states
+12/19/2024 05:42:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]: [ZeRO sharding] DP Rank 13 has 313M out of 10G (3.12%) params' optimizer states
+12/19/2024 05:42:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]: [ZeRO sharding] DP Rank 14 has 313M out of 10G (3.12%) params' optimizer states
+12/19/2024 05:42:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]: [ZeRO sharding] DP Rank 15 has 313M out of 10G (3.12%) params' optimizer states
+12/19/2024 05:42:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]: [ZeRO sharding] DP Rank 16 has 313M out of 10G (3.12%) params' optimizer states
+12/19/2024 05:42:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]: [ZeRO sharding] DP Rank 17 has 313M out of 10G (3.12%) params' optimizer states
+12/19/2024 05:42:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]: [ZeRO sharding] DP Rank 18 has 313M out of 10G (3.12%) params' optimizer states
+12/19/2024 05:42:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]: [ZeRO sharding] DP Rank 19 has 313M out of 10G (3.12%) params' optimizer states
+12/19/2024 05:42:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]: [ZeRO sharding] DP Rank 20 has 313M out of 10G (3.12%) params' optimizer states
+12/19/2024 05:42:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]: [ZeRO sharding] DP Rank 21 has 313M out of 10G (3.12%) params' optimizer states
+12/19/2024 05:42:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]: [ZeRO sharding] DP Rank 22 has 313M out of 10G (3.12%) params' optimizer states
+12/19/2024 05:42:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]: [ZeRO sharding] DP Rank 23 has 313M out of 10G (3.12%) params' optimizer states
+12/19/2024 05:42:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]: [ZeRO sharding] DP Rank 24 has 313M out of 10G (3.12%) params' optimizer states
+12/19/2024 05:42:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]: [ZeRO sharding] DP Rank 25 has 313M out of 10G (3.12%) params' optimizer states
+12/19/2024 05:42:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]: [ZeRO sharding] DP Rank 26 has 313M out of 10G (3.12%) params' optimizer states
+12/19/2024 05:42:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]: [ZeRO sharding] DP Rank 27 has 313M out of 10G (3.12%) params' optimizer states
+12/19/2024 05:42:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]: [ZeRO sharding] DP Rank 28 has 313M out of 10G (3.12%) params' optimizer states
+12/19/2024 05:42:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]: [ZeRO sharding] DP Rank 29 has 313M out of 10G (3.12%) params' optimizer states
+12/19/2024 05:42:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]: [ZeRO sharding] DP Rank 30 has 313M out of 10G (3.12%) params' optimizer states
+12/19/2024 05:42:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]: [ZeRO sharding] DP Rank 31 has 313M out of 10G (3.12%) params' optimizer states
+12/19/2024 05:42:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]: [Training Plan] Stage Stable Training Stage has 99 remaining training steps and has consumed 0 samples
+12/19/2024 05:42:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]: Using dummy data generator
+12/19/2024 05:42:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]: [Training Plan] There are 1 training stages 
+12/19/2024 05:42:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]: [Stage Stable Training Stage] start from step 1 
+12/19/2024 05:42:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]: 
+12/19/2024 05:42:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]: [Start training] datetime: 2024-12-19 05:42:33.959600 | mbs: 8 | grad_accum: 1 | global_batch_size: 256 | sequence_length: 4096 | train_steps: 100 | start_iteration_step: 0 | consumed_train_samples: 0
+12/19/2024 05:42:35 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]: Resuming training from stage Stable Training Stage, it has trained for 0 samples and has 99 remaining train steps
+12/19/2024 05:42:35 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:  Memory usage: 58416.76MiB. Peak allocated 58416.76MiB. Peak reserved: 58488.00MiB
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 723, in _core_forward
+    hidden_states = self.input_layernorm(hidden_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/nn/layer_norm.py", line 44, in forward
+    return layer_norm_fn(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/flash_attn/ops/triton/layer_norm.py", line 875, in layer_norm_fn
+    return LayerNormFn.apply(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/flash_attn/ops/triton/layer_norm.py", line 748, in forward
+    y, y1, mean, rstd, residual_out, seeds, dropout_mask, dropout_mask1 = _layer_norm_fwd(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/flash_attn/ops/triton/layer_norm.py", line 335, in _layer_norm_fwd
+    _layer_norm_fwd_1pass_kernel[(M,)](
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/triton/runtime/jit.py", line 330, in <lambda>
+    return lambda *args, **kwargs: self.run(grid=grid, warmup=False, *args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/triton/runtime/autotuner.py", line 183, in run
+    timings = {config: self._bench(*args, config=config, **kwargs) for config in pruned_configs}
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/triton/runtime/autotuner.py", line 183, in <dictcomp>
+    timings = {config: self._bench(*args, config=config, **kwargs) for config in pruned_configs}
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/triton/runtime/autotuner.py", line 163, in _bench
+    return self.do_bench(kernel_call, quantiles=(0.5, 0.2, 0.8))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/triton/testing.py", line 117, in do_bench
+    fn()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/triton/runtime/autotuner.py", line 149, in kernel_call
+    self.fn.run(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/triton/runtime/autotuner.py", line 369, in run
+    return self.fn.run(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/triton/runtime/autotuner.py", line 369, in run
+    return self.fn.run(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/triton/runtime/autotuner.py", line 369, in run
+    return self.fn.run(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/triton/runtime/jit.py", line 653, in run
+    kernel.run(grid_0, grid_1, grid_2, stream, kernel.function, kernel.packed_metadata, launch_metadata,
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/triton/compiler/compiler.py", line 395, in __getattribute__
+    self._init_handles()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/triton/compiler/compiler.py", line 390, in _init_handles
+    self.module, self.function, self.n_regs, self.n_spills = driver.active.utils.load_binary(
+RuntimeError: Triton Error [CUDA]: device kernel image is invalid
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 723, in _core_forward
+    hidden_states = self.input_layernorm(hidden_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/nn/layer_norm.py", line 44, in forward
+    return layer_norm_fn(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/flash_attn/ops/triton/layer_norm.py", line 875, in layer_norm_fn
+    return LayerNormFn.apply(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/flash_attn/ops/triton/layer_norm.py", line 748, in forward
+    y, y1, mean, rstd, residual_out, seeds, dropout_mask, dropout_mask1 = _layer_norm_fwd(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/flash_attn/ops/triton/layer_norm.py", line 335, in _layer_norm_fwd
+    _layer_norm_fwd_1pass_kernel[(M,)](
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/triton/runtime/jit.py", line 330, in <lambda>
+    return lambda *args, **kwargs: self.run(grid=grid, warmup=False, *args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/triton/runtime/autotuner.py", line 183, in run
+    timings = {config: self._bench(*args, config=config, **kwargs) for config in pruned_configs}
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/triton/runtime/autotuner.py", line 183, in <dictcomp>
+    timings = {config: self._bench(*args, config=config, **kwargs) for config in pruned_configs}
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/triton/runtime/autotuner.py", line 163, in _bench
+    return self.do_bench(kernel_call, quantiles=(0.5, 0.2, 0.8))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/triton/testing.py", line 117, in do_bench
+    fn()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/triton/runtime/autotuner.py", line 149, in kernel_call
+    self.fn.run(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/triton/runtime/autotuner.py", line 369, in run
+    return self.fn.run(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/triton/runtime/autotuner.py", line 369, in run
+    return self.fn.run(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/triton/runtime/autotuner.py", line 369, in run
+    return self.fn.run(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/triton/runtime/jit.py", line 623, in run
+    kernel = self.compile(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/triton/compiler/compiler.py", line 303, in compile
+    return CompiledKernel(src, metadata_group, hash)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/triton/compiler/compiler.py", line 368, in __init__
+    self.asm = AsmDict({
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/triton/compiler/compiler.py", line 369, in <dictcomp>
+    file.suffix[1:]: file.read_bytes() if file.suffix[1:] == binary_ext else file.read_text()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/pathlib.py", line 1126, in read_bytes
+    with self.open(mode='rb') as f:
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/pathlib.py", line 1119, in open
+    return self._accessor.open(self, mode, buffering, encoding, errors,
+FileNotFoundError: [Errno 2] No such file or directory: '/admin/home/nouamane/.triton/cache/fxp9q8mDSfbdhYDc9Ebl-t3ex2yZXs8K7knt1681O0U/_layer_norm_fwd_1pass_kernel.cubin'
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+Traceback (most recent call last):
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    trainer.train(dataloader)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+trainer.train(dataloader)
+    trainer.train(dataloader)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return f(*args, **kwargs)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)      File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+
+outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    outputs = self.pipeline_engine.train_batch_iter(    
+outputs = self.pipeline_engine.train_batch_iter(    
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = model(**micro_batch)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+output = model(**micro_batch)    
+    output = model(**micro_batch)    outputs = self.pipeline_engine.train_batch_iter(  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+
+output = model(**micro_batch)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+        return self._call_impl(*args, **kwargs)return self._call_impl(*args, **kwargs)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    return self._call_impl(*args, **kwargs)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+        return self._call_impl(*args, **kwargs)return forward_call(*args, **kwargs)
+
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    return forward_call(*args, **kwargs)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    sharded_logits = self.model(
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+sharded_logits = self.model(
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+        return self._call_impl(*args, **kwargs)return self._call_impl(*args, **kwargs)
+
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+return self._call_impl(*args, **kwargs)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+        sharded_logits = self.model(return self._call_impl(*args, **kwargs)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+            return forward_call(*args, **kwargs)return forward_call(*args, **kwargs)return forward_call(*args, **kwargs)
+
+    
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+return forward_call(*args, **kwargs)  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+        return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+        hidden_encoder_states = encoder_block(**hidden_encoder_states)return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+        return self._call_impl(*args, **kwargs)return self._call_impl(*args, **kwargs)
+    
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+return self._call_impl(*args, **kwargs)return self._call_impl(*args, **kwargs)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+            return forward_call(*args, **kwargs)return forward_call(*args, **kwargs)    return forward_call(*args, **kwargs)
+
+return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = self.pp_block(**new_kwargs)    
+output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)        
+return self._call_impl(*args, **kwargs)return forward_call(*args, **kwargs)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    return self._call_impl(*args, **kwargs)return self._call_impl(*args, **kwargs)
+
+    return self._call_impl(*args, **kwargs)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+        return forward_call(*args, **kwargs)return forward_call(*args, **kwargs)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+        return forward_call(*args, **kwargs)return forward_call(*args, **kwargs)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+        hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)    
+hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)    
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+        hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+
+    return self._call_impl(*args, **kwargs)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+        return self._call_impl(*args, **kwargs)return forward_call(*args, **kwargs)
+
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+return forward_call(*args, **kwargs)  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+        hidden_states = self.down_proj(self.split_silu_mul(merged_states))hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return self._call_impl(*args, **kwargs)
+          File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+return self._call_impl(*args, **kwargs)return self._call_impl(*args, **kwargs)
+
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+return self._call_impl(*args, **kwargs)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+        return forward_call(*args, **kwargs)return forward_call(*args, **kwargs)
+
+        return forward_call(*args, **kwargs)return forward_call(*args, **kwargs)  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    out = F.linear(tensor, weight, bias)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 4 has a total capacty of 79.33 GiB of which 247.94 MiB is free. Including non-PyTorch memory, this process has 79.08 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 358.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return row_linear(
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+return row_linear(
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+        return row_linear(return row_linear(
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)    
+return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+    out = F.linear(tensor, weight, bias)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 5 has a total capacty of 79.33 GiB of which 183.94 MiB is free. Including non-PyTorch memory, this process has 79.14 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 518.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+        return super().apply(*args, **kwargs)  # type: ignore[misc]out = F.linear(tensor, weight, bias)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 7 has a total capacty of 79.33 GiB of which 415.94 MiB is free. Including non-PyTorch memory, this process has 78.91 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 486.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return super().apply(*args, **kwargs)  # type: ignore[misc]    
+return super().apply(*args, **kwargs)  # type: ignore[misc]  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    out = F.linear(tensor, weight, bias)
+    out = F.linear(tensor, weight, bias)
+torch.cuda    .out = F.linear(tensor, weight, bias)OutOfMemoryError
+: torch.cudaCUDA out of memory. Tried to allocate 512.00 MiB. GPU 3 has a total capacty of 79.33 GiB of which 447.94 MiB is free. Including non-PyTorch memory, this process has 78.88 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 134.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF.
+OutOfMemoryErrortorch.cuda: .CUDA out of memory. Tried to allocate 512.00 MiB. GPU 0 has a total capacty of 79.33 GiB of which 295.94 MiB is free. Including non-PyTorch memory, this process has 79.03 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 358.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONFOutOfMemoryError
+: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 1 has a total capacty of 79.33 GiB of which 215.94 MiB is free. Including non-PyTorch memory, this process has 79.11 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 486.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    out = F.linear(tensor, weight, bias)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 6 has a total capacty of 79.33 GiB of which 167.94 MiB is free. Including non-PyTorch memory, this process has 79.15 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 454.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    out = F.linear(tensor, weight, bias)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 2 has a total capacty of 79.33 GiB of which 199.94 MiB is free. Including non-PyTorch memory, this process has 79.12 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 422.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    trainer.train(dataloader)
+    trainer.train(dataloader)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    
+trainer.train(dataloader)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+        trainer.train(dataloader)  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+outputs = self.pipeline_engine.train_batch_iter(    
+
+outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    return self._call_impl(*args, **kwargs)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+        return self._call_impl(*args, **kwargs)return self._call_impl(*args, **kwargs)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)return forward_call(*args, **kwargs)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+Traceback (most recent call last):
+    return self._call_impl(*args, **kwargs)  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    sharded_logits = self.model(
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self._call_impl(*args, **kwargs)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+trainer.train(dataloader)
+    return forward_call(*args, **kwargs)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+return forward_call(*args, **kwargs)    
+return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    return forward_call(*args, **kwargs)
+    output = self.pp_block(**new_kwargs)  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+        outputs = self.pipeline_engine.train_batch_iter(return forward_call(*args, **kwargs)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+        return self._call_impl(*args, **kwargs)output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    return forward_call(*args, **kwargs)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+return self._call_impl(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    return forward_call(*args, **kwargs)return forward_call(*args, **kwargs)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    return forward_call(*args, **kwargs)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+        hidden_states = self.down_proj(self.split_silu_mul(merged_states))hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+    return self._call_impl(*args, **kwargs)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+        return forward_call(*args, **kwargs)
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    return self._call_impl(*args, **kwargs)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)return forward_call(*args, **kwargs)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    out = F.linear(tensor, weight, bias)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 2 has a total capacty of 79.33 GiB of which 311.94 MiB is free. Including non-PyTorch memory, this process has 79.01 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 230.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return row_linear(
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    out = F.linear(tensor, weight, bias)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 6 has a total capacty of 79.33 GiB of which 375.94 MiB is free. Including non-PyTorch memory, this process has 78.95 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 166.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+    return super().apply(*args, **kwargs)  # type: ignore[misc]  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    return forward_call(*args, **kwargs)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    out = F.linear(tensor, weight, bias)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 1 has a total capacty of 79.33 GiB of which 359.94 MiB is free. Including non-PyTorch memory, this process has 78.97 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 262.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    out = F.linear(tensor, weight, bias)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 4 has a total capacty of 79.33 GiB of which 167.94 MiB is free. Including non-PyTorch memory, this process has 79.15 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 358.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    out = F.linear(tensor, weight, bias)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 0 has a total capacty of 79.33 GiB of which 343.94 MiB is free. Including non-PyTorch memory, this process has 78.98 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 230.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    out = F.linear(tensor, weight, bias)
+    output = self.pp_block(**new_kwargs)
+torch.cuda.OutOfMemoryError  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 3 has a total capacty of 79.33 GiB of which 271.94 MiB is free. Including non-PyTorch memory, this process has 79.05 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 230.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    out = F.linear(tensor, weight, bias)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 5 has a total capacty of 79.33 GiB of which 423.94 MiB is free. Including non-PyTorch memory, this process has 78.90 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 198.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+trainer.train(dataloader)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+    return f(*args, **kwargs)  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+    outputs = self.pipeline_engine.train_batch_iter(  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = model(**micro_batch)
+    output = model(**micro_batch)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+            return self._call_impl(*args, **kwargs)return self._call_impl(*args, **kwargs)return self._call_impl(*args, **kwargs)
+
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return forward_call(*args, **kwargs)
+          File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+return forward_call(*args, **kwargs)return forward_call(*args, **kwargs)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    sharded_logits = self.model(    
+sharded_logits = self.model(
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+sharded_logits = self.model(  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    return self._call_impl(*args, **kwargs)    
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+        return self._call_impl(*args, **kwargs)return self._call_impl(*args, **kwargs)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)    
+return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+        return f(*args, **kwargs)return forward_call(*args, **kwargs)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    trainer.train(dataloader)
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+Traceback (most recent call last):
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+Traceback (most recent call last):
+Traceback (most recent call last):
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return self._call_impl(*args, **kwargs)
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)    
+output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+        return forward_call(*args, **kwargs)return self._call_impl(*args, **kwargs)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    trainer.train(dataloader)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)    
+return forward_call(*args, **kwargs)  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    trainer.train(dataloader)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+trainer.train(dataloader)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    trainer.train(dataloader)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+trainer.train(dataloader)
+    trainer.train(dataloader)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    trainer.train(dataloader)
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)    
+return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+        hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    outputs = self.pipeline_engine.train_batch_iter(    
+hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return f(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return f(*args, **kwargs)
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return self._call_impl(*args, **kwargs)
+return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return f(*args, **kwargs)
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return self._call_impl(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    return forward_call(*args, **kwargs)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+return forward_call(*args, **kwargs)    
+return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+        return forward_call(*args, **kwargs)return forward_call(*args, **kwargs)    
+
+return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+        return forward_call(*args, **kwargs)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return f(*args, **kwargs)
+output = model(**micro_batch)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    output = self.pp_block(**new_kwargs)
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+        output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    sharded_logits = self.model(
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)    
+outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return self._call_impl(*args, **kwargs)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self._call_impl(*args, **kwargs)
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+    return forward_call(*args, **kwargs)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    outputs = self.pipeline_engine.train_batch_iter(
+        return forward_call(*args, **kwargs)return forward_call(*args, **kwargs)
+
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+return forward_call(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+        output = model(**micro_batch)return forward_call(*args, **kwargs)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+        output = self.pp_block(**new_kwargs)hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+            return row_linear(return self._call_impl(*args, **kwargs)return row_linear(
+
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    return self._call_impl(*args, **kwargs)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+hidden_encoder_states = encoder_block(**hidden_encoder_states)
+
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+        return row_linear(sharded_logits = self.model(
+
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return forward_call(*args, **kwargs)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return forward_call(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = model(**micro_batch)
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    return forward_call(*args, **kwargs)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+        sharded_logits = self.model(sharded_logits = self.model(
+
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    out = F.linear(tensor, weight, bias)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 3 has a total capacty of 79.33 GiB of which 271.94 MiB is free. Including non-PyTorch memory, this process has 79.05 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 230.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]    
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    return self._call_impl(*args, **kwargs)
+    out = F.linear(tensor, weight, bias)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 5 has a total capacty of 79.33 GiB of which 455.94 MiB is free. Including non-PyTorch memory, this process has 78.87 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 166.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    out = F.linear(tensor, weight, bias)
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+torch.cuda    .OutOfMemoryErrorhidden_states = self.down_proj(self.split_silu_mul(merged_states)): 
+CUDA out of memory. Tried to allocate 512.00 MiB. GPU 4 has a total capacty of 79.33 GiB of which 39.94 MiB is free. Including non-PyTorch memory, this process has 79.28 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 486.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+        trainer.train(dataloader)return forward_call(*args, **kwargs)
+
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    out = F.linear(tensor, weight, bias)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 0 has a total capacty of 79.33 GiB of which 343.94 MiB is free. Including non-PyTorch memory, this process has 78.98 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 230.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+    return self._call_impl(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return forward_call(*args, **kwargs)
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+return forward_call(*args, **kwargs)    
+return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    return forward_call(*args, **kwargs)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    out = F.linear(tensor, weight, bias)
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 7 has a total capacty of 79.33 GiB of which 319.94 MiB is free. Including non-PyTorch memory, this process has 79.01 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 422.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+        return forward_call(*args, **kwargs)output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    trainer.train(dataloader)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+trainer.train(dataloader)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+trainer.train(dataloader)    trainer.train(dataloader)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+    sharded_logits = self.model(  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    return self._call_impl(*args, **kwargs)      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = model(**micro_batch)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+return row_linear(
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = model(**micro_batch)
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+    return forward_call(*args, **kwargs)  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+
+        return forward_call(*args, **kwargs)hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+        return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]return forward_call(*args, **kwargs)
+
+    out = F.linear(tensor, weight, bias)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 5 has a total capacty of 79.33 GiB of which 119.94 MiB is free. Including non-PyTorch memory, this process has 79.20 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 422.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return f(*args, **kwargs)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return self._call_impl(*args, **kwargs)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+return self._call_impl(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+    return f(*args, **kwargs)
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return forward_call(*args, **kwargs)
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+        return row_linear(out = F.linear(tensor, weight, bias)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 6 has a total capacty of 79.33 GiB of which 407.94 MiB is free. Including non-PyTorch memory, this process has 78.92 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 134.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    output = self.pp_block(**new_kwargs)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+output = self.pp_block(**new_kwargs)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+    return row_linear(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+sharded_logits = self.model(
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    out = F.linear(tensor, weight, bias)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 6 has a total capacty of 79.33 GiB of which 231.94 MiB is free. Including non-PyTorch memory, this process has 79.09 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 390.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+        outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)outputs = self.pipeline_engine.train_batch_iter(
+
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs = self.pipeline_engine.train_batch_iter(
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return self._call_impl(*args, **kwargs)
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    return self._call_impl(*args, **kwargs)
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    return self._call_impl(*args, **kwargs)
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+        return self._call_impl(*args, **kwargs)return self._call_impl(*args, **kwargs)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    return forward_call(*args, **kwargs)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    output = model(**micro_batch)
+    return self._call_impl(*args, **kwargs)
+return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+        return forward_call(*args, **kwargs)
+return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    return self._call_impl(*args, **kwargs)
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+        output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+return self._call_impl(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    out = F.linear(tensor, weight, bias)
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 2 has a total capacty of 79.33 GiB of which 199.94 MiB is free. Including non-PyTorch memory, this process has 79.12 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 422.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+        hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+
+hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+        return self._call_impl(*args, **kwargs)return self._call_impl(*args, **kwargs)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+        hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+        return forward_call(*args, **kwargs)
+return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    out = F.linear(tensor, weight, bias)
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    torch.cudareturn forward_call(*args, **kwargs).
+OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 2 has a total capacty of 79.33 GiB of which 247.94 MiB is free. Including non-PyTorch memory, this process has 79.08 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 294.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+
+    return self._call_impl(*args, **kwargs)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    return forward_call(*args, **kwargs)
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)    
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    return forward_call(*args, **kwargs)
+          File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    return self._call_impl(*args, **kwargs)
+return forward_call(*args, **kwargs)return forward_call(*args, **kwargs)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+        hidden_states = self.down_proj(self.split_silu_mul(merged_states))hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    sharded_logits = self.model(
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    sharded_logits = self.model(
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    out = F.linear(tensor, weight, bias)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 6 has a total capacty of 79.33 GiB of which 215.94 MiB is free. Including non-PyTorch memory, this process has 79.11 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 486.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.pp_block(**new_kwargs)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+        return forward_call(*args, **kwargs)return forward_call(*args, **kwargs)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    output = self.pp_block(**new_kwargs)
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+        return self._call_impl(*args, **kwargs)return self._call_impl(*args, **kwargs)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    sharded_logits = self.model(
+    return forward_call(*args, **kwargs)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+    return forward_call(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+        return forward_call(*args, **kwargs)return forward_call(*args, **kwargs)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+        return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]hidden_encoder_states = encoder_block(**hidden_encoder_states)
+
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+    return row_linear(
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+Traceback (most recent call last):
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+    return self._call_impl(*args, **kwargs)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+    return self._call_impl(*args, **kwargs)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    out = F.linear(tensor, weight, bias)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 7 has a total capacty of 79.33 GiB of which 431.94 MiB is free. Including non-PyTorch memory, this process has 78.90 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 390.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    trainer.train(dataloader)
+    return self._call_impl(*args, **kwargs)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return row_linear(
+hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    return self._call_impl(*args, **kwargs)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = self.pp_block(**new_kwargs)
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    output = self.pp_block(**new_kwargs)
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    output = self.pp_block(**new_kwargs)
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+        return forward_call(*args, **kwargs)return self._call_impl(*args, **kwargs)
+
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    output = self.pp_block(**new_kwargs)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    out = F.linear(tensor, weight, bias)
+            return self._call_impl(*args, **kwargs)return self._call_impl(*args, **kwargs)return self._call_impl(*args, **kwargs)
+
+
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 1 has a total capacty of 79.33 GiB of which 71.94 MiB is free. Including non-PyTorch memory, this process has 79.25 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 550.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+return forward_call(*args, **kwargs)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+        return forward_call(*args, **kwargs)
+return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    return row_linear(
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+        return forward_call(*args, **kwargs)return forward_call(*args, **kwargs)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    out = F.linear(tensor, weight, bias)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 3 has a total capacty of 79.33 GiB of which 239.94 MiB is free. Including non-PyTorch memory, this process has 79.08 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 262.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    return self._call_impl(*args, **kwargs)
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 725, in _core_forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = self.pp_block(**new_kwargs)
+    return row_linear(  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+
+    return row_linear(
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 725, in _core_forward
+    out = F.linear(tensor, weight, bias)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return forward_call(*args, **kwargs)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 0 has a total capacty of 79.33 GiB of which 215.94 MiB is free. Including non-PyTorch memory, this process has 79.11 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 518.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+        hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)output = self.attn(hidden_states=hidden_states, sequence_mask=sequence_mask)
+
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 725, in _core_forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 725, in _core_forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    out = F.linear(tensor, weight, bias)
+    output = self.attn(hidden_states=hidden_states, sequence_mask=sequence_mask)    
+hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 2 has a total capacty of 79.33 GiB of which 471.94 MiB is free. Including non-PyTorch memory, this process has 78.86 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 230.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 725, in _core_forward
+    output = self.attn(hidden_states=hidden_states, sequence_mask=sequence_mask)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.attn(hidden_states=hidden_states, sequence_mask=sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 725, in _core_forward
+    return self._call_impl(*args, **kwargs)
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = self.attn(hidden_states=hidden_states, sequence_mask=sequence_mask)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 690, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return row_linear(
+    output = self.attn(hidden_states=hidden_states, sequence_mask=sequence_mask)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 690, in forward
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 690, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 690, in forward
+    output = model(**micro_batch)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 690, in forward
+    output = self.o_proj(attention_output)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return row_linear(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    out = F.linear(tensor, weight, bias)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 7 has a total capacty of 79.33 GiB of which 271.94 MiB is free. Including non-PyTorch memory, this process has 79.05 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 550.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    output = self.o_proj(attention_output)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+        output = self.o_proj(attention_output)        return self._call_impl(*args, **kwargs)
+    out = F.linear(tensor, weight, bias)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 1 has a total capacty of 79.33 GiB of which 167.94 MiB is free. Including non-PyTorch memory, this process has 79.15 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 454.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+output = self.o_proj(attention_output)
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 690, in forward
+    return self._call_impl(*args, **kwargs)    
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = self.o_proj(attention_output)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    out = F.linear(tensor, weight, bias)
+    return forward_call(*args, **kwargs)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 1 has a total capacty of 79.33 GiB of which 119.94 MiB is free. Including non-PyTorch memory, this process has 79.20 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 422.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    out = F.linear(tensor, weight, bias)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 5 has a total capacty of 79.33 GiB of which 7.94 MiB is free. Including non-PyTorch memory, this process has 79.31 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 614.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    output = self.o_proj(attention_output)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+        return forward_call(*args, **kwargs)return forward_call(*args, **kwargs)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    out = F.linear(tensor, weight, bias)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 3 has a total capacty of 79.33 GiB of which 31.94 MiB is free. Including non-PyTorch memory, this process has 79.29 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 390.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+    out = F.linear(tensor, weight, bias)
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 4 has a total capacty of 79.33 GiB of which 327.94 MiB is free. Including non-PyTorch memory, this process has 79.00 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 358.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    out = F.linear(tensor, weight, bias)
+    torch.cudareturn _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode).
+OutOfMemoryError:   File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+CUDA out of memory. Tried to allocate 512.00 MiB. GPU 1 has a total capacty of 79.33 GiB of which 405.94 MiB is free. Including non-PyTorch memory, this process has 78.92 GiB memory in use. Of the allocated memory 70.35 GiB is allocated by PyTorch, and 230.12 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    return row_linear(
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    out = F.linear(tensor, weight, bias)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 2 has a total capacty of 79.33 GiB of which 341.94 MiB is free. Including non-PyTorch memory, this process has 78.98 GiB memory in use. Of the allocated memory 70.35 GiB is allocated by PyTorch, and 294.12 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    out = F.linear(tensor, weight, bias)
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 4 has a total capacty of 79.33 GiB of which 421.94 MiB is free. Including non-PyTorch memory, this process has 78.91 GiB memory in use. Of the allocated memory 70.35 GiB is allocated by PyTorch, and 198.12 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    out = F.linear(tensor, weight, bias)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 3 has a total capacty of 79.33 GiB of which 317.94 MiB is free. Including non-PyTorch memory, this process has 79.01 GiB memory in use. Of the allocated memory 70.35 GiB is allocated by PyTorch, and 198.12 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return self._call_impl(*args, **kwargs)
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+    out = F.linear(tensor, weight, bias)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 6 has a total capacty of 79.33 GiB of which 405.94 MiB is free. Including non-PyTorch memory, this process has 78.92 GiB memory in use. Of the allocated memory 70.35 GiB is allocated by PyTorch, and 230.12 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    out = F.linear(tensor, weight, bias)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 5 has a total capacty of 79.33 GiB of which 501.94 MiB is free. Including non-PyTorch memory, this process has 78.83 GiB memory in use. Of the allocated memory 70.35 GiB is allocated by PyTorch, and 134.12 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    out = F.linear(tensor, weight, bias)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 4 has a total capacty of 79.33 GiB of which 247.94 MiB is free. Including non-PyTorch memory, this process has 79.08 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 358.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    out = F.linear(tensor, weight, bias)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 0 has a total capacty of 79.33 GiB of which 295.94 MiB is free. Including non-PyTorch memory, this process has 79.03 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 358.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    trainer.train(dataloader)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+trainer.train(dataloader)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+outputs = self.pipeline_engine.train_batch_iter(
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+        outputs = self.pipeline_engine.train_batch_iter(outputs = self.pipeline_engine.train_batch_iter(
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)    
+output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)    
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = model(**micro_batch)    
+    output = model(**micro_batch)output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+        return self._call_impl(*args, **kwargs)return self._call_impl(*args, **kwargs)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+        return forward_call(*args, **kwargs)
+return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+        return forward_call(*args, **kwargs)
+return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    sharded_logits = self.model(
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+sharded_logits = self.model(
+        sharded_logits = self.model(return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return f(*args, **kwargs)    
+    trainer.train(dataloader)  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+trainer.train(dataloader)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+trainer.train(dataloader)    
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+trainer.train(dataloader)    trainer.train(dataloader)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+        return self._call_impl(*args, **kwargs)return self._call_impl(*args, **kwargs)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)    
+return self._call_impl(*args, **kwargs)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+        return forward_call(*args, **kwargs)return forward_call(*args, **kwargs)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    trainer.train(dataloader)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return f(*args, **kwargs)
+    return f(*args, **kwargs)  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return f(*args, **kwargs)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+return f(*args, **kwargs)
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]    
+return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+hidden_encoder_states = encoder_block(**hidden_encoder_states)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return f(*args, **kwargs)
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+return self._call_impl(*args, **kwargs)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return forward_call(*args, **kwargs)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    return self._call_impl(*args, **kwargs)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    outputs = self.pipeline_engine.train_batch_iter(
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    output = self.pp_block(**new_kwargs)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)    
+return self._call_impl(*args, **kwargs)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+        return forward_call(*args, **kwargs)    return self._call_impl(*args, **kwargs)
+return forward_call(*args, **kwargs)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    output = model(**micro_batch)
+    return forward_call(*args, **kwargs)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+        output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+        return self._call_impl(*args, **kwargs)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+return self._call_impl(*args, **kwargs)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+        hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+        return forward_call(*args, **kwargs)return forward_call(*args, **kwargs)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+        hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+        hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+        return self._call_impl(*args, **kwargs)return self._call_impl(*args, **kwargs)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+        return self._call_impl(*args, **kwargs)return forward_call(*args, **kwargs)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+Traceback (most recent call last):
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    sharded_logits = self.model(
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+sharded_logits = self.model(    
+sharded_logits = self.model(
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+        hidden_states = self.down_proj(self.split_silu_mul(merged_states))return forward_call(*args, **kwargs)
+
+    return self._call_impl(*args, **kwargs)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    return forward_call(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+        return forward_call(*args, **kwargs)return self._call_impl(*args, **kwargs)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+        return forward_call(*args, **kwargs)return forward_call(*args, **kwargs)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    out = F.linear(tensor, weight, bias)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 2 has a total capacty of 79.33 GiB of which 135.94 MiB is free. Including non-PyTorch memory, this process has 79.19 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 486.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+        return row_linear(return super().apply(*args, **kwargs)  # type: ignore[misc]
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    out = F.linear(tensor, weight, bias)
+    output = self.pp_block(**new_kwargs)
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 3 has a total capacty of 79.33 GiB of which 159.94 MiB is free. Including non-PyTorch memory, this process has 79.16 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 422.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)    
+return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+
+hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    out = F.linear(tensor, weight, bias)
+out = F.linear(tensor, weight, bias)
+torch.cudatorch.cuda..OutOfMemoryErrorOutOfMemoryError: : CUDA out of memory. Tried to allocate 512.00 MiB. GPU 1 has a total capacty of 79.33 GiB of which 311.94 MiB is free. Including non-PyTorch memory, this process has 79.01 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 390.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONFCUDA out of memory. Tried to allocate 512.00 MiB. GPU 5 has a total capacty of 79.33 GiB of which 279.94 MiB is free. Including non-PyTorch memory, this process has 79.04 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 422.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+    output = model(**micro_batch)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    out = F.linear(tensor, weight, bias)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 4 has a total capacty of 79.33 GiB of which 151.94 MiB is free. Including non-PyTorch memory, this process has 79.17 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 454.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    return self._call_impl(*args, **kwargs)    
+return self._call_impl(*args, **kwargs)    
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+return self._call_impl(*args, **kwargs)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    out = F.linear(tensor, weight, bias)
+    return self._call_impl(*args, **kwargs)return forward_call(*args, **kwargs)    
+
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 0 has a total capacty of 79.33 GiB of which 391.94 MiB is free. Including non-PyTorch memory, this process has 78.94 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 262.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+            return forward_call(*args, **kwargs)return forward_call(*args, **kwargs)return forward_call(*args, **kwargs)
+    
+
+return forward_call(*args, **kwargs)  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    return self._call_impl(*args, **kwargs)    
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+        output = self.pp_block(**new_kwargs)return forward_call(*args, **kwargs)        
+    output = self.pp_block(**new_kwargs)
+return forward_call(*args, **kwargs)    output = self.pp_block(**new_kwargs)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    return self._call_impl(*args, **kwargs)        
+    out = F.linear(tensor, weight, bias)
+return self._call_impl(*args, **kwargs)return self._call_impl(*args, **kwargs)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 6 has a total capacty of 79.33 GiB of which 167.94 MiB is free. Including non-PyTorch memory, this process has 79.15 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 454.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+        hidden_encoder_states = encoder_block(**hidden_encoder_states)
+return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    return self._call_impl(*args, **kwargs)    
+    return self._call_impl(*args, **kwargs)return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+            hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]return forward_call(*args, **kwargs)    
+
+
+hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+return self._call_impl(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    return self._call_impl(*args, **kwargs)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+return self._call_impl(*args, **kwargs)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+        hidden_states = self.down_proj(self.split_silu_mul(merged_states))hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+        return self._call_impl(*args, **kwargs)return self._call_impl(*args, **kwargs)
+
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+return self._call_impl(*args, **kwargs)    
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    return forward_call(*args, **kwargs)    
+return forward_call(*args, **kwargs)  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    out = F.linear(tensor, weight, bias)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 6 has a total capacty of 79.33 GiB of which 455.94 MiB is free. Including non-PyTorch memory, this process has 78.87 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 166.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return row_linear(    
+return row_linear(
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    return row_linear(  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+return self._call_impl(*args, **kwargs)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    out = F.linear(tensor, weight, bias)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 4 has a total capacty of 79.33 GiB of which 375.94 MiB is free. Including non-PyTorch memory, this process has 78.95 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 230.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF    
+return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+    return super().apply(*args, **kwargs)  # type: ignore[misc]  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    out = F.linear(tensor, weight, bias)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 5 has a total capacty of 79.33 GiB of which 247.94 MiB is free. Including non-PyTorch memory, this process has 79.08 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 294.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    out = F.linear(tensor, weight, bias)
+    out = F.linear(tensor, weight, bias)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 0 has a total capacty of 79.33 GiB of which 487.94 MiB is free. Including non-PyTorch memory, this process has 78.84 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 166.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 2 has a total capacty of 79.33 GiB of which 263.94 MiB is free. Including non-PyTorch memory, this process has 79.06 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 358.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    out = F.linear(tensor, weight, bias)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 3 has a total capacty of 79.33 GiB of which 191.94 MiB is free. Including non-PyTorch memory, this process has 79.13 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 230.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    out = F.linear(tensor, weight, bias)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 1 has a total capacty of 79.33 GiB of which 215.94 MiB is free. Including non-PyTorch memory, this process has 79.11 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 326.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 469, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    out = differentiable_reduce_scatter_sum(out, group=group)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/distributed_differentiable_primitives.py", line 146, in differentiable_reduce_scatter_sum
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    return DifferentiableReduceScatterSum.apply(tensor, group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+        hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/distributed_differentiable_primitives.py", line 112, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    sharded_tensor = torch.empty(
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 64.00 MiB. GPU 7 has a total capacty of 79.33 GiB of which 63.94 MiB is free. Including non-PyTorch memory, this process has 79.26 GiB memory in use. Of the allocated memory 70.16 GiB is allocated by PyTorch, and 166.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    return row_linear(
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+    return super().apply(*args, **kwargs)  # type: ignore[misc]  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    out = F.linear(tensor, weight, bias)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 6 has a total capacty of 79.33 GiB of which 439.94 MiB is free. Including non-PyTorch memory, this process has 78.89 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 262.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+Traceback (most recent call last):
+    out = F.linear(tensor, weight, bias)
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 3 has a total capacty of 79.33 GiB of which 111.94 MiB is free. Including non-PyTorch memory, this process has 79.21 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 390.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+Traceback (most recent call last):
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    return self._call_impl(*args, **kwargs)
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return f(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return forward_call(*args, **kwargs)
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)    
+return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    out = F.linear(tensor, weight, bias)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 2 has a total capacty of 79.33 GiB of which 375.94 MiB is free. Including non-PyTorch memory, this process has 78.95 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 326.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return self._call_impl(*args, **kwargs)
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    output = self.pp_block(**new_kwargs)
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    sharded_logits = self.model(
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    return forward_call(*args, **kwargs)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return self._call_impl(*args, **kwargs)
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    output = self.pp_block(**new_kwargs)
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    return f(*args, **kwargs)    
+sharded_logits = self.model(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    trainer.train(dataloader)
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    out = F.linear(tensor, weight, bias)
+    return forward_call(*args, **kwargs)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 1 has a total capacty of 79.33 GiB of which 263.94 MiB is free. Including non-PyTorch memory, this process has 79.06 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 358.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+        return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)outputs = self.pipeline_engine.train_batch_iter(
+
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    trainer.train(dataloader)
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    out = F.linear(tensor, weight, bias)
+    output = model(**micro_batch)torch.cuda
+.OutOfMemoryError  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 7 has a total capacty of 79.33 GiB of which 463.94 MiB is free. Including non-PyTorch memory, this process has 78.87 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 358.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return forward_call(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    out = F.linear(tensor, weight, bias)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 0 has a total capacty of 79.33 GiB of which 375.94 MiB is free. Including non-PyTorch memory, this process has 78.95 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 358.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+        output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)return row_linear(
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+        return self._call_impl(*args, **kwargs)return self._call_impl(*args, **kwargs)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 469, in forward
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return forward_call(*args, **kwargs)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return forward_call(*args, **kwargs)
+    out = differentiable_reduce_scatter_sum(out, group=group)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    out = F.linear(tensor, weight, bias)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 4 has a total capacty of 79.33 GiB of which 199.94 MiB is free. Including non-PyTorch memory, this process has 79.12 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 486.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/distributed_differentiable_primitives.py", line 146, in differentiable_reduce_scatter_sum
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    sharded_logits = self.model(
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+hidden_encoder_states = encoder_block(**hidden_encoder_states)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return f(*args, **kwargs)  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return DifferentiableReduceScatterSum.apply(tensor, group)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs = self.pipeline_engine.train_batch_iter(
+    return self._call_impl(*args, **kwargs)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/distributed_differentiable_primitives.py", line 112, in forward
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    return row_linear(
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self._call_impl(*args, **kwargs)
+    sharded_tensor = torch.empty(
+torch.cuda.OutOfMemoryError    : CUDA out of memory. Tried to allocate 64.00 MiB. GPU 6 has a total capacty of 79.33 GiB of which 55.94 MiB is free. Including non-PyTorch memory, this process has 79.26 GiB memory in use. Of the allocated memory 70.16 GiB is allocated by PyTorch, and 134.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONFreturn self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = model(**micro_batch)
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+    output = model(**micro_batch)
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)    
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = self.pp_block(**new_kwargs)
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    return forward_call(*args, **kwargs)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+return forward_call(*args, **kwargs)
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+Traceback (most recent call last):
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    sharded_logits = self.model(
+    out = F.linear(tensor, weight, bias)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 5 has a total capacty of 79.33 GiB of which 103.94 MiB is free. Including non-PyTorch memory, this process has 79.22 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 518.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    return f(*args, **kwargs)
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    trainer.train(dataloader)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return forward_call(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return forward_call(*args, **kwargs)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return forward_call(*args, **kwargs)
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return self._call_impl(*args, **kwargs)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+        return forward_call(*args, **kwargs)return forward_call(*args, **kwargs)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    return self._call_impl(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.pp_block(**new_kwargs)
+    return row_linear(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    return self._call_impl(*args, **kwargs)
+          File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+return self._call_impl(*args, **kwargs)return forward_call(*args, **kwargs)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+        outputs = self.pipeline_engine.train_batch_iter(return super().apply(*args, **kwargs)  # type: ignore[misc]
+
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    out = F.linear(tensor, weight, bias)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 7 has a total capacty of 79.33 GiB of which 367.94 MiB is free. Including non-PyTorch memory, this process has 78.96 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 454.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    out = F.linear(tensor, weight, bias)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 1 has a total capacty of 79.33 GiB of which 391.94 MiB is free. Including non-PyTorch memory, this process has 78.94 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 230.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = model(**micro_batch)
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+        return super().apply(*args, **kwargs)  # type: ignore[misc]hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    out = F.linear(tensor, weight, bias)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 0 has a total capacty of 79.33 GiB of which 359.94 MiB is free. Including non-PyTorch memory, this process has 78.97 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 294.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    return f(*args, **kwargs)
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    return self._call_impl(*args, **kwargs)
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    out = F.linear(tensor, weight, bias)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 2 has a total capacty of 79.33 GiB of which 215.94 MiB is free. Including non-PyTorch memory, this process has 79.11 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 486.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+        hidden_states = self.down_proj(self.split_silu_mul(merged_states))outputs = self.pipeline_engine.train_batch_iter(
+
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    output = model(**micro_batch)
+    out = F.linear(tensor, weight, bias)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 3 has a total capacty of 79.33 GiB of which 271.94 MiB is free. Including non-PyTorch memory, this process has 79.05 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 230.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    output = model(**micro_batch)
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    return self._call_impl(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return row_linear(
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    out = F.linear(tensor, weight, bias)
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 5 has a total capacty of 79.33 GiB of which 263.94 MiB is free. Including non-PyTorch memory, this process has 79.06 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 358.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+        return row_linear(output = model(**micro_batch)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    trainer.train(dataloader)
+    sharded_logits = self.model(    
+trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    out = F.linear(tensor, weight, bias)
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 1 has a total capacty of 79.33 GiB of which 183.94 MiB is free. Including non-PyTorch memory, this process has 79.14 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 358.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    return forward_call(*args, **kwargs)
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    out = F.linear(tensor, weight, bias)
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 5 has a total capacty of 79.33 GiB of which 23.94 MiB is free. Including non-PyTorch memory, this process has 79.29 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 518.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+    return forward_call(*args, **kwargs)  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+    return self._call_impl(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = model(**micro_batch)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    out = F.linear(tensor, weight, bias)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 4 has a total capacty of 79.33 GiB of which 247.94 MiB is free. Including non-PyTorch memory, this process has 79.08 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 358.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    return self._call_impl(*args, **kwargs)    
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    return forward_call(*args, **kwargs)
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return forward_call(*args, **kwargs)
+    return forward_call(*args, **kwargs)  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return forward_call(*args, **kwargs)
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    sharded_logits = self.model(
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return self._call_impl(*args, **kwargs)
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.pp_block(**new_kwargs)
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return forward_call(*args, **kwargs)
+        return self._call_impl(*args, **kwargs)return self._call_impl(*args, **kwargs)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    return row_linear(
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    return forward_call(*args, **kwargs)
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+    out = F.linear(tensor, weight, bias)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 4 has a total capacty of 79.33 GiB of which 343.94 MiB is free. Including non-PyTorch memory, this process has 78.98 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 262.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+Traceback (most recent call last):
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    out = F.linear(tensor, weight, bias)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 0 has a total capacty of 79.33 GiB of which 343.94 MiB is free. Including non-PyTorch memory, this process has 78.98 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 390.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)    
+output = model(**micro_batch)
+    out = F.linear(tensor, weight, bias)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 4 has a total capacty of 79.33 GiB of which 295.94 MiB is free. Including non-PyTorch memory, this process has 79.03 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 390.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+        trainer.train(dataloader)output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+        return forward_call(*args, **kwargs)return forward_call(*args, **kwargs)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+    trainer.train(dataloader)
+    return self._call_impl(*args, **kwargs)
+    output = self.pp_block(**new_kwargs)  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+        sharded_logits = self.model(output = self.pp_block(**new_kwargs)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return forward_call(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+        return self._call_impl(*args, **kwargs)return self._call_impl(*args, **kwargs)
+
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    return forward_call(*args, **kwargs)
+    return forward_call(*args, **kwargs)  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+        return forward_call(*args, **kwargs)
+return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    return forward_call(*args, **kwargs)
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return f(*args, **kwargs)
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    out = F.linear(tensor, weight, bias)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 7 has a total capacty of 79.33 GiB of which 255.94 MiB is free. Including non-PyTorch memory, this process has 79.07 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 486.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+        outputs = self.pipeline_engine.train_batch_iter(return self._call_impl(*args, **kwargs)
+
+        hidden_encoder_states = encoder_block(**hidden_encoder_states)return self._call_impl(*args, **kwargs)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    trainer.train(dataloader)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return forward_call(*args, **kwargs)    
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    return forward_call(*args, **kwargs)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+    return row_linear(
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+        hidden_states = self.down_proj(self.split_silu_mul(merged_states))return f(*args, **kwargs)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+        outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)output = self.pp_block(**new_kwargs)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+return self._call_impl(*args, **kwargs)
+    output = model(**micro_batch)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return row_linear(
+    return self._call_impl(*args, **kwargs)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    return forward_call(*args, **kwargs)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    out = F.linear(tensor, weight, bias)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 6 has a total capacty of 79.33 GiB of which 359.94 MiB is free. Including non-PyTorch memory, this process has 78.97 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 262.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    outputs = self.pipeline_engine.train_batch_iter(
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    outputs = self.pipeline_engine.train_batch_iter(
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return row_linear(
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    trainer.train(dataloader)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+        return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    output = model(**micro_batch)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    out = F.linear(tensor, weight, bias)
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 5 has a total capacty of 79.33 GiB of which 343.94 MiB is free. Including non-PyTorch memory, this process has 78.98 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 198.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    outputs = self.pipeline_engine.train_batch_iter(
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    out = F.linear(tensor, weight, bias)
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 7 has a total capacty of 79.33 GiB of which 191.94 MiB is free. Including non-PyTorch memory, this process has 79.13 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 550.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    return forward_call(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    sharded_logits = self.model(
+    out = F.linear(tensor, weight, bias)
+    out = F.linear(tensor, weight, bias)
+torch.cudatorch.cuda..OutOfMemoryErrorOutOfMemoryError: : CUDA out of memory. Tried to allocate 512.00 MiB. GPU 2 has a total capacty of 79.33 GiB of which 135.94 MiB is free. Including non-PyTorch memory, this process has 79.19 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 486.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONFCUDA out of memory. Tried to allocate 512.00 MiB. GPU 7 has a total capacty of 79.33 GiB of which 415.94 MiB is free. Including non-PyTorch memory, this process has 78.91 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 486.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    out = F.linear(tensor, weight, bias)    
+return forward_call(*args, **kwargs)
+torch.cuda  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 4 has a total capacty of 79.33 GiB of which 311.94 MiB is free. Including non-PyTorch memory, this process has 79.01 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 294.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    return forward_call(*args, **kwargs)
+    output = model(**micro_batch)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    sharded_logits = self.model(
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+    return self._call_impl(*args, **kwargs)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+        return row_linear(sharded_logits = self.model(
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = model(**micro_batch)
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self._call_impl(*args, **kwargs)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+        return row_linear(return self._call_impl(*args, **kwargs)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    out = F.linear(tensor, weight, bias)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 1 has a total capacty of 79.33 GiB of which 471.94 MiB is free. Including non-PyTorch memory, this process has 78.86 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 230.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    out = F.linear(tensor, weight, bias)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 2 has a total capacty of 79.33 GiB of which 135.94 MiB is free. Including non-PyTorch memory, this process has 79.19 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 486.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+        return forward_call(*args, **kwargs)return self._call_impl(*args, **kwargs)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    out = F.linear(tensor, weight, bias)
+    return forward_call(*args, **kwargs)
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 1 has a total capacty of 79.33 GiB of which 279.94 MiB is free. Including non-PyTorch memory, this process has 79.04 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 262.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+        return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)return self._call_impl(*args, **kwargs)
+
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    return forward_call(*args, **kwargs)    
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    out = F.linear(tensor, weight, bias)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 3 has a total capacty of 79.33 GiB of which 191.94 MiB is free. Including non-PyTorch memory, this process has 79.13 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 230.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)    
+hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    out = F.linear(tensor, weight, bias)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 3 has a total capacty of 79.33 GiB of which 127.94 MiB is free. Including non-PyTorch memory, this process has 79.19 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 294.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return self._call_impl(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return forward_call(*args, **kwargs)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+        hidden_states = self.down_proj(self.split_silu_mul(merged_states))hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+        return forward_call(*args, **kwargs)return super().apply(*args, **kwargs)  # type: ignore[misc]
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+        out = F.linear(tensor, weight, bias)return row_linear(    
+
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+torch.cuda.OutOfMemoryError  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 3 has a total capacty of 79.33 GiB of which 191.94 MiB is free. Including non-PyTorch memory, this process has 79.13 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 390.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    out = F.linear(tensor, weight, bias)
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 0 has a total capacty of 79.33 GiB of which 71.94 MiB is free. Including non-PyTorch memory, this process has 79.25 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 582.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    out = F.linear(tensor, weight, bias)
+    return self._call_impl(*args, **kwargs)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 5 has a total capacty of 79.33 GiB of which 375.94 MiB is free. Including non-PyTorch memory, this process has 78.95 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 326.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    out = F.linear(tensor, weight, bias)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 2 has a total capacty of 79.33 GiB of which 423.94 MiB is free. Including non-PyTorch memory, this process has 78.90 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 198.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    out = F.linear(tensor, weight, bias)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 6 has a total capacty of 79.33 GiB of which 295.94 MiB is free. Including non-PyTorch memory, this process has 79.03 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 326.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    return self._call_impl(*args, **kwargs)
+    out = F.linear(tensor, weight, bias)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 0 has a total capacty of 79.33 GiB of which 167.94 MiB is free. Including non-PyTorch memory, this process has 79.15 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 486.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    out = F.linear(tensor, weight, bias)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 6 has a total capacty of 79.33 GiB of which 327.94 MiB is free. Including non-PyTorch memory, this process has 79.00 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 294.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    outputs = self.pipeline_engine.train_batch_iter(
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    out = F.linear(tensor, weight, bias)
+    output = model(**micro_batch)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 5 has a total capacty of 79.33 GiB of which 103.94 MiB is free. Including non-PyTorch memory, this process has 79.22 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 518.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+        return self._call_impl(*args, **kwargs)trainer.train(dataloader)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+        output = self.pp_block(**new_kwargs)outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    output = model(**micro_batch)
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return self._call_impl(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+        output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)return forward_call(*args, **kwargs)
+    sharded_logits = self.model(
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    sharded_logits = self.model(
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return self._call_impl(*args, **kwargs)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return forward_call(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    out = F.linear(tensor, weight, bias)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 2 has a total capacty of 79.33 GiB of which 119.94 MiB is free. Including non-PyTorch memory, this process has 79.20 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 582.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    output = self.pp_block(**new_kwargs)
+    return self._call_impl(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    out = F.linear(tensor, weight, bias)
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 1 has a total capacty of 79.33 GiB of which 231.94 MiB is free. Including non-PyTorch memory, this process has 79.09 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 390.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    return self._call_impl(*args, **kwargs)
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return self._call_impl(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    return forward_call(*args, **kwargs)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    return row_linear(
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+    sharded_logits = self.model(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    out = F.linear(tensor, weight, bias)
+    return self._call_impl(*args, **kwargs)
+torch.cuda.  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 3 has a total capacty of 79.33 GiB of which 47.94 MiB is free. Including non-PyTorch memory, this process has 79.27 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 454.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    return row_linear(
+    out = F.linear(tensor, weight, bias)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 6 has a total capacty of 79.33 GiB of which 87.94 MiB is free. Including non-PyTorch memory, this process has 79.23 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 614.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+        return self._call_impl(*args, **kwargs)return super().apply(*args, **kwargs)  # type: ignore[misc]
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    out = F.linear(tensor, weight, bias)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 5 has a total capacty of 79.33 GiB of which 391.94 MiB is free. Including non-PyTorch memory, this process has 78.94 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 230.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+        trainer.train(dataloader)hidden_encoder_states = encoder_block(**hidden_encoder_states)
+
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+        return self._call_impl(*args, **kwargs)trainer.train(dataloader)
+
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+        return f(*args, **kwargs)return forward_call(*args, **kwargs)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return f(*args, **kwargs)
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+        outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+
+    return self._call_impl(*args, **kwargs)  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    return self._call_impl(*args, **kwargs)    
+return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    trainer.train(dataloader)
+    return forward_call(*args, **kwargs)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    return forward_call(*args, **kwargs)
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+        hidden_encoder_states = encoder_block(**hidden_encoder_states)sharded_logits = self.model(
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    output = model(**micro_batch)    
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]return f(*args, **kwargs)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+        return self._call_impl(*args, **kwargs)return self._call_impl(*args, **kwargs)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    return self._call_impl(*args, **kwargs)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    out = F.linear(tensor, weight, bias)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 1 has a total capacty of 79.33 GiB of which 295.94 MiB is free. Including non-PyTorch memory, this process has 79.03 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 326.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return forward_call(*args, **kwargs)
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    return forward_call(*args, **kwargs)
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    return forward_call(*args, **kwargs)
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    output = model(**micro_batch)
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+    return self._call_impl(*args, **kwargs)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+return self._call_impl(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    sharded_logits = self.model(
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return forward_call(*args, **kwargs)
+    return forward_call(*args, **kwargs)  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    return self._call_impl(*args, **kwargs)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    out = F.linear(tensor, weight, bias)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 7 has a total capacty of 79.33 GiB of which 303.94 MiB is free. Including non-PyTorch memory, this process has 79.02 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 518.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return forward_call(*args, **kwargs)
+    return row_linear(
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+    sharded_logits = self.model(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    out = F.linear(tensor, weight, bias)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 2 has a total capacty of 79.33 GiB of which 343.94 MiB is free. Including non-PyTorch memory, this process has 78.98 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 358.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    out = F.linear(tensor, weight, bias)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 4 has a total capacty of 79.33 GiB of which 359.94 MiB is free. Including non-PyTorch memory, this process has 78.97 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 326.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return self._call_impl(*args, **kwargs)    
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    return forward_call(*args, **kwargs)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return forward_call(*args, **kwargs)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    output = self.pp_block(**new_kwargs)
+    return row_linear(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    return row_linear(
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+    out = F.linear(tensor, weight, bias)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 4 has a total capacty of 79.33 GiB of which 295.94 MiB is free. Including non-PyTorch memory, this process has 79.03 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 390.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    trainer.train(dataloader)
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    return f(*args, **kwargs)
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return self._call_impl(*args, **kwargs)
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    out = F.linear(tensor, weight, bias)
+    return self._call_impl(*args, **kwargs)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 6 has a total capacty of 79.33 GiB of which 279.94 MiB is free. Including non-PyTorch memory, this process has 79.04 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 422.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+Traceback (most recent call last):
+    out = F.linear(tensor, weight, bias)
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 3 has a total capacty of 79.33 GiB of which 271.94 MiB is free. Including non-PyTorch memory, this process has 79.05 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 230.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    trainer.train(dataloader)    
+outputs = self.pipeline_engine.train_batch_iter(
+    return row_linear(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return self._call_impl(*args, **kwargs)
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return self._call_impl(*args, **kwargs)
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    return self._call_impl(*args, **kwargs)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    out = F.linear(tensor, weight, bias)
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 0 has a total capacty of 79.33 GiB of which 375.94 MiB is free. Including non-PyTorch memory, this process has 78.95 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 358.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    return row_linear(
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    return self._call_impl(*args, **kwargs)
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return row_linear(
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+Traceback (most recent call last):
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    out = F.linear(tensor, weight, bias)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 7 has a total capacty of 79.33 GiB of which 431.94 MiB is free. Including non-PyTorch memory, this process has 78.90 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 390.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    return forward_call(*args, **kwargs)
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    out = F.linear(tensor, weight, bias)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 0 has a total capacty of 79.33 GiB of which 503.94 MiB is free. Including non-PyTorch memory, this process has 78.83 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 230.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return f(*args, **kwargs)
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+Traceback (most recent call last):
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return self._call_impl(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return forward_call(*args, **kwargs)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return self._call_impl(*args, **kwargs)
+        return self._call_impl(*args, **kwargs)return self._call_impl(*args, **kwargs)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+        return forward_call(*args, **kwargs)return forward_call(*args, **kwargs)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    return self._call_impl(*args, **kwargs)
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return self._call_impl(*args, **kwargs)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = self.pp_block(**new_kwargs)
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+    return row_linear(  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    out = F.linear(tensor, weight, bias)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 0 has a total capacty of 79.33 GiB of which 359.94 MiB is free. Including non-PyTorch memory, this process has 78.97 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 294.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+        outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+Traceback (most recent call last):
+        trainer.train(dataloader)return f(*args, **kwargs)
+    
+outputs = self.pipeline_engine.train_batch_iter(  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    out = F.linear(tensor, weight, bias)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 4 has a total capacty of 79.33 GiB of which 55.94 MiB is free. Including non-PyTorch memory, this process has 79.26 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 550.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+        return forward_call(*args, **kwargs)return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+        output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+        output = model(**micro_batch)
+output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return super().apply(*args, **kwargs)  # type: ignore[misc]    
+hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+    return f(*args, **kwargs)
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    out = F.linear(tensor, weight, bias)
+    return self._call_impl(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 5 has a total capacty of 79.33 GiB of which 55.94 MiB is free. Including non-PyTorch memory, this process has 79.26 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 646.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    return forward_call(*args, **kwargs)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+    return forward_call(*args, **kwargs)  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return f(*args, **kwargs)
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]    
+sharded_logits = self.model(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    output = model(**micro_batch)
+    return f(*args, **kwargs)    
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return row_linear(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+        return forward_call(*args, **kwargs)return forward_call(*args, **kwargs)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    return self._call_impl(*args, **kwargs)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return forward_call(*args, **kwargs)
+        return f(*args, **kwargs)outputs = self.pipeline_engine.train_batch_iter(
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    out = F.linear(tensor, weight, bias)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 3 has a total capacty of 79.33 GiB of which 159.94 MiB is free. Including non-PyTorch memory, this process has 79.16 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 422.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    sharded_logits = self.model(
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+        output = self.pp_block(**new_kwargs)return self._call_impl(*args, **kwargs)
+    
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+return forward_call(*args, **kwargs)output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+    sharded_logits = self.model(
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self._call_impl(*args, **kwargs)
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+return forward_call(*args, **kwargs)    
+return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    output = model(**micro_batch)    
+return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+        out = F.linear(tensor, weight, bias)return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 1 has a total capacty of 79.33 GiB of which 343.94 MiB is free. Including non-PyTorch memory, this process has 78.98 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 358.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    trainer.train(dataloader)
+    return forward_call(*args, **kwargs)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+        hidden_states = self.down_proj(self.split_silu_mul(merged_states))output = self.pp_block(**new_kwargs)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    
+sharded_logits = self.model(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+                return self._call_impl(*args, **kwargs)return self._call_impl(*args, **kwargs)return self._call_impl(*args, **kwargs)return self._call_impl(*args, **kwargs)
+
+
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return forward_call(*args, **kwargs)    
+return self._call_impl(*args, **kwargs)      File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+        return forward_call(*args, **kwargs)return forward_call(*args, **kwargs)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+        return forward_call(*args, **kwargs)return forward_call(*args, **kwargs)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    output = self.pp_block(**new_kwargs)
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    
+return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return self._call_impl(*args, **kwargs)
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    out = F.linear(tensor, weight, bias)
+    return self._call_impl(*args, **kwargs)torch.cuda
+.      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+OutOfMemoryErrorreturn self._call_impl(*args, **kwargs): 
+CUDA out of memory. Tried to allocate 512.00 MiB. GPU 2 has a total capacty of 79.33 GiB of which 215.94 MiB is free. Including non-PyTorch memory, this process has 79.11 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 326.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    out = F.linear(tensor, weight, bias)
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+torch.cuda    .OutOfMemoryErrorreturn self._call_impl(*args, **kwargs): 
+CUDA out of memory. Tried to allocate 512.00 MiB. GPU 6 has a total capacty of 79.33 GiB of which 247.94 MiB is free. Including non-PyTorch memory, this process has 79.08 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 294.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+          File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+return self._call_impl(*args, **kwargs)return forward_call(*args, **kwargs)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+Traceback (most recent call last):
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+Traceback (most recent call last):
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    output = self.pp_block(**new_kwargs)
+    sharded_logits = self.model(
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    trainer.train(dataloader)
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return self._call_impl(*args, **kwargs)
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+    return self._call_impl(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    return forward_call(*args, **kwargs)    
+return self._call_impl(*args, **kwargs)  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+
+    return f(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    out = F.linear(tensor, weight, bias)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 3 has a total capacty of 79.33 GiB of which 175.94 MiB is free. Including non-PyTorch memory, this process has 79.15 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 326.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return forward_call(*args, **kwargs)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    out = F.linear(tensor, weight, bias)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 0 has a total capacty of 79.33 GiB of which 247.94 MiB is free. Including non-PyTorch memory, this process has 79.08 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 326.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return self._call_impl(*args, **kwargs)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    output = model(**micro_batch)    
+hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.pp_block(**new_kwargs)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return self._call_impl(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.pp_block(**new_kwargs)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    output = model(**micro_batch)
+    return self._call_impl(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+out = F.linear(tensor, weight, bias)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 4 has a total capacty of 79.33 GiB of which 327.94 MiB is free. Including non-PyTorch memory, this process has 79.00 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 198.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    sharded_logits = self.model(
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+        return forward_call(*args, **kwargs)hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    out = F.linear(tensor, weight, bias)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 2 has a total capacty of 79.33 GiB of which 135.94 MiB is free. Including non-PyTorch memory, this process has 79.19 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 486.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+        return row_linear(return self._call_impl(*args, **kwargs)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    return forward_call(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)    
+    trainer.train(dataloader)output = model(**micro_batch)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    trainer.train(dataloader)
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+        return forward_call(*args, **kwargs)hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return f(*args, **kwargs)
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+Traceback (most recent call last):
+Traceback (most recent call last):
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return self._call_impl(*args, **kwargs)
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    out = F.linear(tensor, weight, bias)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 6 has a total capacty of 79.33 GiB of which 391.94 MiB is free. Including non-PyTorch memory, this process has 78.94 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 230.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    out = F.linear(tensor, weight, bias)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 7 has a total capacty of 79.33 GiB of which 399.94 MiB is free. Including non-PyTorch memory, this process has 78.93 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 422.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    trainer.train(dataloader)    
+trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs = self.pipeline_engine.train_batch_iter(
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    trainer.train(dataloader)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+        trainer.train(dataloader)trainer.train(dataloader)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    trainer.train(dataloader)
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return self._call_impl(*args, **kwargs)
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+        hidden_states = self.down_proj(self.split_silu_mul(merged_states))output = model(**micro_batch)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    return forward_call(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    return f(*args, **kwargs)
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = self.pp_block(**new_kwargs)
+    return f(*args, **kwargs)
+    return f(*args, **kwargs)  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+return f(*args, **kwargs)    
+return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+    return forward_call(*args, **kwargs)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self._call_impl(*args, **kwargs)
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    output = model(**micro_batch)
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+return forward_call(*args, **kwargs)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    out = F.linear(tensor, weight, bias)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 1 has a total capacty of 79.33 GiB of which 295.94 MiB is free. Including non-PyTorch memory, this process has 79.03 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 326.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return self._call_impl(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)    
+outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    output = model(**micro_batch)
+        return row_linear(hidden_encoder_states = encoder_block(**hidden_encoder_states)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+        outputs = self.pipeline_engine.train_batch_iter(outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    return forward_call(*args, **kwargs)
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    outputs = self.pipeline_engine.train_batch_iter(
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return forward_call(*args, **kwargs)
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    return self._call_impl(*args, **kwargs)
+    out = F.linear(tensor, weight, bias)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 2 has a total capacty of 79.33 GiB of which 359.94 MiB is free. Including non-PyTorch memory, this process has 78.97 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 262.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+        output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    
+    output = model(**micro_batch)sharded_logits = self.model(  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    output = model(**micro_batch)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    out = F.linear(tensor, weight, bias)
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+torch.cuda.    OutOfMemoryErrorreturn self._call_impl(*args, **kwargs): 
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+CUDA out of memory. Tried to allocate 512.00 MiB. GPU 0 has a total capacty of 79.33 GiB of which 295.94 MiB is free. Including non-PyTorch memory, this process has 79.03 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 358.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)    
+return forward_call(*args, **kwargs)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    out = F.linear(tensor, weight, bias)
+    output = self.pp_block(**new_kwargs)
+return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 5 has a total capacty of 79.33 GiB of which 199.94 MiB is free. Including non-PyTorch memory, this process has 79.12 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 422.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    return forward_call(*args, **kwargs)
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return forward_call(*args, **kwargs)
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    sharded_logits = self.model(
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+sharded_logits = self.model(
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return row_linear(
+        hidden_encoder_states = encoder_block(**hidden_encoder_states)sharded_logits = self.model(
+
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+        return self._call_impl(*args, **kwargs)return self._call_impl(*args, **kwargs)
+    
+return self._call_impl(*args, **kwargs)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+        return forward_call(*args, **kwargs)return forward_call(*args, **kwargs)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    return forward_call(*args, **kwargs)
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+    return self._call_impl(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+    out = F.linear(tensor, weight, bias)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 3 has a total capacty of 79.33 GiB of which 223.94 MiB is free. Including non-PyTorch memory, this process has 79.10 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 358.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    out = F.linear(tensor, weight, bias)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 5 has a total capacty of 79.33 GiB of which 311.94 MiB is free. Including non-PyTorch memory, this process has 79.01 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 390.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+        hidden_encoder_states = encoder_block(**hidden_encoder_states)
+hidden_encoder_states = encoder_block(**hidden_encoder_states)    
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+hidden_encoder_states = encoder_block(**hidden_encoder_states)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    
+hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    return forward_call(*args, **kwargs)
+    return row_linear(
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    return self._call_impl(*args, **kwargs)    
+    return self._call_impl(*args, **kwargs)
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+        return self._call_impl(*args, **kwargs)return self._call_impl(*args, **kwargs)
+
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+return self._call_impl(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    return forward_call(*args, **kwargs)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+return forward_call(*args, **kwargs)    
+return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    out = F.linear(tensor, weight, bias)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 7 has a total capacty of 79.33 GiB of which 447.94 MiB is free. Including non-PyTorch memory, this process has 78.88 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 454.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return self._call_impl(*args, **kwargs)
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = self.pp_block(**new_kwargs)    
+hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    output = self.pp_block(**new_kwargs)
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+return self._call_impl(*args, **kwargs)    
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return row_linear(
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+return self._call_impl(*args, **kwargs)
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+        return forward_call(*args, **kwargs)return forward_call(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    
+return self._call_impl(*args, **kwargs)  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+return forward_call(*args, **kwargs)      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+
+    return forward_call(*args, **kwargs)
+return forward_call(*args, **kwargs)  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    return forward_call(*args, **kwargs)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+    out = F.linear(tensor, weight, bias)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 6 has a total capacty of 79.33 GiB of which 455.94 MiB is free. Including non-PyTorch memory, this process has 78.87 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 166.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)    
+hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]    
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    out = F.linear(tensor, weight, bias)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 6 has a total capacty of 79.33 GiB of which 279.94 MiB is free. Including non-PyTorch memory, this process has 79.04 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 422.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    out = F.linear(tensor, weight, bias)
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 1 has a total capacty of 79.33 GiB of which 151.94 MiB is free. Including non-PyTorch memory, this process has 79.17 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 550.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    out = F.linear(tensor, weight, bias)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 7 has a total capacty of 79.33 GiB of which 367.94 MiB is free. Including non-PyTorch memory, this process has 78.96 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 454.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)    
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    return forward_call(*args, **kwargs)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+return forward_call(*args, **kwargs)
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    out = F.linear(tensor, weight, bias)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 4 has a total capacty of 79.33 GiB of which 407.94 MiB is free. Including non-PyTorch memory, this process has 78.92 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 198.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    return row_linear(
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    trainer.train(dataloader)
+    trainer.train(dataloader)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    
+trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    out = F.linear(tensor, weight, bias)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 2 has a total capacty of 79.33 GiB of which 375.94 MiB is free. Including non-PyTorch memory, this process has 78.95 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 326.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return f(*args, **kwargs)
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return super().apply(*args, **kwargs)  # type: ignore[misc]    
+return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    out = F.linear(tensor, weight, bias)
+    out = F.linear(tensor, weight, bias)
+torch.cuda.OutOfMemoryErrortorch.cuda: .CUDA out of memory. Tried to allocate 512.00 MiB. GPU 0 has a total capacty of 79.33 GiB of which 247.94 MiB is free. Including non-PyTorch memory, this process has 79.08 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 486.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONFOutOfMemoryError
+: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 4 has a total capacty of 79.33 GiB of which 167.94 MiB is free. Including non-PyTorch memory, this process has 79.15 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 518.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+    out = F.linear(tensor, weight, bias)  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 5 has a total capacty of 79.33 GiB of which 231.94 MiB is free. Including non-PyTorch memory, this process has 79.09 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 390.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    outputs = self.pipeline_engine.train_batch_iter(
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    out = F.linear(tensor, weight, bias)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 1 has a total capacty of 79.33 GiB of which 199.94 MiB is free. Including non-PyTorch memory, this process has 79.12 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 422.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    out = F.linear(tensor, weight, bias)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 3 has a total capacty of 79.33 GiB of which 111.94 MiB is free. Including non-PyTorch memory, this process has 79.21 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 390.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = model(**micro_batch)    
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+output = model(**micro_batch)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+        return self._call_impl(*args, **kwargs)return f(*args, **kwargs)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return self._call_impl(*args, **kwargs)    
+    return forward_call(*args, **kwargs)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)    
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+        return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]hidden_encoder_states = encoder_block(**hidden_encoder_states)
+
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return f(*args, **kwargs)
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return self._call_impl(*args, **kwargs)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+        return forward_call(*args, **kwargs)return self._call_impl(*args, **kwargs)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    trainer.train(dataloader)
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return f(*args, **kwargs)
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    output = self.pp_block(**new_kwargs)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    output = self.pp_block(**new_kwargs)
+    output = self.pp_block(**new_kwargs)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    
+output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+    return forward_call(*args, **kwargs)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+        return self._call_impl(*args, **kwargs)return self._call_impl(*args, **kwargs)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    return self._call_impl(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = model(**micro_batch)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+        return self._call_impl(*args, **kwargs)return forward_call(*args, **kwargs)    
+    
+return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]return self._call_impl(*args, **kwargs)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    return self._call_impl(*args, **kwargs)
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+        hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+        hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]sharded_logits = self.model(
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return forward_call(*args, **kwargs)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    return forward_call(*args, **kwargs)
+          File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+return forward_call(*args, **kwargs)return forward_call(*args, **kwargs)    
+return forward_call(*args, **kwargs)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+        hidden_states = self.down_proj(self.split_silu_mul(merged_states))hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+        return forward_call(*args, **kwargs)return forward_call(*args, **kwargs)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    output = self.pp_block(**new_kwargs)
+    return row_linear(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+        return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+return row_linear(
+
+    return forward_call(*args, **kwargs)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)    
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 725, in _core_forward
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    output = self.attn(hidden_states=hidden_states, sequence_mask=sequence_mask)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    out = F.linear(tensor, weight, bias)
+    out = F.linear(tensor, weight, bias)torch.cuda
+.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 6 has a total capacty of 79.33 GiB of which 375.94 MiB is free. Including non-PyTorch memory, this process has 78.95 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 166.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONFtorch.cuda
+.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 1 has a total capacty of 79.33 GiB of which 359.94 MiB is free. Including non-PyTorch memory, this process has 78.97 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 262.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 725, in _core_forward
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    out = F.linear(tensor, weight, bias)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 4 has a total capacty of 79.33 GiB of which 199.94 MiB is free. Including non-PyTorch memory, this process has 79.12 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 326.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    out = F.linear(tensor, weight, bias)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 0 has a total capacty of 79.33 GiB of which 151.94 MiB is free. Including non-PyTorch memory, this process has 79.17 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 422.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    output = self.attn(hidden_states=hidden_states, sequence_mask=sequence_mask)
+        return super().apply(*args, **kwargs)  # type: ignore[misc]hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 690, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    out = F.linear(tensor, weight, bias)
+    return self._call_impl(*args, **kwargs)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 2 has a total capacty of 79.33 GiB of which 55.94 MiB is free. Including non-PyTorch memory, this process has 79.26 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 486.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)    
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+return self._call_impl(*args, **kwargs)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = self.o_proj(attention_output)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)    
+return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    return forward_call(*args, **kwargs)
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return forward_call(*args, **kwargs)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 690, in forward
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+Traceback (most recent call last):
+    return row_linear(
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    output = self.o_proj(attention_output)    
+return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    sharded_logits = self.model(
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+    out = F.linear(tensor, weight, bias)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 3 has a total capacty of 79.33 GiB of which 335.94 MiB is free. Including non-PyTorch memory, this process has 78.99 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 166.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+    out = F.linear(tensor, weight, bias)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 5 has a total capacty of 79.33 GiB of which 359.94 MiB is free. Including non-PyTorch memory, this process has 78.97 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 262.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+        outputs = self.pipeline_engine.train_batch_iter(output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+    out = F.linear(tensor, weight, bias)
+    torch.cudareturn self._call_impl(*args, **kwargs).
+OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 4 has a total capacty of 79.33 GiB of which 453.94 MiB is free. Including non-PyTorch memory, this process has 78.88 GiB memory in use. Of the allocated memory 70.35 GiB is allocated by PyTorch, and 166.12 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    return self._call_impl(*args, **kwargs)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+        return super().apply(*args, **kwargs)  # type: ignore[misc]return forward_call(*args, **kwargs)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    out = F.linear(tensor, weight, bias)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 6 has a total capacty of 79.33 GiB of which 309.94 MiB is free. Including non-PyTorch memory, this process has 79.02 GiB memory in use. Of the allocated memory 70.35 GiB is allocated by PyTorch, and 326.12 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return forward_call(*args, **kwargs)    
+return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self._call_impl(*args, **kwargs)
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+        return f(*args, **kwargs)return self._call_impl(*args, **kwargs)    
+
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    return forward_call(*args, **kwargs)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 725, in _core_forward
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    outputs = self.pipeline_engine.train_batch_iter(
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    output = self.attn(hidden_states=hidden_states, sequence_mask=sequence_mask)
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 725, in _core_forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 690, in forward
+    return self._call_impl(*args, **kwargs)    
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    return forward_call(*args, **kwargs)
+    out = F.linear(tensor, weight, bias)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    output = self.attn(hidden_states=hidden_states, sequence_mask=sequence_mask)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 7 has a total capacty of 79.33 GiB of which 399.94 MiB is free. Including non-PyTorch memory, this process has 78.93 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 422.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)    
+hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 725, in _core_forward
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 725, in _core_forward
+    output = self.o_proj(attention_output)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.attn(hidden_states=hidden_states, sequence_mask=sequence_mask)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.attn(hidden_states=hidden_states, sequence_mask=sequence_mask)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+        return self._call_impl(*args, **kwargs)return forward_call(*args, **kwargs)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 690, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 690, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 690, in forward
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    output = self.o_proj(attention_output)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+        output = self.o_proj(attention_output)output = self.o_proj(attention_output)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+        return self._call_impl(*args, **kwargs)return self._call_impl(*args, **kwargs)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+        return self._call_impl(*args, **kwargs)return self._call_impl(*args, **kwargs)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    out = F.linear(tensor, weight, bias)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 1 has a total capacty of 79.33 GiB of which 277.94 MiB is free. Including non-PyTorch memory, this process has 79.05 GiB memory in use. Of the allocated memory 70.35 GiB is allocated by PyTorch, and 358.12 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return self._call_impl(*args, **kwargs)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    out = F.linear(tensor, weight, bias)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 5 has a total capacty of 79.33 GiB of which 469.94 MiB is free. Including non-PyTorch memory, this process has 78.86 GiB memory in use. Of the allocated memory 70.35 GiB is allocated by PyTorch, and 166.12 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    out = F.linear(tensor, weight, bias)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 3 has a total capacty of 79.33 GiB of which 189.94 MiB is free. Including non-PyTorch memory, this process has 79.13 GiB memory in use. Of the allocated memory 70.35 GiB is allocated by PyTorch, and 326.12 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 725, in _core_forward
+    out = F.linear(tensor, weight, bias)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 0 has a total capacty of 79.33 GiB of which 469.94 MiB is free. Including non-PyTorch memory, this process has 78.86 GiB memory in use. Of the allocated memory 70.35 GiB is allocated by PyTorch, and 198.12 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    output = self.attn(hidden_states=hidden_states, sequence_mask=sequence_mask)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 690, in forward
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs = self.pipeline_engine.train_batch_iter(
+    output = self.o_proj(attention_output)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return self._call_impl(*args, **kwargs)
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return forward_call(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    output = model(**micro_batch)
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    out = F.linear(tensor, weight, bias)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 2 has a total capacty of 79.33 GiB of which 373.94 MiB is free. Including non-PyTorch memory, this process has 78.95 GiB memory in use. Of the allocated memory 70.35 GiB is allocated by PyTorch, and 262.12 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+        return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]trainer.train(dataloader)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    output = model(**micro_batch)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    out = F.linear(tensor, weight, bias)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 2 has a total capacty of 79.33 GiB of which 359.94 MiB is free. Including non-PyTorch memory, this process has 78.97 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 262.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    out = F.linear(tensor, weight, bias)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 1 has a total capacty of 79.33 GiB of which 439.94 MiB is free. Including non-PyTorch memory, this process has 78.89 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 262.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    return row_linear(
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return f(*args, **kwargs)
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    out = F.linear(tensor, weight, bias)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 3 has a total capacty of 79.33 GiB of which 351.94 MiB is free. Including non-PyTorch memory, this process has 78.97 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 230.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    out = F.linear(tensor, weight, bias)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 5 has a total capacty of 79.33 GiB of which 247.94 MiB is free. Including non-PyTorch memory, this process has 79.08 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 454.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    return self._call_impl(*args, **kwargs)
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    out = F.linear(tensor, weight, bias)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 5 has a total capacty of 79.33 GiB of which 71.94 MiB is free. Including non-PyTorch memory, this process has 79.25 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 550.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+Traceback (most recent call last):
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+Traceback (most recent call last):
+Traceback (most recent call last):
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    outputs = self.pipeline_engine.train_batch_iter(
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    return self._call_impl(*args, **kwargs)
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    output = model(**micro_batch)
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+        trainer.train(dataloader)trainer.train(dataloader)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    trainer.train(dataloader)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+trainer.train(dataloader)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+trainer.train(dataloader)
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    output = self.pp_block(**new_kwargs)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+        return self._call_impl(*args, **kwargs)return self._call_impl(*args, **kwargs)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+        return forward_call(*args, **kwargs)return forward_call(*args, **kwargs)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    trainer.train(dataloader)
+    return forward_call(*args, **kwargs)
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    sharded_logits = self.model(
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+        return f(*args, **kwargs)
+return f(*args, **kwargs)
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return f(*args, **kwargs)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+return f(*args, **kwargs)
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    return f(*args, **kwargs)
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return self._call_impl(*args, **kwargs)    
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    return self._call_impl(*args, **kwargs)
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    trainer.train(dataloader)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    trainer.train(dataloader)
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return f(*args, **kwargs)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+        outputs = self.pipeline_engine.train_batch_iter(outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+Traceback (most recent call last):
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    outputs = self.pipeline_engine.train_batch_iter(
+    outputs = self.pipeline_engine.train_batch_iter(  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    return self._call_impl(*args, **kwargs)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+    return forward_call(*args, **kwargs)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    outputs = self.pipeline_engine.train_batch_iter(
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    trainer.train(dataloader)
+    outputs = self.pipeline_engine.train_batch_iter(
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    output = model(**micro_batch)
+    return forward_call(*args, **kwargs)
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return row_linear(
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    return forward_call(*args, **kwargs)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    return f(*args, **kwargs)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    output = model(**micro_batch)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return row_linear(
+    return f(*args, **kwargs)
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    output = model(**micro_batch)
+    output = model(**micro_batch)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = model(**micro_batch)
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+    outputs = self.pipeline_engine.train_batch_iter(
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+    output = model(**micro_batch)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+sharded_logits = self.model(
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    return self._call_impl(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    out = F.linear(tensor, weight, bias)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+        return self._call_impl(*args, **kwargs)return self._call_impl(*args, **kwargs)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 4 has a total capacty of 79.33 GiB of which 87.94 MiB is free. Including non-PyTorch memory, this process has 79.23 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 518.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    return forward_call(*args, **kwargs)
+            return self._call_impl(*args, **kwargs)return self._call_impl(*args, **kwargs)return self._call_impl(*args, **kwargs)
+
+
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    out = F.linear(tensor, weight, bias)
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    return self._call_impl(*args, **kwargs)    
+outputs = self.pipeline_engine.train_batch_iter(    
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 6 has a total capacty of 79.33 GiB of which 135.94 MiB is free. Including non-PyTorch memory, this process has 79.19 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 486.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+
+        return self._call_impl(*args, **kwargs)return self._call_impl(*args, **kwargs)
+
+          File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+return forward_call(*args, **kwargs)return forward_call(*args, **kwargs)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    outputs = self.pipeline_engine.train_batch_iter(
+    return forward_call(*args, **kwargs)
+          File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+return forward_call(*args, **kwargs)
+return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    output = self.pp_block(**new_kwargs)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+    return forward_call(*args, **kwargs)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    output = model(**micro_batch)
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    sharded_logits = self.model(
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+sharded_logits = self.model(
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+    sharded_logits = self.model(
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+sharded_logits = self.model(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    sharded_logits = self.model(
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+        return forward_call(*args, **kwargs)return self._call_impl(*args, **kwargs)
+
+    return self._call_impl(*args, **kwargs)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    return self._call_impl(*args, **kwargs)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return self._call_impl(*args, **kwargs)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+hidden_encoder_states = encoder_block(**hidden_encoder_states)
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return forward_call(*args, **kwargs)
+        return forward_call(*args, **kwargs)return forward_call(*args, **kwargs)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return row_linear(
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    out = F.linear(tensor, weight, bias)
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 3 has a total capacty of 79.33 GiB of which 95.94 MiB is free. Including non-PyTorch memory, this process has 79.22 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 326.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))    
+return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    sharded_logits = self.model(    
+sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    output = self.pp_block(**new_kwargs)
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return self._call_impl(*args, **kwargs)
+Traceback (most recent call last):
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return row_linear(
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    return self._call_impl(*args, **kwargs)
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 469, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return f(*args, **kwargs)
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]    
+hidden_encoder_states = encoder_block(**hidden_encoder_states)
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+        hidden_encoder_states = encoder_block(**hidden_encoder_states)    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+hidden_encoder_states = encoder_block(**hidden_encoder_states)
+
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+hidden_encoder_states = encoder_block(**hidden_encoder_states)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return forward_call(*args, **kwargs)
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+        return forward_call(*args, **kwargs)return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    trainer.train(dataloader)
+    out = differentiable_reduce_scatter_sum(out, group=group)
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/distributed_differentiable_primitives.py", line 146, in differentiable_reduce_scatter_sum
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return row_linear(
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    return DifferentiableReduceScatterSum.apply(tensor, group)
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/distributed_differentiable_primitives.py", line 112, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+        return self._call_impl(*args, **kwargs)return self._call_impl(*args, **kwargs)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+        return self._call_impl(*args, **kwargs)return self._call_impl(*args, **kwargs)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+        output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)return forward_call(*args, **kwargs)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+        return forward_call(*args, **kwargs)return forward_call(*args, **kwargs)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    return forward_call(*args, **kwargs)
+    out = F.linear(tensor, weight, bias)
+    return forward_call(*args, **kwargs)    
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+return forward_call(*args, **kwargs)      File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+return forward_call(*args, **kwargs)
+        
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+return forward_call(*args, **kwargs)return forward_call(*args, **kwargs)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+        return forward_call(*args, **kwargs)
+return forward_call(*args, **kwargs)
+    sharded_tensor = torch.empty(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 0 has a total capacty of 79.33 GiB of which 135.94 MiB is free. Including non-PyTorch memory, this process has 79.19 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 518.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 64.00 MiB. GPU 0 has a total capacty of 79.33 GiB of which 7.94 MiB is free. Including non-PyTorch memory, this process has 79.31 GiB memory in use. Of the allocated memory 70.16 GiB is allocated by PyTorch, and 134.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    output = model(**micro_batch)
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    output = self.pp_block(**new_kwargs)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    return forward_call(*args, **kwargs)
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    out = F.linear(tensor, weight, bias)
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 5 has a total capacty of 79.33 GiB of which 183.94 MiB is free. Including non-PyTorch memory, this process has 79.14 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 358.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+    out = F.linear(tensor, weight, bias)
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 6 has a total capacty of 79.33 GiB of which 423.94 MiB is free. Including non-PyTorch memory, this process has 78.90 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 198.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    return self._call_impl(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = self.pp_block(**new_kwargs)
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.pp_block(**new_kwargs)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.pp_block(**new_kwargs)    
+output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    return forward_call(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    return self._call_impl(*args, **kwargs)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    return forward_call(*args, **kwargs)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return forward_call(*args, **kwargs)
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    return self._call_impl(*args, **kwargs)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+    return row_linear(
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    sharded_logits = self.model(
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)    
+return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    return forward_call(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return forward_call(*args, **kwargs)
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+    sharded_logits = self.model(
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+    sharded_logits = self.model(
+    return self._call_impl(*args, **kwargs)
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+        return self._call_impl(*args, **kwargs)
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return row_linear(
+    return forward_call(*args, **kwargs)
+    output = model(**micro_batch)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+        return self._call_impl(*args, **kwargs)return self._call_impl(*args, **kwargs)
+
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+Traceback (most recent call last):
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+    return self._call_impl(*args, **kwargs)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+        return forward_call(*args, **kwargs)return forward_call(*args, **kwargs)    
+
+return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    return self._call_impl(*args, **kwargs)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+        trainer.train(dataloader)out = F.linear(tensor, weight, bias)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    out = F.linear(tensor, weight, bias)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 4 has a total capacty of 79.33 GiB of which 295.94 MiB is free. Including non-PyTorch memory, this process has 79.03 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 230.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 5 has a total capacty of 79.33 GiB of which 247.94 MiB is free. Including non-PyTorch memory, this process has 79.08 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 454.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return forward_call(*args, **kwargs)
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    out = F.linear(tensor, weight, bias)
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 2 has a total capacty of 79.33 GiB of which 135.94 MiB is free. Including non-PyTorch memory, this process has 79.19 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 486.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+Traceback (most recent call last):
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    trainer.train(dataloader)    
+        trainer.train(dataloader)return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+return row_linear(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return f(*args, **kwargs)
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self._call_impl(*args, **kwargs)
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+        return self._call_impl(*args, **kwargs)return self._call_impl(*args, **kwargs)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+        return forward_call(*args, **kwargs)outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    out = F.linear(tensor, weight, bias)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return f(*args, **kwargs)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 7 has a total capacty of 79.33 GiB of which 431.94 MiB is free. Including non-PyTorch memory, this process has 78.90 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 390.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    trainer.train(dataloader)    
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return f(*args, **kwargs)
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+        out = F.linear(tensor, weight, bias)return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+torch.cuda  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 2 has a total capacty of 79.33 GiB of which 343.94 MiB is free. Including non-PyTorch memory, this process has 78.98 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 198.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+        return self._call_impl(*args, **kwargs)return self._call_impl(*args, **kwargs)
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 469, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    return forward_call(*args, **kwargs)
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    return self._call_impl(*args, **kwargs)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+return self._call_impl(*args, **kwargs)
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    trainer.train(dataloader)
+    trainer.train(dataloader)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+
+    trainer.train(dataloader)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return self._call_impl(*args, **kwargs)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+return self._call_impl(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    return row_linear(
+    outputs = self.pipeline_engine.train_batch_iter(
+    return self._call_impl(*args, **kwargs)
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+    out = differentiable_reduce_scatter_sum(out, group=group)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    return forward_call(*args, **kwargs)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/distributed_differentiable_primitives.py", line 146, in differentiable_reduce_scatter_sum
+        return forward_call(*args, **kwargs)return forward_call(*args, **kwargs)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    output = self.pp_block(**new_kwargs)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    return forward_call(*args, **kwargs)
+          File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+return forward_call(*args, **kwargs)return forward_call(*args, **kwargs)
+
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return row_linear(
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return f(*args, **kwargs)    
+return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+        return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return DifferentiableReduceScatterSum.apply(tensor, group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    out = F.linear(tensor, weight, bias)
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 7 has a total capacty of 79.33 GiB of which 495.94 MiB is free. Including non-PyTorch memory, this process has 78.83 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 326.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+        trainer.train(dataloader)output = self.pp_block(**new_kwargs)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+        return f(*args, **kwargs)
+return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+        hidden_states = self.down_proj(self.split_silu_mul(merged_states))hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    trainer.train(dataloader)
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    output = model(**micro_batch)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    out = F.linear(tensor, weight, bias)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 3 has a total capacty of 79.33 GiB of which 191.94 MiB is free. Including non-PyTorch memory, this process has 79.13 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 230.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/distributed_differentiable_primitives.py", line 112, in forward
+    return forward_call(*args, **kwargs)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return forward_call(*args, **kwargs)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    out = F.linear(tensor, weight, bias)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    out = F.linear(tensor, weight, bias)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 4 has a total capacty of 79.33 GiB of which 215.94 MiB is free. Including non-PyTorch memory, this process has 79.11 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 390.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    output = model(**micro_batch)
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 5 has a total capacty of 79.33 GiB of which 215.94 MiB is free. Including non-PyTorch memory, this process has 79.11 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 326.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    return f(*args, **kwargs)
+    output = model(**micro_batch)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = self.pp_block(**new_kwargs)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    out = F.linear(tensor, weight, bias)
+    return self._call_impl(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 3 has a total capacty of 79.33 GiB of which 223.94 MiB is free. Including non-PyTorch memory, this process has 79.10 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 358.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+return forward_call(*args, **kwargs)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    return self._call_impl(*args, **kwargs)  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+
+        output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)return self._call_impl(*args, **kwargs)    
+
+output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+Traceback (most recent call last):
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return forward_call(*args, **kwargs)
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+sharded_tensor = torch.empty(
+        return self._call_impl(*args, **kwargs)    return self._call_impl(*args, **kwargs)
+return self._call_impl(*args, **kwargs)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    sharded_logits = self.model(    
+hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+torch.cuda    .out = F.linear(tensor, weight, bias)OutOfMemoryError
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    out = F.linear(tensor, weight, bias)
+: CUDA out of memory. Tried to allocate 64.00 MiB. GPU 7 has a total capacty of 79.33 GiB of which 31.94 MiB is free. Including non-PyTorch memory, this process has 79.29 GiB memory in use. Of the allocated memory 70.16 GiB is allocated by PyTorch, and 198.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 1 has a total capacty of 79.33 GiB of which 151.94 MiB is free. Including non-PyTorch memory, this process has 79.17 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 390.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 2 has a total capacty of 79.33 GiB of which 167.94 MiB is free. Including non-PyTorch memory, this process has 79.15 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 454.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    sharded_logits = self.model(
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+        return self._call_impl(*args, **kwargs)outputs = self.pipeline_engine.train_batch_iter(
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    return forward_call(*args, **kwargs)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    trainer.train(dataloader)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+        return self._call_impl(*args, **kwargs)return forward_call(*args, **kwargs)
+    
+return self._call_impl(*args, **kwargs)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    sharded_logits = self.model(
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    sharded_logits = self.model(
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+return self._call_impl(*args, **kwargs)
+    return forward_call(*args, **kwargs)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    return self._call_impl(*args, **kwargs)
+        return row_linear(return row_linear(
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    return self._call_impl(*args, **kwargs)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return row_linear(
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+        return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    return f(*args, **kwargs)
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    output = model(**micro_batch)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return self._call_impl(*args, **kwargs)
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+        return super().apply(*args, **kwargs)  # type: ignore[misc]out = F.linear(tensor, weight, bias)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 6 has a total capacty of 79.33 GiB of which 263.94 MiB is free. Including non-PyTorch memory, this process has 79.06 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 358.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    output = model(**micro_batch)
+    return forward_call(*args, **kwargs)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+return self._call_impl(*args, **kwargs)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+        return self._call_impl(*args, **kwargs)return forward_call(*args, **kwargs)
+
+    return self._call_impl(*args, **kwargs)  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+Traceback (most recent call last):
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+        return super().apply(*args, **kwargs)  # type: ignore[misc]return super().apply(*args, **kwargs)  # type: ignore[misc]  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    out = F.linear(tensor, weight, bias)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+        out = F.linear(tensor, weight, bias)
+return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 3 has a total capacty of 79.33 GiB of which 335.94 MiB is free. Including non-PyTorch memory, this process has 78.99 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 166.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 1 has a total capacty of 79.33 GiB of which 183.94 MiB is free. Including non-PyTorch memory, this process has 79.14 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 518.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    out = F.linear(tensor, weight, bias)
+    out = F.linear(tensor, weight, bias)
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 5 has a total capacty of 79.33 GiB of which 311.94 MiB is free. Including non-PyTorch memory, this process has 79.01 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 390.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 7 has a total capacty of 79.33 GiB of which 127.94 MiB is free. Including non-PyTorch memory, this process has 79.19 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 774.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+        return self._call_impl(*args, **kwargs)return self._call_impl(*args, **kwargs)
+
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+    return forward_call(*args, **kwargs)
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    out = F.linear(tensor, weight, bias)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 4 has a total capacty of 79.33 GiB of which 343.94 MiB is free. Including non-PyTorch memory, this process has 78.98 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 262.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    trainer.train(dataloader)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+        return forward_call(*args, **kwargs)return forward_call(*args, **kwargs)
+
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    out = F.linear(tensor, weight, bias)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 7 has a total capacty of 79.33 GiB of which 271.94 MiB is free. Including non-PyTorch memory, this process has 79.05 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 550.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+        return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]return forward_call(*args, **kwargs)
+
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    sharded_logits = self.model(return forward_call(*args, **kwargs)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 201, in forward
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    return self.act(gate_states) * up_states
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 224.00 MiB. GPU 5 has a total capacty of 79.33 GiB of which 199.94 MiB is free. Including non-PyTorch memory, this process has 79.12 GiB memory in use. Of the allocated memory 69.44 GiB is allocated by PyTorch, and 646.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    out = F.linear(tensor, weight, bias)
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 1 has a total capacty of 79.33 GiB of which 231.94 MiB is free. Including non-PyTorch memory, this process has 79.09 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 390.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return self._call_impl(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)    
+return self._call_impl(*args, **kwargs)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    out = F.linear(tensor, weight, bias)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    trainer.train(dataloader)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 0 has a total capacty of 79.33 GiB of which 231.94 MiB is free. Including non-PyTorch memory, this process has 79.09 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 422.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return self._call_impl(*args, **kwargs)
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    sharded_logits = self.model(
+            return self._call_impl(*args, **kwargs)return self._call_impl(*args, **kwargs)return self._call_impl(*args, **kwargs)
+
+    
+return forward_call(*args, **kwargs)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return forward_call(*args, **kwargs)
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+        return self._call_impl(*args, **kwargs)return forward_call(*args, **kwargs)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    outputs = self.pipeline_engine.train_batch_iter(    
+return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return f(*args, **kwargs)
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    output = self.pp_block(**new_kwargs)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]    
+hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    output = self.pp_block(**new_kwargs)
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = model(**micro_batch)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return self._call_impl(*args, **kwargs)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    return self._call_impl(*args, **kwargs)    
+return forward_call(*args, **kwargs)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+        return super().apply(*args, **kwargs)  # type: ignore[misc]return f(*args, **kwargs)
+
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return self._call_impl(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    return forward_call(*args, **kwargs)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+return self._call_impl(*args, **kwargs)
+    out = F.linear(tensor, weight, bias)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 6 has a total capacty of 79.33 GiB of which 247.94 MiB is free. Including non-PyTorch memory, this process has 79.08 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 294.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)    
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+        return super().apply(*args, **kwargs)  # type: ignore[misc]output = self.pp_block(**new_kwargs)
+
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+    output = model(**micro_batch)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    out = F.linear(tensor, weight, bias)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 4 has a total capacty of 79.33 GiB of which 247.94 MiB is free. Including non-PyTorch memory, this process has 79.08 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 358.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+Traceback (most recent call last):
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+        hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return self._call_impl(*args, **kwargs)
+        output = self.pp_block(**new_kwargs)hidden_encoder_states = encoder_block(**hidden_encoder_states)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    out = F.linear(tensor, weight, bias)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    out = F.linear(tensor, weight, bias)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 1 has a total capacty of 79.33 GiB of which 247.94 MiB is free. Including non-PyTorch memory, this process has 79.08 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 294.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+        outputs = self.pipeline_engine.train_batch_iter(return self._call_impl(*args, **kwargs)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 4 has a total capacty of 79.33 GiB of which 391.94 MiB is free. Including non-PyTorch memory, this process has 78.94 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 294.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    sharded_logits = self.model(
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return self._call_impl(*args, **kwargs)    
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    return forward_call(*args, **kwargs)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+return forward_call(*args, **kwargs)
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    trainer.train(dataloader)
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    return f(*args, **kwargs)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+    return self._call_impl(*args, **kwargs)  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+
+    out = F.linear(tensor, weight, bias)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 3 has a total capacty of 79.33 GiB of which 111.94 MiB is free. Including non-PyTorch memory, this process has 79.21 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 390.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+        return self._call_impl(*args, **kwargs)return forward_call(*args, **kwargs)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    return forward_call(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return f(*args, **kwargs)
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    return forward_call(*args, **kwargs)
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+        return self._call_impl(*args, **kwargs)return self._call_impl(*args, **kwargs)
+
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+    return forward_call(*args, **kwargs)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))    
+hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+    return forward_call(*args, **kwargs)
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)    return forward_call(*args, **kwargs)
+    
+return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self._call_impl(*args, **kwargs)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+        hidden_encoder_states = encoder_block(**hidden_encoder_states)out = F.linear(tensor, weight, bias)
+
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+return forward_call(*args, **kwargs)torch.cuda
+.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 1 has a total capacty of 79.33 GiB of which 327.94 MiB is free. Including non-PyTorch memory, this process has 79.00 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 294.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    out = F.linear(tensor, weight, bias)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 1 has a total capacty of 79.33 GiB of which 183.94 MiB is free. Including non-PyTorch memory, this process has 79.14 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 518.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    trainer.train(dataloader)
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return forward_call(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    return self._call_impl(*args, **kwargs)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+        return forward_call(*args, **kwargs)return forward_call(*args, **kwargs)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+        return forward_call(*args, **kwargs)return self._call_impl(*args, **kwargs)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+    return f(*args, **kwargs)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+        return forward_call(*args, **kwargs)    return forward_call(*args, **kwargs)
+return forward_call(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+Traceback (most recent call last):
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+        output = self.pp_block(**new_kwargs)return forward_call(*args, **kwargs)
+
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    out = F.linear(tensor, weight, bias)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 0 has a total capacty of 79.33 GiB of which 23.94 MiB is free. Including non-PyTorch memory, this process has 79.29 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 550.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+        return forward_call(*args, **kwargs)hidden_encoder_states = encoder_block(**hidden_encoder_states)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)
+    sharded_logits = self.model(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+return row_linear(
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return row_linear(
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+return row_linear(
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    return self._call_impl(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    output = self.pp_block(**new_kwargs)
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return row_linear(
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)    
+return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)    
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return forward_call(*args, **kwargs)
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+    out = F.linear(tensor, weight, bias)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 3 has a total capacty of 79.33 GiB of which 15.94 MiB is free. Including non-PyTorch memory, this process has 79.30 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 486.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    trainer.train(dataloader)
+        output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+    return forward_call(*args, **kwargs)
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return row_linear(
+    return forward_call(*args, **kwargs)
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+Traceback (most recent call last):
+    return row_linear(
+    output = model(**micro_batch)
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return super().apply(*args, **kwargs)  # type: ignore[misc]    
+return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    out = F.linear(tensor, weight, bias)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 2 has a total capacty of 79.33 GiB of which 71.94 MiB is free. Including non-PyTorch memory, this process has 79.25 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 550.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+Traceback (most recent call last):
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+        out = F.linear(tensor, weight, bias)return forward_call(*args, **kwargs)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return self._call_impl(*args, **kwargs)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 6 has a total capacty of 79.33 GiB of which 487.94 MiB is free. Including non-PyTorch memory, this process has 78.84 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 134.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    out = F.linear(tensor, weight, bias)
+    trainer.train(dataloader)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 7 has a total capacty of 79.33 GiB of which 415.94 MiB is free. Including non-PyTorch memory, this process has 78.91 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 486.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return row_linear(
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return self._call_impl(*args, **kwargs)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+        out = F.linear(tensor, weight, bias)out = F.linear(tensor, weight, bias)
+
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    torch.cudaout = F.linear(tensor, weight, bias)    torch.cuda.
+out = F.linear(tensor, weight, bias).OutOfMemoryError
+OutOfMemoryError: : CUDA out of memory. Tried to allocate 512.00 MiB. GPU 4 has a total capacty of 79.33 GiB of which 295.94 MiB is free. Including non-PyTorch memory, this process has 79.03 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 390.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONFCUDA out of memory. Tried to allocate 512.00 MiB. GPU 1 has a total capacty of 79.33 GiB of which 71.94 MiB is free. Including non-PyTorch memory, this process has 79.25 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 550.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+
+torch.cuda.OutOfMemoryError: torch.cudaCUDA out of memory. Tried to allocate 512.00 MiB. GPU 2 has a total capacty of 79.33 GiB of which 311.94 MiB is free. Including non-PyTorch memory, this process has 79.01 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 390.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF.
+OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 0 has a total capacty of 79.33 GiB of which 471.94 MiB is free. Including non-PyTorch memory, this process has 78.86 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 262.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    trainer.train(dataloader)
+    return forward_call(*args, **kwargs)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return forward_call(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return forward_call(*args, **kwargs)
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    out = F.linear(tensor, weight, bias)
+    out = F.linear(tensor, weight, bias)
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 4 has a total capacty of 79.33 GiB of which 295.94 MiB is free. Including non-PyTorch memory, this process has 79.03 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 230.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 7 has a total capacty of 79.33 GiB of which 159.94 MiB is free. Including non-PyTorch memory, this process has 79.16 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 582.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+    return self._call_impl(*args, **kwargs)
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+        return row_linear(sharded_logits = self.model(
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+        return row_linear(outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    return f(*args, **kwargs)
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+    outputs = self.pipeline_engine.train_batch_iter(    
+return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    out = F.linear(tensor, weight, bias)
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 6 has a total capacty of 79.33 GiB of which 23.94 MiB is free. Including non-PyTorch memory, this process has 79.29 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 678.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    out = F.linear(tensor, weight, bias)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 3 has a total capacty of 79.33 GiB of which 287.94 MiB is free. Including non-PyTorch memory, this process has 79.04 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 294.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return f(*args, **kwargs)
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    out = F.linear(tensor, weight, bias)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 6 has a total capacty of 79.33 GiB of which 423.94 MiB is free. Including non-PyTorch memory, this process has 78.90 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 198.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    return self._call_impl(*args, **kwargs)    
+return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    trainer.train(dataloader)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    out = F.linear(tensor, weight, bias)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 5 has a total capacty of 79.33 GiB of which 167.94 MiB is free. Including non-PyTorch memory, this process has 79.15 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 454.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    out = F.linear(tensor, weight, bias)
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 0 has a total capacty of 79.33 GiB of which 295.94 MiB is free. Including non-PyTorch memory, this process has 79.03 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 358.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)    
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return row_linear(
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = self.pp_block(**new_kwargs)
+        out = F.linear(tensor, weight, bias)outputs = self.pipeline_engine.train_batch_iter(
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    return self._call_impl(*args, **kwargs)
+torch.cuda.  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 2 has a total capacty of 79.33 GiB of which 183.94 MiB is free. Including non-PyTorch memory, this process has 79.14 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 358.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return forward_call(*args, **kwargs)
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+        return self._call_impl(*args, **kwargs)return forward_call(*args, **kwargs)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    out = F.linear(tensor, weight, bias)
+    torch.cudareturn forward_call(*args, **kwargs)
+.OutOfMemoryError  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 5 has a total capacty of 79.33 GiB of which 183.94 MiB is free. Including non-PyTorch memory, this process has 79.14 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 358.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    sharded_logits = self.model(
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return self._call_impl(*args, **kwargs)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    return forward_call(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)    
+return forward_call(*args, **kwargs)  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    sharded_logits = self.model(
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+        output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)hidden_encoder_states = encoder_block(**hidden_encoder_states)    
+
+trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self._call_impl(*args, **kwargs)
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    output = self.pp_block(**new_kwargs)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    output = model(**micro_batch)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    out = F.linear(tensor, weight, bias)
+    return f(*args, **kwargs)    
+return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 0 has a total capacty of 79.33 GiB of which 439.94 MiB is free. Including non-PyTorch memory, this process has 78.89 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 294.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]    
+return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)    
+hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return row_linear(
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    out = F.linear(tensor, weight, bias)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 0 has a total capacty of 79.33 GiB of which 103.94 MiB is free. Including non-PyTorch memory, this process has 79.22 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 550.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return self._call_impl(*args, **kwargs)
+    outputs = self.pipeline_engine.train_batch_iter(
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    out = F.linear(tensor, weight, bias)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    out = F.linear(tensor, weight, bias)
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 3 has a total capacty of 79.33 GiB of which 159.94 MiB is free. Including non-PyTorch memory, this process has 79.16 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 262.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 0 has a total capacty of 79.33 GiB of which 327.94 MiB is free. Including non-PyTorch memory, this process has 79.00 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 326.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    return forward_call(*args, **kwargs)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    return forward_call(*args, **kwargs)    
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    return forward_call(*args, **kwargs)
+    output = self.pp_block(**new_kwargs)
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    out = F.linear(tensor, weight, bias)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 2 has a total capacty of 79.33 GiB of which 135.94 MiB is free. Including non-PyTorch memory, this process has 79.19 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 486.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    sharded_logits = self.model(
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    out = F.linear(tensor, weight, bias)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+    return forward_call(*args, **kwargs)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 4 has a total capacty of 79.33 GiB of which 375.94 MiB is free. Including non-PyTorch memory, this process has 78.95 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 230.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    out = F.linear(tensor, weight, bias)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 2 has a total capacty of 79.33 GiB of which 295.94 MiB is free. Including non-PyTorch memory, this process has 79.03 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 326.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    return forward_call(*args, **kwargs)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+        hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return row_linear(
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+    out = F.linear(tensor, weight, bias)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 6 has a total capacty of 79.33 GiB of which 55.94 MiB is free. Including non-PyTorch memory, this process has 79.26 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 486.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return self._call_impl(*args, **kwargs)
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    sharded_logits = self.model(
+    return self._call_impl(*args, **kwargs)
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return self._call_impl(*args, **kwargs)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    return forward_call(*args, **kwargs)
+    return forward_call(*args, **kwargs)  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    return forward_call(*args, **kwargs)
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    out = F.linear(tensor, weight, bias)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 6 has a total capacty of 79.33 GiB of which 407.94 MiB is free. Including non-PyTorch memory, this process has 78.92 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 294.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return forward_call(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    out = F.linear(tensor, weight, bias)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 2 has a total capacty of 79.33 GiB of which 471.94 MiB is free. Including non-PyTorch memory, this process has 78.86 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 230.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return row_linear(
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    out = F.linear(tensor, weight, bias)
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 7 has a total capacty of 79.33 GiB of which 415.94 MiB is free. Including non-PyTorch memory, this process has 78.91 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 326.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    out = F.linear(tensor, weight, bias)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 3 has a total capacty of 79.33 GiB of which 47.94 MiB is free. Including non-PyTorch memory, this process has 79.27 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 454.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    out = F.linear(tensor, weight, bias)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 0 has a total capacty of 79.33 GiB of which 151.94 MiB is free. Including non-PyTorch memory, this process has 79.17 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 422.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    out = F.linear(tensor, weight, bias)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 1 has a total capacty of 79.33 GiB of which 199.94 MiB is free. Including non-PyTorch memory, this process has 79.12 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 422.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    out = F.linear(tensor, weight, bias)
+    return self._call_impl(*args, **kwargs)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 6 has a total capacty of 79.33 GiB of which 231.94 MiB is free. Including non-PyTorch memory, this process has 79.09 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 390.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    out = F.linear(tensor, weight, bias)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 5 has a total capacty of 79.33 GiB of which 231.94 MiB is free. Including non-PyTorch memory, this process has 79.09 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 390.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    out = F.linear(tensor, weight, bias)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 4 has a total capacty of 79.33 GiB of which 183.94 MiB is free. Including non-PyTorch memory, this process has 79.14 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 422.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    out = F.linear(tensor, weight, bias)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 1 has a total capacty of 79.33 GiB of which 151.94 MiB is free. Including non-PyTorch memory, this process has 79.17 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 390.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+double free or corruption (!prev)
+[2024-12-19 05:42:50,663] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 394443 closing signal SIGTERM
+[2024-12-19 05:42:50,663] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 394444 closing signal SIGTERM
+[2024-12-19 05:42:50,663] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 394445 closing signal SIGTERM
+[2024-12-19 05:42:50,664] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 394446 closing signal SIGTERM
+[2024-12-19 05:42:50,665] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 394447 closing signal SIGTERM
+[2024-12-19 05:42:50,665] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 394448 closing signal SIGTERM
+[2024-12-19 05:42:50,665] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 394449 closing signal SIGTERM
+[2024-12-19 05:42:55,678] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 425494 closing signal SIGTERM
+[2024-12-19 05:42:55,678] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 425495 closing signal SIGTERM
+[2024-12-19 05:42:55,678] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 425496 closing signal SIGTERM
+[2024-12-19 05:42:55,678] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 425498 closing signal SIGTERM
+[2024-12-19 05:42:55,678] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 425499 closing signal SIGTERM
+[2024-12-19 05:42:55,678] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 425500 closing signal SIGTERM
+[2024-12-19 05:42:55,679] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 425501 closing signal SIGTERM
+[2024-12-19 05:43:00,669] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1587920 closing signal SIGTERM
+[2024-12-19 05:43:00,669] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1587921 closing signal SIGTERM
+[2024-12-19 05:43:00,670] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1587922 closing signal SIGTERM
+[2024-12-19 05:43:00,670] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1587924 closing signal SIGTERM
+[2024-12-19 05:43:00,670] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1587925 closing signal SIGTERM
+[2024-12-19 05:43:00,670] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1587926 closing signal SIGTERM
+[2024-12-19 05:43:00,670] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1587927 closing signal SIGTERM
+[2024-12-19 05:43:00,672] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 173698 closing signal SIGTERM
+[2024-12-19 05:43:00,672] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 173700 closing signal SIGTERM
+[2024-12-19 05:43:00,672] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 173701 closing signal SIGTERM
+[2024-12-19 05:43:00,672] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 173702 closing signal SIGTERM
+[2024-12-19 05:43:00,672] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 173703 closing signal SIGTERM
+[2024-12-19 05:43:00,672] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 173704 closing signal SIGTERM
+[2024-12-19 05:43:00,673] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 225588 closing signal SIGTERM
+[2024-12-19 05:43:00,674] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 225589 closing signal SIGTERM
+[2024-12-19 05:43:00,674] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 225590 closing signal SIGTERM
+[2024-12-19 05:43:00,673] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 395357 closing signal SIGTERM
+[2024-12-19 05:43:00,674] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 225591 closing signal SIGTERM
+[2024-12-19 05:43:00,674] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 225592 closing signal SIGTERM
+[2024-12-19 05:43:00,674] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 225593 closing signal SIGTERM
+[2024-12-19 05:43:00,674] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 225594 closing signal SIGTERM
+[2024-12-19 05:43:00,673] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 395358 closing signal SIGTERM
+[2024-12-19 05:43:00,673] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 395359 closing signal SIGTERM
+[2024-12-19 05:43:00,673] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 395360 closing signal SIGTERM
+[2024-12-19 05:43:00,673] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 395361 closing signal SIGTERM
+[2024-12-19 05:43:00,673] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 395362 closing signal SIGTERM
+[2024-12-19 05:43:00,673] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 395364 closing signal SIGTERM
+[2024-12-19 05:43:00,676] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 482605 closing signal SIGTERM
+[2024-12-19 05:43:00,676] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 482606 closing signal SIGTERM
+[2024-12-19 05:43:00,676] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 482608 closing signal SIGTERM
+[2024-12-19 05:43:00,676] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 482609 closing signal SIGTERM
+[2024-12-19 05:43:00,676] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 482610 closing signal SIGTERM
+[2024-12-19 05:43:00,676] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 482611 closing signal SIGTERM
+[2024-12-19 05:43:00,676] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 38292 closing signal SIGTERM
+[2024-12-19 05:43:00,676] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 482612 closing signal SIGTERM
+[2024-12-19 05:43:00,676] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 38293 closing signal SIGTERM
+[2024-12-19 05:43:00,676] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 38295 closing signal SIGTERM
+[2024-12-19 05:43:00,676] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 38296 closing signal SIGTERM
+[2024-12-19 05:43:00,676] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 38297 closing signal SIGTERM
+[2024-12-19 05:43:00,676] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 38298 closing signal SIGTERM
+[2024-12-19 05:43:00,678] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 68471 closing signal SIGTERM
+[2024-12-19 05:43:00,679] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 68472 closing signal SIGTERM
+[2024-12-19 05:43:00,678] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 78262 closing signal SIGTERM
+[2024-12-19 05:43:00,679] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 68473 closing signal SIGTERM
+[2024-12-19 05:43:00,679] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 68475 closing signal SIGTERM
+[2024-12-19 05:43:00,679] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 78263 closing signal SIGTERM
+[2024-12-19 05:43:00,679] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 78265 closing signal SIGTERM
+[2024-12-19 05:43:00,679] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 78266 closing signal SIGTERM
+[2024-12-19 05:43:00,679] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 68476 closing signal SIGTERM
+[2024-12-19 05:43:00,679] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 68477 closing signal SIGTERM
+[2024-12-19 05:43:00,679] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 78267 closing signal SIGTERM
+[2024-12-19 05:43:00,679] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 78268 closing signal SIGTERM
+[2024-12-19 05:43:00,679] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 68478 closing signal SIGTERM
+[2024-12-19 05:43:00,680] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 29630 closing signal SIGTERM
+[2024-12-19 05:43:00,680] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 29631 closing signal SIGTERM
+[2024-12-19 05:43:00,680] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 29633 closing signal SIGTERM
+[2024-12-19 05:43:00,680] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 29634 closing signal SIGTERM
+[2024-12-19 05:43:00,680] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 29635 closing signal SIGTERM
+[2024-12-19 05:43:00,680] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 29636 closing signal SIGTERM
+[2024-12-19 05:43:00,682] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 290897 closing signal SIGTERM
+[2024-12-19 05:43:00,682] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 290898 closing signal SIGTERM
+[2024-12-19 05:43:00,682] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 290899 closing signal SIGTERM
+[2024-12-19 05:43:00,682] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 290900 closing signal SIGTERM
+[2024-12-19 05:43:00,682] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 290901 closing signal SIGTERM
+[2024-12-19 05:43:00,682] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 290902 closing signal SIGTERM
+[2024-12-19 05:43:00,682] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 290903 closing signal SIGTERM
+[2024-12-19 05:43:00,682] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 419207 closing signal SIGTERM
+[2024-12-19 05:43:00,682] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 419208 closing signal SIGTERM
+[2024-12-19 05:43:00,682] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 419209 closing signal SIGTERM
+[2024-12-19 05:43:00,682] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 419210 closing signal SIGTERM
+[2024-12-19 05:43:00,682] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 419211 closing signal SIGTERM
+[2024-12-19 05:43:00,682] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 419213 closing signal SIGTERM
+[2024-12-19 05:43:00,681] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 289698 closing signal SIGTERM
+[2024-12-19 05:43:00,682] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 289699 closing signal SIGTERM
+[2024-12-19 05:43:00,682] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 289701 closing signal SIGTERM
+[2024-12-19 05:43:00,682] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 289702 closing signal SIGTERM
+[2024-12-19 05:43:00,682] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 289703 closing signal SIGTERM
+[2024-12-19 05:43:00,682] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 289704 closing signal SIGTERM
+[2024-12-19 05:43:00,682] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 289705 closing signal SIGTERM
+[2024-12-19 05:43:00,683] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 419986 closing signal SIGTERM
+[2024-12-19 05:43:00,682] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 439185 closing signal SIGTERM
+[2024-12-19 05:43:00,684] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 419987 closing signal SIGTERM
+[2024-12-19 05:43:00,683] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 439186 closing signal SIGTERM
+[2024-12-19 05:43:00,684] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 419988 closing signal SIGTERM
+[2024-12-19 05:43:00,683] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 439187 closing signal SIGTERM
+[2024-12-19 05:43:00,683] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 439188 closing signal SIGTERM
+[2024-12-19 05:43:00,683] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 439190 closing signal SIGTERM
+[2024-12-19 05:43:00,683] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 439191 closing signal SIGTERM
+[2024-12-19 05:43:00,684] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 419989 closing signal SIGTERM
+[2024-12-19 05:43:00,684] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 419990 closing signal SIGTERM
+[2024-12-19 05:43:00,684] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 419991 closing signal SIGTERM
+[2024-12-19 05:43:00,684] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 419993 closing signal SIGTERM
+[2024-12-19 05:43:00,681] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 54285 closing signal SIGTERM
+[2024-12-19 05:43:00,681] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 54286 closing signal SIGTERM
+[2024-12-19 05:43:00,681] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 54288 closing signal SIGTERM
+[2024-12-19 05:43:00,681] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 54289 closing signal SIGTERM
+[2024-12-19 05:43:00,681] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 54290 closing signal SIGTERM
+[2024-12-19 05:43:00,681] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 54291 closing signal SIGTERM
+[2024-12-19 05:43:00,681] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 54292 closing signal SIGTERM
+[2024-12-19 05:43:00,683] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 335359 closing signal SIGTERM
+[2024-12-19 05:43:00,684] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 335360 closing signal SIGTERM
+[2024-12-19 05:43:00,684] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 335361 closing signal SIGTERM
+[2024-12-19 05:43:00,684] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 335362 closing signal SIGTERM
+[2024-12-19 05:43:00,684] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 335363 closing signal SIGTERM
+[2024-12-19 05:43:00,684] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 335364 closing signal SIGTERM
+[2024-12-19 05:43:00,684] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 335365 closing signal SIGTERM
+[2024-12-19 05:43:00,684] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 29345 closing signal SIGTERM
+[2024-12-19 05:43:00,684] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 29347 closing signal SIGTERM
+[2024-12-19 05:43:00,684] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 29348 closing signal SIGTERM
+[2024-12-19 05:43:00,684] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 29349 closing signal SIGTERM
+[2024-12-19 05:43:00,684] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 29350 closing signal SIGTERM
+[2024-12-19 05:43:00,684] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 29351 closing signal SIGTERM
+[2024-12-19 05:43:00,684] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 29352 closing signal SIGTERM
+[2024-12-19 05:43:00,684] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1396487 closing signal SIGTERM
+[2024-12-19 05:43:00,685] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1396488 closing signal SIGTERM
+[2024-12-19 05:43:00,685] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1396490 closing signal SIGTERM
+[2024-12-19 05:43:00,685] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1396491 closing signal SIGTERM
+[2024-12-19 05:43:00,685] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1396492 closing signal SIGTERM
+[2024-12-19 05:43:00,685] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1396493 closing signal SIGTERM
+[2024-12-19 05:43:00,682] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 224370 closing signal SIGTERM
+[2024-12-19 05:43:00,682] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 224371 closing signal SIGTERM
+[2024-12-19 05:43:00,682] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 224372 closing signal SIGTERM
+[2024-12-19 05:43:00,682] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 224373 closing signal SIGTERM
+[2024-12-19 05:43:00,682] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 224374 closing signal SIGTERM
+[2024-12-19 05:43:00,682] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 224375 closing signal SIGTERM
+[2024-12-19 05:43:00,682] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 224377 closing signal SIGTERM
+[2024-12-19 05:43:00,688] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 104529 closing signal SIGTERM
+[2024-12-19 05:43:00,688] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 104530 closing signal SIGTERM
+[2024-12-19 05:43:00,688] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 104531 closing signal SIGTERM
+[2024-12-19 05:43:00,688] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 104532 closing signal SIGTERM
+[2024-12-19 05:43:00,688] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 104533 closing signal SIGTERM
+[2024-12-19 05:43:00,688] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 104534 closing signal SIGTERM
+[2024-12-19 05:43:00,688] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 104535 closing signal SIGTERM
+[2024-12-19 05:43:00,689] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 55244 closing signal SIGTERM
+[2024-12-19 05:43:00,690] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 55251 closing signal SIGTERM
+[2024-12-19 05:43:03,941] torch.distributed.elastic.multiprocessing.api: [ERROR] failed (exitcode: 1) local_rank: 0 (pid: 173697) of binary: /fsx/nouamane/miniconda/envs/2-1-cu121/bin/python
+[2024-12-19 05:43:03,972] torch.distributed.elastic.multiprocessing.errors.error_handler: [ERROR] no error file defined for parent, to copy child error file (/tmp/torchelastic_pf42i90l/13502056_3bzkft6q/attempt_0/0/error.json)
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 264, in launch_agent
+    raise ChildFailedError(
+torch.distributed.elastic.multiprocessing.errors.ChildFailedError: 
+============================================================
+/fsx/nouamane/projects/nanotron/run_train.py FAILED
+------------------------------------------------------------
+Failures:
+[1]:
+  time      : 2024-12-19_05:42:41
+  host      : ip-26-0-161-138.ec2.internal
+  rank      : 18 (local_rank: 2)
+  exitcode  : 1 (pid: 173699)
+  error_file: /tmp/torchelastic_pf42i90l/13502056_3bzkft6q/attempt_0/2/error.json
+  traceback : Traceback (most recent call last):
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+      return f(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+      outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+      outputs = self.pipeline_engine.train_batch_iter(
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+      output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+      output = model(**micro_batch)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+      sharded_logits = self.model(
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+      return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+      hidden_encoder_states = encoder_block(**hidden_encoder_states)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+      output = self.pp_block(**new_kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+      hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+      hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+      hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+      return row_linear(
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+      return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+      return super().apply(*args, **kwargs)  # type: ignore[misc]
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+      out = F.linear(tensor, weight, bias)
+  torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 2 has a total capacty of 79.33 GiB of which 471.94 MiB is free. Including non-PyTorch memory, this process has 78.86 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 230.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  
+------------------------------------------------------------
+Root Cause (first observed failure):
+[0]:
+  time      : 2024-12-19_05:42:41
+  host      : ip-26-0-161-138.ec2.internal
+  rank      : 16 (local_rank: 0)
+  exitcode  : 1 (pid: 173697)
+  error_file: /tmp/torchelastic_pf42i90l/13502056_3bzkft6q/attempt_0/0/error.json
+  traceback : Traceback (most recent call last):
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+      return f(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+      outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+      outputs = self.pipeline_engine.train_batch_iter(
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+      output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+      output = model(**micro_batch)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+      sharded_logits = self.model(
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+      return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+      hidden_encoder_states = encoder_block(**hidden_encoder_states)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+      output = self.pp_block(**new_kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+      hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+      hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+      hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+      return row_linear(
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+      return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+      return super().apply(*args, **kwargs)  # type: ignore[misc]
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+      out = F.linear(tensor, weight, bias)
+  torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 0 has a total capacty of 79.33 GiB of which 215.94 MiB is free. Including non-PyTorch memory, this process has 79.11 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 518.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  
+============================================================
+[2024-12-19 05:43:04,052] torch.distributed.elastic.multiprocessing.api: [ERROR] failed (exitcode: 1) local_rank: 0 (pid: 439184) of binary: /fsx/nouamane/miniconda/envs/2-1-cu121/bin/python
+[2024-12-19 05:43:04,082] torch.distributed.elastic.multiprocessing.api: [ERROR] failed (exitcode: 1) local_rank: 6 (pid: 395363) of binary: /fsx/nouamane/miniconda/envs/2-1-cu121/bin/python
+[2024-12-19 05:43:04,086] torch.distributed.elastic.multiprocessing.errors.error_handler: [ERROR] no error file defined for parent, to copy child error file (/tmp/torchelastic_nw_q3g5k/13502056_f6hom5dg/attempt_0/0/error.json)
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 264, in launch_agent
+    raise ChildFailedError(
+torch.distributed.elastic.multiprocessing.errors.ChildFailedError: 
+============================================================
+/fsx/nouamane/projects/nanotron/run_train.py FAILED
+------------------------------------------------------------
+Failures:
+[1]:
+  time      : 2024-12-19_05:42:42
+  host      : ip-26-0-169-86.ec2.internal
+  rank      : 245 (local_rank: 5)
+  exitcode  : 1 (pid: 439189)
+  error_file: /tmp/torchelastic_nw_q3g5k/13502056_f6hom5dg/attempt_0/5/error.json
+  traceback : Traceback (most recent call last):
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+      return f(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+      outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+      outputs = self.pipeline_engine.train_batch_iter(
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+      output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+      output = model(**micro_batch)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+      sharded_logits = self.model(
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+      return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+      hidden_encoder_states = encoder_block(**hidden_encoder_states)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+      output = self.pp_block(**new_kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+      hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+      hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+      hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+      return row_linear(
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+      return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+      return super().apply(*args, **kwargs)  # type: ignore[misc]
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+      out = F.linear(tensor, weight, bias)
+  torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 5 has a total capacty of 79.33 GiB of which 263.94 MiB is free. Including non-PyTorch memory, this process has 79.06 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 358.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  
+------------------------------------------------------------
+Root Cause (first observed failure):
+[0]:
+  time      : 2024-12-19_05:42:42
+  host      : ip-26-0-169-86.ec2.internal
+  rank      : 240 (local_rank: 0)
+  exitcode  : 1 (pid: 439184)
+  error_file: /tmp/torchelastic_nw_q3g5k/13502056_f6hom5dg/attempt_0/0/error.json
+  traceback : Traceback (most recent call last):
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+      return f(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+      outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+      outputs = self.pipeline_engine.train_batch_iter(
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+      output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+      output = model(**micro_batch)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+      sharded_logits = self.model(
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+      return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+      hidden_encoder_states = encoder_block(**hidden_encoder_states)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+      output = self.pp_block(**new_kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+      hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+      hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+      hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+      return row_linear(
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+      return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+      return super().apply(*args, **kwargs)  # type: ignore[misc]
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+      out = F.linear(tensor, weight, bias)
+  torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 0 has a total capacty of 79.33 GiB of which 343.94 MiB is free. Including non-PyTorch memory, this process has 78.98 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 390.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  
+============================================================
+[2024-12-19 05:43:04,123] torch.distributed.elastic.multiprocessing.errors.error_handler: [ERROR] no error file defined for parent, to copy child error file (/tmp/torchelastic_aq50x5o4/13502056_i8vy33_w/attempt_0/6/error.json)
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 264, in launch_agent
+    raise ChildFailedError(
+torch.distributed.elastic.multiprocessing.errors.ChildFailedError: 
+============================================================
+/fsx/nouamane/projects/nanotron/run_train.py FAILED
+------------------------------------------------------------
+Failures:
+  <NO_OTHER_FAILURES>
+------------------------------------------------------------
+Root Cause (first observed failure):
+[0]:
+  time      : 2024-12-19_05:42:42
+  host      : ip-26-0-169-247.ec2.internal
+  rank      : 238 (local_rank: 6)
+  exitcode  : 1 (pid: 395363)
+  error_file: /tmp/torchelastic_aq50x5o4/13502056_i8vy33_w/attempt_0/6/error.json
+  traceback : Traceback (most recent call last):
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+      return f(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+      outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+      outputs = self.pipeline_engine.train_batch_iter(
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+      output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 50, in forward
+      output = model(**micro_batch)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1013, in forward
+      sharded_logits = self.model(
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+      return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+      hidden_encoder_states = encoder_block(**hidden_encoder_states)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 151, in forward
+      output = self.pp_block(**new_kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+      hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+      hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+      hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+      return row_linear(
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+      return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+      return super().apply(*args, **kwargs)  # type: ignore[misc]
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+      out = F.linear(tensor, weight, bias)
+  torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 6 has a total capacty of 79.33 GiB of which 391.94 MiB is free. Including non-PyTorch memory, this process has 78.94 GiB memory in use. Of the allocated memory 69.66 GiB is allocated by PyTorch, and 230.13 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  
+============================================================
+srun: error: ip-26-0-161-138: task 3: Exited with exit code 1
+srun: launch/slurm: _step_signal: Terminating StepId=13502056.0
+slurmstepd: error: *** STEP 13502056.0 ON ip-26-0-161-78 CANCELLED AT 2024-12-19T05:43:04 ***
+[2024-12-19 05:43:04,235] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-19 05:43:04,234] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-19 05:43:04,234] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 54289 closing signal SIGTERM
+[2024-12-19 05:43:04,235] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 78268 closing signal SIGTERM
+[2024-12-19 05:43:04,235] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-19 05:43:04,234] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-19 05:43:04,235] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 482606 closing signal SIGTERM
+[2024-12-19 05:43:04,234] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 54291 closing signal SIGTERM
+[2024-12-19 05:43:04,235] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1396493 closing signal SIGTERM
+[2024-12-19 05:43:04,234] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 54292 closing signal SIGTERM
+[2024-12-19 05:43:04,235] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 482608 closing signal SIGTERM
+[2024-12-19 05:43:04,235] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 482610 closing signal SIGTERM
+[2024-12-19 05:43:04,234] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-19 05:43:04,235] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 482612 closing signal SIGTERM
+[2024-12-19 05:43:04,237] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-19 05:43:04,234] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1587925 closing signal SIGTERM
+[2024-12-19 05:43:04,237] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 65242 closing signal SIGTERM
+[2024-12-19 05:43:04,236] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-19 05:43:04,235] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1587927 closing signal SIGTERM
+[2024-12-19 05:43:04,237] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 65243 closing signal SIGTERM
+[2024-12-19 05:43:04,235] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-19 05:43:04,237] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 65244 closing signal SIGTERM
+[2024-12-19 05:43:04,236] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-19 05:43:04,237] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 65245 closing signal SIGTERM
+[2024-12-19 05:43:04,237] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 21860 closing signal SIGTERM
+[2024-12-19 05:43:04,237] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 65246 closing signal SIGTERM
+[2024-12-19 05:43:04,235] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-19 05:43:04,237] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 65247 closing signal SIGTERM
+[2024-12-19 05:43:04,235] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 225589 closing signal SIGTERM
+[2024-12-19 05:43:04,237] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 65248 closing signal SIGTERM
+[2024-12-19 05:43:04,236] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 425501 closing signal SIGTERM
+[2024-12-19 05:43:04,235] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 225590 closing signal SIGTERM
+[2024-12-19 05:43:04,235] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 113560 closing signal SIGTERM
+[2024-12-19 05:43:04,237] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 21862 closing signal SIGTERM
+[2024-12-19 05:43:04,236] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 113561 closing signal SIGTERM
+[2024-12-19 05:43:04,235] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 225593 closing signal SIGTERM
+[2024-12-19 05:43:04,237] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 21863 closing signal SIGTERM
+[2024-12-19 05:43:04,236] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 113562 closing signal SIGTERM
+[2024-12-19 05:43:04,232] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-19 05:43:04,236] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-19 05:43:04,237] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 21864 closing signal SIGTERM
+[2024-12-19 05:43:04,236] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 113563 closing signal SIGTERM
+[2024-12-19 05:43:04,233] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 224371 closing signal SIGTERM
+[2024-12-19 05:43:04,237] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-19 05:43:04,236] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 113564 closing signal SIGTERM
+[2024-12-19 05:43:04,236] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 29636 closing signal SIGTERM
+[2024-12-19 05:43:04,236] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 113565 closing signal SIGTERM
+[2024-12-19 05:43:04,237] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 21865 closing signal SIGTERM
+[2024-12-19 05:43:04,236] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-19 05:43:04,233] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 224372 closing signal SIGTERM
+[2024-12-19 05:43:04,237] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 290903 closing signal SIGTERM
+[2024-12-19 05:43:04,233] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 224373 closing signal SIGTERM
+[2024-12-19 05:43:04,237] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 176313 closing signal SIGTERM
+[2024-12-19 05:43:04,233] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 224375 closing signal SIGTERM
+[2024-12-19 05:43:04,236] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-19 05:43:04,237] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 176314 closing signal SIGTERM
+[2024-12-19 05:43:04,233] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 224377 closing signal SIGTERM
+[2024-12-19 05:43:04,236] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-19 05:43:04,237] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 176315 closing signal SIGTERM
+[2024-12-19 05:43:04,235] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-19 05:43:04,237] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-19 05:43:04,234] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-19 05:43:04,237] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 29347 closing signal SIGTERM
+[2024-12-19 05:43:04,237] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 176317 closing signal SIGTERM
+[2024-12-19 05:43:04,236] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 38292 closing signal SIGTERM
+[2024-12-19 05:43:04,237] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1396578 closing signal SIGTERM
+[2024-12-19 05:43:04,236] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-19 05:43:04,237] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 29349 closing signal SIGTERM
+[2024-12-19 05:43:04,235] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 55244 closing signal SIGTERM
+[2024-12-19 05:43:04,238] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 225178 closing signal SIGTERM
+[2024-12-19 05:43:04,237] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 176318 closing signal SIGTERM
+[2024-12-19 05:43:04,237] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1396579 closing signal SIGTERM
+[2024-12-19 05:43:04,237] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 29351 closing signal SIGTERM
+[2024-12-19 05:43:04,237] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 289702 closing signal SIGTERM
+[2024-12-19 05:43:04,238] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 225180 closing signal SIGTERM
+[2024-12-19 05:43:04,237] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1396581 closing signal SIGTERM
+[2024-12-19 05:43:04,237] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 176319 closing signal SIGTERM
+[2024-12-19 05:43:04,235] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 55251 closing signal SIGTERM
+[2024-12-19 05:43:04,238] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 225181 closing signal SIGTERM
+[2024-12-19 05:43:04,237] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1396582 closing signal SIGTERM
+[2024-12-19 05:43:04,237] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-19 05:43:04,238] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-19 05:43:04,237] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-19 05:43:04,237] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 289704 closing signal SIGTERM
+[2024-12-19 05:43:04,238] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 225182 closing signal SIGTERM
+[2024-12-19 05:43:04,237] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1396583 closing signal SIGTERM
+[2024-12-19 05:43:04,237] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 335365 closing signal SIGTERM
+[2024-12-19 05:43:04,237] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-19 05:43:04,238] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 225183 closing signal SIGTERM
+[2024-12-19 05:43:04,238] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 419987 closing signal SIGTERM
+[2024-12-19 05:43:04,238] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 225184 closing signal SIGTERM
+[2024-12-19 05:43:04,237] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 46050 closing signal SIGTERM
+[2024-12-19 05:43:04,237] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1396584 closing signal SIGTERM
+[2024-12-19 05:43:04,237] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 68473 closing signal SIGTERM
+[2024-12-19 05:43:04,238] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 419989 closing signal SIGTERM
+[2024-12-19 05:43:04,237] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 46051 closing signal SIGTERM
+[2024-12-19 05:43:04,237] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-19 05:43:04,238] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 419991 closing signal SIGTERM
+[2024-12-19 05:43:04,235] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-19 05:43:04,237] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 46052 closing signal SIGTERM
+[2024-12-19 05:43:04,236] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-19 05:43:04,237] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 68475 closing signal SIGTERM
+[2024-12-19 05:43:04,238] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 419993 closing signal SIGTERM
+[2024-12-19 05:43:04,238] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 419210 closing signal SIGTERM
+[2024-12-19 05:43:04,238] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 46053 closing signal SIGTERM
+[2024-12-19 05:43:04,236] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 292361 closing signal SIGTERM
+[2024-12-19 05:43:04,236] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 394444 closing signal SIGTERM
+[2024-12-19 05:43:04,236] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 394445 closing signal SIGTERM
+[2024-12-19 05:43:04,238] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 46055 closing signal SIGTERM
+[2024-12-19 05:43:04,237] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 68477 closing signal SIGTERM
+[2024-12-19 05:43:04,236] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 292362 closing signal SIGTERM
+[2024-12-19 05:43:04,236] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 394447 closing signal SIGTERM
+[2024-12-19 05:43:04,238] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 46056 closing signal SIGTERM
+[2024-12-19 05:43:04,236] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 394448 closing signal SIGTERM
+[2024-12-19 05:43:04,236] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 292363 closing signal SIGTERM
+[2024-12-19 05:43:04,236] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 394449 closing signal SIGTERM
+[2024-12-19 05:43:04,236] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 292364 closing signal SIGTERM
+[2024-12-19 05:43:04,236] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 292365 closing signal SIGTERM
+[2024-12-19 05:43:04,236] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 292366 closing signal SIGTERM
+[2024-12-19 05:43:04,236] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 292367 closing signal SIGTERM
+[2024-12-19 05:43:04,238] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-19 05:43:04,239] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 322841 closing signal SIGTERM
+[2024-12-19 05:43:04,239] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 322842 closing signal SIGTERM
+[2024-12-19 05:43:04,239] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 322843 closing signal SIGTERM
+[2024-12-19 05:43:04,239] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 322844 closing signal SIGTERM
+[2024-12-19 05:43:04,239] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 322846 closing signal SIGTERM
+[2024-12-19 05:43:04,239] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 322847 closing signal SIGTERM
+[2024-12-19 05:43:04,238] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-19 05:43:04,238] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 104529 closing signal SIGTERM
+[2024-12-19 05:43:04,238] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 104530 closing signal SIGTERM
+[2024-12-19 05:43:04,239] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 104532 closing signal SIGTERM
+[2024-12-19 05:43:04,239] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 104534 closing signal SIGTERM
+[2024-12-19 05:43:04,239] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 104535 closing signal SIGTERM
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 878, in _invoke_run
+    run_result = self._monitor_workers(self._worker_group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/local_elastic_agent.py", line 307, in _monitor_workers
+    result = self._pcontext.wait(0)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 288, in wait
+    return self._poll()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 668, in _poll
+    self.close()  # terminate all running procs
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 331, in close
+    self._close(death_sig=death_sig, timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 713, in _close
+    handler.proc.wait(time_to_wait)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1209, in wait
+    return self._wait(timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1953, in _wait
+    time.sleep(delay)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 54212 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 878, in _invoke_run
+    run_result = self._monitor_workers(self._worker_group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/local_elastic_agent.py", line 307, in _monitor_workers
+    result = self._pcontext.wait(0)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 288, in wait
+    return self._poll()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 668, in _poll
+    self.close()  # terminate all running procs
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 331, in close
+srun: error: ip-26-0-169-247: task 30: Terminated
+    self._close(death_sig=death_sig, timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 713, in _close
+    handler.proc.wait(time_to_wait)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1209, in wait
+    return self._wait(timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1953, in _wait
+    time.sleep(delay)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 419134 got signal: 15
+srun: error: ip-26-0-169-86: task 25: Terminated
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 878, in _invoke_run
+    run_result = self._monitor_workers(self._worker_group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/local_elastic_agent.py", line 307, in _monitor_workers
+    result = self._pcontext.wait(0)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 288, in wait
+    return self._poll()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 668, in _poll
+    self.close()  # terminate all running procs
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 331, in close
+    self._close(death_sig=death_sig, timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 713, in _close
+    handler.proc.wait(time_to_wait)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1209, in wait
+    return self._wait(timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1953, in _wait
+    time.sleep(delay)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 38218 got signal: 15
+srun: error: ip-26-0-161-78: task 0: Exited with exit code 1
+srun: error: ip-26-0-169-132: task 26: Exited with exit code 1
+[2024-12-19 05:43:04,661] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-161-103.ec2.internal_55173_0' has failed to send a keep-alive heartbeat to the rendezvous '13502056' due to an error of type RendezvousConnectionError.
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 878, in _invoke_run
+    run_result = self._monitor_workers(self._worker_group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/local_elastic_agent.py", line 307, in _monitor_workers
+    result = self._pcontext.wait(0)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 288, in wait
+    return self._poll()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 668, in _poll
+    self.close()  # terminate all running procs
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 331, in close
+    self._close(death_sig=death_sig, timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 713, in _close
+    handler.proc.wait(time_to_wait)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1209, in wait
+    return self._wait(timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1953, in _wait
+    time.sleep(delay)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 225517 got signal: 15
+srun: error: ip-26-0-168-120: task 23: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 878, in _invoke_run
+    run_result = self._monitor_workers(self._worker_group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/local_elastic_agent.py", line 307, in _monitor_workers
+    result = self._pcontext.wait(0)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 288, in wait
+    return self._poll()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 668, in _poll
+    self.close()  # terminate all running procs
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 331, in close
+    self._close(death_sig=death_sig, timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 713, in _close
+    handler.proc.wait(time_to_wait)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1209, in wait
+    return self._wait(timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1953, in _wait
+    time.sleep(delay)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 29273 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 878, in _invoke_run
+    run_result = self._monitor_workers(self._worker_group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/local_elastic_agent.py", line 307, in _monitor_workers
+    result = self._pcontext.wait(0)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 288, in wait
+    return self._poll()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 668, in _poll
+    self.close()  # terminate all running procs
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 331, in close
+    self._close(death_sig=death_sig, timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 713, in _close
+    handler.proc.wait(time_to_wait)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1209, in wait
+    return self._wait(timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1953, in _wait
+    time.sleep(delay)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 419916 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 878, in _invoke_run
+    run_result = self._monitor_workers(self._worker_group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/local_elastic_agent.py", line 307, in _monitor_workers
+    result = self._pcontext.wait(0)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 288, in wait
+    return self._poll()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 668, in _poll
+    self.close()  # terminate all running procs
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 331, in close
+    self._close(death_sig=death_sig, timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 713, in _close
+    handler.proc.wait(time_to_wait)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1209, in wait
+    return self._wait(timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1953, in _wait
+    time.sleep(delay)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 482531 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 878, in _invoke_run
+    run_result = self._monitor_workers(self._worker_group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/local_elastic_agent.py", line 307, in _monitor_workers
+    result = self._pcontext.wait(0)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 288, in wait
+    return self._poll()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 668, in _poll
+    self.close()  # terminate all running procs
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 331, in close
+    self._close(death_sig=death_sig, timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 713, in _close
+    handler.proc.wait(time_to_wait)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1209, in wait
+    return self._wait(timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1953, in _wait
+    time.sleep(delay)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 289626 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 45976 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 878, in _invoke_run
+    run_result = self._monitor_workers(self._worker_group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/local_elastic_agent.py", line 307, in _monitor_workers
+    result = self._pcontext.wait(0)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 288, in wait
+    return self._poll()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 668, in _poll
+    self.close()  # terminate all running procs
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 331, in close
+    self._close(death_sig=death_sig, timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 713, in _close
+    handler.proc.wait(time_to_wait)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1209, in wait
+    return self._wait(timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1953, in _wait
+    time.sleep(delay)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 1587848 got signal: 15
+srun: error: ip-26-0-162-14: task 8: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 878, in _invoke_run
+    run_result = self._monitor_workers(self._worker_group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/local_elastic_agent.py", line 307, in _monitor_workers
+    result = self._pcontext.wait(0)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 288, in wait
+    return self._poll()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 668, in _poll
+    self.close()  # terminate all running procs
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 331, in close
+    self._close(death_sig=death_sig, timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 713, in _close
+    handler.proc.wait(time_to_wait)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1209, in wait
+    return self._wait(timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1953, in _wait
+    time.sleep(delay)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 68399 got signal: 15
+srun: error: ip-26-0-167-51: task 17: Exited with exit code 1
+srun: error: ip-26-0-169-139: task 27: Exited with exit code 1
+srun: error: ip-26-0-161-221: task 7: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 113485 got signal: 15
+srun: error: ip-26-0-167-217: task 20: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 878, in _invoke_run
+    run_result = self._monitor_workers(self._worker_group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/local_elastic_agent.py", line 307, in _monitor_workers
+    result = self._pcontext.wait(0)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 288, in wait
+    return self._poll()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 668, in _poll
+    self.close()  # terminate all running procs
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 331, in close
+    self._close(death_sig=death_sig, timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 713, in _close
+    handler.proc.wait(time_to_wait)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1209, in wait
+    return self._wait(timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1953, in _wait
+    time.sleep(delay)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 224298 got signal: 15
+[2024-12-19 05:43:05,348] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-161-153.ec2.internal_176240_0' has failed to send a keep-alive heartbeat to the rendezvous '13502056' due to an error of type RendezvousConnectionError.
+srun: error: ip-26-0-169-207: task 28: Exited with exit code 1
+[2024-12-19 05:43:05,399] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-161-142.ec2.internal_425422_0' has failed to send a keep-alive heartbeat to the rendezvous '13502056' due to an error of type RendezvousConnectionError.
+srun: error: ip-26-0-163-220: task 12: Exited with exit code 1
+srun: error: ip-26-0-168-30: task 22: Exited with exit code 1
+[2024-12-19 05:43:05,425] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-163-158.ec2.internal_1396505_0' has failed to send a keep-alive heartbeat to the rendezvous '13502056' due to an error of type RendezvousConnectionError.
+[2024-12-19 05:43:05,450] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-164-236.ec2.internal_225106_0' has failed to send a keep-alive heartbeat to the rendezvous '13502056' due to an error of type RendezvousConnectionError.
+[2024-12-19 05:43:05,449] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-168-238.ec2.internal_394372_0' has failed to send a keep-alive heartbeat to the rendezvous '13502056' due to an error of type RendezvousConnectionError.
+[2024-12-19 05:43:05,455] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-163-134.ec2.internal_1396413_0' has failed to send a keep-alive heartbeat to the rendezvous '13502056' due to an error of type RendezvousConnectionError.
+[2024-12-19 05:43:05,498] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-163-226.ec2.internal_65170_0' has failed to send a keep-alive heartbeat to the rendezvous '13502056' due to an error of type RendezvousConnectionError.
+[2024-12-19 05:43:05,523] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-169-239.ec2.internal_104456_0' has failed to send a keep-alive heartbeat to the rendezvous '13502056' due to an error of type RendezvousConnectionError.
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 21783 got signal: 15
+[2024-12-19 05:43:05,542] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-167-177.ec2.internal_292288_0' has failed to send a keep-alive heartbeat to the rendezvous '13502056' due to an error of type RendezvousConnectionError.
+[2024-12-19 05:43:05,550] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-161-178.ec2.internal_78188_0' has failed to send a keep-alive heartbeat to the rendezvous '13502056' due to an error of type RendezvousConnectionError.
+[2024-12-19 05:43:05,593] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-170-31.ec2.internal_335287_0' has failed to send a keep-alive heartbeat to the rendezvous '13502056' due to an error of type RendezvousConnectionError.
+[2024-12-19 05:43:05,600] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-163-147.ec2.internal_29558_0' has failed to send a keep-alive heartbeat to the rendezvous '13502056' due to an error of type RendezvousConnectionError.
+[2024-12-19 05:43:05,627] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-167-175.ec2.internal_290824_0' has failed to send a keep-alive heartbeat to the rendezvous '13502056' due to an error of type RendezvousConnectionError.
+[2024-12-19 05:43:05,631] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-167-245.ec2.internal_322767_0' has failed to send a keep-alive heartbeat to the rendezvous '13502056' due to an error of type RendezvousConnectionError.
+srun: error: ip-26-0-165-24: task 15: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 878, in _invoke_run
+    run_result = self._monitor_workers(self._worker_group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/local_elastic_agent.py", line 307, in _monitor_workers
+    result = self._pcontext.wait(0)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 288, in wait
+    return self._poll()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 668, in _poll
+    self.close()  # terminate all running procs
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 331, in close
+    self._close(death_sig=death_sig, timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 713, in _close
+    handler.proc.wait(time_to_wait)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1209, in wait
+    return self._wait(timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1953, in _wait
+    time.sleep(delay)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 104456 got signal: 15
+srun: error: ip-26-0-165-38: task 16: Exited with exit code 1
+srun: error: ip-26-0-161-123: task 2: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 65170 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 225106 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 1396505 got signal: 15
+srun: error: ip-26-0-169-239: task 29: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 176240 got signal: 15
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 322767 got signal: 15
+srun: error: ip-26-0-164-236: task 14: Exited with exit code 1
+srun: error: ip-26-0-163-226: task 13: Exited with exit code 1
+srun: error: ip-26-0-163-158: task 11: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 878, in _invoke_run
+    run_result = self._monitor_workers(self._worker_group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/local_elastic_agent.py", line 307, in _monitor_workers
+    result = self._pcontext.wait(0)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 288, in wait
+    return self._poll()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 668, in _poll
+    self.close()  # terminate all running procs
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 331, in close
+    self._close(death_sig=death_sig, timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 713, in _close
+    handler.proc.wait(time_to_wait)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1209, in wait
+    return self._wait(timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1953, in _wait
+    time.sleep(delay)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 425422 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 292288 got signal: 15
+srun: error: ip-26-0-167-245: task 21: Exited with exit code 1
+srun: error: ip-26-0-161-153: task 5: Exited with exit code 1
+srun: error: ip-26-0-161-142: task 4: Exited with exit code 1
+srun: error: ip-26-0-167-177: task 19: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 878, in _invoke_run
+    run_result = self._monitor_workers(self._worker_group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/local_elastic_agent.py", line 307, in _monitor_workers
+    result = self._pcontext.wait(0)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 288, in wait
+    return self._poll()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 668, in _poll
+    self.close()  # terminate all running procs
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 331, in close
+    self._close(death_sig=death_sig, timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 713, in _close
+    handler.proc.wait(time_to_wait)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1209, in wait
+    return self._wait(timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1953, in _wait
+    time.sleep(delay)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 394372 got signal: 15
+srun: error: ip-26-0-168-238: task 24: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 878, in _invoke_run
+    run_result = self._monitor_workers(self._worker_group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/local_elastic_agent.py", line 307, in _monitor_workers
+    result = self._pcontext.wait(0)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 288, in wait
+    return self._poll()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 668, in _poll
+    self.close()  # terminate all running procs
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 331, in close
+    self._close(death_sig=death_sig, timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 713, in _close
+    handler.proc.wait(time_to_wait)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1209, in wait
+    return self._wait(timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1953, in _wait
+    time.sleep(delay)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 335287 got signal: 15
+[2024-12-19 05:43:09,663] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-161-103.ec2.internal_55173_0' has failed to send a keep-alive heartbeat to the rendezvous '13502056' due to an error of type RendezvousConnectionError.
+srun: error: ip-26-0-170-31: task 31: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 878, in _invoke_run
+    run_result = self._monitor_workers(self._worker_group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/local_elastic_agent.py", line 307, in _monitor_workers
+    result = self._pcontext.wait(0)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 288, in wait
+    return self._poll()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 668, in _poll
+    self.close()  # terminate all running procs
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 331, in close
+    self._close(death_sig=death_sig, timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 713, in _close
+    handler.proc.wait(time_to_wait)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1209, in wait
+    return self._wait(timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1953, in _wait
+    time.sleep(delay)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 55173 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 878, in _invoke_run
+    run_result = self._monitor_workers(self._worker_group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/local_elastic_agent.py", line 307, in _monitor_workers
+    result = self._pcontext.wait(0)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 288, in wait
+    return self._poll()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 668, in _poll
+    self.close()  # terminate all running procs
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 331, in close
+    self._close(death_sig=death_sig, timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 713, in _close
+    handler.proc.wait(time_to_wait)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1209, in wait
+    return self._wait(timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1953, in _wait
+    time.sleep(delay)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 78188 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 878, in _invoke_run
+    run_result = self._monitor_workers(self._worker_group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/local_elastic_agent.py", line 307, in _monitor_workers
+    result = self._pcontext.wait(0)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 288, in wait
+    return self._poll()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 668, in _poll
+    self.close()  # terminate all running procs
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 331, in close
+    self._close(death_sig=death_sig, timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 713, in _close
+    handler.proc.wait(time_to_wait)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1209, in wait
+    return self._wait(timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1953, in _wait
+    time.sleep(delay)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 1396413 got signal: 15
+[2024-12-19 05:43:10,602] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-163-147.ec2.internal_29558_0' has failed to send a keep-alive heartbeat to the rendezvous '13502056' due to an error of type RendezvousConnectionError.
+[2024-12-19 05:43:10,629] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-167-175.ec2.internal_290824_0' has failed to send a keep-alive heartbeat to the rendezvous '13502056' due to an error of type RendezvousConnectionError.
+srun: error: ip-26-0-161-178: task 6: Exited with exit code 1
+srun: error: ip-26-0-163-134: task 9: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 878, in _invoke_run
+    run_result = self._monitor_workers(self._worker_group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/local_elastic_agent.py", line 307, in _monitor_workers
+    result = self._pcontext.wait(0)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 288, in wait
+    return self._poll()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 668, in _poll
+    self.close()  # terminate all running procs
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 331, in close
+    self._close(death_sig=death_sig, timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 713, in _close
+    handler.proc.wait(time_to_wait)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1209, in wait
+    return self._wait(timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1953, in _wait
+    time.sleep(delay)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 29558 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 878, in _invoke_run
+    run_result = self._monitor_workers(self._worker_group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/local_elastic_agent.py", line 307, in _monitor_workers
+    result = self._pcontext.wait(0)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 288, in wait
+    return self._poll()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 668, in _poll
+    self.close()  # terminate all running procs
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 331, in close
+    self._close(death_sig=death_sig, timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 713, in _close
+    handler.proc.wait(time_to_wait)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1209, in wait
+    return self._wait(timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1953, in _wait
+    time.sleep(delay)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 290824 got signal: 15
+srun: error: ip-26-0-163-147: task 10: Exited with exit code 1
+srun: error: ip-26-0-167-175: task 18: Exited with exit code 1
+srun: error: ip-26-0-161-103: task 1: Exited with exit code 1
+srun: Force Terminated StepId=13502056.0
diff --git a/logs/13502078-bench_3.57G_dp32_tp8_pp1_acc8_mbs1_seq4096_zero1_tpmodeRED_vocab131k.out b/logs/13502078-bench_3.57G_dp32_tp8_pp1_acc8_mbs1_seq4096_zero1_tpmodeRED_vocab131k.out
new file mode 100644
index 0000000000000000000000000000000000000000..5eb4f10b55b0ffd143b8fcd4f37491401bdfe67e
--- /dev/null
+++ b/logs/13502078-bench_3.57G_dp32_tp8_pp1_acc8_mbs1_seq4096_zero1_tpmodeRED_vocab131k.out
@@ -0,0 +1,2929 @@
++ '[' -z 13502078 ']'
++ source /etc/profile.d/modules.sh
+++ . /usr/share/modules/init/bash
++++ unset _mlshdbg
++++ '[' 0 = 1 ']'
++++ unset _mlre _mlIFS
++++ '[' -n x ']'
++++ _mlIFS=' 	
+'
++++ IFS=' '
++++ '[' -n '' ']'
+++++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash autoinit
++++ _mlcode='module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+MODULEPATH_modshare=/etc/environment-modules/modules:1:/usr/share/modules/\$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1; export MODULEPATH_modshare;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+LOADEDMODULES=; export LOADEDMODULES;
+MODULEPATH=/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/\$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles; export MODULEPATH;
+test 0;'
++++ _mlret=0
++++ '[' -n x ']'
++++ IFS=' 	
+'
++++ unset _mlIFS
++++ unset _mlre _mlv _mlrv
++++ '[' 0 -eq 0 ']'
++++ eval 'module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+MODULEPATH_modshare=/etc/environment-modules/modules:1:/usr/share/modules/\$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1; export MODULEPATH_modshare;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+LOADEDMODULES=; export LOADEDMODULES;
+MODULEPATH=/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/\$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles; export MODULEPATH;
+test 0;'
+++++ MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl
+++++ export MODULES_CMD
+++++ ENV=/usr/share/modules/init/profile.sh
+++++ export ENV
+++++ MODULEPATH_modshare='/etc/environment-modules/modules:1:/usr/share/modules/$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1'
+++++ export MODULEPATH_modshare
+++++ BASH_ENV=/usr/share/modules/init/bash
+++++ export BASH_ENV
+++++ MODULESHOME=/usr/share/modules
+++++ export MODULESHOME
+++++ LOADEDMODULES=
+++++ export LOADEDMODULES
+++++ MODULEPATH='/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles'
+++++ export MODULEPATH
+++++ test 0
++++ '[' 0 = 1 ']'
++++ '[' -t 2 ']'
++++ export -f module
++++ export -f switchml
++++ '[' 5 -ge 3 ']'
++++ [[ ehxB =~ i ]]
++++ [[ ! :/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin: =~ :/usr/bin: ]]
++++ '[' '!' -n '' ']'
++++ MANPATH=:
++++ export MANPATH
+++++ manpath
++++ [[ ! :/admin/home/nouamane/.local/share/man:/fsx/nouamane/miniconda/envs/2-1-cu121/man:/fsx/nouamane/miniconda/envs/2-1-cu121/share/man:/fsx/nouamane/miniconda/man:/fsx/nouamane/miniconda/share/man:/opt/slurm/share/man:/opt/amazon/openmpi/share/man:/opt/amazon/efa/share/man:/usr/local/man:/usr/local/share/man:/usr/share/man:/admin/home/nouamane/.fzf/man:: =~ :/usr/share/man: ]]
++++ unset _mlcode _mlret
++++ '[' -n '' ']'
++ module load cuda/12.1
++ unset _mlshdbg
++ '[' 0 = 1 ']'
++ unset _mlre _mlIFS
++ '[' -n x ']'
++ _mlIFS=' 	
+'
++ IFS=' '
++ '[' -n '' ']'
+++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash load cuda/12.1
++ eval 'CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include;' export 'CPATH;
+LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:;' export 'LD_LIBRARY_PATH;
+FI_EFA_FORK_SAFE=1;' export 'FI_EFA_FORK_SAFE;
+MANPATH=/usr/local/cuda-12.1/share/man::;' export 'MANPATH;
+LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64;' export 'LIBRARY_PATH;
+_LMFILES_=/usr/share/modules/modulefiles/cuda/12.1;' export '_LMFILES_;
+LOADEDMODULES=cuda/12.1;' export 'LOADEDMODULES;
+MPI_PATH=/opt/amazon/openmpi;' export 'MPI_PATH;
+NCCL_HOME_modshare=/opt/nccl/build:1;' export 'NCCL_HOME_modshare;
+NCCL_PROTO=simple;' export 'NCCL_PROTO;
+MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1;' export 'MANPATH_modshare;
+LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1;' export 'LIBRARY_PATH_modshare;
+NCCL_SOCKET_IFNAME=enp;' export 'NCCL_SOCKET_IFNAME;
+AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl;' export 'AWS_OFI_NCCL_HOME;
+NCCL_HOME=/opt/nccl/build;' export 'NCCL_HOME;
+FI_PROVIDER=efa;' export 'FI_PROVIDER;
+AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1;' export 'AWS_OFI_NCCL_HOME_modshare;
+CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1;' export 'CPATH_modshare;
+LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2::1:/opt/nccl/build/lib:1:/opt/aws-ofi-nccl/lib:2:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/efa/lib:1;' export 'LD_LIBRARY_PATH_modshare;
+FI_EFA_ENABLE_SHM_TRANSFER=1;' export 'FI_EFA_ENABLE_SHM_TRANSFER;
+_LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1;' export '_LMFILES__modshare;
+LOADEDMODULES_modshare=cuda/12.1:1;' export 'LOADEDMODULES_modshare;
+MPI_PATH_modshare=/opt/amazon/openmpi:1;' export 'MPI_PATH_modshare;
+PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin;' export 'PATH;
+CUDA_HOME=/usr/local/cuda-12.1;' export 'CUDA_HOME;
+PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/usr/local/cuda-12.1/include:1:/opt/amazon/efa/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/usr/sbin:1:/opt/slurm/bin:1:/fsx/nouamane/miniconda/bin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:2:/opt/amazon/openmpi/bin:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1;' export 'PATH_modshare;
+test' '0;'
+++ CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include
+++ export CPATH
+++ LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:
+++ export LD_LIBRARY_PATH
+++ FI_EFA_FORK_SAFE=1
+++ export FI_EFA_FORK_SAFE
+++ MANPATH=/usr/local/cuda-12.1/share/man::
+++ export MANPATH
+++ LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64
+++ export LIBRARY_PATH
+++ _LMFILES_=/usr/share/modules/modulefiles/cuda/12.1
+++ export _LMFILES_
+++ LOADEDMODULES=cuda/12.1
+++ export LOADEDMODULES
+++ MPI_PATH=/opt/amazon/openmpi
+++ export MPI_PATH
+++ NCCL_HOME_modshare=/opt/nccl/build:1
+++ export NCCL_HOME_modshare
+++ NCCL_PROTO=simple
+++ export NCCL_PROTO
+++ MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1
+++ export MANPATH_modshare
+++ LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1
+++ export LIBRARY_PATH_modshare
+++ NCCL_SOCKET_IFNAME=enp
+++ export NCCL_SOCKET_IFNAME
+++ AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl
+++ export AWS_OFI_NCCL_HOME
+++ NCCL_HOME=/opt/nccl/build
+++ export NCCL_HOME
+++ FI_PROVIDER=efa
+++ export FI_PROVIDER
+++ AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1
+++ export AWS_OFI_NCCL_HOME_modshare
+++ CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1
+++ export CPATH_modshare
+++ LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2::1:/opt/nccl/build/lib:1:/opt/aws-ofi-nccl/lib:2:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/efa/lib:1
+++ export LD_LIBRARY_PATH_modshare
+++ FI_EFA_ENABLE_SHM_TRANSFER=1
+++ export FI_EFA_ENABLE_SHM_TRANSFER
+++ _LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1
+++ export _LMFILES__modshare
+++ LOADEDMODULES_modshare=cuda/12.1:1
+++ export LOADEDMODULES_modshare
+++ MPI_PATH_modshare=/opt/amazon/openmpi:1
+++ export MPI_PATH_modshare
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ export PATH
+++ CUDA_HOME=/usr/local/cuda-12.1
+++ export CUDA_HOME
+++ PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/usr/local/cuda-12.1/include:1:/opt/amazon/efa/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/usr/sbin:1:/opt/slurm/bin:1:/fsx/nouamane/miniconda/bin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:2:/opt/amazon/openmpi/bin:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1
+++ export PATH_modshare
+++ test 0
++ _mlstatus=0
++ '[' -n x ']'
++ IFS=' 	
+'
++ unset _mlre _mlv _mlrv _mlIFS
++ '[' -n '' ']'
++ unset _mlshdbg
++ return 0
++ source /fsx/nouamane/miniconda/bin/activate
+++ _CONDA_ROOT=/fsx/nouamane/miniconda
+++ . /fsx/nouamane/miniconda/etc/profile.d/conda.sh
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ '[' -z x ']'
+++ conda activate
+++ local cmd=activate
+++ case "$cmd" in
+++ __conda_activate activate
+++ '[' -n '' ']'
+++ local ask_conda
++++ PS1=
++++ __conda_exe shell.posix activate
++++ /fsx/nouamane/miniconda/bin/conda shell.posix activate
+Error while loading conda entry point: conda-libmamba-solver (module 'libmambapy' has no attribute 'QueryFormat')
+++ ask_conda='. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_2='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/etc/conda/activate.d/libxml2_activate.sh"'
+++ eval '. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_2='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/etc/conda/activate.d/libxml2_activate.sh"'
++++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh
+++++ test -n ''
+++++ unset XML_CATALOG_FILES
+++++ unset xml_catalog_files_libxml2
++++ PS1='(base) '
++++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++++ export CONDA_PREFIX=/fsx/nouamane/miniconda
++++ CONDA_PREFIX=/fsx/nouamane/miniconda
++++ export CONDA_SHLVL=3
++++ CONDA_SHLVL=3
++++ export CONDA_DEFAULT_ENV=base
++++ CONDA_DEFAULT_ENV=base
++++ export 'CONDA_PROMPT_MODIFIER=(base) '
++++ CONDA_PROMPT_MODIFIER='(base) '
++++ export CONDA_PREFIX_2=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ CONDA_PREFIX_2=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ . /fsx/nouamane/miniconda/etc/conda/activate.d/libxml2_activate.sh
+++++ test -n ''
+++++ xml_catalog_files_libxml2=
+++++ XML_CATALOG_FILES=
+++++ conda_catalog_files=
+++++ ifs_libxml2=' 	
+'
+++++ IFS=' '
+++++ rem=/fsx/nouamane/miniconda
+++++ for pre in ${rem}
+++++ test '' = /fsx/nouamane/miniconda
+++++ conda_catalog_files=/fsx/nouamane/miniconda
+++++ rem=
+++++ IFS=' 	
+'
+++++ conda_catalog_files='file:///fsx/nouamane/miniconda/etc/xml/catalog file:///etc/xml/catalog'
+++++ export 'XML_CATALOG_FILES=file:///fsx/nouamane/miniconda/etc/xml/catalog file:///etc/xml/catalog'
+++++ XML_CATALOG_FILES='file:///fsx/nouamane/miniconda/etc/xml/catalog file:///etc/xml/catalog'
+++++ unset conda_catalog_files ifs_libxml2 rem
+++ __conda_hashr
+++ '[' -n '' ']'
+++ '[' -n '' ']'
+++ hash -r
++ conda activate 2-1-cu121
++ local cmd=activate
++ case "$cmd" in
++ __conda_activate activate 2-1-cu121
++ '[' -n '' ']'
++ local ask_conda
+++ PS1='(base) '
+++ __conda_exe shell.posix activate 2-1-cu121
+++ /fsx/nouamane/miniconda/bin/conda shell.posix activate 2-1-cu121
+Error while loading conda entry point: conda-libmamba-solver (module 'libmambapy' has no attribute 'QueryFormat')
++ ask_conda='. "/fsx/nouamane/miniconda/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''4'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_3='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
++ eval '. "/fsx/nouamane/miniconda/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''4'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_3='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
+++ . /fsx/nouamane/miniconda/etc/conda/deactivate.d/libxml2_deactivate.sh
++++ test -n ''
++++ unset XML_CATALOG_FILES
++++ unset xml_catalog_files_libxml2
+++ PS1='(2-1-cu121) '
+++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ export CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ export CONDA_SHLVL=4
+++ CONDA_SHLVL=4
+++ export CONDA_DEFAULT_ENV=2-1-cu121
+++ CONDA_DEFAULT_ENV=2-1-cu121
+++ export 'CONDA_PROMPT_MODIFIER=(2-1-cu121) '
+++ CONDA_PROMPT_MODIFIER='(2-1-cu121) '
+++ export CONDA_PREFIX_3=/fsx/nouamane/miniconda
+++ CONDA_PREFIX_3=/fsx/nouamane/miniconda
+++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ export _CE_M=
+++ _CE_M=
+++ export _CE_CONDA=
+++ _CE_CONDA=
+++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh
++++ test -n ''
++++ xml_catalog_files_libxml2=
++++ XML_CATALOG_FILES=
++++ conda_catalog_files=
++++ ifs_libxml2=' 	
+'
++++ IFS=' '
++++ rem=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ for pre in ${rem}
++++ test '' = /fsx/nouamane/miniconda/envs/2-1-cu121
++++ conda_catalog_files=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ rem=
++++ IFS=' 	
+'
++++ conda_catalog_files='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ export 'XML_CATALOG_FILES=file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ XML_CATALOG_FILES='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ unset conda_catalog_files ifs_libxml2 rem
++ __conda_hashr
++ '[' -n '' ']'
++ '[' -n '' ']'
++ hash -r
++ export PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++ PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++ '[' -z '' ']'
+++ scontrol show hostnames 'ip-26-0-162-[79,180,233],ip-26-0-163-43,ip-26-0-164-236,ip-26-0-165-[24,38,59,131,164,202,213],ip-26-0-166-[15,36,125,214,244],ip-26-0-168-[120,238],ip-26-0-169-[86,132,139,207,239,247],ip-26-0-170-[31,132,143,160],ip-26-0-171-[21,56,62]'
++ export 'NODELIST=ip-26-0-162-79
+ip-26-0-162-180
+ip-26-0-162-233
+ip-26-0-163-43
+ip-26-0-164-236
+ip-26-0-165-24
+ip-26-0-165-38
+ip-26-0-165-59
+ip-26-0-165-131
+ip-26-0-165-164
+ip-26-0-165-202
+ip-26-0-165-213
+ip-26-0-166-15
+ip-26-0-166-36
+ip-26-0-166-125
+ip-26-0-166-214
+ip-26-0-166-244
+ip-26-0-168-120
+ip-26-0-168-238
+ip-26-0-169-86
+ip-26-0-169-132
+ip-26-0-169-139
+ip-26-0-169-207
+ip-26-0-169-239
+ip-26-0-169-247
+ip-26-0-170-31
+ip-26-0-170-132
+ip-26-0-170-143
+ip-26-0-170-160
+ip-26-0-171-21
+ip-26-0-171-56
+ip-26-0-171-62'
++ NODELIST='ip-26-0-162-79
+ip-26-0-162-180
+ip-26-0-162-233
+ip-26-0-163-43
+ip-26-0-164-236
+ip-26-0-165-24
+ip-26-0-165-38
+ip-26-0-165-59
+ip-26-0-165-131
+ip-26-0-165-164
+ip-26-0-165-202
+ip-26-0-165-213
+ip-26-0-166-15
+ip-26-0-166-36
+ip-26-0-166-125
+ip-26-0-166-214
+ip-26-0-166-244
+ip-26-0-168-120
+ip-26-0-168-238
+ip-26-0-169-86
+ip-26-0-169-132
+ip-26-0-169-139
+ip-26-0-169-207
+ip-26-0-169-239
+ip-26-0-169-247
+ip-26-0-170-31
+ip-26-0-170-132
+ip-26-0-170-143
+ip-26-0-170-160
+ip-26-0-171-21
+ip-26-0-171-56
+ip-26-0-171-62'
+++ head -n1
+++ scontrol show hostnames 'ip-26-0-162-[79,180,233],ip-26-0-163-43,ip-26-0-164-236,ip-26-0-165-[24,38,59,131,164,202,213],ip-26-0-166-[15,36,125,214,244],ip-26-0-168-[120,238],ip-26-0-169-[86,132,139,207,239,247],ip-26-0-170-[31,132,143,160],ip-26-0-171-[21,56,62]'
++ export MASTER_NODE=ip-26-0-162-79
++ MASTER_NODE=ip-26-0-162-79
++ export MASTER_PORT=12356
++ MASTER_PORT=12356
++ export NNODES=32
++ NNODES=32
++ export GPUS_PER_NODE=8
++ GPUS_PER_NODE=8
++ export WORLD_SIZE=256
++ WORLD_SIZE=256
++ export CUDA_DEVICE_MAX_CONNECTIONS=1
++ CUDA_DEVICE_MAX_CONNECTIONS=1
++ export NCCL_DEBUG=WARN
++ NCCL_DEBUG=WARN
++ export NANOTRON_BENCHMARK=1
++ NANOTRON_BENCHMARK=1
++ export WANDB_MODE=disabled
++ WANDB_MODE=disabled
++ export TORCH_NCCL_ASYNC_ERROR_HANDLING=1
++ TORCH_NCCL_ASYNC_ERROR_HANDLING=1
++ '[' -z '' ']'
++ echo '=== GPU Topology ==='
+=== GPU Topology ===
++ nvidia-smi topo -m
+	GPU0	GPU1	GPU2	GPU3	GPU4	GPU5	GPU6	GPU7	CPU Affinity	NUMA Affinity	GPU NUMA ID
+GPU0	 X 	NV18	NV18	NV18	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU1	NV18	 X 	NV18	NV18	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU2	NV18	NV18	 X 	NV18	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU3	NV18	NV18	NV18	 X 	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU4	NV18	NV18	NV18	NV18	 X 	NV18	NV18	NV18	48-95	1		N/A
+GPU5	NV18	NV18	NV18	NV18	NV18	 X 	NV18	NV18	48-95	1		N/A
+GPU6	NV18	NV18	NV18	NV18	NV18	NV18	 X 	NV18	48-95	1		N/A
+GPU7	NV18	NV18	NV18	NV18	NV18	NV18	NV18	 X 	48-95	1		N/A
+
+Legend:
+
+  X    = Self
+  SYS  = Connection traversing PCIe as well as the SMP interconnect between NUMA nodes (e.g., QPI/UPI)
+  NODE = Connection traversing PCIe as well as the interconnect between PCIe Host Bridges within a NUMA node
+  PHB  = Connection traversing PCIe as well as a PCIe Host Bridge (typically the CPU)
+  PXB  = Connection traversing multiple PCIe bridges (without traversing the PCIe Host Bridge)
+  PIX  = Connection traversing at most a single PCIe bridge
+  NV#  = Connection traversing a bonded set of # NVLinks
++ echo ==================
+==================
++ export SRUN_ALLOC_ARGS=
++ SRUN_ALLOC_ARGS=
++ echo 'Master node: ip-26-0-162-79'
+Master node: ip-26-0-162-79
++ echo 'All nodes: ip-26-0-162-79
+ip-26-0-162-180
+ip-26-0-162-233
+ip-26-0-163-43
+ip-26-0-164-236
+ip-26-0-165-24
+ip-26-0-165-38
+ip-26-0-165-59
+ip-26-0-165-131
+ip-26-0-165-164
+ip-26-0-165-202
+ip-26-0-165-213
+ip-26-0-166-15
+ip-26-0-166-36
+ip-26-0-166-125
+ip-26-0-166-214
+ip-26-0-166-244
+ip-26-0-168-120
+ip-26-0-168-238
+ip-26-0-169-86
+ip-26-0-169-132
+ip-26-0-169-139
+ip-26-0-169-207
+ip-26-0-169-239
+ip-26-0-169-247
+ip-26-0-170-31
+ip-26-0-170-132
+ip-26-0-170-143
+ip-26-0-170-160
+ip-26-0-171-21
+ip-26-0-171-56
+ip-26-0-171-62'
+All nodes: ip-26-0-162-79
+ip-26-0-162-180
+ip-26-0-162-233
+ip-26-0-163-43
+ip-26-0-164-236
+ip-26-0-165-24
+ip-26-0-165-38
+ip-26-0-165-59
+ip-26-0-165-131
+ip-26-0-165-164
+ip-26-0-165-202
+ip-26-0-165-213
+ip-26-0-166-15
+ip-26-0-166-36
+ip-26-0-166-125
+ip-26-0-166-214
+ip-26-0-166-244
+ip-26-0-168-120
+ip-26-0-168-238
+ip-26-0-169-86
+ip-26-0-169-132
+ip-26-0-169-139
+ip-26-0-169-207
+ip-26-0-169-239
+ip-26-0-169-247
+ip-26-0-170-31
+ip-26-0-170-132
+ip-26-0-170-143
+ip-26-0-170-160
+ip-26-0-171-21
+ip-26-0-171-56
+ip-26-0-171-62
++ echo 'World size: 256'
+World size: 256
++ '[' -n '' ']'
++ srun --wait=0 --kill-on-bad-exit=1 torchrun --nnodes=32 --nproc_per_node=8 --rdzv_id=13502078 --rdzv_backend=c10d --rdzv_endpoint=ip-26-0-162-79:12356 --max_restarts 0 --rdzv_conf timeout=60 /fsx/nouamane/projects/nanotron/run_train.py --config-file benchmark/configs/config_3.57G_dp32_tp8_pp1_acc8_mbs1_seq4096_zero1_tpmodeRED_vocab131k.yaml
+[2024-12-19 06:15:56,877] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-19 06:15:56,877] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-19 06:15:56,876] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-19 06:15:56,877] torch.distributed.run: [WARNING] 
+[2024-12-19 06:15:56,877] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 06:15:56,877] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-19 06:15:56,877] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 06:15:56,876] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-19 06:15:56,877] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-19 06:15:56,879] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-19 06:15:56,879] torch.distributed.run: [WARNING] 
+[2024-12-19 06:15:56,879] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 06:15:56,879] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-19 06:15:56,879] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 06:15:56,879] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-19 06:15:56,881] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-19 06:15:56,882] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-19 06:15:56,883] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-19 06:15:56,881] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-19 06:15:56,883] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-19 06:15:56,881] torch.distributed.run: [WARNING] 
+[2024-12-19 06:15:56,881] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 06:15:56,881] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-19 06:15:56,881] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 06:15:56,885] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-19 06:15:56,886] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-19 06:15:56,885] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-19 06:15:56,887] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-19 06:15:56,885] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-19 06:15:56,885] torch.distributed.run: [WARNING] 
+[2024-12-19 06:15:56,885] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 06:15:56,885] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-19 06:15:56,885] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 06:15:56,885] torch.distributed.run: [WARNING] 
+[2024-12-19 06:15:56,885] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 06:15:56,885] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-19 06:15:56,885] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 06:15:56,885] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-19 06:15:56,887] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-19 06:15:56,888] torch.distributed.run: [WARNING] 
+[2024-12-19 06:15:56,888] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 06:15:56,888] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-19 06:15:56,888] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 06:15:56,884] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-19 06:15:56,887] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-19 06:15:56,892] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-19 06:15:56,895] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-19 06:15:56,895] torch.distributed.run: [WARNING] 
+[2024-12-19 06:15:56,895] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 06:15:56,895] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-19 06:15:56,895] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 06:15:56,898] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-19 06:15:56,898] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-19 06:15:56,903] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-19 06:15:56,913] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-19 06:15:56,877] torch.distributed.run: [WARNING] 
+[2024-12-19 06:15:56,877] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 06:15:56,877] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-19 06:15:56,877] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 06:15:56,876] torch.distributed.run: [WARNING] 
+[2024-12-19 06:15:56,876] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 06:15:56,876] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-19 06:15:56,876] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 06:15:56,876] torch.distributed.run: [WARNING] 
+[2024-12-19 06:15:56,876] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 06:15:56,876] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-19 06:15:56,876] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 06:15:56,878] torch.distributed.run: [WARNING] 
+[2024-12-19 06:15:56,878] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 06:15:56,878] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-19 06:15:56,878] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 06:15:56,881] torch.distributed.run: [WARNING] 
+[2024-12-19 06:15:56,881] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 06:15:56,881] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-19 06:15:56,881] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 06:15:56,879] torch.distributed.run: [WARNING] 
+[2024-12-19 06:15:56,879] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 06:15:56,879] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-19 06:15:56,879] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 06:15:56,882] torch.distributed.run: [WARNING] 
+[2024-12-19 06:15:56,882] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 06:15:56,882] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-19 06:15:56,882] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 06:15:56,924] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-19 06:15:56,923] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-19 06:15:56,885] torch.distributed.run: [WARNING] 
+[2024-12-19 06:15:56,885] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 06:15:56,885] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-19 06:15:56,885] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 06:15:56,883] torch.distributed.run: [WARNING] 
+[2024-12-19 06:15:56,883] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 06:15:56,883] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-19 06:15:56,883] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 06:15:56,883] torch.distributed.run: [WARNING] 
+[2024-12-19 06:15:56,883] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 06:15:56,883] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-19 06:15:56,883] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 06:15:56,886] torch.distributed.run: [WARNING] 
+[2024-12-19 06:15:56,886] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 06:15:56,886] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-19 06:15:56,886] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 06:15:56,885] torch.distributed.run: [WARNING] 
+[2024-12-19 06:15:56,885] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 06:15:56,885] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-19 06:15:56,885] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 06:15:56,887] torch.distributed.run: [WARNING] 
+[2024-12-19 06:15:56,887] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 06:15:56,887] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-19 06:15:56,887] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 06:15:56,884] torch.distributed.run: [WARNING] 
+[2024-12-19 06:15:56,884] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 06:15:56,884] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-19 06:15:56,884] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 06:15:56,932] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-19 06:15:56,888] torch.distributed.run: [WARNING] 
+[2024-12-19 06:15:56,888] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 06:15:56,888] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-19 06:15:56,888] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 06:15:56,898] torch.distributed.run: [WARNING] 
+[2024-12-19 06:15:56,898] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 06:15:56,898] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-19 06:15:56,898] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 06:15:56,892] torch.distributed.run: [WARNING] 
+[2024-12-19 06:15:56,892] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 06:15:56,892] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-19 06:15:56,892] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 06:15:56,898] torch.distributed.run: [WARNING] 
+[2024-12-19 06:15:56,898] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 06:15:56,898] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-19 06:15:56,898] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 06:15:56,903] torch.distributed.run: [WARNING] 
+[2024-12-19 06:15:56,903] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 06:15:56,903] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-19 06:15:56,903] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 06:15:56,948] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-19 06:15:56,913] torch.distributed.run: [WARNING] 
+[2024-12-19 06:15:56,913] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 06:15:56,913] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-19 06:15:56,913] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 06:15:56,923] torch.distributed.run: [WARNING] 
+[2024-12-19 06:15:56,923] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 06:15:56,923] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-19 06:15:56,923] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 06:15:56,924] torch.distributed.run: [WARNING] 
+[2024-12-19 06:15:56,924] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 06:15:56,924] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-19 06:15:56,924] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 06:15:56,932] torch.distributed.run: [WARNING] 
+[2024-12-19 06:15:56,932] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 06:15:56,932] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-19 06:15:56,932] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 06:15:56,948] torch.distributed.run: [WARNING] 
+[2024-12-19 06:15:56,948] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 06:15:56,948] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-19 06:15:56,948] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 06:15:57,033] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-19 06:15:57,033] torch.distributed.run: [WARNING] 
+[2024-12-19 06:15:57,033] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 06:15:57,033] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-19 06:15:57,033] torch.distributed.run: [WARNING] *****************************************
+NCCL version 2.18.5+cuda12.2
+12/19/2024 06:16:28 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: Measuring inter-GPU and intra-node bandwidth...
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+12/19/2024 06:16:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: Bandwidth measurement complete. Time taken: 18.86 seconds
+12/19/2024 06:16:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: Config:
+12/19/2024 06:16:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: Config(general=GeneralArgs(project='debug',
+12/19/2024 06:16:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                            run='3.57G_dp32_tp8_pp1_acc8_mbs1_seq4096_zero1_tpmodeRED_vocab131k',
+12/19/2024 06:16:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                            seed=42,
+12/19/2024 06:16:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                            step=None,
+12/19/2024 06:16:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                            consumed_train_samples=None,
+12/19/2024 06:16:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                            benchmark_csv_path=PosixPath('benchmark/results/bench_final2.csv'),
+12/19/2024 06:16:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                            ignore_sanity_checks=True),
+12/19/2024 06:16:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:        parallelism=ParallelismArgs(dp=32,
+12/19/2024 06:16:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                                    pp=1,
+12/19/2024 06:16:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                                    tp=8,
+12/19/2024 06:16:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                                    pp_engine=<nanotron.parallel.pipeline_parallel.engine.OneForwardOneBackwardPipelineEngine object at 0x7f000a474dc0>,
+12/19/2024 06:16:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                                    tp_mode=<TensorParallelLinearMode.REDUCE_SCATTER: 2>,
+12/19/2024 06:16:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                                    tp_linear_async_communication=True,
+12/19/2024 06:16:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                                    recompute_layer=False,
+12/19/2024 06:16:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                                    tp_recompute_allgather=True,
+12/19/2024 06:16:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                                    expert_parallel_size=1),
+12/19/2024 06:16:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:        model=ModelArgs(model_config=LlamaConfig(bos_token_id=0,
+12/19/2024 06:16:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                                                 eos_token_id=0,
+12/19/2024 06:16:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                                                 hidden_act='silu',
+12/19/2024 06:16:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                                                 hidden_size=3072,
+12/19/2024 06:16:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                                                 initializer_range=0.02,
+12/19/2024 06:16:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                                                 intermediate_size=8192,
+12/19/2024 06:16:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                                                 is_llama_config=True,
+12/19/2024 06:16:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                                                 max_position_embeddings=4096,
+12/19/2024 06:16:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                                                 num_attention_heads=32,
+12/19/2024 06:16:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                                                 num_hidden_layers=28,
+12/19/2024 06:16:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                                                 num_key_value_heads=32,
+12/19/2024 06:16:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                                                 pad_token_id=None,
+12/19/2024 06:16:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                                                 pretraining_tp=1,
+12/19/2024 06:16:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                                                 rms_norm_eps=1e-05,
+12/19/2024 06:16:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                                                 rope_scaling=None,
+12/19/2024 06:16:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                                                 rope_theta=10000.0,
+12/19/2024 06:16:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                                                 rope_interleaved=False,
+12/19/2024 06:16:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                                                 tie_word_embeddings=True,
+12/19/2024 06:16:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                                                 use_cache=True,
+12/19/2024 06:16:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                                                 vocab_size=131072),
+12/19/2024 06:16:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                        init_method=RandomInit(std=0.02),
+12/19/2024 06:16:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                        dtype=torch.bfloat16,
+12/19/2024 06:16:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                        make_vocab_size_divisible_by=1,
+12/19/2024 06:16:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                        ddp_bucket_cap_mb=25),
+12/19/2024 06:16:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:        tokenizer=TokenizerArgs(tokenizer_name_or_path='robot-test/dummy-tokenizer-wordlevel',
+12/19/2024 06:16:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                                tokenizer_revision=None,
+12/19/2024 06:16:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                                tokenizer_max_length=None),
+12/19/2024 06:16:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:        checkpoints=CheckpointsArgs(checkpoints_path=PosixPath('checkpoints'),
+12/19/2024 06:16:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                                    checkpoint_interval=10000,
+12/19/2024 06:16:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                                    save_initial_state=False,
+12/19/2024 06:16:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                                    save_final_state=False,
+12/19/2024 06:16:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                                    resume_checkpoint_path=None,
+12/19/2024 06:16:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                                    checkpoints_path_is_shared_file_system=False),
+12/19/2024 06:16:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:        logging=LoggingArgs(log_level='info',
+12/19/2024 06:16:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                            log_level_replica='info',
+12/19/2024 06:16:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                            iteration_step_info_interval=1),
+12/19/2024 06:16:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:        tokens=TokensArgs(sequence_length=4096,
+12/19/2024 06:16:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                          train_steps=100,
+12/19/2024 06:16:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                          micro_batch_size=1,
+12/19/2024 06:16:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                          batch_accumulation_per_replica=8,
+12/19/2024 06:16:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                          val_check_interval=100,
+12/19/2024 06:16:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                          limit_val_batches=0,
+12/19/2024 06:16:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                          limit_test_batches=0),
+12/19/2024 06:16:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:        optimizer=OptimizerArgs(optimizer_factory=AdamWOptimizerArgs(adam_eps=1e-08,
+12/19/2024 06:16:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                                                                     adam_beta1=0.9,
+12/19/2024 06:16:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                                                                     adam_beta2=0.95,
+12/19/2024 06:16:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                                                                     torch_adam_is_fused=True,
+12/19/2024 06:16:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                                                                     name='adamW'),
+12/19/2024 06:16:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                                zero_stage=1,
+12/19/2024 06:16:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                                weight_decay=0.01,
+12/19/2024 06:16:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                                clip_grad=1.0,
+12/19/2024 06:16:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                                accumulate_grad_in_fp32=True,
+12/19/2024 06:16:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                                learning_rate_scheduler=LRSchedulerArgs(learning_rate=0.0003,
+12/19/2024 06:16:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                                                                        lr_warmup_steps=2,
+12/19/2024 06:16:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                                                                        lr_warmup_style='linear',
+12/19/2024 06:16:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                                                                        lr_decay_style='cosine',
+12/19/2024 06:16:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                                                                        lr_decay_steps=13,
+12/19/2024 06:16:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                                                                        lr_decay_starting_step=None,
+12/19/2024 06:16:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                                                                        min_decay_lr=1e-05)),
+12/19/2024 06:16:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:        data_stages=[DatasetStageArgs(name='Stable Training Stage',
+12/19/2024 06:16:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                                      start_training_step=1,
+12/19/2024 06:16:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                                      data=DataArgs(dataset=None,
+12/19/2024 06:16:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                                                    seed=42,
+12/19/2024 06:16:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                                                    num_loading_workers=1))],
+12/19/2024 06:16:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:        profiler=None,
+12/19/2024 06:16:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:        lighteval=None,
+12/19/2024 06:16:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:        s3_upload=None)
+12/19/2024 06:16:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: Model Config:
+12/19/2024 06:16:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: LlamaConfig(bos_token_id=0,
+12/19/2024 06:16:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:             eos_token_id=0,
+12/19/2024 06:16:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:             hidden_act='silu',
+12/19/2024 06:16:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:             hidden_size=3072,
+12/19/2024 06:16:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:             initializer_range=0.02,
+12/19/2024 06:16:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:             intermediate_size=8192,
+12/19/2024 06:16:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:             is_llama_config=True,
+12/19/2024 06:16:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:             max_position_embeddings=4096,
+12/19/2024 06:16:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:             num_attention_heads=32,
+12/19/2024 06:16:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:             num_hidden_layers=28,
+12/19/2024 06:16:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:             num_key_value_heads=32,
+12/19/2024 06:16:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:             pad_token_id=None,
+12/19/2024 06:16:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:             pretraining_tp=1,
+12/19/2024 06:16:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:             rms_norm_eps=1e-05,
+12/19/2024 06:16:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:             rope_scaling=None,
+12/19/2024 06:16:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:             rope_theta=10000.0,
+12/19/2024 06:16:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:             rope_interleaved=False,
+12/19/2024 06:16:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:             tie_word_embeddings=True,
+12/19/2024 06:16:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:             use_cache=True,
+12/19/2024 06:16:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:             vocab_size=131072)
+12/19/2024 06:16:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: Building model..
+12/19/2024 06:16:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: Initialize RoPE Theta = 10000.0
+12/19/2024 06:16:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: Setting PP block ranks...
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+12/19/2024 06:16:48 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: Total number of parameters: 3.57G (6818.67MiB)
+12/19/2024 06:16:48 [INFO|DP=0|PP=0|TP=1|ip-26-0-162-180]: Local number of parameters: 447M (852.33MiB)
+12/19/2024 06:16:48 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: Local number of parameters: 447M (852.33MiB)
+12/19/2024 06:16:48 [INFO|DP=0|PP=0|TP=7|ip-26-0-162-180]: Local number of parameters: 447M (852.33MiB)
+12/19/2024 06:16:48 [INFO|DP=0|PP=0|TP=2|ip-26-0-162-180]: Local number of parameters: 447M (852.33MiB)
+12/19/2024 06:16:48 [INFO|DP=0|PP=0|TP=6|ip-26-0-162-180]: Local number of parameters: 447M (852.33MiB)
+12/19/2024 06:16:48 [INFO|DP=0|PP=0|TP=3|ip-26-0-162-180]: Local number of parameters: 447M (852.33MiB)
+12/19/2024 06:16:48 [INFO|DP=0|PP=0|TP=5|ip-26-0-162-180]: Local number of parameters: 447M (852.33MiB)
+12/19/2024 06:16:48 [INFO|DP=0|PP=0|TP=4|ip-26-0-162-180]: Local number of parameters: 447M (852.33MiB)
+12/19/2024 06:16:48 [INFO|DP=0|PP=0|TP=7|ip-26-0-162-180]: [After model building] Memory usage: 852.87MiB. Peak allocated: 5384.00MiB Peak reserved: 17686.00MiB
+12/19/2024 06:16:48 [INFO|DP=0|PP=0|TP=1|ip-26-0-162-180]: [After model building] Memory usage: 852.87MiB. Peak allocated: 5384.00MiB Peak reserved: 17686.00MiB
+12/19/2024 06:16:48 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: [After model building] Memory usage: 852.87MiB. Peak allocated: 5384.00MiB Peak reserved: 17686.00MiB
+12/19/2024 06:16:48 [INFO|DP=0|PP=0|TP=2|ip-26-0-162-180]: [After model building] Memory usage: 852.87MiB. Peak allocated: 5384.00MiB Peak reserved: 17686.00MiB
+12/19/2024 06:16:48 [INFO|DP=0|PP=0|TP=6|ip-26-0-162-180]: [After model building] Memory usage: 852.87MiB. Peak allocated: 5384.00MiB Peak reserved: 17686.00MiB
+12/19/2024 06:16:48 [INFO|DP=0|PP=0|TP=3|ip-26-0-162-180]: [After model building] Memory usage: 852.87MiB. Peak allocated: 5384.00MiB Peak reserved: 17686.00MiB
+12/19/2024 06:16:48 [INFO|DP=0|PP=0|TP=5|ip-26-0-162-180]: [After model building] Memory usage: 852.87MiB. Peak allocated: 5384.00MiB Peak reserved: 17686.00MiB
+12/19/2024 06:16:48 [INFO|DP=0|PP=0|TP=4|ip-26-0-162-180]: [After model building] Memory usage: 852.87MiB. Peak allocated: 5384.00MiB Peak reserved: 17686.00MiB
+12/19/2024 06:16:48 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: No checkpoint path provided.
+12/19/2024 06:16:48 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: Parametrizing model parameters using StandardParametrizator
+12/19/2024 06:16:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: [Optimizer Building] Using LearningRateForSP as learning rate
+12/19/2024 06:16:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: [ZeRO sharding] Size of optimizer params per rank:
+12/19/2024 06:16:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: [ZeRO sharding] DP Rank 0 has 14M out of 447M (3.12%) params' optimizer states
+12/19/2024 06:16:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: [ZeRO sharding] DP Rank 1 has 14M out of 447M (3.12%) params' optimizer states
+12/19/2024 06:16:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: [ZeRO sharding] DP Rank 2 has 14M out of 447M (3.12%) params' optimizer states
+12/19/2024 06:16:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: [ZeRO sharding] DP Rank 3 has 14M out of 447M (3.12%) params' optimizer states
+12/19/2024 06:16:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: [ZeRO sharding] DP Rank 4 has 14M out of 447M (3.12%) params' optimizer states
+12/19/2024 06:16:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: [ZeRO sharding] DP Rank 5 has 14M out of 447M (3.12%) params' optimizer states
+12/19/2024 06:16:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: [ZeRO sharding] DP Rank 6 has 14M out of 447M (3.12%) params' optimizer states
+12/19/2024 06:16:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: [ZeRO sharding] DP Rank 7 has 14M out of 447M (3.12%) params' optimizer states
+12/19/2024 06:16:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: [ZeRO sharding] DP Rank 8 has 14M out of 447M (3.12%) params' optimizer states
+12/19/2024 06:16:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: [ZeRO sharding] DP Rank 9 has 14M out of 447M (3.12%) params' optimizer states
+12/19/2024 06:16:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: [ZeRO sharding] DP Rank 10 has 14M out of 447M (3.12%) params' optimizer states
+12/19/2024 06:16:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: [ZeRO sharding] DP Rank 11 has 14M out of 447M (3.12%) params' optimizer states
+12/19/2024 06:16:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: [ZeRO sharding] DP Rank 12 has 14M out of 447M (3.12%) params' optimizer states
+12/19/2024 06:16:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: [ZeRO sharding] DP Rank 13 has 14M out of 447M (3.12%) params' optimizer states
+12/19/2024 06:16:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: [ZeRO sharding] DP Rank 14 has 14M out of 447M (3.12%) params' optimizer states
+12/19/2024 06:16:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: [ZeRO sharding] DP Rank 15 has 14M out of 447M (3.12%) params' optimizer states
+12/19/2024 06:16:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: [ZeRO sharding] DP Rank 16 has 14M out of 447M (3.12%) params' optimizer states
+12/19/2024 06:16:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: [ZeRO sharding] DP Rank 17 has 14M out of 447M (3.12%) params' optimizer states
+12/19/2024 06:16:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: [ZeRO sharding] DP Rank 18 has 14M out of 447M (3.12%) params' optimizer states
+12/19/2024 06:16:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: [ZeRO sharding] DP Rank 19 has 14M out of 447M (3.12%) params' optimizer states
+12/19/2024 06:16:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: [ZeRO sharding] DP Rank 20 has 14M out of 447M (3.12%) params' optimizer states
+12/19/2024 06:16:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: [ZeRO sharding] DP Rank 21 has 14M out of 447M (3.12%) params' optimizer states
+12/19/2024 06:16:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: [ZeRO sharding] DP Rank 22 has 14M out of 447M (3.12%) params' optimizer states
+12/19/2024 06:16:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: [ZeRO sharding] DP Rank 23 has 14M out of 447M (3.12%) params' optimizer states
+12/19/2024 06:16:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: [ZeRO sharding] DP Rank 24 has 14M out of 447M (3.12%) params' optimizer states
+12/19/2024 06:16:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: [ZeRO sharding] DP Rank 25 has 14M out of 447M (3.12%) params' optimizer states
+12/19/2024 06:16:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: [ZeRO sharding] DP Rank 26 has 14M out of 447M (3.12%) params' optimizer states
+12/19/2024 06:16:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: [ZeRO sharding] DP Rank 27 has 14M out of 447M (3.12%) params' optimizer states
+12/19/2024 06:16:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: [ZeRO sharding] DP Rank 28 has 14M out of 447M (3.12%) params' optimizer states
+12/19/2024 06:16:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: [ZeRO sharding] DP Rank 29 has 14M out of 447M (3.12%) params' optimizer states
+12/19/2024 06:16:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: [ZeRO sharding] DP Rank 30 has 14M out of 447M (3.12%) params' optimizer states
+12/19/2024 06:16:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: [ZeRO sharding] DP Rank 31 has 14M out of 447M (3.12%) params' optimizer states
+12/19/2024 06:16:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: [Training Plan] Stage Stable Training Stage has 99 remaining training steps and has consumed 0 samples
+12/19/2024 06:16:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: Using dummy data generator
+12/19/2024 06:16:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: [Training Plan] There are 1 training stages 
+12/19/2024 06:16:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: [Stage Stable Training Stage] start from step 1 
+12/19/2024 06:16:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: 
+12/19/2024 06:16:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: [Start training] datetime: 2024-12-19 06:16:50.950020 | mbs: 1 | grad_accum: 8 | global_batch_size: 256 | sequence_length: 4096 | train_steps: 100 | start_iteration_step: 0 | consumed_train_samples: 0
+12/19/2024 06:16:52 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: Resuming training from stage Stable Training Stage, it has trained for 0 samples and has 99 remaining train steps
+12/19/2024 06:16:52 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:  Memory usage: 2610.81MiB. Peak allocated 5384.00MiB. Peak reserved: 19392.00MiB
+12/19/2024 06:16:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:  Memory usage: 2743.88MiB. Peak allocated 5035.96MiB. Peak reserved: 5688.00MiB
+12/19/2024 06:16:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: iteration: 1 / 100 | consumed_tokens: 1.05M | elapsed_time_per_iteration_ms: 3.67K | tokens_per_sec: 286K | tokens_per_sec_per_gpu: 1.12K | global_batch_size: 256 | lm_loss: 12.4 | lr: 0.00015 | model_tflops_per_gpu: 28.7 | hardware_tflops_per_gpu: 28.7 | grad_norm: 1.14 | cuda_memory_allocated: 2.99G | cuda_max_memory_reserved: 6.96G | hd_total_memory_tb: 312G | hd_used_memory_tb: 69.2G | hd_free_memory_tb: 243G
+12/19/2024 06:16:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:  Memory usage: 2850.52MiB. Peak allocated 4581.84MiB. Peak reserved: 6638.00MiB
+12/19/2024 06:16:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:  Memory usage: 2850.53MiB. Peak allocated 5145.61MiB. Peak reserved: 6638.00MiB
+12/19/2024 06:16:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: iteration: 2 / 100 | consumed_tokens: 2.1M | elapsed_time_per_iteration_ms: 1.29K | tokens_per_sec: 812K | tokens_per_sec_per_gpu: 3.17K | global_batch_size: 256 | lm_loss: 12.4 | lr: 0.0003 | model_tflops_per_gpu: 81.5 | hardware_tflops_per_gpu: 81.5 | grad_norm: 1.14 | cuda_memory_allocated: 2.99G | cuda_max_memory_reserved: 6.96G | hd_total_memory_tb: 312G | hd_used_memory_tb: 69.2G | hd_free_memory_tb: 243G
+12/19/2024 06:16:57 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:  Memory usage: 2850.52MiB. Peak allocated 4581.84MiB. Peak reserved: 6638.00MiB
+12/19/2024 06:16:58 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:  Memory usage: 2850.53MiB. Peak allocated 5141.61MiB. Peak reserved: 6638.00MiB
+12/19/2024 06:16:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: iteration: 3 / 100 | consumed_tokens: 3.15M | elapsed_time_per_iteration_ms: 1.31K | tokens_per_sec: 799K | tokens_per_sec_per_gpu: 3.12K | global_batch_size: 256 | lm_loss: 12.4 | lr: 0.000296 | model_tflops_per_gpu: 80.1 | hardware_tflops_per_gpu: 80.1 | grad_norm: 1.11 | cuda_memory_allocated: 2.99G | cuda_max_memory_reserved: 6.96G | hd_total_memory_tb: 312G | hd_used_memory_tb: 69.2G | hd_free_memory_tb: 243G
+12/19/2024 06:16:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: | job_id   | name                                                           | nodes | seq_len | mbs | batch_accum | gbs | mTFLOPs | hTFLOPs | tok/s/gpu | AllReduce (GB/s) | AllGather (GB/s) | ReduceScatter (GB/s) | AR Intra-node (GB/s) | AG Intra-node (GB/s) | RS Intra-node (GB/s) | Mem Alloc (GB) | Mem Res (GB) | dp | pp | tp | pp_engine                           | tp_mode        | tp_async_comm | hidden_size | hidden_act | num_layers | num_heads | num_kv_heads | max_pos | vocab_size | tie_word_embeddings | dtype          | zero_stage | ddp_bucket_cap_mb | accumulate_grad_in_fp32 | Total Params | Local Params |
+12/19/2024 06:16:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: | -------- | -------------------------------------------------------------- | ----- | ------- | --- | ----------- | --- | ------- | ------- | --------- | ---------------- | ---------------- | -------------------- | -------------------- | -------------------- | -------------------- | -------------- | ------------ | -- | -- | -- | ----------------------------------- | -------------- | ------------- | ----------- | ---------- | ---------- | --------- | ------------ | ------- | ---------- | ------------------- | -------------- | ---------- | ----------------- | ----------------------- | ------------ | ------------ |
+12/19/2024 06:16:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: | 13502078 | 3.57G_dp32_tp8_pp1_acc8_mbs1_seq4096_zero1_tpmodeRED_vocab131k | 32    | 4096    | 1   | 8           | 256 | 80.12   | 80.12   | 3121.14   | 176.76           | 73.86            | 74.64                | 456.96               | 263.58               | 263.62               | 4.47           | 6.48         | 32 | 1  | 8  | OneForwardOneBackwardPipelineEngine | REDUCE_SCATTER | True          | 3072        | silu       | 28         | 32        | 32           | 4096    | 131072     | True                | torch.bfloat16 | 1          | 25                | True                    | 3.57G        | 447M         |
+12/19/2024 06:16:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: Attempting to write benchmark results to CSV file: benchmark/results/bench_final2.csv
+12/19/2024 06:16:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: Acquired lock for CSV file: benchmark/results/bench_final2.csv
+12/19/2024 06:16:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: Successfully wrote to CSV file: benchmark/results/bench_final2.csv. Releasing lock...
+12/19/2024 06:16:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: Throughput logging complete
+[2024-12-19 06:17:19,315] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 420106 closing signal SIGTERM
+[2024-12-19 06:17:19,315] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 420107 closing signal SIGTERM
+[2024-12-19 06:17:19,315] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 420108 closing signal SIGTERM
+[2024-12-19 06:17:19,315] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 420109 closing signal SIGTERM
+[2024-12-19 06:17:19,315] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 420111 closing signal SIGTERM
+[2024-12-19 06:17:19,315] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 420112 closing signal SIGTERM
+[2024-12-19 06:17:19,313] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 407925 closing signal SIGTERM
+[2024-12-19 06:17:19,313] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 407927 closing signal SIGTERM
+[2024-12-19 06:17:19,313] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 407929 closing signal SIGTERM
+[2024-12-19 06:17:19,313] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 407930 closing signal SIGTERM
+[2024-12-19 06:17:19,313] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 407931 closing signal SIGTERM
+[2024-12-19 06:17:19,316] torch.distributed.elastic.multiprocessing.api: [ERROR] failed (exitcode: 1) local_rank: 0 (pid: 41003) of binary: /fsx/nouamane/miniconda/envs/2-1-cu121/bin/python
+[2024-12-19 06:17:19,318] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 432683 closing signal SIGTERM
+[2024-12-19 06:17:19,318] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 432684 closing signal SIGTERM
+[2024-12-19 06:17:19,318] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 432685 closing signal SIGTERM
+[2024-12-19 06:17:19,318] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 432686 closing signal SIGTERM
+[2024-12-19 06:17:19,318] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 432687 closing signal SIGTERM
+[2024-12-19 06:17:19,318] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 432688 closing signal SIGTERM
+[2024-12-19 06:17:19,317] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 126784 closing signal SIGTERM
+[2024-12-19 06:17:19,317] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 126785 closing signal SIGTERM
+[2024-12-19 06:17:19,317] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 126787 closing signal SIGTERM
+[2024-12-19 06:17:19,317] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 126788 closing signal SIGTERM
+[2024-12-19 06:17:19,317] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 126789 closing signal SIGTERM
+[2024-12-19 06:17:19,317] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 322580 closing signal SIGTERM
+[2024-12-19 06:17:19,317] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 322581 closing signal SIGTERM
+[2024-12-19 06:17:19,317] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 322582 closing signal SIGTERM
+[2024-12-19 06:17:19,317] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 322584 closing signal SIGTERM
+[2024-12-19 06:17:19,317] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 322585 closing signal SIGTERM
+[2024-12-19 06:17:19,320] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 81652 closing signal SIGTERM
+[2024-12-19 06:17:19,320] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 81654 closing signal SIGTERM
+[2024-12-19 06:17:19,320] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 81655 closing signal SIGTERM
+[2024-12-19 06:17:19,320] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 81656 closing signal SIGTERM
+[2024-12-19 06:17:19,320] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 81657 closing signal SIGTERM
+[2024-12-19 06:17:19,320] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 81658 closing signal SIGTERM
+[2024-12-19 06:17:19,320] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 525491 closing signal SIGTERM
+[2024-12-19 06:17:19,320] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 525492 closing signal SIGTERM
+[2024-12-19 06:17:19,320] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 525493 closing signal SIGTERM
+[2024-12-19 06:17:19,320] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 525494 closing signal SIGTERM
+[2024-12-19 06:17:19,320] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 525495 closing signal SIGTERM
+[2024-12-19 06:17:19,320] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 525496 closing signal SIGTERM
+[2024-12-19 06:17:19,321] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1391664 closing signal SIGTERM
+[2024-12-19 06:17:19,322] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1391666 closing signal SIGTERM
+[2024-12-19 06:17:19,322] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1391667 closing signal SIGTERM
+[2024-12-19 06:17:19,322] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1391668 closing signal SIGTERM
+[2024-12-19 06:17:19,322] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1391669 closing signal SIGTERM
+[2024-12-19 06:17:19,322] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1391670 closing signal SIGTERM
+[2024-12-19 06:17:19,321] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 326953 closing signal SIGTERM
+[2024-12-19 06:17:19,321] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 326954 closing signal SIGTERM
+[2024-12-19 06:17:19,321] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 326955 closing signal SIGTERM
+[2024-12-19 06:17:19,321] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 326956 closing signal SIGTERM
+[2024-12-19 06:17:19,321] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 326957 closing signal SIGTERM
+[2024-12-19 06:17:19,321] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 326958 closing signal SIGTERM
+[2024-12-19 06:17:19,321] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 540110 closing signal SIGTERM
+[2024-12-19 06:17:19,322] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 540111 closing signal SIGTERM
+[2024-12-19 06:17:19,322] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 540113 closing signal SIGTERM
+[2024-12-19 06:17:19,322] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1501902 closing signal SIGTERM
+[2024-12-19 06:17:19,322] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 540114 closing signal SIGTERM
+[2024-12-19 06:17:19,322] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 540115 closing signal SIGTERM
+[2024-12-19 06:17:19,322] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1501903 closing signal SIGTERM
+[2024-12-19 06:17:19,322] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1501904 closing signal SIGTERM
+[2024-12-19 06:17:19,322] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1501906 closing signal SIGTERM
+[2024-12-19 06:17:19,322] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1501908 closing signal SIGTERM
+[2024-12-19 06:17:19,325] torch.distributed.elastic.multiprocessing.api: [ERROR] failed (exitcode: 1) local_rank: 0 (pid: 221311) of binary: /fsx/nouamane/miniconda/envs/2-1-cu121/bin/python
+[2024-12-19 06:17:19,324] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 429092 closing signal SIGTERM
+[2024-12-19 06:17:19,324] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 429093 closing signal SIGTERM
+[2024-12-19 06:17:19,324] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 429094 closing signal SIGTERM
+[2024-12-19 06:17:19,324] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 429095 closing signal SIGTERM
+[2024-12-19 06:17:19,324] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 429096 closing signal SIGTERM
+[2024-12-19 06:17:19,324] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 429098 closing signal SIGTERM
+[2024-12-19 06:17:19,326] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 238542 closing signal SIGTERM
+[2024-12-19 06:17:19,326] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 238543 closing signal SIGTERM
+[2024-12-19 06:17:19,326] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 238544 closing signal SIGTERM
+[2024-12-19 06:17:19,325] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 223112 closing signal SIGTERM
+[2024-12-19 06:17:19,325] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 223114 closing signal SIGTERM
+[2024-12-19 06:17:19,325] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 223115 closing signal SIGTERM
+[2024-12-19 06:17:19,326] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 238546 closing signal SIGTERM
+[2024-12-19 06:17:19,325] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 223117 closing signal SIGTERM
+[2024-12-19 06:17:19,326] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 238547 closing signal SIGTERM
+[2024-12-19 06:17:19,325] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 223118 closing signal SIGTERM
+[2024-12-19 06:17:19,326] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 238548 closing signal SIGTERM
+[2024-12-19 06:17:19,325] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 438123 closing signal SIGTERM
+[2024-12-19 06:17:19,325] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 350879 closing signal SIGTERM
+[2024-12-19 06:17:19,325] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 438124 closing signal SIGTERM
+[2024-12-19 06:17:19,325] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 350881 closing signal SIGTERM
+[2024-12-19 06:17:19,325] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 438125 closing signal SIGTERM
+[2024-12-19 06:17:19,325] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 350882 closing signal SIGTERM
+[2024-12-19 06:17:19,325] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 438126 closing signal SIGTERM
+[2024-12-19 06:17:19,326] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 350883 closing signal SIGTERM
+[2024-12-19 06:17:19,326] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 350884 closing signal SIGTERM
+[2024-12-19 06:17:19,326] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 117104 closing signal SIGTERM
+[2024-12-19 06:17:19,324] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 407003 closing signal SIGTERM
+[2024-12-19 06:17:19,325] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 438127 closing signal SIGTERM
+[2024-12-19 06:17:19,325] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 438129 closing signal SIGTERM
+[2024-12-19 06:17:19,326] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 350885 closing signal SIGTERM
+[2024-12-19 06:17:19,324] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 407004 closing signal SIGTERM
+[2024-12-19 06:17:19,326] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1392199 closing signal SIGTERM
+[2024-12-19 06:17:19,326] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 117105 closing signal SIGTERM
+[2024-12-19 06:17:19,326] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 117106 closing signal SIGTERM
+[2024-12-19 06:17:19,325] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 407005 closing signal SIGTERM
+[2024-12-19 06:17:19,326] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 117107 closing signal SIGTERM
+[2024-12-19 06:17:19,326] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 117108 closing signal SIGTERM
+[2024-12-19 06:17:19,326] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1392200 closing signal SIGTERM
+[2024-12-19 06:17:19,325] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 407007 closing signal SIGTERM
+[2024-12-19 06:17:19,326] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 117109 closing signal SIGTERM
+[2024-12-19 06:17:19,326] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1392201 closing signal SIGTERM
+[2024-12-19 06:17:19,326] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 117110 closing signal SIGTERM
+[2024-12-19 06:17:19,326] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1392202 closing signal SIGTERM
+[2024-12-19 06:17:19,326] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1392203 closing signal SIGTERM
+[2024-12-19 06:17:19,325] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 407008 closing signal SIGTERM
+[2024-12-19 06:17:19,326] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1392204 closing signal SIGTERM
+[2024-12-19 06:17:19,325] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 407009 closing signal SIGTERM
+[2024-12-19 06:17:19,327] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 114754 closing signal SIGTERM
+[2024-12-19 06:17:19,328] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 114755 closing signal SIGTERM
+[2024-12-19 06:17:19,328] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 114756 closing signal SIGTERM
+[2024-12-19 06:17:19,328] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 114757 closing signal SIGTERM
+[2024-12-19 06:17:19,328] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 114758 closing signal SIGTERM
+[2024-12-19 06:17:19,328] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 114759 closing signal SIGTERM
+[2024-12-19 06:17:19,328] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 114760 closing signal SIGTERM
+[2024-12-19 06:17:19,325] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 50397 closing signal SIGTERM
+[2024-12-19 06:17:19,325] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 50399 closing signal SIGTERM
+[2024-12-19 06:17:19,325] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 50400 closing signal SIGTERM
+[2024-12-19 06:17:19,325] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 50401 closing signal SIGTERM
+[2024-12-19 06:17:19,325] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 50402 closing signal SIGTERM
+[2024-12-19 06:17:19,325] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 50403 closing signal SIGTERM
+[2024-12-19 06:17:19,327] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 41448 closing signal SIGTERM
+[2024-12-19 06:17:19,327] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 41449 closing signal SIGTERM
+[2024-12-19 06:17:19,327] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 41450 closing signal SIGTERM
+[2024-12-19 06:17:19,327] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 21814 closing signal SIGTERM
+[2024-12-19 06:17:19,327] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 41451 closing signal SIGTERM
+[2024-12-19 06:17:19,328] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 41452 closing signal SIGTERM
+[2024-12-19 06:17:19,328] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 41453 closing signal SIGTERM
+[2024-12-19 06:17:19,327] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 21815 closing signal SIGTERM
+[2024-12-19 06:17:19,328] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 41454 closing signal SIGTERM
+[2024-12-19 06:17:19,327] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 21816 closing signal SIGTERM
+[2024-12-19 06:17:19,327] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 21817 closing signal SIGTERM
+[2024-12-19 06:17:19,327] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 21818 closing signal SIGTERM
+[2024-12-19 06:17:19,327] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 580775 closing signal SIGTERM
+[2024-12-19 06:17:19,327] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 21819 closing signal SIGTERM
+[2024-12-19 06:17:19,327] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 580776 closing signal SIGTERM
+[2024-12-19 06:17:19,327] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 580778 closing signal SIGTERM
+[2024-12-19 06:17:19,327] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 580779 closing signal SIGTERM
+[2024-12-19 06:17:19,327] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 580780 closing signal SIGTERM
+[2024-12-19 06:17:19,327] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 580781 closing signal SIGTERM
+[2024-12-19 06:17:19,328] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 39936 closing signal SIGTERM
+[2024-12-19 06:17:19,328] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 39937 closing signal SIGTERM
+[2024-12-19 06:17:19,328] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 39938 closing signal SIGTERM
+[2024-12-19 06:17:19,328] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 39939 closing signal SIGTERM
+[2024-12-19 06:17:19,329] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 431818 closing signal SIGTERM
+[2024-12-19 06:17:19,328] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 39940 closing signal SIGTERM
+[2024-12-19 06:17:19,329] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 431820 closing signal SIGTERM
+[2024-12-19 06:17:19,328] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 39941 closing signal SIGTERM
+[2024-12-19 06:17:19,324] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 236333 closing signal SIGTERM
+[2024-12-19 06:17:19,329] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 431821 closing signal SIGTERM
+[2024-12-19 06:17:19,328] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 39942 closing signal SIGTERM
+[2024-12-19 06:17:19,329] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 431822 closing signal SIGTERM
+[2024-12-19 06:17:19,329] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 431823 closing signal SIGTERM
+[2024-12-19 06:17:19,325] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 236334 closing signal SIGTERM
+[2024-12-19 06:17:19,329] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 431824 closing signal SIGTERM
+[2024-12-19 06:17:19,325] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 236336 closing signal SIGTERM
+[2024-12-19 06:17:19,325] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 236337 closing signal SIGTERM
+[2024-12-19 06:17:19,325] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 236338 closing signal SIGTERM
+[2024-12-19 06:17:19,325] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 236339 closing signal SIGTERM
+[2024-12-19 06:17:19,331] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1812117 closing signal SIGTERM
+[2024-12-19 06:17:19,331] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1812118 closing signal SIGTERM
+[2024-12-19 06:17:19,331] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1812120 closing signal SIGTERM
+[2024-12-19 06:17:19,331] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1812121 closing signal SIGTERM
+[2024-12-19 06:17:19,331] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1812122 closing signal SIGTERM
+[2024-12-19 06:17:19,331] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1812123 closing signal SIGTERM
+[2024-12-19 06:17:19,328] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 405275 closing signal SIGTERM
+[2024-12-19 06:17:19,329] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 406666 closing signal SIGTERM
+[2024-12-19 06:17:19,328] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 405276 closing signal SIGTERM
+[2024-12-19 06:17:19,329] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 406667 closing signal SIGTERM
+[2024-12-19 06:17:19,329] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 406668 closing signal SIGTERM
+[2024-12-19 06:17:19,328] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 405278 closing signal SIGTERM
+[2024-12-19 06:17:19,329] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 406669 closing signal SIGTERM
+[2024-12-19 06:17:19,329] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 406670 closing signal SIGTERM
+[2024-12-19 06:17:19,328] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 405279 closing signal SIGTERM
+[2024-12-19 06:17:19,329] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 406671 closing signal SIGTERM
+[2024-12-19 06:17:19,328] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 405280 closing signal SIGTERM
+[2024-12-19 06:17:19,328] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 405281 closing signal SIGTERM
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 264, in launch_agent
+    raise ChildFailedError(
+torch.distributed.elastic.multiprocessing.errors.ChildFailedError: 
+============================================================
+/fsx/nouamane/projects/nanotron/run_train.py FAILED
+------------------------------------------------------------
+Failures:
+[1]:
+  time      : 2024-12-19_06:17:19
+  host      : ip-26-0-171-62.ec2.internal
+  rank      : 249 (local_rank: 1)
+  exitcode  : 1 (pid: 41004)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[2]:
+  time      : 2024-12-19_06:17:19
+  host      : ip-26-0-171-62.ec2.internal
+  rank      : 250 (local_rank: 2)
+  exitcode  : 1 (pid: 41005)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[3]:
+  time      : 2024-12-19_06:17:19
+  host      : ip-26-0-171-62.ec2.internal
+  rank      : 251 (local_rank: 3)
+  exitcode  : 1 (pid: 41006)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[4]:
+  time      : 2024-12-19_06:17:19
+  host      : ip-26-0-171-62.ec2.internal
+  rank      : 252 (local_rank: 4)
+  exitcode  : 1 (pid: 41007)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[5]:
+  time      : 2024-12-19_06:17:19
+  host      : ip-26-0-171-62.ec2.internal
+  rank      : 253 (local_rank: 5)
+  exitcode  : 1 (pid: 41008)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[6]:
+  time      : 2024-12-19_06:17:19
+  host      : ip-26-0-171-62.ec2.internal
+  rank      : 254 (local_rank: 6)
+  exitcode  : 1 (pid: 41009)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[7]:
+  time      : 2024-12-19_06:17:19
+  host      : ip-26-0-171-62.ec2.internal
+  rank      : 255 (local_rank: 7)
+  exitcode  : 1 (pid: 41010)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+------------------------------------------------------------
+Root Cause (first observed failure):
+[0]:
+  time      : 2024-12-19_06:17:19
+  host      : ip-26-0-171-62.ec2.internal
+  rank      : 248 (local_rank: 0)
+  exitcode  : 1 (pid: 41003)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+============================================================
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 264, in launch_agent
+    raise ChildFailedError(
+torch.distributed.elastic.multiprocessing.errors.ChildFailedError: 
+============================================================
+/fsx/nouamane/projects/nanotron/run_train.py FAILED
+------------------------------------------------------------
+Failures:
+[1]:
+  time      : 2024-12-19_06:17:19
+  host      : ip-26-0-162-180.ec2.internal
+  rank      : 1 (local_rank: 1)
+  exitcode  : 1 (pid: 221312)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[2]:
+  time      : 2024-12-19_06:17:19
+  host      : ip-26-0-162-180.ec2.internal
+  rank      : 2 (local_rank: 2)
+  exitcode  : 1 (pid: 221313)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[3]:
+  time      : 2024-12-19_06:17:19
+  host      : ip-26-0-162-180.ec2.internal
+  rank      : 3 (local_rank: 3)
+  exitcode  : 1 (pid: 221314)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[4]:
+  time      : 2024-12-19_06:17:19
+  host      : ip-26-0-162-180.ec2.internal
+  rank      : 4 (local_rank: 4)
+  exitcode  : 1 (pid: 221315)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[5]:
+  time      : 2024-12-19_06:17:19
+  host      : ip-26-0-162-180.ec2.internal
+  rank      : 5 (local_rank: 5)
+  exitcode  : 1 (pid: 221316)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[6]:
+  time      : 2024-12-19_06:17:19
+  host      : ip-26-0-162-180.ec2.internal
+  rank      : 6 (local_rank: 6)
+  exitcode  : 1 (pid: 221317)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[7]:
+  time      : 2024-12-19_06:17:19
+  host      : ip-26-0-162-180.ec2.internal
+  rank      : 7 (local_rank: 7)
+  exitcode  : 1 (pid: 221318)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+------------------------------------------------------------
+Root Cause (first observed failure):
+[0]:
+  time      : 2024-12-19_06:17:19
+  host      : ip-26-0-162-180.ec2.internal
+  rank      : 0 (local_rank: 0)
+  exitcode  : 1 (pid: 221311)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+============================================================
+srun: error: ip-26-0-171-62: task 31: Exited with exit code 1
+srun: launch/slurm: _step_signal: Terminating StepId=13502078.0
+slurmstepd: error: *** STEP 13502078.0 ON ip-26-0-162-79 CANCELLED AT 2024-12-19T06:17:19 ***
+[2024-12-19 06:17:19,741] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-19 06:17:19,739] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-19 06:17:19,741] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 223112 closing signal SIGTERM
+[2024-12-19 06:17:19,739] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 322580 closing signal SIGTERM
+[2024-12-19 06:17:19,741] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 223114 closing signal SIGTERM
+[2024-12-19 06:17:19,739] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 322581 closing signal SIGTERM
+[2024-12-19 06:17:19,739] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 322582 closing signal SIGTERM
+[2024-12-19 06:17:19,739] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 322584 closing signal SIGTERM
+[2024-12-19 06:17:19,740] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-19 06:17:19,739] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 322585 closing signal SIGTERM
+[2024-12-19 06:17:19,740] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 326953 closing signal SIGTERM
+[2024-12-19 06:17:19,741] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 223118 closing signal SIGTERM
+[2024-12-19 06:17:19,740] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 326954 closing signal SIGTERM
+[2024-12-19 06:17:19,740] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 326955 closing signal SIGTERM
+[2024-12-19 06:17:19,740] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 326956 closing signal SIGTERM
+[2024-12-19 06:17:19,741] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 326957 closing signal SIGTERM
+[2024-12-19 06:17:19,741] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 326958 closing signal SIGTERM
+[2024-12-19 06:17:19,742] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-19 06:17:19,741] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-19 06:17:19,738] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-19 06:17:19,742] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 39936 closing signal SIGTERM
+[2024-12-19 06:17:19,741] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 126784 closing signal SIGTERM
+[2024-12-19 06:17:19,738] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 236333 closing signal SIGTERM
+[2024-12-19 06:17:19,742] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 39937 closing signal SIGTERM
+[2024-12-19 06:17:19,738] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 236334 closing signal SIGTERM
+[2024-12-19 06:17:19,742] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 39938 closing signal SIGTERM
+[2024-12-19 06:17:19,742] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-19 06:17:19,741] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 126785 closing signal SIGTERM
+[2024-12-19 06:17:19,738] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 236336 closing signal SIGTERM
+[2024-12-19 06:17:19,742] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 39939 closing signal SIGTERM
+[2024-12-19 06:17:19,741] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 126787 closing signal SIGTERM
+[2024-12-19 06:17:19,738] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 236337 closing signal SIGTERM
+[2024-12-19 06:17:19,742] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 39940 closing signal SIGTERM
+[2024-12-19 06:17:19,742] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 525491 closing signal SIGTERM
+[2024-12-19 06:17:19,741] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 126788 closing signal SIGTERM
+[2024-12-19 06:17:19,743] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-19 06:17:19,738] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 236338 closing signal SIGTERM
+[2024-12-19 06:17:19,742] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 39941 closing signal SIGTERM
+[2024-12-19 06:17:19,742] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 525492 closing signal SIGTERM
+[2024-12-19 06:17:19,741] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 126789 closing signal SIGTERM
+[2024-12-19 06:17:19,738] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 236339 closing signal SIGTERM
+[2024-12-19 06:17:19,742] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 39942 closing signal SIGTERM
+[2024-12-19 06:17:19,743] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 238542 closing signal SIGTERM
+[2024-12-19 06:17:19,742] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 525493 closing signal SIGTERM
+[2024-12-19 06:17:19,743] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 238543 closing signal SIGTERM
+[2024-12-19 06:17:19,743] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 238544 closing signal SIGTERM
+[2024-12-19 06:17:19,742] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 525494 closing signal SIGTERM
+[2024-12-19 06:17:19,743] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 238546 closing signal SIGTERM
+[2024-12-19 06:17:19,743] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 238547 closing signal SIGTERM
+[2024-12-19 06:17:19,742] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-19 06:17:19,742] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 525495 closing signal SIGTERM
+[2024-12-19 06:17:19,743] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 238548 closing signal SIGTERM
+[2024-12-19 06:17:19,741] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-19 06:17:19,743] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-19 06:17:19,742] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-19 06:17:19,742] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 525496 closing signal SIGTERM
+[2024-12-19 06:17:19,742] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 406666 closing signal SIGTERM
+[2024-12-19 06:17:19,741] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 540110 closing signal SIGTERM
+[2024-12-19 06:17:19,743] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 420106 closing signal SIGTERM
+[2024-12-19 06:17:19,742] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 438123 closing signal SIGTERM
+[2024-12-19 06:17:19,742] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 406667 closing signal SIGTERM
+[2024-12-19 06:17:19,743] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 420107 closing signal SIGTERM
+[2024-12-19 06:17:19,742] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-19 06:17:19,742] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 540111 closing signal SIGTERM
+[2024-12-19 06:17:19,742] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 438124 closing signal SIGTERM
+[2024-12-19 06:17:19,742] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-19 06:17:19,742] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 406668 closing signal SIGTERM
+[2024-12-19 06:17:19,743] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 420108 closing signal SIGTERM
+[2024-12-19 06:17:19,743] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-19 06:17:19,742] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 540113 closing signal SIGTERM
+[2024-12-19 06:17:19,742] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 438125 closing signal SIGTERM
+[2024-12-19 06:17:19,743] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-19 06:17:19,742] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 406669 closing signal SIGTERM
+[2024-12-19 06:17:19,741] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-19 06:17:19,743] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 420109 closing signal SIGTERM
+[2024-12-19 06:17:19,742] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 350879 closing signal SIGTERM
+[2024-12-19 06:17:19,743] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 114754 closing signal SIGTERM
+[2024-12-19 06:17:19,742] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 540114 closing signal SIGTERM
+[2024-12-19 06:17:19,742] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 438126 closing signal SIGTERM
+[2024-12-19 06:17:19,742] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 429092 closing signal SIGTERM
+[2024-12-19 06:17:19,742] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 406670 closing signal SIGTERM
+[2024-12-19 06:17:19,743] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 420111 closing signal SIGTERM
+[2024-12-19 06:17:19,742] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 350881 closing signal SIGTERM
+[2024-12-19 06:17:19,742] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 540115 closing signal SIGTERM
+[2024-12-19 06:17:19,742] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 438127 closing signal SIGTERM
+[2024-12-19 06:17:19,742] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 429093 closing signal SIGTERM
+[2024-12-19 06:17:19,743] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 431818 closing signal SIGTERM
+[2024-12-19 06:17:19,742] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 406671 closing signal SIGTERM
+[2024-12-19 06:17:19,741] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 405275 closing signal SIGTERM
+[2024-12-19 06:17:19,743] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 420112 closing signal SIGTERM
+[2024-12-19 06:17:19,742] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 350882 closing signal SIGTERM
+[2024-12-19 06:17:19,743] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 114755 closing signal SIGTERM
+[2024-12-19 06:17:19,742] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 438129 closing signal SIGTERM
+[2024-12-19 06:17:19,742] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-19 06:17:19,742] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 429094 closing signal SIGTERM
+[2024-12-19 06:17:19,743] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 431820 closing signal SIGTERM
+[2024-12-19 06:17:19,741] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 405276 closing signal SIGTERM
+[2024-12-19 06:17:19,742] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 350883 closing signal SIGTERM
+[2024-12-19 06:17:19,743] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 114756 closing signal SIGTERM
+[2024-12-19 06:17:19,742] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 21814 closing signal SIGTERM
+[2024-12-19 06:17:19,742] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 429095 closing signal SIGTERM
+[2024-12-19 06:17:19,743] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 431821 closing signal SIGTERM
+[2024-12-19 06:17:19,745] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-19 06:17:19,741] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 405278 closing signal SIGTERM
+[2024-12-19 06:17:19,742] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 350884 closing signal SIGTERM
+[2024-12-19 06:17:19,744] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 114757 closing signal SIGTERM
+[2024-12-19 06:17:19,742] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 21815 closing signal SIGTERM
+[2024-12-19 06:17:19,742] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 429096 closing signal SIGTERM
+[2024-12-19 06:17:19,743] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 431822 closing signal SIGTERM
+[2024-12-19 06:17:19,741] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 405279 closing signal SIGTERM
+[2024-12-19 06:17:19,743] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 350885 closing signal SIGTERM
+[2024-12-19 06:17:19,744] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 114758 closing signal SIGTERM
+[2024-12-19 06:17:19,742] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 21816 closing signal SIGTERM
+[2024-12-19 06:17:19,743] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 429098 closing signal SIGTERM
+[2024-12-19 06:17:19,743] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 431823 closing signal SIGTERM
+[2024-12-19 06:17:19,745] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1812117 closing signal SIGTERM
+[2024-12-19 06:17:19,742] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-19 06:17:19,741] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 405280 closing signal SIGTERM
+[2024-12-19 06:17:19,744] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 114759 closing signal SIGTERM
+[2024-12-19 06:17:19,742] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 21817 closing signal SIGTERM
+[2024-12-19 06:17:19,743] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-19 06:17:19,743] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 431824 closing signal SIGTERM
+[2024-12-19 06:17:19,741] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 405281 closing signal SIGTERM
+[2024-12-19 06:17:19,744] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 114760 closing signal SIGTERM
+[2024-12-19 06:17:19,742] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 21818 closing signal SIGTERM
+[2024-12-19 06:17:19,745] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1812118 closing signal SIGTERM
+[2024-12-19 06:17:19,743] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 21819 closing signal SIGTERM
+[2024-12-19 06:17:19,742] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 580775 closing signal SIGTERM
+[2024-12-19 06:17:19,744] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-19 06:17:19,742] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 580776 closing signal SIGTERM
+[2024-12-19 06:17:19,743] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-19 06:17:19,743] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 41448 closing signal SIGTERM
+[2024-12-19 06:17:19,745] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1812120 closing signal SIGTERM
+[2024-12-19 06:17:19,745] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1812121 closing signal SIGTERM
+[2024-12-19 06:17:19,742] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 580778 closing signal SIGTERM
+[2024-12-19 06:17:19,742] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 580779 closing signal SIGTERM
+[2024-12-19 06:17:19,743] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 41449 closing signal SIGTERM
+[2024-12-19 06:17:19,745] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1812122 closing signal SIGTERM
+[2024-12-19 06:17:19,743] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1392199 closing signal SIGTERM
+[2024-12-19 06:17:19,743] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 41450 closing signal SIGTERM
+[2024-12-19 06:17:19,744] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 432683 closing signal SIGTERM
+[2024-12-19 06:17:19,741] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-19 06:17:19,743] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1392200 closing signal SIGTERM
+[2024-12-19 06:17:19,741] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-19 06:17:19,741] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-19 06:17:19,742] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 580780 closing signal SIGTERM
+[2024-12-19 06:17:19,743] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-19 06:17:19,743] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 41451 closing signal SIGTERM
+[2024-12-19 06:17:19,744] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 432684 closing signal SIGTERM
+[2024-12-19 06:17:19,745] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1812123 closing signal SIGTERM
+[2024-12-19 06:17:19,742] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 580781 closing signal SIGTERM
+[2024-12-19 06:17:19,743] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1392201 closing signal SIGTERM
+[2024-12-19 06:17:19,743] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 41452 closing signal SIGTERM
+[2024-12-19 06:17:19,742] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 407003 closing signal SIGTERM
+[2024-12-19 06:17:19,743] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-19 06:17:19,743] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1392202 closing signal SIGTERM
+[2024-12-19 06:17:19,743] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 41453 closing signal SIGTERM
+[2024-12-19 06:17:19,741] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 50397 closing signal SIGTERM
+[2024-12-19 06:17:19,742] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-19 06:17:19,741] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 407925 closing signal SIGTERM
+[2024-12-19 06:17:19,743] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 81652 closing signal SIGTERM
+[2024-12-19 06:17:19,743] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1392203 closing signal SIGTERM
+[2024-12-19 06:17:19,743] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 41454 closing signal SIGTERM
+[2024-12-19 06:17:19,741] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 50399 closing signal SIGTERM
+[2024-12-19 06:17:19,744] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 432685 closing signal SIGTERM
+[2024-12-19 06:17:19,742] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 407004 closing signal SIGTERM
+[2024-12-19 06:17:19,741] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 50400 closing signal SIGTERM
+[2024-12-19 06:17:19,741] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 407929 closing signal SIGTERM
+[2024-12-19 06:17:19,743] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 81654 closing signal SIGTERM
+[2024-12-19 06:17:19,743] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 117104 closing signal SIGTERM
+[2024-12-19 06:17:19,741] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 50401 closing signal SIGTERM
+[2024-12-19 06:17:19,743] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1501902 closing signal SIGTERM
+[2024-12-19 06:17:19,743] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1392204 closing signal SIGTERM
+[2024-12-19 06:17:19,741] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 50402 closing signal SIGTERM
+[2024-12-19 06:17:19,744] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 432686 closing signal SIGTERM
+[2024-12-19 06:17:19,742] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 407005 closing signal SIGTERM
+[2024-12-19 06:17:19,741] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 407930 closing signal SIGTERM
+[2024-12-19 06:17:19,743] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 81655 closing signal SIGTERM
+[2024-12-19 06:17:19,744] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 117105 closing signal SIGTERM
+[2024-12-19 06:17:19,741] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 50403 closing signal SIGTERM
+[2024-12-19 06:17:19,743] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1501903 closing signal SIGTERM
+[2024-12-19 06:17:19,743] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 81656 closing signal SIGTERM
+[2024-12-19 06:17:19,744] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 117106 closing signal SIGTERM
+[2024-12-19 06:17:19,745] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 432687 closing signal SIGTERM
+[2024-12-19 06:17:19,742] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 407007 closing signal SIGTERM
+[2024-12-19 06:17:19,741] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 407931 closing signal SIGTERM
+[2024-12-19 06:17:19,743] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 81657 closing signal SIGTERM
+[2024-12-19 06:17:19,744] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 117107 closing signal SIGTERM
+[2024-12-19 06:17:19,743] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1501904 closing signal SIGTERM
+[2024-12-19 06:17:19,744] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-19 06:17:19,743] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 81658 closing signal SIGTERM
+[2024-12-19 06:17:19,744] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 117108 closing signal SIGTERM
+[2024-12-19 06:17:19,743] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1501906 closing signal SIGTERM
+[2024-12-19 06:17:19,742] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 407008 closing signal SIGTERM
+[2024-12-19 06:17:19,744] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 117109 closing signal SIGTERM
+[2024-12-19 06:17:19,745] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 432688 closing signal SIGTERM
+[2024-12-19 06:17:19,743] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1501908 closing signal SIGTERM
+[2024-12-19 06:17:19,744] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1391664 closing signal SIGTERM
+[2024-12-19 06:17:19,742] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 407009 closing signal SIGTERM
+[2024-12-19 06:17:19,744] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 117110 closing signal SIGTERM
+[2024-12-19 06:17:19,744] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1391666 closing signal SIGTERM
+[2024-12-19 06:17:19,744] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1391667 closing signal SIGTERM
+[2024-12-19 06:17:19,744] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1391668 closing signal SIGTERM
+[2024-12-19 06:17:19,744] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1391669 closing signal SIGTERM
+[2024-12-19 06:17:19,744] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1391670 closing signal SIGTERM
+srun: error: ip-26-0-162-180: task 1: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 878, in _invoke_run
+    run_result = self._monitor_workers(self._worker_group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/local_elastic_agent.py", line 307, in _monitor_workers
+    result = self._pcontext.wait(0)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 288, in wait
+    return self._poll()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 668, in _poll
+    self.close()  # terminate all running procs
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 331, in close
+    self._close(death_sig=death_sig, timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 713, in _close
+    handler.proc.wait(time_to_wait)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1209, in wait
+    return self._wait(timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1953, in _wait
+    time.sleep(delay)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 223038 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 878, in _invoke_run
+    run_result = self._monitor_workers(self._worker_group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/local_elastic_agent.py", line 307, in _monitor_workers
+    result = self._pcontext.wait(0)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 288, in wait
+    return self._poll()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 668, in _poll
+    self.close()  # terminate all running procs
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 331, in close
+    self._close(death_sig=death_sig, timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 713, in _close
+    handler.proc.wait(time_to_wait)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1209, in wait
+    return self._wait(timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1953, in _wait
+    time.sleep(delay)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 407852 got signal: 15
+srun: error: ip-26-0-162-79: task 0: Exited with exit code 1
+srun: error: ip-26-0-169-247: task 24: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 878, in _invoke_run
+    run_result = self._monitor_workers(self._worker_group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/local_elastic_agent.py", line 307, in _monitor_workers
+    result = self._pcontext.wait(0)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 288, in wait
+    return self._poll()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 668, in _poll
+    self.close()  # terminate all running procs
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 331, in close
+    self._close(death_sig=death_sig, timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 713, in _close
+    handler.proc.wait(time_to_wait)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1209, in wait
+    return self._wait(timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1953, in _wait
+    time.sleep(delay)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 50324 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 878, in _invoke_run
+    run_result = self._monitor_workers(self._worker_group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/local_elastic_agent.py", line 307, in _monitor_workers
+    result = self._pcontext.wait(0)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 288, in wait
+    return self._poll()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 668, in _poll
+    self.close()  # terminate all running procs
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 331, in close
+    self._close(death_sig=death_sig, timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 713, in _close
+    handler.proc.wait(time_to_wait)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1209, in wait
+    return self._wait(timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1953, in _wait
+    time.sleep(delay)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 126710 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 878, in _invoke_run
+    run_result = self._monitor_workers(self._worker_group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/local_elastic_agent.py", line 307, in _monitor_workers
+    result = self._pcontext.wait(0)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 288, in wait
+    return self._poll()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 668, in _poll
+    self.close()  # terminate all running procs
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 331, in close
+    self._close(death_sig=death_sig, timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 713, in _close
+    handler.proc.wait(time_to_wait)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1209, in wait
+    return self._wait(timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1953, in _wait
+    time.sleep(delay)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 540035 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 878, in _invoke_run
+    run_result = self._monitor_workers(self._worker_group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/local_elastic_agent.py", line 307, in _monitor_workers
+    result = self._pcontext.wait(0)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 288, in wait
+    return self._poll()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 668, in _poll
+    self.close()  # terminate all running procs
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 331, in close
+    self._close(death_sig=death_sig, timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 713, in _close
+    handler.proc.wait(time_to_wait)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1209, in wait
+    return self._wait(timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1953, in _wait
+    time.sleep(delay)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 322506 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 878, in _invoke_run
+    run_result = self._monitor_workers(self._worker_group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/local_elastic_agent.py", line 307, in _monitor_workers
+    result = self._pcontext.wait(0)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 288, in wait
+    return self._poll()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 668, in _poll
+    self.close()  # terminate all running procs
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 331, in close
+    self._close(death_sig=death_sig, timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 713, in _close
+    handler.proc.wait(time_to_wait)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1209, in wait
+    return self._wait(timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1953, in _wait
+srun: error: ip-26-0-168-120: task 17: Exited with exit code 1
+    time.sleep(delay)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 1501827 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 878, in _invoke_run
+    run_result = self._monitor_workers(self._worker_group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/local_elastic_agent.py", line 307, in _monitor_workers
+    result = self._pcontext.wait(0)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 288, in wait
+    return self._poll()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 668, in _poll
+    self.close()  # terminate all running procs
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 331, in close
+    self._close(death_sig=death_sig, timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 713, in _close
+    handler.proc.wait(time_to_wait)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1209, in wait
+    return self._wait(timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1953, in _wait
+    time.sleep(delay)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 1812044 got signal: 15
+srun: error: ip-26-0-162-233: task 2: Exited with exit code 1
+srun: error: ip-26-0-165-24: task 5: Exited with exit code 1
+srun: error: ip-26-0-170-132: task 26: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    elastic_launch(
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 878, in _invoke_run
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    run_result = self._monitor_workers(self._worker_group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 878, in _invoke_run
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/local_elastic_agent.py", line 307, in _monitor_workers
+    run_result = self._monitor_workers(self._worker_group)
+    result = self._pcontext.wait(0)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 288, in wait
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+    return self._poll()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 668, in _poll
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/local_elastic_agent.py", line 307, in _monitor_workers
+    result = self._pcontext.wait(0)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 288, in wait
+    self.close()  # terminate all running procs
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 331, in close
+    return self._poll()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 668, in _poll
+    self._close(death_sig=death_sig, timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 713, in _close
+    self.close()  # terminate all running procs
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 331, in close
+    handler.proc.wait(time_to_wait)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1209, in wait
+    self._close(death_sig=death_sig, timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 713, in _close
+    return self._wait(timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1953, in _wait
+    handler.proc.wait(time_to_wait)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1209, in wait
+    time.sleep(delay)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    return self._wait(timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1953, in _wait
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 81580 got signal: 15
+    time.sleep(delay)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 405202 got signal: 15
+srun: error: ip-26-0-165-131: task 8: Exited with exit code 1
+srun: error: ip-26-0-171-56: task 30: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 878, in _invoke_run
+    run_result = self._monitor_workers(self._worker_group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/local_elastic_agent.py", line 307, in _monitor_workers
+    result = self._pcontext.wait(0)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 288, in wait
+    return self._poll()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 668, in _poll
+    self.close()  # terminate all running procs
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 331, in close
+    self._close(death_sig=death_sig, timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 713, in _close
+    handler.proc.wait(time_to_wait)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1209, in wait
+    return self._wait(timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1953, in _wait
+    time.sleep(delay)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 525415 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 878, in _invoke_run
+    run_result = self._monitor_workers(self._worker_group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/local_elastic_agent.py", line 307, in _monitor_workers
+    result = self._pcontext.wait(0)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 288, in wait
+    return self._poll()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 668, in _poll
+    self.close()  # terminate all running procs
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 331, in close
+    self._close(death_sig=death_sig, timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 713, in _close
+    handler.proc.wait(time_to_wait)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1209, in wait
+    return self._wait(timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1953, in _wait
+    time.sleep(delay)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 580701 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 878, in _invoke_run
+    run_result = self._monitor_workers(self._worker_group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/local_elastic_agent.py", line 307, in _monitor_workers
+    result = self._pcontext.wait(0)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 288, in wait
+    return self._poll()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 668, in _poll
+    self.close()  # terminate all running procs
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 331, in close
+    self._close(death_sig=death_sig, timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 713, in _close
+    handler.proc.wait(time_to_wait)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1209, in wait
+    return self._wait(timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1953, in _wait
+    time.sleep(delay)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 406930 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 878, in _invoke_run
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    run_result = self._monitor_workers(self._worker_group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 878, in _invoke_run
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/local_elastic_agent.py", line 307, in _monitor_workers
+    result = self._pcontext.wait(0)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 288, in wait
+    run_result = self._monitor_workers(self._worker_group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    return self._poll()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 668, in _poll
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/local_elastic_agent.py", line 307, in _monitor_workers
+    self.close()  # terminate all running procs
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 331, in close
+    result = self._pcontext.wait(0)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 288, in wait
+    self._close(death_sig=death_sig, timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 713, in _close
+    return self._poll()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 668, in _poll
+    handler.proc.wait(time_to_wait)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1209, in wait
+    self.close()  # terminate all running procs
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 331, in close
+    return self._wait(timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1953, in _wait
+    self._close(death_sig=death_sig, timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 713, in _close
+    time.sleep(delay)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    handler.proc.wait(time_to_wait)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1209, in wait
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 429019 got signal: 15
+    return self._wait(timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1953, in _wait
+    time.sleep(delay)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 1392125 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 878, in _invoke_run
+    run_result = self._monitor_workers(self._worker_group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/local_elastic_agent.py", line 307, in _monitor_workers
+    result = self._pcontext.wait(0)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 288, in wait
+    return self._poll()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 668, in _poll
+    self.close()  # terminate all running procs
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 331, in close
+    self._close(death_sig=death_sig, timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 713, in _close
+    handler.proc.wait(time_to_wait)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1209, in wait
+    return self._wait(timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1953, in _wait
+    time.sleep(delay)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 432610 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 878, in _invoke_run
+    run_result = self._monitor_workers(self._worker_group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/local_elastic_agent.py", line 307, in _monitor_workers
+    result = self._pcontext.wait(0)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 288, in wait
+    return self._poll()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 668, in _poll
+    self.close()  # terminate all running procs
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 331, in close
+    self._close(death_sig=death_sig, timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 713, in _close
+    handler.proc.wait(time_to_wait)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1209, in wait
+    return self._wait(timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1953, in _wait
+    time.sleep(delay)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 21742 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 878, in _invoke_run
+    run_result = self._monitor_workers(self._worker_group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/local_elastic_agent.py", line 307, in _monitor_workers
+    result = self._pcontext.wait(0)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 288, in wait
+    return self._poll()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 668, in _poll
+    self.close()  # terminate all running procs
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 331, in close
+    self._close(death_sig=death_sig, timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 713, in _close
+    handler.proc.wait(time_to_wait)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1209, in wait
+    return self._wait(timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1953, in _wait
+    time.sleep(delay)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 420033 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 878, in _invoke_run
+    run_result = self._monitor_workers(self._worker_group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/local_elastic_agent.py", line 307, in _monitor_workers
+    result = self._pcontext.wait(0)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 288, in wait
+    return self._poll()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 668, in _poll
+    self.close()  # terminate all running procs
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 331, in close
+    self._close(death_sig=death_sig, timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 713, in _close
+    handler.proc.wait(time_to_wait)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1209, in wait
+    return self._wait(timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1953, in _wait
+    time.sleep(delay)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 1391591 got signal: 15
+srun: error: ip-26-0-166-15: task 12: Exited with exit code 1
+srun: error: ip-26-0-169-207: task 22: Exited with exit code 1
+srun: error: ip-26-0-166-36: task 13: Exited with exit code 1
+srun: error: ip-26-0-168-238: task 18: Exited with exit code 1
+srun: error: ip-26-0-166-125: task 14: Exited with exit code 1
+srun: error: ip-26-0-165-164: task 9: Exited with exit code 1
+srun: error: ip-26-0-169-139: task 21: Exited with exit code 1
+srun: error: ip-26-0-170-160: task 28: Exited with exit code 1
+srun: error: ip-26-0-165-213: task 11: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 878, in _invoke_run
+    run_result = self._monitor_workers(self._worker_group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/local_elastic_agent.py", line 307, in _monitor_workers
+    result = self._pcontext.wait(0)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 288, in wait
+    return self._poll()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 668, in _poll
+    self.close()  # terminate all running procs
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 331, in close
+    self._close(death_sig=death_sig, timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 713, in _close
+    handler.proc.wait(time_to_wait)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1209, in wait
+    return self._wait(timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1953, in _wait
+    time.sleep(delay)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 236259 got signal: 15
+srun: error: ip-26-0-169-86: task 19: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 878, in _invoke_run
+    run_result = self._monitor_workers(self._worker_group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/local_elastic_agent.py", line 307, in _monitor_workers
+    result = self._pcontext.wait(0)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 288, in wait
+    return self._poll()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 668, in _poll
+    self.close()  # terminate all running procs
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 331, in close
+    self._close(death_sig=death_sig, timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 713, in _close
+    handler.proc.wait(time_to_wait)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1209, in wait
+    return self._wait(timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1953, in _wait
+    time.sleep(delay)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 438049 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 878, in _invoke_run
+    run_result = self._monitor_workers(self._worker_group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/local_elastic_agent.py", line 307, in _monitor_workers
+    result = self._pcontext.wait(0)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 288, in wait
+    return self._poll()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 668, in _poll
+    self.close()  # terminate all running procs
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 331, in close
+    self._close(death_sig=death_sig, timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 713, in _close
+    handler.proc.wait(time_to_wait)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1209, in wait
+    return self._wait(timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1953, in _wait
+    time.sleep(delay)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 326878 got signal: 15
+srun: error: ip-26-0-170-143: task 27: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 878, in _invoke_run
+    run_result = self._monitor_workers(self._worker_group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/local_elastic_agent.py", line 307, in _monitor_workers
+    result = self._pcontext.wait(0)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 288, in wait
+    return self._poll()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 668, in _poll
+    self.close()  # terminate all running procs
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 331, in close
+    self._close(death_sig=death_sig, timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 713, in _close
+    handler.proc.wait(time_to_wait)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1209, in wait
+    return self._wait(timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1953, in _wait
+    time.sleep(delay)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 406594 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 878, in _invoke_run
+    run_result = self._monitor_workers(self._worker_group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/local_elastic_agent.py", line 307, in _monitor_workers
+    result = self._pcontext.wait(0)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 288, in wait
+    return self._poll()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 668, in _poll
+    self.close()  # terminate all running procs
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 331, in close
+    self._close(death_sig=death_sig, timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 713, in _close
+    handler.proc.wait(time_to_wait)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1209, in wait
+    return self._wait(timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1953, in _wait
+    time.sleep(delay)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 350806 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 878, in _invoke_run
+    run_result = self._monitor_workers(self._worker_group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/local_elastic_agent.py", line 307, in _monitor_workers
+    result = self._pcontext.wait(0)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 288, in wait
+    return self._poll()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 668, in _poll
+srun: error: ip-26-0-166-214: task 15: Exited with exit code 1
+    self.close()  # terminate all running procs
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 331, in close
+    self._close(death_sig=death_sig, timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 713, in _close
+    handler.proc.wait(time_to_wait)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1209, in wait
+    return self._wait(timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1953, in _wait
+    time.sleep(delay)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 238469 got signal: 15
+srun: error: ip-26-0-165-38: task 6: Exited with exit code 1
+srun: error: ip-26-0-163-43: task 3: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 878, in _invoke_run
+    run_result = self._monitor_workers(self._worker_group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/local_elastic_agent.py", line 307, in _monitor_workers
+    result = self._pcontext.wait(0)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 288, in wait
+    return self._poll()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 668, in _poll
+    self.close()  # terminate all running procs
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 331, in close
+    self._close(death_sig=death_sig, timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 713, in _close
+    handler.proc.wait(time_to_wait)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1209, in wait
+    return self._wait(timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1953, in _wait
+    time.sleep(delay)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 41374 got signal: 15
+srun: error: ip-26-0-165-202: task 10: Exited with exit code 1
+srun: error: ip-26-0-170-31: task 25: Exited with exit code 1
+srun: error: ip-26-0-164-236: task 4: Exited with exit code 1
+srun: error: ip-26-0-171-21: task 29: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 878, in _invoke_run
+    run_result = self._monitor_workers(self._worker_group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/local_elastic_agent.py", line 307, in _monitor_workers
+    result = self._pcontext.wait(0)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 288, in wait
+    return self._poll()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 668, in _poll
+    self.close()  # terminate all running procs
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 331, in close
+    self._close(death_sig=death_sig, timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 713, in _close
+    handler.proc.wait(time_to_wait)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1209, in wait
+    return self._wait(timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1953, in _wait
+    time.sleep(delay)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 431745 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 878, in _invoke_run
+    run_result = self._monitor_workers(self._worker_group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/local_elastic_agent.py", line 307, in _monitor_workers
+    result = self._pcontext.wait(0)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 288, in wait
+    return self._poll()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 668, in _poll
+    self.close()  # terminate all running procs
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 331, in close
+    self._close(death_sig=death_sig, timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 713, in _close
+    handler.proc.wait(time_to_wait)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1209, in wait
+    return self._wait(timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1953, in _wait
+    time.sleep(delay)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 39863 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 878, in _invoke_run
+    run_result = self._monitor_workers(self._worker_group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/local_elastic_agent.py", line 307, in _monitor_workers
+    result = self._pcontext.wait(0)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 288, in wait
+    return self._poll()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 668, in _poll
+    self.close()  # terminate all running procs
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 331, in close
+    self._close(death_sig=death_sig, timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 713, in _close
+    handler.proc.wait(time_to_wait)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1209, in wait
+    return self._wait(timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1953, in _wait
+    time.sleep(delay)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 114681 got signal: 15
+[2024-12-19 06:17:23,290] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-169-239.ec2.internal_117030_0' has failed to send a keep-alive heartbeat to the rendezvous '13502078' due to an error of type RendezvousConnectionError.
+srun: error: ip-26-0-169-132: task 20: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 878, in _invoke_run
+    run_result = self._monitor_workers(self._worker_group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/local_elastic_agent.py", line 307, in _monitor_workers
+    result = self._pcontext.wait(0)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 288, in wait
+    return self._poll()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 668, in _poll
+    self.close()  # terminate all running procs
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 331, in close
+    self._close(death_sig=death_sig, timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 713, in _close
+    handler.proc.wait(time_to_wait)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1209, in wait
+    return self._wait(timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1953, in _wait
+    time.sleep(delay)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 117030 got signal: 15
+srun: error: ip-26-0-165-59: task 7: Exited with exit code 1
+srun: error: ip-26-0-166-244: task 16: Exited with exit code 1
+srun: error: ip-26-0-169-239: task 23: Exited with exit code 1
diff --git a/logs/13505303-bench_3.57G_dp1_tp32_pp1_acc256_mbs1_seq4096_zero0_tpmodeRED_vocab131k.out b/logs/13505303-bench_3.57G_dp1_tp32_pp1_acc256_mbs1_seq4096_zero0_tpmodeRED_vocab131k.out
new file mode 100644
index 0000000000000000000000000000000000000000..f65e53918451c15e6bb85014e90c903e8a496b7a
--- /dev/null
+++ b/logs/13505303-bench_3.57G_dp1_tp32_pp1_acc256_mbs1_seq4096_zero0_tpmodeRED_vocab131k.out
@@ -0,0 +1,745 @@
++ '[' -z 13505303 ']'
++ source /etc/profile.d/modules.sh
+++ . /usr/share/modules/init/bash
++++ unset _mlshdbg
++++ '[' 0 = 1 ']'
++++ unset _mlre _mlIFS
++++ '[' -n x ']'
++++ _mlIFS=' 	
+'
++++ IFS=' '
++++ '[' -n '' ']'
+++++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash autoinit
++++ _mlcode='module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+MODULEPATH_modshare=/etc/environment-modules/modules:1:/usr/share/modules/\$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1; export MODULEPATH_modshare;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+LOADEDMODULES=; export LOADEDMODULES;
+MODULEPATH=/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/\$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles; export MODULEPATH;
+test 0;'
++++ _mlret=0
++++ '[' -n x ']'
++++ IFS=' 	
+'
++++ unset _mlIFS
++++ unset _mlre _mlv _mlrv
++++ '[' 0 -eq 0 ']'
++++ eval 'module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+MODULEPATH_modshare=/etc/environment-modules/modules:1:/usr/share/modules/\$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1; export MODULEPATH_modshare;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+LOADEDMODULES=; export LOADEDMODULES;
+MODULEPATH=/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/\$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles; export MODULEPATH;
+test 0;'
+++++ MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl
+++++ export MODULES_CMD
+++++ ENV=/usr/share/modules/init/profile.sh
+++++ export ENV
+++++ MODULEPATH_modshare='/etc/environment-modules/modules:1:/usr/share/modules/$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1'
+++++ export MODULEPATH_modshare
+++++ BASH_ENV=/usr/share/modules/init/bash
+++++ export BASH_ENV
+++++ MODULESHOME=/usr/share/modules
+++++ export MODULESHOME
+++++ LOADEDMODULES=
+++++ export LOADEDMODULES
+++++ MODULEPATH='/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles'
+++++ export MODULEPATH
+++++ test 0
++++ '[' 0 = 1 ']'
++++ '[' -t 2 ']'
++++ export -f module
++++ export -f switchml
++++ '[' 5 -ge 3 ']'
++++ [[ ehxB =~ i ]]
++++ [[ ! :/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin: =~ :/usr/bin: ]]
++++ '[' '!' -n '' ']'
++++ MANPATH=:
++++ export MANPATH
+++++ manpath
++++ [[ ! :/admin/home/nouamane/.local/share/man:/fsx/nouamane/miniconda/envs/2-1-cu121/man:/fsx/nouamane/miniconda/envs/2-1-cu121/share/man:/fsx/nouamane/miniconda/man:/fsx/nouamane/miniconda/share/man:/opt/slurm/share/man:/opt/amazon/openmpi/share/man:/opt/amazon/efa/share/man:/usr/local/man:/usr/local/share/man:/usr/share/man:/admin/home/nouamane/.fzf/man:: =~ :/usr/share/man: ]]
++++ unset _mlcode _mlret
++++ '[' -n '' ']'
++ module load cuda/12.1
++ unset _mlshdbg
++ '[' 0 = 1 ']'
++ unset _mlre _mlIFS
++ '[' -n x ']'
++ _mlIFS=' 	
+'
++ IFS=' '
++ '[' -n '' ']'
+++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash load cuda/12.1
++ eval 'CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include;' export 'CPATH;
+LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:;' export 'LD_LIBRARY_PATH;
+FI_EFA_FORK_SAFE=1;' export 'FI_EFA_FORK_SAFE;
+MANPATH=/usr/local/cuda-12.1/share/man::;' export 'MANPATH;
+LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64;' export 'LIBRARY_PATH;
+_LMFILES_=/usr/share/modules/modulefiles/cuda/12.1;' export '_LMFILES_;
+LOADEDMODULES=cuda/12.1;' export 'LOADEDMODULES;
+MPI_PATH=/opt/amazon/openmpi;' export 'MPI_PATH;
+NCCL_HOME_modshare=/opt/nccl/build:1;' export 'NCCL_HOME_modshare;
+NCCL_PROTO=simple;' export 'NCCL_PROTO;
+MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1;' export 'MANPATH_modshare;
+LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1;' export 'LIBRARY_PATH_modshare;
+NCCL_SOCKET_IFNAME=enp;' export 'NCCL_SOCKET_IFNAME;
+AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl;' export 'AWS_OFI_NCCL_HOME;
+NCCL_HOME=/opt/nccl/build;' export 'NCCL_HOME;
+FI_PROVIDER=efa;' export 'FI_PROVIDER;
+AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1;' export 'AWS_OFI_NCCL_HOME_modshare;
+CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1;' export 'CPATH_modshare;
+LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2::1:/opt/nccl/build/lib:1:/opt/aws-ofi-nccl/lib:2:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/efa/lib:1;' export 'LD_LIBRARY_PATH_modshare;
+FI_EFA_ENABLE_SHM_TRANSFER=1;' export 'FI_EFA_ENABLE_SHM_TRANSFER;
+_LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1;' export '_LMFILES__modshare;
+LOADEDMODULES_modshare=cuda/12.1:1;' export 'LOADEDMODULES_modshare;
+MPI_PATH_modshare=/opt/amazon/openmpi:1;' export 'MPI_PATH_modshare;
+PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin;' export 'PATH;
+CUDA_HOME=/usr/local/cuda-12.1;' export 'CUDA_HOME;
+PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/usr/local/cuda-12.1/include:1:/opt/amazon/efa/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/usr/sbin:1:/opt/slurm/bin:1:/fsx/nouamane/miniconda/bin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:2:/opt/amazon/openmpi/bin:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1;' export 'PATH_modshare;
+test' '0;'
+++ CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include
+++ export CPATH
+++ LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:
+++ export LD_LIBRARY_PATH
+++ FI_EFA_FORK_SAFE=1
+++ export FI_EFA_FORK_SAFE
+++ MANPATH=/usr/local/cuda-12.1/share/man::
+++ export MANPATH
+++ LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64
+++ export LIBRARY_PATH
+++ _LMFILES_=/usr/share/modules/modulefiles/cuda/12.1
+++ export _LMFILES_
+++ LOADEDMODULES=cuda/12.1
+++ export LOADEDMODULES
+++ MPI_PATH=/opt/amazon/openmpi
+++ export MPI_PATH
+++ NCCL_HOME_modshare=/opt/nccl/build:1
+++ export NCCL_HOME_modshare
+++ NCCL_PROTO=simple
+++ export NCCL_PROTO
+++ MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1
+++ export MANPATH_modshare
+++ LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1
+++ export LIBRARY_PATH_modshare
+++ NCCL_SOCKET_IFNAME=enp
+++ export NCCL_SOCKET_IFNAME
+++ AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl
+++ export AWS_OFI_NCCL_HOME
+++ NCCL_HOME=/opt/nccl/build
+++ export NCCL_HOME
+++ FI_PROVIDER=efa
+++ export FI_PROVIDER
+++ AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1
+++ export AWS_OFI_NCCL_HOME_modshare
+++ CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1
+++ export CPATH_modshare
+++ LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2::1:/opt/nccl/build/lib:1:/opt/aws-ofi-nccl/lib:2:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/efa/lib:1
+++ export LD_LIBRARY_PATH_modshare
+++ FI_EFA_ENABLE_SHM_TRANSFER=1
+++ export FI_EFA_ENABLE_SHM_TRANSFER
+++ _LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1
+++ export _LMFILES__modshare
+++ LOADEDMODULES_modshare=cuda/12.1:1
+++ export LOADEDMODULES_modshare
+++ MPI_PATH_modshare=/opt/amazon/openmpi:1
+++ export MPI_PATH_modshare
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ export PATH
+++ CUDA_HOME=/usr/local/cuda-12.1
+++ export CUDA_HOME
+++ PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/usr/local/cuda-12.1/include:1:/opt/amazon/efa/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/usr/sbin:1:/opt/slurm/bin:1:/fsx/nouamane/miniconda/bin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:2:/opt/amazon/openmpi/bin:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1
+++ export PATH_modshare
+++ test 0
++ _mlstatus=0
++ '[' -n x ']'
++ IFS=' 	
+'
++ unset _mlre _mlv _mlrv _mlIFS
++ '[' -n '' ']'
++ unset _mlshdbg
++ return 0
++ source /fsx/nouamane/miniconda/bin/activate
+++ _CONDA_ROOT=/fsx/nouamane/miniconda
+++ . /fsx/nouamane/miniconda/etc/profile.d/conda.sh
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ '[' -z x ']'
+++ conda activate
+++ local cmd=activate
+++ case "$cmd" in
+++ __conda_activate activate
+++ '[' -n '' ']'
+++ local ask_conda
++++ PS1=
++++ __conda_exe shell.posix activate
++++ /fsx/nouamane/miniconda/bin/conda shell.posix activate
+Error while loading conda entry point: conda-libmamba-solver (module 'libmambapy' has no attribute 'QueryFormat')
+++ ask_conda='. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_2='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/etc/conda/activate.d/libxml2_activate.sh"'
+++ eval '. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_2='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/etc/conda/activate.d/libxml2_activate.sh"'
++++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh
+++++ test -n ''
+++++ unset XML_CATALOG_FILES
+++++ unset xml_catalog_files_libxml2
++++ PS1='(base) '
++++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++++ export CONDA_PREFIX=/fsx/nouamane/miniconda
++++ CONDA_PREFIX=/fsx/nouamane/miniconda
++++ export CONDA_SHLVL=3
++++ CONDA_SHLVL=3
++++ export CONDA_DEFAULT_ENV=base
++++ CONDA_DEFAULT_ENV=base
++++ export 'CONDA_PROMPT_MODIFIER=(base) '
++++ CONDA_PROMPT_MODIFIER='(base) '
++++ export CONDA_PREFIX_2=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ CONDA_PREFIX_2=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ . /fsx/nouamane/miniconda/etc/conda/activate.d/libxml2_activate.sh
+++++ test -n ''
+++++ xml_catalog_files_libxml2=
+++++ XML_CATALOG_FILES=
+++++ conda_catalog_files=
+++++ ifs_libxml2=' 	
+'
+++++ IFS=' '
+++++ rem=/fsx/nouamane/miniconda
+++++ for pre in ${rem}
+++++ test '' = /fsx/nouamane/miniconda
+++++ conda_catalog_files=/fsx/nouamane/miniconda
+++++ rem=
+++++ IFS=' 	
+'
+++++ conda_catalog_files='file:///fsx/nouamane/miniconda/etc/xml/catalog file:///etc/xml/catalog'
+++++ export 'XML_CATALOG_FILES=file:///fsx/nouamane/miniconda/etc/xml/catalog file:///etc/xml/catalog'
+++++ XML_CATALOG_FILES='file:///fsx/nouamane/miniconda/etc/xml/catalog file:///etc/xml/catalog'
+++++ unset conda_catalog_files ifs_libxml2 rem
+++ __conda_hashr
+++ '[' -n '' ']'
+++ '[' -n '' ']'
+++ hash -r
++ conda activate 2-1-cu121
++ local cmd=activate
++ case "$cmd" in
++ __conda_activate activate 2-1-cu121
++ '[' -n '' ']'
++ local ask_conda
+++ PS1='(base) '
+++ __conda_exe shell.posix activate 2-1-cu121
+++ /fsx/nouamane/miniconda/bin/conda shell.posix activate 2-1-cu121
+Error while loading conda entry point: conda-libmamba-solver (module 'libmambapy' has no attribute 'QueryFormat')
++ ask_conda='. "/fsx/nouamane/miniconda/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''4'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_3='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
++ eval '. "/fsx/nouamane/miniconda/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''4'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_3='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
+++ . /fsx/nouamane/miniconda/etc/conda/deactivate.d/libxml2_deactivate.sh
++++ test -n ''
++++ unset XML_CATALOG_FILES
++++ unset xml_catalog_files_libxml2
+++ PS1='(2-1-cu121) '
+++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
+++ export CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ export CONDA_SHLVL=4
+++ CONDA_SHLVL=4
+++ export CONDA_DEFAULT_ENV=2-1-cu121
+++ CONDA_DEFAULT_ENV=2-1-cu121
+++ export 'CONDA_PROMPT_MODIFIER=(2-1-cu121) '
+++ CONDA_PROMPT_MODIFIER='(2-1-cu121) '
+++ export CONDA_PREFIX_3=/fsx/nouamane/miniconda
+++ CONDA_PREFIX_3=/fsx/nouamane/miniconda
+++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ export _CE_M=
+++ _CE_M=
+++ export _CE_CONDA=
+++ _CE_CONDA=
+++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh
++++ test -n ''
++++ xml_catalog_files_libxml2=
++++ XML_CATALOG_FILES=
++++ conda_catalog_files=
++++ ifs_libxml2=' 	
+'
++++ IFS=' '
++++ rem=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ for pre in ${rem}
++++ test '' = /fsx/nouamane/miniconda/envs/2-1-cu121
++++ conda_catalog_files=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ rem=
++++ IFS=' 	
+'
++++ conda_catalog_files='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ export 'XML_CATALOG_FILES=file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ XML_CATALOG_FILES='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ unset conda_catalog_files ifs_libxml2 rem
++ __conda_hashr
++ '[' -n '' ']'
++ '[' -n '' ']'
++ hash -r
++ export PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++ PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/admin/home/nouamane/.local/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.fzf/bin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin
++ '[' -z '' ']'
+++ scontrol show hostnames 'ip-26-0-162-[14,46,79,180]'
++ export 'NODELIST=ip-26-0-162-14
+ip-26-0-162-46
+ip-26-0-162-79
+ip-26-0-162-180'
++ NODELIST='ip-26-0-162-14
+ip-26-0-162-46
+ip-26-0-162-79
+ip-26-0-162-180'
+++ head -n1
+++ scontrol show hostnames 'ip-26-0-162-[14,46,79,180]'
++ export MASTER_NODE=ip-26-0-162-14
++ MASTER_NODE=ip-26-0-162-14
++ export MASTER_PORT=12356
++ MASTER_PORT=12356
++ export NNODES=4
++ NNODES=4
++ export GPUS_PER_NODE=8
++ GPUS_PER_NODE=8
++ export WORLD_SIZE=32
++ WORLD_SIZE=32
++ export CUDA_DEVICE_MAX_CONNECTIONS=1
++ CUDA_DEVICE_MAX_CONNECTIONS=1
++ export NCCL_DEBUG=WARN
++ NCCL_DEBUG=WARN
++ export NANOTRON_BENCHMARK=1
++ NANOTRON_BENCHMARK=1
++ export WANDB_MODE=disabled
++ WANDB_MODE=disabled
++ export TORCH_NCCL_ASYNC_ERROR_HANDLING=1
++ TORCH_NCCL_ASYNC_ERROR_HANDLING=1
++ '[' -z '' ']'
++ echo '=== GPU Topology ==='
+=== GPU Topology ===
++ nvidia-smi topo -m
+	GPU0	GPU1	GPU2	GPU3	GPU4	GPU5	GPU6	GPU7	CPU Affinity	NUMA Affinity	GPU NUMA ID
+GPU0	 X 	NV18	NV18	NV18	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU1	NV18	 X 	NV18	NV18	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU2	NV18	NV18	 X 	NV18	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU3	NV18	NV18	NV18	 X 	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU4	NV18	NV18	NV18	NV18	 X 	NV18	NV18	NV18	48-95	1		N/A
+GPU5	NV18	NV18	NV18	NV18	NV18	 X 	NV18	NV18	48-95	1		N/A
+GPU6	NV18	NV18	NV18	NV18	NV18	NV18	 X 	NV18	48-95	1		N/A
+GPU7	NV18	NV18	NV18	NV18	NV18	NV18	NV18	 X 	48-95	1		N/A
+
+Legend:
+
+  X    = Self
+  SYS  = Connection traversing PCIe as well as the SMP interconnect between NUMA nodes (e.g., QPI/UPI)
+  NODE = Connection traversing PCIe as well as the interconnect between PCIe Host Bridges within a NUMA node
+  PHB  = Connection traversing PCIe as well as a PCIe Host Bridge (typically the CPU)
+  PXB  = Connection traversing multiple PCIe bridges (without traversing the PCIe Host Bridge)
+  PIX  = Connection traversing at most a single PCIe bridge
+  NV#  = Connection traversing a bonded set of # NVLinks
++ echo ==================
+==================
++ export SRUN_ALLOC_ARGS=
++ SRUN_ALLOC_ARGS=
++ echo 'Master node: ip-26-0-162-14'
+Master node: ip-26-0-162-14
++ echo 'All nodes: ip-26-0-162-14
+ip-26-0-162-46
+ip-26-0-162-79
+ip-26-0-162-180'
+All nodes: ip-26-0-162-14
+ip-26-0-162-46
+ip-26-0-162-79
+ip-26-0-162-180
++ echo 'World size: 32'
+World size: 32
++ '[' -n '' ']'
++ srun --wait=0 --kill-on-bad-exit=1 torchrun --nnodes=4 --nproc_per_node=8 --rdzv_id=13505303 --rdzv_backend=c10d --rdzv_endpoint=ip-26-0-162-14:12356 --max_restarts 0 --rdzv_conf timeout=60 /fsx/nouamane/projects/nanotron/run_train.py --config-file benchmark/configs/config_3.57G_dp1_tp32_pp1_acc256_mbs1_seq4096_zero0_tpmodeRED_vocab131k.yaml
+[2024-12-19 11:09:47,844] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-19 11:09:47,844] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-19 11:09:47,844] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-19 11:09:47,845] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-19 11:09:47,845] torch.distributed.run: [WARNING] 
+[2024-12-19 11:09:47,845] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 11:09:47,845] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-19 11:09:47,845] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 11:09:47,845] torch.distributed.run: [WARNING] 
+[2024-12-19 11:09:47,845] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 11:09:47,845] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-19 11:09:47,845] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 11:09:47,845] torch.distributed.run: [WARNING] 
+[2024-12-19 11:09:47,845] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 11:09:47,845] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-19 11:09:47,845] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 11:09:47,844] torch.distributed.run: [WARNING] 
+[2024-12-19 11:09:47,844] torch.distributed.run: [WARNING] *****************************************
+[2024-12-19 11:09:47,844] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-19 11:09:47,844] torch.distributed.run: [WARNING] *****************************************
+NCCL version 2.18.5+cuda12.2
+12/19/2024 11:10:12 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]: Measuring inter-GPU and intra-node bandwidth...
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+12/19/2024 11:10:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]: Bandwidth measurement complete. Time taken: 17.06 seconds
+12/19/2024 11:10:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]: Config:
+12/19/2024 11:10:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]: Config(general=GeneralArgs(project='debug',
+12/19/2024 11:10:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                            run='3.57G_dp1_tp32_pp1_acc256_mbs1_seq4096_zero0_tpmodeRED_vocab131k',
+12/19/2024 11:10:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                            seed=42,
+12/19/2024 11:10:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                            step=None,
+12/19/2024 11:10:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                            consumed_train_samples=None,
+12/19/2024 11:10:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                            benchmark_csv_path=PosixPath('benchmark/results/bench_final2.csv'),
+12/19/2024 11:10:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                            ignore_sanity_checks=True),
+12/19/2024 11:10:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:        parallelism=ParallelismArgs(dp=1,
+12/19/2024 11:10:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                                    pp=1,
+12/19/2024 11:10:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                                    tp=32,
+12/19/2024 11:10:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                                    pp_engine=<nanotron.parallel.pipeline_parallel.engine.OneForwardOneBackwardPipelineEngine object at 0x7f3318778dc0>,
+12/19/2024 11:10:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                                    tp_mode=<TensorParallelLinearMode.REDUCE_SCATTER: 2>,
+12/19/2024 11:10:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                                    tp_linear_async_communication=True,
+12/19/2024 11:10:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                                    recompute_layer=False,
+12/19/2024 11:10:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                                    tp_recompute_allgather=True,
+12/19/2024 11:10:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                                    expert_parallel_size=1),
+12/19/2024 11:10:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:        model=ModelArgs(model_config=LlamaConfig(bos_token_id=0,
+12/19/2024 11:10:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                                                 eos_token_id=0,
+12/19/2024 11:10:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                                                 hidden_act='silu',
+12/19/2024 11:10:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                                                 hidden_size=3072,
+12/19/2024 11:10:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                                                 initializer_range=0.02,
+12/19/2024 11:10:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                                                 intermediate_size=8192,
+12/19/2024 11:10:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                                                 is_llama_config=True,
+12/19/2024 11:10:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                                                 max_position_embeddings=4096,
+12/19/2024 11:10:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                                                 num_attention_heads=32,
+12/19/2024 11:10:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                                                 num_hidden_layers=28,
+12/19/2024 11:10:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                                                 num_key_value_heads=32,
+12/19/2024 11:10:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                                                 pad_token_id=None,
+12/19/2024 11:10:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                                                 pretraining_tp=1,
+12/19/2024 11:10:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                                                 rms_norm_eps=1e-05,
+12/19/2024 11:10:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                                                 rope_scaling=None,
+12/19/2024 11:10:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                                                 rope_theta=10000.0,
+12/19/2024 11:10:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                                                 rope_interleaved=False,
+12/19/2024 11:10:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                                                 tie_word_embeddings=True,
+12/19/2024 11:10:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                                                 use_cache=True,
+12/19/2024 11:10:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                                                 vocab_size=131072),
+12/19/2024 11:10:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                        init_method=RandomInit(std=0.02),
+12/19/2024 11:10:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                        dtype=torch.bfloat16,
+12/19/2024 11:10:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                        make_vocab_size_divisible_by=1,
+12/19/2024 11:10:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                        ddp_bucket_cap_mb=25),
+12/19/2024 11:10:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:        tokenizer=TokenizerArgs(tokenizer_name_or_path='robot-test/dummy-tokenizer-wordlevel',
+12/19/2024 11:10:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                                tokenizer_revision=None,
+12/19/2024 11:10:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                                tokenizer_max_length=None),
+12/19/2024 11:10:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:        checkpoints=CheckpointsArgs(checkpoints_path=PosixPath('checkpoints'),
+12/19/2024 11:10:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                                    checkpoint_interval=10000,
+12/19/2024 11:10:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                                    save_initial_state=False,
+12/19/2024 11:10:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                                    save_final_state=False,
+12/19/2024 11:10:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                                    resume_checkpoint_path=None,
+12/19/2024 11:10:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                                    checkpoints_path_is_shared_file_system=False),
+12/19/2024 11:10:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:        logging=LoggingArgs(log_level='info',
+12/19/2024 11:10:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                            log_level_replica='info',
+12/19/2024 11:10:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                            iteration_step_info_interval=1),
+12/19/2024 11:10:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:        tokens=TokensArgs(sequence_length=4096,
+12/19/2024 11:10:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                          train_steps=100,
+12/19/2024 11:10:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                          micro_batch_size=1,
+12/19/2024 11:10:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                          batch_accumulation_per_replica=256,
+12/19/2024 11:10:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                          val_check_interval=100,
+12/19/2024 11:10:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                          limit_val_batches=0,
+12/19/2024 11:10:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                          limit_test_batches=0),
+12/19/2024 11:10:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:        optimizer=OptimizerArgs(optimizer_factory=AdamWOptimizerArgs(adam_eps=1e-08,
+12/19/2024 11:10:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                                                                     adam_beta1=0.9,
+12/19/2024 11:10:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                                                                     adam_beta2=0.95,
+12/19/2024 11:10:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                                                                     torch_adam_is_fused=True,
+12/19/2024 11:10:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                                                                     name='adamW'),
+12/19/2024 11:10:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                                zero_stage=0,
+12/19/2024 11:10:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                                weight_decay=0.01,
+12/19/2024 11:10:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                                clip_grad=1.0,
+12/19/2024 11:10:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                                accumulate_grad_in_fp32=True,
+12/19/2024 11:10:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                                learning_rate_scheduler=LRSchedulerArgs(learning_rate=0.0003,
+12/19/2024 11:10:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                                                                        lr_warmup_steps=2,
+12/19/2024 11:10:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                                                                        lr_warmup_style='linear',
+12/19/2024 11:10:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                                                                        lr_decay_style='cosine',
+12/19/2024 11:10:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                                                                        lr_decay_steps=13,
+12/19/2024 11:10:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                                                                        lr_decay_starting_step=None,
+12/19/2024 11:10:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                                                                        min_decay_lr=1e-05)),
+12/19/2024 11:10:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:        data_stages=[DatasetStageArgs(name='Stable Training Stage',
+12/19/2024 11:10:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                                      start_training_step=1,
+12/19/2024 11:10:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                                      data=DataArgs(dataset=None,
+12/19/2024 11:10:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                                                    seed=42,
+12/19/2024 11:10:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                                                    num_loading_workers=1))],
+12/19/2024 11:10:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:        profiler=None,
+12/19/2024 11:10:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:        lighteval=None,
+12/19/2024 11:10:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:        s3_upload=None)
+12/19/2024 11:10:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]: Model Config:
+12/19/2024 11:10:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]: LlamaConfig(bos_token_id=0,
+12/19/2024 11:10:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:             eos_token_id=0,
+12/19/2024 11:10:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:             hidden_act='silu',
+12/19/2024 11:10:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:             hidden_size=3072,
+12/19/2024 11:10:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:             initializer_range=0.02,
+12/19/2024 11:10:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:             intermediate_size=8192,
+12/19/2024 11:10:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:             is_llama_config=True,
+12/19/2024 11:10:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:             max_position_embeddings=4096,
+12/19/2024 11:10:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:             num_attention_heads=32,
+12/19/2024 11:10:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:             num_hidden_layers=28,
+12/19/2024 11:10:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:             num_key_value_heads=32,
+12/19/2024 11:10:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:             pad_token_id=None,
+12/19/2024 11:10:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:             pretraining_tp=1,
+12/19/2024 11:10:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:             rms_norm_eps=1e-05,
+12/19/2024 11:10:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:             rope_scaling=None,
+12/19/2024 11:10:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:             rope_theta=10000.0,
+12/19/2024 11:10:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:             rope_interleaved=False,
+12/19/2024 11:10:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:             tie_word_embeddings=True,
+12/19/2024 11:10:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:             use_cache=True,
+12/19/2024 11:10:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:             vocab_size=131072)
+12/19/2024 11:10:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]: Building model..
+12/19/2024 11:10:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]: Initialize RoPE Theta = 10000.0
+12/19/2024 11:10:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]: Setting PP block ranks...
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+12/19/2024 11:10:38 [INFO|DP=0|PP=0|TP=5|ip-26-0-162-14]: Local number of parameters: 112M (213.33MiB)
+12/19/2024 11:10:38 [INFO|DP=0|PP=0|TP=19|ip-26-0-162-46]: Local number of parameters: 112M (213.33MiB)
+12/19/2024 11:10:38 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]: Total number of parameters: 3.58G (6826.69MiB)
+12/19/2024 11:10:38 [INFO|DP=0|PP=0|TP=6|ip-26-0-162-14]: Local number of parameters: 112M (213.33MiB)
+12/19/2024 11:10:38 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]: Local number of parameters: 112M (213.33MiB)
+12/19/2024 11:10:38 [INFO|DP=0|PP=0|TP=23|ip-26-0-162-46]: Local number of parameters: 112M (213.33MiB)
+12/19/2024 11:10:38 [INFO|DP=0|PP=0|TP=21|ip-26-0-162-46]: Local number of parameters: 112M (213.33MiB)
+12/19/2024 11:10:38 [INFO|DP=0|PP=0|TP=1|ip-26-0-162-14]: Local number of parameters: 112M (213.33MiB)
+12/19/2024 11:10:38 [INFO|DP=0|PP=0|TP=22|ip-26-0-162-46]: Local number of parameters: 112M (213.33MiB)
+12/19/2024 11:10:38 [INFO|DP=0|PP=0|TP=16|ip-26-0-162-46]: Local number of parameters: 112M (213.33MiB)
+12/19/2024 11:10:38 [INFO|DP=0|PP=0|TP=20|ip-26-0-162-46]: Local number of parameters: 112M (213.33MiB)
+12/19/2024 11:10:38 [INFO|DP=0|PP=0|TP=3|ip-26-0-162-14]: Local number of parameters: 112M (213.33MiB)
+12/19/2024 11:10:38 [INFO|DP=0|PP=0|TP=17|ip-26-0-162-46]: Local number of parameters: 112M (213.33MiB)
+12/19/2024 11:10:38 [INFO|DP=0|PP=0|TP=7|ip-26-0-162-14]: Local number of parameters: 112M (213.33MiB)
+12/19/2024 11:10:38 [INFO|DP=0|PP=0|TP=5|ip-26-0-162-14]: [After model building] Memory usage: 213.50MiB. Peak allocated: 5440.00MiB Peak reserved: 22868.00MiB
+12/19/2024 11:10:38 [INFO|DP=0|PP=0|TP=19|ip-26-0-162-46]: [After model building] Memory usage: 213.50MiB. Peak allocated: 5440.00MiB Peak reserved: 22868.00MiB
+12/19/2024 11:10:38 [INFO|DP=0|PP=0|TP=2|ip-26-0-162-14]: Local number of parameters: 112M (213.33MiB)
+12/19/2024 11:10:38 [INFO|DP=0|PP=0|TP=6|ip-26-0-162-14]: [After model building] Memory usage: 213.50MiB. Peak allocated: 5440.00MiB Peak reserved: 22868.00MiB
+12/19/2024 11:10:38 [INFO|DP=0|PP=0|TP=22|ip-26-0-162-46]: [After model building] Memory usage: 213.50MiB. Peak allocated: 5440.00MiB Peak reserved: 22868.00MiB
+12/19/2024 11:10:38 [INFO|DP=0|PP=0|TP=23|ip-26-0-162-46]: [After model building] Memory usage: 213.50MiB. Peak allocated: 5440.00MiB Peak reserved: 22868.00MiB
+12/19/2024 11:10:38 [INFO|DP=0|PP=0|TP=16|ip-26-0-162-46]: [After model building] Memory usage: 213.50MiB. Peak allocated: 5440.00MiB Peak reserved: 22868.00MiB
+12/19/2024 11:10:38 [INFO|DP=0|PP=0|TP=18|ip-26-0-162-46]: Local number of parameters: 112M (213.33MiB)
+12/19/2024 11:10:38 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]: [After model building] Memory usage: 213.50MiB. Peak allocated: 5440.00MiB Peak reserved: 22868.00MiB
+12/19/2024 11:10:38 [INFO|DP=0|PP=0|TP=20|ip-26-0-162-46]: [After model building] Memory usage: 213.50MiB. Peak allocated: 5440.00MiB Peak reserved: 22868.00MiB
+12/19/2024 11:10:38 [INFO|DP=0|PP=0|TP=21|ip-26-0-162-46]: [After model building] Memory usage: 213.50MiB. Peak allocated: 5440.00MiB Peak reserved: 22868.00MiB
+12/19/2024 11:10:38 [INFO|DP=0|PP=0|TP=3|ip-26-0-162-14]: [After model building] Memory usage: 213.50MiB. Peak allocated: 5440.00MiB Peak reserved: 22868.00MiB
+12/19/2024 11:10:38 [INFO|DP=0|PP=0|TP=1|ip-26-0-162-14]: [After model building] Memory usage: 213.50MiB. Peak allocated: 5440.00MiB Peak reserved: 22868.00MiB
+12/19/2024 11:10:38 [INFO|DP=0|PP=0|TP=17|ip-26-0-162-46]: [After model building] Memory usage: 213.50MiB. Peak allocated: 5440.00MiB Peak reserved: 22868.00MiB
+12/19/2024 11:10:38 [INFO|DP=0|PP=0|TP=8|ip-26-0-162-180]: Local number of parameters: 112M (213.33MiB)
+12/19/2024 11:10:38 [INFO|DP=0|PP=0|TP=26|ip-26-0-162-79]: Local number of parameters: 112M (213.33MiB)
+12/19/2024 11:10:38 [INFO|DP=0|PP=0|TP=31|ip-26-0-162-79]: Local number of parameters: 112M (213.33MiB)
+12/19/2024 11:10:38 [INFO|DP=0|PP=0|TP=29|ip-26-0-162-79]: Local number of parameters: 112M (213.33MiB)
+12/19/2024 11:10:38 [INFO|DP=0|PP=0|TP=30|ip-26-0-162-79]: Local number of parameters: 112M (213.33MiB)
+12/19/2024 11:10:38 [INFO|DP=0|PP=0|TP=24|ip-26-0-162-79]: Local number of parameters: 112M (213.33MiB)
+12/19/2024 11:10:38 [INFO|DP=0|PP=0|TP=4|ip-26-0-162-14]: Local number of parameters: 112M (213.33MiB)
+12/19/2024 11:10:38 [INFO|DP=0|PP=0|TP=9|ip-26-0-162-180]: Local number of parameters: 112M (213.33MiB)
+12/19/2024 11:10:38 [INFO|DP=0|PP=0|TP=10|ip-26-0-162-180]: Local number of parameters: 112M (213.33MiB)
+12/19/2024 11:10:38 [INFO|DP=0|PP=0|TP=11|ip-26-0-162-180]: Local number of parameters: 112M (213.33MiB)
+12/19/2024 11:10:38 [INFO|DP=0|PP=0|TP=12|ip-26-0-162-180]: Local number of parameters: 112M (213.33MiB)
+12/19/2024 11:10:38 [INFO|DP=0|PP=0|TP=15|ip-26-0-162-180]: Local number of parameters: 112M (213.33MiB)
+12/19/2024 11:10:38 [INFO|DP=0|PP=0|TP=14|ip-26-0-162-180]: Local number of parameters: 112M (213.33MiB)
+12/19/2024 11:10:38 [INFO|DP=0|PP=0|TP=28|ip-26-0-162-79]: Local number of parameters: 112M (213.33MiB)
+12/19/2024 11:10:38 [INFO|DP=0|PP=0|TP=7|ip-26-0-162-14]: [After model building] Memory usage: 213.50MiB. Peak allocated: 5440.00MiB Peak reserved: 22868.00MiB
+12/19/2024 11:10:38 [INFO|DP=0|PP=0|TP=13|ip-26-0-162-180]: Local number of parameters: 112M (213.33MiB)
+12/19/2024 11:10:38 [INFO|DP=0|PP=0|TP=25|ip-26-0-162-79]: Local number of parameters: 112M (213.33MiB)
+12/19/2024 11:10:38 [INFO|DP=0|PP=0|TP=27|ip-26-0-162-79]: Local number of parameters: 112M (213.33MiB)
+12/19/2024 11:10:38 [INFO|DP=0|PP=0|TP=8|ip-26-0-162-180]: [After model building] Memory usage: 213.50MiB. Peak allocated: 5440.00MiB Peak reserved: 22868.00MiB
+12/19/2024 11:10:38 [INFO|DP=0|PP=0|TP=2|ip-26-0-162-14]: [After model building] Memory usage: 213.50MiB. Peak allocated: 5440.00MiB Peak reserved: 22868.00MiB
+12/19/2024 11:10:38 [INFO|DP=0|PP=0|TP=29|ip-26-0-162-79]: [After model building] Memory usage: 213.50MiB. Peak allocated: 5440.00MiB Peak reserved: 19796.00MiB
+12/19/2024 11:10:38 [INFO|DP=0|PP=0|TP=31|ip-26-0-162-79]: [After model building] Memory usage: 213.50MiB. Peak allocated: 5440.00MiB Peak reserved: 20820.00MiB
+12/19/2024 11:10:38 [INFO|DP=0|PP=0|TP=18|ip-26-0-162-46]: [After model building] Memory usage: 213.50MiB. Peak allocated: 5440.00MiB Peak reserved: 22868.00MiB
+12/19/2024 11:10:38 [INFO|DP=0|PP=0|TP=11|ip-26-0-162-180]: [After model building] Memory usage: 213.50MiB. Peak allocated: 5440.00MiB Peak reserved: 22868.00MiB
+12/19/2024 11:10:38 [INFO|DP=0|PP=0|TP=10|ip-26-0-162-180]: [After model building] Memory usage: 213.50MiB. Peak allocated: 5440.00MiB Peak reserved: 22868.00MiB
+12/19/2024 11:10:38 [INFO|DP=0|PP=0|TP=30|ip-26-0-162-79]: [After model building] Memory usage: 213.50MiB. Peak allocated: 5440.00MiB Peak reserved: 19796.00MiB
+12/19/2024 11:10:38 [INFO|DP=0|PP=0|TP=26|ip-26-0-162-79]: [After model building] Memory usage: 213.50MiB. Peak allocated: 5440.00MiB Peak reserved: 20820.00MiB
+12/19/2024 11:10:38 [INFO|DP=0|PP=0|TP=24|ip-26-0-162-79]: [After model building] Memory usage: 213.50MiB. Peak allocated: 5440.00MiB Peak reserved: 21844.00MiB
+12/19/2024 11:10:38 [INFO|DP=0|PP=0|TP=9|ip-26-0-162-180]: [After model building] Memory usage: 213.50MiB. Peak allocated: 5440.00MiB Peak reserved: 22868.00MiB
+12/19/2024 11:10:38 [INFO|DP=0|PP=0|TP=14|ip-26-0-162-180]: [After model building] Memory usage: 213.50MiB. Peak allocated: 5440.00MiB Peak reserved: 22868.00MiB
+12/19/2024 11:10:38 [INFO|DP=0|PP=0|TP=15|ip-26-0-162-180]: [After model building] Memory usage: 213.50MiB. Peak allocated: 5440.00MiB Peak reserved: 22868.00MiB
+12/19/2024 11:10:38 [INFO|DP=0|PP=0|TP=28|ip-26-0-162-79]: [After model building] Memory usage: 213.50MiB. Peak allocated: 5440.00MiB Peak reserved: 20820.00MiB
+12/19/2024 11:10:38 [INFO|DP=0|PP=0|TP=12|ip-26-0-162-180]: [After model building] Memory usage: 213.50MiB. Peak allocated: 5440.00MiB Peak reserved: 22868.00MiB
+12/19/2024 11:10:38 [INFO|DP=0|PP=0|TP=13|ip-26-0-162-180]: [After model building] Memory usage: 213.50MiB. Peak allocated: 5440.00MiB Peak reserved: 22868.00MiB
+12/19/2024 11:10:38 [INFO|DP=0|PP=0|TP=27|ip-26-0-162-79]: [After model building] Memory usage: 213.50MiB. Peak allocated: 5440.00MiB Peak reserved: 20820.00MiB
+12/19/2024 11:10:38 [INFO|DP=0|PP=0|TP=25|ip-26-0-162-79]: [After model building] Memory usage: 213.50MiB. Peak allocated: 5440.00MiB Peak reserved: 19796.00MiB
+12/19/2024 11:10:38 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]: No checkpoint path provided.
+12/19/2024 11:10:38 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]: Parametrizing model parameters using StandardParametrizator
+12/19/2024 11:10:38 [INFO|DP=0|PP=0|TP=4|ip-26-0-162-14]: [After model building] Memory usage: 213.50MiB. Peak allocated: 5440.00MiB Peak reserved: 22868.00MiB
+12/19/2024 11:10:38 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]: [Optimizer Building] Using LearningRateForSP as learning rate
+12/19/2024 11:10:39 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]: [Training Plan] Stage Stable Training Stage has 99 remaining training steps and has consumed 0 samples
+12/19/2024 11:10:39 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]: Using dummy data generator
+12/19/2024 11:10:39 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]: [Training Plan] There are 1 training stages 
+12/19/2024 11:10:39 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]: [Stage Stable Training Stage] start from step 1 
+12/19/2024 11:10:39 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]: 
+12/19/2024 11:10:39 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]: [Start training] datetime: 2024-12-19 11:10:39.356514 | mbs: 1 | grad_accum: 256 | global_batch_size: 256 | sequence_length: 4096 | train_steps: 100 | start_iteration_step: 0 | consumed_train_samples: 0
+12/19/2024 11:10:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]: Resuming training from stage Stable Training Stage, it has trained for 0 samples and has 99 remaining train steps
+12/19/2024 11:10:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:  Memory usage: 1066.83MiB. Peak allocated 5440.00MiB. Peak reserved: 22868.00MiB
+12/19/2024 11:12:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:  Memory usage: 1200.03MiB. Peak allocated 1902.92MiB. Peak reserved: 3254.00MiB
+12/19/2024 11:12:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]: iteration: 1 / 100 | consumed_tokens: 1.05M | elapsed_time_per_iteration_ms: 111K | tokens_per_sec: 9.41K | tokens_per_sec_per_gpu: 294 | global_batch_size: 256 | lm_loss: 12.4 | lr: 0.00015 | model_tflops_per_gpu: 7.55 | hardware_tflops_per_gpu: 7.55 | grad_norm: 0.75 | cuda_memory_allocated: 2.16G | cuda_max_memory_reserved: 3.41G | hd_total_memory_tb: 312G | hd_used_memory_tb: 70.4G | hd_free_memory_tb: 242G
+12/19/2024 11:12:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:  Memory usage: 2056.13MiB. Peak allocated 2056.13MiB. Peak reserved: 3254.00MiB
+12/19/2024 11:14:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:  Memory usage: 2056.26MiB. Peak allocated 2639.62MiB. Peak reserved: 4092.00MiB
+12/19/2024 11:14:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]: iteration: 2 / 100 | consumed_tokens: 2.1M | elapsed_time_per_iteration_ms: 109K | tokens_per_sec: 9.61K | tokens_per_sec_per_gpu: 300 | global_batch_size: 256 | lm_loss: 12.4 | lr: 0.0003 | model_tflops_per_gpu: 7.71 | hardware_tflops_per_gpu: 7.71 | grad_norm: 0.749 | cuda_memory_allocated: 2.16G | cuda_max_memory_reserved: 4.29G | hd_total_memory_tb: 312G | hd_used_memory_tb: 70.4G | hd_free_memory_tb: 242G
+12/19/2024 11:14:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:  Memory usage: 2056.13MiB. Peak allocated 2056.34MiB. Peak reserved: 4092.00MiB
diff --git a/logs/13803345-bench_8.86G_dp4_tp2_pp2_acc4_mbs16_seq4096_zero1_tpmodeRED_vocab131k.out b/logs/13803345-bench_8.86G_dp4_tp2_pp2_acc4_mbs16_seq4096_zero1_tpmodeRED_vocab131k.out
new file mode 100644
index 0000000000000000000000000000000000000000..db328268768079fa7585135793985967791ec7b3
--- /dev/null
+++ b/logs/13803345-bench_8.86G_dp4_tp2_pp2_acc4_mbs16_seq4096_zero1_tpmodeRED_vocab131k.out
@@ -0,0 +1,5417 @@
++ '[' -z 13803345 ']'
++ unset FI_PROVIDER
++ source /fsx/nouamane/miniconda/bin/activate
+++ _CONDA_ROOT=/fsx/nouamane/miniconda
+++ . /fsx/nouamane/miniconda/etc/profile.d/conda.sh
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ '[' -z x ']'
+++ conda activate
+++ local cmd=activate
+++ case "$cmd" in
+++ __conda_activate activate
+++ '[' -n '' ']'
+++ local ask_conda
++++ PS1=
++++ __conda_exe shell.posix activate
++++ /fsx/nouamane/miniconda/bin/conda shell.posix activate
+Error while loading conda entry point: conda-libmamba-solver (module 'libmambapy' has no attribute 'QueryFormat')
+++ ask_conda='. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''2'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_1='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/etc/conda/activate.d/libxml2_activate.sh"'
+++ eval '. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''2'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_1='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/etc/conda/activate.d/libxml2_activate.sh"'
++++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh
+++++ test -n ''
+++++ unset XML_CATALOG_FILES
+++++ unset xml_catalog_files_libxml2
++++ PS1='(base) '
++++ export PATH=/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
++++ PATH=/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
++++ export CONDA_PREFIX=/fsx/nouamane/miniconda
++++ CONDA_PREFIX=/fsx/nouamane/miniconda
++++ export CONDA_SHLVL=2
++++ CONDA_SHLVL=2
++++ export CONDA_DEFAULT_ENV=base
++++ CONDA_DEFAULT_ENV=base
++++ export 'CONDA_PROMPT_MODIFIER=(base) '
++++ CONDA_PROMPT_MODIFIER='(base) '
++++ export CONDA_PREFIX_1=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ CONDA_PREFIX_1=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ . /fsx/nouamane/miniconda/etc/conda/activate.d/libxml2_activate.sh
+++++ test -n ''
+++++ xml_catalog_files_libxml2=
+++++ XML_CATALOG_FILES=
+++++ conda_catalog_files=
+++++ ifs_libxml2=' 	
+'
+++++ IFS=' '
+++++ rem=/fsx/nouamane/miniconda
+++++ for pre in ${rem}
+++++ test '' = /fsx/nouamane/miniconda
+++++ conda_catalog_files=/fsx/nouamane/miniconda
+++++ rem=
+++++ IFS=' 	
+'
+++++ conda_catalog_files='file:///fsx/nouamane/miniconda/etc/xml/catalog file:///etc/xml/catalog'
+++++ export 'XML_CATALOG_FILES=file:///fsx/nouamane/miniconda/etc/xml/catalog file:///etc/xml/catalog'
+++++ XML_CATALOG_FILES='file:///fsx/nouamane/miniconda/etc/xml/catalog file:///etc/xml/catalog'
+++++ unset conda_catalog_files ifs_libxml2 rem
+++ __conda_hashr
+++ '[' -n '' ']'
+++ '[' -n '' ']'
+++ hash -r
++ conda activate 2-1-cu121
++ local cmd=activate
++ case "$cmd" in
++ __conda_activate activate 2-1-cu121
++ '[' -n '' ']'
++ local ask_conda
+++ PS1='(base) '
+++ __conda_exe shell.posix activate 2-1-cu121
+++ /fsx/nouamane/miniconda/bin/conda shell.posix activate 2-1-cu121
+Error while loading conda entry point: conda-libmamba-solver (module 'libmambapy' has no attribute 'QueryFormat')
++ ask_conda='. "/fsx/nouamane/miniconda/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(2-1-cu121) '\''
+export PATH='\''/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_2='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
++ eval '. "/fsx/nouamane/miniconda/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(2-1-cu121) '\''
+export PATH='\''/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_2='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
+++ . /fsx/nouamane/miniconda/etc/conda/deactivate.d/libxml2_deactivate.sh
++++ test -n ''
++++ unset XML_CATALOG_FILES
++++ unset xml_catalog_files_libxml2
+++ PS1='(2-1-cu121) '
+++ export PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
+++ PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
+++ export CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ export CONDA_SHLVL=3
+++ CONDA_SHLVL=3
+++ export CONDA_DEFAULT_ENV=2-1-cu121
+++ CONDA_DEFAULT_ENV=2-1-cu121
+++ export 'CONDA_PROMPT_MODIFIER=(2-1-cu121) '
+++ CONDA_PROMPT_MODIFIER='(2-1-cu121) '
+++ export CONDA_PREFIX_2=/fsx/nouamane/miniconda
+++ CONDA_PREFIX_2=/fsx/nouamane/miniconda
+++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ export _CE_M=
+++ _CE_M=
+++ export _CE_CONDA=
+++ _CE_CONDA=
+++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh
++++ test -n ''
++++ xml_catalog_files_libxml2=
++++ XML_CATALOG_FILES=
++++ conda_catalog_files=
++++ ifs_libxml2=' 	
+'
++++ IFS=' '
++++ rem=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ for pre in ${rem}
++++ test '' = /fsx/nouamane/miniconda/envs/2-1-cu121
++++ conda_catalog_files=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ rem=
++++ IFS=' 	
+'
++++ conda_catalog_files='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ export 'XML_CATALOG_FILES=file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ XML_CATALOG_FILES='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ unset conda_catalog_files ifs_libxml2 rem
++ __conda_hashr
++ '[' -n '' ']'
++ '[' -n '' ']'
++ hash -r
++ export PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
++ PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
++ '[' -z '' ']'
+++ scontrol show hostnames 'ip-26-0-166-[36,125]'
++ export 'NODELIST=ip-26-0-166-36
+ip-26-0-166-125'
++ NODELIST='ip-26-0-166-36
+ip-26-0-166-125'
+++ head -n1
+++ scontrol show hostnames 'ip-26-0-166-[36,125]'
++ export MASTER_NODE=ip-26-0-166-36
++ MASTER_NODE=ip-26-0-166-36
++ export MASTER_PORT=12356
++ MASTER_PORT=12356
++ export NNODES=2
++ NNODES=2
++ export GPUS_PER_NODE=8
++ GPUS_PER_NODE=8
++ export WORLD_SIZE=16
++ WORLD_SIZE=16
++ export CUDA_DEVICE_MAX_CONNECTIONS=1
++ CUDA_DEVICE_MAX_CONNECTIONS=1
++ export NCCL_DEBUG=INFO
++ NCCL_DEBUG=INFO
++ export NANOTRON_BENCHMARK=1
++ NANOTRON_BENCHMARK=1
++ export WANDB_MODE=disabled
++ WANDB_MODE=disabled
++ export TORCH_NCCL_ASYNC_ERROR_HANDLING=1
++ TORCH_NCCL_ASYNC_ERROR_HANDLING=1
++ export TORCH_DISTRIBUTED_DEBUG=DETAIL
++ TORCH_DISTRIBUTED_DEBUG=DETAIL
++ export NCCL_P2P_LEVEL=LOC
++ NCCL_P2P_LEVEL=LOC
++ '[' -z '' ']'
++ echo '=== GPU Topology ==='
+=== GPU Topology ===
++ nvidia-smi topo -m
+	GPU0	GPU1	GPU2	GPU3	GPU4	GPU5	GPU6	GPU7	CPU Affinity	NUMA Affinity	GPU NUMA ID
+GPU0	 X 	NV18	NV18	NV18	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU1	NV18	 X 	NV18	NV18	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU2	NV18	NV18	 X 	NV18	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU3	NV18	NV18	NV18	 X 	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU4	NV18	NV18	NV18	NV18	 X 	NV18	NV18	NV18	48-95	1		N/A
+GPU5	NV18	NV18	NV18	NV18	NV18	 X 	NV18	NV18	48-95	1		N/A
+GPU6	NV18	NV18	NV18	NV18	NV18	NV18	 X 	NV18	48-95	1		N/A
+GPU7	NV18	NV18	NV18	NV18	NV18	NV18	NV18	 X 	48-95	1		N/A
+
+Legend:
+
+  X    = Self
+  SYS  = Connection traversing PCIe as well as the SMP interconnect between NUMA nodes (e.g., QPI/UPI)
+  NODE = Connection traversing PCIe as well as the interconnect between PCIe Host Bridges within a NUMA node
+  PHB  = Connection traversing PCIe as well as a PCIe Host Bridge (typically the CPU)
+  PXB  = Connection traversing multiple PCIe bridges (without traversing the PCIe Host Bridge)
+  PIX  = Connection traversing at most a single PCIe bridge
+  NV#  = Connection traversing a bonded set of # NVLinks
++ echo ==================
+==================
++ export SRUN_ALLOC_ARGS=
++ SRUN_ALLOC_ARGS=
++ echo 'Master node: ip-26-0-166-36'
+Master node: ip-26-0-166-36
++ echo 'All nodes: ip-26-0-166-36
+ip-26-0-166-125'
+All nodes: ip-26-0-166-36
+ip-26-0-166-125
++ echo 'World size: 16'
+World size: 16
++ '[' -n '' ']'
++ srun --wait=0 --kill-on-bad-exit=1 torchrun --nnodes=2 --nproc_per_node=8 --rdzv_id=13803345 --rdzv_backend=c10d --rdzv_endpoint=ip-26-0-166-36:12356 --max_restarts 0 --rdzv_conf timeout=60 /fsx/nouamane/projects/nanotron/run_train.py --config-file benchmark/configs/config_8.86G_dp4_tp2_pp2_acc4_mbs16_seq4096_zero1_tpmodeRED_vocab131k.yaml
+[2024-12-27 17:31:34,258] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-27 17:31:34,264] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-27 17:31:34,258] torch.distributed.run: [WARNING] 
+[2024-12-27 17:31:34,258] torch.distributed.run: [WARNING] *****************************************
+[2024-12-27 17:31:34,258] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-27 17:31:34,258] torch.distributed.run: [WARNING] *****************************************
+[2024-12-27 17:31:34,264] torch.distributed.run: [WARNING] 
+[2024-12-27 17:31:34,264] torch.distributed.run: [WARNING] *****************************************
+[2024-12-27 17:31:34,264] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-27 17:31:34,264] torch.distributed.run: [WARNING] *****************************************
+ip-26-0-166-125:1332482:1332482 [0] NCCL INFO Bootstrap : Using enp74s0:26.0.172.86<0>
+ip-26-0-166-125:1332482:1332482 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-166-125:1332482:1332482 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-166-125:1332482:1332482 [0] NCCL INFO cudaDriverVersion 12020
+NCCL version 2.18.5+cuda12.2
+ip-26-0-166-125:1332486:1332486 [4] NCCL INFO cudaDriverVersion 12020
+ip-26-0-166-125:1332483:1332483 [1] NCCL INFO cudaDriverVersion 12020
+ip-26-0-166-125:1332488:1332488 [6] NCCL INFO cudaDriverVersion 12020
+ip-26-0-166-36:1061620:1061620 [0] NCCL INFO cudaDriverVersion 12020
+ip-26-0-166-125:1332484:1332484 [2] NCCL INFO cudaDriverVersion 12020
+ip-26-0-166-125:1332489:1332489 [7] NCCL INFO cudaDriverVersion 12020
+ip-26-0-166-125:1332487:1332487 [5] NCCL INFO cudaDriverVersion 12020
+ip-26-0-166-36:1061625:1061625 [5] NCCL INFO cudaDriverVersion 12020
+ip-26-0-166-125:1332485:1332485 [3] NCCL INFO cudaDriverVersion 12020
+ip-26-0-166-36:1061620:1061620 [0] NCCL INFO Bootstrap : Using enp71s0:26.0.166.36<0>
+ip-26-0-166-125:1332483:1332483 [1] NCCL INFO Bootstrap : Using enp74s0:26.0.172.86<0>
+ip-26-0-166-125:1332486:1332486 [4] NCCL INFO Bootstrap : Using enp74s0:26.0.172.86<0>
+ip-26-0-166-36:1061625:1061625 [5] NCCL INFO Bootstrap : Using enp71s0:26.0.166.36<0>
+ip-26-0-166-36:1061620:1061620 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-166-36:1061620:1061620 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-166-125:1332483:1332483 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-166-125:1332483:1332483 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-166-36:1061625:1061625 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-166-36:1061625:1061625 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-166-125:1332486:1332486 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-166-125:1332486:1332486 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-166-125:1332488:1332488 [6] NCCL INFO Bootstrap : Using enp74s0:26.0.172.86<0>
+ip-26-0-166-125:1332488:1332488 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-166-125:1332488:1332488 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-166-36:1061627:1061627 [7] NCCL INFO cudaDriverVersion 12020
+ip-26-0-166-36:1061622:1061622 [2] NCCL INFO cudaDriverVersion 12020
+ip-26-0-166-36:1061623:1061623 [3] NCCL INFO cudaDriverVersion 12020
+ip-26-0-166-36:1061621:1061621 [1] NCCL INFO cudaDriverVersion 12020
+ip-26-0-166-36:1061624:1061624 [4] NCCL INFO cudaDriverVersion 12020
+ip-26-0-166-125:1332484:1332484 [2] NCCL INFO Bootstrap : Using enp74s0:26.0.172.86<0>
+ip-26-0-166-125:1332489:1332489 [7] NCCL INFO Bootstrap : Using enp74s0:26.0.172.86<0>
+ip-26-0-166-36:1061626:1061626 [6] NCCL INFO cudaDriverVersion 12020
+ip-26-0-166-125:1332487:1332487 [5] NCCL INFO Bootstrap : Using enp74s0:26.0.172.86<0>
+ip-26-0-166-125:1332485:1332485 [3] NCCL INFO Bootstrap : Using enp74s0:26.0.172.86<0>
+ip-26-0-166-125:1332484:1332484 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-166-125:1332484:1332484 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-166-125:1332489:1332489 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-166-125:1332489:1332489 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-166-125:1332487:1332487 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-166-125:1332487:1332487 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-166-125:1332485:1332485 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-166-125:1332485:1332485 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-166-36:1061622:1061622 [2] NCCL INFO Bootstrap : Using enp71s0:26.0.166.36<0>
+ip-26-0-166-36:1061623:1061623 [3] NCCL INFO Bootstrap : Using enp71s0:26.0.166.36<0>
+ip-26-0-166-36:1061627:1061627 [7] NCCL INFO Bootstrap : Using enp71s0:26.0.166.36<0>
+ip-26-0-166-36:1061626:1061626 [6] NCCL INFO Bootstrap : Using enp71s0:26.0.166.36<0>
+ip-26-0-166-36:1061624:1061624 [4] NCCL INFO Bootstrap : Using enp71s0:26.0.166.36<0>
+ip-26-0-166-36:1061621:1061621 [1] NCCL INFO Bootstrap : Using enp71s0:26.0.166.36<0>
+ip-26-0-166-36:1061622:1061622 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-166-36:1061622:1061622 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-166-36:1061627:1061627 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-166-36:1061623:1061623 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-166-36:1061627:1061627 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-166-36:1061623:1061623 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-166-36:1061626:1061626 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-166-36:1061626:1061626 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-166-36:1061621:1061621 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-166-36:1061624:1061624 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-166-36:1061621:1061621 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-166-36:1061624:1061624 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-166-125:1332482:1332646 [0] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-166-125:1332482:1332646 [0] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-166-125:1332482:1332646 [0] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-166-125:1332482:1332646 [0] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-166-125:1332482:1332646 [0] NCCL INFO NET/OFI Setting provider_filter to efa
+ip-26-0-166-125:1332482:1332646 [0] NCCL INFO NET/OFI Setting FI_EFA_FORK_SAFE environment variable to 1
+ip-26-0-166-125:1332482:1332646 [0] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-166-125:1332482:1332646 [0] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-166-125:1332482:1332646 [0] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-166-125:1332482:1332646 [0] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-166-125:1332482:1332646 [0] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-166-36:1061625:1061780 [5] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-166-36:1061625:1061780 [5] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-166-36:1061625:1061780 [5] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-166-36:1061625:1061780 [5] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-166-36:1061625:1061780 [5] NCCL INFO NET/OFI Setting provider_filter to efa
+ip-26-0-166-36:1061625:1061780 [5] NCCL INFO NET/OFI Setting FI_EFA_FORK_SAFE environment variable to 1
+ip-26-0-166-36:1061625:1061780 [5] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-166-36:1061625:1061780 [5] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-166-36:1061625:1061780 [5] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-166-36:1061625:1061780 [5] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-166-36:1061625:1061780 [5] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-166-36:1061620:1061779 [0] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-166-36:1061620:1061779 [0] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-166-36:1061620:1061779 [0] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-166-36:1061620:1061779 [0] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-166-36:1061620:1061779 [0] NCCL INFO NET/OFI Setting provider_filter to efa
+ip-26-0-166-36:1061620:1061779 [0] NCCL INFO NET/OFI Setting FI_EFA_FORK_SAFE environment variable to 1
+ip-26-0-166-36:1061620:1061779 [0] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-166-36:1061620:1061779 [0] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-166-36:1061620:1061779 [0] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-166-36:1061620:1061779 [0] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-166-36:1061620:1061779 [0] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-166-125:1332482:1332646 [0] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-166-125:1332484:1332647 [2] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-166-125:1332484:1332647 [2] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-166-36:1061627:1061782 [7] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-166-36:1061627:1061782 [7] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-166-125:1332484:1332647 [2] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-166-125:1332484:1332647 [2] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-166-125:1332484:1332647 [2] NCCL INFO NET/OFI Setting provider_filter to efa
+ip-26-0-166-125:1332484:1332647 [2] NCCL INFO NET/OFI Setting FI_EFA_FORK_SAFE environment variable to 1
+ip-26-0-166-125:1332484:1332647 [2] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-166-125:1332484:1332647 [2] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-166-125:1332484:1332647 [2] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-166-125:1332484:1332647 [2] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-166-125:1332484:1332647 [2] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-166-125:1332488:1332648 [6] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-166-125:1332488:1332648 [6] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-166-36:1061627:1061782 [7] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-166-36:1061627:1061782 [7] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-166-36:1061627:1061782 [7] NCCL INFO NET/OFI Setting provider_filter to efa
+ip-26-0-166-36:1061627:1061782 [7] NCCL INFO NET/OFI Setting FI_EFA_FORK_SAFE environment variable to 1
+ip-26-0-166-36:1061627:1061782 [7] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-166-36:1061627:1061782 [7] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-166-36:1061627:1061782 [7] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-166-36:1061627:1061782 [7] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-166-36:1061627:1061782 [7] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-166-36:1061624:1061785 [4] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-166-36:1061624:1061785 [4] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-166-36:1061626:1061786 [6] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-166-36:1061626:1061786 [6] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-166-125:1332488:1332648 [6] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-166-125:1332488:1332648 [6] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-166-125:1332488:1332648 [6] NCCL INFO NET/OFI Setting provider_filter to efa
+ip-26-0-166-125:1332488:1332648 [6] NCCL INFO NET/OFI Setting FI_EFA_FORK_SAFE environment variable to 1
+ip-26-0-166-125:1332488:1332648 [6] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-166-125:1332488:1332648 [6] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-166-125:1332488:1332648 [6] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-166-125:1332488:1332648 [6] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-166-125:1332488:1332648 [6] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-166-36:1061622:1061781 [2] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-166-36:1061622:1061781 [2] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-166-125:1332483:1332649 [1] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-166-125:1332483:1332649 [1] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-166-125:1332486:1332650 [4] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-166-125:1332486:1332650 [4] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-166-36:1061624:1061785 [4] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-166-36:1061624:1061785 [4] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-166-36:1061624:1061785 [4] NCCL INFO NET/OFI Setting provider_filter to efa
+ip-26-0-166-36:1061624:1061785 [4] NCCL INFO NET/OFI Setting FI_EFA_FORK_SAFE environment variable to 1
+ip-26-0-166-36:1061624:1061785 [4] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-166-36:1061624:1061785 [4] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-166-36:1061624:1061785 [4] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-166-36:1061624:1061785 [4] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-166-36:1061624:1061785 [4] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-166-36:1061626:1061786 [6] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-166-36:1061626:1061786 [6] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-166-36:1061626:1061786 [6] NCCL INFO NET/OFI Setting provider_filter to efa
+ip-26-0-166-36:1061626:1061786 [6] NCCL INFO NET/OFI Setting FI_EFA_FORK_SAFE environment variable to 1
+ip-26-0-166-36:1061626:1061786 [6] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-166-36:1061626:1061786 [6] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-166-36:1061626:1061786 [6] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-166-36:1061626:1061786 [6] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-166-36:1061626:1061786 [6] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-166-125:1332487:1332652 [5] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-166-125:1332487:1332652 [5] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-166-125:1332483:1332649 [1] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-166-125:1332483:1332649 [1] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-166-125:1332483:1332649 [1] NCCL INFO NET/OFI Setting provider_filter to efa
+ip-26-0-166-125:1332483:1332649 [1] NCCL INFO NET/OFI Setting FI_EFA_FORK_SAFE environment variable to 1
+ip-26-0-166-125:1332483:1332649 [1] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-166-125:1332483:1332649 [1] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-166-125:1332483:1332649 [1] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-166-125:1332483:1332649 [1] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-166-125:1332483:1332649 [1] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-166-125:1332486:1332650 [4] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-166-125:1332486:1332650 [4] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-166-125:1332486:1332650 [4] NCCL INFO NET/OFI Setting provider_filter to efa
+ip-26-0-166-125:1332486:1332650 [4] NCCL INFO NET/OFI Setting FI_EFA_FORK_SAFE environment variable to 1
+ip-26-0-166-125:1332486:1332650 [4] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-166-125:1332486:1332650 [4] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-166-125:1332486:1332650 [4] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-166-125:1332486:1332650 [4] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-166-125:1332486:1332650 [4] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-166-36:1061622:1061781 [2] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-166-36:1061622:1061781 [2] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-166-36:1061622:1061781 [2] NCCL INFO NET/OFI Setting provider_filter to efa
+ip-26-0-166-36:1061622:1061781 [2] NCCL INFO NET/OFI Setting FI_EFA_FORK_SAFE environment variable to 1
+ip-26-0-166-36:1061622:1061781 [2] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-166-36:1061622:1061781 [2] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-166-36:1061622:1061781 [2] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-166-36:1061622:1061781 [2] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-166-36:1061622:1061781 [2] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-166-36:1061621:1061783 [1] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-166-36:1061621:1061783 [1] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-166-36:1061623:1061784 [3] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-166-36:1061623:1061784 [3] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-166-125:1332489:1332653 [7] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-166-125:1332489:1332653 [7] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-166-125:1332487:1332652 [5] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-166-125:1332487:1332652 [5] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-166-125:1332487:1332652 [5] NCCL INFO NET/OFI Setting provider_filter to efa
+ip-26-0-166-125:1332487:1332652 [5] NCCL INFO NET/OFI Setting FI_EFA_FORK_SAFE environment variable to 1
+ip-26-0-166-125:1332487:1332652 [5] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-166-125:1332487:1332652 [5] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-166-125:1332487:1332652 [5] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-166-125:1332487:1332652 [5] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-166-125:1332487:1332652 [5] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-166-125:1332485:1332651 [3] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-166-125:1332485:1332651 [3] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-166-36:1061621:1061783 [1] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-166-36:1061621:1061783 [1] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-166-36:1061621:1061783 [1] NCCL INFO NET/OFI Setting provider_filter to efa
+ip-26-0-166-36:1061621:1061783 [1] NCCL INFO NET/OFI Setting FI_EFA_FORK_SAFE environment variable to 1
+ip-26-0-166-36:1061621:1061783 [1] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-166-36:1061621:1061783 [1] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-166-36:1061621:1061783 [1] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-166-36:1061621:1061783 [1] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-166-36:1061621:1061783 [1] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-166-36:1061623:1061784 [3] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-166-36:1061623:1061784 [3] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-166-36:1061623:1061784 [3] NCCL INFO NET/OFI Setting provider_filter to efa
+ip-26-0-166-36:1061623:1061784 [3] NCCL INFO NET/OFI Setting FI_EFA_FORK_SAFE environment variable to 1
+ip-26-0-166-36:1061623:1061784 [3] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-166-36:1061623:1061784 [3] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-166-36:1061623:1061784 [3] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-166-36:1061623:1061784 [3] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-166-36:1061623:1061784 [3] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-166-36:1061625:1061780 [5] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-166-125:1332489:1332653 [7] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-166-125:1332489:1332653 [7] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-166-125:1332489:1332653 [7] NCCL INFO NET/OFI Setting provider_filter to efa
+ip-26-0-166-125:1332489:1332653 [7] NCCL INFO NET/OFI Setting FI_EFA_FORK_SAFE environment variable to 1
+ip-26-0-166-125:1332489:1332653 [7] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-166-125:1332489:1332653 [7] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-166-125:1332489:1332653 [7] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-166-125:1332489:1332653 [7] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-166-125:1332489:1332653 [7] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-166-125:1332485:1332651 [3] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-166-125:1332485:1332651 [3] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-166-125:1332485:1332651 [3] NCCL INFO NET/OFI Setting provider_filter to efa
+ip-26-0-166-125:1332485:1332651 [3] NCCL INFO NET/OFI Setting FI_EFA_FORK_SAFE environment variable to 1
+ip-26-0-166-125:1332485:1332651 [3] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-166-125:1332485:1332651 [3] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-166-125:1332485:1332651 [3] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-166-125:1332485:1332651 [3] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-166-125:1332485:1332651 [3] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-166-125:1332484:1332647 [2] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-166-36:1061620:1061779 [0] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-166-125:1332488:1332648 [6] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-166-36:1061627:1061782 [7] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-166-125:1332483:1332649 [1] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-166-36:1061624:1061785 [4] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-166-36:1061621:1061783 [1] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-166-36:1061626:1061786 [6] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-166-36:1061622:1061781 [2] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-166-36:1061623:1061784 [3] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-166-125:1332482:1332646 [0] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-166-125:1332482:1332646 [0] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-166-125:1332482:1332646 [0] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-166-125:1332482:1332646 [0] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-166-125:1332482:1332646 [0] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-166-125:1332482:1332646 [0] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-166-125:1332482:1332646 [0] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-166-125:1332482:1332646 [0] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-166-125:1332482:1332646 [0] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-166-125:1332482:1332646 [0] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-166-125:1332482:1332646 [0] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-166-125:1332482:1332646 [0] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-166-125:1332482:1332646 [0] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-166-125:1332482:1332646 [0] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-166-125:1332482:1332646 [0] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-166-125:1332482:1332646 [0] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-166-125:1332482:1332646 [0] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-166-125:1332482:1332646 [0] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-166-125:1332482:1332646 [0] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-166-125:1332482:1332646 [0] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-166-125:1332482:1332646 [0] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-166-125:1332482:1332646 [0] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-166-125:1332482:1332646 [0] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-166-125:1332482:1332646 [0] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-166-125:1332482:1332646 [0] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-166-125:1332482:1332646 [0] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-166-125:1332482:1332646 [0] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-166-125:1332482:1332646 [0] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-166-125:1332482:1332646 [0] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-166-125:1332482:1332646 [0] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-166-125:1332482:1332646 [0] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-166-125:1332482:1332646 [0] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-166-125:1332482:1332646 [0] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/205
+ip-26-0-166-125:1332482:1332646 [0] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-166-125:1332485:1332651 [3] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-166-125:1332486:1332650 [4] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-166-125:1332489:1332653 [7] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-166-125:1332487:1332652 [5] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-166-36:1061625:1061780 [5] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-166-36:1061625:1061780 [5] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-166-36:1061625:1061780 [5] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-166-36:1061625:1061780 [5] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-166-36:1061625:1061780 [5] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-166-36:1061625:1061780 [5] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-166-36:1061625:1061780 [5] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-166-36:1061625:1061780 [5] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-166-36:1061625:1061780 [5] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-166-36:1061625:1061780 [5] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-166-36:1061625:1061780 [5] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-166-36:1061625:1061780 [5] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-166-36:1061625:1061780 [5] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-166-36:1061625:1061780 [5] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-166-36:1061625:1061780 [5] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-166-36:1061625:1061780 [5] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-166-36:1061625:1061780 [5] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-166-36:1061625:1061780 [5] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-166-36:1061625:1061780 [5] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-166-36:1061625:1061780 [5] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-166-36:1061625:1061780 [5] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-166-36:1061625:1061780 [5] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-166-36:1061625:1061780 [5] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-166-36:1061625:1061780 [5] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-166-36:1061625:1061780 [5] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-166-36:1061625:1061780 [5] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-166-36:1061625:1061780 [5] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-166-36:1061625:1061780 [5] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-166-36:1061625:1061780 [5] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-166-36:1061625:1061780 [5] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-166-36:1061625:1061780 [5] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-166-36:1061625:1061780 [5] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-166-36:1061625:1061780 [5] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/184
+ip-26-0-166-36:1061625:1061780 [5] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-166-125:1332482:1332646 [0] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-166-125:1332482:1332646 [0] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-166-125:1332482:1332646 [0] NCCL INFO Using network Libfabric
+ip-26-0-166-125:1332483:1332649 [1] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-166-125:1332483:1332649 [1] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-166-125:1332483:1332649 [1] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-166-125:1332483:1332649 [1] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-166-125:1332483:1332649 [1] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-166-125:1332483:1332649 [1] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-166-125:1332483:1332649 [1] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-166-125:1332483:1332649 [1] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-166-125:1332483:1332649 [1] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-166-125:1332483:1332649 [1] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-166-125:1332483:1332649 [1] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-166-125:1332483:1332649 [1] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-166-125:1332483:1332649 [1] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-166-125:1332483:1332649 [1] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-166-125:1332483:1332649 [1] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-166-125:1332483:1332649 [1] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-166-125:1332483:1332649 [1] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-166-125:1332483:1332649 [1] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-166-125:1332483:1332649 [1] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-166-125:1332483:1332649 [1] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-166-125:1332483:1332649 [1] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-166-125:1332483:1332649 [1] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-166-125:1332483:1332649 [1] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-166-125:1332483:1332649 [1] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-166-125:1332483:1332649 [1] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-166-125:1332483:1332649 [1] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-166-125:1332483:1332649 [1] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-166-125:1332483:1332649 [1] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-166-125:1332483:1332649 [1] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-166-125:1332483:1332649 [1] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-166-125:1332483:1332649 [1] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-166-125:1332483:1332649 [1] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-166-125:1332483:1332649 [1] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/184
+ip-26-0-166-125:1332483:1332649 [1] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-166-36:1061625:1061780 [5] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-166-36:1061625:1061780 [5] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-166-36:1061625:1061780 [5] NCCL INFO Using network Libfabric
+ip-26-0-166-125:1332484:1332647 [2] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-166-125:1332484:1332647 [2] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-166-125:1332484:1332647 [2] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-166-125:1332484:1332647 [2] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-166-125:1332484:1332647 [2] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-166-125:1332484:1332647 [2] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-166-125:1332484:1332647 [2] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-166-125:1332484:1332647 [2] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-166-125:1332484:1332647 [2] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-166-125:1332484:1332647 [2] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-166-125:1332484:1332647 [2] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-166-125:1332484:1332647 [2] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-166-125:1332484:1332647 [2] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-166-125:1332484:1332647 [2] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-166-125:1332484:1332647 [2] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-166-125:1332484:1332647 [2] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-166-125:1332484:1332647 [2] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-166-125:1332484:1332647 [2] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-166-125:1332484:1332647 [2] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-166-125:1332484:1332647 [2] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-166-125:1332484:1332647 [2] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-166-125:1332484:1332647 [2] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-166-125:1332484:1332647 [2] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-166-125:1332484:1332647 [2] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-166-125:1332484:1332647 [2] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-166-125:1332484:1332647 [2] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-166-125:1332484:1332647 [2] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-166-125:1332484:1332647 [2] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-166-125:1332484:1332647 [2] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-166-125:1332484:1332647 [2] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-166-125:1332484:1332647 [2] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-166-125:1332484:1332647 [2] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-166-125:1332484:1332647 [2] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/184
+ip-26-0-166-125:1332484:1332647 [2] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-166-125:1332488:1332648 [6] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-166-125:1332488:1332648 [6] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-166-125:1332488:1332648 [6] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-166-125:1332488:1332648 [6] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-166-125:1332488:1332648 [6] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-166-125:1332488:1332648 [6] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-166-125:1332488:1332648 [6] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-166-125:1332488:1332648 [6] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-166-125:1332488:1332648 [6] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-166-125:1332488:1332648 [6] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-166-125:1332488:1332648 [6] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-166-125:1332488:1332648 [6] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-166-125:1332488:1332648 [6] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-166-125:1332488:1332648 [6] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-166-125:1332488:1332648 [6] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-166-125:1332488:1332648 [6] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-166-125:1332488:1332648 [6] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-166-125:1332488:1332648 [6] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-166-125:1332488:1332648 [6] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-166-125:1332488:1332648 [6] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-166-125:1332488:1332648 [6] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-166-125:1332488:1332648 [6] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-166-125:1332488:1332648 [6] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-166-125:1332488:1332648 [6] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-166-125:1332488:1332648 [6] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-166-125:1332488:1332648 [6] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-166-125:1332488:1332648 [6] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-166-125:1332488:1332648 [6] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-166-125:1332488:1332648 [6] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-166-125:1332488:1332648 [6] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-166-125:1332488:1332648 [6] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-166-125:1332488:1332648 [6] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-166-125:1332488:1332648 [6] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/184
+ip-26-0-166-125:1332488:1332648 [6] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-166-36:1061620:1061779 [0] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-166-36:1061620:1061779 [0] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-166-36:1061620:1061779 [0] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-166-36:1061620:1061779 [0] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-166-36:1061620:1061779 [0] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-166-36:1061620:1061779 [0] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-166-36:1061620:1061779 [0] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-166-36:1061620:1061779 [0] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-166-36:1061620:1061779 [0] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-166-36:1061620:1061779 [0] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-166-36:1061620:1061779 [0] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-166-36:1061620:1061779 [0] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-166-36:1061620:1061779 [0] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-166-36:1061620:1061779 [0] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-166-36:1061620:1061779 [0] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-166-36:1061620:1061779 [0] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-166-36:1061620:1061779 [0] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-166-36:1061620:1061779 [0] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-166-36:1061620:1061779 [0] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-166-36:1061620:1061779 [0] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-166-36:1061620:1061779 [0] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-166-36:1061620:1061779 [0] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-166-36:1061620:1061779 [0] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-166-36:1061620:1061779 [0] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-166-36:1061620:1061779 [0] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-166-36:1061620:1061779 [0] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-166-36:1061620:1061779 [0] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-166-36:1061620:1061779 [0] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-166-36:1061620:1061779 [0] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-166-36:1061620:1061779 [0] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-166-36:1061620:1061779 [0] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-166-36:1061620:1061779 [0] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-166-36:1061620:1061779 [0] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/184
+ip-26-0-166-36:1061620:1061779 [0] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-166-125:1332487:1332652 [5] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-166-125:1332487:1332652 [5] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-166-125:1332487:1332652 [5] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-166-125:1332487:1332652 [5] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-166-125:1332487:1332652 [5] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-166-125:1332487:1332652 [5] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-166-125:1332487:1332652 [5] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-166-125:1332487:1332652 [5] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-166-125:1332487:1332652 [5] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-166-125:1332487:1332652 [5] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-166-125:1332487:1332652 [5] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-166-125:1332487:1332652 [5] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-166-125:1332487:1332652 [5] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-166-125:1332487:1332652 [5] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-166-125:1332487:1332652 [5] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-166-125:1332487:1332652 [5] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-166-125:1332487:1332652 [5] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-166-125:1332487:1332652 [5] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-166-125:1332487:1332652 [5] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-166-125:1332487:1332652 [5] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-166-125:1332487:1332652 [5] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-166-125:1332487:1332652 [5] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-166-125:1332487:1332652 [5] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-166-125:1332487:1332652 [5] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-166-125:1332487:1332652 [5] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-166-125:1332487:1332652 [5] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-166-125:1332487:1332652 [5] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-166-125:1332487:1332652 [5] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-166-125:1332487:1332652 [5] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-166-125:1332487:1332652 [5] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-166-125:1332487:1332652 [5] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-166-125:1332487:1332652 [5] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-166-125:1332487:1332652 [5] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/184
+ip-26-0-166-125:1332487:1332652 [5] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-166-125:1332489:1332653 [7] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-166-125:1332489:1332653 [7] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-166-125:1332489:1332653 [7] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-166-125:1332489:1332653 [7] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-166-125:1332489:1332653 [7] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-166-125:1332489:1332653 [7] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-166-125:1332489:1332653 [7] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-166-125:1332489:1332653 [7] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-166-125:1332489:1332653 [7] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-166-125:1332489:1332653 [7] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-166-125:1332489:1332653 [7] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-166-125:1332489:1332653 [7] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-166-125:1332489:1332653 [7] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-166-125:1332489:1332653 [7] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-166-125:1332489:1332653 [7] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-166-125:1332489:1332653 [7] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-166-125:1332489:1332653 [7] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-166-125:1332489:1332653 [7] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-166-125:1332489:1332653 [7] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-166-125:1332489:1332653 [7] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-166-125:1332489:1332653 [7] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-166-125:1332489:1332653 [7] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-166-125:1332489:1332653 [7] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-166-125:1332489:1332653 [7] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-166-125:1332489:1332653 [7] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-166-125:1332489:1332653 [7] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-166-125:1332489:1332653 [7] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-166-125:1332489:1332653 [7] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-166-125:1332489:1332653 [7] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-166-125:1332489:1332653 [7] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-166-125:1332489:1332653 [7] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-166-125:1332489:1332653 [7] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-166-36:1061622:1061781 [2] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-166-36:1061622:1061781 [2] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-166-36:1061622:1061781 [2] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-166-36:1061622:1061781 [2] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-166-36:1061622:1061781 [2] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-166-36:1061622:1061781 [2] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-166-36:1061622:1061781 [2] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-166-36:1061622:1061781 [2] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-166-36:1061622:1061781 [2] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-166-36:1061622:1061781 [2] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-166-36:1061622:1061781 [2] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-166-36:1061622:1061781 [2] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-166-36:1061622:1061781 [2] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-166-36:1061622:1061781 [2] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-166-36:1061622:1061781 [2] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-166-36:1061622:1061781 [2] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-166-36:1061622:1061781 [2] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-166-36:1061622:1061781 [2] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-166-36:1061622:1061781 [2] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-166-36:1061622:1061781 [2] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-166-36:1061622:1061781 [2] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-166-36:1061622:1061781 [2] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-166-36:1061622:1061781 [2] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-166-36:1061622:1061781 [2] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-166-36:1061622:1061781 [2] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-166-36:1061622:1061781 [2] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-166-36:1061622:1061781 [2] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-166-36:1061622:1061781 [2] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-166-36:1061622:1061781 [2] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-166-36:1061622:1061781 [2] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-166-36:1061622:1061781 [2] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-166-36:1061622:1061781 [2] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-166-36:1061622:1061781 [2] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/184
+ip-26-0-166-36:1061622:1061781 [2] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-166-125:1332486:1332650 [4] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-166-125:1332486:1332650 [4] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-166-125:1332486:1332650 [4] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-166-125:1332486:1332650 [4] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-166-125:1332486:1332650 [4] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-166-125:1332486:1332650 [4] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-166-125:1332486:1332650 [4] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-166-125:1332486:1332650 [4] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-166-125:1332486:1332650 [4] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-166-125:1332486:1332650 [4] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-166-125:1332486:1332650 [4] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-166-125:1332486:1332650 [4] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-166-125:1332486:1332650 [4] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-166-125:1332486:1332650 [4] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-166-125:1332486:1332650 [4] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-166-125:1332486:1332650 [4] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-166-125:1332486:1332650 [4] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-166-125:1332486:1332650 [4] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-166-125:1332486:1332650 [4] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-166-125:1332486:1332650 [4] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-166-125:1332486:1332650 [4] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-166-125:1332486:1332650 [4] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-166-125:1332486:1332650 [4] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-166-125:1332486:1332650 [4] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-166-125:1332486:1332650 [4] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-166-125:1332486:1332650 [4] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-166-125:1332486:1332650 [4] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-166-125:1332486:1332650 [4] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-166-125:1332486:1332650 [4] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-166-125:1332486:1332650 [4] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-166-125:1332486:1332650 [4] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-166-125:1332486:1332650 [4] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-166-125:1332485:1332651 [3] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-166-125:1332485:1332651 [3] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-166-125:1332485:1332651 [3] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-166-125:1332485:1332651 [3] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-166-125:1332485:1332651 [3] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-166-125:1332485:1332651 [3] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-166-125:1332485:1332651 [3] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-166-125:1332485:1332651 [3] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-166-125:1332485:1332651 [3] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-166-125:1332485:1332651 [3] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-166-125:1332485:1332651 [3] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-166-125:1332485:1332651 [3] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-166-36:1061621:1061783 [1] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-166-36:1061621:1061783 [1] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-166-36:1061621:1061783 [1] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-166-36:1061621:1061783 [1] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-166-36:1061621:1061783 [1] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-166-36:1061621:1061783 [1] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-166-36:1061621:1061783 [1] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-166-125:1332485:1332651 [3] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-166-125:1332485:1332651 [3] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-166-125:1332485:1332651 [3] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-166-125:1332485:1332651 [3] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-166-36:1061621:1061783 [1] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-166-36:1061621:1061783 [1] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-166-36:1061621:1061783 [1] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-166-36:1061621:1061783 [1] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-166-36:1061621:1061783 [1] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-166-36:1061621:1061783 [1] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-166-36:1061621:1061783 [1] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-166-36:1061621:1061783 [1] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-166-36:1061621:1061783 [1] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-166-36:1061621:1061783 [1] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-166-36:1061621:1061783 [1] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-166-125:1332485:1332651 [3] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-166-125:1332485:1332651 [3] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-166-125:1332485:1332651 [3] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-166-125:1332485:1332651 [3] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-166-36:1061621:1061783 [1] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-166-36:1061621:1061783 [1] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-166-36:1061621:1061783 [1] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-166-36:1061621:1061783 [1] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-166-125:1332485:1332651 [3] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-166-125:1332485:1332651 [3] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-166-125:1332485:1332651 [3] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-166-125:1332485:1332651 [3] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-166-36:1061621:1061783 [1] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-166-36:1061621:1061783 [1] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-166-36:1061621:1061783 [1] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-166-36:1061621:1061783 [1] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-166-36:1061621:1061783 [1] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-166-36:1061621:1061783 [1] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-166-36:1061621:1061783 [1] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-166-36:1061621:1061783 [1] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-166-36:1061621:1061783 [1] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-166-36:1061621:1061783 [1] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-166-125:1332485:1332651 [3] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-166-125:1332485:1332651 [3] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-166-125:1332485:1332651 [3] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-166-125:1332485:1332651 [3] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-166-125:1332485:1332651 [3] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-166-125:1332485:1332651 [3] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-166-125:1332485:1332651 [3] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-166-125:1332485:1332651 [3] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-166-125:1332489:1332653 [7] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/184
+ip-26-0-166-125:1332489:1332653 [7] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-166-125:1332486:1332650 [4] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/184
+ip-26-0-166-125:1332486:1332650 [4] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-166-125:1332485:1332651 [3] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/184
+ip-26-0-166-125:1332485:1332651 [3] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-166-36:1061621:1061783 [1] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/184
+ip-26-0-166-36:1061621:1061783 [1] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-166-36:1061623:1061784 [3] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-166-36:1061623:1061784 [3] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-166-36:1061623:1061784 [3] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-166-36:1061623:1061784 [3] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-166-36:1061623:1061784 [3] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-166-36:1061623:1061784 [3] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-166-36:1061623:1061784 [3] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-166-36:1061623:1061784 [3] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-166-36:1061623:1061784 [3] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-166-36:1061623:1061784 [3] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-166-36:1061623:1061784 [3] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-166-36:1061623:1061784 [3] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-166-36:1061623:1061784 [3] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-166-36:1061623:1061784 [3] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-166-36:1061623:1061784 [3] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-166-36:1061623:1061784 [3] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-166-36:1061623:1061784 [3] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-166-36:1061623:1061784 [3] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-166-36:1061623:1061784 [3] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-166-36:1061623:1061784 [3] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-166-36:1061623:1061784 [3] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-166-36:1061623:1061784 [3] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-166-36:1061623:1061784 [3] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-166-36:1061623:1061784 [3] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-166-36:1061623:1061784 [3] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-166-36:1061623:1061784 [3] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-166-36:1061623:1061784 [3] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-166-36:1061623:1061784 [3] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-166-36:1061623:1061784 [3] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-166-36:1061623:1061784 [3] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-166-36:1061623:1061784 [3] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-166-36:1061623:1061784 [3] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-166-36:1061623:1061784 [3] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/184
+ip-26-0-166-36:1061623:1061784 [3] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-166-36:1061627:1061782 [7] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-166-36:1061627:1061782 [7] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-166-36:1061627:1061782 [7] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-166-36:1061627:1061782 [7] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-166-36:1061627:1061782 [7] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-166-36:1061627:1061782 [7] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-166-36:1061627:1061782 [7] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-166-36:1061627:1061782 [7] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-166-36:1061627:1061782 [7] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-166-36:1061627:1061782 [7] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-166-36:1061627:1061782 [7] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-166-36:1061627:1061782 [7] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-166-36:1061627:1061782 [7] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-166-36:1061627:1061782 [7] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-166-36:1061627:1061782 [7] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-166-36:1061627:1061782 [7] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-166-36:1061627:1061782 [7] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-166-36:1061627:1061782 [7] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-166-36:1061627:1061782 [7] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-166-36:1061627:1061782 [7] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-166-36:1061627:1061782 [7] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-166-36:1061627:1061782 [7] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-166-36:1061627:1061782 [7] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-166-36:1061627:1061782 [7] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-166-36:1061627:1061782 [7] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-166-36:1061627:1061782 [7] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-166-36:1061627:1061782 [7] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-166-36:1061627:1061782 [7] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-166-36:1061627:1061782 [7] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-166-36:1061627:1061782 [7] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-166-36:1061627:1061782 [7] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-166-36:1061627:1061782 [7] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-166-36:1061627:1061782 [7] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/184
+ip-26-0-166-36:1061627:1061782 [7] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-166-36:1061624:1061785 [4] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-166-36:1061624:1061785 [4] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-166-36:1061624:1061785 [4] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-166-36:1061624:1061785 [4] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-166-36:1061624:1061785 [4] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-166-36:1061624:1061785 [4] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-166-36:1061624:1061785 [4] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-166-36:1061624:1061785 [4] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-166-36:1061624:1061785 [4] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-166-36:1061624:1061785 [4] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-166-36:1061624:1061785 [4] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-166-36:1061624:1061785 [4] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-166-36:1061624:1061785 [4] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-166-36:1061624:1061785 [4] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-166-36:1061624:1061785 [4] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-166-36:1061624:1061785 [4] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-166-36:1061624:1061785 [4] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-166-36:1061624:1061785 [4] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-166-36:1061624:1061785 [4] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-166-36:1061624:1061785 [4] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-166-36:1061624:1061785 [4] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-166-36:1061624:1061785 [4] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-166-36:1061624:1061785 [4] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-166-36:1061624:1061785 [4] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-166-36:1061624:1061785 [4] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-166-36:1061624:1061785 [4] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-166-36:1061624:1061785 [4] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-166-36:1061624:1061785 [4] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-166-36:1061624:1061785 [4] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-166-36:1061624:1061785 [4] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-166-36:1061624:1061785 [4] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-166-36:1061624:1061785 [4] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-166-36:1061624:1061785 [4] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/184
+ip-26-0-166-36:1061624:1061785 [4] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-166-36:1061626:1061786 [6] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-166-36:1061626:1061786 [6] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-166-36:1061626:1061786 [6] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-166-36:1061626:1061786 [6] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-166-36:1061626:1061786 [6] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-166-36:1061626:1061786 [6] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-166-36:1061626:1061786 [6] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-166-36:1061626:1061786 [6] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-166-36:1061626:1061786 [6] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-166-36:1061626:1061786 [6] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-166-36:1061626:1061786 [6] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-166-36:1061626:1061786 [6] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-166-36:1061626:1061786 [6] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-166-36:1061626:1061786 [6] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-166-36:1061626:1061786 [6] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-166-36:1061626:1061786 [6] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-166-36:1061626:1061786 [6] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-166-36:1061626:1061786 [6] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-166-36:1061626:1061786 [6] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-166-36:1061626:1061786 [6] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-166-36:1061626:1061786 [6] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-166-36:1061626:1061786 [6] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-166-36:1061626:1061786 [6] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-166-36:1061626:1061786 [6] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-166-36:1061626:1061786 [6] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-166-36:1061626:1061786 [6] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-166-36:1061626:1061786 [6] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-166-36:1061626:1061786 [6] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-166-36:1061626:1061786 [6] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-166-36:1061626:1061786 [6] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-166-36:1061626:1061786 [6] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-166-36:1061626:1061786 [6] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-166-36:1061626:1061786 [6] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/184
+ip-26-0-166-36:1061626:1061786 [6] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-166-125:1332483:1332649 [1] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-166-125:1332483:1332649 [1] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-166-125:1332483:1332649 [1] NCCL INFO Using network Libfabric
+ip-26-0-166-125:1332484:1332647 [2] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-166-125:1332484:1332647 [2] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-166-125:1332484:1332647 [2] NCCL INFO Using network Libfabric
+ip-26-0-166-125:1332488:1332648 [6] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-166-125:1332488:1332648 [6] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-166-125:1332488:1332648 [6] NCCL INFO Using network Libfabric
+ip-26-0-166-36:1061620:1061779 [0] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-166-36:1061620:1061779 [0] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-166-36:1061620:1061779 [0] NCCL INFO Using network Libfabric
+ip-26-0-166-125:1332487:1332652 [5] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-166-125:1332487:1332652 [5] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-166-125:1332487:1332652 [5] NCCL INFO Using network Libfabric
+ip-26-0-166-125:1332486:1332650 [4] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-166-125:1332486:1332650 [4] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-166-125:1332486:1332650 [4] NCCL INFO Using network Libfabric
+ip-26-0-166-125:1332485:1332651 [3] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-166-125:1332485:1332651 [3] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-166-125:1332485:1332651 [3] NCCL INFO Using network Libfabric
+ip-26-0-166-36:1061622:1061781 [2] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-166-36:1061622:1061781 [2] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-166-36:1061622:1061781 [2] NCCL INFO Using network Libfabric
+ip-26-0-166-125:1332489:1332653 [7] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-166-125:1332489:1332653 [7] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-166-125:1332489:1332653 [7] NCCL INFO Using network Libfabric
+ip-26-0-166-36:1061621:1061783 [1] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-166-36:1061621:1061783 [1] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-166-36:1061621:1061783 [1] NCCL INFO Using network Libfabric
+ip-26-0-166-36:1061623:1061784 [3] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-166-36:1061623:1061784 [3] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-166-36:1061623:1061784 [3] NCCL INFO Using network Libfabric
+ip-26-0-166-36:1061627:1061782 [7] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-166-36:1061627:1061782 [7] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-166-36:1061627:1061782 [7] NCCL INFO Using network Libfabric
+ip-26-0-166-36:1061626:1061786 [6] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-166-36:1061626:1061786 [6] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-166-36:1061626:1061786 [6] NCCL INFO Using network Libfabric
+ip-26-0-166-36:1061624:1061785 [4] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-166-36:1061624:1061785 [4] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-166-36:1061624:1061785 [4] NCCL INFO Using network Libfabric
+ip-26-0-166-36:1061627:1061782 [7] NCCL INFO comm 0x927c9f0 rank 15 nranks 16 cudaDev 7 nvmlDev 7 busId ca000 commId 0x635ad392a550523e - Init START
+ip-26-0-166-36:1061626:1061786 [6] NCCL INFO comm 0x85f9750 rank 14 nranks 16 cudaDev 6 nvmlDev 6 busId b9000 commId 0x635ad392a550523e - Init START
+ip-26-0-166-36:1061624:1061785 [4] NCCL INFO comm 0x9eb7aa0 rank 12 nranks 16 cudaDev 4 nvmlDev 4 busId 97000 commId 0x635ad392a550523e - Init START
+ip-26-0-166-36:1061623:1061784 [3] NCCL INFO comm 0x844a9a0 rank 11 nranks 16 cudaDev 3 nvmlDev 3 busId 86000 commId 0x635ad392a550523e - Init START
+ip-26-0-166-36:1061621:1061783 [1] NCCL INFO comm 0x934f4a0 rank 9 nranks 16 cudaDev 1 nvmlDev 1 busId 64000 commId 0x635ad392a550523e - Init START
+ip-26-0-166-36:1061625:1061780 [5] NCCL INFO comm 0x903d9a0 rank 13 nranks 16 cudaDev 5 nvmlDev 5 busId a8000 commId 0x635ad392a550523e - Init START
+ip-26-0-166-36:1061620:1061779 [0] NCCL INFO comm 0x86930b0 rank 8 nranks 16 cudaDev 0 nvmlDev 0 busId 53000 commId 0x635ad392a550523e - Init START
+ip-26-0-166-36:1061622:1061781 [2] NCCL INFO comm 0x8e8a5c0 rank 10 nranks 16 cudaDev 2 nvmlDev 2 busId 75000 commId 0x635ad392a550523e - Init START
+ip-26-0-166-125:1332488:1332648 [6] NCCL INFO comm 0x8684210 rank 6 nranks 16 cudaDev 6 nvmlDev 6 busId b9000 commId 0x635ad392a550523e - Init START
+ip-26-0-166-125:1332486:1332650 [4] NCCL INFO comm 0x8f3d4d0 rank 4 nranks 16 cudaDev 4 nvmlDev 4 busId 97000 commId 0x635ad392a550523e - Init START
+ip-26-0-166-125:1332489:1332653 [7] NCCL INFO comm 0x9d35c00 rank 7 nranks 16 cudaDev 7 nvmlDev 7 busId ca000 commId 0x635ad392a550523e - Init START
+ip-26-0-166-125:1332487:1332652 [5] NCCL INFO comm 0x98afed0 rank 5 nranks 16 cudaDev 5 nvmlDev 5 busId a8000 commId 0x635ad392a550523e - Init START
+ip-26-0-166-125:1332485:1332651 [3] NCCL INFO comm 0x861c2e0 rank 3 nranks 16 cudaDev 3 nvmlDev 3 busId 86000 commId 0x635ad392a550523e - Init START
+ip-26-0-166-125:1332483:1332649 [1] NCCL INFO comm 0x917b6e0 rank 1 nranks 16 cudaDev 1 nvmlDev 1 busId 64000 commId 0x635ad392a550523e - Init START
+ip-26-0-166-125:1332482:1332646 [0] NCCL INFO comm 0x9676150 rank 0 nranks 16 cudaDev 0 nvmlDev 0 busId 53000 commId 0x635ad392a550523e - Init START
+ip-26-0-166-125:1332484:1332647 [2] NCCL INFO comm 0x98a3c80 rank 2 nranks 16 cudaDev 2 nvmlDev 2 busId 75000 commId 0x635ad392a550523e - Init START
+ip-26-0-166-125:1332487:1332652 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-166-125:1332488:1332648 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-166-125:1332489:1332653 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-166-36:1061620:1061779 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-166-36:1061621:1061783 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-166-36:1061626:1061786 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-166-36:1061624:1061785 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-166-36:1061622:1061781 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-166-36:1061625:1061780 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-166-125:1332486:1332650 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-166-36:1061623:1061784 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-166-125:1332485:1332651 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-166-36:1061627:1061782 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-166-125:1332482:1332646 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/205
+ip-26-0-166-125:1332484:1332647 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-166-125:1332483:1332649 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-166-36:1061624:1061785 [4] NCCL INFO NCCL_P2P_LEVEL set by environment to LOC
+ip-26-0-166-36:1061624:1061785 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-166-125:1332489:1332653 [7] NCCL INFO NCCL_P2P_LEVEL set by environment to LOC
+ip-26-0-166-125:1332489:1332653 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,fff00000,00000000
+ip-26-0-166-36:1061626:1061786 [6] NCCL INFO NCCL_P2P_LEVEL set by environment to LOC
+ip-26-0-166-36:1061626:1061786 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-166-36:1061621:1061783 [1] NCCL INFO NCCL_P2P_LEVEL set by environment to LOC
+ip-26-0-166-36:1061621:1061783 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-166-36:1061625:1061780 [5] NCCL INFO NCCL_P2P_LEVEL set by environment to LOC
+ip-26-0-166-36:1061625:1061780 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-166-36:1061622:1061781 [2] NCCL INFO NCCL_P2P_LEVEL set by environment to LOC
+ip-26-0-166-36:1061622:1061781 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-166-36:1061627:1061782 [7] NCCL INFO NCCL_P2P_LEVEL set by environment to LOC
+ip-26-0-166-36:1061627:1061782 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-166-125:1332485:1332651 [3] NCCL INFO NCCL_P2P_LEVEL set by environment to LOC
+ip-26-0-166-125:1332485:1332651 [3] NCCL INFO Setting affinity for GPU 3 to 0fff,ffffffff
+ip-26-0-166-36:1061623:1061784 [3] NCCL INFO NCCL_P2P_LEVEL set by environment to LOC
+ip-26-0-166-36:1061620:1061779 [0] NCCL INFO NCCL_P2P_LEVEL set by environment to LOC
+ip-26-0-166-36:1061623:1061784 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-166-36:1061620:1061779 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-166-125:1332488:1332648 [6] NCCL INFO NCCL_P2P_LEVEL set by environment to LOC
+ip-26-0-166-125:1332488:1332648 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,fff00000,00000000
+ip-26-0-166-125:1332487:1332652 [5] NCCL INFO NCCL_P2P_LEVEL set by environment to LOC
+ip-26-0-166-125:1332487:1332652 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,fff00000,00000000
+ip-26-0-166-125:1332484:1332647 [2] NCCL INFO NCCL_P2P_LEVEL set by environment to LOC
+ip-26-0-166-125:1332484:1332647 [2] NCCL INFO Setting affinity for GPU 2 to 0fff,ffffffff
+ip-26-0-166-125:1332482:1332646 [0] NCCL INFO NCCL_P2P_LEVEL set by environment to LOC
+ip-26-0-166-125:1332482:1332646 [0] NCCL INFO Setting affinity for GPU 0 to 0fff,ffffffff
+ip-26-0-166-125:1332483:1332649 [1] NCCL INFO NCCL_P2P_LEVEL set by environment to LOC
+ip-26-0-166-125:1332483:1332649 [1] NCCL INFO Setting affinity for GPU 1 to 0fff,ffffffff
+ip-26-0-166-125:1332486:1332650 [4] NCCL INFO NCCL_P2P_LEVEL set by environment to LOC
+ip-26-0-166-125:1332486:1332650 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,fff00000,00000000
+ip-26-0-166-125:1332486:1332650 [4] NCCL INFO Trees [0] 2/6/-1->4->8 [1] 2/6/-1->4->8 [2] -1/-1/-1->4->5 [3] -1/-1/-1->4->5
+ip-26-0-166-125:1332486:1332650 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-125:1332488:1332648 [6] NCCL INFO Trees [0] 5/7/-1->6->4 [1] 5/7/-1->6->4 [2] -1/-1/-1->6->5 [3] -1/-1/-1->6->5
+ip-26-0-166-125:1332488:1332648 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-36:1061627:1061782 [7] NCCL INFO Trees [0] -1/-1/-1->15->14 [1] -1/-1/-1->15->14 [2] 7/-1/-1->15->-1 [3] 7/-1/-1->15->-1
+ip-26-0-166-36:1061626:1061786 [6] NCCL INFO Trees [0] 13/15/-1->14->12 [1] 13/15/-1->14->12 [2] -1/-1/-1->14->13 [3] -1/-1/-1->14->13
+ip-26-0-166-36:1061627:1061782 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-36:1061626:1061786 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-36:1061625:1061780 [5] NCCL INFO Trees [0] -1/-1/-1->13->14 [1] -1/-1/-1->13->14 [2] 14/12/-1->13->11 [3] 14/12/-1->13->11
+ip-26-0-166-36:1061621:1061783 [1] NCCL INFO Trees [0] -1/-1/-1->9->10 [1] -1/-1/-1->9->10 [2] 10/8/-1->9->11 [3] 10/8/-1->9->11
+ip-26-0-166-36:1061623:1061784 [3] NCCL INFO Trees [0] -1/-1/-1->11->10 [1] -1/-1/-1->11->10 [2] 13/9/-1->11->7 [3] 13/9/-1->11->7
+ip-26-0-166-36:1061621:1061783 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-36:1061625:1061780 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-36:1061623:1061784 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-125:1332487:1332652 [5] NCCL INFO Trees [0] -1/-1/-1->5->6 [1] -1/-1/-1->5->6 [2] 6/4/-1->5->3 [3] 6/4/-1->5->3
+ip-26-0-166-36:1061624:1061785 [4] NCCL INFO Trees [0] 10/14/-1->12->8 [1] 10/14/-1->12->8 [2] -1/-1/-1->12->13 [3] -1/-1/-1->12->13
+ip-26-0-166-36:1061624:1061785 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-125:1332487:1332652 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-36:1061620:1061779 [0] NCCL INFO Trees [0] 4/12/-1->8->0 [1] 4/12/-1->8->0 [2] -1/-1/-1->8->9 [3] -1/-1/-1->8->9
+ip-26-0-166-36:1061620:1061779 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-36:1061622:1061781 [2] NCCL INFO Trees [0] 9/11/-1->10->12 [1] 9/11/-1->10->12 [2] -1/-1/-1->10->9 [3] -1/-1/-1->10->9
+ip-26-0-166-36:1061622:1061781 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-125:1332482:1332646 [0] NCCL INFO Channel 00/04 :    0   1   2   3   4   5   6   7   8   9  10  11  12  13  14  15
+ip-26-0-166-125:1332489:1332653 [7] NCCL INFO Trees [0] -1/-1/-1->7->6 [1] -1/-1/-1->7->6 [2] 11/3/-1->7->15 [3] 11/3/-1->7->15
+ip-26-0-166-125:1332482:1332646 [0] NCCL INFO Channel 01/04 :    0   1   2   3   4   5   6   7   8   9  10  11  12  13  14  15
+ip-26-0-166-125:1332483:1332649 [1] NCCL INFO Trees [0] -1/-1/-1->1->2 [1] -1/-1/-1->1->2 [2] 2/0/-1->1->3 [3] 2/0/-1->1->3
+ip-26-0-166-125:1332489:1332653 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-125:1332482:1332646 [0] NCCL INFO Channel 02/04 :    0   1   2   3   4   5   6   7   8   9  10  11  12  13  14  15
+ip-26-0-166-125:1332483:1332649 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-125:1332482:1332646 [0] NCCL INFO Channel 03/04 :    0   1   2   3   4   5   6   7   8   9  10  11  12  13  14  15
+ip-26-0-166-125:1332482:1332646 [0] NCCL INFO Trees [0] 8/-1/-1->0->-1 [1] 8/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-166-125:1332482:1332646 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-125:1332485:1332651 [3] NCCL INFO Trees [0] -1/-1/-1->3->2 [1] -1/-1/-1->3->2 [2] 5/1/-1->3->7 [3] 5/1/-1->3->7
+ip-26-0-166-125:1332485:1332651 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-125:1332484:1332647 [2] NCCL INFO Trees [0] 1/3/-1->2->4 [1] 1/3/-1->2->4 [2] -1/-1/-1->2->1 [3] -1/-1/-1->2->1
+ip-26-0-166-125:1332484:1332647 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-125:1332486:1332650 [4] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-166-125:1332488:1332648 [6] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-166-36:1061621:1061783 [1] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-166-125:1332482:1332646 [0] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-166-36:1061627:1061782 [7] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-166-36:1061626:1061786 [6] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-166-125:1332483:1332649 [1] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-166-125:1332489:1332653 [7] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-166-36:1061625:1061780 [5] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-166-36:1061624:1061785 [4] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-166-125:1332487:1332652 [5] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-166-36:1061623:1061784 [3] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-166-125:1332484:1332647 [2] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-166-125:1332485:1332651 [3] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-166-36:1061620:1061779 [0] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-166-36:1061622:1061781 [2] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-166-36:1061624:1061785 [4] NCCL INFO Channel 00/0 : 11[3] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:1061624:1061785 [4] NCCL INFO Channel 01/0 : 11[3] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:1061624:1061785 [4] NCCL INFO Channel 02/0 : 11[3] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:1061624:1061785 [4] NCCL INFO Channel 03/0 : 11[3] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:1061624:1061785 [4] NCCL INFO Channel 00/0 : 12[4] -> 13[5] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:1061624:1061785 [4] NCCL INFO Channel 01/0 : 12[4] -> 13[5] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:1061624:1061785 [4] NCCL INFO Channel 02/0 : 12[4] -> 13[5] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:1061624:1061785 [4] NCCL INFO Channel 03/0 : 12[4] -> 13[5] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:1332488:1332648 [6] NCCL INFO Channel 00/0 : 5[5] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:1332488:1332648 [6] NCCL INFO Channel 01/0 : 5[5] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:1332486:1332650 [4] NCCL INFO Channel 00/0 : 3[3] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:1061625:1061780 [5] NCCL INFO Channel 00/0 : 12[4] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:1332488:1332648 [6] NCCL INFO Channel 02/0 : 5[5] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:1332486:1332650 [4] NCCL INFO Channel 01/0 : 3[3] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:1061625:1061780 [5] NCCL INFO Channel 01/0 : 12[4] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:1332488:1332648 [6] NCCL INFO Channel 03/0 : 5[5] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:1332486:1332650 [4] NCCL INFO Channel 02/0 : 3[3] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:1061625:1061780 [5] NCCL INFO Channel 02/0 : 12[4] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:1332488:1332648 [6] NCCL INFO Channel 00/0 : 6[6] -> 7[7] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:1332486:1332650 [4] NCCL INFO Channel 03/0 : 3[3] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:1332488:1332648 [6] NCCL INFO Channel 01/0 : 6[6] -> 7[7] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:1061627:1061782 [7] NCCL INFO Channel 00/0 : 14[6] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:1061625:1061780 [5] NCCL INFO Channel 03/0 : 12[4] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:1332488:1332648 [6] NCCL INFO Channel 02/0 : 6[6] -> 7[7] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:1332486:1332650 [4] NCCL INFO Channel 00/0 : 4[4] -> 5[5] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:1061625:1061780 [5] NCCL INFO Channel 00/0 : 13[5] -> 14[6] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:1061625:1061780 [5] NCCL INFO Channel 01/0 : 13[5] -> 14[6] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:1332488:1332648 [6] NCCL INFO Channel 03/0 : 6[6] -> 7[7] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:1332486:1332650 [4] NCCL INFO Channel 01/0 : 4[4] -> 5[5] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:1061627:1061782 [7] NCCL INFO Channel 01/0 : 14[6] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:1061625:1061780 [5] NCCL INFO Channel 02/0 : 13[5] -> 14[6] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:1061622:1061781 [2] NCCL INFO Channel 00/0 : 9[1] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:1061620:1061779 [0] NCCL INFO Channel 00/0 : 7[7] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:1061627:1061782 [7] NCCL INFO Channel 02/0 : 14[6] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:1332489:1332653 [7] NCCL INFO Channel 00/0 : 6[6] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:1061625:1061780 [5] NCCL INFO Channel 03/0 : 13[5] -> 14[6] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:1332486:1332650 [4] NCCL INFO Channel 02/0 : 4[4] -> 5[5] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:1061627:1061782 [7] NCCL INFO Channel 03/0 : 14[6] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:1061620:1061779 [0] NCCL INFO Channel 01/0 : 7[7] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:1332486:1332650 [4] NCCL INFO Channel 03/0 : 4[4] -> 5[5] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:1061626:1061786 [6] NCCL INFO Channel 00/0 : 13[5] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:1332489:1332653 [7] NCCL INFO Channel 01/0 : 6[6] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:1061622:1061781 [2] NCCL INFO Channel 01/0 : 9[1] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:1061621:1061783 [1] NCCL INFO Channel 00/0 : 8[0] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:1061627:1061782 [7] NCCL INFO Channel 00/0 : 15[7] -> 0[0] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:1061620:1061779 [0] NCCL INFO Channel 02/0 : 7[7] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:1332489:1332653 [7] NCCL INFO Channel 02/0 : 6[6] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:1061622:1061781 [2] NCCL INFO Channel 02/0 : 9[1] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:1061626:1061786 [6] NCCL INFO Channel 01/0 : 13[5] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:1332489:1332653 [7] NCCL INFO Channel 03/0 : 6[6] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:1061627:1061782 [7] NCCL INFO Channel 01/0 : 15[7] -> 0[0] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:1061620:1061779 [0] NCCL INFO Channel 03/0 : 7[7] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:1061623:1061784 [3] NCCL INFO Channel 00/0 : 10[2] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:1332489:1332653 [7] NCCL INFO Channel 00/0 : 7[7] -> 8[0] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:1061621:1061783 [1] NCCL INFO Channel 01/0 : 8[0] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:1061622:1061781 [2] NCCL INFO Channel 03/0 : 9[1] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:1061626:1061786 [6] NCCL INFO Channel 02/0 : 13[5] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:1332489:1332653 [7] NCCL INFO Channel 01/0 : 7[7] -> 8[0] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:1061627:1061782 [7] NCCL INFO Channel 02/0 : 15[7] -> 0[0] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:1061620:1061779 [0] NCCL INFO Channel 00/0 : 8[0] -> 9[1] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:1061621:1061783 [1] NCCL INFO Channel 02/0 : 8[0] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:1061626:1061786 [6] NCCL INFO Channel 03/0 : 13[5] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:1061622:1061781 [2] NCCL INFO Channel 00/0 : 10[2] -> 11[3] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:1061623:1061784 [3] NCCL INFO Channel 01/0 : 10[2] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:1061627:1061782 [7] NCCL INFO Channel 03/0 : 15[7] -> 0[0] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:1061620:1061779 [0] NCCL INFO Channel 01/0 : 8[0] -> 9[1] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:1332489:1332653 [7] NCCL INFO Channel 02/0 : 7[7] -> 8[0] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:1332489:1332653 [7] NCCL INFO Channel 03/0 : 7[7] -> 8[0] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:1061626:1061786 [6] NCCL INFO Channel 00/0 : 14[6] -> 15[7] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:1061622:1061781 [2] NCCL INFO Channel 01/0 : 10[2] -> 11[3] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:1061621:1061783 [1] NCCL INFO Channel 03/0 : 8[0] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:1061623:1061784 [3] NCCL INFO Channel 02/0 : 10[2] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:1061620:1061779 [0] NCCL INFO Channel 02/0 : 8[0] -> 9[1] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:1332487:1332652 [5] NCCL INFO Channel 00/0 : 4[4] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:1061622:1061781 [2] NCCL INFO Channel 02/0 : 10[2] -> 11[3] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:1061623:1061784 [3] NCCL INFO Channel 03/0 : 10[2] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:1061620:1061779 [0] NCCL INFO Channel 03/0 : 8[0] -> 9[1] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:1061621:1061783 [1] NCCL INFO Channel 00/0 : 9[1] -> 10[2] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:1061626:1061786 [6] NCCL INFO Channel 01/0 : 14[6] -> 15[7] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:1332487:1332652 [5] NCCL INFO Channel 01/0 : 4[4] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:1061622:1061781 [2] NCCL INFO Channel 03/0 : 10[2] -> 11[3] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:1061623:1061784 [3] NCCL INFO Channel 00/0 : 11[3] -> 12[4] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:1061621:1061783 [1] NCCL INFO Channel 01/0 : 9[1] -> 10[2] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:1332487:1332652 [5] NCCL INFO Channel 02/0 : 4[4] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:1061626:1061786 [6] NCCL INFO Channel 02/0 : 14[6] -> 15[7] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:1061623:1061784 [3] NCCL INFO Channel 01/0 : 11[3] -> 12[4] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:1061621:1061783 [1] NCCL INFO Channel 02/0 : 9[1] -> 10[2] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:1332482:1332646 [0] NCCL INFO Channel 00/0 : 15[7] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:1332487:1332652 [5] NCCL INFO Channel 03/0 : 4[4] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:1061623:1061784 [3] NCCL INFO Channel 02/0 : 11[3] -> 12[4] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:1061626:1061786 [6] NCCL INFO Channel 03/0 : 14[6] -> 15[7] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:1332487:1332652 [5] NCCL INFO Channel 00/0 : 5[5] -> 6[6] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:1061621:1061783 [1] NCCL INFO Channel 03/0 : 9[1] -> 10[2] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:1061623:1061784 [3] NCCL INFO Channel 03/0 : 11[3] -> 12[4] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:1332482:1332646 [0] NCCL INFO Channel 01/0 : 15[7] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:1332487:1332652 [5] NCCL INFO Channel 01/0 : 5[5] -> 6[6] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:1332482:1332646 [0] NCCL INFO Channel 02/0 : 15[7] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:1332487:1332652 [5] NCCL INFO Channel 02/0 : 5[5] -> 6[6] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:1332482:1332646 [0] NCCL INFO Channel 03/0 : 15[7] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:1332487:1332652 [5] NCCL INFO Channel 03/0 : 5[5] -> 6[6] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:1332482:1332646 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[1] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:1332482:1332646 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[1] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:1332482:1332646 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[1] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:1332482:1332646 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[1] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:1332483:1332649 [1] NCCL INFO Channel 00/0 : 0[0] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:1332483:1332649 [1] NCCL INFO Channel 01/0 : 0[0] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:1332483:1332649 [1] NCCL INFO Channel 02/0 : 0[0] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:1332483:1332649 [1] NCCL INFO Channel 03/0 : 0[0] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:1332485:1332651 [3] NCCL INFO Channel 00/0 : 2[2] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:1332483:1332649 [1] NCCL INFO Channel 00/0 : 1[1] -> 2[2] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:1332483:1332649 [1] NCCL INFO Channel 01/0 : 1[1] -> 2[2] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:1332485:1332651 [3] NCCL INFO Channel 01/0 : 2[2] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:1332483:1332649 [1] NCCL INFO Channel 02/0 : 1[1] -> 2[2] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:1332485:1332651 [3] NCCL INFO Channel 02/0 : 2[2] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:1332483:1332649 [1] NCCL INFO Channel 03/0 : 1[1] -> 2[2] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:1332485:1332651 [3] NCCL INFO Channel 03/0 : 2[2] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:1332485:1332651 [3] NCCL INFO Channel 00/0 : 3[3] -> 4[4] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:1332485:1332651 [3] NCCL INFO Channel 01/0 : 3[3] -> 4[4] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:1332485:1332651 [3] NCCL INFO Channel 02/0 : 3[3] -> 4[4] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:1332485:1332651 [3] NCCL INFO Channel 03/0 : 3[3] -> 4[4] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:1332484:1332647 [2] NCCL INFO Channel 00/0 : 1[1] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:1332484:1332647 [2] NCCL INFO Channel 01/0 : 1[1] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:1332484:1332647 [2] NCCL INFO Channel 02/0 : 1[1] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:1332484:1332647 [2] NCCL INFO Channel 03/0 : 1[1] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:1332484:1332647 [2] NCCL INFO Channel 00/0 : 2[2] -> 3[3] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:1332484:1332647 [2] NCCL INFO Channel 01/0 : 2[2] -> 3[3] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:1332484:1332647 [2] NCCL INFO Channel 02/0 : 2[2] -> 3[3] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:1332484:1332647 [2] NCCL INFO Channel 03/0 : 2[2] -> 3[3] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:1061621:1061783 [1] NCCL INFO Connected all rings
+ip-26-0-166-36:1061621:1061783 [1] NCCL INFO Channel 02/0 : 9[1] -> 11[3] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:1061621:1061783 [1] NCCL INFO Channel 03/0 : 9[1] -> 11[3] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:1061624:1061785 [4] NCCL INFO Connected all rings
+ip-26-0-166-36:1061624:1061785 [4] NCCL INFO Channel 00/0 : 10[2] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:1061624:1061785 [4] NCCL INFO Channel 01/0 : 10[2] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:1061624:1061785 [4] NCCL INFO Channel 00/0 : 12[4] -> 14[6] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:1061624:1061785 [4] NCCL INFO Channel 01/0 : 12[4] -> 14[6] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:1061625:1061780 [5] NCCL INFO Connected all rings
+ip-26-0-166-36:1061625:1061780 [5] NCCL INFO Channel 02/0 : 11[3] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:1061625:1061780 [5] NCCL INFO Channel 03/0 : 11[3] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:1061623:1061784 [3] NCCL INFO Connected all rings
+ip-26-0-166-36:1061623:1061784 [3] NCCL INFO Channel 02/0 : 9[1] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:1061623:1061784 [3] NCCL INFO Channel 03/0 : 9[1] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:1061623:1061784 [3] NCCL INFO Channel 02/0 : 11[3] -> 13[5] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:1061623:1061784 [3] NCCL INFO Channel 03/0 : 11[3] -> 13[5] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:1061621:1061783 [1] NCCL INFO Channel 02/0 : 11[3] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:1061625:1061780 [5] NCCL INFO Channel 02/0 : 13[5] -> 11[3] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:1061621:1061783 [1] NCCL INFO Channel 03/0 : 11[3] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:1061623:1061784 [3] NCCL INFO Channel 02/0 : 7[7] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:1061625:1061780 [5] NCCL INFO Channel 03/0 : 13[5] -> 11[3] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:1061623:1061784 [3] NCCL INFO Channel 03/0 : 7[7] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:1061626:1061786 [6] NCCL INFO Connected all rings
+ip-26-0-166-36:1061626:1061786 [6] NCCL INFO Channel 00/0 : 12[4] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:1061626:1061786 [6] NCCL INFO Channel 01/0 : 12[4] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:1061626:1061786 [6] NCCL INFO Channel 00/0 : 14[6] -> 12[4] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:1061626:1061786 [6] NCCL INFO Channel 01/0 : 14[6] -> 12[4] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:1061620:1061779 [0] NCCL INFO Connected all rings
+ip-26-0-166-36:1061627:1061782 [7] NCCL INFO Connected all rings
+ip-26-0-166-36:1061620:1061779 [0] NCCL INFO Channel 00/0 : 4[4] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:1061620:1061779 [0] NCCL INFO Channel 01/0 : 4[4] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:1061627:1061782 [7] NCCL INFO Channel 02/0 : 7[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:1061620:1061779 [0] NCCL INFO Channel 00/0 : 8[0] -> 12[4] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:1061627:1061782 [7] NCCL INFO Channel 03/0 : 7[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:1061620:1061779 [0] NCCL INFO Channel 01/0 : 8[0] -> 12[4] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:1061627:1061782 [7] NCCL INFO Channel 02/0 : 15[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:1061627:1061782 [7] NCCL INFO Channel 03/0 : 15[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:1061622:1061781 [2] NCCL INFO Connected all rings
+ip-26-0-166-36:1061622:1061781 [2] NCCL INFO Channel 00/0 : 10[2] -> 12[4] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:1061622:1061781 [2] NCCL INFO Channel 01/0 : 10[2] -> 12[4] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:1061624:1061785 [4] NCCL INFO Channel 00/0 : 8[0] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:1061622:1061781 [2] NCCL INFO Channel 00/0 : 12[4] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:1061624:1061785 [4] NCCL INFO Channel 01/0 : 8[0] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:1061622:1061781 [2] NCCL INFO Channel 01/0 : 12[4] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:1061624:1061785 [4] NCCL INFO Channel 00/0 : 12[4] -> 8[0] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:1061624:1061785 [4] NCCL INFO Channel 01/0 : 12[4] -> 8[0] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:1332487:1332652 [5] NCCL INFO Connected all rings
+ip-26-0-166-125:1332487:1332652 [5] NCCL INFO Channel 02/0 : 3[3] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:1332487:1332652 [5] NCCL INFO Channel 03/0 : 3[3] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:1332488:1332648 [6] NCCL INFO Connected all rings
+ip-26-0-166-125:1332488:1332648 [6] NCCL INFO Channel 00/0 : 4[4] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:1332488:1332648 [6] NCCL INFO Channel 01/0 : 4[4] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:1332489:1332653 [7] NCCL INFO Connected all rings
+ip-26-0-166-125:1332489:1332653 [7] NCCL INFO Channel 02/0 : 3[3] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:1332489:1332653 [7] NCCL INFO Channel 03/0 : 3[3] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:1332482:1332646 [0] NCCL INFO Connected all rings
+ip-26-0-166-125:1332489:1332653 [7] NCCL INFO Channel 02/0 : 7[7] -> 11[3] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:1332489:1332653 [7] NCCL INFO Channel 03/0 : 7[7] -> 11[3] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:1332482:1332646 [0] NCCL INFO Channel 00/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:1332482:1332646 [0] NCCL INFO Channel 01/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:1332482:1332646 [0] NCCL INFO Channel 00/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:1332482:1332646 [0] NCCL INFO Channel 01/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:1061623:1061784 [3] NCCL INFO Channel 02/0 : 11[3] -> 7[7] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:1061623:1061784 [3] NCCL INFO Channel 03/0 : 11[3] -> 7[7] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:1332485:1332651 [3] NCCL INFO Connected all rings
+ip-26-0-166-125:1332484:1332647 [2] NCCL INFO Connected all rings
+ip-26-0-166-125:1332484:1332647 [2] NCCL INFO Channel 00/0 : 2[2] -> 4[4] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:1332484:1332647 [2] NCCL INFO Channel 01/0 : 2[2] -> 4[4] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:1332485:1332651 [3] NCCL INFO Channel 02/0 : 1[1] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:1332485:1332651 [3] NCCL INFO Channel 03/0 : 1[1] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:1332485:1332651 [3] NCCL INFO Channel 02/0 : 3[3] -> 5[5] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:1332485:1332651 [3] NCCL INFO Channel 03/0 : 3[3] -> 5[5] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:1332487:1332652 [5] NCCL INFO Channel 02/0 : 5[5] -> 3[3] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:1332486:1332650 [4] NCCL INFO Connected all rings
+ip-26-0-166-125:1332487:1332652 [5] NCCL INFO Channel 03/0 : 5[5] -> 3[3] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:1332486:1332650 [4] NCCL INFO Channel 00/0 : 2[2] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:1332486:1332650 [4] NCCL INFO Channel 01/0 : 2[2] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:1332486:1332650 [4] NCCL INFO Channel 00/0 : 4[4] -> 6[6] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:1332486:1332650 [4] NCCL INFO Channel 01/0 : 4[4] -> 6[6] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:1332484:1332647 [2] NCCL INFO Channel 00/0 : 4[4] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:1332484:1332647 [2] NCCL INFO Channel 01/0 : 4[4] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:1332488:1332648 [6] NCCL INFO Channel 00/0 : 6[6] -> 4[4] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:1332486:1332650 [4] NCCL INFO Channel 00/0 : 4[4] -> 8[0] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:1332488:1332648 [6] NCCL INFO Channel 01/0 : 6[6] -> 4[4] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:1332486:1332650 [4] NCCL INFO Channel 01/0 : 4[4] -> 8[0] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:1332486:1332650 [4] NCCL INFO Channel 00/0 : 8[0] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:1061620:1061779 [0] NCCL INFO Channel 00/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:1332486:1332650 [4] NCCL INFO Channel 01/0 : 8[0] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:1061620:1061779 [0] NCCL INFO Channel 01/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:1061620:1061779 [0] NCCL INFO Channel 00/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:1061620:1061779 [0] NCCL INFO Channel 01/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:1061620:1061779 [0] NCCL INFO Channel 00/0 : 12[4] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:1332483:1332649 [1] NCCL INFO Connected all rings
+ip-26-0-166-36:1061620:1061779 [0] NCCL INFO Channel 01/0 : 12[4] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:1332482:1332646 [0] NCCL INFO Channel 02/0 : 1[1] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:1061620:1061779 [0] NCCL INFO Channel 00/0 : 8[0] -> 4[4] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:1332483:1332649 [1] NCCL INFO Channel 02/0 : 1[1] -> 3[3] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:1061620:1061779 [0] NCCL INFO Channel 01/0 : 8[0] -> 4[4] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:1332482:1332646 [0] NCCL INFO Channel 03/0 : 1[1] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:1332483:1332649 [1] NCCL INFO Channel 03/0 : 1[1] -> 3[3] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:1061624:1061785 [4] NCCL INFO Channel 00/0 : 14[6] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:1332485:1332651 [3] NCCL INFO Channel 02/0 : 3[3] -> 7[7] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:1061624:1061785 [4] NCCL INFO Channel 01/0 : 14[6] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:1332483:1332649 [1] NCCL INFO Channel 02/0 : 3[3] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:1061624:1061785 [4] NCCL INFO Channel 00/0 : 12[4] -> 10[2] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:1332485:1332651 [3] NCCL INFO Channel 03/0 : 3[3] -> 7[7] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:1332483:1332649 [1] NCCL INFO Channel 03/0 : 3[3] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:1061620:1061779 [0] NCCL INFO Channel 02/0 : 9[1] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:1061624:1061785 [4] NCCL INFO Channel 01/0 : 12[4] -> 10[2] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:1332486:1332650 [4] NCCL INFO Channel 00/0 : 6[6] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:1061620:1061779 [0] NCCL INFO Channel 03/0 : 9[1] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:1332486:1332650 [4] NCCL INFO Channel 01/0 : 6[6] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:1332485:1332651 [3] NCCL INFO Channel 02/0 : 7[7] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:1332489:1332653 [7] NCCL INFO Channel 02/0 : 15[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:1332486:1332650 [4] NCCL INFO Channel 00/0 : 4[4] -> 2[2] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:1061626:1061786 [6] NCCL INFO Channel 00/0 : 15[7] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:1332485:1332651 [3] NCCL INFO Channel 03/0 : 7[7] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:1332489:1332653 [7] NCCL INFO Channel 03/0 : 15[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:1332486:1332650 [4] NCCL INFO Channel 01/0 : 4[4] -> 2[2] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:1061622:1061781 [2] NCCL INFO Channel 00/0 : 11[3] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:1061626:1061786 [6] NCCL INFO Channel 01/0 : 15[7] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:1061624:1061785 [4] NCCL INFO Channel 02/0 : 13[5] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:1332489:1332653 [7] NCCL INFO Channel 02/0 : 7[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:1061622:1061781 [2] NCCL INFO Channel 01/0 : 11[3] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:1061626:1061786 [6] NCCL INFO Channel 00/0 : 14[6] -> 13[5] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:1061624:1061785 [4] NCCL INFO Channel 03/0 : 13[5] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:1332488:1332648 [6] NCCL INFO Channel 00/0 : 7[7] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:1332489:1332653 [7] NCCL INFO Channel 03/0 : 7[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:1061622:1061781 [2] NCCL INFO Channel 00/0 : 10[2] -> 9[1] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:1061626:1061786 [6] NCCL INFO Channel 01/0 : 14[6] -> 13[5] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:1332488:1332648 [6] NCCL INFO Channel 01/0 : 7[7] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:1061622:1061781 [2] NCCL INFO Channel 01/0 : 10[2] -> 9[1] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:1061626:1061786 [6] NCCL INFO Channel 02/0 : 14[6] -> 13[5] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:1332484:1332647 [2] NCCL INFO Channel 00/0 : 3[3] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:1332486:1332650 [4] NCCL INFO Channel 02/0 : 5[5] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:1061622:1061781 [2] NCCL INFO Channel 02/0 : 10[2] -> 9[1] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:1332488:1332648 [6] NCCL INFO Channel 00/0 : 6[6] -> 5[5] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:1061626:1061786 [6] NCCL INFO Channel 03/0 : 14[6] -> 13[5] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:1332484:1332647 [2] NCCL INFO Channel 01/0 : 3[3] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:1332486:1332650 [4] NCCL INFO Channel 03/0 : 5[5] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:1061622:1061781 [2] NCCL INFO Channel 03/0 : 10[2] -> 9[1] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:1332488:1332648 [6] NCCL INFO Channel 01/0 : 6[6] -> 5[5] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:1061627:1061782 [7] NCCL INFO Channel 00/0 : 15[7] -> 14[6] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:1332489:1332653 [7] NCCL INFO Channel 02/0 : 11[3] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:1332488:1332648 [6] NCCL INFO Channel 02/0 : 6[6] -> 5[5] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:1332484:1332647 [2] NCCL INFO Channel 00/0 : 2[2] -> 1[1] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:1061627:1061782 [7] NCCL INFO Channel 01/0 : 15[7] -> 14[6] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:1332488:1332648 [6] NCCL INFO Channel 03/0 : 6[6] -> 5[5] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:1332489:1332653 [7] NCCL INFO Channel 03/0 : 11[3] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:1332484:1332647 [2] NCCL INFO Channel 01/0 : 2[2] -> 1[1] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:1332489:1332653 [7] NCCL INFO Channel 02/0 : 7[7] -> 3[3] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:1332484:1332647 [2] NCCL INFO Channel 02/0 : 2[2] -> 1[1] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:1332489:1332653 [7] NCCL INFO Channel 03/0 : 7[7] -> 3[3] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:1332484:1332647 [2] NCCL INFO Channel 03/0 : 2[2] -> 1[1] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:1061623:1061784 [3] NCCL INFO Channel 02/0 : 13[5] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:1061623:1061784 [3] NCCL INFO Channel 03/0 : 13[5] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:1332485:1332651 [3] NCCL INFO Channel 02/0 : 5[5] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:1061623:1061784 [3] NCCL INFO Channel 02/0 : 11[3] -> 9[1] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:1332489:1332653 [7] NCCL INFO Channel 00/0 : 7[7] -> 6[6] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:1332485:1332651 [3] NCCL INFO Channel 03/0 : 5[5] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:1332489:1332653 [7] NCCL INFO Channel 01/0 : 7[7] -> 6[6] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:1061623:1061784 [3] NCCL INFO Channel 03/0 : 11[3] -> 9[1] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:1332485:1332651 [3] NCCL INFO Channel 02/0 : 3[3] -> 1[1] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:1332485:1332651 [3] NCCL INFO Channel 03/0 : 3[3] -> 1[1] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:1061625:1061780 [5] NCCL INFO Channel 00/0 : 14[6] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:1332487:1332652 [5] NCCL INFO Channel 00/0 : 6[6] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:1061623:1061784 [3] NCCL INFO Channel 00/0 : 11[3] -> 10[2] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:1061621:1061783 [1] NCCL INFO Channel 00/0 : 10[2] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:1061625:1061780 [5] NCCL INFO Channel 01/0 : 14[6] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:1332483:1332649 [1] NCCL INFO Channel 00/0 : 2[2] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:1332485:1332651 [3] NCCL INFO Channel 00/0 : 3[3] -> 2[2] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:1061623:1061784 [3] NCCL INFO Channel 01/0 : 11[3] -> 10[2] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:1061621:1061783 [1] NCCL INFO Channel 01/0 : 10[2] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:1061625:1061780 [5] NCCL INFO Channel 02/0 : 14[6] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:1332487:1332652 [5] NCCL INFO Channel 01/0 : 6[6] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:1332485:1332651 [3] NCCL INFO Channel 01/0 : 3[3] -> 2[2] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:1332483:1332649 [1] NCCL INFO Channel 01/0 : 2[2] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:1332487:1332652 [5] NCCL INFO Channel 02/0 : 6[6] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:1061621:1061783 [1] NCCL INFO Channel 02/0 : 10[2] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:1061625:1061780 [5] NCCL INFO Channel 03/0 : 14[6] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:1332483:1332649 [1] NCCL INFO Channel 02/0 : 2[2] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:1332487:1332652 [5] NCCL INFO Channel 03/0 : 6[6] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:1061625:1061780 [5] NCCL INFO Channel 02/0 : 13[5] -> 12[4] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:1061621:1061783 [1] NCCL INFO Channel 03/0 : 10[2] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:1332487:1332652 [5] NCCL INFO Channel 02/0 : 5[5] -> 4[4] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:1061625:1061780 [5] NCCL INFO Channel 03/0 : 13[5] -> 12[4] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:1061621:1061783 [1] NCCL INFO Channel 02/0 : 9[1] -> 8[0] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:1332483:1332649 [1] NCCL INFO Channel 03/0 : 2[2] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:1332487:1332652 [5] NCCL INFO Channel 03/0 : 5[5] -> 4[4] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:1332483:1332649 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[0] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:1061621:1061783 [1] NCCL INFO Channel 03/0 : 9[1] -> 8[0] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:1332483:1332649 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[0] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:1061627:1061782 [7] NCCL INFO Connected all trees
+ip-26-0-166-36:1061627:1061782 [7] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+ip-26-0-166-36:1061627:1061782 [7] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-36:1061627:1061782 [7] NCCL INFO comm 0x927c9f0 rank 15 nranks 16 cudaDev 7 nvmlDev 7 busId ca000 commId 0x635ad392a550523e - Init COMPLETE
+ip-26-0-166-36:1061622:1061781 [2] NCCL INFO Connected all trees
+ip-26-0-166-36:1061622:1061781 [2] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+ip-26-0-166-36:1061622:1061781 [2] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-36:1061621:1061783 [1] NCCL INFO Connected all trees
+ip-26-0-166-36:1061621:1061783 [1] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+ip-26-0-166-36:1061621:1061783 [1] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-36:1061626:1061786 [6] NCCL INFO Connected all trees
+ip-26-0-166-36:1061626:1061786 [6] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+ip-26-0-166-36:1061626:1061786 [6] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-36:1061625:1061780 [5] NCCL INFO Connected all trees
+ip-26-0-166-36:1061625:1061780 [5] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+ip-26-0-166-36:1061625:1061780 [5] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-36:1061622:1061781 [2] NCCL INFO comm 0x8e8a5c0 rank 10 nranks 16 cudaDev 2 nvmlDev 2 busId 75000 commId 0x635ad392a550523e - Init COMPLETE
+ip-26-0-166-36:1061621:1061783 [1] NCCL INFO comm 0x934f4a0 rank 9 nranks 16 cudaDev 1 nvmlDev 1 busId 64000 commId 0x635ad392a550523e - Init COMPLETE
+ip-26-0-166-36:1061626:1061786 [6] NCCL INFO comm 0x85f9750 rank 14 nranks 16 cudaDev 6 nvmlDev 6 busId b9000 commId 0x635ad392a550523e - Init COMPLETE
+ip-26-0-166-36:1061625:1061780 [5] NCCL INFO comm 0x903d9a0 rank 13 nranks 16 cudaDev 5 nvmlDev 5 busId a8000 commId 0x635ad392a550523e - Init COMPLETE
+ip-26-0-166-36:1061623:1061784 [3] NCCL INFO Connected all trees
+ip-26-0-166-36:1061623:1061784 [3] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+ip-26-0-166-36:1061623:1061784 [3] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-125:1332482:1332646 [0] NCCL INFO Connected all trees
+ip-26-0-166-125:1332482:1332646 [0] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+ip-26-0-166-125:1332482:1332646 [0] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-36:1061624:1061785 [4] NCCL INFO Connected all trees
+ip-26-0-166-36:1061624:1061785 [4] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+ip-26-0-166-36:1061624:1061785 [4] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-36:1061623:1061784 [3] NCCL INFO comm 0x844a9a0 rank 11 nranks 16 cudaDev 3 nvmlDev 3 busId 86000 commId 0x635ad392a550523e - Init COMPLETE
+ip-26-0-166-125:1332482:1332646 [0] NCCL INFO comm 0x9676150 rank 0 nranks 16 cudaDev 0 nvmlDev 0 busId 53000 commId 0x635ad392a550523e - Init COMPLETE
+ip-26-0-166-36:1061620:1061779 [0] NCCL INFO Connected all trees
+ip-26-0-166-36:1061620:1061779 [0] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+ip-26-0-166-36:1061620:1061779 [0] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-36:1061624:1061785 [4] NCCL INFO comm 0x9eb7aa0 rank 12 nranks 16 cudaDev 4 nvmlDev 4 busId 97000 commId 0x635ad392a550523e - Init COMPLETE
+ip-26-0-166-36:1061620:1061779 [0] NCCL INFO comm 0x86930b0 rank 8 nranks 16 cudaDev 0 nvmlDev 0 busId 53000 commId 0x635ad392a550523e - Init COMPLETE
+ip-26-0-166-125:1332484:1332647 [2] NCCL INFO Connected all trees
+ip-26-0-166-125:1332484:1332647 [2] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+ip-26-0-166-125:1332484:1332647 [2] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-125:1332488:1332648 [6] NCCL INFO Connected all trees
+ip-26-0-166-125:1332488:1332648 [6] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+ip-26-0-166-125:1332488:1332648 [6] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-125:1332483:1332649 [1] NCCL INFO Connected all trees
+ip-26-0-166-125:1332483:1332649 [1] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+ip-26-0-166-125:1332483:1332649 [1] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-125:1332487:1332652 [5] NCCL INFO Connected all trees
+ip-26-0-166-125:1332487:1332652 [5] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+ip-26-0-166-125:1332487:1332652 [5] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-125:1332484:1332647 [2] NCCL INFO comm 0x98a3c80 rank 2 nranks 16 cudaDev 2 nvmlDev 2 busId 75000 commId 0x635ad392a550523e - Init COMPLETE
+ip-26-0-166-125:1332488:1332648 [6] NCCL INFO comm 0x8684210 rank 6 nranks 16 cudaDev 6 nvmlDev 6 busId b9000 commId 0x635ad392a550523e - Init COMPLETE
+ip-26-0-166-125:1332483:1332649 [1] NCCL INFO comm 0x917b6e0 rank 1 nranks 16 cudaDev 1 nvmlDev 1 busId 64000 commId 0x635ad392a550523e - Init COMPLETE
+ip-26-0-166-125:1332487:1332652 [5] NCCL INFO comm 0x98afed0 rank 5 nranks 16 cudaDev 5 nvmlDev 5 busId a8000 commId 0x635ad392a550523e - Init COMPLETE
+ip-26-0-166-125:1332485:1332651 [3] NCCL INFO Connected all trees
+ip-26-0-166-125:1332485:1332651 [3] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+ip-26-0-166-125:1332485:1332651 [3] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-125:1332489:1332653 [7] NCCL INFO Connected all trees
+ip-26-0-166-125:1332489:1332653 [7] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+ip-26-0-166-125:1332489:1332653 [7] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-125:1332485:1332651 [3] NCCL INFO comm 0x861c2e0 rank 3 nranks 16 cudaDev 3 nvmlDev 3 busId 86000 commId 0x635ad392a550523e - Init COMPLETE
+ip-26-0-166-125:1332486:1332650 [4] NCCL INFO Connected all trees
+ip-26-0-166-125:1332486:1332650 [4] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+ip-26-0-166-125:1332486:1332650 [4] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-125:1332489:1332653 [7] NCCL INFO comm 0x9d35c00 rank 7 nranks 16 cudaDev 7 nvmlDev 7 busId ca000 commId 0x635ad392a550523e - Init COMPLETE
+ip-26-0-166-125:1332486:1332650 [4] NCCL INFO comm 0x8f3d4d0 rank 4 nranks 16 cudaDev 4 nvmlDev 4 busId 97000 commId 0x635ad392a550523e - Init COMPLETE
+ip-26-0-166-125:1332482:1332879 [0] NCCL INFO Using network Libfabric
+ip-26-0-166-125:1332484:1332881 [2] NCCL INFO Using network Libfabric
+ip-26-0-166-36:1061627:1062008 [7] NCCL INFO Using network Libfabric
+ip-26-0-166-36:1061626:1062009 [6] NCCL INFO Using network Libfabric
+ip-26-0-166-125:1332485:1332882 [3] NCCL INFO Using network Libfabric
+ip-26-0-166-36:1061625:1062010 [5] NCCL INFO Using network Libfabric
+ip-26-0-166-125:1332483:1332880 [1] NCCL INFO Using network Libfabric
+ip-26-0-166-36:1061622:1062012 [2] NCCL INFO Using network Libfabric
+ip-26-0-166-125:1332488:1332883 [6] NCCL INFO Using network Libfabric
+ip-26-0-166-125:1332489:1332886 [7] NCCL INFO Using network Libfabric
+ip-26-0-166-36:1061624:1062011 [4] NCCL INFO Using network Libfabric
+ip-26-0-166-36:1061621:1062014 [1] NCCL INFO Using network Libfabric
+ip-26-0-166-36:1061620:1062015 [0] NCCL INFO Using network Libfabric
+ip-26-0-166-125:1332487:1332884 [5] NCCL INFO Using network Libfabric
+ip-26-0-166-36:1061623:1062013 [3] NCCL INFO Using network Libfabric
+ip-26-0-166-125:1332486:1332885 [4] NCCL INFO Using network Libfabric
+ip-26-0-166-36:1061627:1062008 [7] NCCL INFO comm 0x94b2540 rank 15 nranks 16 cudaDev 7 nvmlDev 7 busId ca000 commId 0xe69d592a22737672 - Init START
+ip-26-0-166-36:1061626:1062009 [6] NCCL INFO comm 0x882e0a0 rank 14 nranks 16 cudaDev 6 nvmlDev 6 busId b9000 commId 0xe69d592a22737672 - Init START
+ip-26-0-166-36:1061625:1062010 [5] NCCL INFO comm 0x9273220 rank 13 nranks 16 cudaDev 5 nvmlDev 5 busId a8000 commId 0xe69d592a22737672 - Init START
+ip-26-0-166-36:1061624:1062011 [4] NCCL INFO comm 0xa0ed2c0 rank 12 nranks 16 cudaDev 4 nvmlDev 4 busId 97000 commId 0xe69d592a22737672 - Init START
+ip-26-0-166-36:1061620:1062015 [0] NCCL INFO comm 0x88c7870 rank 8 nranks 16 cudaDev 0 nvmlDev 0 busId 53000 commId 0xe69d592a22737672 - Init START
+ip-26-0-166-36:1061621:1062014 [1] NCCL INFO comm 0x9584b90 rank 9 nranks 16 cudaDev 1 nvmlDev 1 busId 64000 commId 0xe69d592a22737672 - Init START
+ip-26-0-166-36:1061622:1062012 [2] NCCL INFO comm 0x90c0070 rank 10 nranks 16 cudaDev 2 nvmlDev 2 busId 75000 commId 0xe69d592a22737672 - Init START
+ip-26-0-166-36:1061623:1062013 [3] NCCL INFO comm 0x867fc00 rank 11 nranks 16 cudaDev 3 nvmlDev 3 busId 86000 commId 0xe69d592a22737672 - Init START
+ip-26-0-166-125:1332484:1332881 [2] NCCL INFO comm 0x9ad90a0 rank 2 nranks 16 cudaDev 2 nvmlDev 2 busId 75000 commId 0xe69d592a22737672 - Init START
+ip-26-0-166-125:1332483:1332880 [1] NCCL INFO comm 0x93b0df0 rank 1 nranks 16 cudaDev 1 nvmlDev 1 busId 64000 commId 0xe69d592a22737672 - Init START
+ip-26-0-166-125:1332485:1332882 [3] NCCL INFO comm 0x8851900 rank 3 nranks 16 cudaDev 3 nvmlDev 3 busId 86000 commId 0xe69d592a22737672 - Init START
+ip-26-0-166-125:1332486:1332885 [4] NCCL INFO comm 0x9172d10 rank 4 nranks 16 cudaDev 4 nvmlDev 4 busId 97000 commId 0xe69d592a22737672 - Init START
+ip-26-0-166-125:1332489:1332886 [7] NCCL INFO comm 0x9f6b3b0 rank 7 nranks 16 cudaDev 7 nvmlDev 7 busId ca000 commId 0xe69d592a22737672 - Init START
+ip-26-0-166-125:1332482:1332879 [0] NCCL INFO comm 0x98ae350 rank 0 nranks 16 cudaDev 0 nvmlDev 0 busId 53000 commId 0xe69d592a22737672 - Init START
+ip-26-0-166-125:1332487:1332884 [5] NCCL INFO comm 0x9ae5650 rank 5 nranks 16 cudaDev 5 nvmlDev 5 busId a8000 commId 0xe69d592a22737672 - Init START
+ip-26-0-166-125:1332488:1332883 [6] NCCL INFO comm 0x88b8b90 rank 6 nranks 16 cudaDev 6 nvmlDev 6 busId b9000 commId 0xe69d592a22737672 - Init START
+ip-26-0-166-36:1061625:1062010 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-166-36:1061626:1062009 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-166-125:1332489:1332886 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-166-125:1332485:1332882 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-166-125:1332487:1332884 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-166-125:1332486:1332885 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-166-36:1061627:1062008 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-166-36:1061624:1062011 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-166-36:1061622:1062012 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-166-36:1061621:1062014 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-166-36:1061623:1062013 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-166-36:1061620:1062015 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-166-125:1332483:1332880 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-166-125:1332482:1332879 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/205
+ip-26-0-166-125:1332488:1332883 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-166-125:1332484:1332881 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-166-125:1332486:1332885 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,fff00000,00000000
+ip-26-0-166-125:1332483:1332880 [1] NCCL INFO Setting affinity for GPU 1 to 0fff,ffffffff
+ip-26-0-166-125:1332482:1332879 [0] NCCL INFO Setting affinity for GPU 0 to 0fff,ffffffff
+ip-26-0-166-125:1332488:1332883 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,fff00000,00000000
+ip-26-0-166-125:1332489:1332886 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,fff00000,00000000
+ip-26-0-166-36:1061626:1062009 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-166-125:1332485:1332882 [3] NCCL INFO Setting affinity for GPU 3 to 0fff,ffffffff
+ip-26-0-166-125:1332484:1332881 [2] NCCL INFO Setting affinity for GPU 2 to 0fff,ffffffff
+ip-26-0-166-125:1332487:1332884 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,fff00000,00000000
+ip-26-0-166-36:1061621:1062014 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-166-36:1061620:1062015 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-166-36:1061627:1062008 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-166-36:1061623:1062013 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-166-36:1061625:1062010 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-166-36:1061622:1062012 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-166-36:1061624:1062011 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-166-36:1061624:1062011 [4] NCCL INFO Trees [0] 10/14/-1->12->8 [1] 10/14/-1->12->8 [2] -1/-1/-1->12->13 [3] -1/-1/-1->12->13
+ip-26-0-166-36:1061624:1062011 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-36:1061625:1062010 [5] NCCL INFO Trees [0] -1/-1/-1->13->14 [1] -1/-1/-1->13->14 [2] 14/12/-1->13->11 [3] 14/12/-1->13->11
+ip-26-0-166-36:1061626:1062009 [6] NCCL INFO Trees [0] 13/15/-1->14->12 [1] 13/15/-1->14->12 [2] -1/-1/-1->14->13 [3] -1/-1/-1->14->13
+ip-26-0-166-36:1061625:1062010 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-36:1061626:1062009 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-36:1061627:1062008 [7] NCCL INFO Trees [0] -1/-1/-1->15->14 [1] -1/-1/-1->15->14 [2] 7/-1/-1->15->-1 [3] 7/-1/-1->15->-1
+ip-26-0-166-36:1061627:1062008 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-36:1061623:1062013 [3] NCCL INFO Trees [0] -1/-1/-1->11->10 [1] -1/-1/-1->11->10 [2] 13/9/-1->11->7 [3] 13/9/-1->11->7
+ip-26-0-166-36:1061623:1062013 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-125:1332489:1332886 [7] NCCL INFO Trees [0] -1/-1/-1->7->6 [1] -1/-1/-1->7->6 [2] 11/3/-1->7->15 [3] 11/3/-1->7->15
+ip-26-0-166-125:1332488:1332883 [6] NCCL INFO Trees [0] 5/7/-1->6->4 [1] 5/7/-1->6->4 [2] -1/-1/-1->6->5 [3] -1/-1/-1->6->5
+ip-26-0-166-125:1332489:1332886 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-125:1332488:1332883 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-36:1061622:1062012 [2] NCCL INFO Trees [0] 9/11/-1->10->12 [1] 9/11/-1->10->12 [2] -1/-1/-1->10->9 [3] -1/-1/-1->10->9
+ip-26-0-166-36:1061622:1062012 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-36:1061621:1062014 [1] NCCL INFO Trees [0] -1/-1/-1->9->10 [1] -1/-1/-1->9->10 [2] 10/8/-1->9->11 [3] 10/8/-1->9->11
+ip-26-0-166-36:1061621:1062014 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-125:1332484:1332881 [2] NCCL INFO Trees [0] 1/3/-1->2->4 [1] 1/3/-1->2->4 [2] -1/-1/-1->2->1 [3] -1/-1/-1->2->1
+ip-26-0-166-36:1061620:1062015 [0] NCCL INFO Trees [0] 4/12/-1->8->0 [1] 4/12/-1->8->0 [2] -1/-1/-1->8->9 [3] -1/-1/-1->8->9
+ip-26-0-166-36:1061620:1062015 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-125:1332484:1332881 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-125:1332485:1332882 [3] NCCL INFO Trees [0] -1/-1/-1->3->2 [1] -1/-1/-1->3->2 [2] 5/1/-1->3->7 [3] 5/1/-1->3->7
+ip-26-0-166-125:1332485:1332882 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-125:1332486:1332885 [4] NCCL INFO Trees [0] 2/6/-1->4->8 [1] 2/6/-1->4->8 [2] -1/-1/-1->4->5 [3] -1/-1/-1->4->5
+ip-26-0-166-125:1332483:1332880 [1] NCCL INFO Trees [0] -1/-1/-1->1->2 [1] -1/-1/-1->1->2 [2] 2/0/-1->1->3 [3] 2/0/-1->1->3
+ip-26-0-166-125:1332486:1332885 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-125:1332487:1332884 [5] NCCL INFO Trees [0] -1/-1/-1->5->6 [1] -1/-1/-1->5->6 [2] 6/4/-1->5->3 [3] 6/4/-1->5->3
+ip-26-0-166-125:1332483:1332880 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-125:1332487:1332884 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-125:1332482:1332879 [0] NCCL INFO Channel 00/04 :    0   1   2   3   4   5   6   7   8   9  10  11  12  13  14  15
+ip-26-0-166-125:1332482:1332879 [0] NCCL INFO Channel 01/04 :    0   1   2   3   4   5   6   7   8   9  10  11  12  13  14  15
+ip-26-0-166-125:1332482:1332879 [0] NCCL INFO Channel 02/04 :    0   1   2   3   4   5   6   7   8   9  10  11  12  13  14  15
+ip-26-0-166-125:1332482:1332879 [0] NCCL INFO Channel 03/04 :    0   1   2   3   4   5   6   7   8   9  10  11  12  13  14  15
+ip-26-0-166-125:1332482:1332879 [0] NCCL INFO Trees [0] 8/-1/-1->0->-1 [1] 8/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-166-125:1332482:1332879 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-36:1061623:1062013 [3] NCCL INFO Channel 00/0 : 10[2] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:1061623:1062013 [3] NCCL INFO Channel 01/0 : 10[2] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:1061623:1062013 [3] NCCL INFO Channel 02/0 : 10[2] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:1061622:1062012 [2] NCCL INFO Channel 00/0 : 9[1] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:1061623:1062013 [3] NCCL INFO Channel 03/0 : 10[2] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:1061622:1062012 [2] NCCL INFO Channel 01/0 : 9[1] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:1061623:1062013 [3] NCCL INFO Channel 00/0 : 11[3] -> 12[4] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:1061623:1062013 [3] NCCL INFO Channel 01/0 : 11[3] -> 12[4] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:1061622:1062012 [2] NCCL INFO Channel 02/0 : 9[1] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:1061623:1062013 [3] NCCL INFO Channel 02/0 : 11[3] -> 12[4] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:1061622:1062012 [2] NCCL INFO Channel 03/0 : 9[1] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:1061624:1062011 [4] NCCL INFO Channel 00/0 : 11[3] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:1061623:1062013 [3] NCCL INFO Channel 03/0 : 11[3] -> 12[4] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:1061622:1062012 [2] NCCL INFO Channel 00/0 : 10[2] -> 11[3] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:1061622:1062012 [2] NCCL INFO Channel 01/0 : 10[2] -> 11[3] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:1061624:1062011 [4] NCCL INFO Channel 01/0 : 11[3] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:1061622:1062012 [2] NCCL INFO Channel 02/0 : 10[2] -> 11[3] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:1061624:1062011 [4] NCCL INFO Channel 02/0 : 11[3] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:1061622:1062012 [2] NCCL INFO Channel 03/0 : 10[2] -> 11[3] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:1061624:1062011 [4] NCCL INFO Channel 03/0 : 11[3] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:1061624:1062011 [4] NCCL INFO Channel 00/0 : 12[4] -> 13[5] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:1061624:1062011 [4] NCCL INFO Channel 01/0 : 12[4] -> 13[5] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:1061621:1062014 [1] NCCL INFO Channel 00/0 : 8[0] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:1061624:1062011 [4] NCCL INFO Channel 02/0 : 12[4] -> 13[5] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:1061621:1062014 [1] NCCL INFO Channel 01/0 : 8[0] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:1061624:1062011 [4] NCCL INFO Channel 03/0 : 12[4] -> 13[5] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:1061621:1062014 [1] NCCL INFO Channel 02/0 : 8[0] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:1061621:1062014 [1] NCCL INFO Channel 03/0 : 8[0] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:1061621:1062014 [1] NCCL INFO Channel 00/0 : 9[1] -> 10[2] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:1061621:1062014 [1] NCCL INFO Channel 01/0 : 9[1] -> 10[2] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:1061627:1062008 [7] NCCL INFO Channel 00/0 : 14[6] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:1061625:1062010 [5] NCCL INFO Channel 00/0 : 12[4] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:1061621:1062014 [1] NCCL INFO Channel 02/0 : 9[1] -> 10[2] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:1061621:1062014 [1] NCCL INFO Channel 03/0 : 9[1] -> 10[2] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:1332485:1332882 [3] NCCL INFO Channel 00/0 : 2[2] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:1061627:1062008 [7] NCCL INFO Channel 01/0 : 14[6] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:1332485:1332882 [3] NCCL INFO Channel 01/0 : 2[2] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:1061625:1062010 [5] NCCL INFO Channel 01/0 : 12[4] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:1061627:1062008 [7] NCCL INFO Channel 02/0 : 14[6] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:1061625:1062010 [5] NCCL INFO Channel 02/0 : 12[4] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:1332485:1332882 [3] NCCL INFO Channel 02/0 : 2[2] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:1061627:1062008 [7] NCCL INFO Channel 03/0 : 14[6] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:1061625:1062010 [5] NCCL INFO Channel 03/0 : 12[4] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:1061627:1062008 [7] NCCL INFO Channel 00/0 : 15[7] -> 0[0] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:1332485:1332882 [3] NCCL INFO Channel 03/0 : 2[2] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:1061625:1062010 [5] NCCL INFO Channel 00/0 : 13[5] -> 14[6] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:1332485:1332882 [3] NCCL INFO Channel 00/0 : 3[3] -> 4[4] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:1061627:1062008 [7] NCCL INFO Channel 01/0 : 15[7] -> 0[0] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:1061625:1062010 [5] NCCL INFO Channel 01/0 : 13[5] -> 14[6] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:1332485:1332882 [3] NCCL INFO Channel 01/0 : 3[3] -> 4[4] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:1061627:1062008 [7] NCCL INFO Channel 02/0 : 15[7] -> 0[0] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:1061625:1062010 [5] NCCL INFO Channel 02/0 : 13[5] -> 14[6] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:1332485:1332882 [3] NCCL INFO Channel 02/0 : 3[3] -> 4[4] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:1061626:1062009 [6] NCCL INFO Channel 00/0 : 13[5] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:1061627:1062008 [7] NCCL INFO Channel 03/0 : 15[7] -> 0[0] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:1332485:1332882 [3] NCCL INFO Channel 03/0 : 3[3] -> 4[4] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:1061620:1062015 [0] NCCL INFO Channel 00/0 : 7[7] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:1061625:1062010 [5] NCCL INFO Channel 03/0 : 13[5] -> 14[6] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:1061626:1062009 [6] NCCL INFO Channel 01/0 : 13[5] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:1332483:1332880 [1] NCCL INFO Channel 00/0 : 0[0] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:1061620:1062015 [0] NCCL INFO Channel 01/0 : 7[7] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:1061626:1062009 [6] NCCL INFO Channel 02/0 : 13[5] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:1061620:1062015 [0] NCCL INFO Channel 02/0 : 7[7] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:1332483:1332880 [1] NCCL INFO Channel 01/0 : 0[0] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:1061626:1062009 [6] NCCL INFO Channel 03/0 : 13[5] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:1061620:1062015 [0] NCCL INFO Channel 03/0 : 7[7] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:1332483:1332880 [1] NCCL INFO Channel 02/0 : 0[0] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:1061626:1062009 [6] NCCL INFO Channel 00/0 : 14[6] -> 15[7] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:1061620:1062015 [0] NCCL INFO Channel 00/0 : 8[0] -> 9[1] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:1332483:1332880 [1] NCCL INFO Channel 03/0 : 0[0] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:1061626:1062009 [6] NCCL INFO Channel 01/0 : 14[6] -> 15[7] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:1061620:1062015 [0] NCCL INFO Channel 01/0 : 8[0] -> 9[1] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:1332483:1332880 [1] NCCL INFO Channel 00/0 : 1[1] -> 2[2] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:1061626:1062009 [6] NCCL INFO Channel 02/0 : 14[6] -> 15[7] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:1061620:1062015 [0] NCCL INFO Channel 02/0 : 8[0] -> 9[1] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:1332484:1332881 [2] NCCL INFO Channel 00/0 : 1[1] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:1332483:1332880 [1] NCCL INFO Channel 01/0 : 1[1] -> 2[2] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:1332483:1332880 [1] NCCL INFO Channel 02/0 : 1[1] -> 2[2] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:1061626:1062009 [6] NCCL INFO Channel 03/0 : 14[6] -> 15[7] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:1061620:1062015 [0] NCCL INFO Channel 03/0 : 8[0] -> 9[1] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:1332484:1332881 [2] NCCL INFO Channel 01/0 : 1[1] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:1332483:1332880 [1] NCCL INFO Channel 03/0 : 1[1] -> 2[2] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:1332484:1332881 [2] NCCL INFO Channel 02/0 : 1[1] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:1332484:1332881 [2] NCCL INFO Channel 03/0 : 1[1] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:1332488:1332883 [6] NCCL INFO Channel 00/0 : 5[5] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:1332484:1332881 [2] NCCL INFO Channel 00/0 : 2[2] -> 3[3] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:1332484:1332881 [2] NCCL INFO Channel 01/0 : 2[2] -> 3[3] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:1332488:1332883 [6] NCCL INFO Channel 01/0 : 5[5] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:1332486:1332885 [4] NCCL INFO Channel 00/0 : 3[3] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:1332484:1332881 [2] NCCL INFO Channel 02/0 : 2[2] -> 3[3] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:1332487:1332884 [5] NCCL INFO Channel 00/0 : 4[4] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:1332488:1332883 [6] NCCL INFO Channel 02/0 : 5[5] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:1332484:1332881 [2] NCCL INFO Channel 03/0 : 2[2] -> 3[3] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:1332488:1332883 [6] NCCL INFO Channel 03/0 : 5[5] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:1332486:1332885 [4] NCCL INFO Channel 01/0 : 3[3] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:1332487:1332884 [5] NCCL INFO Channel 01/0 : 4[4] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:1332488:1332883 [6] NCCL INFO Channel 00/0 : 6[6] -> 7[7] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:1332486:1332885 [4] NCCL INFO Channel 02/0 : 3[3] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:1332487:1332884 [5] NCCL INFO Channel 02/0 : 4[4] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:1332488:1332883 [6] NCCL INFO Channel 01/0 : 6[6] -> 7[7] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:1332486:1332885 [4] NCCL INFO Channel 03/0 : 3[3] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:1332487:1332884 [5] NCCL INFO Channel 03/0 : 4[4] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:1332488:1332883 [6] NCCL INFO Channel 02/0 : 6[6] -> 7[7] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:1332486:1332885 [4] NCCL INFO Channel 00/0 : 4[4] -> 5[5] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:1332487:1332884 [5] NCCL INFO Channel 00/0 : 5[5] -> 6[6] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:1332488:1332883 [6] NCCL INFO Channel 03/0 : 6[6] -> 7[7] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:1332486:1332885 [4] NCCL INFO Channel 01/0 : 4[4] -> 5[5] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:1332489:1332886 [7] NCCL INFO Channel 00/0 : 6[6] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:1332487:1332884 [5] NCCL INFO Channel 01/0 : 5[5] -> 6[6] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:1332486:1332885 [4] NCCL INFO Channel 02/0 : 4[4] -> 5[5] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:1332487:1332884 [5] NCCL INFO Channel 02/0 : 5[5] -> 6[6] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:1332486:1332885 [4] NCCL INFO Channel 03/0 : 4[4] -> 5[5] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:1332487:1332884 [5] NCCL INFO Channel 03/0 : 5[5] -> 6[6] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:1332489:1332886 [7] NCCL INFO Channel 01/0 : 6[6] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:1332489:1332886 [7] NCCL INFO Channel 02/0 : 6[6] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:1332489:1332886 [7] NCCL INFO Channel 03/0 : 6[6] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:1332489:1332886 [7] NCCL INFO Channel 00/0 : 7[7] -> 8[0] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:1332489:1332886 [7] NCCL INFO Channel 01/0 : 7[7] -> 8[0] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:1332489:1332886 [7] NCCL INFO Channel 02/0 : 7[7] -> 8[0] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:1332489:1332886 [7] NCCL INFO Channel 03/0 : 7[7] -> 8[0] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:1332482:1332879 [0] NCCL INFO Channel 00/0 : 15[7] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:1332482:1332879 [0] NCCL INFO Channel 01/0 : 15[7] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:1332482:1332879 [0] NCCL INFO Channel 02/0 : 15[7] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:1332482:1332879 [0] NCCL INFO Channel 03/0 : 15[7] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:1332482:1332879 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[1] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:1332482:1332879 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[1] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:1332482:1332879 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[1] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:1332482:1332879 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[1] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:1061623:1062013 [3] NCCL INFO Connected all rings
+ip-26-0-166-36:1061623:1062013 [3] NCCL INFO Channel 02/0 : 9[1] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:1061623:1062013 [3] NCCL INFO Channel 03/0 : 9[1] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:1061623:1062013 [3] NCCL INFO Channel 02/0 : 11[3] -> 13[5] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:1061623:1062013 [3] NCCL INFO Channel 03/0 : 11[3] -> 13[5] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:1061622:1062012 [2] NCCL INFO Connected all rings
+ip-26-0-166-36:1061622:1062012 [2] NCCL INFO Channel 00/0 : 10[2] -> 12[4] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:1061622:1062012 [2] NCCL INFO Channel 01/0 : 10[2] -> 12[4] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:1061621:1062014 [1] NCCL INFO Connected all rings
+ip-26-0-166-36:1061621:1062014 [1] NCCL INFO Channel 02/0 : 9[1] -> 11[3] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:1061621:1062014 [1] NCCL INFO Channel 03/0 : 9[1] -> 11[3] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:1061621:1062014 [1] NCCL INFO Channel 02/0 : 11[3] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:1061621:1062014 [1] NCCL INFO Channel 03/0 : 11[3] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:1061626:1062009 [6] NCCL INFO Connected all rings
+ip-26-0-166-36:1061626:1062009 [6] NCCL INFO Channel 00/0 : 12[4] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:1061626:1062009 [6] NCCL INFO Channel 01/0 : 12[4] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:1061625:1062010 [5] NCCL INFO Connected all rings
+ip-26-0-166-36:1061625:1062010 [5] NCCL INFO Channel 02/0 : 11[3] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:1061625:1062010 [5] NCCL INFO Channel 03/0 : 11[3] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:1061625:1062010 [5] NCCL INFO Channel 02/0 : 13[5] -> 11[3] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:1061623:1062013 [3] NCCL INFO Channel 02/0 : 7[7] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:1061625:1062010 [5] NCCL INFO Channel 03/0 : 13[5] -> 11[3] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:1061623:1062013 [3] NCCL INFO Channel 03/0 : 7[7] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:1332489:1332886 [7] NCCL INFO Connected all rings
+ip-26-0-166-125:1332489:1332886 [7] NCCL INFO Channel 02/0 : 3[3] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:1332489:1332886 [7] NCCL INFO Channel 03/0 : 3[3] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:1332489:1332886 [7] NCCL INFO Channel 02/0 : 7[7] -> 11[3] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:1332489:1332886 [7] NCCL INFO Channel 03/0 : 7[7] -> 11[3] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:1061624:1062011 [4] NCCL INFO Connected all rings
+ip-26-0-166-36:1061624:1062011 [4] NCCL INFO Channel 00/0 : 10[2] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:1061623:1062013 [3] NCCL INFO Channel 02/0 : 11[3] -> 7[7] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:1061624:1062011 [4] NCCL INFO Channel 01/0 : 10[2] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:1061623:1062013 [3] NCCL INFO Channel 03/0 : 11[3] -> 7[7] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:1061624:1062011 [4] NCCL INFO Channel 00/0 : 12[4] -> 14[6] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:1061624:1062011 [4] NCCL INFO Channel 01/0 : 12[4] -> 14[6] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:1332487:1332884 [5] NCCL INFO Connected all rings
+ip-26-0-166-36:1061622:1062012 [2] NCCL INFO Channel 00/0 : 12[4] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:1332487:1332884 [5] NCCL INFO Channel 02/0 : 3[3] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:1061626:1062009 [6] NCCL INFO Channel 00/0 : 14[6] -> 12[4] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:1061622:1062012 [2] NCCL INFO Channel 01/0 : 12[4] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:1061624:1062011 [4] NCCL INFO Channel 00/0 : 8[0] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:1332487:1332884 [5] NCCL INFO Channel 03/0 : 3[3] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:1061626:1062009 [6] NCCL INFO Channel 01/0 : 14[6] -> 12[4] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:1061624:1062011 [4] NCCL INFO Channel 01/0 : 8[0] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:1332488:1332883 [6] NCCL INFO Connected all rings
+ip-26-0-166-125:1332488:1332883 [6] NCCL INFO Channel 00/0 : 4[4] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:1332488:1332883 [6] NCCL INFO Channel 01/0 : 4[4] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:1061620:1062015 [0] NCCL INFO Connected all rings
+ip-26-0-166-36:1061620:1062015 [0] NCCL INFO Channel 00/0 : 4[4] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:1061620:1062015 [0] NCCL INFO Channel 01/0 : 4[4] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:1061620:1062015 [0] NCCL INFO Channel 00/0 : 8[0] -> 12[4] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:1061620:1062015 [0] NCCL INFO Channel 01/0 : 8[0] -> 12[4] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:1061624:1062011 [4] NCCL INFO Channel 00/0 : 12[4] -> 8[0] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:1061624:1062011 [4] NCCL INFO Channel 01/0 : 12[4] -> 8[0] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:1332484:1332881 [2] NCCL INFO Connected all rings
+ip-26-0-166-125:1332484:1332881 [2] NCCL INFO Channel 00/0 : 2[2] -> 4[4] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:1332484:1332881 [2] NCCL INFO Channel 01/0 : 2[2] -> 4[4] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:1332485:1332882 [3] NCCL INFO Connected all rings
+ip-26-0-166-125:1332485:1332882 [3] NCCL INFO Channel 02/0 : 1[1] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:1332485:1332882 [3] NCCL INFO Channel 03/0 : 1[1] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:1332485:1332882 [3] NCCL INFO Channel 02/0 : 3[3] -> 5[5] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:1332485:1332882 [3] NCCL INFO Channel 03/0 : 3[3] -> 5[5] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:1332487:1332884 [5] NCCL INFO Channel 02/0 : 5[5] -> 3[3] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:1332487:1332884 [5] NCCL INFO Channel 03/0 : 5[5] -> 3[3] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:1332486:1332885 [4] NCCL INFO Connected all rings
+ip-26-0-166-125:1332486:1332885 [4] NCCL INFO Channel 00/0 : 2[2] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:1332486:1332885 [4] NCCL INFO Channel 01/0 : 2[2] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:1332486:1332885 [4] NCCL INFO Channel 00/0 : 4[4] -> 6[6] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:1332486:1332885 [4] NCCL INFO Channel 01/0 : 4[4] -> 6[6] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:1332484:1332881 [2] NCCL INFO Channel 00/0 : 4[4] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:1332484:1332881 [2] NCCL INFO Channel 01/0 : 4[4] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:1332488:1332883 [6] NCCL INFO Channel 00/0 : 6[6] -> 4[4] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:1332486:1332885 [4] NCCL INFO Channel 00/0 : 4[4] -> 8[0] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:1332488:1332883 [6] NCCL INFO Channel 01/0 : 6[6] -> 4[4] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:1332486:1332885 [4] NCCL INFO Channel 01/0 : 4[4] -> 8[0] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:1332486:1332885 [4] NCCL INFO Channel 00/0 : 8[0] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:1061620:1062015 [0] NCCL INFO Channel 00/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:1332486:1332885 [4] NCCL INFO Channel 01/0 : 8[0] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:1061620:1062015 [0] NCCL INFO Channel 01/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:1061620:1062015 [0] NCCL INFO Channel 00/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:1061620:1062015 [0] NCCL INFO Channel 01/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:1061627:1062008 [7] NCCL INFO Connected all rings
+ip-26-0-166-36:1061627:1062008 [7] NCCL INFO Channel 02/0 : 7[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:1061627:1062008 [7] NCCL INFO Channel 03/0 : 7[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:1061627:1062008 [7] NCCL INFO Channel 02/0 : 15[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:1061627:1062008 [7] NCCL INFO Channel 03/0 : 15[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:1332483:1332880 [1] NCCL INFO Connected all rings
+ip-26-0-166-125:1332483:1332880 [1] NCCL INFO Channel 02/0 : 1[1] -> 3[3] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:1332483:1332880 [1] NCCL INFO Channel 03/0 : 1[1] -> 3[3] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:1332485:1332882 [3] NCCL INFO Channel 02/0 : 3[3] -> 7[7] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:1332483:1332880 [1] NCCL INFO Channel 02/0 : 3[3] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:1332485:1332882 [3] NCCL INFO Channel 03/0 : 3[3] -> 7[7] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:1332483:1332880 [1] NCCL INFO Channel 03/0 : 3[3] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:1332485:1332882 [3] NCCL INFO Channel 02/0 : 7[7] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:1332489:1332886 [7] NCCL INFO Channel 02/0 : 15[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:1332485:1332882 [3] NCCL INFO Channel 03/0 : 7[7] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:1332489:1332886 [7] NCCL INFO Channel 03/0 : 15[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:1332489:1332886 [7] NCCL INFO Channel 02/0 : 7[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:1332489:1332886 [7] NCCL INFO Channel 03/0 : 7[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:1332489:1332886 [7] NCCL INFO Channel 02/0 : 11[3] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:1061627:1062008 [7] NCCL INFO Channel 00/0 : 15[7] -> 14[6] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:1061627:1062008 [7] NCCL INFO Channel 01/0 : 15[7] -> 14[6] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:1332489:1332886 [7] NCCL INFO Channel 03/0 : 11[3] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:1332489:1332886 [7] NCCL INFO Channel 02/0 : 7[7] -> 3[3] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:1332489:1332886 [7] NCCL INFO Channel 03/0 : 7[7] -> 3[3] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:1061623:1062013 [3] NCCL INFO Channel 02/0 : 13[5] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:1332485:1332882 [3] NCCL INFO Channel 02/0 : 5[5] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:1061623:1062013 [3] NCCL INFO Channel 03/0 : 13[5] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:1332489:1332886 [7] NCCL INFO Channel 00/0 : 7[7] -> 6[6] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:1061623:1062013 [3] NCCL INFO Channel 02/0 : 11[3] -> 9[1] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:1332485:1332882 [3] NCCL INFO Channel 03/0 : 5[5] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:1061623:1062013 [3] NCCL INFO Channel 03/0 : 11[3] -> 9[1] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:1332489:1332886 [7] NCCL INFO Channel 01/0 : 7[7] -> 6[6] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:1332485:1332882 [3] NCCL INFO Channel 02/0 : 3[3] -> 1[1] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:1332485:1332882 [3] NCCL INFO Channel 03/0 : 3[3] -> 1[1] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:1061625:1062010 [5] NCCL INFO Channel 00/0 : 14[6] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:1332482:1332879 [0] NCCL INFO Connected all rings
+ip-26-0-166-125:1332487:1332884 [5] NCCL INFO Channel 00/0 : 6[6] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:1332483:1332880 [1] NCCL INFO Channel 00/0 : 2[2] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:1061621:1062014 [1] NCCL INFO Channel 00/0 : 10[2] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:1061625:1062010 [5] NCCL INFO Channel 01/0 : 14[6] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:1061623:1062013 [3] NCCL INFO Channel 00/0 : 11[3] -> 10[2] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:1332487:1332884 [5] NCCL INFO Channel 01/0 : 6[6] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:1332485:1332882 [3] NCCL INFO Channel 00/0 : 3[3] -> 2[2] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:1332482:1332879 [0] NCCL INFO Channel 00/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:1332483:1332880 [1] NCCL INFO Channel 01/0 : 2[2] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:1061621:1062014 [1] NCCL INFO Channel 01/0 : 10[2] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:1061625:1062010 [5] NCCL INFO Channel 02/0 : 14[6] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:1061623:1062013 [3] NCCL INFO Channel 01/0 : 11[3] -> 10[2] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:1332487:1332884 [5] NCCL INFO Channel 02/0 : 6[6] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:1332485:1332882 [3] NCCL INFO Channel 01/0 : 3[3] -> 2[2] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:1332483:1332880 [1] NCCL INFO Channel 02/0 : 2[2] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:1332482:1332879 [0] NCCL INFO Channel 01/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:1061625:1062010 [5] NCCL INFO Channel 03/0 : 14[6] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:1061621:1062014 [1] NCCL INFO Channel 02/0 : 10[2] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:1332487:1332884 [5] NCCL INFO Channel 03/0 : 6[6] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:1061625:1062010 [5] NCCL INFO Channel 02/0 : 13[5] -> 12[4] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:1332483:1332880 [1] NCCL INFO Channel 03/0 : 2[2] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:1332482:1332879 [0] NCCL INFO Channel 00/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:1332487:1332884 [5] NCCL INFO Channel 02/0 : 5[5] -> 4[4] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:1061621:1062014 [1] NCCL INFO Channel 03/0 : 10[2] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:1061625:1062010 [5] NCCL INFO Channel 03/0 : 13[5] -> 12[4] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:1332482:1332879 [0] NCCL INFO Channel 01/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:1332483:1332880 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[0] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:1332487:1332884 [5] NCCL INFO Channel 03/0 : 5[5] -> 4[4] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:1061621:1062014 [1] NCCL INFO Channel 02/0 : 9[1] -> 8[0] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:1332483:1332880 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[0] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:1061621:1062014 [1] NCCL INFO Channel 03/0 : 9[1] -> 8[0] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:1332482:1332879 [0] NCCL INFO Channel 02/0 : 1[1] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:1061620:1062015 [0] NCCL INFO Channel 00/0 : 12[4] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:1332482:1332879 [0] NCCL INFO Channel 03/0 : 1[1] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:1061620:1062015 [0] NCCL INFO Channel 01/0 : 12[4] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:1061620:1062015 [0] NCCL INFO Channel 00/0 : 8[0] -> 4[4] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:1061620:1062015 [0] NCCL INFO Channel 01/0 : 8[0] -> 4[4] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:1061624:1062011 [4] NCCL INFO Channel 00/0 : 14[6] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:1061624:1062011 [4] NCCL INFO Channel 01/0 : 14[6] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:1332486:1332885 [4] NCCL INFO Channel 00/0 : 6[6] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:1061624:1062011 [4] NCCL INFO Channel 00/0 : 12[4] -> 10[2] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:1332486:1332885 [4] NCCL INFO Channel 01/0 : 6[6] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:1061624:1062011 [4] NCCL INFO Channel 01/0 : 12[4] -> 10[2] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:1332486:1332885 [4] NCCL INFO Channel 00/0 : 4[4] -> 2[2] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:1332486:1332885 [4] NCCL INFO Channel 01/0 : 4[4] -> 2[2] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:1061626:1062009 [6] NCCL INFO Channel 00/0 : 15[7] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:1061622:1062012 [2] NCCL INFO Channel 00/0 : 11[3] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:1061626:1062009 [6] NCCL INFO Channel 01/0 : 15[7] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:1061624:1062011 [4] NCCL INFO Channel 02/0 : 13[5] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:1332488:1332883 [6] NCCL INFO Channel 00/0 : 7[7] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:1061622:1062012 [2] NCCL INFO Channel 01/0 : 11[3] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:1061626:1062009 [6] NCCL INFO Channel 00/0 : 14[6] -> 13[5] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:1061624:1062011 [4] NCCL INFO Channel 03/0 : 13[5] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:1332484:1332881 [2] NCCL INFO Channel 00/0 : 3[3] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:1061622:1062012 [2] NCCL INFO Channel 00/0 : 10[2] -> 9[1] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:1061626:1062009 [6] NCCL INFO Channel 01/0 : 14[6] -> 13[5] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:1332486:1332885 [4] NCCL INFO Channel 02/0 : 5[5] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:1332488:1332883 [6] NCCL INFO Channel 01/0 : 7[7] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:1061622:1062012 [2] NCCL INFO Channel 01/0 : 10[2] -> 9[1] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:1332484:1332881 [2] NCCL INFO Channel 01/0 : 3[3] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:1332486:1332885 [4] NCCL INFO Channel 03/0 : 5[5] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:1332488:1332883 [6] NCCL INFO Channel 00/0 : 6[6] -> 5[5] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:1061626:1062009 [6] NCCL INFO Channel 02/0 : 14[6] -> 13[5] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:1061622:1062012 [2] NCCL INFO Channel 02/0 : 10[2] -> 9[1] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:1332484:1332881 [2] NCCL INFO Channel 00/0 : 2[2] -> 1[1] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:1061626:1062009 [6] NCCL INFO Channel 03/0 : 14[6] -> 13[5] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:1061622:1062012 [2] NCCL INFO Channel 03/0 : 10[2] -> 9[1] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:1332488:1332883 [6] NCCL INFO Channel 01/0 : 6[6] -> 5[5] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:1332484:1332881 [2] NCCL INFO Channel 01/0 : 2[2] -> 1[1] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:1332488:1332883 [6] NCCL INFO Channel 02/0 : 6[6] -> 5[5] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:1332484:1332881 [2] NCCL INFO Channel 02/0 : 2[2] -> 1[1] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:1061620:1062015 [0] NCCL INFO Channel 02/0 : 9[1] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:1332488:1332883 [6] NCCL INFO Channel 03/0 : 6[6] -> 5[5] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:1332484:1332881 [2] NCCL INFO Channel 03/0 : 2[2] -> 1[1] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:1061620:1062015 [0] NCCL INFO Channel 03/0 : 9[1] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:1332482:1332879 [0] NCCL INFO Connected all trees
+ip-26-0-166-125:1332482:1332879 [0] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+ip-26-0-166-125:1332482:1332879 [0] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-125:1332482:1332879 [0] NCCL INFO comm 0x98ae350 rank 0 nranks 16 cudaDev 0 nvmlDev 0 busId 53000 commId 0xe69d592a22737672 - Init COMPLETE
+ip-26-0-166-125:1332484:1332881 [2] NCCL INFO Connected all trees
+ip-26-0-166-125:1332484:1332881 [2] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+ip-26-0-166-125:1332484:1332881 [2] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-125:1332483:1332880 [1] NCCL INFO Connected all trees
+ip-26-0-166-125:1332483:1332880 [1] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+ip-26-0-166-125:1332483:1332880 [1] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-125:1332484:1332881 [2] NCCL INFO comm 0x9ad90a0 rank 2 nranks 16 cudaDev 2 nvmlDev 2 busId 75000 commId 0xe69d592a22737672 - Init COMPLETE
+ip-26-0-166-125:1332483:1332880 [1] NCCL INFO comm 0x93b0df0 rank 1 nranks 16 cudaDev 1 nvmlDev 1 busId 64000 commId 0xe69d592a22737672 - Init COMPLETE
+ip-26-0-166-36:1061627:1062008 [7] NCCL INFO Connected all trees
+ip-26-0-166-36:1061627:1062008 [7] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+ip-26-0-166-36:1061627:1062008 [7] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-36:1061627:1062008 [7] NCCL INFO comm 0x94b2540 rank 15 nranks 16 cudaDev 7 nvmlDev 7 busId ca000 commId 0xe69d592a22737672 - Init COMPLETE
+ip-26-0-166-125:1332488:1332883 [6] NCCL INFO Connected all trees
+ip-26-0-166-125:1332488:1332883 [6] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+ip-26-0-166-125:1332488:1332883 [6] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-125:1332487:1332884 [5] NCCL INFO Connected all trees
+ip-26-0-166-125:1332487:1332884 [5] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+ip-26-0-166-125:1332487:1332884 [5] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-125:1332485:1332882 [3] NCCL INFO Connected all trees
+ip-26-0-166-125:1332485:1332882 [3] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+ip-26-0-166-125:1332485:1332882 [3] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-36:1061626:1062009 [6] NCCL INFO Connected all trees
+ip-26-0-166-36:1061626:1062009 [6] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+ip-26-0-166-36:1061626:1062009 [6] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-125:1332488:1332883 [6] NCCL INFO comm 0x88b8b90 rank 6 nranks 16 cudaDev 6 nvmlDev 6 busId b9000 commId 0xe69d592a22737672 - Init COMPLETE
+ip-26-0-166-125:1332487:1332884 [5] NCCL INFO comm 0x9ae5650 rank 5 nranks 16 cudaDev 5 nvmlDev 5 busId a8000 commId 0xe69d592a22737672 - Init COMPLETE
+ip-26-0-166-125:1332489:1332886 [7] NCCL INFO Connected all trees
+ip-26-0-166-125:1332489:1332886 [7] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+ip-26-0-166-125:1332489:1332886 [7] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-125:1332485:1332882 [3] NCCL INFO comm 0x8851900 rank 3 nranks 16 cudaDev 3 nvmlDev 3 busId 86000 commId 0xe69d592a22737672 - Init COMPLETE
+ip-26-0-166-36:1061626:1062009 [6] NCCL INFO comm 0x882e0a0 rank 14 nranks 16 cudaDev 6 nvmlDev 6 busId b9000 commId 0xe69d592a22737672 - Init COMPLETE
+ip-26-0-166-36:1061622:1062012 [2] NCCL INFO Connected all trees
+ip-26-0-166-36:1061622:1062012 [2] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+ip-26-0-166-36:1061622:1062012 [2] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-125:1332489:1332886 [7] NCCL INFO comm 0x9f6b3b0 rank 7 nranks 16 cudaDev 7 nvmlDev 7 busId ca000 commId 0xe69d592a22737672 - Init COMPLETE
+ip-26-0-166-125:1332486:1332885 [4] NCCL INFO Connected all trees
+ip-26-0-166-125:1332486:1332885 [4] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+ip-26-0-166-125:1332486:1332885 [4] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-36:1061622:1062012 [2] NCCL INFO comm 0x90c0070 rank 10 nranks 16 cudaDev 2 nvmlDev 2 busId 75000 commId 0xe69d592a22737672 - Init COMPLETE
+ip-26-0-166-125:1332486:1332885 [4] NCCL INFO comm 0x9172d10 rank 4 nranks 16 cudaDev 4 nvmlDev 4 busId 97000 commId 0xe69d592a22737672 - Init COMPLETE
+ip-26-0-166-36:1061625:1062010 [5] NCCL INFO Connected all trees
+ip-26-0-166-36:1061625:1062010 [5] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+ip-26-0-166-36:1061625:1062010 [5] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-36:1061625:1062010 [5] NCCL INFO comm 0x9273220 rank 13 nranks 16 cudaDev 5 nvmlDev 5 busId a8000 commId 0xe69d592a22737672 - Init COMPLETE
+ip-26-0-166-36:1061623:1062013 [3] NCCL INFO Connected all trees
+ip-26-0-166-36:1061623:1062013 [3] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+ip-26-0-166-36:1061623:1062013 [3] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-36:1061624:1062011 [4] NCCL INFO Connected all trees
+ip-26-0-166-36:1061624:1062011 [4] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+ip-26-0-166-36:1061624:1062011 [4] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-36:1061623:1062013 [3] NCCL INFO comm 0x867fc00 rank 11 nranks 16 cudaDev 3 nvmlDev 3 busId 86000 commId 0xe69d592a22737672 - Init COMPLETE
+ip-26-0-166-36:1061621:1062014 [1] NCCL INFO Connected all trees
+ip-26-0-166-36:1061621:1062014 [1] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+ip-26-0-166-36:1061621:1062014 [1] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-36:1061624:1062011 [4] NCCL INFO comm 0xa0ed2c0 rank 12 nranks 16 cudaDev 4 nvmlDev 4 busId 97000 commId 0xe69d592a22737672 - Init COMPLETE
+ip-26-0-166-36:1061621:1062014 [1] NCCL INFO comm 0x9584b90 rank 9 nranks 16 cudaDev 1 nvmlDev 1 busId 64000 commId 0xe69d592a22737672 - Init COMPLETE
+ip-26-0-166-36:1061620:1062015 [0] NCCL INFO Connected all trees
+ip-26-0-166-36:1061620:1062015 [0] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+ip-26-0-166-36:1061620:1062015 [0] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-36:1061620:1062015 [0] NCCL INFO comm 0x88c7870 rank 8 nranks 16 cudaDev 0 nvmlDev 0 busId 53000 commId 0xe69d592a22737672 - Init COMPLETE
+NCCL version 2.18.5+cuda12.2
+ip-26-0-166-36:1061620:1062042 [0] NCCL INFO Using network Libfabric
+ip-26-0-166-36:1061624:1062046 [4] NCCL INFO Using network Libfabric
+ip-26-0-166-36:1061623:1062044 [3] NCCL INFO Using network Libfabric
+ip-26-0-166-36:1061622:1062049 [2] NCCL INFO Using network Libfabric
+ip-26-0-166-36:1061626:1062043 [6] NCCL INFO Using network Libfabric
+ip-26-0-166-36:1061621:1062045 [1] NCCL INFO Using network Libfabric
+ip-26-0-166-36:1061625:1062047 [5] NCCL INFO Using network Libfabric
+ip-26-0-166-36:1061627:1062048 [7] NCCL INFO Using network Libfabric
+ip-26-0-166-125:1332482:1332927 [0] NCCL INFO Using network Libfabric
+ip-26-0-166-125:1332489:1332932 [7] NCCL INFO Using network Libfabric
+ip-26-0-166-125:1332488:1332929 [6] NCCL INFO Using network Libfabric
+ip-26-0-166-125:1332485:1332933 [3] NCCL INFO Using network Libfabric
+ip-26-0-166-125:1332484:1332934 [2] NCCL INFO Using network Libfabric
+ip-26-0-166-125:1332483:1332931 [1] NCCL INFO Using network Libfabric
+ip-26-0-166-125:1332486:1332928 [4] NCCL INFO Using network Libfabric
+ip-26-0-166-125:1332487:1332930 [5] NCCL INFO Using network Libfabric
+ip-26-0-166-36:1061620:1062042 [0] NCCL INFO comm 0x88dd7c0 rank 0 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0x3dbd315886848384 - Init START
+ip-26-0-166-36:1061627:1062048 [7] NCCL INFO comm 0x94c63c0 rank 7 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0x3dbd315886848384 - Init START
+ip-26-0-166-36:1061621:1062045 [1] NCCL INFO comm 0x9598880 rank 1 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0x3dbd315886848384 - Init START
+ip-26-0-166-36:1061622:1062049 [2] NCCL INFO comm 0x90d4310 rank 2 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0x3dbd315886848384 - Init START
+ip-26-0-166-36:1061626:1062043 [6] NCCL INFO comm 0x8842650 rank 6 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0x3dbd315886848384 - Init START
+ip-26-0-166-36:1061623:1062044 [3] NCCL INFO comm 0x8693dc0 rank 3 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0x3dbd315886848384 - Init START
+ip-26-0-166-36:1061624:1062046 [4] NCCL INFO comm 0xa1014c0 rank 4 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0x3dbd315886848384 - Init START
+ip-26-0-166-36:1061625:1062047 [5] NCCL INFO comm 0x9286ea0 rank 5 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0x3dbd315886848384 - Init START
+ip-26-0-166-36:1061620:1062042 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-166-36:1061627:1062048 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-166-36:1061626:1062043 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-166-36:1061621:1062045 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-166-36:1061622:1062049 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-166-36:1061625:1062047 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-166-36:1061624:1062046 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-166-36:1061623:1062044 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-166-125:1332488:1332929 [6] NCCL INFO comm 0x88cce50 rank 6 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0x9221e1aa7583a57b - Init START
+ip-26-0-166-125:1332487:1332930 [5] NCCL INFO comm 0x9af9a60 rank 5 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0x9221e1aa7583a57b - Init START
+ip-26-0-166-125:1332489:1332932 [7] NCCL INFO comm 0x9f7f300 rank 7 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0x9221e1aa7583a57b - Init START
+ip-26-0-166-125:1332486:1332928 [4] NCCL INFO comm 0x9187160 rank 4 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0x9221e1aa7583a57b - Init START
+ip-26-0-166-125:1332482:1332927 [0] NCCL INFO comm 0x98c2990 rank 0 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0x9221e1aa7583a57b - Init START
+ip-26-0-166-125:1332485:1332933 [3] NCCL INFO comm 0x8865ad0 rank 3 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0x9221e1aa7583a57b - Init START
+ip-26-0-166-125:1332483:1332931 [1] NCCL INFO comm 0x93c4f50 rank 1 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0x9221e1aa7583a57b - Init START
+ip-26-0-166-125:1332484:1332934 [2] NCCL INFO comm 0x9aece10 rank 2 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0x9221e1aa7583a57b - Init START
+ip-26-0-166-125:1332482:1332927 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/205
+ip-26-0-166-125:1332485:1332933 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-166-125:1332486:1332928 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-166-125:1332488:1332929 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-166-125:1332487:1332930 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-166-125:1332489:1332932 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-166-125:1332484:1332934 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-166-125:1332483:1332931 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-166-125:1332489:1332932 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,fff00000,00000000
+ip-26-0-166-36:1061625:1062047 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-166-36:1061623:1062044 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-166-125:1332486:1332928 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,fff00000,00000000
+ip-26-0-166-125:1332483:1332931 [1] NCCL INFO Setting affinity for GPU 1 to 0fff,ffffffff
+ip-26-0-166-125:1332484:1332934 [2] NCCL INFO Setting affinity for GPU 2 to 0fff,ffffffff
+ip-26-0-166-36:1061622:1062049 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-166-36:1061620:1062042 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-166-36:1061621:1062045 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-166-36:1061626:1062043 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-166-125:1332485:1332933 [3] NCCL INFO Setting affinity for GPU 3 to 0fff,ffffffff
+ip-26-0-166-36:1061624:1062046 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-166-36:1061627:1062048 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-166-36:1061627:1062048 [7] NCCL INFO Trees [0] -1/-1/-1->7->6 [1] -1/-1/-1->7->6 [2] 3/-1/-1->7->-1 [3] 3/-1/-1->7->-1
+ip-26-0-166-36:1061627:1062048 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-36:1061626:1062043 [6] NCCL INFO Trees [0] 5/7/-1->6->4 [1] 5/7/-1->6->4 [2] -1/-1/-1->6->5 [3] -1/-1/-1->6->5
+ip-26-0-166-36:1061625:1062047 [5] NCCL INFO Trees [0] -1/-1/-1->5->6 [1] -1/-1/-1->5->6 [2] 6/4/-1->5->3 [3] 6/4/-1->5->3
+ip-26-0-166-36:1061626:1062043 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-36:1061625:1062047 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-36:1061620:1062042 [0] NCCL INFO Channel 00/04 :    0   1   2   3   4   5   6   7
+ip-26-0-166-36:1061621:1062045 [1] NCCL INFO Trees [0] -1/-1/-1->1->2 [1] -1/-1/-1->1->2 [2] 2/0/-1->1->3 [3] 2/0/-1->1->3
+ip-26-0-166-36:1061620:1062042 [0] NCCL INFO Channel 01/04 :    0   1   2   3   4   5   6   7
+ip-26-0-166-36:1061623:1062044 [3] NCCL INFO Trees [0] -1/-1/-1->3->2 [1] -1/-1/-1->3->2 [2] 5/1/-1->3->7 [3] 5/1/-1->3->7
+ip-26-0-166-36:1061621:1062045 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-36:1061620:1062042 [0] NCCL INFO Channel 02/04 :    0   1   2   3   4   5   6   7
+ip-26-0-166-36:1061624:1062046 [4] NCCL INFO Trees [0] 2/6/-1->4->0 [1] 2/6/-1->4->0 [2] -1/-1/-1->4->5 [3] -1/-1/-1->4->5
+ip-26-0-166-36:1061623:1062044 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-36:1061620:1062042 [0] NCCL INFO Channel 03/04 :    0   1   2   3   4   5   6   7
+ip-26-0-166-36:1061624:1062046 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-36:1061622:1062049 [2] NCCL INFO Trees [0] 1/3/-1->2->4 [1] 1/3/-1->2->4 [2] -1/-1/-1->2->1 [3] -1/-1/-1->2->1
+ip-26-0-166-36:1061620:1062042 [0] NCCL INFO Trees [0] 4/-1/-1->0->-1 [1] 4/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-166-36:1061622:1062049 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-36:1061620:1062042 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-125:1332482:1332927 [0] NCCL INFO Setting affinity for GPU 0 to 0fff,ffffffff
+ip-26-0-166-125:1332488:1332929 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,fff00000,00000000
+ip-26-0-166-125:1332487:1332930 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,fff00000,00000000
+ip-26-0-166-125:1332487:1332930 [5] NCCL INFO Trees [0] -1/-1/-1->5->6 [1] -1/-1/-1->5->6 [2] 6/4/-1->5->3 [3] 6/4/-1->5->3
+ip-26-0-166-125:1332487:1332930 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-125:1332488:1332929 [6] NCCL INFO Trees [0] 5/7/-1->6->4 [1] 5/7/-1->6->4 [2] -1/-1/-1->6->5 [3] -1/-1/-1->6->5
+ip-26-0-166-125:1332488:1332929 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-125:1332483:1332931 [1] NCCL INFO Trees [0] -1/-1/-1->1->2 [1] -1/-1/-1->1->2 [2] 2/0/-1->1->3 [3] 2/0/-1->1->3
+ip-26-0-166-125:1332482:1332927 [0] NCCL INFO Channel 00/04 :    0   1   2   3   4   5   6   7
+ip-26-0-166-125:1332483:1332931 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-125:1332489:1332932 [7] NCCL INFO Trees [0] -1/-1/-1->7->6 [1] -1/-1/-1->7->6 [2] 3/-1/-1->7->-1 [3] 3/-1/-1->7->-1
+ip-26-0-166-125:1332482:1332927 [0] NCCL INFO Channel 01/04 :    0   1   2   3   4   5   6   7
+ip-26-0-166-125:1332489:1332932 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-125:1332482:1332927 [0] NCCL INFO Channel 02/04 :    0   1   2   3   4   5   6   7
+ip-26-0-166-125:1332485:1332933 [3] NCCL INFO Trees [0] -1/-1/-1->3->2 [1] -1/-1/-1->3->2 [2] 5/1/-1->3->7 [3] 5/1/-1->3->7
+ip-26-0-166-125:1332482:1332927 [0] NCCL INFO Channel 03/04 :    0   1   2   3   4   5   6   7
+ip-26-0-166-125:1332485:1332933 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-125:1332482:1332927 [0] NCCL INFO Trees [0] 4/-1/-1->0->-1 [1] 4/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-166-125:1332482:1332927 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-125:1332484:1332934 [2] NCCL INFO Trees [0] 1/3/-1->2->4 [1] 1/3/-1->2->4 [2] -1/-1/-1->2->1 [3] -1/-1/-1->2->1
+ip-26-0-166-125:1332486:1332928 [4] NCCL INFO Trees [0] 2/6/-1->4->0 [1] 2/6/-1->4->0 [2] -1/-1/-1->4->5 [3] -1/-1/-1->4->5
+ip-26-0-166-125:1332484:1332934 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-125:1332486:1332928 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-36:1061625:1062047 [5] NCCL INFO Channel 00/0 : 4[4] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:1061624:1062046 [4] NCCL INFO Channel 00/0 : 3[3] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:1061625:1062047 [5] NCCL INFO Channel 01/0 : 4[4] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:1061624:1062046 [4] NCCL INFO Channel 01/0 : 3[3] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:1061625:1062047 [5] NCCL INFO Channel 02/0 : 4[4] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:1061624:1062046 [4] NCCL INFO Channel 02/0 : 3[3] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:1061625:1062047 [5] NCCL INFO Channel 03/0 : 4[4] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:1061624:1062046 [4] NCCL INFO Channel 03/0 : 3[3] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:1061625:1062047 [5] NCCL INFO Channel 00/0 : 5[5] -> 6[6] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:1061624:1062046 [4] NCCL INFO Channel 00/0 : 4[4] -> 5[5] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:1061625:1062047 [5] NCCL INFO Channel 01/0 : 5[5] -> 6[6] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:1061624:1062046 [4] NCCL INFO Channel 01/0 : 4[4] -> 5[5] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:1061625:1062047 [5] NCCL INFO Channel 02/0 : 5[5] -> 6[6] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:1061624:1062046 [4] NCCL INFO Channel 02/0 : 4[4] -> 5[5] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:1061625:1062047 [5] NCCL INFO Channel 03/0 : 5[5] -> 6[6] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:1061624:1062046 [4] NCCL INFO Channel 03/0 : 4[4] -> 5[5] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:1061622:1062049 [2] NCCL INFO Channel 00/0 : 1[1] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:1061623:1062044 [3] NCCL INFO Channel 00/0 : 2[2] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:1061622:1062049 [2] NCCL INFO Channel 01/0 : 1[1] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:1061622:1062049 [2] NCCL INFO Channel 02/0 : 1[1] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:1061623:1062044 [3] NCCL INFO Channel 01/0 : 2[2] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:1061622:1062049 [2] NCCL INFO Channel 03/0 : 1[1] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:1061623:1062044 [3] NCCL INFO Channel 02/0 : 2[2] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:1061622:1062049 [2] NCCL INFO Channel 00/0 : 2[2] -> 3[3] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:1061623:1062044 [3] NCCL INFO Channel 03/0 : 2[2] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:1061622:1062049 [2] NCCL INFO Channel 01/0 : 2[2] -> 3[3] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:1061622:1062049 [2] NCCL INFO Channel 02/0 : 2[2] -> 3[3] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:1061623:1062044 [3] NCCL INFO Channel 00/0 : 3[3] -> 4[4] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:1061621:1062045 [1] NCCL INFO Channel 00/0 : 0[0] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:1061627:1062048 [7] NCCL INFO Channel 00/0 : 6[6] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:1061622:1062049 [2] NCCL INFO Channel 03/0 : 2[2] -> 3[3] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:1061623:1062044 [3] NCCL INFO Channel 01/0 : 3[3] -> 4[4] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:1061623:1062044 [3] NCCL INFO Channel 02/0 : 3[3] -> 4[4] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:1061621:1062045 [1] NCCL INFO Channel 01/0 : 0[0] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:1061627:1062048 [7] NCCL INFO Channel 01/0 : 6[6] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:1061621:1062045 [1] NCCL INFO Channel 02/0 : 0[0] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:1061627:1062048 [7] NCCL INFO Channel 02/0 : 6[6] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:1061623:1062044 [3] NCCL INFO Channel 03/0 : 3[3] -> 4[4] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:1061621:1062045 [1] NCCL INFO Channel 03/0 : 0[0] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:1061627:1062048 [7] NCCL INFO Channel 03/0 : 6[6] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:1061621:1062045 [1] NCCL INFO Channel 00/0 : 1[1] -> 2[2] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:1061627:1062048 [7] NCCL INFO Channel 00/0 : 7[7] -> 0[0] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:1061620:1062042 [0] NCCL INFO Channel 00/0 : 7[7] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:1061621:1062045 [1] NCCL INFO Channel 01/0 : 1[1] -> 2[2] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:1061627:1062048 [7] NCCL INFO Channel 01/0 : 7[7] -> 0[0] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:1061621:1062045 [1] NCCL INFO Channel 02/0 : 1[1] -> 2[2] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:1061627:1062048 [7] NCCL INFO Channel 02/0 : 7[7] -> 0[0] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:1061620:1062042 [0] NCCL INFO Channel 01/0 : 7[7] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:1061621:1062045 [1] NCCL INFO Channel 03/0 : 1[1] -> 2[2] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:1061627:1062048 [7] NCCL INFO Channel 03/0 : 7[7] -> 0[0] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:1061620:1062042 [0] NCCL INFO Channel 02/0 : 7[7] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:1061626:1062043 [6] NCCL INFO Channel 00/0 : 5[5] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:1061620:1062042 [0] NCCL INFO Channel 03/0 : 7[7] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:1061620:1062042 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[1] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:1061626:1062043 [6] NCCL INFO Channel 01/0 : 5[5] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:1061620:1062042 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[1] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:1061626:1062043 [6] NCCL INFO Channel 02/0 : 5[5] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:1061620:1062042 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[1] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:1061620:1062042 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[1] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:1061626:1062043 [6] NCCL INFO Channel 03/0 : 5[5] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:1061626:1062043 [6] NCCL INFO Channel 00/0 : 6[6] -> 7[7] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:1061626:1062043 [6] NCCL INFO Channel 01/0 : 6[6] -> 7[7] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:1061626:1062043 [6] NCCL INFO Channel 02/0 : 6[6] -> 7[7] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:1061626:1062043 [6] NCCL INFO Channel 03/0 : 6[6] -> 7[7] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:1332485:1332933 [3] NCCL INFO Channel 00/0 : 2[2] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:1332485:1332933 [3] NCCL INFO Channel 01/0 : 2[2] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:1332485:1332933 [3] NCCL INFO Channel 02/0 : 2[2] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:1332485:1332933 [3] NCCL INFO Channel 03/0 : 2[2] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:1332485:1332933 [3] NCCL INFO Channel 00/0 : 3[3] -> 4[4] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:1332485:1332933 [3] NCCL INFO Channel 01/0 : 3[3] -> 4[4] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:1332485:1332933 [3] NCCL INFO Channel 02/0 : 3[3] -> 4[4] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:1332485:1332933 [3] NCCL INFO Channel 03/0 : 3[3] -> 4[4] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:1332484:1332934 [2] NCCL INFO Channel 00/0 : 1[1] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:1332483:1332931 [1] NCCL INFO Channel 00/0 : 0[0] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:1332484:1332934 [2] NCCL INFO Channel 01/0 : 1[1] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:1332483:1332931 [1] NCCL INFO Channel 01/0 : 0[0] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:1332489:1332932 [7] NCCL INFO Channel 00/0 : 6[6] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:1332484:1332934 [2] NCCL INFO Channel 02/0 : 1[1] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:1332483:1332931 [1] NCCL INFO Channel 02/0 : 0[0] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:1332484:1332934 [2] NCCL INFO Channel 03/0 : 1[1] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:1332489:1332932 [7] NCCL INFO Channel 01/0 : 6[6] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:1332483:1332931 [1] NCCL INFO Channel 03/0 : 0[0] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:1332484:1332934 [2] NCCL INFO Channel 00/0 : 2[2] -> 3[3] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:1332489:1332932 [7] NCCL INFO Channel 02/0 : 6[6] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:1332483:1332931 [1] NCCL INFO Channel 00/0 : 1[1] -> 2[2] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:1332484:1332934 [2] NCCL INFO Channel 01/0 : 2[2] -> 3[3] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:1332489:1332932 [7] NCCL INFO Channel 03/0 : 6[6] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:1332483:1332931 [1] NCCL INFO Channel 01/0 : 1[1] -> 2[2] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:1332484:1332934 [2] NCCL INFO Channel 02/0 : 2[2] -> 3[3] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:1332489:1332932 [7] NCCL INFO Channel 00/0 : 7[7] -> 0[0] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:1332483:1332931 [1] NCCL INFO Channel 02/0 : 1[1] -> 2[2] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:1332484:1332934 [2] NCCL INFO Channel 03/0 : 2[2] -> 3[3] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:1332489:1332932 [7] NCCL INFO Channel 01/0 : 7[7] -> 0[0] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:1332483:1332931 [1] NCCL INFO Channel 03/0 : 1[1] -> 2[2] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:1332489:1332932 [7] NCCL INFO Channel 02/0 : 7[7] -> 0[0] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:1332489:1332932 [7] NCCL INFO Channel 03/0 : 7[7] -> 0[0] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:1332488:1332929 [6] NCCL INFO Channel 00/0 : 5[5] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:1332482:1332927 [0] NCCL INFO Channel 00/0 : 7[7] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:1332487:1332930 [5] NCCL INFO Channel 00/0 : 4[4] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:1332486:1332928 [4] NCCL INFO Channel 00/0 : 3[3] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:1332488:1332929 [6] NCCL INFO Channel 01/0 : 5[5] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:1332487:1332930 [5] NCCL INFO Channel 01/0 : 4[4] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:1332482:1332927 [0] NCCL INFO Channel 01/0 : 7[7] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:1332486:1332928 [4] NCCL INFO Channel 01/0 : 3[3] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:1332488:1332929 [6] NCCL INFO Channel 02/0 : 5[5] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:1332487:1332930 [5] NCCL INFO Channel 02/0 : 4[4] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:1332482:1332927 [0] NCCL INFO Channel 02/0 : 7[7] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:1332488:1332929 [6] NCCL INFO Channel 03/0 : 5[5] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:1332486:1332928 [4] NCCL INFO Channel 02/0 : 3[3] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:1332487:1332930 [5] NCCL INFO Channel 03/0 : 4[4] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:1332488:1332929 [6] NCCL INFO Channel 00/0 : 6[6] -> 7[7] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:1332482:1332927 [0] NCCL INFO Channel 03/0 : 7[7] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:1332487:1332930 [5] NCCL INFO Channel 00/0 : 5[5] -> 6[6] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:1332486:1332928 [4] NCCL INFO Channel 03/0 : 3[3] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:1332488:1332929 [6] NCCL INFO Channel 01/0 : 6[6] -> 7[7] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:1332487:1332930 [5] NCCL INFO Channel 01/0 : 5[5] -> 6[6] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:1332486:1332928 [4] NCCL INFO Channel 00/0 : 4[4] -> 5[5] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:1332482:1332927 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[1] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:1332488:1332929 [6] NCCL INFO Channel 02/0 : 6[6] -> 7[7] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:1332487:1332930 [5] NCCL INFO Channel 02/0 : 5[5] -> 6[6] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:1332486:1332928 [4] NCCL INFO Channel 01/0 : 4[4] -> 5[5] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:1332488:1332929 [6] NCCL INFO Channel 03/0 : 6[6] -> 7[7] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:1332482:1332927 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[1] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:1332487:1332930 [5] NCCL INFO Channel 03/0 : 5[5] -> 6[6] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:1332486:1332928 [4] NCCL INFO Channel 02/0 : 4[4] -> 5[5] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:1332482:1332927 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[1] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:1332486:1332928 [4] NCCL INFO Channel 03/0 : 4[4] -> 5[5] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:1332482:1332927 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[1] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:1061623:1062044 [3] NCCL INFO Connected all rings
+ip-26-0-166-36:1061623:1062044 [3] NCCL INFO Channel 02/0 : 1[1] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:1061623:1062044 [3] NCCL INFO Channel 03/0 : 1[1] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:1061623:1062044 [3] NCCL INFO Channel 02/0 : 3[3] -> 5[5] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:1061623:1062044 [3] NCCL INFO Channel 03/0 : 3[3] -> 5[5] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:1061627:1062048 [7] NCCL INFO Connected all rings
+ip-26-0-166-36:1061627:1062048 [7] NCCL INFO Channel 02/0 : 3[3] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:1061627:1062048 [7] NCCL INFO Channel 03/0 : 3[3] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:1061627:1062048 [7] NCCL INFO Channel 02/0 : 7[7] -> 3[3] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:1061627:1062048 [7] NCCL INFO Channel 03/0 : 7[7] -> 3[3] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:1061621:1062045 [1] NCCL INFO Connected all rings
+ip-26-0-166-36:1061621:1062045 [1] NCCL INFO Channel 02/0 : 1[1] -> 3[3] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:1061625:1062047 [5] NCCL INFO Connected all rings
+ip-26-0-166-36:1061621:1062045 [1] NCCL INFO Channel 03/0 : 1[1] -> 3[3] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:1061625:1062047 [5] NCCL INFO Channel 02/0 : 3[3] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:1061625:1062047 [5] NCCL INFO Channel 03/0 : 3[3] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:1061621:1062045 [1] NCCL INFO Channel 02/0 : 3[3] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:1061621:1062045 [1] NCCL INFO Channel 03/0 : 3[3] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:1061625:1062047 [5] NCCL INFO Channel 02/0 : 5[5] -> 3[3] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:1061623:1062044 [3] NCCL INFO Channel 02/0 : 7[7] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:1061625:1062047 [5] NCCL INFO Channel 03/0 : 5[5] -> 3[3] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:1061623:1062044 [3] NCCL INFO Channel 03/0 : 7[7] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:1061623:1062044 [3] NCCL INFO Channel 02/0 : 3[3] -> 7[7] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:1061623:1062044 [3] NCCL INFO Channel 03/0 : 3[3] -> 7[7] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:1061623:1062044 [3] NCCL INFO Channel 02/0 : 5[5] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:1061627:1062048 [7] NCCL INFO Channel 00/0 : 7[7] -> 6[6] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:1061622:1062049 [2] NCCL INFO Connected all rings
+ip-26-0-166-36:1061627:1062048 [7] NCCL INFO Channel 01/0 : 7[7] -> 6[6] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:1061623:1062044 [3] NCCL INFO Channel 03/0 : 5[5] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:1061623:1062044 [3] NCCL INFO Channel 02/0 : 3[3] -> 1[1] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:1061622:1062049 [2] NCCL INFO Channel 00/0 : 2[2] -> 4[4] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:1061623:1062044 [3] NCCL INFO Channel 03/0 : 3[3] -> 1[1] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:1061622:1062049 [2] NCCL INFO Channel 01/0 : 2[2] -> 4[4] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:1061625:1062047 [5] NCCL INFO Channel 00/0 : 6[6] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:1061625:1062047 [5] NCCL INFO Channel 01/0 : 6[6] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:1061621:1062045 [1] NCCL INFO Channel 00/0 : 2[2] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:1061623:1062044 [3] NCCL INFO Channel 00/0 : 3[3] -> 2[2] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:1061625:1062047 [5] NCCL INFO Channel 02/0 : 6[6] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:1061621:1062045 [1] NCCL INFO Channel 01/0 : 2[2] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:1061623:1062044 [3] NCCL INFO Channel 01/0 : 3[3] -> 2[2] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:1061625:1062047 [5] NCCL INFO Channel 03/0 : 6[6] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:1061621:1062045 [1] NCCL INFO Channel 02/0 : 2[2] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:1061625:1062047 [5] NCCL INFO Channel 02/0 : 5[5] -> 4[4] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:1061621:1062045 [1] NCCL INFO Channel 03/0 : 2[2] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:1061625:1062047 [5] NCCL INFO Channel 03/0 : 5[5] -> 4[4] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:1061621:1062045 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[0] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:1061621:1062045 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[0] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:1061626:1062043 [6] NCCL INFO Connected all rings
+ip-26-0-166-36:1061626:1062043 [6] NCCL INFO Channel 00/0 : 4[4] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:1061626:1062043 [6] NCCL INFO Channel 01/0 : 4[4] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:1061624:1062046 [4] NCCL INFO Connected all rings
+ip-26-0-166-36:1061624:1062046 [4] NCCL INFO Channel 00/0 : 2[2] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:1061624:1062046 [4] NCCL INFO Channel 01/0 : 2[2] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:1061624:1062046 [4] NCCL INFO Channel 00/0 : 4[4] -> 6[6] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:1061624:1062046 [4] NCCL INFO Channel 01/0 : 4[4] -> 6[6] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:1061622:1062049 [2] NCCL INFO Channel 00/0 : 4[4] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:1061626:1062043 [6] NCCL INFO Channel 00/0 : 6[6] -> 4[4] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:1061622:1062049 [2] NCCL INFO Channel 01/0 : 4[4] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:1061626:1062043 [6] NCCL INFO Channel 01/0 : 6[6] -> 4[4] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:1061624:1062046 [4] NCCL INFO Channel 00/0 : 0[0] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:1061624:1062046 [4] NCCL INFO Channel 01/0 : 0[0] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:1061624:1062046 [4] NCCL INFO Channel 00/0 : 4[4] -> 0[0] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:1061624:1062046 [4] NCCL INFO Channel 01/0 : 4[4] -> 0[0] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:1061620:1062042 [0] NCCL INFO Connected all rings
+ip-26-0-166-36:1061620:1062042 [0] NCCL INFO Channel 00/0 : 4[4] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:1061620:1062042 [0] NCCL INFO Channel 01/0 : 4[4] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:1061620:1062042 [0] NCCL INFO Channel 00/0 : 0[0] -> 4[4] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:1061620:1062042 [0] NCCL INFO Channel 01/0 : 0[0] -> 4[4] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:1061624:1062046 [4] NCCL INFO Channel 00/0 : 6[6] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:1061620:1062042 [0] NCCL INFO Channel 02/0 : 1[1] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:1061624:1062046 [4] NCCL INFO Channel 01/0 : 6[6] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:1061620:1062042 [0] NCCL INFO Channel 03/0 : 1[1] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:1061624:1062046 [4] NCCL INFO Channel 00/0 : 4[4] -> 2[2] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:1061624:1062046 [4] NCCL INFO Channel 01/0 : 4[4] -> 2[2] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:1061626:1062043 [6] NCCL INFO Channel 00/0 : 7[7] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:1061622:1062049 [2] NCCL INFO Channel 00/0 : 3[3] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:1061626:1062043 [6] NCCL INFO Channel 01/0 : 7[7] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:1061624:1062046 [4] NCCL INFO Channel 02/0 : 5[5] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:1061622:1062049 [2] NCCL INFO Channel 01/0 : 3[3] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:1061624:1062046 [4] NCCL INFO Channel 03/0 : 5[5] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:1061626:1062043 [6] NCCL INFO Channel 00/0 : 6[6] -> 5[5] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:1061622:1062049 [2] NCCL INFO Channel 00/0 : 2[2] -> 1[1] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:1061626:1062043 [6] NCCL INFO Channel 01/0 : 6[6] -> 5[5] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:1061622:1062049 [2] NCCL INFO Channel 01/0 : 2[2] -> 1[1] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:1061626:1062043 [6] NCCL INFO Channel 02/0 : 6[6] -> 5[5] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:1061622:1062049 [2] NCCL INFO Channel 02/0 : 2[2] -> 1[1] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:1061626:1062043 [6] NCCL INFO Channel 03/0 : 6[6] -> 5[5] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:1061622:1062049 [2] NCCL INFO Channel 03/0 : 2[2] -> 1[1] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:1332487:1332930 [5] NCCL INFO Connected all rings
+ip-26-0-166-125:1332487:1332930 [5] NCCL INFO Channel 02/0 : 3[3] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:1332487:1332930 [5] NCCL INFO Channel 03/0 : 3[3] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:1332483:1332931 [1] NCCL INFO Connected all rings
+ip-26-0-166-125:1332483:1332931 [1] NCCL INFO Channel 02/0 : 1[1] -> 3[3] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:1332483:1332931 [1] NCCL INFO Channel 03/0 : 1[1] -> 3[3] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:1332485:1332933 [3] NCCL INFO Connected all rings
+ip-26-0-166-125:1332485:1332933 [3] NCCL INFO Channel 02/0 : 1[1] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:1332485:1332933 [3] NCCL INFO Channel 03/0 : 1[1] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:1332485:1332933 [3] NCCL INFO Channel 02/0 : 3[3] -> 5[5] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:1332485:1332933 [3] NCCL INFO Channel 03/0 : 3[3] -> 5[5] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:1332483:1332931 [1] NCCL INFO Channel 02/0 : 3[3] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:1332483:1332931 [1] NCCL INFO Channel 03/0 : 3[3] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:1332487:1332930 [5] NCCL INFO Channel 02/0 : 5[5] -> 3[3] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:1332485:1332933 [3] NCCL INFO Channel 02/0 : 7[7] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:1332487:1332930 [5] NCCL INFO Channel 03/0 : 5[5] -> 3[3] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:1332485:1332933 [3] NCCL INFO Channel 03/0 : 7[7] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:1332485:1332933 [3] NCCL INFO Channel 02/0 : 3[3] -> 7[7] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:1332485:1332933 [3] NCCL INFO Channel 03/0 : 3[3] -> 7[7] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:1332484:1332934 [2] NCCL INFO Connected all rings
+ip-26-0-166-125:1332484:1332934 [2] NCCL INFO Channel 00/0 : 2[2] -> 4[4] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:1332484:1332934 [2] NCCL INFO Channel 01/0 : 2[2] -> 4[4] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:1332488:1332929 [6] NCCL INFO Connected all rings
+ip-26-0-166-125:1332488:1332929 [6] NCCL INFO Channel 00/0 : 4[4] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:1332488:1332929 [6] NCCL INFO Channel 01/0 : 4[4] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:1332486:1332928 [4] NCCL INFO Connected all rings
+ip-26-0-166-125:1332486:1332928 [4] NCCL INFO Channel 00/0 : 2[2] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:1332486:1332928 [4] NCCL INFO Channel 01/0 : 2[2] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:1332486:1332928 [4] NCCL INFO Channel 00/0 : 4[4] -> 6[6] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:1332486:1332928 [4] NCCL INFO Channel 01/0 : 4[4] -> 6[6] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:1332484:1332934 [2] NCCL INFO Channel 00/0 : 4[4] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:1332488:1332929 [6] NCCL INFO Channel 00/0 : 6[6] -> 4[4] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:1332484:1332934 [2] NCCL INFO Channel 01/0 : 4[4] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:1332486:1332928 [4] NCCL INFO Channel 00/0 : 0[0] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:1332488:1332929 [6] NCCL INFO Channel 01/0 : 6[6] -> 4[4] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:1332486:1332928 [4] NCCL INFO Channel 01/0 : 0[0] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:1332486:1332928 [4] NCCL INFO Channel 00/0 : 4[4] -> 0[0] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:1332486:1332928 [4] NCCL INFO Channel 01/0 : 4[4] -> 0[0] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:1332482:1332927 [0] NCCL INFO Connected all rings
+ip-26-0-166-125:1332482:1332927 [0] NCCL INFO Channel 00/0 : 4[4] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:1332482:1332927 [0] NCCL INFO Channel 01/0 : 4[4] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:1332482:1332927 [0] NCCL INFO Channel 00/0 : 0[0] -> 4[4] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:1332482:1332927 [0] NCCL INFO Channel 01/0 : 0[0] -> 4[4] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:1332486:1332928 [4] NCCL INFO Channel 00/0 : 6[6] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:1332482:1332927 [0] NCCL INFO Channel 02/0 : 1[1] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:1332486:1332928 [4] NCCL INFO Channel 01/0 : 6[6] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:1332482:1332927 [0] NCCL INFO Channel 03/0 : 1[1] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:1332486:1332928 [4] NCCL INFO Channel 00/0 : 4[4] -> 2[2] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:1332486:1332928 [4] NCCL INFO Channel 01/0 : 4[4] -> 2[2] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:1332488:1332929 [6] NCCL INFO Channel 00/0 : 7[7] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:1332488:1332929 [6] NCCL INFO Channel 01/0 : 7[7] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:1332484:1332934 [2] NCCL INFO Channel 00/0 : 3[3] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:1332486:1332928 [4] NCCL INFO Channel 02/0 : 5[5] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:1332488:1332929 [6] NCCL INFO Channel 00/0 : 6[6] -> 5[5] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:1332484:1332934 [2] NCCL INFO Channel 01/0 : 3[3] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:1332486:1332928 [4] NCCL INFO Channel 03/0 : 5[5] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:1332488:1332929 [6] NCCL INFO Channel 01/0 : 6[6] -> 5[5] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:1332484:1332934 [2] NCCL INFO Channel 00/0 : 2[2] -> 1[1] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:1332488:1332929 [6] NCCL INFO Channel 02/0 : 6[6] -> 5[5] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:1332484:1332934 [2] NCCL INFO Channel 01/0 : 2[2] -> 1[1] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:1332488:1332929 [6] NCCL INFO Channel 03/0 : 6[6] -> 5[5] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:1332484:1332934 [2] NCCL INFO Channel 02/0 : 2[2] -> 1[1] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:1332484:1332934 [2] NCCL INFO Channel 03/0 : 2[2] -> 1[1] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:1332489:1332932 [7] NCCL INFO Connected all rings
+ip-26-0-166-125:1332489:1332932 [7] NCCL INFO Channel 02/0 : 3[3] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:1332489:1332932 [7] NCCL INFO Channel 03/0 : 3[3] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:1332489:1332932 [7] NCCL INFO Channel 02/0 : 7[7] -> 3[3] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:1332489:1332932 [7] NCCL INFO Channel 03/0 : 7[7] -> 3[3] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:1332485:1332933 [3] NCCL INFO Channel 02/0 : 5[5] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:1332489:1332932 [7] NCCL INFO Channel 00/0 : 7[7] -> 6[6] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:1332485:1332933 [3] NCCL INFO Channel 03/0 : 5[5] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:1332489:1332932 [7] NCCL INFO Channel 01/0 : 7[7] -> 6[6] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:1332485:1332933 [3] NCCL INFO Channel 02/0 : 3[3] -> 1[1] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:1332485:1332933 [3] NCCL INFO Channel 03/0 : 3[3] -> 1[1] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:1332487:1332930 [5] NCCL INFO Channel 00/0 : 6[6] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:1332485:1332933 [3] NCCL INFO Channel 00/0 : 3[3] -> 2[2] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:1332483:1332931 [1] NCCL INFO Channel 00/0 : 2[2] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:1332487:1332930 [5] NCCL INFO Channel 01/0 : 6[6] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:1332485:1332933 [3] NCCL INFO Channel 01/0 : 3[3] -> 2[2] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:1332483:1332931 [1] NCCL INFO Channel 01/0 : 2[2] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:1332487:1332930 [5] NCCL INFO Channel 02/0 : 6[6] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:1332483:1332931 [1] NCCL INFO Channel 02/0 : 2[2] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:1332487:1332930 [5] NCCL INFO Channel 03/0 : 6[6] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:1332487:1332930 [5] NCCL INFO Channel 02/0 : 5[5] -> 4[4] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:1332483:1332931 [1] NCCL INFO Channel 03/0 : 2[2] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:1332487:1332930 [5] NCCL INFO Channel 03/0 : 5[5] -> 4[4] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:1332483:1332931 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[0] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:1332483:1332931 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[0] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:1061627:1062048 [7] NCCL INFO Connected all trees
+ip-26-0-166-36:1061627:1062048 [7] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-166-36:1061627:1062048 [7] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-36:1061627:1062048 [7] NCCL INFO comm 0x94c63c0 rank 7 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0x3dbd315886848384 - Init COMPLETE
+ip-26-0-166-36:1061620:1062042 [0] NCCL INFO Connected all trees
+ip-26-0-166-36:1061620:1062042 [0] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-166-36:1061620:1062042 [0] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-36:1061620:1062042 [0] NCCL INFO comm 0x88dd7c0 rank 0 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0x3dbd315886848384 - Init COMPLETE
+ip-26-0-166-36:1061621:1062045 [1] NCCL INFO Connected all trees
+ip-26-0-166-36:1061621:1062045 [1] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-166-36:1061621:1062045 [1] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-125:1332489:1332932 [7] NCCL INFO Connected all trees
+ip-26-0-166-125:1332489:1332932 [7] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-166-125:1332489:1332932 [7] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-36:1061621:1062045 [1] NCCL INFO comm 0x9598880 rank 1 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0x3dbd315886848384 - Init COMPLETE
+ip-26-0-166-36:1061626:1062043 [6] NCCL INFO Connected all trees
+ip-26-0-166-36:1061626:1062043 [6] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-166-36:1061626:1062043 [6] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-36:1061622:1062049 [2] NCCL INFO Connected all trees
+ip-26-0-166-36:1061622:1062049 [2] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-166-36:1061622:1062049 [2] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-125:1332489:1332932 [7] NCCL INFO comm 0x9f7f300 rank 7 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0x9221e1aa7583a57b - Init COMPLETE
+ip-26-0-166-36:1061626:1062043 [6] NCCL INFO comm 0x8842650 rank 6 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0x3dbd315886848384 - Init COMPLETE
+ip-26-0-166-36:1061622:1062049 [2] NCCL INFO comm 0x90d4310 rank 2 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0x3dbd315886848384 - Init COMPLETE
+ip-26-0-166-36:1061623:1062044 [3] NCCL INFO Connected all trees
+ip-26-0-166-36:1061623:1062044 [3] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-166-36:1061623:1062044 [3] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-125:1332482:1332927 [0] NCCL INFO Connected all trees
+ip-26-0-166-125:1332482:1332927 [0] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-166-125:1332482:1332927 [0] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-36:1061625:1062047 [5] NCCL INFO Connected all trees
+ip-26-0-166-36:1061625:1062047 [5] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-166-36:1061625:1062047 [5] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-36:1061623:1062044 [3] NCCL INFO comm 0x8693dc0 rank 3 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0x3dbd315886848384 - Init COMPLETE
+ip-26-0-166-36:1061624:1062046 [4] NCCL INFO Connected all trees
+ip-26-0-166-36:1061624:1062046 [4] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-166-36:1061624:1062046 [4] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-125:1332482:1332927 [0] NCCL INFO comm 0x98c2990 rank 0 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0x9221e1aa7583a57b - Init COMPLETE
+ip-26-0-166-36:1061625:1062047 [5] NCCL INFO comm 0x9286ea0 rank 5 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0x3dbd315886848384 - Init COMPLETE
+ip-26-0-166-36:1061624:1062046 [4] NCCL INFO comm 0xa1014c0 rank 4 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0x3dbd315886848384 - Init COMPLETE
+ip-26-0-166-125:1332484:1332934 [2] NCCL INFO Connected all trees
+ip-26-0-166-125:1332484:1332934 [2] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-166-125:1332484:1332934 [2] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-125:1332483:1332931 [1] NCCL INFO Connected all trees
+ip-26-0-166-125:1332483:1332931 [1] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-166-125:1332483:1332931 [1] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-125:1332484:1332934 [2] NCCL INFO comm 0x9aece10 rank 2 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0x9221e1aa7583a57b - Init COMPLETE
+ip-26-0-166-125:1332483:1332931 [1] NCCL INFO comm 0x93c4f50 rank 1 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0x9221e1aa7583a57b - Init COMPLETE
+ip-26-0-166-125:1332488:1332929 [6] NCCL INFO Connected all trees
+ip-26-0-166-125:1332488:1332929 [6] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-166-125:1332488:1332929 [6] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-125:1332485:1332933 [3] NCCL INFO Connected all trees
+ip-26-0-166-125:1332485:1332933 [3] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-166-125:1332485:1332933 [3] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-125:1332488:1332929 [6] NCCL INFO comm 0x88cce50 rank 6 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0x9221e1aa7583a57b - Init COMPLETE
+ip-26-0-166-125:1332485:1332933 [3] NCCL INFO comm 0x8865ad0 rank 3 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0x9221e1aa7583a57b - Init COMPLETE
+ip-26-0-166-125:1332487:1332930 [5] NCCL INFO Connected all trees
+ip-26-0-166-125:1332487:1332930 [5] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-166-125:1332487:1332930 [5] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-125:1332486:1332928 [4] NCCL INFO Connected all trees
+ip-26-0-166-125:1332486:1332928 [4] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-166-125:1332486:1332928 [4] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-125:1332487:1332930 [5] NCCL INFO comm 0x9af9a60 rank 5 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0x9221e1aa7583a57b - Init COMPLETE
+ip-26-0-166-125:1332486:1332928 [4] NCCL INFO comm 0x9187160 rank 4 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0x9221e1aa7583a57b - Init COMPLETE
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+ip-26-0-166-36:1061624:1062083 [4] NCCL INFO Using network Libfabric
+ip-26-0-166-125:1332488:1332976 [6] NCCL INFO Using network Libfabric
+ip-26-0-166-36:1061626:1062084 [6] NCCL INFO Using network Libfabric
+ip-26-0-166-36:1061627:1062089 [7] NCCL INFO Using network Libfabric
+ip-26-0-166-36:1061622:1062085 [2] NCCL INFO Using network Libfabric
+ip-26-0-166-125:1332483:1332977 [1] NCCL INFO Using network Libfabric
+ip-26-0-166-125:1332482:1332974 [0] NCCL INFO Using network Libfabric
+ip-26-0-166-36:1061620:1062090 [0] NCCL INFO Using network Libfabric
+ip-26-0-166-36:1061625:1062088 [5] NCCL INFO Using network Libfabric
+ip-26-0-166-125:1332486:1332970 [4] NCCL INFO Using network Libfabric
+ip-26-0-166-125:1332485:1332978 [3] NCCL INFO Using network Libfabric
+ip-26-0-166-125:1332487:1332973 [5] NCCL INFO Using network Libfabric
+ip-26-0-166-36:1061623:1062087 [3] NCCL INFO Using network Libfabric
+ip-26-0-166-36:1061621:1062091 [1] NCCL INFO Using network Libfabric
+ip-26-0-166-125:1332482:1332974 [0] NCCL INFO comm 0xb7c5c10 rank 0 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0x3302783f73043d45 - Init START
+ip-26-0-166-125:1332483:1332977 [1] NCCL INFO comm 0xb2c5b40 rank 1 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0x3302783f73043d45 - Init START
+ip-26-0-166-36:1061626:1062084 [6] NCCL INFO comm 0xa5763d0 rank 0 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0x20b055b7295e97aa - Init START
+ip-26-0-166-36:1061627:1062089 [7] NCCL INFO comm 0xb1fae50 rank 1 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0x20b055b7295e97aa - Init START
+ip-26-0-166-125:1332483:1332977 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-166-125:1332482:1332974 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/205
+ip-26-0-166-36:1061625:1062088 [5] NCCL INFO comm 0xafbc490 rank 1 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0x640002f1b71c9ffa - Init START
+ip-26-0-166-36:1061624:1062083 [4] NCCL INFO comm 0xbe380e0 rank 0 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0x640002f1b71c9ffa - Init START
+ip-26-0-166-36:1061627:1062089 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-166-36:1061626:1062084 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-166-125:1332487:1332973 [5] NCCL INFO comm 0xb9f8fc0 rank 1 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0x4cef1d6bf2e8f61 - Init START
+ip-26-0-166-125:1332486:1332970 [4] NCCL INFO comm 0xb088290 rank 0 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0x4cef1d6bf2e8f61 - Init START
+ip-26-0-166-36:1061623:1062087 [3] NCCL INFO comm 0xa3c9190 rank 1 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0xcee17cd97bfa3d70 - Init START
+ip-26-0-166-36:1061622:1062085 [2] NCCL INFO comm 0xae0a390 rank 0 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0xcee17cd97bfa3d70 - Init START
+ip-26-0-166-36:1061621:1062091 [1] NCCL INFO comm 0xb2cd190 rank 1 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0x1b1fe3d9b889f4e1 - Init START
+ip-26-0-166-36:1061620:1062090 [0] NCCL INFO comm 0xa611f00 rank 0 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0x1b1fe3d9b889f4e1 - Init START
+ip-26-0-166-36:1061624:1062083 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-166-36:1061625:1062088 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-166-125:1332489:1332979 [7] NCCL INFO Using network Libfabric
+ip-26-0-166-125:1332487:1332973 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-166-125:1332486:1332970 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-166-36:1061620:1062090 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-166-36:1061621:1062091 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-166-36:1061622:1062085 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-166-36:1061623:1062087 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-166-125:1332484:1332975 [2] NCCL INFO Using network Libfabric
+ip-26-0-166-125:1332485:1332978 [3] NCCL INFO comm 0xa767840 rank 1 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0xfa7340fae4b966a6 - Init START
+ip-26-0-166-125:1332484:1332975 [2] NCCL INFO comm 0xb9ecf80 rank 0 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0xfa7340fae4b966a6 - Init START
+ip-26-0-166-125:1332489:1332979 [7] NCCL INFO comm 0xbe7ef80 rank 1 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0xdc3c7ca3884008cc - Init START
+ip-26-0-166-125:1332488:1332976 [6] NCCL INFO comm 0xa7cdb80 rank 0 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0xdc3c7ca3884008cc - Init START
+ip-26-0-166-125:1332484:1332975 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-166-125:1332485:1332978 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-166-125:1332489:1332979 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-166-125:1332488:1332976 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-166-36:1061622:1062085 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-166-125:1332487:1332973 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,fff00000,00000000
+ip-26-0-166-36:1061620:1062090 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-166-36:1061625:1062088 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-166-125:1332483:1332977 [1] NCCL INFO Setting affinity for GPU 1 to 0fff,ffffffff
+ip-26-0-166-125:1332482:1332974 [0] NCCL INFO Setting affinity for GPU 0 to 0fff,ffffffff
+ip-26-0-166-125:1332483:1332977 [1] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-166-125:1332482:1332974 [0] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-166-125:1332483:1332977 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-125:1332482:1332974 [0] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-166-125:1332482:1332974 [0] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-166-125:1332482:1332974 [0] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-166-125:1332482:1332974 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-166-125:1332482:1332974 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-125:1332489:1332979 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,fff00000,00000000
+ip-26-0-166-36:1061627:1062089 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-166-36:1061626:1062084 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-166-36:1061627:1062089 [7] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-166-36:1061626:1062084 [6] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-166-36:1061627:1062089 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-36:1061626:1062084 [6] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-166-36:1061626:1062084 [6] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-166-36:1061626:1062084 [6] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-166-36:1061626:1062084 [6] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-166-36:1061626:1062084 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-36:1061621:1062091 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-166-36:1061623:1062087 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-166-36:1061621:1062091 [1] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-166-36:1061621:1062091 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-36:1061620:1062090 [0] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-166-36:1061620:1062090 [0] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-166-36:1061620:1062090 [0] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-166-36:1061620:1062090 [0] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-166-36:1061620:1062090 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-166-36:1061620:1062090 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-36:1061624:1062083 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-166-36:1061623:1062087 [3] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-166-36:1061623:1062087 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-36:1061622:1062085 [2] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-166-36:1061622:1062085 [2] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-166-36:1061622:1062085 [2] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-166-36:1061622:1062085 [2] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-166-36:1061622:1062085 [2] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-166-36:1061622:1062085 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-36:1061625:1062088 [5] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-166-36:1061625:1062088 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-36:1061624:1062083 [4] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-166-36:1061624:1062083 [4] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-166-36:1061624:1062083 [4] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-166-36:1061624:1062083 [4] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-166-36:1061624:1062083 [4] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-166-36:1061624:1062083 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-125:1332485:1332978 [3] NCCL INFO Setting affinity for GPU 3 to 0fff,ffffffff
+ip-26-0-166-125:1332484:1332975 [2] NCCL INFO Setting affinity for GPU 2 to 0fff,ffffffff
+ip-26-0-166-125:1332484:1332975 [2] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-166-125:1332485:1332978 [3] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-166-125:1332484:1332975 [2] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-166-125:1332485:1332978 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-125:1332484:1332975 [2] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-166-125:1332484:1332975 [2] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-166-125:1332484:1332975 [2] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-166-125:1332484:1332975 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-125:1332488:1332976 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,fff00000,00000000
+ip-26-0-166-125:1332488:1332976 [6] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-166-125:1332488:1332976 [6] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-166-125:1332489:1332979 [7] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-166-125:1332488:1332976 [6] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-166-125:1332489:1332979 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-125:1332488:1332976 [6] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-166-125:1332488:1332976 [6] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-166-125:1332488:1332976 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-125:1332486:1332970 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,fff00000,00000000
+ip-26-0-166-125:1332487:1332973 [5] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-166-125:1332486:1332970 [4] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-166-125:1332487:1332973 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-125:1332486:1332970 [4] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-166-125:1332486:1332970 [4] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-166-125:1332486:1332970 [4] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-166-125:1332486:1332970 [4] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-166-125:1332486:1332970 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-125:1332482:1332974 [0] NCCL INFO Channel 00/0 : 1[1] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:1332482:1332974 [0] NCCL INFO Channel 01/0 : 1[1] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:1332482:1332974 [0] NCCL INFO Channel 02/0 : 1[1] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:1332482:1332974 [0] NCCL INFO Channel 03/0 : 1[1] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:1332482:1332974 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[1] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:1332482:1332974 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[1] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:1332482:1332974 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[1] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:1332482:1332974 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[1] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:1061622:1062085 [2] NCCL INFO Channel 00/0 : 1[3] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:1061622:1062085 [2] NCCL INFO Channel 01/0 : 1[3] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:1061624:1062083 [4] NCCL INFO Channel 00/0 : 1[5] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:1061622:1062085 [2] NCCL INFO Channel 02/0 : 1[3] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:1061623:1062087 [3] NCCL INFO Channel 00/0 : 0[2] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:1061624:1062083 [4] NCCL INFO Channel 01/0 : 1[5] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:1061623:1062087 [3] NCCL INFO Channel 01/0 : 0[2] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:1061622:1062085 [2] NCCL INFO Channel 03/0 : 1[3] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:1061624:1062083 [4] NCCL INFO Channel 02/0 : 1[5] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:1061623:1062087 [3] NCCL INFO Channel 02/0 : 0[2] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:1061624:1062083 [4] NCCL INFO Channel 03/0 : 1[5] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:1061623:1062087 [3] NCCL INFO Channel 03/0 : 0[2] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:1061622:1062085 [2] NCCL INFO Channel 00/0 : 0[2] -> 1[3] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:1061624:1062083 [4] NCCL INFO Channel 00/0 : 0[4] -> 1[5] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:1061623:1062087 [3] NCCL INFO Channel 00/0 : 1[3] -> 0[2] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:1061624:1062083 [4] NCCL INFO Channel 01/0 : 0[4] -> 1[5] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:1061625:1062088 [5] NCCL INFO Channel 00/0 : 0[4] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:1061622:1062085 [2] NCCL INFO Channel 01/0 : 0[2] -> 1[3] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:1061623:1062087 [3] NCCL INFO Channel 01/0 : 1[3] -> 0[2] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:1061624:1062083 [4] NCCL INFO Channel 02/0 : 0[4] -> 1[5] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:1061623:1062087 [3] NCCL INFO Channel 02/0 : 1[3] -> 0[2] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:1061622:1062085 [2] NCCL INFO Channel 02/0 : 0[2] -> 1[3] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:1061625:1062088 [5] NCCL INFO Channel 01/0 : 0[4] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:1061624:1062083 [4] NCCL INFO Channel 03/0 : 0[4] -> 1[5] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:1061621:1062091 [1] NCCL INFO Channel 00/0 : 0[0] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:1061623:1062087 [3] NCCL INFO Channel 03/0 : 1[3] -> 0[2] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:1061622:1062085 [2] NCCL INFO Channel 03/0 : 0[2] -> 1[3] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:1061625:1062088 [5] NCCL INFO Channel 02/0 : 0[4] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:1061625:1062088 [5] NCCL INFO Channel 03/0 : 0[4] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:1061621:1062091 [1] NCCL INFO Channel 01/0 : 0[0] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:1061621:1062091 [1] NCCL INFO Channel 02/0 : 0[0] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:1061625:1062088 [5] NCCL INFO Channel 00/0 : 1[5] -> 0[4] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:1061621:1062091 [1] NCCL INFO Channel 03/0 : 0[0] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:1061625:1062088 [5] NCCL INFO Channel 01/0 : 1[5] -> 0[4] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:1061627:1062089 [7] NCCL INFO Channel 00/0 : 0[6] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:1061621:1062091 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[0] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:1061625:1062088 [5] NCCL INFO Channel 02/0 : 1[5] -> 0[4] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:1061625:1062088 [5] NCCL INFO Channel 03/0 : 1[5] -> 0[4] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:1061621:1062091 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[0] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:1061627:1062089 [7] NCCL INFO Channel 01/0 : 0[6] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:1061620:1062090 [0] NCCL INFO Channel 00/0 : 1[1] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:1061621:1062091 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[0] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:1061627:1062089 [7] NCCL INFO Channel 02/0 : 0[6] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:1061621:1062091 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[0] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:1061627:1062089 [7] NCCL INFO Channel 03/0 : 0[6] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:1061620:1062090 [0] NCCL INFO Channel 01/0 : 1[1] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:1061627:1062089 [7] NCCL INFO Channel 00/0 : 1[7] -> 0[6] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:1061627:1062089 [7] NCCL INFO Channel 01/0 : 1[7] -> 0[6] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:1061620:1062090 [0] NCCL INFO Channel 02/0 : 1[1] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:1061627:1062089 [7] NCCL INFO Channel 02/0 : 1[7] -> 0[6] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:1061620:1062090 [0] NCCL INFO Channel 03/0 : 1[1] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:1061627:1062089 [7] NCCL INFO Channel 03/0 : 1[7] -> 0[6] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:1061620:1062090 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[1] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:1061626:1062084 [6] NCCL INFO Channel 00/0 : 1[7] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:1061620:1062090 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[1] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:1061620:1062090 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[1] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:1061626:1062084 [6] NCCL INFO Channel 01/0 : 1[7] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:1061620:1062090 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[1] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:1061626:1062084 [6] NCCL INFO Channel 02/0 : 1[7] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:1061626:1062084 [6] NCCL INFO Channel 03/0 : 1[7] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:1061626:1062084 [6] NCCL INFO Channel 00/0 : 0[6] -> 1[7] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:1061626:1062084 [6] NCCL INFO Channel 01/0 : 0[6] -> 1[7] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:1061626:1062084 [6] NCCL INFO Channel 02/0 : 0[6] -> 1[7] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:1061626:1062084 [6] NCCL INFO Channel 03/0 : 0[6] -> 1[7] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:1332485:1332978 [3] NCCL INFO Channel 00/0 : 0[2] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:1332485:1332978 [3] NCCL INFO Channel 01/0 : 0[2] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:1332485:1332978 [3] NCCL INFO Channel 02/0 : 0[2] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:1332485:1332978 [3] NCCL INFO Channel 03/0 : 0[2] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:1332485:1332978 [3] NCCL INFO Channel 00/0 : 1[3] -> 0[2] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:1332485:1332978 [3] NCCL INFO Channel 01/0 : 1[3] -> 0[2] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:1332485:1332978 [3] NCCL INFO Channel 02/0 : 1[3] -> 0[2] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:1332485:1332978 [3] NCCL INFO Channel 03/0 : 1[3] -> 0[2] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:1332483:1332977 [1] NCCL INFO Channel 00/0 : 0[0] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:1332483:1332977 [1] NCCL INFO Channel 01/0 : 0[0] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:1332483:1332977 [1] NCCL INFO Channel 02/0 : 0[0] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:1332483:1332977 [1] NCCL INFO Channel 03/0 : 0[0] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:1332483:1332977 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[0] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:1332483:1332977 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[0] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:1332483:1332977 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[0] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:1332483:1332977 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[0] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:1332484:1332975 [2] NCCL INFO Channel 00/0 : 1[3] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:1332484:1332975 [2] NCCL INFO Channel 01/0 : 1[3] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:1332484:1332975 [2] NCCL INFO Channel 02/0 : 1[3] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:1332484:1332975 [2] NCCL INFO Channel 03/0 : 1[3] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:1332484:1332975 [2] NCCL INFO Channel 00/0 : 0[2] -> 1[3] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:1332484:1332975 [2] NCCL INFO Channel 01/0 : 0[2] -> 1[3] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:1332484:1332975 [2] NCCL INFO Channel 02/0 : 0[2] -> 1[3] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:1332484:1332975 [2] NCCL INFO Channel 03/0 : 0[2] -> 1[3] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:1332487:1332973 [5] NCCL INFO Channel 00/0 : 0[4] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:1332487:1332973 [5] NCCL INFO Channel 01/0 : 0[4] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:1332487:1332973 [5] NCCL INFO Channel 02/0 : 0[4] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:1332487:1332973 [5] NCCL INFO Channel 03/0 : 0[4] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:1332487:1332973 [5] NCCL INFO Channel 00/0 : 1[5] -> 0[4] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:1332487:1332973 [5] NCCL INFO Channel 01/0 : 1[5] -> 0[4] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:1332487:1332973 [5] NCCL INFO Channel 02/0 : 1[5] -> 0[4] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:1332489:1332979 [7] NCCL INFO Channel 00/0 : 0[6] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:1332487:1332973 [5] NCCL INFO Channel 03/0 : 1[5] -> 0[4] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:1332489:1332979 [7] NCCL INFO Channel 01/0 : 0[6] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:1332489:1332979 [7] NCCL INFO Channel 02/0 : 0[6] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:1332489:1332979 [7] NCCL INFO Channel 03/0 : 0[6] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:1332486:1332970 [4] NCCL INFO Channel 00/0 : 1[5] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:1332489:1332979 [7] NCCL INFO Channel 00/0 : 1[7] -> 0[6] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:1332486:1332970 [4] NCCL INFO Channel 01/0 : 1[5] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:1332489:1332979 [7] NCCL INFO Channel 01/0 : 1[7] -> 0[6] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:1332486:1332970 [4] NCCL INFO Channel 02/0 : 1[5] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:1332489:1332979 [7] NCCL INFO Channel 02/0 : 1[7] -> 0[6] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:1332489:1332979 [7] NCCL INFO Channel 03/0 : 1[7] -> 0[6] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:1332486:1332970 [4] NCCL INFO Channel 03/0 : 1[5] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:1332488:1332976 [6] NCCL INFO Channel 00/0 : 1[7] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:1332486:1332970 [4] NCCL INFO Channel 00/0 : 0[4] -> 1[5] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:1332486:1332970 [4] NCCL INFO Channel 01/0 : 0[4] -> 1[5] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:1332488:1332976 [6] NCCL INFO Channel 01/0 : 1[7] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:1332486:1332970 [4] NCCL INFO Channel 02/0 : 0[4] -> 1[5] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:1332488:1332976 [6] NCCL INFO Channel 02/0 : 1[7] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:1332486:1332970 [4] NCCL INFO Channel 03/0 : 0[4] -> 1[5] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:1332488:1332976 [6] NCCL INFO Channel 03/0 : 1[7] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:1332488:1332976 [6] NCCL INFO Channel 00/0 : 0[6] -> 1[7] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:1332488:1332976 [6] NCCL INFO Channel 01/0 : 0[6] -> 1[7] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:1332488:1332976 [6] NCCL INFO Channel 02/0 : 0[6] -> 1[7] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:1332488:1332976 [6] NCCL INFO Channel 03/0 : 0[6] -> 1[7] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:1061621:1062091 [1] NCCL INFO Connected all rings
+ip-26-0-166-36:1061621:1062091 [1] NCCL INFO Connected all trees
+ip-26-0-166-36:1061621:1062091 [1] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-36:1061621:1062091 [1] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-36:1061624:1062083 [4] NCCL INFO Connected all rings
+ip-26-0-166-36:1061624:1062083 [4] NCCL INFO Connected all trees
+ip-26-0-166-36:1061624:1062083 [4] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-36:1061624:1062083 [4] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-36:1061623:1062087 [3] NCCL INFO Connected all rings
+ip-26-0-166-36:1061623:1062087 [3] NCCL INFO Connected all trees
+ip-26-0-166-36:1061623:1062087 [3] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-36:1061623:1062087 [3] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-36:1061622:1062085 [2] NCCL INFO Connected all rings
+ip-26-0-166-36:1061622:1062085 [2] NCCL INFO Connected all trees
+ip-26-0-166-36:1061622:1062085 [2] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-36:1061622:1062085 [2] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-36:1061627:1062089 [7] NCCL INFO Connected all rings
+ip-26-0-166-36:1061627:1062089 [7] NCCL INFO Connected all trees
+ip-26-0-166-36:1061627:1062089 [7] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-36:1061627:1062089 [7] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-36:1061625:1062088 [5] NCCL INFO Connected all rings
+ip-26-0-166-36:1061625:1062088 [5] NCCL INFO Connected all trees
+ip-26-0-166-36:1061625:1062088 [5] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-36:1061625:1062088 [5] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-36:1061621:1062091 [1] NCCL INFO comm 0xb2cd190 rank 1 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0x1b1fe3d9b889f4e1 - Init COMPLETE
+ip-26-0-166-36:1061620:1062090 [0] NCCL INFO Connected all rings
+ip-26-0-166-36:1061620:1062090 [0] NCCL INFO Connected all trees
+ip-26-0-166-36:1061620:1062090 [0] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-36:1061620:1062090 [0] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-36:1061626:1062084 [6] NCCL INFO Connected all rings
+ip-26-0-166-36:1061626:1062084 [6] NCCL INFO Connected all trees
+ip-26-0-166-36:1061626:1062084 [6] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-36:1061626:1062084 [6] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-36:1061624:1062083 [4] NCCL INFO comm 0xbe380e0 rank 0 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0x640002f1b71c9ffa - Init COMPLETE
+ip-26-0-166-36:1061623:1062087 [3] NCCL INFO comm 0xa3c9190 rank 1 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0xcee17cd97bfa3d70 - Init COMPLETE
+ip-26-0-166-36:1061622:1062085 [2] NCCL INFO comm 0xae0a390 rank 0 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0xcee17cd97bfa3d70 - Init COMPLETE
+ip-26-0-166-36:1061627:1062089 [7] NCCL INFO comm 0xb1fae50 rank 1 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0x20b055b7295e97aa - Init COMPLETE
+ip-26-0-166-36:1061625:1062088 [5] NCCL INFO comm 0xafbc490 rank 1 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0x640002f1b71c9ffa - Init COMPLETE
+ip-26-0-166-36:1061620:1062090 [0] NCCL INFO comm 0xa611f00 rank 0 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0x1b1fe3d9b889f4e1 - Init COMPLETE
+ip-26-0-166-36:1061626:1062084 [6] NCCL INFO comm 0xa5763d0 rank 0 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0x20b055b7295e97aa - Init COMPLETE
+ip-26-0-166-125:1332483:1332977 [1] NCCL INFO Connected all rings
+ip-26-0-166-125:1332483:1332977 [1] NCCL INFO Connected all trees
+ip-26-0-166-125:1332483:1332977 [1] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-125:1332483:1332977 [1] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-125:1332482:1332974 [0] NCCL INFO Connected all rings
+ip-26-0-166-125:1332482:1332974 [0] NCCL INFO Connected all trees
+ip-26-0-166-125:1332482:1332974 [0] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-125:1332482:1332974 [0] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-125:1332485:1332978 [3] NCCL INFO Connected all rings
+ip-26-0-166-125:1332485:1332978 [3] NCCL INFO Connected all trees
+ip-26-0-166-125:1332485:1332978 [3] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-125:1332485:1332978 [3] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-125:1332483:1332977 [1] NCCL INFO comm 0xb2c5b40 rank 1 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0x3302783f73043d45 - Init COMPLETE
+ip-26-0-166-125:1332484:1332975 [2] NCCL INFO Connected all rings
+ip-26-0-166-125:1332484:1332975 [2] NCCL INFO Connected all trees
+ip-26-0-166-125:1332484:1332975 [2] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-125:1332484:1332975 [2] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+NCCL version 2.18.5+cuda12.2
+ip-26-0-166-125:1332487:1332973 [5] NCCL INFO Connected all rings
+ip-26-0-166-125:1332487:1332973 [5] NCCL INFO Connected all trees
+ip-26-0-166-125:1332489:1332979 [7] NCCL INFO Connected all rings
+ip-26-0-166-125:1332487:1332973 [5] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-125:1332489:1332979 [7] NCCL INFO Connected all trees
+ip-26-0-166-125:1332487:1332973 [5] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-125:1332489:1332979 [7] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-125:1332489:1332979 [7] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-125:1332483:1333001 [1] NCCL INFO Using network Libfabric
+ip-26-0-166-36:1061621:1062110 [1] NCCL INFO Using network Libfabric
+ip-26-0-166-125:1332486:1332970 [4] NCCL INFO Connected all rings
+ip-26-0-166-125:1332486:1332970 [4] NCCL INFO Connected all trees
+ip-26-0-166-125:1332486:1332970 [4] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-125:1332486:1332970 [4] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-125:1332488:1332976 [6] NCCL INFO Connected all rings
+ip-26-0-166-125:1332488:1332976 [6] NCCL INFO Connected all trees
+ip-26-0-166-125:1332488:1332976 [6] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-125:1332488:1332976 [6] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-36:1061621:1062110 [1] NCCL INFO comm 0xb2dfe50 rank 1 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0xdcbfb4bbd3bc53ef - Init START
+ip-26-0-166-125:1332483:1333001 [1] NCCL INFO comm 0xb2d8800 rank 0 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0xdcbfb4bbd3bc53ef - Init START
+ip-26-0-166-125:1332483:1333001 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-166-36:1061621:1062110 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-166-125:1332482:1332974 [0] NCCL INFO comm 0xb7c5c10 rank 0 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0x3302783f73043d45 - Init COMPLETE
+ip-26-0-166-125:1332482:1333003 [0] NCCL INFO Using network Libfabric
+ip-26-0-166-36:1061620:1062111 [0] NCCL INFO Using network Libfabric
+ip-26-0-166-36:1061620:1062111 [0] NCCL INFO comm 0xa624bc0 rank 1 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0x7fbd181fb8c465e2 - Init START
+ip-26-0-166-125:1332482:1333003 [0] NCCL INFO comm 0xb7d88d0 rank 0 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0x7fbd181fb8c465e2 - Init START
+ip-26-0-166-36:1061620:1062111 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-166-125:1332482:1333003 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/205
+ip-26-0-166-125:1332485:1332978 [3] NCCL INFO comm 0xa767840 rank 1 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0xfa7340fae4b966a6 - Init COMPLETE
+NCCL version 2.18.5+cuda12.2
+ip-26-0-166-125:1332485:1333005 [3] NCCL INFO Using network Libfabric
+ip-26-0-166-36:1061623:1062112 [3] NCCL INFO Using network Libfabric
+ip-26-0-166-36:1061623:1062112 [3] NCCL INFO comm 0xa3dbe50 rank 1 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0x635df9e042769414 - Init START
+ip-26-0-166-125:1332485:1333005 [3] NCCL INFO comm 0xa77a500 rank 0 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0x635df9e042769414 - Init START
+ip-26-0-166-36:1061623:1062112 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-166-125:1332485:1333005 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-166-125:1332487:1332973 [5] NCCL INFO comm 0xb9f8fc0 rank 1 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0x4cef1d6bf2e8f61 - Init COMPLETE
+ip-26-0-166-125:1332484:1332975 [2] NCCL INFO comm 0xb9ecf80 rank 0 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0xfa7340fae4b966a6 - Init COMPLETE
+ip-26-0-166-125:1332489:1332979 [7] NCCL INFO comm 0xbe7ef80 rank 1 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0xdc3c7ca3884008cc - Init COMPLETE
+NCCL version 2.18.5+cuda12.2
+ip-26-0-166-125:1332486:1332970 [4] NCCL INFO comm 0xb088290 rank 0 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0x4cef1d6bf2e8f61 - Init COMPLETE
+ip-26-0-166-125:1332488:1332976 [6] NCCL INFO comm 0xa7cdb80 rank 0 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0xdc3c7ca3884008cc - Init COMPLETE
+ip-26-0-166-125:1332487:1333007 [5] NCCL INFO Using network Libfabric
+ip-26-0-166-36:1061625:1062113 [5] NCCL INFO Using network Libfabric
+NCCL version 2.18.5+cuda12.2
+ip-26-0-166-125:1332484:1333009 [2] NCCL INFO Using network Libfabric
+ip-26-0-166-36:1061622:1062114 [2] NCCL INFO Using network Libfabric
+ip-26-0-166-36:1061625:1062113 [5] NCCL INFO comm 0xafcf150 rank 1 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0x7a9d6cbac2e3a486 - Init START
+ip-26-0-166-125:1332487:1333007 [5] NCCL INFO comm 0xba0bc80 rank 0 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0x7a9d6cbac2e3a486 - Init START
+ip-26-0-166-125:1332489:1333011 [7] NCCL INFO Using network Libfabric
+ip-26-0-166-36:1061627:1062115 [7] NCCL INFO Using network Libfabric
+ip-26-0-166-36:1061625:1062113 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-166-125:1332487:1333007 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-166-36:1061622:1062114 [2] NCCL INFO comm 0xae1d050 rank 1 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0x575448574b85ae3f - Init START
+ip-26-0-166-125:1332484:1333009 [2] NCCL INFO comm 0xb9ffc40 rank 0 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0x575448574b85ae3f - Init START
+ip-26-0-166-36:1061622:1062114 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-166-36:1061627:1062115 [7] NCCL INFO comm 0xb20db10 rank 1 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0x7ad8a342bd604d6e - Init START
+ip-26-0-166-125:1332486:1333014 [4] NCCL INFO Using network Libfabric
+ip-26-0-166-125:1332489:1333011 [7] NCCL INFO comm 0xbe91c40 rank 0 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0x7ad8a342bd604d6e - Init START
+ip-26-0-166-125:1332488:1333015 [6] NCCL INFO Using network Libfabric
+ip-26-0-166-125:1332484:1333009 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-166-36:1061624:1062116 [4] NCCL INFO Using network Libfabric
+ip-26-0-166-36:1061626:1062117 [6] NCCL INFO Using network Libfabric
+ip-26-0-166-36:1061627:1062115 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-166-125:1332489:1333011 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-166-36:1061624:1062116 [4] NCCL INFO comm 0xbe4ada0 rank 1 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0xe9ba90538709f333 - Init START
+ip-26-0-166-125:1332486:1333014 [4] NCCL INFO comm 0xb09af50 rank 0 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0xe9ba90538709f333 - Init START
+ip-26-0-166-36:1061626:1062117 [6] NCCL INFO comm 0xa589090 rank 1 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0x74e02cdae634742c - Init START
+ip-26-0-166-36:1061624:1062116 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-166-125:1332486:1333014 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-166-125:1332488:1333015 [6] NCCL INFO comm 0xa7e0840 rank 0 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0x74e02cdae634742c - Init START
+ip-26-0-166-36:1061626:1062117 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-166-125:1332488:1333015 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-166-36:1061620:1062111 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-166-125:1332482:1333003 [0] NCCL INFO Setting affinity for GPU 0 to 0fff,ffffffff
+ip-26-0-166-125:1332482:1333003 [0] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-166-125:1332482:1333003 [0] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-166-125:1332482:1333003 [0] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-166-36:1061620:1062111 [0] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-166-36:1061620:1062111 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-125:1332482:1333003 [0] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-166-125:1332482:1333003 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-166-125:1332482:1333003 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-36:1061621:1062110 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-166-36:1061623:1062112 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-166-36:1061625:1062113 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-166-36:1061620:1062111 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:1061620:1062111 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:1332482:1333003 [0] NCCL INFO Channel 00/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:1061620:1062111 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:1332482:1333003 [0] NCCL INFO Channel 01/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:1061620:1062111 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:1332482:1333003 [0] NCCL INFO Channel 02/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:1061620:1062111 [0] NCCL INFO Channel 00/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:1332482:1333003 [0] NCCL INFO Channel 03/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:1061620:1062111 [0] NCCL INFO Channel 01/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:1332482:1333003 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:1061620:1062111 [0] NCCL INFO Channel 02/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:1332482:1333003 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:1061620:1062111 [0] NCCL INFO Channel 03/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:1332482:1333003 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:1332482:1333003 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:1332489:1333011 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,fff00000,00000000
+ip-26-0-166-36:1061624:1062116 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-166-125:1332483:1333001 [1] NCCL INFO Setting affinity for GPU 1 to 0fff,ffffffff
+ip-26-0-166-125:1332483:1333001 [1] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-166-125:1332483:1333001 [1] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-166-125:1332483:1333001 [1] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-166-125:1332483:1333001 [1] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-166-36:1061621:1062110 [1] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-166-36:1061621:1062110 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-125:1332483:1333001 [1] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-166-125:1332483:1333001 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-36:1061626:1062117 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-166-36:1061627:1062115 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-166-36:1061622:1062114 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-166-36:1061627:1062115 [7] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-166-36:1061627:1062115 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-125:1332489:1333011 [7] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-166-125:1332489:1333011 [7] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-166-125:1332489:1333011 [7] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-166-125:1332489:1333011 [7] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-166-125:1332489:1333011 [7] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-166-125:1332489:1333011 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-125:1332485:1333005 [3] NCCL INFO Setting affinity for GPU 3 to 0fff,ffffffff
+ip-26-0-166-125:1332485:1333005 [3] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-166-125:1332485:1333005 [3] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-166-125:1332485:1333005 [3] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-166-125:1332485:1333005 [3] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-166-36:1061623:1062112 [3] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-166-36:1061623:1062112 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-125:1332485:1333005 [3] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-166-125:1332485:1333005 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-125:1332484:1333009 [2] NCCL INFO Setting affinity for GPU 2 to 0fff,ffffffff
+ip-26-0-166-125:1332484:1333009 [2] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-166-125:1332484:1333009 [2] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-166-125:1332484:1333009 [2] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-166-125:1332484:1333009 [2] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-166-125:1332484:1333009 [2] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-166-36:1061622:1062114 [2] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-166-36:1061622:1062114 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-125:1332484:1333009 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-125:1332487:1333007 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,fff00000,00000000
+ip-26-0-166-125:1332487:1333007 [5] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-166-125:1332487:1333007 [5] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-166-125:1332487:1333007 [5] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-166-125:1332487:1333007 [5] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-166-36:1061625:1062113 [5] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-166-36:1061625:1062113 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-125:1332487:1333007 [5] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-166-125:1332487:1333007 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-125:1332486:1333014 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,fff00000,00000000
+ip-26-0-166-125:1332488:1333015 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,fff00000,00000000
+ip-26-0-166-125:1332486:1333014 [4] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-166-125:1332486:1333014 [4] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-166-125:1332486:1333014 [4] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-166-125:1332486:1333014 [4] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-166-36:1061624:1062116 [4] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-166-36:1061624:1062116 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-125:1332486:1333014 [4] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-166-125:1332486:1333014 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-125:1332488:1333015 [6] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-166-125:1332488:1333015 [6] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-166-125:1332488:1333015 [6] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-166-125:1332488:1333015 [6] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-166-36:1061626:1062117 [6] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-166-36:1061626:1062117 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-125:1332488:1333015 [6] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-166-125:1332488:1333015 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-36:1061621:1062110 [1] NCCL INFO Channel 00/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:1332483:1333001 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:1061621:1062110 [1] NCCL INFO Channel 01/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:1332483:1333001 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:1061621:1062110 [1] NCCL INFO Channel 02/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:1332483:1333001 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:1061621:1062110 [1] NCCL INFO Channel 03/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:1332483:1333001 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:1061621:1062110 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:1332483:1333001 [1] NCCL INFO Channel 00/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:1061621:1062110 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:1332483:1333001 [1] NCCL INFO Channel 01/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:1061621:1062110 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:1332483:1333001 [1] NCCL INFO Channel 02/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:1061621:1062110 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:1332483:1333001 [1] NCCL INFO Channel 03/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:1061627:1062115 [7] NCCL INFO Channel 00/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:1061627:1062115 [7] NCCL INFO Channel 01/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:1061627:1062115 [7] NCCL INFO Channel 02/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:1061627:1062115 [7] NCCL INFO Channel 03/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:1061627:1062115 [7] NCCL INFO Channel 00/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:1061627:1062115 [7] NCCL INFO Channel 01/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:1061627:1062115 [7] NCCL INFO Channel 02/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:1061627:1062115 [7] NCCL INFO Channel 03/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:1332489:1333011 [7] NCCL INFO Channel 00/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:1332489:1333011 [7] NCCL INFO Channel 01/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:1332489:1333011 [7] NCCL INFO Channel 02/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:1332489:1333011 [7] NCCL INFO Channel 03/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:1332489:1333011 [7] NCCL INFO Channel 00/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:1332489:1333011 [7] NCCL INFO Channel 01/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:1332489:1333011 [7] NCCL INFO Channel 02/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:1332489:1333011 [7] NCCL INFO Channel 03/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:1332485:1333005 [3] NCCL INFO Channel 00/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:1061623:1062112 [3] NCCL INFO Channel 00/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:1332485:1333005 [3] NCCL INFO Channel 01/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:1061623:1062112 [3] NCCL INFO Channel 01/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:1061623:1062112 [3] NCCL INFO Channel 02/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:1332485:1333005 [3] NCCL INFO Channel 02/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:1061623:1062112 [3] NCCL INFO Channel 03/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:1332485:1333005 [3] NCCL INFO Channel 03/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:1061623:1062112 [3] NCCL INFO Channel 00/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:1332485:1333005 [3] NCCL INFO Channel 00/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:1061623:1062112 [3] NCCL INFO Channel 01/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:1332485:1333005 [3] NCCL INFO Channel 01/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:1061623:1062112 [3] NCCL INFO Channel 02/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:1332485:1333005 [3] NCCL INFO Channel 02/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:1061623:1062112 [3] NCCL INFO Channel 03/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:1332485:1333005 [3] NCCL INFO Channel 03/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:1332484:1333009 [2] NCCL INFO Channel 00/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:1061622:1062114 [2] NCCL INFO Channel 00/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:1332484:1333009 [2] NCCL INFO Channel 01/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:1061622:1062114 [2] NCCL INFO Channel 01/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:1332484:1333009 [2] NCCL INFO Channel 02/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:1061622:1062114 [2] NCCL INFO Channel 02/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:1332484:1333009 [2] NCCL INFO Channel 03/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:1061622:1062114 [2] NCCL INFO Channel 03/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:1332484:1333009 [2] NCCL INFO Channel 00/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:1061622:1062114 [2] NCCL INFO Channel 00/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:1332484:1333009 [2] NCCL INFO Channel 01/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:1061622:1062114 [2] NCCL INFO Channel 01/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:1332484:1333009 [2] NCCL INFO Channel 02/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:1061622:1062114 [2] NCCL INFO Channel 02/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:1332484:1333009 [2] NCCL INFO Channel 03/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:1061622:1062114 [2] NCCL INFO Channel 03/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:1061625:1062113 [5] NCCL INFO Channel 00/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:1061625:1062113 [5] NCCL INFO Channel 01/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:1061625:1062113 [5] NCCL INFO Channel 02/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:1061625:1062113 [5] NCCL INFO Channel 03/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:1061625:1062113 [5] NCCL INFO Channel 00/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:1061625:1062113 [5] NCCL INFO Channel 01/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:1061625:1062113 [5] NCCL INFO Channel 02/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:1061625:1062113 [5] NCCL INFO Channel 03/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:1332487:1333007 [5] NCCL INFO Channel 00/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:1332487:1333007 [5] NCCL INFO Channel 01/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:1332487:1333007 [5] NCCL INFO Channel 02/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:1332487:1333007 [5] NCCL INFO Channel 03/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:1332487:1333007 [5] NCCL INFO Channel 00/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:1332487:1333007 [5] NCCL INFO Channel 01/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:1332487:1333007 [5] NCCL INFO Channel 02/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:1332487:1333007 [5] NCCL INFO Channel 03/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:1061624:1062116 [4] NCCL INFO Channel 00/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:1061624:1062116 [4] NCCL INFO Channel 01/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:1061624:1062116 [4] NCCL INFO Channel 02/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:1061624:1062116 [4] NCCL INFO Channel 03/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:1061624:1062116 [4] NCCL INFO Channel 00/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:1061624:1062116 [4] NCCL INFO Channel 01/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:1061624:1062116 [4] NCCL INFO Channel 02/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:1061624:1062116 [4] NCCL INFO Channel 03/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:1061626:1062117 [6] NCCL INFO Channel 00/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:1061626:1062117 [6] NCCL INFO Channel 01/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:1061626:1062117 [6] NCCL INFO Channel 02/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:1061626:1062117 [6] NCCL INFO Channel 03/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:1061626:1062117 [6] NCCL INFO Channel 00/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:1061626:1062117 [6] NCCL INFO Channel 01/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:1061626:1062117 [6] NCCL INFO Channel 02/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:1061626:1062117 [6] NCCL INFO Channel 03/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:1332488:1333015 [6] NCCL INFO Channel 00/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:1332488:1333015 [6] NCCL INFO Channel 01/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:1332488:1333015 [6] NCCL INFO Channel 02/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:1332488:1333015 [6] NCCL INFO Channel 03/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:1332488:1333015 [6] NCCL INFO Channel 00/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:1332488:1333015 [6] NCCL INFO Channel 01/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:1332486:1333014 [4] NCCL INFO Channel 00/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:1332488:1333015 [6] NCCL INFO Channel 02/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:1332486:1333014 [4] NCCL INFO Channel 01/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:1332488:1333015 [6] NCCL INFO Channel 03/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:1332486:1333014 [4] NCCL INFO Channel 02/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:1332486:1333014 [4] NCCL INFO Channel 03/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:1332486:1333014 [4] NCCL INFO Channel 00/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:1332486:1333014 [4] NCCL INFO Channel 01/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:1332486:1333014 [4] NCCL INFO Channel 02/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:1332486:1333014 [4] NCCL INFO Channel 03/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:1332482:1333003 [0] NCCL INFO Connected all rings
+ip-26-0-166-125:1332482:1333003 [0] NCCL INFO Connected all trees
+ip-26-0-166-125:1332482:1333003 [0] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-125:1332482:1333003 [0] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-36:1061620:1062111 [0] NCCL INFO Connected all rings
+ip-26-0-166-36:1061620:1062111 [0] NCCL INFO Connected all trees
+ip-26-0-166-36:1061620:1062111 [0] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-36:1061620:1062111 [0] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-125:1332482:1333003 [0] NCCL INFO comm 0xb7d88d0 rank 0 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0x7fbd181fb8c465e2 - Init COMPLETE
+ip-26-0-166-36:1061620:1062111 [0] NCCL INFO comm 0xa624bc0 rank 1 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0x7fbd181fb8c465e2 - Init COMPLETE
+ip-26-0-166-125:1332483:1333001 [1] NCCL INFO Connected all rings
+ip-26-0-166-125:1332483:1333001 [1] NCCL INFO Connected all trees
+ip-26-0-166-125:1332483:1333001 [1] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-125:1332483:1333001 [1] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-36:1061621:1062110 [1] NCCL INFO Connected all rings
+ip-26-0-166-36:1061621:1062110 [1] NCCL INFO Connected all trees
+ip-26-0-166-36:1061621:1062110 [1] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-36:1061621:1062110 [1] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-36:1061623:1062112 [3] NCCL INFO Connected all rings
+ip-26-0-166-36:1061623:1062112 [3] NCCL INFO Connected all trees
+ip-26-0-166-36:1061623:1062112 [3] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-36:1061623:1062112 [3] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-125:1332489:1333011 [7] NCCL INFO Connected all rings
+ip-26-0-166-125:1332489:1333011 [7] NCCL INFO Connected all trees
+ip-26-0-166-125:1332489:1333011 [7] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-125:1332489:1333011 [7] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-125:1332483:1333001 [1] NCCL INFO comm 0xb2d8800 rank 0 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0xdcbfb4bbd3bc53ef - Init COMPLETE
+ip-26-0-166-36:1061627:1062115 [7] NCCL INFO Connected all rings
+ip-26-0-166-36:1061627:1062115 [7] NCCL INFO Connected all trees
+ip-26-0-166-36:1061627:1062115 [7] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-36:1061627:1062115 [7] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-125:1332485:1333005 [3] NCCL INFO Connected all rings
+ip-26-0-166-125:1332485:1333005 [3] NCCL INFO Connected all trees
+ip-26-0-166-125:1332485:1333005 [3] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-125:1332485:1333005 [3] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-36:1061621:1062110 [1] NCCL INFO comm 0xb2dfe50 rank 1 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0xdcbfb4bbd3bc53ef - Init COMPLETE
+ip-26-0-166-125:1332489:1333011 [7] NCCL INFO comm 0xbe91c40 rank 0 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0x7ad8a342bd604d6e - Init COMPLETE
+ip-26-0-166-36:1061623:1062112 [3] NCCL INFO comm 0xa3dbe50 rank 1 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0x635df9e042769414 - Init COMPLETE
+ip-26-0-166-125:1332485:1333005 [3] NCCL INFO comm 0xa77a500 rank 0 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0x635df9e042769414 - Init COMPLETE
+ip-26-0-166-36:1061627:1062115 [7] NCCL INFO comm 0xb20db10 rank 1 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0x7ad8a342bd604d6e - Init COMPLETE
+ip-26-0-166-125:1332484:1333009 [2] NCCL INFO Connected all rings
+ip-26-0-166-125:1332484:1333009 [2] NCCL INFO Connected all trees
+ip-26-0-166-125:1332484:1333009 [2] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-125:1332484:1333009 [2] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-125:1332487:1333007 [5] NCCL INFO Connected all rings
+ip-26-0-166-125:1332487:1333007 [5] NCCL INFO Connected all trees
+ip-26-0-166-125:1332487:1333007 [5] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-125:1332487:1333007 [5] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-36:1061625:1062113 [5] NCCL INFO Connected all rings
+ip-26-0-166-36:1061625:1062113 [5] NCCL INFO Connected all trees
+ip-26-0-166-36:1061625:1062113 [5] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-36:1061625:1062113 [5] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-36:1061622:1062114 [2] NCCL INFO Connected all rings
+ip-26-0-166-36:1061622:1062114 [2] NCCL INFO Connected all trees
+ip-26-0-166-36:1061622:1062114 [2] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-36:1061622:1062114 [2] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-36:1061624:1062116 [4] NCCL INFO Connected all rings
+ip-26-0-166-36:1061624:1062116 [4] NCCL INFO Connected all trees
+ip-26-0-166-36:1061624:1062116 [4] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-36:1061624:1062116 [4] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-125:1332486:1333014 [4] NCCL INFO Connected all rings
+ip-26-0-166-125:1332486:1333014 [4] NCCL INFO Connected all trees
+ip-26-0-166-125:1332486:1333014 [4] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-125:1332486:1333014 [4] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-125:1332484:1333009 [2] NCCL INFO comm 0xb9ffc40 rank 0 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0x575448574b85ae3f - Init COMPLETE
+ip-26-0-166-36:1061626:1062117 [6] NCCL INFO Connected all rings
+ip-26-0-166-36:1061626:1062117 [6] NCCL INFO Connected all trees
+ip-26-0-166-36:1061626:1062117 [6] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-36:1061626:1062117 [6] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-125:1332488:1333015 [6] NCCL INFO Connected all rings
+ip-26-0-166-125:1332488:1333015 [6] NCCL INFO Connected all trees
+ip-26-0-166-125:1332488:1333015 [6] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-125:1332488:1333015 [6] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-36:1061625:1062113 [5] NCCL INFO comm 0xafcf150 rank 1 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0x7a9d6cbac2e3a486 - Init COMPLETE
+ip-26-0-166-125:1332487:1333007 [5] NCCL INFO comm 0xba0bc80 rank 0 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0x7a9d6cbac2e3a486 - Init COMPLETE
+ip-26-0-166-36:1061622:1062114 [2] NCCL INFO comm 0xae1d050 rank 1 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0x575448574b85ae3f - Init COMPLETE
+ip-26-0-166-36:1061624:1062116 [4] NCCL INFO comm 0xbe4ada0 rank 1 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0xe9ba90538709f333 - Init COMPLETE
+ip-26-0-166-36:1061626:1062117 [6] NCCL INFO comm 0xa589090 rank 1 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0x74e02cdae634742c - Init COMPLETE
+ip-26-0-166-125:1332486:1333014 [4] NCCL INFO comm 0xb09af50 rank 0 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0xe9ba90538709f333 - Init COMPLETE
+ip-26-0-166-125:1332488:1333015 [6] NCCL INFO comm 0xa7e0840 rank 0 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0x74e02cdae634742c - Init COMPLETE
+ip-26-0-166-36:1061620:1062136 [0] NCCL INFO Using network Libfabric
+ip-26-0-166-125:1332482:1333034 [0] NCCL INFO Using network Libfabric
+ip-26-0-166-125:1332488:1333036 [6] NCCL INFO Using network Libfabric
+ip-26-0-166-125:1332486:1333035 [4] NCCL INFO Using network Libfabric
+ip-26-0-166-36:1061626:1062137 [6] NCCL INFO Using network Libfabric
+ip-26-0-166-125:1332484:1333037 [2] NCCL INFO Using network Libfabric
+ip-26-0-166-36:1061622:1062139 [2] NCCL INFO Using network Libfabric
+ip-26-0-166-36:1061624:1062138 [4] NCCL INFO Using network Libfabric
+ip-26-0-166-125:1332488:1333036 [6] NCCL INFO comm 0xa7f5510 rank 3 nranks 4 cudaDev 6 nvmlDev 6 busId b9000 commId 0xcf3103ca05a405d6 - Init START
+ip-26-0-166-125:1332486:1333035 [4] NCCL INFO comm 0xb0adc10 rank 2 nranks 4 cudaDev 4 nvmlDev 4 busId 97000 commId 0xcf3103ca05a405d6 - Init START
+ip-26-0-166-36:1061620:1062136 [0] NCCL INFO comm 0xa7f6770 rank 0 nranks 4 cudaDev 0 nvmlDev 0 busId 53000 commId 0x5af0260af03ae61d - Init START
+ip-26-0-166-36:1061622:1062139 [2] NCCL INFO comm 0xafef960 rank 1 nranks 4 cudaDev 2 nvmlDev 2 busId 75000 commId 0x5af0260af03ae61d - Init START
+ip-26-0-166-36:1061624:1062138 [4] NCCL INFO comm 0xc01b970 rank 2 nranks 4 cudaDev 4 nvmlDev 4 busId 97000 commId 0x5af0260af03ae61d - Init START
+ip-26-0-166-36:1061626:1062137 [6] NCCL INFO comm 0xa75c150 rank 3 nranks 4 cudaDev 6 nvmlDev 6 busId b9000 commId 0x5af0260af03ae61d - Init START
+ip-26-0-166-125:1332482:1333034 [0] NCCL INFO comm 0xb7eb590 rank 0 nranks 4 cudaDev 0 nvmlDev 0 busId 53000 commId 0xcf3103ca05a405d6 - Init START
+ip-26-0-166-125:1332484:1333037 [2] NCCL INFO comm 0xba13110 rank 1 nranks 4 cudaDev 2 nvmlDev 2 busId 75000 commId 0xcf3103ca05a405d6 - Init START
+ip-26-0-166-125:1332482:1333034 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/205
+ip-26-0-166-125:1332488:1333036 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-166-125:1332486:1333035 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-166-125:1332484:1333037 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-166-36:1061622:1062139 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-166-36:1061620:1062136 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-166-36:1061624:1062138 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-166-36:1061626:1062137 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+NCCL version 2.18.5+cuda12.2
+ip-26-0-166-125:1332483:1333039 [1] NCCL INFO Using network Libfabric
+ip-26-0-166-125:1332485:1333040 [3] NCCL INFO Using network Libfabric
+ip-26-0-166-125:1332489:1333042 [7] NCCL INFO Using network Libfabric
+ip-26-0-166-125:1332487:1333041 [5] NCCL INFO Using network Libfabric
+ip-26-0-166-36:1061621:1062141 [1] NCCL INFO Using network Libfabric
+ip-26-0-166-36:1061625:1062143 [5] NCCL INFO Using network Libfabric
+ip-26-0-166-36:1061623:1062144 [3] NCCL INFO Using network Libfabric
+ip-26-0-166-36:1061627:1062142 [7] NCCL INFO Using network Libfabric
+ip-26-0-166-125:1332487:1333041 [5] NCCL INFO comm 0xba1e940 rank 2 nranks 4 cudaDev 5 nvmlDev 5 busId a8000 commId 0xc1f08c305513d36b - Init START
+ip-26-0-166-125:1332483:1333039 [1] NCCL INFO comm 0xb2eb4c0 rank 0 nranks 4 cudaDev 1 nvmlDev 1 busId 64000 commId 0xc1f08c305513d36b - Init START
+ip-26-0-166-125:1332485:1333040 [3] NCCL INFO comm 0xa78d1c0 rank 1 nranks 4 cudaDev 3 nvmlDev 3 busId 86000 commId 0xc1f08c305513d36b - Init START
+ip-26-0-166-125:1332489:1333042 [7] NCCL INFO comm 0xbea4900 rank 3 nranks 4 cudaDev 7 nvmlDev 7 busId ca000 commId 0xc1f08c305513d36b - Init START
+ip-26-0-166-125:1332489:1333042 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-166-125:1332487:1333041 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-166-125:1332483:1333039 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-166-125:1332485:1333040 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-166-36:1061623:1062144 [3] NCCL INFO comm 0xa5ac5d0 rank 1 nranks 4 cudaDev 3 nvmlDev 3 busId 86000 commId 0xe7283ca9e98730a - Init START
+ip-26-0-166-36:1061625:1062143 [5] NCCL INFO comm 0xb1a15c0 rank 2 nranks 4 cudaDev 5 nvmlDev 5 busId a8000 commId 0xe7283ca9e98730a - Init START
+ip-26-0-166-36:1061627:1062142 [7] NCCL INFO comm 0xb3df350 rank 3 nranks 4 cudaDev 7 nvmlDev 7 busId ca000 commId 0xe7283ca9e98730a - Init START
+ip-26-0-166-36:1061621:1062141 [1] NCCL INFO comm 0xb4b1f60 rank 0 nranks 4 cudaDev 1 nvmlDev 1 busId 64000 commId 0xe7283ca9e98730a - Init START
+ip-26-0-166-36:1061627:1062142 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-166-36:1061623:1062144 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-166-36:1061625:1062143 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-166-36:1061621:1062141 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-166-36:1061623:1062144 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-166-125:1332485:1333040 [3] NCCL INFO Setting affinity for GPU 3 to 0fff,ffffffff
+ip-26-0-166-36:1061622:1062139 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-166-36:1061620:1062136 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-166-125:1332488:1333036 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,fff00000,00000000
+ip-26-0-166-36:1061625:1062143 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-166-36:1061624:1062138 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-166-36:1061626:1062137 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-166-36:1061626:1062137 [6] NCCL INFO Trees [0] -1/-1/-1->3->2 [1] -1/-1/-1->3->2 [2] 1/-1/-1->3->-1 [3] 1/-1/-1->3->-1
+ip-26-0-166-36:1061626:1062137 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-36:1061620:1062136 [0] NCCL INFO Channel 00/04 :    0   1   2   3
+ip-26-0-166-36:1061622:1062139 [2] NCCL INFO Trees [0] -1/-1/-1->1->2 [1] -1/-1/-1->1->2 [2] 2/0/-1->1->3 [3] 2/0/-1->1->3
+ip-26-0-166-36:1061622:1062139 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-36:1061620:1062136 [0] NCCL INFO Channel 01/04 :    0   1   2   3
+ip-26-0-166-36:1061624:1062138 [4] NCCL INFO Trees [0] 1/3/-1->2->0 [1] 1/3/-1->2->0 [2] -1/-1/-1->2->1 [3] -1/-1/-1->2->1
+ip-26-0-166-36:1061620:1062136 [0] NCCL INFO Channel 02/04 :    0   1   2   3
+ip-26-0-166-36:1061620:1062136 [0] NCCL INFO Channel 03/04 :    0   1   2   3
+ip-26-0-166-36:1061624:1062138 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-36:1061620:1062136 [0] NCCL INFO Trees [0] 2/-1/-1->0->-1 [1] 2/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-166-36:1061620:1062136 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-125:1332487:1333041 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,fff00000,00000000
+ip-26-0-166-36:1061621:1062141 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-166-125:1332483:1333039 [1] NCCL INFO Setting affinity for GPU 1 to 0fff,ffffffff
+ip-26-0-166-36:1061627:1062142 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-166-36:1061627:1062142 [7] NCCL INFO Trees [0] -1/-1/-1->3->2 [1] -1/-1/-1->3->2 [2] 1/-1/-1->3->-1 [3] 1/-1/-1->3->-1
+ip-26-0-166-36:1061627:1062142 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-36:1061625:1062143 [5] NCCL INFO Trees [0] 1/3/-1->2->0 [1] 1/3/-1->2->0 [2] -1/-1/-1->2->1 [3] -1/-1/-1->2->1
+ip-26-0-166-36:1061621:1062141 [1] NCCL INFO Channel 00/04 :    0   1   2   3
+ip-26-0-166-36:1061625:1062143 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-36:1061623:1062144 [3] NCCL INFO Trees [0] -1/-1/-1->1->2 [1] -1/-1/-1->1->2 [2] 2/0/-1->1->3 [3] 2/0/-1->1->3
+ip-26-0-166-36:1061621:1062141 [1] NCCL INFO Channel 01/04 :    0   1   2   3
+ip-26-0-166-36:1061621:1062141 [1] NCCL INFO Channel 02/04 :    0   1   2   3
+ip-26-0-166-36:1061623:1062144 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-36:1061621:1062141 [1] NCCL INFO Channel 03/04 :    0   1   2   3
+ip-26-0-166-36:1061621:1062141 [1] NCCL INFO Trees [0] 2/-1/-1->0->-1 [1] 2/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-166-36:1061621:1062141 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-125:1332482:1333034 [0] NCCL INFO Setting affinity for GPU 0 to 0fff,ffffffff
+ip-26-0-166-36:1061620:1062136 [0] NCCL INFO Channel 00/0 : 3[6] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:1061620:1062136 [0] NCCL INFO Channel 01/0 : 3[6] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:1061620:1062136 [0] NCCL INFO Channel 02/0 : 3[6] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:1061620:1062136 [0] NCCL INFO Channel 03/0 : 3[6] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:1061620:1062136 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[2] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:1061622:1062139 [2] NCCL INFO Channel 00/0 : 0[0] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:1061620:1062136 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[2] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:1061620:1062136 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[2] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:1061622:1062139 [2] NCCL INFO Channel 01/0 : 0[0] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:1061620:1062136 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[2] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:1061622:1062139 [2] NCCL INFO Channel 02/0 : 0[0] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:1061622:1062139 [2] NCCL INFO Channel 03/0 : 0[0] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:1061622:1062139 [2] NCCL INFO Channel 00/0 : 1[2] -> 2[4] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:1061622:1062139 [2] NCCL INFO Channel 01/0 : 1[2] -> 2[4] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:1061626:1062137 [6] NCCL INFO Channel 00/0 : 2[4] -> 3[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:1061622:1062139 [2] NCCL INFO Channel 02/0 : 1[2] -> 2[4] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:1061622:1062139 [2] NCCL INFO Channel 03/0 : 1[2] -> 2[4] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:1061626:1062137 [6] NCCL INFO Channel 01/0 : 2[4] -> 3[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:1061626:1062137 [6] NCCL INFO Channel 02/0 : 2[4] -> 3[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:1061626:1062137 [6] NCCL INFO Channel 03/0 : 2[4] -> 3[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:1061626:1062137 [6] NCCL INFO Channel 00/0 : 3[6] -> 0[0] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:1061626:1062137 [6] NCCL INFO Channel 01/0 : 3[6] -> 0[0] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:1061626:1062137 [6] NCCL INFO Channel 02/0 : 3[6] -> 0[0] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:1061626:1062137 [6] NCCL INFO Channel 03/0 : 3[6] -> 0[0] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:1061624:1062138 [4] NCCL INFO Channel 00/0 : 1[2] -> 2[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:1061624:1062138 [4] NCCL INFO Channel 01/0 : 1[2] -> 2[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:1061624:1062138 [4] NCCL INFO Channel 02/0 : 1[2] -> 2[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:1061624:1062138 [4] NCCL INFO Channel 03/0 : 1[2] -> 2[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:1061624:1062138 [4] NCCL INFO Channel 00/0 : 2[4] -> 3[6] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:1061624:1062138 [4] NCCL INFO Channel 01/0 : 2[4] -> 3[6] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:1061624:1062138 [4] NCCL INFO Channel 02/0 : 2[4] -> 3[6] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:1061624:1062138 [4] NCCL INFO Channel 03/0 : 2[4] -> 3[6] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:1332486:1333035 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,fff00000,00000000
+ip-26-0-166-125:1332484:1333037 [2] NCCL INFO Setting affinity for GPU 2 to 0fff,ffffffff
+ip-26-0-166-125:1332482:1333034 [0] NCCL INFO Channel 00/04 :    0   1   2   3
+ip-26-0-166-125:1332482:1333034 [0] NCCL INFO Channel 01/04 :    0   1   2   3
+ip-26-0-166-125:1332482:1333034 [0] NCCL INFO Channel 02/04 :    0   1   2   3
+ip-26-0-166-125:1332486:1333035 [4] NCCL INFO Trees [0] 1/3/-1->2->0 [1] 1/3/-1->2->0 [2] -1/-1/-1->2->1 [3] -1/-1/-1->2->1
+ip-26-0-166-125:1332482:1333034 [0] NCCL INFO Channel 03/04 :    0   1   2   3
+ip-26-0-166-125:1332484:1333037 [2] NCCL INFO Trees [0] -1/-1/-1->1->2 [1] -1/-1/-1->1->2 [2] 2/0/-1->1->3 [3] 2/0/-1->1->3
+ip-26-0-166-125:1332486:1333035 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-125:1332482:1333034 [0] NCCL INFO Trees [0] 2/-1/-1->0->-1 [1] 2/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-166-125:1332488:1333036 [6] NCCL INFO Trees [0] -1/-1/-1->3->2 [1] -1/-1/-1->3->2 [2] 1/-1/-1->3->-1 [3] 1/-1/-1->3->-1
+ip-26-0-166-125:1332484:1333037 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-125:1332482:1333034 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-125:1332488:1333036 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-125:1332489:1333042 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,fff00000,00000000
+ip-26-0-166-125:1332489:1333042 [7] NCCL INFO Trees [0] -1/-1/-1->3->2 [1] -1/-1/-1->3->2 [2] 1/-1/-1->3->-1 [3] 1/-1/-1->3->-1
+ip-26-0-166-125:1332489:1333042 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-125:1332483:1333039 [1] NCCL INFO Channel 00/04 :    0   1   2   3
+ip-26-0-166-125:1332487:1333041 [5] NCCL INFO Trees [0] 1/3/-1->2->0 [1] 1/3/-1->2->0 [2] -1/-1/-1->2->1 [3] -1/-1/-1->2->1
+ip-26-0-166-125:1332483:1333039 [1] NCCL INFO Channel 01/04 :    0   1   2   3
+ip-26-0-166-125:1332487:1333041 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-125:1332483:1333039 [1] NCCL INFO Channel 02/04 :    0   1   2   3
+ip-26-0-166-125:1332483:1333039 [1] NCCL INFO Channel 03/04 :    0   1   2   3
+ip-26-0-166-125:1332485:1333040 [3] NCCL INFO Trees [0] -1/-1/-1->1->2 [1] -1/-1/-1->1->2 [2] 2/0/-1->1->3 [3] 2/0/-1->1->3
+ip-26-0-166-125:1332483:1333039 [1] NCCL INFO Trees [0] 2/-1/-1->0->-1 [1] 2/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-166-125:1332485:1333040 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-125:1332483:1333039 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-36:1061623:1062144 [3] NCCL INFO Channel 00/0 : 0[1] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:1061623:1062144 [3] NCCL INFO Channel 01/0 : 0[1] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:1061623:1062144 [3] NCCL INFO Channel 02/0 : 0[1] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:1061623:1062144 [3] NCCL INFO Channel 03/0 : 0[1] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:1061623:1062144 [3] NCCL INFO Channel 00/0 : 1[3] -> 2[5] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:1061623:1062144 [3] NCCL INFO Channel 01/0 : 1[3] -> 2[5] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:1061623:1062144 [3] NCCL INFO Channel 02/0 : 1[3] -> 2[5] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:1061623:1062144 [3] NCCL INFO Channel 03/0 : 1[3] -> 2[5] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:1061625:1062143 [5] NCCL INFO Channel 00/0 : 1[3] -> 2[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:1061625:1062143 [5] NCCL INFO Channel 01/0 : 1[3] -> 2[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:1061625:1062143 [5] NCCL INFO Channel 02/0 : 1[3] -> 2[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:1061625:1062143 [5] NCCL INFO Channel 03/0 : 1[3] -> 2[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:1061625:1062143 [5] NCCL INFO Channel 00/0 : 2[5] -> 3[7] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:1061621:1062141 [1] NCCL INFO Channel 00/0 : 3[7] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:1061627:1062142 [7] NCCL INFO Channel 00/0 : 2[5] -> 3[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:1061625:1062143 [5] NCCL INFO Channel 01/0 : 2[5] -> 3[7] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:1061621:1062141 [1] NCCL INFO Channel 01/0 : 3[7] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:1061625:1062143 [5] NCCL INFO Channel 02/0 : 2[5] -> 3[7] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:1061627:1062142 [7] NCCL INFO Channel 01/0 : 2[5] -> 3[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:1061625:1062143 [5] NCCL INFO Channel 03/0 : 2[5] -> 3[7] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:1061621:1062141 [1] NCCL INFO Channel 02/0 : 3[7] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:1061627:1062142 [7] NCCL INFO Channel 02/0 : 2[5] -> 3[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:1061621:1062141 [1] NCCL INFO Channel 03/0 : 3[7] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:1061627:1062142 [7] NCCL INFO Channel 03/0 : 2[5] -> 3[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:1061621:1062141 [1] NCCL INFO Channel 00/0 : 0[1] -> 1[3] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:1061627:1062142 [7] NCCL INFO Channel 00/0 : 3[7] -> 0[1] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:1061621:1062141 [1] NCCL INFO Channel 01/0 : 0[1] -> 1[3] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:1061627:1062142 [7] NCCL INFO Channel 01/0 : 3[7] -> 0[1] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:1061627:1062142 [7] NCCL INFO Channel 02/0 : 3[7] -> 0[1] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:1061621:1062141 [1] NCCL INFO Channel 02/0 : 0[1] -> 1[3] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:1061627:1062142 [7] NCCL INFO Channel 03/0 : 3[7] -> 0[1] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:1061621:1062141 [1] NCCL INFO Channel 03/0 : 0[1] -> 1[3] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:1332484:1333037 [2] NCCL INFO Channel 00/0 : 0[0] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:1332484:1333037 [2] NCCL INFO Channel 01/0 : 0[0] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:1332484:1333037 [2] NCCL INFO Channel 02/0 : 0[0] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:1332484:1333037 [2] NCCL INFO Channel 03/0 : 0[0] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:1332484:1333037 [2] NCCL INFO Channel 00/0 : 1[2] -> 2[4] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:1332488:1333036 [6] NCCL INFO Channel 00/0 : 2[4] -> 3[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:1332482:1333034 [0] NCCL INFO Channel 00/0 : 3[6] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:1332484:1333037 [2] NCCL INFO Channel 01/0 : 1[2] -> 2[4] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:1332486:1333035 [4] NCCL INFO Channel 00/0 : 1[2] -> 2[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:1332488:1333036 [6] NCCL INFO Channel 01/0 : 2[4] -> 3[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:1332484:1333037 [2] NCCL INFO Channel 02/0 : 1[2] -> 2[4] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:1332482:1333034 [0] NCCL INFO Channel 01/0 : 3[6] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:1332486:1333035 [4] NCCL INFO Channel 01/0 : 1[2] -> 2[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:1332488:1333036 [6] NCCL INFO Channel 02/0 : 2[4] -> 3[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:1332484:1333037 [2] NCCL INFO Channel 03/0 : 1[2] -> 2[4] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:1332482:1333034 [0] NCCL INFO Channel 02/0 : 3[6] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:1332486:1333035 [4] NCCL INFO Channel 02/0 : 1[2] -> 2[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:1332488:1333036 [6] NCCL INFO Channel 03/0 : 2[4] -> 3[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:1332482:1333034 [0] NCCL INFO Channel 03/0 : 3[6] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:1332486:1333035 [4] NCCL INFO Channel 03/0 : 1[2] -> 2[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:1332488:1333036 [6] NCCL INFO Channel 00/0 : 3[6] -> 0[0] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:1332482:1333034 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[2] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:1332486:1333035 [4] NCCL INFO Channel 00/0 : 2[4] -> 3[6] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:1332488:1333036 [6] NCCL INFO Channel 01/0 : 3[6] -> 0[0] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:1332482:1333034 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[2] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:1332486:1333035 [4] NCCL INFO Channel 01/0 : 2[4] -> 3[6] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:1332488:1333036 [6] NCCL INFO Channel 02/0 : 3[6] -> 0[0] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:1332482:1333034 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[2] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:1332486:1333035 [4] NCCL INFO Channel 02/0 : 2[4] -> 3[6] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:1332488:1333036 [6] NCCL INFO Channel 03/0 : 3[6] -> 0[0] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:1332482:1333034 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[2] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:1332486:1333035 [4] NCCL INFO Channel 03/0 : 2[4] -> 3[6] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:1332483:1333039 [1] NCCL INFO Channel 00/0 : 3[7] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:1332483:1333039 [1] NCCL INFO Channel 01/0 : 3[7] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:1332483:1333039 [1] NCCL INFO Channel 02/0 : 3[7] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:1332483:1333039 [1] NCCL INFO Channel 03/0 : 3[7] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:1332483:1333039 [1] NCCL INFO Channel 00/0 : 0[1] -> 1[3] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:1332485:1333040 [3] NCCL INFO Channel 00/0 : 0[1] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:1332483:1333039 [1] NCCL INFO Channel 01/0 : 0[1] -> 1[3] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:1332483:1333039 [1] NCCL INFO Channel 02/0 : 0[1] -> 1[3] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:1332485:1333040 [3] NCCL INFO Channel 01/0 : 0[1] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:1332483:1333039 [1] NCCL INFO Channel 03/0 : 0[1] -> 1[3] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:1332485:1333040 [3] NCCL INFO Channel 02/0 : 0[1] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:1332485:1333040 [3] NCCL INFO Channel 03/0 : 0[1] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:1332485:1333040 [3] NCCL INFO Channel 00/0 : 1[3] -> 2[5] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:1332485:1333040 [3] NCCL INFO Channel 01/0 : 1[3] -> 2[5] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:1332485:1333040 [3] NCCL INFO Channel 02/0 : 1[3] -> 2[5] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:1332487:1333041 [5] NCCL INFO Channel 00/0 : 1[3] -> 2[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:1332485:1333040 [3] NCCL INFO Channel 03/0 : 1[3] -> 2[5] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:1332487:1333041 [5] NCCL INFO Channel 01/0 : 1[3] -> 2[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:1332487:1333041 [5] NCCL INFO Channel 02/0 : 1[3] -> 2[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:1332487:1333041 [5] NCCL INFO Channel 03/0 : 1[3] -> 2[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:1332487:1333041 [5] NCCL INFO Channel 00/0 : 2[5] -> 3[7] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:1332487:1333041 [5] NCCL INFO Channel 01/0 : 2[5] -> 3[7] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:1332487:1333041 [5] NCCL INFO Channel 02/0 : 2[5] -> 3[7] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:1332487:1333041 [5] NCCL INFO Channel 03/0 : 2[5] -> 3[7] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:1332489:1333042 [7] NCCL INFO Channel 00/0 : 2[5] -> 3[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:1332489:1333042 [7] NCCL INFO Channel 01/0 : 2[5] -> 3[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:1332489:1333042 [7] NCCL INFO Channel 02/0 : 2[5] -> 3[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:1332489:1333042 [7] NCCL INFO Channel 03/0 : 2[5] -> 3[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:1332489:1333042 [7] NCCL INFO Channel 00/0 : 3[7] -> 0[1] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:1332489:1333042 [7] NCCL INFO Channel 01/0 : 3[7] -> 0[1] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:1332489:1333042 [7] NCCL INFO Channel 02/0 : 3[7] -> 0[1] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:1332489:1333042 [7] NCCL INFO Channel 03/0 : 3[7] -> 0[1] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:1061622:1062139 [2] NCCL INFO Connected all rings
+ip-26-0-166-36:1061622:1062139 [2] NCCL INFO Channel 02/0 : 3[6] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:1061622:1062139 [2] NCCL INFO Channel 03/0 : 3[6] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:1061622:1062139 [2] NCCL INFO Channel 02/0 : 1[2] -> 3[6] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:1061622:1062139 [2] NCCL INFO Channel 03/0 : 1[2] -> 3[6] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:1061620:1062136 [0] NCCL INFO Connected all rings
+ip-26-0-166-36:1061620:1062136 [0] NCCL INFO Channel 00/0 : 2[4] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:1061620:1062136 [0] NCCL INFO Channel 01/0 : 2[4] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:1061620:1062136 [0] NCCL INFO Channel 00/0 : 0[0] -> 2[4] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:1061620:1062136 [0] NCCL INFO Channel 01/0 : 0[0] -> 2[4] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:1061626:1062137 [6] NCCL INFO Connected all rings
+ip-26-0-166-36:1061626:1062137 [6] NCCL INFO Channel 02/0 : 1[2] -> 3[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:1061626:1062137 [6] NCCL INFO Channel 03/0 : 1[2] -> 3[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:1061626:1062137 [6] NCCL INFO Channel 02/0 : 3[6] -> 1[2] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:1061626:1062137 [6] NCCL INFO Channel 03/0 : 3[6] -> 1[2] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:1061622:1062139 [2] NCCL INFO Channel 00/0 : 2[4] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:1061626:1062137 [6] NCCL INFO Channel 00/0 : 3[6] -> 2[4] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:1061622:1062139 [2] NCCL INFO Channel 01/0 : 2[4] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:1061626:1062137 [6] NCCL INFO Channel 01/0 : 3[6] -> 2[4] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:1061622:1062139 [2] NCCL INFO Channel 02/0 : 2[4] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:1061622:1062139 [2] NCCL INFO Channel 03/0 : 2[4] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:1061622:1062139 [2] NCCL INFO Channel 02/0 : 1[2] -> 0[0] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:1061622:1062139 [2] NCCL INFO Channel 03/0 : 1[2] -> 0[0] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:1061624:1062138 [4] NCCL INFO Connected all rings
+ip-26-0-166-36:1061624:1062138 [4] NCCL INFO Channel 00/0 : 0[0] -> 2[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:1061624:1062138 [4] NCCL INFO Channel 01/0 : 0[0] -> 2[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:1061624:1062138 [4] NCCL INFO Channel 00/0 : 2[4] -> 0[0] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:1061624:1062138 [4] NCCL INFO Channel 01/0 : 2[4] -> 0[0] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:1061620:1062136 [0] NCCL INFO Channel 02/0 : 1[2] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:1061624:1062138 [4] NCCL INFO Channel 00/0 : 3[6] -> 2[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:1061620:1062136 [0] NCCL INFO Channel 03/0 : 1[2] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:1061624:1062138 [4] NCCL INFO Channel 01/0 : 3[6] -> 2[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:1061624:1062138 [4] NCCL INFO Channel 00/0 : 2[4] -> 1[2] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:1061624:1062138 [4] NCCL INFO Channel 01/0 : 2[4] -> 1[2] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:1061624:1062138 [4] NCCL INFO Channel 02/0 : 2[4] -> 1[2] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:1061624:1062138 [4] NCCL INFO Channel 03/0 : 2[4] -> 1[2] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:1061623:1062144 [3] NCCL INFO Connected all rings
+ip-26-0-166-36:1061623:1062144 [3] NCCL INFO Channel 02/0 : 3[7] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:1061623:1062144 [3] NCCL INFO Channel 03/0 : 3[7] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:1061623:1062144 [3] NCCL INFO Channel 02/0 : 1[3] -> 3[7] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:1061623:1062144 [3] NCCL INFO Channel 03/0 : 1[3] -> 3[7] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:1061627:1062142 [7] NCCL INFO Connected all rings
+ip-26-0-166-36:1061627:1062142 [7] NCCL INFO Channel 02/0 : 1[3] -> 3[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:1061627:1062142 [7] NCCL INFO Channel 03/0 : 1[3] -> 3[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:1061627:1062142 [7] NCCL INFO Channel 02/0 : 3[7] -> 1[3] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:1061627:1062142 [7] NCCL INFO Channel 03/0 : 3[7] -> 1[3] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:1061621:1062141 [1] NCCL INFO Connected all rings
+ip-26-0-166-36:1061621:1062141 [1] NCCL INFO Channel 00/0 : 2[5] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:1061627:1062142 [7] NCCL INFO Channel 00/0 : 3[7] -> 2[5] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:1061623:1062144 [3] NCCL INFO Channel 00/0 : 2[5] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:1061621:1062141 [1] NCCL INFO Channel 01/0 : 2[5] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:1061627:1062142 [7] NCCL INFO Channel 01/0 : 3[7] -> 2[5] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:1061621:1062141 [1] NCCL INFO Channel 00/0 : 0[1] -> 2[5] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:1061623:1062144 [3] NCCL INFO Channel 01/0 : 2[5] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:1061621:1062141 [1] NCCL INFO Channel 01/0 : 0[1] -> 2[5] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:1061623:1062144 [3] NCCL INFO Channel 02/0 : 2[5] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:1061623:1062144 [3] NCCL INFO Channel 03/0 : 2[5] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:1061623:1062144 [3] NCCL INFO Channel 02/0 : 1[3] -> 0[1] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:1061623:1062144 [3] NCCL INFO Channel 03/0 : 1[3] -> 0[1] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:1061625:1062143 [5] NCCL INFO Connected all rings
+ip-26-0-166-36:1061625:1062143 [5] NCCL INFO Channel 00/0 : 0[1] -> 2[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:1061625:1062143 [5] NCCL INFO Channel 01/0 : 0[1] -> 2[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:1061625:1062143 [5] NCCL INFO Channel 00/0 : 2[5] -> 0[1] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:1061625:1062143 [5] NCCL INFO Channel 01/0 : 2[5] -> 0[1] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:1061621:1062141 [1] NCCL INFO Channel 02/0 : 1[3] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:1061625:1062143 [5] NCCL INFO Channel 00/0 : 3[7] -> 2[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:1061621:1062141 [1] NCCL INFO Channel 03/0 : 1[3] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:1061625:1062143 [5] NCCL INFO Channel 01/0 : 3[7] -> 2[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:1061625:1062143 [5] NCCL INFO Channel 00/0 : 2[5] -> 1[3] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:1061625:1062143 [5] NCCL INFO Channel 01/0 : 2[5] -> 1[3] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:1061625:1062143 [5] NCCL INFO Channel 02/0 : 2[5] -> 1[3] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:1061625:1062143 [5] NCCL INFO Channel 03/0 : 2[5] -> 1[3] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:1332482:1333034 [0] NCCL INFO Connected all rings
+ip-26-0-166-125:1332482:1333034 [0] NCCL INFO Channel 00/0 : 2[4] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:1332482:1333034 [0] NCCL INFO Channel 01/0 : 2[4] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:1332482:1333034 [0] NCCL INFO Channel 00/0 : 0[0] -> 2[4] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:1332482:1333034 [0] NCCL INFO Channel 01/0 : 0[0] -> 2[4] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:1332483:1333039 [1] NCCL INFO Connected all rings
+ip-26-0-166-125:1332483:1333039 [1] NCCL INFO Channel 00/0 : 2[5] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:1332483:1333039 [1] NCCL INFO Channel 01/0 : 2[5] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:1332483:1333039 [1] NCCL INFO Channel 00/0 : 0[1] -> 2[5] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:1332483:1333039 [1] NCCL INFO Channel 01/0 : 0[1] -> 2[5] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:1061626:1062137 [6] NCCL INFO Connected all trees
+ip-26-0-166-36:1061626:1062137 [6] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-166-36:1061626:1062137 [6] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-36:1061620:1062136 [0] NCCL INFO Connected all trees
+ip-26-0-166-36:1061620:1062136 [0] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-166-36:1061620:1062136 [0] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-125:1332484:1333037 [2] NCCL INFO Connected all rings
+ip-26-0-166-125:1332485:1333040 [3] NCCL INFO Connected all rings
+ip-26-0-166-125:1332486:1333035 [4] NCCL INFO Connected all rings
+ip-26-0-166-125:1332484:1333037 [2] NCCL INFO Channel 02/0 : 3[6] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:1332485:1333040 [3] NCCL INFO Channel 02/0 : 3[7] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:1332486:1333035 [4] NCCL INFO Channel 00/0 : 0[0] -> 2[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:1332484:1333037 [2] NCCL INFO Channel 03/0 : 3[6] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:1332485:1333040 [3] NCCL INFO Channel 03/0 : 3[7] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:1332486:1333035 [4] NCCL INFO Channel 01/0 : 0[0] -> 2[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:1332484:1333037 [2] NCCL INFO Channel 02/0 : 1[2] -> 3[6] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:1332486:1333035 [4] NCCL INFO Channel 00/0 : 2[4] -> 0[0] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:1332485:1333040 [3] NCCL INFO Channel 02/0 : 1[3] -> 3[7] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:1332484:1333037 [2] NCCL INFO Channel 03/0 : 1[2] -> 3[6] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:1332486:1333035 [4] NCCL INFO Channel 01/0 : 2[4] -> 0[0] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:1332485:1333040 [3] NCCL INFO Channel 03/0 : 1[3] -> 3[7] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:1332482:1333034 [0] NCCL INFO Channel 02/0 : 1[2] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:1332486:1333035 [4] NCCL INFO Channel 00/0 : 3[6] -> 2[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:1332482:1333034 [0] NCCL INFO Channel 03/0 : 1[2] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:1332486:1333035 [4] NCCL INFO Channel 01/0 : 3[6] -> 2[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:1332486:1333035 [4] NCCL INFO Channel 00/0 : 2[4] -> 1[2] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:1332486:1333035 [4] NCCL INFO Channel 01/0 : 2[4] -> 1[2] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:1332486:1333035 [4] NCCL INFO Channel 02/0 : 2[4] -> 1[2] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:1332486:1333035 [4] NCCL INFO Channel 03/0 : 2[4] -> 1[2] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:1332488:1333036 [6] NCCL INFO Connected all rings
+ip-26-0-166-125:1332488:1333036 [6] NCCL INFO Channel 02/0 : 1[2] -> 3[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:1332488:1333036 [6] NCCL INFO Channel 03/0 : 1[2] -> 3[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:1332488:1333036 [6] NCCL INFO Channel 02/0 : 3[6] -> 1[2] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:1332488:1333036 [6] NCCL INFO Channel 03/0 : 3[6] -> 1[2] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:1332484:1333037 [2] NCCL INFO Channel 00/0 : 2[4] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:1332488:1333036 [6] NCCL INFO Channel 00/0 : 3[6] -> 2[4] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:1332484:1333037 [2] NCCL INFO Channel 01/0 : 2[4] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:1332488:1333036 [6] NCCL INFO Channel 01/0 : 3[6] -> 2[4] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:1332487:1333041 [5] NCCL INFO Connected all rings
+ip-26-0-166-125:1332484:1333037 [2] NCCL INFO Channel 02/0 : 2[4] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:1332484:1333037 [2] NCCL INFO Channel 03/0 : 2[4] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:1332487:1333041 [5] NCCL INFO Channel 00/0 : 0[1] -> 2[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:1332484:1333037 [2] NCCL INFO Channel 02/0 : 1[2] -> 0[0] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:1332487:1333041 [5] NCCL INFO Channel 01/0 : 0[1] -> 2[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:1332484:1333037 [2] NCCL INFO Channel 03/0 : 1[2] -> 0[0] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:1332487:1333041 [5] NCCL INFO Channel 00/0 : 2[5] -> 0[1] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:1332487:1333041 [5] NCCL INFO Channel 01/0 : 2[5] -> 0[1] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:1332483:1333039 [1] NCCL INFO Channel 02/0 : 1[3] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:1332487:1333041 [5] NCCL INFO Channel 00/0 : 3[7] -> 2[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:1332483:1333039 [1] NCCL INFO Channel 03/0 : 1[3] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:1332487:1333041 [5] NCCL INFO Channel 01/0 : 3[7] -> 2[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:1332487:1333041 [5] NCCL INFO Channel 00/0 : 2[5] -> 1[3] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:1332487:1333041 [5] NCCL INFO Channel 01/0 : 2[5] -> 1[3] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:1332487:1333041 [5] NCCL INFO Channel 02/0 : 2[5] -> 1[3] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:1332487:1333041 [5] NCCL INFO Channel 03/0 : 2[5] -> 1[3] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:1061626:1062137 [6] NCCL INFO comm 0xa75c150 rank 3 nranks 4 cudaDev 6 nvmlDev 6 busId b9000 commId 0x5af0260af03ae61d - Init COMPLETE
+ip-26-0-166-36:1061620:1062136 [0] NCCL INFO comm 0xa7f6770 rank 0 nranks 4 cudaDev 0 nvmlDev 0 busId 53000 commId 0x5af0260af03ae61d - Init COMPLETE
+ip-26-0-166-125:1332489:1333042 [7] NCCL INFO Connected all rings
+ip-26-0-166-125:1332489:1333042 [7] NCCL INFO Channel 02/0 : 1[3] -> 3[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:1332489:1333042 [7] NCCL INFO Channel 03/0 : 1[3] -> 3[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:1332489:1333042 [7] NCCL INFO Channel 02/0 : 3[7] -> 1[3] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:1332489:1333042 [7] NCCL INFO Channel 03/0 : 3[7] -> 1[3] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:1332485:1333040 [3] NCCL INFO Channel 00/0 : 2[5] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:1332489:1333042 [7] NCCL INFO Channel 00/0 : 3[7] -> 2[5] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:1332485:1333040 [3] NCCL INFO Channel 01/0 : 2[5] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:1332489:1333042 [7] NCCL INFO Channel 01/0 : 3[7] -> 2[5] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:1332485:1333040 [3] NCCL INFO Channel 02/0 : 2[5] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:1332485:1333040 [3] NCCL INFO Channel 03/0 : 2[5] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:1332485:1333040 [3] NCCL INFO Channel 02/0 : 1[3] -> 0[1] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:1332485:1333040 [3] NCCL INFO Channel 03/0 : 1[3] -> 0[1] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:1061627:1062142 [7] NCCL INFO Connected all trees
+ip-26-0-166-36:1061627:1062142 [7] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-166-36:1061627:1062142 [7] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-36:1061621:1062141 [1] NCCL INFO Connected all trees
+ip-26-0-166-36:1061621:1062141 [1] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-166-36:1061621:1062141 [1] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-36:1061627:1062142 [7] NCCL INFO comm 0xb3df350 rank 3 nranks 4 cudaDev 7 nvmlDev 7 busId ca000 commId 0xe7283ca9e98730a - Init COMPLETE
+ip-26-0-166-36:1061621:1062141 [1] NCCL INFO comm 0xb4b1f60 rank 0 nranks 4 cudaDev 1 nvmlDev 1 busId 64000 commId 0xe7283ca9e98730a - Init COMPLETE
+ip-26-0-166-36:1061624:1062138 [4] NCCL INFO Connected all trees
+ip-26-0-166-36:1061624:1062138 [4] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-166-36:1061624:1062138 [4] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-36:1061622:1062139 [2] NCCL INFO Connected all trees
+ip-26-0-166-36:1061622:1062139 [2] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-166-36:1061622:1062139 [2] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-36:1061624:1062138 [4] NCCL INFO comm 0xc01b970 rank 2 nranks 4 cudaDev 4 nvmlDev 4 busId 97000 commId 0x5af0260af03ae61d - Init COMPLETE
+ip-26-0-166-36:1061622:1062139 [2] NCCL INFO comm 0xafef960 rank 1 nranks 4 cudaDev 2 nvmlDev 2 busId 75000 commId 0x5af0260af03ae61d - Init COMPLETE
+ip-26-0-166-36:1061625:1062143 [5] NCCL INFO Connected all trees
+ip-26-0-166-36:1061625:1062143 [5] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-166-36:1061625:1062143 [5] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-36:1061623:1062144 [3] NCCL INFO Connected all trees
+ip-26-0-166-36:1061623:1062144 [3] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-166-36:1061623:1062144 [3] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-36:1061625:1062143 [5] NCCL INFO comm 0xb1a15c0 rank 2 nranks 4 cudaDev 5 nvmlDev 5 busId a8000 commId 0xe7283ca9e98730a - Init COMPLETE
+ip-26-0-166-36:1061623:1062144 [3] NCCL INFO comm 0xa5ac5d0 rank 1 nranks 4 cudaDev 3 nvmlDev 3 busId 86000 commId 0xe7283ca9e98730a - Init COMPLETE
+ip-26-0-166-125:1332489:1333042 [7] NCCL INFO Connected all trees
+ip-26-0-166-125:1332489:1333042 [7] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-166-125:1332489:1333042 [7] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-125:1332488:1333036 [6] NCCL INFO Connected all trees
+ip-26-0-166-125:1332488:1333036 [6] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-166-125:1332488:1333036 [6] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-125:1332489:1333042 [7] NCCL INFO comm 0xbea4900 rank 3 nranks 4 cudaDev 7 nvmlDev 7 busId ca000 commId 0xc1f08c305513d36b - Init COMPLETE
+ip-26-0-166-125:1332488:1333036 [6] NCCL INFO comm 0xa7f5510 rank 3 nranks 4 cudaDev 6 nvmlDev 6 busId b9000 commId 0xcf3103ca05a405d6 - Init COMPLETE
+ip-26-0-166-125:1332482:1333034 [0] NCCL INFO Connected all trees
+ip-26-0-166-125:1332482:1333034 [0] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-166-125:1332482:1333034 [0] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-125:1332482:1333034 [0] NCCL INFO comm 0xb7eb590 rank 0 nranks 4 cudaDev 0 nvmlDev 0 busId 53000 commId 0xcf3103ca05a405d6 - Init COMPLETE
+ip-26-0-166-125:1332483:1333039 [1] NCCL INFO Connected all trees
+ip-26-0-166-125:1332483:1333039 [1] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-166-125:1332483:1333039 [1] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-125:1332486:1333035 [4] NCCL INFO Connected all trees
+ip-26-0-166-125:1332486:1333035 [4] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-166-125:1332486:1333035 [4] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-125:1332484:1333037 [2] NCCL INFO Connected all trees
+ip-26-0-166-125:1332484:1333037 [2] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-166-125:1332484:1333037 [2] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-125:1332487:1333041 [5] NCCL INFO Connected all trees
+ip-26-0-166-125:1332487:1333041 [5] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-166-125:1332487:1333041 [5] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-125:1332485:1333040 [3] NCCL INFO Connected all trees
+ip-26-0-166-125:1332485:1333040 [3] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-166-125:1332485:1333040 [3] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-125:1332483:1333039 [1] NCCL INFO comm 0xb2eb4c0 rank 0 nranks 4 cudaDev 1 nvmlDev 1 busId 64000 commId 0xc1f08c305513d36b - Init COMPLETE
+ip-26-0-166-125:1332486:1333035 [4] NCCL INFO comm 0xb0adc10 rank 2 nranks 4 cudaDev 4 nvmlDev 4 busId 97000 commId 0xcf3103ca05a405d6 - Init COMPLETE
+ip-26-0-166-125:1332484:1333037 [2] NCCL INFO comm 0xba13110 rank 1 nranks 4 cudaDev 2 nvmlDev 2 busId 75000 commId 0xcf3103ca05a405d6 - Init COMPLETE
+ip-26-0-166-125:1332487:1333041 [5] NCCL INFO comm 0xba1e940 rank 2 nranks 4 cudaDev 5 nvmlDev 5 busId a8000 commId 0xc1f08c305513d36b - Init COMPLETE
+ip-26-0-166-125:1332485:1333040 [3] NCCL INFO comm 0xa78d1c0 rank 1 nranks 4 cudaDev 3 nvmlDev 3 busId 86000 commId 0xc1f08c305513d36b - Init COMPLETE
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+        outputs = self.pipeline_engine.train_batch_iter(return forward_call(*args, **kwargs)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 161, in forward
+    output = self.pp_block(**new_kwargs)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 201, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 161, in forward
+    return self.act(gate_states) * up_states
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 896.00 MiB. GPU 1 has a total capacty of 79.33 GiB of which 103.94 MiB is free. Including non-PyTorch memory, this process has 79.22 GiB memory in use. Of the allocated memory 75.98 GiB is allocated by PyTorch, and 1.48 GiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    out = F.linear(tensor, weight, bias)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 0 has a total capacty of 79.33 GiB of which 95.94 MiB is free. Including non-PyTorch memory, this process has 79.22 GiB memory in use. Of the allocated memory 76.86 GiB is allocated by PyTorch, and 749.35 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 161, in forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 201, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self.act(gate_states) * up_states
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 896.00 MiB. GPU 3 has a total capacty of 79.33 GiB of which 455.94 MiB is free. Including non-PyTorch memory, this process has 78.87 GiB memory in use. Of the allocated memory 75.98 GiB is allocated by PyTorch, and 1005.35 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 161, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+Traceback (most recent call last):
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 201, in forward
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return self.act(gate_states) * up_states
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 896.00 MiB. GPU 2 has a total capacty of 79.33 GiB of which 831.94 MiB is free. Including non-PyTorch memory, this process has 78.51 GiB memory in use. Of the allocated memory 75.98 GiB is allocated by PyTorch, and 749.35 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    output = model(**micro_batch)
+    return forward_call(*args, **kwargs)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 161, in forward
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    trainer.train(dataloader)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+        return forward_call(*args, **kwargs)output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 201, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self.act(gate_states) * up_states
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 896.00 MiB. GPU 5 has a total capacty of 79.33 GiB of which 319.94 MiB is free. Including non-PyTorch memory, this process has 79.01 GiB memory in use. Of the allocated memory 75.98 GiB is allocated by PyTorch, and 1.23 GiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 161, in forward
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 161, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+        hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 161, in forward
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 201, in forward
+    return self.act(gate_states) * up_states
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 896.00 MiB. GPU 4 has a total capacty of 79.33 GiB of which 199.94 MiB is free. Including non-PyTorch memory, this process has 79.12 GiB memory in use. Of the allocated memory 75.98 GiB is allocated by PyTorch, and 1.23 GiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 201, in forward
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self.act(gate_states) * up_states
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 896.00 MiB. GPU 6 has a total capacty of 79.33 GiB of which 359.94 MiB is free. Including non-PyTorch memory, this process has 78.97 GiB memory in use. Of the allocated memory 75.98 GiB is allocated by PyTorch, and 1.23 GiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 201, in forward
+    return self.act(gate_states) * up_states
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 896.00 MiB. GPU 7 has a total capacty of 79.33 GiB of which 319.94 MiB is free. Including non-PyTorch memory, this process has 79.01 GiB memory in use. Of the allocated memory 75.98 GiB is allocated by PyTorch, and 1.23 GiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+12/27/2024 17:32:21 [ERROR|DP=1|PP=1|TP=0|ip-26-0-166-36]: [RECV META] Rank 1: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/27/2024 17:32:21 [ERROR|DP=3|PP=1|TP=0|ip-26-0-166-36]: [RECV META] Rank 1: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/27/2024 17:32:21 [ERROR|DP=3|PP=1|TP=1|ip-26-0-166-36]: [RECV META] Rank 1: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/27/2024 17:32:21 [WARNING|DP=1|PP=1|TP=0|ip-26-0-166-36]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/27/2024 17:32:21 [WARNING|DP=3|PP=1|TP=0|ip-26-0-166-36]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/27/2024 17:32:21 [WARNING|DP=3|PP=1|TP=1|ip-26-0-166-36]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/27/2024 17:32:21 [WARNING|DP=1|PP=1|TP=0|ip-26-0-166-36]: No progress made in communication iteration, attempt 1/5
+12/27/2024 17:32:21 [WARNING|DP=3|PP=1|TP=0|ip-26-0-166-36]: No progress made in communication iteration, attempt 1/5
+12/27/2024 17:32:21 [ERROR|DP=0|PP=1|TP=0|ip-26-0-166-36]: [RECV META] Rank 1: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/27/2024 17:32:21 [WARNING|DP=3|PP=1|TP=1|ip-26-0-166-36]: No progress made in communication iteration, attempt 1/5
+12/27/2024 17:32:21 [WARNING|DP=0|PP=1|TP=0|ip-26-0-166-36]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/27/2024 17:32:21 [WARNING|DP=0|PP=1|TP=0|ip-26-0-166-36]: No progress made in communication iteration, attempt 1/5
+12/27/2024 17:32:21 [ERROR|DP=2|PP=1|TP=0|ip-26-0-166-36]: [RECV META] Rank 1: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/27/2024 17:32:21 [ERROR|DP=2|PP=1|TP=1|ip-26-0-166-36]: [RECV META] Rank 1: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/27/2024 17:32:21 [WARNING|DP=2|PP=1|TP=0|ip-26-0-166-36]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/27/2024 17:32:21 [WARNING|DP=2|PP=1|TP=1|ip-26-0-166-36]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/27/2024 17:32:21 [ERROR|DP=1|PP=1|TP=1|ip-26-0-166-36]: [RECV META] Rank 1: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/27/2024 17:32:21 [WARNING|DP=2|PP=1|TP=0|ip-26-0-166-36]: No progress made in communication iteration, attempt 1/5
+12/27/2024 17:32:21 [WARNING|DP=2|PP=1|TP=1|ip-26-0-166-36]: No progress made in communication iteration, attempt 1/5
+12/27/2024 17:32:21 [WARNING|DP=1|PP=1|TP=1|ip-26-0-166-36]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/27/2024 17:32:21 [WARNING|DP=1|PP=1|TP=1|ip-26-0-166-36]: No progress made in communication iteration, attempt 1/5
+12/27/2024 17:32:21 [ERROR|DP=0|PP=1|TP=1|ip-26-0-166-36]: [RECV META] Rank 1: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/27/2024 17:32:21 [WARNING|DP=0|PP=1|TP=1|ip-26-0-166-36]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/27/2024 17:32:21 [WARNING|DP=0|PP=1|TP=1|ip-26-0-166-36]: No progress made in communication iteration, attempt 1/5
+12/27/2024 17:32:21 [ERROR|DP=1|PP=1|TP=0|ip-26-0-166-36]: [RECV META] Rank 1: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/27/2024 17:32:21 [ERROR|DP=3|PP=1|TP=0|ip-26-0-166-36]: [RECV META] Rank 1: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/27/2024 17:32:21 [WARNING|DP=1|PP=1|TP=0|ip-26-0-166-36]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/27/2024 17:32:21 [ERROR|DP=3|PP=1|TP=1|ip-26-0-166-36]: [RECV META] Rank 1: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/27/2024 17:32:21 [WARNING|DP=3|PP=1|TP=0|ip-26-0-166-36]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/27/2024 17:32:21 [WARNING|DP=1|PP=1|TP=0|ip-26-0-166-36]: No progress made in communication iteration, attempt 2/5
+12/27/2024 17:32:21 [WARNING|DP=3|PP=1|TP=1|ip-26-0-166-36]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/27/2024 17:32:21 [WARNING|DP=3|PP=1|TP=0|ip-26-0-166-36]: No progress made in communication iteration, attempt 2/5
+12/27/2024 17:32:21 [ERROR|DP=0|PP=1|TP=0|ip-26-0-166-36]: [RECV META] Rank 1: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/27/2024 17:32:21 [WARNING|DP=3|PP=1|TP=1|ip-26-0-166-36]: No progress made in communication iteration, attempt 2/5
+12/27/2024 17:32:21 [WARNING|DP=0|PP=1|TP=0|ip-26-0-166-36]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/27/2024 17:32:21 [WARNING|DP=0|PP=1|TP=0|ip-26-0-166-36]: No progress made in communication iteration, attempt 2/5
+12/27/2024 17:32:21 [ERROR|DP=2|PP=1|TP=0|ip-26-0-166-36]: [RECV META] Rank 1: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/27/2024 17:32:21 [WARNING|DP=2|PP=1|TP=0|ip-26-0-166-36]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/27/2024 17:32:21 [ERROR|DP=2|PP=1|TP=1|ip-26-0-166-36]: [RECV META] Rank 1: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/27/2024 17:32:21 [WARNING|DP=2|PP=1|TP=0|ip-26-0-166-36]: No progress made in communication iteration, attempt 2/5
+12/27/2024 17:32:21 [WARNING|DP=2|PP=1|TP=1|ip-26-0-166-36]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/27/2024 17:32:21 [WARNING|DP=2|PP=1|TP=1|ip-26-0-166-36]: No progress made in communication iteration, attempt 2/5
+12/27/2024 17:32:21 [ERROR|DP=1|PP=1|TP=1|ip-26-0-166-36]: [RECV META] Rank 1: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/27/2024 17:32:21 [WARNING|DP=1|PP=1|TP=1|ip-26-0-166-36]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/27/2024 17:32:21 [WARNING|DP=1|PP=1|TP=1|ip-26-0-166-36]: No progress made in communication iteration, attempt 2/5
+12/27/2024 17:32:21 [ERROR|DP=0|PP=1|TP=1|ip-26-0-166-36]: [RECV META] Rank 1: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/27/2024 17:32:21 [WARNING|DP=0|PP=1|TP=1|ip-26-0-166-36]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/27/2024 17:32:21 [WARNING|DP=0|PP=1|TP=1|ip-26-0-166-36]: No progress made in communication iteration, attempt 2/5
+12/27/2024 17:32:21 [ERROR|DP=1|PP=1|TP=0|ip-26-0-166-36]: [RECV META] Rank 1: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Broken pipe. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/27/2024 17:32:21 [ERROR|DP=3|PP=1|TP=0|ip-26-0-166-36]: [RECV META] Rank 1: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Broken pipe. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/27/2024 17:32:21 [WARNING|DP=1|PP=1|TP=0|ip-26-0-166-36]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Broken pipe. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/27/2024 17:32:21 [ERROR|DP=3|PP=1|TP=1|ip-26-0-166-36]: [RECV META] Rank 1: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Broken pipe. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/27/2024 17:32:21 [WARNING|DP=1|PP=1|TP=0|ip-26-0-166-36]: No progress made in communication iteration, attempt 3/5
+12/27/2024 17:32:21 [WARNING|DP=3|PP=1|TP=0|ip-26-0-166-36]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Broken pipe. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/27/2024 17:32:21 [WARNING|DP=3|PP=1|TP=1|ip-26-0-166-36]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Broken pipe. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/27/2024 17:32:21 [WARNING|DP=3|PP=1|TP=0|ip-26-0-166-36]: No progress made in communication iteration, attempt 3/5
+12/27/2024 17:32:21 [ERROR|DP=0|PP=1|TP=0|ip-26-0-166-36]: [RECV META] Rank 1: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Broken pipe. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/27/2024 17:32:21 [WARNING|DP=3|PP=1|TP=1|ip-26-0-166-36]: No progress made in communication iteration, attempt 3/5
+12/27/2024 17:32:21 [WARNING|DP=0|PP=1|TP=0|ip-26-0-166-36]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Broken pipe. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/27/2024 17:32:21 [WARNING|DP=0|PP=1|TP=0|ip-26-0-166-36]: No progress made in communication iteration, attempt 3/5
+12/27/2024 17:32:21 [ERROR|DP=2|PP=1|TP=0|ip-26-0-166-36]: [RECV META] Rank 1: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Broken pipe. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/27/2024 17:32:21 [WARNING|DP=2|PP=1|TP=0|ip-26-0-166-36]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Broken pipe. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/27/2024 17:32:21 [ERROR|DP=2|PP=1|TP=1|ip-26-0-166-36]: [RECV META] Rank 1: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Broken pipe. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/27/2024 17:32:21 [WARNING|DP=2|PP=1|TP=0|ip-26-0-166-36]: No progress made in communication iteration, attempt 3/5
+12/27/2024 17:32:21 [WARNING|DP=2|PP=1|TP=1|ip-26-0-166-36]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Broken pipe. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/27/2024 17:32:21 [WARNING|DP=2|PP=1|TP=1|ip-26-0-166-36]: No progress made in communication iteration, attempt 3/5
+12/27/2024 17:32:21 [ERROR|DP=1|PP=1|TP=1|ip-26-0-166-36]: [RECV META] Rank 1: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Broken pipe. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/27/2024 17:32:21 [WARNING|DP=1|PP=1|TP=1|ip-26-0-166-36]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Broken pipe. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/27/2024 17:32:21 [WARNING|DP=1|PP=1|TP=1|ip-26-0-166-36]: No progress made in communication iteration, attempt 3/5
+12/27/2024 17:32:21 [ERROR|DP=0|PP=1|TP=1|ip-26-0-166-36]: [RECV META] Rank 1: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Broken pipe. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/27/2024 17:32:21 [WARNING|DP=0|PP=1|TP=1|ip-26-0-166-36]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Broken pipe. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/27/2024 17:32:21 [WARNING|DP=0|PP=1|TP=1|ip-26-0-166-36]: No progress made in communication iteration, attempt 3/5
+12/27/2024 17:32:21 [ERROR|DP=1|PP=1|TP=0|ip-26-0-166-36]: [RECV META] Rank 1: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Broken pipe. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/27/2024 17:32:21 [ERROR|DP=3|PP=1|TP=0|ip-26-0-166-36]: [RECV META] Rank 1: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Broken pipe. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/27/2024 17:32:21 [WARNING|DP=1|PP=1|TP=0|ip-26-0-166-36]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Broken pipe. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/27/2024 17:32:21 [ERROR|DP=3|PP=1|TP=1|ip-26-0-166-36]: [RECV META] Rank 1: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Broken pipe. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/27/2024 17:32:21 [WARNING|DP=1|PP=1|TP=0|ip-26-0-166-36]: No progress made in communication iteration, attempt 4/5
+12/27/2024 17:32:21 [WARNING|DP=3|PP=1|TP=0|ip-26-0-166-36]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Broken pipe. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/27/2024 17:32:21 [WARNING|DP=3|PP=1|TP=1|ip-26-0-166-36]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Broken pipe. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/27/2024 17:32:21 [WARNING|DP=3|PP=1|TP=0|ip-26-0-166-36]: No progress made in communication iteration, attempt 4/5
+12/27/2024 17:32:21 [ERROR|DP=0|PP=1|TP=0|ip-26-0-166-36]: [RECV META] Rank 1: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Broken pipe. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/27/2024 17:32:21 [WARNING|DP=3|PP=1|TP=1|ip-26-0-166-36]: No progress made in communication iteration, attempt 4/5
+12/27/2024 17:32:21 [WARNING|DP=0|PP=1|TP=0|ip-26-0-166-36]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Broken pipe. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/27/2024 17:32:21 [WARNING|DP=0|PP=1|TP=0|ip-26-0-166-36]: No progress made in communication iteration, attempt 4/5
+12/27/2024 17:32:21 [ERROR|DP=2|PP=1|TP=0|ip-26-0-166-36]: [RECV META] Rank 1: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Broken pipe. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/27/2024 17:32:21 [WARNING|DP=2|PP=1|TP=0|ip-26-0-166-36]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Broken pipe. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/27/2024 17:32:21 [ERROR|DP=2|PP=1|TP=1|ip-26-0-166-36]: [RECV META] Rank 1: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Broken pipe. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/27/2024 17:32:21 [WARNING|DP=2|PP=1|TP=0|ip-26-0-166-36]: No progress made in communication iteration, attempt 4/5
+12/27/2024 17:32:21 [WARNING|DP=2|PP=1|TP=1|ip-26-0-166-36]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Broken pipe. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/27/2024 17:32:21 [WARNING|DP=2|PP=1|TP=1|ip-26-0-166-36]: No progress made in communication iteration, attempt 4/5
+12/27/2024 17:32:21 [ERROR|DP=1|PP=1|TP=1|ip-26-0-166-36]: [RECV META] Rank 1: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Broken pipe. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/27/2024 17:32:21 [WARNING|DP=1|PP=1|TP=1|ip-26-0-166-36]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Broken pipe. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/27/2024 17:32:21 [WARNING|DP=1|PP=1|TP=1|ip-26-0-166-36]: No progress made in communication iteration, attempt 4/5
+12/27/2024 17:32:21 [ERROR|DP=0|PP=1|TP=1|ip-26-0-166-36]: [RECV META] Rank 1: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Broken pipe. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/27/2024 17:32:21 [WARNING|DP=0|PP=1|TP=1|ip-26-0-166-36]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Broken pipe. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/27/2024 17:32:21 [WARNING|DP=0|PP=1|TP=1|ip-26-0-166-36]: No progress made in communication iteration, attempt 4/5
+12/27/2024 17:32:21 [ERROR|DP=3|PP=1|TP=1|ip-26-0-166-36]: [RECV META] Rank 1: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Broken pipe. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/27/2024 17:32:21 [ERROR|DP=3|PP=1|TP=0|ip-26-0-166-36]: [RECV META] Rank 1: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Broken pipe. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/27/2024 17:32:21 [WARNING|DP=3|PP=1|TP=1|ip-26-0-166-36]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Broken pipe. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/27/2024 17:32:21 [WARNING|DP=3|PP=1|TP=0|ip-26-0-166-36]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Broken pipe. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/27/2024 17:32:21 [ERROR|DP=0|PP=1|TP=0|ip-26-0-166-36]: [RECV META] Rank 1: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Broken pipe. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/27/2024 17:32:21 [WARNING|DP=3|PP=1|TP=1|ip-26-0-166-36]: No progress made in communication iteration, attempt 5/5
+12/27/2024 17:32:21 [WARNING|DP=3|PP=1|TP=0|ip-26-0-166-36]: No progress made in communication iteration, attempt 5/5
+12/27/2024 17:32:21 [ERROR|DP=3|PP=1|TP=1|ip-26-0-166-36]: Max retries reached, giving up on communication
+12/27/2024 17:32:21 [ERROR|DP=3|PP=1|TP=0|ip-26-0-166-36]: Max retries reached, giving up on communication
+12/27/2024 17:32:21 [WARNING|DP=0|PP=1|TP=0|ip-26-0-166-36]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Broken pipe. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/27/2024 17:32:21 [WARNING|DP=0|PP=1|TP=0|ip-26-0-166-36]: No progress made in communication iteration, attempt 5/5
+12/27/2024 17:32:21 [ERROR|DP=0|PP=1|TP=0|ip-26-0-166-36]: Max retries reached, giving up on communication
+12/27/2024 17:32:21 [ERROR|DP=2|PP=1|TP=0|ip-26-0-166-36]: [RECV META] Rank 1: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Broken pipe. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/27/2024 17:32:21 [WARNING|DP=2|PP=1|TP=0|ip-26-0-166-36]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Broken pipe. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/27/2024 17:32:21 [ERROR|DP=2|PP=1|TP=1|ip-26-0-166-36]: [RECV META] Rank 1: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Broken pipe. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/27/2024 17:32:21 [WARNING|DP=2|PP=1|TP=0|ip-26-0-166-36]: No progress made in communication iteration, attempt 5/5
+12/27/2024 17:32:21 [WARNING|DP=2|PP=1|TP=1|ip-26-0-166-36]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Broken pipe. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/27/2024 17:32:21 [ERROR|DP=2|PP=1|TP=0|ip-26-0-166-36]: Max retries reached, giving up on communication
+12/27/2024 17:32:21 [WARNING|DP=2|PP=1|TP=1|ip-26-0-166-36]: No progress made in communication iteration, attempt 5/5
+12/27/2024 17:32:21 [ERROR|DP=2|PP=1|TP=1|ip-26-0-166-36]: Max retries reached, giving up on communication
+12/27/2024 17:32:21 [ERROR|DP=1|PP=1|TP=1|ip-26-0-166-36]: [RECV META] Rank 1: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Broken pipe. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/27/2024 17:32:21 [WARNING|DP=1|PP=1|TP=1|ip-26-0-166-36]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Broken pipe. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/27/2024 17:32:21 [WARNING|DP=1|PP=1|TP=1|ip-26-0-166-36]: No progress made in communication iteration, attempt 5/5
+12/27/2024 17:32:21 [ERROR|DP=1|PP=1|TP=1|ip-26-0-166-36]: Max retries reached, giving up on communication
+12/27/2024 17:32:21 [ERROR|DP=1|PP=1|TP=0|ip-26-0-166-36]: [RECV META] Rank 1: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Broken pipe. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/27/2024 17:32:21 [WARNING|DP=1|PP=1|TP=0|ip-26-0-166-36]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Broken pipe. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/27/2024 17:32:21 [WARNING|DP=1|PP=1|TP=0|ip-26-0-166-36]: No progress made in communication iteration, attempt 5/5
+12/27/2024 17:32:21 [ERROR|DP=1|PP=1|TP=0|ip-26-0-166-36]: Max retries reached, giving up on communication
+12/27/2024 17:32:21 [ERROR|DP=0|PP=1|TP=1|ip-26-0-166-36]: [RECV META] Rank 1: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Broken pipe. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/27/2024 17:32:21 [WARNING|DP=0|PP=1|TP=1|ip-26-0-166-36]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Broken pipe. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/27/2024 17:32:21 [WARNING|DP=0|PP=1|TP=1|ip-26-0-166-36]: No progress made in communication iteration, attempt 5/5
+12/27/2024 17:32:21 [ERROR|DP=0|PP=1|TP=1|ip-26-0-166-36]: Max retries reached, giving up on communication
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 124, in recv_from_pipeline_state_buffer
+    activation = pipeline_state.activations_buffer.popleft()
+IndexError: pop from an empty deque
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    trainer.train(dataloader)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+trainer.train(dataloader)    
+trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    trainer.train(dataloader)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+
+    trainer.train(dataloader)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return f(*args, **kwargs)
+          File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+return f(*args, **kwargs)outputs = self.pipeline_engine.train_batch_iter(
+    
+    return f(*args, **kwargs)  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+return f(*args, **kwargs)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)    
+outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)    
+outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+        outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)outputs = self.pipeline_engine.train_batch_iter(
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+        outputs = self.pipeline_engine.train_batch_iter(outputs = self.pipeline_engine.train_batch_iter(
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+        output = model(**micro_batch)output = model(**micro_batch)
+    
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+return forward_call(*args, **kwargs)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+        return self._call_impl(*args, **kwargs)return self._call_impl(*args, **kwargs)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    sharded_logits = self.model(
+    sharded_logits = self.model(  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    return self._call_impl(*args, **kwargs)
+    return forward_call(*args, **kwargs)      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+return forward_call(*args, **kwargs)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+        return forward_call(*args, **kwargs)
+sharded_logits = self.model(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 124, in recv_from_pipeline_state_buffer
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+    new_kwargs[name] = recv_from_pipeline_state_buffer(  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 124, in recv_from_pipeline_state_buffer
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 124, in recv_from_pipeline_state_buffer
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 124, in recv_from_pipeline_state_buffer
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    activation = pipeline_state.activations_buffer.popleft()
+IndexError: pop from an empty deque
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 124, in recv_from_pipeline_state_buffer
+    activation = pipeline_state.activations_buffer.popleft()
+IndexError: pop from an empty deque
+    activation = pipeline_state.activations_buffer.popleft()
+IndexError: pop from an empty deque
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    activation = pipeline_state.activations_buffer.popleft()
+IndexError: pop from an empty deque
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    activation = pipeline_state.activations_buffer.popleft()
+IndexError: pop from an empty deque
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 124, in recv_from_pipeline_state_buffer
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    activation = pipeline_state.activations_buffer.popleft()
+IndexError: pop from an empty deque
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 124, in recv_from_pipeline_state_buffer
+    activation = pipeline_state.activations_buffer.popleft()
+IndexError: pop from an empty deque
+[2024-12-27 17:32:45,616] torch.distributed.elastic.multiprocessing.api: [ERROR] failed (exitcode: 1) local_rank: 0 (pid: 1061620) of binary: /fsx/nouamane/miniconda/envs/2-1-cu121/bin/python
+[2024-12-27 17:32:45,617] torch.distributed.elastic.multiprocessing.api: [ERROR] failed (exitcode: 1) local_rank: 0 (pid: 1332482) of binary: /fsx/nouamane/miniconda/envs/2-1-cu121/bin/python
+[2024-12-27 17:32:45,666] torch.distributed.elastic.multiprocessing.errors.error_handler: [ERROR] no error file defined for parent, to copy child error file (/tmp/torchelastic_qad6wean/13803345_9e_95_5w/attempt_0/0/error.json)
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+[2024-12-27 17:32:45,668] torch.distributed.elastic.multiprocessing.errors.error_handler: [ERROR] no error file defined for parent, to copy child error file (/tmp/torchelastic_ju3s8ytm/13803345_vts1gr1m/attempt_0/0/error.json)
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 264, in launch_agent
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    raise ChildFailedError(
+torch.distributed.elastic.multiprocessing.errors.ChildFailedError: 
+============================================================
+/fsx/nouamane/projects/nanotron/run_train.py FAILED
+------------------------------------------------------------
+Failures:
+[1]:
+  time      : 2024-12-27_17:32:20
+  host      : ip-26-0-166-125.ec2.internal
+  rank      : 1 (local_rank: 1)
+  exitcode  : 1 (pid: 1332483)
+  error_file: /tmp/torchelastic_qad6wean/13803345_9e_95_5w/attempt_0/1/error.json
+  traceback : Traceback (most recent call last):
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+      return f(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+      outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+      outputs = self.pipeline_engine.train_batch_iter(
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+      output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+      output = model(**micro_batch)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+      sharded_logits = self.model(
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+      return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+      hidden_encoder_states = encoder_block(**hidden_encoder_states)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 161, in forward
+      output = self.pp_block(**new_kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+      hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+      hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+      hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 201, in forward
+      return self.act(gate_states) * up_states
+  torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 896.00 MiB. GPU 1 has a total capacty of 79.33 GiB of which 103.94 MiB is free. Including non-PyTorch memory, this process has 79.22 GiB memory in use. Of the allocated memory 75.98 GiB is allocated by PyTorch, and 1.48 GiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  
+[2]:
+  time      : 2024-12-27_17:32:20
+  host      : ip-26-0-166-125.ec2.internal
+  rank      : 2 (local_rank: 2)
+  exitcode  : 1 (pid: 1332484)
+  error_file: /tmp/torchelastic_qad6wean/13803345_9e_95_5w/attempt_0/2/error.json
+  traceback : Traceback (most recent call last):
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+      return f(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+      outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+      outputs = self.pipeline_engine.train_batch_iter(
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+      output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+      output = model(**micro_batch)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+      sharded_logits = self.model(
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+      return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+      hidden_encoder_states = encoder_block(**hidden_encoder_states)
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 161, in forward
+      output = self.pp_block(**new_kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+      hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+      hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+      hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 201, in forward
+      return self.act(gate_states) * up_states
+  torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 896.00 MiB. GPU 2 has a total capacty of 79.33 GiB of which 831.94 MiB is free. Including non-PyTorch memory, this process has 78.51 GiB memory in use. Of the allocated memory 75.98 GiB is allocated by PyTorch, and 749.35 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  
+[3]:
+  time      : 2024-12-27_17:32:20
+  host      : ip-26-0-166-125.ec2.internal
+  rank      : 3 (local_rank: 3)
+  exitcode  : 1 (pid: 1332485)
+  error_file: /tmp/torchelastic_qad6wean/13803345_9e_95_5w/attempt_0/3/error.json
+  traceback : Traceback (most recent call last):
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+      return f(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+      outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+      outputs = self.pipeline_engine.train_batch_iter(
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+      output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+      output = model(**micro_batch)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+      sharded_logits = self.model(
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+      return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+      hidden_encoder_states = encoder_block(**hidden_encoder_states)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 161, in forward
+      output = self.pp_block(**new_kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+      hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+      hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+      hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 201, in forward
+      return self.act(gate_states) * up_states
+  torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 896.00 MiB. GPU 3 has a total capacty of 79.33 GiB of which 455.94 MiB is free. Including non-PyTorch memory, this process has 78.87 GiB memory in use. Of the allocated memory 75.98 GiB is allocated by PyTorch, and 1005.35 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  
+[4]:
+  time      : 2024-12-27_17:32:20
+  host      : ip-26-0-166-125.ec2.internal
+  rank      : 4 (local_rank: 4)
+  exitcode  : 1 (pid: 1332486)
+  error_file: /tmp/torchelastic_qad6wean/13803345_9e_95_5w/attempt_0/4/error.json
+  traceback : Traceback (most recent call last):
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+      return f(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+      outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+      outputs = self.pipeline_engine.train_batch_iter(
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+      output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+      output = model(**micro_batch)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+      sharded_logits = self.model(
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+      return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+      hidden_encoder_states = encoder_block(**hidden_encoder_states)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 161, in forward
+      output = self.pp_block(**new_kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+      hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+      hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+      hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 201, in forward
+      return self.act(gate_states) * up_states
+  torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 896.00 MiB. GPU 4 has a total capacty of 79.33 GiB of which 199.94 MiB is free. Including non-PyTorch memory, this process has 79.12 GiB memory in use. Of the allocated memory 75.98 GiB is allocated by PyTorch, and 1.23 GiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  
+[5]:
+  time      : 2024-12-27_17:32:20
+  host      : ip-26-0-166-125.ec2.internal
+  rank      : 5 (local_rank: 5)
+  exitcode  : 1 (pid: 1332487)
+  error_file: /tmp/torchelastic_qad6wean/13803345_9e_95_5w/attempt_0/5/error.json
+  traceback : Traceback (most recent call last):
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+      return f(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+      outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+      outputs = self.pipeline_engine.train_batch_iter(
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+      output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+      output = model(**micro_batch)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+      sharded_logits = self.model(
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+      return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+      hidden_encoder_states = encoder_block(**hidden_encoder_states)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 161, in forward
+      output = self.pp_block(**new_kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+      hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+      hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+      hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 201, in forward
+      return self.act(gate_states) * up_states
+  torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 896.00 MiB. GPU 5 has a total capacty of 79.33 GiB of which 319.94 MiB is free. Including non-PyTorch memory, this process has 79.01 GiB memory in use. Of the allocated memory 75.98 GiB is allocated by PyTorch, and 1.23 GiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  
+[6]:
+  time      : 2024-12-27_17:32:20
+  host      : ip-26-0-166-125.ec2.internal
+  rank      : 6 (local_rank: 6)
+  exitcode  : 1 (pid: 1332488)
+  error_file: /tmp/torchelastic_qad6wean/13803345_9e_95_5w/attempt_0/6/error.json
+  traceback : Traceback (most recent call last):
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+      return f(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+      outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+      outputs = self.pipeline_engine.train_batch_iter(
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+      output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+      output = model(**micro_batch)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+      sharded_logits = self.model(
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+      return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+      hidden_encoder_states = encoder_block(**hidden_encoder_states)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 161, in forward
+      output = self.pp_block(**new_kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+      hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+      hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+      hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 201, in forward
+      return self.act(gate_states) * up_states
+  torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 896.00 MiB. GPU 6 has a total capacty of 79.33 GiB of which 359.94 MiB is free. Including non-PyTorch memory, this process has 78.97 GiB memory in use. Of the allocated memory 75.98 GiB is allocated by PyTorch, and 1.23 GiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  
+[7]:
+  time      : 2024-12-27_17:32:20
+  host      : ip-26-0-166-125.ec2.internal
+  rank      : 7 (local_rank: 7)
+  exitcode  : 1 (pid: 1332489)
+  error_file: /tmp/torchelastic_qad6wean/13803345_9e_95_5w/attempt_0/7/error.json
+  traceback : Traceback (most recent call last):
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+      return f(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+      outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+      outputs = self.pipeline_engine.train_batch_iter(
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+      output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+      output = model(**micro_batch)
+    elastic_launch(
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+      sharded_logits = self.model(
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+      return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+      hidden_encoder_states = encoder_block(**hidden_encoder_states)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 161, in forward
+      output = self.pp_block(**new_kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+      hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+      hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+      hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 201, in forward
+      return self.act(gate_states) * up_states
+  torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 896.00 MiB. GPU 7 has a total capacty of 79.33 GiB of which 319.94 MiB is free. Including non-PyTorch memory, this process has 79.01 GiB memory in use. Of the allocated memory 75.98 GiB is allocated by PyTorch, and 1.23 GiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  
+------------------------------------------------------------
+Root Cause (first observed failure):
+[0]:
+  time      : 2024-12-27_17:32:20
+  host      : ip-26-0-166-125.ec2.internal
+  rank      : 0 (local_rank: 0)
+  exitcode  : 1 (pid: 1332482)
+  error_file: /tmp/torchelastic_qad6wean/13803345_9e_95_5w/attempt_0/0/error.json
+  traceback : Traceback (most recent call last):
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+      return f(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+      outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+      outputs = self.pipeline_engine.train_batch_iter(
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+      output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+      output = model(**micro_batch)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+      sharded_logits = self.model(
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+      return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+      hidden_encoder_states = encoder_block(**hidden_encoder_states)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 161, in forward
+      output = self.pp_block(**new_kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+      hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+      hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    return launch_agent(self._config, self._entrypoint, list(args))
+      hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 264, in launch_agent
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+      return row_linear(
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+      return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+      return super().apply(*args, **kwargs)  # type: ignore[misc]
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+      out = F.linear(tensor, weight, bias)
+  torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 512.00 MiB. GPU 0 has a total capacty of 79.33 GiB of which 95.94 MiB is free. Including non-PyTorch memory, this process has 79.22 GiB memory in use. Of the allocated memory 76.86 GiB is allocated by PyTorch, and 749.35 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  
+============================================================
+    raise ChildFailedError(
+torch.distributed.elastic.multiprocessing.errors.ChildFailedError: 
+============================================================
+/fsx/nouamane/projects/nanotron/run_train.py FAILED
+------------------------------------------------------------
+Failures:
+[1]:
+  time      : 2024-12-27_17:32:21
+  host      : ip-26-0-166-36.ec2.internal
+  rank      : 9 (local_rank: 1)
+  exitcode  : 1 (pid: 1061621)
+  error_file: /tmp/torchelastic_ju3s8ytm/13803345_vts1gr1m/attempt_0/1/error.json
+  traceback : Traceback (most recent call last):
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+      return f(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+      outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+      outputs = self.pipeline_engine.train_batch_iter(
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+      output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+      output = model(**micro_batch)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+      sharded_logits = self.model(
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+      return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+      hidden_encoder_states = encoder_block(**hidden_encoder_states)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+      new_kwargs[name] = recv_from_pipeline_state_buffer(
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 124, in recv_from_pipeline_state_buffer
+      activation = pipeline_state.activations_buffer.popleft()
+  IndexError: pop from an empty deque
+  
+[2]:
+  time      : 2024-12-27_17:32:21
+  host      : ip-26-0-166-36.ec2.internal
+  rank      : 10 (local_rank: 2)
+  exitcode  : 1 (pid: 1061622)
+  error_file: /tmp/torchelastic_ju3s8ytm/13803345_vts1gr1m/attempt_0/2/error.json
+  traceback : Traceback (most recent call last):
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+      return f(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+      outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+      outputs = self.pipeline_engine.train_batch_iter(
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+      output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+      output = model(**micro_batch)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+      sharded_logits = self.model(
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+      return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+      hidden_encoder_states = encoder_block(**hidden_encoder_states)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+      new_kwargs[name] = recv_from_pipeline_state_buffer(
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 124, in recv_from_pipeline_state_buffer
+      activation = pipeline_state.activations_buffer.popleft()
+  IndexError: pop from an empty deque
+  
+[3]:
+  time      : 2024-12-27_17:32:21
+  host      : ip-26-0-166-36.ec2.internal
+  rank      : 11 (local_rank: 3)
+  exitcode  : 1 (pid: 1061623)
+  error_file: /tmp/torchelastic_ju3s8ytm/13803345_vts1gr1m/attempt_0/3/error.json
+  traceback : Traceback (most recent call last):
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+      return f(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+      outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+      outputs = self.pipeline_engine.train_batch_iter(
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+      output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+      output = model(**micro_batch)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+      sharded_logits = self.model(
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+      return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+      hidden_encoder_states = encoder_block(**hidden_encoder_states)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+      new_kwargs[name] = recv_from_pipeline_state_buffer(
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 124, in recv_from_pipeline_state_buffer
+      activation = pipeline_state.activations_buffer.popleft()
+  IndexError: pop from an empty deque
+  
+[4]:
+  time      : 2024-12-27_17:32:21
+  host      : ip-26-0-166-36.ec2.internal
+  rank      : 12 (local_rank: 4)
+  exitcode  : 1 (pid: 1061624)
+  error_file: /tmp/torchelastic_ju3s8ytm/13803345_vts1gr1m/attempt_0/4/error.json
+  traceback : Traceback (most recent call last):
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+      return f(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+      outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+      outputs = self.pipeline_engine.train_batch_iter(
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+      output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+      output = model(**micro_batch)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+      sharded_logits = self.model(
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+      return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+      hidden_encoder_states = encoder_block(**hidden_encoder_states)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+      new_kwargs[name] = recv_from_pipeline_state_buffer(
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 124, in recv_from_pipeline_state_buffer
+      activation = pipeline_state.activations_buffer.popleft()
+  IndexError: pop from an empty deque
+  
+[5]:
+  time      : 2024-12-27_17:32:21
+  host      : ip-26-0-166-36.ec2.internal
+  rank      : 13 (local_rank: 5)
+  exitcode  : 1 (pid: 1061625)
+  error_file: /tmp/torchelastic_ju3s8ytm/13803345_vts1gr1m/attempt_0/5/error.json
+  traceback : Traceback (most recent call last):
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+      return f(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+      outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+      outputs = self.pipeline_engine.train_batch_iter(
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+      output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+      output = model(**micro_batch)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+      sharded_logits = self.model(
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+      return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+      hidden_encoder_states = encoder_block(**hidden_encoder_states)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+      new_kwargs[name] = recv_from_pipeline_state_buffer(
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 124, in recv_from_pipeline_state_buffer
+      activation = pipeline_state.activations_buffer.popleft()
+  IndexError: pop from an empty deque
+  
+[6]:
+  time      : 2024-12-27_17:32:21
+  host      : ip-26-0-166-36.ec2.internal
+  rank      : 14 (local_rank: 6)
+  exitcode  : 1 (pid: 1061626)
+  error_file: /tmp/torchelastic_ju3s8ytm/13803345_vts1gr1m/attempt_0/6/error.json
+  traceback : Traceback (most recent call last):
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+      return f(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+      outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+      outputs = self.pipeline_engine.train_batch_iter(
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+      output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+      output = model(**micro_batch)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+      sharded_logits = self.model(
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+      return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+      hidden_encoder_states = encoder_block(**hidden_encoder_states)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+      new_kwargs[name] = recv_from_pipeline_state_buffer(
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 124, in recv_from_pipeline_state_buffer
+      activation = pipeline_state.activations_buffer.popleft()
+  IndexError: pop from an empty deque
+  
+[7]:
+  time      : 2024-12-27_17:32:21
+  host      : ip-26-0-166-36.ec2.internal
+  rank      : 15 (local_rank: 7)
+  exitcode  : 1 (pid: 1061627)
+  error_file: /tmp/torchelastic_ju3s8ytm/13803345_vts1gr1m/attempt_0/7/error.json
+  traceback : Traceback (most recent call last):
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+      return f(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+      outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+      outputs = self.pipeline_engine.train_batch_iter(
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+      output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+      output = model(**micro_batch)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+      sharded_logits = self.model(
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+      return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+      hidden_encoder_states = encoder_block(**hidden_encoder_states)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+      new_kwargs[name] = recv_from_pipeline_state_buffer(
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 124, in recv_from_pipeline_state_buffer
+      activation = pipeline_state.activations_buffer.popleft()
+  IndexError: pop from an empty deque
+  
+------------------------------------------------------------
+Root Cause (first observed failure):
+[0]:
+  time      : 2024-12-27_17:32:21
+  host      : ip-26-0-166-36.ec2.internal
+  rank      : 8 (local_rank: 0)
+  exitcode  : 1 (pid: 1061620)
+  error_file: /tmp/torchelastic_ju3s8ytm/13803345_vts1gr1m/attempt_0/0/error.json
+  traceback : Traceback (most recent call last):
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+      return f(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+      outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+      outputs = self.pipeline_engine.train_batch_iter(
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+      output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+      output = model(**micro_batch)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+      sharded_logits = self.model(
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+      return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+      hidden_encoder_states = encoder_block(**hidden_encoder_states)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+      new_kwargs[name] = recv_from_pipeline_state_buffer(
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 124, in recv_from_pipeline_state_buffer
+      activation = pipeline_state.activations_buffer.popleft()
+  IndexError: pop from an empty deque
+  
+============================================================
+srun: error: ip-26-0-166-125: task 1: Exited with exit code 1
+srun: launch/slurm: _step_signal: Terminating StepId=13803345.0
+srun: error: ip-26-0-166-36: task 0: Exited with exit code 1
diff --git a/logs/13848147-bench_8.86G_dp4_tp1_pp2_acc1_mbs64_seq4096_zero1_tpmodeRED_vocab131k.out b/logs/13848147-bench_8.86G_dp4_tp1_pp2_acc1_mbs64_seq4096_zero1_tpmodeRED_vocab131k.out
new file mode 100644
index 0000000000000000000000000000000000000000..8d4d181066f5dc5363dbbb72ae427e10b315a2c5
--- /dev/null
+++ b/logs/13848147-bench_8.86G_dp4_tp1_pp2_acc1_mbs64_seq4096_zero1_tpmodeRED_vocab131k.out
@@ -0,0 +1,4500 @@
++ '[' -z 13848147 ']'
++ source /etc/profile.d/modules.sh
+++ . /usr/share/modules/init/bash
++++ unset _mlshdbg
++++ '[' 0 = 1 ']'
++++ unset _mlre _mlIFS
++++ '[' -n x ']'
++++ _mlIFS=' 	
+'
++++ IFS=' '
++++ '[' -n '' ']'
+++++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash autoinit
++++ _mlcode='module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+MODULEPATH_modshare=/etc/environment-modules/modules:1:/usr/share/modules/\$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1; export MODULEPATH_modshare;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+LOADEDMODULES=; export LOADEDMODULES;
+MODULEPATH=/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/\$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles; export MODULEPATH;
+test 0;'
++++ _mlret=0
++++ '[' -n x ']'
++++ IFS=' 	
+'
++++ unset _mlIFS
++++ unset _mlre _mlv _mlrv
++++ '[' 0 -eq 0 ']'
++++ eval 'module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+MODULEPATH_modshare=/etc/environment-modules/modules:1:/usr/share/modules/\$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1; export MODULEPATH_modshare;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+LOADEDMODULES=; export LOADEDMODULES;
+MODULEPATH=/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/\$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles; export MODULEPATH;
+test 0;'
+++++ MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl
+++++ export MODULES_CMD
+++++ ENV=/usr/share/modules/init/profile.sh
+++++ export ENV
+++++ MODULEPATH_modshare='/etc/environment-modules/modules:1:/usr/share/modules/$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1'
+++++ export MODULEPATH_modshare
+++++ BASH_ENV=/usr/share/modules/init/bash
+++++ export BASH_ENV
+++++ MODULESHOME=/usr/share/modules
+++++ export MODULESHOME
+++++ LOADEDMODULES=
+++++ export LOADEDMODULES
+++++ MODULEPATH='/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles'
+++++ export MODULEPATH
+++++ test 0
++++ '[' 0 = 1 ']'
++++ '[' -t 2 ']'
++++ export -f module
++++ export -f switchml
++++ '[' 5 -ge 3 ']'
++++ [[ ehxB =~ i ]]
++++ [[ ! :/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin: =~ :/usr/bin: ]]
++++ '[' '!' -n '' ']'
++++ MANPATH=:
++++ export MANPATH
+++++ manpath
++++ [[ ! :/fsx/nouamane/miniconda/envs/2-1-cu121/man:/fsx/nouamane/miniconda/envs/2-1-cu121/share/man:/fsx/nouamane/miniconda/man:/fsx/nouamane/miniconda/share/man:/usr/local/man:/usr/local/share/man:/usr/share/man:/opt/slurm/share/man:/admin/home/nouamane/.local/share/man:/admin/home/nouamane/.fzf/man:: =~ :/usr/share/man: ]]
++++ unset _mlcode _mlret
++++ '[' -n '' ']'
++ module load cuda/12.1
++ unset _mlshdbg
++ '[' 0 = 1 ']'
++ unset _mlre _mlIFS
++ '[' -n x ']'
++ _mlIFS=' 	
+'
++ IFS=' '
++ '[' -n '' ']'
+++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash load cuda/12.1
++ eval 'CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include;' export 'CPATH;
+LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:;' export 'LD_LIBRARY_PATH;
+FI_EFA_FORK_SAFE=1;' export 'FI_EFA_FORK_SAFE;
+MANPATH=/usr/local/cuda-12.1/share/man::;' export 'MANPATH;
+LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64;' export 'LIBRARY_PATH;
+_LMFILES_=/usr/share/modules/modulefiles/cuda/12.1;' export '_LMFILES_;
+LOADEDMODULES=cuda/12.1;' export 'LOADEDMODULES;
+MPI_PATH=/opt/amazon/openmpi;' export 'MPI_PATH;
+NCCL_HOME_modshare=/opt/nccl/build:1;' export 'NCCL_HOME_modshare;
+NCCL_PROTO=simple;' export 'NCCL_PROTO;
+MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1;' export 'MANPATH_modshare;
+LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1;' export 'LIBRARY_PATH_modshare;
+NCCL_SOCKET_IFNAME=enp;' export 'NCCL_SOCKET_IFNAME;
+AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl;' export 'AWS_OFI_NCCL_HOME;
+NCCL_HOME=/opt/nccl/build;' export 'NCCL_HOME;
+FI_PROVIDER=efa;' export 'FI_PROVIDER;
+AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1;' export 'AWS_OFI_NCCL_HOME_modshare;
+CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1;' export 'CPATH_modshare;
+LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2::1:/opt/nccl/build/lib:1:/opt/aws-ofi-nccl/lib:2:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/efa/lib:1;' export 'LD_LIBRARY_PATH_modshare;
+FI_EFA_ENABLE_SHM_TRANSFER=1;' export 'FI_EFA_ENABLE_SHM_TRANSFER;
+_LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1;' export '_LMFILES__modshare;
+LOADEDMODULES_modshare=cuda/12.1:1;' export 'LOADEDMODULES_modshare;
+MPI_PATH_modshare=/opt/amazon/openmpi:1;' export 'MPI_PATH_modshare;
+PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin;' export 'PATH;
+CUDA_HOME=/usr/local/cuda-12.1;' export 'CUDA_HOME;
+PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/opt/slurm/bin:1:/usr/sbin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1;' export 'PATH_modshare;
+test' '0;'
+++ CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include
+++ export CPATH
+++ LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:
+++ export LD_LIBRARY_PATH
+++ FI_EFA_FORK_SAFE=1
+++ export FI_EFA_FORK_SAFE
+++ MANPATH=/usr/local/cuda-12.1/share/man::
+++ export MANPATH
+++ LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64
+++ export LIBRARY_PATH
+++ _LMFILES_=/usr/share/modules/modulefiles/cuda/12.1
+++ export _LMFILES_
+++ LOADEDMODULES=cuda/12.1
+++ export LOADEDMODULES
+++ MPI_PATH=/opt/amazon/openmpi
+++ export MPI_PATH
+++ NCCL_HOME_modshare=/opt/nccl/build:1
+++ export NCCL_HOME_modshare
+++ NCCL_PROTO=simple
+++ export NCCL_PROTO
+++ MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1
+++ export MANPATH_modshare
+++ LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1
+++ export LIBRARY_PATH_modshare
+++ NCCL_SOCKET_IFNAME=enp
+++ export NCCL_SOCKET_IFNAME
+++ AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl
+++ export AWS_OFI_NCCL_HOME
+++ NCCL_HOME=/opt/nccl/build
+++ export NCCL_HOME
+++ FI_PROVIDER=efa
+++ export FI_PROVIDER
+++ AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1
+++ export AWS_OFI_NCCL_HOME_modshare
+++ CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1
+++ export CPATH_modshare
+++ LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2::1:/opt/nccl/build/lib:1:/opt/aws-ofi-nccl/lib:2:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/efa/lib:1
+++ export LD_LIBRARY_PATH_modshare
+++ FI_EFA_ENABLE_SHM_TRANSFER=1
+++ export FI_EFA_ENABLE_SHM_TRANSFER
+++ _LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1
+++ export _LMFILES__modshare
+++ LOADEDMODULES_modshare=cuda/12.1:1
+++ export LOADEDMODULES_modshare
+++ MPI_PATH_modshare=/opt/amazon/openmpi:1
+++ export MPI_PATH_modshare
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
+++ export PATH
+++ CUDA_HOME=/usr/local/cuda-12.1
+++ export CUDA_HOME
+++ PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/opt/slurm/bin:1:/usr/sbin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1
+++ export PATH_modshare
+++ test 0
++ _mlstatus=0
++ '[' -n x ']'
++ IFS=' 	
+'
++ unset _mlre _mlv _mlrv _mlIFS
++ '[' -n '' ']'
++ unset _mlshdbg
++ return 0
++ source /fsx/nouamane/miniconda/bin/activate
+++ _CONDA_ROOT=/fsx/nouamane/miniconda
+++ . /fsx/nouamane/miniconda/etc/profile.d/conda.sh
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ '[' -z x ']'
+++ conda activate
+++ local cmd=activate
+++ case "$cmd" in
+++ __conda_activate activate
+++ '[' -n '' ']'
+++ local ask_conda
++++ PS1=
++++ __conda_exe shell.posix activate
++++ /fsx/nouamane/miniconda/bin/conda shell.posix activate
+Error while loading conda entry point: conda-libmamba-solver (module 'libmambapy' has no attribute 'QueryFormat')
+++ ask_conda='. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''2'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_1='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/etc/conda/activate.d/libxml2_activate.sh"'
+++ eval '. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''2'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_1='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/etc/conda/activate.d/libxml2_activate.sh"'
++++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh
+++++ test -n ''
+++++ unset XML_CATALOG_FILES
+++++ unset xml_catalog_files_libxml2
++++ PS1='(base) '
++++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
++++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
++++ export CONDA_PREFIX=/fsx/nouamane/miniconda
++++ CONDA_PREFIX=/fsx/nouamane/miniconda
++++ export CONDA_SHLVL=2
++++ CONDA_SHLVL=2
++++ export CONDA_DEFAULT_ENV=base
++++ CONDA_DEFAULT_ENV=base
++++ export 'CONDA_PROMPT_MODIFIER=(base) '
++++ CONDA_PROMPT_MODIFIER='(base) '
++++ export CONDA_PREFIX_1=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ CONDA_PREFIX_1=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ . /fsx/nouamane/miniconda/etc/conda/activate.d/libxml2_activate.sh
+++++ test -n ''
+++++ xml_catalog_files_libxml2=
+++++ XML_CATALOG_FILES=
+++++ conda_catalog_files=
+++++ ifs_libxml2=' 	
+'
+++++ IFS=' '
+++++ rem=/fsx/nouamane/miniconda
+++++ for pre in ${rem}
+++++ test '' = /fsx/nouamane/miniconda
+++++ conda_catalog_files=/fsx/nouamane/miniconda
+++++ rem=
+++++ IFS=' 	
+'
+++++ conda_catalog_files='file:///fsx/nouamane/miniconda/etc/xml/catalog file:///etc/xml/catalog'
+++++ export 'XML_CATALOG_FILES=file:///fsx/nouamane/miniconda/etc/xml/catalog file:///etc/xml/catalog'
+++++ XML_CATALOG_FILES='file:///fsx/nouamane/miniconda/etc/xml/catalog file:///etc/xml/catalog'
+++++ unset conda_catalog_files ifs_libxml2 rem
+++ __conda_hashr
+++ '[' -n '' ']'
+++ '[' -n '' ']'
+++ hash -r
++ conda activate 2-1-cu121
++ local cmd=activate
++ case "$cmd" in
++ __conda_activate activate 2-1-cu121
++ '[' -n '' ']'
++ local ask_conda
+++ PS1='(base) '
+++ __conda_exe shell.posix activate 2-1-cu121
+++ /fsx/nouamane/miniconda/bin/conda shell.posix activate 2-1-cu121
+Error while loading conda entry point: conda-libmamba-solver (module 'libmambapy' has no attribute 'QueryFormat')
++ ask_conda='. "/fsx/nouamane/miniconda/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_2='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
++ eval '. "/fsx/nouamane/miniconda/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_2='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
+++ . /fsx/nouamane/miniconda/etc/conda/deactivate.d/libxml2_deactivate.sh
++++ test -n ''
++++ unset XML_CATALOG_FILES
++++ unset xml_catalog_files_libxml2
+++ PS1='(2-1-cu121) '
+++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
+++ export CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ export CONDA_SHLVL=3
+++ CONDA_SHLVL=3
+++ export CONDA_DEFAULT_ENV=2-1-cu121
+++ CONDA_DEFAULT_ENV=2-1-cu121
+++ export 'CONDA_PROMPT_MODIFIER=(2-1-cu121) '
+++ CONDA_PROMPT_MODIFIER='(2-1-cu121) '
+++ export CONDA_PREFIX_2=/fsx/nouamane/miniconda
+++ CONDA_PREFIX_2=/fsx/nouamane/miniconda
+++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ export _CE_M=
+++ _CE_M=
+++ export _CE_CONDA=
+++ _CE_CONDA=
+++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh
++++ test -n ''
++++ xml_catalog_files_libxml2=
++++ XML_CATALOG_FILES=
++++ conda_catalog_files=
++++ ifs_libxml2=' 	
+'
++++ IFS=' '
++++ rem=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ for pre in ${rem}
++++ test '' = /fsx/nouamane/miniconda/envs/2-1-cu121
++++ conda_catalog_files=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ rem=
++++ IFS=' 	
+'
++++ conda_catalog_files='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ export 'XML_CATALOG_FILES=file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ XML_CATALOG_FILES='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ unset conda_catalog_files ifs_libxml2 rem
++ __conda_hashr
++ '[' -n '' ']'
++ '[' -n '' ']'
++ hash -r
++ export PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
++ PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
++ '[' -z '' ']'
+++ scontrol show hostnames ip-26-0-164-0
++ export NODELIST=ip-26-0-164-0
++ NODELIST=ip-26-0-164-0
+++ scontrol show hostnames ip-26-0-164-0
+++ head -n1
++ export MASTER_NODE=ip-26-0-164-0
++ MASTER_NODE=ip-26-0-164-0
++ export MASTER_PORT=12356
++ MASTER_PORT=12356
++ export NNODES=1
++ NNODES=1
++ export GPUS_PER_NODE=8
++ GPUS_PER_NODE=8
++ export WORLD_SIZE=8
++ WORLD_SIZE=8
++ export CUDA_DEVICE_MAX_CONNECTIONS=1
++ CUDA_DEVICE_MAX_CONNECTIONS=1
++ export NCCL_DEBUG=INFO
++ NCCL_DEBUG=INFO
++ export NANOTRON_BENCHMARK=1
++ NANOTRON_BENCHMARK=1
++ export WANDB_MODE=disabled
++ WANDB_MODE=disabled
++ export TORCH_NCCL_ASYNC_ERROR_HANDLING=1
++ TORCH_NCCL_ASYNC_ERROR_HANDLING=1
++ export TORCH_DISTRIBUTED_DEBUG=DETAIL
++ TORCH_DISTRIBUTED_DEBUG=DETAIL
++ '[' -z '' ']'
++ echo '=== GPU Topology ==='
+=== GPU Topology ===
++ nvidia-smi topo -m
+	GPU0	GPU1	GPU2	GPU3	GPU4	GPU5	GPU6	GPU7	CPU Affinity	NUMA Affinity	GPU NUMA ID
+GPU0	 X 	NV18	NV18	NV18	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU1	NV18	 X 	NV18	NV18	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU2	NV18	NV18	 X 	NV18	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU3	NV18	NV18	NV18	 X 	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU4	NV18	NV18	NV18	NV18	 X 	NV18	NV18	NV18	48-95	1		N/A
+GPU5	NV18	NV18	NV18	NV18	NV18	 X 	NV18	NV18	48-95	1		N/A
+GPU6	NV18	NV18	NV18	NV18	NV18	NV18	 X 	NV18	48-95	1		N/A
+GPU7	NV18	NV18	NV18	NV18	NV18	NV18	NV18	 X 	48-95	1		N/A
+
+Legend:
+
+  X    = Self
+  SYS  = Connection traversing PCIe as well as the SMP interconnect between NUMA nodes (e.g., QPI/UPI)
+  NODE = Connection traversing PCIe as well as the interconnect between PCIe Host Bridges within a NUMA node
+  PHB  = Connection traversing PCIe as well as a PCIe Host Bridge (typically the CPU)
+  PXB  = Connection traversing multiple PCIe bridges (without traversing the PCIe Host Bridge)
+  PIX  = Connection traversing at most a single PCIe bridge
+  NV#  = Connection traversing a bonded set of # NVLinks
++ echo ==================
+==================
++ export SRUN_ALLOC_ARGS=
++ SRUN_ALLOC_ARGS=
++ echo 'Master node: ip-26-0-164-0'
+Master node: ip-26-0-164-0
++ echo 'All nodes: ip-26-0-164-0'
+All nodes: ip-26-0-164-0
++ echo 'World size: 8'
+World size: 8
++ '[' -n '' ']'
++ srun --wait=0 --kill-on-bad-exit=1 torchrun --nnodes=1 --nproc_per_node=8 --rdzv_id=13848147 --rdzv_backend=c10d --rdzv_endpoint=ip-26-0-164-0:12356 --max_restarts 0 --rdzv_conf timeout=60 /fsx/nouamane/projects/nanotron/run_train.py --config-file benchmark/configs/config_8.86G_dp4_tp1_pp2_acc1_mbs64_seq4096_zero1_tpmodeRED_vocab131k.yaml
+[2024-12-28 00:51:53,617] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-28 00:51:53,617] torch.distributed.run: [WARNING] 
+[2024-12-28 00:51:53,617] torch.distributed.run: [WARNING] *****************************************
+[2024-12-28 00:51:53,617] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-28 00:51:53,617] torch.distributed.run: [WARNING] *****************************************
+ip-26-0-164-0:2484741:2484741 [0] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-164-0:2484741:2484741 [0] NCCL INFO Bootstrap : Using enp71s0:26.0.164.0<0>
+ip-26-0-164-0:2484741:2484741 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-164-0:2484741:2484741 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-164-0:2484741:2484741 [0] NCCL INFO cudaDriverVersion 12020
+NCCL version 2.18.5+cuda12.2
+ip-26-0-164-0:2484747:2484747 [6] NCCL INFO cudaDriverVersion 12020
+ip-26-0-164-0:2484747:2484747 [6] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-164-0:2484745:2484745 [4] NCCL INFO cudaDriverVersion 12020
+ip-26-0-164-0:2484742:2484742 [1] NCCL INFO cudaDriverVersion 12020
+ip-26-0-164-0:2484745:2484745 [4] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-164-0:2484742:2484742 [1] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-164-0:2484746:2484746 [5] NCCL INFO cudaDriverVersion 12020
+ip-26-0-164-0:2484744:2484744 [3] NCCL INFO cudaDriverVersion 12020
+ip-26-0-164-0:2484746:2484746 [5] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-164-0:2484747:2484747 [6] NCCL INFO Bootstrap : Using enp71s0:26.0.164.0<0>
+ip-26-0-164-0:2484744:2484744 [3] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-164-0:2484748:2484748 [7] NCCL INFO cudaDriverVersion 12020
+ip-26-0-164-0:2484747:2484747 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-164-0:2484747:2484747 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-164-0:2484745:2484745 [4] NCCL INFO Bootstrap : Using enp71s0:26.0.164.0<0>
+ip-26-0-164-0:2484748:2484748 [7] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-164-0:2484742:2484742 [1] NCCL INFO Bootstrap : Using enp71s0:26.0.164.0<0>
+ip-26-0-164-0:2484743:2484743 [2] NCCL INFO cudaDriverVersion 12020
+ip-26-0-164-0:2484745:2484745 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-164-0:2484745:2484745 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-164-0:2484742:2484742 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-164-0:2484742:2484742 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-164-0:2484746:2484746 [5] NCCL INFO Bootstrap : Using enp71s0:26.0.164.0<0>
+ip-26-0-164-0:2484743:2484743 [2] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-164-0:2484744:2484744 [3] NCCL INFO Bootstrap : Using enp71s0:26.0.164.0<0>
+ip-26-0-164-0:2484748:2484748 [7] NCCL INFO Bootstrap : Using enp71s0:26.0.164.0<0>
+ip-26-0-164-0:2484746:2484746 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-164-0:2484746:2484746 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-164-0:2484744:2484744 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-164-0:2484744:2484744 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-164-0:2484748:2484748 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-164-0:2484748:2484748 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-164-0:2484743:2484743 [2] NCCL INFO Bootstrap : Using enp71s0:26.0.164.0<0>
+ip-26-0-164-0:2484743:2484743 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-164-0:2484743:2484743 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-164-0:2484741:2484901 [0] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-164-0:2484741:2484901 [0] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-164-0:2484741:2484901 [0] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-164-0:2484741:2484901 [0] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-164-0:2484741:2484901 [0] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-164-0:2484741:2484901 [0] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-164-0:2484741:2484901 [0] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-164-0:2484741:2484901 [0] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-164-0:2484741:2484901 [0] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-164-0:2484745:2484902 [4] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-164-0:2484745:2484902 [4] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-164-0:2484745:2484902 [4] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-164-0:2484745:2484902 [4] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-164-0:2484745:2484902 [4] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-164-0:2484745:2484902 [4] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-164-0:2484745:2484902 [4] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-164-0:2484745:2484902 [4] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-164-0:2484745:2484902 [4] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-164-0:2484747:2484908 [6] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-164-0:2484747:2484908 [6] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-164-0:2484743:2484903 [2] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-164-0:2484743:2484903 [2] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-164-0:2484746:2484904 [5] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-164-0:2484746:2484904 [5] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-164-0:2484747:2484908 [6] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-164-0:2484747:2484908 [6] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-164-0:2484747:2484908 [6] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-164-0:2484747:2484908 [6] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-164-0:2484747:2484908 [6] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-164-0:2484747:2484908 [6] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-164-0:2484747:2484908 [6] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-164-0:2484748:2484905 [7] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-164-0:2484748:2484905 [7] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-164-0:2484743:2484903 [2] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-164-0:2484743:2484903 [2] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-164-0:2484743:2484903 [2] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-164-0:2484743:2484903 [2] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-164-0:2484743:2484903 [2] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-164-0:2484743:2484903 [2] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-164-0:2484743:2484903 [2] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-164-0:2484746:2484904 [5] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-164-0:2484746:2484904 [5] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-164-0:2484746:2484904 [5] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-164-0:2484746:2484904 [5] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-164-0:2484746:2484904 [5] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-164-0:2484746:2484904 [5] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-164-0:2484746:2484904 [5] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-164-0:2484744:2484906 [3] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-164-0:2484744:2484906 [3] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-164-0:2484742:2484907 [1] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-164-0:2484742:2484907 [1] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-164-0:2484748:2484905 [7] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-164-0:2484748:2484905 [7] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-164-0:2484748:2484905 [7] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-164-0:2484748:2484905 [7] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-164-0:2484748:2484905 [7] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-164-0:2484748:2484905 [7] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-164-0:2484748:2484905 [7] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-164-0:2484742:2484907 [1] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-164-0:2484742:2484907 [1] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-164-0:2484742:2484907 [1] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-164-0:2484742:2484907 [1] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-164-0:2484742:2484907 [1] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-164-0:2484742:2484907 [1] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-164-0:2484742:2484907 [1] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-164-0:2484744:2484906 [3] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-164-0:2484744:2484906 [3] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-164-0:2484744:2484906 [3] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-164-0:2484744:2484906 [3] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-164-0:2484744:2484906 [3] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-164-0:2484744:2484906 [3] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-164-0:2484744:2484906 [3] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-164-0:2484741:2484901 [0] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-164-0:2484745:2484902 [4] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-164-0:2484747:2484908 [6] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-164-0:2484748:2484905 [7] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-164-0:2484744:2484906 [3] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-164-0:2484743:2484903 [2] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-164-0:2484742:2484907 [1] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-164-0:2484746:2484904 [5] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-164-0:2484741:2484901 [0] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-164-0:2484741:2484901 [0] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-164-0:2484741:2484901 [0] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-164-0:2484741:2484901 [0] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-164-0:2484741:2484901 [0] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-164-0:2484741:2484901 [0] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-164-0:2484741:2484901 [0] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-164-0:2484741:2484901 [0] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-164-0:2484741:2484901 [0] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-164-0:2484741:2484901 [0] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-164-0:2484741:2484901 [0] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-164-0:2484741:2484901 [0] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-164-0:2484741:2484901 [0] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-164-0:2484741:2484901 [0] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-164-0:2484741:2484901 [0] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-164-0:2484741:2484901 [0] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-164-0:2484741:2484901 [0] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-164-0:2484741:2484901 [0] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-164-0:2484741:2484901 [0] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-164-0:2484741:2484901 [0] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-164-0:2484741:2484901 [0] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-164-0:2484741:2484901 [0] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-164-0:2484741:2484901 [0] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-164-0:2484741:2484901 [0] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-164-0:2484741:2484901 [0] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-164-0:2484741:2484901 [0] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-164-0:2484741:2484901 [0] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-164-0:2484741:2484901 [0] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-164-0:2484741:2484901 [0] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-164-0:2484741:2484901 [0] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-164-0:2484741:2484901 [0] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-164-0:2484741:2484901 [0] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-164-0:2484741:2484901 [0] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/181
+ip-26-0-164-0:2484741:2484901 [0] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-164-0:2484746:2484904 [5] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-164-0:2484746:2484904 [5] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-164-0:2484746:2484904 [5] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-164-0:2484746:2484904 [5] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-164-0:2484746:2484904 [5] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-164-0:2484746:2484904 [5] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-164-0:2484746:2484904 [5] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-164-0:2484746:2484904 [5] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-164-0:2484746:2484904 [5] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-164-0:2484746:2484904 [5] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-164-0:2484746:2484904 [5] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-164-0:2484746:2484904 [5] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-164-0:2484746:2484904 [5] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-164-0:2484746:2484904 [5] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-164-0:2484746:2484904 [5] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-164-0:2484746:2484904 [5] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-164-0:2484746:2484904 [5] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-164-0:2484746:2484904 [5] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-164-0:2484746:2484904 [5] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-164-0:2484746:2484904 [5] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-164-0:2484746:2484904 [5] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-164-0:2484746:2484904 [5] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-164-0:2484746:2484904 [5] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-164-0:2484746:2484904 [5] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-164-0:2484746:2484904 [5] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-164-0:2484746:2484904 [5] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-164-0:2484746:2484904 [5] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-164-0:2484746:2484904 [5] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-164-0:2484746:2484904 [5] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-164-0:2484746:2484904 [5] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-164-0:2484746:2484904 [5] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-164-0:2484746:2484904 [5] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-164-0:2484746:2484904 [5] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/168
+ip-26-0-164-0:2484746:2484904 [5] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-164-0:2484744:2484906 [3] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-164-0:2484744:2484906 [3] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-164-0:2484744:2484906 [3] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-164-0:2484744:2484906 [3] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-164-0:2484744:2484906 [3] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-164-0:2484744:2484906 [3] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-164-0:2484744:2484906 [3] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-164-0:2484744:2484906 [3] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-164-0:2484744:2484906 [3] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-164-0:2484744:2484906 [3] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-164-0:2484744:2484906 [3] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-164-0:2484744:2484906 [3] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-164-0:2484744:2484906 [3] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-164-0:2484744:2484906 [3] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-164-0:2484744:2484906 [3] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-164-0:2484744:2484906 [3] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-164-0:2484744:2484906 [3] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-164-0:2484744:2484906 [3] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-164-0:2484744:2484906 [3] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-164-0:2484744:2484906 [3] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-164-0:2484744:2484906 [3] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-164-0:2484744:2484906 [3] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-164-0:2484744:2484906 [3] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-164-0:2484744:2484906 [3] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-164-0:2484744:2484906 [3] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-164-0:2484744:2484906 [3] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-164-0:2484744:2484906 [3] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-164-0:2484744:2484906 [3] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-164-0:2484744:2484906 [3] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-164-0:2484744:2484906 [3] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-164-0:2484744:2484906 [3] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-164-0:2484744:2484906 [3] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-164-0:2484744:2484906 [3] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/168
+ip-26-0-164-0:2484744:2484906 [3] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-164-0:2484741:2484901 [0] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-164-0:2484741:2484901 [0] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-164-0:2484741:2484901 [0] NCCL INFO Using network Libfabric
+ip-26-0-164-0:2484748:2484905 [7] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-164-0:2484748:2484905 [7] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-164-0:2484748:2484905 [7] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-164-0:2484748:2484905 [7] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-164-0:2484748:2484905 [7] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-164-0:2484748:2484905 [7] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-164-0:2484748:2484905 [7] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-164-0:2484748:2484905 [7] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-164-0:2484748:2484905 [7] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-164-0:2484748:2484905 [7] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-164-0:2484748:2484905 [7] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-164-0:2484748:2484905 [7] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-164-0:2484748:2484905 [7] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-164-0:2484748:2484905 [7] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-164-0:2484748:2484905 [7] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-164-0:2484748:2484905 [7] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-164-0:2484748:2484905 [7] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-164-0:2484748:2484905 [7] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-164-0:2484748:2484905 [7] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-164-0:2484748:2484905 [7] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-164-0:2484748:2484905 [7] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-164-0:2484748:2484905 [7] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-164-0:2484748:2484905 [7] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-164-0:2484748:2484905 [7] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-164-0:2484748:2484905 [7] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-164-0:2484748:2484905 [7] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-164-0:2484748:2484905 [7] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-164-0:2484748:2484905 [7] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-164-0:2484748:2484905 [7] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-164-0:2484748:2484905 [7] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-164-0:2484748:2484905 [7] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-164-0:2484748:2484905 [7] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-164-0:2484748:2484905 [7] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/168
+ip-26-0-164-0:2484748:2484905 [7] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-164-0:2484745:2484902 [4] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-164-0:2484745:2484902 [4] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-164-0:2484745:2484902 [4] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-164-0:2484745:2484902 [4] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-164-0:2484745:2484902 [4] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-164-0:2484745:2484902 [4] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-164-0:2484745:2484902 [4] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-164-0:2484745:2484902 [4] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-164-0:2484745:2484902 [4] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-164-0:2484745:2484902 [4] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-164-0:2484745:2484902 [4] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-164-0:2484745:2484902 [4] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-164-0:2484745:2484902 [4] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-164-0:2484745:2484902 [4] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-164-0:2484745:2484902 [4] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-164-0:2484745:2484902 [4] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-164-0:2484745:2484902 [4] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-164-0:2484745:2484902 [4] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-164-0:2484745:2484902 [4] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-164-0:2484745:2484902 [4] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-164-0:2484745:2484902 [4] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-164-0:2484745:2484902 [4] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-164-0:2484745:2484902 [4] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-164-0:2484745:2484902 [4] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-164-0:2484745:2484902 [4] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-164-0:2484745:2484902 [4] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-164-0:2484745:2484902 [4] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-164-0:2484745:2484902 [4] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-164-0:2484745:2484902 [4] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-164-0:2484745:2484902 [4] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-164-0:2484745:2484902 [4] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-164-0:2484745:2484902 [4] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-164-0:2484745:2484902 [4] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/168
+ip-26-0-164-0:2484745:2484902 [4] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-164-0:2484743:2484903 [2] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-164-0:2484743:2484903 [2] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-164-0:2484743:2484903 [2] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-164-0:2484743:2484903 [2] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-164-0:2484743:2484903 [2] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-164-0:2484743:2484903 [2] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-164-0:2484743:2484903 [2] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-164-0:2484743:2484903 [2] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-164-0:2484743:2484903 [2] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-164-0:2484743:2484903 [2] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-164-0:2484743:2484903 [2] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-164-0:2484743:2484903 [2] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-164-0:2484743:2484903 [2] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-164-0:2484743:2484903 [2] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-164-0:2484743:2484903 [2] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-164-0:2484743:2484903 [2] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-164-0:2484743:2484903 [2] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-164-0:2484743:2484903 [2] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-164-0:2484743:2484903 [2] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-164-0:2484743:2484903 [2] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-164-0:2484743:2484903 [2] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-164-0:2484743:2484903 [2] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-164-0:2484743:2484903 [2] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-164-0:2484743:2484903 [2] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-164-0:2484743:2484903 [2] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-164-0:2484743:2484903 [2] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-164-0:2484743:2484903 [2] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-164-0:2484743:2484903 [2] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-164-0:2484743:2484903 [2] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-164-0:2484743:2484903 [2] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-164-0:2484743:2484903 [2] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-164-0:2484743:2484903 [2] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-164-0:2484743:2484903 [2] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/168
+ip-26-0-164-0:2484743:2484903 [2] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-164-0:2484747:2484908 [6] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-164-0:2484747:2484908 [6] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-164-0:2484747:2484908 [6] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-164-0:2484747:2484908 [6] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-164-0:2484747:2484908 [6] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-164-0:2484747:2484908 [6] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-164-0:2484747:2484908 [6] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-164-0:2484747:2484908 [6] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-164-0:2484747:2484908 [6] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-164-0:2484747:2484908 [6] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-164-0:2484747:2484908 [6] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-164-0:2484747:2484908 [6] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-164-0:2484747:2484908 [6] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-164-0:2484747:2484908 [6] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-164-0:2484747:2484908 [6] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-164-0:2484747:2484908 [6] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-164-0:2484747:2484908 [6] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-164-0:2484747:2484908 [6] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-164-0:2484747:2484908 [6] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-164-0:2484747:2484908 [6] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-164-0:2484747:2484908 [6] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-164-0:2484747:2484908 [6] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-164-0:2484747:2484908 [6] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-164-0:2484747:2484908 [6] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-164-0:2484747:2484908 [6] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-164-0:2484747:2484908 [6] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-164-0:2484747:2484908 [6] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-164-0:2484747:2484908 [6] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-164-0:2484747:2484908 [6] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-164-0:2484747:2484908 [6] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-164-0:2484747:2484908 [6] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-164-0:2484747:2484908 [6] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-164-0:2484742:2484907 [1] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-164-0:2484742:2484907 [1] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-164-0:2484742:2484907 [1] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-164-0:2484742:2484907 [1] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-164-0:2484742:2484907 [1] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-164-0:2484742:2484907 [1] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-164-0:2484742:2484907 [1] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-164-0:2484742:2484907 [1] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-164-0:2484742:2484907 [1] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-164-0:2484742:2484907 [1] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-164-0:2484742:2484907 [1] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-164-0:2484742:2484907 [1] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-164-0:2484742:2484907 [1] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-164-0:2484742:2484907 [1] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-164-0:2484742:2484907 [1] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-164-0:2484742:2484907 [1] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-164-0:2484742:2484907 [1] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-164-0:2484742:2484907 [1] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-164-0:2484742:2484907 [1] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-164-0:2484742:2484907 [1] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-164-0:2484742:2484907 [1] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-164-0:2484742:2484907 [1] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-164-0:2484742:2484907 [1] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-164-0:2484742:2484907 [1] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-164-0:2484742:2484907 [1] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-164-0:2484742:2484907 [1] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-164-0:2484742:2484907 [1] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-164-0:2484742:2484907 [1] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-164-0:2484742:2484907 [1] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-164-0:2484742:2484907 [1] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-164-0:2484742:2484907 [1] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-164-0:2484742:2484907 [1] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-164-0:2484747:2484908 [6] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/168
+ip-26-0-164-0:2484742:2484907 [1] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/168
+ip-26-0-164-0:2484747:2484908 [6] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-164-0:2484742:2484907 [1] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-164-0:2484746:2484904 [5] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-164-0:2484746:2484904 [5] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-164-0:2484746:2484904 [5] NCCL INFO Using network Libfabric
+ip-26-0-164-0:2484744:2484906 [3] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-164-0:2484744:2484906 [3] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-164-0:2484744:2484906 [3] NCCL INFO Using network Libfabric
+ip-26-0-164-0:2484745:2484902 [4] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-164-0:2484745:2484902 [4] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-164-0:2484745:2484902 [4] NCCL INFO Using network Libfabric
+ip-26-0-164-0:2484743:2484903 [2] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-164-0:2484743:2484903 [2] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-164-0:2484743:2484903 [2] NCCL INFO Using network Libfabric
+ip-26-0-164-0:2484742:2484907 [1] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-164-0:2484742:2484907 [1] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-164-0:2484742:2484907 [1] NCCL INFO Using network Libfabric
+ip-26-0-164-0:2484748:2484905 [7] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-164-0:2484748:2484905 [7] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-164-0:2484748:2484905 [7] NCCL INFO Using network Libfabric
+ip-26-0-164-0:2484747:2484908 [6] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-164-0:2484747:2484908 [6] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-164-0:2484747:2484908 [6] NCCL INFO Using network Libfabric
+ip-26-0-164-0:2484742:2484907 [1] NCCL INFO comm 0x98ddfe0 rank 1 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0xb21dbb55538c0b55 - Init START
+ip-26-0-164-0:2484741:2484901 [0] NCCL INFO comm 0x844f620 rank 0 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0xb21dbb55538c0b55 - Init START
+ip-26-0-164-0:2484745:2484902 [4] NCCL INFO comm 0x956ef50 rank 4 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0xb21dbb55538c0b55 - Init START
+ip-26-0-164-0:2484743:2484903 [2] NCCL INFO comm 0x88fbde0 rank 2 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0xb21dbb55538c0b55 - Init START
+ip-26-0-164-0:2484748:2484905 [7] NCCL INFO comm 0x842d180 rank 7 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0xb21dbb55538c0b55 - Init START
+ip-26-0-164-0:2484747:2484908 [6] NCCL INFO comm 0x9177540 rank 6 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0xb21dbb55538c0b55 - Init START
+ip-26-0-164-0:2484744:2484906 [3] NCCL INFO comm 0x9e84410 rank 3 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0xb21dbb55538c0b55 - Init START
+ip-26-0-164-0:2484746:2484904 [5] NCCL INFO comm 0x86ef420 rank 5 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0xb21dbb55538c0b55 - Init START
+ip-26-0-164-0:2484742:2484907 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/168
+ip-26-0-164-0:2484748:2484905 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/168
+ip-26-0-164-0:2484741:2484901 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/181
+ip-26-0-164-0:2484747:2484908 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/168
+ip-26-0-164-0:2484746:2484904 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/168
+ip-26-0-164-0:2484745:2484902 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/168
+ip-26-0-164-0:2484744:2484906 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/168
+ip-26-0-164-0:2484743:2484903 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/168
+ip-26-0-164-0:2484741:2484901 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-164-0:2484741:2484901 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-164-0:2484748:2484905 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-164-0:2484748:2484905 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-164-0:2484744:2484906 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-164-0:2484744:2484906 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-164-0:2484746:2484904 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-164-0:2484746:2484904 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-164-0:2484745:2484902 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-164-0:2484745:2484902 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-164-0:2484747:2484908 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-164-0:2484747:2484908 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-164-0:2484742:2484907 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-164-0:2484742:2484907 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-164-0:2484743:2484903 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-164-0:2484743:2484903 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-164-0:2484743:2484903 [2] NCCL INFO Trees [0] 3/-1/-1->2->1 [1] 3/-1/-1->2->1 [2] 3/-1/-1->2->1 [3] 3/-1/-1->2->1 [4] 3/-1/-1->2->1 [5] 3/-1/-1->2->1 [6] 3/-1/-1->2->1 [7] 3/-1/-1->2->1 [8] 3/-1/-1->2->1 [9] 3/-1/-1->2->1 [10] 3/-1/-1->2->1 [11] 3/-1/-1->2->1 [12] 3/-1/-1->2->1 [13] 3/-1/-1->2->1 [14] 3/-1/-1->2->1 [15] 3/-1/-1->2->1 [16] 3/-1/-1->2->1 [17] 3/-1/-1->2->1 [18] 3/-1/-1->2->1 [19] 3/-1/-1->2->1 [20] 3/-1/-1->2->1 [21] 3/-1/-1->2->1 [22] 3/-1/-1->2->1 [23] 3/-1/-1->2->1
+ip-26-0-164-0:2484743:2484903 [2] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-164-0:2484745:2484902 [4] NCCL INFO Trees [0] 5/-1/-1->4->3 [1] 5/-1/-1->4->3 [2] 5/-1/-1->4->3 [3] 5/-1/-1->4->3 [4] 5/-1/-1->4->3 [5] 5/-1/-1->4->3 [6] 5/-1/-1->4->3 [7] 5/-1/-1->4->3 [8] 5/-1/-1->4->3 [9] 5/-1/-1->4->3 [10] 5/-1/-1->4->3 [11] 5/-1/-1->4->3 [12] 5/-1/-1->4->3 [13] 5/-1/-1->4->3 [14] 5/-1/-1->4->3 [15] 5/-1/-1->4->3 [16] 5/-1/-1->4->3 [17] 5/-1/-1->4->3 [18] 5/-1/-1->4->3 [19] 5/-1/-1->4->3 [20] 5/-1/-1->4->3 [21] 5/-1/-1->4->3 [22] 5/-1/-1->4->3 [23] 5/-1/-1->4->3
+ip-26-0-164-0:2484742:2484907 [1] NCCL INFO Trees [0] 2/-1/-1->1->0 [1] 2/-1/-1->1->0 [2] 2/-1/-1->1->0 [3] 2/-1/-1->1->0 [4] 2/-1/-1->1->0 [5] 2/-1/-1->1->0 [6] 2/-1/-1->1->0 [7] 2/-1/-1->1->0 [8] 2/-1/-1->1->0 [9] 2/-1/-1->1->0 [10] 2/-1/-1->1->0 [11] 2/-1/-1->1->0 [12] 2/-1/-1->1->0 [13] 2/-1/-1->1->0 [14] 2/-1/-1->1->0 [15] 2/-1/-1->1->0 [16] 2/-1/-1->1->0 [17] 2/-1/-1->1->0 [18] 2/-1/-1->1->0 [19] 2/-1/-1->1->0 [20] 2/-1/-1->1->0 [21] 2/-1/-1->1->0 [22] 2/-1/-1->1->0 [23] 2/-1/-1->1->0
+ip-26-0-164-0:2484746:2484904 [5] NCCL INFO Trees [0] 6/-1/-1->5->4 [1] 6/-1/-1->5->4 [2] 6/-1/-1->5->4 [3] 6/-1/-1->5->4 [4] 6/-1/-1->5->4 [5] 6/-1/-1->5->4 [6] 6/-1/-1->5->4 [7] 6/-1/-1->5->4 [8] 6/-1/-1->5->4 [9] 6/-1/-1->5->4 [10] 6/-1/-1->5->4 [11] 6/-1/-1->5->4 [12] 6/-1/-1->5->4 [13] 6/-1/-1->5->4 [14] 6/-1/-1->5->4 [15] 6/-1/-1->5->4 [16] 6/-1/-1->5->4 [17] 6/-1/-1->5->4 [18] 6/-1/-1->5->4 [19] 6/-1/-1->5->4 [20] 6/-1/-1->5->4 [21] 6/-1/-1->5->4 [22] 6/-1/-1->5->4 [23] 6/-1/-1->5->4
+ip-26-0-164-0:2484745:2484902 [4] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-164-0:2484742:2484907 [1] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-164-0:2484746:2484904 [5] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-164-0:2484747:2484908 [6] NCCL INFO Trees [0] 7/-1/-1->6->5 [1] 7/-1/-1->6->5 [2] 7/-1/-1->6->5 [3] 7/-1/-1->6->5 [4] 7/-1/-1->6->5 [5] 7/-1/-1->6->5 [6] 7/-1/-1->6->5 [7] 7/-1/-1->6->5 [8] 7/-1/-1->6->5 [9] 7/-1/-1->6->5 [10] 7/-1/-1->6->5 [11] 7/-1/-1->6->5 [12] 7/-1/-1->6->5 [13] 7/-1/-1->6->5 [14] 7/-1/-1->6->5 [15] 7/-1/-1->6->5 [16] 7/-1/-1->6->5 [17] 7/-1/-1->6->5 [18] 7/-1/-1->6->5 [19] 7/-1/-1->6->5 [20] 7/-1/-1->6->5 [21] 7/-1/-1->6->5 [22] 7/-1/-1->6->5 [23] 7/-1/-1->6->5
+ip-26-0-164-0:2484747:2484908 [6] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-164-0:2484748:2484905 [7] NCCL INFO Trees [0] -1/-1/-1->7->6 [1] -1/-1/-1->7->6 [2] -1/-1/-1->7->6 [3] -1/-1/-1->7->6 [4] -1/-1/-1->7->6 [5] -1/-1/-1->7->6 [6] -1/-1/-1->7->6 [7] -1/-1/-1->7->6 [8] -1/-1/-1->7->6 [9] -1/-1/-1->7->6 [10] -1/-1/-1->7->6 [11] -1/-1/-1->7->6 [12] -1/-1/-1->7->6 [13] -1/-1/-1->7->6 [14] -1/-1/-1->7->6 [15] -1/-1/-1->7->6 [16] -1/-1/-1->7->6 [17] -1/-1/-1->7->6 [18] -1/-1/-1->7->6 [19] -1/-1/-1->7->6 [20] -1/-1/-1->7->6 [21] -1/-1/-1->7->6 [22] -1/-1/-1->7->6 [23] -1/-1/-1->7->6
+ip-26-0-164-0:2484748:2484905 [7] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-164-0:2484744:2484906 [3] NCCL INFO Trees [0] 4/-1/-1->3->2 [1] 4/-1/-1->3->2 [2] 4/-1/-1->3->2 [3] 4/-1/-1->3->2 [4] 4/-1/-1->3->2 [5] 4/-1/-1->3->2 [6] 4/-1/-1->3->2 [7] 4/-1/-1->3->2 [8] 4/-1/-1->3->2 [9] 4/-1/-1->3->2 [10] 4/-1/-1->3->2 [11] 4/-1/-1->3->2 [12] 4/-1/-1->3->2 [13] 4/-1/-1->3->2 [14] 4/-1/-1->3->2 [15] 4/-1/-1->3->2 [16] 4/-1/-1->3->2 [17] 4/-1/-1->3->2 [18] 4/-1/-1->3->2 [19] 4/-1/-1->3->2 [20] 4/-1/-1->3->2 [21] 4/-1/-1->3->2 [22] 4/-1/-1->3->2 [23] 4/-1/-1->3->2
+ip-26-0-164-0:2484741:2484901 [0] NCCL INFO Channel 00/24 :    0   1   2   3   4   5   6   7
+ip-26-0-164-0:2484744:2484906 [3] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-164-0:2484741:2484901 [0] NCCL INFO Channel 01/24 :    0   1   2   3   4   5   6   7
+ip-26-0-164-0:2484741:2484901 [0] NCCL INFO Channel 02/24 :    0   1   2   3   4   5   6   7
+ip-26-0-164-0:2484741:2484901 [0] NCCL INFO Channel 03/24 :    0   1   2   3   4   5   6   7
+ip-26-0-164-0:2484741:2484901 [0] NCCL INFO Channel 04/24 :    0   1   2   3   4   5   6   7
+ip-26-0-164-0:2484741:2484901 [0] NCCL INFO Channel 05/24 :    0   1   2   3   4   5   6   7
+ip-26-0-164-0:2484741:2484901 [0] NCCL INFO Channel 06/24 :    0   1   2   3   4   5   6   7
+ip-26-0-164-0:2484741:2484901 [0] NCCL INFO Channel 07/24 :    0   1   2   3   4   5   6   7
+ip-26-0-164-0:2484741:2484901 [0] NCCL INFO Channel 08/24 :    0   1   2   3   4   5   6   7
+ip-26-0-164-0:2484741:2484901 [0] NCCL INFO Channel 09/24 :    0   1   2   3   4   5   6   7
+ip-26-0-164-0:2484741:2484901 [0] NCCL INFO Channel 10/24 :    0   1   2   3   4   5   6   7
+ip-26-0-164-0:2484741:2484901 [0] NCCL INFO Channel 11/24 :    0   1   2   3   4   5   6   7
+ip-26-0-164-0:2484741:2484901 [0] NCCL INFO Channel 12/24 :    0   1   2   3   4   5   6   7
+ip-26-0-164-0:2484741:2484901 [0] NCCL INFO Channel 13/24 :    0   1   2   3   4   5   6   7
+ip-26-0-164-0:2484741:2484901 [0] NCCL INFO Channel 14/24 :    0   1   2   3   4   5   6   7
+ip-26-0-164-0:2484741:2484901 [0] NCCL INFO Channel 15/24 :    0   1   2   3   4   5   6   7
+ip-26-0-164-0:2484741:2484901 [0] NCCL INFO Channel 16/24 :    0   1   2   3   4   5   6   7
+ip-26-0-164-0:2484741:2484901 [0] NCCL INFO Channel 17/24 :    0   1   2   3   4   5   6   7
+ip-26-0-164-0:2484741:2484901 [0] NCCL INFO Channel 18/24 :    0   1   2   3   4   5   6   7
+ip-26-0-164-0:2484741:2484901 [0] NCCL INFO Channel 19/24 :    0   1   2   3   4   5   6   7
+ip-26-0-164-0:2484741:2484901 [0] NCCL INFO Channel 20/24 :    0   1   2   3   4   5   6   7
+ip-26-0-164-0:2484741:2484901 [0] NCCL INFO Channel 21/24 :    0   1   2   3   4   5   6   7
+ip-26-0-164-0:2484741:2484901 [0] NCCL INFO Channel 22/24 :    0   1   2   3   4   5   6   7
+ip-26-0-164-0:2484741:2484901 [0] NCCL INFO Channel 23/24 :    0   1   2   3   4   5   6   7
+ip-26-0-164-0:2484741:2484901 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] 1/-1/-1->0->-1 [7] 1/-1/-1->0->-1 [8] 1/-1/-1->0->-1 [9] 1/-1/-1->0->-1 [10] 1/-1/-1->0->-1 [11] 1/-1/-1->0->-1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] 1/-1/-1->0->-1 [19] 1/-1/-1->0->-1 [20] 1/-1/-1->0->-1 [21] 1/-1/-1->0->-1 [22] 1/-1/-1->0->-1 [23] 1/-1/-1->0->-1
+ip-26-0-164-0:2484741:2484901 [0] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-164-0:2484743:2484903 [2] NCCL INFO Channel 00/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484743:2484903 [2] NCCL INFO Channel 01/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484743:2484903 [2] NCCL INFO Channel 02/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484743:2484903 [2] NCCL INFO Channel 03/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484745:2484902 [4] NCCL INFO Channel 00/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484743:2484903 [2] NCCL INFO Channel 04/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484745:2484902 [4] NCCL INFO Channel 01/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484743:2484903 [2] NCCL INFO Channel 05/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484741:2484901 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484742:2484907 [1] NCCL INFO Channel 00/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484745:2484902 [4] NCCL INFO Channel 02/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484743:2484903 [2] NCCL INFO Channel 06/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484741:2484901 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484742:2484907 [1] NCCL INFO Channel 01/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484745:2484902 [4] NCCL INFO Channel 03/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484743:2484903 [2] NCCL INFO Channel 07/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484741:2484901 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484742:2484907 [1] NCCL INFO Channel 02/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484744:2484906 [3] NCCL INFO Channel 00/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484745:2484902 [4] NCCL INFO Channel 04/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484743:2484903 [2] NCCL INFO Channel 08/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484741:2484901 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484742:2484907 [1] NCCL INFO Channel 03/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484744:2484906 [3] NCCL INFO Channel 01/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484745:2484902 [4] NCCL INFO Channel 05/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484743:2484903 [2] NCCL INFO Channel 09/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484741:2484901 [0] NCCL INFO Channel 04/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484742:2484907 [1] NCCL INFO Channel 04/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484744:2484906 [3] NCCL INFO Channel 02/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484745:2484902 [4] NCCL INFO Channel 06/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484743:2484903 [2] NCCL INFO Channel 10/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484741:2484901 [0] NCCL INFO Channel 05/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484742:2484907 [1] NCCL INFO Channel 05/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484744:2484906 [3] NCCL INFO Channel 03/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484745:2484902 [4] NCCL INFO Channel 07/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484743:2484903 [2] NCCL INFO Channel 11/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484741:2484901 [0] NCCL INFO Channel 06/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484742:2484907 [1] NCCL INFO Channel 06/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484744:2484906 [3] NCCL INFO Channel 04/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484745:2484902 [4] NCCL INFO Channel 08/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484748:2484905 [7] NCCL INFO Channel 00/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484746:2484904 [5] NCCL INFO Channel 00/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484743:2484903 [2] NCCL INFO Channel 12/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484747:2484908 [6] NCCL INFO Channel 00/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-164-0:2484741:2484901 [0] NCCL INFO Channel 07/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484742:2484907 [1] NCCL INFO Channel 07/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484744:2484906 [3] NCCL INFO Channel 05/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484745:2484902 [4] NCCL INFO Channel 09/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484748:2484905 [7] NCCL INFO Channel 01/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484743:2484903 [2] NCCL INFO Channel 13/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484746:2484904 [5] NCCL INFO Channel 01/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484747:2484908 [6] NCCL INFO Channel 01/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-164-0:2484741:2484901 [0] NCCL INFO Channel 08/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484742:2484907 [1] NCCL INFO Channel 08/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484744:2484906 [3] NCCL INFO Channel 06/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484745:2484902 [4] NCCL INFO Channel 10/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484748:2484905 [7] NCCL INFO Channel 02/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484743:2484903 [2] NCCL INFO Channel 14/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484746:2484904 [5] NCCL INFO Channel 02/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484747:2484908 [6] NCCL INFO Channel 02/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-164-0:2484742:2484907 [1] NCCL INFO Channel 09/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484744:2484906 [3] NCCL INFO Channel 07/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484745:2484902 [4] NCCL INFO Channel 11/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484748:2484905 [7] NCCL INFO Channel 03/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484743:2484903 [2] NCCL INFO Channel 15/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484741:2484901 [0] NCCL INFO Channel 09/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484746:2484904 [5] NCCL INFO Channel 03/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484747:2484908 [6] NCCL INFO Channel 03/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-164-0:2484742:2484907 [1] NCCL INFO Channel 10/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484745:2484902 [4] NCCL INFO Channel 12/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484744:2484906 [3] NCCL INFO Channel 08/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484748:2484905 [7] NCCL INFO Channel 04/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484743:2484903 [2] NCCL INFO Channel 16/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484741:2484901 [0] NCCL INFO Channel 10/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484746:2484904 [5] NCCL INFO Channel 04/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484747:2484908 [6] NCCL INFO Channel 04/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-164-0:2484742:2484907 [1] NCCL INFO Channel 11/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484745:2484902 [4] NCCL INFO Channel 13/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484744:2484906 [3] NCCL INFO Channel 09/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484748:2484905 [7] NCCL INFO Channel 05/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484743:2484903 [2] NCCL INFO Channel 17/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484741:2484901 [0] NCCL INFO Channel 11/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484746:2484904 [5] NCCL INFO Channel 05/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484747:2484908 [6] NCCL INFO Channel 05/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-164-0:2484742:2484907 [1] NCCL INFO Channel 12/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484745:2484902 [4] NCCL INFO Channel 14/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484744:2484906 [3] NCCL INFO Channel 10/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484748:2484905 [7] NCCL INFO Channel 06/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484743:2484903 [2] NCCL INFO Channel 18/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484741:2484901 [0] NCCL INFO Channel 12/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484746:2484904 [5] NCCL INFO Channel 06/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484747:2484908 [6] NCCL INFO Channel 06/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-164-0:2484742:2484907 [1] NCCL INFO Channel 13/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484745:2484902 [4] NCCL INFO Channel 15/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484748:2484905 [7] NCCL INFO Channel 07/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484744:2484906 [3] NCCL INFO Channel 11/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484743:2484903 [2] NCCL INFO Channel 19/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484741:2484901 [0] NCCL INFO Channel 13/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484746:2484904 [5] NCCL INFO Channel 07/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484747:2484908 [6] NCCL INFO Channel 07/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-164-0:2484742:2484907 [1] NCCL INFO Channel 14/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484745:2484902 [4] NCCL INFO Channel 16/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484748:2484905 [7] NCCL INFO Channel 08/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484744:2484906 [3] NCCL INFO Channel 12/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484743:2484903 [2] NCCL INFO Channel 20/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484741:2484901 [0] NCCL INFO Channel 14/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484746:2484904 [5] NCCL INFO Channel 08/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484747:2484908 [6] NCCL INFO Channel 08/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-164-0:2484742:2484907 [1] NCCL INFO Channel 15/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484748:2484905 [7] NCCL INFO Channel 09/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484745:2484902 [4] NCCL INFO Channel 17/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484744:2484906 [3] NCCL INFO Channel 13/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484746:2484904 [5] NCCL INFO Channel 09/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484743:2484903 [2] NCCL INFO Channel 21/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484741:2484901 [0] NCCL INFO Channel 15/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484747:2484908 [6] NCCL INFO Channel 09/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-164-0:2484742:2484907 [1] NCCL INFO Channel 16/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484748:2484905 [7] NCCL INFO Channel 10/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484745:2484902 [4] NCCL INFO Channel 18/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484744:2484906 [3] NCCL INFO Channel 14/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484746:2484904 [5] NCCL INFO Channel 10/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484743:2484903 [2] NCCL INFO Channel 22/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484741:2484901 [0] NCCL INFO Channel 16/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484747:2484908 [6] NCCL INFO Channel 10/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-164-0:2484742:2484907 [1] NCCL INFO Channel 17/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484748:2484905 [7] NCCL INFO Channel 11/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484745:2484902 [4] NCCL INFO Channel 19/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484744:2484906 [3] NCCL INFO Channel 15/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484746:2484904 [5] NCCL INFO Channel 11/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484743:2484903 [2] NCCL INFO Channel 23/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484741:2484901 [0] NCCL INFO Channel 17/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484747:2484908 [6] NCCL INFO Channel 11/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-164-0:2484742:2484907 [1] NCCL INFO Channel 18/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484748:2484905 [7] NCCL INFO Channel 12/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484745:2484902 [4] NCCL INFO Channel 20/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484744:2484906 [3] NCCL INFO Channel 16/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484746:2484904 [5] NCCL INFO Channel 12/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484741:2484901 [0] NCCL INFO Channel 18/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484747:2484908 [6] NCCL INFO Channel 12/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-164-0:2484742:2484907 [1] NCCL INFO Channel 19/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484748:2484905 [7] NCCL INFO Channel 13/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484745:2484902 [4] NCCL INFO Channel 21/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484744:2484906 [3] NCCL INFO Channel 17/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484746:2484904 [5] NCCL INFO Channel 13/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484741:2484901 [0] NCCL INFO Channel 19/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484747:2484908 [6] NCCL INFO Channel 13/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-164-0:2484742:2484907 [1] NCCL INFO Channel 20/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484748:2484905 [7] NCCL INFO Channel 14/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484745:2484902 [4] NCCL INFO Channel 22/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484744:2484906 [3] NCCL INFO Channel 18/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484746:2484904 [5] NCCL INFO Channel 14/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484747:2484908 [6] NCCL INFO Channel 14/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-164-0:2484742:2484907 [1] NCCL INFO Channel 21/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484741:2484901 [0] NCCL INFO Channel 20/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484748:2484905 [7] NCCL INFO Channel 15/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484745:2484902 [4] NCCL INFO Channel 23/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484744:2484906 [3] NCCL INFO Channel 19/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484746:2484904 [5] NCCL INFO Channel 15/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484747:2484908 [6] NCCL INFO Channel 15/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-164-0:2484742:2484907 [1] NCCL INFO Channel 22/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484741:2484901 [0] NCCL INFO Channel 21/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484748:2484905 [7] NCCL INFO Channel 16/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484744:2484906 [3] NCCL INFO Channel 20/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484746:2484904 [5] NCCL INFO Channel 16/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484747:2484908 [6] NCCL INFO Channel 16/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-164-0:2484742:2484907 [1] NCCL INFO Channel 23/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484741:2484901 [0] NCCL INFO Channel 22/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484748:2484905 [7] NCCL INFO Channel 17/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484744:2484906 [3] NCCL INFO Channel 21/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484746:2484904 [5] NCCL INFO Channel 17/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484747:2484908 [6] NCCL INFO Channel 17/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-164-0:2484741:2484901 [0] NCCL INFO Channel 23/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484748:2484905 [7] NCCL INFO Channel 18/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484744:2484906 [3] NCCL INFO Channel 22/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484746:2484904 [5] NCCL INFO Channel 18/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484747:2484908 [6] NCCL INFO Channel 18/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-164-0:2484748:2484905 [7] NCCL INFO Channel 19/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484744:2484906 [3] NCCL INFO Channel 23/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484746:2484904 [5] NCCL INFO Channel 19/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484747:2484908 [6] NCCL INFO Channel 19/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-164-0:2484748:2484905 [7] NCCL INFO Channel 20/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484746:2484904 [5] NCCL INFO Channel 20/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484747:2484908 [6] NCCL INFO Channel 20/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-164-0:2484748:2484905 [7] NCCL INFO Channel 21/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484746:2484904 [5] NCCL INFO Channel 21/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484747:2484908 [6] NCCL INFO Channel 21/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-164-0:2484748:2484905 [7] NCCL INFO Channel 22/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484746:2484904 [5] NCCL INFO Channel 22/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484746:2484904 [5] NCCL INFO Channel 23/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484747:2484908 [6] NCCL INFO Channel 22/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-164-0:2484748:2484905 [7] NCCL INFO Channel 23/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484747:2484908 [6] NCCL INFO Channel 23/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-164-0:2484744:2484906 [3] NCCL INFO Connected all rings
+ip-26-0-164-0:2484742:2484907 [1] NCCL INFO Connected all rings
+ip-26-0-164-0:2484743:2484903 [2] NCCL INFO Connected all rings
+ip-26-0-164-0:2484745:2484902 [4] NCCL INFO Connected all rings
+ip-26-0-164-0:2484747:2484908 [6] NCCL INFO Connected all rings
+ip-26-0-164-0:2484746:2484904 [5] NCCL INFO Connected all rings
+ip-26-0-164-0:2484748:2484905 [7] NCCL INFO Connected all rings
+ip-26-0-164-0:2484748:2484905 [7] NCCL INFO Channel 00/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484748:2484905 [7] NCCL INFO Channel 01/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484741:2484901 [0] NCCL INFO Connected all rings
+ip-26-0-164-0:2484748:2484905 [7] NCCL INFO Channel 02/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484748:2484905 [7] NCCL INFO Channel 03/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484748:2484905 [7] NCCL INFO Channel 04/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484748:2484905 [7] NCCL INFO Channel 05/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484748:2484905 [7] NCCL INFO Channel 06/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484748:2484905 [7] NCCL INFO Channel 07/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484748:2484905 [7] NCCL INFO Channel 08/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484748:2484905 [7] NCCL INFO Channel 09/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484748:2484905 [7] NCCL INFO Channel 10/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484748:2484905 [7] NCCL INFO Channel 11/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484748:2484905 [7] NCCL INFO Channel 12/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484748:2484905 [7] NCCL INFO Channel 13/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484742:2484907 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484748:2484905 [7] NCCL INFO Channel 14/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484742:2484907 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484748:2484905 [7] NCCL INFO Channel 15/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484742:2484907 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484748:2484905 [7] NCCL INFO Channel 16/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484742:2484907 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484747:2484908 [6] NCCL INFO Channel 00/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484745:2484902 [4] NCCL INFO Channel 00/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484742:2484907 [1] NCCL INFO Channel 04/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484745:2484902 [4] NCCL INFO Channel 01/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484747:2484908 [6] NCCL INFO Channel 01/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484742:2484907 [1] NCCL INFO Channel 05/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484745:2484902 [4] NCCL INFO Channel 02/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484747:2484908 [6] NCCL INFO Channel 02/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484742:2484907 [1] NCCL INFO Channel 06/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484745:2484902 [4] NCCL INFO Channel 03/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484747:2484908 [6] NCCL INFO Channel 03/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484742:2484907 [1] NCCL INFO Channel 07/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484745:2484902 [4] NCCL INFO Channel 04/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484747:2484908 [6] NCCL INFO Channel 04/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484742:2484907 [1] NCCL INFO Channel 08/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484745:2484902 [4] NCCL INFO Channel 05/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484747:2484908 [6] NCCL INFO Channel 05/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484742:2484907 [1] NCCL INFO Channel 09/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484744:2484906 [3] NCCL INFO Channel 00/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484745:2484902 [4] NCCL INFO Channel 06/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484747:2484908 [6] NCCL INFO Channel 06/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484742:2484907 [1] NCCL INFO Channel 10/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484745:2484902 [4] NCCL INFO Channel 07/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484744:2484906 [3] NCCL INFO Channel 01/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484747:2484908 [6] NCCL INFO Channel 07/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484742:2484907 [1] NCCL INFO Channel 11/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484745:2484902 [4] NCCL INFO Channel 08/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484744:2484906 [3] NCCL INFO Channel 02/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484747:2484908 [6] NCCL INFO Channel 08/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484742:2484907 [1] NCCL INFO Channel 12/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484746:2484904 [5] NCCL INFO Channel 00/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484748:2484905 [7] NCCL INFO Channel 17/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484745:2484902 [4] NCCL INFO Channel 09/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484744:2484906 [3] NCCL INFO Channel 03/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484747:2484908 [6] NCCL INFO Channel 09/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484742:2484907 [1] NCCL INFO Channel 13/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484746:2484904 [5] NCCL INFO Channel 01/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484748:2484905 [7] NCCL INFO Channel 18/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484745:2484902 [4] NCCL INFO Channel 10/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484744:2484906 [3] NCCL INFO Channel 04/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484747:2484908 [6] NCCL INFO Channel 10/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484742:2484907 [1] NCCL INFO Channel 14/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484746:2484904 [5] NCCL INFO Channel 02/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484748:2484905 [7] NCCL INFO Channel 19/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484745:2484902 [4] NCCL INFO Channel 11/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484744:2484906 [3] NCCL INFO Channel 05/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484747:2484908 [6] NCCL INFO Channel 11/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484743:2484903 [2] NCCL INFO Channel 00/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484742:2484907 [1] NCCL INFO Channel 15/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484746:2484904 [5] NCCL INFO Channel 03/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484748:2484905 [7] NCCL INFO Channel 20/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484745:2484902 [4] NCCL INFO Channel 12/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484744:2484906 [3] NCCL INFO Channel 06/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484747:2484908 [6] NCCL INFO Channel 12/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484743:2484903 [2] NCCL INFO Channel 01/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484742:2484907 [1] NCCL INFO Channel 16/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484746:2484904 [5] NCCL INFO Channel 04/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484748:2484905 [7] NCCL INFO Channel 21/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484745:2484902 [4] NCCL INFO Channel 13/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484744:2484906 [3] NCCL INFO Channel 07/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484747:2484908 [6] NCCL INFO Channel 13/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484743:2484903 [2] NCCL INFO Channel 02/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484742:2484907 [1] NCCL INFO Channel 17/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484746:2484904 [5] NCCL INFO Channel 05/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484748:2484905 [7] NCCL INFO Channel 22/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484745:2484902 [4] NCCL INFO Channel 14/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484744:2484906 [3] NCCL INFO Channel 08/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484747:2484908 [6] NCCL INFO Channel 14/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484743:2484903 [2] NCCL INFO Channel 03/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484742:2484907 [1] NCCL INFO Channel 18/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484746:2484904 [5] NCCL INFO Channel 06/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484748:2484905 [7] NCCL INFO Channel 23/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484745:2484902 [4] NCCL INFO Channel 15/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484744:2484906 [3] NCCL INFO Channel 09/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484747:2484908 [6] NCCL INFO Channel 15/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484743:2484903 [2] NCCL INFO Channel 04/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484742:2484907 [1] NCCL INFO Channel 19/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484746:2484904 [5] NCCL INFO Channel 07/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484745:2484902 [4] NCCL INFO Channel 16/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484744:2484906 [3] NCCL INFO Channel 10/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484747:2484908 [6] NCCL INFO Channel 16/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484743:2484903 [2] NCCL INFO Channel 05/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484742:2484907 [1] NCCL INFO Channel 20/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484746:2484904 [5] NCCL INFO Channel 08/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484745:2484902 [4] NCCL INFO Channel 17/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484744:2484906 [3] NCCL INFO Channel 11/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484747:2484908 [6] NCCL INFO Channel 17/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484743:2484903 [2] NCCL INFO Channel 06/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484742:2484907 [1] NCCL INFO Channel 21/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484746:2484904 [5] NCCL INFO Channel 09/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484745:2484902 [4] NCCL INFO Channel 18/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484744:2484906 [3] NCCL INFO Channel 12/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484747:2484908 [6] NCCL INFO Channel 18/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484743:2484903 [2] NCCL INFO Channel 07/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484742:2484907 [1] NCCL INFO Channel 22/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484746:2484904 [5] NCCL INFO Channel 10/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484745:2484902 [4] NCCL INFO Channel 19/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484744:2484906 [3] NCCL INFO Channel 13/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484746:2484904 [5] NCCL INFO Channel 11/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484744:2484906 [3] NCCL INFO Channel 14/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484745:2484902 [4] NCCL INFO Channel 20/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484747:2484908 [6] NCCL INFO Channel 19/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484746:2484904 [5] NCCL INFO Channel 12/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484743:2484903 [2] NCCL INFO Channel 08/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484745:2484902 [4] NCCL INFO Channel 21/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484744:2484906 [3] NCCL INFO Channel 15/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484747:2484908 [6] NCCL INFO Channel 20/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484746:2484904 [5] NCCL INFO Channel 13/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484743:2484903 [2] NCCL INFO Channel 09/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484745:2484902 [4] NCCL INFO Channel 22/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484744:2484906 [3] NCCL INFO Channel 16/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484747:2484908 [6] NCCL INFO Channel 21/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484746:2484904 [5] NCCL INFO Channel 14/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484743:2484903 [2] NCCL INFO Channel 10/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484745:2484902 [4] NCCL INFO Channel 23/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484744:2484906 [3] NCCL INFO Channel 17/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484747:2484908 [6] NCCL INFO Channel 22/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484746:2484904 [5] NCCL INFO Channel 15/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484743:2484903 [2] NCCL INFO Channel 11/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484744:2484906 [3] NCCL INFO Channel 18/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484747:2484908 [6] NCCL INFO Channel 23/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484746:2484904 [5] NCCL INFO Channel 16/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484743:2484903 [2] NCCL INFO Channel 12/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484744:2484906 [3] NCCL INFO Channel 19/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484746:2484904 [5] NCCL INFO Channel 17/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484743:2484903 [2] NCCL INFO Channel 13/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484742:2484907 [1] NCCL INFO Channel 23/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484744:2484906 [3] NCCL INFO Channel 20/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484746:2484904 [5] NCCL INFO Channel 18/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484743:2484903 [2] NCCL INFO Channel 14/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484744:2484906 [3] NCCL INFO Channel 21/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484743:2484903 [2] NCCL INFO Channel 15/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484746:2484904 [5] NCCL INFO Channel 19/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484744:2484906 [3] NCCL INFO Channel 22/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484743:2484903 [2] NCCL INFO Channel 16/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484746:2484904 [5] NCCL INFO Channel 20/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484744:2484906 [3] NCCL INFO Channel 23/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484743:2484903 [2] NCCL INFO Channel 17/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484746:2484904 [5] NCCL INFO Channel 21/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484743:2484903 [2] NCCL INFO Channel 18/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484746:2484904 [5] NCCL INFO Channel 22/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484746:2484904 [5] NCCL INFO Channel 23/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484743:2484903 [2] NCCL INFO Channel 19/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484743:2484903 [2] NCCL INFO Channel 20/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484743:2484903 [2] NCCL INFO Channel 21/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484743:2484903 [2] NCCL INFO Channel 22/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484743:2484903 [2] NCCL INFO Channel 23/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484741:2484901 [0] NCCL INFO Connected all trees
+ip-26-0-164-0:2484741:2484901 [0] NCCL INFO NVLS comm 0x844f620 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-164-0:2484748:2484905 [7] NCCL INFO Connected all trees
+ip-26-0-164-0:2484748:2484905 [7] NCCL INFO NVLS comm 0x842d180 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-164-0:2484746:2484904 [5] NCCL INFO Connected all trees
+ip-26-0-164-0:2484746:2484904 [5] NCCL INFO NVLS comm 0x86ef420 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-164-0:2484745:2484902 [4] NCCL INFO Connected all trees
+ip-26-0-164-0:2484745:2484902 [4] NCCL INFO NVLS comm 0x956ef50 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-164-0:2484747:2484908 [6] NCCL INFO Connected all trees
+ip-26-0-164-0:2484747:2484908 [6] NCCL INFO NVLS comm 0x9177540 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-164-0:2484742:2484907 [1] NCCL INFO Connected all trees
+ip-26-0-164-0:2484742:2484907 [1] NCCL INFO NVLS comm 0x98ddfe0 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-164-0:2484744:2484906 [3] NCCL INFO Connected all trees
+ip-26-0-164-0:2484744:2484906 [3] NCCL INFO NVLS comm 0x9e84410 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-164-0:2484743:2484903 [2] NCCL INFO Connected all trees
+ip-26-0-164-0:2484743:2484903 [2] NCCL INFO NVLS comm 0x88fbde0 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-164-0:2484746:2484904 [5] NCCL INFO Connected NVLS tree
+ip-26-0-164-0:2484746:2484904 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-164-0:2484746:2484904 [5] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-164-0:2484746:2484904 [5] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-164-0:2484748:2484905 [7] NCCL INFO Connected NVLS tree
+ip-26-0-164-0:2484748:2484905 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-164-0:2484748:2484905 [7] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-164-0:2484748:2484905 [7] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-164-0:2484742:2484907 [1] NCCL INFO Connected NVLS tree
+ip-26-0-164-0:2484742:2484907 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-164-0:2484741:2484901 [0] NCCL INFO Connected NVLS tree
+ip-26-0-164-0:2484742:2484907 [1] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-164-0:2484742:2484907 [1] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-164-0:2484741:2484901 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-164-0:2484743:2484903 [2] NCCL INFO Connected NVLS tree
+ip-26-0-164-0:2484743:2484903 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-164-0:2484743:2484903 [2] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-164-0:2484743:2484903 [2] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-164-0:2484741:2484901 [0] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-164-0:2484741:2484901 [0] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-164-0:2484745:2484902 [4] NCCL INFO Connected NVLS tree
+ip-26-0-164-0:2484745:2484902 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-164-0:2484745:2484902 [4] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-164-0:2484745:2484902 [4] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-164-0:2484747:2484908 [6] NCCL INFO Connected NVLS tree
+ip-26-0-164-0:2484747:2484908 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-164-0:2484747:2484908 [6] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-164-0:2484747:2484908 [6] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-164-0:2484744:2484906 [3] NCCL INFO Connected NVLS tree
+ip-26-0-164-0:2484744:2484906 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-164-0:2484744:2484906 [3] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-164-0:2484744:2484906 [3] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-164-0:2484742:2484907 [1] NCCL INFO comm 0x98ddfe0 rank 1 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0xb21dbb55538c0b55 - Init COMPLETE
+ip-26-0-164-0:2484745:2484902 [4] NCCL INFO comm 0x956ef50 rank 4 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0xb21dbb55538c0b55 - Init COMPLETE
+ip-26-0-164-0:2484747:2484908 [6] NCCL INFO comm 0x9177540 rank 6 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0xb21dbb55538c0b55 - Init COMPLETE
+ip-26-0-164-0:2484748:2484905 [7] NCCL INFO comm 0x842d180 rank 7 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0xb21dbb55538c0b55 - Init COMPLETE
+ip-26-0-164-0:2484744:2484906 [3] NCCL INFO comm 0x9e84410 rank 3 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0xb21dbb55538c0b55 - Init COMPLETE
+ip-26-0-164-0:2484743:2484903 [2] NCCL INFO comm 0x88fbde0 rank 2 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0xb21dbb55538c0b55 - Init COMPLETE
+ip-26-0-164-0:2484741:2484901 [0] NCCL INFO comm 0x844f620 rank 0 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0xb21dbb55538c0b55 - Init COMPLETE
+ip-26-0-164-0:2484746:2484904 [5] NCCL INFO comm 0x86ef420 rank 5 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0xb21dbb55538c0b55 - Init COMPLETE
+ip-26-0-164-0:2484741:2485069 [0] NCCL INFO Using network Libfabric
+ip-26-0-164-0:2484747:2485070 [6] NCCL INFO Using network Libfabric
+ip-26-0-164-0:2484745:2485072 [4] NCCL INFO Using network Libfabric
+ip-26-0-164-0:2484746:2485071 [5] NCCL INFO Using network Libfabric
+ip-26-0-164-0:2484743:2485076 [2] NCCL INFO Using network Libfabric
+ip-26-0-164-0:2484744:2485074 [3] NCCL INFO Using network Libfabric
+ip-26-0-164-0:2484748:2485073 [7] NCCL INFO Using network Libfabric
+ip-26-0-164-0:2484742:2485075 [1] NCCL INFO Using network Libfabric
+ip-26-0-164-0:2484741:2485069 [0] NCCL INFO comm 0x867f250 rank 0 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0x217e915e5aab251b - Init START
+ip-26-0-164-0:2484742:2485075 [1] NCCL INFO comm 0x9b0c2e0 rank 1 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0x217e915e5aab251b - Init START
+ip-26-0-164-0:2484748:2485073 [7] NCCL INFO comm 0x865a640 rank 7 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0x217e915e5aab251b - Init START
+ip-26-0-164-0:2484746:2485071 [5] NCCL INFO comm 0x891d160 rank 5 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0x217e915e5aab251b - Init START
+ip-26-0-164-0:2484743:2485076 [2] NCCL INFO comm 0x8b2a280 rank 2 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0x217e915e5aab251b - Init START
+ip-26-0-164-0:2484747:2485070 [6] NCCL INFO comm 0x93a5100 rank 6 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0x217e915e5aab251b - Init START
+ip-26-0-164-0:2484745:2485072 [4] NCCL INFO comm 0x979d740 rank 4 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0x217e915e5aab251b - Init START
+ip-26-0-164-0:2484744:2485074 [3] NCCL INFO comm 0xa0b27e0 rank 3 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0x217e915e5aab251b - Init START
+ip-26-0-164-0:2484748:2485073 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/168
+ip-26-0-164-0:2484747:2485070 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/168
+ip-26-0-164-0:2484742:2485075 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/168
+ip-26-0-164-0:2484743:2485076 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/168
+ip-26-0-164-0:2484741:2485069 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/181
+ip-26-0-164-0:2484745:2485072 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/168
+ip-26-0-164-0:2484746:2485071 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/168
+ip-26-0-164-0:2484744:2485074 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/168
+ip-26-0-164-0:2484741:2485069 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-164-0:2484741:2485069 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-164-0:2484743:2485076 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-164-0:2484743:2485076 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-164-0:2484746:2485071 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-164-0:2484746:2485071 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-164-0:2484745:2485072 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-164-0:2484745:2485072 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-164-0:2484742:2485075 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-164-0:2484742:2485075 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-164-0:2484748:2485073 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-164-0:2484748:2485073 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-164-0:2484744:2485074 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-164-0:2484744:2485074 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-164-0:2484747:2485070 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-164-0:2484747:2485070 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-164-0:2484747:2485070 [6] NCCL INFO Trees [0] 7/-1/-1->6->5 [1] 7/-1/-1->6->5 [2] 7/-1/-1->6->5 [3] 7/-1/-1->6->5 [4] 7/-1/-1->6->5 [5] 7/-1/-1->6->5 [6] 7/-1/-1->6->5 [7] 7/-1/-1->6->5 [8] 7/-1/-1->6->5 [9] 7/-1/-1->6->5 [10] 7/-1/-1->6->5 [11] 7/-1/-1->6->5 [12] 7/-1/-1->6->5 [13] 7/-1/-1->6->5 [14] 7/-1/-1->6->5 [15] 7/-1/-1->6->5 [16] 7/-1/-1->6->5 [17] 7/-1/-1->6->5 [18] 7/-1/-1->6->5 [19] 7/-1/-1->6->5 [20] 7/-1/-1->6->5 [21] 7/-1/-1->6->5 [22] 7/-1/-1->6->5 [23] 7/-1/-1->6->5
+ip-26-0-164-0:2484747:2485070 [6] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-164-0:2484748:2485073 [7] NCCL INFO Trees [0] -1/-1/-1->7->6 [1] -1/-1/-1->7->6 [2] -1/-1/-1->7->6 [3] -1/-1/-1->7->6 [4] -1/-1/-1->7->6 [5] -1/-1/-1->7->6 [6] -1/-1/-1->7->6 [7] -1/-1/-1->7->6 [8] -1/-1/-1->7->6 [9] -1/-1/-1->7->6 [10] -1/-1/-1->7->6 [11] -1/-1/-1->7->6 [12] -1/-1/-1->7->6 [13] -1/-1/-1->7->6 [14] -1/-1/-1->7->6 [15] -1/-1/-1->7->6 [16] -1/-1/-1->7->6 [17] -1/-1/-1->7->6 [18] -1/-1/-1->7->6 [19] -1/-1/-1->7->6 [20] -1/-1/-1->7->6 [21] -1/-1/-1->7->6 [22] -1/-1/-1->7->6 [23] -1/-1/-1->7->6
+ip-26-0-164-0:2484748:2485073 [7] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-164-0:2484746:2485071 [5] NCCL INFO Trees [0] 6/-1/-1->5->4 [1] 6/-1/-1->5->4 [2] 6/-1/-1->5->4 [3] 6/-1/-1->5->4 [4] 6/-1/-1->5->4 [5] 6/-1/-1->5->4 [6] 6/-1/-1->5->4 [7] 6/-1/-1->5->4 [8] 6/-1/-1->5->4 [9] 6/-1/-1->5->4 [10] 6/-1/-1->5->4 [11] 6/-1/-1->5->4 [12] 6/-1/-1->5->4 [13] 6/-1/-1->5->4 [14] 6/-1/-1->5->4 [15] 6/-1/-1->5->4 [16] 6/-1/-1->5->4 [17] 6/-1/-1->5->4 [18] 6/-1/-1->5->4 [19] 6/-1/-1->5->4 [20] 6/-1/-1->5->4 [21] 6/-1/-1->5->4 [22] 6/-1/-1->5->4 [23] 6/-1/-1->5->4
+ip-26-0-164-0:2484746:2485071 [5] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-164-0:2484745:2485072 [4] NCCL INFO Trees [0] 5/-1/-1->4->3 [1] 5/-1/-1->4->3 [2] 5/-1/-1->4->3 [3] 5/-1/-1->4->3 [4] 5/-1/-1->4->3 [5] 5/-1/-1->4->3 [6] 5/-1/-1->4->3 [7] 5/-1/-1->4->3 [8] 5/-1/-1->4->3 [9] 5/-1/-1->4->3 [10] 5/-1/-1->4->3 [11] 5/-1/-1->4->3 [12] 5/-1/-1->4->3 [13] 5/-1/-1->4->3 [14] 5/-1/-1->4->3 [15] 5/-1/-1->4->3 [16] 5/-1/-1->4->3 [17] 5/-1/-1->4->3 [18] 5/-1/-1->4->3 [19] 5/-1/-1->4->3 [20] 5/-1/-1->4->3 [21] 5/-1/-1->4->3 [22] 5/-1/-1->4->3 [23] 5/-1/-1->4->3
+ip-26-0-164-0:2484745:2485072 [4] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-164-0:2484743:2485076 [2] NCCL INFO Trees [0] 3/-1/-1->2->1 [1] 3/-1/-1->2->1 [2] 3/-1/-1->2->1 [3] 3/-1/-1->2->1 [4] 3/-1/-1->2->1 [5] 3/-1/-1->2->1 [6] 3/-1/-1->2->1 [7] 3/-1/-1->2->1 [8] 3/-1/-1->2->1 [9] 3/-1/-1->2->1 [10] 3/-1/-1->2->1 [11] 3/-1/-1->2->1 [12] 3/-1/-1->2->1 [13] 3/-1/-1->2->1 [14] 3/-1/-1->2->1 [15] 3/-1/-1->2->1 [16] 3/-1/-1->2->1 [17] 3/-1/-1->2->1 [18] 3/-1/-1->2->1 [19] 3/-1/-1->2->1 [20] 3/-1/-1->2->1 [21] 3/-1/-1->2->1 [22] 3/-1/-1->2->1 [23] 3/-1/-1->2->1
+ip-26-0-164-0:2484744:2485074 [3] NCCL INFO Trees [0] 4/-1/-1->3->2 [1] 4/-1/-1->3->2 [2] 4/-1/-1->3->2 [3] 4/-1/-1->3->2 [4] 4/-1/-1->3->2 [5] 4/-1/-1->3->2 [6] 4/-1/-1->3->2 [7] 4/-1/-1->3->2 [8] 4/-1/-1->3->2 [9] 4/-1/-1->3->2 [10] 4/-1/-1->3->2 [11] 4/-1/-1->3->2 [12] 4/-1/-1->3->2 [13] 4/-1/-1->3->2 [14] 4/-1/-1->3->2 [15] 4/-1/-1->3->2 [16] 4/-1/-1->3->2 [17] 4/-1/-1->3->2 [18] 4/-1/-1->3->2 [19] 4/-1/-1->3->2 [20] 4/-1/-1->3->2 [21] 4/-1/-1->3->2 [22] 4/-1/-1->3->2 [23] 4/-1/-1->3->2
+ip-26-0-164-0:2484742:2485075 [1] NCCL INFO Trees [0] 2/-1/-1->1->0 [1] 2/-1/-1->1->0 [2] 2/-1/-1->1->0 [3] 2/-1/-1->1->0 [4] 2/-1/-1->1->0 [5] 2/-1/-1->1->0 [6] 2/-1/-1->1->0 [7] 2/-1/-1->1->0 [8] 2/-1/-1->1->0 [9] 2/-1/-1->1->0 [10] 2/-1/-1->1->0 [11] 2/-1/-1->1->0 [12] 2/-1/-1->1->0 [13] 2/-1/-1->1->0 [14] 2/-1/-1->1->0 [15] 2/-1/-1->1->0 [16] 2/-1/-1->1->0 [17] 2/-1/-1->1->0 [18] 2/-1/-1->1->0 [19] 2/-1/-1->1->0 [20] 2/-1/-1->1->0 [21] 2/-1/-1->1->0 [22] 2/-1/-1->1->0 [23] 2/-1/-1->1->0
+ip-26-0-164-0:2484743:2485076 [2] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-164-0:2484744:2485074 [3] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-164-0:2484742:2485075 [1] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-164-0:2484741:2485069 [0] NCCL INFO Channel 00/24 :    0   1   2   3   4   5   6   7
+ip-26-0-164-0:2484741:2485069 [0] NCCL INFO Channel 01/24 :    0   1   2   3   4   5   6   7
+ip-26-0-164-0:2484741:2485069 [0] NCCL INFO Channel 02/24 :    0   1   2   3   4   5   6   7
+ip-26-0-164-0:2484741:2485069 [0] NCCL INFO Channel 03/24 :    0   1   2   3   4   5   6   7
+ip-26-0-164-0:2484741:2485069 [0] NCCL INFO Channel 04/24 :    0   1   2   3   4   5   6   7
+ip-26-0-164-0:2484741:2485069 [0] NCCL INFO Channel 05/24 :    0   1   2   3   4   5   6   7
+ip-26-0-164-0:2484741:2485069 [0] NCCL INFO Channel 06/24 :    0   1   2   3   4   5   6   7
+ip-26-0-164-0:2484741:2485069 [0] NCCL INFO Channel 07/24 :    0   1   2   3   4   5   6   7
+ip-26-0-164-0:2484741:2485069 [0] NCCL INFO Channel 08/24 :    0   1   2   3   4   5   6   7
+ip-26-0-164-0:2484741:2485069 [0] NCCL INFO Channel 09/24 :    0   1   2   3   4   5   6   7
+ip-26-0-164-0:2484741:2485069 [0] NCCL INFO Channel 10/24 :    0   1   2   3   4   5   6   7
+ip-26-0-164-0:2484741:2485069 [0] NCCL INFO Channel 11/24 :    0   1   2   3   4   5   6   7
+ip-26-0-164-0:2484741:2485069 [0] NCCL INFO Channel 12/24 :    0   1   2   3   4   5   6   7
+ip-26-0-164-0:2484741:2485069 [0] NCCL INFO Channel 13/24 :    0   1   2   3   4   5   6   7
+ip-26-0-164-0:2484741:2485069 [0] NCCL INFO Channel 14/24 :    0   1   2   3   4   5   6   7
+ip-26-0-164-0:2484741:2485069 [0] NCCL INFO Channel 15/24 :    0   1   2   3   4   5   6   7
+ip-26-0-164-0:2484741:2485069 [0] NCCL INFO Channel 16/24 :    0   1   2   3   4   5   6   7
+ip-26-0-164-0:2484741:2485069 [0] NCCL INFO Channel 17/24 :    0   1   2   3   4   5   6   7
+ip-26-0-164-0:2484741:2485069 [0] NCCL INFO Channel 18/24 :    0   1   2   3   4   5   6   7
+ip-26-0-164-0:2484741:2485069 [0] NCCL INFO Channel 19/24 :    0   1   2   3   4   5   6   7
+ip-26-0-164-0:2484741:2485069 [0] NCCL INFO Channel 20/24 :    0   1   2   3   4   5   6   7
+ip-26-0-164-0:2484741:2485069 [0] NCCL INFO Channel 21/24 :    0   1   2   3   4   5   6   7
+ip-26-0-164-0:2484741:2485069 [0] NCCL INFO Channel 22/24 :    0   1   2   3   4   5   6   7
+ip-26-0-164-0:2484741:2485069 [0] NCCL INFO Channel 23/24 :    0   1   2   3   4   5   6   7
+ip-26-0-164-0:2484741:2485069 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] 1/-1/-1->0->-1 [7] 1/-1/-1->0->-1 [8] 1/-1/-1->0->-1 [9] 1/-1/-1->0->-1 [10] 1/-1/-1->0->-1 [11] 1/-1/-1->0->-1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] 1/-1/-1->0->-1 [19] 1/-1/-1->0->-1 [20] 1/-1/-1->0->-1 [21] 1/-1/-1->0->-1 [22] 1/-1/-1->0->-1 [23] 1/-1/-1->0->-1
+ip-26-0-164-0:2484741:2485069 [0] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-164-0:2484744:2485074 [3] NCCL INFO Channel 00/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484741:2485069 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484745:2485072 [4] NCCL INFO Channel 00/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484744:2485074 [3] NCCL INFO Channel 01/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484748:2485073 [7] NCCL INFO Channel 00/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484741:2485069 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484745:2485072 [4] NCCL INFO Channel 01/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484744:2485074 [3] NCCL INFO Channel 02/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484748:2485073 [7] NCCL INFO Channel 01/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484741:2485069 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484743:2485076 [2] NCCL INFO Channel 00/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484745:2485072 [4] NCCL INFO Channel 02/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484742:2485075 [1] NCCL INFO Channel 00/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484744:2485074 [3] NCCL INFO Channel 03/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484748:2485073 [7] NCCL INFO Channel 02/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484741:2485069 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484743:2485076 [2] NCCL INFO Channel 01/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484745:2485072 [4] NCCL INFO Channel 03/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484746:2485071 [5] NCCL INFO Channel 00/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484742:2485075 [1] NCCL INFO Channel 01/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484744:2485074 [3] NCCL INFO Channel 04/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484748:2485073 [7] NCCL INFO Channel 03/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484741:2485069 [0] NCCL INFO Channel 04/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484743:2485076 [2] NCCL INFO Channel 02/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484745:2485072 [4] NCCL INFO Channel 04/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484747:2485070 [6] NCCL INFO Channel 00/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-164-0:2484746:2485071 [5] NCCL INFO Channel 01/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484742:2485075 [1] NCCL INFO Channel 02/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484744:2485074 [3] NCCL INFO Channel 05/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484748:2485073 [7] NCCL INFO Channel 04/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484741:2485069 [0] NCCL INFO Channel 05/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484743:2485076 [2] NCCL INFO Channel 03/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484745:2485072 [4] NCCL INFO Channel 05/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484747:2485070 [6] NCCL INFO Channel 01/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-164-0:2484746:2485071 [5] NCCL INFO Channel 02/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484742:2485075 [1] NCCL INFO Channel 03/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484745:2485072 [4] NCCL INFO Channel 06/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484742:2485075 [1] NCCL INFO Channel 04/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484741:2485069 [0] NCCL INFO Channel 06/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484745:2485072 [4] NCCL INFO Channel 07/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484742:2485075 [1] NCCL INFO Channel 05/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484741:2485069 [0] NCCL INFO Channel 07/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484748:2485073 [7] NCCL INFO Channel 05/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484745:2485072 [4] NCCL INFO Channel 08/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484742:2485075 [1] NCCL INFO Channel 06/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484741:2485069 [0] NCCL INFO Channel 08/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484748:2485073 [7] NCCL INFO Channel 06/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484744:2485074 [3] NCCL INFO Channel 06/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484745:2485072 [4] NCCL INFO Channel 09/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484747:2485070 [6] NCCL INFO Channel 02/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-164-0:2484742:2485075 [1] NCCL INFO Channel 07/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484746:2485071 [5] NCCL INFO Channel 03/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484748:2485073 [7] NCCL INFO Channel 07/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484741:2485069 [0] NCCL INFO Channel 09/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484744:2485074 [3] NCCL INFO Channel 07/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484745:2485072 [4] NCCL INFO Channel 10/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484747:2485070 [6] NCCL INFO Channel 03/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-164-0:2484742:2485075 [1] NCCL INFO Channel 08/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484746:2485071 [5] NCCL INFO Channel 04/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484741:2485069 [0] NCCL INFO Channel 10/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484748:2485073 [7] NCCL INFO Channel 08/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484744:2485074 [3] NCCL INFO Channel 08/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484745:2485072 [4] NCCL INFO Channel 11/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484747:2485070 [6] NCCL INFO Channel 04/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-164-0:2484746:2485071 [5] NCCL INFO Channel 05/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484741:2485069 [0] NCCL INFO Channel 11/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484748:2485073 [7] NCCL INFO Channel 09/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484744:2485074 [3] NCCL INFO Channel 09/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484743:2485076 [2] NCCL INFO Channel 04/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484745:2485072 [4] NCCL INFO Channel 12/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484747:2485070 [6] NCCL INFO Channel 05/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-164-0:2484742:2485075 [1] NCCL INFO Channel 09/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484746:2485071 [5] NCCL INFO Channel 06/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484741:2485069 [0] NCCL INFO Channel 12/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484748:2485073 [7] NCCL INFO Channel 10/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484744:2485074 [3] NCCL INFO Channel 10/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484743:2485076 [2] NCCL INFO Channel 05/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484745:2485072 [4] NCCL INFO Channel 13/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484742:2485075 [1] NCCL INFO Channel 10/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484747:2485070 [6] NCCL INFO Channel 06/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-164-0:2484746:2485071 [5] NCCL INFO Channel 07/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484741:2485069 [0] NCCL INFO Channel 13/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484748:2485073 [7] NCCL INFO Channel 11/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484743:2485076 [2] NCCL INFO Channel 06/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484744:2485074 [3] NCCL INFO Channel 11/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484745:2485072 [4] NCCL INFO Channel 14/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484747:2485070 [6] NCCL INFO Channel 07/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-164-0:2484742:2485075 [1] NCCL INFO Channel 11/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484741:2485069 [0] NCCL INFO Channel 14/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484746:2485071 [5] NCCL INFO Channel 08/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484743:2485076 [2] NCCL INFO Channel 07/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484748:2485073 [7] NCCL INFO Channel 12/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484744:2485074 [3] NCCL INFO Channel 12/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484747:2485070 [6] NCCL INFO Channel 08/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-164-0:2484742:2485075 [1] NCCL INFO Channel 12/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484745:2485072 [4] NCCL INFO Channel 15/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484741:2485069 [0] NCCL INFO Channel 15/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484746:2485071 [5] NCCL INFO Channel 09/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484743:2485076 [2] NCCL INFO Channel 08/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484748:2485073 [7] NCCL INFO Channel 13/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484744:2485074 [3] NCCL INFO Channel 13/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484747:2485070 [6] NCCL INFO Channel 09/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-164-0:2484742:2485075 [1] NCCL INFO Channel 13/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484745:2485072 [4] NCCL INFO Channel 16/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484741:2485069 [0] NCCL INFO Channel 16/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484746:2485071 [5] NCCL INFO Channel 10/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484743:2485076 [2] NCCL INFO Channel 09/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484748:2485073 [7] NCCL INFO Channel 14/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484744:2485074 [3] NCCL INFO Channel 14/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484747:2485070 [6] NCCL INFO Channel 10/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-164-0:2484742:2485075 [1] NCCL INFO Channel 14/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484745:2485072 [4] NCCL INFO Channel 17/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484741:2485069 [0] NCCL INFO Channel 17/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484743:2485076 [2] NCCL INFO Channel 10/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484746:2485071 [5] NCCL INFO Channel 11/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484748:2485073 [7] NCCL INFO Channel 15/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484747:2485070 [6] NCCL INFO Channel 11/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-164-0:2484744:2485074 [3] NCCL INFO Channel 15/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484742:2485075 [1] NCCL INFO Channel 15/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484745:2485072 [4] NCCL INFO Channel 18/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484741:2485069 [0] NCCL INFO Channel 18/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484743:2485076 [2] NCCL INFO Channel 11/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484746:2485071 [5] NCCL INFO Channel 12/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484748:2485073 [7] NCCL INFO Channel 16/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484747:2485070 [6] NCCL INFO Channel 12/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-164-0:2484744:2485074 [3] NCCL INFO Channel 16/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484742:2485075 [1] NCCL INFO Channel 16/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484745:2485072 [4] NCCL INFO Channel 19/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484741:2485069 [0] NCCL INFO Channel 19/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484744:2485074 [3] NCCL INFO Channel 17/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484743:2485076 [2] NCCL INFO Channel 12/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484746:2485071 [5] NCCL INFO Channel 13/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484748:2485073 [7] NCCL INFO Channel 17/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484747:2485070 [6] NCCL INFO Channel 13/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-164-0:2484742:2485075 [1] NCCL INFO Channel 17/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484745:2485072 [4] NCCL INFO Channel 20/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484741:2485069 [0] NCCL INFO Channel 20/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484744:2485074 [3] NCCL INFO Channel 18/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484743:2485076 [2] NCCL INFO Channel 13/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484746:2485071 [5] NCCL INFO Channel 14/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484747:2485070 [6] NCCL INFO Channel 14/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-164-0:2484748:2485073 [7] NCCL INFO Channel 18/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484742:2485075 [1] NCCL INFO Channel 18/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484745:2485072 [4] NCCL INFO Channel 21/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484741:2485069 [0] NCCL INFO Channel 21/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484744:2485074 [3] NCCL INFO Channel 19/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484743:2485076 [2] NCCL INFO Channel 14/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484746:2485071 [5] NCCL INFO Channel 15/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484747:2485070 [6] NCCL INFO Channel 15/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-164-0:2484748:2485073 [7] NCCL INFO Channel 19/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484745:2485072 [4] NCCL INFO Channel 22/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484742:2485075 [1] NCCL INFO Channel 19/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484741:2485069 [0] NCCL INFO Channel 22/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484744:2485074 [3] NCCL INFO Channel 20/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484743:2485076 [2] NCCL INFO Channel 15/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484747:2485070 [6] NCCL INFO Channel 16/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-164-0:2484746:2485071 [5] NCCL INFO Channel 16/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484748:2485073 [7] NCCL INFO Channel 20/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484745:2485072 [4] NCCL INFO Channel 23/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484742:2485075 [1] NCCL INFO Channel 20/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484741:2485069 [0] NCCL INFO Channel 23/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484744:2485074 [3] NCCL INFO Channel 21/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484743:2485076 [2] NCCL INFO Channel 16/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484747:2485070 [6] NCCL INFO Channel 17/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-164-0:2484746:2485071 [5] NCCL INFO Channel 17/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484748:2485073 [7] NCCL INFO Channel 21/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484742:2485075 [1] NCCL INFO Channel 21/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484744:2485074 [3] NCCL INFO Channel 22/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484743:2485076 [2] NCCL INFO Channel 17/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484747:2485070 [6] NCCL INFO Channel 18/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-164-0:2484746:2485071 [5] NCCL INFO Channel 18/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484748:2485073 [7] NCCL INFO Channel 22/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484742:2485075 [1] NCCL INFO Channel 22/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484743:2485076 [2] NCCL INFO Channel 18/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484744:2485074 [3] NCCL INFO Channel 23/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484747:2485070 [6] NCCL INFO Channel 19/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-164-0:2484746:2485071 [5] NCCL INFO Channel 19/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484748:2485073 [7] NCCL INFO Channel 23/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484742:2485075 [1] NCCL INFO Channel 23/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484743:2485076 [2] NCCL INFO Channel 19/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484747:2485070 [6] NCCL INFO Channel 20/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-164-0:2484746:2485071 [5] NCCL INFO Channel 20/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484743:2485076 [2] NCCL INFO Channel 20/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484746:2485071 [5] NCCL INFO Channel 21/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484747:2485070 [6] NCCL INFO Channel 21/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-164-0:2484743:2485076 [2] NCCL INFO Channel 21/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484746:2485071 [5] NCCL INFO Channel 22/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484747:2485070 [6] NCCL INFO Channel 22/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-164-0:2484743:2485076 [2] NCCL INFO Channel 22/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484746:2485071 [5] NCCL INFO Channel 23/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484747:2485070 [6] NCCL INFO Channel 23/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-164-0:2484743:2485076 [2] NCCL INFO Channel 23/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484741:2485069 [0] NCCL INFO Connected all rings
+ip-26-0-164-0:2484742:2485075 [1] NCCL INFO Connected all rings
+ip-26-0-164-0:2484747:2485070 [6] NCCL INFO Connected all rings
+ip-26-0-164-0:2484746:2485071 [5] NCCL INFO Connected all rings
+ip-26-0-164-0:2484743:2485076 [2] NCCL INFO Connected all rings
+ip-26-0-164-0:2484744:2485074 [3] NCCL INFO Connected all rings
+ip-26-0-164-0:2484745:2485072 [4] NCCL INFO Connected all rings
+ip-26-0-164-0:2484748:2485073 [7] NCCL INFO Connected all rings
+ip-26-0-164-0:2484748:2485073 [7] NCCL INFO Channel 00/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484748:2485073 [7] NCCL INFO Channel 01/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484748:2485073 [7] NCCL INFO Channel 02/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484748:2485073 [7] NCCL INFO Channel 03/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484748:2485073 [7] NCCL INFO Channel 04/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484748:2485073 [7] NCCL INFO Channel 05/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484748:2485073 [7] NCCL INFO Channel 06/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484748:2485073 [7] NCCL INFO Channel 07/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484748:2485073 [7] NCCL INFO Channel 08/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484748:2485073 [7] NCCL INFO Channel 09/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484748:2485073 [7] NCCL INFO Channel 10/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484748:2485073 [7] NCCL INFO Channel 11/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484748:2485073 [7] NCCL INFO Channel 12/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484748:2485073 [7] NCCL INFO Channel 13/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484748:2485073 [7] NCCL INFO Channel 14/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484748:2485073 [7] NCCL INFO Channel 15/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484748:2485073 [7] NCCL INFO Channel 16/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484748:2485073 [7] NCCL INFO Channel 17/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484748:2485073 [7] NCCL INFO Channel 18/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484742:2485075 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484742:2485075 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484742:2485075 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484742:2485075 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484746:2485071 [5] NCCL INFO Channel 00/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484748:2485073 [7] NCCL INFO Channel 19/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484746:2485071 [5] NCCL INFO Channel 01/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484748:2485073 [7] NCCL INFO Channel 20/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484746:2485071 [5] NCCL INFO Channel 02/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484748:2485073 [7] NCCL INFO Channel 21/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484743:2485076 [2] NCCL INFO Channel 00/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484746:2485071 [5] NCCL INFO Channel 03/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484748:2485073 [7] NCCL INFO Channel 22/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484743:2485076 [2] NCCL INFO Channel 01/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484746:2485071 [5] NCCL INFO Channel 04/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484748:2485073 [7] NCCL INFO Channel 23/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484743:2485076 [2] NCCL INFO Channel 02/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484746:2485071 [5] NCCL INFO Channel 05/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484747:2485070 [6] NCCL INFO Channel 00/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484742:2485075 [1] NCCL INFO Channel 04/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484743:2485076 [2] NCCL INFO Channel 03/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484746:2485071 [5] NCCL INFO Channel 06/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484747:2485070 [6] NCCL INFO Channel 01/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484742:2485075 [1] NCCL INFO Channel 05/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484743:2485076 [2] NCCL INFO Channel 04/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484746:2485071 [5] NCCL INFO Channel 07/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484747:2485070 [6] NCCL INFO Channel 02/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484742:2485075 [1] NCCL INFO Channel 06/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484745:2485072 [4] NCCL INFO Channel 00/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484743:2485076 [2] NCCL INFO Channel 05/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484746:2485071 [5] NCCL INFO Channel 08/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484744:2485074 [3] NCCL INFO Channel 00/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484747:2485070 [6] NCCL INFO Channel 03/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484742:2485075 [1] NCCL INFO Channel 07/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484745:2485072 [4] NCCL INFO Channel 01/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484743:2485076 [2] NCCL INFO Channel 06/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484746:2485071 [5] NCCL INFO Channel 09/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484744:2485074 [3] NCCL INFO Channel 01/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484747:2485070 [6] NCCL INFO Channel 04/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484742:2485075 [1] NCCL INFO Channel 08/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484745:2485072 [4] NCCL INFO Channel 02/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484743:2485076 [2] NCCL INFO Channel 07/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484746:2485071 [5] NCCL INFO Channel 10/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484744:2485074 [3] NCCL INFO Channel 02/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484747:2485070 [6] NCCL INFO Channel 05/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484742:2485075 [1] NCCL INFO Channel 09/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484745:2485072 [4] NCCL INFO Channel 03/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484743:2485076 [2] NCCL INFO Channel 08/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484746:2485071 [5] NCCL INFO Channel 11/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484744:2485074 [3] NCCL INFO Channel 03/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484747:2485070 [6] NCCL INFO Channel 06/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484742:2485075 [1] NCCL INFO Channel 10/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484743:2485076 [2] NCCL INFO Channel 09/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484746:2485071 [5] NCCL INFO Channel 12/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484745:2485072 [4] NCCL INFO Channel 04/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484744:2485074 [3] NCCL INFO Channel 04/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484747:2485070 [6] NCCL INFO Channel 07/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484742:2485075 [1] NCCL INFO Channel 11/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484743:2485076 [2] NCCL INFO Channel 10/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484746:2485071 [5] NCCL INFO Channel 13/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484745:2485072 [4] NCCL INFO Channel 05/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484744:2485074 [3] NCCL INFO Channel 05/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484747:2485070 [6] NCCL INFO Channel 08/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484742:2485075 [1] NCCL INFO Channel 12/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484743:2485076 [2] NCCL INFO Channel 11/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484746:2485071 [5] NCCL INFO Channel 14/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484745:2485072 [4] NCCL INFO Channel 06/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484747:2485070 [6] NCCL INFO Channel 09/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484744:2485074 [3] NCCL INFO Channel 06/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484742:2485075 [1] NCCL INFO Channel 13/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484743:2485076 [2] NCCL INFO Channel 12/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484746:2485071 [5] NCCL INFO Channel 15/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484745:2485072 [4] NCCL INFO Channel 07/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484747:2485070 [6] NCCL INFO Channel 10/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484744:2485074 [3] NCCL INFO Channel 07/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484742:2485075 [1] NCCL INFO Channel 14/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484743:2485076 [2] NCCL INFO Channel 13/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484746:2485071 [5] NCCL INFO Channel 16/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484745:2485072 [4] NCCL INFO Channel 08/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484747:2485070 [6] NCCL INFO Channel 11/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484744:2485074 [3] NCCL INFO Channel 08/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484742:2485075 [1] NCCL INFO Channel 15/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484744:2485074 [3] NCCL INFO Channel 09/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484746:2485071 [5] NCCL INFO Channel 17/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484743:2485076 [2] NCCL INFO Channel 14/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484742:2485075 [1] NCCL INFO Channel 16/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484744:2485074 [3] NCCL INFO Channel 10/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484745:2485072 [4] NCCL INFO Channel 09/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484747:2485070 [6] NCCL INFO Channel 12/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484742:2485075 [1] NCCL INFO Channel 17/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484745:2485072 [4] NCCL INFO Channel 10/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484744:2485074 [3] NCCL INFO Channel 11/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484747:2485070 [6] NCCL INFO Channel 13/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484742:2485075 [1] NCCL INFO Channel 18/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484745:2485072 [4] NCCL INFO Channel 11/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484744:2485074 [3] NCCL INFO Channel 12/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484747:2485070 [6] NCCL INFO Channel 14/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484742:2485075 [1] NCCL INFO Channel 19/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484745:2485072 [4] NCCL INFO Channel 12/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484744:2485074 [3] NCCL INFO Channel 13/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484747:2485070 [6] NCCL INFO Channel 15/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484742:2485075 [1] NCCL INFO Channel 20/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484745:2485072 [4] NCCL INFO Channel 13/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484744:2485074 [3] NCCL INFO Channel 14/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484747:2485070 [6] NCCL INFO Channel 16/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484742:2485075 [1] NCCL INFO Channel 21/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484746:2485071 [5] NCCL INFO Channel 18/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484743:2485076 [2] NCCL INFO Channel 15/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484745:2485072 [4] NCCL INFO Channel 14/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484744:2485074 [3] NCCL INFO Channel 15/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484747:2485070 [6] NCCL INFO Channel 17/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484742:2485075 [1] NCCL INFO Channel 22/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484746:2485071 [5] NCCL INFO Channel 19/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484743:2485076 [2] NCCL INFO Channel 16/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484745:2485072 [4] NCCL INFO Channel 15/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484744:2485074 [3] NCCL INFO Channel 16/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484747:2485070 [6] NCCL INFO Channel 18/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484742:2485075 [1] NCCL INFO Channel 23/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484746:2485071 [5] NCCL INFO Channel 20/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484743:2485076 [2] NCCL INFO Channel 17/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484745:2485072 [4] NCCL INFO Channel 16/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484744:2485074 [3] NCCL INFO Channel 17/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484747:2485070 [6] NCCL INFO Channel 19/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484746:2485071 [5] NCCL INFO Channel 21/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484743:2485076 [2] NCCL INFO Channel 18/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484745:2485072 [4] NCCL INFO Channel 17/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484744:2485074 [3] NCCL INFO Channel 18/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484747:2485070 [6] NCCL INFO Channel 20/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484746:2485071 [5] NCCL INFO Channel 22/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484743:2485076 [2] NCCL INFO Channel 19/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484745:2485072 [4] NCCL INFO Channel 18/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484744:2485074 [3] NCCL INFO Channel 19/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484747:2485070 [6] NCCL INFO Channel 21/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484745:2485072 [4] NCCL INFO Channel 19/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484747:2485070 [6] NCCL INFO Channel 22/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484745:2485072 [4] NCCL INFO Channel 20/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484747:2485070 [6] NCCL INFO Channel 23/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484745:2485072 [4] NCCL INFO Channel 21/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484743:2485076 [2] NCCL INFO Channel 20/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484746:2485071 [5] NCCL INFO Channel 23/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484744:2485074 [3] NCCL INFO Channel 20/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484743:2485076 [2] NCCL INFO Channel 21/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484745:2485072 [4] NCCL INFO Channel 22/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484743:2485076 [2] NCCL INFO Channel 22/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484745:2485072 [4] NCCL INFO Channel 23/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484744:2485074 [3] NCCL INFO Channel 21/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484743:2485076 [2] NCCL INFO Channel 23/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484744:2485074 [3] NCCL INFO Channel 22/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484744:2485074 [3] NCCL INFO Channel 23/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484741:2485069 [0] NCCL INFO Connected all trees
+ip-26-0-164-0:2484741:2485069 [0] NCCL INFO NVLS comm 0x867f250 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-164-0:2484748:2485073 [7] NCCL INFO Connected all trees
+ip-26-0-164-0:2484748:2485073 [7] NCCL INFO NVLS comm 0x865a640 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-164-0:2484744:2485074 [3] NCCL INFO Connected all trees
+ip-26-0-164-0:2484744:2485074 [3] NCCL INFO NVLS comm 0xa0b27e0 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-164-0:2484746:2485071 [5] NCCL INFO Connected all trees
+ip-26-0-164-0:2484746:2485071 [5] NCCL INFO NVLS comm 0x891d160 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-164-0:2484742:2485075 [1] NCCL INFO Connected all trees
+ip-26-0-164-0:2484742:2485075 [1] NCCL INFO NVLS comm 0x9b0c2e0 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-164-0:2484743:2485076 [2] NCCL INFO Connected all trees
+ip-26-0-164-0:2484743:2485076 [2] NCCL INFO NVLS comm 0x8b2a280 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-164-0:2484747:2485070 [6] NCCL INFO Connected all trees
+ip-26-0-164-0:2484747:2485070 [6] NCCL INFO NVLS comm 0x93a5100 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-164-0:2484745:2485072 [4] NCCL INFO Connected all trees
+ip-26-0-164-0:2484745:2485072 [4] NCCL INFO NVLS comm 0x979d740 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-164-0:2484746:2485071 [5] NCCL INFO Connected NVLS tree
+ip-26-0-164-0:2484746:2485071 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-164-0:2484746:2485071 [5] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-164-0:2484746:2485071 [5] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-164-0:2484748:2485073 [7] NCCL INFO Connected NVLS tree
+ip-26-0-164-0:2484748:2485073 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-164-0:2484748:2485073 [7] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-164-0:2484748:2485073 [7] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-164-0:2484742:2485075 [1] NCCL INFO Connected NVLS tree
+ip-26-0-164-0:2484745:2485072 [4] NCCL INFO Connected NVLS tree
+ip-26-0-164-0:2484744:2485074 [3] NCCL INFO Connected NVLS tree
+ip-26-0-164-0:2484742:2485075 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-164-0:2484745:2485072 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-164-0:2484742:2485075 [1] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-164-0:2484742:2485075 [1] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-164-0:2484745:2485072 [4] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-164-0:2484745:2485072 [4] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-164-0:2484744:2485074 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-164-0:2484744:2485074 [3] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-164-0:2484744:2485074 [3] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-164-0:2484743:2485076 [2] NCCL INFO Connected NVLS tree
+ip-26-0-164-0:2484743:2485076 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-164-0:2484743:2485076 [2] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-164-0:2484743:2485076 [2] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-164-0:2484747:2485070 [6] NCCL INFO Connected NVLS tree
+ip-26-0-164-0:2484747:2485070 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-164-0:2484747:2485070 [6] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-164-0:2484747:2485070 [6] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-164-0:2484741:2485069 [0] NCCL INFO Connected NVLS tree
+ip-26-0-164-0:2484741:2485069 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-164-0:2484741:2485069 [0] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-164-0:2484741:2485069 [0] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-164-0:2484748:2485073 [7] NCCL INFO comm 0x865a640 rank 7 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0x217e915e5aab251b - Init COMPLETE
+ip-26-0-164-0:2484746:2485071 [5] NCCL INFO comm 0x891d160 rank 5 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0x217e915e5aab251b - Init COMPLETE
+ip-26-0-164-0:2484747:2485070 [6] NCCL INFO comm 0x93a5100 rank 6 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0x217e915e5aab251b - Init COMPLETE
+ip-26-0-164-0:2484744:2485074 [3] NCCL INFO comm 0xa0b27e0 rank 3 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0x217e915e5aab251b - Init COMPLETE
+ip-26-0-164-0:2484743:2485076 [2] NCCL INFO comm 0x8b2a280 rank 2 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0x217e915e5aab251b - Init COMPLETE
+ip-26-0-164-0:2484742:2485075 [1] NCCL INFO comm 0x9b0c2e0 rank 1 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0x217e915e5aab251b - Init COMPLETE
+ip-26-0-164-0:2484741:2485069 [0] NCCL INFO comm 0x867f250 rank 0 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0x217e915e5aab251b - Init COMPLETE
+ip-26-0-164-0:2484745:2485072 [4] NCCL INFO comm 0x979d740 rank 4 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0x217e915e5aab251b - Init COMPLETE
+ip-26-0-164-0:2484747:2485104 [6] NCCL INFO Using network Libfabric
+ip-26-0-164-0:2484741:2485103 [0] NCCL INFO Using network Libfabric
+ip-26-0-164-0:2484746:2485106 [5] NCCL INFO Using network Libfabric
+ip-26-0-164-0:2484745:2485108 [4] NCCL INFO Using network Libfabric
+ip-26-0-164-0:2484748:2485105 [7] NCCL INFO Using network Libfabric
+ip-26-0-164-0:2484743:2485109 [2] NCCL INFO Using network Libfabric
+ip-26-0-164-0:2484742:2485110 [1] NCCL INFO Using network Libfabric
+ip-26-0-164-0:2484744:2485107 [3] NCCL INFO Using network Libfabric
+ip-26-0-164-0:2484742:2485110 [1] NCCL INFO comm 0x9b210d0 rank 1 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0x48681016f9ce6d74 - Init START
+ip-26-0-164-0:2484748:2485105 [7] NCCL INFO comm 0x866f6d0 rank 7 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0x48681016f9ce6d74 - Init START
+ip-26-0-164-0:2484741:2485103 [0] NCCL INFO comm 0x8694f40 rank 0 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0x48681016f9ce6d74 - Init START
+ip-26-0-164-0:2484747:2485104 [6] NCCL INFO comm 0x93ba060 rank 6 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0x48681016f9ce6d74 - Init START
+ip-26-0-164-0:2484746:2485106 [5] NCCL INFO comm 0x89321e0 rank 5 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0x48681016f9ce6d74 - Init START
+ip-26-0-164-0:2484743:2485109 [2] NCCL INFO comm 0x8b3ee10 rank 2 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0x48681016f9ce6d74 - Init START
+ip-26-0-164-0:2484745:2485108 [4] NCCL INFO comm 0x97b2900 rank 4 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0x48681016f9ce6d74 - Init START
+ip-26-0-164-0:2484744:2485107 [3] NCCL INFO comm 0xa0c7a40 rank 3 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0x48681016f9ce6d74 - Init START
+ip-26-0-164-0:2484742:2485110 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/168
+ip-26-0-164-0:2484748:2485105 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/168
+ip-26-0-164-0:2484747:2485104 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/168
+ip-26-0-164-0:2484741:2485103 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/181
+ip-26-0-164-0:2484744:2485107 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/168
+ip-26-0-164-0:2484745:2485108 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/168
+ip-26-0-164-0:2484743:2485109 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/168
+ip-26-0-164-0:2484746:2485106 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/168
+ip-26-0-164-0:2484744:2485107 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-164-0:2484744:2485107 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-164-0:2484741:2485103 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-164-0:2484741:2485103 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-164-0:2484748:2485105 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-164-0:2484748:2485105 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-164-0:2484743:2485109 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-164-0:2484743:2485109 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-164-0:2484745:2485108 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-164-0:2484745:2485108 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-164-0:2484747:2485104 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-164-0:2484747:2485104 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-164-0:2484746:2485106 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-164-0:2484746:2485106 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-164-0:2484742:2485110 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-164-0:2484742:2485110 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-164-0:2484742:2485110 [1] NCCL INFO Trees [0] 2/-1/-1->1->0 [1] 2/-1/-1->1->0 [2] 2/-1/-1->1->0 [3] 2/-1/-1->1->0 [4] 2/-1/-1->1->0 [5] 2/-1/-1->1->0 [6] 2/-1/-1->1->0 [7] 2/-1/-1->1->0 [8] 2/-1/-1->1->0 [9] 2/-1/-1->1->0 [10] 2/-1/-1->1->0 [11] 2/-1/-1->1->0 [12] 2/-1/-1->1->0 [13] 2/-1/-1->1->0 [14] 2/-1/-1->1->0 [15] 2/-1/-1->1->0 [16] 2/-1/-1->1->0 [17] 2/-1/-1->1->0 [18] 2/-1/-1->1->0 [19] 2/-1/-1->1->0 [20] 2/-1/-1->1->0 [21] 2/-1/-1->1->0 [22] 2/-1/-1->1->0 [23] 2/-1/-1->1->0
+ip-26-0-164-0:2484742:2485110 [1] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-164-0:2484743:2485109 [2] NCCL INFO Trees [0] 3/-1/-1->2->1 [1] 3/-1/-1->2->1 [2] 3/-1/-1->2->1 [3] 3/-1/-1->2->1 [4] 3/-1/-1->2->1 [5] 3/-1/-1->2->1 [6] 3/-1/-1->2->1 [7] 3/-1/-1->2->1 [8] 3/-1/-1->2->1 [9] 3/-1/-1->2->1 [10] 3/-1/-1->2->1 [11] 3/-1/-1->2->1 [12] 3/-1/-1->2->1 [13] 3/-1/-1->2->1 [14] 3/-1/-1->2->1 [15] 3/-1/-1->2->1 [16] 3/-1/-1->2->1 [17] 3/-1/-1->2->1 [18] 3/-1/-1->2->1 [19] 3/-1/-1->2->1 [20] 3/-1/-1->2->1 [21] 3/-1/-1->2->1 [22] 3/-1/-1->2->1 [23] 3/-1/-1->2->1
+ip-26-0-164-0:2484743:2485109 [2] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-164-0:2484748:2485105 [7] NCCL INFO Trees [0] -1/-1/-1->7->6 [1] -1/-1/-1->7->6 [2] -1/-1/-1->7->6 [3] -1/-1/-1->7->6 [4] -1/-1/-1->7->6 [5] -1/-1/-1->7->6 [6] -1/-1/-1->7->6 [7] -1/-1/-1->7->6 [8] -1/-1/-1->7->6 [9] -1/-1/-1->7->6 [10] -1/-1/-1->7->6 [11] -1/-1/-1->7->6 [12] -1/-1/-1->7->6 [13] -1/-1/-1->7->6 [14] -1/-1/-1->7->6 [15] -1/-1/-1->7->6 [16] -1/-1/-1->7->6 [17] -1/-1/-1->7->6 [18] -1/-1/-1->7->6 [19] -1/-1/-1->7->6 [20] -1/-1/-1->7->6 [21] -1/-1/-1->7->6 [22] -1/-1/-1->7->6 [23] -1/-1/-1->7->6
+ip-26-0-164-0:2484748:2485105 [7] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-164-0:2484741:2485103 [0] NCCL INFO Channel 00/24 :    0   1   2   3   4   5   6   7
+ip-26-0-164-0:2484747:2485104 [6] NCCL INFO Trees [0] 7/-1/-1->6->5 [1] 7/-1/-1->6->5 [2] 7/-1/-1->6->5 [3] 7/-1/-1->6->5 [4] 7/-1/-1->6->5 [5] 7/-1/-1->6->5 [6] 7/-1/-1->6->5 [7] 7/-1/-1->6->5 [8] 7/-1/-1->6->5 [9] 7/-1/-1->6->5 [10] 7/-1/-1->6->5 [11] 7/-1/-1->6->5 [12] 7/-1/-1->6->5 [13] 7/-1/-1->6->5 [14] 7/-1/-1->6->5 [15] 7/-1/-1->6->5 [16] 7/-1/-1->6->5 [17] 7/-1/-1->6->5 [18] 7/-1/-1->6->5 [19] 7/-1/-1->6->5 [20] 7/-1/-1->6->5 [21] 7/-1/-1->6->5 [22] 7/-1/-1->6->5 [23] 7/-1/-1->6->5
+ip-26-0-164-0:2484741:2485103 [0] NCCL INFO Channel 01/24 :    0   1   2   3   4   5   6   7
+ip-26-0-164-0:2484745:2485108 [4] NCCL INFO Trees [0] 5/-1/-1->4->3 [1] 5/-1/-1->4->3 [2] 5/-1/-1->4->3 [3] 5/-1/-1->4->3 [4] 5/-1/-1->4->3 [5] 5/-1/-1->4->3 [6] 5/-1/-1->4->3 [7] 5/-1/-1->4->3 [8] 5/-1/-1->4->3 [9] 5/-1/-1->4->3 [10] 5/-1/-1->4->3 [11] 5/-1/-1->4->3 [12] 5/-1/-1->4->3 [13] 5/-1/-1->4->3 [14] 5/-1/-1->4->3 [15] 5/-1/-1->4->3 [16] 5/-1/-1->4->3 [17] 5/-1/-1->4->3 [18] 5/-1/-1->4->3 [19] 5/-1/-1->4->3 [20] 5/-1/-1->4->3 [21] 5/-1/-1->4->3 [22] 5/-1/-1->4->3 [23] 5/-1/-1->4->3
+ip-26-0-164-0:2484746:2485106 [5] NCCL INFO Trees [0] 6/-1/-1->5->4 [1] 6/-1/-1->5->4 [2] 6/-1/-1->5->4 [3] 6/-1/-1->5->4 [4] 6/-1/-1->5->4 [5] 6/-1/-1->5->4 [6] 6/-1/-1->5->4 [7] 6/-1/-1->5->4 [8] 6/-1/-1->5->4 [9] 6/-1/-1->5->4 [10] 6/-1/-1->5->4 [11] 6/-1/-1->5->4 [12] 6/-1/-1->5->4 [13] 6/-1/-1->5->4 [14] 6/-1/-1->5->4 [15] 6/-1/-1->5->4 [16] 6/-1/-1->5->4 [17] 6/-1/-1->5->4 [18] 6/-1/-1->5->4 [19] 6/-1/-1->5->4 [20] 6/-1/-1->5->4 [21] 6/-1/-1->5->4 [22] 6/-1/-1->5->4 [23] 6/-1/-1->5->4
+ip-26-0-164-0:2484747:2485104 [6] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-164-0:2484741:2485103 [0] NCCL INFO Channel 02/24 :    0   1   2   3   4   5   6   7
+ip-26-0-164-0:2484745:2485108 [4] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-164-0:2484746:2485106 [5] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-164-0:2484741:2485103 [0] NCCL INFO Channel 03/24 :    0   1   2   3   4   5   6   7
+ip-26-0-164-0:2484741:2485103 [0] NCCL INFO Channel 04/24 :    0   1   2   3   4   5   6   7
+ip-26-0-164-0:2484741:2485103 [0] NCCL INFO Channel 05/24 :    0   1   2   3   4   5   6   7
+ip-26-0-164-0:2484741:2485103 [0] NCCL INFO Channel 06/24 :    0   1   2   3   4   5   6   7
+ip-26-0-164-0:2484741:2485103 [0] NCCL INFO Channel 07/24 :    0   1   2   3   4   5   6   7
+ip-26-0-164-0:2484741:2485103 [0] NCCL INFO Channel 08/24 :    0   1   2   3   4   5   6   7
+ip-26-0-164-0:2484744:2485107 [3] NCCL INFO Trees [0] 4/-1/-1->3->2 [1] 4/-1/-1->3->2 [2] 4/-1/-1->3->2 [3] 4/-1/-1->3->2 [4] 4/-1/-1->3->2 [5] 4/-1/-1->3->2 [6] 4/-1/-1->3->2 [7] 4/-1/-1->3->2 [8] 4/-1/-1->3->2 [9] 4/-1/-1->3->2 [10] 4/-1/-1->3->2 [11] 4/-1/-1->3->2 [12] 4/-1/-1->3->2 [13] 4/-1/-1->3->2 [14] 4/-1/-1->3->2 [15] 4/-1/-1->3->2 [16] 4/-1/-1->3->2 [17] 4/-1/-1->3->2 [18] 4/-1/-1->3->2 [19] 4/-1/-1->3->2 [20] 4/-1/-1->3->2 [21] 4/-1/-1->3->2 [22] 4/-1/-1->3->2 [23] 4/-1/-1->3->2
+ip-26-0-164-0:2484741:2485103 [0] NCCL INFO Channel 09/24 :    0   1   2   3   4   5   6   7
+ip-26-0-164-0:2484741:2485103 [0] NCCL INFO Channel 10/24 :    0   1   2   3   4   5   6   7
+ip-26-0-164-0:2484741:2485103 [0] NCCL INFO Channel 11/24 :    0   1   2   3   4   5   6   7
+ip-26-0-164-0:2484741:2485103 [0] NCCL INFO Channel 12/24 :    0   1   2   3   4   5   6   7
+ip-26-0-164-0:2484744:2485107 [3] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-164-0:2484741:2485103 [0] NCCL INFO Channel 13/24 :    0   1   2   3   4   5   6   7
+ip-26-0-164-0:2484741:2485103 [0] NCCL INFO Channel 14/24 :    0   1   2   3   4   5   6   7
+ip-26-0-164-0:2484741:2485103 [0] NCCL INFO Channel 15/24 :    0   1   2   3   4   5   6   7
+ip-26-0-164-0:2484741:2485103 [0] NCCL INFO Channel 16/24 :    0   1   2   3   4   5   6   7
+ip-26-0-164-0:2484741:2485103 [0] NCCL INFO Channel 17/24 :    0   1   2   3   4   5   6   7
+ip-26-0-164-0:2484741:2485103 [0] NCCL INFO Channel 18/24 :    0   1   2   3   4   5   6   7
+ip-26-0-164-0:2484741:2485103 [0] NCCL INFO Channel 19/24 :    0   1   2   3   4   5   6   7
+ip-26-0-164-0:2484741:2485103 [0] NCCL INFO Channel 20/24 :    0   1   2   3   4   5   6   7
+ip-26-0-164-0:2484741:2485103 [0] NCCL INFO Channel 21/24 :    0   1   2   3   4   5   6   7
+ip-26-0-164-0:2484741:2485103 [0] NCCL INFO Channel 22/24 :    0   1   2   3   4   5   6   7
+ip-26-0-164-0:2484741:2485103 [0] NCCL INFO Channel 23/24 :    0   1   2   3   4   5   6   7
+ip-26-0-164-0:2484741:2485103 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] 1/-1/-1->0->-1 [7] 1/-1/-1->0->-1 [8] 1/-1/-1->0->-1 [9] 1/-1/-1->0->-1 [10] 1/-1/-1->0->-1 [11] 1/-1/-1->0->-1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] 1/-1/-1->0->-1 [19] 1/-1/-1->0->-1 [20] 1/-1/-1->0->-1 [21] 1/-1/-1->0->-1 [22] 1/-1/-1->0->-1 [23] 1/-1/-1->0->-1
+ip-26-0-164-0:2484741:2485103 [0] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-164-0:2484746:2485106 [5] NCCL INFO Channel 00/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484746:2485106 [5] NCCL INFO Channel 01/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484746:2485106 [5] NCCL INFO Channel 02/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484746:2485106 [5] NCCL INFO Channel 03/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484741:2485103 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484746:2485106 [5] NCCL INFO Channel 04/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484741:2485103 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484746:2485106 [5] NCCL INFO Channel 05/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484743:2485109 [2] NCCL INFO Channel 00/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484741:2485103 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484748:2485105 [7] NCCL INFO Channel 00/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484745:2485108 [4] NCCL INFO Channel 00/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484747:2485104 [6] NCCL INFO Channel 00/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-164-0:2484746:2485106 [5] NCCL INFO Channel 06/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484742:2485110 [1] NCCL INFO Channel 00/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484744:2485107 [3] NCCL INFO Channel 00/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484743:2485109 [2] NCCL INFO Channel 01/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484741:2485103 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484748:2485105 [7] NCCL INFO Channel 01/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484745:2485108 [4] NCCL INFO Channel 01/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484747:2485104 [6] NCCL INFO Channel 01/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-164-0:2484746:2485106 [5] NCCL INFO Channel 07/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484742:2485110 [1] NCCL INFO Channel 01/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484743:2485109 [2] NCCL INFO Channel 02/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484741:2485103 [0] NCCL INFO Channel 04/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484744:2485107 [3] NCCL INFO Channel 01/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484748:2485105 [7] NCCL INFO Channel 02/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484745:2485108 [4] NCCL INFO Channel 02/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484747:2485104 [6] NCCL INFO Channel 02/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-164-0:2484742:2485110 [1] NCCL INFO Channel 02/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484746:2485106 [5] NCCL INFO Channel 08/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484743:2485109 [2] NCCL INFO Channel 03/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484747:2485104 [6] NCCL INFO Channel 03/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-164-0:2484742:2485110 [1] NCCL INFO Channel 03/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484746:2485106 [5] NCCL INFO Channel 09/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484743:2485109 [2] NCCL INFO Channel 04/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484747:2485104 [6] NCCL INFO Channel 04/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-164-0:2484742:2485110 [1] NCCL INFO Channel 04/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484746:2485106 [5] NCCL INFO Channel 10/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484743:2485109 [2] NCCL INFO Channel 05/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484747:2485104 [6] NCCL INFO Channel 05/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-164-0:2484742:2485110 [1] NCCL INFO Channel 05/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484746:2485106 [5] NCCL INFO Channel 11/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484743:2485109 [2] NCCL INFO Channel 06/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484747:2485104 [6] NCCL INFO Channel 06/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-164-0:2484741:2485103 [0] NCCL INFO Channel 05/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484747:2485104 [6] NCCL INFO Channel 07/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-164-0:2484748:2485105 [7] NCCL INFO Channel 03/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484741:2485103 [0] NCCL INFO Channel 06/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484745:2485108 [4] NCCL INFO Channel 03/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484747:2485104 [6] NCCL INFO Channel 08/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-164-0:2484748:2485105 [7] NCCL INFO Channel 04/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484745:2485108 [4] NCCL INFO Channel 04/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484744:2485107 [3] NCCL INFO Channel 02/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484746:2485106 [5] NCCL INFO Channel 12/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484743:2485109 [2] NCCL INFO Channel 07/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484745:2485108 [4] NCCL INFO Channel 05/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484744:2485107 [3] NCCL INFO Channel 03/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484743:2485109 [2] NCCL INFO Channel 08/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484746:2485106 [5] NCCL INFO Channel 13/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484745:2485108 [4] NCCL INFO Channel 06/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484744:2485107 [3] NCCL INFO Channel 04/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484743:2485109 [2] NCCL INFO Channel 09/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484746:2485106 [5] NCCL INFO Channel 14/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484745:2485108 [4] NCCL INFO Channel 07/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484744:2485107 [3] NCCL INFO Channel 05/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484743:2485109 [2] NCCL INFO Channel 10/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484746:2485106 [5] NCCL INFO Channel 15/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484742:2485110 [1] NCCL INFO Channel 06/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484745:2485108 [4] NCCL INFO Channel 08/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484748:2485105 [7] NCCL INFO Channel 05/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484741:2485103 [0] NCCL INFO Channel 07/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484742:2485110 [1] NCCL INFO Channel 07/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484745:2485108 [4] NCCL INFO Channel 09/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484747:2485104 [6] NCCL INFO Channel 09/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-164-0:2484746:2485106 [5] NCCL INFO Channel 16/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484741:2485103 [0] NCCL INFO Channel 08/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484748:2485105 [7] NCCL INFO Channel 06/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484742:2485110 [1] NCCL INFO Channel 08/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484745:2485108 [4] NCCL INFO Channel 10/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484747:2485104 [6] NCCL INFO Channel 10/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-164-0:2484746:2485106 [5] NCCL INFO Channel 17/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484741:2485103 [0] NCCL INFO Channel 09/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484748:2485105 [7] NCCL INFO Channel 07/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484742:2485110 [1] NCCL INFO Channel 09/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484745:2485108 [4] NCCL INFO Channel 11/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484747:2485104 [6] NCCL INFO Channel 11/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-164-0:2484741:2485103 [0] NCCL INFO Channel 10/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484746:2485106 [5] NCCL INFO Channel 18/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484748:2485105 [7] NCCL INFO Channel 08/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484742:2485110 [1] NCCL INFO Channel 10/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484747:2485104 [6] NCCL INFO Channel 12/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-164-0:2484743:2485109 [2] NCCL INFO Channel 11/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484745:2485108 [4] NCCL INFO Channel 12/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484746:2485106 [5] NCCL INFO Channel 19/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484741:2485103 [0] NCCL INFO Channel 11/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484748:2485105 [7] NCCL INFO Channel 09/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484747:2485104 [6] NCCL INFO Channel 13/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-164-0:2484743:2485109 [2] NCCL INFO Channel 12/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484745:2485108 [4] NCCL INFO Channel 13/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484746:2485106 [5] NCCL INFO Channel 20/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484741:2485103 [0] NCCL INFO Channel 12/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484748:2485105 [7] NCCL INFO Channel 10/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484747:2485104 [6] NCCL INFO Channel 14/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-164-0:2484743:2485109 [2] NCCL INFO Channel 13/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484746:2485106 [5] NCCL INFO Channel 21/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484745:2485108 [4] NCCL INFO Channel 14/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484741:2485103 [0] NCCL INFO Channel 13/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484748:2485105 [7] NCCL INFO Channel 11/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484747:2485104 [6] NCCL INFO Channel 15/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-164-0:2484743:2485109 [2] NCCL INFO Channel 14/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484746:2485106 [5] NCCL INFO Channel 22/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484745:2485108 [4] NCCL INFO Channel 15/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484741:2485103 [0] NCCL INFO Channel 14/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484748:2485105 [7] NCCL INFO Channel 12/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484747:2485104 [6] NCCL INFO Channel 16/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-164-0:2484743:2485109 [2] NCCL INFO Channel 15/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484746:2485106 [5] NCCL INFO Channel 23/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484745:2485108 [4] NCCL INFO Channel 16/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484741:2485103 [0] NCCL INFO Channel 15/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484748:2485105 [7] NCCL INFO Channel 13/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484747:2485104 [6] NCCL INFO Channel 17/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-164-0:2484743:2485109 [2] NCCL INFO Channel 16/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484745:2485108 [4] NCCL INFO Channel 17/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484741:2485103 [0] NCCL INFO Channel 16/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484748:2485105 [7] NCCL INFO Channel 14/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484744:2485107 [3] NCCL INFO Channel 06/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484747:2485104 [6] NCCL INFO Channel 18/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-164-0:2484745:2485108 [4] NCCL INFO Channel 18/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484743:2485109 [2] NCCL INFO Channel 17/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484742:2485110 [1] NCCL INFO Channel 11/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484741:2485103 [0] NCCL INFO Channel 17/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484748:2485105 [7] NCCL INFO Channel 15/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484744:2485107 [3] NCCL INFO Channel 07/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484747:2485104 [6] NCCL INFO Channel 19/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-164-0:2484745:2485108 [4] NCCL INFO Channel 19/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484743:2485109 [2] NCCL INFO Channel 18/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484742:2485110 [1] NCCL INFO Channel 12/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484741:2485103 [0] NCCL INFO Channel 18/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484748:2485105 [7] NCCL INFO Channel 16/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484744:2485107 [3] NCCL INFO Channel 08/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484747:2485104 [6] NCCL INFO Channel 20/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-164-0:2484745:2485108 [4] NCCL INFO Channel 20/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484743:2485109 [2] NCCL INFO Channel 19/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484742:2485110 [1] NCCL INFO Channel 13/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484741:2485103 [0] NCCL INFO Channel 19/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484748:2485105 [7] NCCL INFO Channel 17/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484744:2485107 [3] NCCL INFO Channel 09/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484747:2485104 [6] NCCL INFO Channel 21/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-164-0:2484745:2485108 [4] NCCL INFO Channel 21/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484743:2485109 [2] NCCL INFO Channel 20/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484742:2485110 [1] NCCL INFO Channel 14/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484748:2485105 [7] NCCL INFO Channel 18/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484741:2485103 [0] NCCL INFO Channel 20/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484744:2485107 [3] NCCL INFO Channel 10/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484747:2485104 [6] NCCL INFO Channel 22/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-164-0:2484745:2485108 [4] NCCL INFO Channel 22/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484743:2485109 [2] NCCL INFO Channel 21/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484742:2485110 [1] NCCL INFO Channel 15/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484748:2485105 [7] NCCL INFO Channel 19/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484741:2485103 [0] NCCL INFO Channel 21/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484744:2485107 [3] NCCL INFO Channel 11/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484747:2485104 [6] NCCL INFO Channel 23/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-164-0:2484745:2485108 [4] NCCL INFO Channel 23/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484743:2485109 [2] NCCL INFO Channel 22/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484742:2485110 [1] NCCL INFO Channel 16/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484748:2485105 [7] NCCL INFO Channel 20/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484741:2485103 [0] NCCL INFO Channel 22/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484744:2485107 [3] NCCL INFO Channel 12/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484743:2485109 [2] NCCL INFO Channel 23/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484742:2485110 [1] NCCL INFO Channel 17/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484741:2485103 [0] NCCL INFO Channel 23/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484748:2485105 [7] NCCL INFO Channel 21/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484744:2485107 [3] NCCL INFO Channel 13/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484748:2485105 [7] NCCL INFO Channel 22/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484742:2485110 [1] NCCL INFO Channel 18/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484744:2485107 [3] NCCL INFO Channel 14/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484742:2485110 [1] NCCL INFO Channel 19/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484748:2485105 [7] NCCL INFO Channel 23/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484744:2485107 [3] NCCL INFO Channel 15/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484744:2485107 [3] NCCL INFO Channel 16/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484742:2485110 [1] NCCL INFO Channel 20/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484744:2485107 [3] NCCL INFO Channel 17/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484742:2485110 [1] NCCL INFO Channel 21/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484742:2485110 [1] NCCL INFO Channel 22/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484742:2485110 [1] NCCL INFO Channel 23/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484744:2485107 [3] NCCL INFO Channel 18/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484744:2485107 [3] NCCL INFO Channel 19/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484744:2485107 [3] NCCL INFO Channel 20/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484744:2485107 [3] NCCL INFO Channel 21/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484744:2485107 [3] NCCL INFO Channel 22/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484744:2485107 [3] NCCL INFO Channel 23/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484748:2485105 [7] NCCL INFO Connected all rings
+ip-26-0-164-0:2484748:2485105 [7] NCCL INFO Channel 00/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484746:2485106 [5] NCCL INFO Connected all rings
+ip-26-0-164-0:2484748:2485105 [7] NCCL INFO Channel 01/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484747:2485104 [6] NCCL INFO Connected all rings
+ip-26-0-164-0:2484748:2485105 [7] NCCL INFO Channel 02/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484741:2485103 [0] NCCL INFO Connected all rings
+ip-26-0-164-0:2484742:2485110 [1] NCCL INFO Connected all rings
+ip-26-0-164-0:2484748:2485105 [7] NCCL INFO Channel 03/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484748:2485105 [7] NCCL INFO Channel 04/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484745:2485108 [4] NCCL INFO Connected all rings
+ip-26-0-164-0:2484748:2485105 [7] NCCL INFO Channel 05/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484748:2485105 [7] NCCL INFO Channel 06/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484748:2485105 [7] NCCL INFO Channel 07/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484743:2485109 [2] NCCL INFO Connected all rings
+ip-26-0-164-0:2484748:2485105 [7] NCCL INFO Channel 08/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484744:2485107 [3] NCCL INFO Connected all rings
+ip-26-0-164-0:2484748:2485105 [7] NCCL INFO Channel 09/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484748:2485105 [7] NCCL INFO Channel 10/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484748:2485105 [7] NCCL INFO Channel 11/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484748:2485105 [7] NCCL INFO Channel 12/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484748:2485105 [7] NCCL INFO Channel 13/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484748:2485105 [7] NCCL INFO Channel 14/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484748:2485105 [7] NCCL INFO Channel 15/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484748:2485105 [7] NCCL INFO Channel 16/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484748:2485105 [7] NCCL INFO Channel 17/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484748:2485105 [7] NCCL INFO Channel 18/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484748:2485105 [7] NCCL INFO Channel 19/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484748:2485105 [7] NCCL INFO Channel 20/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484748:2485105 [7] NCCL INFO Channel 21/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484748:2485105 [7] NCCL INFO Channel 22/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484748:2485105 [7] NCCL INFO Channel 23/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-164-0:2484747:2485104 [6] NCCL INFO Channel 00/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484747:2485104 [6] NCCL INFO Channel 01/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484747:2485104 [6] NCCL INFO Channel 02/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484745:2485108 [4] NCCL INFO Channel 00/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484746:2485106 [5] NCCL INFO Channel 00/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484747:2485104 [6] NCCL INFO Channel 03/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484745:2485108 [4] NCCL INFO Channel 01/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484746:2485106 [5] NCCL INFO Channel 01/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484747:2485104 [6] NCCL INFO Channel 04/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484745:2485108 [4] NCCL INFO Channel 02/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484746:2485106 [5] NCCL INFO Channel 02/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484747:2485104 [6] NCCL INFO Channel 05/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484745:2485108 [4] NCCL INFO Channel 03/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484746:2485106 [5] NCCL INFO Channel 03/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484747:2485104 [6] NCCL INFO Channel 06/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484745:2485108 [4] NCCL INFO Channel 04/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484746:2485106 [5] NCCL INFO Channel 04/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484746:2485106 [5] NCCL INFO Channel 05/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484747:2485104 [6] NCCL INFO Channel 07/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484747:2485104 [6] NCCL INFO Channel 08/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484745:2485108 [4] NCCL INFO Channel 05/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484743:2485109 [2] NCCL INFO Channel 00/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484746:2485106 [5] NCCL INFO Channel 06/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484747:2485104 [6] NCCL INFO Channel 09/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484745:2485108 [4] NCCL INFO Channel 06/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484743:2485109 [2] NCCL INFO Channel 01/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484746:2485106 [5] NCCL INFO Channel 07/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484747:2485104 [6] NCCL INFO Channel 10/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484745:2485108 [4] NCCL INFO Channel 07/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484743:2485109 [2] NCCL INFO Channel 02/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484746:2485106 [5] NCCL INFO Channel 08/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484745:2485108 [4] NCCL INFO Channel 08/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484747:2485104 [6] NCCL INFO Channel 11/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484743:2485109 [2] NCCL INFO Channel 03/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484747:2485104 [6] NCCL INFO Channel 12/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484743:2485109 [2] NCCL INFO Channel 04/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484747:2485104 [6] NCCL INFO Channel 13/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484743:2485109 [2] NCCL INFO Channel 05/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484746:2485106 [5] NCCL INFO Channel 09/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484745:2485108 [4] NCCL INFO Channel 09/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484746:2485106 [5] NCCL INFO Channel 10/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484745:2485108 [4] NCCL INFO Channel 10/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484746:2485106 [5] NCCL INFO Channel 11/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484742:2485110 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484744:2485107 [3] NCCL INFO Channel 00/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484745:2485108 [4] NCCL INFO Channel 11/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484746:2485106 [5] NCCL INFO Channel 12/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484742:2485110 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484744:2485107 [3] NCCL INFO Channel 01/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484745:2485108 [4] NCCL INFO Channel 12/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484746:2485106 [5] NCCL INFO Channel 13/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484742:2485110 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484744:2485107 [3] NCCL INFO Channel 02/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484745:2485108 [4] NCCL INFO Channel 13/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484746:2485106 [5] NCCL INFO Channel 14/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484747:2485104 [6] NCCL INFO Channel 14/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484743:2485109 [2] NCCL INFO Channel 06/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484742:2485110 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484744:2485107 [3] NCCL INFO Channel 03/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484745:2485108 [4] NCCL INFO Channel 14/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484746:2485106 [5] NCCL INFO Channel 15/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484747:2485104 [6] NCCL INFO Channel 15/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484743:2485109 [2] NCCL INFO Channel 07/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484742:2485110 [1] NCCL INFO Channel 04/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484744:2485107 [3] NCCL INFO Channel 04/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484745:2485108 [4] NCCL INFO Channel 15/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484746:2485106 [5] NCCL INFO Channel 16/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484747:2485104 [6] NCCL INFO Channel 16/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484743:2485109 [2] NCCL INFO Channel 08/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484742:2485110 [1] NCCL INFO Channel 05/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484744:2485107 [3] NCCL INFO Channel 05/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484746:2485106 [5] NCCL INFO Channel 17/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484742:2485110 [1] NCCL INFO Channel 06/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484744:2485107 [3] NCCL INFO Channel 06/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484746:2485106 [5] NCCL INFO Channel 18/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484743:2485109 [2] NCCL INFO Channel 09/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484742:2485110 [1] NCCL INFO Channel 07/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484744:2485107 [3] NCCL INFO Channel 07/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484746:2485106 [5] NCCL INFO Channel 19/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484743:2485109 [2] NCCL INFO Channel 10/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484742:2485110 [1] NCCL INFO Channel 08/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484744:2485107 [3] NCCL INFO Channel 08/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484746:2485106 [5] NCCL INFO Channel 20/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484743:2485109 [2] NCCL INFO Channel 11/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484742:2485110 [1] NCCL INFO Channel 09/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484744:2485107 [3] NCCL INFO Channel 09/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484746:2485106 [5] NCCL INFO Channel 21/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484743:2485109 [2] NCCL INFO Channel 12/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484742:2485110 [1] NCCL INFO Channel 10/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484744:2485107 [3] NCCL INFO Channel 10/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484746:2485106 [5] NCCL INFO Channel 22/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484743:2485109 [2] NCCL INFO Channel 13/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484742:2485110 [1] NCCL INFO Channel 11/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484744:2485107 [3] NCCL INFO Channel 11/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484746:2485106 [5] NCCL INFO Channel 23/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-164-0:2484743:2485109 [2] NCCL INFO Channel 14/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484742:2485110 [1] NCCL INFO Channel 12/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484744:2485107 [3] NCCL INFO Channel 12/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484743:2485109 [2] NCCL INFO Channel 15/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484742:2485110 [1] NCCL INFO Channel 13/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484744:2485107 [3] NCCL INFO Channel 13/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484745:2485108 [4] NCCL INFO Channel 16/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484747:2485104 [6] NCCL INFO Channel 17/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484743:2485109 [2] NCCL INFO Channel 16/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484742:2485110 [1] NCCL INFO Channel 14/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484744:2485107 [3] NCCL INFO Channel 14/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484745:2485108 [4] NCCL INFO Channel 17/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484747:2485104 [6] NCCL INFO Channel 18/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484743:2485109 [2] NCCL INFO Channel 17/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484742:2485110 [1] NCCL INFO Channel 15/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484744:2485107 [3] NCCL INFO Channel 15/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484745:2485108 [4] NCCL INFO Channel 18/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484747:2485104 [6] NCCL INFO Channel 19/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484743:2485109 [2] NCCL INFO Channel 18/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484742:2485110 [1] NCCL INFO Channel 16/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484744:2485107 [3] NCCL INFO Channel 16/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484745:2485108 [4] NCCL INFO Channel 19/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484747:2485104 [6] NCCL INFO Channel 20/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484743:2485109 [2] NCCL INFO Channel 19/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484742:2485110 [1] NCCL INFO Channel 17/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484744:2485107 [3] NCCL INFO Channel 17/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484745:2485108 [4] NCCL INFO Channel 20/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484747:2485104 [6] NCCL INFO Channel 21/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484743:2485109 [2] NCCL INFO Channel 20/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484744:2485107 [3] NCCL INFO Channel 18/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484742:2485110 [1] NCCL INFO Channel 18/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484745:2485108 [4] NCCL INFO Channel 21/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484747:2485104 [6] NCCL INFO Channel 22/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484743:2485109 [2] NCCL INFO Channel 21/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484744:2485107 [3] NCCL INFO Channel 19/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484742:2485110 [1] NCCL INFO Channel 19/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484745:2485108 [4] NCCL INFO Channel 22/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484747:2485104 [6] NCCL INFO Channel 23/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-164-0:2484743:2485109 [2] NCCL INFO Channel 22/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484744:2485107 [3] NCCL INFO Channel 20/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484742:2485110 [1] NCCL INFO Channel 20/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484745:2485108 [4] NCCL INFO Channel 23/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484743:2485109 [2] NCCL INFO Channel 23/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484744:2485107 [3] NCCL INFO Channel 21/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484742:2485110 [1] NCCL INFO Channel 21/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484744:2485107 [3] NCCL INFO Channel 22/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484744:2485107 [3] NCCL INFO Channel 23/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484742:2485110 [1] NCCL INFO Channel 22/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484742:2485110 [1] NCCL INFO Channel 23/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484741:2485103 [0] NCCL INFO Connected all trees
+ip-26-0-164-0:2484741:2485103 [0] NCCL INFO NVLS comm 0x8694f40 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-164-0:2484748:2485105 [7] NCCL INFO Connected all trees
+ip-26-0-164-0:2484748:2485105 [7] NCCL INFO NVLS comm 0x866f6d0 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-164-0:2484746:2485106 [5] NCCL INFO Connected all trees
+ip-26-0-164-0:2484746:2485106 [5] NCCL INFO NVLS comm 0x89321e0 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-164-0:2484747:2485104 [6] NCCL INFO Connected all trees
+ip-26-0-164-0:2484747:2485104 [6] NCCL INFO NVLS comm 0x93ba060 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-164-0:2484742:2485110 [1] NCCL INFO Connected all trees
+ip-26-0-164-0:2484742:2485110 [1] NCCL INFO NVLS comm 0x9b210d0 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-164-0:2484744:2485107 [3] NCCL INFO Connected all trees
+ip-26-0-164-0:2484744:2485107 [3] NCCL INFO NVLS comm 0xa0c7a40 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-164-0:2484745:2485108 [4] NCCL INFO Connected all trees
+ip-26-0-164-0:2484745:2485108 [4] NCCL INFO NVLS comm 0x97b2900 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-164-0:2484743:2485109 [2] NCCL INFO Connected all trees
+ip-26-0-164-0:2484743:2485109 [2] NCCL INFO NVLS comm 0x8b3ee10 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-164-0:2484743:2485109 [2] NCCL INFO Connected NVLS tree
+ip-26-0-164-0:2484743:2485109 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-164-0:2484743:2485109 [2] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-164-0:2484743:2485109 [2] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-164-0:2484746:2485106 [5] NCCL INFO Connected NVLS tree
+ip-26-0-164-0:2484746:2485106 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-164-0:2484746:2485106 [5] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-164-0:2484746:2485106 [5] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-164-0:2484744:2485107 [3] NCCL INFO Connected NVLS tree
+ip-26-0-164-0:2484744:2485107 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-164-0:2484744:2485107 [3] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-164-0:2484744:2485107 [3] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-164-0:2484742:2485110 [1] NCCL INFO Connected NVLS tree
+ip-26-0-164-0:2484742:2485110 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-164-0:2484742:2485110 [1] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-164-0:2484742:2485110 [1] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-164-0:2484745:2485108 [4] NCCL INFO Connected NVLS tree
+ip-26-0-164-0:2484745:2485108 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-164-0:2484745:2485108 [4] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-164-0:2484745:2485108 [4] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-164-0:2484747:2485104 [6] NCCL INFO Connected NVLS tree
+ip-26-0-164-0:2484747:2485104 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-164-0:2484747:2485104 [6] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-164-0:2484747:2485104 [6] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-164-0:2484748:2485105 [7] NCCL INFO Connected NVLS tree
+ip-26-0-164-0:2484748:2485105 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-164-0:2484748:2485105 [7] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-164-0:2484748:2485105 [7] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-164-0:2484741:2485103 [0] NCCL INFO Connected NVLS tree
+ip-26-0-164-0:2484741:2485103 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-164-0:2484741:2485103 [0] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-164-0:2484741:2485103 [0] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-164-0:2484746:2485106 [5] NCCL INFO comm 0x89321e0 rank 5 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0x48681016f9ce6d74 - Init COMPLETE
+ip-26-0-164-0:2484748:2485105 [7] NCCL INFO comm 0x866f6d0 rank 7 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0x48681016f9ce6d74 - Init COMPLETE
+ip-26-0-164-0:2484744:2485107 [3] NCCL INFO comm 0xa0c7a40 rank 3 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0x48681016f9ce6d74 - Init COMPLETE
+ip-26-0-164-0:2484742:2485110 [1] NCCL INFO comm 0x9b210d0 rank 1 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0x48681016f9ce6d74 - Init COMPLETE
+ip-26-0-164-0:2484741:2485103 [0] NCCL INFO comm 0x8694f40 rank 0 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0x48681016f9ce6d74 - Init COMPLETE
+ip-26-0-164-0:2484743:2485109 [2] NCCL INFO comm 0x8b3ee10 rank 2 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0x48681016f9ce6d74 - Init COMPLETE
+ip-26-0-164-0:2484745:2485108 [4] NCCL INFO comm 0x97b2900 rank 4 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0x48681016f9ce6d74 - Init COMPLETE
+ip-26-0-164-0:2484747:2485104 [6] NCCL INFO comm 0x93ba060 rank 6 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0x48681016f9ce6d74 - Init COMPLETE
+12/28/2024 00:52:28 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]: Config:
+12/28/2024 00:52:28 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]: Config(general=GeneralArgs(project='debug',
+12/28/2024 00:52:28 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]:                            run='8.86G_dp4_tp1_pp2_acc1_mbs64_seq4096_zero1_tpmodeRED_vocab131k',
+12/28/2024 00:52:28 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]:                            seed=42,
+12/28/2024 00:52:28 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]:                            step=None,
+12/28/2024 00:52:28 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]:                            consumed_train_samples=None,
+12/28/2024 00:52:28 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]:                            benchmark_csv_path=PosixPath('benchmark/results/bench_final2.csv'),
+12/28/2024 00:52:28 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]:                            ignore_sanity_checks=True),
+12/28/2024 00:52:28 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]:        parallelism=ParallelismArgs(dp=4,
+12/28/2024 00:52:28 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]:                                    pp=2,
+12/28/2024 00:52:28 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]:                                    tp=1,
+12/28/2024 00:52:28 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]:                                    pp_engine=<nanotron.parallel.pipeline_parallel.engine.OneForwardOneBackwardPipelineEngine object at 0x7fe09aab8dc0>,
+12/28/2024 00:52:28 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]:                                    tp_mode=<TensorParallelLinearMode.REDUCE_SCATTER: 2>,
+12/28/2024 00:52:28 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]:                                    tp_linear_async_communication=True,
+12/28/2024 00:52:28 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]:                                    recompute_layer=False,
+12/28/2024 00:52:28 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]:                                    tp_recompute_allgather=True,
+12/28/2024 00:52:28 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]:                                    expert_parallel_size=1),
+12/28/2024 00:52:28 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]:        model=ModelArgs(model_config=LlamaConfig(bos_token_id=0,
+12/28/2024 00:52:28 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]:                                                 eos_token_id=0,
+12/28/2024 00:52:28 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]:                                                 hidden_act='silu',
+12/28/2024 00:52:28 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]:                                                 hidden_size=4096,
+12/28/2024 00:52:28 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]:                                                 initializer_range=0.02,
+12/28/2024 00:52:28 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]:                                                 intermediate_size=14336,
+12/28/2024 00:52:28 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]:                                                 is_llama_config=True,
+12/28/2024 00:52:28 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]:                                                 max_position_embeddings=4096,
+12/28/2024 00:52:28 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]:                                                 num_attention_heads=32,
+12/28/2024 00:52:28 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]:                                                 num_hidden_layers=32,
+12/28/2024 00:52:28 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]:                                                 num_key_value_heads=32,
+12/28/2024 00:52:28 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]:                                                 pad_token_id=None,
+12/28/2024 00:52:28 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]:                                                 pretraining_tp=1,
+12/28/2024 00:52:28 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]:                                                 rms_norm_eps=1e-05,
+12/28/2024 00:52:28 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]:                                                 rope_scaling=None,
+12/28/2024 00:52:28 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]:                                                 rope_theta=10000.0,
+12/28/2024 00:52:28 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]:                                                 rope_interleaved=False,
+12/28/2024 00:52:28 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]:                                                 tie_word_embeddings=False,
+12/28/2024 00:52:28 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]:                                                 use_cache=True,
+12/28/2024 00:52:28 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]:                                                 vocab_size=131072),
+12/28/2024 00:52:28 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]:                        init_method=RandomInit(std=0.02),
+12/28/2024 00:52:28 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]:                        dtype=torch.bfloat16,
+12/28/2024 00:52:28 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]:                        make_vocab_size_divisible_by=1,
+12/28/2024 00:52:28 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]:                        ddp_bucket_cap_mb=25),
+12/28/2024 00:52:28 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]:        tokenizer=TokenizerArgs(tokenizer_name_or_path='robot-test/dummy-tokenizer-wordlevel',
+12/28/2024 00:52:28 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]:                                tokenizer_revision=None,
+12/28/2024 00:52:28 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]:                                tokenizer_max_length=None),
+12/28/2024 00:52:28 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]:        checkpoints=CheckpointsArgs(checkpoints_path=PosixPath('checkpoints'),
+12/28/2024 00:52:28 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]:                                    checkpoint_interval=10000,
+12/28/2024 00:52:28 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]:                                    save_initial_state=False,
+12/28/2024 00:52:28 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]:                                    save_final_state=False,
+12/28/2024 00:52:28 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]:                                    resume_checkpoint_path=None,
+12/28/2024 00:52:28 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]:                                    checkpoints_path_is_shared_file_system=False),
+12/28/2024 00:52:28 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]:        logging=LoggingArgs(log_level='info',
+12/28/2024 00:52:28 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]:                            log_level_replica='info',
+12/28/2024 00:52:28 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]:                            iteration_step_info_interval=1),
+12/28/2024 00:52:28 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]:        tokens=TokensArgs(sequence_length=4096,
+12/28/2024 00:52:28 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]:                          train_steps=100,
+12/28/2024 00:52:28 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]:                          micro_batch_size=64,
+12/28/2024 00:52:28 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]:                          batch_accumulation_per_replica=1,
+12/28/2024 00:52:28 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]:                          val_check_interval=100,
+12/28/2024 00:52:28 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]:                          limit_val_batches=0,
+12/28/2024 00:52:28 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]:                          limit_test_batches=0),
+12/28/2024 00:52:28 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]:        optimizer=OptimizerArgs(optimizer_factory=AdamWOptimizerArgs(adam_eps=1e-08,
+12/28/2024 00:52:28 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]:                                                                     adam_beta1=0.9,
+12/28/2024 00:52:28 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]:                                                                     adam_beta2=0.95,
+12/28/2024 00:52:28 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]:                                                                     torch_adam_is_fused=True,
+12/28/2024 00:52:28 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]:                                                                     name='adamW'),
+12/28/2024 00:52:28 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]:                                zero_stage=1,
+12/28/2024 00:52:28 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]:                                weight_decay=0.01,
+12/28/2024 00:52:28 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]:                                clip_grad=1.0,
+12/28/2024 00:52:28 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]:                                accumulate_grad_in_fp32=True,
+12/28/2024 00:52:28 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]:                                learning_rate_scheduler=LRSchedulerArgs(learning_rate=0.0003,
+12/28/2024 00:52:28 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]:                                                                        lr_warmup_steps=2,
+12/28/2024 00:52:28 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]:                                                                        lr_warmup_style='linear',
+12/28/2024 00:52:28 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]:                                                                        lr_decay_style='cosine',
+12/28/2024 00:52:28 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]:                                                                        lr_decay_steps=13,
+12/28/2024 00:52:28 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]:                                                                        lr_decay_starting_step=None,
+12/28/2024 00:52:28 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]:                                                                        min_decay_lr=1e-05)),
+12/28/2024 00:52:28 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]:        data_stages=[DatasetStageArgs(name='Stable Training Stage',
+12/28/2024 00:52:28 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]:                                      start_training_step=1,
+12/28/2024 00:52:28 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]:                                      data=DataArgs(dataset=None,
+12/28/2024 00:52:28 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]:                                                    seed=42,
+12/28/2024 00:52:28 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]:                                                    num_loading_workers=1))],
+12/28/2024 00:52:28 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]:        profiler=None,
+12/28/2024 00:52:28 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]:        lighteval=None,
+12/28/2024 00:52:28 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]:        s3_upload=None)
+12/28/2024 00:52:28 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]: Model Config:
+12/28/2024 00:52:28 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]: LlamaConfig(bos_token_id=0,
+12/28/2024 00:52:28 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]:             eos_token_id=0,
+12/28/2024 00:52:28 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]:             hidden_act='silu',
+12/28/2024 00:52:28 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]:             hidden_size=4096,
+12/28/2024 00:52:28 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]:             initializer_range=0.02,
+12/28/2024 00:52:28 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]:             intermediate_size=14336,
+12/28/2024 00:52:28 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]:             is_llama_config=True,
+12/28/2024 00:52:28 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]:             max_position_embeddings=4096,
+12/28/2024 00:52:28 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]:             num_attention_heads=32,
+12/28/2024 00:52:28 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]:             num_hidden_layers=32,
+12/28/2024 00:52:28 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]:             num_key_value_heads=32,
+12/28/2024 00:52:28 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]:             pad_token_id=None,
+12/28/2024 00:52:28 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]:             pretraining_tp=1,
+12/28/2024 00:52:28 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]:             rms_norm_eps=1e-05,
+12/28/2024 00:52:28 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]:             rope_scaling=None,
+12/28/2024 00:52:28 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]:             rope_theta=10000.0,
+12/28/2024 00:52:28 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]:             rope_interleaved=False,
+12/28/2024 00:52:28 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]:             tie_word_embeddings=False,
+12/28/2024 00:52:28 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]:             use_cache=True,
+12/28/2024 00:52:28 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]:             vocab_size=131072)
+12/28/2024 00:52:28 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]: Building model..
+12/28/2024 00:52:28 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]: Initialize RoPE Theta = 10000.0
+12/28/2024 00:52:28 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]: Setting PP block ranks...
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+ip-26-0-164-0:2484745:2485144 [4] NCCL INFO Using network Libfabric
+ip-26-0-164-0:2484747:2485142 [6] NCCL INFO Using network Libfabric
+ip-26-0-164-0:2484746:2485145 [5] NCCL INFO Using network Libfabric
+ip-26-0-164-0:2484748:2485146 [7] NCCL INFO Using network Libfabric
+ip-26-0-164-0:2484741:2485147 [0] NCCL INFO Using network Libfabric
+ip-26-0-164-0:2484745:2485144 [4] NCCL INFO comm 0xb42c5a0 rank 0 nranks 1 cudaDev 4 nvmlDev 4 busId 97000 commId 0x1e090979c5241f34 - Init START
+ip-26-0-164-0:2484746:2485145 [5] NCCL INFO comm 0xa5ac3a0 rank 0 nranks 1 cudaDev 5 nvmlDev 5 busId a8000 commId 0xef2e24f43444c710 - Init START
+ip-26-0-164-0:2484747:2485142 [6] NCCL INFO comm 0xb035860 rank 0 nranks 1 cudaDev 6 nvmlDev 6 busId b9000 commId 0xa5336209729d3acd - Init START
+ip-26-0-164-0:2484745:2485144 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/168
+ip-26-0-164-0:2484744:2485153 [3] NCCL INFO Using network Libfabric
+ip-26-0-164-0:2484746:2485145 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/168
+ip-26-0-164-0:2484747:2485142 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/168
+ip-26-0-164-0:2484748:2485146 [7] NCCL INFO comm 0xa2eace0 rank 0 nranks 1 cudaDev 7 nvmlDev 7 busId ca000 commId 0xf518d5765d366d39 - Init START
+ip-26-0-164-0:2484742:2485152 [1] NCCL INFO Using network Libfabric
+ip-26-0-164-0:2484748:2485146 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/168
+ip-26-0-164-0:2484741:2485147 [0] NCCL INFO comm 0xa4dc170 rank 0 nranks 1 cudaDev 0 nvmlDev 0 busId 53000 commId 0xedcacd5a9bb9c2c9 - Init START
+ip-26-0-164-0:2484743:2485151 [2] NCCL INFO Using network Libfabric
+ip-26-0-164-0:2484741:2485147 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/181
+ip-26-0-164-0:2484744:2485153 [3] NCCL INFO comm 0xbf0dbb0 rank 0 nranks 1 cudaDev 3 nvmlDev 3 busId 86000 commId 0xa8d63340555d0554 - Init START
+ip-26-0-164-0:2484744:2485153 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/168
+ip-26-0-164-0:2484742:2485152 [1] NCCL INFO comm 0xb965d90 rank 0 nranks 1 cudaDev 1 nvmlDev 1 busId 64000 commId 0x839c4a55810fd6a4 - Init START
+ip-26-0-164-0:2484743:2485151 [2] NCCL INFO comm 0xa9821a0 rank 0 nranks 1 cudaDev 2 nvmlDev 2 busId 75000 commId 0xd613ae441411ab6f - Init START
+ip-26-0-164-0:2484742:2485152 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/168
+ip-26-0-164-0:2484743:2485151 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/168
+ip-26-0-164-0:2484742:2485152 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-164-0:2484742:2485152 [1] NCCL INFO Channel 00/32 :    0
+ip-26-0-164-0:2484742:2485152 [1] NCCL INFO Channel 01/32 :    0
+ip-26-0-164-0:2484742:2485152 [1] NCCL INFO Channel 02/32 :    0
+ip-26-0-164-0:2484742:2485152 [1] NCCL INFO Channel 03/32 :    0
+ip-26-0-164-0:2484742:2485152 [1] NCCL INFO Channel 04/32 :    0
+ip-26-0-164-0:2484742:2485152 [1] NCCL INFO Channel 05/32 :    0
+ip-26-0-164-0:2484742:2485152 [1] NCCL INFO Channel 06/32 :    0
+ip-26-0-164-0:2484742:2485152 [1] NCCL INFO Channel 07/32 :    0
+ip-26-0-164-0:2484742:2485152 [1] NCCL INFO Channel 08/32 :    0
+ip-26-0-164-0:2484742:2485152 [1] NCCL INFO Channel 09/32 :    0
+ip-26-0-164-0:2484742:2485152 [1] NCCL INFO Channel 10/32 :    0
+ip-26-0-164-0:2484742:2485152 [1] NCCL INFO Channel 11/32 :    0
+ip-26-0-164-0:2484742:2485152 [1] NCCL INFO Channel 12/32 :    0
+ip-26-0-164-0:2484742:2485152 [1] NCCL INFO Channel 13/32 :    0
+ip-26-0-164-0:2484742:2485152 [1] NCCL INFO Channel 14/32 :    0
+ip-26-0-164-0:2484742:2485152 [1] NCCL INFO Channel 15/32 :    0
+ip-26-0-164-0:2484742:2485152 [1] NCCL INFO Channel 16/32 :    0
+ip-26-0-164-0:2484742:2485152 [1] NCCL INFO Channel 17/32 :    0
+ip-26-0-164-0:2484742:2485152 [1] NCCL INFO Channel 18/32 :    0
+ip-26-0-164-0:2484742:2485152 [1] NCCL INFO Channel 19/32 :    0
+ip-26-0-164-0:2484742:2485152 [1] NCCL INFO Channel 20/32 :    0
+ip-26-0-164-0:2484742:2485152 [1] NCCL INFO Channel 21/32 :    0
+ip-26-0-164-0:2484742:2485152 [1] NCCL INFO Channel 22/32 :    0
+ip-26-0-164-0:2484742:2485152 [1] NCCL INFO Channel 23/32 :    0
+ip-26-0-164-0:2484742:2485152 [1] NCCL INFO Channel 24/32 :    0
+ip-26-0-164-0:2484742:2485152 [1] NCCL INFO Channel 25/32 :    0
+ip-26-0-164-0:2484742:2485152 [1] NCCL INFO Channel 26/32 :    0
+ip-26-0-164-0:2484742:2485152 [1] NCCL INFO Channel 27/32 :    0
+ip-26-0-164-0:2484742:2485152 [1] NCCL INFO Channel 28/32 :    0
+ip-26-0-164-0:2484742:2485152 [1] NCCL INFO Channel 29/32 :    0
+ip-26-0-164-0:2484742:2485152 [1] NCCL INFO Channel 30/32 :    0
+ip-26-0-164-0:2484742:2485152 [1] NCCL INFO Channel 31/32 :    0
+ip-26-0-164-0:2484742:2485152 [1] NCCL INFO Trees [0] -1/-1/-1->0->-1 [1] -1/-1/-1->0->-1 [2] -1/-1/-1->0->-1 [3] -1/-1/-1->0->-1 [4] -1/-1/-1->0->-1 [5] -1/-1/-1->0->-1 [6] -1/-1/-1->0->-1 [7] -1/-1/-1->0->-1 [8] -1/-1/-1->0->-1 [9] -1/-1/-1->0->-1 [10] -1/-1/-1->0->-1 [11] -1/-1/-1->0->-1 [12] -1/-1/-1->0->-1 [13] -1/-1/-1->0->-1 [14] -1/-1/-1->0->-1 [15] -1/-1/-1->0->-1 [16] -1/-1/-1->0->-1 [17] -1/-1/-1->0->-1 [18] -1/-1/-1->0->-1 [19] -1/-1/-1->0->-1 [20] -1/-1/-1->0->-1 [21] -1/-1/-1->0->-1 [22] -1/-1/-1->0->-1 [23] -1/-1/-1->0->-1 [24] -1/-1/-1->0->-1 [25] -1/-1/-1->0->-1 [26] -1/-1/-1->0->-1 [27] -1/-1/-1->0->-1 [28] -1/-1/-1->0->-1 [29] -1/-1/-1->0->-1 [30] -1/-1/-1->0->-1 [31] -1/-1/-1->0->-1
+ip-26-0-164-0:2484742:2485152 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-164-0:2484742:2485152 [1] NCCL INFO Connected all rings
+ip-26-0-164-0:2484742:2485152 [1] NCCL INFO Connected all trees
+ip-26-0-164-0:2484742:2485152 [1] NCCL INFO 32 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-164-0:2484743:2485151 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-164-0:2484743:2485151 [2] NCCL INFO Channel 00/32 :    0
+ip-26-0-164-0:2484743:2485151 [2] NCCL INFO Channel 01/32 :    0
+ip-26-0-164-0:2484743:2485151 [2] NCCL INFO Channel 02/32 :    0
+ip-26-0-164-0:2484743:2485151 [2] NCCL INFO Channel 03/32 :    0
+ip-26-0-164-0:2484743:2485151 [2] NCCL INFO Channel 04/32 :    0
+ip-26-0-164-0:2484743:2485151 [2] NCCL INFO Channel 05/32 :    0
+ip-26-0-164-0:2484743:2485151 [2] NCCL INFO Channel 06/32 :    0
+ip-26-0-164-0:2484743:2485151 [2] NCCL INFO Channel 07/32 :    0
+ip-26-0-164-0:2484743:2485151 [2] NCCL INFO Channel 08/32 :    0
+ip-26-0-164-0:2484743:2485151 [2] NCCL INFO Channel 09/32 :    0
+ip-26-0-164-0:2484743:2485151 [2] NCCL INFO Channel 10/32 :    0
+ip-26-0-164-0:2484743:2485151 [2] NCCL INFO Channel 11/32 :    0
+ip-26-0-164-0:2484743:2485151 [2] NCCL INFO Channel 12/32 :    0
+ip-26-0-164-0:2484743:2485151 [2] NCCL INFO Channel 13/32 :    0
+ip-26-0-164-0:2484743:2485151 [2] NCCL INFO Channel 14/32 :    0
+ip-26-0-164-0:2484743:2485151 [2] NCCL INFO Channel 15/32 :    0
+ip-26-0-164-0:2484743:2485151 [2] NCCL INFO Channel 16/32 :    0
+ip-26-0-164-0:2484743:2485151 [2] NCCL INFO Channel 17/32 :    0
+ip-26-0-164-0:2484743:2485151 [2] NCCL INFO Channel 18/32 :    0
+ip-26-0-164-0:2484743:2485151 [2] NCCL INFO Channel 19/32 :    0
+ip-26-0-164-0:2484743:2485151 [2] NCCL INFO Channel 20/32 :    0
+ip-26-0-164-0:2484743:2485151 [2] NCCL INFO Channel 21/32 :    0
+ip-26-0-164-0:2484743:2485151 [2] NCCL INFO Channel 22/32 :    0
+ip-26-0-164-0:2484743:2485151 [2] NCCL INFO Channel 23/32 :    0
+ip-26-0-164-0:2484743:2485151 [2] NCCL INFO Channel 24/32 :    0
+ip-26-0-164-0:2484743:2485151 [2] NCCL INFO Channel 25/32 :    0
+ip-26-0-164-0:2484743:2485151 [2] NCCL INFO Channel 26/32 :    0
+ip-26-0-164-0:2484743:2485151 [2] NCCL INFO Channel 27/32 :    0
+ip-26-0-164-0:2484743:2485151 [2] NCCL INFO Channel 28/32 :    0
+ip-26-0-164-0:2484743:2485151 [2] NCCL INFO Channel 29/32 :    0
+ip-26-0-164-0:2484743:2485151 [2] NCCL INFO Channel 30/32 :    0
+ip-26-0-164-0:2484743:2485151 [2] NCCL INFO Channel 31/32 :    0
+ip-26-0-164-0:2484743:2485151 [2] NCCL INFO Trees [0] -1/-1/-1->0->-1 [1] -1/-1/-1->0->-1 [2] -1/-1/-1->0->-1 [3] -1/-1/-1->0->-1 [4] -1/-1/-1->0->-1 [5] -1/-1/-1->0->-1 [6] -1/-1/-1->0->-1 [7] -1/-1/-1->0->-1 [8] -1/-1/-1->0->-1 [9] -1/-1/-1->0->-1 [10] -1/-1/-1->0->-1 [11] -1/-1/-1->0->-1 [12] -1/-1/-1->0->-1 [13] -1/-1/-1->0->-1 [14] -1/-1/-1->0->-1 [15] -1/-1/-1->0->-1 [16] -1/-1/-1->0->-1 [17] -1/-1/-1->0->-1 [18] -1/-1/-1->0->-1 [19] -1/-1/-1->0->-1 [20] -1/-1/-1->0->-1 [21] -1/-1/-1->0->-1 [22] -1/-1/-1->0->-1 [23] -1/-1/-1->0->-1 [24] -1/-1/-1->0->-1 [25] -1/-1/-1->0->-1 [26] -1/-1/-1->0->-1 [27] -1/-1/-1->0->-1 [28] -1/-1/-1->0->-1 [29] -1/-1/-1->0->-1 [30] -1/-1/-1->0->-1 [31] -1/-1/-1->0->-1
+ip-26-0-164-0:2484743:2485151 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-164-0:2484743:2485151 [2] NCCL INFO Connected all rings
+ip-26-0-164-0:2484743:2485151 [2] NCCL INFO Connected all trees
+ip-26-0-164-0:2484743:2485151 [2] NCCL INFO 32 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-164-0:2484741:2485147 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-164-0:2484741:2485147 [0] NCCL INFO Channel 00/32 :    0
+ip-26-0-164-0:2484741:2485147 [0] NCCL INFO Channel 01/32 :    0
+ip-26-0-164-0:2484741:2485147 [0] NCCL INFO Channel 02/32 :    0
+ip-26-0-164-0:2484741:2485147 [0] NCCL INFO Channel 03/32 :    0
+ip-26-0-164-0:2484741:2485147 [0] NCCL INFO Channel 04/32 :    0
+ip-26-0-164-0:2484741:2485147 [0] NCCL INFO Channel 05/32 :    0
+ip-26-0-164-0:2484741:2485147 [0] NCCL INFO Channel 06/32 :    0
+ip-26-0-164-0:2484741:2485147 [0] NCCL INFO Channel 07/32 :    0
+ip-26-0-164-0:2484741:2485147 [0] NCCL INFO Channel 08/32 :    0
+ip-26-0-164-0:2484741:2485147 [0] NCCL INFO Channel 09/32 :    0
+ip-26-0-164-0:2484741:2485147 [0] NCCL INFO Channel 10/32 :    0
+ip-26-0-164-0:2484741:2485147 [0] NCCL INFO Channel 11/32 :    0
+ip-26-0-164-0:2484741:2485147 [0] NCCL INFO Channel 12/32 :    0
+ip-26-0-164-0:2484741:2485147 [0] NCCL INFO Channel 13/32 :    0
+ip-26-0-164-0:2484741:2485147 [0] NCCL INFO Channel 14/32 :    0
+ip-26-0-164-0:2484741:2485147 [0] NCCL INFO Channel 15/32 :    0
+ip-26-0-164-0:2484741:2485147 [0] NCCL INFO Channel 16/32 :    0
+ip-26-0-164-0:2484741:2485147 [0] NCCL INFO Channel 17/32 :    0
+ip-26-0-164-0:2484741:2485147 [0] NCCL INFO Channel 18/32 :    0
+ip-26-0-164-0:2484741:2485147 [0] NCCL INFO Channel 19/32 :    0
+ip-26-0-164-0:2484741:2485147 [0] NCCL INFO Channel 20/32 :    0
+ip-26-0-164-0:2484741:2485147 [0] NCCL INFO Channel 21/32 :    0
+ip-26-0-164-0:2484741:2485147 [0] NCCL INFO Channel 22/32 :    0
+ip-26-0-164-0:2484741:2485147 [0] NCCL INFO Channel 23/32 :    0
+ip-26-0-164-0:2484741:2485147 [0] NCCL INFO Channel 24/32 :    0
+ip-26-0-164-0:2484741:2485147 [0] NCCL INFO Channel 25/32 :    0
+ip-26-0-164-0:2484741:2485147 [0] NCCL INFO Channel 26/32 :    0
+ip-26-0-164-0:2484741:2485147 [0] NCCL INFO Channel 27/32 :    0
+ip-26-0-164-0:2484741:2485147 [0] NCCL INFO Channel 28/32 :    0
+ip-26-0-164-0:2484741:2485147 [0] NCCL INFO Channel 29/32 :    0
+ip-26-0-164-0:2484741:2485147 [0] NCCL INFO Channel 30/32 :    0
+ip-26-0-164-0:2484741:2485147 [0] NCCL INFO Channel 31/32 :    0
+ip-26-0-164-0:2484741:2485147 [0] NCCL INFO Trees [0] -1/-1/-1->0->-1 [1] -1/-1/-1->0->-1 [2] -1/-1/-1->0->-1 [3] -1/-1/-1->0->-1 [4] -1/-1/-1->0->-1 [5] -1/-1/-1->0->-1 [6] -1/-1/-1->0->-1 [7] -1/-1/-1->0->-1 [8] -1/-1/-1->0->-1 [9] -1/-1/-1->0->-1 [10] -1/-1/-1->0->-1 [11] -1/-1/-1->0->-1 [12] -1/-1/-1->0->-1 [13] -1/-1/-1->0->-1 [14] -1/-1/-1->0->-1 [15] -1/-1/-1->0->-1 [16] -1/-1/-1->0->-1 [17] -1/-1/-1->0->-1 [18] -1/-1/-1->0->-1 [19] -1/-1/-1->0->-1 [20] -1/-1/-1->0->-1 [21] -1/-1/-1->0->-1 [22] -1/-1/-1->0->-1 [23] -1/-1/-1->0->-1 [24] -1/-1/-1->0->-1 [25] -1/-1/-1->0->-1 [26] -1/-1/-1->0->-1 [27] -1/-1/-1->0->-1 [28] -1/-1/-1->0->-1 [29] -1/-1/-1->0->-1 [30] -1/-1/-1->0->-1 [31] -1/-1/-1->0->-1
+ip-26-0-164-0:2484741:2485147 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-164-0:2484741:2485147 [0] NCCL INFO Connected all rings
+ip-26-0-164-0:2484741:2485147 [0] NCCL INFO Connected all trees
+ip-26-0-164-0:2484741:2485147 [0] NCCL INFO 32 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-164-0:2484743:2485151 [2] NCCL INFO comm 0xa9821a0 rank 0 nranks 1 cudaDev 2 nvmlDev 2 busId 75000 commId 0xd613ae441411ab6f - Init COMPLETE
+ip-26-0-164-0:2484742:2485152 [1] NCCL INFO comm 0xb965d90 rank 0 nranks 1 cudaDev 1 nvmlDev 1 busId 64000 commId 0x839c4a55810fd6a4 - Init COMPLETE
+ip-26-0-164-0:2484746:2485145 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-164-0:2484746:2485145 [5] NCCL INFO Channel 00/32 :    0
+ip-26-0-164-0:2484746:2485145 [5] NCCL INFO Channel 01/32 :    0
+ip-26-0-164-0:2484746:2485145 [5] NCCL INFO Channel 02/32 :    0
+ip-26-0-164-0:2484746:2485145 [5] NCCL INFO Channel 03/32 :    0
+ip-26-0-164-0:2484746:2485145 [5] NCCL INFO Channel 04/32 :    0
+ip-26-0-164-0:2484746:2485145 [5] NCCL INFO Channel 05/32 :    0
+ip-26-0-164-0:2484746:2485145 [5] NCCL INFO Channel 06/32 :    0
+ip-26-0-164-0:2484746:2485145 [5] NCCL INFO Channel 07/32 :    0
+ip-26-0-164-0:2484746:2485145 [5] NCCL INFO Channel 08/32 :    0
+ip-26-0-164-0:2484746:2485145 [5] NCCL INFO Channel 09/32 :    0
+ip-26-0-164-0:2484746:2485145 [5] NCCL INFO Channel 10/32 :    0
+ip-26-0-164-0:2484746:2485145 [5] NCCL INFO Channel 11/32 :    0
+ip-26-0-164-0:2484746:2485145 [5] NCCL INFO Channel 12/32 :    0
+ip-26-0-164-0:2484746:2485145 [5] NCCL INFO Channel 13/32 :    0
+ip-26-0-164-0:2484746:2485145 [5] NCCL INFO Channel 14/32 :    0
+ip-26-0-164-0:2484746:2485145 [5] NCCL INFO Channel 15/32 :    0
+ip-26-0-164-0:2484746:2485145 [5] NCCL INFO Channel 16/32 :    0
+ip-26-0-164-0:2484746:2485145 [5] NCCL INFO Channel 17/32 :    0
+ip-26-0-164-0:2484746:2485145 [5] NCCL INFO Channel 18/32 :    0
+ip-26-0-164-0:2484746:2485145 [5] NCCL INFO Channel 19/32 :    0
+ip-26-0-164-0:2484746:2485145 [5] NCCL INFO Channel 20/32 :    0
+ip-26-0-164-0:2484746:2485145 [5] NCCL INFO Channel 21/32 :    0
+ip-26-0-164-0:2484746:2485145 [5] NCCL INFO Channel 22/32 :    0
+ip-26-0-164-0:2484746:2485145 [5] NCCL INFO Channel 23/32 :    0
+ip-26-0-164-0:2484746:2485145 [5] NCCL INFO Channel 24/32 :    0
+ip-26-0-164-0:2484746:2485145 [5] NCCL INFO Channel 25/32 :    0
+ip-26-0-164-0:2484746:2485145 [5] NCCL INFO Channel 26/32 :    0
+ip-26-0-164-0:2484746:2485145 [5] NCCL INFO Channel 27/32 :    0
+ip-26-0-164-0:2484746:2485145 [5] NCCL INFO Channel 28/32 :    0
+ip-26-0-164-0:2484746:2485145 [5] NCCL INFO Channel 29/32 :    0
+ip-26-0-164-0:2484746:2485145 [5] NCCL INFO Channel 30/32 :    0
+ip-26-0-164-0:2484746:2485145 [5] NCCL INFO Channel 31/32 :    0
+ip-26-0-164-0:2484746:2485145 [5] NCCL INFO Trees [0] -1/-1/-1->0->-1 [1] -1/-1/-1->0->-1 [2] -1/-1/-1->0->-1 [3] -1/-1/-1->0->-1 [4] -1/-1/-1->0->-1 [5] -1/-1/-1->0->-1 [6] -1/-1/-1->0->-1 [7] -1/-1/-1->0->-1 [8] -1/-1/-1->0->-1 [9] -1/-1/-1->0->-1 [10] -1/-1/-1->0->-1 [11] -1/-1/-1->0->-1 [12] -1/-1/-1->0->-1 [13] -1/-1/-1->0->-1 [14] -1/-1/-1->0->-1 [15] -1/-1/-1->0->-1 [16] -1/-1/-1->0->-1 [17] -1/-1/-1->0->-1 [18] -1/-1/-1->0->-1 [19] -1/-1/-1->0->-1 [20] -1/-1/-1->0->-1 [21] -1/-1/-1->0->-1 [22] -1/-1/-1->0->-1 [23] -1/-1/-1->0->-1 [24] -1/-1/-1->0->-1 [25] -1/-1/-1->0->-1 [26] -1/-1/-1->0->-1 [27] -1/-1/-1->0->-1 [28] -1/-1/-1->0->-1 [29] -1/-1/-1->0->-1 [30] -1/-1/-1->0->-1 [31] -1/-1/-1->0->-1
+ip-26-0-164-0:2484746:2485145 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-164-0:2484746:2485145 [5] NCCL INFO Connected all rings
+ip-26-0-164-0:2484746:2485145 [5] NCCL INFO Connected all trees
+ip-26-0-164-0:2484746:2485145 [5] NCCL INFO 32 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-164-0:2484745:2485144 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-164-0:2484745:2485144 [4] NCCL INFO Channel 00/32 :    0
+ip-26-0-164-0:2484745:2485144 [4] NCCL INFO Channel 01/32 :    0
+ip-26-0-164-0:2484745:2485144 [4] NCCL INFO Channel 02/32 :    0
+ip-26-0-164-0:2484745:2485144 [4] NCCL INFO Channel 03/32 :    0
+ip-26-0-164-0:2484745:2485144 [4] NCCL INFO Channel 04/32 :    0
+ip-26-0-164-0:2484745:2485144 [4] NCCL INFO Channel 05/32 :    0
+ip-26-0-164-0:2484745:2485144 [4] NCCL INFO Channel 06/32 :    0
+ip-26-0-164-0:2484745:2485144 [4] NCCL INFO Channel 07/32 :    0
+ip-26-0-164-0:2484745:2485144 [4] NCCL INFO Channel 08/32 :    0
+ip-26-0-164-0:2484745:2485144 [4] NCCL INFO Channel 09/32 :    0
+ip-26-0-164-0:2484745:2485144 [4] NCCL INFO Channel 10/32 :    0
+ip-26-0-164-0:2484745:2485144 [4] NCCL INFO Channel 11/32 :    0
+ip-26-0-164-0:2484745:2485144 [4] NCCL INFO Channel 12/32 :    0
+ip-26-0-164-0:2484745:2485144 [4] NCCL INFO Channel 13/32 :    0
+ip-26-0-164-0:2484745:2485144 [4] NCCL INFO Channel 14/32 :    0
+ip-26-0-164-0:2484745:2485144 [4] NCCL INFO Channel 15/32 :    0
+ip-26-0-164-0:2484745:2485144 [4] NCCL INFO Channel 16/32 :    0
+ip-26-0-164-0:2484745:2485144 [4] NCCL INFO Channel 17/32 :    0
+ip-26-0-164-0:2484745:2485144 [4] NCCL INFO Channel 18/32 :    0
+ip-26-0-164-0:2484745:2485144 [4] NCCL INFO Channel 19/32 :    0
+ip-26-0-164-0:2484745:2485144 [4] NCCL INFO Channel 20/32 :    0
+ip-26-0-164-0:2484745:2485144 [4] NCCL INFO Channel 21/32 :    0
+ip-26-0-164-0:2484745:2485144 [4] NCCL INFO Channel 22/32 :    0
+ip-26-0-164-0:2484745:2485144 [4] NCCL INFO Channel 23/32 :    0
+ip-26-0-164-0:2484745:2485144 [4] NCCL INFO Channel 24/32 :    0
+ip-26-0-164-0:2484745:2485144 [4] NCCL INFO Channel 25/32 :    0
+ip-26-0-164-0:2484745:2485144 [4] NCCL INFO Channel 26/32 :    0
+ip-26-0-164-0:2484745:2485144 [4] NCCL INFO Channel 27/32 :    0
+ip-26-0-164-0:2484745:2485144 [4] NCCL INFO Channel 28/32 :    0
+ip-26-0-164-0:2484745:2485144 [4] NCCL INFO Channel 29/32 :    0
+ip-26-0-164-0:2484745:2485144 [4] NCCL INFO Channel 30/32 :    0
+ip-26-0-164-0:2484745:2485144 [4] NCCL INFO Channel 31/32 :    0
+ip-26-0-164-0:2484745:2485144 [4] NCCL INFO Trees [0] -1/-1/-1->0->-1 [1] -1/-1/-1->0->-1 [2] -1/-1/-1->0->-1 [3] -1/-1/-1->0->-1 [4] -1/-1/-1->0->-1 [5] -1/-1/-1->0->-1 [6] -1/-1/-1->0->-1 [7] -1/-1/-1->0->-1 [8] -1/-1/-1->0->-1 [9] -1/-1/-1->0->-1 [10] -1/-1/-1->0->-1 [11] -1/-1/-1->0->-1 [12] -1/-1/-1->0->-1 [13] -1/-1/-1->0->-1 [14] -1/-1/-1->0->-1 [15] -1/-1/-1->0->-1 [16] -1/-1/-1->0->-1 [17] -1/-1/-1->0->-1 [18] -1/-1/-1->0->-1 [19] -1/-1/-1->0->-1 [20] -1/-1/-1->0->-1 [21] -1/-1/-1->0->-1 [22] -1/-1/-1->0->-1 [23] -1/-1/-1->0->-1 [24] -1/-1/-1->0->-1 [25] -1/-1/-1->0->-1 [26] -1/-1/-1->0->-1 [27] -1/-1/-1->0->-1 [28] -1/-1/-1->0->-1 [29] -1/-1/-1->0->-1 [30] -1/-1/-1->0->-1 [31] -1/-1/-1->0->-1
+ip-26-0-164-0:2484745:2485144 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-164-0:2484745:2485144 [4] NCCL INFO Connected all rings
+ip-26-0-164-0:2484745:2485144 [4] NCCL INFO Connected all trees
+ip-26-0-164-0:2484745:2485144 [4] NCCL INFO 32 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-164-0:2484744:2485153 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-164-0:2484744:2485153 [3] NCCL INFO Channel 00/32 :    0
+ip-26-0-164-0:2484744:2485153 [3] NCCL INFO Channel 01/32 :    0
+ip-26-0-164-0:2484744:2485153 [3] NCCL INFO Channel 02/32 :    0
+ip-26-0-164-0:2484744:2485153 [3] NCCL INFO Channel 03/32 :    0
+ip-26-0-164-0:2484744:2485153 [3] NCCL INFO Channel 04/32 :    0
+ip-26-0-164-0:2484744:2485153 [3] NCCL INFO Channel 05/32 :    0
+ip-26-0-164-0:2484744:2485153 [3] NCCL INFO Channel 06/32 :    0
+ip-26-0-164-0:2484744:2485153 [3] NCCL INFO Channel 07/32 :    0
+ip-26-0-164-0:2484744:2485153 [3] NCCL INFO Channel 08/32 :    0
+ip-26-0-164-0:2484744:2485153 [3] NCCL INFO Channel 09/32 :    0
+ip-26-0-164-0:2484744:2485153 [3] NCCL INFO Channel 10/32 :    0
+ip-26-0-164-0:2484744:2485153 [3] NCCL INFO Channel 11/32 :    0
+ip-26-0-164-0:2484744:2485153 [3] NCCL INFO Channel 12/32 :    0
+ip-26-0-164-0:2484744:2485153 [3] NCCL INFO Channel 13/32 :    0
+ip-26-0-164-0:2484744:2485153 [3] NCCL INFO Channel 14/32 :    0
+ip-26-0-164-0:2484744:2485153 [3] NCCL INFO Channel 15/32 :    0
+ip-26-0-164-0:2484744:2485153 [3] NCCL INFO Channel 16/32 :    0
+ip-26-0-164-0:2484744:2485153 [3] NCCL INFO Channel 17/32 :    0
+ip-26-0-164-0:2484744:2485153 [3] NCCL INFO Channel 18/32 :    0
+ip-26-0-164-0:2484744:2485153 [3] NCCL INFO Channel 19/32 :    0
+ip-26-0-164-0:2484744:2485153 [3] NCCL INFO Channel 20/32 :    0
+ip-26-0-164-0:2484744:2485153 [3] NCCL INFO Channel 21/32 :    0
+ip-26-0-164-0:2484744:2485153 [3] NCCL INFO Channel 22/32 :    0
+ip-26-0-164-0:2484744:2485153 [3] NCCL INFO Channel 23/32 :    0
+ip-26-0-164-0:2484744:2485153 [3] NCCL INFO Channel 24/32 :    0
+ip-26-0-164-0:2484744:2485153 [3] NCCL INFO Channel 25/32 :    0
+ip-26-0-164-0:2484744:2485153 [3] NCCL INFO Channel 26/32 :    0
+ip-26-0-164-0:2484744:2485153 [3] NCCL INFO Channel 27/32 :    0
+ip-26-0-164-0:2484744:2485153 [3] NCCL INFO Channel 28/32 :    0
+ip-26-0-164-0:2484744:2485153 [3] NCCL INFO Channel 29/32 :    0
+ip-26-0-164-0:2484744:2485153 [3] NCCL INFO Channel 30/32 :    0
+ip-26-0-164-0:2484744:2485153 [3] NCCL INFO Channel 31/32 :    0
+ip-26-0-164-0:2484744:2485153 [3] NCCL INFO Trees [0] -1/-1/-1->0->-1 [1] -1/-1/-1->0->-1 [2] -1/-1/-1->0->-1 [3] -1/-1/-1->0->-1 [4] -1/-1/-1->0->-1 [5] -1/-1/-1->0->-1 [6] -1/-1/-1->0->-1 [7] -1/-1/-1->0->-1 [8] -1/-1/-1->0->-1 [9] -1/-1/-1->0->-1 [10] -1/-1/-1->0->-1 [11] -1/-1/-1->0->-1 [12] -1/-1/-1->0->-1 [13] -1/-1/-1->0->-1 [14] -1/-1/-1->0->-1 [15] -1/-1/-1->0->-1 [16] -1/-1/-1->0->-1 [17] -1/-1/-1->0->-1 [18] -1/-1/-1->0->-1 [19] -1/-1/-1->0->-1 [20] -1/-1/-1->0->-1 [21] -1/-1/-1->0->-1 [22] -1/-1/-1->0->-1 [23] -1/-1/-1->0->-1 [24] -1/-1/-1->0->-1 [25] -1/-1/-1->0->-1 [26] -1/-1/-1->0->-1 [27] -1/-1/-1->0->-1 [28] -1/-1/-1->0->-1 [29] -1/-1/-1->0->-1 [30] -1/-1/-1->0->-1 [31] -1/-1/-1->0->-1
+ip-26-0-164-0:2484744:2485153 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-164-0:2484747:2485142 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-164-0:2484747:2485142 [6] NCCL INFO Channel 00/32 :    0
+ip-26-0-164-0:2484747:2485142 [6] NCCL INFO Channel 01/32 :    0
+ip-26-0-164-0:2484747:2485142 [6] NCCL INFO Channel 02/32 :    0
+ip-26-0-164-0:2484747:2485142 [6] NCCL INFO Channel 03/32 :    0
+ip-26-0-164-0:2484747:2485142 [6] NCCL INFO Channel 04/32 :    0
+ip-26-0-164-0:2484747:2485142 [6] NCCL INFO Channel 05/32 :    0
+ip-26-0-164-0:2484747:2485142 [6] NCCL INFO Channel 06/32 :    0
+ip-26-0-164-0:2484747:2485142 [6] NCCL INFO Channel 07/32 :    0
+ip-26-0-164-0:2484747:2485142 [6] NCCL INFO Channel 08/32 :    0
+ip-26-0-164-0:2484747:2485142 [6] NCCL INFO Channel 09/32 :    0
+ip-26-0-164-0:2484747:2485142 [6] NCCL INFO Channel 10/32 :    0
+ip-26-0-164-0:2484747:2485142 [6] NCCL INFO Channel 11/32 :    0
+ip-26-0-164-0:2484747:2485142 [6] NCCL INFO Channel 12/32 :    0
+ip-26-0-164-0:2484747:2485142 [6] NCCL INFO Channel 13/32 :    0
+ip-26-0-164-0:2484747:2485142 [6] NCCL INFO Channel 14/32 :    0
+ip-26-0-164-0:2484747:2485142 [6] NCCL INFO Channel 15/32 :    0
+ip-26-0-164-0:2484747:2485142 [6] NCCL INFO Channel 16/32 :    0
+ip-26-0-164-0:2484747:2485142 [6] NCCL INFO Channel 17/32 :    0
+ip-26-0-164-0:2484747:2485142 [6] NCCL INFO Channel 18/32 :    0
+ip-26-0-164-0:2484747:2485142 [6] NCCL INFO Channel 19/32 :    0
+ip-26-0-164-0:2484747:2485142 [6] NCCL INFO Channel 20/32 :    0
+ip-26-0-164-0:2484747:2485142 [6] NCCL INFO Channel 21/32 :    0
+ip-26-0-164-0:2484747:2485142 [6] NCCL INFO Channel 22/32 :    0
+ip-26-0-164-0:2484747:2485142 [6] NCCL INFO Channel 23/32 :    0
+ip-26-0-164-0:2484747:2485142 [6] NCCL INFO Channel 24/32 :    0
+ip-26-0-164-0:2484747:2485142 [6] NCCL INFO Channel 25/32 :    0
+ip-26-0-164-0:2484747:2485142 [6] NCCL INFO Channel 26/32 :    0
+ip-26-0-164-0:2484747:2485142 [6] NCCL INFO Channel 27/32 :    0
+ip-26-0-164-0:2484747:2485142 [6] NCCL INFO Channel 28/32 :    0
+ip-26-0-164-0:2484747:2485142 [6] NCCL INFO Channel 29/32 :    0
+ip-26-0-164-0:2484747:2485142 [6] NCCL INFO Channel 30/32 :    0
+ip-26-0-164-0:2484747:2485142 [6] NCCL INFO Channel 31/32 :    0
+ip-26-0-164-0:2484747:2485142 [6] NCCL INFO Trees [0] -1/-1/-1->0->-1 [1] -1/-1/-1->0->-1 [2] -1/-1/-1->0->-1 [3] -1/-1/-1->0->-1 [4] -1/-1/-1->0->-1 [5] -1/-1/-1->0->-1 [6] -1/-1/-1->0->-1 [7] -1/-1/-1->0->-1 [8] -1/-1/-1->0->-1 [9] -1/-1/-1->0->-1 [10] -1/-1/-1->0->-1 [11] -1/-1/-1->0->-1 [12] -1/-1/-1->0->-1 [13] -1/-1/-1->0->-1 [14] -1/-1/-1->0->-1 [15] -1/-1/-1->0->-1 [16] -1/-1/-1->0->-1 [17] -1/-1/-1->0->-1 [18] -1/-1/-1->0->-1 [19] -1/-1/-1->0->-1 [20] -1/-1/-1->0->-1 [21] -1/-1/-1->0->-1 [22] -1/-1/-1->0->-1 [23] -1/-1/-1->0->-1 [24] -1/-1/-1->0->-1 [25] -1/-1/-1->0->-1 [26] -1/-1/-1->0->-1 [27] -1/-1/-1->0->-1 [28] -1/-1/-1->0->-1 [29] -1/-1/-1->0->-1 [30] -1/-1/-1->0->-1 [31] -1/-1/-1->0->-1
+ip-26-0-164-0:2484747:2485142 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-164-0:2484744:2485153 [3] NCCL INFO Connected all rings
+ip-26-0-164-0:2484744:2485153 [3] NCCL INFO Connected all trees
+ip-26-0-164-0:2484744:2485153 [3] NCCL INFO 32 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-164-0:2484741:2485147 [0] NCCL INFO comm 0xa4dc170 rank 0 nranks 1 cudaDev 0 nvmlDev 0 busId 53000 commId 0xedcacd5a9bb9c2c9 - Init COMPLETE
+ip-26-0-164-0:2484747:2485142 [6] NCCL INFO Connected all rings
+ip-26-0-164-0:2484747:2485142 [6] NCCL INFO Connected all trees
+ip-26-0-164-0:2484747:2485142 [6] NCCL INFO 32 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-164-0:2484748:2485146 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-164-0:2484748:2485146 [7] NCCL INFO Channel 00/32 :    0
+ip-26-0-164-0:2484748:2485146 [7] NCCL INFO Channel 01/32 :    0
+ip-26-0-164-0:2484748:2485146 [7] NCCL INFO Channel 02/32 :    0
+ip-26-0-164-0:2484748:2485146 [7] NCCL INFO Channel 03/32 :    0
+ip-26-0-164-0:2484748:2485146 [7] NCCL INFO Channel 04/32 :    0
+ip-26-0-164-0:2484748:2485146 [7] NCCL INFO Channel 05/32 :    0
+ip-26-0-164-0:2484748:2485146 [7] NCCL INFO Channel 06/32 :    0
+ip-26-0-164-0:2484748:2485146 [7] NCCL INFO Channel 07/32 :    0
+ip-26-0-164-0:2484748:2485146 [7] NCCL INFO Channel 08/32 :    0
+ip-26-0-164-0:2484748:2485146 [7] NCCL INFO Channel 09/32 :    0
+ip-26-0-164-0:2484748:2485146 [7] NCCL INFO Channel 10/32 :    0
+ip-26-0-164-0:2484748:2485146 [7] NCCL INFO Channel 11/32 :    0
+ip-26-0-164-0:2484748:2485146 [7] NCCL INFO Channel 12/32 :    0
+ip-26-0-164-0:2484748:2485146 [7] NCCL INFO Channel 13/32 :    0
+ip-26-0-164-0:2484748:2485146 [7] NCCL INFO Channel 14/32 :    0
+ip-26-0-164-0:2484748:2485146 [7] NCCL INFO Channel 15/32 :    0
+ip-26-0-164-0:2484748:2485146 [7] NCCL INFO Channel 16/32 :    0
+ip-26-0-164-0:2484748:2485146 [7] NCCL INFO Channel 17/32 :    0
+ip-26-0-164-0:2484748:2485146 [7] NCCL INFO Channel 18/32 :    0
+ip-26-0-164-0:2484748:2485146 [7] NCCL INFO Channel 19/32 :    0
+ip-26-0-164-0:2484748:2485146 [7] NCCL INFO Channel 20/32 :    0
+ip-26-0-164-0:2484748:2485146 [7] NCCL INFO Channel 21/32 :    0
+ip-26-0-164-0:2484748:2485146 [7] NCCL INFO Channel 22/32 :    0
+ip-26-0-164-0:2484748:2485146 [7] NCCL INFO Channel 23/32 :    0
+ip-26-0-164-0:2484748:2485146 [7] NCCL INFO Channel 24/32 :    0
+ip-26-0-164-0:2484748:2485146 [7] NCCL INFO Channel 25/32 :    0
+ip-26-0-164-0:2484748:2485146 [7] NCCL INFO Channel 26/32 :    0
+ip-26-0-164-0:2484748:2485146 [7] NCCL INFO Channel 27/32 :    0
+ip-26-0-164-0:2484748:2485146 [7] NCCL INFO Channel 28/32 :    0
+ip-26-0-164-0:2484748:2485146 [7] NCCL INFO Channel 29/32 :    0
+ip-26-0-164-0:2484748:2485146 [7] NCCL INFO Channel 30/32 :    0
+ip-26-0-164-0:2484748:2485146 [7] NCCL INFO Channel 31/32 :    0
+ip-26-0-164-0:2484748:2485146 [7] NCCL INFO Trees [0] -1/-1/-1->0->-1 [1] -1/-1/-1->0->-1 [2] -1/-1/-1->0->-1 [3] -1/-1/-1->0->-1 [4] -1/-1/-1->0->-1 [5] -1/-1/-1->0->-1 [6] -1/-1/-1->0->-1 [7] -1/-1/-1->0->-1 [8] -1/-1/-1->0->-1 [9] -1/-1/-1->0->-1 [10] -1/-1/-1->0->-1 [11] -1/-1/-1->0->-1 [12] -1/-1/-1->0->-1 [13] -1/-1/-1->0->-1 [14] -1/-1/-1->0->-1 [15] -1/-1/-1->0->-1 [16] -1/-1/-1->0->-1 [17] -1/-1/-1->0->-1 [18] -1/-1/-1->0->-1 [19] -1/-1/-1->0->-1 [20] -1/-1/-1->0->-1 [21] -1/-1/-1->0->-1 [22] -1/-1/-1->0->-1 [23] -1/-1/-1->0->-1 [24] -1/-1/-1->0->-1 [25] -1/-1/-1->0->-1 [26] -1/-1/-1->0->-1 [27] -1/-1/-1->0->-1 [28] -1/-1/-1->0->-1 [29] -1/-1/-1->0->-1 [30] -1/-1/-1->0->-1 [31] -1/-1/-1->0->-1
+ip-26-0-164-0:2484748:2485146 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-164-0:2484748:2485146 [7] NCCL INFO Connected all rings
+ip-26-0-164-0:2484748:2485146 [7] NCCL INFO Connected all trees
+ip-26-0-164-0:2484748:2485146 [7] NCCL INFO 32 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-164-0:2484745:2485144 [4] NCCL INFO comm 0xb42c5a0 rank 0 nranks 1 cudaDev 4 nvmlDev 4 busId 97000 commId 0x1e090979c5241f34 - Init COMPLETE
+ip-26-0-164-0:2484746:2485145 [5] NCCL INFO comm 0xa5ac3a0 rank 0 nranks 1 cudaDev 5 nvmlDev 5 busId a8000 commId 0xef2e24f43444c710 - Init COMPLETE
+ip-26-0-164-0:2484745:2485171 [4] NCCL INFO Using network Libfabric
+ip-26-0-164-0:2484741:2485170 [0] NCCL INFO Using network Libfabric
+ip-26-0-164-0:2484741:2485170 [0] NCCL INFO comm 0xa4dee20 rank 0 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0x91eff57b42c2199c - Init START
+ip-26-0-164-0:2484745:2485171 [4] NCCL INFO comm 0xb42f250 rank 1 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0x91eff57b42c2199c - Init START
+ip-26-0-164-0:2484745:2485171 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/168
+ip-26-0-164-0:2484741:2485170 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/181
+ip-26-0-164-0:2484742:2485174 [1] NCCL INFO Using network Libfabric
+ip-26-0-164-0:2484746:2485175 [5] NCCL INFO Using network Libfabric
+ip-26-0-164-0:2484746:2485175 [5] NCCL INFO comm 0xa5af050 rank 1 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0x2d884fdd270407ae - Init START
+ip-26-0-164-0:2484742:2485174 [1] NCCL INFO comm 0xb9690b0 rank 0 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0x2d884fdd270407ae - Init START
+ip-26-0-164-0:2484746:2485175 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/168
+ip-26-0-164-0:2484742:2485174 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/168
+ip-26-0-164-0:2484744:2485153 [3] NCCL INFO comm 0xbf0dbb0 rank 0 nranks 1 cudaDev 3 nvmlDev 3 busId 86000 commId 0xa8d63340555d0554 - Init COMPLETE
+ip-26-0-164-0:2484747:2485142 [6] NCCL INFO comm 0xb035860 rank 0 nranks 1 cudaDev 6 nvmlDev 6 busId b9000 commId 0xa5336209729d3acd - Init COMPLETE
+ip-26-0-164-0:2484748:2485146 [7] NCCL INFO comm 0xa2eace0 rank 0 nranks 1 cudaDev 7 nvmlDev 7 busId ca000 commId 0xf518d5765d366d39 - Init COMPLETE
+ip-26-0-164-0:2484743:2485177 [2] NCCL INFO Using network Libfabric
+ip-26-0-164-0:2484747:2485178 [6] NCCL INFO Using network Libfabric
+ip-26-0-164-0:2484747:2485178 [6] NCCL INFO comm 0xb038510 rank 1 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0x86bf22e4e2bbfb6 - Init START
+ip-26-0-164-0:2484743:2485177 [2] NCCL INFO comm 0xa9854c0 rank 0 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0x86bf22e4e2bbfb6 - Init START
+ip-26-0-164-0:2484747:2485178 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/168
+ip-26-0-164-0:2484743:2485177 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/168
+ip-26-0-164-0:2484744:2485180 [3] NCCL INFO Using network Libfabric
+ip-26-0-164-0:2484748:2485181 [7] NCCL INFO Using network Libfabric
+ip-26-0-164-0:2484744:2485180 [3] NCCL INFO comm 0xbf10860 rank 0 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0xb5b8c07d17244084 - Init START
+ip-26-0-164-0:2484748:2485181 [7] NCCL INFO comm 0xa2ed990 rank 1 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0xb5b8c07d17244084 - Init START
+ip-26-0-164-0:2484744:2485180 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/168
+ip-26-0-164-0:2484748:2485181 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/168
+ip-26-0-164-0:2484741:2485170 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-164-0:2484745:2485171 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-164-0:2484745:2485171 [4] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] -1/-1/-1->1->0 [3] -1/-1/-1->1->0 [4] -1/-1/-1->1->0 [5] -1/-1/-1->1->0 [6] 0/-1/-1->1->-1 [7] 0/-1/-1->1->-1 [8] 0/-1/-1->1->-1 [9] 0/-1/-1->1->-1 [10] 0/-1/-1->1->-1 [11] 0/-1/-1->1->-1 [12] -1/-1/-1->1->0 [13] -1/-1/-1->1->0 [14] -1/-1/-1->1->0 [15] -1/-1/-1->1->0 [16] -1/-1/-1->1->0 [17] -1/-1/-1->1->0 [18] 0/-1/-1->1->-1 [19] 0/-1/-1->1->-1 [20] 0/-1/-1->1->-1 [21] 0/-1/-1->1->-1 [22] 0/-1/-1->1->-1 [23] 0/-1/-1->1->-1
+ip-26-0-164-0:2484741:2485170 [0] NCCL INFO Channel 00/24 :    0   1
+ip-26-0-164-0:2484745:2485171 [4] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-164-0:2484741:2485170 [0] NCCL INFO Channel 01/24 :    0   1
+ip-26-0-164-0:2484741:2485170 [0] NCCL INFO Channel 02/24 :    0   1
+ip-26-0-164-0:2484741:2485170 [0] NCCL INFO Channel 03/24 :    0   1
+ip-26-0-164-0:2484741:2485170 [0] NCCL INFO Channel 04/24 :    0   1
+ip-26-0-164-0:2484741:2485170 [0] NCCL INFO Channel 05/24 :    0   1
+ip-26-0-164-0:2484741:2485170 [0] NCCL INFO Channel 06/24 :    0   1
+ip-26-0-164-0:2484741:2485170 [0] NCCL INFO Channel 07/24 :    0   1
+ip-26-0-164-0:2484741:2485170 [0] NCCL INFO Channel 08/24 :    0   1
+ip-26-0-164-0:2484741:2485170 [0] NCCL INFO Channel 09/24 :    0   1
+ip-26-0-164-0:2484741:2485170 [0] NCCL INFO Channel 10/24 :    0   1
+ip-26-0-164-0:2484741:2485170 [0] NCCL INFO Channel 11/24 :    0   1
+ip-26-0-164-0:2484741:2485170 [0] NCCL INFO Channel 12/24 :    0   1
+ip-26-0-164-0:2484741:2485170 [0] NCCL INFO Channel 13/24 :    0   1
+ip-26-0-164-0:2484741:2485170 [0] NCCL INFO Channel 14/24 :    0   1
+ip-26-0-164-0:2484741:2485170 [0] NCCL INFO Channel 15/24 :    0   1
+ip-26-0-164-0:2484741:2485170 [0] NCCL INFO Channel 16/24 :    0   1
+ip-26-0-164-0:2484741:2485170 [0] NCCL INFO Channel 17/24 :    0   1
+ip-26-0-164-0:2484741:2485170 [0] NCCL INFO Channel 18/24 :    0   1
+ip-26-0-164-0:2484741:2485170 [0] NCCL INFO Channel 19/24 :    0   1
+ip-26-0-164-0:2484741:2485170 [0] NCCL INFO Channel 20/24 :    0   1
+ip-26-0-164-0:2484741:2485170 [0] NCCL INFO Channel 21/24 :    0   1
+ip-26-0-164-0:2484741:2485170 [0] NCCL INFO Channel 22/24 :    0   1
+ip-26-0-164-0:2484741:2485170 [0] NCCL INFO Channel 23/24 :    0   1
+ip-26-0-164-0:2484741:2485170 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] -1/-1/-1->0->1 [7] -1/-1/-1->0->1 [8] -1/-1/-1->0->1 [9] -1/-1/-1->0->1 [10] -1/-1/-1->0->1 [11] -1/-1/-1->0->1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] -1/-1/-1->0->1 [19] -1/-1/-1->0->1 [20] -1/-1/-1->0->1 [21] -1/-1/-1->0->1 [22] -1/-1/-1->0->1 [23] -1/-1/-1->0->1
+ip-26-0-164-0:2484741:2485170 [0] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-164-0:2484742:2485174 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-164-0:2484748:2485181 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-164-0:2484746:2485175 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-164-0:2484742:2485174 [1] NCCL INFO Channel 00/24 :    0   1
+ip-26-0-164-0:2484742:2485174 [1] NCCL INFO Channel 01/24 :    0   1
+ip-26-0-164-0:2484742:2485174 [1] NCCL INFO Channel 02/24 :    0   1
+ip-26-0-164-0:2484746:2485175 [5] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] -1/-1/-1->1->0 [3] -1/-1/-1->1->0 [4] -1/-1/-1->1->0 [5] -1/-1/-1->1->0 [6] 0/-1/-1->1->-1 [7] 0/-1/-1->1->-1 [8] 0/-1/-1->1->-1 [9] 0/-1/-1->1->-1 [10] 0/-1/-1->1->-1 [11] 0/-1/-1->1->-1 [12] -1/-1/-1->1->0 [13] -1/-1/-1->1->0 [14] -1/-1/-1->1->0 [15] -1/-1/-1->1->0 [16] -1/-1/-1->1->0 [17] -1/-1/-1->1->0 [18] 0/-1/-1->1->-1 [19] 0/-1/-1->1->-1 [20] 0/-1/-1->1->-1 [21] 0/-1/-1->1->-1 [22] 0/-1/-1->1->-1 [23] 0/-1/-1->1->-1
+ip-26-0-164-0:2484742:2485174 [1] NCCL INFO Channel 03/24 :    0   1
+ip-26-0-164-0:2484742:2485174 [1] NCCL INFO Channel 04/24 :    0   1
+ip-26-0-164-0:2484746:2485175 [5] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-164-0:2484742:2485174 [1] NCCL INFO Channel 05/24 :    0   1
+ip-26-0-164-0:2484742:2485174 [1] NCCL INFO Channel 06/24 :    0   1
+ip-26-0-164-0:2484742:2485174 [1] NCCL INFO Channel 07/24 :    0   1
+ip-26-0-164-0:2484742:2485174 [1] NCCL INFO Channel 08/24 :    0   1
+ip-26-0-164-0:2484742:2485174 [1] NCCL INFO Channel 09/24 :    0   1
+ip-26-0-164-0:2484742:2485174 [1] NCCL INFO Channel 10/24 :    0   1
+ip-26-0-164-0:2484742:2485174 [1] NCCL INFO Channel 11/24 :    0   1
+ip-26-0-164-0:2484742:2485174 [1] NCCL INFO Channel 12/24 :    0   1
+ip-26-0-164-0:2484742:2485174 [1] NCCL INFO Channel 13/24 :    0   1
+ip-26-0-164-0:2484742:2485174 [1] NCCL INFO Channel 14/24 :    0   1
+ip-26-0-164-0:2484742:2485174 [1] NCCL INFO Channel 15/24 :    0   1
+ip-26-0-164-0:2484742:2485174 [1] NCCL INFO Channel 16/24 :    0   1
+ip-26-0-164-0:2484742:2485174 [1] NCCL INFO Channel 17/24 :    0   1
+ip-26-0-164-0:2484742:2485174 [1] NCCL INFO Channel 18/24 :    0   1
+ip-26-0-164-0:2484742:2485174 [1] NCCL INFO Channel 19/24 :    0   1
+ip-26-0-164-0:2484742:2485174 [1] NCCL INFO Channel 20/24 :    0   1
+ip-26-0-164-0:2484742:2485174 [1] NCCL INFO Channel 21/24 :    0   1
+ip-26-0-164-0:2484742:2485174 [1] NCCL INFO Channel 22/24 :    0   1
+ip-26-0-164-0:2484742:2485174 [1] NCCL INFO Channel 23/24 :    0   1
+ip-26-0-164-0:2484742:2485174 [1] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] -1/-1/-1->0->1 [7] -1/-1/-1->0->1 [8] -1/-1/-1->0->1 [9] -1/-1/-1->0->1 [10] -1/-1/-1->0->1 [11] -1/-1/-1->0->1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] -1/-1/-1->0->1 [19] -1/-1/-1->0->1 [20] -1/-1/-1->0->1 [21] -1/-1/-1->0->1 [22] -1/-1/-1->0->1 [23] -1/-1/-1->0->1
+ip-26-0-164-0:2484742:2485174 [1] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-164-0:2484744:2485180 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-164-0:2484744:2485180 [3] NCCL INFO Channel 00/24 :    0   1
+ip-26-0-164-0:2484748:2485181 [7] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] -1/-1/-1->1->0 [3] -1/-1/-1->1->0 [4] -1/-1/-1->1->0 [5] -1/-1/-1->1->0 [6] 0/-1/-1->1->-1 [7] 0/-1/-1->1->-1 [8] 0/-1/-1->1->-1 [9] 0/-1/-1->1->-1 [10] 0/-1/-1->1->-1 [11] 0/-1/-1->1->-1 [12] -1/-1/-1->1->0 [13] -1/-1/-1->1->0 [14] -1/-1/-1->1->0 [15] -1/-1/-1->1->0 [16] -1/-1/-1->1->0 [17] -1/-1/-1->1->0 [18] 0/-1/-1->1->-1 [19] 0/-1/-1->1->-1 [20] 0/-1/-1->1->-1 [21] 0/-1/-1->1->-1 [22] 0/-1/-1->1->-1 [23] 0/-1/-1->1->-1
+ip-26-0-164-0:2484744:2485180 [3] NCCL INFO Channel 01/24 :    0   1
+ip-26-0-164-0:2484748:2485181 [7] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-164-0:2484744:2485180 [3] NCCL INFO Channel 02/24 :    0   1
+ip-26-0-164-0:2484744:2485180 [3] NCCL INFO Channel 03/24 :    0   1
+ip-26-0-164-0:2484744:2485180 [3] NCCL INFO Channel 04/24 :    0   1
+ip-26-0-164-0:2484744:2485180 [3] NCCL INFO Channel 05/24 :    0   1
+ip-26-0-164-0:2484744:2485180 [3] NCCL INFO Channel 06/24 :    0   1
+ip-26-0-164-0:2484744:2485180 [3] NCCL INFO Channel 07/24 :    0   1
+ip-26-0-164-0:2484744:2485180 [3] NCCL INFO Channel 08/24 :    0   1
+ip-26-0-164-0:2484744:2485180 [3] NCCL INFO Channel 09/24 :    0   1
+ip-26-0-164-0:2484744:2485180 [3] NCCL INFO Channel 10/24 :    0   1
+ip-26-0-164-0:2484744:2485180 [3] NCCL INFO Channel 11/24 :    0   1
+ip-26-0-164-0:2484744:2485180 [3] NCCL INFO Channel 12/24 :    0   1
+ip-26-0-164-0:2484744:2485180 [3] NCCL INFO Channel 13/24 :    0   1
+ip-26-0-164-0:2484744:2485180 [3] NCCL INFO Channel 14/24 :    0   1
+ip-26-0-164-0:2484744:2485180 [3] NCCL INFO Channel 15/24 :    0   1
+ip-26-0-164-0:2484744:2485180 [3] NCCL INFO Channel 16/24 :    0   1
+ip-26-0-164-0:2484744:2485180 [3] NCCL INFO Channel 17/24 :    0   1
+ip-26-0-164-0:2484744:2485180 [3] NCCL INFO Channel 18/24 :    0   1
+ip-26-0-164-0:2484744:2485180 [3] NCCL INFO Channel 19/24 :    0   1
+ip-26-0-164-0:2484744:2485180 [3] NCCL INFO Channel 20/24 :    0   1
+ip-26-0-164-0:2484744:2485180 [3] NCCL INFO Channel 21/24 :    0   1
+ip-26-0-164-0:2484744:2485180 [3] NCCL INFO Channel 22/24 :    0   1
+ip-26-0-164-0:2484744:2485180 [3] NCCL INFO Channel 23/24 :    0   1
+ip-26-0-164-0:2484744:2485180 [3] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] -1/-1/-1->0->1 [7] -1/-1/-1->0->1 [8] -1/-1/-1->0->1 [9] -1/-1/-1->0->1 [10] -1/-1/-1->0->1 [11] -1/-1/-1->0->1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] -1/-1/-1->0->1 [19] -1/-1/-1->0->1 [20] -1/-1/-1->0->1 [21] -1/-1/-1->0->1 [22] -1/-1/-1->0->1 [23] -1/-1/-1->0->1
+ip-26-0-164-0:2484744:2485180 [3] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-164-0:2484747:2485178 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-164-0:2484743:2485177 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-164-0:2484743:2485177 [2] NCCL INFO Channel 00/24 :    0   1
+ip-26-0-164-0:2484743:2485177 [2] NCCL INFO Channel 01/24 :    0   1
+ip-26-0-164-0:2484743:2485177 [2] NCCL INFO Channel 02/24 :    0   1
+ip-26-0-164-0:2484743:2485177 [2] NCCL INFO Channel 03/24 :    0   1
+ip-26-0-164-0:2484743:2485177 [2] NCCL INFO Channel 04/24 :    0   1
+ip-26-0-164-0:2484743:2485177 [2] NCCL INFO Channel 05/24 :    0   1
+ip-26-0-164-0:2484743:2485177 [2] NCCL INFO Channel 06/24 :    0   1
+ip-26-0-164-0:2484743:2485177 [2] NCCL INFO Channel 07/24 :    0   1
+ip-26-0-164-0:2484743:2485177 [2] NCCL INFO Channel 08/24 :    0   1
+ip-26-0-164-0:2484743:2485177 [2] NCCL INFO Channel 09/24 :    0   1
+ip-26-0-164-0:2484747:2485178 [6] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] -1/-1/-1->1->0 [3] -1/-1/-1->1->0 [4] -1/-1/-1->1->0 [5] -1/-1/-1->1->0 [6] 0/-1/-1->1->-1 [7] 0/-1/-1->1->-1 [8] 0/-1/-1->1->-1 [9] 0/-1/-1->1->-1 [10] 0/-1/-1->1->-1 [11] 0/-1/-1->1->-1 [12] -1/-1/-1->1->0 [13] -1/-1/-1->1->0 [14] -1/-1/-1->1->0 [15] -1/-1/-1->1->0 [16] -1/-1/-1->1->0 [17] -1/-1/-1->1->0 [18] 0/-1/-1->1->-1 [19] 0/-1/-1->1->-1 [20] 0/-1/-1->1->-1 [21] 0/-1/-1->1->-1 [22] 0/-1/-1->1->-1 [23] 0/-1/-1->1->-1
+ip-26-0-164-0:2484743:2485177 [2] NCCL INFO Channel 10/24 :    0   1
+ip-26-0-164-0:2484747:2485178 [6] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-164-0:2484743:2485177 [2] NCCL INFO Channel 11/24 :    0   1
+ip-26-0-164-0:2484743:2485177 [2] NCCL INFO Channel 12/24 :    0   1
+ip-26-0-164-0:2484743:2485177 [2] NCCL INFO Channel 13/24 :    0   1
+ip-26-0-164-0:2484743:2485177 [2] NCCL INFO Channel 14/24 :    0   1
+ip-26-0-164-0:2484743:2485177 [2] NCCL INFO Channel 15/24 :    0   1
+ip-26-0-164-0:2484743:2485177 [2] NCCL INFO Channel 16/24 :    0   1
+ip-26-0-164-0:2484743:2485177 [2] NCCL INFO Channel 17/24 :    0   1
+ip-26-0-164-0:2484743:2485177 [2] NCCL INFO Channel 18/24 :    0   1
+ip-26-0-164-0:2484743:2485177 [2] NCCL INFO Channel 19/24 :    0   1
+ip-26-0-164-0:2484743:2485177 [2] NCCL INFO Channel 20/24 :    0   1
+ip-26-0-164-0:2484743:2485177 [2] NCCL INFO Channel 21/24 :    0   1
+ip-26-0-164-0:2484743:2485177 [2] NCCL INFO Channel 22/24 :    0   1
+ip-26-0-164-0:2484743:2485177 [2] NCCL INFO Channel 23/24 :    0   1
+ip-26-0-164-0:2484743:2485177 [2] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] -1/-1/-1->0->1 [7] -1/-1/-1->0->1 [8] -1/-1/-1->0->1 [9] -1/-1/-1->0->1 [10] -1/-1/-1->0->1 [11] -1/-1/-1->0->1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] -1/-1/-1->0->1 [19] -1/-1/-1->0->1 [20] -1/-1/-1->0->1 [21] -1/-1/-1->0->1 [22] -1/-1/-1->0->1 [23] -1/-1/-1->0->1
+ip-26-0-164-0:2484743:2485177 [2] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-164-0:2484741:2485170 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[4] via P2P/IPC
+ip-26-0-164-0:2484741:2485170 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[4] via P2P/IPC
+ip-26-0-164-0:2484741:2485170 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[4] via P2P/IPC
+ip-26-0-164-0:2484745:2485171 [4] NCCL INFO Channel 00/0 : 1[4] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484741:2485170 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[4] via P2P/IPC
+ip-26-0-164-0:2484745:2485171 [4] NCCL INFO Channel 01/0 : 1[4] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484741:2485170 [0] NCCL INFO Channel 04/0 : 0[0] -> 1[4] via P2P/IPC
+ip-26-0-164-0:2484745:2485171 [4] NCCL INFO Channel 02/0 : 1[4] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484741:2485170 [0] NCCL INFO Channel 05/0 : 0[0] -> 1[4] via P2P/IPC
+ip-26-0-164-0:2484745:2485171 [4] NCCL INFO Channel 03/0 : 1[4] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484741:2485170 [0] NCCL INFO Channel 06/0 : 0[0] -> 1[4] via P2P/IPC
+ip-26-0-164-0:2484745:2485171 [4] NCCL INFO Channel 04/0 : 1[4] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484746:2485175 [5] NCCL INFO Channel 00/0 : 1[5] -> 0[1] via P2P/IPC
+ip-26-0-164-0:2484741:2485170 [0] NCCL INFO Channel 07/0 : 0[0] -> 1[4] via P2P/IPC
+ip-26-0-164-0:2484741:2485170 [0] NCCL INFO Channel 08/0 : 0[0] -> 1[4] via P2P/IPC
+ip-26-0-164-0:2484746:2485175 [5] NCCL INFO Channel 01/0 : 1[5] -> 0[1] via P2P/IPC
+ip-26-0-164-0:2484746:2485175 [5] NCCL INFO Channel 02/0 : 1[5] -> 0[1] via P2P/IPC
+ip-26-0-164-0:2484741:2485170 [0] NCCL INFO Channel 09/0 : 0[0] -> 1[4] via P2P/IPC
+ip-26-0-164-0:2484748:2485181 [7] NCCL INFO Channel 00/0 : 1[7] -> 0[3] via P2P/IPC
+ip-26-0-164-0:2484746:2485175 [5] NCCL INFO Channel 03/0 : 1[5] -> 0[1] via P2P/IPC
+ip-26-0-164-0:2484741:2485170 [0] NCCL INFO Channel 10/0 : 0[0] -> 1[4] via P2P/IPC
+ip-26-0-164-0:2484748:2485181 [7] NCCL INFO Channel 01/0 : 1[7] -> 0[3] via P2P/IPC
+ip-26-0-164-0:2484746:2485175 [5] NCCL INFO Channel 04/0 : 1[5] -> 0[1] via P2P/IPC
+ip-26-0-164-0:2484741:2485170 [0] NCCL INFO Channel 11/0 : 0[0] -> 1[4] via P2P/IPC
+ip-26-0-164-0:2484748:2485181 [7] NCCL INFO Channel 02/0 : 1[7] -> 0[3] via P2P/IPC
+ip-26-0-164-0:2484741:2485170 [0] NCCL INFO Channel 12/0 : 0[0] -> 1[4] via P2P/IPC
+ip-26-0-164-0:2484746:2485175 [5] NCCL INFO Channel 05/0 : 1[5] -> 0[1] via P2P/IPC
+ip-26-0-164-0:2484748:2485181 [7] NCCL INFO Channel 03/0 : 1[7] -> 0[3] via P2P/IPC
+ip-26-0-164-0:2484741:2485170 [0] NCCL INFO Channel 13/0 : 0[0] -> 1[4] via P2P/IPC
+ip-26-0-164-0:2484746:2485175 [5] NCCL INFO Channel 06/0 : 1[5] -> 0[1] via P2P/IPC
+ip-26-0-164-0:2484748:2485181 [7] NCCL INFO Channel 04/0 : 1[7] -> 0[3] via P2P/IPC
+ip-26-0-164-0:2484741:2485170 [0] NCCL INFO Channel 14/0 : 0[0] -> 1[4] via P2P/IPC
+ip-26-0-164-0:2484746:2485175 [5] NCCL INFO Channel 07/0 : 1[5] -> 0[1] via P2P/IPC
+ip-26-0-164-0:2484748:2485181 [7] NCCL INFO Channel 05/0 : 1[7] -> 0[3] via P2P/IPC
+ip-26-0-164-0:2484747:2485178 [6] NCCL INFO Channel 00/0 : 1[6] -> 0[2] via P2P/IPC
+ip-26-0-164-0:2484745:2485171 [4] NCCL INFO Channel 05/0 : 1[4] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484741:2485170 [0] NCCL INFO Channel 15/0 : 0[0] -> 1[4] via P2P/IPC
+ip-26-0-164-0:2484746:2485175 [5] NCCL INFO Channel 08/0 : 1[5] -> 0[1] via P2P/IPC
+ip-26-0-164-0:2484742:2485174 [1] NCCL INFO Channel 00/0 : 0[1] -> 1[5] via P2P/IPC
+ip-26-0-164-0:2484748:2485181 [7] NCCL INFO Channel 06/0 : 1[7] -> 0[3] via P2P/IPC
+ip-26-0-164-0:2484747:2485178 [6] NCCL INFO Channel 01/0 : 1[6] -> 0[2] via P2P/IPC
+ip-26-0-164-0:2484745:2485171 [4] NCCL INFO Channel 06/0 : 1[4] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484741:2485170 [0] NCCL INFO Channel 16/0 : 0[0] -> 1[4] via P2P/IPC
+ip-26-0-164-0:2484746:2485175 [5] NCCL INFO Channel 09/0 : 1[5] -> 0[1] via P2P/IPC
+ip-26-0-164-0:2484742:2485174 [1] NCCL INFO Channel 01/0 : 0[1] -> 1[5] via P2P/IPC
+ip-26-0-164-0:2484748:2485181 [7] NCCL INFO Channel 07/0 : 1[7] -> 0[3] via P2P/IPC
+ip-26-0-164-0:2484747:2485178 [6] NCCL INFO Channel 02/0 : 1[6] -> 0[2] via P2P/IPC
+ip-26-0-164-0:2484745:2485171 [4] NCCL INFO Channel 07/0 : 1[4] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484741:2485170 [0] NCCL INFO Channel 17/0 : 0[0] -> 1[4] via P2P/IPC
+ip-26-0-164-0:2484746:2485175 [5] NCCL INFO Channel 10/0 : 1[5] -> 0[1] via P2P/IPC
+ip-26-0-164-0:2484742:2485174 [1] NCCL INFO Channel 02/0 : 0[1] -> 1[5] via P2P/IPC
+ip-26-0-164-0:2484748:2485181 [7] NCCL INFO Channel 08/0 : 1[7] -> 0[3] via P2P/IPC
+ip-26-0-164-0:2484747:2485178 [6] NCCL INFO Channel 03/0 : 1[6] -> 0[2] via P2P/IPC
+ip-26-0-164-0:2484745:2485171 [4] NCCL INFO Channel 08/0 : 1[4] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484741:2485170 [0] NCCL INFO Channel 18/0 : 0[0] -> 1[4] via P2P/IPC
+ip-26-0-164-0:2484746:2485175 [5] NCCL INFO Channel 11/0 : 1[5] -> 0[1] via P2P/IPC
+ip-26-0-164-0:2484744:2485180 [3] NCCL INFO Channel 00/0 : 0[3] -> 1[7] via P2P/IPC
+ip-26-0-164-0:2484742:2485174 [1] NCCL INFO Channel 03/0 : 0[1] -> 1[5] via P2P/IPC
+ip-26-0-164-0:2484748:2485181 [7] NCCL INFO Channel 09/0 : 1[7] -> 0[3] via P2P/IPC
+ip-26-0-164-0:2484747:2485178 [6] NCCL INFO Channel 04/0 : 1[6] -> 0[2] via P2P/IPC
+ip-26-0-164-0:2484745:2485171 [4] NCCL INFO Channel 09/0 : 1[4] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484741:2485170 [0] NCCL INFO Channel 19/0 : 0[0] -> 1[4] via P2P/IPC
+ip-26-0-164-0:2484746:2485175 [5] NCCL INFO Channel 12/0 : 1[5] -> 0[1] via P2P/IPC
+ip-26-0-164-0:2484744:2485180 [3] NCCL INFO Channel 01/0 : 0[3] -> 1[7] via P2P/IPC
+ip-26-0-164-0:2484742:2485174 [1] NCCL INFO Channel 04/0 : 0[1] -> 1[5] via P2P/IPC
+ip-26-0-164-0:2484748:2485181 [7] NCCL INFO Channel 10/0 : 1[7] -> 0[3] via P2P/IPC
+ip-26-0-164-0:2484745:2485171 [4] NCCL INFO Channel 10/0 : 1[4] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484747:2485178 [6] NCCL INFO Channel 05/0 : 1[6] -> 0[2] via P2P/IPC
+ip-26-0-164-0:2484741:2485170 [0] NCCL INFO Channel 20/0 : 0[0] -> 1[4] via P2P/IPC
+ip-26-0-164-0:2484746:2485175 [5] NCCL INFO Channel 13/0 : 1[5] -> 0[1] via P2P/IPC
+ip-26-0-164-0:2484744:2485180 [3] NCCL INFO Channel 02/0 : 0[3] -> 1[7] via P2P/IPC
+ip-26-0-164-0:2484742:2485174 [1] NCCL INFO Channel 05/0 : 0[1] -> 1[5] via P2P/IPC
+ip-26-0-164-0:2484743:2485177 [2] NCCL INFO Channel 00/0 : 0[2] -> 1[6] via P2P/IPC
+ip-26-0-164-0:2484748:2485181 [7] NCCL INFO Channel 11/0 : 1[7] -> 0[3] via P2P/IPC
+ip-26-0-164-0:2484745:2485171 [4] NCCL INFO Channel 11/0 : 1[4] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484747:2485178 [6] NCCL INFO Channel 06/0 : 1[6] -> 0[2] via P2P/IPC
+ip-26-0-164-0:2484744:2485180 [3] NCCL INFO Channel 03/0 : 0[3] -> 1[7] via P2P/IPC
+ip-26-0-164-0:2484742:2485174 [1] NCCL INFO Channel 06/0 : 0[1] -> 1[5] via P2P/IPC
+ip-26-0-164-0:2484743:2485177 [2] NCCL INFO Channel 01/0 : 0[2] -> 1[6] via P2P/IPC
+ip-26-0-164-0:2484744:2485180 [3] NCCL INFO Channel 04/0 : 0[3] -> 1[7] via P2P/IPC
+ip-26-0-164-0:2484742:2485174 [1] NCCL INFO Channel 07/0 : 0[1] -> 1[5] via P2P/IPC
+ip-26-0-164-0:2484748:2485181 [7] NCCL INFO Channel 12/0 : 1[7] -> 0[3] via P2P/IPC
+ip-26-0-164-0:2484743:2485177 [2] NCCL INFO Channel 02/0 : 0[2] -> 1[6] via P2P/IPC
+ip-26-0-164-0:2484744:2485180 [3] NCCL INFO Channel 05/0 : 0[3] -> 1[7] via P2P/IPC
+ip-26-0-164-0:2484742:2485174 [1] NCCL INFO Channel 08/0 : 0[1] -> 1[5] via P2P/IPC
+ip-26-0-164-0:2484743:2485177 [2] NCCL INFO Channel 03/0 : 0[2] -> 1[6] via P2P/IPC
+ip-26-0-164-0:2484748:2485181 [7] NCCL INFO Channel 13/0 : 1[7] -> 0[3] via P2P/IPC
+ip-26-0-164-0:2484744:2485180 [3] NCCL INFO Channel 06/0 : 0[3] -> 1[7] via P2P/IPC
+ip-26-0-164-0:2484742:2485174 [1] NCCL INFO Channel 09/0 : 0[1] -> 1[5] via P2P/IPC
+ip-26-0-164-0:2484743:2485177 [2] NCCL INFO Channel 04/0 : 0[2] -> 1[6] via P2P/IPC
+ip-26-0-164-0:2484748:2485181 [7] NCCL INFO Channel 14/0 : 1[7] -> 0[3] via P2P/IPC
+ip-26-0-164-0:2484744:2485180 [3] NCCL INFO Channel 07/0 : 0[3] -> 1[7] via P2P/IPC
+ip-26-0-164-0:2484745:2485171 [4] NCCL INFO Channel 12/0 : 1[4] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484742:2485174 [1] NCCL INFO Channel 10/0 : 0[1] -> 1[5] via P2P/IPC
+ip-26-0-164-0:2484743:2485177 [2] NCCL INFO Channel 05/0 : 0[2] -> 1[6] via P2P/IPC
+ip-26-0-164-0:2484748:2485181 [7] NCCL INFO Channel 15/0 : 1[7] -> 0[3] via P2P/IPC
+ip-26-0-164-0:2484744:2485180 [3] NCCL INFO Channel 08/0 : 0[3] -> 1[7] via P2P/IPC
+ip-26-0-164-0:2484745:2485171 [4] NCCL INFO Channel 13/0 : 1[4] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484742:2485174 [1] NCCL INFO Channel 11/0 : 0[1] -> 1[5] via P2P/IPC
+ip-26-0-164-0:2484743:2485177 [2] NCCL INFO Channel 06/0 : 0[2] -> 1[6] via P2P/IPC
+ip-26-0-164-0:2484748:2485181 [7] NCCL INFO Channel 16/0 : 1[7] -> 0[3] via P2P/IPC
+ip-26-0-164-0:2484744:2485180 [3] NCCL INFO Channel 09/0 : 0[3] -> 1[7] via P2P/IPC
+ip-26-0-164-0:2484745:2485171 [4] NCCL INFO Channel 14/0 : 1[4] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484742:2485174 [1] NCCL INFO Channel 12/0 : 0[1] -> 1[5] via P2P/IPC
+ip-26-0-164-0:2484743:2485177 [2] NCCL INFO Channel 07/0 : 0[2] -> 1[6] via P2P/IPC
+ip-26-0-164-0:2484748:2485181 [7] NCCL INFO Channel 17/0 : 1[7] -> 0[3] via P2P/IPC
+ip-26-0-164-0:2484746:2485175 [5] NCCL INFO Channel 14/0 : 1[5] -> 0[1] via P2P/IPC
+ip-26-0-164-0:2484741:2485170 [0] NCCL INFO Channel 21/0 : 0[0] -> 1[4] via P2P/IPC
+ip-26-0-164-0:2484745:2485171 [4] NCCL INFO Channel 15/0 : 1[4] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484744:2485180 [3] NCCL INFO Channel 10/0 : 0[3] -> 1[7] via P2P/IPC
+ip-26-0-164-0:2484742:2485174 [1] NCCL INFO Channel 13/0 : 0[1] -> 1[5] via P2P/IPC
+ip-26-0-164-0:2484747:2485178 [6] NCCL INFO Channel 07/0 : 1[6] -> 0[2] via P2P/IPC
+ip-26-0-164-0:2484743:2485177 [2] NCCL INFO Channel 08/0 : 0[2] -> 1[6] via P2P/IPC
+ip-26-0-164-0:2484748:2485181 [7] NCCL INFO Channel 18/0 : 1[7] -> 0[3] via P2P/IPC
+ip-26-0-164-0:2484746:2485175 [5] NCCL INFO Channel 15/0 : 1[5] -> 0[1] via P2P/IPC
+ip-26-0-164-0:2484741:2485170 [0] NCCL INFO Channel 22/0 : 0[0] -> 1[4] via P2P/IPC
+ip-26-0-164-0:2484745:2485171 [4] NCCL INFO Channel 16/0 : 1[4] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484744:2485180 [3] NCCL INFO Channel 11/0 : 0[3] -> 1[7] via P2P/IPC
+ip-26-0-164-0:2484742:2485174 [1] NCCL INFO Channel 14/0 : 0[1] -> 1[5] via P2P/IPC
+ip-26-0-164-0:2484743:2485177 [2] NCCL INFO Channel 09/0 : 0[2] -> 1[6] via P2P/IPC
+ip-26-0-164-0:2484747:2485178 [6] NCCL INFO Channel 08/0 : 1[6] -> 0[2] via P2P/IPC
+ip-26-0-164-0:2484748:2485181 [7] NCCL INFO Channel 19/0 : 1[7] -> 0[3] via P2P/IPC
+ip-26-0-164-0:2484746:2485175 [5] NCCL INFO Channel 16/0 : 1[5] -> 0[1] via P2P/IPC
+ip-26-0-164-0:2484741:2485170 [0] NCCL INFO Channel 23/0 : 0[0] -> 1[4] via P2P/IPC
+ip-26-0-164-0:2484745:2485171 [4] NCCL INFO Channel 17/0 : 1[4] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484744:2485180 [3] NCCL INFO Channel 12/0 : 0[3] -> 1[7] via P2P/IPC
+ip-26-0-164-0:2484742:2485174 [1] NCCL INFO Channel 15/0 : 0[1] -> 1[5] via P2P/IPC
+ip-26-0-164-0:2484743:2485177 [2] NCCL INFO Channel 10/0 : 0[2] -> 1[6] via P2P/IPC
+ip-26-0-164-0:2484747:2485178 [6] NCCL INFO Channel 09/0 : 1[6] -> 0[2] via P2P/IPC
+ip-26-0-164-0:2484748:2485181 [7] NCCL INFO Channel 20/0 : 1[7] -> 0[3] via P2P/IPC
+ip-26-0-164-0:2484746:2485175 [5] NCCL INFO Channel 17/0 : 1[5] -> 0[1] via P2P/IPC
+ip-26-0-164-0:2484745:2485171 [4] NCCL INFO Channel 18/0 : 1[4] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484744:2485180 [3] NCCL INFO Channel 13/0 : 0[3] -> 1[7] via P2P/IPC
+ip-26-0-164-0:2484742:2485174 [1] NCCL INFO Channel 16/0 : 0[1] -> 1[5] via P2P/IPC
+ip-26-0-164-0:2484743:2485177 [2] NCCL INFO Channel 11/0 : 0[2] -> 1[6] via P2P/IPC
+ip-26-0-164-0:2484747:2485178 [6] NCCL INFO Channel 10/0 : 1[6] -> 0[2] via P2P/IPC
+ip-26-0-164-0:2484748:2485181 [7] NCCL INFO Channel 21/0 : 1[7] -> 0[3] via P2P/IPC
+ip-26-0-164-0:2484746:2485175 [5] NCCL INFO Channel 18/0 : 1[5] -> 0[1] via P2P/IPC
+ip-26-0-164-0:2484745:2485171 [4] NCCL INFO Channel 19/0 : 1[4] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484744:2485180 [3] NCCL INFO Channel 14/0 : 0[3] -> 1[7] via P2P/IPC
+ip-26-0-164-0:2484742:2485174 [1] NCCL INFO Channel 17/0 : 0[1] -> 1[5] via P2P/IPC
+ip-26-0-164-0:2484743:2485177 [2] NCCL INFO Channel 12/0 : 0[2] -> 1[6] via P2P/IPC
+ip-26-0-164-0:2484747:2485178 [6] NCCL INFO Channel 11/0 : 1[6] -> 0[2] via P2P/IPC
+ip-26-0-164-0:2484748:2485181 [7] NCCL INFO Channel 22/0 : 1[7] -> 0[3] via P2P/IPC
+ip-26-0-164-0:2484746:2485175 [5] NCCL INFO Channel 19/0 : 1[5] -> 0[1] via P2P/IPC
+ip-26-0-164-0:2484745:2485171 [4] NCCL INFO Channel 20/0 : 1[4] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484744:2485180 [3] NCCL INFO Channel 15/0 : 0[3] -> 1[7] via P2P/IPC
+ip-26-0-164-0:2484742:2485174 [1] NCCL INFO Channel 18/0 : 0[1] -> 1[5] via P2P/IPC
+ip-26-0-164-0:2484743:2485177 [2] NCCL INFO Channel 13/0 : 0[2] -> 1[6] via P2P/IPC
+ip-26-0-164-0:2484747:2485178 [6] NCCL INFO Channel 12/0 : 1[6] -> 0[2] via P2P/IPC
+ip-26-0-164-0:2484748:2485181 [7] NCCL INFO Channel 23/0 : 1[7] -> 0[3] via P2P/IPC
+ip-26-0-164-0:2484746:2485175 [5] NCCL INFO Channel 20/0 : 1[5] -> 0[1] via P2P/IPC
+ip-26-0-164-0:2484745:2485171 [4] NCCL INFO Channel 21/0 : 1[4] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484744:2485180 [3] NCCL INFO Channel 16/0 : 0[3] -> 1[7] via P2P/IPC
+ip-26-0-164-0:2484743:2485177 [2] NCCL INFO Channel 14/0 : 0[2] -> 1[6] via P2P/IPC
+ip-26-0-164-0:2484745:2485171 [4] NCCL INFO Channel 22/0 : 1[4] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484747:2485178 [6] NCCL INFO Channel 13/0 : 1[6] -> 0[2] via P2P/IPC
+ip-26-0-164-0:2484744:2485180 [3] NCCL INFO Channel 17/0 : 0[3] -> 1[7] via P2P/IPC
+ip-26-0-164-0:2484743:2485177 [2] NCCL INFO Channel 15/0 : 0[2] -> 1[6] via P2P/IPC
+ip-26-0-164-0:2484745:2485171 [4] NCCL INFO Channel 23/0 : 1[4] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484747:2485178 [6] NCCL INFO Channel 14/0 : 1[6] -> 0[2] via P2P/IPC
+ip-26-0-164-0:2484744:2485180 [3] NCCL INFO Channel 18/0 : 0[3] -> 1[7] via P2P/IPC
+ip-26-0-164-0:2484743:2485177 [2] NCCL INFO Channel 16/0 : 0[2] -> 1[6] via P2P/IPC
+ip-26-0-164-0:2484747:2485178 [6] NCCL INFO Channel 15/0 : 1[6] -> 0[2] via P2P/IPC
+ip-26-0-164-0:2484744:2485180 [3] NCCL INFO Channel 19/0 : 0[3] -> 1[7] via P2P/IPC
+ip-26-0-164-0:2484743:2485177 [2] NCCL INFO Channel 17/0 : 0[2] -> 1[6] via P2P/IPC
+ip-26-0-164-0:2484747:2485178 [6] NCCL INFO Channel 16/0 : 1[6] -> 0[2] via P2P/IPC
+ip-26-0-164-0:2484744:2485180 [3] NCCL INFO Channel 20/0 : 0[3] -> 1[7] via P2P/IPC
+ip-26-0-164-0:2484743:2485177 [2] NCCL INFO Channel 18/0 : 0[2] -> 1[6] via P2P/IPC
+ip-26-0-164-0:2484742:2485174 [1] NCCL INFO Channel 19/0 : 0[1] -> 1[5] via P2P/IPC
+ip-26-0-164-0:2484747:2485178 [6] NCCL INFO Channel 17/0 : 1[6] -> 0[2] via P2P/IPC
+ip-26-0-164-0:2484744:2485180 [3] NCCL INFO Channel 21/0 : 0[3] -> 1[7] via P2P/IPC
+ip-26-0-164-0:2484746:2485175 [5] NCCL INFO Channel 21/0 : 1[5] -> 0[1] via P2P/IPC
+ip-26-0-164-0:2484743:2485177 [2] NCCL INFO Channel 19/0 : 0[2] -> 1[6] via P2P/IPC
+ip-26-0-164-0:2484742:2485174 [1] NCCL INFO Channel 20/0 : 0[1] -> 1[5] via P2P/IPC
+ip-26-0-164-0:2484747:2485178 [6] NCCL INFO Channel 18/0 : 1[6] -> 0[2] via P2P/IPC
+ip-26-0-164-0:2484744:2485180 [3] NCCL INFO Channel 22/0 : 0[3] -> 1[7] via P2P/IPC
+ip-26-0-164-0:2484746:2485175 [5] NCCL INFO Channel 22/0 : 1[5] -> 0[1] via P2P/IPC
+ip-26-0-164-0:2484743:2485177 [2] NCCL INFO Channel 20/0 : 0[2] -> 1[6] via P2P/IPC
+ip-26-0-164-0:2484742:2485174 [1] NCCL INFO Channel 21/0 : 0[1] -> 1[5] via P2P/IPC
+ip-26-0-164-0:2484747:2485178 [6] NCCL INFO Channel 19/0 : 1[6] -> 0[2] via P2P/IPC
+ip-26-0-164-0:2484744:2485180 [3] NCCL INFO Channel 23/0 : 0[3] -> 1[7] via P2P/IPC
+ip-26-0-164-0:2484746:2485175 [5] NCCL INFO Channel 23/0 : 1[5] -> 0[1] via P2P/IPC
+ip-26-0-164-0:2484743:2485177 [2] NCCL INFO Channel 21/0 : 0[2] -> 1[6] via P2P/IPC
+ip-26-0-164-0:2484742:2485174 [1] NCCL INFO Channel 22/0 : 0[1] -> 1[5] via P2P/IPC
+ip-26-0-164-0:2484747:2485178 [6] NCCL INFO Channel 20/0 : 1[6] -> 0[2] via P2P/IPC
+ip-26-0-164-0:2484742:2485174 [1] NCCL INFO Channel 23/0 : 0[1] -> 1[5] via P2P/IPC
+ip-26-0-164-0:2484747:2485178 [6] NCCL INFO Channel 21/0 : 1[6] -> 0[2] via P2P/IPC
+ip-26-0-164-0:2484747:2485178 [6] NCCL INFO Channel 22/0 : 1[6] -> 0[2] via P2P/IPC
+ip-26-0-164-0:2484743:2485177 [2] NCCL INFO Channel 22/0 : 0[2] -> 1[6] via P2P/IPC
+ip-26-0-164-0:2484747:2485178 [6] NCCL INFO Channel 23/0 : 1[6] -> 0[2] via P2P/IPC
+ip-26-0-164-0:2484743:2485177 [2] NCCL INFO Channel 23/0 : 0[2] -> 1[6] via P2P/IPC
+ip-26-0-164-0:2484741:2485170 [0] NCCL INFO Connected all rings
+ip-26-0-164-0:2484741:2485170 [0] NCCL INFO Connected all trees
+ip-26-0-164-0:2484741:2485170 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-164-0:2484741:2485170 [0] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-164-0:2484741:2485170 [0] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-164-0:2484744:2485180 [3] NCCL INFO Connected all rings
+ip-26-0-164-0:2484744:2485180 [3] NCCL INFO Connected all trees
+ip-26-0-164-0:2484744:2485180 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-164-0:2484744:2485180 [3] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-164-0:2484744:2485180 [3] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-164-0:2484745:2485171 [4] NCCL INFO Connected all rings
+ip-26-0-164-0:2484745:2485171 [4] NCCL INFO Connected all trees
+ip-26-0-164-0:2484745:2485171 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-164-0:2484745:2485171 [4] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-164-0:2484745:2485171 [4] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-164-0:2484742:2485174 [1] NCCL INFO Connected all rings
+ip-26-0-164-0:2484742:2485174 [1] NCCL INFO Connected all trees
+ip-26-0-164-0:2484742:2485174 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-164-0:2484742:2485174 [1] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-164-0:2484742:2485174 [1] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-164-0:2484741:2485170 [0] NCCL INFO comm 0xa4dee20 rank 0 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0x91eff57b42c2199c - Init COMPLETE
+ip-26-0-164-0:2484745:2485171 [4] NCCL INFO comm 0xb42f250 rank 1 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0x91eff57b42c2199c - Init COMPLETE
+ip-26-0-164-0:2484747:2485178 [6] NCCL INFO Connected all rings
+ip-26-0-164-0:2484747:2485178 [6] NCCL INFO Connected all trees
+ip-26-0-164-0:2484747:2485178 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-164-0:2484747:2485178 [6] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-164-0:2484747:2485178 [6] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+12/28/2024 00:52:31 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]: Total number of parameters: 8.86G (16896.51MiB)
+12/28/2024 00:52:31 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]: Local number of parameters: 4.92G (9376.28MiB)
+ip-26-0-164-0:2484746:2485175 [5] NCCL INFO Connected all rings
+ip-26-0-164-0:2484746:2485175 [5] NCCL INFO Connected all trees
+ip-26-0-164-0:2484746:2485175 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-164-0:2484746:2485175 [5] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-164-0:2484746:2485175 [5] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+12/28/2024 00:52:31 [INFO|DP=0|PP=1|TP=0|ip-26-0-164-0]: Local number of parameters: 3.94G (7520.23MiB)
+12/28/2024 00:52:31 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]: [After model building] Memory usage: 9376.31MiB. Peak allocated: 9376.31MiB Peak reserved: 12802.00MiB
+12/28/2024 00:52:31 [INFO|DP=0|PP=1|TP=0|ip-26-0-164-0]: [After model building] Memory usage: 7520.25MiB. Peak allocated: 7520.25MiB Peak reserved: 12802.00MiB
+12/28/2024 00:52:31 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]: No checkpoint path provided.
+12/28/2024 00:52:31 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]: Parametrizing model parameters using StandardParametrizator
+ip-26-0-164-0:2484743:2485177 [2] NCCL INFO Connected all rings
+ip-26-0-164-0:2484743:2485177 [2] NCCL INFO Connected all trees
+ip-26-0-164-0:2484743:2485177 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-164-0:2484743:2485177 [2] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-164-0:2484743:2485177 [2] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-164-0:2484748:2485181 [7] NCCL INFO Connected all rings
+ip-26-0-164-0:2484748:2485181 [7] NCCL INFO Connected all trees
+ip-26-0-164-0:2484748:2485181 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-164-0:2484748:2485181 [7] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-164-0:2484748:2485181 [7] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-164-0:2484742:2485174 [1] NCCL INFO comm 0xb9690b0 rank 0 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0x2d884fdd270407ae - Init COMPLETE
+ip-26-0-164-0:2484746:2485175 [5] NCCL INFO comm 0xa5af050 rank 1 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0x2d884fdd270407ae - Init COMPLETE
+ip-26-0-164-0:2484747:2485178 [6] NCCL INFO comm 0xb038510 rank 1 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0x86bf22e4e2bbfb6 - Init COMPLETE
+ip-26-0-164-0:2484743:2485177 [2] NCCL INFO comm 0xa9854c0 rank 0 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0x86bf22e4e2bbfb6 - Init COMPLETE
+ip-26-0-164-0:2484744:2485180 [3] NCCL INFO comm 0xbf10860 rank 0 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0xb5b8c07d17244084 - Init COMPLETE
+ip-26-0-164-0:2484748:2485181 [7] NCCL INFO comm 0xa2ed990 rank 1 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0xb5b8c07d17244084 - Init COMPLETE
+ip-26-0-164-0:2484745:2485201 [4] NCCL INFO Using network Libfabric
+ip-26-0-164-0:2484746:2485202 [5] NCCL INFO Using network Libfabric
+ip-26-0-164-0:2484747:2485204 [6] NCCL INFO Using network Libfabric
+ip-26-0-164-0:2484748:2485203 [7] NCCL INFO Using network Libfabric
+ip-26-0-164-0:2484741:2485206 [0] NCCL INFO Using network Libfabric
+ip-26-0-164-0:2484744:2485207 [3] NCCL INFO Using network Libfabric
+ip-26-0-164-0:2484743:2485209 [2] NCCL INFO Using network Libfabric
+ip-26-0-164-0:2484745:2485201 [4] NCCL INFO comm 0xb605980 rank 0 nranks 4 cudaDev 4 nvmlDev 4 busId 97000 commId 0x341dd36c0358578 - Init START
+ip-26-0-164-0:2484748:2485203 [7] NCCL INFO comm 0xa4c3fc0 rank 3 nranks 4 cudaDev 7 nvmlDev 7 busId ca000 commId 0x341dd36c0358578 - Init START
+ip-26-0-164-0:2484746:2485202 [5] NCCL INFO comm 0xa785800 rank 1 nranks 4 cudaDev 5 nvmlDev 5 busId a8000 commId 0x341dd36c0358578 - Init START
+ip-26-0-164-0:2484747:2485204 [6] NCCL INFO comm 0xb20e820 rank 2 nranks 4 cudaDev 6 nvmlDev 6 busId b9000 commId 0x341dd36c0358578 - Init START
+ip-26-0-164-0:2484742:2485208 [1] NCCL INFO Using network Libfabric
+ip-26-0-164-0:2484747:2485204 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/168
+ip-26-0-164-0:2484748:2485203 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/168
+ip-26-0-164-0:2484745:2485201 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/168
+ip-26-0-164-0:2484746:2485202 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/168
+ip-26-0-164-0:2484743:2485209 [2] NCCL INFO comm 0xa999d70 rank 2 nranks 4 cudaDev 2 nvmlDev 2 busId 75000 commId 0x428ec3d4f02a242d - Init START
+ip-26-0-164-0:2484741:2485206 [0] NCCL INFO comm 0xa4f3140 rank 0 nranks 4 cudaDev 0 nvmlDev 0 busId 53000 commId 0x428ec3d4f02a242d - Init START
+ip-26-0-164-0:2484742:2485208 [1] NCCL INFO comm 0xb97d960 rank 1 nranks 4 cudaDev 1 nvmlDev 1 busId 64000 commId 0x428ec3d4f02a242d - Init START
+ip-26-0-164-0:2484744:2485207 [3] NCCL INFO comm 0xbf25640 rank 3 nranks 4 cudaDev 3 nvmlDev 3 busId 86000 commId 0x428ec3d4f02a242d - Init START
+ip-26-0-164-0:2484743:2485209 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/168
+ip-26-0-164-0:2484742:2485208 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/168
+ip-26-0-164-0:2484744:2485207 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/168
+ip-26-0-164-0:2484741:2485206 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/181
+ip-26-0-164-0:2484741:2485206 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-164-0:2484741:2485206 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-164-0:2484747:2485204 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-164-0:2484747:2485204 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-164-0:2484743:2485209 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-164-0:2484743:2485209 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-164-0:2484748:2485203 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-164-0:2484748:2485203 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-164-0:2484745:2485201 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-164-0:2484745:2485201 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-164-0:2484746:2485202 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-164-0:2484746:2485202 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-164-0:2484744:2485207 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-164-0:2484744:2485207 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-164-0:2484745:2485201 [4] NCCL INFO Channel 00/24 :    0   1   2   3
+ip-26-0-164-0:2484746:2485202 [5] NCCL INFO Trees [0] 2/-1/-1->1->0 [1] 2/-1/-1->1->0 [2] 2/-1/-1->1->0 [3] 2/-1/-1->1->0 [4] 2/-1/-1->1->0 [5] 2/-1/-1->1->0 [6] 2/-1/-1->1->0 [7] 2/-1/-1->1->0 [8] 2/-1/-1->1->0 [9] 2/-1/-1->1->0 [10] 2/-1/-1->1->0 [11] 2/-1/-1->1->0 [12] 2/-1/-1->1->0 [13] 2/-1/-1->1->0 [14] 2/-1/-1->1->0 [15] 2/-1/-1->1->0 [16] 2/-1/-1->1->0 [17] 2/-1/-1->1->0 [18] 2/-1/-1->1->0 [19] 2/-1/-1->1->0 [20] 2/-1/-1->1->0 [21] 2/-1/-1->1->0 [22] 2/-1/-1->1->0 [23] 2/-1/-1->1->0
+ip-26-0-164-0:2484745:2485201 [4] NCCL INFO Channel 01/24 :    0   1   2   3
+ip-26-0-164-0:2484746:2485202 [5] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-164-0:2484745:2485201 [4] NCCL INFO Channel 02/24 :    0   1   2   3
+ip-26-0-164-0:2484745:2485201 [4] NCCL INFO Channel 03/24 :    0   1   2   3
+ip-26-0-164-0:2484745:2485201 [4] NCCL INFO Channel 04/24 :    0   1   2   3
+ip-26-0-164-0:2484745:2485201 [4] NCCL INFO Channel 05/24 :    0   1   2   3
+ip-26-0-164-0:2484745:2485201 [4] NCCL INFO Channel 06/24 :    0   1   2   3
+ip-26-0-164-0:2484745:2485201 [4] NCCL INFO Channel 07/24 :    0   1   2   3
+ip-26-0-164-0:2484748:2485203 [7] NCCL INFO Trees [0] -1/-1/-1->3->2 [1] -1/-1/-1->3->2 [2] -1/-1/-1->3->2 [3] -1/-1/-1->3->2 [4] -1/-1/-1->3->2 [5] -1/-1/-1->3->2 [6] -1/-1/-1->3->2 [7] -1/-1/-1->3->2 [8] -1/-1/-1->3->2 [9] -1/-1/-1->3->2 [10] -1/-1/-1->3->2 [11] -1/-1/-1->3->2 [12] -1/-1/-1->3->2 [13] -1/-1/-1->3->2 [14] -1/-1/-1->3->2 [15] -1/-1/-1->3->2 [16] -1/-1/-1->3->2 [17] -1/-1/-1->3->2 [18] -1/-1/-1->3->2 [19] -1/-1/-1->3->2 [20] -1/-1/-1->3->2 [21] -1/-1/-1->3->2 [22] -1/-1/-1->3->2 [23] -1/-1/-1->3->2
+ip-26-0-164-0:2484745:2485201 [4] NCCL INFO Channel 08/24 :    0   1   2   3
+ip-26-0-164-0:2484747:2485204 [6] NCCL INFO Trees [0] 3/-1/-1->2->1 [1] 3/-1/-1->2->1 [2] 3/-1/-1->2->1 [3] 3/-1/-1->2->1 [4] 3/-1/-1->2->1 [5] 3/-1/-1->2->1 [6] 3/-1/-1->2->1 [7] 3/-1/-1->2->1 [8] 3/-1/-1->2->1 [9] 3/-1/-1->2->1 [10] 3/-1/-1->2->1 [11] 3/-1/-1->2->1 [12] 3/-1/-1->2->1 [13] 3/-1/-1->2->1 [14] 3/-1/-1->2->1 [15] 3/-1/-1->2->1 [16] 3/-1/-1->2->1 [17] 3/-1/-1->2->1 [18] 3/-1/-1->2->1 [19] 3/-1/-1->2->1 [20] 3/-1/-1->2->1 [21] 3/-1/-1->2->1 [22] 3/-1/-1->2->1 [23] 3/-1/-1->2->1
+ip-26-0-164-0:2484748:2485203 [7] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-164-0:2484745:2485201 [4] NCCL INFO Channel 09/24 :    0   1   2   3
+ip-26-0-164-0:2484747:2485204 [6] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-164-0:2484745:2485201 [4] NCCL INFO Channel 10/24 :    0   1   2   3
+ip-26-0-164-0:2484745:2485201 [4] NCCL INFO Channel 11/24 :    0   1   2   3
+ip-26-0-164-0:2484745:2485201 [4] NCCL INFO Channel 12/24 :    0   1   2   3
+ip-26-0-164-0:2484745:2485201 [4] NCCL INFO Channel 13/24 :    0   1   2   3
+ip-26-0-164-0:2484745:2485201 [4] NCCL INFO Channel 14/24 :    0   1   2   3
+ip-26-0-164-0:2484745:2485201 [4] NCCL INFO Channel 15/24 :    0   1   2   3
+ip-26-0-164-0:2484745:2485201 [4] NCCL INFO Channel 16/24 :    0   1   2   3
+ip-26-0-164-0:2484745:2485201 [4] NCCL INFO Channel 17/24 :    0   1   2   3
+ip-26-0-164-0:2484745:2485201 [4] NCCL INFO Channel 18/24 :    0   1   2   3
+ip-26-0-164-0:2484745:2485201 [4] NCCL INFO Channel 19/24 :    0   1   2   3
+ip-26-0-164-0:2484745:2485201 [4] NCCL INFO Channel 20/24 :    0   1   2   3
+ip-26-0-164-0:2484745:2485201 [4] NCCL INFO Channel 21/24 :    0   1   2   3
+ip-26-0-164-0:2484745:2485201 [4] NCCL INFO Channel 22/24 :    0   1   2   3
+ip-26-0-164-0:2484745:2485201 [4] NCCL INFO Channel 23/24 :    0   1   2   3
+ip-26-0-164-0:2484745:2485201 [4] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] 1/-1/-1->0->-1 [7] 1/-1/-1->0->-1 [8] 1/-1/-1->0->-1 [9] 1/-1/-1->0->-1 [10] 1/-1/-1->0->-1 [11] 1/-1/-1->0->-1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] 1/-1/-1->0->-1 [19] 1/-1/-1->0->-1 [20] 1/-1/-1->0->-1 [21] 1/-1/-1->0->-1 [22] 1/-1/-1->0->-1 [23] 1/-1/-1->0->-1
+ip-26-0-164-0:2484745:2485201 [4] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-164-0:2484742:2485208 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-164-0:2484742:2485208 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-164-0:2484741:2485206 [0] NCCL INFO Channel 00/24 :    0   1   2   3
+ip-26-0-164-0:2484742:2485208 [1] NCCL INFO Trees [0] 2/-1/-1->1->0 [1] 2/-1/-1->1->0 [2] 2/-1/-1->1->0 [3] 2/-1/-1->1->0 [4] 2/-1/-1->1->0 [5] 2/-1/-1->1->0 [6] 2/-1/-1->1->0 [7] 2/-1/-1->1->0 [8] 2/-1/-1->1->0 [9] 2/-1/-1->1->0 [10] 2/-1/-1->1->0 [11] 2/-1/-1->1->0 [12] 2/-1/-1->1->0 [13] 2/-1/-1->1->0 [14] 2/-1/-1->1->0 [15] 2/-1/-1->1->0 [16] 2/-1/-1->1->0 [17] 2/-1/-1->1->0 [18] 2/-1/-1->1->0 [19] 2/-1/-1->1->0 [20] 2/-1/-1->1->0 [21] 2/-1/-1->1->0 [22] 2/-1/-1->1->0 [23] 2/-1/-1->1->0
+ip-26-0-164-0:2484741:2485206 [0] NCCL INFO Channel 01/24 :    0   1   2   3
+ip-26-0-164-0:2484743:2485209 [2] NCCL INFO Trees [0] 3/-1/-1->2->1 [1] 3/-1/-1->2->1 [2] 3/-1/-1->2->1 [3] 3/-1/-1->2->1 [4] 3/-1/-1->2->1 [5] 3/-1/-1->2->1 [6] 3/-1/-1->2->1 [7] 3/-1/-1->2->1 [8] 3/-1/-1->2->1 [9] 3/-1/-1->2->1 [10] 3/-1/-1->2->1 [11] 3/-1/-1->2->1 [12] 3/-1/-1->2->1 [13] 3/-1/-1->2->1 [14] 3/-1/-1->2->1 [15] 3/-1/-1->2->1 [16] 3/-1/-1->2->1 [17] 3/-1/-1->2->1 [18] 3/-1/-1->2->1 [19] 3/-1/-1->2->1 [20] 3/-1/-1->2->1 [21] 3/-1/-1->2->1 [22] 3/-1/-1->2->1 [23] 3/-1/-1->2->1
+ip-26-0-164-0:2484742:2485208 [1] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-164-0:2484741:2485206 [0] NCCL INFO Channel 02/24 :    0   1   2   3
+ip-26-0-164-0:2484743:2485209 [2] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-164-0:2484741:2485206 [0] NCCL INFO Channel 03/24 :    0   1   2   3
+ip-26-0-164-0:2484741:2485206 [0] NCCL INFO Channel 04/24 :    0   1   2   3
+ip-26-0-164-0:2484741:2485206 [0] NCCL INFO Channel 05/24 :    0   1   2   3
+ip-26-0-164-0:2484741:2485206 [0] NCCL INFO Channel 06/24 :    0   1   2   3
+ip-26-0-164-0:2484744:2485207 [3] NCCL INFO Trees [0] -1/-1/-1->3->2 [1] -1/-1/-1->3->2 [2] -1/-1/-1->3->2 [3] -1/-1/-1->3->2 [4] -1/-1/-1->3->2 [5] -1/-1/-1->3->2 [6] -1/-1/-1->3->2 [7] -1/-1/-1->3->2 [8] -1/-1/-1->3->2 [9] -1/-1/-1->3->2 [10] -1/-1/-1->3->2 [11] -1/-1/-1->3->2 [12] -1/-1/-1->3->2 [13] -1/-1/-1->3->2 [14] -1/-1/-1->3->2 [15] -1/-1/-1->3->2 [16] -1/-1/-1->3->2 [17] -1/-1/-1->3->2 [18] -1/-1/-1->3->2 [19] -1/-1/-1->3->2 [20] -1/-1/-1->3->2 [21] -1/-1/-1->3->2 [22] -1/-1/-1->3->2 [23] -1/-1/-1->3->2
+ip-26-0-164-0:2484741:2485206 [0] NCCL INFO Channel 07/24 :    0   1   2   3
+ip-26-0-164-0:2484741:2485206 [0] NCCL INFO Channel 08/24 :    0   1   2   3
+ip-26-0-164-0:2484744:2485207 [3] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-164-0:2484741:2485206 [0] NCCL INFO Channel 09/24 :    0   1   2   3
+ip-26-0-164-0:2484741:2485206 [0] NCCL INFO Channel 10/24 :    0   1   2   3
+ip-26-0-164-0:2484741:2485206 [0] NCCL INFO Channel 11/24 :    0   1   2   3
+ip-26-0-164-0:2484741:2485206 [0] NCCL INFO Channel 12/24 :    0   1   2   3
+ip-26-0-164-0:2484741:2485206 [0] NCCL INFO Channel 13/24 :    0   1   2   3
+ip-26-0-164-0:2484741:2485206 [0] NCCL INFO Channel 14/24 :    0   1   2   3
+ip-26-0-164-0:2484741:2485206 [0] NCCL INFO Channel 15/24 :    0   1   2   3
+ip-26-0-164-0:2484741:2485206 [0] NCCL INFO Channel 16/24 :    0   1   2   3
+ip-26-0-164-0:2484741:2485206 [0] NCCL INFO Channel 17/24 :    0   1   2   3
+ip-26-0-164-0:2484741:2485206 [0] NCCL INFO Channel 18/24 :    0   1   2   3
+ip-26-0-164-0:2484741:2485206 [0] NCCL INFO Channel 19/24 :    0   1   2   3
+ip-26-0-164-0:2484741:2485206 [0] NCCL INFO Channel 20/24 :    0   1   2   3
+ip-26-0-164-0:2484741:2485206 [0] NCCL INFO Channel 21/24 :    0   1   2   3
+ip-26-0-164-0:2484741:2485206 [0] NCCL INFO Channel 22/24 :    0   1   2   3
+ip-26-0-164-0:2484741:2485206 [0] NCCL INFO Channel 23/24 :    0   1   2   3
+ip-26-0-164-0:2484741:2485206 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] 1/-1/-1->0->-1 [7] 1/-1/-1->0->-1 [8] 1/-1/-1->0->-1 [9] 1/-1/-1->0->-1 [10] 1/-1/-1->0->-1 [11] 1/-1/-1->0->-1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] 1/-1/-1->0->-1 [19] 1/-1/-1->0->-1 [20] 1/-1/-1->0->-1 [21] 1/-1/-1->0->-1 [22] 1/-1/-1->0->-1 [23] 1/-1/-1->0->-1
+ip-26-0-164-0:2484741:2485206 [0] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-164-0:2484746:2485202 [5] NCCL INFO Channel 00/0 : 1[5] -> 2[6] via P2P/IPC
+ip-26-0-164-0:2484747:2485204 [6] NCCL INFO Channel 00/0 : 2[6] -> 3[7] via P2P/IPC
+ip-26-0-164-0:2484748:2485203 [7] NCCL INFO Channel 00/0 : 3[7] -> 0[4] via P2P/IPC
+ip-26-0-164-0:2484746:2485202 [5] NCCL INFO Channel 01/0 : 1[5] -> 2[6] via P2P/IPC
+ip-26-0-164-0:2484745:2485201 [4] NCCL INFO Channel 00/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-164-0:2484747:2485204 [6] NCCL INFO Channel 01/0 : 2[6] -> 3[7] via P2P/IPC
+ip-26-0-164-0:2484748:2485203 [7] NCCL INFO Channel 01/0 : 3[7] -> 0[4] via P2P/IPC
+ip-26-0-164-0:2484746:2485202 [5] NCCL INFO Channel 02/0 : 1[5] -> 2[6] via P2P/IPC
+ip-26-0-164-0:2484745:2485201 [4] NCCL INFO Channel 01/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-164-0:2484747:2485204 [6] NCCL INFO Channel 02/0 : 2[6] -> 3[7] via P2P/IPC
+ip-26-0-164-0:2484748:2485203 [7] NCCL INFO Channel 02/0 : 3[7] -> 0[4] via P2P/IPC
+ip-26-0-164-0:2484745:2485201 [4] NCCL INFO Channel 02/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-164-0:2484747:2485204 [6] NCCL INFO Channel 03/0 : 2[6] -> 3[7] via P2P/IPC
+ip-26-0-164-0:2484746:2485202 [5] NCCL INFO Channel 03/0 : 1[5] -> 2[6] via P2P/IPC
+ip-26-0-164-0:2484748:2485203 [7] NCCL INFO Channel 03/0 : 3[7] -> 0[4] via P2P/IPC
+ip-26-0-164-0:2484745:2485201 [4] NCCL INFO Channel 03/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-164-0:2484747:2485204 [6] NCCL INFO Channel 04/0 : 2[6] -> 3[7] via P2P/IPC
+ip-26-0-164-0:2484746:2485202 [5] NCCL INFO Channel 04/0 : 1[5] -> 2[6] via P2P/IPC
+ip-26-0-164-0:2484748:2485203 [7] NCCL INFO Channel 04/0 : 3[7] -> 0[4] via P2P/IPC
+ip-26-0-164-0:2484745:2485201 [4] NCCL INFO Channel 04/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-164-0:2484747:2485204 [6] NCCL INFO Channel 05/0 : 2[6] -> 3[7] via P2P/IPC
+ip-26-0-164-0:2484746:2485202 [5] NCCL INFO Channel 05/0 : 1[5] -> 2[6] via P2P/IPC
+ip-26-0-164-0:2484748:2485203 [7] NCCL INFO Channel 05/0 : 3[7] -> 0[4] via P2P/IPC
+ip-26-0-164-0:2484745:2485201 [4] NCCL INFO Channel 05/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-164-0:2484747:2485204 [6] NCCL INFO Channel 06/0 : 2[6] -> 3[7] via P2P/IPC
+ip-26-0-164-0:2484746:2485202 [5] NCCL INFO Channel 06/0 : 1[5] -> 2[6] via P2P/IPC
+ip-26-0-164-0:2484748:2485203 [7] NCCL INFO Channel 06/0 : 3[7] -> 0[4] via P2P/IPC
+ip-26-0-164-0:2484747:2485204 [6] NCCL INFO Channel 07/0 : 2[6] -> 3[7] via P2P/IPC
+ip-26-0-164-0:2484745:2485201 [4] NCCL INFO Channel 06/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-164-0:2484746:2485202 [5] NCCL INFO Channel 07/0 : 1[5] -> 2[6] via P2P/IPC
+ip-26-0-164-0:2484748:2485203 [7] NCCL INFO Channel 07/0 : 3[7] -> 0[4] via P2P/IPC
+ip-26-0-164-0:2484745:2485201 [4] NCCL INFO Channel 07/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-164-0:2484747:2485204 [6] NCCL INFO Channel 08/0 : 2[6] -> 3[7] via P2P/IPC
+ip-26-0-164-0:2484746:2485202 [5] NCCL INFO Channel 08/0 : 1[5] -> 2[6] via P2P/IPC
+ip-26-0-164-0:2484748:2485203 [7] NCCL INFO Channel 08/0 : 3[7] -> 0[4] via P2P/IPC
+ip-26-0-164-0:2484745:2485201 [4] NCCL INFO Channel 08/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-164-0:2484746:2485202 [5] NCCL INFO Channel 09/0 : 1[5] -> 2[6] via P2P/IPC
+ip-26-0-164-0:2484747:2485204 [6] NCCL INFO Channel 09/0 : 2[6] -> 3[7] via P2P/IPC
+ip-26-0-164-0:2484748:2485203 [7] NCCL INFO Channel 09/0 : 3[7] -> 0[4] via P2P/IPC
+ip-26-0-164-0:2484745:2485201 [4] NCCL INFO Channel 09/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-164-0:2484746:2485202 [5] NCCL INFO Channel 10/0 : 1[5] -> 2[6] via P2P/IPC
+ip-26-0-164-0:2484747:2485204 [6] NCCL INFO Channel 10/0 : 2[6] -> 3[7] via P2P/IPC
+ip-26-0-164-0:2484748:2485203 [7] NCCL INFO Channel 10/0 : 3[7] -> 0[4] via P2P/IPC
+ip-26-0-164-0:2484745:2485201 [4] NCCL INFO Channel 10/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-164-0:2484746:2485202 [5] NCCL INFO Channel 11/0 : 1[5] -> 2[6] via P2P/IPC
+ip-26-0-164-0:2484747:2485204 [6] NCCL INFO Channel 11/0 : 2[6] -> 3[7] via P2P/IPC
+ip-26-0-164-0:2484748:2485203 [7] NCCL INFO Channel 11/0 : 3[7] -> 0[4] via P2P/IPC
+ip-26-0-164-0:2484745:2485201 [4] NCCL INFO Channel 11/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-164-0:2484746:2485202 [5] NCCL INFO Channel 12/0 : 1[5] -> 2[6] via P2P/IPC
+ip-26-0-164-0:2484747:2485204 [6] NCCL INFO Channel 12/0 : 2[6] -> 3[7] via P2P/IPC
+ip-26-0-164-0:2484748:2485203 [7] NCCL INFO Channel 12/0 : 3[7] -> 0[4] via P2P/IPC
+ip-26-0-164-0:2484746:2485202 [5] NCCL INFO Channel 13/0 : 1[5] -> 2[6] via P2P/IPC
+ip-26-0-164-0:2484745:2485201 [4] NCCL INFO Channel 12/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-164-0:2484747:2485204 [6] NCCL INFO Channel 13/0 : 2[6] -> 3[7] via P2P/IPC
+ip-26-0-164-0:2484748:2485203 [7] NCCL INFO Channel 13/0 : 3[7] -> 0[4] via P2P/IPC
+ip-26-0-164-0:2484745:2485201 [4] NCCL INFO Channel 13/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-164-0:2484746:2485202 [5] NCCL INFO Channel 14/0 : 1[5] -> 2[6] via P2P/IPC
+ip-26-0-164-0:2484747:2485204 [6] NCCL INFO Channel 14/0 : 2[6] -> 3[7] via P2P/IPC
+ip-26-0-164-0:2484748:2485203 [7] NCCL INFO Channel 14/0 : 3[7] -> 0[4] via P2P/IPC
+ip-26-0-164-0:2484745:2485201 [4] NCCL INFO Channel 14/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-164-0:2484746:2485202 [5] NCCL INFO Channel 15/0 : 1[5] -> 2[6] via P2P/IPC
+ip-26-0-164-0:2484747:2485204 [6] NCCL INFO Channel 15/0 : 2[6] -> 3[7] via P2P/IPC
+ip-26-0-164-0:2484748:2485203 [7] NCCL INFO Channel 15/0 : 3[7] -> 0[4] via P2P/IPC
+ip-26-0-164-0:2484745:2485201 [4] NCCL INFO Channel 15/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-164-0:2484746:2485202 [5] NCCL INFO Channel 16/0 : 1[5] -> 2[6] via P2P/IPC
+ip-26-0-164-0:2484747:2485204 [6] NCCL INFO Channel 16/0 : 2[6] -> 3[7] via P2P/IPC
+ip-26-0-164-0:2484748:2485203 [7] NCCL INFO Channel 16/0 : 3[7] -> 0[4] via P2P/IPC
+ip-26-0-164-0:2484745:2485201 [4] NCCL INFO Channel 16/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-164-0:2484746:2485202 [5] NCCL INFO Channel 17/0 : 1[5] -> 2[6] via P2P/IPC
+ip-26-0-164-0:2484747:2485204 [6] NCCL INFO Channel 17/0 : 2[6] -> 3[7] via P2P/IPC
+ip-26-0-164-0:2484748:2485203 [7] NCCL INFO Channel 17/0 : 3[7] -> 0[4] via P2P/IPC
+ip-26-0-164-0:2484746:2485202 [5] NCCL INFO Channel 18/0 : 1[5] -> 2[6] via P2P/IPC
+ip-26-0-164-0:2484745:2485201 [4] NCCL INFO Channel 17/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-164-0:2484747:2485204 [6] NCCL INFO Channel 18/0 : 2[6] -> 3[7] via P2P/IPC
+ip-26-0-164-0:2484748:2485203 [7] NCCL INFO Channel 18/0 : 3[7] -> 0[4] via P2P/IPC
+ip-26-0-164-0:2484746:2485202 [5] NCCL INFO Channel 19/0 : 1[5] -> 2[6] via P2P/IPC
+ip-26-0-164-0:2484745:2485201 [4] NCCL INFO Channel 18/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-164-0:2484747:2485204 [6] NCCL INFO Channel 19/0 : 2[6] -> 3[7] via P2P/IPC
+ip-26-0-164-0:2484746:2485202 [5] NCCL INFO Channel 20/0 : 1[5] -> 2[6] via P2P/IPC
+ip-26-0-164-0:2484745:2485201 [4] NCCL INFO Channel 19/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-164-0:2484743:2485209 [2] NCCL INFO Channel 00/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484747:2485204 [6] NCCL INFO Channel 20/0 : 2[6] -> 3[7] via P2P/IPC
+ip-26-0-164-0:2484746:2485202 [5] NCCL INFO Channel 21/0 : 1[5] -> 2[6] via P2P/IPC
+ip-26-0-164-0:2484745:2485201 [4] NCCL INFO Channel 20/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-164-0:2484743:2485209 [2] NCCL INFO Channel 01/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484747:2485204 [6] NCCL INFO Channel 21/0 : 2[6] -> 3[7] via P2P/IPC
+ip-26-0-164-0:2484746:2485202 [5] NCCL INFO Channel 22/0 : 1[5] -> 2[6] via P2P/IPC
+ip-26-0-164-0:2484745:2485201 [4] NCCL INFO Channel 21/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-164-0:2484743:2485209 [2] NCCL INFO Channel 02/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484747:2485204 [6] NCCL INFO Channel 22/0 : 2[6] -> 3[7] via P2P/IPC
+ip-26-0-164-0:2484748:2485203 [7] NCCL INFO Channel 19/0 : 3[7] -> 0[4] via P2P/IPC
+ip-26-0-164-0:2484746:2485202 [5] NCCL INFO Channel 23/0 : 1[5] -> 2[6] via P2P/IPC
+ip-26-0-164-0:2484745:2485201 [4] NCCL INFO Channel 22/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-164-0:2484743:2485209 [2] NCCL INFO Channel 03/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484742:2485208 [1] NCCL INFO Channel 00/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484747:2485204 [6] NCCL INFO Channel 23/0 : 2[6] -> 3[7] via P2P/IPC
+ip-26-0-164-0:2484748:2485203 [7] NCCL INFO Channel 20/0 : 3[7] -> 0[4] via P2P/IPC
+ip-26-0-164-0:2484741:2485206 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484744:2485207 [3] NCCL INFO Channel 00/0 : 3[3] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484745:2485201 [4] NCCL INFO Channel 23/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-164-0:2484743:2485209 [2] NCCL INFO Channel 04/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484742:2485208 [1] NCCL INFO Channel 01/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484748:2485203 [7] NCCL INFO Channel 21/0 : 3[7] -> 0[4] via P2P/IPC
+ip-26-0-164-0:2484744:2485207 [3] NCCL INFO Channel 01/0 : 3[3] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484741:2485206 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484742:2485208 [1] NCCL INFO Channel 02/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484743:2485209 [2] NCCL INFO Channel 05/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484744:2485207 [3] NCCL INFO Channel 02/0 : 3[3] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484748:2485203 [7] NCCL INFO Channel 22/0 : 3[7] -> 0[4] via P2P/IPC
+ip-26-0-164-0:2484741:2485206 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484742:2485208 [1] NCCL INFO Channel 03/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484741:2485206 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484744:2485207 [3] NCCL INFO Channel 03/0 : 3[3] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484743:2485209 [2] NCCL INFO Channel 06/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484741:2485206 [0] NCCL INFO Channel 04/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484744:2485207 [3] NCCL INFO Channel 04/0 : 3[3] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484743:2485209 [2] NCCL INFO Channel 07/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484741:2485206 [0] NCCL INFO Channel 05/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484744:2485207 [3] NCCL INFO Channel 05/0 : 3[3] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484743:2485209 [2] NCCL INFO Channel 08/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484748:2485203 [7] NCCL INFO Channel 23/0 : 3[7] -> 0[4] via P2P/IPC
+ip-26-0-164-0:2484742:2485208 [1] NCCL INFO Channel 04/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484741:2485206 [0] NCCL INFO Channel 06/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484744:2485207 [3] NCCL INFO Channel 06/0 : 3[3] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484743:2485209 [2] NCCL INFO Channel 09/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484742:2485208 [1] NCCL INFO Channel 05/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484741:2485206 [0] NCCL INFO Channel 07/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484744:2485207 [3] NCCL INFO Channel 07/0 : 3[3] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484743:2485209 [2] NCCL INFO Channel 10/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484743:2485209 [2] NCCL INFO Channel 11/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484744:2485207 [3] NCCL INFO Channel 08/0 : 3[3] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484743:2485209 [2] NCCL INFO Channel 12/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484743:2485209 [2] NCCL INFO Channel 13/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484742:2485208 [1] NCCL INFO Channel 06/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484744:2485207 [3] NCCL INFO Channel 09/0 : 3[3] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484742:2485208 [1] NCCL INFO Channel 07/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484744:2485207 [3] NCCL INFO Channel 10/0 : 3[3] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484741:2485206 [0] NCCL INFO Channel 08/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484743:2485209 [2] NCCL INFO Channel 14/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484743:2485209 [2] NCCL INFO Channel 15/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484742:2485208 [1] NCCL INFO Channel 08/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484744:2485207 [3] NCCL INFO Channel 11/0 : 3[3] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484741:2485206 [0] NCCL INFO Channel 09/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484743:2485209 [2] NCCL INFO Channel 16/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484742:2485208 [1] NCCL INFO Channel 09/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484744:2485207 [3] NCCL INFO Channel 12/0 : 3[3] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484741:2485206 [0] NCCL INFO Channel 10/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484743:2485209 [2] NCCL INFO Channel 17/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484742:2485208 [1] NCCL INFO Channel 10/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484744:2485207 [3] NCCL INFO Channel 13/0 : 3[3] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484741:2485206 [0] NCCL INFO Channel 11/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484743:2485209 [2] NCCL INFO Channel 18/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484742:2485208 [1] NCCL INFO Channel 11/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484744:2485207 [3] NCCL INFO Channel 14/0 : 3[3] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484741:2485206 [0] NCCL INFO Channel 12/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484741:2485206 [0] NCCL INFO Channel 13/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484744:2485207 [3] NCCL INFO Channel 15/0 : 3[3] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484741:2485206 [0] NCCL INFO Channel 14/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484744:2485207 [3] NCCL INFO Channel 16/0 : 3[3] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484742:2485208 [1] NCCL INFO Channel 12/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484743:2485209 [2] NCCL INFO Channel 19/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484741:2485206 [0] NCCL INFO Channel 15/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484744:2485207 [3] NCCL INFO Channel 17/0 : 3[3] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484742:2485208 [1] NCCL INFO Channel 13/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484743:2485209 [2] NCCL INFO Channel 20/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484741:2485206 [0] NCCL INFO Channel 16/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484744:2485207 [3] NCCL INFO Channel 18/0 : 3[3] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484742:2485208 [1] NCCL INFO Channel 14/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484743:2485209 [2] NCCL INFO Channel 21/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484742:2485208 [1] NCCL INFO Channel 15/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484741:2485206 [0] NCCL INFO Channel 17/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484744:2485207 [3] NCCL INFO Channel 19/0 : 3[3] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484742:2485208 [1] NCCL INFO Channel 16/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484741:2485206 [0] NCCL INFO Channel 18/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484744:2485207 [3] NCCL INFO Channel 20/0 : 3[3] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484743:2485209 [2] NCCL INFO Channel 22/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484742:2485208 [1] NCCL INFO Channel 17/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484741:2485206 [0] NCCL INFO Channel 19/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484744:2485207 [3] NCCL INFO Channel 21/0 : 3[3] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484743:2485209 [2] NCCL INFO Channel 23/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-164-0:2484742:2485208 [1] NCCL INFO Channel 18/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484744:2485207 [3] NCCL INFO Channel 22/0 : 3[3] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484741:2485206 [0] NCCL INFO Channel 20/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484744:2485207 [3] NCCL INFO Channel 23/0 : 3[3] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484742:2485208 [1] NCCL INFO Channel 19/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484741:2485206 [0] NCCL INFO Channel 21/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484742:2485208 [1] NCCL INFO Channel 20/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484741:2485206 [0] NCCL INFO Channel 22/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484742:2485208 [1] NCCL INFO Channel 21/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484741:2485206 [0] NCCL INFO Channel 23/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484742:2485208 [1] NCCL INFO Channel 22/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484742:2485208 [1] NCCL INFO Channel 23/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484748:2485203 [7] NCCL INFO Connected all rings
+ip-26-0-164-0:2484748:2485203 [7] NCCL INFO Channel 00/0 : 3[7] -> 2[6] via P2P/IPC
+ip-26-0-164-0:2484748:2485203 [7] NCCL INFO Channel 01/0 : 3[7] -> 2[6] via P2P/IPC
+ip-26-0-164-0:2484746:2485202 [5] NCCL INFO Connected all rings
+ip-26-0-164-0:2484748:2485203 [7] NCCL INFO Channel 02/0 : 3[7] -> 2[6] via P2P/IPC
+ip-26-0-164-0:2484748:2485203 [7] NCCL INFO Channel 03/0 : 3[7] -> 2[6] via P2P/IPC
+ip-26-0-164-0:2484748:2485203 [7] NCCL INFO Channel 04/0 : 3[7] -> 2[6] via P2P/IPC
+ip-26-0-164-0:2484745:2485201 [4] NCCL INFO Connected all rings
+ip-26-0-164-0:2484747:2485204 [6] NCCL INFO Connected all rings
+ip-26-0-164-0:2484748:2485203 [7] NCCL INFO Channel 05/0 : 3[7] -> 2[6] via P2P/IPC
+ip-26-0-164-0:2484748:2485203 [7] NCCL INFO Channel 06/0 : 3[7] -> 2[6] via P2P/IPC
+ip-26-0-164-0:2484748:2485203 [7] NCCL INFO Channel 07/0 : 3[7] -> 2[6] via P2P/IPC
+ip-26-0-164-0:2484748:2485203 [7] NCCL INFO Channel 08/0 : 3[7] -> 2[6] via P2P/IPC
+ip-26-0-164-0:2484748:2485203 [7] NCCL INFO Channel 09/0 : 3[7] -> 2[6] via P2P/IPC
+ip-26-0-164-0:2484748:2485203 [7] NCCL INFO Channel 10/0 : 3[7] -> 2[6] via P2P/IPC
+ip-26-0-164-0:2484748:2485203 [7] NCCL INFO Channel 11/0 : 3[7] -> 2[6] via P2P/IPC
+ip-26-0-164-0:2484748:2485203 [7] NCCL INFO Channel 12/0 : 3[7] -> 2[6] via P2P/IPC
+ip-26-0-164-0:2484748:2485203 [7] NCCL INFO Channel 13/0 : 3[7] -> 2[6] via P2P/IPC
+ip-26-0-164-0:2484748:2485203 [7] NCCL INFO Channel 14/0 : 3[7] -> 2[6] via P2P/IPC
+ip-26-0-164-0:2484748:2485203 [7] NCCL INFO Channel 15/0 : 3[7] -> 2[6] via P2P/IPC
+ip-26-0-164-0:2484748:2485203 [7] NCCL INFO Channel 16/0 : 3[7] -> 2[6] via P2P/IPC
+ip-26-0-164-0:2484748:2485203 [7] NCCL INFO Channel 17/0 : 3[7] -> 2[6] via P2P/IPC
+ip-26-0-164-0:2484748:2485203 [7] NCCL INFO Channel 18/0 : 3[7] -> 2[6] via P2P/IPC
+ip-26-0-164-0:2484748:2485203 [7] NCCL INFO Channel 19/0 : 3[7] -> 2[6] via P2P/IPC
+ip-26-0-164-0:2484748:2485203 [7] NCCL INFO Channel 20/0 : 3[7] -> 2[6] via P2P/IPC
+ip-26-0-164-0:2484748:2485203 [7] NCCL INFO Channel 21/0 : 3[7] -> 2[6] via P2P/IPC
+ip-26-0-164-0:2484748:2485203 [7] NCCL INFO Channel 22/0 : 3[7] -> 2[6] via P2P/IPC
+ip-26-0-164-0:2484742:2485208 [1] NCCL INFO Connected all rings
+ip-26-0-164-0:2484743:2485209 [2] NCCL INFO Connected all rings
+ip-26-0-164-0:2484748:2485203 [7] NCCL INFO Channel 23/0 : 3[7] -> 2[6] via P2P/IPC
+ip-26-0-164-0:2484741:2485206 [0] NCCL INFO Connected all rings
+ip-26-0-164-0:2484744:2485207 [3] NCCL INFO Connected all rings
+ip-26-0-164-0:2484744:2485207 [3] NCCL INFO Channel 00/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484744:2485207 [3] NCCL INFO Channel 01/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484744:2485207 [3] NCCL INFO Channel 02/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484744:2485207 [3] NCCL INFO Channel 03/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484744:2485207 [3] NCCL INFO Channel 04/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484744:2485207 [3] NCCL INFO Channel 05/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484744:2485207 [3] NCCL INFO Channel 06/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484744:2485207 [3] NCCL INFO Channel 07/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484746:2485202 [5] NCCL INFO Channel 00/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-164-0:2484747:2485204 [6] NCCL INFO Channel 00/0 : 2[6] -> 1[5] via P2P/IPC
+ip-26-0-164-0:2484746:2485202 [5] NCCL INFO Channel 01/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-164-0:2484746:2485202 [5] NCCL INFO Channel 02/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-164-0:2484744:2485207 [3] NCCL INFO Channel 08/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484746:2485202 [5] NCCL INFO Channel 03/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-164-0:2484744:2485207 [3] NCCL INFO Channel 09/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484746:2485202 [5] NCCL INFO Channel 04/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-164-0:2484744:2485207 [3] NCCL INFO Channel 10/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484746:2485202 [5] NCCL INFO Channel 05/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-164-0:2484747:2485204 [6] NCCL INFO Channel 01/0 : 2[6] -> 1[5] via P2P/IPC
+ip-26-0-164-0:2484744:2485207 [3] NCCL INFO Channel 11/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484746:2485202 [5] NCCL INFO Channel 06/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-164-0:2484747:2485204 [6] NCCL INFO Channel 02/0 : 2[6] -> 1[5] via P2P/IPC
+ip-26-0-164-0:2484744:2485207 [3] NCCL INFO Channel 12/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484746:2485202 [5] NCCL INFO Channel 07/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-164-0:2484747:2485204 [6] NCCL INFO Channel 03/0 : 2[6] -> 1[5] via P2P/IPC
+ip-26-0-164-0:2484744:2485207 [3] NCCL INFO Channel 13/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484746:2485202 [5] NCCL INFO Channel 08/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-164-0:2484747:2485204 [6] NCCL INFO Channel 04/0 : 2[6] -> 1[5] via P2P/IPC
+ip-26-0-164-0:2484744:2485207 [3] NCCL INFO Channel 14/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484746:2485202 [5] NCCL INFO Channel 09/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-164-0:2484747:2485204 [6] NCCL INFO Channel 05/0 : 2[6] -> 1[5] via P2P/IPC
+ip-26-0-164-0:2484744:2485207 [3] NCCL INFO Channel 15/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484746:2485202 [5] NCCL INFO Channel 10/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-164-0:2484747:2485204 [6] NCCL INFO Channel 06/0 : 2[6] -> 1[5] via P2P/IPC
+ip-26-0-164-0:2484744:2485207 [3] NCCL INFO Channel 16/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484746:2485202 [5] NCCL INFO Channel 11/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-164-0:2484747:2485204 [6] NCCL INFO Channel 07/0 : 2[6] -> 1[5] via P2P/IPC
+ip-26-0-164-0:2484744:2485207 [3] NCCL INFO Channel 17/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484746:2485202 [5] NCCL INFO Channel 12/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-164-0:2484747:2485204 [6] NCCL INFO Channel 08/0 : 2[6] -> 1[5] via P2P/IPC
+ip-26-0-164-0:2484743:2485209 [2] NCCL INFO Channel 00/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484744:2485207 [3] NCCL INFO Channel 18/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484746:2485202 [5] NCCL INFO Channel 13/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-164-0:2484747:2485204 [6] NCCL INFO Channel 09/0 : 2[6] -> 1[5] via P2P/IPC
+ip-26-0-164-0:2484743:2485209 [2] NCCL INFO Channel 01/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484744:2485207 [3] NCCL INFO Channel 19/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484746:2485202 [5] NCCL INFO Channel 14/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-164-0:2484743:2485209 [2] NCCL INFO Channel 02/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484747:2485204 [6] NCCL INFO Channel 10/0 : 2[6] -> 1[5] via P2P/IPC
+ip-26-0-164-0:2484744:2485207 [3] NCCL INFO Channel 20/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484746:2485202 [5] NCCL INFO Channel 15/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-164-0:2484747:2485204 [6] NCCL INFO Channel 11/0 : 2[6] -> 1[5] via P2P/IPC
+ip-26-0-164-0:2484743:2485209 [2] NCCL INFO Channel 03/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484744:2485207 [3] NCCL INFO Channel 21/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484746:2485202 [5] NCCL INFO Channel 16/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-164-0:2484747:2485204 [6] NCCL INFO Channel 12/0 : 2[6] -> 1[5] via P2P/IPC
+ip-26-0-164-0:2484743:2485209 [2] NCCL INFO Channel 04/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484744:2485207 [3] NCCL INFO Channel 22/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484742:2485208 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484746:2485202 [5] NCCL INFO Channel 17/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-164-0:2484743:2485209 [2] NCCL INFO Channel 05/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484747:2485204 [6] NCCL INFO Channel 13/0 : 2[6] -> 1[5] via P2P/IPC
+ip-26-0-164-0:2484744:2485207 [3] NCCL INFO Channel 23/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-164-0:2484742:2485208 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484746:2485202 [5] NCCL INFO Channel 18/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-164-0:2484747:2485204 [6] NCCL INFO Channel 14/0 : 2[6] -> 1[5] via P2P/IPC
+ip-26-0-164-0:2484743:2485209 [2] NCCL INFO Channel 06/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484742:2485208 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484746:2485202 [5] NCCL INFO Channel 19/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-164-0:2484747:2485204 [6] NCCL INFO Channel 15/0 : 2[6] -> 1[5] via P2P/IPC
+ip-26-0-164-0:2484742:2485208 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484743:2485209 [2] NCCL INFO Channel 07/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484746:2485202 [5] NCCL INFO Channel 20/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-164-0:2484747:2485204 [6] NCCL INFO Channel 16/0 : 2[6] -> 1[5] via P2P/IPC
+ip-26-0-164-0:2484742:2485208 [1] NCCL INFO Channel 04/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484743:2485209 [2] NCCL INFO Channel 08/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484746:2485202 [5] NCCL INFO Channel 21/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-164-0:2484747:2485204 [6] NCCL INFO Channel 17/0 : 2[6] -> 1[5] via P2P/IPC
+ip-26-0-164-0:2484742:2485208 [1] NCCL INFO Channel 05/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484743:2485209 [2] NCCL INFO Channel 09/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484746:2485202 [5] NCCL INFO Channel 22/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-164-0:2484747:2485204 [6] NCCL INFO Channel 18/0 : 2[6] -> 1[5] via P2P/IPC
+ip-26-0-164-0:2484742:2485208 [1] NCCL INFO Channel 06/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484743:2485209 [2] NCCL INFO Channel 10/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484746:2485202 [5] NCCL INFO Channel 23/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-164-0:2484747:2485204 [6] NCCL INFO Channel 19/0 : 2[6] -> 1[5] via P2P/IPC
+ip-26-0-164-0:2484742:2485208 [1] NCCL INFO Channel 07/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484743:2485209 [2] NCCL INFO Channel 11/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484742:2485208 [1] NCCL INFO Channel 08/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484747:2485204 [6] NCCL INFO Channel 20/0 : 2[6] -> 1[5] via P2P/IPC
+ip-26-0-164-0:2484743:2485209 [2] NCCL INFO Channel 12/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484742:2485208 [1] NCCL INFO Channel 09/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484747:2485204 [6] NCCL INFO Channel 21/0 : 2[6] -> 1[5] via P2P/IPC
+ip-26-0-164-0:2484743:2485209 [2] NCCL INFO Channel 13/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484742:2485208 [1] NCCL INFO Channel 10/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484747:2485204 [6] NCCL INFO Channel 22/0 : 2[6] -> 1[5] via P2P/IPC
+ip-26-0-164-0:2484743:2485209 [2] NCCL INFO Channel 14/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484742:2485208 [1] NCCL INFO Channel 11/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484743:2485209 [2] NCCL INFO Channel 15/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484747:2485204 [6] NCCL INFO Channel 23/0 : 2[6] -> 1[5] via P2P/IPC
+ip-26-0-164-0:2484742:2485208 [1] NCCL INFO Channel 12/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484743:2485209 [2] NCCL INFO Channel 16/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484742:2485208 [1] NCCL INFO Channel 13/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484743:2485209 [2] NCCL INFO Channel 17/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484743:2485209 [2] NCCL INFO Channel 18/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484742:2485208 [1] NCCL INFO Channel 14/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484743:2485209 [2] NCCL INFO Channel 19/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484742:2485208 [1] NCCL INFO Channel 15/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484742:2485208 [1] NCCL INFO Channel 16/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484742:2485208 [1] NCCL INFO Channel 17/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484743:2485209 [2] NCCL INFO Channel 20/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484742:2485208 [1] NCCL INFO Channel 18/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484743:2485209 [2] NCCL INFO Channel 21/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484742:2485208 [1] NCCL INFO Channel 19/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484742:2485208 [1] NCCL INFO Channel 20/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484743:2485209 [2] NCCL INFO Channel 22/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484743:2485209 [2] NCCL INFO Channel 23/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-164-0:2484745:2485201 [4] NCCL INFO Connected all trees
+ip-26-0-164-0:2484745:2485201 [4] NCCL INFO NVLS comm 0xb605980 headRank 0 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-164-0:2484742:2485208 [1] NCCL INFO Channel 21/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484748:2485203 [7] NCCL INFO Connected all trees
+ip-26-0-164-0:2484748:2485203 [7] NCCL INFO NVLS comm 0xa4c3fc0 headRank 3 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-164-0:2484742:2485208 [1] NCCL INFO Channel 22/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484742:2485208 [1] NCCL INFO Channel 23/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-164-0:2484744:2485207 [3] NCCL INFO Connected all trees
+ip-26-0-164-0:2484744:2485207 [3] NCCL INFO NVLS comm 0xbf25640 headRank 3 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-164-0:2484747:2485204 [6] NCCL INFO Connected all trees
+ip-26-0-164-0:2484747:2485204 [6] NCCL INFO NVLS comm 0xb20e820 headRank 2 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-164-0:2484741:2485206 [0] NCCL INFO Connected all trees
+ip-26-0-164-0:2484741:2485206 [0] NCCL INFO NVLS comm 0xa4f3140 headRank 0 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-164-0:2484746:2485202 [5] NCCL INFO Connected all trees
+ip-26-0-164-0:2484746:2485202 [5] NCCL INFO NVLS comm 0xa785800 headRank 1 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-164-0:2484748:2485203 [7] NCCL INFO Connected NVLS tree
+ip-26-0-164-0:2484748:2485203 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-164-0:2484748:2485203 [7] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-164-0:2484748:2485203 [7] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-164-0:2484746:2485202 [5] NCCL INFO Connected NVLS tree
+ip-26-0-164-0:2484746:2485202 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-164-0:2484746:2485202 [5] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-164-0:2484746:2485202 [5] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-164-0:2484747:2485204 [6] NCCL INFO Connected NVLS tree
+ip-26-0-164-0:2484747:2485204 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-164-0:2484747:2485204 [6] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-164-0:2484747:2485204 [6] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-164-0:2484745:2485201 [4] NCCL INFO Connected NVLS tree
+ip-26-0-164-0:2484745:2485201 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-164-0:2484745:2485201 [4] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-164-0:2484745:2485201 [4] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-164-0:2484742:2485208 [1] NCCL INFO Connected all trees
+ip-26-0-164-0:2484742:2485208 [1] NCCL INFO NVLS comm 0xb97d960 headRank 1 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-164-0:2484743:2485209 [2] NCCL INFO Connected all trees
+ip-26-0-164-0:2484743:2485209 [2] NCCL INFO NVLS comm 0xa999d70 headRank 2 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-164-0:2484746:2485202 [5] NCCL INFO comm 0xa785800 rank 1 nranks 4 cudaDev 5 nvmlDev 5 busId a8000 commId 0x341dd36c0358578 - Init COMPLETE
+ip-26-0-164-0:2484747:2485204 [6] NCCL INFO comm 0xb20e820 rank 2 nranks 4 cudaDev 6 nvmlDev 6 busId b9000 commId 0x341dd36c0358578 - Init COMPLETE
+ip-26-0-164-0:2484748:2485203 [7] NCCL INFO comm 0xa4c3fc0 rank 3 nranks 4 cudaDev 7 nvmlDev 7 busId ca000 commId 0x341dd36c0358578 - Init COMPLETE
+ip-26-0-164-0:2484745:2485201 [4] NCCL INFO comm 0xb605980 rank 0 nranks 4 cudaDev 4 nvmlDev 4 busId 97000 commId 0x341dd36c0358578 - Init COMPLETE
+ip-26-0-164-0:2484744:2485207 [3] NCCL INFO Connected NVLS tree
+ip-26-0-164-0:2484743:2485209 [2] NCCL INFO Connected NVLS tree
+ip-26-0-164-0:2484743:2485209 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-164-0:2484743:2485209 [2] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-164-0:2484744:2485207 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-164-0:2484743:2485209 [2] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-164-0:2484744:2485207 [3] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-164-0:2484744:2485207 [3] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-164-0:2484742:2485208 [1] NCCL INFO Connected NVLS tree
+ip-26-0-164-0:2484741:2485206 [0] NCCL INFO Connected NVLS tree
+ip-26-0-164-0:2484741:2485206 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-164-0:2484742:2485208 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-164-0:2484742:2485208 [1] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-164-0:2484742:2485208 [1] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-164-0:2484741:2485206 [0] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-164-0:2484741:2485206 [0] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-164-0:2484744:2485207 [3] NCCL INFO comm 0xbf25640 rank 3 nranks 4 cudaDev 3 nvmlDev 3 busId 86000 commId 0x428ec3d4f02a242d - Init COMPLETE
+ip-26-0-164-0:2484742:2485208 [1] NCCL INFO comm 0xb97d960 rank 1 nranks 4 cudaDev 1 nvmlDev 1 busId 64000 commId 0x428ec3d4f02a242d - Init COMPLETE
+ip-26-0-164-0:2484743:2485209 [2] NCCL INFO comm 0xa999d70 rank 2 nranks 4 cudaDev 2 nvmlDev 2 busId 75000 commId 0x428ec3d4f02a242d - Init COMPLETE
+ip-26-0-164-0:2484741:2485206 [0] NCCL INFO comm 0xa4f3140 rank 0 nranks 4 cudaDev 0 nvmlDev 0 busId 53000 commId 0x428ec3d4f02a242d - Init COMPLETE
+12/28/2024 00:52:35 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]: [Optimizer Building] Using LearningRateForSP as learning rate
+12/28/2024 00:52:35 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]: [ZeRO sharding] Size of optimizer params per rank:
+12/28/2024 00:52:35 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]: [ZeRO sharding] DP Rank 0 has 1.23G out of 4.92G (25.00%) params' optimizer states
+12/28/2024 00:52:35 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]: [ZeRO sharding] DP Rank 1 has 1.23G out of 4.92G (25.00%) params' optimizer states
+12/28/2024 00:52:35 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]: [ZeRO sharding] DP Rank 2 has 1.23G out of 4.92G (25.00%) params' optimizer states
+12/28/2024 00:52:35 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]: [ZeRO sharding] DP Rank 3 has 1.23G out of 4.92G (25.00%) params' optimizer states
+12/28/2024 00:52:35 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]: module_name: model.token_position_embeddings | PP: 0/2 | Block rank: 0
+12/28/2024 00:52:35 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]: module_name: model.decoder.0 | PP: 0/2 | Block rank: 1
+12/28/2024 00:52:35 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]: module_name: model.decoder.1 | PP: 0/2 | Block rank: 2
+12/28/2024 00:52:35 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]: module_name: model.decoder.2 | PP: 0/2 | Block rank: 3
+12/28/2024 00:52:35 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]: module_name: model.decoder.3 | PP: 0/2 | Block rank: 4
+12/28/2024 00:52:35 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]: module_name: model.decoder.4 | PP: 0/2 | Block rank: 5
+12/28/2024 00:52:35 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]: module_name: model.decoder.5 | PP: 0/2 | Block rank: 6
+12/28/2024 00:52:35 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]: module_name: model.decoder.6 | PP: 0/2 | Block rank: 7
+12/28/2024 00:52:35 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]: module_name: model.decoder.7 | PP: 0/2 | Block rank: 8
+12/28/2024 00:52:35 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]: module_name: model.decoder.8 | PP: 0/2 | Block rank: 9
+12/28/2024 00:52:35 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]: module_name: model.decoder.9 | PP: 0/2 | Block rank: 10
+12/28/2024 00:52:35 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]: module_name: model.decoder.10 | PP: 0/2 | Block rank: 11
+12/28/2024 00:52:35 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]: module_name: model.decoder.11 | PP: 0/2 | Block rank: 12
+12/28/2024 00:52:35 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]: module_name: model.decoder.12 | PP: 0/2 | Block rank: 13
+12/28/2024 00:52:35 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]: module_name: model.decoder.13 | PP: 0/2 | Block rank: 14
+12/28/2024 00:52:35 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]: module_name: model.decoder.14 | PP: 0/2 | Block rank: 15
+12/28/2024 00:52:35 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]: module_name: model.decoder.15 | PP: 0/2 | Block rank: 16
+12/28/2024 00:52:35 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]: module_name: model.decoder.16 | PP: 0/2 | Block rank: 17
+12/28/2024 00:52:35 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]: module_name: model.decoder.17 | PP: 0/2 | Block rank: 18
+12/28/2024 00:52:35 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]: module_name: model.decoder.18 | PP: 1/2 | Block rank: 0
+12/28/2024 00:52:35 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]: module_name: model.decoder.19 | PP: 1/2 | Block rank: 1
+12/28/2024 00:52:35 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]: module_name: model.decoder.20 | PP: 1/2 | Block rank: 2
+12/28/2024 00:52:35 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]: module_name: model.decoder.21 | PP: 1/2 | Block rank: 3
+12/28/2024 00:52:35 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]: module_name: model.decoder.22 | PP: 1/2 | Block rank: 4
+12/28/2024 00:52:35 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]: module_name: model.decoder.23 | PP: 1/2 | Block rank: 5
+12/28/2024 00:52:35 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]: module_name: model.decoder.24 | PP: 1/2 | Block rank: 6
+12/28/2024 00:52:35 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]: module_name: model.decoder.25 | PP: 1/2 | Block rank: 7
+12/28/2024 00:52:35 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]: module_name: model.decoder.26 | PP: 1/2 | Block rank: 8
+12/28/2024 00:52:35 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]: module_name: model.decoder.27 | PP: 1/2 | Block rank: 9
+12/28/2024 00:52:35 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]: module_name: model.decoder.28 | PP: 1/2 | Block rank: 10
+12/28/2024 00:52:35 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]: module_name: model.decoder.29 | PP: 1/2 | Block rank: 11
+12/28/2024 00:52:35 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]: module_name: model.decoder.30 | PP: 1/2 | Block rank: 12
+12/28/2024 00:52:35 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]: module_name: model.decoder.31 | PP: 1/2 | Block rank: 13
+12/28/2024 00:52:35 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]: module_name: model.final_layer_norm | PP: 1/2 | Block rank: 14
+12/28/2024 00:52:35 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]: module_name: model.lm_head | PP: 1/2 | Block rank: 15
+12/28/2024 00:52:35 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]: module_name: model.cast_to_fp32 | PP: 1/2 | Block rank: 16
+12/28/2024 00:52:35 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]: module_name: loss | PP: 1/2 | Block rank: 17
+12/28/2024 00:52:35 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]: [Training Plan] Stage Stable Training Stage has 99 remaining training steps and has consumed 0 samples
+12/28/2024 00:52:35 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]: Using dummy data generator
+12/28/2024 00:52:35 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]: [Training Plan] There are 1 training stages 
+12/28/2024 00:52:35 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]: [Stage Stable Training Stage] start from step 1 
+12/28/2024 00:52:35 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]: 
+12/28/2024 00:52:35 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]: [Start training] datetime: 2024-12-28 00:52:35.746012 | mbs: 64 | grad_accum: 1 | global_batch_size: 256 | sequence_length: 4096 | train_steps: 100 | start_iteration_step: 0 | consumed_train_samples: 0
+12/28/2024 00:52:35 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]: Resuming training from stage Stable Training Stage, it has trained for 0 samples and has 99 remaining train steps
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+12/28/2024 00:52:35 [INFO|DP=0|PP=0|TP=0|ip-26-0-164-0]:  Memory usage: 32817.01MiB. Peak allocated 32817.01MiB. Peak reserved: 36246.00MiB
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+        return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]trainer.train(dataloader)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 160, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 244, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    merged_states = self.gate_up_proj(hidden_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 89, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 160, in forward
+    return column_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 441, in column_linear
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return _ColumnLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode, tp_recompute_allgather)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/utils.py", line 40, in wrapper
+    return func(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 140, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    return F.linear(gathered_tensor, weight, bias)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 14.00 GiB. GPU 2 has a total capacty of 79.33 GiB of which 10.69 GiB is free. Including non-PyTorch memory, this process has 68.63 GiB memory in use. Of the allocated memory 54.12 GiB is allocated by PyTorch, and 4.28 GiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 244, in forward
+    merged_states = self.gate_up_proj(hidden_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 89, in forward
+    return column_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 441, in column_linear
+    return _ColumnLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode, tp_recompute_allgather)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/utils.py", line 40, in wrapper
+    return func(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 140, in forward
+    return F.linear(gathered_tensor, weight, bias)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 14.00 GiB. GPU 1 has a total capacty of 79.33 GiB of which 10.69 GiB is free. Including non-PyTorch memory, this process has 68.63 GiB memory in use. Of the allocated memory 54.12 GiB is allocated by PyTorch, and 4.28 GiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 160, in forward
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 244, in forward
+    merged_states = self.gate_up_proj(hidden_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 89, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return column_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 441, in column_linear
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return _ColumnLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode, tp_recompute_allgather)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/utils.py", line 40, in wrapper
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return func(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 140, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return F.linear(gathered_tensor, weight, bias)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 14.00 GiB. GPU 3 has a total capacty of 79.33 GiB of which 10.92 GiB is free. Including non-PyTorch memory, this process has 68.40 GiB memory in use. Of the allocated memory 54.12 GiB is allocated by PyTorch, and 4.28 GiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 160, in forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 244, in forward
+    merged_states = self.gate_up_proj(hidden_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 89, in forward
+    return column_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 441, in column_linear
+    return _ColumnLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode, tp_recompute_allgather)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/utils.py", line 40, in wrapper
+    return func(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 140, in forward
+    return F.linear(gathered_tensor, weight, bias)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 14.00 GiB. GPU 0 has a total capacty of 79.33 GiB of which 10.87 GiB is free. Including non-PyTorch memory, this process has 68.44 GiB memory in use. Of the allocated memory 54.12 GiB is allocated by PyTorch, and 4.28 GiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+12/28/2024 00:52:38 [ERROR|DP=1|PP=1|TP=0|ip-26-0-164-0]: [RECV META] Rank 1: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 00:52:38 [ERROR|DP=0|PP=1|TP=0|ip-26-0-164-0]: [RECV META] Rank 1: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 00:52:38 [ERROR|DP=3|PP=1|TP=0|ip-26-0-164-0]: [RECV META] Rank 1: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 00:52:38 [ERROR|DP=2|PP=1|TP=0|ip-26-0-164-0]: [RECV META] Rank 1: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 00:52:38 [WARNING|DP=1|PP=1|TP=0|ip-26-0-164-0]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 00:52:38 [WARNING|DP=3|PP=1|TP=0|ip-26-0-164-0]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 00:52:38 [WARNING|DP=2|PP=1|TP=0|ip-26-0-164-0]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 00:52:38 [WARNING|DP=1|PP=1|TP=0|ip-26-0-164-0]: No progress made in communication iteration, attempt 1/1
+12/28/2024 00:52:38 [WARNING|DP=0|PP=1|TP=0|ip-26-0-164-0]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 00:52:38 [ERROR|DP=1|PP=1|TP=0|ip-26-0-164-0]: Max retries reached, giving up on communication
+12/28/2024 00:52:38 [WARNING|DP=3|PP=1|TP=0|ip-26-0-164-0]: No progress made in communication iteration, attempt 1/1
+12/28/2024 00:52:38 [WARNING|DP=2|PP=1|TP=0|ip-26-0-164-0]: No progress made in communication iteration, attempt 1/1
+12/28/2024 00:52:38 [WARNING|DP=0|PP=1|TP=0|ip-26-0-164-0]: No progress made in communication iteration, attempt 1/1
+12/28/2024 00:52:38 [ERROR|DP=3|PP=1|TP=0|ip-26-0-164-0]: Max retries reached, giving up on communication
+12/28/2024 00:52:38 [ERROR|DP=2|PP=1|TP=0|ip-26-0-164-0]: Max retries reached, giving up on communication
+12/28/2024 00:52:38 [ERROR|DP=0|PP=1|TP=0|ip-26-0-164-0]: Max retries reached, giving up on communication
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+    activation = pipeline_state.activations_buffer.popleft()
+IndexError: pop from an empty deque
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+        outputs = self.pipeline_engine.train_batch_iter(outputs = self.pipeline_engine.train_batch_iter(
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+        return forward_call(*args, **kwargs)hidden_encoder_states = encoder_block(**hidden_encoder_states)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+    activation = pipeline_state.activations_buffer.popleft()
+IndexError: pop from an empty deque
+    activation = pipeline_state.activations_buffer.popleft()
+IndexError: pop from an empty deque
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+    activation = pipeline_state.activations_buffer.popleft()
+IndexError: pop from an empty deque
+ip-26-0-164-0:2484745:2485217 [4] NCCL INFO [Service thread] Connection closed by localRank 1
+ip-26-0-164-0:2484745:2485217 [4] NCCL INFO [Service thread] Connection closed by localRank 2
+ip-26-0-164-0:2484745:2485217 [4] NCCL INFO [Service thread] Connection closed by localRank 3
+[2024-12-28 00:52:48,969] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2484744 closing signal SIGTERM
+[2024-12-28 00:52:48,969] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2484745 closing signal SIGTERM
+[2024-12-28 00:52:48,969] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2484746 closing signal SIGTERM
+[2024-12-28 00:52:48,969] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2484747 closing signal SIGTERM
+[2024-12-28 00:52:49,647] torch.distributed.elastic.multiprocessing.api: [ERROR] failed (exitcode: 1) local_rank: 0 (pid: 2484741) of binary: /fsx/nouamane/miniconda/envs/2-1-cu121/bin/python
+[2024-12-28 00:52:49,688] torch.distributed.elastic.multiprocessing.errors.error_handler: [ERROR] no error file defined for parent, to copy child error file (/tmp/torchelastic_buoiir12/13848147_w41p6kz4/attempt_0/0/error.json)
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 264, in launch_agent
+    raise ChildFailedError(
+torch.distributed.elastic.multiprocessing.errors.ChildFailedError: 
+============================================================
+/fsx/nouamane/projects/nanotron/run_train.py FAILED
+------------------------------------------------------------
+Failures:
+[1]:
+  time      : 2024-12-28_00:52:37
+  host      : ip-26-0-164-0.ec2.internal
+  rank      : 1 (local_rank: 1)
+  exitcode  : 1 (pid: 2484742)
+  error_file: /tmp/torchelastic_buoiir12/13848147_w41p6kz4/attempt_0/1/error.json
+  traceback : Traceback (most recent call last):
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+      return f(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+      outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+      outputs = self.pipeline_engine.train_batch_iter(
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+      output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+      output = model(**micro_batch)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+      sharded_logits = self.model(
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+      return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+      hidden_encoder_states = encoder_block(**hidden_encoder_states)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 160, in forward
+      output = self.pp_block(**new_kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+      hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+      hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 244, in forward
+      merged_states = self.gate_up_proj(hidden_states)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 89, in forward
+      return column_linear(
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 441, in column_linear
+      return _ColumnLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode, tp_recompute_allgather)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+      return super().apply(*args, **kwargs)  # type: ignore[misc]
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/utils.py", line 40, in wrapper
+      return func(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 140, in forward
+      return F.linear(gathered_tensor, weight, bias)
+  torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 14.00 GiB. GPU 1 has a total capacty of 79.33 GiB of which 10.69 GiB is free. Including non-PyTorch memory, this process has 68.63 GiB memory in use. Of the allocated memory 54.12 GiB is allocated by PyTorch, and 4.28 GiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  
+[2]:
+  time      : 2024-12-28_00:52:37
+  host      : ip-26-0-164-0.ec2.internal
+  rank      : 2 (local_rank: 2)
+  exitcode  : 1 (pid: 2484743)
+  error_file: /tmp/torchelastic_buoiir12/13848147_w41p6kz4/attempt_0/2/error.json
+  traceback : Traceback (most recent call last):
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+      return f(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+      outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+      outputs = self.pipeline_engine.train_batch_iter(
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+      output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+      output = model(**micro_batch)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+      sharded_logits = self.model(
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+      return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+      hidden_encoder_states = encoder_block(**hidden_encoder_states)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 160, in forward
+      output = self.pp_block(**new_kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+      hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+      hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 244, in forward
+      merged_states = self.gate_up_proj(hidden_states)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 89, in forward
+      return column_linear(
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 441, in column_linear
+      return _ColumnLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode, tp_recompute_allgather)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+      return super().apply(*args, **kwargs)  # type: ignore[misc]
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/utils.py", line 40, in wrapper
+      return func(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 140, in forward
+      return F.linear(gathered_tensor, weight, bias)
+  torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 14.00 GiB. GPU 2 has a total capacty of 79.33 GiB of which 10.69 GiB is free. Including non-PyTorch memory, this process has 68.63 GiB memory in use. Of the allocated memory 54.12 GiB is allocated by PyTorch, and 4.28 GiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  
+[3]:
+  time      : 2024-12-28_00:52:38
+  host      : ip-26-0-164-0.ec2.internal
+  rank      : 7 (local_rank: 7)
+  exitcode  : 1 (pid: 2484748)
+  error_file: /tmp/torchelastic_buoiir12/13848147_w41p6kz4/attempt_0/7/error.json
+  traceback : Traceback (most recent call last):
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+      return f(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+      outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+      outputs = self.pipeline_engine.train_batch_iter(
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+      output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+      output = model(**micro_batch)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+      sharded_logits = self.model(
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+      return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+      hidden_encoder_states = encoder_block(**hidden_encoder_states)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+      new_kwargs[name] = recv_from_pipeline_state_buffer(
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+      activation = pipeline_state.activations_buffer.popleft()
+  IndexError: pop from an empty deque
+  
+------------------------------------------------------------
+Root Cause (first observed failure):
+[0]:
+  time      : 2024-12-28_00:52:37
+  host      : ip-26-0-164-0.ec2.internal
+  rank      : 0 (local_rank: 0)
+  exitcode  : 1 (pid: 2484741)
+  error_file: /tmp/torchelastic_buoiir12/13848147_w41p6kz4/attempt_0/0/error.json
+  traceback : Traceback (most recent call last):
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+      return f(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+      outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+      outputs = self.pipeline_engine.train_batch_iter(
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+      output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+      output = model(**micro_batch)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+      sharded_logits = self.model(
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+      return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+      hidden_encoder_states = encoder_block(**hidden_encoder_states)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 160, in forward
+      output = self.pp_block(**new_kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+      hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+      hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 244, in forward
+      merged_states = self.gate_up_proj(hidden_states)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 89, in forward
+      return column_linear(
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 441, in column_linear
+      return _ColumnLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode, tp_recompute_allgather)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+      return super().apply(*args, **kwargs)  # type: ignore[misc]
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/utils.py", line 40, in wrapper
+      return func(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 140, in forward
+      return F.linear(gathered_tensor, weight, bias)
+  torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 14.00 GiB. GPU 0 has a total capacty of 79.33 GiB of which 10.87 GiB is free. Including non-PyTorch memory, this process has 68.44 GiB memory in use. Of the allocated memory 54.12 GiB is allocated by PyTorch, and 4.28 GiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  
+============================================================
+srun: error: ip-26-0-164-0: task 0: Exited with exit code 1
+srun: launch/slurm: _step_signal: Terminating StepId=13848147.0
diff --git a/logs/13848421-bench_469G_dp2_tp4_pp4_acc4_mbs32_seq4096_zero1_tpmodeRED_vocab131k.out b/logs/13848421-bench_469G_dp2_tp4_pp4_acc4_mbs32_seq4096_zero1_tpmodeRED_vocab131k.out
new file mode 100644
index 0000000000000000000000000000000000000000..35d3539b4f866b10704f256f27616e9117eebea7
--- /dev/null
+++ b/logs/13848421-bench_469G_dp2_tp4_pp4_acc4_mbs32_seq4096_zero1_tpmodeRED_vocab131k.out
@@ -0,0 +1,16276 @@
++ '[' -z 13848421 ']'
++ source /etc/profile.d/modules.sh
+++ . /usr/share/modules/init/bash
++++ unset _mlshdbg
++++ '[' 0 = 1 ']'
++++ unset _mlre _mlIFS
++++ '[' -n x ']'
++++ _mlIFS=' 	
+'
++++ IFS=' '
++++ '[' -n '' ']'
+++++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash autoinit
++++ _mlcode='module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+MODULEPATH_modshare=/etc/environment-modules/modules:1:/usr/share/modules/\$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1; export MODULEPATH_modshare;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+LOADEDMODULES=; export LOADEDMODULES;
+MODULEPATH=/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/\$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles; export MODULEPATH;
+test 0;'
++++ _mlret=0
++++ '[' -n x ']'
++++ IFS=' 	
+'
++++ unset _mlIFS
++++ unset _mlre _mlv _mlrv
++++ '[' 0 -eq 0 ']'
++++ eval 'module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+MODULEPATH_modshare=/etc/environment-modules/modules:1:/usr/share/modules/\$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1; export MODULEPATH_modshare;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+LOADEDMODULES=; export LOADEDMODULES;
+MODULEPATH=/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/\$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles; export MODULEPATH;
+test 0;'
+++++ MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl
+++++ export MODULES_CMD
+++++ ENV=/usr/share/modules/init/profile.sh
+++++ export ENV
+++++ MODULEPATH_modshare='/etc/environment-modules/modules:1:/usr/share/modules/$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1'
+++++ export MODULEPATH_modshare
+++++ BASH_ENV=/usr/share/modules/init/bash
+++++ export BASH_ENV
+++++ MODULESHOME=/usr/share/modules
+++++ export MODULESHOME
+++++ LOADEDMODULES=
+++++ export LOADEDMODULES
+++++ MODULEPATH='/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles'
+++++ export MODULEPATH
+++++ test 0
++++ '[' 0 = 1 ']'
++++ '[' -t 2 ']'
++++ export -f module
++++ export -f switchml
++++ '[' 5 -ge 3 ']'
++++ [[ ehxB =~ i ]]
++++ [[ ! :/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin: =~ :/usr/bin: ]]
++++ '[' '!' -n '' ']'
++++ MANPATH=:
++++ export MANPATH
+++++ manpath
++++ [[ ! :/fsx/nouamane/miniconda/envs/2-1-cu121/man:/fsx/nouamane/miniconda/envs/2-1-cu121/share/man:/fsx/nouamane/miniconda/man:/fsx/nouamane/miniconda/share/man:/usr/local/man:/usr/local/share/man:/usr/share/man:/opt/slurm/share/man:/admin/home/nouamane/.local/share/man:/admin/home/nouamane/.fzf/man:: =~ :/usr/share/man: ]]
++++ unset _mlcode _mlret
++++ '[' -n '' ']'
++ module load cuda/12.1
++ unset _mlshdbg
++ '[' 0 = 1 ']'
++ unset _mlre _mlIFS
++ '[' -n x ']'
++ _mlIFS=' 	
+'
++ IFS=' '
++ '[' -n '' ']'
+++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash load cuda/12.1
++ eval 'CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include;' export 'CPATH;
+LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:;' export 'LD_LIBRARY_PATH;
+FI_EFA_FORK_SAFE=1;' export 'FI_EFA_FORK_SAFE;
+MANPATH=/usr/local/cuda-12.1/share/man::;' export 'MANPATH;
+LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64;' export 'LIBRARY_PATH;
+_LMFILES_=/usr/share/modules/modulefiles/cuda/12.1;' export '_LMFILES_;
+LOADEDMODULES=cuda/12.1;' export 'LOADEDMODULES;
+MPI_PATH=/opt/amazon/openmpi;' export 'MPI_PATH;
+NCCL_HOME_modshare=/opt/nccl/build:1;' export 'NCCL_HOME_modshare;
+NCCL_PROTO=simple;' export 'NCCL_PROTO;
+MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1;' export 'MANPATH_modshare;
+LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1;' export 'LIBRARY_PATH_modshare;
+NCCL_SOCKET_IFNAME=enp;' export 'NCCL_SOCKET_IFNAME;
+AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl;' export 'AWS_OFI_NCCL_HOME;
+NCCL_HOME=/opt/nccl/build;' export 'NCCL_HOME;
+FI_PROVIDER=efa;' export 'FI_PROVIDER;
+AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1;' export 'AWS_OFI_NCCL_HOME_modshare;
+CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1;' export 'CPATH_modshare;
+LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2::1:/opt/nccl/build/lib:1:/opt/aws-ofi-nccl/lib:2:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/efa/lib:1;' export 'LD_LIBRARY_PATH_modshare;
+FI_EFA_ENABLE_SHM_TRANSFER=1;' export 'FI_EFA_ENABLE_SHM_TRANSFER;
+_LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1;' export '_LMFILES__modshare;
+LOADEDMODULES_modshare=cuda/12.1:1;' export 'LOADEDMODULES_modshare;
+MPI_PATH_modshare=/opt/amazon/openmpi:1;' export 'MPI_PATH_modshare;
+PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin;' export 'PATH;
+CUDA_HOME=/usr/local/cuda-12.1;' export 'CUDA_HOME;
+PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/opt/slurm/bin:1:/usr/sbin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1;' export 'PATH_modshare;
+test' '0;'
+++ CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include
+++ export CPATH
+++ LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:
+++ export LD_LIBRARY_PATH
+++ FI_EFA_FORK_SAFE=1
+++ export FI_EFA_FORK_SAFE
+++ MANPATH=/usr/local/cuda-12.1/share/man::
+++ export MANPATH
+++ LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64
+++ export LIBRARY_PATH
+++ _LMFILES_=/usr/share/modules/modulefiles/cuda/12.1
+++ export _LMFILES_
+++ LOADEDMODULES=cuda/12.1
+++ export LOADEDMODULES
+++ MPI_PATH=/opt/amazon/openmpi
+++ export MPI_PATH
+++ NCCL_HOME_modshare=/opt/nccl/build:1
+++ export NCCL_HOME_modshare
+++ NCCL_PROTO=simple
+++ export NCCL_PROTO
+++ MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1
+++ export MANPATH_modshare
+++ LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1
+++ export LIBRARY_PATH_modshare
+++ NCCL_SOCKET_IFNAME=enp
+++ export NCCL_SOCKET_IFNAME
+++ AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl
+++ export AWS_OFI_NCCL_HOME
+++ NCCL_HOME=/opt/nccl/build
+++ export NCCL_HOME
+++ FI_PROVIDER=efa
+++ export FI_PROVIDER
+++ AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1
+++ export AWS_OFI_NCCL_HOME_modshare
+++ CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1
+++ export CPATH_modshare
+++ LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2::1:/opt/nccl/build/lib:1:/opt/aws-ofi-nccl/lib:2:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/efa/lib:1
+++ export LD_LIBRARY_PATH_modshare
+++ FI_EFA_ENABLE_SHM_TRANSFER=1
+++ export FI_EFA_ENABLE_SHM_TRANSFER
+++ _LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1
+++ export _LMFILES__modshare
+++ LOADEDMODULES_modshare=cuda/12.1:1
+++ export LOADEDMODULES_modshare
+++ MPI_PATH_modshare=/opt/amazon/openmpi:1
+++ export MPI_PATH_modshare
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
+++ export PATH
+++ CUDA_HOME=/usr/local/cuda-12.1
+++ export CUDA_HOME
+++ PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/opt/slurm/bin:1:/usr/sbin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1
+++ export PATH_modshare
+++ test 0
++ _mlstatus=0
++ '[' -n x ']'
++ IFS=' 	
+'
++ unset _mlre _mlv _mlrv _mlIFS
++ '[' -n '' ']'
++ unset _mlshdbg
++ return 0
++ source /fsx/nouamane/miniconda/bin/activate
+++ _CONDA_ROOT=/fsx/nouamane/miniconda
+++ . /fsx/nouamane/miniconda/etc/profile.d/conda.sh
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ '[' -z x ']'
+++ conda activate
+++ local cmd=activate
+++ case "$cmd" in
+++ __conda_activate activate
+++ '[' -n '' ']'
+++ local ask_conda
++++ PS1=
++++ __conda_exe shell.posix activate
++++ /fsx/nouamane/miniconda/bin/conda shell.posix activate
+Error while loading conda entry point: conda-libmamba-solver (module 'libmambapy' has no attribute 'QueryFormat')
+++ ask_conda='. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''2'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_1='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/etc/conda/activate.d/libxml2_activate.sh"'
+++ eval '. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''2'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_1='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/etc/conda/activate.d/libxml2_activate.sh"'
++++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh
+++++ test -n ''
+++++ unset XML_CATALOG_FILES
+++++ unset xml_catalog_files_libxml2
++++ PS1='(base) '
++++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
++++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
++++ export CONDA_PREFIX=/fsx/nouamane/miniconda
++++ CONDA_PREFIX=/fsx/nouamane/miniconda
++++ export CONDA_SHLVL=2
++++ CONDA_SHLVL=2
++++ export CONDA_DEFAULT_ENV=base
++++ CONDA_DEFAULT_ENV=base
++++ export 'CONDA_PROMPT_MODIFIER=(base) '
++++ CONDA_PROMPT_MODIFIER='(base) '
++++ export CONDA_PREFIX_1=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ CONDA_PREFIX_1=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ . /fsx/nouamane/miniconda/etc/conda/activate.d/libxml2_activate.sh
+++++ test -n ''
+++++ xml_catalog_files_libxml2=
+++++ XML_CATALOG_FILES=
+++++ conda_catalog_files=
+++++ ifs_libxml2=' 	
+'
+++++ IFS=' '
+++++ rem=/fsx/nouamane/miniconda
+++++ for pre in ${rem}
+++++ test '' = /fsx/nouamane/miniconda
+++++ conda_catalog_files=/fsx/nouamane/miniconda
+++++ rem=
+++++ IFS=' 	
+'
+++++ conda_catalog_files='file:///fsx/nouamane/miniconda/etc/xml/catalog file:///etc/xml/catalog'
+++++ export 'XML_CATALOG_FILES=file:///fsx/nouamane/miniconda/etc/xml/catalog file:///etc/xml/catalog'
+++++ XML_CATALOG_FILES='file:///fsx/nouamane/miniconda/etc/xml/catalog file:///etc/xml/catalog'
+++++ unset conda_catalog_files ifs_libxml2 rem
+++ __conda_hashr
+++ '[' -n '' ']'
+++ '[' -n '' ']'
+++ hash -r
++ conda activate 2-1-cu121
++ local cmd=activate
++ case "$cmd" in
++ __conda_activate activate 2-1-cu121
++ '[' -n '' ']'
++ local ask_conda
+++ PS1='(base) '
+++ __conda_exe shell.posix activate 2-1-cu121
+++ /fsx/nouamane/miniconda/bin/conda shell.posix activate 2-1-cu121
+Error while loading conda entry point: conda-libmamba-solver (module 'libmambapy' has no attribute 'QueryFormat')
++ ask_conda='. "/fsx/nouamane/miniconda/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_2='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
++ eval '. "/fsx/nouamane/miniconda/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_2='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
+++ . /fsx/nouamane/miniconda/etc/conda/deactivate.d/libxml2_deactivate.sh
++++ test -n ''
++++ unset XML_CATALOG_FILES
++++ unset xml_catalog_files_libxml2
+++ PS1='(2-1-cu121) '
+++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
+++ export CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ export CONDA_SHLVL=3
+++ CONDA_SHLVL=3
+++ export CONDA_DEFAULT_ENV=2-1-cu121
+++ CONDA_DEFAULT_ENV=2-1-cu121
+++ export 'CONDA_PROMPT_MODIFIER=(2-1-cu121) '
+++ CONDA_PROMPT_MODIFIER='(2-1-cu121) '
+++ export CONDA_PREFIX_2=/fsx/nouamane/miniconda
+++ CONDA_PREFIX_2=/fsx/nouamane/miniconda
+++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ export _CE_M=
+++ _CE_M=
+++ export _CE_CONDA=
+++ _CE_CONDA=
+++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh
++++ test -n ''
++++ xml_catalog_files_libxml2=
++++ XML_CATALOG_FILES=
++++ conda_catalog_files=
++++ ifs_libxml2=' 	
+'
++++ IFS=' '
++++ rem=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ for pre in ${rem}
++++ test '' = /fsx/nouamane/miniconda/envs/2-1-cu121
++++ conda_catalog_files=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ rem=
++++ IFS=' 	
+'
++++ conda_catalog_files='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ export 'XML_CATALOG_FILES=file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ XML_CATALOG_FILES='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ unset conda_catalog_files ifs_libxml2 rem
++ __conda_hashr
++ '[' -n '' ']'
++ '[' -n '' ']'
++ hash -r
++ export PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
++ PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
++ '[' -z '' ']'
+++ scontrol show hostnames 'ip-26-0-162-14,ip-26-0-163-58,ip-26-0-170-[31,132]'
++ export 'NODELIST=ip-26-0-162-14
+ip-26-0-163-58
+ip-26-0-170-31
+ip-26-0-170-132'
++ NODELIST='ip-26-0-162-14
+ip-26-0-163-58
+ip-26-0-170-31
+ip-26-0-170-132'
+++ head -n1
+++ scontrol show hostnames 'ip-26-0-162-14,ip-26-0-163-58,ip-26-0-170-[31,132]'
++ export MASTER_NODE=ip-26-0-162-14
++ MASTER_NODE=ip-26-0-162-14
++ export MASTER_PORT=12356
++ MASTER_PORT=12356
++ export NNODES=4
++ NNODES=4
++ export GPUS_PER_NODE=8
++ GPUS_PER_NODE=8
++ export WORLD_SIZE=32
++ WORLD_SIZE=32
++ export CUDA_DEVICE_MAX_CONNECTIONS=1
++ CUDA_DEVICE_MAX_CONNECTIONS=1
++ export NCCL_DEBUG=INFO
++ NCCL_DEBUG=INFO
++ export NANOTRON_BENCHMARK=1
++ NANOTRON_BENCHMARK=1
++ export WANDB_MODE=disabled
++ WANDB_MODE=disabled
++ export TORCH_NCCL_ASYNC_ERROR_HANDLING=1
++ TORCH_NCCL_ASYNC_ERROR_HANDLING=1
++ export TORCH_DISTRIBUTED_DEBUG=DETAIL
++ TORCH_DISTRIBUTED_DEBUG=DETAIL
++ '[' -z '' ']'
++ echo '=== GPU Topology ==='
+=== GPU Topology ===
++ nvidia-smi topo -m
+	GPU0	GPU1	GPU2	GPU3	GPU4	GPU5	GPU6	GPU7	CPU Affinity	NUMA Affinity	GPU NUMA ID
+GPU0	 X 	NV18	NV18	NV18	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU1	NV18	 X 	NV18	NV18	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU2	NV18	NV18	 X 	NV18	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU3	NV18	NV18	NV18	 X 	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU4	NV18	NV18	NV18	NV18	 X 	NV18	NV18	NV18	48-95	1		N/A
+GPU5	NV18	NV18	NV18	NV18	NV18	 X 	NV18	NV18	48-95	1		N/A
+GPU6	NV18	NV18	NV18	NV18	NV18	NV18	 X 	NV18	48-95	1		N/A
+GPU7	NV18	NV18	NV18	NV18	NV18	NV18	NV18	 X 	48-95	1		N/A
+
+Legend:
+
+  X    = Self
+  SYS  = Connection traversing PCIe as well as the SMP interconnect between NUMA nodes (e.g., QPI/UPI)
+  NODE = Connection traversing PCIe as well as the interconnect between PCIe Host Bridges within a NUMA node
+  PHB  = Connection traversing PCIe as well as a PCIe Host Bridge (typically the CPU)
+  PXB  = Connection traversing multiple PCIe bridges (without traversing the PCIe Host Bridge)
+  PIX  = Connection traversing at most a single PCIe bridge
+  NV#  = Connection traversing a bonded set of # NVLinks
++ echo ==================
+==================
++ export SRUN_ALLOC_ARGS=
++ SRUN_ALLOC_ARGS=
++ echo 'Master node: ip-26-0-162-14'
+Master node: ip-26-0-162-14
++ echo 'All nodes: ip-26-0-162-14
+ip-26-0-163-58
+ip-26-0-170-31
+ip-26-0-170-132'
+All nodes: ip-26-0-162-14
+ip-26-0-163-58
+ip-26-0-170-31
+ip-26-0-170-132
++ echo 'World size: 32'
+World size: 32
++ '[' -n '' ']'
++ srun --wait=0 --kill-on-bad-exit=1 torchrun --nnodes=4 --nproc_per_node=8 --rdzv_id=13848421 --rdzv_backend=c10d --rdzv_endpoint=ip-26-0-162-14:12356 --max_restarts 0 --rdzv_conf timeout=60 /fsx/nouamane/projects/nanotron/run_train.py --config-file benchmark/configs/config_469G_dp2_tp4_pp4_acc4_mbs32_seq4096_zero1_tpmodeRED_vocab131k.yaml
+[2024-12-28 01:35:53,668] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-28 01:35:53,670] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-28 01:35:53,670] torch.distributed.run: [WARNING] 
+[2024-12-28 01:35:53,670] torch.distributed.run: [WARNING] *****************************************
+[2024-12-28 01:35:53,670] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-28 01:35:53,670] torch.distributed.run: [WARNING] *****************************************
+[2024-12-28 01:35:53,668] torch.distributed.run: [WARNING] 
+[2024-12-28 01:35:53,668] torch.distributed.run: [WARNING] *****************************************
+[2024-12-28 01:35:53,668] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-28 01:35:53,668] torch.distributed.run: [WARNING] *****************************************
+[2024-12-28 01:35:53,714] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-28 01:35:53,714] torch.distributed.run: [WARNING] 
+[2024-12-28 01:35:53,714] torch.distributed.run: [WARNING] *****************************************
+[2024-12-28 01:35:53,714] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-28 01:35:53,714] torch.distributed.run: [WARNING] *****************************************
+[2024-12-28 01:35:53,884] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-28 01:35:53,884] torch.distributed.run: [WARNING] 
+[2024-12-28 01:35:53,884] torch.distributed.run: [WARNING] *****************************************
+[2024-12-28 01:35:53,884] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-28 01:35:53,884] torch.distributed.run: [WARNING] *****************************************
+ip-26-0-162-14:2525075:2525075 [0] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-162-14:2525075:2525075 [0] NCCL INFO Bootstrap : Using enp71s0:26.0.162.14<0>
+ip-26-0-162-14:2525075:2525075 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-162-14:2525075:2525075 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-162-14:2525075:2525075 [0] NCCL INFO cudaDriverVersion 12020
+NCCL version 2.18.5+cuda12.2
+ip-26-0-170-31:3197185:3197185 [5] NCCL INFO cudaDriverVersion 12020
+ip-26-0-170-31:3197186:3197186 [6] NCCL INFO cudaDriverVersion 12020
+ip-26-0-170-31:3197185:3197185 [5] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-170-31:3197186:3197186 [6] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-170-31:3197186:3197186 [6] NCCL INFO Bootstrap : Using enp71s0:26.0.170.31<0>
+ip-26-0-170-31:3197185:3197185 [5] NCCL INFO Bootstrap : Using enp71s0:26.0.170.31<0>
+ip-26-0-170-132:3330188:3330188 [5] NCCL INFO cudaDriverVersion 12020
+ip-26-0-170-132:3330189:3330189 [6] NCCL INFO cudaDriverVersion 12020
+ip-26-0-170-31:3197186:3197186 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-170-31:3197186:3197186 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-170-31:3197185:3197185 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-170-31:3197185:3197185 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-170-132:3330188:3330188 [5] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-170-132:3330189:3330189 [6] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-170-132:3330185:3330185 [2] NCCL INFO cudaDriverVersion 12020
+ip-26-0-162-14:2525082:2525082 [7] NCCL INFO cudaDriverVersion 12020
+ip-26-0-170-132:3330185:3330185 [2] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-162-14:2525077:2525077 [2] NCCL INFO cudaDriverVersion 12020
+ip-26-0-162-14:2525076:2525076 [1] NCCL INFO cudaDriverVersion 12020
+ip-26-0-162-14:2525080:2525080 [5] NCCL INFO cudaDriverVersion 12020
+ip-26-0-162-14:2525082:2525082 [7] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-162-14:2525077:2525077 [2] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-162-14:2525076:2525076 [1] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-162-14:2525080:2525080 [5] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-162-14:2525079:2525079 [4] NCCL INFO cudaDriverVersion 12020
+ip-26-0-162-14:2525078:2525078 [3] NCCL INFO cudaDriverVersion 12020
+ip-26-0-170-132:3330189:3330189 [6] NCCL INFO Bootstrap : Using enp71s0:26.0.170.132<0>
+ip-26-0-170-132:3330188:3330188 [5] NCCL INFO Bootstrap : Using enp71s0:26.0.170.132<0>
+ip-26-0-162-14:2525078:2525078 [3] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-162-14:2525079:2525079 [4] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-162-14:2525081:2525081 [6] NCCL INFO cudaDriverVersion 12020
+ip-26-0-170-132:3330185:3330185 [2] NCCL INFO Bootstrap : Using enp71s0:26.0.170.132<0>
+ip-26-0-170-132:3330190:3330190 [7] NCCL INFO cudaDriverVersion 12020
+ip-26-0-170-132:3330187:3330187 [4] NCCL INFO cudaDriverVersion 12020
+ip-26-0-162-14:2525081:2525081 [6] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-170-132:3330190:3330190 [7] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-170-132:3330187:3330187 [4] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-170-132:3330188:3330188 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-170-132:3330189:3330189 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-170-132:3330185:3330185 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-170-132:3330188:3330188 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-170-132:3330185:3330185 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-170-132:3330189:3330189 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-170-31:3197187:3197187 [7] NCCL INFO cudaDriverVersion 12020
+ip-26-0-170-132:3330183:3330183 [0] NCCL INFO cudaDriverVersion 12020
+ip-26-0-170-31:3197187:3197187 [7] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-170-132:3330183:3330183 [0] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-170-31:3197181:3197181 [1] NCCL INFO cudaDriverVersion 12020
+ip-26-0-170-31:3197187:3197187 [7] NCCL INFO Bootstrap : Using enp71s0:26.0.170.31<0>
+ip-26-0-170-31:3197181:3197181 [1] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-170-132:3330190:3330190 [7] NCCL INFO Bootstrap : Using enp71s0:26.0.170.132<0>
+ip-26-0-170-132:3330187:3330187 [4] NCCL INFO Bootstrap : Using enp71s0:26.0.170.132<0>
+ip-26-0-162-14:2525082:2525082 [7] NCCL INFO Bootstrap : Using enp71s0:26.0.162.14<0>
+ip-26-0-162-14:2525077:2525077 [2] NCCL INFO Bootstrap : Using enp71s0:26.0.162.14<0>
+ip-26-0-170-132:3330183:3330183 [0] NCCL INFO Bootstrap : Using enp71s0:26.0.170.132<0>
+ip-26-0-162-14:2525076:2525076 [1] NCCL INFO Bootstrap : Using enp71s0:26.0.162.14<0>
+ip-26-0-162-14:2525080:2525080 [5] NCCL INFO Bootstrap : Using enp71s0:26.0.162.14<0>
+ip-26-0-170-31:3197187:3197187 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-170-31:3197187:3197187 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-170-31:3197181:3197181 [1] NCCL INFO Bootstrap : Using enp71s0:26.0.170.31<0>
+ip-26-0-170-132:3330190:3330190 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-170-132:3330187:3330187 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-170-132:3330190:3330190 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-162-14:2525079:2525079 [4] NCCL INFO Bootstrap : Using enp71s0:26.0.162.14<0>
+ip-26-0-162-14:2525078:2525078 [3] NCCL INFO Bootstrap : Using enp71s0:26.0.162.14<0>
+ip-26-0-162-14:2525081:2525081 [6] NCCL INFO Bootstrap : Using enp71s0:26.0.162.14<0>
+ip-26-0-170-132:3330187:3330187 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-170-132:3330183:3330183 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-170-132:3330183:3330183 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-162-14:2525077:2525077 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-162-14:2525082:2525082 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-162-14:2525077:2525077 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-162-14:2525082:2525082 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-162-14:2525076:2525076 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-162-14:2525076:2525076 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-162-14:2525080:2525080 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-162-14:2525080:2525080 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-163-58:934725:934725 [5] NCCL INFO cudaDriverVersion 12020
+ip-26-0-163-58:934723:934723 [3] NCCL INFO cudaDriverVersion 12020
+ip-26-0-170-31:3197181:3197181 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-170-31:3197181:3197181 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-163-58:934724:934724 [4] NCCL INFO cudaDriverVersion 12020
+ip-26-0-163-58:934722:934722 [2] NCCL INFO cudaDriverVersion 12020
+ip-26-0-163-58:934727:934727 [7] NCCL INFO cudaDriverVersion 12020
+ip-26-0-163-58:934720:934720 [0] NCCL INFO cudaDriverVersion 12020
+ip-26-0-162-14:2525081:2525081 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-162-14:2525081:2525081 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-163-58:934721:934721 [1] NCCL INFO cudaDriverVersion 12020
+ip-26-0-163-58:934726:934726 [6] NCCL INFO cudaDriverVersion 12020
+ip-26-0-163-58:934725:934725 [5] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-163-58:934724:934724 [4] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-162-14:2525078:2525078 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-162-14:2525079:2525079 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-162-14:2525078:2525078 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-162-14:2525079:2525079 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-163-58:934723:934723 [3] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-163-58:934720:934720 [0] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-163-58:934722:934722 [2] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-163-58:934727:934727 [7] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-163-58:934726:934726 [6] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-163-58:934721:934721 [1] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-170-31:3197182:3197182 [2] NCCL INFO cudaDriverVersion 12020
+ip-26-0-170-132:3330184:3330184 [1] NCCL INFO cudaDriverVersion 12020
+ip-26-0-170-31:3197184:3197184 [4] NCCL INFO cudaDriverVersion 12020
+ip-26-0-170-31:3197182:3197182 [2] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-170-132:3330184:3330184 [1] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-170-31:3197184:3197184 [4] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-170-132:3330184:3330184 [1] NCCL INFO Bootstrap : Using enp71s0:26.0.170.132<0>
+ip-26-0-170-31:3197182:3197182 [2] NCCL INFO Bootstrap : Using enp71s0:26.0.170.31<0>
+ip-26-0-170-31:3197184:3197184 [4] NCCL INFO Bootstrap : Using enp71s0:26.0.170.31<0>
+ip-26-0-163-58:934724:934724 [4] NCCL INFO Bootstrap : Using enp71s0:26.0.163.58<0>
+ip-26-0-163-58:934725:934725 [5] NCCL INFO Bootstrap : Using enp71s0:26.0.163.58<0>
+ip-26-0-163-58:934723:934723 [3] NCCL INFO Bootstrap : Using enp71s0:26.0.163.58<0>
+ip-26-0-163-58:934720:934720 [0] NCCL INFO Bootstrap : Using enp71s0:26.0.163.58<0>
+ip-26-0-163-58:934726:934726 [6] NCCL INFO Bootstrap : Using enp71s0:26.0.163.58<0>
+ip-26-0-163-58:934722:934722 [2] NCCL INFO Bootstrap : Using enp71s0:26.0.163.58<0>
+ip-26-0-163-58:934727:934727 [7] NCCL INFO Bootstrap : Using enp71s0:26.0.163.58<0>
+ip-26-0-163-58:934721:934721 [1] NCCL INFO Bootstrap : Using enp71s0:26.0.163.58<0>
+ip-26-0-170-132:3330184:3330184 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-170-132:3330184:3330184 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-170-31:3197182:3197182 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-170-31:3197182:3197182 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-170-132:3330186:3330186 [3] NCCL INFO cudaDriverVersion 12020
+ip-26-0-170-31:3197180:3197180 [0] NCCL INFO cudaDriverVersion 12020
+ip-26-0-170-31:3197184:3197184 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-170-31:3197184:3197184 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-170-132:3330186:3330186 [3] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-163-58:934724:934724 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-163-58:934720:934720 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-163-58:934724:934724 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-163-58:934720:934720 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-163-58:934725:934725 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-163-58:934725:934725 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-163-58:934723:934723 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-163-58:934723:934723 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-163-58:934721:934721 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-163-58:934721:934721 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-170-31:3197180:3197180 [0] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-163-58:934726:934726 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-163-58:934727:934727 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-163-58:934726:934726 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-163-58:934727:934727 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-163-58:934722:934722 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-163-58:934722:934722 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-170-132:3330186:3330186 [3] NCCL INFO Bootstrap : Using enp71s0:26.0.170.132<0>
+ip-26-0-170-31:3197180:3197180 [0] NCCL INFO Bootstrap : Using enp71s0:26.0.170.31<0>
+ip-26-0-170-132:3330186:3330186 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-170-132:3330186:3330186 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-170-31:3197180:3197180 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-170-31:3197180:3197180 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-170-31:3197183:3197183 [3] NCCL INFO cudaDriverVersion 12020
+ip-26-0-170-31:3197183:3197183 [3] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-170-31:3197183:3197183 [3] NCCL INFO Bootstrap : Using enp71s0:26.0.170.31<0>
+ip-26-0-170-31:3197183:3197183 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-170-31:3197183:3197183 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-170-31:3197185:3197353 [5] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-170-31:3197185:3197353 [5] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-170-31:3197185:3197353 [5] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-170-31:3197185:3197353 [5] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-170-31:3197185:3197353 [5] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-170-31:3197185:3197353 [5] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-170-31:3197185:3197353 [5] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-170-31:3197185:3197353 [5] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-170-31:3197185:3197353 [5] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-170-31:3197186:3197354 [6] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-170-31:3197186:3197354 [6] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-170-31:3197181:3197357 [1] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-170-31:3197181:3197357 [1] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-170-31:3197180:3197360 [0] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-170-31:3197180:3197360 [0] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-170-31:3197187:3197356 [7] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-170-31:3197187:3197356 [7] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-170-31:3197181:3197357 [1] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-170-31:3197181:3197357 [1] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-170-31:3197181:3197357 [1] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-170-31:3197181:3197357 [1] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-170-31:3197181:3197357 [1] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-170-31:3197181:3197357 [1] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-170-31:3197181:3197357 [1] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-170-31:3197186:3197354 [6] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-170-31:3197186:3197354 [6] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-170-31:3197186:3197354 [6] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-170-31:3197186:3197354 [6] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-170-31:3197186:3197354 [6] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-170-31:3197186:3197354 [6] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-170-31:3197186:3197354 [6] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-170-31:3197185:3197353 [5] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-163-58:934726:934884 [6] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-163-58:934726:934884 [6] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-170-31:3197183:3197358 [3] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-170-31:3197183:3197358 [3] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-163-58:934722:934880 [2] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-163-58:934722:934880 [2] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-163-58:934726:934884 [6] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-163-58:934726:934884 [6] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-163-58:934726:934884 [6] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-163-58:934726:934884 [6] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-163-58:934726:934884 [6] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-163-58:934726:934884 [6] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-163-58:934726:934884 [6] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-163-58:934727:934886 [7] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-163-58:934727:934886 [7] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-163-58:934725:934883 [5] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-163-58:934725:934883 [5] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-170-31:3197180:3197360 [0] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-170-31:3197180:3197360 [0] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-170-31:3197180:3197360 [0] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-170-31:3197180:3197360 [0] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-170-31:3197180:3197360 [0] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-170-31:3197180:3197360 [0] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-170-31:3197180:3197360 [0] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-163-58:934721:934885 [1] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-163-58:934721:934885 [1] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-163-58:934724:934881 [4] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-163-58:934724:934881 [4] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-163-58:934722:934880 [2] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-163-58:934722:934880 [2] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-163-58:934722:934880 [2] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-163-58:934722:934880 [2] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-163-58:934722:934880 [2] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-163-58:934722:934880 [2] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-163-58:934722:934880 [2] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-170-31:3197187:3197356 [7] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-170-31:3197187:3197356 [7] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-170-31:3197187:3197356 [7] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-170-31:3197187:3197356 [7] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-170-31:3197187:3197356 [7] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-170-31:3197187:3197356 [7] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-170-31:3197187:3197356 [7] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-170-31:3197183:3197358 [3] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-170-31:3197183:3197358 [3] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-170-31:3197183:3197358 [3] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-170-31:3197183:3197358 [3] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-170-31:3197183:3197358 [3] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-170-31:3197183:3197358 [3] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-170-31:3197183:3197358 [3] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-170-31:3197182:3197359 [2] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-170-31:3197182:3197359 [2] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-163-58:934720:934882 [0] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-163-58:934720:934882 [0] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-163-58:934727:934886 [7] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-163-58:934727:934886 [7] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-163-58:934727:934886 [7] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-163-58:934727:934886 [7] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-163-58:934727:934886 [7] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-163-58:934727:934886 [7] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-163-58:934727:934886 [7] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-163-58:934725:934883 [5] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-163-58:934725:934883 [5] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-163-58:934725:934883 [5] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-163-58:934725:934883 [5] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-163-58:934725:934883 [5] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-163-58:934725:934883 [5] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-163-58:934725:934883 [5] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-170-31:3197182:3197359 [2] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-170-31:3197182:3197359 [2] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-170-31:3197182:3197359 [2] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-170-31:3197182:3197359 [2] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-170-31:3197182:3197359 [2] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-170-31:3197182:3197359 [2] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-170-31:3197182:3197359 [2] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-163-58:934724:934881 [4] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-163-58:934724:934881 [4] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-163-58:934724:934881 [4] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-163-58:934724:934881 [4] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-163-58:934724:934881 [4] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-163-58:934724:934881 [4] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-163-58:934724:934881 [4] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-163-58:934721:934885 [1] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-163-58:934721:934885 [1] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-163-58:934721:934885 [1] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-163-58:934721:934885 [1] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-163-58:934721:934885 [1] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-163-58:934721:934885 [1] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-163-58:934721:934885 [1] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-163-58:934720:934882 [0] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-163-58:934720:934882 [0] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-163-58:934720:934882 [0] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-163-58:934720:934882 [0] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-163-58:934720:934882 [0] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-163-58:934720:934882 [0] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-163-58:934720:934882 [0] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-163-58:934723:934887 [3] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-163-58:934723:934887 [3] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-170-31:3197186:3197354 [6] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-163-58:934723:934887 [3] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-163-58:934723:934887 [3] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-163-58:934723:934887 [3] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-163-58:934723:934887 [3] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-163-58:934723:934887 [3] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-163-58:934723:934887 [3] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-163-58:934723:934887 [3] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-170-31:3197181:3197357 [1] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-170-31:3197180:3197360 [0] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-170-31:3197187:3197356 [7] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-163-58:934726:934884 [6] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-170-31:3197183:3197358 [3] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-170-31:3197182:3197359 [2] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-163-58:934722:934880 [2] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-163-58:934727:934886 [7] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-163-58:934725:934883 [5] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-163-58:934724:934881 [4] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-163-58:934721:934885 [1] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-163-58:934720:934882 [0] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-163-58:934723:934887 [3] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/253
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO Using network Libfabric
+ip-26-0-170-31:3197185:3197353 [5] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-170-31:3197185:3197353 [5] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-170-31:3197185:3197353 [5] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-170-31:3197185:3197353 [5] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-170-31:3197185:3197353 [5] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-170-31:3197185:3197353 [5] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-170-31:3197185:3197353 [5] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-170-31:3197185:3197353 [5] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-170-31:3197185:3197353 [5] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-170-31:3197185:3197353 [5] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-170-31:3197185:3197353 [5] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-170-31:3197185:3197353 [5] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-170-31:3197185:3197353 [5] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-170-31:3197185:3197353 [5] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-170-31:3197185:3197353 [5] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-170-31:3197185:3197353 [5] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-170-31:3197185:3197353 [5] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-170-31:3197185:3197353 [5] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-170-31:3197185:3197353 [5] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-170-31:3197185:3197353 [5] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-170-31:3197185:3197353 [5] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-170-31:3197185:3197353 [5] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-170-31:3197185:3197353 [5] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-170-31:3197185:3197353 [5] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-170-31:3197185:3197353 [5] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-170-31:3197185:3197353 [5] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-170-31:3197185:3197353 [5] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-170-31:3197185:3197353 [5] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-170-31:3197185:3197353 [5] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-170-31:3197185:3197353 [5] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-170-31:3197185:3197353 [5] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-170-31:3197185:3197353 [5] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-170-31:3197185:3197353 [5] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/216
+ip-26-0-170-31:3197185:3197353 [5] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/216
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/216
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/216
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-170-31:3197185:3197353 [5] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-170-31:3197185:3197353 [5] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-170-31:3197185:3197353 [5] NCCL INFO Using network Libfabric
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/216
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/216
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/216
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-170-31:3197180:3197360 [0] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-170-31:3197180:3197360 [0] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-170-31:3197180:3197360 [0] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-170-31:3197180:3197360 [0] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-170-31:3197180:3197360 [0] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-170-31:3197180:3197360 [0] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-170-31:3197180:3197360 [0] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-170-31:3197180:3197360 [0] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-170-31:3197180:3197360 [0] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-170-31:3197180:3197360 [0] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-170-31:3197180:3197360 [0] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-163-58:934722:934880 [2] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-163-58:934722:934880 [2] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-163-58:934722:934880 [2] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-163-58:934722:934880 [2] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-163-58:934722:934880 [2] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-163-58:934722:934880 [2] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-163-58:934722:934880 [2] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-163-58:934722:934880 [2] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-163-58:934722:934880 [2] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-163-58:934722:934880 [2] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-163-58:934722:934880 [2] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-170-31:3197180:3197360 [0] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-170-31:3197180:3197360 [0] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-170-31:3197180:3197360 [0] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-170-31:3197180:3197360 [0] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-163-58:934722:934880 [2] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-163-58:934722:934880 [2] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-170-31:3197180:3197360 [0] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-170-31:3197180:3197360 [0] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-170-31:3197180:3197360 [0] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-170-31:3197180:3197360 [0] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-170-31:3197180:3197360 [0] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-170-31:3197180:3197360 [0] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-170-31:3197180:3197360 [0] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-170-31:3197180:3197360 [0] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-170-31:3197180:3197360 [0] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-170-31:3197180:3197360 [0] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-170-31:3197180:3197360 [0] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-163-58:934722:934880 [2] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-163-58:934722:934880 [2] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-163-58:934722:934880 [2] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-163-58:934722:934880 [2] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-163-58:934722:934880 [2] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-163-58:934722:934880 [2] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-163-58:934722:934880 [2] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-163-58:934722:934880 [2] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-163-58:934722:934880 [2] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-163-58:934722:934880 [2] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-163-58:934722:934880 [2] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-170-31:3197180:3197360 [0] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-170-31:3197180:3197360 [0] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-170-31:3197180:3197360 [0] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-170-31:3197180:3197360 [0] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-170-31:3197180:3197360 [0] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-170-31:3197180:3197360 [0] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-163-58:934722:934880 [2] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-163-58:934722:934880 [2] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-163-58:934722:934880 [2] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-163-58:934722:934880 [2] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-163-58:934722:934880 [2] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-163-58:934722:934880 [2] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-163-58:934722:934880 [2] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-163-58:934722:934880 [2] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/216
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/216
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-170-31:3197180:3197360 [0] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/216
+ip-26-0-170-31:3197180:3197360 [0] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-163-58:934722:934880 [2] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/216
+ip-26-0-163-58:934722:934880 [2] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/216
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-170-31:3197181:3197357 [1] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-170-31:3197181:3197357 [1] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-170-31:3197181:3197357 [1] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-170-31:3197181:3197357 [1] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-170-31:3197181:3197357 [1] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-170-31:3197181:3197357 [1] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-170-31:3197181:3197357 [1] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-170-31:3197181:3197357 [1] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-170-31:3197181:3197357 [1] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-170-31:3197181:3197357 [1] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-170-31:3197181:3197357 [1] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-170-31:3197181:3197357 [1] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-170-31:3197181:3197357 [1] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-170-31:3197181:3197357 [1] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-170-31:3197181:3197357 [1] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-170-31:3197181:3197357 [1] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-170-31:3197181:3197357 [1] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-170-31:3197181:3197357 [1] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-170-31:3197181:3197357 [1] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-170-31:3197181:3197357 [1] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-170-31:3197181:3197357 [1] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-170-31:3197181:3197357 [1] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-170-31:3197181:3197357 [1] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-170-31:3197181:3197357 [1] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-170-31:3197181:3197357 [1] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-170-31:3197181:3197357 [1] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-170-31:3197181:3197357 [1] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-170-31:3197181:3197357 [1] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-170-31:3197181:3197357 [1] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-170-31:3197181:3197357 [1] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-170-31:3197181:3197357 [1] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-170-31:3197181:3197357 [1] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-170-31:3197181:3197357 [1] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/216
+ip-26-0-170-31:3197181:3197357 [1] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-170-31:3197182:3197359 [2] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-170-31:3197182:3197359 [2] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-170-31:3197182:3197359 [2] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-170-31:3197182:3197359 [2] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-170-31:3197182:3197359 [2] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-170-31:3197182:3197359 [2] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-170-31:3197182:3197359 [2] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-170-31:3197182:3197359 [2] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-170-31:3197182:3197359 [2] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-170-31:3197182:3197359 [2] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-170-31:3197182:3197359 [2] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-170-31:3197182:3197359 [2] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-170-31:3197182:3197359 [2] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-170-31:3197182:3197359 [2] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-170-31:3197182:3197359 [2] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-170-31:3197182:3197359 [2] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-170-31:3197182:3197359 [2] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-170-31:3197182:3197359 [2] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-170-31:3197182:3197359 [2] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-170-31:3197182:3197359 [2] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-170-31:3197182:3197359 [2] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-170-31:3197182:3197359 [2] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-170-31:3197182:3197359 [2] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-170-31:3197182:3197359 [2] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-170-31:3197182:3197359 [2] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-170-31:3197182:3197359 [2] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-170-31:3197182:3197359 [2] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-170-31:3197182:3197359 [2] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-170-31:3197182:3197359 [2] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-170-31:3197182:3197359 [2] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-170-31:3197182:3197359 [2] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-170-31:3197182:3197359 [2] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-163-58:934725:934883 [5] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-163-58:934725:934883 [5] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-163-58:934725:934883 [5] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-163-58:934725:934883 [5] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-163-58:934725:934883 [5] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-163-58:934725:934883 [5] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-163-58:934725:934883 [5] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-163-58:934725:934883 [5] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-163-58:934725:934883 [5] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-163-58:934725:934883 [5] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-163-58:934725:934883 [5] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-163-58:934725:934883 [5] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-163-58:934725:934883 [5] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-163-58:934725:934883 [5] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-163-58:934725:934883 [5] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-163-58:934725:934883 [5] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-163-58:934725:934883 [5] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-163-58:934725:934883 [5] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-163-58:934725:934883 [5] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-163-58:934725:934883 [5] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-163-58:934725:934883 [5] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-163-58:934725:934883 [5] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-163-58:934725:934883 [5] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-163-58:934725:934883 [5] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-163-58:934725:934883 [5] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-163-58:934725:934883 [5] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-163-58:934725:934883 [5] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-163-58:934725:934883 [5] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-163-58:934725:934883 [5] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-163-58:934725:934883 [5] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-163-58:934725:934883 [5] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-163-58:934725:934883 [5] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-163-58:934725:934883 [5] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/216
+ip-26-0-163-58:934725:934883 [5] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-170-31:3197182:3197359 [2] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/216
+ip-26-0-170-31:3197182:3197359 [2] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-170-31:3197186:3197354 [6] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-170-31:3197186:3197354 [6] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-170-31:3197186:3197354 [6] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-170-31:3197186:3197354 [6] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-170-31:3197186:3197354 [6] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-170-31:3197186:3197354 [6] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-170-31:3197186:3197354 [6] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-170-31:3197186:3197354 [6] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-170-31:3197186:3197354 [6] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-170-31:3197186:3197354 [6] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-170-31:3197186:3197354 [6] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-170-31:3197186:3197354 [6] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-170-31:3197186:3197354 [6] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-170-31:3197186:3197354 [6] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-170-31:3197186:3197354 [6] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-170-31:3197186:3197354 [6] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-170-31:3197186:3197354 [6] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-170-31:3197186:3197354 [6] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-170-31:3197186:3197354 [6] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-170-31:3197186:3197354 [6] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-170-31:3197186:3197354 [6] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-170-31:3197186:3197354 [6] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-170-31:3197186:3197354 [6] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-170-31:3197186:3197354 [6] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-170-31:3197186:3197354 [6] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-170-31:3197186:3197354 [6] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-170-31:3197186:3197354 [6] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-170-31:3197186:3197354 [6] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-170-31:3197186:3197354 [6] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-170-31:3197186:3197354 [6] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-170-31:3197186:3197354 [6] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-170-31:3197186:3197354 [6] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-170-31:3197186:3197354 [6] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/216
+ip-26-0-170-31:3197186:3197354 [6] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-170-31:3197187:3197356 [7] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-170-31:3197187:3197356 [7] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-170-31:3197187:3197356 [7] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-170-31:3197187:3197356 [7] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-170-31:3197187:3197356 [7] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-170-31:3197187:3197356 [7] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-170-31:3197187:3197356 [7] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-170-31:3197187:3197356 [7] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-170-31:3197187:3197356 [7] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-170-31:3197187:3197356 [7] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-170-31:3197187:3197356 [7] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-170-31:3197187:3197356 [7] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-170-31:3197187:3197356 [7] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-170-31:3197187:3197356 [7] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-170-31:3197187:3197356 [7] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-170-31:3197187:3197356 [7] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-170-31:3197187:3197356 [7] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-170-31:3197187:3197356 [7] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-170-31:3197187:3197356 [7] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-170-31:3197187:3197356 [7] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-170-31:3197187:3197356 [7] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-170-31:3197187:3197356 [7] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-170-31:3197187:3197356 [7] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-170-31:3197187:3197356 [7] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-170-31:3197187:3197356 [7] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-170-31:3197187:3197356 [7] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-170-31:3197187:3197356 [7] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-170-31:3197187:3197356 [7] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-170-31:3197187:3197356 [7] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-170-31:3197187:3197356 [7] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-170-31:3197187:3197356 [7] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-170-31:3197187:3197356 [7] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-170-31:3197183:3197358 [3] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-170-31:3197183:3197358 [3] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-170-31:3197183:3197358 [3] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-170-31:3197183:3197358 [3] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-170-31:3197183:3197358 [3] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-170-31:3197183:3197358 [3] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-170-31:3197183:3197358 [3] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-170-31:3197183:3197358 [3] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-170-31:3197183:3197358 [3] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-170-31:3197183:3197358 [3] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-170-31:3197183:3197358 [3] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-170-31:3197183:3197358 [3] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-170-31:3197183:3197358 [3] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-170-31:3197183:3197358 [3] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-170-31:3197183:3197358 [3] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-170-31:3197183:3197358 [3] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-170-31:3197183:3197358 [3] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-170-31:3197183:3197358 [3] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-170-31:3197183:3197358 [3] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-170-31:3197183:3197358 [3] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-170-31:3197183:3197358 [3] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-170-31:3197183:3197358 [3] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-170-31:3197183:3197358 [3] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-170-31:3197183:3197358 [3] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-170-31:3197183:3197358 [3] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-170-31:3197183:3197358 [3] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-170-31:3197183:3197358 [3] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-170-31:3197183:3197358 [3] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-170-31:3197183:3197358 [3] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-170-31:3197183:3197358 [3] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-170-31:3197183:3197358 [3] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-170-31:3197183:3197358 [3] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-163-58:934726:934884 [6] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-163-58:934726:934884 [6] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-163-58:934726:934884 [6] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-163-58:934726:934884 [6] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-163-58:934726:934884 [6] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-163-58:934726:934884 [6] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-163-58:934726:934884 [6] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-163-58:934726:934884 [6] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-163-58:934726:934884 [6] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-163-58:934726:934884 [6] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-163-58:934726:934884 [6] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-163-58:934726:934884 [6] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-163-58:934726:934884 [6] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-163-58:934726:934884 [6] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-170-31:3197187:3197356 [7] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/216
+ip-26-0-163-58:934726:934884 [6] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-163-58:934726:934884 [6] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-163-58:934726:934884 [6] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-163-58:934726:934884 [6] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-163-58:934726:934884 [6] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-163-58:934726:934884 [6] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-163-58:934726:934884 [6] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-163-58:934726:934884 [6] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-163-58:934726:934884 [6] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-163-58:934726:934884 [6] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-163-58:934726:934884 [6] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-163-58:934726:934884 [6] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-163-58:934726:934884 [6] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-163-58:934726:934884 [6] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-163-58:934726:934884 [6] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-163-58:934726:934884 [6] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-163-58:934726:934884 [6] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-163-58:934726:934884 [6] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-170-31:3197187:3197356 [7] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-170-31:3197183:3197358 [3] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/216
+ip-26-0-170-31:3197183:3197358 [3] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-163-58:934726:934884 [6] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/216
+ip-26-0-163-58:934726:934884 [6] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-163-58:934727:934886 [7] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-163-58:934727:934886 [7] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-163-58:934727:934886 [7] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-163-58:934727:934886 [7] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-163-58:934727:934886 [7] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-163-58:934727:934886 [7] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-163-58:934727:934886 [7] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-163-58:934727:934886 [7] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-163-58:934727:934886 [7] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-163-58:934727:934886 [7] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-163-58:934727:934886 [7] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-163-58:934727:934886 [7] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-163-58:934727:934886 [7] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-163-58:934727:934886 [7] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-163-58:934727:934886 [7] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-163-58:934727:934886 [7] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-163-58:934727:934886 [7] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-163-58:934727:934886 [7] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-163-58:934727:934886 [7] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-163-58:934727:934886 [7] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-163-58:934727:934886 [7] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-163-58:934727:934886 [7] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-163-58:934727:934886 [7] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-163-58:934727:934886 [7] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-163-58:934727:934886 [7] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-163-58:934727:934886 [7] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-163-58:934727:934886 [7] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-163-58:934727:934886 [7] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-163-58:934727:934886 [7] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-163-58:934727:934886 [7] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-163-58:934727:934886 [7] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-163-58:934727:934886 [7] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-163-58:934721:934885 [1] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-163-58:934721:934885 [1] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-163-58:934721:934885 [1] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-163-58:934721:934885 [1] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-163-58:934721:934885 [1] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-163-58:934721:934885 [1] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-163-58:934721:934885 [1] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-163-58:934721:934885 [1] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-163-58:934721:934885 [1] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-163-58:934721:934885 [1] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-163-58:934721:934885 [1] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-163-58:934721:934885 [1] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-163-58:934721:934885 [1] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-163-58:934721:934885 [1] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-163-58:934721:934885 [1] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-163-58:934721:934885 [1] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-163-58:934721:934885 [1] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-163-58:934721:934885 [1] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-163-58:934721:934885 [1] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-163-58:934721:934885 [1] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-163-58:934721:934885 [1] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-163-58:934721:934885 [1] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-163-58:934721:934885 [1] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-163-58:934721:934885 [1] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-163-58:934721:934885 [1] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-163-58:934721:934885 [1] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-163-58:934721:934885 [1] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-163-58:934721:934885 [1] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-163-58:934721:934885 [1] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-163-58:934721:934885 [1] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-163-58:934721:934885 [1] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-163-58:934721:934885 [1] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-163-58:934727:934886 [7] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/216
+ip-26-0-163-58:934727:934886 [7] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-163-58:934721:934885 [1] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/216
+ip-26-0-163-58:934721:934885 [1] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-163-58:934723:934887 [3] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-163-58:934723:934887 [3] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-163-58:934723:934887 [3] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-163-58:934723:934887 [3] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-163-58:934723:934887 [3] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-163-58:934723:934887 [3] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-163-58:934723:934887 [3] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-163-58:934723:934887 [3] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-163-58:934723:934887 [3] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-163-58:934723:934887 [3] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-163-58:934723:934887 [3] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-163-58:934723:934887 [3] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-163-58:934723:934887 [3] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-163-58:934723:934887 [3] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-163-58:934723:934887 [3] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-163-58:934723:934887 [3] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-163-58:934723:934887 [3] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-163-58:934723:934887 [3] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-163-58:934723:934887 [3] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-163-58:934723:934887 [3] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-163-58:934723:934887 [3] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-163-58:934723:934887 [3] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-163-58:934723:934887 [3] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-163-58:934723:934887 [3] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-163-58:934723:934887 [3] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-163-58:934723:934887 [3] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-163-58:934723:934887 [3] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-163-58:934723:934887 [3] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-163-58:934723:934887 [3] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-163-58:934723:934887 [3] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-163-58:934723:934887 [3] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-163-58:934723:934887 [3] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/216
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO Using network Libfabric
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/216
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-163-58:934723:934887 [3] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/216
+ip-26-0-163-58:934723:934887 [3] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-163-58:934724:934881 [4] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-163-58:934724:934881 [4] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-163-58:934724:934881 [4] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-163-58:934724:934881 [4] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-163-58:934724:934881 [4] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-163-58:934724:934881 [4] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-163-58:934724:934881 [4] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-163-58:934724:934881 [4] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-163-58:934724:934881 [4] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-163-58:934724:934881 [4] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-163-58:934724:934881 [4] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-163-58:934724:934881 [4] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-163-58:934724:934881 [4] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-163-58:934724:934881 [4] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-163-58:934724:934881 [4] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-163-58:934724:934881 [4] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-163-58:934724:934881 [4] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-163-58:934724:934881 [4] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-163-58:934724:934881 [4] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-163-58:934724:934881 [4] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-163-58:934724:934881 [4] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-163-58:934724:934881 [4] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-163-58:934724:934881 [4] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-163-58:934724:934881 [4] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-163-58:934724:934881 [4] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-163-58:934724:934881 [4] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-163-58:934724:934881 [4] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-163-58:934724:934881 [4] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-163-58:934724:934881 [4] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-163-58:934724:934881 [4] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-163-58:934724:934881 [4] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-163-58:934724:934881 [4] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-163-58:934724:934881 [4] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/216
+ip-26-0-163-58:934724:934881 [4] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-163-58:934720:934882 [0] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-163-58:934720:934882 [0] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-163-58:934720:934882 [0] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-163-58:934720:934882 [0] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-163-58:934720:934882 [0] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-163-58:934720:934882 [0] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-163-58:934720:934882 [0] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-163-58:934720:934882 [0] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-163-58:934720:934882 [0] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-163-58:934720:934882 [0] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-163-58:934720:934882 [0] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-163-58:934720:934882 [0] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-163-58:934720:934882 [0] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-163-58:934720:934882 [0] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-163-58:934720:934882 [0] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-163-58:934720:934882 [0] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-163-58:934720:934882 [0] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-163-58:934720:934882 [0] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-163-58:934720:934882 [0] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-163-58:934720:934882 [0] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-163-58:934720:934882 [0] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-163-58:934720:934882 [0] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-163-58:934720:934882 [0] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-163-58:934720:934882 [0] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-163-58:934720:934882 [0] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-163-58:934720:934882 [0] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-163-58:934720:934882 [0] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-163-58:934720:934882 [0] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-163-58:934720:934882 [0] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-163-58:934720:934882 [0] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-163-58:934720:934882 [0] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-163-58:934720:934882 [0] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-163-58:934720:934882 [0] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/216
+ip-26-0-163-58:934720:934882 [0] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/216
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO Using network Libfabric
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/216
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/216
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/216
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/216
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-163-58:934722:934880 [2] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-163-58:934722:934880 [2] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-163-58:934722:934880 [2] NCCL INFO Using network Libfabric
+ip-26-0-163-58:934725:934883 [5] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-163-58:934725:934883 [5] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-163-58:934725:934883 [5] NCCL INFO Using network Libfabric
+ip-26-0-163-58:934726:934884 [6] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-163-58:934726:934884 [6] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-163-58:934726:934884 [6] NCCL INFO Using network Libfabric
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO Using network Libfabric
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO Using network Libfabric
+ip-26-0-163-58:934723:934887 [3] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-163-58:934723:934887 [3] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-163-58:934723:934887 [3] NCCL INFO Using network Libfabric
+ip-26-0-163-58:934721:934885 [1] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-163-58:934721:934885 [1] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-163-58:934721:934885 [1] NCCL INFO Using network Libfabric
+ip-26-0-163-58:934727:934886 [7] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-163-58:934727:934886 [7] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-163-58:934727:934886 [7] NCCL INFO Using network Libfabric
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO Using network Libfabric
+ip-26-0-163-58:934724:934881 [4] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-163-58:934724:934881 [4] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-163-58:934724:934881 [4] NCCL INFO Using network Libfabric
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO Using network Libfabric
+ip-26-0-170-31:3197181:3197357 [1] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-170-31:3197181:3197357 [1] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-170-31:3197181:3197357 [1] NCCL INFO Using network Libfabric
+ip-26-0-170-31:3197180:3197360 [0] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-170-31:3197180:3197360 [0] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-170-31:3197180:3197360 [0] NCCL INFO Using network Libfabric
+ip-26-0-170-31:3197183:3197358 [3] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-170-31:3197183:3197358 [3] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-170-31:3197183:3197358 [3] NCCL INFO Using network Libfabric
+ip-26-0-170-31:3197182:3197359 [2] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-170-31:3197182:3197359 [2] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-170-31:3197182:3197359 [2] NCCL INFO Using network Libfabric
+ip-26-0-170-31:3197187:3197356 [7] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-170-31:3197187:3197356 [7] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-170-31:3197187:3197356 [7] NCCL INFO Using network Libfabric
+ip-26-0-170-31:3197186:3197354 [6] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-170-31:3197186:3197354 [6] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-170-31:3197186:3197354 [6] NCCL INFO Using network Libfabric
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO Using network Libfabric
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO Using network Libfabric
+ip-26-0-163-58:934720:934882 [0] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-163-58:934720:934882 [0] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-163-58:934720:934882 [0] NCCL INFO Using network Libfabric
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO Using network Libfabric
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO Using network Libfabric
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO Using network Libfabric
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO Using network Libfabric
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO Using network Libfabric
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO Using network Libfabric
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO Using network Libfabric
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO Using network Libfabric
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO comm 0x9c385f0 rank 7 nranks 32 cudaDev 7 nvmlDev 7 busId ca000 commId 0x1214b375c22bac3c - Init START
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO comm 0x9a0b900 rank 5 nranks 32 cudaDev 5 nvmlDev 5 busId a8000 commId 0x1214b375c22bac3c - Init START
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO comm 0x92b7800 rank 6 nranks 32 cudaDev 6 nvmlDev 6 busId b9000 commId 0x1214b375c22bac3c - Init START
+ip-26-0-163-58:934720:934882 [0] NCCL INFO comm 0x95c46a0 rank 8 nranks 32 cudaDev 0 nvmlDev 0 busId 53000 commId 0x1214b375c22bac3c - Init START
+ip-26-0-163-58:934721:934885 [1] NCCL INFO comm 0xa3091e0 rank 9 nranks 32 cudaDev 1 nvmlDev 1 busId 64000 commId 0x1214b375c22bac3c - Init START
+ip-26-0-163-58:934723:934887 [3] NCCL INFO comm 0x98861d0 rank 11 nranks 32 cudaDev 3 nvmlDev 3 busId 86000 commId 0x1214b375c22bac3c - Init START
+ip-26-0-163-58:934727:934886 [7] NCCL INFO comm 0x97de120 rank 15 nranks 32 cudaDev 7 nvmlDev 7 busId ca000 commId 0x1214b375c22bac3c - Init START
+ip-26-0-163-58:934725:934883 [5] NCCL INFO comm 0x886a710 rank 13 nranks 32 cudaDev 5 nvmlDev 5 busId a8000 commId 0x1214b375c22bac3c - Init START
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO comm 0x8fe8d30 rank 3 nranks 32 cudaDev 3 nvmlDev 3 busId 86000 commId 0x1214b375c22bac3c - Init START
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO comm 0x94864f0 rank 4 nranks 32 cudaDev 4 nvmlDev 4 busId 97000 commId 0x1214b375c22bac3c - Init START
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO comm 0xa132980 rank 1 nranks 32 cudaDev 1 nvmlDev 1 busId 64000 commId 0x1214b375c22bac3c - Init START
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO comm 0x962f910 rank 0 nranks 32 cudaDev 0 nvmlDev 0 busId 53000 commId 0x1214b375c22bac3c - Init START
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO comm 0x93cf5d0 rank 17 nranks 32 cudaDev 1 nvmlDev 1 busId 64000 commId 0x1214b375c22bac3c - Init START
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO comm 0x9296ef0 rank 16 nranks 32 cudaDev 0 nvmlDev 0 busId 53000 commId 0x1214b375c22bac3c - Init START
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO comm 0x8cd8e90 rank 19 nranks 32 cudaDev 3 nvmlDev 3 busId 86000 commId 0x1214b375c22bac3c - Init START
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO comm 0x856fc20 rank 22 nranks 32 cudaDev 6 nvmlDev 6 busId b9000 commId 0x1214b375c22bac3c - Init START
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO comm 0xa2b01f0 rank 21 nranks 32 cudaDev 5 nvmlDev 5 busId a8000 commId 0x1214b375c22bac3c - Init START
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO comm 0xa2e0240 rank 23 nranks 32 cudaDev 7 nvmlDev 7 busId ca000 commId 0x1214b375c22bac3c - Init START
+ip-26-0-170-31:3197187:3197356 [7] NCCL INFO comm 0xa426700 rank 31 nranks 32 cudaDev 7 nvmlDev 7 busId ca000 commId 0x1214b375c22bac3c - Init START
+ip-26-0-170-31:3197186:3197354 [6] NCCL INFO comm 0x9df89f0 rank 30 nranks 32 cudaDev 6 nvmlDev 6 busId b9000 commId 0x1214b375c22bac3c - Init START
+ip-26-0-170-31:3197185:3197353 [5] NCCL INFO comm 0x8610a20 rank 29 nranks 32 cudaDev 5 nvmlDev 5 busId a8000 commId 0x1214b375c22bac3c - Init START
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO comm 0x9ec45d0 rank 28 nranks 32 cudaDev 4 nvmlDev 4 busId 97000 commId 0x1214b375c22bac3c - Init START
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO comm 0x9f7c5a0 rank 2 nranks 32 cudaDev 2 nvmlDev 2 busId 75000 commId 0x1214b375c22bac3c - Init START
+ip-26-0-163-58:934724:934881 [4] NCCL INFO comm 0x967ca40 rank 12 nranks 32 cudaDev 4 nvmlDev 4 busId 97000 commId 0x1214b375c22bac3c - Init START
+ip-26-0-163-58:934722:934880 [2] NCCL INFO comm 0x8886e30 rank 10 nranks 32 cudaDev 2 nvmlDev 2 busId 75000 commId 0x1214b375c22bac3c - Init START
+ip-26-0-163-58:934726:934884 [6] NCCL INFO comm 0x9bcf110 rank 14 nranks 32 cudaDev 6 nvmlDev 6 busId b9000 commId 0x1214b375c22bac3c - Init START
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO comm 0x97fbc80 rank 20 nranks 32 cudaDev 4 nvmlDev 4 busId 97000 commId 0x1214b375c22bac3c - Init START
+ip-26-0-170-31:3197180:3197360 [0] NCCL INFO comm 0x9f60150 rank 24 nranks 32 cudaDev 0 nvmlDev 0 busId 53000 commId 0x1214b375c22bac3c - Init START
+ip-26-0-170-31:3197182:3197359 [2] NCCL INFO comm 0x86c2450 rank 26 nranks 32 cudaDev 2 nvmlDev 2 busId 75000 commId 0x1214b375c22bac3c - Init START
+ip-26-0-170-31:3197181:3197357 [1] NCCL INFO comm 0x998b390 rank 25 nranks 32 cudaDev 1 nvmlDev 1 busId 64000 commId 0x1214b375c22bac3c - Init START
+ip-26-0-170-31:3197183:3197358 [3] NCCL INFO comm 0x9d302b0 rank 27 nranks 32 cudaDev 3 nvmlDev 3 busId 86000 commId 0x1214b375c22bac3c - Init START
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO comm 0x985f5f0 rank 18 nranks 32 cudaDev 2 nvmlDev 2 busId 75000 commId 0x1214b375c22bac3c - Init START
+ip-26-0-163-58:934725:934883 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-163-58:934726:934884 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-163-58:934727:934886 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-163-58:934724:934881 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-163-58:934723:934887 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-163-58:934721:934885 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-170-31:3197186:3197354 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-170-31:3197182:3197359 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-170-31:3197183:3197358 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-170-31:3197187:3197356 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-170-31:3197181:3197357 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-170-31:3197180:3197360 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-170-31:3197185:3197353 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/253
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-163-58:934722:934880 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-163-58:934720:934882 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-163-58:934721:934885 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-163-58:934721:934885 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-170-31:3197181:3197357 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-170-31:3197181:3197357 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-170-31:3197183:3197358 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-170-31:3197183:3197358 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-163-58:934727:934886 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-163-58:934727:934886 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-170-31:3197187:3197356 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-170-31:3197187:3197356 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-163-58:934723:934887 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-163-58:934723:934887 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-170-31:3197180:3197360 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-170-31:3197180:3197360 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-170-31:3197185:3197353 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-170-31:3197185:3197353 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-170-31:3197182:3197359 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-170-31:3197182:3197359 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-170-31:3197186:3197354 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-170-31:3197186:3197354 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-163-58:934726:934884 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-163-58:934726:934884 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-163-58:934720:934882 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-163-58:934720:934882 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-163-58:934724:934881 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-163-58:934724:934881 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-163-58:934725:934883 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-163-58:934725:934883 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-163-58:934722:934880 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-163-58:934722:934880 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO NVLS Head  0:  0  8 16 24
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO NVLS Head  1:  1  9 17 25
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO NVLS Head  2:  2 10 18 26
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO NVLS Head  3:  3 11 19 27
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO NVLS Head  4:  4 12 20 28
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO NVLS Head  5:  5 13 21 29
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO NVLS Head  6:  6 14 22 30
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO NVLS Head  7:  7 15 23 31
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO Trees [0] 3/-1/-1->2->1 [1] 3/-1/-1->2->1 [2] 3/18/-1->2->-1 [3] -1/-1/-1->2->1 [4] 3/-1/-1->2->1 [5] 3/-1/-1->2->1 [6] 3/-1/-1->2->1 [7] 3/-1/-1->2->1 [8] 3/-1/-1->2->1 [9] 3/-1/-1->2->1 [10] 3/-1/-1->2->10 [11] -1/-1/-1->2->1 [12] 3/-1/-1->2->1 [13] 3/-1/-1->2->1 [14] 3/-1/-1->2->1 [15] 3/-1/-1->2->1
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO NVLS Head  0:  0  8 16 24
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO NVLS Head  1:  1  9 17 25
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO NVLS Head  2:  2 10 18 26
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO NVLS Head  3:  3 11 19 27
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO NVLS Head  4:  4 12 20 28
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO NVLS Head  5:  5 13 21 29
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO NVLS Head  6:  6 14 22 30
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO NVLS Head  7:  7 15 23 31
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO Trees [0] 4/-1/-1->3->2 [1] 4/-1/-1->3->2 [2] 4/-1/-1->3->2 [3] 4/19/-1->3->-1 [4] -1/-1/-1->3->2 [5] 4/-1/-1->3->2 [6] 4/-1/-1->3->2 [7] 4/-1/-1->3->2 [8] 4/-1/-1->3->2 [9] 4/-1/-1->3->2 [10] 4/-1/-1->3->2 [11] 4/-1/-1->3->11 [12] -1/-1/-1->3->2 [13] 4/-1/-1->3->2 [14] 4/-1/-1->3->2 [15] 4/-1/-1->3->2
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO NVLS Head  0:  0  8 16 24
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO NVLS Head  0:  0  8 16 24
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO NVLS Head  1:  1  9 17 25
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO NVLS Head  1:  1  9 17 25
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO NVLS Head  2:  2 10 18 26
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO NVLS Head  2:  2 10 18 26
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO NVLS Head  3:  3 11 19 27
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO NVLS Head  3:  3 11 19 27
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO NVLS Head  0:  0  8 16 24
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO NVLS Head  4:  4 12 20 28
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO NVLS Head  4:  4 12 20 28
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO NVLS Head  0:  0  8 16 24
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO NVLS Head  5:  5 13 21 29
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO NVLS Head  5:  5 13 21 29
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO NVLS Head  1:  1  9 17 25
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO NVLS Head  1:  1  9 17 25
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO NVLS Head  6:  6 14 22 30
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO NVLS Head  6:  6 14 22 30
+ip-26-0-170-31:3197187:3197356 [7] NCCL INFO Trees [0] -1/-1/-1->31->30 [1] 24/-1/-1->31->30 [2] 24/-1/-1->31->30 [3] 24/-1/-1->31->30 [4] 24/-1/-1->31->30 [5] 24/-1/-1->31->30 [6] 24/-1/-1->31->30 [7] 24/-1/-1->31->23 [8] -1/-1/-1->31->30 [9] 24/-1/-1->31->30 [10] 24/-1/-1->31->30 [11] 24/-1/-1->31->30 [12] 24/-1/-1->31->30 [13] 24/-1/-1->31->30 [14] 24/-1/-1->31->30 [15] 24/15/-1->31->-1
+ip-26-0-170-31:3197187:3197356 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO NVLS Head  2:  2 10 18 26
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO NVLS Head  2:  2 10 18 26
+ip-26-0-163-58:934720:934882 [0] NCCL INFO Trees [0] 9/-1/-1->8->16 [1] -1/-1/-1->8->15 [2] 9/-1/-1->8->15 [3] 9/-1/-1->8->15 [4] 9/-1/-1->8->15 [5] 9/-1/-1->8->15 [6] 9/-1/-1->8->15 [7] 9/-1/-1->8->15 [8] 9/16/0->8->24 [9] -1/-1/-1->8->15 [10] 9/-1/-1->8->15 [11] 9/-1/-1->8->15 [12] 9/-1/-1->8->15 [13] 9/-1/-1->8->15 [14] 9/-1/-1->8->15 [15] 9/-1/-1->8->15
+ip-26-0-163-58:934727:934886 [7] NCCL INFO Trees [0] -1/-1/-1->15->14 [1] 8/-1/-1->15->14 [2] 8/-1/-1->15->14 [3] 8/-1/-1->15->14 [4] 8/-1/-1->15->14 [5] 8/-1/-1->15->14 [6] 8/-1/-1->15->14 [7] 8/-1/-1->15->23 [8] -1/-1/-1->15->14 [9] 8/-1/-1->15->14 [10] 8/-1/-1->15->14 [11] 8/-1/-1->15->14 [12] 8/-1/-1->15->14 [13] 8/-1/-1->15->14 [14] 8/-1/-1->15->14 [15] 8/23/7->15->31
+ip-26-0-163-58:934720:934882 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO Trees [0] 18/-1/-1->17->16 [1] 18/9/25->17->1 [2] -1/-1/-1->17->16 [3] 18/-1/-1->17->16 [4] 18/-1/-1->17->16 [5] 18/-1/-1->17->16 [6] 18/-1/-1->17->16 [7] 18/-1/-1->17->16 [8] 18/-1/-1->17->16 [9] 18/-1/-1->17->9 [10] -1/-1/-1->17->16 [11] 18/-1/-1->17->16 [12] 18/-1/-1->17->16 [13] 18/-1/-1->17->16 [14] 18/-1/-1->17->16 [15] 18/-1/-1->17->16
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO Trees [0] 19/-1/-1->18->17 [1] 19/-1/-1->18->17 [2] 19/10/26->18->2 [3] -1/-1/-1->18->17 [4] 19/-1/-1->18->17 [5] 19/-1/-1->18->17 [6] 19/-1/-1->18->17 [7] 19/-1/-1->18->17 [8] 19/-1/-1->18->17 [9] 19/-1/-1->18->17 [10] 19/-1/-1->18->10 [11] -1/-1/-1->18->17 [12] 19/-1/-1->18->17 [13] 19/-1/-1->18->17 [14] 19/-1/-1->18->17 [15] 19/-1/-1->18->17
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO Trees [0] 29/-1/-1->28->27 [1] 29/-1/-1->28->27 [2] 29/-1/-1->28->27 [3] 29/-1/-1->28->27 [4] 29/-1/-1->28->20 [5] -1/-1/-1->28->27 [6] 29/-1/-1->28->27 [7] 29/-1/-1->28->27 [8] 29/-1/-1->28->27 [9] 29/-1/-1->28->27 [10] 29/-1/-1->28->27 [11] 29/-1/-1->28->27 [12] 29/12/-1->28->-1 [13] -1/-1/-1->28->27 [14] 29/-1/-1->28->27 [15] 29/-1/-1->28->27
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-170-31:3197180:3197360 [0] NCCL INFO Trees [0] 25/-1/-1->24->16 [1] -1/-1/-1->24->31 [2] 25/-1/-1->24->31 [3] 25/-1/-1->24->31 [4] 25/-1/-1->24->31 [5] 25/-1/-1->24->31 [6] 25/-1/-1->24->31 [7] 25/-1/-1->24->31 [8] 25/8/-1->24->-1 [9] -1/-1/-1->24->31 [10] 25/-1/-1->24->31 [11] 25/-1/-1->24->31 [12] 25/-1/-1->24->31 [13] 25/-1/-1->24->31 [14] 25/-1/-1->24->31 [15] 25/-1/-1->24->31
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO NVLS Head  7:  7 15 23 31
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO NVLS Head  7:  7 15 23 31
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO NVLS Head  3:  3 11 19 27
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO NVLS Head  3:  3 11 19 27
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO NVLS Head  4:  4 12 20 28
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO NVLS Head  4:  4 12 20 28
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO NVLS Head  5:  5 13 21 29
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO NVLS Head  5:  5 13 21 29
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO NVLS Head  6:  6 14 22 30
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO NVLS Head  6:  6 14 22 30
+ip-26-0-163-58:934725:934883 [5] NCCL INFO Trees [0] 14/-1/-1->13->12 [1] 14/-1/-1->13->12 [2] 14/-1/-1->13->12 [3] 14/-1/-1->13->12 [4] 14/-1/-1->13->12 [5] 14/-1/-1->13->21 [6] -1/-1/-1->13->12 [7] 14/-1/-1->13->12 [8] 14/-1/-1->13->12 [9] 14/-1/-1->13->12 [10] 14/-1/-1->13->12 [11] 14/-1/-1->13->12 [12] 14/-1/-1->13->12 [13] 14/21/5->13->29 [14] -1/-1/-1->13->12 [15] 14/-1/-1->13->12
+ip-26-0-163-58:934727:934886 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-163-58:934722:934880 [2] NCCL INFO Trees [0] 11/-1/-1->10->9 [1] 11/-1/-1->10->9 [2] 11/-1/-1->10->18 [3] -1/-1/-1->10->9 [4] 11/-1/-1->10->9 [5] 11/-1/-1->10->9 [6] 11/-1/-1->10->9 [7] 11/-1/-1->10->9 [8] 11/-1/-1->10->9 [9] 11/-1/-1->10->9 [10] 11/18/2->10->26 [11] -1/-1/-1->10->9 [12] 11/-1/-1->10->9 [13] 11/-1/-1->10->9 [14] 11/-1/-1->10->9 [15] 11/-1/-1->10->9
+ip-26-0-163-58:934725:934883 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO Trees [0] -1/-1/-1->23->22 [1] 16/-1/-1->23->22 [2] 16/-1/-1->23->22 [3] 16/-1/-1->23->22 [4] 16/-1/-1->23->22 [5] 16/-1/-1->23->22 [6] 16/-1/-1->23->22 [7] 16/15/31->23->7 [8] -1/-1/-1->23->22 [9] 16/-1/-1->23->22 [10] 16/-1/-1->23->22 [11] 16/-1/-1->23->22 [12] 16/-1/-1->23->22 [13] 16/-1/-1->23->22 [14] 16/-1/-1->23->22 [15] 16/-1/-1->23->15
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-170-31:3197183:3197358 [3] NCCL INFO Trees [0] 28/-1/-1->27->26 [1] 28/-1/-1->27->26 [2] 28/-1/-1->27->26 [3] 28/-1/-1->27->19 [4] -1/-1/-1->27->26 [5] 28/-1/-1->27->26 [6] 28/-1/-1->27->26 [7] 28/-1/-1->27->26 [8] 28/-1/-1->27->26 [9] 28/-1/-1->27->26 [10] 28/-1/-1->27->26 [11] 28/11/-1->27->-1 [12] -1/-1/-1->27->26 [13] 28/-1/-1->27->26 [14] 28/-1/-1->27->26 [15] 28/-1/-1->27->26
+ip-26-0-170-31:3197180:3197360 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-170-31:3197183:3197358 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-170-31:3197186:3197354 [6] NCCL INFO Trees [0] 31/-1/-1->30->29 [1] 31/-1/-1->30->29 [2] 31/-1/-1->30->29 [3] 31/-1/-1->30->29 [4] 31/-1/-1->30->29 [5] 31/-1/-1->30->29 [6] 31/-1/-1->30->22 [7] -1/-1/-1->30->29 [8] 31/-1/-1->30->29 [9] 31/-1/-1->30->29 [10] 31/-1/-1->30->29 [11] 31/-1/-1->30->29 [12] 31/-1/-1->30->29 [13] 31/-1/-1->30->29 [14] 31/14/-1->30->-1 [15] -1/-1/-1->30->29
+ip-26-0-170-31:3197186:3197354 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO NVLS Head  7:  7 15 23 31
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO NVLS Head  7:  7 15 23 31
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO NVLS Head  0:  0  8 16 24
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO NVLS Head  0:  0  8 16 24
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO NVLS Head  1:  1  9 17 25
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO NVLS Head  1:  1  9 17 25
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO NVLS Head  2:  2 10 18 26
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO NVLS Head  2:  2 10 18 26
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO Trees [0] -1/-1/-1->7->6 [1] 0/-1/-1->7->6 [2] 0/-1/-1->7->6 [3] 0/-1/-1->7->6 [4] 0/-1/-1->7->6 [5] 0/-1/-1->7->6 [6] 0/-1/-1->7->6 [7] 0/23/-1->7->-1 [8] -1/-1/-1->7->6 [9] 0/-1/-1->7->6 [10] 0/-1/-1->7->6 [11] 0/-1/-1->7->6 [12] 0/-1/-1->7->6 [13] 0/-1/-1->7->6 [14] 0/-1/-1->7->6 [15] 0/-1/-1->7->15
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO NVLS Head  3:  3 11 19 27
+ip-26-0-163-58:934724:934881 [4] NCCL INFO Trees [0] 13/-1/-1->12->11 [1] 13/-1/-1->12->11 [2] 13/-1/-1->12->11 [3] 13/-1/-1->12->11 [4] 13/-1/-1->12->20 [5] -1/-1/-1->12->11 [6] 13/-1/-1->12->11 [7] 13/-1/-1->12->11 [8] 13/-1/-1->12->11 [9] 13/-1/-1->12->11 [10] 13/-1/-1->12->11 [11] 13/-1/-1->12->11 [12] 13/20/4->12->28 [13] -1/-1/-1->12->11 [14] 13/-1/-1->12->11 [15] 13/-1/-1->12->11
+ip-26-0-163-58:934726:934884 [6] NCCL INFO Trees [0] 15/-1/-1->14->13 [1] 15/-1/-1->14->13 [2] 15/-1/-1->14->13 [3] 15/-1/-1->14->13 [4] 15/-1/-1->14->13 [5] 15/-1/-1->14->13 [6] 15/-1/-1->14->22 [7] -1/-1/-1->14->13 [8] 15/-1/-1->14->13 [9] 15/-1/-1->14->13 [10] 15/-1/-1->14->13 [11] 15/-1/-1->14->13 [12] 15/-1/-1->14->13 [13] 15/-1/-1->14->13 [14] 15/22/6->14->30 [15] -1/-1/-1->14->13
+ip-26-0-163-58:934722:934880 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-163-58:934724:934881 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-163-58:934726:934884 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO Trees [0] 21/-1/-1->20->19 [1] 21/-1/-1->20->19 [2] 21/-1/-1->20->19 [3] 21/-1/-1->20->19 [4] 21/12/28->20->4 [5] -1/-1/-1->20->19 [6] 21/-1/-1->20->19 [7] 21/-1/-1->20->19 [8] 21/-1/-1->20->19 [9] 21/-1/-1->20->19 [10] 21/-1/-1->20->19 [11] 21/-1/-1->20->19 [12] 21/-1/-1->20->12 [13] -1/-1/-1->20->19 [14] 21/-1/-1->20->19 [15] 21/-1/-1->20->19
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO Trees [0] 17/8/24->16->0 [1] -1/-1/-1->16->23 [2] 17/-1/-1->16->23 [3] 17/-1/-1->16->23 [4] 17/-1/-1->16->23 [5] 17/-1/-1->16->23 [6] 17/-1/-1->16->23 [7] 17/-1/-1->16->23 [8] 17/-1/-1->16->8 [9] -1/-1/-1->16->23 [10] 17/-1/-1->16->23 [11] 17/-1/-1->16->23 [12] 17/-1/-1->16->23 [13] 17/-1/-1->16->23 [14] 17/-1/-1->16->23 [15] 17/-1/-1->16->23
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO NVLS Head  3:  3 11 19 27
+ip-26-0-163-58:934721:934885 [1] NCCL INFO Trees [0] 10/-1/-1->9->8 [1] 10/-1/-1->9->17 [2] -1/-1/-1->9->8 [3] 10/-1/-1->9->8 [4] 10/-1/-1->9->8 [5] 10/-1/-1->9->8 [6] 10/-1/-1->9->8 [7] 10/-1/-1->9->8 [8] 10/-1/-1->9->8 [9] 10/17/1->9->25 [10] -1/-1/-1->9->8 [11] 10/-1/-1->9->8 [12] 10/-1/-1->9->8 [13] 10/-1/-1->9->8 [14] 10/-1/-1->9->8 [15] 10/-1/-1->9->8
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO Trees [0] 20/-1/-1->19->18 [1] 20/-1/-1->19->18 [2] 20/-1/-1->19->18 [3] 20/11/27->19->3 [4] -1/-1/-1->19->18 [5] 20/-1/-1->19->18 [6] 20/-1/-1->19->18 [7] 20/-1/-1->19->18 [8] 20/-1/-1->19->18 [9] 20/-1/-1->19->18 [10] 20/-1/-1->19->18 [11] 20/-1/-1->19->11 [12] -1/-1/-1->19->18 [13] 20/-1/-1->19->18 [14] 20/-1/-1->19->18 [15] 20/-1/-1->19->18
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-163-58:934721:934885 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-163-58:934723:934887 [3] NCCL INFO Trees [0] 12/-1/-1->11->10 [1] 12/-1/-1->11->10 [2] 12/-1/-1->11->10 [3] 12/-1/-1->11->19 [4] -1/-1/-1->11->10 [5] 12/-1/-1->11->10 [6] 12/-1/-1->11->10 [7] 12/-1/-1->11->10 [8] 12/-1/-1->11->10 [9] 12/-1/-1->11->10 [10] 12/-1/-1->11->10 [11] 12/19/3->11->27 [12] -1/-1/-1->11->10 [13] 12/-1/-1->11->10 [14] 12/-1/-1->11->10 [15] 12/-1/-1->11->10
+ip-26-0-170-31:3197182:3197359 [2] NCCL INFO Trees [0] 27/-1/-1->26->25 [1] 27/-1/-1->26->25 [2] 27/-1/-1->26->18 [3] -1/-1/-1->26->25 [4] 27/-1/-1->26->25 [5] 27/-1/-1->26->25 [6] 27/-1/-1->26->25 [7] 27/-1/-1->26->25 [8] 27/-1/-1->26->25 [9] 27/-1/-1->26->25 [10] 27/10/-1->26->-1 [11] -1/-1/-1->26->25 [12] 27/-1/-1->26->25 [13] 27/-1/-1->26->25 [14] 27/-1/-1->26->25 [15] 27/-1/-1->26->25
+ip-26-0-170-31:3197185:3197353 [5] NCCL INFO Trees [0] 30/-1/-1->29->28 [1] 30/-1/-1->29->28 [2] 30/-1/-1->29->28 [3] 30/-1/-1->29->28 [4] 30/-1/-1->29->28 [5] 30/-1/-1->29->21 [6] -1/-1/-1->29->28 [7] 30/-1/-1->29->28 [8] 30/-1/-1->29->28 [9] 30/-1/-1->29->28 [10] 30/-1/-1->29->28 [11] 30/-1/-1->29->28 [12] 30/-1/-1->29->28 [13] 30/13/-1->29->-1 [14] -1/-1/-1->29->28 [15] 30/-1/-1->29->28
+ip-26-0-170-31:3197182:3197359 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-170-31:3197185:3197353 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-163-58:934723:934887 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO Trees [0] 23/-1/-1->22->21 [1] 23/-1/-1->22->21 [2] 23/-1/-1->22->21 [3] 23/-1/-1->22->21 [4] 23/-1/-1->22->21 [5] 23/-1/-1->22->21 [6] 23/14/30->22->6 [7] -1/-1/-1->22->21 [8] 23/-1/-1->22->21 [9] 23/-1/-1->22->21 [10] 23/-1/-1->22->21 [11] 23/-1/-1->22->21 [12] 23/-1/-1->22->21 [13] 23/-1/-1->22->21 [14] 23/-1/-1->22->14 [15] -1/-1/-1->22->21
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-170-31:3197181:3197357 [1] NCCL INFO Trees [0] 26/-1/-1->25->24 [1] 26/-1/-1->25->17 [2] -1/-1/-1->25->24 [3] 26/-1/-1->25->24 [4] 26/-1/-1->25->24 [5] 26/-1/-1->25->24 [6] 26/-1/-1->25->24 [7] 26/-1/-1->25->24 [8] 26/-1/-1->25->24 [9] 26/9/-1->25->-1 [10] -1/-1/-1->25->24 [11] 26/-1/-1->25->24 [12] 26/-1/-1->25->24 [13] 26/-1/-1->25->24 [14] 26/-1/-1->25->24 [15] 26/-1/-1->25->24
+ip-26-0-170-31:3197181:3197357 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO Trees [0] 22/-1/-1->21->20 [1] 22/-1/-1->21->20 [2] 22/-1/-1->21->20 [3] 22/-1/-1->21->20 [4] 22/-1/-1->21->20 [5] 22/13/29->21->5 [6] -1/-1/-1->21->20 [7] 22/-1/-1->21->20 [8] 22/-1/-1->21->20 [9] 22/-1/-1->21->20 [10] 22/-1/-1->21->20 [11] 22/-1/-1->21->20 [12] 22/-1/-1->21->20 [13] 22/-1/-1->21->13 [14] -1/-1/-1->21->20 [15] 22/-1/-1->21->20
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-170-31:3197180:3197360 [0] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-163-58:934720:934882 [0] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO Channel 00/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-163-58:934725:934883 [5] NCCL INFO Channel 00/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-163-58:934722:934880 [2] NCCL INFO Channel 00/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-163-58:934726:934884 [6] NCCL INFO Channel 00/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO Channel 08/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-170-31:3197183:3197358 [3] NCCL INFO Channel 00/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO Channel 00/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-163-58:934725:934883 [5] NCCL INFO Channel 08/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-170-31:3197185:3197353 [5] NCCL INFO Channel 00/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-163-58:934722:934880 [2] NCCL INFO Channel 08/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-170-31:3197183:3197358 [3] NCCL INFO Channel 08/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-163-58:934726:934884 [6] NCCL INFO Channel 08/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO Channel 08/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-170-31:3197186:3197354 [6] NCCL INFO Channel 00/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-170-31:3197185:3197353 [5] NCCL INFO Channel 08/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO Channel 00/0 : 23[7] -> 24[0] [send] via NET/Libfabric/0(16)/GDRDMA
+ip-26-0-163-58:934724:934881 [4] NCCL INFO Channel 00/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-163-58:934723:934887 [3] NCCL INFO Channel 00/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-163-58:934724:934881 [4] NCCL INFO Channel 08/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-170-31:3197186:3197354 [6] NCCL INFO Channel 08/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO Channel 05/0 : 28[4] -> 31[7] via P2P/IPC
+ip-26-0-163-58:934721:934885 [1] NCCL INFO Channel 00/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-170-31:3197182:3197359 [2] NCCL INFO Channel 00/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-163-58:934723:934887 [3] NCCL INFO Channel 08/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO Channel 00/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-163-58:934721:934885 [1] NCCL INFO Channel 08/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO Channel 00/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO Channel 00/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO Channel 06/0 : 28[4] -> 31[7] via P2P/IPC
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO Channel 00/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-170-31:3197182:3197359 [2] NCCL INFO Channel 08/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-170-31:3197181:3197357 [1] NCCL INFO Channel 00/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO Channel 08/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-163-58:934724:934881 [4] NCCL INFO Channel 05/0 : 12[4] -> 15[7] via P2P/IPC
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO Channel 08/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO Channel 13/0 : 28[4] -> 31[7] via P2P/IPC
+ip-26-0-163-58:934722:934880 [2] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-163-58:934723:934887 [3] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO Channel 08/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-163-58:934724:934881 [4] NCCL INFO Channel 06/0 : 12[4] -> 15[7] via P2P/IPC
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO Channel 08/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-163-58:934724:934881 [4] NCCL INFO Channel 13/0 : 12[4] -> 15[7] via P2P/IPC
+ip-26-0-170-31:3197181:3197357 [1] NCCL INFO Channel 08/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO Channel 14/0 : 28[4] -> 31[7] via P2P/IPC
+ip-26-0-170-31:3197183:3197358 [3] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-163-58:934724:934881 [4] NCCL INFO Channel 14/0 : 12[4] -> 15[7] via P2P/IPC
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-170-31:3197182:3197359 [2] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO Channel 00/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO Channel 08/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO Channel 05/0 : 20[4] -> 23[7] via P2P/IPC
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO Channel 06/0 : 20[4] -> 23[7] via P2P/IPC
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO Channel 13/0 : 20[4] -> 23[7] via P2P/IPC
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO Channel 14/0 : 20[4] -> 23[7] via P2P/IPC
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO NVLS Head  4:  4 12 20 28
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO Trees [0] 5/-1/-1->4->3 [1] 5/-1/-1->4->3 [2] 5/-1/-1->4->3 [3] 5/-1/-1->4->3 [4] 5/20/-1->4->-1 [5] -1/-1/-1->4->3 [6] 5/-1/-1->4->3 [7] 5/-1/-1->4->3 [8] 5/-1/-1->4->3 [9] 5/-1/-1->4->3 [10] 5/-1/-1->4->3 [11] 5/-1/-1->4->3 [12] 5/-1/-1->4->12 [13] -1/-1/-1->4->3 [14] 5/-1/-1->4->3 [15] 5/-1/-1->4->3
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO Trees [0] 6/-1/-1->5->4 [1] 6/-1/-1->5->4 [2] 6/-1/-1->5->4 [3] 6/-1/-1->5->4 [4] 6/-1/-1->5->4 [5] 6/21/-1->5->-1 [6] -1/-1/-1->5->4 [7] 6/-1/-1->5->4 [8] 6/-1/-1->5->4 [9] 6/-1/-1->5->4 [10] 6/-1/-1->5->4 [11] 6/-1/-1->5->4 [12] 6/-1/-1->5->4 [13] 6/-1/-1->5->13 [14] -1/-1/-1->5->4 [15] 6/-1/-1->5->4
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO NVLS Head  4:  4 12 20 28
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO Trees [0] 7/-1/-1->6->5 [1] 7/-1/-1->6->5 [2] 7/-1/-1->6->5 [3] 7/-1/-1->6->5 [4] 7/-1/-1->6->5 [5] 7/-1/-1->6->5 [6] 7/22/-1->6->-1 [7] -1/-1/-1->6->5 [8] 7/-1/-1->6->5 [9] 7/-1/-1->6->5 [10] 7/-1/-1->6->5 [11] 7/-1/-1->6->5 [12] 7/-1/-1->6->5 [13] 7/-1/-1->6->5 [14] 7/-1/-1->6->14 [15] -1/-1/-1->6->5
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO NVLS Head  5:  5 13 21 29
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO NVLS Head  6:  6 14 22 30
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO NVLS Head  5:  5 13 21 29
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO NVLS Head  7:  7 15 23 31
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO NVLS Head  6:  6 14 22 30
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO NVLS Head  7:  7 15 23 31
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO Channel 00/16 :    0   1   2   3   4   5   6   7   8   9  10  11  12  13  14  15  16  17  18  19
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO Channel 01/16 :    0   7   6   5   4   3   2   9   8  15  14  13  12  11  10  17  16  23  22  21
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO Channel 02/16 :    0   7   6   5   4   3  10   9   8  15  14  13  12  11  18  17  16  23  22  21
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO Channel 03/16 :    0   7   6   5   4  11  10   9   8  15  14  13  12  19  18  17  16  23  22  21
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO Channel 04/16 :    0   7   6   5  12  11  10   9   8  15  14  13  20  19  18  17  16  23  22  21
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO Channel 05/16 :    0   4   7   6  13  11  10   9   8  12  15  14  21  19  18  17  16  20  23  22
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO Trees [0] 2/-1/-1->1->0 [1] 2/17/-1->1->-1 [2] -1/-1/-1->1->0 [3] 2/-1/-1->1->0 [4] 2/-1/-1->1->0 [5] 2/-1/-1->1->0 [6] 2/-1/-1->1->0 [7] 2/-1/-1->1->0 [8] 2/-1/-1->1->0 [9] 2/-1/-1->1->9 [10] -1/-1/-1->1->0 [11] 2/-1/-1->1->0 [12] 2/-1/-1->1->0 [13] 2/-1/-1->1->0 [14] 2/-1/-1->1->0 [15] 2/-1/-1->1->0
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO Channel 06/16 :    0   5   4   7  14  11  10   9   8  13  12  15  22  19  18  17  16  21  20  23
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO Channel 07/16 :    0   6   5   4  15  11  10   9   8  14  13  12  23  19  18  17  16  22  21  20
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO Channel 08/16 :    0   1   2   3   4   5   6   7   8   9  10  11  12  13  14  15  16  17  18  19
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO Channel 09/16 :    0   7   6   5   4   3   2   9   8  15  14  13  12  11  10  17  16  23  22  21
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO Channel 10/16 :    0   7   6   5   4   3  10   9   8  15  14  13  12  11  18  17  16  23  22  21
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO Channel 11/16 :    0   7   6   5   4  11  10   9   8  15  14  13  12  19  18  17  16  23  22  21
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO Channel 12/16 :    0   7   6   5  12  11  10   9   8  15  14  13  20  19  18  17  16  23  22  21
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO Channel 13/16 :    0   4   7   6  13  11  10   9   8  12  15  14  21  19  18  17  16  20  23  22
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO Channel 14/16 :    0   5   4   7  14  11  10   9   8  13  12  15  22  19  18  17  16  21  20  23
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO Channel 15/16 :    0   6   5   4  15  11  10   9   8  14  13  12  23  19  18  17  16  22  21  20
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO Trees [0] 1/16/-1->0->-1 [1] -1/-1/-1->0->7 [2] 1/-1/-1->0->7 [3] 1/-1/-1->0->7 [4] 1/-1/-1->0->7 [5] 1/-1/-1->0->7 [6] 1/-1/-1->0->7 [7] 1/-1/-1->0->7 [8] 1/-1/-1->0->8 [9] -1/-1/-1->0->7 [10] 1/-1/-1->0->7 [11] 1/-1/-1->0->7 [12] 1/-1/-1->0->7 [13] 1/-1/-1->0->7 [14] 1/-1/-1->0->7 [15] 1/-1/-1->0->7
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO Channel 00/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO Channel 08/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO Channel 00/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO Channel 00/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO Channel 00/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO Channel 00/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO Channel 08/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO Channel 00/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO Channel 08/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO Channel 08/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO Channel 08/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO Channel 08/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO Channel 05/0 : 4[4] -> 7[7] via P2P/IPC
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO Channel 06/0 : 4[4] -> 7[7] via P2P/IPC
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO Channel 13/0 : 4[4] -> 7[7] via P2P/IPC
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO Channel 14/0 : 4[4] -> 7[7] via P2P/IPC
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO Channel 00/0 : 15[7] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO Channel 08/0 : 23[7] -> 24[0] [send] via NET/Libfabric/0(16)/GDRDMA
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO Channel 08/0 : 15[7] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO Channel 00/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-163-58:934720:934882 [0] NCCL INFO Channel 00/0 : 7[7] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934727:934886 [7] NCCL INFO Channel 00/0 : 15[7] -> 16[0] [send] via NET/Libfabric/0(8)/GDRDMA
+ip-26-0-163-58:934720:934882 [0] NCCL INFO Channel 08/0 : 7[7] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934720:934882 [0] NCCL INFO Channel 00/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-170-31:3197180:3197360 [0] NCCL INFO Channel 00/0 : 23[7] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934727:934886 [7] NCCL INFO Channel 08/0 : 15[7] -> 16[0] [send] via NET/Libfabric/0(8)/GDRDMA
+ip-26-0-170-31:3197187:3197356 [7] NCCL INFO Channel 00/0 : 31[7] -> 0[0] [send] via NET/Libfabric/0(24)/GDRDMA
+ip-26-0-170-31:3197180:3197360 [0] NCCL INFO Channel 08/0 : 23[7] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-31:3197180:3197360 [0] NCCL INFO Channel 00/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO Channel 08/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-170-31:3197187:3197356 [7] NCCL INFO Channel 08/0 : 31[7] -> 0[0] [send] via NET/Libfabric/0(24)/GDRDMA
+ip-26-0-163-58:934720:934882 [0] NCCL INFO Channel 08/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO Channel 05/0 : 16[0] -> 20[4] via P2P/IPC
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO Channel 13/0 : 16[0] -> 20[4] via P2P/IPC
+ip-26-0-170-31:3197180:3197360 [0] NCCL INFO Channel 08/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-163-58:934721:934885 [1] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-170-31:3197181:3197357 [1] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-170-31:3197180:3197360 [0] NCCL INFO Channel 05/0 : 24[0] -> 28[4] via P2P/IPC
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO Channel 03/0 : 12[4] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO Channel 00/0 : 31[7] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO Channel 11/0 : 12[4] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO Channel 00/0 : 7[7] -> 8[0] [send] via NET/Libfabric/0(0)/GDRDMA
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO Channel 08/0 : 31[7] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-31:3197180:3197360 [0] NCCL INFO Channel 13/0 : 24[0] -> 28[4] via P2P/IPC
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO Channel 08/0 : 7[7] -> 8[0] [send] via NET/Libfabric/0(0)/GDRDMA
+ip-26-0-163-58:934720:934882 [0] NCCL INFO Channel 05/0 : 8[0] -> 12[4] via P2P/IPC
+ip-26-0-163-58:934720:934882 [0] NCCL INFO Channel 13/0 : 8[0] -> 12[4] via P2P/IPC
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO Channel 08/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-163-58:934720:934882 [0] NCCL INFO Channel 06/0 : 8[0] -> 13[5] via P2P/IPC
+ip-26-0-163-58:934724:934881 [4] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO Channel 02/0 : 11[3] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934720:934882 [0] NCCL INFO Channel 14/0 : 8[0] -> 13[5] via P2P/IPC
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO Channel 02/0 : 19[3] -> 26[2] [send] via NET/Libfabric/2(18)/GDRDMA
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO Channel 10/0 : 11[3] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO Channel 10/0 : 19[3] -> 26[2] [send] via NET/Libfabric/2(18)/GDRDMA
+ip-26-0-163-58:934725:934883 [5] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-163-58:934720:934882 [0] NCCL INFO Channel 07/0 : 8[0] -> 14[6] via P2P/IPC
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO Channel 05/0 : 0[0] -> 4[4] via P2P/IPC
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO Channel 01/0 : 18[2] -> 25[1] [send] via NET/Libfabric/1(17)/GDRDMA
+ip-26-0-170-31:3197182:3197359 [2] NCCL INFO Channel 02/0 : 19[3] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO Channel 09/0 : 18[2] -> 25[1] [send] via NET/Libfabric/1(17)/GDRDMA
+ip-26-0-163-58:934720:934882 [0] NCCL INFO Channel 15/0 : 8[0] -> 14[6] via P2P/IPC
+ip-26-0-170-31:3197182:3197359 [2] NCCL INFO Channel 10/0 : 19[3] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO Channel 13/0 : 0[0] -> 4[4] via P2P/IPC
+ip-26-0-163-58:934720:934882 [0] NCCL INFO Channel 01/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-170-31:3197182:3197359 [2] NCCL INFO Channel 01/0 : 26[2] -> 1[1] [send] via NET/Libfabric/1(25)/GDRDMA
+ip-26-0-163-58:934726:934884 [6] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-163-58:934722:934880 [2] NCCL INFO Channel 02/0 : 3[3] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-31:3197182:3197359 [2] NCCL INFO Channel 09/0 : 26[2] -> 1[1] [send] via NET/Libfabric/1(25)/GDRDMA
+ip-26-0-163-58:934722:934880 [2] NCCL INFO Channel 10/0 : 3[3] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-31:3197183:3197358 [3] NCCL INFO Channel 03/0 : 20[4] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934720:934882 [0] NCCL INFO Channel 02/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-170-31:3197183:3197358 [3] NCCL INFO Channel 11/0 : 20[4] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934727:934886 [7] NCCL INFO Channel 06/0 : 15[7] -> 22[6] [send] via NET/Libfabric/6(14)/GDRDMA
+ip-26-0-163-58:934720:934882 [0] NCCL INFO Channel 03/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-170-31:3197183:3197358 [3] NCCL INFO Channel 02/0 : 27[3] -> 2[2] [send] via NET/Libfabric/2(26)/GDRDMA
+ip-26-0-163-58:934727:934886 [7] NCCL INFO Channel 14/0 : 15[7] -> 22[6] [send] via NET/Libfabric/6(14)/GDRDMA
+ip-26-0-170-31:3197183:3197358 [3] NCCL INFO Channel 10/0 : 27[3] -> 2[2] [send] via NET/Libfabric/2(26)/GDRDMA
+ip-26-0-163-58:934720:934882 [0] NCCL INFO Channel 04/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-163-58:934723:934887 [3] NCCL INFO Channel 03/0 : 4[4] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934723:934887 [3] NCCL INFO Channel 11/0 : 4[4] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO Channel 02/0 : 27[3] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO Channel 10/0 : 27[3] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934723:934887 [3] NCCL INFO Channel 02/0 : 11[3] -> 18[2] [send] via NET/Libfabric/2(10)/GDRDMA
+ip-26-0-163-58:934723:934887 [3] NCCL INFO Channel 10/0 : 11[3] -> 18[2] [send] via NET/Libfabric/2(10)/GDRDMA
+ip-26-0-163-58:934720:934882 [0] NCCL INFO Channel 09/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-163-58:934720:934882 [0] NCCL INFO Channel 10/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO Channel 06/0 : 23[7] -> 30[6] [send] via NET/Libfabric/6(22)/GDRDMA
+ip-26-0-163-58:934720:934882 [0] NCCL INFO Channel 11/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO Channel 14/0 : 23[7] -> 30[6] [send] via NET/Libfabric/6(22)/GDRDMA
+ip-26-0-163-58:934720:934882 [0] NCCL INFO Channel 12/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO Channel 03/0 : 28[4] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO Channel 06/0 : 16[0] -> 21[5] via P2P/IPC
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO Channel 11/0 : 28[4] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO Channel 06/0 : 7[7] -> 14[6] [send] via NET/Libfabric/6(6)/GDRDMA
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO Channel 14/0 : 16[0] -> 21[5] via P2P/IPC
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO Channel 02/0 : 3[3] -> 10[2] [send] via NET/Libfabric/2(2)/GDRDMA
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO Channel 14/0 : 7[7] -> 14[6] [send] via NET/Libfabric/6(6)/GDRDMA
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO Channel 10/0 : 3[3] -> 10[2] [send] via NET/Libfabric/2(2)/GDRDMA
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO Channel 01/0 : 2[2] -> 9[1] [send] via NET/Libfabric/1(1)/GDRDMA
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO Channel 09/0 : 2[2] -> 9[1] [send] via NET/Libfabric/1(1)/GDRDMA
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO Channel 07/0 : 16[0] -> 22[6] via P2P/IPC
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO Channel 15/0 : 16[0] -> 22[6] via P2P/IPC
+ip-26-0-170-31:3197180:3197360 [0] NCCL INFO Channel 06/0 : 24[0] -> 29[5] via P2P/IPC
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO Channel 01/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-170-31:3197180:3197360 [0] NCCL INFO Channel 14/0 : 24[0] -> 29[5] via P2P/IPC
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO Channel 02/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-170-31:3197180:3197360 [0] NCCL INFO Channel 07/0 : 24[0] -> 30[6] via P2P/IPC
+ip-26-0-170-31:3197185:3197353 [5] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO Channel 03/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-170-31:3197180:3197360 [0] NCCL INFO Channel 15/0 : 24[0] -> 30[6] via P2P/IPC
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO Channel 04/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-170-31:3197180:3197360 [0] NCCL INFO Channel 01/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-170-31:3197186:3197354 [6] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-170-31:3197180:3197360 [0] NCCL INFO Channel 02/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-170-31:3197180:3197360 [0] NCCL INFO Channel 03/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-170-31:3197187:3197356 [7] NCCL INFO Channel 06/0 : 31[7] -> 6[6] [send] via NET/Libfabric/6(30)/GDRDMA
+ip-26-0-170-31:3197180:3197360 [0] NCCL INFO Channel 04/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO Channel 09/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-170-31:3197180:3197360 [0] NCCL INFO Channel 09/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO Channel 10/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-170-31:3197180:3197360 [0] NCCL INFO Channel 10/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO Channel 11/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-170-31:3197180:3197360 [0] NCCL INFO Channel 11/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-170-31:3197180:3197360 [0] NCCL INFO Channel 12/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO Channel 12/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO Channel 06/0 : 0[0] -> 5[5] via P2P/IPC
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO Channel 14/0 : 0[0] -> 5[5] via P2P/IPC
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO Channel 07/0 : 0[0] -> 6[6] via P2P/IPC
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO Channel 15/0 : 0[0] -> 6[6] via P2P/IPC
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO Channel 01/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO Channel 02/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO Channel 03/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO Channel 04/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO Channel 09/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO Channel 10/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO Channel 11/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO Channel 12/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-163-58:934721:934885 [1] NCCL INFO Channel 01/0 : 2[2] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934722:934880 [2] NCCL INFO Channel 01/0 : 10[2] -> 17[1] [send] via NET/Libfabric/1(9)/GDRDMA
+ip-26-0-163-58:934721:934885 [1] NCCL INFO Channel 09/0 : 2[2] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934722:934880 [2] NCCL INFO Channel 09/0 : 10[2] -> 17[1] [send] via NET/Libfabric/1(9)/GDRDMA
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO Channel 01/0 : 10[2] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO Channel 09/0 : 10[2] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934725:934883 [5] NCCL INFO Channel 05/0 : 6[6] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934724:934881 [4] NCCL INFO Channel 04/0 : 5[5] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934725:934883 [5] NCCL INFO Channel 13/0 : 6[6] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934724:934881 [4] NCCL INFO Channel 12/0 : 5[5] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934725:934883 [5] NCCL INFO Channel 04/0 : 13[5] -> 20[4] [send] via NET/Libfabric/4(12)/GDRDMA
+ip-26-0-163-58:934724:934881 [4] NCCL INFO Channel 03/0 : 12[4] -> 19[3] [send] via NET/Libfabric/3(11)/GDRDMA
+ip-26-0-163-58:934725:934883 [5] NCCL INFO Channel 12/0 : 13[5] -> 20[4] [send] via NET/Libfabric/4(12)/GDRDMA
+ip-26-0-163-58:934724:934881 [4] NCCL INFO Channel 11/0 : 12[4] -> 19[3] [send] via NET/Libfabric/3(11)/GDRDMA
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO Channel 02/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-163-58:934721:934885 [1] NCCL INFO Channel 01/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO Channel 01/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-163-58:934722:934880 [2] NCCL INFO Channel 02/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO Channel 03/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-163-58:934726:934884 [6] NCCL INFO Channel 06/0 : 7[7] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934721:934885 [1] NCCL INFO Channel 02/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO Channel 04/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO Channel 02/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-163-58:934722:934880 [2] NCCL INFO Channel 03/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-163-58:934726:934884 [6] NCCL INFO Channel 14/0 : 7[7] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934721:934885 [1] NCCL INFO Channel 03/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO Channel 05/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO Channel 03/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-163-58:934726:934884 [6] NCCL INFO Channel 05/0 : 14[6] -> 21[5] [send] via NET/Libfabric/5(13)/GDRDMA
+ip-26-0-163-58:934726:934884 [6] NCCL INFO Channel 13/0 : 14[6] -> 21[5] [send] via NET/Libfabric/5(13)/GDRDMA
+ip-26-0-163-58:934722:934880 [2] NCCL INFO Channel 04/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-163-58:934721:934885 [1] NCCL INFO Channel 04/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO Channel 06/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO Channel 04/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-163-58:934722:934880 [2] NCCL INFO Channel 05/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO Channel 07/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO Channel 05/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-163-58:934721:934885 [1] NCCL INFO Channel 05/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO Channel 10/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO Channel 06/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-163-58:934722:934880 [2] NCCL INFO Channel 06/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-163-58:934721:934885 [1] NCCL INFO Channel 06/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO Channel 11/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO Channel 07/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-163-58:934722:934880 [2] NCCL INFO Channel 07/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO Channel 12/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-163-58:934721:934885 [1] NCCL INFO Channel 07/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO Channel 09/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO Channel 13/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO Channel 10/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-163-58:934722:934880 [2] NCCL INFO Channel 10/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-163-58:934721:934885 [1] NCCL INFO Channel 09/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO Channel 14/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO Channel 11/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-170-31:3197181:3197357 [1] NCCL INFO Channel 01/0 : 18[2] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934722:934880 [2] NCCL INFO Channel 11/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO Channel 04/0 : 13[5] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934721:934885 [1] NCCL INFO Channel 10/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-170-31:3197181:3197357 [1] NCCL INFO Channel 09/0 : 18[2] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO Channel 12/0 : 13[5] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO Channel 15/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO Channel 12/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-163-58:934722:934880 [2] NCCL INFO Channel 12/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO Channel 03/0 : 20[4] -> 27[3] [send] via NET/Libfabric/3(19)/GDRDMA
+ip-26-0-163-58:934721:934885 [1] NCCL INFO Channel 11/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO Channel 13/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO Channel 11/0 : 20[4] -> 27[3] [send] via NET/Libfabric/3(19)/GDRDMA
+ip-26-0-163-58:934722:934880 [2] NCCL INFO Channel 13/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-163-58:934721:934885 [1] NCCL INFO Channel 12/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO Channel 14/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-163-58:934722:934880 [2] NCCL INFO Channel 14/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-163-58:934721:934885 [1] NCCL INFO Channel 13/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO Channel 15/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-170-31:3197185:3197353 [5] NCCL INFO Channel 05/0 : 22[6] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-31:3197185:3197353 [5] NCCL INFO Channel 13/0 : 22[6] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934722:934880 [2] NCCL INFO Channel 15/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-163-58:934721:934885 [1] NCCL INFO Channel 14/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO Channel 06/0 : 15[7] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-31:3197186:3197354 [6] NCCL INFO Channel 06/0 : 23[7] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO Channel 04/0 : 21[5] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-31:3197187:3197356 [7] NCCL INFO Channel 14/0 : 31[7] -> 6[6] [send] via NET/Libfabric/6(30)/GDRDMA
+ip-26-0-163-58:934721:934885 [1] NCCL INFO Channel 15/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO Channel 14/0 : 15[7] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO Channel 12/0 : 21[5] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-31:3197186:3197354 [6] NCCL INFO Channel 14/0 : 23[7] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-31:3197185:3197353 [5] NCCL INFO Channel 04/0 : 29[5] -> 4[4] [send] via NET/Libfabric/4(28)/GDRDMA
+ip-26-0-170-31:3197185:3197353 [5] NCCL INFO Channel 12/0 : 29[5] -> 4[4] [send] via NET/Libfabric/4(28)/GDRDMA
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO Channel 03/0 : 28[4] -> 3[3] [send] via NET/Libfabric/3(27)/GDRDMA
+ip-26-0-170-31:3197186:3197354 [6] NCCL INFO Channel 05/0 : 30[6] -> 5[5] [send] via NET/Libfabric/5(29)/GDRDMA
+ip-26-0-170-31:3197186:3197354 [6] NCCL INFO Channel 13/0 : 30[6] -> 5[5] [send] via NET/Libfabric/5(29)/GDRDMA
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO Channel 11/0 : 28[4] -> 3[3] [send] via NET/Libfabric/3(27)/GDRDMA
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO Channel 07/0 : 20[4] -> 31[7] [send] via NET/Libfabric/7(23)/GDRDMA
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO Channel 05/0 : 14[6] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO Channel 05/0 : 22[6] -> 29[5] [send] via NET/Libfabric/5(21)/GDRDMA
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO Channel 13/0 : 14[6] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO Channel 13/0 : 22[6] -> 29[5] [send] via NET/Libfabric/5(21)/GDRDMA
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO Channel 04/0 : 21[5] -> 28[4] [send] via NET/Libfabric/4(20)/GDRDMA
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO Channel 12/0 : 21[5] -> 28[4] [send] via NET/Libfabric/4(20)/GDRDMA
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO Channel 06/0 : 22[6] -> 19[3] via P2P/IPC
+ip-26-0-170-31:3197181:3197357 [1] NCCL INFO Channel 01/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-163-58:934726:934884 [6] NCCL INFO Channel 06/0 : 14[6] -> 11[3] via P2P/IPC
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO Channel 05/0 : 21[5] -> 19[3] via P2P/IPC
+ip-26-0-163-58:934727:934886 [7] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-170-31:3197181:3197357 [1] NCCL INFO Channel 02/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO Channel 01/0 : 26[2] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO Channel 09/0 : 26[2] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-31:3197181:3197357 [1] NCCL INFO Channel 03/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-170-31:3197181:3197357 [1] NCCL INFO Channel 04/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-170-31:3197181:3197357 [1] NCCL INFO Channel 05/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-170-31:3197181:3197357 [1] NCCL INFO Channel 06/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO Channel 07/0 : 28[4] -> 7[7] [send] via NET/Libfabric/7(31)/GDRDMA
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO Channel 15/0 : 28[4] -> 7[7] [send] via NET/Libfabric/7(31)/GDRDMA
+ip-26-0-170-31:3197181:3197357 [1] NCCL INFO Channel 07/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO Channel 05/0 : 30[6] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO Channel 13/0 : 30[6] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-31:3197181:3197357 [1] NCCL INFO Channel 09/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO Channel 04/0 : 29[5] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO Channel 12/0 : 29[5] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO Channel 04/0 : 5[5] -> 12[4] [send] via NET/Libfabric/4(4)/GDRDMA
+ip-26-0-170-31:3197181:3197357 [1] NCCL INFO Channel 10/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO Channel 06/0 : 31[7] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO Channel 12/0 : 5[5] -> 12[4] [send] via NET/Libfabric/4(4)/GDRDMA
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO Channel 14/0 : 31[7] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO Channel 03/0 : 4[4] -> 11[3] [send] via NET/Libfabric/3(3)/GDRDMA
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO Channel 11/0 : 4[4] -> 11[3] [send] via NET/Libfabric/3(3)/GDRDMA
+ip-26-0-170-31:3197181:3197357 [1] NCCL INFO Channel 11/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO Channel 05/0 : 6[6] -> 13[5] [send] via NET/Libfabric/5(5)/GDRDMA
+ip-26-0-170-31:3197186:3197354 [6] NCCL INFO Channel 06/0 : 30[6] -> 27[3] via P2P/IPC
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO Channel 13/0 : 6[6] -> 13[5] [send] via NET/Libfabric/5(5)/GDRDMA
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO Channel 05/0 : 5[5] -> 3[3] via P2P/IPC
+ip-26-0-170-31:3197185:3197353 [5] NCCL INFO Channel 05/0 : 29[5] -> 27[3] via P2P/IPC
+ip-26-0-170-31:3197187:3197356 [7] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-170-31:3197181:3197357 [1] NCCL INFO Channel 12/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO Channel 06/0 : 6[6] -> 3[3] via P2P/IPC
+ip-26-0-163-58:934725:934883 [5] NCCL INFO Channel 05/0 : 13[5] -> 11[3] via P2P/IPC
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO Channel 13/0 : 5[5] -> 3[3] via P2P/IPC
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO Channel 14/0 : 6[6] -> 3[3] via P2P/IPC
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-163-58:934726:934884 [6] NCCL INFO Channel 14/0 : 14[6] -> 11[3] via P2P/IPC
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO Channel 13/0 : 21[5] -> 19[3] via P2P/IPC
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO Channel 14/0 : 22[6] -> 19[3] via P2P/IPC
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-170-31:3197186:3197354 [6] NCCL INFO Channel 14/0 : 30[6] -> 27[3] via P2P/IPC
+ip-26-0-170-31:3197185:3197353 [5] NCCL INFO Channel 13/0 : 29[5] -> 27[3] via P2P/IPC
+ip-26-0-170-31:3197181:3197357 [1] NCCL INFO Channel 13/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO Channel 04/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO Channel 05/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-170-31:3197181:3197357 [1] NCCL INFO Channel 14/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO Channel 06/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-170-31:3197181:3197357 [1] NCCL INFO Channel 15/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO Channel 07/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO Channel 02/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO Channel 09/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO Channel 03/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO Channel 10/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO Channel 04/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO Channel 11/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO Channel 05/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO Channel 12/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-163-58:934725:934883 [5] NCCL INFO Channel 13/0 : 13[5] -> 11[3] via P2P/IPC
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO Channel 06/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO Channel 13/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO Channel 14/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO Channel 15/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO Channel 07/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-170-31:3197182:3197359 [2] NCCL INFO Channel 02/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO Channel 10/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO Channel 11/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-170-31:3197182:3197359 [2] NCCL INFO Channel 03/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-170-31:3197182:3197359 [2] NCCL INFO Channel 04/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO Channel 12/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO Channel 13/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-170-31:3197182:3197359 [2] NCCL INFO Channel 05/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO Channel 14/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-170-31:3197182:3197359 [2] NCCL INFO Channel 06/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO Channel 15/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-170-31:3197182:3197359 [2] NCCL INFO Channel 07/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-170-31:3197182:3197359 [2] NCCL INFO Channel 10/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-170-31:3197182:3197359 [2] NCCL INFO Channel 11/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-170-31:3197182:3197359 [2] NCCL INFO Channel 12/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-170-31:3197182:3197359 [2] NCCL INFO Channel 13/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-170-31:3197182:3197359 [2] NCCL INFO Channel 14/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-170-31:3197182:3197359 [2] NCCL INFO Channel 15/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO Channel 07/0 : 28[4] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO Channel 15/0 : 28[4] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO Channel 07/0 : 4[4] -> 15[7] [send] via NET/Libfabric/7(7)/GDRDMA
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO Channel 15/0 : 4[4] -> 15[7] [send] via NET/Libfabric/7(7)/GDRDMA
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO Channel 07/0 : 12[4] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO Channel 15/0 : 20[4] -> 31[7] [send] via NET/Libfabric/7(23)/GDRDMA
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO Channel 07/0 : 7[7] -> 3[3] via P2P/IPC
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO Channel 15/0 : 12[4] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934727:934886 [7] NCCL INFO Channel 07/0 : 4[4] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934727:934886 [7] NCCL INFO Channel 15/0 : 4[4] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934724:934881 [4] NCCL INFO Channel 07/0 : 12[4] -> 23[7] [send] via NET/Libfabric/7(15)/GDRDMA
+ip-26-0-163-58:934724:934881 [4] NCCL INFO Channel 15/0 : 12[4] -> 23[7] [send] via NET/Libfabric/7(15)/GDRDMA
+ip-26-0-163-58:934727:934886 [7] NCCL INFO Channel 07/0 : 15[7] -> 11[3] via P2P/IPC
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO Channel 07/0 : 23[7] -> 19[3] via P2P/IPC
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO Channel 15/0 : 7[7] -> 3[3] via P2P/IPC
+ip-26-0-170-31:3197187:3197356 [7] NCCL INFO Channel 07/0 : 20[4] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-31:3197187:3197356 [7] NCCL INFO Channel 15/0 : 20[4] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-31:3197187:3197356 [7] NCCL INFO Channel 07/0 : 31[7] -> 27[3] via P2P/IPC
+ip-26-0-170-31:3197187:3197356 [7] NCCL INFO Channel 15/0 : 31[7] -> 27[3] via P2P/IPC
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO Channel 15/0 : 23[7] -> 19[3] via P2P/IPC
+ip-26-0-163-58:934727:934886 [7] NCCL INFO Channel 15/0 : 15[7] -> 11[3] via P2P/IPC
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO Channel 01/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-170-31:3197187:3197356 [7] NCCL INFO Channel 01/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO Channel 01/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-170-31:3197187:3197356 [7] NCCL INFO Channel 02/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO Channel 02/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO Channel 02/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-170-31:3197187:3197356 [7] NCCL INFO Channel 03/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-163-58:934727:934886 [7] NCCL INFO Channel 01/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-170-31:3197187:3197356 [7] NCCL INFO Channel 04/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO Channel 03/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO Channel 03/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-170-31:3197187:3197356 [7] NCCL INFO Channel 05/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO Channel 04/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-163-58:934727:934886 [7] NCCL INFO Channel 02/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO Channel 04/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-163-58:934727:934886 [7] NCCL INFO Channel 03/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-163-58:934727:934886 [7] NCCL INFO Channel 04/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-170-31:3197187:3197356 [7] NCCL INFO Channel 09/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-163-58:934727:934886 [7] NCCL INFO Channel 05/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO Channel 05/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-163-58:934727:934886 [7] NCCL INFO Channel 09/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-163-58:934727:934886 [7] NCCL INFO Channel 10/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO Channel 05/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-170-31:3197187:3197356 [7] NCCL INFO Channel 10/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO Channel 01/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-163-58:934727:934886 [7] NCCL INFO Channel 11/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-170-31:3197187:3197356 [7] NCCL INFO Channel 11/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO Channel 09/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO Channel 02/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-163-58:934724:934881 [4] NCCL INFO Channel 01/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-163-58:934727:934886 [7] NCCL INFO Channel 12/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-170-31:3197187:3197356 [7] NCCL INFO Channel 12/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO Channel 04/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-163-58:934724:934881 [4] NCCL INFO Channel 02/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-163-58:934727:934886 [7] NCCL INFO Channel 13/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-170-31:3197187:3197356 [7] NCCL INFO Channel 13/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO Channel 09/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-163-58:934724:934881 [4] NCCL INFO Channel 04/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO Channel 09/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-163-58:934724:934881 [4] NCCL INFO Channel 09/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO Channel 10/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-163-58:934724:934881 [4] NCCL INFO Channel 10/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO Channel 11/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-163-58:934724:934881 [4] NCCL INFO Channel 12/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO Channel 12/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO Channel 10/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-163-58:934723:934887 [3] NCCL INFO Channel 01/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO Channel 13/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-163-58:934726:934884 [6] NCCL INFO Channel 01/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-163-58:934723:934887 [3] NCCL INFO Channel 03/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO Channel 01/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-163-58:934726:934884 [6] NCCL INFO Channel 02/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO Channel 02/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-163-58:934723:934887 [3] NCCL INFO Channel 04/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-163-58:934726:934884 [6] NCCL INFO Channel 03/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO Channel 10/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO Channel 04/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-163-58:934723:934887 [3] NCCL INFO Channel 05/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-163-58:934726:934884 [6] NCCL INFO Channel 04/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-163-58:934723:934887 [3] NCCL INFO Channel 06/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO Channel 09/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-163-58:934726:934884 [6] NCCL INFO Channel 07/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO Channel 01/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-163-58:934725:934883 [5] NCCL INFO Channel 01/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-163-58:934723:934887 [3] NCCL INFO Channel 07/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-163-58:934726:934884 [6] NCCL INFO Channel 09/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO Channel 10/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-163-58:934725:934883 [5] NCCL INFO Channel 02/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO Channel 03/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-163-58:934726:934884 [6] NCCL INFO Channel 10/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-163-58:934723:934887 [3] NCCL INFO Channel 09/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-163-58:934725:934883 [5] NCCL INFO Channel 03/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO Channel 12/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO Channel 04/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-163-58:934726:934884 [6] NCCL INFO Channel 11/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO Channel 12/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-163-58:934723:934887 [3] NCCL INFO Channel 11/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-163-58:934725:934883 [5] NCCL INFO Channel 06/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO Channel 01/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO Channel 01/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-163-58:934726:934884 [6] NCCL INFO Channel 12/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-163-58:934723:934887 [3] NCCL INFO Channel 12/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO Channel 05/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-163-58:934725:934883 [5] NCCL INFO Channel 07/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO Channel 02/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO Channel 02/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-163-58:934726:934884 [6] NCCL INFO Channel 15/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO Channel 06/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-163-58:934723:934887 [3] NCCL INFO Channel 13/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-163-58:934725:934883 [5] NCCL INFO Channel 09/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO Channel 03/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO Channel 03/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO Channel 07/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-170-31:3197183:3197358 [3] NCCL INFO Channel 01/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO Channel 11/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-163-58:934723:934887 [3] NCCL INFO Channel 14/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-163-58:934725:934883 [5] NCCL INFO Channel 10/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO Channel 12/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO Channel 04/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO Channel 09/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO Channel 06/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO Channel 07/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO Channel 11/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-170-31:3197183:3197358 [3] NCCL INFO Channel 03/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO Channel 01/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO Channel 13/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO Channel 02/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO Channel 07/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-163-58:934723:934887 [3] NCCL INFO Channel 15/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-163-58:934725:934883 [5] NCCL INFO Channel 11/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO Channel 09/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO Channel 12/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-170-31:3197183:3197358 [3] NCCL INFO Channel 04/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO Channel 04/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO Channel 09/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO Channel 01/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-170-31:3197186:3197354 [6] NCCL INFO Channel 01/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO Channel 10/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO Channel 13/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO Channel 10/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO Channel 09/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO Channel 03/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO Channel 11/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO Channel 14/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO Channel 11/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO Channel 10/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO Channel 01/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO Channel 12/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO Channel 04/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO Channel 15/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO Channel 14/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-163-58:934725:934883 [5] NCCL INFO Channel 14/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO Channel 15/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO Channel 12/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO Channel 02/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO Channel 15/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO Channel 05/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO Channel 03/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO Channel 06/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO Channel 04/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO Channel 07/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO Channel 01/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO Channel 07/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO Channel 09/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO Channel 02/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO Channel 09/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO Channel 11/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO Channel 03/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-170-31:3197183:3197358 [3] NCCL INFO Channel 05/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO Channel 12/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO Channel 10/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO Channel 06/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO Channel 13/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO Channel 11/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO Channel 07/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO Channel 14/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO Channel 12/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO Channel 09/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-170-31:3197186:3197354 [6] NCCL INFO Channel 02/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO Channel 15/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO Channel 15/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-170-31:3197183:3197358 [3] NCCL INFO Channel 06/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO Channel 10/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-170-31:3197186:3197354 [6] NCCL INFO Channel 03/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO Channel 11/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-170-31:3197183:3197358 [3] NCCL INFO Channel 07/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-170-31:3197186:3197354 [6] NCCL INFO Channel 04/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-170-31:3197183:3197358 [3] NCCL INFO Channel 09/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-170-31:3197186:3197354 [6] NCCL INFO Channel 07/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-170-31:3197185:3197353 [5] NCCL INFO Channel 01/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-170-31:3197183:3197358 [3] NCCL INFO Channel 11/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-170-31:3197186:3197354 [6] NCCL INFO Channel 09/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-170-31:3197185:3197353 [5] NCCL INFO Channel 02/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-170-31:3197183:3197358 [3] NCCL INFO Channel 12/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-170-31:3197186:3197354 [6] NCCL INFO Channel 10/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-170-31:3197185:3197353 [5] NCCL INFO Channel 03/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-170-31:3197183:3197358 [3] NCCL INFO Channel 13/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-170-31:3197186:3197354 [6] NCCL INFO Channel 11/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-170-31:3197185:3197353 [5] NCCL INFO Channel 06/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-170-31:3197183:3197358 [3] NCCL INFO Channel 14/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-170-31:3197186:3197354 [6] NCCL INFO Channel 12/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-170-31:3197185:3197353 [5] NCCL INFO Channel 07/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-163-58:934725:934883 [5] NCCL INFO Channel 15/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-170-31:3197183:3197358 [3] NCCL INFO Channel 15/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-170-31:3197185:3197353 [5] NCCL INFO Channel 09/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-170-31:3197186:3197354 [6] NCCL INFO Channel 15/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-170-31:3197185:3197353 [5] NCCL INFO Channel 10/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-170-31:3197185:3197353 [5] NCCL INFO Channel 11/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO Channel 14/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-170-31:3197185:3197353 [5] NCCL INFO Channel 14/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-170-31:3197185:3197353 [5] NCCL INFO Channel 15/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO Channel 15/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO Connected all rings
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO Channel 02/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-170-31:3197181:3197357 [1] NCCL INFO Connected all rings
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO Connected all rings
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO Connected all rings
+ip-26-0-163-58:934724:934881 [4] NCCL INFO Connected all rings
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO Connected all rings
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO Channel 04/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-170-31:3197186:3197354 [6] NCCL INFO Connected all rings
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO Channel 05/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO Channel 06/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO Channel 07/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO Channel 10/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO Channel 11/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO Channel 12/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-163-58:934720:934882 [0] NCCL INFO Connected all rings
+ip-26-0-163-58:934720:934882 [0] NCCL INFO Channel 02/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-163-58:934720:934882 [0] NCCL INFO Channel 03/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO Channel 13/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-163-58:934721:934885 [1] NCCL INFO Connected all rings
+ip-26-0-163-58:934720:934882 [0] NCCL INFO Channel 04/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO Channel 14/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO Channel 15/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-163-58:934720:934882 [0] NCCL INFO Channel 05/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-163-58:934720:934882 [0] NCCL INFO Channel 06/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO Connected all rings
+ip-26-0-163-58:934720:934882 [0] NCCL INFO Channel 07/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-163-58:934720:934882 [0] NCCL INFO Channel 10/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-163-58:934720:934882 [0] NCCL INFO Channel 11/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-163-58:934720:934882 [0] NCCL INFO Channel 12/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-163-58:934720:934882 [0] NCCL INFO Channel 13/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-163-58:934722:934880 [2] NCCL INFO Connected all rings
+ip-26-0-163-58:934720:934882 [0] NCCL INFO Channel 14/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-163-58:934720:934882 [0] NCCL INFO Channel 15/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-163-58:934721:934885 [1] NCCL INFO Channel 01/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-163-58:934721:934885 [1] NCCL INFO Channel 03/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO Connected all rings
+ip-26-0-163-58:934721:934885 [1] NCCL INFO Channel 04/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-163-58:934721:934885 [1] NCCL INFO Channel 05/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-163-58:934721:934885 [1] NCCL INFO Channel 06/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-163-58:934721:934885 [1] NCCL INFO Channel 07/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-163-58:934721:934885 [1] NCCL INFO Channel 09/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-163-58:934721:934885 [1] NCCL INFO Channel 11/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO Connected all rings
+ip-26-0-163-58:934721:934885 [1] NCCL INFO Channel 12/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-163-58:934722:934880 [2] NCCL INFO Channel 01/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-170-31:3197180:3197360 [0] NCCL INFO Connected all rings
+ip-26-0-170-31:3197180:3197360 [0] NCCL INFO Channel 02/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-163-58:934721:934885 [1] NCCL INFO Channel 13/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-163-58:934722:934880 [2] NCCL INFO Channel 02/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-170-31:3197180:3197360 [0] NCCL INFO Channel 03/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-170-31:3197180:3197360 [0] NCCL INFO Channel 04/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-170-31:3197180:3197360 [0] NCCL INFO Channel 05/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-163-58:934721:934885 [1] NCCL INFO Channel 14/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-163-58:934722:934880 [2] NCCL INFO Channel 04/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-163-58:934721:934885 [1] NCCL INFO Channel 15/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-170-31:3197180:3197360 [0] NCCL INFO Channel 06/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-170-31:3197180:3197360 [0] NCCL INFO Channel 07/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO Channel 03/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-163-58:934722:934880 [2] NCCL INFO Channel 05/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-170-31:3197180:3197360 [0] NCCL INFO Channel 10/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO Channel 01/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-163-58:934722:934880 [2] NCCL INFO Channel 06/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-163-58:934720:934882 [0] NCCL INFO Channel 05/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO Channel 01/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-170-31:3197180:3197360 [0] NCCL INFO Channel 11/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO Channel 02/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-163-58:934722:934880 [2] NCCL INFO Channel 07/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO Channel 02/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO Channel 04/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-163-58:934720:934882 [0] NCCL INFO Channel 06/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-163-58:934722:934880 [2] NCCL INFO Channel 09/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO Channel 03/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-163-58:934720:934882 [0] NCCL INFO Channel 07/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-170-31:3197180:3197360 [0] NCCL INFO Channel 12/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-163-58:934722:934880 [2] NCCL INFO Channel 10/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO Channel 05/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO Channel 04/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-163-58:934720:934882 [0] NCCL INFO Channel 13/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-163-58:934722:934880 [2] NCCL INFO Channel 12/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-170-31:3197180:3197360 [0] NCCL INFO Channel 13/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-163-58:934720:934882 [0] NCCL INFO Channel 14/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-163-58:934722:934880 [2] NCCL INFO Channel 13/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-170-31:3197180:3197360 [0] NCCL INFO Channel 14/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-163-58:934720:934882 [0] NCCL INFO Channel 15/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO Channel 06/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-163-58:934722:934880 [2] NCCL INFO Channel 14/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO Channel 05/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-170-31:3197180:3197360 [0] NCCL INFO Channel 15/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO Channel 07/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO Channel 07/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO Channel 09/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO Channel 04/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO Channel 09/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO Channel 10/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO Channel 10/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-163-58:934725:934883 [5] NCCL INFO Connected all rings
+ip-26-0-163-58:934722:934880 [2] NCCL INFO Channel 15/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO Channel 12/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO Channel 05/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO Channel 11/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-163-58:934721:934885 [1] NCCL INFO Channel 09/0 : 1[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934721:934885 [1] NCCL INFO Channel 01/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934721:934885 [1] NCCL INFO Channel 09/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO Channel 13/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO Channel 06/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO Channel 12/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-170-31:3197185:3197353 [5] NCCL INFO Connected all rings
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO Channel 14/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO Channel 07/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO Channel 13/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO Channel 15/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO Channel 10/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO Connected all rings
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO Channel 15/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO Connected all rings
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO Channel 11/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO Channel 12/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-170-31:3197182:3197359 [2] NCCL INFO Connected all rings
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO Channel 13/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO Connected all rings
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO Channel 14/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO Connected all rings
+ip-26-0-170-31:3197181:3197357 [1] NCCL INFO Channel 01/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO Channel 15/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-163-58:934727:934886 [7] NCCL INFO Connected all rings
+ip-26-0-170-31:3197181:3197357 [1] NCCL INFO Channel 03/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-163-58:934724:934881 [4] NCCL INFO Channel 01/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-170-31:3197187:3197356 [7] NCCL INFO Connected all rings
+ip-26-0-170-31:3197181:3197357 [1] NCCL INFO Channel 04/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO Connected all rings
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO Connected all rings
+ip-26-0-170-31:3197181:3197357 [1] NCCL INFO Channel 05/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-163-58:934726:934884 [6] NCCL INFO Connected all rings
+ip-26-0-163-58:934724:934881 [4] NCCL INFO Channel 02/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-170-31:3197181:3197357 [1] NCCL INFO Channel 06/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-163-58:934724:934881 [4] NCCL INFO Channel 03/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO Connected all rings
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO Channel 01/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-170-31:3197181:3197357 [1] NCCL INFO Channel 07/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-163-58:934724:934881 [4] NCCL INFO Channel 04/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-170-31:3197183:3197358 [3] NCCL INFO Connected all rings
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO Channel 02/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-163-58:934724:934881 [4] NCCL INFO Channel 06/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-163-58:934723:934887 [3] NCCL INFO Connected all rings
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO Channel 01/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO Channel 03/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-170-31:3197181:3197357 [1] NCCL INFO Channel 09/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-163-58:934724:934881 [4] NCCL INFO Channel 07/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO Channel 02/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO Channel 04/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO Channel 01/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-163-58:934724:934881 [4] NCCL INFO Channel 09/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO Channel 03/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO Channel 05/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-170-31:3197181:3197357 [1] NCCL INFO Channel 11/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO Channel 02/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO Channel 06/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-163-58:934724:934881 [4] NCCL INFO Channel 10/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO Channel 05/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO Channel 03/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO Channel 09/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-170-31:3197181:3197357 [1] NCCL INFO Channel 12/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO Channel 01/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO Channel 06/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-163-58:934724:934881 [4] NCCL INFO Channel 11/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO Channel 04/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO Channel 10/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO Channel 07/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-163-58:934724:934881 [4] NCCL INFO Channel 12/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-170-31:3197181:3197357 [1] NCCL INFO Channel 13/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO Channel 03/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO Channel 06/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO Channel 11/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO Channel 09/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-170-31:3197185:3197353 [5] NCCL INFO Channel 01/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-170-31:3197186:3197354 [6] NCCL INFO Channel 01/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO Connected all rings
+ip-26-0-163-58:934724:934881 [4] NCCL INFO Channel 14/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO Channel 07/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO Channel 01/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO Channel 12/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO Channel 10/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO Channel 04/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-170-31:3197181:3197357 [1] NCCL INFO Channel 14/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-170-31:3197185:3197353 [5] NCCL INFO Channel 02/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-170-31:3197186:3197354 [6] NCCL INFO Channel 02/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-163-58:934725:934883 [5] NCCL INFO Channel 01/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO Channel 09/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-163-58:934724:934881 [4] NCCL INFO Channel 15/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO Channel 11/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO Channel 03/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO Channel 13/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO Channel 05/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO Channel 10/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-170-31:3197182:3197359 [2] NCCL INFO Channel 01/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-163-58:934725:934883 [5] NCCL INFO Channel 02/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-170-31:3197181:3197357 [1] NCCL INFO Channel 15/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO Channel 13/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO Channel 04/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO Channel 14/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-170-31:3197185:3197353 [5] NCCL INFO Channel 03/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-170-31:3197186:3197354 [6] NCCL INFO Channel 03/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO Channel 01/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO Channel 11/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-163-58:934725:934883 [5] NCCL INFO Channel 03/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-163-58:934726:934884 [6] NCCL INFO Channel 01/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO Channel 06/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO Channel 14/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO Channel 05/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-163-58:934725:934883 [5] NCCL INFO Channel 04/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-163-58:934726:934884 [6] NCCL INFO Channel 02/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-170-31:3197182:3197359 [2] NCCL INFO Channel 02/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO Channel 12/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-170-31:3197180:3197360 [0] NCCL INFO Channel 05/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO Channel 14/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO Channel 02/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-170-31:3197185:3197353 [5] NCCL INFO Channel 04/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-163-58:934726:934884 [6] NCCL INFO Channel 03/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-163-58:934725:934883 [5] NCCL INFO Channel 05/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-170-31:3197186:3197354 [6] NCCL INFO Channel 04/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO Channel 15/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO Channel 07/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO Channel 06/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-163-58:934726:934884 [6] NCCL INFO Channel 04/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-163-58:934725:934883 [5] NCCL INFO Channel 07/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-170-31:3197182:3197359 [2] NCCL INFO Channel 04/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-163-58:934723:934887 [3] NCCL INFO Channel 01/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO Channel 14/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO Channel 04/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-163-58:934725:934883 [5] NCCL INFO Channel 09/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-163-58:934726:934884 [6] NCCL INFO Channel 05/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO Channel 07/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-170-31:3197180:3197360 [0] NCCL INFO Channel 06/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-163-58:934723:934887 [3] NCCL INFO Channel 02/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO Channel 15/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-170-31:3197185:3197353 [5] NCCL INFO Channel 05/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-170-31:3197186:3197354 [6] NCCL INFO Channel 05/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO Channel 09/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-163-58:934725:934883 [5] NCCL INFO Channel 10/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-163-58:934726:934884 [6] NCCL INFO Channel 06/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO Channel 09/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-163-58:934723:934887 [3] NCCL INFO Channel 03/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-170-31:3197182:3197359 [2] NCCL INFO Channel 05/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO Connected all rings
+ip-26-0-163-58:934726:934884 [6] NCCL INFO Channel 09/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-163-58:934725:934883 [5] NCCL INFO Channel 11/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO Channel 05/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO Channel 11/0 : 3[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO Channel 11/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO Channel 13/0 : 5[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO Channel 12/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934723:934887 [3] NCCL INFO Channel 05/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-170-31:3197180:3197360 [0] NCCL INFO Channel 07/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-170-31:3197185:3197353 [5] NCCL INFO Channel 07/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-170-31:3197186:3197354 [6] NCCL INFO Channel 06/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO Channel 11/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO Channel 12/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-163-58:934726:934884 [6] NCCL INFO Channel 10/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-163-58:934725:934883 [5] NCCL INFO Channel 12/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-163-58:934723:934887 [3] NCCL INFO Channel 06/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO Channel 01/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO Channel 13/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-170-31:3197182:3197359 [2] NCCL INFO Channel 06/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-163-58:934726:934884 [6] NCCL INFO Channel 11/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-163-58:934725:934883 [5] NCCL INFO Channel 13/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO Channel 06/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO Channel 14/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-163-58:934723:934887 [3] NCCL INFO Channel 07/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-170-31:3197185:3197353 [5] NCCL INFO Channel 09/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO Channel 15/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-170-31:3197186:3197354 [6] NCCL INFO Channel 09/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-163-58:934725:934883 [5] NCCL INFO Channel 15/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-163-58:934726:934884 [6] NCCL INFO Channel 12/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-170-31:3197180:3197360 [0] NCCL INFO Channel 13/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-163-58:934723:934887 [3] NCCL INFO Channel 09/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO Channel 12/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO Channel 02/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-170-31:3197182:3197359 [2] NCCL INFO Channel 07/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO Channel 05/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-163-58:934726:934884 [6] NCCL INFO Channel 13/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO Channel 10/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934723:934887 [3] NCCL INFO Channel 10/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO Channel 09/0 : 1[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO Connected all rings
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO Channel 07/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO Channel 06/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-170-31:3197185:3197353 [5] NCCL INFO Channel 10/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-170-31:3197186:3197354 [6] NCCL INFO Channel 10/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-163-58:934726:934884 [6] NCCL INFO Channel 14/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-163-58:934723:934887 [3] NCCL INFO Channel 11/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-170-31:3197180:3197360 [0] NCCL INFO Channel 14/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO Channel 01/0 : 17[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO Channel 01/0 : 1[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO Channel 07/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO Channel 03/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-170-31:3197182:3197359 [2] NCCL INFO Channel 09/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO Channel 13/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-163-58:934723:934887 [3] NCCL INFO Channel 13/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO Channel 13/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-163-58:934725:934883 [5] NCCL INFO Channel 13/0 : 5[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934725:934883 [5] NCCL INFO Channel 05/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934723:934887 [3] NCCL INFO Channel 14/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO Channel 14/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-170-31:3197185:3197353 [5] NCCL INFO Channel 11/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-163-58:934726:934884 [6] NCCL INFO Channel 14/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934725:934883 [5] NCCL INFO Channel 13/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934726:934884 [6] NCCL INFO Channel 06/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-31:3197186:3197354 [6] NCCL INFO Channel 11/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO Channel 09/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO Channel 01/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-163-58:934726:934884 [6] NCCL INFO Channel 14/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO Channel 15/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-163-58:934723:934887 [3] NCCL INFO Channel 15/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-170-31:3197180:3197360 [0] NCCL INFO Channel 15/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-170-31:3197183:3197358 [3] NCCL INFO Channel 01/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO Channel 05/0 : 21[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO Channel 04/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO Channel 06/0 : 22[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO Channel 05/0 : 5[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-31:3197182:3197359 [2] NCCL INFO Channel 10/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO Channel 06/0 : 6[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO Channel 15/0 : 7[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO Channel 08/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO Channel 14/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-163-58:934722:934880 [2] NCCL INFO Channel 10/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-31:3197185:3197353 [5] NCCL INFO Channel 12/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-163-58:934724:934881 [4] NCCL INFO Channel 12/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-31:3197186:3197354 [6] NCCL INFO Channel 12/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-163-58:934722:934880 [2] NCCL INFO Channel 02/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934723:934887 [3] NCCL INFO Channel 11/0 : 3[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO Channel 10/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-163-58:934724:934881 [4] NCCL INFO Channel 04/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934722:934880 [2] NCCL INFO Channel 10/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO Channel 02/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-163-58:934723:934887 [3] NCCL INFO Channel 03/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934724:934881 [4] NCCL INFO Channel 12/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934723:934887 [3] NCCL INFO Channel 11/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-31:3197183:3197358 [3] NCCL INFO Channel 02/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO Channel 06/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO Channel 02/0 : 18[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-31:3197182:3197359 [2] NCCL INFO Channel 12/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO Channel 02/0 : 2[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO Channel 04/0 : 20[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO Channel 03/0 : 19[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO Channel 04/0 : 4[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO Channel 03/0 : 3[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO Channel 15/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-170-31:3197185:3197353 [5] NCCL INFO Channel 13/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-163-58:934727:934886 [7] NCCL INFO Channel 15/0 : 7[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-31:3197186:3197354 [6] NCCL INFO Channel 13/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-163-58:934720:934882 [0] NCCL INFO Channel 08/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934727:934886 [7] NCCL INFO Channel 07/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934720:934882 [0] NCCL INFO Channel 00/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934727:934886 [7] NCCL INFO Channel 15/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO Channel 12/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-163-58:934720:934882 [0] NCCL INFO Channel 08/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO Channel 03/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-170-31:3197183:3197358 [3] NCCL INFO Channel 03/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO Channel 07/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO Channel 07/0 : 23[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-31:3197182:3197359 [2] NCCL INFO Channel 13/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO Channel 07/0 : 7[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO Channel 00/0 : 16[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-31:3197185:3197353 [5] NCCL INFO Channel 15/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO Channel 00/0 : 0[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-31:3197186:3197354 [6] NCCL INFO Channel 14/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-170-31:3197183:3197358 [3] NCCL INFO Channel 05/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO Channel 09/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO Channel 05/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO Channel 13/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO Channel 05/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-170-31:3197182:3197359 [2] NCCL INFO Channel 14/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-170-31:3197183:3197358 [3] NCCL INFO Channel 06/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO Channel 10/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-170-31:3197182:3197359 [2] NCCL INFO Channel 15/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-170-31:3197186:3197354 [6] NCCL INFO Channel 06/0 : 22[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO Channel 14/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO Channel 06/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO Channel 06/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO Channel 11/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-170-31:3197183:3197358 [3] NCCL INFO Channel 07/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO Channel 01/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-170-31:3197181:3197357 [1] NCCL INFO Channel 01/0 : 17[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO Channel 12/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-170-31:3197183:3197358 [3] NCCL INFO Channel 09/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO Channel 15/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO Channel 07/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO Channel 07/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO Channel 14/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-170-31:3197183:3197358 [3] NCCL INFO Channel 10/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO Channel 02/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO Channel 15/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-170-31:3197183:3197358 [3] NCCL INFO Channel 11/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO Channel 09/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO Channel 01/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO Channel 13/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO Channel 09/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO Channel 01/0 : 17[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO Channel 03/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-170-31:3197183:3197358 [3] NCCL INFO Channel 13/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-170-31:3197185:3197353 [5] NCCL INFO Channel 05/0 : 21[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO Channel 10/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-170-31:3197183:3197358 [3] NCCL INFO Channel 14/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO Channel 01/0 : 1[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-31:3197181:3197357 [1] NCCL INFO Channel 09/0 : 9[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO Channel 01/0 : 17[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-31:3197187:3197356 [7] NCCL INFO Channel 07/0 : 23[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-31:3197180:3197360 [0] NCCL INFO Channel 00/0 : 16[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO Channel 14/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-170-31:3197181:3197357 [1] NCCL INFO Channel 09/0 : 25[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-31:3197183:3197358 [3] NCCL INFO Channel 15/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO Channel 04/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO Channel 01/0 : 25[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO Channel 09/0 : 9[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO Channel 01/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO Channel 09/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-31:3197182:3197359 [2] NCCL INFO Channel 02/0 : 18[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO Channel 04/0 : 20[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-31:3197183:3197358 [3] NCCL INFO Channel 03/0 : 19[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO Channel 11/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO Channel 15/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO Channel 13/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO Channel 14/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO Channel 01/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO Channel 06/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO Channel 15/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO Channel 02/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO Channel 01/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO Channel 07/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO Channel 03/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO Channel 02/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO Channel 02/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO Channel 10/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO Channel 09/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO Channel 02/0 : 18[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO Channel 04/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO Channel 03/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO Channel 10/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-163-58:934722:934880 [2] NCCL INFO Channel 10/0 : 26[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO Channel 05/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO Channel 02/0 : 2[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934722:934880 [2] NCCL INFO Channel 10/0 : 10[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-31:3197182:3197359 [2] NCCL INFO Channel 10/0 : 10[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO Channel 02/0 : 18[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO Channel 04/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-170-31:3197182:3197359 [2] NCCL INFO Channel 10/0 : 26[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934721:934885 [1] NCCL INFO Channel 09/0 : 25[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO Channel 11/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO Channel 02/0 : 26[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-31:3197182:3197359 [2] NCCL INFO Channel 02/0 : 26[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO Channel 10/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934721:934885 [1] NCCL INFO Channel 09/0 : 9[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO Channel 06/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO Channel 02/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO Channel 10/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934722:934880 [2] NCCL INFO Channel 02/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO Channel 05/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-163-58:934722:934880 [2] NCCL INFO Channel 10/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934721:934885 [1] NCCL INFO Channel 01/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO Channel 12/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-170-31:3197181:3197357 [1] NCCL INFO Channel 01/0 : 25[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934722:934880 [2] NCCL INFO Channel 10/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO Channel 09/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-163-58:934721:934885 [1] NCCL INFO Channel 09/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934721:934885 [1] NCCL INFO Channel 09/0 : 9[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO Channel 07/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO Channel 14/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO Channel 10/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO Channel 09/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO Channel 15/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO Channel 11/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-170-31:3197182:3197359 [2] NCCL INFO Channel 00/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-163-58:934722:934880 [2] NCCL INFO Channel 00/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-163-58:934722:934880 [2] NCCL INFO Channel 01/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO Channel 00/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-163-58:934722:934880 [2] NCCL INFO Channel 08/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-163-58:934722:934880 [2] NCCL INFO Channel 09/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-163-58:934721:934885 [1] NCCL INFO Channel 00/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-163-58:934721:934885 [1] NCCL INFO Channel 08/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-170-31:3197182:3197359 [2] NCCL INFO Channel 01/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-170-31:3197182:3197359 [2] NCCL INFO Channel 08/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-170-31:3197181:3197357 [1] NCCL INFO Channel 00/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO Channel 01/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO Channel 08/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO Channel 08/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO Channel 09/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-170-31:3197182:3197359 [2] NCCL INFO Channel 09/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-170-31:3197181:3197357 [1] NCCL INFO Channel 08/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO Channel 10/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO Channel 12/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO Channel 03/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO Channel 11/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO Channel 03/0 : 19[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO Channel 11/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO Channel 13/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-163-58:934723:934887 [3] NCCL INFO Channel 11/0 : 27[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO Channel 00/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-163-58:934723:934887 [3] NCCL INFO Channel 11/0 : 11[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO Channel 03/0 : 3[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-31:3197183:3197358 [3] NCCL INFO Channel 11/0 : 11[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO Channel 03/0 : 19[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-31:3197183:3197358 [3] NCCL INFO Channel 11/0 : 27[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO Channel 14/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO Channel 11/0 : 11[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO Channel 03/0 : 27[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO Channel 12/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-163-58:934723:934887 [3] NCCL INFO Channel 03/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO Channel 03/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO Channel 00/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-170-31:3197183:3197358 [3] NCCL INFO Channel 03/0 : 27[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO Channel 08/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO Channel 11/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934723:934887 [3] NCCL INFO Channel 11/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934723:934887 [3] NCCL INFO Channel 11/0 : 11[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO Channel 13/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO Channel 01/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO Channel 15/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO Channel 08/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO Channel 09/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO Channel 04/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO Channel 06/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO Channel 12/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO Channel 14/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO Channel 05/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO Channel 04/0 : 20[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO Channel 06/0 : 22[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO Channel 13/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO Channel 05/0 : 21[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934724:934881 [4] NCCL INFO Channel 12/0 : 28[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934726:934884 [6] NCCL INFO Channel 14/0 : 30[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934724:934881 [4] NCCL INFO Channel 12/0 : 12[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO Channel 06/0 : 6[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO Channel 04/0 : 4[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO Channel 12/0 : 12[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934726:934884 [6] NCCL INFO Channel 14/0 : 14[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934725:934883 [5] NCCL INFO Channel 13/0 : 29[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO Channel 06/0 : 22[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO Channel 04/0 : 20[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-31:3197186:3197354 [6] NCCL INFO Channel 14/0 : 14[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934725:934883 [5] NCCL INFO Channel 13/0 : 13[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO Channel 12/0 : 28[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-31:3197186:3197354 [6] NCCL INFO Channel 14/0 : 30[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO Channel 05/0 : 5[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-31:3197185:3197353 [5] NCCL INFO Channel 13/0 : 13[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO Channel 05/0 : 21[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-31:3197185:3197353 [5] NCCL INFO Channel 13/0 : 29[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO Channel 04/0 : 28[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO Channel 12/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO Channel 14/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO Channel 06/0 : 30[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO Channel 04/0 : 28[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO Channel 04/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO Channel 00/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO Channel 06/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO Channel 05/0 : 29[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO Channel 14/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO Channel 12/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO Channel 13/0 : 13[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO Channel 05/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO Channel 02/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO Channel 13/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO Channel 07/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO Channel 15/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO Channel 08/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-163-58:934723:934887 [3] NCCL INFO Channel 00/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO Channel 07/0 : 23[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO Channel 10/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-163-58:934723:934887 [3] NCCL INFO Channel 02/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-163-58:934725:934883 [5] NCCL INFO Channel 05/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-31:3197185:3197353 [5] NCCL INFO Channel 05/0 : 29[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934726:934884 [6] NCCL INFO Channel 06/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-31:3197186:3197354 [6] NCCL INFO Channel 06/0 : 30[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO Channel 07/0 : 7[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934725:934883 [5] NCCL INFO Channel 13/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934723:934887 [3] NCCL INFO Channel 08/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-163-58:934726:934884 [6] NCCL INFO Channel 14/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-31:3197187:3197356 [7] NCCL INFO Channel 15/0 : 15[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934725:934883 [5] NCCL INFO Channel 13/0 : 13[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO Channel 07/0 : 23[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934726:934884 [6] NCCL INFO Channel 14/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-31:3197187:3197356 [7] NCCL INFO Channel 15/0 : 31[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934723:934887 [3] NCCL INFO Channel 10/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO Channel 07/0 : 31[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO Channel 07/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO Channel 15/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO Channel 15/0 : 15[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO Channel 00/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO Channel 02/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-163-58:934727:934886 [7] NCCL INFO Channel 15/0 : 31[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO Channel 00/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-163-58:934727:934886 [7] NCCL INFO Channel 15/0 : 15[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO Channel 08/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-163-58:934727:934886 [7] NCCL INFO Channel 07/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO Channel 04/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-170-31:3197187:3197356 [7] NCCL INFO Channel 07/0 : 31[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934724:934881 [4] NCCL INFO Channel 04/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934727:934886 [7] NCCL INFO Channel 15/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934724:934881 [4] NCCL INFO Channel 12/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO Channel 10/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-170-31:3197187:3197356 [7] NCCL INFO Channel 01/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-163-58:934727:934886 [7] NCCL INFO Channel 15/0 : 15[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934724:934881 [4] NCCL INFO Channel 12/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO Channel 08/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO Channel 01/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-163-58:934727:934886 [7] NCCL INFO Channel 01/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO Channel 01/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO Channel 12/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-170-31:3197183:3197358 [3] NCCL INFO Channel 00/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO Channel 00/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-170-31:3197185:3197353 [5] NCCL INFO Channel 00/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-170-31:3197187:3197356 [7] NCCL INFO Channel 02/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-170-31:3197186:3197354 [6] NCCL INFO Channel 00/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-170-31:3197183:3197358 [3] NCCL INFO Channel 02/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO Channel 03/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-170-31:3197185:3197353 [5] NCCL INFO Channel 04/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-170-31:3197187:3197356 [7] NCCL INFO Channel 03/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-170-31:3197186:3197354 [6] NCCL INFO Channel 05/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO Channel 00/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO Channel 08/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO Channel 00/0 : 16[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-31:3197183:3197358 [3] NCCL INFO Channel 08/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO Channel 05/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-170-31:3197185:3197353 [5] NCCL INFO Channel 08/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-163-58:934720:934882 [0] NCCL INFO Channel 08/0 : 24[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-31:3197187:3197356 [7] NCCL INFO Channel 04/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO Channel 00/0 : 0[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934720:934882 [0] NCCL INFO Channel 08/0 : 8[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-31:3197186:3197354 [6] NCCL INFO Channel 08/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO Channel 00/0 : 16[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-31:3197180:3197360 [0] NCCL INFO Channel 08/0 : 8[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-31:3197180:3197360 [0] NCCL INFO Channel 08/0 : 24[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO Channel 00/0 : 24[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO Channel 08/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO Channel 00/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-31:3197183:3197358 [3] NCCL INFO Channel 10/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO Channel 08/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934727:934886 [7] NCCL INFO Channel 02/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-170-31:3197186:3197354 [6] NCCL INFO Channel 13/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO Channel 02/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-170-31:3197180:3197360 [0] NCCL INFO Channel 00/0 : 24[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934720:934882 [0] NCCL INFO Channel 00/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934720:934882 [0] NCCL INFO Channel 08/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934720:934882 [0] NCCL INFO Channel 08/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-31:3197185:3197353 [5] NCCL INFO Channel 12/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO Channel 06/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO Channel 03/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-170-31:3197187:3197356 [7] NCCL INFO Channel 05/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO Channel 04/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-163-58:934727:934886 [7] NCCL INFO Channel 03/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO Channel 07/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-163-58:934727:934886 [7] NCCL INFO Channel 04/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-163-58:934726:934884 [6] NCCL INFO Channel 00/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-163-58:934725:934883 [5] NCCL INFO Channel 00/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-163-58:934727:934886 [7] NCCL INFO Channel 05/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-163-58:934725:934883 [5] NCCL INFO Channel 04/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-163-58:934726:934884 [6] NCCL INFO Channel 05/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-163-58:934724:934881 [4] NCCL INFO Channel 00/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-163-58:934727:934886 [7] NCCL INFO Channel 06/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-163-58:934726:934884 [6] NCCL INFO Channel 08/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-163-58:934725:934883 [5] NCCL INFO Channel 08/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-163-58:934724:934881 [4] NCCL INFO Channel 03/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-163-58:934727:934886 [7] NCCL INFO Channel 07/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-163-58:934725:934883 [5] NCCL INFO Channel 12/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-163-58:934724:934881 [4] NCCL INFO Channel 05/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-170-31:3197187:3197356 [7] NCCL INFO Channel 06/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-163-58:934727:934886 [7] NCCL INFO Channel 09/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-163-58:934726:934884 [6] NCCL INFO Channel 13/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO Channel 02/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO Channel 05/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-163-58:934724:934881 [4] NCCL INFO Channel 06/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-170-31:3197187:3197356 [7] NCCL INFO Channel 07/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-163-58:934727:934886 [7] NCCL INFO Channel 10/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO Channel 08/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO Channel 03/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO Channel 00/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO Channel 00/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO Channel 06/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO Channel 04/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO Channel 11/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-163-58:934724:934881 [4] NCCL INFO Channel 07/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-163-58:934727:934886 [7] NCCL INFO Channel 11/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO Channel 00/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO Channel 00/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO Channel 13/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO Channel 03/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-170-31:3197187:3197356 [7] NCCL INFO Channel 09/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-163-58:934724:934881 [4] NCCL INFO Channel 08/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO Channel 05/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO Channel 05/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-163-58:934727:934886 [7] NCCL INFO Channel 12/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO Channel 07/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO Channel 05/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO Channel 08/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO Channel 05/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO Channel 04/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO Channel 09/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO Channel 13/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO Channel 06/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO Channel 14/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-170-31:3197187:3197356 [7] NCCL INFO Channel 10/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO Channel 10/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO Channel 07/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO Channel 00/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-163-58:934724:934881 [4] NCCL INFO Channel 11/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO Channel 11/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-163-58:934727:934886 [7] NCCL INFO Channel 13/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO Channel 08/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO Channel 06/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-163-58:934724:934881 [4] NCCL INFO Channel 13/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO Channel 08/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO Channel 15/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-170-31:3197187:3197356 [7] NCCL INFO Channel 11/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO Channel 12/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-163-58:934727:934886 [7] NCCL INFO Channel 14/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO Channel 03/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO Channel 11/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO Channel 07/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-163-58:934724:934881 [4] NCCL INFO Channel 14/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO Channel 13/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-170-31:3197187:3197356 [7] NCCL INFO Channel 12/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-163-58:934727:934886 [7] NCCL INFO Channel 15/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO Channel 13/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO Channel 05/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO Channel 08/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO Channel 09/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-163-58:934724:934881 [4] NCCL INFO Channel 15/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-170-31:3197187:3197356 [7] NCCL INFO Channel 13/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO Channel 14/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO Channel 13/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO Channel 06/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO Channel 12/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO Channel 10/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-170-31:3197187:3197356 [7] NCCL INFO Channel 14/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO Channel 15/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO Channel 14/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO Channel 07/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO Channel 11/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO Channel 08/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO Channel 15/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO Channel 12/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO Channel 11/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-170-31:3197187:3197356 [7] NCCL INFO Channel 15/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO Channel 13/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO Channel 13/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO Channel 14/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO Channel 14/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO Channel 15/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO Channel 00/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO Channel 15/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO Channel 06/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO Channel 08/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-170-31:3197187:3197356 [7] NCCL INFO Channel 00/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-170-31:3197187:3197356 [7] NCCL INFO Channel 06/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO Channel 14/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO Channel 00/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-170-31:3197187:3197356 [7] NCCL INFO Channel 08/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-170-31:3197187:3197356 [7] NCCL INFO Channel 14/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO Channel 06/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO Channel 08/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-163-58:934727:934886 [7] NCCL INFO Channel 00/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-163-58:934727:934886 [7] NCCL INFO Channel 06/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-163-58:934727:934886 [7] NCCL INFO Channel 08/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO Channel 14/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-163-58:934727:934886 [7] NCCL INFO Channel 14/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO Connected all trees
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO NVLS comm 0x9f7c5a0 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-170-31:3197182:3197359 [2] NCCL INFO Connected all trees
+ip-26-0-170-31:3197182:3197359 [2] NCCL INFO NVLS comm 0x86c2450 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-170-31:3197187:3197356 [7] NCCL INFO Connected all trees
+ip-26-0-170-31:3197187:3197356 [7] NCCL INFO NVLS comm 0xa426700 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-170-31:3197180:3197360 [0] NCCL INFO Connected all trees
+ip-26-0-170-31:3197180:3197360 [0] NCCL INFO NVLS comm 0x9f60150 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO Connected all trees
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO NVLS comm 0xa132980 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-170-31:3197181:3197357 [1] NCCL INFO Connected all trees
+ip-26-0-170-31:3197181:3197357 [1] NCCL INFO NVLS comm 0x998b390 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO Connected all trees
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO NVLS comm 0x9a0b900 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-170-31:3197185:3197353 [5] NCCL INFO Connected all trees
+ip-26-0-170-31:3197185:3197353 [5] NCCL INFO NVLS comm 0x8610a20 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO Connected all trees
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO Connected all trees
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO NVLS comm 0x9c385f0 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO NVLS comm 0x8fe8d30 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-170-31:3197183:3197358 [3] NCCL INFO Connected all trees
+ip-26-0-170-31:3197183:3197358 [3] NCCL INFO NVLS comm 0x9d302b0 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO Connected all trees
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO NVLS comm 0x94864f0 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO Connected all trees
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO NVLS comm 0x9ec45d0 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO Connected all trees
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO NVLS comm 0x92b7800 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-163-58:934722:934880 [2] NCCL INFO Connected all trees
+ip-26-0-163-58:934722:934880 [2] NCCL INFO NVLS comm 0x8886e30 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO Connected all trees
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO NVLS comm 0x962f910 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-170-31:3197186:3197354 [6] NCCL INFO Connected all trees
+ip-26-0-170-31:3197186:3197354 [6] NCCL INFO NVLS comm 0x9df89f0 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-163-58:934720:934882 [0] NCCL INFO Connected all trees
+ip-26-0-163-58:934727:934886 [7] NCCL INFO Connected all trees
+ip-26-0-163-58:934727:934886 [7] NCCL INFO NVLS comm 0x97de120 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-163-58:934720:934882 [0] NCCL INFO NVLS comm 0x95c46a0 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-163-58:934721:934885 [1] NCCL INFO Connected all trees
+ip-26-0-163-58:934721:934885 [1] NCCL INFO NVLS comm 0xa3091e0 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO Connected all trees
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO NVLS comm 0x985f5f0 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO Connected all trees
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO NVLS comm 0xa2b01f0 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-163-58:934725:934883 [5] NCCL INFO Connected all trees
+ip-26-0-163-58:934726:934884 [6] NCCL INFO Connected all trees
+ip-26-0-163-58:934725:934883 [5] NCCL INFO NVLS comm 0x886a710 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-163-58:934726:934884 [6] NCCL INFO NVLS comm 0x9bcf110 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO Connected all trees
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO NVLS comm 0x8cd8e90 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO Connected all trees
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO Connected all trees
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO NVLS comm 0x97fbc80 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO NVLS comm 0x856fc20 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO Connected all trees
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO NVLS comm 0x9296ef0 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-163-58:934723:934887 [3] NCCL INFO Connected all trees
+ip-26-0-163-58:934723:934887 [3] NCCL INFO NVLS comm 0x98861d0 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-163-58:934724:934881 [4] NCCL INFO Connected all trees
+ip-26-0-163-58:934724:934881 [4] NCCL INFO NVLS comm 0x967ca40 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO Connected all trees
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO NVLS comm 0xa2e0240 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO Connected all trees
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO NVLS comm 0x93cf5d0 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-170-31:3197185:3197353 [5] NCCL INFO Channel 00/0 : 21[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-31:3197185:3197353 [5] NCCL INFO Channel 02/0 : 21[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-31:3197181:3197357 [1] NCCL INFO Channel 00/0 : 17[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-31:3197185:3197353 [5] NCCL INFO Channel 04/0 : 21[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-31:3197181:3197357 [1] NCCL INFO Channel 02/0 : 17[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-31:3197185:3197353 [5] NCCL INFO Channel 06/0 : 21[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-31:3197181:3197357 [1] NCCL INFO Channel 04/0 : 17[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-31:3197185:3197353 [5] NCCL INFO Channel 08/0 : 21[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-31:3197181:3197357 [1] NCCL INFO Channel 06/0 : 17[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-31:3197185:3197353 [5] NCCL INFO Channel 10/0 : 21[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-31:3197185:3197353 [5] NCCL INFO Channel 12/0 : 21[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-31:3197181:3197357 [1] NCCL INFO Channel 08/0 : 17[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-31:3197185:3197353 [5] NCCL INFO Channel 14/0 : 21[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-31:3197181:3197357 [1] NCCL INFO Channel 10/0 : 17[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-31:3197181:3197357 [1] NCCL INFO Channel 12/0 : 17[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-31:3197181:3197357 [1] NCCL INFO Channel 14/0 : 17[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-31:3197187:3197356 [7] NCCL INFO Channel 00/0 : 23[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-31:3197187:3197356 [7] NCCL INFO Channel 02/0 : 23[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO Channel 01/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO Channel 00/0 : 20[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO Channel 03/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-31:3197187:3197356 [7] NCCL INFO Channel 04/0 : 23[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO Channel 05/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO Channel 02/0 : 20[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO Channel 01/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-31:3197187:3197356 [7] NCCL INFO Channel 06/0 : 23[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO Channel 06/0 : 20[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-31:3197187:3197356 [7] NCCL INFO Channel 08/0 : 23[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO Channel 07/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-31:3197180:3197360 [0] NCCL INFO Channel 02/0 : 16[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO Channel 01/0 : 5[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO Channel 08/0 : 20[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO Channel 01/0 : 1[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO Channel 03/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO Channel 01/0 : 7[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO Channel 01/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO Channel 09/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO Channel 01/0 : 3[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-31:3197183:3197358 [3] NCCL INFO Channel 00/0 : 19[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO Channel 03/0 : 5[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-31:3197187:3197356 [7] NCCL INFO Channel 10/0 : 23[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO Channel 05/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-31:3197186:3197354 [6] NCCL INFO Channel 00/0 : 22[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO Channel 03/0 : 1[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-31:3197180:3197360 [0] NCCL INFO Channel 04/0 : 16[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO Channel 03/0 : 7[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO Channel 10/0 : 20[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO Channel 03/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-31:3197182:3197359 [2] NCCL INFO Channel 00/0 : 18[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-31:3197187:3197356 [7] NCCL INFO Channel 12/0 : 23[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-31:3197183:3197358 [3] NCCL INFO Channel 02/0 : 19[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO Channel 01/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO Channel 11/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO Channel 03/0 : 3[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO Channel 05/0 : 5[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-31:3197180:3197360 [0] NCCL INFO Channel 06/0 : 16[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-31:3197186:3197354 [6] NCCL INFO Channel 02/0 : 22[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO Channel 05/0 : 1[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO Channel 12/0 : 20[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO Channel 07/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-31:3197187:3197356 [7] NCCL INFO Channel 14/0 : 23[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-31:3197182:3197359 [2] NCCL INFO Channel 04/0 : 18[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-31:3197183:3197358 [3] NCCL INFO Channel 04/0 : 19[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO Channel 05/0 : 7[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-31:3197180:3197360 [0] NCCL INFO Channel 08/0 : 16[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO Channel 13/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO Channel 05/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO Channel 03/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO Channel 05/0 : 3[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO Channel 14/0 : 20[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-31:3197186:3197354 [6] NCCL INFO Channel 04/0 : 22[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO Channel 07/0 : 5[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-31:3197183:3197358 [3] NCCL INFO Channel 06/0 : 19[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO Channel 07/0 : 1[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO Channel 09/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-31:3197180:3197360 [0] NCCL INFO Channel 10/0 : 16[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO Channel 07/0 : 7[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-31:3197182:3197359 [2] NCCL INFO Channel 06/0 : 18[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO Channel 15/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-31:3197186:3197354 [6] NCCL INFO Channel 08/0 : 22[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO Channel 07/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO Channel 05/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO Channel 07/0 : 3[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO Channel 09/0 : 5[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-31:3197183:3197358 [3] NCCL INFO Channel 08/0 : 19[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-31:3197180:3197360 [0] NCCL INFO Channel 12/0 : 16[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO Channel 11/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-31:3197186:3197354 [6] NCCL INFO Channel 10/0 : 22[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO Channel 11/0 : 1[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-31:3197182:3197359 [2] NCCL INFO Channel 08/0 : 18[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO Channel 09/0 : 7[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-31:3197180:3197360 [0] NCCL INFO Channel 14/0 : 16[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-31:3197183:3197358 [3] NCCL INFO Channel 10/0 : 19[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO Channel 09/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-31:3197186:3197354 [6] NCCL INFO Channel 12/0 : 22[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO Channel 07/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO Channel 09/0 : 3[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO Channel 11/0 : 5[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-31:3197182:3197359 [2] NCCL INFO Channel 10/0 : 18[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO Channel 13/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-31:3197183:3197358 [3] NCCL INFO Channel 12/0 : 19[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-31:3197186:3197354 [6] NCCL INFO Channel 14/0 : 22[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-31:3197182:3197359 [2] NCCL INFO Channel 12/0 : 18[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO Channel 11/0 : 7[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO Channel 13/0 : 1[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO Channel 11/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO Channel 09/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-31:3197183:3197358 [3] NCCL INFO Channel 14/0 : 19[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO Channel 15/0 : 5[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-31:3197182:3197359 [2] NCCL INFO Channel 14/0 : 18[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO Channel 13/0 : 3[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO Channel 15/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO Channel 13/0 : 7[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO Channel 13/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO Channel 15/0 : 1[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO Channel 11/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO Channel 15/0 : 3[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO Channel 15/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO Channel 13/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO Channel 15/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934724:934881 [4] NCCL INFO Channel 01/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934720:934882 [0] NCCL INFO Channel 01/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934724:934881 [4] NCCL INFO Channel 03/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934720:934882 [0] NCCL INFO Channel 03/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934724:934881 [4] NCCL INFO Channel 05/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934720:934882 [0] NCCL INFO Channel 05/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934724:934881 [4] NCCL INFO Channel 07/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934720:934882 [0] NCCL INFO Channel 07/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934724:934881 [4] NCCL INFO Channel 09/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934720:934882 [0] NCCL INFO Channel 09/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934724:934881 [4] NCCL INFO Channel 11/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934720:934882 [0] NCCL INFO Channel 11/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934724:934881 [4] NCCL INFO Channel 13/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934720:934882 [0] NCCL INFO Channel 13/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934724:934881 [4] NCCL INFO Channel 15/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934720:934882 [0] NCCL INFO Channel 15/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934724:934881 [4] NCCL INFO Channel 00/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934720:934882 [0] NCCL INFO Channel 01/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934724:934881 [4] NCCL INFO Channel 01/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934720:934882 [0] NCCL INFO Channel 02/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934724:934881 [4] NCCL INFO Channel 02/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934720:934882 [0] NCCL INFO Channel 03/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934720:934882 [0] NCCL INFO Channel 04/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934724:934881 [4] NCCL INFO Channel 03/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934720:934882 [0] NCCL INFO Channel 05/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934724:934881 [4] NCCL INFO Channel 05/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934720:934882 [0] NCCL INFO Channel 06/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934724:934881 [4] NCCL INFO Channel 06/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934720:934882 [0] NCCL INFO Channel 07/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934724:934881 [4] NCCL INFO Channel 07/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934720:934882 [0] NCCL INFO Channel 09/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934724:934881 [4] NCCL INFO Channel 08/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934720:934882 [0] NCCL INFO Channel 10/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934726:934884 [6] NCCL INFO Channel 01/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934724:934881 [4] NCCL INFO Channel 09/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934720:934882 [0] NCCL INFO Channel 11/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934724:934881 [4] NCCL INFO Channel 10/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934726:934884 [6] NCCL INFO Channel 03/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934720:934882 [0] NCCL INFO Channel 12/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934724:934881 [4] NCCL INFO Channel 11/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934726:934884 [6] NCCL INFO Channel 05/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934720:934882 [0] NCCL INFO Channel 13/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934724:934881 [4] NCCL INFO Channel 13/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934726:934884 [6] NCCL INFO Channel 07/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934720:934882 [0] NCCL INFO Channel 14/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934724:934881 [4] NCCL INFO Channel 14/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934720:934882 [0] NCCL INFO Channel 15/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934726:934884 [6] NCCL INFO Channel 09/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934724:934881 [4] NCCL INFO Channel 15/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934726:934884 [6] NCCL INFO Channel 11/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934726:934884 [6] NCCL INFO Channel 13/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934726:934884 [6] NCCL INFO Channel 15/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934726:934884 [6] NCCL INFO Channel 00/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO Channel 00/0 : 20[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO Channel 02/0 : 16[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934726:934884 [6] NCCL INFO Channel 01/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO Channel 02/0 : 20[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO Channel 04/0 : 16[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934726:934884 [6] NCCL INFO Channel 02/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934726:934884 [6] NCCL INFO Channel 03/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO Channel 06/0 : 20[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO Channel 06/0 : 16[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934726:934884 [6] NCCL INFO Channel 04/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO Channel 08/0 : 20[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO Channel 08/0 : 16[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934726:934884 [6] NCCL INFO Channel 05/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934726:934884 [6] NCCL INFO Channel 07/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO Channel 10/0 : 20[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO Channel 10/0 : 16[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934726:934884 [6] NCCL INFO Channel 08/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO Channel 12/0 : 16[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO Channel 12/0 : 20[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934727:934886 [7] NCCL INFO Channel 01/0 : 7[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934725:934883 [5] NCCL INFO Channel 01/0 : 5[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO Channel 14/0 : 16[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934726:934884 [6] NCCL INFO Channel 09/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO Channel 14/0 : 20[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934727:934886 [7] NCCL INFO Channel 03/0 : 7[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO Channel 02/0 : 0[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934725:934883 [5] NCCL INFO Channel 03/0 : 5[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934726:934884 [6] NCCL INFO Channel 10/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO Channel 00/0 : 4[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO Channel 04/0 : 0[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934727:934886 [7] NCCL INFO Channel 05/0 : 7[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934721:934885 [1] NCCL INFO Channel 01/0 : 1[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934726:934884 [6] NCCL INFO Channel 11/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934725:934883 [5] NCCL INFO Channel 05/0 : 5[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO Channel 02/0 : 4[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO Channel 06/0 : 0[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934727:934886 [7] NCCL INFO Channel 07/0 : 7[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO Channel 06/0 : 4[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO Channel 08/0 : 0[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934726:934884 [6] NCCL INFO Channel 12/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934721:934885 [1] NCCL INFO Channel 03/0 : 1[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934725:934883 [5] NCCL INFO Channel 07/0 : 5[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934722:934880 [2] NCCL INFO Channel 01/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934723:934887 [3] NCCL INFO Channel 01/0 : 3[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO Channel 08/0 : 4[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO Channel 10/0 : 0[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934727:934886 [7] NCCL INFO Channel 09/0 : 7[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934726:934884 [6] NCCL INFO Channel 13/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934721:934885 [1] NCCL INFO Channel 05/0 : 1[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO Channel 10/0 : 4[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934725:934883 [5] NCCL INFO Channel 09/0 : 5[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO Channel 12/0 : 0[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934722:934880 [2] NCCL INFO Channel 03/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934723:934887 [3] NCCL INFO Channel 03/0 : 3[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934726:934884 [6] NCCL INFO Channel 15/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934727:934886 [7] NCCL INFO Channel 11/0 : 7[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934721:934885 [1] NCCL INFO Channel 07/0 : 1[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO Channel 00/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO Channel 12/0 : 4[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO Channel 14/0 : 0[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934725:934883 [5] NCCL INFO Channel 11/0 : 5[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO Channel 01/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934722:934880 [2] NCCL INFO Channel 05/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934723:934887 [3] NCCL INFO Channel 05/0 : 3[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934727:934886 [7] NCCL INFO Channel 13/0 : 7[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO Channel 14/0 : 4[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934721:934885 [1] NCCL INFO Channel 11/0 : 1[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934725:934883 [5] NCCL INFO Channel 15/0 : 5[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934722:934880 [2] NCCL INFO Channel 07/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO Channel 02/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934723:934887 [3] NCCL INFO Channel 07/0 : 3[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934727:934886 [7] NCCL INFO Channel 00/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934721:934885 [1] NCCL INFO Channel 13/0 : 1[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934725:934883 [5] NCCL INFO Channel 00/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934722:934880 [2] NCCL INFO Channel 09/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO Channel 00/0 : 22[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO Channel 03/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934723:934887 [3] NCCL INFO Channel 09/0 : 3[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934727:934886 [7] NCCL INFO Channel 01/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934725:934883 [5] NCCL INFO Channel 01/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934721:934885 [1] NCCL INFO Channel 15/0 : 1[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934722:934880 [2] NCCL INFO Channel 11/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO Channel 02/0 : 22[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934723:934887 [3] NCCL INFO Channel 13/0 : 3[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO Channel 05/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934727:934886 [7] NCCL INFO Channel 02/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934725:934883 [5] NCCL INFO Channel 02/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO Channel 04/0 : 22[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934721:934885 [1] NCCL INFO Channel 00/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934722:934880 [2] NCCL INFO Channel 13/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934723:934887 [3] NCCL INFO Channel 15/0 : 3[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934727:934886 [7] NCCL INFO Channel 03/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO Channel 06/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934725:934883 [5] NCCL INFO Channel 03/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO Channel 08/0 : 22[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934721:934885 [1] NCCL INFO Channel 02/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934722:934880 [2] NCCL INFO Channel 15/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO Channel 07/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934723:934887 [3] NCCL INFO Channel 00/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO Channel 10/0 : 22[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934727:934886 [7] NCCL INFO Channel 04/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934725:934883 [5] NCCL INFO Channel 04/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO Channel 08/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934721:934885 [1] NCCL INFO Channel 03/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934722:934880 [2] NCCL INFO Channel 00/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934723:934887 [3] NCCL INFO Channel 01/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO Channel 12/0 : 22[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934727:934886 [7] NCCL INFO Channel 05/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934725:934883 [5] NCCL INFO Channel 06/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO Channel 09/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO Channel 14/0 : 22[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934721:934885 [1] NCCL INFO Channel 04/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934723:934887 [3] NCCL INFO Channel 02/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO Channel 10/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO Channel 00/0 : 6[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934727:934886 [7] NCCL INFO Channel 06/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934722:934880 [2] NCCL INFO Channel 01/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934725:934883 [5] NCCL INFO Channel 07/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934721:934885 [1] NCCL INFO Channel 05/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934723:934887 [3] NCCL INFO Channel 04/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO Channel 11/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO Channel 02/0 : 6[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934727:934886 [7] NCCL INFO Channel 08/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934721:934885 [1] NCCL INFO Channel 06/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934722:934880 [2] NCCL INFO Channel 03/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934725:934883 [5] NCCL INFO Channel 08/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO Channel 13/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO Channel 04/0 : 6[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934723:934887 [3] NCCL INFO Channel 05/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934727:934886 [7] NCCL INFO Channel 09/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934721:934885 [1] NCCL INFO Channel 07/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO Channel 14/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO Channel 08/0 : 6[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934725:934883 [5] NCCL INFO Channel 09/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934722:934880 [2] NCCL INFO Channel 04/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934723:934887 [3] NCCL INFO Channel 06/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934727:934886 [7] NCCL INFO Channel 10/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934721:934885 [1] NCCL INFO Channel 08/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO Channel 15/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO Channel 10/0 : 6[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934725:934883 [5] NCCL INFO Channel 10/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934722:934880 [2] NCCL INFO Channel 05/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934723:934887 [3] NCCL INFO Channel 07/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO Channel 00/0 : 20[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO Channel 12/0 : 6[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934721:934885 [1] NCCL INFO Channel 10/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934727:934886 [7] NCCL INFO Channel 11/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO Channel 02/0 : 20[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO Channel 14/0 : 6[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934722:934880 [2] NCCL INFO Channel 06/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934725:934883 [5] NCCL INFO Channel 11/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO Channel 01/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO Channel 00/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934723:934887 [3] NCCL INFO Channel 08/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934721:934885 [1] NCCL INFO Channel 11/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934727:934886 [7] NCCL INFO Channel 12/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO Channel 06/0 : 20[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934722:934880 [2] NCCL INFO Channel 07/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO Channel 01/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO Channel 08/0 : 20[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO Channel 02/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934725:934883 [5] NCCL INFO Channel 12/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934723:934887 [3] NCCL INFO Channel 09/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934721:934885 [1] NCCL INFO Channel 12/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934727:934886 [7] NCCL INFO Channel 13/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934722:934880 [2] NCCL INFO Channel 08/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO Channel 02/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934721:934885 [1] NCCL INFO Channel 13/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934725:934883 [5] NCCL INFO Channel 14/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934723:934887 [3] NCCL INFO Channel 10/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO Channel 10/0 : 20[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO Channel 03/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934722:934880 [2] NCCL INFO Channel 09/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934727:934886 [7] NCCL INFO Channel 14/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO Channel 03/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO Channel 12/0 : 20[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO Channel 04/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934721:934885 [1] NCCL INFO Channel 14/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO Channel 04/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO Channel 14/0 : 20[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934725:934883 [5] NCCL INFO Channel 15/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934723:934887 [3] NCCL INFO Channel 12/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934722:934880 [2] NCCL INFO Channel 11/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934721:934885 [1] NCCL INFO Channel 15/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO Channel 05/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO Channel 06/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934722:934880 [2] NCCL INFO Channel 12/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934723:934887 [3] NCCL INFO Channel 13/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO Channel 06/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO Channel 07/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934722:934880 [2] NCCL INFO Channel 13/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934723:934887 [3] NCCL INFO Channel 14/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO Channel 00/0 : 23[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO Channel 07/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO Channel 08/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934722:934880 [2] NCCL INFO Channel 14/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934723:934887 [3] NCCL INFO Channel 15/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO Channel 09/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO Channel 02/0 : 23[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934724:934881 [4] NCCL INFO Channel 01/0 : 28[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO Channel 00/0 : 21[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO Channel 00/0 : 17[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO Channel 09/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934722:934880 [2] NCCL INFO Channel 15/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO Channel 04/0 : 23[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934724:934881 [4] NCCL INFO Channel 03/0 : 28[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO Channel 10/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO Channel 02/0 : 21[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO Channel 10/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO Channel 00/0 : 4[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO Channel 02/0 : 17[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934724:934881 [4] NCCL INFO Channel 05/0 : 28[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO Channel 06/0 : 23[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO Channel 11/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO Channel 04/0 : 21[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO Channel 01/0 : 12[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO Channel 04/0 : 17[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO Channel 11/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO Channel 02/0 : 4[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO Channel 08/0 : 23[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO Channel 12/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934724:934881 [4] NCCL INFO Channel 07/0 : 28[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO Channel 03/0 : 12[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO Channel 06/0 : 21[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO Channel 00/0 : 19[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO Channel 12/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO Channel 06/0 : 17[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO Channel 06/0 : 4[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO Channel 14/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934724:934881 [4] NCCL INFO Channel 09/0 : 28[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO Channel 00/0 : 18[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO Channel 10/0 : 23[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO Channel 05/0 : 12[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO Channel 08/0 : 21[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO Channel 13/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO Channel 02/0 : 19[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO Channel 08/0 : 17[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934724:934881 [4] NCCL INFO Channel 11/0 : 28[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO Channel 08/0 : 4[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO Channel 15/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO Channel 12/0 : 23[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO Channel 07/0 : 12[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO Channel 04/0 : 18[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO Channel 10/0 : 21[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO Channel 04/0 : 19[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO Channel 10/0 : 17[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934724:934881 [4] NCCL INFO Channel 13/0 : 28[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO Channel 14/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO Channel 09/0 : 12[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO Channel 14/0 : 23[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO Channel 00/0 : 21[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO Channel 10/0 : 4[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO Channel 06/0 : 18[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934724:934881 [4] NCCL INFO Channel 15/0 : 28[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO Channel 15/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO Channel 12/0 : 21[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO Channel 12/0 : 17[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO Channel 06/0 : 19[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO Channel 11/0 : 12[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO Channel 00/0 : 7[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO Channel 02/0 : 21[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO Channel 12/0 : 4[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO Channel 08/0 : 18[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934724:934881 [4] NCCL INFO Channel 01/0 : 12[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO Channel 02/0 : 16[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO Channel 14/0 : 21[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO Channel 14/0 : 17[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO Channel 08/0 : 19[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO Channel 02/0 : 7[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO Channel 13/0 : 12[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO Channel 04/0 : 21[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO Channel 00/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO Channel 14/0 : 4[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO Channel 10/0 : 18[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO Channel 04/0 : 16[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO Channel 00/0 : 5[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934724:934881 [4] NCCL INFO Channel 03/0 : 12[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO Channel 06/0 : 21[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO Channel 15/0 : 12[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO Channel 00/0 : 1[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO Channel 10/0 : 19[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934724:934881 [4] NCCL INFO Channel 05/0 : 12[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO Channel 04/0 : 7[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO Channel 12/0 : 18[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO Channel 02/0 : 5[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO Channel 00/0 : 20[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO Channel 02/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO Channel 06/0 : 16[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO Channel 00/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO Channel 00/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO Channel 01/0 : 28[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934724:934881 [4] NCCL INFO Channel 07/0 : 12[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO Channel 02/0 : 1[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO Channel 06/0 : 7[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO Channel 12/0 : 19[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO Channel 08/0 : 21[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO Channel 00/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO Channel 03/0 : 28[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO Channel 04/0 : 5[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO Channel 14/0 : 18[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934724:934881 [4] NCCL INFO Channel 09/0 : 12[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO Channel 03/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO Channel 02/0 : 20[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO Channel 08/0 : 16[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO Channel 00/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO Channel 01/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO Channel 01/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO Channel 10/0 : 21[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO Channel 08/0 : 7[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO Channel 04/0 : 1[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO Channel 01/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO Channel 05/0 : 28[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934724:934881 [4] NCCL INFO Channel 11/0 : 12[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO Channel 14/0 : 19[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO Channel 06/0 : 5[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO Channel 00/0 : 2[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO Channel 10/0 : 16[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934724:934881 [4] NCCL INFO Channel 13/0 : 12[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO Channel 07/0 : 28[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO Channel 10/0 : 7[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO Channel 06/0 : 20[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO Channel 06/0 : 1[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO Channel 04/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO Channel 00/0 : 3[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934724:934881 [4] NCCL INFO Channel 15/0 : 12[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO Channel 08/0 : 5[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO Channel 04/0 : 2[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO Channel 09/0 : 28[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO Channel 12/0 : 7[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO Channel 12/0 : 21[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO Channel 02/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO Channel 01/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO Channel 08/0 : 1[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO Channel 02/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO Channel 02/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO Channel 02/0 : 3[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO Channel 12/0 : 16[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO Channel 08/0 : 20[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO Channel 05/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO Channel 11/0 : 28[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO Channel 06/0 : 2[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO Channel 10/0 : 5[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO Channel 14/0 : 21[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO Channel 14/0 : 7[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO Channel 10/0 : 1[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO Channel 04/0 : 3[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO Channel 13/0 : 28[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO Channel 03/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO Channel 03/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO Channel 04/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO Channel 14/0 : 16[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO Channel 03/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO Channel 10/0 : 20[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO Channel 06/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO Channel 15/0 : 28[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO Channel 08/0 : 2[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO Channel 12/0 : 5[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO Channel 12/0 : 1[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO Channel 06/0 : 3[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO Channel 04/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO Channel 14/0 : 5[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO Channel 10/0 : 2[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO Channel 04/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO Channel 12/0 : 20[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO Channel 04/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO Channel 07/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO Channel 05/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO Channel 14/0 : 1[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO Channel 08/0 : 3[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO Channel 05/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO Channel 12/0 : 2[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934725:934883 [5] NCCL INFO Channel 01/0 : 29[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO Channel 05/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO Channel 10/0 : 3[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO Channel 14/0 : 2[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO Channel 12/0 : 3[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934720:934882 [0] NCCL INFO Channel 01/0 : 24[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934725:934883 [5] NCCL INFO Channel 03/0 : 29[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO Channel 08/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO Channel 14/0 : 20[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO Channel 05/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO Channel 06/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO Channel 06/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO Channel 06/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934720:934882 [0] NCCL INFO Channel 03/0 : 24[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO Channel 14/0 : 3[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934725:934883 [5] NCCL INFO Channel 05/0 : 29[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO Channel 10/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO Channel 07/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO Channel 07/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO Channel 07/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934720:934882 [0] NCCL INFO Channel 05/0 : 24[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO Channel 00/0 : 5[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO Channel 08/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO Channel 02/0 : 0[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO Channel 11/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO Channel 08/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-31:3197185:3197353 [5] NCCL INFO Channel 01/0 : 13[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO Channel 01/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934725:934883 [5] NCCL INFO Channel 07/0 : 29[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO Channel 08/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO Channel 08/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO Channel 03/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934720:934882 [0] NCCL INFO Channel 07/0 : 24[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934725:934883 [5] NCCL INFO Channel 09/0 : 29[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-31:3197185:3197353 [5] NCCL INFO Channel 03/0 : 13[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO Channel 09/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO Channel 02/0 : 5[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO Channel 00/0 : 28[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO Channel 04/0 : 0[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO Channel 12/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO Channel 09/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO Channel 09/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934720:934882 [0] NCCL INFO Channel 09/0 : 24[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO Channel 05/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934725:934883 [5] NCCL INFO Channel 11/0 : 29[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-31:3197185:3197353 [5] NCCL INFO Channel 05/0 : 13[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO Channel 09/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO Channel 07/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934720:934882 [0] NCCL INFO Channel 11/0 : 24[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934725:934883 [5] NCCL INFO Channel 15/0 : 29[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO Channel 10/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO Channel 04/0 : 5[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-31:3197185:3197353 [5] NCCL INFO Channel 07/0 : 13[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO Channel 09/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934720:934882 [0] NCCL INFO Channel 13/0 : 24[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934725:934883 [5] NCCL INFO Channel 01/0 : 13[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO Channel 06/0 : 0[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO Channel 13/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO Channel 02/0 : 28[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO Channel 11/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO Channel 10/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO Channel 10/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO Channel 11/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO Channel 06/0 : 5[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-31:3197185:3197353 [5] NCCL INFO Channel 09/0 : 13[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934720:934882 [0] NCCL INFO Channel 15/0 : 24[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO Channel 11/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO Channel 08/0 : 0[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934725:934883 [5] NCCL INFO Channel 03/0 : 13[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-31:3197185:3197353 [5] NCCL INFO Channel 11/0 : 13[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO Channel 13/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO Channel 14/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO Channel 06/0 : 28[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO Channel 12/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO Channel 12/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO Channel 11/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO Channel 12/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO Channel 08/0 : 5[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934720:934882 [0] NCCL INFO Channel 01/0 : 8[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO Channel 10/0 : 0[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934725:934883 [5] NCCL INFO Channel 05/0 : 13[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-31:3197185:3197353 [5] NCCL INFO Channel 15/0 : 13[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO Channel 15/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO Channel 15/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934720:934882 [0] NCCL INFO Channel 03/0 : 8[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934725:934883 [5] NCCL INFO Channel 07/0 : 13[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-31:3197185:3197353 [5] NCCL INFO Channel 01/0 : 29[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO Channel 08/0 : 28[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO Channel 13/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO Channel 13/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934720:934882 [0] NCCL INFO Channel 05/0 : 8[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934725:934883 [5] NCCL INFO Channel 09/0 : 13[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-31:3197185:3197353 [5] NCCL INFO Channel 03/0 : 29[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO Channel 12/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO Channel 13/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO Channel 10/0 : 5[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO Channel 12/0 : 0[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO Channel 00/0 : 17[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO Channel 10/0 : 28[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934720:934882 [0] NCCL INFO Channel 07/0 : 8[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934725:934883 [5] NCCL INFO Channel 11/0 : 13[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-31:3197185:3197353 [5] NCCL INFO Channel 05/0 : 29[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO Channel 14/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934720:934882 [0] NCCL INFO Channel 09/0 : 8[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934725:934883 [5] NCCL INFO Channel 15/0 : 13[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO Channel 14/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-31:3197185:3197353 [5] NCCL INFO Channel 07/0 : 29[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934720:934882 [0] NCCL INFO Channel 11/0 : 8[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO Channel 13/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO Channel 02/0 : 17[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO Channel 12/0 : 5[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO Channel 14/0 : 0[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO Channel 14/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO Channel 12/0 : 28[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO Channel 15/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO Channel 15/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934720:934882 [0] NCCL INFO Channel 13/0 : 8[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-31:3197185:3197353 [5] NCCL INFO Channel 09/0 : 29[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO Channel 15/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934720:934882 [0] NCCL INFO Channel 15/0 : 8[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-31:3197185:3197353 [5] NCCL INFO Channel 11/0 : 29[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO Channel 04/0 : 17[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO Channel 14/0 : 5[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO Channel 02/0 : 16[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO Channel 00/0 : 23[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO Channel 00/0 : 19[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO Channel 14/0 : 28[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO Channel 00/0 : 18[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO Channel 00/0 : 22[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-31:3197185:3197353 [5] NCCL INFO Channel 15/0 : 29[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO Channel 06/0 : 17[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO Channel 00/0 : 21[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO Channel 04/0 : 16[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO Channel 02/0 : 19[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO Channel 02/0 : 23[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO Channel 04/0 : 18[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO Channel 00/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO Channel 02/0 : 22[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO Channel 08/0 : 17[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO Channel 02/0 : 21[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO Channel 06/0 : 16[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO Channel 04/0 : 19[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-31:3197185:3197353 [5] NCCL INFO Channel 00/0 : 29[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934725:934883 [5] NCCL INFO Channel 00/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO Channel 04/0 : 23[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO Channel 06/0 : 18[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO Channel 01/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO Channel 10/0 : 17[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO Channel 04/0 : 22[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-31:3197185:3197353 [5] NCCL INFO Channel 02/0 : 29[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934725:934883 [5] NCCL INFO Channel 01/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO Channel 08/0 : 16[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO Channel 04/0 : 21[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-31:3197185:3197353 [5] NCCL INFO Channel 04/0 : 29[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO Channel 06/0 : 19[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934725:934883 [5] NCCL INFO Channel 02/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO Channel 06/0 : 23[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO Channel 08/0 : 18[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-31:3197185:3197353 [5] NCCL INFO Channel 06/0 : 29[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO Channel 12/0 : 17[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO Channel 02/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934725:934883 [5] NCCL INFO Channel 03/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO Channel 10/0 : 16[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO Channel 08/0 : 19[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO Channel 08/0 : 22[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-31:3197185:3197353 [5] NCCL INFO Channel 08/0 : 29[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO Channel 06/0 : 21[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO Channel 08/0 : 23[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO Channel 10/0 : 18[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934725:934883 [5] NCCL INFO Channel 04/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO Channel 14/0 : 17[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-31:3197185:3197353 [5] NCCL INFO Channel 10/0 : 29[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934725:934883 [5] NCCL INFO Channel 06/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO Channel 03/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO Channel 12/0 : 16[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-31:3197185:3197353 [5] NCCL INFO Channel 12/0 : 29[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO Channel 10/0 : 19[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO Channel 12/0 : 18[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO Channel 10/0 : 22[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO Channel 10/0 : 23[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO Channel 08/0 : 21[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934725:934883 [5] NCCL INFO Channel 07/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-31:3197185:3197353 [5] NCCL INFO Channel 14/0 : 29[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO Channel 05/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO Channel 14/0 : 16[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934725:934883 [5] NCCL INFO Channel 08/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO Channel 12/0 : 19[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO Channel 14/0 : 18[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO Channel 10/0 : 21[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO Channel 12/0 : 22[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO Channel 12/0 : 23[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO Channel 06/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934725:934883 [5] NCCL INFO Channel 09/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934721:934885 [1] NCCL INFO Channel 01/0 : 25[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO Channel 14/0 : 19[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO Channel 12/0 : 21[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO Channel 14/0 : 22[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO Channel 14/0 : 23[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO Channel 07/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934724:934881 [4] NCCL INFO Channel 00/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO Channel 00/0 : 28[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO Channel 14/0 : 21[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934725:934883 [5] NCCL INFO Channel 10/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934721:934885 [1] NCCL INFO Channel 03/0 : 25[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO Channel 08/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-31:3197181:3197357 [1] NCCL INFO Channel 01/0 : 9[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO Channel 02/0 : 24[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934725:934883 [5] NCCL INFO Channel 11/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934724:934881 [4] NCCL INFO Channel 01/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934721:934885 [1] NCCL INFO Channel 05/0 : 25[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO Channel 02/0 : 28[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO Channel 00/0 : 1[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO Channel 09/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934725:934883 [5] NCCL INFO Channel 12/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO Channel 04/0 : 24[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO Channel 06/0 : 28[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-31:3197181:3197357 [1] NCCL INFO Channel 03/0 : 9[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO Channel 10/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO Channel 02/0 : 1[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934721:934885 [1] NCCL INFO Channel 07/0 : 25[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934724:934881 [4] NCCL INFO Channel 02/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934722:934880 [2] NCCL INFO Channel 01/0 : 26[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO Channel 00/0 : 2[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO Channel 08/0 : 28[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934723:934887 [3] NCCL INFO Channel 01/0 : 27[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934725:934883 [5] NCCL INFO Channel 14/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934726:934884 [6] NCCL INFO Channel 01/0 : 30[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934721:934885 [1] NCCL INFO Channel 11/0 : 25[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934727:934886 [7] NCCL INFO Channel 01/0 : 31[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO Channel 00/0 : 29[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO Channel 06/0 : 24[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO Channel 04/0 : 1[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-31:3197181:3197357 [1] NCCL INFO Channel 05/0 : 9[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-31:3197182:3197359 [2] NCCL INFO Channel 01/0 : 10[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO Channel 00/0 : 3[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO Channel 04/0 : 2[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO Channel 11/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO Channel 10/0 : 28[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-31:3197181:3197357 [1] NCCL INFO Channel 07/0 : 9[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934722:934880 [2] NCCL INFO Channel 03/0 : 26[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934724:934881 [4] NCCL INFO Channel 03/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO Channel 00/0 : 6[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO Channel 00/0 : 7[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO Channel 02/0 : 29[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO Channel 06/0 : 1[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO Channel 08/0 : 24[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934725:934883 [5] NCCL INFO Channel 15/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934723:934887 [3] NCCL INFO Channel 03/0 : 27[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-31:3197186:3197354 [6] NCCL INFO Channel 01/0 : 14[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934721:934885 [1] NCCL INFO Channel 13/0 : 25[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934726:934884 [6] NCCL INFO Channel 03/0 : 30[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO Channel 12/0 : 28[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934727:934886 [7] NCCL INFO Channel 03/0 : 31[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-31:3197182:3197359 [2] NCCL INFO Channel 03/0 : 10[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934722:934880 [2] NCCL INFO Channel 05/0 : 26[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO Channel 02/0 : 3[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934724:934881 [4] NCCL INFO Channel 05/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO Channel 06/0 : 2[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO Channel 13/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO Channel 02/0 : 7[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO Channel 02/0 : 6[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-31:3197183:3197358 [3] NCCL INFO Channel 01/0 : 11[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-31:3197181:3197357 [1] NCCL INFO Channel 11/0 : 9[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-31:3197187:3197356 [7] NCCL INFO Channel 01/0 : 15[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934723:934887 [3] NCCL INFO Channel 05/0 : 27[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934725:934883 [5] NCCL INFO Channel 01/0 : 13[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-31:3197186:3197354 [6] NCCL INFO Channel 03/0 : 14[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934721:934885 [1] NCCL INFO Channel 15/0 : 25[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO Channel 04/0 : 29[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO Channel 08/0 : 1[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO Channel 10/0 : 24[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO Channel 14/0 : 28[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-31:3197182:3197359 [2] NCCL INFO Channel 05/0 : 10[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934726:934884 [6] NCCL INFO Channel 05/0 : 30[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934727:934886 [7] NCCL INFO Channel 05/0 : 31[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO Channel 04/0 : 3[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO Channel 08/0 : 2[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934722:934880 [2] NCCL INFO Channel 07/0 : 26[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934724:934881 [4] NCCL INFO Channel 06/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-31:3197181:3197357 [1] NCCL INFO Channel 13/0 : 9[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO Channel 14/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-31:3197183:3197358 [3] NCCL INFO Channel 03/0 : 11[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934725:934883 [5] NCCL INFO Channel 03/0 : 13[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO Channel 04/0 : 7[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-31:3197187:3197356 [7] NCCL INFO Channel 03/0 : 15[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934723:934887 [3] NCCL INFO Channel 07/0 : 27[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO Channel 04/0 : 6[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-31:3197186:3197354 [6] NCCL INFO Channel 05/0 : 14[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934721:934885 [1] NCCL INFO Channel 01/0 : 9[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO Channel 12/0 : 24[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO Channel 10/0 : 1[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-31:3197182:3197359 [2] NCCL INFO Channel 07/0 : 10[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934726:934884 [6] NCCL INFO Channel 07/0 : 30[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO Channel 06/0 : 29[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934727:934886 [7] NCCL INFO Channel 07/0 : 31[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934722:934880 [2] NCCL INFO Channel 09/0 : 26[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934724:934881 [4] NCCL INFO Channel 07/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-31:3197181:3197357 [1] NCCL INFO Channel 15/0 : 9[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO Channel 06/0 : 3[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-31:3197183:3197358 [3] NCCL INFO Channel 05/0 : 11[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO Channel 10/0 : 2[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-31:3197187:3197356 [7] NCCL INFO Channel 05/0 : 15[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-31:3197186:3197354 [6] NCCL INFO Channel 07/0 : 14[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-31:3197182:3197359 [2] NCCL INFO Channel 09/0 : 10[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO Channel 15/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-31:3197181:3197357 [1] NCCL INFO Channel 01/0 : 25[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934725:934883 [5] NCCL INFO Channel 05/0 : 13[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO Channel 06/0 : 7[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO Channel 08/0 : 6[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934723:934887 [3] NCCL INFO Channel 09/0 : 27[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934721:934885 [1] NCCL INFO Channel 03/0 : 9[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO Channel 14/0 : 24[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934726:934884 [6] NCCL INFO Channel 09/0 : 30[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO Channel 12/0 : 1[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934727:934886 [7] NCCL INFO Channel 09/0 : 31[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO Channel 08/0 : 29[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934722:934880 [2] NCCL INFO Channel 11/0 : 26[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-31:3197187:3197356 [7] NCCL INFO Channel 07/0 : 15[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO Channel 08/0 : 3[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934724:934881 [4] NCCL INFO Channel 08/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO Channel 12/0 : 2[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934725:934883 [5] NCCL INFO Channel 07/0 : 13[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO Channel 08/0 : 7[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-31:3197182:3197359 [2] NCCL INFO Channel 11/0 : 10[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-31:3197186:3197354 [6] NCCL INFO Channel 09/0 : 14[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-31:3197183:3197358 [3] NCCL INFO Channel 07/0 : 11[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934721:934885 [1] NCCL INFO Channel 05/0 : 9[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO Channel 10/0 : 6[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-31:3197181:3197357 [1] NCCL INFO Channel 03/0 : 25[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934723:934887 [3] NCCL INFO Channel 13/0 : 27[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO Channel 01/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934726:934884 [6] NCCL INFO Channel 11/0 : 30[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934727:934886 [7] NCCL INFO Channel 11/0 : 31[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO Channel 14/0 : 1[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO Channel 10/0 : 29[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-31:3197187:3197356 [7] NCCL INFO Channel 09/0 : 15[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-31:3197186:3197354 [6] NCCL INFO Channel 11/0 : 14[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO Channel 14/0 : 2[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO Channel 10/0 : 7[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO Channel 12/0 : 6[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO Channel 02/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO Channel 00/0 : 17[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO Channel 10/0 : 3[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO Channel 12/0 : 29[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO Channel 00/0 : 18[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-31:3197182:3197359 [2] NCCL INFO Channel 13/0 : 10[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-31:3197181:3197357 [1] NCCL INFO Channel 05/0 : 25[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-31:3197183:3197358 [3] NCCL INFO Channel 09/0 : 11[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-31:3197180:3197360 [0] NCCL INFO Channel 01/0 : 8[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-31:3197187:3197356 [7] NCCL INFO Channel 11/0 : 15[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-31:3197186:3197354 [6] NCCL INFO Channel 13/0 : 14[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO Channel 12/0 : 7[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO Channel 14/0 : 6[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO Channel 03/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO Channel 02/0 : 17[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-31:3197182:3197359 [2] NCCL INFO Channel 15/0 : 10[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-31:3197183:3197358 [3] NCCL INFO Channel 13/0 : 11[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-31:3197181:3197357 [1] NCCL INFO Channel 07/0 : 25[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-31:3197187:3197356 [7] NCCL INFO Channel 13/0 : 15[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-31:3197180:3197360 [0] NCCL INFO Channel 03/0 : 8[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO Channel 14/0 : 29[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO Channel 04/0 : 18[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-31:3197186:3197354 [6] NCCL INFO Channel 15/0 : 14[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO Channel 14/0 : 7[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO Channel 04/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO Channel 12/0 : 3[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO Channel 00/0 : 22[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO Channel 04/0 : 17[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO Channel 06/0 : 18[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO Channel 00/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO Channel 05/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO Channel 00/0 : 23[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO Channel 14/0 : 3[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO Channel 06/0 : 17[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO Channel 02/0 : 22[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO Channel 08/0 : 18[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO Channel 01/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO Channel 06/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO Channel 02/0 : 23[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO Channel 00/0 : 19[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO Channel 08/0 : 17[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO Channel 04/0 : 22[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO Channel 10/0 : 18[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO Channel 07/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO Channel 02/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO Channel 04/0 : 23[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO Channel 02/0 : 19[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO Channel 10/0 : 17[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO Channel 12/0 : 18[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO Channel 08/0 : 22[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO Channel 09/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO Channel 03/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO Channel 06/0 : 23[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO Channel 04/0 : 19[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO Channel 12/0 : 17[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO Channel 14/0 : 18[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO Channel 10/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO Channel 10/0 : 22[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO Channel 04/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO Channel 06/0 : 19[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO Channel 14/0 : 17[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO Channel 08/0 : 23[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO Channel 11/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO Channel 12/0 : 22[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO Channel 06/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO Channel 08/0 : 19[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO Channel 10/0 : 23[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO Channel 12/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO Channel 14/0 : 22[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO Channel 07/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO Channel 00/0 : 26[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO Channel 10/0 : 19[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO Channel 13/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO Channel 12/0 : 23[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO Channel 08/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO Channel 04/0 : 26[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO Channel 12/0 : 19[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO Channel 14/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO Channel 14/0 : 23[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO Channel 09/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934722:934880 [2] NCCL INFO Channel 13/0 : 26[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934724:934881 [4] NCCL INFO Channel 09/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934725:934883 [5] NCCL INFO Channel 09/0 : 13[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934721:934885 [1] NCCL INFO Channel 07/0 : 9[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934723:934887 [3] NCCL INFO Channel 15/0 : 27[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934726:934884 [6] NCCL INFO Channel 13/0 : 30[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934727:934886 [7] NCCL INFO Channel 13/0 : 31[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934722:934880 [2] NCCL INFO Channel 15/0 : 26[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934724:934881 [4] NCCL INFO Channel 10/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934725:934883 [5] NCCL INFO Channel 11/0 : 13[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934721:934885 [1] NCCL INFO Channel 11/0 : 9[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934723:934887 [3] NCCL INFO Channel 01/0 : 11[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934727:934886 [7] NCCL INFO Channel 01/0 : 15[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934726:934884 [6] NCCL INFO Channel 15/0 : 30[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934722:934880 [2] NCCL INFO Channel 01/0 : 10[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934724:934881 [4] NCCL INFO Channel 11/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934725:934883 [5] NCCL INFO Channel 15/0 : 13[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934721:934885 [1] NCCL INFO Channel 13/0 : 9[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934723:934887 [3] NCCL INFO Channel 03/0 : 11[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934722:934880 [2] NCCL INFO Channel 03/0 : 10[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934726:934884 [6] NCCL INFO Channel 01/0 : 14[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934727:934886 [7] NCCL INFO Channel 03/0 : 15[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934724:934881 [4] NCCL INFO Channel 13/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934721:934885 [1] NCCL INFO Channel 15/0 : 9[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934723:934887 [3] NCCL INFO Channel 05/0 : 11[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934722:934880 [2] NCCL INFO Channel 05/0 : 10[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934726:934884 [6] NCCL INFO Channel 03/0 : 14[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934727:934886 [7] NCCL INFO Channel 05/0 : 15[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934724:934881 [4] NCCL INFO Channel 14/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934723:934887 [3] NCCL INFO Channel 07/0 : 11[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934722:934880 [2] NCCL INFO Channel 07/0 : 10[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934727:934886 [7] NCCL INFO Channel 07/0 : 15[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934726:934884 [6] NCCL INFO Channel 05/0 : 14[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934723:934887 [3] NCCL INFO Channel 09/0 : 11[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934724:934881 [4] NCCL INFO Channel 15/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934722:934880 [2] NCCL INFO Channel 09/0 : 10[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934727:934886 [7] NCCL INFO Channel 09/0 : 15[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934726:934884 [6] NCCL INFO Channel 07/0 : 14[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934723:934887 [3] NCCL INFO Channel 13/0 : 11[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934722:934880 [2] NCCL INFO Channel 11/0 : 10[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934724:934881 [4] NCCL INFO Channel 01/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934727:934886 [7] NCCL INFO Channel 11/0 : 15[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934726:934884 [6] NCCL INFO Channel 09/0 : 14[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934723:934887 [3] NCCL INFO Channel 15/0 : 11[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934724:934881 [4] NCCL INFO Channel 03/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934726:934884 [6] NCCL INFO Channel 11/0 : 14[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934727:934886 [7] NCCL INFO Channel 13/0 : 15[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO Channel 01/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934721:934885 [1] NCCL INFO Channel 00/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934726:934884 [6] NCCL INFO Channel 13/0 : 14[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934724:934881 [4] NCCL INFO Channel 05/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934721:934885 [1] NCCL INFO Channel 02/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934722:934880 [2] NCCL INFO Channel 13/0 : 10[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934724:934881 [4] NCCL INFO Channel 07/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934726:934884 [6] NCCL INFO Channel 15/0 : 14[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934721:934885 [1] NCCL INFO Channel 03/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934722:934880 [2] NCCL INFO Channel 15/0 : 10[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934724:934881 [4] NCCL INFO Channel 09/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934721:934885 [1] NCCL INFO Channel 04/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO Channel 03/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934727:934886 [7] NCCL INFO Channel 00/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934724:934881 [4] NCCL INFO Channel 11/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934721:934885 [1] NCCL INFO Channel 05/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934727:934886 [7] NCCL INFO Channel 01/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934724:934881 [4] NCCL INFO Channel 13/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934727:934886 [7] NCCL INFO Channel 02/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934724:934881 [4] NCCL INFO Channel 15/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934721:934885 [1] NCCL INFO Channel 06/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934727:934886 [7] NCCL INFO Channel 03/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934721:934885 [1] NCCL INFO Channel 07/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934723:934887 [3] NCCL INFO Channel 00/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934727:934886 [7] NCCL INFO Channel 04/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934721:934885 [1] NCCL INFO Channel 08/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934723:934887 [3] NCCL INFO Channel 01/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934727:934886 [7] NCCL INFO Channel 05/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO Channel 05/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934721:934885 [1] NCCL INFO Channel 10/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934723:934887 [3] NCCL INFO Channel 02/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934727:934886 [7] NCCL INFO Channel 06/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934721:934885 [1] NCCL INFO Channel 11/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934723:934887 [3] NCCL INFO Channel 04/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934727:934886 [7] NCCL INFO Channel 08/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934720:934882 [0] NCCL INFO Channel 01/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934721:934885 [1] NCCL INFO Channel 12/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO Channel 07/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934723:934887 [3] NCCL INFO Channel 05/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934727:934886 [7] NCCL INFO Channel 09/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934720:934882 [0] NCCL INFO Channel 02/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934721:934885 [1] NCCL INFO Channel 13/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934723:934887 [3] NCCL INFO Channel 06/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934727:934886 [7] NCCL INFO Channel 10/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934721:934885 [1] NCCL INFO Channel 14/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934720:934882 [0] NCCL INFO Channel 03/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934723:934887 [3] NCCL INFO Channel 07/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934727:934886 [7] NCCL INFO Channel 11/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO Channel 09/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934721:934885 [1] NCCL INFO Channel 15/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934720:934882 [0] NCCL INFO Channel 04/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934723:934887 [3] NCCL INFO Channel 08/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934727:934886 [7] NCCL INFO Channel 12/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934721:934885 [1] NCCL INFO Channel 01/0 : 9[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934720:934882 [0] NCCL INFO Channel 05/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934723:934887 [3] NCCL INFO Channel 09/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934727:934886 [7] NCCL INFO Channel 13/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934721:934885 [1] NCCL INFO Channel 03/0 : 9[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934720:934882 [0] NCCL INFO Channel 06/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934723:934887 [3] NCCL INFO Channel 10/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934727:934886 [7] NCCL INFO Channel 14/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO Channel 11/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934721:934885 [1] NCCL INFO Channel 05/0 : 9[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934720:934882 [0] NCCL INFO Channel 07/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934723:934887 [3] NCCL INFO Channel 12/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934727:934886 [7] NCCL INFO Channel 01/0 : 15[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934721:934885 [1] NCCL INFO Channel 07/0 : 9[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO Channel 13/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934720:934882 [0] NCCL INFO Channel 09/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934723:934887 [3] NCCL INFO Channel 13/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934727:934886 [7] NCCL INFO Channel 03/0 : 15[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934721:934885 [1] NCCL INFO Channel 11/0 : 9[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO Channel 15/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934720:934882 [0] NCCL INFO Channel 10/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934723:934887 [3] NCCL INFO Channel 14/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934727:934886 [7] NCCL INFO Channel 05/0 : 15[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934721:934885 [1] NCCL INFO Channel 13/0 : 9[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934720:934882 [0] NCCL INFO Channel 11/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934723:934887 [3] NCCL INFO Channel 15/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934727:934886 [7] NCCL INFO Channel 07/0 : 15[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934721:934885 [1] NCCL INFO Channel 15/0 : 9[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934720:934882 [0] NCCL INFO Channel 12/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934723:934887 [3] NCCL INFO Channel 01/0 : 11[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934722:934880 [2] NCCL INFO Channel 00/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934727:934886 [7] NCCL INFO Channel 09/0 : 15[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934720:934882 [0] NCCL INFO Channel 13/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934723:934887 [3] NCCL INFO Channel 03/0 : 11[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934727:934886 [7] NCCL INFO Channel 11/0 : 15[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934722:934880 [2] NCCL INFO Channel 01/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934720:934882 [0] NCCL INFO Channel 14/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934723:934887 [3] NCCL INFO Channel 05/0 : 11[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934727:934886 [7] NCCL INFO Channel 13/0 : 15[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-31:3197182:3197359 [2] NCCL INFO Channel 01/0 : 26[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934723:934887 [3] NCCL INFO Channel 07/0 : 11[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934720:934882 [0] NCCL INFO Channel 15/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-31:3197183:3197358 [3] NCCL INFO Channel 15/0 : 11[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934722:934880 [2] NCCL INFO Channel 03/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-31:3197181:3197357 [1] NCCL INFO Channel 11/0 : 25[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-31:3197187:3197356 [7] NCCL INFO Channel 01/0 : 31[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-31:3197186:3197354 [6] NCCL INFO Channel 01/0 : 30[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-31:3197182:3197359 [2] NCCL INFO Channel 03/0 : 26[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-31:3197180:3197360 [0] NCCL INFO Channel 05/0 : 8[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO Channel 06/0 : 26[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO Channel 14/0 : 19[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-31:3197181:3197357 [1] NCCL INFO Channel 13/0 : 25[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO Channel 15/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-31:3197187:3197356 [7] NCCL INFO Channel 03/0 : 31[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO Channel 10/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO Channel 00/0 : 30[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO Channel 08/0 : 26[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-31:3197180:3197360 [0] NCCL INFO Channel 07/0 : 8[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934723:934887 [3] NCCL INFO Channel 09/0 : 11[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934720:934882 [0] NCCL INFO Channel 01/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO Channel 02/0 : 30[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO Channel 11/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO Channel 10/0 : 26[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-31:3197181:3197357 [1] NCCL INFO Channel 15/0 : 25[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-31:3197187:3197356 [7] NCCL INFO Channel 05/0 : 31[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934722:934880 [2] NCCL INFO Channel 04/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO Channel 00/0 : 31[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-31:3197180:3197360 [0] NCCL INFO Channel 09/0 : 8[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO Channel 00/0 : 27[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-31:3197187:3197356 [7] NCCL INFO Channel 07/0 : 31[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO Channel 12/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO Channel 04/0 : 30[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO Channel 12/0 : 26[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-31:3197180:3197360 [0] NCCL INFO Channel 11/0 : 8[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO Channel 02/0 : 31[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-31:3197183:3197358 [3] NCCL INFO Channel 01/0 : 27[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO Channel 02/0 : 27[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-31:3197187:3197356 [7] NCCL INFO Channel 09/0 : 31[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO Channel 14/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-31:3197180:3197360 [0] NCCL INFO Channel 13/0 : 8[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO Channel 14/0 : 26[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO Channel 08/0 : 30[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-31:3197183:3197358 [3] NCCL INFO Channel 03/0 : 27[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934720:934882 [0] NCCL INFO Channel 03/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934723:934887 [3] NCCL INFO Channel 13/0 : 11[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO Channel 04/0 : 31[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO Channel 04/0 : 27[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-31:3197187:3197356 [7] NCCL INFO Channel 11/0 : 31[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934722:934880 [2] NCCL INFO Channel 05/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO Channel 15/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-31:3197181:3197357 [1] NCCL INFO Channel 00/0 : 25[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO Channel 00/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-31:3197180:3197360 [0] NCCL INFO Channel 15/0 : 8[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO Channel 10/0 : 30[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO Channel 06/0 : 27[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-31:3197183:3197358 [3] NCCL INFO Channel 05/0 : 27[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO Channel 06/0 : 31[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-31:3197187:3197356 [7] NCCL INFO Channel 13/0 : 31[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO Channel 01/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-31:3197180:3197360 [0] NCCL INFO Channel 01/0 : 24[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-31:3197181:3197357 [1] NCCL INFO Channel 02/0 : 25[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO Channel 08/0 : 27[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO Channel 12/0 : 30[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO Channel 08/0 : 31[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-31:3197183:3197358 [3] NCCL INFO Channel 07/0 : 27[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934720:934882 [0] NCCL INFO Channel 05/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO Channel 03/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-31:3197180:3197360 [0] NCCL INFO Channel 03/0 : 24[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-31:3197181:3197357 [1] NCCL INFO Channel 04/0 : 25[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-31:3197183:3197358 [3] NCCL INFO Channel 09/0 : 27[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934723:934887 [3] NCCL INFO Channel 15/0 : 11[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO Channel 10/0 : 27[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO Channel 10/0 : 31[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-31:3197180:3197360 [0] NCCL INFO Channel 05/0 : 24[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934722:934880 [2] NCCL INFO Channel 06/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO Channel 14/0 : 30[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO Channel 04/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-31:3197183:3197358 [3] NCCL INFO Channel 13/0 : 27[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-31:3197181:3197357 [1] NCCL INFO Channel 06/0 : 25[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO Channel 12/0 : 27[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-31:3197187:3197356 [7] NCCL INFO Channel 00/0 : 31[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO Channel 00/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO Channel 05/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO Channel 12/0 : 31[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-31:3197180:3197360 [0] NCCL INFO Channel 07/0 : 24[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO Channel 14/0 : 27[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-31:3197183:3197358 [3] NCCL INFO Channel 15/0 : 27[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-31:3197181:3197357 [1] NCCL INFO Channel 08/0 : 25[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO Channel 01/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-31:3197187:3197356 [7] NCCL INFO Channel 02/0 : 31[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO Channel 06/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO Channel 14/0 : 31[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-31:3197180:3197360 [0] NCCL INFO Channel 09/0 : 24[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934720:934882 [0] NCCL INFO Channel 07/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO Channel 00/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-31:3197181:3197357 [1] NCCL INFO Channel 10/0 : 25[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934722:934880 [2] NCCL INFO Channel 07/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO Channel 02/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-31:3197187:3197356 [7] NCCL INFO Channel 04/0 : 31[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO Channel 07/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO Channel 00/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-31:3197180:3197360 [0] NCCL INFO Channel 11/0 : 24[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO Channel 01/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-31:3197181:3197357 [1] NCCL INFO Channel 12/0 : 25[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO Channel 00/0 : 25[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-31:3197187:3197356 [7] NCCL INFO Channel 06/0 : 31[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO Channel 03/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-31:3197180:3197360 [0] NCCL INFO Channel 13/0 : 24[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO Channel 08/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO Channel 01/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-31:3197187:3197356 [7] NCCL INFO Channel 08/0 : 31[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-31:3197181:3197357 [1] NCCL INFO Channel 14/0 : 25[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-31:3197183:3197358 [3] NCCL INFO Channel 00/0 : 27[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934720:934882 [0] NCCL INFO Channel 09/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO Channel 02/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-31:3197180:3197360 [0] NCCL INFO Channel 15/0 : 24[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934722:934880 [2] NCCL INFO Channel 08/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO Channel 02/0 : 25[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO Channel 04/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-31:3197187:3197356 [7] NCCL INFO Channel 10/0 : 31[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO Channel 09/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-31:3197183:3197358 [3] NCCL INFO Channel 02/0 : 27[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO Channel 02/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO Channel 04/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-31:3197187:3197356 [7] NCCL INFO Channel 12/0 : 31[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO Channel 05/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO Channel 04/0 : 25[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-31:3197183:3197358 [3] NCCL INFO Channel 04/0 : 27[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO Channel 11/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-31:3197187:3197356 [7] NCCL INFO Channel 14/0 : 31[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-31:3197183:3197358 [3] NCCL INFO Channel 06/0 : 27[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-31:3197180:3197360 [0] NCCL INFO Channel 02/0 : 24[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO Channel 05/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO Channel 03/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-31:3197180:3197360 [0] NCCL INFO Channel 04/0 : 24[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-31:3197183:3197358 [3] NCCL INFO Channel 08/0 : 27[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934720:934882 [0] NCCL INFO Channel 11/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO Channel 07/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO Channel 06/0 : 25[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO Channel 12/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-31:3197180:3197360 [0] NCCL INFO Channel 06/0 : 24[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-31:3197183:3197358 [3] NCCL INFO Channel 10/0 : 27[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO Channel 06/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO Channel 04/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-31:3197180:3197360 [0] NCCL INFO Channel 08/0 : 24[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-31:3197183:3197358 [3] NCCL INFO Channel 12/0 : 27[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO Channel 13/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO Channel 08/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO Channel 08/0 : 25[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-31:3197180:3197360 [0] NCCL INFO Channel 10/0 : 24[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-31:3197183:3197358 [3] NCCL INFO Channel 14/0 : 27[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO Channel 07/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO Channel 05/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-31:3197182:3197359 [2] NCCL INFO Channel 05/0 : 26[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO Channel 14/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-31:3197180:3197360 [0] NCCL INFO Channel 12/0 : 24[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934722:934880 [2] NCCL INFO Channel 09/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO Channel 09/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-31:3197182:3197359 [2] NCCL INFO Channel 07/0 : 26[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934720:934882 [0] NCCL INFO Channel 13/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO Channel 10/0 : 25[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-31:3197180:3197360 [0] NCCL INFO Channel 14/0 : 24[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO Channel 08/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO Channel 06/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-31:3197182:3197359 [2] NCCL INFO Channel 09/0 : 26[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO Channel 15/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO Channel 10/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-31:3197182:3197359 [2] NCCL INFO Channel 11/0 : 26[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-31:3197182:3197359 [2] NCCL INFO Channel 13/0 : 26[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-31:3197182:3197359 [2] NCCL INFO Channel 15/0 : 26[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934720:934882 [0] NCCL INFO Channel 15/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934722:934880 [2] NCCL INFO Channel 11/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-31:3197182:3197359 [2] NCCL INFO Channel 00/0 : 26[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-31:3197182:3197359 [2] NCCL INFO Channel 04/0 : 26[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO Channel 12/0 : 25[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-31:3197182:3197359 [2] NCCL INFO Channel 06/0 : 26[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO Channel 09/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO Channel 08/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-31:3197182:3197359 [2] NCCL INFO Channel 08/0 : 26[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO Channel 11/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-31:3197182:3197359 [2] NCCL INFO Channel 10/0 : 26[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-31:3197186:3197354 [6] NCCL INFO Channel 03/0 : 30[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934722:934880 [2] NCCL INFO Channel 12/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO Channel 14/0 : 25[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO Channel 10/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO Channel 09/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO Channel 12/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO Channel 00/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO Channel 12/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO Channel 13/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO Channel 10/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-31:3197182:3197359 [2] NCCL INFO Channel 12/0 : 26[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934722:934880 [2] NCCL INFO Channel 13/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO Channel 02/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-31:3197186:3197354 [6] NCCL INFO Channel 05/0 : 30[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO Channel 13/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-31:3197182:3197359 [2] NCCL INFO Channel 14/0 : 26[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO Channel 15/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-31:3197186:3197354 [6] NCCL INFO Channel 07/0 : 30[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO Channel 11/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO Channel 03/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-31:3197186:3197354 [6] NCCL INFO Channel 09/0 : 30[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO Channel 14/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-31:3197186:3197354 [6] NCCL INFO Channel 11/0 : 30[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO Channel 12/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO Channel 04/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO Channel 15/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-31:3197186:3197354 [6] NCCL INFO Channel 13/0 : 30[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO Channel 05/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO Channel 13/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-31:3197186:3197354 [6] NCCL INFO Channel 15/0 : 30[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO Channel 01/0 : 15[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934722:934880 [2] NCCL INFO Channel 14/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-31:3197186:3197354 [6] NCCL INFO Channel 00/0 : 30[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO Channel 06/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO Channel 14/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO Channel 07/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934726:934884 [6] NCCL INFO Channel 00/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-31:3197186:3197354 [6] NCCL INFO Channel 02/0 : 30[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO Channel 03/0 : 15[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934722:934880 [2] NCCL INFO Channel 15/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO Channel 08/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO Channel 05/0 : 15[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934726:934884 [6] NCCL INFO Channel 01/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934722:934880 [2] NCCL INFO Channel 01/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO Channel 10/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-31:3197186:3197354 [6] NCCL INFO Channel 04/0 : 30[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934726:934884 [6] NCCL INFO Channel 02/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934722:934880 [2] NCCL INFO Channel 03/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO Channel 11/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO Channel 07/0 : 15[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-31:3197186:3197354 [6] NCCL INFO Channel 08/0 : 30[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-31:3197186:3197354 [6] NCCL INFO Channel 10/0 : 30[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934726:934884 [6] NCCL INFO Channel 03/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934722:934880 [2] NCCL INFO Channel 05/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO Channel 12/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO Channel 09/0 : 15[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-31:3197186:3197354 [6] NCCL INFO Channel 12/0 : 30[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934726:934884 [6] NCCL INFO Channel 04/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934722:934880 [2] NCCL INFO Channel 07/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO Channel 13/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO Channel 11/0 : 15[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-31:3197186:3197354 [6] NCCL INFO Channel 14/0 : 30[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934726:934884 [6] NCCL INFO Channel 05/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934722:934880 [2] NCCL INFO Channel 09/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO Channel 14/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO Channel 13/0 : 15[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO Channel 15/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934726:934884 [6] NCCL INFO Channel 07/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934722:934880 [2] NCCL INFO Channel 11/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934726:934884 [6] NCCL INFO Channel 08/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934722:934880 [2] NCCL INFO Channel 13/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934726:934884 [6] NCCL INFO Channel 09/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934722:934880 [2] NCCL INFO Channel 15/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934726:934884 [6] NCCL INFO Channel 10/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934726:934884 [6] NCCL INFO Channel 11/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934726:934884 [6] NCCL INFO Channel 12/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934726:934884 [6] NCCL INFO Channel 13/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934726:934884 [6] NCCL INFO Channel 15/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934726:934884 [6] NCCL INFO Channel 01/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934726:934884 [6] NCCL INFO Channel 03/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934726:934884 [6] NCCL INFO Channel 05/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934726:934884 [6] NCCL INFO Channel 07/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO Channel 01/0 : 9[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934726:934884 [6] NCCL INFO Channel 09/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934726:934884 [6] NCCL INFO Channel 11/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO Channel 03/0 : 9[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934726:934884 [6] NCCL INFO Channel 13/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO Channel 05/0 : 9[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934726:934884 [6] NCCL INFO Channel 15/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO Channel 07/0 : 9[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO Channel 11/0 : 9[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO Channel 01/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO Channel 13/0 : 9[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO Channel 03/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO Channel 15/0 : 9[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO Channel 05/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO Channel 07/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO Channel 09/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO Channel 11/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO Channel 13/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO Channel 15/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO Channel 01/0 : 11[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO Channel 03/0 : 11[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO Channel 01/0 : 13[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO Channel 03/0 : 13[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO Channel 05/0 : 11[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO Channel 05/0 : 13[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO Channel 07/0 : 11[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO Channel 07/0 : 13[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO Channel 09/0 : 11[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO Channel 09/0 : 13[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO Channel 13/0 : 11[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO Channel 11/0 : 13[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO Channel 15/0 : 11[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO Channel 15/0 : 13[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO Channel 01/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO Channel 03/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO Channel 05/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO Channel 07/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO Channel 09/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO Channel 11/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO Channel 13/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO Channel 15/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-31:3197181:3197357 [1] NCCL INFO Connected NVLS tree
+ip-26-0-170-31:3197181:3197357 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-170-31:3197181:3197357 [1] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-170-31:3197181:3197357 [1] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-170-31:3197185:3197353 [5] NCCL INFO Connected NVLS tree
+ip-26-0-170-31:3197185:3197353 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-170-31:3197185:3197353 [5] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-170-31:3197185:3197353 [5] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-170-31:3197183:3197358 [3] NCCL INFO Connected NVLS tree
+ip-26-0-170-31:3197183:3197358 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-170-31:3197183:3197358 [3] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-170-31:3197183:3197358 [3] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO Connected NVLS tree
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-170-31:3197186:3197354 [6] NCCL INFO Connected NVLS tree
+ip-26-0-170-31:3197186:3197354 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-170-31:3197186:3197354 [6] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-170-31:3197186:3197354 [6] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO Connected NVLS tree
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-170-31:3197180:3197360 [0] NCCL INFO Connected NVLS tree
+ip-26-0-170-31:3197180:3197360 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-170-31:3197180:3197360 [0] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-170-31:3197180:3197360 [0] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-170-31:3197182:3197359 [2] NCCL INFO Connected NVLS tree
+ip-26-0-170-31:3197182:3197359 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-170-31:3197182:3197359 [2] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-170-31:3197182:3197359 [2] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO Connected NVLS tree
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO Connected NVLS tree
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-170-31:3197187:3197356 [7] NCCL INFO Connected NVLS tree
+ip-26-0-170-31:3197187:3197356 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-170-31:3197187:3197356 [7] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-170-31:3197187:3197356 [7] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO Connected NVLS tree
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-163-58:934721:934885 [1] NCCL INFO Connected NVLS tree
+ip-26-0-163-58:934721:934885 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-163-58:934721:934885 [1] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-163-58:934721:934885 [1] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO Connected NVLS tree
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO Connected NVLS tree
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-170-31:3197184:3197355 [4] NCCL INFO comm 0x9ec45d0 rank 28 nranks 32 cudaDev 4 nvmlDev 4 busId 97000 commId 0x1214b375c22bac3c - Init COMPLETE
+ip-26-0-170-31:3197182:3197359 [2] NCCL INFO comm 0x86c2450 rank 26 nranks 32 cudaDev 2 nvmlDev 2 busId 75000 commId 0x1214b375c22bac3c - Init COMPLETE
+ip-26-0-170-31:3197186:3197354 [6] NCCL INFO comm 0x9df89f0 rank 30 nranks 32 cudaDev 6 nvmlDev 6 busId b9000 commId 0x1214b375c22bac3c - Init COMPLETE
+ip-26-0-170-31:3197180:3197360 [0] NCCL INFO comm 0x9f60150 rank 24 nranks 32 cudaDev 0 nvmlDev 0 busId 53000 commId 0x1214b375c22bac3c - Init COMPLETE
+ip-26-0-170-31:3197183:3197358 [3] NCCL INFO comm 0x9d302b0 rank 27 nranks 32 cudaDev 3 nvmlDev 3 busId 86000 commId 0x1214b375c22bac3c - Init COMPLETE
+ip-26-0-170-31:3197187:3197356 [7] NCCL INFO comm 0xa426700 rank 31 nranks 32 cudaDev 7 nvmlDev 7 busId ca000 commId 0x1214b375c22bac3c - Init COMPLETE
+ip-26-0-170-31:3197185:3197353 [5] NCCL INFO comm 0x8610a20 rank 29 nranks 32 cudaDev 5 nvmlDev 5 busId a8000 commId 0x1214b375c22bac3c - Init COMPLETE
+ip-26-0-170-31:3197181:3197357 [1] NCCL INFO comm 0x998b390 rank 25 nranks 32 cudaDev 1 nvmlDev 1 busId 64000 commId 0x1214b375c22bac3c - Init COMPLETE
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO Connected NVLS tree
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO Connected NVLS tree
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-163-58:934725:934883 [5] NCCL INFO Connected NVLS tree
+ip-26-0-163-58:934725:934883 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-163-58:934725:934883 [5] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-163-58:934725:934883 [5] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-163-58:934723:934887 [3] NCCL INFO Connected NVLS tree
+ip-26-0-163-58:934723:934887 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-163-58:934723:934887 [3] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-163-58:934723:934887 [3] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-163-58:934727:934886 [7] NCCL INFO Connected NVLS tree
+ip-26-0-163-58:934727:934886 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-163-58:934727:934886 [7] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-163-58:934727:934886 [7] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO Connected NVLS tree
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO Connected NVLS tree
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO Connected NVLS tree
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO Connected NVLS tree
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-163-58:934722:934880 [2] NCCL INFO Connected NVLS tree
+ip-26-0-163-58:934722:934880 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-163-58:934722:934880 [2] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-163-58:934722:934880 [2] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO Connected NVLS tree
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO Connected NVLS tree
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO Connected NVLS tree
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-163-58:934726:934884 [6] NCCL INFO Connected NVLS tree
+ip-26-0-163-58:934726:934884 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-163-58:934726:934884 [6] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-163-58:934726:934884 [6] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-162-14:2525081:2525240 [6] NCCL INFO comm 0x92b7800 rank 6 nranks 32 cudaDev 6 nvmlDev 6 busId b9000 commId 0x1214b375c22bac3c - Init COMPLETE
+ip-26-0-162-14:2525078:2525244 [3] NCCL INFO comm 0x8fe8d30 rank 3 nranks 32 cudaDev 3 nvmlDev 3 busId 86000 commId 0x1214b375c22bac3c - Init COMPLETE
+ip-26-0-162-14:2525080:2525241 [5] NCCL INFO comm 0x9a0b900 rank 5 nranks 32 cudaDev 5 nvmlDev 5 busId a8000 commId 0x1214b375c22bac3c - Init COMPLETE
+ip-26-0-162-14:2525077:2525242 [2] NCCL INFO comm 0x9f7c5a0 rank 2 nranks 32 cudaDev 2 nvmlDev 2 busId 75000 commId 0x1214b375c22bac3c - Init COMPLETE
+ip-26-0-162-14:2525079:2525243 [4] NCCL INFO comm 0x94864f0 rank 4 nranks 32 cudaDev 4 nvmlDev 4 busId 97000 commId 0x1214b375c22bac3c - Init COMPLETE
+ip-26-0-162-14:2525076:2525238 [1] NCCL INFO comm 0xa132980 rank 1 nranks 32 cudaDev 1 nvmlDev 1 busId 64000 commId 0x1214b375c22bac3c - Init COMPLETE
+ip-26-0-162-14:2525075:2525237 [0] NCCL INFO comm 0x962f910 rank 0 nranks 32 cudaDev 0 nvmlDev 0 busId 53000 commId 0x1214b375c22bac3c - Init COMPLETE
+ip-26-0-162-14:2525082:2525239 [7] NCCL INFO comm 0x9c385f0 rank 7 nranks 32 cudaDev 7 nvmlDev 7 busId ca000 commId 0x1214b375c22bac3c - Init COMPLETE
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO Connected NVLS tree
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-163-58:934724:934881 [4] NCCL INFO Connected NVLS tree
+ip-26-0-163-58:934724:934881 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-163-58:934724:934881 [4] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-163-58:934724:934881 [4] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-163-58:934720:934882 [0] NCCL INFO Connected NVLS tree
+ip-26-0-163-58:934720:934882 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-163-58:934720:934882 [0] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-163-58:934720:934882 [0] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-170-132:3330190:3330358 [7] NCCL INFO comm 0xa2e0240 rank 23 nranks 32 cudaDev 7 nvmlDev 7 busId ca000 commId 0x1214b375c22bac3c - Init COMPLETE
+ip-26-0-170-132:3330186:3330361 [3] NCCL INFO comm 0x8cd8e90 rank 19 nranks 32 cudaDev 3 nvmlDev 3 busId 86000 commId 0x1214b375c22bac3c - Init COMPLETE
+ip-26-0-170-132:3330184:3330363 [1] NCCL INFO comm 0x93cf5d0 rank 17 nranks 32 cudaDev 1 nvmlDev 1 busId 64000 commId 0x1214b375c22bac3c - Init COMPLETE
+ip-26-0-170-132:3330188:3330359 [5] NCCL INFO comm 0xa2b01f0 rank 21 nranks 32 cudaDev 5 nvmlDev 5 busId a8000 commId 0x1214b375c22bac3c - Init COMPLETE
+ip-26-0-170-132:3330187:3330360 [4] NCCL INFO comm 0x97fbc80 rank 20 nranks 32 cudaDev 4 nvmlDev 4 busId 97000 commId 0x1214b375c22bac3c - Init COMPLETE
+ip-26-0-170-132:3330189:3330356 [6] NCCL INFO comm 0x856fc20 rank 22 nranks 32 cudaDev 6 nvmlDev 6 busId b9000 commId 0x1214b375c22bac3c - Init COMPLETE
+ip-26-0-170-132:3330183:3330362 [0] NCCL INFO comm 0x9296ef0 rank 16 nranks 32 cudaDev 0 nvmlDev 0 busId 53000 commId 0x1214b375c22bac3c - Init COMPLETE
+ip-26-0-170-132:3330185:3330357 [2] NCCL INFO comm 0x985f5f0 rank 18 nranks 32 cudaDev 2 nvmlDev 2 busId 75000 commId 0x1214b375c22bac3c - Init COMPLETE
+ip-26-0-163-58:934725:934883 [5] NCCL INFO comm 0x886a710 rank 13 nranks 32 cudaDev 5 nvmlDev 5 busId a8000 commId 0x1214b375c22bac3c - Init COMPLETE
+ip-26-0-163-58:934724:934881 [4] NCCL INFO comm 0x967ca40 rank 12 nranks 32 cudaDev 4 nvmlDev 4 busId 97000 commId 0x1214b375c22bac3c - Init COMPLETE
+ip-26-0-163-58:934722:934880 [2] NCCL INFO comm 0x8886e30 rank 10 nranks 32 cudaDev 2 nvmlDev 2 busId 75000 commId 0x1214b375c22bac3c - Init COMPLETE
+ip-26-0-163-58:934726:934884 [6] NCCL INFO comm 0x9bcf110 rank 14 nranks 32 cudaDev 6 nvmlDev 6 busId b9000 commId 0x1214b375c22bac3c - Init COMPLETE
+ip-26-0-163-58:934727:934886 [7] NCCL INFO comm 0x97de120 rank 15 nranks 32 cudaDev 7 nvmlDev 7 busId ca000 commId 0x1214b375c22bac3c - Init COMPLETE
+ip-26-0-163-58:934720:934882 [0] NCCL INFO comm 0x95c46a0 rank 8 nranks 32 cudaDev 0 nvmlDev 0 busId 53000 commId 0x1214b375c22bac3c - Init COMPLETE
+ip-26-0-163-58:934721:934885 [1] NCCL INFO comm 0xa3091e0 rank 9 nranks 32 cudaDev 1 nvmlDev 1 busId 64000 commId 0x1214b375c22bac3c - Init COMPLETE
+ip-26-0-163-58:934723:934887 [3] NCCL INFO comm 0x98861d0 rank 11 nranks 32 cudaDev 3 nvmlDev 3 busId 86000 commId 0x1214b375c22bac3c - Init COMPLETE
+ip-26-0-162-14:2525075:2525471 [0] NCCL INFO Using network Libfabric
+ip-26-0-163-58:934724:935113 [4] NCCL INFO Using network Libfabric
+ip-26-0-163-58:934727:935114 [7] NCCL INFO Using network Libfabric
+ip-26-0-163-58:934721:935115 [1] NCCL INFO Using network Libfabric
+ip-26-0-162-14:2525082:2525472 [7] NCCL INFO Using network Libfabric
+ip-26-0-170-31:3197184:3197586 [4] NCCL INFO Using network Libfabric
+ip-26-0-162-14:2525081:2525473 [6] NCCL INFO Using network Libfabric
+ip-26-0-162-14:2525077:2525475 [2] NCCL INFO Using network Libfabric
+ip-26-0-162-14:2525080:2525476 [5] NCCL INFO Using network Libfabric
+ip-26-0-163-58:934723:935117 [3] NCCL INFO Using network Libfabric
+ip-26-0-170-31:3197181:3197587 [1] NCCL INFO Using network Libfabric
+ip-26-0-170-31:3197182:3197588 [2] NCCL INFO Using network Libfabric
+ip-26-0-163-58:934725:935118 [5] NCCL INFO Using network Libfabric
+ip-26-0-170-132:3330187:3330591 [4] NCCL INFO Using network Libfabric
+ip-26-0-162-14:2525078:2525474 [3] NCCL INFO Using network Libfabric
+ip-26-0-170-31:3197180:3197589 [0] NCCL INFO Using network Libfabric
+ip-26-0-162-14:2525079:2525477 [4] NCCL INFO Using network Libfabric
+ip-26-0-170-132:3330186:3330592 [3] NCCL INFO Using network Libfabric
+ip-26-0-170-132:3330185:3330593 [2] NCCL INFO Using network Libfabric
+ip-26-0-170-132:3330190:3330589 [7] NCCL INFO Using network Libfabric
+ip-26-0-170-31:3197187:3197590 [7] NCCL INFO Using network Libfabric
+ip-26-0-163-58:934720:935120 [0] NCCL INFO Using network Libfabric
+ip-26-0-163-58:934722:935116 [2] NCCL INFO Using network Libfabric
+ip-26-0-170-132:3330189:3330594 [6] NCCL INFO Using network Libfabric
+ip-26-0-170-132:3330184:3330596 [1] NCCL INFO Using network Libfabric
+ip-26-0-170-31:3197186:3197593 [6] NCCL INFO Using network Libfabric
+ip-26-0-163-58:934726:935119 [6] NCCL INFO Using network Libfabric
+ip-26-0-170-31:3197183:3197592 [3] NCCL INFO Using network Libfabric
+ip-26-0-170-31:3197185:3197591 [5] NCCL INFO Using network Libfabric
+ip-26-0-170-132:3330188:3330595 [5] NCCL INFO Using network Libfabric
+ip-26-0-170-132:3330183:3330590 [0] NCCL INFO Using network Libfabric
+ip-26-0-162-14:2525076:2525478 [1] NCCL INFO Using network Libfabric
+ip-26-0-163-58:934727:935114 [7] NCCL INFO comm 0x9a1dc80 rank 15 nranks 32 cudaDev 7 nvmlDev 7 busId ca000 commId 0xd3b880bde5f1bc69 - Init START
+ip-26-0-163-58:934726:935119 [6] NCCL INFO comm 0x9e0f0e0 rank 14 nranks 32 cudaDev 6 nvmlDev 6 busId b9000 commId 0xd3b880bde5f1bc69 - Init START
+ip-26-0-163-58:934725:935118 [5] NCCL INFO comm 0x8aa9ac0 rank 13 nranks 32 cudaDev 5 nvmlDev 5 busId a8000 commId 0xd3b880bde5f1bc69 - Init START
+ip-26-0-163-58:934724:935113 [4] NCCL INFO comm 0x98bc760 rank 12 nranks 32 cudaDev 4 nvmlDev 4 busId 97000 commId 0xd3b880bde5f1bc69 - Init START
+ip-26-0-162-14:2525082:2525472 [7] NCCL INFO comm 0x9e76320 rank 7 nranks 32 cudaDev 7 nvmlDev 7 busId ca000 commId 0xd3b880bde5f1bc69 - Init START
+ip-26-0-162-14:2525081:2525473 [6] NCCL INFO comm 0x94f5e60 rank 6 nranks 32 cudaDev 6 nvmlDev 6 busId b9000 commId 0xd3b880bde5f1bc69 - Init START
+ip-26-0-162-14:2525080:2525476 [5] NCCL INFO comm 0x9c491a0 rank 5 nranks 32 cudaDev 5 nvmlDev 5 busId a8000 commId 0xd3b880bde5f1bc69 - Init START
+ip-26-0-162-14:2525079:2525477 [4] NCCL INFO comm 0x96c46e0 rank 4 nranks 32 cudaDev 4 nvmlDev 4 busId 97000 commId 0xd3b880bde5f1bc69 - Init START
+ip-26-0-163-58:934721:935115 [1] NCCL INFO comm 0xa549100 rank 9 nranks 32 cudaDev 1 nvmlDev 1 busId 64000 commId 0xd3b880bde5f1bc69 - Init START
+ip-26-0-163-58:934723:935117 [3] NCCL INFO comm 0x9ac5f80 rank 11 nranks 32 cudaDev 3 nvmlDev 3 busId 86000 commId 0xd3b880bde5f1bc69 - Init START
+ip-26-0-163-58:934722:935116 [2] NCCL INFO comm 0x8ac6a60 rank 10 nranks 32 cudaDev 2 nvmlDev 2 busId 75000 commId 0xd3b880bde5f1bc69 - Init START
+ip-26-0-170-132:3330183:3330590 [0] NCCL INFO comm 0x94d59c0 rank 16 nranks 32 cudaDev 0 nvmlDev 0 busId 53000 commId 0xd3b880bde5f1bc69 - Init START
+ip-26-0-170-31:3197187:3197590 [7] NCCL INFO comm 0xa665180 rank 31 nranks 32 cudaDev 7 nvmlDev 7 busId ca000 commId 0xd3b880bde5f1bc69 - Init START
+ip-26-0-170-31:3197186:3197593 [6] NCCL INFO comm 0xa034d40 rank 30 nranks 32 cudaDev 6 nvmlDev 6 busId b9000 commId 0xd3b880bde5f1bc69 - Init START
+ip-26-0-170-31:3197185:3197591 [5] NCCL INFO comm 0x884f520 rank 29 nranks 32 cudaDev 5 nvmlDev 5 busId a8000 commId 0xd3b880bde5f1bc69 - Init START
+ip-26-0-170-31:3197184:3197586 [4] NCCL INFO comm 0xa1031e0 rank 28 nranks 32 cudaDev 4 nvmlDev 4 busId 97000 commId 0xd3b880bde5f1bc69 - Init START
+ip-26-0-170-31:3197181:3197587 [1] NCCL INFO comm 0x9bc77b0 rank 25 nranks 32 cudaDev 1 nvmlDev 1 busId 64000 commId 0xd3b880bde5f1bc69 - Init START
+ip-26-0-170-31:3197180:3197589 [0] NCCL INFO comm 0xa19e6a0 rank 24 nranks 32 cudaDev 0 nvmlDev 0 busId 53000 commId 0xd3b880bde5f1bc69 - Init START
+ip-26-0-162-14:2525078:2525474 [3] NCCL INFO comm 0x9224d20 rank 3 nranks 32 cudaDev 3 nvmlDev 3 busId 86000 commId 0xd3b880bde5f1bc69 - Init START
+ip-26-0-163-58:934720:935120 [0] NCCL INFO comm 0x98054a0 rank 8 nranks 32 cudaDev 0 nvmlDev 0 busId 53000 commId 0xd3b880bde5f1bc69 - Init START
+ip-26-0-162-14:2525077:2525475 [2] NCCL INFO comm 0xa1b95c0 rank 2 nranks 32 cudaDev 2 nvmlDev 2 busId 75000 commId 0xd3b880bde5f1bc69 - Init START
+ip-26-0-170-132:3330184:3330596 [1] NCCL INFO comm 0x960d840 rank 17 nranks 32 cudaDev 1 nvmlDev 1 busId 64000 commId 0xd3b880bde5f1bc69 - Init START
+ip-26-0-170-31:3197183:3197592 [3] NCCL INFO comm 0x9f6fc60 rank 27 nranks 32 cudaDev 3 nvmlDev 3 busId 86000 commId 0xd3b880bde5f1bc69 - Init START
+ip-26-0-170-31:3197182:3197588 [2] NCCL INFO comm 0x88fe880 rank 26 nranks 32 cudaDev 2 nvmlDev 2 busId 75000 commId 0xd3b880bde5f1bc69 - Init START
+ip-26-0-162-14:2525076:2525478 [1] NCCL INFO comm 0xa370ac0 rank 1 nranks 32 cudaDev 1 nvmlDev 1 busId 64000 commId 0xd3b880bde5f1bc69 - Init START
+ip-26-0-170-132:3330189:3330594 [6] NCCL INFO comm 0x87afc80 rank 22 nranks 32 cudaDev 6 nvmlDev 6 busId b9000 commId 0xd3b880bde5f1bc69 - Init START
+ip-26-0-162-14:2525075:2525471 [0] NCCL INFO comm 0x98759b0 rank 0 nranks 32 cudaDev 0 nvmlDev 0 busId 53000 commId 0xd3b880bde5f1bc69 - Init START
+ip-26-0-163-58:934727:935114 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-170-132:3330190:3330589 [7] NCCL INFO comm 0xa51d2b0 rank 23 nranks 32 cudaDev 7 nvmlDev 7 busId ca000 commId 0xd3b880bde5f1bc69 - Init START
+ip-26-0-170-132:3330188:3330595 [5] NCCL INFO comm 0xa4ee330 rank 21 nranks 32 cudaDev 5 nvmlDev 5 busId a8000 commId 0xd3b880bde5f1bc69 - Init START
+ip-26-0-162-14:2525082:2525472 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-163-58:934726:935119 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-163-58:934725:935118 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-170-132:3330187:3330591 [4] NCCL INFO comm 0x9a3ab00 rank 20 nranks 32 cudaDev 4 nvmlDev 4 busId 97000 commId 0xd3b880bde5f1bc69 - Init START
+ip-26-0-170-31:3197187:3197590 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-162-14:2525081:2525473 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-170-132:3330186:3330592 [3] NCCL INFO comm 0x8f16e30 rank 19 nranks 32 cudaDev 3 nvmlDev 3 busId 86000 commId 0xd3b880bde5f1bc69 - Init START
+ip-26-0-170-132:3330185:3330593 [2] NCCL INFO comm 0x9a9fba0 rank 18 nranks 32 cudaDev 2 nvmlDev 2 busId 75000 commId 0xd3b880bde5f1bc69 - Init START
+ip-26-0-162-14:2525079:2525477 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-162-14:2525075:2525471 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/253
+ip-26-0-162-14:2525077:2525475 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-162-14:2525080:2525476 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-162-14:2525076:2525478 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-162-14:2525078:2525474 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-163-58:934724:935113 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-170-31:3197186:3197593 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-170-31:3197185:3197591 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-163-58:934723:935117 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-170-132:3330184:3330596 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-170-132:3330183:3330590 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-163-58:934721:935115 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-163-58:934722:935116 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-163-58:934720:935120 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-170-132:3330188:3330595 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-170-132:3330190:3330589 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-170-132:3330186:3330592 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-170-132:3330187:3330591 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-170-31:3197184:3197586 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-170-31:3197183:3197592 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-170-31:3197182:3197588 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-170-132:3330189:3330594 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-170-31:3197180:3197589 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-170-31:3197181:3197587 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-170-132:3330185:3330593 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-170-31:3197184:3197586 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-170-31:3197184:3197586 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-170-31:3197180:3197589 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-170-31:3197180:3197589 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-170-31:3197186:3197593 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-170-31:3197186:3197593 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-170-31:3197182:3197588 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-170-31:3197182:3197588 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-170-132:3330185:3330593 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-170-132:3330185:3330593 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-162-14:2525077:2525475 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-162-14:2525077:2525475 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-170-31:3197183:3197592 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-170-31:3197183:3197592 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-163-58:934721:935115 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-163-58:934721:935115 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-162-14:2525082:2525472 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-162-14:2525082:2525472 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-170-31:3197187:3197590 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-170-31:3197187:3197590 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-170-31:3197185:3197591 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-170-31:3197185:3197591 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-170-31:3197181:3197587 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-170-31:3197181:3197587 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-163-58:934727:935114 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-163-58:934727:935114 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-163-58:934725:935118 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-163-58:934725:935118 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-162-14:2525075:2525471 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-162-14:2525075:2525471 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-163-58:934722:935116 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-163-58:934722:935116 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-163-58:934720:935120 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-163-58:934720:935120 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-163-58:934723:935117 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-163-58:934723:935117 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-163-58:934724:935113 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-163-58:934724:935113 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-163-58:934726:935119 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-163-58:934726:935119 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-162-14:2525076:2525478 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-162-14:2525076:2525478 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-162-14:2525079:2525477 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-162-14:2525079:2525477 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-162-14:2525078:2525474 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-162-14:2525078:2525474 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-162-14:2525081:2525473 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-162-14:2525081:2525473 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-162-14:2525080:2525476 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-162-14:2525080:2525476 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-170-132:3330184:3330596 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-170-132:3330184:3330596 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-170-132:3330188:3330595 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-170-132:3330188:3330595 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-170-132:3330187:3330591 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-170-132:3330187:3330591 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-170-132:3330189:3330594 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-170-132:3330189:3330594 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-170-132:3330186:3330592 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-170-132:3330186:3330592 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-170-132:3330183:3330590 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-170-132:3330183:3330590 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-170-132:3330190:3330589 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-170-132:3330190:3330589 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-170-132:3330186:3330592 [3] NCCL INFO Trees [0] 20/-1/-1->19->18 [1] 20/-1/-1->19->18 [2] 20/-1/-1->19->18 [3] 20/11/27->19->3 [4] -1/-1/-1->19->18 [5] 20/-1/-1->19->18 [6] 20/-1/-1->19->18 [7] 20/-1/-1->19->18 [8] 20/-1/-1->19->18 [9] 20/-1/-1->19->18 [10] 20/-1/-1->19->18 [11] 20/-1/-1->19->11 [12] -1/-1/-1->19->18 [13] 20/-1/-1->19->18 [14] 20/-1/-1->19->18 [15] 20/-1/-1->19->18
+ip-26-0-170-132:3330186:3330592 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-170-132:3330187:3330591 [4] NCCL INFO Trees [0] 21/-1/-1->20->19 [1] 21/-1/-1->20->19 [2] 21/-1/-1->20->19 [3] 21/-1/-1->20->19 [4] 21/12/28->20->4 [5] -1/-1/-1->20->19 [6] 21/-1/-1->20->19 [7] 21/-1/-1->20->19 [8] 21/-1/-1->20->19 [9] 21/-1/-1->20->19 [10] 21/-1/-1->20->19 [11] 21/-1/-1->20->19 [12] 21/-1/-1->20->12 [13] -1/-1/-1->20->19 [14] 21/-1/-1->20->19 [15] 21/-1/-1->20->19
+ip-26-0-170-132:3330187:3330591 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-170-132:3330188:3330595 [5] NCCL INFO Trees [0] 22/-1/-1->21->20 [1] 22/-1/-1->21->20 [2] 22/-1/-1->21->20 [3] 22/-1/-1->21->20 [4] 22/-1/-1->21->20 [5] 22/13/29->21->5 [6] -1/-1/-1->21->20 [7] 22/-1/-1->21->20 [8] 22/-1/-1->21->20 [9] 22/-1/-1->21->20 [10] 22/-1/-1->21->20 [11] 22/-1/-1->21->20 [12] 22/-1/-1->21->20 [13] 22/-1/-1->21->13 [14] -1/-1/-1->21->20 [15] 22/-1/-1->21->20
+ip-26-0-170-132:3330188:3330595 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-170-132:3330189:3330594 [6] NCCL INFO Trees [0] 23/-1/-1->22->21 [1] 23/-1/-1->22->21 [2] 23/-1/-1->22->21 [3] 23/-1/-1->22->21 [4] 23/-1/-1->22->21 [5] 23/-1/-1->22->21 [6] 23/14/30->22->6 [7] -1/-1/-1->22->21 [8] 23/-1/-1->22->21 [9] 23/-1/-1->22->21 [10] 23/-1/-1->22->21 [11] 23/-1/-1->22->21 [12] 23/-1/-1->22->21 [13] 23/-1/-1->22->21 [14] 23/-1/-1->22->14 [15] -1/-1/-1->22->21
+ip-26-0-170-132:3330189:3330594 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-170-132:3330190:3330589 [7] NCCL INFO Trees [0] -1/-1/-1->23->22 [1] 16/-1/-1->23->22 [2] 16/-1/-1->23->22 [3] 16/-1/-1->23->22 [4] 16/-1/-1->23->22 [5] 16/-1/-1->23->22 [6] 16/-1/-1->23->22 [7] 16/15/31->23->7 [8] -1/-1/-1->23->22 [9] 16/-1/-1->23->22 [10] 16/-1/-1->23->22 [11] 16/-1/-1->23->22 [12] 16/-1/-1->23->22 [13] 16/-1/-1->23->22 [14] 16/-1/-1->23->22 [15] 16/-1/-1->23->15
+ip-26-0-170-132:3330190:3330589 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-163-58:934727:935114 [7] NCCL INFO Trees [0] -1/-1/-1->15->14 [1] 8/-1/-1->15->14 [2] 8/-1/-1->15->14 [3] 8/-1/-1->15->14 [4] 8/-1/-1->15->14 [5] 8/-1/-1->15->14 [6] 8/-1/-1->15->14 [7] 8/-1/-1->15->23 [8] -1/-1/-1->15->14 [9] 8/-1/-1->15->14 [10] 8/-1/-1->15->14 [11] 8/-1/-1->15->14 [12] 8/-1/-1->15->14 [13] 8/-1/-1->15->14 [14] 8/-1/-1->15->14 [15] 8/23/7->15->31
+ip-26-0-163-58:934727:935114 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-162-14:2525082:2525472 [7] NCCL INFO NVLS Head  0:  0  8 16 24
+ip-26-0-162-14:2525082:2525472 [7] NCCL INFO NVLS Head  1:  1  9 17 25
+ip-26-0-162-14:2525082:2525472 [7] NCCL INFO NVLS Head  2:  2 10 18 26
+ip-26-0-162-14:2525082:2525472 [7] NCCL INFO NVLS Head  3:  3 11 19 27
+ip-26-0-162-14:2525082:2525472 [7] NCCL INFO NVLS Head  4:  4 12 20 28
+ip-26-0-162-14:2525082:2525472 [7] NCCL INFO NVLS Head  5:  5 13 21 29
+ip-26-0-162-14:2525076:2525478 [1] NCCL INFO NVLS Head  0:  0  8 16 24
+ip-26-0-162-14:2525082:2525472 [7] NCCL INFO NVLS Head  6:  6 14 22 30
+ip-26-0-170-31:3197182:3197588 [2] NCCL INFO Trees [0] 27/-1/-1->26->25 [1] 27/-1/-1->26->25 [2] 27/-1/-1->26->18 [3] -1/-1/-1->26->25 [4] 27/-1/-1->26->25 [5] 27/-1/-1->26->25 [6] 27/-1/-1->26->25 [7] 27/-1/-1->26->25 [8] 27/-1/-1->26->25 [9] 27/-1/-1->26->25 [10] 27/10/-1->26->-1 [11] -1/-1/-1->26->25 [12] 27/-1/-1->26->25 [13] 27/-1/-1->26->25 [14] 27/-1/-1->26->25 [15] 27/-1/-1->26->25
+ip-26-0-170-31:3197182:3197588 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-170-31:3197187:3197590 [7] NCCL INFO Trees [0] -1/-1/-1->31->30 [1] 24/-1/-1->31->30 [2] 24/-1/-1->31->30 [3] 24/-1/-1->31->30 [4] 24/-1/-1->31->30 [5] 24/-1/-1->31->30 [6] 24/-1/-1->31->30 [7] 24/-1/-1->31->23 [8] -1/-1/-1->31->30 [9] 24/-1/-1->31->30 [10] 24/-1/-1->31->30 [11] 24/-1/-1->31->30 [12] 24/-1/-1->31->30 [13] 24/-1/-1->31->30 [14] 24/-1/-1->31->30 [15] 24/15/-1->31->-1
+ip-26-0-170-31:3197187:3197590 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-170-132:3330183:3330590 [0] NCCL INFO Trees [0] 17/8/24->16->0 [1] -1/-1/-1->16->23 [2] 17/-1/-1->16->23 [3] 17/-1/-1->16->23 [4] 17/-1/-1->16->23 [5] 17/-1/-1->16->23 [6] 17/-1/-1->16->23 [7] 17/-1/-1->16->23 [8] 17/-1/-1->16->8 [9] -1/-1/-1->16->23 [10] 17/-1/-1->16->23 [11] 17/-1/-1->16->23 [12] 17/-1/-1->16->23 [13] 17/-1/-1->16->23 [14] 17/-1/-1->16->23 [15] 17/-1/-1->16->23
+ip-26-0-170-132:3330183:3330590 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-170-132:3330184:3330596 [1] NCCL INFO Trees [0] 18/-1/-1->17->16 [1] 18/9/25->17->1 [2] -1/-1/-1->17->16 [3] 18/-1/-1->17->16 [4] 18/-1/-1->17->16 [5] 18/-1/-1->17->16 [6] 18/-1/-1->17->16 [7] 18/-1/-1->17->16 [8] 18/-1/-1->17->16 [9] 18/-1/-1->17->9 [10] -1/-1/-1->17->16 [11] 18/-1/-1->17->16 [12] 18/-1/-1->17->16 [13] 18/-1/-1->17->16 [14] 18/-1/-1->17->16 [15] 18/-1/-1->17->16
+ip-26-0-163-58:934725:935118 [5] NCCL INFO Trees [0] 14/-1/-1->13->12 [1] 14/-1/-1->13->12 [2] 14/-1/-1->13->12 [3] 14/-1/-1->13->12 [4] 14/-1/-1->13->12 [5] 14/-1/-1->13->21 [6] -1/-1/-1->13->12 [7] 14/-1/-1->13->12 [8] 14/-1/-1->13->12 [9] 14/-1/-1->13->12 [10] 14/-1/-1->13->12 [11] 14/-1/-1->13->12 [12] 14/-1/-1->13->12 [13] 14/21/5->13->29 [14] -1/-1/-1->13->12 [15] 14/-1/-1->13->12
+ip-26-0-163-58:934725:935118 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-163-58:934726:935119 [6] NCCL INFO Trees [0] 15/-1/-1->14->13 [1] 15/-1/-1->14->13 [2] 15/-1/-1->14->13 [3] 15/-1/-1->14->13 [4] 15/-1/-1->14->13 [5] 15/-1/-1->14->13 [6] 15/-1/-1->14->22 [7] -1/-1/-1->14->13 [8] 15/-1/-1->14->13 [9] 15/-1/-1->14->13 [10] 15/-1/-1->14->13 [11] 15/-1/-1->14->13 [12] 15/-1/-1->14->13 [13] 15/-1/-1->14->13 [14] 15/22/6->14->30 [15] -1/-1/-1->14->13
+ip-26-0-163-58:934726:935119 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-170-132:3330185:3330593 [2] NCCL INFO Trees [0] 19/-1/-1->18->17 [1] 19/-1/-1->18->17 [2] 19/10/26->18->2 [3] -1/-1/-1->18->17 [4] 19/-1/-1->18->17 [5] 19/-1/-1->18->17 [6] 19/-1/-1->18->17 [7] 19/-1/-1->18->17 [8] 19/-1/-1->18->17 [9] 19/-1/-1->18->17 [10] 19/-1/-1->18->10 [11] -1/-1/-1->18->17 [12] 19/-1/-1->18->17 [13] 19/-1/-1->18->17 [14] 19/-1/-1->18->17 [15] 19/-1/-1->18->17
+ip-26-0-170-132:3330184:3330596 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-170-31:3197180:3197589 [0] NCCL INFO Trees [0] 25/-1/-1->24->16 [1] -1/-1/-1->24->31 [2] 25/-1/-1->24->31 [3] 25/-1/-1->24->31 [4] 25/-1/-1->24->31 [5] 25/-1/-1->24->31 [6] 25/-1/-1->24->31 [7] 25/-1/-1->24->31 [8] 25/8/-1->24->-1 [9] -1/-1/-1->24->31 [10] 25/-1/-1->24->31 [11] 25/-1/-1->24->31 [12] 25/-1/-1->24->31 [13] 25/-1/-1->24->31 [14] 25/-1/-1->24->31 [15] 25/-1/-1->24->31
+ip-26-0-162-14:2525082:2525472 [7] NCCL INFO NVLS Head  7:  7 15 23 31
+ip-26-0-162-14:2525081:2525473 [6] NCCL INFO NVLS Head  0:  0  8 16 24
+ip-26-0-162-14:2525076:2525478 [1] NCCL INFO NVLS Head  1:  1  9 17 25
+ip-26-0-162-14:2525077:2525475 [2] NCCL INFO NVLS Head  0:  0  8 16 24
+ip-26-0-162-14:2525076:2525478 [1] NCCL INFO NVLS Head  2:  2 10 18 26
+ip-26-0-162-14:2525081:2525473 [6] NCCL INFO NVLS Head  1:  1  9 17 25
+ip-26-0-162-14:2525077:2525475 [2] NCCL INFO NVLS Head  1:  1  9 17 25
+ip-26-0-162-14:2525078:2525474 [3] NCCL INFO NVLS Head  0:  0  8 16 24
+ip-26-0-162-14:2525081:2525473 [6] NCCL INFO NVLS Head  2:  2 10 18 26
+ip-26-0-162-14:2525080:2525476 [5] NCCL INFO NVLS Head  0:  0  8 16 24
+ip-26-0-162-14:2525078:2525474 [3] NCCL INFO NVLS Head  1:  1  9 17 25
+ip-26-0-162-14:2525076:2525478 [1] NCCL INFO NVLS Head  3:  3 11 19 27
+ip-26-0-162-14:2525077:2525475 [2] NCCL INFO NVLS Head  2:  2 10 18 26
+ip-26-0-162-14:2525081:2525473 [6] NCCL INFO NVLS Head  3:  3 11 19 27
+ip-26-0-163-58:934722:935116 [2] NCCL INFO Trees [0] 11/-1/-1->10->9 [1] 11/-1/-1->10->9 [2] 11/-1/-1->10->18 [3] -1/-1/-1->10->9 [4] 11/-1/-1->10->9 [5] 11/-1/-1->10->9 [6] 11/-1/-1->10->9 [7] 11/-1/-1->10->9 [8] 11/-1/-1->10->9 [9] 11/-1/-1->10->9 [10] 11/18/2->10->26 [11] -1/-1/-1->10->9 [12] 11/-1/-1->10->9 [13] 11/-1/-1->10->9 [14] 11/-1/-1->10->9 [15] 11/-1/-1->10->9
+ip-26-0-163-58:934722:935116 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-170-132:3330185:3330593 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-170-31:3197184:3197586 [4] NCCL INFO Trees [0] 29/-1/-1->28->27 [1] 29/-1/-1->28->27 [2] 29/-1/-1->28->27 [3] 29/-1/-1->28->27 [4] 29/-1/-1->28->20 [5] -1/-1/-1->28->27 [6] 29/-1/-1->28->27 [7] 29/-1/-1->28->27 [8] 29/-1/-1->28->27 [9] 29/-1/-1->28->27 [10] 29/-1/-1->28->27 [11] 29/-1/-1->28->27 [12] 29/12/-1->28->-1 [13] -1/-1/-1->28->27 [14] 29/-1/-1->28->27 [15] 29/-1/-1->28->27
+ip-26-0-170-31:3197180:3197589 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-170-31:3197183:3197592 [3] NCCL INFO Trees [0] 28/-1/-1->27->26 [1] 28/-1/-1->27->26 [2] 28/-1/-1->27->26 [3] 28/-1/-1->27->19 [4] -1/-1/-1->27->26 [5] 28/-1/-1->27->26 [6] 28/-1/-1->27->26 [7] 28/-1/-1->27->26 [8] 28/-1/-1->27->26 [9] 28/-1/-1->27->26 [10] 28/-1/-1->27->26 [11] 28/11/-1->27->-1 [12] -1/-1/-1->27->26 [13] 28/-1/-1->27->26 [14] 28/-1/-1->27->26 [15] 28/-1/-1->27->26
+ip-26-0-170-31:3197184:3197586 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-170-31:3197183:3197592 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-162-14:2525080:2525476 [5] NCCL INFO NVLS Head  1:  1  9 17 25
+ip-26-0-162-14:2525078:2525474 [3] NCCL INFO NVLS Head  2:  2 10 18 26
+ip-26-0-162-14:2525076:2525478 [1] NCCL INFO NVLS Head  4:  4 12 20 28
+ip-26-0-162-14:2525080:2525476 [5] NCCL INFO NVLS Head  2:  2 10 18 26
+ip-26-0-162-14:2525078:2525474 [3] NCCL INFO NVLS Head  3:  3 11 19 27
+ip-26-0-162-14:2525082:2525472 [7] NCCL INFO Trees [0] -1/-1/-1->7->6 [1] 0/-1/-1->7->6 [2] 0/-1/-1->7->6 [3] 0/-1/-1->7->6 [4] 0/-1/-1->7->6 [5] 0/-1/-1->7->6 [6] 0/-1/-1->7->6 [7] 0/23/-1->7->-1 [8] -1/-1/-1->7->6 [9] 0/-1/-1->7->6 [10] 0/-1/-1->7->6 [11] 0/-1/-1->7->6 [12] 0/-1/-1->7->6 [13] 0/-1/-1->7->6 [14] 0/-1/-1->7->6 [15] 0/-1/-1->7->15
+ip-26-0-162-14:2525080:2525476 [5] NCCL INFO NVLS Head  3:  3 11 19 27
+ip-26-0-162-14:2525078:2525474 [3] NCCL INFO NVLS Head  4:  4 12 20 28
+ip-26-0-162-14:2525077:2525475 [2] NCCL INFO NVLS Head  3:  3 11 19 27
+ip-26-0-162-14:2525078:2525474 [3] NCCL INFO NVLS Head  5:  5 13 21 29
+ip-26-0-163-58:934723:935117 [3] NCCL INFO Trees [0] 12/-1/-1->11->10 [1] 12/-1/-1->11->10 [2] 12/-1/-1->11->10 [3] 12/-1/-1->11->19 [4] -1/-1/-1->11->10 [5] 12/-1/-1->11->10 [6] 12/-1/-1->11->10 [7] 12/-1/-1->11->10 [8] 12/-1/-1->11->10 [9] 12/-1/-1->11->10 [10] 12/-1/-1->11->10 [11] 12/19/3->11->27 [12] -1/-1/-1->11->10 [13] 12/-1/-1->11->10 [14] 12/-1/-1->11->10 [15] 12/-1/-1->11->10
+ip-26-0-170-31:3197181:3197587 [1] NCCL INFO Trees [0] 26/-1/-1->25->24 [1] 26/-1/-1->25->17 [2] -1/-1/-1->25->24 [3] 26/-1/-1->25->24 [4] 26/-1/-1->25->24 [5] 26/-1/-1->25->24 [6] 26/-1/-1->25->24 [7] 26/-1/-1->25->24 [8] 26/-1/-1->25->24 [9] 26/9/-1->25->-1 [10] -1/-1/-1->25->24 [11] 26/-1/-1->25->24 [12] 26/-1/-1->25->24 [13] 26/-1/-1->25->24 [14] 26/-1/-1->25->24 [15] 26/-1/-1->25->24
+ip-26-0-170-31:3197186:3197593 [6] NCCL INFO Trees [0] 31/-1/-1->30->29 [1] 31/-1/-1->30->29 [2] 31/-1/-1->30->29 [3] 31/-1/-1->30->29 [4] 31/-1/-1->30->29 [5] 31/-1/-1->30->29 [6] 31/-1/-1->30->22 [7] -1/-1/-1->30->29 [8] 31/-1/-1->30->29 [9] 31/-1/-1->30->29 [10] 31/-1/-1->30->29 [11] 31/-1/-1->30->29 [12] 31/-1/-1->30->29 [13] 31/-1/-1->30->29 [14] 31/14/-1->30->-1 [15] -1/-1/-1->30->29
+ip-26-0-170-31:3197181:3197587 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-170-31:3197186:3197593 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-162-14:2525081:2525473 [6] NCCL INFO NVLS Head  4:  4 12 20 28
+ip-26-0-162-14:2525076:2525478 [1] NCCL INFO NVLS Head  5:  5 13 21 29
+ip-26-0-162-14:2525079:2525477 [4] NCCL INFO NVLS Head  0:  0  8 16 24
+ip-26-0-162-14:2525082:2525472 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-162-14:2525080:2525476 [5] NCCL INFO NVLS Head  4:  4 12 20 28
+ip-26-0-163-58:934723:935117 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-163-58:934724:935113 [4] NCCL INFO Trees [0] 13/-1/-1->12->11 [1] 13/-1/-1->12->11 [2] 13/-1/-1->12->11 [3] 13/-1/-1->12->11 [4] 13/-1/-1->12->20 [5] -1/-1/-1->12->11 [6] 13/-1/-1->12->11 [7] 13/-1/-1->12->11 [8] 13/-1/-1->12->11 [9] 13/-1/-1->12->11 [10] 13/-1/-1->12->11 [11] 13/-1/-1->12->11 [12] 13/20/4->12->28 [13] -1/-1/-1->12->11 [14] 13/-1/-1->12->11 [15] 13/-1/-1->12->11
+ip-26-0-163-58:934724:935113 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-170-31:3197185:3197591 [5] NCCL INFO Trees [0] 30/-1/-1->29->28 [1] 30/-1/-1->29->28 [2] 30/-1/-1->29->28 [3] 30/-1/-1->29->28 [4] 30/-1/-1->29->28 [5] 30/-1/-1->29->21 [6] -1/-1/-1->29->28 [7] 30/-1/-1->29->28 [8] 30/-1/-1->29->28 [9] 30/-1/-1->29->28 [10] 30/-1/-1->29->28 [11] 30/-1/-1->29->28 [12] 30/-1/-1->29->28 [13] 30/13/-1->29->-1 [14] -1/-1/-1->29->28 [15] 30/-1/-1->29->28
+ip-26-0-170-31:3197185:3197591 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-162-14:2525080:2525476 [5] NCCL INFO NVLS Head  5:  5 13 21 29
+ip-26-0-162-14:2525077:2525475 [2] NCCL INFO NVLS Head  4:  4 12 20 28
+ip-26-0-162-14:2525078:2525474 [3] NCCL INFO NVLS Head  6:  6 14 22 30
+ip-26-0-162-14:2525081:2525473 [6] NCCL INFO NVLS Head  5:  5 13 21 29
+ip-26-0-162-14:2525076:2525478 [1] NCCL INFO NVLS Head  6:  6 14 22 30
+ip-26-0-162-14:2525079:2525477 [4] NCCL INFO NVLS Head  1:  1  9 17 25
+ip-26-0-162-14:2525080:2525476 [5] NCCL INFO NVLS Head  6:  6 14 22 30
+ip-26-0-162-14:2525076:2525478 [1] NCCL INFO NVLS Head  7:  7 15 23 31
+ip-26-0-162-14:2525079:2525477 [4] NCCL INFO NVLS Head  2:  2 10 18 26
+ip-26-0-162-14:2525077:2525475 [2] NCCL INFO NVLS Head  5:  5 13 21 29
+ip-26-0-162-14:2525078:2525474 [3] NCCL INFO NVLS Head  7:  7 15 23 31
+ip-26-0-162-14:2525081:2525473 [6] NCCL INFO NVLS Head  6:  6 14 22 30
+ip-26-0-162-14:2525080:2525476 [5] NCCL INFO NVLS Head  7:  7 15 23 31
+ip-26-0-162-14:2525079:2525477 [4] NCCL INFO NVLS Head  3:  3 11 19 27
+ip-26-0-163-58:934720:935120 [0] NCCL INFO Trees [0] 9/-1/-1->8->16 [1] -1/-1/-1->8->15 [2] 9/-1/-1->8->15 [3] 9/-1/-1->8->15 [4] 9/-1/-1->8->15 [5] 9/-1/-1->8->15 [6] 9/-1/-1->8->15 [7] 9/-1/-1->8->15 [8] 9/16/0->8->24 [9] -1/-1/-1->8->15 [10] 9/-1/-1->8->15 [11] 9/-1/-1->8->15 [12] 9/-1/-1->8->15 [13] 9/-1/-1->8->15 [14] 9/-1/-1->8->15 [15] 9/-1/-1->8->15
+ip-26-0-162-14:2525081:2525473 [6] NCCL INFO NVLS Head  7:  7 15 23 31
+ip-26-0-162-14:2525077:2525475 [2] NCCL INFO NVLS Head  6:  6 14 22 30
+ip-26-0-162-14:2525079:2525477 [4] NCCL INFO NVLS Head  4:  4 12 20 28
+ip-26-0-162-14:2525077:2525475 [2] NCCL INFO NVLS Head  7:  7 15 23 31
+ip-26-0-162-14:2525079:2525477 [4] NCCL INFO NVLS Head  5:  5 13 21 29
+ip-26-0-162-14:2525075:2525471 [0] NCCL INFO NVLS Head  0:  0  8 16 24
+ip-26-0-162-14:2525079:2525477 [4] NCCL INFO NVLS Head  6:  6 14 22 30
+ip-26-0-162-14:2525079:2525477 [4] NCCL INFO NVLS Head  7:  7 15 23 31
+ip-26-0-162-14:2525075:2525471 [0] NCCL INFO NVLS Head  1:  1  9 17 25
+ip-26-0-162-14:2525075:2525471 [0] NCCL INFO NVLS Head  2:  2 10 18 26
+ip-26-0-162-14:2525075:2525471 [0] NCCL INFO NVLS Head  3:  3 11 19 27
+ip-26-0-163-58:934720:935120 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-163-58:934721:935115 [1] NCCL INFO Trees [0] 10/-1/-1->9->8 [1] 10/-1/-1->9->17 [2] -1/-1/-1->9->8 [3] 10/-1/-1->9->8 [4] 10/-1/-1->9->8 [5] 10/-1/-1->9->8 [6] 10/-1/-1->9->8 [7] 10/-1/-1->9->8 [8] 10/-1/-1->9->8 [9] 10/17/1->9->25 [10] -1/-1/-1->9->8 [11] 10/-1/-1->9->8 [12] 10/-1/-1->9->8 [13] 10/-1/-1->9->8 [14] 10/-1/-1->9->8 [15] 10/-1/-1->9->8
+ip-26-0-163-58:934721:935115 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-162-14:2525076:2525478 [1] NCCL INFO Trees [0] 2/-1/-1->1->0 [1] 2/17/-1->1->-1 [2] -1/-1/-1->1->0 [3] 2/-1/-1->1->0 [4] 2/-1/-1->1->0 [5] 2/-1/-1->1->0 [6] 2/-1/-1->1->0 [7] 2/-1/-1->1->0 [8] 2/-1/-1->1->0 [9] 2/-1/-1->1->9 [10] -1/-1/-1->1->0 [11] 2/-1/-1->1->0 [12] 2/-1/-1->1->0 [13] 2/-1/-1->1->0 [14] 2/-1/-1->1->0 [15] 2/-1/-1->1->0
+ip-26-0-162-14:2525078:2525474 [3] NCCL INFO Trees [0] 4/-1/-1->3->2 [1] 4/-1/-1->3->2 [2] 4/-1/-1->3->2 [3] 4/19/-1->3->-1 [4] -1/-1/-1->3->2 [5] 4/-1/-1->3->2 [6] 4/-1/-1->3->2 [7] 4/-1/-1->3->2 [8] 4/-1/-1->3->2 [9] 4/-1/-1->3->2 [10] 4/-1/-1->3->2 [11] 4/-1/-1->3->11 [12] -1/-1/-1->3->2 [13] 4/-1/-1->3->2 [14] 4/-1/-1->3->2 [15] 4/-1/-1->3->2
+ip-26-0-162-14:2525075:2525471 [0] NCCL INFO NVLS Head  4:  4 12 20 28
+ip-26-0-162-14:2525076:2525478 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-162-14:2525078:2525474 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-162-14:2525080:2525476 [5] NCCL INFO Trees [0] 6/-1/-1->5->4 [1] 6/-1/-1->5->4 [2] 6/-1/-1->5->4 [3] 6/-1/-1->5->4 [4] 6/-1/-1->5->4 [5] 6/21/-1->5->-1 [6] -1/-1/-1->5->4 [7] 6/-1/-1->5->4 [8] 6/-1/-1->5->4 [9] 6/-1/-1->5->4 [10] 6/-1/-1->5->4 [11] 6/-1/-1->5->4 [12] 6/-1/-1->5->4 [13] 6/-1/-1->5->13 [14] -1/-1/-1->5->4 [15] 6/-1/-1->5->4
+ip-26-0-162-14:2525075:2525471 [0] NCCL INFO NVLS Head  5:  5 13 21 29
+ip-26-0-162-14:2525081:2525473 [6] NCCL INFO Trees [0] 7/-1/-1->6->5 [1] 7/-1/-1->6->5 [2] 7/-1/-1->6->5 [3] 7/-1/-1->6->5 [4] 7/-1/-1->6->5 [5] 7/-1/-1->6->5 [6] 7/22/-1->6->-1 [7] -1/-1/-1->6->5 [8] 7/-1/-1->6->5 [9] 7/-1/-1->6->5 [10] 7/-1/-1->6->5 [11] 7/-1/-1->6->5 [12] 7/-1/-1->6->5 [13] 7/-1/-1->6->5 [14] 7/-1/-1->6->14 [15] -1/-1/-1->6->5
+ip-26-0-162-14:2525080:2525476 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-162-14:2525075:2525471 [0] NCCL INFO NVLS Head  6:  6 14 22 30
+ip-26-0-162-14:2525077:2525475 [2] NCCL INFO Trees [0] 3/-1/-1->2->1 [1] 3/-1/-1->2->1 [2] 3/18/-1->2->-1 [3] -1/-1/-1->2->1 [4] 3/-1/-1->2->1 [5] 3/-1/-1->2->1 [6] 3/-1/-1->2->1 [7] 3/-1/-1->2->1 [8] 3/-1/-1->2->1 [9] 3/-1/-1->2->1 [10] 3/-1/-1->2->10 [11] -1/-1/-1->2->1 [12] 3/-1/-1->2->1 [13] 3/-1/-1->2->1 [14] 3/-1/-1->2->1 [15] 3/-1/-1->2->1
+ip-26-0-162-14:2525081:2525473 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-162-14:2525075:2525471 [0] NCCL INFO NVLS Head  7:  7 15 23 31
+ip-26-0-162-14:2525077:2525475 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-162-14:2525079:2525477 [4] NCCL INFO Trees [0] 5/-1/-1->4->3 [1] 5/-1/-1->4->3 [2] 5/-1/-1->4->3 [3] 5/-1/-1->4->3 [4] 5/20/-1->4->-1 [5] -1/-1/-1->4->3 [6] 5/-1/-1->4->3 [7] 5/-1/-1->4->3 [8] 5/-1/-1->4->3 [9] 5/-1/-1->4->3 [10] 5/-1/-1->4->3 [11] 5/-1/-1->4->3 [12] 5/-1/-1->4->12 [13] -1/-1/-1->4->3 [14] 5/-1/-1->4->3 [15] 5/-1/-1->4->3
+ip-26-0-162-14:2525079:2525477 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-162-14:2525075:2525471 [0] NCCL INFO Channel 00/16 :    0   1   2   3   4   5   6   7   8   9  10  11  12  13  14  15  16  17  18  19
+ip-26-0-162-14:2525075:2525471 [0] NCCL INFO Channel 01/16 :    0   7   6   5   4   3   2   9   8  15  14  13  12  11  10  17  16  23  22  21
+ip-26-0-162-14:2525075:2525471 [0] NCCL INFO Channel 02/16 :    0   7   6   5   4   3  10   9   8  15  14  13  12  11  18  17  16  23  22  21
+ip-26-0-162-14:2525075:2525471 [0] NCCL INFO Channel 03/16 :    0   7   6   5   4  11  10   9   8  15  14  13  12  19  18  17  16  23  22  21
+ip-26-0-162-14:2525075:2525471 [0] NCCL INFO Channel 04/16 :    0   7   6   5  12  11  10   9   8  15  14  13  20  19  18  17  16  23  22  21
+ip-26-0-162-14:2525075:2525471 [0] NCCL INFO Channel 05/16 :    0   4   7   6  13  11  10   9   8  12  15  14  21  19  18  17  16  20  23  22
+ip-26-0-162-14:2525075:2525471 [0] NCCL INFO Channel 06/16 :    0   5   4   7  14  11  10   9   8  13  12  15  22  19  18  17  16  21  20  23
+ip-26-0-162-14:2525075:2525471 [0] NCCL INFO Channel 07/16 :    0   6   5   4  15  11  10   9   8  14  13  12  23  19  18  17  16  22  21  20
+ip-26-0-162-14:2525075:2525471 [0] NCCL INFO Channel 08/16 :    0   1   2   3   4   5   6   7   8   9  10  11  12  13  14  15  16  17  18  19
+ip-26-0-162-14:2525075:2525471 [0] NCCL INFO Channel 09/16 :    0   7   6   5   4   3   2   9   8  15  14  13  12  11  10  17  16  23  22  21
+ip-26-0-162-14:2525075:2525471 [0] NCCL INFO Channel 10/16 :    0   7   6   5   4   3  10   9   8  15  14  13  12  11  18  17  16  23  22  21
+ip-26-0-162-14:2525075:2525471 [0] NCCL INFO Channel 11/16 :    0   7   6   5   4  11  10   9   8  15  14  13  12  19  18  17  16  23  22  21
+ip-26-0-162-14:2525075:2525471 [0] NCCL INFO Channel 12/16 :    0   7   6   5  12  11  10   9   8  15  14  13  20  19  18  17  16  23  22  21
+ip-26-0-162-14:2525075:2525471 [0] NCCL INFO Channel 13/16 :    0   4   7   6  13  11  10   9   8  12  15  14  21  19  18  17  16  20  23  22
+ip-26-0-162-14:2525075:2525471 [0] NCCL INFO Channel 14/16 :    0   5   4   7  14  11  10   9   8  13  12  15  22  19  18  17  16  21  20  23
+ip-26-0-162-14:2525075:2525471 [0] NCCL INFO Channel 15/16 :    0   6   5   4  15  11  10   9   8  14  13  12  23  19  18  17  16  22  21  20
+ip-26-0-162-14:2525075:2525471 [0] NCCL INFO Trees [0] 1/16/-1->0->-1 [1] -1/-1/-1->0->7 [2] 1/-1/-1->0->7 [3] 1/-1/-1->0->7 [4] 1/-1/-1->0->7 [5] 1/-1/-1->0->7 [6] 1/-1/-1->0->7 [7] 1/-1/-1->0->7 [8] 1/-1/-1->0->8 [9] -1/-1/-1->0->7 [10] 1/-1/-1->0->7 [11] 1/-1/-1->0->7 [12] 1/-1/-1->0->7 [13] 1/-1/-1->0->7 [14] 1/-1/-1->0->7 [15] 1/-1/-1->0->7
+ip-26-0-162-14:2525075:2525471 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-162-14:2525078:2525474 [3] NCCL INFO Channel 00/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-162-14:2525081:2525473 [6] NCCL INFO Channel 00/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-170-132:3330186:3330592 [3] NCCL INFO Channel 00/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-162-14:2525080:2525476 [5] NCCL INFO Channel 00/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-170-132:3330187:3330591 [4] NCCL INFO Channel 00/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-163-58:934722:935116 [2] NCCL INFO Channel 00/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-162-14:2525078:2525474 [3] NCCL INFO Channel 08/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-162-14:2525081:2525473 [6] NCCL INFO Channel 08/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-163-58:934726:935119 [6] NCCL INFO Channel 00/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-170-31:3197185:3197591 [5] NCCL INFO Channel 00/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-170-31:3197184:3197586 [4] NCCL INFO Channel 00/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-170-132:3330185:3330593 [2] NCCL INFO Channel 00/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-170-31:3197186:3197593 [6] NCCL INFO Channel 00/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-170-132:3330189:3330594 [6] NCCL INFO Channel 00/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-170-132:3330188:3330595 [5] NCCL INFO Channel 00/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-170-132:3330184:3330596 [1] NCCL INFO Channel 00/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-162-14:2525080:2525476 [5] NCCL INFO Channel 08/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-163-58:934725:935118 [5] NCCL INFO Channel 00/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-170-132:3330186:3330592 [3] NCCL INFO Channel 08/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-163-58:934723:935117 [3] NCCL INFO Channel 00/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-170-31:3197183:3197592 [3] NCCL INFO Channel 00/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-170-31:3197182:3197588 [2] NCCL INFO Channel 00/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-163-58:934724:935113 [4] NCCL INFO Channel 00/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-170-31:3197181:3197587 [1] NCCL INFO Channel 00/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-163-58:934721:935115 [1] NCCL INFO Channel 00/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-170-132:3330187:3330591 [4] NCCL INFO Channel 08/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-163-58:934722:935116 [2] NCCL INFO Channel 08/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-163-58:934726:935119 [6] NCCL INFO Channel 08/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-170-31:3197185:3197591 [5] NCCL INFO Channel 08/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-170-31:3197184:3197586 [4] NCCL INFO Channel 08/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-170-132:3330185:3330593 [2] NCCL INFO Channel 08/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-170-31:3197186:3197593 [6] NCCL INFO Channel 08/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-170-132:3330189:3330594 [6] NCCL INFO Channel 08/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-170-132:3330188:3330595 [5] NCCL INFO Channel 08/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-163-58:934725:935118 [5] NCCL INFO Channel 08/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-170-132:3330184:3330596 [1] NCCL INFO Channel 08/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-170-31:3197183:3197592 [3] NCCL INFO Channel 08/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-163-58:934723:935117 [3] NCCL INFO Channel 08/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-170-31:3197182:3197588 [2] NCCL INFO Channel 08/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-163-58:934724:935113 [4] NCCL INFO Channel 08/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-170-31:3197181:3197587 [1] NCCL INFO Channel 08/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-163-58:934721:935115 [1] NCCL INFO Channel 08/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-162-14:2525076:2525478 [1] NCCL INFO Channel 00/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-162-14:2525079:2525477 [4] NCCL INFO Channel 00/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-170-132:3330187:3330591 [4] NCCL INFO Channel 05/0 : 20[4] -> 23[7] via P2P/IPC
+ip-26-0-162-14:2525076:2525478 [1] NCCL INFO Channel 08/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-162-14:2525077:2525475 [2] NCCL INFO Channel 00/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-162-14:2525079:2525477 [4] NCCL INFO Channel 08/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-170-31:3197184:3197586 [4] NCCL INFO Channel 05/0 : 28[4] -> 31[7] via P2P/IPC
+ip-26-0-163-58:934724:935113 [4] NCCL INFO Channel 05/0 : 12[4] -> 15[7] via P2P/IPC
+ip-26-0-170-132:3330187:3330591 [4] NCCL INFO Channel 06/0 : 20[4] -> 23[7] via P2P/IPC
+ip-26-0-162-14:2525077:2525475 [2] NCCL INFO Channel 08/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-170-31:3197184:3197586 [4] NCCL INFO Channel 06/0 : 28[4] -> 31[7] via P2P/IPC
+ip-26-0-163-58:934724:935113 [4] NCCL INFO Channel 06/0 : 12[4] -> 15[7] via P2P/IPC
+ip-26-0-170-132:3330187:3330591 [4] NCCL INFO Channel 13/0 : 20[4] -> 23[7] via P2P/IPC
+ip-26-0-162-14:2525079:2525477 [4] NCCL INFO Channel 05/0 : 4[4] -> 7[7] via P2P/IPC
+ip-26-0-170-31:3197184:3197586 [4] NCCL INFO Channel 13/0 : 28[4] -> 31[7] via P2P/IPC
+ip-26-0-170-132:3330187:3330591 [4] NCCL INFO Channel 14/0 : 20[4] -> 23[7] via P2P/IPC
+ip-26-0-163-58:934724:935113 [4] NCCL INFO Channel 13/0 : 12[4] -> 15[7] via P2P/IPC
+ip-26-0-170-31:3197184:3197586 [4] NCCL INFO Channel 14/0 : 28[4] -> 31[7] via P2P/IPC
+ip-26-0-162-14:2525079:2525477 [4] NCCL INFO Channel 06/0 : 4[4] -> 7[7] via P2P/IPC
+ip-26-0-163-58:934724:935113 [4] NCCL INFO Channel 14/0 : 12[4] -> 15[7] via P2P/IPC
+ip-26-0-162-14:2525079:2525477 [4] NCCL INFO Channel 13/0 : 4[4] -> 7[7] via P2P/IPC
+ip-26-0-162-14:2525079:2525477 [4] NCCL INFO Channel 14/0 : 4[4] -> 7[7] via P2P/IPC
+ip-26-0-170-132:3330183:3330590 [0] NCCL INFO Channel 00/0 : 15[7] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330183:3330590 [0] NCCL INFO Channel 08/0 : 15[7] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330183:3330590 [0] NCCL INFO Channel 00/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-170-132:3330190:3330589 [7] NCCL INFO Channel 00/0 : 23[7] -> 24[0] [send] via NET/Libfabric/0(16)/GDRDMA
+ip-26-0-163-58:934720:935120 [0] NCCL INFO Channel 00/0 : 7[7] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934720:935120 [0] NCCL INFO Channel 08/0 : 7[7] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934720:935120 [0] NCCL INFO Channel 00/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-163-58:934727:935114 [7] NCCL INFO Channel 00/0 : 15[7] -> 16[0] [send] via NET/Libfabric/0(8)/GDRDMA
+ip-26-0-170-132:3330183:3330590 [0] NCCL INFO Channel 08/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-170-132:3330190:3330589 [7] NCCL INFO Channel 08/0 : 23[7] -> 24[0] [send] via NET/Libfabric/0(16)/GDRDMA
+ip-26-0-163-58:934720:935120 [0] NCCL INFO Channel 08/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-163-58:934727:935114 [7] NCCL INFO Channel 08/0 : 15[7] -> 16[0] [send] via NET/Libfabric/0(8)/GDRDMA
+ip-26-0-170-132:3330183:3330590 [0] NCCL INFO Channel 05/0 : 16[0] -> 20[4] via P2P/IPC
+ip-26-0-170-132:3330186:3330592 [3] NCCL INFO Channel 03/0 : 12[4] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330186:3330592 [3] NCCL INFO Channel 11/0 : 12[4] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-31:3197180:3197589 [0] NCCL INFO Channel 00/0 : 23[7] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-31:3197180:3197589 [0] NCCL INFO Channel 08/0 : 23[7] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-31:3197180:3197589 [0] NCCL INFO Channel 00/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-170-31:3197187:3197590 [7] NCCL INFO Channel 00/0 : 31[7] -> 0[0] [send] via NET/Libfabric/0(24)/GDRDMA
+ip-26-0-170-31:3197180:3197589 [0] NCCL INFO Channel 08/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-170-31:3197187:3197590 [7] NCCL INFO Channel 08/0 : 31[7] -> 0[0] [send] via NET/Libfabric/0(24)/GDRDMA
+ip-26-0-170-132:3330185:3330593 [2] NCCL INFO Channel 02/0 : 11[3] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330185:3330593 [2] NCCL INFO Channel 10/0 : 11[3] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330186:3330592 [3] NCCL INFO Channel 02/0 : 19[3] -> 26[2] [send] via NET/Libfabric/2(18)/GDRDMA
+ip-26-0-170-132:3330186:3330592 [3] NCCL INFO Channel 10/0 : 19[3] -> 26[2] [send] via NET/Libfabric/2(18)/GDRDMA
+ip-26-0-170-31:3197180:3197589 [0] NCCL INFO Channel 05/0 : 24[0] -> 28[4] via P2P/IPC
+ip-26-0-162-14:2525075:2525471 [0] NCCL INFO Channel 00/0 : 31[7] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-31:3197180:3197589 [0] NCCL INFO Channel 13/0 : 24[0] -> 28[4] via P2P/IPC
+ip-26-0-162-14:2525075:2525471 [0] NCCL INFO Channel 08/0 : 31[7] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-14:2525075:2525471 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-162-14:2525082:2525472 [7] NCCL INFO Channel 00/0 : 7[7] -> 8[0] [send] via NET/Libfabric/0(0)/GDRDMA
+ip-26-0-163-58:934722:935116 [2] NCCL INFO Channel 02/0 : 3[3] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934722:935116 [2] NCCL INFO Channel 10/0 : 3[3] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-31:3197182:3197588 [2] NCCL INFO Channel 02/0 : 19[3] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-14:2525075:2525471 [0] NCCL INFO Channel 08/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-170-31:3197182:3197588 [2] NCCL INFO Channel 10/0 : 19[3] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330183:3330590 [0] NCCL INFO Channel 13/0 : 16[0] -> 20[4] via P2P/IPC
+ip-26-0-162-14:2525082:2525472 [7] NCCL INFO Channel 08/0 : 7[7] -> 8[0] [send] via NET/Libfabric/0(0)/GDRDMA
+ip-26-0-162-14:2525075:2525471 [0] NCCL INFO Channel 05/0 : 0[0] -> 4[4] via P2P/IPC
+ip-26-0-163-58:934720:935120 [0] NCCL INFO Channel 05/0 : 8[0] -> 12[4] via P2P/IPC
+ip-26-0-170-31:3197183:3197592 [3] NCCL INFO Channel 03/0 : 20[4] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934723:935117 [3] NCCL INFO Channel 03/0 : 4[4] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-31:3197183:3197592 [3] NCCL INFO Channel 11/0 : 20[4] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934723:935117 [3] NCCL INFO Channel 11/0 : 4[4] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934720:935120 [0] NCCL INFO Channel 13/0 : 8[0] -> 12[4] via P2P/IPC
+ip-26-0-162-14:2525075:2525471 [0] NCCL INFO Channel 13/0 : 0[0] -> 4[4] via P2P/IPC
+ip-26-0-170-31:3197183:3197592 [3] NCCL INFO Channel 02/0 : 27[3] -> 2[2] [send] via NET/Libfabric/2(26)/GDRDMA
+ip-26-0-163-58:934723:935117 [3] NCCL INFO Channel 02/0 : 11[3] -> 18[2] [send] via NET/Libfabric/2(10)/GDRDMA
+ip-26-0-163-58:934723:935117 [3] NCCL INFO Channel 10/0 : 11[3] -> 18[2] [send] via NET/Libfabric/2(10)/GDRDMA
+ip-26-0-170-31:3197183:3197592 [3] NCCL INFO Channel 10/0 : 27[3] -> 2[2] [send] via NET/Libfabric/2(26)/GDRDMA
+ip-26-0-163-58:934720:935120 [0] NCCL INFO Channel 06/0 : 8[0] -> 13[5] via P2P/IPC
+ip-26-0-163-58:934720:935120 [0] NCCL INFO Channel 14/0 : 8[0] -> 13[5] via P2P/IPC
+ip-26-0-163-58:934727:935114 [7] NCCL INFO Channel 06/0 : 15[7] -> 22[6] [send] via NET/Libfabric/6(14)/GDRDMA
+ip-26-0-163-58:934720:935120 [0] NCCL INFO Channel 07/0 : 8[0] -> 14[6] via P2P/IPC
+ip-26-0-163-58:934727:935114 [7] NCCL INFO Channel 14/0 : 15[7] -> 22[6] [send] via NET/Libfabric/6(14)/GDRDMA
+ip-26-0-163-58:934720:935120 [0] NCCL INFO Channel 15/0 : 8[0] -> 14[6] via P2P/IPC
+ip-26-0-170-31:3197182:3197588 [2] NCCL INFO Channel 01/0 : 26[2] -> 1[1] [send] via NET/Libfabric/1(25)/GDRDMA
+ip-26-0-163-58:934720:935120 [0] NCCL INFO Channel 01/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-162-14:2525077:2525475 [2] NCCL INFO Channel 02/0 : 27[3] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934720:935120 [0] NCCL INFO Channel 02/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-162-14:2525077:2525475 [2] NCCL INFO Channel 10/0 : 27[3] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934720:935120 [0] NCCL INFO Channel 03/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-162-14:2525078:2525474 [3] NCCL INFO Channel 03/0 : 28[4] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-14:2525075:2525471 [0] NCCL INFO Channel 06/0 : 0[0] -> 5[5] via P2P/IPC
+ip-26-0-162-14:2525078:2525474 [3] NCCL INFO Channel 11/0 : 28[4] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934720:935120 [0] NCCL INFO Channel 04/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-162-14:2525078:2525474 [3] NCCL INFO Channel 02/0 : 3[3] -> 10[2] [send] via NET/Libfabric/2(2)/GDRDMA
+ip-26-0-163-58:934720:935120 [0] NCCL INFO Channel 09/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-162-14:2525078:2525474 [3] NCCL INFO Channel 10/0 : 3[3] -> 10[2] [send] via NET/Libfabric/2(2)/GDRDMA
+ip-26-0-162-14:2525075:2525471 [0] NCCL INFO Channel 14/0 : 0[0] -> 5[5] via P2P/IPC
+ip-26-0-163-58:934720:935120 [0] NCCL INFO Channel 10/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-163-58:934720:935120 [0] NCCL INFO Channel 11/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-162-14:2525075:2525471 [0] NCCL INFO Channel 07/0 : 0[0] -> 6[6] via P2P/IPC
+ip-26-0-163-58:934720:935120 [0] NCCL INFO Channel 12/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-162-14:2525075:2525471 [0] NCCL INFO Channel 15/0 : 0[0] -> 6[6] via P2P/IPC
+ip-26-0-162-14:2525075:2525471 [0] NCCL INFO Channel 01/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-170-132:3330190:3330589 [7] NCCL INFO Channel 06/0 : 23[7] -> 30[6] [send] via NET/Libfabric/6(22)/GDRDMA
+ip-26-0-170-132:3330190:3330589 [7] NCCL INFO Channel 14/0 : 23[7] -> 30[6] [send] via NET/Libfabric/6(22)/GDRDMA
+ip-26-0-170-31:3197180:3197589 [0] NCCL INFO Channel 06/0 : 24[0] -> 29[5] via P2P/IPC
+ip-26-0-170-31:3197187:3197590 [7] NCCL INFO Channel 06/0 : 31[7] -> 6[6] [send] via NET/Libfabric/6(30)/GDRDMA
+ip-26-0-170-31:3197187:3197590 [7] NCCL INFO Channel 14/0 : 31[7] -> 6[6] [send] via NET/Libfabric/6(30)/GDRDMA
+ip-26-0-170-31:3197180:3197589 [0] NCCL INFO Channel 14/0 : 24[0] -> 29[5] via P2P/IPC
+ip-26-0-170-31:3197180:3197589 [0] NCCL INFO Channel 07/0 : 24[0] -> 30[6] via P2P/IPC
+ip-26-0-162-14:2525075:2525471 [0] NCCL INFO Channel 02/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-162-14:2525075:2525471 [0] NCCL INFO Channel 03/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-170-31:3197180:3197589 [0] NCCL INFO Channel 15/0 : 24[0] -> 30[6] via P2P/IPC
+ip-26-0-162-14:2525075:2525471 [0] NCCL INFO Channel 04/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-162-14:2525082:2525472 [7] NCCL INFO Channel 06/0 : 7[7] -> 14[6] [send] via NET/Libfabric/6(6)/GDRDMA
+ip-26-0-170-31:3197180:3197589 [0] NCCL INFO Channel 01/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-162-14:2525075:2525471 [0] NCCL INFO Channel 09/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-170-31:3197180:3197589 [0] NCCL INFO Channel 02/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-162-14:2525075:2525471 [0] NCCL INFO Channel 10/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-162-14:2525075:2525471 [0] NCCL INFO Channel 11/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-170-132:3330183:3330590 [0] NCCL INFO Channel 06/0 : 16[0] -> 21[5] via P2P/IPC
+ip-26-0-162-14:2525075:2525471 [0] NCCL INFO Channel 12/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-170-132:3330183:3330590 [0] NCCL INFO Channel 14/0 : 16[0] -> 21[5] via P2P/IPC
+ip-26-0-170-132:3330183:3330590 [0] NCCL INFO Channel 07/0 : 16[0] -> 22[6] via P2P/IPC
+ip-26-0-170-132:3330183:3330590 [0] NCCL INFO Channel 15/0 : 16[0] -> 22[6] via P2P/IPC
+ip-26-0-170-31:3197180:3197589 [0] NCCL INFO Channel 03/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-170-31:3197180:3197589 [0] NCCL INFO Channel 04/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-170-132:3330183:3330590 [0] NCCL INFO Channel 01/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-170-31:3197180:3197589 [0] NCCL INFO Channel 09/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-170-31:3197180:3197589 [0] NCCL INFO Channel 10/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-170-132:3330183:3330590 [0] NCCL INFO Channel 02/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-170-31:3197180:3197589 [0] NCCL INFO Channel 11/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-170-31:3197180:3197589 [0] NCCL INFO Channel 12/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-170-132:3330183:3330590 [0] NCCL INFO Channel 03/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-170-132:3330183:3330590 [0] NCCL INFO Channel 04/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-170-132:3330183:3330590 [0] NCCL INFO Channel 09/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-170-132:3330183:3330590 [0] NCCL INFO Channel 10/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-170-132:3330183:3330590 [0] NCCL INFO Channel 11/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-170-132:3330183:3330590 [0] NCCL INFO Channel 12/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-170-132:3330184:3330596 [1] NCCL INFO Channel 01/0 : 10[2] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330185:3330593 [2] NCCL INFO Channel 01/0 : 18[2] -> 25[1] [send] via NET/Libfabric/1(17)/GDRDMA
+ip-26-0-170-132:3330184:3330596 [1] NCCL INFO Channel 09/0 : 10[2] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330185:3330593 [2] NCCL INFO Channel 09/0 : 18[2] -> 25[1] [send] via NET/Libfabric/1(17)/GDRDMA
+ip-26-0-163-58:934721:935115 [1] NCCL INFO Channel 01/0 : 2[2] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934721:935115 [1] NCCL INFO Channel 09/0 : 2[2] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934722:935116 [2] NCCL INFO Channel 01/0 : 10[2] -> 17[1] [send] via NET/Libfabric/1(9)/GDRDMA
+ip-26-0-163-58:934722:935116 [2] NCCL INFO Channel 09/0 : 10[2] -> 17[1] [send] via NET/Libfabric/1(9)/GDRDMA
+ip-26-0-163-58:934722:935116 [2] NCCL INFO Channel 02/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-162-14:2525076:2525478 [1] NCCL INFO Channel 01/0 : 26[2] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-31:3197181:3197587 [1] NCCL INFO Channel 01/0 : 18[2] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934722:935116 [2] NCCL INFO Channel 03/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-170-132:3330184:3330596 [1] NCCL INFO Channel 01/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-162-14:2525076:2525478 [1] NCCL INFO Channel 09/0 : 26[2] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-14:2525077:2525475 [2] NCCL INFO Channel 01/0 : 2[2] -> 9[1] [send] via NET/Libfabric/1(1)/GDRDMA
+ip-26-0-170-31:3197182:3197588 [2] NCCL INFO Channel 09/0 : 26[2] -> 1[1] [send] via NET/Libfabric/1(25)/GDRDMA
+ip-26-0-170-31:3197181:3197587 [1] NCCL INFO Channel 09/0 : 18[2] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-14:2525077:2525475 [2] NCCL INFO Channel 09/0 : 2[2] -> 9[1] [send] via NET/Libfabric/1(1)/GDRDMA
+ip-26-0-170-132:3330184:3330596 [1] NCCL INFO Channel 02/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-163-58:934722:935116 [2] NCCL INFO Channel 04/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-170-132:3330184:3330596 [1] NCCL INFO Channel 03/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-163-58:934722:935116 [2] NCCL INFO Channel 05/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-163-58:934724:935113 [4] NCCL INFO Channel 04/0 : 5[5] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934724:935113 [4] NCCL INFO Channel 12/0 : 5[5] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934722:935116 [2] NCCL INFO Channel 06/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-170-132:3330184:3330596 [1] NCCL INFO Channel 04/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-163-58:934724:935113 [4] NCCL INFO Channel 03/0 : 12[4] -> 19[3] [send] via NET/Libfabric/3(11)/GDRDMA
+ip-26-0-170-132:3330184:3330596 [1] NCCL INFO Channel 05/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-163-58:934722:935116 [2] NCCL INFO Channel 07/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-163-58:934724:935113 [4] NCCL INFO Channel 11/0 : 12[4] -> 19[3] [send] via NET/Libfabric/3(11)/GDRDMA
+ip-26-0-170-132:3330184:3330596 [1] NCCL INFO Channel 06/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-163-58:934726:935119 [6] NCCL INFO Channel 06/0 : 7[7] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934722:935116 [2] NCCL INFO Channel 10/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-163-58:934726:935119 [6] NCCL INFO Channel 14/0 : 7[7] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934725:935118 [5] NCCL INFO Channel 05/0 : 6[6] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330184:3330596 [1] NCCL INFO Channel 07/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-163-58:934725:935118 [5] NCCL INFO Channel 13/0 : 6[6] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934722:935116 [2] NCCL INFO Channel 11/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-163-58:934726:935119 [6] NCCL INFO Channel 05/0 : 14[6] -> 21[5] [send] via NET/Libfabric/5(13)/GDRDMA
+ip-26-0-163-58:934725:935118 [5] NCCL INFO Channel 04/0 : 13[5] -> 20[4] [send] via NET/Libfabric/4(12)/GDRDMA
+ip-26-0-163-58:934726:935119 [6] NCCL INFO Channel 13/0 : 14[6] -> 21[5] [send] via NET/Libfabric/5(13)/GDRDMA
+ip-26-0-163-58:934725:935118 [5] NCCL INFO Channel 12/0 : 13[5] -> 20[4] [send] via NET/Libfabric/4(12)/GDRDMA
+ip-26-0-170-132:3330184:3330596 [1] NCCL INFO Channel 09/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-163-58:934722:935116 [2] NCCL INFO Channel 12/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-163-58:934722:935116 [2] NCCL INFO Channel 13/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-170-132:3330184:3330596 [1] NCCL INFO Channel 10/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-170-132:3330184:3330596 [1] NCCL INFO Channel 11/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-170-132:3330184:3330596 [1] NCCL INFO Channel 12/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-163-58:934722:935116 [2] NCCL INFO Channel 14/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-170-132:3330184:3330596 [1] NCCL INFO Channel 13/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-163-58:934722:935116 [2] NCCL INFO Channel 15/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-170-132:3330184:3330596 [1] NCCL INFO Channel 14/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-170-31:3197184:3197586 [4] NCCL INFO Channel 04/0 : 21[5] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-132:3330184:3330596 [1] NCCL INFO Channel 15/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-170-31:3197184:3197586 [4] NCCL INFO Channel 12/0 : 21[5] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-31:3197182:3197588 [2] NCCL INFO Channel 02/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-170-31:3197181:3197587 [1] NCCL INFO Channel 01/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-170-31:3197184:3197586 [4] NCCL INFO Channel 03/0 : 28[4] -> 3[3] [send] via NET/Libfabric/3(27)/GDRDMA
+ip-26-0-170-31:3197184:3197586 [4] NCCL INFO Channel 11/0 : 28[4] -> 3[3] [send] via NET/Libfabric/3(27)/GDRDMA
+ip-26-0-170-31:3197182:3197588 [2] NCCL INFO Channel 03/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-170-31:3197181:3197587 [1] NCCL INFO Channel 02/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-170-132:3330185:3330593 [2] NCCL INFO Channel 02/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-170-31:3197186:3197593 [6] NCCL INFO Channel 06/0 : 23[7] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934721:935115 [1] NCCL INFO Channel 01/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-170-31:3197182:3197588 [2] NCCL INFO Channel 04/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-170-31:3197181:3197587 [1] NCCL INFO Channel 03/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-170-31:3197186:3197593 [6] NCCL INFO Channel 14/0 : 23[7] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-14:2525080:2525476 [5] NCCL INFO Channel 05/0 : 30[6] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-14:2525080:2525476 [5] NCCL INFO Channel 13/0 : 30[6] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934721:935115 [1] NCCL INFO Channel 02/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-162-14:2525079:2525477 [4] NCCL INFO Channel 04/0 : 29[5] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-132:3330189:3330594 [6] NCCL INFO Channel 06/0 : 15[7] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-31:3197182:3197588 [2] NCCL INFO Channel 05/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-162-14:2525079:2525477 [4] NCCL INFO Channel 12/0 : 29[5] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-132:3330189:3330594 [6] NCCL INFO Channel 14/0 : 15[7] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-31:3197181:3197587 [1] NCCL INFO Channel 04/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-162-14:2525080:2525476 [5] NCCL INFO Channel 04/0 : 5[5] -> 12[4] [send] via NET/Libfabric/4(4)/GDRDMA
+ip-26-0-163-58:934721:935115 [1] NCCL INFO Channel 03/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-170-31:3197185:3197591 [5] NCCL INFO Channel 05/0 : 22[6] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-14:2525080:2525476 [5] NCCL INFO Channel 12/0 : 5[5] -> 12[4] [send] via NET/Libfabric/4(4)/GDRDMA
+ip-26-0-170-31:3197185:3197591 [5] NCCL INFO Channel 13/0 : 22[6] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-31:3197186:3197593 [6] NCCL INFO Channel 05/0 : 30[6] -> 5[5] [send] via NET/Libfabric/5(29)/GDRDMA
+ip-26-0-163-58:934721:935115 [1] NCCL INFO Channel 04/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-170-31:3197182:3197588 [2] NCCL INFO Channel 06/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-170-31:3197186:3197593 [6] NCCL INFO Channel 13/0 : 30[6] -> 5[5] [send] via NET/Libfabric/5(29)/GDRDMA
+ip-26-0-162-14:2525081:2525473 [6] NCCL INFO Channel 06/0 : 31[7] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-31:3197181:3197587 [1] NCCL INFO Channel 05/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-170-31:3197185:3197591 [5] NCCL INFO Channel 04/0 : 29[5] -> 4[4] [send] via NET/Libfabric/4(28)/GDRDMA
+ip-26-0-170-31:3197185:3197591 [5] NCCL INFO Channel 12/0 : 29[5] -> 4[4] [send] via NET/Libfabric/4(28)/GDRDMA
+ip-26-0-162-14:2525082:2525472 [7] NCCL INFO Channel 14/0 : 7[7] -> 14[6] [send] via NET/Libfabric/6(6)/GDRDMA
+ip-26-0-163-58:934721:935115 [1] NCCL INFO Channel 05/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-170-132:3330187:3330591 [4] NCCL INFO Channel 04/0 : 13[5] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-14:2525081:2525473 [6] NCCL INFO Channel 14/0 : 31[7] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330187:3330591 [4] NCCL INFO Channel 12/0 : 13[5] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-31:3197182:3197588 [2] NCCL INFO Channel 07/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-170-31:3197181:3197587 [1] NCCL INFO Channel 06/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-162-14:2525080:2525476 [5] NCCL INFO Channel 05/0 : 5[5] -> 3[3] via P2P/IPC
+ip-26-0-162-14:2525081:2525473 [6] NCCL INFO Channel 05/0 : 6[6] -> 13[5] [send] via NET/Libfabric/5(5)/GDRDMA
+ip-26-0-170-31:3197186:3197593 [6] NCCL INFO Channel 06/0 : 30[6] -> 27[3] via P2P/IPC
+ip-26-0-163-58:934721:935115 [1] NCCL INFO Channel 06/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-162-14:2525081:2525473 [6] NCCL INFO Channel 13/0 : 6[6] -> 13[5] [send] via NET/Libfabric/5(5)/GDRDMA
+ip-26-0-170-132:3330188:3330595 [5] NCCL INFO Channel 05/0 : 14[6] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330187:3330591 [4] NCCL INFO Channel 03/0 : 20[4] -> 27[3] [send] via NET/Libfabric/3(19)/GDRDMA
+ip-26-0-170-132:3330188:3330595 [5] NCCL INFO Channel 13/0 : 14[6] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934721:935115 [1] NCCL INFO Channel 07/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-170-132:3330187:3330591 [4] NCCL INFO Channel 11/0 : 20[4] -> 27[3] [send] via NET/Libfabric/3(19)/GDRDMA
+ip-26-0-170-132:3330189:3330594 [6] NCCL INFO Channel 05/0 : 22[6] -> 29[5] [send] via NET/Libfabric/5(21)/GDRDMA
+ip-26-0-170-132:3330189:3330594 [6] NCCL INFO Channel 13/0 : 22[6] -> 29[5] [send] via NET/Libfabric/5(21)/GDRDMA
+ip-26-0-170-31:3197182:3197588 [2] NCCL INFO Channel 10/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-170-132:3330188:3330595 [5] NCCL INFO Channel 04/0 : 21[5] -> 28[4] [send] via NET/Libfabric/4(20)/GDRDMA
+ip-26-0-162-14:2525081:2525473 [6] NCCL INFO Channel 06/0 : 6[6] -> 3[3] via P2P/IPC
+ip-26-0-163-58:934721:935115 [1] NCCL INFO Channel 09/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-170-31:3197181:3197587 [1] NCCL INFO Channel 07/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-163-58:934725:935118 [5] NCCL INFO Channel 05/0 : 13[5] -> 11[3] via P2P/IPC
+ip-26-0-170-132:3330188:3330595 [5] NCCL INFO Channel 12/0 : 21[5] -> 28[4] [send] via NET/Libfabric/4(20)/GDRDMA
+ip-26-0-170-31:3197186:3197593 [6] NCCL INFO Channel 14/0 : 30[6] -> 27[3] via P2P/IPC
+ip-26-0-163-58:934721:935115 [1] NCCL INFO Channel 10/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-170-132:3330189:3330594 [6] NCCL INFO Channel 06/0 : 22[6] -> 19[3] via P2P/IPC
+ip-26-0-163-58:934726:935119 [6] NCCL INFO Channel 06/0 : 14[6] -> 11[3] via P2P/IPC
+ip-26-0-170-31:3197182:3197588 [2] NCCL INFO Channel 11/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-170-132:3330188:3330595 [5] NCCL INFO Channel 05/0 : 21[5] -> 19[3] via P2P/IPC
+ip-26-0-170-31:3197181:3197587 [1] NCCL INFO Channel 09/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-163-58:934725:935118 [5] NCCL INFO Channel 13/0 : 13[5] -> 11[3] via P2P/IPC
+ip-26-0-163-58:934721:935115 [1] NCCL INFO Channel 11/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-170-31:3197182:3197588 [2] NCCL INFO Channel 12/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-170-31:3197181:3197587 [1] NCCL INFO Channel 10/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-163-58:934726:935119 [6] NCCL INFO Channel 14/0 : 14[6] -> 11[3] via P2P/IPC
+ip-26-0-163-58:934724:935113 [4] NCCL INFO Channel 07/0 : 12[4] -> 23[7] [send] via NET/Libfabric/7(15)/GDRDMA
+ip-26-0-170-132:3330187:3330591 [4] NCCL INFO Channel 07/0 : 20[4] -> 31[7] [send] via NET/Libfabric/7(23)/GDRDMA
+ip-26-0-170-31:3197182:3197588 [2] NCCL INFO Channel 13/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-163-58:934721:935115 [1] NCCL INFO Channel 12/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-170-132:3330187:3330591 [4] NCCL INFO Channel 15/0 : 20[4] -> 31[7] [send] via NET/Libfabric/7(23)/GDRDMA
+ip-26-0-170-31:3197181:3197587 [1] NCCL INFO Channel 11/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-163-58:934721:935115 [1] NCCL INFO Channel 13/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-163-58:934721:935115 [1] NCCL INFO Channel 14/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-162-14:2525077:2525475 [2] NCCL INFO Channel 02/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-162-14:2525079:2525477 [4] NCCL INFO Channel 03/0 : 4[4] -> 11[3] [send] via NET/Libfabric/3(3)/GDRDMA
+ip-26-0-163-58:934721:935115 [1] NCCL INFO Channel 15/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-162-14:2525079:2525477 [4] NCCL INFO Channel 11/0 : 4[4] -> 11[3] [send] via NET/Libfabric/3(3)/GDRDMA
+ip-26-0-162-14:2525076:2525478 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-170-31:3197184:3197586 [4] NCCL INFO Channel 07/0 : 28[4] -> 7[7] [send] via NET/Libfabric/7(31)/GDRDMA
+ip-26-0-170-31:3197185:3197591 [5] NCCL INFO Channel 05/0 : 29[5] -> 27[3] via P2P/IPC
+ip-26-0-162-14:2525077:2525475 [2] NCCL INFO Channel 03/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-170-31:3197182:3197588 [2] NCCL INFO Channel 14/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-162-14:2525080:2525476 [5] NCCL INFO Channel 13/0 : 5[5] -> 3[3] via P2P/IPC
+ip-26-0-162-14:2525076:2525478 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-170-31:3197181:3197587 [1] NCCL INFO Channel 12/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-162-14:2525077:2525475 [2] NCCL INFO Channel 04/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-170-31:3197185:3197591 [5] NCCL INFO Channel 13/0 : 29[5] -> 27[3] via P2P/IPC
+ip-26-0-170-31:3197182:3197588 [2] NCCL INFO Channel 15/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-162-14:2525081:2525473 [6] NCCL INFO Channel 14/0 : 6[6] -> 3[3] via P2P/IPC
+ip-26-0-170-31:3197181:3197587 [1] NCCL INFO Channel 13/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-162-14:2525076:2525478 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-170-31:3197181:3197587 [1] NCCL INFO Channel 14/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-170-132:3330185:3330593 [2] NCCL INFO Channel 03/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-162-14:2525077:2525475 [2] NCCL INFO Channel 05/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-170-31:3197181:3197587 [1] NCCL INFO Channel 15/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-162-14:2525076:2525478 [1] NCCL INFO Channel 04/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-162-14:2525077:2525475 [2] NCCL INFO Channel 06/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-162-14:2525076:2525478 [1] NCCL INFO Channel 05/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330189:3330594 [6] NCCL INFO Channel 14/0 : 22[6] -> 19[3] via P2P/IPC
+ip-26-0-162-14:2525077:2525475 [2] NCCL INFO Channel 07/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-170-132:3330188:3330595 [5] NCCL INFO Channel 13/0 : 21[5] -> 19[3] via P2P/IPC
+ip-26-0-162-14:2525076:2525478 [1] NCCL INFO Channel 06/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-162-14:2525077:2525475 [2] NCCL INFO Channel 10/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-162-14:2525076:2525478 [1] NCCL INFO Channel 07/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-162-14:2525077:2525475 [2] NCCL INFO Channel 11/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-162-14:2525076:2525478 [1] NCCL INFO Channel 09/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-162-14:2525077:2525475 [2] NCCL INFO Channel 12/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-162-14:2525076:2525478 [1] NCCL INFO Channel 10/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-162-14:2525077:2525475 [2] NCCL INFO Channel 13/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-162-14:2525076:2525478 [1] NCCL INFO Channel 11/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-162-14:2525077:2525475 [2] NCCL INFO Channel 14/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-170-132:3330185:3330593 [2] NCCL INFO Channel 04/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-162-14:2525076:2525478 [1] NCCL INFO Channel 12/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-162-14:2525077:2525475 [2] NCCL INFO Channel 15/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-162-14:2525076:2525478 [1] NCCL INFO Channel 13/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-162-14:2525076:2525478 [1] NCCL INFO Channel 14/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-162-14:2525076:2525478 [1] NCCL INFO Channel 15/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330185:3330593 [2] NCCL INFO Channel 05/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-170-132:3330185:3330593 [2] NCCL INFO Channel 06/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-170-132:3330185:3330593 [2] NCCL INFO Channel 07/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-170-132:3330185:3330593 [2] NCCL INFO Channel 10/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-170-132:3330185:3330593 [2] NCCL INFO Channel 11/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-170-132:3330185:3330593 [2] NCCL INFO Channel 12/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-170-132:3330185:3330593 [2] NCCL INFO Channel 13/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-170-132:3330185:3330593 [2] NCCL INFO Channel 14/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-170-132:3330185:3330593 [2] NCCL INFO Channel 15/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-163-58:934727:935114 [7] NCCL INFO Channel 07/0 : 4[4] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934727:935114 [7] NCCL INFO Channel 15/0 : 4[4] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934724:935113 [4] NCCL INFO Channel 15/0 : 12[4] -> 23[7] [send] via NET/Libfabric/7(15)/GDRDMA
+ip-26-0-162-14:2525082:2525472 [7] NCCL INFO Channel 07/0 : 28[4] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-31:3197187:3197590 [7] NCCL INFO Channel 07/0 : 20[4] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-14:2525082:2525472 [7] NCCL INFO Channel 15/0 : 28[4] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-14:2525079:2525477 [4] NCCL INFO Channel 07/0 : 4[4] -> 15[7] [send] via NET/Libfabric/7(7)/GDRDMA
+ip-26-0-170-31:3197184:3197586 [4] NCCL INFO Channel 15/0 : 28[4] -> 7[7] [send] via NET/Libfabric/7(31)/GDRDMA
+ip-26-0-162-14:2525079:2525477 [4] NCCL INFO Channel 15/0 : 4[4] -> 15[7] [send] via NET/Libfabric/7(7)/GDRDMA
+ip-26-0-170-31:3197187:3197590 [7] NCCL INFO Channel 15/0 : 20[4] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-14:2525082:2525472 [7] NCCL INFO Channel 07/0 : 7[7] -> 3[3] via P2P/IPC
+ip-26-0-170-132:3330190:3330589 [7] NCCL INFO Channel 07/0 : 12[4] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-31:3197187:3197590 [7] NCCL INFO Channel 07/0 : 31[7] -> 27[3] via P2P/IPC
+ip-26-0-170-132:3330190:3330589 [7] NCCL INFO Channel 15/0 : 12[4] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934727:935114 [7] NCCL INFO Channel 07/0 : 15[7] -> 11[3] via P2P/IPC
+ip-26-0-170-132:3330190:3330589 [7] NCCL INFO Channel 07/0 : 23[7] -> 19[3] via P2P/IPC
+ip-26-0-162-14:2525082:2525472 [7] NCCL INFO Channel 15/0 : 7[7] -> 3[3] via P2P/IPC
+ip-26-0-170-132:3330190:3330589 [7] NCCL INFO Channel 15/0 : 23[7] -> 19[3] via P2P/IPC
+ip-26-0-163-58:934727:935114 [7] NCCL INFO Channel 15/0 : 15[7] -> 11[3] via P2P/IPC
+ip-26-0-170-132:3330190:3330589 [7] NCCL INFO Channel 01/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-163-58:934727:935114 [7] NCCL INFO Channel 01/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-170-132:3330190:3330589 [7] NCCL INFO Channel 02/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-163-58:934727:935114 [7] NCCL INFO Channel 02/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-170-31:3197187:3197590 [7] NCCL INFO Channel 15/0 : 31[7] -> 27[3] via P2P/IPC
+ip-26-0-163-58:934727:935114 [7] NCCL INFO Channel 03/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-163-58:934727:935114 [7] NCCL INFO Channel 04/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-162-14:2525082:2525472 [7] NCCL INFO Channel 01/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-163-58:934727:935114 [7] NCCL INFO Channel 05/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-163-58:934727:935114 [7] NCCL INFO Channel 09/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-170-132:3330190:3330589 [7] NCCL INFO Channel 03/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-163-58:934727:935114 [7] NCCL INFO Channel 10/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-170-31:3197187:3197590 [7] NCCL INFO Channel 01/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-163-58:934727:935114 [7] NCCL INFO Channel 11/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-170-31:3197187:3197590 [7] NCCL INFO Channel 02/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-162-14:2525082:2525472 [7] NCCL INFO Channel 02/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-170-31:3197187:3197590 [7] NCCL INFO Channel 03/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-163-58:934727:935114 [7] NCCL INFO Channel 12/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-170-132:3330190:3330589 [7] NCCL INFO Channel 04/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-170-31:3197187:3197590 [7] NCCL INFO Channel 04/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-163-58:934724:935113 [4] NCCL INFO Channel 01/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-162-14:2525082:2525472 [7] NCCL INFO Channel 03/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-163-58:934727:935114 [7] NCCL INFO Channel 13/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-162-14:2525082:2525472 [7] NCCL INFO Channel 04/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-170-31:3197187:3197590 [7] NCCL INFO Channel 05/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-170-31:3197187:3197590 [7] NCCL INFO Channel 09/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-170-31:3197187:3197590 [7] NCCL INFO Channel 10/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-163-58:934724:935113 [4] NCCL INFO Channel 02/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-162-14:2525082:2525472 [7] NCCL INFO Channel 05/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-170-31:3197187:3197590 [7] NCCL INFO Channel 11/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-163-58:934724:935113 [4] NCCL INFO Channel 04/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-170-132:3330190:3330589 [7] NCCL INFO Channel 05/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-162-14:2525082:2525472 [7] NCCL INFO Channel 09/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-170-31:3197184:3197586 [4] NCCL INFO Channel 01/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-163-58:934724:935113 [4] NCCL INFO Channel 09/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-170-31:3197187:3197590 [7] NCCL INFO Channel 12/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-163-58:934723:935117 [3] NCCL INFO Channel 01/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-162-14:2525082:2525472 [7] NCCL INFO Channel 10/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-170-31:3197184:3197586 [4] NCCL INFO Channel 02/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-170-31:3197187:3197590 [7] NCCL INFO Channel 13/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-163-58:934724:935113 [4] NCCL INFO Channel 10/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-163-58:934723:935117 [3] NCCL INFO Channel 03/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-170-31:3197184:3197586 [4] NCCL INFO Channel 04/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-162-14:2525082:2525472 [7] NCCL INFO Channel 11/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-170-31:3197184:3197586 [4] NCCL INFO Channel 09/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-163-58:934724:935113 [4] NCCL INFO Channel 12/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-162-14:2525079:2525477 [4] NCCL INFO Channel 01/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-163-58:934726:935119 [6] NCCL INFO Channel 01/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-162-14:2525082:2525472 [7] NCCL INFO Channel 12/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-163-58:934723:935117 [3] NCCL INFO Channel 04/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-170-31:3197184:3197586 [4] NCCL INFO Channel 10/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-162-14:2525079:2525477 [4] NCCL INFO Channel 02/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-163-58:934725:935118 [5] NCCL INFO Channel 01/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-162-14:2525082:2525472 [7] NCCL INFO Channel 13/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-163-58:934726:935119 [6] NCCL INFO Channel 02/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-163-58:934723:935117 [3] NCCL INFO Channel 05/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-170-31:3197184:3197586 [4] NCCL INFO Channel 12/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-163-58:934725:935118 [5] NCCL INFO Channel 02/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-170-132:3330190:3330589 [7] NCCL INFO Channel 09/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-163-58:934726:935119 [6] NCCL INFO Channel 03/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-163-58:934723:935117 [3] NCCL INFO Channel 06/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-163-58:934725:935118 [5] NCCL INFO Channel 03/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-170-31:3197183:3197592 [3] NCCL INFO Channel 01/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-163-58:934726:935119 [6] NCCL INFO Channel 04/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-163-58:934723:935117 [3] NCCL INFO Channel 07/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-163-58:934725:935118 [5] NCCL INFO Channel 06/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-162-14:2525079:2525477 [4] NCCL INFO Channel 04/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-170-31:3197183:3197592 [3] NCCL INFO Channel 03/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-163-58:934726:935119 [6] NCCL INFO Channel 07/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-163-58:934723:935117 [3] NCCL INFO Channel 09/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-170-132:3330190:3330589 [7] NCCL INFO Channel 10/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-163-58:934725:935118 [5] NCCL INFO Channel 07/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-162-14:2525079:2525477 [4] NCCL INFO Channel 09/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-163-58:934726:935119 [6] NCCL INFO Channel 09/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-163-58:934723:935117 [3] NCCL INFO Channel 11/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-163-58:934725:935118 [5] NCCL INFO Channel 09/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-170-132:3330190:3330589 [7] NCCL INFO Channel 11/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-162-14:2525079:2525477 [4] NCCL INFO Channel 10/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-163-58:934723:935117 [3] NCCL INFO Channel 12/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-163-58:934726:935119 [6] NCCL INFO Channel 10/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-163-58:934725:935118 [5] NCCL INFO Channel 10/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-170-31:3197183:3197592 [3] NCCL INFO Channel 04/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-170-132:3330190:3330589 [7] NCCL INFO Channel 12/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-170-132:3330187:3330591 [4] NCCL INFO Channel 01/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-170-31:3197183:3197592 [3] NCCL INFO Channel 05/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-162-14:2525079:2525477 [4] NCCL INFO Channel 12/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-170-31:3197186:3197593 [6] NCCL INFO Channel 01/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-163-58:934723:935117 [3] NCCL INFO Channel 13/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-163-58:934726:935119 [6] NCCL INFO Channel 11/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-163-58:934725:935118 [5] NCCL INFO Channel 11/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-170-31:3197183:3197592 [3] NCCL INFO Channel 06/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-170-132:3330190:3330589 [7] NCCL INFO Channel 13/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-162-14:2525078:2525474 [3] NCCL INFO Channel 01/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-170-31:3197186:3197593 [6] NCCL INFO Channel 02/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-170-132:3330187:3330591 [4] NCCL INFO Channel 02/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-170-31:3197183:3197592 [3] NCCL INFO Channel 07/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-170-31:3197186:3197593 [6] NCCL INFO Channel 03/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-162-14:2525078:2525474 [3] NCCL INFO Channel 03/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-170-31:3197183:3197592 [3] NCCL INFO Channel 09/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-170-31:3197186:3197593 [6] NCCL INFO Channel 04/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-170-132:3330187:3330591 [4] NCCL INFO Channel 04/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-170-31:3197183:3197592 [3] NCCL INFO Channel 11/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-170-31:3197186:3197593 [6] NCCL INFO Channel 07/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-170-31:3197183:3197592 [3] NCCL INFO Channel 12/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-170-31:3197186:3197593 [6] NCCL INFO Channel 09/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-170-31:3197183:3197592 [3] NCCL INFO Channel 13/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-170-31:3197186:3197593 [6] NCCL INFO Channel 10/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-163-58:934723:935117 [3] NCCL INFO Channel 14/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-170-31:3197183:3197592 [3] NCCL INFO Channel 14/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-162-14:2525081:2525473 [6] NCCL INFO Channel 01/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-162-14:2525078:2525474 [3] NCCL INFO Channel 04/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-170-31:3197186:3197593 [6] NCCL INFO Channel 11/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-163-58:934723:935117 [3] NCCL INFO Channel 15/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-170-31:3197183:3197592 [3] NCCL INFO Channel 15/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-163-58:934725:935118 [5] NCCL INFO Channel 14/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-170-132:3330187:3330591 [4] NCCL INFO Channel 09/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-162-14:2525081:2525473 [6] NCCL INFO Channel 02/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-170-132:3330186:3330592 [3] NCCL INFO Channel 01/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-170-31:3197185:3197591 [5] NCCL INFO Channel 01/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-162-14:2525078:2525474 [3] NCCL INFO Channel 05/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-163-58:934725:935118 [5] NCCL INFO Channel 15/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-163-58:934726:935119 [6] NCCL INFO Channel 12/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-162-14:2525081:2525473 [6] NCCL INFO Channel 03/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-170-31:3197185:3197591 [5] NCCL INFO Channel 02/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-170-132:3330187:3330591 [4] NCCL INFO Channel 10/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-162-14:2525078:2525474 [3] NCCL INFO Channel 06/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-170-132:3330186:3330592 [3] NCCL INFO Channel 03/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-170-31:3197185:3197591 [5] NCCL INFO Channel 03/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-162-14:2525081:2525473 [6] NCCL INFO Channel 04/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-170-31:3197185:3197591 [5] NCCL INFO Channel 06/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-162-14:2525078:2525474 [3] NCCL INFO Channel 07/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-162-14:2525080:2525476 [5] NCCL INFO Channel 01/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-170-132:3330187:3330591 [4] NCCL INFO Channel 12/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-162-14:2525081:2525473 [6] NCCL INFO Channel 07/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-170-132:3330186:3330592 [3] NCCL INFO Channel 04/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-162-14:2525078:2525474 [3] NCCL INFO Channel 09/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-162-14:2525080:2525476 [5] NCCL INFO Channel 02/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-162-14:2525081:2525473 [6] NCCL INFO Channel 09/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-162-14:2525078:2525474 [3] NCCL INFO Channel 11/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-162-14:2525080:2525476 [5] NCCL INFO Channel 03/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-162-14:2525081:2525473 [6] NCCL INFO Channel 10/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-170-31:3197186:3197593 [6] NCCL INFO Channel 12/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-162-14:2525078:2525474 [3] NCCL INFO Channel 12/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-162-14:2525080:2525476 [5] NCCL INFO Channel 06/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-170-31:3197185:3197591 [5] NCCL INFO Channel 07/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-162-14:2525081:2525473 [6] NCCL INFO Channel 11/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-170-132:3330189:3330594 [6] NCCL INFO Channel 01/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-170-132:3330186:3330592 [3] NCCL INFO Channel 05/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-162-14:2525078:2525474 [3] NCCL INFO Channel 13/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-162-14:2525080:2525476 [5] NCCL INFO Channel 07/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-162-14:2525081:2525473 [6] NCCL INFO Channel 12/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-163-58:934726:935119 [6] NCCL INFO Channel 15/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-170-132:3330189:3330594 [6] NCCL INFO Channel 02/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-170-132:3330186:3330592 [3] NCCL INFO Channel 06/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-162-14:2525078:2525474 [3] NCCL INFO Channel 14/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-162-14:2525080:2525476 [5] NCCL INFO Channel 09/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-170-132:3330189:3330594 [6] NCCL INFO Channel 03/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-162-14:2525081:2525473 [6] NCCL INFO Channel 15/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-170-132:3330186:3330592 [3] NCCL INFO Channel 07/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-162-14:2525078:2525474 [3] NCCL INFO Channel 15/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-170-132:3330189:3330594 [6] NCCL INFO Channel 04/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-162-14:2525080:2525476 [5] NCCL INFO Channel 10/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-170-132:3330186:3330592 [3] NCCL INFO Channel 09/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-170-132:3330189:3330594 [6] NCCL INFO Channel 07/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-170-132:3330186:3330592 [3] NCCL INFO Channel 11/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-170-132:3330186:3330592 [3] NCCL INFO Channel 12/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-170-132:3330189:3330594 [6] NCCL INFO Channel 09/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-170-31:3197186:3197593 [6] NCCL INFO Channel 15/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-162-14:2525080:2525476 [5] NCCL INFO Channel 11/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-170-132:3330188:3330595 [5] NCCL INFO Channel 01/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-170-132:3330186:3330592 [3] NCCL INFO Channel 13/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-170-132:3330189:3330594 [6] NCCL INFO Channel 10/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-170-132:3330188:3330595 [5] NCCL INFO Channel 02/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-170-132:3330186:3330592 [3] NCCL INFO Channel 14/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-170-132:3330189:3330594 [6] NCCL INFO Channel 11/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-170-132:3330188:3330595 [5] NCCL INFO Channel 03/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-170-132:3330186:3330592 [3] NCCL INFO Channel 15/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-170-132:3330189:3330594 [6] NCCL INFO Channel 12/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-170-132:3330188:3330595 [5] NCCL INFO Channel 06/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-170-132:3330189:3330594 [6] NCCL INFO Channel 15/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-170-132:3330188:3330595 [5] NCCL INFO Channel 07/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-170-31:3197185:3197591 [5] NCCL INFO Channel 09/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-162-14:2525080:2525476 [5] NCCL INFO Channel 14/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-170-132:3330188:3330595 [5] NCCL INFO Channel 09/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-170-132:3330188:3330595 [5] NCCL INFO Channel 10/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-170-132:3330188:3330595 [5] NCCL INFO Channel 11/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-170-132:3330188:3330595 [5] NCCL INFO Channel 14/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-170-132:3330188:3330595 [5] NCCL INFO Channel 15/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-170-31:3197185:3197591 [5] NCCL INFO Channel 10/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-170-31:3197185:3197591 [5] NCCL INFO Channel 11/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-162-14:2525080:2525476 [5] NCCL INFO Channel 15/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-170-31:3197185:3197591 [5] NCCL INFO Channel 14/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-170-31:3197185:3197591 [5] NCCL INFO Channel 15/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-170-31:3197180:3197589 [0] NCCL INFO Connected all rings
+ip-26-0-170-31:3197180:3197589 [0] NCCL INFO Channel 02/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-170-31:3197183:3197592 [3] NCCL INFO Connected all rings
+ip-26-0-170-132:3330184:3330596 [1] NCCL INFO Connected all rings
+ip-26-0-170-31:3197183:3197592 [3] NCCL INFO Channel 01/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-163-58:934720:935120 [0] NCCL INFO Connected all rings
+ip-26-0-163-58:934720:935120 [0] NCCL INFO Channel 02/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-170-31:3197183:3197592 [3] NCCL INFO Channel 02/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-170-31:3197183:3197592 [3] NCCL INFO Channel 03/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-170-31:3197183:3197592 [3] NCCL INFO Channel 05/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-170-31:3197183:3197592 [3] NCCL INFO Channel 06/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-170-31:3197183:3197592 [3] NCCL INFO Channel 07/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-170-31:3197183:3197592 [3] NCCL INFO Channel 09/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-170-31:3197183:3197592 [3] NCCL INFO Channel 10/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-170-31:3197183:3197592 [3] NCCL INFO Channel 11/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-170-31:3197183:3197592 [3] NCCL INFO Channel 13/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-170-31:3197183:3197592 [3] NCCL INFO Channel 14/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-170-31:3197183:3197592 [3] NCCL INFO Channel 15/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-163-58:934722:935116 [2] NCCL INFO Connected all rings
+ip-26-0-170-132:3330186:3330592 [3] NCCL INFO Connected all rings
+ip-26-0-170-31:3197182:3197588 [2] NCCL INFO Connected all rings
+ip-26-0-170-31:3197180:3197589 [0] NCCL INFO Channel 03/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-170-132:3330183:3330590 [0] NCCL INFO Connected all rings
+ip-26-0-170-132:3330183:3330590 [0] NCCL INFO Channel 02/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-170-31:3197180:3197589 [0] NCCL INFO Channel 04/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-170-132:3330183:3330590 [0] NCCL INFO Channel 03/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-170-31:3197180:3197589 [0] NCCL INFO Channel 05/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-170-31:3197180:3197589 [0] NCCL INFO Channel 06/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-170-31:3197180:3197589 [0] NCCL INFO Channel 07/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-163-58:934721:935115 [1] NCCL INFO Connected all rings
+ip-26-0-170-31:3197182:3197588 [2] NCCL INFO Channel 01/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-170-31:3197180:3197589 [0] NCCL INFO Channel 10/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-170-31:3197182:3197588 [2] NCCL INFO Channel 02/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-170-31:3197180:3197589 [0] NCCL INFO Channel 11/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-170-31:3197182:3197588 [2] NCCL INFO Channel 04/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-170-31:3197180:3197589 [0] NCCL INFO Channel 12/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-170-31:3197182:3197588 [2] NCCL INFO Channel 05/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-170-31:3197180:3197589 [0] NCCL INFO Channel 13/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-170-31:3197182:3197588 [2] NCCL INFO Channel 06/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-163-58:934720:935120 [0] NCCL INFO Channel 03/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-170-31:3197180:3197589 [0] NCCL INFO Channel 14/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-170-31:3197182:3197588 [2] NCCL INFO Channel 07/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-170-31:3197180:3197589 [0] NCCL INFO Channel 15/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-170-31:3197182:3197588 [2] NCCL INFO Channel 09/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-170-31:3197182:3197588 [2] NCCL INFO Channel 10/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-170-132:3330183:3330590 [0] NCCL INFO Channel 04/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-163-58:934720:935120 [0] NCCL INFO Channel 04/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-170-132:3330183:3330590 [0] NCCL INFO Channel 05/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-170-31:3197182:3197588 [2] NCCL INFO Channel 12/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-162-14:2525076:2525478 [1] NCCL INFO Connected all rings
+ip-26-0-170-31:3197182:3197588 [2] NCCL INFO Channel 13/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-163-58:934720:935120 [0] NCCL INFO Channel 05/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-170-132:3330183:3330590 [0] NCCL INFO Channel 06/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-170-31:3197182:3197588 [2] NCCL INFO Channel 14/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-163-58:934720:935120 [0] NCCL INFO Channel 06/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-163-58:934720:935120 [0] NCCL INFO Channel 07/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-170-31:3197182:3197588 [2] NCCL INFO Channel 15/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-170-132:3330183:3330590 [0] NCCL INFO Channel 07/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-163-58:934720:935120 [0] NCCL INFO Channel 10/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-170-132:3330183:3330590 [0] NCCL INFO Channel 10/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-163-58:934720:935120 [0] NCCL INFO Channel 11/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-170-132:3330183:3330590 [0] NCCL INFO Channel 11/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-163-58:934720:935120 [0] NCCL INFO Channel 12/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-170-132:3330183:3330590 [0] NCCL INFO Channel 12/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-170-132:3330183:3330590 [0] NCCL INFO Channel 13/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-170-132:3330183:3330590 [0] NCCL INFO Channel 14/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-170-31:3197185:3197591 [5] NCCL INFO Connected all rings
+ip-26-0-170-132:3330184:3330596 [1] NCCL INFO Channel 01/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-170-132:3330183:3330590 [0] NCCL INFO Channel 15/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-170-132:3330184:3330596 [1] NCCL INFO Channel 03/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-163-58:934721:935115 [1] NCCL INFO Channel 01/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-163-58:934720:935120 [0] NCCL INFO Channel 13/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-163-58:934721:935115 [1] NCCL INFO Channel 03/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-162-14:2525076:2525478 [1] NCCL INFO Channel 01/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-162-14:2525076:2525478 [1] NCCL INFO Channel 03/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-170-31:3197181:3197587 [1] NCCL INFO Connected all rings
+ip-26-0-163-58:934720:935120 [0] NCCL INFO Channel 14/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-170-132:3330184:3330596 [1] NCCL INFO Channel 04/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-163-58:934721:935115 [1] NCCL INFO Channel 04/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-170-132:3330184:3330596 [1] NCCL INFO Channel 05/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-163-58:934720:935120 [0] NCCL INFO Channel 15/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-163-58:934726:935119 [6] NCCL INFO Connected all rings
+ip-26-0-163-58:934721:935115 [1] NCCL INFO Channel 05/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-162-14:2525076:2525478 [1] NCCL INFO Channel 04/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-162-14:2525076:2525478 [1] NCCL INFO Channel 05/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-170-132:3330184:3330596 [1] NCCL INFO Channel 06/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-170-132:3330184:3330596 [1] NCCL INFO Channel 07/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-163-58:934721:935115 [1] NCCL INFO Channel 06/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-162-14:2525076:2525478 [1] NCCL INFO Channel 06/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-170-132:3330185:3330593 [2] NCCL INFO Connected all rings
+ip-26-0-163-58:934721:935115 [1] NCCL INFO Channel 07/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-170-132:3330184:3330596 [1] NCCL INFO Channel 09/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-162-14:2525076:2525478 [1] NCCL INFO Channel 07/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-162-14:2525079:2525477 [4] NCCL INFO Connected all rings
+ip-26-0-162-14:2525077:2525475 [2] NCCL INFO Connected all rings
+ip-26-0-163-58:934721:935115 [1] NCCL INFO Channel 09/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-162-14:2525076:2525478 [1] NCCL INFO Channel 09/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-170-31:3197181:3197587 [1] NCCL INFO Channel 01/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-163-58:934725:935118 [5] NCCL INFO Connected all rings
+ip-26-0-170-31:3197181:3197587 [1] NCCL INFO Channel 03/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-170-132:3330188:3330595 [5] NCCL INFO Connected all rings
+ip-26-0-170-31:3197181:3197587 [1] NCCL INFO Channel 04/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-162-14:2525076:2525478 [1] NCCL INFO Channel 11/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-162-14:2525075:2525471 [0] NCCL INFO Connected all rings
+ip-26-0-163-58:934721:935115 [1] NCCL INFO Channel 11/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-162-14:2525075:2525471 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-170-31:3197181:3197587 [1] NCCL INFO Channel 05/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-170-31:3197181:3197587 [1] NCCL INFO Channel 06/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-163-58:934727:935114 [7] NCCL INFO Connected all rings
+ip-26-0-162-14:2525076:2525478 [1] NCCL INFO Channel 12/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-170-31:3197181:3197587 [1] NCCL INFO Channel 07/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-163-58:934721:935115 [1] NCCL INFO Channel 12/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-170-132:3330184:3330596 [1] NCCL INFO Channel 11/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-170-31:3197181:3197587 [1] NCCL INFO Channel 09/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-162-14:2525075:2525471 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-162-14:2525076:2525478 [1] NCCL INFO Channel 13/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-170-31:3197181:3197587 [1] NCCL INFO Channel 11/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-162-14:2525082:2525472 [7] NCCL INFO Connected all rings
+ip-26-0-163-58:934721:935115 [1] NCCL INFO Channel 13/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-170-132:3330184:3330596 [1] NCCL INFO Channel 12/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-163-58:934722:935116 [2] NCCL INFO Channel 01/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-170-31:3197181:3197587 [1] NCCL INFO Channel 12/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-162-14:2525075:2525471 [0] NCCL INFO Channel 04/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-170-132:3330184:3330596 [1] NCCL INFO Channel 13/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-162-14:2525076:2525478 [1] NCCL INFO Channel 14/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-163-58:934721:935115 [1] NCCL INFO Channel 14/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-163-58:934722:935116 [2] NCCL INFO Channel 02/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-163-58:934724:935113 [4] NCCL INFO Connected all rings
+ip-26-0-170-132:3330184:3330596 [1] NCCL INFO Channel 14/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-162-14:2525075:2525471 [0] NCCL INFO Channel 05/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-170-31:3197181:3197587 [1] NCCL INFO Channel 13/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-163-58:934723:935117 [3] NCCL INFO Connected all rings
+ip-26-0-162-14:2525076:2525478 [1] NCCL INFO Channel 15/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-170-31:3197181:3197587 [1] NCCL INFO Channel 14/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-163-58:934721:935115 [1] NCCL INFO Channel 15/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-163-58:934722:935116 [2] NCCL INFO Channel 04/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-170-132:3330184:3330596 [1] NCCL INFO Channel 15/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-170-31:3197181:3197587 [1] NCCL INFO Channel 15/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-162-14:2525075:2525471 [0] NCCL INFO Channel 06/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-170-31:3197180:3197589 [0] NCCL INFO Channel 05/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-170-31:3197181:3197587 [1] NCCL INFO Channel 01/0 : 17[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-31:3197182:3197588 [2] NCCL INFO Channel 02/0 : 18[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-14:2525081:2525473 [6] NCCL INFO Connected all rings
+ip-26-0-170-31:3197180:3197589 [0] NCCL INFO Channel 06/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-163-58:934722:935116 [2] NCCL INFO Channel 05/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-170-132:3330183:3330590 [0] NCCL INFO Channel 05/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-163-58:934720:935120 [0] NCCL INFO Channel 05/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-162-14:2525075:2525471 [0] NCCL INFO Channel 07/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-162-14:2525080:2525476 [5] NCCL INFO Connected all rings
+ip-26-0-170-31:3197180:3197589 [0] NCCL INFO Channel 07/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-163-58:934722:935116 [2] NCCL INFO Channel 06/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-163-58:934720:935120 [0] NCCL INFO Channel 06/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-162-14:2525075:2525471 [0] NCCL INFO Channel 10/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-170-132:3330189:3330594 [6] NCCL INFO Connected all rings
+ip-26-0-170-132:3330183:3330590 [0] NCCL INFO Channel 06/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-163-58:934722:935116 [2] NCCL INFO Channel 07/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-162-14:2525075:2525471 [0] NCCL INFO Channel 11/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-163-58:934720:935120 [0] NCCL INFO Channel 07/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-163-58:934726:935119 [6] NCCL INFO Channel 01/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-170-132:3330187:3330591 [4] NCCL INFO Connected all rings
+ip-26-0-163-58:934722:935116 [2] NCCL INFO Channel 09/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-162-14:2525075:2525471 [0] NCCL INFO Channel 12/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-170-132:3330183:3330590 [0] NCCL INFO Channel 07/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-170-132:3330183:3330590 [0] NCCL INFO Channel 13/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-162-14:2525075:2525471 [0] NCCL INFO Channel 13/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-163-58:934720:935120 [0] NCCL INFO Channel 13/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-163-58:934726:935119 [6] NCCL INFO Channel 02/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-170-132:3330183:3330590 [0] NCCL INFO Channel 14/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-163-58:934720:935120 [0] NCCL INFO Channel 14/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-162-14:2525079:2525477 [4] NCCL INFO Channel 01/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-163-58:934726:935119 [6] NCCL INFO Channel 03/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-162-14:2525075:2525471 [0] NCCL INFO Channel 14/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-170-132:3330183:3330590 [0] NCCL INFO Channel 15/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-162-14:2525077:2525475 [2] NCCL INFO Channel 01/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-163-58:934720:935120 [0] NCCL INFO Channel 15/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-163-58:934726:935119 [6] NCCL INFO Channel 04/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-162-14:2525079:2525477 [4] NCCL INFO Channel 02/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-162-14:2525075:2525471 [0] NCCL INFO Channel 15/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-170-31:3197180:3197589 [0] NCCL INFO Channel 13/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-162-14:2525077:2525475 [2] NCCL INFO Channel 02/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-170-132:3330185:3330593 [2] NCCL INFO Channel 01/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-162-14:2525079:2525477 [4] NCCL INFO Channel 03/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-170-31:3197180:3197589 [0] NCCL INFO Channel 14/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-162-14:2525075:2525471 [0] NCCL INFO Channel 05/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-162-14:2525077:2525475 [2] NCCL INFO Channel 04/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-170-31:3197180:3197589 [0] NCCL INFO Channel 15/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-162-14:2525079:2525477 [4] NCCL INFO Channel 04/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-163-58:934726:935119 [6] NCCL INFO Channel 05/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-162-14:2525075:2525471 [0] NCCL INFO Channel 06/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-162-14:2525077:2525475 [2] NCCL INFO Channel 05/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-170-132:3330190:3330589 [7] NCCL INFO Connected all rings
+ip-26-0-170-132:3330185:3330593 [2] NCCL INFO Channel 02/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-163-58:934726:935119 [6] NCCL INFO Channel 06/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-163-58:934725:935118 [5] NCCL INFO Channel 01/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-162-14:2525079:2525477 [4] NCCL INFO Channel 06/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-162-14:2525075:2525471 [0] NCCL INFO Channel 07/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-163-58:934725:935118 [5] NCCL INFO Channel 02/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-163-58:934726:935119 [6] NCCL INFO Channel 09/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-162-14:2525077:2525475 [2] NCCL INFO Channel 06/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-170-132:3330185:3330593 [2] NCCL INFO Channel 04/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-162-14:2525079:2525477 [4] NCCL INFO Channel 07/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-162-14:2525075:2525471 [0] NCCL INFO Channel 13/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-162-14:2525077:2525475 [2] NCCL INFO Channel 07/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-170-132:3330185:3330593 [2] NCCL INFO Channel 05/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-162-14:2525079:2525477 [4] NCCL INFO Channel 09/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-163-58:934722:935116 [2] NCCL INFO Channel 10/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-170-31:3197184:3197586 [4] NCCL INFO Connected all rings
+ip-26-0-162-14:2525075:2525471 [0] NCCL INFO Channel 14/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-170-132:3330188:3330595 [5] NCCL INFO Channel 01/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-163-58:934725:935118 [5] NCCL INFO Channel 03/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-162-14:2525077:2525475 [2] NCCL INFO Channel 09/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-163-58:934722:935116 [2] NCCL INFO Channel 12/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-163-58:934726:935119 [6] NCCL INFO Channel 10/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-162-14:2525081:2525473 [6] NCCL INFO Channel 01/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-170-132:3330185:3330593 [2] NCCL INFO Channel 06/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-162-14:2525079:2525477 [4] NCCL INFO Channel 10/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-162-14:2525075:2525471 [0] NCCL INFO Channel 15/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-163-58:934725:935118 [5] NCCL INFO Channel 04/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-170-132:3330188:3330595 [5] NCCL INFO Channel 02/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-163-58:934722:935116 [2] NCCL INFO Channel 13/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-163-58:934726:935119 [6] NCCL INFO Channel 11/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-170-132:3330186:3330592 [3] NCCL INFO Channel 01/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-162-14:2525077:2525475 [2] NCCL INFO Channel 10/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-162-14:2525080:2525476 [5] NCCL INFO Channel 01/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-163-58:934725:935118 [5] NCCL INFO Channel 05/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-162-14:2525081:2525473 [6] NCCL INFO Channel 02/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-162-14:2525079:2525477 [4] NCCL INFO Channel 11/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-170-132:3330185:3330593 [2] NCCL INFO Channel 07/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-170-132:3330189:3330594 [6] NCCL INFO Channel 01/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-170-132:3330188:3330595 [5] NCCL INFO Channel 03/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-163-58:934722:935116 [2] NCCL INFO Channel 14/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-163-58:934726:935119 [6] NCCL INFO Channel 12/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-163-58:934724:935113 [4] NCCL INFO Channel 01/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-163-58:934725:935118 [5] NCCL INFO Channel 07/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-162-14:2525077:2525475 [2] NCCL INFO Channel 12/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-162-14:2525080:2525476 [5] NCCL INFO Channel 02/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-162-14:2525081:2525473 [6] NCCL INFO Channel 03/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-162-14:2525079:2525477 [4] NCCL INFO Channel 12/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-170-132:3330186:3330592 [3] NCCL INFO Channel 02/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-170-132:3330185:3330593 [2] NCCL INFO Channel 09/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-163-58:934722:935116 [2] NCCL INFO Channel 15/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-163-58:934726:935119 [6] NCCL INFO Channel 13/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-170-132:3330189:3330594 [6] NCCL INFO Channel 02/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-163-58:934724:935113 [4] NCCL INFO Channel 02/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-163-58:934725:935118 [5] NCCL INFO Channel 09/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-170-132:3330188:3330595 [5] NCCL INFO Channel 04/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-162-14:2525077:2525475 [2] NCCL INFO Channel 13/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-162-14:2525080:2525476 [5] NCCL INFO Channel 03/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-162-14:2525081:2525473 [6] NCCL INFO Channel 04/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-162-14:2525079:2525477 [4] NCCL INFO Channel 14/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-170-31:3197186:3197593 [6] NCCL INFO Connected all rings
+ip-26-0-163-58:934726:935119 [6] NCCL INFO Channel 14/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-162-14:2525077:2525475 [2] NCCL INFO Channel 14/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-170-132:3330186:3330592 [3] NCCL INFO Channel 03/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-170-132:3330187:3330591 [4] NCCL INFO Channel 01/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-170-132:3330185:3330593 [2] NCCL INFO Channel 10/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-163-58:934724:935113 [4] NCCL INFO Channel 03/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-163-58:934725:935118 [5] NCCL INFO Channel 10/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-163-58:934721:935115 [1] NCCL INFO Channel 09/0 : 1[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934721:935115 [1] NCCL INFO Channel 01/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330189:3330594 [6] NCCL INFO Channel 03/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-170-132:3330188:3330595 [5] NCCL INFO Channel 05/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-163-58:934721:935115 [1] NCCL INFO Channel 09/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-14:2525081:2525473 [6] NCCL INFO Channel 05/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-162-14:2525079:2525477 [4] NCCL INFO Channel 15/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-162-14:2525080:2525476 [5] NCCL INFO Channel 04/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-162-14:2525077:2525475 [2] NCCL INFO Channel 15/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-163-58:934724:935113 [4] NCCL INFO Channel 04/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-163-58:934725:935118 [5] NCCL INFO Channel 11/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-170-132:3330187:3330591 [4] NCCL INFO Channel 02/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-170-132:3330186:3330592 [3] NCCL INFO Channel 05/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-170-132:3330185:3330593 [2] NCCL INFO Channel 12/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-162-14:2525080:2525476 [5] NCCL INFO Channel 05/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-162-14:2525081:2525473 [6] NCCL INFO Channel 06/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-170-132:3330189:3330594 [6] NCCL INFO Channel 04/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-170-132:3330188:3330595 [5] NCCL INFO Channel 07/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-163-58:934724:935113 [4] NCCL INFO Channel 06/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-163-58:934725:935118 [5] NCCL INFO Channel 12/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-163-58:934723:935117 [3] NCCL INFO Channel 01/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-162-14:2525080:2525476 [5] NCCL INFO Channel 07/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-162-14:2525076:2525478 [1] NCCL INFO Channel 09/0 : 1[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-14:2525081:2525473 [6] NCCL INFO Channel 09/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-163-58:934724:935113 [4] NCCL INFO Channel 07/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-163-58:934725:935118 [5] NCCL INFO Channel 13/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-170-132:3330187:3330591 [4] NCCL INFO Channel 03/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-162-14:2525076:2525478 [1] NCCL INFO Channel 01/0 : 17[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330186:3330592 [3] NCCL INFO Channel 06/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-170-132:3330185:3330593 [2] NCCL INFO Channel 13/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-162-14:2525076:2525478 [1] NCCL INFO Channel 01/0 : 1[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-14:2525080:2525476 [5] NCCL INFO Channel 09/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-162-14:2525081:2525473 [6] NCCL INFO Channel 10/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-170-132:3330189:3330594 [6] NCCL INFO Channel 05/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-170-132:3330188:3330595 [5] NCCL INFO Channel 09/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-163-58:934723:935117 [3] NCCL INFO Channel 02/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-163-58:934724:935113 [4] NCCL INFO Channel 09/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-163-58:934725:935118 [5] NCCL INFO Channel 15/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-162-14:2525080:2525476 [5] NCCL INFO Channel 10/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-162-14:2525081:2525473 [6] NCCL INFO Channel 11/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-163-58:934723:935117 [3] NCCL INFO Channel 03/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-170-132:3330187:3330591 [4] NCCL INFO Channel 04/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-162-14:2525080:2525476 [5] NCCL INFO Channel 11/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-162-14:2525081:2525473 [6] NCCL INFO Channel 12/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-170-132:3330186:3330592 [3] NCCL INFO Channel 07/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-170-132:3330185:3330593 [2] NCCL INFO Channel 14/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-170-132:3330189:3330594 [6] NCCL INFO Channel 06/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-170-132:3330188:3330595 [5] NCCL INFO Channel 10/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-170-31:3197184:3197586 [4] NCCL INFO Channel 01/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-163-58:934724:935113 [4] NCCL INFO Channel 10/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-163-58:934723:935117 [3] NCCL INFO Channel 05/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-162-14:2525080:2525476 [5] NCCL INFO Channel 12/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-163-58:934726:935119 [6] NCCL INFO Channel 14/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-14:2525081:2525473 [6] NCCL INFO Channel 13/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-163-58:934726:935119 [6] NCCL INFO Channel 06/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934726:935119 [6] NCCL INFO Channel 14/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934724:935113 [4] NCCL INFO Channel 11/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-162-14:2525080:2525476 [5] NCCL INFO Channel 13/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-162-14:2525081:2525473 [6] NCCL INFO Channel 14/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-163-58:934723:935117 [3] NCCL INFO Channel 06/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-170-132:3330187:3330591 [4] NCCL INFO Channel 06/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-170-132:3330186:3330592 [3] NCCL INFO Channel 09/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-170-132:3330185:3330593 [2] NCCL INFO Channel 15/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-170-132:3330189:3330594 [6] NCCL INFO Channel 09/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-163-58:934724:935113 [4] NCCL INFO Channel 12/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-170-132:3330188:3330595 [5] NCCL INFO Channel 11/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-163-58:934723:935117 [3] NCCL INFO Channel 07/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-162-14:2525080:2525476 [5] NCCL INFO Channel 15/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-170-31:3197187:3197590 [7] NCCL INFO Connected all rings
+ip-26-0-163-58:934727:935114 [7] NCCL INFO Channel 15/0 : 7[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934720:935120 [0] NCCL INFO Channel 08/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934724:935113 [4] NCCL INFO Channel 14/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-163-58:934727:935114 [7] NCCL INFO Channel 07/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934720:935120 [0] NCCL INFO Channel 00/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934727:935114 [7] NCCL INFO Channel 15/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934723:935117 [3] NCCL INFO Channel 09/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-163-58:934720:935120 [0] NCCL INFO Channel 08/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-31:3197184:3197586 [4] NCCL INFO Channel 02/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-163-58:934724:935113 [4] NCCL INFO Channel 15/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-162-14:2525081:2525473 [6] NCCL INFO Channel 14/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934723:935117 [3] NCCL INFO Channel 10/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-162-14:2525080:2525476 [5] NCCL INFO Channel 13/0 : 5[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330187:3330591 [4] NCCL INFO Channel 07/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-170-132:3330186:3330592 [3] NCCL INFO Channel 10/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-170-132:3330189:3330594 [6] NCCL INFO Channel 10/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-170-132:3330188:3330595 [5] NCCL INFO Channel 12/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-162-14:2525081:2525473 [6] NCCL INFO Channel 06/0 : 22[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-31:3197184:3197586 [4] NCCL INFO Channel 03/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-163-58:934723:935117 [3] NCCL INFO Channel 11/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-162-14:2525081:2525473 [6] NCCL INFO Channel 06/0 : 6[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330184:3330596 [1] NCCL INFO Channel 01/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330184:3330596 [1] NCCL INFO Channel 09/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934725:935118 [5] NCCL INFO Channel 13/0 : 5[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330184:3330596 [1] NCCL INFO Channel 01/0 : 17[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934723:935117 [3] NCCL INFO Channel 13/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-170-31:3197185:3197591 [5] NCCL INFO Channel 01/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-163-58:934725:935118 [5] NCCL INFO Channel 05/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934725:935118 [5] NCCL INFO Channel 13/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934723:935117 [3] NCCL INFO Channel 14/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-163-58:934721:935115 [1] NCCL INFO Channel 09/0 : 25[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330187:3330591 [4] NCCL INFO Channel 09/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-163-58:934721:935115 [1] NCCL INFO Channel 09/0 : 9[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-14:2525080:2525476 [5] NCCL INFO Channel 05/0 : 21[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330184:3330596 [1] NCCL INFO Channel 01/0 : 1[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330186:3330592 [3] NCCL INFO Channel 11/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-162-14:2525080:2525476 [5] NCCL INFO Channel 05/0 : 5[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330189:3330594 [6] NCCL INFO Channel 11/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-163-58:934723:935117 [3] NCCL INFO Channel 15/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-170-132:3330184:3330596 [1] NCCL INFO Channel 01/0 : 17[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330188:3330595 [5] NCCL INFO Channel 13/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-170-132:3330184:3330596 [1] NCCL INFO Channel 01/0 : 25[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-14:2525076:2525478 [1] NCCL INFO Channel 09/0 : 9[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330184:3330596 [1] NCCL INFO Channel 01/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330184:3330596 [1] NCCL INFO Channel 09/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-31:3197184:3197586 [4] NCCL INFO Channel 04/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-162-14:2525082:2525472 [7] NCCL INFO Channel 15/0 : 7[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-14:2525075:2525471 [0] NCCL INFO Channel 08/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934722:935116 [2] NCCL INFO Channel 10/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330187:3330591 [4] NCCL INFO Channel 10/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-170-132:3330186:3330592 [3] NCCL INFO Channel 13/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-163-58:934724:935113 [4] NCCL INFO Channel 12/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-132:3330189:3330594 [6] NCCL INFO Channel 12/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-163-58:934722:935116 [2] NCCL INFO Channel 02/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-31:3197181:3197587 [1] NCCL INFO Channel 09/0 : 9[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934723:935117 [3] NCCL INFO Channel 11/0 : 3[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330188:3330595 [5] NCCL INFO Channel 15/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-163-58:934724:935113 [4] NCCL INFO Channel 04/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-31:3197181:3197587 [1] NCCL INFO Channel 09/0 : 25[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-14:2525075:2525471 [0] NCCL INFO Channel 00/0 : 16[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934722:935116 [2] NCCL INFO Channel 10/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934723:935117 [3] NCCL INFO Channel 03/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-14:2525082:2525472 [7] NCCL INFO Channel 07/0 : 23[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-14:2525075:2525471 [0] NCCL INFO Channel 00/0 : 0[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934724:935113 [4] NCCL INFO Channel 12/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934723:935117 [3] NCCL INFO Channel 11/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-14:2525082:2525472 [7] NCCL INFO Channel 07/0 : 7[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-31:3197185:3197591 [5] NCCL INFO Channel 02/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-170-31:3197181:3197587 [1] NCCL INFO Channel 01/0 : 25[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934721:935115 [1] NCCL INFO Channel 01/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-31:3197184:3197586 [4] NCCL INFO Channel 06/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-163-58:934721:935115 [1] NCCL INFO Channel 09/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934721:935115 [1] NCCL INFO Channel 09/0 : 9[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330187:3330591 [4] NCCL INFO Channel 11/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-170-132:3330186:3330592 [3] NCCL INFO Channel 14/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-170-132:3330189:3330594 [6] NCCL INFO Channel 13/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-170-31:3197185:3197591 [5] NCCL INFO Channel 03/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-170-31:3197184:3197586 [4] NCCL INFO Channel 07/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-170-132:3330187:3330591 [4] NCCL INFO Channel 12/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-170-132:3330189:3330594 [6] NCCL INFO Channel 14/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-170-132:3330186:3330592 [3] NCCL INFO Channel 15/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-170-31:3197185:3197591 [5] NCCL INFO Channel 04/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-170-31:3197184:3197586 [4] NCCL INFO Channel 09/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-170-132:3330187:3330591 [4] NCCL INFO Channel 14/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-162-14:2525076:2525478 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330185:3330593 [2] NCCL INFO Channel 02/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330189:3330594 [6] NCCL INFO Channel 06/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330185:3330593 [2] NCCL INFO Channel 10/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330187:3330591 [4] NCCL INFO Channel 15/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-162-14:2525078:2525474 [3] NCCL INFO Connected all rings
+ip-26-0-162-14:2525076:2525478 [1] NCCL INFO Channel 08/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330189:3330594 [6] NCCL INFO Channel 14/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330185:3330593 [2] NCCL INFO Channel 02/0 : 18[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-31:3197184:3197586 [4] NCCL INFO Channel 10/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-163-58:934721:935115 [1] NCCL INFO Channel 00/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-170-132:3330189:3330594 [6] NCCL INFO Channel 06/0 : 22[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-31:3197185:3197591 [5] NCCL INFO Channel 05/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-163-58:934721:935115 [1] NCCL INFO Channel 08/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-163-58:934726:935119 [6] NCCL INFO Channel 14/0 : 30[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330184:3330596 [1] NCCL INFO Channel 00/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-170-132:3330185:3330593 [2] NCCL INFO Channel 02/0 : 2[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934726:935119 [6] NCCL INFO Channel 14/0 : 14[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330185:3330593 [2] NCCL INFO Channel 02/0 : 18[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330186:3330592 [3] NCCL INFO Channel 03/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330184:3330596 [1] NCCL INFO Channel 08/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-170-132:3330188:3330595 [5] NCCL INFO Channel 05/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-31:3197184:3197586 [4] NCCL INFO Channel 11/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-170-31:3197185:3197591 [5] NCCL INFO Channel 07/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-170-132:3330186:3330592 [3] NCCL INFO Channel 11/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330187:3330591 [4] NCCL INFO Channel 04/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-132:3330188:3330595 [5] NCCL INFO Channel 13/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330186:3330592 [3] NCCL INFO Channel 03/0 : 19[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330187:3330591 [4] NCCL INFO Channel 12/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-132:3330188:3330595 [5] NCCL INFO Channel 05/0 : 21[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330187:3330591 [4] NCCL INFO Channel 04/0 : 20[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934725:935118 [5] NCCL INFO Channel 13/0 : 29[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934725:935118 [5] NCCL INFO Channel 13/0 : 13[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-31:3197184:3197586 [4] NCCL INFO Channel 12/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-170-31:3197185:3197591 [5] NCCL INFO Channel 09/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-170-31:3197182:3197588 [2] NCCL INFO Channel 10/0 : 10[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-31:3197182:3197588 [2] NCCL INFO Channel 10/0 : 26[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-31:3197185:3197591 [5] NCCL INFO Channel 10/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-170-31:3197184:3197586 [4] NCCL INFO Channel 14/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-170-31:3197186:3197593 [6] NCCL INFO Channel 01/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-170-31:3197184:3197586 [4] NCCL INFO Channel 15/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-170-31:3197185:3197591 [5] NCCL INFO Channel 11/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-170-132:3330190:3330589 [7] NCCL INFO Channel 07/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-14:2525078:2525474 [3] NCCL INFO Channel 01/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-170-132:3330183:3330590 [0] NCCL INFO Channel 00/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330190:3330589 [7] NCCL INFO Channel 15/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330183:3330590 [0] NCCL INFO Channel 08/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-31:3197186:3197593 [6] NCCL INFO Channel 02/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-170-132:3330190:3330589 [7] NCCL INFO Channel 07/0 : 23[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330183:3330590 [0] NCCL INFO Channel 00/0 : 16[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-14:2525078:2525474 [3] NCCL INFO Channel 02/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-170-31:3197181:3197587 [1] NCCL INFO Channel 00/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-163-58:934727:935114 [7] NCCL INFO Channel 15/0 : 31[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-14:2525078:2525474 [3] NCCL INFO Channel 03/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-163-58:934720:935120 [0] NCCL INFO Channel 08/0 : 24[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934727:935114 [7] NCCL INFO Channel 15/0 : 15[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934720:935120 [0] NCCL INFO Channel 08/0 : 8[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-31:3197185:3197591 [5] NCCL INFO Channel 12/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-162-14:2525078:2525474 [3] NCCL INFO Channel 05/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-170-31:3197186:3197593 [6] NCCL INFO Channel 03/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-170-31:3197181:3197587 [1] NCCL INFO Channel 08/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-170-31:3197183:3197592 [3] NCCL INFO Channel 03/0 : 19[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-14:2525078:2525474 [3] NCCL INFO Channel 06/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-170-31:3197185:3197591 [5] NCCL INFO Channel 13/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-170-31:3197183:3197592 [3] NCCL INFO Channel 11/0 : 11[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330186:3330592 [3] NCCL INFO Channel 03/0 : 3[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-14:2525078:2525474 [3] NCCL INFO Channel 07/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-170-31:3197183:3197592 [3] NCCL INFO Channel 11/0 : 27[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330186:3330592 [3] NCCL INFO Channel 03/0 : 19[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-14:2525078:2525474 [3] NCCL INFO Channel 09/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-170-31:3197186:3197593 [6] NCCL INFO Channel 04/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-170-31:3197185:3197591 [5] NCCL INFO Channel 15/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-162-14:2525078:2525474 [3] NCCL INFO Channel 10/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-170-31:3197186:3197593 [6] NCCL INFO Channel 05/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-162-14:2525078:2525474 [3] NCCL INFO Channel 11/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-170-31:3197186:3197593 [6] NCCL INFO Channel 06/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-170-31:3197184:3197586 [4] NCCL INFO Channel 04/0 : 20[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-14:2525078:2525474 [3] NCCL INFO Channel 13/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-170-31:3197186:3197593 [6] NCCL INFO Channel 09/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-170-31:3197184:3197586 [4] NCCL INFO Channel 12/0 : 12[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-14:2525078:2525474 [3] NCCL INFO Channel 14/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-170-132:3330187:3330591 [4] NCCL INFO Channel 04/0 : 4[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-31:3197184:3197586 [4] NCCL INFO Channel 12/0 : 28[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-132:3330187:3330591 [4] NCCL INFO Channel 04/0 : 20[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-31:3197186:3197593 [6] NCCL INFO Channel 10/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-162-14:2525078:2525474 [3] NCCL INFO Channel 15/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-170-31:3197186:3197593 [6] NCCL INFO Channel 11/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-170-31:3197186:3197593 [6] NCCL INFO Channel 12/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-162-14:2525077:2525475 [2] NCCL INFO Channel 10/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-31:3197186:3197593 [6] NCCL INFO Channel 13/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-162-14:2525079:2525477 [4] NCCL INFO Channel 12/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-14:2525078:2525474 [3] NCCL INFO Channel 11/0 : 3[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-14:2525077:2525475 [2] NCCL INFO Channel 02/0 : 18[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-31:3197186:3197593 [6] NCCL INFO Channel 14/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-162-14:2525077:2525475 [2] NCCL INFO Channel 02/0 : 2[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-14:2525079:2525477 [4] NCCL INFO Channel 04/0 : 20[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-14:2525078:2525474 [3] NCCL INFO Channel 03/0 : 19[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934722:935116 [2] NCCL INFO Channel 10/0 : 26[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934724:935113 [4] NCCL INFO Channel 12/0 : 28[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934723:935117 [3] NCCL INFO Channel 11/0 : 27[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-14:2525078:2525474 [3] NCCL INFO Channel 03/0 : 3[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-14:2525079:2525477 [4] NCCL INFO Channel 04/0 : 4[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934722:935116 [2] NCCL INFO Channel 10/0 : 10[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934724:935113 [4] NCCL INFO Channel 12/0 : 12[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-14:2525077:2525475 [2] NCCL INFO Channel 10/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934723:935117 [3] NCCL INFO Channel 11/0 : 11[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330185:3330593 [2] NCCL INFO Channel 02/0 : 26[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934722:935116 [2] NCCL INFO Channel 02/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-14:2525078:2525474 [3] NCCL INFO Channel 11/0 : 11[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-14:2525079:2525477 [4] NCCL INFO Channel 12/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-132:3330185:3330593 [2] NCCL INFO Channel 02/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330186:3330592 [3] NCCL INFO Channel 03/0 : 27[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934724:935113 [4] NCCL INFO Channel 04/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-31:3197184:3197586 [4] NCCL INFO Channel 04/0 : 28[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934723:935117 [3] NCCL INFO Channel 03/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330185:3330593 [2] NCCL INFO Channel 10/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-31:3197185:3197591 [5] NCCL INFO Channel 05/0 : 21[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-31:3197183:3197592 [3] NCCL INFO Channel 03/0 : 27[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-31:3197182:3197588 [2] NCCL INFO Channel 02/0 : 26[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-31:3197186:3197593 [6] NCCL INFO Channel 06/0 : 22[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934722:935116 [2] NCCL INFO Channel 10/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330186:3330592 [3] NCCL INFO Channel 03/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934724:935113 [4] NCCL INFO Channel 12/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934723:935117 [3] NCCL INFO Channel 11/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934722:935116 [2] NCCL INFO Channel 10/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934724:935113 [4] NCCL INFO Channel 12/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934723:935117 [3] NCCL INFO Channel 11/0 : 11[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330186:3330592 [3] NCCL INFO Channel 11/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330187:3330591 [4] NCCL INFO Channel 04/0 : 28[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-31:3197185:3197591 [5] NCCL INFO Channel 13/0 : 13[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330188:3330595 [5] NCCL INFO Channel 05/0 : 5[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-31:3197186:3197593 [6] NCCL INFO Channel 14/0 : 14[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330187:3330591 [4] NCCL INFO Channel 04/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-31:3197185:3197591 [5] NCCL INFO Channel 13/0 : 29[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330189:3330594 [6] NCCL INFO Channel 06/0 : 6[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330188:3330595 [5] NCCL INFO Channel 05/0 : 21[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-31:3197186:3197593 [6] NCCL INFO Channel 14/0 : 30[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330187:3330591 [4] NCCL INFO Channel 12/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-132:3330189:3330594 [6] NCCL INFO Channel 06/0 : 22[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-31:3197185:3197591 [5] NCCL INFO Channel 05/0 : 29[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330188:3330595 [5] NCCL INFO Channel 05/0 : 29[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934725:935118 [5] NCCL INFO Channel 05/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-31:3197186:3197593 [6] NCCL INFO Channel 06/0 : 30[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-14:2525080:2525476 [5] NCCL INFO Channel 13/0 : 13[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934725:935118 [5] NCCL INFO Channel 13/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330189:3330594 [6] NCCL INFO Channel 06/0 : 30[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330188:3330595 [5] NCCL INFO Channel 05/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-14:2525081:2525473 [6] NCCL INFO Channel 14/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934726:935119 [6] NCCL INFO Channel 06/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330188:3330595 [5] NCCL INFO Channel 13/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330189:3330594 [6] NCCL INFO Channel 06/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934725:935118 [5] NCCL INFO Channel 13/0 : 13[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330189:3330594 [6] NCCL INFO Channel 14/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934726:935119 [6] NCCL INFO Channel 14/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934726:935119 [6] NCCL INFO Channel 14/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-31:3197182:3197588 [2] NCCL INFO Channel 00/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-162-14:2525079:2525477 [4] NCCL INFO Channel 00/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-162-14:2525077:2525475 [2] NCCL INFO Channel 00/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-163-58:934722:935116 [2] NCCL INFO Channel 00/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-170-31:3197187:3197590 [7] NCCL INFO Channel 07/0 : 23[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-31:3197180:3197589 [0] NCCL INFO Channel 00/0 : 16[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-31:3197184:3197586 [4] NCCL INFO Channel 00/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-163-58:934724:935113 [4] NCCL INFO Channel 00/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-170-31:3197182:3197588 [2] NCCL INFO Channel 01/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-170-31:3197187:3197590 [7] NCCL INFO Channel 15/0 : 15[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330190:3330589 [7] NCCL INFO Channel 07/0 : 7[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-31:3197187:3197590 [7] NCCL INFO Channel 15/0 : 31[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330190:3330589 [7] NCCL INFO Channel 07/0 : 23[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-31:3197180:3197589 [0] NCCL INFO Channel 08/0 : 8[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330183:3330590 [0] NCCL INFO Channel 00/0 : 0[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-14:2525079:2525477 [4] NCCL INFO Channel 03/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-170-31:3197180:3197589 [0] NCCL INFO Channel 08/0 : 24[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934722:935116 [2] NCCL INFO Channel 01/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-170-132:3330183:3330590 [0] NCCL INFO Channel 00/0 : 16[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-31:3197186:3197593 [6] NCCL INFO Channel 00/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-170-31:3197187:3197590 [7] NCCL INFO Channel 07/0 : 31[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934727:935114 [7] NCCL INFO Channel 07/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330190:3330589 [7] NCCL INFO Channel 07/0 : 31[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-31:3197180:3197589 [0] NCCL INFO Channel 00/0 : 24[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-14:2525082:2525472 [7] NCCL INFO Channel 15/0 : 15[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934727:935114 [7] NCCL INFO Channel 15/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330190:3330589 [7] NCCL INFO Channel 07/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934720:935120 [0] NCCL INFO Channel 00/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330183:3330590 [0] NCCL INFO Channel 00/0 : 24[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-14:2525075:2525471 [0] NCCL INFO Channel 08/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934727:935114 [7] NCCL INFO Channel 15/0 : 15[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934720:935120 [0] NCCL INFO Channel 08/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330190:3330589 [7] NCCL INFO Channel 15/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934720:935120 [0] NCCL INFO Channel 08/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330183:3330590 [0] NCCL INFO Channel 00/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-31:3197184:3197586 [4] NCCL INFO Channel 03/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-163-58:934725:935118 [5] NCCL INFO Channel 00/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-163-58:934724:935113 [4] NCCL INFO Channel 03/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-170-132:3330183:3330590 [0] NCCL INFO Channel 08/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-31:3197187:3197590 [7] NCCL INFO Channel 01/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-170-31:3197182:3197588 [2] NCCL INFO Channel 08/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-163-58:934726:935119 [6] NCCL INFO Channel 00/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-162-14:2525082:2525472 [7] NCCL INFO Channel 01/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330190:3330589 [7] NCCL INFO Channel 01/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-163-58:934727:935114 [7] NCCL INFO Channel 01/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-170-31:3197186:3197593 [6] NCCL INFO Channel 05/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-163-58:934722:935116 [2] NCCL INFO Channel 08/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-163-58:934724:935113 [4] NCCL INFO Channel 05/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-163-58:934725:935118 [5] NCCL INFO Channel 04/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-170-31:3197184:3197586 [4] NCCL INFO Channel 05/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-170-132:3330185:3330593 [2] NCCL INFO Channel 00/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-163-58:934726:935119 [6] NCCL INFO Channel 05/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-170-31:3197187:3197590 [7] NCCL INFO Channel 02/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-170-31:3197182:3197588 [2] NCCL INFO Channel 09/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-170-132:3330188:3330595 [5] NCCL INFO Channel 00/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-170-132:3330190:3330589 [7] NCCL INFO Channel 02/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-163-58:934727:935114 [7] NCCL INFO Channel 02/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-170-31:3197186:3197593 [6] NCCL INFO Channel 08/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-163-58:934722:935116 [2] NCCL INFO Channel 09/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-170-132:3330189:3330594 [6] NCCL INFO Channel 00/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-170-31:3197185:3197591 [5] NCCL INFO Channel 00/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-163-58:934724:935113 [4] NCCL INFO Channel 06/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-170-31:3197184:3197586 [4] NCCL INFO Channel 06/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-163-58:934725:935118 [5] NCCL INFO Channel 08/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-170-132:3330185:3330593 [2] NCCL INFO Channel 01/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-163-58:934726:935119 [6] NCCL INFO Channel 08/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-170-31:3197187:3197590 [7] NCCL INFO Channel 03/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-170-132:3330188:3330595 [5] NCCL INFO Channel 04/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-170-132:3330190:3330589 [7] NCCL INFO Channel 03/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-163-58:934727:935114 [7] NCCL INFO Channel 03/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-170-31:3197186:3197593 [6] NCCL INFO Channel 13/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-170-132:3330189:3330594 [6] NCCL INFO Channel 05/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-170-31:3197185:3197591 [5] NCCL INFO Channel 04/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-170-31:3197184:3197586 [4] NCCL INFO Channel 07/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-163-58:934724:935113 [4] NCCL INFO Channel 07/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-163-58:934725:935118 [5] NCCL INFO Channel 12/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-162-14:2525077:2525475 [2] NCCL INFO Channel 01/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-162-14:2525080:2525476 [5] NCCL INFO Channel 00/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-163-58:934726:935119 [6] NCCL INFO Channel 13/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-170-31:3197187:3197590 [7] NCCL INFO Channel 04/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-163-58:934727:935114 [7] NCCL INFO Channel 04/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-170-132:3330188:3330595 [5] NCCL INFO Channel 08/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-170-132:3330185:3330593 [2] NCCL INFO Channel 08/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-170-132:3330190:3330589 [7] NCCL INFO Channel 04/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-162-14:2525081:2525473 [6] NCCL INFO Channel 00/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-170-31:3197185:3197591 [5] NCCL INFO Channel 08/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-170-132:3330189:3330594 [6] NCCL INFO Channel 08/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-170-31:3197184:3197586 [4] NCCL INFO Channel 08/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-170-132:3330187:3330591 [4] NCCL INFO Channel 00/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-162-14:2525077:2525475 [2] NCCL INFO Channel 08/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-170-31:3197187:3197590 [7] NCCL INFO Channel 05/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-162-14:2525082:2525472 [7] NCCL INFO Channel 02/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-162-14:2525081:2525473 [6] NCCL INFO Channel 05/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-163-58:934724:935113 [4] NCCL INFO Channel 08/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-170-31:3197185:3197591 [5] NCCL INFO Channel 12/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-170-31:3197184:3197586 [4] NCCL INFO Channel 11/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-170-31:3197187:3197590 [7] NCCL INFO Channel 06/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-170-132:3330185:3330593 [2] NCCL INFO Channel 09/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-170-132:3330190:3330589 [7] NCCL INFO Channel 05/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-163-58:934723:935117 [3] NCCL INFO Channel 00/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-170-132:3330188:3330595 [5] NCCL INFO Channel 12/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-163-58:934724:935113 [4] NCCL INFO Channel 11/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-170-132:3330189:3330594 [6] NCCL INFO Channel 13/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-170-132:3330187:3330591 [4] NCCL INFO Channel 03/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-170-31:3197184:3197586 [4] NCCL INFO Channel 13/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-170-31:3197187:3197590 [7] NCCL INFO Channel 07/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-163-58:934727:935114 [7] NCCL INFO Channel 05/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-163-58:934723:935117 [3] NCCL INFO Channel 02/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-163-58:934724:935113 [4] NCCL INFO Channel 13/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-170-31:3197183:3197592 [3] NCCL INFO Channel 00/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-162-14:2525077:2525475 [2] NCCL INFO Channel 09/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-170-31:3197187:3197590 [7] NCCL INFO Channel 09/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-170-132:3330190:3330589 [7] NCCL INFO Channel 06/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-163-58:934723:935117 [3] NCCL INFO Channel 08/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-163-58:934727:935114 [7] NCCL INFO Channel 06/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-170-31:3197184:3197586 [4] NCCL INFO Channel 14/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-163-58:934724:935113 [4] NCCL INFO Channel 14/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-162-14:2525079:2525477 [4] NCCL INFO Channel 05/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-170-132:3330187:3330591 [4] NCCL INFO Channel 05/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-162-14:2525080:2525476 [5] NCCL INFO Channel 04/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-170-31:3197183:3197592 [3] NCCL INFO Channel 02/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-170-31:3197187:3197590 [7] NCCL INFO Channel 10/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-170-31:3197184:3197586 [4] NCCL INFO Channel 15/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-162-14:2525079:2525477 [4] NCCL INFO Channel 06/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-162-14:2525080:2525476 [5] NCCL INFO Channel 08/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-170-132:3330187:3330591 [4] NCCL INFO Channel 06/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-170-132:3330190:3330589 [7] NCCL INFO Channel 07/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-170-31:3197183:3197592 [3] NCCL INFO Channel 08/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-170-132:3330186:3330592 [3] NCCL INFO Channel 00/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-170-31:3197187:3197590 [7] NCCL INFO Channel 11/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-170-132:3330187:3330591 [4] NCCL INFO Channel 07/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-170-132:3330190:3330589 [7] NCCL INFO Channel 09/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-162-14:2525079:2525477 [4] NCCL INFO Channel 07/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-162-14:2525080:2525476 [5] NCCL INFO Channel 12/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-163-58:934723:935117 [3] NCCL INFO Channel 10/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-170-31:3197183:3197592 [3] NCCL INFO Channel 10/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-170-132:3330186:3330592 [3] NCCL INFO Channel 02/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-163-58:934724:935113 [4] NCCL INFO Channel 15/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-163-58:934727:935114 [7] NCCL INFO Channel 07/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-162-14:2525079:2525477 [4] NCCL INFO Channel 08/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-170-132:3330190:3330589 [7] NCCL INFO Channel 10/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-170-132:3330187:3330591 [4] NCCL INFO Channel 08/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-170-132:3330186:3330592 [3] NCCL INFO Channel 08/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-162-14:2525079:2525477 [4] NCCL INFO Channel 11/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-162-14:2525082:2525472 [7] NCCL INFO Channel 03/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330190:3330589 [7] NCCL INFO Channel 11/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-170-132:3330187:3330591 [4] NCCL INFO Channel 11/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-162-14:2525078:2525474 [3] NCCL INFO Channel 00/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-170-31:3197187:3197590 [7] NCCL INFO Channel 12/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-162-14:2525079:2525477 [4] NCCL INFO Channel 13/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-162-14:2525082:2525472 [7] NCCL INFO Channel 04/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330186:3330592 [3] NCCL INFO Channel 10/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-170-132:3330190:3330589 [7] NCCL INFO Channel 12/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-170-132:3330187:3330591 [4] NCCL INFO Channel 13/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-162-14:2525078:2525474 [3] NCCL INFO Channel 02/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-170-132:3330187:3330591 [4] NCCL INFO Channel 14/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-162-14:2525081:2525473 [6] NCCL INFO Channel 08/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-170-132:3330187:3330591 [4] NCCL INFO Channel 15/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-162-14:2525079:2525477 [4] NCCL INFO Channel 14/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-162-14:2525078:2525474 [3] NCCL INFO Channel 08/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-162-14:2525081:2525473 [6] NCCL INFO Channel 13/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-170-31:3197187:3197590 [7] NCCL INFO Channel 13/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-170-132:3330190:3330589 [7] NCCL INFO Channel 13/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-162-14:2525079:2525477 [4] NCCL INFO Channel 15/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-162-14:2525078:2525474 [3] NCCL INFO Channel 10/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-163-58:934727:935114 [7] NCCL INFO Channel 09/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-162-14:2525082:2525472 [7] NCCL INFO Channel 05/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330190:3330589 [7] NCCL INFO Channel 14/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-163-58:934727:935114 [7] NCCL INFO Channel 10/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-162-14:2525082:2525472 [7] NCCL INFO Channel 06/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-170-31:3197187:3197590 [7] NCCL INFO Channel 14/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-170-132:3330190:3330589 [7] NCCL INFO Channel 15/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-163-58:934727:935114 [7] NCCL INFO Channel 11/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-162-14:2525082:2525472 [7] NCCL INFO Channel 07/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-170-31:3197187:3197590 [7] NCCL INFO Channel 15/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-163-58:934727:935114 [7] NCCL INFO Channel 12/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-170-132:3330190:3330589 [7] NCCL INFO Channel 00/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-163-58:934727:935114 [7] NCCL INFO Channel 13/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-162-14:2525082:2525472 [7] NCCL INFO Channel 09/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330190:3330589 [7] NCCL INFO Channel 06/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-163-58:934727:935114 [7] NCCL INFO Channel 14/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-162-14:2525082:2525472 [7] NCCL INFO Channel 10/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-163-58:934727:935114 [7] NCCL INFO Channel 15/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-170-132:3330190:3330589 [7] NCCL INFO Channel 08/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-170-132:3330190:3330589 [7] NCCL INFO Channel 14/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-170-31:3197187:3197590 [7] NCCL INFO Channel 00/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-170-31:3197187:3197590 [7] NCCL INFO Channel 06/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-162-14:2525082:2525472 [7] NCCL INFO Channel 11/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-170-31:3197187:3197590 [7] NCCL INFO Channel 08/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-162-14:2525082:2525472 [7] NCCL INFO Channel 12/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-170-31:3197187:3197590 [7] NCCL INFO Channel 14/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-162-14:2525082:2525472 [7] NCCL INFO Channel 13/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-162-14:2525082:2525472 [7] NCCL INFO Channel 14/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-163-58:934727:935114 [7] NCCL INFO Channel 00/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-162-14:2525082:2525472 [7] NCCL INFO Channel 15/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-163-58:934727:935114 [7] NCCL INFO Channel 06/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-162-14:2525082:2525472 [7] NCCL INFO Channel 00/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-163-58:934727:935114 [7] NCCL INFO Channel 08/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-162-14:2525082:2525472 [7] NCCL INFO Channel 06/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-163-58:934727:935114 [7] NCCL INFO Channel 14/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-162-14:2525082:2525472 [7] NCCL INFO Channel 08/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-162-14:2525082:2525472 [7] NCCL INFO Channel 14/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-170-31:3197182:3197588 [2] NCCL INFO Connected all trees
+ip-26-0-170-31:3197182:3197588 [2] NCCL INFO NVLS comm 0x88fe880 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-170-31:3197185:3197591 [5] NCCL INFO Connected all trees
+ip-26-0-170-31:3197185:3197591 [5] NCCL INFO NVLS comm 0x884f520 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-170-31:3197180:3197589 [0] NCCL INFO Connected all trees
+ip-26-0-170-31:3197180:3197589 [0] NCCL INFO NVLS comm 0xa19e6a0 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-170-31:3197183:3197592 [3] NCCL INFO Connected all trees
+ip-26-0-170-31:3197183:3197592 [3] NCCL INFO NVLS comm 0x9f6fc60 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-170-31:3197184:3197586 [4] NCCL INFO Connected all trees
+ip-26-0-170-31:3197184:3197586 [4] NCCL INFO NVLS comm 0xa1031e0 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-170-31:3197187:3197590 [7] NCCL INFO Connected all trees
+ip-26-0-170-31:3197187:3197590 [7] NCCL INFO NVLS comm 0xa665180 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-162-14:2525080:2525476 [5] NCCL INFO Connected all trees
+ip-26-0-162-14:2525080:2525476 [5] NCCL INFO NVLS comm 0x9c491a0 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-162-14:2525078:2525474 [3] NCCL INFO Connected all trees
+ip-26-0-162-14:2525077:2525475 [2] NCCL INFO Connected all trees
+ip-26-0-162-14:2525079:2525477 [4] NCCL INFO Connected all trees
+ip-26-0-162-14:2525078:2525474 [3] NCCL INFO NVLS comm 0x9224d20 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-170-31:3197186:3197593 [6] NCCL INFO Connected all trees
+ip-26-0-162-14:2525079:2525477 [4] NCCL INFO NVLS comm 0x96c46e0 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-162-14:2525077:2525475 [2] NCCL INFO NVLS comm 0xa1b95c0 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-170-31:3197186:3197593 [6] NCCL INFO NVLS comm 0xa034d40 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-162-14:2525076:2525478 [1] NCCL INFO Connected all trees
+ip-26-0-162-14:2525076:2525478 [1] NCCL INFO NVLS comm 0xa370ac0 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-162-14:2525082:2525472 [7] NCCL INFO Connected all trees
+ip-26-0-162-14:2525082:2525472 [7] NCCL INFO NVLS comm 0x9e76320 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-162-14:2525081:2525473 [6] NCCL INFO Connected all trees
+ip-26-0-162-14:2525081:2525473 [6] NCCL INFO NVLS comm 0x94f5e60 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-162-14:2525075:2525471 [0] NCCL INFO Connected all trees
+ip-26-0-162-14:2525075:2525471 [0] NCCL INFO NVLS comm 0x98759b0 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-170-31:3197181:3197587 [1] NCCL INFO Connected all trees
+ip-26-0-170-31:3197181:3197587 [1] NCCL INFO NVLS comm 0x9bc77b0 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-170-132:3330185:3330593 [2] NCCL INFO Connected all trees
+ip-26-0-170-132:3330185:3330593 [2] NCCL INFO NVLS comm 0x9a9fba0 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-170-132:3330188:3330595 [5] NCCL INFO Connected all trees
+ip-26-0-170-132:3330186:3330592 [3] NCCL INFO Connected all trees
+ip-26-0-170-132:3330188:3330595 [5] NCCL INFO NVLS comm 0xa4ee330 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-170-132:3330186:3330592 [3] NCCL INFO NVLS comm 0x8f16e30 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-163-58:934722:935116 [2] NCCL INFO Connected all trees
+ip-26-0-163-58:934722:935116 [2] NCCL INFO NVLS comm 0x8ac6a60 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-170-132:3330184:3330596 [1] NCCL INFO Connected all trees
+ip-26-0-170-132:3330184:3330596 [1] NCCL INFO NVLS comm 0x960d840 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-170-132:3330187:3330591 [4] NCCL INFO Connected all trees
+ip-26-0-170-132:3330187:3330591 [4] NCCL INFO NVLS comm 0x9a3ab00 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-163-58:934725:935118 [5] NCCL INFO Connected all trees
+ip-26-0-163-58:934725:935118 [5] NCCL INFO NVLS comm 0x8aa9ac0 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-163-58:934723:935117 [3] NCCL INFO Connected all trees
+ip-26-0-163-58:934723:935117 [3] NCCL INFO NVLS comm 0x9ac5f80 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-170-132:3330190:3330589 [7] NCCL INFO Connected all trees
+ip-26-0-170-132:3330190:3330589 [7] NCCL INFO NVLS comm 0xa51d2b0 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-170-132:3330189:3330594 [6] NCCL INFO Connected all trees
+ip-26-0-170-132:3330189:3330594 [6] NCCL INFO NVLS comm 0x87afc80 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-163-58:934724:935113 [4] NCCL INFO Connected all trees
+ip-26-0-163-58:934724:935113 [4] NCCL INFO NVLS comm 0x98bc760 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-170-132:3330183:3330590 [0] NCCL INFO Connected all trees
+ip-26-0-170-132:3330183:3330590 [0] NCCL INFO NVLS comm 0x94d59c0 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-163-58:934721:935115 [1] NCCL INFO Connected all trees
+ip-26-0-163-58:934721:935115 [1] NCCL INFO NVLS comm 0xa549100 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-163-58:934727:935114 [7] NCCL INFO Connected all trees
+ip-26-0-163-58:934727:935114 [7] NCCL INFO NVLS comm 0x9a1dc80 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-163-58:934726:935119 [6] NCCL INFO Connected all trees
+ip-26-0-163-58:934726:935119 [6] NCCL INFO NVLS comm 0x9e0f0e0 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-163-58:934720:935120 [0] NCCL INFO Connected all trees
+ip-26-0-163-58:934720:935120 [0] NCCL INFO NVLS comm 0x98054a0 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-162-14:2525078:2525474 [3] NCCL INFO Channel 01/0 : 3[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-14:2525078:2525474 [3] NCCL INFO Channel 03/0 : 3[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-14:2525080:2525476 [5] NCCL INFO Channel 01/0 : 5[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-14:2525078:2525474 [3] NCCL INFO Channel 05/0 : 3[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-14:2525082:2525472 [7] NCCL INFO Channel 01/0 : 7[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-14:2525080:2525476 [5] NCCL INFO Channel 03/0 : 5[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-14:2525078:2525474 [3] NCCL INFO Channel 07/0 : 3[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-14:2525082:2525472 [7] NCCL INFO Channel 03/0 : 7[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-14:2525080:2525476 [5] NCCL INFO Channel 05/0 : 5[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-14:2525078:2525474 [3] NCCL INFO Channel 09/0 : 3[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-14:2525082:2525472 [7] NCCL INFO Channel 05/0 : 7[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-14:2525080:2525476 [5] NCCL INFO Channel 07/0 : 5[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-14:2525079:2525477 [4] NCCL INFO Channel 01/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-14:2525078:2525474 [3] NCCL INFO Channel 13/0 : 3[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-14:2525075:2525471 [0] NCCL INFO Channel 01/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-14:2525077:2525475 [2] NCCL INFO Channel 01/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-14:2525082:2525472 [7] NCCL INFO Channel 07/0 : 7[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-14:2525080:2525476 [5] NCCL INFO Channel 09/0 : 5[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-14:2525079:2525477 [4] NCCL INFO Channel 03/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-14:2525078:2525474 [3] NCCL INFO Channel 15/0 : 3[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-14:2525075:2525471 [0] NCCL INFO Channel 03/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-14:2525082:2525472 [7] NCCL INFO Channel 09/0 : 7[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-14:2525077:2525475 [2] NCCL INFO Channel 03/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-14:2525080:2525476 [5] NCCL INFO Channel 11/0 : 5[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-14:2525075:2525471 [0] NCCL INFO Channel 05/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-14:2525079:2525477 [4] NCCL INFO Channel 05/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-14:2525082:2525472 [7] NCCL INFO Channel 11/0 : 7[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-14:2525077:2525475 [2] NCCL INFO Channel 05/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-14:2525080:2525476 [5] NCCL INFO Channel 15/0 : 5[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-14:2525081:2525473 [6] NCCL INFO Channel 01/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-14:2525075:2525471 [0] NCCL INFO Channel 07/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-14:2525079:2525477 [4] NCCL INFO Channel 07/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-14:2525082:2525472 [7] NCCL INFO Channel 13/0 : 7[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-14:2525077:2525475 [2] NCCL INFO Channel 07/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-14:2525075:2525471 [0] NCCL INFO Channel 09/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-14:2525079:2525477 [4] NCCL INFO Channel 09/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-14:2525081:2525473 [6] NCCL INFO Channel 03/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-14:2525076:2525478 [1] NCCL INFO Channel 01/0 : 1[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-14:2525077:2525475 [2] NCCL INFO Channel 09/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-14:2525075:2525471 [0] NCCL INFO Channel 11/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-14:2525079:2525477 [4] NCCL INFO Channel 11/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-14:2525081:2525473 [6] NCCL INFO Channel 05/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-14:2525075:2525471 [0] NCCL INFO Channel 13/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-14:2525077:2525475 [2] NCCL INFO Channel 11/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-14:2525076:2525478 [1] NCCL INFO Channel 03/0 : 1[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-14:2525079:2525477 [4] NCCL INFO Channel 13/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-14:2525081:2525473 [6] NCCL INFO Channel 07/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-14:2525075:2525471 [0] NCCL INFO Channel 15/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-14:2525077:2525475 [2] NCCL INFO Channel 13/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-14:2525079:2525477 [4] NCCL INFO Channel 15/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-14:2525076:2525478 [1] NCCL INFO Channel 05/0 : 1[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-14:2525081:2525473 [6] NCCL INFO Channel 09/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-14:2525077:2525475 [2] NCCL INFO Channel 15/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-14:2525076:2525478 [1] NCCL INFO Channel 07/0 : 1[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-14:2525081:2525473 [6] NCCL INFO Channel 11/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-14:2525076:2525478 [1] NCCL INFO Channel 11/0 : 1[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-14:2525081:2525473 [6] NCCL INFO Channel 13/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-14:2525076:2525478 [1] NCCL INFO Channel 13/0 : 1[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-14:2525081:2525473 [6] NCCL INFO Channel 15/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-14:2525076:2525478 [1] NCCL INFO Channel 15/0 : 1[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-31:3197186:3197593 [6] NCCL INFO Channel 00/0 : 22[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-31:3197186:3197593 [6] NCCL INFO Channel 02/0 : 22[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-31:3197186:3197593 [6] NCCL INFO Channel 04/0 : 22[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-31:3197186:3197593 [6] NCCL INFO Channel 08/0 : 22[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-31:3197186:3197593 [6] NCCL INFO Channel 10/0 : 22[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-31:3197186:3197593 [6] NCCL INFO Channel 12/0 : 22[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-31:3197186:3197593 [6] NCCL INFO Channel 14/0 : 22[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-31:3197180:3197589 [0] NCCL INFO Channel 02/0 : 16[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-31:3197180:3197589 [0] NCCL INFO Channel 04/0 : 16[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-31:3197180:3197589 [0] NCCL INFO Channel 06/0 : 16[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-31:3197180:3197589 [0] NCCL INFO Channel 08/0 : 16[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-31:3197180:3197589 [0] NCCL INFO Channel 10/0 : 16[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-31:3197180:3197589 [0] NCCL INFO Channel 12/0 : 16[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-31:3197180:3197589 [0] NCCL INFO Channel 14/0 : 16[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-31:3197183:3197592 [3] NCCL INFO Channel 00/0 : 19[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-31:3197185:3197591 [5] NCCL INFO Channel 00/0 : 21[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-31:3197183:3197592 [3] NCCL INFO Channel 02/0 : 19[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-31:3197184:3197586 [4] NCCL INFO Channel 00/0 : 20[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-31:3197185:3197591 [5] NCCL INFO Channel 02/0 : 21[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-31:3197183:3197592 [3] NCCL INFO Channel 04/0 : 19[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-31:3197184:3197586 [4] NCCL INFO Channel 02/0 : 20[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-31:3197185:3197591 [5] NCCL INFO Channel 04/0 : 21[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-31:3197183:3197592 [3] NCCL INFO Channel 06/0 : 19[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-31:3197182:3197588 [2] NCCL INFO Channel 00/0 : 18[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-31:3197187:3197590 [7] NCCL INFO Channel 00/0 : 23[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-31:3197184:3197586 [4] NCCL INFO Channel 06/0 : 20[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-31:3197181:3197587 [1] NCCL INFO Channel 00/0 : 17[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-31:3197183:3197592 [3] NCCL INFO Channel 08/0 : 19[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-31:3197185:3197591 [5] NCCL INFO Channel 06/0 : 21[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-31:3197187:3197590 [7] NCCL INFO Channel 02/0 : 23[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-31:3197182:3197588 [2] NCCL INFO Channel 04/0 : 18[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-31:3197184:3197586 [4] NCCL INFO Channel 08/0 : 20[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-31:3197181:3197587 [1] NCCL INFO Channel 02/0 : 17[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-31:3197183:3197592 [3] NCCL INFO Channel 10/0 : 19[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-31:3197185:3197591 [5] NCCL INFO Channel 08/0 : 21[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-31:3197187:3197590 [7] NCCL INFO Channel 04/0 : 23[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-31:3197182:3197588 [2] NCCL INFO Channel 06/0 : 18[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-31:3197184:3197586 [4] NCCL INFO Channel 10/0 : 20[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-31:3197185:3197591 [5] NCCL INFO Channel 10/0 : 21[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-31:3197181:3197587 [1] NCCL INFO Channel 04/0 : 17[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-31:3197183:3197592 [3] NCCL INFO Channel 12/0 : 19[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-31:3197187:3197590 [7] NCCL INFO Channel 06/0 : 23[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-31:3197182:3197588 [2] NCCL INFO Channel 08/0 : 18[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-31:3197184:3197586 [4] NCCL INFO Channel 12/0 : 20[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-31:3197185:3197591 [5] NCCL INFO Channel 12/0 : 21[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-31:3197181:3197587 [1] NCCL INFO Channel 06/0 : 17[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-31:3197187:3197590 [7] NCCL INFO Channel 08/0 : 23[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-31:3197183:3197592 [3] NCCL INFO Channel 14/0 : 19[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-31:3197184:3197586 [4] NCCL INFO Channel 14/0 : 20[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-31:3197182:3197588 [2] NCCL INFO Channel 10/0 : 18[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-31:3197185:3197591 [5] NCCL INFO Channel 14/0 : 21[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-31:3197187:3197590 [7] NCCL INFO Channel 10/0 : 23[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-31:3197182:3197588 [2] NCCL INFO Channel 12/0 : 18[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-31:3197181:3197587 [1] NCCL INFO Channel 08/0 : 17[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-31:3197187:3197590 [7] NCCL INFO Channel 12/0 : 23[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-31:3197182:3197588 [2] NCCL INFO Channel 14/0 : 18[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-31:3197181:3197587 [1] NCCL INFO Channel 10/0 : 17[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-31:3197187:3197590 [7] NCCL INFO Channel 14/0 : 23[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-31:3197181:3197587 [1] NCCL INFO Channel 12/0 : 17[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-31:3197181:3197587 [1] NCCL INFO Channel 14/0 : 17[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330183:3330590 [0] NCCL INFO Channel 01/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330183:3330590 [0] NCCL INFO Channel 02/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330183:3330590 [0] NCCL INFO Channel 03/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330183:3330590 [0] NCCL INFO Channel 04/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330183:3330590 [0] NCCL INFO Channel 05/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330185:3330593 [2] NCCL INFO Channel 00/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330187:3330591 [4] NCCL INFO Channel 00/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-132:3330190:3330589 [7] NCCL INFO Channel 00/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330189:3330594 [6] NCCL INFO Channel 00/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330183:3330590 [0] NCCL INFO Channel 06/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330186:3330592 [3] NCCL INFO Channel 00/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330187:3330591 [4] NCCL INFO Channel 01/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-132:3330185:3330593 [2] NCCL INFO Channel 01/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330190:3330589 [7] NCCL INFO Channel 01/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330189:3330594 [6] NCCL INFO Channel 01/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330183:3330590 [0] NCCL INFO Channel 07/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330186:3330592 [3] NCCL INFO Channel 01/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330187:3330591 [4] NCCL INFO Channel 02/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-132:3330188:3330595 [5] NCCL INFO Channel 00/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330189:3330594 [6] NCCL INFO Channel 02/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330183:3330590 [0] NCCL INFO Channel 09/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330185:3330593 [2] NCCL INFO Channel 03/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330190:3330589 [7] NCCL INFO Channel 02/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330186:3330592 [3] NCCL INFO Channel 02/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330187:3330591 [4] NCCL INFO Channel 03/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-132:3330184:3330596 [1] NCCL INFO Channel 00/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330188:3330595 [5] NCCL INFO Channel 01/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330183:3330590 [0] NCCL INFO Channel 10/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330189:3330594 [6] NCCL INFO Channel 03/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330185:3330593 [2] NCCL INFO Channel 04/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330186:3330592 [3] NCCL INFO Channel 04/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330190:3330589 [7] NCCL INFO Channel 03/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330184:3330596 [1] NCCL INFO Channel 02/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330187:3330591 [4] NCCL INFO Channel 05/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-132:3330188:3330595 [5] NCCL INFO Channel 02/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330183:3330590 [0] NCCL INFO Channel 11/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330189:3330594 [6] NCCL INFO Channel 04/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330185:3330593 [2] NCCL INFO Channel 05/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330186:3330592 [3] NCCL INFO Channel 05/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330190:3330589 [7] NCCL INFO Channel 04/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330184:3330596 [1] NCCL INFO Channel 03/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330187:3330591 [4] NCCL INFO Channel 06/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-132:3330189:3330594 [6] NCCL INFO Channel 05/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330183:3330590 [0] NCCL INFO Channel 12/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330188:3330595 [5] NCCL INFO Channel 03/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330185:3330593 [2] NCCL INFO Channel 06/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330186:3330592 [3] NCCL INFO Channel 06/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330190:3330589 [7] NCCL INFO Channel 05/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330189:3330594 [6] NCCL INFO Channel 07/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330183:3330590 [0] NCCL INFO Channel 13/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330184:3330596 [1] NCCL INFO Channel 04/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330187:3330591 [4] NCCL INFO Channel 07/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-132:3330186:3330592 [3] NCCL INFO Channel 07/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330188:3330595 [5] NCCL INFO Channel 04/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330185:3330593 [2] NCCL INFO Channel 07/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330189:3330594 [6] NCCL INFO Channel 08/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330190:3330589 [7] NCCL INFO Channel 06/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330183:3330590 [0] NCCL INFO Channel 14/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330184:3330596 [1] NCCL INFO Channel 05/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330187:3330591 [4] NCCL INFO Channel 08/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-132:3330186:3330592 [3] NCCL INFO Channel 08/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330188:3330595 [5] NCCL INFO Channel 06/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330185:3330593 [2] NCCL INFO Channel 08/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330189:3330594 [6] NCCL INFO Channel 09/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330190:3330589 [7] NCCL INFO Channel 08/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330183:3330590 [0] NCCL INFO Channel 15/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330184:3330596 [1] NCCL INFO Channel 06/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330187:3330591 [4] NCCL INFO Channel 09/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-132:3330186:3330592 [3] NCCL INFO Channel 09/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330189:3330594 [6] NCCL INFO Channel 10/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330185:3330593 [2] NCCL INFO Channel 09/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330188:3330595 [5] NCCL INFO Channel 07/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330183:3330590 [0] NCCL INFO Channel 02/0 : 16[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330190:3330589 [7] NCCL INFO Channel 09/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330184:3330596 [1] NCCL INFO Channel 07/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330187:3330591 [4] NCCL INFO Channel 10/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-132:3330186:3330592 [3] NCCL INFO Channel 10/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330189:3330594 [6] NCCL INFO Channel 11/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330188:3330595 [5] NCCL INFO Channel 08/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330185:3330593 [2] NCCL INFO Channel 11/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330183:3330590 [0] NCCL INFO Channel 04/0 : 16[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330190:3330589 [7] NCCL INFO Channel 10/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330184:3330596 [1] NCCL INFO Channel 08/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330187:3330591 [4] NCCL INFO Channel 11/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-132:3330186:3330592 [3] NCCL INFO Channel 12/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330189:3330594 [6] NCCL INFO Channel 12/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330188:3330595 [5] NCCL INFO Channel 09/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330185:3330593 [2] NCCL INFO Channel 12/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330183:3330590 [0] NCCL INFO Channel 06/0 : 16[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330190:3330589 [7] NCCL INFO Channel 11/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330187:3330591 [4] NCCL INFO Channel 13/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-132:3330186:3330592 [3] NCCL INFO Channel 13/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330184:3330596 [1] NCCL INFO Channel 10/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330189:3330594 [6] NCCL INFO Channel 13/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330188:3330595 [5] NCCL INFO Channel 10/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330183:3330590 [0] NCCL INFO Channel 08/0 : 16[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330185:3330593 [2] NCCL INFO Channel 13/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330187:3330591 [4] NCCL INFO Channel 14/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-132:3330190:3330589 [7] NCCL INFO Channel 12/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330189:3330594 [6] NCCL INFO Channel 15/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330186:3330592 [3] NCCL INFO Channel 14/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330184:3330596 [1] NCCL INFO Channel 11/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330183:3330590 [0] NCCL INFO Channel 10/0 : 16[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330188:3330595 [5] NCCL INFO Channel 11/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330185:3330593 [2] NCCL INFO Channel 14/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330187:3330591 [4] NCCL INFO Channel 15/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-132:3330190:3330589 [7] NCCL INFO Channel 13/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330184:3330596 [1] NCCL INFO Channel 12/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330186:3330592 [3] NCCL INFO Channel 15/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330183:3330590 [0] NCCL INFO Channel 12/0 : 16[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330189:3330594 [6] NCCL INFO Channel 00/0 : 22[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330185:3330593 [2] NCCL INFO Channel 15/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330188:3330595 [5] NCCL INFO Channel 12/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330187:3330591 [4] NCCL INFO Channel 00/0 : 20[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-132:3330190:3330589 [7] NCCL INFO Channel 14/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330183:3330590 [0] NCCL INFO Channel 14/0 : 16[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330189:3330594 [6] NCCL INFO Channel 02/0 : 22[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330186:3330592 [3] NCCL INFO Channel 00/0 : 19[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330184:3330596 [1] NCCL INFO Channel 13/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330185:3330593 [2] NCCL INFO Channel 00/0 : 18[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330187:3330591 [4] NCCL INFO Channel 02/0 : 20[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-132:3330190:3330589 [7] NCCL INFO Channel 00/0 : 23[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330188:3330595 [5] NCCL INFO Channel 14/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330189:3330594 [6] NCCL INFO Channel 04/0 : 22[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330184:3330596 [1] NCCL INFO Channel 14/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330186:3330592 [3] NCCL INFO Channel 02/0 : 19[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330190:3330589 [7] NCCL INFO Channel 02/0 : 23[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330185:3330593 [2] NCCL INFO Channel 04/0 : 18[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330187:3330591 [4] NCCL INFO Channel 06/0 : 20[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-132:3330188:3330595 [5] NCCL INFO Channel 15/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330189:3330594 [6] NCCL INFO Channel 08/0 : 22[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330184:3330596 [1] NCCL INFO Channel 15/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330190:3330589 [7] NCCL INFO Channel 04/0 : 23[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330186:3330592 [3] NCCL INFO Channel 04/0 : 19[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330187:3330591 [4] NCCL INFO Channel 08/0 : 20[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-132:3330185:3330593 [2] NCCL INFO Channel 06/0 : 18[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330189:3330594 [6] NCCL INFO Channel 10/0 : 22[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330188:3330595 [5] NCCL INFO Channel 00/0 : 21[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330184:3330596 [1] NCCL INFO Channel 00/0 : 17[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330190:3330589 [7] NCCL INFO Channel 06/0 : 23[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-31:3197180:3197589 [0] NCCL INFO Channel 01/0 : 8[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330186:3330592 [3] NCCL INFO Channel 06/0 : 19[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330187:3330591 [4] NCCL INFO Channel 10/0 : 20[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-132:3330189:3330594 [6] NCCL INFO Channel 12/0 : 22[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330184:3330596 [1] NCCL INFO Channel 02/0 : 17[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330185:3330593 [2] NCCL INFO Channel 08/0 : 18[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330188:3330595 [5] NCCL INFO Channel 02/0 : 21[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-31:3197180:3197589 [0] NCCL INFO Channel 03/0 : 8[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330190:3330589 [7] NCCL INFO Channel 08/0 : 23[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330187:3330591 [4] NCCL INFO Channel 12/0 : 20[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-132:3330186:3330592 [3] NCCL INFO Channel 08/0 : 19[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-31:3197180:3197589 [0] NCCL INFO Channel 05/0 : 8[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330189:3330594 [6] NCCL INFO Channel 14/0 : 22[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330184:3330596 [1] NCCL INFO Channel 04/0 : 17[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330185:3330593 [2] NCCL INFO Channel 10/0 : 18[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330188:3330595 [5] NCCL INFO Channel 04/0 : 21[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330190:3330589 [7] NCCL INFO Channel 10/0 : 23[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-31:3197180:3197589 [0] NCCL INFO Channel 07/0 : 8[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330186:3330592 [3] NCCL INFO Channel 10/0 : 19[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330187:3330591 [4] NCCL INFO Channel 14/0 : 20[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-132:3330184:3330596 [1] NCCL INFO Channel 06/0 : 17[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330185:3330593 [2] NCCL INFO Channel 12/0 : 18[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-31:3197180:3197589 [0] NCCL INFO Channel 09/0 : 8[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330188:3330595 [5] NCCL INFO Channel 06/0 : 21[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330190:3330589 [7] NCCL INFO Channel 12/0 : 23[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330186:3330592 [3] NCCL INFO Channel 12/0 : 19[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330184:3330596 [1] NCCL INFO Channel 08/0 : 17[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330185:3330593 [2] NCCL INFO Channel 14/0 : 18[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-31:3197180:3197589 [0] NCCL INFO Channel 11/0 : 8[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330190:3330589 [7] NCCL INFO Channel 14/0 : 23[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330186:3330592 [3] NCCL INFO Channel 14/0 : 19[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330188:3330595 [5] NCCL INFO Channel 08/0 : 21[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330184:3330596 [1] NCCL INFO Channel 10/0 : 17[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-31:3197180:3197589 [0] NCCL INFO Channel 13/0 : 8[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330188:3330595 [5] NCCL INFO Channel 10/0 : 21[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330184:3330596 [1] NCCL INFO Channel 12/0 : 17[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330188:3330595 [5] NCCL INFO Channel 12/0 : 21[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-31:3197180:3197589 [0] NCCL INFO Channel 15/0 : 8[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330184:3330596 [1] NCCL INFO Channel 14/0 : 17[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-31:3197186:3197593 [6] NCCL INFO Channel 01/0 : 14[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330188:3330595 [5] NCCL INFO Channel 14/0 : 21[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-31:3197180:3197589 [0] NCCL INFO Channel 01/0 : 24[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-31:3197186:3197593 [6] NCCL INFO Channel 03/0 : 14[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-31:3197184:3197586 [4] NCCL INFO Channel 01/0 : 12[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-31:3197180:3197589 [0] NCCL INFO Channel 03/0 : 24[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-31:3197186:3197593 [6] NCCL INFO Channel 05/0 : 14[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-31:3197184:3197586 [4] NCCL INFO Channel 03/0 : 12[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-31:3197180:3197589 [0] NCCL INFO Channel 05/0 : 24[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-31:3197182:3197588 [2] NCCL INFO Channel 01/0 : 10[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-31:3197183:3197592 [3] NCCL INFO Channel 01/0 : 11[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-31:3197186:3197593 [6] NCCL INFO Channel 07/0 : 14[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-31:3197187:3197590 [7] NCCL INFO Channel 01/0 : 15[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-31:3197180:3197589 [0] NCCL INFO Channel 07/0 : 24[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-31:3197184:3197586 [4] NCCL INFO Channel 05/0 : 12[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-31:3197183:3197592 [3] NCCL INFO Channel 03/0 : 11[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-31:3197182:3197588 [2] NCCL INFO Channel 03/0 : 10[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-31:3197187:3197590 [7] NCCL INFO Channel 03/0 : 15[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-31:3197186:3197593 [6] NCCL INFO Channel 09/0 : 14[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-31:3197180:3197589 [0] NCCL INFO Channel 09/0 : 24[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-31:3197184:3197586 [4] NCCL INFO Channel 07/0 : 12[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-31:3197183:3197592 [3] NCCL INFO Channel 05/0 : 11[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-31:3197182:3197588 [2] NCCL INFO Channel 05/0 : 10[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-31:3197187:3197590 [7] NCCL INFO Channel 05/0 : 15[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-31:3197186:3197593 [6] NCCL INFO Channel 11/0 : 14[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-31:3197185:3197591 [5] NCCL INFO Channel 01/0 : 13[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-31:3197181:3197587 [1] NCCL INFO Channel 01/0 : 9[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-31:3197180:3197589 [0] NCCL INFO Channel 11/0 : 24[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-31:3197184:3197586 [4] NCCL INFO Channel 09/0 : 12[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-31:3197182:3197588 [2] NCCL INFO Channel 07/0 : 10[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-31:3197183:3197592 [3] NCCL INFO Channel 07/0 : 11[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-31:3197187:3197590 [7] NCCL INFO Channel 07/0 : 15[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-31:3197186:3197593 [6] NCCL INFO Channel 13/0 : 14[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-31:3197180:3197589 [0] NCCL INFO Channel 13/0 : 24[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-31:3197185:3197591 [5] NCCL INFO Channel 03/0 : 13[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-31:3197184:3197586 [4] NCCL INFO Channel 11/0 : 12[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-31:3197181:3197587 [1] NCCL INFO Channel 03/0 : 9[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-31:3197187:3197590 [7] NCCL INFO Channel 09/0 : 15[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-31:3197182:3197588 [2] NCCL INFO Channel 09/0 : 10[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-31:3197183:3197592 [3] NCCL INFO Channel 09/0 : 11[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-31:3197186:3197593 [6] NCCL INFO Channel 15/0 : 14[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-31:3197180:3197589 [0] NCCL INFO Channel 15/0 : 24[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-31:3197185:3197591 [5] NCCL INFO Channel 05/0 : 13[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-31:3197184:3197586 [4] NCCL INFO Channel 13/0 : 12[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-31:3197187:3197590 [7] NCCL INFO Channel 11/0 : 15[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-31:3197181:3197587 [1] NCCL INFO Channel 05/0 : 9[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-31:3197186:3197593 [6] NCCL INFO Channel 01/0 : 30[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-31:3197183:3197592 [3] NCCL INFO Channel 13/0 : 11[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-31:3197182:3197588 [2] NCCL INFO Channel 11/0 : 10[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-31:3197185:3197591 [5] NCCL INFO Channel 07/0 : 13[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-31:3197184:3197586 [4] NCCL INFO Channel 15/0 : 12[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-31:3197187:3197590 [7] NCCL INFO Channel 13/0 : 15[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-31:3197181:3197587 [1] NCCL INFO Channel 07/0 : 9[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-31:3197186:3197593 [6] NCCL INFO Channel 03/0 : 30[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-31:3197185:3197591 [5] NCCL INFO Channel 09/0 : 13[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-31:3197183:3197592 [3] NCCL INFO Channel 15/0 : 11[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-31:3197182:3197588 [2] NCCL INFO Channel 13/0 : 10[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-31:3197187:3197590 [7] NCCL INFO Channel 01/0 : 31[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-31:3197184:3197586 [4] NCCL INFO Channel 01/0 : 28[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-31:3197186:3197593 [6] NCCL INFO Channel 05/0 : 30[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-31:3197181:3197587 [1] NCCL INFO Channel 11/0 : 9[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-31:3197185:3197591 [5] NCCL INFO Channel 11/0 : 13[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-31:3197183:3197592 [3] NCCL INFO Channel 01/0 : 27[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-31:3197182:3197588 [2] NCCL INFO Channel 15/0 : 10[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-31:3197187:3197590 [7] NCCL INFO Channel 03/0 : 31[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-31:3197184:3197586 [4] NCCL INFO Channel 03/0 : 28[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-31:3197186:3197593 [6] NCCL INFO Channel 07/0 : 30[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-31:3197181:3197587 [1] NCCL INFO Channel 13/0 : 9[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-31:3197185:3197591 [5] NCCL INFO Channel 15/0 : 13[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-31:3197183:3197592 [3] NCCL INFO Channel 03/0 : 27[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-31:3197187:3197590 [7] NCCL INFO Channel 05/0 : 31[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-31:3197182:3197588 [2] NCCL INFO Channel 01/0 : 26[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-31:3197184:3197586 [4] NCCL INFO Channel 05/0 : 28[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-31:3197186:3197593 [6] NCCL INFO Channel 09/0 : 30[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-31:3197181:3197587 [1] NCCL INFO Channel 15/0 : 9[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-31:3197185:3197591 [5] NCCL INFO Channel 01/0 : 29[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-31:3197187:3197590 [7] NCCL INFO Channel 07/0 : 31[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-31:3197183:3197592 [3] NCCL INFO Channel 05/0 : 27[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-31:3197182:3197588 [2] NCCL INFO Channel 03/0 : 26[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-31:3197186:3197593 [6] NCCL INFO Channel 11/0 : 30[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-31:3197184:3197586 [4] NCCL INFO Channel 07/0 : 28[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-31:3197181:3197587 [1] NCCL INFO Channel 01/0 : 25[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-31:3197185:3197591 [5] NCCL INFO Channel 03/0 : 29[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-31:3197187:3197590 [7] NCCL INFO Channel 09/0 : 31[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-31:3197183:3197592 [3] NCCL INFO Channel 07/0 : 27[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-31:3197184:3197586 [4] NCCL INFO Channel 09/0 : 28[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-31:3197186:3197593 [6] NCCL INFO Channel 13/0 : 30[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-31:3197182:3197588 [2] NCCL INFO Channel 05/0 : 26[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-31:3197185:3197591 [5] NCCL INFO Channel 05/0 : 29[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-31:3197181:3197587 [1] NCCL INFO Channel 03/0 : 25[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-31:3197187:3197590 [7] NCCL INFO Channel 11/0 : 31[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-31:3197183:3197592 [3] NCCL INFO Channel 09/0 : 27[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-31:3197184:3197586 [4] NCCL INFO Channel 11/0 : 28[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-31:3197186:3197593 [6] NCCL INFO Channel 15/0 : 30[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-31:3197182:3197588 [2] NCCL INFO Channel 07/0 : 26[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-31:3197185:3197591 [5] NCCL INFO Channel 07/0 : 29[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-31:3197187:3197590 [7] NCCL INFO Channel 13/0 : 31[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-31:3197181:3197587 [1] NCCL INFO Channel 05/0 : 25[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-31:3197183:3197592 [3] NCCL INFO Channel 13/0 : 27[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-31:3197184:3197586 [4] NCCL INFO Channel 13/0 : 28[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-31:3197182:3197588 [2] NCCL INFO Channel 09/0 : 26[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-31:3197185:3197591 [5] NCCL INFO Channel 09/0 : 29[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-31:3197184:3197586 [4] NCCL INFO Channel 15/0 : 28[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-31:3197183:3197592 [3] NCCL INFO Channel 15/0 : 27[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-31:3197181:3197587 [1] NCCL INFO Channel 07/0 : 25[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-31:3197182:3197588 [2] NCCL INFO Channel 11/0 : 26[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-31:3197185:3197591 [5] NCCL INFO Channel 11/0 : 29[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-31:3197181:3197587 [1] NCCL INFO Channel 11/0 : 25[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-31:3197185:3197591 [5] NCCL INFO Channel 15/0 : 29[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-31:3197182:3197588 [2] NCCL INFO Channel 13/0 : 26[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-31:3197181:3197587 [1] NCCL INFO Channel 13/0 : 25[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-31:3197182:3197588 [2] NCCL INFO Channel 15/0 : 26[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-31:3197181:3197587 [1] NCCL INFO Channel 15/0 : 25[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934725:935118 [5] NCCL INFO Channel 01/0 : 5[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934721:935115 [1] NCCL INFO Channel 01/0 : 1[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934725:935118 [5] NCCL INFO Channel 03/0 : 5[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934721:935115 [1] NCCL INFO Channel 03/0 : 1[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934725:935118 [5] NCCL INFO Channel 05/0 : 5[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934721:935115 [1] NCCL INFO Channel 05/0 : 1[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934725:935118 [5] NCCL INFO Channel 07/0 : 5[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934727:935114 [7] NCCL INFO Channel 01/0 : 7[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934724:935113 [4] NCCL INFO Channel 01/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934725:935118 [5] NCCL INFO Channel 09/0 : 5[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934721:935115 [1] NCCL INFO Channel 07/0 : 1[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934727:935114 [7] NCCL INFO Channel 03/0 : 7[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934722:935116 [2] NCCL INFO Channel 01/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934724:935113 [4] NCCL INFO Channel 03/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934725:935118 [5] NCCL INFO Channel 11/0 : 5[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934726:935119 [6] NCCL INFO Channel 01/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934721:935115 [1] NCCL INFO Channel 11/0 : 1[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934727:935114 [7] NCCL INFO Channel 05/0 : 7[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934722:935116 [2] NCCL INFO Channel 03/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934724:935113 [4] NCCL INFO Channel 05/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934720:935120 [0] NCCL INFO Channel 01/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934725:935118 [5] NCCL INFO Channel 15/0 : 5[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934726:935119 [6] NCCL INFO Channel 03/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934721:935115 [1] NCCL INFO Channel 13/0 : 1[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934727:935114 [7] NCCL INFO Channel 07/0 : 7[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934722:935116 [2] NCCL INFO Channel 05/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934724:935113 [4] NCCL INFO Channel 07/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934720:935120 [0] NCCL INFO Channel 03/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934726:935119 [6] NCCL INFO Channel 05/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934721:935115 [1] NCCL INFO Channel 15/0 : 1[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934725:935118 [5] NCCL INFO Channel 00/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934727:935114 [7] NCCL INFO Channel 09/0 : 7[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934722:935116 [2] NCCL INFO Channel 07/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934724:935113 [4] NCCL INFO Channel 09/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934720:935120 [0] NCCL INFO Channel 05/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934725:935118 [5] NCCL INFO Channel 01/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934726:935119 [6] NCCL INFO Channel 07/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934721:935115 [1] NCCL INFO Channel 00/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934722:935116 [2] NCCL INFO Channel 09/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934723:935117 [3] NCCL INFO Channel 01/0 : 3[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934727:935114 [7] NCCL INFO Channel 11/0 : 7[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934720:935120 [0] NCCL INFO Channel 07/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934724:935113 [4] NCCL INFO Channel 11/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934725:935118 [5] NCCL INFO Channel 02/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934726:935119 [6] NCCL INFO Channel 09/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934721:935115 [1] NCCL INFO Channel 02/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934722:935116 [2] NCCL INFO Channel 11/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934723:935117 [3] NCCL INFO Channel 03/0 : 3[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934724:935113 [4] NCCL INFO Channel 13/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934725:935118 [5] NCCL INFO Channel 03/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934727:935114 [7] NCCL INFO Channel 13/0 : 7[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934720:935120 [0] NCCL INFO Channel 09/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934721:935115 [1] NCCL INFO Channel 03/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934726:935119 [6] NCCL INFO Channel 11/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934722:935116 [2] NCCL INFO Channel 13/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934723:935117 [3] NCCL INFO Channel 05/0 : 3[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934725:935118 [5] NCCL INFO Channel 04/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934720:935120 [0] NCCL INFO Channel 11/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934727:935114 [7] NCCL INFO Channel 00/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934721:935115 [1] NCCL INFO Channel 04/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934724:935113 [4] NCCL INFO Channel 15/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934726:935119 [6] NCCL INFO Channel 13/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934722:935116 [2] NCCL INFO Channel 15/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934723:935117 [3] NCCL INFO Channel 07/0 : 3[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934725:935118 [5] NCCL INFO Channel 06/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934720:935120 [0] NCCL INFO Channel 13/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934721:935115 [1] NCCL INFO Channel 05/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934722:935116 [2] NCCL INFO Channel 00/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934727:935114 [7] NCCL INFO Channel 01/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934724:935113 [4] NCCL INFO Channel 00/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934726:935119 [6] NCCL INFO Channel 15/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934723:935117 [3] NCCL INFO Channel 09/0 : 3[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934725:935118 [5] NCCL INFO Channel 07/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934720:935120 [0] NCCL INFO Channel 15/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934721:935115 [1] NCCL INFO Channel 06/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934722:935116 [2] NCCL INFO Channel 01/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934726:935119 [6] NCCL INFO Channel 00/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934727:935114 [7] NCCL INFO Channel 02/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934723:935117 [3] NCCL INFO Channel 13/0 : 3[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934724:935113 [4] NCCL INFO Channel 01/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934725:935118 [5] NCCL INFO Channel 08/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934720:935120 [0] NCCL INFO Channel 01/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934721:935115 [1] NCCL INFO Channel 07/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934722:935116 [2] NCCL INFO Channel 03/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934726:935119 [6] NCCL INFO Channel 01/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934724:935113 [4] NCCL INFO Channel 02/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934723:935117 [3] NCCL INFO Channel 15/0 : 3[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934727:935114 [7] NCCL INFO Channel 03/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934720:935120 [0] NCCL INFO Channel 02/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934721:935115 [1] NCCL INFO Channel 08/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934722:935116 [2] NCCL INFO Channel 04/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934725:935118 [5] NCCL INFO Channel 09/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934726:935119 [6] NCCL INFO Channel 02/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934724:935113 [4] NCCL INFO Channel 03/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934723:935117 [3] NCCL INFO Channel 00/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934727:935114 [7] NCCL INFO Channel 04/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934720:935120 [0] NCCL INFO Channel 03/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934721:935115 [1] NCCL INFO Channel 10/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934722:935116 [2] NCCL INFO Channel 05/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934725:935118 [5] NCCL INFO Channel 10/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934726:935119 [6] NCCL INFO Channel 03/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934724:935113 [4] NCCL INFO Channel 05/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934723:935117 [3] NCCL INFO Channel 01/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934720:935120 [0] NCCL INFO Channel 04/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934727:935114 [7] NCCL INFO Channel 05/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934721:935115 [1] NCCL INFO Channel 11/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934722:935116 [2] NCCL INFO Channel 06/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934725:935118 [5] NCCL INFO Channel 11/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934726:935119 [6] NCCL INFO Channel 04/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934724:935113 [4] NCCL INFO Channel 06/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934723:935117 [3] NCCL INFO Channel 02/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934720:935120 [0] NCCL INFO Channel 05/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934727:935114 [7] NCCL INFO Channel 06/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934721:935115 [1] NCCL INFO Channel 12/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934722:935116 [2] NCCL INFO Channel 07/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934725:935118 [5] NCCL INFO Channel 12/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934723:935117 [3] NCCL INFO Channel 04/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934724:935113 [4] NCCL INFO Channel 07/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934726:935119 [6] NCCL INFO Channel 05/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934720:935120 [0] NCCL INFO Channel 06/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934721:935115 [1] NCCL INFO Channel 13/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934727:935114 [7] NCCL INFO Channel 08/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934722:935116 [2] NCCL INFO Channel 08/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934725:935118 [5] NCCL INFO Channel 14/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934723:935117 [3] NCCL INFO Channel 05/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934724:935113 [4] NCCL INFO Channel 08/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934726:935119 [6] NCCL INFO Channel 07/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934721:935115 [1] NCCL INFO Channel 14/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934722:935116 [2] NCCL INFO Channel 09/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934720:935120 [0] NCCL INFO Channel 07/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934727:935114 [7] NCCL INFO Channel 09/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934723:935117 [3] NCCL INFO Channel 06/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934725:935118 [5] NCCL INFO Channel 15/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934724:935113 [4] NCCL INFO Channel 09/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934726:935119 [6] NCCL INFO Channel 08/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934721:935115 [1] NCCL INFO Channel 15/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934722:935116 [2] NCCL INFO Channel 11/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934720:935120 [0] NCCL INFO Channel 09/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934727:935114 [7] NCCL INFO Channel 10/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934723:935117 [3] NCCL INFO Channel 07/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934724:935113 [4] NCCL INFO Channel 10/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934726:935119 [6] NCCL INFO Channel 09/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934722:935116 [2] NCCL INFO Channel 12/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934727:935114 [7] NCCL INFO Channel 11/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934720:935120 [0] NCCL INFO Channel 10/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934723:935117 [3] NCCL INFO Channel 08/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934724:935113 [4] NCCL INFO Channel 11/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934726:935119 [6] NCCL INFO Channel 10/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934722:935116 [2] NCCL INFO Channel 13/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934727:935114 [7] NCCL INFO Channel 12/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934723:935117 [3] NCCL INFO Channel 09/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934724:935113 [4] NCCL INFO Channel 13/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934726:935119 [6] NCCL INFO Channel 11/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934720:935120 [0] NCCL INFO Channel 11/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934722:935116 [2] NCCL INFO Channel 14/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934727:935114 [7] NCCL INFO Channel 13/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-14:2525080:2525476 [5] NCCL INFO Channel 00/0 : 21[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-14:2525076:2525478 [1] NCCL INFO Channel 00/0 : 17[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330188:3330595 [5] NCCL INFO Channel 00/0 : 5[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934725:935118 [5] NCCL INFO Channel 01/0 : 29[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-14:2525076:2525478 [1] NCCL INFO Channel 02/0 : 17[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330184:3330596 [1] NCCL INFO Channel 00/0 : 1[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-14:2525080:2525476 [5] NCCL INFO Channel 02/0 : 21[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934723:935117 [3] NCCL INFO Channel 10/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934724:935113 [4] NCCL INFO Channel 14/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934726:935119 [6] NCCL INFO Channel 12/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934721:935115 [1] NCCL INFO Channel 01/0 : 25[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934722:935116 [2] NCCL INFO Channel 15/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934727:935114 [7] NCCL INFO Channel 14/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330188:3330595 [5] NCCL INFO Channel 02/0 : 5[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934720:935120 [0] NCCL INFO Channel 12/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330184:3330596 [1] NCCL INFO Channel 02/0 : 1[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-14:2525076:2525478 [1] NCCL INFO Channel 04/0 : 17[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-14:2525080:2525476 [5] NCCL INFO Channel 04/0 : 21[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934725:935118 [5] NCCL INFO Channel 03/0 : 29[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-14:2525076:2525478 [1] NCCL INFO Channel 06/0 : 17[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934724:935113 [4] NCCL INFO Channel 15/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934723:935117 [3] NCCL INFO Channel 12/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330188:3330595 [5] NCCL INFO Channel 04/0 : 5[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934726:935119 [6] NCCL INFO Channel 13/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934721:935115 [1] NCCL INFO Channel 03/0 : 25[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330184:3330596 [1] NCCL INFO Channel 04/0 : 1[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-14:2525080:2525476 [5] NCCL INFO Channel 06/0 : 21[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-14:2525076:2525478 [1] NCCL INFO Channel 08/0 : 17[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330188:3330595 [5] NCCL INFO Channel 06/0 : 5[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330184:3330596 [1] NCCL INFO Channel 06/0 : 1[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-14:2525080:2525476 [5] NCCL INFO Channel 08/0 : 21[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934720:935120 [0] NCCL INFO Channel 13/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934725:935118 [5] NCCL INFO Channel 05/0 : 29[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934723:935117 [3] NCCL INFO Channel 13/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934726:935119 [6] NCCL INFO Channel 15/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934721:935115 [1] NCCL INFO Channel 05/0 : 25[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330188:3330595 [5] NCCL INFO Channel 08/0 : 5[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330184:3330596 [1] NCCL INFO Channel 08/0 : 1[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-14:2525076:2525478 [1] NCCL INFO Channel 10/0 : 17[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934725:935118 [5] NCCL INFO Channel 07/0 : 29[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934720:935120 [0] NCCL INFO Channel 14/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-14:2525080:2525476 [5] NCCL INFO Channel 10/0 : 21[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934723:935117 [3] NCCL INFO Channel 14/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330188:3330595 [5] NCCL INFO Channel 10/0 : 5[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-14:2525082:2525472 [7] NCCL INFO Channel 00/0 : 23[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934721:935115 [1] NCCL INFO Channel 07/0 : 25[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330184:3330596 [1] NCCL INFO Channel 10/0 : 1[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934725:935118 [5] NCCL INFO Channel 09/0 : 29[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-14:2525076:2525478 [1] NCCL INFO Channel 12/0 : 17[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-14:2525077:2525475 [2] NCCL INFO Channel 00/0 : 18[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934720:935120 [0] NCCL INFO Channel 15/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-14:2525079:2525477 [4] NCCL INFO Channel 00/0 : 20[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-14:2525080:2525476 [5] NCCL INFO Channel 12/0 : 21[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330188:3330595 [5] NCCL INFO Channel 12/0 : 5[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-14:2525082:2525472 [7] NCCL INFO Channel 02/0 : 23[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330184:3330596 [1] NCCL INFO Channel 12/0 : 1[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330185:3330593 [2] NCCL INFO Channel 00/0 : 2[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330188:3330595 [5] NCCL INFO Channel 14/0 : 5[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-14:2525077:2525475 [2] NCCL INFO Channel 04/0 : 18[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-14:2525076:2525478 [1] NCCL INFO Channel 14/0 : 17[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934722:935116 [2] NCCL INFO Channel 01/0 : 26[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934721:935115 [1] NCCL INFO Channel 11/0 : 25[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-14:2525079:2525477 [4] NCCL INFO Channel 02/0 : 20[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934725:935118 [5] NCCL INFO Channel 11/0 : 29[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934723:935117 [3] NCCL INFO Channel 15/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-14:2525080:2525476 [5] NCCL INFO Channel 14/0 : 21[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934727:935114 [7] NCCL INFO Channel 01/0 : 31[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330190:3330589 [7] NCCL INFO Channel 00/0 : 7[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-14:2525082:2525472 [7] NCCL INFO Channel 04/0 : 23[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934721:935115 [1] NCCL INFO Channel 13/0 : 25[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330184:3330596 [1] NCCL INFO Channel 14/0 : 1[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934722:935116 [2] NCCL INFO Channel 03/0 : 26[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330188:3330595 [5] NCCL INFO Channel 00/0 : 21[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330185:3330593 [2] NCCL INFO Channel 04/0 : 2[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934725:935118 [5] NCCL INFO Channel 15/0 : 29[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330187:3330591 [4] NCCL INFO Channel 00/0 : 4[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-132:3330190:3330589 [7] NCCL INFO Channel 02/0 : 7[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330184:3330596 [1] NCCL INFO Channel 00/0 : 17[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934724:935113 [4] NCCL INFO Channel 01/0 : 28[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934727:935114 [7] NCCL INFO Channel 03/0 : 31[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934722:935116 [2] NCCL INFO Channel 05/0 : 26[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934721:935115 [1] NCCL INFO Channel 15/0 : 25[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934725:935118 [5] NCCL INFO Channel 01/0 : 13[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-14:2525079:2525477 [4] NCCL INFO Channel 06/0 : 20[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-132:3330188:3330595 [5] NCCL INFO Channel 02/0 : 21[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-14:2525076:2525478 [1] NCCL INFO Channel 00/0 : 1[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330185:3330593 [2] NCCL INFO Channel 06/0 : 2[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-14:2525082:2525472 [7] NCCL INFO Channel 06/0 : 23[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-14:2525080:2525476 [5] NCCL INFO Channel 00/0 : 5[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-14:2525075:2525471 [0] NCCL INFO Channel 02/0 : 16[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-14:2525077:2525475 [2] NCCL INFO Channel 06/0 : 18[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-14:2525081:2525473 [6] NCCL INFO Channel 00/0 : 22[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934724:935113 [4] NCCL INFO Channel 03/0 : 28[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-14:2525078:2525474 [3] NCCL INFO Channel 00/0 : 19[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934727:935114 [7] NCCL INFO Channel 05/0 : 31[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-14:2525079:2525477 [4] NCCL INFO Channel 08/0 : 20[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934722:935116 [2] NCCL INFO Channel 07/0 : 26[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330184:3330596 [1] NCCL INFO Channel 02/0 : 17[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330190:3330589 [7] NCCL INFO Channel 04/0 : 7[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330187:3330591 [4] NCCL INFO Channel 02/0 : 4[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934721:935115 [1] NCCL INFO Channel 01/0 : 9[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934726:935119 [6] NCCL INFO Channel 01/0 : 30[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934725:935118 [5] NCCL INFO Channel 03/0 : 13[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330189:3330594 [6] NCCL INFO Channel 00/0 : 6[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330188:3330595 [5] NCCL INFO Channel 04/0 : 21[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-14:2525076:2525478 [1] NCCL INFO Channel 02/0 : 1[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330183:3330590 [0] NCCL INFO Channel 02/0 : 0[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330185:3330593 [2] NCCL INFO Channel 08/0 : 2[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934724:935113 [4] NCCL INFO Channel 05/0 : 28[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934720:935120 [0] NCCL INFO Channel 01/0 : 24[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330184:3330596 [1] NCCL INFO Channel 04/0 : 17[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934727:935114 [7] NCCL INFO Channel 07/0 : 31[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-14:2525082:2525472 [7] NCCL INFO Channel 08/0 : 23[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934721:935115 [1] NCCL INFO Channel 03/0 : 9[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934722:935116 [2] NCCL INFO Channel 09/0 : 26[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-14:2525080:2525476 [5] NCCL INFO Channel 02/0 : 5[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934725:935118 [5] NCCL INFO Channel 05/0 : 13[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934726:935119 [6] NCCL INFO Channel 03/0 : 30[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330187:3330591 [4] NCCL INFO Channel 06/0 : 4[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-132:3330188:3330595 [5] NCCL INFO Channel 06/0 : 21[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-14:2525075:2525471 [0] NCCL INFO Channel 04/0 : 16[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-14:2525081:2525473 [6] NCCL INFO Channel 02/0 : 22[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-14:2525078:2525474 [3] NCCL INFO Channel 02/0 : 19[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330189:3330594 [6] NCCL INFO Channel 02/0 : 6[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-14:2525077:2525475 [2] NCCL INFO Channel 08/0 : 18[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-14:2525079:2525477 [4] NCCL INFO Channel 10/0 : 20[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-14:2525076:2525478 [1] NCCL INFO Channel 04/0 : 1[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-14:2525082:2525472 [7] NCCL INFO Channel 10/0 : 23[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-14:2525080:2525476 [5] NCCL INFO Channel 04/0 : 5[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934723:935117 [3] NCCL INFO Channel 01/0 : 27[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934724:935113 [4] NCCL INFO Channel 07/0 : 28[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-14:2525075:2525471 [0] NCCL INFO Channel 06/0 : 16[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934720:935120 [0] NCCL INFO Channel 03/0 : 24[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934727:935114 [7] NCCL INFO Channel 09/0 : 31[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330190:3330589 [7] NCCL INFO Channel 06/0 : 7[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330183:3330590 [0] NCCL INFO Channel 04/0 : 0[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934721:935115 [1] NCCL INFO Channel 05/0 : 9[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934722:935116 [2] NCCL INFO Channel 11/0 : 26[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330185:3330593 [2] NCCL INFO Channel 10/0 : 2[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330184:3330596 [1] NCCL INFO Channel 06/0 : 17[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330186:3330592 [3] NCCL INFO Channel 00/0 : 3[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934725:935118 [5] NCCL INFO Channel 07/0 : 13[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330188:3330595 [5] NCCL INFO Channel 08/0 : 21[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934726:935119 [6] NCCL INFO Channel 05/0 : 30[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330187:3330591 [4] NCCL INFO Channel 08/0 : 4[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-132:3330189:3330594 [6] NCCL INFO Channel 04/0 : 6[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-14:2525081:2525473 [6] NCCL INFO Channel 04/0 : 22[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-14:2525078:2525474 [3] NCCL INFO Channel 04/0 : 19[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-14:2525076:2525478 [1] NCCL INFO Channel 06/0 : 1[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934723:935117 [3] NCCL INFO Channel 03/0 : 27[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-14:2525079:2525477 [4] NCCL INFO Channel 12/0 : 20[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-14:2525077:2525475 [2] NCCL INFO Channel 10/0 : 18[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-14:2525082:2525472 [7] NCCL INFO Channel 12/0 : 23[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-14:2525080:2525476 [5] NCCL INFO Channel 06/0 : 5[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934720:935120 [0] NCCL INFO Channel 05/0 : 24[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934724:935113 [4] NCCL INFO Channel 09/0 : 28[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934727:935114 [7] NCCL INFO Channel 11/0 : 31[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330190:3330589 [7] NCCL INFO Channel 08/0 : 7[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-14:2525075:2525471 [0] NCCL INFO Channel 08/0 : 16[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934721:935115 [1] NCCL INFO Channel 07/0 : 9[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330184:3330596 [1] NCCL INFO Channel 08/0 : 17[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330183:3330590 [0] NCCL INFO Channel 06/0 : 0[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934722:935116 [2] NCCL INFO Channel 13/0 : 26[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330186:3330592 [3] NCCL INFO Channel 02/0 : 3[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330185:3330593 [2] NCCL INFO Channel 12/0 : 2[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330188:3330595 [5] NCCL INFO Channel 10/0 : 21[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330187:3330591 [4] NCCL INFO Channel 10/0 : 4[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-132:3330189:3330594 [6] NCCL INFO Channel 08/0 : 6[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-14:2525078:2525474 [3] NCCL INFO Channel 06/0 : 19[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-14:2525081:2525473 [6] NCCL INFO Channel 08/0 : 22[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934725:935118 [5] NCCL INFO Channel 09/0 : 13[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934726:935119 [6] NCCL INFO Channel 07/0 : 30[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934723:935117 [3] NCCL INFO Channel 05/0 : 27[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330190:3330589 [7] NCCL INFO Channel 10/0 : 7[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330184:3330596 [1] NCCL INFO Channel 10/0 : 17[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934727:935114 [7] NCCL INFO Channel 13/0 : 31[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934720:935120 [0] NCCL INFO Channel 07/0 : 24[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934724:935113 [4] NCCL INFO Channel 11/0 : 28[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-132:3330183:3330590 [0] NCCL INFO Channel 08/0 : 0[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934721:935115 [1] NCCL INFO Channel 11/0 : 9[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934722:935116 [2] NCCL INFO Channel 15/0 : 26[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-14:2525076:2525478 [1] NCCL INFO Channel 08/0 : 1[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330188:3330595 [5] NCCL INFO Channel 12/0 : 21[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330185:3330593 [2] NCCL INFO Channel 14/0 : 2[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-14:2525082:2525472 [7] NCCL INFO Channel 14/0 : 23[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-14:2525079:2525477 [4] NCCL INFO Channel 14/0 : 20[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-14:2525080:2525476 [5] NCCL INFO Channel 08/0 : 5[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934726:935119 [6] NCCL INFO Channel 09/0 : 30[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934725:935118 [5] NCCL INFO Channel 11/0 : 13[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330186:3330592 [3] NCCL INFO Channel 04/0 : 3[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-14:2525077:2525475 [2] NCCL INFO Channel 12/0 : 18[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934723:935117 [3] NCCL INFO Channel 07/0 : 27[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330187:3330591 [4] NCCL INFO Channel 12/0 : 4[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-14:2525078:2525474 [3] NCCL INFO Channel 08/0 : 19[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934727:935114 [7] NCCL INFO Channel 01/0 : 15[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934720:935120 [0] NCCL INFO Channel 09/0 : 24[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934724:935113 [4] NCCL INFO Channel 13/0 : 28[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-132:3330184:3330596 [1] NCCL INFO Channel 12/0 : 17[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330189:3330594 [6] NCCL INFO Channel 10/0 : 6[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330183:3330590 [0] NCCL INFO Channel 10/0 : 0[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-14:2525081:2525473 [6] NCCL INFO Channel 10/0 : 22[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-14:2525076:2525478 [1] NCCL INFO Channel 10/0 : 1[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330190:3330589 [7] NCCL INFO Channel 12/0 : 7[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-14:2525082:2525472 [7] NCCL INFO Channel 00/0 : 7[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934721:935115 [1] NCCL INFO Channel 13/0 : 9[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330188:3330595 [5] NCCL INFO Channel 14/0 : 21[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934722:935116 [2] NCCL INFO Channel 01/0 : 10[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330185:3330593 [2] NCCL INFO Channel 00/0 : 18[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-14:2525080:2525476 [5] NCCL INFO Channel 10/0 : 5[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934725:935118 [5] NCCL INFO Channel 15/0 : 13[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934726:935119 [6] NCCL INFO Channel 11/0 : 30[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934723:935117 [3] NCCL INFO Channel 09/0 : 27[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330186:3330592 [3] NCCL INFO Channel 06/0 : 3[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-14:2525079:2525477 [4] NCCL INFO Channel 00/0 : 4[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-14:2525077:2525475 [2] NCCL INFO Channel 14/0 : 18[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-14:2525081:2525473 [6] NCCL INFO Channel 12/0 : 22[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-14:2525078:2525474 [3] NCCL INFO Channel 10/0 : 19[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-14:2525076:2525478 [1] NCCL INFO Channel 12/0 : 1[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934727:935114 [7] NCCL INFO Channel 03/0 : 15[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330187:3330591 [4] NCCL INFO Channel 14/0 : 4[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-132:3330189:3330594 [6] NCCL INFO Channel 12/0 : 6[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-14:2525082:2525472 [7] NCCL INFO Channel 02/0 : 7[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934721:935115 [1] NCCL INFO Channel 15/0 : 9[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934722:935116 [2] NCCL INFO Channel 03/0 : 10[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330183:3330590 [0] NCCL INFO Channel 12/0 : 0[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330184:3330596 [1] NCCL INFO Channel 14/0 : 17[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-14:2525080:2525476 [5] NCCL INFO Channel 12/0 : 5[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934724:935113 [4] NCCL INFO Channel 15/0 : 28[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934720:935120 [0] NCCL INFO Channel 11/0 : 24[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330190:3330589 [7] NCCL INFO Channel 14/0 : 7[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-14:2525079:2525477 [4] NCCL INFO Channel 02/0 : 4[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934723:935117 [3] NCCL INFO Channel 13/0 : 27[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330185:3330593 [2] NCCL INFO Channel 04/0 : 18[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934727:935114 [7] NCCL INFO Channel 05/0 : 15[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934726:935119 [6] NCCL INFO Channel 13/0 : 30[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330186:3330592 [3] NCCL INFO Channel 08/0 : 3[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-14:2525077:2525475 [2] NCCL INFO Channel 00/0 : 2[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-14:2525081:2525473 [6] NCCL INFO Channel 14/0 : 22[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-14:2525078:2525474 [3] NCCL INFO Channel 12/0 : 19[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-14:2525076:2525478 [1] NCCL INFO Channel 14/0 : 1[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-14:2525082:2525472 [7] NCCL INFO Channel 04/0 : 7[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934722:935116 [2] NCCL INFO Channel 05/0 : 10[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330187:3330591 [4] NCCL INFO Channel 00/0 : 20[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-14:2525080:2525476 [5] NCCL INFO Channel 14/0 : 5[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-14:2525079:2525477 [4] NCCL INFO Channel 06/0 : 4[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-132:3330189:3330594 [6] NCCL INFO Channel 14/0 : 6[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330183:3330590 [0] NCCL INFO Channel 14/0 : 0[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330190:3330589 [7] NCCL INFO Channel 00/0 : 23[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-14:2525077:2525475 [2] NCCL INFO Channel 04/0 : 2[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330185:3330593 [2] NCCL INFO Channel 06/0 : 18[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934724:935113 [4] NCCL INFO Channel 01/0 : 12[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-132:3330186:3330592 [3] NCCL INFO Channel 10/0 : 3[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934723:935117 [3] NCCL INFO Channel 15/0 : 27[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934720:935120 [0] NCCL INFO Channel 13/0 : 24[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330187:3330591 [4] NCCL INFO Channel 02/0 : 20[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934727:935114 [7] NCCL INFO Channel 07/0 : 15[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934726:935119 [6] NCCL INFO Channel 15/0 : 30[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934725:935118 [5] NCCL INFO Channel 00/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-14:2525082:2525472 [7] NCCL INFO Channel 06/0 : 7[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934722:935116 [2] NCCL INFO Channel 07/0 : 10[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-14:2525081:2525473 [6] NCCL INFO Channel 00/0 : 6[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-14:2525078:2525474 [3] NCCL INFO Channel 14/0 : 19[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-14:2525079:2525477 [4] NCCL INFO Channel 08/0 : 4[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-132:3330189:3330594 [6] NCCL INFO Channel 00/0 : 22[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-14:2525077:2525475 [2] NCCL INFO Channel 06/0 : 2[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330183:3330590 [0] NCCL INFO Channel 02/0 : 16[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-31:3197181:3197587 [1] NCCL INFO Channel 00/0 : 25[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934724:935113 [4] NCCL INFO Channel 03/0 : 12[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-132:3330185:3330593 [2] NCCL INFO Channel 08/0 : 18[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330186:3330592 [3] NCCL INFO Channel 12/0 : 3[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330190:3330589 [7] NCCL INFO Channel 02/0 : 23[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330187:3330591 [4] NCCL INFO Channel 06/0 : 20[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934723:935117 [3] NCCL INFO Channel 01/0 : 11[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934721:935115 [1] NCCL INFO Channel 00/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330189:3330594 [6] NCCL INFO Channel 02/0 : 22[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330183:3330590 [0] NCCL INFO Channel 04/0 : 16[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-14:2525082:2525472 [7] NCCL INFO Channel 08/0 : 7[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-31:3197181:3197587 [1] NCCL INFO Channel 02/0 : 25[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-14:2525079:2525477 [4] NCCL INFO Channel 10/0 : 4[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-14:2525081:2525473 [6] NCCL INFO Channel 02/0 : 6[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934727:935114 [7] NCCL INFO Channel 09/0 : 15[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934726:935119 [6] NCCL INFO Channel 01/0 : 14[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330187:3330591 [4] NCCL INFO Channel 08/0 : 20[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-132:3330186:3330592 [3] NCCL INFO Channel 14/0 : 3[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330190:3330589 [7] NCCL INFO Channel 04/0 : 23[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-14:2525078:2525474 [3] NCCL INFO Channel 00/0 : 3[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-14:2525077:2525475 [2] NCCL INFO Channel 08/0 : 2[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934725:935118 [5] NCCL INFO Channel 01/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934720:935120 [0] NCCL INFO Channel 15/0 : 24[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934722:935116 [2] NCCL INFO Channel 09/0 : 10[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934724:935113 [4] NCCL INFO Channel 05/0 : 12[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-132:3330189:3330594 [6] NCCL INFO Channel 04/0 : 22[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-14:2525082:2525472 [7] NCCL INFO Channel 10/0 : 7[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330183:3330590 [0] NCCL INFO Channel 06/0 : 16[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-31:3197181:3197587 [1] NCCL INFO Channel 04/0 : 25[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-14:2525080:2525476 [5] NCCL INFO Channel 01/0 : 13[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-14:2525079:2525477 [4] NCCL INFO Channel 12/0 : 4[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934723:935117 [3] NCCL INFO Channel 03/0 : 11[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934721:935115 [1] NCCL INFO Channel 02/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934727:935114 [7] NCCL INFO Channel 11/0 : 15[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934726:935119 [6] NCCL INFO Channel 03/0 : 14[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-14:2525081:2525473 [6] NCCL INFO Channel 04/0 : 6[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330185:3330593 [2] NCCL INFO Channel 10/0 : 18[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-14:2525078:2525474 [3] NCCL INFO Channel 02/0 : 3[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-14:2525077:2525475 [2] NCCL INFO Channel 10/0 : 2[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-14:2525082:2525472 [7] NCCL INFO Channel 12/0 : 7[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934725:935118 [5] NCCL INFO Channel 02/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-31:3197181:3197587 [1] NCCL INFO Channel 06/0 : 25[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-14:2525080:2525476 [5] NCCL INFO Channel 03/0 : 13[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934722:935116 [2] NCCL INFO Channel 11/0 : 10[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934724:935113 [4] NCCL INFO Channel 07/0 : 12[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934720:935120 [0] NCCL INFO Channel 01/0 : 8[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934723:935117 [3] NCCL INFO Channel 05/0 : 11[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-14:2525079:2525477 [4] NCCL INFO Channel 14/0 : 4[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934721:935115 [1] NCCL INFO Channel 03/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330187:3330591 [4] NCCL INFO Channel 10/0 : 20[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-14:2525081:2525473 [6] NCCL INFO Channel 08/0 : 6[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330189:3330594 [6] NCCL INFO Channel 08/0 : 22[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330186:3330592 [3] NCCL INFO Channel 00/0 : 19[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330190:3330589 [7] NCCL INFO Channel 06/0 : 23[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330183:3330590 [0] NCCL INFO Channel 08/0 : 16[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330185:3330593 [2] NCCL INFO Channel 12/0 : 18[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330188:3330595 [5] NCCL INFO Channel 00/0 : 29[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-31:3197181:3197587 [1] NCCL INFO Channel 08/0 : 25[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-14:2525082:2525472 [7] NCCL INFO Channel 14/0 : 7[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934727:935114 [7] NCCL INFO Channel 13/0 : 15[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-14:2525078:2525474 [3] NCCL INFO Channel 04/0 : 3[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934726:935119 [6] NCCL INFO Channel 05/0 : 14[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-14:2525077:2525475 [2] NCCL INFO Channel 12/0 : 2[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934725:935118 [5] NCCL INFO Channel 03/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330187:3330591 [4] NCCL INFO Channel 12/0 : 20[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-14:2525080:2525476 [5] NCCL INFO Channel 05/0 : 13[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-14:2525081:2525473 [6] NCCL INFO Channel 10/0 : 6[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934722:935116 [2] NCCL INFO Channel 13/0 : 10[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-14:2525078:2525474 [3] NCCL INFO Channel 06/0 : 3[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330189:3330594 [6] NCCL INFO Channel 10/0 : 22[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-31:3197181:3197587 [1] NCCL INFO Channel 10/0 : 25[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-14:2525077:2525475 [2] NCCL INFO Channel 14/0 : 2[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934723:935117 [3] NCCL INFO Channel 07/0 : 11[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934724:935113 [4] NCCL INFO Channel 09/0 : 12[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934721:935115 [1] NCCL INFO Channel 04/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934720:935120 [0] NCCL INFO Channel 03/0 : 8[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-14:2525081:2525473 [6] NCCL INFO Channel 12/0 : 6[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934726:935119 [6] NCCL INFO Channel 07/0 : 14[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-14:2525080:2525476 [5] NCCL INFO Channel 07/0 : 13[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934725:935118 [5] NCCL INFO Channel 04/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934722:935116 [2] NCCL INFO Channel 15/0 : 10[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330186:3330592 [3] NCCL INFO Channel 02/0 : 19[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330183:3330590 [0] NCCL INFO Channel 10/0 : 16[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330185:3330593 [2] NCCL INFO Channel 14/0 : 18[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-14:2525078:2525474 [3] NCCL INFO Channel 08/0 : 3[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330190:3330589 [7] NCCL INFO Channel 08/0 : 23[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330187:3330591 [4] NCCL INFO Channel 14/0 : 20[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-132:3330188:3330595 [5] NCCL INFO Channel 02/0 : 29[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-31:3197181:3197587 [1] NCCL INFO Channel 12/0 : 25[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-14:2525081:2525473 [6] NCCL INFO Channel 14/0 : 6[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934723:935117 [3] NCCL INFO Channel 09/0 : 11[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-14:2525080:2525476 [5] NCCL INFO Channel 09/0 : 13[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330189:3330594 [6] NCCL INFO Channel 12/0 : 22[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-14:2525078:2525474 [3] NCCL INFO Channel 10/0 : 3[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330186:3330592 [3] NCCL INFO Channel 04/0 : 19[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330183:3330590 [0] NCCL INFO Channel 12/0 : 16[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-31:3197187:3197590 [7] NCCL INFO Channel 00/0 : 31[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934720:935120 [0] NCCL INFO Channel 05/0 : 8[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934721:935115 [1] NCCL INFO Channel 05/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934724:935113 [4] NCCL INFO Channel 11/0 : 12[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934726:935119 [6] NCCL INFO Channel 09/0 : 14[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934725:935118 [5] NCCL INFO Channel 06/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934727:935114 [7] NCCL INFO Channel 00/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-14:2525080:2525476 [5] NCCL INFO Channel 11/0 : 13[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330190:3330589 [7] NCCL INFO Channel 10/0 : 23[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330188:3330595 [5] NCCL INFO Channel 04/0 : 29[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330189:3330594 [6] NCCL INFO Channel 14/0 : 22[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-31:3197181:3197587 [1] NCCL INFO Channel 14/0 : 25[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-14:2525078:2525474 [3] NCCL INFO Channel 12/0 : 3[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934723:935117 [3] NCCL INFO Channel 13/0 : 11[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-14:2525079:2525477 [4] NCCL INFO Channel 01/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-132:3330183:3330590 [0] NCCL INFO Channel 14/0 : 16[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330186:3330592 [3] NCCL INFO Channel 06/0 : 19[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-31:3197187:3197590 [7] NCCL INFO Channel 02/0 : 31[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330190:3330589 [7] NCCL INFO Channel 12/0 : 23[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330188:3330595 [5] NCCL INFO Channel 06/0 : 29[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-31:3197182:3197588 [2] NCCL INFO Channel 00/0 : 26[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934721:935115 [1] NCCL INFO Channel 06/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330187:3330591 [4] NCCL INFO Channel 00/0 : 28[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-14:2525080:2525476 [5] NCCL INFO Channel 15/0 : 13[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934720:935120 [0] NCCL INFO Channel 07/0 : 8[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934724:935113 [4] NCCL INFO Channel 13/0 : 12[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-14:2525078:2525474 [3] NCCL INFO Channel 14/0 : 3[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-14:2525077:2525475 [2] NCCL INFO Channel 01/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-14:2525079:2525477 [4] NCCL INFO Channel 03/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-132:3330188:3330595 [5] NCCL INFO Channel 08/0 : 29[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-31:3197187:3197590 [7] NCCL INFO Channel 04/0 : 31[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-31:3197182:3197588 [2] NCCL INFO Channel 04/0 : 26[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934726:935119 [6] NCCL INFO Channel 11/0 : 14[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934725:935118 [5] NCCL INFO Channel 07/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-14:2525077:2525475 [2] NCCL INFO Channel 03/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934722:935116 [2] NCCL INFO Channel 00/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934727:935114 [7] NCCL INFO Channel 01/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934723:935117 [3] NCCL INFO Channel 15/0 : 11[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330190:3330589 [7] NCCL INFO Channel 14/0 : 23[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-14:2525081:2525473 [6] NCCL INFO Channel 01/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934721:935115 [1] NCCL INFO Channel 07/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330187:3330591 [4] NCCL INFO Channel 02/0 : 28[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-132:3330185:3330593 [2] NCCL INFO Channel 00/0 : 26[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330186:3330592 [3] NCCL INFO Channel 08/0 : 19[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-14:2525079:2525477 [4] NCCL INFO Channel 05/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-132:3330189:3330594 [6] NCCL INFO Channel 00/0 : 30[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-31:3197187:3197590 [7] NCCL INFO Channel 06/0 : 31[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-14:2525076:2525478 [1] NCCL INFO Channel 01/0 : 9[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-31:3197182:3197588 [2] NCCL INFO Channel 06/0 : 26[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-14:2525077:2525475 [2] NCCL INFO Channel 05/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-31:3197187:3197590 [7] NCCL INFO Channel 08/0 : 31[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-14:2525081:2525473 [6] NCCL INFO Channel 03/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934724:935113 [4] NCCL INFO Channel 15/0 : 12[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-132:3330188:3330595 [5] NCCL INFO Channel 10/0 : 29[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-14:2525079:2525477 [4] NCCL INFO Channel 07/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934720:935120 [0] NCCL INFO Channel 09/0 : 8[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330187:3330591 [4] NCCL INFO Channel 06/0 : 28[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-14:2525076:2525478 [1] NCCL INFO Channel 03/0 : 9[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934726:935119 [6] NCCL INFO Channel 13/0 : 14[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330186:3330592 [3] NCCL INFO Channel 10/0 : 19[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-14:2525075:2525471 [0] NCCL INFO Channel 10/0 : 16[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934722:935116 [2] NCCL INFO Channel 01/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934725:935118 [5] NCCL INFO Channel 08/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330185:3330593 [2] NCCL INFO Channel 04/0 : 26[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-31:3197182:3197588 [2] NCCL INFO Channel 08/0 : 26[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934727:935114 [7] NCCL INFO Channel 02/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330189:3330594 [6] NCCL INFO Channel 02/0 : 30[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330184:3330596 [1] NCCL INFO Channel 00/0 : 25[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-31:3197187:3197590 [7] NCCL INFO Channel 10/0 : 31[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-14:2525077:2525475 [2] NCCL INFO Channel 07/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-14:2525081:2525473 [6] NCCL INFO Channel 05/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-14:2525079:2525477 [4] NCCL INFO Channel 09/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934721:935115 [1] NCCL INFO Channel 08/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330188:3330595 [5] NCCL INFO Channel 12/0 : 29[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-14:2525082:2525472 [7] NCCL INFO Channel 01/0 : 15[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934720:935120 [0] NCCL INFO Channel 11/0 : 8[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-14:2525076:2525478 [1] NCCL INFO Channel 05/0 : 9[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-14:2525075:2525471 [0] NCCL INFO Channel 12/0 : 16[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934726:935119 [6] NCCL INFO Channel 15/0 : 14[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-31:3197183:3197592 [3] NCCL INFO Channel 00/0 : 27[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-31:3197182:3197588 [2] NCCL INFO Channel 10/0 : 26[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330187:3330591 [4] NCCL INFO Channel 08/0 : 28[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-132:3330186:3330592 [3] NCCL INFO Channel 12/0 : 19[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-31:3197187:3197590 [7] NCCL INFO Channel 12/0 : 31[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330185:3330593 [2] NCCL INFO Channel 06/0 : 26[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-31:3197183:3197592 [3] NCCL INFO Channel 02/0 : 27[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-14:2525077:2525475 [2] NCCL INFO Channel 09/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330189:3330594 [6] NCCL INFO Channel 04/0 : 30[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-14:2525079:2525477 [4] NCCL INFO Channel 11/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934723:935117 [3] NCCL INFO Channel 00/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330190:3330589 [7] NCCL INFO Channel 00/0 : 31[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330184:3330596 [1] NCCL INFO Channel 02/0 : 25[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-14:2525081:2525473 [6] NCCL INFO Channel 07/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934722:935116 [2] NCCL INFO Channel 03/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330188:3330595 [5] NCCL INFO Channel 14/0 : 29[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-14:2525082:2525472 [7] NCCL INFO Channel 03/0 : 15[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934725:935118 [5] NCCL INFO Channel 09/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934727:935114 [7] NCCL INFO Channel 03/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934720:935120 [0] NCCL INFO Channel 13/0 : 8[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-14:2525076:2525478 [1] NCCL INFO Channel 07/0 : 9[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-14:2525075:2525471 [0] NCCL INFO Channel 14/0 : 16[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934721:935115 [1] NCCL INFO Channel 10/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-31:3197182:3197588 [2] NCCL INFO Channel 12/0 : 26[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-31:3197187:3197590 [7] NCCL INFO Channel 14/0 : 31[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934724:935113 [4] NCCL INFO Channel 00/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-31:3197184:3197586 [4] NCCL INFO Channel 00/0 : 28[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934723:935117 [3] NCCL INFO Channel 01/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330186:3330592 [3] NCCL INFO Channel 14/0 : 19[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-31:3197183:3197592 [3] NCCL INFO Channel 04/0 : 27[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934722:935116 [2] NCCL INFO Channel 04/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330187:3330591 [4] NCCL INFO Channel 10/0 : 28[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-132:3330185:3330593 [2] NCCL INFO Channel 08/0 : 26[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-14:2525077:2525475 [2] NCCL INFO Channel 11/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934725:935118 [5] NCCL INFO Channel 10/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330189:3330594 [6] NCCL INFO Channel 08/0 : 30[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-14:2525079:2525477 [4] NCCL INFO Channel 13/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934720:935120 [0] NCCL INFO Channel 15/0 : 8[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330190:3330589 [7] NCCL INFO Channel 02/0 : 31[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330184:3330596 [1] NCCL INFO Channel 04/0 : 25[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-14:2525081:2525473 [6] NCCL INFO Channel 09/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934727:935114 [7] NCCL INFO Channel 04/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330188:3330595 [5] NCCL INFO Channel 00/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-31:3197184:3197586 [4] NCCL INFO Channel 02/0 : 28[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-31:3197182:3197588 [2] NCCL INFO Channel 14/0 : 26[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-14:2525082:2525472 [7] NCCL INFO Channel 05/0 : 15[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-31:3197183:3197592 [3] NCCL INFO Channel 06/0 : 27[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-14:2525075:2525471 [0] NCCL INFO Channel 02/0 : 0[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-31:3197184:3197586 [4] NCCL INFO Channel 06/0 : 28[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-14:2525076:2525478 [1] NCCL INFO Channel 11/0 : 9[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-14:2525077:2525475 [2] NCCL INFO Channel 13/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934721:935115 [1] NCCL INFO Channel 11/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934724:935113 [4] NCCL INFO Channel 01/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-132:3330187:3330591 [4] NCCL INFO Channel 12/0 : 28[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934723:935117 [3] NCCL INFO Channel 02/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330189:3330594 [6] NCCL INFO Channel 10/0 : 30[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330185:3330593 [2] NCCL INFO Channel 10/0 : 26[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-14:2525079:2525477 [4] NCCL INFO Channel 15/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-14:2525081:2525473 [6] NCCL INFO Channel 11/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934722:935116 [2] NCCL INFO Channel 05/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330184:3330596 [1] NCCL INFO Channel 06/0 : 25[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330190:3330589 [7] NCCL INFO Channel 04/0 : 31[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330188:3330595 [5] NCCL INFO Channel 01/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-31:3197184:3197586 [4] NCCL INFO Channel 08/0 : 28[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934726:935119 [6] NCCL INFO Channel 00/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934725:935118 [5] NCCL INFO Channel 11/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330187:3330591 [4] NCCL INFO Channel 14/0 : 28[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934727:935114 [7] NCCL INFO Channel 05/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330189:3330594 [6] NCCL INFO Channel 12/0 : 30[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330185:3330593 [2] NCCL INFO Channel 12/0 : 26[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-14:2525082:2525472 [7] NCCL INFO Channel 07/0 : 15[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-14:2525078:2525474 [3] NCCL INFO Channel 01/0 : 11[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-14:2525075:2525471 [0] NCCL INFO Channel 04/0 : 0[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934721:935115 [1] NCCL INFO Channel 12/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330186:3330592 [3] NCCL INFO Channel 00/0 : 27[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-31:3197183:3197592 [3] NCCL INFO Channel 08/0 : 27[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-14:2525076:2525478 [1] NCCL INFO Channel 13/0 : 9[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934724:935113 [4] NCCL INFO Channel 02/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-14:2525077:2525475 [2] NCCL INFO Channel 15/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934723:935117 [3] NCCL INFO Channel 04/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934722:935116 [2] NCCL INFO Channel 06/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-14:2525081:2525473 [6] NCCL INFO Channel 13/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934726:935119 [6] NCCL INFO Channel 01/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934725:935118 [5] NCCL INFO Channel 12/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934727:935114 [7] NCCL INFO Channel 06/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330188:3330595 [5] NCCL INFO Channel 02/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-14:2525082:2525472 [7] NCCL INFO Channel 09/0 : 15[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330184:3330596 [1] NCCL INFO Channel 08/0 : 25[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330190:3330589 [7] NCCL INFO Channel 06/0 : 31[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-31:3197183:3197592 [3] NCCL INFO Channel 10/0 : 27[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-31:3197184:3197586 [4] NCCL INFO Channel 10/0 : 28[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-132:3330187:3330591 [4] NCCL INFO Channel 00/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-14:2525078:2525474 [3] NCCL INFO Channel 03/0 : 11[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-14:2525075:2525471 [0] NCCL INFO Channel 06/0 : 0[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934721:935115 [1] NCCL INFO Channel 13/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330185:3330593 [2] NCCL INFO Channel 14/0 : 26[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-14:2525076:2525478 [1] NCCL INFO Channel 15/0 : 9[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-14:2525081:2525473 [6] NCCL INFO Channel 15/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934724:935113 [4] NCCL INFO Channel 03/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-132:3330189:3330594 [6] NCCL INFO Channel 14/0 : 30[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934720:935120 [0] NCCL INFO Channel 01/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330186:3330592 [3] NCCL INFO Channel 02/0 : 27[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-14:2525082:2525472 [7] NCCL INFO Channel 11/0 : 15[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934722:935116 [2] NCCL INFO Channel 07/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934723:935117 [3] NCCL INFO Channel 05/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330188:3330595 [5] NCCL INFO Channel 03/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-31:3197183:3197592 [3] NCCL INFO Channel 12/0 : 27[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934725:935118 [5] NCCL INFO Channel 14/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934726:935119 [6] NCCL INFO Channel 02/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330190:3330589 [7] NCCL INFO Channel 08/0 : 31[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330184:3330596 [1] NCCL INFO Channel 10/0 : 25[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-31:3197184:3197586 [4] NCCL INFO Channel 12/0 : 28[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934727:935114 [7] NCCL INFO Channel 08/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934721:935115 [1] NCCL INFO Channel 14/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934724:935113 [4] NCCL INFO Channel 05/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-14:2525075:2525471 [0] NCCL INFO Channel 08/0 : 0[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-14:2525078:2525474 [3] NCCL INFO Channel 05/0 : 11[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934720:935120 [0] NCCL INFO Channel 02/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934722:935116 [2] NCCL INFO Channel 08/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330187:3330591 [4] NCCL INFO Channel 01/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-14:2525082:2525472 [7] NCCL INFO Channel 13/0 : 15[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330185:3330593 [2] NCCL INFO Channel 00/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330186:3330592 [3] NCCL INFO Channel 04/0 : 27[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330189:3330594 [6] NCCL INFO Channel 00/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-31:3197183:3197592 [3] NCCL INFO Channel 14/0 : 27[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330188:3330595 [5] NCCL INFO Channel 04/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-31:3197184:3197586 [4] NCCL INFO Channel 14/0 : 28[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-14:2525075:2525471 [0] NCCL INFO Channel 10/0 : 0[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330190:3330589 [7] NCCL INFO Channel 10/0 : 31[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330184:3330596 [1] NCCL INFO Channel 12/0 : 25[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-14:2525078:2525474 [3] NCCL INFO Channel 07/0 : 11[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934723:935117 [3] NCCL INFO Channel 06/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330187:3330591 [4] NCCL INFO Channel 02/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934726:935119 [6] NCCL INFO Channel 03/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934725:935118 [5] NCCL INFO Channel 15/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-14:2525075:2525471 [0] NCCL INFO Channel 12/0 : 0[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934727:935114 [7] NCCL INFO Channel 09/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934721:935115 [1] NCCL INFO Channel 15/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-14:2525078:2525474 [3] NCCL INFO Channel 09/0 : 11[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934724:935113 [4] NCCL INFO Channel 06/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934720:935120 [0] NCCL INFO Channel 03/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934722:935116 [2] NCCL INFO Channel 09/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330185:3330593 [2] NCCL INFO Channel 01/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330186:3330592 [3] NCCL INFO Channel 06/0 : 27[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330189:3330594 [6] NCCL INFO Channel 01/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-14:2525075:2525471 [0] NCCL INFO Channel 14/0 : 0[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330188:3330595 [5] NCCL INFO Channel 06/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-14:2525078:2525474 [3] NCCL INFO Channel 13/0 : 11[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330190:3330589 [7] NCCL INFO Channel 12/0 : 31[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330187:3330591 [4] NCCL INFO Channel 03/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934726:935119 [6] NCCL INFO Channel 04/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330184:3330596 [1] NCCL INFO Channel 14/0 : 25[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934723:935117 [3] NCCL INFO Channel 07/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330185:3330593 [2] NCCL INFO Channel 03/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934725:935118 [5] NCCL INFO Channel 01/0 : 13[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934727:935114 [7] NCCL INFO Channel 10/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330186:3330592 [3] NCCL INFO Channel 08/0 : 27[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934721:935115 [1] NCCL INFO Channel 01/0 : 9[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-14:2525078:2525474 [3] NCCL INFO Channel 15/0 : 11[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934720:935120 [0] NCCL INFO Channel 04/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934722:935116 [2] NCCL INFO Channel 11/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934724:935113 [4] NCCL INFO Channel 07/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934726:935119 [6] NCCL INFO Channel 05/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330189:3330594 [6] NCCL INFO Channel 02/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330188:3330595 [5] NCCL INFO Channel 07/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934723:935117 [3] NCCL INFO Channel 08/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330190:3330589 [7] NCCL INFO Channel 14/0 : 31[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330184:3330596 [1] NCCL INFO Channel 00/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330187:3330591 [4] NCCL INFO Channel 05/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934725:935118 [5] NCCL INFO Channel 03/0 : 13[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934721:935115 [1] NCCL INFO Channel 03/0 : 9[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934727:935114 [7] NCCL INFO Channel 11/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330185:3330593 [2] NCCL INFO Channel 04/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330186:3330592 [3] NCCL INFO Channel 10/0 : 27[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934722:935116 [2] NCCL INFO Channel 12/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934720:935120 [0] NCCL INFO Channel 05/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934724:935113 [4] NCCL INFO Channel 08/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934726:935119 [6] NCCL INFO Channel 07/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330189:3330594 [6] NCCL INFO Channel 03/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330188:3330595 [5] NCCL INFO Channel 08/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-14:2525075:2525471 [0] NCCL INFO Channel 01/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934723:935117 [3] NCCL INFO Channel 09/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934725:935118 [5] NCCL INFO Channel 05/0 : 13[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330184:3330596 [1] NCCL INFO Channel 02/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330187:3330591 [4] NCCL INFO Channel 06/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-132:3330190:3330589 [7] NCCL INFO Channel 00/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330186:3330592 [3] NCCL INFO Channel 12/0 : 27[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330189:3330594 [6] NCCL INFO Channel 04/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330188:3330595 [5] NCCL INFO Channel 09/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330185:3330593 [2] NCCL INFO Channel 05/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330184:3330596 [1] NCCL INFO Channel 03/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330187:3330591 [4] NCCL INFO Channel 07/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-14:2525075:2525471 [0] NCCL INFO Channel 03/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934721:935115 [1] NCCL INFO Channel 05/0 : 9[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330190:3330589 [7] NCCL INFO Channel 01/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934727:935114 [7] NCCL INFO Channel 12/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934722:935116 [2] NCCL INFO Channel 13/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934720:935120 [0] NCCL INFO Channel 06/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934724:935113 [4] NCCL INFO Channel 09/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934723:935117 [3] NCCL INFO Channel 10/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934726:935119 [6] NCCL INFO Channel 08/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934725:935118 [5] NCCL INFO Channel 07/0 : 13[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330189:3330594 [6] NCCL INFO Channel 05/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330186:3330592 [3] NCCL INFO Channel 14/0 : 27[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-14:2525075:2525471 [0] NCCL INFO Channel 05/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330185:3330593 [2] NCCL INFO Channel 06/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330188:3330595 [5] NCCL INFO Channel 10/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330183:3330590 [0] NCCL INFO Channel 02/0 : 24[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934721:935115 [1] NCCL INFO Channel 07/0 : 9[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934727:935114 [7] NCCL INFO Channel 13/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934722:935116 [2] NCCL INFO Channel 14/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-14:2525075:2525471 [0] NCCL INFO Channel 07/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934720:935120 [0] NCCL INFO Channel 07/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330184:3330596 [1] NCCL INFO Channel 04/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330187:3330591 [4] NCCL INFO Channel 08/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-132:3330190:3330589 [7] NCCL INFO Channel 02/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934724:935113 [4] NCCL INFO Channel 10/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934726:935119 [6] NCCL INFO Channel 09/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934723:935117 [3] NCCL INFO Channel 12/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934725:935118 [5] NCCL INFO Channel 09/0 : 13[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934721:935115 [1] NCCL INFO Channel 11/0 : 9[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330186:3330592 [3] NCCL INFO Channel 00/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330189:3330594 [6] NCCL INFO Channel 07/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-14:2525075:2525471 [0] NCCL INFO Channel 09/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934727:935114 [7] NCCL INFO Channel 14/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934722:935116 [2] NCCL INFO Channel 15/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934720:935120 [0] NCCL INFO Channel 09/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330188:3330595 [5] NCCL INFO Channel 11/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330185:3330593 [2] NCCL INFO Channel 07/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330183:3330590 [0] NCCL INFO Channel 04/0 : 24[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330184:3330596 [1] NCCL INFO Channel 05/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330187:3330591 [4] NCCL INFO Channel 09/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-132:3330190:3330589 [7] NCCL INFO Channel 03/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-14:2525075:2525471 [0] NCCL INFO Channel 11/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934726:935119 [6] NCCL INFO Channel 10/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934723:935117 [3] NCCL INFO Channel 13/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934724:935113 [4] NCCL INFO Channel 11/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-132:3330186:3330592 [3] NCCL INFO Channel 01/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330189:3330594 [6] NCCL INFO Channel 08/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-14:2525075:2525471 [0] NCCL INFO Channel 13/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934725:935118 [5] NCCL INFO Channel 11/0 : 13[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934721:935115 [1] NCCL INFO Channel 13/0 : 9[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934722:935116 [2] NCCL INFO Channel 01/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330188:3330595 [5] NCCL INFO Channel 12/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934720:935120 [0] NCCL INFO Channel 10/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934727:935114 [7] NCCL INFO Channel 01/0 : 15[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934726:935119 [6] NCCL INFO Channel 11/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-14:2525075:2525471 [0] NCCL INFO Channel 15/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330184:3330596 [1] NCCL INFO Channel 06/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330187:3330591 [4] NCCL INFO Channel 10/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-132:3330183:3330590 [0] NCCL INFO Channel 06/0 : 24[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330185:3330593 [2] NCCL INFO Channel 08/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934723:935117 [3] NCCL INFO Channel 14/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934724:935113 [4] NCCL INFO Channel 13/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934725:935118 [5] NCCL INFO Channel 15/0 : 13[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934721:935115 [1] NCCL INFO Channel 15/0 : 9[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330190:3330589 [7] NCCL INFO Channel 04/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330186:3330592 [3] NCCL INFO Channel 02/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330189:3330594 [6] NCCL INFO Channel 09/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330188:3330595 [5] NCCL INFO Channel 14/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934722:935116 [2] NCCL INFO Channel 03/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934727:935114 [7] NCCL INFO Channel 03/0 : 15[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934720:935120 [0] NCCL INFO Channel 11/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934726:935119 [6] NCCL INFO Channel 12/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934724:935113 [4] NCCL INFO Channel 14/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934723:935117 [3] NCCL INFO Channel 15/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330184:3330596 [1] NCCL INFO Channel 07/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934722:935116 [2] NCCL INFO Channel 05/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934727:935114 [7] NCCL INFO Channel 05/0 : 15[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330183:3330590 [0] NCCL INFO Channel 08/0 : 24[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330185:3330593 [2] NCCL INFO Channel 09/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934720:935120 [0] NCCL INFO Channel 12/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934726:935119 [6] NCCL INFO Channel 13/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934723:935117 [3] NCCL INFO Channel 01/0 : 11[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934724:935113 [4] NCCL INFO Channel 15/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-132:3330190:3330589 [7] NCCL INFO Channel 05/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330187:3330591 [4] NCCL INFO Channel 11/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-132:3330189:3330594 [6] NCCL INFO Channel 10/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330186:3330592 [3] NCCL INFO Channel 04/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330188:3330595 [5] NCCL INFO Channel 15/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330184:3330596 [1] NCCL INFO Channel 08/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934722:935116 [2] NCCL INFO Channel 07/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934727:935114 [7] NCCL INFO Channel 07/0 : 15[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330183:3330590 [0] NCCL INFO Channel 10/0 : 24[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934720:935120 [0] NCCL INFO Channel 13/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934723:935117 [3] NCCL INFO Channel 03/0 : 11[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934724:935113 [4] NCCL INFO Channel 01/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934726:935119 [6] NCCL INFO Channel 15/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330185:3330593 [2] NCCL INFO Channel 11/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934727:935114 [7] NCCL INFO Channel 09/0 : 15[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934722:935116 [2] NCCL INFO Channel 09/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934720:935120 [0] NCCL INFO Channel 14/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934723:935117 [3] NCCL INFO Channel 05/0 : 11[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330189:3330594 [6] NCCL INFO Channel 11/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330187:3330591 [4] NCCL INFO Channel 13/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-132:3330184:3330596 [1] NCCL INFO Channel 10/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330190:3330589 [7] NCCL INFO Channel 06/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330186:3330592 [3] NCCL INFO Channel 05/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330183:3330590 [0] NCCL INFO Channel 12/0 : 24[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934724:935113 [4] NCCL INFO Channel 03/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934726:935119 [6] NCCL INFO Channel 01/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934722:935116 [2] NCCL INFO Channel 11/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934727:935114 [7] NCCL INFO Channel 11/0 : 15[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330185:3330593 [2] NCCL INFO Channel 12/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330184:3330596 [1] NCCL INFO Channel 11/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330189:3330594 [6] NCCL INFO Channel 12/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934720:935120 [0] NCCL INFO Channel 15/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934723:935117 [3] NCCL INFO Channel 07/0 : 11[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-31:3197185:3197591 [5] NCCL INFO Channel 00/0 : 29[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330187:3330591 [4] NCCL INFO Channel 14/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-132:3330190:3330589 [7] NCCL INFO Channel 08/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330186:3330592 [3] NCCL INFO Channel 06/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330183:3330590 [0] NCCL INFO Channel 14/0 : 24[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934724:935113 [4] NCCL INFO Channel 05/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-31:3197185:3197591 [5] NCCL INFO Channel 02/0 : 29[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934726:935119 [6] NCCL INFO Channel 03/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934722:935116 [2] NCCL INFO Channel 13/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934727:935114 [7] NCCL INFO Channel 13/0 : 15[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934723:935117 [3] NCCL INFO Channel 09/0 : 11[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934720:935120 [0] NCCL INFO Channel 01/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330189:3330594 [6] NCCL INFO Channel 13/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-31:3197185:3197591 [5] NCCL INFO Channel 04/0 : 29[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934724:935113 [4] NCCL INFO Channel 07/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934726:935119 [6] NCCL INFO Channel 05/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330185:3330593 [2] NCCL INFO Channel 13/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934723:935117 [3] NCCL INFO Channel 13/0 : 11[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330184:3330596 [1] NCCL INFO Channel 12/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330186:3330592 [3] NCCL INFO Channel 07/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330187:3330591 [4] NCCL INFO Channel 15/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-132:3330190:3330589 [7] NCCL INFO Channel 09/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330183:3330590 [0] NCCL INFO Channel 01/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934722:935116 [2] NCCL INFO Channel 15/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934720:935120 [0] NCCL INFO Channel 03/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934724:935113 [4] NCCL INFO Channel 09/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934726:935119 [6] NCCL INFO Channel 07/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330189:3330594 [6] NCCL INFO Channel 15/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-31:3197185:3197591 [5] NCCL INFO Channel 06/0 : 29[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330185:3330593 [2] NCCL INFO Channel 14/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330184:3330596 [1] NCCL INFO Channel 13/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934723:935117 [3] NCCL INFO Channel 15/0 : 11[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934720:935120 [0] NCCL INFO Channel 05/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330186:3330592 [3] NCCL INFO Channel 08/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934724:935113 [4] NCCL INFO Channel 11/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934726:935119 [6] NCCL INFO Channel 09/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-31:3197185:3197591 [5] NCCL INFO Channel 08/0 : 29[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330183:3330590 [0] NCCL INFO Channel 02/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330190:3330589 [7] NCCL INFO Channel 10/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330185:3330593 [2] NCCL INFO Channel 15/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330184:3330596 [1] NCCL INFO Channel 14/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934720:935120 [0] NCCL INFO Channel 07/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934724:935113 [4] NCCL INFO Channel 13/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934726:935119 [6] NCCL INFO Channel 11/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330183:3330590 [0] NCCL INFO Channel 03/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-31:3197185:3197591 [5] NCCL INFO Channel 10/0 : 29[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934724:935113 [4] NCCL INFO Channel 15/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-132:3330190:3330589 [7] NCCL INFO Channel 11/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934726:935119 [6] NCCL INFO Channel 13/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330184:3330596 [1] NCCL INFO Channel 15/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934720:935120 [0] NCCL INFO Channel 09/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330186:3330592 [3] NCCL INFO Channel 09/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-31:3197185:3197591 [5] NCCL INFO Channel 12/0 : 29[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934726:935119 [6] NCCL INFO Channel 15/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330183:3330590 [0] NCCL INFO Channel 04/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330190:3330589 [7] NCCL INFO Channel 12/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330186:3330592 [3] NCCL INFO Channel 10/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934720:935120 [0] NCCL INFO Channel 11/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934720:935120 [0] NCCL INFO Channel 13/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330183:3330590 [0] NCCL INFO Channel 05/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330190:3330589 [7] NCCL INFO Channel 13/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-31:3197185:3197591 [5] NCCL INFO Channel 14/0 : 29[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330186:3330592 [3] NCCL INFO Channel 12/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330183:3330590 [0] NCCL INFO Channel 06/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330190:3330589 [7] NCCL INFO Channel 14/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934720:935120 [0] NCCL INFO Channel 15/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330186:3330592 [3] NCCL INFO Channel 13/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330183:3330590 [0] NCCL INFO Channel 07/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330186:3330592 [3] NCCL INFO Channel 14/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330183:3330590 [0] NCCL INFO Channel 09/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330186:3330592 [3] NCCL INFO Channel 15/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330183:3330590 [0] NCCL INFO Channel 10/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330183:3330590 [0] NCCL INFO Channel 11/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330183:3330590 [0] NCCL INFO Channel 12/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330183:3330590 [0] NCCL INFO Channel 13/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330183:3330590 [0] NCCL INFO Channel 14/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330183:3330590 [0] NCCL INFO Channel 15/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-31:3197180:3197589 [0] NCCL INFO Channel 02/0 : 24[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-31:3197186:3197593 [6] NCCL INFO Channel 00/0 : 30[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-31:3197180:3197589 [0] NCCL INFO Channel 04/0 : 24[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-31:3197180:3197589 [0] NCCL INFO Channel 06/0 : 24[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-31:3197186:3197593 [6] NCCL INFO Channel 02/0 : 30[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-31:3197180:3197589 [0] NCCL INFO Channel 08/0 : 24[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-31:3197180:3197589 [0] NCCL INFO Channel 10/0 : 24[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-31:3197186:3197593 [6] NCCL INFO Channel 04/0 : 30[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-31:3197180:3197589 [0] NCCL INFO Channel 12/0 : 24[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-31:3197186:3197593 [6] NCCL INFO Channel 08/0 : 30[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-31:3197180:3197589 [0] NCCL INFO Channel 14/0 : 24[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-31:3197186:3197593 [6] NCCL INFO Channel 10/0 : 30[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-31:3197186:3197593 [6] NCCL INFO Channel 12/0 : 30[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-31:3197186:3197593 [6] NCCL INFO Channel 14/0 : 30[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-14:2525078:2525474 [3] NCCL INFO Connected NVLS tree
+ip-26-0-162-14:2525078:2525474 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-162-14:2525078:2525474 [3] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-162-14:2525078:2525474 [3] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-162-14:2525076:2525478 [1] NCCL INFO Connected NVLS tree
+ip-26-0-162-14:2525076:2525478 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-162-14:2525076:2525478 [1] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-162-14:2525076:2525478 [1] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-162-14:2525080:2525476 [5] NCCL INFO Connected NVLS tree
+ip-26-0-162-14:2525080:2525476 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-162-14:2525080:2525476 [5] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-162-14:2525080:2525476 [5] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-162-14:2525082:2525472 [7] NCCL INFO Connected NVLS tree
+ip-26-0-162-14:2525082:2525472 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-162-14:2525082:2525472 [7] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-162-14:2525082:2525472 [7] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-170-31:3197182:3197588 [2] NCCL INFO Connected NVLS tree
+ip-26-0-170-31:3197182:3197588 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-170-31:3197182:3197588 [2] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-170-31:3197182:3197588 [2] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-170-31:3197186:3197593 [6] NCCL INFO Connected NVLS tree
+ip-26-0-170-31:3197186:3197593 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-170-31:3197186:3197593 [6] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-170-31:3197186:3197593 [6] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-170-31:3197180:3197589 [0] NCCL INFO Connected NVLS tree
+ip-26-0-170-31:3197180:3197589 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-170-31:3197180:3197589 [0] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-170-31:3197180:3197589 [0] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-170-31:3197184:3197586 [4] NCCL INFO Connected NVLS tree
+ip-26-0-170-31:3197184:3197586 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-170-31:3197184:3197586 [4] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-170-31:3197184:3197586 [4] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-170-31:3197185:3197591 [5] NCCL INFO Connected NVLS tree
+ip-26-0-170-31:3197185:3197591 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-170-31:3197185:3197591 [5] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-170-31:3197185:3197591 [5] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-170-31:3197183:3197592 [3] NCCL INFO Connected NVLS tree
+ip-26-0-170-31:3197183:3197592 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-170-31:3197183:3197592 [3] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-170-31:3197183:3197592 [3] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-170-31:3197187:3197590 [7] NCCL INFO Connected NVLS tree
+ip-26-0-170-31:3197187:3197590 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-170-31:3197187:3197590 [7] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-170-31:3197187:3197590 [7] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-170-31:3197181:3197587 [1] NCCL INFO Connected NVLS tree
+ip-26-0-170-31:3197181:3197587 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-170-31:3197181:3197587 [1] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-170-31:3197181:3197587 [1] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-170-31:3197181:3197587 [1] NCCL INFO comm 0x9bc77b0 rank 25 nranks 32 cudaDev 1 nvmlDev 1 busId 64000 commId 0xd3b880bde5f1bc69 - Init COMPLETE
+ip-26-0-170-31:3197182:3197588 [2] NCCL INFO comm 0x88fe880 rank 26 nranks 32 cudaDev 2 nvmlDev 2 busId 75000 commId 0xd3b880bde5f1bc69 - Init COMPLETE
+ip-26-0-170-31:3197180:3197589 [0] NCCL INFO comm 0xa19e6a0 rank 24 nranks 32 cudaDev 0 nvmlDev 0 busId 53000 commId 0xd3b880bde5f1bc69 - Init COMPLETE
+ip-26-0-170-31:3197184:3197586 [4] NCCL INFO comm 0xa1031e0 rank 28 nranks 32 cudaDev 4 nvmlDev 4 busId 97000 commId 0xd3b880bde5f1bc69 - Init COMPLETE
+ip-26-0-170-31:3197185:3197591 [5] NCCL INFO comm 0x884f520 rank 29 nranks 32 cudaDev 5 nvmlDev 5 busId a8000 commId 0xd3b880bde5f1bc69 - Init COMPLETE
+ip-26-0-170-132:3330186:3330592 [3] NCCL INFO Connected NVLS tree
+ip-26-0-170-31:3197186:3197593 [6] NCCL INFO comm 0xa034d40 rank 30 nranks 32 cudaDev 6 nvmlDev 6 busId b9000 commId 0xd3b880bde5f1bc69 - Init COMPLETE
+ip-26-0-170-132:3330186:3330592 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-170-132:3330186:3330592 [3] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-170-132:3330186:3330592 [3] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-170-31:3197183:3197592 [3] NCCL INFO comm 0x9f6fc60 rank 27 nranks 32 cudaDev 3 nvmlDev 3 busId 86000 commId 0xd3b880bde5f1bc69 - Init COMPLETE
+ip-26-0-170-31:3197187:3197590 [7] NCCL INFO comm 0xa665180 rank 31 nranks 32 cudaDev 7 nvmlDev 7 busId ca000 commId 0xd3b880bde5f1bc69 - Init COMPLETE
+ip-26-0-170-132:3330184:3330596 [1] NCCL INFO Connected NVLS tree
+ip-26-0-170-132:3330184:3330596 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-170-132:3330184:3330596 [1] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-170-132:3330184:3330596 [1] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-170-132:3330190:3330589 [7] NCCL INFO Connected NVLS tree
+ip-26-0-170-132:3330190:3330589 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-170-132:3330190:3330589 [7] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-170-132:3330190:3330589 [7] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-170-132:3330189:3330594 [6] NCCL INFO Connected NVLS tree
+ip-26-0-170-132:3330189:3330594 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-170-132:3330189:3330594 [6] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-170-132:3330189:3330594 [6] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-162-14:2525081:2525473 [6] NCCL INFO Connected NVLS tree
+ip-26-0-162-14:2525081:2525473 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-162-14:2525081:2525473 [6] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-162-14:2525081:2525473 [6] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-163-58:934721:935115 [1] NCCL INFO Connected NVLS tree
+ip-26-0-163-58:934721:935115 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-163-58:934721:935115 [1] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-163-58:934721:935115 [1] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-162-14:2525077:2525475 [2] NCCL INFO Connected NVLS tree
+ip-26-0-162-14:2525077:2525475 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-162-14:2525077:2525475 [2] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-162-14:2525077:2525475 [2] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-163-58:934723:935117 [3] NCCL INFO Connected NVLS tree
+ip-26-0-163-58:934723:935117 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-163-58:934723:935117 [3] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-163-58:934723:935117 [3] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-170-132:3330185:3330593 [2] NCCL INFO Connected NVLS tree
+ip-26-0-170-132:3330185:3330593 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-170-132:3330185:3330593 [2] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-170-132:3330185:3330593 [2] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-170-132:3330188:3330595 [5] NCCL INFO Connected NVLS tree
+ip-26-0-170-132:3330188:3330595 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-170-132:3330188:3330595 [5] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-170-132:3330188:3330595 [5] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-163-58:934725:935118 [5] NCCL INFO Connected NVLS tree
+ip-26-0-163-58:934725:935118 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-163-58:934725:935118 [5] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-163-58:934725:935118 [5] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-162-14:2525075:2525471 [0] NCCL INFO Connected NVLS tree
+ip-26-0-162-14:2525075:2525471 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-162-14:2525075:2525471 [0] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-162-14:2525075:2525471 [0] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-163-58:934727:935114 [7] NCCL INFO Connected NVLS tree
+ip-26-0-163-58:934727:935114 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-163-58:934727:935114 [7] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-163-58:934727:935114 [7] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-163-58:934722:935116 [2] NCCL INFO Connected NVLS tree
+ip-26-0-163-58:934722:935116 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-163-58:934722:935116 [2] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-163-58:934722:935116 [2] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-163-58:934726:935119 [6] NCCL INFO Connected NVLS tree
+ip-26-0-163-58:934726:935119 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-163-58:934726:935119 [6] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-163-58:934726:935119 [6] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-162-14:2525079:2525477 [4] NCCL INFO Connected NVLS tree
+ip-26-0-162-14:2525079:2525477 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-162-14:2525079:2525477 [4] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-162-14:2525079:2525477 [4] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-170-132:3330187:3330591 [4] NCCL INFO Connected NVLS tree
+ip-26-0-170-132:3330187:3330591 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-170-132:3330187:3330591 [4] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-170-132:3330187:3330591 [4] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-163-58:934724:935113 [4] NCCL INFO Connected NVLS tree
+ip-26-0-163-58:934724:935113 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-163-58:934724:935113 [4] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-163-58:934724:935113 [4] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-170-132:3330183:3330590 [0] NCCL INFO Connected NVLS tree
+ip-26-0-170-132:3330183:3330590 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-170-132:3330183:3330590 [0] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-170-132:3330183:3330590 [0] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-162-14:2525080:2525476 [5] NCCL INFO comm 0x9c491a0 rank 5 nranks 32 cudaDev 5 nvmlDev 5 busId a8000 commId 0xd3b880bde5f1bc69 - Init COMPLETE
+ip-26-0-162-14:2525079:2525477 [4] NCCL INFO comm 0x96c46e0 rank 4 nranks 32 cudaDev 4 nvmlDev 4 busId 97000 commId 0xd3b880bde5f1bc69 - Init COMPLETE
+ip-26-0-162-14:2525075:2525471 [0] NCCL INFO comm 0x98759b0 rank 0 nranks 32 cudaDev 0 nvmlDev 0 busId 53000 commId 0xd3b880bde5f1bc69 - Init COMPLETE
+ip-26-0-163-58:934720:935120 [0] NCCL INFO Connected NVLS tree
+ip-26-0-162-14:2525081:2525473 [6] NCCL INFO comm 0x94f5e60 rank 6 nranks 32 cudaDev 6 nvmlDev 6 busId b9000 commId 0xd3b880bde5f1bc69 - Init COMPLETE
+ip-26-0-162-14:2525077:2525475 [2] NCCL INFO comm 0xa1b95c0 rank 2 nranks 32 cudaDev 2 nvmlDev 2 busId 75000 commId 0xd3b880bde5f1bc69 - Init COMPLETE
+ip-26-0-162-14:2525078:2525474 [3] NCCL INFO comm 0x9224d20 rank 3 nranks 32 cudaDev 3 nvmlDev 3 busId 86000 commId 0xd3b880bde5f1bc69 - Init COMPLETE
+ip-26-0-162-14:2525076:2525478 [1] NCCL INFO comm 0xa370ac0 rank 1 nranks 32 cudaDev 1 nvmlDev 1 busId 64000 commId 0xd3b880bde5f1bc69 - Init COMPLETE
+ip-26-0-162-14:2525082:2525472 [7] NCCL INFO comm 0x9e76320 rank 7 nranks 32 cudaDev 7 nvmlDev 7 busId ca000 commId 0xd3b880bde5f1bc69 - Init COMPLETE
+ip-26-0-163-58:934720:935120 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-163-58:934720:935120 [0] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-163-58:934720:935120 [0] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-170-132:3330187:3330591 [4] NCCL INFO comm 0x9a3ab00 rank 20 nranks 32 cudaDev 4 nvmlDev 4 busId 97000 commId 0xd3b880bde5f1bc69 - Init COMPLETE
+ip-26-0-170-132:3330183:3330590 [0] NCCL INFO comm 0x94d59c0 rank 16 nranks 32 cudaDev 0 nvmlDev 0 busId 53000 commId 0xd3b880bde5f1bc69 - Init COMPLETE
+ip-26-0-170-132:3330189:3330594 [6] NCCL INFO comm 0x87afc80 rank 22 nranks 32 cudaDev 6 nvmlDev 6 busId b9000 commId 0xd3b880bde5f1bc69 - Init COMPLETE
+ip-26-0-170-132:3330185:3330593 [2] NCCL INFO comm 0x9a9fba0 rank 18 nranks 32 cudaDev 2 nvmlDev 2 busId 75000 commId 0xd3b880bde5f1bc69 - Init COMPLETE
+ip-26-0-170-132:3330186:3330592 [3] NCCL INFO comm 0x8f16e30 rank 19 nranks 32 cudaDev 3 nvmlDev 3 busId 86000 commId 0xd3b880bde5f1bc69 - Init COMPLETE
+ip-26-0-170-132:3330190:3330589 [7] NCCL INFO comm 0xa51d2b0 rank 23 nranks 32 cudaDev 7 nvmlDev 7 busId ca000 commId 0xd3b880bde5f1bc69 - Init COMPLETE
+ip-26-0-170-132:3330188:3330595 [5] NCCL INFO comm 0xa4ee330 rank 21 nranks 32 cudaDev 5 nvmlDev 5 busId a8000 commId 0xd3b880bde5f1bc69 - Init COMPLETE
+ip-26-0-170-132:3330184:3330596 [1] NCCL INFO comm 0x960d840 rank 17 nranks 32 cudaDev 1 nvmlDev 1 busId 64000 commId 0xd3b880bde5f1bc69 - Init COMPLETE
+ip-26-0-163-58:934727:935114 [7] NCCL INFO comm 0x9a1dc80 rank 15 nranks 32 cudaDev 7 nvmlDev 7 busId ca000 commId 0xd3b880bde5f1bc69 - Init COMPLETE
+ip-26-0-163-58:934725:935118 [5] NCCL INFO comm 0x8aa9ac0 rank 13 nranks 32 cudaDev 5 nvmlDev 5 busId a8000 commId 0xd3b880bde5f1bc69 - Init COMPLETE
+ip-26-0-163-58:934721:935115 [1] NCCL INFO comm 0xa549100 rank 9 nranks 32 cudaDev 1 nvmlDev 1 busId 64000 commId 0xd3b880bde5f1bc69 - Init COMPLETE
+ip-26-0-163-58:934723:935117 [3] NCCL INFO comm 0x9ac5f80 rank 11 nranks 32 cudaDev 3 nvmlDev 3 busId 86000 commId 0xd3b880bde5f1bc69 - Init COMPLETE
+ip-26-0-163-58:934722:935116 [2] NCCL INFO comm 0x8ac6a60 rank 10 nranks 32 cudaDev 2 nvmlDev 2 busId 75000 commId 0xd3b880bde5f1bc69 - Init COMPLETE
+ip-26-0-163-58:934726:935119 [6] NCCL INFO comm 0x9e0f0e0 rank 14 nranks 32 cudaDev 6 nvmlDev 6 busId b9000 commId 0xd3b880bde5f1bc69 - Init COMPLETE
+ip-26-0-163-58:934724:935113 [4] NCCL INFO comm 0x98bc760 rank 12 nranks 32 cudaDev 4 nvmlDev 4 busId 97000 commId 0xd3b880bde5f1bc69 - Init COMPLETE
+ip-26-0-163-58:934720:935120 [0] NCCL INFO comm 0x98054a0 rank 8 nranks 32 cudaDev 0 nvmlDev 0 busId 53000 commId 0xd3b880bde5f1bc69 - Init COMPLETE
+ip-26-0-162-14:2525075:2525509 [0] NCCL INFO Using network Libfabric
+ip-26-0-162-14:2525078:2525510 [3] NCCL INFO Using network Libfabric
+ip-26-0-162-14:2525080:2525511 [5] NCCL INFO Using network Libfabric
+ip-26-0-162-14:2525082:2525515 [7] NCCL INFO Using network Libfabric
+ip-26-0-162-14:2525081:2525514 [6] NCCL INFO Using network Libfabric
+ip-26-0-162-14:2525076:2525513 [1] NCCL INFO Using network Libfabric
+ip-26-0-162-14:2525077:2525512 [2] NCCL INFO Using network Libfabric
+ip-26-0-162-14:2525079:2525516 [4] NCCL INFO Using network Libfabric
+ip-26-0-162-14:2525082:2525515 [7] NCCL INFO comm 0x9e8b8b0 rank 7 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0xcc04d7b16f1339ed - Init START
+ip-26-0-162-14:2525075:2525509 [0] NCCL INFO comm 0x988a6e0 rank 0 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0xcc04d7b16f1339ed - Init START
+ip-26-0-162-14:2525078:2525510 [3] NCCL INFO comm 0x923c7b0 rank 3 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0xcc04d7b16f1339ed - Init START
+ip-26-0-162-14:2525081:2525514 [6] NCCL INFO comm 0x950b8e0 rank 6 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0xcc04d7b16f1339ed - Init START
+ip-26-0-162-14:2525079:2525516 [4] NCCL INFO comm 0x96da6c0 rank 4 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0xcc04d7b16f1339ed - Init START
+ip-26-0-162-14:2525077:2525512 [2] NCCL INFO comm 0xa1cf600 rank 2 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0xcc04d7b16f1339ed - Init START
+ip-26-0-162-14:2525076:2525513 [1] NCCL INFO comm 0xa386440 rank 1 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0xcc04d7b16f1339ed - Init START
+ip-26-0-162-14:2525080:2525511 [5] NCCL INFO comm 0x9c5e6a0 rank 5 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0xcc04d7b16f1339ed - Init START
+ip-26-0-162-14:2525082:2525515 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-162-14:2525079:2525516 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-162-14:2525078:2525510 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-162-14:2525081:2525514 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-162-14:2525077:2525512 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-162-14:2525075:2525509 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/253
+ip-26-0-162-14:2525076:2525513 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-162-14:2525080:2525511 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+NCCL version 2.18.5+cuda12.2
+ip-26-0-170-132:3330190:3330627 [7] NCCL INFO Using network Libfabric
+ip-26-0-170-132:3330183:3330626 [0] NCCL INFO Using network Libfabric
+ip-26-0-170-132:3330187:3330630 [4] NCCL INFO Using network Libfabric
+ip-26-0-170-132:3330189:3330633 [6] NCCL INFO Using network Libfabric
+ip-26-0-170-132:3330184:3330631 [1] NCCL INFO Using network Libfabric
+ip-26-0-170-132:3330188:3330632 [5] NCCL INFO Using network Libfabric
+ip-26-0-170-132:3330185:3330628 [2] NCCL INFO Using network Libfabric
+ip-26-0-170-132:3330186:3330629 [3] NCCL INFO Using network Libfabric
+ip-26-0-170-132:3330187:3330630 [4] NCCL INFO comm 0x9a52690 rank 4 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0xa8d0c6af6511c913 - Init START
+ip-26-0-170-132:3330186:3330629 [3] NCCL INFO comm 0x8f2f440 rank 3 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0xa8d0c6af6511c913 - Init START
+ip-26-0-170-132:3330185:3330628 [2] NCCL INFO comm 0x9ab5de0 rank 2 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0xa8d0c6af6511c913 - Init START
+ip-26-0-170-132:3330184:3330631 [1] NCCL INFO comm 0x9625de0 rank 1 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0xa8d0c6af6511c913 - Init START
+ip-26-0-170-132:3330183:3330626 [0] NCCL INFO comm 0x94ede40 rank 0 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0xa8d0c6af6511c913 - Init START
+ip-26-0-170-132:3330188:3330632 [5] NCCL INFO comm 0xa5074f0 rank 5 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0xa8d0c6af6511c913 - Init START
+ip-26-0-170-132:3330189:3330633 [6] NCCL INFO comm 0x87c6270 rank 6 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0xa8d0c6af6511c913 - Init START
+ip-26-0-170-132:3330190:3330627 [7] NCCL INFO comm 0xa535420 rank 7 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0xa8d0c6af6511c913 - Init START
+ip-26-0-170-132:3330183:3330626 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-170-132:3330189:3330633 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-170-132:3330186:3330629 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-170-132:3330185:3330628 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-170-132:3330188:3330632 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-170-132:3330184:3330631 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-170-132:3330187:3330630 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-170-132:3330190:3330627 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+NCCL version 2.18.5+cuda12.2
+ip-26-0-170-31:3197180:3197624 [0] NCCL INFO Using network Libfabric
+ip-26-0-170-31:3197187:3197626 [7] NCCL INFO Using network Libfabric
+ip-26-0-170-31:3197181:3197628 [1] NCCL INFO Using network Libfabric
+ip-26-0-170-31:3197184:3197630 [4] NCCL INFO Using network Libfabric
+ip-26-0-170-31:3197185:3197627 [5] NCCL INFO Using network Libfabric
+ip-26-0-170-31:3197182:3197631 [2] NCCL INFO Using network Libfabric
+ip-26-0-170-31:3197186:3197629 [6] NCCL INFO Using network Libfabric
+ip-26-0-170-31:3197183:3197625 [3] NCCL INFO Using network Libfabric
+ip-26-0-170-31:3197182:3197631 [2] NCCL INFO comm 0x89157c0 rank 2 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0x41c9367c841b7051 - Init START
+ip-26-0-170-31:3197183:3197625 [3] NCCL INFO comm 0x9f84250 rank 3 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0x41c9367c841b7051 - Init START
+ip-26-0-170-31:3197184:3197630 [4] NCCL INFO comm 0xa117e20 rank 4 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0x41c9367c841b7051 - Init START
+ip-26-0-170-31:3197185:3197627 [5] NCCL INFO comm 0x8864ad0 rank 5 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0x41c9367c841b7051 - Init START
+ip-26-0-170-31:3197181:3197628 [1] NCCL INFO comm 0x9be0100 rank 1 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0x41c9367c841b7051 - Init START
+ip-26-0-170-31:3197187:3197626 [7] NCCL INFO comm 0xa6793c0 rank 7 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0x41c9367c841b7051 - Init START
+ip-26-0-170-31:3197186:3197629 [6] NCCL INFO comm 0xa04bb00 rank 6 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0x41c9367c841b7051 - Init START
+ip-26-0-170-31:3197180:3197624 [0] NCCL INFO comm 0xa1b4140 rank 0 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0x41c9367c841b7051 - Init START
+ip-26-0-170-31:3197185:3197627 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-170-31:3197183:3197625 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-170-31:3197184:3197630 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-170-31:3197181:3197628 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-170-31:3197180:3197624 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-170-31:3197187:3197626 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-170-31:3197182:3197631 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-170-31:3197186:3197629 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+NCCL version 2.18.5+cuda12.2
+ip-26-0-163-58:934720:935150 [0] NCCL INFO Using network Libfabric
+ip-26-0-163-58:934722:935155 [2] NCCL INFO Using network Libfabric
+ip-26-0-163-58:934726:935151 [6] NCCL INFO Using network Libfabric
+ip-26-0-163-58:934724:935154 [4] NCCL INFO Using network Libfabric
+ip-26-0-163-58:934727:935152 [7] NCCL INFO Using network Libfabric
+ip-26-0-163-58:934721:935153 [1] NCCL INFO Using network Libfabric
+ip-26-0-163-58:934723:935156 [3] NCCL INFO Using network Libfabric
+ip-26-0-163-58:934725:935157 [5] NCCL INFO Using network Libfabric
+ip-26-0-163-58:934723:935156 [3] NCCL INFO comm 0x9adc790 rank 3 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0x37fac877b5d17063 - Init START
+ip-26-0-163-58:934726:935151 [6] NCCL INFO comm 0x9e24e90 rank 6 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0x37fac877b5d17063 - Init START
+ip-26-0-163-58:934724:935154 [4] NCCL INFO comm 0x98d2f70 rank 4 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0x37fac877b5d17063 - Init START
+ip-26-0-163-58:934720:935150 [0] NCCL INFO comm 0x9818c80 rank 0 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0x37fac877b5d17063 - Init START
+ip-26-0-163-58:934725:935157 [5] NCCL INFO comm 0x8abff80 rank 5 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0x37fac877b5d17063 - Init START
+ip-26-0-163-58:934722:935155 [2] NCCL INFO comm 0x8adcf20 rank 2 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0x37fac877b5d17063 - Init START
+ip-26-0-163-58:934721:935153 [1] NCCL INFO comm 0xa55f070 rank 1 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0x37fac877b5d17063 - Init START
+ip-26-0-163-58:934727:935152 [7] NCCL INFO comm 0x9a33d80 rank 7 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0x37fac877b5d17063 - Init START
+ip-26-0-163-58:934720:935150 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-163-58:934721:935153 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-163-58:934727:935152 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-163-58:934726:935151 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-163-58:934725:935157 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-163-58:934723:935156 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-163-58:934724:935154 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-163-58:934722:935155 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-163-58:934726:935151 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-163-58:934726:935151 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-163-58:934722:935155 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-163-58:934722:935155 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-162-14:2525079:2525516 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-162-14:2525079:2525516 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-163-58:934723:935156 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-163-58:934723:935156 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-170-31:3197182:3197631 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-170-31:3197182:3197631 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-170-132:3330188:3330632 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-170-132:3330188:3330632 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-170-31:3197187:3197626 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-170-31:3197187:3197626 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-163-58:934727:935152 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-163-58:934727:935152 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-163-58:934725:935157 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-163-58:934725:935157 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-163-58:934724:935154 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-163-58:934724:935154 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-163-58:934721:935153 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-163-58:934721:935153 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-163-58:934720:935150 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-163-58:934720:935150 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-163-58:934720:935150 [0] NCCL INFO Channel 00/24 :    0   1   2   3   4   5   6   7
+ip-26-0-163-58:934720:935150 [0] NCCL INFO Channel 01/24 :    0   1   2   3   4   5   6   7
+ip-26-0-163-58:934720:935150 [0] NCCL INFO Channel 02/24 :    0   1   2   3   4   5   6   7
+ip-26-0-163-58:934720:935150 [0] NCCL INFO Channel 03/24 :    0   1   2   3   4   5   6   7
+ip-26-0-163-58:934720:935150 [0] NCCL INFO Channel 04/24 :    0   1   2   3   4   5   6   7
+ip-26-0-163-58:934720:935150 [0] NCCL INFO Channel 05/24 :    0   1   2   3   4   5   6   7
+ip-26-0-163-58:934720:935150 [0] NCCL INFO Channel 06/24 :    0   1   2   3   4   5   6   7
+ip-26-0-163-58:934720:935150 [0] NCCL INFO Channel 07/24 :    0   1   2   3   4   5   6   7
+ip-26-0-163-58:934720:935150 [0] NCCL INFO Channel 08/24 :    0   1   2   3   4   5   6   7
+ip-26-0-163-58:934720:935150 [0] NCCL INFO Channel 09/24 :    0   1   2   3   4   5   6   7
+ip-26-0-163-58:934720:935150 [0] NCCL INFO Channel 10/24 :    0   1   2   3   4   5   6   7
+ip-26-0-163-58:934720:935150 [0] NCCL INFO Channel 11/24 :    0   1   2   3   4   5   6   7
+ip-26-0-163-58:934720:935150 [0] NCCL INFO Channel 12/24 :    0   1   2   3   4   5   6   7
+ip-26-0-163-58:934720:935150 [0] NCCL INFO Channel 13/24 :    0   1   2   3   4   5   6   7
+ip-26-0-163-58:934720:935150 [0] NCCL INFO Channel 14/24 :    0   1   2   3   4   5   6   7
+ip-26-0-163-58:934720:935150 [0] NCCL INFO Channel 15/24 :    0   1   2   3   4   5   6   7
+ip-26-0-163-58:934720:935150 [0] NCCL INFO Channel 16/24 :    0   1   2   3   4   5   6   7
+ip-26-0-163-58:934720:935150 [0] NCCL INFO Channel 17/24 :    0   1   2   3   4   5   6   7
+ip-26-0-163-58:934720:935150 [0] NCCL INFO Channel 18/24 :    0   1   2   3   4   5   6   7
+ip-26-0-163-58:934720:935150 [0] NCCL INFO Channel 19/24 :    0   1   2   3   4   5   6   7
+ip-26-0-163-58:934720:935150 [0] NCCL INFO Channel 20/24 :    0   1   2   3   4   5   6   7
+ip-26-0-163-58:934720:935150 [0] NCCL INFO Channel 21/24 :    0   1   2   3   4   5   6   7
+ip-26-0-163-58:934720:935150 [0] NCCL INFO Channel 22/24 :    0   1   2   3   4   5   6   7
+ip-26-0-163-58:934720:935150 [0] NCCL INFO Channel 23/24 :    0   1   2   3   4   5   6   7
+ip-26-0-163-58:934720:935150 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] 1/-1/-1->0->-1 [7] 1/-1/-1->0->-1 [8] 1/-1/-1->0->-1 [9] 1/-1/-1->0->-1 [10] 1/-1/-1->0->-1 [11] 1/-1/-1->0->-1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] 1/-1/-1->0->-1 [19] 1/-1/-1->0->-1 [20] 1/-1/-1->0->-1 [21] 1/-1/-1->0->-1 [22] 1/-1/-1->0->-1 [23] 1/-1/-1->0->-1
+ip-26-0-163-58:934720:935150 [0] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-163-58:934721:935153 [1] NCCL INFO Trees [0] 2/-1/-1->1->0 [1] 2/-1/-1->1->0 [2] 2/-1/-1->1->0 [3] 2/-1/-1->1->0 [4] 2/-1/-1->1->0 [5] 2/-1/-1->1->0 [6] 2/-1/-1->1->0 [7] 2/-1/-1->1->0 [8] 2/-1/-1->1->0 [9] 2/-1/-1->1->0 [10] 2/-1/-1->1->0 [11] 2/-1/-1->1->0 [12] 2/-1/-1->1->0 [13] 2/-1/-1->1->0 [14] 2/-1/-1->1->0 [15] 2/-1/-1->1->0 [16] 2/-1/-1->1->0 [17] 2/-1/-1->1->0 [18] 2/-1/-1->1->0 [19] 2/-1/-1->1->0 [20] 2/-1/-1->1->0 [21] 2/-1/-1->1->0 [22] 2/-1/-1->1->0 [23] 2/-1/-1->1->0
+ip-26-0-163-58:934727:935152 [7] NCCL INFO Trees [0] -1/-1/-1->7->6 [1] -1/-1/-1->7->6 [2] -1/-1/-1->7->6 [3] -1/-1/-1->7->6 [4] -1/-1/-1->7->6 [5] -1/-1/-1->7->6 [6] -1/-1/-1->7->6 [7] -1/-1/-1->7->6 [8] -1/-1/-1->7->6 [9] -1/-1/-1->7->6 [10] -1/-1/-1->7->6 [11] -1/-1/-1->7->6 [12] -1/-1/-1->7->6 [13] -1/-1/-1->7->6 [14] -1/-1/-1->7->6 [15] -1/-1/-1->7->6 [16] -1/-1/-1->7->6 [17] -1/-1/-1->7->6 [18] -1/-1/-1->7->6 [19] -1/-1/-1->7->6 [20] -1/-1/-1->7->6 [21] -1/-1/-1->7->6 [22] -1/-1/-1->7->6 [23] -1/-1/-1->7->6
+ip-26-0-163-58:934721:935153 [1] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-163-58:934726:935151 [6] NCCL INFO Trees [0] 7/-1/-1->6->5 [1] 7/-1/-1->6->5 [2] 7/-1/-1->6->5 [3] 7/-1/-1->6->5 [4] 7/-1/-1->6->5 [5] 7/-1/-1->6->5 [6] 7/-1/-1->6->5 [7] 7/-1/-1->6->5 [8] 7/-1/-1->6->5 [9] 7/-1/-1->6->5 [10] 7/-1/-1->6->5 [11] 7/-1/-1->6->5 [12] 7/-1/-1->6->5 [13] 7/-1/-1->6->5 [14] 7/-1/-1->6->5 [15] 7/-1/-1->6->5 [16] 7/-1/-1->6->5 [17] 7/-1/-1->6->5 [18] 7/-1/-1->6->5 [19] 7/-1/-1->6->5 [20] 7/-1/-1->6->5 [21] 7/-1/-1->6->5 [22] 7/-1/-1->6->5 [23] 7/-1/-1->6->5
+ip-26-0-163-58:934727:935152 [7] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-163-58:934725:935157 [5] NCCL INFO Trees [0] 6/-1/-1->5->4 [1] 6/-1/-1->5->4 [2] 6/-1/-1->5->4 [3] 6/-1/-1->5->4 [4] 6/-1/-1->5->4 [5] 6/-1/-1->5->4 [6] 6/-1/-1->5->4 [7] 6/-1/-1->5->4 [8] 6/-1/-1->5->4 [9] 6/-1/-1->5->4 [10] 6/-1/-1->5->4 [11] 6/-1/-1->5->4 [12] 6/-1/-1->5->4 [13] 6/-1/-1->5->4 [14] 6/-1/-1->5->4 [15] 6/-1/-1->5->4 [16] 6/-1/-1->5->4 [17] 6/-1/-1->5->4 [18] 6/-1/-1->5->4 [19] 6/-1/-1->5->4 [20] 6/-1/-1->5->4 [21] 6/-1/-1->5->4 [22] 6/-1/-1->5->4 [23] 6/-1/-1->5->4
+ip-26-0-163-58:934726:935151 [6] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-163-58:934722:935155 [2] NCCL INFO Trees [0] 3/-1/-1->2->1 [1] 3/-1/-1->2->1 [2] 3/-1/-1->2->1 [3] 3/-1/-1->2->1 [4] 3/-1/-1->2->1 [5] 3/-1/-1->2->1 [6] 3/-1/-1->2->1 [7] 3/-1/-1->2->1 [8] 3/-1/-1->2->1 [9] 3/-1/-1->2->1 [10] 3/-1/-1->2->1 [11] 3/-1/-1->2->1 [12] 3/-1/-1->2->1 [13] 3/-1/-1->2->1 [14] 3/-1/-1->2->1 [15] 3/-1/-1->2->1 [16] 3/-1/-1->2->1 [17] 3/-1/-1->2->1 [18] 3/-1/-1->2->1 [19] 3/-1/-1->2->1 [20] 3/-1/-1->2->1 [21] 3/-1/-1->2->1 [22] 3/-1/-1->2->1 [23] 3/-1/-1->2->1
+ip-26-0-163-58:934725:935157 [5] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-163-58:934722:935155 [2] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-163-58:934723:935156 [3] NCCL INFO Trees [0] 4/-1/-1->3->2 [1] 4/-1/-1->3->2 [2] 4/-1/-1->3->2 [3] 4/-1/-1->3->2 [4] 4/-1/-1->3->2 [5] 4/-1/-1->3->2 [6] 4/-1/-1->3->2 [7] 4/-1/-1->3->2 [8] 4/-1/-1->3->2 [9] 4/-1/-1->3->2 [10] 4/-1/-1->3->2 [11] 4/-1/-1->3->2 [12] 4/-1/-1->3->2 [13] 4/-1/-1->3->2 [14] 4/-1/-1->3->2 [15] 4/-1/-1->3->2 [16] 4/-1/-1->3->2 [17] 4/-1/-1->3->2 [18] 4/-1/-1->3->2 [19] 4/-1/-1->3->2 [20] 4/-1/-1->3->2 [21] 4/-1/-1->3->2 [22] 4/-1/-1->3->2 [23] 4/-1/-1->3->2
+ip-26-0-163-58:934723:935156 [3] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-163-58:934724:935154 [4] NCCL INFO Trees [0] 5/-1/-1->4->3 [1] 5/-1/-1->4->3 [2] 5/-1/-1->4->3 [3] 5/-1/-1->4->3 [4] 5/-1/-1->4->3 [5] 5/-1/-1->4->3 [6] 5/-1/-1->4->3 [7] 5/-1/-1->4->3 [8] 5/-1/-1->4->3 [9] 5/-1/-1->4->3 [10] 5/-1/-1->4->3 [11] 5/-1/-1->4->3 [12] 5/-1/-1->4->3 [13] 5/-1/-1->4->3 [14] 5/-1/-1->4->3 [15] 5/-1/-1->4->3 [16] 5/-1/-1->4->3 [17] 5/-1/-1->4->3 [18] 5/-1/-1->4->3 [19] 5/-1/-1->4->3 [20] 5/-1/-1->4->3 [21] 5/-1/-1->4->3 [22] 5/-1/-1->4->3 [23] 5/-1/-1->4->3
+ip-26-0-163-58:934724:935154 [4] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-170-31:3197181:3197628 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-170-31:3197181:3197628 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-163-58:934720:935150 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-163-58:934720:935150 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-163-58:934720:935150 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-163-58:934723:935156 [3] NCCL INFO Channel 00/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-163-58:934720:935150 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-163-58:934726:935151 [6] NCCL INFO Channel 00/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-163-58:934723:935156 [3] NCCL INFO Channel 01/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-163-58:934720:935150 [0] NCCL INFO Channel 04/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-163-58:934726:935151 [6] NCCL INFO Channel 01/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-163-58:934720:935150 [0] NCCL INFO Channel 05/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-163-58:934723:935156 [3] NCCL INFO Channel 02/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-163-58:934726:935151 [6] NCCL INFO Channel 02/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-163-58:934723:935156 [3] NCCL INFO Channel 03/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-163-58:934720:935150 [0] NCCL INFO Channel 06/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-163-58:934727:935152 [7] NCCL INFO Channel 00/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-163-58:934722:935155 [2] NCCL INFO Channel 00/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-163-58:934724:935154 [4] NCCL INFO Channel 00/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-163-58:934726:935151 [6] NCCL INFO Channel 03/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-163-58:934727:935152 [7] NCCL INFO Channel 01/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-163-58:934720:935150 [0] NCCL INFO Channel 07/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-163-58:934723:935156 [3] NCCL INFO Channel 04/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-163-58:934721:935153 [1] NCCL INFO Channel 00/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-163-58:934722:935155 [2] NCCL INFO Channel 01/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-163-58:934727:935152 [7] NCCL INFO Channel 02/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-163-58:934720:935150 [0] NCCL INFO Channel 08/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-163-58:934723:935156 [3] NCCL INFO Channel 05/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-163-58:934721:935153 [1] NCCL INFO Channel 01/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-163-58:934722:935155 [2] NCCL INFO Channel 02/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-163-58:934724:935154 [4] NCCL INFO Channel 01/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-163-58:934727:935152 [7] NCCL INFO Channel 03/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-163-58:934720:935150 [0] NCCL INFO Channel 09/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-163-58:934723:935156 [3] NCCL INFO Channel 06/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-163-58:934721:935153 [1] NCCL INFO Channel 02/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-163-58:934722:935155 [2] NCCL INFO Channel 03/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-163-58:934720:935150 [0] NCCL INFO Channel 10/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-163-58:934726:935151 [6] NCCL INFO Channel 04/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-170-31:3197180:3197624 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-170-31:3197180:3197624 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-163-58:934722:935155 [2] NCCL INFO Channel 04/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-163-58:934723:935156 [3] NCCL INFO Channel 07/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-163-58:934720:935150 [0] NCCL INFO Channel 11/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-163-58:934726:935151 [6] NCCL INFO Channel 05/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-163-58:934723:935156 [3] NCCL INFO Channel 08/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-163-58:934722:935155 [2] NCCL INFO Channel 05/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-163-58:934720:935150 [0] NCCL INFO Channel 12/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-163-58:934726:935151 [6] NCCL INFO Channel 06/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-163-58:934723:935156 [3] NCCL INFO Channel 09/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-163-58:934722:935155 [2] NCCL INFO Channel 06/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-163-58:934720:935150 [0] NCCL INFO Channel 13/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-163-58:934726:935151 [6] NCCL INFO Channel 07/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-163-58:934723:935156 [3] NCCL INFO Channel 10/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-163-58:934722:935155 [2] NCCL INFO Channel 07/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-163-58:934720:935150 [0] NCCL INFO Channel 14/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-163-58:934726:935151 [6] NCCL INFO Channel 08/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-163-58:934723:935156 [3] NCCL INFO Channel 11/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-163-58:934722:935155 [2] NCCL INFO Channel 08/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-163-58:934720:935150 [0] NCCL INFO Channel 15/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-163-58:934726:935151 [6] NCCL INFO Channel 09/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-163-58:934723:935156 [3] NCCL INFO Channel 12/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-163-58:934722:935155 [2] NCCL INFO Channel 09/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-163-58:934720:935150 [0] NCCL INFO Channel 16/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-163-58:934726:935151 [6] NCCL INFO Channel 10/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-163-58:934723:935156 [3] NCCL INFO Channel 13/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-163-58:934722:935155 [2] NCCL INFO Channel 10/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-163-58:934720:935150 [0] NCCL INFO Channel 17/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-163-58:934726:935151 [6] NCCL INFO Channel 11/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-163-58:934724:935154 [4] NCCL INFO Channel 02/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-163-58:934727:935152 [7] NCCL INFO Channel 04/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-163-58:934723:935156 [3] NCCL INFO Channel 14/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-163-58:934722:935155 [2] NCCL INFO Channel 11/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-163-58:934721:935153 [1] NCCL INFO Channel 03/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-163-58:934720:935150 [0] NCCL INFO Channel 18/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-163-58:934726:935151 [6] NCCL INFO Channel 12/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-163-58:934724:935154 [4] NCCL INFO Channel 03/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-163-58:934727:935152 [7] NCCL INFO Channel 05/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-163-58:934723:935156 [3] NCCL INFO Channel 15/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-163-58:934725:935157 [5] NCCL INFO Channel 00/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-163-58:934722:935155 [2] NCCL INFO Channel 12/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-163-58:934721:935153 [1] NCCL INFO Channel 04/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-163-58:934720:935150 [0] NCCL INFO Channel 19/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-163-58:934726:935151 [6] NCCL INFO Channel 13/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-163-58:934724:935154 [4] NCCL INFO Channel 04/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-163-58:934727:935152 [7] NCCL INFO Channel 06/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-163-58:934723:935156 [3] NCCL INFO Channel 16/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-163-58:934725:935157 [5] NCCL INFO Channel 01/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-163-58:934722:935155 [2] NCCL INFO Channel 13/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-163-58:934721:935153 [1] NCCL INFO Channel 05/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-163-58:934720:935150 [0] NCCL INFO Channel 20/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-163-58:934726:935151 [6] NCCL INFO Channel 14/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-163-58:934724:935154 [4] NCCL INFO Channel 05/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-163-58:934727:935152 [7] NCCL INFO Channel 07/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-163-58:934723:935156 [3] NCCL INFO Channel 17/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-163-58:934725:935157 [5] NCCL INFO Channel 02/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-163-58:934722:935155 [2] NCCL INFO Channel 14/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-163-58:934721:935153 [1] NCCL INFO Channel 06/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-163-58:934720:935150 [0] NCCL INFO Channel 21/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-163-58:934724:935154 [4] NCCL INFO Channel 06/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-163-58:934726:935151 [6] NCCL INFO Channel 15/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-163-58:934727:935152 [7] NCCL INFO Channel 08/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-163-58:934723:935156 [3] NCCL INFO Channel 18/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-163-58:934725:935157 [5] NCCL INFO Channel 03/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-163-58:934722:935155 [2] NCCL INFO Channel 15/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-163-58:934721:935153 [1] NCCL INFO Channel 07/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-163-58:934720:935150 [0] NCCL INFO Channel 22/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-163-58:934724:935154 [4] NCCL INFO Channel 07/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-163-58:934726:935151 [6] NCCL INFO Channel 16/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-163-58:934727:935152 [7] NCCL INFO Channel 09/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-163-58:934723:935156 [3] NCCL INFO Channel 19/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-163-58:934725:935157 [5] NCCL INFO Channel 04/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-163-58:934722:935155 [2] NCCL INFO Channel 16/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-163-58:934721:935153 [1] NCCL INFO Channel 08/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-163-58:934720:935150 [0] NCCL INFO Channel 23/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-163-58:934724:935154 [4] NCCL INFO Channel 08/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-163-58:934727:935152 [7] NCCL INFO Channel 10/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-163-58:934726:935151 [6] NCCL INFO Channel 17/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-163-58:934723:935156 [3] NCCL INFO Channel 20/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-163-58:934725:935157 [5] NCCL INFO Channel 05/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-163-58:934722:935155 [2] NCCL INFO Channel 17/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-163-58:934721:935153 [1] NCCL INFO Channel 09/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-163-58:934724:935154 [4] NCCL INFO Channel 09/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-163-58:934727:935152 [7] NCCL INFO Channel 11/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-163-58:934726:935151 [6] NCCL INFO Channel 18/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-163-58:934723:935156 [3] NCCL INFO Channel 21/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-163-58:934725:935157 [5] NCCL INFO Channel 06/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-163-58:934722:935155 [2] NCCL INFO Channel 18/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-163-58:934721:935153 [1] NCCL INFO Channel 10/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-163-58:934724:935154 [4] NCCL INFO Channel 10/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-163-58:934727:935152 [7] NCCL INFO Channel 12/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-163-58:934726:935151 [6] NCCL INFO Channel 19/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-163-58:934723:935156 [3] NCCL INFO Channel 22/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-163-58:934725:935157 [5] NCCL INFO Channel 07/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-163-58:934722:935155 [2] NCCL INFO Channel 19/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-163-58:934721:935153 [1] NCCL INFO Channel 11/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-163-58:934727:935152 [7] NCCL INFO Channel 13/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-163-58:934724:935154 [4] NCCL INFO Channel 11/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-163-58:934726:935151 [6] NCCL INFO Channel 20/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-163-58:934723:935156 [3] NCCL INFO Channel 23/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-163-58:934722:935155 [2] NCCL INFO Channel 20/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-163-58:934725:935157 [5] NCCL INFO Channel 08/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-163-58:934721:935153 [1] NCCL INFO Channel 12/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-163-58:934724:935154 [4] NCCL INFO Channel 12/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-163-58:934727:935152 [7] NCCL INFO Channel 14/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-163-58:934726:935151 [6] NCCL INFO Channel 21/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-163-58:934722:935155 [2] NCCL INFO Channel 21/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-163-58:934725:935157 [5] NCCL INFO Channel 09/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-163-58:934721:935153 [1] NCCL INFO Channel 13/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-163-58:934724:935154 [4] NCCL INFO Channel 13/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-163-58:934727:935152 [7] NCCL INFO Channel 15/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-163-58:934722:935155 [2] NCCL INFO Channel 22/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-163-58:934726:935151 [6] NCCL INFO Channel 22/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-163-58:934725:935157 [5] NCCL INFO Channel 10/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-163-58:934721:935153 [1] NCCL INFO Channel 14/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-163-58:934727:935152 [7] NCCL INFO Channel 16/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-163-58:934724:935154 [4] NCCL INFO Channel 14/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-163-58:934722:935155 [2] NCCL INFO Channel 23/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-163-58:934726:935151 [6] NCCL INFO Channel 23/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-163-58:934725:935157 [5] NCCL INFO Channel 11/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-163-58:934721:935153 [1] NCCL INFO Channel 15/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-170-132:3330183:3330626 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-170-132:3330183:3330626 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-163-58:934727:935152 [7] NCCL INFO Channel 17/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-163-58:934725:935157 [5] NCCL INFO Channel 12/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-163-58:934724:935154 [4] NCCL INFO Channel 15/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-163-58:934721:935153 [1] NCCL INFO Channel 16/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-163-58:934725:935157 [5] NCCL INFO Channel 13/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-163-58:934727:935152 [7] NCCL INFO Channel 18/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-163-58:934724:935154 [4] NCCL INFO Channel 16/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-163-58:934721:935153 [1] NCCL INFO Channel 17/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-163-58:934725:935157 [5] NCCL INFO Channel 14/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-163-58:934727:935152 [7] NCCL INFO Channel 19/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-163-58:934724:935154 [4] NCCL INFO Channel 17/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-163-58:934721:935153 [1] NCCL INFO Channel 18/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-163-58:934727:935152 [7] NCCL INFO Channel 20/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-163-58:934725:935157 [5] NCCL INFO Channel 15/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-163-58:934724:935154 [4] NCCL INFO Channel 18/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-163-58:934721:935153 [1] NCCL INFO Channel 19/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-163-58:934725:935157 [5] NCCL INFO Channel 16/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-163-58:934724:935154 [4] NCCL INFO Channel 19/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-163-58:934727:935152 [7] NCCL INFO Channel 21/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-163-58:934721:935153 [1] NCCL INFO Channel 20/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-163-58:934725:935157 [5] NCCL INFO Channel 17/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-163-58:934724:935154 [4] NCCL INFO Channel 20/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-163-58:934727:935152 [7] NCCL INFO Channel 22/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-163-58:934721:935153 [1] NCCL INFO Channel 21/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-163-58:934725:935157 [5] NCCL INFO Channel 18/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-163-58:934724:935154 [4] NCCL INFO Channel 21/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-163-58:934727:935152 [7] NCCL INFO Channel 23/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-163-58:934721:935153 [1] NCCL INFO Channel 22/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-163-58:934725:935157 [5] NCCL INFO Channel 19/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-163-58:934724:935154 [4] NCCL INFO Channel 22/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-163-58:934721:935153 [1] NCCL INFO Channel 23/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-163-58:934725:935157 [5] NCCL INFO Channel 20/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-163-58:934724:935154 [4] NCCL INFO Channel 23/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-163-58:934725:935157 [5] NCCL INFO Channel 21/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-170-132:3330190:3330627 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-170-132:3330190:3330627 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-170-31:3197185:3197627 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-170-31:3197185:3197627 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-162-14:2525080:2525511 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-162-14:2525080:2525511 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-170-132:3330186:3330629 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-170-132:3330186:3330629 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-163-58:934725:935157 [5] NCCL INFO Channel 22/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-170-31:3197186:3197629 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-170-31:3197186:3197629 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-170-132:3330185:3330628 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-170-132:3330185:3330628 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-170-132:3330184:3330631 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-170-132:3330184:3330631 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-162-14:2525076:2525513 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-162-14:2525076:2525513 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-162-14:2525075:2525509 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-162-14:2525075:2525509 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-163-58:934725:935157 [5] NCCL INFO Channel 23/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-170-31:3197184:3197630 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-170-31:3197184:3197630 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-170-31:3197183:3197625 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-170-31:3197183:3197625 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-170-31:3197183:3197625 [3] NCCL INFO Trees [0] 4/-1/-1->3->2 [1] 4/-1/-1->3->2 [2] 4/-1/-1->3->2 [3] 4/-1/-1->3->2 [4] 4/-1/-1->3->2 [5] 4/-1/-1->3->2 [6] 4/-1/-1->3->2 [7] 4/-1/-1->3->2 [8] 4/-1/-1->3->2 [9] 4/-1/-1->3->2 [10] 4/-1/-1->3->2 [11] 4/-1/-1->3->2 [12] 4/-1/-1->3->2 [13] 4/-1/-1->3->2 [14] 4/-1/-1->3->2 [15] 4/-1/-1->3->2 [16] 4/-1/-1->3->2 [17] 4/-1/-1->3->2 [18] 4/-1/-1->3->2 [19] 4/-1/-1->3->2 [20] 4/-1/-1->3->2 [21] 4/-1/-1->3->2 [22] 4/-1/-1->3->2 [23] 4/-1/-1->3->2
+ip-26-0-170-31:3197180:3197624 [0] NCCL INFO Channel 00/24 :    0   1   2   3   4   5   6   7
+ip-26-0-170-31:3197183:3197625 [3] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-170-31:3197180:3197624 [0] NCCL INFO Channel 01/24 :    0   1   2   3   4   5   6   7
+ip-26-0-170-31:3197180:3197624 [0] NCCL INFO Channel 02/24 :    0   1   2   3   4   5   6   7
+ip-26-0-170-31:3197187:3197626 [7] NCCL INFO Trees [0] -1/-1/-1->7->6 [1] -1/-1/-1->7->6 [2] -1/-1/-1->7->6 [3] -1/-1/-1->7->6 [4] -1/-1/-1->7->6 [5] -1/-1/-1->7->6 [6] -1/-1/-1->7->6 [7] -1/-1/-1->7->6 [8] -1/-1/-1->7->6 [9] -1/-1/-1->7->6 [10] -1/-1/-1->7->6 [11] -1/-1/-1->7->6 [12] -1/-1/-1->7->6 [13] -1/-1/-1->7->6 [14] -1/-1/-1->7->6 [15] -1/-1/-1->7->6 [16] -1/-1/-1->7->6 [17] -1/-1/-1->7->6 [18] -1/-1/-1->7->6 [19] -1/-1/-1->7->6 [20] -1/-1/-1->7->6 [21] -1/-1/-1->7->6 [22] -1/-1/-1->7->6 [23] -1/-1/-1->7->6
+ip-26-0-170-31:3197180:3197624 [0] NCCL INFO Channel 03/24 :    0   1   2   3   4   5   6   7
+ip-26-0-170-31:3197180:3197624 [0] NCCL INFO Channel 04/24 :    0   1   2   3   4   5   6   7
+ip-26-0-170-31:3197182:3197631 [2] NCCL INFO Trees [0] 3/-1/-1->2->1 [1] 3/-1/-1->2->1 [2] 3/-1/-1->2->1 [3] 3/-1/-1->2->1 [4] 3/-1/-1->2->1 [5] 3/-1/-1->2->1 [6] 3/-1/-1->2->1 [7] 3/-1/-1->2->1 [8] 3/-1/-1->2->1 [9] 3/-1/-1->2->1 [10] 3/-1/-1->2->1 [11] 3/-1/-1->2->1 [12] 3/-1/-1->2->1 [13] 3/-1/-1->2->1 [14] 3/-1/-1->2->1 [15] 3/-1/-1->2->1 [16] 3/-1/-1->2->1 [17] 3/-1/-1->2->1 [18] 3/-1/-1->2->1 [19] 3/-1/-1->2->1 [20] 3/-1/-1->2->1 [21] 3/-1/-1->2->1 [22] 3/-1/-1->2->1 [23] 3/-1/-1->2->1
+ip-26-0-170-31:3197187:3197626 [7] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-170-31:3197180:3197624 [0] NCCL INFO Channel 05/24 :    0   1   2   3   4   5   6   7
+ip-26-0-170-31:3197180:3197624 [0] NCCL INFO Channel 06/24 :    0   1   2   3   4   5   6   7
+ip-26-0-170-31:3197181:3197628 [1] NCCL INFO Trees [0] 2/-1/-1->1->0 [1] 2/-1/-1->1->0 [2] 2/-1/-1->1->0 [3] 2/-1/-1->1->0 [4] 2/-1/-1->1->0 [5] 2/-1/-1->1->0 [6] 2/-1/-1->1->0 [7] 2/-1/-1->1->0 [8] 2/-1/-1->1->0 [9] 2/-1/-1->1->0 [10] 2/-1/-1->1->0 [11] 2/-1/-1->1->0 [12] 2/-1/-1->1->0 [13] 2/-1/-1->1->0 [14] 2/-1/-1->1->0 [15] 2/-1/-1->1->0 [16] 2/-1/-1->1->0 [17] 2/-1/-1->1->0 [18] 2/-1/-1->1->0 [19] 2/-1/-1->1->0 [20] 2/-1/-1->1->0 [21] 2/-1/-1->1->0 [22] 2/-1/-1->1->0 [23] 2/-1/-1->1->0
+ip-26-0-170-31:3197185:3197627 [5] NCCL INFO Trees [0] 6/-1/-1->5->4 [1] 6/-1/-1->5->4 [2] 6/-1/-1->5->4 [3] 6/-1/-1->5->4 [4] 6/-1/-1->5->4 [5] 6/-1/-1->5->4 [6] 6/-1/-1->5->4 [7] 6/-1/-1->5->4 [8] 6/-1/-1->5->4 [9] 6/-1/-1->5->4 [10] 6/-1/-1->5->4 [11] 6/-1/-1->5->4 [12] 6/-1/-1->5->4 [13] 6/-1/-1->5->4 [14] 6/-1/-1->5->4 [15] 6/-1/-1->5->4 [16] 6/-1/-1->5->4 [17] 6/-1/-1->5->4 [18] 6/-1/-1->5->4 [19] 6/-1/-1->5->4 [20] 6/-1/-1->5->4 [21] 6/-1/-1->5->4 [22] 6/-1/-1->5->4 [23] 6/-1/-1->5->4
+ip-26-0-170-31:3197184:3197630 [4] NCCL INFO Trees [0] 5/-1/-1->4->3 [1] 5/-1/-1->4->3 [2] 5/-1/-1->4->3 [3] 5/-1/-1->4->3 [4] 5/-1/-1->4->3 [5] 5/-1/-1->4->3 [6] 5/-1/-1->4->3 [7] 5/-1/-1->4->3 [8] 5/-1/-1->4->3 [9] 5/-1/-1->4->3 [10] 5/-1/-1->4->3 [11] 5/-1/-1->4->3 [12] 5/-1/-1->4->3 [13] 5/-1/-1->4->3 [14] 5/-1/-1->4->3 [15] 5/-1/-1->4->3 [16] 5/-1/-1->4->3 [17] 5/-1/-1->4->3 [18] 5/-1/-1->4->3 [19] 5/-1/-1->4->3 [20] 5/-1/-1->4->3 [21] 5/-1/-1->4->3 [22] 5/-1/-1->4->3 [23] 5/-1/-1->4->3
+ip-26-0-170-31:3197182:3197631 [2] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-170-31:3197180:3197624 [0] NCCL INFO Channel 07/24 :    0   1   2   3   4   5   6   7
+ip-26-0-170-31:3197185:3197627 [5] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-170-31:3197184:3197630 [4] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-170-31:3197186:3197629 [6] NCCL INFO Trees [0] 7/-1/-1->6->5 [1] 7/-1/-1->6->5 [2] 7/-1/-1->6->5 [3] 7/-1/-1->6->5 [4] 7/-1/-1->6->5 [5] 7/-1/-1->6->5 [6] 7/-1/-1->6->5 [7] 7/-1/-1->6->5 [8] 7/-1/-1->6->5 [9] 7/-1/-1->6->5 [10] 7/-1/-1->6->5 [11] 7/-1/-1->6->5 [12] 7/-1/-1->6->5 [13] 7/-1/-1->6->5 [14] 7/-1/-1->6->5 [15] 7/-1/-1->6->5 [16] 7/-1/-1->6->5 [17] 7/-1/-1->6->5 [18] 7/-1/-1->6->5 [19] 7/-1/-1->6->5 [20] 7/-1/-1->6->5 [21] 7/-1/-1->6->5 [22] 7/-1/-1->6->5 [23] 7/-1/-1->6->5
+ip-26-0-170-31:3197180:3197624 [0] NCCL INFO Channel 08/24 :    0   1   2   3   4   5   6   7
+ip-26-0-170-31:3197181:3197628 [1] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-170-31:3197180:3197624 [0] NCCL INFO Channel 09/24 :    0   1   2   3   4   5   6   7
+ip-26-0-170-31:3197186:3197629 [6] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-170-31:3197180:3197624 [0] NCCL INFO Channel 10/24 :    0   1   2   3   4   5   6   7
+ip-26-0-170-31:3197180:3197624 [0] NCCL INFO Channel 11/24 :    0   1   2   3   4   5   6   7
+ip-26-0-170-31:3197180:3197624 [0] NCCL INFO Channel 12/24 :    0   1   2   3   4   5   6   7
+ip-26-0-170-31:3197180:3197624 [0] NCCL INFO Channel 13/24 :    0   1   2   3   4   5   6   7
+ip-26-0-170-31:3197180:3197624 [0] NCCL INFO Channel 14/24 :    0   1   2   3   4   5   6   7
+ip-26-0-170-31:3197180:3197624 [0] NCCL INFO Channel 15/24 :    0   1   2   3   4   5   6   7
+ip-26-0-170-31:3197180:3197624 [0] NCCL INFO Channel 16/24 :    0   1   2   3   4   5   6   7
+ip-26-0-170-31:3197180:3197624 [0] NCCL INFO Channel 17/24 :    0   1   2   3   4   5   6   7
+ip-26-0-170-31:3197180:3197624 [0] NCCL INFO Channel 18/24 :    0   1   2   3   4   5   6   7
+ip-26-0-170-31:3197180:3197624 [0] NCCL INFO Channel 19/24 :    0   1   2   3   4   5   6   7
+ip-26-0-170-31:3197180:3197624 [0] NCCL INFO Channel 20/24 :    0   1   2   3   4   5   6   7
+ip-26-0-170-31:3197180:3197624 [0] NCCL INFO Channel 21/24 :    0   1   2   3   4   5   6   7
+ip-26-0-170-31:3197180:3197624 [0] NCCL INFO Channel 22/24 :    0   1   2   3   4   5   6   7
+ip-26-0-170-31:3197180:3197624 [0] NCCL INFO Channel 23/24 :    0   1   2   3   4   5   6   7
+ip-26-0-170-31:3197180:3197624 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] 1/-1/-1->0->-1 [7] 1/-1/-1->0->-1 [8] 1/-1/-1->0->-1 [9] 1/-1/-1->0->-1 [10] 1/-1/-1->0->-1 [11] 1/-1/-1->0->-1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] 1/-1/-1->0->-1 [19] 1/-1/-1->0->-1 [20] 1/-1/-1->0->-1 [21] 1/-1/-1->0->-1 [22] 1/-1/-1->0->-1 [23] 1/-1/-1->0->-1
+ip-26-0-170-31:3197180:3197624 [0] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-162-14:2525078:2525510 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-162-14:2525078:2525510 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-170-132:3330189:3330633 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-170-132:3330189:3330633 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-170-31:3197187:3197626 [7] NCCL INFO Channel 00/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-170-31:3197180:3197624 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-170-31:3197182:3197631 [2] NCCL INFO Channel 00/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-170-31:3197187:3197626 [7] NCCL INFO Channel 01/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-170-31:3197186:3197629 [6] NCCL INFO Channel 00/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-170-31:3197180:3197624 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-170-31:3197182:3197631 [2] NCCL INFO Channel 01/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-170-31:3197187:3197626 [7] NCCL INFO Channel 02/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-170-31:3197180:3197624 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-170-31:3197186:3197629 [6] NCCL INFO Channel 01/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-170-31:3197182:3197631 [2] NCCL INFO Channel 02/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-170-31:3197187:3197626 [7] NCCL INFO Channel 03/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-170-31:3197187:3197626 [7] NCCL INFO Channel 04/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-170-31:3197187:3197626 [7] NCCL INFO Channel 05/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-170-31:3197186:3197629 [6] NCCL INFO Channel 02/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-170-31:3197181:3197628 [1] NCCL INFO Channel 00/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-170-31:3197187:3197626 [7] NCCL INFO Channel 06/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-170-31:3197186:3197629 [6] NCCL INFO Channel 03/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-170-31:3197181:3197628 [1] NCCL INFO Channel 01/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-170-31:3197187:3197626 [7] NCCL INFO Channel 07/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-170-31:3197186:3197629 [6] NCCL INFO Channel 04/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-170-31:3197181:3197628 [1] NCCL INFO Channel 02/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-170-31:3197187:3197626 [7] NCCL INFO Channel 08/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-170-31:3197186:3197629 [6] NCCL INFO Channel 05/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-170-31:3197181:3197628 [1] NCCL INFO Channel 03/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-170-31:3197187:3197626 [7] NCCL INFO Channel 09/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-170-31:3197186:3197629 [6] NCCL INFO Channel 06/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-170-132:3330187:3330630 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-170-132:3330187:3330630 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-170-31:3197181:3197628 [1] NCCL INFO Channel 04/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-170-132:3330187:3330630 [4] NCCL INFO Trees [0] 5/-1/-1->4->3 [1] 5/-1/-1->4->3 [2] 5/-1/-1->4->3 [3] 5/-1/-1->4->3 [4] 5/-1/-1->4->3 [5] 5/-1/-1->4->3 [6] 5/-1/-1->4->3 [7] 5/-1/-1->4->3 [8] 5/-1/-1->4->3 [9] 5/-1/-1->4->3 [10] 5/-1/-1->4->3 [11] 5/-1/-1->4->3 [12] 5/-1/-1->4->3 [13] 5/-1/-1->4->3 [14] 5/-1/-1->4->3 [15] 5/-1/-1->4->3 [16] 5/-1/-1->4->3 [17] 5/-1/-1->4->3 [18] 5/-1/-1->4->3 [19] 5/-1/-1->4->3 [20] 5/-1/-1->4->3 [21] 5/-1/-1->4->3 [22] 5/-1/-1->4->3 [23] 5/-1/-1->4->3
+ip-26-0-170-132:3330187:3330630 [4] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-170-132:3330188:3330632 [5] NCCL INFO Trees [0] 6/-1/-1->5->4 [1] 6/-1/-1->5->4 [2] 6/-1/-1->5->4 [3] 6/-1/-1->5->4 [4] 6/-1/-1->5->4 [5] 6/-1/-1->5->4 [6] 6/-1/-1->5->4 [7] 6/-1/-1->5->4 [8] 6/-1/-1->5->4 [9] 6/-1/-1->5->4 [10] 6/-1/-1->5->4 [11] 6/-1/-1->5->4 [12] 6/-1/-1->5->4 [13] 6/-1/-1->5->4 [14] 6/-1/-1->5->4 [15] 6/-1/-1->5->4 [16] 6/-1/-1->5->4 [17] 6/-1/-1->5->4 [18] 6/-1/-1->5->4 [19] 6/-1/-1->5->4 [20] 6/-1/-1->5->4 [21] 6/-1/-1->5->4 [22] 6/-1/-1->5->4 [23] 6/-1/-1->5->4
+ip-26-0-170-132:3330188:3330632 [5] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-170-132:3330189:3330633 [6] NCCL INFO Trees [0] 7/-1/-1->6->5 [1] 7/-1/-1->6->5 [2] 7/-1/-1->6->5 [3] 7/-1/-1->6->5 [4] 7/-1/-1->6->5 [5] 7/-1/-1->6->5 [6] 7/-1/-1->6->5 [7] 7/-1/-1->6->5 [8] 7/-1/-1->6->5 [9] 7/-1/-1->6->5 [10] 7/-1/-1->6->5 [11] 7/-1/-1->6->5 [12] 7/-1/-1->6->5 [13] 7/-1/-1->6->5 [14] 7/-1/-1->6->5 [15] 7/-1/-1->6->5 [16] 7/-1/-1->6->5 [17] 7/-1/-1->6->5 [18] 7/-1/-1->6->5 [19] 7/-1/-1->6->5 [20] 7/-1/-1->6->5 [21] 7/-1/-1->6->5 [22] 7/-1/-1->6->5 [23] 7/-1/-1->6->5
+ip-26-0-170-132:3330189:3330633 [6] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-170-132:3330190:3330627 [7] NCCL INFO Trees [0] -1/-1/-1->7->6 [1] -1/-1/-1->7->6 [2] -1/-1/-1->7->6 [3] -1/-1/-1->7->6 [4] -1/-1/-1->7->6 [5] -1/-1/-1->7->6 [6] -1/-1/-1->7->6 [7] -1/-1/-1->7->6 [8] -1/-1/-1->7->6 [9] -1/-1/-1->7->6 [10] -1/-1/-1->7->6 [11] -1/-1/-1->7->6 [12] -1/-1/-1->7->6 [13] -1/-1/-1->7->6 [14] -1/-1/-1->7->6 [15] -1/-1/-1->7->6 [16] -1/-1/-1->7->6 [17] -1/-1/-1->7->6 [18] -1/-1/-1->7->6 [19] -1/-1/-1->7->6 [20] -1/-1/-1->7->6 [21] -1/-1/-1->7->6 [22] -1/-1/-1->7->6 [23] -1/-1/-1->7->6
+ip-26-0-170-132:3330190:3330627 [7] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-170-31:3197187:3197626 [7] NCCL INFO Channel 10/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-170-31:3197186:3197629 [6] NCCL INFO Channel 07/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-170-31:3197181:3197628 [1] NCCL INFO Channel 05/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-170-31:3197186:3197629 [6] NCCL INFO Channel 08/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-170-31:3197187:3197626 [7] NCCL INFO Channel 11/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-170-31:3197182:3197631 [2] NCCL INFO Channel 03/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-170-31:3197180:3197624 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-170-31:3197181:3197628 [1] NCCL INFO Channel 06/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-170-31:3197187:3197626 [7] NCCL INFO Channel 12/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-170-31:3197186:3197629 [6] NCCL INFO Channel 09/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-170-132:3330183:3330626 [0] NCCL INFO Channel 00/24 :    0   1   2   3   4   5   6   7
+ip-26-0-170-132:3330183:3330626 [0] NCCL INFO Channel 01/24 :    0   1   2   3   4   5   6   7
+ip-26-0-170-132:3330183:3330626 [0] NCCL INFO Channel 02/24 :    0   1   2   3   4   5   6   7
+ip-26-0-170-132:3330183:3330626 [0] NCCL INFO Channel 03/24 :    0   1   2   3   4   5   6   7
+ip-26-0-170-132:3330183:3330626 [0] NCCL INFO Channel 04/24 :    0   1   2   3   4   5   6   7
+ip-26-0-170-132:3330183:3330626 [0] NCCL INFO Channel 05/24 :    0   1   2   3   4   5   6   7
+ip-26-0-170-132:3330183:3330626 [0] NCCL INFO Channel 06/24 :    0   1   2   3   4   5   6   7
+ip-26-0-170-132:3330183:3330626 [0] NCCL INFO Channel 07/24 :    0   1   2   3   4   5   6   7
+ip-26-0-170-132:3330183:3330626 [0] NCCL INFO Channel 08/24 :    0   1   2   3   4   5   6   7
+ip-26-0-170-132:3330186:3330629 [3] NCCL INFO Trees [0] 4/-1/-1->3->2 [1] 4/-1/-1->3->2 [2] 4/-1/-1->3->2 [3] 4/-1/-1->3->2 [4] 4/-1/-1->3->2 [5] 4/-1/-1->3->2 [6] 4/-1/-1->3->2 [7] 4/-1/-1->3->2 [8] 4/-1/-1->3->2 [9] 4/-1/-1->3->2 [10] 4/-1/-1->3->2 [11] 4/-1/-1->3->2 [12] 4/-1/-1->3->2 [13] 4/-1/-1->3->2 [14] 4/-1/-1->3->2 [15] 4/-1/-1->3->2 [16] 4/-1/-1->3->2 [17] 4/-1/-1->3->2 [18] 4/-1/-1->3->2 [19] 4/-1/-1->3->2 [20] 4/-1/-1->3->2 [21] 4/-1/-1->3->2 [22] 4/-1/-1->3->2 [23] 4/-1/-1->3->2
+ip-26-0-170-132:3330183:3330626 [0] NCCL INFO Channel 09/24 :    0   1   2   3   4   5   6   7
+ip-26-0-170-132:3330183:3330626 [0] NCCL INFO Channel 10/24 :    0   1   2   3   4   5   6   7
+ip-26-0-170-132:3330186:3330629 [3] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-170-132:3330183:3330626 [0] NCCL INFO Channel 11/24 :    0   1   2   3   4   5   6   7
+ip-26-0-170-132:3330185:3330628 [2] NCCL INFO Trees [0] 3/-1/-1->2->1 [1] 3/-1/-1->2->1 [2] 3/-1/-1->2->1 [3] 3/-1/-1->2->1 [4] 3/-1/-1->2->1 [5] 3/-1/-1->2->1 [6] 3/-1/-1->2->1 [7] 3/-1/-1->2->1 [8] 3/-1/-1->2->1 [9] 3/-1/-1->2->1 [10] 3/-1/-1->2->1 [11] 3/-1/-1->2->1 [12] 3/-1/-1->2->1 [13] 3/-1/-1->2->1 [14] 3/-1/-1->2->1 [15] 3/-1/-1->2->1 [16] 3/-1/-1->2->1 [17] 3/-1/-1->2->1 [18] 3/-1/-1->2->1 [19] 3/-1/-1->2->1 [20] 3/-1/-1->2->1 [21] 3/-1/-1->2->1 [22] 3/-1/-1->2->1 [23] 3/-1/-1->2->1
+ip-26-0-170-132:3330183:3330626 [0] NCCL INFO Channel 12/24 :    0   1   2   3   4   5   6   7
+ip-26-0-170-132:3330183:3330626 [0] NCCL INFO Channel 13/24 :    0   1   2   3   4   5   6   7
+ip-26-0-170-132:3330185:3330628 [2] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-170-132:3330183:3330626 [0] NCCL INFO Channel 14/24 :    0   1   2   3   4   5   6   7
+ip-26-0-170-132:3330184:3330631 [1] NCCL INFO Trees [0] 2/-1/-1->1->0 [1] 2/-1/-1->1->0 [2] 2/-1/-1->1->0 [3] 2/-1/-1->1->0 [4] 2/-1/-1->1->0 [5] 2/-1/-1->1->0 [6] 2/-1/-1->1->0 [7] 2/-1/-1->1->0 [8] 2/-1/-1->1->0 [9] 2/-1/-1->1->0 [10] 2/-1/-1->1->0 [11] 2/-1/-1->1->0 [12] 2/-1/-1->1->0 [13] 2/-1/-1->1->0 [14] 2/-1/-1->1->0 [15] 2/-1/-1->1->0 [16] 2/-1/-1->1->0 [17] 2/-1/-1->1->0 [18] 2/-1/-1->1->0 [19] 2/-1/-1->1->0 [20] 2/-1/-1->1->0 [21] 2/-1/-1->1->0 [22] 2/-1/-1->1->0 [23] 2/-1/-1->1->0
+ip-26-0-170-132:3330183:3330626 [0] NCCL INFO Channel 15/24 :    0   1   2   3   4   5   6   7
+ip-26-0-170-132:3330184:3330631 [1] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-170-132:3330183:3330626 [0] NCCL INFO Channel 16/24 :    0   1   2   3   4   5   6   7
+ip-26-0-170-132:3330183:3330626 [0] NCCL INFO Channel 17/24 :    0   1   2   3   4   5   6   7
+ip-26-0-170-132:3330183:3330626 [0] NCCL INFO Channel 18/24 :    0   1   2   3   4   5   6   7
+ip-26-0-170-132:3330183:3330626 [0] NCCL INFO Channel 19/24 :    0   1   2   3   4   5   6   7
+ip-26-0-170-132:3330183:3330626 [0] NCCL INFO Channel 20/24 :    0   1   2   3   4   5   6   7
+ip-26-0-170-132:3330183:3330626 [0] NCCL INFO Channel 21/24 :    0   1   2   3   4   5   6   7
+ip-26-0-170-132:3330183:3330626 [0] NCCL INFO Channel 22/24 :    0   1   2   3   4   5   6   7
+ip-26-0-170-132:3330183:3330626 [0] NCCL INFO Channel 23/24 :    0   1   2   3   4   5   6   7
+ip-26-0-170-132:3330183:3330626 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] 1/-1/-1->0->-1 [7] 1/-1/-1->0->-1 [8] 1/-1/-1->0->-1 [9] 1/-1/-1->0->-1 [10] 1/-1/-1->0->-1 [11] 1/-1/-1->0->-1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] 1/-1/-1->0->-1 [19] 1/-1/-1->0->-1 [20] 1/-1/-1->0->-1 [21] 1/-1/-1->0->-1 [22] 1/-1/-1->0->-1 [23] 1/-1/-1->0->-1
+ip-26-0-170-132:3330183:3330626 [0] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-170-31:3197182:3197631 [2] NCCL INFO Channel 04/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-170-31:3197180:3197624 [0] NCCL INFO Channel 04/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-170-31:3197181:3197628 [1] NCCL INFO Channel 07/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-162-14:2525077:2525512 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-170-31:3197187:3197626 [7] NCCL INFO Channel 13/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-170-31:3197186:3197629 [6] NCCL INFO Channel 10/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-162-14:2525077:2525512 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-170-31:3197182:3197631 [2] NCCL INFO Channel 05/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-170-31:3197180:3197624 [0] NCCL INFO Channel 05/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-170-31:3197181:3197628 [1] NCCL INFO Channel 08/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-170-31:3197187:3197626 [7] NCCL INFO Channel 14/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-170-31:3197186:3197629 [6] NCCL INFO Channel 11/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-162-14:2525081:2525514 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-162-14:2525081:2525514 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-170-31:3197182:3197631 [2] NCCL INFO Channel 06/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-170-31:3197181:3197628 [1] NCCL INFO Channel 09/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-170-31:3197180:3197624 [0] NCCL INFO Channel 06/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-170-31:3197183:3197625 [3] NCCL INFO Channel 00/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-170-31:3197184:3197630 [4] NCCL INFO Channel 00/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-170-31:3197187:3197626 [7] NCCL INFO Channel 15/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-170-31:3197186:3197629 [6] NCCL INFO Channel 12/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-170-31:3197181:3197628 [1] NCCL INFO Channel 10/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-170-31:3197183:3197625 [3] NCCL INFO Channel 01/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-170-31:3197184:3197630 [4] NCCL INFO Channel 01/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-170-31:3197187:3197626 [7] NCCL INFO Channel 16/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-170-31:3197186:3197629 [6] NCCL INFO Channel 13/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-170-31:3197181:3197628 [1] NCCL INFO Channel 11/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-170-31:3197183:3197625 [3] NCCL INFO Channel 02/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-170-31:3197187:3197626 [7] NCCL INFO Channel 17/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-170-31:3197184:3197630 [4] NCCL INFO Channel 02/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-170-31:3197186:3197629 [6] NCCL INFO Channel 14/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-170-31:3197185:3197627 [5] NCCL INFO Channel 00/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-170-31:3197181:3197628 [1] NCCL INFO Channel 12/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-170-31:3197183:3197625 [3] NCCL INFO Channel 03/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-170-31:3197187:3197626 [7] NCCL INFO Channel 18/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-170-31:3197184:3197630 [4] NCCL INFO Channel 03/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-170-31:3197186:3197629 [6] NCCL INFO Channel 15/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-162-14:2525082:2525515 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-162-14:2525082:2525515 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-170-31:3197185:3197627 [5] NCCL INFO Channel 01/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-162-14:2525082:2525515 [7] NCCL INFO Trees [0] -1/-1/-1->7->6 [1] -1/-1/-1->7->6 [2] -1/-1/-1->7->6 [3] -1/-1/-1->7->6 [4] -1/-1/-1->7->6 [5] -1/-1/-1->7->6 [6] -1/-1/-1->7->6 [7] -1/-1/-1->7->6 [8] -1/-1/-1->7->6 [9] -1/-1/-1->7->6 [10] -1/-1/-1->7->6 [11] -1/-1/-1->7->6 [12] -1/-1/-1->7->6 [13] -1/-1/-1->7->6 [14] -1/-1/-1->7->6 [15] -1/-1/-1->7->6 [16] -1/-1/-1->7->6 [17] -1/-1/-1->7->6 [18] -1/-1/-1->7->6 [19] -1/-1/-1->7->6 [20] -1/-1/-1->7->6 [21] -1/-1/-1->7->6 [22] -1/-1/-1->7->6 [23] -1/-1/-1->7->6
+ip-26-0-162-14:2525082:2525515 [7] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-162-14:2525081:2525514 [6] NCCL INFO Trees [0] 7/-1/-1->6->5 [1] 7/-1/-1->6->5 [2] 7/-1/-1->6->5 [3] 7/-1/-1->6->5 [4] 7/-1/-1->6->5 [5] 7/-1/-1->6->5 [6] 7/-1/-1->6->5 [7] 7/-1/-1->6->5 [8] 7/-1/-1->6->5 [9] 7/-1/-1->6->5 [10] 7/-1/-1->6->5 [11] 7/-1/-1->6->5 [12] 7/-1/-1->6->5 [13] 7/-1/-1->6->5 [14] 7/-1/-1->6->5 [15] 7/-1/-1->6->5 [16] 7/-1/-1->6->5 [17] 7/-1/-1->6->5 [18] 7/-1/-1->6->5 [19] 7/-1/-1->6->5 [20] 7/-1/-1->6->5 [21] 7/-1/-1->6->5 [22] 7/-1/-1->6->5 [23] 7/-1/-1->6->5
+ip-26-0-162-14:2525081:2525514 [6] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-162-14:2525077:2525512 [2] NCCL INFO Trees [0] 3/-1/-1->2->1 [1] 3/-1/-1->2->1 [2] 3/-1/-1->2->1 [3] 3/-1/-1->2->1 [4] 3/-1/-1->2->1 [5] 3/-1/-1->2->1 [6] 3/-1/-1->2->1 [7] 3/-1/-1->2->1 [8] 3/-1/-1->2->1 [9] 3/-1/-1->2->1 [10] 3/-1/-1->2->1 [11] 3/-1/-1->2->1 [12] 3/-1/-1->2->1 [13] 3/-1/-1->2->1 [14] 3/-1/-1->2->1 [15] 3/-1/-1->2->1 [16] 3/-1/-1->2->1 [17] 3/-1/-1->2->1 [18] 3/-1/-1->2->1 [19] 3/-1/-1->2->1 [20] 3/-1/-1->2->1 [21] 3/-1/-1->2->1 [22] 3/-1/-1->2->1 [23] 3/-1/-1->2->1
+ip-26-0-162-14:2525076:2525513 [1] NCCL INFO Trees [0] 2/-1/-1->1->0 [1] 2/-1/-1->1->0 [2] 2/-1/-1->1->0 [3] 2/-1/-1->1->0 [4] 2/-1/-1->1->0 [5] 2/-1/-1->1->0 [6] 2/-1/-1->1->0 [7] 2/-1/-1->1->0 [8] 2/-1/-1->1->0 [9] 2/-1/-1->1->0 [10] 2/-1/-1->1->0 [11] 2/-1/-1->1->0 [12] 2/-1/-1->1->0 [13] 2/-1/-1->1->0 [14] 2/-1/-1->1->0 [15] 2/-1/-1->1->0 [16] 2/-1/-1->1->0 [17] 2/-1/-1->1->0 [18] 2/-1/-1->1->0 [19] 2/-1/-1->1->0 [20] 2/-1/-1->1->0 [21] 2/-1/-1->1->0 [22] 2/-1/-1->1->0 [23] 2/-1/-1->1->0
+ip-26-0-162-14:2525080:2525511 [5] NCCL INFO Trees [0] 6/-1/-1->5->4 [1] 6/-1/-1->5->4 [2] 6/-1/-1->5->4 [3] 6/-1/-1->5->4 [4] 6/-1/-1->5->4 [5] 6/-1/-1->5->4 [6] 6/-1/-1->5->4 [7] 6/-1/-1->5->4 [8] 6/-1/-1->5->4 [9] 6/-1/-1->5->4 [10] 6/-1/-1->5->4 [11] 6/-1/-1->5->4 [12] 6/-1/-1->5->4 [13] 6/-1/-1->5->4 [14] 6/-1/-1->5->4 [15] 6/-1/-1->5->4 [16] 6/-1/-1->5->4 [17] 6/-1/-1->5->4 [18] 6/-1/-1->5->4 [19] 6/-1/-1->5->4 [20] 6/-1/-1->5->4 [21] 6/-1/-1->5->4 [22] 6/-1/-1->5->4 [23] 6/-1/-1->5->4
+ip-26-0-162-14:2525078:2525510 [3] NCCL INFO Trees [0] 4/-1/-1->3->2 [1] 4/-1/-1->3->2 [2] 4/-1/-1->3->2 [3] 4/-1/-1->3->2 [4] 4/-1/-1->3->2 [5] 4/-1/-1->3->2 [6] 4/-1/-1->3->2 [7] 4/-1/-1->3->2 [8] 4/-1/-1->3->2 [9] 4/-1/-1->3->2 [10] 4/-1/-1->3->2 [11] 4/-1/-1->3->2 [12] 4/-1/-1->3->2 [13] 4/-1/-1->3->2 [14] 4/-1/-1->3->2 [15] 4/-1/-1->3->2 [16] 4/-1/-1->3->2 [17] 4/-1/-1->3->2 [18] 4/-1/-1->3->2 [19] 4/-1/-1->3->2 [20] 4/-1/-1->3->2 [21] 4/-1/-1->3->2 [22] 4/-1/-1->3->2 [23] 4/-1/-1->3->2
+ip-26-0-162-14:2525076:2525513 [1] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-162-14:2525077:2525512 [2] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-162-14:2525079:2525516 [4] NCCL INFO Trees [0] 5/-1/-1->4->3 [1] 5/-1/-1->4->3 [2] 5/-1/-1->4->3 [3] 5/-1/-1->4->3 [4] 5/-1/-1->4->3 [5] 5/-1/-1->4->3 [6] 5/-1/-1->4->3 [7] 5/-1/-1->4->3 [8] 5/-1/-1->4->3 [9] 5/-1/-1->4->3 [10] 5/-1/-1->4->3 [11] 5/-1/-1->4->3 [12] 5/-1/-1->4->3 [13] 5/-1/-1->4->3 [14] 5/-1/-1->4->3 [15] 5/-1/-1->4->3 [16] 5/-1/-1->4->3 [17] 5/-1/-1->4->3 [18] 5/-1/-1->4->3 [19] 5/-1/-1->4->3 [20] 5/-1/-1->4->3 [21] 5/-1/-1->4->3 [22] 5/-1/-1->4->3 [23] 5/-1/-1->4->3
+ip-26-0-162-14:2525078:2525510 [3] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-162-14:2525080:2525511 [5] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-162-14:2525075:2525509 [0] NCCL INFO Channel 00/24 :    0   1   2   3   4   5   6   7
+ip-26-0-162-14:2525079:2525516 [4] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-162-14:2525075:2525509 [0] NCCL INFO Channel 01/24 :    0   1   2   3   4   5   6   7
+ip-26-0-162-14:2525075:2525509 [0] NCCL INFO Channel 02/24 :    0   1   2   3   4   5   6   7
+ip-26-0-162-14:2525075:2525509 [0] NCCL INFO Channel 03/24 :    0   1   2   3   4   5   6   7
+ip-26-0-162-14:2525075:2525509 [0] NCCL INFO Channel 04/24 :    0   1   2   3   4   5   6   7
+ip-26-0-162-14:2525075:2525509 [0] NCCL INFO Channel 05/24 :    0   1   2   3   4   5   6   7
+ip-26-0-162-14:2525075:2525509 [0] NCCL INFO Channel 06/24 :    0   1   2   3   4   5   6   7
+ip-26-0-162-14:2525075:2525509 [0] NCCL INFO Channel 07/24 :    0   1   2   3   4   5   6   7
+ip-26-0-162-14:2525075:2525509 [0] NCCL INFO Channel 08/24 :    0   1   2   3   4   5   6   7
+ip-26-0-162-14:2525075:2525509 [0] NCCL INFO Channel 09/24 :    0   1   2   3   4   5   6   7
+ip-26-0-162-14:2525075:2525509 [0] NCCL INFO Channel 10/24 :    0   1   2   3   4   5   6   7
+ip-26-0-162-14:2525075:2525509 [0] NCCL INFO Channel 11/24 :    0   1   2   3   4   5   6   7
+ip-26-0-162-14:2525075:2525509 [0] NCCL INFO Channel 12/24 :    0   1   2   3   4   5   6   7
+ip-26-0-162-14:2525075:2525509 [0] NCCL INFO Channel 13/24 :    0   1   2   3   4   5   6   7
+ip-26-0-162-14:2525075:2525509 [0] NCCL INFO Channel 14/24 :    0   1   2   3   4   5   6   7
+ip-26-0-162-14:2525075:2525509 [0] NCCL INFO Channel 15/24 :    0   1   2   3   4   5   6   7
+ip-26-0-162-14:2525075:2525509 [0] NCCL INFO Channel 16/24 :    0   1   2   3   4   5   6   7
+ip-26-0-162-14:2525075:2525509 [0] NCCL INFO Channel 17/24 :    0   1   2   3   4   5   6   7
+ip-26-0-162-14:2525075:2525509 [0] NCCL INFO Channel 18/24 :    0   1   2   3   4   5   6   7
+ip-26-0-162-14:2525075:2525509 [0] NCCL INFO Channel 19/24 :    0   1   2   3   4   5   6   7
+ip-26-0-162-14:2525075:2525509 [0] NCCL INFO Channel 20/24 :    0   1   2   3   4   5   6   7
+ip-26-0-162-14:2525075:2525509 [0] NCCL INFO Channel 21/24 :    0   1   2   3   4   5   6   7
+ip-26-0-162-14:2525075:2525509 [0] NCCL INFO Channel 22/24 :    0   1   2   3   4   5   6   7
+ip-26-0-162-14:2525075:2525509 [0] NCCL INFO Channel 23/24 :    0   1   2   3   4   5   6   7
+ip-26-0-162-14:2525075:2525509 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] 1/-1/-1->0->-1 [7] 1/-1/-1->0->-1 [8] 1/-1/-1->0->-1 [9] 1/-1/-1->0->-1 [10] 1/-1/-1->0->-1 [11] 1/-1/-1->0->-1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] 1/-1/-1->0->-1 [19] 1/-1/-1->0->-1 [20] 1/-1/-1->0->-1 [21] 1/-1/-1->0->-1 [22] 1/-1/-1->0->-1 [23] 1/-1/-1->0->-1
+ip-26-0-162-14:2525075:2525509 [0] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-170-31:3197181:3197628 [1] NCCL INFO Channel 13/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-170-31:3197180:3197624 [0] NCCL INFO Channel 07/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-170-31:3197183:3197625 [3] NCCL INFO Channel 04/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-170-31:3197187:3197626 [7] NCCL INFO Channel 19/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-170-31:3197186:3197629 [6] NCCL INFO Channel 16/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-170-31:3197184:3197630 [4] NCCL INFO Channel 04/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-170-31:3197182:3197631 [2] NCCL INFO Channel 07/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-170-31:3197185:3197627 [5] NCCL INFO Channel 02/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-170-31:3197181:3197628 [1] NCCL INFO Channel 14/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-170-31:3197180:3197624 [0] NCCL INFO Channel 08/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-170-31:3197183:3197625 [3] NCCL INFO Channel 05/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-170-31:3197187:3197626 [7] NCCL INFO Channel 20/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-170-31:3197186:3197629 [6] NCCL INFO Channel 17/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-170-31:3197184:3197630 [4] NCCL INFO Channel 05/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-170-31:3197182:3197631 [2] NCCL INFO Channel 08/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-170-31:3197185:3197627 [5] NCCL INFO Channel 03/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-170-31:3197181:3197628 [1] NCCL INFO Channel 15/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-170-31:3197180:3197624 [0] NCCL INFO Channel 09/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-170-31:3197183:3197625 [3] NCCL INFO Channel 06/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-170-31:3197187:3197626 [7] NCCL INFO Channel 21/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-170-31:3197186:3197629 [6] NCCL INFO Channel 18/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-170-31:3197184:3197630 [4] NCCL INFO Channel 06/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-170-31:3197185:3197627 [5] NCCL INFO Channel 04/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-170-31:3197182:3197631 [2] NCCL INFO Channel 09/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-170-31:3197181:3197628 [1] NCCL INFO Channel 16/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-170-31:3197180:3197624 [0] NCCL INFO Channel 10/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-170-31:3197183:3197625 [3] NCCL INFO Channel 07/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-170-31:3197187:3197626 [7] NCCL INFO Channel 22/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-170-31:3197186:3197629 [6] NCCL INFO Channel 19/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-170-31:3197184:3197630 [4] NCCL INFO Channel 07/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-170-31:3197185:3197627 [5] NCCL INFO Channel 05/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-170-31:3197182:3197631 [2] NCCL INFO Channel 10/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-170-31:3197181:3197628 [1] NCCL INFO Channel 17/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-170-31:3197183:3197625 [3] NCCL INFO Channel 08/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-170-31:3197180:3197624 [0] NCCL INFO Channel 11/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-170-31:3197187:3197626 [7] NCCL INFO Channel 23/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-170-31:3197186:3197629 [6] NCCL INFO Channel 20/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-170-31:3197184:3197630 [4] NCCL INFO Channel 08/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-170-31:3197185:3197627 [5] NCCL INFO Channel 06/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-170-31:3197182:3197631 [2] NCCL INFO Channel 11/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-170-31:3197181:3197628 [1] NCCL INFO Channel 18/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-170-31:3197183:3197625 [3] NCCL INFO Channel 09/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-170-31:3197180:3197624 [0] NCCL INFO Channel 12/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-170-31:3197186:3197629 [6] NCCL INFO Channel 21/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-170-31:3197184:3197630 [4] NCCL INFO Channel 09/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-170-31:3197185:3197627 [5] NCCL INFO Channel 07/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-170-31:3197182:3197631 [2] NCCL INFO Channel 12/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-170-31:3197181:3197628 [1] NCCL INFO Channel 19/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-170-31:3197180:3197624 [0] NCCL INFO Channel 13/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-170-31:3197184:3197630 [4] NCCL INFO Channel 10/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-170-31:3197181:3197628 [1] NCCL INFO Channel 20/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-170-31:3197180:3197624 [0] NCCL INFO Channel 14/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-170-31:3197186:3197629 [6] NCCL INFO Channel 22/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-170-31:3197181:3197628 [1] NCCL INFO Channel 21/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-170-31:3197183:3197625 [3] NCCL INFO Channel 10/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-170-31:3197185:3197627 [5] NCCL INFO Channel 08/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-170-31:3197181:3197628 [1] NCCL INFO Channel 22/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-170-31:3197184:3197630 [4] NCCL INFO Channel 11/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-170-31:3197183:3197625 [3] NCCL INFO Channel 11/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-170-31:3197180:3197624 [0] NCCL INFO Channel 15/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-170-31:3197184:3197630 [4] NCCL INFO Channel 12/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-170-31:3197183:3197625 [3] NCCL INFO Channel 12/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-170-31:3197180:3197624 [0] NCCL INFO Channel 16/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-170-31:3197186:3197629 [6] NCCL INFO Channel 23/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-170-31:3197184:3197630 [4] NCCL INFO Channel 13/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-170-31:3197183:3197625 [3] NCCL INFO Channel 13/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-170-31:3197182:3197631 [2] NCCL INFO Channel 13/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-170-31:3197180:3197624 [0] NCCL INFO Channel 17/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-170-31:3197185:3197627 [5] NCCL INFO Channel 09/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-170-31:3197181:3197628 [1] NCCL INFO Channel 23/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-170-31:3197184:3197630 [4] NCCL INFO Channel 14/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-170-31:3197183:3197625 [3] NCCL INFO Channel 14/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-170-31:3197185:3197627 [5] NCCL INFO Channel 10/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-170-31:3197184:3197630 [4] NCCL INFO Channel 15/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-170-31:3197183:3197625 [3] NCCL INFO Channel 15/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-170-132:3330187:3330630 [4] NCCL INFO Channel 00/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-170-132:3330188:3330632 [5] NCCL INFO Channel 00/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-170-31:3197185:3197627 [5] NCCL INFO Channel 11/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-170-132:3330187:3330630 [4] NCCL INFO Channel 01/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-170-132:3330188:3330632 [5] NCCL INFO Channel 01/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-170-31:3197180:3197624 [0] NCCL INFO Channel 18/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-170-31:3197182:3197631 [2] NCCL INFO Channel 14/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-170-31:3197185:3197627 [5] NCCL INFO Channel 12/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-170-31:3197184:3197630 [4] NCCL INFO Channel 16/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-170-132:3330187:3330630 [4] NCCL INFO Channel 02/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-170-132:3330188:3330632 [5] NCCL INFO Channel 02/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-170-31:3197180:3197624 [0] NCCL INFO Channel 19/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-170-132:3330187:3330630 [4] NCCL INFO Channel 03/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-170-132:3330188:3330632 [5] NCCL INFO Channel 03/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-170-31:3197185:3197627 [5] NCCL INFO Channel 13/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-170-31:3197182:3197631 [2] NCCL INFO Channel 15/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-170-132:3330190:3330627 [7] NCCL INFO Channel 00/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330185:3330628 [2] NCCL INFO Channel 00/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-170-31:3197184:3197630 [4] NCCL INFO Channel 17/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-170-31:3197183:3197625 [3] NCCL INFO Channel 16/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-170-132:3330187:3330630 [4] NCCL INFO Channel 04/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-170-132:3330188:3330632 [5] NCCL INFO Channel 04/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-170-31:3197185:3197627 [5] NCCL INFO Channel 14/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-170-31:3197182:3197631 [2] NCCL INFO Channel 16/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-170-31:3197184:3197630 [4] NCCL INFO Channel 18/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-170-132:3330190:3330627 [7] NCCL INFO Channel 01/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-170-31:3197183:3197625 [3] NCCL INFO Channel 17/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-170-132:3330185:3330628 [2] NCCL INFO Channel 01/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-170-31:3197185:3197627 [5] NCCL INFO Channel 15/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-170-132:3330187:3330630 [4] NCCL INFO Channel 05/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-170-31:3197182:3197631 [2] NCCL INFO Channel 17/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-170-132:3330188:3330632 [5] NCCL INFO Channel 05/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-170-132:3330186:3330629 [3] NCCL INFO Channel 00/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-170-132:3330184:3330631 [1] NCCL INFO Channel 00/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-170-31:3197184:3197630 [4] NCCL INFO Channel 19/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-170-31:3197183:3197625 [3] NCCL INFO Channel 18/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-170-31:3197185:3197627 [5] NCCL INFO Channel 16/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-170-31:3197182:3197631 [2] NCCL INFO Channel 18/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-170-132:3330190:3330627 [7] NCCL INFO Channel 02/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330185:3330628 [2] NCCL INFO Channel 02/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-170-132:3330187:3330630 [4] NCCL INFO Channel 06/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-170-31:3197184:3197630 [4] NCCL INFO Channel 20/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-170-132:3330188:3330632 [5] NCCL INFO Channel 06/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-170-132:3330186:3330629 [3] NCCL INFO Channel 01/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-170-31:3197183:3197625 [3] NCCL INFO Channel 19/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-170-132:3330184:3330631 [1] NCCL INFO Channel 01/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-170-31:3197185:3197627 [5] NCCL INFO Channel 17/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-170-31:3197182:3197631 [2] NCCL INFO Channel 19/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-170-31:3197184:3197630 [4] NCCL INFO Channel 21/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-170-31:3197183:3197625 [3] NCCL INFO Channel 20/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-170-132:3330190:3330627 [7] NCCL INFO Channel 03/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330187:3330630 [4] NCCL INFO Channel 07/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-170-132:3330188:3330632 [5] NCCL INFO Channel 07/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-170-132:3330184:3330631 [1] NCCL INFO Channel 02/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-170-132:3330187:3330630 [4] NCCL INFO Channel 08/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-170-132:3330190:3330627 [7] NCCL INFO Channel 04/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330189:3330633 [6] NCCL INFO Channel 00/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-170-132:3330188:3330632 [5] NCCL INFO Channel 08/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-170-132:3330187:3330630 [4] NCCL INFO Channel 09/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-170-132:3330184:3330631 [1] NCCL INFO Channel 03/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-170-132:3330190:3330627 [7] NCCL INFO Channel 05/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330189:3330633 [6] NCCL INFO Channel 01/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-170-31:3197182:3197631 [2] NCCL INFO Channel 20/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-170-132:3330188:3330632 [5] NCCL INFO Channel 09/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-170-31:3197180:3197624 [0] NCCL INFO Channel 20/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-170-31:3197183:3197625 [3] NCCL INFO Channel 21/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-170-132:3330185:3330628 [2] NCCL INFO Channel 03/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-170-31:3197182:3197631 [2] NCCL INFO Channel 21/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-170-31:3197185:3197627 [5] NCCL INFO Channel 18/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-170-132:3330187:3330630 [4] NCCL INFO Channel 10/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-170-132:3330184:3330631 [1] NCCL INFO Channel 04/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-170-132:3330190:3330627 [7] NCCL INFO Channel 06/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330189:3330633 [6] NCCL INFO Channel 02/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-170-31:3197184:3197630 [4] NCCL INFO Channel 22/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-170-132:3330188:3330632 [5] NCCL INFO Channel 10/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-170-132:3330185:3330628 [2] NCCL INFO Channel 04/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-170-132:3330187:3330630 [4] NCCL INFO Channel 11/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-170-132:3330184:3330631 [1] NCCL INFO Channel 05/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-170-132:3330190:3330627 [7] NCCL INFO Channel 07/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330189:3330633 [6] NCCL INFO Channel 03/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-170-31:3197183:3197625 [3] NCCL INFO Channel 22/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-170-31:3197180:3197624 [0] NCCL INFO Channel 21/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-170-31:3197182:3197631 [2] NCCL INFO Channel 22/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-170-31:3197184:3197630 [4] NCCL INFO Channel 23/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-170-31:3197185:3197627 [5] NCCL INFO Channel 19/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-170-132:3330188:3330632 [5] NCCL INFO Channel 11/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-170-132:3330185:3330628 [2] NCCL INFO Channel 05/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-170-31:3197183:3197625 [3] NCCL INFO Channel 23/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-170-132:3330187:3330630 [4] NCCL INFO Channel 12/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-170-31:3197182:3197631 [2] NCCL INFO Channel 23/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-170-31:3197180:3197624 [0] NCCL INFO Channel 22/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-170-132:3330190:3330627 [7] NCCL INFO Channel 08/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-170-31:3197185:3197627 [5] NCCL INFO Channel 20/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-170-132:3330184:3330631 [1] NCCL INFO Channel 06/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-170-132:3330189:3330633 [6] NCCL INFO Channel 04/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-170-31:3197180:3197624 [0] NCCL INFO Channel 23/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-170-132:3330188:3330632 [5] NCCL INFO Channel 12/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-170-31:3197185:3197627 [5] NCCL INFO Channel 21/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-170-132:3330185:3330628 [2] NCCL INFO Channel 06/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-170-132:3330187:3330630 [4] NCCL INFO Channel 13/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-170-132:3330190:3330627 [7] NCCL INFO Channel 09/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330184:3330631 [1] NCCL INFO Channel 07/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-170-132:3330189:3330633 [6] NCCL INFO Channel 05/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-170-132:3330186:3330629 [3] NCCL INFO Channel 02/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-170-132:3330188:3330632 [5] NCCL INFO Channel 13/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-170-132:3330185:3330628 [2] NCCL INFO Channel 07/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-170-132:3330183:3330626 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-170-31:3197185:3197627 [5] NCCL INFO Channel 22/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-162-14:2525081:2525514 [6] NCCL INFO Channel 00/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-170-132:3330187:3330630 [4] NCCL INFO Channel 14/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-170-132:3330190:3330627 [7] NCCL INFO Channel 10/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330184:3330631 [1] NCCL INFO Channel 08/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-170-132:3330189:3330633 [6] NCCL INFO Channel 06/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-170-132:3330186:3330629 [3] NCCL INFO Channel 03/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-162-14:2525079:2525516 [4] NCCL INFO Channel 00/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-162-14:2525081:2525514 [6] NCCL INFO Channel 01/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-170-132:3330188:3330632 [5] NCCL INFO Channel 14/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-170-132:3330185:3330628 [2] NCCL INFO Channel 08/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-170-132:3330183:3330626 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-170-132:3330187:3330630 [4] NCCL INFO Channel 15/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-162-14:2525079:2525516 [4] NCCL INFO Channel 01/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-170-132:3330190:3330627 [7] NCCL INFO Channel 11/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330184:3330631 [1] NCCL INFO Channel 09/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-162-14:2525081:2525514 [6] NCCL INFO Channel 02/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-170-132:3330186:3330629 [3] NCCL INFO Channel 04/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-170-132:3330189:3330633 [6] NCCL INFO Channel 07/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-162-14:2525077:2525512 [2] NCCL INFO Channel 00/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-162-14:2525076:2525513 [1] NCCL INFO Channel 00/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-170-132:3330185:3330628 [2] NCCL INFO Channel 09/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-170-132:3330188:3330632 [5] NCCL INFO Channel 15/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-170-132:3330183:3330626 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-162-14:2525082:2525515 [7] NCCL INFO Channel 00/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330187:3330630 [4] NCCL INFO Channel 16/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-170-132:3330190:3330627 [7] NCCL INFO Channel 12/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-162-14:2525081:2525514 [6] NCCL INFO Channel 03/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-162-14:2525079:2525516 [4] NCCL INFO Channel 02/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-170-132:3330184:3330631 [1] NCCL INFO Channel 10/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-162-14:2525077:2525512 [2] NCCL INFO Channel 01/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-170-132:3330186:3330629 [3] NCCL INFO Channel 05/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-170-132:3330189:3330633 [6] NCCL INFO Channel 08/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-162-14:2525076:2525513 [1] NCCL INFO Channel 01/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-162-14:2525078:2525510 [3] NCCL INFO Channel 00/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-162-14:2525082:2525515 [7] NCCL INFO Channel 01/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330185:3330628 [2] NCCL INFO Channel 10/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-170-132:3330183:3330626 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-170-132:3330188:3330632 [5] NCCL INFO Channel 16/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-162-14:2525081:2525514 [6] NCCL INFO Channel 04/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-162-14:2525079:2525516 [4] NCCL INFO Channel 03/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-170-132:3330190:3330627 [7] NCCL INFO Channel 13/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330187:3330630 [4] NCCL INFO Channel 17/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-162-14:2525077:2525512 [2] NCCL INFO Channel 02/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-170-132:3330184:3330631 [1] NCCL INFO Channel 11/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-170-132:3330186:3330629 [3] NCCL INFO Channel 06/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-170-132:3330189:3330633 [6] NCCL INFO Channel 09/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-162-14:2525078:2525510 [3] NCCL INFO Channel 01/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-162-14:2525076:2525513 [1] NCCL INFO Channel 02/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-162-14:2525082:2525515 [7] NCCL INFO Channel 02/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330185:3330628 [2] NCCL INFO Channel 11/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-170-132:3330183:3330626 [0] NCCL INFO Channel 04/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-170-132:3330188:3330632 [5] NCCL INFO Channel 17/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-162-14:2525081:2525514 [6] NCCL INFO Channel 05/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-162-14:2525079:2525516 [4] NCCL INFO Channel 04/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-170-132:3330186:3330629 [3] NCCL INFO Channel 07/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-162-14:2525077:2525512 [2] NCCL INFO Channel 03/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-170-132:3330190:3330627 [7] NCCL INFO Channel 14/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330187:3330630 [4] NCCL INFO Channel 18/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-170-132:3330184:3330631 [1] NCCL INFO Channel 12/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-170-132:3330189:3330633 [6] NCCL INFO Channel 10/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-170-132:3330185:3330628 [2] NCCL INFO Channel 12/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-162-14:2525078:2525510 [3] NCCL INFO Channel 02/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-162-14:2525076:2525513 [1] NCCL INFO Channel 03/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-162-14:2525082:2525515 [7] NCCL INFO Channel 03/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330183:3330626 [0] NCCL INFO Channel 05/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-170-132:3330188:3330632 [5] NCCL INFO Channel 18/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-162-14:2525081:2525514 [6] NCCL INFO Channel 06/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-162-14:2525079:2525516 [4] NCCL INFO Channel 05/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-162-14:2525077:2525512 [2] NCCL INFO Channel 04/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-162-14:2525076:2525513 [1] NCCL INFO Channel 04/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-162-14:2525078:2525510 [3] NCCL INFO Channel 03/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-162-14:2525082:2525515 [7] NCCL INFO Channel 04/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330186:3330629 [3] NCCL INFO Channel 08/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-170-132:3330190:3330627 [7] NCCL INFO Channel 15/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330187:3330630 [4] NCCL INFO Channel 19/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-170-132:3330184:3330631 [1] NCCL INFO Channel 13/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-170-132:3330189:3330633 [6] NCCL INFO Channel 11/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-170-132:3330185:3330628 [2] NCCL INFO Channel 13/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-170-132:3330183:3330626 [0] NCCL INFO Channel 06/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-162-14:2525079:2525516 [4] NCCL INFO Channel 06/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-162-14:2525081:2525514 [6] NCCL INFO Channel 07/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-162-14:2525077:2525512 [2] NCCL INFO Channel 05/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-162-14:2525075:2525509 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-170-132:3330188:3330632 [5] NCCL INFO Channel 19/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-170-31:3197185:3197627 [5] NCCL INFO Channel 23/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-162-14:2525080:2525511 [5] NCCL INFO Channel 00/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-162-14:2525076:2525513 [1] NCCL INFO Channel 05/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-162-14:2525078:2525510 [3] NCCL INFO Channel 04/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-162-14:2525082:2525515 [7] NCCL INFO Channel 05/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330186:3330629 [3] NCCL INFO Channel 09/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-170-132:3330190:3330627 [7] NCCL INFO Channel 16/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330187:3330630 [4] NCCL INFO Channel 20/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-170-132:3330184:3330631 [1] NCCL INFO Channel 14/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-170-132:3330189:3330633 [6] NCCL INFO Channel 12/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-170-132:3330185:3330628 [2] NCCL INFO Channel 14/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-162-14:2525077:2525512 [2] NCCL INFO Channel 06/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-162-14:2525081:2525514 [6] NCCL INFO Channel 08/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-170-132:3330183:3330626 [0] NCCL INFO Channel 07/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-170-132:3330188:3330632 [5] NCCL INFO Channel 20/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-162-14:2525075:2525509 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-162-14:2525079:2525516 [4] NCCL INFO Channel 07/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-162-14:2525080:2525511 [5] NCCL INFO Channel 01/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-162-14:2525076:2525513 [1] NCCL INFO Channel 06/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-162-14:2525078:2525510 [3] NCCL INFO Channel 05/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-162-14:2525082:2525515 [7] NCCL INFO Channel 06/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330186:3330629 [3] NCCL INFO Channel 10/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-170-132:3330190:3330627 [7] NCCL INFO Channel 17/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330187:3330630 [4] NCCL INFO Channel 21/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-170-132:3330184:3330631 [1] NCCL INFO Channel 15/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-170-132:3330189:3330633 [6] NCCL INFO Channel 13/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-170-132:3330185:3330628 [2] NCCL INFO Channel 15/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-170-132:3330183:3330626 [0] NCCL INFO Channel 08/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-162-14:2525077:2525512 [2] NCCL INFO Channel 07/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-162-14:2525081:2525514 [6] NCCL INFO Channel 09/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-170-132:3330188:3330632 [5] NCCL INFO Channel 21/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-162-14:2525079:2525516 [4] NCCL INFO Channel 08/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-162-14:2525075:2525509 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-162-14:2525080:2525511 [5] NCCL INFO Channel 02/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-162-14:2525078:2525510 [3] NCCL INFO Channel 06/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-162-14:2525076:2525513 [1] NCCL INFO Channel 07/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-162-14:2525082:2525515 [7] NCCL INFO Channel 07/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330186:3330629 [3] NCCL INFO Channel 11/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-170-132:3330190:3330627 [7] NCCL INFO Channel 18/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330184:3330631 [1] NCCL INFO Channel 16/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-170-132:3330187:3330630 [4] NCCL INFO Channel 22/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-170-132:3330189:3330633 [6] NCCL INFO Channel 14/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-170-132:3330185:3330628 [2] NCCL INFO Channel 16/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-170-132:3330183:3330626 [0] NCCL INFO Channel 09/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-170-132:3330188:3330632 [5] NCCL INFO Channel 22/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-162-14:2525077:2525512 [2] NCCL INFO Channel 08/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-162-14:2525079:2525516 [4] NCCL INFO Channel 09/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-162-14:2525075:2525509 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-162-14:2525080:2525511 [5] NCCL INFO Channel 03/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-162-14:2525078:2525510 [3] NCCL INFO Channel 07/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-162-14:2525082:2525515 [7] NCCL INFO Channel 08/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-162-14:2525076:2525513 [1] NCCL INFO Channel 08/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-170-132:3330186:3330629 [3] NCCL INFO Channel 12/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-170-132:3330190:3330627 [7] NCCL INFO Channel 19/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330187:3330630 [4] NCCL INFO Channel 23/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-170-132:3330184:3330631 [1] NCCL INFO Channel 17/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-170-132:3330189:3330633 [6] NCCL INFO Channel 15/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-170-132:3330185:3330628 [2] NCCL INFO Channel 17/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-170-132:3330183:3330626 [0] NCCL INFO Channel 10/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-162-14:2525081:2525514 [6] NCCL INFO Channel 10/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-170-132:3330188:3330632 [5] NCCL INFO Channel 23/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-162-14:2525075:2525509 [0] NCCL INFO Channel 04/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-162-14:2525078:2525510 [3] NCCL INFO Channel 08/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-162-14:2525081:2525514 [6] NCCL INFO Channel 11/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-162-14:2525082:2525515 [7] NCCL INFO Channel 09/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-162-14:2525078:2525510 [3] NCCL INFO Channel 09/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-162-14:2525077:2525512 [2] NCCL INFO Channel 09/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-162-14:2525080:2525511 [5] NCCL INFO Channel 04/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-170-132:3330186:3330629 [3] NCCL INFO Channel 13/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-170-132:3330190:3330627 [7] NCCL INFO Channel 20/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330184:3330631 [1] NCCL INFO Channel 18/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-170-132:3330185:3330628 [2] NCCL INFO Channel 18/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-170-132:3330183:3330626 [0] NCCL INFO Channel 11/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-170-132:3330189:3330633 [6] NCCL INFO Channel 16/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-162-14:2525078:2525510 [3] NCCL INFO Channel 10/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-162-14:2525077:2525512 [2] NCCL INFO Channel 10/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-162-14:2525080:2525511 [5] NCCL INFO Channel 05/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-162-14:2525075:2525509 [0] NCCL INFO Channel 05/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-162-14:2525079:2525516 [4] NCCL INFO Channel 10/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-162-14:2525076:2525513 [1] NCCL INFO Channel 09/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-162-14:2525078:2525510 [3] NCCL INFO Channel 11/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-162-14:2525077:2525512 [2] NCCL INFO Channel 11/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-162-14:2525080:2525511 [5] NCCL INFO Channel 06/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-170-132:3330186:3330629 [3] NCCL INFO Channel 14/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-170-132:3330184:3330631 [1] NCCL INFO Channel 19/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-170-132:3330190:3330627 [7] NCCL INFO Channel 21/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330185:3330628 [2] NCCL INFO Channel 19/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-170-132:3330183:3330626 [0] NCCL INFO Channel 12/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-170-132:3330189:3330633 [6] NCCL INFO Channel 17/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-162-14:2525075:2525509 [0] NCCL INFO Channel 06/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-162-14:2525079:2525516 [4] NCCL INFO Channel 11/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-162-14:2525076:2525513 [1] NCCL INFO Channel 10/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-162-14:2525078:2525510 [3] NCCL INFO Channel 12/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-162-14:2525077:2525512 [2] NCCL INFO Channel 12/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-162-14:2525080:2525511 [5] NCCL INFO Channel 07/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-170-132:3330186:3330629 [3] NCCL INFO Channel 15/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-170-132:3330184:3330631 [1] NCCL INFO Channel 20/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-170-132:3330185:3330628 [2] NCCL INFO Channel 20/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-170-132:3330190:3330627 [7] NCCL INFO Channel 22/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330183:3330626 [0] NCCL INFO Channel 13/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-170-132:3330189:3330633 [6] NCCL INFO Channel 18/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-162-14:2525075:2525509 [0] NCCL INFO Channel 07/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-162-14:2525079:2525516 [4] NCCL INFO Channel 12/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-162-14:2525076:2525513 [1] NCCL INFO Channel 11/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-162-14:2525078:2525510 [3] NCCL INFO Channel 13/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-162-14:2525077:2525512 [2] NCCL INFO Channel 13/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-162-14:2525080:2525511 [5] NCCL INFO Channel 08/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-162-14:2525075:2525509 [0] NCCL INFO Channel 08/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-162-14:2525079:2525516 [4] NCCL INFO Channel 13/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-170-132:3330186:3330629 [3] NCCL INFO Channel 16/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-170-132:3330184:3330631 [1] NCCL INFO Channel 21/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-170-132:3330190:3330627 [7] NCCL INFO Channel 23/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330185:3330628 [2] NCCL INFO Channel 21/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-170-132:3330183:3330626 [0] NCCL INFO Channel 14/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-170-132:3330189:3330633 [6] NCCL INFO Channel 19/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-162-14:2525076:2525513 [1] NCCL INFO Channel 12/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-162-14:2525078:2525510 [3] NCCL INFO Channel 14/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-162-14:2525077:2525512 [2] NCCL INFO Channel 14/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-162-14:2525080:2525511 [5] NCCL INFO Channel 09/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-162-14:2525075:2525509 [0] NCCL INFO Channel 09/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-162-14:2525079:2525516 [4] NCCL INFO Channel 14/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-170-132:3330186:3330629 [3] NCCL INFO Channel 17/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-170-132:3330184:3330631 [1] NCCL INFO Channel 22/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-170-132:3330185:3330628 [2] NCCL INFO Channel 22/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-170-132:3330183:3330626 [0] NCCL INFO Channel 15/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-170-132:3330189:3330633 [6] NCCL INFO Channel 20/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-162-14:2525076:2525513 [1] NCCL INFO Channel 13/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-162-14:2525082:2525515 [7] NCCL INFO Channel 10/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-162-14:2525078:2525510 [3] NCCL INFO Channel 15/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-162-14:2525081:2525514 [6] NCCL INFO Channel 12/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-162-14:2525077:2525512 [2] NCCL INFO Channel 15/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-162-14:2525080:2525511 [5] NCCL INFO Channel 10/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-170-132:3330184:3330631 [1] NCCL INFO Channel 23/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-170-132:3330185:3330628 [2] NCCL INFO Channel 23/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-170-132:3330186:3330629 [3] NCCL INFO Channel 18/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-170-132:3330183:3330626 [0] NCCL INFO Channel 16/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-170-132:3330189:3330633 [6] NCCL INFO Channel 21/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-162-14:2525075:2525509 [0] NCCL INFO Channel 10/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-162-14:2525079:2525516 [4] NCCL INFO Channel 15/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-162-14:2525076:2525513 [1] NCCL INFO Channel 14/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-170-132:3330186:3330629 [3] NCCL INFO Channel 19/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-170-132:3330183:3330626 [0] NCCL INFO Channel 17/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-170-132:3330189:3330633 [6] NCCL INFO Channel 22/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-162-14:2525082:2525515 [7] NCCL INFO Channel 11/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-162-14:2525077:2525512 [2] NCCL INFO Channel 16/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-162-14:2525078:2525510 [3] NCCL INFO Channel 16/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-162-14:2525081:2525514 [6] NCCL INFO Channel 13/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-162-14:2525080:2525511 [5] NCCL INFO Channel 11/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-170-132:3330186:3330629 [3] NCCL INFO Channel 20/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-170-132:3330183:3330626 [0] NCCL INFO Channel 18/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-162-14:2525075:2525509 [0] NCCL INFO Channel 11/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-170-132:3330189:3330633 [6] NCCL INFO Channel 23/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-162-14:2525079:2525516 [4] NCCL INFO Channel 16/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-170-132:3330183:3330626 [0] NCCL INFO Channel 19/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-170-132:3330186:3330629 [3] NCCL INFO Channel 21/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-162-14:2525076:2525513 [1] NCCL INFO Channel 15/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-162-14:2525081:2525514 [6] NCCL INFO Channel 14/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-162-14:2525082:2525515 [7] NCCL INFO Channel 12/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-162-14:2525077:2525512 [2] NCCL INFO Channel 17/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-162-14:2525078:2525510 [3] NCCL INFO Channel 17/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-162-14:2525080:2525511 [5] NCCL INFO Channel 12/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-162-14:2525075:2525509 [0] NCCL INFO Channel 12/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-162-14:2525079:2525516 [4] NCCL INFO Channel 17/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-162-14:2525076:2525513 [1] NCCL INFO Channel 16/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-170-132:3330183:3330626 [0] NCCL INFO Channel 20/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-162-14:2525081:2525514 [6] NCCL INFO Channel 15/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-162-14:2525082:2525515 [7] NCCL INFO Channel 13/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-162-14:2525077:2525512 [2] NCCL INFO Channel 18/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-162-14:2525078:2525510 [3] NCCL INFO Channel 18/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-162-14:2525080:2525511 [5] NCCL INFO Channel 13/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-162-14:2525075:2525509 [0] NCCL INFO Channel 13/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-162-14:2525079:2525516 [4] NCCL INFO Channel 18/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-162-14:2525076:2525513 [1] NCCL INFO Channel 17/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-170-132:3330186:3330629 [3] NCCL INFO Channel 22/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-162-14:2525081:2525514 [6] NCCL INFO Channel 16/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-162-14:2525082:2525515 [7] NCCL INFO Channel 14/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-162-14:2525078:2525510 [3] NCCL INFO Channel 19/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-162-14:2525080:2525511 [5] NCCL INFO Channel 14/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-162-14:2525077:2525512 [2] NCCL INFO Channel 19/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-162-14:2525079:2525516 [4] NCCL INFO Channel 19/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-162-14:2525075:2525509 [0] NCCL INFO Channel 14/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-162-14:2525076:2525513 [1] NCCL INFO Channel 18/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-162-14:2525081:2525514 [6] NCCL INFO Channel 17/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-162-14:2525082:2525515 [7] NCCL INFO Channel 15/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-162-14:2525078:2525510 [3] NCCL INFO Channel 20/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-162-14:2525080:2525511 [5] NCCL INFO Channel 15/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-162-14:2525079:2525516 [4] NCCL INFO Channel 20/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-162-14:2525077:2525512 [2] NCCL INFO Channel 20/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-162-14:2525075:2525509 [0] NCCL INFO Channel 15/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-162-14:2525076:2525513 [1] NCCL INFO Channel 19/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-170-132:3330183:3330626 [0] NCCL INFO Channel 21/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-162-14:2525081:2525514 [6] NCCL INFO Channel 18/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-162-14:2525082:2525515 [7] NCCL INFO Channel 16/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-162-14:2525080:2525511 [5] NCCL INFO Channel 16/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-162-14:2525079:2525516 [4] NCCL INFO Channel 21/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-162-14:2525078:2525510 [3] NCCL INFO Channel 21/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-162-14:2525077:2525512 [2] NCCL INFO Channel 21/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-162-14:2525075:2525509 [0] NCCL INFO Channel 16/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-170-132:3330186:3330629 [3] NCCL INFO Channel 23/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-162-14:2525076:2525513 [1] NCCL INFO Channel 20/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-162-14:2525081:2525514 [6] NCCL INFO Channel 19/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-162-14:2525082:2525515 [7] NCCL INFO Channel 17/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-162-14:2525080:2525511 [5] NCCL INFO Channel 17/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-162-14:2525079:2525516 [4] NCCL INFO Channel 22/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-162-14:2525078:2525510 [3] NCCL INFO Channel 22/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-162-14:2525077:2525512 [2] NCCL INFO Channel 22/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-162-14:2525075:2525509 [0] NCCL INFO Channel 17/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-162-14:2525076:2525513 [1] NCCL INFO Channel 21/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-162-14:2525081:2525514 [6] NCCL INFO Channel 20/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-162-14:2525082:2525515 [7] NCCL INFO Channel 18/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-162-14:2525080:2525511 [5] NCCL INFO Channel 18/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-162-14:2525079:2525516 [4] NCCL INFO Channel 23/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-162-14:2525078:2525510 [3] NCCL INFO Channel 23/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-162-14:2525077:2525512 [2] NCCL INFO Channel 23/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-162-14:2525075:2525509 [0] NCCL INFO Channel 18/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-162-14:2525076:2525513 [1] NCCL INFO Channel 22/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-162-14:2525081:2525514 [6] NCCL INFO Channel 21/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-162-14:2525082:2525515 [7] NCCL INFO Channel 19/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-162-14:2525080:2525511 [5] NCCL INFO Channel 19/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-162-14:2525075:2525509 [0] NCCL INFO Channel 19/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-162-14:2525076:2525513 [1] NCCL INFO Channel 23/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-162-14:2525081:2525514 [6] NCCL INFO Channel 22/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-162-14:2525082:2525515 [7] NCCL INFO Channel 20/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-162-14:2525080:2525511 [5] NCCL INFO Channel 20/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-162-14:2525075:2525509 [0] NCCL INFO Channel 20/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-162-14:2525081:2525514 [6] NCCL INFO Channel 23/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-162-14:2525082:2525515 [7] NCCL INFO Channel 21/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330183:3330626 [0] NCCL INFO Channel 22/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-162-14:2525080:2525511 [5] NCCL INFO Channel 21/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-162-14:2525075:2525509 [0] NCCL INFO Channel 21/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-170-132:3330183:3330626 [0] NCCL INFO Channel 23/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-162-14:2525082:2525515 [7] NCCL INFO Channel 22/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-162-14:2525080:2525511 [5] NCCL INFO Channel 22/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-162-14:2525075:2525509 [0] NCCL INFO Channel 22/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-162-14:2525082:2525515 [7] NCCL INFO Channel 23/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-162-14:2525080:2525511 [5] NCCL INFO Channel 23/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-162-14:2525075:2525509 [0] NCCL INFO Channel 23/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-163-58:934727:935152 [7] NCCL INFO Connected all rings
+ip-26-0-163-58:934727:935152 [7] NCCL INFO Channel 00/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-163-58:934727:935152 [7] NCCL INFO Channel 01/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-163-58:934727:935152 [7] NCCL INFO Channel 02/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-163-58:934721:935153 [1] NCCL INFO Connected all rings
+ip-26-0-163-58:934723:935156 [3] NCCL INFO Connected all rings
+ip-26-0-163-58:934727:935152 [7] NCCL INFO Channel 03/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-163-58:934727:935152 [7] NCCL INFO Channel 04/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-163-58:934720:935150 [0] NCCL INFO Connected all rings
+ip-26-0-163-58:934727:935152 [7] NCCL INFO Channel 05/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-163-58:934727:935152 [7] NCCL INFO Channel 06/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-163-58:934727:935152 [7] NCCL INFO Channel 07/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-163-58:934727:935152 [7] NCCL INFO Channel 08/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-163-58:934727:935152 [7] NCCL INFO Channel 09/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-163-58:934722:935155 [2] NCCL INFO Connected all rings
+ip-26-0-163-58:934726:935151 [6] NCCL INFO Connected all rings
+ip-26-0-163-58:934727:935152 [7] NCCL INFO Channel 10/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-163-58:934725:935157 [5] NCCL INFO Connected all rings
+ip-26-0-163-58:934724:935154 [4] NCCL INFO Connected all rings
+ip-26-0-163-58:934727:935152 [7] NCCL INFO Channel 11/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-163-58:934727:935152 [7] NCCL INFO Channel 12/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-163-58:934727:935152 [7] NCCL INFO Channel 13/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-163-58:934727:935152 [7] NCCL INFO Channel 14/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-163-58:934727:935152 [7] NCCL INFO Channel 15/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-163-58:934727:935152 [7] NCCL INFO Channel 16/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-163-58:934727:935152 [7] NCCL INFO Channel 17/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-163-58:934727:935152 [7] NCCL INFO Channel 18/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-163-58:934727:935152 [7] NCCL INFO Channel 19/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-163-58:934727:935152 [7] NCCL INFO Channel 20/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-163-58:934727:935152 [7] NCCL INFO Channel 21/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-163-58:934727:935152 [7] NCCL INFO Channel 22/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-163-58:934727:935152 [7] NCCL INFO Channel 23/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-163-58:934721:935153 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-163-58:934721:935153 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-163-58:934721:935153 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-163-58:934721:935153 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-163-58:934721:935153 [1] NCCL INFO Channel 04/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-163-58:934722:935155 [2] NCCL INFO Channel 00/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-163-58:934726:935151 [6] NCCL INFO Channel 00/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-163-58:934721:935153 [1] NCCL INFO Channel 05/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-163-58:934725:935157 [5] NCCL INFO Channel 00/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-163-58:934722:935155 [2] NCCL INFO Channel 01/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-163-58:934726:935151 [6] NCCL INFO Channel 01/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-163-58:934724:935154 [4] NCCL INFO Channel 00/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-163-58:934721:935153 [1] NCCL INFO Channel 06/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-163-58:934725:935157 [5] NCCL INFO Channel 01/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-163-58:934722:935155 [2] NCCL INFO Channel 02/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-163-58:934726:935151 [6] NCCL INFO Channel 02/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-163-58:934724:935154 [4] NCCL INFO Channel 01/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-163-58:934721:935153 [1] NCCL INFO Channel 07/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-163-58:934725:935157 [5] NCCL INFO Channel 02/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-163-58:934722:935155 [2] NCCL INFO Channel 03/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-163-58:934726:935151 [6] NCCL INFO Channel 03/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-163-58:934724:935154 [4] NCCL INFO Channel 02/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-163-58:934721:935153 [1] NCCL INFO Channel 08/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-163-58:934725:935157 [5] NCCL INFO Channel 03/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-163-58:934722:935155 [2] NCCL INFO Channel 04/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-163-58:934726:935151 [6] NCCL INFO Channel 04/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-163-58:934724:935154 [4] NCCL INFO Channel 03/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-163-58:934723:935156 [3] NCCL INFO Channel 00/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-163-58:934721:935153 [1] NCCL INFO Channel 09/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-163-58:934722:935155 [2] NCCL INFO Channel 05/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-163-58:934725:935157 [5] NCCL INFO Channel 04/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-163-58:934726:935151 [6] NCCL INFO Channel 05/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-163-58:934724:935154 [4] NCCL INFO Channel 04/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-163-58:934723:935156 [3] NCCL INFO Channel 01/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-163-58:934721:935153 [1] NCCL INFO Channel 10/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-163-58:934722:935155 [2] NCCL INFO Channel 06/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-163-58:934725:935157 [5] NCCL INFO Channel 05/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-163-58:934726:935151 [6] NCCL INFO Channel 06/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-163-58:934724:935154 [4] NCCL INFO Channel 05/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-162-14:2525078:2525510 [3] NCCL INFO Connected all rings
+ip-26-0-163-58:934721:935153 [1] NCCL INFO Channel 11/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-163-58:934722:935155 [2] NCCL INFO Channel 07/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-163-58:934723:935156 [3] NCCL INFO Channel 02/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-163-58:934725:935157 [5] NCCL INFO Channel 06/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-163-58:934722:935155 [2] NCCL INFO Channel 08/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-163-58:934721:935153 [1] NCCL INFO Channel 12/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-163-58:934723:935156 [3] NCCL INFO Channel 03/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-163-58:934726:935151 [6] NCCL INFO Channel 07/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-163-58:934724:935154 [4] NCCL INFO Channel 06/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-163-58:934725:935157 [5] NCCL INFO Channel 07/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-163-58:934723:935156 [3] NCCL INFO Channel 04/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-163-58:934726:935151 [6] NCCL INFO Channel 08/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-163-58:934724:935154 [4] NCCL INFO Channel 07/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-163-58:934725:935157 [5] NCCL INFO Channel 08/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-163-58:934722:935155 [2] NCCL INFO Channel 09/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-163-58:934721:935153 [1] NCCL INFO Channel 13/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-163-58:934723:935156 [3] NCCL INFO Channel 05/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-163-58:934726:935151 [6] NCCL INFO Channel 09/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-163-58:934724:935154 [4] NCCL INFO Channel 08/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-163-58:934725:935157 [5] NCCL INFO Channel 09/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-163-58:934721:935153 [1] NCCL INFO Channel 14/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-163-58:934722:935155 [2] NCCL INFO Channel 10/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-163-58:934723:935156 [3] NCCL INFO Channel 06/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-163-58:934725:935157 [5] NCCL INFO Channel 10/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-163-58:934722:935155 [2] NCCL INFO Channel 11/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-163-58:934726:935151 [6] NCCL INFO Channel 10/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-163-58:934723:935156 [3] NCCL INFO Channel 07/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-163-58:934725:935157 [5] NCCL INFO Channel 11/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-163-58:934722:935155 [2] NCCL INFO Channel 12/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-163-58:934726:935151 [6] NCCL INFO Channel 11/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-163-58:934723:935156 [3] NCCL INFO Channel 08/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-163-58:934725:935157 [5] NCCL INFO Channel 12/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-163-58:934722:935155 [2] NCCL INFO Channel 13/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-163-58:934726:935151 [6] NCCL INFO Channel 12/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-163-58:934723:935156 [3] NCCL INFO Channel 09/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-163-58:934725:935157 [5] NCCL INFO Channel 13/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-163-58:934722:935155 [2] NCCL INFO Channel 14/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-163-58:934726:935151 [6] NCCL INFO Channel 13/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-163-58:934723:935156 [3] NCCL INFO Channel 10/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-163-58:934724:935154 [4] NCCL INFO Channel 09/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-163-58:934725:935157 [5] NCCL INFO Channel 14/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-163-58:934721:935153 [1] NCCL INFO Channel 15/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-163-58:934722:935155 [2] NCCL INFO Channel 15/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-163-58:934726:935151 [6] NCCL INFO Channel 14/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-163-58:934724:935154 [4] NCCL INFO Channel 10/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-163-58:934723:935156 [3] NCCL INFO Channel 11/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-163-58:934726:935151 [6] NCCL INFO Channel 15/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-163-58:934721:935153 [1] NCCL INFO Channel 16/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-163-58:934723:935156 [3] NCCL INFO Channel 12/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-163-58:934724:935154 [4] NCCL INFO Channel 11/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-163-58:934726:935151 [6] NCCL INFO Channel 16/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-163-58:934721:935153 [1] NCCL INFO Channel 17/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-163-58:934723:935156 [3] NCCL INFO Channel 13/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-163-58:934724:935154 [4] NCCL INFO Channel 12/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-163-58:934721:935153 [1] NCCL INFO Channel 18/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-163-58:934726:935151 [6] NCCL INFO Channel 17/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-163-58:934723:935156 [3] NCCL INFO Channel 14/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-163-58:934724:935154 [4] NCCL INFO Channel 13/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-163-58:934721:935153 [1] NCCL INFO Channel 19/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-163-58:934726:935151 [6] NCCL INFO Channel 18/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-163-58:934723:935156 [3] NCCL INFO Channel 15/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-163-58:934724:935154 [4] NCCL INFO Channel 14/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-163-58:934721:935153 [1] NCCL INFO Channel 20/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-163-58:934726:935151 [6] NCCL INFO Channel 19/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-163-58:934723:935156 [3] NCCL INFO Channel 16/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-163-58:934724:935154 [4] NCCL INFO Channel 15/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-163-58:934721:935153 [1] NCCL INFO Channel 21/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-163-58:934726:935151 [6] NCCL INFO Channel 20/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-163-58:934723:935156 [3] NCCL INFO Channel 17/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-163-58:934724:935154 [4] NCCL INFO Channel 16/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-163-58:934721:935153 [1] NCCL INFO Channel 22/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-163-58:934726:935151 [6] NCCL INFO Channel 21/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-163-58:934725:935157 [5] NCCL INFO Channel 15/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-163-58:934723:935156 [3] NCCL INFO Channel 18/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-163-58:934724:935154 [4] NCCL INFO Channel 17/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-163-58:934722:935155 [2] NCCL INFO Channel 16/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-163-58:934726:935151 [6] NCCL INFO Channel 22/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-163-58:934721:935153 [1] NCCL INFO Channel 23/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-163-58:934725:935157 [5] NCCL INFO Channel 16/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-163-58:934723:935156 [3] NCCL INFO Channel 19/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-163-58:934724:935154 [4] NCCL INFO Channel 18/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-163-58:934722:935155 [2] NCCL INFO Channel 17/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-163-58:934726:935151 [6] NCCL INFO Channel 23/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-163-58:934725:935157 [5] NCCL INFO Channel 17/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-163-58:934723:935156 [3] NCCL INFO Channel 20/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-163-58:934724:935154 [4] NCCL INFO Channel 19/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-163-58:934722:935155 [2] NCCL INFO Channel 18/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-170-31:3197187:3197626 [7] NCCL INFO Connected all rings
+ip-26-0-170-31:3197187:3197626 [7] NCCL INFO Channel 00/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-163-58:934725:935157 [5] NCCL INFO Channel 18/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-163-58:934724:935154 [4] NCCL INFO Channel 20/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-163-58:934725:935157 [5] NCCL INFO Channel 19/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-163-58:934724:935154 [4] NCCL INFO Channel 21/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-163-58:934725:935157 [5] NCCL INFO Channel 20/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-163-58:934724:935154 [4] NCCL INFO Channel 22/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-163-58:934723:935156 [3] NCCL INFO Channel 21/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-163-58:934722:935155 [2] NCCL INFO Channel 19/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-163-58:934725:935157 [5] NCCL INFO Channel 21/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-163-58:934724:935154 [4] NCCL INFO Channel 23/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-163-58:934723:935156 [3] NCCL INFO Channel 22/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-163-58:934722:935155 [2] NCCL INFO Channel 20/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-170-31:3197187:3197626 [7] NCCL INFO Channel 01/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-163-58:934725:935157 [5] NCCL INFO Channel 22/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-170-31:3197180:3197624 [0] NCCL INFO Connected all rings
+ip-26-0-163-58:934723:935156 [3] NCCL INFO Channel 23/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-163-58:934722:935155 [2] NCCL INFO Channel 21/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-163-58:934725:935157 [5] NCCL INFO Channel 23/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-170-132:3330189:3330633 [6] NCCL INFO Connected all rings
+ip-26-0-163-58:934722:935155 [2] NCCL INFO Channel 22/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-170-31:3197187:3197626 [7] NCCL INFO Channel 02/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-170-132:3330188:3330632 [5] NCCL INFO Connected all rings
+ip-26-0-163-58:934722:935155 [2] NCCL INFO Channel 23/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-170-31:3197187:3197626 [7] NCCL INFO Channel 03/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-162-14:2525077:2525512 [2] NCCL INFO Connected all rings
+ip-26-0-170-31:3197183:3197625 [3] NCCL INFO Connected all rings
+ip-26-0-170-31:3197187:3197626 [7] NCCL INFO Channel 04/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-170-31:3197184:3197630 [4] NCCL INFO Connected all rings
+ip-26-0-170-31:3197187:3197626 [7] NCCL INFO Channel 05/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-170-31:3197187:3197626 [7] NCCL INFO Channel 06/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-170-31:3197182:3197631 [2] NCCL INFO Connected all rings
+ip-26-0-170-31:3197181:3197628 [1] NCCL INFO Connected all rings
+ip-26-0-170-31:3197187:3197626 [7] NCCL INFO Channel 07/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-170-31:3197187:3197626 [7] NCCL INFO Channel 08/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-170-31:3197187:3197626 [7] NCCL INFO Channel 09/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-170-31:3197186:3197629 [6] NCCL INFO Connected all rings
+ip-26-0-170-31:3197185:3197627 [5] NCCL INFO Connected all rings
+ip-26-0-170-31:3197187:3197626 [7] NCCL INFO Channel 10/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-170-132:3330183:3330626 [0] NCCL INFO Connected all rings
+ip-26-0-170-31:3197187:3197626 [7] NCCL INFO Channel 11/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-170-31:3197187:3197626 [7] NCCL INFO Channel 12/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-170-31:3197187:3197626 [7] NCCL INFO Channel 13/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-162-14:2525075:2525509 [0] NCCL INFO Connected all rings
+ip-26-0-170-31:3197187:3197626 [7] NCCL INFO Channel 14/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-170-31:3197187:3197626 [7] NCCL INFO Channel 15/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-170-31:3197187:3197626 [7] NCCL INFO Channel 16/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-170-31:3197187:3197626 [7] NCCL INFO Channel 17/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-170-31:3197187:3197626 [7] NCCL INFO Channel 18/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-170-31:3197187:3197626 [7] NCCL INFO Channel 19/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-170-132:3330185:3330628 [2] NCCL INFO Connected all rings
+ip-26-0-162-14:2525076:2525513 [1] NCCL INFO Connected all rings
+ip-26-0-170-31:3197187:3197626 [7] NCCL INFO Channel 20/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-170-31:3197187:3197626 [7] NCCL INFO Channel 21/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-170-31:3197187:3197626 [7] NCCL INFO Channel 22/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-170-31:3197187:3197626 [7] NCCL INFO Channel 23/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-162-14:2525082:2525515 [7] NCCL INFO Connected all rings
+ip-26-0-162-14:2525082:2525515 [7] NCCL INFO Channel 00/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-170-31:3197183:3197625 [3] NCCL INFO Channel 00/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-170-31:3197183:3197625 [3] NCCL INFO Channel 01/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-170-31:3197184:3197630 [4] NCCL INFO Channel 00/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-170-31:3197183:3197625 [3] NCCL INFO Channel 02/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-170-31:3197184:3197630 [4] NCCL INFO Channel 01/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-170-31:3197183:3197625 [3] NCCL INFO Channel 03/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-170-132:3330190:3330627 [7] NCCL INFO Connected all rings
+ip-26-0-170-132:3330190:3330627 [7] NCCL INFO Channel 00/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-162-14:2525079:2525516 [4] NCCL INFO Connected all rings
+ip-26-0-170-31:3197184:3197630 [4] NCCL INFO Channel 02/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-170-31:3197183:3197625 [3] NCCL INFO Channel 04/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-170-132:3330184:3330631 [1] NCCL INFO Connected all rings
+ip-26-0-170-31:3197184:3197630 [4] NCCL INFO Channel 03/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-162-14:2525082:2525515 [7] NCCL INFO Channel 01/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-170-31:3197183:3197625 [3] NCCL INFO Channel 05/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-170-31:3197182:3197631 [2] NCCL INFO Channel 00/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-170-132:3330187:3330630 [4] NCCL INFO Connected all rings
+ip-26-0-170-31:3197186:3197629 [6] NCCL INFO Channel 00/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-170-31:3197184:3197630 [4] NCCL INFO Channel 04/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-170-31:3197181:3197628 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-170-31:3197185:3197627 [5] NCCL INFO Channel 00/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-170-31:3197183:3197625 [3] NCCL INFO Channel 06/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-170-31:3197182:3197631 [2] NCCL INFO Channel 01/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-170-31:3197186:3197629 [6] NCCL INFO Channel 01/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-170-31:3197184:3197630 [4] NCCL INFO Channel 05/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-170-31:3197181:3197628 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-170-31:3197185:3197627 [5] NCCL INFO Channel 01/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-170-31:3197183:3197625 [3] NCCL INFO Channel 07/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-170-31:3197182:3197631 [2] NCCL INFO Channel 02/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-170-31:3197186:3197629 [6] NCCL INFO Channel 02/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-170-31:3197184:3197630 [4] NCCL INFO Channel 06/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-170-31:3197181:3197628 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-170-31:3197185:3197627 [5] NCCL INFO Channel 02/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-170-31:3197183:3197625 [3] NCCL INFO Channel 08/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-170-31:3197182:3197631 [2] NCCL INFO Channel 03/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-170-31:3197186:3197629 [6] NCCL INFO Channel 03/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-170-31:3197184:3197630 [4] NCCL INFO Channel 07/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-170-31:3197181:3197628 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-162-14:2525080:2525511 [5] NCCL INFO Connected all rings
+ip-26-0-162-14:2525082:2525515 [7] NCCL INFO Channel 02/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-170-31:3197185:3197627 [5] NCCL INFO Channel 03/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-170-31:3197183:3197625 [3] NCCL INFO Channel 09/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-170-31:3197182:3197631 [2] NCCL INFO Channel 04/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-170-31:3197186:3197629 [6] NCCL INFO Channel 04/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-170-31:3197184:3197630 [4] NCCL INFO Channel 08/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-170-132:3330186:3330629 [3] NCCL INFO Connected all rings
+ip-26-0-170-31:3197181:3197628 [1] NCCL INFO Channel 04/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-162-14:2525081:2525514 [6] NCCL INFO Connected all rings
+ip-26-0-170-31:3197185:3197627 [5] NCCL INFO Channel 04/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-170-31:3197183:3197625 [3] NCCL INFO Channel 10/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-170-31:3197182:3197631 [2] NCCL INFO Channel 05/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-170-31:3197186:3197629 [6] NCCL INFO Channel 05/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-170-31:3197184:3197630 [4] NCCL INFO Channel 09/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-170-31:3197181:3197628 [1] NCCL INFO Channel 05/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-162-14:2525082:2525515 [7] NCCL INFO Channel 03/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-170-132:3330190:3330627 [7] NCCL INFO Channel 01/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-170-31:3197185:3197627 [5] NCCL INFO Channel 05/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-170-31:3197183:3197625 [3] NCCL INFO Channel 11/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-170-31:3197186:3197629 [6] NCCL INFO Channel 06/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-170-31:3197182:3197631 [2] NCCL INFO Channel 06/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-170-31:3197184:3197630 [4] NCCL INFO Channel 10/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-170-31:3197181:3197628 [1] NCCL INFO Channel 06/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330190:3330627 [7] NCCL INFO Channel 02/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-162-14:2525082:2525515 [7] NCCL INFO Channel 04/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-170-31:3197185:3197627 [5] NCCL INFO Channel 06/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-170-31:3197183:3197625 [3] NCCL INFO Channel 12/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-170-132:3330190:3330627 [7] NCCL INFO Channel 03/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-170-31:3197186:3197629 [6] NCCL INFO Channel 07/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-170-31:3197182:3197631 [2] NCCL INFO Channel 07/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-170-31:3197184:3197630 [4] NCCL INFO Channel 11/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-170-31:3197181:3197628 [1] NCCL INFO Channel 07/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330190:3330627 [7] NCCL INFO Channel 04/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-170-31:3197185:3197627 [5] NCCL INFO Channel 07/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-162-14:2525082:2525515 [7] NCCL INFO Channel 05/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-170-31:3197183:3197625 [3] NCCL INFO Channel 13/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-170-31:3197186:3197629 [6] NCCL INFO Channel 08/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-170-31:3197182:3197631 [2] NCCL INFO Channel 08/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-170-31:3197184:3197630 [4] NCCL INFO Channel 12/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-170-31:3197181:3197628 [1] NCCL INFO Channel 08/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330190:3330627 [7] NCCL INFO Channel 05/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-170-31:3197185:3197627 [5] NCCL INFO Channel 08/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-170-31:3197183:3197625 [3] NCCL INFO Channel 14/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-162-14:2525082:2525515 [7] NCCL INFO Channel 06/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-170-31:3197186:3197629 [6] NCCL INFO Channel 09/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-170-31:3197182:3197631 [2] NCCL INFO Channel 09/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-170-31:3197184:3197630 [4] NCCL INFO Channel 13/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-170-31:3197181:3197628 [1] NCCL INFO Channel 09/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330190:3330627 [7] NCCL INFO Channel 06/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-170-31:3197185:3197627 [5] NCCL INFO Channel 09/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-170-31:3197183:3197625 [3] NCCL INFO Channel 15/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-170-31:3197186:3197629 [6] NCCL INFO Channel 10/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-170-31:3197182:3197631 [2] NCCL INFO Channel 10/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-170-31:3197184:3197630 [4] NCCL INFO Channel 14/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-170-31:3197181:3197628 [1] NCCL INFO Channel 10/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-162-14:2525082:2525515 [7] NCCL INFO Channel 07/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-170-31:3197185:3197627 [5] NCCL INFO Channel 10/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-170-31:3197183:3197625 [3] NCCL INFO Channel 16/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-170-31:3197186:3197629 [6] NCCL INFO Channel 11/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-170-31:3197182:3197631 [2] NCCL INFO Channel 11/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-170-31:3197181:3197628 [1] NCCL INFO Channel 11/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-170-31:3197184:3197630 [4] NCCL INFO Channel 15/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-162-14:2525082:2525515 [7] NCCL INFO Channel 08/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-170-31:3197185:3197627 [5] NCCL INFO Channel 11/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-170-31:3197183:3197625 [3] NCCL INFO Channel 17/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-170-31:3197186:3197629 [6] NCCL INFO Channel 12/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-170-31:3197182:3197631 [2] NCCL INFO Channel 12/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-170-31:3197181:3197628 [1] NCCL INFO Channel 12/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-170-31:3197184:3197630 [4] NCCL INFO Channel 16/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-162-14:2525082:2525515 [7] NCCL INFO Channel 09/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-170-31:3197185:3197627 [5] NCCL INFO Channel 12/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-170-31:3197183:3197625 [3] NCCL INFO Channel 18/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-170-31:3197186:3197629 [6] NCCL INFO Channel 13/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-170-31:3197182:3197631 [2] NCCL INFO Channel 13/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-170-31:3197181:3197628 [1] NCCL INFO Channel 13/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-170-31:3197184:3197630 [4] NCCL INFO Channel 17/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-162-14:2525082:2525515 [7] NCCL INFO Channel 10/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-170-31:3197185:3197627 [5] NCCL INFO Channel 13/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-170-31:3197183:3197625 [3] NCCL INFO Channel 19/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-170-31:3197186:3197629 [6] NCCL INFO Channel 14/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-170-31:3197182:3197631 [2] NCCL INFO Channel 14/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-170-31:3197181:3197628 [1] NCCL INFO Channel 14/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-170-31:3197184:3197630 [4] NCCL INFO Channel 18/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-162-14:2525082:2525515 [7] NCCL INFO Channel 11/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-170-31:3197185:3197627 [5] NCCL INFO Channel 14/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-170-132:3330190:3330627 [7] NCCL INFO Channel 07/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-170-31:3197183:3197625 [3] NCCL INFO Channel 20/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-170-31:3197186:3197629 [6] NCCL INFO Channel 15/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-170-31:3197182:3197631 [2] NCCL INFO Channel 15/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-170-31:3197181:3197628 [1] NCCL INFO Channel 15/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-170-31:3197184:3197630 [4] NCCL INFO Channel 19/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-170-132:3330190:3330627 [7] NCCL INFO Channel 08/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-170-31:3197185:3197627 [5] NCCL INFO Channel 15/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-162-14:2525082:2525515 [7] NCCL INFO Channel 12/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-170-31:3197183:3197625 [3] NCCL INFO Channel 21/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-170-31:3197186:3197629 [6] NCCL INFO Channel 16/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-170-31:3197182:3197631 [2] NCCL INFO Channel 16/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-170-31:3197181:3197628 [1] NCCL INFO Channel 16/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-170-31:3197184:3197630 [4] NCCL INFO Channel 20/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-170-132:3330190:3330627 [7] NCCL INFO Channel 09/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-162-14:2525082:2525515 [7] NCCL INFO Channel 13/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-170-31:3197185:3197627 [5] NCCL INFO Channel 16/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-170-31:3197183:3197625 [3] NCCL INFO Channel 22/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-170-31:3197186:3197629 [6] NCCL INFO Channel 17/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-170-132:3330190:3330627 [7] NCCL INFO Channel 10/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-170-31:3197182:3197631 [2] NCCL INFO Channel 17/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-170-31:3197184:3197630 [4] NCCL INFO Channel 21/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-170-31:3197181:3197628 [1] NCCL INFO Channel 17/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-170-31:3197185:3197627 [5] NCCL INFO Channel 17/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-162-14:2525082:2525515 [7] NCCL INFO Channel 14/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-170-132:3330190:3330627 [7] NCCL INFO Channel 11/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-170-31:3197183:3197625 [3] NCCL INFO Channel 23/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-170-31:3197186:3197629 [6] NCCL INFO Channel 18/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-170-31:3197181:3197628 [1] NCCL INFO Channel 18/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-170-31:3197182:3197631 [2] NCCL INFO Channel 18/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-170-31:3197184:3197630 [4] NCCL INFO Channel 22/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-170-31:3197185:3197627 [5] NCCL INFO Channel 18/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-162-14:2525082:2525515 [7] NCCL INFO Channel 15/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-170-132:3330190:3330627 [7] NCCL INFO Channel 12/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-170-31:3197181:3197628 [1] NCCL INFO Channel 19/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-170-31:3197186:3197629 [6] NCCL INFO Channel 19/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-170-31:3197184:3197630 [4] NCCL INFO Channel 23/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-170-31:3197182:3197631 [2] NCCL INFO Channel 19/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-170-31:3197185:3197627 [5] NCCL INFO Channel 19/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-162-14:2525078:2525510 [3] NCCL INFO Channel 00/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-162-14:2525082:2525515 [7] NCCL INFO Channel 16/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-170-31:3197186:3197629 [6] NCCL INFO Channel 20/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-170-31:3197181:3197628 [1] NCCL INFO Channel 20/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-170-31:3197182:3197631 [2] NCCL INFO Channel 20/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-170-31:3197185:3197627 [5] NCCL INFO Channel 20/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-170-132:3330190:3330627 [7] NCCL INFO Channel 13/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-162-14:2525078:2525510 [3] NCCL INFO Channel 01/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-170-31:3197186:3197629 [6] NCCL INFO Channel 21/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-170-31:3197181:3197628 [1] NCCL INFO Channel 21/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-170-31:3197182:3197631 [2] NCCL INFO Channel 21/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-170-31:3197185:3197627 [5] NCCL INFO Channel 21/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-162-14:2525082:2525515 [7] NCCL INFO Channel 17/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-170-132:3330190:3330627 [7] NCCL INFO Channel 14/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-170-31:3197186:3197629 [6] NCCL INFO Channel 22/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-170-31:3197181:3197628 [1] NCCL INFO Channel 22/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-170-31:3197182:3197631 [2] NCCL INFO Channel 22/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-170-31:3197185:3197627 [5] NCCL INFO Channel 22/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-162-14:2525078:2525510 [3] NCCL INFO Channel 02/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-162-14:2525082:2525515 [7] NCCL INFO Channel 18/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-170-31:3197186:3197629 [6] NCCL INFO Channel 23/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-170-31:3197181:3197628 [1] NCCL INFO Channel 23/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-170-31:3197182:3197631 [2] NCCL INFO Channel 23/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-170-31:3197185:3197627 [5] NCCL INFO Channel 23/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-170-132:3330190:3330627 [7] NCCL INFO Channel 15/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-162-14:2525078:2525510 [3] NCCL INFO Channel 03/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-162-14:2525082:2525515 [7] NCCL INFO Channel 19/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-170-132:3330190:3330627 [7] NCCL INFO Channel 16/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-170-132:3330190:3330627 [7] NCCL INFO Channel 17/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-170-132:3330185:3330628 [2] NCCL INFO Channel 00/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-170-132:3330184:3330631 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330190:3330627 [7] NCCL INFO Channel 18/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-170-132:3330188:3330632 [5] NCCL INFO Channel 00/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-170-132:3330185:3330628 [2] NCCL INFO Channel 01/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-162-14:2525076:2525513 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330184:3330631 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-162-14:2525076:2525513 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-162-14:2525077:2525512 [2] NCCL INFO Channel 00/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-170-132:3330190:3330627 [7] NCCL INFO Channel 19/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-170-132:3330185:3330628 [2] NCCL INFO Channel 02/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-170-132:3330188:3330632 [5] NCCL INFO Channel 01/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-170-132:3330184:3330631 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-162-14:2525076:2525513 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-162-14:2525077:2525512 [2] NCCL INFO Channel 01/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-162-14:2525078:2525510 [3] NCCL INFO Channel 04/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-162-14:2525079:2525516 [4] NCCL INFO Channel 00/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-162-14:2525080:2525511 [5] NCCL INFO Channel 00/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-162-14:2525082:2525515 [7] NCCL INFO Channel 20/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-170-132:3330185:3330628 [2] NCCL INFO Channel 03/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-170-132:3330190:3330627 [7] NCCL INFO Channel 20/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-170-132:3330188:3330632 [5] NCCL INFO Channel 02/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-170-132:3330184:3330631 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-162-14:2525076:2525513 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-162-14:2525077:2525512 [2] NCCL INFO Channel 02/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-162-14:2525078:2525510 [3] NCCL INFO Channel 05/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-162-14:2525079:2525516 [4] NCCL INFO Channel 01/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-162-14:2525080:2525511 [5] NCCL INFO Channel 01/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-162-14:2525081:2525514 [6] NCCL INFO Channel 00/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-162-14:2525082:2525515 [7] NCCL INFO Channel 21/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-170-132:3330185:3330628 [2] NCCL INFO Channel 04/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-170-132:3330190:3330627 [7] NCCL INFO Channel 21/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-170-132:3330188:3330632 [5] NCCL INFO Channel 03/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-170-132:3330184:3330631 [1] NCCL INFO Channel 04/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-162-14:2525077:2525512 [2] NCCL INFO Channel 03/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-162-14:2525076:2525513 [1] NCCL INFO Channel 04/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-162-14:2525078:2525510 [3] NCCL INFO Channel 06/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-170-132:3330185:3330628 [2] NCCL INFO Channel 05/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-162-14:2525080:2525511 [5] NCCL INFO Channel 02/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-162-14:2525079:2525516 [4] NCCL INFO Channel 02/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-162-14:2525081:2525514 [6] NCCL INFO Channel 01/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-170-132:3330190:3330627 [7] NCCL INFO Channel 22/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-170-132:3330188:3330632 [5] NCCL INFO Channel 04/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-162-14:2525082:2525515 [7] NCCL INFO Channel 22/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-170-132:3330184:3330631 [1] NCCL INFO Channel 05/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330187:3330630 [4] NCCL INFO Channel 00/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-162-14:2525077:2525512 [2] NCCL INFO Channel 04/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-162-14:2525076:2525513 [1] NCCL INFO Channel 05/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-162-14:2525078:2525510 [3] NCCL INFO Channel 07/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-170-132:3330185:3330628 [2] NCCL INFO Channel 06/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-162-14:2525080:2525511 [5] NCCL INFO Channel 03/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-162-14:2525079:2525516 [4] NCCL INFO Channel 03/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-170-132:3330190:3330627 [7] NCCL INFO Channel 23/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-170-132:3330188:3330632 [5] NCCL INFO Channel 05/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-162-14:2525081:2525514 [6] NCCL INFO Channel 02/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-162-14:2525082:2525515 [7] NCCL INFO Channel 23/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-170-132:3330184:3330631 [1] NCCL INFO Channel 06/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330187:3330630 [4] NCCL INFO Channel 01/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-162-14:2525077:2525512 [2] NCCL INFO Channel 05/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-162-14:2525076:2525513 [1] NCCL INFO Channel 06/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-162-14:2525078:2525510 [3] NCCL INFO Channel 08/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-170-132:3330185:3330628 [2] NCCL INFO Channel 07/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-162-14:2525079:2525516 [4] NCCL INFO Channel 04/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-162-14:2525080:2525511 [5] NCCL INFO Channel 04/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-170-132:3330188:3330632 [5] NCCL INFO Channel 06/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-162-14:2525081:2525514 [6] NCCL INFO Channel 03/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-170-132:3330184:3330631 [1] NCCL INFO Channel 07/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330189:3330633 [6] NCCL INFO Channel 00/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-170-132:3330187:3330630 [4] NCCL INFO Channel 02/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-162-14:2525077:2525512 [2] NCCL INFO Channel 06/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-162-14:2525076:2525513 [1] NCCL INFO Channel 07/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-162-14:2525078:2525510 [3] NCCL INFO Channel 09/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-162-14:2525079:2525516 [4] NCCL INFO Channel 05/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-162-14:2525080:2525511 [5] NCCL INFO Channel 05/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-162-14:2525081:2525514 [6] NCCL INFO Channel 04/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-170-132:3330185:3330628 [2] NCCL INFO Channel 08/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-170-132:3330188:3330632 [5] NCCL INFO Channel 07/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-170-132:3330184:3330631 [1] NCCL INFO Channel 08/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330189:3330633 [6] NCCL INFO Channel 01/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-170-132:3330187:3330630 [4] NCCL INFO Channel 03/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-162-14:2525077:2525512 [2] NCCL INFO Channel 07/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-162-14:2525076:2525513 [1] NCCL INFO Channel 08/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-162-14:2525078:2525510 [3] NCCL INFO Channel 10/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-162-14:2525079:2525516 [4] NCCL INFO Channel 06/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-162-14:2525080:2525511 [5] NCCL INFO Channel 06/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-162-14:2525081:2525514 [6] NCCL INFO Channel 05/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-170-132:3330185:3330628 [2] NCCL INFO Channel 09/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-170-132:3330188:3330632 [5] NCCL INFO Channel 08/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-162-14:2525077:2525512 [2] NCCL INFO Channel 08/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-170-132:3330184:3330631 [1] NCCL INFO Channel 09/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330187:3330630 [4] NCCL INFO Channel 04/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-162-14:2525076:2525513 [1] NCCL INFO Channel 09/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-162-14:2525078:2525510 [3] NCCL INFO Channel 11/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-170-132:3330189:3330633 [6] NCCL INFO Channel 02/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-162-14:2525079:2525516 [4] NCCL INFO Channel 07/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-162-14:2525080:2525511 [5] NCCL INFO Channel 07/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-162-14:2525081:2525514 [6] NCCL INFO Channel 06/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-170-132:3330185:3330628 [2] NCCL INFO Channel 10/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-170-132:3330188:3330632 [5] NCCL INFO Channel 09/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-162-14:2525077:2525512 [2] NCCL INFO Channel 09/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-162-14:2525076:2525513 [1] NCCL INFO Channel 10/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-162-14:2525078:2525510 [3] NCCL INFO Channel 12/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-162-14:2525079:2525516 [4] NCCL INFO Channel 08/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-162-14:2525080:2525511 [5] NCCL INFO Channel 08/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-170-132:3330184:3330631 [1] NCCL INFO Channel 10/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-162-14:2525081:2525514 [6] NCCL INFO Channel 07/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-170-132:3330187:3330630 [4] NCCL INFO Channel 05/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-170-132:3330189:3330633 [6] NCCL INFO Channel 03/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-170-132:3330185:3330628 [2] NCCL INFO Channel 11/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-170-132:3330188:3330632 [5] NCCL INFO Channel 10/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-162-14:2525077:2525512 [2] NCCL INFO Channel 10/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-162-14:2525076:2525513 [1] NCCL INFO Channel 11/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-162-14:2525078:2525510 [3] NCCL INFO Channel 13/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-162-14:2525079:2525516 [4] NCCL INFO Channel 09/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-162-14:2525080:2525511 [5] NCCL INFO Channel 09/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-162-14:2525081:2525514 [6] NCCL INFO Channel 08/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-170-132:3330184:3330631 [1] NCCL INFO Channel 11/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330187:3330630 [4] NCCL INFO Channel 06/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-170-132:3330189:3330633 [6] NCCL INFO Channel 04/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-170-132:3330185:3330628 [2] NCCL INFO Channel 12/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-170-132:3330188:3330632 [5] NCCL INFO Channel 11/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-162-14:2525076:2525513 [1] NCCL INFO Channel 12/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-162-14:2525077:2525512 [2] NCCL INFO Channel 11/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-162-14:2525079:2525516 [4] NCCL INFO Channel 10/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-162-14:2525078:2525510 [3] NCCL INFO Channel 14/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-162-14:2525080:2525511 [5] NCCL INFO Channel 10/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-162-14:2525081:2525514 [6] NCCL INFO Channel 09/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-170-132:3330187:3330630 [4] NCCL INFO Channel 07/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-170-132:3330184:3330631 [1] NCCL INFO Channel 12/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330185:3330628 [2] NCCL INFO Channel 13/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-170-132:3330188:3330632 [5] NCCL INFO Channel 12/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-170-132:3330187:3330630 [4] NCCL INFO Channel 08/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-162-14:2525076:2525513 [1] NCCL INFO Channel 13/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-162-14:2525077:2525512 [2] NCCL INFO Channel 12/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-162-14:2525079:2525516 [4] NCCL INFO Channel 11/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-162-14:2525080:2525511 [5] NCCL INFO Channel 11/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-170-132:3330184:3330631 [1] NCCL INFO Channel 13/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-162-14:2525078:2525510 [3] NCCL INFO Channel 15/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-162-14:2525081:2525514 [6] NCCL INFO Channel 10/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-170-132:3330185:3330628 [2] NCCL INFO Channel 14/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-170-132:3330188:3330632 [5] NCCL INFO Channel 13/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-170-132:3330187:3330630 [4] NCCL INFO Channel 09/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-170-132:3330184:3330631 [1] NCCL INFO Channel 14/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-162-14:2525076:2525513 [1] NCCL INFO Channel 14/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-162-14:2525080:2525511 [5] NCCL INFO Channel 12/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-162-14:2525077:2525512 [2] NCCL INFO Channel 13/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-162-14:2525078:2525510 [3] NCCL INFO Channel 16/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-162-14:2525079:2525516 [4] NCCL INFO Channel 12/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-162-14:2525081:2525514 [6] NCCL INFO Channel 11/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-170-132:3330185:3330628 [2] NCCL INFO Channel 15/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-170-132:3330188:3330632 [5] NCCL INFO Channel 14/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-170-132:3330187:3330630 [4] NCCL INFO Channel 10/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-170-132:3330184:3330631 [1] NCCL INFO Channel 15/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330186:3330629 [3] NCCL INFO Channel 00/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-162-14:2525077:2525512 [2] NCCL INFO Channel 14/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-162-14:2525078:2525510 [3] NCCL INFO Channel 17/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-162-14:2525081:2525514 [6] NCCL INFO Channel 12/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-170-132:3330185:3330628 [2] NCCL INFO Channel 16/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-170-132:3330188:3330632 [5] NCCL INFO Channel 15/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-162-14:2525076:2525513 [1] NCCL INFO Channel 15/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-162-14:2525080:2525511 [5] NCCL INFO Channel 13/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-170-132:3330187:3330630 [4] NCCL INFO Channel 11/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-170-132:3330184:3330631 [1] NCCL INFO Channel 16/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330186:3330629 [3] NCCL INFO Channel 01/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-162-14:2525078:2525510 [3] NCCL INFO Channel 18/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-162-14:2525077:2525512 [2] NCCL INFO Channel 15/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-170-132:3330185:3330628 [2] NCCL INFO Channel 17/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-162-14:2525081:2525514 [6] NCCL INFO Channel 13/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-170-132:3330188:3330632 [5] NCCL INFO Channel 16/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-163-58:934720:935150 [0] NCCL INFO Connected all trees
+ip-26-0-162-14:2525076:2525513 [1] NCCL INFO Channel 16/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-162-14:2525080:2525511 [5] NCCL INFO Channel 14/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-170-132:3330187:3330630 [4] NCCL INFO Channel 12/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-170-132:3330184:3330631 [1] NCCL INFO Channel 17/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-163-58:934720:935150 [0] NCCL INFO NVLS comm 0x9818c80 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-170-132:3330186:3330629 [3] NCCL INFO Channel 02/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-162-14:2525078:2525510 [3] NCCL INFO Channel 19/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-162-14:2525077:2525512 [2] NCCL INFO Channel 16/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-170-132:3330185:3330628 [2] NCCL INFO Channel 18/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-170-132:3330188:3330632 [5] NCCL INFO Channel 17/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-162-14:2525081:2525514 [6] NCCL INFO Channel 14/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-162-14:2525076:2525513 [1] NCCL INFO Channel 17/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-162-14:2525080:2525511 [5] NCCL INFO Channel 15/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-170-132:3330187:3330630 [4] NCCL INFO Channel 13/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-170-132:3330184:3330631 [1] NCCL INFO Channel 18/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330186:3330629 [3] NCCL INFO Channel 03/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-162-14:2525078:2525510 [3] NCCL INFO Channel 20/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-162-14:2525077:2525512 [2] NCCL INFO Channel 17/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-162-14:2525081:2525514 [6] NCCL INFO Channel 15/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-170-132:3330185:3330628 [2] NCCL INFO Channel 19/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-170-132:3330188:3330632 [5] NCCL INFO Channel 18/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-162-14:2525076:2525513 [1] NCCL INFO Channel 18/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-162-14:2525080:2525511 [5] NCCL INFO Channel 16/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-170-132:3330187:3330630 [4] NCCL INFO Channel 14/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-170-132:3330184:3330631 [1] NCCL INFO Channel 19/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330186:3330629 [3] NCCL INFO Channel 04/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-162-14:2525078:2525510 [3] NCCL INFO Channel 21/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-170-132:3330185:3330628 [2] NCCL INFO Channel 20/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-162-14:2525081:2525514 [6] NCCL INFO Channel 16/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-162-14:2525077:2525512 [2] NCCL INFO Channel 18/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-162-14:2525076:2525513 [1] NCCL INFO Channel 19/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-162-14:2525080:2525511 [5] NCCL INFO Channel 17/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-170-132:3330188:3330632 [5] NCCL INFO Channel 19/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-162-14:2525079:2525516 [4] NCCL INFO Channel 13/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-170-132:3330187:3330630 [4] NCCL INFO Channel 15/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-170-132:3330184:3330631 [1] NCCL INFO Channel 20/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330186:3330629 [3] NCCL INFO Channel 05/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-162-14:2525078:2525510 [3] NCCL INFO Channel 22/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-170-132:3330185:3330628 [2] NCCL INFO Channel 21/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-170-132:3330188:3330632 [5] NCCL INFO Channel 20/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-162-14:2525081:2525514 [6] NCCL INFO Channel 17/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-162-14:2525077:2525512 [2] NCCL INFO Channel 19/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-162-14:2525076:2525513 [1] NCCL INFO Channel 20/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330189:3330633 [6] NCCL INFO Channel 05/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-162-14:2525080:2525511 [5] NCCL INFO Channel 18/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-162-14:2525079:2525516 [4] NCCL INFO Channel 14/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-170-132:3330184:3330631 [1] NCCL INFO Channel 21/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330187:3330630 [4] NCCL INFO Channel 16/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-170-132:3330186:3330629 [3] NCCL INFO Channel 06/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-170-132:3330185:3330628 [2] NCCL INFO Channel 22/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-170-132:3330188:3330632 [5] NCCL INFO Channel 21/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-170-132:3330189:3330633 [6] NCCL INFO Channel 06/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-170-132:3330184:3330631 [1] NCCL INFO Channel 22/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330187:3330630 [4] NCCL INFO Channel 17/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-170-132:3330186:3330629 [3] NCCL INFO Channel 07/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-162-14:2525081:2525514 [6] NCCL INFO Channel 18/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-162-14:2525077:2525512 [2] NCCL INFO Channel 20/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-162-14:2525078:2525510 [3] NCCL INFO Channel 23/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-170-132:3330185:3330628 [2] NCCL INFO Channel 23/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-170-132:3330188:3330632 [5] NCCL INFO Channel 22/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-162-14:2525076:2525513 [1] NCCL INFO Channel 21/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330189:3330633 [6] NCCL INFO Channel 07/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-162-14:2525079:2525516 [4] NCCL INFO Channel 15/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-170-132:3330184:3330631 [1] NCCL INFO Channel 23/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330187:3330630 [4] NCCL INFO Channel 18/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-170-132:3330186:3330629 [3] NCCL INFO Channel 08/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-170-132:3330188:3330632 [5] NCCL INFO Channel 23/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-170-132:3330189:3330633 [6] NCCL INFO Channel 08/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-170-132:3330187:3330630 [4] NCCL INFO Channel 19/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-170-132:3330186:3330629 [3] NCCL INFO Channel 09/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-162-14:2525077:2525512 [2] NCCL INFO Channel 21/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-162-14:2525080:2525511 [5] NCCL INFO Channel 19/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-162-14:2525076:2525513 [1] NCCL INFO Channel 22/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330189:3330633 [6] NCCL INFO Channel 09/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-162-14:2525079:2525516 [4] NCCL INFO Channel 16/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-162-14:2525077:2525512 [2] NCCL INFO Channel 22/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-162-14:2525080:2525511 [5] NCCL INFO Channel 20/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-162-14:2525081:2525514 [6] NCCL INFO Channel 19/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-170-132:3330187:3330630 [4] NCCL INFO Channel 20/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-170-132:3330186:3330629 [3] NCCL INFO Channel 10/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-162-14:2525079:2525516 [4] NCCL INFO Channel 17/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-162-14:2525076:2525513 [1] NCCL INFO Channel 23/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-162-14:2525080:2525511 [5] NCCL INFO Channel 21/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-162-14:2525081:2525514 [6] NCCL INFO Channel 20/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-162-14:2525079:2525516 [4] NCCL INFO Channel 18/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-162-14:2525080:2525511 [5] NCCL INFO Channel 22/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-162-14:2525077:2525512 [2] NCCL INFO Channel 23/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-170-132:3330189:3330633 [6] NCCL INFO Channel 10/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-162-14:2525081:2525514 [6] NCCL INFO Channel 21/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-162-14:2525079:2525516 [4] NCCL INFO Channel 19/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-162-14:2525080:2525511 [5] NCCL INFO Channel 23/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-170-132:3330187:3330630 [4] NCCL INFO Channel 21/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-170-132:3330186:3330629 [3] NCCL INFO Channel 11/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-162-14:2525081:2525514 [6] NCCL INFO Channel 22/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-162-14:2525079:2525516 [4] NCCL INFO Channel 20/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-162-14:2525081:2525514 [6] NCCL INFO Channel 23/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-170-132:3330189:3330633 [6] NCCL INFO Channel 11/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-170-132:3330187:3330630 [4] NCCL INFO Channel 22/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-170-132:3330186:3330629 [3] NCCL INFO Channel 12/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-162-14:2525079:2525516 [4] NCCL INFO Channel 21/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-170-132:3330189:3330633 [6] NCCL INFO Channel 12/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-170-132:3330187:3330630 [4] NCCL INFO Channel 23/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-170-132:3330186:3330629 [3] NCCL INFO Channel 13/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-163-58:934727:935152 [7] NCCL INFO Connected all trees
+ip-26-0-163-58:934727:935152 [7] NCCL INFO NVLS comm 0x9a33d80 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-170-132:3330189:3330633 [6] NCCL INFO Channel 13/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-170-132:3330186:3330629 [3] NCCL INFO Channel 14/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-170-132:3330189:3330633 [6] NCCL INFO Channel 14/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-170-132:3330186:3330629 [3] NCCL INFO Channel 15/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-162-14:2525079:2525516 [4] NCCL INFO Channel 22/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-170-132:3330189:3330633 [6] NCCL INFO Channel 15/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-170-132:3330186:3330629 [3] NCCL INFO Channel 16/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-170-132:3330189:3330633 [6] NCCL INFO Channel 16/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-170-132:3330186:3330629 [3] NCCL INFO Channel 17/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-162-14:2525079:2525516 [4] NCCL INFO Channel 23/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-170-132:3330186:3330629 [3] NCCL INFO Channel 18/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-170-132:3330189:3330633 [6] NCCL INFO Channel 17/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-170-132:3330186:3330629 [3] NCCL INFO Channel 19/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-170-132:3330186:3330629 [3] NCCL INFO Channel 20/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-170-132:3330189:3330633 [6] NCCL INFO Channel 18/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-170-132:3330186:3330629 [3] NCCL INFO Channel 21/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-170-132:3330183:3330626 [0] NCCL INFO Connected all trees
+ip-26-0-170-132:3330183:3330626 [0] NCCL INFO NVLS comm 0x94ede40 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-170-132:3330189:3330633 [6] NCCL INFO Channel 19/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-170-132:3330186:3330629 [3] NCCL INFO Channel 22/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-170-132:3330189:3330633 [6] NCCL INFO Channel 20/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-170-132:3330186:3330629 [3] NCCL INFO Channel 23/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-170-132:3330189:3330633 [6] NCCL INFO Channel 21/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-170-132:3330189:3330633 [6] NCCL INFO Channel 22/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-170-132:3330189:3330633 [6] NCCL INFO Channel 23/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-170-31:3197187:3197626 [7] NCCL INFO Connected all trees
+ip-26-0-170-31:3197187:3197626 [7] NCCL INFO NVLS comm 0xa6793c0 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-170-132:3330184:3330631 [1] NCCL INFO Connected all trees
+ip-26-0-170-132:3330184:3330631 [1] NCCL INFO NVLS comm 0x9625de0 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-170-31:3197180:3197624 [0] NCCL INFO Connected all trees
+ip-26-0-170-31:3197180:3197624 [0] NCCL INFO NVLS comm 0xa1b4140 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-162-14:2525075:2525509 [0] NCCL INFO Connected all trees
+ip-26-0-162-14:2525075:2525509 [0] NCCL INFO NVLS comm 0x988a6e0 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-163-58:934722:935155 [2] NCCL INFO Connected all trees
+ip-26-0-163-58:934722:935155 [2] NCCL INFO NVLS comm 0x8adcf20 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-162-14:2525082:2525515 [7] NCCL INFO Connected all trees
+ip-26-0-162-14:2525082:2525515 [7] NCCL INFO NVLS comm 0x9e8b8b0 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-163-58:934723:935156 [3] NCCL INFO Connected all trees
+ip-26-0-163-58:934723:935156 [3] NCCL INFO NVLS comm 0x9adc790 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-163-58:934724:935154 [4] NCCL INFO Connected all trees
+ip-26-0-163-58:934724:935154 [4] NCCL INFO NVLS comm 0x98d2f70 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-163-58:934725:935157 [5] NCCL INFO Connected all trees
+ip-26-0-163-58:934725:935157 [5] NCCL INFO NVLS comm 0x8abff80 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-170-132:3330190:3330627 [7] NCCL INFO Connected all trees
+ip-26-0-170-132:3330190:3330627 [7] NCCL INFO NVLS comm 0xa535420 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-163-58:934721:935153 [1] NCCL INFO Connected all trees
+ip-26-0-163-58:934721:935153 [1] NCCL INFO NVLS comm 0xa55f070 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-163-58:934726:935151 [6] NCCL INFO Connected all trees
+ip-26-0-163-58:934726:935151 [6] NCCL INFO NVLS comm 0x9e24e90 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-170-31:3197185:3197627 [5] NCCL INFO Connected all trees
+ip-26-0-170-31:3197185:3197627 [5] NCCL INFO NVLS comm 0x8864ad0 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-163-58:934726:935151 [6] NCCL INFO Connected NVLS tree
+ip-26-0-163-58:934726:935151 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-163-58:934726:935151 [6] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-163-58:934726:935151 [6] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-163-58:934720:935150 [0] NCCL INFO Connected NVLS tree
+ip-26-0-163-58:934720:935150 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-163-58:934724:935154 [4] NCCL INFO Connected NVLS tree
+ip-26-0-163-58:934724:935154 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-163-58:934722:935155 [2] NCCL INFO Connected NVLS tree
+ip-26-0-163-58:934722:935155 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-163-58:934720:935150 [0] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-163-58:934720:935150 [0] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-163-58:934722:935155 [2] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-163-58:934724:935154 [4] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-163-58:934722:935155 [2] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-163-58:934724:935154 [4] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-163-58:934727:935152 [7] NCCL INFO Connected NVLS tree
+ip-26-0-163-58:934727:935152 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-163-58:934727:935152 [7] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-163-58:934727:935152 [7] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-163-58:934725:935157 [5] NCCL INFO Connected NVLS tree
+ip-26-0-163-58:934725:935157 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-163-58:934725:935157 [5] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-163-58:934725:935157 [5] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-163-58:934723:935156 [3] NCCL INFO Connected NVLS tree
+ip-26-0-163-58:934723:935156 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-163-58:934723:935156 [3] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-163-58:934723:935156 [3] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-163-58:934721:935153 [1] NCCL INFO Connected NVLS tree
+ip-26-0-163-58:934721:935153 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-163-58:934721:935153 [1] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-163-58:934721:935153 [1] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-170-31:3197182:3197631 [2] NCCL INFO Connected all trees
+ip-26-0-170-31:3197182:3197631 [2] NCCL INFO NVLS comm 0x89157c0 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-162-14:2525076:2525513 [1] NCCL INFO Connected all trees
+ip-26-0-162-14:2525076:2525513 [1] NCCL INFO NVLS comm 0xa386440 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-170-31:3197184:3197630 [4] NCCL INFO Connected all trees
+ip-26-0-170-31:3197184:3197630 [4] NCCL INFO NVLS comm 0xa117e20 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-163-58:934722:935155 [2] NCCL INFO comm 0x8adcf20 rank 2 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0x37fac877b5d17063 - Init COMPLETE
+ip-26-0-163-58:934720:935150 [0] NCCL INFO comm 0x9818c80 rank 0 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0x37fac877b5d17063 - Init COMPLETE
+ip-26-0-163-58:934726:935151 [6] NCCL INFO comm 0x9e24e90 rank 6 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0x37fac877b5d17063 - Init COMPLETE
+ip-26-0-163-58:934724:935154 [4] NCCL INFO comm 0x98d2f70 rank 4 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0x37fac877b5d17063 - Init COMPLETE
+ip-26-0-163-58:934721:935153 [1] NCCL INFO comm 0xa55f070 rank 1 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0x37fac877b5d17063 - Init COMPLETE
+ip-26-0-163-58:934725:935157 [5] NCCL INFO comm 0x8abff80 rank 5 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0x37fac877b5d17063 - Init COMPLETE
+ip-26-0-163-58:934723:935156 [3] NCCL INFO comm 0x9adc790 rank 3 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0x37fac877b5d17063 - Init COMPLETE
+ip-26-0-163-58:934727:935152 [7] NCCL INFO comm 0x9a33d80 rank 7 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0x37fac877b5d17063 - Init COMPLETE
+ip-26-0-170-31:3197183:3197625 [3] NCCL INFO Connected all trees
+ip-26-0-170-31:3197183:3197625 [3] NCCL INFO NVLS comm 0x9f84250 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-170-31:3197186:3197629 [6] NCCL INFO Connected all trees
+ip-26-0-170-31:3197186:3197629 [6] NCCL INFO NVLS comm 0xa04bb00 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-170-31:3197181:3197628 [1] NCCL INFO Connected all trees
+ip-26-0-170-31:3197181:3197628 [1] NCCL INFO NVLS comm 0x9be0100 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-162-14:2525081:2525514 [6] NCCL INFO Connected all trees
+ip-26-0-162-14:2525081:2525514 [6] NCCL INFO NVLS comm 0x950b8e0 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-170-132:3330186:3330629 [3] NCCL INFO Connected all trees
+ip-26-0-170-132:3330186:3330629 [3] NCCL INFO NVLS comm 0x8f2f440 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-162-14:2525077:2525512 [2] NCCL INFO Connected all trees
+ip-26-0-162-14:2525077:2525512 [2] NCCL INFO NVLS comm 0xa1cf600 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-170-132:3330189:3330633 [6] NCCL INFO Connected all trees
+ip-26-0-162-14:2525080:2525511 [5] NCCL INFO Connected all trees
+ip-26-0-170-132:3330189:3330633 [6] NCCL INFO NVLS comm 0x87c6270 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-162-14:2525080:2525511 [5] NCCL INFO NVLS comm 0x9c5e6a0 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-170-132:3330185:3330628 [2] NCCL INFO Connected all trees
+ip-26-0-170-132:3330185:3330628 [2] NCCL INFO NVLS comm 0x9ab5de0 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-162-14:2525079:2525516 [4] NCCL INFO Connected all trees
+ip-26-0-162-14:2525079:2525516 [4] NCCL INFO NVLS comm 0x96da6c0 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-170-132:3330187:3330630 [4] NCCL INFO Connected all trees
+ip-26-0-170-132:3330187:3330630 [4] NCCL INFO NVLS comm 0x9a52690 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-170-132:3330188:3330632 [5] NCCL INFO Connected all trees
+ip-26-0-170-132:3330188:3330632 [5] NCCL INFO NVLS comm 0xa5074f0 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-162-14:2525078:2525510 [3] NCCL INFO Connected all trees
+ip-26-0-162-14:2525078:2525510 [3] NCCL INFO NVLS comm 0x923c7b0 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-170-31:3197180:3197624 [0] NCCL INFO Connected NVLS tree
+ip-26-0-170-31:3197180:3197624 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-170-31:3197184:3197630 [4] NCCL INFO Connected NVLS tree
+ip-26-0-170-31:3197184:3197630 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-170-31:3197184:3197630 [4] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-170-31:3197184:3197630 [4] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-170-31:3197180:3197624 [0] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-170-31:3197180:3197624 [0] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-170-31:3197182:3197631 [2] NCCL INFO Connected NVLS tree
+ip-26-0-170-31:3197182:3197631 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-170-31:3197182:3197631 [2] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-170-31:3197182:3197631 [2] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-170-31:3197186:3197629 [6] NCCL INFO Connected NVLS tree
+ip-26-0-170-31:3197186:3197629 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-170-31:3197186:3197629 [6] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-170-31:3197186:3197629 [6] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-170-31:3197185:3197627 [5] NCCL INFO Connected NVLS tree
+ip-26-0-170-31:3197185:3197627 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-170-31:3197185:3197627 [5] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-170-31:3197185:3197627 [5] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-170-31:3197181:3197628 [1] NCCL INFO Connected NVLS tree
+ip-26-0-170-31:3197181:3197628 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-170-31:3197181:3197628 [1] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-170-31:3197181:3197628 [1] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-170-31:3197183:3197625 [3] NCCL INFO Connected NVLS tree
+ip-26-0-170-31:3197183:3197625 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-170-31:3197183:3197625 [3] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-170-31:3197183:3197625 [3] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-170-31:3197187:3197626 [7] NCCL INFO Connected NVLS tree
+ip-26-0-170-31:3197187:3197626 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-170-31:3197187:3197626 [7] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-170-31:3197187:3197626 [7] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-170-31:3197180:3197624 [0] NCCL INFO comm 0xa1b4140 rank 0 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0x41c9367c841b7051 - Init COMPLETE
+ip-26-0-170-31:3197183:3197625 [3] NCCL INFO comm 0x9f84250 rank 3 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0x41c9367c841b7051 - Init COMPLETE
+ip-26-0-170-31:3197187:3197626 [7] NCCL INFO comm 0xa6793c0 rank 7 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0x41c9367c841b7051 - Init COMPLETE
+ip-26-0-170-31:3197184:3197630 [4] NCCL INFO comm 0xa117e20 rank 4 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0x41c9367c841b7051 - Init COMPLETE
+ip-26-0-170-31:3197185:3197627 [5] NCCL INFO comm 0x8864ad0 rank 5 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0x41c9367c841b7051 - Init COMPLETE
+ip-26-0-170-31:3197186:3197629 [6] NCCL INFO comm 0xa04bb00 rank 6 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0x41c9367c841b7051 - Init COMPLETE
+ip-26-0-170-31:3197182:3197631 [2] NCCL INFO comm 0x89157c0 rank 2 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0x41c9367c841b7051 - Init COMPLETE
+ip-26-0-170-31:3197181:3197628 [1] NCCL INFO comm 0x9be0100 rank 1 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0x41c9367c841b7051 - Init COMPLETE
+ip-26-0-162-14:2525075:2525509 [0] NCCL INFO Connected NVLS tree
+ip-26-0-162-14:2525075:2525509 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-162-14:2525081:2525514 [6] NCCL INFO Connected NVLS tree
+ip-26-0-162-14:2525075:2525509 [0] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-162-14:2525075:2525509 [0] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-162-14:2525081:2525514 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-162-14:2525081:2525514 [6] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-162-14:2525081:2525514 [6] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-162-14:2525079:2525516 [4] NCCL INFO Connected NVLS tree
+ip-26-0-162-14:2525079:2525516 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-162-14:2525077:2525512 [2] NCCL INFO Connected NVLS tree
+ip-26-0-162-14:2525079:2525516 [4] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-162-14:2525079:2525516 [4] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-162-14:2525078:2525510 [3] NCCL INFO Connected NVLS tree
+ip-26-0-162-14:2525077:2525512 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-162-14:2525078:2525510 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-162-14:2525077:2525512 [2] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-162-14:2525077:2525512 [2] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-162-14:2525078:2525510 [3] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-162-14:2525078:2525510 [3] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-162-14:2525082:2525515 [7] NCCL INFO Connected NVLS tree
+ip-26-0-162-14:2525082:2525515 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-162-14:2525082:2525515 [7] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-162-14:2525082:2525515 [7] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-162-14:2525080:2525511 [5] NCCL INFO Connected NVLS tree
+ip-26-0-162-14:2525080:2525511 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-162-14:2525080:2525511 [5] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-162-14:2525080:2525511 [5] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-162-14:2525076:2525513 [1] NCCL INFO Connected NVLS tree
+ip-26-0-162-14:2525076:2525513 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-162-14:2525076:2525513 [1] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-162-14:2525076:2525513 [1] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-170-132:3330188:3330632 [5] NCCL INFO Connected NVLS tree
+ip-26-0-170-132:3330188:3330632 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-170-132:3330188:3330632 [5] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-170-132:3330188:3330632 [5] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-170-132:3330187:3330630 [4] NCCL INFO Connected NVLS tree
+ip-26-0-170-132:3330189:3330633 [6] NCCL INFO Connected NVLS tree
+ip-26-0-170-132:3330190:3330627 [7] NCCL INFO Connected NVLS tree
+ip-26-0-170-132:3330187:3330630 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-170-132:3330190:3330627 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-170-132:3330187:3330630 [4] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-170-132:3330187:3330630 [4] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-170-132:3330189:3330633 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-170-132:3330190:3330627 [7] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-170-132:3330190:3330627 [7] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-170-132:3330189:3330633 [6] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-170-132:3330189:3330633 [6] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-170-132:3330186:3330629 [3] NCCL INFO Connected NVLS tree
+ip-26-0-170-132:3330186:3330629 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-170-132:3330186:3330629 [3] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-170-132:3330186:3330629 [3] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-170-132:3330183:3330626 [0] NCCL INFO Connected NVLS tree
+ip-26-0-170-132:3330183:3330626 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-170-132:3330183:3330626 [0] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-170-132:3330183:3330626 [0] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-170-132:3330185:3330628 [2] NCCL INFO Connected NVLS tree
+ip-26-0-170-132:3330185:3330628 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-170-132:3330185:3330628 [2] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-170-132:3330185:3330628 [2] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-170-132:3330184:3330631 [1] NCCL INFO Connected NVLS tree
+ip-26-0-170-132:3330184:3330631 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-170-132:3330184:3330631 [1] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-170-132:3330184:3330631 [1] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-162-14:2525082:2525515 [7] NCCL INFO comm 0x9e8b8b0 rank 7 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0xcc04d7b16f1339ed - Init COMPLETE
+ip-26-0-162-14:2525076:2525513 [1] NCCL INFO comm 0xa386440 rank 1 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0xcc04d7b16f1339ed - Init COMPLETE
+ip-26-0-162-14:2525080:2525511 [5] NCCL INFO comm 0x9c5e6a0 rank 5 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0xcc04d7b16f1339ed - Init COMPLETE
+ip-26-0-162-14:2525081:2525514 [6] NCCL INFO comm 0x950b8e0 rank 6 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0xcc04d7b16f1339ed - Init COMPLETE
+ip-26-0-162-14:2525078:2525510 [3] NCCL INFO comm 0x923c7b0 rank 3 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0xcc04d7b16f1339ed - Init COMPLETE
+ip-26-0-162-14:2525075:2525509 [0] NCCL INFO comm 0x988a6e0 rank 0 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0xcc04d7b16f1339ed - Init COMPLETE
+ip-26-0-162-14:2525079:2525516 [4] NCCL INFO comm 0x96da6c0 rank 4 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0xcc04d7b16f1339ed - Init COMPLETE
+ip-26-0-162-14:2525077:2525512 [2] NCCL INFO comm 0xa1cf600 rank 2 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0xcc04d7b16f1339ed - Init COMPLETE
+ip-26-0-170-132:3330188:3330632 [5] NCCL INFO comm 0xa5074f0 rank 5 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0xa8d0c6af6511c913 - Init COMPLETE
+ip-26-0-170-132:3330190:3330627 [7] NCCL INFO comm 0xa535420 rank 7 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0xa8d0c6af6511c913 - Init COMPLETE
+ip-26-0-170-132:3330186:3330629 [3] NCCL INFO comm 0x8f2f440 rank 3 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0xa8d0c6af6511c913 - Init COMPLETE
+ip-26-0-170-132:3330184:3330631 [1] NCCL INFO comm 0x9625de0 rank 1 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0xa8d0c6af6511c913 - Init COMPLETE
+ip-26-0-170-132:3330189:3330633 [6] NCCL INFO comm 0x87c6270 rank 6 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0xa8d0c6af6511c913 - Init COMPLETE
+ip-26-0-170-132:3330185:3330628 [2] NCCL INFO comm 0x9ab5de0 rank 2 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0xa8d0c6af6511c913 - Init COMPLETE
+ip-26-0-170-132:3330183:3330626 [0] NCCL INFO comm 0x94ede40 rank 0 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0xa8d0c6af6511c913 - Init COMPLETE
+ip-26-0-170-132:3330187:3330630 [4] NCCL INFO comm 0x9a52690 rank 4 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0xa8d0c6af6511c913 - Init COMPLETE
+12/28/2024 01:36:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]: Config:
+12/28/2024 01:36:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]: Config(general=GeneralArgs(project='debug',
+12/28/2024 01:36:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                            run='469G_dp2_tp4_pp4_acc4_mbs32_seq4096_zero1_tpmodeRED_vocab131k',
+12/28/2024 01:36:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                            seed=42,
+12/28/2024 01:36:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                            step=None,
+12/28/2024 01:36:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                            consumed_train_samples=None,
+12/28/2024 01:36:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                            benchmark_csv_path=PosixPath('benchmark/results/bench_final2.csv'),
+12/28/2024 01:36:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                            ignore_sanity_checks=True),
+12/28/2024 01:36:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:        parallelism=ParallelismArgs(dp=2,
+12/28/2024 01:36:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                                    pp=4,
+12/28/2024 01:36:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                                    tp=4,
+12/28/2024 01:36:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                                    pp_engine=<nanotron.parallel.pipeline_parallel.engine.OneForwardOneBackwardPipelineEngine object at 0x7f4dcd684dc0>,
+12/28/2024 01:36:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                                    tp_mode=<TensorParallelLinearMode.REDUCE_SCATTER: 2>,
+12/28/2024 01:36:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                                    tp_linear_async_communication=True,
+12/28/2024 01:36:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                                    recompute_layer=False,
+12/28/2024 01:36:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                                    tp_recompute_allgather=True,
+12/28/2024 01:36:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                                    expert_parallel_size=1),
+12/28/2024 01:36:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:        model=ModelArgs(model_config=LlamaConfig(bos_token_id=0,
+12/28/2024 01:36:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                                                 eos_token_id=0,
+12/28/2024 01:36:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                                                 hidden_act='silu',
+12/28/2024 01:36:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                                                 hidden_size=16384,
+12/28/2024 01:36:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                                                 initializer_range=0.02,
+12/28/2024 01:36:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                                                 intermediate_size=53248,
+12/28/2024 01:36:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                                                 is_llama_config=True,
+12/28/2024 01:36:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                                                 max_position_embeddings=4096,
+12/28/2024 01:36:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                                                 num_attention_heads=128,
+12/28/2024 01:36:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                                                 num_hidden_layers=126,
+12/28/2024 01:36:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                                                 num_key_value_heads=128,
+12/28/2024 01:36:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                                                 pad_token_id=None,
+12/28/2024 01:36:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                                                 pretraining_tp=1,
+12/28/2024 01:36:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                                                 rms_norm_eps=1e-05,
+12/28/2024 01:36:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                                                 rope_scaling=None,
+12/28/2024 01:36:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                                                 rope_theta=10000.0,
+12/28/2024 01:36:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                                                 rope_interleaved=False,
+12/28/2024 01:36:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                                                 tie_word_embeddings=False,
+12/28/2024 01:36:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                                                 use_cache=True,
+12/28/2024 01:36:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                                                 vocab_size=131072),
+12/28/2024 01:36:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                        init_method=RandomInit(std=0.02),
+12/28/2024 01:36:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                        dtype=torch.bfloat16,
+12/28/2024 01:36:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                        make_vocab_size_divisible_by=1,
+12/28/2024 01:36:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                        ddp_bucket_cap_mb=25),
+12/28/2024 01:36:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:        tokenizer=TokenizerArgs(tokenizer_name_or_path='robot-test/dummy-tokenizer-wordlevel',
+12/28/2024 01:36:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                                tokenizer_revision=None,
+12/28/2024 01:36:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                                tokenizer_max_length=None),
+12/28/2024 01:36:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:        checkpoints=CheckpointsArgs(checkpoints_path=PosixPath('checkpoints'),
+12/28/2024 01:36:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                                    checkpoint_interval=10000,
+12/28/2024 01:36:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                                    save_initial_state=False,
+12/28/2024 01:36:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                                    save_final_state=False,
+12/28/2024 01:36:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                                    resume_checkpoint_path=None,
+12/28/2024 01:36:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                                    checkpoints_path_is_shared_file_system=False),
+12/28/2024 01:36:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:        logging=LoggingArgs(log_level='info',
+12/28/2024 01:36:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                            log_level_replica='info',
+12/28/2024 01:36:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                            iteration_step_info_interval=1),
+12/28/2024 01:36:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:        tokens=TokensArgs(sequence_length=4096,
+12/28/2024 01:36:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                          train_steps=100,
+12/28/2024 01:36:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                          micro_batch_size=32,
+12/28/2024 01:36:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                          batch_accumulation_per_replica=4,
+12/28/2024 01:36:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                          val_check_interval=100,
+12/28/2024 01:36:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                          limit_val_batches=0,
+12/28/2024 01:36:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                          limit_test_batches=0),
+12/28/2024 01:36:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:        optimizer=OptimizerArgs(optimizer_factory=AdamWOptimizerArgs(adam_eps=1e-08,
+12/28/2024 01:36:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                                                                     adam_beta1=0.9,
+12/28/2024 01:36:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                                                                     adam_beta2=0.95,
+12/28/2024 01:36:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                                                                     torch_adam_is_fused=True,
+12/28/2024 01:36:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                                                                     name='adamW'),
+12/28/2024 01:36:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                                zero_stage=1,
+12/28/2024 01:36:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                                weight_decay=0.01,
+12/28/2024 01:36:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                                clip_grad=1.0,
+12/28/2024 01:36:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                                accumulate_grad_in_fp32=True,
+12/28/2024 01:36:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                                learning_rate_scheduler=LRSchedulerArgs(learning_rate=0.0003,
+12/28/2024 01:36:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                                                                        lr_warmup_steps=2,
+12/28/2024 01:36:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                                                                        lr_warmup_style='linear',
+12/28/2024 01:36:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                                                                        lr_decay_style='cosine',
+12/28/2024 01:36:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                                                                        lr_decay_steps=13,
+12/28/2024 01:36:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                                                                        lr_decay_starting_step=None,
+12/28/2024 01:36:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                                                                        min_decay_lr=1e-05)),
+12/28/2024 01:36:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:        data_stages=[DatasetStageArgs(name='Stable Training Stage',
+12/28/2024 01:36:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                                      start_training_step=1,
+12/28/2024 01:36:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                                      data=DataArgs(dataset=None,
+12/28/2024 01:36:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                                                    seed=42,
+12/28/2024 01:36:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:                                                    num_loading_workers=1))],
+12/28/2024 01:36:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:        profiler=None,
+12/28/2024 01:36:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:        lighteval=None,
+12/28/2024 01:36:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:        s3_upload=None)
+12/28/2024 01:36:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]: Model Config:
+12/28/2024 01:36:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]: LlamaConfig(bos_token_id=0,
+12/28/2024 01:36:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:             eos_token_id=0,
+12/28/2024 01:36:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:             hidden_act='silu',
+12/28/2024 01:36:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:             hidden_size=16384,
+12/28/2024 01:36:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:             initializer_range=0.02,
+12/28/2024 01:36:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:             intermediate_size=53248,
+12/28/2024 01:36:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:             is_llama_config=True,
+12/28/2024 01:36:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:             max_position_embeddings=4096,
+12/28/2024 01:36:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:             num_attention_heads=128,
+12/28/2024 01:36:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:             num_hidden_layers=126,
+12/28/2024 01:36:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:             num_key_value_heads=128,
+12/28/2024 01:36:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:             pad_token_id=None,
+12/28/2024 01:36:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:             pretraining_tp=1,
+12/28/2024 01:36:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:             rms_norm_eps=1e-05,
+12/28/2024 01:36:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:             rope_scaling=None,
+12/28/2024 01:36:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:             rope_theta=10000.0,
+12/28/2024 01:36:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:             rope_interleaved=False,
+12/28/2024 01:36:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:             tie_word_embeddings=False,
+12/28/2024 01:36:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:             use_cache=True,
+12/28/2024 01:36:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]:             vocab_size=131072)
+12/28/2024 01:36:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]: Building model..
+12/28/2024 01:36:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]: Initialize RoPE Theta = 10000.0
+12/28/2024 01:36:36 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]: Setting PP block ranks...
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+ip-26-0-170-132:3330186:3330663 [3] NCCL INFO Using network Libfabric
+ip-26-0-170-132:3330183:3330662 [0] NCCL INFO Using network Libfabric
+ip-26-0-170-132:3330185:3330664 [2] NCCL INFO Using network Libfabric
+ip-26-0-170-132:3330187:3330667 [4] NCCL INFO Using network Libfabric
+ip-26-0-170-132:3330184:3330665 [1] NCCL INFO Using network Libfabric
+ip-26-0-163-58:934727:935188 [7] NCCL INFO Using network Libfabric
+ip-26-0-163-58:934724:935186 [4] NCCL INFO Using network Libfabric
+ip-26-0-170-132:3330189:3330668 [6] NCCL INFO Using network Libfabric
+ip-26-0-163-58:934725:935190 [5] NCCL INFO Using network Libfabric
+ip-26-0-163-58:934726:935189 [6] NCCL INFO Using network Libfabric
+ip-26-0-170-31:3197187:3197764 [7] NCCL INFO Using network Libfabric
+ip-26-0-170-31:3197184:3197762 [4] NCCL INFO Using network Libfabric
+ip-26-0-170-132:3330190:3330670 [7] NCCL INFO Using network Libfabric
+ip-26-0-170-132:3330188:3330669 [5] NCCL INFO Using network Libfabric
+ip-26-0-163-58:934720:935191 [0] NCCL INFO Using network Libfabric
+ip-26-0-162-14:2525081:2525549 [6] NCCL INFO Using network Libfabric
+ip-26-0-170-31:3197180:3197763 [0] NCCL INFO Using network Libfabric
+ip-26-0-162-14:2525082:2525548 [7] NCCL INFO Using network Libfabric
+ip-26-0-170-31:3197185:3197767 [5] NCCL INFO Using network Libfabric
+ip-26-0-162-14:2525079:2525546 [4] NCCL INFO Using network Libfabric
+ip-26-0-170-132:3330183:3330662 [0] NCCL INFO comm 0xa828b10 rank 0 nranks 4 cudaDev 0 nvmlDev 0 busId 53000 commId 0x302943922abb6c5a - Init START
+ip-26-0-170-132:3330186:3330663 [3] NCCL INFO comm 0xa269dd0 rank 3 nranks 4 cudaDev 3 nvmlDev 3 busId 86000 commId 0x302943922abb6c5a - Init START
+ip-26-0-170-132:3330184:3330665 [1] NCCL INFO comm 0xa91e090 rank 1 nranks 4 cudaDev 1 nvmlDev 1 busId 64000 commId 0x302943922abb6c5a - Init START
+ip-26-0-170-132:3330185:3330664 [2] NCCL INFO comm 0xadb09f0 rank 2 nranks 4 cudaDev 2 nvmlDev 2 busId 75000 commId 0x302943922abb6c5a - Init START
+ip-26-0-163-58:934721:935193 [1] NCCL INFO Using network Libfabric
+ip-26-0-162-14:2525080:2525550 [5] NCCL INFO Using network Libfabric
+ip-26-0-162-14:2525075:2525551 [0] NCCL INFO Using network Libfabric
+ip-26-0-162-14:2525076:2525552 [1] NCCL INFO Using network Libfabric
+ip-26-0-162-14:2525078:2525554 [3] NCCL INFO Using network Libfabric
+ip-26-0-163-58:934723:935194 [3] NCCL INFO Using network Libfabric
+ip-26-0-170-132:3330185:3330664 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-170-132:3330184:3330665 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-170-132:3330183:3330662 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-170-132:3330186:3330663 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-170-31:3197181:3197768 [1] NCCL INFO Using network Libfabric
+ip-26-0-162-14:2525077:2525553 [2] NCCL INFO Using network Libfabric
+ip-26-0-163-58:934722:935192 [2] NCCL INFO Using network Libfabric
+ip-26-0-163-58:934726:935189 [6] NCCL INFO comm 0xb169000 rank 2 nranks 4 cudaDev 6 nvmlDev 6 busId b9000 commId 0xbda347e6977b7d76 - Init START
+ip-26-0-163-58:934724:935186 [4] NCCL INFO comm 0xac186b0 rank 0 nranks 4 cudaDev 4 nvmlDev 4 busId 97000 commId 0xbda347e6977b7d76 - Init START
+ip-26-0-163-58:934727:935188 [7] NCCL INFO comm 0xad794e0 rank 3 nranks 4 cudaDev 7 nvmlDev 7 busId ca000 commId 0xbda347e6977b7d76 - Init START
+ip-26-0-163-58:934725:935190 [5] NCCL INFO comm 0x9e04790 rank 1 nranks 4 cudaDev 5 nvmlDev 5 busId a8000 commId 0xbda347e6977b7d76 - Init START
+ip-26-0-170-31:3197182:3197769 [2] NCCL INFO Using network Libfabric
+ip-26-0-170-31:3197186:3197766 [6] NCCL INFO Using network Libfabric
+ip-26-0-170-132:3330188:3330669 [5] NCCL INFO comm 0xb841e10 rank 1 nranks 4 cudaDev 5 nvmlDev 5 busId a8000 commId 0xecbe8b333a0a7265 - Init START
+ip-26-0-170-132:3330187:3330667 [4] NCCL INFO comm 0xad482e0 rank 0 nranks 4 cudaDev 4 nvmlDev 4 busId 97000 commId 0xecbe8b333a0a7265 - Init START
+ip-26-0-170-132:3330189:3330668 [6] NCCL INFO comm 0x9b016b0 rank 2 nranks 4 cudaDev 6 nvmlDev 6 busId b9000 commId 0xecbe8b333a0a7265 - Init START
+ip-26-0-170-132:3330190:3330670 [7] NCCL INFO comm 0xb86f370 rank 3 nranks 4 cudaDev 7 nvmlDev 7 busId ca000 commId 0xecbe8b333a0a7265 - Init START
+ip-26-0-170-31:3197183:3197765 [3] NCCL INFO Using network Libfabric
+ip-26-0-163-58:934726:935189 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-163-58:934727:935188 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-163-58:934725:935190 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-163-58:934724:935186 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-170-132:3330188:3330669 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-170-132:3330190:3330670 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-170-132:3330187:3330667 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-170-132:3330189:3330668 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-162-14:2525082:2525548 [7] NCCL INFO comm 0xb3960f0 rank 3 nranks 4 cudaDev 7 nvmlDev 7 busId ca000 commId 0xe0ce9c074ef535b1 - Init START
+ip-26-0-162-14:2525079:2525546 [4] NCCL INFO comm 0xabe60b0 rank 0 nranks 4 cudaDev 4 nvmlDev 4 busId 97000 commId 0xe0ce9c074ef535b1 - Init START
+ip-26-0-162-14:2525081:2525549 [6] NCCL INFO comm 0xaa18180 rank 2 nranks 4 cudaDev 6 nvmlDev 6 busId b9000 commId 0xe0ce9c074ef535b1 - Init START
+ip-26-0-162-14:2525080:2525550 [5] NCCL INFO comm 0xb16a6a0 rank 1 nranks 4 cudaDev 5 nvmlDev 5 busId a8000 commId 0xe0ce9c074ef535b1 - Init START
+ip-26-0-163-58:934721:935193 [1] NCCL INFO comm 0xb8a3740 rank 1 nranks 4 cudaDev 1 nvmlDev 1 busId 64000 commId 0xa9cee0599e4ded26 - Init START
+ip-26-0-163-58:934720:935191 [0] NCCL INFO comm 0xab5f260 rank 0 nranks 4 cudaDev 0 nvmlDev 0 busId 53000 commId 0xa9cee0599e4ded26 - Init START
+ip-26-0-163-58:934722:935192 [2] NCCL INFO comm 0x9e22940 rank 2 nranks 4 cudaDev 2 nvmlDev 2 busId 75000 commId 0xa9cee0599e4ded26 - Init START
+ip-26-0-163-58:934723:935194 [3] NCCL INFO comm 0xae236c0 rank 3 nranks 4 cudaDev 3 nvmlDev 3 busId 86000 commId 0xa9cee0599e4ded26 - Init START
+ip-26-0-162-14:2525076:2525552 [1] NCCL INFO comm 0xb891cc0 rank 1 nranks 4 cudaDev 1 nvmlDev 1 busId 64000 commId 0xf2cf793020f06054 - Init START
+ip-26-0-162-14:2525077:2525553 [2] NCCL INFO comm 0xc1217b0 rank 2 nranks 4 cudaDev 2 nvmlDev 2 busId 75000 commId 0xf2cf793020f06054 - Init START
+ip-26-0-162-14:2525075:2525551 [0] NCCL INFO comm 0xad97080 rank 0 nranks 4 cudaDev 0 nvmlDev 0 busId 53000 commId 0xf2cf793020f06054 - Init START
+ip-26-0-162-14:2525078:2525554 [3] NCCL INFO comm 0xa7490e0 rank 3 nranks 4 cudaDev 3 nvmlDev 3 busId 86000 commId 0xf2cf793020f06054 - Init START
+ip-26-0-170-31:3197180:3197763 [0] NCCL INFO comm 0xb4ae6f0 rank 0 nranks 4 cudaDev 0 nvmlDev 0 busId 53000 commId 0xdf4467e4af4c2d67 - Init START
+ip-26-0-170-31:3197181:3197768 [1] NCCL INFO comm 0xaed8a00 rank 1 nranks 4 cudaDev 1 nvmlDev 1 busId 64000 commId 0xdf4467e4af4c2d67 - Init START
+ip-26-0-170-31:3197182:3197769 [2] NCCL INFO comm 0x9c0ef50 rank 2 nranks 4 cudaDev 2 nvmlDev 2 busId 75000 commId 0xdf4467e4af4c2d67 - Init START
+ip-26-0-170-31:3197183:3197765 [3] NCCL INFO comm 0xb27e620 rank 3 nranks 4 cudaDev 3 nvmlDev 3 busId 86000 commId 0xdf4467e4af4c2d67 - Init START
+ip-26-0-170-31:3197185:3197767 [5] NCCL INFO comm 0x9b5cf70 rank 1 nranks 4 cudaDev 5 nvmlDev 5 busId a8000 commId 0xe5acb1f64c970582 - Init START
+ip-26-0-170-31:3197186:3197766 [6] NCCL INFO comm 0xb345070 rank 2 nranks 4 cudaDev 6 nvmlDev 6 busId b9000 commId 0xe5acb1f64c970582 - Init START
+ip-26-0-163-58:934720:935191 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-163-58:934721:935193 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-170-31:3197187:3197764 [7] NCCL INFO comm 0xb973ff0 rank 3 nranks 4 cudaDev 7 nvmlDev 7 busId ca000 commId 0xe5acb1f64c970582 - Init START
+ip-26-0-170-31:3197184:3197762 [4] NCCL INFO comm 0xb411040 rank 0 nranks 4 cudaDev 4 nvmlDev 4 busId 97000 commId 0xe5acb1f64c970582 - Init START
+ip-26-0-163-58:934722:935192 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-163-58:934723:935194 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-170-31:3197181:3197768 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-170-31:3197180:3197763 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-170-31:3197182:3197769 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-170-31:3197184:3197762 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-170-31:3197183:3197765 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-170-31:3197185:3197767 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-170-31:3197187:3197764 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-170-31:3197186:3197766 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-162-14:2525081:2525549 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-162-14:2525082:2525548 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-162-14:2525079:2525546 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-162-14:2525080:2525550 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-162-14:2525077:2525553 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-162-14:2525075:2525551 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/253
+ip-26-0-162-14:2525076:2525552 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-162-14:2525078:2525554 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-170-132:3330183:3330662 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-170-132:3330183:3330662 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-170-132:3330189:3330668 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-170-132:3330189:3330668 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-170-132:3330188:3330669 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-170-132:3330188:3330669 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-162-14:2525080:2525550 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-162-14:2525080:2525550 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-170-132:3330186:3330663 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-170-132:3330186:3330663 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-162-14:2525082:2525548 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-162-14:2525082:2525548 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-162-14:2525079:2525546 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-162-14:2525079:2525546 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-162-14:2525081:2525549 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-162-14:2525081:2525549 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-170-132:3330185:3330664 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-170-132:3330185:3330664 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-170-132:3330187:3330667 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-170-132:3330187:3330667 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-162-14:2525077:2525553 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-162-14:2525077:2525553 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-162-14:2525081:2525549 [6] NCCL INFO Trees [0] 3/-1/-1->2->1 [1] 3/-1/-1->2->1 [2] 3/-1/-1->2->1 [3] 3/-1/-1->2->1 [4] 3/-1/-1->2->1 [5] 3/-1/-1->2->1 [6] 3/-1/-1->2->1 [7] 3/-1/-1->2->1 [8] 3/-1/-1->2->1 [9] 3/-1/-1->2->1 [10] 3/-1/-1->2->1 [11] 3/-1/-1->2->1 [12] 3/-1/-1->2->1 [13] 3/-1/-1->2->1 [14] 3/-1/-1->2->1 [15] 3/-1/-1->2->1 [16] 3/-1/-1->2->1 [17] 3/-1/-1->2->1 [18] 3/-1/-1->2->1 [19] 3/-1/-1->2->1 [20] 3/-1/-1->2->1 [21] 3/-1/-1->2->1 [22] 3/-1/-1->2->1 [23] 3/-1/-1->2->1
+ip-26-0-162-14:2525079:2525546 [4] NCCL INFO Channel 00/24 :    0   1   2   3
+ip-26-0-162-14:2525081:2525549 [6] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-162-14:2525082:2525548 [7] NCCL INFO Trees [0] -1/-1/-1->3->2 [1] -1/-1/-1->3->2 [2] -1/-1/-1->3->2 [3] -1/-1/-1->3->2 [4] -1/-1/-1->3->2 [5] -1/-1/-1->3->2 [6] -1/-1/-1->3->2 [7] -1/-1/-1->3->2 [8] -1/-1/-1->3->2 [9] -1/-1/-1->3->2 [10] -1/-1/-1->3->2 [11] -1/-1/-1->3->2 [12] -1/-1/-1->3->2 [13] -1/-1/-1->3->2 [14] -1/-1/-1->3->2 [15] -1/-1/-1->3->2 [16] -1/-1/-1->3->2 [17] -1/-1/-1->3->2 [18] -1/-1/-1->3->2 [19] -1/-1/-1->3->2 [20] -1/-1/-1->3->2 [21] -1/-1/-1->3->2 [22] -1/-1/-1->3->2 [23] -1/-1/-1->3->2
+ip-26-0-162-14:2525079:2525546 [4] NCCL INFO Channel 01/24 :    0   1   2   3
+ip-26-0-162-14:2525080:2525550 [5] NCCL INFO Trees [0] 2/-1/-1->1->0 [1] 2/-1/-1->1->0 [2] 2/-1/-1->1->0 [3] 2/-1/-1->1->0 [4] 2/-1/-1->1->0 [5] 2/-1/-1->1->0 [6] 2/-1/-1->1->0 [7] 2/-1/-1->1->0 [8] 2/-1/-1->1->0 [9] 2/-1/-1->1->0 [10] 2/-1/-1->1->0 [11] 2/-1/-1->1->0 [12] 2/-1/-1->1->0 [13] 2/-1/-1->1->0 [14] 2/-1/-1->1->0 [15] 2/-1/-1->1->0 [16] 2/-1/-1->1->0 [17] 2/-1/-1->1->0 [18] 2/-1/-1->1->0 [19] 2/-1/-1->1->0 [20] 2/-1/-1->1->0 [21] 2/-1/-1->1->0 [22] 2/-1/-1->1->0 [23] 2/-1/-1->1->0
+ip-26-0-162-14:2525082:2525548 [7] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-162-14:2525079:2525546 [4] NCCL INFO Channel 02/24 :    0   1   2   3
+ip-26-0-162-14:2525080:2525550 [5] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-162-14:2525079:2525546 [4] NCCL INFO Channel 03/24 :    0   1   2   3
+ip-26-0-162-14:2525079:2525546 [4] NCCL INFO Channel 04/24 :    0   1   2   3
+ip-26-0-162-14:2525079:2525546 [4] NCCL INFO Channel 05/24 :    0   1   2   3
+ip-26-0-162-14:2525079:2525546 [4] NCCL INFO Channel 06/24 :    0   1   2   3
+ip-26-0-162-14:2525079:2525546 [4] NCCL INFO Channel 07/24 :    0   1   2   3
+ip-26-0-162-14:2525079:2525546 [4] NCCL INFO Channel 08/24 :    0   1   2   3
+ip-26-0-162-14:2525079:2525546 [4] NCCL INFO Channel 09/24 :    0   1   2   3
+ip-26-0-162-14:2525079:2525546 [4] NCCL INFO Channel 10/24 :    0   1   2   3
+ip-26-0-162-14:2525079:2525546 [4] NCCL INFO Channel 11/24 :    0   1   2   3
+ip-26-0-162-14:2525079:2525546 [4] NCCL INFO Channel 12/24 :    0   1   2   3
+ip-26-0-162-14:2525079:2525546 [4] NCCL INFO Channel 13/24 :    0   1   2   3
+ip-26-0-162-14:2525079:2525546 [4] NCCL INFO Channel 14/24 :    0   1   2   3
+ip-26-0-162-14:2525079:2525546 [4] NCCL INFO Channel 15/24 :    0   1   2   3
+ip-26-0-162-14:2525079:2525546 [4] NCCL INFO Channel 16/24 :    0   1   2   3
+ip-26-0-162-14:2525079:2525546 [4] NCCL INFO Channel 17/24 :    0   1   2   3
+ip-26-0-162-14:2525079:2525546 [4] NCCL INFO Channel 18/24 :    0   1   2   3
+ip-26-0-162-14:2525079:2525546 [4] NCCL INFO Channel 19/24 :    0   1   2   3
+ip-26-0-162-14:2525079:2525546 [4] NCCL INFO Channel 20/24 :    0   1   2   3
+ip-26-0-162-14:2525079:2525546 [4] NCCL INFO Channel 21/24 :    0   1   2   3
+ip-26-0-162-14:2525079:2525546 [4] NCCL INFO Channel 22/24 :    0   1   2   3
+ip-26-0-162-14:2525079:2525546 [4] NCCL INFO Channel 23/24 :    0   1   2   3
+ip-26-0-162-14:2525079:2525546 [4] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] 1/-1/-1->0->-1 [7] 1/-1/-1->0->-1 [8] 1/-1/-1->0->-1 [9] 1/-1/-1->0->-1 [10] 1/-1/-1->0->-1 [11] 1/-1/-1->0->-1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] 1/-1/-1->0->-1 [19] 1/-1/-1->0->-1 [20] 1/-1/-1->0->-1 [21] 1/-1/-1->0->-1 [22] 1/-1/-1->0->-1 [23] 1/-1/-1->0->-1
+ip-26-0-162-14:2525079:2525546 [4] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-170-132:3330184:3330665 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-170-132:3330184:3330665 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-170-132:3330190:3330670 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-170-132:3330190:3330670 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-162-14:2525075:2525551 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-162-14:2525075:2525551 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-170-132:3330184:3330665 [1] NCCL INFO Trees [0] 2/-1/-1->1->0 [1] 2/-1/-1->1->0 [2] 2/-1/-1->1->0 [3] 2/-1/-1->1->0 [4] 2/-1/-1->1->0 [5] 2/-1/-1->1->0 [6] 2/-1/-1->1->0 [7] 2/-1/-1->1->0 [8] 2/-1/-1->1->0 [9] 2/-1/-1->1->0 [10] 2/-1/-1->1->0 [11] 2/-1/-1->1->0 [12] 2/-1/-1->1->0 [13] 2/-1/-1->1->0 [14] 2/-1/-1->1->0 [15] 2/-1/-1->1->0 [16] 2/-1/-1->1->0 [17] 2/-1/-1->1->0 [18] 2/-1/-1->1->0 [19] 2/-1/-1->1->0 [20] 2/-1/-1->1->0 [21] 2/-1/-1->1->0 [22] 2/-1/-1->1->0 [23] 2/-1/-1->1->0
+ip-26-0-170-132:3330183:3330662 [0] NCCL INFO Channel 00/24 :    0   1   2   3
+ip-26-0-170-132:3330184:3330665 [1] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-170-132:3330183:3330662 [0] NCCL INFO Channel 01/24 :    0   1   2   3
+ip-26-0-170-132:3330185:3330664 [2] NCCL INFO Trees [0] 3/-1/-1->2->1 [1] 3/-1/-1->2->1 [2] 3/-1/-1->2->1 [3] 3/-1/-1->2->1 [4] 3/-1/-1->2->1 [5] 3/-1/-1->2->1 [6] 3/-1/-1->2->1 [7] 3/-1/-1->2->1 [8] 3/-1/-1->2->1 [9] 3/-1/-1->2->1 [10] 3/-1/-1->2->1 [11] 3/-1/-1->2->1 [12] 3/-1/-1->2->1 [13] 3/-1/-1->2->1 [14] 3/-1/-1->2->1 [15] 3/-1/-1->2->1 [16] 3/-1/-1->2->1 [17] 3/-1/-1->2->1 [18] 3/-1/-1->2->1 [19] 3/-1/-1->2->1 [20] 3/-1/-1->2->1 [21] 3/-1/-1->2->1 [22] 3/-1/-1->2->1 [23] 3/-1/-1->2->1
+ip-26-0-170-132:3330186:3330663 [3] NCCL INFO Trees [0] -1/-1/-1->3->2 [1] -1/-1/-1->3->2 [2] -1/-1/-1->3->2 [3] -1/-1/-1->3->2 [4] -1/-1/-1->3->2 [5] -1/-1/-1->3->2 [6] -1/-1/-1->3->2 [7] -1/-1/-1->3->2 [8] -1/-1/-1->3->2 [9] -1/-1/-1->3->2 [10] -1/-1/-1->3->2 [11] -1/-1/-1->3->2 [12] -1/-1/-1->3->2 [13] -1/-1/-1->3->2 [14] -1/-1/-1->3->2 [15] -1/-1/-1->3->2 [16] -1/-1/-1->3->2 [17] -1/-1/-1->3->2 [18] -1/-1/-1->3->2 [19] -1/-1/-1->3->2 [20] -1/-1/-1->3->2 [21] -1/-1/-1->3->2 [22] -1/-1/-1->3->2 [23] -1/-1/-1->3->2
+ip-26-0-162-14:2525078:2525554 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-162-14:2525078:2525554 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-170-132:3330183:3330662 [0] NCCL INFO Channel 02/24 :    0   1   2   3
+ip-26-0-170-132:3330186:3330663 [3] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-170-132:3330185:3330664 [2] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-170-132:3330183:3330662 [0] NCCL INFO Channel 03/24 :    0   1   2   3
+ip-26-0-170-132:3330183:3330662 [0] NCCL INFO Channel 04/24 :    0   1   2   3
+ip-26-0-170-132:3330183:3330662 [0] NCCL INFO Channel 05/24 :    0   1   2   3
+ip-26-0-170-132:3330183:3330662 [0] NCCL INFO Channel 06/24 :    0   1   2   3
+ip-26-0-170-132:3330183:3330662 [0] NCCL INFO Channel 07/24 :    0   1   2   3
+ip-26-0-170-132:3330183:3330662 [0] NCCL INFO Channel 08/24 :    0   1   2   3
+ip-26-0-170-132:3330183:3330662 [0] NCCL INFO Channel 09/24 :    0   1   2   3
+ip-26-0-170-132:3330183:3330662 [0] NCCL INFO Channel 10/24 :    0   1   2   3
+ip-26-0-170-132:3330183:3330662 [0] NCCL INFO Channel 11/24 :    0   1   2   3
+ip-26-0-170-132:3330183:3330662 [0] NCCL INFO Channel 12/24 :    0   1   2   3
+ip-26-0-170-132:3330183:3330662 [0] NCCL INFO Channel 13/24 :    0   1   2   3
+ip-26-0-170-132:3330183:3330662 [0] NCCL INFO Channel 14/24 :    0   1   2   3
+ip-26-0-170-132:3330183:3330662 [0] NCCL INFO Channel 15/24 :    0   1   2   3
+ip-26-0-170-132:3330183:3330662 [0] NCCL INFO Channel 16/24 :    0   1   2   3
+ip-26-0-170-132:3330183:3330662 [0] NCCL INFO Channel 17/24 :    0   1   2   3
+ip-26-0-170-132:3330183:3330662 [0] NCCL INFO Channel 18/24 :    0   1   2   3
+ip-26-0-170-132:3330183:3330662 [0] NCCL INFO Channel 19/24 :    0   1   2   3
+ip-26-0-170-132:3330183:3330662 [0] NCCL INFO Channel 20/24 :    0   1   2   3
+ip-26-0-170-132:3330183:3330662 [0] NCCL INFO Channel 21/24 :    0   1   2   3
+ip-26-0-170-132:3330183:3330662 [0] NCCL INFO Channel 22/24 :    0   1   2   3
+ip-26-0-170-132:3330183:3330662 [0] NCCL INFO Channel 23/24 :    0   1   2   3
+ip-26-0-170-132:3330183:3330662 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] 1/-1/-1->0->-1 [7] 1/-1/-1->0->-1 [8] 1/-1/-1->0->-1 [9] 1/-1/-1->0->-1 [10] 1/-1/-1->0->-1 [11] 1/-1/-1->0->-1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] 1/-1/-1->0->-1 [19] 1/-1/-1->0->-1 [20] 1/-1/-1->0->-1 [21] 1/-1/-1->0->-1 [22] 1/-1/-1->0->-1 [23] 1/-1/-1->0->-1
+ip-26-0-170-132:3330183:3330662 [0] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-162-14:2525076:2525552 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-162-14:2525076:2525552 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-163-58:934722:935192 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-163-58:934722:935192 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-170-132:3330187:3330667 [4] NCCL INFO Channel 00/24 :    0   1   2   3
+ip-26-0-170-132:3330187:3330667 [4] NCCL INFO Channel 01/24 :    0   1   2   3
+ip-26-0-170-132:3330187:3330667 [4] NCCL INFO Channel 02/24 :    0   1   2   3
+ip-26-0-170-132:3330187:3330667 [4] NCCL INFO Channel 03/24 :    0   1   2   3
+ip-26-0-170-132:3330190:3330670 [7] NCCL INFO Trees [0] -1/-1/-1->3->2 [1] -1/-1/-1->3->2 [2] -1/-1/-1->3->2 [3] -1/-1/-1->3->2 [4] -1/-1/-1->3->2 [5] -1/-1/-1->3->2 [6] -1/-1/-1->3->2 [7] -1/-1/-1->3->2 [8] -1/-1/-1->3->2 [9] -1/-1/-1->3->2 [10] -1/-1/-1->3->2 [11] -1/-1/-1->3->2 [12] -1/-1/-1->3->2 [13] -1/-1/-1->3->2 [14] -1/-1/-1->3->2 [15] -1/-1/-1->3->2 [16] -1/-1/-1->3->2 [17] -1/-1/-1->3->2 [18] -1/-1/-1->3->2 [19] -1/-1/-1->3->2 [20] -1/-1/-1->3->2 [21] -1/-1/-1->3->2 [22] -1/-1/-1->3->2 [23] -1/-1/-1->3->2
+ip-26-0-170-132:3330187:3330667 [4] NCCL INFO Channel 04/24 :    0   1   2   3
+ip-26-0-170-132:3330190:3330670 [7] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-170-132:3330187:3330667 [4] NCCL INFO Channel 05/24 :    0   1   2   3
+ip-26-0-170-132:3330188:3330669 [5] NCCL INFO Trees [0] 2/-1/-1->1->0 [1] 2/-1/-1->1->0 [2] 2/-1/-1->1->0 [3] 2/-1/-1->1->0 [4] 2/-1/-1->1->0 [5] 2/-1/-1->1->0 [6] 2/-1/-1->1->0 [7] 2/-1/-1->1->0 [8] 2/-1/-1->1->0 [9] 2/-1/-1->1->0 [10] 2/-1/-1->1->0 [11] 2/-1/-1->1->0 [12] 2/-1/-1->1->0 [13] 2/-1/-1->1->0 [14] 2/-1/-1->1->0 [15] 2/-1/-1->1->0 [16] 2/-1/-1->1->0 [17] 2/-1/-1->1->0 [18] 2/-1/-1->1->0 [19] 2/-1/-1->1->0 [20] 2/-1/-1->1->0 [21] 2/-1/-1->1->0 [22] 2/-1/-1->1->0 [23] 2/-1/-1->1->0
+ip-26-0-170-132:3330187:3330667 [4] NCCL INFO Channel 06/24 :    0   1   2   3
+ip-26-0-170-132:3330188:3330669 [5] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-170-132:3330187:3330667 [4] NCCL INFO Channel 07/24 :    0   1   2   3
+ip-26-0-170-132:3330187:3330667 [4] NCCL INFO Channel 08/24 :    0   1   2   3
+ip-26-0-170-132:3330187:3330667 [4] NCCL INFO Channel 09/24 :    0   1   2   3
+ip-26-0-170-132:3330187:3330667 [4] NCCL INFO Channel 10/24 :    0   1   2   3
+ip-26-0-170-132:3330187:3330667 [4] NCCL INFO Channel 11/24 :    0   1   2   3
+ip-26-0-170-132:3330187:3330667 [4] NCCL INFO Channel 12/24 :    0   1   2   3
+ip-26-0-170-132:3330189:3330668 [6] NCCL INFO Trees [0] 3/-1/-1->2->1 [1] 3/-1/-1->2->1 [2] 3/-1/-1->2->1 [3] 3/-1/-1->2->1 [4] 3/-1/-1->2->1 [5] 3/-1/-1->2->1 [6] 3/-1/-1->2->1 [7] 3/-1/-1->2->1 [8] 3/-1/-1->2->1 [9] 3/-1/-1->2->1 [10] 3/-1/-1->2->1 [11] 3/-1/-1->2->1 [12] 3/-1/-1->2->1 [13] 3/-1/-1->2->1 [14] 3/-1/-1->2->1 [15] 3/-1/-1->2->1 [16] 3/-1/-1->2->1 [17] 3/-1/-1->2->1 [18] 3/-1/-1->2->1 [19] 3/-1/-1->2->1 [20] 3/-1/-1->2->1 [21] 3/-1/-1->2->1 [22] 3/-1/-1->2->1 [23] 3/-1/-1->2->1
+ip-26-0-170-132:3330187:3330667 [4] NCCL INFO Channel 13/24 :    0   1   2   3
+ip-26-0-170-132:3330187:3330667 [4] NCCL INFO Channel 14/24 :    0   1   2   3
+ip-26-0-170-132:3330189:3330668 [6] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-170-132:3330187:3330667 [4] NCCL INFO Channel 15/24 :    0   1   2   3
+ip-26-0-170-132:3330187:3330667 [4] NCCL INFO Channel 16/24 :    0   1   2   3
+ip-26-0-170-132:3330187:3330667 [4] NCCL INFO Channel 17/24 :    0   1   2   3
+ip-26-0-170-132:3330187:3330667 [4] NCCL INFO Channel 18/24 :    0   1   2   3
+ip-26-0-170-132:3330187:3330667 [4] NCCL INFO Channel 19/24 :    0   1   2   3
+ip-26-0-170-132:3330187:3330667 [4] NCCL INFO Channel 20/24 :    0   1   2   3
+ip-26-0-170-132:3330187:3330667 [4] NCCL INFO Channel 21/24 :    0   1   2   3
+ip-26-0-170-132:3330187:3330667 [4] NCCL INFO Channel 22/24 :    0   1   2   3
+ip-26-0-170-132:3330187:3330667 [4] NCCL INFO Channel 23/24 :    0   1   2   3
+ip-26-0-170-132:3330187:3330667 [4] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] 1/-1/-1->0->-1 [7] 1/-1/-1->0->-1 [8] 1/-1/-1->0->-1 [9] 1/-1/-1->0->-1 [10] 1/-1/-1->0->-1 [11] 1/-1/-1->0->-1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] 1/-1/-1->0->-1 [19] 1/-1/-1->0->-1 [20] 1/-1/-1->0->-1 [21] 1/-1/-1->0->-1 [22] 1/-1/-1->0->-1 [23] 1/-1/-1->0->-1
+ip-26-0-170-132:3330187:3330667 [4] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-162-14:2525081:2525549 [6] NCCL INFO Channel 00/0 : 2[6] -> 3[7] via P2P/IPC
+ip-26-0-162-14:2525080:2525550 [5] NCCL INFO Channel 00/0 : 1[5] -> 2[6] via P2P/IPC
+ip-26-0-162-14:2525081:2525549 [6] NCCL INFO Channel 01/0 : 2[6] -> 3[7] via P2P/IPC
+ip-26-0-162-14:2525082:2525548 [7] NCCL INFO Channel 00/0 : 3[7] -> 0[4] via P2P/IPC
+ip-26-0-162-14:2525080:2525550 [5] NCCL INFO Channel 01/0 : 1[5] -> 2[6] via P2P/IPC
+ip-26-0-162-14:2525079:2525546 [4] NCCL INFO Channel 00/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-162-14:2525075:2525551 [0] NCCL INFO Channel 00/24 :    0   1   2   3
+ip-26-0-162-14:2525075:2525551 [0] NCCL INFO Channel 01/24 :    0   1   2   3
+ip-26-0-162-14:2525076:2525552 [1] NCCL INFO Trees [0] 2/-1/-1->1->0 [1] 2/-1/-1->1->0 [2] 2/-1/-1->1->0 [3] 2/-1/-1->1->0 [4] 2/-1/-1->1->0 [5] 2/-1/-1->1->0 [6] 2/-1/-1->1->0 [7] 2/-1/-1->1->0 [8] 2/-1/-1->1->0 [9] 2/-1/-1->1->0 [10] 2/-1/-1->1->0 [11] 2/-1/-1->1->0 [12] 2/-1/-1->1->0 [13] 2/-1/-1->1->0 [14] 2/-1/-1->1->0 [15] 2/-1/-1->1->0 [16] 2/-1/-1->1->0 [17] 2/-1/-1->1->0 [18] 2/-1/-1->1->0 [19] 2/-1/-1->1->0 [20] 2/-1/-1->1->0 [21] 2/-1/-1->1->0 [22] 2/-1/-1->1->0 [23] 2/-1/-1->1->0
+ip-26-0-162-14:2525075:2525551 [0] NCCL INFO Channel 02/24 :    0   1   2   3
+ip-26-0-162-14:2525076:2525552 [1] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-162-14:2525075:2525551 [0] NCCL INFO Channel 03/24 :    0   1   2   3
+ip-26-0-162-14:2525078:2525554 [3] NCCL INFO Trees [0] -1/-1/-1->3->2 [1] -1/-1/-1->3->2 [2] -1/-1/-1->3->2 [3] -1/-1/-1->3->2 [4] -1/-1/-1->3->2 [5] -1/-1/-1->3->2 [6] -1/-1/-1->3->2 [7] -1/-1/-1->3->2 [8] -1/-1/-1->3->2 [9] -1/-1/-1->3->2 [10] -1/-1/-1->3->2 [11] -1/-1/-1->3->2 [12] -1/-1/-1->3->2 [13] -1/-1/-1->3->2 [14] -1/-1/-1->3->2 [15] -1/-1/-1->3->2 [16] -1/-1/-1->3->2 [17] -1/-1/-1->3->2 [18] -1/-1/-1->3->2 [19] -1/-1/-1->3->2 [20] -1/-1/-1->3->2 [21] -1/-1/-1->3->2 [22] -1/-1/-1->3->2 [23] -1/-1/-1->3->2
+ip-26-0-162-14:2525075:2525551 [0] NCCL INFO Channel 04/24 :    0   1   2   3
+ip-26-0-162-14:2525075:2525551 [0] NCCL INFO Channel 05/24 :    0   1   2   3
+ip-26-0-162-14:2525078:2525554 [3] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-162-14:2525075:2525551 [0] NCCL INFO Channel 06/24 :    0   1   2   3
+ip-26-0-162-14:2525075:2525551 [0] NCCL INFO Channel 07/24 :    0   1   2   3
+ip-26-0-162-14:2525075:2525551 [0] NCCL INFO Channel 08/24 :    0   1   2   3
+ip-26-0-162-14:2525075:2525551 [0] NCCL INFO Channel 09/24 :    0   1   2   3
+ip-26-0-162-14:2525077:2525553 [2] NCCL INFO Trees [0] 3/-1/-1->2->1 [1] 3/-1/-1->2->1 [2] 3/-1/-1->2->1 [3] 3/-1/-1->2->1 [4] 3/-1/-1->2->1 [5] 3/-1/-1->2->1 [6] 3/-1/-1->2->1 [7] 3/-1/-1->2->1 [8] 3/-1/-1->2->1 [9] 3/-1/-1->2->1 [10] 3/-1/-1->2->1 [11] 3/-1/-1->2->1 [12] 3/-1/-1->2->1 [13] 3/-1/-1->2->1 [14] 3/-1/-1->2->1 [15] 3/-1/-1->2->1 [16] 3/-1/-1->2->1 [17] 3/-1/-1->2->1 [18] 3/-1/-1->2->1 [19] 3/-1/-1->2->1 [20] 3/-1/-1->2->1 [21] 3/-1/-1->2->1 [22] 3/-1/-1->2->1 [23] 3/-1/-1->2->1
+ip-26-0-162-14:2525075:2525551 [0] NCCL INFO Channel 10/24 :    0   1   2   3
+ip-26-0-162-14:2525075:2525551 [0] NCCL INFO Channel 11/24 :    0   1   2   3
+ip-26-0-162-14:2525077:2525553 [2] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-162-14:2525075:2525551 [0] NCCL INFO Channel 12/24 :    0   1   2   3
+ip-26-0-162-14:2525075:2525551 [0] NCCL INFO Channel 13/24 :    0   1   2   3
+ip-26-0-162-14:2525075:2525551 [0] NCCL INFO Channel 14/24 :    0   1   2   3
+ip-26-0-162-14:2525075:2525551 [0] NCCL INFO Channel 15/24 :    0   1   2   3
+ip-26-0-162-14:2525075:2525551 [0] NCCL INFO Channel 16/24 :    0   1   2   3
+ip-26-0-170-132:3330185:3330664 [2] NCCL INFO Channel 00/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-170-132:3330185:3330664 [2] NCCL INFO Channel 01/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-170-132:3330183:3330662 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-170-132:3330185:3330664 [2] NCCL INFO Channel 02/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-170-132:3330183:3330662 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-170-132:3330185:3330664 [2] NCCL INFO Channel 03/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-170-132:3330183:3330662 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-170-132:3330183:3330662 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-170-132:3330183:3330662 [0] NCCL INFO Channel 04/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-170-132:3330183:3330662 [0] NCCL INFO Channel 05/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-170-132:3330183:3330662 [0] NCCL INFO Channel 06/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-170-132:3330184:3330665 [1] NCCL INFO Channel 00/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-170-132:3330186:3330663 [3] NCCL INFO Channel 00/0 : 3[3] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330185:3330664 [2] NCCL INFO Channel 04/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-170-132:3330183:3330662 [0] NCCL INFO Channel 07/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-170-132:3330184:3330665 [1] NCCL INFO Channel 01/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-170-132:3330186:3330663 [3] NCCL INFO Channel 01/0 : 3[3] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330185:3330664 [2] NCCL INFO Channel 05/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-170-132:3330183:3330662 [0] NCCL INFO Channel 08/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-170-132:3330184:3330665 [1] NCCL INFO Channel 02/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-170-132:3330186:3330663 [3] NCCL INFO Channel 02/0 : 3[3] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330185:3330664 [2] NCCL INFO Channel 06/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-170-132:3330183:3330662 [0] NCCL INFO Channel 09/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-170-132:3330184:3330665 [1] NCCL INFO Channel 03/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-170-132:3330186:3330663 [3] NCCL INFO Channel 03/0 : 3[3] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330185:3330664 [2] NCCL INFO Channel 07/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-170-132:3330183:3330662 [0] NCCL INFO Channel 10/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-170-132:3330184:3330665 [1] NCCL INFO Channel 04/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-170-132:3330186:3330663 [3] NCCL INFO Channel 04/0 : 3[3] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330185:3330664 [2] NCCL INFO Channel 08/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-170-132:3330183:3330662 [0] NCCL INFO Channel 11/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-170-132:3330184:3330665 [1] NCCL INFO Channel 05/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-170-132:3330186:3330663 [3] NCCL INFO Channel 05/0 : 3[3] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330185:3330664 [2] NCCL INFO Channel 09/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-170-132:3330183:3330662 [0] NCCL INFO Channel 12/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-170-132:3330184:3330665 [1] NCCL INFO Channel 06/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-170-132:3330186:3330663 [3] NCCL INFO Channel 06/0 : 3[3] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330187:3330667 [4] NCCL INFO Channel 00/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-170-132:3330185:3330664 [2] NCCL INFO Channel 10/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-170-132:3330183:3330662 [0] NCCL INFO Channel 13/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-170-132:3330184:3330665 [1] NCCL INFO Channel 07/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-170-132:3330187:3330667 [4] NCCL INFO Channel 01/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-170-132:3330186:3330663 [3] NCCL INFO Channel 07/0 : 3[3] -> 0[0] via P2P/IPC
+ip-26-0-162-14:2525075:2525551 [0] NCCL INFO Channel 17/24 :    0   1   2   3
+ip-26-0-162-14:2525075:2525551 [0] NCCL INFO Channel 18/24 :    0   1   2   3
+ip-26-0-162-14:2525075:2525551 [0] NCCL INFO Channel 19/24 :    0   1   2   3
+ip-26-0-162-14:2525075:2525551 [0] NCCL INFO Channel 20/24 :    0   1   2   3
+ip-26-0-162-14:2525075:2525551 [0] NCCL INFO Channel 21/24 :    0   1   2   3
+ip-26-0-162-14:2525075:2525551 [0] NCCL INFO Channel 22/24 :    0   1   2   3
+ip-26-0-162-14:2525075:2525551 [0] NCCL INFO Channel 23/24 :    0   1   2   3
+ip-26-0-162-14:2525075:2525551 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] 1/-1/-1->0->-1 [7] 1/-1/-1->0->-1 [8] 1/-1/-1->0->-1 [9] 1/-1/-1->0->-1 [10] 1/-1/-1->0->-1 [11] 1/-1/-1->0->-1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] 1/-1/-1->0->-1 [19] 1/-1/-1->0->-1 [20] 1/-1/-1->0->-1 [21] 1/-1/-1->0->-1 [22] 1/-1/-1->0->-1 [23] 1/-1/-1->0->-1
+ip-26-0-162-14:2525075:2525551 [0] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-162-14:2525081:2525549 [6] NCCL INFO Channel 02/0 : 2[6] -> 3[7] via P2P/IPC
+ip-26-0-162-14:2525082:2525548 [7] NCCL INFO Channel 01/0 : 3[7] -> 0[4] via P2P/IPC
+ip-26-0-162-14:2525080:2525550 [5] NCCL INFO Channel 02/0 : 1[5] -> 2[6] via P2P/IPC
+ip-26-0-162-14:2525079:2525546 [4] NCCL INFO Channel 01/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-162-14:2525081:2525549 [6] NCCL INFO Channel 03/0 : 2[6] -> 3[7] via P2P/IPC
+ip-26-0-162-14:2525082:2525548 [7] NCCL INFO Channel 02/0 : 3[7] -> 0[4] via P2P/IPC
+ip-26-0-162-14:2525080:2525550 [5] NCCL INFO Channel 03/0 : 1[5] -> 2[6] via P2P/IPC
+ip-26-0-162-14:2525079:2525546 [4] NCCL INFO Channel 02/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-162-14:2525081:2525549 [6] NCCL INFO Channel 04/0 : 2[6] -> 3[7] via P2P/IPC
+ip-26-0-162-14:2525082:2525548 [7] NCCL INFO Channel 03/0 : 3[7] -> 0[4] via P2P/IPC
+ip-26-0-162-14:2525080:2525550 [5] NCCL INFO Channel 04/0 : 1[5] -> 2[6] via P2P/IPC
+ip-26-0-162-14:2525079:2525546 [4] NCCL INFO Channel 03/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-162-14:2525081:2525549 [6] NCCL INFO Channel 05/0 : 2[6] -> 3[7] via P2P/IPC
+ip-26-0-162-14:2525082:2525548 [7] NCCL INFO Channel 04/0 : 3[7] -> 0[4] via P2P/IPC
+ip-26-0-162-14:2525080:2525550 [5] NCCL INFO Channel 05/0 : 1[5] -> 2[6] via P2P/IPC
+ip-26-0-162-14:2525079:2525546 [4] NCCL INFO Channel 04/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-162-14:2525081:2525549 [6] NCCL INFO Channel 06/0 : 2[6] -> 3[7] via P2P/IPC
+ip-26-0-162-14:2525082:2525548 [7] NCCL INFO Channel 05/0 : 3[7] -> 0[4] via P2P/IPC
+ip-26-0-162-14:2525080:2525550 [5] NCCL INFO Channel 06/0 : 1[5] -> 2[6] via P2P/IPC
+ip-26-0-162-14:2525079:2525546 [4] NCCL INFO Channel 05/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-162-14:2525081:2525549 [6] NCCL INFO Channel 07/0 : 2[6] -> 3[7] via P2P/IPC
+ip-26-0-162-14:2525082:2525548 [7] NCCL INFO Channel 06/0 : 3[7] -> 0[4] via P2P/IPC
+ip-26-0-162-14:2525080:2525550 [5] NCCL INFO Channel 07/0 : 1[5] -> 2[6] via P2P/IPC
+ip-26-0-162-14:2525079:2525546 [4] NCCL INFO Channel 06/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-162-14:2525082:2525548 [7] NCCL INFO Channel 07/0 : 3[7] -> 0[4] via P2P/IPC
+ip-26-0-162-14:2525081:2525549 [6] NCCL INFO Channel 08/0 : 2[6] -> 3[7] via P2P/IPC
+ip-26-0-162-14:2525080:2525550 [5] NCCL INFO Channel 08/0 : 1[5] -> 2[6] via P2P/IPC
+ip-26-0-162-14:2525079:2525546 [4] NCCL INFO Channel 07/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-162-14:2525082:2525548 [7] NCCL INFO Channel 08/0 : 3[7] -> 0[4] via P2P/IPC
+ip-26-0-162-14:2525081:2525549 [6] NCCL INFO Channel 09/0 : 2[6] -> 3[7] via P2P/IPC
+ip-26-0-162-14:2525080:2525550 [5] NCCL INFO Channel 09/0 : 1[5] -> 2[6] via P2P/IPC
+ip-26-0-162-14:2525079:2525546 [4] NCCL INFO Channel 08/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-162-14:2525082:2525548 [7] NCCL INFO Channel 09/0 : 3[7] -> 0[4] via P2P/IPC
+ip-26-0-162-14:2525081:2525549 [6] NCCL INFO Channel 10/0 : 2[6] -> 3[7] via P2P/IPC
+ip-26-0-162-14:2525080:2525550 [5] NCCL INFO Channel 10/0 : 1[5] -> 2[6] via P2P/IPC
+ip-26-0-162-14:2525079:2525546 [4] NCCL INFO Channel 09/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-162-14:2525082:2525548 [7] NCCL INFO Channel 10/0 : 3[7] -> 0[4] via P2P/IPC
+ip-26-0-162-14:2525081:2525549 [6] NCCL INFO Channel 11/0 : 2[6] -> 3[7] via P2P/IPC
+ip-26-0-162-14:2525080:2525550 [5] NCCL INFO Channel 11/0 : 1[5] -> 2[6] via P2P/IPC
+ip-26-0-162-14:2525079:2525546 [4] NCCL INFO Channel 10/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-170-132:3330188:3330669 [5] NCCL INFO Channel 00/0 : 1[5] -> 2[6] via P2P/IPC
+ip-26-0-162-14:2525082:2525548 [7] NCCL INFO Channel 11/0 : 3[7] -> 0[4] via P2P/IPC
+ip-26-0-162-14:2525081:2525549 [6] NCCL INFO Channel 12/0 : 2[6] -> 3[7] via P2P/IPC
+ip-26-0-162-14:2525080:2525550 [5] NCCL INFO Channel 12/0 : 1[5] -> 2[6] via P2P/IPC
+ip-26-0-162-14:2525079:2525546 [4] NCCL INFO Channel 11/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-162-14:2525079:2525546 [4] NCCL INFO Channel 12/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-162-14:2525079:2525546 [4] NCCL INFO Channel 13/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-162-14:2525079:2525546 [4] NCCL INFO Channel 14/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-162-14:2525079:2525546 [4] NCCL INFO Channel 15/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-162-14:2525082:2525548 [7] NCCL INFO Channel 12/0 : 3[7] -> 0[4] via P2P/IPC
+ip-26-0-162-14:2525081:2525549 [6] NCCL INFO Channel 13/0 : 2[6] -> 3[7] via P2P/IPC
+ip-26-0-162-14:2525080:2525550 [5] NCCL INFO Channel 13/0 : 1[5] -> 2[6] via P2P/IPC
+ip-26-0-162-14:2525079:2525546 [4] NCCL INFO Channel 16/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-162-14:2525082:2525548 [7] NCCL INFO Channel 13/0 : 3[7] -> 0[4] via P2P/IPC
+ip-26-0-162-14:2525081:2525549 [6] NCCL INFO Channel 14/0 : 2[6] -> 3[7] via P2P/IPC
+ip-26-0-162-14:2525080:2525550 [5] NCCL INFO Channel 14/0 : 1[5] -> 2[6] via P2P/IPC
+ip-26-0-170-132:3330185:3330664 [2] NCCL INFO Channel 11/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-170-132:3330183:3330662 [0] NCCL INFO Channel 14/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-162-14:2525079:2525546 [4] NCCL INFO Channel 17/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-162-14:2525082:2525548 [7] NCCL INFO Channel 14/0 : 3[7] -> 0[4] via P2P/IPC
+ip-26-0-162-14:2525080:2525550 [5] NCCL INFO Channel 15/0 : 1[5] -> 2[6] via P2P/IPC
+ip-26-0-162-14:2525079:2525546 [4] NCCL INFO Channel 18/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-170-132:3330184:3330665 [1] NCCL INFO Channel 08/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-170-132:3330187:3330667 [4] NCCL INFO Channel 02/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-170-132:3330186:3330663 [3] NCCL INFO Channel 08/0 : 3[3] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330188:3330669 [5] NCCL INFO Channel 01/0 : 1[5] -> 2[6] via P2P/IPC
+ip-26-0-170-132:3330190:3330670 [7] NCCL INFO Channel 00/0 : 3[7] -> 0[4] via P2P/IPC
+ip-26-0-170-132:3330185:3330664 [2] NCCL INFO Channel 12/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-170-132:3330183:3330662 [0] NCCL INFO Channel 15/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-170-132:3330184:3330665 [1] NCCL INFO Channel 09/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-170-132:3330187:3330667 [4] NCCL INFO Channel 03/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-170-132:3330186:3330663 [3] NCCL INFO Channel 09/0 : 3[3] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330190:3330670 [7] NCCL INFO Channel 01/0 : 3[7] -> 0[4] via P2P/IPC
+ip-26-0-170-132:3330188:3330669 [5] NCCL INFO Channel 02/0 : 1[5] -> 2[6] via P2P/IPC
+ip-26-0-170-132:3330185:3330664 [2] NCCL INFO Channel 13/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-170-132:3330183:3330662 [0] NCCL INFO Channel 16/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-170-132:3330184:3330665 [1] NCCL INFO Channel 10/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-170-132:3330186:3330663 [3] NCCL INFO Channel 10/0 : 3[3] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330187:3330667 [4] NCCL INFO Channel 04/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-170-132:3330190:3330670 [7] NCCL INFO Channel 02/0 : 3[7] -> 0[4] via P2P/IPC
+ip-26-0-170-132:3330188:3330669 [5] NCCL INFO Channel 03/0 : 1[5] -> 2[6] via P2P/IPC
+ip-26-0-170-132:3330185:3330664 [2] NCCL INFO Channel 14/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-170-132:3330183:3330662 [0] NCCL INFO Channel 17/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-170-132:3330184:3330665 [1] NCCL INFO Channel 11/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-170-132:3330186:3330663 [3] NCCL INFO Channel 11/0 : 3[3] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330187:3330667 [4] NCCL INFO Channel 05/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-170-132:3330190:3330670 [7] NCCL INFO Channel 03/0 : 3[7] -> 0[4] via P2P/IPC
+ip-26-0-170-132:3330185:3330664 [2] NCCL INFO Channel 15/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-170-132:3330188:3330669 [5] NCCL INFO Channel 04/0 : 1[5] -> 2[6] via P2P/IPC
+ip-26-0-170-132:3330183:3330662 [0] NCCL INFO Channel 18/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-170-132:3330184:3330665 [1] NCCL INFO Channel 12/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-170-132:3330186:3330663 [3] NCCL INFO Channel 12/0 : 3[3] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330187:3330667 [4] NCCL INFO Channel 06/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-170-132:3330190:3330670 [7] NCCL INFO Channel 04/0 : 3[7] -> 0[4] via P2P/IPC
+ip-26-0-170-132:3330185:3330664 [2] NCCL INFO Channel 16/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-170-132:3330188:3330669 [5] NCCL INFO Channel 05/0 : 1[5] -> 2[6] via P2P/IPC
+ip-26-0-170-132:3330183:3330662 [0] NCCL INFO Channel 19/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-170-132:3330184:3330665 [1] NCCL INFO Channel 13/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-170-132:3330186:3330663 [3] NCCL INFO Channel 13/0 : 3[3] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330187:3330667 [4] NCCL INFO Channel 07/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-170-132:3330190:3330670 [7] NCCL INFO Channel 05/0 : 3[7] -> 0[4] via P2P/IPC
+ip-26-0-170-132:3330185:3330664 [2] NCCL INFO Channel 17/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-170-132:3330188:3330669 [5] NCCL INFO Channel 06/0 : 1[5] -> 2[6] via P2P/IPC
+ip-26-0-170-132:3330183:3330662 [0] NCCL INFO Channel 20/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-170-132:3330184:3330665 [1] NCCL INFO Channel 14/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-170-132:3330186:3330663 [3] NCCL INFO Channel 14/0 : 3[3] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330187:3330667 [4] NCCL INFO Channel 08/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-170-132:3330190:3330670 [7] NCCL INFO Channel 06/0 : 3[7] -> 0[4] via P2P/IPC
+ip-26-0-170-132:3330185:3330664 [2] NCCL INFO Channel 18/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-170-132:3330188:3330669 [5] NCCL INFO Channel 07/0 : 1[5] -> 2[6] via P2P/IPC
+ip-26-0-170-132:3330183:3330662 [0] NCCL INFO Channel 21/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-170-132:3330186:3330663 [3] NCCL INFO Channel 15/0 : 3[3] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330184:3330665 [1] NCCL INFO Channel 15/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-170-132:3330187:3330667 [4] NCCL INFO Channel 09/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-170-132:3330190:3330670 [7] NCCL INFO Channel 07/0 : 3[7] -> 0[4] via P2P/IPC
+ip-26-0-170-132:3330185:3330664 [2] NCCL INFO Channel 19/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-170-132:3330188:3330669 [5] NCCL INFO Channel 08/0 : 1[5] -> 2[6] via P2P/IPC
+ip-26-0-170-132:3330183:3330662 [0] NCCL INFO Channel 22/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-170-132:3330189:3330668 [6] NCCL INFO Channel 00/0 : 2[6] -> 3[7] via P2P/IPC
+ip-26-0-170-132:3330186:3330663 [3] NCCL INFO Channel 16/0 : 3[3] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330184:3330665 [1] NCCL INFO Channel 16/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-170-132:3330187:3330667 [4] NCCL INFO Channel 10/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-170-132:3330190:3330670 [7] NCCL INFO Channel 08/0 : 3[7] -> 0[4] via P2P/IPC
+ip-26-0-170-132:3330185:3330664 [2] NCCL INFO Channel 20/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-170-132:3330188:3330669 [5] NCCL INFO Channel 09/0 : 1[5] -> 2[6] via P2P/IPC
+ip-26-0-170-132:3330183:3330662 [0] NCCL INFO Channel 23/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-170-132:3330189:3330668 [6] NCCL INFO Channel 01/0 : 2[6] -> 3[7] via P2P/IPC
+ip-26-0-170-132:3330186:3330663 [3] NCCL INFO Channel 17/0 : 3[3] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330184:3330665 [1] NCCL INFO Channel 17/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-170-132:3330187:3330667 [4] NCCL INFO Channel 11/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-170-132:3330190:3330670 [7] NCCL INFO Channel 09/0 : 3[7] -> 0[4] via P2P/IPC
+ip-26-0-170-132:3330185:3330664 [2] NCCL INFO Channel 21/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-170-132:3330188:3330669 [5] NCCL INFO Channel 10/0 : 1[5] -> 2[6] via P2P/IPC
+ip-26-0-170-132:3330184:3330665 [1] NCCL INFO Channel 18/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-170-132:3330189:3330668 [6] NCCL INFO Channel 02/0 : 2[6] -> 3[7] via P2P/IPC
+ip-26-0-170-132:3330186:3330663 [3] NCCL INFO Channel 18/0 : 3[3] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330187:3330667 [4] NCCL INFO Channel 12/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-170-132:3330190:3330670 [7] NCCL INFO Channel 10/0 : 3[7] -> 0[4] via P2P/IPC
+ip-26-0-170-132:3330185:3330664 [2] NCCL INFO Channel 22/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-170-132:3330188:3330669 [5] NCCL INFO Channel 11/0 : 1[5] -> 2[6] via P2P/IPC
+ip-26-0-170-132:3330189:3330668 [6] NCCL INFO Channel 03/0 : 2[6] -> 3[7] via P2P/IPC
+ip-26-0-170-132:3330184:3330665 [1] NCCL INFO Channel 19/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-170-132:3330187:3330667 [4] NCCL INFO Channel 13/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-170-132:3330186:3330663 [3] NCCL INFO Channel 19/0 : 3[3] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330190:3330670 [7] NCCL INFO Channel 11/0 : 3[7] -> 0[4] via P2P/IPC
+ip-26-0-170-132:3330185:3330664 [2] NCCL INFO Channel 23/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-170-132:3330188:3330669 [5] NCCL INFO Channel 12/0 : 1[5] -> 2[6] via P2P/IPC
+ip-26-0-170-132:3330189:3330668 [6] NCCL INFO Channel 04/0 : 2[6] -> 3[7] via P2P/IPC
+ip-26-0-170-132:3330184:3330665 [1] NCCL INFO Channel 20/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-170-132:3330186:3330663 [3] NCCL INFO Channel 20/0 : 3[3] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330187:3330667 [4] NCCL INFO Channel 14/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-170-132:3330190:3330670 [7] NCCL INFO Channel 12/0 : 3[7] -> 0[4] via P2P/IPC
+ip-26-0-170-132:3330188:3330669 [5] NCCL INFO Channel 13/0 : 1[5] -> 2[6] via P2P/IPC
+ip-26-0-170-132:3330189:3330668 [6] NCCL INFO Channel 05/0 : 2[6] -> 3[7] via P2P/IPC
+ip-26-0-170-132:3330184:3330665 [1] NCCL INFO Channel 21/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-170-132:3330186:3330663 [3] NCCL INFO Channel 21/0 : 3[3] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330187:3330667 [4] NCCL INFO Channel 15/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-170-132:3330190:3330670 [7] NCCL INFO Channel 13/0 : 3[7] -> 0[4] via P2P/IPC
+ip-26-0-170-132:3330188:3330669 [5] NCCL INFO Channel 14/0 : 1[5] -> 2[6] via P2P/IPC
+ip-26-0-170-132:3330189:3330668 [6] NCCL INFO Channel 06/0 : 2[6] -> 3[7] via P2P/IPC
+ip-26-0-170-132:3330184:3330665 [1] NCCL INFO Channel 22/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-170-132:3330186:3330663 [3] NCCL INFO Channel 22/0 : 3[3] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330187:3330667 [4] NCCL INFO Channel 16/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-162-14:2525082:2525548 [7] NCCL INFO Channel 15/0 : 3[7] -> 0[4] via P2P/IPC
+ip-26-0-162-14:2525080:2525550 [5] NCCL INFO Channel 16/0 : 1[5] -> 2[6] via P2P/IPC
+ip-26-0-162-14:2525079:2525546 [4] NCCL INFO Channel 19/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-162-14:2525082:2525548 [7] NCCL INFO Channel 16/0 : 3[7] -> 0[4] via P2P/IPC
+ip-26-0-162-14:2525080:2525550 [5] NCCL INFO Channel 17/0 : 1[5] -> 2[6] via P2P/IPC
+ip-26-0-162-14:2525079:2525546 [4] NCCL INFO Channel 20/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-162-14:2525081:2525549 [6] NCCL INFO Channel 15/0 : 2[6] -> 3[7] via P2P/IPC
+ip-26-0-162-14:2525082:2525548 [7] NCCL INFO Channel 17/0 : 3[7] -> 0[4] via P2P/IPC
+ip-26-0-170-132:3330190:3330670 [7] NCCL INFO Channel 14/0 : 3[7] -> 0[4] via P2P/IPC
+ip-26-0-162-14:2525079:2525546 [4] NCCL INFO Channel 21/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-162-14:2525082:2525548 [7] NCCL INFO Channel 18/0 : 3[7] -> 0[4] via P2P/IPC
+ip-26-0-162-14:2525080:2525550 [5] NCCL INFO Channel 18/0 : 1[5] -> 2[6] via P2P/IPC
+ip-26-0-162-14:2525079:2525546 [4] NCCL INFO Channel 22/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-162-14:2525082:2525548 [7] NCCL INFO Channel 19/0 : 3[7] -> 0[4] via P2P/IPC
+ip-26-0-162-14:2525080:2525550 [5] NCCL INFO Channel 19/0 : 1[5] -> 2[6] via P2P/IPC
+ip-26-0-162-14:2525081:2525549 [6] NCCL INFO Channel 16/0 : 2[6] -> 3[7] via P2P/IPC
+ip-26-0-162-14:2525082:2525548 [7] NCCL INFO Channel 20/0 : 3[7] -> 0[4] via P2P/IPC
+ip-26-0-162-14:2525082:2525548 [7] NCCL INFO Channel 21/0 : 3[7] -> 0[4] via P2P/IPC
+ip-26-0-162-14:2525081:2525549 [6] NCCL INFO Channel 17/0 : 2[6] -> 3[7] via P2P/IPC
+ip-26-0-162-14:2525077:2525553 [2] NCCL INFO Channel 00/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-162-14:2525078:2525554 [3] NCCL INFO Channel 00/0 : 3[3] -> 0[0] via P2P/IPC
+ip-26-0-162-14:2525080:2525550 [5] NCCL INFO Channel 20/0 : 1[5] -> 2[6] via P2P/IPC
+ip-26-0-162-14:2525079:2525546 [4] NCCL INFO Channel 23/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-162-14:2525077:2525553 [2] NCCL INFO Channel 01/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-162-14:2525078:2525554 [3] NCCL INFO Channel 01/0 : 3[3] -> 0[0] via P2P/IPC
+ip-26-0-162-14:2525082:2525548 [7] NCCL INFO Channel 22/0 : 3[7] -> 0[4] via P2P/IPC
+ip-26-0-162-14:2525081:2525549 [6] NCCL INFO Channel 18/0 : 2[6] -> 3[7] via P2P/IPC
+ip-26-0-162-14:2525082:2525548 [7] NCCL INFO Channel 23/0 : 3[7] -> 0[4] via P2P/IPC
+ip-26-0-162-14:2525081:2525549 [6] NCCL INFO Channel 19/0 : 2[6] -> 3[7] via P2P/IPC
+ip-26-0-162-14:2525080:2525550 [5] NCCL INFO Channel 21/0 : 1[5] -> 2[6] via P2P/IPC
+ip-26-0-162-14:2525081:2525549 [6] NCCL INFO Channel 20/0 : 2[6] -> 3[7] via P2P/IPC
+ip-26-0-162-14:2525080:2525550 [5] NCCL INFO Channel 22/0 : 1[5] -> 2[6] via P2P/IPC
+ip-26-0-162-14:2525075:2525551 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-162-14:2525081:2525549 [6] NCCL INFO Channel 21/0 : 2[6] -> 3[7] via P2P/IPC
+ip-26-0-162-14:2525076:2525552 [1] NCCL INFO Channel 00/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-162-14:2525077:2525553 [2] NCCL INFO Channel 02/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-162-14:2525080:2525550 [5] NCCL INFO Channel 23/0 : 1[5] -> 2[6] via P2P/IPC
+ip-26-0-162-14:2525075:2525551 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-162-14:2525078:2525554 [3] NCCL INFO Channel 02/0 : 3[3] -> 0[0] via P2P/IPC
+ip-26-0-162-14:2525081:2525549 [6] NCCL INFO Channel 22/0 : 2[6] -> 3[7] via P2P/IPC
+ip-26-0-162-14:2525076:2525552 [1] NCCL INFO Channel 01/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-162-14:2525077:2525553 [2] NCCL INFO Channel 03/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-162-14:2525075:2525551 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-162-14:2525078:2525554 [3] NCCL INFO Channel 03/0 : 3[3] -> 0[0] via P2P/IPC
+ip-26-0-162-14:2525081:2525549 [6] NCCL INFO Channel 23/0 : 2[6] -> 3[7] via P2P/IPC
+ip-26-0-162-14:2525076:2525552 [1] NCCL INFO Channel 02/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-162-14:2525077:2525553 [2] NCCL INFO Channel 04/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-162-14:2525075:2525551 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-162-14:2525078:2525554 [3] NCCL INFO Channel 04/0 : 3[3] -> 0[0] via P2P/IPC
+ip-26-0-162-14:2525076:2525552 [1] NCCL INFO Channel 03/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-162-14:2525077:2525553 [2] NCCL INFO Channel 05/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-162-14:2525075:2525551 [0] NCCL INFO Channel 04/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-162-14:2525078:2525554 [3] NCCL INFO Channel 05/0 : 3[3] -> 0[0] via P2P/IPC
+ip-26-0-162-14:2525076:2525552 [1] NCCL INFO Channel 04/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-162-14:2525077:2525553 [2] NCCL INFO Channel 06/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-170-132:3330189:3330668 [6] NCCL INFO Channel 07/0 : 2[6] -> 3[7] via P2P/IPC
+ip-26-0-170-132:3330186:3330663 [3] NCCL INFO Channel 23/0 : 3[3] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330188:3330669 [5] NCCL INFO Channel 15/0 : 1[5] -> 2[6] via P2P/IPC
+ip-26-0-170-132:3330184:3330665 [1] NCCL INFO Channel 23/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-162-14:2525076:2525552 [1] NCCL INFO Channel 05/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-170-132:3330190:3330670 [7] NCCL INFO Channel 15/0 : 3[7] -> 0[4] via P2P/IPC
+ip-26-0-170-132:3330187:3330667 [4] NCCL INFO Channel 17/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-162-14:2525078:2525554 [3] NCCL INFO Channel 06/0 : 3[3] -> 0[0] via P2P/IPC
+ip-26-0-162-14:2525077:2525553 [2] NCCL INFO Channel 07/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-163-58:934726:935189 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-163-58:934726:935189 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-170-132:3330188:3330669 [5] NCCL INFO Channel 16/0 : 1[5] -> 2[6] via P2P/IPC
+ip-26-0-162-14:2525076:2525552 [1] NCCL INFO Channel 06/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-162-14:2525078:2525554 [3] NCCL INFO Channel 07/0 : 3[3] -> 0[0] via P2P/IPC
+ip-26-0-162-14:2525077:2525553 [2] NCCL INFO Channel 08/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-170-132:3330188:3330669 [5] NCCL INFO Channel 17/0 : 1[5] -> 2[6] via P2P/IPC
+ip-26-0-162-14:2525075:2525551 [0] NCCL INFO Channel 05/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-162-14:2525076:2525552 [1] NCCL INFO Channel 07/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-170-132:3330189:3330668 [6] NCCL INFO Channel 08/0 : 2[6] -> 3[7] via P2P/IPC
+ip-26-0-162-14:2525078:2525554 [3] NCCL INFO Channel 08/0 : 3[3] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330188:3330669 [5] NCCL INFO Channel 18/0 : 1[5] -> 2[6] via P2P/IPC
+ip-26-0-162-14:2525077:2525553 [2] NCCL INFO Channel 09/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-170-132:3330190:3330670 [7] NCCL INFO Channel 16/0 : 3[7] -> 0[4] via P2P/IPC
+ip-26-0-162-14:2525075:2525551 [0] NCCL INFO Channel 06/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-170-132:3330187:3330667 [4] NCCL INFO Channel 18/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-162-14:2525076:2525552 [1] NCCL INFO Channel 08/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-170-132:3330187:3330667 [4] NCCL INFO Channel 19/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-170-31:3197180:3197763 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-170-31:3197180:3197763 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-162-14:2525078:2525554 [3] NCCL INFO Channel 09/0 : 3[3] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330188:3330669 [5] NCCL INFO Channel 19/0 : 1[5] -> 2[6] via P2P/IPC
+ip-26-0-162-14:2525077:2525553 [2] NCCL INFO Channel 10/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-162-14:2525075:2525551 [0] NCCL INFO Channel 07/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-162-14:2525076:2525552 [1] NCCL INFO Channel 09/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-162-14:2525078:2525554 [3] NCCL INFO Channel 10/0 : 3[3] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330189:3330668 [6] NCCL INFO Channel 09/0 : 2[6] -> 3[7] via P2P/IPC
+ip-26-0-170-132:3330187:3330667 [4] NCCL INFO Channel 20/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-162-14:2525077:2525553 [2] NCCL INFO Channel 11/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-170-132:3330190:3330670 [7] NCCL INFO Channel 17/0 : 3[7] -> 0[4] via P2P/IPC
+ip-26-0-162-14:2525075:2525551 [0] NCCL INFO Channel 08/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-170-132:3330189:3330668 [6] NCCL INFO Channel 10/0 : 2[6] -> 3[7] via P2P/IPC
+ip-26-0-170-132:3330188:3330669 [5] NCCL INFO Channel 20/0 : 1[5] -> 2[6] via P2P/IPC
+ip-26-0-170-132:3330187:3330667 [4] NCCL INFO Channel 21/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-170-132:3330190:3330670 [7] NCCL INFO Channel 18/0 : 3[7] -> 0[4] via P2P/IPC
+ip-26-0-162-14:2525076:2525552 [1] NCCL INFO Channel 10/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-162-14:2525078:2525554 [3] NCCL INFO Channel 11/0 : 3[3] -> 0[0] via P2P/IPC
+ip-26-0-163-58:934721:935193 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-163-58:934721:935193 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-162-14:2525077:2525553 [2] NCCL INFO Channel 12/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-170-132:3330188:3330669 [5] NCCL INFO Channel 21/0 : 1[5] -> 2[6] via P2P/IPC
+ip-26-0-170-132:3330189:3330668 [6] NCCL INFO Channel 11/0 : 2[6] -> 3[7] via P2P/IPC
+ip-26-0-162-14:2525075:2525551 [0] NCCL INFO Channel 09/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-170-132:3330187:3330667 [4] NCCL INFO Channel 22/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-170-132:3330190:3330670 [7] NCCL INFO Channel 19/0 : 3[7] -> 0[4] via P2P/IPC
+ip-26-0-162-14:2525076:2525552 [1] NCCL INFO Channel 11/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-162-14:2525075:2525551 [0] NCCL INFO Channel 10/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-170-132:3330189:3330668 [6] NCCL INFO Channel 12/0 : 2[6] -> 3[7] via P2P/IPC
+ip-26-0-170-132:3330188:3330669 [5] NCCL INFO Channel 22/0 : 1[5] -> 2[6] via P2P/IPC
+ip-26-0-170-132:3330187:3330667 [4] NCCL INFO Channel 23/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-162-14:2525078:2525554 [3] NCCL INFO Channel 12/0 : 3[3] -> 0[0] via P2P/IPC
+ip-26-0-162-14:2525077:2525553 [2] NCCL INFO Channel 13/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-170-132:3330190:3330670 [7] NCCL INFO Channel 20/0 : 3[7] -> 0[4] via P2P/IPC
+ip-26-0-170-132:3330188:3330669 [5] NCCL INFO Channel 23/0 : 1[5] -> 2[6] via P2P/IPC
+ip-26-0-170-132:3330189:3330668 [6] NCCL INFO Channel 13/0 : 2[6] -> 3[7] via P2P/IPC
+ip-26-0-162-14:2525078:2525554 [3] NCCL INFO Channel 13/0 : 3[3] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330190:3330670 [7] NCCL INFO Channel 21/0 : 3[7] -> 0[4] via P2P/IPC
+ip-26-0-162-14:2525078:2525554 [3] NCCL INFO Channel 14/0 : 3[3] -> 0[0] via P2P/IPC
+ip-26-0-162-14:2525076:2525552 [1] NCCL INFO Channel 12/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-162-14:2525077:2525553 [2] NCCL INFO Channel 14/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-170-132:3330189:3330668 [6] NCCL INFO Channel 14/0 : 2[6] -> 3[7] via P2P/IPC
+ip-26-0-162-14:2525078:2525554 [3] NCCL INFO Channel 15/0 : 3[3] -> 0[0] via P2P/IPC
+ip-26-0-162-14:2525075:2525551 [0] NCCL INFO Channel 11/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-170-132:3330190:3330670 [7] NCCL INFO Channel 22/0 : 3[7] -> 0[4] via P2P/IPC
+ip-26-0-162-14:2525076:2525552 [1] NCCL INFO Channel 13/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-162-14:2525077:2525553 [2] NCCL INFO Channel 15/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-163-58:934727:935188 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-163-58:934727:935188 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-170-132:3330189:3330668 [6] NCCL INFO Channel 15/0 : 2[6] -> 3[7] via P2P/IPC
+ip-26-0-170-132:3330190:3330670 [7] NCCL INFO Channel 23/0 : 3[7] -> 0[4] via P2P/IPC
+ip-26-0-162-14:2525078:2525554 [3] NCCL INFO Channel 16/0 : 3[3] -> 0[0] via P2P/IPC
+ip-26-0-162-14:2525075:2525551 [0] NCCL INFO Channel 12/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-162-14:2525076:2525552 [1] NCCL INFO Channel 14/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-162-14:2525077:2525553 [2] NCCL INFO Channel 16/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-170-132:3330189:3330668 [6] NCCL INFO Channel 16/0 : 2[6] -> 3[7] via P2P/IPC
+ip-26-0-170-31:3197187:3197764 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-170-31:3197187:3197764 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-162-14:2525078:2525554 [3] NCCL INFO Channel 17/0 : 3[3] -> 0[0] via P2P/IPC
+ip-26-0-162-14:2525075:2525551 [0] NCCL INFO Channel 13/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-162-14:2525076:2525552 [1] NCCL INFO Channel 15/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-162-14:2525077:2525553 [2] NCCL INFO Channel 17/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-170-132:3330189:3330668 [6] NCCL INFO Channel 17/0 : 2[6] -> 3[7] via P2P/IPC
+ip-26-0-162-14:2525075:2525551 [0] NCCL INFO Channel 14/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-162-14:2525076:2525552 [1] NCCL INFO Channel 16/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-162-14:2525078:2525554 [3] NCCL INFO Channel 18/0 : 3[3] -> 0[0] via P2P/IPC
+ip-26-0-162-14:2525075:2525551 [0] NCCL INFO Channel 15/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-170-132:3330189:3330668 [6] NCCL INFO Channel 18/0 : 2[6] -> 3[7] via P2P/IPC
+ip-26-0-162-14:2525077:2525553 [2] NCCL INFO Channel 18/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-162-14:2525076:2525552 [1] NCCL INFO Channel 17/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-163-58:934723:935194 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-163-58:934723:935194 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-162-14:2525078:2525554 [3] NCCL INFO Channel 19/0 : 3[3] -> 0[0] via P2P/IPC
+ip-26-0-162-14:2525075:2525551 [0] NCCL INFO Channel 16/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-162-14:2525077:2525553 [2] NCCL INFO Channel 19/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-162-14:2525076:2525552 [1] NCCL INFO Channel 18/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-170-31:3197184:3197762 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-170-31:3197184:3197762 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-162-14:2525078:2525554 [3] NCCL INFO Channel 20/0 : 3[3] -> 0[0] via P2P/IPC
+ip-26-0-162-14:2525076:2525552 [1] NCCL INFO Channel 19/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-162-14:2525075:2525551 [0] NCCL INFO Channel 17/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-170-132:3330189:3330668 [6] NCCL INFO Channel 19/0 : 2[6] -> 3[7] via P2P/IPC
+ip-26-0-162-14:2525077:2525553 [2] NCCL INFO Channel 20/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-162-14:2525078:2525554 [3] NCCL INFO Channel 21/0 : 3[3] -> 0[0] via P2P/IPC
+ip-26-0-163-58:934724:935186 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-163-58:934724:935186 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-170-31:3197185:3197767 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-170-31:3197185:3197767 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-170-132:3330189:3330668 [6] NCCL INFO Channel 20/0 : 2[6] -> 3[7] via P2P/IPC
+ip-26-0-162-14:2525078:2525554 [3] NCCL INFO Channel 22/0 : 3[3] -> 0[0] via P2P/IPC
+ip-26-0-162-14:2525075:2525551 [0] NCCL INFO Channel 18/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-163-58:934725:935190 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-163-58:934725:935190 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-162-14:2525077:2525553 [2] NCCL INFO Channel 21/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-162-14:2525076:2525552 [1] NCCL INFO Channel 20/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-170-132:3330189:3330668 [6] NCCL INFO Channel 21/0 : 2[6] -> 3[7] via P2P/IPC
+ip-26-0-170-31:3197182:3197769 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-170-31:3197182:3197769 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-162-14:2525077:2525553 [2] NCCL INFO Channel 22/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-170-31:3197186:3197766 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-170-31:3197186:3197766 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-170-132:3330189:3330668 [6] NCCL INFO Channel 22/0 : 2[6] -> 3[7] via P2P/IPC
+ip-26-0-162-14:2525078:2525554 [3] NCCL INFO Channel 23/0 : 3[3] -> 0[0] via P2P/IPC
+ip-26-0-163-58:934725:935190 [5] NCCL INFO Trees [0] 2/-1/-1->1->0 [1] 2/-1/-1->1->0 [2] 2/-1/-1->1->0 [3] 2/-1/-1->1->0 [4] 2/-1/-1->1->0 [5] 2/-1/-1->1->0 [6] 2/-1/-1->1->0 [7] 2/-1/-1->1->0 [8] 2/-1/-1->1->0 [9] 2/-1/-1->1->0 [10] 2/-1/-1->1->0 [11] 2/-1/-1->1->0 [12] 2/-1/-1->1->0 [13] 2/-1/-1->1->0 [14] 2/-1/-1->1->0 [15] 2/-1/-1->1->0 [16] 2/-1/-1->1->0 [17] 2/-1/-1->1->0 [18] 2/-1/-1->1->0 [19] 2/-1/-1->1->0 [20] 2/-1/-1->1->0 [21] 2/-1/-1->1->0 [22] 2/-1/-1->1->0 [23] 2/-1/-1->1->0
+ip-26-0-163-58:934724:935186 [4] NCCL INFO Channel 00/24 :    0   1   2   3
+ip-26-0-163-58:934725:935190 [5] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-163-58:934724:935186 [4] NCCL INFO Channel 01/24 :    0   1   2   3
+ip-26-0-163-58:934724:935186 [4] NCCL INFO Channel 02/24 :    0   1   2   3
+ip-26-0-163-58:934724:935186 [4] NCCL INFO Channel 03/24 :    0   1   2   3
+ip-26-0-163-58:934724:935186 [4] NCCL INFO Channel 04/24 :    0   1   2   3
+ip-26-0-163-58:934724:935186 [4] NCCL INFO Channel 05/24 :    0   1   2   3
+ip-26-0-162-14:2525075:2525551 [0] NCCL INFO Channel 19/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-163-58:934724:935186 [4] NCCL INFO Channel 06/24 :    0   1   2   3
+ip-26-0-163-58:934724:935186 [4] NCCL INFO Channel 07/24 :    0   1   2   3
+ip-26-0-163-58:934726:935189 [6] NCCL INFO Trees [0] 3/-1/-1->2->1 [1] 3/-1/-1->2->1 [2] 3/-1/-1->2->1 [3] 3/-1/-1->2->1 [4] 3/-1/-1->2->1 [5] 3/-1/-1->2->1 [6] 3/-1/-1->2->1 [7] 3/-1/-1->2->1 [8] 3/-1/-1->2->1 [9] 3/-1/-1->2->1 [10] 3/-1/-1->2->1 [11] 3/-1/-1->2->1 [12] 3/-1/-1->2->1 [13] 3/-1/-1->2->1 [14] 3/-1/-1->2->1 [15] 3/-1/-1->2->1 [16] 3/-1/-1->2->1 [17] 3/-1/-1->2->1 [18] 3/-1/-1->2->1 [19] 3/-1/-1->2->1 [20] 3/-1/-1->2->1 [21] 3/-1/-1->2->1 [22] 3/-1/-1->2->1 [23] 3/-1/-1->2->1
+ip-26-0-163-58:934727:935188 [7] NCCL INFO Trees [0] -1/-1/-1->3->2 [1] -1/-1/-1->3->2 [2] -1/-1/-1->3->2 [3] -1/-1/-1->3->2 [4] -1/-1/-1->3->2 [5] -1/-1/-1->3->2 [6] -1/-1/-1->3->2 [7] -1/-1/-1->3->2 [8] -1/-1/-1->3->2 [9] -1/-1/-1->3->2 [10] -1/-1/-1->3->2 [11] -1/-1/-1->3->2 [12] -1/-1/-1->3->2 [13] -1/-1/-1->3->2 [14] -1/-1/-1->3->2 [15] -1/-1/-1->3->2 [16] -1/-1/-1->3->2 [17] -1/-1/-1->3->2 [18] -1/-1/-1->3->2 [19] -1/-1/-1->3->2 [20] -1/-1/-1->3->2 [21] -1/-1/-1->3->2 [22] -1/-1/-1->3->2 [23] -1/-1/-1->3->2
+ip-26-0-163-58:934724:935186 [4] NCCL INFO Channel 08/24 :    0   1   2   3
+ip-26-0-163-58:934726:935189 [6] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-163-58:934727:935188 [7] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-163-58:934724:935186 [4] NCCL INFO Channel 09/24 :    0   1   2   3
+ip-26-0-163-58:934724:935186 [4] NCCL INFO Channel 10/24 :    0   1   2   3
+ip-26-0-163-58:934724:935186 [4] NCCL INFO Channel 11/24 :    0   1   2   3
+ip-26-0-163-58:934724:935186 [4] NCCL INFO Channel 12/24 :    0   1   2   3
+ip-26-0-163-58:934724:935186 [4] NCCL INFO Channel 13/24 :    0   1   2   3
+ip-26-0-163-58:934724:935186 [4] NCCL INFO Channel 14/24 :    0   1   2   3
+ip-26-0-163-58:934724:935186 [4] NCCL INFO Channel 15/24 :    0   1   2   3
+ip-26-0-163-58:934724:935186 [4] NCCL INFO Channel 16/24 :    0   1   2   3
+ip-26-0-163-58:934724:935186 [4] NCCL INFO Channel 17/24 :    0   1   2   3
+ip-26-0-163-58:934724:935186 [4] NCCL INFO Channel 18/24 :    0   1   2   3
+ip-26-0-163-58:934724:935186 [4] NCCL INFO Channel 19/24 :    0   1   2   3
+ip-26-0-163-58:934724:935186 [4] NCCL INFO Channel 20/24 :    0   1   2   3
+ip-26-0-163-58:934724:935186 [4] NCCL INFO Channel 21/24 :    0   1   2   3
+ip-26-0-163-58:934724:935186 [4] NCCL INFO Channel 22/24 :    0   1   2   3
+ip-26-0-163-58:934724:935186 [4] NCCL INFO Channel 23/24 :    0   1   2   3
+ip-26-0-163-58:934724:935186 [4] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] 1/-1/-1->0->-1 [7] 1/-1/-1->0->-1 [8] 1/-1/-1->0->-1 [9] 1/-1/-1->0->-1 [10] 1/-1/-1->0->-1 [11] 1/-1/-1->0->-1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] 1/-1/-1->0->-1 [19] 1/-1/-1->0->-1 [20] 1/-1/-1->0->-1 [21] 1/-1/-1->0->-1 [22] 1/-1/-1->0->-1 [23] 1/-1/-1->0->-1
+ip-26-0-163-58:934724:935186 [4] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-162-14:2525077:2525553 [2] NCCL INFO Channel 23/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-170-31:3197181:3197768 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-170-31:3197181:3197768 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-162-14:2525076:2525552 [1] NCCL INFO Channel 21/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-170-132:3330189:3330668 [6] NCCL INFO Channel 23/0 : 2[6] -> 3[7] via P2P/IPC
+ip-26-0-163-58:934720:935191 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-163-58:934720:935191 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-170-31:3197183:3197765 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-170-31:3197183:3197765 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-162-14:2525075:2525551 [0] NCCL INFO Channel 20/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-170-31:3197184:3197762 [4] NCCL INFO Channel 00/24 :    0   1   2   3
+ip-26-0-170-31:3197186:3197766 [6] NCCL INFO Trees [0] 3/-1/-1->2->1 [1] 3/-1/-1->2->1 [2] 3/-1/-1->2->1 [3] 3/-1/-1->2->1 [4] 3/-1/-1->2->1 [5] 3/-1/-1->2->1 [6] 3/-1/-1->2->1 [7] 3/-1/-1->2->1 [8] 3/-1/-1->2->1 [9] 3/-1/-1->2->1 [10] 3/-1/-1->2->1 [11] 3/-1/-1->2->1 [12] 3/-1/-1->2->1 [13] 3/-1/-1->2->1 [14] 3/-1/-1->2->1 [15] 3/-1/-1->2->1 [16] 3/-1/-1->2->1 [17] 3/-1/-1->2->1 [18] 3/-1/-1->2->1 [19] 3/-1/-1->2->1 [20] 3/-1/-1->2->1 [21] 3/-1/-1->2->1 [22] 3/-1/-1->2->1 [23] 3/-1/-1->2->1
+ip-26-0-170-31:3197184:3197762 [4] NCCL INFO Channel 01/24 :    0   1   2   3
+ip-26-0-170-31:3197184:3197762 [4] NCCL INFO Channel 02/24 :    0   1   2   3
+ip-26-0-170-31:3197186:3197766 [6] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-170-31:3197184:3197762 [4] NCCL INFO Channel 03/24 :    0   1   2   3
+ip-26-0-170-31:3197185:3197767 [5] NCCL INFO Trees [0] 2/-1/-1->1->0 [1] 2/-1/-1->1->0 [2] 2/-1/-1->1->0 [3] 2/-1/-1->1->0 [4] 2/-1/-1->1->0 [5] 2/-1/-1->1->0 [6] 2/-1/-1->1->0 [7] 2/-1/-1->1->0 [8] 2/-1/-1->1->0 [9] 2/-1/-1->1->0 [10] 2/-1/-1->1->0 [11] 2/-1/-1->1->0 [12] 2/-1/-1->1->0 [13] 2/-1/-1->1->0 [14] 2/-1/-1->1->0 [15] 2/-1/-1->1->0 [16] 2/-1/-1->1->0 [17] 2/-1/-1->1->0 [18] 2/-1/-1->1->0 [19] 2/-1/-1->1->0 [20] 2/-1/-1->1->0 [21] 2/-1/-1->1->0 [22] 2/-1/-1->1->0 [23] 2/-1/-1->1->0
+ip-26-0-170-31:3197184:3197762 [4] NCCL INFO Channel 04/24 :    0   1   2   3
+ip-26-0-170-31:3197185:3197767 [5] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-170-31:3197184:3197762 [4] NCCL INFO Channel 05/24 :    0   1   2   3
+ip-26-0-170-31:3197187:3197764 [7] NCCL INFO Trees [0] -1/-1/-1->3->2 [1] -1/-1/-1->3->2 [2] -1/-1/-1->3->2 [3] -1/-1/-1->3->2 [4] -1/-1/-1->3->2 [5] -1/-1/-1->3->2 [6] -1/-1/-1->3->2 [7] -1/-1/-1->3->2 [8] -1/-1/-1->3->2 [9] -1/-1/-1->3->2 [10] -1/-1/-1->3->2 [11] -1/-1/-1->3->2 [12] -1/-1/-1->3->2 [13] -1/-1/-1->3->2 [14] -1/-1/-1->3->2 [15] -1/-1/-1->3->2 [16] -1/-1/-1->3->2 [17] -1/-1/-1->3->2 [18] -1/-1/-1->3->2 [19] -1/-1/-1->3->2 [20] -1/-1/-1->3->2 [21] -1/-1/-1->3->2 [22] -1/-1/-1->3->2 [23] -1/-1/-1->3->2
+ip-26-0-170-31:3197184:3197762 [4] NCCL INFO Channel 06/24 :    0   1   2   3
+ip-26-0-170-31:3197184:3197762 [4] NCCL INFO Channel 07/24 :    0   1   2   3
+ip-26-0-170-31:3197187:3197764 [7] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-170-31:3197184:3197762 [4] NCCL INFO Channel 08/24 :    0   1   2   3
+ip-26-0-170-31:3197184:3197762 [4] NCCL INFO Channel 09/24 :    0   1   2   3
+ip-26-0-170-31:3197184:3197762 [4] NCCL INFO Channel 10/24 :    0   1   2   3
+ip-26-0-170-31:3197184:3197762 [4] NCCL INFO Channel 11/24 :    0   1   2   3
+ip-26-0-170-31:3197184:3197762 [4] NCCL INFO Channel 12/24 :    0   1   2   3
+ip-26-0-170-31:3197184:3197762 [4] NCCL INFO Channel 13/24 :    0   1   2   3
+ip-26-0-170-31:3197184:3197762 [4] NCCL INFO Channel 14/24 :    0   1   2   3
+ip-26-0-170-31:3197184:3197762 [4] NCCL INFO Channel 15/24 :    0   1   2   3
+ip-26-0-170-31:3197184:3197762 [4] NCCL INFO Channel 16/24 :    0   1   2   3
+ip-26-0-170-31:3197184:3197762 [4] NCCL INFO Channel 17/24 :    0   1   2   3
+ip-26-0-170-31:3197184:3197762 [4] NCCL INFO Channel 18/24 :    0   1   2   3
+ip-26-0-170-31:3197184:3197762 [4] NCCL INFO Channel 19/24 :    0   1   2   3
+ip-26-0-170-31:3197184:3197762 [4] NCCL INFO Channel 20/24 :    0   1   2   3
+ip-26-0-170-31:3197184:3197762 [4] NCCL INFO Channel 21/24 :    0   1   2   3
+ip-26-0-170-31:3197184:3197762 [4] NCCL INFO Channel 22/24 :    0   1   2   3
+ip-26-0-170-31:3197184:3197762 [4] NCCL INFO Channel 23/24 :    0   1   2   3
+ip-26-0-170-31:3197184:3197762 [4] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] 1/-1/-1->0->-1 [7] 1/-1/-1->0->-1 [8] 1/-1/-1->0->-1 [9] 1/-1/-1->0->-1 [10] 1/-1/-1->0->-1 [11] 1/-1/-1->0->-1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] 1/-1/-1->0->-1 [19] 1/-1/-1->0->-1 [20] 1/-1/-1->0->-1 [21] 1/-1/-1->0->-1 [22] 1/-1/-1->0->-1 [23] 1/-1/-1->0->-1
+ip-26-0-170-31:3197184:3197762 [4] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-162-14:2525076:2525552 [1] NCCL INFO Channel 22/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-162-14:2525075:2525551 [0] NCCL INFO Channel 21/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-163-58:934725:935190 [5] NCCL INFO Channel 00/0 : 1[5] -> 2[6] via P2P/IPC
+ip-26-0-170-31:3197183:3197765 [3] NCCL INFO Trees [0] -1/-1/-1->3->2 [1] -1/-1/-1->3->2 [2] -1/-1/-1->3->2 [3] -1/-1/-1->3->2 [4] -1/-1/-1->3->2 [5] -1/-1/-1->3->2 [6] -1/-1/-1->3->2 [7] -1/-1/-1->3->2 [8] -1/-1/-1->3->2 [9] -1/-1/-1->3->2 [10] -1/-1/-1->3->2 [11] -1/-1/-1->3->2 [12] -1/-1/-1->3->2 [13] -1/-1/-1->3->2 [14] -1/-1/-1->3->2 [15] -1/-1/-1->3->2 [16] -1/-1/-1->3->2 [17] -1/-1/-1->3->2 [18] -1/-1/-1->3->2 [19] -1/-1/-1->3->2 [20] -1/-1/-1->3->2 [21] -1/-1/-1->3->2 [22] -1/-1/-1->3->2 [23] -1/-1/-1->3->2
+ip-26-0-170-31:3197180:3197763 [0] NCCL INFO Channel 00/24 :    0   1   2   3
+ip-26-0-170-31:3197183:3197765 [3] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-170-31:3197180:3197763 [0] NCCL INFO Channel 01/24 :    0   1   2   3
+ip-26-0-170-31:3197180:3197763 [0] NCCL INFO Channel 02/24 :    0   1   2   3
+ip-26-0-170-31:3197180:3197763 [0] NCCL INFO Channel 03/24 :    0   1   2   3
+ip-26-0-170-31:3197180:3197763 [0] NCCL INFO Channel 04/24 :    0   1   2   3
+ip-26-0-170-31:3197180:3197763 [0] NCCL INFO Channel 05/24 :    0   1   2   3
+ip-26-0-170-31:3197181:3197768 [1] NCCL INFO Trees [0] 2/-1/-1->1->0 [1] 2/-1/-1->1->0 [2] 2/-1/-1->1->0 [3] 2/-1/-1->1->0 [4] 2/-1/-1->1->0 [5] 2/-1/-1->1->0 [6] 2/-1/-1->1->0 [7] 2/-1/-1->1->0 [8] 2/-1/-1->1->0 [9] 2/-1/-1->1->0 [10] 2/-1/-1->1->0 [11] 2/-1/-1->1->0 [12] 2/-1/-1->1->0 [13] 2/-1/-1->1->0 [14] 2/-1/-1->1->0 [15] 2/-1/-1->1->0 [16] 2/-1/-1->1->0 [17] 2/-1/-1->1->0 [18] 2/-1/-1->1->0 [19] 2/-1/-1->1->0 [20] 2/-1/-1->1->0 [21] 2/-1/-1->1->0 [22] 2/-1/-1->1->0 [23] 2/-1/-1->1->0
+ip-26-0-170-31:3197180:3197763 [0] NCCL INFO Channel 06/24 :    0   1   2   3
+ip-26-0-170-31:3197182:3197769 [2] NCCL INFO Trees [0] 3/-1/-1->2->1 [1] 3/-1/-1->2->1 [2] 3/-1/-1->2->1 [3] 3/-1/-1->2->1 [4] 3/-1/-1->2->1 [5] 3/-1/-1->2->1 [6] 3/-1/-1->2->1 [7] 3/-1/-1->2->1 [8] 3/-1/-1->2->1 [9] 3/-1/-1->2->1 [10] 3/-1/-1->2->1 [11] 3/-1/-1->2->1 [12] 3/-1/-1->2->1 [13] 3/-1/-1->2->1 [14] 3/-1/-1->2->1 [15] 3/-1/-1->2->1 [16] 3/-1/-1->2->1 [17] 3/-1/-1->2->1 [18] 3/-1/-1->2->1 [19] 3/-1/-1->2->1 [20] 3/-1/-1->2->1 [21] 3/-1/-1->2->1 [22] 3/-1/-1->2->1 [23] 3/-1/-1->2->1
+ip-26-0-170-31:3197181:3197768 [1] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-170-31:3197180:3197763 [0] NCCL INFO Channel 07/24 :    0   1   2   3
+ip-26-0-170-31:3197180:3197763 [0] NCCL INFO Channel 08/24 :    0   1   2   3
+ip-26-0-170-31:3197182:3197769 [2] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-170-31:3197180:3197763 [0] NCCL INFO Channel 09/24 :    0   1   2   3
+ip-26-0-170-31:3197180:3197763 [0] NCCL INFO Channel 10/24 :    0   1   2   3
+ip-26-0-170-31:3197180:3197763 [0] NCCL INFO Channel 11/24 :    0   1   2   3
+ip-26-0-170-31:3197180:3197763 [0] NCCL INFO Channel 12/24 :    0   1   2   3
+ip-26-0-170-31:3197180:3197763 [0] NCCL INFO Channel 13/24 :    0   1   2   3
+ip-26-0-170-31:3197180:3197763 [0] NCCL INFO Channel 14/24 :    0   1   2   3
+ip-26-0-170-31:3197180:3197763 [0] NCCL INFO Channel 15/24 :    0   1   2   3
+ip-26-0-170-31:3197180:3197763 [0] NCCL INFO Channel 16/24 :    0   1   2   3
+ip-26-0-170-31:3197180:3197763 [0] NCCL INFO Channel 17/24 :    0   1   2   3
+ip-26-0-170-31:3197180:3197763 [0] NCCL INFO Channel 18/24 :    0   1   2   3
+ip-26-0-170-31:3197180:3197763 [0] NCCL INFO Channel 19/24 :    0   1   2   3
+ip-26-0-170-31:3197180:3197763 [0] NCCL INFO Channel 20/24 :    0   1   2   3
+ip-26-0-170-31:3197180:3197763 [0] NCCL INFO Channel 21/24 :    0   1   2   3
+ip-26-0-170-31:3197180:3197763 [0] NCCL INFO Channel 22/24 :    0   1   2   3
+ip-26-0-170-31:3197180:3197763 [0] NCCL INFO Channel 23/24 :    0   1   2   3
+ip-26-0-170-31:3197180:3197763 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] 1/-1/-1->0->-1 [7] 1/-1/-1->0->-1 [8] 1/-1/-1->0->-1 [9] 1/-1/-1->0->-1 [10] 1/-1/-1->0->-1 [11] 1/-1/-1->0->-1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] 1/-1/-1->0->-1 [19] 1/-1/-1->0->-1 [20] 1/-1/-1->0->-1 [21] 1/-1/-1->0->-1 [22] 1/-1/-1->0->-1 [23] 1/-1/-1->0->-1
+ip-26-0-170-31:3197180:3197763 [0] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-163-58:934720:935191 [0] NCCL INFO Channel 00/24 :    0   1   2   3
+ip-26-0-163-58:934720:935191 [0] NCCL INFO Channel 01/24 :    0   1   2   3
+ip-26-0-163-58:934720:935191 [0] NCCL INFO Channel 02/24 :    0   1   2   3
+ip-26-0-163-58:934720:935191 [0] NCCL INFO Channel 03/24 :    0   1   2   3
+ip-26-0-163-58:934720:935191 [0] NCCL INFO Channel 04/24 :    0   1   2   3
+ip-26-0-163-58:934720:935191 [0] NCCL INFO Channel 05/24 :    0   1   2   3
+ip-26-0-163-58:934720:935191 [0] NCCL INFO Channel 06/24 :    0   1   2   3
+ip-26-0-163-58:934720:935191 [0] NCCL INFO Channel 07/24 :    0   1   2   3
+ip-26-0-163-58:934720:935191 [0] NCCL INFO Channel 08/24 :    0   1   2   3
+ip-26-0-163-58:934720:935191 [0] NCCL INFO Channel 09/24 :    0   1   2   3
+ip-26-0-163-58:934720:935191 [0] NCCL INFO Channel 10/24 :    0   1   2   3
+ip-26-0-163-58:934720:935191 [0] NCCL INFO Channel 11/24 :    0   1   2   3
+ip-26-0-163-58:934720:935191 [0] NCCL INFO Channel 12/24 :    0   1   2   3
+ip-26-0-163-58:934720:935191 [0] NCCL INFO Channel 13/24 :    0   1   2   3
+ip-26-0-163-58:934720:935191 [0] NCCL INFO Channel 14/24 :    0   1   2   3
+ip-26-0-163-58:934720:935191 [0] NCCL INFO Channel 15/24 :    0   1   2   3
+ip-26-0-163-58:934720:935191 [0] NCCL INFO Channel 16/24 :    0   1   2   3
+ip-26-0-163-58:934720:935191 [0] NCCL INFO Channel 17/24 :    0   1   2   3
+ip-26-0-163-58:934720:935191 [0] NCCL INFO Channel 18/24 :    0   1   2   3
+ip-26-0-163-58:934720:935191 [0] NCCL INFO Channel 19/24 :    0   1   2   3
+ip-26-0-163-58:934721:935193 [1] NCCL INFO Trees [0] 2/-1/-1->1->0 [1] 2/-1/-1->1->0 [2] 2/-1/-1->1->0 [3] 2/-1/-1->1->0 [4] 2/-1/-1->1->0 [5] 2/-1/-1->1->0 [6] 2/-1/-1->1->0 [7] 2/-1/-1->1->0 [8] 2/-1/-1->1->0 [9] 2/-1/-1->1->0 [10] 2/-1/-1->1->0 [11] 2/-1/-1->1->0 [12] 2/-1/-1->1->0 [13] 2/-1/-1->1->0 [14] 2/-1/-1->1->0 [15] 2/-1/-1->1->0 [16] 2/-1/-1->1->0 [17] 2/-1/-1->1->0 [18] 2/-1/-1->1->0 [19] 2/-1/-1->1->0 [20] 2/-1/-1->1->0 [21] 2/-1/-1->1->0 [22] 2/-1/-1->1->0 [23] 2/-1/-1->1->0
+ip-26-0-163-58:934720:935191 [0] NCCL INFO Channel 20/24 :    0   1   2   3
+ip-26-0-163-58:934720:935191 [0] NCCL INFO Channel 21/24 :    0   1   2   3
+ip-26-0-163-58:934721:935193 [1] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-163-58:934720:935191 [0] NCCL INFO Channel 22/24 :    0   1   2   3
+ip-26-0-163-58:934720:935191 [0] NCCL INFO Channel 23/24 :    0   1   2   3
+ip-26-0-163-58:934722:935192 [2] NCCL INFO Trees [0] 3/-1/-1->2->1 [1] 3/-1/-1->2->1 [2] 3/-1/-1->2->1 [3] 3/-1/-1->2->1 [4] 3/-1/-1->2->1 [5] 3/-1/-1->2->1 [6] 3/-1/-1->2->1 [7] 3/-1/-1->2->1 [8] 3/-1/-1->2->1 [9] 3/-1/-1->2->1 [10] 3/-1/-1->2->1 [11] 3/-1/-1->2->1 [12] 3/-1/-1->2->1 [13] 3/-1/-1->2->1 [14] 3/-1/-1->2->1 [15] 3/-1/-1->2->1 [16] 3/-1/-1->2->1 [17] 3/-1/-1->2->1 [18] 3/-1/-1->2->1 [19] 3/-1/-1->2->1 [20] 3/-1/-1->2->1 [21] 3/-1/-1->2->1 [22] 3/-1/-1->2->1 [23] 3/-1/-1->2->1
+ip-26-0-163-58:934723:935194 [3] NCCL INFO Trees [0] -1/-1/-1->3->2 [1] -1/-1/-1->3->2 [2] -1/-1/-1->3->2 [3] -1/-1/-1->3->2 [4] -1/-1/-1->3->2 [5] -1/-1/-1->3->2 [6] -1/-1/-1->3->2 [7] -1/-1/-1->3->2 [8] -1/-1/-1->3->2 [9] -1/-1/-1->3->2 [10] -1/-1/-1->3->2 [11] -1/-1/-1->3->2 [12] -1/-1/-1->3->2 [13] -1/-1/-1->3->2 [14] -1/-1/-1->3->2 [15] -1/-1/-1->3->2 [16] -1/-1/-1->3->2 [17] -1/-1/-1->3->2 [18] -1/-1/-1->3->2 [19] -1/-1/-1->3->2 [20] -1/-1/-1->3->2 [21] -1/-1/-1->3->2 [22] -1/-1/-1->3->2 [23] -1/-1/-1->3->2
+ip-26-0-163-58:934722:935192 [2] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-163-58:934723:935194 [3] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-163-58:934720:935191 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] 1/-1/-1->0->-1 [7] 1/-1/-1->0->-1 [8] 1/-1/-1->0->-1 [9] 1/-1/-1->0->-1 [10] 1/-1/-1->0->-1 [11] 1/-1/-1->0->-1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] 1/-1/-1->0->-1 [19] 1/-1/-1->0->-1 [20] 1/-1/-1->0->-1 [21] 1/-1/-1->0->-1 [22] 1/-1/-1->0->-1 [23] 1/-1/-1->0->-1
+ip-26-0-163-58:934720:935191 [0] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-163-58:934726:935189 [6] NCCL INFO Channel 00/0 : 2[6] -> 3[7] via P2P/IPC
+ip-26-0-163-58:934727:935188 [7] NCCL INFO Channel 00/0 : 3[7] -> 0[4] via P2P/IPC
+ip-26-0-163-58:934724:935186 [4] NCCL INFO Channel 00/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-163-58:934725:935190 [5] NCCL INFO Channel 01/0 : 1[5] -> 2[6] via P2P/IPC
+ip-26-0-163-58:934727:935188 [7] NCCL INFO Channel 01/0 : 3[7] -> 0[4] via P2P/IPC
+ip-26-0-163-58:934724:935186 [4] NCCL INFO Channel 01/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-163-58:934726:935189 [6] NCCL INFO Channel 01/0 : 2[6] -> 3[7] via P2P/IPC
+ip-26-0-163-58:934725:935190 [5] NCCL INFO Channel 02/0 : 1[5] -> 2[6] via P2P/IPC
+ip-26-0-163-58:934724:935186 [4] NCCL INFO Channel 02/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-163-58:934726:935189 [6] NCCL INFO Channel 02/0 : 2[6] -> 3[7] via P2P/IPC
+ip-26-0-163-58:934727:935188 [7] NCCL INFO Channel 02/0 : 3[7] -> 0[4] via P2P/IPC
+ip-26-0-162-14:2525076:2525552 [1] NCCL INFO Channel 23/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-163-58:934725:935190 [5] NCCL INFO Channel 03/0 : 1[5] -> 2[6] via P2P/IPC
+ip-26-0-170-31:3197185:3197767 [5] NCCL INFO Channel 00/0 : 1[5] -> 2[6] via P2P/IPC
+ip-26-0-170-31:3197187:3197764 [7] NCCL INFO Channel 00/0 : 3[7] -> 0[4] via P2P/IPC
+ip-26-0-163-58:934724:935186 [4] NCCL INFO Channel 03/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-163-58:934726:935189 [6] NCCL INFO Channel 03/0 : 2[6] -> 3[7] via P2P/IPC
+ip-26-0-163-58:934727:935188 [7] NCCL INFO Channel 03/0 : 3[7] -> 0[4] via P2P/IPC
+ip-26-0-163-58:934725:935190 [5] NCCL INFO Channel 04/0 : 1[5] -> 2[6] via P2P/IPC
+ip-26-0-170-31:3197186:3197766 [6] NCCL INFO Channel 00/0 : 2[6] -> 3[7] via P2P/IPC
+ip-26-0-170-31:3197184:3197762 [4] NCCL INFO Channel 00/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-162-14:2525075:2525551 [0] NCCL INFO Channel 22/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-170-31:3197185:3197767 [5] NCCL INFO Channel 01/0 : 1[5] -> 2[6] via P2P/IPC
+ip-26-0-170-31:3197187:3197764 [7] NCCL INFO Channel 01/0 : 3[7] -> 0[4] via P2P/IPC
+ip-26-0-163-58:934724:935186 [4] NCCL INFO Channel 04/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-163-58:934726:935189 [6] NCCL INFO Channel 04/0 : 2[6] -> 3[7] via P2P/IPC
+ip-26-0-163-58:934725:935190 [5] NCCL INFO Channel 05/0 : 1[5] -> 2[6] via P2P/IPC
+ip-26-0-163-58:934727:935188 [7] NCCL INFO Channel 04/0 : 3[7] -> 0[4] via P2P/IPC
+ip-26-0-170-31:3197186:3197766 [6] NCCL INFO Channel 01/0 : 2[6] -> 3[7] via P2P/IPC
+ip-26-0-170-31:3197184:3197762 [4] NCCL INFO Channel 01/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-170-31:3197185:3197767 [5] NCCL INFO Channel 02/0 : 1[5] -> 2[6] via P2P/IPC
+ip-26-0-170-31:3197187:3197764 [7] NCCL INFO Channel 02/0 : 3[7] -> 0[4] via P2P/IPC
+ip-26-0-163-58:934726:935189 [6] NCCL INFO Channel 05/0 : 2[6] -> 3[7] via P2P/IPC
+ip-26-0-163-58:934724:935186 [4] NCCL INFO Channel 05/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-163-58:934725:935190 [5] NCCL INFO Channel 06/0 : 1[5] -> 2[6] via P2P/IPC
+ip-26-0-163-58:934727:935188 [7] NCCL INFO Channel 05/0 : 3[7] -> 0[4] via P2P/IPC
+ip-26-0-170-31:3197186:3197766 [6] NCCL INFO Channel 02/0 : 2[6] -> 3[7] via P2P/IPC
+ip-26-0-170-31:3197184:3197762 [4] NCCL INFO Channel 02/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-170-31:3197185:3197767 [5] NCCL INFO Channel 03/0 : 1[5] -> 2[6] via P2P/IPC
+ip-26-0-163-58:934725:935190 [5] NCCL INFO Channel 07/0 : 1[5] -> 2[6] via P2P/IPC
+ip-26-0-170-31:3197187:3197764 [7] NCCL INFO Channel 03/0 : 3[7] -> 0[4] via P2P/IPC
+ip-26-0-163-58:934727:935188 [7] NCCL INFO Channel 06/0 : 3[7] -> 0[4] via P2P/IPC
+ip-26-0-170-31:3197186:3197766 [6] NCCL INFO Channel 03/0 : 2[6] -> 3[7] via P2P/IPC
+ip-26-0-170-31:3197184:3197762 [4] NCCL INFO Channel 03/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-163-58:934724:935186 [4] NCCL INFO Channel 06/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-163-58:934726:935189 [6] NCCL INFO Channel 06/0 : 2[6] -> 3[7] via P2P/IPC
+ip-26-0-163-58:934725:935190 [5] NCCL INFO Channel 08/0 : 1[5] -> 2[6] via P2P/IPC
+ip-26-0-163-58:934727:935188 [7] NCCL INFO Channel 07/0 : 3[7] -> 0[4] via P2P/IPC
+ip-26-0-170-31:3197185:3197767 [5] NCCL INFO Channel 04/0 : 1[5] -> 2[6] via P2P/IPC
+ip-26-0-170-31:3197187:3197764 [7] NCCL INFO Channel 04/0 : 3[7] -> 0[4] via P2P/IPC
+ip-26-0-170-31:3197186:3197766 [6] NCCL INFO Channel 04/0 : 2[6] -> 3[7] via P2P/IPC
+ip-26-0-170-31:3197184:3197762 [4] NCCL INFO Channel 04/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-163-58:934724:935186 [4] NCCL INFO Channel 07/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-163-58:934726:935189 [6] NCCL INFO Channel 07/0 : 2[6] -> 3[7] via P2P/IPC
+ip-26-0-163-58:934725:935190 [5] NCCL INFO Channel 09/0 : 1[5] -> 2[6] via P2P/IPC
+ip-26-0-163-58:934727:935188 [7] NCCL INFO Channel 08/0 : 3[7] -> 0[4] via P2P/IPC
+ip-26-0-170-31:3197185:3197767 [5] NCCL INFO Channel 05/0 : 1[5] -> 2[6] via P2P/IPC
+ip-26-0-170-31:3197187:3197764 [7] NCCL INFO Channel 05/0 : 3[7] -> 0[4] via P2P/IPC
+ip-26-0-162-14:2525075:2525551 [0] NCCL INFO Channel 23/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-163-58:934724:935186 [4] NCCL INFO Channel 08/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-170-31:3197186:3197766 [6] NCCL INFO Channel 05/0 : 2[6] -> 3[7] via P2P/IPC
+ip-26-0-170-31:3197184:3197762 [4] NCCL INFO Channel 05/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-163-58:934726:935189 [6] NCCL INFO Channel 08/0 : 2[6] -> 3[7] via P2P/IPC
+ip-26-0-163-58:934725:935190 [5] NCCL INFO Channel 10/0 : 1[5] -> 2[6] via P2P/IPC
+ip-26-0-170-31:3197185:3197767 [5] NCCL INFO Channel 06/0 : 1[5] -> 2[6] via P2P/IPC
+ip-26-0-163-58:934727:935188 [7] NCCL INFO Channel 09/0 : 3[7] -> 0[4] via P2P/IPC
+ip-26-0-170-31:3197187:3197764 [7] NCCL INFO Channel 06/0 : 3[7] -> 0[4] via P2P/IPC
+ip-26-0-170-31:3197185:3197767 [5] NCCL INFO Channel 07/0 : 1[5] -> 2[6] via P2P/IPC
+ip-26-0-170-31:3197187:3197764 [7] NCCL INFO Channel 07/0 : 3[7] -> 0[4] via P2P/IPC
+ip-26-0-163-58:934724:935186 [4] NCCL INFO Channel 09/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-163-58:934725:935190 [5] NCCL INFO Channel 11/0 : 1[5] -> 2[6] via P2P/IPC
+ip-26-0-163-58:934726:935189 [6] NCCL INFO Channel 09/0 : 2[6] -> 3[7] via P2P/IPC
+ip-26-0-163-58:934727:935188 [7] NCCL INFO Channel 10/0 : 3[7] -> 0[4] via P2P/IPC
+ip-26-0-170-31:3197184:3197762 [4] NCCL INFO Channel 06/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-170-31:3197185:3197767 [5] NCCL INFO Channel 08/0 : 1[5] -> 2[6] via P2P/IPC
+ip-26-0-170-31:3197187:3197764 [7] NCCL INFO Channel 08/0 : 3[7] -> 0[4] via P2P/IPC
+ip-26-0-163-58:934724:935186 [4] NCCL INFO Channel 10/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-163-58:934726:935189 [6] NCCL INFO Channel 10/0 : 2[6] -> 3[7] via P2P/IPC
+ip-26-0-163-58:934725:935190 [5] NCCL INFO Channel 12/0 : 1[5] -> 2[6] via P2P/IPC
+ip-26-0-163-58:934727:935188 [7] NCCL INFO Channel 11/0 : 3[7] -> 0[4] via P2P/IPC
+ip-26-0-170-31:3197186:3197766 [6] NCCL INFO Channel 06/0 : 2[6] -> 3[7] via P2P/IPC
+ip-26-0-170-31:3197185:3197767 [5] NCCL INFO Channel 09/0 : 1[5] -> 2[6] via P2P/IPC
+ip-26-0-170-31:3197184:3197762 [4] NCCL INFO Channel 07/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-170-31:3197187:3197764 [7] NCCL INFO Channel 09/0 : 3[7] -> 0[4] via P2P/IPC
+ip-26-0-163-58:934724:935186 [4] NCCL INFO Channel 11/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-163-58:934726:935189 [6] NCCL INFO Channel 11/0 : 2[6] -> 3[7] via P2P/IPC
+ip-26-0-163-58:934725:935190 [5] NCCL INFO Channel 13/0 : 1[5] -> 2[6] via P2P/IPC
+ip-26-0-163-58:934727:935188 [7] NCCL INFO Channel 12/0 : 3[7] -> 0[4] via P2P/IPC
+ip-26-0-170-31:3197186:3197766 [6] NCCL INFO Channel 07/0 : 2[6] -> 3[7] via P2P/IPC
+ip-26-0-170-31:3197185:3197767 [5] NCCL INFO Channel 10/0 : 1[5] -> 2[6] via P2P/IPC
+ip-26-0-170-31:3197184:3197762 [4] NCCL INFO Channel 08/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-170-31:3197187:3197764 [7] NCCL INFO Channel 10/0 : 3[7] -> 0[4] via P2P/IPC
+ip-26-0-163-58:934724:935186 [4] NCCL INFO Channel 12/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-163-58:934725:935190 [5] NCCL INFO Channel 14/0 : 1[5] -> 2[6] via P2P/IPC
+ip-26-0-163-58:934726:935189 [6] NCCL INFO Channel 12/0 : 2[6] -> 3[7] via P2P/IPC
+ip-26-0-163-58:934727:935188 [7] NCCL INFO Channel 13/0 : 3[7] -> 0[4] via P2P/IPC
+ip-26-0-170-31:3197186:3197766 [6] NCCL INFO Channel 08/0 : 2[6] -> 3[7] via P2P/IPC
+ip-26-0-170-31:3197185:3197767 [5] NCCL INFO Channel 11/0 : 1[5] -> 2[6] via P2P/IPC
+ip-26-0-170-31:3197184:3197762 [4] NCCL INFO Channel 09/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-170-31:3197187:3197764 [7] NCCL INFO Channel 11/0 : 3[7] -> 0[4] via P2P/IPC
+ip-26-0-163-58:934727:935188 [7] NCCL INFO Channel 14/0 : 3[7] -> 0[4] via P2P/IPC
+ip-26-0-170-31:3197186:3197766 [6] NCCL INFO Channel 09/0 : 2[6] -> 3[7] via P2P/IPC
+ip-26-0-170-31:3197185:3197767 [5] NCCL INFO Channel 12/0 : 1[5] -> 2[6] via P2P/IPC
+ip-26-0-170-31:3197184:3197762 [4] NCCL INFO Channel 10/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-170-31:3197187:3197764 [7] NCCL INFO Channel 12/0 : 3[7] -> 0[4] via P2P/IPC
+ip-26-0-163-58:934724:935186 [4] NCCL INFO Channel 13/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-163-58:934726:935189 [6] NCCL INFO Channel 13/0 : 2[6] -> 3[7] via P2P/IPC
+ip-26-0-170-31:3197185:3197767 [5] NCCL INFO Channel 13/0 : 1[5] -> 2[6] via P2P/IPC
+ip-26-0-170-31:3197186:3197766 [6] NCCL INFO Channel 10/0 : 2[6] -> 3[7] via P2P/IPC
+ip-26-0-170-31:3197184:3197762 [4] NCCL INFO Channel 11/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-170-31:3197187:3197764 [7] NCCL INFO Channel 13/0 : 3[7] -> 0[4] via P2P/IPC
+ip-26-0-163-58:934725:935190 [5] NCCL INFO Channel 15/0 : 1[5] -> 2[6] via P2P/IPC
+ip-26-0-163-58:934727:935188 [7] NCCL INFO Channel 15/0 : 3[7] -> 0[4] via P2P/IPC
+ip-26-0-170-31:3197184:3197762 [4] NCCL INFO Channel 12/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-170-31:3197186:3197766 [6] NCCL INFO Channel 11/0 : 2[6] -> 3[7] via P2P/IPC
+ip-26-0-170-31:3197185:3197767 [5] NCCL INFO Channel 14/0 : 1[5] -> 2[6] via P2P/IPC
+ip-26-0-170-31:3197187:3197764 [7] NCCL INFO Channel 14/0 : 3[7] -> 0[4] via P2P/IPC
+ip-26-0-163-58:934724:935186 [4] NCCL INFO Channel 14/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-163-58:934727:935188 [7] NCCL INFO Channel 16/0 : 3[7] -> 0[4] via P2P/IPC
+ip-26-0-163-58:934724:935186 [4] NCCL INFO Channel 15/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-163-58:934726:935189 [6] NCCL INFO Channel 14/0 : 2[6] -> 3[7] via P2P/IPC
+ip-26-0-170-31:3197184:3197762 [4] NCCL INFO Channel 13/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-170-31:3197186:3197766 [6] NCCL INFO Channel 12/0 : 2[6] -> 3[7] via P2P/IPC
+ip-26-0-170-31:3197185:3197767 [5] NCCL INFO Channel 15/0 : 1[5] -> 2[6] via P2P/IPC
+ip-26-0-170-31:3197187:3197764 [7] NCCL INFO Channel 15/0 : 3[7] -> 0[4] via P2P/IPC
+ip-26-0-163-58:934727:935188 [7] NCCL INFO Channel 17/0 : 3[7] -> 0[4] via P2P/IPC
+ip-26-0-163-58:934724:935186 [4] NCCL INFO Channel 16/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-163-58:934726:935189 [6] NCCL INFO Channel 15/0 : 2[6] -> 3[7] via P2P/IPC
+ip-26-0-170-31:3197184:3197762 [4] NCCL INFO Channel 14/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-170-31:3197185:3197767 [5] NCCL INFO Channel 16/0 : 1[5] -> 2[6] via P2P/IPC
+ip-26-0-170-31:3197186:3197766 [6] NCCL INFO Channel 13/0 : 2[6] -> 3[7] via P2P/IPC
+ip-26-0-170-31:3197187:3197764 [7] NCCL INFO Channel 16/0 : 3[7] -> 0[4] via P2P/IPC
+ip-26-0-163-58:934727:935188 [7] NCCL INFO Channel 18/0 : 3[7] -> 0[4] via P2P/IPC
+ip-26-0-163-58:934724:935186 [4] NCCL INFO Channel 17/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-163-58:934726:935189 [6] NCCL INFO Channel 16/0 : 2[6] -> 3[7] via P2P/IPC
+ip-26-0-163-58:934725:935190 [5] NCCL INFO Channel 16/0 : 1[5] -> 2[6] via P2P/IPC
+ip-26-0-163-58:934727:935188 [7] NCCL INFO Channel 19/0 : 3[7] -> 0[4] via P2P/IPC
+ip-26-0-163-58:934726:935189 [6] NCCL INFO Channel 17/0 : 2[6] -> 3[7] via P2P/IPC
+ip-26-0-163-58:934724:935186 [4] NCCL INFO Channel 18/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-170-31:3197184:3197762 [4] NCCL INFO Channel 15/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-170-31:3197186:3197766 [6] NCCL INFO Channel 14/0 : 2[6] -> 3[7] via P2P/IPC
+ip-26-0-170-31:3197185:3197767 [5] NCCL INFO Channel 17/0 : 1[5] -> 2[6] via P2P/IPC
+ip-26-0-170-31:3197187:3197764 [7] NCCL INFO Channel 17/0 : 3[7] -> 0[4] via P2P/IPC
+ip-26-0-163-58:934725:935190 [5] NCCL INFO Channel 17/0 : 1[5] -> 2[6] via P2P/IPC
+ip-26-0-163-58:934727:935188 [7] NCCL INFO Channel 20/0 : 3[7] -> 0[4] via P2P/IPC
+ip-26-0-163-58:934726:935189 [6] NCCL INFO Channel 18/0 : 2[6] -> 3[7] via P2P/IPC
+ip-26-0-163-58:934724:935186 [4] NCCL INFO Channel 19/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-170-31:3197184:3197762 [4] NCCL INFO Channel 16/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-170-31:3197186:3197766 [6] NCCL INFO Channel 15/0 : 2[6] -> 3[7] via P2P/IPC
+ip-26-0-170-31:3197185:3197767 [5] NCCL INFO Channel 18/0 : 1[5] -> 2[6] via P2P/IPC
+ip-26-0-170-31:3197187:3197764 [7] NCCL INFO Channel 18/0 : 3[7] -> 0[4] via P2P/IPC
+ip-26-0-170-31:3197184:3197762 [4] NCCL INFO Channel 17/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-170-31:3197185:3197767 [5] NCCL INFO Channel 19/0 : 1[5] -> 2[6] via P2P/IPC
+ip-26-0-170-31:3197187:3197764 [7] NCCL INFO Channel 19/0 : 3[7] -> 0[4] via P2P/IPC
+ip-26-0-170-31:3197186:3197766 [6] NCCL INFO Channel 16/0 : 2[6] -> 3[7] via P2P/IPC
+ip-26-0-163-58:934725:935190 [5] NCCL INFO Channel 18/0 : 1[5] -> 2[6] via P2P/IPC
+ip-26-0-163-58:934727:935188 [7] NCCL INFO Channel 21/0 : 3[7] -> 0[4] via P2P/IPC
+ip-26-0-163-58:934726:935189 [6] NCCL INFO Channel 19/0 : 2[6] -> 3[7] via P2P/IPC
+ip-26-0-163-58:934724:935186 [4] NCCL INFO Channel 20/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-163-58:934725:935190 [5] NCCL INFO Channel 19/0 : 1[5] -> 2[6] via P2P/IPC
+ip-26-0-163-58:934727:935188 [7] NCCL INFO Channel 22/0 : 3[7] -> 0[4] via P2P/IPC
+ip-26-0-163-58:934726:935189 [6] NCCL INFO Channel 20/0 : 2[6] -> 3[7] via P2P/IPC
+ip-26-0-163-58:934724:935186 [4] NCCL INFO Channel 21/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-170-31:3197186:3197766 [6] NCCL INFO Channel 17/0 : 2[6] -> 3[7] via P2P/IPC
+ip-26-0-170-31:3197183:3197765 [3] NCCL INFO Channel 00/0 : 3[3] -> 0[0] via P2P/IPC
+ip-26-0-170-31:3197184:3197762 [4] NCCL INFO Channel 18/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-170-31:3197185:3197767 [5] NCCL INFO Channel 20/0 : 1[5] -> 2[6] via P2P/IPC
+ip-26-0-170-31:3197187:3197764 [7] NCCL INFO Channel 20/0 : 3[7] -> 0[4] via P2P/IPC
+ip-26-0-170-31:3197180:3197763 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-163-58:934725:935190 [5] NCCL INFO Channel 20/0 : 1[5] -> 2[6] via P2P/IPC
+ip-26-0-163-58:934727:935188 [7] NCCL INFO Channel 23/0 : 3[7] -> 0[4] via P2P/IPC
+ip-26-0-163-58:934726:935189 [6] NCCL INFO Channel 21/0 : 2[6] -> 3[7] via P2P/IPC
+ip-26-0-163-58:934724:935186 [4] NCCL INFO Channel 22/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-170-31:3197184:3197762 [4] NCCL INFO Channel 19/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-170-31:3197185:3197767 [5] NCCL INFO Channel 21/0 : 1[5] -> 2[6] via P2P/IPC
+ip-26-0-163-58:934725:935190 [5] NCCL INFO Channel 21/0 : 1[5] -> 2[6] via P2P/IPC
+ip-26-0-163-58:934723:935194 [3] NCCL INFO Channel 00/0 : 3[3] -> 0[0] via P2P/IPC
+ip-26-0-163-58:934724:935186 [4] NCCL INFO Channel 23/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-163-58:934726:935189 [6] NCCL INFO Channel 22/0 : 2[6] -> 3[7] via P2P/IPC
+ip-26-0-170-31:3197184:3197762 [4] NCCL INFO Channel 20/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-170-31:3197181:3197768 [1] NCCL INFO Channel 00/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-170-31:3197180:3197763 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-163-58:934725:935190 [5] NCCL INFO Channel 22/0 : 1[5] -> 2[6] via P2P/IPC
+ip-26-0-163-58:934722:935192 [2] NCCL INFO Channel 00/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-163-58:934726:935189 [6] NCCL INFO Channel 23/0 : 2[6] -> 3[7] via P2P/IPC
+ip-26-0-163-58:934723:935194 [3] NCCL INFO Channel 01/0 : 3[3] -> 0[0] via P2P/IPC
+ip-26-0-163-58:934721:935193 [1] NCCL INFO Channel 00/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-170-31:3197186:3197766 [6] NCCL INFO Channel 18/0 : 2[6] -> 3[7] via P2P/IPC
+ip-26-0-170-31:3197183:3197765 [3] NCCL INFO Channel 01/0 : 3[3] -> 0[0] via P2P/IPC
+ip-26-0-170-31:3197181:3197768 [1] NCCL INFO Channel 01/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-170-31:3197180:3197763 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-170-31:3197186:3197766 [6] NCCL INFO Channel 19/0 : 2[6] -> 3[7] via P2P/IPC
+ip-26-0-170-31:3197182:3197769 [2] NCCL INFO Channel 00/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-170-31:3197180:3197763 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-163-58:934722:935192 [2] NCCL INFO Channel 01/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-163-58:934725:935190 [5] NCCL INFO Channel 23/0 : 1[5] -> 2[6] via P2P/IPC
+ip-26-0-163-58:934723:935194 [3] NCCL INFO Channel 02/0 : 3[3] -> 0[0] via P2P/IPC
+ip-26-0-163-58:934721:935193 [1] NCCL INFO Channel 01/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-170-31:3197185:3197767 [5] NCCL INFO Channel 22/0 : 1[5] -> 2[6] via P2P/IPC
+ip-26-0-170-31:3197184:3197762 [4] NCCL INFO Channel 21/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-170-31:3197180:3197763 [0] NCCL INFO Channel 04/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-170-31:3197182:3197769 [2] NCCL INFO Channel 01/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-170-31:3197185:3197767 [5] NCCL INFO Channel 23/0 : 1[5] -> 2[6] via P2P/IPC
+ip-26-0-170-31:3197187:3197764 [7] NCCL INFO Channel 21/0 : 3[7] -> 0[4] via P2P/IPC
+ip-26-0-163-58:934723:935194 [3] NCCL INFO Channel 03/0 : 3[3] -> 0[0] via P2P/IPC
+ip-26-0-163-58:934721:935193 [1] NCCL INFO Channel 02/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-163-58:934722:935192 [2] NCCL INFO Channel 02/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-170-31:3197182:3197769 [2] NCCL INFO Channel 02/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-170-31:3197180:3197763 [0] NCCL INFO Channel 05/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-170-31:3197184:3197762 [4] NCCL INFO Channel 22/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-170-31:3197187:3197764 [7] NCCL INFO Channel 22/0 : 3[7] -> 0[4] via P2P/IPC
+ip-26-0-170-31:3197182:3197769 [2] NCCL INFO Channel 03/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-170-31:3197183:3197765 [3] NCCL INFO Channel 02/0 : 3[3] -> 0[0] via P2P/IPC
+ip-26-0-170-31:3197180:3197763 [0] NCCL INFO Channel 06/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-170-31:3197181:3197768 [1] NCCL INFO Channel 02/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-170-31:3197184:3197762 [4] NCCL INFO Channel 23/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-170-31:3197187:3197764 [7] NCCL INFO Channel 23/0 : 3[7] -> 0[4] via P2P/IPC
+ip-26-0-162-14:2525079:2525546 [4] NCCL INFO Connected all rings
+ip-26-0-170-31:3197186:3197766 [6] NCCL INFO Channel 20/0 : 2[6] -> 3[7] via P2P/IPC
+ip-26-0-170-31:3197182:3197769 [2] NCCL INFO Channel 04/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-170-31:3197183:3197765 [3] NCCL INFO Channel 03/0 : 3[3] -> 0[0] via P2P/IPC
+ip-26-0-170-31:3197180:3197763 [0] NCCL INFO Channel 07/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-170-31:3197181:3197768 [1] NCCL INFO Channel 03/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-170-31:3197186:3197766 [6] NCCL INFO Channel 21/0 : 2[6] -> 3[7] via P2P/IPC
+ip-26-0-170-31:3197182:3197769 [2] NCCL INFO Channel 05/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-170-31:3197180:3197763 [0] NCCL INFO Channel 08/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-170-31:3197183:3197765 [3] NCCL INFO Channel 04/0 : 3[3] -> 0[0] via P2P/IPC
+ip-26-0-163-58:934723:935194 [3] NCCL INFO Channel 04/0 : 3[3] -> 0[0] via P2P/IPC
+ip-26-0-163-58:934721:935193 [1] NCCL INFO Channel 03/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-163-58:934722:935192 [2] NCCL INFO Channel 03/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-170-31:3197181:3197768 [1] NCCL INFO Channel 04/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-170-31:3197180:3197763 [0] NCCL INFO Channel 09/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-170-31:3197182:3197769 [2] NCCL INFO Channel 06/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-170-31:3197186:3197766 [6] NCCL INFO Channel 22/0 : 2[6] -> 3[7] via P2P/IPC
+ip-26-0-170-31:3197183:3197765 [3] NCCL INFO Channel 05/0 : 3[3] -> 0[0] via P2P/IPC
+ip-26-0-170-31:3197181:3197768 [1] NCCL INFO Channel 05/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-170-31:3197182:3197769 [2] NCCL INFO Channel 07/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-170-31:3197180:3197763 [0] NCCL INFO Channel 10/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-170-31:3197186:3197766 [6] NCCL INFO Channel 23/0 : 2[6] -> 3[7] via P2P/IPC
+ip-26-0-170-31:3197183:3197765 [3] NCCL INFO Channel 06/0 : 3[3] -> 0[0] via P2P/IPC
+ip-26-0-170-31:3197181:3197768 [1] NCCL INFO Channel 06/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-163-58:934723:935194 [3] NCCL INFO Channel 05/0 : 3[3] -> 0[0] via P2P/IPC
+ip-26-0-163-58:934720:935191 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-170-31:3197182:3197769 [2] NCCL INFO Channel 08/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-170-31:3197180:3197763 [0] NCCL INFO Channel 11/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-170-31:3197181:3197768 [1] NCCL INFO Channel 07/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-170-31:3197183:3197765 [3] NCCL INFO Channel 07/0 : 3[3] -> 0[0] via P2P/IPC
+ip-26-0-163-58:934721:935193 [1] NCCL INFO Channel 04/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-163-58:934722:935192 [2] NCCL INFO Channel 04/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-163-58:934721:935193 [1] NCCL INFO Channel 05/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-170-31:3197182:3197769 [2] NCCL INFO Channel 09/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-162-14:2525081:2525549 [6] NCCL INFO Connected all rings
+ip-26-0-170-31:3197181:3197768 [1] NCCL INFO Channel 08/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-170-31:3197180:3197763 [0] NCCL INFO Channel 12/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-170-31:3197183:3197765 [3] NCCL INFO Channel 08/0 : 3[3] -> 0[0] via P2P/IPC
+ip-26-0-163-58:934722:935192 [2] NCCL INFO Channel 05/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-163-58:934721:935193 [1] NCCL INFO Channel 06/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-163-58:934720:935191 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-163-58:934723:935194 [3] NCCL INFO Channel 06/0 : 3[3] -> 0[0] via P2P/IPC
+ip-26-0-170-31:3197181:3197768 [1] NCCL INFO Channel 09/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-170-31:3197181:3197768 [1] NCCL INFO Channel 10/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-163-58:934722:935192 [2] NCCL INFO Channel 06/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-163-58:934721:935193 [1] NCCL INFO Channel 07/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-163-58:934723:935194 [3] NCCL INFO Channel 07/0 : 3[3] -> 0[0] via P2P/IPC
+ip-26-0-170-31:3197181:3197768 [1] NCCL INFO Channel 11/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-163-58:934720:935191 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-170-31:3197181:3197768 [1] NCCL INFO Channel 12/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-170-31:3197181:3197768 [1] NCCL INFO Channel 13/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-170-31:3197181:3197768 [1] NCCL INFO Channel 14/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-170-31:3197181:3197768 [1] NCCL INFO Channel 15/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-170-31:3197181:3197768 [1] NCCL INFO Channel 16/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-170-31:3197182:3197769 [2] NCCL INFO Channel 10/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-170-31:3197180:3197763 [0] NCCL INFO Channel 13/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-170-31:3197183:3197765 [3] NCCL INFO Channel 09/0 : 3[3] -> 0[0] via P2P/IPC
+ip-26-0-162-14:2525082:2525548 [7] NCCL INFO Connected all rings
+ip-26-0-162-14:2525082:2525548 [7] NCCL INFO Channel 00/0 : 3[7] -> 2[6] via P2P/IPC
+ip-26-0-163-58:934722:935192 [2] NCCL INFO Channel 07/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-170-31:3197181:3197768 [1] NCCL INFO Channel 17/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-163-58:934721:935193 [1] NCCL INFO Channel 08/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-163-58:934723:935194 [3] NCCL INFO Channel 08/0 : 3[3] -> 0[0] via P2P/IPC
+ip-26-0-163-58:934720:935191 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-162-14:2525080:2525550 [5] NCCL INFO Connected all rings
+ip-26-0-170-31:3197182:3197769 [2] NCCL INFO Channel 11/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-170-31:3197180:3197763 [0] NCCL INFO Channel 14/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-170-132:3330185:3330664 [2] NCCL INFO Connected all rings
+ip-26-0-170-31:3197183:3197765 [3] NCCL INFO Channel 10/0 : 3[3] -> 0[0] via P2P/IPC
+ip-26-0-163-58:934723:935194 [3] NCCL INFO Channel 09/0 : 3[3] -> 0[0] via P2P/IPC
+ip-26-0-170-31:3197182:3197769 [2] NCCL INFO Channel 12/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-163-58:934722:935192 [2] NCCL INFO Channel 08/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-170-132:3330186:3330663 [3] NCCL INFO Connected all rings
+ip-26-0-170-132:3330186:3330663 [3] NCCL INFO Channel 00/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-163-58:934720:935191 [0] NCCL INFO Channel 04/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-163-58:934721:935193 [1] NCCL INFO Channel 09/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-170-31:3197182:3197769 [2] NCCL INFO Channel 13/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-170-31:3197181:3197768 [1] NCCL INFO Channel 18/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-162-14:2525082:2525548 [7] NCCL INFO Channel 01/0 : 3[7] -> 2[6] via P2P/IPC
+ip-26-0-170-31:3197180:3197763 [0] NCCL INFO Channel 15/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-170-31:3197183:3197765 [3] NCCL INFO Channel 11/0 : 3[3] -> 0[0] via P2P/IPC
+ip-26-0-170-31:3197182:3197769 [2] NCCL INFO Channel 14/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-170-31:3197181:3197768 [1] NCCL INFO Channel 19/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-170-132:3330186:3330663 [3] NCCL INFO Channel 01/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-170-132:3330183:3330662 [0] NCCL INFO Connected all rings
+ip-26-0-163-58:934720:935191 [0] NCCL INFO Channel 05/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-163-58:934721:935193 [1] NCCL INFO Channel 10/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-163-58:934722:935192 [2] NCCL INFO Channel 09/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-163-58:934723:935194 [3] NCCL INFO Channel 10/0 : 3[3] -> 0[0] via P2P/IPC
+ip-26-0-170-31:3197180:3197763 [0] NCCL INFO Channel 16/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-162-14:2525082:2525548 [7] NCCL INFO Channel 02/0 : 3[7] -> 2[6] via P2P/IPC
+ip-26-0-170-31:3197183:3197765 [3] NCCL INFO Channel 12/0 : 3[3] -> 0[0] via P2P/IPC
+ip-26-0-163-58:934720:935191 [0] NCCL INFO Channel 06/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-170-31:3197182:3197769 [2] NCCL INFO Channel 15/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-170-132:3330186:3330663 [3] NCCL INFO Channel 02/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-163-58:934721:935193 [1] NCCL INFO Channel 11/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-163-58:934722:935192 [2] NCCL INFO Channel 10/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-163-58:934723:935194 [3] NCCL INFO Channel 11/0 : 3[3] -> 0[0] via P2P/IPC
+ip-26-0-170-31:3197180:3197763 [0] NCCL INFO Channel 17/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-170-31:3197181:3197768 [1] NCCL INFO Channel 20/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-170-31:3197183:3197765 [3] NCCL INFO Channel 13/0 : 3[3] -> 0[0] via P2P/IPC
+ip-26-0-163-58:934720:935191 [0] NCCL INFO Channel 07/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-162-14:2525082:2525548 [7] NCCL INFO Channel 03/0 : 3[7] -> 2[6] via P2P/IPC
+ip-26-0-170-31:3197180:3197763 [0] NCCL INFO Channel 18/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-170-31:3197182:3197769 [2] NCCL INFO Channel 16/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-170-132:3330184:3330665 [1] NCCL INFO Connected all rings
+ip-26-0-170-132:3330186:3330663 [3] NCCL INFO Channel 03/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-163-58:934721:935193 [1] NCCL INFO Channel 12/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-163-58:934722:935192 [2] NCCL INFO Channel 11/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-163-58:934723:935194 [3] NCCL INFO Channel 12/0 : 3[3] -> 0[0] via P2P/IPC
+ip-26-0-170-31:3197180:3197763 [0] NCCL INFO Channel 19/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-163-58:934720:935191 [0] NCCL INFO Channel 08/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-163-58:934721:935193 [1] NCCL INFO Channel 13/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-163-58:934722:935192 [2] NCCL INFO Channel 12/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-163-58:934723:935194 [3] NCCL INFO Channel 13/0 : 3[3] -> 0[0] via P2P/IPC
+ip-26-0-170-31:3197180:3197763 [0] NCCL INFO Channel 20/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-162-14:2525082:2525548 [7] NCCL INFO Channel 04/0 : 3[7] -> 2[6] via P2P/IPC
+ip-26-0-170-31:3197181:3197768 [1] NCCL INFO Channel 21/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-170-31:3197183:3197765 [3] NCCL INFO Channel 14/0 : 3[3] -> 0[0] via P2P/IPC
+ip-26-0-163-58:934722:935192 [2] NCCL INFO Channel 13/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-170-31:3197182:3197769 [2] NCCL INFO Channel 17/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-170-31:3197180:3197763 [0] NCCL INFO Channel 21/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-162-14:2525082:2525548 [7] NCCL INFO Channel 05/0 : 3[7] -> 2[6] via P2P/IPC
+ip-26-0-170-132:3330186:3330663 [3] NCCL INFO Channel 04/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-163-58:934722:935192 [2] NCCL INFO Channel 14/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-170-31:3197181:3197768 [1] NCCL INFO Channel 22/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-170-31:3197183:3197765 [3] NCCL INFO Channel 15/0 : 3[3] -> 0[0] via P2P/IPC
+ip-26-0-163-58:934720:935191 [0] NCCL INFO Channel 09/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-163-58:934723:935194 [3] NCCL INFO Channel 14/0 : 3[3] -> 0[0] via P2P/IPC
+ip-26-0-163-58:934721:935193 [1] NCCL INFO Channel 14/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-170-31:3197182:3197769 [2] NCCL INFO Channel 18/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-163-58:934722:935192 [2] NCCL INFO Channel 15/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-170-31:3197180:3197763 [0] NCCL INFO Channel 22/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-170-31:3197181:3197768 [1] NCCL INFO Channel 23/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-170-31:3197183:3197765 [3] NCCL INFO Channel 16/0 : 3[3] -> 0[0] via P2P/IPC
+ip-26-0-163-58:934720:935191 [0] NCCL INFO Channel 10/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-163-58:934723:935194 [3] NCCL INFO Channel 15/0 : 3[3] -> 0[0] via P2P/IPC
+ip-26-0-162-14:2525082:2525548 [7] NCCL INFO Channel 06/0 : 3[7] -> 2[6] via P2P/IPC
+ip-26-0-163-58:934721:935193 [1] NCCL INFO Channel 15/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-170-31:3197182:3197769 [2] NCCL INFO Channel 19/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-170-31:3197180:3197763 [0] NCCL INFO Channel 23/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-163-58:934722:935192 [2] NCCL INFO Channel 16/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-170-31:3197183:3197765 [3] NCCL INFO Channel 17/0 : 3[3] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330186:3330663 [3] NCCL INFO Channel 05/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-162-14:2525082:2525548 [7] NCCL INFO Channel 07/0 : 3[7] -> 2[6] via P2P/IPC
+ip-26-0-163-58:934723:935194 [3] NCCL INFO Channel 16/0 : 3[3] -> 0[0] via P2P/IPC
+ip-26-0-163-58:934720:935191 [0] NCCL INFO Channel 11/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-163-58:934721:935193 [1] NCCL INFO Channel 16/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-170-132:3330186:3330663 [3] NCCL INFO Channel 06/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-163-58:934722:935192 [2] NCCL INFO Channel 17/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-170-31:3197182:3197769 [2] NCCL INFO Channel 20/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-163-58:934723:935194 [3] NCCL INFO Channel 17/0 : 3[3] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330186:3330663 [3] NCCL INFO Channel 07/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-163-58:934720:935191 [0] NCCL INFO Channel 12/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-170-31:3197183:3197765 [3] NCCL INFO Channel 18/0 : 3[3] -> 0[0] via P2P/IPC
+ip-26-0-163-58:934721:935193 [1] NCCL INFO Channel 17/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-170-31:3197182:3197769 [2] NCCL INFO Channel 21/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-170-31:3197183:3197765 [3] NCCL INFO Channel 19/0 : 3[3] -> 0[0] via P2P/IPC
+ip-26-0-162-14:2525082:2525548 [7] NCCL INFO Channel 08/0 : 3[7] -> 2[6] via P2P/IPC
+ip-26-0-163-58:934722:935192 [2] NCCL INFO Channel 18/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-170-31:3197182:3197769 [2] NCCL INFO Channel 22/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-170-31:3197183:3197765 [3] NCCL INFO Channel 20/0 : 3[3] -> 0[0] via P2P/IPC
+ip-26-0-163-58:934723:935194 [3] NCCL INFO Channel 18/0 : 3[3] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330187:3330667 [4] NCCL INFO Connected all rings
+ip-26-0-163-58:934720:935191 [0] NCCL INFO Channel 13/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-170-31:3197183:3197765 [3] NCCL INFO Channel 21/0 : 3[3] -> 0[0] via P2P/IPC
+ip-26-0-170-31:3197182:3197769 [2] NCCL INFO Channel 23/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-170-132:3330186:3330663 [3] NCCL INFO Channel 08/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-162-14:2525082:2525548 [7] NCCL INFO Channel 09/0 : 3[7] -> 2[6] via P2P/IPC
+ip-26-0-163-58:934721:935193 [1] NCCL INFO Channel 18/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-163-58:934722:935192 [2] NCCL INFO Channel 19/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-170-31:3197183:3197765 [3] NCCL INFO Channel 22/0 : 3[3] -> 0[0] via P2P/IPC
+ip-26-0-163-58:934720:935191 [0] NCCL INFO Channel 14/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-170-132:3330186:3330663 [3] NCCL INFO Channel 09/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-162-14:2525082:2525548 [7] NCCL INFO Channel 10/0 : 3[7] -> 2[6] via P2P/IPC
+ip-26-0-162-14:2525082:2525548 [7] NCCL INFO Channel 11/0 : 3[7] -> 2[6] via P2P/IPC
+ip-26-0-163-58:934723:935194 [3] NCCL INFO Channel 19/0 : 3[3] -> 0[0] via P2P/IPC
+ip-26-0-163-58:934722:935192 [2] NCCL INFO Channel 20/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-170-132:3330186:3330663 [3] NCCL INFO Channel 10/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-163-58:934721:935193 [1] NCCL INFO Channel 19/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-163-58:934720:935191 [0] NCCL INFO Channel 15/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-170-132:3330186:3330663 [3] NCCL INFO Channel 11/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-163-58:934723:935194 [3] NCCL INFO Channel 20/0 : 3[3] -> 0[0] via P2P/IPC
+ip-26-0-162-14:2525082:2525548 [7] NCCL INFO Channel 12/0 : 3[7] -> 2[6] via P2P/IPC
+ip-26-0-163-58:934722:935192 [2] NCCL INFO Channel 21/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-170-31:3197183:3197765 [3] NCCL INFO Channel 23/0 : 3[3] -> 0[0] via P2P/IPC
+ip-26-0-163-58:934721:935193 [1] NCCL INFO Channel 20/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-162-14:2525082:2525548 [7] NCCL INFO Channel 13/0 : 3[7] -> 2[6] via P2P/IPC
+ip-26-0-163-58:934720:935191 [0] NCCL INFO Channel 16/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-163-58:934723:935194 [3] NCCL INFO Channel 21/0 : 3[3] -> 0[0] via P2P/IPC
+ip-26-0-163-58:934722:935192 [2] NCCL INFO Channel 22/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-170-132:3330186:3330663 [3] NCCL INFO Channel 12/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-163-58:934721:935193 [1] NCCL INFO Channel 21/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-162-14:2525082:2525548 [7] NCCL INFO Channel 14/0 : 3[7] -> 2[6] via P2P/IPC
+ip-26-0-163-58:934720:935191 [0] NCCL INFO Channel 17/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-170-132:3330186:3330663 [3] NCCL INFO Channel 13/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-163-58:934723:935194 [3] NCCL INFO Channel 22/0 : 3[3] -> 0[0] via P2P/IPC
+ip-26-0-163-58:934722:935192 [2] NCCL INFO Channel 23/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-162-14:2525082:2525548 [7] NCCL INFO Channel 15/0 : 3[7] -> 2[6] via P2P/IPC
+ip-26-0-163-58:934720:935191 [0] NCCL INFO Channel 18/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-163-58:934723:935194 [3] NCCL INFO Channel 23/0 : 3[3] -> 0[0] via P2P/IPC
+ip-26-0-163-58:934720:935191 [0] NCCL INFO Channel 19/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-163-58:934721:935193 [1] NCCL INFO Channel 22/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-170-132:3330186:3330663 [3] NCCL INFO Channel 14/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-162-14:2525082:2525548 [7] NCCL INFO Channel 16/0 : 3[7] -> 2[6] via P2P/IPC
+ip-26-0-170-132:3330188:3330669 [5] NCCL INFO Connected all rings
+ip-26-0-170-132:3330186:3330663 [3] NCCL INFO Channel 15/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-163-58:934720:935191 [0] NCCL INFO Channel 20/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-163-58:934721:935193 [1] NCCL INFO Channel 23/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-170-132:3330186:3330663 [3] NCCL INFO Channel 16/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-170-132:3330189:3330668 [6] NCCL INFO Connected all rings
+ip-26-0-162-14:2525082:2525548 [7] NCCL INFO Channel 17/0 : 3[7] -> 2[6] via P2P/IPC
+ip-26-0-163-58:934720:935191 [0] NCCL INFO Channel 21/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-170-132:3330186:3330663 [3] NCCL INFO Channel 17/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-162-14:2525082:2525548 [7] NCCL INFO Channel 18/0 : 3[7] -> 2[6] via P2P/IPC
+ip-26-0-163-58:934720:935191 [0] NCCL INFO Channel 22/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-170-132:3330186:3330663 [3] NCCL INFO Channel 18/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-170-132:3330190:3330670 [7] NCCL INFO Connected all rings
+ip-26-0-170-132:3330190:3330670 [7] NCCL INFO Channel 00/0 : 3[7] -> 2[6] via P2P/IPC
+ip-26-0-170-132:3330186:3330663 [3] NCCL INFO Channel 19/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-162-14:2525082:2525548 [7] NCCL INFO Channel 19/0 : 3[7] -> 2[6] via P2P/IPC
+ip-26-0-170-132:3330190:3330670 [7] NCCL INFO Channel 01/0 : 3[7] -> 2[6] via P2P/IPC
+ip-26-0-170-132:3330186:3330663 [3] NCCL INFO Channel 20/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-170-132:3330190:3330670 [7] NCCL INFO Channel 02/0 : 3[7] -> 2[6] via P2P/IPC
+ip-26-0-170-132:3330186:3330663 [3] NCCL INFO Channel 21/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-170-132:3330185:3330664 [2] NCCL INFO Channel 00/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-170-132:3330190:3330670 [7] NCCL INFO Channel 03/0 : 3[7] -> 2[6] via P2P/IPC
+ip-26-0-170-132:3330186:3330663 [3] NCCL INFO Channel 22/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-170-132:3330185:3330664 [2] NCCL INFO Channel 01/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-170-132:3330190:3330670 [7] NCCL INFO Channel 04/0 : 3[7] -> 2[6] via P2P/IPC
+ip-26-0-170-132:3330185:3330664 [2] NCCL INFO Channel 02/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-170-132:3330186:3330663 [3] NCCL INFO Channel 23/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-170-132:3330190:3330670 [7] NCCL INFO Channel 05/0 : 3[7] -> 2[6] via P2P/IPC
+ip-26-0-170-132:3330185:3330664 [2] NCCL INFO Channel 03/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-162-14:2525078:2525554 [3] NCCL INFO Connected all rings
+ip-26-0-162-14:2525078:2525554 [3] NCCL INFO Channel 00/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-163-58:934720:935191 [0] NCCL INFO Channel 23/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-162-14:2525082:2525548 [7] NCCL INFO Channel 20/0 : 3[7] -> 2[6] via P2P/IPC
+ip-26-0-170-132:3330184:3330665 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330185:3330664 [2] NCCL INFO Channel 04/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-170-132:3330184:3330665 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330185:3330664 [2] NCCL INFO Channel 05/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-170-132:3330184:3330665 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330185:3330664 [2] NCCL INFO Channel 06/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-170-132:3330184:3330665 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330185:3330664 [2] NCCL INFO Channel 07/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-170-132:3330190:3330670 [7] NCCL INFO Channel 06/0 : 3[7] -> 2[6] via P2P/IPC
+ip-26-0-170-132:3330184:3330665 [1] NCCL INFO Channel 04/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-163-58:934724:935186 [4] NCCL INFO Connected all rings
+ip-26-0-170-132:3330185:3330664 [2] NCCL INFO Channel 08/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-162-14:2525082:2525548 [7] NCCL INFO Channel 21/0 : 3[7] -> 2[6] via P2P/IPC
+ip-26-0-170-132:3330190:3330670 [7] NCCL INFO Channel 07/0 : 3[7] -> 2[6] via P2P/IPC
+ip-26-0-170-132:3330184:3330665 [1] NCCL INFO Channel 05/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330185:3330664 [2] NCCL INFO Channel 09/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-170-132:3330190:3330670 [7] NCCL INFO Channel 08/0 : 3[7] -> 2[6] via P2P/IPC
+ip-26-0-170-132:3330184:3330665 [1] NCCL INFO Channel 06/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330185:3330664 [2] NCCL INFO Channel 10/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-170-132:3330190:3330670 [7] NCCL INFO Channel 09/0 : 3[7] -> 2[6] via P2P/IPC
+ip-26-0-162-14:2525077:2525553 [2] NCCL INFO Connected all rings
+ip-26-0-163-58:934727:935188 [7] NCCL INFO Connected all rings
+ip-26-0-163-58:934727:935188 [7] NCCL INFO Channel 00/0 : 3[7] -> 2[6] via P2P/IPC
+ip-26-0-170-132:3330184:3330665 [1] NCCL INFO Channel 07/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330185:3330664 [2] NCCL INFO Channel 11/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-170-132:3330190:3330670 [7] NCCL INFO Channel 10/0 : 3[7] -> 2[6] via P2P/IPC
+ip-26-0-170-132:3330184:3330665 [1] NCCL INFO Channel 08/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330185:3330664 [2] NCCL INFO Channel 12/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-170-132:3330190:3330670 [7] NCCL INFO Channel 11/0 : 3[7] -> 2[6] via P2P/IPC
+ip-26-0-162-14:2525078:2525554 [3] NCCL INFO Channel 01/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-170-132:3330184:3330665 [1] NCCL INFO Channel 09/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330185:3330664 [2] NCCL INFO Channel 13/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-170-132:3330190:3330670 [7] NCCL INFO Channel 12/0 : 3[7] -> 2[6] via P2P/IPC
+ip-26-0-162-14:2525082:2525548 [7] NCCL INFO Channel 22/0 : 3[7] -> 2[6] via P2P/IPC
+ip-26-0-170-132:3330184:3330665 [1] NCCL INFO Channel 10/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330185:3330664 [2] NCCL INFO Channel 14/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-170-132:3330190:3330670 [7] NCCL INFO Channel 13/0 : 3[7] -> 2[6] via P2P/IPC
+ip-26-0-170-132:3330185:3330664 [2] NCCL INFO Channel 15/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-170-132:3330190:3330670 [7] NCCL INFO Channel 14/0 : 3[7] -> 2[6] via P2P/IPC
+ip-26-0-170-132:3330184:3330665 [1] NCCL INFO Channel 11/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330185:3330664 [2] NCCL INFO Channel 16/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-170-132:3330184:3330665 [1] NCCL INFO Channel 12/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330190:3330670 [7] NCCL INFO Channel 15/0 : 3[7] -> 2[6] via P2P/IPC
+ip-26-0-162-14:2525076:2525552 [1] NCCL INFO Connected all rings
+ip-26-0-170-132:3330185:3330664 [2] NCCL INFO Channel 17/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-170-132:3330184:3330665 [1] NCCL INFO Channel 13/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330190:3330670 [7] NCCL INFO Channel 16/0 : 3[7] -> 2[6] via P2P/IPC
+ip-26-0-162-14:2525078:2525554 [3] NCCL INFO Channel 02/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-163-58:934725:935190 [5] NCCL INFO Connected all rings
+ip-26-0-170-132:3330185:3330664 [2] NCCL INFO Channel 18/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-170-132:3330184:3330665 [1] NCCL INFO Channel 14/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330190:3330670 [7] NCCL INFO Channel 17/0 : 3[7] -> 2[6] via P2P/IPC
+ip-26-0-170-132:3330189:3330668 [6] NCCL INFO Channel 00/0 : 2[6] -> 1[5] via P2P/IPC
+ip-26-0-162-14:2525075:2525551 [0] NCCL INFO Connected all rings
+ip-26-0-163-58:934727:935188 [7] NCCL INFO Channel 01/0 : 3[7] -> 2[6] via P2P/IPC
+ip-26-0-162-14:2525082:2525548 [7] NCCL INFO Channel 23/0 : 3[7] -> 2[6] via P2P/IPC
+ip-26-0-162-14:2525078:2525554 [3] NCCL INFO Channel 03/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-170-132:3330185:3330664 [2] NCCL INFO Channel 19/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-170-132:3330184:3330665 [1] NCCL INFO Channel 15/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330188:3330669 [5] NCCL INFO Channel 00/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-170-132:3330190:3330670 [7] NCCL INFO Channel 18/0 : 3[7] -> 2[6] via P2P/IPC
+ip-26-0-170-132:3330189:3330668 [6] NCCL INFO Channel 01/0 : 2[6] -> 1[5] via P2P/IPC
+ip-26-0-162-14:2525080:2525550 [5] NCCL INFO Channel 00/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-162-14:2525078:2525554 [3] NCCL INFO Channel 04/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-170-132:3330185:3330664 [2] NCCL INFO Channel 20/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-170-132:3330188:3330669 [5] NCCL INFO Channel 01/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-170-132:3330190:3330670 [7] NCCL INFO Channel 19/0 : 3[7] -> 2[6] via P2P/IPC
+ip-26-0-170-132:3330189:3330668 [6] NCCL INFO Channel 02/0 : 2[6] -> 1[5] via P2P/IPC
+ip-26-0-162-14:2525078:2525554 [3] NCCL INFO Channel 05/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-162-14:2525080:2525550 [5] NCCL INFO Channel 01/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-170-132:3330184:3330665 [1] NCCL INFO Channel 16/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-162-14:2525081:2525549 [6] NCCL INFO Channel 00/0 : 2[6] -> 1[5] via P2P/IPC
+ip-26-0-170-132:3330185:3330664 [2] NCCL INFO Channel 21/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-170-132:3330188:3330669 [5] NCCL INFO Channel 02/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-170-132:3330190:3330670 [7] NCCL INFO Channel 20/0 : 3[7] -> 2[6] via P2P/IPC
+ip-26-0-170-132:3330189:3330668 [6] NCCL INFO Channel 03/0 : 2[6] -> 1[5] via P2P/IPC
+ip-26-0-162-14:2525080:2525550 [5] NCCL INFO Channel 02/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-162-14:2525078:2525554 [3] NCCL INFO Channel 06/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-170-132:3330184:3330665 [1] NCCL INFO Channel 17/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330188:3330669 [5] NCCL INFO Channel 03/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-162-14:2525081:2525549 [6] NCCL INFO Channel 01/0 : 2[6] -> 1[5] via P2P/IPC
+ip-26-0-162-14:2525080:2525550 [5] NCCL INFO Channel 03/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-162-14:2525078:2525554 [3] NCCL INFO Channel 07/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-170-132:3330185:3330664 [2] NCCL INFO Channel 22/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-170-132:3330190:3330670 [7] NCCL INFO Channel 21/0 : 3[7] -> 2[6] via P2P/IPC
+ip-26-0-170-132:3330184:3330665 [1] NCCL INFO Channel 18/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330188:3330669 [5] NCCL INFO Channel 04/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-162-14:2525081:2525549 [6] NCCL INFO Channel 02/0 : 2[6] -> 1[5] via P2P/IPC
+ip-26-0-162-14:2525080:2525550 [5] NCCL INFO Channel 04/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-162-14:2525078:2525554 [3] NCCL INFO Channel 08/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-170-132:3330185:3330664 [2] NCCL INFO Channel 23/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-170-132:3330189:3330668 [6] NCCL INFO Channel 04/0 : 2[6] -> 1[5] via P2P/IPC
+ip-26-0-170-132:3330190:3330670 [7] NCCL INFO Channel 22/0 : 3[7] -> 2[6] via P2P/IPC
+ip-26-0-170-132:3330184:3330665 [1] NCCL INFO Channel 19/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330188:3330669 [5] NCCL INFO Channel 05/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-163-58:934727:935188 [7] NCCL INFO Channel 02/0 : 3[7] -> 2[6] via P2P/IPC
+ip-26-0-162-14:2525081:2525549 [6] NCCL INFO Channel 03/0 : 2[6] -> 1[5] via P2P/IPC
+ip-26-0-162-14:2525080:2525550 [5] NCCL INFO Channel 05/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-162-14:2525078:2525554 [3] NCCL INFO Channel 09/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-170-132:3330189:3330668 [6] NCCL INFO Channel 05/0 : 2[6] -> 1[5] via P2P/IPC
+ip-26-0-170-132:3330190:3330670 [7] NCCL INFO Channel 23/0 : 3[7] -> 2[6] via P2P/IPC
+ip-26-0-170-132:3330184:3330665 [1] NCCL INFO Channel 20/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330188:3330669 [5] NCCL INFO Channel 06/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-162-14:2525081:2525549 [6] NCCL INFO Channel 04/0 : 2[6] -> 1[5] via P2P/IPC
+ip-26-0-162-14:2525080:2525550 [5] NCCL INFO Channel 06/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-162-14:2525078:2525554 [3] NCCL INFO Channel 10/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-162-14:2525081:2525549 [6] NCCL INFO Channel 05/0 : 2[6] -> 1[5] via P2P/IPC
+ip-26-0-162-14:2525080:2525550 [5] NCCL INFO Channel 07/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-162-14:2525078:2525554 [3] NCCL INFO Channel 11/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-162-14:2525081:2525549 [6] NCCL INFO Channel 06/0 : 2[6] -> 1[5] via P2P/IPC
+ip-26-0-162-14:2525080:2525550 [5] NCCL INFO Channel 08/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-162-14:2525078:2525554 [3] NCCL INFO Channel 12/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-170-132:3330189:3330668 [6] NCCL INFO Channel 06/0 : 2[6] -> 1[5] via P2P/IPC
+ip-26-0-170-132:3330184:3330665 [1] NCCL INFO Channel 21/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330188:3330669 [5] NCCL INFO Channel 07/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-162-14:2525081:2525549 [6] NCCL INFO Channel 07/0 : 2[6] -> 1[5] via P2P/IPC
+ip-26-0-162-14:2525080:2525550 [5] NCCL INFO Channel 09/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-162-14:2525078:2525554 [3] NCCL INFO Channel 13/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-162-14:2525080:2525550 [5] NCCL INFO Channel 10/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-162-14:2525081:2525549 [6] NCCL INFO Channel 08/0 : 2[6] -> 1[5] via P2P/IPC
+ip-26-0-162-14:2525078:2525554 [3] NCCL INFO Channel 14/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-162-14:2525080:2525550 [5] NCCL INFO Channel 11/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-162-14:2525081:2525549 [6] NCCL INFO Channel 09/0 : 2[6] -> 1[5] via P2P/IPC
+ip-26-0-162-14:2525078:2525554 [3] NCCL INFO Channel 15/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-163-58:934727:935188 [7] NCCL INFO Channel 03/0 : 3[7] -> 2[6] via P2P/IPC
+ip-26-0-170-132:3330188:3330669 [5] NCCL INFO Channel 08/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-170-132:3330189:3330668 [6] NCCL INFO Channel 07/0 : 2[6] -> 1[5] via P2P/IPC
+ip-26-0-170-132:3330184:3330665 [1] NCCL INFO Channel 22/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-162-14:2525080:2525550 [5] NCCL INFO Channel 12/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-162-14:2525081:2525549 [6] NCCL INFO Channel 10/0 : 2[6] -> 1[5] via P2P/IPC
+ip-26-0-162-14:2525078:2525554 [3] NCCL INFO Channel 16/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-162-14:2525080:2525550 [5] NCCL INFO Channel 13/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-162-14:2525081:2525549 [6] NCCL INFO Channel 11/0 : 2[6] -> 1[5] via P2P/IPC
+ip-26-0-162-14:2525078:2525554 [3] NCCL INFO Channel 17/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-162-14:2525080:2525550 [5] NCCL INFO Channel 14/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-162-14:2525081:2525549 [6] NCCL INFO Channel 12/0 : 2[6] -> 1[5] via P2P/IPC
+ip-26-0-162-14:2525078:2525554 [3] NCCL INFO Channel 18/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-170-132:3330188:3330669 [5] NCCL INFO Channel 09/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-170-132:3330189:3330668 [6] NCCL INFO Channel 08/0 : 2[6] -> 1[5] via P2P/IPC
+ip-26-0-170-132:3330184:3330665 [1] NCCL INFO Channel 23/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-162-14:2525080:2525550 [5] NCCL INFO Channel 15/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-162-14:2525081:2525549 [6] NCCL INFO Channel 13/0 : 2[6] -> 1[5] via P2P/IPC
+ip-26-0-162-14:2525078:2525554 [3] NCCL INFO Channel 19/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-162-14:2525080:2525550 [5] NCCL INFO Channel 16/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-162-14:2525081:2525549 [6] NCCL INFO Channel 14/0 : 2[6] -> 1[5] via P2P/IPC
+ip-26-0-162-14:2525078:2525554 [3] NCCL INFO Channel 20/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-170-132:3330188:3330669 [5] NCCL INFO Channel 10/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-170-132:3330189:3330668 [6] NCCL INFO Channel 09/0 : 2[6] -> 1[5] via P2P/IPC
+ip-26-0-162-14:2525080:2525550 [5] NCCL INFO Channel 17/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-162-14:2525081:2525549 [6] NCCL INFO Channel 15/0 : 2[6] -> 1[5] via P2P/IPC
+ip-26-0-163-58:934727:935188 [7] NCCL INFO Channel 04/0 : 3[7] -> 2[6] via P2P/IPC
+ip-26-0-162-14:2525078:2525554 [3] NCCL INFO Channel 21/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-162-14:2525080:2525550 [5] NCCL INFO Channel 18/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-162-14:2525081:2525549 [6] NCCL INFO Channel 16/0 : 2[6] -> 1[5] via P2P/IPC
+ip-26-0-162-14:2525078:2525554 [3] NCCL INFO Channel 22/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-162-14:2525080:2525550 [5] NCCL INFO Channel 19/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-162-14:2525081:2525549 [6] NCCL INFO Channel 17/0 : 2[6] -> 1[5] via P2P/IPC
+ip-26-0-162-14:2525078:2525554 [3] NCCL INFO Channel 23/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-162-14:2525080:2525550 [5] NCCL INFO Channel 20/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-162-14:2525081:2525549 [6] NCCL INFO Channel 18/0 : 2[6] -> 1[5] via P2P/IPC
+ip-26-0-162-14:2525080:2525550 [5] NCCL INFO Channel 21/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-162-14:2525081:2525549 [6] NCCL INFO Channel 19/0 : 2[6] -> 1[5] via P2P/IPC
+ip-26-0-162-14:2525081:2525549 [6] NCCL INFO Channel 20/0 : 2[6] -> 1[5] via P2P/IPC
+ip-26-0-162-14:2525080:2525550 [5] NCCL INFO Channel 22/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-170-132:3330189:3330668 [6] NCCL INFO Channel 10/0 : 2[6] -> 1[5] via P2P/IPC
+ip-26-0-162-14:2525080:2525550 [5] NCCL INFO Channel 23/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-162-14:2525081:2525549 [6] NCCL INFO Channel 21/0 : 2[6] -> 1[5] via P2P/IPC
+ip-26-0-162-14:2525081:2525549 [6] NCCL INFO Channel 22/0 : 2[6] -> 1[5] via P2P/IPC
+ip-26-0-163-58:934727:935188 [7] NCCL INFO Channel 05/0 : 3[7] -> 2[6] via P2P/IPC
+ip-26-0-162-14:2525081:2525549 [6] NCCL INFO Channel 23/0 : 2[6] -> 1[5] via P2P/IPC
+ip-26-0-163-58:934726:935189 [6] NCCL INFO Connected all rings
+ip-26-0-170-132:3330188:3330669 [5] NCCL INFO Channel 11/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-162-14:2525077:2525553 [2] NCCL INFO Channel 00/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-163-58:934727:935188 [7] NCCL INFO Channel 06/0 : 3[7] -> 2[6] via P2P/IPC
+ip-26-0-170-132:3330188:3330669 [5] NCCL INFO Channel 12/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-170-132:3330189:3330668 [6] NCCL INFO Channel 11/0 : 2[6] -> 1[5] via P2P/IPC
+ip-26-0-162-14:2525076:2525552 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-162-14:2525077:2525553 [2] NCCL INFO Channel 01/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-170-31:3197184:3197762 [4] NCCL INFO Connected all rings
+ip-26-0-170-132:3330188:3330669 [5] NCCL INFO Channel 13/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-170-132:3330189:3330668 [6] NCCL INFO Channel 12/0 : 2[6] -> 1[5] via P2P/IPC
+ip-26-0-162-14:2525076:2525552 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-162-14:2525077:2525553 [2] NCCL INFO Channel 02/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-163-58:934727:935188 [7] NCCL INFO Channel 07/0 : 3[7] -> 2[6] via P2P/IPC
+ip-26-0-170-132:3330188:3330669 [5] NCCL INFO Channel 14/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-170-132:3330189:3330668 [6] NCCL INFO Channel 13/0 : 2[6] -> 1[5] via P2P/IPC
+ip-26-0-170-31:3197186:3197766 [6] NCCL INFO Connected all rings
+ip-26-0-170-132:3330188:3330669 [5] NCCL INFO Channel 15/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-170-132:3330189:3330668 [6] NCCL INFO Channel 14/0 : 2[6] -> 1[5] via P2P/IPC
+ip-26-0-170-132:3330188:3330669 [5] NCCL INFO Channel 16/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-170-132:3330189:3330668 [6] NCCL INFO Channel 15/0 : 2[6] -> 1[5] via P2P/IPC
+ip-26-0-163-58:934727:935188 [7] NCCL INFO Channel 08/0 : 3[7] -> 2[6] via P2P/IPC
+ip-26-0-162-14:2525076:2525552 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-162-14:2525077:2525553 [2] NCCL INFO Channel 03/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-170-132:3330188:3330669 [5] NCCL INFO Channel 17/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-170-132:3330189:3330668 [6] NCCL INFO Channel 16/0 : 2[6] -> 1[5] via P2P/IPC
+ip-26-0-170-31:3197185:3197767 [5] NCCL INFO Connected all rings
+ip-26-0-163-58:934727:935188 [7] NCCL INFO Channel 09/0 : 3[7] -> 2[6] via P2P/IPC
+ip-26-0-170-132:3330186:3330663 [3] NCCL INFO Connected all trees
+ip-26-0-170-132:3330186:3330663 [3] NCCL INFO NVLS comm 0xa269dd0 headRank 3 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-162-14:2525077:2525553 [2] NCCL INFO Channel 04/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-162-14:2525076:2525552 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-163-58:934727:935188 [7] NCCL INFO Channel 10/0 : 3[7] -> 2[6] via P2P/IPC
+ip-26-0-170-31:3197187:3197764 [7] NCCL INFO Connected all rings
+ip-26-0-170-31:3197187:3197764 [7] NCCL INFO Channel 00/0 : 3[7] -> 2[6] via P2P/IPC
+ip-26-0-162-14:2525077:2525553 [2] NCCL INFO Channel 05/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-170-132:3330188:3330669 [5] NCCL INFO Channel 18/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-170-132:3330189:3330668 [6] NCCL INFO Channel 17/0 : 2[6] -> 1[5] via P2P/IPC
+ip-26-0-162-14:2525076:2525552 [1] NCCL INFO Channel 04/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-162-14:2525077:2525553 [2] NCCL INFO Channel 06/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-170-31:3197187:3197764 [7] NCCL INFO Channel 01/0 : 3[7] -> 2[6] via P2P/IPC
+ip-26-0-170-132:3330189:3330668 [6] NCCL INFO Channel 18/0 : 2[6] -> 1[5] via P2P/IPC
+ip-26-0-170-132:3330188:3330669 [5] NCCL INFO Channel 19/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-162-14:2525076:2525552 [1] NCCL INFO Channel 05/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-162-14:2525077:2525553 [2] NCCL INFO Channel 07/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-170-132:3330189:3330668 [6] NCCL INFO Channel 19/0 : 2[6] -> 1[5] via P2P/IPC
+ip-26-0-170-31:3197187:3197764 [7] NCCL INFO Channel 02/0 : 3[7] -> 2[6] via P2P/IPC
+ip-26-0-170-132:3330188:3330669 [5] NCCL INFO Channel 20/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-163-58:934727:935188 [7] NCCL INFO Channel 11/0 : 3[7] -> 2[6] via P2P/IPC
+ip-26-0-162-14:2525076:2525552 [1] NCCL INFO Channel 06/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-163-58:934727:935188 [7] NCCL INFO Channel 12/0 : 3[7] -> 2[6] via P2P/IPC
+ip-26-0-162-14:2525077:2525553 [2] NCCL INFO Channel 08/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-170-132:3330189:3330668 [6] NCCL INFO Channel 20/0 : 2[6] -> 1[5] via P2P/IPC
+ip-26-0-170-132:3330188:3330669 [5] NCCL INFO Channel 21/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-170-132:3330189:3330668 [6] NCCL INFO Channel 21/0 : 2[6] -> 1[5] via P2P/IPC
+ip-26-0-170-132:3330188:3330669 [5] NCCL INFO Channel 22/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-163-58:934727:935188 [7] NCCL INFO Channel 13/0 : 3[7] -> 2[6] via P2P/IPC
+ip-26-0-162-14:2525076:2525552 [1] NCCL INFO Channel 07/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-170-31:3197187:3197764 [7] NCCL INFO Channel 03/0 : 3[7] -> 2[6] via P2P/IPC
+ip-26-0-162-14:2525077:2525553 [2] NCCL INFO Channel 09/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-170-132:3330189:3330668 [6] NCCL INFO Channel 22/0 : 2[6] -> 1[5] via P2P/IPC
+ip-26-0-170-132:3330188:3330669 [5] NCCL INFO Channel 23/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-163-58:934727:935188 [7] NCCL INFO Channel 14/0 : 3[7] -> 2[6] via P2P/IPC
+ip-26-0-162-14:2525076:2525552 [1] NCCL INFO Channel 08/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-162-14:2525077:2525553 [2] NCCL INFO Channel 10/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-170-31:3197187:3197764 [7] NCCL INFO Channel 04/0 : 3[7] -> 2[6] via P2P/IPC
+ip-26-0-163-58:934727:935188 [7] NCCL INFO Channel 15/0 : 3[7] -> 2[6] via P2P/IPC
+ip-26-0-170-132:3330183:3330662 [0] NCCL INFO Connected all trees
+ip-26-0-162-14:2525076:2525552 [1] NCCL INFO Channel 09/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330183:3330662 [0] NCCL INFO NVLS comm 0xa828b10 headRank 0 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-162-14:2525077:2525553 [2] NCCL INFO Channel 11/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-170-132:3330189:3330668 [6] NCCL INFO Channel 23/0 : 2[6] -> 1[5] via P2P/IPC
+ip-26-0-170-31:3197187:3197764 [7] NCCL INFO Channel 05/0 : 3[7] -> 2[6] via P2P/IPC
+ip-26-0-163-58:934727:935188 [7] NCCL INFO Channel 16/0 : 3[7] -> 2[6] via P2P/IPC
+ip-26-0-162-14:2525076:2525552 [1] NCCL INFO Channel 10/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-163-58:934727:935188 [7] NCCL INFO Channel 17/0 : 3[7] -> 2[6] via P2P/IPC
+ip-26-0-162-14:2525077:2525553 [2] NCCL INFO Channel 12/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-170-31:3197187:3197764 [7] NCCL INFO Channel 06/0 : 3[7] -> 2[6] via P2P/IPC
+ip-26-0-162-14:2525079:2525546 [4] NCCL INFO Connected all trees
+ip-26-0-162-14:2525076:2525552 [1] NCCL INFO Channel 11/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-162-14:2525079:2525546 [4] NCCL INFO NVLS comm 0xabe60b0 headRank 0 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-170-31:3197187:3197764 [7] NCCL INFO Channel 07/0 : 3[7] -> 2[6] via P2P/IPC
+ip-26-0-162-14:2525077:2525553 [2] NCCL INFO Channel 13/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-163-58:934727:935188 [7] NCCL INFO Channel 18/0 : 3[7] -> 2[6] via P2P/IPC
+ip-26-0-162-14:2525076:2525552 [1] NCCL INFO Channel 12/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-162-14:2525082:2525548 [7] NCCL INFO Connected all trees
+ip-26-0-162-14:2525082:2525548 [7] NCCL INFO NVLS comm 0xb3960f0 headRank 3 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-162-14:2525077:2525553 [2] NCCL INFO Channel 14/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-163-58:934727:935188 [7] NCCL INFO Channel 19/0 : 3[7] -> 2[6] via P2P/IPC
+ip-26-0-162-14:2525077:2525553 [2] NCCL INFO Channel 15/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-162-14:2525076:2525552 [1] NCCL INFO Channel 13/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-162-14:2525076:2525552 [1] NCCL INFO Channel 14/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-163-58:934727:935188 [7] NCCL INFO Channel 20/0 : 3[7] -> 2[6] via P2P/IPC
+ip-26-0-162-14:2525076:2525552 [1] NCCL INFO Channel 15/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-170-31:3197187:3197764 [7] NCCL INFO Channel 08/0 : 3[7] -> 2[6] via P2P/IPC
+ip-26-0-170-31:3197183:3197765 [3] NCCL INFO Connected all rings
+ip-26-0-170-31:3197183:3197765 [3] NCCL INFO Channel 00/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-162-14:2525077:2525553 [2] NCCL INFO Channel 16/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-163-58:934727:935188 [7] NCCL INFO Channel 21/0 : 3[7] -> 2[6] via P2P/IPC
+ip-26-0-162-14:2525076:2525552 [1] NCCL INFO Channel 16/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-170-31:3197187:3197764 [7] NCCL INFO Channel 09/0 : 3[7] -> 2[6] via P2P/IPC
+ip-26-0-162-14:2525077:2525553 [2] NCCL INFO Channel 17/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-170-31:3197183:3197765 [3] NCCL INFO Channel 01/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-162-14:2525076:2525552 [1] NCCL INFO Channel 17/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-163-58:934726:935189 [6] NCCL INFO Channel 00/0 : 2[6] -> 1[5] via P2P/IPC
+ip-26-0-170-31:3197181:3197768 [1] NCCL INFO Connected all rings
+ip-26-0-163-58:934726:935189 [6] NCCL INFO Channel 01/0 : 2[6] -> 1[5] via P2P/IPC
+ip-26-0-163-58:934727:935188 [7] NCCL INFO Channel 22/0 : 3[7] -> 2[6] via P2P/IPC
+ip-26-0-162-14:2525077:2525553 [2] NCCL INFO Channel 18/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-170-31:3197187:3197764 [7] NCCL INFO Channel 10/0 : 3[7] -> 2[6] via P2P/IPC
+ip-26-0-170-31:3197183:3197765 [3] NCCL INFO Channel 02/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-170-31:3197187:3197764 [7] NCCL INFO Channel 11/0 : 3[7] -> 2[6] via P2P/IPC
+ip-26-0-170-31:3197183:3197765 [3] NCCL INFO Channel 03/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-170-31:3197187:3197764 [7] NCCL INFO Channel 12/0 : 3[7] -> 2[6] via P2P/IPC
+ip-26-0-170-31:3197183:3197765 [3] NCCL INFO Channel 04/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-163-58:934726:935189 [6] NCCL INFO Channel 02/0 : 2[6] -> 1[5] via P2P/IPC
+ip-26-0-162-14:2525077:2525553 [2] NCCL INFO Channel 19/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-162-14:2525076:2525552 [1] NCCL INFO Channel 18/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-170-31:3197187:3197764 [7] NCCL INFO Channel 13/0 : 3[7] -> 2[6] via P2P/IPC
+ip-26-0-163-58:934722:935192 [2] NCCL INFO Connected all rings
+ip-26-0-170-31:3197183:3197765 [3] NCCL INFO Channel 05/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-162-14:2525076:2525552 [1] NCCL INFO Channel 19/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-162-14:2525077:2525553 [2] NCCL INFO Channel 20/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-163-58:934727:935188 [7] NCCL INFO Channel 23/0 : 3[7] -> 2[6] via P2P/IPC
+ip-26-0-170-31:3197187:3197764 [7] NCCL INFO Channel 14/0 : 3[7] -> 2[6] via P2P/IPC
+ip-26-0-163-58:934726:935189 [6] NCCL INFO Channel 03/0 : 2[6] -> 1[5] via P2P/IPC
+ip-26-0-163-58:934726:935189 [6] NCCL INFO Channel 04/0 : 2[6] -> 1[5] via P2P/IPC
+ip-26-0-162-14:2525077:2525553 [2] NCCL INFO Channel 21/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-162-14:2525076:2525552 [1] NCCL INFO Channel 20/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-170-31:3197183:3197765 [3] NCCL INFO Channel 06/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-170-31:3197187:3197764 [7] NCCL INFO Channel 15/0 : 3[7] -> 2[6] via P2P/IPC
+ip-26-0-170-31:3197183:3197765 [3] NCCL INFO Channel 07/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-162-14:2525077:2525553 [2] NCCL INFO Channel 22/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-162-14:2525076:2525552 [1] NCCL INFO Channel 21/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-170-31:3197187:3197764 [7] NCCL INFO Channel 16/0 : 3[7] -> 2[6] via P2P/IPC
+ip-26-0-163-58:934726:935189 [6] NCCL INFO Channel 05/0 : 2[6] -> 1[5] via P2P/IPC
+ip-26-0-162-14:2525081:2525549 [6] NCCL INFO Connected all trees
+ip-26-0-162-14:2525081:2525549 [6] NCCL INFO NVLS comm 0xaa18180 headRank 2 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-170-31:3197180:3197763 [0] NCCL INFO Connected all rings
+ip-26-0-162-14:2525077:2525553 [2] NCCL INFO Channel 23/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-170-31:3197187:3197764 [7] NCCL INFO Channel 17/0 : 3[7] -> 2[6] via P2P/IPC
+ip-26-0-162-14:2525076:2525552 [1] NCCL INFO Channel 22/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-170-31:3197183:3197765 [3] NCCL INFO Channel 08/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-170-31:3197187:3197764 [7] NCCL INFO Channel 18/0 : 3[7] -> 2[6] via P2P/IPC
+ip-26-0-170-31:3197183:3197765 [3] NCCL INFO Channel 09/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-162-14:2525076:2525552 [1] NCCL INFO Channel 23/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-170-31:3197182:3197769 [2] NCCL INFO Connected all rings
+ip-26-0-170-31:3197187:3197764 [7] NCCL INFO Channel 19/0 : 3[7] -> 2[6] via P2P/IPC
+ip-26-0-170-31:3197187:3197764 [7] NCCL INFO Channel 20/0 : 3[7] -> 2[6] via P2P/IPC
+ip-26-0-170-31:3197185:3197767 [5] NCCL INFO Channel 00/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-170-31:3197183:3197765 [3] NCCL INFO Channel 10/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-163-58:934721:935193 [1] NCCL INFO Connected all rings
+ip-26-0-163-58:934726:935189 [6] NCCL INFO Channel 06/0 : 2[6] -> 1[5] via P2P/IPC
+ip-26-0-163-58:934725:935190 [5] NCCL INFO Channel 00/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-162-14:2525080:2525550 [5] NCCL INFO Connected all trees
+ip-26-0-162-14:2525080:2525550 [5] NCCL INFO NVLS comm 0xb16a6a0 headRank 1 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-170-31:3197187:3197764 [7] NCCL INFO Channel 21/0 : 3[7] -> 2[6] via P2P/IPC
+ip-26-0-170-31:3197185:3197767 [5] NCCL INFO Channel 01/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-170-31:3197183:3197765 [3] NCCL INFO Channel 11/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-170-31:3197187:3197764 [7] NCCL INFO Channel 22/0 : 3[7] -> 2[6] via P2P/IPC
+ip-26-0-170-31:3197185:3197767 [5] NCCL INFO Channel 02/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-170-31:3197183:3197765 [3] NCCL INFO Channel 12/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-170-31:3197187:3197764 [7] NCCL INFO Channel 23/0 : 3[7] -> 2[6] via P2P/IPC
+ip-26-0-170-31:3197186:3197766 [6] NCCL INFO Channel 00/0 : 2[6] -> 1[5] via P2P/IPC
+ip-26-0-170-31:3197185:3197767 [5] NCCL INFO Channel 03/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-170-31:3197183:3197765 [3] NCCL INFO Channel 13/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-163-58:934726:935189 [6] NCCL INFO Channel 07/0 : 2[6] -> 1[5] via P2P/IPC
+ip-26-0-170-31:3197185:3197767 [5] NCCL INFO Channel 04/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-170-31:3197186:3197766 [6] NCCL INFO Channel 01/0 : 2[6] -> 1[5] via P2P/IPC
+ip-26-0-170-31:3197183:3197765 [3] NCCL INFO Channel 14/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-163-58:934726:935189 [6] NCCL INFO Channel 08/0 : 2[6] -> 1[5] via P2P/IPC
+ip-26-0-170-31:3197185:3197767 [5] NCCL INFO Channel 05/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-170-31:3197186:3197766 [6] NCCL INFO Channel 02/0 : 2[6] -> 1[5] via P2P/IPC
+ip-26-0-170-31:3197183:3197765 [3] NCCL INFO Channel 15/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-163-58:934725:935190 [5] NCCL INFO Channel 01/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-163-58:934720:935191 [0] NCCL INFO Connected all rings
+ip-26-0-170-31:3197185:3197767 [5] NCCL INFO Channel 06/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-170-31:3197186:3197766 [6] NCCL INFO Channel 03/0 : 2[6] -> 1[5] via P2P/IPC
+ip-26-0-170-31:3197183:3197765 [3] NCCL INFO Channel 16/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-163-58:934726:935189 [6] NCCL INFO Channel 09/0 : 2[6] -> 1[5] via P2P/IPC
+ip-26-0-170-31:3197186:3197766 [6] NCCL INFO Channel 04/0 : 2[6] -> 1[5] via P2P/IPC
+ip-26-0-170-31:3197185:3197767 [5] NCCL INFO Channel 07/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-170-31:3197183:3197765 [3] NCCL INFO Channel 17/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-163-58:934723:935194 [3] NCCL INFO Connected all rings
+ip-26-0-163-58:934723:935194 [3] NCCL INFO Channel 00/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-163-58:934726:935189 [6] NCCL INFO Channel 10/0 : 2[6] -> 1[5] via P2P/IPC
+ip-26-0-163-58:934723:935194 [3] NCCL INFO Channel 01/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-170-31:3197186:3197766 [6] NCCL INFO Channel 05/0 : 2[6] -> 1[5] via P2P/IPC
+ip-26-0-170-31:3197185:3197767 [5] NCCL INFO Channel 08/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-170-31:3197183:3197765 [3] NCCL INFO Channel 18/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-163-58:934725:935190 [5] NCCL INFO Channel 02/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-163-58:934726:935189 [6] NCCL INFO Channel 11/0 : 2[6] -> 1[5] via P2P/IPC
+ip-26-0-170-31:3197186:3197766 [6] NCCL INFO Channel 06/0 : 2[6] -> 1[5] via P2P/IPC
+ip-26-0-163-58:934725:935190 [5] NCCL INFO Channel 03/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-163-58:934726:935189 [6] NCCL INFO Channel 12/0 : 2[6] -> 1[5] via P2P/IPC
+ip-26-0-163-58:934725:935190 [5] NCCL INFO Channel 04/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-163-58:934726:935189 [6] NCCL INFO Channel 13/0 : 2[6] -> 1[5] via P2P/IPC
+ip-26-0-170-31:3197186:3197766 [6] NCCL INFO Channel 07/0 : 2[6] -> 1[5] via P2P/IPC
+ip-26-0-163-58:934725:935190 [5] NCCL INFO Channel 05/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-163-58:934726:935189 [6] NCCL INFO Channel 14/0 : 2[6] -> 1[5] via P2P/IPC
+ip-26-0-170-31:3197185:3197767 [5] NCCL INFO Channel 09/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-163-58:934725:935190 [5] NCCL INFO Channel 06/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-163-58:934726:935189 [6] NCCL INFO Channel 15/0 : 2[6] -> 1[5] via P2P/IPC
+ip-26-0-170-31:3197183:3197765 [3] NCCL INFO Channel 19/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-170-31:3197186:3197766 [6] NCCL INFO Channel 08/0 : 2[6] -> 1[5] via P2P/IPC
+ip-26-0-163-58:934725:935190 [5] NCCL INFO Channel 07/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-163-58:934726:935189 [6] NCCL INFO Channel 16/0 : 2[6] -> 1[5] via P2P/IPC
+ip-26-0-170-31:3197185:3197767 [5] NCCL INFO Channel 10/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-163-58:934723:935194 [3] NCCL INFO Channel 02/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-163-58:934725:935190 [5] NCCL INFO Channel 08/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-163-58:934726:935189 [6] NCCL INFO Channel 17/0 : 2[6] -> 1[5] via P2P/IPC
+ip-26-0-163-58:934723:935194 [3] NCCL INFO Channel 03/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-170-31:3197183:3197765 [3] NCCL INFO Channel 20/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-170-31:3197186:3197766 [6] NCCL INFO Channel 09/0 : 2[6] -> 1[5] via P2P/IPC
+ip-26-0-163-58:934725:935190 [5] NCCL INFO Channel 09/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-163-58:934726:935189 [6] NCCL INFO Channel 18/0 : 2[6] -> 1[5] via P2P/IPC
+ip-26-0-163-58:934723:935194 [3] NCCL INFO Channel 04/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-170-31:3197185:3197767 [5] NCCL INFO Channel 11/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-170-31:3197183:3197765 [3] NCCL INFO Channel 21/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-163-58:934725:935190 [5] NCCL INFO Channel 10/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-163-58:934726:935189 [6] NCCL INFO Channel 19/0 : 2[6] -> 1[5] via P2P/IPC
+ip-26-0-163-58:934723:935194 [3] NCCL INFO Channel 05/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-170-31:3197185:3197767 [5] NCCL INFO Channel 12/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-170-31:3197183:3197765 [3] NCCL INFO Channel 22/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-163-58:934725:935190 [5] NCCL INFO Channel 11/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-163-58:934726:935189 [6] NCCL INFO Channel 20/0 : 2[6] -> 1[5] via P2P/IPC
+ip-26-0-163-58:934723:935194 [3] NCCL INFO Channel 06/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-170-31:3197185:3197767 [5] NCCL INFO Channel 13/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-170-31:3197183:3197765 [3] NCCL INFO Channel 23/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-163-58:934725:935190 [5] NCCL INFO Channel 12/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-163-58:934726:935189 [6] NCCL INFO Channel 21/0 : 2[6] -> 1[5] via P2P/IPC
+ip-26-0-163-58:934723:935194 [3] NCCL INFO Channel 07/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-170-31:3197186:3197766 [6] NCCL INFO Channel 10/0 : 2[6] -> 1[5] via P2P/IPC
+ip-26-0-170-31:3197181:3197768 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-170-31:3197185:3197767 [5] NCCL INFO Channel 14/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-170-31:3197186:3197766 [6] NCCL INFO Channel 11/0 : 2[6] -> 1[5] via P2P/IPC
+ip-26-0-170-31:3197185:3197767 [5] NCCL INFO Channel 15/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-170-31:3197181:3197768 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-163-58:934725:935190 [5] NCCL INFO Channel 13/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-163-58:934726:935189 [6] NCCL INFO Channel 22/0 : 2[6] -> 1[5] via P2P/IPC
+ip-26-0-163-58:934723:935194 [3] NCCL INFO Channel 08/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-170-31:3197185:3197767 [5] NCCL INFO Channel 16/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-170-31:3197186:3197766 [6] NCCL INFO Channel 12/0 : 2[6] -> 1[5] via P2P/IPC
+ip-26-0-170-31:3197181:3197768 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-163-58:934725:935190 [5] NCCL INFO Channel 14/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-163-58:934726:935189 [6] NCCL INFO Channel 23/0 : 2[6] -> 1[5] via P2P/IPC
+ip-26-0-163-58:934723:935194 [3] NCCL INFO Channel 09/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-170-31:3197185:3197767 [5] NCCL INFO Channel 17/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-170-31:3197186:3197766 [6] NCCL INFO Channel 13/0 : 2[6] -> 1[5] via P2P/IPC
+ip-26-0-170-31:3197181:3197768 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-163-58:934725:935190 [5] NCCL INFO Channel 15/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-163-58:934723:935194 [3] NCCL INFO Channel 10/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-170-31:3197186:3197766 [6] NCCL INFO Channel 14/0 : 2[6] -> 1[5] via P2P/IPC
+ip-26-0-170-31:3197185:3197767 [5] NCCL INFO Channel 18/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-170-31:3197181:3197768 [1] NCCL INFO Channel 04/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-163-58:934725:935190 [5] NCCL INFO Channel 16/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-163-58:934723:935194 [3] NCCL INFO Channel 11/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-170-31:3197186:3197766 [6] NCCL INFO Channel 15/0 : 2[6] -> 1[5] via P2P/IPC
+ip-26-0-170-31:3197185:3197767 [5] NCCL INFO Channel 19/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-170-31:3197181:3197768 [1] NCCL INFO Channel 05/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-170-31:3197186:3197766 [6] NCCL INFO Channel 16/0 : 2[6] -> 1[5] via P2P/IPC
+ip-26-0-170-31:3197185:3197767 [5] NCCL INFO Channel 20/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-170-31:3197181:3197768 [1] NCCL INFO Channel 06/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-163-58:934725:935190 [5] NCCL INFO Channel 17/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-163-58:934723:935194 [3] NCCL INFO Channel 12/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-170-31:3197186:3197766 [6] NCCL INFO Channel 17/0 : 2[6] -> 1[5] via P2P/IPC
+ip-26-0-170-31:3197181:3197768 [1] NCCL INFO Channel 07/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-170-31:3197185:3197767 [5] NCCL INFO Channel 21/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-163-58:934725:935190 [5] NCCL INFO Channel 18/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-163-58:934723:935194 [3] NCCL INFO Channel 13/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-170-31:3197186:3197766 [6] NCCL INFO Channel 18/0 : 2[6] -> 1[5] via P2P/IPC
+ip-26-0-170-31:3197181:3197768 [1] NCCL INFO Channel 08/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-170-31:3197185:3197767 [5] NCCL INFO Channel 22/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-170-31:3197186:3197766 [6] NCCL INFO Channel 19/0 : 2[6] -> 1[5] via P2P/IPC
+ip-26-0-170-31:3197185:3197767 [5] NCCL INFO Channel 23/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-170-31:3197181:3197768 [1] NCCL INFO Channel 09/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-170-31:3197182:3197769 [2] NCCL INFO Channel 00/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-163-58:934722:935192 [2] NCCL INFO Channel 00/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-163-58:934725:935190 [5] NCCL INFO Channel 19/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-163-58:934723:935194 [3] NCCL INFO Channel 14/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-170-31:3197181:3197768 [1] NCCL INFO Channel 10/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-170-31:3197186:3197766 [6] NCCL INFO Channel 20/0 : 2[6] -> 1[5] via P2P/IPC
+ip-26-0-170-31:3197182:3197769 [2] NCCL INFO Channel 01/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-170-31:3197186:3197766 [6] NCCL INFO Channel 21/0 : 2[6] -> 1[5] via P2P/IPC
+ip-26-0-170-31:3197181:3197768 [1] NCCL INFO Channel 11/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-170-31:3197182:3197769 [2] NCCL INFO Channel 02/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-163-58:934722:935192 [2] NCCL INFO Channel 01/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-163-58:934722:935192 [2] NCCL INFO Channel 02/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-163-58:934722:935192 [2] NCCL INFO Channel 03/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-163-58:934725:935190 [5] NCCL INFO Channel 20/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-163-58:934722:935192 [2] NCCL INFO Channel 04/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-170-31:3197181:3197768 [1] NCCL INFO Channel 12/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-170-31:3197186:3197766 [6] NCCL INFO Channel 22/0 : 2[6] -> 1[5] via P2P/IPC
+ip-26-0-163-58:934725:935190 [5] NCCL INFO Channel 21/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-163-58:934723:935194 [3] NCCL INFO Channel 15/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-170-31:3197182:3197769 [2] NCCL INFO Channel 03/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-163-58:934725:935190 [5] NCCL INFO Channel 22/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-163-58:934722:935192 [2] NCCL INFO Channel 05/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-163-58:934723:935194 [3] NCCL INFO Channel 16/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-170-31:3197181:3197768 [1] NCCL INFO Channel 13/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-163-58:934725:935190 [5] NCCL INFO Channel 23/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-163-58:934722:935192 [2] NCCL INFO Channel 06/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-170-31:3197186:3197766 [6] NCCL INFO Channel 23/0 : 2[6] -> 1[5] via P2P/IPC
+ip-26-0-163-58:934723:935194 [3] NCCL INFO Channel 17/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-163-58:934722:935192 [2] NCCL INFO Channel 07/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-163-58:934723:935194 [3] NCCL INFO Channel 18/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-170-132:3330184:3330665 [1] NCCL INFO Connected all trees
+ip-26-0-170-132:3330184:3330665 [1] NCCL INFO NVLS comm 0xa91e090 headRank 1 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-170-31:3197181:3197768 [1] NCCL INFO Channel 14/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-162-14:2525082:2525548 [7] NCCL INFO Connected NVLS tree
+ip-26-0-162-14:2525082:2525548 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-162-14:2525082:2525548 [7] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-162-14:2525082:2525548 [7] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-162-14:2525080:2525550 [5] NCCL INFO Connected NVLS tree
+ip-26-0-162-14:2525080:2525550 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-162-14:2525080:2525550 [5] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-162-14:2525080:2525550 [5] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-162-14:2525081:2525549 [6] NCCL INFO Connected NVLS tree
+ip-26-0-162-14:2525081:2525549 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-162-14:2525081:2525549 [6] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-162-14:2525081:2525549 [6] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-162-14:2525079:2525546 [4] NCCL INFO Connected NVLS tree
+ip-26-0-162-14:2525079:2525546 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-162-14:2525079:2525546 [4] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-162-14:2525079:2525546 [4] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-170-31:3197181:3197768 [1] NCCL INFO Channel 15/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-170-31:3197181:3197768 [1] NCCL INFO Channel 16/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-170-31:3197181:3197768 [1] NCCL INFO Channel 17/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-170-31:3197182:3197769 [2] NCCL INFO Channel 04/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-163-58:934722:935192 [2] NCCL INFO Channel 08/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-163-58:934723:935194 [3] NCCL INFO Channel 19/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-170-31:3197181:3197768 [1] NCCL INFO Channel 18/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-163-58:934721:935193 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-163-58:934721:935193 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-163-58:934721:935193 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330185:3330664 [2] NCCL INFO Connected all trees
+ip-26-0-170-132:3330185:3330664 [2] NCCL INFO NVLS comm 0xadb09f0 headRank 2 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-163-58:934722:935192 [2] NCCL INFO Channel 09/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-163-58:934723:935194 [3] NCCL INFO Channel 20/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-170-132:3330187:3330667 [4] NCCL INFO Connected all trees
+ip-26-0-170-132:3330187:3330667 [4] NCCL INFO NVLS comm 0xad482e0 headRank 0 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-163-58:934721:935193 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-170-31:3197182:3197769 [2] NCCL INFO Channel 05/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-170-31:3197181:3197768 [1] NCCL INFO Channel 19/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330190:3330670 [7] NCCL INFO Connected all trees
+ip-26-0-170-132:3330190:3330670 [7] NCCL INFO NVLS comm 0xb86f370 headRank 3 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-170-31:3197181:3197768 [1] NCCL INFO Channel 20/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-170-31:3197181:3197768 [1] NCCL INFO Channel 21/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-170-31:3197181:3197768 [1] NCCL INFO Channel 22/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-170-31:3197182:3197769 [2] NCCL INFO Channel 06/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-163-58:934722:935192 [2] NCCL INFO Channel 10/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-162-14:2525079:2525546 [4] NCCL INFO comm 0xabe60b0 rank 0 nranks 4 cudaDev 4 nvmlDev 4 busId 97000 commId 0xe0ce9c074ef535b1 - Init COMPLETE
+ip-26-0-162-14:2525081:2525549 [6] NCCL INFO comm 0xaa18180 rank 2 nranks 4 cudaDev 6 nvmlDev 6 busId b9000 commId 0xe0ce9c074ef535b1 - Init COMPLETE
+ip-26-0-162-14:2525082:2525548 [7] NCCL INFO comm 0xb3960f0 rank 3 nranks 4 cudaDev 7 nvmlDev 7 busId ca000 commId 0xe0ce9c074ef535b1 - Init COMPLETE
+ip-26-0-162-14:2525080:2525550 [5] NCCL INFO comm 0xb16a6a0 rank 1 nranks 4 cudaDev 5 nvmlDev 5 busId a8000 commId 0xe0ce9c074ef535b1 - Init COMPLETE
+ip-26-0-163-58:934723:935194 [3] NCCL INFO Channel 21/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-163-58:934721:935193 [1] NCCL INFO Channel 04/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-170-31:3197181:3197768 [1] NCCL INFO Channel 23/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-170-31:3197182:3197769 [2] NCCL INFO Channel 07/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-163-58:934722:935192 [2] NCCL INFO Channel 11/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-163-58:934723:935194 [3] NCCL INFO Channel 22/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-163-58:934721:935193 [1] NCCL INFO Channel 05/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-170-31:3197182:3197769 [2] NCCL INFO Channel 08/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-163-58:934722:935192 [2] NCCL INFO Channel 12/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-163-58:934723:935194 [3] NCCL INFO Channel 23/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-163-58:934721:935193 [1] NCCL INFO Channel 06/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-170-31:3197182:3197769 [2] NCCL INFO Channel 09/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-163-58:934722:935192 [2] NCCL INFO Channel 13/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-170-31:3197182:3197769 [2] NCCL INFO Channel 10/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-170-31:3197182:3197769 [2] NCCL INFO Channel 11/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-170-31:3197182:3197769 [2] NCCL INFO Channel 12/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-163-58:934722:935192 [2] NCCL INFO Channel 14/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-163-58:934721:935193 [1] NCCL INFO Channel 07/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-170-31:3197182:3197769 [2] NCCL INFO Channel 13/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-170-31:3197182:3197769 [2] NCCL INFO Channel 14/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-170-31:3197182:3197769 [2] NCCL INFO Channel 15/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-163-58:934727:935188 [7] NCCL INFO Connected all trees
+ip-26-0-163-58:934722:935192 [2] NCCL INFO Channel 15/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-163-58:934727:935188 [7] NCCL INFO NVLS comm 0xad794e0 headRank 3 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-170-31:3197182:3197769 [2] NCCL INFO Channel 16/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-163-58:934722:935192 [2] NCCL INFO Channel 16/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-163-58:934721:935193 [1] NCCL INFO Channel 08/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-163-58:934721:935193 [1] NCCL INFO Channel 09/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-170-31:3197182:3197769 [2] NCCL INFO Channel 17/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-170-132:3330186:3330663 [3] NCCL INFO Connected NVLS tree
+ip-26-0-170-132:3330186:3330663 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-170-132:3330186:3330663 [3] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-170-132:3330186:3330663 [3] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-162-14:2525078:2525554 [3] NCCL INFO Connected all trees
+ip-26-0-162-14:2525078:2525554 [3] NCCL INFO NVLS comm 0xa7490e0 headRank 3 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-163-58:934722:935192 [2] NCCL INFO Channel 17/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-163-58:934721:935193 [1] NCCL INFO Channel 10/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-170-31:3197184:3197762 [4] NCCL INFO Connected all trees
+ip-26-0-170-31:3197184:3197762 [4] NCCL INFO NVLS comm 0xb411040 headRank 0 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-170-132:3330184:3330665 [1] NCCL INFO Connected NVLS tree
+ip-26-0-170-132:3330184:3330665 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-170-132:3330184:3330665 [1] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-170-132:3330184:3330665 [1] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-170-132:3330185:3330664 [2] NCCL INFO Connected NVLS tree
+ip-26-0-170-132:3330185:3330664 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-170-132:3330185:3330664 [2] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-170-132:3330185:3330664 [2] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-170-132:3330183:3330662 [0] NCCL INFO Connected NVLS tree
+ip-26-0-170-132:3330183:3330662 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-170-132:3330183:3330662 [0] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-170-132:3330183:3330662 [0] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-162-14:2525075:2525551 [0] NCCL INFO Connected all trees
+ip-26-0-162-14:2525075:2525551 [0] NCCL INFO NVLS comm 0xad97080 headRank 0 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-170-31:3197182:3197769 [2] NCCL INFO Channel 18/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-163-58:934721:935193 [1] NCCL INFO Channel 11/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-163-58:934722:935192 [2] NCCL INFO Channel 18/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-170-31:3197182:3197769 [2] NCCL INFO Channel 19/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-163-58:934721:935193 [1] NCCL INFO Channel 12/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-163-58:934722:935192 [2] NCCL INFO Channel 19/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-170-31:3197182:3197769 [2] NCCL INFO Channel 20/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-170-132:3330185:3330664 [2] NCCL INFO comm 0xadb09f0 rank 2 nranks 4 cudaDev 2 nvmlDev 2 busId 75000 commId 0x302943922abb6c5a - Init COMPLETE
+ip-26-0-170-132:3330186:3330663 [3] NCCL INFO comm 0xa269dd0 rank 3 nranks 4 cudaDev 3 nvmlDev 3 busId 86000 commId 0x302943922abb6c5a - Init COMPLETE
+ip-26-0-170-132:3330183:3330662 [0] NCCL INFO comm 0xa828b10 rank 0 nranks 4 cudaDev 0 nvmlDev 0 busId 53000 commId 0x302943922abb6c5a - Init COMPLETE
+ip-26-0-170-132:3330184:3330665 [1] NCCL INFO comm 0xa91e090 rank 1 nranks 4 cudaDev 1 nvmlDev 1 busId 64000 commId 0x302943922abb6c5a - Init COMPLETE
+ip-26-0-170-132:3330189:3330668 [6] NCCL INFO Connected all trees
+ip-26-0-170-132:3330189:3330668 [6] NCCL INFO NVLS comm 0x9b016b0 headRank 2 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-163-58:934722:935192 [2] NCCL INFO Channel 20/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-163-58:934721:935193 [1] NCCL INFO Channel 13/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-170-31:3197187:3197764 [7] NCCL INFO Connected all trees
+ip-26-0-170-31:3197187:3197764 [7] NCCL INFO NVLS comm 0xb973ff0 headRank 3 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-170-132:3330188:3330669 [5] NCCL INFO Connected all trees
+ip-26-0-163-58:934724:935186 [4] NCCL INFO Connected all trees
+ip-26-0-170-132:3330188:3330669 [5] NCCL INFO NVLS comm 0xb841e10 headRank 1 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-163-58:934724:935186 [4] NCCL INFO NVLS comm 0xac186b0 headRank 0 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-163-58:934722:935192 [2] NCCL INFO Channel 21/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-163-58:934721:935193 [1] NCCL INFO Channel 14/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-162-14:2525077:2525553 [2] NCCL INFO Connected all trees
+ip-26-0-170-31:3197182:3197769 [2] NCCL INFO Channel 21/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-162-14:2525077:2525553 [2] NCCL INFO NVLS comm 0xc1217b0 headRank 2 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-163-58:934721:935193 [1] NCCL INFO Channel 15/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-163-58:934722:935192 [2] NCCL INFO Channel 22/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-170-31:3197182:3197769 [2] NCCL INFO Channel 22/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-163-58:934721:935193 [1] NCCL INFO Channel 16/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-163-58:934722:935192 [2] NCCL INFO Channel 23/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-162-14:2525076:2525552 [1] NCCL INFO Connected all trees
+ip-26-0-162-14:2525076:2525552 [1] NCCL INFO NVLS comm 0xb891cc0 headRank 1 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-163-58:934721:935193 [1] NCCL INFO Channel 17/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-163-58:934721:935193 [1] NCCL INFO Channel 18/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-170-31:3197182:3197769 [2] NCCL INFO Channel 23/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-163-58:934721:935193 [1] NCCL INFO Channel 19/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-170-31:3197180:3197763 [0] NCCL INFO Connected all trees
+ip-26-0-170-31:3197180:3197763 [0] NCCL INFO NVLS comm 0xb4ae6f0 headRank 0 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-163-58:934721:935193 [1] NCCL INFO Channel 20/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-163-58:934721:935193 [1] NCCL INFO Channel 21/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330188:3330669 [5] NCCL INFO Connected NVLS tree
+ip-26-0-170-132:3330188:3330669 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-170-132:3330188:3330669 [5] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-170-132:3330188:3330669 [5] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-170-132:3330190:3330670 [7] NCCL INFO Connected NVLS tree
+ip-26-0-170-132:3330190:3330670 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-170-132:3330190:3330670 [7] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-170-132:3330187:3330667 [4] NCCL INFO Connected NVLS tree
+ip-26-0-170-132:3330190:3330670 [7] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-170-132:3330189:3330668 [6] NCCL INFO Connected NVLS tree
+ip-26-0-170-132:3330189:3330668 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-170-132:3330187:3330667 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-170-132:3330189:3330668 [6] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-170-132:3330189:3330668 [6] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-170-132:3330187:3330667 [4] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-170-132:3330187:3330667 [4] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-163-58:934721:935193 [1] NCCL INFO Channel 22/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-163-58:934721:935193 [1] NCCL INFO Channel 23/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330190:3330670 [7] NCCL INFO comm 0xb86f370 rank 3 nranks 4 cudaDev 7 nvmlDev 7 busId ca000 commId 0xecbe8b333a0a7265 - Init COMPLETE
+ip-26-0-170-132:3330189:3330668 [6] NCCL INFO comm 0x9b016b0 rank 2 nranks 4 cudaDev 6 nvmlDev 6 busId b9000 commId 0xecbe8b333a0a7265 - Init COMPLETE
+ip-26-0-170-132:3330188:3330669 [5] NCCL INFO comm 0xb841e10 rank 1 nranks 4 cudaDev 5 nvmlDev 5 busId a8000 commId 0xecbe8b333a0a7265 - Init COMPLETE
+ip-26-0-170-132:3330187:3330667 [4] NCCL INFO comm 0xad482e0 rank 0 nranks 4 cudaDev 4 nvmlDev 4 busId 97000 commId 0xecbe8b333a0a7265 - Init COMPLETE
+ip-26-0-163-58:934726:935189 [6] NCCL INFO Connected all trees
+ip-26-0-163-58:934726:935189 [6] NCCL INFO NVLS comm 0xb169000 headRank 2 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-162-14:2525077:2525553 [2] NCCL INFO Connected NVLS tree
+ip-26-0-162-14:2525077:2525553 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-162-14:2525077:2525553 [2] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-162-14:2525077:2525553 [2] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-162-14:2525076:2525552 [1] NCCL INFO Connected NVLS tree
+ip-26-0-162-14:2525076:2525552 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-162-14:2525076:2525552 [1] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-162-14:2525076:2525552 [1] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-162-14:2525078:2525554 [3] NCCL INFO Connected NVLS tree
+ip-26-0-162-14:2525078:2525554 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-162-14:2525078:2525554 [3] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-162-14:2525078:2525554 [3] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-162-14:2525075:2525551 [0] NCCL INFO Connected NVLS tree
+ip-26-0-162-14:2525075:2525551 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-162-14:2525075:2525551 [0] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-162-14:2525075:2525551 [0] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-163-58:934725:935190 [5] NCCL INFO Connected all trees
+ip-26-0-163-58:934725:935190 [5] NCCL INFO NVLS comm 0x9e04790 headRank 1 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-162-14:2525078:2525554 [3] NCCL INFO comm 0xa7490e0 rank 3 nranks 4 cudaDev 3 nvmlDev 3 busId 86000 commId 0xf2cf793020f06054 - Init COMPLETE
+ip-26-0-162-14:2525076:2525552 [1] NCCL INFO comm 0xb891cc0 rank 1 nranks 4 cudaDev 1 nvmlDev 1 busId 64000 commId 0xf2cf793020f06054 - Init COMPLETE
+ip-26-0-162-14:2525077:2525553 [2] NCCL INFO comm 0xc1217b0 rank 2 nranks 4 cudaDev 2 nvmlDev 2 busId 75000 commId 0xf2cf793020f06054 - Init COMPLETE
+ip-26-0-162-14:2525075:2525551 [0] NCCL INFO comm 0xad97080 rank 0 nranks 4 cudaDev 0 nvmlDev 0 busId 53000 commId 0xf2cf793020f06054 - Init COMPLETE
+ip-26-0-170-31:3197186:3197766 [6] NCCL INFO Connected all trees
+ip-26-0-170-31:3197186:3197766 [6] NCCL INFO NVLS comm 0xb345070 headRank 2 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-163-58:934726:935189 [6] NCCL INFO Connected NVLS tree
+ip-26-0-163-58:934726:935189 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-163-58:934726:935189 [6] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-163-58:934726:935189 [6] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-163-58:934724:935186 [4] NCCL INFO Connected NVLS tree
+ip-26-0-163-58:934724:935186 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-163-58:934724:935186 [4] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-163-58:934724:935186 [4] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-163-58:934725:935190 [5] NCCL INFO Connected NVLS tree
+ip-26-0-163-58:934725:935190 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-163-58:934725:935190 [5] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-163-58:934725:935190 [5] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-163-58:934727:935188 [7] NCCL INFO Connected NVLS tree
+ip-26-0-163-58:934727:935188 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-163-58:934727:935188 [7] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-163-58:934727:935188 [7] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-170-31:3197185:3197767 [5] NCCL INFO Connected all trees
+ip-26-0-170-31:3197185:3197767 [5] NCCL INFO NVLS comm 0x9b5cf70 headRank 1 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-163-58:934723:935194 [3] NCCL INFO Connected all trees
+ip-26-0-163-58:934723:935194 [3] NCCL INFO NVLS comm 0xae236c0 headRank 3 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-163-58:934726:935189 [6] NCCL INFO comm 0xb169000 rank 2 nranks 4 cudaDev 6 nvmlDev 6 busId b9000 commId 0xbda347e6977b7d76 - Init COMPLETE
+ip-26-0-163-58:934724:935186 [4] NCCL INFO comm 0xac186b0 rank 0 nranks 4 cudaDev 4 nvmlDev 4 busId 97000 commId 0xbda347e6977b7d76 - Init COMPLETE
+ip-26-0-163-58:934725:935190 [5] NCCL INFO comm 0x9e04790 rank 1 nranks 4 cudaDev 5 nvmlDev 5 busId a8000 commId 0xbda347e6977b7d76 - Init COMPLETE
+ip-26-0-163-58:934727:935188 [7] NCCL INFO comm 0xad794e0 rank 3 nranks 4 cudaDev 7 nvmlDev 7 busId ca000 commId 0xbda347e6977b7d76 - Init COMPLETE
+ip-26-0-170-31:3197183:3197765 [3] NCCL INFO Connected all trees
+ip-26-0-170-31:3197183:3197765 [3] NCCL INFO NVLS comm 0xb27e620 headRank 3 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-170-31:3197187:3197764 [7] NCCL INFO Connected NVLS tree
+ip-26-0-170-31:3197187:3197764 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-170-31:3197187:3197764 [7] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-170-31:3197187:3197764 [7] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-170-31:3197186:3197766 [6] NCCL INFO Connected NVLS tree
+ip-26-0-170-31:3197186:3197766 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-170-31:3197186:3197766 [6] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-170-31:3197186:3197766 [6] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-170-31:3197185:3197767 [5] NCCL INFO Connected NVLS tree
+ip-26-0-170-31:3197185:3197767 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-170-31:3197185:3197767 [5] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-170-31:3197185:3197767 [5] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-170-31:3197184:3197762 [4] NCCL INFO Connected NVLS tree
+ip-26-0-170-31:3197184:3197762 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-170-31:3197184:3197762 [4] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-170-31:3197184:3197762 [4] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-163-58:934720:935191 [0] NCCL INFO Connected all trees
+ip-26-0-163-58:934720:935191 [0] NCCL INFO NVLS comm 0xab5f260 headRank 0 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-170-31:3197185:3197767 [5] NCCL INFO comm 0x9b5cf70 rank 1 nranks 4 cudaDev 5 nvmlDev 5 busId a8000 commId 0xe5acb1f64c970582 - Init COMPLETE
+ip-26-0-170-31:3197187:3197764 [7] NCCL INFO comm 0xb973ff0 rank 3 nranks 4 cudaDev 7 nvmlDev 7 busId ca000 commId 0xe5acb1f64c970582 - Init COMPLETE
+ip-26-0-170-31:3197184:3197762 [4] NCCL INFO comm 0xb411040 rank 0 nranks 4 cudaDev 4 nvmlDev 4 busId 97000 commId 0xe5acb1f64c970582 - Init COMPLETE
+ip-26-0-170-31:3197186:3197766 [6] NCCL INFO comm 0xb345070 rank 2 nranks 4 cudaDev 6 nvmlDev 6 busId b9000 commId 0xe5acb1f64c970582 - Init COMPLETE
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+ip-26-0-162-14:2525080:2525583 [5] NCCL INFO Using network Libfabric
+ip-26-0-162-14:2525079:2525580 [4] NCCL INFO Using network Libfabric
+ip-26-0-163-58:934724:935213 [4] NCCL INFO Using network Libfabric
+ip-26-0-170-31:3197184:3197789 [4] NCCL INFO Using network Libfabric
+ip-26-0-170-132:3330187:3330694 [4] NCCL INFO Using network Libfabric
+ip-26-0-170-31:3197187:3197788 [7] NCCL INFO Using network Libfabric
+ip-26-0-162-14:2525081:2525584 [6] NCCL INFO Using network Libfabric
+ip-26-0-170-132:3330188:3330696 [5] NCCL INFO Using network Libfabric
+ip-26-0-162-14:2525082:2525581 [7] NCCL INFO Using network Libfabric
+ip-26-0-170-132:3330190:3330695 [7] NCCL INFO Using network Libfabric
+ip-26-0-170-31:3197186:3197791 [6] NCCL INFO Using network Libfabric
+ip-26-0-170-31:3197185:3197790 [5] NCCL INFO Using network Libfabric
+ip-26-0-163-58:934727:935214 [7] NCCL INFO Using network Libfabric
+ip-26-0-163-58:934725:935215 [5] NCCL INFO Using network Libfabric
+ip-26-0-170-132:3330189:3330697 [6] NCCL INFO Using network Libfabric
+ip-26-0-163-58:934724:935213 [4] NCCL INFO comm 0xb65df60 rank 1 nranks 4 cudaDev 4 nvmlDev 4 busId 97000 commId 0xdb7e12b19ea1b0d1 - Init START
+ip-26-0-170-132:3330187:3330694 [4] NCCL INFO comm 0xb7d2640 rank 2 nranks 4 cudaDev 4 nvmlDev 4 busId 97000 commId 0xdb7e12b19ea1b0d1 - Init START
+ip-26-0-162-14:2525079:2525580 [4] NCCL INFO comm 0xb62b0a0 rank 0 nranks 4 cudaDev 4 nvmlDev 4 busId 97000 commId 0xdb7e12b19ea1b0d1 - Init START
+ip-26-0-170-31:3197184:3197789 [4] NCCL INFO comm 0xbea2d40 rank 3 nranks 4 cudaDev 4 nvmlDev 4 busId 97000 commId 0xdb7e12b19ea1b0d1 - Init START
+ip-26-0-162-14:2525079:2525580 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-163-58:934724:935213 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-170-132:3330187:3330694 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-170-31:3197187:3197788 [7] NCCL INFO comm 0xc405430 rank 3 nranks 4 cudaDev 7 nvmlDev 7 busId ca000 commId 0x434f5bae1e1954ec - Init START
+ip-26-0-170-31:3197184:3197789 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-162-14:2525082:2525581 [7] NCCL INFO comm 0xbddb140 rank 0 nranks 4 cudaDev 7 nvmlDev 7 busId ca000 commId 0x434f5bae1e1954ec - Init START
+ip-26-0-162-14:2525080:2525583 [5] NCCL INFO comm 0xbbafcf0 rank 0 nranks 4 cudaDev 5 nvmlDev 5 busId a8000 commId 0x630d4051fd216a10 - Init START
+ip-26-0-163-58:934727:935214 [7] NCCL INFO comm 0xb7be4b0 rank 1 nranks 4 cudaDev 7 nvmlDev 7 busId ca000 commId 0x434f5bae1e1954ec - Init START
+ip-26-0-170-132:3330190:3330695 [7] NCCL INFO comm 0xc2b5e60 rank 2 nranks 4 cudaDev 7 nvmlDev 7 busId ca000 commId 0x434f5bae1e1954ec - Init START
+ip-26-0-170-31:3197185:3197790 [5] NCCL INFO comm 0xa5ed570 rank 3 nranks 4 cudaDev 5 nvmlDev 5 busId a8000 commId 0x630d4051fd216a10 - Init START
+ip-26-0-163-58:934725:935215 [5] NCCL INFO comm 0xa8498a0 rank 1 nranks 4 cudaDev 5 nvmlDev 5 busId a8000 commId 0x630d4051fd216a10 - Init START
+ip-26-0-170-132:3330188:3330696 [5] NCCL INFO comm 0xc288720 rank 2 nranks 4 cudaDev 5 nvmlDev 5 busId a8000 commId 0x630d4051fd216a10 - Init START
+ip-26-0-170-132:3330190:3330695 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-170-31:3197187:3197788 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-162-14:2525082:2525581 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-170-132:3330188:3330696 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-163-58:934727:935214 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-170-31:3197185:3197790 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-162-14:2525080:2525583 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-163-58:934725:935215 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-163-58:934726:935216 [6] NCCL INFO Using network Libfabric
+ip-26-0-170-31:3197186:3197791 [6] NCCL INFO comm 0xbdd67c0 rank 3 nranks 4 cudaDev 6 nvmlDev 6 busId b9000 commId 0x383416e98fe27c57 - Init START
+ip-26-0-162-14:2525081:2525584 [6] NCCL INFO comm 0xb45d830 rank 0 nranks 4 cudaDev 6 nvmlDev 6 busId b9000 commId 0x383416e98fe27c57 - Init START
+ip-26-0-163-58:934726:935216 [6] NCCL INFO comm 0xbbadcc0 rank 1 nranks 4 cudaDev 6 nvmlDev 6 busId b9000 commId 0x383416e98fe27c57 - Init START
+ip-26-0-170-132:3330189:3330697 [6] NCCL INFO comm 0xa547f50 rank 2 nranks 4 cudaDev 6 nvmlDev 6 busId b9000 commId 0x383416e98fe27c57 - Init START
+ip-26-0-170-132:3330189:3330697 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-170-31:3197186:3197791 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-162-14:2525081:2525584 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-163-58:934726:935216 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-170-31:3197182:3197769 [2] NCCL INFO Connected all trees
+ip-26-0-170-31:3197182:3197769 [2] NCCL INFO NVLS comm 0x9c0ef50 headRank 2 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-170-31:3197181:3197768 [1] NCCL INFO Connected all trees
+ip-26-0-170-31:3197181:3197768 [1] NCCL INFO NVLS comm 0xaed8a00 headRank 1 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-163-58:934721:935193 [1] NCCL INFO Connected all trees
+ip-26-0-163-58:934721:935193 [1] NCCL INFO NVLS comm 0xb8a3740 headRank 1 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-163-58:934722:935192 [2] NCCL INFO Connected all trees
+ip-26-0-163-58:934722:935192 [2] NCCL INFO NVLS comm 0x9e22940 headRank 2 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-170-31:3197181:3197768 [1] NCCL INFO Connected NVLS tree
+ip-26-0-170-31:3197182:3197769 [2] NCCL INFO Connected NVLS tree
+ip-26-0-170-31:3197182:3197769 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-170-31:3197182:3197769 [2] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-170-31:3197182:3197769 [2] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-170-31:3197181:3197768 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-170-31:3197180:3197763 [0] NCCL INFO Connected NVLS tree
+ip-26-0-170-31:3197180:3197763 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-170-31:3197181:3197768 [1] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-170-31:3197181:3197768 [1] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-170-31:3197180:3197763 [0] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-170-31:3197180:3197763 [0] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-170-31:3197183:3197765 [3] NCCL INFO Connected NVLS tree
+ip-26-0-170-31:3197183:3197765 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-170-31:3197183:3197765 [3] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-170-31:3197183:3197765 [3] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-163-58:934722:935192 [2] NCCL INFO Connected NVLS tree
+ip-26-0-163-58:934722:935192 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-163-58:934721:935193 [1] NCCL INFO Connected NVLS tree
+ip-26-0-163-58:934722:935192 [2] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-163-58:934722:935192 [2] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-163-58:934721:935193 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-163-58:934721:935193 [1] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-163-58:934721:935193 [1] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-163-58:934720:935191 [0] NCCL INFO Connected NVLS tree
+ip-26-0-163-58:934720:935191 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-163-58:934720:935191 [0] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-163-58:934720:935191 [0] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-163-58:934723:935194 [3] NCCL INFO Connected NVLS tree
+ip-26-0-163-58:934723:935194 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-163-58:934723:935194 [3] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-163-58:934723:935194 [3] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-170-31:3197182:3197769 [2] NCCL INFO comm 0x9c0ef50 rank 2 nranks 4 cudaDev 2 nvmlDev 2 busId 75000 commId 0xdf4467e4af4c2d67 - Init COMPLETE
+ip-26-0-170-31:3197181:3197768 [1] NCCL INFO comm 0xaed8a00 rank 1 nranks 4 cudaDev 1 nvmlDev 1 busId 64000 commId 0xdf4467e4af4c2d67 - Init COMPLETE
+ip-26-0-170-31:3197180:3197763 [0] NCCL INFO comm 0xb4ae6f0 rank 0 nranks 4 cudaDev 0 nvmlDev 0 busId 53000 commId 0xdf4467e4af4c2d67 - Init COMPLETE
+ip-26-0-170-31:3197183:3197765 [3] NCCL INFO comm 0xb27e620 rank 3 nranks 4 cudaDev 3 nvmlDev 3 busId 86000 commId 0xdf4467e4af4c2d67 - Init COMPLETE
+ip-26-0-163-58:934722:935192 [2] NCCL INFO comm 0x9e22940 rank 2 nranks 4 cudaDev 2 nvmlDev 2 busId 75000 commId 0xa9cee0599e4ded26 - Init COMPLETE
+ip-26-0-163-58:934721:935193 [1] NCCL INFO comm 0xb8a3740 rank 1 nranks 4 cudaDev 1 nvmlDev 1 busId 64000 commId 0xa9cee0599e4ded26 - Init COMPLETE
+ip-26-0-163-58:934720:935191 [0] NCCL INFO comm 0xab5f260 rank 0 nranks 4 cudaDev 0 nvmlDev 0 busId 53000 commId 0xa9cee0599e4ded26 - Init COMPLETE
+ip-26-0-163-58:934723:935194 [3] NCCL INFO comm 0xae236c0 rank 3 nranks 4 cudaDev 3 nvmlDev 3 busId 86000 commId 0xa9cee0599e4ded26 - Init COMPLETE
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+ip-26-0-162-14:2525076:2525588 [1] NCCL INFO Using network Libfabric
+ip-26-0-163-58:934722:935222 [2] NCCL INFO Using network Libfabric
+ip-26-0-170-31:3197182:3197797 [2] NCCL INFO Using network Libfabric
+ip-26-0-170-132:3330185:3330698 [2] NCCL INFO Using network Libfabric
+ip-26-0-162-14:2525077:2525586 [2] NCCL INFO Using network Libfabric
+ip-26-0-170-132:3330184:3330699 [1] NCCL INFO Using network Libfabric
+ip-26-0-170-31:3197181:3197798 [1] NCCL INFO Using network Libfabric
+NCCL version 2.18.5+cuda12.2
+ip-26-0-163-58:934721:935223 [1] NCCL INFO Using network Libfabric
+ip-26-0-170-31:3197182:3197797 [2] NCCL INFO comm 0xa69f1d0 rank 3 nranks 4 cudaDev 2 nvmlDev 2 busId 75000 commId 0x82c4762eb1811da0 - Init START
+ip-26-0-162-14:2525077:2525586 [2] NCCL INFO comm 0xc13bea0 rank 0 nranks 4 cudaDev 2 nvmlDev 2 busId 75000 commId 0x82c4762eb1811da0 - Init START
+ip-26-0-163-58:934722:935222 [2] NCCL INFO comm 0xa867ed0 rank 1 nranks 4 cudaDev 2 nvmlDev 2 busId 75000 commId 0x82c4762eb1811da0 - Init START
+ip-26-0-170-132:3330185:3330698 [2] NCCL INFO comm 0xb83aa70 rank 2 nranks 4 cudaDev 2 nvmlDev 2 busId 75000 commId 0x82c4762eb1811da0 - Init START
+ip-26-0-170-132:3330185:3330698 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-162-14:2525078:2525591 [3] NCCL INFO Using network Libfabric
+ip-26-0-163-58:934722:935222 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-162-14:2525077:2525586 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-170-31:3197180:3197800 [0] NCCL INFO Using network Libfabric
+ip-26-0-170-31:3197182:3197797 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-163-58:934720:935224 [0] NCCL INFO Using network Libfabric
+ip-26-0-170-31:3197183:3197799 [3] NCCL INFO Using network Libfabric
+ip-26-0-162-14:2525075:2525592 [0] NCCL INFO Using network Libfabric
+ip-26-0-162-14:2525076:2525588 [1] NCCL INFO comm 0xc2d65a0 rank 0 nranks 4 cudaDev 1 nvmlDev 1 busId 64000 commId 0xcdbbe7a2eb7417f4 - Init START
+ip-26-0-170-31:3197181:3197798 [1] NCCL INFO comm 0xb969e60 rank 3 nranks 4 cudaDev 1 nvmlDev 1 busId 64000 commId 0xcdbbe7a2eb7417f4 - Init START
+ip-26-0-170-132:3330186:3330700 [3] NCCL INFO Using network Libfabric
+ip-26-0-170-132:3330183:3330701 [0] NCCL INFO Using network Libfabric
+ip-26-0-163-58:934721:935223 [1] NCCL INFO comm 0xc2e8db0 rank 1 nranks 4 cudaDev 1 nvmlDev 1 busId 64000 commId 0xcdbbe7a2eb7417f4 - Init START
+ip-26-0-170-132:3330184:3330699 [1] NCCL INFO comm 0xb3a7d00 rank 2 nranks 4 cudaDev 1 nvmlDev 1 busId 64000 commId 0xcdbbe7a2eb7417f4 - Init START
+ip-26-0-170-132:3330184:3330699 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-170-31:3197181:3197798 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-162-14:2525076:2525588 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-163-58:934723:935225 [3] NCCL INFO Using network Libfabric
+ip-26-0-163-58:934721:935223 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-170-31:3197180:3197800 [0] NCCL INFO comm 0xbf3f770 rank 3 nranks 4 cudaDev 0 nvmlDev 0 busId 53000 commId 0x655e0359b702bd7a - Init START
+ip-26-0-162-14:2525075:2525592 [0] NCCL INFO comm 0xb7dc430 rank 0 nranks 4 cudaDev 0 nvmlDev 0 busId 53000 commId 0x655e0359b702bd7a - Init START
+ip-26-0-163-58:934720:935224 [0] NCCL INFO comm 0xb5a4670 rank 1 nranks 4 cudaDev 0 nvmlDev 0 busId 53000 commId 0x655e0359b702bd7a - Init START
+ip-26-0-170-132:3330183:3330701 [0] NCCL INFO comm 0xb26f790 rank 2 nranks 4 cudaDev 0 nvmlDev 0 busId 53000 commId 0x655e0359b702bd7a - Init START
+ip-26-0-170-132:3330183:3330701 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-162-14:2525078:2525591 [3] NCCL INFO comm 0xb18deb0 rank 0 nranks 4 cudaDev 3 nvmlDev 3 busId 86000 commId 0x26d03f9b08f1ec31 - Init START
+ip-26-0-162-14:2525075:2525592 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/253
+ip-26-0-163-58:934720:935224 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-170-31:3197183:3197799 [3] NCCL INFO comm 0xbd0ed80 rank 3 nranks 4 cudaDev 3 nvmlDev 3 busId 86000 commId 0x26d03f9b08f1ec31 - Init START
+ip-26-0-170-31:3197180:3197800 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-170-132:3330186:3330700 [3] NCCL INFO comm 0xacb0aa0 rank 2 nranks 4 cudaDev 3 nvmlDev 3 busId 86000 commId 0x26d03f9b08f1ec31 - Init START
+ip-26-0-163-58:934723:935225 [3] NCCL INFO comm 0xb868d80 rank 1 nranks 4 cudaDev 3 nvmlDev 3 busId 86000 commId 0x26d03f9b08f1ec31 - Init START
+ip-26-0-170-132:3330186:3330700 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-162-14:2525078:2525591 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-170-31:3197183:3197799 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-163-58:934723:935225 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-162-14:2525082:2525581 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-170-132:3330190:3330695 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-170-132:3330188:3330696 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-162-14:2525079:2525580 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-162-14:2525080:2525583 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-170-132:3330187:3330694 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-162-14:2525081:2525584 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-170-132:3330189:3330697 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-170-31:3197187:3197788 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-170-31:3197186:3197791 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-170-31:3197184:3197789 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-163-58:934724:935213 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-163-58:934724:935213 [4] NCCL INFO Trees [0] -1/-1/-1->1->2 [1] -1/-1/-1->1->2 [2] 2/0/-1->1->3 [3] 2/0/-1->1->3
+ip-26-0-163-58:934724:935213 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-170-132:3330187:3330694 [4] NCCL INFO Trees [0] 1/3/-1->2->0 [1] 1/3/-1->2->0 [2] -1/-1/-1->2->1 [3] -1/-1/-1->2->1
+ip-26-0-170-132:3330187:3330694 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-170-31:3197184:3197789 [4] NCCL INFO Trees [0] -1/-1/-1->3->2 [1] -1/-1/-1->3->2 [2] 1/-1/-1->3->-1 [3] 1/-1/-1->3->-1
+ip-26-0-170-31:3197184:3197789 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-162-14:2525079:2525580 [4] NCCL INFO Channel 00/04 :    0   1   2   3
+ip-26-0-162-14:2525079:2525580 [4] NCCL INFO Channel 01/04 :    0   1   2   3
+ip-26-0-162-14:2525079:2525580 [4] NCCL INFO Channel 02/04 :    0   1   2   3
+ip-26-0-162-14:2525079:2525580 [4] NCCL INFO Channel 03/04 :    0   1   2   3
+ip-26-0-162-14:2525079:2525580 [4] NCCL INFO Trees [0] 2/-1/-1->0->-1 [1] 2/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-162-14:2525079:2525580 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-162-14:2525076:2525588 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-170-31:3197185:3197790 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-163-58:934727:935214 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-163-58:934727:935214 [7] NCCL INFO Trees [0] -1/-1/-1->1->2 [1] -1/-1/-1->1->2 [2] 2/0/-1->1->3 [3] 2/0/-1->1->3
+ip-26-0-163-58:934727:935214 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-170-132:3330190:3330695 [7] NCCL INFO Trees [0] 1/3/-1->2->0 [1] 1/3/-1->2->0 [2] -1/-1/-1->2->1 [3] -1/-1/-1->2->1
+ip-26-0-170-132:3330190:3330695 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-170-31:3197187:3197788 [7] NCCL INFO Trees [0] -1/-1/-1->3->2 [1] -1/-1/-1->3->2 [2] 1/-1/-1->3->-1 [3] 1/-1/-1->3->-1
+ip-26-0-170-31:3197187:3197788 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-162-14:2525082:2525581 [7] NCCL INFO Channel 00/04 :    0   1   2   3
+ip-26-0-162-14:2525082:2525581 [7] NCCL INFO Channel 01/04 :    0   1   2   3
+ip-26-0-162-14:2525082:2525581 [7] NCCL INFO Channel 02/04 :    0   1   2   3
+ip-26-0-162-14:2525082:2525581 [7] NCCL INFO Channel 03/04 :    0   1   2   3
+ip-26-0-162-14:2525082:2525581 [7] NCCL INFO Trees [0] 2/-1/-1->0->-1 [1] 2/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-162-14:2525082:2525581 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-162-14:2525077:2525586 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-162-14:2525075:2525592 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-162-14:2525078:2525591 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-163-58:934725:935215 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-163-58:934725:935215 [5] NCCL INFO Trees [0] -1/-1/-1->1->2 [1] -1/-1/-1->1->2 [2] 2/0/-1->1->3 [3] 2/0/-1->1->3
+ip-26-0-163-58:934725:935215 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-170-132:3330188:3330696 [5] NCCL INFO Trees [0] 1/3/-1->2->0 [1] 1/3/-1->2->0 [2] -1/-1/-1->2->1 [3] -1/-1/-1->2->1
+ip-26-0-170-132:3330188:3330696 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-170-31:3197185:3197790 [5] NCCL INFO Trees [0] -1/-1/-1->3->2 [1] -1/-1/-1->3->2 [2] 1/-1/-1->3->-1 [3] 1/-1/-1->3->-1
+ip-26-0-170-31:3197185:3197790 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-162-14:2525080:2525583 [5] NCCL INFO Channel 00/04 :    0   1   2   3
+ip-26-0-162-14:2525080:2525583 [5] NCCL INFO Channel 01/04 :    0   1   2   3
+ip-26-0-162-14:2525080:2525583 [5] NCCL INFO Channel 02/04 :    0   1   2   3
+ip-26-0-162-14:2525080:2525583 [5] NCCL INFO Channel 03/04 :    0   1   2   3
+ip-26-0-162-14:2525080:2525583 [5] NCCL INFO Trees [0] 2/-1/-1->0->-1 [1] 2/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-162-14:2525080:2525583 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-163-58:934724:935213 [4] NCCL INFO Channel 00/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934724:935213 [4] NCCL INFO Channel 01/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934724:935213 [4] NCCL INFO Channel 02/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934724:935213 [4] NCCL INFO Channel 03/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934724:935213 [4] NCCL INFO Channel 00/0 : 1[4] -> 2[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-132:3330187:3330694 [4] NCCL INFO Channel 00/0 : 1[4] -> 2[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934724:935213 [4] NCCL INFO Channel 01/0 : 1[4] -> 2[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-132:3330187:3330694 [4] NCCL INFO Channel 01/0 : 1[4] -> 2[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934724:935213 [4] NCCL INFO Channel 02/0 : 1[4] -> 2[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934724:935213 [4] NCCL INFO Channel 03/0 : 1[4] -> 2[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-132:3330187:3330694 [4] NCCL INFO Channel 02/0 : 1[4] -> 2[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-132:3330187:3330694 [4] NCCL INFO Channel 03/0 : 1[4] -> 2[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-31:3197184:3197789 [4] NCCL INFO Channel 00/0 : 2[4] -> 3[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-132:3330187:3330694 [4] NCCL INFO Channel 00/0 : 2[4] -> 3[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-31:3197184:3197789 [4] NCCL INFO Channel 01/0 : 2[4] -> 3[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-31:3197184:3197789 [4] NCCL INFO Channel 02/0 : 2[4] -> 3[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-132:3330187:3330694 [4] NCCL INFO Channel 01/0 : 2[4] -> 3[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-31:3197184:3197789 [4] NCCL INFO Channel 03/0 : 2[4] -> 3[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-132:3330187:3330694 [4] NCCL INFO Channel 02/0 : 2[4] -> 3[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-132:3330187:3330694 [4] NCCL INFO Channel 03/0 : 2[4] -> 3[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-31:3197184:3197789 [4] NCCL INFO Channel 00/0 : 3[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-31:3197184:3197789 [4] NCCL INFO Channel 01/0 : 3[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-31:3197184:3197789 [4] NCCL INFO Channel 02/0 : 3[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-31:3197184:3197789 [4] NCCL INFO Channel 03/0 : 3[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-14:2525079:2525580 [4] NCCL INFO Channel 00/0 : 3[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-14:2525079:2525580 [4] NCCL INFO Channel 01/0 : 3[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-14:2525079:2525580 [4] NCCL INFO Channel 02/0 : 3[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-14:2525079:2525580 [4] NCCL INFO Channel 03/0 : 3[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-14:2525079:2525580 [4] NCCL INFO Channel 00/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-14:2525079:2525580 [4] NCCL INFO Channel 01/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-14:2525079:2525580 [4] NCCL INFO Channel 02/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-14:2525079:2525580 [4] NCCL INFO Channel 03/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-132:3330184:3330699 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-163-58:934726:935216 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-163-58:934726:935216 [6] NCCL INFO Trees [0] -1/-1/-1->1->2 [1] -1/-1/-1->1->2 [2] 2/0/-1->1->3 [3] 2/0/-1->1->3
+ip-26-0-163-58:934726:935216 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-170-132:3330189:3330697 [6] NCCL INFO Trees [0] 1/3/-1->2->0 [1] 1/3/-1->2->0 [2] -1/-1/-1->2->1 [3] -1/-1/-1->2->1
+ip-26-0-170-132:3330189:3330697 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-162-14:2525081:2525584 [6] NCCL INFO Channel 00/04 :    0   1   2   3
+ip-26-0-162-14:2525081:2525584 [6] NCCL INFO Channel 01/04 :    0   1   2   3
+ip-26-0-162-14:2525081:2525584 [6] NCCL INFO Channel 02/04 :    0   1   2   3
+ip-26-0-162-14:2525081:2525584 [6] NCCL INFO Channel 03/04 :    0   1   2   3
+ip-26-0-170-31:3197186:3197791 [6] NCCL INFO Trees [0] -1/-1/-1->3->2 [1] -1/-1/-1->3->2 [2] 1/-1/-1->3->-1 [3] 1/-1/-1->3->-1
+ip-26-0-170-31:3197186:3197791 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-162-14:2525081:2525584 [6] NCCL INFO Trees [0] 2/-1/-1->0->-1 [1] 2/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-162-14:2525081:2525584 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-170-132:3330185:3330698 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-170-132:3330183:3330701 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-170-132:3330186:3330700 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-163-58:934727:935214 [7] NCCL INFO Channel 00/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934727:935214 [7] NCCL INFO Channel 01/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330190:3330695 [7] NCCL INFO Channel 00/0 : 1[7] -> 2[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934727:935214 [7] NCCL INFO Channel 02/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934727:935214 [7] NCCL INFO Channel 03/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330190:3330695 [7] NCCL INFO Channel 01/0 : 1[7] -> 2[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934727:935214 [7] NCCL INFO Channel 00/0 : 1[7] -> 2[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330190:3330695 [7] NCCL INFO Channel 02/0 : 1[7] -> 2[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934727:935214 [7] NCCL INFO Channel 01/0 : 1[7] -> 2[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330190:3330695 [7] NCCL INFO Channel 03/0 : 1[7] -> 2[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934727:935214 [7] NCCL INFO Channel 02/0 : 1[7] -> 2[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330190:3330695 [7] NCCL INFO Channel 00/0 : 2[7] -> 3[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934727:935214 [7] NCCL INFO Channel 03/0 : 1[7] -> 2[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330190:3330695 [7] NCCL INFO Channel 01/0 : 2[7] -> 3[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330190:3330695 [7] NCCL INFO Channel 02/0 : 2[7] -> 3[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330190:3330695 [7] NCCL INFO Channel 03/0 : 2[7] -> 3[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-31:3197187:3197788 [7] NCCL INFO Channel 00/0 : 2[7] -> 3[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-31:3197187:3197788 [7] NCCL INFO Channel 01/0 : 2[7] -> 3[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-31:3197187:3197788 [7] NCCL INFO Channel 02/0 : 2[7] -> 3[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-31:3197187:3197788 [7] NCCL INFO Channel 03/0 : 2[7] -> 3[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-31:3197187:3197788 [7] NCCL INFO Channel 00/0 : 3[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-31:3197187:3197788 [7] NCCL INFO Channel 01/0 : 3[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-31:3197187:3197788 [7] NCCL INFO Channel 02/0 : 3[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-31:3197187:3197788 [7] NCCL INFO Channel 03/0 : 3[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-14:2525082:2525581 [7] NCCL INFO Channel 00/0 : 3[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-14:2525082:2525581 [7] NCCL INFO Channel 01/0 : 3[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-14:2525082:2525581 [7] NCCL INFO Channel 02/0 : 3[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-14:2525082:2525581 [7] NCCL INFO Channel 03/0 : 3[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-14:2525082:2525581 [7] NCCL INFO Channel 00/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-14:2525082:2525581 [7] NCCL INFO Channel 01/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-14:2525082:2525581 [7] NCCL INFO Channel 02/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-14:2525082:2525581 [7] NCCL INFO Channel 03/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-31:3197180:3197800 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-170-31:3197181:3197798 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-170-31:3197183:3197799 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-163-58:934725:935215 [5] NCCL INFO Channel 00/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934725:935215 [5] NCCL INFO Channel 01/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934725:935215 [5] NCCL INFO Channel 02/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934725:935215 [5] NCCL INFO Channel 03/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934725:935215 [5] NCCL INFO Channel 00/0 : 1[5] -> 2[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934725:935215 [5] NCCL INFO Channel 01/0 : 1[5] -> 2[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934725:935215 [5] NCCL INFO Channel 02/0 : 1[5] -> 2[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934725:935215 [5] NCCL INFO Channel 03/0 : 1[5] -> 2[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-14:2525080:2525583 [5] NCCL INFO Channel 00/0 : 3[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-14:2525080:2525583 [5] NCCL INFO Channel 01/0 : 3[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-14:2525080:2525583 [5] NCCL INFO Channel 02/0 : 3[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-14:2525080:2525583 [5] NCCL INFO Channel 03/0 : 3[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-14:2525080:2525583 [5] NCCL INFO Channel 00/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-31:3197185:3197790 [5] NCCL INFO Channel 00/0 : 2[5] -> 3[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-14:2525080:2525583 [5] NCCL INFO Channel 01/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-31:3197185:3197790 [5] NCCL INFO Channel 01/0 : 2[5] -> 3[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-14:2525080:2525583 [5] NCCL INFO Channel 02/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-31:3197185:3197790 [5] NCCL INFO Channel 02/0 : 2[5] -> 3[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-14:2525080:2525583 [5] NCCL INFO Channel 03/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-31:3197185:3197790 [5] NCCL INFO Channel 03/0 : 2[5] -> 3[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-31:3197185:3197790 [5] NCCL INFO Channel 00/0 : 3[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-31:3197185:3197790 [5] NCCL INFO Channel 01/0 : 3[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-31:3197185:3197790 [5] NCCL INFO Channel 02/0 : 3[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-31:3197185:3197790 [5] NCCL INFO Channel 03/0 : 3[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-31:3197182:3197797 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-170-132:3330188:3330696 [5] NCCL INFO Channel 00/0 : 1[5] -> 2[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330188:3330696 [5] NCCL INFO Channel 01/0 : 1[5] -> 2[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330188:3330696 [5] NCCL INFO Channel 02/0 : 1[5] -> 2[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330188:3330696 [5] NCCL INFO Channel 03/0 : 1[5] -> 2[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330188:3330696 [5] NCCL INFO Channel 00/0 : 2[5] -> 3[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330188:3330696 [5] NCCL INFO Channel 01/0 : 2[5] -> 3[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330188:3330696 [5] NCCL INFO Channel 02/0 : 2[5] -> 3[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330188:3330696 [5] NCCL INFO Channel 03/0 : 2[5] -> 3[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-31:3197186:3197791 [6] NCCL INFO Channel 00/0 : 2[6] -> 3[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-31:3197186:3197791 [6] NCCL INFO Channel 01/0 : 2[6] -> 3[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-31:3197186:3197791 [6] NCCL INFO Channel 02/0 : 2[6] -> 3[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-31:3197186:3197791 [6] NCCL INFO Channel 03/0 : 2[6] -> 3[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-31:3197186:3197791 [6] NCCL INFO Channel 00/0 : 3[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-31:3197186:3197791 [6] NCCL INFO Channel 01/0 : 3[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-31:3197186:3197791 [6] NCCL INFO Channel 02/0 : 3[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-31:3197186:3197791 [6] NCCL INFO Channel 03/0 : 3[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330189:3330697 [6] NCCL INFO Channel 00/0 : 1[6] -> 2[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330189:3330697 [6] NCCL INFO Channel 01/0 : 1[6] -> 2[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934726:935216 [6] NCCL INFO Channel 00/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330189:3330697 [6] NCCL INFO Channel 02/0 : 1[6] -> 2[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934726:935216 [6] NCCL INFO Channel 01/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330189:3330697 [6] NCCL INFO Channel 03/0 : 1[6] -> 2[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934726:935216 [6] NCCL INFO Channel 02/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330189:3330697 [6] NCCL INFO Channel 00/0 : 2[6] -> 3[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934726:935216 [6] NCCL INFO Channel 03/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330189:3330697 [6] NCCL INFO Channel 01/0 : 2[6] -> 3[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330189:3330697 [6] NCCL INFO Channel 02/0 : 2[6] -> 3[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934726:935216 [6] NCCL INFO Channel 00/0 : 1[6] -> 2[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330189:3330697 [6] NCCL INFO Channel 03/0 : 2[6] -> 3[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934726:935216 [6] NCCL INFO Channel 01/0 : 1[6] -> 2[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934726:935216 [6] NCCL INFO Channel 02/0 : 1[6] -> 2[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934726:935216 [6] NCCL INFO Channel 03/0 : 1[6] -> 2[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-14:2525081:2525584 [6] NCCL INFO Channel 00/0 : 3[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-14:2525081:2525584 [6] NCCL INFO Channel 01/0 : 3[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-14:2525081:2525584 [6] NCCL INFO Channel 02/0 : 3[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-14:2525081:2525584 [6] NCCL INFO Channel 03/0 : 3[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-14:2525081:2525584 [6] NCCL INFO Channel 00/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-14:2525081:2525584 [6] NCCL INFO Channel 01/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-14:2525081:2525584 [6] NCCL INFO Channel 02/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-14:2525081:2525584 [6] NCCL INFO Channel 03/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934722:935222 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-163-58:934722:935222 [2] NCCL INFO Trees [0] -1/-1/-1->1->2 [1] -1/-1/-1->1->2 [2] 2/0/-1->1->3 [3] 2/0/-1->1->3
+ip-26-0-163-58:934722:935222 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-170-132:3330185:3330698 [2] NCCL INFO Trees [0] 1/3/-1->2->0 [1] 1/3/-1->2->0 [2] -1/-1/-1->2->1 [3] -1/-1/-1->2->1
+ip-26-0-170-132:3330185:3330698 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-170-31:3197182:3197797 [2] NCCL INFO Trees [0] -1/-1/-1->3->2 [1] -1/-1/-1->3->2 [2] 1/-1/-1->3->-1 [3] 1/-1/-1->3->-1
+ip-26-0-170-31:3197182:3197797 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-162-14:2525077:2525586 [2] NCCL INFO Channel 00/04 :    0   1   2   3
+ip-26-0-162-14:2525077:2525586 [2] NCCL INFO Channel 01/04 :    0   1   2   3
+ip-26-0-162-14:2525077:2525586 [2] NCCL INFO Channel 02/04 :    0   1   2   3
+ip-26-0-162-14:2525077:2525586 [2] NCCL INFO Channel 03/04 :    0   1   2   3
+ip-26-0-162-14:2525077:2525586 [2] NCCL INFO Trees [0] 2/-1/-1->0->-1 [1] 2/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-162-14:2525077:2525586 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-163-58:934720:935224 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-163-58:934720:935224 [0] NCCL INFO Trees [0] -1/-1/-1->1->2 [1] -1/-1/-1->1->2 [2] 2/0/-1->1->3 [3] 2/0/-1->1->3
+ip-26-0-163-58:934720:935224 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-170-132:3330183:3330701 [0] NCCL INFO Trees [0] 1/3/-1->2->0 [1] 1/3/-1->2->0 [2] -1/-1/-1->2->1 [3] -1/-1/-1->2->1
+ip-26-0-170-132:3330183:3330701 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-170-31:3197180:3197800 [0] NCCL INFO Trees [0] -1/-1/-1->3->2 [1] -1/-1/-1->3->2 [2] 1/-1/-1->3->-1 [3] 1/-1/-1->3->-1
+ip-26-0-170-31:3197180:3197800 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-162-14:2525075:2525592 [0] NCCL INFO Channel 00/04 :    0   1   2   3
+ip-26-0-162-14:2525075:2525592 [0] NCCL INFO Channel 01/04 :    0   1   2   3
+ip-26-0-162-14:2525075:2525592 [0] NCCL INFO Channel 02/04 :    0   1   2   3
+ip-26-0-162-14:2525075:2525592 [0] NCCL INFO Channel 03/04 :    0   1   2   3
+ip-26-0-162-14:2525075:2525592 [0] NCCL INFO Trees [0] 2/-1/-1->0->-1 [1] 2/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-162-14:2525075:2525592 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-163-58:934721:935223 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-163-58:934721:935223 [1] NCCL INFO Trees [0] -1/-1/-1->1->2 [1] -1/-1/-1->1->2 [2] 2/0/-1->1->3 [3] 2/0/-1->1->3
+ip-26-0-163-58:934721:935223 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-170-132:3330184:3330699 [1] NCCL INFO Trees [0] 1/3/-1->2->0 [1] 1/3/-1->2->0 [2] -1/-1/-1->2->1 [3] -1/-1/-1->2->1
+ip-26-0-170-132:3330184:3330699 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-170-31:3197181:3197798 [1] NCCL INFO Trees [0] -1/-1/-1->3->2 [1] -1/-1/-1->3->2 [2] 1/-1/-1->3->-1 [3] 1/-1/-1->3->-1
+ip-26-0-170-31:3197181:3197798 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-162-14:2525076:2525588 [1] NCCL INFO Channel 00/04 :    0   1   2   3
+ip-26-0-162-14:2525076:2525588 [1] NCCL INFO Channel 01/04 :    0   1   2   3
+ip-26-0-162-14:2525076:2525588 [1] NCCL INFO Channel 02/04 :    0   1   2   3
+ip-26-0-162-14:2525076:2525588 [1] NCCL INFO Channel 03/04 :    0   1   2   3
+ip-26-0-162-14:2525076:2525588 [1] NCCL INFO Trees [0] 2/-1/-1->0->-1 [1] 2/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-162-14:2525076:2525588 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-163-58:934723:935225 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-163-58:934723:935225 [3] NCCL INFO Trees [0] -1/-1/-1->1->2 [1] -1/-1/-1->1->2 [2] 2/0/-1->1->3 [3] 2/0/-1->1->3
+ip-26-0-163-58:934723:935225 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-170-132:3330186:3330700 [3] NCCL INFO Trees [0] 1/3/-1->2->0 [1] 1/3/-1->2->0 [2] -1/-1/-1->2->1 [3] -1/-1/-1->2->1
+ip-26-0-170-132:3330186:3330700 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-170-31:3197183:3197799 [3] NCCL INFO Trees [0] -1/-1/-1->3->2 [1] -1/-1/-1->3->2 [2] 1/-1/-1->3->-1 [3] 1/-1/-1->3->-1
+ip-26-0-170-31:3197183:3197799 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-162-14:2525078:2525591 [3] NCCL INFO Channel 00/04 :    0   1   2   3
+ip-26-0-162-14:2525078:2525591 [3] NCCL INFO Channel 01/04 :    0   1   2   3
+ip-26-0-162-14:2525078:2525591 [3] NCCL INFO Channel 02/04 :    0   1   2   3
+ip-26-0-162-14:2525078:2525591 [3] NCCL INFO Channel 03/04 :    0   1   2   3
+ip-26-0-162-14:2525078:2525591 [3] NCCL INFO Trees [0] 2/-1/-1->0->-1 [1] 2/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-162-14:2525078:2525591 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-170-31:3197184:3197789 [4] NCCL INFO Connected all rings
+ip-26-0-170-31:3197184:3197789 [4] NCCL INFO Channel 02/0 : 1[4] -> 3[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-31:3197184:3197789 [4] NCCL INFO Channel 03/0 : 1[4] -> 3[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-31:3197184:3197789 [4] NCCL INFO Channel 02/0 : 3[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-31:3197184:3197789 [4] NCCL INFO Channel 03/0 : 3[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-132:3330187:3330694 [4] NCCL INFO Connected all rings
+ip-26-0-170-132:3330187:3330694 [4] NCCL INFO Channel 00/0 : 0[4] -> 2[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-132:3330187:3330694 [4] NCCL INFO Channel 01/0 : 0[4] -> 2[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-132:3330187:3330694 [4] NCCL INFO Channel 00/0 : 2[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-132:3330187:3330694 [4] NCCL INFO Channel 01/0 : 2[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-31:3197182:3197797 [2] NCCL INFO Channel 00/0 : 2[2] -> 3[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-31:3197182:3197797 [2] NCCL INFO Channel 01/0 : 2[2] -> 3[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-31:3197182:3197797 [2] NCCL INFO Channel 02/0 : 2[2] -> 3[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934722:935222 [2] NCCL INFO Channel 00/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-31:3197182:3197797 [2] NCCL INFO Channel 03/0 : 2[2] -> 3[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-31:3197182:3197797 [2] NCCL INFO Channel 00/0 : 3[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934722:935222 [2] NCCL INFO Channel 01/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-31:3197182:3197797 [2] NCCL INFO Channel 01/0 : 3[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934722:935222 [2] NCCL INFO Channel 02/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-31:3197182:3197797 [2] NCCL INFO Channel 02/0 : 3[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934722:935222 [2] NCCL INFO Channel 03/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-31:3197182:3197797 [2] NCCL INFO Channel 03/0 : 3[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934722:935222 [2] NCCL INFO Channel 00/0 : 1[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934722:935222 [2] NCCL INFO Channel 01/0 : 1[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934722:935222 [2] NCCL INFO Channel 02/0 : 1[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330185:3330698 [2] NCCL INFO Channel 00/0 : 1[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934722:935222 [2] NCCL INFO Channel 03/0 : 1[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330185:3330698 [2] NCCL INFO Channel 01/0 : 1[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330185:3330698 [2] NCCL INFO Channel 02/0 : 1[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330185:3330698 [2] NCCL INFO Channel 03/0 : 1[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330185:3330698 [2] NCCL INFO Channel 00/0 : 2[2] -> 3[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330185:3330698 [2] NCCL INFO Channel 01/0 : 2[2] -> 3[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330185:3330698 [2] NCCL INFO Channel 02/0 : 2[2] -> 3[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330185:3330698 [2] NCCL INFO Channel 03/0 : 2[2] -> 3[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-14:2525077:2525586 [2] NCCL INFO Channel 00/0 : 3[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-14:2525077:2525586 [2] NCCL INFO Channel 01/0 : 3[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-14:2525077:2525586 [2] NCCL INFO Channel 02/0 : 3[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-14:2525077:2525586 [2] NCCL INFO Channel 03/0 : 3[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-14:2525077:2525586 [2] NCCL INFO Channel 00/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-14:2525077:2525586 [2] NCCL INFO Channel 01/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-14:2525077:2525586 [2] NCCL INFO Channel 02/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-14:2525077:2525586 [2] NCCL INFO Channel 03/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-31:3197180:3197800 [0] NCCL INFO Channel 00/0 : 2[0] -> 3[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-31:3197180:3197800 [0] NCCL INFO Channel 01/0 : 2[0] -> 3[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-31:3197180:3197800 [0] NCCL INFO Channel 02/0 : 2[0] -> 3[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934720:935224 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-31:3197180:3197800 [0] NCCL INFO Channel 03/0 : 2[0] -> 3[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-31:3197180:3197800 [0] NCCL INFO Channel 00/0 : 3[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934720:935224 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330183:3330701 [0] NCCL INFO Channel 00/0 : 1[0] -> 2[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-31:3197180:3197800 [0] NCCL INFO Channel 01/0 : 3[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934720:935224 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-31:3197180:3197800 [0] NCCL INFO Channel 02/0 : 3[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934720:935224 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330183:3330701 [0] NCCL INFO Channel 01/0 : 1[0] -> 2[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-31:3197180:3197800 [0] NCCL INFO Channel 03/0 : 3[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934720:935224 [0] NCCL INFO Channel 00/0 : 1[0] -> 2[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330183:3330701 [0] NCCL INFO Channel 02/0 : 1[0] -> 2[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934720:935224 [0] NCCL INFO Channel 01/0 : 1[0] -> 2[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330183:3330701 [0] NCCL INFO Channel 03/0 : 1[0] -> 2[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934720:935224 [0] NCCL INFO Channel 02/0 : 1[0] -> 2[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330183:3330701 [0] NCCL INFO Channel 00/0 : 2[0] -> 3[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934720:935224 [0] NCCL INFO Channel 03/0 : 1[0] -> 2[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330183:3330701 [0] NCCL INFO Channel 01/0 : 2[0] -> 3[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330183:3330701 [0] NCCL INFO Channel 02/0 : 2[0] -> 3[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330183:3330701 [0] NCCL INFO Channel 03/0 : 2[0] -> 3[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-14:2525075:2525592 [0] NCCL INFO Channel 00/0 : 3[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-14:2525075:2525592 [0] NCCL INFO Channel 01/0 : 3[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-14:2525075:2525592 [0] NCCL INFO Channel 02/0 : 3[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-14:2525075:2525592 [0] NCCL INFO Channel 03/0 : 3[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-14:2525075:2525592 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-14:2525075:2525592 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-14:2525075:2525592 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-14:2525075:2525592 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934721:935223 [1] NCCL INFO Channel 00/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934721:935223 [1] NCCL INFO Channel 01/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934721:935223 [1] NCCL INFO Channel 02/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934721:935223 [1] NCCL INFO Channel 03/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934721:935223 [1] NCCL INFO Channel 00/0 : 1[1] -> 2[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934721:935223 [1] NCCL INFO Channel 01/0 : 1[1] -> 2[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934721:935223 [1] NCCL INFO Channel 02/0 : 1[1] -> 2[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934721:935223 [1] NCCL INFO Channel 03/0 : 1[1] -> 2[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330184:3330699 [1] NCCL INFO Channel 00/0 : 1[1] -> 2[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-31:3197181:3197798 [1] NCCL INFO Channel 00/0 : 2[1] -> 3[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330184:3330699 [1] NCCL INFO Channel 01/0 : 1[1] -> 2[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-31:3197181:3197798 [1] NCCL INFO Channel 01/0 : 2[1] -> 3[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330184:3330699 [1] NCCL INFO Channel 02/0 : 1[1] -> 2[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-31:3197181:3197798 [1] NCCL INFO Channel 02/0 : 2[1] -> 3[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-31:3197181:3197798 [1] NCCL INFO Channel 03/0 : 2[1] -> 3[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330184:3330699 [1] NCCL INFO Channel 03/0 : 1[1] -> 2[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-31:3197181:3197798 [1] NCCL INFO Channel 00/0 : 3[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-31:3197181:3197798 [1] NCCL INFO Channel 01/0 : 3[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330184:3330699 [1] NCCL INFO Channel 00/0 : 2[1] -> 3[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-31:3197181:3197798 [1] NCCL INFO Channel 02/0 : 3[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330184:3330699 [1] NCCL INFO Channel 01/0 : 2[1] -> 3[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-31:3197181:3197798 [1] NCCL INFO Channel 03/0 : 3[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330184:3330699 [1] NCCL INFO Channel 02/0 : 2[1] -> 3[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330184:3330699 [1] NCCL INFO Channel 03/0 : 2[1] -> 3[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-14:2525076:2525588 [1] NCCL INFO Channel 00/0 : 3[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-14:2525076:2525588 [1] NCCL INFO Channel 01/0 : 3[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-14:2525076:2525588 [1] NCCL INFO Channel 02/0 : 3[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-14:2525076:2525588 [1] NCCL INFO Channel 03/0 : 3[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-14:2525076:2525588 [1] NCCL INFO Channel 00/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-14:2525076:2525588 [1] NCCL INFO Channel 01/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-14:2525076:2525588 [1] NCCL INFO Channel 02/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-14:2525076:2525588 [1] NCCL INFO Channel 03/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330186:3330700 [3] NCCL INFO Channel 00/0 : 1[3] -> 2[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330186:3330700 [3] NCCL INFO Channel 01/0 : 1[3] -> 2[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330186:3330700 [3] NCCL INFO Channel 02/0 : 1[3] -> 2[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330186:3330700 [3] NCCL INFO Channel 03/0 : 1[3] -> 2[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330186:3330700 [3] NCCL INFO Channel 00/0 : 2[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330186:3330700 [3] NCCL INFO Channel 01/0 : 2[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330186:3330700 [3] NCCL INFO Channel 02/0 : 2[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330186:3330700 [3] NCCL INFO Channel 03/0 : 2[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934723:935225 [3] NCCL INFO Channel 00/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934723:935225 [3] NCCL INFO Channel 01/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934723:935225 [3] NCCL INFO Channel 02/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-31:3197183:3197799 [3] NCCL INFO Channel 00/0 : 2[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934723:935225 [3] NCCL INFO Channel 03/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-31:3197183:3197799 [3] NCCL INFO Channel 01/0 : 2[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934723:935225 [3] NCCL INFO Channel 00/0 : 1[3] -> 2[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-31:3197183:3197799 [3] NCCL INFO Channel 02/0 : 2[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934723:935225 [3] NCCL INFO Channel 01/0 : 1[3] -> 2[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-31:3197183:3197799 [3] NCCL INFO Channel 03/0 : 2[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934723:935225 [3] NCCL INFO Channel 02/0 : 1[3] -> 2[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-31:3197183:3197799 [3] NCCL INFO Channel 00/0 : 3[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934723:935225 [3] NCCL INFO Channel 03/0 : 1[3] -> 2[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-31:3197183:3197799 [3] NCCL INFO Channel 01/0 : 3[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-31:3197183:3197799 [3] NCCL INFO Channel 02/0 : 3[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-31:3197183:3197799 [3] NCCL INFO Channel 03/0 : 3[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-14:2525078:2525591 [3] NCCL INFO Channel 00/0 : 3[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-14:2525078:2525591 [3] NCCL INFO Channel 01/0 : 3[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-14:2525078:2525591 [3] NCCL INFO Channel 02/0 : 3[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-14:2525078:2525591 [3] NCCL INFO Channel 03/0 : 3[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-14:2525078:2525591 [3] NCCL INFO Channel 00/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-14:2525078:2525591 [3] NCCL INFO Channel 01/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-14:2525078:2525591 [3] NCCL INFO Channel 02/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-14:2525078:2525591 [3] NCCL INFO Channel 03/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-14:2525082:2525581 [7] NCCL INFO Connected all rings
+ip-26-0-162-14:2525082:2525581 [7] NCCL INFO Channel 00/0 : 2[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-14:2525082:2525581 [7] NCCL INFO Channel 01/0 : 2[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-14:2525082:2525581 [7] NCCL INFO Channel 00/0 : 0[7] -> 2[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-14:2525082:2525581 [7] NCCL INFO Channel 01/0 : 0[7] -> 2[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-31:3197187:3197788 [7] NCCL INFO Connected all rings
+ip-26-0-170-31:3197187:3197788 [7] NCCL INFO Channel 02/0 : 1[7] -> 3[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-31:3197187:3197788 [7] NCCL INFO Channel 03/0 : 1[7] -> 3[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-31:3197187:3197788 [7] NCCL INFO Channel 02/0 : 3[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-31:3197187:3197788 [7] NCCL INFO Channel 03/0 : 3[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330190:3330695 [7] NCCL INFO Connected all rings
+ip-26-0-170-132:3330190:3330695 [7] NCCL INFO Channel 00/0 : 0[7] -> 2[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330190:3330695 [7] NCCL INFO Channel 01/0 : 0[7] -> 2[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330190:3330695 [7] NCCL INFO Channel 00/0 : 2[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330190:3330695 [7] NCCL INFO Channel 01/0 : 2[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330190:3330695 [7] NCCL INFO Channel 00/0 : 3[7] -> 2[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-14:2525082:2525581 [7] NCCL INFO Channel 02/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330190:3330695 [7] NCCL INFO Channel 01/0 : 3[7] -> 2[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-14:2525082:2525581 [7] NCCL INFO Channel 03/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330190:3330695 [7] NCCL INFO Channel 00/0 : 2[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330190:3330695 [7] NCCL INFO Channel 01/0 : 2[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330190:3330695 [7] NCCL INFO Channel 02/0 : 2[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-132:3330190:3330695 [7] NCCL INFO Channel 03/0 : 2[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-14:2525079:2525580 [4] NCCL INFO Connected all rings
+ip-26-0-162-14:2525079:2525580 [4] NCCL INFO Channel 00/0 : 2[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-14:2525079:2525580 [4] NCCL INFO Channel 01/0 : 2[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-14:2525079:2525580 [4] NCCL INFO Channel 00/0 : 0[4] -> 2[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-14:2525079:2525580 [4] NCCL INFO Channel 01/0 : 0[4] -> 2[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-14:2525079:2525580 [4] NCCL INFO Channel 02/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-14:2525079:2525580 [4] NCCL INFO Channel 03/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-132:3330187:3330694 [4] NCCL INFO Channel 00/0 : 3[4] -> 2[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-132:3330187:3330694 [4] NCCL INFO Channel 01/0 : 3[4] -> 2[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-132:3330187:3330694 [4] NCCL INFO Channel 00/0 : 2[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-132:3330187:3330694 [4] NCCL INFO Channel 01/0 : 2[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-132:3330187:3330694 [4] NCCL INFO Channel 02/0 : 2[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-132:3330187:3330694 [4] NCCL INFO Channel 03/0 : 2[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-14:2525080:2525583 [5] NCCL INFO Connected all rings
+ip-26-0-162-14:2525080:2525583 [5] NCCL INFO Channel 00/0 : 2[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-14:2525080:2525583 [5] NCCL INFO Channel 01/0 : 2[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-14:2525080:2525583 [5] NCCL INFO Channel 00/0 : 0[5] -> 2[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-14:2525080:2525583 [5] NCCL INFO Channel 01/0 : 0[5] -> 2[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-31:3197185:3197790 [5] NCCL INFO Connected all rings
+ip-26-0-170-31:3197185:3197790 [5] NCCL INFO Channel 02/0 : 1[5] -> 3[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-31:3197185:3197790 [5] NCCL INFO Channel 03/0 : 1[5] -> 3[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-31:3197185:3197790 [5] NCCL INFO Channel 02/0 : 3[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-31:3197185:3197790 [5] NCCL INFO Channel 03/0 : 3[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330188:3330696 [5] NCCL INFO Connected all rings
+ip-26-0-170-132:3330188:3330696 [5] NCCL INFO Channel 00/0 : 0[5] -> 2[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330188:3330696 [5] NCCL INFO Channel 01/0 : 0[5] -> 2[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330188:3330696 [5] NCCL INFO Channel 00/0 : 2[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330188:3330696 [5] NCCL INFO Channel 01/0 : 2[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330188:3330696 [5] NCCL INFO Channel 00/0 : 3[5] -> 2[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-14:2525080:2525583 [5] NCCL INFO Channel 02/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330188:3330696 [5] NCCL INFO Channel 01/0 : 3[5] -> 2[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-14:2525080:2525583 [5] NCCL INFO Channel 03/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330188:3330696 [5] NCCL INFO Channel 00/0 : 2[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330188:3330696 [5] NCCL INFO Channel 01/0 : 2[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330188:3330696 [5] NCCL INFO Channel 02/0 : 2[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-132:3330188:3330696 [5] NCCL INFO Channel 03/0 : 2[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934727:935214 [7] NCCL INFO Connected all rings
+ip-26-0-163-58:934727:935214 [7] NCCL INFO Channel 02/0 : 3[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934727:935214 [7] NCCL INFO Channel 03/0 : 3[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934727:935214 [7] NCCL INFO Channel 02/0 : 1[7] -> 3[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934727:935214 [7] NCCL INFO Channel 03/0 : 1[7] -> 3[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934727:935214 [7] NCCL INFO Channel 00/0 : 2[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934727:935214 [7] NCCL INFO Channel 01/0 : 2[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-31:3197187:3197788 [7] NCCL INFO Channel 00/0 : 3[7] -> 2[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934727:935214 [7] NCCL INFO Channel 02/0 : 2[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-170-31:3197187:3197788 [7] NCCL INFO Channel 01/0 : 3[7] -> 2[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934727:935214 [7] NCCL INFO Channel 03/0 : 2[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934727:935214 [7] NCCL INFO Channel 02/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934727:935214 [7] NCCL INFO Channel 03/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-163-58:934724:935213 [4] NCCL INFO Connected all rings
+ip-26-0-163-58:934724:935213 [4] NCCL INFO Channel 02/0 : 3[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934724:935213 [4] NCCL INFO Channel 03/0 : 3[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934724:935213 [4] NCCL INFO Channel 02/0 : 1[4] -> 3[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934724:935213 [4] NCCL INFO Channel 03/0 : 1[4] -> 3[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934724:935213 [4] NCCL INFO Channel 00/0 : 2[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934724:935213 [4] NCCL INFO Channel 01/0 : 2[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-31:3197184:3197789 [4] NCCL INFO Channel 00/0 : 3[4] -> 2[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934724:935213 [4] NCCL INFO Channel 02/0 : 2[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-170-31:3197184:3197789 [4] NCCL INFO Channel 01/0 : 3[4] -> 2[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934724:935213 [4] NCCL INFO Channel 03/0 : 2[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934724:935213 [4] NCCL INFO Channel 02/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934724:935213 [4] NCCL INFO Channel 03/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-163-58:934725:935215 [5] NCCL INFO Connected all rings
+ip-26-0-163-58:934725:935215 [5] NCCL INFO Channel 02/0 : 3[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934725:935215 [5] NCCL INFO Channel 03/0 : 3[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934725:935215 [5] NCCL INFO Channel 02/0 : 1[5] -> 3[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934725:935215 [5] NCCL INFO Channel 03/0 : 1[5] -> 3[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934725:935215 [5] NCCL INFO Channel 00/0 : 2[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-31:3197185:3197790 [5] NCCL INFO Channel 00/0 : 3[5] -> 2[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934725:935215 [5] NCCL INFO Channel 01/0 : 2[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-31:3197185:3197790 [5] NCCL INFO Channel 01/0 : 3[5] -> 2[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934725:935215 [5] NCCL INFO Channel 02/0 : 2[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934725:935215 [5] NCCL INFO Channel 03/0 : 2[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934725:935215 [5] NCCL INFO Channel 02/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-163-58:934725:935215 [5] NCCL INFO Channel 03/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-170-31:3197186:3197791 [6] NCCL INFO Connected all rings
+ip-26-0-170-31:3197186:3197791 [6] NCCL INFO Channel 02/0 : 1[6] -> 3[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-31:3197186:3197791 [6] NCCL INFO Channel 03/0 : 1[6] -> 3[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-31:3197186:3197791 [6] NCCL INFO Channel 02/0 : 3[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-31:3197186:3197791 [6] NCCL INFO Channel 03/0 : 3[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330189:3330697 [6] NCCL INFO Connected all rings
+ip-26-0-170-132:3330189:3330697 [6] NCCL INFO Channel 00/0 : 0[6] -> 2[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330189:3330697 [6] NCCL INFO Channel 01/0 : 0[6] -> 2[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330189:3330697 [6] NCCL INFO Channel 00/0 : 2[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330189:3330697 [6] NCCL INFO Channel 01/0 : 2[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-14:2525081:2525584 [6] NCCL INFO Connected all rings
+ip-26-0-162-14:2525081:2525584 [6] NCCL INFO Channel 00/0 : 2[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-14:2525081:2525584 [6] NCCL INFO Channel 01/0 : 2[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-14:2525081:2525584 [6] NCCL INFO Channel 00/0 : 0[6] -> 2[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-14:2525081:2525584 [6] NCCL INFO Channel 01/0 : 0[6] -> 2[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-14:2525081:2525584 [6] NCCL INFO Channel 02/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330189:3330697 [6] NCCL INFO Channel 00/0 : 3[6] -> 2[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-14:2525081:2525584 [6] NCCL INFO Channel 03/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330189:3330697 [6] NCCL INFO Channel 01/0 : 3[6] -> 2[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330189:3330697 [6] NCCL INFO Channel 00/0 : 2[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330189:3330697 [6] NCCL INFO Channel 01/0 : 2[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330189:3330697 [6] NCCL INFO Channel 02/0 : 2[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-132:3330189:3330697 [6] NCCL INFO Channel 03/0 : 2[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934726:935216 [6] NCCL INFO Connected all rings
+ip-26-0-163-58:934726:935216 [6] NCCL INFO Channel 02/0 : 3[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934726:935216 [6] NCCL INFO Channel 03/0 : 3[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934726:935216 [6] NCCL INFO Channel 02/0 : 1[6] -> 3[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934726:935216 [6] NCCL INFO Channel 03/0 : 1[6] -> 3[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934726:935216 [6] NCCL INFO Channel 00/0 : 2[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934726:935216 [6] NCCL INFO Channel 01/0 : 2[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-31:3197186:3197791 [6] NCCL INFO Channel 00/0 : 3[6] -> 2[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934726:935216 [6] NCCL INFO Channel 02/0 : 2[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-31:3197186:3197791 [6] NCCL INFO Channel 01/0 : 3[6] -> 2[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934726:935216 [6] NCCL INFO Channel 03/0 : 2[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934726:935216 [6] NCCL INFO Channel 02/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-163-58:934726:935216 [6] NCCL INFO Channel 03/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-170-31:3197187:3197788 [7] NCCL INFO Connected all trees
+ip-26-0-170-31:3197187:3197788 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-170-31:3197187:3197788 [7] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-170-31:3197187:3197788 [7] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-170-31:3197184:3197789 [4] NCCL INFO Connected all trees
+ip-26-0-170-31:3197184:3197789 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-170-31:3197184:3197789 [4] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-170-31:3197184:3197789 [4] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-170-31:3197187:3197788 [7] NCCL INFO comm 0xc405430 rank 3 nranks 4 cudaDev 7 nvmlDev 7 busId ca000 commId 0x434f5bae1e1954ec - Init COMPLETE
+ip-26-0-170-31:3197185:3197790 [5] NCCL INFO Connected all trees
+ip-26-0-170-31:3197185:3197790 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-170-31:3197185:3197790 [5] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-170-31:3197185:3197790 [5] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-170-31:3197184:3197789 [4] NCCL INFO comm 0xbea2d40 rank 3 nranks 4 cudaDev 4 nvmlDev 4 busId 97000 commId 0xdb7e12b19ea1b0d1 - Init COMPLETE
+ip-26-0-170-31:3197185:3197790 [5] NCCL INFO comm 0xa5ed570 rank 3 nranks 4 cudaDev 5 nvmlDev 5 busId a8000 commId 0x630d4051fd216a10 - Init COMPLETE
+ip-26-0-170-31:3197186:3197791 [6] NCCL INFO Connected all trees
+ip-26-0-170-31:3197186:3197791 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-170-31:3197186:3197791 [6] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-170-31:3197186:3197791 [6] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-162-14:2525077:2525586 [2] NCCL INFO Connected all rings
+ip-26-0-162-14:2525077:2525586 [2] NCCL INFO Channel 00/0 : 2[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-14:2525077:2525586 [2] NCCL INFO Channel 01/0 : 2[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-14:2525077:2525586 [2] NCCL INFO Channel 00/0 : 0[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-14:2525077:2525586 [2] NCCL INFO Channel 01/0 : 0[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-14:2525075:2525592 [0] NCCL INFO Connected all rings
+ip-26-0-162-14:2525075:2525592 [0] NCCL INFO Channel 00/0 : 2[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-14:2525075:2525592 [0] NCCL INFO Channel 01/0 : 2[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-14:2525075:2525592 [0] NCCL INFO Channel 00/0 : 0[0] -> 2[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-14:2525075:2525592 [0] NCCL INFO Channel 01/0 : 0[0] -> 2[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-14:2525079:2525580 [4] NCCL INFO Connected all trees
+ip-26-0-162-14:2525079:2525580 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-162-14:2525079:2525580 [4] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-162-14:2525079:2525580 [4] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-170-31:3197186:3197791 [6] NCCL INFO comm 0xbdd67c0 rank 3 nranks 4 cudaDev 6 nvmlDev 6 busId b9000 commId 0x383416e98fe27c57 - Init COMPLETE
+ip-26-0-170-31:3197182:3197797 [2] NCCL INFO Connected all rings
+ip-26-0-162-14:2525076:2525588 [1] NCCL INFO Connected all rings
+ip-26-0-170-31:3197182:3197797 [2] NCCL INFO Channel 02/0 : 1[2] -> 3[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-31:3197182:3197797 [2] NCCL INFO Channel 03/0 : 1[2] -> 3[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-14:2525076:2525588 [1] NCCL INFO Channel 00/0 : 2[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-31:3197182:3197797 [2] NCCL INFO Channel 02/0 : 3[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-14:2525076:2525588 [1] NCCL INFO Channel 01/0 : 2[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-31:3197182:3197797 [2] NCCL INFO Channel 03/0 : 3[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-14:2525076:2525588 [1] NCCL INFO Channel 00/0 : 0[1] -> 2[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-14:2525076:2525588 [1] NCCL INFO Channel 01/0 : 0[1] -> 2[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-14:2525080:2525583 [5] NCCL INFO Connected all trees
+ip-26-0-162-14:2525080:2525583 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-162-14:2525080:2525583 [5] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-162-14:2525080:2525583 [5] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-162-14:2525078:2525591 [3] NCCL INFO Connected all rings
+ip-26-0-162-14:2525078:2525591 [3] NCCL INFO Channel 00/0 : 2[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-14:2525078:2525591 [3] NCCL INFO Channel 01/0 : 2[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-14:2525078:2525591 [3] NCCL INFO Channel 00/0 : 0[3] -> 2[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-14:2525078:2525591 [3] NCCL INFO Channel 01/0 : 0[3] -> 2[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-14:2525079:2525580 [4] NCCL INFO comm 0xb62b0a0 rank 0 nranks 4 cudaDev 4 nvmlDev 4 busId 97000 commId 0xdb7e12b19ea1b0d1 - Init COMPLETE
+ip-26-0-170-31:3197180:3197800 [0] NCCL INFO Connected all rings
+ip-26-0-170-31:3197180:3197800 [0] NCCL INFO Channel 02/0 : 1[0] -> 3[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-31:3197180:3197800 [0] NCCL INFO Channel 03/0 : 1[0] -> 3[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-31:3197180:3197800 [0] NCCL INFO Channel 02/0 : 3[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-31:3197180:3197800 [0] NCCL INFO Channel 03/0 : 3[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-31:3197181:3197798 [1] NCCL INFO Connected all rings
+ip-26-0-170-31:3197181:3197798 [1] NCCL INFO Channel 02/0 : 1[1] -> 3[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-31:3197181:3197798 [1] NCCL INFO Channel 03/0 : 1[1] -> 3[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-31:3197181:3197798 [1] NCCL INFO Channel 02/0 : 3[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-31:3197181:3197798 [1] NCCL INFO Channel 03/0 : 3[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-31:3197183:3197799 [3] NCCL INFO Connected all rings
+ip-26-0-170-31:3197183:3197799 [3] NCCL INFO Channel 02/0 : 1[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-31:3197183:3197799 [3] NCCL INFO Channel 03/0 : 1[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-31:3197183:3197799 [3] NCCL INFO Channel 02/0 : 3[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-31:3197183:3197799 [3] NCCL INFO Channel 03/0 : 3[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-14:2525080:2525583 [5] NCCL INFO comm 0xbbafcf0 rank 0 nranks 4 cudaDev 5 nvmlDev 5 busId a8000 commId 0x630d4051fd216a10 - Init COMPLETE
+ip-26-0-163-58:934722:935222 [2] NCCL INFO Connected all rings
+ip-26-0-163-58:934722:935222 [2] NCCL INFO Channel 02/0 : 3[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934722:935222 [2] NCCL INFO Channel 03/0 : 3[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934722:935222 [2] NCCL INFO Channel 02/0 : 1[2] -> 3[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934722:935222 [2] NCCL INFO Channel 03/0 : 1[2] -> 3[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934722:935222 [2] NCCL INFO Channel 00/0 : 2[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-31:3197182:3197797 [2] NCCL INFO Channel 00/0 : 3[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934722:935222 [2] NCCL INFO Channel 01/0 : 2[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-31:3197182:3197797 [2] NCCL INFO Channel 01/0 : 3[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934722:935222 [2] NCCL INFO Channel 02/0 : 2[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934722:935222 [2] NCCL INFO Channel 03/0 : 2[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934722:935222 [2] NCCL INFO Channel 02/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-163-58:934722:935222 [2] NCCL INFO Channel 03/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330185:3330698 [2] NCCL INFO Connected all rings
+ip-26-0-170-132:3330185:3330698 [2] NCCL INFO Channel 00/0 : 0[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330185:3330698 [2] NCCL INFO Channel 01/0 : 0[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330185:3330698 [2] NCCL INFO Channel 00/0 : 2[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330185:3330698 [2] NCCL INFO Channel 01/0 : 2[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330185:3330698 [2] NCCL INFO Channel 00/0 : 3[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330185:3330698 [2] NCCL INFO Channel 01/0 : 3[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330185:3330698 [2] NCCL INFO Channel 00/0 : 2[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330185:3330698 [2] NCCL INFO Channel 01/0 : 2[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330185:3330698 [2] NCCL INFO Channel 02/0 : 2[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330185:3330698 [2] NCCL INFO Channel 03/0 : 2[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-14:2525077:2525586 [2] NCCL INFO Channel 02/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-14:2525077:2525586 [2] NCCL INFO Channel 03/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-170-132:3330183:3330701 [0] NCCL INFO Connected all rings
+ip-26-0-170-132:3330183:3330701 [0] NCCL INFO Channel 00/0 : 0[0] -> 2[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330183:3330701 [0] NCCL INFO Channel 01/0 : 0[0] -> 2[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330183:3330701 [0] NCCL INFO Channel 00/0 : 2[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330183:3330701 [0] NCCL INFO Channel 01/0 : 2[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330183:3330701 [0] NCCL INFO Channel 00/0 : 3[0] -> 2[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330183:3330701 [0] NCCL INFO Channel 01/0 : 3[0] -> 2[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330183:3330701 [0] NCCL INFO Channel 00/0 : 2[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330183:3330701 [0] NCCL INFO Channel 01/0 : 2[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330183:3330701 [0] NCCL INFO Channel 02/0 : 2[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330183:3330701 [0] NCCL INFO Channel 03/0 : 2[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-14:2525075:2525592 [0] NCCL INFO Channel 02/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-14:2525075:2525592 [0] NCCL INFO Channel 03/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934720:935224 [0] NCCL INFO Connected all rings
+ip-26-0-163-58:934720:935224 [0] NCCL INFO Channel 02/0 : 3[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934720:935224 [0] NCCL INFO Channel 03/0 : 3[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934720:935224 [0] NCCL INFO Channel 02/0 : 1[0] -> 3[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934720:935224 [0] NCCL INFO Channel 03/0 : 1[0] -> 3[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934720:935224 [0] NCCL INFO Channel 00/0 : 2[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-31:3197180:3197800 [0] NCCL INFO Channel 00/0 : 3[0] -> 2[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934720:935224 [0] NCCL INFO Channel 01/0 : 2[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-31:3197180:3197800 [0] NCCL INFO Channel 01/0 : 3[0] -> 2[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934720:935224 [0] NCCL INFO Channel 02/0 : 2[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934720:935224 [0] NCCL INFO Channel 03/0 : 2[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934720:935224 [0] NCCL INFO Channel 02/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-163-58:934720:935224 [0] NCCL INFO Channel 03/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-170-132:3330186:3330700 [3] NCCL INFO Connected all rings
+ip-26-0-170-132:3330186:3330700 [3] NCCL INFO Channel 00/0 : 0[3] -> 2[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330186:3330700 [3] NCCL INFO Channel 01/0 : 0[3] -> 2[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330186:3330700 [3] NCCL INFO Channel 00/0 : 2[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330186:3330700 [3] NCCL INFO Channel 01/0 : 2[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-14:2525082:2525581 [7] NCCL INFO Connected all trees
+ip-26-0-162-14:2525082:2525581 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-162-14:2525082:2525581 [7] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-162-14:2525082:2525581 [7] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-170-132:3330184:3330699 [1] NCCL INFO Connected all rings
+ip-26-0-170-132:3330184:3330699 [1] NCCL INFO Channel 00/0 : 0[1] -> 2[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330186:3330700 [3] NCCL INFO Channel 00/0 : 3[3] -> 2[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-14:2525078:2525591 [3] NCCL INFO Channel 02/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330184:3330699 [1] NCCL INFO Channel 01/0 : 0[1] -> 2[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330186:3330700 [3] NCCL INFO Channel 01/0 : 3[3] -> 2[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-14:2525078:2525591 [3] NCCL INFO Channel 03/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330184:3330699 [1] NCCL INFO Channel 00/0 : 2[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330186:3330700 [3] NCCL INFO Channel 00/0 : 2[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330184:3330699 [1] NCCL INFO Channel 01/0 : 2[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330186:3330700 [3] NCCL INFO Channel 01/0 : 2[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330186:3330700 [3] NCCL INFO Channel 02/0 : 2[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330186:3330700 [3] NCCL INFO Channel 03/0 : 2[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330184:3330699 [1] NCCL INFO Channel 00/0 : 3[1] -> 2[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-14:2525076:2525588 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330184:3330699 [1] NCCL INFO Channel 01/0 : 3[1] -> 2[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-14:2525076:2525588 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330184:3330699 [1] NCCL INFO Channel 00/0 : 2[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330184:3330699 [1] NCCL INFO Channel 01/0 : 2[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330184:3330699 [1] NCCL INFO Channel 02/0 : 2[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330184:3330699 [1] NCCL INFO Channel 03/0 : 2[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330190:3330695 [7] NCCL INFO Connected all trees
+ip-26-0-170-132:3330190:3330695 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-170-132:3330190:3330695 [7] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-170-132:3330190:3330695 [7] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-162-14:2525082:2525581 [7] NCCL INFO comm 0xbddb140 rank 0 nranks 4 cudaDev 7 nvmlDev 7 busId ca000 commId 0x434f5bae1e1954ec - Init COMPLETE
+ip-26-0-163-58:934721:935223 [1] NCCL INFO Connected all rings
+ip-26-0-163-58:934721:935223 [1] NCCL INFO Channel 02/0 : 3[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934721:935223 [1] NCCL INFO Channel 03/0 : 3[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934721:935223 [1] NCCL INFO Channel 02/0 : 1[1] -> 3[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934721:935223 [1] NCCL INFO Channel 03/0 : 1[1] -> 3[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934721:935223 [1] NCCL INFO Channel 00/0 : 2[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934721:935223 [1] NCCL INFO Channel 01/0 : 2[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934721:935223 [1] NCCL INFO Channel 02/0 : 2[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934721:935223 [1] NCCL INFO Channel 03/0 : 2[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934721:935223 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-163-58:934721:935223 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-31:3197181:3197798 [1] NCCL INFO Channel 00/0 : 3[1] -> 2[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-31:3197181:3197798 [1] NCCL INFO Channel 01/0 : 3[1] -> 2[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-170-132:3330190:3330695 [7] NCCL INFO comm 0xc2b5e60 rank 2 nranks 4 cudaDev 7 nvmlDev 7 busId ca000 commId 0x434f5bae1e1954ec - Init COMPLETE
+ip-26-0-170-132:3330187:3330694 [4] NCCL INFO Connected all trees
+ip-26-0-170-132:3330187:3330694 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-170-132:3330187:3330694 [4] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-170-132:3330187:3330694 [4] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-163-58:934723:935225 [3] NCCL INFO Connected all rings
+ip-26-0-163-58:934723:935225 [3] NCCL INFO Channel 02/0 : 3[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934723:935225 [3] NCCL INFO Channel 03/0 : 3[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934723:935225 [3] NCCL INFO Channel 02/0 : 1[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934723:935225 [3] NCCL INFO Channel 03/0 : 1[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-14:2525081:2525584 [6] NCCL INFO Connected all trees
+ip-26-0-162-14:2525081:2525584 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-162-14:2525081:2525584 [6] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-162-14:2525081:2525584 [6] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-170-132:3330188:3330696 [5] NCCL INFO Connected all trees
+ip-26-0-170-132:3330188:3330696 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-170-132:3330188:3330696 [5] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-170-132:3330188:3330696 [5] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-163-58:934723:935225 [3] NCCL INFO Channel 00/0 : 2[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-31:3197183:3197799 [3] NCCL INFO Channel 00/0 : 3[3] -> 2[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934723:935225 [3] NCCL INFO Channel 01/0 : 2[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-31:3197183:3197799 [3] NCCL INFO Channel 01/0 : 3[3] -> 2[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934723:935225 [3] NCCL INFO Channel 02/0 : 2[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934723:935225 [3] NCCL INFO Channel 03/0 : 2[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934723:935225 [3] NCCL INFO Channel 02/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-163-58:934723:935225 [3] NCCL INFO Channel 03/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-170-132:3330187:3330694 [4] NCCL INFO comm 0xb7d2640 rank 2 nranks 4 cudaDev 4 nvmlDev 4 busId 97000 commId 0xdb7e12b19ea1b0d1 - Init COMPLETE
+ip-26-0-162-14:2525081:2525584 [6] NCCL INFO comm 0xb45d830 rank 0 nranks 4 cudaDev 6 nvmlDev 6 busId b9000 commId 0x383416e98fe27c57 - Init COMPLETE
+ip-26-0-163-58:934724:935213 [4] NCCL INFO Connected all trees
+ip-26-0-163-58:934724:935213 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-163-58:934724:935213 [4] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-163-58:934724:935213 [4] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-170-132:3330188:3330696 [5] NCCL INFO comm 0xc288720 rank 2 nranks 4 cudaDev 5 nvmlDev 5 busId a8000 commId 0x630d4051fd216a10 - Init COMPLETE
+ip-26-0-163-58:934727:935214 [7] NCCL INFO Connected all trees
+ip-26-0-163-58:934727:935214 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-163-58:934727:935214 [7] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-163-58:934727:935214 [7] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-163-58:934725:935215 [5] NCCL INFO Connected all trees
+ip-26-0-163-58:934725:935215 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-163-58:934725:935215 [5] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-163-58:934725:935215 [5] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-163-58:934724:935213 [4] NCCL INFO comm 0xb65df60 rank 1 nranks 4 cudaDev 4 nvmlDev 4 busId 97000 commId 0xdb7e12b19ea1b0d1 - Init COMPLETE
+ip-26-0-163-58:934727:935214 [7] NCCL INFO comm 0xb7be4b0 rank 1 nranks 4 cudaDev 7 nvmlDev 7 busId ca000 commId 0x434f5bae1e1954ec - Init COMPLETE
+ip-26-0-163-58:934725:935215 [5] NCCL INFO comm 0xa8498a0 rank 1 nranks 4 cudaDev 5 nvmlDev 5 busId a8000 commId 0x630d4051fd216a10 - Init COMPLETE
+ip-26-0-170-132:3330189:3330697 [6] NCCL INFO Connected all trees
+ip-26-0-170-132:3330189:3330697 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-170-132:3330189:3330697 [6] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-170-132:3330189:3330697 [6] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-170-31:3197182:3197797 [2] NCCL INFO Connected all trees
+ip-26-0-170-31:3197182:3197797 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-170-31:3197182:3197797 [2] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-170-31:3197182:3197797 [2] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-170-132:3330189:3330697 [6] NCCL INFO comm 0xa547f50 rank 2 nranks 4 cudaDev 6 nvmlDev 6 busId b9000 commId 0x383416e98fe27c57 - Init COMPLETE
+ip-26-0-170-31:3197182:3197797 [2] NCCL INFO comm 0xa69f1d0 rank 3 nranks 4 cudaDev 2 nvmlDev 2 busId 75000 commId 0x82c4762eb1811da0 - Init COMPLETE
+ip-26-0-163-58:934726:935216 [6] NCCL INFO Connected all trees
+ip-26-0-163-58:934726:935216 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-163-58:934726:935216 [6] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-163-58:934726:935216 [6] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-170-31:3197180:3197800 [0] NCCL INFO Connected all trees
+ip-26-0-170-31:3197180:3197800 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-170-31:3197180:3197800 [0] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-170-31:3197180:3197800 [0] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-163-58:934726:935216 [6] NCCL INFO comm 0xbbadcc0 rank 1 nranks 4 cudaDev 6 nvmlDev 6 busId b9000 commId 0x383416e98fe27c57 - Init COMPLETE
+ip-26-0-170-31:3197180:3197800 [0] NCCL INFO comm 0xbf3f770 rank 3 nranks 4 cudaDev 0 nvmlDev 0 busId 53000 commId 0x655e0359b702bd7a - Init COMPLETE
+ip-26-0-162-14:2525076:2525588 [1] NCCL INFO Connected all trees
+ip-26-0-162-14:2525076:2525588 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-162-14:2525076:2525588 [1] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-162-14:2525076:2525588 [1] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-162-14:2525076:2525588 [1] NCCL INFO comm 0xc2d65a0 rank 0 nranks 4 cudaDev 1 nvmlDev 1 busId 64000 commId 0xcdbbe7a2eb7417f4 - Init COMPLETE
+ip-26-0-170-31:3197181:3197798 [1] NCCL INFO Connected all trees
+ip-26-0-170-31:3197181:3197798 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-170-31:3197181:3197798 [1] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-170-31:3197181:3197798 [1] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-170-31:3197183:3197799 [3] NCCL INFO Connected all trees
+ip-26-0-170-31:3197183:3197799 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-170-31:3197183:3197799 [3] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-170-31:3197183:3197799 [3] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-162-14:2525077:2525586 [2] NCCL INFO Connected all trees
+ip-26-0-162-14:2525077:2525586 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-162-14:2525077:2525586 [2] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-162-14:2525077:2525586 [2] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-170-31:3197181:3197798 [1] NCCL INFO comm 0xb969e60 rank 3 nranks 4 cudaDev 1 nvmlDev 1 busId 64000 commId 0xcdbbe7a2eb7417f4 - Init COMPLETE
+ip-26-0-170-31:3197183:3197799 [3] NCCL INFO comm 0xbd0ed80 rank 3 nranks 4 cudaDev 3 nvmlDev 3 busId 86000 commId 0x26d03f9b08f1ec31 - Init COMPLETE
+ip-26-0-162-14:2525077:2525586 [2] NCCL INFO comm 0xc13bea0 rank 0 nranks 4 cudaDev 2 nvmlDev 2 busId 75000 commId 0x82c4762eb1811da0 - Init COMPLETE
+ip-26-0-162-14:2525078:2525591 [3] NCCL INFO Connected all trees
+ip-26-0-162-14:2525078:2525591 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-162-14:2525078:2525591 [3] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-162-14:2525078:2525591 [3] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-170-132:3330185:3330698 [2] NCCL INFO Connected all trees
+ip-26-0-170-132:3330185:3330698 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-170-132:3330185:3330698 [2] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-170-132:3330185:3330698 [2] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-162-14:2525078:2525591 [3] NCCL INFO comm 0xb18deb0 rank 0 nranks 4 cudaDev 3 nvmlDev 3 busId 86000 commId 0x26d03f9b08f1ec31 - Init COMPLETE
+ip-26-0-163-58:934722:935222 [2] NCCL INFO Connected all trees
+ip-26-0-163-58:934722:935222 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-163-58:934722:935222 [2] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-163-58:934722:935222 [2] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-162-14:2525075:2525592 [0] NCCL INFO Connected all trees
+ip-26-0-162-14:2525075:2525592 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-162-14:2525075:2525592 [0] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-162-14:2525075:2525592 [0] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-170-132:3330185:3330698 [2] NCCL INFO comm 0xb83aa70 rank 2 nranks 4 cudaDev 2 nvmlDev 2 busId 75000 commId 0x82c4762eb1811da0 - Init COMPLETE
+ip-26-0-170-132:3330183:3330701 [0] NCCL INFO Connected all trees
+ip-26-0-170-132:3330183:3330701 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-170-132:3330183:3330701 [0] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-170-132:3330183:3330701 [0] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-163-58:934722:935222 [2] NCCL INFO comm 0xa867ed0 rank 1 nranks 4 cudaDev 2 nvmlDev 2 busId 75000 commId 0x82c4762eb1811da0 - Init COMPLETE
+ip-26-0-162-14:2525075:2525592 [0] NCCL INFO comm 0xb7dc430 rank 0 nranks 4 cudaDev 0 nvmlDev 0 busId 53000 commId 0x655e0359b702bd7a - Init COMPLETE
+ip-26-0-170-132:3330183:3330701 [0] NCCL INFO comm 0xb26f790 rank 2 nranks 4 cudaDev 0 nvmlDev 0 busId 53000 commId 0x655e0359b702bd7a - Init COMPLETE
+ip-26-0-170-132:3330186:3330700 [3] NCCL INFO Connected all trees
+ip-26-0-170-132:3330186:3330700 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-170-132:3330186:3330700 [3] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-170-132:3330186:3330700 [3] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-163-58:934721:935223 [1] NCCL INFO Connected all trees
+ip-26-0-163-58:934721:935223 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-163-58:934721:935223 [1] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-163-58:934721:935223 [1] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-163-58:934720:935224 [0] NCCL INFO Connected all trees
+ip-26-0-163-58:934720:935224 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-163-58:934720:935224 [0] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-163-58:934720:935224 [0] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-170-132:3330184:3330699 [1] NCCL INFO Connected all trees
+ip-26-0-170-132:3330184:3330699 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-170-132:3330184:3330699 [1] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-170-132:3330184:3330699 [1] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-170-132:3330186:3330700 [3] NCCL INFO comm 0xacb0aa0 rank 2 nranks 4 cudaDev 3 nvmlDev 3 busId 86000 commId 0x26d03f9b08f1ec31 - Init COMPLETE
+ip-26-0-163-58:934721:935223 [1] NCCL INFO comm 0xc2e8db0 rank 1 nranks 4 cudaDev 1 nvmlDev 1 busId 64000 commId 0xcdbbe7a2eb7417f4 - Init COMPLETE
+ip-26-0-163-58:934720:935224 [0] NCCL INFO comm 0xb5a4670 rank 1 nranks 4 cudaDev 0 nvmlDev 0 busId 53000 commId 0x655e0359b702bd7a - Init COMPLETE
+ip-26-0-163-58:934723:935225 [3] NCCL INFO Connected all trees
+ip-26-0-163-58:934723:935225 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-163-58:934723:935225 [3] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-163-58:934723:935225 [3] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-170-132:3330184:3330699 [1] NCCL INFO comm 0xb3a7d00 rank 2 nranks 4 cudaDev 1 nvmlDev 1 busId 64000 commId 0xcdbbe7a2eb7417f4 - Init COMPLETE
+ip-26-0-163-58:934723:935225 [3] NCCL INFO comm 0xb868d80 rank 1 nranks 4 cudaDev 3 nvmlDev 3 busId 86000 commId 0x26d03f9b08f1ec31 - Init COMPLETE
+12/28/2024 01:36:42 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]: Total number of parameters: 469G (895263.62MiB)
+12/28/2024 01:36:42 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]: Local number of parameters: 30.1G (57346.00MiB)
+12/28/2024 01:36:42 [INFO|DP=0|PP=2|TP=0|ip-26-0-170-132]: Local number of parameters: 28.6G (54561.94MiB)
+12/28/2024 01:36:42 [INFO|DP=0|PP=1|TP=0|ip-26-0-163-58]: Local number of parameters: 29.5G (56322.00MiB)
+12/28/2024 01:36:42 [INFO|DP=0|PP=1|TP=0|ip-26-0-163-58]: [After model building] Memory usage: 56322.04MiB. Peak allocated: 56322.05MiB Peak reserved: 57444.00MiB
+12/28/2024 01:36:42 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]: [After model building] Memory usage: 57346.04MiB. Peak allocated: 57346.05MiB Peak reserved: 58372.00MiB
+12/28/2024 01:36:42 [INFO|DP=0|PP=2|TP=0|ip-26-0-170-132]: [After model building] Memory usage: 54561.98MiB. Peak allocated: 54561.99MiB Peak reserved: 55682.00MiB
+12/28/2024 01:36:42 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]: No checkpoint path provided.
+12/28/2024 01:36:42 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]: Parametrizing model parameters using StandardParametrizator
+12/28/2024 01:36:42 [INFO|DP=0|PP=3|TP=0|ip-26-0-170-31]: Local number of parameters: 29.1G (55585.97MiB)
+12/28/2024 01:36:42 [INFO|DP=0|PP=3|TP=0|ip-26-0-170-31]: [After model building] Memory usage: 55586.01MiB. Peak allocated: 55586.02MiB Peak reserved: 56612.00MiB
+12/28/2024 01:36:42 [INFO|DP=0|PP=0|TP=2|ip-26-0-162-14]: Local number of parameters: 30.1G (57346.00MiB)
+12/28/2024 01:36:42 [INFO|DP=0|PP=0|TP=1|ip-26-0-162-14]: Local number of parameters: 30.1G (57346.00MiB)
+12/28/2024 01:36:42 [INFO|DP=0|PP=0|TP=2|ip-26-0-162-14]: [After model building] Memory usage: 57346.04MiB. Peak allocated: 57346.05MiB Peak reserved: 58372.00MiB
+12/28/2024 01:36:42 [INFO|DP=0|PP=0|TP=1|ip-26-0-162-14]: [After model building] Memory usage: 57346.04MiB. Peak allocated: 57346.05MiB Peak reserved: 58372.00MiB
+12/28/2024 01:36:42 [INFO|DP=0|PP=0|TP=3|ip-26-0-162-14]: Local number of parameters: 30.1G (57346.00MiB)
+12/28/2024 01:36:42 [INFO|DP=0|PP=0|TP=3|ip-26-0-162-14]: [After model building] Memory usage: 57346.04MiB. Peak allocated: 57346.05MiB Peak reserved: 58372.00MiB
+12/28/2024 01:36:42 [INFO|DP=0|PP=3|TP=2|ip-26-0-170-31]: Local number of parameters: 29.1G (55585.97MiB)
+12/28/2024 01:36:42 [INFO|DP=0|PP=1|TP=2|ip-26-0-163-58]: Local number of parameters: 29.5G (56322.00MiB)
+12/28/2024 01:36:42 [INFO|DP=0|PP=2|TP=2|ip-26-0-170-132]: Local number of parameters: 28.6G (54561.94MiB)
+ip-26-0-163-58:934720:935245 [0] NCCL INFO Using network Libfabric
+ip-26-0-163-58:934724:935246 [4] NCCL INFO Using network Libfabric
+ip-26-0-170-132:3330183:3330721 [0] NCCL INFO Using network Libfabric
+12/28/2024 01:36:42 [INFO|DP=0|PP=2|TP=3|ip-26-0-170-132]: Local number of parameters: 28.6G (54561.94MiB)
+ip-26-0-170-132:3330187:3330722 [4] NCCL INFO Using network Libfabric
+12/28/2024 01:36:42 [INFO|DP=0|PP=3|TP=2|ip-26-0-170-31]: [After model building] Memory usage: 55586.01MiB. Peak allocated: 55586.02MiB Peak reserved: 56612.00MiB
+12/28/2024 01:36:42 [INFO|DP=0|PP=1|TP=3|ip-26-0-163-58]: Local number of parameters: 29.5G (56322.00MiB)
+ip-26-0-162-14:2525079:2525613 [4] NCCL INFO Using network Libfabric
+12/28/2024 01:36:42 [INFO|DP=0|PP=3|TP=3|ip-26-0-170-31]: Local number of parameters: 29.1G (55585.97MiB)
+ip-26-0-162-14:2525075:2525612 [0] NCCL INFO Using network Libfabric
+12/28/2024 01:36:42 [INFO|DP=0|PP=1|TP=2|ip-26-0-163-58]: [After model building] Memory usage: 56322.04MiB. Peak allocated: 56322.05MiB Peak reserved: 57444.00MiB
+12/28/2024 01:36:42 [INFO|DP=0|PP=2|TP=2|ip-26-0-170-132]: [After model building] Memory usage: 54561.98MiB. Peak allocated: 54561.99MiB Peak reserved: 55682.00MiB
+ip-26-0-163-58:934720:935245 [0] NCCL INFO comm 0xb5e0780 rank 0 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0x9171fa6542ce32fa - Init START
+ip-26-0-163-58:934724:935246 [4] NCCL INFO comm 0xb699a90 rank 1 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0x9171fa6542ce32fa - Init START
+ip-26-0-163-58:934724:935246 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-163-58:934720:935245 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-170-132:3330183:3330721 [0] NCCL INFO comm 0xb2af600 rank 0 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0x4da867b891ba2670 - Init START
+ip-26-0-170-132:3330187:3330722 [4] NCCL INFO comm 0xb8116d0 rank 1 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0x4da867b891ba2670 - Init START
+12/28/2024 01:36:42 [INFO|DP=0|PP=2|TP=3|ip-26-0-170-132]: [After model building] Memory usage: 54561.98MiB. Peak allocated: 54561.99MiB Peak reserved: 55682.00MiB
+12/28/2024 01:36:42 [INFO|DP=0|PP=1|TP=3|ip-26-0-163-58]: [After model building] Memory usage: 56322.04MiB. Peak allocated: 56322.05MiB Peak reserved: 57444.00MiB
+ip-26-0-170-132:3330187:3330722 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-170-132:3330183:3330721 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-162-14:2525079:2525613 [4] NCCL INFO comm 0xb658f10 rank 1 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0x228c3daf39bf52b8 - Init START
+ip-26-0-162-14:2525075:2525612 [0] NCCL INFO comm 0xb80ecd0 rank 0 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0x228c3daf39bf52b8 - Init START
+12/28/2024 01:36:42 [INFO|DP=0|PP=3|TP=3|ip-26-0-170-31]: [After model building] Memory usage: 55586.01MiB. Peak allocated: 55586.02MiB Peak reserved: 56612.00MiB
+ip-26-0-162-14:2525079:2525613 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-162-14:2525075:2525612 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/253
+12/28/2024 01:36:42 [INFO|DP=0|PP=2|TP=1|ip-26-0-170-132]: Local number of parameters: 28.6G (54561.94MiB)
+12/28/2024 01:36:42 [INFO|DP=0|PP=3|TP=1|ip-26-0-170-31]: Local number of parameters: 29.1G (55585.97MiB)
+12/28/2024 01:36:42 [INFO|DP=0|PP=1|TP=1|ip-26-0-163-58]: Local number of parameters: 29.5G (56322.00MiB)
+12/28/2024 01:36:42 [INFO|DP=0|PP=1|TP=1|ip-26-0-163-58]: [After model building] Memory usage: 56322.04MiB. Peak allocated: 56322.05MiB Peak reserved: 57444.00MiB
+12/28/2024 01:36:42 [INFO|DP=0|PP=2|TP=1|ip-26-0-170-132]: [After model building] Memory usage: 54561.98MiB. Peak allocated: 54561.99MiB Peak reserved: 55682.00MiB
+12/28/2024 01:36:42 [INFO|DP=0|PP=3|TP=1|ip-26-0-170-31]: [After model building] Memory usage: 55586.01MiB. Peak allocated: 55586.02MiB Peak reserved: 56612.00MiB
+ip-26-0-170-31:3197180:3197820 [0] NCCL INFO Using network Libfabric
+ip-26-0-170-31:3197184:3197821 [4] NCCL INFO Using network Libfabric
+ip-26-0-170-31:3197184:3197821 [4] NCCL INFO comm 0xbedda70 rank 1 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0xa8f4448dd46d93b1 - Init START
+ip-26-0-170-31:3197180:3197820 [0] NCCL INFO comm 0xbf79c50 rank 0 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0xa8f4448dd46d93b1 - Init START
+ip-26-0-170-31:3197184:3197821 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-170-31:3197180:3197820 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-162-14:2525076:2525615 [1] NCCL INFO Using network Libfabric
+ip-26-0-162-14:2525080:2525616 [5] NCCL INFO Using network Libfabric
+ip-26-0-162-14:2525078:2525618 [3] NCCL INFO Using network Libfabric
+ip-26-0-162-14:2525076:2525615 [1] NCCL INFO comm 0xc303810 rank 0 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0xc23f2d54f1a22bd4 - Init START
+ip-26-0-162-14:2525080:2525616 [5] NCCL INFO comm 0xbbdcd90 rank 1 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0xc23f2d54f1a22bd4 - Init START
+ip-26-0-162-14:2525076:2525615 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-162-14:2525080:2525616 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-162-14:2525077:2525620 [2] NCCL INFO Using network Libfabric
+ip-26-0-162-14:2525081:2525622 [6] NCCL INFO Using network Libfabric
+ip-26-0-162-14:2525082:2525621 [7] NCCL INFO Using network Libfabric
+ip-26-0-162-14:2525081:2525622 [6] NCCL INFO comm 0xb48b4d0 rank 1 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0x603d878f636bb0f4 - Init START
+ip-26-0-162-14:2525077:2525620 [2] NCCL INFO comm 0xc14fe00 rank 0 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0x603d878f636bb0f4 - Init START
+ip-26-0-162-14:2525082:2525621 [7] NCCL INFO comm 0xbe081b0 rank 1 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0x57a6b54ba58bf2d8 - Init START
+ip-26-0-162-14:2525078:2525618 [3] NCCL INFO comm 0xb1bbb00 rank 0 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0x57a6b54ba58bf2d8 - Init START
+ip-26-0-162-14:2525077:2525620 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-162-14:2525081:2525622 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-162-14:2525078:2525618 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-162-14:2525082:2525621 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+ip-26-0-170-31:3197182:3197823 [2] NCCL INFO Using network Libfabric
+ip-26-0-163-58:934726:935249 [6] NCCL INFO Using network Libfabric
+ip-26-0-170-31:3197186:3197824 [6] NCCL INFO Using network Libfabric
+NCCL version 2.18.5+cuda12.2
+ip-26-0-170-132:3330186:3330725 [3] NCCL INFO Using network Libfabric
+ip-26-0-163-58:934722:935248 [2] NCCL INFO Using network Libfabric
+ip-26-0-170-132:3330190:3330726 [7] NCCL INFO Using network Libfabric
+NCCL version 2.18.5+cuda12.2
+ip-26-0-170-132:3330185:3330727 [2] NCCL INFO Using network Libfabric
+ip-26-0-170-31:3197186:3197824 [6] NCCL INFO comm 0xbe11790 rank 1 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0x453cb490910f7b74 - Init START
+ip-26-0-170-31:3197182:3197823 [2] NCCL INFO comm 0xa6da550 rank 0 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0x453cb490910f7b74 - Init START
+NCCL version 2.18.5+cuda12.2
+ip-26-0-170-132:3330189:3330728 [6] NCCL INFO Using network Libfabric
+ip-26-0-170-31:3197186:3197824 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-170-31:3197182:3197823 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-163-58:934726:935249 [6] NCCL INFO comm 0xbbea240 rank 1 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0x1b908c1f11d5daeb - Init START
+ip-26-0-163-58:934722:935248 [2] NCCL INFO comm 0xa8a5a70 rank 0 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0x1b908c1f11d5daeb - Init START
+ip-26-0-170-132:3330190:3330726 [7] NCCL INFO comm 0xc2f4190 rank 1 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0x7fd882a83b32ac48 - Init START
+ip-26-0-170-132:3330186:3330725 [3] NCCL INFO comm 0xacef840 rank 0 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0x7fd882a83b32ac48 - Init START
+NCCL version 2.18.5+cuda12.2
+ip-26-0-163-58:934726:935249 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-163-58:934722:935248 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-170-132:3330185:3330727 [2] NCCL INFO comm 0xb879ad0 rank 0 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0xbb4dd0623a3af179 - Init START
+ip-26-0-170-132:3330189:3330728 [6] NCCL INFO comm 0xa5862f0 rank 1 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0xbb4dd0623a3af179 - Init START
+ip-26-0-163-58:934723:935251 [3] NCCL INFO Using network Libfabric
+ip-26-0-170-132:3330186:3330725 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-170-132:3330190:3330726 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-170-132:3330185:3330727 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-163-58:934727:935252 [7] NCCL INFO Using network Libfabric
+ip-26-0-170-132:3330189:3330728 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+NCCL version 2.18.5+cuda12.2
+ip-26-0-170-31:3197183:3197826 [3] NCCL INFO Using network Libfabric
+ip-26-0-170-132:3330184:3330730 [1] NCCL INFO Using network Libfabric
+ip-26-0-170-132:3330188:3330731 [5] NCCL INFO Using network Libfabric
+ip-26-0-163-58:934727:935252 [7] NCCL INFO comm 0xb7f9b50 rank 1 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0x9d71cac1fadb657d - Init START
+ip-26-0-163-58:934723:935251 [3] NCCL INFO comm 0xb8a40e0 rank 0 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0x9d71cac1fadb657d - Init START
+ip-26-0-163-58:934727:935252 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-163-58:934723:935251 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-163-58:934721:935254 [1] NCCL INFO Using network Libfabric
+ip-26-0-163-58:934725:935255 [5] NCCL INFO Using network Libfabric
+ip-26-0-170-132:3330188:3330731 [5] NCCL INFO comm 0xc2c69b0 rank 1 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0x7a9ebadf14aae2f9 - Init START
+ip-26-0-170-132:3330184:3330730 [1] NCCL INFO comm 0xb3e6490 rank 0 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0x7a9ebadf14aae2f9 - Init START
+ip-26-0-170-31:3197187:3197827 [7] NCCL INFO Using network Libfabric
+ip-26-0-170-132:3330188:3330731 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-170-132:3330184:3330730 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-170-31:3197181:3197829 [1] NCCL INFO Using network Libfabric
+ip-26-0-163-58:934721:935254 [1] NCCL INFO comm 0xc323ca0 rank 0 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0xb44c20b3113eb843 - Init START
+ip-26-0-163-58:934725:935255 [5] NCCL INFO comm 0xa885490 rank 1 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0xb44c20b3113eb843 - Init START
+ip-26-0-170-31:3197183:3197826 [3] NCCL INFO comm 0xbd49970 rank 0 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0xce3d779e80e630f - Init START
+ip-26-0-170-31:3197187:3197827 [7] NCCL INFO comm 0xc4401d0 rank 1 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0xce3d779e80e630f - Init START
+ip-26-0-163-58:934725:935255 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-163-58:934721:935254 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-170-31:3197185:3197830 [5] NCCL INFO Using network Libfabric
+ip-26-0-170-31:3197187:3197827 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-170-31:3197183:3197826 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-170-31:3197181:3197829 [1] NCCL INFO comm 0xb9a5230 rank 0 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0x8915cf04ef157888 - Init START
+ip-26-0-170-31:3197185:3197830 [5] NCCL INFO comm 0xa627ed0 rank 1 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0x8915cf04ef157888 - Init START
+ip-26-0-170-31:3197181:3197829 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-170-31:3197185:3197830 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/216
+ip-26-0-170-31:3197180:3197820 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-170-31:3197184:3197821 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-170-31:3197180:3197820 [0] NCCL INFO Channel 00/24 :    0   1
+ip-26-0-170-31:3197180:3197820 [0] NCCL INFO Channel 01/24 :    0   1
+ip-26-0-170-31:3197180:3197820 [0] NCCL INFO Channel 02/24 :    0   1
+ip-26-0-170-31:3197180:3197820 [0] NCCL INFO Channel 03/24 :    0   1
+ip-26-0-170-31:3197180:3197820 [0] NCCL INFO Channel 04/24 :    0   1
+ip-26-0-170-31:3197180:3197820 [0] NCCL INFO Channel 05/24 :    0   1
+ip-26-0-170-31:3197180:3197820 [0] NCCL INFO Channel 06/24 :    0   1
+ip-26-0-170-31:3197180:3197820 [0] NCCL INFO Channel 07/24 :    0   1
+ip-26-0-170-31:3197180:3197820 [0] NCCL INFO Channel 08/24 :    0   1
+ip-26-0-170-31:3197180:3197820 [0] NCCL INFO Channel 09/24 :    0   1
+ip-26-0-170-31:3197180:3197820 [0] NCCL INFO Channel 10/24 :    0   1
+ip-26-0-170-31:3197180:3197820 [0] NCCL INFO Channel 11/24 :    0   1
+ip-26-0-170-31:3197180:3197820 [0] NCCL INFO Channel 12/24 :    0   1
+ip-26-0-170-31:3197184:3197821 [4] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] -1/-1/-1->1->0 [3] -1/-1/-1->1->0 [4] -1/-1/-1->1->0 [5] -1/-1/-1->1->0 [6] 0/-1/-1->1->-1 [7] 0/-1/-1->1->-1 [8] 0/-1/-1->1->-1 [9] 0/-1/-1->1->-1 [10] 0/-1/-1->1->-1 [11] 0/-1/-1->1->-1 [12] -1/-1/-1->1->0 [13] -1/-1/-1->1->0 [14] -1/-1/-1->1->0 [15] -1/-1/-1->1->0 [16] -1/-1/-1->1->0 [17] -1/-1/-1->1->0 [18] 0/-1/-1->1->-1 [19] 0/-1/-1->1->-1 [20] 0/-1/-1->1->-1 [21] 0/-1/-1->1->-1 [22] 0/-1/-1->1->-1 [23] 0/-1/-1->1->-1
+ip-26-0-170-31:3197180:3197820 [0] NCCL INFO Channel 13/24 :    0   1
+ip-26-0-170-31:3197184:3197821 [4] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-170-31:3197180:3197820 [0] NCCL INFO Channel 14/24 :    0   1
+ip-26-0-170-31:3197180:3197820 [0] NCCL INFO Channel 15/24 :    0   1
+ip-26-0-170-31:3197180:3197820 [0] NCCL INFO Channel 16/24 :    0   1
+ip-26-0-170-31:3197180:3197820 [0] NCCL INFO Channel 17/24 :    0   1
+ip-26-0-170-31:3197180:3197820 [0] NCCL INFO Channel 18/24 :    0   1
+ip-26-0-170-31:3197180:3197820 [0] NCCL INFO Channel 19/24 :    0   1
+ip-26-0-170-31:3197180:3197820 [0] NCCL INFO Channel 20/24 :    0   1
+ip-26-0-170-31:3197180:3197820 [0] NCCL INFO Channel 21/24 :    0   1
+ip-26-0-170-31:3197180:3197820 [0] NCCL INFO Channel 22/24 :    0   1
+ip-26-0-170-31:3197180:3197820 [0] NCCL INFO Channel 23/24 :    0   1
+ip-26-0-170-31:3197180:3197820 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] -1/-1/-1->0->1 [7] -1/-1/-1->0->1 [8] -1/-1/-1->0->1 [9] -1/-1/-1->0->1 [10] -1/-1/-1->0->1 [11] -1/-1/-1->0->1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] -1/-1/-1->0->1 [19] -1/-1/-1->0->1 [20] -1/-1/-1->0->1 [21] -1/-1/-1->0->1 [22] -1/-1/-1->0->1 [23] -1/-1/-1->0->1
+ip-26-0-170-31:3197180:3197820 [0] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-170-31:3197182:3197823 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-170-31:3197187:3197827 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-170-31:3197185:3197830 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-162-14:2525080:2525616 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-163-58:934725:935255 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-162-14:2525078:2525618 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-170-31:3197181:3197829 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-162-14:2525075:2525612 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-170-31:3197181:3197829 [1] NCCL INFO Channel 00/24 :    0   1
+ip-26-0-170-31:3197185:3197830 [5] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] -1/-1/-1->1->0 [3] -1/-1/-1->1->0 [4] -1/-1/-1->1->0 [5] -1/-1/-1->1->0 [6] 0/-1/-1->1->-1 [7] 0/-1/-1->1->-1 [8] 0/-1/-1->1->-1 [9] 0/-1/-1->1->-1 [10] 0/-1/-1->1->-1 [11] 0/-1/-1->1->-1 [12] -1/-1/-1->1->0 [13] -1/-1/-1->1->0 [14] -1/-1/-1->1->0 [15] -1/-1/-1->1->0 [16] -1/-1/-1->1->0 [17] -1/-1/-1->1->0 [18] 0/-1/-1->1->-1 [19] 0/-1/-1->1->-1 [20] 0/-1/-1->1->-1 [21] 0/-1/-1->1->-1 [22] 0/-1/-1->1->-1 [23] 0/-1/-1->1->-1
+ip-26-0-170-31:3197185:3197830 [5] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-170-31:3197181:3197829 [1] NCCL INFO Channel 01/24 :    0   1
+ip-26-0-170-31:3197181:3197829 [1] NCCL INFO Channel 02/24 :    0   1
+ip-26-0-170-31:3197181:3197829 [1] NCCL INFO Channel 03/24 :    0   1
+ip-26-0-170-31:3197181:3197829 [1] NCCL INFO Channel 04/24 :    0   1
+ip-26-0-170-31:3197181:3197829 [1] NCCL INFO Channel 05/24 :    0   1
+ip-26-0-170-31:3197181:3197829 [1] NCCL INFO Channel 06/24 :    0   1
+ip-26-0-170-31:3197181:3197829 [1] NCCL INFO Channel 07/24 :    0   1
+ip-26-0-170-31:3197181:3197829 [1] NCCL INFO Channel 08/24 :    0   1
+ip-26-0-170-31:3197181:3197829 [1] NCCL INFO Channel 09/24 :    0   1
+ip-26-0-170-31:3197181:3197829 [1] NCCL INFO Channel 10/24 :    0   1
+ip-26-0-170-31:3197181:3197829 [1] NCCL INFO Channel 11/24 :    0   1
+ip-26-0-170-31:3197181:3197829 [1] NCCL INFO Channel 12/24 :    0   1
+ip-26-0-170-31:3197181:3197829 [1] NCCL INFO Channel 13/24 :    0   1
+ip-26-0-170-31:3197181:3197829 [1] NCCL INFO Channel 14/24 :    0   1
+ip-26-0-170-31:3197181:3197829 [1] NCCL INFO Channel 15/24 :    0   1
+ip-26-0-170-31:3197181:3197829 [1] NCCL INFO Channel 16/24 :    0   1
+ip-26-0-170-31:3197181:3197829 [1] NCCL INFO Channel 17/24 :    0   1
+ip-26-0-170-31:3197181:3197829 [1] NCCL INFO Channel 18/24 :    0   1
+ip-26-0-170-31:3197181:3197829 [1] NCCL INFO Channel 19/24 :    0   1
+ip-26-0-170-31:3197181:3197829 [1] NCCL INFO Channel 20/24 :    0   1
+ip-26-0-170-31:3197181:3197829 [1] NCCL INFO Channel 21/24 :    0   1
+ip-26-0-170-31:3197181:3197829 [1] NCCL INFO Channel 22/24 :    0   1
+ip-26-0-170-31:3197181:3197829 [1] NCCL INFO Channel 23/24 :    0   1
+ip-26-0-170-31:3197181:3197829 [1] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] -1/-1/-1->0->1 [7] -1/-1/-1->0->1 [8] -1/-1/-1->0->1 [9] -1/-1/-1->0->1 [10] -1/-1/-1->0->1 [11] -1/-1/-1->0->1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] -1/-1/-1->0->1 [19] -1/-1/-1->0->1 [20] -1/-1/-1->0->1 [21] -1/-1/-1->0->1 [22] -1/-1/-1->0->1 [23] -1/-1/-1->0->1
+ip-26-0-170-31:3197181:3197829 [1] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-170-31:3197183:3197826 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-170-31:3197183:3197826 [3] NCCL INFO Channel 00/24 :    0   1
+ip-26-0-170-31:3197183:3197826 [3] NCCL INFO Channel 01/24 :    0   1
+ip-26-0-170-31:3197183:3197826 [3] NCCL INFO Channel 02/24 :    0   1
+ip-26-0-170-31:3197183:3197826 [3] NCCL INFO Channel 03/24 :    0   1
+ip-26-0-170-31:3197183:3197826 [3] NCCL INFO Channel 04/24 :    0   1
+ip-26-0-170-31:3197183:3197826 [3] NCCL INFO Channel 05/24 :    0   1
+ip-26-0-170-31:3197183:3197826 [3] NCCL INFO Channel 06/24 :    0   1
+ip-26-0-170-31:3197187:3197827 [7] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] -1/-1/-1->1->0 [3] -1/-1/-1->1->0 [4] -1/-1/-1->1->0 [5] -1/-1/-1->1->0 [6] 0/-1/-1->1->-1 [7] 0/-1/-1->1->-1 [8] 0/-1/-1->1->-1 [9] 0/-1/-1->1->-1 [10] 0/-1/-1->1->-1 [11] 0/-1/-1->1->-1 [12] -1/-1/-1->1->0 [13] -1/-1/-1->1->0 [14] -1/-1/-1->1->0 [15] -1/-1/-1->1->0 [16] -1/-1/-1->1->0 [17] -1/-1/-1->1->0 [18] 0/-1/-1->1->-1 [19] 0/-1/-1->1->-1 [20] 0/-1/-1->1->-1 [21] 0/-1/-1->1->-1 [22] 0/-1/-1->1->-1 [23] 0/-1/-1->1->-1
+ip-26-0-170-31:3197183:3197826 [3] NCCL INFO Channel 07/24 :    0   1
+ip-26-0-170-31:3197187:3197827 [7] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-170-31:3197183:3197826 [3] NCCL INFO Channel 08/24 :    0   1
+ip-26-0-170-31:3197183:3197826 [3] NCCL INFO Channel 09/24 :    0   1
+ip-26-0-170-31:3197183:3197826 [3] NCCL INFO Channel 10/24 :    0   1
+ip-26-0-170-31:3197183:3197826 [3] NCCL INFO Channel 11/24 :    0   1
+ip-26-0-170-31:3197183:3197826 [3] NCCL INFO Channel 12/24 :    0   1
+ip-26-0-170-31:3197183:3197826 [3] NCCL INFO Channel 13/24 :    0   1
+ip-26-0-170-31:3197183:3197826 [3] NCCL INFO Channel 14/24 :    0   1
+ip-26-0-170-31:3197183:3197826 [3] NCCL INFO Channel 15/24 :    0   1
+ip-26-0-170-31:3197183:3197826 [3] NCCL INFO Channel 16/24 :    0   1
+ip-26-0-170-31:3197183:3197826 [3] NCCL INFO Channel 17/24 :    0   1
+ip-26-0-170-31:3197183:3197826 [3] NCCL INFO Channel 18/24 :    0   1
+ip-26-0-170-31:3197183:3197826 [3] NCCL INFO Channel 19/24 :    0   1
+ip-26-0-170-31:3197183:3197826 [3] NCCL INFO Channel 20/24 :    0   1
+ip-26-0-170-31:3197183:3197826 [3] NCCL INFO Channel 21/24 :    0   1
+ip-26-0-170-31:3197183:3197826 [3] NCCL INFO Channel 22/24 :    0   1
+ip-26-0-170-31:3197183:3197826 [3] NCCL INFO Channel 23/24 :    0   1
+ip-26-0-170-31:3197183:3197826 [3] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] -1/-1/-1->0->1 [7] -1/-1/-1->0->1 [8] -1/-1/-1->0->1 [9] -1/-1/-1->0->1 [10] -1/-1/-1->0->1 [11] -1/-1/-1->0->1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] -1/-1/-1->0->1 [19] -1/-1/-1->0->1 [20] -1/-1/-1->0->1 [21] -1/-1/-1->0->1 [22] -1/-1/-1->0->1 [23] -1/-1/-1->0->1
+ip-26-0-170-31:3197183:3197826 [3] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-163-58:934722:935248 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-163-58:934724:935246 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-170-31:3197186:3197824 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-170-31:3197182:3197823 [2] NCCL INFO Channel 00/24 :    0   1
+ip-26-0-170-31:3197182:3197823 [2] NCCL INFO Channel 01/24 :    0   1
+ip-26-0-170-31:3197182:3197823 [2] NCCL INFO Channel 02/24 :    0   1
+ip-26-0-170-31:3197182:3197823 [2] NCCL INFO Channel 03/24 :    0   1
+ip-26-0-170-31:3197182:3197823 [2] NCCL INFO Channel 04/24 :    0   1
+ip-26-0-170-31:3197182:3197823 [2] NCCL INFO Channel 05/24 :    0   1
+ip-26-0-170-31:3197182:3197823 [2] NCCL INFO Channel 06/24 :    0   1
+ip-26-0-170-31:3197182:3197823 [2] NCCL INFO Channel 07/24 :    0   1
+ip-26-0-170-31:3197182:3197823 [2] NCCL INFO Channel 08/24 :    0   1
+ip-26-0-170-31:3197182:3197823 [2] NCCL INFO Channel 09/24 :    0   1
+ip-26-0-170-31:3197182:3197823 [2] NCCL INFO Channel 10/24 :    0   1
+ip-26-0-170-31:3197182:3197823 [2] NCCL INFO Channel 11/24 :    0   1
+ip-26-0-170-31:3197182:3197823 [2] NCCL INFO Channel 12/24 :    0   1
+ip-26-0-170-31:3197182:3197823 [2] NCCL INFO Channel 13/24 :    0   1
+ip-26-0-170-31:3197186:3197824 [6] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] -1/-1/-1->1->0 [3] -1/-1/-1->1->0 [4] -1/-1/-1->1->0 [5] -1/-1/-1->1->0 [6] 0/-1/-1->1->-1 [7] 0/-1/-1->1->-1 [8] 0/-1/-1->1->-1 [9] 0/-1/-1->1->-1 [10] 0/-1/-1->1->-1 [11] 0/-1/-1->1->-1 [12] -1/-1/-1->1->0 [13] -1/-1/-1->1->0 [14] -1/-1/-1->1->0 [15] -1/-1/-1->1->0 [16] -1/-1/-1->1->0 [17] -1/-1/-1->1->0 [18] 0/-1/-1->1->-1 [19] 0/-1/-1->1->-1 [20] 0/-1/-1->1->-1 [21] 0/-1/-1->1->-1 [22] 0/-1/-1->1->-1 [23] 0/-1/-1->1->-1
+ip-26-0-170-31:3197182:3197823 [2] NCCL INFO Channel 14/24 :    0   1
+ip-26-0-170-31:3197182:3197823 [2] NCCL INFO Channel 15/24 :    0   1
+ip-26-0-170-31:3197186:3197824 [6] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-170-31:3197182:3197823 [2] NCCL INFO Channel 16/24 :    0   1
+ip-26-0-170-31:3197182:3197823 [2] NCCL INFO Channel 17/24 :    0   1
+ip-26-0-170-31:3197182:3197823 [2] NCCL INFO Channel 18/24 :    0   1
+ip-26-0-170-31:3197182:3197823 [2] NCCL INFO Channel 19/24 :    0   1
+ip-26-0-170-31:3197182:3197823 [2] NCCL INFO Channel 20/24 :    0   1
+ip-26-0-170-31:3197182:3197823 [2] NCCL INFO Channel 21/24 :    0   1
+ip-26-0-170-31:3197182:3197823 [2] NCCL INFO Channel 22/24 :    0   1
+ip-26-0-170-31:3197182:3197823 [2] NCCL INFO Channel 23/24 :    0   1
+ip-26-0-170-31:3197182:3197823 [2] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] -1/-1/-1->0->1 [7] -1/-1/-1->0->1 [8] -1/-1/-1->0->1 [9] -1/-1/-1->0->1 [10] -1/-1/-1->0->1 [11] -1/-1/-1->0->1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] -1/-1/-1->0->1 [19] -1/-1/-1->0->1 [20] -1/-1/-1->0->1 [21] -1/-1/-1->0->1 [22] -1/-1/-1->0->1 [23] -1/-1/-1->0->1
+ip-26-0-170-31:3197182:3197823 [2] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-163-58:934726:935249 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-163-58:934722:935248 [2] NCCL INFO Channel 00/24 :    0   1
+ip-26-0-163-58:934722:935248 [2] NCCL INFO Channel 01/24 :    0   1
+ip-26-0-163-58:934722:935248 [2] NCCL INFO Channel 02/24 :    0   1
+ip-26-0-163-58:934722:935248 [2] NCCL INFO Channel 03/24 :    0   1
+ip-26-0-163-58:934722:935248 [2] NCCL INFO Channel 04/24 :    0   1
+ip-26-0-163-58:934722:935248 [2] NCCL INFO Channel 05/24 :    0   1
+ip-26-0-163-58:934722:935248 [2] NCCL INFO Channel 06/24 :    0   1
+ip-26-0-163-58:934722:935248 [2] NCCL INFO Channel 07/24 :    0   1
+ip-26-0-163-58:934722:935248 [2] NCCL INFO Channel 08/24 :    0   1
+ip-26-0-163-58:934726:935249 [6] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] -1/-1/-1->1->0 [3] -1/-1/-1->1->0 [4] -1/-1/-1->1->0 [5] -1/-1/-1->1->0 [6] 0/-1/-1->1->-1 [7] 0/-1/-1->1->-1 [8] 0/-1/-1->1->-1 [9] 0/-1/-1->1->-1 [10] 0/-1/-1->1->-1 [11] 0/-1/-1->1->-1 [12] -1/-1/-1->1->0 [13] -1/-1/-1->1->0 [14] -1/-1/-1->1->0 [15] -1/-1/-1->1->0 [16] -1/-1/-1->1->0 [17] -1/-1/-1->1->0 [18] 0/-1/-1->1->-1 [19] 0/-1/-1->1->-1 [20] 0/-1/-1->1->-1 [21] 0/-1/-1->1->-1 [22] 0/-1/-1->1->-1 [23] 0/-1/-1->1->-1
+ip-26-0-163-58:934722:935248 [2] NCCL INFO Channel 09/24 :    0   1
+ip-26-0-163-58:934722:935248 [2] NCCL INFO Channel 10/24 :    0   1
+ip-26-0-163-58:934726:935249 [6] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-163-58:934722:935248 [2] NCCL INFO Channel 11/24 :    0   1
+ip-26-0-163-58:934722:935248 [2] NCCL INFO Channel 12/24 :    0   1
+ip-26-0-163-58:934722:935248 [2] NCCL INFO Channel 13/24 :    0   1
+ip-26-0-163-58:934722:935248 [2] NCCL INFO Channel 14/24 :    0   1
+ip-26-0-163-58:934722:935248 [2] NCCL INFO Channel 15/24 :    0   1
+ip-26-0-163-58:934722:935248 [2] NCCL INFO Channel 16/24 :    0   1
+ip-26-0-163-58:934722:935248 [2] NCCL INFO Channel 17/24 :    0   1
+ip-26-0-163-58:934722:935248 [2] NCCL INFO Channel 18/24 :    0   1
+ip-26-0-163-58:934722:935248 [2] NCCL INFO Channel 19/24 :    0   1
+ip-26-0-163-58:934722:935248 [2] NCCL INFO Channel 20/24 :    0   1
+ip-26-0-163-58:934722:935248 [2] NCCL INFO Channel 21/24 :    0   1
+ip-26-0-163-58:934722:935248 [2] NCCL INFO Channel 22/24 :    0   1
+ip-26-0-163-58:934722:935248 [2] NCCL INFO Channel 23/24 :    0   1
+ip-26-0-163-58:934722:935248 [2] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] -1/-1/-1->0->1 [7] -1/-1/-1->0->1 [8] -1/-1/-1->0->1 [9] -1/-1/-1->0->1 [10] -1/-1/-1->0->1 [11] -1/-1/-1->0->1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] -1/-1/-1->0->1 [19] -1/-1/-1->0->1 [20] -1/-1/-1->0->1 [21] -1/-1/-1->0->1 [22] -1/-1/-1->0->1 [23] -1/-1/-1->0->1
+ip-26-0-163-58:934722:935248 [2] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-163-58:934727:935252 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-162-14:2525082:2525621 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-162-14:2525078:2525618 [3] NCCL INFO Channel 00/24 :    0   1
+ip-26-0-162-14:2525078:2525618 [3] NCCL INFO Channel 01/24 :    0   1
+ip-26-0-162-14:2525078:2525618 [3] NCCL INFO Channel 02/24 :    0   1
+ip-26-0-162-14:2525078:2525618 [3] NCCL INFO Channel 03/24 :    0   1
+ip-26-0-162-14:2525078:2525618 [3] NCCL INFO Channel 04/24 :    0   1
+ip-26-0-162-14:2525078:2525618 [3] NCCL INFO Channel 05/24 :    0   1
+ip-26-0-162-14:2525078:2525618 [3] NCCL INFO Channel 06/24 :    0   1
+ip-26-0-162-14:2525078:2525618 [3] NCCL INFO Channel 07/24 :    0   1
+ip-26-0-162-14:2525078:2525618 [3] NCCL INFO Channel 08/24 :    0   1
+ip-26-0-162-14:2525078:2525618 [3] NCCL INFO Channel 09/24 :    0   1
+ip-26-0-162-14:2525078:2525618 [3] NCCL INFO Channel 10/24 :    0   1
+ip-26-0-162-14:2525078:2525618 [3] NCCL INFO Channel 11/24 :    0   1
+ip-26-0-162-14:2525082:2525621 [7] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] -1/-1/-1->1->0 [3] -1/-1/-1->1->0 [4] -1/-1/-1->1->0 [5] -1/-1/-1->1->0 [6] 0/-1/-1->1->-1 [7] 0/-1/-1->1->-1 [8] 0/-1/-1->1->-1 [9] 0/-1/-1->1->-1 [10] 0/-1/-1->1->-1 [11] 0/-1/-1->1->-1 [12] -1/-1/-1->1->0 [13] -1/-1/-1->1->0 [14] -1/-1/-1->1->0 [15] -1/-1/-1->1->0 [16] -1/-1/-1->1->0 [17] -1/-1/-1->1->0 [18] 0/-1/-1->1->-1 [19] 0/-1/-1->1->-1 [20] 0/-1/-1->1->-1 [21] 0/-1/-1->1->-1 [22] 0/-1/-1->1->-1 [23] 0/-1/-1->1->-1
+ip-26-0-162-14:2525078:2525618 [3] NCCL INFO Channel 12/24 :    0   1
+ip-26-0-162-14:2525082:2525621 [7] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-162-14:2525078:2525618 [3] NCCL INFO Channel 13/24 :    0   1
+ip-26-0-162-14:2525078:2525618 [3] NCCL INFO Channel 14/24 :    0   1
+ip-26-0-162-14:2525078:2525618 [3] NCCL INFO Channel 15/24 :    0   1
+ip-26-0-162-14:2525078:2525618 [3] NCCL INFO Channel 16/24 :    0   1
+ip-26-0-162-14:2525078:2525618 [3] NCCL INFO Channel 17/24 :    0   1
+ip-26-0-162-14:2525078:2525618 [3] NCCL INFO Channel 18/24 :    0   1
+ip-26-0-162-14:2525078:2525618 [3] NCCL INFO Channel 19/24 :    0   1
+ip-26-0-162-14:2525078:2525618 [3] NCCL INFO Channel 20/24 :    0   1
+ip-26-0-162-14:2525078:2525618 [3] NCCL INFO Channel 21/24 :    0   1
+ip-26-0-162-14:2525078:2525618 [3] NCCL INFO Channel 22/24 :    0   1
+ip-26-0-162-14:2525078:2525618 [3] NCCL INFO Channel 23/24 :    0   1
+ip-26-0-162-14:2525078:2525618 [3] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] -1/-1/-1->0->1 [7] -1/-1/-1->0->1 [8] -1/-1/-1->0->1 [9] -1/-1/-1->0->1 [10] -1/-1/-1->0->1 [11] -1/-1/-1->0->1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] -1/-1/-1->0->1 [19] -1/-1/-1->0->1 [20] -1/-1/-1->0->1 [21] -1/-1/-1->0->1 [22] -1/-1/-1->0->1 [23] -1/-1/-1->0->1
+ip-26-0-162-14:2525078:2525618 [3] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-163-58:934720:935245 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-163-58:934724:935246 [4] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] -1/-1/-1->1->0 [3] -1/-1/-1->1->0 [4] -1/-1/-1->1->0 [5] -1/-1/-1->1->0 [6] 0/-1/-1->1->-1 [7] 0/-1/-1->1->-1 [8] 0/-1/-1->1->-1 [9] 0/-1/-1->1->-1 [10] 0/-1/-1->1->-1 [11] 0/-1/-1->1->-1 [12] -1/-1/-1->1->0 [13] -1/-1/-1->1->0 [14] -1/-1/-1->1->0 [15] -1/-1/-1->1->0 [16] -1/-1/-1->1->0 [17] -1/-1/-1->1->0 [18] 0/-1/-1->1->-1 [19] 0/-1/-1->1->-1 [20] 0/-1/-1->1->-1 [21] 0/-1/-1->1->-1 [22] 0/-1/-1->1->-1 [23] 0/-1/-1->1->-1
+ip-26-0-163-58:934724:935246 [4] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-163-58:934720:935245 [0] NCCL INFO Channel 00/24 :    0   1
+ip-26-0-163-58:934720:935245 [0] NCCL INFO Channel 01/24 :    0   1
+ip-26-0-163-58:934720:935245 [0] NCCL INFO Channel 02/24 :    0   1
+ip-26-0-163-58:934720:935245 [0] NCCL INFO Channel 03/24 :    0   1
+ip-26-0-163-58:934720:935245 [0] NCCL INFO Channel 04/24 :    0   1
+ip-26-0-163-58:934720:935245 [0] NCCL INFO Channel 05/24 :    0   1
+ip-26-0-163-58:934720:935245 [0] NCCL INFO Channel 06/24 :    0   1
+ip-26-0-163-58:934720:935245 [0] NCCL INFO Channel 07/24 :    0   1
+ip-26-0-163-58:934720:935245 [0] NCCL INFO Channel 08/24 :    0   1
+ip-26-0-163-58:934720:935245 [0] NCCL INFO Channel 09/24 :    0   1
+ip-26-0-163-58:934720:935245 [0] NCCL INFO Channel 10/24 :    0   1
+ip-26-0-163-58:934720:935245 [0] NCCL INFO Channel 11/24 :    0   1
+ip-26-0-163-58:934720:935245 [0] NCCL INFO Channel 12/24 :    0   1
+ip-26-0-163-58:934720:935245 [0] NCCL INFO Channel 13/24 :    0   1
+ip-26-0-163-58:934720:935245 [0] NCCL INFO Channel 14/24 :    0   1
+ip-26-0-163-58:934720:935245 [0] NCCL INFO Channel 15/24 :    0   1
+ip-26-0-163-58:934720:935245 [0] NCCL INFO Channel 16/24 :    0   1
+ip-26-0-163-58:934720:935245 [0] NCCL INFO Channel 17/24 :    0   1
+ip-26-0-163-58:934720:935245 [0] NCCL INFO Channel 18/24 :    0   1
+ip-26-0-163-58:934720:935245 [0] NCCL INFO Channel 19/24 :    0   1
+ip-26-0-163-58:934720:935245 [0] NCCL INFO Channel 20/24 :    0   1
+ip-26-0-163-58:934720:935245 [0] NCCL INFO Channel 21/24 :    0   1
+ip-26-0-163-58:934720:935245 [0] NCCL INFO Channel 22/24 :    0   1
+ip-26-0-163-58:934720:935245 [0] NCCL INFO Channel 23/24 :    0   1
+ip-26-0-163-58:934720:935245 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] -1/-1/-1->0->1 [7] -1/-1/-1->0->1 [8] -1/-1/-1->0->1 [9] -1/-1/-1->0->1 [10] -1/-1/-1->0->1 [11] -1/-1/-1->0->1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] -1/-1/-1->0->1 [19] -1/-1/-1->0->1 [20] -1/-1/-1->0->1 [21] -1/-1/-1->0->1 [22] -1/-1/-1->0->1 [23] -1/-1/-1->0->1
+ip-26-0-163-58:934720:935245 [0] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-170-31:3197180:3197820 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[4] via P2P/IPC
+ip-26-0-170-31:3197180:3197820 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[4] via P2P/IPC
+ip-26-0-162-14:2525076:2525615 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-162-14:2525076:2525615 [1] NCCL INFO Channel 00/24 :    0   1
+ip-26-0-162-14:2525076:2525615 [1] NCCL INFO Channel 01/24 :    0   1
+ip-26-0-162-14:2525076:2525615 [1] NCCL INFO Channel 02/24 :    0   1
+ip-26-0-162-14:2525076:2525615 [1] NCCL INFO Channel 03/24 :    0   1
+ip-26-0-162-14:2525080:2525616 [5] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] -1/-1/-1->1->0 [3] -1/-1/-1->1->0 [4] -1/-1/-1->1->0 [5] -1/-1/-1->1->0 [6] 0/-1/-1->1->-1 [7] 0/-1/-1->1->-1 [8] 0/-1/-1->1->-1 [9] 0/-1/-1->1->-1 [10] 0/-1/-1->1->-1 [11] 0/-1/-1->1->-1 [12] -1/-1/-1->1->0 [13] -1/-1/-1->1->0 [14] -1/-1/-1->1->0 [15] -1/-1/-1->1->0 [16] -1/-1/-1->1->0 [17] -1/-1/-1->1->0 [18] 0/-1/-1->1->-1 [19] 0/-1/-1->1->-1 [20] 0/-1/-1->1->-1 [21] 0/-1/-1->1->-1 [22] 0/-1/-1->1->-1 [23] 0/-1/-1->1->-1
+ip-26-0-162-14:2525076:2525615 [1] NCCL INFO Channel 04/24 :    0   1
+ip-26-0-162-14:2525080:2525616 [5] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-162-14:2525076:2525615 [1] NCCL INFO Channel 05/24 :    0   1
+ip-26-0-162-14:2525076:2525615 [1] NCCL INFO Channel 06/24 :    0   1
+ip-26-0-162-14:2525076:2525615 [1] NCCL INFO Channel 07/24 :    0   1
+ip-26-0-162-14:2525076:2525615 [1] NCCL INFO Channel 08/24 :    0   1
+ip-26-0-162-14:2525076:2525615 [1] NCCL INFO Channel 09/24 :    0   1
+ip-26-0-162-14:2525076:2525615 [1] NCCL INFO Channel 10/24 :    0   1
+ip-26-0-162-14:2525076:2525615 [1] NCCL INFO Channel 11/24 :    0   1
+ip-26-0-162-14:2525076:2525615 [1] NCCL INFO Channel 12/24 :    0   1
+ip-26-0-162-14:2525076:2525615 [1] NCCL INFO Channel 13/24 :    0   1
+ip-26-0-162-14:2525076:2525615 [1] NCCL INFO Channel 14/24 :    0   1
+ip-26-0-162-14:2525076:2525615 [1] NCCL INFO Channel 15/24 :    0   1
+ip-26-0-162-14:2525076:2525615 [1] NCCL INFO Channel 16/24 :    0   1
+ip-26-0-162-14:2525076:2525615 [1] NCCL INFO Channel 17/24 :    0   1
+ip-26-0-162-14:2525076:2525615 [1] NCCL INFO Channel 18/24 :    0   1
+ip-26-0-162-14:2525076:2525615 [1] NCCL INFO Channel 19/24 :    0   1
+ip-26-0-162-14:2525076:2525615 [1] NCCL INFO Channel 20/24 :    0   1
+ip-26-0-162-14:2525076:2525615 [1] NCCL INFO Channel 21/24 :    0   1
+ip-26-0-162-14:2525076:2525615 [1] NCCL INFO Channel 22/24 :    0   1
+ip-26-0-162-14:2525076:2525615 [1] NCCL INFO Channel 23/24 :    0   1
+ip-26-0-162-14:2525076:2525615 [1] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] -1/-1/-1->0->1 [7] -1/-1/-1->0->1 [8] -1/-1/-1->0->1 [9] -1/-1/-1->0->1 [10] -1/-1/-1->0->1 [11] -1/-1/-1->0->1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] -1/-1/-1->0->1 [19] -1/-1/-1->0->1 [20] -1/-1/-1->0->1 [21] -1/-1/-1->0->1 [22] -1/-1/-1->0->1 [23] -1/-1/-1->0->1
+ip-26-0-162-14:2525076:2525615 [1] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-163-58:934723:935251 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-163-58:934727:935252 [7] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] -1/-1/-1->1->0 [3] -1/-1/-1->1->0 [4] -1/-1/-1->1->0 [5] -1/-1/-1->1->0 [6] 0/-1/-1->1->-1 [7] 0/-1/-1->1->-1 [8] 0/-1/-1->1->-1 [9] 0/-1/-1->1->-1 [10] 0/-1/-1->1->-1 [11] 0/-1/-1->1->-1 [12] -1/-1/-1->1->0 [13] -1/-1/-1->1->0 [14] -1/-1/-1->1->0 [15] -1/-1/-1->1->0 [16] -1/-1/-1->1->0 [17] -1/-1/-1->1->0 [18] 0/-1/-1->1->-1 [19] 0/-1/-1->1->-1 [20] 0/-1/-1->1->-1 [21] 0/-1/-1->1->-1 [22] 0/-1/-1->1->-1 [23] 0/-1/-1->1->-1
+ip-26-0-163-58:934723:935251 [3] NCCL INFO Channel 00/24 :    0   1
+ip-26-0-163-58:934727:935252 [7] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-163-58:934723:935251 [3] NCCL INFO Channel 01/24 :    0   1
+ip-26-0-163-58:934723:935251 [3] NCCL INFO Channel 02/24 :    0   1
+ip-26-0-163-58:934723:935251 [3] NCCL INFO Channel 03/24 :    0   1
+ip-26-0-163-58:934723:935251 [3] NCCL INFO Channel 04/24 :    0   1
+ip-26-0-163-58:934723:935251 [3] NCCL INFO Channel 05/24 :    0   1
+ip-26-0-163-58:934723:935251 [3] NCCL INFO Channel 06/24 :    0   1
+ip-26-0-163-58:934723:935251 [3] NCCL INFO Channel 07/24 :    0   1
+ip-26-0-163-58:934723:935251 [3] NCCL INFO Channel 08/24 :    0   1
+ip-26-0-163-58:934723:935251 [3] NCCL INFO Channel 09/24 :    0   1
+ip-26-0-163-58:934723:935251 [3] NCCL INFO Channel 10/24 :    0   1
+ip-26-0-163-58:934723:935251 [3] NCCL INFO Channel 11/24 :    0   1
+ip-26-0-163-58:934723:935251 [3] NCCL INFO Channel 12/24 :    0   1
+ip-26-0-163-58:934723:935251 [3] NCCL INFO Channel 13/24 :    0   1
+ip-26-0-163-58:934723:935251 [3] NCCL INFO Channel 14/24 :    0   1
+ip-26-0-163-58:934723:935251 [3] NCCL INFO Channel 15/24 :    0   1
+ip-26-0-163-58:934723:935251 [3] NCCL INFO Channel 16/24 :    0   1
+ip-26-0-163-58:934723:935251 [3] NCCL INFO Channel 17/24 :    0   1
+ip-26-0-163-58:934723:935251 [3] NCCL INFO Channel 18/24 :    0   1
+ip-26-0-163-58:934723:935251 [3] NCCL INFO Channel 19/24 :    0   1
+ip-26-0-163-58:934723:935251 [3] NCCL INFO Channel 20/24 :    0   1
+ip-26-0-163-58:934723:935251 [3] NCCL INFO Channel 21/24 :    0   1
+ip-26-0-163-58:934723:935251 [3] NCCL INFO Channel 22/24 :    0   1
+ip-26-0-163-58:934723:935251 [3] NCCL INFO Channel 23/24 :    0   1
+ip-26-0-163-58:934723:935251 [3] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] -1/-1/-1->0->1 [7] -1/-1/-1->0->1 [8] -1/-1/-1->0->1 [9] -1/-1/-1->0->1 [10] -1/-1/-1->0->1 [11] -1/-1/-1->0->1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] -1/-1/-1->0->1 [19] -1/-1/-1->0->1 [20] -1/-1/-1->0->1 [21] -1/-1/-1->0->1 [22] -1/-1/-1->0->1 [23] -1/-1/-1->0->1
+ip-26-0-163-58:934723:935251 [3] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-170-31:3197180:3197820 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[4] via P2P/IPC
+ip-26-0-170-31:3197184:3197821 [4] NCCL INFO Channel 00/0 : 1[4] -> 0[0] via P2P/IPC
+ip-26-0-170-31:3197180:3197820 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[4] via P2P/IPC
+ip-26-0-170-31:3197184:3197821 [4] NCCL INFO Channel 01/0 : 1[4] -> 0[0] via P2P/IPC
+ip-26-0-170-31:3197180:3197820 [0] NCCL INFO Channel 04/0 : 0[0] -> 1[4] via P2P/IPC
+ip-26-0-170-31:3197184:3197821 [4] NCCL INFO Channel 02/0 : 1[4] -> 0[0] via P2P/IPC
+ip-26-0-170-31:3197184:3197821 [4] NCCL INFO Channel 03/0 : 1[4] -> 0[0] via P2P/IPC
+ip-26-0-170-31:3197184:3197821 [4] NCCL INFO Channel 04/0 : 1[4] -> 0[0] via P2P/IPC
+ip-26-0-162-14:2525081:2525622 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-170-31:3197184:3197821 [4] NCCL INFO Channel 05/0 : 1[4] -> 0[0] via P2P/IPC
+ip-26-0-170-31:3197184:3197821 [4] NCCL INFO Channel 06/0 : 1[4] -> 0[0] via P2P/IPC
+ip-26-0-170-31:3197184:3197821 [4] NCCL INFO Channel 07/0 : 1[4] -> 0[0] via P2P/IPC
+ip-26-0-170-31:3197180:3197820 [0] NCCL INFO Channel 05/0 : 0[0] -> 1[4] via P2P/IPC
+ip-26-0-170-31:3197184:3197821 [4] NCCL INFO Channel 08/0 : 1[4] -> 0[0] via P2P/IPC
+ip-26-0-163-58:934721:935254 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-163-58:934721:935254 [1] NCCL INFO Channel 00/24 :    0   1
+ip-26-0-163-58:934725:935255 [5] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] -1/-1/-1->1->0 [3] -1/-1/-1->1->0 [4] -1/-1/-1->1->0 [5] -1/-1/-1->1->0 [6] 0/-1/-1->1->-1 [7] 0/-1/-1->1->-1 [8] 0/-1/-1->1->-1 [9] 0/-1/-1->1->-1 [10] 0/-1/-1->1->-1 [11] 0/-1/-1->1->-1 [12] -1/-1/-1->1->0 [13] -1/-1/-1->1->0 [14] -1/-1/-1->1->0 [15] -1/-1/-1->1->0 [16] -1/-1/-1->1->0 [17] -1/-1/-1->1->0 [18] 0/-1/-1->1->-1 [19] 0/-1/-1->1->-1 [20] 0/-1/-1->1->-1 [21] 0/-1/-1->1->-1 [22] 0/-1/-1->1->-1 [23] 0/-1/-1->1->-1
+ip-26-0-163-58:934721:935254 [1] NCCL INFO Channel 01/24 :    0   1
+ip-26-0-163-58:934725:935255 [5] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-163-58:934721:935254 [1] NCCL INFO Channel 02/24 :    0   1
+ip-26-0-163-58:934721:935254 [1] NCCL INFO Channel 03/24 :    0   1
+ip-26-0-163-58:934721:935254 [1] NCCL INFO Channel 04/24 :    0   1
+ip-26-0-163-58:934721:935254 [1] NCCL INFO Channel 05/24 :    0   1
+ip-26-0-163-58:934721:935254 [1] NCCL INFO Channel 06/24 :    0   1
+ip-26-0-163-58:934721:935254 [1] NCCL INFO Channel 07/24 :    0   1
+ip-26-0-163-58:934721:935254 [1] NCCL INFO Channel 08/24 :    0   1
+ip-26-0-163-58:934721:935254 [1] NCCL INFO Channel 09/24 :    0   1
+ip-26-0-163-58:934721:935254 [1] NCCL INFO Channel 10/24 :    0   1
+ip-26-0-163-58:934721:935254 [1] NCCL INFO Channel 11/24 :    0   1
+ip-26-0-163-58:934721:935254 [1] NCCL INFO Channel 12/24 :    0   1
+ip-26-0-163-58:934721:935254 [1] NCCL INFO Channel 13/24 :    0   1
+ip-26-0-163-58:934721:935254 [1] NCCL INFO Channel 14/24 :    0   1
+ip-26-0-163-58:934721:935254 [1] NCCL INFO Channel 15/24 :    0   1
+ip-26-0-163-58:934721:935254 [1] NCCL INFO Channel 16/24 :    0   1
+ip-26-0-163-58:934721:935254 [1] NCCL INFO Channel 17/24 :    0   1
+ip-26-0-163-58:934721:935254 [1] NCCL INFO Channel 18/24 :    0   1
+ip-26-0-163-58:934721:935254 [1] NCCL INFO Channel 19/24 :    0   1
+ip-26-0-163-58:934721:935254 [1] NCCL INFO Channel 20/24 :    0   1
+ip-26-0-163-58:934721:935254 [1] NCCL INFO Channel 21/24 :    0   1
+ip-26-0-163-58:934721:935254 [1] NCCL INFO Channel 22/24 :    0   1
+ip-26-0-163-58:934721:935254 [1] NCCL INFO Channel 23/24 :    0   1
+ip-26-0-163-58:934721:935254 [1] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] -1/-1/-1->0->1 [7] -1/-1/-1->0->1 [8] -1/-1/-1->0->1 [9] -1/-1/-1->0->1 [10] -1/-1/-1->0->1 [11] -1/-1/-1->0->1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] -1/-1/-1->0->1 [19] -1/-1/-1->0->1 [20] -1/-1/-1->0->1 [21] -1/-1/-1->0->1 [22] -1/-1/-1->0->1 [23] -1/-1/-1->0->1
+ip-26-0-163-58:934721:935254 [1] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-170-31:3197180:3197820 [0] NCCL INFO Channel 06/0 : 0[0] -> 1[4] via P2P/IPC
+ip-26-0-162-14:2525077:2525620 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-162-14:2525081:2525622 [6] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] -1/-1/-1->1->0 [3] -1/-1/-1->1->0 [4] -1/-1/-1->1->0 [5] -1/-1/-1->1->0 [6] 0/-1/-1->1->-1 [7] 0/-1/-1->1->-1 [8] 0/-1/-1->1->-1 [9] 0/-1/-1->1->-1 [10] 0/-1/-1->1->-1 [11] 0/-1/-1->1->-1 [12] -1/-1/-1->1->0 [13] -1/-1/-1->1->0 [14] -1/-1/-1->1->0 [15] -1/-1/-1->1->0 [16] -1/-1/-1->1->0 [17] -1/-1/-1->1->0 [18] 0/-1/-1->1->-1 [19] 0/-1/-1->1->-1 [20] 0/-1/-1->1->-1 [21] 0/-1/-1->1->-1 [22] 0/-1/-1->1->-1 [23] 0/-1/-1->1->-1
+ip-26-0-162-14:2525077:2525620 [2] NCCL INFO Channel 00/24 :    0   1
+ip-26-0-162-14:2525081:2525622 [6] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-162-14:2525077:2525620 [2] NCCL INFO Channel 01/24 :    0   1
+ip-26-0-162-14:2525077:2525620 [2] NCCL INFO Channel 02/24 :    0   1
+ip-26-0-162-14:2525077:2525620 [2] NCCL INFO Channel 03/24 :    0   1
+ip-26-0-162-14:2525077:2525620 [2] NCCL INFO Channel 04/24 :    0   1
+ip-26-0-162-14:2525077:2525620 [2] NCCL INFO Channel 05/24 :    0   1
+ip-26-0-162-14:2525077:2525620 [2] NCCL INFO Channel 06/24 :    0   1
+ip-26-0-162-14:2525077:2525620 [2] NCCL INFO Channel 07/24 :    0   1
+ip-26-0-162-14:2525077:2525620 [2] NCCL INFO Channel 08/24 :    0   1
+ip-26-0-162-14:2525077:2525620 [2] NCCL INFO Channel 09/24 :    0   1
+ip-26-0-162-14:2525077:2525620 [2] NCCL INFO Channel 10/24 :    0   1
+ip-26-0-162-14:2525077:2525620 [2] NCCL INFO Channel 11/24 :    0   1
+ip-26-0-162-14:2525077:2525620 [2] NCCL INFO Channel 12/24 :    0   1
+ip-26-0-162-14:2525077:2525620 [2] NCCL INFO Channel 13/24 :    0   1
+ip-26-0-162-14:2525077:2525620 [2] NCCL INFO Channel 14/24 :    0   1
+ip-26-0-162-14:2525077:2525620 [2] NCCL INFO Channel 15/24 :    0   1
+ip-26-0-162-14:2525077:2525620 [2] NCCL INFO Channel 16/24 :    0   1
+ip-26-0-162-14:2525077:2525620 [2] NCCL INFO Channel 17/24 :    0   1
+ip-26-0-162-14:2525077:2525620 [2] NCCL INFO Channel 18/24 :    0   1
+ip-26-0-162-14:2525077:2525620 [2] NCCL INFO Channel 19/24 :    0   1
+ip-26-0-162-14:2525077:2525620 [2] NCCL INFO Channel 20/24 :    0   1
+ip-26-0-162-14:2525077:2525620 [2] NCCL INFO Channel 21/24 :    0   1
+ip-26-0-162-14:2525077:2525620 [2] NCCL INFO Channel 22/24 :    0   1
+ip-26-0-162-14:2525077:2525620 [2] NCCL INFO Channel 23/24 :    0   1
+ip-26-0-162-14:2525077:2525620 [2] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] -1/-1/-1->0->1 [7] -1/-1/-1->0->1 [8] -1/-1/-1->0->1 [9] -1/-1/-1->0->1 [10] -1/-1/-1->0->1 [11] -1/-1/-1->0->1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] -1/-1/-1->0->1 [19] -1/-1/-1->0->1 [20] -1/-1/-1->0->1 [21] -1/-1/-1->0->1 [22] -1/-1/-1->0->1 [23] -1/-1/-1->0->1
+ip-26-0-162-14:2525077:2525620 [2] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-170-31:3197184:3197821 [4] NCCL INFO Channel 09/0 : 1[4] -> 0[0] via P2P/IPC
+ip-26-0-170-31:3197180:3197820 [0] NCCL INFO Channel 07/0 : 0[0] -> 1[4] via P2P/IPC
+ip-26-0-170-31:3197184:3197821 [4] NCCL INFO Channel 10/0 : 1[4] -> 0[0] via P2P/IPC
+ip-26-0-170-31:3197185:3197830 [5] NCCL INFO Channel 00/0 : 1[5] -> 0[1] via P2P/IPC
+ip-26-0-170-31:3197180:3197820 [0] NCCL INFO Channel 08/0 : 0[0] -> 1[4] via P2P/IPC
+ip-26-0-170-31:3197184:3197821 [4] NCCL INFO Channel 11/0 : 1[4] -> 0[0] via P2P/IPC
+ip-26-0-170-31:3197185:3197830 [5] NCCL INFO Channel 01/0 : 1[5] -> 0[1] via P2P/IPC
+ip-26-0-170-31:3197180:3197820 [0] NCCL INFO Channel 09/0 : 0[0] -> 1[4] via P2P/IPC
+ip-26-0-162-14:2525079:2525613 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-162-14:2525075:2525612 [0] NCCL INFO Channel 00/24 :    0   1
+ip-26-0-162-14:2525075:2525612 [0] NCCL INFO Channel 01/24 :    0   1
+ip-26-0-162-14:2525075:2525612 [0] NCCL INFO Channel 02/24 :    0   1
+ip-26-0-162-14:2525075:2525612 [0] NCCL INFO Channel 03/24 :    0   1
+ip-26-0-162-14:2525075:2525612 [0] NCCL INFO Channel 04/24 :    0   1
+ip-26-0-162-14:2525075:2525612 [0] NCCL INFO Channel 05/24 :    0   1
+ip-26-0-162-14:2525075:2525612 [0] NCCL INFO Channel 06/24 :    0   1
+ip-26-0-162-14:2525075:2525612 [0] NCCL INFO Channel 07/24 :    0   1
+ip-26-0-162-14:2525075:2525612 [0] NCCL INFO Channel 08/24 :    0   1
+ip-26-0-162-14:2525075:2525612 [0] NCCL INFO Channel 09/24 :    0   1
+ip-26-0-162-14:2525075:2525612 [0] NCCL INFO Channel 10/24 :    0   1
+ip-26-0-162-14:2525075:2525612 [0] NCCL INFO Channel 11/24 :    0   1
+ip-26-0-162-14:2525075:2525612 [0] NCCL INFO Channel 12/24 :    0   1
+ip-26-0-162-14:2525075:2525612 [0] NCCL INFO Channel 13/24 :    0   1
+ip-26-0-162-14:2525075:2525612 [0] NCCL INFO Channel 14/24 :    0   1
+ip-26-0-162-14:2525075:2525612 [0] NCCL INFO Channel 15/24 :    0   1
+ip-26-0-162-14:2525079:2525613 [4] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] -1/-1/-1->1->0 [3] -1/-1/-1->1->0 [4] -1/-1/-1->1->0 [5] -1/-1/-1->1->0 [6] 0/-1/-1->1->-1 [7] 0/-1/-1->1->-1 [8] 0/-1/-1->1->-1 [9] 0/-1/-1->1->-1 [10] 0/-1/-1->1->-1 [11] 0/-1/-1->1->-1 [12] -1/-1/-1->1->0 [13] -1/-1/-1->1->0 [14] -1/-1/-1->1->0 [15] -1/-1/-1->1->0 [16] -1/-1/-1->1->0 [17] -1/-1/-1->1->0 [18] 0/-1/-1->1->-1 [19] 0/-1/-1->1->-1 [20] 0/-1/-1->1->-1 [21] 0/-1/-1->1->-1 [22] 0/-1/-1->1->-1 [23] 0/-1/-1->1->-1
+ip-26-0-162-14:2525075:2525612 [0] NCCL INFO Channel 16/24 :    0   1
+ip-26-0-162-14:2525075:2525612 [0] NCCL INFO Channel 17/24 :    0   1
+ip-26-0-162-14:2525079:2525613 [4] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-162-14:2525075:2525612 [0] NCCL INFO Channel 18/24 :    0   1
+ip-26-0-162-14:2525075:2525612 [0] NCCL INFO Channel 19/24 :    0   1
+ip-26-0-162-14:2525075:2525612 [0] NCCL INFO Channel 20/24 :    0   1
+ip-26-0-162-14:2525075:2525612 [0] NCCL INFO Channel 21/24 :    0   1
+ip-26-0-162-14:2525075:2525612 [0] NCCL INFO Channel 22/24 :    0   1
+ip-26-0-162-14:2525075:2525612 [0] NCCL INFO Channel 23/24 :    0   1
+ip-26-0-162-14:2525075:2525612 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] -1/-1/-1->0->1 [7] -1/-1/-1->0->1 [8] -1/-1/-1->0->1 [9] -1/-1/-1->0->1 [10] -1/-1/-1->0->1 [11] -1/-1/-1->0->1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] -1/-1/-1->0->1 [19] -1/-1/-1->0->1 [20] -1/-1/-1->0->1 [21] -1/-1/-1->0->1 [22] -1/-1/-1->0->1 [23] -1/-1/-1->0->1
+ip-26-0-162-14:2525075:2525612 [0] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-170-31:3197184:3197821 [4] NCCL INFO Channel 12/0 : 1[4] -> 0[0] via P2P/IPC
+ip-26-0-170-31:3197185:3197830 [5] NCCL INFO Channel 02/0 : 1[5] -> 0[1] via P2P/IPC
+ip-26-0-170-31:3197180:3197820 [0] NCCL INFO Channel 10/0 : 0[0] -> 1[4] via P2P/IPC
+ip-26-0-170-31:3197181:3197829 [1] NCCL INFO Channel 00/0 : 0[1] -> 1[5] via P2P/IPC
+ip-26-0-170-31:3197184:3197821 [4] NCCL INFO Channel 13/0 : 1[4] -> 0[0] via P2P/IPC
+ip-26-0-170-31:3197185:3197830 [5] NCCL INFO Channel 03/0 : 1[5] -> 0[1] via P2P/IPC
+ip-26-0-170-31:3197180:3197820 [0] NCCL INFO Channel 11/0 : 0[0] -> 1[4] via P2P/IPC
+ip-26-0-170-31:3197184:3197821 [4] NCCL INFO Channel 14/0 : 1[4] -> 0[0] via P2P/IPC
+ip-26-0-170-31:3197185:3197830 [5] NCCL INFO Channel 04/0 : 1[5] -> 0[1] via P2P/IPC
+ip-26-0-170-31:3197181:3197829 [1] NCCL INFO Channel 01/0 : 0[1] -> 1[5] via P2P/IPC
+ip-26-0-170-31:3197180:3197820 [0] NCCL INFO Channel 12/0 : 0[0] -> 1[4] via P2P/IPC
+ip-26-0-170-31:3197187:3197827 [7] NCCL INFO Channel 00/0 : 1[7] -> 0[3] via P2P/IPC
+ip-26-0-170-31:3197184:3197821 [4] NCCL INFO Channel 15/0 : 1[4] -> 0[0] via P2P/IPC
+ip-26-0-170-31:3197183:3197826 [3] NCCL INFO Channel 00/0 : 0[3] -> 1[7] via P2P/IPC
+ip-26-0-170-31:3197185:3197830 [5] NCCL INFO Channel 05/0 : 1[5] -> 0[1] via P2P/IPC
+ip-26-0-170-31:3197180:3197820 [0] NCCL INFO Channel 13/0 : 0[0] -> 1[4] via P2P/IPC
+ip-26-0-170-31:3197181:3197829 [1] NCCL INFO Channel 02/0 : 0[1] -> 1[5] via P2P/IPC
+ip-26-0-163-58:934722:935248 [2] NCCL INFO Channel 00/0 : 0[2] -> 1[6] via P2P/IPC
+ip-26-0-170-31:3197184:3197821 [4] NCCL INFO Channel 16/0 : 1[4] -> 0[0] via P2P/IPC
+ip-26-0-170-31:3197183:3197826 [3] NCCL INFO Channel 01/0 : 0[3] -> 1[7] via P2P/IPC
+ip-26-0-170-31:3197185:3197830 [5] NCCL INFO Channel 06/0 : 1[5] -> 0[1] via P2P/IPC
+ip-26-0-170-31:3197181:3197829 [1] NCCL INFO Channel 03/0 : 0[1] -> 1[5] via P2P/IPC
+ip-26-0-170-31:3197182:3197823 [2] NCCL INFO Channel 00/0 : 0[2] -> 1[6] via P2P/IPC
+ip-26-0-170-31:3197180:3197820 [0] NCCL INFO Channel 14/0 : 0[0] -> 1[4] via P2P/IPC
+ip-26-0-163-58:934722:935248 [2] NCCL INFO Channel 01/0 : 0[2] -> 1[6] via P2P/IPC
+ip-26-0-163-58:934724:935246 [4] NCCL INFO Channel 00/0 : 1[4] -> 0[0] via P2P/IPC
+ip-26-0-170-31:3197187:3197827 [7] NCCL INFO Channel 01/0 : 1[7] -> 0[3] via P2P/IPC
+ip-26-0-170-31:3197180:3197820 [0] NCCL INFO Channel 15/0 : 0[0] -> 1[4] via P2P/IPC
+ip-26-0-163-58:934722:935248 [2] NCCL INFO Channel 02/0 : 0[2] -> 1[6] via P2P/IPC
+ip-26-0-170-31:3197186:3197824 [6] NCCL INFO Channel 00/0 : 1[6] -> 0[2] via P2P/IPC
+ip-26-0-170-31:3197184:3197821 [4] NCCL INFO Channel 17/0 : 1[4] -> 0[0] via P2P/IPC
+ip-26-0-170-31:3197183:3197826 [3] NCCL INFO Channel 02/0 : 0[3] -> 1[7] via P2P/IPC
+ip-26-0-163-58:934724:935246 [4] NCCL INFO Channel 01/0 : 1[4] -> 0[0] via P2P/IPC
+ip-26-0-163-58:934722:935248 [2] NCCL INFO Channel 03/0 : 0[2] -> 1[6] via P2P/IPC
+ip-26-0-163-58:934724:935246 [4] NCCL INFO Channel 02/0 : 1[4] -> 0[0] via P2P/IPC
+ip-26-0-163-58:934722:935248 [2] NCCL INFO Channel 04/0 : 0[2] -> 1[6] via P2P/IPC
+ip-26-0-170-31:3197185:3197830 [5] NCCL INFO Channel 07/0 : 1[5] -> 0[1] via P2P/IPC
+ip-26-0-170-31:3197182:3197823 [2] NCCL INFO Channel 01/0 : 0[2] -> 1[6] via P2P/IPC
+ip-26-0-163-58:934720:935245 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[4] via P2P/IPC
+ip-26-0-170-31:3197181:3197829 [1] NCCL INFO Channel 04/0 : 0[1] -> 1[5] via P2P/IPC
+ip-26-0-163-58:934724:935246 [4] NCCL INFO Channel 03/0 : 1[4] -> 0[0] via P2P/IPC
+ip-26-0-170-31:3197187:3197827 [7] NCCL INFO Channel 02/0 : 1[7] -> 0[3] via P2P/IPC
+ip-26-0-163-58:934723:935251 [3] NCCL INFO Channel 00/0 : 0[3] -> 1[7] via P2P/IPC
+ip-26-0-163-58:934722:935248 [2] NCCL INFO Channel 05/0 : 0[2] -> 1[6] via P2P/IPC
+ip-26-0-163-58:934720:935245 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[4] via P2P/IPC
+ip-26-0-170-31:3197185:3197830 [5] NCCL INFO Channel 08/0 : 1[5] -> 0[1] via P2P/IPC
+ip-26-0-170-31:3197182:3197823 [2] NCCL INFO Channel 02/0 : 0[2] -> 1[6] via P2P/IPC
+ip-26-0-170-31:3197181:3197829 [1] NCCL INFO Channel 05/0 : 0[1] -> 1[5] via P2P/IPC
+ip-26-0-162-14:2525080:2525616 [5] NCCL INFO Channel 00/0 : 1[5] -> 0[1] via P2P/IPC
+ip-26-0-162-14:2525082:2525621 [7] NCCL INFO Channel 00/0 : 1[7] -> 0[3] via P2P/IPC
+ip-26-0-163-58:934724:935246 [4] NCCL INFO Channel 04/0 : 1[4] -> 0[0] via P2P/IPC
+ip-26-0-163-58:934723:935251 [3] NCCL INFO Channel 01/0 : 0[3] -> 1[7] via P2P/IPC
+ip-26-0-163-58:934722:935248 [2] NCCL INFO Channel 06/0 : 0[2] -> 1[6] via P2P/IPC
+ip-26-0-163-58:934720:935245 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[4] via P2P/IPC
+ip-26-0-170-31:3197186:3197824 [6] NCCL INFO Channel 01/0 : 1[6] -> 0[2] via P2P/IPC
+ip-26-0-170-31:3197180:3197820 [0] NCCL INFO Channel 16/0 : 0[0] -> 1[4] via P2P/IPC
+ip-26-0-170-31:3197187:3197827 [7] NCCL INFO Channel 03/0 : 1[7] -> 0[3] via P2P/IPC
+ip-26-0-170-31:3197182:3197823 [2] NCCL INFO Channel 03/0 : 0[2] -> 1[6] via P2P/IPC
+ip-26-0-170-31:3197185:3197830 [5] NCCL INFO Channel 09/0 : 1[5] -> 0[1] via P2P/IPC
+ip-26-0-170-31:3197184:3197821 [4] NCCL INFO Channel 18/0 : 1[4] -> 0[0] via P2P/IPC
+ip-26-0-170-31:3197183:3197826 [3] NCCL INFO Channel 03/0 : 0[3] -> 1[7] via P2P/IPC
+ip-26-0-170-31:3197181:3197829 [1] NCCL INFO Channel 06/0 : 0[1] -> 1[5] via P2P/IPC
+ip-26-0-162-14:2525080:2525616 [5] NCCL INFO Channel 01/0 : 1[5] -> 0[1] via P2P/IPC
+ip-26-0-162-14:2525082:2525621 [7] NCCL INFO Channel 01/0 : 1[7] -> 0[3] via P2P/IPC
+ip-26-0-163-58:934724:935246 [4] NCCL INFO Channel 05/0 : 1[4] -> 0[0] via P2P/IPC
+ip-26-0-163-58:934723:935251 [3] NCCL INFO Channel 02/0 : 0[3] -> 1[7] via P2P/IPC
+ip-26-0-163-58:934722:935248 [2] NCCL INFO Channel 07/0 : 0[2] -> 1[6] via P2P/IPC
+ip-26-0-163-58:934720:935245 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[4] via P2P/IPC
+ip-26-0-170-31:3197182:3197823 [2] NCCL INFO Channel 04/0 : 0[2] -> 1[6] via P2P/IPC
+ip-26-0-162-14:2525080:2525616 [5] NCCL INFO Channel 02/0 : 1[5] -> 0[1] via P2P/IPC
+ip-26-0-162-14:2525082:2525621 [7] NCCL INFO Channel 02/0 : 1[7] -> 0[3] via P2P/IPC
+ip-26-0-170-31:3197184:3197821 [4] NCCL INFO Channel 19/0 : 1[4] -> 0[0] via P2P/IPC
+ip-26-0-163-58:934724:935246 [4] NCCL INFO Channel 06/0 : 1[4] -> 0[0] via P2P/IPC
+ip-26-0-163-58:934723:935251 [3] NCCL INFO Channel 03/0 : 0[3] -> 1[7] via P2P/IPC
+ip-26-0-163-58:934722:935248 [2] NCCL INFO Channel 08/0 : 0[2] -> 1[6] via P2P/IPC
+ip-26-0-163-58:934720:935245 [0] NCCL INFO Channel 04/0 : 0[0] -> 1[4] via P2P/IPC
+ip-26-0-170-31:3197186:3197824 [6] NCCL INFO Channel 02/0 : 1[6] -> 0[2] via P2P/IPC
+ip-26-0-162-14:2525081:2525622 [6] NCCL INFO Channel 00/0 : 1[6] -> 0[2] via P2P/IPC
+ip-26-0-170-31:3197184:3197821 [4] NCCL INFO Channel 20/0 : 1[4] -> 0[0] via P2P/IPC
+ip-26-0-170-31:3197183:3197826 [3] NCCL INFO Channel 04/0 : 0[3] -> 1[7] via P2P/IPC
+ip-26-0-162-14:2525080:2525616 [5] NCCL INFO Channel 03/0 : 1[5] -> 0[1] via P2P/IPC
+ip-26-0-162-14:2525082:2525621 [7] NCCL INFO Channel 03/0 : 1[7] -> 0[3] via P2P/IPC
+ip-26-0-170-31:3197186:3197824 [6] NCCL INFO Channel 03/0 : 1[6] -> 0[2] via P2P/IPC
+ip-26-0-170-31:3197180:3197820 [0] NCCL INFO Channel 17/0 : 0[0] -> 1[4] via P2P/IPC
+ip-26-0-163-58:934724:935246 [4] NCCL INFO Channel 07/0 : 1[4] -> 0[0] via P2P/IPC
+ip-26-0-163-58:934723:935251 [3] NCCL INFO Channel 04/0 : 0[3] -> 1[7] via P2P/IPC
+ip-26-0-170-31:3197184:3197821 [4] NCCL INFO Channel 21/0 : 1[4] -> 0[0] via P2P/IPC
+ip-26-0-170-31:3197183:3197826 [3] NCCL INFO Channel 05/0 : 0[3] -> 1[7] via P2P/IPC
+ip-26-0-163-58:934722:935248 [2] NCCL INFO Channel 09/0 : 0[2] -> 1[6] via P2P/IPC
+ip-26-0-162-14:2525081:2525622 [6] NCCL INFO Channel 01/0 : 1[6] -> 0[2] via P2P/IPC
+ip-26-0-162-14:2525076:2525615 [1] NCCL INFO Channel 00/0 : 0[1] -> 1[5] via P2P/IPC
+ip-26-0-162-14:2525080:2525616 [5] NCCL INFO Channel 04/0 : 1[5] -> 0[1] via P2P/IPC
+ip-26-0-162-14:2525082:2525621 [7] NCCL INFO Channel 04/0 : 1[7] -> 0[3] via P2P/IPC
+ip-26-0-163-58:934724:935246 [4] NCCL INFO Channel 08/0 : 1[4] -> 0[0] via P2P/IPC
+ip-26-0-163-58:934723:935251 [3] NCCL INFO Channel 05/0 : 0[3] -> 1[7] via P2P/IPC
+ip-26-0-170-31:3197186:3197824 [6] NCCL INFO Channel 04/0 : 1[6] -> 0[2] via P2P/IPC
+ip-26-0-170-31:3197180:3197820 [0] NCCL INFO Channel 18/0 : 0[0] -> 1[4] via P2P/IPC
+ip-26-0-170-31:3197184:3197821 [4] NCCL INFO Channel 22/0 : 1[4] -> 0[0] via P2P/IPC
+ip-26-0-170-31:3197183:3197826 [3] NCCL INFO Channel 06/0 : 0[3] -> 1[7] via P2P/IPC
+ip-26-0-162-14:2525078:2525618 [3] NCCL INFO Channel 00/0 : 0[3] -> 1[7] via P2P/IPC
+ip-26-0-162-14:2525081:2525622 [6] NCCL INFO Channel 02/0 : 1[6] -> 0[2] via P2P/IPC
+ip-26-0-162-14:2525076:2525615 [1] NCCL INFO Channel 01/0 : 0[1] -> 1[5] via P2P/IPC
+ip-26-0-162-14:2525080:2525616 [5] NCCL INFO Channel 05/0 : 1[5] -> 0[1] via P2P/IPC
+ip-26-0-162-14:2525082:2525621 [7] NCCL INFO Channel 05/0 : 1[7] -> 0[3] via P2P/IPC
+ip-26-0-163-58:934720:935245 [0] NCCL INFO Channel 05/0 : 0[0] -> 1[4] via P2P/IPC
+ip-26-0-170-31:3197187:3197827 [7] NCCL INFO Channel 04/0 : 1[7] -> 0[3] via P2P/IPC
+ip-26-0-170-31:3197181:3197829 [1] NCCL INFO Channel 07/0 : 0[1] -> 1[5] via P2P/IPC
+ip-26-0-170-31:3197185:3197830 [5] NCCL INFO Channel 10/0 : 1[5] -> 0[1] via P2P/IPC
+ip-26-0-170-31:3197182:3197823 [2] NCCL INFO Channel 05/0 : 0[2] -> 1[6] via P2P/IPC
+ip-26-0-163-58:934727:935252 [7] NCCL INFO Channel 00/0 : 1[7] -> 0[3] via P2P/IPC
+ip-26-0-170-31:3197186:3197824 [6] NCCL INFO Channel 05/0 : 1[6] -> 0[2] via P2P/IPC
+ip-26-0-163-58:934720:935245 [0] NCCL INFO Channel 06/0 : 0[0] -> 1[4] via P2P/IPC
+ip-26-0-170-31:3197184:3197821 [4] NCCL INFO Channel 23/0 : 1[4] -> 0[0] via P2P/IPC
+ip-26-0-162-14:2525078:2525618 [3] NCCL INFO Channel 01/0 : 0[3] -> 1[7] via P2P/IPC
+ip-26-0-162-14:2525079:2525613 [4] NCCL INFO Channel 00/0 : 1[4] -> 0[0] via P2P/IPC
+ip-26-0-162-14:2525081:2525622 [6] NCCL INFO Channel 03/0 : 1[6] -> 0[2] via P2P/IPC
+ip-26-0-163-58:934725:935255 [5] NCCL INFO Channel 00/0 : 1[5] -> 0[1] via P2P/IPC
+ip-26-0-163-58:934727:935252 [7] NCCL INFO Channel 01/0 : 1[7] -> 0[3] via P2P/IPC
+ip-26-0-170-31:3197180:3197820 [0] NCCL INFO Channel 19/0 : 0[0] -> 1[4] via P2P/IPC
+ip-26-0-162-14:2525080:2525616 [5] NCCL INFO Channel 06/0 : 1[5] -> 0[1] via P2P/IPC
+ip-26-0-162-14:2525076:2525615 [1] NCCL INFO Channel 02/0 : 0[1] -> 1[5] via P2P/IPC
+ip-26-0-162-14:2525082:2525621 [7] NCCL INFO Channel 06/0 : 1[7] -> 0[3] via P2P/IPC
+ip-26-0-170-31:3197187:3197827 [7] NCCL INFO Channel 05/0 : 1[7] -> 0[3] via P2P/IPC
+ip-26-0-170-31:3197181:3197829 [1] NCCL INFO Channel 08/0 : 0[1] -> 1[5] via P2P/IPC
+ip-26-0-170-31:3197180:3197820 [0] NCCL INFO Channel 20/0 : 0[0] -> 1[4] via P2P/IPC
+ip-26-0-163-58:934720:935245 [0] NCCL INFO Channel 07/0 : 0[0] -> 1[4] via P2P/IPC
+ip-26-0-170-31:3197183:3197826 [3] NCCL INFO Channel 07/0 : 0[3] -> 1[7] via P2P/IPC
+ip-26-0-163-58:934722:935248 [2] NCCL INFO Channel 10/0 : 0[2] -> 1[6] via P2P/IPC
+ip-26-0-163-58:934721:935254 [1] NCCL INFO Channel 00/0 : 0[1] -> 1[5] via P2P/IPC
+ip-26-0-170-31:3197187:3197827 [7] NCCL INFO Channel 06/0 : 1[7] -> 0[3] via P2P/IPC
+ip-26-0-162-14:2525078:2525618 [3] NCCL INFO Channel 02/0 : 0[3] -> 1[7] via P2P/IPC
+ip-26-0-170-31:3197181:3197829 [1] NCCL INFO Channel 09/0 : 0[1] -> 1[5] via P2P/IPC
+ip-26-0-162-14:2525079:2525613 [4] NCCL INFO Channel 01/0 : 1[4] -> 0[0] via P2P/IPC
+ip-26-0-162-14:2525081:2525622 [6] NCCL INFO Channel 04/0 : 1[6] -> 0[2] via P2P/IPC
+ip-26-0-170-31:3197180:3197820 [0] NCCL INFO Channel 21/0 : 0[0] -> 1[4] via P2P/IPC
+ip-26-0-162-14:2525080:2525616 [5] NCCL INFO Channel 07/0 : 1[5] -> 0[1] via P2P/IPC
+ip-26-0-163-58:934724:935246 [4] NCCL INFO Channel 09/0 : 1[4] -> 0[0] via P2P/IPC
+ip-26-0-163-58:934725:935255 [5] NCCL INFO Channel 01/0 : 1[5] -> 0[1] via P2P/IPC
+ip-26-0-163-58:934727:935252 [7] NCCL INFO Channel 02/0 : 1[7] -> 0[3] via P2P/IPC
+ip-26-0-163-58:934723:935251 [3] NCCL INFO Channel 06/0 : 0[3] -> 1[7] via P2P/IPC
+ip-26-0-170-31:3197183:3197826 [3] NCCL INFO Channel 08/0 : 0[3] -> 1[7] via P2P/IPC
+ip-26-0-170-31:3197187:3197827 [7] NCCL INFO Channel 07/0 : 1[7] -> 0[3] via P2P/IPC
+ip-26-0-162-14:2525076:2525615 [1] NCCL INFO Channel 03/0 : 0[1] -> 1[5] via P2P/IPC
+ip-26-0-162-14:2525082:2525621 [7] NCCL INFO Channel 07/0 : 1[7] -> 0[3] via P2P/IPC
+ip-26-0-170-31:3197181:3197829 [1] NCCL INFO Channel 10/0 : 0[1] -> 1[5] via P2P/IPC
+ip-26-0-170-31:3197180:3197820 [0] NCCL INFO Channel 22/0 : 0[0] -> 1[4] via P2P/IPC
+ip-26-0-163-58:934720:935245 [0] NCCL INFO Channel 08/0 : 0[0] -> 1[4] via P2P/IPC
+ip-26-0-163-58:934726:935249 [6] NCCL INFO Channel 00/0 : 1[6] -> 0[2] via P2P/IPC
+ip-26-0-163-58:934722:935248 [2] NCCL INFO Channel 11/0 : 0[2] -> 1[6] via P2P/IPC
+ip-26-0-163-58:934721:935254 [1] NCCL INFO Channel 01/0 : 0[1] -> 1[5] via P2P/IPC
+ip-26-0-170-31:3197187:3197827 [7] NCCL INFO Channel 08/0 : 1[7] -> 0[3] via P2P/IPC
+ip-26-0-170-31:3197183:3197826 [3] NCCL INFO Channel 09/0 : 0[3] -> 1[7] via P2P/IPC
+ip-26-0-162-14:2525078:2525618 [3] NCCL INFO Channel 03/0 : 0[3] -> 1[7] via P2P/IPC
+ip-26-0-162-14:2525079:2525613 [4] NCCL INFO Channel 02/0 : 1[4] -> 0[0] via P2P/IPC
+ip-26-0-162-14:2525077:2525620 [2] NCCL INFO Channel 00/0 : 0[2] -> 1[6] via P2P/IPC
+ip-26-0-162-14:2525081:2525622 [6] NCCL INFO Channel 05/0 : 1[6] -> 0[2] via P2P/IPC
+ip-26-0-162-14:2525080:2525616 [5] NCCL INFO Channel 08/0 : 1[5] -> 0[1] via P2P/IPC
+ip-26-0-162-14:2525076:2525615 [1] NCCL INFO Channel 04/0 : 0[1] -> 1[5] via P2P/IPC
+ip-26-0-162-14:2525082:2525621 [7] NCCL INFO Channel 08/0 : 1[7] -> 0[3] via P2P/IPC
+ip-26-0-163-58:934724:935246 [4] NCCL INFO Channel 10/0 : 1[4] -> 0[0] via P2P/IPC
+ip-26-0-163-58:934727:935252 [7] NCCL INFO Channel 03/0 : 1[7] -> 0[3] via P2P/IPC
+ip-26-0-163-58:934725:935255 [5] NCCL INFO Channel 02/0 : 1[5] -> 0[1] via P2P/IPC
+ip-26-0-163-58:934723:935251 [3] NCCL INFO Channel 07/0 : 0[3] -> 1[7] via P2P/IPC
+ip-26-0-170-31:3197186:3197824 [6] NCCL INFO Channel 06/0 : 1[6] -> 0[2] via P2P/IPC
+ip-26-0-170-31:3197181:3197829 [1] NCCL INFO Channel 11/0 : 0[1] -> 1[5] via P2P/IPC
+ip-26-0-170-31:3197180:3197820 [0] NCCL INFO Channel 23/0 : 0[0] -> 1[4] via P2P/IPC
+ip-26-0-170-31:3197185:3197830 [5] NCCL INFO Channel 11/0 : 1[5] -> 0[1] via P2P/IPC
+ip-26-0-170-31:3197182:3197823 [2] NCCL INFO Channel 06/0 : 0[2] -> 1[6] via P2P/IPC
+ip-26-0-170-31:3197187:3197827 [7] NCCL INFO Channel 09/0 : 1[7] -> 0[3] via P2P/IPC
+ip-26-0-170-31:3197183:3197826 [3] NCCL INFO Channel 10/0 : 0[3] -> 1[7] via P2P/IPC
+ip-26-0-163-58:934720:935245 [0] NCCL INFO Channel 09/0 : 0[0] -> 1[4] via P2P/IPC
+ip-26-0-163-58:934726:935249 [6] NCCL INFO Channel 01/0 : 1[6] -> 0[2] via P2P/IPC
+ip-26-0-163-58:934722:935248 [2] NCCL INFO Channel 12/0 : 0[2] -> 1[6] via P2P/IPC
+ip-26-0-163-58:934721:935254 [1] NCCL INFO Channel 02/0 : 0[1] -> 1[5] via P2P/IPC
+ip-26-0-162-14:2525078:2525618 [3] NCCL INFO Channel 04/0 : 0[3] -> 1[7] via P2P/IPC
+ip-26-0-162-14:2525079:2525613 [4] NCCL INFO Channel 03/0 : 1[4] -> 0[0] via P2P/IPC
+ip-26-0-162-14:2525081:2525622 [6] NCCL INFO Channel 06/0 : 1[6] -> 0[2] via P2P/IPC
+ip-26-0-162-14:2525077:2525620 [2] NCCL INFO Channel 01/0 : 0[2] -> 1[6] via P2P/IPC
+ip-26-0-162-14:2525080:2525616 [5] NCCL INFO Channel 09/0 : 1[5] -> 0[1] via P2P/IPC
+ip-26-0-162-14:2525076:2525615 [1] NCCL INFO Channel 05/0 : 0[1] -> 1[5] via P2P/IPC
+ip-26-0-162-14:2525082:2525621 [7] NCCL INFO Channel 09/0 : 1[7] -> 0[3] via P2P/IPC
+ip-26-0-163-58:934724:935246 [4] NCCL INFO Channel 11/0 : 1[4] -> 0[0] via P2P/IPC
+ip-26-0-163-58:934725:935255 [5] NCCL INFO Channel 03/0 : 1[5] -> 0[1] via P2P/IPC
+ip-26-0-163-58:934723:935251 [3] NCCL INFO Channel 08/0 : 0[3] -> 1[7] via P2P/IPC
+ip-26-0-163-58:934727:935252 [7] NCCL INFO Channel 04/0 : 1[7] -> 0[3] via P2P/IPC
+ip-26-0-170-31:3197186:3197824 [6] NCCL INFO Channel 07/0 : 1[6] -> 0[2] via P2P/IPC
+ip-26-0-170-31:3197181:3197829 [1] NCCL INFO Channel 12/0 : 0[1] -> 1[5] via P2P/IPC
+ip-26-0-170-31:3197182:3197823 [2] NCCL INFO Channel 07/0 : 0[2] -> 1[6] via P2P/IPC
+ip-26-0-170-31:3197185:3197830 [5] NCCL INFO Channel 12/0 : 1[5] -> 0[1] via P2P/IPC
+ip-26-0-170-31:3197187:3197827 [7] NCCL INFO Channel 10/0 : 1[7] -> 0[3] via P2P/IPC
+ip-26-0-170-31:3197183:3197826 [3] NCCL INFO Channel 11/0 : 0[3] -> 1[7] via P2P/IPC
+ip-26-0-163-58:934720:935245 [0] NCCL INFO Channel 10/0 : 0[0] -> 1[4] via P2P/IPC
+ip-26-0-163-58:934726:935249 [6] NCCL INFO Channel 02/0 : 1[6] -> 0[2] via P2P/IPC
+ip-26-0-163-58:934722:935248 [2] NCCL INFO Channel 13/0 : 0[2] -> 1[6] via P2P/IPC
+ip-26-0-163-58:934721:935254 [1] NCCL INFO Channel 03/0 : 0[1] -> 1[5] via P2P/IPC
+ip-26-0-162-14:2525078:2525618 [3] NCCL INFO Channel 05/0 : 0[3] -> 1[7] via P2P/IPC
+ip-26-0-163-58:934724:935246 [4] NCCL INFO Channel 12/0 : 1[4] -> 0[0] via P2P/IPC
+ip-26-0-162-14:2525082:2525621 [7] NCCL INFO Channel 10/0 : 1[7] -> 0[3] via P2P/IPC
+ip-26-0-162-14:2525079:2525613 [4] NCCL INFO Channel 04/0 : 1[4] -> 0[0] via P2P/IPC
+ip-26-0-163-58:934725:935255 [5] NCCL INFO Channel 04/0 : 1[5] -> 0[1] via P2P/IPC
+ip-26-0-162-14:2525081:2525622 [6] NCCL INFO Channel 07/0 : 1[6] -> 0[2] via P2P/IPC
+ip-26-0-163-58:934723:935251 [3] NCCL INFO Channel 09/0 : 0[3] -> 1[7] via P2P/IPC
+ip-26-0-163-58:934727:935252 [7] NCCL INFO Channel 05/0 : 1[7] -> 0[3] via P2P/IPC
+ip-26-0-162-14:2525077:2525620 [2] NCCL INFO Channel 02/0 : 0[2] -> 1[6] via P2P/IPC
+ip-26-0-163-58:934726:935249 [6] NCCL INFO Channel 03/0 : 1[6] -> 0[2] via P2P/IPC
+ip-26-0-162-14:2525080:2525616 [5] NCCL INFO Channel 10/0 : 1[5] -> 0[1] via P2P/IPC
+ip-26-0-162-14:2525076:2525615 [1] NCCL INFO Channel 06/0 : 0[1] -> 1[5] via P2P/IPC
+ip-26-0-163-58:934722:935248 [2] NCCL INFO Channel 14/0 : 0[2] -> 1[6] via P2P/IPC
+ip-26-0-170-31:3197186:3197824 [6] NCCL INFO Channel 08/0 : 1[6] -> 0[2] via P2P/IPC
+ip-26-0-162-14:2525081:2525622 [6] NCCL INFO Channel 08/0 : 1[6] -> 0[2] via P2P/IPC
+ip-26-0-170-31:3197183:3197826 [3] NCCL INFO Channel 12/0 : 0[3] -> 1[7] via P2P/IPC
+ip-26-0-170-31:3197181:3197829 [1] NCCL INFO Channel 13/0 : 0[1] -> 1[5] via P2P/IPC
+ip-26-0-170-31:3197182:3197823 [2] NCCL INFO Channel 08/0 : 0[2] -> 1[6] via P2P/IPC
+ip-26-0-163-58:934724:935246 [4] NCCL INFO Channel 13/0 : 1[4] -> 0[0] via P2P/IPC
+ip-26-0-162-14:2525079:2525613 [4] NCCL INFO Channel 05/0 : 1[4] -> 0[0] via P2P/IPC
+ip-26-0-162-14:2525080:2525616 [5] NCCL INFO Channel 11/0 : 1[5] -> 0[1] via P2P/IPC
+ip-26-0-170-31:3197186:3197824 [6] NCCL INFO Channel 09/0 : 1[6] -> 0[2] via P2P/IPC
+ip-26-0-170-31:3197183:3197826 [3] NCCL INFO Channel 13/0 : 0[3] -> 1[7] via P2P/IPC
+ip-26-0-163-58:934721:935254 [1] NCCL INFO Channel 04/0 : 0[1] -> 1[5] via P2P/IPC
+ip-26-0-163-58:934720:935245 [0] NCCL INFO Channel 11/0 : 0[0] -> 1[4] via P2P/IPC
+ip-26-0-170-31:3197181:3197829 [1] NCCL INFO Channel 14/0 : 0[1] -> 1[5] via P2P/IPC
+ip-26-0-170-31:3197182:3197823 [2] NCCL INFO Channel 09/0 : 0[2] -> 1[6] via P2P/IPC
+ip-26-0-162-14:2525082:2525621 [7] NCCL INFO Channel 11/0 : 1[7] -> 0[3] via P2P/IPC
+ip-26-0-163-58:934727:935252 [7] NCCL INFO Channel 06/0 : 1[7] -> 0[3] via P2P/IPC
+ip-26-0-163-58:934724:935246 [4] NCCL INFO Channel 14/0 : 1[4] -> 0[0] via P2P/IPC
+ip-26-0-170-31:3197187:3197827 [7] NCCL INFO Channel 11/0 : 1[7] -> 0[3] via P2P/IPC
+ip-26-0-162-14:2525077:2525620 [2] NCCL INFO Channel 03/0 : 0[2] -> 1[6] via P2P/IPC
+ip-26-0-163-58:934720:935245 [0] NCCL INFO Channel 12/0 : 0[0] -> 1[4] via P2P/IPC
+ip-26-0-163-58:934725:935255 [5] NCCL INFO Channel 05/0 : 1[5] -> 0[1] via P2P/IPC
+ip-26-0-170-31:3197186:3197824 [6] NCCL INFO Channel 10/0 : 1[6] -> 0[2] via P2P/IPC
+ip-26-0-170-31:3197183:3197826 [3] NCCL INFO Channel 14/0 : 0[3] -> 1[7] via P2P/IPC
+ip-26-0-170-31:3197181:3197829 [1] NCCL INFO Channel 15/0 : 0[1] -> 1[5] via P2P/IPC
+ip-26-0-170-31:3197182:3197823 [2] NCCL INFO Channel 10/0 : 0[2] -> 1[6] via P2P/IPC
+ip-26-0-170-31:3197187:3197827 [7] NCCL INFO Channel 12/0 : 1[7] -> 0[3] via P2P/IPC
+ip-26-0-162-14:2525076:2525615 [1] NCCL INFO Channel 07/0 : 0[1] -> 1[5] via P2P/IPC
+ip-26-0-163-58:934727:935252 [7] NCCL INFO Channel 07/0 : 1[7] -> 0[3] via P2P/IPC
+ip-26-0-163-58:934724:935246 [4] NCCL INFO Channel 15/0 : 1[4] -> 0[0] via P2P/IPC
+ip-26-0-163-58:934721:935254 [1] NCCL INFO Channel 05/0 : 0[1] -> 1[5] via P2P/IPC
+ip-26-0-170-31:3197185:3197830 [5] NCCL INFO Channel 13/0 : 1[5] -> 0[1] via P2P/IPC
+ip-26-0-163-58:934720:935245 [0] NCCL INFO Channel 13/0 : 0[0] -> 1[4] via P2P/IPC
+ip-26-0-163-58:934725:935255 [5] NCCL INFO Channel 06/0 : 1[5] -> 0[1] via P2P/IPC
+ip-26-0-162-14:2525078:2525618 [3] NCCL INFO Channel 06/0 : 0[3] -> 1[7] via P2P/IPC
+ip-26-0-163-58:934727:935252 [7] NCCL INFO Channel 08/0 : 1[7] -> 0[3] via P2P/IPC
+ip-26-0-163-58:934724:935246 [4] NCCL INFO Channel 16/0 : 1[4] -> 0[0] via P2P/IPC
+ip-26-0-162-14:2525079:2525613 [4] NCCL INFO Channel 06/0 : 1[4] -> 0[0] via P2P/IPC
+ip-26-0-162-14:2525081:2525622 [6] NCCL INFO Channel 09/0 : 1[6] -> 0[2] via P2P/IPC
+ip-26-0-163-58:934721:935254 [1] NCCL INFO Channel 06/0 : 0[1] -> 1[5] via P2P/IPC
+ip-26-0-162-14:2525080:2525616 [5] NCCL INFO Channel 12/0 : 1[5] -> 0[1] via P2P/IPC
+ip-26-0-162-14:2525078:2525618 [3] NCCL INFO Channel 07/0 : 0[3] -> 1[7] via P2P/IPC
+ip-26-0-162-14:2525082:2525621 [7] NCCL INFO Channel 12/0 : 1[7] -> 0[3] via P2P/IPC
+ip-26-0-163-58:934720:935245 [0] NCCL INFO Channel 14/0 : 0[0] -> 1[4] via P2P/IPC
+ip-26-0-163-58:934725:935255 [5] NCCL INFO Channel 07/0 : 1[5] -> 0[1] via P2P/IPC
+ip-26-0-163-58:934723:935251 [3] NCCL INFO Channel 10/0 : 0[3] -> 1[7] via P2P/IPC
+ip-26-0-163-58:934722:935248 [2] NCCL INFO Channel 15/0 : 0[2] -> 1[6] via P2P/IPC
+ip-26-0-163-58:934726:935249 [6] NCCL INFO Channel 04/0 : 1[6] -> 0[2] via P2P/IPC
+ip-26-0-170-31:3197183:3197826 [3] NCCL INFO Channel 15/0 : 0[3] -> 1[7] via P2P/IPC
+ip-26-0-170-31:3197186:3197824 [6] NCCL INFO Channel 11/0 : 1[6] -> 0[2] via P2P/IPC
+ip-26-0-163-58:934727:935252 [7] NCCL INFO Channel 09/0 : 1[7] -> 0[3] via P2P/IPC
+ip-26-0-170-31:3197181:3197829 [1] NCCL INFO Channel 16/0 : 0[1] -> 1[5] via P2P/IPC
+ip-26-0-170-31:3197182:3197823 [2] NCCL INFO Channel 11/0 : 0[2] -> 1[6] via P2P/IPC
+ip-26-0-170-31:3197187:3197827 [7] NCCL INFO Channel 13/0 : 1[7] -> 0[3] via P2P/IPC
+ip-26-0-162-14:2525079:2525613 [4] NCCL INFO Channel 07/0 : 1[4] -> 0[0] via P2P/IPC
+ip-26-0-162-14:2525081:2525622 [6] NCCL INFO Channel 10/0 : 1[6] -> 0[2] via P2P/IPC
+ip-26-0-163-58:934724:935246 [4] NCCL INFO Channel 17/0 : 1[4] -> 0[0] via P2P/IPC
+ip-26-0-162-14:2525077:2525620 [2] NCCL INFO Channel 04/0 : 0[2] -> 1[6] via P2P/IPC
+ip-26-0-162-14:2525075:2525612 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[4] via P2P/IPC
+ip-26-0-163-58:934725:935255 [5] NCCL INFO Channel 08/0 : 1[5] -> 0[1] via P2P/IPC
+ip-26-0-170-31:3197185:3197830 [5] NCCL INFO Channel 14/0 : 1[5] -> 0[1] via P2P/IPC
+ip-26-0-163-58:934723:935251 [3] NCCL INFO Channel 11/0 : 0[3] -> 1[7] via P2P/IPC
+ip-26-0-163-58:934720:935245 [0] NCCL INFO Channel 15/0 : 0[0] -> 1[4] via P2P/IPC
+ip-26-0-163-58:934722:935248 [2] NCCL INFO Channel 16/0 : 0[2] -> 1[6] via P2P/IPC
+ip-26-0-163-58:934725:935255 [5] NCCL INFO Channel 09/0 : 1[5] -> 0[1] via P2P/IPC
+ip-26-0-163-58:934723:935251 [3] NCCL INFO Channel 12/0 : 0[3] -> 1[7] via P2P/IPC
+ip-26-0-162-14:2525080:2525616 [5] NCCL INFO Channel 13/0 : 1[5] -> 0[1] via P2P/IPC
+ip-26-0-163-58:934720:935245 [0] NCCL INFO Channel 16/0 : 0[0] -> 1[4] via P2P/IPC
+ip-26-0-163-58:934722:935248 [2] NCCL INFO Channel 17/0 : 0[2] -> 1[6] via P2P/IPC
+ip-26-0-162-14:2525076:2525615 [1] NCCL INFO Channel 08/0 : 0[1] -> 1[5] via P2P/IPC
+ip-26-0-162-14:2525079:2525613 [4] NCCL INFO Channel 08/0 : 1[4] -> 0[0] via P2P/IPC
+ip-26-0-162-14:2525078:2525618 [3] NCCL INFO Channel 08/0 : 0[3] -> 1[7] via P2P/IPC
+ip-26-0-163-58:934725:935255 [5] NCCL INFO Channel 10/0 : 1[5] -> 0[1] via P2P/IPC
+ip-26-0-163-58:934723:935251 [3] NCCL INFO Channel 13/0 : 0[3] -> 1[7] via P2P/IPC
+ip-26-0-162-14:2525080:2525616 [5] NCCL INFO Channel 14/0 : 1[5] -> 0[1] via P2P/IPC
+ip-26-0-163-58:934720:935245 [0] NCCL INFO Channel 17/0 : 0[0] -> 1[4] via P2P/IPC
+ip-26-0-163-58:934722:935248 [2] NCCL INFO Channel 18/0 : 0[2] -> 1[6] via P2P/IPC
+ip-26-0-162-14:2525076:2525615 [1] NCCL INFO Channel 09/0 : 0[1] -> 1[5] via P2P/IPC
+ip-26-0-170-31:3197183:3197826 [3] NCCL INFO Channel 16/0 : 0[3] -> 1[7] via P2P/IPC
+ip-26-0-170-31:3197181:3197829 [1] NCCL INFO Channel 17/0 : 0[1] -> 1[5] via P2P/IPC
+ip-26-0-170-31:3197182:3197823 [2] NCCL INFO Channel 12/0 : 0[2] -> 1[6] via P2P/IPC
+ip-26-0-170-31:3197187:3197827 [7] NCCL INFO Channel 14/0 : 1[7] -> 0[3] via P2P/IPC
+ip-26-0-162-14:2525079:2525613 [4] NCCL INFO Channel 09/0 : 1[4] -> 0[0] via P2P/IPC
+ip-26-0-162-14:2525078:2525618 [3] NCCL INFO Channel 09/0 : 0[3] -> 1[7] via P2P/IPC
+ip-26-0-162-14:2525080:2525616 [5] NCCL INFO Channel 15/0 : 1[5] -> 0[1] via P2P/IPC
+ip-26-0-163-58:934725:935255 [5] NCCL INFO Channel 11/0 : 1[5] -> 0[1] via P2P/IPC
+ip-26-0-163-58:934723:935251 [3] NCCL INFO Channel 14/0 : 0[3] -> 1[7] via P2P/IPC
+ip-26-0-163-58:934720:935245 [0] NCCL INFO Channel 18/0 : 0[0] -> 1[4] via P2P/IPC
+ip-26-0-163-58:934721:935254 [1] NCCL INFO Channel 07/0 : 0[1] -> 1[5] via P2P/IPC
+ip-26-0-163-58:934722:935248 [2] NCCL INFO Channel 19/0 : 0[2] -> 1[6] via P2P/IPC
+ip-26-0-170-31:3197185:3197830 [5] NCCL INFO Channel 15/0 : 1[5] -> 0[1] via P2P/IPC
+ip-26-0-162-14:2525082:2525621 [7] NCCL INFO Channel 13/0 : 1[7] -> 0[3] via P2P/IPC
+ip-26-0-162-14:2525076:2525615 [1] NCCL INFO Channel 10/0 : 0[1] -> 1[5] via P2P/IPC
+ip-26-0-162-14:2525081:2525622 [6] NCCL INFO Channel 11/0 : 1[6] -> 0[2] via P2P/IPC
+ip-26-0-163-58:934726:935249 [6] NCCL INFO Channel 05/0 : 1[6] -> 0[2] via P2P/IPC
+ip-26-0-163-58:934725:935255 [5] NCCL INFO Channel 12/0 : 1[5] -> 0[1] via P2P/IPC
+ip-26-0-163-58:934727:935252 [7] NCCL INFO Channel 10/0 : 1[7] -> 0[3] via P2P/IPC
+ip-26-0-163-58:934720:935245 [0] NCCL INFO Channel 19/0 : 0[0] -> 1[4] via P2P/IPC
+ip-26-0-163-58:934723:935251 [3] NCCL INFO Channel 15/0 : 0[3] -> 1[7] via P2P/IPC
+ip-26-0-163-58:934721:935254 [1] NCCL INFO Channel 08/0 : 0[1] -> 1[5] via P2P/IPC
+ip-26-0-163-58:934724:935246 [4] NCCL INFO Channel 18/0 : 1[4] -> 0[0] via P2P/IPC
+ip-26-0-163-58:934722:935248 [2] NCCL INFO Channel 20/0 : 0[2] -> 1[6] via P2P/IPC
+ip-26-0-162-14:2525079:2525613 [4] NCCL INFO Channel 10/0 : 1[4] -> 0[0] via P2P/IPC
+ip-26-0-162-14:2525078:2525618 [3] NCCL INFO Channel 10/0 : 0[3] -> 1[7] via P2P/IPC
+ip-26-0-162-14:2525080:2525616 [5] NCCL INFO Channel 16/0 : 1[5] -> 0[1] via P2P/IPC
+ip-26-0-162-14:2525077:2525620 [2] NCCL INFO Channel 05/0 : 0[2] -> 1[6] via P2P/IPC
+ip-26-0-162-14:2525075:2525612 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[4] via P2P/IPC
+ip-26-0-162-14:2525082:2525621 [7] NCCL INFO Channel 14/0 : 1[7] -> 0[3] via P2P/IPC
+ip-26-0-162-14:2525076:2525615 [1] NCCL INFO Channel 11/0 : 0[1] -> 1[5] via P2P/IPC
+ip-26-0-170-31:3197181:3197829 [1] NCCL INFO Channel 18/0 : 0[1] -> 1[5] via P2P/IPC
+ip-26-0-170-31:3197183:3197826 [3] NCCL INFO Channel 17/0 : 0[3] -> 1[7] via P2P/IPC
+ip-26-0-163-58:934726:935249 [6] NCCL INFO Channel 06/0 : 1[6] -> 0[2] via P2P/IPC
+ip-26-0-162-14:2525081:2525622 [6] NCCL INFO Channel 12/0 : 1[6] -> 0[2] via P2P/IPC
+ip-26-0-163-58:934725:935255 [5] NCCL INFO Channel 13/0 : 1[5] -> 0[1] via P2P/IPC
+ip-26-0-163-58:934727:935252 [7] NCCL INFO Channel 11/0 : 1[7] -> 0[3] via P2P/IPC
+ip-26-0-163-58:934720:935245 [0] NCCL INFO Channel 20/0 : 0[0] -> 1[4] via P2P/IPC
+ip-26-0-163-58:934723:935251 [3] NCCL INFO Channel 16/0 : 0[3] -> 1[7] via P2P/IPC
+ip-26-0-163-58:934722:935248 [2] NCCL INFO Channel 21/0 : 0[2] -> 1[6] via P2P/IPC
+ip-26-0-163-58:934724:935246 [4] NCCL INFO Channel 19/0 : 1[4] -> 0[0] via P2P/IPC
+ip-26-0-162-14:2525079:2525613 [4] NCCL INFO Channel 11/0 : 1[4] -> 0[0] via P2P/IPC
+ip-26-0-162-14:2525078:2525618 [3] NCCL INFO Channel 11/0 : 0[3] -> 1[7] via P2P/IPC
+ip-26-0-170-31:3197182:3197823 [2] NCCL INFO Channel 13/0 : 0[2] -> 1[6] via P2P/IPC
+ip-26-0-162-14:2525080:2525616 [5] NCCL INFO Channel 17/0 : 1[5] -> 0[1] via P2P/IPC
+ip-26-0-170-31:3197187:3197827 [7] NCCL INFO Channel 15/0 : 1[7] -> 0[3] via P2P/IPC
+ip-26-0-162-14:2525077:2525620 [2] NCCL INFO Channel 06/0 : 0[2] -> 1[6] via P2P/IPC
+ip-26-0-162-14:2525075:2525612 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[4] via P2P/IPC
+ip-26-0-170-31:3197186:3197824 [6] NCCL INFO Channel 12/0 : 1[6] -> 0[2] via P2P/IPC
+ip-26-0-170-31:3197185:3197830 [5] NCCL INFO Channel 16/0 : 1[5] -> 0[1] via P2P/IPC
+ip-26-0-163-58:934721:935254 [1] NCCL INFO Channel 09/0 : 0[1] -> 1[5] via P2P/IPC
+ip-26-0-162-14:2525082:2525621 [7] NCCL INFO Channel 15/0 : 1[7] -> 0[3] via P2P/IPC
+ip-26-0-162-14:2525076:2525615 [1] NCCL INFO Channel 12/0 : 0[1] -> 1[5] via P2P/IPC
+ip-26-0-163-58:934726:935249 [6] NCCL INFO Channel 07/0 : 1[6] -> 0[2] via P2P/IPC
+ip-26-0-162-14:2525081:2525622 [6] NCCL INFO Channel 13/0 : 1[6] -> 0[2] via P2P/IPC
+ip-26-0-163-58:934725:935255 [5] NCCL INFO Channel 14/0 : 1[5] -> 0[1] via P2P/IPC
+ip-26-0-163-58:934727:935252 [7] NCCL INFO Channel 12/0 : 1[7] -> 0[3] via P2P/IPC
+ip-26-0-163-58:934720:935245 [0] NCCL INFO Channel 21/0 : 0[0] -> 1[4] via P2P/IPC
+ip-26-0-163-58:934723:935251 [3] NCCL INFO Channel 17/0 : 0[3] -> 1[7] via P2P/IPC
+ip-26-0-163-58:934722:935248 [2] NCCL INFO Channel 22/0 : 0[2] -> 1[6] via P2P/IPC
+ip-26-0-163-58:934724:935246 [4] NCCL INFO Channel 20/0 : 1[4] -> 0[0] via P2P/IPC
+ip-26-0-162-14:2525079:2525613 [4] NCCL INFO Channel 12/0 : 1[4] -> 0[0] via P2P/IPC
+ip-26-0-162-14:2525078:2525618 [3] NCCL INFO Channel 12/0 : 0[3] -> 1[7] via P2P/IPC
+ip-26-0-162-14:2525080:2525616 [5] NCCL INFO Channel 18/0 : 1[5] -> 0[1] via P2P/IPC
+ip-26-0-162-14:2525075:2525612 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[4] via P2P/IPC
+ip-26-0-162-14:2525077:2525620 [2] NCCL INFO Channel 07/0 : 0[2] -> 1[6] via P2P/IPC
+ip-26-0-163-58:934721:935254 [1] NCCL INFO Channel 10/0 : 0[1] -> 1[5] via P2P/IPC
+ip-26-0-162-14:2525082:2525621 [7] NCCL INFO Channel 16/0 : 1[7] -> 0[3] via P2P/IPC
+ip-26-0-163-58:934726:935249 [6] NCCL INFO Channel 08/0 : 1[6] -> 0[2] via P2P/IPC
+ip-26-0-162-14:2525076:2525615 [1] NCCL INFO Channel 13/0 : 0[1] -> 1[5] via P2P/IPC
+ip-26-0-170-31:3197181:3197829 [1] NCCL INFO Channel 19/0 : 0[1] -> 1[5] via P2P/IPC
+ip-26-0-170-31:3197183:3197826 [3] NCCL INFO Channel 18/0 : 0[3] -> 1[7] via P2P/IPC
+ip-26-0-162-14:2525081:2525622 [6] NCCL INFO Channel 14/0 : 1[6] -> 0[2] via P2P/IPC
+ip-26-0-170-31:3197182:3197823 [2] NCCL INFO Channel 14/0 : 0[2] -> 1[6] via P2P/IPC
+ip-26-0-170-31:3197187:3197827 [7] NCCL INFO Channel 16/0 : 1[7] -> 0[3] via P2P/IPC
+ip-26-0-170-31:3197185:3197830 [5] NCCL INFO Channel 17/0 : 1[5] -> 0[1] via P2P/IPC
+ip-26-0-170-31:3197186:3197824 [6] NCCL INFO Channel 13/0 : 1[6] -> 0[2] via P2P/IPC
+ip-26-0-162-14:2525079:2525613 [4] NCCL INFO Channel 13/0 : 1[4] -> 0[0] via P2P/IPC
+ip-26-0-162-14:2525078:2525618 [3] NCCL INFO Channel 13/0 : 0[3] -> 1[7] via P2P/IPC
+ip-26-0-162-14:2525080:2525616 [5] NCCL INFO Channel 19/0 : 1[5] -> 0[1] via P2P/IPC
+ip-26-0-162-14:2525075:2525612 [0] NCCL INFO Channel 04/0 : 0[0] -> 1[4] via P2P/IPC
+ip-26-0-162-14:2525077:2525620 [2] NCCL INFO Channel 08/0 : 0[2] -> 1[6] via P2P/IPC
+ip-26-0-163-58:934725:935255 [5] NCCL INFO Channel 15/0 : 1[5] -> 0[1] via P2P/IPC
+ip-26-0-163-58:934727:935252 [7] NCCL INFO Channel 13/0 : 1[7] -> 0[3] via P2P/IPC
+ip-26-0-162-14:2525082:2525621 [7] NCCL INFO Channel 17/0 : 1[7] -> 0[3] via P2P/IPC
+ip-26-0-163-58:934721:935254 [1] NCCL INFO Channel 11/0 : 0[1] -> 1[5] via P2P/IPC
+ip-26-0-163-58:934720:935245 [0] NCCL INFO Channel 22/0 : 0[0] -> 1[4] via P2P/IPC
+ip-26-0-162-14:2525076:2525615 [1] NCCL INFO Channel 14/0 : 0[1] -> 1[5] via P2P/IPC
+ip-26-0-163-58:934727:935252 [7] NCCL INFO Channel 14/0 : 1[7] -> 0[3] via P2P/IPC
+ip-26-0-163-58:934725:935255 [5] NCCL INFO Channel 16/0 : 1[5] -> 0[1] via P2P/IPC
+ip-26-0-162-14:2525081:2525622 [6] NCCL INFO Channel 15/0 : 1[6] -> 0[2] via P2P/IPC
+ip-26-0-163-58:934720:935245 [0] NCCL INFO Channel 23/0 : 0[0] -> 1[4] via P2P/IPC
+ip-26-0-163-58:934721:935254 [1] NCCL INFO Channel 12/0 : 0[1] -> 1[5] via P2P/IPC
+ip-26-0-163-58:934723:935251 [3] NCCL INFO Channel 18/0 : 0[3] -> 1[7] via P2P/IPC
+ip-26-0-162-14:2525079:2525613 [4] NCCL INFO Channel 14/0 : 1[4] -> 0[0] via P2P/IPC
+ip-26-0-162-14:2525078:2525618 [3] NCCL INFO Channel 14/0 : 0[3] -> 1[7] via P2P/IPC
+ip-26-0-162-14:2525080:2525616 [5] NCCL INFO Channel 20/0 : 1[5] -> 0[1] via P2P/IPC
+ip-26-0-162-14:2525075:2525612 [0] NCCL INFO Channel 05/0 : 0[0] -> 1[4] via P2P/IPC
+ip-26-0-162-14:2525077:2525620 [2] NCCL INFO Channel 09/0 : 0[2] -> 1[6] via P2P/IPC
+ip-26-0-162-14:2525082:2525621 [7] NCCL INFO Channel 18/0 : 1[7] -> 0[3] via P2P/IPC
+ip-26-0-163-58:934727:935252 [7] NCCL INFO Channel 15/0 : 1[7] -> 0[3] via P2P/IPC
+ip-26-0-163-58:934725:935255 [5] NCCL INFO Channel 17/0 : 1[5] -> 0[1] via P2P/IPC
+ip-26-0-162-14:2525076:2525615 [1] NCCL INFO Channel 15/0 : 0[1] -> 1[5] via P2P/IPC
+ip-26-0-163-58:934726:935249 [6] NCCL INFO Channel 09/0 : 1[6] -> 0[2] via P2P/IPC
+ip-26-0-163-58:934721:935254 [1] NCCL INFO Channel 13/0 : 0[1] -> 1[5] via P2P/IPC
+ip-26-0-162-14:2525081:2525622 [6] NCCL INFO Channel 16/0 : 1[6] -> 0[2] via P2P/IPC
+ip-26-0-163-58:934723:935251 [3] NCCL INFO Channel 19/0 : 0[3] -> 1[7] via P2P/IPC
+ip-26-0-170-31:3197181:3197829 [1] NCCL INFO Channel 20/0 : 0[1] -> 1[5] via P2P/IPC
+ip-26-0-170-31:3197183:3197826 [3] NCCL INFO Channel 19/0 : 0[3] -> 1[7] via P2P/IPC
+ip-26-0-170-31:3197186:3197824 [6] NCCL INFO Channel 14/0 : 1[6] -> 0[2] via P2P/IPC
+ip-26-0-170-31:3197182:3197823 [2] NCCL INFO Channel 15/0 : 0[2] -> 1[6] via P2P/IPC
+ip-26-0-170-31:3197185:3197830 [5] NCCL INFO Channel 18/0 : 1[5] -> 0[1] via P2P/IPC
+ip-26-0-170-31:3197187:3197827 [7] NCCL INFO Channel 17/0 : 1[7] -> 0[3] via P2P/IPC
+ip-26-0-162-14:2525079:2525613 [4] NCCL INFO Channel 15/0 : 1[4] -> 0[0] via P2P/IPC
+ip-26-0-162-14:2525078:2525618 [3] NCCL INFO Channel 15/0 : 0[3] -> 1[7] via P2P/IPC
+ip-26-0-163-58:934727:935252 [7] NCCL INFO Channel 16/0 : 1[7] -> 0[3] via P2P/IPC
+ip-26-0-162-14:2525077:2525620 [2] NCCL INFO Channel 10/0 : 0[2] -> 1[6] via P2P/IPC
+ip-26-0-162-14:2525075:2525612 [0] NCCL INFO Channel 06/0 : 0[0] -> 1[4] via P2P/IPC
+ip-26-0-162-14:2525080:2525616 [5] NCCL INFO Channel 21/0 : 1[5] -> 0[1] via P2P/IPC
+ip-26-0-163-58:934724:935246 [4] NCCL INFO Channel 21/0 : 1[4] -> 0[0] via P2P/IPC
+ip-26-0-163-58:934722:935248 [2] NCCL INFO Channel 23/0 : 0[2] -> 1[6] via P2P/IPC
+ip-26-0-162-14:2525082:2525621 [7] NCCL INFO Channel 19/0 : 1[7] -> 0[3] via P2P/IPC
+ip-26-0-163-58:934725:935255 [5] NCCL INFO Channel 18/0 : 1[5] -> 0[1] via P2P/IPC
+ip-26-0-163-58:934726:935249 [6] NCCL INFO Channel 10/0 : 1[6] -> 0[2] via P2P/IPC
+ip-26-0-162-14:2525076:2525615 [1] NCCL INFO Channel 16/0 : 0[1] -> 1[5] via P2P/IPC
+ip-26-0-162-14:2525081:2525622 [6] NCCL INFO Channel 17/0 : 1[6] -> 0[2] via P2P/IPC
+ip-26-0-163-58:934724:935246 [4] NCCL INFO Channel 22/0 : 1[4] -> 0[0] via P2P/IPC
+ip-26-0-163-58:934727:935252 [7] NCCL INFO Channel 17/0 : 1[7] -> 0[3] via P2P/IPC
+ip-26-0-163-58:934723:935251 [3] NCCL INFO Channel 20/0 : 0[3] -> 1[7] via P2P/IPC
+ip-26-0-162-14:2525078:2525618 [3] NCCL INFO Channel 16/0 : 0[3] -> 1[7] via P2P/IPC
+ip-26-0-162-14:2525079:2525613 [4] NCCL INFO Channel 16/0 : 1[4] -> 0[0] via P2P/IPC
+ip-26-0-162-14:2525075:2525612 [0] NCCL INFO Channel 07/0 : 0[0] -> 1[4] via P2P/IPC
+ip-26-0-162-14:2525077:2525620 [2] NCCL INFO Channel 11/0 : 0[2] -> 1[6] via P2P/IPC
+ip-26-0-162-14:2525080:2525616 [5] NCCL INFO Channel 22/0 : 1[5] -> 0[1] via P2P/IPC
+ip-26-0-162-14:2525082:2525621 [7] NCCL INFO Channel 20/0 : 1[7] -> 0[3] via P2P/IPC
+ip-26-0-162-14:2525076:2525615 [1] NCCL INFO Channel 17/0 : 0[1] -> 1[5] via P2P/IPC
+ip-26-0-163-58:934724:935246 [4] NCCL INFO Channel 23/0 : 1[4] -> 0[0] via P2P/IPC
+ip-26-0-163-58:934725:935255 [5] NCCL INFO Channel 19/0 : 1[5] -> 0[1] via P2P/IPC
+ip-26-0-163-58:934721:935254 [1] NCCL INFO Channel 14/0 : 0[1] -> 1[5] via P2P/IPC
+ip-26-0-162-14:2525081:2525622 [6] NCCL INFO Channel 18/0 : 1[6] -> 0[2] via P2P/IPC
+ip-26-0-163-58:934727:935252 [7] NCCL INFO Channel 18/0 : 1[7] -> 0[3] via P2P/IPC
+ip-26-0-170-31:3197181:3197829 [1] NCCL INFO Channel 21/0 : 0[1] -> 1[5] via P2P/IPC
+ip-26-0-170-31:3197183:3197826 [3] NCCL INFO Channel 20/0 : 0[3] -> 1[7] via P2P/IPC
+ip-26-0-170-31:3197186:3197824 [6] NCCL INFO Channel 15/0 : 1[6] -> 0[2] via P2P/IPC
+ip-26-0-170-31:3197185:3197830 [5] NCCL INFO Channel 19/0 : 1[5] -> 0[1] via P2P/IPC
+ip-26-0-170-31:3197187:3197827 [7] NCCL INFO Channel 18/0 : 1[7] -> 0[3] via P2P/IPC
+ip-26-0-170-31:3197182:3197823 [2] NCCL INFO Channel 16/0 : 0[2] -> 1[6] via P2P/IPC
+ip-26-0-163-58:934721:935254 [1] NCCL INFO Channel 15/0 : 0[1] -> 1[5] via P2P/IPC
+ip-26-0-162-14:2525078:2525618 [3] NCCL INFO Channel 17/0 : 0[3] -> 1[7] via P2P/IPC
+ip-26-0-163-58:934725:935255 [5] NCCL INFO Channel 20/0 : 1[5] -> 0[1] via P2P/IPC
+ip-26-0-162-14:2525075:2525612 [0] NCCL INFO Channel 08/0 : 0[0] -> 1[4] via P2P/IPC
+ip-26-0-162-14:2525077:2525620 [2] NCCL INFO Channel 12/0 : 0[2] -> 1[6] via P2P/IPC
+ip-26-0-162-14:2525080:2525616 [5] NCCL INFO Channel 23/0 : 1[5] -> 0[1] via P2P/IPC
+ip-26-0-163-58:934727:935252 [7] NCCL INFO Channel 19/0 : 1[7] -> 0[3] via P2P/IPC
+ip-26-0-162-14:2525082:2525621 [7] NCCL INFO Channel 21/0 : 1[7] -> 0[3] via P2P/IPC
+ip-26-0-162-14:2525076:2525615 [1] NCCL INFO Channel 18/0 : 0[1] -> 1[5] via P2P/IPC
+ip-26-0-162-14:2525081:2525622 [6] NCCL INFO Channel 19/0 : 1[6] -> 0[2] via P2P/IPC
+ip-26-0-162-14:2525079:2525613 [4] NCCL INFO Channel 17/0 : 1[4] -> 0[0] via P2P/IPC
+ip-26-0-162-14:2525078:2525618 [3] NCCL INFO Channel 18/0 : 0[3] -> 1[7] via P2P/IPC
+ip-26-0-162-14:2525075:2525612 [0] NCCL INFO Channel 09/0 : 0[0] -> 1[4] via P2P/IPC
+ip-26-0-162-14:2525077:2525620 [2] NCCL INFO Channel 13/0 : 0[2] -> 1[6] via P2P/IPC
+ip-26-0-162-14:2525082:2525621 [7] NCCL INFO Channel 22/0 : 1[7] -> 0[3] via P2P/IPC
+ip-26-0-162-14:2525076:2525615 [1] NCCL INFO Channel 19/0 : 0[1] -> 1[5] via P2P/IPC
+ip-26-0-163-58:934721:935254 [1] NCCL INFO Channel 16/0 : 0[1] -> 1[5] via P2P/IPC
+ip-26-0-163-58:934725:935255 [5] NCCL INFO Channel 21/0 : 1[5] -> 0[1] via P2P/IPC
+ip-26-0-162-14:2525081:2525622 [6] NCCL INFO Channel 20/0 : 1[6] -> 0[2] via P2P/IPC
+ip-26-0-162-14:2525078:2525618 [3] NCCL INFO Channel 19/0 : 0[3] -> 1[7] via P2P/IPC
+ip-26-0-162-14:2525081:2525622 [6] NCCL INFO Channel 21/0 : 1[6] -> 0[2] via P2P/IPC
+ip-26-0-170-31:3197187:3197827 [7] NCCL INFO Channel 19/0 : 1[7] -> 0[3] via P2P/IPC
+ip-26-0-163-58:934727:935252 [7] NCCL INFO Channel 20/0 : 1[7] -> 0[3] via P2P/IPC
+ip-26-0-162-14:2525082:2525621 [7] NCCL INFO Channel 23/0 : 1[7] -> 0[3] via P2P/IPC
+ip-26-0-162-14:2525079:2525613 [4] NCCL INFO Channel 18/0 : 1[4] -> 0[0] via P2P/IPC
+ip-26-0-162-14:2525078:2525618 [3] NCCL INFO Channel 20/0 : 0[3] -> 1[7] via P2P/IPC
+ip-26-0-162-14:2525081:2525622 [6] NCCL INFO Channel 22/0 : 1[6] -> 0[2] via P2P/IPC
+ip-26-0-163-58:934726:935249 [6] NCCL INFO Channel 11/0 : 1[6] -> 0[2] via P2P/IPC
+ip-26-0-163-58:934723:935251 [3] NCCL INFO Channel 21/0 : 0[3] -> 1[7] via P2P/IPC
+ip-26-0-162-14:2525076:2525615 [1] NCCL INFO Channel 20/0 : 0[1] -> 1[5] via P2P/IPC
+ip-26-0-162-14:2525079:2525613 [4] NCCL INFO Channel 19/0 : 1[4] -> 0[0] via P2P/IPC
+ip-26-0-162-14:2525078:2525618 [3] NCCL INFO Channel 21/0 : 0[3] -> 1[7] via P2P/IPC
+ip-26-0-162-14:2525081:2525622 [6] NCCL INFO Channel 23/0 : 1[6] -> 0[2] via P2P/IPC
+ip-26-0-162-14:2525075:2525612 [0] NCCL INFO Channel 10/0 : 0[0] -> 1[4] via P2P/IPC
+ip-26-0-163-58:934725:935255 [5] NCCL INFO Channel 22/0 : 1[5] -> 0[1] via P2P/IPC
+ip-26-0-162-14:2525077:2525620 [2] NCCL INFO Channel 14/0 : 0[2] -> 1[6] via P2P/IPC
+ip-26-0-162-14:2525076:2525615 [1] NCCL INFO Channel 21/0 : 0[1] -> 1[5] via P2P/IPC
+ip-26-0-162-14:2525079:2525613 [4] NCCL INFO Channel 20/0 : 1[4] -> 0[0] via P2P/IPC
+ip-26-0-162-14:2525078:2525618 [3] NCCL INFO Channel 22/0 : 0[3] -> 1[7] via P2P/IPC
+ip-26-0-170-31:3197183:3197826 [3] NCCL INFO Channel 21/0 : 0[3] -> 1[7] via P2P/IPC
+ip-26-0-170-31:3197186:3197824 [6] NCCL INFO Channel 16/0 : 1[6] -> 0[2] via P2P/IPC
+ip-26-0-170-31:3197185:3197830 [5] NCCL INFO Channel 20/0 : 1[5] -> 0[1] via P2P/IPC
+ip-26-0-163-58:934727:935252 [7] NCCL INFO Channel 21/0 : 1[7] -> 0[3] via P2P/IPC
+ip-26-0-162-14:2525075:2525612 [0] NCCL INFO Channel 11/0 : 0[0] -> 1[4] via P2P/IPC
+ip-26-0-170-31:3197187:3197827 [7] NCCL INFO Channel 20/0 : 1[7] -> 0[3] via P2P/IPC
+ip-26-0-162-14:2525077:2525620 [2] NCCL INFO Channel 15/0 : 0[2] -> 1[6] via P2P/IPC
+ip-26-0-162-14:2525076:2525615 [1] NCCL INFO Channel 22/0 : 0[1] -> 1[5] via P2P/IPC
+ip-26-0-162-14:2525079:2525613 [4] NCCL INFO Channel 21/0 : 1[4] -> 0[0] via P2P/IPC
+ip-26-0-162-14:2525078:2525618 [3] NCCL INFO Channel 23/0 : 0[3] -> 1[7] via P2P/IPC
+ip-26-0-163-58:934725:935255 [5] NCCL INFO Channel 23/0 : 1[5] -> 0[1] via P2P/IPC
+ip-26-0-170-31:3197182:3197823 [2] NCCL INFO Channel 17/0 : 0[2] -> 1[6] via P2P/IPC
+ip-26-0-170-31:3197183:3197826 [3] NCCL INFO Channel 22/0 : 0[3] -> 1[7] via P2P/IPC
+ip-26-0-170-31:3197181:3197829 [1] NCCL INFO Channel 22/0 : 0[1] -> 1[5] via P2P/IPC
+ip-26-0-162-14:2525075:2525612 [0] NCCL INFO Channel 12/0 : 0[0] -> 1[4] via P2P/IPC
+ip-26-0-163-58:934727:935252 [7] NCCL INFO Channel 22/0 : 1[7] -> 0[3] via P2P/IPC
+ip-26-0-162-14:2525077:2525620 [2] NCCL INFO Channel 16/0 : 0[2] -> 1[6] via P2P/IPC
+ip-26-0-162-14:2525076:2525615 [1] NCCL INFO Channel 23/0 : 0[1] -> 1[5] via P2P/IPC
+ip-26-0-162-14:2525079:2525613 [4] NCCL INFO Channel 22/0 : 1[4] -> 0[0] via P2P/IPC
+ip-26-0-162-14:2525075:2525612 [0] NCCL INFO Channel 13/0 : 0[0] -> 1[4] via P2P/IPC
+ip-26-0-163-58:934727:935252 [7] NCCL INFO Channel 23/0 : 1[7] -> 0[3] via P2P/IPC
+ip-26-0-170-31:3197182:3197823 [2] NCCL INFO Channel 18/0 : 0[2] -> 1[6] via P2P/IPC
+ip-26-0-170-31:3197183:3197826 [3] NCCL INFO Channel 23/0 : 0[3] -> 1[7] via P2P/IPC
+ip-26-0-170-31:3197181:3197829 [1] NCCL INFO Channel 23/0 : 0[1] -> 1[5] via P2P/IPC
+ip-26-0-163-58:934721:935254 [1] NCCL INFO Channel 17/0 : 0[1] -> 1[5] via P2P/IPC
+ip-26-0-162-14:2525077:2525620 [2] NCCL INFO Channel 17/0 : 0[2] -> 1[6] via P2P/IPC
+ip-26-0-162-14:2525079:2525613 [4] NCCL INFO Channel 23/0 : 1[4] -> 0[0] via P2P/IPC
+ip-26-0-162-14:2525075:2525612 [0] NCCL INFO Channel 14/0 : 0[0] -> 1[4] via P2P/IPC
+ip-26-0-163-58:934726:935249 [6] NCCL INFO Channel 12/0 : 1[6] -> 0[2] via P2P/IPC
+ip-26-0-170-31:3197185:3197830 [5] NCCL INFO Channel 21/0 : 1[5] -> 0[1] via P2P/IPC
+ip-26-0-163-58:934723:935251 [3] NCCL INFO Channel 22/0 : 0[3] -> 1[7] via P2P/IPC
+ip-26-0-163-58:934721:935254 [1] NCCL INFO Channel 18/0 : 0[1] -> 1[5] via P2P/IPC
+ip-26-0-170-31:3197182:3197823 [2] NCCL INFO Channel 19/0 : 0[2] -> 1[6] via P2P/IPC
+ip-26-0-170-31:3197186:3197824 [6] NCCL INFO Channel 17/0 : 1[6] -> 0[2] via P2P/IPC
+ip-26-0-170-132:3330184:3330730 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-170-31:3197182:3197823 [2] NCCL INFO Channel 20/0 : 0[2] -> 1[6] via P2P/IPC
+ip-26-0-162-14:2525077:2525620 [2] NCCL INFO Channel 18/0 : 0[2] -> 1[6] via P2P/IPC
+ip-26-0-162-14:2525075:2525612 [0] NCCL INFO Channel 15/0 : 0[0] -> 1[4] via P2P/IPC
+ip-26-0-170-31:3197186:3197824 [6] NCCL INFO Channel 18/0 : 1[6] -> 0[2] via P2P/IPC
+ip-26-0-170-31:3197187:3197827 [7] NCCL INFO Channel 21/0 : 1[7] -> 0[3] via P2P/IPC
+ip-26-0-170-31:3197185:3197830 [5] NCCL INFO Channel 22/0 : 1[5] -> 0[1] via P2P/IPC
+ip-26-0-163-58:934726:935249 [6] NCCL INFO Channel 13/0 : 1[6] -> 0[2] via P2P/IPC
+ip-26-0-170-31:3197182:3197823 [2] NCCL INFO Channel 21/0 : 0[2] -> 1[6] via P2P/IPC
+ip-26-0-163-58:934721:935254 [1] NCCL INFO Channel 19/0 : 0[1] -> 1[5] via P2P/IPC
+ip-26-0-163-58:934723:935251 [3] NCCL INFO Channel 23/0 : 0[3] -> 1[7] via P2P/IPC
+ip-26-0-170-31:3197187:3197827 [7] NCCL INFO Channel 22/0 : 1[7] -> 0[3] via P2P/IPC
+ip-26-0-170-31:3197186:3197824 [6] NCCL INFO Channel 19/0 : 1[6] -> 0[2] via P2P/IPC
+ip-26-0-170-31:3197185:3197830 [5] NCCL INFO Channel 23/0 : 1[5] -> 0[1] via P2P/IPC
+ip-26-0-162-14:2525077:2525620 [2] NCCL INFO Channel 19/0 : 0[2] -> 1[6] via P2P/IPC
+ip-26-0-170-31:3197182:3197823 [2] NCCL INFO Channel 22/0 : 0[2] -> 1[6] via P2P/IPC
+ip-26-0-162-14:2525075:2525612 [0] NCCL INFO Channel 16/0 : 0[0] -> 1[4] via P2P/IPC
+ip-26-0-170-31:3197187:3197827 [7] NCCL INFO Channel 23/0 : 1[7] -> 0[3] via P2P/IPC
+ip-26-0-170-31:3197186:3197824 [6] NCCL INFO Channel 20/0 : 1[6] -> 0[2] via P2P/IPC
+ip-26-0-170-31:3197182:3197823 [2] NCCL INFO Channel 23/0 : 0[2] -> 1[6] via P2P/IPC
+ip-26-0-162-14:2525077:2525620 [2] NCCL INFO Channel 20/0 : 0[2] -> 1[6] via P2P/IPC
+ip-26-0-162-14:2525075:2525612 [0] NCCL INFO Channel 17/0 : 0[0] -> 1[4] via P2P/IPC
+ip-26-0-170-31:3197186:3197824 [6] NCCL INFO Channel 21/0 : 1[6] -> 0[2] via P2P/IPC
+ip-26-0-162-14:2525077:2525620 [2] NCCL INFO Channel 21/0 : 0[2] -> 1[6] via P2P/IPC
+ip-26-0-162-14:2525077:2525620 [2] NCCL INFO Channel 22/0 : 0[2] -> 1[6] via P2P/IPC
+ip-26-0-170-31:3197186:3197824 [6] NCCL INFO Channel 22/0 : 1[6] -> 0[2] via P2P/IPC
+ip-26-0-162-14:2525075:2525612 [0] NCCL INFO Channel 18/0 : 0[0] -> 1[4] via P2P/IPC
+ip-26-0-170-132:3330187:3330722 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-162-14:2525075:2525612 [0] NCCL INFO Channel 19/0 : 0[0] -> 1[4] via P2P/IPC
+ip-26-0-170-31:3197186:3197824 [6] NCCL INFO Channel 23/0 : 1[6] -> 0[2] via P2P/IPC
+ip-26-0-162-14:2525077:2525620 [2] NCCL INFO Channel 23/0 : 0[2] -> 1[6] via P2P/IPC
+ip-26-0-162-14:2525075:2525612 [0] NCCL INFO Channel 20/0 : 0[0] -> 1[4] via P2P/IPC
+ip-26-0-163-58:934726:935249 [6] NCCL INFO Channel 14/0 : 1[6] -> 0[2] via P2P/IPC
+ip-26-0-163-58:934721:935254 [1] NCCL INFO Channel 20/0 : 0[1] -> 1[5] via P2P/IPC
+ip-26-0-170-132:3330183:3330721 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-170-132:3330183:3330721 [0] NCCL INFO Channel 00/24 :    0   1
+ip-26-0-170-132:3330187:3330722 [4] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] -1/-1/-1->1->0 [3] -1/-1/-1->1->0 [4] -1/-1/-1->1->0 [5] -1/-1/-1->1->0 [6] 0/-1/-1->1->-1 [7] 0/-1/-1->1->-1 [8] 0/-1/-1->1->-1 [9] 0/-1/-1->1->-1 [10] 0/-1/-1->1->-1 [11] 0/-1/-1->1->-1 [12] -1/-1/-1->1->0 [13] -1/-1/-1->1->0 [14] -1/-1/-1->1->0 [15] -1/-1/-1->1->0 [16] -1/-1/-1->1->0 [17] -1/-1/-1->1->0 [18] 0/-1/-1->1->-1 [19] 0/-1/-1->1->-1 [20] 0/-1/-1->1->-1 [21] 0/-1/-1->1->-1 [22] 0/-1/-1->1->-1 [23] 0/-1/-1->1->-1
+ip-26-0-170-132:3330183:3330721 [0] NCCL INFO Channel 01/24 :    0   1
+ip-26-0-170-132:3330183:3330721 [0] NCCL INFO Channel 02/24 :    0   1
+ip-26-0-170-132:3330187:3330722 [4] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-170-132:3330183:3330721 [0] NCCL INFO Channel 03/24 :    0   1
+ip-26-0-170-132:3330183:3330721 [0] NCCL INFO Channel 04/24 :    0   1
+ip-26-0-170-132:3330183:3330721 [0] NCCL INFO Channel 05/24 :    0   1
+ip-26-0-170-132:3330183:3330721 [0] NCCL INFO Channel 06/24 :    0   1
+ip-26-0-170-132:3330183:3330721 [0] NCCL INFO Channel 07/24 :    0   1
+ip-26-0-170-132:3330183:3330721 [0] NCCL INFO Channel 08/24 :    0   1
+ip-26-0-170-132:3330183:3330721 [0] NCCL INFO Channel 09/24 :    0   1
+ip-26-0-170-132:3330183:3330721 [0] NCCL INFO Channel 10/24 :    0   1
+ip-26-0-170-132:3330183:3330721 [0] NCCL INFO Channel 11/24 :    0   1
+ip-26-0-170-132:3330183:3330721 [0] NCCL INFO Channel 12/24 :    0   1
+ip-26-0-170-132:3330183:3330721 [0] NCCL INFO Channel 13/24 :    0   1
+ip-26-0-170-132:3330183:3330721 [0] NCCL INFO Channel 14/24 :    0   1
+ip-26-0-170-132:3330183:3330721 [0] NCCL INFO Channel 15/24 :    0   1
+ip-26-0-170-132:3330183:3330721 [0] NCCL INFO Channel 16/24 :    0   1
+ip-26-0-170-132:3330183:3330721 [0] NCCL INFO Channel 17/24 :    0   1
+ip-26-0-170-132:3330183:3330721 [0] NCCL INFO Channel 18/24 :    0   1
+ip-26-0-170-132:3330183:3330721 [0] NCCL INFO Channel 19/24 :    0   1
+ip-26-0-170-132:3330183:3330721 [0] NCCL INFO Channel 20/24 :    0   1
+ip-26-0-170-132:3330183:3330721 [0] NCCL INFO Channel 21/24 :    0   1
+ip-26-0-170-132:3330183:3330721 [0] NCCL INFO Channel 22/24 :    0   1
+ip-26-0-170-132:3330183:3330721 [0] NCCL INFO Channel 23/24 :    0   1
+ip-26-0-170-132:3330183:3330721 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] -1/-1/-1->0->1 [7] -1/-1/-1->0->1 [8] -1/-1/-1->0->1 [9] -1/-1/-1->0->1 [10] -1/-1/-1->0->1 [11] -1/-1/-1->0->1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] -1/-1/-1->0->1 [19] -1/-1/-1->0->1 [20] -1/-1/-1->0->1 [21] -1/-1/-1->0->1 [22] -1/-1/-1->0->1 [23] -1/-1/-1->0->1
+ip-26-0-170-132:3330183:3330721 [0] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-170-132:3330186:3330725 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-163-58:934726:935249 [6] NCCL INFO Channel 15/0 : 1[6] -> 0[2] via P2P/IPC
+ip-26-0-162-14:2525075:2525612 [0] NCCL INFO Channel 21/0 : 0[0] -> 1[4] via P2P/IPC
+ip-26-0-170-132:3330190:3330726 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-170-132:3330186:3330725 [3] NCCL INFO Channel 00/24 :    0   1
+ip-26-0-170-132:3330186:3330725 [3] NCCL INFO Channel 01/24 :    0   1
+ip-26-0-170-132:3330186:3330725 [3] NCCL INFO Channel 02/24 :    0   1
+ip-26-0-170-132:3330186:3330725 [3] NCCL INFO Channel 03/24 :    0   1
+ip-26-0-170-132:3330186:3330725 [3] NCCL INFO Channel 04/24 :    0   1
+ip-26-0-170-132:3330186:3330725 [3] NCCL INFO Channel 05/24 :    0   1
+ip-26-0-170-132:3330186:3330725 [3] NCCL INFO Channel 06/24 :    0   1
+ip-26-0-170-132:3330186:3330725 [3] NCCL INFO Channel 07/24 :    0   1
+ip-26-0-170-132:3330186:3330725 [3] NCCL INFO Channel 08/24 :    0   1
+ip-26-0-170-132:3330186:3330725 [3] NCCL INFO Channel 09/24 :    0   1
+ip-26-0-170-132:3330186:3330725 [3] NCCL INFO Channel 10/24 :    0   1
+ip-26-0-170-132:3330186:3330725 [3] NCCL INFO Channel 11/24 :    0   1
+ip-26-0-170-132:3330186:3330725 [3] NCCL INFO Channel 12/24 :    0   1
+ip-26-0-170-132:3330190:3330726 [7] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] -1/-1/-1->1->0 [3] -1/-1/-1->1->0 [4] -1/-1/-1->1->0 [5] -1/-1/-1->1->0 [6] 0/-1/-1->1->-1 [7] 0/-1/-1->1->-1 [8] 0/-1/-1->1->-1 [9] 0/-1/-1->1->-1 [10] 0/-1/-1->1->-1 [11] 0/-1/-1->1->-1 [12] -1/-1/-1->1->0 [13] -1/-1/-1->1->0 [14] -1/-1/-1->1->0 [15] -1/-1/-1->1->0 [16] -1/-1/-1->1->0 [17] -1/-1/-1->1->0 [18] 0/-1/-1->1->-1 [19] 0/-1/-1->1->-1 [20] 0/-1/-1->1->-1 [21] 0/-1/-1->1->-1 [22] 0/-1/-1->1->-1 [23] 0/-1/-1->1->-1
+ip-26-0-170-132:3330186:3330725 [3] NCCL INFO Channel 13/24 :    0   1
+ip-26-0-170-132:3330190:3330726 [7] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-170-132:3330186:3330725 [3] NCCL INFO Channel 14/24 :    0   1
+ip-26-0-170-132:3330186:3330725 [3] NCCL INFO Channel 15/24 :    0   1
+ip-26-0-170-132:3330186:3330725 [3] NCCL INFO Channel 16/24 :    0   1
+ip-26-0-170-132:3330186:3330725 [3] NCCL INFO Channel 17/24 :    0   1
+ip-26-0-170-132:3330186:3330725 [3] NCCL INFO Channel 18/24 :    0   1
+ip-26-0-170-132:3330186:3330725 [3] NCCL INFO Channel 19/24 :    0   1
+ip-26-0-170-132:3330186:3330725 [3] NCCL INFO Channel 20/24 :    0   1
+ip-26-0-170-132:3330186:3330725 [3] NCCL INFO Channel 21/24 :    0   1
+ip-26-0-170-132:3330186:3330725 [3] NCCL INFO Channel 22/24 :    0   1
+ip-26-0-170-132:3330186:3330725 [3] NCCL INFO Channel 23/24 :    0   1
+ip-26-0-170-132:3330186:3330725 [3] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] -1/-1/-1->0->1 [7] -1/-1/-1->0->1 [8] -1/-1/-1->0->1 [9] -1/-1/-1->0->1 [10] -1/-1/-1->0->1 [11] -1/-1/-1->0->1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] -1/-1/-1->0->1 [19] -1/-1/-1->0->1 [20] -1/-1/-1->0->1 [21] -1/-1/-1->0->1 [22] -1/-1/-1->0->1 [23] -1/-1/-1->0->1
+ip-26-0-170-132:3330186:3330725 [3] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-163-58:934721:935254 [1] NCCL INFO Channel 21/0 : 0[1] -> 1[5] via P2P/IPC
+ip-26-0-170-132:3330189:3330728 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-163-58:934726:935249 [6] NCCL INFO Channel 16/0 : 1[6] -> 0[2] via P2P/IPC
+ip-26-0-170-132:3330188:3330731 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-170-132:3330184:3330730 [1] NCCL INFO Channel 00/24 :    0   1
+ip-26-0-170-132:3330188:3330731 [5] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] -1/-1/-1->1->0 [3] -1/-1/-1->1->0 [4] -1/-1/-1->1->0 [5] -1/-1/-1->1->0 [6] 0/-1/-1->1->-1 [7] 0/-1/-1->1->-1 [8] 0/-1/-1->1->-1 [9] 0/-1/-1->1->-1 [10] 0/-1/-1->1->-1 [11] 0/-1/-1->1->-1 [12] -1/-1/-1->1->0 [13] -1/-1/-1->1->0 [14] -1/-1/-1->1->0 [15] -1/-1/-1->1->0 [16] -1/-1/-1->1->0 [17] -1/-1/-1->1->0 [18] 0/-1/-1->1->-1 [19] 0/-1/-1->1->-1 [20] 0/-1/-1->1->-1 [21] 0/-1/-1->1->-1 [22] 0/-1/-1->1->-1 [23] 0/-1/-1->1->-1
+ip-26-0-170-132:3330184:3330730 [1] NCCL INFO Channel 01/24 :    0   1
+ip-26-0-170-132:3330188:3330731 [5] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-170-132:3330184:3330730 [1] NCCL INFO Channel 02/24 :    0   1
+ip-26-0-170-132:3330184:3330730 [1] NCCL INFO Channel 03/24 :    0   1
+ip-26-0-170-132:3330184:3330730 [1] NCCL INFO Channel 04/24 :    0   1
+ip-26-0-170-132:3330184:3330730 [1] NCCL INFO Channel 05/24 :    0   1
+ip-26-0-170-132:3330184:3330730 [1] NCCL INFO Channel 06/24 :    0   1
+ip-26-0-170-132:3330184:3330730 [1] NCCL INFO Channel 07/24 :    0   1
+ip-26-0-170-132:3330184:3330730 [1] NCCL INFO Channel 08/24 :    0   1
+ip-26-0-170-132:3330184:3330730 [1] NCCL INFO Channel 09/24 :    0   1
+ip-26-0-170-132:3330184:3330730 [1] NCCL INFO Channel 10/24 :    0   1
+ip-26-0-170-132:3330184:3330730 [1] NCCL INFO Channel 11/24 :    0   1
+ip-26-0-170-132:3330184:3330730 [1] NCCL INFO Channel 12/24 :    0   1
+ip-26-0-170-132:3330184:3330730 [1] NCCL INFO Channel 13/24 :    0   1
+ip-26-0-170-132:3330184:3330730 [1] NCCL INFO Channel 14/24 :    0   1
+ip-26-0-170-132:3330184:3330730 [1] NCCL INFO Channel 15/24 :    0   1
+ip-26-0-170-132:3330184:3330730 [1] NCCL INFO Channel 16/24 :    0   1
+ip-26-0-170-132:3330184:3330730 [1] NCCL INFO Channel 17/24 :    0   1
+ip-26-0-170-132:3330184:3330730 [1] NCCL INFO Channel 18/24 :    0   1
+ip-26-0-170-132:3330184:3330730 [1] NCCL INFO Channel 19/24 :    0   1
+ip-26-0-170-132:3330184:3330730 [1] NCCL INFO Channel 20/24 :    0   1
+ip-26-0-170-132:3330184:3330730 [1] NCCL INFO Channel 21/24 :    0   1
+ip-26-0-170-132:3330184:3330730 [1] NCCL INFO Channel 22/24 :    0   1
+ip-26-0-170-132:3330184:3330730 [1] NCCL INFO Channel 23/24 :    0   1
+ip-26-0-170-132:3330184:3330730 [1] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] -1/-1/-1->0->1 [7] -1/-1/-1->0->1 [8] -1/-1/-1->0->1 [9] -1/-1/-1->0->1 [10] -1/-1/-1->0->1 [11] -1/-1/-1->0->1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] -1/-1/-1->0->1 [19] -1/-1/-1->0->1 [20] -1/-1/-1->0->1 [21] -1/-1/-1->0->1 [22] -1/-1/-1->0->1 [23] -1/-1/-1->0->1
+ip-26-0-170-132:3330184:3330730 [1] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-162-14:2525075:2525612 [0] NCCL INFO Channel 22/0 : 0[0] -> 1[4] via P2P/IPC
+ip-26-0-170-132:3330185:3330727 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-170-132:3330185:3330727 [2] NCCL INFO Channel 00/24 :    0   1
+ip-26-0-170-132:3330189:3330728 [6] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] -1/-1/-1->1->0 [3] -1/-1/-1->1->0 [4] -1/-1/-1->1->0 [5] -1/-1/-1->1->0 [6] 0/-1/-1->1->-1 [7] 0/-1/-1->1->-1 [8] 0/-1/-1->1->-1 [9] 0/-1/-1->1->-1 [10] 0/-1/-1->1->-1 [11] 0/-1/-1->1->-1 [12] -1/-1/-1->1->0 [13] -1/-1/-1->1->0 [14] -1/-1/-1->1->0 [15] -1/-1/-1->1->0 [16] -1/-1/-1->1->0 [17] -1/-1/-1->1->0 [18] 0/-1/-1->1->-1 [19] 0/-1/-1->1->-1 [20] 0/-1/-1->1->-1 [21] 0/-1/-1->1->-1 [22] 0/-1/-1->1->-1 [23] 0/-1/-1->1->-1
+ip-26-0-170-132:3330185:3330727 [2] NCCL INFO Channel 01/24 :    0   1
+ip-26-0-170-132:3330185:3330727 [2] NCCL INFO Channel 02/24 :    0   1
+ip-26-0-170-132:3330189:3330728 [6] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-170-132:3330185:3330727 [2] NCCL INFO Channel 03/24 :    0   1
+ip-26-0-170-132:3330185:3330727 [2] NCCL INFO Channel 04/24 :    0   1
+ip-26-0-170-132:3330185:3330727 [2] NCCL INFO Channel 05/24 :    0   1
+ip-26-0-170-132:3330185:3330727 [2] NCCL INFO Channel 06/24 :    0   1
+ip-26-0-170-132:3330185:3330727 [2] NCCL INFO Channel 07/24 :    0   1
+ip-26-0-170-132:3330185:3330727 [2] NCCL INFO Channel 08/24 :    0   1
+ip-26-0-170-132:3330185:3330727 [2] NCCL INFO Channel 09/24 :    0   1
+ip-26-0-170-132:3330185:3330727 [2] NCCL INFO Channel 10/24 :    0   1
+ip-26-0-170-132:3330185:3330727 [2] NCCL INFO Channel 11/24 :    0   1
+ip-26-0-170-132:3330185:3330727 [2] NCCL INFO Channel 12/24 :    0   1
+ip-26-0-170-132:3330185:3330727 [2] NCCL INFO Channel 13/24 :    0   1
+ip-26-0-170-132:3330185:3330727 [2] NCCL INFO Channel 14/24 :    0   1
+ip-26-0-170-132:3330185:3330727 [2] NCCL INFO Channel 15/24 :    0   1
+ip-26-0-170-132:3330185:3330727 [2] NCCL INFO Channel 16/24 :    0   1
+ip-26-0-170-132:3330185:3330727 [2] NCCL INFO Channel 17/24 :    0   1
+ip-26-0-170-132:3330185:3330727 [2] NCCL INFO Channel 18/24 :    0   1
+ip-26-0-170-132:3330185:3330727 [2] NCCL INFO Channel 19/24 :    0   1
+ip-26-0-170-132:3330185:3330727 [2] NCCL INFO Channel 20/24 :    0   1
+ip-26-0-170-132:3330185:3330727 [2] NCCL INFO Channel 21/24 :    0   1
+ip-26-0-170-132:3330185:3330727 [2] NCCL INFO Channel 22/24 :    0   1
+ip-26-0-170-132:3330185:3330727 [2] NCCL INFO Channel 23/24 :    0   1
+ip-26-0-170-132:3330185:3330727 [2] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] -1/-1/-1->0->1 [7] -1/-1/-1->0->1 [8] -1/-1/-1->0->1 [9] -1/-1/-1->0->1 [10] -1/-1/-1->0->1 [11] -1/-1/-1->0->1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] -1/-1/-1->0->1 [19] -1/-1/-1->0->1 [20] -1/-1/-1->0->1 [21] -1/-1/-1->0->1 [22] -1/-1/-1->0->1 [23] -1/-1/-1->0->1
+ip-26-0-170-132:3330185:3330727 [2] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-162-14:2525075:2525612 [0] NCCL INFO Channel 23/0 : 0[0] -> 1[4] via P2P/IPC
+ip-26-0-163-58:934721:935254 [1] NCCL INFO Channel 22/0 : 0[1] -> 1[5] via P2P/IPC
+ip-26-0-163-58:934726:935249 [6] NCCL INFO Channel 17/0 : 1[6] -> 0[2] via P2P/IPC
+ip-26-0-163-58:934721:935254 [1] NCCL INFO Channel 23/0 : 0[1] -> 1[5] via P2P/IPC
+ip-26-0-163-58:934726:935249 [6] NCCL INFO Channel 18/0 : 1[6] -> 0[2] via P2P/IPC
+ip-26-0-170-132:3330190:3330726 [7] NCCL INFO Channel 00/0 : 1[7] -> 0[3] via P2P/IPC
+ip-26-0-170-132:3330183:3330721 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[4] via P2P/IPC
+ip-26-0-170-132:3330186:3330725 [3] NCCL INFO Channel 00/0 : 0[3] -> 1[7] via P2P/IPC
+ip-26-0-170-132:3330190:3330726 [7] NCCL INFO Channel 01/0 : 1[7] -> 0[3] via P2P/IPC
+ip-26-0-170-132:3330183:3330721 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[4] via P2P/IPC
+ip-26-0-170-132:3330190:3330726 [7] NCCL INFO Channel 02/0 : 1[7] -> 0[3] via P2P/IPC
+ip-26-0-170-132:3330186:3330725 [3] NCCL INFO Channel 01/0 : 0[3] -> 1[7] via P2P/IPC
+ip-26-0-170-132:3330183:3330721 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[4] via P2P/IPC
+ip-26-0-170-132:3330190:3330726 [7] NCCL INFO Channel 03/0 : 1[7] -> 0[3] via P2P/IPC
+ip-26-0-170-132:3330186:3330725 [3] NCCL INFO Channel 02/0 : 0[3] -> 1[7] via P2P/IPC
+ip-26-0-170-132:3330183:3330721 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[4] via P2P/IPC
+ip-26-0-170-132:3330190:3330726 [7] NCCL INFO Channel 04/0 : 1[7] -> 0[3] via P2P/IPC
+ip-26-0-170-132:3330186:3330725 [3] NCCL INFO Channel 03/0 : 0[3] -> 1[7] via P2P/IPC
+ip-26-0-170-132:3330190:3330726 [7] NCCL INFO Channel 05/0 : 1[7] -> 0[3] via P2P/IPC
+ip-26-0-170-132:3330186:3330725 [3] NCCL INFO Channel 04/0 : 0[3] -> 1[7] via P2P/IPC
+ip-26-0-163-58:934726:935249 [6] NCCL INFO Channel 19/0 : 1[6] -> 0[2] via P2P/IPC
+ip-26-0-170-132:3330190:3330726 [7] NCCL INFO Channel 06/0 : 1[7] -> 0[3] via P2P/IPC
+ip-26-0-170-132:3330186:3330725 [3] NCCL INFO Channel 05/0 : 0[3] -> 1[7] via P2P/IPC
+ip-26-0-170-132:3330190:3330726 [7] NCCL INFO Channel 07/0 : 1[7] -> 0[3] via P2P/IPC
+ip-26-0-170-132:3330186:3330725 [3] NCCL INFO Channel 06/0 : 0[3] -> 1[7] via P2P/IPC
+ip-26-0-170-132:3330184:3330730 [1] NCCL INFO Channel 00/0 : 0[1] -> 1[5] via P2P/IPC
+ip-26-0-170-132:3330183:3330721 [0] NCCL INFO Channel 04/0 : 0[0] -> 1[4] via P2P/IPC
+ip-26-0-170-132:3330187:3330722 [4] NCCL INFO Channel 00/0 : 1[4] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330190:3330726 [7] NCCL INFO Channel 08/0 : 1[7] -> 0[3] via P2P/IPC
+ip-26-0-170-132:3330186:3330725 [3] NCCL INFO Channel 07/0 : 0[3] -> 1[7] via P2P/IPC
+ip-26-0-170-132:3330184:3330730 [1] NCCL INFO Channel 01/0 : 0[1] -> 1[5] via P2P/IPC
+ip-26-0-170-132:3330183:3330721 [0] NCCL INFO Channel 05/0 : 0[0] -> 1[4] via P2P/IPC
+ip-26-0-170-132:3330187:3330722 [4] NCCL INFO Channel 01/0 : 1[4] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330190:3330726 [7] NCCL INFO Channel 09/0 : 1[7] -> 0[3] via P2P/IPC
+ip-26-0-170-132:3330186:3330725 [3] NCCL INFO Channel 08/0 : 0[3] -> 1[7] via P2P/IPC
+ip-26-0-170-132:3330184:3330730 [1] NCCL INFO Channel 02/0 : 0[1] -> 1[5] via P2P/IPC
+ip-26-0-170-132:3330187:3330722 [4] NCCL INFO Channel 02/0 : 1[4] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330183:3330721 [0] NCCL INFO Channel 06/0 : 0[0] -> 1[4] via P2P/IPC
+ip-26-0-170-132:3330187:3330722 [4] NCCL INFO Channel 03/0 : 1[4] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330183:3330721 [0] NCCL INFO Channel 07/0 : 0[0] -> 1[4] via P2P/IPC
+ip-26-0-170-132:3330184:3330730 [1] NCCL INFO Channel 03/0 : 0[1] -> 1[5] via P2P/IPC
+ip-26-0-170-132:3330183:3330721 [0] NCCL INFO Channel 08/0 : 0[0] -> 1[4] via P2P/IPC
+ip-26-0-170-132:3330187:3330722 [4] NCCL INFO Channel 04/0 : 1[4] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330184:3330730 [1] NCCL INFO Channel 04/0 : 0[1] -> 1[5] via P2P/IPC
+ip-26-0-170-132:3330190:3330726 [7] NCCL INFO Channel 10/0 : 1[7] -> 0[3] via P2P/IPC
+ip-26-0-170-132:3330183:3330721 [0] NCCL INFO Channel 09/0 : 0[0] -> 1[4] via P2P/IPC
+ip-26-0-170-132:3330184:3330730 [1] NCCL INFO Channel 05/0 : 0[1] -> 1[5] via P2P/IPC
+ip-26-0-170-132:3330187:3330722 [4] NCCL INFO Channel 05/0 : 1[4] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330186:3330725 [3] NCCL INFO Channel 09/0 : 0[3] -> 1[7] via P2P/IPC
+ip-26-0-170-132:3330190:3330726 [7] NCCL INFO Channel 11/0 : 1[7] -> 0[3] via P2P/IPC
+ip-26-0-170-132:3330183:3330721 [0] NCCL INFO Channel 10/0 : 0[0] -> 1[4] via P2P/IPC
+ip-26-0-170-132:3330184:3330730 [1] NCCL INFO Channel 06/0 : 0[1] -> 1[5] via P2P/IPC
+ip-26-0-170-132:3330187:3330722 [4] NCCL INFO Channel 06/0 : 1[4] -> 0[0] via P2P/IPC
+ip-26-0-163-58:934726:935249 [6] NCCL INFO Channel 20/0 : 1[6] -> 0[2] via P2P/IPC
+ip-26-0-170-132:3330189:3330728 [6] NCCL INFO Channel 00/0 : 1[6] -> 0[2] via P2P/IPC
+ip-26-0-170-132:3330186:3330725 [3] NCCL INFO Channel 10/0 : 0[3] -> 1[7] via P2P/IPC
+ip-26-0-170-132:3330190:3330726 [7] NCCL INFO Channel 12/0 : 1[7] -> 0[3] via P2P/IPC
+ip-26-0-170-132:3330183:3330721 [0] NCCL INFO Channel 11/0 : 0[0] -> 1[4] via P2P/IPC
+ip-26-0-170-132:3330184:3330730 [1] NCCL INFO Channel 07/0 : 0[1] -> 1[5] via P2P/IPC
+ip-26-0-170-132:3330187:3330722 [4] NCCL INFO Channel 07/0 : 1[4] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330189:3330728 [6] NCCL INFO Channel 01/0 : 1[6] -> 0[2] via P2P/IPC
+ip-26-0-170-132:3330188:3330731 [5] NCCL INFO Channel 00/0 : 1[5] -> 0[1] via P2P/IPC
+ip-26-0-170-132:3330190:3330726 [7] NCCL INFO Channel 13/0 : 1[7] -> 0[3] via P2P/IPC
+ip-26-0-170-132:3330183:3330721 [0] NCCL INFO Channel 12/0 : 0[0] -> 1[4] via P2P/IPC
+ip-26-0-170-132:3330184:3330730 [1] NCCL INFO Channel 08/0 : 0[1] -> 1[5] via P2P/IPC
+ip-26-0-170-132:3330187:3330722 [4] NCCL INFO Channel 08/0 : 1[4] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330189:3330728 [6] NCCL INFO Channel 02/0 : 1[6] -> 0[2] via P2P/IPC
+ip-26-0-170-132:3330188:3330731 [5] NCCL INFO Channel 01/0 : 1[5] -> 0[1] via P2P/IPC
+ip-26-0-170-132:3330186:3330725 [3] NCCL INFO Channel 11/0 : 0[3] -> 1[7] via P2P/IPC
+ip-26-0-170-132:3330190:3330726 [7] NCCL INFO Channel 14/0 : 1[7] -> 0[3] via P2P/IPC
+ip-26-0-170-132:3330185:3330727 [2] NCCL INFO Channel 00/0 : 0[2] -> 1[6] via P2P/IPC
+ip-26-0-170-132:3330183:3330721 [0] NCCL INFO Channel 13/0 : 0[0] -> 1[4] via P2P/IPC
+ip-26-0-170-132:3330184:3330730 [1] NCCL INFO Channel 09/0 : 0[1] -> 1[5] via P2P/IPC
+ip-26-0-170-132:3330187:3330722 [4] NCCL INFO Channel 09/0 : 1[4] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330189:3330728 [6] NCCL INFO Channel 03/0 : 1[6] -> 0[2] via P2P/IPC
+ip-26-0-163-58:934726:935249 [6] NCCL INFO Channel 21/0 : 1[6] -> 0[2] via P2P/IPC
+ip-26-0-170-132:3330188:3330731 [5] NCCL INFO Channel 02/0 : 1[5] -> 0[1] via P2P/IPC
+ip-26-0-170-132:3330190:3330726 [7] NCCL INFO Channel 15/0 : 1[7] -> 0[3] via P2P/IPC
+ip-26-0-170-132:3330186:3330725 [3] NCCL INFO Channel 12/0 : 0[3] -> 1[7] via P2P/IPC
+ip-26-0-170-132:3330183:3330721 [0] NCCL INFO Channel 14/0 : 0[0] -> 1[4] via P2P/IPC
+ip-26-0-170-132:3330185:3330727 [2] NCCL INFO Channel 01/0 : 0[2] -> 1[6] via P2P/IPC
+ip-26-0-170-132:3330187:3330722 [4] NCCL INFO Channel 10/0 : 1[4] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330189:3330728 [6] NCCL INFO Channel 04/0 : 1[6] -> 0[2] via P2P/IPC
+ip-26-0-170-132:3330188:3330731 [5] NCCL INFO Channel 03/0 : 1[5] -> 0[1] via P2P/IPC
+ip-26-0-170-132:3330183:3330721 [0] NCCL INFO Channel 15/0 : 0[0] -> 1[4] via P2P/IPC
+ip-26-0-170-132:3330185:3330727 [2] NCCL INFO Channel 02/0 : 0[2] -> 1[6] via P2P/IPC
+ip-26-0-170-132:3330187:3330722 [4] NCCL INFO Channel 11/0 : 1[4] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330189:3330728 [6] NCCL INFO Channel 05/0 : 1[6] -> 0[2] via P2P/IPC
+ip-26-0-170-132:3330188:3330731 [5] NCCL INFO Channel 04/0 : 1[5] -> 0[1] via P2P/IPC
+ip-26-0-170-132:3330184:3330730 [1] NCCL INFO Channel 10/0 : 0[1] -> 1[5] via P2P/IPC
+ip-26-0-170-132:3330186:3330725 [3] NCCL INFO Channel 13/0 : 0[3] -> 1[7] via P2P/IPC
+ip-26-0-170-132:3330189:3330728 [6] NCCL INFO Channel 06/0 : 1[6] -> 0[2] via P2P/IPC
+ip-26-0-170-132:3330185:3330727 [2] NCCL INFO Channel 03/0 : 0[2] -> 1[6] via P2P/IPC
+ip-26-0-170-132:3330183:3330721 [0] NCCL INFO Channel 16/0 : 0[0] -> 1[4] via P2P/IPC
+ip-26-0-170-132:3330187:3330722 [4] NCCL INFO Channel 12/0 : 1[4] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330190:3330726 [7] NCCL INFO Channel 16/0 : 1[7] -> 0[3] via P2P/IPC
+ip-26-0-170-132:3330188:3330731 [5] NCCL INFO Channel 05/0 : 1[5] -> 0[1] via P2P/IPC
+ip-26-0-170-132:3330184:3330730 [1] NCCL INFO Channel 11/0 : 0[1] -> 1[5] via P2P/IPC
+ip-26-0-170-132:3330186:3330725 [3] NCCL INFO Channel 14/0 : 0[3] -> 1[7] via P2P/IPC
+ip-26-0-170-132:3330185:3330727 [2] NCCL INFO Channel 04/0 : 0[2] -> 1[6] via P2P/IPC
+ip-26-0-170-132:3330183:3330721 [0] NCCL INFO Channel 17/0 : 0[0] -> 1[4] via P2P/IPC
+ip-26-0-170-132:3330187:3330722 [4] NCCL INFO Channel 13/0 : 1[4] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330190:3330726 [7] NCCL INFO Channel 17/0 : 1[7] -> 0[3] via P2P/IPC
+ip-26-0-170-132:3330189:3330728 [6] NCCL INFO Channel 07/0 : 1[6] -> 0[2] via P2P/IPC
+ip-26-0-170-132:3330188:3330731 [5] NCCL INFO Channel 06/0 : 1[5] -> 0[1] via P2P/IPC
+ip-26-0-170-132:3330184:3330730 [1] NCCL INFO Channel 12/0 : 0[1] -> 1[5] via P2P/IPC
+ip-26-0-170-132:3330185:3330727 [2] NCCL INFO Channel 05/0 : 0[2] -> 1[6] via P2P/IPC
+ip-26-0-170-132:3330183:3330721 [0] NCCL INFO Channel 18/0 : 0[0] -> 1[4] via P2P/IPC
+ip-26-0-170-132:3330190:3330726 [7] NCCL INFO Channel 18/0 : 1[7] -> 0[3] via P2P/IPC
+ip-26-0-170-132:3330187:3330722 [4] NCCL INFO Channel 14/0 : 1[4] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330189:3330728 [6] NCCL INFO Channel 08/0 : 1[6] -> 0[2] via P2P/IPC
+ip-26-0-170-132:3330188:3330731 [5] NCCL INFO Channel 07/0 : 1[5] -> 0[1] via P2P/IPC
+ip-26-0-170-132:3330184:3330730 [1] NCCL INFO Channel 13/0 : 0[1] -> 1[5] via P2P/IPC
+ip-26-0-170-132:3330185:3330727 [2] NCCL INFO Channel 06/0 : 0[2] -> 1[6] via P2P/IPC
+ip-26-0-170-132:3330190:3330726 [7] NCCL INFO Channel 19/0 : 1[7] -> 0[3] via P2P/IPC
+ip-26-0-170-132:3330183:3330721 [0] NCCL INFO Channel 19/0 : 0[0] -> 1[4] via P2P/IPC
+ip-26-0-170-132:3330187:3330722 [4] NCCL INFO Channel 15/0 : 1[4] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330189:3330728 [6] NCCL INFO Channel 09/0 : 1[6] -> 0[2] via P2P/IPC
+ip-26-0-163-58:934726:935249 [6] NCCL INFO Channel 22/0 : 1[6] -> 0[2] via P2P/IPC
+ip-26-0-170-132:3330188:3330731 [5] NCCL INFO Channel 08/0 : 1[5] -> 0[1] via P2P/IPC
+ip-26-0-170-132:3330184:3330730 [1] NCCL INFO Channel 14/0 : 0[1] -> 1[5] via P2P/IPC
+ip-26-0-170-132:3330185:3330727 [2] NCCL INFO Channel 07/0 : 0[2] -> 1[6] via P2P/IPC
+ip-26-0-170-132:3330190:3330726 [7] NCCL INFO Channel 20/0 : 1[7] -> 0[3] via P2P/IPC
+ip-26-0-170-132:3330183:3330721 [0] NCCL INFO Channel 20/0 : 0[0] -> 1[4] via P2P/IPC
+ip-26-0-170-132:3330187:3330722 [4] NCCL INFO Channel 16/0 : 1[4] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330189:3330728 [6] NCCL INFO Channel 10/0 : 1[6] -> 0[2] via P2P/IPC
+ip-26-0-170-132:3330188:3330731 [5] NCCL INFO Channel 09/0 : 1[5] -> 0[1] via P2P/IPC
+ip-26-0-170-132:3330184:3330730 [1] NCCL INFO Channel 15/0 : 0[1] -> 1[5] via P2P/IPC
+ip-26-0-170-132:3330186:3330725 [3] NCCL INFO Channel 15/0 : 0[3] -> 1[7] via P2P/IPC
+ip-26-0-170-132:3330185:3330727 [2] NCCL INFO Channel 08/0 : 0[2] -> 1[6] via P2P/IPC
+ip-26-0-170-132:3330187:3330722 [4] NCCL INFO Channel 17/0 : 1[4] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330190:3330726 [7] NCCL INFO Channel 21/0 : 1[7] -> 0[3] via P2P/IPC
+ip-26-0-170-132:3330183:3330721 [0] NCCL INFO Channel 21/0 : 0[0] -> 1[4] via P2P/IPC
+ip-26-0-170-132:3330189:3330728 [6] NCCL INFO Channel 11/0 : 1[6] -> 0[2] via P2P/IPC
+ip-26-0-170-132:3330184:3330730 [1] NCCL INFO Channel 16/0 : 0[1] -> 1[5] via P2P/IPC
+ip-26-0-170-132:3330188:3330731 [5] NCCL INFO Channel 10/0 : 1[5] -> 0[1] via P2P/IPC
+ip-26-0-170-132:3330186:3330725 [3] NCCL INFO Channel 16/0 : 0[3] -> 1[7] via P2P/IPC
+ip-26-0-170-132:3330185:3330727 [2] NCCL INFO Channel 09/0 : 0[2] -> 1[6] via P2P/IPC
+ip-26-0-170-132:3330187:3330722 [4] NCCL INFO Channel 18/0 : 1[4] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330190:3330726 [7] NCCL INFO Channel 22/0 : 1[7] -> 0[3] via P2P/IPC
+ip-26-0-170-132:3330183:3330721 [0] NCCL INFO Channel 22/0 : 0[0] -> 1[4] via P2P/IPC
+ip-26-0-170-132:3330189:3330728 [6] NCCL INFO Channel 12/0 : 1[6] -> 0[2] via P2P/IPC
+ip-26-0-170-132:3330188:3330731 [5] NCCL INFO Channel 11/0 : 1[5] -> 0[1] via P2P/IPC
+ip-26-0-170-132:3330186:3330725 [3] NCCL INFO Channel 17/0 : 0[3] -> 1[7] via P2P/IPC
+ip-26-0-170-132:3330184:3330730 [1] NCCL INFO Channel 17/0 : 0[1] -> 1[5] via P2P/IPC
+ip-26-0-170-132:3330185:3330727 [2] NCCL INFO Channel 10/0 : 0[2] -> 1[6] via P2P/IPC
+ip-26-0-170-132:3330187:3330722 [4] NCCL INFO Channel 19/0 : 1[4] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330190:3330726 [7] NCCL INFO Channel 23/0 : 1[7] -> 0[3] via P2P/IPC
+ip-26-0-170-132:3330183:3330721 [0] NCCL INFO Channel 23/0 : 0[0] -> 1[4] via P2P/IPC
+ip-26-0-170-132:3330189:3330728 [6] NCCL INFO Channel 13/0 : 1[6] -> 0[2] via P2P/IPC
+ip-26-0-170-132:3330186:3330725 [3] NCCL INFO Channel 18/0 : 0[3] -> 1[7] via P2P/IPC
+ip-26-0-163-58:934726:935249 [6] NCCL INFO Channel 23/0 : 1[6] -> 0[2] via P2P/IPC
+ip-26-0-170-132:3330184:3330730 [1] NCCL INFO Channel 18/0 : 0[1] -> 1[5] via P2P/IPC
+ip-26-0-170-132:3330188:3330731 [5] NCCL INFO Channel 12/0 : 1[5] -> 0[1] via P2P/IPC
+ip-26-0-170-132:3330185:3330727 [2] NCCL INFO Channel 11/0 : 0[2] -> 1[6] via P2P/IPC
+ip-26-0-170-132:3330187:3330722 [4] NCCL INFO Channel 20/0 : 1[4] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330189:3330728 [6] NCCL INFO Channel 14/0 : 1[6] -> 0[2] via P2P/IPC
+ip-26-0-170-132:3330186:3330725 [3] NCCL INFO Channel 19/0 : 0[3] -> 1[7] via P2P/IPC
+ip-26-0-170-132:3330184:3330730 [1] NCCL INFO Channel 19/0 : 0[1] -> 1[5] via P2P/IPC
+ip-26-0-170-132:3330188:3330731 [5] NCCL INFO Channel 13/0 : 1[5] -> 0[1] via P2P/IPC
+ip-26-0-170-132:3330187:3330722 [4] NCCL INFO Channel 21/0 : 1[4] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330185:3330727 [2] NCCL INFO Channel 12/0 : 0[2] -> 1[6] via P2P/IPC
+ip-26-0-170-132:3330186:3330725 [3] NCCL INFO Channel 20/0 : 0[3] -> 1[7] via P2P/IPC
+ip-26-0-170-132:3330189:3330728 [6] NCCL INFO Channel 15/0 : 1[6] -> 0[2] via P2P/IPC
+ip-26-0-170-132:3330184:3330730 [1] NCCL INFO Channel 20/0 : 0[1] -> 1[5] via P2P/IPC
+ip-26-0-170-132:3330188:3330731 [5] NCCL INFO Channel 14/0 : 1[5] -> 0[1] via P2P/IPC
+ip-26-0-170-132:3330187:3330722 [4] NCCL INFO Channel 22/0 : 1[4] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330185:3330727 [2] NCCL INFO Channel 13/0 : 0[2] -> 1[6] via P2P/IPC
+ip-26-0-170-132:3330186:3330725 [3] NCCL INFO Channel 21/0 : 0[3] -> 1[7] via P2P/IPC
+ip-26-0-170-132:3330189:3330728 [6] NCCL INFO Channel 16/0 : 1[6] -> 0[2] via P2P/IPC
+ip-26-0-170-132:3330184:3330730 [1] NCCL INFO Channel 21/0 : 0[1] -> 1[5] via P2P/IPC
+ip-26-0-170-132:3330188:3330731 [5] NCCL INFO Channel 15/0 : 1[5] -> 0[1] via P2P/IPC
+ip-26-0-170-132:3330186:3330725 [3] NCCL INFO Channel 22/0 : 0[3] -> 1[7] via P2P/IPC
+ip-26-0-170-132:3330184:3330730 [1] NCCL INFO Channel 22/0 : 0[1] -> 1[5] via P2P/IPC
+ip-26-0-170-132:3330189:3330728 [6] NCCL INFO Channel 17/0 : 1[6] -> 0[2] via P2P/IPC
+ip-26-0-170-132:3330185:3330727 [2] NCCL INFO Channel 14/0 : 0[2] -> 1[6] via P2P/IPC
+ip-26-0-170-132:3330188:3330731 [5] NCCL INFO Channel 16/0 : 1[5] -> 0[1] via P2P/IPC
+ip-26-0-170-132:3330187:3330722 [4] NCCL INFO Channel 23/0 : 1[4] -> 0[0] via P2P/IPC
+ip-26-0-170-132:3330184:3330730 [1] NCCL INFO Channel 23/0 : 0[1] -> 1[5] via P2P/IPC
+ip-26-0-170-132:3330186:3330725 [3] NCCL INFO Channel 23/0 : 0[3] -> 1[7] via P2P/IPC
+ip-26-0-170-132:3330189:3330728 [6] NCCL INFO Channel 18/0 : 1[6] -> 0[2] via P2P/IPC
+ip-26-0-170-132:3330188:3330731 [5] NCCL INFO Channel 17/0 : 1[5] -> 0[1] via P2P/IPC
+ip-26-0-170-132:3330189:3330728 [6] NCCL INFO Channel 19/0 : 1[6] -> 0[2] via P2P/IPC
+ip-26-0-170-132:3330188:3330731 [5] NCCL INFO Channel 18/0 : 1[5] -> 0[1] via P2P/IPC
+ip-26-0-170-132:3330185:3330727 [2] NCCL INFO Channel 15/0 : 0[2] -> 1[6] via P2P/IPC
+ip-26-0-170-132:3330189:3330728 [6] NCCL INFO Channel 20/0 : 1[6] -> 0[2] via P2P/IPC
+ip-26-0-170-132:3330188:3330731 [5] NCCL INFO Channel 19/0 : 1[5] -> 0[1] via P2P/IPC
+ip-26-0-170-132:3330185:3330727 [2] NCCL INFO Channel 16/0 : 0[2] -> 1[6] via P2P/IPC
+ip-26-0-170-132:3330188:3330731 [5] NCCL INFO Channel 20/0 : 1[5] -> 0[1] via P2P/IPC
+ip-26-0-170-132:3330185:3330727 [2] NCCL INFO Channel 17/0 : 0[2] -> 1[6] via P2P/IPC
+ip-26-0-170-132:3330189:3330728 [6] NCCL INFO Channel 21/0 : 1[6] -> 0[2] via P2P/IPC
+ip-26-0-170-132:3330188:3330731 [5] NCCL INFO Channel 21/0 : 1[5] -> 0[1] via P2P/IPC
+ip-26-0-170-132:3330185:3330727 [2] NCCL INFO Channel 18/0 : 0[2] -> 1[6] via P2P/IPC
+ip-26-0-170-132:3330189:3330728 [6] NCCL INFO Channel 22/0 : 1[6] -> 0[2] via P2P/IPC
+ip-26-0-170-132:3330188:3330731 [5] NCCL INFO Channel 22/0 : 1[5] -> 0[1] via P2P/IPC
+ip-26-0-170-132:3330185:3330727 [2] NCCL INFO Channel 19/0 : 0[2] -> 1[6] via P2P/IPC
+ip-26-0-170-132:3330189:3330728 [6] NCCL INFO Channel 23/0 : 1[6] -> 0[2] via P2P/IPC
+ip-26-0-170-132:3330188:3330731 [5] NCCL INFO Channel 23/0 : 1[5] -> 0[1] via P2P/IPC
+ip-26-0-170-132:3330185:3330727 [2] NCCL INFO Channel 20/0 : 0[2] -> 1[6] via P2P/IPC
+ip-26-0-170-132:3330185:3330727 [2] NCCL INFO Channel 21/0 : 0[2] -> 1[6] via P2P/IPC
+ip-26-0-170-132:3330185:3330727 [2] NCCL INFO Channel 22/0 : 0[2] -> 1[6] via P2P/IPC
+ip-26-0-170-132:3330185:3330727 [2] NCCL INFO Channel 23/0 : 0[2] -> 1[6] via P2P/IPC
+ip-26-0-170-31:3197180:3197820 [0] NCCL INFO Connected all rings
+ip-26-0-170-31:3197180:3197820 [0] NCCL INFO Connected all trees
+ip-26-0-170-31:3197180:3197820 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-170-31:3197180:3197820 [0] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-170-31:3197180:3197820 [0] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-170-31:3197184:3197821 [4] NCCL INFO Connected all rings
+ip-26-0-170-31:3197184:3197821 [4] NCCL INFO Connected all trees
+ip-26-0-170-31:3197184:3197821 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-170-31:3197184:3197821 [4] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-170-31:3197184:3197821 [4] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-170-31:3197184:3197821 [4] NCCL INFO comm 0xbedda70 rank 1 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0xa8f4448dd46d93b1 - Init COMPLETE
+ip-26-0-170-31:3197180:3197820 [0] NCCL INFO comm 0xbf79c50 rank 0 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0xa8f4448dd46d93b1 - Init COMPLETE
+ip-26-0-163-58:934724:935246 [4] NCCL INFO Connected all rings
+ip-26-0-163-58:934724:935246 [4] NCCL INFO Connected all trees
+ip-26-0-163-58:934724:935246 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-163-58:934724:935246 [4] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-163-58:934724:935246 [4] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-162-14:2525082:2525621 [7] NCCL INFO Connected all rings
+ip-26-0-162-14:2525082:2525621 [7] NCCL INFO Connected all trees
+ip-26-0-162-14:2525082:2525621 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-162-14:2525082:2525621 [7] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-162-14:2525082:2525621 [7] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-163-58:934720:935245 [0] NCCL INFO Connected all rings
+ip-26-0-163-58:934720:935245 [0] NCCL INFO Connected all trees
+ip-26-0-163-58:934720:935245 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-163-58:934720:935245 [0] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-163-58:934720:935245 [0] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-163-58:934727:935252 [7] NCCL INFO Connected all rings
+ip-26-0-163-58:934727:935252 [7] NCCL INFO Connected all trees
+ip-26-0-163-58:934727:935252 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-163-58:934727:935252 [7] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-163-58:934727:935252 [7] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-163-58:934724:935246 [4] NCCL INFO comm 0xb699a90 rank 1 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0x9171fa6542ce32fa - Init COMPLETE
+ip-26-0-163-58:934720:935245 [0] NCCL INFO comm 0xb5e0780 rank 0 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0x9171fa6542ce32fa - Init COMPLETE
+ip-26-0-170-31:3197181:3197829 [1] NCCL INFO Connected all rings
+ip-26-0-170-31:3197181:3197829 [1] NCCL INFO Connected all trees
+ip-26-0-170-31:3197181:3197829 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-170-31:3197181:3197829 [1] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-170-31:3197181:3197829 [1] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-162-14:2525078:2525618 [3] NCCL INFO Connected all rings
+ip-26-0-162-14:2525078:2525618 [3] NCCL INFO Connected all trees
+ip-26-0-162-14:2525078:2525618 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-162-14:2525078:2525618 [3] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-162-14:2525078:2525618 [3] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-170-31:3197185:3197830 [5] NCCL INFO Connected all rings
+ip-26-0-170-31:3197185:3197830 [5] NCCL INFO Connected all trees
+ip-26-0-170-31:3197185:3197830 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-170-31:3197185:3197830 [5] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-170-31:3197185:3197830 [5] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-162-14:2525076:2525615 [1] NCCL INFO Connected all rings
+ip-26-0-162-14:2525076:2525615 [1] NCCL INFO Connected all trees
+ip-26-0-162-14:2525076:2525615 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-162-14:2525076:2525615 [1] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-162-14:2525076:2525615 [1] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-162-14:2525080:2525616 [5] NCCL INFO Connected all rings
+ip-26-0-162-14:2525080:2525616 [5] NCCL INFO Connected all trees
+ip-26-0-162-14:2525080:2525616 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-162-14:2525080:2525616 [5] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-162-14:2525080:2525616 [5] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-162-14:2525082:2525621 [7] NCCL INFO comm 0xbe081b0 rank 1 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0x57a6b54ba58bf2d8 - Init COMPLETE
+ip-26-0-162-14:2525078:2525618 [3] NCCL INFO comm 0xb1bbb00 rank 0 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0x57a6b54ba58bf2d8 - Init COMPLETE
+ip-26-0-163-58:934723:935251 [3] NCCL INFO Connected all rings
+ip-26-0-163-58:934723:935251 [3] NCCL INFO Connected all trees
+ip-26-0-163-58:934723:935251 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-163-58:934723:935251 [3] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-163-58:934723:935251 [3] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-170-31:3197187:3197827 [7] NCCL INFO Connected all rings
+ip-26-0-170-31:3197187:3197827 [7] NCCL INFO Connected all trees
+ip-26-0-170-31:3197187:3197827 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-170-31:3197187:3197827 [7] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-170-31:3197187:3197827 [7] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-170-31:3197181:3197829 [1] NCCL INFO comm 0xb9a5230 rank 0 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0x8915cf04ef157888 - Init COMPLETE
+ip-26-0-170-31:3197185:3197830 [5] NCCL INFO comm 0xa627ed0 rank 1 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0x8915cf04ef157888 - Init COMPLETE
+ip-26-0-170-31:3197183:3197826 [3] NCCL INFO Connected all rings
+ip-26-0-170-31:3197183:3197826 [3] NCCL INFO Connected all trees
+ip-26-0-170-31:3197183:3197826 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-170-31:3197183:3197826 [3] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-170-31:3197183:3197826 [3] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-170-31:3197186:3197824 [6] NCCL INFO Connected all rings
+ip-26-0-170-31:3197186:3197824 [6] NCCL INFO Connected all trees
+ip-26-0-170-31:3197186:3197824 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-170-31:3197186:3197824 [6] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-170-31:3197186:3197824 [6] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-170-31:3197182:3197823 [2] NCCL INFO Connected all rings
+ip-26-0-170-31:3197182:3197823 [2] NCCL INFO Connected all trees
+ip-26-0-170-31:3197182:3197823 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-170-31:3197182:3197823 [2] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-170-31:3197182:3197823 [2] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-170-31:3197187:3197827 [7] NCCL INFO comm 0xc4401d0 rank 1 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0xce3d779e80e630f - Init COMPLETE
+ip-26-0-170-31:3197183:3197826 [3] NCCL INFO comm 0xbd49970 rank 0 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0xce3d779e80e630f - Init COMPLETE
+ip-26-0-162-14:2525076:2525615 [1] NCCL INFO comm 0xc303810 rank 0 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0xc23f2d54f1a22bd4 - Init COMPLETE
+ip-26-0-162-14:2525080:2525616 [5] NCCL INFO comm 0xbbdcd90 rank 1 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0xc23f2d54f1a22bd4 - Init COMPLETE
+ip-26-0-163-58:934721:935254 [1] NCCL INFO Connected all rings
+ip-26-0-163-58:934721:935254 [1] NCCL INFO Connected all trees
+ip-26-0-163-58:934721:935254 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-163-58:934721:935254 [1] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-163-58:934721:935254 [1] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-170-31:3197186:3197824 [6] NCCL INFO comm 0xbe11790 rank 1 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0x453cb490910f7b74 - Init COMPLETE
+ip-26-0-170-31:3197182:3197823 [2] NCCL INFO comm 0xa6da550 rank 0 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0x453cb490910f7b74 - Init COMPLETE
+ip-26-0-162-14:2525081:2525622 [6] NCCL INFO Connected all rings
+ip-26-0-162-14:2525081:2525622 [6] NCCL INFO Connected all trees
+ip-26-0-162-14:2525081:2525622 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-162-14:2525081:2525622 [6] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-162-14:2525081:2525622 [6] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-162-14:2525079:2525613 [4] NCCL INFO Connected all rings
+ip-26-0-162-14:2525079:2525613 [4] NCCL INFO Connected all trees
+ip-26-0-162-14:2525079:2525613 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-162-14:2525079:2525613 [4] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-162-14:2525079:2525613 [4] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-163-58:934725:935255 [5] NCCL INFO Connected all rings
+ip-26-0-163-58:934725:935255 [5] NCCL INFO Connected all trees
+ip-26-0-163-58:934725:935255 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-163-58:934725:935255 [5] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-163-58:934725:935255 [5] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-163-58:934727:935252 [7] NCCL INFO comm 0xb7f9b50 rank 1 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0x9d71cac1fadb657d - Init COMPLETE
+ip-26-0-163-58:934723:935251 [3] NCCL INFO comm 0xb8a40e0 rank 0 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0x9d71cac1fadb657d - Init COMPLETE
+ip-26-0-162-14:2525077:2525620 [2] NCCL INFO Connected all rings
+ip-26-0-162-14:2525077:2525620 [2] NCCL INFO Connected all trees
+ip-26-0-162-14:2525077:2525620 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-162-14:2525077:2525620 [2] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-162-14:2525077:2525620 [2] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-162-14:2525075:2525612 [0] NCCL INFO Connected all rings
+ip-26-0-162-14:2525075:2525612 [0] NCCL INFO Connected all trees
+ip-26-0-162-14:2525075:2525612 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-162-14:2525075:2525612 [0] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-162-14:2525075:2525612 [0] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-163-58:934721:935254 [1] NCCL INFO comm 0xc323ca0 rank 0 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0xb44c20b3113eb843 - Init COMPLETE
+ip-26-0-163-58:934725:935255 [5] NCCL INFO comm 0xa885490 rank 1 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0xb44c20b3113eb843 - Init COMPLETE
+ip-26-0-162-14:2525081:2525622 [6] NCCL INFO comm 0xb48b4d0 rank 1 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0x603d878f636bb0f4 - Init COMPLETE
+ip-26-0-162-14:2525077:2525620 [2] NCCL INFO comm 0xc14fe00 rank 0 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0x603d878f636bb0f4 - Init COMPLETE
+ip-26-0-162-14:2525079:2525613 [4] NCCL INFO comm 0xb658f10 rank 1 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0x228c3daf39bf52b8 - Init COMPLETE
+ip-26-0-162-14:2525075:2525612 [0] NCCL INFO comm 0xb80ecd0 rank 0 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0x228c3daf39bf52b8 - Init COMPLETE
+ip-26-0-163-58:934726:935249 [6] NCCL INFO Connected all rings
+ip-26-0-163-58:934726:935249 [6] NCCL INFO Connected all trees
+ip-26-0-163-58:934726:935249 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-163-58:934726:935249 [6] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-163-58:934726:935249 [6] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-163-58:934722:935248 [2] NCCL INFO Connected all rings
+ip-26-0-163-58:934722:935248 [2] NCCL INFO Connected all trees
+ip-26-0-163-58:934722:935248 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-163-58:934722:935248 [2] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-163-58:934722:935248 [2] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-163-58:934726:935249 [6] NCCL INFO comm 0xbbea240 rank 1 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0x1b908c1f11d5daeb - Init COMPLETE
+ip-26-0-163-58:934722:935248 [2] NCCL INFO comm 0xa8a5a70 rank 0 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0x1b908c1f11d5daeb - Init COMPLETE
+ip-26-0-170-132:3330186:3330725 [3] NCCL INFO Connected all rings
+ip-26-0-170-132:3330186:3330725 [3] NCCL INFO Connected all trees
+ip-26-0-170-132:3330186:3330725 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-170-132:3330186:3330725 [3] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-170-132:3330186:3330725 [3] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-170-132:3330190:3330726 [7] NCCL INFO Connected all rings
+ip-26-0-170-132:3330190:3330726 [7] NCCL INFO Connected all trees
+ip-26-0-170-132:3330190:3330726 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-170-132:3330190:3330726 [7] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-170-132:3330190:3330726 [7] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-170-132:3330187:3330722 [4] NCCL INFO Connected all rings
+ip-26-0-170-132:3330187:3330722 [4] NCCL INFO Connected all trees
+ip-26-0-170-132:3330187:3330722 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-170-132:3330187:3330722 [4] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-170-132:3330187:3330722 [4] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+12/28/2024 01:36:44 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]: [Optimizer Building] Using LearningRateForSP as learning rate
+ip-26-0-170-132:3330190:3330726 [7] NCCL INFO comm 0xc2f4190 rank 1 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0x7fd882a83b32ac48 - Init COMPLETE
+12/28/2024 01:36:44 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]: [ZeRO sharding] Size of optimizer params per rank:
+ip-26-0-170-132:3330186:3330725 [3] NCCL INFO comm 0xacef840 rank 0 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0x7fd882a83b32ac48 - Init COMPLETE
+12/28/2024 01:36:44 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]: [ZeRO sharding] DP Rank 0 has 15G out of 30.1G (50.00%) params' optimizer states
+12/28/2024 01:36:44 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-14]: [ZeRO sharding] DP Rank 1 has 15G out of 30.1G (50.00%) params' optimizer states
+ip-26-0-170-132:3330183:3330721 [0] NCCL INFO Connected all rings
+ip-26-0-170-132:3330183:3330721 [0] NCCL INFO Connected all trees
+ip-26-0-170-132:3330183:3330721 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-170-132:3330183:3330721 [0] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-170-132:3330183:3330721 [0] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 383, in init_optimizer_and_grad_accumulator
+    optimizer = ZeroDistributedOptimizer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/zero.py", line 91, in __init__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    optimizer = optimizer_builder(param_groups_in_rank)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 383, in init_optimizer_and_grad_accumulator
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    optimizer = ZeroDistributedOptimizer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/zero.py", line 91, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    optimizer = optimizer_builder(param_groups_in_rank)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+ip-26-0-170-132:3330187:3330722 [4] NCCL INFO comm 0xb8116d0 rank 1 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0x4da867b891ba2670 - Init COMPLETE
+    result = OptimizerFromGradientAccumulator(
+ip-26-0-170-132:3330183:3330721 [0] NCCL INFO comm 0xb2af600 rank 0 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0x4da867b891ba2670 - Init COMPLETE
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 108.57 GiB. GPU 0 has a total capacty of 79.33 GiB of which 13.61 GiB is free. Including non-PyTorch memory, this process has 65.71 GiB memory in use. Of the allocated memory 54.28 GiB is allocated by PyTorch, and 961.99 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 108.57 GiB. GPU 3 has a total capacty of 79.33 GiB of which 13.64 GiB is free. Including non-PyTorch memory, this process has 65.67 GiB memory in use. Of the allocated memory 54.28 GiB is allocated by PyTorch, and 961.99 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 383, in init_optimizer_and_grad_accumulator
+    optimizer = ZeroDistributedOptimizer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/zero.py", line 91, in __init__
+    optimizer = optimizer_builder(param_groups_in_rank)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+ip-26-0-170-132:3330185:3330727 [2] NCCL INFO Connected all rings
+ip-26-0-170-132:3330185:3330727 [2] NCCL INFO Connected all trees
+ip-26-0-170-132:3330185:3330727 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-170-132:3330185:3330727 [2] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-170-132:3330185:3330727 [2] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 108.57 GiB. GPU 2 has a total capacty of 79.33 GiB of which 13.53 GiB is free. Including non-PyTorch memory, this process has 65.79 GiB memory in use. Of the allocated memory 54.28 GiB is allocated by PyTorch, and 961.99 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 383, in init_optimizer_and_grad_accumulator
+    optimizer = ZeroDistributedOptimizer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/zero.py", line 91, in __init__
+    optimizer = optimizer_builder(param_groups_in_rank)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 108.57 GiB. GPU 1 has a total capacty of 79.33 GiB of which 13.53 GiB is free. Including non-PyTorch memory, this process has 65.79 GiB memory in use. Of the allocated memory 54.28 GiB is allocated by PyTorch, and 961.99 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 383, in init_optimizer_and_grad_accumulator
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    optimizer = ZeroDistributedOptimizer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/zero.py", line 91, in __init__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 383, in init_optimizer_and_grad_accumulator
+    optimizer = optimizer_builder(param_groups_in_rank)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    optimizer = ZeroDistributedOptimizer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/zero.py", line 91, in __init__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    optimizer = optimizer_builder(param_groups_in_rank)    
+self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 383, in init_optimizer_and_grad_accumulator
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 383, in init_optimizer_and_grad_accumulator
+    optimizer = ZeroDistributedOptimizer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/zero.py", line 91, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    optimizer = ZeroDistributedOptimizer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/zero.py", line 91, in __init__
+    optimizer = optimizer_builder(param_groups_in_rank)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    optimizer = optimizer_builder(param_groups_in_rank)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 108.57 GiB. GPU 7 has a total capacty of 79.33 GiB of which 13.96 GiB is free. Including non-PyTorch memory, this process has 65.36 GiB memory in use. Of the allocated memory 54.28 GiB is allocated by PyTorch, and 961.99 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 108.57 GiB. GPU 6 has a total capacty of 79.33 GiB of which 13.53 GiB is free. Including non-PyTorch memory, this process has 65.79 GiB memory in use. Of the allocated memory 54.28 GiB is allocated by PyTorch, and 961.99 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 108.57 GiB. GPU 4 has a total capacty of 79.33 GiB of which 13.56 GiB is free. Including non-PyTorch memory, this process has 65.76 GiB memory in use. Of the allocated memory 54.28 GiB is allocated by PyTorch, and 961.99 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 108.57 GiB. GPU 5 has a total capacty of 79.33 GiB of which 13.53 GiB is free. Including non-PyTorch memory, this process has 65.79 GiB memory in use. Of the allocated memory 54.28 GiB is allocated by PyTorch, and 961.99 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+ip-26-0-170-132:3330188:3330731 [5] NCCL INFO Connected all rings
+ip-26-0-170-132:3330188:3330731 [5] NCCL INFO Connected all trees
+ip-26-0-170-132:3330188:3330731 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-170-132:3330188:3330731 [5] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-170-132:3330188:3330731 [5] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-170-132:3330189:3330728 [6] NCCL INFO Connected all rings
+ip-26-0-170-132:3330189:3330728 [6] NCCL INFO Connected all trees
+ip-26-0-170-132:3330189:3330728 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-170-132:3330189:3330728 [6] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-170-132:3330189:3330728 [6] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-170-132:3330184:3330730 [1] NCCL INFO Connected all rings
+ip-26-0-170-132:3330184:3330730 [1] NCCL INFO Connected all trees
+ip-26-0-170-132:3330184:3330730 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-170-132:3330184:3330730 [1] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-170-132:3330184:3330730 [1] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-170-132:3330185:3330727 [2] NCCL INFO comm 0xb879ad0 rank 0 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0xbb4dd0623a3af179 - Init COMPLETE
+ip-26-0-170-132:3330189:3330728 [6] NCCL INFO comm 0xa5862f0 rank 1 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0xbb4dd0623a3af179 - Init COMPLETE
+ip-26-0-170-132:3330184:3330730 [1] NCCL INFO comm 0xb3e6490 rank 0 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0x7a9ebadf14aae2f9 - Init COMPLETE
+ip-26-0-170-132:3330188:3330731 [5] NCCL INFO comm 0xc2c69b0 rank 1 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0x7a9ebadf14aae2f9 - Init COMPLETE
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 383, in init_optimizer_and_grad_accumulator
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 383, in init_optimizer_and_grad_accumulator
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    optimizer = ZeroDistributedOptimizer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/zero.py", line 91, in __init__
+    optimizer = ZeroDistributedOptimizer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/zero.py", line 91, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 383, in init_optimizer_and_grad_accumulator
+    optimizer = optimizer_builder(param_groups_in_rank)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    optimizer = optimizer_builder(param_groups_in_rank)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    optimizer = ZeroDistributedOptimizer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/zero.py", line 91, in __init__
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    optimizer = optimizer_builder(param_groups_in_rank)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 112.00 GiB. GPU 1 has a total capacty of 79.33 GiB of which 11.81 GiB is free. Including non-PyTorch memory, this process has 67.51 GiB memory in use. Of the allocated memory 56.00 GiB is allocated by PyTorch, and 961.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONFtorch.cuda
+.OutOfMemoryError: CUDA out of memory. Tried to allocate 112.00 GiB. GPU 5 has a total capacty of 79.33 GiB of which 11.81 GiB is free. Including non-PyTorch memory, this process has 67.51 GiB memory in use. Of the allocated memory 56.00 GiB is allocated by PyTorch, and 961.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 112.00 GiB. GPU 2 has a total capacty of 79.33 GiB of which 11.81 GiB is free. Including non-PyTorch memory, this process has 67.51 GiB memory in use. Of the allocated memory 56.00 GiB is allocated by PyTorch, and 961.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 383, in init_optimizer_and_grad_accumulator
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    trainer = DistributedTrainer(config_file)
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 383, in init_optimizer_and_grad_accumulator
+    optimizer = ZeroDistributedOptimizer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/zero.py", line 91, in __init__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    optimizer = ZeroDistributedOptimizer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/zero.py", line 91, in __init__
+    optimizer = optimizer_builder(param_groups_in_rank)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 383, in init_optimizer_and_grad_accumulator
+    optimizer = optimizer_builder(param_groups_in_rank)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    trainer = DistributedTrainer(config_file)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 383, in init_optimizer_and_grad_accumulator
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 383, in init_optimizer_and_grad_accumulator
+    optimizer = ZeroDistributedOptimizer(
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    optimizer = ZeroDistributedOptimizer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/zero.py", line 91, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/zero.py", line 91, in __init__
+    optimizer = ZeroDistributedOptimizer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/zero.py", line 91, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    optimizer = optimizer_builder(param_groups_in_rank)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    optimizer = optimizer_builder(param_groups_in_rank)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    optimizer = optimizer_builder(param_groups_in_rank)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 112.00 GiB. GPU 4 has a total capacty of 79.33 GiB of which 11.84 GiB is free. Including non-PyTorch memory, this process has 67.48 GiB memory in use. Of the allocated memory 56.00 GiB is allocated by PyTorch, and 961.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 112.00 GiB. GPU 6 has a total capacty of 79.33 GiB of which 11.81 GiB is free. Including non-PyTorch memory, this process has 67.51 GiB memory in use. Of the allocated memory 56.00 GiB is allocated by PyTorch, and 961.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 110.00 GiB. GPU 0 has a total capacty of 79.33 GiB of which 12.09 GiB is free. Including non-PyTorch memory, this process has 67.23 GiB memory in use. Of the allocated memory 55.00 GiB is allocated by PyTorch, and 1.03 GiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 112.00 GiB. GPU 0 has a total capacty of 79.33 GiB of which 11.89 GiB is free. Including non-PyTorch memory, this process has 67.43 GiB memory in use. Of the allocated memory 56.00 GiB is allocated by PyTorch, and 961.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 112.00 GiB. GPU 7 has a total capacty of 79.33 GiB of which 12.14 GiB is free. Including non-PyTorch memory, this process has 67.17 GiB memory in use. Of the allocated memory 56.00 GiB is allocated by PyTorch, and 1.03 GiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 383, in init_optimizer_and_grad_accumulator
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    optimizer = ZeroDistributedOptimizer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/zero.py", line 91, in __init__
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    optimizer = optimizer_builder(param_groups_in_rank)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 383, in init_optimizer_and_grad_accumulator
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    optimizer = ZeroDistributedOptimizer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/zero.py", line 91, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    optimizer = optimizer_builder(param_groups_in_rank)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 383, in init_optimizer_and_grad_accumulator
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 110.00 GiB. GPU 2 has a total capacty of 79.33 GiB of which 12.01 GiB is free. Including non-PyTorch memory, this process has 67.31 GiB memory in use. Of the allocated memory 55.00 GiB is allocated by PyTorch, and 1.03 GiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    optimizer = ZeroDistributedOptimizer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/zero.py", line 91, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    optimizer = optimizer_builder(param_groups_in_rank)
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 112.00 GiB. GPU 3 has a total capacty of 79.33 GiB of which 11.93 GiB is free. Including non-PyTorch memory, this process has 67.39 GiB memory in use. Of the allocated memory 56.00 GiB is allocated by PyTorch, and 961.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 383, in init_optimizer_and_grad_accumulator
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    optimizer = ZeroDistributedOptimizer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/zero.py", line 91, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    optimizer = optimizer_builder(param_groups_in_rank)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 110.00 GiB. GPU 1 has a total capacty of 79.33 GiB of which 12.09 GiB is free. Including non-PyTorch memory, this process has 67.23 GiB memory in use. Of the allocated memory 55.00 GiB is allocated by PyTorch, and 1.03 GiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    trainer = DistributedTrainer(config_file)
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 110.00 GiB. GPU 5 has a total capacty of 79.33 GiB of which 12.09 GiB is free. Including non-PyTorch memory, this process has 67.23 GiB memory in use. Of the allocated memory 55.00 GiB is allocated by PyTorch, and 1.03 GiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 383, in init_optimizer_and_grad_accumulator
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    optimizer = ZeroDistributedOptimizer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/zero.py", line 91, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 383, in init_optimizer_and_grad_accumulator
+    optimizer = optimizer_builder(param_groups_in_rank)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    optimizer = ZeroDistributedOptimizer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/zero.py", line 91, in __init__
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    optimizer = optimizer_builder(param_groups_in_rank)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 110.00 GiB. GPU 7 has a total capacty of 79.33 GiB of which 12.52 GiB is free. Including non-PyTorch memory, this process has 66.80 GiB memory in use. Of the allocated memory 55.00 GiB is allocated by PyTorch, and 1.03 GiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 110.00 GiB. GPU 4 has a total capacty of 79.33 GiB of which 12.04 GiB is free. Including non-PyTorch memory, this process has 67.28 GiB memory in use. Of the allocated memory 55.00 GiB is allocated by PyTorch, and 1.03 GiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 383, in init_optimizer_and_grad_accumulator
+    optimizer = ZeroDistributedOptimizer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/zero.py", line 91, in __init__
+    optimizer = optimizer_builder(param_groups_in_rank)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 110.00 GiB. GPU 6 has a total capacty of 79.33 GiB of which 12.01 GiB is free. Including non-PyTorch memory, this process has 67.31 GiB memory in use. Of the allocated memory 55.00 GiB is allocated by PyTorch, and 1.03 GiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 383, in init_optimizer_and_grad_accumulator
+    optimizer = ZeroDistributedOptimizer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/zero.py", line 91, in __init__
+    optimizer = optimizer_builder(param_groups_in_rank)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 110.00 GiB. GPU 3 has a total capacty of 79.33 GiB of which 12.21 GiB is free. Including non-PyTorch memory, this process has 67.11 GiB memory in use. Of the allocated memory 55.00 GiB is allocated by PyTorch, and 1.03 GiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 383, in init_optimizer_and_grad_accumulator
+    optimizer = ZeroDistributedOptimizer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/zero.py", line 91, in __init__
+    optimizer = optimizer_builder(param_groups_in_rank)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 106.57 GiB. GPU 3 has a total capacty of 79.33 GiB of which 13.85 GiB is free. Including non-PyTorch memory, this process has 65.47 GiB memory in use. Of the allocated memory 53.28 GiB is allocated by PyTorch, and 1.03 GiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 383, in init_optimizer_and_grad_accumulator
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    optimizer = ZeroDistributedOptimizer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/zero.py", line 91, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 383, in init_optimizer_and_grad_accumulator
+    optimizer = optimizer_builder(param_groups_in_rank)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    optimizer = ZeroDistributedOptimizer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/zero.py", line 91, in __init__
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    optimizer = optimizer_builder(param_groups_in_rank)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 106.57 GiB. GPU 6 has a total capacty of 79.33 GiB of which 13.81 GiB is free. Including non-PyTorch memory, this process has 65.51 GiB memory in use. Of the allocated memory 53.28 GiB is allocated by PyTorch, and 1.03 GiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 106.57 GiB. GPU 1 has a total capacty of 79.33 GiB of which 13.73 GiB is free. Including non-PyTorch memory, this process has 65.59 GiB memory in use. Of the allocated memory 53.28 GiB is allocated by PyTorch, and 1.03 GiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 383, in init_optimizer_and_grad_accumulator
+    optimizer = ZeroDistributedOptimizer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/zero.py", line 91, in __init__
+    optimizer = optimizer_builder(param_groups_in_rank)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 383, in init_optimizer_and_grad_accumulator
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 383, in init_optimizer_and_grad_accumulator
+    optimizer = ZeroDistributedOptimizer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/zero.py", line 91, in __init__
+    optimizer = ZeroDistributedOptimizer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/zero.py", line 91, in __init__
+    optimizer = optimizer_builder(param_groups_in_rank)    
+self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    optimizer = optimizer_builder(param_groups_in_rank)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 106.57 GiB. GPU 5 has a total capacty of 79.33 GiB of which 13.73 GiB is free. Including non-PyTorch memory, this process has 65.59 GiB memory in use. Of the allocated memory 53.28 GiB is allocated by PyTorch, and 1.03 GiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 106.57 GiB. GPU 4 has a total capacty of 79.33 GiB of which 13.84 GiB is free. Including non-PyTorch memory, this process has 65.48 GiB memory in use. Of the allocated memory 53.28 GiB is allocated by PyTorch, and 1.03 GiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 106.57 GiB. GPU 0 has a total capacty of 79.33 GiB of which 13.89 GiB is free. Including non-PyTorch memory, this process has 65.43 GiB memory in use. Of the allocated memory 53.28 GiB is allocated by PyTorch, and 1.03 GiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 383, in init_optimizer_and_grad_accumulator
+    optimizer = ZeroDistributedOptimizer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/zero.py", line 91, in __init__
+    optimizer = optimizer_builder(param_groups_in_rank)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 383, in init_optimizer_and_grad_accumulator
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 106.57 GiB. GPU 7 has a total capacty of 79.33 GiB of which 14.16 GiB is free. Including non-PyTorch memory, this process has 65.16 GiB memory in use. Of the allocated memory 53.28 GiB is allocated by PyTorch, and 1.03 GiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    optimizer = ZeroDistributedOptimizer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/zero.py", line 91, in __init__
+    optimizer = optimizer_builder(param_groups_in_rank)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 106.57 GiB. GPU 2 has a total capacty of 79.33 GiB of which 13.81 GiB is free. Including non-PyTorch memory, this process has 65.51 GiB memory in use. Of the allocated memory 53.28 GiB is allocated by PyTorch, and 1.03 GiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+double free or corruption (!prev)
+terminate called after throwing an instance of 'c10::Error'
+  what():  CUDA driver error: unknown error
+Exception raised from _hasPrimaryContext at ../aten/src/ATen/cuda/detail/CUDAHooks.cpp:67 (most recent call first):
+frame #0: c10::Error::Error(c10::SourceLocation, std::string) + 0x57 (0x7f8f5a1d4617 in /fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/lib/libc10.so)
+frame #1: c10::detail::torchCheckFail(char const*, char const*, unsigned int, char const*) + 0x68 (0x7f8f5a18fa56 in /fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/lib/libc10.so)
+frame #2: <unknown function> + 0x11a00bf (0x7f8f5b4f00bf in /fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/lib/libtorch_cuda.so)
+frame #3: c10::cuda::MaybeSetDevice(int) + 0xc (0x7f8f5a285bdc in /fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/lib/libc10_cuda.so)
+frame #4: std::_Sp_counted_ptr_inplace<std::vector<at::cuda::CUDAEvent, std::allocator<at::cuda::CUDAEvent> >, std::allocator<std::vector<at::cuda::CUDAEvent, std::allocator<at::cuda::CUDAEvent> > >, (__gnu_cxx::_Lock_policy)2>::_M_dispose() + 0x98 (0x7f8f5b5ffae8 in /fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/lib/libtorch_cuda.so)
+frame #5: std::_Sp_counted_base<(__gnu_cxx::_Lock_policy)2>::_M_release() + 0x48 (0x7f8fb9552048 in /fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/lib/libtorch_python.so)
+frame #6: c10d::ProcessGroupNCCL::WorkNCCL::~WorkNCCL() + 0x135 (0x7f8f5b5cac55 in /fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/lib/libtorch_cuda.so)
+frame #7: c10d::ProcessGroupNCCL::workCleanupLoop() + 0x3c5 (0x7f8f5b5e0585 in /fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/lib/libtorch_cuda.so)
+frame #8: c10d::ProcessGroupNCCL::ncclCommWatchdog() + 0x78 (0x7f8f5b5e0718 in /fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/lib/libtorch_cuda.so)
+frame #9: <unknown function> + 0xd3e95 (0x7f8fe92d8e95 in /fsx/nouamane/miniconda/envs/2-1-cu121/bin/../lib/libstdc++.so.6)
+frame #10: <unknown function> + 0x8609 (0x7f8fe9748609 in /lib/x86_64-linux-gnu/libpthread.so.0)
+frame #11: clone + 0x43 (0x7f8fe9511353 in /lib/x86_64-linux-gnu/libc.so.6)
+
+ip-26-0-170-31:3197184:3197777 [0] NCCL INFO [Service thread] Connection closed by localRank 2
+ip-26-0-170-31:3197184:3197605 [0] NCCL INFO [Service thread] Connection closed by localRank 6
+ip-26-0-170-31:3197187:3197604 [0] NCCL INFO [Service thread] Connection closed by localRank 6
+ip-26-0-170-31:3197183:3197606 [0] NCCL INFO [Service thread] Connection closed by localRank 6
+ip-26-0-170-31:3197182:3197603 [0] NCCL INFO [Service thread] Connection closed by localRank 6
+ip-26-0-170-31:3197181:3197610 [0] NCCL INFO [Service thread] Connection closed by localRank 6
+ip-26-0-170-31:3197185:3197609 [0] NCCL INFO [Service thread] Connection closed by localRank 6
+ip-26-0-170-31:3197187:3197371 [0] NCCL INFO [Service thread] Connection closed by localRank 6
+ip-26-0-170-31:3197184:3197372 [0] NCCL INFO [Service thread] Connection closed by localRank 6
+ip-26-0-170-31:3197183:3197373 [0] NCCL INFO [Service thread] Connection closed by localRank 6
+ip-26-0-170-31:3197182:3197377 [0] NCCL INFO [Service thread] Connection closed by localRank 6
+ip-26-0-170-31:3197181:3197378 [0] NCCL INFO [Service thread] Connection closed by localRank 6
+ip-26-0-170-31:3197185:3197376 [0] NCCL INFO [Service thread] Connection closed by localRank 6
+ip-26-0-170-31:3197185:3197609 [0] NCCL INFO [Service thread] Connection closed by localRank 0
+ip-26-0-170-31:3197183:3197606 [0] NCCL INFO [Service thread] Connection closed by localRank 0
+ip-26-0-170-31:3197182:3197603 [0] NCCL INFO [Service thread] Connection closed by localRank 0
+ip-26-0-170-31:3197181:3197610 [0] NCCL INFO [Service thread] Connection closed by localRank 0
+ip-26-0-170-31:3197185:3197376 [0] NCCL INFO [Service thread] Connection closed by localRank 0
+ip-26-0-170-31:3197183:3197373 [0] NCCL INFO [Service thread] Connection closed by localRank 0
+ip-26-0-170-31:3197182:3197377 [0] NCCL INFO [Service thread] Connection closed by localRank 0
+ip-26-0-170-31:3197181:3197378 [0] NCCL INFO [Service thread] Connection closed by localRank 0
+ip-26-0-170-31:3197185:3197609 [0] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-170-31:3197183:3197606 [0] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-170-31:3197182:3197603 [0] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-170-31:3197185:3197376 [0] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-170-31:3197183:3197373 [0] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-170-31:3197182:3197377 [0] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-170-31:3197185:3197609 [0] NCCL INFO [Service thread] Connection closed by localRank 4
+ip-26-0-170-31:3197182:3197603 [0] NCCL INFO [Service thread] Connection closed by localRank 4
+ip-26-0-170-31:3197183:3197606 [0] NCCL INFO [Service thread] Connection closed by localRank 4
+ip-26-0-170-31:3197185:3197376 [0] NCCL INFO [Service thread] Connection closed by localRank 4
+ip-26-0-170-31:3197182:3197377 [0] NCCL INFO [Service thread] Connection closed by localRank 4
+ip-26-0-170-31:3197183:3197373 [0] NCCL INFO [Service thread] Connection closed by localRank 4
+ip-26-0-162-14:2525079:2525494 [0] NCCL INFO [Service thread] Connection closed by localRank 2
+ip-26-0-162-14:2525078:2525490 [0] NCCL INFO [Service thread] Connection closed by localRank 2
+ip-26-0-162-14:2525076:2525489 [0] NCCL INFO [Service thread] Connection closed by localRank 2
+ip-26-0-162-14:2525078:2525258 [0] NCCL INFO [Service thread] Connection closed by localRank 2
+ip-26-0-162-14:2525079:2525262 [0] NCCL INFO [Service thread] Connection closed by localRank 2
+ip-26-0-162-14:2525076:2525263 [0] NCCL INFO [Service thread] Connection closed by localRank 2
+[E TraceUtils.h:35] Store is down while updating #52 with key NCCL_1_trace_end
+[E TraceUtils.h:35] Store is down while updating #59 with key NCCL_2_trace_end
+[E TraceUtils.h:35] Store is down while updating #58 with key NCCL_2_trace_start
+[E TraceUtils.h:35] Store is down while updating #59 with key NCCL_0_trace_start
+[E TraceUtils.h:35] Store is down while updating #51 with key NCCL_0_trace_start
+[E TraceUtils.h:35] Store is down while updating #52 with key NCCL_0_trace_start
+[E TraceUtils.h:35] Store is down while updating #59 with key NCCL_1_trace_start
+[E TraceUtils.h:35] Store is down while updating #52 with key NCCL_1_trace_start
+[E TraceUtils.h:35] Store is down while updating #101 with key NCCL_1_trace_start
+[E TraceUtils.h:35] Store is down while updating #51 with key NCCL_0_trace_start
+[E TraceUtils.h:35] Store is down while updating #100 with key NCCL_0_trace_start
+[E TraceUtils.h:35] Store is down while updating #52 with key NCCL_1_trace_start
+[E TraceUtils.h:35] Store is down while updating #60 with key NCCL_0_trace_start
+[E TraceUtils.h:35] Store is down while updating #59 with key NCCL_1_trace_start
+ip-26-0-163-58:934720:935206 [0] NCCL INFO [Service thread] Connection closed by localRank 1
+ip-26-0-163-58:934720:935166 [0] NCCL INFO [Service thread] Connection closed by localRank 1
+ip-26-0-163-58:934727:935129 [0] NCCL INFO [Service thread] Connection closed by localRank 1
+ip-26-0-163-58:934725:935132 [0] NCCL INFO [Service thread] Connection closed by localRank 1
+ip-26-0-163-58:934724:935135 [0] NCCL INFO [Service thread] Connection closed by localRank 1
+ip-26-0-163-58:934720:935136 [0] NCCL INFO [Service thread] Connection closed by localRank 1
+ip-26-0-163-58:934727:934901 [0] NCCL INFO [Service thread] Connection closed by localRank 1
+ip-26-0-163-58:934725:934900 [0] NCCL INFO [Service thread] Connection closed by localRank 1
+ip-26-0-163-58:934724:934904 [0] NCCL INFO [Service thread] Connection closed by localRank 1
+ip-26-0-163-58:934720:934903 [0] NCCL INFO [Service thread] Connection closed by localRank 1
+ip-26-0-170-132:3330190:3330609 [0] NCCL INFO [Service thread] Connection closed by localRank 3
+ip-26-0-170-132:3330188:3330607 [0] NCCL INFO [Service thread] Connection closed by localRank 3
+ip-26-0-170-132:3330184:3330611 [0] NCCL INFO [Service thread] Connection closed by localRank 3
+ip-26-0-170-132:3330185:3330610 [0] NCCL INFO [Service thread] Connection closed by localRank 3
+ip-26-0-170-132:3330190:3330377 [0] NCCL INFO [Service thread] Connection closed by localRank 3
+ip-26-0-170-132:3330188:3330382 [0] NCCL INFO [Service thread] Connection closed by localRank 3
+ip-26-0-170-132:3330184:3330375 [0] NCCL INFO [Service thread] Connection closed by localRank 3
+ip-26-0-170-132:3330185:3330376 [0] NCCL INFO [Service thread] Connection closed by localRank 3
+ip-26-0-170-132:3330185:3330610 [0] NCCL INFO [Service thread] Connection closed by localRank 6
+ip-26-0-170-132:3330188:3330607 [0] NCCL INFO [Service thread] Connection closed by localRank 6
+ip-26-0-170-132:3330184:3330611 [0] NCCL INFO [Service thread] Connection closed by localRank 6
+ip-26-0-170-132:3330185:3330376 [0] NCCL INFO [Service thread] Connection closed by localRank 6
+ip-26-0-170-132:3330184:3330375 [0] NCCL INFO [Service thread] Connection closed by localRank 6
+ip-26-0-170-132:3330188:3330382 [0] NCCL INFO [Service thread] Connection closed by localRank 6
+[2024-12-28 01:37:00,951] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3197180 closing signal SIGTERM
+[2024-12-28 01:37:00,952] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3197181 closing signal SIGTERM
+[2024-12-28 01:37:00,952] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3197182 closing signal SIGTERM
+[2024-12-28 01:37:00,952] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3197183 closing signal SIGTERM
+[2024-12-28 01:37:00,952] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3197184 closing signal SIGTERM
+[2024-12-28 01:37:00,952] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3197185 closing signal SIGTERM
+[2024-12-28 01:37:00,952] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3197187 closing signal SIGTERM
+[2024-12-28 01:37:04,197] torch.distributed.elastic.multiprocessing.api: [ERROR] failed (exitcode: -6) local_rank: 6 (pid: 3197186) of binary: /fsx/nouamane/miniconda/envs/2-1-cu121/bin/python
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 264, in launch_agent
+    raise ChildFailedError(
+torch.distributed.elastic.multiprocessing.errors.ChildFailedError: 
+========================================================
+/fsx/nouamane/projects/nanotron/run_train.py FAILED
+--------------------------------------------------------
+Failures:
+  <NO_OTHER_FAILURES>
+--------------------------------------------------------
+Root Cause (first observed failure):
+[0]:
+  time      : 2024-12-28_01:37:00
+  host      : ip-26-0-170-31.ec2.internal
+  rank      : 30 (local_rank: 6)
+  exitcode  : -6 (pid: 3197186)
+  error_file: <N/A>
+  traceback : Signal 6 (SIGABRT) received by PID 3197186
+========================================================
+srun: error: ip-26-0-170-31: task 2: Exited with exit code 1
+srun: launch/slurm: _step_signal: Terminating StepId=13848421.0
+slurmstepd: error: *** STEP 13848421.0 ON ip-26-0-162-14 CANCELLED AT 2024-12-28T01:37:04 ***
+[2024-12-28 01:37:04,604] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-28 01:37:04,605] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-28 01:37:04,605] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 934720 closing signal SIGTERM
+[2024-12-28 01:37:04,605] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 934721 closing signal SIGTERM
+[2024-12-28 01:37:04,606] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-28 01:37:04,605] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 934722 closing signal SIGTERM
+[2024-12-28 01:37:04,607] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3330183 closing signal SIGTERM
+[2024-12-28 01:37:04,605] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 934723 closing signal SIGTERM
+[2024-12-28 01:37:04,605] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 934724 closing signal SIGTERM
+[2024-12-28 01:37:04,607] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3330184 closing signal SIGTERM
+[2024-12-28 01:37:04,607] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3330185 closing signal SIGTERM
+[2024-12-28 01:37:04,605] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 934725 closing signal SIGTERM
+[2024-12-28 01:37:04,605] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2525081 closing signal SIGTERM
+[2024-12-28 01:37:04,605] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 934726 closing signal SIGTERM
+[2024-12-28 01:37:04,607] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3330187 closing signal SIGTERM
+[2024-12-28 01:37:04,605] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 934727 closing signal SIGTERM
+[2024-12-28 01:37:04,607] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3330188 closing signal SIGTERM
+[2024-12-28 01:37:04,607] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3330189 closing signal SIGTERM
+[2024-12-28 01:37:04,607] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3330190 closing signal SIGTERM
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 2525002 got signal: 15
+srun: error: ip-26-0-162-14: task 0: Exited with exit code 1
+[2024-12-28 01:37:05,929] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-163-58.ec2.internal_934649_0' has failed to send a keep-alive heartbeat to the rendezvous '13848421' due to an error of type RendezvousConnectionError.
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 3330010 got signal: 15
+srun: error: ip-26-0-170-132: task 3: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 934649 got signal: 15
+srun: error: ip-26-0-163-58: task 1: Exited with exit code 1
diff --git a/logs/13849095-bench_469G_dp2_tp32_pp2_acc2_mbs64_seq4096_zero1_tpmodeRED_vocab131k.out b/logs/13849095-bench_469G_dp2_tp32_pp2_acc2_mbs64_seq4096_zero1_tpmodeRED_vocab131k.out
new file mode 100644
index 0000000000000000000000000000000000000000..3d1b3787a07c14064a8645f021d97c04644dc2ae
--- /dev/null
+++ b/logs/13849095-bench_469G_dp2_tp32_pp2_acc2_mbs64_seq4096_zero1_tpmodeRED_vocab131k.out
@@ -0,0 +1,70303 @@
++ '[' -z 13849095 ']'
++ source /etc/profile.d/modules.sh
+++ . /usr/share/modules/init/bash
++++ unset _mlshdbg
++++ '[' 0 = 1 ']'
++++ unset _mlre _mlIFS
++++ '[' -n x ']'
++++ _mlIFS=' 	
+'
++++ IFS=' '
++++ '[' -n '' ']'
+++++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash autoinit
++++ _mlcode='module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+MODULEPATH_modshare=/etc/environment-modules/modules:1:/usr/share/modules/\$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1; export MODULEPATH_modshare;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+LOADEDMODULES=; export LOADEDMODULES;
+MODULEPATH=/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/\$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles; export MODULEPATH;
+test 0;'
++++ _mlret=0
++++ '[' -n x ']'
++++ IFS=' 	
+'
++++ unset _mlIFS
++++ unset _mlre _mlv _mlrv
++++ '[' 0 -eq 0 ']'
++++ eval 'module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+MODULEPATH_modshare=/etc/environment-modules/modules:1:/usr/share/modules/\$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1; export MODULEPATH_modshare;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+LOADEDMODULES=; export LOADEDMODULES;
+MODULEPATH=/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/\$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles; export MODULEPATH;
+test 0;'
+++++ MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl
+++++ export MODULES_CMD
+++++ ENV=/usr/share/modules/init/profile.sh
+++++ export ENV
+++++ MODULEPATH_modshare='/etc/environment-modules/modules:1:/usr/share/modules/$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1'
+++++ export MODULEPATH_modshare
+++++ BASH_ENV=/usr/share/modules/init/bash
+++++ export BASH_ENV
+++++ MODULESHOME=/usr/share/modules
+++++ export MODULESHOME
+++++ LOADEDMODULES=
+++++ export LOADEDMODULES
+++++ MODULEPATH='/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles'
+++++ export MODULEPATH
+++++ test 0
++++ '[' 0 = 1 ']'
++++ '[' -t 2 ']'
++++ export -f module
++++ export -f switchml
++++ '[' 5 -ge 3 ']'
++++ [[ ehxB =~ i ]]
++++ [[ ! :/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin: =~ :/usr/bin: ]]
++++ '[' '!' -n '' ']'
++++ MANPATH=:
++++ export MANPATH
+++++ manpath
++++ [[ ! :/fsx/nouamane/miniconda/envs/2-1-cu121/man:/fsx/nouamane/miniconda/envs/2-1-cu121/share/man:/fsx/nouamane/miniconda/man:/fsx/nouamane/miniconda/share/man:/usr/local/man:/usr/local/share/man:/usr/share/man:/opt/slurm/share/man:/admin/home/nouamane/.local/share/man:/admin/home/nouamane/.fzf/man:: =~ :/usr/share/man: ]]
++++ unset _mlcode _mlret
++++ '[' -n '' ']'
++ module load cuda/12.1
++ unset _mlshdbg
++ '[' 0 = 1 ']'
++ unset _mlre _mlIFS
++ '[' -n x ']'
++ _mlIFS=' 	
+'
++ IFS=' '
++ '[' -n '' ']'
+++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash load cuda/12.1
++ eval 'CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include;' export 'CPATH;
+LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:;' export 'LD_LIBRARY_PATH;
+FI_EFA_FORK_SAFE=1;' export 'FI_EFA_FORK_SAFE;
+MANPATH=/usr/local/cuda-12.1/share/man::;' export 'MANPATH;
+LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64;' export 'LIBRARY_PATH;
+_LMFILES_=/usr/share/modules/modulefiles/cuda/12.1;' export '_LMFILES_;
+LOADEDMODULES=cuda/12.1;' export 'LOADEDMODULES;
+MPI_PATH=/opt/amazon/openmpi;' export 'MPI_PATH;
+NCCL_HOME_modshare=/opt/nccl/build:1;' export 'NCCL_HOME_modshare;
+NCCL_PROTO=simple;' export 'NCCL_PROTO;
+MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1;' export 'MANPATH_modshare;
+LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1;' export 'LIBRARY_PATH_modshare;
+NCCL_SOCKET_IFNAME=enp;' export 'NCCL_SOCKET_IFNAME;
+AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl;' export 'AWS_OFI_NCCL_HOME;
+NCCL_HOME=/opt/nccl/build;' export 'NCCL_HOME;
+FI_PROVIDER=efa;' export 'FI_PROVIDER;
+AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1;' export 'AWS_OFI_NCCL_HOME_modshare;
+CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1;' export 'CPATH_modshare;
+LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2::1:/opt/nccl/build/lib:1:/opt/aws-ofi-nccl/lib:2:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/efa/lib:1;' export 'LD_LIBRARY_PATH_modshare;
+FI_EFA_ENABLE_SHM_TRANSFER=1;' export 'FI_EFA_ENABLE_SHM_TRANSFER;
+_LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1;' export '_LMFILES__modshare;
+LOADEDMODULES_modshare=cuda/12.1:1;' export 'LOADEDMODULES_modshare;
+MPI_PATH_modshare=/opt/amazon/openmpi:1;' export 'MPI_PATH_modshare;
+PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin;' export 'PATH;
+CUDA_HOME=/usr/local/cuda-12.1;' export 'CUDA_HOME;
+PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/opt/slurm/bin:1:/usr/sbin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1;' export 'PATH_modshare;
+test' '0;'
+++ CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include
+++ export CPATH
+++ LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:
+++ export LD_LIBRARY_PATH
+++ FI_EFA_FORK_SAFE=1
+++ export FI_EFA_FORK_SAFE
+++ MANPATH=/usr/local/cuda-12.1/share/man::
+++ export MANPATH
+++ LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64
+++ export LIBRARY_PATH
+++ _LMFILES_=/usr/share/modules/modulefiles/cuda/12.1
+++ export _LMFILES_
+++ LOADEDMODULES=cuda/12.1
+++ export LOADEDMODULES
+++ MPI_PATH=/opt/amazon/openmpi
+++ export MPI_PATH
+++ NCCL_HOME_modshare=/opt/nccl/build:1
+++ export NCCL_HOME_modshare
+++ NCCL_PROTO=simple
+++ export NCCL_PROTO
+++ MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1
+++ export MANPATH_modshare
+++ LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1
+++ export LIBRARY_PATH_modshare
+++ NCCL_SOCKET_IFNAME=enp
+++ export NCCL_SOCKET_IFNAME
+++ AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl
+++ export AWS_OFI_NCCL_HOME
+++ NCCL_HOME=/opt/nccl/build
+++ export NCCL_HOME
+++ FI_PROVIDER=efa
+++ export FI_PROVIDER
+++ AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1
+++ export AWS_OFI_NCCL_HOME_modshare
+++ CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1
+++ export CPATH_modshare
+++ LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2::1:/opt/nccl/build/lib:1:/opt/aws-ofi-nccl/lib:2:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/efa/lib:1
+++ export LD_LIBRARY_PATH_modshare
+++ FI_EFA_ENABLE_SHM_TRANSFER=1
+++ export FI_EFA_ENABLE_SHM_TRANSFER
+++ _LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1
+++ export _LMFILES__modshare
+++ LOADEDMODULES_modshare=cuda/12.1:1
+++ export LOADEDMODULES_modshare
+++ MPI_PATH_modshare=/opt/amazon/openmpi:1
+++ export MPI_PATH_modshare
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
+++ export PATH
+++ CUDA_HOME=/usr/local/cuda-12.1
+++ export CUDA_HOME
+++ PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/opt/slurm/bin:1:/usr/sbin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1
+++ export PATH_modshare
+++ test 0
++ _mlstatus=0
++ '[' -n x ']'
++ IFS=' 	
+'
++ unset _mlre _mlv _mlrv _mlIFS
++ '[' -n '' ']'
++ unset _mlshdbg
++ return 0
++ source /fsx/nouamane/miniconda/bin/activate
+++ _CONDA_ROOT=/fsx/nouamane/miniconda
+++ . /fsx/nouamane/miniconda/etc/profile.d/conda.sh
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ '[' -z x ']'
+++ conda activate
+++ local cmd=activate
+++ case "$cmd" in
+++ __conda_activate activate
+++ '[' -n '' ']'
+++ local ask_conda
++++ PS1=
++++ __conda_exe shell.posix activate
++++ /fsx/nouamane/miniconda/bin/conda shell.posix activate
+Error while loading conda entry point: conda-libmamba-solver (module 'libmambapy' has no attribute 'QueryFormat')
+++ ask_conda='. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''2'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_1='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/etc/conda/activate.d/libxml2_activate.sh"'
+++ eval '. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''2'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_1='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/etc/conda/activate.d/libxml2_activate.sh"'
++++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh
+++++ test -n ''
+++++ unset XML_CATALOG_FILES
+++++ unset xml_catalog_files_libxml2
++++ PS1='(base) '
++++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
++++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
++++ export CONDA_PREFIX=/fsx/nouamane/miniconda
++++ CONDA_PREFIX=/fsx/nouamane/miniconda
++++ export CONDA_SHLVL=2
++++ CONDA_SHLVL=2
++++ export CONDA_DEFAULT_ENV=base
++++ CONDA_DEFAULT_ENV=base
++++ export 'CONDA_PROMPT_MODIFIER=(base) '
++++ CONDA_PROMPT_MODIFIER='(base) '
++++ export CONDA_PREFIX_1=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ CONDA_PREFIX_1=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ . /fsx/nouamane/miniconda/etc/conda/activate.d/libxml2_activate.sh
+++++ test -n ''
+++++ xml_catalog_files_libxml2=
+++++ XML_CATALOG_FILES=
+++++ conda_catalog_files=
+++++ ifs_libxml2=' 	
+'
+++++ IFS=' '
+++++ rem=/fsx/nouamane/miniconda
+++++ for pre in ${rem}
+++++ test '' = /fsx/nouamane/miniconda
+++++ conda_catalog_files=/fsx/nouamane/miniconda
+++++ rem=
+++++ IFS=' 	
+'
+++++ conda_catalog_files='file:///fsx/nouamane/miniconda/etc/xml/catalog file:///etc/xml/catalog'
+++++ export 'XML_CATALOG_FILES=file:///fsx/nouamane/miniconda/etc/xml/catalog file:///etc/xml/catalog'
+++++ XML_CATALOG_FILES='file:///fsx/nouamane/miniconda/etc/xml/catalog file:///etc/xml/catalog'
+++++ unset conda_catalog_files ifs_libxml2 rem
+++ __conda_hashr
+++ '[' -n '' ']'
+++ '[' -n '' ']'
+++ hash -r
++ conda activate 2-1-cu121
++ local cmd=activate
++ case "$cmd" in
++ __conda_activate activate 2-1-cu121
++ '[' -n '' ']'
++ local ask_conda
+++ PS1='(base) '
+++ __conda_exe shell.posix activate 2-1-cu121
+++ /fsx/nouamane/miniconda/bin/conda shell.posix activate 2-1-cu121
+Error while loading conda entry point: conda-libmamba-solver (module 'libmambapy' has no attribute 'QueryFormat')
++ ask_conda='. "/fsx/nouamane/miniconda/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_2='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
++ eval '. "/fsx/nouamane/miniconda/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_2='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
+++ . /fsx/nouamane/miniconda/etc/conda/deactivate.d/libxml2_deactivate.sh
++++ test -n ''
++++ unset XML_CATALOG_FILES
++++ unset xml_catalog_files_libxml2
+++ PS1='(2-1-cu121) '
+++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
+++ export CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ export CONDA_SHLVL=3
+++ CONDA_SHLVL=3
+++ export CONDA_DEFAULT_ENV=2-1-cu121
+++ CONDA_DEFAULT_ENV=2-1-cu121
+++ export 'CONDA_PROMPT_MODIFIER=(2-1-cu121) '
+++ CONDA_PROMPT_MODIFIER='(2-1-cu121) '
+++ export CONDA_PREFIX_2=/fsx/nouamane/miniconda
+++ CONDA_PREFIX_2=/fsx/nouamane/miniconda
+++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ export _CE_M=
+++ _CE_M=
+++ export _CE_CONDA=
+++ _CE_CONDA=
+++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh
++++ test -n ''
++++ xml_catalog_files_libxml2=
++++ XML_CATALOG_FILES=
++++ conda_catalog_files=
++++ ifs_libxml2=' 	
+'
++++ IFS=' '
++++ rem=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ for pre in ${rem}
++++ test '' = /fsx/nouamane/miniconda/envs/2-1-cu121
++++ conda_catalog_files=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ rem=
++++ IFS=' 	
+'
++++ conda_catalog_files='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ export 'XML_CATALOG_FILES=file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ XML_CATALOG_FILES='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ unset conda_catalog_files ifs_libxml2 rem
++ __conda_hashr
++ '[' -n '' ']'
++ '[' -n '' ']'
++ hash -r
++ export PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
++ PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/a846435528b4b760494a836f96f0739889253530/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
++ '[' -z '' ']'
+++ scontrol show hostnames 'ip-26-0-160-[225,242],ip-26-0-161-[78,103,123,138,142,153,178],ip-26-0-165-[59,131,164,202],ip-26-0-168-[30,34,52]'
++ export 'NODELIST=ip-26-0-160-225
+ip-26-0-160-242
+ip-26-0-161-78
+ip-26-0-161-103
+ip-26-0-161-123
+ip-26-0-161-138
+ip-26-0-161-142
+ip-26-0-161-153
+ip-26-0-161-178
+ip-26-0-165-59
+ip-26-0-165-131
+ip-26-0-165-164
+ip-26-0-165-202
+ip-26-0-168-30
+ip-26-0-168-34
+ip-26-0-168-52'
++ NODELIST='ip-26-0-160-225
+ip-26-0-160-242
+ip-26-0-161-78
+ip-26-0-161-103
+ip-26-0-161-123
+ip-26-0-161-138
+ip-26-0-161-142
+ip-26-0-161-153
+ip-26-0-161-178
+ip-26-0-165-59
+ip-26-0-165-131
+ip-26-0-165-164
+ip-26-0-165-202
+ip-26-0-168-30
+ip-26-0-168-34
+ip-26-0-168-52'
+++ head -n1
+++ scontrol show hostnames 'ip-26-0-160-[225,242],ip-26-0-161-[78,103,123,138,142,153,178],ip-26-0-165-[59,131,164,202],ip-26-0-168-[30,34,52]'
++ export MASTER_NODE=ip-26-0-160-225
++ MASTER_NODE=ip-26-0-160-225
++ export MASTER_PORT=12356
++ MASTER_PORT=12356
++ export NNODES=16
++ NNODES=16
++ export GPUS_PER_NODE=8
++ GPUS_PER_NODE=8
++ export WORLD_SIZE=128
++ WORLD_SIZE=128
++ export CUDA_DEVICE_MAX_CONNECTIONS=1
++ CUDA_DEVICE_MAX_CONNECTIONS=1
++ export NCCL_DEBUG=INFO
++ NCCL_DEBUG=INFO
++ export NANOTRON_BENCHMARK=1
++ NANOTRON_BENCHMARK=1
++ export WANDB_MODE=disabled
++ WANDB_MODE=disabled
++ export TORCH_NCCL_ASYNC_ERROR_HANDLING=1
++ TORCH_NCCL_ASYNC_ERROR_HANDLING=1
++ export TORCH_DISTRIBUTED_DEBUG=DETAIL
++ TORCH_DISTRIBUTED_DEBUG=DETAIL
++ '[' -z '' ']'
++ echo '=== GPU Topology ==='
+=== GPU Topology ===
++ nvidia-smi topo -m
+	GPU0	GPU1	GPU2	GPU3	GPU4	GPU5	GPU6	GPU7	CPU Affinity	NUMA Affinity	GPU NUMA ID
+GPU0	 X 	NV18	NV18	NV18	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU1	NV18	 X 	NV18	NV18	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU2	NV18	NV18	 X 	NV18	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU3	NV18	NV18	NV18	 X 	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU4	NV18	NV18	NV18	NV18	 X 	NV18	NV18	NV18	48-95	1		N/A
+GPU5	NV18	NV18	NV18	NV18	NV18	 X 	NV18	NV18	48-95	1		N/A
+GPU6	NV18	NV18	NV18	NV18	NV18	NV18	 X 	NV18	48-95	1		N/A
+GPU7	NV18	NV18	NV18	NV18	NV18	NV18	NV18	 X 	48-95	1		N/A
+
+Legend:
+
+  X    = Self
+  SYS  = Connection traversing PCIe as well as the SMP interconnect between NUMA nodes (e.g., QPI/UPI)
+  NODE = Connection traversing PCIe as well as the interconnect between PCIe Host Bridges within a NUMA node
+  PHB  = Connection traversing PCIe as well as a PCIe Host Bridge (typically the CPU)
+  PXB  = Connection traversing multiple PCIe bridges (without traversing the PCIe Host Bridge)
+  PIX  = Connection traversing at most a single PCIe bridge
+  NV#  = Connection traversing a bonded set of # NVLinks
++ echo ==================
+==================
++ export SRUN_ALLOC_ARGS=
++ SRUN_ALLOC_ARGS=
++ echo 'Master node: ip-26-0-160-225'
+Master node: ip-26-0-160-225
++ echo 'All nodes: ip-26-0-160-225
+ip-26-0-160-242
+ip-26-0-161-78
+ip-26-0-161-103
+ip-26-0-161-123
+ip-26-0-161-138
+ip-26-0-161-142
+ip-26-0-161-153
+ip-26-0-161-178
+ip-26-0-165-59
+ip-26-0-165-131
+ip-26-0-165-164
+ip-26-0-165-202
+ip-26-0-168-30
+ip-26-0-168-34
+ip-26-0-168-52'
+All nodes: ip-26-0-160-225
+ip-26-0-160-242
+ip-26-0-161-78
+ip-26-0-161-103
+ip-26-0-161-123
+ip-26-0-161-138
+ip-26-0-161-142
+ip-26-0-161-153
+ip-26-0-161-178
+ip-26-0-165-59
+ip-26-0-165-131
+ip-26-0-165-164
+ip-26-0-165-202
+ip-26-0-168-30
+ip-26-0-168-34
+ip-26-0-168-52
++ echo 'World size: 128'
+World size: 128
++ '[' -n '' ']'
++ srun --wait=0 --kill-on-bad-exit=1 torchrun --nnodes=16 --nproc_per_node=8 --rdzv_id=13849095 --rdzv_backend=c10d --rdzv_endpoint=ip-26-0-160-225:12356 --max_restarts 0 --rdzv_conf timeout=60 /fsx/nouamane/projects/nanotron/run_train.py --config-file benchmark/configs/config_469G_dp2_tp32_pp2_acc2_mbs64_seq4096_zero1_tpmodeRED_vocab131k.yaml
+[2024-12-28 11:54:53,199] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-28 11:54:53,199] torch.distributed.run: [WARNING] 
+[2024-12-28 11:54:53,199] torch.distributed.run: [WARNING] *****************************************
+[2024-12-28 11:54:53,199] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-28 11:54:53,199] torch.distributed.run: [WARNING] *****************************************
+[2024-12-28 11:54:55,067] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-28 11:54:55,068] torch.distributed.run: [WARNING] 
+[2024-12-28 11:54:55,068] torch.distributed.run: [WARNING] *****************************************
+[2024-12-28 11:54:55,068] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-28 11:54:55,068] torch.distributed.run: [WARNING] *****************************************
+[2024-12-28 11:54:55,071] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-28 11:54:55,071] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-28 11:54:55,071] torch.distributed.run: [WARNING] 
+[2024-12-28 11:54:55,071] torch.distributed.run: [WARNING] *****************************************
+[2024-12-28 11:54:55,071] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-28 11:54:55,071] torch.distributed.run: [WARNING] *****************************************
+[2024-12-28 11:54:55,072] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-28 11:54:55,071] torch.distributed.run: [WARNING] 
+[2024-12-28 11:54:55,071] torch.distributed.run: [WARNING] *****************************************
+[2024-12-28 11:54:55,071] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-28 11:54:55,071] torch.distributed.run: [WARNING] *****************************************
+[2024-12-28 11:54:55,073] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-28 11:54:55,073] torch.distributed.run: [WARNING] 
+[2024-12-28 11:54:55,073] torch.distributed.run: [WARNING] *****************************************
+[2024-12-28 11:54:55,073] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-28 11:54:55,073] torch.distributed.run: [WARNING] *****************************************
+[2024-12-28 11:54:55,077] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-28 11:54:55,078] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-28 11:54:55,078] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-28 11:54:55,077] torch.distributed.run: [WARNING] 
+[2024-12-28 11:54:55,077] torch.distributed.run: [WARNING] *****************************************
+[2024-12-28 11:54:55,077] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-28 11:54:55,077] torch.distributed.run: [WARNING] *****************************************
+[2024-12-28 11:54:55,079] torch.distributed.run: [WARNING] 
+[2024-12-28 11:54:55,079] torch.distributed.run: [WARNING] *****************************************
+[2024-12-28 11:54:55,079] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-28 11:54:55,079] torch.distributed.run: [WARNING] *****************************************
+[2024-12-28 11:54:55,078] torch.distributed.run: [WARNING] 
+[2024-12-28 11:54:55,078] torch.distributed.run: [WARNING] *****************************************
+[2024-12-28 11:54:55,078] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-28 11:54:55,078] torch.distributed.run: [WARNING] *****************************************
+[2024-12-28 11:54:55,082] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-28 11:54:55,082] torch.distributed.run: [WARNING] 
+[2024-12-28 11:54:55,082] torch.distributed.run: [WARNING] *****************************************
+[2024-12-28 11:54:55,082] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-28 11:54:55,082] torch.distributed.run: [WARNING] *****************************************
+[2024-12-28 11:54:55,083] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-28 11:54:55,083] torch.distributed.run: [WARNING] 
+[2024-12-28 11:54:55,083] torch.distributed.run: [WARNING] *****************************************
+[2024-12-28 11:54:55,083] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-28 11:54:55,083] torch.distributed.run: [WARNING] *****************************************
+[2024-12-28 11:54:55,093] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-28 11:54:55,093] torch.distributed.run: [WARNING] 
+[2024-12-28 11:54:55,093] torch.distributed.run: [WARNING] *****************************************
+[2024-12-28 11:54:55,093] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-28 11:54:55,093] torch.distributed.run: [WARNING] *****************************************
+[2024-12-28 11:54:55,092] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-28 11:54:55,092] torch.distributed.run: [WARNING] 
+[2024-12-28 11:54:55,092] torch.distributed.run: [WARNING] *****************************************
+[2024-12-28 11:54:55,092] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-28 11:54:55,092] torch.distributed.run: [WARNING] *****************************************
+[2024-12-28 11:54:55,108] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-28 11:54:55,072] torch.distributed.run: [WARNING] 
+[2024-12-28 11:54:55,072] torch.distributed.run: [WARNING] *****************************************
+[2024-12-28 11:54:55,072] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-28 11:54:55,072] torch.distributed.run: [WARNING] *****************************************
+[2024-12-28 11:54:55,108] torch.distributed.run: [WARNING] 
+[2024-12-28 11:54:55,108] torch.distributed.run: [WARNING] *****************************************
+[2024-12-28 11:54:55,108] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-28 11:54:55,108] torch.distributed.run: [WARNING] *****************************************
+[2024-12-28 11:54:55,155] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-28 11:54:55,155] torch.distributed.run: [WARNING] 
+[2024-12-28 11:54:55,155] torch.distributed.run: [WARNING] *****************************************
+[2024-12-28 11:54:55,155] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-28 11:54:55,155] torch.distributed.run: [WARNING] *****************************************
+[2024-12-28 11:54:55,160] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2024-12-28 11:54:55,160] torch.distributed.run: [WARNING] 
+[2024-12-28 11:54:55,160] torch.distributed.run: [WARNING] *****************************************
+[2024-12-28 11:54:55,160] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2024-12-28 11:54:55,160] torch.distributed.run: [WARNING] *****************************************
+ip-26-0-160-225:124160:124160 [0] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-160-225:124160:124160 [0] NCCL INFO Bootstrap : Using enp71s0:26.0.160.225<0>
+ip-26-0-160-225:124160:124160 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-160-225:124160:124160 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-160-225:124160:124160 [0] NCCL INFO cudaDriverVersion 12020
+NCCL version 2.18.5+cuda12.2
+ip-26-0-168-34:272156:272156 [6] NCCL INFO cudaDriverVersion 12020
+ip-26-0-168-34:272157:272157 [7] NCCL INFO cudaDriverVersion 12020
+ip-26-0-168-34:272157:272157 [7] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-168-34:272156:272156 [6] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-165-59:1557273:1557273 [7] NCCL INFO cudaDriverVersion 12020
+ip-26-0-165-59:1557271:1557271 [5] NCCL INFO cudaDriverVersion 12020
+ip-26-0-165-59:1557273:1557273 [7] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-165-59:1557271:1557271 [5] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-168-34:272157:272157 [7] NCCL INFO Bootstrap : Using enp71s0:26.0.168.34<0>
+ip-26-0-168-34:272156:272156 [6] NCCL INFO Bootstrap : Using enp71s0:26.0.168.34<0>
+ip-26-0-161-78:114518:114518 [6] NCCL INFO cudaDriverVersion 12020
+ip-26-0-161-78:114519:114519 [7] NCCL INFO cudaDriverVersion 12020
+ip-26-0-165-59:1557271:1557271 [5] NCCL INFO Bootstrap : Using enp71s0:26.0.165.59<0>
+ip-26-0-165-59:1557273:1557273 [7] NCCL INFO Bootstrap : Using enp71s0:26.0.165.59<0>
+ip-26-0-161-78:114518:114518 [6] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-161-78:114519:114519 [7] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-165-202:153564:153564 [2] NCCL INFO cudaDriverVersion 12020
+ip-26-0-165-202:153566:153566 [4] NCCL INFO cudaDriverVersion 12020
+ip-26-0-161-78:114515:114515 [3] NCCL INFO cudaDriverVersion 12020
+ip-26-0-168-34:272157:272157 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-168-34:272157:272157 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-168-34:272156:272156 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-168-34:272156:272156 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-161-78:114512:114512 [0] NCCL INFO cudaDriverVersion 12020
+ip-26-0-165-202:153564:153564 [2] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-161-123:32708:32708 [4] NCCL INFO cudaDriverVersion 12020
+ip-26-0-165-202:153566:153566 [4] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-161-78:114515:114515 [3] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-161-78:114519:114519 [7] NCCL INFO Bootstrap : Using enp71s0:26.0.161.78<0>
+ip-26-0-161-78:114518:114518 [6] NCCL INFO Bootstrap : Using enp71s0:26.0.161.78<0>
+ip-26-0-161-78:114512:114512 [0] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-165-202:153562:153562 [0] NCCL INFO cudaDriverVersion 12020
+ip-26-0-165-202:153562:153562 [0] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-168-30:235648:235648 [0] NCCL INFO cudaDriverVersion 12020
+ip-26-0-165-202:153568:153568 [6] NCCL INFO cudaDriverVersion 12020
+ip-26-0-161-123:32708:32708 [4] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-168-30:235654:235654 [6] NCCL INFO cudaDriverVersion 12020
+ip-26-0-165-202:153565:153565 [3] NCCL INFO cudaDriverVersion 12020
+ip-26-0-161-123:32707:32707 [3] NCCL INFO cudaDriverVersion 12020
+ip-26-0-168-30:235649:235649 [1] NCCL INFO cudaDriverVersion 12020
+ip-26-0-168-30:235655:235655 [7] NCCL INFO cudaDriverVersion 12020
+ip-26-0-161-103:96888:96888 [0] NCCL INFO cudaDriverVersion 12020
+ip-26-0-161-103:96892:96892 [4] NCCL INFO cudaDriverVersion 12020
+ip-26-0-168-30:235653:235653 [5] NCCL INFO cudaDriverVersion 12020
+ip-26-0-161-78:114515:114515 [3] NCCL INFO Bootstrap : Using enp71s0:26.0.161.78<0>
+ip-26-0-165-59:1557271:1557271 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-165-59:1557271:1557271 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-168-30:235652:235652 [4] NCCL INFO cudaDriverVersion 12020
+ip-26-0-168-30:235651:235651 [3] NCCL INFO cudaDriverVersion 12020
+ip-26-0-165-164:1227464:1227464 [1] NCCL INFO cudaDriverVersion 12020
+ip-26-0-168-30:235650:235650 [2] NCCL INFO cudaDriverVersion 12020
+ip-26-0-165-202:153568:153568 [6] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-165-202:153565:153565 [3] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-161-123:32705:32705 [1] NCCL INFO cudaDriverVersion 12020
+ip-26-0-161-178:2815966:2815966 [3] NCCL INFO cudaDriverVersion 12020
+ip-26-0-165-59:1557273:1557273 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-165-59:1557273:1557273 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-161-78:114512:114512 [0] NCCL INFO Bootstrap : Using enp71s0:26.0.161.78<0>
+ip-26-0-161-123:32707:32707 [3] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-161-153:2636090:2636090 [6] NCCL INFO cudaDriverVersion 12020
+ip-26-0-161-103:96889:96889 [1] NCCL INFO cudaDriverVersion 12020
+ip-26-0-161-103:96891:96891 [3] NCCL INFO cudaDriverVersion 12020
+ip-26-0-161-103:96890:96890 [2] NCCL INFO cudaDriverVersion 12020
+ip-26-0-161-142:62885:62885 [4] NCCL INFO cudaDriverVersion 12020
+ip-26-0-161-103:96893:96893 [5] NCCL INFO cudaDriverVersion 12020
+ip-26-0-161-123:32705:32705 [1] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-165-164:1227469:1227469 [6] NCCL INFO cudaDriverVersion 12020
+ip-26-0-161-178:2815967:2815967 [4] NCCL INFO cudaDriverVersion 12020
+ip-26-0-161-103:96894:96894 [6] NCCL INFO cudaDriverVersion 12020
+ip-26-0-161-103:96895:96895 [7] NCCL INFO cudaDriverVersion 12020
+ip-26-0-161-142:62887:62887 [6] NCCL INFO cudaDriverVersion 12020
+ip-26-0-161-178:2815968:2815968 [5] NCCL INFO cudaDriverVersion 12020
+ip-26-0-161-178:2815963:2815963 [0] NCCL INFO cudaDriverVersion 12020
+ip-26-0-161-178:2815964:2815964 [1] NCCL INFO cudaDriverVersion 12020
+ip-26-0-161-142:62881:62881 [0] NCCL INFO cudaDriverVersion 12020
+ip-26-0-161-142:62886:62886 [5] NCCL INFO cudaDriverVersion 12020
+ip-26-0-161-142:62884:62884 [3] NCCL INFO cudaDriverVersion 12020
+ip-26-0-161-178:2815970:2815970 [7] NCCL INFO cudaDriverVersion 12020
+ip-26-0-168-30:235651:235651 [3] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-161-142:62883:62883 [2] NCCL INFO cudaDriverVersion 12020
+ip-26-0-161-178:2815965:2815965 [2] NCCL INFO cudaDriverVersion 12020
+ip-26-0-168-30:235648:235648 [0] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-161-138:96570:96570 [7] NCCL INFO cudaDriverVersion 12020
+ip-26-0-161-142:62882:62882 [1] NCCL INFO cudaDriverVersion 12020
+ip-26-0-161-178:2815969:2815969 [6] NCCL INFO cudaDriverVersion 12020
+ip-26-0-161-142:62888:62888 [7] NCCL INFO cudaDriverVersion 12020
+ip-26-0-161-138:96569:96569 [6] NCCL INFO cudaDriverVersion 12020
+ip-26-0-161-153:2636090:2636090 [6] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-168-30:235650:235650 [2] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-161-138:96566:96566 [3] NCCL INFO cudaDriverVersion 12020
+ip-26-0-168-30:235654:235654 [6] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-168-30:235655:235655 [7] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-168-30:235652:235652 [4] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-168-30:235653:235653 [5] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-161-138:96563:96563 [0] NCCL INFO cudaDriverVersion 12020
+ip-26-0-168-30:235649:235649 [1] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-161-123:32708:32708 [4] NCCL INFO Bootstrap : Using enp71s0:26.0.161.123<0>
+ip-26-0-161-103:96895:96895 [7] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-161-103:96892:96892 [4] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-161-103:96893:96893 [5] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-161-103:96888:96888 [0] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-161-103:96889:96889 [1] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-161-138:96564:96564 [1] NCCL INFO cudaDriverVersion 12020
+ip-26-0-161-103:96890:96890 [2] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-161-103:96891:96891 [3] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-165-164:1227464:1227464 [1] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-161-138:96568:96568 [5] NCCL INFO cudaDriverVersion 12020
+ip-26-0-161-103:96894:96894 [6] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-160-225:124166:124166 [6] NCCL INFO cudaDriverVersion 12020
+ip-26-0-165-202:153569:153569 [7] NCCL INFO cudaDriverVersion 12020
+ip-26-0-161-178:2815966:2815966 [3] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-161-138:96565:96565 [2] NCCL INFO cudaDriverVersion 12020
+ip-26-0-165-164:1227469:1227469 [6] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-161-178:2815964:2815964 [1] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-161-178:2815963:2815963 [0] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-161-178:2815967:2815967 [4] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-161-178:2815968:2815968 [5] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-161-178:2815969:2815969 [6] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-161-178:2815965:2815965 [2] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-168-34:272151:272151 [1] NCCL INFO cudaDriverVersion 12020
+ip-26-0-161-178:2815970:2815970 [7] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-161-138:96567:96567 [4] NCCL INFO cudaDriverVersion 12020
+ip-26-0-165-202:153564:153564 [2] NCCL INFO Bootstrap : Using enp71s0:26.0.165.202<0>
+ip-26-0-160-225:124161:124161 [1] NCCL INFO cudaDriverVersion 12020
+ip-26-0-165-202:153566:153566 [4] NCCL INFO Bootstrap : Using enp71s0:26.0.165.202<0>
+ip-26-0-160-225:124164:124164 [4] NCCL INFO cudaDriverVersion 12020
+ip-26-0-160-225:124167:124167 [7] NCCL INFO cudaDriverVersion 12020
+ip-26-0-165-202:153569:153569 [7] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-168-52:2722046:2722046 [1] NCCL INFO cudaDriverVersion 12020
+ip-26-0-160-225:124166:124166 [6] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-160-225:124165:124165 [5] NCCL INFO cudaDriverVersion 12020
+ip-26-0-165-202:153567:153567 [5] NCCL INFO cudaDriverVersion 12020
+ip-26-0-160-225:124162:124162 [2] NCCL INFO cudaDriverVersion 12020
+ip-26-0-165-202:153563:153563 [1] NCCL INFO cudaDriverVersion 12020
+ip-26-0-161-123:32706:32706 [2] NCCL INFO cudaDriverVersion 12020
+ip-26-0-168-52:2722048:2722048 [3] NCCL INFO cudaDriverVersion 12020
+ip-26-0-168-34:272151:272151 [1] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-161-78:114515:114515 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-161-78:114519:114519 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-161-78:114518:114518 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-161-78:114519:114519 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-161-78:114515:114515 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-161-78:114518:114518 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-168-52:2722047:2722047 [2] NCCL INFO cudaDriverVersion 12020
+ip-26-0-161-142:62885:62885 [4] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-161-78:114512:114512 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-161-78:114512:114512 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-161-153:2636090:2636090 [6] NCCL INFO Bootstrap : Using enp71s0:26.0.161.153<0>
+ip-26-0-161-142:62888:62888 [7] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-161-142:62887:62887 [6] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-168-52:2722045:2722045 [0] NCCL INFO cudaDriverVersion 12020
+ip-26-0-161-142:62886:62886 [5] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-165-202:153567:153567 [5] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-168-52:2722052:2722052 [7] NCCL INFO cudaDriverVersion 12020
+ip-26-0-161-142:62884:62884 [3] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-161-142:62882:62882 [1] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-161-142:62883:62883 [2] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-161-142:62881:62881 [0] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-168-52:2722049:2722049 [4] NCCL INFO cudaDriverVersion 12020
+ip-26-0-165-202:153563:153563 [1] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-160-225:124164:124164 [4] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-160-225:124167:124167 [7] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-168-52:2722051:2722051 [6] NCCL INFO cudaDriverVersion 12020
+ip-26-0-160-225:124165:124165 [5] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-161-123:32707:32707 [3] NCCL INFO Bootstrap : Using enp71s0:26.0.161.123<0>
+ip-26-0-160-225:124162:124162 [2] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-161-138:96569:96569 [6] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-168-52:2722050:2722050 [5] NCCL INFO cudaDriverVersion 12020
+ip-26-0-161-138:96567:96567 [4] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-161-138:96563:96563 [0] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-161-138:96565:96565 [2] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-160-225:124161:124161 [1] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-161-123:32705:32705 [1] NCCL INFO Bootstrap : Using enp71s0:26.0.161.123<0>
+ip-26-0-161-123:32706:32706 [2] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-161-138:96568:96568 [5] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-161-138:96564:96564 [1] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-161-138:96566:96566 [3] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-161-138:96570:96570 [7] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-168-34:272155:272155 [5] NCCL INFO cudaDriverVersion 12020
+ip-26-0-161-78:114514:114514 [2] NCCL INFO cudaDriverVersion 12020
+ip-26-0-165-131:1630205:1630205 [0] NCCL INFO cudaDriverVersion 12020
+ip-26-0-160-225:124163:124163 [3] NCCL INFO cudaDriverVersion 12020
+ip-26-0-168-34:272151:272151 [1] NCCL INFO Bootstrap : Using enp71s0:26.0.168.34<0>
+ip-26-0-165-131:1630206:1630206 [1] NCCL INFO cudaDriverVersion 12020
+ip-26-0-161-78:114514:114514 [2] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-165-131:1630209:1630209 [4] NCCL INFO cudaDriverVersion 12020
+ip-26-0-161-153:2636090:2636090 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-161-153:2636090:2636090 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-165-202:153562:153562 [0] NCCL INFO Bootstrap : Using enp71s0:26.0.165.202<0>
+ip-26-0-160-225:124163:124163 [3] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-165-131:1630208:1630208 [3] NCCL INFO cudaDriverVersion 12020
+ip-26-0-165-131:1630212:1630212 [7] NCCL INFO cudaDriverVersion 12020
+ip-26-0-168-34:272155:272155 [5] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-165-131:1630211:1630211 [6] NCCL INFO cudaDriverVersion 12020
+ip-26-0-168-52:2722051:2722051 [6] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-165-131:1630207:1630207 [2] NCCL INFO cudaDriverVersion 12020
+ip-26-0-168-52:2722052:2722052 [7] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-161-123:32706:32706 [2] NCCL INFO Bootstrap : Using enp71s0:26.0.161.123<0>
+ip-26-0-165-131:1630210:1630210 [5] NCCL INFO cudaDriverVersion 12020
+ip-26-0-168-52:2722047:2722047 [2] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-168-52:2722050:2722050 [5] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-168-52:2722045:2722045 [0] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-168-52:2722048:2722048 [3] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-165-164:1227464:1227464 [1] NCCL INFO Bootstrap : Using enp71s0:26.0.165.164<0>
+ip-26-0-165-164:1227469:1227469 [6] NCCL INFO Bootstrap : Using enp71s0:26.0.165.164<0>
+ip-26-0-168-52:2722049:2722049 [4] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-168-52:2722046:2722046 [1] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-161-78:114513:114513 [1] NCCL INFO cudaDriverVersion 12020
+ip-26-0-160-242:1364908:1364908 [6] NCCL INFO cudaDriverVersion 12020
+ip-26-0-161-78:114514:114514 [2] NCCL INFO Bootstrap : Using enp71s0:26.0.161.78<0>
+ip-26-0-168-34:272151:272151 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-168-34:272151:272151 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-161-123:32709:32709 [5] NCCL INFO cudaDriverVersion 12020
+ip-26-0-160-242:1364903:1364903 [1] NCCL INFO cudaDriverVersion 12020
+ip-26-0-165-164:1227463:1227463 [0] NCCL INFO cudaDriverVersion 12020
+ip-26-0-165-202:153568:153568 [6] NCCL INFO Bootstrap : Using enp71s0:26.0.165.202<0>
+ip-26-0-165-202:153565:153565 [3] NCCL INFO Bootstrap : Using enp71s0:26.0.165.202<0>
+ip-26-0-168-34:272150:272150 [0] NCCL INFO cudaDriverVersion 12020
+ip-26-0-161-78:114513:114513 [1] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-160-242:1364909:1364909 [7] NCCL INFO cudaDriverVersion 12020
+ip-26-0-160-242:1364902:1364902 [0] NCCL INFO cudaDriverVersion 12020
+ip-26-0-160-242:1364904:1364904 [2] NCCL INFO cudaDriverVersion 12020
+ip-26-0-165-164:1227463:1227463 [0] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-161-123:32704:32704 [0] NCCL INFO cudaDriverVersion 12020
+ip-26-0-161-123:32709:32709 [5] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-165-164:1227470:1227470 [7] NCCL INFO cudaDriverVersion 12020
+ip-26-0-168-34:272150:272150 [0] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-165-164:1227465:1227465 [2] NCCL INFO cudaDriverVersion 12020
+ip-26-0-161-78:114514:114514 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-161-78:114514:114514 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-165-164:1227466:1227466 [3] NCCL INFO cudaDriverVersion 12020
+ip-26-0-161-78:114517:114517 [5] NCCL INFO cudaDriverVersion 12020
+ip-26-0-168-34:272155:272155 [5] NCCL INFO Bootstrap : Using enp71s0:26.0.168.34<0>
+ip-26-0-165-59:1557266:1557266 [0] NCCL INFO cudaDriverVersion 12020
+ip-26-0-161-78:114513:114513 [1] NCCL INFO Bootstrap : Using enp71s0:26.0.161.78<0>
+ip-26-0-161-123:32704:32704 [0] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-161-123:32708:32708 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-161-123:32705:32705 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-161-123:32708:32708 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-161-123:32705:32705 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-165-164:1227468:1227468 [5] NCCL INFO cudaDriverVersion 12020
+ip-26-0-161-123:32706:32706 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-161-123:32707:32707 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-161-123:32706:32706 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-161-123:32707:32707 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-165-131:1630205:1630205 [0] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-165-59:1557267:1557267 [1] NCCL INFO cudaDriverVersion 12020
+ip-26-0-165-131:1630211:1630211 [6] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-165-202:153569:153569 [7] NCCL INFO Bootstrap : Using enp71s0:26.0.165.202<0>
+ip-26-0-165-131:1630208:1630208 [3] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-165-131:1630209:1630209 [4] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-165-164:1227463:1227463 [0] NCCL INFO Bootstrap : Using enp71s0:26.0.165.164<0>
+ip-26-0-165-202:153567:153567 [5] NCCL INFO Bootstrap : Using enp71s0:26.0.165.202<0>
+ip-26-0-165-202:153563:153563 [1] NCCL INFO Bootstrap : Using enp71s0:26.0.165.202<0>
+ip-26-0-165-131:1630207:1630207 [2] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-165-131:1630212:1630212 [7] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-168-34:272153:272153 [3] NCCL INFO cudaDriverVersion 12020
+ip-26-0-161-78:114516:114516 [4] NCCL INFO cudaDriverVersion 12020
+ip-26-0-165-164:1227465:1227465 [2] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-168-34:272152:272152 [2] NCCL INFO cudaDriverVersion 12020
+ip-26-0-165-131:1630206:1630206 [1] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-165-131:1630210:1630210 [5] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-165-59:1557266:1557266 [0] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-165-164:1227470:1227470 [7] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-165-164:1227467:1227467 [4] NCCL INFO cudaDriverVersion 12020
+ip-26-0-161-123:32711:32711 [7] NCCL INFO cudaDriverVersion 12020
+ip-26-0-160-242:1364906:1364906 [4] NCCL INFO cudaDriverVersion 12020
+ip-26-0-165-164:1227466:1227466 [3] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-161-78:114517:114517 [5] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-168-34:272150:272150 [0] NCCL INFO Bootstrap : Using enp71s0:26.0.168.34<0>
+ip-26-0-165-59:1557269:1557269 [3] NCCL INFO cudaDriverVersion 12020
+ip-26-0-160-242:1364908:1364908 [6] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-165-59:1557267:1557267 [1] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-160-242:1364904:1364904 [2] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-160-242:1364909:1364909 [7] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-160-242:1364902:1364902 [0] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-160-242:1364903:1364903 [1] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-161-78:114513:114513 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-165-164:1227468:1227468 [5] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-161-78:114513:114513 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-168-34:272152:272152 [2] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-160-242:1364905:1364905 [3] NCCL INFO cudaDriverVersion 12020
+ip-26-0-165-202:153566:153566 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-165-202:153568:153568 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-165-202:153565:153565 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-165-202:153566:153566 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-165-202:153568:153568 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-165-202:153565:153565 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-161-78:114516:114516 [4] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-168-34:272153:272153 [3] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-165-202:153562:153562 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-165-202:153564:153564 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-165-202:153562:153562 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-165-202:153564:153564 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-168-34:272154:272154 [4] NCCL INFO cudaDriverVersion 12020
+ip-26-0-168-34:272155:272155 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-165-59:1557269:1557269 [3] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-168-34:272155:272155 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-160-242:1364906:1364906 [4] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-161-123:32711:32711 [7] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-165-164:1227467:1227467 [4] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-160-242:1364907:1364907 [5] NCCL INFO cudaDriverVersion 12020
+ip-26-0-161-153:2636087:2636087 [3] NCCL INFO cudaDriverVersion 12020
+ip-26-0-165-202:153567:153567 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-165-202:153567:153567 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-168-34:272150:272150 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-161-153:2636086:2636086 [2] NCCL INFO cudaDriverVersion 12020
+ip-26-0-165-202:153569:153569 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-165-202:153563:153563 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-165-202:153569:153569 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-165-202:153563:153563 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-168-34:272150:272150 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-160-242:1364905:1364905 [3] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-161-153:2636089:2636089 [5] NCCL INFO cudaDriverVersion 12020
+ip-26-0-160-242:1364907:1364907 [5] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-168-34:272154:272154 [4] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-161-153:2636085:2636085 [1] NCCL INFO cudaDriverVersion 12020
+ip-26-0-161-153:2636087:2636087 [3] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-161-123:32709:32709 [5] NCCL INFO Bootstrap : Using enp71s0:26.0.161.123<0>
+ip-26-0-161-153:2636086:2636086 [2] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-168-30:235651:235651 [3] NCCL INFO Bootstrap : Using enp71s0:26.0.168.30<0>
+ip-26-0-161-153:2636084:2636084 [0] NCCL INFO cudaDriverVersion 12020
+ip-26-0-165-59:1557266:1557266 [0] NCCL INFO Bootstrap : Using enp71s0:26.0.165.59<0>
+ip-26-0-160-225:124166:124166 [6] NCCL INFO Bootstrap : Using enp71s0:26.0.160.225<0>
+ip-26-0-161-153:2636089:2636089 [5] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-168-30:235655:235655 [7] NCCL INFO Bootstrap : Using enp71s0:26.0.168.30<0>
+ip-26-0-161-123:32704:32704 [0] NCCL INFO Bootstrap : Using enp71s0:26.0.161.123<0>
+ip-26-0-168-30:235648:235648 [0] NCCL INFO Bootstrap : Using enp71s0:26.0.168.30<0>
+ip-26-0-168-30:235650:235650 [2] NCCL INFO Bootstrap : Using enp71s0:26.0.168.30<0>
+ip-26-0-168-30:235654:235654 [6] NCCL INFO Bootstrap : Using enp71s0:26.0.168.30<0>
+ip-26-0-168-30:235653:235653 [5] NCCL INFO Bootstrap : Using enp71s0:26.0.168.30<0>
+ip-26-0-168-30:235649:235649 [1] NCCL INFO Bootstrap : Using enp71s0:26.0.168.30<0>
+ip-26-0-168-30:235652:235652 [4] NCCL INFO Bootstrap : Using enp71s0:26.0.168.30<0>
+ip-26-0-165-59:1557267:1557267 [1] NCCL INFO Bootstrap : Using enp71s0:26.0.165.59<0>
+ip-26-0-161-153:2636085:2636085 [1] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-165-59:1557269:1557269 [3] NCCL INFO Bootstrap : Using enp71s0:26.0.165.59<0>
+ip-26-0-161-153:2636084:2636084 [0] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-168-34:272152:272152 [2] NCCL INFO Bootstrap : Using enp71s0:26.0.168.34<0>
+ip-26-0-168-34:272153:272153 [3] NCCL INFO Bootstrap : Using enp71s0:26.0.168.34<0>
+ip-26-0-161-78:114517:114517 [5] NCCL INFO Bootstrap : Using enp71s0:26.0.161.78<0>
+ip-26-0-161-123:32710:32710 [6] NCCL INFO cudaDriverVersion 12020
+ip-26-0-161-153:2636091:2636091 [7] NCCL INFO cudaDriverVersion 12020
+ip-26-0-161-78:114516:114516 [4] NCCL INFO Bootstrap : Using enp71s0:26.0.161.78<0>
+ip-26-0-161-153:2636088:2636088 [4] NCCL INFO cudaDriverVersion 12020
+ip-26-0-161-123:32711:32711 [7] NCCL INFO Bootstrap : Using enp71s0:26.0.161.123<0>
+ip-26-0-168-34:272154:272154 [4] NCCL INFO Bootstrap : Using enp71s0:26.0.168.34<0>
+ip-26-0-165-164:1227463:1227463 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-165-164:1227464:1227464 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-165-164:1227463:1227463 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-165-164:1227464:1227464 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-165-164:1227469:1227469 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-165-59:1557266:1557266 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-165-164:1227469:1227469 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-165-59:1557266:1557266 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-161-123:32709:32709 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-161-123:32704:32704 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-161-123:32709:32709 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-161-123:32704:32704 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-165-59:1557267:1557267 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-165-59:1557267:1557267 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-165-59:1557270:1557270 [4] NCCL INFO cudaDriverVersion 12020
+ip-26-0-161-123:32710:32710 [6] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-165-59:1557269:1557269 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-165-59:1557269:1557269 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-160-225:124164:124164 [4] NCCL INFO Bootstrap : Using enp71s0:26.0.160.225<0>
+ip-26-0-160-225:124167:124167 [7] NCCL INFO Bootstrap : Using enp71s0:26.0.160.225<0>
+ip-26-0-160-225:124162:124162 [2] NCCL INFO Bootstrap : Using enp71s0:26.0.160.225<0>
+ip-26-0-160-225:124166:124166 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-160-225:124165:124165 [5] NCCL INFO Bootstrap : Using enp71s0:26.0.160.225<0>
+ip-26-0-160-225:124161:124161 [1] NCCL INFO Bootstrap : Using enp71s0:26.0.160.225<0>
+ip-26-0-160-225:124166:124166 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-160-225:124163:124163 [3] NCCL INFO Bootstrap : Using enp71s0:26.0.160.225<0>
+ip-26-0-168-34:272152:272152 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-168-34:272153:272153 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-168-34:272152:272152 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-168-34:272153:272153 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-161-153:2636088:2636088 [4] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-161-153:2636091:2636091 [7] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-161-103:96895:96895 [7] NCCL INFO Bootstrap : Using enp71s0:26.0.161.103<0>
+ip-26-0-161-103:96893:96893 [5] NCCL INFO Bootstrap : Using enp71s0:26.0.161.103<0>
+ip-26-0-161-103:96890:96890 [2] NCCL INFO Bootstrap : Using enp71s0:26.0.161.103<0>
+ip-26-0-161-103:96892:96892 [4] NCCL INFO Bootstrap : Using enp71s0:26.0.161.103<0>
+ip-26-0-161-103:96888:96888 [0] NCCL INFO Bootstrap : Using enp71s0:26.0.161.103<0>
+ip-26-0-161-103:96889:96889 [1] NCCL INFO Bootstrap : Using enp71s0:26.0.161.103<0>
+ip-26-0-161-103:96891:96891 [3] NCCL INFO Bootstrap : Using enp71s0:26.0.161.103<0>
+ip-26-0-161-103:96894:96894 [6] NCCL INFO Bootstrap : Using enp71s0:26.0.161.103<0>
+ip-26-0-161-142:62885:62885 [4] NCCL INFO Bootstrap : Using enp71s0:26.0.161.142<0>
+ip-26-0-161-78:114516:114516 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-161-78:114516:114516 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-161-142:62887:62887 [6] NCCL INFO Bootstrap : Using enp71s0:26.0.161.142<0>
+ip-26-0-161-142:62888:62888 [7] NCCL INFO Bootstrap : Using enp71s0:26.0.161.142<0>
+ip-26-0-161-142:62886:62886 [5] NCCL INFO Bootstrap : Using enp71s0:26.0.161.142<0>
+ip-26-0-161-142:62882:62882 [1] NCCL INFO Bootstrap : Using enp71s0:26.0.161.142<0>
+ip-26-0-161-142:62884:62884 [3] NCCL INFO Bootstrap : Using enp71s0:26.0.161.142<0>
+ip-26-0-161-142:62881:62881 [0] NCCL INFO Bootstrap : Using enp71s0:26.0.161.142<0>
+ip-26-0-161-142:62883:62883 [2] NCCL INFO Bootstrap : Using enp71s0:26.0.161.142<0>
+ip-26-0-165-59:1557270:1557270 [4] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-168-34:272154:272154 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-168-34:272154:272154 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-161-123:32710:32710 [6] NCCL INFO Bootstrap : Using enp71s0:26.0.161.123<0>
+ip-26-0-161-78:114517:114517 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-161-78:114517:114517 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-161-123:32711:32711 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-161-123:32711:32711 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-165-164:1227465:1227465 [2] NCCL INFO Bootstrap : Using enp71s0:26.0.165.164<0>
+ip-26-0-165-59:1557268:1557268 [2] NCCL INFO cudaDriverVersion 12020
+ip-26-0-165-164:1227470:1227470 [7] NCCL INFO Bootstrap : Using enp71s0:26.0.165.164<0>
+ip-26-0-165-164:1227466:1227466 [3] NCCL INFO Bootstrap : Using enp71s0:26.0.165.164<0>
+ip-26-0-165-164:1227468:1227468 [5] NCCL INFO Bootstrap : Using enp71s0:26.0.165.164<0>
+ip-26-0-165-164:1227467:1227467 [4] NCCL INFO Bootstrap : Using enp71s0:26.0.165.164<0>
+ip-26-0-165-59:1557270:1557270 [4] NCCL INFO Bootstrap : Using enp71s0:26.0.165.59<0>
+ip-26-0-165-59:1557272:1557272 [6] NCCL INFO cudaDriverVersion 12020
+ip-26-0-160-225:124164:124164 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-160-225:124167:124167 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-160-225:124162:124162 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-160-225:124163:124163 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-160-225:124165:124165 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-160-225:124164:124164 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-160-225:124167:124167 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-160-225:124162:124162 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-160-225:124165:124165 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-161-123:32710:32710 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-161-123:32710:32710 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-160-225:124163:124163 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-160-225:124161:124161 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-160-225:124161:124161 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-165-164:1227465:1227465 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-165-164:1227465:1227465 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-165-59:1557268:1557268 [2] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-161-178:2815966:2815966 [3] NCCL INFO Bootstrap : Using enp73s0:26.0.175.184<0>
+ip-26-0-161-178:2815963:2815963 [0] NCCL INFO Bootstrap : Using enp73s0:26.0.175.184<0>
+ip-26-0-161-178:2815964:2815964 [1] NCCL INFO Bootstrap : Using enp73s0:26.0.175.184<0>
+ip-26-0-165-59:1557272:1557272 [6] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-161-178:2815967:2815967 [4] NCCL INFO Bootstrap : Using enp73s0:26.0.175.184<0>
+ip-26-0-161-178:2815968:2815968 [5] NCCL INFO Bootstrap : Using enp73s0:26.0.175.184<0>
+ip-26-0-161-178:2815965:2815965 [2] NCCL INFO Bootstrap : Using enp73s0:26.0.175.184<0>
+ip-26-0-161-178:2815969:2815969 [6] NCCL INFO Bootstrap : Using enp73s0:26.0.175.184<0>
+ip-26-0-161-178:2815970:2815970 [7] NCCL INFO Bootstrap : Using enp73s0:26.0.175.184<0>
+ip-26-0-165-164:1227470:1227470 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-165-164:1227470:1227470 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-165-59:1557270:1557270 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-165-59:1557270:1557270 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-165-164:1227468:1227468 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-165-164:1227466:1227466 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-165-164:1227468:1227468 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-165-164:1227466:1227466 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-165-164:1227467:1227467 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-165-59:1557268:1557268 [2] NCCL INFO Bootstrap : Using enp71s0:26.0.165.59<0>
+ip-26-0-165-164:1227467:1227467 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-165-59:1557272:1557272 [6] NCCL INFO Bootstrap : Using enp71s0:26.0.165.59<0>
+ip-26-0-161-138:96569:96569 [6] NCCL INFO Bootstrap : Using enp74s0:26.0.173.132<0>
+ip-26-0-168-52:2722051:2722051 [6] NCCL INFO Bootstrap : Using enp71s0:26.0.168.52<0>
+ip-26-0-168-52:2722052:2722052 [7] NCCL INFO Bootstrap : Using enp71s0:26.0.168.52<0>
+ip-26-0-168-52:2722047:2722047 [2] NCCL INFO Bootstrap : Using enp71s0:26.0.168.52<0>
+ip-26-0-168-52:2722050:2722050 [5] NCCL INFO Bootstrap : Using enp71s0:26.0.168.52<0>
+ip-26-0-168-52:2722045:2722045 [0] NCCL INFO Bootstrap : Using enp71s0:26.0.168.52<0>
+ip-26-0-168-52:2722046:2722046 [1] NCCL INFO Bootstrap : Using enp71s0:26.0.168.52<0>
+ip-26-0-168-52:2722049:2722049 [4] NCCL INFO Bootstrap : Using enp71s0:26.0.168.52<0>
+ip-26-0-168-52:2722048:2722048 [3] NCCL INFO Bootstrap : Using enp71s0:26.0.168.52<0>
+ip-26-0-161-138:96565:96565 [2] NCCL INFO Bootstrap : Using enp74s0:26.0.173.132<0>
+ip-26-0-161-138:96568:96568 [5] NCCL INFO Bootstrap : Using enp74s0:26.0.173.132<0>
+ip-26-0-161-138:96567:96567 [4] NCCL INFO Bootstrap : Using enp74s0:26.0.173.132<0>
+ip-26-0-161-138:96563:96563 [0] NCCL INFO Bootstrap : Using enp74s0:26.0.173.132<0>
+ip-26-0-161-138:96564:96564 [1] NCCL INFO Bootstrap : Using enp74s0:26.0.173.132<0>
+ip-26-0-165-59:1557268:1557268 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-161-138:96570:96570 [7] NCCL INFO Bootstrap : Using enp74s0:26.0.173.132<0>
+ip-26-0-161-138:96566:96566 [3] NCCL INFO Bootstrap : Using enp74s0:26.0.173.132<0>
+ip-26-0-165-59:1557268:1557268 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-165-59:1557272:1557272 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-165-59:1557272:1557272 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-168-30:235650:235650 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-168-30:235649:235649 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-168-30:235653:235653 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-168-30:235652:235652 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-168-30:235651:235651 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-168-30:235655:235655 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-168-30:235650:235650 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-168-30:235653:235653 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-168-30:235649:235649 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-168-30:235651:235651 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-168-30:235652:235652 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-168-30:235655:235655 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-168-30:235648:235648 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-168-30:235648:235648 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-168-30:235654:235654 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-168-30:235654:235654 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-165-131:1630205:1630205 [0] NCCL INFO Bootstrap : Using enp71s0:26.0.165.131<0>
+ip-26-0-165-131:1630211:1630211 [6] NCCL INFO Bootstrap : Using enp71s0:26.0.165.131<0>
+ip-26-0-165-131:1630210:1630210 [5] NCCL INFO Bootstrap : Using enp71s0:26.0.165.131<0>
+ip-26-0-165-131:1630209:1630209 [4] NCCL INFO Bootstrap : Using enp71s0:26.0.165.131<0>
+ip-26-0-165-131:1630208:1630208 [3] NCCL INFO Bootstrap : Using enp71s0:26.0.165.131<0>
+ip-26-0-165-131:1630206:1630206 [1] NCCL INFO Bootstrap : Using enp71s0:26.0.165.131<0>
+ip-26-0-160-242:1364908:1364908 [6] NCCL INFO Bootstrap : Using enp71s0:26.0.160.242<0>
+ip-26-0-165-131:1630212:1630212 [7] NCCL INFO Bootstrap : Using enp71s0:26.0.165.131<0>
+ip-26-0-165-131:1630207:1630207 [2] NCCL INFO Bootstrap : Using enp71s0:26.0.165.131<0>
+ip-26-0-160-242:1364909:1364909 [7] NCCL INFO Bootstrap : Using enp71s0:26.0.160.242<0>
+ip-26-0-160-242:1364904:1364904 [2] NCCL INFO Bootstrap : Using enp71s0:26.0.160.242<0>
+ip-26-0-160-242:1364902:1364902 [0] NCCL INFO Bootstrap : Using enp71s0:26.0.160.242<0>
+ip-26-0-160-242:1364903:1364903 [1] NCCL INFO Bootstrap : Using enp71s0:26.0.160.242<0>
+ip-26-0-160-242:1364906:1364906 [4] NCCL INFO Bootstrap : Using enp71s0:26.0.160.242<0>
+ip-26-0-160-242:1364907:1364907 [5] NCCL INFO Bootstrap : Using enp71s0:26.0.160.242<0>
+ip-26-0-160-242:1364905:1364905 [3] NCCL INFO Bootstrap : Using enp71s0:26.0.160.242<0>
+ip-26-0-161-103:96893:96893 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-161-103:96892:96892 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-161-103:96889:96889 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-161-103:96888:96888 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-161-103:96890:96890 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-161-103:96893:96893 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-161-103:96892:96892 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-161-103:96889:96889 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-161-103:96888:96888 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-161-103:96890:96890 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-161-103:96891:96891 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-161-103:96891:96891 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-161-103:96895:96895 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-161-103:96894:96894 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-161-103:96895:96895 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-161-103:96894:96894 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-161-142:62883:62883 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-161-142:62881:62881 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-161-142:62886:62886 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-161-142:62882:62882 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-161-142:62884:62884 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-161-153:2636087:2636087 [3] NCCL INFO Bootstrap : Using enp71s0:26.0.161.153<0>
+ip-26-0-161-142:62883:62883 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-161-142:62881:62881 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-161-142:62886:62886 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-161-142:62884:62884 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-161-142:62882:62882 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-161-142:62888:62888 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-161-142:62887:62887 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-161-142:62885:62885 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-161-142:62887:62887 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-161-142:62888:62888 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-161-142:62885:62885 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-161-153:2636086:2636086 [2] NCCL INFO Bootstrap : Using enp71s0:26.0.161.153<0>
+ip-26-0-161-153:2636085:2636085 [1] NCCL INFO Bootstrap : Using enp71s0:26.0.161.153<0>
+ip-26-0-161-153:2636089:2636089 [5] NCCL INFO Bootstrap : Using enp71s0:26.0.161.153<0>
+ip-26-0-161-153:2636084:2636084 [0] NCCL INFO Bootstrap : Using enp71s0:26.0.161.153<0>
+ip-26-0-161-153:2636088:2636088 [4] NCCL INFO Bootstrap : Using enp71s0:26.0.161.153<0>
+ip-26-0-161-153:2636091:2636091 [7] NCCL INFO Bootstrap : Using enp71s0:26.0.161.153<0>
+ip-26-0-161-178:2815963:2815963 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-161-178:2815967:2815967 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-161-178:2815964:2815964 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-161-178:2815967:2815967 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-161-178:2815964:2815964 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-161-178:2815963:2815963 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-161-178:2815968:2815968 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-161-178:2815968:2815968 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-161-153:2636087:2636087 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-161-153:2636085:2636085 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-161-153:2636087:2636087 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-161-153:2636085:2636085 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-161-153:2636089:2636089 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-161-153:2636084:2636084 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-161-153:2636089:2636089 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-161-153:2636084:2636084 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-161-178:2815966:2815966 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-161-178:2815966:2815966 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-161-178:2815969:2815969 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-161-178:2815965:2815965 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-161-178:2815969:2815969 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-161-178:2815965:2815965 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-161-178:2815970:2815970 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-161-178:2815970:2815970 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-161-153:2636086:2636086 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-161-153:2636086:2636086 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-161-153:2636088:2636088 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-161-153:2636091:2636091 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-161-153:2636091:2636091 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-161-153:2636088:2636088 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-161-138:96569:96569 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-161-138:96566:96566 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-161-138:96569:96569 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-161-138:96566:96566 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-161-138:96565:96565 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-161-138:96567:96567 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-161-138:96568:96568 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-161-138:96564:96564 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-161-138:96565:96565 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-161-138:96567:96567 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-161-138:96568:96568 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-161-138:96564:96564 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-161-138:96570:96570 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-161-138:96570:96570 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-168-52:2722050:2722050 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-168-52:2722050:2722050 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-161-138:96563:96563 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-161-138:96563:96563 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-168-52:2722047:2722047 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-168-52:2722052:2722052 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-168-52:2722045:2722045 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-168-52:2722049:2722049 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-168-52:2722047:2722047 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-168-52:2722052:2722052 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-168-52:2722045:2722045 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-168-52:2722049:2722049 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-168-52:2722046:2722046 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-168-52:2722046:2722046 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-168-52:2722048:2722048 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-168-52:2722048:2722048 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-168-52:2722051:2722051 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-168-52:2722051:2722051 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-165-131:1630206:1630206 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-165-131:1630206:1630206 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-160-242:1364902:1364902 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-160-242:1364909:1364909 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-160-242:1364909:1364909 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-160-242:1364902:1364902 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-165-131:1630205:1630205 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-165-131:1630205:1630205 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-160-242:1364906:1364906 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-165-131:1630210:1630210 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-165-131:1630210:1630210 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-160-242:1364906:1364906 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-165-131:1630208:1630208 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-165-131:1630207:1630207 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-165-131:1630211:1630211 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-165-131:1630212:1630212 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-165-131:1630208:1630208 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-165-131:1630207:1630207 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-165-131:1630212:1630212 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-165-131:1630211:1630211 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-160-242:1364907:1364907 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-160-242:1364908:1364908 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-160-242:1364907:1364907 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-160-242:1364908:1364908 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-165-131:1630209:1630209 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-165-131:1630209:1630209 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-160-242:1364904:1364904 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-160-242:1364904:1364904 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-160-242:1364905:1364905 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-160-242:1364903:1364903 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-160-242:1364905:1364905 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-160-242:1364903:1364903 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-160-225:124160:124335 [0] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-160-225:124160:124335 [0] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-160-225:124160:124335 [0] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-160-225:124160:124335 [0] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-160-225:124160:124335 [0] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-160-225:124160:124335 [0] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-160-225:124160:124335 [0] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-160-225:124160:124335 [0] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-160-225:124160:124335 [0] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-161-123:32706:32877 [2] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-161-123:32706:32877 [2] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-161-123:32707:32878 [3] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-161-123:32707:32878 [3] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-168-34:272156:272325 [6] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-168-34:272156:272325 [6] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-161-142:62885:63054 [4] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-161-142:62885:63054 [4] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-168-34:272156:272325 [6] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-168-34:272156:272325 [6] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-168-34:272156:272325 [6] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-168-34:272156:272325 [6] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-168-34:272156:272325 [6] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-168-34:272156:272325 [6] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-168-34:272156:272325 [6] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-160-225:124162:124342 [2] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-160-225:124162:124342 [2] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-161-142:62885:63054 [4] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-161-142:62885:63054 [4] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-161-142:62885:63054 [4] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-161-142:62885:63054 [4] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-161-142:62885:63054 [4] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-161-142:62885:63054 [4] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-161-142:62885:63054 [4] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-160-225:124162:124342 [2] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-160-225:124162:124342 [2] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-161-123:32707:32878 [3] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-161-123:32707:32878 [3] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-161-123:32706:32877 [2] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-161-123:32706:32877 [2] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-160-225:124162:124342 [2] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-160-225:124162:124342 [2] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-160-225:124162:124342 [2] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-161-123:32707:32878 [3] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-161-123:32707:32878 [3] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-160-225:124162:124342 [2] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-160-225:124162:124342 [2] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-161-123:32707:32878 [3] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-161-123:32707:32878 [3] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-161-123:32707:32878 [3] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-161-123:32706:32877 [2] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-161-123:32706:32877 [2] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-161-123:32706:32877 [2] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-161-123:32706:32877 [2] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-161-123:32706:32877 [2] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-160-225:124165:124341 [5] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-160-225:124165:124341 [5] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-168-30:235648:235826 [0] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-168-30:235648:235826 [0] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-161-142:62887:63055 [6] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-161-142:62887:63055 [6] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-165-202:153568:153736 [6] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-165-202:153568:153736 [6] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-168-30:235648:235826 [0] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-168-30:235648:235826 [0] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-168-30:235648:235826 [0] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-168-30:235648:235826 [0] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-168-30:235648:235826 [0] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-168-30:235648:235826 [0] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-168-30:235648:235826 [0] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-165-202:153565:153738 [3] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-165-202:153565:153738 [3] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-168-30:235653:235825 [5] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-168-30:235653:235825 [5] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-161-78:114517:114687 [5] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-161-78:114517:114687 [5] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-161-142:62887:63055 [6] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-161-142:62887:63055 [6] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-161-142:62887:63055 [6] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-161-142:62887:63055 [6] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-161-142:62887:63055 [6] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-161-142:62887:63055 [6] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-161-142:62887:63055 [6] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-160-225:124166:124339 [6] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-160-225:124166:124339 [6] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-160-225:124165:124341 [5] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-160-225:124165:124341 [5] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-160-225:124165:124341 [5] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-160-225:124165:124341 [5] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-160-225:124165:124341 [5] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-160-225:124165:124341 [5] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-160-225:124165:124341 [5] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-165-202:153564:153737 [2] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-165-202:153564:153737 [2] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-168-34:272157:272324 [7] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-168-34:272157:272324 [7] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-161-138:96568:96753 [5] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-161-138:96568:96753 [5] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-161-78:114517:114687 [5] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-165-202:153562:153735 [0] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-165-202:153562:153735 [0] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-161-78:114517:114687 [5] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-161-78:114517:114687 [5] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-161-78:114517:114687 [5] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-161-78:114517:114687 [5] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-161-78:114517:114687 [5] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-161-78:114517:114687 [5] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-168-30:235654:235821 [6] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-168-30:235654:235821 [6] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-168-34:272154:272330 [4] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-168-34:272154:272330 [4] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-161-142:62883:63058 [2] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-161-142:62883:63058 [2] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-160-225:124164:124336 [4] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-160-225:124164:124336 [4] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-160-225:124161:124337 [1] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-160-225:124161:124337 [1] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-161-123:32709:32882 [5] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-161-123:32709:32882 [5] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-161-103:96893:97067 [5] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-161-103:96893:97067 [5] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-165-202:153568:153736 [6] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-165-202:153568:153736 [6] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-165-202:153568:153736 [6] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-165-202:153568:153736 [6] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-165-202:153568:153736 [6] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-165-202:153568:153736 [6] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-165-202:153568:153736 [6] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-161-123:32711:32884 [7] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-161-123:32711:32884 [7] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-161-78:114514:114689 [2] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-161-78:114514:114689 [2] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-161-123:32704:32881 [0] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-161-123:32704:32881 [0] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-161-138:96568:96753 [5] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-161-138:96568:96753 [5] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-161-138:96568:96753 [5] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-161-138:96568:96753 [5] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-161-138:96568:96753 [5] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-161-138:96568:96753 [5] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-161-138:96568:96753 [5] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-161-142:62882:63056 [1] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-161-142:62882:63056 [1] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-160-225:124163:124340 [3] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-160-225:124163:124340 [3] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-161-142:62886:63057 [5] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-161-142:62886:63057 [5] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-160-225:124167:124338 [7] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-160-225:124167:124338 [7] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-161-142:62883:63058 [2] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-161-142:62883:63058 [2] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-161-142:62883:63058 [2] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-161-142:62883:63058 [2] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-161-142:62883:63058 [2] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-161-142:62883:63058 [2] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-161-142:62883:63058 [2] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-168-34:272153:272328 [3] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-168-34:272153:272328 [3] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-161-103:96888:97066 [0] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-161-103:96888:97066 [0] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-168-30:235653:235825 [5] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-168-30:235653:235825 [5] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-168-30:235653:235825 [5] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-168-30:235653:235825 [5] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-168-30:235653:235825 [5] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-168-30:235653:235825 [5] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-168-30:235653:235825 [5] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-161-78:114512:114686 [0] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-161-78:114512:114686 [0] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-161-123:32710:32883 [6] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-161-123:32710:32883 [6] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-161-103:96893:97067 [5] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-161-103:96893:97067 [5] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-161-103:96893:97067 [5] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-161-103:96893:97067 [5] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-161-103:96893:97067 [5] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-161-103:96893:97067 [5] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-161-103:96893:97067 [5] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-165-202:153565:153738 [3] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-165-202:153565:153738 [3] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-165-202:153565:153738 [3] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-165-202:153565:153738 [3] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-165-202:153565:153738 [3] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-165-202:153565:153738 [3] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-165-202:153565:153738 [3] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-168-34:272157:272324 [7] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-168-34:272157:272324 [7] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-168-34:272157:272324 [7] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-161-138:96569:96758 [6] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-161-138:96569:96758 [6] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-168-34:272157:272324 [7] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-168-34:272157:272324 [7] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-168-34:272157:272324 [7] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-168-34:272157:272324 [7] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-168-34:272155:272329 [5] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-168-34:272155:272329 [5] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-168-30:235654:235821 [6] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-168-30:235654:235821 [6] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-161-103:96888:97066 [0] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-168-30:235654:235821 [6] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-168-30:235654:235821 [6] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-161-103:96888:97066 [0] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-168-30:235654:235821 [6] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-161-103:96888:97066 [0] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-168-30:235654:235821 [6] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-168-30:235654:235821 [6] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-161-103:96888:97066 [0] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-161-103:96888:97066 [0] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-161-103:96888:97066 [0] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-161-103:96888:97066 [0] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-160-225:124166:124339 [6] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-160-225:124166:124339 [6] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-160-225:124166:124339 [6] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-160-225:124166:124339 [6] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-160-225:124166:124339 [6] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-160-225:124166:124339 [6] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-160-225:124166:124339 [6] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-165-202:153569:153739 [7] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-165-202:153569:153739 [7] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-161-123:32705:32880 [1] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-161-123:32705:32880 [1] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-168-34:272154:272330 [4] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-168-34:272154:272330 [4] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-168-34:272154:272330 [4] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-168-34:272154:272330 [4] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-168-34:272154:272330 [4] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-168-34:272154:272330 [4] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-168-34:272154:272330 [4] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-168-30:235652:235822 [4] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-168-30:235652:235822 [4] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-161-138:96567:96760 [4] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-161-138:96567:96760 [4] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-168-34:272150:272327 [0] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-168-34:272150:272327 [0] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-161-103:96889:97065 [1] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-161-103:96889:97065 [1] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-165-202:153564:153737 [2] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-165-202:153564:153737 [2] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-168-34:272151:272326 [1] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-168-34:272151:272326 [1] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-165-202:153564:153737 [2] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-165-202:153564:153737 [2] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-165-202:153564:153737 [2] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-165-202:153564:153737 [2] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-165-202:153564:153737 [2] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-161-142:62881:63059 [0] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-161-142:62881:63059 [0] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-165-202:153567:153742 [5] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-165-202:153567:153742 [5] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-161-142:62884:63061 [3] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-161-142:62884:63061 [3] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-161-103:96895:97069 [7] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-161-103:96895:97069 [7] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-165-202:153562:153735 [0] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-165-202:153562:153735 [0] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-165-202:153562:153735 [0] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-165-202:153562:153735 [0] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-165-202:153562:153735 [0] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-165-202:153562:153735 [0] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-165-202:153562:153735 [0] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-161-123:32709:32882 [5] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-161-123:32709:32882 [5] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-161-123:32709:32882 [5] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-161-123:32709:32882 [5] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-161-123:32709:32882 [5] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-161-123:32709:32882 [5] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-161-123:32709:32882 [5] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-161-142:62882:63056 [1] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-168-52:2722046:2722223 [1] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-168-52:2722046:2722223 [1] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-161-142:62882:63056 [1] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-161-142:62882:63056 [1] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-161-142:62882:63056 [1] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-161-142:62882:63056 [1] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-161-123:32711:32884 [7] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-161-142:62882:63056 [1] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-161-142:62882:63056 [1] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-161-78:114514:114689 [2] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-161-123:32711:32884 [7] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-161-78:114514:114689 [2] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-161-123:32711:32884 [7] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-161-123:32711:32884 [7] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-161-123:32711:32884 [7] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-161-123:32711:32884 [7] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-161-78:114514:114689 [2] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-161-78:114514:114689 [2] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-161-78:114514:114689 [2] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-161-123:32711:32884 [7] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-161-78:114514:114689 [2] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-161-78:114514:114689 [2] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-161-103:96892:97064 [4] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-161-103:96892:97064 [4] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-168-30:235649:235827 [1] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-168-30:235649:235827 [1] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-161-142:62886:63057 [5] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-161-142:62886:63057 [5] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-161-142:62886:63057 [5] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-161-78:114518:114693 [6] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-161-78:114518:114693 [6] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-161-142:62886:63057 [5] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-161-142:62886:63057 [5] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-161-142:62886:63057 [5] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-161-142:62886:63057 [5] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-161-138:96569:96758 [6] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-161-138:96569:96758 [6] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-161-138:96569:96758 [6] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-161-138:96569:96758 [6] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-161-138:96569:96758 [6] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-161-138:96569:96758 [6] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-161-138:96569:96758 [6] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-168-30:235650:235824 [2] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-168-30:235650:235824 [2] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-161-78:114519:114692 [7] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-161-78:114519:114692 [7] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-161-123:32704:32881 [0] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-161-123:32704:32881 [0] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-161-123:32704:32881 [0] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-161-123:32704:32881 [0] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-161-123:32704:32881 [0] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-161-123:32704:32881 [0] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-161-123:32704:32881 [0] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-168-30:235651:235823 [3] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-168-30:235651:235823 [3] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-165-202:153566:153741 [4] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-165-202:153566:153741 [4] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-161-138:96566:96755 [3] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-161-138:96566:96755 [3] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-161-138:96567:96760 [4] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-161-138:96567:96760 [4] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-161-138:96567:96760 [4] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-161-138:96567:96760 [4] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-161-138:96567:96760 [4] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-161-138:96567:96760 [4] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-161-138:96567:96760 [4] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-161-78:114512:114686 [0] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-165-202:153563:153740 [1] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-165-202:153563:153740 [1] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-161-78:114512:114686 [0] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-161-78:114512:114686 [0] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-161-78:114512:114686 [0] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-161-78:114512:114686 [0] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-161-78:114512:114686 [0] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-161-78:114512:114686 [0] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-161-138:96570:96759 [7] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-161-138:96570:96759 [7] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-168-34:272152:272331 [2] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-168-34:272152:272331 [2] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-160-225:124164:124336 [4] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-160-225:124164:124336 [4] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-160-225:124164:124336 [4] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-160-225:124164:124336 [4] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-160-225:124164:124336 [4] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-160-225:124164:124336 [4] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-160-225:124164:124336 [4] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-160-225:124161:124337 [1] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-160-225:124161:124337 [1] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-168-30:235652:235822 [4] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-160-225:124161:124337 [1] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-160-225:124161:124337 [1] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-160-225:124161:124337 [1] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-160-225:124161:124337 [1] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-160-225:124161:124337 [1] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-168-30:235652:235822 [4] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-168-30:235652:235822 [4] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-168-30:235652:235822 [4] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-168-30:235652:235822 [4] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-168-30:235652:235822 [4] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-168-30:235652:235822 [4] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-161-123:32710:32883 [6] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-161-123:32710:32883 [6] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-161-123:32710:32883 [6] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-161-123:32710:32883 [6] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-161-123:32710:32883 [6] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-161-123:32710:32883 [6] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-161-123:32710:32883 [6] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-168-52:2722048:2722220 [3] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-168-52:2722048:2722220 [3] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-160-225:124163:124340 [3] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-168-34:272153:272328 [3] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-160-225:124163:124340 [3] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-160-225:124163:124340 [3] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-168-34:272153:272328 [3] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-160-225:124163:124340 [3] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-160-225:124163:124340 [3] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-168-34:272153:272328 [3] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-160-225:124163:124340 [3] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-160-225:124163:124340 [3] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-168-34:272153:272328 [3] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-168-34:272153:272328 [3] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-160-225:124167:124338 [7] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-160-225:124167:124338 [7] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-168-34:272153:272328 [3] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-168-34:272153:272328 [3] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-160-225:124167:124338 [7] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-161-78:114516:114691 [4] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-161-78:114516:114691 [4] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-160-225:124167:124338 [7] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-160-225:124167:124338 [7] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-160-225:124167:124338 [7] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-160-225:124167:124338 [7] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-161-103:96889:97065 [1] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-168-52:2722050:2722225 [5] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-168-52:2722050:2722225 [5] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-161-103:96889:97065 [1] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-161-103:96889:97065 [1] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-161-103:96889:97065 [1] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-161-103:96889:97065 [1] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-161-103:96889:97065 [1] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-161-103:96889:97065 [1] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-161-142:62881:63059 [0] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-161-142:62881:63059 [0] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-161-142:62881:63059 [0] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-161-142:62881:63059 [0] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-161-142:62881:63059 [0] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-161-142:62881:63059 [0] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-161-142:62881:63059 [0] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-160-225:124162:124342 [2] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-161-142:62884:63061 [3] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-161-142:62884:63061 [3] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-161-142:62884:63061 [3] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-161-123:32705:32880 [1] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-161-142:62884:63061 [3] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-161-142:62884:63061 [3] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-168-52:2722051:2722224 [6] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-168-52:2722051:2722224 [6] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-161-142:62884:63061 [3] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-161-142:62884:63061 [3] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-161-103:96895:97069 [7] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-161-123:32705:32880 [1] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-161-103:96895:97069 [7] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-161-103:96895:97069 [7] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-161-103:96895:97069 [7] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-161-103:96895:97069 [7] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-161-123:32705:32880 [1] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-161-123:32705:32880 [1] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-161-123:32705:32880 [1] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-161-123:32705:32880 [1] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-161-123:32705:32880 [1] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-161-103:96895:97069 [7] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-161-103:96895:97069 [7] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-168-34:272155:272329 [5] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-168-34:272155:272329 [5] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-168-34:272155:272329 [5] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-168-52:2722047:2722221 [2] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-168-52:2722047:2722221 [2] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-168-34:272155:272329 [5] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-168-34:272155:272329 [5] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-168-34:272155:272329 [5] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-168-34:272155:272329 [5] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-161-138:96563:96756 [0] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-161-138:96563:96756 [0] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-161-103:96890:97070 [2] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-161-103:96890:97070 [2] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-168-30:235655:235828 [7] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-168-30:235655:235828 [7] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-161-103:96894:97068 [6] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-161-103:96894:97068 [6] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-168-52:2722046:2722223 [1] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-168-52:2722046:2722223 [1] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-168-52:2722046:2722223 [1] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-168-52:2722046:2722223 [1] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-168-52:2722046:2722223 [1] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-168-52:2722046:2722223 [1] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-168-52:2722046:2722223 [1] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-168-34:272150:272327 [0] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-168-34:272150:272327 [0] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-168-34:272150:272327 [0] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-168-34:272150:272327 [0] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-168-34:272150:272327 [0] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-168-34:272150:272327 [0] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-168-34:272150:272327 [0] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-168-30:235649:235827 [1] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-168-30:235649:235827 [1] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-168-30:235649:235827 [1] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-168-30:235649:235827 [1] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-168-30:235649:235827 [1] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-168-30:235649:235827 [1] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-168-30:235649:235827 [1] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-168-34:272151:272326 [1] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-168-34:272151:272326 [1] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-168-34:272151:272326 [1] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-168-34:272151:272326 [1] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-168-34:272151:272326 [1] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-168-34:272151:272326 [1] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-168-34:272151:272326 [1] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-161-103:96892:97064 [4] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-161-103:96892:97064 [4] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-161-103:96892:97064 [4] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-161-103:96892:97064 [4] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-161-103:96892:97064 [4] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-161-103:96892:97064 [4] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-161-103:96892:97064 [4] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-161-138:96566:96755 [3] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-161-142:62888:63060 [7] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-161-142:62888:63060 [7] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-161-138:96566:96755 [3] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-168-34:272156:272325 [6] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-161-138:96566:96755 [3] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-161-138:96566:96755 [3] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-161-138:96566:96755 [3] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-168-30:235650:235824 [2] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-161-138:96566:96755 [3] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-161-138:96566:96755 [3] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-168-30:235650:235824 [2] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-168-30:235650:235824 [2] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-168-30:235650:235824 [2] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-168-30:235650:235824 [2] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-168-30:235650:235824 [2] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-168-30:235650:235824 [2] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-161-78:114513:114688 [1] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-161-78:114513:114688 [1] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-161-78:114515:114690 [3] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-161-78:114515:114690 [3] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-168-30:235651:235823 [3] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-168-30:235651:235823 [3] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-168-30:235651:235823 [3] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-168-30:235651:235823 [3] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-168-30:235651:235823 [3] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-168-30:235651:235823 [3] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-168-30:235651:235823 [3] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-161-138:96570:96759 [7] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-161-138:96570:96759 [7] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-168-52:2722052:2722226 [7] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-168-52:2722052:2722226 [7] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-161-138:96570:96759 [7] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-161-138:96570:96759 [7] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-161-138:96570:96759 [7] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-161-138:96570:96759 [7] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-161-138:96570:96759 [7] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-168-52:2722045:2722219 [0] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-168-52:2722045:2722219 [0] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-161-142:62885:63054 [4] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-161-142:62887:63055 [6] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-161-138:96565:96754 [2] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-161-138:96565:96754 [2] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-161-142:62888:63060 [7] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-161-142:62888:63060 [7] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-161-142:62888:63060 [7] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-161-142:62888:63060 [7] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-161-142:62888:63060 [7] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-161-142:62888:63060 [7] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-161-142:62888:63060 [7] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-161-78:114519:114692 [7] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-161-78:114519:114692 [7] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-161-78:114519:114692 [7] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-161-78:114519:114692 [7] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-161-78:114519:114692 [7] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-161-78:114519:114692 [7] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-161-78:114519:114692 [7] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-168-34:272152:272331 [2] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-168-34:272152:272331 [2] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-168-34:272152:272331 [2] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-168-34:272152:272331 [2] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-168-34:272152:272331 [2] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-168-34:272152:272331 [2] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-168-34:272152:272331 [2] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-161-138:96564:96757 [1] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-161-138:96564:96757 [1] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-161-78:114516:114691 [4] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-161-78:114516:114691 [4] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-161-78:114516:114691 [4] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-161-78:114516:114691 [4] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-161-78:114516:114691 [4] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-161-78:114516:114691 [4] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-161-78:114516:114691 [4] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-161-138:96563:96756 [0] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-161-138:96563:96756 [0] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-161-138:96563:96756 [0] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-161-138:96563:96756 [0] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-161-138:96563:96756 [0] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-161-138:96563:96756 [0] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-161-138:96563:96756 [0] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-161-103:96890:97070 [2] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-161-103:96890:97070 [2] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-161-103:96890:97070 [2] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-161-103:96890:97070 [2] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-161-103:96890:97070 [2] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-161-103:96890:97070 [2] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-161-103:96890:97070 [2] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-165-202:153569:153739 [7] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-165-202:153569:153739 [7] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-165-202:153569:153739 [7] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-165-202:153569:153739 [7] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-165-202:153569:153739 [7] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-165-202:153569:153739 [7] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-165-202:153569:153739 [7] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-168-52:2722048:2722220 [3] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-168-52:2722048:2722220 [3] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-168-52:2722048:2722220 [3] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-168-52:2722048:2722220 [3] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-168-52:2722048:2722220 [3] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-168-52:2722048:2722220 [3] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-168-52:2722048:2722220 [3] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-165-202:153567:153742 [5] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-165-202:153567:153742 [5] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-165-202:153567:153742 [5] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-165-202:153567:153742 [5] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-165-202:153567:153742 [5] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-165-202:153567:153742 [5] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-165-202:153567:153742 [5] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-168-30:235655:235828 [7] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-168-30:235655:235828 [7] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-168-30:235655:235828 [7] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-168-30:235655:235828 [7] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-168-30:235655:235828 [7] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-168-30:235655:235828 [7] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-168-30:235655:235828 [7] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-161-78:114518:114693 [6] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-161-78:114518:114693 [6] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-161-78:114518:114693 [6] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-161-78:114518:114693 [6] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-161-78:114518:114693 [6] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-161-78:114518:114693 [6] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-161-78:114518:114693 [6] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-160-225:124165:124341 [5] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-161-123:32706:32877 [2] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-161-123:32707:32878 [3] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-161-123:32708:32879 [4] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-161-123:32708:32879 [4] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-168-52:2722050:2722225 [5] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-168-52:2722050:2722225 [5] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-168-52:2722050:2722225 [5] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-168-52:2722050:2722225 [5] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-168-52:2722050:2722225 [5] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-168-52:2722050:2722225 [5] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-168-52:2722050:2722225 [5] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-165-202:153566:153741 [4] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-165-202:153566:153741 [4] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-165-202:153566:153741 [4] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-165-202:153566:153741 [4] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-165-202:153566:153741 [4] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-165-202:153566:153741 [4] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-165-202:153566:153741 [4] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-165-202:153563:153740 [1] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-165-202:153563:153740 [1] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-161-142:62883:63058 [2] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-165-202:153563:153740 [1] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-165-202:153563:153740 [1] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-165-202:153563:153740 [1] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-165-202:153563:153740 [1] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-165-202:153563:153740 [1] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-168-52:2722051:2722224 [6] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-168-52:2722051:2722224 [6] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-168-52:2722051:2722224 [6] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-168-52:2722051:2722224 [6] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-168-52:2722051:2722224 [6] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-168-52:2722051:2722224 [6] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-168-52:2722051:2722224 [6] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-161-123:32708:32879 [4] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-161-123:32708:32879 [4] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-161-123:32708:32879 [4] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-161-123:32708:32879 [4] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-161-123:32708:32879 [4] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-161-123:32708:32879 [4] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-161-123:32708:32879 [4] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-161-78:114513:114688 [1] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-161-78:114513:114688 [1] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-161-78:114513:114688 [1] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-161-78:114513:114688 [1] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-161-78:114513:114688 [1] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-168-52:2722047:2722221 [2] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-161-78:114513:114688 [1] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-161-78:114513:114688 [1] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-168-52:2722047:2722221 [2] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-168-52:2722047:2722221 [2] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-168-52:2722047:2722221 [2] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-168-52:2722047:2722221 [2] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-168-52:2722047:2722221 [2] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-168-52:2722047:2722221 [2] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-161-138:96565:96754 [2] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-161-138:96565:96754 [2] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-161-138:96565:96754 [2] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-161-138:96565:96754 [2] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-161-138:96565:96754 [2] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-161-138:96565:96754 [2] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-161-138:96565:96754 [2] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-161-138:96564:96757 [1] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-161-138:96564:96757 [1] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-161-138:96564:96757 [1] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-161-138:96564:96757 [1] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-161-138:96564:96757 [1] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-161-138:96564:96757 [1] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-161-138:96564:96757 [1] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-161-103:96894:97068 [6] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-161-103:96894:97068 [6] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-161-103:96894:97068 [6] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-161-103:96894:97068 [6] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-161-103:96894:97068 [6] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-161-103:96894:97068 [6] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-161-103:96894:97068 [6] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-168-30:235648:235826 [0] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-168-30:235653:235825 [5] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-168-30:235654:235821 [6] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-161-78:114515:114690 [3] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-161-78:114515:114690 [3] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-161-78:114515:114690 [3] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-161-78:114515:114690 [3] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-161-78:114515:114690 [3] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-161-78:114515:114690 [3] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-161-78:114515:114690 [3] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-161-78:114517:114687 [5] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-168-34:272157:272324 [7] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-161-78:114514:114689 [2] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-161-138:96568:96753 [5] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-161-78:114512:114686 [0] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-168-52:2722045:2722219 [0] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-168-52:2722045:2722219 [0] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-168-52:2722045:2722219 [0] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-168-52:2722045:2722219 [0] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-168-52:2722045:2722219 [0] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-168-52:2722045:2722219 [0] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-168-52:2722045:2722219 [0] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-161-142:62882:63056 [1] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-161-138:96569:96758 [6] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-161-142:62886:63057 [5] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-168-34:272154:272330 [4] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-160-225:124166:124339 [6] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-161-103:96888:97066 [0] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-161-103:96893:97067 [5] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-165-202:153568:153736 [6] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-161-138:96567:96760 [4] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-161-123:32709:32882 [5] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-160-225:124160:124335 [0] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-168-52:2722052:2722226 [7] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-161-123:32711:32884 [7] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-168-52:2722052:2722226 [7] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-168-52:2722052:2722226 [7] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-168-52:2722052:2722226 [7] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-168-52:2722052:2722226 [7] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-168-52:2722052:2722226 [7] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-168-52:2722052:2722226 [7] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-161-123:32704:32881 [0] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-161-142:62881:63059 [0] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-161-142:62884:63061 [3] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-161-142:62888:63060 [7] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-165-202:153565:153738 [3] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-161-123:32710:32883 [6] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-168-30:235652:235822 [4] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-161-103:96891:97063 [3] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-161-103:96891:97063 [3] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-161-103:96895:97069 [7] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-161-123:32705:32880 [1] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-160-225:124164:124336 [4] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-168-34:272153:272328 [3] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-161-103:96891:97063 [3] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-161-103:96891:97063 [3] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-161-103:96891:97063 [3] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-161-103:96891:97063 [3] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-161-103:96891:97063 [3] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-161-103:96891:97063 [3] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-161-103:96891:97063 [3] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-168-34:272155:272329 [5] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-160-225:124161:124337 [1] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-165-202:153564:153737 [2] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-160-225:124167:124338 [7] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-160-225:124163:124340 [3] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-168-34:272151:272326 [1] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-161-138:96566:96755 [3] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-168-34:272150:272327 [0] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-168-30:235649:235827 [1] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-168-34:272152:272331 [2] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-161-78:114519:114692 [7] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-165-202:153562:153735 [0] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-161-103:96889:97065 [1] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-161-138:96570:96759 [7] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-168-30:235650:235824 [2] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-168-30:235651:235823 [3] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-161-138:96563:96756 [0] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-168-30:235655:235828 [7] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-161-123:32708:32879 [4] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-161-103:96892:97064 [4] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-161-78:114516:114691 [4] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-161-103:96890:97070 [2] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-161-103:96894:97068 [6] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-165-202:153569:153739 [7] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-161-78:114518:114693 [6] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-161-78:114513:114688 [1] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-165-202:153567:153742 [5] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-161-78:114515:114690 [3] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-165-202:153566:153741 [4] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-165-202:153563:153740 [1] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-161-138:96565:96754 [2] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-161-138:96564:96757 [1] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-168-52:2722046:2722223 [1] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-168-52:2722048:2722220 [3] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-168-52:2722050:2722225 [5] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-161-103:96891:97063 [3] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-168-52:2722051:2722224 [6] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-168-52:2722045:2722219 [0] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-168-52:2722047:2722221 [2] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-168-52:2722052:2722226 [7] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/408
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO Using network Libfabric
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/408
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-165-202:153563:153740 [1] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-165-202:153563:153740 [1] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-165-202:153563:153740 [1] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-165-202:153563:153740 [1] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-165-202:153563:153740 [1] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-165-202:153563:153740 [1] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-165-202:153563:153740 [1] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-165-202:153563:153740 [1] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-165-202:153563:153740 [1] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-165-202:153563:153740 [1] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-165-202:153563:153740 [1] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-165-202:153563:153740 [1] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-165-202:153563:153740 [1] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-165-202:153563:153740 [1] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-165-202:153563:153740 [1] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-165-202:153563:153740 [1] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-165-202:153563:153740 [1] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-165-202:153563:153740 [1] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-165-202:153563:153740 [1] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-165-202:153563:153740 [1] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-165-202:153563:153740 [1] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-165-202:153563:153740 [1] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-165-202:153563:153740 [1] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-165-202:153563:153740 [1] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-165-202:153563:153740 [1] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-165-202:153563:153740 [1] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-165-202:153563:153740 [1] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-165-202:153563:153740 [1] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-165-202:153563:153740 [1] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-165-202:153563:153740 [1] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-165-202:153563:153740 [1] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-165-202:153563:153740 [1] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-165-202:153563:153740 [1] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/408
+ip-26-0-165-202:153563:153740 [1] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/409
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/408
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/408
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/408
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/408
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-160-225:124160:124335 [0] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-160-225:124160:124335 [0] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-160-225:124160:124335 [0] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-160-225:124160:124335 [0] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-160-225:124160:124335 [0] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-160-225:124160:124335 [0] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-160-225:124160:124335 [0] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-160-225:124160:124335 [0] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-160-225:124160:124335 [0] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-160-225:124160:124335 [0] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-160-225:124160:124335 [0] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-160-225:124160:124335 [0] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-160-225:124160:124335 [0] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-160-225:124160:124335 [0] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-160-225:124160:124335 [0] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-160-225:124160:124335 [0] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-160-225:124160:124335 [0] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-160-225:124160:124335 [0] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-160-225:124160:124335 [0] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-160-225:124160:124335 [0] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-160-225:124160:124335 [0] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-160-225:124160:124335 [0] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-160-225:124160:124335 [0] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-160-225:124160:124335 [0] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-160-225:124160:124335 [0] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-160-225:124160:124335 [0] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-160-225:124160:124335 [0] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-160-225:124160:124335 [0] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-160-225:124160:124335 [0] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-160-225:124160:124335 [0] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-160-225:124160:124335 [0] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-160-225:124160:124335 [0] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-160-225:124160:124335 [0] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/541
+ip-26-0-160-225:124160:124335 [0] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-165-202:153563:153740 [1] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-165-202:153563:153740 [1] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-165-202:153563:153740 [1] NCCL INFO Using network Libfabric
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO Using network Libfabric
+ip-26-0-161-123:32706:32877 [2] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-161-123:32706:32877 [2] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-161-123:32706:32877 [2] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-161-123:32706:32877 [2] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-161-123:32706:32877 [2] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-161-123:32706:32877 [2] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-161-123:32706:32877 [2] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-161-123:32706:32877 [2] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-161-123:32706:32877 [2] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-161-123:32706:32877 [2] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-161-123:32706:32877 [2] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-161-123:32706:32877 [2] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-161-123:32706:32877 [2] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-161-123:32706:32877 [2] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-161-123:32706:32877 [2] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-161-123:32706:32877 [2] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-161-123:32706:32877 [2] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-161-123:32706:32877 [2] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-161-123:32706:32877 [2] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-161-123:32706:32877 [2] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-161-123:32706:32877 [2] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-161-123:32706:32877 [2] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-161-123:32706:32877 [2] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-161-123:32706:32877 [2] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-161-123:32706:32877 [2] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-161-123:32706:32877 [2] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-161-123:32706:32877 [2] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-161-123:32706:32877 [2] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-161-123:32706:32877 [2] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-161-123:32706:32877 [2] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-161-123:32706:32877 [2] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-161-123:32706:32877 [2] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-161-123:32706:32877 [2] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/408
+ip-26-0-161-123:32706:32877 [2] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/408
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/408
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-168-34:272152:272331 [2] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-168-34:272152:272331 [2] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-168-34:272152:272331 [2] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-168-34:272152:272331 [2] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-168-34:272152:272331 [2] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-168-34:272152:272331 [2] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-168-34:272152:272331 [2] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-168-34:272152:272331 [2] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-168-34:272152:272331 [2] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-168-34:272152:272331 [2] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-168-34:272152:272331 [2] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-168-34:272152:272331 [2] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-168-34:272152:272331 [2] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-168-34:272152:272331 [2] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-168-34:272152:272331 [2] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-168-34:272152:272331 [2] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-168-34:272152:272331 [2] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-168-34:272152:272331 [2] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-168-34:272152:272331 [2] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-168-34:272152:272331 [2] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-168-34:272152:272331 [2] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-168-34:272152:272331 [2] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-168-34:272152:272331 [2] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-168-34:272152:272331 [2] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-168-34:272152:272331 [2] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-168-34:272152:272331 [2] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-168-34:272152:272331 [2] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-168-34:272152:272331 [2] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-168-34:272152:272331 [2] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-168-34:272152:272331 [2] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-168-34:272152:272331 [2] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-168-34:272152:272331 [2] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-168-34:272152:272331 [2] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/408
+ip-26-0-168-34:272152:272331 [2] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/408
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-161-78:114517:114687 [5] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-161-78:114517:114687 [5] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-161-78:114517:114687 [5] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-161-78:114517:114687 [5] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-161-78:114517:114687 [5] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-161-78:114517:114687 [5] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-161-78:114517:114687 [5] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-161-78:114517:114687 [5] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-161-78:114517:114687 [5] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-161-78:114517:114687 [5] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-161-78:114517:114687 [5] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-161-78:114517:114687 [5] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-161-78:114517:114687 [5] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-161-78:114517:114687 [5] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-161-78:114517:114687 [5] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-161-78:114517:114687 [5] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-161-78:114517:114687 [5] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-161-78:114517:114687 [5] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-161-78:114517:114687 [5] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-161-78:114517:114687 [5] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-161-78:114517:114687 [5] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-161-78:114517:114687 [5] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-161-78:114517:114687 [5] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-161-78:114517:114687 [5] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-161-78:114517:114687 [5] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-161-78:114517:114687 [5] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-161-78:114517:114687 [5] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-161-78:114517:114687 [5] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-161-78:114517:114687 [5] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-161-78:114517:114687 [5] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-161-78:114517:114687 [5] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-161-78:114517:114687 [5] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-161-78:114517:114687 [5] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/408
+ip-26-0-161-78:114517:114687 [5] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-161-78:114514:114689 [2] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-161-78:114514:114689 [2] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-161-78:114514:114689 [2] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-161-78:114514:114689 [2] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-161-78:114514:114689 [2] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-161-78:114514:114689 [2] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-161-78:114514:114689 [2] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-161-78:114514:114689 [2] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-161-78:114514:114689 [2] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-161-78:114514:114689 [2] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-161-78:114514:114689 [2] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-161-78:114514:114689 [2] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-161-78:114514:114689 [2] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-161-78:114514:114689 [2] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-161-78:114514:114689 [2] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-161-78:114514:114689 [2] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-161-78:114514:114689 [2] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-161-78:114514:114689 [2] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-161-78:114514:114689 [2] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-161-78:114514:114689 [2] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-161-78:114514:114689 [2] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-161-78:114514:114689 [2] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-161-78:114514:114689 [2] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-161-78:114514:114689 [2] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-161-78:114514:114689 [2] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-161-78:114514:114689 [2] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-161-78:114514:114689 [2] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-161-78:114514:114689 [2] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-161-78:114514:114689 [2] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-161-78:114514:114689 [2] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-161-78:114514:114689 [2] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-161-78:114514:114689 [2] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-160-225:124161:124337 [1] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-160-225:124161:124337 [1] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-160-225:124161:124337 [1] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-160-225:124161:124337 [1] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-160-225:124161:124337 [1] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-160-225:124161:124337 [1] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-160-225:124161:124337 [1] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-160-225:124161:124337 [1] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-160-225:124161:124337 [1] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-160-225:124161:124337 [1] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-160-225:124161:124337 [1] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-160-225:124161:124337 [1] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-160-225:124161:124337 [1] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-160-225:124161:124337 [1] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-160-225:124161:124337 [1] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-160-225:124161:124337 [1] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-160-225:124161:124337 [1] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-160-225:124161:124337 [1] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-160-225:124161:124337 [1] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-160-225:124161:124337 [1] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-160-225:124161:124337 [1] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-160-225:124161:124337 [1] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-160-225:124161:124337 [1] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-160-225:124161:124337 [1] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-160-225:124161:124337 [1] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-160-225:124161:124337 [1] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-160-225:124161:124337 [1] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-160-225:124161:124337 [1] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-160-225:124161:124337 [1] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-160-225:124161:124337 [1] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-160-225:124161:124337 [1] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-160-225:124161:124337 [1] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-160-225:124161:124337 [1] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/408
+ip-26-0-160-225:124161:124337 [1] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-161-78:114514:114689 [2] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/408
+ip-26-0-161-78:114514:114689 [2] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/408
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-161-142:62885:63054 [4] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-161-142:62885:63054 [4] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-161-142:62885:63054 [4] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-161-142:62885:63054 [4] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-161-142:62885:63054 [4] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-161-142:62885:63054 [4] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-161-142:62885:63054 [4] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-161-142:62885:63054 [4] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-161-142:62885:63054 [4] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-161-142:62885:63054 [4] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-161-142:62885:63054 [4] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-161-142:62885:63054 [4] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-161-142:62885:63054 [4] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-161-142:62885:63054 [4] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-161-142:62885:63054 [4] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-161-142:62885:63054 [4] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-161-142:62885:63054 [4] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-161-142:62885:63054 [4] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-161-142:62885:63054 [4] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-161-142:62885:63054 [4] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-161-142:62885:63054 [4] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-161-142:62885:63054 [4] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-161-142:62885:63054 [4] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-161-142:62885:63054 [4] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-161-142:62885:63054 [4] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-161-142:62885:63054 [4] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-161-142:62885:63054 [4] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-161-142:62885:63054 [4] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-161-142:62885:63054 [4] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-161-142:62885:63054 [4] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-161-142:62885:63054 [4] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-161-142:62885:63054 [4] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-161-142:62885:63054 [4] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/408
+ip-26-0-161-142:62885:63054 [4] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-161-142:62887:63055 [6] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-161-142:62887:63055 [6] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-161-142:62887:63055 [6] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-161-142:62887:63055 [6] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-161-142:62887:63055 [6] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-161-142:62887:63055 [6] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-161-142:62887:63055 [6] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-161-142:62887:63055 [6] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-161-142:62887:63055 [6] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-161-142:62887:63055 [6] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-161-142:62887:63055 [6] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-161-142:62887:63055 [6] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-161-142:62887:63055 [6] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-161-142:62887:63055 [6] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-161-142:62887:63055 [6] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-161-142:62887:63055 [6] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-161-142:62887:63055 [6] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-161-142:62887:63055 [6] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-161-142:62887:63055 [6] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-161-142:62887:63055 [6] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-161-142:62887:63055 [6] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-161-142:62887:63055 [6] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-161-142:62887:63055 [6] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-161-142:62887:63055 [6] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-161-142:62887:63055 [6] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-161-142:62887:63055 [6] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-161-142:62887:63055 [6] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-161-142:62887:63055 [6] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-161-142:62887:63055 [6] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-161-142:62887:63055 [6] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-161-142:62887:63055 [6] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-161-142:62887:63055 [6] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-161-142:62887:63055 [6] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/408
+ip-26-0-161-142:62887:63055 [6] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-168-34:272154:272330 [4] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-168-34:272154:272330 [4] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-168-34:272154:272330 [4] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-168-34:272154:272330 [4] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-168-34:272154:272330 [4] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-168-34:272154:272330 [4] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-168-34:272154:272330 [4] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-168-34:272154:272330 [4] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-168-34:272154:272330 [4] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-168-34:272154:272330 [4] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-168-34:272154:272330 [4] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-168-34:272154:272330 [4] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-168-34:272154:272330 [4] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-168-34:272154:272330 [4] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-168-34:272154:272330 [4] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-168-34:272154:272330 [4] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-168-34:272154:272330 [4] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-168-34:272154:272330 [4] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-168-34:272154:272330 [4] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-168-34:272154:272330 [4] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-168-34:272154:272330 [4] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-168-34:272154:272330 [4] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-168-34:272154:272330 [4] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-168-34:272154:272330 [4] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-168-34:272154:272330 [4] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-168-34:272154:272330 [4] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-168-34:272154:272330 [4] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-168-34:272154:272330 [4] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-168-34:272154:272330 [4] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-168-34:272154:272330 [4] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-168-34:272154:272330 [4] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-168-34:272154:272330 [4] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-161-78:114516:114691 [4] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-161-78:114516:114691 [4] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-161-78:114516:114691 [4] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-161-78:114516:114691 [4] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-161-78:114516:114691 [4] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-161-78:114516:114691 [4] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-161-78:114516:114691 [4] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-161-78:114516:114691 [4] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-161-78:114516:114691 [4] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-161-78:114516:114691 [4] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-161-78:114516:114691 [4] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-161-78:114516:114691 [4] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-161-78:114516:114691 [4] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-161-78:114516:114691 [4] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-161-78:114516:114691 [4] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-161-78:114516:114691 [4] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-161-78:114516:114691 [4] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-161-78:114516:114691 [4] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-161-78:114516:114691 [4] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-161-78:114516:114691 [4] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-161-78:114516:114691 [4] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-161-78:114516:114691 [4] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-161-78:114516:114691 [4] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-161-78:114516:114691 [4] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-161-78:114516:114691 [4] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-161-78:114516:114691 [4] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-161-78:114516:114691 [4] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-161-78:114516:114691 [4] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-161-78:114516:114691 [4] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-161-78:114516:114691 [4] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-161-78:114516:114691 [4] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-161-78:114516:114691 [4] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-168-34:272154:272330 [4] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/408
+ip-26-0-168-34:272154:272330 [4] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-161-78:114516:114691 [4] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/408
+ip-26-0-161-78:114516:114691 [4] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-161-123:32705:32880 [1] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-161-123:32705:32880 [1] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-161-123:32705:32880 [1] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-161-123:32705:32880 [1] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-161-123:32705:32880 [1] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-161-123:32705:32880 [1] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-161-123:32705:32880 [1] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-161-123:32705:32880 [1] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-161-123:32705:32880 [1] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-161-123:32705:32880 [1] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-161-123:32705:32880 [1] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-161-123:32705:32880 [1] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-161-123:32705:32880 [1] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-161-123:32705:32880 [1] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-161-123:32705:32880 [1] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-161-123:32705:32880 [1] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-161-123:32705:32880 [1] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-161-123:32705:32880 [1] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-161-123:32705:32880 [1] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-161-123:32705:32880 [1] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-161-123:32705:32880 [1] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-161-123:32705:32880 [1] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-161-123:32705:32880 [1] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-161-123:32705:32880 [1] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-161-123:32705:32880 [1] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-161-123:32705:32880 [1] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-161-123:32705:32880 [1] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-161-123:32705:32880 [1] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-161-123:32705:32880 [1] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-161-123:32705:32880 [1] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-161-123:32705:32880 [1] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-161-123:32705:32880 [1] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-161-78:114518:114693 [6] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-161-78:114518:114693 [6] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-161-78:114518:114693 [6] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-161-78:114518:114693 [6] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-161-78:114515:114690 [3] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-161-78:114518:114693 [6] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-161-78:114518:114693 [6] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-161-78:114515:114690 [3] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-161-78:114518:114693 [6] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-161-78:114515:114690 [3] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-161-78:114518:114693 [6] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-161-78:114515:114690 [3] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-161-78:114518:114693 [6] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-161-78:114515:114690 [3] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-161-78:114518:114693 [6] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-161-78:114515:114690 [3] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-161-78:114518:114693 [6] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-161-78:114515:114690 [3] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-161-78:114518:114693 [6] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-161-78:114515:114690 [3] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-161-78:114518:114693 [6] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-161-78:114515:114690 [3] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-161-78:114518:114693 [6] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-161-78:114515:114690 [3] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-161-78:114518:114693 [6] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-161-78:114515:114690 [3] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-161-78:114518:114693 [6] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-161-78:114515:114690 [3] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-161-78:114518:114693 [6] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-161-78:114515:114690 [3] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-161-78:114518:114693 [6] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-161-78:114515:114690 [3] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-161-78:114518:114693 [6] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-161-78:114515:114690 [3] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-161-78:114518:114693 [6] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-161-78:114515:114690 [3] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-161-78:114518:114693 [6] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-161-78:114515:114690 [3] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-161-78:114518:114693 [6] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-161-78:114515:114690 [3] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-161-78:114518:114693 [6] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-161-78:114515:114690 [3] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-161-78:114518:114693 [6] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-161-78:114515:114690 [3] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-161-78:114518:114693 [6] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-161-78:114515:114690 [3] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-161-78:114518:114693 [6] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-161-78:114515:114690 [3] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-161-78:114518:114693 [6] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-161-78:114515:114690 [3] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-161-78:114518:114693 [6] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-161-78:114515:114690 [3] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-161-78:114518:114693 [6] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-161-78:114515:114690 [3] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-161-78:114518:114693 [6] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-161-78:114515:114690 [3] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-161-78:114518:114693 [6] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-161-78:114515:114690 [3] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-161-78:114518:114693 [6] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-161-78:114515:114690 [3] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-161-78:114515:114690 [3] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-161-78:114515:114690 [3] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-161-78:114515:114690 [3] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-161-78:114515:114690 [3] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-168-34:272153:272328 [3] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-168-34:272153:272328 [3] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-168-34:272153:272328 [3] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-168-34:272153:272328 [3] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-168-34:272153:272328 [3] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-168-34:272153:272328 [3] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-168-34:272153:272328 [3] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-168-34:272153:272328 [3] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-168-34:272153:272328 [3] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-168-34:272153:272328 [3] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-168-34:272153:272328 [3] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-168-34:272153:272328 [3] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-168-34:272153:272328 [3] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-168-34:272153:272328 [3] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-168-34:272153:272328 [3] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-168-34:272153:272328 [3] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-168-34:272153:272328 [3] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-168-34:272153:272328 [3] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-168-34:272153:272328 [3] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-168-34:272153:272328 [3] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-168-34:272153:272328 [3] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-168-34:272153:272328 [3] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-168-34:272153:272328 [3] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-168-34:272153:272328 [3] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-168-34:272153:272328 [3] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-168-34:272153:272328 [3] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-168-34:272153:272328 [3] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-168-34:272153:272328 [3] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-168-34:272153:272328 [3] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-168-34:272153:272328 [3] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-168-34:272153:272328 [3] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-168-34:272153:272328 [3] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-161-123:32705:32880 [1] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/408
+ip-26-0-161-123:32705:32880 [1] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/408
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-168-34:272153:272328 [3] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/408
+ip-26-0-168-34:272153:272328 [3] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-160-225:124160:124335 [0] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-160-225:124160:124335 [0] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-160-225:124160:124335 [0] NCCL INFO Using network Libfabric
+ip-26-0-165-202:153567:153742 [5] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-165-202:153567:153742 [5] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-165-202:153567:153742 [5] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-165-202:153567:153742 [5] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-165-202:153567:153742 [5] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-165-202:153567:153742 [5] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-165-202:153567:153742 [5] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-165-202:153567:153742 [5] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-165-202:153567:153742 [5] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-165-202:153567:153742 [5] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-165-202:153567:153742 [5] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-165-202:153567:153742 [5] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-165-202:153567:153742 [5] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-165-202:153567:153742 [5] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-165-202:153567:153742 [5] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-165-202:153567:153742 [5] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-165-202:153567:153742 [5] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-165-202:153567:153742 [5] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-165-202:153567:153742 [5] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-165-202:153567:153742 [5] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-165-202:153567:153742 [5] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-165-202:153567:153742 [5] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-165-202:153567:153742 [5] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-165-202:153567:153742 [5] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-165-202:153567:153742 [5] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-165-202:153567:153742 [5] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-165-202:153567:153742 [5] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-165-202:153567:153742 [5] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-165-202:153567:153742 [5] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-165-202:153567:153742 [5] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-165-202:153567:153742 [5] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-165-202:153567:153742 [5] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-165-202:153567:153742 [5] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/408
+ip-26-0-165-202:153567:153742 [5] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/408
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/408
+ip-26-0-168-34:272157:272324 [7] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-168-34:272157:272324 [7] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-168-34:272157:272324 [7] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-168-34:272157:272324 [7] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-168-34:272157:272324 [7] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-168-34:272157:272324 [7] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-168-34:272157:272324 [7] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-168-34:272157:272324 [7] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-168-34:272157:272324 [7] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-168-34:272157:272324 [7] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-168-34:272157:272324 [7] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-168-34:272157:272324 [7] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-168-34:272157:272324 [7] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-168-34:272157:272324 [7] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-168-34:272157:272324 [7] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-168-34:272157:272324 [7] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-168-34:272157:272324 [7] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-168-34:272157:272324 [7] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-168-34:272157:272324 [7] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-168-34:272157:272324 [7] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-168-34:272157:272324 [7] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-168-34:272157:272324 [7] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-168-34:272157:272324 [7] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-168-34:272157:272324 [7] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-168-34:272157:272324 [7] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-168-34:272157:272324 [7] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-168-34:272157:272324 [7] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-168-34:272157:272324 [7] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-168-34:272157:272324 [7] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-168-34:272157:272324 [7] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-168-34:272157:272324 [7] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-168-34:272157:272324 [7] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/408
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-168-34:272157:272324 [7] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/408
+ip-26-0-168-34:272157:272324 [7] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-168-52:2722046:2722223 [1] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-168-52:2722046:2722223 [1] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-168-52:2722046:2722223 [1] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-168-52:2722046:2722223 [1] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-168-52:2722046:2722223 [1] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-168-52:2722046:2722223 [1] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-168-52:2722046:2722223 [1] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-168-52:2722046:2722223 [1] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-168-52:2722046:2722223 [1] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-168-52:2722046:2722223 [1] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-168-52:2722046:2722223 [1] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-168-52:2722046:2722223 [1] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-168-52:2722046:2722223 [1] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-168-52:2722046:2722223 [1] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-168-52:2722046:2722223 [1] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-168-52:2722046:2722223 [1] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-168-52:2722046:2722223 [1] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-168-52:2722046:2722223 [1] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-168-52:2722046:2722223 [1] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-168-52:2722046:2722223 [1] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-168-52:2722046:2722223 [1] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-168-52:2722046:2722223 [1] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-168-52:2722046:2722223 [1] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-168-52:2722046:2722223 [1] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-168-52:2722046:2722223 [1] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-168-52:2722046:2722223 [1] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-168-52:2722046:2722223 [1] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-168-52:2722046:2722223 [1] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-168-52:2722046:2722223 [1] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-168-52:2722046:2722223 [1] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-168-52:2722046:2722223 [1] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-168-52:2722046:2722223 [1] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-165-202:153564:153737 [2] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-165-202:153564:153737 [2] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-165-202:153564:153737 [2] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-165-202:153564:153737 [2] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-165-202:153564:153737 [2] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-165-202:153564:153737 [2] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-165-202:153564:153737 [2] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-165-202:153564:153737 [2] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-165-202:153564:153737 [2] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-165-202:153564:153737 [2] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-165-202:153564:153737 [2] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-165-202:153564:153737 [2] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-165-202:153564:153737 [2] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-165-202:153564:153737 [2] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-165-202:153564:153737 [2] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-165-202:153564:153737 [2] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-165-202:153564:153737 [2] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-165-202:153564:153737 [2] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-165-202:153564:153737 [2] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-165-202:153564:153737 [2] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-165-202:153564:153737 [2] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-165-202:153564:153737 [2] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-165-202:153564:153737 [2] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-165-202:153564:153737 [2] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-165-202:153564:153737 [2] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-165-202:153564:153737 [2] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-165-202:153564:153737 [2] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-165-202:153564:153737 [2] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-165-202:153564:153737 [2] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-165-202:153564:153737 [2] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-165-202:153564:153737 [2] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-165-202:153564:153737 [2] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-168-52:2722046:2722223 [1] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/408
+ip-26-0-168-52:2722046:2722223 [1] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-168-34:272155:272329 [5] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-168-34:272155:272329 [5] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-168-34:272155:272329 [5] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-168-34:272155:272329 [5] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-168-34:272155:272329 [5] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-168-34:272155:272329 [5] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-161-78:114515:114690 [3] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/408
+ip-26-0-161-78:114515:114690 [3] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-161-78:114518:114693 [6] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/408
+ip-26-0-161-78:114518:114693 [6] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-168-34:272155:272329 [5] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-168-34:272155:272329 [5] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-168-34:272155:272329 [5] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-168-34:272155:272329 [5] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-168-34:272155:272329 [5] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-161-78:114513:114688 [1] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-161-78:114513:114688 [1] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-161-78:114513:114688 [1] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-161-78:114513:114688 [1] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-161-78:114513:114688 [1] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-168-34:272155:272329 [5] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-168-34:272155:272329 [5] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-168-34:272155:272329 [5] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-168-34:272155:272329 [5] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-168-34:272155:272329 [5] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-168-34:272155:272329 [5] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-168-34:272155:272329 [5] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-168-34:272155:272329 [5] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-168-34:272155:272329 [5] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-168-34:272155:272329 [5] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-168-34:272155:272329 [5] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-161-78:114513:114688 [1] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-161-78:114513:114688 [1] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-161-78:114513:114688 [1] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-161-78:114513:114688 [1] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-168-34:272155:272329 [5] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-168-34:272155:272329 [5] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-161-78:114513:114688 [1] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-161-78:114513:114688 [1] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-161-78:114513:114688 [1] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-161-78:114513:114688 [1] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-168-34:272155:272329 [5] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-168-34:272155:272329 [5] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-168-34:272155:272329 [5] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-168-34:272155:272329 [5] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-161-78:114513:114688 [1] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-161-78:114513:114688 [1] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-161-78:114513:114688 [1] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-161-78:114513:114688 [1] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-161-78:114513:114688 [1] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-161-78:114513:114688 [1] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-161-78:114513:114688 [1] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-168-34:272155:272329 [5] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-161-78:114513:114688 [1] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-161-78:114513:114688 [1] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-161-78:114513:114688 [1] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-161-78:114513:114688 [1] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-168-34:272155:272329 [5] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-168-34:272155:272329 [5] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-168-34:272155:272329 [5] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-161-78:114513:114688 [1] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-161-78:114513:114688 [1] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-161-78:114513:114688 [1] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-161-78:114513:114688 [1] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-161-78:114513:114688 [1] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-161-78:114513:114688 [1] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-161-78:114513:114688 [1] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-161-78:114513:114688 [1] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-161-78:114513:114688 [1] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/408
+ip-26-0-161-78:114513:114688 [1] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-161-78:114519:114692 [7] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-161-78:114519:114692 [7] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-161-78:114519:114692 [7] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-161-78:114519:114692 [7] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-161-78:114519:114692 [7] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-161-78:114519:114692 [7] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-161-78:114519:114692 [7] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-161-78:114519:114692 [7] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-161-78:114519:114692 [7] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-161-78:114519:114692 [7] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-161-78:114519:114692 [7] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-161-78:114519:114692 [7] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-161-78:114519:114692 [7] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-161-78:114519:114692 [7] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-161-78:114519:114692 [7] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-161-78:114519:114692 [7] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-161-78:114519:114692 [7] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-161-78:114519:114692 [7] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-161-78:114519:114692 [7] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-161-78:114519:114692 [7] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-161-78:114519:114692 [7] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-161-78:114519:114692 [7] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-161-78:114519:114692 [7] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-161-78:114519:114692 [7] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-161-78:114519:114692 [7] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-161-78:114519:114692 [7] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-161-78:114519:114692 [7] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-161-78:114519:114692 [7] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-161-78:114519:114692 [7] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-161-78:114519:114692 [7] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-161-78:114519:114692 [7] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-161-78:114519:114692 [7] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-161-78:114519:114692 [7] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/408
+ip-26-0-161-78:114519:114692 [7] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-161-78:114512:114686 [0] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-161-78:114512:114686 [0] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-161-78:114512:114686 [0] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-161-78:114512:114686 [0] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-161-78:114512:114686 [0] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-161-78:114512:114686 [0] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-161-78:114512:114686 [0] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-161-78:114512:114686 [0] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-161-78:114512:114686 [0] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-161-78:114512:114686 [0] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-161-78:114512:114686 [0] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-161-78:114512:114686 [0] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-161-78:114512:114686 [0] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-161-78:114512:114686 [0] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-161-78:114512:114686 [0] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-161-78:114512:114686 [0] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-161-78:114512:114686 [0] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-161-78:114512:114686 [0] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-161-78:114512:114686 [0] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-161-78:114512:114686 [0] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-161-78:114512:114686 [0] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-161-78:114512:114686 [0] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-161-78:114512:114686 [0] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-161-78:114512:114686 [0] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-161-78:114512:114686 [0] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-161-78:114512:114686 [0] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-161-78:114512:114686 [0] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-161-78:114512:114686 [0] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-161-78:114512:114686 [0] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-161-78:114512:114686 [0] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-161-78:114512:114686 [0] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-161-78:114512:114686 [0] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-168-34:272156:272325 [6] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-168-34:272156:272325 [6] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-168-34:272156:272325 [6] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-168-34:272156:272325 [6] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-168-34:272156:272325 [6] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-168-34:272156:272325 [6] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-168-34:272156:272325 [6] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-168-34:272156:272325 [6] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-168-34:272156:272325 [6] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-168-34:272156:272325 [6] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-168-34:272156:272325 [6] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-168-34:272156:272325 [6] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-168-34:272156:272325 [6] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-168-34:272156:272325 [6] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-168-34:272156:272325 [6] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-168-34:272156:272325 [6] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-168-34:272156:272325 [6] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-168-34:272156:272325 [6] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-168-34:272156:272325 [6] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-168-34:272156:272325 [6] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-168-34:272156:272325 [6] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-168-34:272156:272325 [6] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-168-34:272156:272325 [6] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-168-34:272156:272325 [6] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-168-34:272156:272325 [6] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-168-34:272156:272325 [6] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-168-34:272156:272325 [6] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-168-34:272156:272325 [6] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-168-34:272156:272325 [6] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-168-34:272156:272325 [6] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-168-34:272156:272325 [6] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-168-34:272156:272325 [6] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-168-34:272150:272327 [0] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-168-34:272150:272327 [0] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-168-34:272150:272327 [0] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-168-34:272150:272327 [0] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-168-34:272150:272327 [0] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-168-34:272150:272327 [0] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-168-34:272150:272327 [0] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-168-34:272150:272327 [0] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-168-34:272150:272327 [0] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-168-34:272150:272327 [0] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-168-34:272150:272327 [0] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-168-34:272150:272327 [0] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-168-34:272150:272327 [0] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-168-34:272150:272327 [0] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-168-34:272150:272327 [0] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-168-34:272150:272327 [0] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-168-34:272150:272327 [0] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-168-34:272150:272327 [0] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-168-34:272150:272327 [0] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-168-34:272150:272327 [0] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-168-34:272150:272327 [0] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-168-34:272150:272327 [0] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-168-34:272150:272327 [0] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-168-34:272150:272327 [0] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-168-34:272150:272327 [0] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-168-34:272150:272327 [0] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-168-34:272150:272327 [0] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-168-34:272150:272327 [0] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-168-34:272150:272327 [0] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-168-34:272150:272327 [0] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-168-34:272150:272327 [0] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-168-34:272150:272327 [0] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/408
+ip-26-0-161-78:114512:114686 [0] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/408
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-161-78:114512:114686 [0] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-165-202:153564:153737 [2] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/408
+ip-26-0-165-202:153564:153737 [2] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-160-225:124163:124340 [3] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-160-225:124163:124340 [3] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-160-225:124163:124340 [3] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-160-225:124163:124340 [3] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-160-225:124163:124340 [3] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-160-225:124163:124340 [3] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-160-225:124163:124340 [3] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-160-225:124163:124340 [3] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-160-225:124163:124340 [3] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-160-225:124163:124340 [3] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-160-225:124163:124340 [3] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-160-225:124163:124340 [3] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-160-225:124163:124340 [3] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-160-225:124163:124340 [3] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-160-225:124163:124340 [3] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-160-225:124163:124340 [3] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-160-225:124163:124340 [3] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-160-225:124163:124340 [3] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-160-225:124163:124340 [3] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-160-225:124163:124340 [3] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-160-225:124163:124340 [3] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-160-225:124163:124340 [3] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-160-225:124163:124340 [3] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-160-225:124163:124340 [3] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-160-225:124163:124340 [3] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-160-225:124163:124340 [3] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-160-225:124163:124340 [3] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-160-225:124163:124340 [3] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-160-225:124163:124340 [3] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-160-225:124163:124340 [3] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-160-225:124163:124340 [3] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-160-225:124163:124340 [3] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-160-225:124165:124341 [5] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-160-225:124165:124341 [5] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-160-225:124165:124341 [5] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-160-225:124165:124341 [5] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-160-225:124165:124341 [5] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-160-225:124165:124341 [5] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-160-225:124165:124341 [5] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-160-225:124165:124341 [5] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-160-225:124165:124341 [5] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-160-225:124165:124341 [5] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-168-34:272155:272329 [5] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/408
+ip-26-0-160-225:124165:124341 [5] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-160-225:124165:124341 [5] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-160-225:124165:124341 [5] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-160-225:124165:124341 [5] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-160-225:124165:124341 [5] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-160-225:124165:124341 [5] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-160-225:124165:124341 [5] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-160-225:124165:124341 [5] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-160-225:124165:124341 [5] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-160-225:124165:124341 [5] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-160-225:124165:124341 [5] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-160-225:124165:124341 [5] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-160-225:124165:124341 [5] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-160-225:124165:124341 [5] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-168-34:272155:272329 [5] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-160-225:124165:124341 [5] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-160-225:124165:124341 [5] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-160-225:124165:124341 [5] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-160-225:124165:124341 [5] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-160-225:124165:124341 [5] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-160-225:124165:124341 [5] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-160-225:124165:124341 [5] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-160-225:124165:124341 [5] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-168-34:272151:272326 [1] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-168-34:272151:272326 [1] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-168-34:272151:272326 [1] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-168-34:272151:272326 [1] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-168-34:272151:272326 [1] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-168-34:272151:272326 [1] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-168-34:272151:272326 [1] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-168-34:272151:272326 [1] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-168-34:272151:272326 [1] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-168-34:272151:272326 [1] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-168-34:272151:272326 [1] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-168-34:272151:272326 [1] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-168-34:272151:272326 [1] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-168-34:272151:272326 [1] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-168-34:272151:272326 [1] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-168-34:272151:272326 [1] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-168-34:272151:272326 [1] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-168-34:272151:272326 [1] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-168-34:272151:272326 [1] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-168-34:272151:272326 [1] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-168-34:272151:272326 [1] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-168-34:272151:272326 [1] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-168-34:272151:272326 [1] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-168-34:272151:272326 [1] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-168-34:272151:272326 [1] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-168-34:272151:272326 [1] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-168-34:272151:272326 [1] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-168-34:272151:272326 [1] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-168-34:272151:272326 [1] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-168-34:272151:272326 [1] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-168-34:272151:272326 [1] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-168-34:272151:272326 [1] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-168-34:272156:272325 [6] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/408
+ip-26-0-168-34:272156:272325 [6] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-168-34:272150:272327 [0] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/408
+ip-26-0-168-34:272150:272327 [0] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/408
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-168-34:272151:272326 [1] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/408
+ip-26-0-168-34:272151:272326 [1] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/408
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/408
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-160-225:124163:124340 [3] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/408
+ip-26-0-160-225:124163:124340 [3] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-160-225:124165:124341 [5] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/408
+ip-26-0-160-225:124165:124341 [5] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-161-123:32707:32878 [3] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-161-123:32707:32878 [3] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-161-123:32707:32878 [3] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-161-123:32707:32878 [3] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-161-123:32707:32878 [3] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-161-123:32707:32878 [3] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-161-123:32707:32878 [3] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-161-123:32707:32878 [3] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-161-123:32707:32878 [3] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-161-123:32707:32878 [3] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-161-123:32707:32878 [3] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-161-123:32707:32878 [3] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-161-123:32707:32878 [3] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-161-123:32707:32878 [3] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-161-123:32707:32878 [3] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-161-123:32707:32878 [3] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-161-123:32707:32878 [3] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-161-123:32707:32878 [3] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-161-123:32707:32878 [3] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-161-123:32707:32878 [3] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-161-123:32707:32878 [3] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-161-123:32707:32878 [3] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-161-123:32707:32878 [3] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-161-123:32707:32878 [3] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-161-123:32707:32878 [3] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-161-123:32707:32878 [3] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-161-123:32707:32878 [3] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-161-123:32707:32878 [3] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-161-123:32707:32878 [3] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-161-123:32707:32878 [3] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-161-123:32707:32878 [3] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-161-123:32707:32878 [3] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-165-202:153566:153741 [4] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-165-202:153566:153741 [4] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-165-202:153566:153741 [4] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-165-202:153566:153741 [4] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-165-202:153566:153741 [4] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-165-202:153566:153741 [4] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-165-202:153566:153741 [4] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-165-202:153566:153741 [4] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-165-202:153566:153741 [4] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-165-202:153566:153741 [4] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-161-123:32707:32878 [3] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/408
+ip-26-0-165-202:153566:153741 [4] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-165-202:153566:153741 [4] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-165-202:153566:153741 [4] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-165-202:153566:153741 [4] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-165-202:153566:153741 [4] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-165-202:153566:153741 [4] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-165-202:153566:153741 [4] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-165-202:153566:153741 [4] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-165-202:153566:153741 [4] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-165-202:153566:153741 [4] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-165-202:153566:153741 [4] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-165-202:153566:153741 [4] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-165-202:153566:153741 [4] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-165-202:153566:153741 [4] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-165-202:153566:153741 [4] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-165-202:153566:153741 [4] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-165-202:153566:153741 [4] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-165-202:153566:153741 [4] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-165-202:153566:153741 [4] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-165-202:153566:153741 [4] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-161-123:32707:32878 [3] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-165-202:153566:153741 [4] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-165-202:153566:153741 [4] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/408
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-165-202:153566:153741 [4] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/408
+ip-26-0-165-202:153566:153741 [4] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-161-123:32704:32881 [0] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-161-123:32704:32881 [0] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-161-123:32704:32881 [0] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-161-123:32704:32881 [0] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-161-123:32704:32881 [0] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-161-123:32704:32881 [0] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-161-123:32704:32881 [0] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-161-123:32704:32881 [0] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-161-123:32704:32881 [0] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-161-123:32704:32881 [0] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-161-123:32704:32881 [0] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-161-123:32704:32881 [0] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-161-123:32704:32881 [0] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-161-123:32704:32881 [0] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-161-123:32704:32881 [0] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-161-123:32704:32881 [0] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-161-123:32704:32881 [0] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-161-123:32704:32881 [0] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-161-123:32704:32881 [0] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-161-123:32704:32881 [0] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-161-123:32704:32881 [0] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-161-123:32704:32881 [0] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-161-123:32704:32881 [0] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-161-123:32704:32881 [0] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-161-123:32704:32881 [0] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-161-123:32704:32881 [0] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-161-123:32704:32881 [0] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-161-123:32704:32881 [0] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-161-123:32704:32881 [0] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-161-123:32704:32881 [0] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-161-123:32704:32881 [0] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-161-123:32704:32881 [0] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-165-202:153568:153736 [6] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-165-202:153568:153736 [6] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-165-202:153568:153736 [6] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-165-202:153568:153736 [6] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-165-202:153568:153736 [6] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-165-202:153568:153736 [6] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-165-202:153568:153736 [6] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-165-202:153568:153736 [6] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-165-202:153568:153736 [6] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-165-202:153568:153736 [6] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-165-202:153568:153736 [6] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-165-202:153568:153736 [6] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-165-202:153568:153736 [6] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-165-202:153568:153736 [6] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-165-202:153568:153736 [6] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-165-202:153568:153736 [6] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-165-202:153568:153736 [6] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-165-202:153568:153736 [6] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-165-202:153568:153736 [6] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-165-202:153568:153736 [6] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-165-202:153568:153736 [6] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-165-202:153568:153736 [6] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-165-202:153568:153736 [6] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-165-202:153568:153736 [6] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-165-202:153568:153736 [6] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-165-202:153568:153736 [6] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-165-202:153568:153736 [6] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-165-202:153568:153736 [6] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-165-202:153568:153736 [6] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-165-202:153568:153736 [6] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-165-202:153568:153736 [6] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-165-202:153568:153736 [6] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-161-103:96895:97069 [7] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-161-103:96895:97069 [7] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-161-103:96895:97069 [7] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-161-103:96895:97069 [7] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-161-103:96895:97069 [7] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-161-103:96895:97069 [7] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-161-103:96895:97069 [7] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-161-103:96895:97069 [7] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-161-103:96895:97069 [7] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-161-103:96895:97069 [7] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-161-103:96895:97069 [7] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-161-103:96895:97069 [7] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-161-103:96895:97069 [7] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-161-103:96895:97069 [7] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-161-103:96895:97069 [7] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-161-103:96895:97069 [7] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-161-103:96895:97069 [7] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-161-103:96895:97069 [7] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-161-103:96895:97069 [7] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-161-103:96895:97069 [7] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-161-103:96895:97069 [7] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-161-103:96895:97069 [7] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-161-103:96895:97069 [7] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-161-103:96895:97069 [7] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-161-103:96895:97069 [7] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-161-103:96895:97069 [7] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-161-103:96895:97069 [7] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-161-103:96895:97069 [7] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-161-103:96895:97069 [7] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-161-103:96895:97069 [7] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-161-103:96895:97069 [7] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-161-103:96895:97069 [7] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-165-202:153562:153735 [0] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-165-202:153562:153735 [0] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-165-202:153562:153735 [0] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-165-202:153562:153735 [0] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-165-202:153562:153735 [0] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-165-202:153562:153735 [0] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-165-202:153562:153735 [0] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-165-202:153562:153735 [0] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-165-202:153562:153735 [0] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-165-202:153562:153735 [0] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-165-202:153562:153735 [0] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-165-202:153562:153735 [0] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-165-202:153562:153735 [0] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-165-202:153562:153735 [0] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-165-202:153562:153735 [0] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-165-202:153562:153735 [0] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-165-202:153562:153735 [0] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-165-202:153562:153735 [0] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-165-202:153562:153735 [0] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-165-202:153562:153735 [0] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-165-202:153562:153735 [0] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-165-202:153562:153735 [0] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-165-202:153562:153735 [0] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-165-202:153562:153735 [0] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-165-202:153562:153735 [0] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-165-202:153562:153735 [0] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-165-202:153562:153735 [0] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-165-202:153562:153735 [0] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-165-202:153562:153735 [0] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-165-202:153562:153735 [0] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-165-202:153562:153735 [0] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-165-202:153562:153735 [0] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-161-123:32704:32881 [0] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/408
+ip-26-0-161-123:32704:32881 [0] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-161-103:96895:97069 [7] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/408
+ip-26-0-165-202:153568:153736 [6] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/408
+ip-26-0-161-103:96895:97069 [7] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-165-202:153568:153736 [6] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-165-202:153565:153738 [3] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-165-202:153565:153738 [3] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-165-202:153565:153738 [3] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-165-202:153565:153738 [3] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-165-202:153565:153738 [3] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-165-202:153565:153738 [3] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-165-202:153565:153738 [3] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-165-202:153565:153738 [3] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-165-202:153565:153738 [3] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-165-202:153565:153738 [3] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-165-202:153565:153738 [3] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-165-202:153565:153738 [3] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-165-202:153565:153738 [3] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-165-202:153565:153738 [3] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-165-202:153565:153738 [3] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-165-202:153565:153738 [3] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-165-202:153565:153738 [3] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-165-202:153565:153738 [3] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-165-202:153565:153738 [3] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-165-202:153565:153738 [3] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-165-202:153565:153738 [3] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-165-202:153565:153738 [3] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-165-202:153565:153738 [3] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-165-202:153565:153738 [3] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-165-202:153565:153738 [3] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-165-202:153565:153738 [3] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-165-202:153565:153738 [3] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-165-202:153565:153738 [3] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-165-202:153565:153738 [3] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-165-202:153565:153738 [3] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-165-202:153565:153738 [3] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-165-202:153565:153738 [3] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-161-123:32706:32877 [2] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-161-123:32706:32877 [2] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-161-123:32706:32877 [2] NCCL INFO Using network Libfabric
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/408
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-165-202:153562:153735 [0] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/408
+ip-26-0-165-202:153562:153735 [0] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-165-202:153569:153739 [7] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-165-202:153569:153739 [7] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-165-202:153569:153739 [7] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-165-202:153569:153739 [7] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-165-202:153569:153739 [7] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-165-202:153569:153739 [7] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-165-202:153569:153739 [7] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-165-202:153569:153739 [7] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-165-202:153569:153739 [7] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-165-202:153569:153739 [7] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-165-202:153569:153739 [7] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-165-202:153569:153739 [7] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-165-202:153569:153739 [7] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-165-202:153569:153739 [7] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-165-202:153569:153739 [7] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-165-202:153569:153739 [7] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-165-202:153569:153739 [7] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-165-202:153569:153739 [7] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-165-202:153569:153739 [7] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-165-202:153569:153739 [7] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-165-202:153569:153739 [7] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-165-202:153569:153739 [7] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-165-202:153569:153739 [7] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-165-202:153569:153739 [7] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-165-202:153569:153739 [7] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-165-202:153569:153739 [7] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-165-202:153569:153739 [7] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-165-202:153569:153739 [7] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-165-202:153569:153739 [7] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-165-202:153569:153739 [7] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-165-202:153569:153739 [7] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-165-202:153569:153739 [7] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-160-225:124164:124336 [4] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-160-225:124164:124336 [4] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-160-225:124164:124336 [4] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-160-225:124164:124336 [4] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-160-225:124164:124336 [4] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-160-225:124164:124336 [4] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-160-225:124164:124336 [4] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-160-225:124164:124336 [4] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-160-225:124164:124336 [4] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-160-225:124164:124336 [4] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-160-225:124164:124336 [4] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-160-225:124164:124336 [4] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-160-225:124164:124336 [4] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-160-225:124164:124336 [4] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-160-225:124164:124336 [4] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-160-225:124164:124336 [4] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-160-225:124164:124336 [4] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-160-225:124164:124336 [4] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-160-225:124164:124336 [4] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-160-225:124164:124336 [4] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-160-225:124164:124336 [4] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-160-225:124164:124336 [4] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-160-225:124164:124336 [4] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-160-225:124164:124336 [4] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-160-225:124164:124336 [4] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-160-225:124164:124336 [4] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-160-225:124164:124336 [4] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-160-225:124164:124336 [4] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-160-225:124164:124336 [4] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-160-225:124164:124336 [4] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-160-225:124164:124336 [4] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-160-225:124164:124336 [4] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-165-202:153565:153738 [3] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/408
+ip-26-0-165-202:153565:153738 [3] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-165-202:153569:153739 [7] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/408
+ip-26-0-165-202:153569:153739 [7] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-160-225:124167:124338 [7] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-160-225:124167:124338 [7] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-160-225:124167:124338 [7] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-160-225:124167:124338 [7] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-160-225:124167:124338 [7] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-160-225:124167:124338 [7] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-160-225:124167:124338 [7] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-160-225:124167:124338 [7] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-160-225:124167:124338 [7] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-160-225:124167:124338 [7] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-160-225:124167:124338 [7] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-160-225:124167:124338 [7] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-160-225:124167:124338 [7] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-160-225:124167:124338 [7] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-160-225:124167:124338 [7] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-160-225:124167:124338 [7] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-160-225:124167:124338 [7] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-160-225:124167:124338 [7] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-160-225:124167:124338 [7] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-160-225:124167:124338 [7] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-160-225:124167:124338 [7] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-160-225:124167:124338 [7] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-160-225:124167:124338 [7] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-160-225:124167:124338 [7] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-160-225:124167:124338 [7] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-160-225:124167:124338 [7] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-160-225:124167:124338 [7] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-160-225:124167:124338 [7] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-160-225:124167:124338 [7] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-160-225:124167:124338 [7] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-160-225:124167:124338 [7] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-160-225:124167:124338 [7] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-161-142:62882:63056 [1] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-161-142:62882:63056 [1] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-161-142:62882:63056 [1] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-161-142:62882:63056 [1] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-161-142:62882:63056 [1] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-161-142:62882:63056 [1] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-161-142:62882:63056 [1] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-161-142:62882:63056 [1] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-161-142:62882:63056 [1] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-161-142:62882:63056 [1] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-161-142:62882:63056 [1] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-161-142:62882:63056 [1] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-161-142:62882:63056 [1] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-161-142:62882:63056 [1] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-161-142:62882:63056 [1] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-161-142:62882:63056 [1] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-161-142:62882:63056 [1] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-161-142:62882:63056 [1] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-161-142:62882:63056 [1] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-161-142:62882:63056 [1] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-161-142:62882:63056 [1] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-161-142:62882:63056 [1] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-161-142:62882:63056 [1] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-161-142:62882:63056 [1] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-161-142:62882:63056 [1] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-161-142:62882:63056 [1] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-161-142:62882:63056 [1] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-161-142:62882:63056 [1] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-161-142:62882:63056 [1] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-161-142:62882:63056 [1] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-161-142:62882:63056 [1] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-161-142:62882:63056 [1] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-161-123:32709:32882 [5] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-161-123:32709:32882 [5] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-161-123:32709:32882 [5] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-161-123:32709:32882 [5] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-161-123:32709:32882 [5] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-161-123:32709:32882 [5] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-161-123:32709:32882 [5] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-161-123:32709:32882 [5] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-161-123:32709:32882 [5] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-161-123:32709:32882 [5] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-161-123:32709:32882 [5] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-161-123:32709:32882 [5] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-161-123:32709:32882 [5] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-161-123:32709:32882 [5] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-161-123:32709:32882 [5] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-161-123:32709:32882 [5] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-161-123:32709:32882 [5] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-161-123:32709:32882 [5] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-161-123:32709:32882 [5] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-161-123:32709:32882 [5] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-161-123:32709:32882 [5] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-161-123:32709:32882 [5] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-161-123:32709:32882 [5] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-161-123:32709:32882 [5] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-161-123:32709:32882 [5] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-161-123:32709:32882 [5] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-161-123:32709:32882 [5] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-161-123:32709:32882 [5] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-161-123:32709:32882 [5] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-161-123:32709:32882 [5] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-161-123:32709:32882 [5] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-161-123:32709:32882 [5] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-160-225:124164:124336 [4] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/408
+ip-26-0-160-225:124164:124336 [4] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/408
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-160-225:124166:124339 [6] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-160-225:124166:124339 [6] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-160-225:124166:124339 [6] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-160-225:124166:124339 [6] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-160-225:124166:124339 [6] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-160-225:124166:124339 [6] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-168-52:2722052:2722226 [7] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-168-52:2722052:2722226 [7] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-168-52:2722052:2722226 [7] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-168-52:2722052:2722226 [7] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-168-52:2722052:2722226 [7] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-160-225:124166:124339 [6] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-160-225:124166:124339 [6] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-160-225:124166:124339 [6] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-160-225:124166:124339 [6] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-160-225:124166:124339 [6] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-160-225:124166:124339 [6] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-160-225:124166:124339 [6] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-160-225:124166:124339 [6] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-160-225:124166:124339 [6] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-160-225:124166:124339 [6] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-168-52:2722052:2722226 [7] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-168-52:2722052:2722226 [7] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-168-52:2722052:2722226 [7] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-168-52:2722052:2722226 [7] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-168-52:2722052:2722226 [7] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-160-225:124166:124339 [6] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-160-225:124166:124339 [6] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-160-225:124166:124339 [6] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-160-225:124166:124339 [6] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-160-225:124166:124339 [6] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-168-52:2722052:2722226 [7] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-168-52:2722052:2722226 [7] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-168-52:2722052:2722226 [7] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-168-52:2722052:2722226 [7] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-168-52:2722052:2722226 [7] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-168-52:2722052:2722226 [7] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-168-52:2722052:2722226 [7] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-168-52:2722052:2722226 [7] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-168-52:2722052:2722226 [7] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-168-52:2722052:2722226 [7] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-168-52:2722052:2722226 [7] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-160-225:124166:124339 [6] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-160-225:124166:124339 [6] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-160-225:124166:124339 [6] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-160-225:124166:124339 [6] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-160-225:124166:124339 [6] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-160-225:124166:124339 [6] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-160-225:124166:124339 [6] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-160-225:124166:124339 [6] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-160-225:124166:124339 [6] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-168-52:2722052:2722226 [7] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-160-225:124166:124339 [6] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-160-225:124166:124339 [6] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-168-52:2722052:2722226 [7] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-168-52:2722052:2722226 [7] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-168-52:2722052:2722226 [7] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-168-52:2722052:2722226 [7] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-168-52:2722052:2722226 [7] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-168-52:2722052:2722226 [7] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-168-52:2722052:2722226 [7] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-168-52:2722052:2722226 [7] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-168-52:2722052:2722226 [7] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-168-52:2722052:2722226 [7] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-160-225:124162:124342 [2] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-160-225:124162:124342 [2] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-160-225:124162:124342 [2] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-160-225:124162:124342 [2] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-160-225:124162:124342 [2] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-160-225:124162:124342 [2] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-160-225:124162:124342 [2] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-160-225:124162:124342 [2] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-160-225:124162:124342 [2] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-160-225:124162:124342 [2] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-160-225:124162:124342 [2] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-160-225:124162:124342 [2] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-160-225:124162:124342 [2] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-160-225:124162:124342 [2] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-160-225:124162:124342 [2] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-160-225:124162:124342 [2] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-160-225:124162:124342 [2] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-160-225:124162:124342 [2] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-160-225:124162:124342 [2] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-160-225:124162:124342 [2] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-160-225:124162:124342 [2] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-160-225:124162:124342 [2] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-160-225:124162:124342 [2] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-160-225:124162:124342 [2] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-160-225:124162:124342 [2] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-160-225:124162:124342 [2] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-160-225:124162:124342 [2] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-160-225:124162:124342 [2] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-160-225:124162:124342 [2] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-160-225:124162:124342 [2] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-160-225:124162:124342 [2] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-160-225:124162:124342 [2] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-161-123:32709:32882 [5] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/408
+ip-26-0-161-123:32709:32882 [5] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-161-142:62882:63056 [1] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/408
+ip-26-0-161-142:62882:63056 [1] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-161-123:32710:32883 [6] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-161-123:32710:32883 [6] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-161-123:32710:32883 [6] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-161-123:32710:32883 [6] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-161-123:32710:32883 [6] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-161-123:32710:32883 [6] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-161-123:32710:32883 [6] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-161-123:32710:32883 [6] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-161-123:32710:32883 [6] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-161-123:32710:32883 [6] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-161-123:32710:32883 [6] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-161-123:32710:32883 [6] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-168-52:2722052:2722226 [7] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/408
+ip-26-0-161-123:32710:32883 [6] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-168-52:2722052:2722226 [7] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-161-123:32710:32883 [6] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-161-123:32710:32883 [6] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-161-123:32710:32883 [6] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-161-123:32710:32883 [6] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-161-123:32710:32883 [6] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-161-123:32710:32883 [6] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-161-123:32710:32883 [6] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-161-123:32710:32883 [6] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-161-123:32710:32883 [6] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-161-123:32710:32883 [6] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-161-123:32710:32883 [6] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-161-123:32710:32883 [6] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-161-123:32710:32883 [6] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-161-123:32710:32883 [6] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-161-123:32710:32883 [6] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-161-123:32710:32883 [6] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-161-123:32710:32883 [6] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-161-123:32710:32883 [6] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-161-123:32710:32883 [6] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-160-225:124167:124338 [7] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/408
+ip-26-0-160-225:124167:124338 [7] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-160-225:124166:124339 [6] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/408
+ip-26-0-160-225:124166:124339 [6] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-161-123:32711:32884 [7] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-161-123:32711:32884 [7] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-161-123:32711:32884 [7] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-161-123:32711:32884 [7] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-161-123:32711:32884 [7] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-161-123:32711:32884 [7] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-161-123:32711:32884 [7] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-161-123:32711:32884 [7] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-161-123:32711:32884 [7] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-161-123:32711:32884 [7] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-161-123:32711:32884 [7] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-161-123:32711:32884 [7] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-161-123:32711:32884 [7] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-161-123:32711:32884 [7] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-161-123:32711:32884 [7] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-161-123:32711:32884 [7] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-161-123:32711:32884 [7] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-161-123:32711:32884 [7] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-161-123:32711:32884 [7] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-161-123:32711:32884 [7] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-161-123:32711:32884 [7] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-161-123:32711:32884 [7] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-161-123:32711:32884 [7] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-161-123:32711:32884 [7] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-161-123:32711:32884 [7] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-161-123:32711:32884 [7] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-161-123:32711:32884 [7] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-161-123:32711:32884 [7] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-161-123:32711:32884 [7] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-161-123:32711:32884 [7] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-161-123:32711:32884 [7] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-161-123:32711:32884 [7] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-160-225:124162:124342 [2] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/408
+ip-26-0-160-225:124162:124342 [2] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-161-123:32710:32883 [6] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/408
+ip-26-0-161-123:32710:32883 [6] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-168-34:272152:272331 [2] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-168-34:272152:272331 [2] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-168-34:272152:272331 [2] NCCL INFO Using network Libfabric
+ip-26-0-161-123:32711:32884 [7] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/408
+ip-26-0-161-123:32711:32884 [7] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-161-123:32708:32879 [4] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-161-123:32708:32879 [4] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-161-123:32708:32879 [4] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-161-123:32708:32879 [4] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-161-123:32708:32879 [4] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-161-123:32708:32879 [4] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-161-123:32708:32879 [4] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-161-123:32708:32879 [4] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-161-123:32708:32879 [4] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-161-123:32708:32879 [4] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-161-123:32708:32879 [4] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-161-123:32708:32879 [4] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-161-123:32708:32879 [4] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-161-123:32708:32879 [4] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-161-123:32708:32879 [4] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-161-123:32708:32879 [4] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-161-123:32708:32879 [4] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-161-123:32708:32879 [4] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-161-123:32708:32879 [4] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-161-123:32708:32879 [4] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-161-123:32708:32879 [4] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-161-123:32708:32879 [4] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-161-123:32708:32879 [4] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-161-123:32708:32879 [4] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-161-123:32708:32879 [4] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-161-123:32708:32879 [4] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-161-123:32708:32879 [4] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-161-123:32708:32879 [4] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-161-123:32708:32879 [4] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-161-123:32708:32879 [4] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-161-123:32708:32879 [4] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-161-123:32708:32879 [4] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-161-123:32708:32879 [4] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/408
+ip-26-0-161-123:32708:32879 [4] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/408
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-161-142:62884:63061 [3] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-161-142:62884:63061 [3] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-161-142:62884:63061 [3] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-161-142:62884:63061 [3] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-161-142:62884:63061 [3] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-161-142:62884:63061 [3] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-161-142:62884:63061 [3] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/408
+ip-26-0-161-142:62884:63061 [3] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-161-142:62884:63061 [3] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-161-142:62884:63061 [3] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-161-142:62884:63061 [3] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-161-142:62884:63061 [3] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-161-142:62884:63061 [3] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-161-142:62884:63061 [3] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-161-142:62884:63061 [3] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-161-142:62884:63061 [3] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-161-142:62884:63061 [3] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-161-142:62884:63061 [3] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-161-142:62884:63061 [3] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-161-142:62884:63061 [3] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-161-142:62884:63061 [3] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-161-142:62884:63061 [3] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-161-142:62884:63061 [3] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-161-142:62884:63061 [3] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-161-142:62884:63061 [3] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-161-142:62884:63061 [3] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-161-142:62884:63061 [3] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-161-142:62884:63061 [3] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-161-142:62884:63061 [3] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-161-142:62884:63061 [3] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-161-142:62884:63061 [3] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-161-142:62884:63061 [3] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO Using network Libfabric
+ip-26-0-161-142:62884:63061 [3] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/408
+ip-26-0-161-142:62884:63061 [3] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/408
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-161-142:62888:63060 [7] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-161-142:62888:63060 [7] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-161-142:62888:63060 [7] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-161-142:62888:63060 [7] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-161-142:62888:63060 [7] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-161-142:62888:63060 [7] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-161-142:62888:63060 [7] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-161-142:62888:63060 [7] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-161-142:62888:63060 [7] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-161-142:62888:63060 [7] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-161-142:62888:63060 [7] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-161-142:62888:63060 [7] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-161-142:62888:63060 [7] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-161-142:62888:63060 [7] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-161-142:62888:63060 [7] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-161-142:62888:63060 [7] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-161-142:62888:63060 [7] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-161-142:62888:63060 [7] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-161-142:62888:63060 [7] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-161-142:62888:63060 [7] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-161-142:62888:63060 [7] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-161-142:62888:63060 [7] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-161-142:62888:63060 [7] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-161-142:62888:63060 [7] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-161-142:62888:63060 [7] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-161-142:62888:63060 [7] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-161-142:62888:63060 [7] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-161-142:62888:63060 [7] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-161-142:62888:63060 [7] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-161-142:62888:63060 [7] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-161-142:62888:63060 [7] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-161-142:62888:63060 [7] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/409
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-161-142:62883:63058 [2] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-161-142:62883:63058 [2] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-161-142:62883:63058 [2] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-161-142:62883:63058 [2] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-161-142:62883:63058 [2] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-161-142:62883:63058 [2] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-161-142:62883:63058 [2] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-161-142:62883:63058 [2] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-161-142:62883:63058 [2] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-161-142:62883:63058 [2] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-161-142:62883:63058 [2] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-161-142:62883:63058 [2] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-161-142:62883:63058 [2] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-161-142:62883:63058 [2] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-161-142:62883:63058 [2] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-161-142:62883:63058 [2] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-161-142:62883:63058 [2] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-161-142:62883:63058 [2] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-161-142:62883:63058 [2] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-161-142:62883:63058 [2] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-161-142:62883:63058 [2] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-161-142:62883:63058 [2] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-161-142:62883:63058 [2] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-161-142:62883:63058 [2] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-161-142:62883:63058 [2] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-161-142:62883:63058 [2] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-161-142:62883:63058 [2] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-161-142:62883:63058 [2] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-161-142:62883:63058 [2] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-161-142:62883:63058 [2] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-161-142:62883:63058 [2] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-161-142:62883:63058 [2] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/408
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-161-142:62888:63060 [7] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/408
+ip-26-0-161-142:62888:63060 [7] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-168-52:2722047:2722221 [2] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-168-52:2722047:2722221 [2] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-168-52:2722047:2722221 [2] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-168-52:2722047:2722221 [2] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-168-52:2722047:2722221 [2] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-168-52:2722047:2722221 [2] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-168-52:2722047:2722221 [2] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-168-52:2722047:2722221 [2] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-168-52:2722047:2722221 [2] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-168-52:2722047:2722221 [2] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-168-52:2722047:2722221 [2] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-168-52:2722047:2722221 [2] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-168-52:2722047:2722221 [2] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-168-52:2722047:2722221 [2] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-168-52:2722047:2722221 [2] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-168-52:2722047:2722221 [2] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-168-52:2722047:2722221 [2] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-168-52:2722047:2722221 [2] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-168-52:2722047:2722221 [2] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-168-52:2722047:2722221 [2] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-168-52:2722047:2722221 [2] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-168-52:2722047:2722221 [2] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-168-52:2722047:2722221 [2] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-168-52:2722047:2722221 [2] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-168-52:2722047:2722221 [2] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-168-52:2722047:2722221 [2] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-168-52:2722047:2722221 [2] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-168-52:2722047:2722221 [2] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-168-52:2722047:2722221 [2] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-168-52:2722047:2722221 [2] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-168-52:2722047:2722221 [2] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-168-52:2722047:2722221 [2] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-161-138:96565:96754 [2] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-161-138:96565:96754 [2] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-161-138:96565:96754 [2] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-161-138:96565:96754 [2] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-161-138:96565:96754 [2] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-161-138:96565:96754 [2] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-161-138:96565:96754 [2] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-161-138:96565:96754 [2] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-161-138:96565:96754 [2] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-161-138:96565:96754 [2] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-161-138:96565:96754 [2] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-161-138:96565:96754 [2] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-161-138:96565:96754 [2] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-161-138:96565:96754 [2] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-161-138:96565:96754 [2] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-161-138:96565:96754 [2] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-161-138:96565:96754 [2] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-161-138:96565:96754 [2] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-161-138:96565:96754 [2] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-161-138:96565:96754 [2] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-161-138:96565:96754 [2] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-161-138:96565:96754 [2] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-161-138:96565:96754 [2] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-161-138:96565:96754 [2] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-161-138:96565:96754 [2] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-161-138:96565:96754 [2] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-161-138:96565:96754 [2] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-161-138:96565:96754 [2] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-161-138:96565:96754 [2] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-161-138:96565:96754 [2] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-161-138:96565:96754 [2] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-161-138:96565:96754 [2] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-161-142:62883:63058 [2] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/408
+ip-26-0-161-142:62883:63058 [2] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/408
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-161-138:96565:96754 [2] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/408
+ip-26-0-161-138:96565:96754 [2] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-168-52:2722047:2722221 [2] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/408
+ip-26-0-168-52:2722047:2722221 [2] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-168-52:2722050:2722225 [5] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-168-52:2722050:2722225 [5] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-168-52:2722050:2722225 [5] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-168-52:2722050:2722225 [5] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-168-52:2722050:2722225 [5] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-168-52:2722050:2722225 [5] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-168-52:2722050:2722225 [5] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-168-52:2722050:2722225 [5] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-168-52:2722050:2722225 [5] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-168-52:2722050:2722225 [5] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-168-52:2722050:2722225 [5] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-168-52:2722050:2722225 [5] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-168-52:2722050:2722225 [5] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-168-52:2722050:2722225 [5] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-168-52:2722050:2722225 [5] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-168-52:2722050:2722225 [5] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-168-52:2722050:2722225 [5] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-168-52:2722050:2722225 [5] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-168-52:2722050:2722225 [5] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-168-52:2722050:2722225 [5] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-168-52:2722050:2722225 [5] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-168-52:2722050:2722225 [5] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-168-52:2722050:2722225 [5] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-168-52:2722050:2722225 [5] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-168-52:2722050:2722225 [5] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-168-52:2722050:2722225 [5] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-168-52:2722050:2722225 [5] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-168-52:2722050:2722225 [5] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-168-52:2722050:2722225 [5] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-168-52:2722050:2722225 [5] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-168-52:2722050:2722225 [5] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-168-52:2722050:2722225 [5] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/408
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-161-142:62886:63057 [5] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-161-142:62886:63057 [5] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-161-142:62886:63057 [5] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-161-142:62886:63057 [5] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-161-142:62886:63057 [5] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-161-142:62886:63057 [5] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-161-142:62886:63057 [5] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-161-142:62886:63057 [5] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-161-142:62886:63057 [5] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-161-142:62886:63057 [5] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-161-142:62886:63057 [5] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-161-142:62886:63057 [5] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-161-142:62886:63057 [5] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-161-142:62886:63057 [5] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-161-142:62886:63057 [5] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-161-142:62886:63057 [5] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-161-142:62886:63057 [5] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-161-142:62886:63057 [5] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-161-142:62886:63057 [5] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-161-142:62886:63057 [5] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-161-142:62886:63057 [5] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-161-142:62886:63057 [5] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-161-142:62886:63057 [5] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-161-142:62886:63057 [5] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-161-142:62886:63057 [5] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-161-142:62886:63057 [5] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-161-142:62886:63057 [5] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-161-142:62886:63057 [5] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-161-142:62886:63057 [5] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-161-142:62886:63057 [5] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-161-142:62886:63057 [5] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-161-142:62886:63057 [5] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-168-52:2722050:2722225 [5] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/408
+ip-26-0-161-142:62881:63059 [0] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-161-142:62881:63059 [0] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-161-142:62881:63059 [0] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-161-142:62881:63059 [0] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-161-142:62881:63059 [0] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-161-142:62881:63059 [0] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-161-142:62881:63059 [0] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-161-142:62881:63059 [0] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-161-142:62881:63059 [0] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-161-142:62881:63059 [0] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-161-142:62881:63059 [0] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-161-142:62881:63059 [0] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-161-142:62881:63059 [0] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-161-142:62881:63059 [0] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-161-142:62881:63059 [0] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-161-142:62881:63059 [0] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-161-142:62881:63059 [0] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-161-142:62881:63059 [0] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-161-142:62881:63059 [0] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-161-142:62881:63059 [0] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-168-52:2722050:2722225 [5] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-161-142:62881:63059 [0] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-161-142:62881:63059 [0] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-161-142:62881:63059 [0] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-161-142:62881:63059 [0] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-161-142:62881:63059 [0] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-161-142:62881:63059 [0] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-161-142:62881:63059 [0] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-161-142:62881:63059 [0] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-161-142:62881:63059 [0] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-161-142:62881:63059 [0] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-161-142:62881:63059 [0] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-161-142:62881:63059 [0] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-161-103:96889:97065 [1] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-161-103:96889:97065 [1] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-161-103:96889:97065 [1] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-161-103:96889:97065 [1] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-161-103:96889:97065 [1] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-161-103:96889:97065 [1] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-161-103:96889:97065 [1] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-161-103:96889:97065 [1] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-161-103:96889:97065 [1] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-161-103:96889:97065 [1] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-161-103:96889:97065 [1] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-161-103:96889:97065 [1] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-161-103:96889:97065 [1] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-161-103:96889:97065 [1] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-161-103:96889:97065 [1] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-161-103:96889:97065 [1] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-161-103:96889:97065 [1] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-161-103:96889:97065 [1] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-161-103:96889:97065 [1] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-161-103:96889:97065 [1] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-161-103:96889:97065 [1] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-161-103:96889:97065 [1] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-161-103:96889:97065 [1] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-161-103:96889:97065 [1] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-161-103:96889:97065 [1] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-161-103:96889:97065 [1] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-161-103:96889:97065 [1] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-161-103:96889:97065 [1] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-161-103:96889:97065 [1] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-161-103:96889:97065 [1] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-161-103:96889:97065 [1] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-161-103:96889:97065 [1] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-161-142:62886:63057 [5] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/408
+ip-26-0-161-142:62886:63057 [5] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-161-142:62881:63059 [0] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/408
+ip-26-0-161-142:62881:63059 [0] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/408
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-161-103:96889:97065 [1] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/408
+ip-26-0-161-103:96889:97065 [1] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-168-30:235651:235823 [3] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-168-30:235651:235823 [3] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-168-30:235651:235823 [3] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-168-30:235651:235823 [3] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-168-30:235651:235823 [3] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-168-30:235651:235823 [3] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-168-30:235651:235823 [3] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-168-30:235651:235823 [3] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-168-30:235651:235823 [3] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-168-30:235651:235823 [3] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-168-30:235651:235823 [3] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-168-30:235651:235823 [3] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-168-30:235651:235823 [3] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-168-30:235651:235823 [3] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-168-30:235651:235823 [3] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-168-30:235651:235823 [3] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-168-30:235651:235823 [3] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-168-30:235651:235823 [3] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-168-30:235651:235823 [3] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-168-30:235651:235823 [3] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-168-30:235651:235823 [3] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-168-30:235651:235823 [3] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-168-30:235651:235823 [3] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-168-30:235651:235823 [3] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-168-30:235651:235823 [3] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-168-30:235651:235823 [3] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-168-30:235651:235823 [3] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-168-30:235651:235823 [3] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-168-30:235651:235823 [3] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-168-30:235651:235823 [3] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-168-30:235651:235823 [3] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-168-30:235651:235823 [3] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-160-225:124161:124337 [1] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-160-225:124161:124337 [1] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-160-225:124161:124337 [1] NCCL INFO Using network Libfabric
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/408
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-168-30:235651:235823 [3] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/408
+ip-26-0-168-30:235651:235823 [3] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-168-52:2722051:2722224 [6] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-168-52:2722051:2722224 [6] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-168-52:2722051:2722224 [6] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-168-52:2722051:2722224 [6] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-168-52:2722051:2722224 [6] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-168-52:2722051:2722224 [6] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-168-52:2722051:2722224 [6] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-168-52:2722051:2722224 [6] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-168-52:2722051:2722224 [6] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-168-52:2722051:2722224 [6] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-168-52:2722051:2722224 [6] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-168-52:2722051:2722224 [6] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-168-52:2722051:2722224 [6] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-168-52:2722051:2722224 [6] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-168-52:2722051:2722224 [6] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-168-52:2722051:2722224 [6] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-168-52:2722051:2722224 [6] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-168-52:2722051:2722224 [6] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-168-52:2722051:2722224 [6] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-168-52:2722051:2722224 [6] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-168-52:2722051:2722224 [6] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-168-52:2722051:2722224 [6] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-168-52:2722051:2722224 [6] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-168-52:2722051:2722224 [6] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-168-52:2722051:2722224 [6] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-168-52:2722051:2722224 [6] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-168-52:2722051:2722224 [6] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-168-52:2722051:2722224 [6] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-168-52:2722051:2722224 [6] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-168-52:2722051:2722224 [6] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-168-52:2722051:2722224 [6] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-168-52:2722051:2722224 [6] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-161-78:114517:114687 [5] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-161-78:114517:114687 [5] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-161-78:114517:114687 [5] NCCL INFO Using network Libfabric
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/408
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/408
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-161-103:96891:97063 [3] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-161-103:96891:97063 [3] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-161-103:96891:97063 [3] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-161-103:96891:97063 [3] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-161-103:96891:97063 [3] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-161-103:96891:97063 [3] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-161-103:96891:97063 [3] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-161-103:96891:97063 [3] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-161-103:96891:97063 [3] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-161-103:96891:97063 [3] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-161-103:96891:97063 [3] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-161-103:96891:97063 [3] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-161-103:96891:97063 [3] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-161-103:96891:97063 [3] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-161-103:96891:97063 [3] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-161-103:96891:97063 [3] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-161-103:96891:97063 [3] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-161-103:96891:97063 [3] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-161-103:96891:97063 [3] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-161-103:96891:97063 [3] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-161-103:96891:97063 [3] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-161-103:96891:97063 [3] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-161-103:96891:97063 [3] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-161-103:96891:97063 [3] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-161-103:96891:97063 [3] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-161-103:96891:97063 [3] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-161-103:96891:97063 [3] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-161-103:96891:97063 [3] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-161-103:96891:97063 [3] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-161-103:96891:97063 [3] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-161-103:96891:97063 [3] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-161-103:96891:97063 [3] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-168-52:2722051:2722224 [6] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/408
+ip-26-0-168-52:2722051:2722224 [6] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/408
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-161-103:96891:97063 [3] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/408
+ip-26-0-161-103:96891:97063 [3] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-161-103:96893:97067 [5] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-161-103:96893:97067 [5] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-161-103:96893:97067 [5] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-161-103:96893:97067 [5] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-161-103:96893:97067 [5] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-161-103:96893:97067 [5] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-161-103:96893:97067 [5] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-161-103:96893:97067 [5] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-161-103:96893:97067 [5] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-161-103:96893:97067 [5] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-161-103:96893:97067 [5] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-161-103:96893:97067 [5] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-161-103:96893:97067 [5] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-161-103:96893:97067 [5] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-161-103:96893:97067 [5] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-161-103:96893:97067 [5] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-161-103:96893:97067 [5] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-161-103:96893:97067 [5] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-161-103:96893:97067 [5] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-161-103:96893:97067 [5] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-161-103:96893:97067 [5] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-161-103:96893:97067 [5] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-161-103:96893:97067 [5] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-161-103:96893:97067 [5] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-161-103:96893:97067 [5] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-161-103:96893:97067 [5] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-161-103:96893:97067 [5] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-161-103:96893:97067 [5] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-161-103:96893:97067 [5] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-161-103:96893:97067 [5] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-161-103:96893:97067 [5] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-161-103:96893:97067 [5] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/408
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-161-103:96893:97067 [5] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/408
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/408
+ip-26-0-161-103:96893:97067 [5] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO Using network Libfabric
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/408
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/409
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/408
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-168-52:2722045:2722219 [0] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-168-52:2722045:2722219 [0] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-168-52:2722045:2722219 [0] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-168-52:2722045:2722219 [0] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-168-52:2722045:2722219 [0] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-168-52:2722045:2722219 [0] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-168-52:2722045:2722219 [0] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-168-52:2722045:2722219 [0] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-168-52:2722045:2722219 [0] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-168-52:2722045:2722219 [0] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-168-52:2722045:2722219 [0] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-168-52:2722045:2722219 [0] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-168-52:2722045:2722219 [0] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-168-52:2722045:2722219 [0] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-168-52:2722045:2722219 [0] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-168-52:2722045:2722219 [0] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-168-52:2722045:2722219 [0] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-168-52:2722045:2722219 [0] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-168-52:2722045:2722219 [0] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-168-52:2722045:2722219 [0] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-168-52:2722045:2722219 [0] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-168-52:2722045:2722219 [0] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-168-52:2722045:2722219 [0] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-168-52:2722045:2722219 [0] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-168-52:2722045:2722219 [0] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-168-52:2722045:2722219 [0] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-168-52:2722045:2722219 [0] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-168-52:2722045:2722219 [0] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-168-52:2722045:2722219 [0] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-168-52:2722045:2722219 [0] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-168-52:2722045:2722219 [0] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-168-52:2722045:2722219 [0] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-168-52:2722048:2722220 [3] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-168-52:2722048:2722220 [3] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-168-52:2722048:2722220 [3] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-168-52:2722048:2722220 [3] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-168-52:2722048:2722220 [3] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-168-52:2722048:2722220 [3] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-168-52:2722048:2722220 [3] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-168-52:2722048:2722220 [3] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-168-52:2722048:2722220 [3] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-168-52:2722048:2722220 [3] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-168-52:2722048:2722220 [3] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-168-52:2722048:2722220 [3] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-168-52:2722048:2722220 [3] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-168-52:2722048:2722220 [3] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-168-52:2722048:2722220 [3] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-168-52:2722048:2722220 [3] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-168-52:2722048:2722220 [3] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-168-52:2722048:2722220 [3] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-168-52:2722048:2722220 [3] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-168-52:2722048:2722220 [3] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-168-52:2722048:2722220 [3] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-168-52:2722048:2722220 [3] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-168-52:2722048:2722220 [3] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-168-52:2722048:2722220 [3] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-168-52:2722048:2722220 [3] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-168-52:2722048:2722220 [3] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-168-52:2722048:2722220 [3] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-168-52:2722048:2722220 [3] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-168-52:2722048:2722220 [3] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-168-52:2722048:2722220 [3] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-168-52:2722048:2722220 [3] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-168-52:2722048:2722220 [3] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/408
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-168-52:2722045:2722219 [0] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/408
+ip-26-0-168-52:2722045:2722219 [0] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-168-52:2722048:2722220 [3] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/408
+ip-26-0-168-52:2722048:2722220 [3] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/408
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-161-103:96890:97070 [2] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-161-103:96890:97070 [2] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-161-103:96890:97070 [2] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-161-103:96890:97070 [2] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-161-103:96890:97070 [2] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-161-103:96890:97070 [2] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-161-103:96890:97070 [2] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-161-103:96890:97070 [2] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-161-103:96890:97070 [2] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-161-103:96890:97070 [2] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-161-103:96890:97070 [2] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-161-103:96890:97070 [2] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-161-103:96890:97070 [2] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-161-103:96890:97070 [2] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-161-103:96890:97070 [2] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-161-103:96890:97070 [2] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-161-103:96890:97070 [2] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-161-103:96890:97070 [2] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-161-103:96890:97070 [2] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-161-103:96890:97070 [2] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-161-103:96890:97070 [2] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-161-103:96890:97070 [2] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-161-103:96890:97070 [2] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-161-103:96890:97070 [2] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-161-103:96890:97070 [2] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-161-103:96890:97070 [2] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-161-103:96890:97070 [2] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-161-103:96890:97070 [2] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-161-103:96890:97070 [2] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-161-103:96890:97070 [2] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-161-103:96890:97070 [2] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-161-103:96890:97070 [2] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/409
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-161-103:96890:97070 [2] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/408
+ip-26-0-161-103:96890:97070 [2] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/408
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-161-138:96569:96758 [6] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-161-138:96569:96758 [6] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-161-138:96569:96758 [6] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-161-138:96569:96758 [6] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-161-138:96569:96758 [6] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-161-138:96569:96758 [6] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-161-138:96569:96758 [6] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-161-138:96569:96758 [6] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-161-138:96569:96758 [6] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-161-138:96569:96758 [6] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-161-138:96569:96758 [6] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-161-138:96569:96758 [6] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-161-138:96569:96758 [6] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-161-138:96569:96758 [6] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-161-138:96569:96758 [6] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-161-138:96569:96758 [6] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-161-138:96569:96758 [6] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-161-138:96569:96758 [6] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-161-138:96569:96758 [6] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-161-138:96569:96758 [6] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-161-138:96569:96758 [6] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-161-138:96569:96758 [6] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-161-138:96569:96758 [6] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-161-138:96569:96758 [6] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-161-138:96569:96758 [6] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-161-138:96569:96758 [6] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-161-138:96569:96758 [6] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-161-138:96569:96758 [6] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-161-138:96569:96758 [6] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-161-138:96569:96758 [6] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-161-138:96569:96758 [6] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-161-138:96569:96758 [6] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/408
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-161-138:96566:96755 [3] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-161-138:96566:96755 [3] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-161-138:96566:96755 [3] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-161-138:96566:96755 [3] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-161-138:96566:96755 [3] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-161-138:96566:96755 [3] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-161-138:96566:96755 [3] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-161-138:96566:96755 [3] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-161-138:96566:96755 [3] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-161-138:96566:96755 [3] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-161-138:96566:96755 [3] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-161-138:96566:96755 [3] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-161-138:96566:96755 [3] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-161-138:96566:96755 [3] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-161-138:96566:96755 [3] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-161-138:96566:96755 [3] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-161-138:96566:96755 [3] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-161-138:96566:96755 [3] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-161-138:96566:96755 [3] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-161-138:96566:96755 [3] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-161-138:96566:96755 [3] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-161-138:96566:96755 [3] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-161-138:96566:96755 [3] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-161-138:96566:96755 [3] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-161-138:96566:96755 [3] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-161-138:96566:96755 [3] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-161-138:96566:96755 [3] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-161-138:96566:96755 [3] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-161-138:96566:96755 [3] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-161-138:96566:96755 [3] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-161-138:96566:96755 [3] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-161-138:96566:96755 [3] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-161-142:62885:63054 [4] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-161-142:62885:63054 [4] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-161-142:62885:63054 [4] NCCL INFO Using network Libfabric
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-161-138:96569:96758 [6] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/408
+ip-26-0-161-138:96569:96758 [6] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/409
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-161-138:96566:96755 [3] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/408
+ip-26-0-161-138:96566:96755 [3] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/408
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-161-103:96888:97066 [0] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-161-103:96888:97066 [0] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-161-103:96888:97066 [0] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-161-103:96888:97066 [0] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-161-103:96888:97066 [0] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-161-103:96888:97066 [0] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-161-103:96888:97066 [0] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-161-103:96888:97066 [0] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-161-103:96888:97066 [0] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-161-103:96888:97066 [0] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-161-103:96888:97066 [0] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-161-103:96888:97066 [0] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-161-103:96888:97066 [0] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-161-103:96888:97066 [0] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-161-103:96888:97066 [0] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-161-103:96888:97066 [0] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-161-103:96888:97066 [0] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-161-103:96888:97066 [0] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-161-103:96888:97066 [0] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-161-103:96888:97066 [0] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-161-103:96888:97066 [0] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-161-103:96888:97066 [0] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-161-103:96888:97066 [0] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-161-103:96888:97066 [0] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-161-103:96888:97066 [0] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-161-103:96888:97066 [0] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-161-103:96888:97066 [0] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-161-103:96888:97066 [0] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-161-103:96888:97066 [0] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-161-103:96888:97066 [0] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-161-103:96888:97066 [0] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-161-103:96888:97066 [0] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/409
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO Using network Libfabric
+ip-26-0-161-103:96894:97068 [6] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-161-103:96894:97068 [6] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-161-103:96894:97068 [6] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-161-103:96894:97068 [6] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-161-103:96894:97068 [6] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-161-103:96894:97068 [6] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-161-103:96894:97068 [6] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-161-103:96894:97068 [6] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-161-103:96894:97068 [6] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-161-103:96894:97068 [6] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-161-103:96894:97068 [6] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-161-103:96894:97068 [6] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-161-103:96894:97068 [6] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-161-103:96894:97068 [6] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-161-103:96894:97068 [6] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-161-103:96894:97068 [6] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-161-103:96894:97068 [6] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-161-103:96894:97068 [6] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-161-103:96894:97068 [6] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-161-103:96894:97068 [6] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-161-103:96894:97068 [6] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-161-103:96894:97068 [6] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-161-103:96894:97068 [6] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-161-103:96894:97068 [6] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-161-103:96894:97068 [6] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-161-103:96894:97068 [6] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-161-103:96894:97068 [6] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-161-103:96894:97068 [6] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-161-103:96894:97068 [6] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-161-103:96894:97068 [6] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-161-103:96894:97068 [6] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-161-103:96894:97068 [6] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-161-103:96888:97066 [0] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/408
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-161-103:96888:97066 [0] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-161-103:96892:97064 [4] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-161-103:96892:97064 [4] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-161-103:96892:97064 [4] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-161-103:96892:97064 [4] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-161-103:96892:97064 [4] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-161-103:96892:97064 [4] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-161-103:96892:97064 [4] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-161-103:96892:97064 [4] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-161-103:96892:97064 [4] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-161-103:96892:97064 [4] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-161-103:96892:97064 [4] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-161-103:96892:97064 [4] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-161-103:96892:97064 [4] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-161-103:96892:97064 [4] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-161-103:96892:97064 [4] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-161-103:96892:97064 [4] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-161-103:96892:97064 [4] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-161-103:96892:97064 [4] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-161-103:96892:97064 [4] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-161-103:96892:97064 [4] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-161-103:96892:97064 [4] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-161-103:96892:97064 [4] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-161-103:96892:97064 [4] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-161-103:96892:97064 [4] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-161-103:96892:97064 [4] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-161-103:96892:97064 [4] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-161-103:96892:97064 [4] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-161-103:96892:97064 [4] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-161-103:96892:97064 [4] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-161-103:96892:97064 [4] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-161-103:96892:97064 [4] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-161-103:96892:97064 [4] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/409
+ip-26-0-161-103:96894:97068 [6] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/408
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-161-103:96894:97068 [6] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-161-103:96892:97064 [4] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/408
+ip-26-0-161-142:62887:63055 [6] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-161-142:62887:63055 [6] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-161-142:62887:63055 [6] NCCL INFO Using network Libfabric
+ip-26-0-161-103:96892:97064 [4] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-161-138:96570:96759 [7] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-161-138:96570:96759 [7] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-161-138:96570:96759 [7] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-161-138:96570:96759 [7] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-161-138:96570:96759 [7] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-161-138:96570:96759 [7] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-161-138:96570:96759 [7] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-161-138:96570:96759 [7] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-161-138:96570:96759 [7] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-161-138:96570:96759 [7] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-161-138:96570:96759 [7] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-161-138:96570:96759 [7] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-161-138:96570:96759 [7] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-161-138:96570:96759 [7] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-161-138:96570:96759 [7] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-161-138:96570:96759 [7] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-161-138:96570:96759 [7] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-161-138:96570:96759 [7] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-161-138:96570:96759 [7] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-161-138:96570:96759 [7] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-161-138:96570:96759 [7] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-161-138:96570:96759 [7] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-161-138:96570:96759 [7] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-161-138:96570:96759 [7] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-161-138:96570:96759 [7] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-161-138:96570:96759 [7] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-161-138:96570:96759 [7] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-161-138:96570:96759 [7] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-168-30:235650:235824 [2] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-168-30:235650:235824 [2] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-168-30:235650:235824 [2] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-161-138:96570:96759 [7] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-161-138:96570:96759 [7] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-161-138:96570:96759 [7] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-161-138:96570:96759 [7] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-168-30:235650:235824 [2] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-168-30:235650:235824 [2] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-168-30:235650:235824 [2] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-168-30:235650:235824 [2] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-168-30:235650:235824 [2] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-168-30:235650:235824 [2] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-168-30:235650:235824 [2] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-168-30:235650:235824 [2] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-168-30:235650:235824 [2] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-168-30:235650:235824 [2] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-168-30:235650:235824 [2] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-168-30:235650:235824 [2] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-168-30:235650:235824 [2] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-168-30:235650:235824 [2] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-168-30:235650:235824 [2] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-168-30:235650:235824 [2] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-168-30:235650:235824 [2] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-168-30:235650:235824 [2] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-168-30:235650:235824 [2] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-168-30:235650:235824 [2] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-168-30:235650:235824 [2] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-168-30:235650:235824 [2] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-168-30:235650:235824 [2] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-168-30:235650:235824 [2] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-161-138:96567:96760 [4] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-161-138:96567:96760 [4] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-161-138:96567:96760 [4] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-161-138:96567:96760 [4] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-168-30:235650:235824 [2] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-168-30:235650:235824 [2] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-168-30:235650:235824 [2] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-168-30:235650:235824 [2] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-168-30:235650:235824 [2] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/409
+ip-26-0-161-138:96567:96760 [4] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-161-138:96567:96760 [4] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-161-138:96567:96760 [4] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-161-138:96567:96760 [4] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-161-138:96567:96760 [4] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-161-138:96567:96760 [4] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-161-138:96567:96760 [4] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-161-138:96567:96760 [4] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-161-138:96567:96760 [4] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-161-138:96567:96760 [4] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-161-138:96567:96760 [4] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-161-138:96567:96760 [4] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-161-138:96567:96760 [4] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-161-138:96567:96760 [4] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-161-138:96567:96760 [4] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-161-138:96567:96760 [4] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-161-138:96567:96760 [4] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-161-138:96567:96760 [4] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-161-138:96567:96760 [4] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-161-138:96567:96760 [4] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-161-138:96567:96760 [4] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-161-138:96567:96760 [4] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-161-138:96567:96760 [4] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-161-138:96567:96760 [4] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-161-138:96567:96760 [4] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-161-138:96567:96760 [4] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-161-138:96567:96760 [4] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-161-138:96567:96760 [4] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-161-138:96563:96756 [0] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-161-138:96563:96756 [0] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-161-138:96563:96756 [0] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-161-138:96563:96756 [0] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-161-138:96563:96756 [0] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-161-138:96563:96756 [0] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-161-138:96563:96756 [0] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-161-138:96563:96756 [0] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-161-138:96563:96756 [0] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-161-138:96563:96756 [0] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-161-138:96563:96756 [0] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-161-138:96563:96756 [0] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-161-138:96563:96756 [0] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-161-138:96563:96756 [0] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-161-138:96563:96756 [0] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-161-138:96563:96756 [0] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-161-138:96563:96756 [0] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-161-138:96563:96756 [0] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-161-138:96563:96756 [0] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-161-138:96563:96756 [0] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-161-138:96563:96756 [0] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-161-138:96563:96756 [0] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-161-138:96563:96756 [0] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-161-138:96563:96756 [0] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-161-138:96563:96756 [0] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-161-138:96563:96756 [0] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-161-138:96563:96756 [0] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-161-138:96563:96756 [0] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-161-138:96563:96756 [0] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-161-138:96563:96756 [0] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-161-138:96563:96756 [0] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-161-138:96563:96756 [0] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-168-30:235650:235824 [2] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/408
+ip-26-0-168-30:235650:235824 [2] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-161-138:96570:96759 [7] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/408
+ip-26-0-161-138:96570:96759 [7] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-161-138:96567:96760 [4] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/408
+ip-26-0-161-138:96567:96760 [4] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-161-78:114514:114689 [2] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-161-78:114514:114689 [2] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-161-78:114514:114689 [2] NCCL INFO Using network Libfabric
+ip-26-0-161-138:96563:96756 [0] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/408
+ip-26-0-161-138:96563:96756 [0] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-161-138:96568:96753 [5] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-161-138:96568:96753 [5] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-161-138:96568:96753 [5] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-161-138:96568:96753 [5] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-161-138:96568:96753 [5] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-161-138:96568:96753 [5] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-161-138:96568:96753 [5] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-161-138:96568:96753 [5] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-161-138:96568:96753 [5] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-161-138:96568:96753 [5] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-161-138:96568:96753 [5] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-161-138:96568:96753 [5] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-161-138:96568:96753 [5] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-161-138:96568:96753 [5] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-161-138:96568:96753 [5] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-161-138:96568:96753 [5] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-161-138:96568:96753 [5] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-161-138:96568:96753 [5] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-161-138:96568:96753 [5] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-161-138:96568:96753 [5] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-161-138:96568:96753 [5] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-161-138:96568:96753 [5] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-161-138:96568:96753 [5] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-161-138:96568:96753 [5] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-161-138:96568:96753 [5] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-161-138:96568:96753 [5] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-161-138:96568:96753 [5] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-161-138:96568:96753 [5] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-161-138:96568:96753 [5] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-161-138:96568:96753 [5] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-161-138:96568:96753 [5] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-161-138:96568:96753 [5] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-161-138:96564:96757 [1] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-161-138:96564:96757 [1] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-161-138:96564:96757 [1] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-161-138:96564:96757 [1] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-161-138:96564:96757 [1] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-161-138:96564:96757 [1] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-161-138:96564:96757 [1] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-161-138:96564:96757 [1] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-161-138:96564:96757 [1] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-161-138:96564:96757 [1] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-161-138:96564:96757 [1] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-161-138:96564:96757 [1] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-161-138:96564:96757 [1] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-161-138:96564:96757 [1] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-161-138:96564:96757 [1] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-161-138:96564:96757 [1] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-161-138:96564:96757 [1] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-161-138:96564:96757 [1] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-161-138:96564:96757 [1] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-161-138:96564:96757 [1] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-161-138:96564:96757 [1] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-161-138:96564:96757 [1] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-161-138:96564:96757 [1] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-161-138:96564:96757 [1] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-161-138:96564:96757 [1] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-161-138:96564:96757 [1] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-161-138:96564:96757 [1] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-161-138:96564:96757 [1] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-161-138:96564:96757 [1] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-161-138:96564:96757 [1] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-161-138:96564:96757 [1] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-161-138:96564:96757 [1] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-161-138:96568:96753 [5] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/408
+ip-26-0-161-138:96568:96753 [5] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-161-138:96564:96757 [1] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/408
+ip-26-0-161-138:96564:96757 [1] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-161-123:32705:32880 [1] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-161-123:32705:32880 [1] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-161-123:32705:32880 [1] NCCL INFO Using network Libfabric
+ip-26-0-168-30:235652:235822 [4] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-168-30:235652:235822 [4] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-168-30:235652:235822 [4] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-168-30:235652:235822 [4] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-168-30:235652:235822 [4] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-168-30:235652:235822 [4] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-168-30:235652:235822 [4] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-168-30:235652:235822 [4] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-168-30:235652:235822 [4] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-168-30:235652:235822 [4] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-168-30:235652:235822 [4] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-168-30:235652:235822 [4] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-168-30:235652:235822 [4] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-168-30:235652:235822 [4] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-168-30:235652:235822 [4] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-168-30:235652:235822 [4] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-168-30:235652:235822 [4] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-168-30:235652:235822 [4] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-168-30:235652:235822 [4] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-168-30:235652:235822 [4] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-168-30:235652:235822 [4] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-168-30:235652:235822 [4] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-168-30:235652:235822 [4] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-168-30:235652:235822 [4] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-168-30:235652:235822 [4] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-168-30:235652:235822 [4] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-168-30:235652:235822 [4] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-168-30:235652:235822 [4] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-168-30:235652:235822 [4] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-168-30:235652:235822 [4] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-168-30:235652:235822 [4] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-168-30:235652:235822 [4] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-168-30:235652:235822 [4] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/408
+ip-26-0-168-30:235652:235822 [4] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO Using network Libfabric
+ip-26-0-168-30:235654:235821 [6] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-168-30:235654:235821 [6] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-168-30:235654:235821 [6] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-168-30:235654:235821 [6] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-168-30:235654:235821 [6] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-168-30:235654:235821 [6] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-168-30:235654:235821 [6] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-168-30:235654:235821 [6] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-168-30:235654:235821 [6] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-168-30:235654:235821 [6] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-168-30:235654:235821 [6] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-168-30:235654:235821 [6] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-168-30:235654:235821 [6] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-168-30:235654:235821 [6] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-168-30:235654:235821 [6] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-168-30:235654:235821 [6] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-168-30:235654:235821 [6] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-168-30:235654:235821 [6] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-168-30:235654:235821 [6] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-168-30:235654:235821 [6] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-168-30:235654:235821 [6] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-168-30:235654:235821 [6] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-168-30:235654:235821 [6] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-168-30:235654:235821 [6] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-168-30:235654:235821 [6] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-168-30:235654:235821 [6] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-168-30:235654:235821 [6] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-168-30:235654:235821 [6] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-168-30:235654:235821 [6] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-168-30:235654:235821 [6] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-168-30:235654:235821 [6] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-168-30:235654:235821 [6] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-168-30:235654:235821 [6] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/408
+ip-26-0-168-30:235654:235821 [6] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-168-30:235649:235827 [1] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-168-30:235649:235827 [1] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-168-30:235649:235827 [1] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-168-30:235649:235827 [1] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-168-30:235649:235827 [1] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-168-30:235649:235827 [1] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-168-30:235655:235828 [7] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-168-30:235649:235827 [1] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-168-30:235649:235827 [1] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-168-30:235655:235828 [7] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-168-30:235649:235827 [1] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-168-30:235655:235828 [7] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-168-30:235649:235827 [1] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-168-30:235655:235828 [7] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-168-30:235649:235827 [1] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-168-30:235655:235828 [7] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-168-30:235649:235827 [1] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-168-30:235655:235828 [7] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-168-30:235649:235827 [1] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-168-30:235655:235828 [7] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-168-30:235649:235827 [1] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-168-30:235655:235828 [7] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-168-30:235649:235827 [1] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-168-30:235655:235828 [7] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-168-30:235649:235827 [1] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-168-30:235655:235828 [7] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-168-30:235649:235827 [1] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-168-30:235655:235828 [7] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-168-30:235649:235827 [1] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-168-30:235655:235828 [7] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-168-30:235649:235827 [1] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-168-30:235655:235828 [7] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-168-30:235649:235827 [1] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-168-30:235655:235828 [7] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-168-30:235649:235827 [1] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-168-30:235655:235828 [7] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-168-30:235649:235827 [1] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-168-30:235655:235828 [7] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-168-30:235649:235827 [1] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-168-30:235655:235828 [7] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-168-30:235649:235827 [1] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-168-30:235655:235828 [7] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-168-30:235649:235827 [1] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-168-30:235655:235828 [7] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-168-30:235649:235827 [1] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-168-30:235655:235828 [7] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-168-30:235649:235827 [1] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-168-30:235655:235828 [7] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-168-30:235649:235827 [1] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-168-30:235655:235828 [7] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-168-30:235649:235827 [1] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-168-30:235655:235828 [7] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-168-30:235649:235827 [1] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-168-30:235655:235828 [7] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-168-30:235655:235828 [7] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-168-30:235655:235828 [7] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-168-30:235655:235828 [7] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-168-30:235649:235827 [1] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-168-30:235655:235828 [7] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-168-30:235649:235827 [1] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-168-30:235655:235828 [7] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-168-30:235655:235828 [7] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-168-30:235655:235828 [7] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO Using network Libfabric
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO Using network Libfabric
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO Using network Libfabric
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO Using network Libfabric
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO Using network Libfabric
+ip-26-0-168-52:2722046:2722223 [1] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-168-52:2722046:2722223 [1] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-168-52:2722046:2722223 [1] NCCL INFO Using network Libfabric
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO Using network Libfabric
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO Using network Libfabric
+ip-26-0-161-103:96895:97069 [7] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-161-103:96895:97069 [7] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-161-103:96895:97069 [7] NCCL INFO Using network Libfabric
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO Using network Libfabric
+ip-26-0-168-30:235655:235828 [7] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-168-30:235653:235825 [5] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-168-30:235653:235825 [5] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-168-30:235653:235825 [5] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-168-30:235653:235825 [5] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-168-30:235653:235825 [5] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-168-30:235653:235825 [5] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-168-30:235653:235825 [5] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-168-30:235653:235825 [5] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-168-30:235653:235825 [5] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-168-30:235653:235825 [5] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-168-30:235653:235825 [5] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-168-30:235653:235825 [5] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-168-30:235653:235825 [5] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-168-30:235653:235825 [5] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-168-30:235653:235825 [5] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-168-30:235653:235825 [5] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-168-30:235653:235825 [5] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-168-30:235653:235825 [5] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-168-30:235653:235825 [5] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-168-30:235653:235825 [5] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-168-30:235653:235825 [5] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-168-30:235653:235825 [5] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-168-30:235653:235825 [5] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-168-30:235653:235825 [5] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-168-30:235653:235825 [5] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-168-30:235653:235825 [5] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-168-30:235653:235825 [5] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-168-30:235653:235825 [5] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-168-30:235653:235825 [5] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-168-30:235653:235825 [5] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-168-30:235653:235825 [5] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-168-30:235653:235825 [5] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-168-30:235648:235826 [0] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-168-30:235648:235826 [0] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-168-30:235648:235826 [0] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-168-30:235648:235826 [0] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-168-30:235648:235826 [0] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-168-30:235648:235826 [0] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-168-30:235648:235826 [0] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-168-30:235648:235826 [0] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-168-30:235648:235826 [0] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-168-30:235648:235826 [0] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-168-30:235648:235826 [0] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-168-30:235648:235826 [0] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-168-30:235648:235826 [0] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-168-30:235648:235826 [0] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-168-30:235648:235826 [0] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-168-30:235648:235826 [0] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-168-30:235648:235826 [0] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-168-30:235648:235826 [0] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-168-30:235648:235826 [0] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-168-30:235648:235826 [0] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-168-30:235648:235826 [0] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-168-30:235648:235826 [0] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-168-30:235648:235826 [0] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-168-30:235648:235826 [0] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-168-30:235648:235826 [0] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-168-30:235648:235826 [0] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-168-30:235648:235826 [0] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-168-30:235648:235826 [0] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-168-30:235648:235826 [0] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-168-30:235648:235826 [0] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-168-30:235648:235826 [0] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-168-30:235648:235826 [0] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-168-30:235655:235828 [7] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/408
+ip-26-0-168-30:235649:235827 [1] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/408
+ip-26-0-168-30:235653:235825 [5] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/408
+ip-26-0-168-30:235655:235828 [7] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-168-30:235649:235827 [1] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-168-30:235653:235825 [5] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-168-30:235648:235826 [0] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/408
+ip-26-0-168-30:235648:235826 [0] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-168-34:272154:272330 [4] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-168-34:272154:272330 [4] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-168-34:272154:272330 [4] NCCL INFO Using network Libfabric
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO Using network Libfabric
+ip-26-0-161-78:114516:114691 [4] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-161-78:114516:114691 [4] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-161-78:114516:114691 [4] NCCL INFO Using network Libfabric
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO Using network Libfabric
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO Using network Libfabric
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO Using network Libfabric
+ip-26-0-168-30:235651:235823 [3] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-168-30:235651:235823 [3] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-168-30:235651:235823 [3] NCCL INFO Using network Libfabric
+ip-26-0-161-138:96565:96754 [2] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-161-138:96565:96754 [2] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-161-138:96565:96754 [2] NCCL INFO Using network Libfabric
+ip-26-0-168-52:2722052:2722226 [7] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-168-52:2722052:2722226 [7] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-168-52:2722052:2722226 [7] NCCL INFO Using network Libfabric
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO Using network Libfabric
+ip-26-0-165-202:153564:153737 [2] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-165-202:153564:153737 [2] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-165-202:153564:153737 [2] NCCL INFO Using network Libfabric
+ip-26-0-165-202:153566:153741 [4] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-165-202:153566:153741 [4] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-165-202:153566:153741 [4] NCCL INFO Using network Libfabric
+ip-26-0-165-202:153567:153742 [5] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-165-202:153567:153742 [5] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-165-202:153567:153742 [5] NCCL INFO Using network Libfabric
+ip-26-0-165-202:153562:153735 [0] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-165-202:153562:153735 [0] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-165-202:153562:153735 [0] NCCL INFO Using network Libfabric
+ip-26-0-165-202:153565:153738 [3] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-165-202:153565:153738 [3] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-165-202:153565:153738 [3] NCCL INFO Using network Libfabric
+ip-26-0-161-103:96889:97065 [1] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-161-103:96889:97065 [1] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-161-103:96889:97065 [1] NCCL INFO Using network Libfabric
+ip-26-0-165-202:153569:153739 [7] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-165-202:153569:153739 [7] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-165-202:153569:153739 [7] NCCL INFO Using network Libfabric
+ip-26-0-165-202:153568:153736 [6] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-165-202:153568:153736 [6] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-165-202:153568:153736 [6] NCCL INFO Using network Libfabric
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO Using network Libfabric
+ip-26-0-168-30:235650:235824 [2] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-168-30:235650:235824 [2] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-168-30:235650:235824 [2] NCCL INFO Using network Libfabric
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO Using network Libfabric
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO Using network Libfabric
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO Using network Libfabric
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO Using network Libfabric
+ip-26-0-161-138:96569:96758 [6] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-161-138:96569:96758 [6] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-161-138:96569:96758 [6] NCCL INFO Using network Libfabric
+ip-26-0-161-138:96566:96755 [3] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-161-138:96566:96755 [3] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-161-138:96566:96755 [3] NCCL INFO Using network Libfabric
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO Using network Libfabric
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO Using network Libfabric
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO Using network Libfabric
+ip-26-0-161-103:96891:97063 [3] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-161-103:96891:97063 [3] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-161-103:96891:97063 [3] NCCL INFO Using network Libfabric
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO Using network Libfabric
+ip-26-0-161-78:114515:114690 [3] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-161-78:114515:114690 [3] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-161-78:114515:114690 [3] NCCL INFO Using network Libfabric
+ip-26-0-161-78:114518:114693 [6] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-161-78:114518:114693 [6] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-161-78:114518:114693 [6] NCCL INFO Using network Libfabric
+ip-26-0-161-103:96893:97067 [5] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-161-103:96893:97067 [5] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-161-103:96893:97067 [5] NCCL INFO Using network Libfabric
+ip-26-0-161-138:96567:96760 [4] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-161-138:96567:96760 [4] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-161-138:96567:96760 [4] NCCL INFO Using network Libfabric
+ip-26-0-168-34:272153:272328 [3] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-168-34:272153:272328 [3] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-168-34:272153:272328 [3] NCCL INFO Using network Libfabric
+ip-26-0-161-138:96563:96756 [0] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-161-138:96563:96756 [0] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-161-138:96563:96756 [0] NCCL INFO Using network Libfabric
+ip-26-0-161-138:96570:96759 [7] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-161-138:96570:96759 [7] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-161-138:96570:96759 [7] NCCL INFO Using network Libfabric
+ip-26-0-161-138:96568:96753 [5] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-161-138:96568:96753 [5] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-161-138:96568:96753 [5] NCCL INFO Using network Libfabric
+ip-26-0-161-138:96564:96757 [1] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-161-138:96564:96757 [1] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-161-138:96564:96757 [1] NCCL INFO Using network Libfabric
+ip-26-0-161-142:62882:63056 [1] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-161-142:62882:63056 [1] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-161-142:62882:63056 [1] NCCL INFO Using network Libfabric
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO Using network Libfabric
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO Using network Libfabric
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO Using network Libfabric
+ip-26-0-168-52:2722047:2722221 [2] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-168-52:2722047:2722221 [2] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-168-52:2722047:2722221 [2] NCCL INFO Using network Libfabric
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO Using network Libfabric
+ip-26-0-168-52:2722050:2722225 [5] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-168-52:2722050:2722225 [5] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-168-52:2722050:2722225 [5] NCCL INFO Using network Libfabric
+ip-26-0-160-225:124165:124341 [5] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-160-225:124165:124341 [5] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-160-225:124165:124341 [5] NCCL INFO Using network Libfabric
+ip-26-0-160-225:124166:124339 [6] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-160-225:124166:124339 [6] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-160-225:124166:124339 [6] NCCL INFO Using network Libfabric
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO Using network Libfabric
+ip-26-0-160-225:124163:124340 [3] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-160-225:124163:124340 [3] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-160-225:124163:124340 [3] NCCL INFO Using network Libfabric
+ip-26-0-161-123:32707:32878 [3] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-161-123:32707:32878 [3] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-161-123:32707:32878 [3] NCCL INFO Using network Libfabric
+ip-26-0-160-225:124162:124342 [2] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-160-225:124162:124342 [2] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-160-225:124162:124342 [2] NCCL INFO Using network Libfabric
+ip-26-0-160-225:124164:124336 [4] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-160-225:124164:124336 [4] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-160-225:124164:124336 [4] NCCL INFO Using network Libfabric
+ip-26-0-160-225:124167:124338 [7] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-160-225:124167:124338 [7] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-160-225:124167:124338 [7] NCCL INFO Using network Libfabric
+ip-26-0-161-123:32704:32881 [0] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-161-123:32704:32881 [0] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-161-123:32704:32881 [0] NCCL INFO Using network Libfabric
+ip-26-0-161-123:32711:32884 [7] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-161-123:32711:32884 [7] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-161-123:32711:32884 [7] NCCL INFO Using network Libfabric
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO Using network Libfabric
+ip-26-0-161-123:32708:32879 [4] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-161-123:32708:32879 [4] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-161-123:32708:32879 [4] NCCL INFO Using network Libfabric
+ip-26-0-161-123:32710:32883 [6] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-161-123:32710:32883 [6] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-161-123:32710:32883 [6] NCCL INFO Using network Libfabric
+ip-26-0-161-123:32709:32882 [5] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-161-123:32709:32882 [5] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-161-123:32709:32882 [5] NCCL INFO Using network Libfabric
+ip-26-0-168-30:235652:235822 [4] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-168-30:235652:235822 [4] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-168-30:235652:235822 [4] NCCL INFO Using network Libfabric
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO Using network Libfabric
+ip-26-0-168-52:2722051:2722224 [6] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-168-52:2722051:2722224 [6] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-168-52:2722051:2722224 [6] NCCL INFO Using network Libfabric
+ip-26-0-161-78:114513:114688 [1] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-161-78:114513:114688 [1] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-161-78:114513:114688 [1] NCCL INFO Using network Libfabric
+ip-26-0-161-142:62884:63061 [3] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-161-142:62884:63061 [3] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-161-142:62884:63061 [3] NCCL INFO Using network Libfabric
+ip-26-0-168-34:272157:272324 [7] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-168-34:272157:272324 [7] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-168-34:272157:272324 [7] NCCL INFO Using network Libfabric
+ip-26-0-168-30:235654:235821 [6] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-168-30:235654:235821 [6] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-168-30:235654:235821 [6] NCCL INFO Using network Libfabric
+ip-26-0-168-34:272151:272326 [1] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-168-34:272151:272326 [1] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-168-34:272151:272326 [1] NCCL INFO Using network Libfabric
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO Using network Libfabric
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO Using network Libfabric
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO Using network Libfabric
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO Using network Libfabric
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO Using network Libfabric
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO Using network Libfabric
+ip-26-0-168-34:272156:272325 [6] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-168-34:272156:272325 [6] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-168-34:272156:272325 [6] NCCL INFO Using network Libfabric
+ip-26-0-168-34:272150:272327 [0] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-168-34:272150:272327 [0] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-168-34:272150:272327 [0] NCCL INFO Using network Libfabric
+ip-26-0-168-34:272155:272329 [5] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-168-34:272155:272329 [5] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-168-34:272155:272329 [5] NCCL INFO Using network Libfabric
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO Using network Libfabric
+ip-26-0-161-78:114519:114692 [7] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-161-78:114519:114692 [7] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-161-78:114519:114692 [7] NCCL INFO Using network Libfabric
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO Using network Libfabric
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO Using network Libfabric
+ip-26-0-161-142:62883:63058 [2] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-161-142:62883:63058 [2] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-161-142:62883:63058 [2] NCCL INFO Using network Libfabric
+ip-26-0-161-142:62888:63060 [7] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-161-142:62888:63060 [7] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-161-142:62888:63060 [7] NCCL INFO Using network Libfabric
+ip-26-0-161-103:96890:97070 [2] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-161-103:96890:97070 [2] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-161-103:96890:97070 [2] NCCL INFO Using network Libfabric
+ip-26-0-168-30:235653:235825 [5] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-168-30:235653:235825 [5] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-168-30:235653:235825 [5] NCCL INFO Using network Libfabric
+ip-26-0-161-142:62881:63059 [0] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-161-142:62881:63059 [0] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-161-142:62881:63059 [0] NCCL INFO Using network Libfabric
+ip-26-0-161-142:62886:63057 [5] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-161-142:62886:63057 [5] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-161-142:62886:63057 [5] NCCL INFO Using network Libfabric
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO Using network Libfabric
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO Using network Libfabric
+ip-26-0-168-52:2722045:2722219 [0] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-168-52:2722045:2722219 [0] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-168-52:2722045:2722219 [0] NCCL INFO Using network Libfabric
+ip-26-0-168-30:235648:235826 [0] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-168-30:235648:235826 [0] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-168-30:235648:235826 [0] NCCL INFO Using network Libfabric
+ip-26-0-168-52:2722048:2722220 [3] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-168-52:2722048:2722220 [3] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-168-52:2722048:2722220 [3] NCCL INFO Using network Libfabric
+ip-26-0-168-30:235655:235828 [7] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-168-30:235655:235828 [7] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-168-30:235655:235828 [7] NCCL INFO Using network Libfabric
+ip-26-0-168-30:235649:235827 [1] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-168-30:235649:235827 [1] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-168-30:235649:235827 [1] NCCL INFO Using network Libfabric
+ip-26-0-161-103:96888:97066 [0] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-161-103:96888:97066 [0] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-161-103:96888:97066 [0] NCCL INFO Using network Libfabric
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO Using network Libfabric
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO Using network Libfabric
+ip-26-0-161-103:96894:97068 [6] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-161-103:96894:97068 [6] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-161-103:96894:97068 [6] NCCL INFO Using network Libfabric
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO Using network Libfabric
+ip-26-0-161-103:96892:97064 [4] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-161-103:96892:97064 [4] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-161-103:96892:97064 [4] NCCL INFO Using network Libfabric
+ip-26-0-161-78:114512:114686 [0] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-161-78:114512:114686 [0] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-161-78:114512:114686 [0] NCCL INFO Using network Libfabric
+ip-26-0-161-142:62888:63060 [7] NCCL INFO comm 0x91600d0 rank 47 nranks 128 cudaDev 7 nvmlDev 7 busId ca000 commId 0xe67d0548a5bfb9af - Init START
+ip-26-0-161-138:96570:96759 [7] NCCL INFO comm 0xa0c08e0 rank 39 nranks 128 cudaDev 7 nvmlDev 7 busId ca000 commId 0xe67d0548a5bfb9af - Init START
+ip-26-0-161-138:96568:96753 [5] NCCL INFO comm 0x88d80e0 rank 37 nranks 128 cudaDev 5 nvmlDev 5 busId a8000 commId 0xe67d0548a5bfb9af - Init START
+ip-26-0-161-138:96569:96758 [6] NCCL INFO comm 0x8e85570 rank 38 nranks 128 cudaDev 6 nvmlDev 6 busId b9000 commId 0xe67d0548a5bfb9af - Init START
+ip-26-0-161-142:62887:63055 [6] NCCL INFO comm 0x867a2a0 rank 46 nranks 128 cudaDev 6 nvmlDev 6 busId b9000 commId 0xe67d0548a5bfb9af - Init START
+ip-26-0-161-142:62886:63057 [5] NCCL INFO comm 0x9de8db0 rank 45 nranks 128 cudaDev 5 nvmlDev 5 busId a8000 commId 0xe67d0548a5bfb9af - Init START
+ip-26-0-161-142:62885:63054 [4] NCCL INFO comm 0x9e18d20 rank 44 nranks 128 cudaDev 4 nvmlDev 4 busId 97000 commId 0xe67d0548a5bfb9af - Init START
+ip-26-0-161-142:62883:63058 [2] NCCL INFO comm 0x8b38920 rank 42 nranks 128 cudaDev 2 nvmlDev 2 busId 75000 commId 0xe67d0548a5bfb9af - Init START
+ip-26-0-161-142:62884:63061 [3] NCCL INFO comm 0x95b82f0 rank 43 nranks 128 cudaDev 3 nvmlDev 3 busId 86000 commId 0xe67d0548a5bfb9af - Init START
+ip-26-0-161-142:62881:63059 [0] NCCL INFO comm 0xa10dc70 rank 40 nranks 128 cudaDev 0 nvmlDev 0 busId 53000 commId 0xe67d0548a5bfb9af - Init START
+ip-26-0-168-34:272157:272324 [7] NCCL INFO comm 0x93fc8f0 rank 119 nranks 128 cudaDev 7 nvmlDev 7 busId ca000 commId 0xe67d0548a5bfb9af - Init START
+ip-26-0-161-142:62882:63056 [1] NCCL INFO comm 0x8ed2140 rank 41 nranks 128 cudaDev 1 nvmlDev 1 busId 64000 commId 0xe67d0548a5bfb9af - Init START
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO comm 0x88a2150 rank 51 nranks 128 cudaDev 3 nvmlDev 3 busId 86000 commId 0xe67d0548a5bfb9af - Init START
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO comm 0x994f620 rank 50 nranks 128 cudaDev 2 nvmlDev 2 busId 75000 commId 0xe67d0548a5bfb9af - Init START
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO comm 0x8f46e20 rank 48 nranks 128 cudaDev 0 nvmlDev 0 busId 53000 commId 0xe67d0548a5bfb9af - Init START
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO comm 0xa1c3e40 rank 49 nranks 128 cudaDev 1 nvmlDev 1 busId 64000 commId 0xe67d0548a5bfb9af - Init START
+ip-26-0-168-52:2722045:2722219 [0] NCCL INFO comm 0x852ebd0 rank 120 nranks 128 cudaDev 0 nvmlDev 0 busId 53000 commId 0xe67d0548a5bfb9af - Init START
+ip-26-0-168-34:272156:272325 [6] NCCL INFO comm 0xa194790 rank 118 nranks 128 cudaDev 6 nvmlDev 6 busId b9000 commId 0xe67d0548a5bfb9af - Init START
+ip-26-0-168-34:272155:272329 [5] NCCL INFO comm 0x91a3e50 rank 117 nranks 128 cudaDev 5 nvmlDev 5 busId a8000 commId 0xe67d0548a5bfb9af - Init START
+ip-26-0-161-138:96567:96760 [4] NCCL INFO comm 0x886b530 rank 36 nranks 128 cudaDev 4 nvmlDev 4 busId 97000 commId 0xe67d0548a5bfb9af - Init START
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO comm 0x84bbf10 rank 86 nranks 128 cudaDev 6 nvmlDev 6 busId b9000 commId 0xe67d0548a5bfb9af - Init START
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO comm 0x91fcaa0 rank 84 nranks 128 cudaDev 4 nvmlDev 4 busId 97000 commId 0xe67d0548a5bfb9af - Init START
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO comm 0x8c25390 rank 82 nranks 128 cudaDev 2 nvmlDev 2 busId 75000 commId 0xe67d0548a5bfb9af - Init START
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO comm 0x9298e10 rank 87 nranks 128 cudaDev 7 nvmlDev 7 busId ca000 commId 0xe67d0548a5bfb9af - Init START
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO comm 0x9b878e0 rank 85 nranks 128 cudaDev 5 nvmlDev 5 busId a8000 commId 0xe67d0548a5bfb9af - Init START
+ip-26-0-168-34:272154:272330 [4] NCCL INFO comm 0x9279c80 rank 116 nranks 128 cudaDev 4 nvmlDev 4 busId 97000 commId 0xe67d0548a5bfb9af - Init START
+ip-26-0-168-34:272153:272328 [3] NCCL INFO comm 0x85b56b0 rank 115 nranks 128 cudaDev 3 nvmlDev 3 busId 86000 commId 0xe67d0548a5bfb9af - Init START
+ip-26-0-168-34:272151:272326 [1] NCCL INFO comm 0x9494aa0 rank 113 nranks 128 cudaDev 1 nvmlDev 1 busId 64000 commId 0xe67d0548a5bfb9af - Init START
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO comm 0x84a7790 rank 101 nranks 128 cudaDev 5 nvmlDev 5 busId a8000 commId 0xe67d0548a5bfb9af - Init START
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO comm 0x9ef1450 rank 99 nranks 128 cudaDev 3 nvmlDev 3 busId 86000 commId 0xe67d0548a5bfb9af - Init START
+ip-26-0-168-30:235655:235828 [7] NCCL INFO comm 0x9447f80 rank 111 nranks 128 cudaDev 7 nvmlDev 7 busId ca000 commId 0xe67d0548a5bfb9af - Init START
+ip-26-0-168-30:235654:235821 [6] NCCL INFO comm 0x994c420 rank 110 nranks 128 cudaDev 6 nvmlDev 6 busId b9000 commId 0xe67d0548a5bfb9af - Init START
+ip-26-0-168-30:235653:235825 [5] NCCL INFO comm 0x91002f0 rank 109 nranks 128 cudaDev 5 nvmlDev 5 busId a8000 commId 0xe67d0548a5bfb9af - Init START
+ip-26-0-168-34:272152:272331 [2] NCCL INFO comm 0xa466440 rank 114 nranks 128 cudaDev 2 nvmlDev 2 busId 75000 commId 0xe67d0548a5bfb9af - Init START
+ip-26-0-168-34:272150:272327 [0] NCCL INFO comm 0x9d62500 rank 112 nranks 128 cudaDev 0 nvmlDev 0 busId 53000 commId 0xe67d0548a5bfb9af - Init START
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO comm 0x90c0bf0 rank 63 nranks 128 cudaDev 7 nvmlDev 7 busId ca000 commId 0xe67d0548a5bfb9af - Init START
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO comm 0x9a50ce0 rank 62 nranks 128 cudaDev 6 nvmlDev 6 busId b9000 commId 0xe67d0548a5bfb9af - Init START
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO comm 0x8629c70 rank 60 nranks 128 cudaDev 4 nvmlDev 4 busId 97000 commId 0xe67d0548a5bfb9af - Init START
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO comm 0xa08a840 rank 61 nranks 128 cudaDev 5 nvmlDev 5 busId a8000 commId 0xe67d0548a5bfb9af - Init START
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO comm 0x8e7c100 rank 83 nranks 128 cudaDev 3 nvmlDev 3 busId 86000 commId 0xe67d0548a5bfb9af - Init START
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO comm 0x9061f10 rank 81 nranks 128 cudaDev 1 nvmlDev 1 busId 64000 commId 0xe67d0548a5bfb9af - Init START
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO comm 0xa46efe0 rank 78 nranks 128 cudaDev 6 nvmlDev 6 busId b9000 commId 0xe67d0548a5bfb9af - Init START
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO comm 0x86bcdf0 rank 79 nranks 128 cudaDev 7 nvmlDev 7 busId ca000 commId 0xe67d0548a5bfb9af - Init START
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO comm 0x9e94a10 rank 77 nranks 128 cudaDev 5 nvmlDev 5 busId a8000 commId 0xe67d0548a5bfb9af - Init START
+ip-26-0-165-202:153565:153738 [3] NCCL INFO comm 0x9998430 rank 91 nranks 128 cudaDev 3 nvmlDev 3 busId 86000 commId 0xe67d0548a5bfb9af - Init START
+ip-26-0-165-202:153563:153740 [1] NCCL INFO comm 0xa3c8c40 rank 89 nranks 128 cudaDev 1 nvmlDev 1 busId 64000 commId 0xe67d0548a5bfb9af - Init START
+ip-26-0-165-202:153562:153735 [0] NCCL INFO comm 0x948d480 rank 88 nranks 128 cudaDev 0 nvmlDev 0 busId 53000 commId 0xe67d0548a5bfb9af - Init START
+ip-26-0-165-202:153564:153737 [2] NCCL INFO comm 0x87bf990 rank 90 nranks 128 cudaDev 2 nvmlDev 2 busId 75000 commId 0xe67d0548a5bfb9af - Init START
+ip-26-0-165-202:153566:153741 [4] NCCL INFO comm 0x8a05910 rank 92 nranks 128 cudaDev 4 nvmlDev 4 busId 97000 commId 0xe67d0548a5bfb9af - Init START
+ip-26-0-161-78:114519:114692 [7] NCCL INFO comm 0x84be170 rank 71 nranks 128 cudaDev 7 nvmlDev 7 busId ca000 commId 0xe67d0548a5bfb9af - Init START
+ip-26-0-161-78:114517:114687 [5] NCCL INFO comm 0xa282530 rank 69 nranks 128 cudaDev 5 nvmlDev 5 busId a8000 commId 0xe67d0548a5bfb9af - Init START
+ip-26-0-161-78:114515:114690 [3] NCCL INFO comm 0x8dcd620 rank 67 nranks 128 cudaDev 3 nvmlDev 3 busId 86000 commId 0xe67d0548a5bfb9af - Init START
+ip-26-0-161-78:114514:114689 [2] NCCL INFO comm 0x9cf4bf0 rank 66 nranks 128 cudaDev 2 nvmlDev 2 busId 75000 commId 0xe67d0548a5bfb9af - Init START
+ip-26-0-161-78:114518:114693 [6] NCCL INFO comm 0x9878640 rank 70 nranks 128 cudaDev 6 nvmlDev 6 busId b9000 commId 0xe67d0548a5bfb9af - Init START
+ip-26-0-161-78:114512:114686 [0] NCCL INFO comm 0x9281680 rank 64 nranks 128 cudaDev 0 nvmlDev 0 busId 53000 commId 0xe67d0548a5bfb9af - Init START
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO comm 0x8b40680 rank 98 nranks 128 cudaDev 2 nvmlDev 2 busId 75000 commId 0xe67d0548a5bfb9af - Init START
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO comm 0x9406880 rank 102 nranks 128 cudaDev 6 nvmlDev 6 busId b9000 commId 0xe67d0548a5bfb9af - Init START
+ip-26-0-168-30:235649:235827 [1] NCCL INFO comm 0x9bd9090 rank 105 nranks 128 cudaDev 1 nvmlDev 1 busId 64000 commId 0xe67d0548a5bfb9af - Init START
+ip-26-0-168-30:235652:235822 [4] NCCL INFO comm 0x9c80d90 rank 108 nranks 128 cudaDev 4 nvmlDev 4 busId 97000 commId 0xe67d0548a5bfb9af - Init START
+ip-26-0-168-30:235650:235824 [2] NCCL INFO comm 0x8ef34a0 rank 106 nranks 128 cudaDev 2 nvmlDev 2 busId 75000 commId 0xe67d0548a5bfb9af - Init START
+ip-26-0-168-30:235651:235823 [3] NCCL INFO comm 0x8ed2f20 rank 107 nranks 128 cudaDev 3 nvmlDev 3 busId 86000 commId 0xe67d0548a5bfb9af - Init START
+ip-26-0-168-30:235648:235826 [0] NCCL INFO comm 0x890ea90 rank 104 nranks 128 cudaDev 0 nvmlDev 0 busId 53000 commId 0xe67d0548a5bfb9af - Init START
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO comm 0x8bcfca0 rank 58 nranks 128 cudaDev 2 nvmlDev 2 busId 75000 commId 0xe67d0548a5bfb9af - Init START
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO comm 0x8793c10 rank 57 nranks 128 cudaDev 1 nvmlDev 1 busId 64000 commId 0xe67d0548a5bfb9af - Init START
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO comm 0x94a1240 rank 59 nranks 128 cudaDev 3 nvmlDev 3 busId 86000 commId 0xe67d0548a5bfb9af - Init START
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO comm 0x8798710 rank 80 nranks 128 cudaDev 0 nvmlDev 0 busId 53000 commId 0xe67d0548a5bfb9af - Init START
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO comm 0x91b2fc0 rank 76 nranks 128 cudaDev 4 nvmlDev 4 busId 97000 commId 0xe67d0548a5bfb9af - Init START
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO comm 0x84c9ed0 rank 75 nranks 128 cudaDev 3 nvmlDev 3 busId 86000 commId 0xe67d0548a5bfb9af - Init START
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO comm 0x954a920 rank 74 nranks 128 cudaDev 2 nvmlDev 2 busId 75000 commId 0xe67d0548a5bfb9af - Init START
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO comm 0x9b32d60 rank 73 nranks 128 cudaDev 1 nvmlDev 1 busId 64000 commId 0xe67d0548a5bfb9af - Init START
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO comm 0x929efe0 rank 52 nranks 128 cudaDev 4 nvmlDev 4 busId 97000 commId 0xe67d0548a5bfb9af - Init START
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO comm 0x871cb90 rank 53 nranks 128 cudaDev 5 nvmlDev 5 busId a8000 commId 0xe67d0548a5bfb9af - Init START
+ip-26-0-165-202:153568:153736 [6] NCCL INFO comm 0x87ae600 rank 94 nranks 128 cudaDev 6 nvmlDev 6 busId b9000 commId 0xe67d0548a5bfb9af - Init START
+ip-26-0-165-202:153569:153739 [7] NCCL INFO comm 0x87276c0 rank 95 nranks 128 cudaDev 7 nvmlDev 7 busId ca000 commId 0xe67d0548a5bfb9af - Init START
+ip-26-0-161-78:114513:114688 [1] NCCL INFO comm 0x9e36040 rank 65 nranks 128 cudaDev 1 nvmlDev 1 busId 64000 commId 0xe67d0548a5bfb9af - Init START
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO comm 0x9ff3aa0 rank 103 nranks 128 cudaDev 7 nvmlDev 7 busId ca000 commId 0xe67d0548a5bfb9af - Init START
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO comm 0x889c010 rank 97 nranks 128 cudaDev 1 nvmlDev 1 busId 64000 commId 0xe67d0548a5bfb9af - Init START
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO comm 0x937b130 rank 56 nranks 128 cudaDev 0 nvmlDev 0 busId 53000 commId 0xe67d0548a5bfb9af - Init START
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO comm 0x90028a0 rank 55 nranks 128 cudaDev 7 nvmlDev 7 busId ca000 commId 0xe67d0548a5bfb9af - Init START
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO comm 0x9a07a60 rank 54 nranks 128 cudaDev 6 nvmlDev 6 busId b9000 commId 0xe67d0548a5bfb9af - Init START
+ip-26-0-161-78:114516:114691 [4] NCCL INFO comm 0x936c450 rank 68 nranks 128 cudaDev 4 nvmlDev 4 busId 97000 commId 0xe67d0548a5bfb9af - Init START
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO comm 0xa120e40 rank 96 nranks 128 cudaDev 0 nvmlDev 0 busId 53000 commId 0xe67d0548a5bfb9af - Init START
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO comm 0x8917950 rank 100 nranks 128 cudaDev 4 nvmlDev 4 busId 97000 commId 0xe67d0548a5bfb9af - Init START
+ip-26-0-165-202:153567:153742 [5] NCCL INFO comm 0xa1c25f0 rank 93 nranks 128 cudaDev 5 nvmlDev 5 busId a8000 commId 0xe67d0548a5bfb9af - Init START
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO comm 0x8818910 rank 124 nranks 128 cudaDev 4 nvmlDev 4 busId 97000 commId 0xe67d0548a5bfb9af - Init START
+ip-26-0-168-52:2722048:2722220 [3] NCCL INFO comm 0x864def0 rank 123 nranks 128 cudaDev 3 nvmlDev 3 busId 86000 commId 0xe67d0548a5bfb9af - Init START
+ip-26-0-168-52:2722047:2722221 [2] NCCL INFO comm 0x94226d0 rank 122 nranks 128 cudaDev 2 nvmlDev 2 busId 75000 commId 0xe67d0548a5bfb9af - Init START
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO comm 0x8e6e9d0 rank 72 nranks 128 cudaDev 0 nvmlDev 0 busId 53000 commId 0xe67d0548a5bfb9af - Init START
+ip-26-0-160-225:124160:124335 [0] NCCL INFO comm 0x921d930 rank 0 nranks 128 cudaDev 0 nvmlDev 0 busId 53000 commId 0xe67d0548a5bfb9af - Init START
+ip-26-0-160-225:124164:124336 [4] NCCL INFO comm 0x8f6da90 rank 4 nranks 128 cudaDev 4 nvmlDev 4 busId 97000 commId 0xe67d0548a5bfb9af - Init START
+ip-26-0-160-225:124162:124342 [2] NCCL INFO comm 0xa2aa6e0 rank 2 nranks 128 cudaDev 2 nvmlDev 2 busId 75000 commId 0xe67d0548a5bfb9af - Init START
+ip-26-0-160-225:124166:124339 [6] NCCL INFO comm 0x975d3b0 rank 6 nranks 128 cudaDev 6 nvmlDev 6 busId b9000 commId 0xe67d0548a5bfb9af - Init START
+ip-26-0-168-52:2722050:2722225 [5] NCCL INFO comm 0x86815d0 rank 125 nranks 128 cudaDev 5 nvmlDev 5 busId a8000 commId 0xe67d0548a5bfb9af - Init START
+ip-26-0-168-52:2722046:2722223 [1] NCCL INFO comm 0x9800e30 rank 121 nranks 128 cudaDev 1 nvmlDev 1 busId 64000 commId 0xe67d0548a5bfb9af - Init START
+ip-26-0-168-52:2722052:2722226 [7] NCCL INFO comm 0x9675870 rank 127 nranks 128 cudaDev 7 nvmlDev 7 busId ca000 commId 0xe67d0548a5bfb9af - Init START
+ip-26-0-168-52:2722051:2722224 [6] NCCL INFO comm 0x92559c0 rank 126 nranks 128 cudaDev 6 nvmlDev 6 busId b9000 commId 0xe67d0548a5bfb9af - Init START
+ip-26-0-160-225:124163:124340 [3] NCCL INFO comm 0x94c8a80 rank 3 nranks 128 cudaDev 3 nvmlDev 3 busId 86000 commId 0xe67d0548a5bfb9af - Init START
+ip-26-0-160-225:124167:124338 [7] NCCL INFO comm 0x9c467b0 rank 7 nranks 128 cudaDev 7 nvmlDev 7 busId ca000 commId 0xe67d0548a5bfb9af - Init START
+ip-26-0-160-225:124161:124337 [1] NCCL INFO comm 0x87faca0 rank 1 nranks 128 cudaDev 1 nvmlDev 1 busId 64000 commId 0xe67d0548a5bfb9af - Init START
+ip-26-0-160-225:124165:124341 [5] NCCL INFO comm 0x8838dc0 rank 5 nranks 128 cudaDev 5 nvmlDev 5 busId a8000 commId 0xe67d0548a5bfb9af - Init START
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO comm 0x9ad1540 rank 9 nranks 128 cudaDev 1 nvmlDev 1 busId 64000 commId 0xe67d0548a5bfb9af - Init START
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO comm 0x97add10 rank 8 nranks 128 cudaDev 0 nvmlDev 0 busId 53000 commId 0xe67d0548a5bfb9af - Init START
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO comm 0x9f93670 rank 12 nranks 128 cudaDev 4 nvmlDev 4 busId 97000 commId 0xe67d0548a5bfb9af - Init START
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO comm 0xa3ae1e0 rank 14 nranks 128 cudaDev 6 nvmlDev 6 busId b9000 commId 0xe67d0548a5bfb9af - Init START
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO comm 0x9e64d60 rank 15 nranks 128 cudaDev 7 nvmlDev 7 busId ca000 commId 0xe67d0548a5bfb9af - Init START
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO comm 0x963e6f0 rank 10 nranks 128 cudaDev 2 nvmlDev 2 busId 75000 commId 0xe67d0548a5bfb9af - Init START
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO comm 0x982c5c0 rank 11 nranks 128 cudaDev 3 nvmlDev 3 busId 86000 commId 0xe67d0548a5bfb9af - Init START
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO comm 0x8690d60 rank 13 nranks 128 cudaDev 5 nvmlDev 5 busId a8000 commId 0xe67d0548a5bfb9af - Init START
+ip-26-0-161-138:96563:96756 [0] NCCL INFO comm 0x99b2260 rank 32 nranks 128 cudaDev 0 nvmlDev 0 busId 53000 commId 0xe67d0548a5bfb9af - Init START
+ip-26-0-161-138:96566:96755 [3] NCCL INFO comm 0x9315300 rank 35 nranks 128 cudaDev 3 nvmlDev 3 busId 86000 commId 0xe67d0548a5bfb9af - Init START
+ip-26-0-161-138:96565:96754 [2] NCCL INFO comm 0xa208380 rank 34 nranks 128 cudaDev 2 nvmlDev 2 busId 75000 commId 0xe67d0548a5bfb9af - Init START
+ip-26-0-161-123:32710:32883 [6] NCCL INFO comm 0x991e950 rank 30 nranks 128 cudaDev 6 nvmlDev 6 busId b9000 commId 0xe67d0548a5bfb9af - Init START
+ip-26-0-161-123:32711:32884 [7] NCCL INFO comm 0x9e7ffa0 rank 31 nranks 128 cudaDev 7 nvmlDev 7 busId ca000 commId 0xe67d0548a5bfb9af - Init START
+ip-26-0-161-123:32704:32881 [0] NCCL INFO comm 0xa45bb70 rank 24 nranks 128 cudaDev 0 nvmlDev 0 busId 53000 commId 0xe67d0548a5bfb9af - Init START
+ip-26-0-161-123:32709:32882 [5] NCCL INFO comm 0xa23fe10 rank 29 nranks 128 cudaDev 5 nvmlDev 5 busId a8000 commId 0xe67d0548a5bfb9af - Init START
+ip-26-0-161-123:32708:32879 [4] NCCL INFO comm 0xa27e140 rank 28 nranks 128 cudaDev 4 nvmlDev 4 busId 97000 commId 0xe67d0548a5bfb9af - Init START
+ip-26-0-161-123:32707:32878 [3] NCCL INFO comm 0x887a500 rank 27 nranks 128 cudaDev 3 nvmlDev 3 busId 86000 commId 0xe67d0548a5bfb9af - Init START
+ip-26-0-161-103:96892:97064 [4] NCCL INFO comm 0x9b9ac80 rank 20 nranks 128 cudaDev 4 nvmlDev 4 busId 97000 commId 0xe67d0548a5bfb9af - Init START
+ip-26-0-161-103:96889:97065 [1] NCCL INFO comm 0x87d4fe0 rank 17 nranks 128 cudaDev 1 nvmlDev 1 busId 64000 commId 0xe67d0548a5bfb9af - Init START
+ip-26-0-161-103:96888:97066 [0] NCCL INFO comm 0x8e3cea0 rank 16 nranks 128 cudaDev 0 nvmlDev 0 busId 53000 commId 0xe67d0548a5bfb9af - Init START
+ip-26-0-161-103:96893:97067 [5] NCCL INFO comm 0x86ea090 rank 21 nranks 128 cudaDev 5 nvmlDev 5 busId a8000 commId 0xe67d0548a5bfb9af - Init START
+ip-26-0-161-103:96895:97069 [7] NCCL INFO comm 0xa4213d0 rank 23 nranks 128 cudaDev 7 nvmlDev 7 busId ca000 commId 0xe67d0548a5bfb9af - Init START
+ip-26-0-161-103:96890:97070 [2] NCCL INFO comm 0x8e18080 rank 18 nranks 128 cudaDev 2 nvmlDev 2 busId 75000 commId 0xe67d0548a5bfb9af - Init START
+ip-26-0-161-138:96564:96757 [1] NCCL INFO comm 0x96232d0 rank 33 nranks 128 cudaDev 1 nvmlDev 1 busId 64000 commId 0xe67d0548a5bfb9af - Init START
+ip-26-0-161-123:32705:32880 [1] NCCL INFO comm 0x99eb180 rank 25 nranks 128 cudaDev 1 nvmlDev 1 busId 64000 commId 0xe67d0548a5bfb9af - Init START
+ip-26-0-161-123:32706:32877 [2] NCCL INFO comm 0x9a13240 rank 26 nranks 128 cudaDev 2 nvmlDev 2 busId 75000 commId 0xe67d0548a5bfb9af - Init START
+ip-26-0-161-103:96894:97068 [6] NCCL INFO comm 0x8a7aba0 rank 22 nranks 128 cudaDev 6 nvmlDev 6 busId b9000 commId 0xe67d0548a5bfb9af - Init START
+ip-26-0-161-103:96891:97063 [3] NCCL INFO comm 0x8c696e0 rank 19 nranks 128 cudaDev 3 nvmlDev 3 busId 86000 commId 0xe67d0548a5bfb9af - Init START
+ip-26-0-161-138:96566:96755 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-138:96565:96754 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-138:96564:96757 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-138:96563:96756 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-123:32711:32884 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-123:32709:32882 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-123:32710:32883 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-123:32708:32879 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-123:32707:32878 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-123:32706:32877 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-103:96895:97069 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-103:96893:97067 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-103:96894:97068 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-103:96892:97064 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-103:96891:97063 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-123:32705:32880 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-123:32704:32881 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-160-225:124166:124339 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-160-225:124167:124338 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-103:96888:97066 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-103:96890:97070 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-103:96889:97065 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-34:272157:272324 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-34:272156:272325 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-34:272155:272329 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-34:272154:272330 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-34:272153:272328 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-30:235655:235828 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-30:235654:235821 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-30:235653:235825 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-34:272152:272331 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-34:272151:272326 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-34:272150:272327 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-30:235652:235822 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-30:235651:235823 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-30:235650:235824 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-30:235649:235827 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-30:235648:235826 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-202:153566:153741 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-202:153565:153738 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-202:153569:153739 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-202:153568:153736 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/409
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-202:153567:153742 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-202:153564:153737 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/409
+ip-26-0-165-202:153563:153740 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-202:153562:153735 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-78:114519:114692 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/409
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/409
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/409
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/409
+ip-26-0-160-225:124165:124341 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-78:114517:114687 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-78:114518:114693 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/409
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/409
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-78:114516:114691 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-78:114515:114690 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-78:114514:114689 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-78:114513:114688 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-78:114512:114686 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-142:62888:63060 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-142:62887:63055 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-142:62886:63057 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-138:96570:96759 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-142:62885:63054 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-52:2722045:2722219 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-142:62884:63061 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-138:96569:96758 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-138:96568:96753 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-142:62883:63058 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-142:62882:63056 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-142:62881:63059 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-160-225:124164:124336 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-138:96567:96760 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-160-225:124163:124340 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-52:2722047:2722221 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-52:2722046:2722223 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-52:2722048:2722220 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-160-225:124162:124342 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-160-225:124160:124335 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/541
+ip-26-0-160-225:124161:124337 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-52:2722052:2722226 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-52:2722050:2722225 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-52:2722051:2722224 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO Setting affinity for GPU 3 to 0fff,ffffffff
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-161-123:32710:32883 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-161-123:32710:32883 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-168-30:235648:235826 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-168-30:235648:235826 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-168-30:235654:235821 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-168-30:235654:235821 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,fff00000,00000000
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-161-123:32705:32880 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-161-123:32705:32880 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-168-30:235650:235824 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-168-30:235650:235824 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO Setting affinity for GPU 1 to 0fff,ffffffff
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-168-34:272154:272330 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-168-34:272154:272330 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO Setting affinity for GPU 0 to 0fff,ffffffff
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-168-30:235649:235827 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-168-30:235649:235827 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-168-30:235652:235822 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-168-30:235652:235822 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-161-78:114519:114692 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-161-78:114519:114692 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-161-123:32707:32878 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-161-123:32707:32878 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-165-202:153564:153737 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-165-202:153564:153737 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-161-78:114512:114686 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-161-78:114512:114686 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO Setting affinity for GPU 2 to 0fff,ffffffff
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,fff00000,00000000
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-168-34:272153:272328 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-168-34:272153:272328 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-168-30:235651:235823 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-168-30:235651:235823 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-161-123:32709:32882 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-161-123:32709:32882 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-161-142:62881:63059 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-161-142:62881:63059 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-168-30:235653:235825 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-168-30:235653:235825 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-160-225:124165:124341 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-160-225:124165:124341 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-168-30:235655:235828 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-168-30:235655:235828 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-161-123:32708:32879 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-161-123:32708:32879 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,fff00000,00000000
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,fff00000,00000000
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-161-142:62885:63054 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-161-142:62885:63054 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-168-52:2722046:2722223 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-168-52:2722046:2722223 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-161-78:114518:114693 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-161-78:114518:114693 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-161-123:32704:32881 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-161-123:32704:32881 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-161-142:62888:63060 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-161-142:62884:63061 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-161-142:62884:63061 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-161-142:62888:63060 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-161-123:32706:32877 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-161-123:32706:32877 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-161-123:32711:32884 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-161-123:32711:32884 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-168-52:2722047:2722221 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-168-52:2722047:2722221 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-161-138:96565:96754 [2] NCCL INFO Setting affinity for GPU 2 to 0fff,ffffffff
+ip-26-0-161-138:96565:96754 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-160-225:124162:124342 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-160-225:124162:124342 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO Setting affinity for GPU 1 to 0fff,ffffffff
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-165-202:153569:153739 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-165-202:153569:153739 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-168-34:272151:272326 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-168-34:272151:272326 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-168-34:272155:272329 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-168-34:272155:272329 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,fff00000,00000000
+ip-26-0-168-34:272150:272327 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-168-34:272150:272327 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-161-138:96566:96755 [3] NCCL INFO Setting affinity for GPU 3 to 0fff,ffffffff
+ip-26-0-161-138:96566:96755 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-168-34:272152:272331 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-161-142:62882:63056 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-161-142:62882:63056 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-168-34:272152:272331 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-161-78:114515:114690 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-161-103:96895:97069 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-161-103:96895:97069 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-161-78:114515:114690 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-168-34:272156:272325 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-168-34:272156:272325 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-161-78:114517:114687 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-161-78:114517:114687 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-161-142:62887:63055 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-161-142:62887:63055 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-161-138:96568:96753 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,fff00000,00000000
+ip-26-0-161-138:96568:96753 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-161-142:62883:63058 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-161-142:62883:63058 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-161-78:114516:114691 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-161-78:114516:114691 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,fff00000,00000000
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,fff00000,00000000
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-165-202:153566:153741 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-165-202:153566:153741 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-161-103:96891:97063 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-161-103:96891:97063 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO Setting affinity for GPU 2 to 0fff,ffffffff
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-168-34:272157:272324 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-168-34:272157:272324 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-161-138:96570:96759 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,fff00000,00000000
+ip-26-0-161-138:96570:96759 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,fff00000,00000000
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-165-202:153567:153742 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-161-138:96564:96757 [1] NCCL INFO Setting affinity for GPU 1 to 0fff,ffffffff
+ip-26-0-161-138:96564:96757 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-161-142:62886:63057 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-161-142:62886:63057 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-165-202:153562:153735 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-165-202:153562:153735 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-165-202:153567:153742 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-161-138:96563:96756 [0] NCCL INFO Setting affinity for GPU 0 to 0fff,ffffffff
+ip-26-0-161-138:96563:96756 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO Setting affinity for GPU 3 to 0fff,ffffffff
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-161-78:114513:114688 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-161-78:114513:114688 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-165-202:153565:153738 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-165-202:153565:153738 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO Setting affinity for GPU 0 to 0fff,ffffffff
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-161-78:114514:114689 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-161-138:96567:96760 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,fff00000,00000000
+ip-26-0-161-138:96567:96760 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-161-78:114514:114689 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-168-52:2722051:2722224 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-168-52:2722051:2722224 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-161-103:96890:97070 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-161-103:96890:97070 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-160-225:124164:124336 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-160-225:124164:124336 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-161-138:96569:96758 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,fff00000,00000000
+ip-26-0-161-138:96569:96758 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-168-52:2722052:2722226 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-168-52:2722052:2722226 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-165-202:153563:153740 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-165-202:153563:153740 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-161-103:96892:97064 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-161-103:96892:97064 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-165-202:153568:153736 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-165-202:153568:153736 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-161-103:96888:97066 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-161-103:96888:97066 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-168-52:2722045:2722219 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-168-52:2722045:2722219 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-168-52:2722048:2722220 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-168-52:2722050:2722225 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-168-52:2722050:2722225 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-168-52:2722048:2722220 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-160-225:124161:124337 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-160-225:124161:124337 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-160-225:124160:124335 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-160-225:124160:124335 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-160-225:124167:124338 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-160-225:124167:124338 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-160-225:124166:124339 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-160-225:124166:124339 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-161-103:96889:97065 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-161-103:96889:97065 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-161-103:96894:97068 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-161-103:96894:97068 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-160-225:124163:124340 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-160-225:124163:124340 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-161-103:96893:97067 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-161-103:96893:97067 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO Trees [0] 97/80/112->96->64 [1] -1/-1/-1->96->103 [2] 97/-1/-1->96->103 [3] 97/-1/-1->96->103 [4] 97/-1/-1->96->103 [5] 97/-1/-1->96->103 [6] 97/-1/-1->96->103 [7] 97/-1/-1->96->103 [8] 97/-1/-1->96->104 [9] -1/-1/-1->96->103 [10] 97/-1/-1->96->103 [11] 97/-1/-1->96->103 [12] 97/-1/-1->96->103 [13] 97/-1/-1->96->103 [14] 97/-1/-1->96->103 [15] 97/-1/-1->96->103
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO Trees [0] 98/-1/-1->97->96 [1] 98/81/113->97->65 [2] -1/-1/-1->97->96 [3] 98/-1/-1->97->96 [4] 98/-1/-1->97->96 [5] 98/-1/-1->97->96 [6] 98/-1/-1->97->96 [7] 98/-1/-1->97->96 [8] 98/-1/-1->97->96 [9] 98/-1/-1->97->105 [10] -1/-1/-1->97->96 [11] 98/-1/-1->97->96 [12] 98/-1/-1->97->96 [13] 98/-1/-1->97->96 [14] 98/-1/-1->97->96 [15] 98/-1/-1->97->96
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO Trees [0] 99/-1/-1->98->97 [1] 99/-1/-1->98->97 [2] 99/82/114->98->66 [3] -1/-1/-1->98->97 [4] 99/-1/-1->98->97 [5] 99/-1/-1->98->97 [6] 99/-1/-1->98->97 [7] 99/-1/-1->98->97 [8] 99/-1/-1->98->97 [9] 99/-1/-1->98->97 [10] 99/-1/-1->98->106 [11] -1/-1/-1->98->97 [12] 99/-1/-1->98->97 [13] 99/-1/-1->98->97 [14] 99/-1/-1->98->97 [15] 99/-1/-1->98->97
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO Trees [0] 100/-1/-1->99->98 [1] 100/-1/-1->99->98 [2] 100/-1/-1->99->98 [3] 100/83/115->99->67 [4] -1/-1/-1->99->98 [5] 100/-1/-1->99->98 [6] 100/-1/-1->99->98 [7] 100/-1/-1->99->98 [8] 100/-1/-1->99->98 [9] 100/-1/-1->99->98 [10] 100/-1/-1->99->98 [11] 100/-1/-1->99->107 [12] -1/-1/-1->99->98 [13] 100/-1/-1->99->98 [14] 100/-1/-1->99->98 [15] 100/-1/-1->99->98
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO Trees [0] 101/-1/-1->100->99 [1] 101/-1/-1->100->99 [2] 101/-1/-1->100->99 [3] 101/-1/-1->100->99 [4] 101/84/116->100->68 [5] -1/-1/-1->100->99 [6] 101/-1/-1->100->99 [7] 101/-1/-1->100->99 [8] 101/-1/-1->100->99 [9] 101/-1/-1->100->99 [10] 101/-1/-1->100->99 [11] 101/-1/-1->100->99 [12] 101/-1/-1->100->108 [13] -1/-1/-1->100->99 [14] 101/-1/-1->100->99 [15] 101/-1/-1->100->99
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO Trees [0] 102/-1/-1->101->100 [1] 102/-1/-1->101->100 [2] 102/-1/-1->101->100 [3] 102/-1/-1->101->100 [4] 102/-1/-1->101->100 [5] 102/85/117->101->69 [6] -1/-1/-1->101->100 [7] 102/-1/-1->101->100 [8] 102/-1/-1->101->100 [9] 102/-1/-1->101->100 [10] 102/-1/-1->101->100 [11] 102/-1/-1->101->100 [12] 102/-1/-1->101->100 [13] 102/-1/-1->101->109 [14] -1/-1/-1->101->100 [15] 102/-1/-1->101->100
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO Trees [0] 103/-1/-1->102->101 [1] 103/-1/-1->102->101 [2] 103/-1/-1->102->101 [3] 103/-1/-1->102->101 [4] 103/-1/-1->102->101 [5] 103/-1/-1->102->101 [6] 103/86/118->102->70 [7] -1/-1/-1->102->101 [8] 103/-1/-1->102->101 [9] 103/-1/-1->102->101 [10] 103/-1/-1->102->101 [11] 103/-1/-1->102->101 [12] 103/-1/-1->102->101 [13] 103/-1/-1->102->101 [14] 103/-1/-1->102->110 [15] -1/-1/-1->102->101
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO Trees [0] -1/-1/-1->103->102 [1] 96/-1/-1->103->102 [2] 96/-1/-1->103->102 [3] 96/-1/-1->103->102 [4] 96/-1/-1->103->102 [5] 96/-1/-1->103->102 [6] 96/-1/-1->103->102 [7] 96/87/119->103->71 [8] -1/-1/-1->103->102 [9] 96/-1/-1->103->102 [10] 96/-1/-1->103->102 [11] 96/-1/-1->103->102 [12] 96/-1/-1->103->102 [13] 96/-1/-1->103->102 [14] 96/-1/-1->103->102 [15] 96/-1/-1->103->111
+ip-26-0-168-30:235648:235826 [0] NCCL INFO Trees [0] 105/-1/-1->104->112 [1] -1/-1/-1->104->111 [2] 105/-1/-1->104->111 [3] 105/-1/-1->104->111 [4] 105/-1/-1->104->111 [5] 105/-1/-1->104->111 [6] 105/-1/-1->104->111 [7] 105/-1/-1->104->111 [8] 105/112/96->104->88 [9] -1/-1/-1->104->111 [10] 105/-1/-1->104->111 [11] 105/-1/-1->104->111 [12] 105/-1/-1->104->111 [13] 105/-1/-1->104->111 [14] 105/-1/-1->104->111 [15] 105/-1/-1->104->111
+ip-26-0-168-30:235648:235826 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO Trees [0] 87/-1/-1->86->85 [1] 87/-1/-1->86->85 [2] 87/-1/-1->86->85 [3] 87/-1/-1->86->85 [4] 87/-1/-1->86->85 [5] 87/-1/-1->86->85 [6] 87/78/94->86->102 [7] -1/-1/-1->86->85 [8] 87/-1/-1->86->85 [9] 87/-1/-1->86->85 [10] 87/-1/-1->86->85 [11] 87/-1/-1->86->85 [12] 87/-1/-1->86->85 [13] 87/-1/-1->86->85 [14] 87/-1/-1->86->78 [15] -1/-1/-1->86->85
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-202:153569:153739 [7] NCCL INFO Trees [0] -1/-1/-1->95->94 [1] 88/-1/-1->95->94 [2] 88/-1/-1->95->94 [3] 88/-1/-1->95->94 [4] 88/-1/-1->95->94 [5] 88/-1/-1->95->94 [6] 88/-1/-1->95->94 [7] 88/-1/-1->95->87 [8] -1/-1/-1->95->94 [9] 88/-1/-1->95->94 [10] 88/-1/-1->95->94 [11] 88/-1/-1->95->94 [12] 88/-1/-1->95->94 [13] 88/-1/-1->95->94 [14] 88/-1/-1->95->94 [15] 88/111/79->95->63
+ip-26-0-165-202:153569:153739 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO Trees [0] 86/-1/-1->85->84 [1] 86/-1/-1->85->84 [2] 86/-1/-1->85->84 [3] 86/-1/-1->85->84 [4] 86/-1/-1->85->84 [5] 86/77/93->85->101 [6] -1/-1/-1->85->84 [7] 86/-1/-1->85->84 [8] 86/-1/-1->85->84 [9] 86/-1/-1->85->84 [10] 86/-1/-1->85->84 [11] 86/-1/-1->85->84 [12] 86/-1/-1->85->84 [13] 86/-1/-1->85->77 [14] -1/-1/-1->85->84 [15] 86/-1/-1->85->84
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-202:153565:153738 [3] NCCL INFO Trees [0] 92/-1/-1->91->90 [1] 92/-1/-1->91->90 [2] 92/-1/-1->91->90 [3] 92/-1/-1->91->83 [4] -1/-1/-1->91->90 [5] 92/-1/-1->91->90 [6] 92/-1/-1->91->90 [7] 92/-1/-1->91->90 [8] 92/-1/-1->91->90 [9] 92/-1/-1->91->90 [10] 92/-1/-1->91->90 [11] 92/107/75->91->59 [12] -1/-1/-1->91->90 [13] 92/-1/-1->91->90 [14] 92/-1/-1->91->90 [15] 92/-1/-1->91->90
+ip-26-0-165-202:153565:153738 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-202:153568:153736 [6] NCCL INFO Trees [0] 95/-1/-1->94->93 [1] 95/-1/-1->94->93 [2] 95/-1/-1->94->93 [3] 95/-1/-1->94->93 [4] 95/-1/-1->94->93 [5] 95/-1/-1->94->93 [6] 95/-1/-1->94->86 [7] -1/-1/-1->94->93 [8] 95/-1/-1->94->93 [9] 95/-1/-1->94->93 [10] 95/-1/-1->94->93 [11] 95/-1/-1->94->93 [12] 95/-1/-1->94->93 [13] 95/-1/-1->94->93 [14] 95/110/78->94->62 [15] -1/-1/-1->94->93
+ip-26-0-165-202:153562:153735 [0] NCCL INFO Trees [0] 89/-1/-1->88->80 [1] -1/-1/-1->88->95 [2] 89/-1/-1->88->95 [3] 89/-1/-1->88->95 [4] 89/-1/-1->88->95 [5] 89/-1/-1->88->95 [6] 89/-1/-1->88->95 [7] 89/-1/-1->88->95 [8] 89/104/72->88->56 [9] -1/-1/-1->88->95 [10] 89/-1/-1->88->95 [11] 89/-1/-1->88->95 [12] 89/-1/-1->88->95 [13] 89/-1/-1->88->95 [14] 89/-1/-1->88->95 [15] 89/-1/-1->88->95
+ip-26-0-165-202:153568:153736 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-202:153562:153735 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-202:153566:153741 [4] NCCL INFO Trees [0] 93/-1/-1->92->91 [1] 93/-1/-1->92->91 [2] 93/-1/-1->92->91 [3] 93/-1/-1->92->91 [4] 93/-1/-1->92->84 [5] -1/-1/-1->92->91 [6] 93/-1/-1->92->91 [7] 93/-1/-1->92->91 [8] 93/-1/-1->92->91 [9] 93/-1/-1->92->91 [10] 93/-1/-1->92->91 [11] 93/-1/-1->92->91 [12] 93/108/76->92->60 [13] -1/-1/-1->92->91 [14] 93/-1/-1->92->91 [15] 93/-1/-1->92->91
+ip-26-0-165-202:153566:153741 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO Trees [0] 84/-1/-1->83->82 [1] 84/-1/-1->83->82 [2] 84/-1/-1->83->82 [3] 84/75/91->83->99 [4] -1/-1/-1->83->82 [5] 84/-1/-1->83->82 [6] 84/-1/-1->83->82 [7] 84/-1/-1->83->82 [8] 84/-1/-1->83->82 [9] 84/-1/-1->83->82 [10] 84/-1/-1->83->82 [11] 84/-1/-1->83->75 [12] -1/-1/-1->83->82 [13] 84/-1/-1->83->82 [14] 84/-1/-1->83->82 [15] 84/-1/-1->83->82
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-202:153564:153737 [2] NCCL INFO Trees [0] 91/-1/-1->90->89 [1] 91/-1/-1->90->89 [2] 91/-1/-1->90->82 [3] -1/-1/-1->90->89 [4] 91/-1/-1->90->89 [5] 91/-1/-1->90->89 [6] 91/-1/-1->90->89 [7] 91/-1/-1->90->89 [8] 91/-1/-1->90->89 [9] 91/-1/-1->90->89 [10] 91/106/74->90->58 [11] -1/-1/-1->90->89 [12] 91/-1/-1->90->89 [13] 91/-1/-1->90->89 [14] 91/-1/-1->90->89 [15] 91/-1/-1->90->89
+ip-26-0-165-202:153564:153737 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-202:153567:153742 [5] NCCL INFO Trees [0] 94/-1/-1->93->92 [1] 94/-1/-1->93->92 [2] 94/-1/-1->93->92 [3] 94/-1/-1->93->92 [4] 94/-1/-1->93->92 [5] 94/-1/-1->93->85 [6] -1/-1/-1->93->92 [7] 94/-1/-1->93->92 [8] 94/-1/-1->93->92 [9] 94/-1/-1->93->92 [10] 94/-1/-1->93->92 [11] 94/-1/-1->93->92 [12] 94/-1/-1->93->92 [13] 94/109/77->93->61 [14] -1/-1/-1->93->92 [15] 94/-1/-1->93->92
+ip-26-0-165-202:153567:153742 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO Trees [0] -1/-1/-1->79->78 [1] 72/-1/-1->79->78 [2] 72/-1/-1->79->78 [3] 72/-1/-1->79->78 [4] 72/-1/-1->79->78 [5] 72/-1/-1->79->78 [6] 72/-1/-1->79->78 [7] 72/-1/-1->79->87 [8] -1/-1/-1->79->78 [9] 72/-1/-1->79->78 [10] 72/-1/-1->79->78 [11] 72/-1/-1->79->78 [12] 72/-1/-1->79->78 [13] 72/-1/-1->79->78 [14] 72/-1/-1->79->78 [15] 72/87/71->79->95
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-202:153563:153740 [1] NCCL INFO Trees [0] 90/-1/-1->89->88 [1] 90/-1/-1->89->81 [2] -1/-1/-1->89->88 [3] 90/-1/-1->89->88 [4] 90/-1/-1->89->88 [5] 90/-1/-1->89->88 [6] 90/-1/-1->89->88 [7] 90/-1/-1->89->88 [8] 90/-1/-1->89->88 [9] 90/105/73->89->57 [10] -1/-1/-1->89->88 [11] 90/-1/-1->89->88 [12] 90/-1/-1->89->88 [13] 90/-1/-1->89->88 [14] 90/-1/-1->89->88 [15] 90/-1/-1->89->88
+ip-26-0-165-202:153563:153740 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-78:114519:114692 [7] NCCL INFO Trees [0] -1/-1/-1->71->70 [1] 64/-1/-1->71->70 [2] 64/-1/-1->71->70 [3] 64/-1/-1->71->70 [4] 64/-1/-1->71->70 [5] 64/-1/-1->71->70 [6] 64/-1/-1->71->70 [7] 64/39/103->71->7 [8] -1/-1/-1->71->70 [9] 64/-1/-1->71->70 [10] 64/-1/-1->71->70 [11] 64/-1/-1->71->70 [12] 64/-1/-1->71->70 [13] 64/-1/-1->71->70 [14] 64/-1/-1->71->70 [15] 64/-1/-1->71->79
+ip-26-0-161-78:114519:114692 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO Trees [0] 79/-1/-1->78->77 [1] 79/-1/-1->78->77 [2] 79/-1/-1->78->77 [3] 79/-1/-1->78->77 [4] 79/-1/-1->78->77 [5] 79/-1/-1->78->77 [6] 79/-1/-1->78->86 [7] -1/-1/-1->78->77 [8] 79/-1/-1->78->77 [9] 79/-1/-1->78->77 [10] 79/-1/-1->78->77 [11] 79/-1/-1->78->77 [12] 79/-1/-1->78->77 [13] 79/-1/-1->78->77 [14] 79/86/70->78->94 [15] -1/-1/-1->78->77
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO Trees [0] 81/72/88->80->96 [1] -1/-1/-1->80->87 [2] 81/-1/-1->80->87 [3] 81/-1/-1->80->87 [4] 81/-1/-1->80->87 [5] 81/-1/-1->80->87 [6] 81/-1/-1->80->87 [7] 81/-1/-1->80->87 [8] 81/-1/-1->80->72 [9] -1/-1/-1->80->87 [10] 81/-1/-1->80->87 [11] 81/-1/-1->80->87 [12] 81/-1/-1->80->87 [13] 81/-1/-1->80->87 [14] 81/-1/-1->80->87 [15] 81/-1/-1->80->87
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO Trees [0] 74/-1/-1->73->72 [1] 74/-1/-1->73->81 [2] -1/-1/-1->73->72 [3] 74/-1/-1->73->72 [4] 74/-1/-1->73->72 [5] 74/-1/-1->73->72 [6] 74/-1/-1->73->72 [7] 74/-1/-1->73->72 [8] 74/-1/-1->73->72 [9] 74/81/65->73->89 [10] -1/-1/-1->73->72 [11] 74/-1/-1->73->72 [12] 74/-1/-1->73->72 [13] 74/-1/-1->73->72 [14] 74/-1/-1->73->72 [15] 74/-1/-1->73->72
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO Trees [0] 75/-1/-1->74->73 [1] 75/-1/-1->74->73 [2] 75/-1/-1->74->82 [3] -1/-1/-1->74->73 [4] 75/-1/-1->74->73 [5] 75/-1/-1->74->73 [6] 75/-1/-1->74->73 [7] 75/-1/-1->74->73 [8] 75/-1/-1->74->73 [9] 75/-1/-1->74->73 [10] 75/82/66->74->90 [11] -1/-1/-1->74->73 [12] 75/-1/-1->74->73 [13] 75/-1/-1->74->73 [14] 75/-1/-1->74->73 [15] 75/-1/-1->74->73
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO Trees [0] 82/-1/-1->81->80 [1] 82/73/89->81->97 [2] -1/-1/-1->81->80 [3] 82/-1/-1->81->80 [4] 82/-1/-1->81->80 [5] 82/-1/-1->81->80 [6] 82/-1/-1->81->80 [7] 82/-1/-1->81->80 [8] 82/-1/-1->81->80 [9] 82/-1/-1->81->73 [10] -1/-1/-1->81->80 [11] 82/-1/-1->81->80 [12] 82/-1/-1->81->80 [13] 82/-1/-1->81->80 [14] 82/-1/-1->81->80 [15] 82/-1/-1->81->80
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO Trees [0] -1/-1/-1->87->86 [1] 80/-1/-1->87->86 [2] 80/-1/-1->87->86 [3] 80/-1/-1->87->86 [4] 80/-1/-1->87->86 [5] 80/-1/-1->87->86 [6] 80/-1/-1->87->86 [7] 80/79/95->87->103 [8] -1/-1/-1->87->86 [9] 80/-1/-1->87->86 [10] 80/-1/-1->87->86 [11] 80/-1/-1->87->86 [12] 80/-1/-1->87->86 [13] 80/-1/-1->87->86 [14] 80/-1/-1->87->86 [15] 80/-1/-1->87->79
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO Trees [0] 78/-1/-1->77->76 [1] 78/-1/-1->77->76 [2] 78/-1/-1->77->76 [3] 78/-1/-1->77->76 [4] 78/-1/-1->77->76 [5] 78/-1/-1->77->85 [6] -1/-1/-1->77->76 [7] 78/-1/-1->77->76 [8] 78/-1/-1->77->76 [9] 78/-1/-1->77->76 [10] 78/-1/-1->77->76 [11] 78/-1/-1->77->76 [12] 78/-1/-1->77->76 [13] 78/85/69->77->93 [14] -1/-1/-1->77->76 [15] 78/-1/-1->77->76
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-78:114518:114693 [6] NCCL INFO Trees [0] 71/-1/-1->70->69 [1] 71/-1/-1->70->69 [2] 71/-1/-1->70->69 [3] 71/-1/-1->70->69 [4] 71/-1/-1->70->69 [5] 71/-1/-1->70->69 [6] 71/38/102->70->6 [7] -1/-1/-1->70->69 [8] 71/-1/-1->70->69 [9] 71/-1/-1->70->69 [10] 71/-1/-1->70->69 [11] 71/-1/-1->70->69 [12] 71/-1/-1->70->69 [13] 71/-1/-1->70->69 [14] 71/-1/-1->70->78 [15] -1/-1/-1->70->69
+ip-26-0-161-78:114518:114693 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO Trees [0] 83/-1/-1->82->81 [1] 83/-1/-1->82->81 [2] 83/74/90->82->98 [3] -1/-1/-1->82->81 [4] 83/-1/-1->82->81 [5] 83/-1/-1->82->81 [6] 83/-1/-1->82->81 [7] 83/-1/-1->82->81 [8] 83/-1/-1->82->81 [9] 83/-1/-1->82->81 [10] 83/-1/-1->82->74 [11] -1/-1/-1->82->81 [12] 83/-1/-1->82->81 [13] 83/-1/-1->82->81 [14] 83/-1/-1->82->81 [15] 83/-1/-1->82->81
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO Trees [0] 76/-1/-1->75->74 [1] 76/-1/-1->75->74 [2] 76/-1/-1->75->74 [3] 76/-1/-1->75->83 [4] -1/-1/-1->75->74 [5] 76/-1/-1->75->74 [6] 76/-1/-1->75->74 [7] 76/-1/-1->75->74 [8] 76/-1/-1->75->74 [9] 76/-1/-1->75->74 [10] 76/-1/-1->75->74 [11] 76/83/67->75->91 [12] -1/-1/-1->75->74 [13] 76/-1/-1->75->74 [14] 76/-1/-1->75->74 [15] 76/-1/-1->75->74
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO Trees [0] 85/-1/-1->84->83 [1] 85/-1/-1->84->83 [2] 85/-1/-1->84->83 [3] 85/-1/-1->84->83 [4] 85/76/92->84->100 [5] -1/-1/-1->84->83 [6] 85/-1/-1->84->83 [7] 85/-1/-1->84->83 [8] 85/-1/-1->84->83 [9] 85/-1/-1->84->83 [10] 85/-1/-1->84->83 [11] 85/-1/-1->84->83 [12] 85/-1/-1->84->76 [13] -1/-1/-1->84->83 [14] 85/-1/-1->84->83 [15] 85/-1/-1->84->83
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO Trees [0] 73/-1/-1->72->80 [1] -1/-1/-1->72->79 [2] 73/-1/-1->72->79 [3] 73/-1/-1->72->79 [4] 73/-1/-1->72->79 [5] 73/-1/-1->72->79 [6] 73/-1/-1->72->79 [7] 73/-1/-1->72->79 [8] 73/80/64->72->88 [9] -1/-1/-1->72->79 [10] 73/-1/-1->72->79 [11] 73/-1/-1->72->79 [12] 73/-1/-1->72->79 [13] 73/-1/-1->72->79 [14] 73/-1/-1->72->79 [15] 73/-1/-1->72->79
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-225:124167:124338 [7] NCCL INFO NVLS Head  0:  0  8 16 24 32 40 48 56 64 72 80 88 96 104 112 120
+ip-26-0-160-225:124166:124339 [6] NCCL INFO NVLS Head  0:  0  8 16 24 32 40 48 56 64 72 80 88 96 104 112 120
+ip-26-0-160-225:124166:124339 [6] NCCL INFO NVLS Head  1:  1  9 17 25 33 41 49 57 65 73 81 89 97 105 113 121
+ip-26-0-160-225:124167:124338 [7] NCCL INFO NVLS Head  1:  1  9 17 25 33 41 49 57 65 73 81 89 97 105 113 121
+ip-26-0-160-225:124166:124339 [6] NCCL INFO NVLS Head  2:  2 10 18 26 34 42 50 58 66 74 82 90 98 106 114 122
+ip-26-0-160-225:124167:124338 [7] NCCL INFO NVLS Head  2:  2 10 18 26 34 42 50 58 66 74 82 90 98 106 114 122
+ip-26-0-160-225:124166:124339 [6] NCCL INFO NVLS Head  3:  3 11 19 27 35 43 51 59 67 75 83 91 99 107 115 123
+ip-26-0-160-225:124167:124338 [7] NCCL INFO NVLS Head  3:  3 11 19 27 35 43 51 59 67 75 83 91 99 107 115 123
+ip-26-0-160-225:124166:124339 [6] NCCL INFO NVLS Head  4:  4 12 20 28 36 44 52 60 68 76 84 92 100 108 116 124
+ip-26-0-161-78:114517:114687 [5] NCCL INFO Trees [0] 70/-1/-1->69->68 [1] 70/-1/-1->69->68 [2] 70/-1/-1->69->68 [3] 70/-1/-1->69->68 [4] 70/-1/-1->69->68 [5] 70/37/101->69->5 [6] -1/-1/-1->69->68 [7] 70/-1/-1->69->68 [8] 70/-1/-1->69->68 [9] 70/-1/-1->69->68 [10] 70/-1/-1->69->68 [11] 70/-1/-1->69->68 [12] 70/-1/-1->69->68 [13] 70/-1/-1->69->77 [14] -1/-1/-1->69->68 [15] 70/-1/-1->69->68
+ip-26-0-161-78:114517:114687 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-225:124167:124338 [7] NCCL INFO NVLS Head  4:  4 12 20 28 36 44 52 60 68 76 84 92 100 108 116 124
+ip-26-0-161-78:114516:114691 [4] NCCL INFO Trees [0] 69/-1/-1->68->67 [1] 69/-1/-1->68->67 [2] 69/-1/-1->68->67 [3] 69/-1/-1->68->67 [4] 69/36/100->68->4 [5] -1/-1/-1->68->67 [6] 69/-1/-1->68->67 [7] 69/-1/-1->68->67 [8] 69/-1/-1->68->67 [9] 69/-1/-1->68->67 [10] 69/-1/-1->68->67 [11] 69/-1/-1->68->67 [12] 69/-1/-1->68->76 [13] -1/-1/-1->68->67 [14] 69/-1/-1->68->67 [15] 69/-1/-1->68->67
+ip-26-0-161-78:114516:114691 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-225:124166:124339 [6] NCCL INFO NVLS Head  5:  5 13 21 29 37 45 53 61 69 77 85 93 101 109 117 125
+ip-26-0-160-225:124167:124338 [7] NCCL INFO NVLS Head  5:  5 13 21 29 37 45 53 61 69 77 85 93 101 109 117 125
+ip-26-0-160-225:124165:124341 [5] NCCL INFO NVLS Head  0:  0  8 16 24 32 40 48 56 64 72 80 88 96 104 112 120
+ip-26-0-160-225:124166:124339 [6] NCCL INFO NVLS Head  6:  6 14 22 30 38 46 54 62 70 78 86 94 102 110 118 126
+ip-26-0-160-225:124167:124338 [7] NCCL INFO NVLS Head  6:  6 14 22 30 38 46 54 62 70 78 86 94 102 110 118 126
+ip-26-0-160-225:124166:124339 [6] NCCL INFO NVLS Head  7:  7 15 23 31 39 47 55 63 71 79 87 95 103 111 119 127
+ip-26-0-160-225:124165:124341 [5] NCCL INFO NVLS Head  1:  1  9 17 25 33 41 49 57 65 73 81 89 97 105 113 121
+ip-26-0-160-225:124167:124338 [7] NCCL INFO NVLS Head  7:  7 15 23 31 39 47 55 63 71 79 87 95 103 111 119 127
+ip-26-0-160-225:124165:124341 [5] NCCL INFO NVLS Head  2:  2 10 18 26 34 42 50 58 66 74 82 90 98 106 114 122
+ip-26-0-160-225:124165:124341 [5] NCCL INFO NVLS Head  3:  3 11 19 27 35 43 51 59 67 75 83 91 99 107 115 123
+ip-26-0-160-225:124165:124341 [5] NCCL INFO NVLS Head  4:  4 12 20 28 36 44 52 60 68 76 84 92 100 108 116 124
+ip-26-0-160-225:124165:124341 [5] NCCL INFO NVLS Head  5:  5 13 21 29 37 45 53 61 69 77 85 93 101 109 117 125
+ip-26-0-160-225:124165:124341 [5] NCCL INFO NVLS Head  6:  6 14 22 30 38 46 54 62 70 78 86 94 102 110 118 126
+ip-26-0-160-225:124165:124341 [5] NCCL INFO NVLS Head  7:  7 15 23 31 39 47 55 63 71 79 87 95 103 111 119 127
+ip-26-0-160-225:124160:124335 [0] NCCL INFO NVLS Head  0:  0  8 16 24 32 40 48 56 64 72 80 88 96 104 112 120
+ip-26-0-160-225:124160:124335 [0] NCCL INFO NVLS Head  1:  1  9 17 25 33 41 49 57 65 73 81 89 97 105 113 121
+ip-26-0-160-225:124164:124336 [4] NCCL INFO NVLS Head  0:  0  8 16 24 32 40 48 56 64 72 80 88 96 104 112 120
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO Trees [0] 55/-1/-1->54->53 [1] 55/-1/-1->54->53 [2] 55/-1/-1->54->53 [3] 55/-1/-1->54->53 [4] 55/-1/-1->54->53 [5] 55/-1/-1->54->53 [6] 55/46/62->54->38 [7] -1/-1/-1->54->53 [8] 55/-1/-1->54->53 [9] 55/-1/-1->54->53 [10] 55/-1/-1->54->53 [11] 55/-1/-1->54->53 [12] 55/-1/-1->54->53 [13] 55/-1/-1->54->53 [14] 55/-1/-1->54->46 [15] -1/-1/-1->54->53
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO Trees [0] -1/-1/-1->55->54 [1] 48/-1/-1->55->54 [2] 48/-1/-1->55->54 [3] 48/-1/-1->55->54 [4] 48/-1/-1->55->54 [5] 48/-1/-1->55->54 [6] 48/-1/-1->55->54 [7] 48/47/63->55->39 [8] -1/-1/-1->55->54 [9] 48/-1/-1->55->54 [10] 48/-1/-1->55->54 [11] 48/-1/-1->55->54 [12] 48/-1/-1->55->54 [13] 48/-1/-1->55->54 [14] 48/-1/-1->55->54 [15] 48/-1/-1->55->47
+ip-26-0-160-225:124160:124335 [0] NCCL INFO NVLS Head  2:  2 10 18 26 34 42 50 58 66 74 82 90 98 106 114 122
+ip-26-0-160-225:124164:124336 [4] NCCL INFO NVLS Head  1:  1  9 17 25 33 41 49 57 65 73 81 89 97 105 113 121
+ip-26-0-160-225:124160:124335 [0] NCCL INFO NVLS Head  3:  3 11 19 27 35 43 51 59 67 75 83 91 99 107 115 123
+ip-26-0-160-225:124164:124336 [4] NCCL INFO NVLS Head  2:  2 10 18 26 34 42 50 58 66 74 82 90 98 106 114 122
+ip-26-0-160-225:124160:124335 [0] NCCL INFO NVLS Head  4:  4 12 20 28 36 44 52 60 68 76 84 92 100 108 116 124
+ip-26-0-160-225:124164:124336 [4] NCCL INFO NVLS Head  3:  3 11 19 27 35 43 51 59 67 75 83 91 99 107 115 123
+ip-26-0-160-225:124160:124335 [0] NCCL INFO NVLS Head  5:  5 13 21 29 37 45 53 61 69 77 85 93 101 109 117 125
+ip-26-0-160-225:124164:124336 [4] NCCL INFO NVLS Head  4:  4 12 20 28 36 44 52 60 68 76 84 92 100 108 116 124
+ip-26-0-160-225:124160:124335 [0] NCCL INFO NVLS Head  6:  6 14 22 30 38 46 54 62 70 78 86 94 102 110 118 126
+ip-26-0-161-78:114515:114690 [3] NCCL INFO Trees [0] 68/-1/-1->67->66 [1] 68/-1/-1->67->66 [2] 68/-1/-1->67->66 [3] 68/35/99->67->3 [4] -1/-1/-1->67->66 [5] 68/-1/-1->67->66 [6] 68/-1/-1->67->66 [7] 68/-1/-1->67->66 [8] 68/-1/-1->67->66 [9] 68/-1/-1->67->66 [10] 68/-1/-1->67->66 [11] 68/-1/-1->67->75 [12] -1/-1/-1->67->66 [13] 68/-1/-1->67->66 [14] 68/-1/-1->67->66 [15] 68/-1/-1->67->66
+ip-26-0-161-78:114514:114689 [2] NCCL INFO Trees [0] 67/-1/-1->66->65 [1] 67/-1/-1->66->65 [2] 67/34/98->66->2 [3] -1/-1/-1->66->65 [4] 67/-1/-1->66->65 [5] 67/-1/-1->66->65 [6] 67/-1/-1->66->65 [7] 67/-1/-1->66->65 [8] 67/-1/-1->66->65 [9] 67/-1/-1->66->65 [10] 67/-1/-1->66->74 [11] -1/-1/-1->66->65 [12] 67/-1/-1->66->65 [13] 67/-1/-1->66->65 [14] 67/-1/-1->66->65 [15] 67/-1/-1->66->65
+ip-26-0-161-78:114515:114690 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-78:114514:114689 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-168-30:235649:235827 [1] NCCL INFO Trees [0] 106/-1/-1->105->104 [1] 106/-1/-1->105->113 [2] -1/-1/-1->105->104 [3] 106/-1/-1->105->104 [4] 106/-1/-1->105->104 [5] 106/-1/-1->105->104 [6] 106/-1/-1->105->104 [7] 106/-1/-1->105->104 [8] 106/-1/-1->105->104 [9] 106/113/97->105->89 [10] -1/-1/-1->105->104 [11] 106/-1/-1->105->104 [12] 106/-1/-1->105->104 [13] 106/-1/-1->105->104 [14] 106/-1/-1->105->104 [15] 106/-1/-1->105->104
+ip-26-0-168-30:235651:235823 [3] NCCL INFO Trees [0] 108/-1/-1->107->106 [1] 108/-1/-1->107->106 [2] 108/-1/-1->107->106 [3] 108/-1/-1->107->115 [4] -1/-1/-1->107->106 [5] 108/-1/-1->107->106 [6] 108/-1/-1->107->106 [7] 108/-1/-1->107->106 [8] 108/-1/-1->107->106 [9] 108/-1/-1->107->106 [10] 108/-1/-1->107->106 [11] 108/115/99->107->91 [12] -1/-1/-1->107->106 [13] 108/-1/-1->107->106 [14] 108/-1/-1->107->106 [15] 108/-1/-1->107->106
+ip-26-0-168-30:235649:235827 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO Trees [0] 77/-1/-1->76->75 [1] 77/-1/-1->76->75 [2] 77/-1/-1->76->75 [3] 77/-1/-1->76->75 [4] 77/-1/-1->76->84 [5] -1/-1/-1->76->75 [6] 77/-1/-1->76->75 [7] 77/-1/-1->76->75 [8] 77/-1/-1->76->75 [9] 77/-1/-1->76->75 [10] 77/-1/-1->76->75 [11] 77/-1/-1->76->75 [12] 77/84/68->76->92 [13] -1/-1/-1->76->75 [14] 77/-1/-1->76->75 [15] 77/-1/-1->76->75
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-225:124164:124336 [4] NCCL INFO NVLS Head  5:  5 13 21 29 37 45 53 61 69 77 85 93 101 109 117 125
+ip-26-0-160-225:124160:124335 [0] NCCL INFO NVLS Head  7:  7 15 23 31 39 47 55 63 71 79 87 95 103 111 119 127
+ip-26-0-160-225:124164:124336 [4] NCCL INFO NVLS Head  6:  6 14 22 30 38 46 54 62 70 78 86 94 102 110 118 126
+ip-26-0-160-225:124164:124336 [4] NCCL INFO NVLS Head  7:  7 15 23 31 39 47 55 63 71 79 87 95 103 111 119 127
+ip-26-0-160-225:124166:124339 [6] NCCL INFO Trees [0] 7/-1/-1->6->5 [1] 7/-1/-1->6->5 [2] 7/-1/-1->6->5 [3] 7/-1/-1->6->5 [4] 7/-1/-1->6->5 [5] 7/-1/-1->6->5 [6] 7/70/-1->6->-1 [7] -1/-1/-1->6->5 [8] 7/-1/-1->6->5 [9] 7/-1/-1->6->5 [10] 7/-1/-1->6->5 [11] 7/-1/-1->6->5 [12] 7/-1/-1->6->5 [13] 7/-1/-1->6->5 [14] 7/-1/-1->6->14 [15] -1/-1/-1->6->5
+ip-26-0-168-52:2722052:2722226 [7] NCCL INFO Trees [0] -1/-1/-1->127->126 [1] 120/-1/-1->127->126 [2] 120/-1/-1->127->126 [3] 120/-1/-1->127->126 [4] 120/-1/-1->127->126 [5] 120/-1/-1->127->126 [6] 120/-1/-1->127->126 [7] 120/-1/-1->127->119 [8] -1/-1/-1->127->126 [9] 120/-1/-1->127->126 [10] 120/-1/-1->127->126 [11] 120/-1/-1->127->126 [12] 120/-1/-1->127->126 [13] 120/-1/-1->127->126 [14] 120/-1/-1->127->126 [15] 120/63/-1->127->-1
+ip-26-0-168-52:2722051:2722224 [6] NCCL INFO Trees [0] 127/-1/-1->126->125 [1] 127/-1/-1->126->125 [2] 127/-1/-1->126->125 [3] 127/-1/-1->126->125 [4] 127/-1/-1->126->125 [5] 127/-1/-1->126->125 [6] 127/-1/-1->126->118 [7] -1/-1/-1->126->125 [8] 127/-1/-1->126->125 [9] 127/-1/-1->126->125 [10] 127/-1/-1->126->125 [11] 127/-1/-1->126->125 [12] 127/-1/-1->126->125 [13] 127/-1/-1->126->125 [14] 127/62/-1->126->-1 [15] -1/-1/-1->126->125
+ip-26-0-168-52:2722052:2722226 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-168-52:2722051:2722224 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-142:62888:63060 [7] NCCL INFO Trees [0] -1/-1/-1->47->46 [1] 40/-1/-1->47->46 [2] 40/-1/-1->47->46 [3] 40/-1/-1->47->46 [4] 40/-1/-1->47->46 [5] 40/-1/-1->47->46 [6] 40/-1/-1->47->46 [7] 40/-1/-1->47->55 [8] -1/-1/-1->47->46 [9] 40/-1/-1->47->46 [10] 40/-1/-1->47->46 [11] 40/-1/-1->47->46 [12] 40/-1/-1->47->46 [13] 40/-1/-1->47->46 [14] 40/-1/-1->47->46 [15] 40/55/39->47->31
+ip-26-0-161-142:62888:63060 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-168-30:235652:235822 [4] NCCL INFO Trees [0] 109/-1/-1->108->107 [1] 109/-1/-1->108->107 [2] 109/-1/-1->108->107 [3] 109/-1/-1->108->107 [4] 109/-1/-1->108->116 [5] -1/-1/-1->108->107 [6] 109/-1/-1->108->107 [7] 109/-1/-1->108->107 [8] 109/-1/-1->108->107 [9] 109/-1/-1->108->107 [10] 109/-1/-1->108->107 [11] 109/-1/-1->108->107 [12] 109/116/100->108->92 [13] -1/-1/-1->108->107 [14] 109/-1/-1->108->107 [15] 109/-1/-1->108->107
+ip-26-0-168-30:235651:235823 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-168-30:235652:235822 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-168-30:235655:235828 [7] NCCL INFO Trees [0] -1/-1/-1->111->110 [1] 104/-1/-1->111->110 [2] 104/-1/-1->111->110 [3] 104/-1/-1->111->110 [4] 104/-1/-1->111->110 [5] 104/-1/-1->111->110 [6] 104/-1/-1->111->110 [7] 104/-1/-1->111->119 [8] -1/-1/-1->111->110 [9] 104/-1/-1->111->110 [10] 104/-1/-1->111->110 [11] 104/-1/-1->111->110 [12] 104/-1/-1->111->110 [13] 104/-1/-1->111->110 [14] 104/-1/-1->111->110 [15] 104/119/103->111->95
+ip-26-0-168-34:272152:272331 [2] NCCL INFO Trees [0] 115/-1/-1->114->113 [1] 115/-1/-1->114->113 [2] 115/106/122->114->98 [3] -1/-1/-1->114->113 [4] 115/-1/-1->114->113 [5] 115/-1/-1->114->113 [6] 115/-1/-1->114->113 [7] 115/-1/-1->114->113 [8] 115/-1/-1->114->113 [9] 115/-1/-1->114->113 [10] 115/-1/-1->114->106 [11] -1/-1/-1->114->113 [12] 115/-1/-1->114->113 [13] 115/-1/-1->114->113 [14] 115/-1/-1->114->113 [15] 115/-1/-1->114->113
+ip-26-0-168-34:272152:272331 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO Trees [0] 61/-1/-1->60->59 [1] 61/-1/-1->60->59 [2] 61/-1/-1->60->59 [3] 61/-1/-1->60->59 [4] 61/-1/-1->60->52 [5] -1/-1/-1->60->59 [6] 61/-1/-1->60->59 [7] 61/-1/-1->60->59 [8] 61/-1/-1->60->59 [9] 61/-1/-1->60->59 [10] 61/-1/-1->60->59 [11] 61/-1/-1->60->59 [12] 61/92/28->60->124 [13] -1/-1/-1->60->59 [14] 61/-1/-1->60->59 [15] 61/-1/-1->60->59
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO Trees [0] -1/-1/-1->63->62 [1] 56/-1/-1->63->62 [2] 56/-1/-1->63->62 [3] 56/-1/-1->63->62 [4] 56/-1/-1->63->62 [5] 56/-1/-1->63->62 [6] 56/-1/-1->63->62 [7] 56/-1/-1->63->55 [8] -1/-1/-1->63->62 [9] 56/-1/-1->63->62 [10] 56/-1/-1->63->62 [11] 56/-1/-1->63->62 [12] 56/-1/-1->63->62 [13] 56/-1/-1->63->62 [14] 56/-1/-1->63->62 [15] 56/95/31->63->127
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO Trees [0] 9/-1/-1->8->16 [1] -1/-1/-1->8->15 [2] 9/-1/-1->8->15 [3] 9/-1/-1->8->15 [4] 9/-1/-1->8->15 [5] 9/-1/-1->8->15 [6] 9/-1/-1->8->15 [7] 9/-1/-1->8->15 [8] 9/16/0->8->24 [9] -1/-1/-1->8->15 [10] 9/-1/-1->8->15 [11] 9/-1/-1->8->15 [12] 9/-1/-1->8->15 [13] 9/-1/-1->8->15 [14] 9/-1/-1->8->15 [15] 9/-1/-1->8->15
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO Trees [0] 54/-1/-1->53->52 [1] 54/-1/-1->53->52 [2] 54/-1/-1->53->52 [3] 54/-1/-1->53->52 [4] 54/-1/-1->53->52 [5] 54/45/61->53->37 [6] -1/-1/-1->53->52 [7] 54/-1/-1->53->52 [8] 54/-1/-1->53->52 [9] 54/-1/-1->53->52 [10] 54/-1/-1->53->52 [11] 54/-1/-1->53->52 [12] 54/-1/-1->53->52 [13] 54/-1/-1->53->45 [14] -1/-1/-1->53->52 [15] 54/-1/-1->53->52
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-225:124167:124338 [7] NCCL INFO Trees [0] -1/-1/-1->7->6 [1] 0/-1/-1->7->6 [2] 0/-1/-1->7->6 [3] 0/-1/-1->7->6 [4] 0/-1/-1->7->6 [5] 0/-1/-1->7->6 [6] 0/-1/-1->7->6 [7] 0/71/-1->7->-1 [8] -1/-1/-1->7->6 [9] 0/-1/-1->7->6 [10] 0/-1/-1->7->6 [11] 0/-1/-1->7->6 [12] 0/-1/-1->7->6 [13] 0/-1/-1->7->6 [14] 0/-1/-1->7->6 [15] 0/-1/-1->7->15
+ip-26-0-160-225:124166:124339 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-225:124167:124338 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-225:124160:124335 [0] NCCL INFO Channel 00/16 :    0   1   2   3   4   5   6   7   8   9  10  11  12  13  14  15  16  17  18  19
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO Trees [0] 125/-1/-1->124->123 [1] 125/-1/-1->124->123 [2] 125/-1/-1->124->123 [3] 125/-1/-1->124->123 [4] 125/-1/-1->124->116 [5] -1/-1/-1->124->123 [6] 125/-1/-1->124->123 [7] 125/-1/-1->124->123 [8] 125/-1/-1->124->123 [9] 125/-1/-1->124->123 [10] 125/-1/-1->124->123 [11] 125/-1/-1->124->123 [12] 125/60/-1->124->-1 [13] -1/-1/-1->124->123 [14] 125/-1/-1->124->123 [15] 125/-1/-1->124->123
+ip-26-0-161-78:114512:114686 [0] NCCL INFO Trees [0] 65/32/96->64->0 [1] -1/-1/-1->64->71 [2] 65/-1/-1->64->71 [3] 65/-1/-1->64->71 [4] 65/-1/-1->64->71 [5] 65/-1/-1->64->71 [6] 65/-1/-1->64->71 [7] 65/-1/-1->64->71 [8] 65/-1/-1->64->72 [9] -1/-1/-1->64->71 [10] 65/-1/-1->64->71 [11] 65/-1/-1->64->71 [12] 65/-1/-1->64->71 [13] 65/-1/-1->64->71 [14] 65/-1/-1->64->71 [15] 65/-1/-1->64->71
+ip-26-0-161-78:114512:114686 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-142:62887:63055 [6] NCCL INFO Trees [0] 47/-1/-1->46->45 [1] 47/-1/-1->46->45 [2] 47/-1/-1->46->45 [3] 47/-1/-1->46->45 [4] 47/-1/-1->46->45 [5] 47/-1/-1->46->45 [6] 47/-1/-1->46->54 [7] -1/-1/-1->46->45 [8] 47/-1/-1->46->45 [9] 47/-1/-1->46->45 [10] 47/-1/-1->46->45 [11] 47/-1/-1->46->45 [12] 47/-1/-1->46->45 [13] 47/-1/-1->46->45 [14] 47/54/38->46->30 [15] -1/-1/-1->46->45
+ip-26-0-161-142:62887:63055 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-168-30:235655:235828 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-168-30:235653:235825 [5] NCCL INFO Trees [0] 110/-1/-1->109->108 [1] 110/-1/-1->109->108 [2] 110/-1/-1->109->108 [3] 110/-1/-1->109->108 [4] 110/-1/-1->109->108 [5] 110/-1/-1->109->117 [6] -1/-1/-1->109->108 [7] 110/-1/-1->109->108 [8] 110/-1/-1->109->108 [9] 110/-1/-1->109->108 [10] 110/-1/-1->109->108 [11] 110/-1/-1->109->108 [12] 110/-1/-1->109->108 [13] 110/117/101->109->93 [14] -1/-1/-1->109->108 [15] 110/-1/-1->109->108
+ip-26-0-168-30:235653:235825 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-168-34:272154:272330 [4] NCCL INFO Trees [0] 117/-1/-1->116->115 [1] 117/-1/-1->116->115 [2] 117/-1/-1->116->115 [3] 117/-1/-1->116->115 [4] 117/108/124->116->100 [5] -1/-1/-1->116->115 [6] 117/-1/-1->116->115 [7] 117/-1/-1->116->115 [8] 117/-1/-1->116->115 [9] 117/-1/-1->116->115 [10] 117/-1/-1->116->115 [11] 117/-1/-1->116->115 [12] 117/-1/-1->116->108 [13] -1/-1/-1->116->115 [14] 117/-1/-1->116->115 [15] 117/-1/-1->116->115
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO Trees [0] 62/-1/-1->61->60 [1] 62/-1/-1->61->60 [2] 62/-1/-1->61->60 [3] 62/-1/-1->61->60 [4] 62/-1/-1->61->60 [5] 62/-1/-1->61->53 [6] -1/-1/-1->61->60 [7] 62/-1/-1->61->60 [8] 62/-1/-1->61->60 [9] 62/-1/-1->61->60 [10] 62/-1/-1->61->60 [11] 62/-1/-1->61->60 [12] 62/-1/-1->61->60 [13] 62/93/29->61->125 [14] -1/-1/-1->61->60 [15] 62/-1/-1->61->60
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO Trees [0] 53/-1/-1->52->51 [1] 53/-1/-1->52->51 [2] 53/-1/-1->52->51 [3] 53/-1/-1->52->51 [4] 53/44/60->52->36 [5] -1/-1/-1->52->51 [6] 53/-1/-1->52->51 [7] 53/-1/-1->52->51 [8] 53/-1/-1->52->51 [9] 53/-1/-1->52->51 [10] 53/-1/-1->52->51 [11] 53/-1/-1->52->51 [12] 53/-1/-1->52->44 [13] -1/-1/-1->52->51 [14] 53/-1/-1->52->51 [15] 53/-1/-1->52->51
+ip-26-0-160-225:124161:124337 [1] NCCL INFO NVLS Head  0:  0  8 16 24 32 40 48 56 64 72 80 88 96 104 112 120
+ip-26-0-160-225:124160:124335 [0] NCCL INFO Channel 01/16 :    0   7   6   5   4   3   2   9   8  15  14  13  12  11  10  17  16  23  22  21
+ip-26-0-160-225:124163:124340 [3] NCCL INFO NVLS Head  0:  0  8 16 24 32 40 48 56 64 72 80 88 96 104 112 120
+ip-26-0-160-225:124161:124337 [1] NCCL INFO NVLS Head  1:  1  9 17 25 33 41 49 57 65 73 81 89 97 105 113 121
+ip-26-0-160-225:124161:124337 [1] NCCL INFO NVLS Head  2:  2 10 18 26 34 42 50 58 66 74 82 90 98 106 114 122
+ip-26-0-160-225:124163:124340 [3] NCCL INFO NVLS Head  1:  1  9 17 25 33 41 49 57 65 73 81 89 97 105 113 121
+ip-26-0-160-225:124160:124335 [0] NCCL INFO Channel 02/16 :    0   7   6   5   4   3  10   9   8  15  14  13  12  11  18  17  16  23  22  21
+ip-26-0-160-225:124161:124337 [1] NCCL INFO NVLS Head  3:  3 11 19 27 35 43 51 59 67 75 83 91 99 107 115 123
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-138:96568:96753 [5] NCCL INFO Trees [0] 38/-1/-1->37->36 [1] 38/-1/-1->37->36 [2] 38/-1/-1->37->36 [3] 38/-1/-1->37->36 [4] 38/-1/-1->37->36 [5] 38/21/53->37->69 [6] -1/-1/-1->37->36 [7] 38/-1/-1->37->36 [8] 38/-1/-1->37->36 [9] 38/-1/-1->37->36 [10] 38/-1/-1->37->36 [11] 38/-1/-1->37->36 [12] 38/-1/-1->37->36 [13] 38/-1/-1->37->45 [14] -1/-1/-1->37->36 [15] 38/-1/-1->37->36
+ip-26-0-161-138:96568:96753 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-78:114513:114688 [1] NCCL INFO Trees [0] 66/-1/-1->65->64 [1] 66/33/97->65->1 [2] -1/-1/-1->65->64 [3] 66/-1/-1->65->64 [4] 66/-1/-1->65->64 [5] 66/-1/-1->65->64 [6] 66/-1/-1->65->64 [7] 66/-1/-1->65->64 [8] 66/-1/-1->65->64 [9] 66/-1/-1->65->73 [10] -1/-1/-1->65->64 [11] 66/-1/-1->65->64 [12] 66/-1/-1->65->64 [13] 66/-1/-1->65->64 [14] 66/-1/-1->65->64 [15] 66/-1/-1->65->64
+ip-26-0-161-78:114513:114688 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-142:62886:63057 [5] NCCL INFO Trees [0] 46/-1/-1->45->44 [1] 46/-1/-1->45->44 [2] 46/-1/-1->45->44 [3] 46/-1/-1->45->44 [4] 46/-1/-1->45->44 [5] 46/-1/-1->45->53 [6] -1/-1/-1->45->44 [7] 46/-1/-1->45->44 [8] 46/-1/-1->45->44 [9] 46/-1/-1->45->44 [10] 46/-1/-1->45->44 [11] 46/-1/-1->45->44 [12] 46/-1/-1->45->44 [13] 46/53/37->45->29 [14] -1/-1/-1->45->44 [15] 46/-1/-1->45->44
+ip-26-0-161-142:62886:63057 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-168-34:272154:272330 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-168-34:272156:272325 [6] NCCL INFO Trees [0] 119/-1/-1->118->117 [1] 119/-1/-1->118->117 [2] 119/-1/-1->118->117 [3] 119/-1/-1->118->117 [4] 119/-1/-1->118->117 [5] 119/-1/-1->118->117 [6] 119/110/126->118->102 [7] -1/-1/-1->118->117 [8] 119/-1/-1->118->117 [9] 119/-1/-1->118->117 [10] 119/-1/-1->118->117 [11] 119/-1/-1->118->117 [12] 119/-1/-1->118->117 [13] 119/-1/-1->118->117 [14] 119/-1/-1->118->110 [15] -1/-1/-1->118->117
+ip-26-0-168-34:272156:272325 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-168-34:272157:272324 [7] NCCL INFO Trees [0] -1/-1/-1->119->118 [1] 112/-1/-1->119->118 [2] 112/-1/-1->119->118 [3] 112/-1/-1->119->118 [4] 112/-1/-1->119->118 [5] 112/-1/-1->119->118 [6] 112/-1/-1->119->118 [7] 112/111/127->119->103 [8] -1/-1/-1->119->118 [9] 112/-1/-1->119->118 [10] 112/-1/-1->119->118 [11] 112/-1/-1->119->118 [12] 112/-1/-1->119->118 [13] 112/-1/-1->119->118 [14] 112/-1/-1->119->118 [15] 112/-1/-1->119->111
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO Trees [0] 59/-1/-1->58->57 [1] 59/-1/-1->58->57 [2] 59/-1/-1->58->50 [3] -1/-1/-1->58->57 [4] 59/-1/-1->58->57 [5] 59/-1/-1->58->57 [6] 59/-1/-1->58->57 [7] 59/-1/-1->58->57 [8] 59/-1/-1->58->57 [9] 59/-1/-1->58->57 [10] 59/90/26->58->122 [11] -1/-1/-1->58->57 [12] 59/-1/-1->58->57 [13] 59/-1/-1->58->57 [14] 59/-1/-1->58->57 [15] 59/-1/-1->58->57
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-225:124163:124340 [3] NCCL INFO NVLS Head  2:  2 10 18 26 34 42 50 58 66 74 82 90 98 106 114 122
+ip-26-0-160-225:124165:124341 [5] NCCL INFO Trees [0] 6/-1/-1->5->4 [1] 6/-1/-1->5->4 [2] 6/-1/-1->5->4 [3] 6/-1/-1->5->4 [4] 6/-1/-1->5->4 [5] 6/69/-1->5->-1 [6] -1/-1/-1->5->4 [7] 6/-1/-1->5->4 [8] 6/-1/-1->5->4 [9] 6/-1/-1->5->4 [10] 6/-1/-1->5->4 [11] 6/-1/-1->5->4 [12] 6/-1/-1->5->4 [13] 6/-1/-1->5->13 [14] -1/-1/-1->5->4 [15] 6/-1/-1->5->4
+ip-26-0-160-225:124161:124337 [1] NCCL INFO NVLS Head  4:  4 12 20 28 36 44 52 60 68 76 84 92 100 108 116 124
+ip-26-0-160-225:124163:124340 [3] NCCL INFO NVLS Head  3:  3 11 19 27 35 43 51 59 67 75 83 91 99 107 115 123
+ip-26-0-160-225:124160:124335 [0] NCCL INFO Channel 03/16 :    0   7   6   5   4  11  10   9   8  15  14  13  12  19  18  17  16  23  22  21
+ip-26-0-160-225:124165:124341 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-225:124161:124337 [1] NCCL INFO NVLS Head  5:  5 13 21 29 37 45 53 61 69 77 85 93 101 109 117 125
+ip-26-0-168-52:2722046:2722223 [1] NCCL INFO Trees [0] 122/-1/-1->121->120 [1] 122/-1/-1->121->113 [2] -1/-1/-1->121->120 [3] 122/-1/-1->121->120 [4] 122/-1/-1->121->120 [5] 122/-1/-1->121->120 [6] 122/-1/-1->121->120 [7] 122/-1/-1->121->120 [8] 122/-1/-1->121->120 [9] 122/57/-1->121->-1 [10] -1/-1/-1->121->120 [11] 122/-1/-1->121->120 [12] 122/-1/-1->121->120 [13] 122/-1/-1->121->120 [14] 122/-1/-1->121->120 [15] 122/-1/-1->121->120
+ip-26-0-168-52:2722050:2722225 [5] NCCL INFO Trees [0] 126/-1/-1->125->124 [1] 126/-1/-1->125->124 [2] 126/-1/-1->125->124 [3] 126/-1/-1->125->124 [4] 126/-1/-1->125->124 [5] 126/-1/-1->125->117 [6] -1/-1/-1->125->124 [7] 126/-1/-1->125->124 [8] 126/-1/-1->125->124 [9] 126/-1/-1->125->124 [10] 126/-1/-1->125->124 [11] 126/-1/-1->125->124 [12] 126/-1/-1->125->124 [13] 126/61/-1->125->-1 [14] -1/-1/-1->125->124 [15] 126/-1/-1->125->124
+ip-26-0-168-34:272157:272324 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO Trees [0] 63/-1/-1->62->61 [1] 63/-1/-1->62->61 [2] 63/-1/-1->62->61 [3] 63/-1/-1->62->61 [4] 63/-1/-1->62->61 [5] 63/-1/-1->62->61 [6] 63/-1/-1->62->54 [7] -1/-1/-1->62->61 [8] 63/-1/-1->62->61 [9] 63/-1/-1->62->61 [10] 63/-1/-1->62->61 [11] 63/-1/-1->62->61 [12] 63/-1/-1->62->61 [13] 63/-1/-1->62->61 [14] 63/94/30->62->126 [15] -1/-1/-1->62->61
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-225:124163:124340 [3] NCCL INFO NVLS Head  4:  4 12 20 28 36 44 52 60 68 76 84 92 100 108 116 124
+ip-26-0-160-225:124161:124337 [1] NCCL INFO NVLS Head  6:  6 14 22 30 38 46 54 62 70 78 86 94 102 110 118 126
+ip-26-0-168-52:2722046:2722223 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-168-52:2722048:2722220 [3] NCCL INFO Trees [0] 124/-1/-1->123->122 [1] 124/-1/-1->123->122 [2] 124/-1/-1->123->122 [3] 124/-1/-1->123->115 [4] -1/-1/-1->123->122 [5] 124/-1/-1->123->122 [6] 124/-1/-1->123->122 [7] 124/-1/-1->123->122 [8] 124/-1/-1->123->122 [9] 124/-1/-1->123->122 [10] 124/-1/-1->123->122 [11] 124/59/-1->123->-1 [12] -1/-1/-1->123->122 [13] 124/-1/-1->123->122 [14] 124/-1/-1->123->122 [15] 124/-1/-1->123->122
+ip-26-0-168-52:2722050:2722225 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-168-52:2722048:2722220 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-142:62884:63061 [3] NCCL INFO Trees [0] 44/-1/-1->43->42 [1] 44/-1/-1->43->42 [2] 44/-1/-1->43->42 [3] 44/-1/-1->43->51 [4] -1/-1/-1->43->42 [5] 44/-1/-1->43->42 [6] 44/-1/-1->43->42 [7] 44/-1/-1->43->42 [8] 44/-1/-1->43->42 [9] 44/-1/-1->43->42 [10] 44/-1/-1->43->42 [11] 44/51/35->43->27 [12] -1/-1/-1->43->42 [13] 44/-1/-1->43->42 [14] 44/-1/-1->43->42 [15] 44/-1/-1->43->42
+ip-26-0-161-142:62882:63056 [1] NCCL INFO Trees [0] 42/-1/-1->41->40 [1] 42/-1/-1->41->49 [2] -1/-1/-1->41->40 [3] 42/-1/-1->41->40 [4] 42/-1/-1->41->40 [5] 42/-1/-1->41->40 [6] 42/-1/-1->41->40 [7] 42/-1/-1->41->40 [8] 42/-1/-1->41->40 [9] 42/49/33->41->25 [10] -1/-1/-1->41->40 [11] 42/-1/-1->41->40 [12] 42/-1/-1->41->40 [13] 42/-1/-1->41->40 [14] 42/-1/-1->41->40 [15] 42/-1/-1->41->40
+ip-26-0-161-142:62884:63061 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-142:62882:63056 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-168-30:235650:235824 [2] NCCL INFO Trees [0] 107/-1/-1->106->105 [1] 107/-1/-1->106->105 [2] 107/-1/-1->106->114 [3] -1/-1/-1->106->105 [4] 107/-1/-1->106->105 [5] 107/-1/-1->106->105 [6] 107/-1/-1->106->105 [7] 107/-1/-1->106->105 [8] 107/-1/-1->106->105 [9] 107/-1/-1->106->105 [10] 107/114/98->106->90 [11] -1/-1/-1->106->105 [12] 107/-1/-1->106->105 [13] 107/-1/-1->106->105 [14] 107/-1/-1->106->105 [15] 107/-1/-1->106->105
+ip-26-0-168-30:235650:235824 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-168-34:272155:272329 [5] NCCL INFO Trees [0] 118/-1/-1->117->116 [1] 118/-1/-1->117->116 [2] 118/-1/-1->117->116 [3] 118/-1/-1->117->116 [4] 118/-1/-1->117->116 [5] 118/109/125->117->101 [6] -1/-1/-1->117->116 [7] 118/-1/-1->117->116 [8] 118/-1/-1->117->116 [9] 118/-1/-1->117->116 [10] 118/-1/-1->117->116 [11] 118/-1/-1->117->116 [12] 118/-1/-1->117->116 [13] 118/-1/-1->117->109 [14] -1/-1/-1->117->116 [15] 118/-1/-1->117->116
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO Trees [0] 10/-1/-1->9->8 [1] 10/-1/-1->9->17 [2] -1/-1/-1->9->8 [3] 10/-1/-1->9->8 [4] 10/-1/-1->9->8 [5] 10/-1/-1->9->8 [6] 10/-1/-1->9->8 [7] 10/-1/-1->9->8 [8] 10/-1/-1->9->8 [9] 10/17/1->9->25 [10] -1/-1/-1->9->8 [11] 10/-1/-1->9->8 [12] 10/-1/-1->9->8 [13] 10/-1/-1->9->8 [14] 10/-1/-1->9->8 [15] 10/-1/-1->9->8
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-225:124160:124335 [0] NCCL INFO Channel 04/16 :    0   7   6   5  12  11  10   9   8  15  14  13  20  19  18  17  16  23  22  21
+ip-26-0-160-225:124163:124340 [3] NCCL INFO NVLS Head  5:  5 13 21 29 37 45 53 61 69 77 85 93 101 109 117 125
+ip-26-0-160-225:124161:124337 [1] NCCL INFO NVLS Head  7:  7 15 23 31 39 47 55 63 71 79 87 95 103 111 119 127
+ip-26-0-160-225:124163:124340 [3] NCCL INFO NVLS Head  6:  6 14 22 30 38 46 54 62 70 78 86 94 102 110 118 126
+ip-26-0-160-225:124163:124340 [3] NCCL INFO NVLS Head  7:  7 15 23 31 39 47 55 63 71 79 87 95 103 111 119 127
+ip-26-0-160-225:124160:124335 [0] NCCL INFO Channel 05/16 :    0   4   7   6  13  11  10   9   8  12  15  14  21  19  18  17  16  20  23  22
+ip-26-0-160-225:124160:124335 [0] NCCL INFO Channel 06/16 :    0   5   4   7  14  11  10   9   8  13  12  15  22  19  18  17  16  21  20  23
+ip-26-0-160-225:124160:124335 [0] NCCL INFO Channel 07/16 :    0   6   5   4  15  11  10   9   8  14  13  12  23  19  18  17  16  22  21  20
+ip-26-0-168-52:2722047:2722221 [2] NCCL INFO Trees [0] 123/-1/-1->122->121 [1] 123/-1/-1->122->121 [2] 123/-1/-1->122->114 [3] -1/-1/-1->122->121 [4] 123/-1/-1->122->121 [5] 123/-1/-1->122->121 [6] 123/-1/-1->122->121 [7] 123/-1/-1->122->121 [8] 123/-1/-1->122->121 [9] 123/-1/-1->122->121 [10] 123/58/-1->122->-1 [11] -1/-1/-1->122->121 [12] 123/-1/-1->122->121 [13] 123/-1/-1->122->121 [14] 123/-1/-1->122->121 [15] 123/-1/-1->122->121
+ip-26-0-168-52:2722045:2722219 [0] NCCL INFO Trees [0] 121/-1/-1->120->112 [1] -1/-1/-1->120->127 [2] 121/-1/-1->120->127 [3] 121/-1/-1->120->127 [4] 121/-1/-1->120->127 [5] 121/-1/-1->120->127 [6] 121/-1/-1->120->127 [7] 121/-1/-1->120->127 [8] 121/56/-1->120->-1 [9] -1/-1/-1->120->127 [10] 121/-1/-1->120->127 [11] 121/-1/-1->120->127 [12] 121/-1/-1->120->127 [13] 121/-1/-1->120->127 [14] 121/-1/-1->120->127 [15] 121/-1/-1->120->127
+ip-26-0-168-52:2722047:2722221 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-168-52:2722045:2722219 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-138:96566:96755 [3] NCCL INFO Trees [0] 36/-1/-1->35->34 [1] 36/-1/-1->35->34 [2] 36/-1/-1->35->34 [3] 36/19/51->35->67 [4] -1/-1/-1->35->34 [5] 36/-1/-1->35->34 [6] 36/-1/-1->35->34 [7] 36/-1/-1->35->34 [8] 36/-1/-1->35->34 [9] 36/-1/-1->35->34 [10] 36/-1/-1->35->34 [11] 36/-1/-1->35->43 [12] -1/-1/-1->35->34 [13] 36/-1/-1->35->34 [14] 36/-1/-1->35->34 [15] 36/-1/-1->35->34
+ip-26-0-161-138:96566:96755 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-142:62881:63059 [0] NCCL INFO Trees [0] 41/-1/-1->40->48 [1] -1/-1/-1->40->47 [2] 41/-1/-1->40->47 [3] 41/-1/-1->40->47 [4] 41/-1/-1->40->47 [5] 41/-1/-1->40->47 [6] 41/-1/-1->40->47 [7] 41/-1/-1->40->47 [8] 41/48/32->40->24 [9] -1/-1/-1->40->47 [10] 41/-1/-1->40->47 [11] 41/-1/-1->40->47 [12] 41/-1/-1->40->47 [13] 41/-1/-1->40->47 [14] 41/-1/-1->40->47 [15] 41/-1/-1->40->47
+ip-26-0-161-142:62881:63059 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-168-30:235654:235821 [6] NCCL INFO Trees [0] 111/-1/-1->110->109 [1] 111/-1/-1->110->109 [2] 111/-1/-1->110->109 [3] 111/-1/-1->110->109 [4] 111/-1/-1->110->109 [5] 111/-1/-1->110->109 [6] 111/-1/-1->110->118 [7] -1/-1/-1->110->109 [8] 111/-1/-1->110->109 [9] 111/-1/-1->110->109 [10] 111/-1/-1->110->109 [11] 111/-1/-1->110->109 [12] 111/-1/-1->110->109 [13] 111/-1/-1->110->109 [14] 111/118/102->110->94 [15] -1/-1/-1->110->109
+ip-26-0-168-30:235654:235821 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-168-34:272155:272329 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO Trees [0] 58/-1/-1->57->56 [1] 58/-1/-1->57->49 [2] -1/-1/-1->57->56 [3] 58/-1/-1->57->56 [4] 58/-1/-1->57->56 [5] 58/-1/-1->57->56 [6] 58/-1/-1->57->56 [7] 58/-1/-1->57->56 [8] 58/-1/-1->57->56 [9] 58/89/25->57->121 [10] -1/-1/-1->57->56 [11] 58/-1/-1->57->56 [12] 58/-1/-1->57->56 [13] 58/-1/-1->57->56 [14] 58/-1/-1->57->56 [15] 58/-1/-1->57->56
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO Trees [0] 11/-1/-1->10->9 [1] 11/-1/-1->10->9 [2] 11/-1/-1->10->18 [3] -1/-1/-1->10->9 [4] 11/-1/-1->10->9 [5] 11/-1/-1->10->9 [6] 11/-1/-1->10->9 [7] 11/-1/-1->10->9 [8] 11/-1/-1->10->9 [9] 11/-1/-1->10->9 [10] 11/18/2->10->26 [11] -1/-1/-1->10->9 [12] 11/-1/-1->10->9 [13] 11/-1/-1->10->9 [14] 11/-1/-1->10->9 [15] 11/-1/-1->10->9
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO Trees [0] 52/-1/-1->51->50 [1] 52/-1/-1->51->50 [2] 52/-1/-1->51->50 [3] 52/43/59->51->35 [4] -1/-1/-1->51->50 [5] 52/-1/-1->51->50 [6] 52/-1/-1->51->50 [7] 52/-1/-1->51->50 [8] 52/-1/-1->51->50 [9] 52/-1/-1->51->50 [10] 52/-1/-1->51->50 [11] 52/-1/-1->51->43 [12] -1/-1/-1->51->50 [13] 52/-1/-1->51->50 [14] 52/-1/-1->51->50 [15] 52/-1/-1->51->50
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-225:124162:124342 [2] NCCL INFO NVLS Head  0:  0  8 16 24 32 40 48 56 64 72 80 88 96 104 112 120
+ip-26-0-160-225:124160:124335 [0] NCCL INFO Channel 08/16 :    0   1   2   3   4   5   6   7   8   9  10  11  12  13  14  15  16  17  18  19
+ip-26-0-160-225:124162:124342 [2] NCCL INFO NVLS Head  1:  1  9 17 25 33 41 49 57 65 73 81 89 97 105 113 121
+ip-26-0-160-225:124162:124342 [2] NCCL INFO NVLS Head  2:  2 10 18 26 34 42 50 58 66 74 82 90 98 106 114 122
+ip-26-0-160-225:124160:124335 [0] NCCL INFO Channel 09/16 :    0   7   6   5   4   3   2   9   8  15  14  13  12  11  10  17  16  23  22  21
+ip-26-0-160-225:124162:124342 [2] NCCL INFO NVLS Head  3:  3 11 19 27 35 43 51 59 67 75 83 91 99 107 115 123
+ip-26-0-160-225:124162:124342 [2] NCCL INFO NVLS Head  4:  4 12 20 28 36 44 52 60 68 76 84 92 100 108 116 124
+ip-26-0-160-225:124160:124335 [0] NCCL INFO Channel 10/16 :    0   7   6   5   4   3  10   9   8  15  14  13  12  11  18  17  16  23  22  21
+ip-26-0-161-138:96567:96760 [4] NCCL INFO Trees [0] 37/-1/-1->36->35 [1] 37/-1/-1->36->35 [2] 37/-1/-1->36->35 [3] 37/-1/-1->36->35 [4] 37/20/52->36->68 [5] -1/-1/-1->36->35 [6] 37/-1/-1->36->35 [7] 37/-1/-1->36->35 [8] 37/-1/-1->36->35 [9] 37/-1/-1->36->35 [10] 37/-1/-1->36->35 [11] 37/-1/-1->36->35 [12] 37/-1/-1->36->44 [13] -1/-1/-1->36->35 [14] 37/-1/-1->36->35 [15] 37/-1/-1->36->35
+ip-26-0-161-138:96567:96760 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-142:62883:63058 [2] NCCL INFO Trees [0] 43/-1/-1->42->41 [1] 43/-1/-1->42->41 [2] 43/-1/-1->42->50 [3] -1/-1/-1->42->41 [4] 43/-1/-1->42->41 [5] 43/-1/-1->42->41 [6] 43/-1/-1->42->41 [7] 43/-1/-1->42->41 [8] 43/-1/-1->42->41 [9] 43/-1/-1->42->41 [10] 43/50/34->42->26 [11] -1/-1/-1->42->41 [12] 43/-1/-1->42->41 [13] 43/-1/-1->42->41 [14] 43/-1/-1->42->41 [15] 43/-1/-1->42->41
+ip-26-0-168-34:272151:272326 [1] NCCL INFO Trees [0] 114/-1/-1->113->112 [1] 114/105/121->113->97 [2] -1/-1/-1->113->112 [3] 114/-1/-1->113->112 [4] 114/-1/-1->113->112 [5] 114/-1/-1->113->112 [6] 114/-1/-1->113->112 [7] 114/-1/-1->113->112 [8] 114/-1/-1->113->112 [9] 114/-1/-1->113->105 [10] -1/-1/-1->113->112 [11] 114/-1/-1->113->112 [12] 114/-1/-1->113->112 [13] 114/-1/-1->113->112 [14] 114/-1/-1->113->112 [15] 114/-1/-1->113->112
+ip-26-0-168-34:272153:272328 [3] NCCL INFO Trees [0] 116/-1/-1->115->114 [1] 116/-1/-1->115->114 [2] 116/-1/-1->115->114 [3] 116/107/123->115->99 [4] -1/-1/-1->115->114 [5] 116/-1/-1->115->114 [6] 116/-1/-1->115->114 [7] 116/-1/-1->115->114 [8] 116/-1/-1->115->114 [9] 116/-1/-1->115->114 [10] 116/-1/-1->115->114 [11] 116/-1/-1->115->107 [12] -1/-1/-1->115->114 [13] 116/-1/-1->115->114 [14] 116/-1/-1->115->114 [15] 116/-1/-1->115->114
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO Trees [0] 15/-1/-1->14->13 [1] 15/-1/-1->14->13 [2] 15/-1/-1->14->13 [3] 15/-1/-1->14->13 [4] 15/-1/-1->14->13 [5] 15/-1/-1->14->13 [6] 15/-1/-1->14->22 [7] -1/-1/-1->14->13 [8] 15/-1/-1->14->13 [9] 15/-1/-1->14->13 [10] 15/-1/-1->14->13 [11] 15/-1/-1->14->13 [12] 15/-1/-1->14->13 [13] 15/-1/-1->14->13 [14] 15/22/6->14->30 [15] -1/-1/-1->14->13
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO Trees [0] 49/40/56->48->32 [1] -1/-1/-1->48->55 [2] 49/-1/-1->48->55 [3] 49/-1/-1->48->55 [4] 49/-1/-1->48->55 [5] 49/-1/-1->48->55 [6] 49/-1/-1->48->55 [7] 49/-1/-1->48->55 [8] 49/-1/-1->48->40 [9] -1/-1/-1->48->55 [10] 49/-1/-1->48->55 [11] 49/-1/-1->48->55 [12] 49/-1/-1->48->55 [13] 49/-1/-1->48->55 [14] 49/-1/-1->48->55 [15] 49/-1/-1->48->55
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO Trees [0] 50/-1/-1->49->48 [1] 50/41/57->49->33 [2] -1/-1/-1->49->48 [3] 50/-1/-1->49->48 [4] 50/-1/-1->49->48 [5] 50/-1/-1->49->48 [6] 50/-1/-1->49->48 [7] 50/-1/-1->49->48 [8] 50/-1/-1->49->48 [9] 50/-1/-1->49->41 [10] -1/-1/-1->49->48 [11] 50/-1/-1->49->48 [12] 50/-1/-1->49->48 [13] 50/-1/-1->49->48 [14] 50/-1/-1->49->48 [15] 50/-1/-1->49->48
+ip-26-0-160-225:124164:124336 [4] NCCL INFO Trees [0] 5/-1/-1->4->3 [1] 5/-1/-1->4->3 [2] 5/-1/-1->4->3 [3] 5/-1/-1->4->3 [4] 5/68/-1->4->-1 [5] -1/-1/-1->4->3 [6] 5/-1/-1->4->3 [7] 5/-1/-1->4->3 [8] 5/-1/-1->4->3 [9] 5/-1/-1->4->3 [10] 5/-1/-1->4->3 [11] 5/-1/-1->4->3 [12] 5/-1/-1->4->12 [13] -1/-1/-1->4->3 [14] 5/-1/-1->4->3 [15] 5/-1/-1->4->3
+ip-26-0-160-225:124162:124342 [2] NCCL INFO NVLS Head  5:  5 13 21 29 37 45 53 61 69 77 85 93 101 109 117 125
+ip-26-0-161-142:62883:63058 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-168-34:272150:272327 [0] NCCL INFO Trees [0] 113/104/120->112->96 [1] -1/-1/-1->112->119 [2] 113/-1/-1->112->119 [3] 113/-1/-1->112->119 [4] 113/-1/-1->112->119 [5] 113/-1/-1->112->119 [6] 113/-1/-1->112->119 [7] 113/-1/-1->112->119 [8] 113/-1/-1->112->104 [9] -1/-1/-1->112->119 [10] 113/-1/-1->112->119 [11] 113/-1/-1->112->119 [12] 113/-1/-1->112->119 [13] 113/-1/-1->112->119 [14] 113/-1/-1->112->119 [15] 113/-1/-1->112->119
+ip-26-0-168-34:272153:272328 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-168-34:272151:272326 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-103:96893:97067 [5] NCCL INFO Trees [0] 22/-1/-1->21->20 [1] 22/-1/-1->21->20 [2] 22/-1/-1->21->20 [3] 22/-1/-1->21->20 [4] 22/-1/-1->21->20 [5] 22/13/29->21->37 [6] -1/-1/-1->21->20 [7] 22/-1/-1->21->20 [8] 22/-1/-1->21->20 [9] 22/-1/-1->21->20 [10] 22/-1/-1->21->20 [11] 22/-1/-1->21->20 [12] 22/-1/-1->21->20 [13] 22/-1/-1->21->13 [14] -1/-1/-1->21->20 [15] 22/-1/-1->21->20
+ip-26-0-161-103:96893:97067 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO Trees [0] 60/-1/-1->59->58 [1] 60/-1/-1->59->58 [2] 60/-1/-1->59->58 [3] 60/-1/-1->59->51 [4] -1/-1/-1->59->58 [5] 60/-1/-1->59->58 [6] 60/-1/-1->59->58 [7] 60/-1/-1->59->58 [8] 60/-1/-1->59->58 [9] 60/-1/-1->59->58 [10] 60/-1/-1->59->58 [11] 60/91/27->59->123 [12] -1/-1/-1->59->58 [13] 60/-1/-1->59->58 [14] 60/-1/-1->59->58 [15] 60/-1/-1->59->58
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO Trees [0] 14/-1/-1->13->12 [1] 14/-1/-1->13->12 [2] 14/-1/-1->13->12 [3] 14/-1/-1->13->12 [4] 14/-1/-1->13->12 [5] 14/-1/-1->13->21 [6] -1/-1/-1->13->12 [7] 14/-1/-1->13->12 [8] 14/-1/-1->13->12 [9] 14/-1/-1->13->12 [10] 14/-1/-1->13->12 [11] 14/-1/-1->13->12 [12] 14/-1/-1->13->12 [13] 14/21/5->13->29 [14] -1/-1/-1->13->12 [15] 14/-1/-1->13->12
+ip-26-0-161-123:32711:32884 [7] NCCL INFO Trees [0] -1/-1/-1->31->30 [1] 24/-1/-1->31->30 [2] 24/-1/-1->31->30 [3] 24/-1/-1->31->30 [4] 24/-1/-1->31->30 [5] 24/-1/-1->31->30 [6] 24/-1/-1->31->30 [7] 24/-1/-1->31->23 [8] -1/-1/-1->31->30 [9] 24/-1/-1->31->30 [10] 24/-1/-1->31->30 [11] 24/-1/-1->31->30 [12] 24/-1/-1->31->30 [13] 24/-1/-1->31->30 [14] 24/-1/-1->31->30 [15] 24/47/15->31->63
+ip-26-0-161-123:32711:32884 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-123:32710:32883 [6] NCCL INFO Trees [0] 31/-1/-1->30->29 [1] 31/-1/-1->30->29 [2] 31/-1/-1->30->29 [3] 31/-1/-1->30->29 [4] 31/-1/-1->30->29 [5] 31/-1/-1->30->29 [6] 31/-1/-1->30->22 [7] -1/-1/-1->30->29 [8] 31/-1/-1->30->29 [9] 31/-1/-1->30->29 [10] 31/-1/-1->30->29 [11] 31/-1/-1->30->29 [12] 31/-1/-1->30->29 [13] 31/-1/-1->30->29 [14] 31/46/14->30->62 [15] -1/-1/-1->30->29
+ip-26-0-161-123:32710:32883 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-225:124164:124336 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-225:124162:124342 [2] NCCL INFO NVLS Head  6:  6 14 22 30 38 46 54 62 70 78 86 94 102 110 118 126
+ip-26-0-160-225:124160:124335 [0] NCCL INFO Channel 11/16 :    0   7   6   5   4  11  10   9   8  15  14  13  12  19  18  17  16  23  22  21
+ip-26-0-160-225:124162:124342 [2] NCCL INFO NVLS Head  7:  7 15 23 31 39 47 55 63 71 79 87 95 103 111 119 127
+ip-26-0-160-225:124160:124335 [0] NCCL INFO Channel 12/16 :    0   7   6   5  12  11  10   9   8  15  14  13  20  19  18  17  16  23  22  21
+ip-26-0-160-225:124160:124335 [0] NCCL INFO Channel 13/16 :    0   4   7   6  13  11  10   9   8  12  15  14  21  19  18  17  16  20  23  22
+ip-26-0-161-138:96563:96756 [0] NCCL INFO Trees [0] 33/16/48->32->64 [1] -1/-1/-1->32->39 [2] 33/-1/-1->32->39 [3] 33/-1/-1->32->39 [4] 33/-1/-1->32->39 [5] 33/-1/-1->32->39 [6] 33/-1/-1->32->39 [7] 33/-1/-1->32->39 [8] 33/-1/-1->32->40 [9] -1/-1/-1->32->39 [10] 33/-1/-1->32->39 [11] 33/-1/-1->32->39 [12] 33/-1/-1->32->39 [13] 33/-1/-1->32->39 [14] 33/-1/-1->32->39 [15] 33/-1/-1->32->39
+ip-26-0-161-138:96563:96756 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-138:96570:96759 [7] NCCL INFO Trees [0] -1/-1/-1->39->38 [1] 32/-1/-1->39->38 [2] 32/-1/-1->39->38 [3] 32/-1/-1->39->38 [4] 32/-1/-1->39->38 [5] 32/-1/-1->39->38 [6] 32/-1/-1->39->38 [7] 32/23/55->39->71 [8] -1/-1/-1->39->38 [9] 32/-1/-1->39->38 [10] 32/-1/-1->39->38 [11] 32/-1/-1->39->38 [12] 32/-1/-1->39->38 [13] 32/-1/-1->39->38 [14] 32/-1/-1->39->38 [15] 32/-1/-1->39->47
+ip-26-0-168-34:272150:272327 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO Trees [0] 57/-1/-1->56->48 [1] -1/-1/-1->56->63 [2] 57/-1/-1->56->63 [3] 57/-1/-1->56->63 [4] 57/-1/-1->56->63 [5] 57/-1/-1->56->63 [6] 57/-1/-1->56->63 [7] 57/-1/-1->56->63 [8] 57/88/24->56->120 [9] -1/-1/-1->56->63 [10] 57/-1/-1->56->63 [11] 57/-1/-1->56->63 [12] 57/-1/-1->56->63 [13] 57/-1/-1->56->63 [14] 57/-1/-1->56->63 [15] 57/-1/-1->56->63
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO Trees [0] -1/-1/-1->15->14 [1] 8/-1/-1->15->14 [2] 8/-1/-1->15->14 [3] 8/-1/-1->15->14 [4] 8/-1/-1->15->14 [5] 8/-1/-1->15->14 [6] 8/-1/-1->15->14 [7] 8/-1/-1->15->23 [8] -1/-1/-1->15->14 [9] 8/-1/-1->15->14 [10] 8/-1/-1->15->14 [11] 8/-1/-1->15->14 [12] 8/-1/-1->15->14 [13] 8/-1/-1->15->14 [14] 8/-1/-1->15->14 [15] 8/23/7->15->31
+ip-26-0-160-225:124160:124335 [0] NCCL INFO Channel 14/16 :    0   5   4   7  14  11  10   9   8  13  12  15  22  19  18  17  16  21  20  23
+ip-26-0-160-225:124160:124335 [0] NCCL INFO Channel 15/16 :    0   6   5   4  15  11  10   9   8  14  13  12  23  19  18  17  16  22  21  20
+ip-26-0-161-138:96569:96758 [6] NCCL INFO Trees [0] 39/-1/-1->38->37 [1] 39/-1/-1->38->37 [2] 39/-1/-1->38->37 [3] 39/-1/-1->38->37 [4] 39/-1/-1->38->37 [5] 39/-1/-1->38->37 [6] 39/22/54->38->70 [7] -1/-1/-1->38->37 [8] 39/-1/-1->38->37 [9] 39/-1/-1->38->37 [10] 39/-1/-1->38->37 [11] 39/-1/-1->38->37 [12] 39/-1/-1->38->37 [13] 39/-1/-1->38->37 [14] 39/-1/-1->38->46 [15] -1/-1/-1->38->37
+ip-26-0-161-138:96570:96759 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-142:62885:63054 [4] NCCL INFO Trees [0] 45/-1/-1->44->43 [1] 45/-1/-1->44->43 [2] 45/-1/-1->44->43 [3] 45/-1/-1->44->43 [4] 45/-1/-1->44->52 [5] -1/-1/-1->44->43 [6] 45/-1/-1->44->43 [7] 45/-1/-1->44->43 [8] 45/-1/-1->44->43 [9] 45/-1/-1->44->43 [10] 45/-1/-1->44->43 [11] 45/-1/-1->44->43 [12] 45/52/36->44->28 [13] -1/-1/-1->44->43 [14] 45/-1/-1->44->43 [15] 45/-1/-1->44->43
+ip-26-0-161-142:62885:63054 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-103:96895:97069 [7] NCCL INFO Trees [0] -1/-1/-1->23->22 [1] 16/-1/-1->23->22 [2] 16/-1/-1->23->22 [3] 16/-1/-1->23->22 [4] 16/-1/-1->23->22 [5] 16/-1/-1->23->22 [6] 16/-1/-1->23->22 [7] 16/15/31->23->39 [8] -1/-1/-1->23->22 [9] 16/-1/-1->23->22 [10] 16/-1/-1->23->22 [11] 16/-1/-1->23->22 [12] 16/-1/-1->23->22 [13] 16/-1/-1->23->22 [14] 16/-1/-1->23->22 [15] 16/-1/-1->23->15
+ip-26-0-161-103:96894:97068 [6] NCCL INFO Trees [0] 23/-1/-1->22->21 [1] 23/-1/-1->22->21 [2] 23/-1/-1->22->21 [3] 23/-1/-1->22->21 [4] 23/-1/-1->22->21 [5] 23/-1/-1->22->21 [6] 23/14/30->22->38 [7] -1/-1/-1->22->21 [8] 23/-1/-1->22->21 [9] 23/-1/-1->22->21 [10] 23/-1/-1->22->21 [11] 23/-1/-1->22->21 [12] 23/-1/-1->22->21 [13] 23/-1/-1->22->21 [14] 23/-1/-1->22->14 [15] -1/-1/-1->22->21
+ip-26-0-161-103:96895:97069 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-103:96894:97068 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-123:32708:32879 [4] NCCL INFO Trees [0] 29/-1/-1->28->27 [1] 29/-1/-1->28->27 [2] 29/-1/-1->28->27 [3] 29/-1/-1->28->27 [4] 29/-1/-1->28->20 [5] -1/-1/-1->28->27 [6] 29/-1/-1->28->27 [7] 29/-1/-1->28->27 [8] 29/-1/-1->28->27 [9] 29/-1/-1->28->27 [10] 29/-1/-1->28->27 [11] 29/-1/-1->28->27 [12] 29/44/12->28->60 [13] -1/-1/-1->28->27 [14] 29/-1/-1->28->27 [15] 29/-1/-1->28->27
+ip-26-0-161-123:32708:32879 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-123:32706:32877 [2] NCCL INFO Trees [0] 27/-1/-1->26->25 [1] 27/-1/-1->26->25 [2] 27/-1/-1->26->18 [3] -1/-1/-1->26->25 [4] 27/-1/-1->26->25 [5] 27/-1/-1->26->25 [6] 27/-1/-1->26->25 [7] 27/-1/-1->26->25 [8] 27/-1/-1->26->25 [9] 27/-1/-1->26->25 [10] 27/42/10->26->58 [11] -1/-1/-1->26->25 [12] 27/-1/-1->26->25 [13] 27/-1/-1->26->25 [14] 27/-1/-1->26->25 [15] 27/-1/-1->26->25
+ip-26-0-161-123:32706:32877 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-225:124161:124337 [1] NCCL INFO Trees [0] 2/-1/-1->1->0 [1] 2/65/-1->1->-1 [2] -1/-1/-1->1->0 [3] 2/-1/-1->1->0 [4] 2/-1/-1->1->0 [5] 2/-1/-1->1->0 [6] 2/-1/-1->1->0 [7] 2/-1/-1->1->0 [8] 2/-1/-1->1->0 [9] 2/-1/-1->1->9 [10] -1/-1/-1->1->0 [11] 2/-1/-1->1->0 [12] 2/-1/-1->1->0 [13] 2/-1/-1->1->0 [14] 2/-1/-1->1->0 [15] 2/-1/-1->1->0
+ip-26-0-160-225:124161:124337 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-225:124163:124340 [3] NCCL INFO Trees [0] 4/-1/-1->3->2 [1] 4/-1/-1->3->2 [2] 4/-1/-1->3->2 [3] 4/67/-1->3->-1 [4] -1/-1/-1->3->2 [5] 4/-1/-1->3->2 [6] 4/-1/-1->3->2 [7] 4/-1/-1->3->2 [8] 4/-1/-1->3->2 [9] 4/-1/-1->3->2 [10] 4/-1/-1->3->2 [11] 4/-1/-1->3->11 [12] -1/-1/-1->3->2 [13] 4/-1/-1->3->2 [14] 4/-1/-1->3->2 [15] 4/-1/-1->3->2
+ip-26-0-161-138:96569:96758 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-103:96892:97064 [4] NCCL INFO Trees [0] 21/-1/-1->20->19 [1] 21/-1/-1->20->19 [2] 21/-1/-1->20->19 [3] 21/-1/-1->20->19 [4] 21/12/28->20->36 [5] -1/-1/-1->20->19 [6] 21/-1/-1->20->19 [7] 21/-1/-1->20->19 [8] 21/-1/-1->20->19 [9] 21/-1/-1->20->19 [10] 21/-1/-1->20->19 [11] 21/-1/-1->20->19 [12] 21/-1/-1->20->12 [13] -1/-1/-1->20->19 [14] 21/-1/-1->20->19 [15] 21/-1/-1->20->19
+ip-26-0-161-103:96889:97065 [1] NCCL INFO Trees [0] 18/-1/-1->17->16 [1] 18/9/25->17->33 [2] -1/-1/-1->17->16 [3] 18/-1/-1->17->16 [4] 18/-1/-1->17->16 [5] 18/-1/-1->17->16 [6] 18/-1/-1->17->16 [7] 18/-1/-1->17->16 [8] 18/-1/-1->17->16 [9] 18/-1/-1->17->9 [10] -1/-1/-1->17->16 [11] 18/-1/-1->17->16 [12] 18/-1/-1->17->16 [13] 18/-1/-1->17->16 [14] 18/-1/-1->17->16 [15] 18/-1/-1->17->16
+ip-26-0-161-103:96892:97064 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-103:96889:97065 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO Trees [0] 13/-1/-1->12->11 [1] 13/-1/-1->12->11 [2] 13/-1/-1->12->11 [3] 13/-1/-1->12->11 [4] 13/-1/-1->12->20 [5] -1/-1/-1->12->11 [6] 13/-1/-1->12->11 [7] 13/-1/-1->12->11 [8] 13/-1/-1->12->11 [9] 13/-1/-1->12->11 [10] 13/-1/-1->12->11 [11] 13/-1/-1->12->11 [12] 13/20/4->12->28 [13] -1/-1/-1->12->11 [14] 13/-1/-1->12->11 [15] 13/-1/-1->12->11
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-123:32707:32878 [3] NCCL INFO Trees [0] 28/-1/-1->27->26 [1] 28/-1/-1->27->26 [2] 28/-1/-1->27->26 [3] 28/-1/-1->27->19 [4] -1/-1/-1->27->26 [5] 28/-1/-1->27->26 [6] 28/-1/-1->27->26 [7] 28/-1/-1->27->26 [8] 28/-1/-1->27->26 [9] 28/-1/-1->27->26 [10] 28/-1/-1->27->26 [11] 28/43/11->27->59 [12] -1/-1/-1->27->26 [13] 28/-1/-1->27->26 [14] 28/-1/-1->27->26 [15] 28/-1/-1->27->26
+ip-26-0-161-123:32709:32882 [5] NCCL INFO Trees [0] 30/-1/-1->29->28 [1] 30/-1/-1->29->28 [2] 30/-1/-1->29->28 [3] 30/-1/-1->29->28 [4] 30/-1/-1->29->28 [5] 30/-1/-1->29->21 [6] -1/-1/-1->29->28 [7] 30/-1/-1->29->28 [8] 30/-1/-1->29->28 [9] 30/-1/-1->29->28 [10] 30/-1/-1->29->28 [11] 30/-1/-1->29->28 [12] 30/-1/-1->29->28 [13] 30/45/13->29->61 [14] -1/-1/-1->29->28 [15] 30/-1/-1->29->28
+ip-26-0-161-123:32707:32878 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-123:32709:32882 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO Trees [0] 51/-1/-1->50->49 [1] 51/-1/-1->50->49 [2] 51/42/58->50->34 [3] -1/-1/-1->50->49 [4] 51/-1/-1->50->49 [5] 51/-1/-1->50->49 [6] 51/-1/-1->50->49 [7] 51/-1/-1->50->49 [8] 51/-1/-1->50->49 [9] 51/-1/-1->50->49 [10] 51/-1/-1->50->42 [11] -1/-1/-1->50->49 [12] 51/-1/-1->50->49 [13] 51/-1/-1->50->49 [14] 51/-1/-1->50->49 [15] 51/-1/-1->50->49
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-225:124160:124335 [0] NCCL INFO Trees [0] 1/64/-1->0->-1 [1] -1/-1/-1->0->7 [2] 1/-1/-1->0->7 [3] 1/-1/-1->0->7 [4] 1/-1/-1->0->7 [5] 1/-1/-1->0->7 [6] 1/-1/-1->0->7 [7] 1/-1/-1->0->7 [8] 1/-1/-1->0->8 [9] -1/-1/-1->0->7 [10] 1/-1/-1->0->7 [11] 1/-1/-1->0->7 [12] 1/-1/-1->0->7 [13] 1/-1/-1->0->7 [14] 1/-1/-1->0->7 [15] 1/-1/-1->0->7
+ip-26-0-160-225:124163:124340 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-225:124160:124335 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-138:96564:96757 [1] NCCL INFO Trees [0] 34/-1/-1->33->32 [1] 34/17/49->33->65 [2] -1/-1/-1->33->32 [3] 34/-1/-1->33->32 [4] 34/-1/-1->33->32 [5] 34/-1/-1->33->32 [6] 34/-1/-1->33->32 [7] 34/-1/-1->33->32 [8] 34/-1/-1->33->32 [9] 34/-1/-1->33->41 [10] -1/-1/-1->33->32 [11] 34/-1/-1->33->32 [12] 34/-1/-1->33->32 [13] 34/-1/-1->33->32 [14] 34/-1/-1->33->32 [15] 34/-1/-1->33->32
+ip-26-0-161-103:96888:97066 [0] NCCL INFO Trees [0] 17/8/24->16->32 [1] -1/-1/-1->16->23 [2] 17/-1/-1->16->23 [3] 17/-1/-1->16->23 [4] 17/-1/-1->16->23 [5] 17/-1/-1->16->23 [6] 17/-1/-1->16->23 [7] 17/-1/-1->16->23 [8] 17/-1/-1->16->8 [9] -1/-1/-1->16->23 [10] 17/-1/-1->16->23 [11] 17/-1/-1->16->23 [12] 17/-1/-1->16->23 [13] 17/-1/-1->16->23 [14] 17/-1/-1->16->23 [15] 17/-1/-1->16->23
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO Trees [0] 12/-1/-1->11->10 [1] 12/-1/-1->11->10 [2] 12/-1/-1->11->10 [3] 12/-1/-1->11->19 [4] -1/-1/-1->11->10 [5] 12/-1/-1->11->10 [6] 12/-1/-1->11->10 [7] 12/-1/-1->11->10 [8] 12/-1/-1->11->10 [9] 12/-1/-1->11->10 [10] 12/-1/-1->11->10 [11] 12/19/3->11->27 [12] -1/-1/-1->11->10 [13] 12/-1/-1->11->10 [14] 12/-1/-1->11->10 [15] 12/-1/-1->11->10
+ip-26-0-161-123:32704:32881 [0] NCCL INFO Trees [0] 25/-1/-1->24->16 [1] -1/-1/-1->24->31 [2] 25/-1/-1->24->31 [3] 25/-1/-1->24->31 [4] 25/-1/-1->24->31 [5] 25/-1/-1->24->31 [6] 25/-1/-1->24->31 [7] 25/-1/-1->24->31 [8] 25/40/8->24->56 [9] -1/-1/-1->24->31 [10] 25/-1/-1->24->31 [11] 25/-1/-1->24->31 [12] 25/-1/-1->24->31 [13] 25/-1/-1->24->31 [14] 25/-1/-1->24->31 [15] 25/-1/-1->24->31
+ip-26-0-161-123:32704:32881 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-138:96564:96757 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-103:96888:97066 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-123:32705:32880 [1] NCCL INFO Trees [0] 26/-1/-1->25->24 [1] 26/-1/-1->25->17 [2] -1/-1/-1->25->24 [3] 26/-1/-1->25->24 [4] 26/-1/-1->25->24 [5] 26/-1/-1->25->24 [6] 26/-1/-1->25->24 [7] 26/-1/-1->25->24 [8] 26/-1/-1->25->24 [9] 26/41/9->25->57 [10] -1/-1/-1->25->24 [11] 26/-1/-1->25->24 [12] 26/-1/-1->25->24 [13] 26/-1/-1->25->24 [14] 26/-1/-1->25->24 [15] 26/-1/-1->25->24
+ip-26-0-161-123:32705:32880 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-138:96565:96754 [2] NCCL INFO Trees [0] 35/-1/-1->34->33 [1] 35/-1/-1->34->33 [2] 35/18/50->34->66 [3] -1/-1/-1->34->33 [4] 35/-1/-1->34->33 [5] 35/-1/-1->34->33 [6] 35/-1/-1->34->33 [7] 35/-1/-1->34->33 [8] 35/-1/-1->34->33 [9] 35/-1/-1->34->33 [10] 35/-1/-1->34->42 [11] -1/-1/-1->34->33 [12] 35/-1/-1->34->33 [13] 35/-1/-1->34->33 [14] 35/-1/-1->34->33 [15] 35/-1/-1->34->33
+ip-26-0-161-103:96890:97070 [2] NCCL INFO Trees [0] 19/-1/-1->18->17 [1] 19/-1/-1->18->17 [2] 19/10/26->18->34 [3] -1/-1/-1->18->17 [4] 19/-1/-1->18->17 [5] 19/-1/-1->18->17 [6] 19/-1/-1->18->17 [7] 19/-1/-1->18->17 [8] 19/-1/-1->18->17 [9] 19/-1/-1->18->17 [10] 19/-1/-1->18->10 [11] -1/-1/-1->18->17 [12] 19/-1/-1->18->17 [13] 19/-1/-1->18->17 [14] 19/-1/-1->18->17 [15] 19/-1/-1->18->17
+ip-26-0-161-103:96890:97070 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-138:96565:96754 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-103:96891:97063 [3] NCCL INFO Trees [0] 20/-1/-1->19->18 [1] 20/-1/-1->19->18 [2] 20/-1/-1->19->18 [3] 20/11/27->19->35 [4] -1/-1/-1->19->18 [5] 20/-1/-1->19->18 [6] 20/-1/-1->19->18 [7] 20/-1/-1->19->18 [8] 20/-1/-1->19->18 [9] 20/-1/-1->19->18 [10] 20/-1/-1->19->18 [11] 20/-1/-1->19->11 [12] -1/-1/-1->19->18 [13] 20/-1/-1->19->18 [14] 20/-1/-1->19->18 [15] 20/-1/-1->19->18
+ip-26-0-161-103:96891:97063 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-168-30:235648:235826 [0] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-165-202:153562:153735 [0] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO Channel 00/0 : 86[6] -> 87[7] via P2P/IPC
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-161-78:114512:114686 [0] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO Channel 08/0 : 86[6] -> 87[7] via P2P/IPC
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO Channel 00/0 : 85[5] -> 86[6] via P2P/IPC
+ip-26-0-165-202:153568:153736 [6] NCCL INFO Channel 00/0 : 94[6] -> 95[7] via P2P/IPC
+ip-26-0-161-138:96568:96753 [5] NCCL INFO Channel 00/0 : 37[5] -> 38[6] via P2P/IPC
+ip-26-0-161-78:114518:114693 [6] NCCL INFO Channel 00/0 : 70[6] -> 71[7] via P2P/IPC
+ip-26-0-161-103:96893:97067 [5] NCCL INFO Channel 00/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO Channel 08/0 : 85[5] -> 86[6] via P2P/IPC
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-161-138:96563:96756 [0] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-161-78:114518:114693 [6] NCCL INFO Channel 08/0 : 70[6] -> 71[7] via P2P/IPC
+ip-26-0-161-142:62881:63059 [0] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-168-52:2722051:2722224 [6] NCCL INFO Channel 00/0 : 126[6] -> 127[7] via P2P/IPC
+ip-26-0-161-138:96568:96753 [5] NCCL INFO Channel 08/0 : 37[5] -> 38[6] via P2P/IPC
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO Channel 00/0 : 97[1] -> 98[2] via P2P/IPC
+ip-26-0-168-52:2722045:2722219 [0] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-161-103:96888:97066 [0] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-168-30:235652:235822 [4] NCCL INFO Channel 00/0 : 108[4] -> 109[5] via P2P/IPC
+ip-26-0-161-103:96893:97067 [5] NCCL INFO Channel 08/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-165-202:153568:153736 [6] NCCL INFO Channel 08/0 : 94[6] -> 95[7] via P2P/IPC
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO Channel 00/0 : 83[3] -> 84[4] via P2P/IPC
+ip-26-0-161-103:96892:97064 [4] NCCL INFO Channel 00/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO Channel 00/0 : 60[4] -> 61[5] via P2P/IPC
+ip-26-0-161-138:96567:96760 [4] NCCL INFO Channel 00/0 : 36[4] -> 37[5] via P2P/IPC
+ip-26-0-161-78:114517:114687 [5] NCCL INFO Channel 00/0 : 69[5] -> 70[6] via P2P/IPC
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO Channel 08/0 : 83[3] -> 84[4] via P2P/IPC
+ip-26-0-161-138:96566:96755 [3] NCCL INFO Channel 00/0 : 35[3] -> 36[4] via P2P/IPC
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO Channel 08/0 : 97[1] -> 98[2] via P2P/IPC
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-161-142:62886:63057 [5] NCCL INFO Channel 00/0 : 45[5] -> 46[6] via P2P/IPC
+ip-26-0-161-123:32708:32879 [4] NCCL INFO Channel 00/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-161-142:62887:63055 [6] NCCL INFO Channel 00/0 : 46[6] -> 47[7] via P2P/IPC
+ip-26-0-165-202:153564:153737 [2] NCCL INFO Channel 00/0 : 90[2] -> 91[3] via P2P/IPC
+ip-26-0-168-34:272150:272327 [0] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-165-202:153566:153741 [4] NCCL INFO Channel 00/0 : 92[4] -> 93[5] via P2P/IPC
+ip-26-0-161-103:96892:97064 [4] NCCL INFO Channel 08/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO Channel 00/0 : 73[1] -> 74[2] via P2P/IPC
+ip-26-0-161-138:96567:96760 [4] NCCL INFO Channel 08/0 : 36[4] -> 37[5] via P2P/IPC
+ip-26-0-161-138:96566:96755 [3] NCCL INFO Channel 08/0 : 35[3] -> 36[4] via P2P/IPC
+ip-26-0-168-52:2722051:2722224 [6] NCCL INFO Channel 08/0 : 126[6] -> 127[7] via P2P/IPC
+ip-26-0-161-123:32704:32881 [0] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO Channel 08/0 : 60[4] -> 61[5] via P2P/IPC
+ip-26-0-161-78:114517:114687 [5] NCCL INFO Channel 08/0 : 69[5] -> 70[6] via P2P/IPC
+ip-26-0-161-142:62886:63057 [5] NCCL INFO Channel 08/0 : 45[5] -> 46[6] via P2P/IPC
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO Channel 00/0 : 124[4] -> 125[5] via P2P/IPC
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO Channel 00/0 : 61[5] -> 62[6] via P2P/IPC
+ip-26-0-165-202:153564:153737 [2] NCCL INFO Channel 08/0 : 90[2] -> 91[3] via P2P/IPC
+ip-26-0-165-202:153566:153741 [4] NCCL INFO Channel 08/0 : 92[4] -> 93[5] via P2P/IPC
+ip-26-0-161-142:62887:63055 [6] NCCL INFO Channel 08/0 : 46[6] -> 47[7] via P2P/IPC
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO Channel 00/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-161-123:32708:32879 [4] NCCL INFO Channel 08/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO Channel 00/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-161-123:32706:32877 [2] NCCL INFO Channel 00/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-168-52:2722046:2722223 [1] NCCL INFO Channel 00/0 : 121[1] -> 122[2] via P2P/IPC
+ip-26-0-161-78:114514:114689 [2] NCCL INFO Channel 00/0 : 66[2] -> 67[3] via P2P/IPC
+ip-26-0-165-202:153565:153738 [3] NCCL INFO Channel 00/0 : 91[3] -> 92[4] via P2P/IPC
+ip-26-0-161-123:32710:32883 [6] NCCL INFO Channel 00/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-165-202:153567:153742 [5] NCCL INFO Channel 00/0 : 93[5] -> 94[6] via P2P/IPC
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO Channel 00/0 : 101[5] -> 102[6] via P2P/IPC
+ip-26-0-168-52:2722050:2722225 [5] NCCL INFO Channel 00/0 : 125[5] -> 126[6] via P2P/IPC
+ip-26-0-161-138:96567:96760 [4] NCCL INFO Channel 05/0 : 36[4] -> 39[7] via P2P/IPC
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO Channel 00/0 : 81[1] -> 82[2] via P2P/IPC
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO Channel 00/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO Channel 00/0 : 78[6] -> 79[7] via P2P/IPC
+ip-26-0-165-202:153563:153740 [1] NCCL INFO Channel 00/0 : 89[1] -> 90[2] via P2P/IPC
+ip-26-0-161-138:96564:96757 [1] NCCL INFO Channel 00/0 : 33[1] -> 34[2] via P2P/IPC
+ip-26-0-168-30:235652:235822 [4] NCCL INFO Channel 08/0 : 108[4] -> 109[5] via P2P/IPC
+ip-26-0-161-123:32709:32882 [5] NCCL INFO Channel 00/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO Channel 08/0 : 61[5] -> 62[6] via P2P/IPC
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO Channel 00/0 : 82[2] -> 83[3] via P2P/IPC
+ip-26-0-161-103:96894:97068 [6] NCCL INFO Channel 00/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO Channel 08/0 : 73[1] -> 74[2] via P2P/IPC
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO Channel 00/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO Channel 00/0 : 84[4] -> 85[5] via P2P/IPC
+ip-26-0-165-202:153567:153742 [5] NCCL INFO Channel 08/0 : 93[5] -> 94[6] via P2P/IPC
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO Channel 08/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO Channel 00/0 : 62[6] -> 63[7] via P2P/IPC
+ip-26-0-165-202:153565:153738 [3] NCCL INFO Channel 08/0 : 91[3] -> 92[4] via P2P/IPC
+ip-26-0-168-34:272154:272330 [4] NCCL INFO Channel 00/0 : 116[4] -> 117[5] via P2P/IPC
+ip-26-0-161-142:62884:63061 [3] NCCL INFO Channel 00/0 : 43[3] -> 44[4] via P2P/IPC
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO Channel 08/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-161-123:32706:32877 [2] NCCL INFO Channel 08/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-161-103:96889:97065 [1] NCCL INFO Channel 00/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO Channel 08/0 : 101[5] -> 102[6] via P2P/IPC
+ip-26-0-165-202:153563:153740 [1] NCCL INFO Channel 08/0 : 89[1] -> 90[2] via P2P/IPC
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO Channel 08/0 : 78[6] -> 79[7] via P2P/IPC
+ip-26-0-161-103:96894:97068 [6] NCCL INFO Channel 08/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-161-138:96567:96760 [4] NCCL INFO Channel 06/0 : 36[4] -> 39[7] via P2P/IPC
+ip-26-0-161-123:32710:32883 [6] NCCL INFO Channel 08/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO Channel 00/0 : 75[3] -> 76[4] via P2P/IPC
+ip-26-0-161-138:96564:96757 [1] NCCL INFO Channel 08/0 : 33[1] -> 34[2] via P2P/IPC
+ip-26-0-168-34:272156:272325 [6] NCCL INFO Channel 00/0 : 118[6] -> 119[7] via P2P/IPC
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO Channel 08/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-161-142:62883:63058 [2] NCCL INFO Channel 00/0 : 42[2] -> 43[3] via P2P/IPC
+ip-26-0-161-123:32709:32882 [5] NCCL INFO Channel 08/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO Channel 08/0 : 81[1] -> 82[2] via P2P/IPC
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO Channel 00/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-161-138:96569:96758 [6] NCCL INFO Channel 00/0 : 38[6] -> 39[7] via P2P/IPC
+ip-26-0-161-103:96891:97063 [3] NCCL INFO Channel 00/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO Channel 08/0 : 82[2] -> 83[3] via P2P/IPC
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO Channel 08/0 : 84[4] -> 85[5] via P2P/IPC
+ip-26-0-161-103:96890:97070 [2] NCCL INFO Channel 00/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO Channel 00/0 : 74[2] -> 75[3] via P2P/IPC
+ip-26-0-161-142:62885:63054 [4] NCCL INFO Channel 00/0 : 44[4] -> 45[5] via P2P/IPC
+ip-26-0-168-30:235653:235825 [5] NCCL INFO Channel 00/0 : 109[5] -> 110[6] via P2P/IPC
+ip-26-0-168-30:235649:235827 [1] NCCL INFO Channel 00/0 : 105[1] -> 106[2] via P2P/IPC
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO Channel 08/0 : 62[6] -> 63[7] via P2P/IPC
+ip-26-0-161-142:62884:63061 [3] NCCL INFO Channel 08/0 : 43[3] -> 44[4] via P2P/IPC
+ip-26-0-161-142:62882:63056 [1] NCCL INFO Channel 00/0 : 41[1] -> 42[2] via P2P/IPC
+ip-26-0-161-103:96889:97065 [1] NCCL INFO Channel 08/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO Channel 08/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO Channel 00/0 : 59[3] -> 60[4] via P2P/IPC
+ip-26-0-161-123:32705:32880 [1] NCCL INFO Channel 00/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO Channel 00/0 : 57[1] -> 58[2] via P2P/IPC
+ip-26-0-161-138:96565:96754 [2] NCCL INFO Channel 00/0 : 34[2] -> 35[3] via P2P/IPC
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO Channel 00/0 : 58[2] -> 59[3] via P2P/IPC
+ip-26-0-161-142:62883:63058 [2] NCCL INFO Channel 08/0 : 42[2] -> 43[3] via P2P/IPC
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO Channel 08/0 : 75[3] -> 76[4] via P2P/IPC
+ip-26-0-165-202:153566:153741 [4] NCCL INFO Channel 05/0 : 92[4] -> 95[7] via P2P/IPC
+ip-26-0-165-202:153564:153737 [2] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-168-34:272154:272330 [4] NCCL INFO Channel 08/0 : 116[4] -> 117[5] via P2P/IPC
+ip-26-0-165-202:153565:153738 [3] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO Channel 00/0 : 98[2] -> 99[3] via P2P/IPC
+ip-26-0-161-138:96567:96760 [4] NCCL INFO Channel 13/0 : 36[4] -> 39[7] via P2P/IPC
+ip-26-0-161-78:114515:114690 [3] NCCL INFO Channel 00/0 : 67[3] -> 68[4] via P2P/IPC
+ip-26-0-161-78:114514:114689 [2] NCCL INFO Channel 08/0 : 66[2] -> 67[3] via P2P/IPC
+ip-26-0-168-34:272153:272328 [3] NCCL INFO Channel 00/0 : 115[3] -> 116[4] via P2P/IPC
+ip-26-0-168-34:272152:272331 [2] NCCL INFO Channel 00/0 : 114[2] -> 115[3] via P2P/IPC
+ip-26-0-161-78:114516:114691 [4] NCCL INFO Channel 00/0 : 68[4] -> 69[5] via P2P/IPC
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO Channel 00/0 : 100[4] -> 101[5] via P2P/IPC
+ip-26-0-161-103:96891:97063 [3] NCCL INFO Channel 08/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO Channel 00/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-161-103:96890:97070 [2] NCCL INFO Channel 08/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-165-202:153566:153741 [4] NCCL INFO Channel 06/0 : 92[4] -> 95[7] via P2P/IPC
+ip-26-0-161-78:114513:114688 [1] NCCL INFO Channel 00/0 : 65[1] -> 66[2] via P2P/IPC
+ip-26-0-168-34:272156:272325 [6] NCCL INFO Channel 08/0 : 118[6] -> 119[7] via P2P/IPC
+ip-26-0-161-123:32705:32880 [1] NCCL INFO Channel 08/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO Channel 08/0 : 74[2] -> 75[3] via P2P/IPC
+ip-26-0-168-30:235653:235825 [5] NCCL INFO Channel 08/0 : 109[5] -> 110[6] via P2P/IPC
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO Channel 08/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-161-142:62885:63054 [4] NCCL INFO Channel 08/0 : 44[4] -> 45[5] via P2P/IPC
+ip-26-0-168-30:235649:235827 [1] NCCL INFO Channel 08/0 : 105[1] -> 106[2] via P2P/IPC
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO Channel 00/0 : 77[5] -> 78[6] via P2P/IPC
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO Channel 08/0 : 59[3] -> 60[4] via P2P/IPC
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO Channel 08/0 : 58[2] -> 59[3] via P2P/IPC
+ip-26-0-161-123:32707:32878 [3] NCCL INFO Channel 00/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO Channel 05/0 : 84[4] -> 87[7] via P2P/IPC
+ip-26-0-161-142:62882:63056 [1] NCCL INFO Channel 08/0 : 41[1] -> 42[2] via P2P/IPC
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO Channel 08/0 : 57[1] -> 58[2] via P2P/IPC
+ip-26-0-165-202:153566:153741 [4] NCCL INFO Channel 13/0 : 92[4] -> 95[7] via P2P/IPC
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO Channel 08/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO Channel 00/0 : 102[6] -> 103[7] via P2P/IPC
+ip-26-0-165-202:153566:153741 [4] NCCL INFO Channel 14/0 : 92[4] -> 95[7] via P2P/IPC
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO Channel 06/0 : 84[4] -> 87[7] via P2P/IPC
+ip-26-0-161-78:114516:114691 [4] NCCL INFO Channel 08/0 : 68[4] -> 69[5] via P2P/IPC
+ip-26-0-168-34:272152:272331 [2] NCCL INFO Channel 08/0 : 114[2] -> 115[3] via P2P/IPC
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO Channel 00/0 : 99[3] -> 100[4] via P2P/IPC
+ip-26-0-161-142:62884:63061 [3] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-168-30:235651:235823 [3] NCCL INFO Channel 00/0 : 107[3] -> 108[4] via P2P/IPC
+ip-26-0-161-78:114515:114690 [3] NCCL INFO Channel 08/0 : 67[3] -> 68[4] via P2P/IPC
+ip-26-0-168-34:272153:272328 [3] NCCL INFO Channel 08/0 : 115[3] -> 116[4] via P2P/IPC
+ip-26-0-161-78:114513:114688 [1] NCCL INFO Channel 08/0 : 65[1] -> 66[2] via P2P/IPC
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO Channel 08/0 : 98[2] -> 99[3] via P2P/IPC
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO Channel 08/0 : 100[4] -> 101[5] via P2P/IPC
+ip-26-0-161-103:96892:97064 [4] NCCL INFO Channel 05/0 : 20[4] -> 23[7] via P2P/IPC
+ip-26-0-161-123:32707:32878 [3] NCCL INFO Channel 08/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO Channel 08/0 : 77[5] -> 78[6] via P2P/IPC
+ip-26-0-161-142:62885:63054 [4] NCCL INFO Channel 05/0 : 44[4] -> 47[7] via P2P/IPC
+ip-26-0-161-142:62883:63058 [2] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-161-103:96891:97063 [3] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-161-103:96890:97070 [2] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-168-34:272155:272329 [5] NCCL INFO Channel 00/0 : 117[5] -> 118[6] via P2P/IPC
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO Channel 08/0 : 102[6] -> 103[7] via P2P/IPC
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO Channel 05/0 : 60[4] -> 63[7] via P2P/IPC
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO Channel 13/0 : 84[4] -> 87[7] via P2P/IPC
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-161-103:96892:97064 [4] NCCL INFO Channel 06/0 : 20[4] -> 23[7] via P2P/IPC
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO Channel 05/0 : 12[4] -> 15[7] via P2P/IPC
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-161-142:62885:63054 [4] NCCL INFO Channel 06/0 : 44[4] -> 47[7] via P2P/IPC
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-168-30:235650:235824 [2] NCCL INFO Channel 00/0 : 106[2] -> 107[3] via P2P/IPC
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO Channel 14/0 : 84[4] -> 87[7] via P2P/IPC
+ip-26-0-161-123:32706:32877 [2] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO Channel 00/0 : 76[4] -> 77[5] via P2P/IPC
+ip-26-0-168-30:235651:235823 [3] NCCL INFO Channel 08/0 : 107[3] -> 108[4] via P2P/IPC
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-161-103:96892:97064 [4] NCCL INFO Channel 13/0 : 20[4] -> 23[7] via P2P/IPC
+ip-26-0-161-123:32708:32879 [4] NCCL INFO Channel 05/0 : 28[4] -> 31[7] via P2P/IPC
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO Channel 08/0 : 99[3] -> 100[4] via P2P/IPC
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO Channel 06/0 : 60[4] -> 63[7] via P2P/IPC
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO Channel 06/0 : 12[4] -> 15[7] via P2P/IPC
+ip-26-0-161-123:32707:32878 [3] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-161-142:62885:63054 [4] NCCL INFO Channel 13/0 : 44[4] -> 47[7] via P2P/IPC
+ip-26-0-168-34:272155:272329 [5] NCCL INFO Channel 08/0 : 117[5] -> 118[6] via P2P/IPC
+ip-26-0-161-103:96892:97064 [4] NCCL INFO Channel 14/0 : 20[4] -> 23[7] via P2P/IPC
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO Channel 08/0 : 76[4] -> 77[5] via P2P/IPC
+ip-26-0-168-30:235650:235824 [2] NCCL INFO Channel 08/0 : 106[2] -> 107[3] via P2P/IPC
+ip-26-0-168-34:272153:272328 [3] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO Channel 13/0 : 60[4] -> 63[7] via P2P/IPC
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO Channel 13/0 : 12[4] -> 15[7] via P2P/IPC
+ip-26-0-161-123:32708:32879 [4] NCCL INFO Channel 06/0 : 28[4] -> 31[7] via P2P/IPC
+ip-26-0-161-142:62885:63054 [4] NCCL INFO Channel 14/0 : 44[4] -> 47[7] via P2P/IPC
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-168-30:235654:235821 [6] NCCL INFO Channel 00/0 : 110[6] -> 111[7] via P2P/IPC
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO Channel 14/0 : 60[4] -> 63[7] via P2P/IPC
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO Channel 14/0 : 12[4] -> 15[7] via P2P/IPC
+ip-26-0-161-123:32708:32879 [4] NCCL INFO Channel 13/0 : 28[4] -> 31[7] via P2P/IPC
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO Channel 05/0 : 100[4] -> 103[7] via P2P/IPC
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO Channel 00/0 : 54[6] -> 55[7] via P2P/IPC
+ip-26-0-168-34:272154:272330 [4] NCCL INFO Channel 05/0 : 116[4] -> 119[7] via P2P/IPC
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-168-30:235652:235822 [4] NCCL INFO Channel 05/0 : 108[4] -> 111[7] via P2P/IPC
+ip-26-0-161-123:32708:32879 [4] NCCL INFO Channel 14/0 : 28[4] -> 31[7] via P2P/IPC
+ip-26-0-168-30:235651:235823 [3] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-168-34:272151:272326 [1] NCCL INFO Channel 00/0 : 113[1] -> 114[2] via P2P/IPC
+ip-26-0-168-30:235650:235824 [2] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-168-30:235654:235821 [6] NCCL INFO Channel 08/0 : 110[6] -> 111[7] via P2P/IPC
+ip-26-0-168-34:272154:272330 [4] NCCL INFO Channel 06/0 : 116[4] -> 119[7] via P2P/IPC
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO Channel 06/0 : 100[4] -> 103[7] via P2P/IPC
+ip-26-0-168-30:235652:235822 [4] NCCL INFO Channel 06/0 : 108[4] -> 111[7] via P2P/IPC
+ip-26-0-161-78:114516:114691 [4] NCCL INFO Channel 05/0 : 68[4] -> 71[7] via P2P/IPC
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO Channel 13/0 : 100[4] -> 103[7] via P2P/IPC
+ip-26-0-161-78:114514:114689 [2] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-161-78:114515:114690 [3] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-168-30:235652:235822 [4] NCCL INFO Channel 13/0 : 108[4] -> 111[7] via P2P/IPC
+ip-26-0-168-52:2722046:2722223 [1] NCCL INFO Channel 08/0 : 121[1] -> 122[2] via P2P/IPC
+ip-26-0-168-34:272151:272326 [1] NCCL INFO Channel 08/0 : 113[1] -> 114[2] via P2P/IPC
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO Channel 00/0 : 51[3] -> 52[4] via P2P/IPC
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO Channel 08/0 : 124[4] -> 125[5] via P2P/IPC
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO Channel 14/0 : 100[4] -> 103[7] via P2P/IPC
+ip-26-0-168-34:272154:272330 [4] NCCL INFO Channel 13/0 : 116[4] -> 119[7] via P2P/IPC
+ip-26-0-161-78:114516:114691 [4] NCCL INFO Channel 06/0 : 68[4] -> 71[7] via P2P/IPC
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO Channel 08/0 : 54[6] -> 55[7] via P2P/IPC
+ip-26-0-168-30:235652:235822 [4] NCCL INFO Channel 14/0 : 108[4] -> 111[7] via P2P/IPC
+ip-26-0-168-52:2722047:2722221 [2] NCCL INFO Channel 00/0 : 122[2] -> 123[3] via P2P/IPC
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO Channel 00/0 : 52[4] -> 53[5] via P2P/IPC
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-168-34:272154:272330 [4] NCCL INFO Channel 14/0 : 116[4] -> 119[7] via P2P/IPC
+ip-26-0-168-52:2722050:2722225 [5] NCCL INFO Channel 08/0 : 125[5] -> 126[6] via P2P/IPC
+ip-26-0-161-78:114516:114691 [4] NCCL INFO Channel 13/0 : 68[4] -> 71[7] via P2P/IPC
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO Channel 05/0 : 76[4] -> 79[7] via P2P/IPC
+ip-26-0-168-34:272152:272331 [2] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO Channel 00/0 : 49[1] -> 50[2] via P2P/IPC
+ip-26-0-161-78:114516:114691 [4] NCCL INFO Channel 14/0 : 68[4] -> 71[7] via P2P/IPC
+ip-26-0-168-52:2722047:2722221 [2] NCCL INFO Channel 08/0 : 122[2] -> 123[3] via P2P/IPC
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO Channel 06/0 : 76[4] -> 79[7] via P2P/IPC
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO Channel 08/0 : 51[3] -> 52[4] via P2P/IPC
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO Channel 13/0 : 76[4] -> 79[7] via P2P/IPC
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO Channel 08/0 : 52[4] -> 53[5] via P2P/IPC
+ip-26-0-168-52:2722048:2722220 [3] NCCL INFO Channel 00/0 : 123[3] -> 124[4] via P2P/IPC
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO Channel 08/0 : 49[1] -> 50[2] via P2P/IPC
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO Channel 00/0 : 53[5] -> 54[6] via P2P/IPC
+ip-26-0-168-52:2722048:2722220 [3] NCCL INFO Channel 08/0 : 123[3] -> 124[4] via P2P/IPC
+ip-26-0-168-52:2722047:2722221 [2] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO Channel 05/0 : 124[4] -> 127[7] via P2P/IPC
+ip-26-0-161-138:96569:96758 [6] NCCL INFO Channel 08/0 : 38[6] -> 39[7] via P2P/IPC
+ip-26-0-161-138:96565:96754 [2] NCCL INFO Channel 08/0 : 34[2] -> 35[3] via P2P/IPC
+ip-26-0-161-138:96567:96760 [4] NCCL INFO Channel 14/0 : 36[4] -> 39[7] via P2P/IPC
+ip-26-0-161-138:96566:96755 [3] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-161-138:96565:96754 [2] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-168-52:2722048:2722220 [3] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO Channel 00/0 : 50[2] -> 51[3] via P2P/IPC
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO Channel 06/0 : 124[4] -> 127[7] via P2P/IPC
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO Channel 14/0 : 76[4] -> 79[7] via P2P/IPC
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO Channel 08/0 : 53[5] -> 54[6] via P2P/IPC
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO Channel 13/0 : 124[4] -> 127[7] via P2P/IPC
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO Channel 14/0 : 124[4] -> 127[7] via P2P/IPC
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO Channel 08/0 : 50[2] -> 51[3] via P2P/IPC
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO Channel 05/0 : 52[4] -> 55[7] via P2P/IPC
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO Channel 06/0 : 52[4] -> 55[7] via P2P/IPC
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO Channel 13/0 : 52[4] -> 55[7] via P2P/IPC
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO Channel 14/0 : 52[4] -> 55[7] via P2P/IPC
+ip-26-0-160-225:124162:124342 [2] NCCL INFO Trees [0] 3/-1/-1->2->1 [1] 3/-1/-1->2->1 [2] 3/66/-1->2->-1 [3] -1/-1/-1->2->1 [4] 3/-1/-1->2->1 [5] 3/-1/-1->2->1 [6] 3/-1/-1->2->1 [7] 3/-1/-1->2->1 [8] 3/-1/-1->2->1 [9] 3/-1/-1->2->1 [10] 3/-1/-1->2->10 [11] -1/-1/-1->2->1 [12] 3/-1/-1->2->1 [13] 3/-1/-1->2->1 [14] 3/-1/-1->2->1 [15] 3/-1/-1->2->1
+ip-26-0-160-225:124162:124342 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-225:124160:124335 [0] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-160-225:124166:124339 [6] NCCL INFO Channel 00/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-225:124167:124338 [7] NCCL INFO Channel 00/0 : 7[7] -> 8[0] [send] via NET/Libfabric/0(0)/GDRDMA
+ip-26-0-160-225:124165:124341 [5] NCCL INFO Channel 00/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-225:124163:124340 [3] NCCL INFO Channel 00/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-225:124164:124336 [4] NCCL INFO Channel 00/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-225:124166:124339 [6] NCCL INFO Channel 08/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-225:124161:124337 [1] NCCL INFO Channel 00/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-225:124162:124342 [2] NCCL INFO Channel 00/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-225:124165:124341 [5] NCCL INFO Channel 08/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-225:124163:124340 [3] NCCL INFO Channel 08/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-225:124164:124336 [4] NCCL INFO Channel 08/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-225:124161:124337 [1] NCCL INFO Channel 08/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-225:124162:124342 [2] NCCL INFO Channel 08/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-225:124164:124336 [4] NCCL INFO Channel 05/0 : 4[4] -> 7[7] via P2P/IPC
+ip-26-0-160-225:124163:124340 [3] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-160-225:124162:124342 [2] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-160-225:124164:124336 [4] NCCL INFO Channel 06/0 : 4[4] -> 7[7] via P2P/IPC
+ip-26-0-160-225:124164:124336 [4] NCCL INFO Channel 13/0 : 4[4] -> 7[7] via P2P/IPC
+ip-26-0-160-225:124164:124336 [4] NCCL INFO Channel 14/0 : 4[4] -> 7[7] via P2P/IPC
+ip-26-0-165-202:153562:153735 [0] NCCL INFO Channel 00/0 : 87[7] -> 88[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153569:153739 [7] NCCL INFO Channel 00/0 : 95[7] -> 96[0] [send] via NET/Libfabric/0(88)/GDRDMA
+ip-26-0-165-202:153562:153735 [0] NCCL INFO Channel 08/0 : 87[7] -> 88[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153562:153735 [0] NCCL INFO Channel 00/0 : 88[0] -> 89[1] via P2P/IPC
+ip-26-0-165-202:153569:153739 [7] NCCL INFO Channel 08/0 : 95[7] -> 96[0] [send] via NET/Libfabric/0(88)/GDRDMA
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO Channel 00/0 : 79[7] -> 80[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153562:153735 [0] NCCL INFO Channel 08/0 : 88[0] -> 89[1] via P2P/IPC
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO Channel 08/0 : 79[7] -> 80[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO Channel 00/0 : 80[0] -> 81[1] via P2P/IPC
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO Channel 00/0 : 87[7] -> 88[0] [send] via NET/Libfabric/0(80)/GDRDMA
+ip-26-0-161-142:62881:63059 [0] NCCL INFO Channel 00/0 : 39[7] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62888:63060 [7] NCCL INFO Channel 00/0 : 47[7] -> 48[0] [send] via NET/Libfabric/0(40)/GDRDMA
+ip-26-0-161-142:62881:63059 [0] NCCL INFO Channel 08/0 : 39[7] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62881:63059 [0] NCCL INFO Channel 00/0 : 40[0] -> 41[1] via P2P/IPC
+ip-26-0-161-142:62888:63060 [7] NCCL INFO Channel 08/0 : 47[7] -> 48[0] [send] via NET/Libfabric/0(40)/GDRDMA
+ip-26-0-165-202:153563:153740 [1] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO Channel 00/0 : 95[7] -> 96[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO Channel 08/0 : 80[0] -> 81[1] via P2P/IPC
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO Channel 08/0 : 87[7] -> 88[0] [send] via NET/Libfabric/0(80)/GDRDMA
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO Channel 08/0 : 95[7] -> 96[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96888:97066 [0] NCCL INFO Channel 00/0 : 15[7] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235648:235826 [0] NCCL INFO Channel 00/0 : 103[7] -> 104[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO Channel 00/0 : 96[0] -> 97[1] via P2P/IPC
+ip-26-0-161-103:96888:97066 [0] NCCL INFO Channel 08/0 : 15[7] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235648:235826 [0] NCCL INFO Channel 08/0 : 103[7] -> 104[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO Channel 00/0 : 103[7] -> 104[0] [send] via NET/Libfabric/0(96)/GDRDMA
+ip-26-0-161-103:96888:97066 [0] NCCL INFO Channel 00/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-168-30:235648:235826 [0] NCCL INFO Channel 00/0 : 104[0] -> 105[1] via P2P/IPC
+ip-26-0-161-103:96895:97069 [7] NCCL INFO Channel 00/0 : 23[7] -> 24[0] [send] via NET/Libfabric/0(16)/GDRDMA
+ip-26-0-168-30:235655:235828 [7] NCCL INFO Channel 00/0 : 111[7] -> 112[0] [send] via NET/Libfabric/0(104)/GDRDMA
+ip-26-0-165-202:153562:153735 [0] NCCL INFO Channel 05/0 : 88[0] -> 92[4] via P2P/IPC
+ip-26-0-165-202:153562:153735 [0] NCCL INFO Channel 13/0 : 88[0] -> 92[4] via P2P/IPC
+ip-26-0-161-103:96888:97066 [0] NCCL INFO Channel 08/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-168-30:235648:235826 [0] NCCL INFO Channel 08/0 : 104[0] -> 105[1] via P2P/IPC
+ip-26-0-168-30:235655:235828 [7] NCCL INFO Channel 08/0 : 111[7] -> 112[0] [send] via NET/Libfabric/0(104)/GDRDMA
+ip-26-0-161-103:96895:97069 [7] NCCL INFO Channel 08/0 : 23[7] -> 24[0] [send] via NET/Libfabric/0(16)/GDRDMA
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO Channel 08/0 : 96[0] -> 97[1] via P2P/IPC
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO Channel 08/0 : 103[7] -> 104[0] [send] via NET/Libfabric/0(96)/GDRDMA
+ip-26-0-161-103:96889:97065 [1] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-161-142:62881:63059 [0] NCCL INFO Channel 08/0 : 40[0] -> 41[1] via P2P/IPC
+ip-26-0-168-30:235649:235827 [1] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-168-30:235648:235826 [0] NCCL INFO Channel 05/0 : 104[0] -> 108[4] via P2P/IPC
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO Channel 05/0 : 96[0] -> 100[4] via P2P/IPC
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-168-30:235648:235826 [0] NCCL INFO Channel 13/0 : 104[0] -> 108[4] via P2P/IPC
+ip-26-0-161-138:96563:96756 [0] NCCL INFO Channel 00/0 : 31[7] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114512:114686 [0] NCCL INFO Channel 00/0 : 63[7] -> 64[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96563:96756 [0] NCCL INFO Channel 08/0 : 31[7] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114519:114692 [7] NCCL INFO Channel 00/0 : 71[7] -> 72[0] [send] via NET/Libfabric/0(64)/GDRDMA
+ip-26-0-161-142:62882:63056 [1] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-161-78:114512:114686 [0] NCCL INFO Channel 08/0 : 63[7] -> 64[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96563:96756 [0] NCCL INFO Channel 00/0 : 32[0] -> 33[1] via P2P/IPC
+ip-26-0-161-78:114512:114686 [0] NCCL INFO Channel 00/0 : 64[0] -> 65[1] via P2P/IPC
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO Channel 13/0 : 96[0] -> 100[4] via P2P/IPC
+ip-26-0-161-138:96570:96759 [7] NCCL INFO Channel 00/0 : 39[7] -> 40[0] [send] via NET/Libfabric/0(32)/GDRDMA
+ip-26-0-161-78:114519:114692 [7] NCCL INFO Channel 08/0 : 71[7] -> 72[0] [send] via NET/Libfabric/0(64)/GDRDMA
+ip-26-0-161-138:96563:96756 [0] NCCL INFO Channel 08/0 : 32[0] -> 33[1] via P2P/IPC
+ip-26-0-161-78:114512:114686 [0] NCCL INFO Channel 08/0 : 64[0] -> 65[1] via P2P/IPC
+ip-26-0-161-138:96570:96759 [7] NCCL INFO Channel 08/0 : 39[7] -> 40[0] [send] via NET/Libfabric/0(32)/GDRDMA
+ip-26-0-168-52:2722045:2722219 [0] NCCL INFO Channel 00/0 : 119[7] -> 120[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO Channel 00/0 : 55[7] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-52:2722052:2722226 [7] NCCL INFO Channel 00/0 : 127[7] -> 0[0] [send] via NET/Libfabric/0(120)/GDRDMA
+ip-26-0-161-142:62881:63059 [0] NCCL INFO Channel 05/0 : 40[0] -> 44[4] via P2P/IPC
+ip-26-0-168-52:2722045:2722219 [0] NCCL INFO Channel 08/0 : 119[7] -> 120[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO Channel 00/0 : 63[7] -> 64[0] [send] via NET/Libfabric/0(56)/GDRDMA
+ip-26-0-168-52:2722045:2722219 [0] NCCL INFO Channel 00/0 : 120[0] -> 121[1] via P2P/IPC
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO Channel 08/0 : 55[7] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-52:2722052:2722226 [7] NCCL INFO Channel 08/0 : 127[7] -> 0[0] [send] via NET/Libfabric/0(120)/GDRDMA
+ip-26-0-161-78:114513:114688 [1] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO Channel 00/0 : 56[0] -> 57[1] via P2P/IPC
+ip-26-0-161-142:62881:63059 [0] NCCL INFO Channel 13/0 : 40[0] -> 44[4] via P2P/IPC
+ip-26-0-161-138:96564:96757 [1] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO Channel 08/0 : 63[7] -> 64[0] [send] via NET/Libfabric/0(56)/GDRDMA
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO Channel 06/0 : 96[0] -> 101[5] via P2P/IPC
+ip-26-0-165-202:153562:153735 [0] NCCL INFO Channel 06/0 : 88[0] -> 93[5] via P2P/IPC
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-165-202:153566:153741 [4] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO Channel 00/0 : 71[7] -> 72[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO Channel 00/0 : 79[7] -> 80[0] [send] via NET/Libfabric/0(72)/GDRDMA
+ip-26-0-161-78:114512:114686 [0] NCCL INFO Channel 05/0 : 64[0] -> 68[4] via P2P/IPC
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO Channel 08/0 : 71[7] -> 72[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO Channel 00/0 : 72[0] -> 73[1] via P2P/IPC
+ip-26-0-168-52:2722045:2722219 [0] NCCL INFO Channel 08/0 : 120[0] -> 121[1] via P2P/IPC
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO Channel 08/0 : 79[7] -> 80[0] [send] via NET/Libfabric/0(72)/GDRDMA
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO Channel 14/0 : 96[0] -> 101[5] via P2P/IPC
+ip-26-0-165-202:153562:153735 [0] NCCL INFO Channel 14/0 : 88[0] -> 93[5] via P2P/IPC
+ip-26-0-161-78:114512:114686 [0] NCCL INFO Channel 13/0 : 64[0] -> 68[4] via P2P/IPC
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO Channel 08/0 : 56[0] -> 57[1] via P2P/IPC
+ip-26-0-161-142:62883:63058 [2] NCCL INFO Channel 02/0 : 35[3] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO Channel 05/0 : 80[0] -> 84[4] via P2P/IPC
+ip-26-0-168-34:272150:272327 [0] NCCL INFO Channel 00/0 : 111[7] -> 112[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62883:63058 [2] NCCL INFO Channel 10/0 : 35[3] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272150:272327 [0] NCCL INFO Channel 08/0 : 111[7] -> 112[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96890:97070 [2] NCCL INFO Channel 02/0 : 11[3] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153562:153735 [0] NCCL INFO Channel 07/0 : 88[0] -> 94[6] via P2P/IPC
+ip-26-0-168-34:272150:272327 [0] NCCL INFO Channel 00/0 : 112[0] -> 113[1] via P2P/IPC
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO Channel 13/0 : 80[0] -> 84[4] via P2P/IPC
+ip-26-0-161-103:96890:97070 [2] NCCL INFO Channel 10/0 : 11[3] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272157:272324 [7] NCCL INFO Channel 00/0 : 119[7] -> 120[0] [send] via NET/Libfabric/0(112)/GDRDMA
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO Channel 03/0 : 76[4] -> 83[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-52:2722046:2722223 [1] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-161-123:32704:32881 [0] NCCL INFO Channel 00/0 : 23[7] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32711:32884 [7] NCCL INFO Channel 00/0 : 31[7] -> 32[0] [send] via NET/Libfabric/0(24)/GDRDMA
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO Channel 11/0 : 76[4] -> 83[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32704:32881 [0] NCCL INFO Channel 08/0 : 23[7] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153562:153735 [0] NCCL INFO Channel 15/0 : 88[0] -> 94[6] via P2P/IPC
+ip-26-0-161-123:32704:32881 [0] NCCL INFO Channel 00/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-161-123:32711:32884 [7] NCCL INFO Channel 08/0 : 31[7] -> 32[0] [send] via NET/Libfabric/0(24)/GDRDMA
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO Channel 08/0 : 72[0] -> 73[1] via P2P/IPC
+ip-26-0-161-138:96565:96754 [2] NCCL INFO Channel 02/0 : 27[3] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96565:96754 [2] NCCL INFO Channel 10/0 : 27[3] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153567:153742 [5] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-168-34:272150:272327 [0] NCCL INFO Channel 08/0 : 112[0] -> 113[1] via P2P/IPC
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO Channel 00/0 : 7[7] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153562:153735 [0] NCCL INFO Channel 01/0 : 88[0] -> 95[7] via P2P/IPC
+ip-26-0-165-202:153568:153736 [6] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO Channel 07/0 : 96[0] -> 102[6] via P2P/IPC
+ip-26-0-168-34:272157:272324 [7] NCCL INFO Channel 08/0 : 119[7] -> 120[0] [send] via NET/Libfabric/0(112)/GDRDMA
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO Channel 08/0 : 7[7] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO Channel 00/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO Channel 00/0 : 15[7] -> 16[0] [send] via NET/Libfabric/0(8)/GDRDMA
+ip-26-0-165-202:153562:153735 [0] NCCL INFO Channel 02/0 : 88[0] -> 95[7] via P2P/IPC
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO Channel 15/0 : 96[0] -> 102[6] via P2P/IPC
+ip-26-0-168-52:2722045:2722219 [0] NCCL INFO Channel 05/0 : 120[0] -> 124[4] via P2P/IPC
+ip-26-0-160-225:124160:124335 [0] NCCL INFO Channel 00/0 : 127[7] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153562:153735 [0] NCCL INFO Channel 03/0 : 88[0] -> 95[7] via P2P/IPC
+ip-26-0-160-225:124167:124338 [7] NCCL INFO Channel 08/0 : 7[7] -> 8[0] [send] via NET/Libfabric/0(0)/GDRDMA
+ip-26-0-160-225:124160:124335 [0] NCCL INFO Channel 08/0 : 127[7] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:124160:124335 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-225:124160:124335 [0] NCCL INFO Channel 08/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-225:124161:124337 [1] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-160-225:124160:124335 [0] NCCL INFO Channel 05/0 : 0[0] -> 4[4] via P2P/IPC
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-160-225:124160:124335 [0] NCCL INFO Channel 13/0 : 0[0] -> 4[4] via P2P/IPC
+ip-26-0-161-123:32704:32881 [0] NCCL INFO Channel 08/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO Channel 05/0 : 72[0] -> 76[4] via P2P/IPC
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO Channel 06/0 : 80[0] -> 85[5] via P2P/IPC
+ip-26-0-160-225:124163:124340 [3] NCCL INFO Channel 03/0 : 124[4] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153562:153735 [0] NCCL INFO Channel 04/0 : 88[0] -> 95[7] via P2P/IPC
+ip-26-0-168-34:272150:272327 [0] NCCL INFO Channel 05/0 : 112[0] -> 116[4] via P2P/IPC
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO Channel 01/0 : 96[0] -> 103[7] via P2P/IPC
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-160-225:124163:124340 [3] NCCL INFO Channel 11/0 : 124[4] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO Channel 13/0 : 72[0] -> 76[4] via P2P/IPC
+ip-26-0-165-202:153564:153737 [2] NCCL INFO Channel 02/0 : 83[3] -> 90[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO Channel 08/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-165-202:153564:153737 [2] NCCL INFO Channel 10/0 : 83[3] -> 90[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO Channel 14/0 : 80[0] -> 85[5] via P2P/IPC
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO Channel 08/0 : 15[7] -> 16[0] [send] via NET/Libfabric/0(8)/GDRDMA
+ip-26-0-165-202:153562:153735 [0] NCCL INFO Channel 09/0 : 88[0] -> 95[7] via P2P/IPC
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO Channel 06/0 : 87[7] -> 94[6] [send] via NET/Libfabric/6(86)/GDRDMA
+ip-26-0-161-138:96563:96756 [0] NCCL INFO Channel 05/0 : 32[0] -> 36[4] via P2P/IPC
+ip-26-0-161-142:62884:63061 [3] NCCL INFO Channel 03/0 : 36[4] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153565:153738 [3] NCCL INFO Channel 03/0 : 84[4] -> 91[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO Channel 02/0 : 96[0] -> 103[7] via P2P/IPC
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO Channel 14/0 : 87[7] -> 94[6] [send] via NET/Libfabric/6(86)/GDRDMA
+ip-26-0-168-34:272151:272326 [1] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-165-202:153562:153735 [0] NCCL INFO Channel 10/0 : 88[0] -> 95[7] via P2P/IPC
+ip-26-0-161-142:62884:63061 [3] NCCL INFO Channel 11/0 : 36[4] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32705:32880 [1] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-165-202:153565:153738 [3] NCCL INFO Channel 11/0 : 84[4] -> 91[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272150:272327 [0] NCCL INFO Channel 13/0 : 112[0] -> 116[4] via P2P/IPC
+ip-26-0-161-103:96891:97063 [3] NCCL INFO Channel 03/0 : 12[4] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32704:32881 [0] NCCL INFO Channel 05/0 : 24[0] -> 28[4] via P2P/IPC
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO Channel 03/0 : 96[0] -> 103[7] via P2P/IPC
+ip-26-0-161-138:96566:96755 [3] NCCL INFO Channel 03/0 : 28[4] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96888:97066 [0] NCCL INFO Channel 05/0 : 16[0] -> 20[4] via P2P/IPC
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO Channel 02/0 : 75[3] -> 82[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153562:153735 [0] NCCL INFO Channel 11/0 : 88[0] -> 95[7] via P2P/IPC
+ip-26-0-161-138:96563:96756 [0] NCCL INFO Channel 13/0 : 32[0] -> 36[4] via P2P/IPC
+ip-26-0-161-103:96891:97063 [3] NCCL INFO Channel 11/0 : 12[4] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO Channel 07/0 : 80[0] -> 86[6] via P2P/IPC
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-161-142:62884:63061 [3] NCCL INFO Channel 02/0 : 43[3] -> 50[2] [send] via NET/Libfabric/2(42)/GDRDMA
+ip-26-0-165-202:153565:153738 [3] NCCL INFO Channel 02/0 : 91[3] -> 98[2] [send] via NET/Libfabric/2(90)/GDRDMA
+ip-26-0-161-138:96566:96755 [3] NCCL INFO Channel 11/0 : 28[4] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO Channel 10/0 : 75[3] -> 82[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-52:2722045:2722219 [0] NCCL INFO Channel 13/0 : 120[0] -> 124[4] via P2P/IPC
+ip-26-0-165-202:153565:153738 [3] NCCL INFO Channel 10/0 : 91[3] -> 98[2] [send] via NET/Libfabric/2(90)/GDRDMA
+ip-26-0-161-142:62884:63061 [3] NCCL INFO Channel 10/0 : 43[3] -> 50[2] [send] via NET/Libfabric/2(42)/GDRDMA
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO Channel 05/0 : 8[0] -> 12[4] via P2P/IPC
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO Channel 02/0 : 83[3] -> 90[2] [send] via NET/Libfabric/2(82)/GDRDMA
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO Channel 04/0 : 96[0] -> 103[7] via P2P/IPC
+ip-26-0-165-202:153562:153735 [0] NCCL INFO Channel 12/0 : 88[0] -> 95[7] via P2P/IPC
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO Channel 15/0 : 80[0] -> 86[6] via P2P/IPC
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO Channel 10/0 : 83[3] -> 90[2] [send] via NET/Libfabric/2(82)/GDRDMA
+ip-26-0-161-103:96891:97063 [3] NCCL INFO Channel 02/0 : 19[3] -> 26[2] [send] via NET/Libfabric/2(18)/GDRDMA
+ip-26-0-161-123:32704:32881 [0] NCCL INFO Channel 13/0 : 24[0] -> 28[4] via P2P/IPC
+ip-26-0-161-138:96566:96755 [3] NCCL INFO Channel 02/0 : 35[3] -> 42[2] [send] via NET/Libfabric/2(34)/GDRDMA
+ip-26-0-161-103:96888:97066 [0] NCCL INFO Channel 13/0 : 16[0] -> 20[4] via P2P/IPC
+ip-26-0-161-138:96563:96756 [0] NCCL INFO Channel 06/0 : 32[0] -> 37[5] via P2P/IPC
+ip-26-0-161-103:96891:97063 [3] NCCL INFO Channel 10/0 : 19[3] -> 26[2] [send] via NET/Libfabric/2(18)/GDRDMA
+ip-26-0-165-202:153569:153739 [7] NCCL INFO Channel 06/0 : 95[7] -> 102[6] [send] via NET/Libfabric/6(94)/GDRDMA
+ip-26-0-161-138:96566:96755 [3] NCCL INFO Channel 10/0 : 35[3] -> 42[2] [send] via NET/Libfabric/2(34)/GDRDMA
+ip-26-0-161-138:96567:96760 [4] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO Channel 13/0 : 8[0] -> 12[4] via P2P/IPC
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO Channel 06/0 : 72[0] -> 77[5] via P2P/IPC
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO Channel 01/0 : 80[0] -> 87[7] via P2P/IPC
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-168-52:2722045:2722219 [0] NCCL INFO Channel 06/0 : 120[0] -> 125[5] via P2P/IPC
+ip-26-0-161-138:96563:96756 [0] NCCL INFO Channel 14/0 : 32[0] -> 37[5] via P2P/IPC
+ip-26-0-168-52:2722045:2722219 [0] NCCL INFO Channel 14/0 : 120[0] -> 125[5] via P2P/IPC
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO Channel 02/0 : 80[0] -> 87[7] via P2P/IPC
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO Channel 14/0 : 72[0] -> 77[5] via P2P/IPC
+ip-26-0-161-138:96563:96756 [0] NCCL INFO Channel 07/0 : 32[0] -> 38[6] via P2P/IPC
+ip-26-0-161-138:96568:96753 [5] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO Channel 03/0 : 80[0] -> 87[7] via P2P/IPC
+ip-26-0-161-138:96563:96756 [0] NCCL INFO Channel 15/0 : 32[0] -> 38[6] via P2P/IPC
+ip-26-0-161-78:114512:114686 [0] NCCL INFO Channel 06/0 : 64[0] -> 69[5] via P2P/IPC
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO Channel 02/0 : 91[3] -> 98[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114516:114691 [4] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO Channel 02/0 : 51[3] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO Channel 04/0 : 80[0] -> 87[7] via P2P/IPC
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO Channel 07/0 : 72[0] -> 78[6] via P2P/IPC
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO Channel 10/0 : 91[3] -> 98[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO Channel 02/0 : 3[3] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO Channel 10/0 : 51[3] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96563:96756 [0] NCCL INFO Channel 01/0 : 32[0] -> 39[7] via P2P/IPC
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO Channel 10/0 : 3[3] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO Channel 09/0 : 96[0] -> 103[7] via P2P/IPC
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO Channel 09/0 : 80[0] -> 87[7] via P2P/IPC
+ip-26-0-161-138:96569:96758 [6] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-161-78:114512:114686 [0] NCCL INFO Channel 14/0 : 64[0] -> 69[5] via P2P/IPC
+ip-26-0-168-34:272154:272330 [4] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-168-34:272150:272327 [0] NCCL INFO Channel 06/0 : 112[0] -> 117[5] via P2P/IPC
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO Channel 15/0 : 72[0] -> 78[6] via P2P/IPC
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO Channel 10/0 : 96[0] -> 103[7] via P2P/IPC
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO Channel 10/0 : 80[0] -> 87[7] via P2P/IPC
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-161-138:96563:96756 [0] NCCL INFO Channel 02/0 : 32[0] -> 39[7] via P2P/IPC
+ip-26-0-161-103:96888:97066 [0] NCCL INFO Channel 06/0 : 16[0] -> 21[5] via P2P/IPC
+ip-26-0-161-103:96892:97064 [4] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-161-123:32708:32879 [4] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO Channel 11/0 : 96[0] -> 103[7] via P2P/IPC
+ip-26-0-161-78:114512:114686 [0] NCCL INFO Channel 07/0 : 64[0] -> 70[6] via P2P/IPC
+ip-26-0-168-34:272150:272327 [0] NCCL INFO Channel 14/0 : 112[0] -> 117[5] via P2P/IPC
+ip-26-0-161-123:32704:32881 [0] NCCL INFO Channel 06/0 : 24[0] -> 29[5] via P2P/IPC
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO Channel 11/0 : 80[0] -> 87[7] via P2P/IPC
+ip-26-0-161-138:96563:96756 [0] NCCL INFO Channel 03/0 : 32[0] -> 39[7] via P2P/IPC
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO Channel 06/0 : 63[7] -> 70[6] [send] via NET/Libfabric/6(62)/GDRDMA
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO Channel 01/0 : 72[0] -> 79[7] via P2P/IPC
+ip-26-0-161-78:114517:114687 [5] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-168-52:2722050:2722225 [5] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-168-52:2722045:2722219 [0] NCCL INFO Channel 07/0 : 120[0] -> 126[6] via P2P/IPC
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO Channel 03/0 : 52[4] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO Channel 03/0 : 4[4] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96570:96759 [7] NCCL INFO Channel 06/0 : 39[7] -> 46[6] [send] via NET/Libfabric/6(38)/GDRDMA
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO Channel 12/0 : 96[0] -> 103[7] via P2P/IPC
+ip-26-0-161-103:96888:97066 [0] NCCL INFO Channel 14/0 : 16[0] -> 21[5] via P2P/IPC
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO Channel 11/0 : 52[4] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO Channel 12/0 : 80[0] -> 87[7] via P2P/IPC
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO Channel 11/0 : 4[4] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96563:96756 [0] NCCL INFO Channel 04/0 : 32[0] -> 39[7] via P2P/IPC
+ip-26-0-161-123:32704:32881 [0] NCCL INFO Channel 14/0 : 24[0] -> 29[5] via P2P/IPC
+ip-26-0-160-225:124160:124335 [0] NCCL INFO Channel 06/0 : 0[0] -> 5[5] via P2P/IPC
+ip-26-0-160-225:124164:124336 [4] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-168-52:2722045:2722219 [0] NCCL INFO Channel 15/0 : 120[0] -> 126[6] via P2P/IPC
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO Channel 02/0 : 59[3] -> 66[2] [send] via NET/Libfabric/2(58)/GDRDMA
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO Channel 02/0 : 72[0] -> 79[7] via P2P/IPC
+ip-26-0-161-78:114512:114686 [0] NCCL INFO Channel 15/0 : 64[0] -> 70[6] via P2P/IPC
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO Channel 02/0 : 11[3] -> 18[2] [send] via NET/Libfabric/2(10)/GDRDMA
+ip-26-0-161-138:96563:96756 [0] NCCL INFO Channel 09/0 : 32[0] -> 39[7] via P2P/IPC
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO Channel 10/0 : 59[3] -> 66[2] [send] via NET/Libfabric/2(58)/GDRDMA
+ip-26-0-168-34:272150:272327 [0] NCCL INFO Channel 07/0 : 112[0] -> 118[6] via P2P/IPC
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO Channel 14/0 : 63[7] -> 70[6] [send] via NET/Libfabric/6(62)/GDRDMA
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO Channel 10/0 : 11[3] -> 18[2] [send] via NET/Libfabric/2(10)/GDRDMA
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO Channel 02/0 : 67[3] -> 74[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272153:272328 [3] NCCL INFO Channel 03/0 : 108[4] -> 115[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO Channel 03/0 : 72[0] -> 79[7] via P2P/IPC
+ip-26-0-168-34:272155:272329 [5] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-161-103:96888:97066 [0] NCCL INFO Channel 07/0 : 16[0] -> 22[6] via P2P/IPC
+ip-26-0-161-123:32704:32881 [0] NCCL INFO Channel 07/0 : 24[0] -> 30[6] via P2P/IPC
+ip-26-0-160-225:124162:124342 [2] NCCL INFO Channel 02/0 : 123[3] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96563:96756 [0] NCCL INFO Channel 10/0 : 32[0] -> 39[7] via P2P/IPC
+ip-26-0-168-34:272153:272328 [3] NCCL INFO Channel 11/0 : 108[4] -> 115[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96893:97067 [5] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO Channel 10/0 : 67[3] -> 74[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:124160:124335 [0] NCCL INFO Channel 14/0 : 0[0] -> 5[5] via P2P/IPC
+ip-26-0-168-52:2722045:2722219 [0] NCCL INFO Channel 01/0 : 120[0] -> 127[7] via P2P/IPC
+ip-26-0-168-30:235648:235826 [0] NCCL INFO Channel 06/0 : 104[0] -> 109[5] via P2P/IPC
+ip-26-0-161-123:32709:32882 [5] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-160-225:124162:124342 [2] NCCL INFO Channel 10/0 : 123[3] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114512:114686 [0] NCCL INFO Channel 01/0 : 64[0] -> 71[7] via P2P/IPC
+ip-26-0-168-34:272150:272327 [0] NCCL INFO Channel 15/0 : 112[0] -> 118[6] via P2P/IPC
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-160-225:124163:124340 [3] NCCL INFO Channel 02/0 : 3[3] -> 10[2] [send] via NET/Libfabric/2(2)/GDRDMA
+ip-26-0-168-52:2722051:2722224 [6] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-168-30:235652:235822 [4] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO Channel 04/0 : 72[0] -> 79[7] via P2P/IPC
+ip-26-0-161-138:96563:96756 [0] NCCL INFO Channel 11/0 : 32[0] -> 39[7] via P2P/IPC
+ip-26-0-168-30:235651:235823 [3] NCCL INFO Channel 03/0 : 100[4] -> 107[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:124163:124340 [3] NCCL INFO Channel 10/0 : 3[3] -> 10[2] [send] via NET/Libfabric/2(2)/GDRDMA
+ip-26-0-161-103:96888:97066 [0] NCCL INFO Channel 15/0 : 16[0] -> 22[6] via P2P/IPC
+ip-26-0-161-123:32704:32881 [0] NCCL INFO Channel 15/0 : 24[0] -> 30[6] via P2P/IPC
+ip-26-0-168-30:235651:235823 [3] NCCL INFO Channel 11/0 : 100[4] -> 107[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235648:235826 [0] NCCL INFO Channel 14/0 : 104[0] -> 109[5] via P2P/IPC
+ip-26-0-168-52:2722045:2722219 [0] NCCL INFO Channel 02/0 : 120[0] -> 127[7] via P2P/IPC
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO Channel 09/0 : 72[0] -> 79[7] via P2P/IPC
+ip-26-0-161-78:114518:114693 [6] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-161-78:114512:114686 [0] NCCL INFO Channel 02/0 : 64[0] -> 71[7] via P2P/IPC
+ip-26-0-161-138:96563:96756 [0] NCCL INFO Channel 12/0 : 32[0] -> 39[7] via P2P/IPC
+ip-26-0-168-34:272156:272325 [6] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-168-34:272150:272327 [0] NCCL INFO Channel 01/0 : 112[0] -> 119[7] via P2P/IPC
+ip-26-0-160-225:124165:124341 [5] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO Channel 10/0 : 72[0] -> 79[7] via P2P/IPC
+ip-26-0-160-225:124160:124335 [0] NCCL INFO Channel 07/0 : 0[0] -> 6[6] via P2P/IPC
+ip-26-0-168-52:2722045:2722219 [0] NCCL INFO Channel 03/0 : 120[0] -> 127[7] via P2P/IPC
+ip-26-0-161-78:114512:114686 [0] NCCL INFO Channel 03/0 : 64[0] -> 71[7] via P2P/IPC
+ip-26-0-161-103:96888:97066 [0] NCCL INFO Channel 01/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-161-123:32710:32883 [6] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-161-123:32704:32881 [0] NCCL INFO Channel 01/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-168-52:2722045:2722219 [0] NCCL INFO Channel 04/0 : 120[0] -> 127[7] via P2P/IPC
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO Channel 03/0 : 92[4] -> 99[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32707:32878 [3] NCCL INFO Channel 03/0 : 20[4] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO Channel 11/0 : 72[0] -> 79[7] via P2P/IPC
+ip-26-0-161-78:114512:114686 [0] NCCL INFO Channel 04/0 : 64[0] -> 71[7] via P2P/IPC
+ip-26-0-160-225:124160:124335 [0] NCCL INFO Channel 15/0 : 0[0] -> 6[6] via P2P/IPC
+ip-26-0-168-34:272150:272327 [0] NCCL INFO Channel 02/0 : 112[0] -> 119[7] via P2P/IPC
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO Channel 01/0 : 10[2] -> 17[1] [send] via NET/Libfabric/1(9)/GDRDMA
+ip-26-0-161-123:32707:32878 [3] NCCL INFO Channel 11/0 : 20[4] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO Channel 11/0 : 92[4] -> 99[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96888:97066 [0] NCCL INFO Channel 02/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-168-52:2722045:2722219 [0] NCCL INFO Channel 09/0 : 120[0] -> 127[7] via P2P/IPC
+ip-26-0-161-78:114514:114689 [2] NCCL INFO Channel 02/0 : 59[3] -> 66[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO Channel 12/0 : 72[0] -> 79[7] via P2P/IPC
+ip-26-0-161-78:114512:114686 [0] NCCL INFO Channel 09/0 : 64[0] -> 71[7] via P2P/IPC
+ip-26-0-168-34:272150:272327 [0] NCCL INFO Channel 03/0 : 112[0] -> 119[7] via P2P/IPC
+ip-26-0-161-78:114514:114689 [2] NCCL INFO Channel 10/0 : 59[3] -> 66[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96888:97066 [0] NCCL INFO Channel 03/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO Channel 02/0 : 99[3] -> 106[2] [send] via NET/Libfabric/2(98)/GDRDMA
+ip-26-0-161-123:32704:32881 [0] NCCL INFO Channel 02/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-168-52:2722045:2722219 [0] NCCL INFO Channel 10/0 : 120[0] -> 127[7] via P2P/IPC
+ip-26-0-160-225:124160:124335 [0] NCCL INFO Channel 01/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-160-225:124166:124339 [6] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-161-78:114512:114686 [0] NCCL INFO Channel 10/0 : 64[0] -> 71[7] via P2P/IPC
+ip-26-0-161-123:32706:32877 [2] NCCL INFO Channel 02/0 : 19[3] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO Channel 10/0 : 99[3] -> 106[2] [send] via NET/Libfabric/2(98)/GDRDMA
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO Channel 03/0 : 68[4] -> 75[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-52:2722052:2722226 [7] NCCL INFO Channel 06/0 : 127[7] -> 6[6] [send] via NET/Libfabric/6(126)/GDRDMA
+ip-26-0-168-34:272150:272327 [0] NCCL INFO Channel 04/0 : 112[0] -> 119[7] via P2P/IPC
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO Channel 06/0 : 79[7] -> 86[6] [send] via NET/Libfabric/6(78)/GDRDMA
+ip-26-0-161-103:96888:97066 [0] NCCL INFO Channel 04/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-161-123:32706:32877 [2] NCCL INFO Channel 10/0 : 19[3] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO Channel 11/0 : 68[4] -> 75[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32707:32878 [3] NCCL INFO Channel 02/0 : 27[3] -> 34[2] [send] via NET/Libfabric/2(26)/GDRDMA
+ip-26-0-161-78:114512:114686 [0] NCCL INFO Channel 11/0 : 64[0] -> 71[7] via P2P/IPC
+ip-26-0-168-52:2722045:2722219 [0] NCCL INFO Channel 11/0 : 120[0] -> 127[7] via P2P/IPC
+ip-26-0-161-123:32704:32881 [0] NCCL INFO Channel 03/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-160-225:124160:124335 [0] NCCL INFO Channel 02/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-168-34:272150:272327 [0] NCCL INFO Channel 09/0 : 112[0] -> 119[7] via P2P/IPC
+ip-26-0-161-103:96888:97066 [0] NCCL INFO Channel 09/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-161-123:32707:32878 [3] NCCL INFO Channel 10/0 : 27[3] -> 34[2] [send] via NET/Libfabric/2(26)/GDRDMA
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO Channel 02/0 : 75[3] -> 82[2] [send] via NET/Libfabric/2(74)/GDRDMA
+ip-26-0-168-30:235650:235824 [2] NCCL INFO Channel 02/0 : 99[3] -> 106[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO Channel 10/0 : 75[3] -> 82[2] [send] via NET/Libfabric/2(74)/GDRDMA
+ip-26-0-168-52:2722045:2722219 [0] NCCL INFO Channel 12/0 : 120[0] -> 127[7] via P2P/IPC
+ip-26-0-161-78:114512:114686 [0] NCCL INFO Channel 12/0 : 64[0] -> 71[7] via P2P/IPC
+ip-26-0-168-30:235655:235828 [7] NCCL INFO Channel 06/0 : 111[7] -> 118[6] [send] via NET/Libfabric/6(110)/GDRDMA
+ip-26-0-161-123:32704:32881 [0] NCCL INFO Channel 04/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-161-78:114519:114692 [7] NCCL INFO Channel 06/0 : 71[7] -> 78[6] [send] via NET/Libfabric/6(70)/GDRDMA
+ip-26-0-168-30:235648:235826 [0] NCCL INFO Channel 07/0 : 104[0] -> 110[6] via P2P/IPC
+ip-26-0-160-225:124160:124335 [0] NCCL INFO Channel 03/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-161-103:96888:97066 [0] NCCL INFO Channel 10/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-168-34:272150:272327 [0] NCCL INFO Channel 10/0 : 112[0] -> 119[7] via P2P/IPC
+ip-26-0-168-30:235651:235823 [3] NCCL INFO Channel 02/0 : 107[3] -> 114[2] [send] via NET/Libfabric/2(106)/GDRDMA
+ip-26-0-168-30:235655:235828 [7] NCCL INFO Channel 14/0 : 111[7] -> 118[6] [send] via NET/Libfabric/6(110)/GDRDMA
+ip-26-0-168-30:235650:235824 [2] NCCL INFO Channel 10/0 : 99[3] -> 106[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235648:235826 [0] NCCL INFO Channel 15/0 : 104[0] -> 110[6] via P2P/IPC
+ip-26-0-161-123:32704:32881 [0] NCCL INFO Channel 09/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-160-225:124160:124335 [0] NCCL INFO Channel 04/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-168-30:235651:235823 [3] NCCL INFO Channel 10/0 : 107[3] -> 114[2] [send] via NET/Libfabric/2(106)/GDRDMA
+ip-26-0-168-34:272157:272324 [7] NCCL INFO Channel 06/0 : 119[7] -> 126[6] [send] via NET/Libfabric/6(118)/GDRDMA
+ip-26-0-161-103:96894:97068 [6] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-161-103:96888:97066 [0] NCCL INFO Channel 11/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-168-34:272150:272327 [0] NCCL INFO Channel 11/0 : 112[0] -> 119[7] via P2P/IPC
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO Channel 06/0 : 15[7] -> 22[6] [send] via NET/Libfabric/6(14)/GDRDMA
+ip-26-0-161-78:114515:114690 [3] NCCL INFO Channel 03/0 : 60[4] -> 67[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32704:32881 [0] NCCL INFO Channel 10/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-160-225:124160:124335 [0] NCCL INFO Channel 09/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO Channel 14/0 : 15[7] -> 22[6] [send] via NET/Libfabric/6(14)/GDRDMA
+ip-26-0-161-103:96888:97066 [0] NCCL INFO Channel 12/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-168-30:235648:235826 [0] NCCL INFO Channel 01/0 : 104[0] -> 111[7] via P2P/IPC
+ip-26-0-161-78:114515:114690 [3] NCCL INFO Channel 11/0 : 60[4] -> 67[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272152:272331 [2] NCCL INFO Channel 02/0 : 107[3] -> 114[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32704:32881 [0] NCCL INFO Channel 11/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-160-225:124160:124335 [0] NCCL INFO Channel 10/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-161-78:114515:114690 [3] NCCL INFO Channel 02/0 : 67[3] -> 74[2] [send] via NET/Libfabric/2(66)/GDRDMA
+ip-26-0-168-34:272152:272331 [2] NCCL INFO Channel 10/0 : 107[3] -> 114[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272153:272328 [3] NCCL INFO Channel 02/0 : 115[3] -> 122[2] [send] via NET/Libfabric/2(114)/GDRDMA
+ip-26-0-161-78:114515:114690 [3] NCCL INFO Channel 10/0 : 67[3] -> 74[2] [send] via NET/Libfabric/2(66)/GDRDMA
+ip-26-0-168-30:235654:235821 [6] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-161-123:32711:32884 [7] NCCL INFO Channel 06/0 : 31[7] -> 38[6] [send] via NET/Libfabric/6(30)/GDRDMA
+ip-26-0-161-103:96895:97069 [7] NCCL INFO Channel 06/0 : 23[7] -> 30[6] [send] via NET/Libfabric/6(22)/GDRDMA
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO Channel 00/0 : 47[7] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO Channel 06/0 : 8[0] -> 13[5] via P2P/IPC
+ip-26-0-168-30:235648:235826 [0] NCCL INFO Channel 02/0 : 104[0] -> 111[7] via P2P/IPC
+ip-26-0-161-123:32704:32881 [0] NCCL INFO Channel 12/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-168-34:272153:272328 [3] NCCL INFO Channel 10/0 : 115[3] -> 122[2] [send] via NET/Libfabric/2(114)/GDRDMA
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO Channel 00/0 : 55[7] -> 56[0] [send] via NET/Libfabric/0(48)/GDRDMA
+ip-26-0-160-225:124160:124335 [0] NCCL INFO Channel 11/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-161-123:32711:32884 [7] NCCL INFO Channel 14/0 : 31[7] -> 38[6] [send] via NET/Libfabric/6(30)/GDRDMA
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO Channel 08/0 : 47[7] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO Channel 00/0 : 48[0] -> 49[1] via P2P/IPC
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO Channel 08/0 : 55[7] -> 56[0] [send] via NET/Libfabric/0(48)/GDRDMA
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO Channel 14/0 : 8[0] -> 13[5] via P2P/IPC
+ip-26-0-160-225:124160:124335 [0] NCCL INFO Channel 12/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-160-225:124167:124338 [7] NCCL INFO Channel 06/0 : 7[7] -> 14[6] [send] via NET/Libfabric/6(6)/GDRDMA
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO Channel 05/0 : 56[0] -> 60[4] via P2P/IPC
+ip-26-0-168-34:272150:272327 [0] NCCL INFO Channel 12/0 : 112[0] -> 119[7] via P2P/IPC
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO Channel 07/0 : 8[0] -> 14[6] via P2P/IPC
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO Channel 13/0 : 56[0] -> 60[4] via P2P/IPC
+ip-26-0-168-30:235648:235826 [0] NCCL INFO Channel 03/0 : 104[0] -> 111[7] via P2P/IPC
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO Channel 15/0 : 8[0] -> 14[6] via P2P/IPC
+ip-26-0-168-30:235648:235826 [0] NCCL INFO Channel 04/0 : 104[0] -> 111[7] via P2P/IPC
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO Channel 06/0 : 56[0] -> 61[5] via P2P/IPC
+ip-26-0-168-30:235653:235825 [5] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-168-30:235648:235826 [0] NCCL INFO Channel 09/0 : 104[0] -> 111[7] via P2P/IPC
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO Channel 01/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO Channel 14/0 : 56[0] -> 61[5] via P2P/IPC
+ip-26-0-168-30:235648:235826 [0] NCCL INFO Channel 10/0 : 104[0] -> 111[7] via P2P/IPC
+ip-26-0-168-52:2722048:2722220 [3] NCCL INFO Channel 03/0 : 116[4] -> 123[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO Channel 08/0 : 48[0] -> 49[1] via P2P/IPC
+ip-26-0-168-52:2722048:2722220 [3] NCCL INFO Channel 11/0 : 116[4] -> 123[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235648:235826 [0] NCCL INFO Channel 11/0 : 104[0] -> 111[7] via P2P/IPC
+ip-26-0-168-52:2722047:2722221 [2] NCCL INFO Channel 02/0 : 115[3] -> 122[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO Channel 02/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO Channel 07/0 : 56[0] -> 62[6] via P2P/IPC
+ip-26-0-168-52:2722047:2722221 [2] NCCL INFO Channel 10/0 : 115[3] -> 122[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-168-30:235648:235826 [0] NCCL INFO Channel 12/0 : 104[0] -> 111[7] via P2P/IPC
+ip-26-0-168-52:2722048:2722220 [3] NCCL INFO Channel 02/0 : 123[3] -> 2[2] [send] via NET/Libfabric/2(122)/GDRDMA
+ip-26-0-168-52:2722048:2722220 [3] NCCL INFO Channel 10/0 : 123[3] -> 2[2] [send] via NET/Libfabric/2(122)/GDRDMA
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO Channel 15/0 : 56[0] -> 62[6] via P2P/IPC
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO Channel 03/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO Channel 05/0 : 48[0] -> 52[4] via P2P/IPC
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO Channel 04/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO Channel 01/0 : 56[0] -> 63[7] via P2P/IPC
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO Channel 09/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO Channel 10/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO Channel 13/0 : 48[0] -> 52[4] via P2P/IPC
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO Channel 11/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO Channel 12/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO Channel 06/0 : 48[0] -> 53[5] via P2P/IPC
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO Channel 14/0 : 48[0] -> 53[5] via P2P/IPC
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO Channel 03/0 : 44[4] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62881:63059 [0] NCCL INFO Channel 06/0 : 40[0] -> 45[5] via P2P/IPC
+ip-26-0-161-142:62885:63054 [4] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO Channel 11/0 : 44[4] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO Channel 07/0 : 48[0] -> 54[6] via P2P/IPC
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-161-142:62881:63059 [0] NCCL INFO Channel 14/0 : 40[0] -> 45[5] via P2P/IPC
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO Channel 02/0 : 56[0] -> 63[7] via P2P/IPC
+ip-26-0-161-142:62881:63059 [0] NCCL INFO Channel 07/0 : 40[0] -> 46[6] via P2P/IPC
+ip-26-0-161-142:62886:63057 [5] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO Channel 02/0 : 43[3] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO Channel 15/0 : 48[0] -> 54[6] via P2P/IPC
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO Channel 10/0 : 43[3] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO Channel 02/0 : 51[3] -> 58[2] [send] via NET/Libfabric/2(50)/GDRDMA
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO Channel 03/0 : 56[0] -> 63[7] via P2P/IPC
+ip-26-0-161-142:62881:63059 [0] NCCL INFO Channel 15/0 : 40[0] -> 46[6] via P2P/IPC
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO Channel 01/0 : 48[0] -> 55[7] via P2P/IPC
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO Channel 10/0 : 51[3] -> 58[2] [send] via NET/Libfabric/2(50)/GDRDMA
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO Channel 04/0 : 56[0] -> 63[7] via P2P/IPC
+ip-26-0-161-142:62881:63059 [0] NCCL INFO Channel 01/0 : 40[0] -> 47[7] via P2P/IPC
+ip-26-0-161-142:62887:63055 [6] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO Channel 09/0 : 56[0] -> 63[7] via P2P/IPC
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO Channel 10/0 : 56[0] -> 63[7] via P2P/IPC
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO Channel 02/0 : 48[0] -> 55[7] via P2P/IPC
+ip-26-0-161-142:62881:63059 [0] NCCL INFO Channel 02/0 : 40[0] -> 47[7] via P2P/IPC
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO Channel 11/0 : 56[0] -> 63[7] via P2P/IPC
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO Channel 03/0 : 48[0] -> 55[7] via P2P/IPC
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO Channel 12/0 : 56[0] -> 63[7] via P2P/IPC
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-161-142:62881:63059 [0] NCCL INFO Channel 03/0 : 40[0] -> 47[7] via P2P/IPC
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO Channel 04/0 : 48[0] -> 55[7] via P2P/IPC
+ip-26-0-161-142:62881:63059 [0] NCCL INFO Channel 04/0 : 40[0] -> 47[7] via P2P/IPC
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO Channel 06/0 : 55[7] -> 62[6] [send] via NET/Libfabric/6(54)/GDRDMA
+ip-26-0-161-142:62881:63059 [0] NCCL INFO Channel 09/0 : 40[0] -> 47[7] via P2P/IPC
+ip-26-0-161-142:62881:63059 [0] NCCL INFO Channel 10/0 : 40[0] -> 47[7] via P2P/IPC
+ip-26-0-161-142:62881:63059 [0] NCCL INFO Channel 11/0 : 40[0] -> 47[7] via P2P/IPC
+ip-26-0-161-142:62888:63060 [7] NCCL INFO Channel 06/0 : 47[7] -> 54[6] [send] via NET/Libfabric/6(46)/GDRDMA
+ip-26-0-161-142:62881:63059 [0] NCCL INFO Channel 12/0 : 40[0] -> 47[7] via P2P/IPC
+ip-26-0-161-142:62888:63060 [7] NCCL INFO Channel 14/0 : 47[7] -> 54[6] [send] via NET/Libfabric/6(46)/GDRDMA
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO Channel 09/0 : 48[0] -> 55[7] via P2P/IPC
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO Channel 10/0 : 48[0] -> 55[7] via P2P/IPC
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO Channel 11/0 : 48[0] -> 55[7] via P2P/IPC
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO Channel 12/0 : 48[0] -> 55[7] via P2P/IPC
+ip-26-0-165-202:153563:153740 [1] NCCL INFO Channel 01/0 : 82[2] -> 89[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153563:153740 [1] NCCL INFO Channel 09/0 : 82[2] -> 89[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153564:153737 [2] NCCL INFO Channel 01/0 : 90[2] -> 97[1] [send] via NET/Libfabric/1(89)/GDRDMA
+ip-26-0-161-142:62882:63056 [1] NCCL INFO Channel 01/0 : 34[2] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153564:153737 [2] NCCL INFO Channel 09/0 : 90[2] -> 97[1] [send] via NET/Libfabric/1(89)/GDRDMA
+ip-26-0-161-142:62882:63056 [1] NCCL INFO Channel 09/0 : 34[2] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62883:63058 [2] NCCL INFO Channel 01/0 : 42[2] -> 49[1] [send] via NET/Libfabric/1(41)/GDRDMA
+ip-26-0-161-142:62883:63058 [2] NCCL INFO Channel 09/0 : 42[2] -> 49[1] [send] via NET/Libfabric/1(41)/GDRDMA
+ip-26-0-161-138:96564:96757 [1] NCCL INFO Channel 01/0 : 26[2] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96565:96754 [2] NCCL INFO Channel 01/0 : 34[2] -> 41[1] [send] via NET/Libfabric/1(33)/GDRDMA
+ip-26-0-161-138:96564:96757 [1] NCCL INFO Channel 09/0 : 26[2] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96565:96754 [2] NCCL INFO Channel 09/0 : 34[2] -> 41[1] [send] via NET/Libfabric/1(33)/GDRDMA
+ip-26-0-161-103:96889:97065 [1] NCCL INFO Channel 01/0 : 10[2] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96889:97065 [1] NCCL INFO Channel 09/0 : 10[2] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96890:97070 [2] NCCL INFO Channel 01/0 : 18[2] -> 25[1] [send] via NET/Libfabric/1(17)/GDRDMA
+ip-26-0-161-103:96890:97070 [2] NCCL INFO Channel 09/0 : 18[2] -> 25[1] [send] via NET/Libfabric/1(17)/GDRDMA
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO Channel 01/0 : 74[2] -> 81[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO Channel 09/0 : 74[2] -> 81[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO Channel 01/0 : 82[2] -> 89[1] [send] via NET/Libfabric/1(81)/GDRDMA
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO Channel 09/0 : 82[2] -> 89[1] [send] via NET/Libfabric/1(81)/GDRDMA
+ip-26-0-168-30:235649:235827 [1] NCCL INFO Channel 01/0 : 98[2] -> 105[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235649:235827 [1] NCCL INFO Channel 09/0 : 98[2] -> 105[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235650:235824 [2] NCCL INFO Channel 01/0 : 106[2] -> 113[1] [send] via NET/Libfabric/1(105)/GDRDMA
+ip-26-0-168-30:235650:235824 [2] NCCL INFO Channel 09/0 : 106[2] -> 113[1] [send] via NET/Libfabric/1(105)/GDRDMA
+ip-26-0-161-78:114513:114688 [1] NCCL INFO Channel 01/0 : 58[2] -> 65[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114513:114688 [1] NCCL INFO Channel 09/0 : 58[2] -> 65[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114514:114689 [2] NCCL INFO Channel 01/0 : 66[2] -> 73[1] [send] via NET/Libfabric/1(65)/GDRDMA
+ip-26-0-161-78:114514:114689 [2] NCCL INFO Channel 09/0 : 66[2] -> 73[1] [send] via NET/Libfabric/1(65)/GDRDMA
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO Channel 01/0 : 90[2] -> 97[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO Channel 09/0 : 90[2] -> 97[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO Channel 01/0 : 98[2] -> 105[1] [send] via NET/Libfabric/1(97)/GDRDMA
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO Channel 09/0 : 98[2] -> 105[1] [send] via NET/Libfabric/1(97)/GDRDMA
+ip-26-0-160-225:124161:124337 [1] NCCL INFO Channel 01/0 : 122[2] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO Channel 04/0 : 93[5] -> 100[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:124161:124337 [1] NCCL INFO Channel 09/0 : 122[2] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO Channel 06/0 : 79[7] -> 86[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:124162:124342 [2] NCCL INFO Channel 01/0 : 2[2] -> 9[1] [send] via NET/Libfabric/1(1)/GDRDMA
+ip-26-0-165-202:153566:153741 [4] NCCL INFO Channel 04/0 : 85[5] -> 92[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:124162:124342 [2] NCCL INFO Channel 09/0 : 2[2] -> 9[1] [send] via NET/Libfabric/1(1)/GDRDMA
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO Channel 12/0 : 93[5] -> 100[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153566:153741 [4] NCCL INFO Channel 12/0 : 85[5] -> 92[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO Channel 14/0 : 79[7] -> 86[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96565:96754 [2] NCCL INFO Channel 02/0 : 34[2] -> 33[1] via P2P/IPC
+ip-26-0-168-52:2722046:2722223 [1] NCCL INFO Channel 01/0 : 114[2] -> 121[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO Channel 03/0 : 100[4] -> 107[3] [send] via NET/Libfabric/3(99)/GDRDMA
+ip-26-0-161-142:62882:63056 [1] NCCL INFO Channel 01/0 : 41[1] -> 40[0] via P2P/IPC
+ip-26-0-168-52:2722047:2722221 [2] NCCL INFO Channel 01/0 : 122[2] -> 1[1] [send] via NET/Libfabric/1(121)/GDRDMA
+ip-26-0-165-202:153566:153741 [4] NCCL INFO Channel 03/0 : 92[4] -> 99[3] [send] via NET/Libfabric/3(91)/GDRDMA
+ip-26-0-161-138:96565:96754 [2] NCCL INFO Channel 03/0 : 34[2] -> 33[1] via P2P/IPC
+ip-26-0-168-52:2722046:2722223 [1] NCCL INFO Channel 09/0 : 114[2] -> 121[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO Channel 11/0 : 100[4] -> 107[3] [send] via NET/Libfabric/3(99)/GDRDMA
+ip-26-0-168-52:2722047:2722221 [2] NCCL INFO Channel 09/0 : 122[2] -> 1[1] [send] via NET/Libfabric/1(121)/GDRDMA
+ip-26-0-165-202:153566:153741 [4] NCCL INFO Channel 11/0 : 92[4] -> 99[3] [send] via NET/Libfabric/3(91)/GDRDMA
+ip-26-0-161-142:62882:63056 [1] NCCL INFO Channel 02/0 : 41[1] -> 40[0] via P2P/IPC
+ip-26-0-161-138:96565:96754 [2] NCCL INFO Channel 04/0 : 34[2] -> 33[1] via P2P/IPC
+ip-26-0-161-142:62882:63056 [1] NCCL INFO Channel 03/0 : 41[1] -> 40[0] via P2P/IPC
+ip-26-0-161-138:96565:96754 [2] NCCL INFO Channel 05/0 : 34[2] -> 33[1] via P2P/IPC
+ip-26-0-161-142:62882:63056 [1] NCCL INFO Channel 04/0 : 41[1] -> 40[0] via P2P/IPC
+ip-26-0-161-138:96565:96754 [2] NCCL INFO Channel 06/0 : 34[2] -> 33[1] via P2P/IPC
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO Channel 04/0 : 77[5] -> 84[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153568:153736 [6] NCCL INFO Channel 06/0 : 87[7] -> 94[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO Channel 12/0 : 77[5] -> 84[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153569:153739 [7] NCCL INFO Channel 14/0 : 95[7] -> 102[6] [send] via NET/Libfabric/6(94)/GDRDMA
+ip-26-0-161-142:62882:63056 [1] NCCL INFO Channel 05/0 : 41[1] -> 40[0] via P2P/IPC
+ip-26-0-161-138:96565:96754 [2] NCCL INFO Channel 07/0 : 34[2] -> 33[1] via P2P/IPC
+ip-26-0-165-202:153568:153736 [6] NCCL INFO Channel 14/0 : 87[7] -> 94[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96569:96758 [6] NCCL INFO Channel 06/0 : 31[7] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO Channel 03/0 : 84[4] -> 91[3] [send] via NET/Libfabric/3(83)/GDRDMA
+ip-26-0-161-138:96565:96754 [2] NCCL INFO Channel 10/0 : 34[2] -> 33[1] via P2P/IPC
+ip-26-0-161-142:62882:63056 [1] NCCL INFO Channel 06/0 : 41[1] -> 40[0] via P2P/IPC
+ip-26-0-161-138:96569:96758 [6] NCCL INFO Channel 14/0 : 31[7] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO Channel 05/0 : 78[6] -> 85[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96570:96759 [7] NCCL INFO Channel 14/0 : 39[7] -> 46[6] [send] via NET/Libfabric/6(38)/GDRDMA
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO Channel 11/0 : 84[4] -> 91[3] [send] via NET/Libfabric/3(83)/GDRDMA
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO Channel 02/0 : 82[2] -> 81[1] via P2P/IPC
+ip-26-0-161-138:96565:96754 [2] NCCL INFO Channel 11/0 : 34[2] -> 33[1] via P2P/IPC
+ip-26-0-161-142:62882:63056 [1] NCCL INFO Channel 07/0 : 41[1] -> 40[0] via P2P/IPC
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO Channel 13/0 : 78[6] -> 85[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO Channel 05/0 : 86[6] -> 93[5] [send] via NET/Libfabric/5(85)/GDRDMA
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO Channel 03/0 : 82[2] -> 81[1] via P2P/IPC
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO Channel 13/0 : 86[6] -> 93[5] [send] via NET/Libfabric/5(85)/GDRDMA
+ip-26-0-161-138:96565:96754 [2] NCCL INFO Channel 12/0 : 34[2] -> 33[1] via P2P/IPC
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO Channel 04/0 : 85[5] -> 92[4] [send] via NET/Libfabric/4(84)/GDRDMA
+ip-26-0-161-142:62882:63056 [1] NCCL INFO Channel 09/0 : 41[1] -> 40[0] via P2P/IPC
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO Channel 12/0 : 85[5] -> 92[4] [send] via NET/Libfabric/4(84)/GDRDMA
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO Channel 04/0 : 82[2] -> 81[1] via P2P/IPC
+ip-26-0-161-138:96565:96754 [2] NCCL INFO Channel 13/0 : 34[2] -> 33[1] via P2P/IPC
+ip-26-0-161-142:62882:63056 [1] NCCL INFO Channel 10/0 : 41[1] -> 40[0] via P2P/IPC
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO Channel 05/0 : 82[2] -> 81[1] via P2P/IPC
+ip-26-0-161-138:96565:96754 [2] NCCL INFO Channel 14/0 : 34[2] -> 33[1] via P2P/IPC
+ip-26-0-161-142:62882:63056 [1] NCCL INFO Channel 11/0 : 41[1] -> 40[0] via P2P/IPC
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO Channel 06/0 : 82[2] -> 81[1] via P2P/IPC
+ip-26-0-161-138:96565:96754 [2] NCCL INFO Channel 15/0 : 34[2] -> 33[1] via P2P/IPC
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO Channel 01/0 : 66[2] -> 73[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96892:97064 [4] NCCL INFO Channel 04/0 : 13[5] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62882:63056 [1] NCCL INFO Channel 12/0 : 41[1] -> 40[0] via P2P/IPC
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO Channel 09/0 : 66[2] -> 73[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO Channel 07/0 : 82[2] -> 81[1] via P2P/IPC
+ip-26-0-161-103:96892:97064 [4] NCCL INFO Channel 12/0 : 13[5] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO Channel 01/0 : 74[2] -> 81[1] [send] via NET/Libfabric/1(73)/GDRDMA
+ip-26-0-168-34:272151:272326 [1] NCCL INFO Channel 01/0 : 106[2] -> 113[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO Channel 09/0 : 74[2] -> 81[1] [send] via NET/Libfabric/1(73)/GDRDMA
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO Channel 10/0 : 82[2] -> 81[1] via P2P/IPC
+ip-26-0-168-34:272151:272326 [1] NCCL INFO Channel 09/0 : 106[2] -> 113[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62882:63056 [1] NCCL INFO Channel 13/0 : 41[1] -> 40[0] via P2P/IPC
+ip-26-0-168-34:272152:272331 [2] NCCL INFO Channel 01/0 : 114[2] -> 121[1] [send] via NET/Libfabric/1(113)/GDRDMA
+ip-26-0-161-103:96892:97064 [4] NCCL INFO Channel 03/0 : 20[4] -> 27[3] [send] via NET/Libfabric/3(19)/GDRDMA
+ip-26-0-160-225:124164:124336 [4] NCCL INFO Channel 04/0 : 125[5] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272152:272331 [2] NCCL INFO Channel 09/0 : 114[2] -> 121[1] [send] via NET/Libfabric/1(113)/GDRDMA
+ip-26-0-161-103:96892:97064 [4] NCCL INFO Channel 11/0 : 20[4] -> 27[3] [send] via NET/Libfabric/3(19)/GDRDMA
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO Channel 11/0 : 82[2] -> 81[1] via P2P/IPC
+ip-26-0-161-142:62882:63056 [1] NCCL INFO Channel 14/0 : 41[1] -> 40[0] via P2P/IPC
+ip-26-0-160-225:124164:124336 [4] NCCL INFO Channel 12/0 : 125[5] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96567:96760 [4] NCCL INFO Channel 04/0 : 29[5] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO Channel 12/0 : 82[2] -> 81[1] via P2P/IPC
+ip-26-0-161-142:62882:63056 [1] NCCL INFO Channel 15/0 : 41[1] -> 40[0] via P2P/IPC
+ip-26-0-161-138:96567:96760 [4] NCCL INFO Channel 12/0 : 29[5] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153567:153742 [5] NCCL INFO Channel 05/0 : 86[6] -> 93[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO Channel 05/0 : 94[6] -> 101[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153568:153736 [6] NCCL INFO Channel 05/0 : 94[6] -> 101[5] [send] via NET/Libfabric/5(93)/GDRDMA
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO Channel 13/0 : 82[2] -> 81[1] via P2P/IPC
+ip-26-0-160-225:124164:124336 [4] NCCL INFO Channel 03/0 : 4[4] -> 11[3] [send] via NET/Libfabric/3(3)/GDRDMA
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO Channel 13/0 : 94[6] -> 101[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153567:153742 [5] NCCL INFO Channel 13/0 : 86[6] -> 93[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124164:124336 [4] NCCL INFO Channel 11/0 : 4[4] -> 11[3] [send] via NET/Libfabric/3(3)/GDRDMA
+ip-26-0-165-202:153568:153736 [6] NCCL INFO Channel 13/0 : 94[6] -> 101[5] [send] via NET/Libfabric/5(93)/GDRDMA
+ip-26-0-161-138:96568:96753 [5] NCCL INFO Channel 05/0 : 30[6] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO Channel 14/0 : 82[2] -> 81[1] via P2P/IPC
+ip-26-0-165-202:153567:153742 [5] NCCL INFO Channel 04/0 : 93[5] -> 100[4] [send] via NET/Libfabric/4(92)/GDRDMA
+ip-26-0-161-138:96567:96760 [4] NCCL INFO Channel 03/0 : 36[4] -> 43[3] [send] via NET/Libfabric/3(35)/GDRDMA
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO Channel 04/0 : 101[5] -> 108[4] [send] via NET/Libfabric/4(100)/GDRDMA
+ip-26-0-165-202:153567:153742 [5] NCCL INFO Channel 12/0 : 93[5] -> 100[4] [send] via NET/Libfabric/4(92)/GDRDMA
+ip-26-0-161-78:114517:114687 [5] NCCL INFO Channel 05/0 : 62[6] -> 69[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96568:96753 [5] NCCL INFO Channel 13/0 : 30[6] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO Channel 12/0 : 101[5] -> 108[4] [send] via NET/Libfabric/4(100)/GDRDMA
+ip-26-0-161-78:114517:114687 [5] NCCL INFO Channel 13/0 : 62[6] -> 69[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235649:235827 [1] NCCL INFO Channel 01/0 : 105[1] -> 104[0] via P2P/IPC
+ip-26-0-161-138:96569:96758 [6] NCCL INFO Channel 05/0 : 38[6] -> 45[5] [send] via NET/Libfabric/5(37)/GDRDMA
+ip-26-0-161-138:96567:96760 [4] NCCL INFO Channel 11/0 : 36[4] -> 43[3] [send] via NET/Libfabric/3(35)/GDRDMA
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-161-103:96893:97067 [5] NCCL INFO Channel 05/0 : 14[6] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96569:96758 [6] NCCL INFO Channel 13/0 : 38[6] -> 45[5] [send] via NET/Libfabric/5(37)/GDRDMA
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO Channel 15/0 : 82[2] -> 81[1] via P2P/IPC
+ip-26-0-161-138:96568:96753 [5] NCCL INFO Channel 04/0 : 37[5] -> 44[4] [send] via NET/Libfabric/4(36)/GDRDMA
+ip-26-0-161-103:96893:97067 [5] NCCL INFO Channel 13/0 : 14[6] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96568:96753 [5] NCCL INFO Channel 12/0 : 37[5] -> 44[4] [send] via NET/Libfabric/4(36)/GDRDMA
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO Channel 02/0 : 98[2] -> 97[1] via P2P/IPC
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO Channel 04/0 : 69[5] -> 76[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235649:235827 [1] NCCL INFO Channel 02/0 : 105[1] -> 104[0] via P2P/IPC
+ip-26-0-161-123:32711:32884 [7] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-161-103:96893:97067 [5] NCCL INFO Channel 04/0 : 21[5] -> 28[4] [send] via NET/Libfabric/4(20)/GDRDMA
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO Channel 05/0 : 70[6] -> 77[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153568:153736 [6] NCCL INFO Channel 06/0 : 94[6] -> 91[3] via P2P/IPC
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO Channel 12/0 : 69[5] -> 76[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153567:153742 [5] NCCL INFO Channel 05/0 : 93[5] -> 91[3] via P2P/IPC
+ip-26-0-161-103:96893:97067 [5] NCCL INFO Channel 12/0 : 21[5] -> 28[4] [send] via NET/Libfabric/4(20)/GDRDMA
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO Channel 01/0 : 50[2] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO Channel 13/0 : 70[6] -> 77[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153566:153741 [4] NCCL INFO Channel 07/0 : 92[4] -> 103[7] [send] via NET/Libfabric/7(95)/GDRDMA
+ip-26-0-168-52:2722047:2722221 [2] NCCL INFO Channel 02/0 : 122[2] -> 121[1] via P2P/IPC
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO Channel 03/0 : 98[2] -> 97[1] via P2P/IPC
+ip-26-0-165-202:153566:153741 [4] NCCL INFO Channel 15/0 : 92[4] -> 103[7] [send] via NET/Libfabric/7(95)/GDRDMA
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO Channel 01/0 : 58[2] -> 65[1] [send] via NET/Libfabric/1(57)/GDRDMA
+ip-26-0-161-103:96894:97068 [6] NCCL INFO Channel 06/0 : 15[7] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO Channel 09/0 : 50[2] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO Channel 04/0 : 77[5] -> 84[4] [send] via NET/Libfabric/4(76)/GDRDMA
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO Channel 09/0 : 58[2] -> 65[1] [send] via NET/Libfabric/1(57)/GDRDMA
+ip-26-0-168-30:235649:235827 [1] NCCL INFO Channel 03/0 : 105[1] -> 104[0] via P2P/IPC
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO Channel 03/0 : 76[4] -> 83[3] [send] via NET/Libfabric/3(75)/GDRDMA
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO Channel 06/0 : 95[7] -> 102[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32705:32880 [1] NCCL INFO Channel 01/0 : 18[2] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96894:97068 [6] NCCL INFO Channel 14/0 : 15[7] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO Channel 12/0 : 77[5] -> 84[4] [send] via NET/Libfabric/4(76)/GDRDMA
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO Channel 04/0 : 98[2] -> 97[1] via P2P/IPC
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO Channel 14/0 : 95[7] -> 102[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32705:32880 [1] NCCL INFO Channel 09/0 : 18[2] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO Channel 11/0 : 76[4] -> 83[3] [send] via NET/Libfabric/3(75)/GDRDMA
+ip-26-0-161-103:96895:97069 [7] NCCL INFO Channel 14/0 : 23[7] -> 30[6] [send] via NET/Libfabric/6(22)/GDRDMA
+ip-26-0-168-52:2722047:2722221 [2] NCCL INFO Channel 03/0 : 122[2] -> 121[1] via P2P/IPC
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO Channel 06/0 : 103[7] -> 110[6] [send] via NET/Libfabric/6(102)/GDRDMA
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO Channel 01/0 : 97[1] -> 96[0] via P2P/IPC
+ip-26-0-161-123:32706:32877 [2] NCCL INFO Channel 01/0 : 26[2] -> 33[1] [send] via NET/Libfabric/1(25)/GDRDMA
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO Channel 01/0 : 2[2] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32706:32877 [2] NCCL INFO Channel 09/0 : 26[2] -> 33[1] [send] via NET/Libfabric/1(25)/GDRDMA
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO Channel 14/0 : 103[7] -> 110[6] [send] via NET/Libfabric/6(102)/GDRDMA
+ip-26-0-168-30:235649:235827 [1] NCCL INFO Channel 04/0 : 105[1] -> 104[0] via P2P/IPC
+ip-26-0-161-103:96894:97068 [6] NCCL INFO Channel 05/0 : 22[6] -> 29[5] [send] via NET/Libfabric/5(21)/GDRDMA
+ip-26-0-165-202:153563:153740 [1] NCCL INFO Channel 01/0 : 89[1] -> 88[0] via P2P/IPC
+ip-26-0-161-78:114516:114691 [4] NCCL INFO Channel 04/0 : 61[5] -> 68[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO Channel 05/0 : 102[6] -> 109[5] [send] via NET/Libfabric/5(101)/GDRDMA
+ip-26-0-160-225:124166:124339 [6] NCCL INFO Channel 06/0 : 127[7] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO Channel 09/0 : 10[2] -> 17[1] [send] via NET/Libfabric/1(9)/GDRDMA
+ip-26-0-165-202:153568:153736 [6] NCCL INFO Channel 14/0 : 94[6] -> 91[3] via P2P/IPC
+ip-26-0-161-78:114516:114691 [4] NCCL INFO Channel 12/0 : 61[5] -> 68[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96894:97068 [6] NCCL INFO Channel 13/0 : 22[6] -> 29[5] [send] via NET/Libfabric/5(21)/GDRDMA
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO Channel 13/0 : 102[6] -> 109[5] [send] via NET/Libfabric/5(101)/GDRDMA
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO Channel 09/0 : 2[2] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114517:114687 [5] NCCL INFO Channel 04/0 : 69[5] -> 76[4] [send] via NET/Libfabric/4(68)/GDRDMA
+ip-26-0-160-225:124167:124338 [7] NCCL INFO Channel 14/0 : 7[7] -> 14[6] [send] via NET/Libfabric/6(6)/GDRDMA
+ip-26-0-160-225:124166:124339 [6] NCCL INFO Channel 14/0 : 127[7] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-52:2722047:2722221 [2] NCCL INFO Channel 04/0 : 122[2] -> 121[1] via P2P/IPC
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO Channel 05/0 : 98[2] -> 97[1] via P2P/IPC
+ip-26-0-165-202:153567:153742 [5] NCCL INFO Channel 13/0 : 93[5] -> 91[3] via P2P/IPC
+ip-26-0-161-78:114517:114687 [5] NCCL INFO Channel 12/0 : 69[5] -> 76[4] [send] via NET/Libfabric/4(68)/GDRDMA
+ip-26-0-168-30:235649:235827 [1] NCCL INFO Channel 05/0 : 105[1] -> 104[0] via P2P/IPC
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO Channel 02/0 : 97[1] -> 96[0] via P2P/IPC
+ip-26-0-161-78:114518:114693 [6] NCCL INFO Channel 06/0 : 63[7] -> 70[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-165-202:153569:153739 [7] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-161-78:114516:114691 [4] NCCL INFO Channel 03/0 : 68[4] -> 75[3] [send] via NET/Libfabric/3(67)/GDRDMA
+ip-26-0-161-78:114518:114693 [6] NCCL INFO Channel 14/0 : 63[7] -> 70[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:124161:124337 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-168-52:2722047:2722221 [2] NCCL INFO Channel 05/0 : 122[2] -> 121[1] via P2P/IPC
+ip-26-0-161-78:114516:114691 [4] NCCL INFO Channel 11/0 : 68[4] -> 75[3] [send] via NET/Libfabric/3(67)/GDRDMA
+ip-26-0-161-78:114519:114692 [7] NCCL INFO Channel 14/0 : 71[7] -> 78[6] [send] via NET/Libfabric/6(70)/GDRDMA
+ip-26-0-168-30:235649:235827 [1] NCCL INFO Channel 06/0 : 105[1] -> 104[0] via P2P/IPC
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO Channel 06/0 : 98[2] -> 97[1] via P2P/IPC
+ip-26-0-165-202:153563:153740 [1] NCCL INFO Channel 02/0 : 89[1] -> 88[0] via P2P/IPC
+ip-26-0-160-225:124165:124341 [5] NCCL INFO Channel 05/0 : 126[6] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114518:114693 [6] NCCL INFO Channel 05/0 : 70[6] -> 77[5] [send] via NET/Libfabric/5(69)/GDRDMA
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO Channel 03/0 : 97[1] -> 96[0] via P2P/IPC
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO Channel 06/0 : 71[7] -> 78[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:124165:124341 [5] NCCL INFO Channel 13/0 : 126[6] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114518:114693 [6] NCCL INFO Channel 13/0 : 70[6] -> 77[5] [send] via NET/Libfabric/5(69)/GDRDMA
+ip-26-0-160-225:124166:124339 [6] NCCL INFO Channel 05/0 : 6[6] -> 13[5] [send] via NET/Libfabric/5(5)/GDRDMA
+ip-26-0-168-52:2722050:2722225 [5] NCCL INFO Channel 05/0 : 118[6] -> 125[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO Channel 07/0 : 84[4] -> 95[7] [send] via NET/Libfabric/7(87)/GDRDMA
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO Channel 14/0 : 71[7] -> 78[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235649:235827 [1] NCCL INFO Channel 07/0 : 105[1] -> 104[0] via P2P/IPC
+ip-26-0-160-225:124166:124339 [6] NCCL INFO Channel 13/0 : 6[6] -> 13[5] [send] via NET/Libfabric/5(5)/GDRDMA
+ip-26-0-168-52:2722047:2722221 [2] NCCL INFO Channel 06/0 : 122[2] -> 121[1] via P2P/IPC
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO Channel 15/0 : 84[4] -> 95[7] [send] via NET/Libfabric/7(87)/GDRDMA
+ip-26-0-168-52:2722050:2722225 [5] NCCL INFO Channel 13/0 : 118[6] -> 125[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO Channel 14/0 : 79[7] -> 86[6] [send] via NET/Libfabric/6(78)/GDRDMA
+ip-26-0-160-225:124161:124337 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-225:124165:124341 [5] NCCL INFO Channel 04/0 : 5[5] -> 12[4] [send] via NET/Libfabric/4(4)/GDRDMA
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO Channel 07/0 : 98[2] -> 97[1] via P2P/IPC
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO Channel 05/0 : 78[6] -> 85[5] [send] via NET/Libfabric/5(77)/GDRDMA
+ip-26-0-160-225:124165:124341 [5] NCCL INFO Channel 12/0 : 5[5] -> 12[4] [send] via NET/Libfabric/4(4)/GDRDMA
+ip-26-0-165-202:153563:153740 [1] NCCL INFO Channel 03/0 : 89[1] -> 88[0] via P2P/IPC
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO Channel 04/0 : 97[1] -> 96[0] via P2P/IPC
+ip-26-0-168-52:2722047:2722221 [2] NCCL INFO Channel 07/0 : 122[2] -> 121[1] via P2P/IPC
+ip-26-0-168-30:235654:235821 [6] NCCL INFO Channel 06/0 : 103[7] -> 110[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO Channel 07/0 : 100[4] -> 111[7] [send] via NET/Libfabric/7(103)/GDRDMA
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO Channel 13/0 : 78[6] -> 85[5] [send] via NET/Libfabric/5(77)/GDRDMA
+ip-26-0-168-30:235649:235827 [1] NCCL INFO Channel 09/0 : 105[1] -> 104[0] via P2P/IPC
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO Channel 15/0 : 100[4] -> 111[7] [send] via NET/Libfabric/7(103)/GDRDMA
+ip-26-0-168-34:272154:272330 [4] NCCL INFO Channel 04/0 : 109[5] -> 116[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272156:272325 [6] NCCL INFO Channel 06/0 : 111[7] -> 118[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114518:114693 [6] NCCL INFO Channel 06/0 : 70[6] -> 67[3] via P2P/IPC
+ip-26-0-165-202:153564:153737 [2] NCCL INFO Channel 02/0 : 90[2] -> 89[1] via P2P/IPC
+ip-26-0-168-30:235654:235821 [6] NCCL INFO Channel 14/0 : 103[7] -> 110[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO Channel 10/0 : 98[2] -> 97[1] via P2P/IPC
+ip-26-0-160-225:124161:124337 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-168-34:272154:272330 [4] NCCL INFO Channel 12/0 : 109[5] -> 116[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272157:272324 [7] NCCL INFO Channel 14/0 : 119[7] -> 126[6] [send] via NET/Libfabric/6(118)/GDRDMA
+ip-26-0-161-123:32708:32879 [4] NCCL INFO Channel 04/0 : 21[5] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO Channel 05/0 : 77[5] -> 75[3] via P2P/IPC
+ip-26-0-168-52:2722047:2722221 [2] NCCL INFO Channel 10/0 : 122[2] -> 121[1] via P2P/IPC
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO Channel 05/0 : 97[1] -> 96[0] via P2P/IPC
+ip-26-0-168-30:235649:235827 [1] NCCL INFO Channel 10/0 : 105[1] -> 104[0] via P2P/IPC
+ip-26-0-168-34:272156:272325 [6] NCCL INFO Channel 14/0 : 111[7] -> 118[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272155:272329 [5] NCCL INFO Channel 05/0 : 110[6] -> 117[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO Channel 06/0 : 86[6] -> 83[3] via P2P/IPC
+ip-26-0-161-123:32708:32879 [4] NCCL INFO Channel 12/0 : 21[5] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153563:153740 [1] NCCL INFO Channel 04/0 : 89[1] -> 88[0] via P2P/IPC
+ip-26-0-161-78:114519:114692 [7] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-168-34:272155:272329 [5] NCCL INFO Channel 13/0 : 110[6] -> 117[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-52:2722051:2722224 [6] NCCL INFO Channel 06/0 : 119[7] -> 126[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO Channel 11/0 : 98[2] -> 97[1] via P2P/IPC
+ip-26-0-165-202:153564:153737 [2] NCCL INFO Channel 03/0 : 90[2] -> 89[1] via P2P/IPC
+ip-26-0-168-34:272154:272330 [4] NCCL INFO Channel 03/0 : 116[4] -> 123[3] [send] via NET/Libfabric/3(115)/GDRDMA
+ip-26-0-160-225:124161:124337 [1] NCCL INFO Channel 04/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-168-30:235649:235827 [1] NCCL INFO Channel 11/0 : 105[1] -> 104[0] via P2P/IPC
+ip-26-0-161-123:32708:32879 [4] NCCL INFO Channel 03/0 : 28[4] -> 35[3] [send] via NET/Libfabric/3(27)/GDRDMA
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO Channel 05/0 : 85[5] -> 83[3] via P2P/IPC
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO Channel 06/0 : 78[6] -> 75[3] via P2P/IPC
+ip-26-0-168-52:2722047:2722221 [2] NCCL INFO Channel 11/0 : 122[2] -> 121[1] via P2P/IPC
+ip-26-0-168-34:272156:272325 [6] NCCL INFO Channel 05/0 : 118[6] -> 125[5] [send] via NET/Libfabric/5(117)/GDRDMA
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO Channel 06/0 : 97[1] -> 96[0] via P2P/IPC
+ip-26-0-161-78:114518:114693 [6] NCCL INFO Channel 14/0 : 70[6] -> 67[3] via P2P/IPC
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO Channel 01/0 : 73[1] -> 72[0] via P2P/IPC
+ip-26-0-168-52:2722052:2722226 [7] NCCL INFO Channel 14/0 : 127[7] -> 6[6] [send] via NET/Libfabric/6(126)/GDRDMA
+ip-26-0-168-34:272155:272329 [5] NCCL INFO Channel 04/0 : 117[5] -> 124[4] [send] via NET/Libfabric/4(116)/GDRDMA
+ip-26-0-168-52:2722051:2722224 [6] NCCL INFO Channel 14/0 : 119[7] -> 126[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153563:153740 [1] NCCL INFO Channel 05/0 : 89[1] -> 88[0] via P2P/IPC
+ip-26-0-161-123:32708:32879 [4] NCCL INFO Channel 11/0 : 28[4] -> 35[3] [send] via NET/Libfabric/3(27)/GDRDMA
+ip-26-0-168-34:272156:272325 [6] NCCL INFO Channel 13/0 : 118[6] -> 125[5] [send] via NET/Libfabric/5(117)/GDRDMA
+ip-26-0-168-34:272154:272330 [4] NCCL INFO Channel 11/0 : 116[4] -> 123[3] [send] via NET/Libfabric/3(115)/GDRDMA
+ip-26-0-168-34:272155:272329 [5] NCCL INFO Channel 12/0 : 117[5] -> 124[4] [send] via NET/Libfabric/4(116)/GDRDMA
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO Channel 12/0 : 98[2] -> 97[1] via P2P/IPC
+ip-26-0-168-52:2722047:2722221 [2] NCCL INFO Channel 12/0 : 122[2] -> 121[1] via P2P/IPC
+ip-26-0-168-30:235649:235827 [1] NCCL INFO Channel 12/0 : 105[1] -> 104[0] via P2P/IPC
+ip-26-0-165-202:153564:153737 [2] NCCL INFO Channel 04/0 : 90[2] -> 89[1] via P2P/IPC
+ip-26-0-161-78:114514:114689 [2] NCCL INFO Channel 02/0 : 66[2] -> 65[1] via P2P/IPC
+ip-26-0-160-225:124161:124337 [1] NCCL INFO Channel 05/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-168-52:2722051:2722224 [6] NCCL INFO Channel 05/0 : 126[6] -> 5[5] [send] via NET/Libfabric/5(125)/GDRDMA
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO Channel 07/0 : 97[1] -> 96[0] via P2P/IPC
+ip-26-0-168-30:235655:235828 [7] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-168-52:2722052:2722226 [7] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-168-52:2722051:2722224 [6] NCCL INFO Channel 13/0 : 126[6] -> 5[5] [send] via NET/Libfabric/5(125)/GDRDMA
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO Channel 04/0 : 5[5] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32710:32883 [6] NCCL INFO Channel 06/0 : 23[7] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235650:235824 [2] NCCL INFO Channel 02/0 : 106[2] -> 105[1] via P2P/IPC
+ip-26-0-165-202:153563:153740 [1] NCCL INFO Channel 06/0 : 89[1] -> 88[0] via P2P/IPC
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO Channel 12/0 : 5[5] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO Channel 13/0 : 98[2] -> 97[1] via P2P/IPC
+ip-26-0-168-30:235649:235827 [1] NCCL INFO Channel 13/0 : 105[1] -> 104[0] via P2P/IPC
+ip-26-0-161-123:32710:32883 [6] NCCL INFO Channel 14/0 : 23[7] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:124161:124337 [1] NCCL INFO Channel 06/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-202:153564:153737 [2] NCCL INFO Channel 05/0 : 90[2] -> 89[1] via P2P/IPC
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO Channel 09/0 : 97[1] -> 96[0] via P2P/IPC
+ip-26-0-168-34:272157:272324 [7] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO Channel 06/0 : 55[7] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32708:32879 [4] NCCL INFO Channel 07/0 : 28[4] -> 39[7] [send] via NET/Libfabric/7(31)/GDRDMA
+ip-26-0-161-78:114514:114689 [2] NCCL INFO Channel 03/0 : 66[2] -> 65[1] via P2P/IPC
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO Channel 13/0 : 77[5] -> 75[3] via P2P/IPC
+ip-26-0-161-123:32708:32879 [4] NCCL INFO Channel 15/0 : 28[4] -> 39[7] [send] via NET/Libfabric/7(31)/GDRDMA
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO Channel 14/0 : 55[7] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO Channel 03/0 : 12[4] -> 19[3] [send] via NET/Libfabric/3(11)/GDRDMA
+ip-26-0-165-202:153563:153740 [1] NCCL INFO Channel 07/0 : 89[1] -> 88[0] via P2P/IPC
+ip-26-0-168-30:235650:235824 [2] NCCL INFO Channel 03/0 : 106[2] -> 105[1] via P2P/IPC
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO Channel 07/0 : 76[4] -> 87[7] [send] via NET/Libfabric/7(79)/GDRDMA
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO Channel 02/0 : 73[1] -> 72[0] via P2P/IPC
+ip-26-0-168-30:235649:235827 [1] NCCL INFO Channel 14/0 : 105[1] -> 104[0] via P2P/IPC
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO Channel 14/0 : 98[2] -> 97[1] via P2P/IPC
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO Channel 11/0 : 12[4] -> 19[3] [send] via NET/Libfabric/3(11)/GDRDMA
+ip-26-0-168-52:2722051:2722224 [6] NCCL INFO Channel 06/0 : 126[6] -> 123[3] via P2P/IPC
+ip-26-0-165-202:153564:153737 [2] NCCL INFO Channel 06/0 : 90[2] -> 89[1] via P2P/IPC
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO Channel 10/0 : 97[1] -> 96[0] via P2P/IPC
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO Channel 14/0 : 78[6] -> 75[3] via P2P/IPC
+ip-26-0-160-225:124161:124337 [1] NCCL INFO Channel 07/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-202:153563:153740 [1] NCCL INFO Channel 09/0 : 89[1] -> 88[0] via P2P/IPC
+ip-26-0-161-78:114514:114689 [2] NCCL INFO Channel 04/0 : 66[2] -> 65[1] via P2P/IPC
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO Channel 02/0 : 58[2] -> 57[1] via P2P/IPC
+ip-26-0-168-30:235650:235824 [2] NCCL INFO Channel 04/0 : 106[2] -> 105[1] via P2P/IPC
+ip-26-0-168-30:235649:235827 [1] NCCL INFO Channel 15/0 : 105[1] -> 104[0] via P2P/IPC
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO Channel 06/0 : 7[7] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235652:235822 [4] NCCL INFO Channel 04/0 : 101[5] -> 108[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO Channel 15/0 : 98[2] -> 97[1] via P2P/IPC
+ip-26-0-165-202:153564:153737 [2] NCCL INFO Channel 07/0 : 90[2] -> 89[1] via P2P/IPC
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO Channel 14/0 : 7[7] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:124165:124341 [5] NCCL INFO Channel 05/0 : 5[5] -> 3[3] via P2P/IPC
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO Channel 11/0 : 97[1] -> 96[0] via P2P/IPC
+ip-26-0-168-30:235652:235822 [4] NCCL INFO Channel 12/0 : 101[5] -> 108[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO Channel 03/0 : 58[2] -> 57[1] via P2P/IPC
+ip-26-0-165-202:153563:153740 [1] NCCL INFO Channel 10/0 : 89[1] -> 88[0] via P2P/IPC
+ip-26-0-160-225:124161:124337 [1] NCCL INFO Channel 09/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO Channel 04/0 : 53[5] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114514:114689 [2] NCCL INFO Channel 05/0 : 66[2] -> 65[1] via P2P/IPC
+ip-26-0-168-30:235650:235824 [2] NCCL INFO Channel 05/0 : 106[2] -> 105[1] via P2P/IPC
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO Channel 04/0 : 117[5] -> 124[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO Channel 12/0 : 53[5] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153564:153737 [2] NCCL INFO Channel 10/0 : 90[2] -> 89[1] via P2P/IPC
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO Channel 12/0 : 117[5] -> 124[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO Channel 04/0 : 58[2] -> 57[1] via P2P/IPC
+ip-26-0-168-30:235652:235822 [4] NCCL INFO Channel 03/0 : 108[4] -> 115[3] [send] via NET/Libfabric/3(107)/GDRDMA
+ip-26-0-168-52:2722050:2722225 [5] NCCL INFO Channel 04/0 : 125[5] -> 4[4] [send] via NET/Libfabric/4(124)/GDRDMA
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO Channel 12/0 : 97[1] -> 96[0] via P2P/IPC
+ip-26-0-165-202:153563:153740 [1] NCCL INFO Channel 11/0 : 89[1] -> 88[0] via P2P/IPC
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO Channel 03/0 : 73[1] -> 72[0] via P2P/IPC
+ip-26-0-161-78:114514:114689 [2] NCCL INFO Channel 06/0 : 66[2] -> 65[1] via P2P/IPC
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO Channel 05/0 : 58[2] -> 57[1] via P2P/IPC
+ip-26-0-168-30:235652:235822 [4] NCCL INFO Channel 11/0 : 108[4] -> 115[3] [send] via NET/Libfabric/3(107)/GDRDMA
+ip-26-0-168-52:2722050:2722225 [5] NCCL INFO Channel 12/0 : 125[5] -> 4[4] [send] via NET/Libfabric/4(124)/GDRDMA
+ip-26-0-165-202:153564:153737 [2] NCCL INFO Channel 11/0 : 90[2] -> 89[1] via P2P/IPC
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO Channel 03/0 : 60[4] -> 67[3] [send] via NET/Libfabric/3(59)/GDRDMA
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO Channel 03/0 : 124[4] -> 3[3] [send] via NET/Libfabric/3(123)/GDRDMA
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO Channel 14/0 : 86[6] -> 83[3] via P2P/IPC
+ip-26-0-165-202:153563:153740 [1] NCCL INFO Channel 12/0 : 89[1] -> 88[0] via P2P/IPC
+ip-26-0-161-123:32705:32880 [1] NCCL INFO Channel 01/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO Channel 11/0 : 60[4] -> 67[3] [send] via NET/Libfabric/3(59)/GDRDMA
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO Channel 06/0 : 58[2] -> 57[1] via P2P/IPC
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO Channel 13/0 : 97[1] -> 96[0] via P2P/IPC
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO Channel 05/0 : 54[6] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO Channel 04/0 : 73[1] -> 72[0] via P2P/IPC
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO Channel 01/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-160-225:124161:124337 [1] NCCL INFO Channel 10/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-225:124165:124341 [5] NCCL INFO Channel 13/0 : 5[5] -> 3[3] via P2P/IPC
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO Channel 13/0 : 54[6] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32705:32880 [1] NCCL INFO Channel 02/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-161-78:114514:114689 [2] NCCL INFO Channel 07/0 : 66[2] -> 65[1] via P2P/IPC
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO Channel 05/0 : 101[5] -> 99[3] via P2P/IPC
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO Channel 05/0 : 62[6] -> 69[5] [send] via NET/Libfabric/5(61)/GDRDMA
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO Channel 07/0 : 58[2] -> 57[1] via P2P/IPC
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO Channel 02/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-168-34:272156:272325 [6] NCCL INFO Channel 06/0 : 118[6] -> 115[3] via P2P/IPC
+ip-26-0-161-123:32706:32877 [2] NCCL INFO Channel 02/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO Channel 05/0 : 6[6] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO Channel 13/0 : 85[5] -> 83[3] via P2P/IPC
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO Channel 13/0 : 62[6] -> 69[5] [send] via NET/Libfabric/5(61)/GDRDMA
+ip-26-0-168-52:2722050:2722225 [5] NCCL INFO Channel 05/0 : 125[5] -> 123[3] via P2P/IPC
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO Channel 02/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO Channel 10/0 : 58[2] -> 57[1] via P2P/IPC
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO Channel 04/0 : 61[5] -> 68[4] [send] via NET/Libfabric/4(60)/GDRDMA
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO Channel 13/0 : 6[6] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32705:32880 [1] NCCL INFO Channel 03/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO Channel 05/0 : 73[1] -> 72[0] via P2P/IPC
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO Channel 05/0 : 14[6] -> 21[5] [send] via NET/Libfabric/5(13)/GDRDMA
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO Channel 03/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO Channel 12/0 : 61[5] -> 68[4] [send] via NET/Libfabric/4(60)/GDRDMA
+ip-26-0-161-123:32706:32877 [2] NCCL INFO Channel 03/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO Channel 13/0 : 14[6] -> 21[5] [send] via NET/Libfabric/5(13)/GDRDMA
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO Channel 11/0 : 58[2] -> 57[1] via P2P/IPC
+ip-26-0-168-34:272151:272326 [1] NCCL INFO Channel 01/0 : 113[1] -> 112[0] via P2P/IPC
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO Channel 03/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-161-78:114514:114689 [2] NCCL INFO Channel 10/0 : 66[2] -> 65[1] via P2P/IPC
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO Channel 04/0 : 13[5] -> 20[4] [send] via NET/Libfabric/4(12)/GDRDMA
+ip-26-0-161-123:32705:32880 [1] NCCL INFO Channel 04/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-160-225:124161:124337 [1] NCCL INFO Channel 11/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO Channel 12/0 : 13[5] -> 20[4] [send] via NET/Libfabric/4(12)/GDRDMA
+ip-26-0-168-34:272156:272325 [6] NCCL INFO Channel 14/0 : 118[6] -> 115[3] via P2P/IPC
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO Channel 04/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-161-123:32709:32882 [5] NCCL INFO Channel 05/0 : 22[6] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153564:153737 [2] NCCL INFO Channel 12/0 : 90[2] -> 89[1] via P2P/IPC
+ip-26-0-168-30:235650:235824 [2] NCCL INFO Channel 06/0 : 106[2] -> 105[1] via P2P/IPC
+ip-26-0-161-123:32706:32877 [2] NCCL INFO Channel 04/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO Channel 12/0 : 58[2] -> 57[1] via P2P/IPC
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO Channel 04/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-160-225:124167:124338 [7] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-165-202:153563:153740 [1] NCCL INFO Channel 13/0 : 89[1] -> 88[0] via P2P/IPC
+ip-26-0-161-123:32709:32882 [5] NCCL INFO Channel 13/0 : 22[6] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO Channel 06/0 : 73[1] -> 72[0] via P2P/IPC
+ip-26-0-168-34:272151:272326 [1] NCCL INFO Channel 02/0 : 113[1] -> 112[0] via P2P/IPC
+ip-26-0-161-123:32710:32883 [6] NCCL INFO Channel 05/0 : 30[6] -> 37[5] [send] via NET/Libfabric/5(29)/GDRDMA
+ip-26-0-161-78:114517:114687 [5] NCCL INFO Channel 05/0 : 69[5] -> 67[3] via P2P/IPC
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO Channel 05/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-161-123:32705:32880 [1] NCCL INFO Channel 05/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-165-202:153564:153737 [2] NCCL INFO Channel 13/0 : 90[2] -> 89[1] via P2P/IPC
+ip-26-0-160-225:124161:124337 [1] NCCL INFO Channel 12/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO Channel 13/0 : 58[2] -> 57[1] via P2P/IPC
+ip-26-0-161-123:32706:32877 [2] NCCL INFO Channel 05/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-160-225:124166:124339 [6] NCCL INFO Channel 06/0 : 6[6] -> 3[3] via P2P/IPC
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO Channel 05/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-161-123:32709:32882 [5] NCCL INFO Channel 04/0 : 29[5] -> 36[4] [send] via NET/Libfabric/4(28)/GDRDMA
+ip-26-0-161-123:32710:32883 [6] NCCL INFO Channel 13/0 : 30[6] -> 37[5] [send] via NET/Libfabric/5(29)/GDRDMA
+ip-26-0-165-202:153563:153740 [1] NCCL INFO Channel 14/0 : 89[1] -> 88[0] via P2P/IPC
+ip-26-0-161-103:96893:97067 [5] NCCL INFO Channel 05/0 : 21[5] -> 19[3] via P2P/IPC
+ip-26-0-168-34:272152:272331 [2] NCCL INFO Channel 02/0 : 114[2] -> 113[1] via P2P/IPC
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO Channel 06/0 : 14[6] -> 11[3] via P2P/IPC
+ip-26-0-161-123:32709:32882 [5] NCCL INFO Channel 12/0 : 29[5] -> 36[4] [send] via NET/Libfabric/4(28)/GDRDMA
+ip-26-0-168-30:235653:235825 [5] NCCL INFO Channel 05/0 : 102[6] -> 109[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO Channel 06/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO Channel 14/0 : 58[2] -> 57[1] via P2P/IPC
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO Channel 07/0 : 73[1] -> 72[0] via P2P/IPC
+ip-26-0-161-123:32705:32880 [1] NCCL INFO Channel 06/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-168-34:272151:272326 [1] NCCL INFO Channel 03/0 : 113[1] -> 112[0] via P2P/IPC
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO Channel 05/0 : 13[5] -> 11[3] via P2P/IPC
+ip-26-0-165-202:153564:153737 [2] NCCL INFO Channel 14/0 : 90[2] -> 89[1] via P2P/IPC
+ip-26-0-168-30:235653:235825 [5] NCCL INFO Channel 13/0 : 102[6] -> 109[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32706:32877 [2] NCCL INFO Channel 06/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-168-30:235654:235821 [6] NCCL INFO Channel 05/0 : 110[6] -> 117[5] [send] via NET/Libfabric/5(109)/GDRDMA
+ip-26-0-165-202:153563:153740 [1] NCCL INFO Channel 15/0 : 89[1] -> 88[0] via P2P/IPC
+ip-26-0-161-78:114514:114689 [2] NCCL INFO Channel 11/0 : 66[2] -> 65[1] via P2P/IPC
+ip-26-0-168-34:272152:272331 [2] NCCL INFO Channel 03/0 : 114[2] -> 113[1] via P2P/IPC
+ip-26-0-161-103:96895:97069 [7] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO Channel 14/0 : 97[1] -> 96[0] via P2P/IPC
+ip-26-0-161-142:62885:63054 [4] NCCL INFO Channel 04/0 : 37[5] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO Channel 15/0 : 58[2] -> 57[1] via P2P/IPC
+ip-26-0-168-30:235654:235821 [6] NCCL INFO Channel 13/0 : 110[6] -> 117[5] [send] via NET/Libfabric/5(109)/GDRDMA
+ip-26-0-161-123:32705:32880 [1] NCCL INFO Channel 07/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-168-34:272151:272326 [1] NCCL INFO Channel 04/0 : 113[1] -> 112[0] via P2P/IPC
+ip-26-0-160-225:124161:124337 [1] NCCL INFO Channel 13/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-138:96564:96757 [1] NCCL INFO Channel 01/0 : 33[1] -> 32[0] via P2P/IPC
+ip-26-0-161-142:62885:63054 [4] NCCL INFO Channel 12/0 : 37[5] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235653:235825 [5] NCCL INFO Channel 04/0 : 109[5] -> 116[4] [send] via NET/Libfabric/4(108)/GDRDMA
+ip-26-0-161-123:32706:32877 [2] NCCL INFO Channel 07/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-161-103:96890:97070 [2] NCCL INFO Channel 02/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-161-103:96894:97068 [6] NCCL INFO Channel 06/0 : 22[6] -> 19[3] via P2P/IPC
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO Channel 06/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO Channel 09/0 : 73[1] -> 72[0] via P2P/IPC
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO Channel 02/0 : 74[2] -> 73[1] via P2P/IPC
+ip-26-0-165-202:153564:153737 [2] NCCL INFO Channel 15/0 : 90[2] -> 89[1] via P2P/IPC
+ip-26-0-168-34:272152:272331 [2] NCCL INFO Channel 04/0 : 114[2] -> 113[1] via P2P/IPC
+ip-26-0-168-30:235653:235825 [5] NCCL INFO Channel 12/0 : 109[5] -> 116[4] [send] via NET/Libfabric/4(108)/GDRDMA
+ip-26-0-161-123:32710:32883 [6] NCCL INFO Channel 06/0 : 30[6] -> 27[3] via P2P/IPC
+ip-26-0-160-225:124166:124339 [6] NCCL INFO Channel 14/0 : 6[6] -> 3[3] via P2P/IPC
+ip-26-0-161-138:96564:96757 [1] NCCL INFO Channel 02/0 : 33[1] -> 32[0] via P2P/IPC
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-161-123:32705:32880 [1] NCCL INFO Channel 09/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-161-123:32709:32882 [5] NCCL INFO Channel 05/0 : 29[5] -> 27[3] via P2P/IPC
+ip-26-0-168-34:272151:272326 [1] NCCL INFO Channel 05/0 : 113[1] -> 112[0] via P2P/IPC
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO Channel 14/0 : 14[6] -> 11[3] via P2P/IPC
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO Channel 07/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-161-78:114517:114687 [5] NCCL INFO Channel 13/0 : 69[5] -> 67[3] via P2P/IPC
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO Channel 15/0 : 97[1] -> 96[0] via P2P/IPC
+ip-26-0-161-142:62885:63054 [4] NCCL INFO Channel 03/0 : 44[4] -> 51[3] [send] via NET/Libfabric/3(43)/GDRDMA
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO Channel 13/0 : 101[5] -> 99[3] via P2P/IPC
+ip-26-0-161-103:96893:97067 [5] NCCL INFO Channel 13/0 : 21[5] -> 19[3] via P2P/IPC
+ip-26-0-161-123:32706:32877 [2] NCCL INFO Channel 10/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-168-34:272152:272331 [2] NCCL INFO Channel 05/0 : 114[2] -> 113[1] via P2P/IPC
+ip-26-0-161-138:96564:96757 [1] NCCL INFO Channel 03/0 : 33[1] -> 32[0] via P2P/IPC
+ip-26-0-161-142:62885:63054 [4] NCCL INFO Channel 11/0 : 44[4] -> 51[3] [send] via NET/Libfabric/3(43)/GDRDMA
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO Channel 10/0 : 73[1] -> 72[0] via P2P/IPC
+ip-26-0-160-225:124164:124336 [4] NCCL INFO Channel 07/0 : 4[4] -> 15[7] [send] via NET/Libfabric/7(7)/GDRDMA
+ip-26-0-168-34:272151:272326 [1] NCCL INFO Channel 06/0 : 113[1] -> 112[0] via P2P/IPC
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO Channel 13/0 : 13[5] -> 11[3] via P2P/IPC
+ip-26-0-160-225:124161:124337 [1] NCCL INFO Channel 14/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-78:114514:114689 [2] NCCL INFO Channel 12/0 : 66[2] -> 65[1] via P2P/IPC
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO Channel 06/0 : 102[6] -> 99[3] via P2P/IPC
+ip-26-0-168-30:235654:235821 [6] NCCL INFO Channel 06/0 : 110[6] -> 107[3] via P2P/IPC
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO Channel 03/0 : 74[2] -> 73[1] via P2P/IPC
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO Channel 07/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-161-138:96564:96757 [1] NCCL INFO Channel 04/0 : 33[1] -> 32[0] via P2P/IPC
+ip-26-0-168-34:272152:272331 [2] NCCL INFO Channel 06/0 : 114[2] -> 113[1] via P2P/IPC
+ip-26-0-168-30:235650:235824 [2] NCCL INFO Channel 07/0 : 106[2] -> 105[1] via P2P/IPC
+ip-26-0-168-30:235653:235825 [5] NCCL INFO Channel 05/0 : 109[5] -> 107[3] via P2P/IPC
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO Channel 10/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-168-34:272151:272326 [1] NCCL INFO Channel 07/0 : 113[1] -> 112[0] via P2P/IPC
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO Channel 11/0 : 73[1] -> 72[0] via P2P/IPC
+ip-26-0-168-52:2722051:2722224 [6] NCCL INFO Channel 14/0 : 126[6] -> 123[3] via P2P/IPC
+ip-26-0-161-138:96564:96757 [1] NCCL INFO Channel 05/0 : 33[1] -> 32[0] via P2P/IPC
+ip-26-0-161-123:32705:32880 [1] NCCL INFO Channel 10/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-161-123:32710:32883 [6] NCCL INFO Channel 14/0 : 30[6] -> 27[3] via P2P/IPC
+ip-26-0-168-34:272152:272331 [2] NCCL INFO Channel 07/0 : 114[2] -> 113[1] via P2P/IPC
+ip-26-0-161-138:96568:96753 [5] NCCL INFO Channel 05/0 : 37[5] -> 35[3] via P2P/IPC
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO Channel 04/0 : 74[2] -> 73[1] via P2P/IPC
+ip-26-0-160-225:124161:124337 [1] NCCL INFO Channel 15/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-123:32706:32877 [2] NCCL INFO Channel 11/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-168-52:2722046:2722223 [1] NCCL INFO Channel 01/0 : 121[1] -> 120[0] via P2P/IPC
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO Channel 09/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO Channel 14/0 : 102[6] -> 99[3] via P2P/IPC
+ip-26-0-168-34:272151:272326 [1] NCCL INFO Channel 09/0 : 113[1] -> 112[0] via P2P/IPC
+ip-26-0-161-78:114514:114689 [2] NCCL INFO Channel 13/0 : 66[2] -> 65[1] via P2P/IPC
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO Channel 12/0 : 73[1] -> 72[0] via P2P/IPC
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO Channel 01/0 : 42[2] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96564:96757 [1] NCCL INFO Channel 06/0 : 33[1] -> 32[0] via P2P/IPC
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO Channel 11/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-168-34:272152:272331 [2] NCCL INFO Channel 10/0 : 114[2] -> 113[1] via P2P/IPC
+ip-26-0-160-225:124162:124342 [2] NCCL INFO Channel 02/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-78:114513:114688 [1] NCCL INFO Channel 01/0 : 65[1] -> 64[0] via P2P/IPC
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO Channel 09/0 : 42[2] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO Channel 10/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-161-123:32709:32882 [5] NCCL INFO Channel 13/0 : 29[5] -> 27[3] via P2P/IPC
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO Channel 05/0 : 74[2] -> 73[1] via P2P/IPC
+ip-26-0-168-34:272151:272326 [1] NCCL INFO Channel 10/0 : 113[1] -> 112[0] via P2P/IPC
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO Channel 01/0 : 50[2] -> 57[1] [send] via NET/Libfabric/1(49)/GDRDMA
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO Channel 12/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-161-123:32705:32880 [1] NCCL INFO Channel 11/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-168-34:272152:272331 [2] NCCL INFO Channel 11/0 : 114[2] -> 113[1] via P2P/IPC
+ip-26-0-168-52:2722046:2722223 [1] NCCL INFO Channel 02/0 : 121[1] -> 120[0] via P2P/IPC
+ip-26-0-161-123:32706:32877 [2] NCCL INFO Channel 12/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-161-138:96564:96757 [1] NCCL INFO Channel 07/0 : 33[1] -> 32[0] via P2P/IPC
+ip-26-0-161-138:96568:96753 [5] NCCL INFO Channel 13/0 : 37[5] -> 35[3] via P2P/IPC
+ip-26-0-161-78:114514:114689 [2] NCCL INFO Channel 14/0 : 66[2] -> 65[1] via P2P/IPC
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO Channel 13/0 : 73[1] -> 72[0] via P2P/IPC
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO Channel 11/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-161-78:114513:114688 [1] NCCL INFO Channel 02/0 : 65[1] -> 64[0] via P2P/IPC
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO Channel 01/0 : 81[1] -> 80[0] via P2P/IPC
+ip-26-0-168-34:272151:272326 [1] NCCL INFO Channel 11/0 : 113[1] -> 112[0] via P2P/IPC
+ip-26-0-160-225:124162:124342 [2] NCCL INFO Channel 03/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO Channel 13/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO Channel 11/0 : 124[4] -> 3[3] [send] via NET/Libfabric/3(123)/GDRDMA
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO Channel 06/0 : 74[2] -> 73[1] via P2P/IPC
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO Channel 09/0 : 50[2] -> 57[1] [send] via NET/Libfabric/1(49)/GDRDMA
+ip-26-0-161-142:62886:63057 [5] NCCL INFO Channel 05/0 : 38[6] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272152:272331 [2] NCCL INFO Channel 12/0 : 114[2] -> 113[1] via P2P/IPC
+ip-26-0-161-103:96892:97064 [4] NCCL INFO Channel 07/0 : 20[4] -> 31[7] [send] via NET/Libfabric/7(23)/GDRDMA
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO Channel 12/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-161-138:96564:96757 [1] NCCL INFO Channel 09/0 : 33[1] -> 32[0] via P2P/IPC
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO Channel 02/0 : 81[1] -> 80[0] via P2P/IPC
+ip-26-0-161-142:62886:63057 [5] NCCL INFO Channel 13/0 : 38[6] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO Channel 14/0 : 73[1] -> 72[0] via P2P/IPC
+ip-26-0-160-225:124162:124342 [2] NCCL INFO Channel 04/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-168-34:272151:272326 [1] NCCL INFO Channel 12/0 : 113[1] -> 112[0] via P2P/IPC
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO Channel 14/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-161-123:32705:32880 [1] NCCL INFO Channel 12/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-161-78:114514:114689 [2] NCCL INFO Channel 15/0 : 66[2] -> 65[1] via P2P/IPC
+ip-26-0-161-123:32706:32877 [2] NCCL INFO Channel 13/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-168-34:272152:272331 [2] NCCL INFO Channel 13/0 : 114[2] -> 113[1] via P2P/IPC
+ip-26-0-168-52:2722046:2722223 [1] NCCL INFO Channel 03/0 : 121[1] -> 120[0] via P2P/IPC
+ip-26-0-161-78:114513:114688 [1] NCCL INFO Channel 03/0 : 65[1] -> 64[0] via P2P/IPC
+ip-26-0-161-142:62887:63055 [6] NCCL INFO Channel 06/0 : 39[7] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96564:96757 [1] NCCL INFO Channel 10/0 : 33[1] -> 32[0] via P2P/IPC
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO Channel 03/0 : 81[1] -> 80[0] via P2P/IPC
+ip-26-0-168-52:2722050:2722225 [5] NCCL INFO Channel 13/0 : 125[5] -> 123[3] via P2P/IPC
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO Channel 07/0 : 74[2] -> 73[1] via P2P/IPC
+ip-26-0-161-138:96567:96760 [4] NCCL INFO Channel 07/0 : 36[4] -> 47[7] [send] via NET/Libfabric/7(39)/GDRDMA
+ip-26-0-161-142:62886:63057 [5] NCCL INFO Channel 04/0 : 45[5] -> 52[4] [send] via NET/Libfabric/4(44)/GDRDMA
+ip-26-0-160-225:124162:124342 [2] NCCL INFO Channel 05/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-103:96890:97070 [2] NCCL INFO Channel 03/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-161-138:96567:96760 [4] NCCL INFO Channel 15/0 : 36[4] -> 47[7] [send] via NET/Libfabric/7(39)/GDRDMA
+ip-26-0-168-34:272155:272329 [5] NCCL INFO Channel 05/0 : 117[5] -> 115[3] via P2P/IPC
+ip-26-0-161-142:62887:63055 [6] NCCL INFO Channel 14/0 : 39[7] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO Channel 15/0 : 73[1] -> 72[0] via P2P/IPC
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO Channel 13/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-161-103:96889:97065 [1] NCCL INFO Channel 01/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-161-103:96894:97068 [6] NCCL INFO Channel 14/0 : 22[6] -> 19[3] via P2P/IPC
+ip-26-0-168-52:2722047:2722221 [2] NCCL INFO Channel 13/0 : 122[2] -> 121[1] via P2P/IPC
+ip-26-0-161-142:62886:63057 [5] NCCL INFO Channel 12/0 : 45[5] -> 52[4] [send] via NET/Libfabric/4(44)/GDRDMA
+ip-26-0-161-138:96564:96757 [1] NCCL INFO Channel 11/0 : 33[1] -> 32[0] via P2P/IPC
+ip-26-0-161-123:32706:32877 [2] NCCL INFO Channel 14/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO Channel 15/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO Channel 04/0 : 81[1] -> 80[0] via P2P/IPC
+ip-26-0-161-123:32705:32880 [1] NCCL INFO Channel 13/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-161-142:62887:63055 [6] NCCL INFO Channel 05/0 : 46[6] -> 53[5] [send] via NET/Libfabric/5(45)/GDRDMA
+ip-26-0-160-225:124162:124342 [2] NCCL INFO Channel 06/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-142:62887:63055 [6] NCCL INFO Channel 13/0 : 46[6] -> 53[5] [send] via NET/Libfabric/5(45)/GDRDMA
+ip-26-0-161-78:114513:114688 [1] NCCL INFO Channel 04/0 : 65[1] -> 64[0] via P2P/IPC
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO Channel 10/0 : 74[2] -> 73[1] via P2P/IPC
+ip-26-0-161-138:96564:96757 [1] NCCL INFO Channel 12/0 : 33[1] -> 32[0] via P2P/IPC
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO Channel 05/0 : 81[1] -> 80[0] via P2P/IPC
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO Channel 14/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-160-225:124162:124342 [2] NCCL INFO Channel 07/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-123:32706:32877 [2] NCCL INFO Channel 15/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-168-52:2722046:2722223 [1] NCCL INFO Channel 04/0 : 121[1] -> 120[0] via P2P/IPC
+ip-26-0-168-34:272152:272331 [2] NCCL INFO Channel 14/0 : 114[2] -> 113[1] via P2P/IPC
+ip-26-0-168-34:272151:272326 [1] NCCL INFO Channel 13/0 : 113[1] -> 112[0] via P2P/IPC
+ip-26-0-161-138:96569:96758 [6] NCCL INFO Channel 06/0 : 38[6] -> 35[3] via P2P/IPC
+ip-26-0-161-123:32705:32880 [1] NCCL INFO Channel 14/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-161-103:96890:97070 [2] NCCL INFO Channel 04/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO Channel 11/0 : 74[2] -> 73[1] via P2P/IPC
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO Channel 06/0 : 81[1] -> 80[0] via P2P/IPC
+ip-26-0-168-52:2722047:2722221 [2] NCCL INFO Channel 14/0 : 122[2] -> 121[1] via P2P/IPC
+ip-26-0-161-138:96570:96759 [7] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO Channel 15/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-161-78:114513:114688 [1] NCCL INFO Channel 05/0 : 65[1] -> 64[0] via P2P/IPC
+ip-26-0-161-138:96564:96757 [1] NCCL INFO Channel 13/0 : 33[1] -> 32[0] via P2P/IPC
+ip-26-0-161-142:62885:63054 [4] NCCL INFO Channel 07/0 : 44[4] -> 55[7] [send] via NET/Libfabric/7(47)/GDRDMA
+ip-26-0-160-225:124162:124342 [2] NCCL INFO Channel 10/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-103:96889:97065 [1] NCCL INFO Channel 02/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-161-142:62885:63054 [4] NCCL INFO Channel 15/0 : 44[4] -> 55[7] [send] via NET/Libfabric/7(47)/GDRDMA
+ip-26-0-168-52:2722046:2722223 [1] NCCL INFO Channel 05/0 : 121[1] -> 120[0] via P2P/IPC
+ip-26-0-161-123:32705:32880 [1] NCCL INFO Channel 15/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO Channel 07/0 : 81[1] -> 80[0] via P2P/IPC
+ip-26-0-168-34:272155:272329 [5] NCCL INFO Channel 13/0 : 117[5] -> 115[3] via P2P/IPC
+ip-26-0-161-138:96569:96758 [6] NCCL INFO Channel 14/0 : 38[6] -> 35[3] via P2P/IPC
+ip-26-0-161-103:96890:97070 [2] NCCL INFO Channel 05/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-161-78:114513:114688 [1] NCCL INFO Channel 06/0 : 65[1] -> 64[0] via P2P/IPC
+ip-26-0-160-225:124162:124342 [2] NCCL INFO Channel 11/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-138:96564:96757 [1] NCCL INFO Channel 14/0 : 33[1] -> 32[0] via P2P/IPC
+ip-26-0-168-34:272151:272326 [1] NCCL INFO Channel 14/0 : 113[1] -> 112[0] via P2P/IPC
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO Channel 09/0 : 81[1] -> 80[0] via P2P/IPC
+ip-26-0-168-52:2722047:2722221 [2] NCCL INFO Channel 15/0 : 122[2] -> 121[1] via P2P/IPC
+ip-26-0-168-34:272152:272331 [2] NCCL INFO Channel 15/0 : 114[2] -> 113[1] via P2P/IPC
+ip-26-0-161-103:96889:97065 [1] NCCL INFO Channel 03/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-168-52:2722046:2722223 [1] NCCL INFO Channel 06/0 : 121[1] -> 120[0] via P2P/IPC
+ip-26-0-160-225:124162:124342 [2] NCCL INFO Channel 12/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-103:96890:97070 [2] NCCL INFO Channel 06/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-161-78:114513:114688 [1] NCCL INFO Channel 07/0 : 65[1] -> 64[0] via P2P/IPC
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO Channel 10/0 : 81[1] -> 80[0] via P2P/IPC
+ip-26-0-161-138:96564:96757 [1] NCCL INFO Channel 15/0 : 33[1] -> 32[0] via P2P/IPC
+ip-26-0-168-34:272151:272326 [1] NCCL INFO Channel 15/0 : 113[1] -> 112[0] via P2P/IPC
+ip-26-0-160-225:124162:124342 [2] NCCL INFO Channel 13/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-78:114513:114688 [1] NCCL INFO Channel 09/0 : 65[1] -> 64[0] via P2P/IPC
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO Channel 11/0 : 81[1] -> 80[0] via P2P/IPC
+ip-26-0-168-30:235650:235824 [2] NCCL INFO Channel 10/0 : 106[2] -> 105[1] via P2P/IPC
+ip-26-0-161-103:96889:97065 [1] NCCL INFO Channel 04/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-168-30:235654:235821 [6] NCCL INFO Channel 14/0 : 110[6] -> 107[3] via P2P/IPC
+ip-26-0-161-103:96890:97070 [2] NCCL INFO Channel 07/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO Channel 12/0 : 81[1] -> 80[0] via P2P/IPC
+ip-26-0-161-78:114513:114688 [1] NCCL INFO Channel 10/0 : 65[1] -> 64[0] via P2P/IPC
+ip-26-0-161-103:96889:97065 [1] NCCL INFO Channel 05/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-161-103:96890:97070 [2] NCCL INFO Channel 10/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO Channel 13/0 : 81[1] -> 80[0] via P2P/IPC
+ip-26-0-161-78:114513:114688 [1] NCCL INFO Channel 11/0 : 65[1] -> 64[0] via P2P/IPC
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO Channel 14/0 : 81[1] -> 80[0] via P2P/IPC
+ip-26-0-161-78:114513:114688 [1] NCCL INFO Channel 12/0 : 65[1] -> 64[0] via P2P/IPC
+ip-26-0-161-103:96889:97065 [1] NCCL INFO Channel 06/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-161-103:96890:97070 [2] NCCL INFO Channel 11/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-161-103:96889:97065 [1] NCCL INFO Channel 07/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-161-78:114513:114688 [1] NCCL INFO Channel 13/0 : 65[1] -> 64[0] via P2P/IPC
+ip-26-0-161-103:96890:97070 [2] NCCL INFO Channel 12/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-161-103:96889:97065 [1] NCCL INFO Channel 09/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-168-30:235653:235825 [5] NCCL INFO Channel 13/0 : 109[5] -> 107[3] via P2P/IPC
+ip-26-0-161-103:96890:97070 [2] NCCL INFO Channel 13/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-161-103:96889:97065 [1] NCCL INFO Channel 10/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-168-30:235650:235824 [2] NCCL INFO Channel 11/0 : 106[2] -> 105[1] via P2P/IPC
+ip-26-0-161-103:96890:97070 [2] NCCL INFO Channel 14/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO Channel 15/0 : 81[1] -> 80[0] via P2P/IPC
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO Channel 01/0 : 57[1] -> 56[0] via P2P/IPC
+ip-26-0-161-103:96890:97070 [2] NCCL INFO Channel 15/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO Channel 02/0 : 57[1] -> 56[0] via P2P/IPC
+ip-26-0-161-103:96889:97065 [1] NCCL INFO Channel 11/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-161-78:114513:114688 [1] NCCL INFO Channel 14/0 : 65[1] -> 64[0] via P2P/IPC
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO Channel 04/0 : 45[5] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO Channel 03/0 : 57[1] -> 56[0] via P2P/IPC
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO Channel 06/0 : 47[7] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO Channel 12/0 : 45[5] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114513:114688 [1] NCCL INFO Channel 15/0 : 65[1] -> 64[0] via P2P/IPC
+ip-26-0-161-103:96889:97065 [1] NCCL INFO Channel 12/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO Channel 05/0 : 46[6] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO Channel 04/0 : 57[1] -> 56[0] via P2P/IPC
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO Channel 14/0 : 55[7] -> 62[6] [send] via NET/Libfabric/6(54)/GDRDMA
+ip-26-0-160-225:124162:124342 [2] NCCL INFO Channel 14/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO Channel 14/0 : 47[7] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO Channel 13/0 : 46[6] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO Channel 05/0 : 57[1] -> 56[0] via P2P/IPC
+ip-26-0-161-103:96889:97065 [1] NCCL INFO Channel 13/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO Channel 03/0 : 52[4] -> 59[3] [send] via NET/Libfabric/3(51)/GDRDMA
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO Channel 05/0 : 54[6] -> 61[5] [send] via NET/Libfabric/5(53)/GDRDMA
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO Channel 04/0 : 53[5] -> 60[4] [send] via NET/Libfabric/4(52)/GDRDMA
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO Channel 01/0 : 49[1] -> 48[0] via P2P/IPC
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO Channel 06/0 : 57[1] -> 56[0] via P2P/IPC
+ip-26-0-161-103:96889:97065 [1] NCCL INFO Channel 14/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO Channel 13/0 : 54[6] -> 61[5] [send] via NET/Libfabric/5(53)/GDRDMA
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO Channel 11/0 : 52[4] -> 59[3] [send] via NET/Libfabric/3(51)/GDRDMA
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO Channel 12/0 : 53[5] -> 60[4] [send] via NET/Libfabric/4(52)/GDRDMA
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO Channel 07/0 : 57[1] -> 56[0] via P2P/IPC
+ip-26-0-161-103:96889:97065 [1] NCCL INFO Channel 15/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO Channel 06/0 : 62[6] -> 59[3] via P2P/IPC
+ip-26-0-161-142:62888:63060 [7] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO Channel 02/0 : 49[1] -> 48[0] via P2P/IPC
+ip-26-0-161-142:62886:63057 [5] NCCL INFO Channel 05/0 : 45[5] -> 43[3] via P2P/IPC
+ip-26-0-161-142:62887:63055 [6] NCCL INFO Channel 06/0 : 46[6] -> 43[3] via P2P/IPC
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO Channel 09/0 : 57[1] -> 56[0] via P2P/IPC
+ip-26-0-168-52:2722046:2722223 [1] NCCL INFO Channel 07/0 : 121[1] -> 120[0] via P2P/IPC
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO Channel 05/0 : 61[5] -> 59[3] via P2P/IPC
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO Channel 02/0 : 50[2] -> 49[1] via P2P/IPC
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO Channel 06/0 : 54[6] -> 51[3] via P2P/IPC
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO Channel 05/0 : 53[5] -> 51[3] via P2P/IPC
+ip-26-0-168-52:2722046:2722223 [1] NCCL INFO Channel 09/0 : 121[1] -> 120[0] via P2P/IPC
+ip-26-0-160-225:124162:124342 [2] NCCL INFO Channel 15/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO Channel 10/0 : 57[1] -> 56[0] via P2P/IPC
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO Channel 14/0 : 62[6] -> 59[3] via P2P/IPC
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO Channel 03/0 : 49[1] -> 48[0] via P2P/IPC
+ip-26-0-168-52:2722046:2722223 [1] NCCL INFO Channel 10/0 : 121[1] -> 120[0] via P2P/IPC
+ip-26-0-168-52:2722046:2722223 [1] NCCL INFO Channel 11/0 : 121[1] -> 120[0] via P2P/IPC
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO Channel 13/0 : 61[5] -> 59[3] via P2P/IPC
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO Channel 03/0 : 50[2] -> 49[1] via P2P/IPC
+ip-26-0-168-52:2722046:2722223 [1] NCCL INFO Channel 12/0 : 121[1] -> 120[0] via P2P/IPC
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO Channel 11/0 : 57[1] -> 56[0] via P2P/IPC
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO Channel 14/0 : 54[6] -> 51[3] via P2P/IPC
+ip-26-0-168-52:2722046:2722223 [1] NCCL INFO Channel 13/0 : 121[1] -> 120[0] via P2P/IPC
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO Channel 13/0 : 53[5] -> 51[3] via P2P/IPC
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO Channel 12/0 : 57[1] -> 56[0] via P2P/IPC
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO Channel 04/0 : 49[1] -> 48[0] via P2P/IPC
+ip-26-0-168-52:2722046:2722223 [1] NCCL INFO Channel 14/0 : 121[1] -> 120[0] via P2P/IPC
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO Channel 12/0 : 74[2] -> 73[1] via P2P/IPC
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO Channel 13/0 : 57[1] -> 56[0] via P2P/IPC
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO Channel 04/0 : 50[2] -> 49[1] via P2P/IPC
+ip-26-0-168-52:2722046:2722223 [1] NCCL INFO Channel 15/0 : 121[1] -> 120[0] via P2P/IPC
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO Channel 14/0 : 57[1] -> 56[0] via P2P/IPC
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO Channel 15/0 : 57[1] -> 56[0] via P2P/IPC
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO Channel 05/0 : 49[1] -> 48[0] via P2P/IPC
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO Channel 05/0 : 50[2] -> 49[1] via P2P/IPC
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO Channel 06/0 : 49[1] -> 48[0] via P2P/IPC
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO Channel 06/0 : 50[2] -> 49[1] via P2P/IPC
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO Channel 07/0 : 49[1] -> 48[0] via P2P/IPC
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO Channel 13/0 : 74[2] -> 73[1] via P2P/IPC
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO Channel 07/0 : 50[2] -> 49[1] via P2P/IPC
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO Channel 09/0 : 49[1] -> 48[0] via P2P/IPC
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO Channel 10/0 : 50[2] -> 49[1] via P2P/IPC
+ip-26-0-161-142:62887:63055 [6] NCCL INFO Channel 14/0 : 46[6] -> 43[3] via P2P/IPC
+ip-26-0-161-142:62886:63057 [5] NCCL INFO Channel 13/0 : 45[5] -> 43[3] via P2P/IPC
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO Channel 10/0 : 49[1] -> 48[0] via P2P/IPC
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO Channel 11/0 : 50[2] -> 49[1] via P2P/IPC
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO Channel 11/0 : 49[1] -> 48[0] via P2P/IPC
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO Channel 12/0 : 50[2] -> 49[1] via P2P/IPC
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO Channel 12/0 : 49[1] -> 48[0] via P2P/IPC
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO Channel 13/0 : 50[2] -> 49[1] via P2P/IPC
+ip-26-0-168-30:235650:235824 [2] NCCL INFO Channel 12/0 : 106[2] -> 105[1] via P2P/IPC
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO Channel 13/0 : 49[1] -> 48[0] via P2P/IPC
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO Channel 14/0 : 50[2] -> 49[1] via P2P/IPC
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO Channel 14/0 : 49[1] -> 48[0] via P2P/IPC
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO Channel 14/0 : 74[2] -> 73[1] via P2P/IPC
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO Channel 15/0 : 50[2] -> 49[1] via P2P/IPC
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO Channel 15/0 : 49[1] -> 48[0] via P2P/IPC
+ip-26-0-168-30:235650:235824 [2] NCCL INFO Channel 13/0 : 106[2] -> 105[1] via P2P/IPC
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO Channel 15/0 : 74[2] -> 73[1] via P2P/IPC
+ip-26-0-168-30:235650:235824 [2] NCCL INFO Channel 14/0 : 106[2] -> 105[1] via P2P/IPC
+ip-26-0-168-30:235650:235824 [2] NCCL INFO Channel 15/0 : 106[2] -> 105[1] via P2P/IPC
+ip-26-0-161-142:62883:63058 [2] NCCL INFO Channel 02/0 : 42[2] -> 41[1] via P2P/IPC
+ip-26-0-161-142:62883:63058 [2] NCCL INFO Channel 03/0 : 42[2] -> 41[1] via P2P/IPC
+ip-26-0-161-142:62883:63058 [2] NCCL INFO Channel 04/0 : 42[2] -> 41[1] via P2P/IPC
+ip-26-0-161-142:62883:63058 [2] NCCL INFO Channel 05/0 : 42[2] -> 41[1] via P2P/IPC
+ip-26-0-161-142:62883:63058 [2] NCCL INFO Channel 06/0 : 42[2] -> 41[1] via P2P/IPC
+ip-26-0-161-142:62883:63058 [2] NCCL INFO Channel 07/0 : 42[2] -> 41[1] via P2P/IPC
+ip-26-0-161-142:62883:63058 [2] NCCL INFO Channel 10/0 : 42[2] -> 41[1] via P2P/IPC
+ip-26-0-161-142:62883:63058 [2] NCCL INFO Channel 11/0 : 42[2] -> 41[1] via P2P/IPC
+ip-26-0-161-142:62883:63058 [2] NCCL INFO Channel 12/0 : 42[2] -> 41[1] via P2P/IPC
+ip-26-0-161-142:62883:63058 [2] NCCL INFO Channel 13/0 : 42[2] -> 41[1] via P2P/IPC
+ip-26-0-161-142:62883:63058 [2] NCCL INFO Channel 14/0 : 42[2] -> 41[1] via P2P/IPC
+ip-26-0-161-142:62883:63058 [2] NCCL INFO Channel 15/0 : 42[2] -> 41[1] via P2P/IPC
+ip-26-0-165-202:153569:153739 [7] NCCL INFO Channel 07/0 : 84[4] -> 95[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153569:153739 [7] NCCL INFO Channel 15/0 : 84[4] -> 95[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153569:153739 [7] NCCL INFO Channel 07/0 : 95[7] -> 91[3] via P2P/IPC
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO Channel 07/0 : 52[4] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO Channel 07/0 : 60[4] -> 71[7] [send] via NET/Libfabric/7(63)/GDRDMA
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO Channel 15/0 : 52[4] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO Channel 15/0 : 60[4] -> 71[7] [send] via NET/Libfabric/7(63)/GDRDMA
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO Channel 07/0 : 76[4] -> 87[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO Channel 15/0 : 76[4] -> 87[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO Channel 07/0 : 4[4] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO Channel 15/0 : 4[4] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO Channel 07/0 : 12[4] -> 23[7] [send] via NET/Libfabric/7(15)/GDRDMA
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO Channel 15/0 : 12[4] -> 23[7] [send] via NET/Libfabric/7(15)/GDRDMA
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO Channel 07/0 : 68[4] -> 79[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO Channel 15/0 : 68[4] -> 79[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO Channel 15/0 : 76[4] -> 87[7] [send] via NET/Libfabric/7(79)/GDRDMA
+ip-26-0-168-34:272157:272324 [7] NCCL INFO Channel 07/0 : 108[4] -> 119[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272157:272324 [7] NCCL INFO Channel 15/0 : 108[4] -> 119[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272154:272330 [4] NCCL INFO Channel 07/0 : 116[4] -> 127[7] [send] via NET/Libfabric/7(119)/GDRDMA
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO Channel 07/0 : 87[7] -> 83[3] via P2P/IPC
+ip-26-0-168-34:272154:272330 [4] NCCL INFO Channel 15/0 : 116[4] -> 127[7] [send] via NET/Libfabric/7(119)/GDRDMA
+ip-26-0-161-78:114519:114692 [7] NCCL INFO Channel 07/0 : 60[4] -> 71[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114519:114692 [7] NCCL INFO Channel 15/0 : 60[4] -> 71[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114516:114691 [4] NCCL INFO Channel 07/0 : 68[4] -> 79[7] [send] via NET/Libfabric/7(71)/GDRDMA
+ip-26-0-161-78:114516:114691 [4] NCCL INFO Channel 15/0 : 68[4] -> 79[7] [send] via NET/Libfabric/7(71)/GDRDMA
+ip-26-0-161-78:114519:114692 [7] NCCL INFO Channel 07/0 : 71[7] -> 67[3] via P2P/IPC
+ip-26-0-160-225:124167:124338 [7] NCCL INFO Channel 07/0 : 124[4] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124164:124336 [4] NCCL INFO Channel 15/0 : 4[4] -> 15[7] [send] via NET/Libfabric/7(7)/GDRDMA
+ip-26-0-160-225:124167:124338 [7] NCCL INFO Channel 15/0 : 124[4] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153569:153739 [7] NCCL INFO Channel 15/0 : 95[7] -> 91[3] via P2P/IPC
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO Channel 07/0 : 15[7] -> 11[3] via P2P/IPC
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO Channel 07/0 : 79[7] -> 75[3] via P2P/IPC
+ip-26-0-168-52:2722052:2722226 [7] NCCL INFO Channel 07/0 : 116[4] -> 127[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722052:2722226 [7] NCCL INFO Channel 15/0 : 116[4] -> 127[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO Channel 07/0 : 124[4] -> 7[7] [send] via NET/Libfabric/7(127)/GDRDMA
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO Channel 15/0 : 124[4] -> 7[7] [send] via NET/Libfabric/7(127)/GDRDMA
+ip-26-0-161-103:96895:97069 [7] NCCL INFO Channel 07/0 : 12[4] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235655:235828 [7] NCCL INFO Channel 07/0 : 100[4] -> 111[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96892:97064 [4] NCCL INFO Channel 15/0 : 20[4] -> 31[7] [send] via NET/Libfabric/7(23)/GDRDMA
+ip-26-0-168-52:2722052:2722226 [7] NCCL INFO Channel 07/0 : 127[7] -> 123[3] via P2P/IPC
+ip-26-0-161-103:96895:97069 [7] NCCL INFO Channel 15/0 : 12[4] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124167:124338 [7] NCCL INFO Channel 07/0 : 7[7] -> 3[3] via P2P/IPC
+ip-26-0-161-138:96570:96759 [7] NCCL INFO Channel 07/0 : 28[4] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235652:235822 [4] NCCL INFO Channel 07/0 : 108[4] -> 119[7] [send] via NET/Libfabric/7(111)/GDRDMA
+ip-26-0-168-30:235655:235828 [7] NCCL INFO Channel 15/0 : 100[4] -> 111[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96570:96759 [7] NCCL INFO Channel 15/0 : 28[4] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235652:235822 [4] NCCL INFO Channel 15/0 : 108[4] -> 119[7] [send] via NET/Libfabric/7(111)/GDRDMA
+ip-26-0-161-103:96895:97069 [7] NCCL INFO Channel 07/0 : 23[7] -> 19[3] via P2P/IPC
+ip-26-0-168-30:235655:235828 [7] NCCL INFO Channel 07/0 : 111[7] -> 107[3] via P2P/IPC
+ip-26-0-161-138:96570:96759 [7] NCCL INFO Channel 07/0 : 39[7] -> 35[3] via P2P/IPC
+ip-26-0-168-34:272157:272324 [7] NCCL INFO Channel 07/0 : 119[7] -> 115[3] via P2P/IPC
+ip-26-0-161-78:114519:114692 [7] NCCL INFO Channel 15/0 : 71[7] -> 67[3] via P2P/IPC
+ip-26-0-161-123:32711:32884 [7] NCCL INFO Channel 07/0 : 20[4] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32711:32884 [7] NCCL INFO Channel 15/0 : 20[4] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32711:32884 [7] NCCL INFO Channel 07/0 : 31[7] -> 27[3] via P2P/IPC
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO Channel 07/0 : 92[4] -> 103[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO Channel 15/0 : 92[4] -> 103[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235655:235828 [7] NCCL INFO Channel 15/0 : 111[7] -> 107[3] via P2P/IPC
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO Channel 07/0 : 103[7] -> 99[3] via P2P/IPC
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO Channel 15/0 : 87[7] -> 83[3] via P2P/IPC
+ip-26-0-168-52:2722052:2722226 [7] NCCL INFO Channel 15/0 : 127[7] -> 123[3] via P2P/IPC
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO Channel 15/0 : 79[7] -> 75[3] via P2P/IPC
+ip-26-0-165-202:153569:153739 [7] NCCL INFO Channel 01/0 : 95[7] -> 94[6] via P2P/IPC
+ip-26-0-168-30:235655:235828 [7] NCCL INFO Channel 01/0 : 111[7] -> 110[6] via P2P/IPC
+ip-26-0-161-142:62888:63060 [7] NCCL INFO Channel 07/0 : 36[4] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62888:63060 [7] NCCL INFO Channel 15/0 : 36[4] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62888:63060 [7] NCCL INFO Channel 07/0 : 47[7] -> 43[3] via P2P/IPC
+ip-26-0-161-78:114519:114692 [7] NCCL INFO Channel 01/0 : 71[7] -> 70[6] via P2P/IPC
+ip-26-0-161-123:32711:32884 [7] NCCL INFO Channel 15/0 : 31[7] -> 27[3] via P2P/IPC
+ip-26-0-168-30:235655:235828 [7] NCCL INFO Channel 02/0 : 111[7] -> 110[6] via P2P/IPC
+ip-26-0-168-34:272157:272324 [7] NCCL INFO Channel 15/0 : 119[7] -> 115[3] via P2P/IPC
+ip-26-0-165-202:153569:153739 [7] NCCL INFO Channel 02/0 : 95[7] -> 94[6] via P2P/IPC
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO Channel 01/0 : 87[7] -> 86[6] via P2P/IPC
+ip-26-0-168-52:2722052:2722226 [7] NCCL INFO Channel 01/0 : 127[7] -> 126[6] via P2P/IPC
+ip-26-0-161-142:62888:63060 [7] NCCL INFO Channel 15/0 : 47[7] -> 43[3] via P2P/IPC
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO Channel 15/0 : 103[7] -> 99[3] via P2P/IPC
+ip-26-0-160-225:124167:124338 [7] NCCL INFO Channel 15/0 : 7[7] -> 3[3] via P2P/IPC
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO Channel 01/0 : 79[7] -> 78[6] via P2P/IPC
+ip-26-0-161-142:62888:63060 [7] NCCL INFO Channel 01/0 : 47[7] -> 46[6] via P2P/IPC
+ip-26-0-168-30:235655:235828 [7] NCCL INFO Channel 03/0 : 111[7] -> 110[6] via P2P/IPC
+ip-26-0-161-103:96895:97069 [7] NCCL INFO Channel 15/0 : 23[7] -> 19[3] via P2P/IPC
+ip-26-0-161-78:114519:114692 [7] NCCL INFO Channel 02/0 : 71[7] -> 70[6] via P2P/IPC
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO Channel 02/0 : 87[7] -> 86[6] via P2P/IPC
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO Channel 07/0 : 44[4] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO Channel 07/0 : 52[4] -> 63[7] [send] via NET/Libfabric/7(55)/GDRDMA
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO Channel 15/0 : 15[7] -> 11[3] via P2P/IPC
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO Channel 15/0 : 44[4] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62888:63060 [7] NCCL INFO Channel 02/0 : 47[7] -> 46[6] via P2P/IPC
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO Channel 15/0 : 52[4] -> 63[7] [send] via NET/Libfabric/7(55)/GDRDMA
+ip-26-0-161-123:32711:32884 [7] NCCL INFO Channel 01/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-161-138:96570:96759 [7] NCCL INFO Channel 15/0 : 39[7] -> 35[3] via P2P/IPC
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO Channel 07/0 : 55[7] -> 51[3] via P2P/IPC
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO Channel 07/0 : 63[7] -> 59[3] via P2P/IPC
+ip-26-0-168-34:272157:272324 [7] NCCL INFO Channel 01/0 : 119[7] -> 118[6] via P2P/IPC
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO Channel 02/0 : 79[7] -> 78[6] via P2P/IPC
+ip-26-0-165-202:153569:153739 [7] NCCL INFO Channel 03/0 : 95[7] -> 94[6] via P2P/IPC
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO Channel 03/0 : 87[7] -> 86[6] via P2P/IPC
+ip-26-0-168-30:235655:235828 [7] NCCL INFO Channel 04/0 : 111[7] -> 110[6] via P2P/IPC
+ip-26-0-165-202:153569:153739 [7] NCCL INFO Channel 04/0 : 95[7] -> 94[6] via P2P/IPC
+ip-26-0-168-52:2722052:2722226 [7] NCCL INFO Channel 02/0 : 127[7] -> 126[6] via P2P/IPC
+ip-26-0-161-103:96895:97069 [7] NCCL INFO Channel 01/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO Channel 01/0 : 103[7] -> 102[6] via P2P/IPC
+ip-26-0-165-202:153569:153739 [7] NCCL INFO Channel 05/0 : 95[7] -> 94[6] via P2P/IPC
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO Channel 01/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-161-142:62888:63060 [7] NCCL INFO Channel 03/0 : 47[7] -> 46[6] via P2P/IPC
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO Channel 02/0 : 103[7] -> 102[6] via P2P/IPC
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO Channel 03/0 : 79[7] -> 78[6] via P2P/IPC
+ip-26-0-165-202:153569:153739 [7] NCCL INFO Channel 09/0 : 95[7] -> 94[6] via P2P/IPC
+ip-26-0-161-138:96570:96759 [7] NCCL INFO Channel 01/0 : 39[7] -> 38[6] via P2P/IPC
+ip-26-0-161-142:62888:63060 [7] NCCL INFO Channel 04/0 : 47[7] -> 46[6] via P2P/IPC
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO Channel 03/0 : 103[7] -> 102[6] via P2P/IPC
+ip-26-0-168-30:235655:235828 [7] NCCL INFO Channel 05/0 : 111[7] -> 110[6] via P2P/IPC
+ip-26-0-165-202:153569:153739 [7] NCCL INFO Channel 10/0 : 95[7] -> 94[6] via P2P/IPC
+ip-26-0-161-138:96570:96759 [7] NCCL INFO Channel 02/0 : 39[7] -> 38[6] via P2P/IPC
+ip-26-0-161-142:62888:63060 [7] NCCL INFO Channel 05/0 : 47[7] -> 46[6] via P2P/IPC
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO Channel 04/0 : 103[7] -> 102[6] via P2P/IPC
+ip-26-0-168-34:272157:272324 [7] NCCL INFO Channel 02/0 : 119[7] -> 118[6] via P2P/IPC
+ip-26-0-161-138:96570:96759 [7] NCCL INFO Channel 03/0 : 39[7] -> 38[6] via P2P/IPC
+ip-26-0-165-202:153569:153739 [7] NCCL INFO Channel 11/0 : 95[7] -> 94[6] via P2P/IPC
+ip-26-0-161-142:62888:63060 [7] NCCL INFO Channel 09/0 : 47[7] -> 46[6] via P2P/IPC
+ip-26-0-161-138:96570:96759 [7] NCCL INFO Channel 04/0 : 39[7] -> 38[6] via P2P/IPC
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO Channel 04/0 : 87[7] -> 86[6] via P2P/IPC
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO Channel 05/0 : 103[7] -> 102[6] via P2P/IPC
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO Channel 04/0 : 79[7] -> 78[6] via P2P/IPC
+ip-26-0-161-103:96895:97069 [7] NCCL INFO Channel 02/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-161-78:114519:114692 [7] NCCL INFO Channel 03/0 : 71[7] -> 70[6] via P2P/IPC
+ip-26-0-165-202:153569:153739 [7] NCCL INFO Channel 12/0 : 95[7] -> 94[6] via P2P/IPC
+ip-26-0-161-138:96570:96759 [7] NCCL INFO Channel 05/0 : 39[7] -> 38[6] via P2P/IPC
+ip-26-0-161-142:62888:63060 [7] NCCL INFO Channel 10/0 : 47[7] -> 46[6] via P2P/IPC
+ip-26-0-165-202:153569:153739 [7] NCCL INFO Channel 13/0 : 95[7] -> 94[6] via P2P/IPC
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO Channel 09/0 : 103[7] -> 102[6] via P2P/IPC
+ip-26-0-161-123:32711:32884 [7] NCCL INFO Channel 02/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-168-30:235655:235828 [7] NCCL INFO Channel 09/0 : 111[7] -> 110[6] via P2P/IPC
+ip-26-0-161-138:96570:96759 [7] NCCL INFO Channel 09/0 : 39[7] -> 38[6] via P2P/IPC
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO Channel 02/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-161-142:62888:63060 [7] NCCL INFO Channel 11/0 : 47[7] -> 46[6] via P2P/IPC
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO Channel 15/0 : 63[7] -> 59[3] via P2P/IPC
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO Channel 03/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO Channel 10/0 : 103[7] -> 102[6] via P2P/IPC
+ip-26-0-168-52:2722052:2722226 [7] NCCL INFO Channel 03/0 : 127[7] -> 126[6] via P2P/IPC
+ip-26-0-161-138:96570:96759 [7] NCCL INFO Channel 10/0 : 39[7] -> 38[6] via P2P/IPC
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO Channel 04/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-161-142:62888:63060 [7] NCCL INFO Channel 12/0 : 47[7] -> 46[6] via P2P/IPC
+ip-26-0-160-225:124167:124338 [7] NCCL INFO Channel 01/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-138:96570:96759 [7] NCCL INFO Channel 11/0 : 39[7] -> 38[6] via P2P/IPC
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO Channel 15/0 : 55[7] -> 51[3] via P2P/IPC
+ip-26-0-161-138:96570:96759 [7] NCCL INFO Channel 12/0 : 39[7] -> 38[6] via P2P/IPC
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO Channel 11/0 : 103[7] -> 102[6] via P2P/IPC
+ip-26-0-161-138:96570:96759 [7] NCCL INFO Channel 13/0 : 39[7] -> 38[6] via P2P/IPC
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO Channel 05/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-168-34:272157:272324 [7] NCCL INFO Channel 03/0 : 119[7] -> 118[6] via P2P/IPC
+ip-26-0-165-202:153565:153738 [3] NCCL INFO Channel 01/0 : 91[3] -> 90[2] via P2P/IPC
+ip-26-0-161-142:62888:63060 [7] NCCL INFO Channel 13/0 : 47[7] -> 46[6] via P2P/IPC
+ip-26-0-165-202:153566:153741 [4] NCCL INFO Channel 01/0 : 92[4] -> 91[3] via P2P/IPC
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO Channel 01/0 : 100[4] -> 99[3] via P2P/IPC
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO Channel 09/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO Channel 05/0 : 79[7] -> 78[6] via P2P/IPC
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO Channel 12/0 : 103[7] -> 102[6] via P2P/IPC
+ip-26-0-165-202:153565:153738 [3] NCCL INFO Channel 03/0 : 91[3] -> 90[2] via P2P/IPC
+ip-26-0-165-202:153566:153741 [4] NCCL INFO Channel 02/0 : 92[4] -> 91[3] via P2P/IPC
+ip-26-0-161-103:96895:97069 [7] NCCL INFO Channel 03/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-165-202:153567:153742 [5] NCCL INFO Channel 01/0 : 93[5] -> 92[4] via P2P/IPC
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO Channel 10/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-165-202:153565:153738 [3] NCCL INFO Channel 04/0 : 91[3] -> 90[2] via P2P/IPC
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO Channel 02/0 : 100[4] -> 99[3] via P2P/IPC
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO Channel 05/0 : 87[7] -> 86[6] via P2P/IPC
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO Channel 13/0 : 103[7] -> 102[6] via P2P/IPC
+ip-26-0-161-123:32711:32884 [7] NCCL INFO Channel 03/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-168-30:235655:235828 [7] NCCL INFO Channel 10/0 : 111[7] -> 110[6] via P2P/IPC
+ip-26-0-161-138:96567:96760 [4] NCCL INFO Channel 01/0 : 36[4] -> 35[3] via P2P/IPC
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO Channel 11/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO Channel 04/0 : 100[4] -> 99[3] via P2P/IPC
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO Channel 09/0 : 87[7] -> 86[6] via P2P/IPC
+ip-26-0-161-138:96567:96760 [4] NCCL INFO Channel 02/0 : 36[4] -> 35[3] via P2P/IPC
+ip-26-0-160-225:124167:124338 [7] NCCL INFO Channel 02/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-168-30:235655:235828 [7] NCCL INFO Channel 11/0 : 111[7] -> 110[6] via P2P/IPC
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO Channel 01/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-165-202:153566:153741 [4] NCCL INFO Channel 04/0 : 92[4] -> 91[3] via P2P/IPC
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO Channel 12/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-161-103:96895:97069 [7] NCCL INFO Channel 04/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-165-202:153567:153742 [5] NCCL INFO Channel 02/0 : 93[5] -> 92[4] via P2P/IPC
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO Channel 09/0 : 79[7] -> 78[6] via P2P/IPC
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO Channel 01/0 : 63[7] -> 62[6] via P2P/IPC
+ip-26-0-168-34:272157:272324 [7] NCCL INFO Channel 04/0 : 119[7] -> 118[6] via P2P/IPC
+ip-26-0-161-138:96567:96760 [4] NCCL INFO Channel 04/0 : 36[4] -> 35[3] via P2P/IPC
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO Channel 01/0 : 55[7] -> 54[6] via P2P/IPC
+ip-26-0-168-30:235652:235822 [4] NCCL INFO Channel 01/0 : 108[4] -> 107[3] via P2P/IPC
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO Channel 09/0 : 100[4] -> 99[3] via P2P/IPC
+ip-26-0-165-202:153565:153738 [3] NCCL INFO Channel 05/0 : 91[3] -> 90[2] via P2P/IPC
+ip-26-0-161-142:62885:63054 [4] NCCL INFO Channel 01/0 : 44[4] -> 43[3] via P2P/IPC
+ip-26-0-168-30:235655:235828 [7] NCCL INFO Channel 12/0 : 111[7] -> 110[6] via P2P/IPC
+ip-26-0-165-202:153566:153741 [4] NCCL INFO Channel 09/0 : 92[4] -> 91[3] via P2P/IPC
+ip-26-0-165-202:153567:153742 [5] NCCL INFO Channel 03/0 : 93[5] -> 92[4] via P2P/IPC
+ip-26-0-161-103:96895:97069 [7] NCCL INFO Channel 05/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO Channel 02/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-161-123:32711:32884 [7] NCCL INFO Channel 04/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO Channel 13/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-161-78:114519:114692 [7] NCCL INFO Channel 04/0 : 71[7] -> 70[6] via P2P/IPC
+ip-26-0-168-34:272157:272324 [7] NCCL INFO Channel 05/0 : 119[7] -> 118[6] via P2P/IPC
+ip-26-0-161-142:62885:63054 [4] NCCL INFO Channel 02/0 : 44[4] -> 43[3] via P2P/IPC
+ip-26-0-161-138:96567:96760 [4] NCCL INFO Channel 09/0 : 36[4] -> 35[3] via P2P/IPC
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO Channel 10/0 : 100[4] -> 99[3] via P2P/IPC
+ip-26-0-165-202:153565:153738 [3] NCCL INFO Channel 06/0 : 91[3] -> 90[2] via P2P/IPC
+ip-26-0-161-103:96895:97069 [7] NCCL INFO Channel 09/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO Channel 04/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO Channel 01/0 : 99[3] -> 98[2] via P2P/IPC
+ip-26-0-168-30:235652:235822 [4] NCCL INFO Channel 02/0 : 108[4] -> 107[3] via P2P/IPC
+ip-26-0-165-202:153566:153741 [4] NCCL INFO Channel 10/0 : 92[4] -> 91[3] via P2P/IPC
+ip-26-0-168-30:235655:235828 [7] NCCL INFO Channel 13/0 : 111[7] -> 110[6] via P2P/IPC
+ip-26-0-165-202:153567:153742 [5] NCCL INFO Channel 06/0 : 93[5] -> 92[4] via P2P/IPC
+ip-26-0-165-202:153568:153736 [6] NCCL INFO Channel 01/0 : 94[6] -> 93[5] via P2P/IPC
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO Channel 01/0 : 84[4] -> 83[3] via P2P/IPC
+ip-26-0-161-123:32711:32884 [7] NCCL INFO Channel 05/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO Channel 09/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO Channel 10/0 : 87[7] -> 86[6] via P2P/IPC
+ip-26-0-165-202:153565:153738 [3] NCCL INFO Channel 07/0 : 91[3] -> 90[2] via P2P/IPC
+ip-26-0-168-34:272157:272324 [7] NCCL INFO Channel 09/0 : 119[7] -> 118[6] via P2P/IPC
+ip-26-0-161-138:96567:96760 [4] NCCL INFO Channel 10/0 : 36[4] -> 35[3] via P2P/IPC
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO Channel 12/0 : 100[4] -> 99[3] via P2P/IPC
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO Channel 10/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO Channel 03/0 : 99[3] -> 98[2] via P2P/IPC
+ip-26-0-165-202:153566:153741 [4] NCCL INFO Channel 12/0 : 92[4] -> 91[3] via P2P/IPC
+ip-26-0-161-138:96566:96755 [3] NCCL INFO Channel 01/0 : 35[3] -> 34[2] via P2P/IPC
+ip-26-0-161-142:62884:63061 [3] NCCL INFO Channel 01/0 : 43[3] -> 42[2] via P2P/IPC
+ip-26-0-161-123:32711:32884 [7] NCCL INFO Channel 09/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-165-202:153567:153742 [5] NCCL INFO Channel 07/0 : 93[5] -> 92[4] via P2P/IPC
+ip-26-0-165-202:153568:153736 [6] NCCL INFO Channel 02/0 : 94[6] -> 93[5] via P2P/IPC
+ip-26-0-168-52:2722052:2722226 [7] NCCL INFO Channel 04/0 : 127[7] -> 126[6] via P2P/IPC
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO Channel 12/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-165-202:153565:153738 [3] NCCL INFO Channel 09/0 : 91[3] -> 90[2] via P2P/IPC
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO Channel 10/0 : 79[7] -> 78[6] via P2P/IPC
+ip-26-0-168-30:235652:235822 [4] NCCL INFO Channel 04/0 : 108[4] -> 107[3] via P2P/IPC
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO Channel 01/0 : 102[6] -> 101[5] via P2P/IPC
+ip-26-0-161-138:96567:96760 [4] NCCL INFO Channel 12/0 : 36[4] -> 35[3] via P2P/IPC
+ip-26-0-161-142:62885:63054 [4] NCCL INFO Channel 04/0 : 44[4] -> 43[3] via P2P/IPC
+ip-26-0-168-52:2722052:2722226 [7] NCCL INFO Channel 05/0 : 127[7] -> 126[6] via P2P/IPC
+ip-26-0-168-34:272157:272324 [7] NCCL INFO Channel 10/0 : 119[7] -> 118[6] via P2P/IPC
+ip-26-0-168-30:235652:235822 [4] NCCL INFO Channel 09/0 : 108[4] -> 107[3] via P2P/IPC
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO Channel 04/0 : 99[3] -> 98[2] via P2P/IPC
+ip-26-0-161-138:96566:96755 [3] NCCL INFO Channel 03/0 : 35[3] -> 34[2] via P2P/IPC
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO Channel 01/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-161-123:32711:32884 [7] NCCL INFO Channel 10/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-165-202:153567:153742 [5] NCCL INFO Channel 09/0 : 93[5] -> 92[4] via P2P/IPC
+ip-26-0-161-142:62884:63061 [3] NCCL INFO Channel 03/0 : 43[3] -> 42[2] via P2P/IPC
+ip-26-0-165-202:153568:153736 [6] NCCL INFO Channel 03/0 : 94[6] -> 93[5] via P2P/IPC
+ip-26-0-168-30:235652:235822 [4] NCCL INFO Channel 10/0 : 108[4] -> 107[3] via P2P/IPC
+ip-26-0-165-202:153565:153738 [3] NCCL INFO Channel 11/0 : 91[3] -> 90[2] via P2P/IPC
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO Channel 02/0 : 102[6] -> 101[5] via P2P/IPC
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO Channel 11/0 : 79[7] -> 78[6] via P2P/IPC
+ip-26-0-161-142:62887:63055 [6] NCCL INFO Channel 01/0 : 46[6] -> 45[5] via P2P/IPC
+ip-26-0-161-78:114519:114692 [7] NCCL INFO Channel 05/0 : 71[7] -> 70[6] via P2P/IPC
+ip-26-0-161-142:62885:63054 [4] NCCL INFO Channel 09/0 : 44[4] -> 43[3] via P2P/IPC
+ip-26-0-168-30:235652:235822 [4] NCCL INFO Channel 12/0 : 108[4] -> 107[3] via P2P/IPC
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO Channel 05/0 : 99[3] -> 98[2] via P2P/IPC
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO Channel 02/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-160-225:124167:124338 [7] NCCL INFO Channel 03/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-168-34:272157:272324 [7] NCCL INFO Channel 11/0 : 119[7] -> 118[6] via P2P/IPC
+ip-26-0-161-138:96566:96755 [3] NCCL INFO Channel 04/0 : 35[3] -> 34[2] via P2P/IPC
+ip-26-0-165-202:153567:153742 [5] NCCL INFO Channel 10/0 : 93[5] -> 92[4] via P2P/IPC
+ip-26-0-168-52:2722052:2722226 [7] NCCL INFO Channel 09/0 : 127[7] -> 126[6] via P2P/IPC
+ip-26-0-161-123:32711:32884 [7] NCCL INFO Channel 11/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-165-202:153568:153736 [6] NCCL INFO Channel 04/0 : 94[6] -> 93[5] via P2P/IPC
+ip-26-0-161-103:96895:97069 [7] NCCL INFO Channel 10/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO Channel 03/0 : 102[6] -> 101[5] via P2P/IPC
+ip-26-0-168-30:235651:235823 [3] NCCL INFO Channel 01/0 : 107[3] -> 106[2] via P2P/IPC
+ip-26-0-165-202:153565:153738 [3] NCCL INFO Channel 12/0 : 91[3] -> 90[2] via P2P/IPC
+ip-26-0-161-138:96569:96758 [6] NCCL INFO Channel 01/0 : 38[6] -> 37[5] via P2P/IPC
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO Channel 01/0 : 101[5] -> 100[4] via P2P/IPC
+ip-26-0-161-142:62884:63061 [3] NCCL INFO Channel 04/0 : 43[3] -> 42[2] via P2P/IPC
+ip-26-0-161-142:62887:63055 [6] NCCL INFO Channel 02/0 : 46[6] -> 45[5] via P2P/IPC
+ip-26-0-165-202:153567:153742 [5] NCCL INFO Channel 11/0 : 93[5] -> 92[4] via P2P/IPC
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO Channel 06/0 : 99[3] -> 98[2] via P2P/IPC
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO Channel 03/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-161-138:96566:96755 [3] NCCL INFO Channel 05/0 : 35[3] -> 34[2] via P2P/IPC
+ip-26-0-161-142:62885:63054 [4] NCCL INFO Channel 10/0 : 44[4] -> 43[3] via P2P/IPC
+ip-26-0-168-30:235651:235823 [3] NCCL INFO Channel 03/0 : 107[3] -> 106[2] via P2P/IPC
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO Channel 12/0 : 79[7] -> 78[6] via P2P/IPC
+ip-26-0-168-34:272157:272324 [7] NCCL INFO Channel 12/0 : 119[7] -> 118[6] via P2P/IPC
+ip-26-0-161-138:96568:96753 [5] NCCL INFO Channel 01/0 : 37[5] -> 36[4] via P2P/IPC
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO Channel 01/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-165-202:153568:153736 [6] NCCL INFO Channel 07/0 : 94[6] -> 93[5] via P2P/IPC
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO Channel 01/0 : 76[4] -> 75[3] via P2P/IPC
+ip-26-0-161-138:96569:96758 [6] NCCL INFO Channel 02/0 : 38[6] -> 37[5] via P2P/IPC
+ip-26-0-165-202:153565:153738 [3] NCCL INFO Channel 13/0 : 91[3] -> 90[2] via P2P/IPC
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO Channel 04/0 : 102[6] -> 101[5] via P2P/IPC
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO Channel 02/0 : 101[5] -> 100[4] via P2P/IPC
+ip-26-0-168-52:2722052:2722226 [7] NCCL INFO Channel 10/0 : 127[7] -> 126[6] via P2P/IPC
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO Channel 13/0 : 79[7] -> 78[6] via P2P/IPC
+ip-26-0-161-138:96566:96755 [3] NCCL INFO Channel 06/0 : 35[3] -> 34[2] via P2P/IPC
+ip-26-0-168-30:235651:235823 [3] NCCL INFO Channel 04/0 : 107[3] -> 106[2] via P2P/IPC
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO Channel 07/0 : 99[3] -> 98[2] via P2P/IPC
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO Channel 04/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-161-123:32711:32884 [7] NCCL INFO Channel 12/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-161-103:96895:97069 [7] NCCL INFO Channel 11/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-161-138:96568:96753 [5] NCCL INFO Channel 02/0 : 37[5] -> 36[4] via P2P/IPC
+ip-26-0-161-142:62884:63061 [3] NCCL INFO Channel 05/0 : 43[3] -> 42[2] via P2P/IPC
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO Channel 03/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-161-142:62887:63055 [6] NCCL INFO Channel 03/0 : 46[6] -> 45[5] via P2P/IPC
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO Channel 02/0 : 76[4] -> 75[3] via P2P/IPC
+ip-26-0-168-34:272154:272330 [4] NCCL INFO Channel 01/0 : 116[4] -> 115[3] via P2P/IPC
+ip-26-0-161-142:62885:63054 [4] NCCL INFO Channel 12/0 : 44[4] -> 43[3] via P2P/IPC
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO Channel 07/0 : 102[6] -> 101[5] via P2P/IPC
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO Channel 03/0 : 101[5] -> 100[4] via P2P/IPC
+ip-26-0-168-52:2722052:2722226 [7] NCCL INFO Channel 11/0 : 127[7] -> 126[6] via P2P/IPC
+ip-26-0-168-30:235651:235823 [3] NCCL INFO Channel 05/0 : 107[3] -> 106[2] via P2P/IPC
+ip-26-0-161-138:96569:96758 [6] NCCL INFO Channel 03/0 : 38[6] -> 37[5] via P2P/IPC
+ip-26-0-168-34:272157:272324 [7] NCCL INFO Channel 13/0 : 119[7] -> 118[6] via P2P/IPC
+ip-26-0-161-138:96566:96755 [3] NCCL INFO Channel 07/0 : 35[3] -> 34[2] via P2P/IPC
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO Channel 09/0 : 99[3] -> 98[2] via P2P/IPC
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO Channel 07/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO Channel 04/0 : 76[4] -> 75[3] via P2P/IPC
+ip-26-0-161-138:96568:96753 [5] NCCL INFO Channel 03/0 : 37[5] -> 36[4] via P2P/IPC
+ip-26-0-161-123:32708:32879 [4] NCCL INFO Channel 01/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO Channel 04/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-168-34:272154:272330 [4] NCCL INFO Channel 02/0 : 116[4] -> 115[3] via P2P/IPC
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO Channel 09/0 : 102[6] -> 101[5] via P2P/IPC
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO Channel 06/0 : 101[5] -> 100[4] via P2P/IPC
+ip-26-0-161-142:62884:63061 [3] NCCL INFO Channel 06/0 : 43[3] -> 42[2] via P2P/IPC
+ip-26-0-161-142:62887:63055 [6] NCCL INFO Channel 04/0 : 46[6] -> 45[5] via P2P/IPC
+ip-26-0-161-138:96569:96758 [6] NCCL INFO Channel 04/0 : 38[6] -> 37[5] via P2P/IPC
+ip-26-0-168-30:235651:235823 [3] NCCL INFO Channel 06/0 : 107[3] -> 106[2] via P2P/IPC
+ip-26-0-161-123:32711:32884 [7] NCCL INFO Channel 13/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-160-225:124167:124338 [7] NCCL INFO Channel 04/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO Channel 01/0 : 124[4] -> 123[3] via P2P/IPC
+ip-26-0-161-103:96895:97069 [7] NCCL INFO Channel 12/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-161-138:96566:96755 [3] NCCL INFO Channel 09/0 : 35[3] -> 34[2] via P2P/IPC
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO Channel 11/0 : 99[3] -> 98[2] via P2P/IPC
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO Channel 09/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-168-52:2722052:2722226 [7] NCCL INFO Channel 12/0 : 127[7] -> 126[6] via P2P/IPC
+ip-26-0-161-138:96568:96753 [5] NCCL INFO Channel 06/0 : 37[5] -> 36[4] via P2P/IPC
+ip-26-0-161-142:62886:63057 [5] NCCL INFO Channel 01/0 : 45[5] -> 44[4] via P2P/IPC
+ip-26-0-168-30:235654:235821 [6] NCCL INFO Channel 01/0 : 110[6] -> 109[5] via P2P/IPC
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO Channel 09/0 : 76[4] -> 75[3] via P2P/IPC
+ip-26-0-168-34:272154:272330 [4] NCCL INFO Channel 04/0 : 116[4] -> 115[3] via P2P/IPC
+ip-26-0-161-103:96892:97064 [4] NCCL INFO Channel 01/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO Channel 05/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO Channel 10/0 : 102[6] -> 101[5] via P2P/IPC
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO Channel 07/0 : 101[5] -> 100[4] via P2P/IPC
+ip-26-0-161-123:32708:32879 [4] NCCL INFO Channel 02/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-168-30:235651:235823 [3] NCCL INFO Channel 07/0 : 107[3] -> 106[2] via P2P/IPC
+ip-26-0-161-142:62887:63055 [6] NCCL INFO Channel 07/0 : 46[6] -> 45[5] via P2P/IPC
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO Channel 12/0 : 99[3] -> 98[2] via P2P/IPC
+ip-26-0-161-142:62884:63061 [3] NCCL INFO Channel 07/0 : 43[3] -> 42[2] via P2P/IPC
+ip-26-0-161-138:96569:96758 [6] NCCL INFO Channel 07/0 : 38[6] -> 37[5] via P2P/IPC
+ip-26-0-161-142:62886:63057 [5] NCCL INFO Channel 02/0 : 45[5] -> 44[4] via P2P/IPC
+ip-26-0-168-30:235654:235821 [6] NCCL INFO Channel 02/0 : 110[6] -> 109[5] via P2P/IPC
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO Channel 01/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-160-225:124167:124338 [7] NCCL INFO Channel 05/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-165-202:153568:153736 [6] NCCL INFO Channel 09/0 : 94[6] -> 93[5] via P2P/IPC
+ip-26-0-168-34:272153:272328 [3] NCCL INFO Channel 01/0 : 115[3] -> 114[2] via P2P/IPC
+ip-26-0-168-34:272154:272330 [4] NCCL INFO Channel 09/0 : 116[4] -> 115[3] via P2P/IPC
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO Channel 06/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO Channel 02/0 : 63[7] -> 62[6] via P2P/IPC
+ip-26-0-165-202:153565:153738 [3] NCCL INFO Channel 14/0 : 91[3] -> 90[2] via P2P/IPC
+ip-26-0-161-142:62887:63055 [6] NCCL INFO Channel 09/0 : 46[6] -> 45[5] via P2P/IPC
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO Channel 09/0 : 101[5] -> 100[4] via P2P/IPC
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO Channel 10/0 : 76[4] -> 75[3] via P2P/IPC
+ip-26-0-161-103:96895:97069 [7] NCCL INFO Channel 13/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO Channel 02/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-165-202:153568:153736 [6] NCCL INFO Channel 10/0 : 94[6] -> 93[5] via P2P/IPC
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO Channel 11/0 : 102[6] -> 101[5] via P2P/IPC
+ip-26-0-161-138:96569:96758 [6] NCCL INFO Channel 09/0 : 38[6] -> 37[5] via P2P/IPC
+ip-26-0-168-30:235651:235823 [3] NCCL INFO Channel 09/0 : 107[3] -> 106[2] via P2P/IPC
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO Channel 13/0 : 99[3] -> 98[2] via P2P/IPC
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO Channel 07/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-161-142:62884:63061 [3] NCCL INFO Channel 09/0 : 43[3] -> 42[2] via P2P/IPC
+ip-26-0-168-30:235654:235821 [6] NCCL INFO Channel 03/0 : 110[6] -> 109[5] via P2P/IPC
+ip-26-0-161-123:32707:32878 [3] NCCL INFO Channel 01/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-165-202:153567:153742 [5] NCCL INFO Channel 14/0 : 93[5] -> 92[4] via P2P/IPC
+ip-26-0-161-142:62886:63057 [5] NCCL INFO Channel 03/0 : 45[5] -> 44[4] via P2P/IPC
+ip-26-0-168-34:272154:272330 [4] NCCL INFO Channel 10/0 : 116[4] -> 115[3] via P2P/IPC
+ip-26-0-161-103:96892:97064 [4] NCCL INFO Channel 02/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-161-123:32708:32879 [4] NCCL INFO Channel 04/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-161-78:114519:114692 [7] NCCL INFO Channel 09/0 : 71[7] -> 70[6] via P2P/IPC
+ip-26-0-161-138:96569:96758 [6] NCCL INFO Channel 10/0 : 38[6] -> 37[5] via P2P/IPC
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO Channel 03/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-168-34:272153:272328 [3] NCCL INFO Channel 03/0 : 115[3] -> 114[2] via P2P/IPC
+ip-26-0-160-225:124167:124338 [7] NCCL INFO Channel 09/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO Channel 10/0 : 101[5] -> 100[4] via P2P/IPC
+ip-26-0-161-142:62887:63055 [6] NCCL INFO Channel 10/0 : 46[6] -> 45[5] via P2P/IPC
+ip-26-0-165-202:153568:153736 [6] NCCL INFO Channel 11/0 : 94[6] -> 93[5] via P2P/IPC
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO Channel 01/0 : 60[4] -> 59[3] via P2P/IPC
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO Channel 09/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-168-30:235653:235825 [5] NCCL INFO Channel 01/0 : 109[5] -> 108[4] via P2P/IPC
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO Channel 02/0 : 55[7] -> 54[6] via P2P/IPC
+ip-26-0-165-202:153565:153738 [3] NCCL INFO Channel 15/0 : 91[3] -> 90[2] via P2P/IPC
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO Channel 03/0 : 63[7] -> 62[6] via P2P/IPC
+ip-26-0-161-138:96569:96758 [6] NCCL INFO Channel 11/0 : 38[6] -> 37[5] via P2P/IPC
+ip-26-0-168-30:235651:235823 [3] NCCL INFO Channel 11/0 : 107[3] -> 106[2] via P2P/IPC
+ip-26-0-161-142:62886:63057 [5] NCCL INFO Channel 06/0 : 45[5] -> 44[4] via P2P/IPC
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO Channel 06/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-168-30:235654:235821 [6] NCCL INFO Channel 04/0 : 110[6] -> 109[5] via P2P/IPC
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO Channel 12/0 : 76[4] -> 75[3] via P2P/IPC
+ip-26-0-161-142:62884:63061 [3] NCCL INFO Channel 11/0 : 43[3] -> 42[2] via P2P/IPC
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO Channel 11/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-161-103:96892:97064 [4] NCCL INFO Channel 04/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-161-138:96569:96758 [6] NCCL INFO Channel 12/0 : 38[6] -> 37[5] via P2P/IPC
+ip-26-0-168-34:272154:272330 [4] NCCL INFO Channel 12/0 : 116[4] -> 115[3] via P2P/IPC
+ip-26-0-165-202:153567:153742 [5] NCCL INFO Channel 15/0 : 93[5] -> 92[4] via P2P/IPC
+ip-26-0-161-142:62887:63055 [6] NCCL INFO Channel 11/0 : 46[6] -> 45[5] via P2P/IPC
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO Channel 10/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-161-123:32708:32879 [4] NCCL INFO Channel 09/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-161-138:96568:96753 [5] NCCL INFO Channel 07/0 : 37[5] -> 36[4] via P2P/IPC
+ip-26-0-168-52:2722052:2722226 [7] NCCL INFO Channel 13/0 : 127[7] -> 126[6] via P2P/IPC
+ip-26-0-165-202:153568:153736 [6] NCCL INFO Channel 12/0 : 94[6] -> 93[5] via P2P/IPC
+ip-26-0-168-34:272153:272328 [3] NCCL INFO Channel 04/0 : 115[3] -> 114[2] via P2P/IPC
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO Channel 02/0 : 60[4] -> 59[3] via P2P/IPC
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO Channel 07/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO Channel 02/0 : 124[4] -> 123[3] via P2P/IPC
+ip-26-0-160-225:124167:124338 [7] NCCL INFO Channel 10/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-168-30:235653:235825 [5] NCCL INFO Channel 02/0 : 109[5] -> 108[4] via P2P/IPC
+ip-26-0-161-142:62886:63057 [5] NCCL INFO Channel 07/0 : 45[5] -> 44[4] via P2P/IPC
+ip-26-0-161-138:96566:96755 [3] NCCL INFO Channel 11/0 : 35[3] -> 34[2] via P2P/IPC
+ip-26-0-161-78:114519:114692 [7] NCCL INFO Channel 10/0 : 71[7] -> 70[6] via P2P/IPC
+ip-26-0-168-30:235651:235823 [3] NCCL INFO Channel 12/0 : 107[3] -> 106[2] via P2P/IPC
+ip-26-0-161-138:96568:96753 [5] NCCL INFO Channel 09/0 : 37[5] -> 36[4] via P2P/IPC
+ip-26-0-161-123:32708:32879 [4] NCCL INFO Channel 10/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-161-142:62884:63061 [3] NCCL INFO Channel 12/0 : 43[3] -> 42[2] via P2P/IPC
+ip-26-0-161-103:96891:97063 [3] NCCL INFO Channel 01/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-168-30:235654:235821 [6] NCCL INFO Channel 07/0 : 110[6] -> 109[5] via P2P/IPC
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO Channel 12/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO Channel 04/0 : 63[7] -> 62[6] via P2P/IPC
+ip-26-0-161-138:96569:96758 [6] NCCL INFO Channel 15/0 : 38[6] -> 37[5] via P2P/IPC
+ip-26-0-165-202:153568:153736 [6] NCCL INFO Channel 15/0 : 94[6] -> 93[5] via P2P/IPC
+ip-26-0-161-103:96892:97064 [4] NCCL INFO Channel 09/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO Channel 11/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-161-142:62887:63055 [6] NCCL INFO Channel 12/0 : 46[6] -> 45[5] via P2P/IPC
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO Channel 04/0 : 60[4] -> 59[3] via P2P/IPC
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO Channel 09/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-161-138:96566:96755 [3] NCCL INFO Channel 12/0 : 35[3] -> 34[2] via P2P/IPC
+ip-26-0-161-142:62886:63057 [5] NCCL INFO Channel 09/0 : 45[5] -> 44[4] via P2P/IPC
+ip-26-0-161-123:32708:32879 [4] NCCL INFO Channel 12/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-168-30:235653:235825 [5] NCCL INFO Channel 03/0 : 109[5] -> 108[4] via P2P/IPC
+ip-26-0-168-34:272153:272328 [3] NCCL INFO Channel 05/0 : 115[3] -> 114[2] via P2P/IPC
+ip-26-0-161-138:96568:96753 [5] NCCL INFO Channel 10/0 : 37[5] -> 36[4] via P2P/IPC
+ip-26-0-168-30:235651:235823 [3] NCCL INFO Channel 13/0 : 107[3] -> 106[2] via P2P/IPC
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO Channel 04/0 : 124[4] -> 123[3] via P2P/IPC
+ip-26-0-161-142:62884:63061 [3] NCCL INFO Channel 13/0 : 43[3] -> 42[2] via P2P/IPC
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO Channel 13/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-168-30:235654:235821 [6] NCCL INFO Channel 09/0 : 110[6] -> 109[5] via P2P/IPC
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO Channel 12/0 : 102[6] -> 101[5] via P2P/IPC
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO Channel 12/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-161-142:62887:63055 [6] NCCL INFO Channel 15/0 : 46[6] -> 45[5] via P2P/IPC
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO Channel 05/0 : 63[7] -> 62[6] via P2P/IPC
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO Channel 14/0 : 99[3] -> 98[2] via P2P/IPC
+ip-26-0-161-103:96891:97063 [3] NCCL INFO Channel 03/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO Channel 10/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-160-225:124167:124338 [7] NCCL INFO Channel 11/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-138:96566:96755 [3] NCCL INFO Channel 13/0 : 35[3] -> 34[2] via P2P/IPC
+ip-26-0-161-78:114519:114692 [7] NCCL INFO Channel 11/0 : 71[7] -> 70[6] via P2P/IPC
+ip-26-0-168-30:235653:235825 [5] NCCL INFO Channel 06/0 : 109[5] -> 108[4] via P2P/IPC
+ip-26-0-161-123:32710:32883 [6] NCCL INFO Channel 01/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-161-103:96892:97064 [4] NCCL INFO Channel 10/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-161-142:62886:63057 [5] NCCL INFO Channel 10/0 : 45[5] -> 44[4] via P2P/IPC
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO Channel 11/0 : 101[5] -> 100[4] via P2P/IPC
+ip-26-0-168-34:272153:272328 [3] NCCL INFO Channel 06/0 : 115[3] -> 114[2] via P2P/IPC
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO Channel 14/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-161-142:62884:63061 [3] NCCL INFO Channel 14/0 : 43[3] -> 42[2] via P2P/IPC
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO Channel 15/0 : 102[6] -> 101[5] via P2P/IPC
+ip-26-0-168-30:235651:235823 [3] NCCL INFO Channel 14/0 : 107[3] -> 106[2] via P2P/IPC
+ip-26-0-168-34:272156:272325 [6] NCCL INFO Channel 01/0 : 118[6] -> 117[5] via P2P/IPC
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO Channel 01/0 : 75[3] -> 74[2] via P2P/IPC
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO Channel 09/0 : 124[4] -> 123[3] via P2P/IPC
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO Channel 15/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-168-30:235654:235821 [6] NCCL INFO Channel 10/0 : 110[6] -> 109[5] via P2P/IPC
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO Channel 09/0 : 60[4] -> 59[3] via P2P/IPC
+ip-26-0-161-123:32709:32882 [5] NCCL INFO Channel 01/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO Channel 11/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-161-142:62886:63057 [5] NCCL INFO Channel 11/0 : 45[5] -> 44[4] via P2P/IPC
+ip-26-0-168-30:235653:235825 [5] NCCL INFO Channel 07/0 : 109[5] -> 108[4] via P2P/IPC
+ip-26-0-161-123:32710:32883 [6] NCCL INFO Channel 02/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-168-34:272153:272328 [3] NCCL INFO Channel 07/0 : 115[3] -> 114[2] via P2P/IPC
+ip-26-0-161-103:96891:97063 [3] NCCL INFO Channel 04/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO Channel 01/0 : 78[6] -> 77[5] via P2P/IPC
+ip-26-0-161-103:96892:97064 [4] NCCL INFO Channel 12/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO Channel 09/0 : 63[7] -> 62[6] via P2P/IPC
+ip-26-0-161-123:32709:32882 [5] NCCL INFO Channel 02/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-168-34:272156:272325 [6] NCCL INFO Channel 02/0 : 118[6] -> 117[5] via P2P/IPC
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO Channel 15/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-168-30:235654:235821 [6] NCCL INFO Channel 11/0 : 110[6] -> 109[5] via P2P/IPC
+ip-26-0-161-123:32710:32883 [6] NCCL INFO Channel 03/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-161-78:114519:114692 [7] NCCL INFO Channel 12/0 : 71[7] -> 70[6] via P2P/IPC
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO Channel 03/0 : 75[3] -> 74[2] via P2P/IPC
+ip-26-0-168-30:235651:235823 [3] NCCL INFO Channel 15/0 : 107[3] -> 106[2] via P2P/IPC
+ip-26-0-160-225:124167:124338 [7] NCCL INFO Channel 12/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO Channel 14/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-168-30:235653:235825 [5] NCCL INFO Channel 09/0 : 109[5] -> 108[4] via P2P/IPC
+ip-26-0-168-34:272153:272328 [3] NCCL INFO Channel 09/0 : 115[3] -> 114[2] via P2P/IPC
+ip-26-0-161-123:32709:32882 [5] NCCL INFO Channel 03/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO Channel 02/0 : 78[6] -> 77[5] via P2P/IPC
+ip-26-0-161-123:32710:32883 [6] NCCL INFO Channel 04/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-168-34:272156:272325 [6] NCCL INFO Channel 03/0 : 118[6] -> 117[5] via P2P/IPC
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO Channel 10/0 : 60[4] -> 59[3] via P2P/IPC
+ip-26-0-168-30:235654:235821 [6] NCCL INFO Channel 12/0 : 110[6] -> 109[5] via P2P/IPC
+ip-26-0-161-103:96894:97068 [6] NCCL INFO Channel 01/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-161-103:96891:97063 [3] NCCL INFO Channel 05/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-161-138:96568:96753 [5] NCCL INFO Channel 11/0 : 37[5] -> 36[4] via P2P/IPC
+ip-26-0-161-123:32709:32882 [5] NCCL INFO Channel 06/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO Channel 04/0 : 75[3] -> 74[2] via P2P/IPC
+ip-26-0-168-30:235653:235825 [5] NCCL INFO Channel 10/0 : 109[5] -> 108[4] via P2P/IPC
+ip-26-0-161-123:32707:32878 [3] NCCL INFO Channel 03/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO Channel 10/0 : 63[7] -> 62[6] via P2P/IPC
+ip-26-0-161-123:32710:32883 [6] NCCL INFO Channel 07/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO Channel 03/0 : 78[6] -> 77[5] via P2P/IPC
+ip-26-0-168-34:272156:272325 [6] NCCL INFO Channel 04/0 : 118[6] -> 117[5] via P2P/IPC
+ip-26-0-168-30:235654:235821 [6] NCCL INFO Channel 15/0 : 110[6] -> 109[5] via P2P/IPC
+ip-26-0-161-78:114516:114691 [4] NCCL INFO Channel 01/0 : 68[4] -> 67[3] via P2P/IPC
+ip-26-0-160-225:124167:124338 [7] NCCL INFO Channel 13/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-78:114519:114692 [7] NCCL INFO Channel 13/0 : 71[7] -> 70[6] via P2P/IPC
+ip-26-0-168-34:272155:272329 [5] NCCL INFO Channel 01/0 : 117[5] -> 116[4] via P2P/IPC
+ip-26-0-160-225:124164:124336 [4] NCCL INFO Channel 01/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO Channel 05/0 : 75[3] -> 74[2] via P2P/IPC
+ip-26-0-161-123:32709:32882 [5] NCCL INFO Channel 07/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-161-103:96891:97063 [3] NCCL INFO Channel 06/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-161-103:96894:97068 [6] NCCL INFO Channel 02/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-168-34:272156:272325 [6] NCCL INFO Channel 07/0 : 118[6] -> 117[5] via P2P/IPC
+ip-26-0-161-123:32707:32878 [3] NCCL INFO Channel 04/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO Channel 12/0 : 60[4] -> 59[3] via P2P/IPC
+ip-26-0-161-123:32710:32883 [6] NCCL INFO Channel 09/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-168-34:272155:272329 [5] NCCL INFO Channel 02/0 : 117[5] -> 116[4] via P2P/IPC
+ip-26-0-161-103:96893:97067 [5] NCCL INFO Channel 01/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO Channel 04/0 : 78[6] -> 77[5] via P2P/IPC
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO Channel 10/0 : 124[4] -> 123[3] via P2P/IPC
+ip-26-0-168-34:272156:272325 [6] NCCL INFO Channel 09/0 : 118[6] -> 117[5] via P2P/IPC
+ip-26-0-161-123:32709:32882 [5] NCCL INFO Channel 09/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO Channel 11/0 : 63[7] -> 62[6] via P2P/IPC
+ip-26-0-161-142:62884:63061 [3] NCCL INFO Channel 15/0 : 43[3] -> 42[2] via P2P/IPC
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO Channel 06/0 : 75[3] -> 74[2] via P2P/IPC
+ip-26-0-168-34:272155:272329 [5] NCCL INFO Channel 03/0 : 117[5] -> 116[4] via P2P/IPC
+ip-26-0-161-103:96891:97063 [3] NCCL INFO Channel 07/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-161-103:96894:97068 [6] NCCL INFO Channel 03/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO Channel 14/0 : 101[5] -> 100[4] via P2P/IPC
+ip-26-0-161-78:114516:114691 [4] NCCL INFO Channel 02/0 : 68[4] -> 67[3] via P2P/IPC
+ip-26-0-161-123:32707:32878 [3] NCCL INFO Channel 05/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO Channel 15/0 : 99[3] -> 98[2] via P2P/IPC
+ip-26-0-168-34:272156:272325 [6] NCCL INFO Channel 10/0 : 118[6] -> 117[5] via P2P/IPC
+ip-26-0-161-123:32710:32883 [6] NCCL INFO Channel 10/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-161-103:96893:97067 [5] NCCL INFO Channel 02/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-168-34:272155:272329 [5] NCCL INFO Channel 06/0 : 117[5] -> 116[4] via P2P/IPC
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO Channel 01/0 : 77[5] -> 76[4] via P2P/IPC
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO Channel 07/0 : 78[6] -> 77[5] via P2P/IPC
+ip-26-0-160-225:124164:124336 [4] NCCL INFO Channel 02/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-123:32709:32882 [5] NCCL INFO Channel 10/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO Channel 07/0 : 75[3] -> 74[2] via P2P/IPC
+ip-26-0-168-34:272156:272325 [6] NCCL INFO Channel 11/0 : 118[6] -> 117[5] via P2P/IPC
+ip-26-0-168-52:2722048:2722220 [3] NCCL INFO Channel 01/0 : 123[3] -> 122[2] via P2P/IPC
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO Channel 12/0 : 124[4] -> 123[3] via P2P/IPC
+ip-26-0-161-78:114516:114691 [4] NCCL INFO Channel 04/0 : 68[4] -> 67[3] via P2P/IPC
+ip-26-0-168-34:272153:272328 [3] NCCL INFO Channel 11/0 : 115[3] -> 114[2] via P2P/IPC
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO Channel 12/0 : 63[7] -> 62[6] via P2P/IPC
+ip-26-0-168-34:272155:272329 [5] NCCL INFO Channel 07/0 : 117[5] -> 116[4] via P2P/IPC
+ip-26-0-161-123:32707:32878 [3] NCCL INFO Channel 06/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-161-123:32710:32883 [6] NCCL INFO Channel 11/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO Channel 09/0 : 78[6] -> 77[5] via P2P/IPC
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO Channel 02/0 : 77[5] -> 76[4] via P2P/IPC
+ip-26-0-161-138:96566:96755 [3] NCCL INFO Channel 14/0 : 35[3] -> 34[2] via P2P/IPC
+ip-26-0-168-34:272156:272325 [6] NCCL INFO Channel 12/0 : 118[6] -> 117[5] via P2P/IPC
+ip-26-0-161-123:32709:32882 [5] NCCL INFO Channel 11/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-161-78:114515:114690 [3] NCCL INFO Channel 01/0 : 67[3] -> 66[2] via P2P/IPC
+ip-26-0-161-138:96568:96753 [5] NCCL INFO Channel 14/0 : 37[5] -> 36[4] via P2P/IPC
+ip-26-0-160-225:124164:124336 [4] NCCL INFO Channel 04/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-78:114516:114691 [4] NCCL INFO Channel 09/0 : 68[4] -> 67[3] via P2P/IPC
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO Channel 09/0 : 75[3] -> 74[2] via P2P/IPC
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO Channel 03/0 : 55[7] -> 54[6] via P2P/IPC
+ip-26-0-168-34:272153:272328 [3] NCCL INFO Channel 12/0 : 115[3] -> 114[2] via P2P/IPC
+ip-26-0-168-34:272155:272329 [5] NCCL INFO Channel 09/0 : 117[5] -> 116[4] via P2P/IPC
+ip-26-0-161-123:32707:32878 [3] NCCL INFO Channel 07/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO Channel 13/0 : 63[7] -> 62[6] via P2P/IPC
+ip-26-0-160-225:124164:124336 [4] NCCL INFO Channel 09/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-168-34:272156:272325 [6] NCCL INFO Channel 15/0 : 118[6] -> 117[5] via P2P/IPC
+ip-26-0-168-52:2722048:2722220 [3] NCCL INFO Channel 03/0 : 123[3] -> 122[2] via P2P/IPC
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO Channel 10/0 : 78[6] -> 77[5] via P2P/IPC
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO Channel 03/0 : 77[5] -> 76[4] via P2P/IPC
+ip-26-0-161-142:62886:63057 [5] NCCL INFO Channel 14/0 : 45[5] -> 44[4] via P2P/IPC
+ip-26-0-161-123:32707:32878 [3] NCCL INFO Channel 09/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO Channel 11/0 : 75[3] -> 74[2] via P2P/IPC
+ip-26-0-160-225:124164:124336 [4] NCCL INFO Channel 10/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-78:114515:114690 [3] NCCL INFO Channel 03/0 : 67[3] -> 66[2] via P2P/IPC
+ip-26-0-168-34:272153:272328 [3] NCCL INFO Channel 13/0 : 115[3] -> 114[2] via P2P/IPC
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO Channel 04/0 : 55[7] -> 54[6] via P2P/IPC
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO Channel 01/0 : 59[3] -> 58[2] via P2P/IPC
+ip-26-0-161-78:114516:114691 [4] NCCL INFO Channel 10/0 : 68[4] -> 67[3] via P2P/IPC
+ip-26-0-168-34:272155:272329 [5] NCCL INFO Channel 10/0 : 117[5] -> 116[4] via P2P/IPC
+ip-26-0-161-123:32707:32878 [3] NCCL INFO Channel 11/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-160-225:124164:124336 [4] NCCL INFO Channel 12/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO Channel 11/0 : 78[6] -> 77[5] via P2P/IPC
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO Channel 06/0 : 77[5] -> 76[4] via P2P/IPC
+ip-26-0-168-52:2722048:2722220 [3] NCCL INFO Channel 04/0 : 123[3] -> 122[2] via P2P/IPC
+ip-26-0-161-123:32707:32878 [3] NCCL INFO Channel 12/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO Channel 12/0 : 75[3] -> 74[2] via P2P/IPC
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO Channel 03/0 : 59[3] -> 58[2] via P2P/IPC
+ip-26-0-161-78:114518:114693 [6] NCCL INFO Channel 01/0 : 70[6] -> 69[5] via P2P/IPC
+ip-26-0-168-52:2722048:2722220 [3] NCCL INFO Channel 05/0 : 123[3] -> 122[2] via P2P/IPC
+ip-26-0-161-123:32707:32878 [3] NCCL INFO Channel 13/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-161-78:114515:114690 [3] NCCL INFO Channel 04/0 : 67[3] -> 66[2] via P2P/IPC
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO Channel 12/0 : 78[6] -> 77[5] via P2P/IPC
+ip-26-0-161-78:114516:114691 [4] NCCL INFO Channel 12/0 : 68[4] -> 67[3] via P2P/IPC
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO Channel 01/0 : 62[6] -> 61[5] via P2P/IPC
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO Channel 05/0 : 55[7] -> 54[6] via P2P/IPC
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO Channel 07/0 : 77[5] -> 76[4] via P2P/IPC
+ip-26-0-161-123:32707:32878 [3] NCCL INFO Channel 14/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-168-52:2722048:2722220 [3] NCCL INFO Channel 06/0 : 123[3] -> 122[2] via P2P/IPC
+ip-26-0-161-78:114518:114693 [6] NCCL INFO Channel 02/0 : 70[6] -> 69[5] via P2P/IPC
+ip-26-0-161-123:32707:32878 [3] NCCL INFO Channel 15/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-168-52:2722048:2722220 [3] NCCL INFO Channel 07/0 : 123[3] -> 122[2] via P2P/IPC
+ip-26-0-168-30:235653:235825 [5] NCCL INFO Channel 11/0 : 109[5] -> 108[4] via P2P/IPC
+ip-26-0-161-123:32710:32883 [6] NCCL INFO Channel 12/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-161-103:96894:97068 [6] NCCL INFO Channel 04/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-161-78:114515:114690 [3] NCCL INFO Channel 05/0 : 67[3] -> 66[2] via P2P/IPC
+ip-26-0-161-103:96893:97067 [5] NCCL INFO Channel 03/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO Channel 15/0 : 101[5] -> 100[4] via P2P/IPC
+ip-26-0-168-52:2722048:2722220 [3] NCCL INFO Channel 09/0 : 123[3] -> 122[2] via P2P/IPC
+ip-26-0-161-103:96894:97068 [6] NCCL INFO Channel 07/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-161-78:114517:114687 [5] NCCL INFO Channel 01/0 : 69[5] -> 68[4] via P2P/IPC
+ip-26-0-161-123:32710:32883 [6] NCCL INFO Channel 15/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO Channel 09/0 : 55[7] -> 54[6] via P2P/IPC
+ip-26-0-161-103:96891:97063 [3] NCCL INFO Channel 09/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-161-78:114518:114693 [6] NCCL INFO Channel 03/0 : 70[6] -> 69[5] via P2P/IPC
+ip-26-0-161-103:96893:97067 [5] NCCL INFO Channel 06/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO Channel 13/0 : 75[3] -> 74[2] via P2P/IPC
+ip-26-0-168-34:272153:272328 [3] NCCL INFO Channel 14/0 : 115[3] -> 114[2] via P2P/IPC
+ip-26-0-161-78:114515:114690 [3] NCCL INFO Channel 06/0 : 67[3] -> 66[2] via P2P/IPC
+ip-26-0-161-123:32709:32882 [5] NCCL INFO Channel 14/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-168-34:272155:272329 [5] NCCL INFO Channel 11/0 : 117[5] -> 116[4] via P2P/IPC
+ip-26-0-168-52:2722048:2722220 [3] NCCL INFO Channel 11/0 : 123[3] -> 122[2] via P2P/IPC
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO Channel 14/0 : 75[3] -> 74[2] via P2P/IPC
+ip-26-0-161-78:114517:114687 [5] NCCL INFO Channel 02/0 : 69[5] -> 68[4] via P2P/IPC
+ip-26-0-161-103:96894:97068 [6] NCCL INFO Channel 09/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-161-78:114518:114693 [6] NCCL INFO Channel 04/0 : 70[6] -> 69[5] via P2P/IPC
+ip-26-0-168-52:2722050:2722225 [5] NCCL INFO Channel 01/0 : 125[5] -> 124[4] via P2P/IPC
+ip-26-0-161-103:96891:97063 [3] NCCL INFO Channel 11/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-161-103:96893:97067 [5] NCCL INFO Channel 07/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-160-225:124165:124341 [5] NCCL INFO Channel 01/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-168-52:2722048:2722220 [3] NCCL INFO Channel 12/0 : 123[3] -> 122[2] via P2P/IPC
+ip-26-0-161-138:96566:96755 [3] NCCL INFO Channel 15/0 : 35[3] -> 34[2] via P2P/IPC
+ip-26-0-161-78:114515:114690 [3] NCCL INFO Channel 07/0 : 67[3] -> 66[2] via P2P/IPC
+ip-26-0-161-103:96894:97068 [6] NCCL INFO Channel 10/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-168-52:2722051:2722224 [6] NCCL INFO Channel 01/0 : 126[6] -> 125[5] via P2P/IPC
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO Channel 15/0 : 75[3] -> 74[2] via P2P/IPC
+ip-26-0-161-78:114517:114687 [5] NCCL INFO Channel 03/0 : 69[5] -> 68[4] via P2P/IPC
+ip-26-0-168-52:2722050:2722225 [5] NCCL INFO Channel 02/0 : 125[5] -> 124[4] via P2P/IPC
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO Channel 04/0 : 59[3] -> 58[2] via P2P/IPC
+ip-26-0-161-78:114518:114693 [6] NCCL INFO Channel 07/0 : 70[6] -> 69[5] via P2P/IPC
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO Channel 10/0 : 55[7] -> 54[6] via P2P/IPC
+ip-26-0-161-103:96891:97063 [3] NCCL INFO Channel 12/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-161-103:96893:97067 [5] NCCL INFO Channel 09/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-160-225:124166:124339 [6] NCCL INFO Channel 01/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-225:124165:124341 [5] NCCL INFO Channel 02/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO Channel 09/0 : 77[5] -> 76[4] via P2P/IPC
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO Channel 15/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO Channel 05/0 : 59[3] -> 58[2] via P2P/IPC
+ip-26-0-161-78:114515:114690 [3] NCCL INFO Channel 09/0 : 67[3] -> 66[2] via P2P/IPC
+ip-26-0-161-103:96894:97068 [6] NCCL INFO Channel 11/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO Channel 02/0 : 62[6] -> 61[5] via P2P/IPC
+ip-26-0-168-52:2722048:2722220 [3] NCCL INFO Channel 13/0 : 123[3] -> 122[2] via P2P/IPC
+ip-26-0-160-225:124163:124340 [3] NCCL INFO Channel 01/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-168-52:2722051:2722224 [6] NCCL INFO Channel 02/0 : 126[6] -> 125[5] via P2P/IPC
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO Channel 15/0 : 78[6] -> 77[5] via P2P/IPC
+ip-26-0-161-78:114518:114693 [6] NCCL INFO Channel 09/0 : 70[6] -> 69[5] via P2P/IPC
+ip-26-0-161-103:96891:97063 [3] NCCL INFO Channel 13/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-161-103:96893:97067 [5] NCCL INFO Channel 10/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-160-225:124166:124339 [6] NCCL INFO Channel 02/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-168-52:2722050:2722225 [5] NCCL INFO Channel 03/0 : 125[5] -> 124[4] via P2P/IPC
+ip-26-0-161-78:114515:114690 [3] NCCL INFO Channel 11/0 : 67[3] -> 66[2] via P2P/IPC
+ip-26-0-160-225:124165:124341 [5] NCCL INFO Channel 03/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO Channel 03/0 : 62[6] -> 61[5] via P2P/IPC
+ip-26-0-161-103:96894:97068 [6] NCCL INFO Channel 12/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO Channel 06/0 : 59[3] -> 58[2] via P2P/IPC
+ip-26-0-161-78:114518:114693 [6] NCCL INFO Channel 10/0 : 70[6] -> 69[5] via P2P/IPC
+ip-26-0-160-225:124163:124340 [3] NCCL INFO Channel 03/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-78:114515:114690 [3] NCCL INFO Channel 12/0 : 67[3] -> 66[2] via P2P/IPC
+ip-26-0-161-103:96893:97067 [5] NCCL INFO Channel 11/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO Channel 04/0 : 62[6] -> 61[5] via P2P/IPC
+ip-26-0-161-78:114518:114693 [6] NCCL INFO Channel 11/0 : 70[6] -> 69[5] via P2P/IPC
+ip-26-0-168-34:272155:272329 [5] NCCL INFO Channel 14/0 : 117[5] -> 116[4] via P2P/IPC
+ip-26-0-161-103:96891:97063 [3] NCCL INFO Channel 14/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO Channel 07/0 : 59[3] -> 58[2] via P2P/IPC
+ip-26-0-160-225:124166:124339 [6] NCCL INFO Channel 03/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-225:124165:124341 [5] NCCL INFO Channel 06/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-103:96894:97068 [6] NCCL INFO Channel 15/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-161-78:114515:114690 [3] NCCL INFO Channel 13/0 : 67[3] -> 66[2] via P2P/IPC
+ip-26-0-161-103:96893:97067 [5] NCCL INFO Channel 14/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-161-78:114518:114693 [6] NCCL INFO Channel 12/0 : 70[6] -> 69[5] via P2P/IPC
+ip-26-0-160-225:124163:124340 [3] NCCL INFO Channel 04/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO Channel 07/0 : 62[6] -> 61[5] via P2P/IPC
+ip-26-0-161-78:114515:114690 [3] NCCL INFO Channel 14/0 : 67[3] -> 66[2] via P2P/IPC
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO Channel 11/0 : 55[7] -> 54[6] via P2P/IPC
+ip-26-0-161-138:96568:96753 [5] NCCL INFO Channel 15/0 : 37[5] -> 36[4] via P2P/IPC
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO Channel 09/0 : 59[3] -> 58[2] via P2P/IPC
+ip-26-0-160-225:124166:124339 [6] NCCL INFO Channel 04/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-103:96891:97063 [3] NCCL INFO Channel 15/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-160-225:124165:124341 [5] NCCL INFO Channel 07/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-78:114518:114693 [6] NCCL INFO Channel 15/0 : 70[6] -> 69[5] via P2P/IPC
+ip-26-0-161-103:96893:97067 [5] NCCL INFO Channel 15/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO Channel 10/0 : 77[5] -> 76[4] via P2P/IPC
+ip-26-0-160-225:124163:124340 [3] NCCL INFO Channel 05/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-168-52:2722048:2722220 [3] NCCL INFO Channel 14/0 : 123[3] -> 122[2] via P2P/IPC
+ip-26-0-168-52:2722051:2722224 [6] NCCL INFO Channel 03/0 : 126[6] -> 125[5] via P2P/IPC
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO Channel 01/0 : 61[5] -> 60[4] via P2P/IPC
+ip-26-0-160-225:124166:124339 [6] NCCL INFO Channel 07/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-78:114515:114690 [3] NCCL INFO Channel 15/0 : 67[3] -> 66[2] via P2P/IPC
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO Channel 09/0 : 62[6] -> 61[5] via P2P/IPC
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO Channel 11/0 : 59[3] -> 58[2] via P2P/IPC
+ip-26-0-160-225:124165:124341 [5] NCCL INFO Channel 09/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-168-52:2722050:2722225 [5] NCCL INFO Channel 06/0 : 125[5] -> 124[4] via P2P/IPC
+ip-26-0-168-34:272153:272328 [3] NCCL INFO Channel 15/0 : 115[3] -> 114[2] via P2P/IPC
+ip-26-0-160-225:124163:124340 [3] NCCL INFO Channel 06/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO Channel 02/0 : 61[5] -> 60[4] via P2P/IPC
+ip-26-0-168-52:2722048:2722220 [3] NCCL INFO Channel 15/0 : 123[3] -> 122[2] via P2P/IPC
+ip-26-0-160-225:124166:124339 [6] NCCL INFO Channel 09/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-168-52:2722051:2722224 [6] NCCL INFO Channel 04/0 : 126[6] -> 125[5] via P2P/IPC
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO Channel 10/0 : 62[6] -> 61[5] via P2P/IPC
+ip-26-0-160-225:124165:124341 [5] NCCL INFO Channel 10/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-168-52:2722050:2722225 [5] NCCL INFO Channel 07/0 : 125[5] -> 124[4] via P2P/IPC
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO Channel 12/0 : 59[3] -> 58[2] via P2P/IPC
+ip-26-0-161-78:114517:114687 [5] NCCL INFO Channel 06/0 : 69[5] -> 68[4] via P2P/IPC
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO Channel 03/0 : 61[5] -> 60[4] via P2P/IPC
+ip-26-0-160-225:124163:124340 [3] NCCL INFO Channel 07/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-225:124166:124339 [6] NCCL INFO Channel 10/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-168-52:2722051:2722224 [6] NCCL INFO Channel 07/0 : 126[6] -> 125[5] via P2P/IPC
+ip-26-0-160-225:124165:124341 [5] NCCL INFO Channel 11/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-168-52:2722050:2722225 [5] NCCL INFO Channel 09/0 : 125[5] -> 124[4] via P2P/IPC
+ip-26-0-161-142:62886:63057 [5] NCCL INFO Channel 15/0 : 45[5] -> 44[4] via P2P/IPC
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO Channel 11/0 : 62[6] -> 61[5] via P2P/IPC
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO Channel 13/0 : 59[3] -> 58[2] via P2P/IPC
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO Channel 01/0 : 52[4] -> 51[3] via P2P/IPC
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO Channel 06/0 : 61[5] -> 60[4] via P2P/IPC
+ip-26-0-160-225:124163:124340 [3] NCCL INFO Channel 09/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-225:124166:124339 [6] NCCL INFO Channel 11/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO Channel 12/0 : 62[6] -> 61[5] via P2P/IPC
+ip-26-0-160-225:124165:124341 [5] NCCL INFO Channel 14/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO Channel 14/0 : 59[3] -> 58[2] via P2P/IPC
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO Channel 12/0 : 55[7] -> 54[6] via P2P/IPC
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO Channel 07/0 : 61[5] -> 60[4] via P2P/IPC
+ip-26-0-160-225:124163:124340 [3] NCCL INFO Channel 11/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-225:124166:124339 [6] NCCL INFO Channel 12/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-168-34:272155:272329 [5] NCCL INFO Channel 15/0 : 117[5] -> 116[4] via P2P/IPC
+ip-26-0-160-225:124165:124341 [5] NCCL INFO Channel 15/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO Channel 02/0 : 52[4] -> 51[3] via P2P/IPC
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO Channel 09/0 : 61[5] -> 60[4] via P2P/IPC
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO Channel 15/0 : 62[6] -> 61[5] via P2P/IPC
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO Channel 15/0 : 59[3] -> 58[2] via P2P/IPC
+ip-26-0-160-225:124163:124340 [3] NCCL INFO Channel 12/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-225:124166:124339 [6] NCCL INFO Channel 15/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO Channel 13/0 : 55[7] -> 54[6] via P2P/IPC
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO Channel 10/0 : 61[5] -> 60[4] via P2P/IPC
+ip-26-0-160-225:124163:124340 [3] NCCL INFO Channel 13/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO Channel 04/0 : 52[4] -> 51[3] via P2P/IPC
+ip-26-0-161-78:114517:114687 [5] NCCL INFO Channel 07/0 : 69[5] -> 68[4] via P2P/IPC
+ip-26-0-168-52:2722051:2722224 [6] NCCL INFO Channel 09/0 : 126[6] -> 125[5] via P2P/IPC
+ip-26-0-168-52:2722050:2722225 [5] NCCL INFO Channel 10/0 : 125[5] -> 124[4] via P2P/IPC
+ip-26-0-160-225:124163:124340 [3] NCCL INFO Channel 14/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO Channel 09/0 : 52[4] -> 51[3] via P2P/IPC
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO Channel 10/0 : 52[4] -> 51[3] via P2P/IPC
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO Channel 01/0 : 51[3] -> 50[2] via P2P/IPC
+ip-26-0-168-52:2722051:2722224 [6] NCCL INFO Channel 10/0 : 126[6] -> 125[5] via P2P/IPC
+ip-26-0-168-52:2722051:2722224 [6] NCCL INFO Channel 11/0 : 126[6] -> 125[5] via P2P/IPC
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO Channel 12/0 : 52[4] -> 51[3] via P2P/IPC
+ip-26-0-168-52:2722051:2722224 [6] NCCL INFO Channel 12/0 : 126[6] -> 125[5] via P2P/IPC
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO Channel 01/0 : 54[6] -> 53[5] via P2P/IPC
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO Channel 03/0 : 51[3] -> 50[2] via P2P/IPC
+ip-26-0-168-52:2722051:2722224 [6] NCCL INFO Channel 15/0 : 126[6] -> 125[5] via P2P/IPC
+ip-26-0-161-123:32709:32882 [5] NCCL INFO Channel 15/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO Channel 02/0 : 54[6] -> 53[5] via P2P/IPC
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO Channel 04/0 : 51[3] -> 50[2] via P2P/IPC
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO Channel 03/0 : 54[6] -> 53[5] via P2P/IPC
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO Channel 05/0 : 51[3] -> 50[2] via P2P/IPC
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO Channel 01/0 : 53[5] -> 52[4] via P2P/IPC
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO Channel 11/0 : 77[5] -> 76[4] via P2P/IPC
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO Channel 11/0 : 61[5] -> 60[4] via P2P/IPC
+ip-26-0-161-78:114517:114687 [5] NCCL INFO Channel 09/0 : 69[5] -> 68[4] via P2P/IPC
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO Channel 06/0 : 51[3] -> 50[2] via P2P/IPC
+ip-26-0-168-52:2722050:2722225 [5] NCCL INFO Channel 11/0 : 125[5] -> 124[4] via P2P/IPC
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO Channel 02/0 : 53[5] -> 52[4] via P2P/IPC
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO Channel 04/0 : 54[6] -> 53[5] via P2P/IPC
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO Channel 07/0 : 51[3] -> 50[2] via P2P/IPC
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO Channel 03/0 : 53[5] -> 52[4] via P2P/IPC
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO Channel 07/0 : 54[6] -> 53[5] via P2P/IPC
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO Channel 09/0 : 51[3] -> 50[2] via P2P/IPC
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO Channel 06/0 : 53[5] -> 52[4] via P2P/IPC
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO Channel 09/0 : 54[6] -> 53[5] via P2P/IPC
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO Channel 11/0 : 51[3] -> 50[2] via P2P/IPC
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO Channel 10/0 : 54[6] -> 53[5] via P2P/IPC
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO Channel 07/0 : 53[5] -> 52[4] via P2P/IPC
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO Channel 12/0 : 51[3] -> 50[2] via P2P/IPC
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO Channel 11/0 : 54[6] -> 53[5] via P2P/IPC
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO Channel 09/0 : 53[5] -> 52[4] via P2P/IPC
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO Channel 13/0 : 51[3] -> 50[2] via P2P/IPC
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO Channel 12/0 : 54[6] -> 53[5] via P2P/IPC
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO Channel 10/0 : 53[5] -> 52[4] via P2P/IPC
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO Channel 14/0 : 51[3] -> 50[2] via P2P/IPC
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO Channel 11/0 : 53[5] -> 52[4] via P2P/IPC
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO Channel 15/0 : 54[6] -> 53[5] via P2P/IPC
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO Channel 15/0 : 51[3] -> 50[2] via P2P/IPC
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO Channel 14/0 : 53[5] -> 52[4] via P2P/IPC
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO Channel 14/0 : 77[5] -> 76[4] via P2P/IPC
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO Channel 15/0 : 53[5] -> 52[4] via P2P/IPC
+ip-26-0-160-225:124163:124340 [3] NCCL INFO Channel 15/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-168-30:235653:235825 [5] NCCL INFO Channel 14/0 : 109[5] -> 108[4] via P2P/IPC
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO Channel 14/0 : 61[5] -> 60[4] via P2P/IPC
+ip-26-0-168-30:235653:235825 [5] NCCL INFO Channel 15/0 : 109[5] -> 108[4] via P2P/IPC
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO Channel 02/0 : 84[4] -> 83[3] via P2P/IPC
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO Channel 11/0 : 87[7] -> 86[6] via P2P/IPC
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO Channel 04/0 : 84[4] -> 83[3] via P2P/IPC
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO Channel 12/0 : 87[7] -> 86[6] via P2P/IPC
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO Channel 09/0 : 84[4] -> 83[3] via P2P/IPC
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO Channel 15/0 : 61[5] -> 60[4] via P2P/IPC
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO Channel 13/0 : 87[7] -> 86[6] via P2P/IPC
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO Channel 15/0 : 77[5] -> 76[4] via P2P/IPC
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO Channel 10/0 : 84[4] -> 83[3] via P2P/IPC
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO Channel 01/0 : 83[3] -> 82[2] via P2P/IPC
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO Channel 12/0 : 84[4] -> 83[3] via P2P/IPC
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO Channel 03/0 : 83[3] -> 82[2] via P2P/IPC
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO Channel 04/0 : 83[3] -> 82[2] via P2P/IPC
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO Channel 01/0 : 86[6] -> 85[5] via P2P/IPC
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO Channel 05/0 : 83[3] -> 82[2] via P2P/IPC
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO Channel 02/0 : 86[6] -> 85[5] via P2P/IPC
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO Channel 06/0 : 83[3] -> 82[2] via P2P/IPC
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO Channel 01/0 : 85[5] -> 84[4] via P2P/IPC
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO Channel 03/0 : 86[6] -> 85[5] via P2P/IPC
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO Channel 07/0 : 83[3] -> 82[2] via P2P/IPC
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO Channel 02/0 : 85[5] -> 84[4] via P2P/IPC
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO Channel 04/0 : 86[6] -> 85[5] via P2P/IPC
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO Channel 09/0 : 83[3] -> 82[2] via P2P/IPC
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO Channel 03/0 : 85[5] -> 84[4] via P2P/IPC
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO Channel 07/0 : 86[6] -> 85[5] via P2P/IPC
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO Channel 11/0 : 83[3] -> 82[2] via P2P/IPC
+ip-26-0-161-78:114517:114687 [5] NCCL INFO Channel 10/0 : 69[5] -> 68[4] via P2P/IPC
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO Channel 06/0 : 85[5] -> 84[4] via P2P/IPC
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO Channel 09/0 : 86[6] -> 85[5] via P2P/IPC
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO Channel 12/0 : 83[3] -> 82[2] via P2P/IPC
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO Channel 07/0 : 85[5] -> 84[4] via P2P/IPC
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO Channel 10/0 : 86[6] -> 85[5] via P2P/IPC
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO Channel 13/0 : 83[3] -> 82[2] via P2P/IPC
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO Channel 09/0 : 85[5] -> 84[4] via P2P/IPC
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO Channel 11/0 : 86[6] -> 85[5] via P2P/IPC
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO Channel 14/0 : 83[3] -> 82[2] via P2P/IPC
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO Channel 10/0 : 85[5] -> 84[4] via P2P/IPC
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO Channel 12/0 : 86[6] -> 85[5] via P2P/IPC
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO Channel 15/0 : 83[3] -> 82[2] via P2P/IPC
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO Channel 11/0 : 85[5] -> 84[4] via P2P/IPC
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO Channel 15/0 : 86[6] -> 85[5] via P2P/IPC
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO Channel 14/0 : 85[5] -> 84[4] via P2P/IPC
+ip-26-0-168-52:2722050:2722225 [5] NCCL INFO Channel 14/0 : 125[5] -> 124[4] via P2P/IPC
+ip-26-0-161-78:114517:114687 [5] NCCL INFO Channel 11/0 : 69[5] -> 68[4] via P2P/IPC
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO Channel 15/0 : 85[5] -> 84[4] via P2P/IPC
+ip-26-0-168-52:2722050:2722225 [5] NCCL INFO Channel 15/0 : 125[5] -> 124[4] via P2P/IPC
+ip-26-0-161-78:114517:114687 [5] NCCL INFO Channel 14/0 : 69[5] -> 68[4] via P2P/IPC
+ip-26-0-161-78:114517:114687 [5] NCCL INFO Channel 15/0 : 69[5] -> 68[4] via P2P/IPC
+ip-26-0-161-142:62881:63059 [0] NCCL INFO Connected all rings
+ip-26-0-161-142:62881:63059 [0] NCCL INFO Channel 02/0 : 40[0] -> 41[1] via P2P/IPC
+ip-26-0-168-52:2722045:2722219 [0] NCCL INFO Connected all rings
+ip-26-0-168-52:2722045:2722219 [0] NCCL INFO Channel 02/0 : 120[0] -> 121[1] via P2P/IPC
+ip-26-0-161-103:96888:97066 [0] NCCL INFO Connected all rings
+ip-26-0-161-103:96888:97066 [0] NCCL INFO Channel 02/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-161-142:62883:63058 [2] NCCL INFO Connected all rings
+ip-26-0-160-225:124161:124337 [1] NCCL INFO Connected all rings
+ip-26-0-161-103:96889:97065 [1] NCCL INFO Connected all rings
+ip-26-0-161-123:32706:32877 [2] NCCL INFO Connected all rings
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO Connected all rings
+ip-26-0-161-138:96564:96757 [1] NCCL INFO Connected all rings
+ip-26-0-168-52:2722047:2722221 [2] NCCL INFO Connected all rings
+ip-26-0-160-225:124166:124339 [6] NCCL INFO Connected all rings
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO Connected all rings
+ip-26-0-160-225:124161:124337 [1] NCCL INFO Channel 01/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-225:124161:124337 [1] NCCL INFO Channel 03/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-168-34:272151:272326 [1] NCCL INFO Connected all rings
+ip-26-0-160-225:124161:124337 [1] NCCL INFO Channel 04/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-225:124161:124337 [1] NCCL INFO Channel 05/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-225:124161:124337 [1] NCCL INFO Channel 06/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-168-52:2722045:2722219 [0] NCCL INFO Channel 03/0 : 120[0] -> 121[1] via P2P/IPC
+ip-26-0-160-225:124161:124337 [1] NCCL INFO Channel 07/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-168-30:235648:235826 [0] NCCL INFO Connected all rings
+ip-26-0-168-30:235648:235826 [0] NCCL INFO Channel 02/0 : 104[0] -> 105[1] via P2P/IPC
+ip-26-0-168-52:2722045:2722219 [0] NCCL INFO Channel 04/0 : 120[0] -> 121[1] via P2P/IPC
+ip-26-0-161-103:96889:97065 [1] NCCL INFO Channel 01/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-168-52:2722045:2722219 [0] NCCL INFO Channel 05/0 : 120[0] -> 121[1] via P2P/IPC
+ip-26-0-168-52:2722045:2722219 [0] NCCL INFO Channel 06/0 : 120[0] -> 121[1] via P2P/IPC
+ip-26-0-161-103:96889:97065 [1] NCCL INFO Channel 03/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-168-52:2722045:2722219 [0] NCCL INFO Channel 07/0 : 120[0] -> 121[1] via P2P/IPC
+ip-26-0-168-52:2722045:2722219 [0] NCCL INFO Channel 10/0 : 120[0] -> 121[1] via P2P/IPC
+ip-26-0-161-138:96564:96757 [1] NCCL INFO Channel 01/0 : 33[1] -> 34[2] via P2P/IPC
+ip-26-0-161-103:96894:97068 [6] NCCL INFO Connected all rings
+ip-26-0-161-123:32704:32881 [0] NCCL INFO Connected all rings
+ip-26-0-161-123:32704:32881 [0] NCCL INFO Channel 02/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-168-52:2722045:2722219 [0] NCCL INFO Channel 11/0 : 120[0] -> 121[1] via P2P/IPC
+ip-26-0-161-123:32704:32881 [0] NCCL INFO Channel 03/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-161-142:62881:63059 [0] NCCL INFO Channel 03/0 : 40[0] -> 41[1] via P2P/IPC
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO Connected all rings
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO Channel 02/0 : 48[0] -> 49[1] via P2P/IPC
+ip-26-0-161-123:32704:32881 [0] NCCL INFO Channel 04/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-161-123:32704:32881 [0] NCCL INFO Channel 05/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-161-123:32704:32881 [0] NCCL INFO Channel 06/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-161-142:62881:63059 [0] NCCL INFO Channel 04/0 : 40[0] -> 41[1] via P2P/IPC
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO Connected all rings
+ip-26-0-161-123:32704:32881 [0] NCCL INFO Channel 07/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-161-142:62881:63059 [0] NCCL INFO Channel 05/0 : 40[0] -> 41[1] via P2P/IPC
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO Channel 03/0 : 48[0] -> 49[1] via P2P/IPC
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO Connected all rings
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO Channel 04/0 : 48[0] -> 49[1] via P2P/IPC
+ip-26-0-161-142:62883:63058 [2] NCCL INFO Channel 01/0 : 42[2] -> 43[3] via P2P/IPC
+ip-26-0-161-142:62881:63059 [0] NCCL INFO Channel 06/0 : 40[0] -> 41[1] via P2P/IPC
+ip-26-0-161-142:62883:63058 [2] NCCL INFO Channel 02/0 : 42[2] -> 43[3] via P2P/IPC
+ip-26-0-161-123:32704:32881 [0] NCCL INFO Channel 10/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-161-142:62881:63059 [0] NCCL INFO Channel 07/0 : 40[0] -> 41[1] via P2P/IPC
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO Connected all rings
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO Channel 02/0 : 80[0] -> 81[1] via P2P/IPC
+ip-26-0-168-34:272154:272330 [4] NCCL INFO Connected all rings
+ip-26-0-161-142:62883:63058 [2] NCCL INFO Channel 04/0 : 42[2] -> 43[3] via P2P/IPC
+ip-26-0-161-123:32704:32881 [0] NCCL INFO Channel 11/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-161-142:62881:63059 [0] NCCL INFO Channel 10/0 : 40[0] -> 41[1] via P2P/IPC
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO Channel 05/0 : 48[0] -> 49[1] via P2P/IPC
+ip-26-0-161-142:62883:63058 [2] NCCL INFO Channel 05/0 : 42[2] -> 43[3] via P2P/IPC
+ip-26-0-161-142:62881:63059 [0] NCCL INFO Channel 11/0 : 40[0] -> 41[1] via P2P/IPC
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO Connected all rings
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO Channel 03/0 : 80[0] -> 81[1] via P2P/IPC
+ip-26-0-161-142:62883:63058 [2] NCCL INFO Channel 06/0 : 42[2] -> 43[3] via P2P/IPC
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO Channel 06/0 : 48[0] -> 49[1] via P2P/IPC
+ip-26-0-161-142:62881:63059 [0] NCCL INFO Channel 12/0 : 40[0] -> 41[1] via P2P/IPC
+ip-26-0-161-123:32704:32881 [0] NCCL INFO Channel 12/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-161-142:62883:63058 [2] NCCL INFO Channel 07/0 : 42[2] -> 43[3] via P2P/IPC
+ip-26-0-161-142:62881:63059 [0] NCCL INFO Channel 13/0 : 40[0] -> 41[1] via P2P/IPC
+ip-26-0-161-123:32704:32881 [0] NCCL INFO Channel 13/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-161-142:62883:63058 [2] NCCL INFO Channel 09/0 : 42[2] -> 43[3] via P2P/IPC
+ip-26-0-161-142:62881:63059 [0] NCCL INFO Channel 14/0 : 40[0] -> 41[1] via P2P/IPC
+ip-26-0-161-142:62883:63058 [2] NCCL INFO Channel 10/0 : 42[2] -> 43[3] via P2P/IPC
+ip-26-0-161-142:62881:63059 [0] NCCL INFO Channel 15/0 : 40[0] -> 41[1] via P2P/IPC
+ip-26-0-161-123:32704:32881 [0] NCCL INFO Channel 14/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO Channel 04/0 : 80[0] -> 81[1] via P2P/IPC
+ip-26-0-161-142:62883:63058 [2] NCCL INFO Channel 12/0 : 42[2] -> 43[3] via P2P/IPC
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO Channel 07/0 : 48[0] -> 49[1] via P2P/IPC
+ip-26-0-161-142:62883:63058 [2] NCCL INFO Channel 13/0 : 42[2] -> 43[3] via P2P/IPC
+ip-26-0-161-123:32704:32881 [0] NCCL INFO Channel 15/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-161-142:62883:63058 [2] NCCL INFO Channel 14/0 : 42[2] -> 43[3] via P2P/IPC
+ip-26-0-161-142:62883:63058 [2] NCCL INFO Channel 15/0 : 42[2] -> 43[3] via P2P/IPC
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO Channel 05/0 : 80[0] -> 81[1] via P2P/IPC
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO Channel 06/0 : 80[0] -> 81[1] via P2P/IPC
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO Channel 07/0 : 80[0] -> 81[1] via P2P/IPC
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO Channel 10/0 : 80[0] -> 81[1] via P2P/IPC
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO Channel 11/0 : 80[0] -> 81[1] via P2P/IPC
+ip-26-0-161-78:114518:114693 [6] NCCL INFO Connected all rings
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO Channel 12/0 : 80[0] -> 81[1] via P2P/IPC
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO Channel 01/0 : 49[1] -> 50[2] via P2P/IPC
+ip-26-0-161-123:32707:32878 [3] NCCL INFO Connected all rings
+ip-26-0-168-34:272154:272330 [4] NCCL INFO Channel 01/0 : 116[4] -> 117[5] via P2P/IPC
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO Connected all rings
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO Connected all rings
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO Channel 02/0 : 72[0] -> 73[1] via P2P/IPC
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO Channel 13/0 : 80[0] -> 81[1] via P2P/IPC
+ip-26-0-168-30:235654:235821 [6] NCCL INFO Connected all rings
+ip-26-0-168-34:272154:272330 [4] NCCL INFO Channel 02/0 : 116[4] -> 117[5] via P2P/IPC
+ip-26-0-168-34:272154:272330 [4] NCCL INFO Channel 03/0 : 116[4] -> 117[5] via P2P/IPC
+ip-26-0-161-138:96563:96756 [0] NCCL INFO Connected all rings
+ip-26-0-161-138:96563:96756 [0] NCCL INFO Channel 02/0 : 32[0] -> 33[1] via P2P/IPC
+ip-26-0-161-103:96888:97066 [0] NCCL INFO Channel 03/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO Channel 14/0 : 80[0] -> 81[1] via P2P/IPC
+ip-26-0-161-138:96563:96756 [0] NCCL INFO Channel 03/0 : 32[0] -> 33[1] via P2P/IPC
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO Connected all rings
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO Channel 02/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-168-34:272154:272330 [4] NCCL INFO Channel 04/0 : 116[4] -> 117[5] via P2P/IPC
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO Connected all rings
+ip-26-0-168-34:272154:272330 [4] NCCL INFO Channel 06/0 : 116[4] -> 117[5] via P2P/IPC
+ip-26-0-161-138:96563:96756 [0] NCCL INFO Channel 04/0 : 32[0] -> 33[1] via P2P/IPC
+ip-26-0-161-142:62885:63054 [4] NCCL INFO Connected all rings
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO Channel 03/0 : 72[0] -> 73[1] via P2P/IPC
+ip-26-0-168-34:272154:272330 [4] NCCL INFO Channel 07/0 : 116[4] -> 117[5] via P2P/IPC
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO Channel 03/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO Channel 04/0 : 72[0] -> 73[1] via P2P/IPC
+ip-26-0-168-34:272154:272330 [4] NCCL INFO Channel 09/0 : 116[4] -> 117[5] via P2P/IPC
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO Channel 04/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO Channel 05/0 : 72[0] -> 73[1] via P2P/IPC
+ip-26-0-168-34:272154:272330 [4] NCCL INFO Channel 10/0 : 116[4] -> 117[5] via P2P/IPC
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO Channel 06/0 : 72[0] -> 73[1] via P2P/IPC
+ip-26-0-161-138:96563:96756 [0] NCCL INFO Channel 05/0 : 32[0] -> 33[1] via P2P/IPC
+ip-26-0-161-103:96888:97066 [0] NCCL INFO Channel 04/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO Channel 05/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-168-34:272154:272330 [4] NCCL INFO Channel 11/0 : 116[4] -> 117[5] via P2P/IPC
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO Channel 07/0 : 72[0] -> 73[1] via P2P/IPC
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO Channel 06/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-165-202:153567:153742 [5] NCCL INFO Connected all rings
+ip-26-0-161-138:96563:96756 [0] NCCL INFO Channel 06/0 : 32[0] -> 33[1] via P2P/IPC
+ip-26-0-168-52:2722045:2722219 [0] NCCL INFO Channel 12/0 : 120[0] -> 121[1] via P2P/IPC
+ip-26-0-168-34:272154:272330 [4] NCCL INFO Channel 12/0 : 116[4] -> 117[5] via P2P/IPC
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO Channel 10/0 : 72[0] -> 73[1] via P2P/IPC
+ip-26-0-161-103:96888:97066 [0] NCCL INFO Channel 05/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-161-78:114515:114690 [3] NCCL INFO Connected all rings
+ip-26-0-160-225:124160:124335 [0] NCCL INFO Connected all rings
+ip-26-0-160-225:124160:124335 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO Connected all rings
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO Channel 07/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO Channel 02/0 : 96[0] -> 97[1] via P2P/IPC
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO Connected all rings
+ip-26-0-168-34:272154:272330 [4] NCCL INFO Channel 14/0 : 116[4] -> 117[5] via P2P/IPC
+ip-26-0-161-138:96563:96756 [0] NCCL INFO Channel 07/0 : 32[0] -> 33[1] via P2P/IPC
+ip-26-0-160-225:124160:124335 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-168-30:235648:235826 [0] NCCL INFO Channel 03/0 : 104[0] -> 105[1] via P2P/IPC
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO Channel 10/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-168-52:2722052:2722226 [7] NCCL INFO Connected all rings
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO Channel 11/0 : 72[0] -> 73[1] via P2P/IPC
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO Channel 03/0 : 96[0] -> 97[1] via P2P/IPC
+ip-26-0-160-225:124160:124335 [0] NCCL INFO Channel 04/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-168-52:2722045:2722219 [0] NCCL INFO Channel 13/0 : 120[0] -> 121[1] via P2P/IPC
+ip-26-0-168-34:272154:272330 [4] NCCL INFO Channel 15/0 : 116[4] -> 117[5] via P2P/IPC
+ip-26-0-161-138:96563:96756 [0] NCCL INFO Channel 10/0 : 32[0] -> 33[1] via P2P/IPC
+ip-26-0-168-52:2722045:2722219 [0] NCCL INFO Channel 14/0 : 120[0] -> 121[1] via P2P/IPC
+ip-26-0-168-52:2722045:2722219 [0] NCCL INFO Channel 15/0 : 120[0] -> 121[1] via P2P/IPC
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO Channel 11/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-161-138:96563:96756 [0] NCCL INFO Channel 11/0 : 32[0] -> 33[1] via P2P/IPC
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO Channel 04/0 : 96[0] -> 97[1] via P2P/IPC
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO Connected all rings
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO Channel 12/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-161-138:96563:96756 [0] NCCL INFO Channel 12/0 : 32[0] -> 33[1] via P2P/IPC
+ip-26-0-161-103:96888:97066 [0] NCCL INFO Channel 06/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-168-30:235648:235826 [0] NCCL INFO Channel 04/0 : 104[0] -> 105[1] via P2P/IPC
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO Channel 05/0 : 96[0] -> 97[1] via P2P/IPC
+ip-26-0-161-103:96888:97066 [0] NCCL INFO Channel 07/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO Channel 13/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-168-30:235648:235826 [0] NCCL INFO Channel 05/0 : 104[0] -> 105[1] via P2P/IPC
+ip-26-0-161-138:96563:96756 [0] NCCL INFO Channel 13/0 : 32[0] -> 33[1] via P2P/IPC
+ip-26-0-160-225:124160:124335 [0] NCCL INFO Channel 05/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO Channel 01/0 : 51[3] -> 52[4] via P2P/IPC
+ip-26-0-161-138:96563:96756 [0] NCCL INFO Channel 14/0 : 32[0] -> 33[1] via P2P/IPC
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO Channel 12/0 : 72[0] -> 73[1] via P2P/IPC
+ip-26-0-160-225:124160:124335 [0] NCCL INFO Channel 06/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-138:96563:96756 [0] NCCL INFO Channel 15/0 : 32[0] -> 33[1] via P2P/IPC
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO Channel 14/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-160-225:124160:124335 [0] NCCL INFO Channel 07/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-202:153564:153737 [2] NCCL INFO Connected all rings
+ip-26-0-161-103:96888:97066 [0] NCCL INFO Channel 10/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO Channel 15/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-161-78:114512:114686 [0] NCCL INFO Connected all rings
+ip-26-0-161-78:114512:114686 [0] NCCL INFO Channel 02/0 : 64[0] -> 65[1] via P2P/IPC
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO Channel 02/0 : 51[3] -> 52[4] via P2P/IPC
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO Channel 03/0 : 51[3] -> 52[4] via P2P/IPC
+ip-26-0-161-103:96888:97066 [0] NCCL INFO Channel 11/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO Channel 05/0 : 51[3] -> 52[4] via P2P/IPC
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO Connected all rings
+ip-26-0-160-225:124160:124335 [0] NCCL INFO Channel 10/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO Channel 13/0 : 72[0] -> 73[1] via P2P/IPC
+ip-26-0-161-103:96888:97066 [0] NCCL INFO Channel 12/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-160-225:124160:124335 [0] NCCL INFO Channel 11/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO Channel 14/0 : 72[0] -> 73[1] via P2P/IPC
+ip-26-0-168-52:2722048:2722220 [3] NCCL INFO Connected all rings
+ip-26-0-161-123:32706:32877 [2] NCCL INFO Channel 01/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-161-103:96888:97066 [0] NCCL INFO Channel 13/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO Channel 15/0 : 72[0] -> 73[1] via P2P/IPC
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO Channel 06/0 : 51[3] -> 52[4] via P2P/IPC
+ip-26-0-165-202:153566:153741 [4] NCCL INFO Connected all rings
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO Channel 01/0 : 97[1] -> 98[2] via P2P/IPC
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO Channel 03/0 : 97[1] -> 98[2] via P2P/IPC
+ip-26-0-165-202:153568:153736 [6] NCCL INFO Connected all rings
+ip-26-0-161-78:114513:114688 [1] NCCL INFO Connected all rings
+ip-26-0-161-123:32706:32877 [2] NCCL INFO Channel 02/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO Channel 06/0 : 96[0] -> 97[1] via P2P/IPC
+ip-26-0-161-103:96888:97066 [0] NCCL INFO Channel 14/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-160-225:124160:124335 [0] NCCL INFO Channel 12/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-123:32706:32877 [2] NCCL INFO Channel 04/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-161-123:32706:32877 [2] NCCL INFO Channel 05/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-165-202:153567:153742 [5] NCCL INFO Channel 01/0 : 93[5] -> 94[6] via P2P/IPC
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO Channel 04/0 : 97[1] -> 98[2] via P2P/IPC
+ip-26-0-161-103:96888:97066 [0] NCCL INFO Channel 15/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-161-78:114512:114686 [0] NCCL INFO Channel 03/0 : 64[0] -> 65[1] via P2P/IPC
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO Channel 07/0 : 96[0] -> 97[1] via P2P/IPC
+ip-26-0-165-202:153567:153742 [5] NCCL INFO Channel 02/0 : 93[5] -> 94[6] via P2P/IPC
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO Channel 05/0 : 97[1] -> 98[2] via P2P/IPC
+ip-26-0-161-123:32706:32877 [2] NCCL INFO Channel 06/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-161-123:32706:32877 [2] NCCL INFO Channel 07/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-161-78:114512:114686 [0] NCCL INFO Channel 04/0 : 64[0] -> 65[1] via P2P/IPC
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO Channel 10/0 : 96[0] -> 97[1] via P2P/IPC
+ip-26-0-160-225:124167:124338 [7] NCCL INFO Connected all rings
+ip-26-0-168-30:235648:235826 [0] NCCL INFO Channel 06/0 : 104[0] -> 105[1] via P2P/IPC
+ip-26-0-161-138:96568:96753 [5] NCCL INFO Connected all rings
+ip-26-0-161-123:32706:32877 [2] NCCL INFO Channel 09/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-165-202:153567:153742 [5] NCCL INFO Channel 03/0 : 93[5] -> 94[6] via P2P/IPC
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO Channel 06/0 : 97[1] -> 98[2] via P2P/IPC
+ip-26-0-160-225:124160:124335 [0] NCCL INFO Channel 13/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO Channel 11/0 : 96[0] -> 97[1] via P2P/IPC
+ip-26-0-161-78:114512:114686 [0] NCCL INFO Channel 05/0 : 64[0] -> 65[1] via P2P/IPC
+ip-26-0-161-103:96891:97063 [3] NCCL INFO Connected all rings
+ip-26-0-161-123:32706:32877 [2] NCCL INFO Channel 10/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-161-103:96894:97068 [6] NCCL INFO Channel 01/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO Channel 07/0 : 97[1] -> 98[2] via P2P/IPC
+ip-26-0-165-202:153567:153742 [5] NCCL INFO Channel 04/0 : 93[5] -> 94[6] via P2P/IPC
+ip-26-0-161-78:114512:114686 [0] NCCL INFO Channel 06/0 : 64[0] -> 65[1] via P2P/IPC
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO Channel 12/0 : 96[0] -> 97[1] via P2P/IPC
+ip-26-0-160-225:124160:124335 [0] NCCL INFO Channel 14/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO Channel 01/0 : 73[1] -> 74[2] via P2P/IPC
+ip-26-0-161-103:96894:97068 [6] NCCL INFO Channel 02/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO Channel 09/0 : 97[1] -> 98[2] via P2P/IPC
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO Channel 13/0 : 96[0] -> 97[1] via P2P/IPC
+ip-26-0-161-123:32706:32877 [2] NCCL INFO Channel 12/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-161-138:96564:96757 [1] NCCL INFO Channel 03/0 : 33[1] -> 34[2] via P2P/IPC
+ip-26-0-161-138:96569:96758 [6] NCCL INFO Connected all rings
+ip-26-0-161-78:114512:114686 [0] NCCL INFO Channel 07/0 : 64[0] -> 65[1] via P2P/IPC
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO Channel 11/0 : 97[1] -> 98[2] via P2P/IPC
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO Channel 03/0 : 73[1] -> 74[2] via P2P/IPC
+ip-26-0-160-225:124160:124335 [0] NCCL INFO Channel 15/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO Channel 14/0 : 96[0] -> 97[1] via P2P/IPC
+ip-26-0-165-202:153567:153742 [5] NCCL INFO Channel 05/0 : 93[5] -> 94[6] via P2P/IPC
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO Channel 01/0 : 58[2] -> 59[3] via P2P/IPC
+ip-26-0-161-78:114512:114686 [0] NCCL INFO Channel 10/0 : 64[0] -> 65[1] via P2P/IPC
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO Channel 04/0 : 73[1] -> 74[2] via P2P/IPC
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO Channel 02/0 : 58[2] -> 59[3] via P2P/IPC
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO Channel 12/0 : 97[1] -> 98[2] via P2P/IPC
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO Channel 01/0 : 98[2] -> 99[3] via P2P/IPC
+ip-26-0-161-123:32706:32877 [2] NCCL INFO Channel 13/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO Channel 15/0 : 96[0] -> 97[1] via P2P/IPC
+ip-26-0-161-103:96894:97068 [6] NCCL INFO Channel 03/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-165-202:153567:153742 [5] NCCL INFO Channel 07/0 : 93[5] -> 94[6] via P2P/IPC
+ip-26-0-161-123:32706:32877 [2] NCCL INFO Channel 14/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-161-138:96565:96754 [2] NCCL INFO Connected all rings
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO Channel 04/0 : 58[2] -> 59[3] via P2P/IPC
+ip-26-0-161-103:96894:97068 [6] NCCL INFO Channel 04/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-161-138:96564:96757 [1] NCCL INFO Channel 04/0 : 33[1] -> 34[2] via P2P/IPC
+ip-26-0-161-78:114512:114686 [0] NCCL INFO Channel 11/0 : 64[0] -> 65[1] via P2P/IPC
+ip-26-0-165-202:153567:153742 [5] NCCL INFO Channel 09/0 : 93[5] -> 94[6] via P2P/IPC
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO Channel 02/0 : 98[2] -> 99[3] via P2P/IPC
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO Channel 13/0 : 97[1] -> 98[2] via P2P/IPC
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO Channel 05/0 : 58[2] -> 59[3] via P2P/IPC
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO Channel 04/0 : 98[2] -> 99[3] via P2P/IPC
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO Channel 14/0 : 97[1] -> 98[2] via P2P/IPC
+ip-26-0-161-78:114512:114686 [0] NCCL INFO Channel 12/0 : 64[0] -> 65[1] via P2P/IPC
+ip-26-0-161-103:96894:97068 [6] NCCL INFO Channel 05/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-161-138:96564:96757 [1] NCCL INFO Channel 05/0 : 33[1] -> 34[2] via P2P/IPC
+ip-26-0-165-202:153567:153742 [5] NCCL INFO Channel 10/0 : 93[5] -> 94[6] via P2P/IPC
+ip-26-0-161-123:32706:32877 [2] NCCL INFO Channel 15/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-161-123:32711:32884 [7] NCCL INFO Connected all rings
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO Channel 05/0 : 73[1] -> 74[2] via P2P/IPC
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO Channel 05/0 : 98[2] -> 99[3] via P2P/IPC
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO Channel 15/0 : 97[1] -> 98[2] via P2P/IPC
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO Channel 06/0 : 58[2] -> 59[3] via P2P/IPC
+ip-26-0-165-202:153567:153742 [5] NCCL INFO Channel 11/0 : 93[5] -> 94[6] via P2P/IPC
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO Channel 06/0 : 98[2] -> 99[3] via P2P/IPC
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO Channel 06/0 : 73[1] -> 74[2] via P2P/IPC
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO Channel 05/0 : 96[0] -> 103[7] via P2P/IPC
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO Channel 07/0 : 58[2] -> 59[3] via P2P/IPC
+ip-26-0-161-78:114512:114686 [0] NCCL INFO Channel 13/0 : 64[0] -> 65[1] via P2P/IPC
+ip-26-0-161-103:96894:97068 [6] NCCL INFO Channel 06/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-165-202:153567:153742 [5] NCCL INFO Channel 12/0 : 93[5] -> 94[6] via P2P/IPC
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO Channel 09/0 : 58[2] -> 59[3] via P2P/IPC
+ip-26-0-165-202:153564:153737 [2] NCCL INFO Channel 01/0 : 90[2] -> 91[3] via P2P/IPC
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO Channel 07/0 : 73[1] -> 74[2] via P2P/IPC
+ip-26-0-165-202:153567:153742 [5] NCCL INFO Channel 13/0 : 93[5] -> 94[6] via P2P/IPC
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO Connected all rings
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO Channel 09/0 : 73[1] -> 74[2] via P2P/IPC
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO Connected all rings
+ip-26-0-165-202:153564:153737 [2] NCCL INFO Channel 02/0 : 90[2] -> 91[3] via P2P/IPC
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO Channel 10/0 : 58[2] -> 59[3] via P2P/IPC
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO Connected all rings
+ip-26-0-161-103:96889:97065 [1] NCCL INFO Channel 04/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO Connected all rings
+ip-26-0-161-103:96893:97067 [5] NCCL INFO Connected all rings
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO Channel 12/0 : 58[2] -> 59[3] via P2P/IPC
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO Channel 07/0 : 98[2] -> 99[3] via P2P/IPC
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO Connected all rings
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO Channel 06/0 : 96[0] -> 103[7] via P2P/IPC
+ip-26-0-165-202:153568:153736 [6] NCCL INFO Channel 01/0 : 94[6] -> 95[7] via P2P/IPC
+ip-26-0-161-78:114512:114686 [0] NCCL INFO Channel 14/0 : 64[0] -> 65[1] via P2P/IPC
+ip-26-0-161-138:96564:96757 [1] NCCL INFO Channel 06/0 : 33[1] -> 34[2] via P2P/IPC
+ip-26-0-165-202:153567:153742 [5] NCCL INFO Channel 15/0 : 93[5] -> 94[6] via P2P/IPC
+ip-26-0-160-225:124161:124337 [1] NCCL INFO Channel 09/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO Channel 09/0 : 98[2] -> 99[3] via P2P/IPC
+ip-26-0-168-52:2722047:2722221 [2] NCCL INFO Channel 01/0 : 122[2] -> 123[3] via P2P/IPC
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO Channel 11/0 : 73[1] -> 74[2] via P2P/IPC
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO Channel 07/0 : 96[0] -> 103[7] via P2P/IPC
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO Channel 13/0 : 58[2] -> 59[3] via P2P/IPC
+ip-26-0-161-103:96890:97070 [2] NCCL INFO Connected all rings
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO Channel 10/0 : 98[2] -> 99[3] via P2P/IPC
+ip-26-0-161-142:62884:63061 [3] NCCL INFO Connected all rings
+ip-26-0-165-202:153564:153737 [2] NCCL INFO Channel 04/0 : 90[2] -> 91[3] via P2P/IPC
+ip-26-0-168-52:2722047:2722221 [2] NCCL INFO Channel 02/0 : 122[2] -> 123[3] via P2P/IPC
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO Channel 14/0 : 58[2] -> 59[3] via P2P/IPC
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO Channel 13/0 : 96[0] -> 103[7] via P2P/IPC
+ip-26-0-161-103:96889:97065 [1] NCCL INFO Channel 05/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-168-52:2722047:2722221 [2] NCCL INFO Channel 04/0 : 122[2] -> 123[3] via P2P/IPC
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO Connected all rings
+ip-26-0-165-202:153568:153736 [6] NCCL INFO Channel 02/0 : 94[6] -> 95[7] via P2P/IPC
+ip-26-0-165-202:153564:153737 [2] NCCL INFO Channel 05/0 : 90[2] -> 91[3] via P2P/IPC
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO Channel 12/0 : 98[2] -> 99[3] via P2P/IPC
+ip-26-0-161-103:96889:97065 [1] NCCL INFO Channel 06/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO Channel 14/0 : 96[0] -> 103[7] via P2P/IPC
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO Channel 15/0 : 58[2] -> 59[3] via P2P/IPC
+ip-26-0-168-52:2722047:2722221 [2] NCCL INFO Channel 05/0 : 122[2] -> 123[3] via P2P/IPC
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO Connected all rings
+ip-26-0-161-78:114512:114686 [0] NCCL INFO Channel 15/0 : 64[0] -> 65[1] via P2P/IPC
+ip-26-0-165-202:153568:153736 [6] NCCL INFO Channel 03/0 : 94[6] -> 95[7] via P2P/IPC
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO Channel 13/0 : 98[2] -> 99[3] via P2P/IPC
+ip-26-0-168-52:2722047:2722221 [2] NCCL INFO Channel 06/0 : 122[2] -> 123[3] via P2P/IPC
+ip-26-0-165-202:153564:153737 [2] NCCL INFO Channel 06/0 : 90[2] -> 91[3] via P2P/IPC
+ip-26-0-161-103:96891:97063 [3] NCCL INFO Channel 01/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-161-103:96889:97065 [1] NCCL INFO Channel 07/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO Channel 12/0 : 73[1] -> 74[2] via P2P/IPC
+ip-26-0-168-52:2722047:2722221 [2] NCCL INFO Channel 07/0 : 122[2] -> 123[3] via P2P/IPC
+ip-26-0-160-225:124161:124337 [1] NCCL INFO Channel 11/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-202:153568:153736 [6] NCCL INFO Channel 04/0 : 94[6] -> 95[7] via P2P/IPC
+ip-26-0-161-78:114513:114688 [1] NCCL INFO Channel 01/0 : 65[1] -> 66[2] via P2P/IPC
+ip-26-0-161-142:62882:63056 [1] NCCL INFO Connected all rings
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO Channel 14/0 : 98[2] -> 99[3] via P2P/IPC
+ip-26-0-160-225:124162:124342 [2] NCCL INFO Connected all rings
+ip-26-0-165-202:153564:153737 [2] NCCL INFO Channel 07/0 : 90[2] -> 91[3] via P2P/IPC
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO Channel 13/0 : 73[1] -> 74[2] via P2P/IPC
+ip-26-0-160-225:124161:124337 [1] NCCL INFO Channel 12/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-168-52:2722047:2722221 [2] NCCL INFO Channel 09/0 : 122[2] -> 123[3] via P2P/IPC
+ip-26-0-160-225:124166:124339 [6] NCCL INFO Channel 01/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-78:114513:114688 [1] NCCL INFO Channel 03/0 : 65[1] -> 66[2] via P2P/IPC
+ip-26-0-161-103:96892:97064 [4] NCCL INFO Connected all rings
+ip-26-0-168-52:2722047:2722221 [2] NCCL INFO Channel 10/0 : 122[2] -> 123[3] via P2P/IPC
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO Channel 15/0 : 96[0] -> 103[7] via P2P/IPC
+ip-26-0-165-202:153568:153736 [6] NCCL INFO Channel 05/0 : 94[6] -> 95[7] via P2P/IPC
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO Channel 15/0 : 98[2] -> 99[3] via P2P/IPC
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO Connected all rings
+ip-26-0-165-202:153564:153737 [2] NCCL INFO Channel 09/0 : 90[2] -> 91[3] via P2P/IPC
+ip-26-0-161-103:96891:97063 [3] NCCL INFO Channel 02/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-161-103:96889:97065 [1] NCCL INFO Channel 09/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-161-78:114513:114688 [1] NCCL INFO Channel 04/0 : 65[1] -> 66[2] via P2P/IPC
+ip-26-0-168-34:272152:272331 [2] NCCL INFO Connected all rings
+ip-26-0-165-202:153566:153741 [4] NCCL INFO Channel 01/0 : 92[4] -> 93[5] via P2P/IPC
+ip-26-0-161-123:32708:32879 [4] NCCL INFO Connected all rings
+ip-26-0-168-52:2722047:2722221 [2] NCCL INFO Channel 12/0 : 122[2] -> 123[3] via P2P/IPC
+ip-26-0-160-225:124161:124337 [1] NCCL INFO Channel 13/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-225:124166:124339 [6] NCCL INFO Channel 02/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-103:96891:97063 [3] NCCL INFO Channel 03/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-168-34:272151:272326 [1] NCCL INFO Channel 01/0 : 113[1] -> 114[2] via P2P/IPC
+ip-26-0-161-103:96889:97065 [1] NCCL INFO Channel 11/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO Connected all rings
+ip-26-0-160-225:124161:124337 [1] NCCL INFO Channel 14/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-202:153564:153737 [2] NCCL INFO Channel 10/0 : 90[2] -> 91[3] via P2P/IPC
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO Channel 09/0 : 97[1] -> 105[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO Channel 14/0 : 73[1] -> 74[2] via P2P/IPC
+ip-26-0-165-202:153568:153736 [6] NCCL INFO Channel 06/0 : 94[6] -> 95[7] via P2P/IPC
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO Connected all rings
+ip-26-0-160-225:124166:124339 [6] NCCL INFO Channel 03/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-78:114513:114688 [1] NCCL INFO Channel 05/0 : 65[1] -> 66[2] via P2P/IPC
+ip-26-0-165-202:153566:153741 [4] NCCL INFO Channel 02/0 : 92[4] -> 93[5] via P2P/IPC
+ip-26-0-161-138:96564:96757 [1] NCCL INFO Channel 07/0 : 33[1] -> 34[2] via P2P/IPC
+ip-26-0-168-34:272151:272326 [1] NCCL INFO Channel 03/0 : 113[1] -> 114[2] via P2P/IPC
+ip-26-0-161-103:96889:97065 [1] NCCL INFO Channel 12/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-161-103:96891:97063 [3] NCCL INFO Channel 05/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO Channel 15/0 : 73[1] -> 74[2] via P2P/IPC
+ip-26-0-161-138:96564:96757 [1] NCCL INFO Channel 09/0 : 33[1] -> 34[2] via P2P/IPC
+ip-26-0-161-78:114513:114688 [1] NCCL INFO Channel 06/0 : 65[1] -> 66[2] via P2P/IPC
+ip-26-0-165-202:153564:153737 [2] NCCL INFO Channel 12/0 : 90[2] -> 91[3] via P2P/IPC
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO Connected all rings
+ip-26-0-165-202:153568:153736 [6] NCCL INFO Channel 09/0 : 94[6] -> 95[7] via P2P/IPC
+ip-26-0-160-225:124161:124337 [1] NCCL INFO Channel 15/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO Channel 05/0 : 72[0] -> 79[7] via P2P/IPC
+ip-26-0-161-103:96889:97065 [1] NCCL INFO Channel 13/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-165-202:153566:153741 [4] NCCL INFO Channel 03/0 : 92[4] -> 93[5] via P2P/IPC
+ip-26-0-161-103:96891:97063 [3] NCCL INFO Channel 06/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-161-138:96564:96757 [1] NCCL INFO Channel 11/0 : 33[1] -> 34[2] via P2P/IPC
+ip-26-0-168-52:2722047:2722221 [2] NCCL INFO Channel 13/0 : 122[2] -> 123[3] via P2P/IPC
+ip-26-0-168-52:2722051:2722224 [6] NCCL INFO Connected all rings
+ip-26-0-160-225:124166:124339 [6] NCCL INFO Channel 04/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-168-34:272151:272326 [1] NCCL INFO Channel 04/0 : 113[1] -> 114[2] via P2P/IPC
+ip-26-0-161-78:114513:114688 [1] NCCL INFO Channel 07/0 : 65[1] -> 66[2] via P2P/IPC
+ip-26-0-161-103:96889:97065 [1] NCCL INFO Channel 14/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-161-103:96891:97063 [3] NCCL INFO Channel 07/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-165-202:153564:153737 [2] NCCL INFO Channel 13/0 : 90[2] -> 91[3] via P2P/IPC
+ip-26-0-165-202:153568:153736 [6] NCCL INFO Channel 10/0 : 94[6] -> 95[7] via P2P/IPC
+ip-26-0-161-138:96564:96757 [1] NCCL INFO Channel 12/0 : 33[1] -> 34[2] via P2P/IPC
+ip-26-0-165-202:153566:153741 [4] NCCL INFO Channel 04/0 : 92[4] -> 93[5] via P2P/IPC
+ip-26-0-160-225:124165:124341 [5] NCCL INFO Connected all rings
+ip-26-0-168-34:272153:272328 [3] NCCL INFO Connected all rings
+ip-26-0-161-138:96566:96755 [3] NCCL INFO Connected all rings
+ip-26-0-168-34:272151:272326 [1] NCCL INFO Channel 05/0 : 113[1] -> 114[2] via P2P/IPC
+ip-26-0-161-123:32707:32878 [3] NCCL INFO Channel 01/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-161-103:96889:97065 [1] NCCL INFO Channel 15/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-161-78:114513:114688 [1] NCCL INFO Channel 09/0 : 65[1] -> 66[2] via P2P/IPC
+ip-26-0-165-202:153564:153737 [2] NCCL INFO Channel 14/0 : 90[2] -> 91[3] via P2P/IPC
+ip-26-0-161-103:96891:97063 [3] NCCL INFO Channel 09/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-161-138:96568:96753 [5] NCCL INFO Channel 01/0 : 37[5] -> 38[6] via P2P/IPC
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO Connected all rings
+ip-26-0-165-202:153568:153736 [6] NCCL INFO Channel 11/0 : 94[6] -> 95[7] via P2P/IPC
+ip-26-0-165-202:153566:153741 [4] NCCL INFO Channel 06/0 : 92[4] -> 93[5] via P2P/IPC
+ip-26-0-161-123:32707:32878 [3] NCCL INFO Channel 02/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO Channel 06/0 : 72[0] -> 79[7] via P2P/IPC
+ip-26-0-161-138:96564:96757 [1] NCCL INFO Channel 13/0 : 33[1] -> 34[2] via P2P/IPC
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO Connected all rings
+ip-26-0-161-123:32707:32878 [3] NCCL INFO Channel 03/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-161-103:96888:97066 [0] NCCL INFO Channel 05/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-161-138:96568:96753 [5] NCCL INFO Channel 02/0 : 37[5] -> 38[6] via P2P/IPC
+ip-26-0-161-103:96891:97063 [3] NCCL INFO Channel 10/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-161-138:96570:96759 [7] NCCL INFO Connected all rings
+ip-26-0-161-78:114513:114688 [1] NCCL INFO Channel 11/0 : 65[1] -> 66[2] via P2P/IPC
+ip-26-0-165-202:153564:153737 [2] NCCL INFO Channel 15/0 : 90[2] -> 91[3] via P2P/IPC
+ip-26-0-165-202:153568:153736 [6] NCCL INFO Channel 12/0 : 94[6] -> 95[7] via P2P/IPC
+ip-26-0-161-138:96564:96757 [1] NCCL INFO Channel 14/0 : 33[1] -> 34[2] via P2P/IPC
+ip-26-0-168-52:2722047:2722221 [2] NCCL INFO Channel 14/0 : 122[2] -> 123[3] via P2P/IPC
+ip-26-0-165-202:153566:153741 [4] NCCL INFO Channel 07/0 : 92[4] -> 93[5] via P2P/IPC
+ip-26-0-161-123:32707:32878 [3] NCCL INFO Channel 05/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-161-103:96888:97066 [0] NCCL INFO Channel 06/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-161-103:96891:97063 [3] NCCL INFO Channel 11/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-168-34:272151:272326 [1] NCCL INFO Channel 06/0 : 113[1] -> 114[2] via P2P/IPC
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO Channel 10/0 : 48[0] -> 49[1] via P2P/IPC
+ip-26-0-161-138:96568:96753 [5] NCCL INFO Channel 03/0 : 37[5] -> 38[6] via P2P/IPC
+ip-26-0-161-78:114514:114689 [2] NCCL INFO Connected all rings
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO Channel 01/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-165-202:153568:153736 [6] NCCL INFO Channel 13/0 : 94[6] -> 95[7] via P2P/IPC
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO Channel 07/0 : 72[0] -> 79[7] via P2P/IPC
+ip-26-0-161-78:114513:114688 [1] NCCL INFO Channel 12/0 : 65[1] -> 66[2] via P2P/IPC
+ip-26-0-168-34:272151:272326 [1] NCCL INFO Channel 07/0 : 113[1] -> 114[2] via P2P/IPC
+ip-26-0-165-202:153566:153741 [4] NCCL INFO Channel 09/0 : 92[4] -> 93[5] via P2P/IPC
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO Channel 01/0 : 101[5] -> 102[6] via P2P/IPC
+ip-26-0-161-103:96888:97066 [0] NCCL INFO Channel 07/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-161-103:96891:97063 [3] NCCL INFO Channel 13/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO Channel 02/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-161-142:62887:63055 [6] NCCL INFO Connected all rings
+ip-26-0-160-225:124166:124339 [6] NCCL INFO Channel 05/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-202:153568:153736 [6] NCCL INFO Channel 14/0 : 94[6] -> 95[7] via P2P/IPC
+ip-26-0-161-78:114513:114688 [1] NCCL INFO Channel 13/0 : 65[1] -> 66[2] via P2P/IPC
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO Channel 01/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-168-52:2722047:2722221 [2] NCCL INFO Channel 15/0 : 122[2] -> 123[3] via P2P/IPC
+ip-26-0-161-142:62884:63061 [3] NCCL INFO Channel 01/0 : 43[3] -> 44[4] via P2P/IPC
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO Channel 13/0 : 72[0] -> 79[7] via P2P/IPC
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO Channel 02/0 : 101[5] -> 102[6] via P2P/IPC
+ip-26-0-161-103:96888:97066 [0] NCCL INFO Channel 13/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-165-202:153566:153741 [4] NCCL INFO Channel 10/0 : 92[4] -> 93[5] via P2P/IPC
+ip-26-0-161-123:32709:32882 [5] NCCL INFO Connected all rings
+ip-26-0-161-142:62884:63061 [3] NCCL INFO Channel 02/0 : 43[3] -> 44[4] via P2P/IPC
+ip-26-0-160-225:124166:124339 [6] NCCL INFO Channel 06/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-103:96891:97063 [3] NCCL INFO Channel 14/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO Connected all rings
+ip-26-0-168-30:235649:235827 [1] NCCL INFO Connected all rings
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO Channel 03/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-161-123:32707:32878 [3] NCCL INFO Channel 06/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-161-103:96890:97070 [2] NCCL INFO Channel 01/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO Connected all rings
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO Channel 02/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-165-202:153566:153741 [4] NCCL INFO Channel 11/0 : 92[4] -> 93[5] via P2P/IPC
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO Channel 03/0 : 101[5] -> 102[6] via P2P/IPC
+ip-26-0-160-225:124166:124339 [6] NCCL INFO Channel 09/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-103:96888:97066 [0] NCCL INFO Channel 14/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-161-78:114513:114688 [1] NCCL INFO Channel 14/0 : 65[1] -> 66[2] via P2P/IPC
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO Channel 04/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO Channel 01/0 : 102[6] -> 103[7] via P2P/IPC
+ip-26-0-161-103:96891:97063 [3] NCCL INFO Channel 15/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-168-34:272151:272326 [1] NCCL INFO Channel 09/0 : 113[1] -> 114[2] via P2P/IPC
+ip-26-0-161-103:96890:97070 [2] NCCL INFO Channel 02/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO Channel 03/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-161-123:32707:32878 [3] NCCL INFO Channel 07/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-165-202:153566:153741 [4] NCCL INFO Channel 12/0 : 92[4] -> 93[5] via P2P/IPC
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO Connected all rings
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO Channel 04/0 : 101[5] -> 102[6] via P2P/IPC
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO Channel 11/0 : 48[0] -> 49[1] via P2P/IPC
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO Channel 02/0 : 102[6] -> 103[7] via P2P/IPC
+ip-26-0-160-225:124166:124339 [6] NCCL INFO Channel 10/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO Channel 01/0 : 60[4] -> 61[5] via P2P/IPC
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO Channel 05/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-161-103:96888:97066 [0] NCCL INFO Channel 15/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-161-123:32707:32878 [3] NCCL INFO Channel 09/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-161-78:114513:114688 [1] NCCL INFO Channel 15/0 : 65[1] -> 66[2] via P2P/IPC
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO Channel 04/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO Channel 12/0 : 48[0] -> 49[1] via P2P/IPC
+ip-26-0-161-103:96890:97070 [2] NCCL INFO Channel 04/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-168-34:272151:272326 [1] NCCL INFO Channel 11/0 : 113[1] -> 114[2] via P2P/IPC
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO Channel 05/0 : 101[5] -> 102[6] via P2P/IPC
+ip-26-0-165-202:153566:153741 [4] NCCL INFO Channel 14/0 : 92[4] -> 93[5] via P2P/IPC
+ip-26-0-161-103:96893:97067 [5] NCCL INFO Channel 01/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO Channel 03/0 : 102[6] -> 103[7] via P2P/IPC
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO Channel 14/0 : 72[0] -> 79[7] via P2P/IPC
+ip-26-0-161-103:96892:97064 [4] NCCL INFO Channel 01/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-160-225:124166:124339 [6] NCCL INFO Channel 11/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-202:153566:153741 [4] NCCL INFO Channel 15/0 : 92[4] -> 93[5] via P2P/IPC
+ip-26-0-168-34:272151:272326 [1] NCCL INFO Channel 12/0 : 113[1] -> 114[2] via P2P/IPC
+ip-26-0-161-142:62886:63057 [5] NCCL INFO Connected all rings
+ip-26-0-160-225:124160:124335 [0] NCCL INFO Channel 05/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO Channel 01/0 : 99[3] -> 100[4] via P2P/IPC
+ip-26-0-161-103:96890:97070 [2] NCCL INFO Channel 05/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO Channel 02/0 : 60[4] -> 61[5] via P2P/IPC
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO Channel 07/0 : 101[5] -> 102[6] via P2P/IPC
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO Connected all rings
+ip-26-0-161-123:32707:32878 [3] NCCL INFO Channel 10/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-161-103:96893:97067 [5] NCCL INFO Channel 02/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO Channel 04/0 : 102[6] -> 103[7] via P2P/IPC
+ip-26-0-161-103:96892:97064 [4] NCCL INFO Channel 02/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO Channel 15/0 : 72[0] -> 79[7] via P2P/IPC
+ip-26-0-161-138:96564:96757 [1] NCCL INFO Channel 15/0 : 33[1] -> 34[2] via P2P/IPC
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO Channel 02/0 : 99[3] -> 100[4] via P2P/IPC
+ip-26-0-161-123:32707:32878 [3] NCCL INFO Channel 11/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-168-30:235648:235826 [0] NCCL INFO Channel 07/0 : 104[0] -> 105[1] via P2P/IPC
+ip-26-0-161-103:96890:97070 [2] NCCL INFO Channel 06/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO Channel 01/0 : 74[2] -> 75[3] via P2P/IPC
+ip-26-0-160-225:124166:124339 [6] NCCL INFO Channel 12/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-78:114512:114686 [0] NCCL INFO Channel 05/0 : 64[0] -> 71[7] via P2P/IPC
+ip-26-0-161-103:96893:97067 [5] NCCL INFO Channel 03/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO Channel 09/0 : 101[5] -> 102[6] via P2P/IPC
+ip-26-0-161-103:96892:97064 [4] NCCL INFO Channel 03/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO Channel 05/0 : 102[6] -> 103[7] via P2P/IPC
+ip-26-0-168-34:272151:272326 [1] NCCL INFO Channel 13/0 : 113[1] -> 114[2] via P2P/IPC
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO Channel 03/0 : 60[4] -> 61[5] via P2P/IPC
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO Channel 13/0 : 48[0] -> 49[1] via P2P/IPC
+ip-26-0-161-103:96890:97070 [2] NCCL INFO Channel 07/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO Channel 03/0 : 99[3] -> 100[4] via P2P/IPC
+ip-26-0-165-202:153567:153742 [5] NCCL INFO Channel 05/0 : 85[5] -> 93[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124160:124335 [0] NCCL INFO Channel 06/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-161-103:96893:97067 [5] NCCL INFO Channel 04/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO Channel 06/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-161-142:62884:63061 [3] NCCL INFO Channel 03/0 : 43[3] -> 44[4] via P2P/IPC
+ip-26-0-161-138:96563:96756 [0] NCCL INFO Channel 05/0 : 32[0] -> 39[7] via P2P/IPC
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO Channel 01/0 : 100[4] -> 101[5] via P2P/IPC
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO Channel 10/0 : 101[5] -> 102[6] via P2P/IPC
+ip-26-0-161-103:96892:97064 [4] NCCL INFO Channel 04/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO Channel 02/0 : 74[2] -> 75[3] via P2P/IPC
+ip-26-0-161-123:32707:32878 [3] NCCL INFO Channel 13/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO Channel 06/0 : 102[6] -> 103[7] via P2P/IPC
+ip-26-0-168-34:272151:272326 [1] NCCL INFO Channel 14/0 : 113[1] -> 114[2] via P2P/IPC
+ip-26-0-168-34:272150:272327 [0] NCCL INFO Connected all rings
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO Channel 06/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-168-34:272150:272327 [0] NCCL INFO Channel 02/0 : 112[0] -> 113[1] via P2P/IPC
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO Channel 05/0 : 99[3] -> 100[4] via P2P/IPC
+ip-26-0-161-123:32707:32878 [3] NCCL INFO Channel 14/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-161-103:96890:97070 [2] NCCL INFO Channel 09/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-161-138:96563:96756 [0] NCCL INFO Channel 06/0 : 32[0] -> 39[7] via P2P/IPC
+ip-26-0-161-103:96894:97068 [6] NCCL INFO Channel 09/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO Channel 02/0 : 100[4] -> 101[5] via P2P/IPC
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO Channel 11/0 : 101[5] -> 102[6] via P2P/IPC
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO Channel 09/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-161-103:96893:97067 [5] NCCL INFO Channel 05/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO Channel 09/0 : 102[6] -> 103[7] via P2P/IPC
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO Channel 04/0 : 74[2] -> 75[3] via P2P/IPC
+ip-26-0-161-123:32705:32880 [1] NCCL INFO Connected all rings
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO Channel 06/0 : 99[3] -> 100[4] via P2P/IPC
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO Channel 07/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-168-34:272151:272326 [1] NCCL INFO Channel 15/0 : 113[1] -> 114[2] via P2P/IPC
+ip-26-0-161-123:32707:32878 [3] NCCL INFO Channel 15/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-161-138:96563:96756 [0] NCCL INFO Channel 07/0 : 32[0] -> 39[7] via P2P/IPC
+ip-26-0-161-78:114512:114686 [0] NCCL INFO Channel 06/0 : 64[0] -> 71[7] via P2P/IPC
+ip-26-0-161-138:96565:96754 [2] NCCL INFO Channel 01/0 : 34[2] -> 35[3] via P2P/IPC
+ip-26-0-160-225:124166:124339 [6] NCCL INFO Channel 13/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO Channel 12/0 : 101[5] -> 102[6] via P2P/IPC
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO Channel 05/0 : 74[2] -> 75[3] via P2P/IPC
+ip-26-0-168-34:272152:272331 [2] NCCL INFO Channel 01/0 : 114[2] -> 115[3] via P2P/IPC
+ip-26-0-168-34:272150:272327 [0] NCCL INFO Channel 03/0 : 112[0] -> 113[1] via P2P/IPC
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO Channel 10/0 : 102[6] -> 103[7] via P2P/IPC
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO Channel 01/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO Channel 07/0 : 99[3] -> 100[4] via P2P/IPC
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO Channel 10/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-161-142:62888:63060 [7] NCCL INFO Connected all rings
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO Channel 13/0 : 101[5] -> 102[6] via P2P/IPC
+ip-26-0-160-225:124160:124335 [0] NCCL INFO Channel 07/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO Channel 11/0 : 102[6] -> 103[7] via P2P/IPC
+ip-26-0-161-138:96563:96756 [0] NCCL INFO Channel 13/0 : 32[0] -> 39[7] via P2P/IPC
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO Channel 09/0 : 99[3] -> 100[4] via P2P/IPC
+ip-26-0-160-225:124166:124339 [6] NCCL INFO Channel 14/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-138:96568:96753 [5] NCCL INFO Channel 04/0 : 37[5] -> 38[6] via P2P/IPC
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO Channel 06/0 : 74[2] -> 75[3] via P2P/IPC
+ip-26-0-161-138:96565:96754 [2] NCCL INFO Channel 02/0 : 34[2] -> 35[3] via P2P/IPC
+ip-26-0-161-103:96892:97064 [4] NCCL INFO Channel 06/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO Channel 15/0 : 101[5] -> 102[6] via P2P/IPC
+ip-26-0-161-103:96895:97069 [7] NCCL INFO Connected all rings
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO Channel 12/0 : 102[6] -> 103[7] via P2P/IPC
+ip-26-0-168-34:272152:272331 [2] NCCL INFO Channel 02/0 : 114[2] -> 115[3] via P2P/IPC
+ip-26-0-168-34:272150:272327 [0] NCCL INFO Channel 04/0 : 112[0] -> 113[1] via P2P/IPC
+ip-26-0-160-225:124160:124335 [0] NCCL INFO Channel 13/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO Channel 10/0 : 99[3] -> 100[4] via P2P/IPC
+ip-26-0-161-78:114512:114686 [0] NCCL INFO Channel 07/0 : 64[0] -> 71[7] via P2P/IPC
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO Channel 14/0 : 48[0] -> 49[1] via P2P/IPC
+ip-26-0-161-138:96563:96756 [0] NCCL INFO Channel 14/0 : 32[0] -> 39[7] via P2P/IPC
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO Channel 07/0 : 74[2] -> 75[3] via P2P/IPC
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO Channel 13/0 : 102[6] -> 103[7] via P2P/IPC
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO Channel 09/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO Channel 03/0 : 100[4] -> 101[5] via P2P/IPC
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO Channel 11/0 : 99[3] -> 100[4] via P2P/IPC
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO Channel 15/0 : 48[0] -> 49[1] via P2P/IPC
+ip-26-0-161-78:114517:114687 [5] NCCL INFO Connected all rings
+ip-26-0-161-138:96568:96753 [5] NCCL INFO Channel 05/0 : 37[5] -> 38[6] via P2P/IPC
+ip-26-0-160-225:124160:124335 [0] NCCL INFO Channel 14/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-165-202:153569:153739 [7] NCCL INFO Connected all rings
+ip-26-0-168-34:272152:272331 [2] NCCL INFO Channel 04/0 : 114[2] -> 115[3] via P2P/IPC
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO Channel 09/0 : 74[2] -> 75[3] via P2P/IPC
+ip-26-0-168-34:272150:272327 [0] NCCL INFO Channel 05/0 : 112[0] -> 113[1] via P2P/IPC
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO Channel 14/0 : 102[6] -> 103[7] via P2P/IPC
+ip-26-0-161-138:96565:96754 [2] NCCL INFO Channel 04/0 : 34[2] -> 35[3] via P2P/IPC
+ip-26-0-161-142:62884:63061 [3] NCCL INFO Channel 05/0 : 43[3] -> 44[4] via P2P/IPC
+ip-26-0-168-30:235648:235826 [0] NCCL INFO Channel 10/0 : 104[0] -> 105[1] via P2P/IPC
+ip-26-0-161-123:32708:32879 [4] NCCL INFO Channel 01/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO Channel 13/0 : 99[3] -> 100[4] via P2P/IPC
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO Channel 10/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO Channel 15/0 : 80[0] -> 81[1] via P2P/IPC
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO Channel 04/0 : 100[4] -> 101[5] via P2P/IPC
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO Channel 14/0 : 99[3] -> 100[4] via P2P/IPC
+ip-26-0-160-225:124160:124335 [0] NCCL INFO Channel 15/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-161-78:114512:114686 [0] NCCL INFO Channel 13/0 : 64[0] -> 71[7] via P2P/IPC
+ip-26-0-168-34:272152:272331 [2] NCCL INFO Channel 05/0 : 114[2] -> 115[3] via P2P/IPC
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO Channel 10/0 : 74[2] -> 75[3] via P2P/IPC
+ip-26-0-161-103:96890:97070 [2] NCCL INFO Channel 10/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO Channel 14/0 : 102[6] -> 110[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96893:97067 [5] NCCL INFO Channel 07/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-168-34:272150:272327 [0] NCCL INFO Channel 06/0 : 112[0] -> 113[1] via P2P/IPC
+ip-26-0-161-123:32708:32879 [4] NCCL INFO Channel 02/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-161-142:62884:63061 [3] NCCL INFO Channel 06/0 : 43[3] -> 44[4] via P2P/IPC
+ip-26-0-161-103:96894:97068 [6] NCCL INFO Channel 10/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-168-52:2722051:2722224 [6] NCCL INFO Channel 01/0 : 126[6] -> 127[7] via P2P/IPC
+ip-26-0-161-138:96568:96753 [5] NCCL INFO Channel 07/0 : 37[5] -> 38[6] via P2P/IPC
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO Channel 15/0 : 99[3] -> 100[4] via P2P/IPC
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO Channel 06/0 : 100[4] -> 101[5] via P2P/IPC
+ip-26-0-161-103:96892:97064 [4] NCCL INFO Channel 07/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO Channel 11/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-161-138:96565:96754 [2] NCCL INFO Channel 05/0 : 34[2] -> 35[3] via P2P/IPC
+ip-26-0-168-30:235648:235826 [0] NCCL INFO Channel 11/0 : 104[0] -> 105[1] via P2P/IPC
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO Channel 12/0 : 74[2] -> 75[3] via P2P/IPC
+ip-26-0-161-138:96569:96758 [6] NCCL INFO Channel 01/0 : 38[6] -> 39[7] via P2P/IPC
+ip-26-0-161-123:32708:32879 [4] NCCL INFO Channel 03/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO Channel 01/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-161-123:32709:32882 [5] NCCL INFO Channel 01/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO Channel 07/0 : 100[4] -> 101[5] via P2P/IPC
+ip-26-0-161-103:96890:97070 [2] NCCL INFO Channel 12/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-168-52:2722048:2722220 [3] NCCL INFO Channel 01/0 : 123[3] -> 124[4] via P2P/IPC
+ip-26-0-168-30:235650:235824 [2] NCCL INFO Connected all rings
+ip-26-0-168-34:272152:272331 [2] NCCL INFO Channel 06/0 : 114[2] -> 115[3] via P2P/IPC
+ip-26-0-161-103:96893:97067 [5] NCCL INFO Channel 09/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-168-52:2722051:2722224 [6] NCCL INFO Channel 02/0 : 126[6] -> 127[7] via P2P/IPC
+ip-26-0-161-142:62884:63061 [3] NCCL INFO Channel 07/0 : 43[3] -> 44[4] via P2P/IPC
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO Channel 12/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO Channel 10/0 : 98[2] -> 106[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96894:97068 [6] NCCL INFO Channel 11/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO Channel 09/0 : 100[4] -> 101[5] via P2P/IPC
+ip-26-0-161-103:96892:97064 [4] NCCL INFO Channel 09/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-161-138:96568:96753 [5] NCCL INFO Channel 09/0 : 37[5] -> 38[6] via P2P/IPC
+ip-26-0-161-138:96565:96754 [2] NCCL INFO Channel 06/0 : 34[2] -> 35[3] via P2P/IPC
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO Channel 13/0 : 74[2] -> 75[3] via P2P/IPC
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO Channel 02/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-161-78:114519:114692 [7] NCCL INFO Connected all rings
+ip-26-0-161-123:32708:32879 [4] NCCL INFO Channel 04/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-161-123:32709:32882 [5] NCCL INFO Channel 02/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO Channel 10/0 : 100[4] -> 101[5] via P2P/IPC
+ip-26-0-161-138:96569:96758 [6] NCCL INFO Channel 02/0 : 38[6] -> 39[7] via P2P/IPC
+ip-26-0-168-52:2722048:2722220 [3] NCCL INFO Channel 02/0 : 123[3] -> 124[4] via P2P/IPC
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO Channel 14/0 : 74[2] -> 75[3] via P2P/IPC
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO Channel 14/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-168-52:2722051:2722224 [6] NCCL INFO Channel 03/0 : 126[6] -> 127[7] via P2P/IPC
+ip-26-0-168-30:235648:235826 [0] NCCL INFO Channel 12/0 : 104[0] -> 105[1] via P2P/IPC
+ip-26-0-161-103:96890:97070 [2] NCCL INFO Channel 13/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-161-103:96893:97067 [5] NCCL INFO Channel 10/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO Channel 11/0 : 100[4] -> 101[5] via P2P/IPC
+ip-26-0-161-142:62884:63061 [3] NCCL INFO Channel 09/0 : 43[3] -> 44[4] via P2P/IPC
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO Channel 01/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-161-103:96892:97064 [4] NCCL INFO Channel 10/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-161-123:32709:32882 [5] NCCL INFO Channel 03/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-161-123:32708:32879 [4] NCCL INFO Channel 06/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-161-103:96894:97068 [6] NCCL INFO Channel 12/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO Channel 03/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-168-52:2722048:2722220 [3] NCCL INFO Channel 03/0 : 123[3] -> 124[4] via P2P/IPC
+ip-26-0-161-138:96568:96753 [5] NCCL INFO Channel 10/0 : 37[5] -> 38[6] via P2P/IPC
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO Channel 12/0 : 100[4] -> 101[5] via P2P/IPC
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO Channel 15/0 : 74[2] -> 75[3] via P2P/IPC
+ip-26-0-161-138:96565:96754 [2] NCCL INFO Channel 07/0 : 34[2] -> 35[3] via P2P/IPC
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO Channel 15/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO Channel 04/0 : 60[4] -> 61[5] via P2P/IPC
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO Channel 06/0 : 60[4] -> 61[5] via P2P/IPC
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO Channel 07/0 : 60[4] -> 61[5] via P2P/IPC
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO Channel 01/0 : 61[5] -> 62[6] via P2P/IPC
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO Channel 09/0 : 60[4] -> 61[5] via P2P/IPC
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO Channel 02/0 : 61[5] -> 62[6] via P2P/IPC
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO Channel 10/0 : 60[4] -> 61[5] via P2P/IPC
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO Channel 01/0 : 62[6] -> 63[7] via P2P/IPC
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO Channel 03/0 : 61[5] -> 62[6] via P2P/IPC
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO Channel 11/0 : 60[4] -> 61[5] via P2P/IPC
+ip-26-0-161-138:96569:96758 [6] NCCL INFO Channel 03/0 : 38[6] -> 39[7] via P2P/IPC
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO Channel 14/0 : 100[4] -> 101[5] via P2P/IPC
+ip-26-0-168-52:2722051:2722224 [6] NCCL INFO Channel 04/0 : 126[6] -> 127[7] via P2P/IPC
+ip-26-0-161-78:114512:114686 [0] NCCL INFO Channel 14/0 : 64[0] -> 71[7] via P2P/IPC
+ip-26-0-161-103:96890:97070 [2] NCCL INFO Channel 14/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO Channel 02/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-168-34:272156:272325 [6] NCCL INFO Connected all rings
+ip-26-0-161-103:96893:97067 [5] NCCL INFO Channel 11/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO Channel 02/0 : 62[6] -> 63[7] via P2P/IPC
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO Channel 15/0 : 103[7] -> 111[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO Channel 08/0 : 96[0] -> 104[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272150:272327 [0] NCCL INFO Channel 07/0 : 112[0] -> 113[1] via P2P/IPC
+ip-26-0-161-123:32709:32882 [5] NCCL INFO Channel 04/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO Channel 04/0 : 61[5] -> 62[6] via P2P/IPC
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO Channel 04/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-168-52:2722048:2722220 [3] NCCL INFO Channel 05/0 : 123[3] -> 124[4] via P2P/IPC
+ip-26-0-161-103:96892:97064 [4] NCCL INFO Channel 11/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-161-142:62884:63061 [3] NCCL INFO Channel 10/0 : 43[3] -> 44[4] via P2P/IPC
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO Channel 15/0 : 100[4] -> 101[5] via P2P/IPC
+ip-26-0-161-123:32708:32879 [4] NCCL INFO Channel 07/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-161-103:96894:97068 [6] NCCL INFO Channel 13/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-161-138:96566:96755 [3] NCCL INFO Channel 01/0 : 35[3] -> 36[4] via P2P/IPC
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO Channel 12/0 : 60[4] -> 61[5] via P2P/IPC
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO Channel 04/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-168-52:2722051:2722224 [6] NCCL INFO Channel 05/0 : 126[6] -> 127[7] via P2P/IPC
+ip-26-0-168-30:235648:235826 [0] NCCL INFO Channel 13/0 : 104[0] -> 105[1] via P2P/IPC
+ip-26-0-161-123:32709:32882 [5] NCCL INFO Channel 05/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO Channel 03/0 : 62[6] -> 63[7] via P2P/IPC
+ip-26-0-161-123:32708:32879 [4] NCCL INFO Channel 09/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-161-138:96568:96753 [5] NCCL INFO Channel 11/0 : 37[5] -> 38[6] via P2P/IPC
+ip-26-0-168-34:272152:272331 [2] NCCL INFO Channel 07/0 : 114[2] -> 115[3] via P2P/IPC
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO Channel 05/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-161-138:96565:96754 [2] NCCL INFO Channel 09/0 : 34[2] -> 35[3] via P2P/IPC
+ip-26-0-168-52:2722046:2722223 [1] NCCL INFO Connected all rings
+ip-26-0-168-52:2722048:2722220 [3] NCCL INFO Channel 06/0 : 123[3] -> 124[4] via P2P/IPC
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO Channel 05/0 : 61[5] -> 62[6] via P2P/IPC
+ip-26-0-161-103:96890:97070 [2] NCCL INFO Channel 15/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-161-103:96893:97067 [5] NCCL INFO Channel 12/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO Connected all rings
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO Channel 11/0 : 99[3] -> 107[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO Channel 09/0 : 65[1] -> 73[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-52:2722051:2722224 [6] NCCL INFO Channel 06/0 : 126[6] -> 127[7] via P2P/IPC
+ip-26-0-161-123:32710:32883 [6] NCCL INFO Connected all rings
+ip-26-0-168-34:272150:272327 [0] NCCL INFO Channel 10/0 : 112[0] -> 113[1] via P2P/IPC
+ip-26-0-161-138:96569:96758 [6] NCCL INFO Channel 04/0 : 38[6] -> 39[7] via P2P/IPC
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO Channel 13/0 : 101[5] -> 109[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO Connected all rings
+ip-26-0-161-103:96892:97064 [4] NCCL INFO Channel 12/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO Channel 01/0 : 73[1] -> 81[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32709:32882 [5] NCCL INFO Channel 07/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO Channel 12/0 : 100[4] -> 108[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62884:63061 [3] NCCL INFO Channel 11/0 : 43[3] -> 44[4] via P2P/IPC
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO Channel 02/0 : 56[0] -> 57[1] via P2P/IPC
+ip-26-0-161-78:114512:114686 [0] NCCL INFO Channel 15/0 : 64[0] -> 71[7] via P2P/IPC
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO Channel 09/0 : 73[1] -> 81[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96894:97068 [6] NCCL INFO Channel 14/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-161-123:32708:32879 [4] NCCL INFO Channel 10/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO Channel 05/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO Channel 03/0 : 49[1] -> 50[2] via P2P/IPC
+ip-26-0-161-138:96566:96755 [3] NCCL INFO Channel 02/0 : 35[3] -> 36[4] via P2P/IPC
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO Channel 14/0 : 60[4] -> 61[5] via P2P/IPC
+ip-26-0-168-30:235653:235825 [5] NCCL INFO Connected all rings
+ip-26-0-161-138:96567:96760 [4] NCCL INFO Connected all rings
+ip-26-0-161-138:96565:96754 [2] NCCL INFO Channel 10/0 : 34[2] -> 35[3] via P2P/IPC
+ip-26-0-161-138:96568:96753 [5] NCCL INFO Channel 12/0 : 37[5] -> 38[6] via P2P/IPC
+ip-26-0-168-30:235648:235826 [0] NCCL INFO Channel 14/0 : 104[0] -> 105[1] via P2P/IPC
+ip-26-0-168-34:272152:272331 [2] NCCL INFO Channel 09/0 : 114[2] -> 115[3] via P2P/IPC
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO Channel 07/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-161-123:32709:32882 [5] NCCL INFO Channel 09/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-168-52:2722048:2722220 [3] NCCL INFO Channel 07/0 : 123[3] -> 124[4] via P2P/IPC
+ip-26-0-161-123:32708:32879 [4] NCCL INFO Channel 11/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-161-138:96569:96758 [6] NCCL INFO Channel 05/0 : 38[6] -> 39[7] via P2P/IPC
+ip-26-0-161-142:62884:63061 [3] NCCL INFO Channel 13/0 : 43[3] -> 44[4] via P2P/IPC
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO Channel 01/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-161-123:32705:32880 [1] NCCL INFO Channel 01/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO Connected all rings
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO Channel 02/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO Channel 04/0 : 62[6] -> 63[7] via P2P/IPC
+ip-26-0-160-225:124165:124341 [5] NCCL INFO Channel 01/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO Channel 11/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-168-52:2722051:2722224 [6] NCCL INFO Channel 09/0 : 126[6] -> 127[7] via P2P/IPC
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO Channel 07/0 : 61[5] -> 62[6] via P2P/IPC
+ip-26-0-161-138:96566:96755 [3] NCCL INFO Channel 03/0 : 35[3] -> 36[4] via P2P/IPC
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO Channel 06/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-168-34:272152:272331 [2] NCCL INFO Channel 10/0 : 114[2] -> 115[3] via P2P/IPC
+ip-26-0-161-123:32709:32882 [5] NCCL INFO Channel 10/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-168-34:272150:272327 [0] NCCL INFO Channel 11/0 : 112[0] -> 113[1] via P2P/IPC
+ip-26-0-161-138:96565:96754 [2] NCCL INFO Channel 12/0 : 34[2] -> 35[3] via P2P/IPC
+ip-26-0-161-123:32708:32879 [4] NCCL INFO Channel 12/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-168-52:2722048:2722220 [3] NCCL INFO Channel 09/0 : 123[3] -> 124[4] via P2P/IPC
+ip-26-0-161-138:96568:96753 [5] NCCL INFO Channel 13/0 : 37[5] -> 38[6] via P2P/IPC
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO Channel 03/0 : 56[0] -> 57[1] via P2P/IPC
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO Channel 09/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO Channel 15/0 : 60[4] -> 61[5] via P2P/IPC
+ip-26-0-161-123:32705:32880 [1] NCCL INFO Channel 03/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-168-52:2722051:2722224 [6] NCCL INFO Channel 10/0 : 126[6] -> 127[7] via P2P/IPC
+ip-26-0-161-138:96569:96758 [6] NCCL INFO Channel 06/0 : 38[6] -> 39[7] via P2P/IPC
+ip-26-0-168-34:272157:272324 [7] NCCL INFO Connected all rings
+ip-26-0-160-225:124165:124341 [5] NCCL INFO Channel 02/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-142:62884:63061 [3] NCCL INFO Channel 14/0 : 43[3] -> 44[4] via P2P/IPC
+ip-26-0-161-103:96893:97067 [5] NCCL INFO Channel 13/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-161-138:96563:96756 [0] NCCL INFO Channel 15/0 : 32[0] -> 39[7] via P2P/IPC
+ip-26-0-168-34:272153:272328 [3] NCCL INFO Channel 01/0 : 115[3] -> 116[4] via P2P/IPC
+ip-26-0-168-34:272152:272331 [2] NCCL INFO Channel 12/0 : 114[2] -> 115[3] via P2P/IPC
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO Channel 03/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-161-123:32709:32882 [5] NCCL INFO Channel 11/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-161-138:96566:96755 [3] NCCL INFO Channel 05/0 : 35[3] -> 36[4] via P2P/IPC
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO Channel 03/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-161-123:32708:32879 [4] NCCL INFO Channel 14/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-168-30:235648:235826 [0] NCCL INFO Channel 15/0 : 104[0] -> 105[1] via P2P/IPC
+ip-26-0-161-103:96892:97064 [4] NCCL INFO Channel 14/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-168-34:272150:272327 [0] NCCL INFO Channel 12/0 : 112[0] -> 113[1] via P2P/IPC
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO Channel 12/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO Channel 09/0 : 61[5] -> 62[6] via P2P/IPC
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO Channel 05/0 : 62[6] -> 63[7] via P2P/IPC
+ip-26-0-168-52:2722048:2722220 [3] NCCL INFO Channel 10/0 : 123[3] -> 124[4] via P2P/IPC
+ip-26-0-161-103:96889:97065 [1] NCCL INFO Channel 01/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO Channel 07/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-161-123:32705:32880 [1] NCCL INFO Channel 04/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-161-138:96568:96753 [5] NCCL INFO Channel 15/0 : 37[5] -> 38[6] via P2P/IPC
+ip-26-0-161-103:96889:97065 [1] NCCL INFO Channel 09/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153562:153735 [0] NCCL INFO Connected all rings
+ip-26-0-161-103:96890:97070 [2] NCCL INFO Channel 02/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153562:153735 [0] NCCL INFO Channel 02/0 : 88[0] -> 89[1] via P2P/IPC
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO Channel 10/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-161-103:96889:97065 [1] NCCL INFO Channel 01/0 : 17[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96890:97070 [2] NCCL INFO Channel 10/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-52:2722051:2722224 [6] NCCL INFO Channel 11/0 : 126[6] -> 127[7] via P2P/IPC
+ip-26-0-161-138:96565:96754 [2] NCCL INFO Channel 13/0 : 34[2] -> 35[3] via P2P/IPC
+ip-26-0-161-103:96893:97067 [5] NCCL INFO Channel 15/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-161-103:96890:97070 [2] NCCL INFO Channel 02/0 : 18[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32709:32882 [5] NCCL INFO Channel 12/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-161-103:96892:97064 [4] NCCL INFO Channel 15/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO Channel 04/0 : 56[0] -> 57[1] via P2P/IPC
+ip-26-0-161-138:96569:96758 [6] NCCL INFO Channel 09/0 : 38[6] -> 39[7] via P2P/IPC
+ip-26-0-161-123:32708:32879 [4] NCCL INFO Channel 15/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-161-142:62884:63061 [3] NCCL INFO Channel 15/0 : 43[3] -> 44[4] via P2P/IPC
+ip-26-0-161-123:32705:32880 [1] NCCL INFO Channel 05/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-168-52:2722048:2722220 [3] NCCL INFO Channel 11/0 : 123[3] -> 124[4] via P2P/IPC
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO Channel 04/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-168-34:272153:272328 [3] NCCL INFO Channel 02/0 : 115[3] -> 116[4] via P2P/IPC
+ip-26-0-168-52:2722051:2722224 [6] NCCL INFO Channel 12/0 : 126[6] -> 127[7] via P2P/IPC
+ip-26-0-161-138:96566:96755 [3] NCCL INFO Channel 06/0 : 35[3] -> 36[4] via P2P/IPC
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO Channel 04/0 : 49[1] -> 50[2] via P2P/IPC
+ip-26-0-165-202:153568:153736 [6] NCCL INFO Channel 06/0 : 86[6] -> 94[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272152:272331 [2] NCCL INFO Channel 13/0 : 114[2] -> 115[3] via P2P/IPC
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO Channel 05/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO Channel 10/0 : 61[5] -> 62[6] via P2P/IPC
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO Channel 06/0 : 62[6] -> 63[7] via P2P/IPC
+ip-26-0-165-202:153562:153735 [0] NCCL INFO Channel 03/0 : 88[0] -> 89[1] via P2P/IPC
+ip-26-0-168-34:272150:272327 [0] NCCL INFO Channel 13/0 : 112[0] -> 113[1] via P2P/IPC
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO Channel 13/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO Channel 09/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO Channel 05/0 : 56[0] -> 57[1] via P2P/IPC
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO Channel 11/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-161-123:32709:32882 [5] NCCL INFO Channel 13/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-161-138:96565:96754 [2] NCCL INFO Channel 14/0 : 34[2] -> 35[3] via P2P/IPC
+ip-26-0-161-123:32705:32880 [1] NCCL INFO Channel 06/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-165-202:153562:153735 [0] NCCL INFO Channel 04/0 : 88[0] -> 89[1] via P2P/IPC
+ip-26-0-161-103:96891:97063 [3] NCCL INFO Channel 03/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96893:97067 [5] NCCL INFO Channel 05/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96892:97064 [4] NCCL INFO Channel 04/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96569:96758 [6] NCCL INFO Channel 10/0 : 38[6] -> 39[7] via P2P/IPC
+ip-26-0-161-103:96891:97063 [3] NCCL INFO Channel 11/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96893:97067 [5] NCCL INFO Channel 13/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-52:2722048:2722220 [3] NCCL INFO Channel 13/0 : 123[3] -> 124[4] via P2P/IPC
+ip-26-0-161-123:32707:32878 [3] NCCL INFO Channel 03/0 : 19[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96566:96755 [3] NCCL INFO Channel 07/0 : 35[3] -> 36[4] via P2P/IPC
+ip-26-0-161-103:96892:97064 [4] NCCL INFO Channel 12/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62887:63055 [6] NCCL INFO Channel 01/0 : 46[6] -> 47[7] via P2P/IPC
+ip-26-0-160-225:124165:124341 [5] NCCL INFO Channel 03/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO Channel 11/0 : 61[5] -> 62[6] via P2P/IPC
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO Channel 09/0 : 62[6] -> 63[7] via P2P/IPC
+ip-26-0-168-52:2722051:2722224 [6] NCCL INFO Channel 13/0 : 126[6] -> 127[7] via P2P/IPC
+ip-26-0-161-103:96893:97067 [5] NCCL INFO Channel 05/0 : 21[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96891:97063 [3] NCCL INFO Channel 03/0 : 19[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96892:97064 [4] NCCL INFO Channel 04/0 : 20[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32709:32882 [5] NCCL INFO Channel 15/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-161-78:114514:114689 [2] NCCL INFO Channel 01/0 : 66[2] -> 67[3] via P2P/IPC
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO Channel 05/0 : 49[1] -> 50[2] via P2P/IPC
+ip-26-0-161-142:62885:63054 [4] NCCL INFO Channel 01/0 : 44[4] -> 45[5] via P2P/IPC
+ip-26-0-168-34:272153:272328 [3] NCCL INFO Channel 03/0 : 115[3] -> 116[4] via P2P/IPC
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO Channel 05/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-161-123:32705:32880 [1] NCCL INFO Channel 07/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-161-138:96565:96754 [2] NCCL INFO Channel 15/0 : 34[2] -> 35[3] via P2P/IPC
+ip-26-0-168-34:272152:272331 [2] NCCL INFO Channel 14/0 : 114[2] -> 115[3] via P2P/IPC
+ip-26-0-168-34:272150:272327 [0] NCCL INFO Channel 14/0 : 112[0] -> 113[1] via P2P/IPC
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO Channel 06/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO Channel 06/0 : 56[0] -> 57[1] via P2P/IPC
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO Channel 14/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-168-52:2722048:2722220 [3] NCCL INFO Channel 14/0 : 123[3] -> 124[4] via P2P/IPC
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO Channel 10/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-161-138:96569:96758 [6] NCCL INFO Channel 11/0 : 38[6] -> 39[7] via P2P/IPC
+ip-26-0-168-52:2722051:2722224 [6] NCCL INFO Channel 14/0 : 126[6] -> 127[7] via P2P/IPC
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO Channel 12/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-161-123:32707:32878 [3] NCCL INFO Channel 11/0 : 11[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96566:96755 [3] NCCL INFO Channel 09/0 : 35[3] -> 36[4] via P2P/IPC
+ip-26-0-161-142:62887:63055 [6] NCCL INFO Channel 02/0 : 46[6] -> 47[7] via P2P/IPC
+ip-26-0-161-123:32705:32880 [1] NCCL INFO Channel 09/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-161-123:32707:32878 [3] NCCL INFO Channel 11/0 : 27[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO Channel 06/0 : 49[1] -> 50[2] via P2P/IPC
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO Channel 10/0 : 62[6] -> 63[7] via P2P/IPC
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO Channel 12/0 : 61[5] -> 62[6] via P2P/IPC
+ip-26-0-161-123:32708:32879 [4] NCCL INFO Channel 04/0 : 20[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62885:63054 [4] NCCL INFO Channel 02/0 : 44[4] -> 45[5] via P2P/IPC
+ip-26-0-161-103:96894:97068 [6] NCCL INFO Channel 06/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO Channel 07/0 : 56[0] -> 57[1] via P2P/IPC
+ip-26-0-161-78:114514:114689 [2] NCCL INFO Channel 02/0 : 66[2] -> 67[3] via P2P/IPC
+ip-26-0-161-103:96894:97068 [6] NCCL INFO Channel 14/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272153:272328 [3] NCCL INFO Channel 05/0 : 115[3] -> 116[4] via P2P/IPC
+ip-26-0-161-142:62886:63057 [5] NCCL INFO Channel 01/0 : 45[5] -> 46[6] via P2P/IPC
+ip-26-0-168-52:2722048:2722220 [3] NCCL INFO Channel 15/0 : 123[3] -> 124[4] via P2P/IPC
+ip-26-0-161-103:96894:97068 [6] NCCL INFO Channel 06/0 : 22[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO Channel 06/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-161-123:32708:32879 [4] NCCL INFO Channel 12/0 : 12[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272150:272327 [0] NCCL INFO Channel 15/0 : 112[0] -> 113[1] via P2P/IPC
+ip-26-0-168-34:272152:272331 [2] NCCL INFO Channel 15/0 : 114[2] -> 115[3] via P2P/IPC
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO Channel 07/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-161-123:32705:32880 [1] NCCL INFO Channel 11/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO Connected all rings
+ip-26-0-161-138:96569:96758 [6] NCCL INFO Channel 12/0 : 38[6] -> 39[7] via P2P/IPC
+ip-26-0-161-142:62882:63056 [1] NCCL INFO Channel 01/0 : 41[1] -> 42[2] via P2P/IPC
+ip-26-0-161-123:32708:32879 [4] NCCL INFO Channel 12/0 : 28[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO Channel 13/0 : 61[5] -> 62[6] via P2P/IPC
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO Channel 11/0 : 62[6] -> 63[7] via P2P/IPC
+ip-26-0-161-138:96566:96755 [3] NCCL INFO Channel 10/0 : 35[3] -> 36[4] via P2P/IPC
+ip-26-0-161-138:96564:96757 [1] NCCL INFO Channel 09/0 : 33[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62887:63055 [6] NCCL INFO Channel 03/0 : 46[6] -> 47[7] via P2P/IPC
+ip-26-0-160-225:124164:124336 [4] NCCL INFO Connected all rings
+ip-26-0-161-123:32705:32880 [1] NCCL INFO Channel 12/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO Channel 13/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-161-138:96569:96758 [6] NCCL INFO Channel 13/0 : 38[6] -> 39[7] via P2P/IPC
+ip-26-0-161-142:62885:63054 [4] NCCL INFO Channel 03/0 : 44[4] -> 45[5] via P2P/IPC
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO Channel 10/0 : 56[0] -> 57[1] via P2P/IPC
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO Channel 07/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-161-138:96566:96755 [3] NCCL INFO Channel 11/0 : 35[3] -> 36[4] via P2P/IPC
+ip-26-0-161-142:62886:63057 [5] NCCL INFO Channel 02/0 : 45[5] -> 46[6] via P2P/IPC
+ip-26-0-160-225:124165:124341 [5] NCCL INFO Channel 04/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-165-202:153562:153735 [0] NCCL INFO Channel 05/0 : 88[0] -> 89[1] via P2P/IPC
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO Channel 09/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-161-123:32705:32880 [1] NCCL INFO Channel 13/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO Channel 07/0 : 49[1] -> 50[2] via P2P/IPC
+ip-26-0-161-78:114514:114689 [2] NCCL INFO Channel 04/0 : 66[2] -> 67[3] via P2P/IPC
+ip-26-0-168-34:272153:272328 [3] NCCL INFO Channel 06/0 : 115[3] -> 116[4] via P2P/IPC
+ip-26-0-161-142:62882:63056 [1] NCCL INFO Channel 03/0 : 41[1] -> 42[2] via P2P/IPC
+ip-26-0-161-138:96569:96758 [6] NCCL INFO Channel 14/0 : 38[6] -> 39[7] via P2P/IPC
+ip-26-0-161-123:32710:32883 [6] NCCL INFO Channel 01/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO Connected all rings
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO Channel 01/0 : 81[1] -> 82[2] via P2P/IPC
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO Channel 09/0 : 49[1] -> 50[2] via P2P/IPC
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO Channel 15/0 : 61[5] -> 62[6] via P2P/IPC
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO Channel 12/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-161-142:62887:63055 [6] NCCL INFO Channel 04/0 : 46[6] -> 47[7] via P2P/IPC
+ip-26-0-161-123:32705:32880 [1] NCCL INFO Channel 14/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO Channel 12/0 : 62[6] -> 63[7] via P2P/IPC
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO Channel 15/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO Channel 11/0 : 56[0] -> 57[1] via P2P/IPC
+ip-26-0-161-138:96566:96755 [3] NCCL INFO Channel 13/0 : 35[3] -> 36[4] via P2P/IPC
+ip-26-0-161-123:32710:32883 [6] NCCL INFO Channel 02/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-161-142:62885:63054 [4] NCCL INFO Channel 04/0 : 44[4] -> 45[5] via P2P/IPC
+ip-26-0-165-202:153562:153735 [0] NCCL INFO Channel 06/0 : 88[0] -> 89[1] via P2P/IPC
+ip-26-0-168-34:272150:272327 [0] NCCL INFO Channel 05/0 : 112[0] -> 119[7] via P2P/IPC
+ip-26-0-160-225:124165:124341 [5] NCCL INFO Channel 05/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-142:62886:63057 [5] NCCL INFO Channel 03/0 : 45[5] -> 46[6] via P2P/IPC
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO Channel 09/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-161-123:32705:32880 [1] NCCL INFO Channel 15/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-168-34:272151:272326 [1] NCCL INFO Channel 01/0 : 105[1] -> 113[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235649:235827 [1] NCCL INFO Channel 01/0 : 105[1] -> 106[2] via P2P/IPC
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO Channel 10/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-168-34:272153:272328 [3] NCCL INFO Channel 07/0 : 115[3] -> 116[4] via P2P/IPC
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO Channel 03/0 : 81[1] -> 82[2] via P2P/IPC
+ip-26-0-161-138:96566:96755 [3] NCCL INFO Channel 14/0 : 35[3] -> 36[4] via P2P/IPC
+ip-26-0-161-142:62882:63056 [1] NCCL INFO Channel 04/0 : 41[1] -> 42[2] via P2P/IPC
+ip-26-0-161-123:32710:32883 [6] NCCL INFO Channel 03/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-168-34:272151:272326 [1] NCCL INFO Channel 09/0 : 105[1] -> 113[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114514:114689 [2] NCCL INFO Channel 05/0 : 66[2] -> 67[3] via P2P/IPC
+ip-26-0-161-103:96895:97069 [7] NCCL INFO Channel 07/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272151:272326 [1] NCCL INFO Channel 01/0 : 113[1] -> 121[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO Channel 13/0 : 62[6] -> 63[7] via P2P/IPC
+ip-26-0-161-142:62887:63055 [6] NCCL INFO Channel 05/0 : 46[6] -> 47[7] via P2P/IPC
+ip-26-0-161-103:96888:97066 [0] NCCL INFO Channel 00/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96895:97069 [7] NCCL INFO Channel 15/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96569:96758 [6] NCCL INFO Channel 14/0 : 38[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96895:97069 [7] NCCL INFO Channel 07/0 : 23[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO Channel 04/0 : 81[1] -> 82[2] via P2P/IPC
+ip-26-0-161-142:62885:63054 [4] NCCL INFO Channel 06/0 : 44[4] -> 45[5] via P2P/IPC
+ip-26-0-161-123:32710:32883 [6] NCCL INFO Channel 04/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO Channel 12/0 : 56[0] -> 57[1] via P2P/IPC
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO Channel 13/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-161-103:96888:97066 [0] NCCL INFO Channel 08/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272150:272327 [0] NCCL INFO Channel 06/0 : 112[0] -> 119[7] via P2P/IPC
+ip-26-0-161-103:96888:97066 [0] NCCL INFO Channel 00/0 : 16[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62886:63057 [5] NCCL INFO Channel 04/0 : 45[5] -> 46[6] via P2P/IPC
+ip-26-0-161-138:96566:96755 [3] NCCL INFO Channel 15/0 : 35[3] -> 36[4] via P2P/IPC
+ip-26-0-165-202:153562:153735 [0] NCCL INFO Channel 07/0 : 88[0] -> 89[1] via P2P/IPC
+ip-26-0-161-123:32705:32880 [1] NCCL INFO Channel 01/0 : 17[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32710:32883 [6] NCCL INFO Channel 05/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-160-225:124165:124341 [5] NCCL INFO Channel 07/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-168-34:272153:272328 [3] NCCL INFO Channel 09/0 : 115[3] -> 116[4] via P2P/IPC
+ip-26-0-160-225:124167:124338 [7] NCCL INFO Channel 15/0 : 7[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62882:63056 [1] NCCL INFO Channel 05/0 : 41[1] -> 42[2] via P2P/IPC
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO Channel 11/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO Channel 11/0 : 49[1] -> 50[2] via P2P/IPC
+ip-26-0-161-142:62887:63055 [6] NCCL INFO Channel 06/0 : 46[6] -> 47[7] via P2P/IPC
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO Channel 11/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO Channel 14/0 : 62[6] -> 63[7] via P2P/IPC
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO Channel 14/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO Channel 13/0 : 56[0] -> 57[1] via P2P/IPC
+ip-26-0-161-123:32710:32883 [6] NCCL INFO Channel 06/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO Channel 13/0 : 5[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235649:235827 [1] NCCL INFO Channel 03/0 : 105[1] -> 106[2] via P2P/IPC
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO Channel 06/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153562:153735 [0] NCCL INFO Channel 10/0 : 88[0] -> 89[1] via P2P/IPC
+ip-26-0-161-123:32705:32880 [1] NCCL INFO Channel 09/0 : 9[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62885:63054 [4] NCCL INFO Channel 07/0 : 44[4] -> 45[5] via P2P/IPC
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO Channel 05/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO Channel 14/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO Channel 14/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-161-123:32705:32880 [1] NCCL INFO Channel 09/0 : 25[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114514:114689 [2] NCCL INFO Channel 06/0 : 66[2] -> 67[3] via P2P/IPC
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO Channel 13/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62886:63057 [5] NCCL INFO Channel 05/0 : 45[5] -> 46[6] via P2P/IPC
+ip-26-0-161-123:32710:32883 [6] NCCL INFO Channel 09/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-161-138:96565:96754 [2] NCCL INFO Channel 10/0 : 34[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:124165:124341 [5] NCCL INFO Channel 09/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO Channel 12/0 : 49[1] -> 50[2] via P2P/IPC
+ip-26-0-165-202:153562:153735 [0] NCCL INFO Channel 11/0 : 88[0] -> 89[1] via P2P/IPC
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO Channel 14/0 : 56[0] -> 57[1] via P2P/IPC
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO Channel 12/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-165-202:153563:153740 [1] NCCL INFO Connected all rings
+ip-26-0-161-142:62882:63056 [1] NCCL INFO Channel 06/0 : 41[1] -> 42[2] via P2P/IPC
+ip-26-0-161-142:62887:63055 [6] NCCL INFO Channel 09/0 : 46[6] -> 47[7] via P2P/IPC
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO Channel 13/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-161-123:32710:32883 [6] NCCL INFO Channel 10/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-168-34:272150:272327 [0] NCCL INFO Channel 07/0 : 112[0] -> 119[7] via P2P/IPC
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO Channel 05/0 : 53[5] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO Channel 06/0 : 54[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272153:272328 [3] NCCL INFO Channel 10/0 : 115[3] -> 116[4] via P2P/IPC
+ip-26-0-161-138:96567:96760 [4] NCCL INFO Channel 01/0 : 36[4] -> 37[5] via P2P/IPC
+ip-26-0-161-142:62885:63054 [4] NCCL INFO Channel 09/0 : 44[4] -> 45[5] via P2P/IPC
+ip-26-0-161-123:32704:32881 [0] NCCL INFO Channel 05/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-168-52:2722046:2722223 [1] NCCL INFO Channel 01/0 : 121[1] -> 122[2] via P2P/IPC
+ip-26-0-161-78:114514:114689 [2] NCCL INFO Channel 07/0 : 66[2] -> 67[3] via P2P/IPC
+ip-26-0-165-202:153562:153735 [0] NCCL INFO Channel 12/0 : 88[0] -> 89[1] via P2P/IPC
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO Channel 15/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO Channel 13/0 : 49[1] -> 50[2] via P2P/IPC
+ip-26-0-161-123:32710:32883 [6] NCCL INFO Channel 11/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-161-142:62886:63057 [5] NCCL INFO Channel 07/0 : 45[5] -> 46[6] via P2P/IPC
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO Channel 05/0 : 81[1] -> 82[2] via P2P/IPC
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO Connected all rings
+ip-26-0-160-225:124165:124341 [5] NCCL INFO Channel 10/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO Channel 13/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-161-138:96567:96760 [4] NCCL INFO Channel 02/0 : 36[4] -> 37[5] via P2P/IPC
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO Channel 15/0 : 56[0] -> 57[1] via P2P/IPC
+ip-26-0-165-202:153562:153735 [0] NCCL INFO Channel 13/0 : 88[0] -> 89[1] via P2P/IPC
+ip-26-0-161-142:62882:63056 [1] NCCL INFO Channel 07/0 : 41[1] -> 42[2] via P2P/IPC
+ip-26-0-161-142:62887:63055 [6] NCCL INFO Channel 10/0 : 46[6] -> 47[7] via P2P/IPC
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO Channel 14/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-168-52:2722046:2722223 [1] NCCL INFO Channel 03/0 : 121[1] -> 122[2] via P2P/IPC
+ip-26-0-160-225:124165:124341 [5] NCCL INFO Channel 11/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-123:32704:32881 [0] NCCL INFO Channel 06/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-161-123:32710:32883 [6] NCCL INFO Channel 12/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-161-138:96567:96760 [4] NCCL INFO Channel 03/0 : 36[4] -> 37[5] via P2P/IPC
+ip-26-0-161-142:62885:63054 [4] NCCL INFO Channel 10/0 : 44[4] -> 45[5] via P2P/IPC
+ip-26-0-161-78:114518:114693 [6] NCCL INFO Channel 01/0 : 70[6] -> 71[7] via P2P/IPC
+ip-26-0-161-142:62886:63057 [5] NCCL INFO Channel 09/0 : 45[5] -> 46[6] via P2P/IPC
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO Connected all rings
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO Channel 14/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-168-52:2722046:2722223 [1] NCCL INFO Channel 04/0 : 121[1] -> 122[2] via P2P/IPC
+ip-26-0-161-78:114514:114689 [2] NCCL INFO Channel 09/0 : 66[2] -> 67[3] via P2P/IPC
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO Channel 15/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-161-123:32710:32883 [6] NCCL INFO Channel 13/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-161-123:32704:32881 [0] NCCL INFO Channel 07/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-161-142:62882:63056 [1] NCCL INFO Channel 09/0 : 41[1] -> 42[2] via P2P/IPC
+ip-26-0-161-138:96567:96760 [4] NCCL INFO Channel 04/0 : 36[4] -> 37[5] via P2P/IPC
+ip-26-0-161-142:62887:63055 [6] NCCL INFO Channel 11/0 : 46[6] -> 47[7] via P2P/IPC
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO Channel 14/0 : 49[1] -> 50[2] via P2P/IPC
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO Channel 15/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-161-138:96563:96756 [0] NCCL INFO Channel 08/0 : 32[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62885:63054 [4] NCCL INFO Channel 11/0 : 44[4] -> 45[5] via P2P/IPC
+ip-26-0-161-123:32710:32883 [6] NCCL INFO Channel 14/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-161-78:114517:114687 [5] NCCL INFO Channel 01/0 : 69[5] -> 70[6] via P2P/IPC
+ip-26-0-161-138:96570:96759 [7] NCCL INFO Channel 15/0 : 39[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62886:63057 [5] NCCL INFO Channel 10/0 : 45[5] -> 46[6] via P2P/IPC
+ip-26-0-161-138:96567:96760 [4] NCCL INFO Channel 06/0 : 36[4] -> 37[5] via P2P/IPC
+ip-26-0-161-123:32704:32881 [0] NCCL INFO Channel 13/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-161-78:114518:114693 [6] NCCL INFO Channel 02/0 : 70[6] -> 71[7] via P2P/IPC
+ip-26-0-168-52:2722046:2722223 [1] NCCL INFO Channel 05/0 : 121[1] -> 122[2] via P2P/IPC
+ip-26-0-168-34:272155:272329 [5] NCCL INFO Connected all rings
+ip-26-0-161-123:32706:32877 [2] NCCL INFO Channel 02/0 : 18[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114514:114689 [2] NCCL INFO Channel 10/0 : 66[2] -> 67[3] via P2P/IPC
+ip-26-0-160-225:124165:124341 [5] NCCL INFO Channel 12/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-168-30:235655:235828 [7] NCCL INFO Connected all rings
+ip-26-0-161-138:96567:96760 [4] NCCL INFO Channel 07/0 : 36[4] -> 37[5] via P2P/IPC
+ip-26-0-161-142:62882:63056 [1] NCCL INFO Channel 11/0 : 41[1] -> 42[2] via P2P/IPC
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO Channel 05/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-161-123:32704:32881 [0] NCCL INFO Channel 14/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO Channel 12/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32706:32877 [2] NCCL INFO Channel 10/0 : 10[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96567:96760 [4] NCCL INFO Channel 09/0 : 36[4] -> 37[5] via P2P/IPC
+ip-26-0-161-142:62887:63055 [6] NCCL INFO Channel 12/0 : 46[6] -> 47[7] via P2P/IPC
+ip-26-0-168-52:2722046:2722223 [1] NCCL INFO Channel 06/0 : 121[1] -> 122[2] via P2P/IPC
+ip-26-0-168-30:235649:235827 [1] NCCL INFO Channel 04/0 : 105[1] -> 106[2] via P2P/IPC
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO Channel 11/0 : 3[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO Channel 01/0 : 76[4] -> 77[5] via P2P/IPC
+ip-26-0-161-123:32706:32877 [2] NCCL INFO Channel 10/0 : 26[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153562:153735 [0] NCCL INFO Channel 14/0 : 88[0] -> 89[1] via P2P/IPC
+ip-26-0-161-142:62885:63054 [4] NCCL INFO Channel 12/0 : 44[4] -> 45[5] via P2P/IPC
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO Channel 03/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO Channel 04/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO Channel 15/0 : 49[1] -> 50[2] via P2P/IPC
+ip-26-0-161-123:32709:32882 [5] NCCL INFO Channel 05/0 : 21[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272150:272327 [0] NCCL INFO Channel 13/0 : 112[0] -> 119[7] via P2P/IPC
+ip-26-0-168-34:272153:272328 [3] NCCL INFO Channel 11/0 : 115[3] -> 116[4] via P2P/IPC
+ip-26-0-161-142:62886:63057 [5] NCCL INFO Channel 11/0 : 45[5] -> 46[6] via P2P/IPC
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO Channel 10/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO Channel 09/0 : 1[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32704:32881 [0] NCCL INFO Channel 15/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO Connected all rings
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO Channel 06/0 : 81[1] -> 82[2] via P2P/IPC
+ip-26-0-161-78:114517:114687 [5] NCCL INFO Channel 02/0 : 69[5] -> 70[6] via P2P/IPC
+ip-26-0-161-138:96567:96760 [4] NCCL INFO Channel 10/0 : 36[4] -> 37[5] via P2P/IPC
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO Channel 06/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO Channel 11/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32710:32883 [6] NCCL INFO Channel 06/0 : 22[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO Channel 02/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO Channel 12/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO Channel 01/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-52:2722046:2722223 [1] NCCL INFO Channel 07/0 : 121[1] -> 122[2] via P2P/IPC
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO Channel 10/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153562:153735 [0] NCCL INFO Channel 15/0 : 88[0] -> 89[1] via P2P/IPC
+ip-26-0-161-78:114518:114693 [6] NCCL INFO Channel 03/0 : 70[6] -> 71[7] via P2P/IPC
+ip-26-0-161-142:62882:63056 [1] NCCL INFO Channel 12/0 : 41[1] -> 42[2] via P2P/IPC
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO Channel 09/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32709:32882 [5] NCCL INFO Channel 13/0 : 13[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96567:96760 [4] NCCL INFO Channel 11/0 : 36[4] -> 37[5] via P2P/IPC
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO Channel 07/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-161-142:62887:63055 [6] NCCL INFO Channel 13/0 : 46[6] -> 47[7] via P2P/IPC
+ip-26-0-161-103:96893:97067 [5] NCCL INFO Channel 05/0 : 21[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32709:32882 [5] NCCL INFO Channel 13/0 : 29[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO Channel 02/0 : 76[4] -> 77[5] via P2P/IPC
+ip-26-0-161-78:114514:114689 [2] NCCL INFO Channel 12/0 : 66[2] -> 67[3] via P2P/IPC
+ip-26-0-161-123:32710:32883 [6] NCCL INFO Channel 14/0 : 14[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96894:97068 [6] NCCL INFO Channel 06/0 : 22[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96567:96760 [4] NCCL INFO Channel 12/0 : 36[4] -> 37[5] via P2P/IPC
+ip-26-0-161-142:62885:63054 [4] NCCL INFO Channel 14/0 : 44[4] -> 45[5] via P2P/IPC
+ip-26-0-161-142:62886:63057 [5] NCCL INFO Channel 12/0 : 45[5] -> 46[6] via P2P/IPC
+ip-26-0-161-123:32710:32883 [6] NCCL INFO Channel 14/0 : 30[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO Channel 13/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-161-103:96892:97064 [4] NCCL INFO Channel 04/0 : 20[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96891:97063 [3] NCCL INFO Channel 03/0 : 19[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235649:235827 [1] NCCL INFO Channel 05/0 : 105[1] -> 106[2] via P2P/IPC
+ip-26-0-161-78:114517:114687 [5] NCCL INFO Channel 03/0 : 69[5] -> 70[6] via P2P/IPC
+ip-26-0-161-103:96889:97065 [1] NCCL INFO Channel 01/0 : 17[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96567:96760 [4] NCCL INFO Channel 14/0 : 36[4] -> 37[5] via P2P/IPC
+ip-26-0-161-142:62882:63056 [1] NCCL INFO Channel 13/0 : 41[1] -> 42[2] via P2P/IPC
+ip-26-0-161-103:96890:97070 [2] NCCL INFO Channel 02/0 : 18[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO Channel 14/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-161-78:114518:114693 [6] NCCL INFO Channel 04/0 : 70[6] -> 71[7] via P2P/IPC
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO Channel 07/0 : 81[1] -> 82[2] via P2P/IPC
+ip-26-0-161-142:62887:63055 [6] NCCL INFO Channel 14/0 : 46[6] -> 47[7] via P2P/IPC
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO Channel 03/0 : 76[4] -> 77[5] via P2P/IPC
+ip-26-0-161-138:96567:96760 [4] NCCL INFO Channel 15/0 : 36[4] -> 37[5] via P2P/IPC
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO Channel 15/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-161-78:114514:114689 [2] NCCL INFO Channel 13/0 : 66[2] -> 67[3] via P2P/IPC
+ip-26-0-161-142:62886:63057 [5] NCCL INFO Channel 13/0 : 45[5] -> 46[6] via P2P/IPC
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO Channel 05/0 : 48[0] -> 55[7] via P2P/IPC
+ip-26-0-161-142:62885:63054 [4] NCCL INFO Channel 15/0 : 44[4] -> 45[5] via P2P/IPC
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO Channel 09/0 : 81[1] -> 82[2] via P2P/IPC
+ip-26-0-168-30:235650:235824 [2] NCCL INFO Channel 01/0 : 106[2] -> 107[3] via P2P/IPC
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO Connected all rings
+ip-26-0-160-225:124165:124341 [5] NCCL INFO Channel 13/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-78:114517:114687 [5] NCCL INFO Channel 04/0 : 69[5] -> 70[6] via P2P/IPC
+ip-26-0-168-52:2722046:2722223 [1] NCCL INFO Channel 09/0 : 121[1] -> 122[2] via P2P/IPC
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO Channel 04/0 : 76[4] -> 77[5] via P2P/IPC
+ip-26-0-161-138:96566:96755 [3] NCCL INFO Channel 11/0 : 35[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235649:235827 [1] NCCL INFO Channel 06/0 : 105[1] -> 106[2] via P2P/IPC
+ip-26-0-168-34:272156:272325 [6] NCCL INFO Channel 01/0 : 118[6] -> 119[7] via P2P/IPC
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO Channel 08/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO Channel 15/0 : 7[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114518:114693 [6] NCCL INFO Channel 05/0 : 70[6] -> 71[7] via P2P/IPC
+ip-26-0-161-138:96568:96753 [5] NCCL INFO Channel 13/0 : 37[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO Channel 07/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124165:124341 [5] NCCL INFO Channel 15/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-78:114514:114689 [2] NCCL INFO Channel 14/0 : 66[2] -> 67[3] via P2P/IPC
+ip-26-0-168-34:272150:272327 [0] NCCL INFO Channel 14/0 : 112[0] -> 119[7] via P2P/IPC
+ip-26-0-161-138:96567:96760 [4] NCCL INFO Channel 12/0 : 36[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO Channel 00/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-52:2722046:2722223 [1] NCCL INFO Channel 11/0 : 121[1] -> 122[2] via P2P/IPC
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO Channel 15/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO Channel 08/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114517:114687 [5] NCCL INFO Channel 05/0 : 69[5] -> 70[6] via P2P/IPC
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO Channel 06/0 : 76[4] -> 77[5] via P2P/IPC
+ip-26-0-161-123:32704:32881 [0] NCCL INFO Channel 00/0 : 16[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114518:114693 [6] NCCL INFO Channel 06/0 : 70[6] -> 71[7] via P2P/IPC
+ip-26-0-161-123:32711:32884 [7] NCCL INFO Channel 07/0 : 23[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124167:124338 [7] NCCL INFO Channel 07/0 : 71[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722046:2722223 [1] NCCL INFO Channel 12/0 : 121[1] -> 122[2] via P2P/IPC
+ip-26-0-168-30:235650:235824 [2] NCCL INFO Channel 02/0 : 106[2] -> 107[3] via P2P/IPC
+ip-26-0-168-34:272156:272325 [6] NCCL INFO Channel 02/0 : 118[6] -> 119[7] via P2P/IPC
+ip-26-0-160-225:124167:124338 [7] NCCL INFO Channel 07/0 : 7[7] -> 71[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114514:114689 [2] NCCL INFO Channel 15/0 : 66[2] -> 67[3] via P2P/IPC
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO Channel 15/0 : 15[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272150:272327 [0] NCCL INFO Channel 15/0 : 112[0] -> 119[7] via P2P/IPC
+ip-26-0-161-123:32704:32881 [0] NCCL INFO Channel 08/0 : 8[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO Connected all rings
+ip-26-0-168-30:235649:235827 [1] NCCL INFO Channel 07/0 : 105[1] -> 106[2] via P2P/IPC
+ip-26-0-161-103:96888:97066 [0] NCCL INFO Channel 00/0 : 16[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62882:63056 [1] NCCL INFO Channel 14/0 : 41[1] -> 42[2] via P2P/IPC
+ip-26-0-161-123:32704:32881 [0] NCCL INFO Channel 08/0 : 24[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:124166:124339 [6] NCCL INFO Channel 14/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:124160:124335 [0] NCCL INFO Channel 08/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96895:97069 [7] NCCL INFO Channel 07/0 : 23[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32711:32884 [7] NCCL INFO Channel 15/0 : 15[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO Channel 01/0 : 77[5] -> 78[6] via P2P/IPC
+ip-26-0-161-78:114517:114687 [5] NCCL INFO Channel 07/0 : 69[5] -> 70[6] via P2P/IPC
+ip-26-0-161-123:32711:32884 [7] NCCL INFO Channel 15/0 : 31[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62886:63057 [5] NCCL INFO Channel 15/0 : 45[5] -> 46[6] via P2P/IPC
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO Channel 01/0 : 57[1] -> 58[2] via P2P/IPC
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO Channel 07/0 : 76[4] -> 77[5] via P2P/IPC
+ip-26-0-168-34:272156:272325 [6] NCCL INFO Channel 03/0 : 118[6] -> 119[7] via P2P/IPC
+ip-26-0-160-225:124166:124339 [6] NCCL INFO Channel 06/0 : 70[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:124160:124335 [0] NCCL INFO Channel 00/0 : 64[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO Channel 14/0 : 14[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:124166:124339 [6] NCCL INFO Channel 06/0 : 6[6] -> 70[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO Channel 01/0 : 78[6] -> 79[7] via P2P/IPC
+ip-26-0-160-225:124160:124335 [0] NCCL INFO Channel 00/0 : 0[0] -> 64[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62882:63056 [1] NCCL INFO Channel 15/0 : 41[1] -> 42[2] via P2P/IPC
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO Channel 08/0 : 8[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114518:114693 [6] NCCL INFO Channel 09/0 : 70[6] -> 71[7] via P2P/IPC
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO Channel 15/0 : 31[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235650:235824 [2] NCCL INFO Channel 04/0 : 106[2] -> 107[3] via P2P/IPC
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO Channel 03/0 : 57[1] -> 58[2] via P2P/IPC
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO Channel 02/0 : 77[5] -> 78[6] via P2P/IPC
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO Channel 14/0 : 30[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO Channel 11/0 : 81[1] -> 82[2] via P2P/IPC
+ip-26-0-168-30:235649:235827 [1] NCCL INFO Channel 09/0 : 105[1] -> 106[2] via P2P/IPC
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO Channel 08/0 : 24[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272156:272325 [6] NCCL INFO Channel 04/0 : 118[6] -> 119[7] via P2P/IPC
+ip-26-0-161-78:114517:114687 [5] NCCL INFO Channel 09/0 : 69[5] -> 70[6] via P2P/IPC
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO Channel 09/0 : 76[4] -> 77[5] via P2P/IPC
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO Channel 04/0 : 57[1] -> 58[2] via P2P/IPC
+ip-26-0-160-225:124164:124336 [4] NCCL INFO Channel 01/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-142:62884:63061 [3] NCCL INFO Channel 11/0 : 35[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235653:235825 [5] NCCL INFO Channel 01/0 : 109[5] -> 110[6] via P2P/IPC
+ip-26-0-165-202:153563:153740 [1] NCCL INFO Channel 01/0 : 89[1] -> 90[2] via P2P/IPC
+ip-26-0-161-142:62885:63054 [4] NCCL INFO Channel 12/0 : 36[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO Channel 02/0 : 78[6] -> 79[7] via P2P/IPC
+ip-26-0-161-78:114513:114688 [1] NCCL INFO Channel 09/0 : 65[1] -> 73[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62884:63061 [3] NCCL INFO Channel 03/0 : 43[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62887:63055 [6] NCCL INFO Channel 14/0 : 38[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO Channel 12/0 : 81[1] -> 82[2] via P2P/IPC
+ip-26-0-168-52:2722046:2722223 [1] NCCL INFO Channel 13/0 : 121[1] -> 122[2] via P2P/IPC
+ip-26-0-161-142:62886:63057 [5] NCCL INFO Channel 13/0 : 37[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62885:63054 [4] NCCL INFO Channel 04/0 : 44[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62881:63059 [0] NCCL INFO Channel 05/0 : 40[0] -> 47[7] via P2P/IPC
+ip-26-0-161-142:62887:63055 [6] NCCL INFO Channel 06/0 : 46[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO Channel 06/0 : 48[0] -> 55[7] via P2P/IPC
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO Channel 05/0 : 57[1] -> 58[2] via P2P/IPC
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO Channel 03/0 : 77[5] -> 78[6] via P2P/IPC
+ip-26-0-161-142:62884:63061 [3] NCCL INFO Channel 11/0 : 43[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62886:63057 [5] NCCL INFO Channel 05/0 : 45[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62885:63054 [4] NCCL INFO Channel 12/0 : 44[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272153:272328 [3] NCCL INFO Channel 13/0 : 115[3] -> 116[4] via P2P/IPC
+ip-26-0-161-142:62887:63055 [6] NCCL INFO Channel 14/0 : 46[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153563:153740 [1] NCCL INFO Channel 03/0 : 89[1] -> 90[2] via P2P/IPC
+ip-26-0-161-142:62886:63057 [5] NCCL INFO Channel 13/0 : 45[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114513:114688 [1] NCCL INFO Channel 01/0 : 33[1] -> 65[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272156:272325 [6] NCCL INFO Channel 05/0 : 118[6] -> 119[7] via P2P/IPC
+ip-26-0-165-202:153565:153738 [3] NCCL INFO Connected all rings
+ip-26-0-161-142:62883:63058 [2] NCCL INFO Channel 10/0 : 34[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO Channel 07/0 : 51[3] -> 52[4] via P2P/IPC
+ip-26-0-161-142:62882:63056 [1] NCCL INFO Channel 09/0 : 33[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124164:124336 [4] NCCL INFO Channel 02/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-168-30:235650:235824 [2] NCCL INFO Channel 05/0 : 106[2] -> 107[3] via P2P/IPC
+ip-26-0-168-52:2722046:2722223 [1] NCCL INFO Channel 14/0 : 121[1] -> 122[2] via P2P/IPC
+ip-26-0-161-78:114513:114688 [1] NCCL INFO Channel 01/0 : 65[1] -> 97[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62882:63056 [1] NCCL INFO Channel 01/0 : 41[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62883:63058 [2] NCCL INFO Channel 02/0 : 42[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO Channel 10/0 : 76[4] -> 77[5] via P2P/IPC
+ip-26-0-161-138:96566:96755 [3] NCCL INFO Channel 03/0 : 19[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO Channel 06/0 : 57[1] -> 58[2] via P2P/IPC
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO Connected all rings
+ip-26-0-165-202:153563:153740 [1] NCCL INFO Channel 04/0 : 89[1] -> 90[2] via P2P/IPC
+ip-26-0-161-138:96567:96760 [4] NCCL INFO Channel 04/0 : 20[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96569:96758 [6] NCCL INFO Channel 06/0 : 22[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62882:63056 [1] NCCL INFO Channel 09/0 : 41[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62883:63058 [2] NCCL INFO Channel 10/0 : 42[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO Channel 03/0 : 78[6] -> 79[7] via P2P/IPC
+ip-26-0-161-138:96566:96755 [3] NCCL INFO Channel 03/0 : 35[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96568:96753 [5] NCCL INFO Channel 05/0 : 21[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272153:272328 [3] NCCL INFO Channel 14/0 : 115[3] -> 116[4] via P2P/IPC
+ip-26-0-160-225:124164:124336 [4] NCCL INFO Channel 03/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-138:96567:96760 [4] NCCL INFO Channel 04/0 : 36[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722046:2722223 [1] NCCL INFO Channel 15/0 : 121[1] -> 122[2] via P2P/IPC
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO Channel 04/0 : 77[5] -> 78[6] via P2P/IPC
+ip-26-0-161-138:96569:96758 [6] NCCL INFO Channel 06/0 : 38[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96568:96753 [5] NCCL INFO Channel 05/0 : 37[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153563:153740 [1] NCCL INFO Channel 05/0 : 89[1] -> 90[2] via P2P/IPC
+ip-26-0-168-30:235649:235827 [1] NCCL INFO Channel 11/0 : 105[1] -> 106[2] via P2P/IPC
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO Channel 07/0 : 57[1] -> 58[2] via P2P/IPC
+ip-26-0-168-34:272156:272325 [6] NCCL INFO Channel 06/0 : 118[6] -> 119[7] via P2P/IPC
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO Channel 13/0 : 81[1] -> 82[2] via P2P/IPC
+ip-26-0-161-138:96565:96754 [2] NCCL INFO Channel 02/0 : 18[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO Channel 11/0 : 76[4] -> 77[5] via P2P/IPC
+ip-26-0-161-103:96892:97064 [4] NCCL INFO Channel 04/0 : 36[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272153:272328 [3] NCCL INFO Channel 15/0 : 115[3] -> 116[4] via P2P/IPC
+ip-26-0-161-138:96565:96754 [2] NCCL INFO Channel 02/0 : 34[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96894:97068 [6] NCCL INFO Channel 06/0 : 38[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235653:235825 [5] NCCL INFO Channel 02/0 : 109[5] -> 110[6] via P2P/IPC
+ip-26-0-165-202:153563:153740 [1] NCCL INFO Channel 06/0 : 89[1] -> 90[2] via P2P/IPC
+ip-26-0-161-103:96893:97067 [5] NCCL INFO Channel 05/0 : 37[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124164:124336 [4] NCCL INFO Channel 04/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO Channel 09/0 : 57[1] -> 58[2] via P2P/IPC
+ip-26-0-161-142:62881:63059 [0] NCCL INFO Channel 06/0 : 40[0] -> 47[7] via P2P/IPC
+ip-26-0-168-34:272156:272325 [6] NCCL INFO Channel 09/0 : 118[6] -> 119[7] via P2P/IPC
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO Channel 04/0 : 78[6] -> 79[7] via P2P/IPC
+ip-26-0-161-78:114516:114691 [4] NCCL INFO Connected all rings
+ip-26-0-168-52:2722045:2722219 [0] NCCL INFO Channel 05/0 : 120[0] -> 127[7] via P2P/IPC
+ip-26-0-161-78:114518:114693 [6] NCCL INFO Channel 10/0 : 70[6] -> 71[7] via P2P/IPC
+ip-26-0-165-202:153563:153740 [1] NCCL INFO Channel 07/0 : 89[1] -> 90[2] via P2P/IPC
+ip-26-0-161-103:96890:97070 [2] NCCL INFO Channel 02/0 : 34[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO Connected all rings
+ip-26-0-161-78:114517:114687 [5] NCCL INFO Channel 10/0 : 69[5] -> 70[6] via P2P/IPC
+ip-26-0-161-142:62881:63059 [0] NCCL INFO Channel 07/0 : 40[0] -> 47[7] via P2P/IPC
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO Channel 05/0 : 77[5] -> 78[6] via P2P/IPC
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO Channel 11/0 : 57[1] -> 58[2] via P2P/IPC
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO Channel 07/0 : 48[0] -> 55[7] via P2P/IPC
+ip-26-0-160-225:124164:124336 [4] NCCL INFO Channel 06/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO Channel 12/0 : 76[4] -> 77[5] via P2P/IPC
+ip-26-0-168-52:2722047:2722221 [2] NCCL INFO Channel 02/0 : 114[2] -> 122[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-52:2722046:2722223 [1] NCCL INFO Channel 01/0 : 113[1] -> 121[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO Channel 14/0 : 81[1] -> 82[2] via P2P/IPC
+ip-26-0-168-34:272156:272325 [6] NCCL INFO Channel 10/0 : 118[6] -> 119[7] via P2P/IPC
+ip-26-0-165-202:153563:153740 [1] NCCL INFO Channel 09/0 : 89[1] -> 90[2] via P2P/IPC
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO Channel 05/0 : 78[6] -> 79[7] via P2P/IPC
+ip-26-0-168-34:272152:272331 [2] NCCL INFO Channel 02/0 : 106[2] -> 114[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-52:2722045:2722219 [0] NCCL INFO Channel 06/0 : 120[0] -> 127[7] via P2P/IPC
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO Channel 12/0 : 57[1] -> 58[2] via P2P/IPC
+ip-26-0-168-34:272152:272331 [2] NCCL INFO Channel 10/0 : 106[2] -> 114[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62881:63059 [0] NCCL INFO Channel 13/0 : 40[0] -> 47[7] via P2P/IPC
+ip-26-0-160-225:124164:124336 [4] NCCL INFO Channel 07/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-168-52:2722046:2722223 [1] NCCL INFO Channel 09/0 : 57[1] -> 121[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272153:272328 [3] NCCL INFO Channel 03/0 : 107[3] -> 115[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-52:2722046:2722223 [1] NCCL INFO Channel 09/0 : 121[1] -> 57[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272152:272331 [2] NCCL INFO Channel 02/0 : 114[2] -> 122[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO Channel 07/0 : 77[5] -> 78[6] via P2P/IPC
+ip-26-0-161-78:114518:114693 [6] NCCL INFO Channel 11/0 : 70[6] -> 71[7] via P2P/IPC
+ip-26-0-168-34:272153:272328 [3] NCCL INFO Channel 11/0 : 107[3] -> 115[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114517:114687 [5] NCCL INFO Channel 11/0 : 69[5] -> 70[6] via P2P/IPC
+ip-26-0-168-34:272156:272325 [6] NCCL INFO Channel 11/0 : 118[6] -> 119[7] via P2P/IPC
+ip-26-0-168-34:272153:272328 [3] NCCL INFO Channel 03/0 : 115[3] -> 123[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO Channel 13/0 : 57[1] -> 58[2] via P2P/IPC
+ip-26-0-161-142:62881:63059 [0] NCCL INFO Channel 14/0 : 40[0] -> 47[7] via P2P/IPC
+ip-26-0-165-202:153563:153740 [1] NCCL INFO Channel 11/0 : 89[1] -> 90[2] via P2P/IPC
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO Channel 14/0 : 76[4] -> 77[5] via P2P/IPC
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO Connected all rings
+ip-26-0-161-103:96891:97063 [3] NCCL INFO Channel 03/0 : 35[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:124164:124336 [4] NCCL INFO Channel 09/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-168-34:272156:272325 [6] NCCL INFO Channel 12/0 : 118[6] -> 119[7] via P2P/IPC
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO Channel 06/0 : 78[6] -> 79[7] via P2P/IPC
+ip-26-0-161-142:62881:63059 [0] NCCL INFO Channel 15/0 : 40[0] -> 47[7] via P2P/IPC
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO Channel 14/0 : 57[1] -> 58[2] via P2P/IPC
+ip-26-0-168-52:2722047:2722221 [2] NCCL INFO Channel 10/0 : 58[2] -> 122[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114515:114690 [3] NCCL INFO Channel 01/0 : 67[3] -> 68[4] via P2P/IPC
+ip-26-0-165-202:153563:153740 [1] NCCL INFO Channel 12/0 : 89[1] -> 90[2] via P2P/IPC
+ip-26-0-168-52:2722047:2722221 [2] NCCL INFO Channel 10/0 : 122[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114518:114693 [6] NCCL INFO Channel 12/0 : 70[6] -> 71[7] via P2P/IPC
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO Channel 09/0 : 77[5] -> 78[6] via P2P/IPC
+ip-26-0-160-225:124164:124336 [4] NCCL INFO Channel 10/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-78:114517:114687 [5] NCCL INFO Channel 12/0 : 69[5] -> 70[6] via P2P/IPC
+ip-26-0-168-30:235652:235822 [4] NCCL INFO Connected all rings
+ip-26-0-168-34:272156:272325 [6] NCCL INFO Channel 13/0 : 118[6] -> 119[7] via P2P/IPC
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO Channel 15/0 : 57[1] -> 58[2] via P2P/IPC
+ip-26-0-168-30:235650:235824 [2] NCCL INFO Channel 06/0 : 106[2] -> 107[3] via P2P/IPC
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO Channel 01/0 : 59[3] -> 60[4] via P2P/IPC
+ip-26-0-165-202:153563:153740 [1] NCCL INFO Channel 13/0 : 89[1] -> 90[2] via P2P/IPC
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO Channel 15/0 : 76[4] -> 77[5] via P2P/IPC
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO Channel 09/0 : 51[3] -> 52[4] via P2P/IPC
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO Channel 09/0 : 78[6] -> 79[7] via P2P/IPC
+ip-26-0-161-142:62888:63060 [7] NCCL INFO Channel 15/0 : 39[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272155:272329 [5] NCCL INFO Channel 01/0 : 117[5] -> 118[6] via P2P/IPC
+ip-26-0-160-225:124164:124336 [4] NCCL INFO Channel 11/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-142:62881:63059 [0] NCCL INFO Channel 08/0 : 32[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO Channel 10/0 : 77[5] -> 78[6] via P2P/IPC
+ip-26-0-161-138:96564:96757 [1] NCCL INFO Channel 01/0 : 17[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272156:272325 [6] NCCL INFO Channel 14/0 : 118[6] -> 119[7] via P2P/IPC
+ip-26-0-161-78:114515:114690 [3] NCCL INFO Channel 02/0 : 67[3] -> 68[4] via P2P/IPC
+ip-26-0-161-78:114518:114693 [6] NCCL INFO Channel 13/0 : 70[6] -> 71[7] via P2P/IPC
+ip-26-0-161-142:62888:63060 [7] NCCL INFO Channel 07/0 : 47[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO Channel 02/0 : 59[3] -> 60[4] via P2P/IPC
+ip-26-0-161-138:96564:96757 [1] NCCL INFO Channel 01/0 : 33[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62881:63059 [0] NCCL INFO Channel 00/0 : 40[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235649:235827 [1] NCCL INFO Channel 12/0 : 105[1] -> 106[2] via P2P/IPC
+ip-26-0-165-202:153563:153740 [1] NCCL INFO Channel 14/0 : 89[1] -> 90[2] via P2P/IPC
+ip-26-0-161-78:114517:114687 [5] NCCL INFO Channel 13/0 : 69[5] -> 70[6] via P2P/IPC
+ip-26-0-161-142:62888:63060 [7] NCCL INFO Channel 15/0 : 47[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722045:2722219 [0] NCCL INFO Channel 07/0 : 120[0] -> 127[7] via P2P/IPC
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO Channel 05/0 : 56[0] -> 63[7] via P2P/IPC
+ip-26-0-161-142:62881:63059 [0] NCCL INFO Channel 08/0 : 40[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272155:272329 [5] NCCL INFO Channel 02/0 : 117[5] -> 118[6] via P2P/IPC
+ip-26-0-168-30:235653:235825 [5] NCCL INFO Channel 03/0 : 109[5] -> 110[6] via P2P/IPC
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO Channel 15/0 : 81[1] -> 82[2] via P2P/IPC
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO Channel 03/0 : 59[3] -> 60[4] via P2P/IPC
+ip-26-0-160-225:124164:124336 [4] NCCL INFO Channel 12/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-138:96570:96759 [7] NCCL INFO Channel 07/0 : 23[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO Channel 01/0 : 49[1] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272155:272329 [5] NCCL INFO Channel 03/0 : 117[5] -> 118[6] via P2P/IPC
+ip-26-0-165-202:153563:153740 [1] NCCL INFO Channel 15/0 : 89[1] -> 90[2] via P2P/IPC
+ip-26-0-161-138:96570:96759 [7] NCCL INFO Channel 07/0 : 39[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO Channel 06/0 : 56[0] -> 63[7] via P2P/IPC
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO Channel 10/0 : 78[6] -> 79[7] via P2P/IPC
+ip-26-0-168-52:2722045:2722219 [0] NCCL INFO Channel 13/0 : 120[0] -> 127[7] via P2P/IPC
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO Channel 11/0 : 77[5] -> 78[6] via P2P/IPC
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO Channel 05/0 : 59[3] -> 60[4] via P2P/IPC
+ip-26-0-161-78:114515:114690 [3] NCCL INFO Channel 03/0 : 67[3] -> 68[4] via P2P/IPC
+ip-26-0-160-225:124164:124336 [4] NCCL INFO Channel 14/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO Channel 13/0 : 48[0] -> 55[7] via P2P/IPC
+ip-26-0-161-78:114518:114693 [6] NCCL INFO Channel 14/0 : 70[6] -> 71[7] via P2P/IPC
+ip-26-0-168-34:272155:272329 [5] NCCL INFO Channel 04/0 : 117[5] -> 118[6] via P2P/IPC
+ip-26-0-161-78:114517:114687 [5] NCCL INFO Channel 15/0 : 69[5] -> 70[6] via P2P/IPC
+ip-26-0-161-103:96895:97069 [7] NCCL INFO Channel 07/0 : 39[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO Channel 11/0 : 78[6] -> 79[7] via P2P/IPC
+ip-26-0-168-30:235650:235824 [2] NCCL INFO Channel 07/0 : 106[2] -> 107[3] via P2P/IPC
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO Channel 07/0 : 56[0] -> 63[7] via P2P/IPC
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO Channel 10/0 : 51[3] -> 52[4] via P2P/IPC
+ip-26-0-168-52:2722045:2722219 [0] NCCL INFO Channel 14/0 : 120[0] -> 127[7] via P2P/IPC
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO Channel 06/0 : 59[3] -> 60[4] via P2P/IPC
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO Channel 12/0 : 77[5] -> 78[6] via P2P/IPC
+ip-26-0-165-202:153562:153735 [0] NCCL INFO Channel 05/0 : 88[0] -> 95[7] via P2P/IPC
+ip-26-0-168-34:272155:272329 [5] NCCL INFO Channel 05/0 : 117[5] -> 118[6] via P2P/IPC
+ip-26-0-165-202:153565:153738 [3] NCCL INFO Channel 01/0 : 91[3] -> 92[4] via P2P/IPC
+ip-26-0-161-103:96889:97065 [1] NCCL INFO Channel 01/0 : 33[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO Channel 12/0 : 78[6] -> 79[7] via P2P/IPC
+ip-26-0-168-30:235649:235827 [1] NCCL INFO Channel 13/0 : 105[1] -> 106[2] via P2P/IPC
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO Channel 13/0 : 56[0] -> 63[7] via P2P/IPC
+ip-26-0-165-202:153563:153740 [1] NCCL INFO Channel 01/0 : 81[1] -> 89[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235653:235825 [5] NCCL INFO Channel 04/0 : 109[5] -> 110[6] via P2P/IPC
+ip-26-0-168-52:2722045:2722219 [0] NCCL INFO Channel 15/0 : 120[0] -> 127[7] via P2P/IPC
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO Channel 07/0 : 59[3] -> 60[4] via P2P/IPC
+ip-26-0-161-78:114515:114690 [3] NCCL INFO Channel 05/0 : 67[3] -> 68[4] via P2P/IPC
+ip-26-0-165-202:153562:153735 [0] NCCL INFO Channel 06/0 : 88[0] -> 95[7] via P2P/IPC
+ip-26-0-168-34:272155:272329 [5] NCCL INFO Channel 07/0 : 117[5] -> 118[6] via P2P/IPC
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO Channel 13/0 : 77[5] -> 78[6] via P2P/IPC
+ip-26-0-160-225:124164:124336 [4] NCCL INFO Channel 15/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-202:153565:153738 [3] NCCL INFO Channel 02/0 : 91[3] -> 92[4] via P2P/IPC
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO Channel 14/0 : 56[0] -> 63[7] via P2P/IPC
+ip-26-0-161-78:114515:114690 [3] NCCL INFO Channel 06/0 : 67[3] -> 68[4] via P2P/IPC
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO Channel 09/0 : 59[3] -> 60[4] via P2P/IPC
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO Channel 13/0 : 78[6] -> 79[7] via P2P/IPC
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO Channel 14/0 : 48[0] -> 55[7] via P2P/IPC
+ip-26-0-161-78:114518:114693 [6] NCCL INFO Channel 14/0 : 70[6] -> 78[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153562:153735 [0] NCCL INFO Channel 07/0 : 88[0] -> 95[7] via P2P/IPC
+ip-26-0-168-34:272155:272329 [5] NCCL INFO Channel 09/0 : 117[5] -> 118[6] via P2P/IPC
+ip-26-0-165-202:153565:153738 [3] NCCL INFO Channel 03/0 : 91[3] -> 92[4] via P2P/IPC
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO Channel 15/0 : 77[5] -> 78[6] via P2P/IPC
+ip-26-0-168-52:2722045:2722219 [0] NCCL INFO Channel 00/0 : 112[0] -> 120[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235650:235824 [2] NCCL INFO Channel 09/0 : 106[2] -> 107[3] via P2P/IPC
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO Channel 15/0 : 56[0] -> 63[7] via P2P/IPC
+ip-26-0-168-52:2722052:2722226 [7] NCCL INFO Channel 07/0 : 119[7] -> 127[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO Channel 14/0 : 78[6] -> 79[7] via P2P/IPC
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO Channel 11/0 : 51[3] -> 52[4] via P2P/IPC
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO Channel 10/0 : 59[3] -> 60[4] via P2P/IPC
+ip-26-0-160-225:124165:124341 [5] NCCL INFO Channel 13/0 : 5[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114515:114690 [3] NCCL INFO Channel 07/0 : 67[3] -> 68[4] via P2P/IPC
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO Channel 05/0 : 80[0] -> 87[7] via P2P/IPC
+ip-26-0-165-202:153562:153735 [0] NCCL INFO Channel 13/0 : 88[0] -> 95[7] via P2P/IPC
+ip-26-0-168-34:272155:272329 [5] NCCL INFO Channel 10/0 : 117[5] -> 118[6] via P2P/IPC
+ip-26-0-165-202:153565:153738 [3] NCCL INFO Channel 05/0 : 91[3] -> 92[4] via P2P/IPC
+ip-26-0-168-30:235649:235827 [1] NCCL INFO Channel 14/0 : 105[1] -> 106[2] via P2P/IPC
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO Channel 11/0 : 59[3] -> 60[4] via P2P/IPC
+ip-26-0-168-30:235653:235825 [5] NCCL INFO Channel 05/0 : 109[5] -> 110[6] via P2P/IPC
+ip-26-0-160-225:124165:124341 [5] NCCL INFO Channel 05/0 : 69[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO Channel 13/0 : 13[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114515:114690 [3] NCCL INFO Channel 09/0 : 67[3] -> 68[4] via P2P/IPC
+ip-26-0-160-225:124165:124341 [5] NCCL INFO Channel 05/0 : 5[5] -> 69[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153562:153735 [0] NCCL INFO Channel 14/0 : 88[0] -> 95[7] via P2P/IPC
+ip-26-0-168-34:272155:272329 [5] NCCL INFO Channel 11/0 : 117[5] -> 118[6] via P2P/IPC
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO Channel 13/0 : 59[3] -> 60[4] via P2P/IPC
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO Channel 13/0 : 69[5] -> 77[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153565:153738 [3] NCCL INFO Channel 06/0 : 91[3] -> 92[4] via P2P/IPC
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO Channel 07/0 : 55[7] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO Channel 00/0 : 48[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO Channel 05/0 : 77[5] -> 85[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO Channel 13/0 : 29[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO Channel 01/0 : 124[4] -> 125[5] via P2P/IPC
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO Channel 15/0 : 48[0] -> 55[7] via P2P/IPC
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO Channel 14/0 : 70[6] -> 78[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO Channel 13/0 : 77[5] -> 85[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO Channel 14/0 : 59[3] -> 60[4] via P2P/IPC
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO Channel 06/0 : 78[6] -> 86[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114515:114690 [3] NCCL INFO Channel 10/0 : 67[3] -> 68[4] via P2P/IPC
+ip-26-0-165-202:153562:153735 [0] NCCL INFO Channel 15/0 : 88[0] -> 95[7] via P2P/IPC
+ip-26-0-168-30:235650:235824 [2] NCCL INFO Channel 10/0 : 106[2] -> 107[3] via P2P/IPC
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO Channel 14/0 : 78[6] -> 86[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO Channel 02/0 : 124[4] -> 125[5] via P2P/IPC
+ip-26-0-168-34:272155:272329 [5] NCCL INFO Channel 12/0 : 117[5] -> 118[6] via P2P/IPC
+ip-26-0-165-202:153565:153738 [3] NCCL INFO Channel 07/0 : 91[3] -> 92[4] via P2P/IPC
+ip-26-0-161-138:96563:96756 [0] NCCL INFO Channel 00/0 : 16[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO Channel 15/0 : 59[3] -> 60[4] via P2P/IPC
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO Channel 13/0 : 51[3] -> 52[4] via P2P/IPC
+ip-26-0-161-138:96563:96756 [0] NCCL INFO Channel 00/0 : 32[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114518:114693 [6] NCCL INFO Channel 06/0 : 38[6] -> 70[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272157:272324 [7] NCCL INFO Channel 07/0 : 111[7] -> 119[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272155:272329 [5] NCCL INFO Channel 13/0 : 117[5] -> 118[6] via P2P/IPC
+ip-26-0-165-202:153565:153738 [3] NCCL INFO Channel 09/0 : 91[3] -> 92[4] via P2P/IPC
+ip-26-0-161-78:114518:114693 [6] NCCL INFO Channel 06/0 : 70[6] -> 102[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235649:235827 [1] NCCL INFO Channel 15/0 : 105[1] -> 106[2] via P2P/IPC
+ip-26-0-168-34:272150:272327 [0] NCCL INFO Channel 00/0 : 104[0] -> 112[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272157:272324 [7] NCCL INFO Channel 15/0 : 111[7] -> 119[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114515:114690 [3] NCCL INFO Channel 11/0 : 67[3] -> 68[4] via P2P/IPC
+ip-26-0-168-34:272150:272327 [0] NCCL INFO Channel 08/0 : 104[0] -> 112[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO Channel 06/0 : 80[0] -> 87[7] via P2P/IPC
+ip-26-0-161-103:96888:97066 [0] NCCL INFO Channel 00/0 : 32[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235653:235825 [5] NCCL INFO Channel 07/0 : 109[5] -> 110[6] via P2P/IPC
+ip-26-0-168-34:272157:272324 [7] NCCL INFO Channel 07/0 : 119[7] -> 127[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272150:272327 [0] NCCL INFO Channel 00/0 : 112[0] -> 120[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272155:272329 [5] NCCL INFO Channel 15/0 : 117[5] -> 118[6] via P2P/IPC
+ip-26-0-165-202:153565:153738 [3] NCCL INFO Channel 10/0 : 91[3] -> 92[4] via P2P/IPC
+ip-26-0-165-202:153562:153735 [0] NCCL INFO Channel 00/0 : 80[0] -> 88[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153569:153739 [7] NCCL INFO Channel 07/0 : 87[7] -> 95[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO Channel 04/0 : 52[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO Channel 03/0 : 51[3] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153565:153738 [3] NCCL INFO Channel 11/0 : 91[3] -> 92[4] via P2P/IPC
+ip-26-0-161-78:114515:114690 [3] NCCL INFO Channel 13/0 : 67[3] -> 68[4] via P2P/IPC
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO Connected all rings
+ip-26-0-165-202:153565:153738 [3] NCCL INFO Channel 13/0 : 91[3] -> 92[4] via P2P/IPC
+ip-26-0-168-30:235650:235824 [2] NCCL INFO Channel 12/0 : 106[2] -> 107[3] via P2P/IPC
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO Channel 07/0 : 80[0] -> 87[7] via P2P/IPC
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO Channel 14/0 : 51[3] -> 52[4] via P2P/IPC
+ip-26-0-168-34:272156:272325 [6] NCCL INFO Channel 06/0 : 110[6] -> 118[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114515:114690 [3] NCCL INFO Channel 14/0 : 67[3] -> 68[4] via P2P/IPC
+ip-26-0-168-34:272155:272329 [5] NCCL INFO Channel 05/0 : 109[5] -> 117[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272154:272330 [4] NCCL INFO Channel 04/0 : 108[4] -> 116[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153565:153738 [3] NCCL INFO Channel 14/0 : 91[3] -> 92[4] via P2P/IPC
+ip-26-0-168-34:272156:272325 [6] NCCL INFO Channel 14/0 : 110[6] -> 118[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272155:272329 [5] NCCL INFO Channel 13/0 : 109[5] -> 117[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272154:272330 [4] NCCL INFO Channel 12/0 : 108[4] -> 116[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114519:114692 [7] NCCL INFO Channel 15/0 : 71[7] -> 79[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272156:272325 [6] NCCL INFO Channel 06/0 : 118[6] -> 126[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272155:272329 [5] NCCL INFO Channel 05/0 : 117[5] -> 125[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114516:114691 [4] NCCL INFO Channel 01/0 : 68[4] -> 69[5] via P2P/IPC
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO Channel 03/0 : 124[4] -> 125[5] via P2P/IPC
+ip-26-0-168-34:272154:272330 [4] NCCL INFO Channel 04/0 : 116[4] -> 124[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114515:114690 [3] NCCL INFO Channel 15/0 : 67[3] -> 68[4] via P2P/IPC
+ip-26-0-168-52:2722045:2722219 [0] NCCL INFO Channel 08/0 : 56[0] -> 120[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO Channel 13/0 : 80[0] -> 87[7] via P2P/IPC
+ip-26-0-168-52:2722045:2722219 [0] NCCL INFO Channel 08/0 : 120[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153565:153738 [3] NCCL INFO Channel 15/0 : 91[3] -> 92[4] via P2P/IPC
+ip-26-0-161-78:114516:114691 [4] NCCL INFO Channel 02/0 : 68[4] -> 69[5] via P2P/IPC
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO Channel 04/0 : 124[4] -> 125[5] via P2P/IPC
+ip-26-0-168-30:235653:235825 [5] NCCL INFO Channel 09/0 : 109[5] -> 110[6] via P2P/IPC
+ip-26-0-168-30:235648:235826 [0] NCCL INFO Channel 05/0 : 104[0] -> 111[7] via P2P/IPC
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO Channel 15/0 : 51[3] -> 52[4] via P2P/IPC
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO Channel 06/0 : 124[4] -> 125[5] via P2P/IPC
+ip-26-0-161-78:114516:114691 [4] NCCL INFO Channel 03/0 : 68[4] -> 69[5] via P2P/IPC
+ip-26-0-168-30:235654:235821 [6] NCCL INFO Channel 01/0 : 110[6] -> 111[7] via P2P/IPC
+ip-26-0-161-78:114514:114689 [2] NCCL INFO Channel 10/0 : 66[2] -> 74[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO Channel 14/0 : 80[0] -> 87[7] via P2P/IPC
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO Channel 07/0 : 124[4] -> 125[5] via P2P/IPC
+ip-26-0-165-202:153564:153737 [2] NCCL INFO Channel 02/0 : 82[2] -> 90[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114516:114691 [4] NCCL INFO Channel 04/0 : 68[4] -> 69[5] via P2P/IPC
+ip-26-0-168-30:235650:235824 [2] NCCL INFO Channel 13/0 : 106[2] -> 107[3] via P2P/IPC
+ip-26-0-165-202:153565:153738 [3] NCCL INFO Channel 03/0 : 83[3] -> 91[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153566:153741 [4] NCCL INFO Channel 04/0 : 84[4] -> 92[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114516:114691 [4] NCCL INFO Channel 06/0 : 68[4] -> 69[5] via P2P/IPC
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO Channel 09/0 : 124[4] -> 125[5] via P2P/IPC
+ip-26-0-168-52:2722052:2722226 [7] NCCL INFO Channel 15/0 : 63[7] -> 127[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235653:235825 [5] NCCL INFO Channel 10/0 : 109[5] -> 110[6] via P2P/IPC
+ip-26-0-168-52:2722052:2722226 [7] NCCL INFO Channel 15/0 : 127[7] -> 63[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114516:114691 [4] NCCL INFO Channel 07/0 : 68[4] -> 69[5] via P2P/IPC
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO Channel 10/0 : 124[4] -> 125[5] via P2P/IPC
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO Channel 15/0 : 80[0] -> 87[7] via P2P/IPC
+ip-26-0-161-78:114516:114691 [4] NCCL INFO Channel 09/0 : 68[4] -> 69[5] via P2P/IPC
+ip-26-0-168-30:235648:235826 [0] NCCL INFO Channel 06/0 : 104[0] -> 111[7] via P2P/IPC
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO Channel 11/0 : 124[4] -> 125[5] via P2P/IPC
+ip-26-0-168-30:235654:235821 [6] NCCL INFO Channel 02/0 : 110[6] -> 111[7] via P2P/IPC
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO Channel 15/0 : 71[7] -> 79[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO Channel 08/0 : 64[0] -> 72[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114516:114691 [4] NCCL INFO Channel 10/0 : 68[4] -> 69[5] via P2P/IPC
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO Channel 01/0 : 53[5] -> 54[6] via P2P/IPC
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO Channel 07/0 : 79[7] -> 87[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO Channel 00/0 : 72[0] -> 80[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO Channel 15/0 : 79[7] -> 87[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO Channel 08/0 : 72[0] -> 80[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO Channel 12/0 : 124[4] -> 125[5] via P2P/IPC
+ip-26-0-168-30:235650:235824 [2] NCCL INFO Channel 14/0 : 106[2] -> 107[3] via P2P/IPC
+ip-26-0-161-78:114512:114686 [0] NCCL INFO Channel 08/0 : 64[0] -> 72[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114516:114691 [4] NCCL INFO Channel 11/0 : 68[4] -> 69[5] via P2P/IPC
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO Channel 14/0 : 124[4] -> 125[5] via P2P/IPC
+ip-26-0-161-78:114519:114692 [7] NCCL INFO Channel 07/0 : 39[7] -> 71[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235653:235825 [5] NCCL INFO Channel 11/0 : 109[5] -> 110[6] via P2P/IPC
+ip-26-0-161-78:114516:114691 [4] NCCL INFO Channel 12/0 : 68[4] -> 69[5] via P2P/IPC
+ip-26-0-161-78:114519:114692 [7] NCCL INFO Channel 07/0 : 71[7] -> 103[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114512:114686 [0] NCCL INFO Channel 00/0 : 32[0] -> 64[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO Channel 02/0 : 53[5] -> 54[6] via P2P/IPC
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO Channel 15/0 : 124[4] -> 125[5] via P2P/IPC
+ip-26-0-161-78:114512:114686 [0] NCCL INFO Channel 00/0 : 64[0] -> 96[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114516:114691 [4] NCCL INFO Channel 14/0 : 68[4] -> 69[5] via P2P/IPC
+ip-26-0-168-30:235648:235826 [0] NCCL INFO Channel 07/0 : 104[0] -> 111[7] via P2P/IPC
+ip-26-0-168-30:235654:235821 [6] NCCL INFO Channel 03/0 : 110[6] -> 111[7] via P2P/IPC
+ip-26-0-161-78:114516:114691 [4] NCCL INFO Channel 15/0 : 68[4] -> 69[5] via P2P/IPC
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO Channel 03/0 : 53[5] -> 54[6] via P2P/IPC
+ip-26-0-168-30:235650:235824 [2] NCCL INFO Channel 15/0 : 106[2] -> 107[3] via P2P/IPC
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO Channel 01/0 : 82[2] -> 83[3] via P2P/IPC
+ip-26-0-168-30:235653:235825 [5] NCCL INFO Channel 12/0 : 109[5] -> 110[6] via P2P/IPC
+ip-26-0-161-78:114515:114690 [3] NCCL INFO Channel 11/0 : 67[3] -> 75[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114517:114687 [5] NCCL INFO Channel 13/0 : 69[5] -> 77[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO Channel 04/0 : 53[5] -> 54[6] via P2P/IPC
+ip-26-0-161-78:114516:114691 [4] NCCL INFO Channel 12/0 : 68[4] -> 76[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235648:235826 [0] NCCL INFO Channel 13/0 : 104[0] -> 111[7] via P2P/IPC
+ip-26-0-168-30:235654:235821 [6] NCCL INFO Channel 04/0 : 110[6] -> 111[7] via P2P/IPC
+ip-26-0-161-78:114517:114687 [5] NCCL INFO Channel 05/0 : 37[5] -> 69[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114517:114687 [5] NCCL INFO Channel 05/0 : 69[5] -> 101[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO Channel 02/0 : 50[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO Channel 05/0 : 53[5] -> 54[6] via P2P/IPC
+ip-26-0-168-30:235653:235825 [5] NCCL INFO Channel 13/0 : 109[5] -> 110[6] via P2P/IPC
+ip-26-0-168-52:2722048:2722220 [3] NCCL INFO Channel 03/0 : 115[3] -> 123[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235649:235827 [1] NCCL INFO Channel 09/0 : 97[1] -> 105[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235649:235827 [1] NCCL INFO Channel 01/0 : 105[1] -> 113[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235649:235827 [1] NCCL INFO Channel 09/0 : 105[1] -> 113[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-52:2722048:2722220 [3] NCCL INFO Channel 11/0 : 59[3] -> 123[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235648:235826 [0] NCCL INFO Channel 14/0 : 104[0] -> 111[7] via P2P/IPC
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO Channel 01/0 : 75[3] -> 76[4] via P2P/IPC
+ip-26-0-168-30:235654:235821 [6] NCCL INFO Channel 05/0 : 110[6] -> 111[7] via P2P/IPC
+ip-26-0-168-52:2722048:2722220 [3] NCCL INFO Channel 11/0 : 123[3] -> 59[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO Channel 07/0 : 53[5] -> 54[6] via P2P/IPC
+ip-26-0-168-52:2722050:2722225 [5] NCCL INFO Connected all rings
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO Connected all rings
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO Channel 02/0 : 75[3] -> 76[4] via P2P/IPC
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO Channel 01/0 : 86[6] -> 87[7] via P2P/IPC
+ip-26-0-168-30:235653:235825 [5] NCCL INFO Channel 15/0 : 109[5] -> 110[6] via P2P/IPC
+ip-26-0-168-30:235649:235827 [1] NCCL INFO Channel 09/0 : 89[1] -> 105[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO Channel 03/0 : 75[3] -> 76[4] via P2P/IPC
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO Channel 01/0 : 54[6] -> 55[7] via P2P/IPC
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO Channel 01/0 : 50[2] -> 51[3] via P2P/IPC
+ip-26-0-168-30:235654:235821 [6] NCCL INFO Channel 06/0 : 110[6] -> 111[7] via P2P/IPC
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO Channel 05/0 : 75[3] -> 76[4] via P2P/IPC
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO Channel 09/0 : 53[5] -> 54[6] via P2P/IPC
+ip-26-0-168-30:235648:235826 [0] NCCL INFO Channel 15/0 : 104[0] -> 111[7] via P2P/IPC
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO Channel 02/0 : 82[2] -> 83[3] via P2P/IPC
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO Channel 06/0 : 75[3] -> 76[4] via P2P/IPC
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO Channel 02/0 : 86[6] -> 87[7] via P2P/IPC
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO Channel 02/0 : 54[6] -> 55[7] via P2P/IPC
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO Connected all rings
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO Channel 07/0 : 75[3] -> 76[4] via P2P/IPC
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO Channel 02/0 : 50[2] -> 51[3] via P2P/IPC
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO Channel 10/0 : 53[5] -> 54[6] via P2P/IPC
+ip-26-0-168-34:272151:272326 [1] NCCL INFO Channel 01/0 : 97[1] -> 113[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235654:235821 [6] NCCL INFO Channel 09/0 : 110[6] -> 111[7] via P2P/IPC
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO Channel 09/0 : 75[3] -> 76[4] via P2P/IPC
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO Channel 04/0 : 82[2] -> 83[3] via P2P/IPC
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO Channel 03/0 : 54[6] -> 55[7] via P2P/IPC
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO Channel 10/0 : 75[3] -> 76[4] via P2P/IPC
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO Channel 04/0 : 50[2] -> 51[3] via P2P/IPC
+ip-26-0-168-30:235654:235821 [6] NCCL INFO Channel 10/0 : 110[6] -> 111[7] via P2P/IPC
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO Channel 11/0 : 53[5] -> 54[6] via P2P/IPC
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO Channel 03/0 : 86[6] -> 87[7] via P2P/IPC
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO Channel 11/0 : 75[3] -> 76[4] via P2P/IPC
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO Channel 04/0 : 54[6] -> 55[7] via P2P/IPC
+ip-26-0-168-30:235654:235821 [6] NCCL INFO Channel 11/0 : 110[6] -> 111[7] via P2P/IPC
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO Channel 13/0 : 75[3] -> 76[4] via P2P/IPC
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO Channel 05/0 : 82[2] -> 83[3] via P2P/IPC
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO Channel 05/0 : 50[2] -> 51[3] via P2P/IPC
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO Channel 14/0 : 75[3] -> 76[4] via P2P/IPC
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO Channel 12/0 : 53[5] -> 54[6] via P2P/IPC
+ip-26-0-168-30:235654:235821 [6] NCCL INFO Channel 12/0 : 110[6] -> 111[7] via P2P/IPC
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO Channel 05/0 : 54[6] -> 55[7] via P2P/IPC
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO Channel 15/0 : 75[3] -> 76[4] via P2P/IPC
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO Channel 04/0 : 86[6] -> 87[7] via P2P/IPC
+ip-26-0-168-30:235654:235821 [6] NCCL INFO Channel 13/0 : 110[6] -> 111[7] via P2P/IPC
+ip-26-0-168-30:235652:235822 [4] NCCL INFO Channel 01/0 : 108[4] -> 109[5] via P2P/IPC
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO Channel 06/0 : 82[2] -> 83[3] via P2P/IPC
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO Channel 10/0 : 66[2] -> 74[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-52:2722050:2722225 [5] NCCL INFO Channel 01/0 : 125[5] -> 126[6] via P2P/IPC
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO Channel 12/0 : 68[4] -> 76[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO Channel 02/0 : 74[2] -> 82[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO Channel 11/0 : 67[3] -> 75[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO Channel 04/0 : 76[4] -> 84[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO Channel 10/0 : 74[2] -> 82[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO Channel 03/0 : 75[3] -> 83[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO Channel 12/0 : 76[4] -> 84[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO Channel 11/0 : 75[3] -> 83[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO Channel 05/0 : 86[6] -> 87[7] via P2P/IPC
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO Channel 13/0 : 53[5] -> 54[6] via P2P/IPC
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO Channel 06/0 : 50[2] -> 51[3] via P2P/IPC
+ip-26-0-161-78:114514:114689 [2] NCCL INFO Channel 02/0 : 34[2] -> 66[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114516:114691 [4] NCCL INFO Channel 04/0 : 36[4] -> 68[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114515:114690 [3] NCCL INFO Channel 03/0 : 35[3] -> 67[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO Channel 06/0 : 54[6] -> 55[7] via P2P/IPC
+ip-26-0-161-78:114514:114689 [2] NCCL INFO Channel 02/0 : 66[2] -> 98[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO Channel 07/0 : 82[2] -> 83[3] via P2P/IPC
+ip-26-0-161-78:114516:114691 [4] NCCL INFO Channel 04/0 : 68[4] -> 100[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114515:114690 [3] NCCL INFO Channel 03/0 : 67[3] -> 99[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO Channel 15/0 : 53[5] -> 54[6] via P2P/IPC
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO Channel 06/0 : 86[6] -> 87[7] via P2P/IPC
+ip-26-0-168-52:2722050:2722225 [5] NCCL INFO Channel 02/0 : 125[5] -> 126[6] via P2P/IPC
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO Channel 07/0 : 50[2] -> 51[3] via P2P/IPC
+ip-26-0-168-52:2722050:2722225 [5] NCCL INFO Channel 03/0 : 125[5] -> 126[6] via P2P/IPC
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO Channel 09/0 : 82[2] -> 83[3] via P2P/IPC
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO Channel 09/0 : 54[6] -> 55[7] via P2P/IPC
+ip-26-0-168-52:2722050:2722225 [5] NCCL INFO Channel 04/0 : 125[5] -> 126[6] via P2P/IPC
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO Channel 09/0 : 86[6] -> 87[7] via P2P/IPC
+ip-26-0-168-52:2722050:2722225 [5] NCCL INFO Channel 05/0 : 125[5] -> 126[6] via P2P/IPC
+ip-26-0-168-52:2722050:2722225 [5] NCCL INFO Channel 07/0 : 125[5] -> 126[6] via P2P/IPC
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO Channel 10/0 : 82[2] -> 83[3] via P2P/IPC
+ip-26-0-168-52:2722050:2722225 [5] NCCL INFO Channel 09/0 : 125[5] -> 126[6] via P2P/IPC
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO Channel 10/0 : 86[6] -> 87[7] via P2P/IPC
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO Channel 01/0 : 81[1] -> 97[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-52:2722050:2722225 [5] NCCL INFO Channel 10/0 : 125[5] -> 126[6] via P2P/IPC
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO Channel 01/0 : 97[1] -> 113[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-52:2722050:2722225 [5] NCCL INFO Channel 11/0 : 125[5] -> 126[6] via P2P/IPC
+ip-26-0-160-225:124162:124342 [2] NCCL INFO Channel 01/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO Channel 12/0 : 82[2] -> 83[3] via P2P/IPC
+ip-26-0-168-30:235654:235821 [6] NCCL INFO Channel 14/0 : 110[6] -> 111[7] via P2P/IPC
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO Connected all rings
+ip-26-0-168-52:2722050:2722225 [5] NCCL INFO Channel 12/0 : 125[5] -> 126[6] via P2P/IPC
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO Channel 09/0 : 50[2] -> 51[3] via P2P/IPC
+ip-26-0-160-225:124162:124342 [2] NCCL INFO Channel 02/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO Channel 10/0 : 54[6] -> 55[7] via P2P/IPC
+ip-26-0-168-30:235652:235822 [4] NCCL INFO Channel 02/0 : 108[4] -> 109[5] via P2P/IPC
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO Channel 11/0 : 86[6] -> 87[7] via P2P/IPC
+ip-26-0-168-30:235652:235822 [4] NCCL INFO Channel 03/0 : 108[4] -> 109[5] via P2P/IPC
+ip-26-0-160-225:124162:124342 [2] NCCL INFO Channel 04/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO Channel 13/0 : 82[2] -> 83[3] via P2P/IPC
+ip-26-0-168-30:235654:235821 [6] NCCL INFO Channel 14/0 : 102[6] -> 110[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235654:235821 [6] NCCL INFO Channel 06/0 : 110[6] -> 118[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:124162:124342 [2] NCCL INFO Channel 05/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-168-30:235654:235821 [6] NCCL INFO Channel 14/0 : 110[6] -> 118[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO Channel 10/0 : 50[2] -> 51[3] via P2P/IPC
+ip-26-0-168-30:235652:235822 [4] NCCL INFO Channel 04/0 : 108[4] -> 109[5] via P2P/IPC
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO Channel 11/0 : 54[6] -> 55[7] via P2P/IPC
+ip-26-0-168-34:272151:272326 [1] NCCL INFO Channel 01/0 : 113[1] -> 97[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO Channel 12/0 : 86[6] -> 87[7] via P2P/IPC
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO Channel 14/0 : 82[2] -> 83[3] via P2P/IPC
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO Channel 06/0 : 86[6] -> 102[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:124162:124342 [2] NCCL INFO Channel 06/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-168-30:235654:235821 [6] NCCL INFO Channel 14/0 : 94[6] -> 110[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO Channel 06/0 : 102[6] -> 118[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:124162:124342 [2] NCCL INFO Channel 07/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO Channel 12/0 : 50[2] -> 51[3] via P2P/IPC
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO Channel 12/0 : 54[6] -> 55[7] via P2P/IPC
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO Channel 13/0 : 86[6] -> 87[7] via P2P/IPC
+ip-26-0-160-225:124162:124342 [2] NCCL INFO Channel 09/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO Channel 15/0 : 82[2] -> 83[3] via P2P/IPC
+ip-26-0-160-225:124162:124342 [2] NCCL INFO Channel 10/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-168-30:235652:235822 [4] NCCL INFO Channel 06/0 : 108[4] -> 109[5] via P2P/IPC
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO Channel 14/0 : 86[6] -> 87[7] via P2P/IPC
+ip-26-0-160-225:124162:124342 [2] NCCL INFO Channel 12/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO Channel 01/0 : 83[3] -> 84[4] via P2P/IPC
+ip-26-0-168-30:235652:235822 [4] NCCL INFO Channel 07/0 : 108[4] -> 109[5] via P2P/IPC
+ip-26-0-160-225:124162:124342 [2] NCCL INFO Channel 13/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO Channel 01/0 : 73[1] -> 81[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO Channel 09/0 : 73[1] -> 81[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235652:235822 [4] NCCL INFO Channel 09/0 : 108[4] -> 109[5] via P2P/IPC
+ip-26-0-160-225:124162:124342 [2] NCCL INFO Channel 14/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO Channel 02/0 : 83[3] -> 84[4] via P2P/IPC
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO Channel 01/0 : 81[1] -> 89[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235652:235822 [4] NCCL INFO Channel 10/0 : 108[4] -> 109[5] via P2P/IPC
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO Channel 13/0 : 54[6] -> 55[7] via P2P/IPC
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO Channel 01/0 : 81[1] -> 97[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO Channel 13/0 : 50[2] -> 51[3] via P2P/IPC
+ip-26-0-168-30:235652:235822 [4] NCCL INFO Channel 11/0 : 108[4] -> 109[5] via P2P/IPC
+ip-26-0-165-202:153563:153740 [1] NCCL INFO Channel 09/0 : 73[1] -> 89[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153563:153740 [1] NCCL INFO Channel 09/0 : 89[1] -> 105[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO Channel 03/0 : 83[3] -> 84[4] via P2P/IPC
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO Channel 01/0 : 85[5] -> 86[6] via P2P/IPC
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO Channel 01/0 : 97[1] -> 81[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO Channel 01/0 : 65[1] -> 97[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235652:235822 [4] NCCL INFO Channel 12/0 : 108[4] -> 109[5] via P2P/IPC
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO Channel 14/0 : 54[6] -> 55[7] via P2P/IPC
+ip-26-0-168-30:235655:235828 [7] NCCL INFO Channel 15/0 : 103[7] -> 111[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235648:235826 [0] NCCL INFO Channel 08/0 : 96[0] -> 104[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO Channel 01/0 : 97[1] -> 65[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235655:235828 [7] NCCL INFO Channel 07/0 : 111[7] -> 119[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235648:235826 [0] NCCL INFO Channel 00/0 : 104[0] -> 112[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:124162:124342 [2] NCCL INFO Channel 15/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-168-30:235649:235827 [1] NCCL INFO Channel 09/0 : 105[1] -> 89[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235655:235828 [7] NCCL INFO Channel 15/0 : 111[7] -> 119[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235648:235826 [0] NCCL INFO Channel 08/0 : 104[0] -> 112[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO Channel 05/0 : 83[3] -> 84[4] via P2P/IPC
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO Channel 14/0 : 50[2] -> 51[3] via P2P/IPC
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO Channel 02/0 : 85[5] -> 86[6] via P2P/IPC
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO Channel 09/0 : 73[1] -> 89[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO Channel 01/0 : 84[4] -> 85[5] via P2P/IPC
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO Channel 07/0 : 87[7] -> 103[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153563:153740 [1] NCCL INFO Channel 09/0 : 57[1] -> 89[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO Channel 09/0 : 89[1] -> 73[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO Channel 07/0 : 103[7] -> 119[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235655:235828 [7] NCCL INFO Channel 15/0 : 95[7] -> 111[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235648:235826 [0] NCCL INFO Channel 08/0 : 88[0] -> 104[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:124161:124337 [1] NCCL INFO Channel 09/0 : 1[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-52:2722050:2722225 [5] NCCL INFO Channel 13/0 : 125[5] -> 126[6] via P2P/IPC
+ip-26-0-168-34:272157:272324 [7] NCCL INFO Channel 07/0 : 103[7] -> 119[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272150:272327 [0] NCCL INFO Channel 00/0 : 96[0] -> 112[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO Channel 00/0 : 80[0] -> 96[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO Channel 06/0 : 83[3] -> 84[4] via P2P/IPC
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO Channel 00/0 : 96[0] -> 112[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO Channel 03/0 : 85[5] -> 86[6] via P2P/IPC
+ip-26-0-160-225:124161:124337 [1] NCCL INFO Channel 01/0 : 65[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO Channel 02/0 : 84[4] -> 85[5] via P2P/IPC
+ip-26-0-160-225:124161:124337 [1] NCCL INFO Channel 01/0 : 1[1] -> 65[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO Channel 09/0 : 9[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272157:272324 [7] NCCL INFO Channel 07/0 : 119[7] -> 103[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272150:272327 [0] NCCL INFO Channel 00/0 : 112[0] -> 96[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO Channel 07/0 : 83[3] -> 84[4] via P2P/IPC
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO Channel 04/0 : 85[5] -> 86[6] via P2P/IPC
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO Channel 03/0 : 84[4] -> 85[5] via P2P/IPC
+ip-26-0-168-52:2722050:2722225 [5] NCCL INFO Channel 15/0 : 125[5] -> 126[6] via P2P/IPC
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO Channel 09/0 : 25[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235652:235822 [4] NCCL INFO Channel 14/0 : 108[4] -> 109[5] via P2P/IPC
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO Channel 09/0 : 83[3] -> 84[4] via P2P/IPC
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO Channel 15/0 : 50[2] -> 51[3] via P2P/IPC
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO Channel 05/0 : 85[5] -> 86[6] via P2P/IPC
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO Channel 04/0 : 116[4] -> 124[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO Channel 04/0 : 84[4] -> 85[5] via P2P/IPC
+ip-26-0-168-52:2722050:2722225 [5] NCCL INFO Channel 05/0 : 117[5] -> 125[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235652:235822 [4] NCCL INFO Channel 15/0 : 108[4] -> 109[5] via P2P/IPC
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO Channel 10/0 : 83[3] -> 84[4] via P2P/IPC
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO Channel 12/0 : 60[4] -> 124[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO Channel 12/0 : 124[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722050:2722225 [5] NCCL INFO Channel 13/0 : 61[5] -> 125[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO Channel 07/0 : 85[5] -> 86[6] via P2P/IPC
+ip-26-0-168-52:2722051:2722224 [6] NCCL INFO Channel 06/0 : 118[6] -> 126[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-52:2722050:2722225 [5] NCCL INFO Channel 13/0 : 125[5] -> 61[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO Channel 06/0 : 84[4] -> 85[5] via P2P/IPC
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO Channel 11/0 : 83[3] -> 84[4] via P2P/IPC
+ip-26-0-168-30:235653:235825 [5] NCCL INFO Channel 13/0 : 101[5] -> 109[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO Channel 01/0 : 41[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO Channel 07/0 : 79[7] -> 87[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO Channel 00/0 : 72[0] -> 80[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-52:2722051:2722224 [6] NCCL INFO Channel 14/0 : 62[6] -> 126[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO Channel 02/0 : 42[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235653:235825 [5] NCCL INFO Channel 05/0 : 109[5] -> 117[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO Channel 09/0 : 85[5] -> 86[6] via P2P/IPC
+ip-26-0-168-52:2722051:2722224 [6] NCCL INFO Channel 14/0 : 126[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO Channel 09/0 : 41[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235653:235825 [5] NCCL INFO Channel 13/0 : 109[5] -> 117[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO Channel 15/0 : 79[7] -> 87[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO Channel 08/0 : 72[0] -> 80[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO Channel 10/0 : 42[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO Channel 01/0 : 49[1] -> 57[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO Channel 07/0 : 84[4] -> 85[5] via P2P/IPC
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO Channel 07/0 : 87[7] -> 95[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO Channel 00/0 : 80[0] -> 88[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO Channel 02/0 : 50[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO Channel 13/0 : 83[3] -> 84[4] via P2P/IPC
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO Channel 05/0 : 85[5] -> 101[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO Channel 10/0 : 85[5] -> 86[6] via P2P/IPC
+ip-26-0-161-142:62882:63056 [1] NCCL INFO Channel 09/0 : 25[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO Channel 08/0 : 72[0] -> 88[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO Channel 05/0 : 101[5] -> 117[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235653:235825 [5] NCCL INFO Channel 13/0 : 93[5] -> 109[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62883:63058 [2] NCCL INFO Channel 10/0 : 26[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO Channel 09/0 : 84[4] -> 85[5] via P2P/IPC
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO Channel 15/0 : 79[7] -> 95[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO Channel 01/0 : 33[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO Channel 02/0 : 34[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO Channel 00/0 : 80[0] -> 96[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO Channel 07/0 : 87[7] -> 103[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153562:153735 [0] NCCL INFO Channel 08/0 : 72[0] -> 88[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO Channel 09/0 : 25[1] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO Channel 14/0 : 83[3] -> 84[4] via P2P/IPC
+ip-26-0-165-202:153569:153739 [7] NCCL INFO Channel 15/0 : 79[7] -> 95[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62882:63056 [1] NCCL INFO Channel 09/0 : 41[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO Channel 10/0 : 26[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153562:153735 [0] NCCL INFO Channel 08/0 : 88[0] -> 104[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272156:272325 [6] NCCL INFO Channel 06/0 : 102[6] -> 118[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153569:153739 [7] NCCL INFO Channel 15/0 : 95[7] -> 111[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO Channel 09/0 : 57[1] -> 89[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32705:32880 [1] NCCL INFO Channel 09/0 : 25[1] -> 57[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62883:63058 [2] NCCL INFO Channel 10/0 : 42[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO Channel 10/0 : 58[2] -> 90[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO Channel 00/0 : 96[0] -> 80[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO Channel 01/0 : 49[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO Channel 11/0 : 85[5] -> 86[6] via P2P/IPC
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO Channel 07/0 : 103[7] -> 87[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO Channel 00/0 : 64[0] -> 96[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO Channel 07/0 : 71[7] -> 103[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO Channel 02/0 : 50[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO Channel 10/0 : 84[4] -> 85[5] via P2P/IPC
+ip-26-0-161-138:96564:96757 [1] NCCL INFO Channel 01/0 : 33[1] -> 65[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO Channel 08/0 : 88[0] -> 72[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96565:96754 [2] NCCL INFO Channel 02/0 : 34[2] -> 66[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO Channel 15/0 : 83[3] -> 84[4] via P2P/IPC
+ip-26-0-168-34:272156:272325 [6] NCCL INFO Channel 06/0 : 118[6] -> 102[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO Channel 15/0 : 95[7] -> 79[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32705:32880 [1] NCCL INFO Channel 09/0 : 57[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153562:153735 [0] NCCL INFO Channel 08/0 : 56[0] -> 88[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO Channel 00/0 : 96[0] -> 64[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235648:235826 [0] NCCL INFO Channel 08/0 : 104[0] -> 88[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153569:153739 [7] NCCL INFO Channel 15/0 : 63[7] -> 95[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO Channel 09/0 : 121[1] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO Channel 07/0 : 103[7] -> 71[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235655:235828 [7] NCCL INFO Channel 15/0 : 111[7] -> 95[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153563:153740 [1] NCCL INFO Channel 09/0 : 89[1] -> 57[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO Channel 09/0 : 57[1] -> 121[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO Channel 01/0 : 52[4] -> 53[5] via P2P/IPC
+ip-26-0-161-138:96564:96757 [1] NCCL INFO Channel 01/0 : 65[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO Channel 12/0 : 85[5] -> 86[6] via P2P/IPC
+ip-26-0-168-34:272155:272329 [5] NCCL INFO Channel 05/0 : 101[5] -> 117[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114513:114688 [1] NCCL INFO Channel 01/0 : 1[1] -> 65[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96565:96754 [2] NCCL INFO Channel 02/0 : 66[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO Channel 11/0 : 84[4] -> 85[5] via P2P/IPC
+ip-26-0-161-78:114513:114688 [1] NCCL INFO Channel 01/0 : 65[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO Channel 09/0 : 89[1] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO Channel 02/0 : 52[4] -> 53[5] via P2P/IPC
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO Channel 13/0 : 85[5] -> 86[6] via P2P/IPC
+ip-26-0-168-34:272155:272329 [5] NCCL INFO Channel 05/0 : 117[5] -> 101[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO Channel 09/0 : 57[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-52:2722046:2722223 [1] NCCL INFO Channel 01/0 : 121[1] -> 113[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114513:114688 [1] NCCL INFO Channel 01/0 : 97[1] -> 65[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO Channel 12/0 : 84[4] -> 85[5] via P2P/IPC
+ip-26-0-161-78:114513:114688 [1] NCCL INFO Channel 01/0 : 65[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO Channel 06/0 : 46[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO Channel 02/0 : 74[2] -> 82[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO Channel 03/0 : 52[4] -> 53[5] via P2P/IPC
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO Channel 15/0 : 85[5] -> 86[6] via P2P/IPC
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO Channel 14/0 : 46[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153563:153740 [1] NCCL INFO Channel 09/0 : 105[1] -> 89[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO Channel 10/0 : 74[2] -> 82[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO Channel 06/0 : 54[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153563:153740 [1] NCCL INFO Channel 09/0 : 89[1] -> 73[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO Channel 14/0 : 84[4] -> 85[5] via P2P/IPC
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO Channel 02/0 : 82[2] -> 90[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO Channel 01/0 : 57[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO Channel 01/0 : 113[1] -> 97[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO Channel 04/0 : 52[4] -> 53[5] via P2P/IPC
+ip-26-0-161-78:114513:114688 [1] NCCL INFO Channel 09/0 : 73[1] -> 65[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO Channel 01/0 : 97[1] -> 81[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96564:96757 [1] NCCL INFO Channel 01/0 : 49[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124163:124340 [3] NCCL INFO Connected all rings
+ip-26-0-161-138:96564:96757 [1] NCCL INFO Channel 01/0 : 33[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO Channel 15/0 : 84[4] -> 85[5] via P2P/IPC
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO Channel 10/0 : 74[2] -> 90[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:124161:124337 [1] NCCL INFO Channel 09/0 : 9[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO Channel 06/0 : 52[4] -> 53[5] via P2P/IPC
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO Channel 02/0 : 82[2] -> 98[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153563:153740 [1] NCCL INFO Channel 01/0 : 89[1] -> 81[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO Channel 01/0 : 81[1] -> 73[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272151:272326 [1] NCCL INFO Channel 01/0 : 121[1] -> 113[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153564:153737 [2] NCCL INFO Channel 10/0 : 74[2] -> 90[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO Channel 06/0 : 78[6] -> 86[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO Channel 09/0 : 81[1] -> 73[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153564:153737 [2] NCCL INFO Channel 10/0 : 90[2] -> 106[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO Channel 14/0 : 78[6] -> 86[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO Channel 09/0 : 105[1] -> 97[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272151:272326 [1] NCCL INFO Channel 01/0 : 113[1] -> 105[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO Channel 09/0 : 73[1] -> 65[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO Channel 01/0 : 57[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96564:96757 [1] NCCL INFO Channel 09/0 : 41[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO Channel 06/0 : 86[6] -> 94[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272151:272326 [1] NCCL INFO Channel 09/0 : 113[1] -> 105[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235651:235823 [3] NCCL INFO Connected all rings
+ip-26-0-161-103:96889:97065 [1] NCCL INFO Channel 01/0 : 25[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO Channel 07/0 : 52[4] -> 53[5] via P2P/IPC
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO Channel 01/0 : 49[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235649:235827 [1] NCCL INFO Channel 01/0 : 113[1] -> 105[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96889:97065 [1] NCCL INFO Channel 01/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO Channel 09/0 : 49[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO Channel 05/0 : 77[5] -> 85[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO Channel 10/0 : 90[2] -> 74[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235649:235827 [1] NCCL INFO Channel 09/0 : 113[1] -> 105[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96889:97065 [1] NCCL INFO Channel 09/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO Channel 03/0 : 75[3] -> 83[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235649:235827 [1] NCCL INFO Channel 09/0 : 105[1] -> 97[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO Channel 04/0 : 76[4] -> 84[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO Channel 13/0 : 77[5] -> 85[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO Channel 11/0 : 75[3] -> 83[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO Channel 09/0 : 52[4] -> 53[5] via P2P/IPC
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO Channel 12/0 : 76[4] -> 84[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO Channel 05/0 : 85[5] -> 93[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62887:63055 [6] NCCL INFO Channel 14/0 : 30[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO Channel 03/0 : 83[3] -> 91[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO Channel 04/0 : 84[4] -> 92[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO Channel 06/0 : 38[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO Channel 14/0 : 30[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO Channel 14/0 : 78[6] -> 94[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO Channel 14/0 : 62[6] -> 94[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO Channel 13/0 : 77[5] -> 93[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO Channel 10/0 : 52[4] -> 53[5] via P2P/IPC
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO Channel 11/0 : 75[3] -> 91[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO Channel 05/0 : 85[5] -> 101[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO Channel 06/0 : 54[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO Channel 03/0 : 83[3] -> 99[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96569:96758 [6] NCCL INFO Channel 06/0 : 38[6] -> 70[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO Channel 04/0 : 84[4] -> 100[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153566:153741 [4] NCCL INFO Channel 12/0 : 76[4] -> 92[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO Channel 05/0 : 101[5] -> 85[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153566:153741 [4] NCCL INFO Channel 12/0 : 92[4] -> 108[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96569:96758 [6] NCCL INFO Channel 06/0 : 70[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-52:2722046:2722223 [1] NCCL INFO Channel 00/0 : 121[1] -> 120[0] via P2P/IPC
+ip-26-0-161-123:32705:32880 [1] NCCL INFO Channel 09/0 : 41[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO Channel 12/0 : 76[4] -> 92[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32705:32880 [1] NCCL INFO Channel 09/0 : 25[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-52:2722046:2722223 [1] NCCL INFO Channel 08/0 : 121[1] -> 120[0] via P2P/IPC
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO Channel 00/0 : 97[1] -> 96[0] via P2P/IPC
+ip-26-0-161-123:32705:32880 [1] NCCL INFO Channel 01/0 : 25[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO Channel 08/0 : 97[1] -> 96[0] via P2P/IPC
+ip-26-0-168-30:235649:235827 [1] NCCL INFO Channel 00/0 : 105[1] -> 104[0] via P2P/IPC
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO Channel 01/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO Channel 09/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO Channel 12/0 : 92[4] -> 76[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153565:153738 [3] NCCL INFO Channel 11/0 : 75[3] -> 91[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153565:153738 [3] NCCL INFO Channel 11/0 : 91[3] -> 107[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO Channel 09/0 : 9[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO Channel 06/0 : 86[6] -> 102[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235649:235827 [1] NCCL INFO Channel 08/0 : 105[1] -> 104[0] via P2P/IPC
+ip-26-0-165-202:153567:153742 [5] NCCL INFO Channel 13/0 : 77[5] -> 93[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62882:63056 [1] NCCL INFO Channel 01/0 : 49[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO Channel 06/0 : 102[6] -> 86[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO Channel 11/0 : 91[3] -> 75[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO Channel 06/0 : 70[6] -> 102[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62882:63056 [1] NCCL INFO Channel 09/0 : 49[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153567:153742 [5] NCCL INFO Channel 13/0 : 93[5] -> 109[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO Channel 05/0 : 69[5] -> 101[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62882:63056 [1] NCCL INFO Channel 09/0 : 41[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114513:114688 [1] NCCL INFO Channel 00/0 : 65[1] -> 64[0] via P2P/IPC
+ip-26-0-168-34:272151:272326 [1] NCCL INFO Channel 00/0 : 113[1] -> 112[0] via P2P/IPC
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO Channel 06/0 : 102[6] -> 70[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO Channel 13/0 : 93[5] -> 77[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114518:114693 [6] NCCL INFO Channel 06/0 : 6[6] -> 70[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO Channel 05/0 : 101[5] -> 69[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114513:114688 [1] NCCL INFO Channel 08/0 : 65[1] -> 64[0] via P2P/IPC
+ip-26-0-165-202:153567:153742 [5] NCCL INFO Channel 13/0 : 61[5] -> 93[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114518:114693 [6] NCCL INFO Channel 06/0 : 70[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235653:235825 [5] NCCL INFO Channel 13/0 : 109[5] -> 93[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124163:124340 [3] NCCL INFO Channel 01/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-168-34:272151:272326 [1] NCCL INFO Channel 08/0 : 113[1] -> 112[0] via P2P/IPC
+ip-26-0-161-142:62887:63055 [6] NCCL INFO Channel 14/0 : 46[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32710:32883 [6] NCCL INFO Channel 14/0 : 30[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114518:114693 [6] NCCL INFO Channel 06/0 : 102[6] -> 70[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114518:114693 [6] NCCL INFO Channel 06/0 : 70[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32705:32880 [1] NCCL INFO Channel 00/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-160-225:124163:124340 [3] NCCL INFO Channel 02/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-103:96889:97065 [1] NCCL INFO Channel 00/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO Channel 00/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO Channel 06/0 : 118[6] -> 102[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:124163:124340 [3] NCCL INFO Channel 03/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-202:153568:153736 [6] NCCL INFO Channel 14/0 : 78[6] -> 94[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO Channel 06/0 : 102[6] -> 86[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153568:153736 [6] NCCL INFO Channel 14/0 : 94[6] -> 110[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114518:114693 [6] NCCL INFO Channel 14/0 : 78[6] -> 70[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO Channel 08/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-161-103:96889:97065 [1] NCCL INFO Channel 08/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-161-142:62882:63056 [1] NCCL INFO Channel 00/0 : 41[1] -> 40[0] via P2P/IPC
+ip-26-0-160-225:124163:124340 [3] NCCL INFO Channel 05/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO Channel 00/0 : 57[1] -> 56[0] via P2P/IPC
+ip-26-0-168-34:272156:272325 [6] NCCL INFO Channel 06/0 : 126[6] -> 118[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO Channel 14/0 : 94[6] -> 78[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235651:235823 [3] NCCL INFO Channel 01/0 : 107[3] -> 108[4] via P2P/IPC
+ip-26-0-168-34:272156:272325 [6] NCCL INFO Channel 06/0 : 118[6] -> 110[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO Channel 14/0 : 110[6] -> 102[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:124161:124337 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-142:62882:63056 [1] NCCL INFO Channel 08/0 : 41[1] -> 40[0] via P2P/IPC
+ip-26-0-161-123:32710:32883 [6] NCCL INFO Channel 14/0 : 62[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO Channel 06/0 : 94[6] -> 86[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272156:272325 [6] NCCL INFO Channel 14/0 : 118[6] -> 110[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153568:153736 [6] NCCL INFO Channel 14/0 : 62[6] -> 94[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235654:235821 [6] NCCL INFO Channel 14/0 : 110[6] -> 94[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO Channel 08/0 : 57[1] -> 56[0] via P2P/IPC
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO Channel 06/0 : 86[6] -> 78[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:124163:124340 [3] NCCL INFO Channel 06/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO Channel 14/0 : 86[6] -> 78[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235651:235823 [3] NCCL INFO Channel 02/0 : 107[3] -> 108[4] via P2P/IPC
+ip-26-0-160-225:124161:124337 [1] NCCL INFO Channel 08/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-202:153568:153736 [6] NCCL INFO Channel 14/0 : 94[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO Channel 01/0 : 89[1] -> 81[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124163:124340 [3] NCCL INFO Channel 07/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO Channel 14/0 : 126[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:124166:124339 [6] NCCL INFO Channel 14/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235651:235823 [3] NCCL INFO Channel 03/0 : 107[3] -> 108[4] via P2P/IPC
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO Channel 01/0 : 81[1] -> 73[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO Channel 14/0 : 62[6] -> 126[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO Channel 09/0 : 81[1] -> 73[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96569:96758 [6] NCCL INFO Channel 06/0 : 54[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235651:235823 [3] NCCL INFO Channel 05/0 : 107[3] -> 108[4] via P2P/IPC
+ip-26-0-160-225:124163:124340 [3] NCCL INFO Channel 09/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-123:32705:32880 [1] NCCL INFO Channel 08/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-161-138:96569:96758 [6] NCCL INFO Channel 06/0 : 38[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO Channel 14/0 : 94[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO Channel 14/0 : 62[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-52:2722051:2722224 [6] NCCL INFO Channel 06/0 : 126[6] -> 118[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO Channel 11/0 : 52[4] -> 53[5] via P2P/IPC
+ip-26-0-160-225:124163:124340 [3] NCCL INFO Channel 10/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-168-30:235651:235823 [3] NCCL INFO Channel 06/0 : 107[3] -> 108[4] via P2P/IPC
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO Channel 07/0 : 47[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO Channel 06/0 : 62[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:124163:124340 [3] NCCL INFO Channel 11/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-168-30:235651:235823 [3] NCCL INFO Channel 07/0 : 107[3] -> 108[4] via P2P/IPC
+ip-26-0-165-202:153568:153736 [6] NCCL INFO Channel 14/0 : 110[6] -> 94[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96569:96758 [6] NCCL INFO Channel 14/0 : 46[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO Channel 15/0 : 47[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153568:153736 [6] NCCL INFO Channel 14/0 : 94[6] -> 78[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96894:97068 [6] NCCL INFO Channel 06/0 : 30[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO Channel 06/0 : 62[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO Channel 06/0 : 54[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO Channel 12/0 : 52[4] -> 53[5] via P2P/IPC
+ip-26-0-161-103:96894:97068 [6] NCCL INFO Channel 06/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO Channel 07/0 : 55[7] -> 63[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO Channel 14/0 : 54[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32710:32883 [6] NCCL INFO Channel 14/0 : 46[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235651:235823 [3] NCCL INFO Channel 09/0 : 107[3] -> 108[4] via P2P/IPC
+ip-26-0-160-225:124163:124340 [3] NCCL INFO Channel 13/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-103:96894:97068 [6] NCCL INFO Channel 14/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32710:32883 [6] NCCL INFO Channel 14/0 : 30[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235651:235823 [3] NCCL INFO Channel 10/0 : 107[3] -> 108[4] via P2P/IPC
+ip-26-0-160-225:124163:124340 [3] NCCL INFO Channel 14/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO Channel 14/0 : 52[4] -> 53[5] via P2P/IPC
+ip-26-0-161-142:62888:63060 [7] NCCL INFO Channel 15/0 : 31[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO Channel 07/0 : 39[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235651:235823 [3] NCCL INFO Channel 11/0 : 107[3] -> 108[4] via P2P/IPC
+ip-26-0-161-138:96564:96757 [1] NCCL INFO Channel 00/0 : 33[1] -> 32[0] via P2P/IPC
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO Channel 15/0 : 31[7] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62887:63055 [6] NCCL INFO Channel 06/0 : 54[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO Channel 00/0 : 81[1] -> 80[0] via P2P/IPC
+ip-26-0-160-225:124163:124340 [3] NCCL INFO Channel 15/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-142:62887:63055 [6] NCCL INFO Channel 14/0 : 54[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO Channel 15/0 : 63[7] -> 95[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32710:32883 [6] NCCL INFO Channel 06/0 : 30[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-52:2722051:2722224 [6] NCCL INFO Channel 00/0 : 126[6] -> 125[5] via P2P/IPC
+ip-26-0-161-142:62887:63055 [6] NCCL INFO Channel 14/0 : 46[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO Channel 06/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62888:63060 [7] NCCL INFO Channel 15/0 : 47[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235651:235823 [3] NCCL INFO Channel 13/0 : 107[3] -> 108[4] via P2P/IPC
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO Channel 08/0 : 81[1] -> 80[0] via P2P/IPC
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO Channel 07/0 : 55[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96564:96757 [1] NCCL INFO Channel 08/0 : 33[1] -> 32[0] via P2P/IPC
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO Channel 14/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32711:32884 [7] NCCL INFO Channel 15/0 : 31[7] -> 63[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96570:96759 [7] NCCL INFO Channel 07/0 : 39[7] -> 71[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO Channel 15/0 : 52[4] -> 53[5] via P2P/IPC
+ip-26-0-168-30:235651:235823 [3] NCCL INFO Channel 14/0 : 107[3] -> 108[4] via P2P/IPC
+ip-26-0-165-202:153569:153739 [7] NCCL INFO Channel 15/0 : 95[7] -> 63[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124162:124342 [2] NCCL INFO Channel 10/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32711:32884 [7] NCCL INFO Channel 15/0 : 63[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96570:96759 [7] NCCL INFO Channel 07/0 : 71[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO Channel 15/0 : 127[7] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO Channel 00/0 : 49[1] -> 48[0] via P2P/IPC
+ip-26-0-160-225:124164:124336 [4] NCCL INFO Channel 12/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235651:235823 [3] NCCL INFO Channel 15/0 : 107[3] -> 108[4] via P2P/IPC
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO Channel 00/0 : 62[6] -> 61[5] via P2P/IPC
+ip-26-0-160-225:124163:124340 [3] NCCL INFO Channel 11/0 : 3[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114519:114692 [7] NCCL INFO Channel 07/0 : 7[7] -> 71[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO Channel 15/0 : 63[7] -> 127[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114519:114692 [7] NCCL INFO Channel 07/0 : 71[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO Channel 05/0 : 62[6] -> 61[5] via P2P/IPC
+ip-26-0-160-225:124162:124342 [2] NCCL INFO Channel 02/0 : 66[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:124164:124336 [4] NCCL INFO Channel 04/0 : 68[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO Channel 15/0 : 95[7] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO Channel 00/0 : 73[1] -> 72[0] via P2P/IPC
+ip-26-0-160-225:124163:124340 [3] NCCL INFO Channel 03/0 : 67[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO Channel 08/0 : 49[1] -> 48[0] via P2P/IPC
+ip-26-0-160-225:124162:124342 [2] NCCL INFO Channel 02/0 : 2[2] -> 66[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:124163:124340 [3] NCCL INFO Channel 03/0 : 3[3] -> 67[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-52:2722052:2722226 [7] NCCL INFO Channel 07/0 : 127[7] -> 119[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114519:114692 [7] NCCL INFO Channel 07/0 : 103[7] -> 71[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO Channel 15/0 : 63[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124167:124338 [7] NCCL INFO Channel 15/0 : 15[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO Channel 08/0 : 62[6] -> 61[5] via P2P/IPC
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO Channel 03/0 : 43[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114519:114692 [7] NCCL INFO Channel 07/0 : 71[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO Channel 08/0 : 73[1] -> 72[0] via P2P/IPC
+ip-26-0-168-30:235654:235821 [6] NCCL INFO Channel 06/0 : 118[6] -> 110[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO Channel 05/0 : 45[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235650:235824 [2] NCCL INFO Channel 10/0 : 98[2] -> 106[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235652:235822 [4] NCCL INFO Channel 12/0 : 100[4] -> 108[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO Channel 11/0 : 43[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153568:153736 [6] NCCL INFO Channel 06/0 : 94[6] -> 86[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO Channel 13/0 : 62[6] -> 61[5] via P2P/IPC
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO Channel 04/0 : 44[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO Channel 06/0 : 86[6] -> 78[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO Channel 13/0 : 45[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235654:235821 [6] NCCL INFO Channel 14/0 : 118[6] -> 110[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235651:235823 [3] NCCL INFO Channel 11/0 : 99[3] -> 107[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153569:153739 [7] NCCL INFO Channel 15/0 : 111[7] -> 95[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO Channel 07/0 : 119[7] -> 103[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO Channel 14/0 : 86[6] -> 78[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO Channel 03/0 : 51[3] -> 59[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO Channel 07/0 : 63[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32710:32883 [6] NCCL INFO Channel 00/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO Channel 12/0 : 44[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96569:96758 [6] NCCL INFO Channel 00/0 : 38[6] -> 37[5] via P2P/IPC
+ip-26-0-161-142:62887:63055 [6] NCCL INFO Channel 00/0 : 46[6] -> 45[5] via P2P/IPC
+ip-26-0-168-30:235650:235824 [2] NCCL INFO Channel 02/0 : 106[2] -> 114[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235652:235822 [4] NCCL INFO Channel 04/0 : 108[4] -> 116[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO Channel 05/0 : 53[5] -> 61[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-52:2722051:2722224 [6] NCCL INFO Channel 05/0 : 126[6] -> 125[5] via P2P/IPC
+ip-26-0-165-202:153569:153739 [7] NCCL INFO Channel 15/0 : 95[7] -> 79[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO Channel 07/0 : 103[7] -> 87[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235654:235821 [6] NCCL INFO Channel 14/0 : 110[6] -> 102[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32711:32884 [7] NCCL INFO Channel 15/0 : 47[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO Channel 14/0 : 78[6] -> 70[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114519:114692 [7] NCCL INFO Channel 15/0 : 79[7] -> 71[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235651:235823 [3] NCCL INFO Channel 03/0 : 107[3] -> 115[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO Channel 04/0 : 52[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96570:96759 [7] NCCL INFO Channel 07/0 : 55[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235650:235824 [2] NCCL INFO Channel 10/0 : 106[2] -> 114[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235652:235822 [4] NCCL INFO Channel 12/0 : 108[4] -> 116[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235651:235823 [3] NCCL INFO Channel 11/0 : 107[3] -> 115[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96894:97068 [6] NCCL INFO Channel 00/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-161-138:96570:96759 [7] NCCL INFO Channel 07/0 : 39[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62887:63055 [6] NCCL INFO Channel 05/0 : 46[6] -> 45[5] via P2P/IPC
+ip-26-0-160-225:124164:124336 [4] NCCL INFO Channel 04/0 : 4[4] -> 68[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722051:2722224 [6] NCCL INFO Channel 08/0 : 126[6] -> 125[5] via P2P/IPC
+ip-26-0-161-123:32710:32883 [6] NCCL INFO Channel 05/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO Channel 00/0 : 54[6] -> 53[5] via P2P/IPC
+ip-26-0-161-138:96569:96758 [6] NCCL INFO Channel 05/0 : 38[6] -> 37[5] via P2P/IPC
+ip-26-0-161-142:62886:63057 [5] NCCL INFO Channel 13/0 : 29[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272157:272324 [7] NCCL INFO Channel 07/0 : 127[7] -> 119[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153569:153739 [7] NCCL INFO Channel 07/0 : 95[7] -> 87[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62885:63054 [4] NCCL INFO Channel 12/0 : 28[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO Channel 03/0 : 83[3] -> 99[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO Channel 04/0 : 84[4] -> 100[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO Channel 07/0 : 87[7] -> 79[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO Channel 05/0 : 37[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272157:272324 [7] NCCL INFO Channel 07/0 : 119[7] -> 111[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96894:97068 [6] NCCL INFO Channel 05/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-168-30:235651:235823 [3] NCCL INFO Channel 11/0 : 91[3] -> 107[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-52:2722051:2722224 [6] NCCL INFO Channel 13/0 : 126[6] -> 125[5] via P2P/IPC
+ip-26-0-161-138:96569:96758 [6] NCCL INFO Channel 08/0 : 38[6] -> 37[5] via P2P/IPC
+ip-26-0-161-142:62887:63055 [6] NCCL INFO Channel 08/0 : 46[6] -> 45[5] via P2P/IPC
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO Channel 15/0 : 111[7] -> 103[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO Channel 13/0 : 29[5] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32710:32883 [6] NCCL INFO Channel 08/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO Channel 07/0 : 63[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO Channel 15/0 : 87[7] -> 79[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO Channel 04/0 : 36[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235652:235822 [4] NCCL INFO Channel 12/0 : 92[4] -> 108[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272157:272324 [7] NCCL INFO Channel 15/0 : 119[7] -> 111[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO Channel 07/0 : 95[7] -> 87[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96570:96759 [7] NCCL INFO Channel 15/0 : 47[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62886:63057 [5] NCCL INFO Channel 13/0 : 45[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO Channel 03/0 : 99[3] -> 115[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO Channel 12/0 : 28[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153563:153740 [1] NCCL INFO Channel 00/0 : 89[1] -> 88[0] via P2P/IPC
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO Channel 04/0 : 100[4] -> 116[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO Channel 15/0 : 79[7] -> 71[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO Channel 05/0 : 54[6] -> 53[5] via P2P/IPC
+ip-26-0-168-34:272153:272328 [3] NCCL INFO Channel 03/0 : 99[3] -> 115[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272154:272330 [4] NCCL INFO Channel 04/0 : 100[4] -> 116[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO Channel 07/0 : 87[7] -> 79[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO Channel 07/0 : 55[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62885:63054 [4] NCCL INFO Channel 12/0 : 44[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96894:97068 [6] NCCL INFO Channel 08/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO Channel 13/0 : 61[5] -> 93[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO Channel 14/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96569:96758 [6] NCCL INFO Channel 13/0 : 38[6] -> 37[5] via P2P/IPC
+ip-26-0-168-30:235651:235823 [3] NCCL INFO Channel 11/0 : 107[3] -> 91[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO Channel 12/0 : 60[4] -> 92[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO Channel 12/0 : 12[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32710:32883 [6] NCCL INFO Channel 13/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO Channel 00/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO Channel 15/0 : 87[7] -> 79[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO Channel 15/0 : 55[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62887:63055 [6] NCCL INFO Channel 13/0 : 46[6] -> 45[5] via P2P/IPC
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO Channel 12/0 : 28[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO Channel 05/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO Channel 05/0 : 53[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96568:96753 [5] NCCL INFO Channel 05/0 : 37[5] -> 69[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235652:235822 [4] NCCL INFO Channel 12/0 : 108[4] -> 92[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722052:2722226 [7] NCCL INFO Channel 01/0 : 127[7] -> 120[0] via P2P/IPC
+ip-26-0-165-202:153565:153738 [3] NCCL INFO Channel 11/0 : 59[3] -> 91[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO Channel 03/0 : 99[3] -> 83[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO Channel 08/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO Channel 01/0 : 63[7] -> 56[0] via P2P/IPC
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO Channel 04/0 : 52[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96567:96760 [4] NCCL INFO Channel 04/0 : 36[4] -> 68[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96894:97068 [6] NCCL INFO Channel 13/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO Channel 04/0 : 100[4] -> 84[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO Channel 08/0 : 54[6] -> 53[5] via P2P/IPC
+ip-26-0-165-202:153566:153741 [4] NCCL INFO Channel 12/0 : 60[4] -> 92[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO Channel 03/0 : 67[3] -> 99[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO Channel 10/0 : 10[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114519:114692 [7] NCCL INFO Channel 01/0 : 71[7] -> 64[0] via P2P/IPC
+ip-26-0-168-34:272153:272328 [3] NCCL INFO Channel 03/0 : 115[3] -> 99[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153563:153740 [1] NCCL INFO Channel 08/0 : 89[1] -> 88[0] via P2P/IPC
+ip-26-0-165-202:153569:153739 [7] NCCL INFO Channel 01/0 : 95[7] -> 88[0] via P2P/IPC
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO Channel 04/0 : 68[4] -> 100[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO Channel 13/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-168-34:272154:272330 [4] NCCL INFO Channel 04/0 : 116[4] -> 100[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO Channel 11/0 : 11[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO Channel 13/0 : 54[6] -> 53[5] via P2P/IPC
+ip-26-0-161-138:96568:96753 [5] NCCL INFO Channel 05/0 : 69[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62884:63061 [3] NCCL INFO Channel 11/0 : 27[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-52:2722052:2722226 [7] NCCL INFO Channel 02/0 : 127[7] -> 120[0] via P2P/IPC
+ip-26-0-161-78:114517:114687 [5] NCCL INFO Channel 05/0 : 5[5] -> 69[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO Channel 01/0 : 79[7] -> 72[0] via P2P/IPC
+ip-26-0-165-202:153567:153742 [5] NCCL INFO Channel 13/0 : 93[5] -> 61[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO Channel 02/0 : 63[7] -> 56[0] via P2P/IPC
+ip-26-0-161-138:96567:96760 [4] NCCL INFO Channel 04/0 : 68[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:124166:124339 [6] NCCL INFO Channel 00/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO Channel 03/0 : 35[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114517:114687 [5] NCCL INFO Channel 05/0 : 69[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO Channel 03/0 : 99[3] -> 67[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96895:97069 [7] NCCL INFO Channel 07/0 : 31[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO Channel 10/0 : 26[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153566:153741 [4] NCCL INFO Channel 12/0 : 92[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272156:272325 [6] NCCL INFO Channel 00/0 : 118[6] -> 117[5] via P2P/IPC
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO Channel 11/0 : 27[3] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32706:32877 [2] NCCL INFO Channel 10/0 : 26[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO Channel 04/0 : 100[4] -> 68[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96895:97069 [7] NCCL INFO Channel 07/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO Channel 11/0 : 27[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62884:63061 [3] NCCL INFO Channel 11/0 : 43[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO Channel 11/0 : 59[3] -> 91[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-52:2722052:2722226 [7] NCCL INFO Channel 03/0 : 127[7] -> 120[0] via P2P/IPC
+ip-26-0-161-78:114516:114691 [4] NCCL INFO Channel 04/0 : 4[4] -> 68[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96895:97069 [7] NCCL INFO Channel 15/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32711:32884 [7] NCCL INFO Channel 15/0 : 31[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124166:124339 [6] NCCL INFO Channel 05/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO Channel 00/0 : 86[6] -> 85[5] via P2P/IPC
+ip-26-0-161-78:114519:114692 [7] NCCL INFO Channel 02/0 : 71[7] -> 64[0] via P2P/IPC
+ip-26-0-168-34:272156:272325 [6] NCCL INFO Channel 05/0 : 118[6] -> 117[5] via P2P/IPC
+ip-26-0-165-202:153569:153739 [7] NCCL INFO Channel 02/0 : 95[7] -> 88[0] via P2P/IPC
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO Channel 03/0 : 51[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:124165:124341 [5] NCCL INFO Channel 13/0 : 13[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32707:32878 [3] NCCL INFO Channel 11/0 : 27[3] -> 59[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114516:114691 [4] NCCL INFO Channel 04/0 : 68[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153568:153736 [6] NCCL INFO Channel 00/0 : 94[6] -> 93[5] via P2P/IPC
+ip-26-0-161-78:114517:114687 [5] NCCL INFO Channel 05/0 : 101[5] -> 69[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96566:96755 [3] NCCL INFO Channel 03/0 : 35[3] -> 67[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-52:2722052:2722226 [7] NCCL INFO Channel 04/0 : 127[7] -> 120[0] via P2P/IPC
+ip-26-0-161-123:32706:32877 [2] NCCL INFO Channel 10/0 : 58[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114517:114687 [5] NCCL INFO Channel 05/0 : 69[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO Channel 05/0 : 86[6] -> 85[5] via P2P/IPC
+ip-26-0-160-225:124166:124339 [6] NCCL INFO Channel 08/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-168-34:272156:272325 [6] NCCL INFO Channel 08/0 : 118[6] -> 117[5] via P2P/IPC
+ip-26-0-161-78:114519:114692 [7] NCCL INFO Channel 03/0 : 71[7] -> 64[0] via P2P/IPC
+ip-26-0-161-142:62888:63060 [7] NCCL INFO Channel 07/0 : 55[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124164:124336 [4] NCCL INFO Channel 12/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114516:114691 [4] NCCL INFO Channel 04/0 : 100[4] -> 68[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32709:32882 [5] NCCL INFO Channel 13/0 : 29[5] -> 61[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153569:153739 [7] NCCL INFO Channel 03/0 : 95[7] -> 88[0] via P2P/IPC
+ip-26-0-161-78:114518:114693 [6] NCCL INFO Channel 00/0 : 70[6] -> 69[5] via P2P/IPC
+ip-26-0-168-52:2722052:2722226 [7] NCCL INFO Channel 05/0 : 127[7] -> 120[0] via P2P/IPC
+ip-26-0-161-138:96566:96755 [3] NCCL INFO Channel 03/0 : 67[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32711:32884 [7] NCCL INFO Channel 07/0 : 31[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153568:153736 [6] NCCL INFO Channel 05/0 : 94[6] -> 93[5] via P2P/IPC
+ip-26-0-161-78:114516:114691 [4] NCCL INFO Channel 04/0 : 68[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO Channel 07/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272156:272325 [6] NCCL INFO Channel 13/0 : 118[6] -> 117[5] via P2P/IPC
+ip-26-0-160-225:124166:124339 [6] NCCL INFO Channel 13/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-78:114515:114690 [3] NCCL INFO Channel 03/0 : 3[3] -> 67[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO Channel 15/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96568:96753 [5] NCCL INFO Channel 05/0 : 53[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32711:32884 [7] NCCL INFO Channel 01/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-161-78:114517:114687 [5] NCCL INFO Channel 13/0 : 77[5] -> 69[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114515:114690 [3] NCCL INFO Channel 03/0 : 67[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114519:114692 [7] NCCL INFO Channel 04/0 : 71[7] -> 64[0] via P2P/IPC
+ip-26-0-168-52:2722052:2722226 [7] NCCL INFO Channel 06/0 : 127[7] -> 120[0] via P2P/IPC
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO Channel 15/0 : 15[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32709:32882 [5] NCCL INFO Channel 13/0 : 61[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32708:32879 [4] NCCL INFO Channel 12/0 : 28[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96568:96753 [5] NCCL INFO Channel 05/0 : 37[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114518:114693 [6] NCCL INFO Channel 05/0 : 70[6] -> 69[5] via P2P/IPC
+ip-26-0-168-30:235655:235828 [7] NCCL INFO Channel 07/0 : 119[7] -> 111[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO Channel 01/0 : 87[7] -> 80[0] via P2P/IPC
+ip-26-0-161-138:96567:96760 [4] NCCL INFO Channel 04/0 : 52[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114516:114691 [4] NCCL INFO Channel 12/0 : 76[4] -> 68[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO Channel 00/0 : 40[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235655:235828 [7] NCCL INFO Channel 15/0 : 119[7] -> 111[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722052:2722226 [7] NCCL INFO Channel 07/0 : 127[7] -> 120[0] via P2P/IPC
+ip-26-0-161-123:32711:32884 [7] NCCL INFO Channel 02/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-160-225:124163:124340 [3] NCCL INFO Channel 11/0 : 11[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96567:96760 [4] NCCL INFO Channel 04/0 : 36[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114515:114690 [3] NCCL INFO Channel 03/0 : 99[3] -> 67[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235655:235828 [7] NCCL INFO Channel 15/0 : 111[7] -> 103[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96895:97069 [7] NCCL INFO Channel 01/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO Channel 08/0 : 40[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32708:32879 [4] NCCL INFO Channel 12/0 : 60[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114519:114692 [7] NCCL INFO Channel 05/0 : 71[7] -> 64[0] via P2P/IPC
+ip-26-0-161-78:114515:114690 [3] NCCL INFO Channel 03/0 : 67[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114518:114693 [6] NCCL INFO Channel 08/0 : 70[6] -> 69[5] via P2P/IPC
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO Channel 02/0 : 82[2] -> 98[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO Channel 01/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO Channel 05/0 : 61[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124167:124338 [7] NCCL INFO Channel 01/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-168-30:235650:235824 [2] NCCL INFO Channel 10/0 : 90[2] -> 106[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO Channel 02/0 : 98[2] -> 114[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96568:96753 [5] NCCL INFO Channel 13/0 : 45[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO Channel 00/0 : 48[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO Channel 05/0 : 53[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-52:2722052:2722226 [7] NCCL INFO Channel 09/0 : 127[7] -> 120[0] via P2P/IPC
+ip-26-0-161-103:96893:97067 [5] NCCL INFO Channel 05/0 : 29[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32711:32884 [7] NCCL INFO Channel 03/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO Channel 13/0 : 53[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96895:97069 [7] NCCL INFO Channel 02/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-165-202:153569:153739 [7] NCCL INFO Channel 04/0 : 95[7] -> 88[0] via P2P/IPC
+ip-26-0-161-78:114519:114692 [7] NCCL INFO Channel 06/0 : 71[7] -> 64[0] via P2P/IPC
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO Channel 04/0 : 60[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235655:235828 [7] NCCL INFO Channel 01/0 : 111[7] -> 104[0] via P2P/IPC
+ip-26-0-161-103:96893:97067 [5] NCCL INFO Channel 05/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO Channel 02/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-161-78:114518:114693 [6] NCCL INFO Channel 13/0 : 70[6] -> 69[5] via P2P/IPC
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO Channel 04/0 : 52[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96567:96760 [4] NCCL INFO Channel 12/0 : 44[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96893:97067 [5] NCCL INFO Channel 13/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124167:124338 [7] NCCL INFO Channel 02/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-168-52:2722052:2722226 [7] NCCL INFO Channel 10/0 : 127[7] -> 120[0] via P2P/IPC
+ip-26-0-168-30:235650:235824 [2] NCCL INFO Channel 10/0 : 106[2] -> 90[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO Channel 02/0 : 79[7] -> 72[0] via P2P/IPC
+ip-26-0-161-78:114515:114690 [3] NCCL INFO Channel 11/0 : 75[3] -> 67[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32711:32884 [7] NCCL INFO Channel 04/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO Channel 12/0 : 52[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96566:96755 [3] NCCL INFO Channel 03/0 : 51[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96892:97064 [4] NCCL INFO Channel 04/0 : 28[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO Channel 02/0 : 98[2] -> 82[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96895:97069 [7] NCCL INFO Channel 03/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-165-202:153569:153739 [7] NCCL INFO Channel 05/0 : 95[7] -> 88[0] via P2P/IPC
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO Channel 00/0 : 32[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96566:96755 [3] NCCL INFO Channel 03/0 : 35[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO Channel 08/0 : 24[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO Channel 03/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-161-103:96892:97064 [4] NCCL INFO Channel 04/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235655:235828 [7] NCCL INFO Channel 02/0 : 111[7] -> 104[0] via P2P/IPC
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO Channel 12/0 : 124[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722052:2722226 [7] NCCL INFO Channel 11/0 : 127[7] -> 120[0] via P2P/IPC
+ip-26-0-161-103:96892:97064 [4] NCCL INFO Channel 12/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32711:32884 [7] NCCL INFO Channel 05/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO Channel 03/0 : 79[7] -> 72[0] via P2P/IPC
+ip-26-0-160-225:124167:124338 [7] NCCL INFO Channel 03/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO Channel 01/0 : 103[7] -> 96[0] via P2P/IPC
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO Channel 08/0 : 56[0] -> 88[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO Channel 12/0 : 60[4] -> 124[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96895:97069 [7] NCCL INFO Channel 04/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO Channel 00/0 : 48[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153569:153739 [7] NCCL INFO Channel 06/0 : 95[7] -> 88[0] via P2P/IPC
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO Channel 03/0 : 63[7] -> 56[0] via P2P/IPC
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO Channel 04/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO Channel 03/0 : 59[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-52:2722052:2722226 [7] NCCL INFO Channel 12/0 : 127[7] -> 120[0] via P2P/IPC
+ip-26-0-161-142:62888:63060 [7] NCCL INFO Channel 15/0 : 55[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235655:235828 [7] NCCL INFO Channel 03/0 : 111[7] -> 104[0] via P2P/IPC
+ip-26-0-165-202:153568:153736 [6] NCCL INFO Channel 08/0 : 94[6] -> 93[5] via P2P/IPC
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO Channel 04/0 : 79[7] -> 72[0] via P2P/IPC
+ip-26-0-161-138:96566:96755 [3] NCCL INFO Channel 11/0 : 43[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96895:97069 [7] NCCL INFO Channel 05/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO Channel 04/0 : 63[7] -> 56[0] via P2P/IPC
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO Channel 03/0 : 51[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:124167:124338 [7] NCCL INFO Channel 04/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-142:62888:63060 [7] NCCL INFO Channel 15/0 : 47[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32711:32884 [7] NCCL INFO Channel 06/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-161-103:96891:97063 [3] NCCL INFO Channel 03/0 : 27[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO Channel 12/0 : 92[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO Channel 08/0 : 86[6] -> 85[5] via P2P/IPC
+ip-26-0-165-202:153564:153737 [2] NCCL INFO Channel 10/0 : 58[2] -> 90[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO Channel 11/0 : 51[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO Channel 05/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-161-103:96891:97063 [3] NCCL INFO Channel 03/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235654:235821 [6] NCCL INFO Channel 00/0 : 110[6] -> 109[5] via P2P/IPC
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO Channel 12/0 : 60[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722052:2722226 [7] NCCL INFO Channel 13/0 : 127[7] -> 120[0] via P2P/IPC
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO Channel 04/0 : 124[4] -> 116[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62881:63059 [0] NCCL INFO Channel 08/0 : 24[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO Channel 02/0 : 103[7] -> 96[0] via P2P/IPC
+ip-26-0-165-202:153569:153739 [7] NCCL INFO Channel 07/0 : 95[7] -> 88[0] via P2P/IPC
+ip-26-0-161-103:96891:97063 [3] NCCL INFO Channel 11/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO Channel 02/0 : 87[7] -> 80[0] via P2P/IPC
+ip-26-0-168-30:235655:235828 [7] NCCL INFO Channel 04/0 : 111[7] -> 104[0] via P2P/IPC
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO Channel 05/0 : 63[7] -> 56[0] via P2P/IPC
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO Channel 05/0 : 79[7] -> 72[0] via P2P/IPC
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO Channel 03/0 : 115[3] -> 99[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32711:32884 [7] NCCL INFO Channel 07/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO Channel 01/0 : 55[7] -> 48[0] via P2P/IPC
+ip-26-0-160-225:124167:124338 [7] NCCL INFO Channel 05/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-202:153568:153736 [6] NCCL INFO Channel 13/0 : 94[6] -> 93[5] via P2P/IPC
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO Channel 03/0 : 99[3] -> 83[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96895:97069 [7] NCCL INFO Channel 06/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO Channel 13/0 : 86[6] -> 85[5] via P2P/IPC
+ip-26-0-161-142:62888:63060 [7] NCCL INFO Channel 01/0 : 47[7] -> 40[0] via P2P/IPC
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO Channel 06/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-165-202:153564:153737 [2] NCCL INFO Channel 10/0 : 90[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO Channel 05/0 : 117[5] -> 101[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-52:2722052:2722226 [7] NCCL INFO Channel 14/0 : 127[7] -> 120[0] via P2P/IPC
+ip-26-0-161-138:96563:96756 [0] NCCL INFO Channel 00/0 : 32[0] -> 64[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62881:63059 [0] NCCL INFO Channel 08/0 : 40[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO Channel 05/0 : 101[5] -> 85[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153569:153739 [7] NCCL INFO Channel 09/0 : 95[7] -> 88[0] via P2P/IPC
+ip-26-0-168-30:235654:235821 [6] NCCL INFO Channel 05/0 : 110[6] -> 109[5] via P2P/IPC
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO Channel 06/0 : 63[7] -> 56[0] via P2P/IPC
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO Channel 06/0 : 79[7] -> 72[0] via P2P/IPC
+ip-26-0-161-103:96895:97069 [7] NCCL INFO Channel 07/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-161-123:32711:32884 [7] NCCL INFO Channel 09/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO Channel 03/0 : 87[7] -> 80[0] via P2P/IPC
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO Channel 03/0 : 103[7] -> 96[0] via P2P/IPC
+ip-26-0-165-202:153566:153741 [4] NCCL INFO Channel 12/0 : 108[4] -> 92[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235655:235828 [7] NCCL INFO Channel 05/0 : 111[7] -> 104[0] via P2P/IPC
+ip-26-0-165-202:153566:153741 [4] NCCL INFO Channel 12/0 : 92[4] -> 76[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO Channel 07/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO Channel 02/0 : 55[7] -> 48[0] via P2P/IPC
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO Channel 04/0 : 60[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32704:32881 [0] NCCL INFO Channel 08/0 : 24[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-52:2722052:2722226 [7] NCCL INFO Channel 15/0 : 127[7] -> 120[0] via P2P/IPC
+ip-26-0-161-142:62888:63060 [7] NCCL INFO Channel 02/0 : 47[7] -> 40[0] via P2P/IPC
+ip-26-0-165-202:153569:153739 [7] NCCL INFO Channel 10/0 : 95[7] -> 88[0] via P2P/IPC
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO Channel 11/0 : 107[3] -> 99[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272152:272331 [2] NCCL INFO Channel 02/0 : 98[2] -> 114[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO Channel 10/0 : 122[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32708:32879 [4] NCCL INFO Channel 12/0 : 44[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32707:32878 [3] NCCL INFO Channel 11/0 : 59[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO Channel 07/0 : 79[7] -> 72[0] via P2P/IPC
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO Channel 03/0 : 91[3] -> 83[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272155:272329 [5] NCCL INFO Channel 05/0 : 125[5] -> 117[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96895:97069 [7] NCCL INFO Channel 09/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO Channel 13/0 : 125[5] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235654:235821 [6] NCCL INFO Channel 08/0 : 110[6] -> 109[5] via P2P/IPC
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO Channel 07/0 : 63[7] -> 56[0] via P2P/IPC
+ip-26-0-161-123:32708:32879 [4] NCCL INFO Channel 12/0 : 28[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32711:32884 [7] NCCL INFO Channel 10/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO Channel 04/0 : 87[7] -> 80[0] via P2P/IPC
+ip-26-0-168-34:272155:272329 [5] NCCL INFO Channel 05/0 : 117[5] -> 109[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO Channel 09/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-165-202:153565:153738 [3] NCCL INFO Channel 11/0 : 91[3] -> 59[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62888:63060 [7] NCCL INFO Channel 03/0 : 47[7] -> 40[0] via P2P/IPC
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO Channel 04/0 : 103[7] -> 96[0] via P2P/IPC
+ip-26-0-168-30:235652:235822 [4] NCCL INFO Channel 04/0 : 116[4] -> 108[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO Channel 11/0 : 123[3] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO Channel 10/0 : 58[2] -> 122[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32704:32881 [0] NCCL INFO Channel 08/0 : 56[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO Channel 03/0 : 83[3] -> 75[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO Channel 03/0 : 55[7] -> 48[0] via P2P/IPC
+ip-26-0-161-78:114519:114692 [7] NCCL INFO Channel 07/0 : 71[7] -> 64[0] via P2P/IPC
+ip-26-0-168-34:272155:272329 [5] NCCL INFO Channel 13/0 : 117[5] -> 109[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO Channel 13/0 : 61[5] -> 125[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153569:153739 [7] NCCL INFO Channel 11/0 : 95[7] -> 88[0] via P2P/IPC
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO Channel 13/0 : 109[5] -> 101[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235655:235828 [7] NCCL INFO Channel 06/0 : 111[7] -> 104[0] via P2P/IPC
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO Channel 09/0 : 79[7] -> 72[0] via P2P/IPC
+ip-26-0-161-103:96895:97069 [7] NCCL INFO Channel 10/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO Channel 05/0 : 93[5] -> 85[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62885:63054 [4] NCCL INFO Channel 04/0 : 52[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272152:272331 [2] NCCL INFO Channel 02/0 : 114[2] -> 98[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO Channel 11/0 : 59[3] -> 123[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO Channel 11/0 : 83[3] -> 75[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO Channel 02/0 : 66[2] -> 98[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235652:235822 [4] NCCL INFO Channel 12/0 : 116[4] -> 108[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO Channel 08/0 : 120[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32711:32884 [7] NCCL INFO Channel 11/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-165-202:153566:153741 [4] NCCL INFO Channel 04/0 : 92[4] -> 84[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO Channel 00/0 : 102[6] -> 101[5] via P2P/IPC
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO Channel 09/0 : 63[7] -> 56[0] via P2P/IPC
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO Channel 08/0 : 56[0] -> 120[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO Channel 10/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-165-202:153562:153735 [0] NCCL INFO Channel 08/0 : 88[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62888:63060 [7] NCCL INFO Channel 04/0 : 47[7] -> 40[0] via P2P/IPC
+ip-26-0-168-30:235652:235822 [4] NCCL INFO Channel 12/0 : 108[4] -> 100[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235654:235821 [6] NCCL INFO Channel 13/0 : 110[6] -> 109[5] via P2P/IPC
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO Channel 05/0 : 87[7] -> 80[0] via P2P/IPC
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO Channel 05/0 : 85[5] -> 77[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO Channel 05/0 : 103[7] -> 96[0] via P2P/IPC
+ip-26-0-161-123:32708:32879 [4] NCCL INFO Channel 04/0 : 28[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114519:114692 [7] NCCL INFO Channel 09/0 : 71[7] -> 64[0] via P2P/IPC
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO Channel 13/0 : 93[5] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO Channel 04/0 : 55[7] -> 48[0] via P2P/IPC
+ip-26-0-165-202:153569:153739 [7] NCCL INFO Channel 12/0 : 95[7] -> 88[0] via P2P/IPC
+ip-26-0-161-142:62885:63054 [4] NCCL INFO Channel 12/0 : 52[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235655:235828 [7] NCCL INFO Channel 07/0 : 111[7] -> 104[0] via P2P/IPC
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO Channel 10/0 : 90[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO Channel 04/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO Channel 13/0 : 85[5] -> 77[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO Channel 10/0 : 79[7] -> 72[0] via P2P/IPC
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO Channel 04/0 : 116[4] -> 100[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96895:97069 [7] NCCL INFO Channel 11/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-168-52:2722047:2722221 [2] NCCL INFO Channel 02/0 : 122[2] -> 114[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-52:2722050:2722225 [5] NCCL INFO Channel 05/0 : 125[5] -> 117[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62885:63054 [4] NCCL INFO Channel 12/0 : 44[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32711:32884 [7] NCCL INFO Channel 12/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO Channel 11/0 : 91[3] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO Channel 12/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO Channel 04/0 : 84[4] -> 76[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO Channel 02/0 : 98[2] -> 66[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO Channel 10/0 : 63[7] -> 56[0] via P2P/IPC
+ip-26-0-165-202:153569:153739 [7] NCCL INFO Channel 13/0 : 95[7] -> 88[0] via P2P/IPC
+ip-26-0-161-142:62888:63060 [7] NCCL INFO Channel 05/0 : 47[7] -> 40[0] via P2P/IPC
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO Channel 05/0 : 102[6] -> 101[5] via P2P/IPC
+ip-26-0-168-52:2722048:2722220 [3] NCCL INFO Channel 03/0 : 123[3] -> 115[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114519:114692 [7] NCCL INFO Channel 10/0 : 71[7] -> 64[0] via P2P/IPC
+ip-26-0-161-78:114514:114689 [2] NCCL INFO Channel 02/0 : 2[2] -> 66[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO Channel 04/0 : 100[4] -> 84[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO Channel 11/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO Channel 12/0 : 84[4] -> 76[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96563:96756 [0] NCCL INFO Channel 00/0 : 64[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO Channel 08/0 : 88[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO Channel 13/0 : 61[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO Channel 10/0 : 58[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-52:2722045:2722219 [0] NCCL INFO Channel 00/0 : 120[0] -> 112[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO Channel 06/0 : 103[7] -> 96[0] via P2P/IPC
+ip-26-0-168-30:235655:235828 [7] NCCL INFO Channel 09/0 : 111[7] -> 104[0] via P2P/IPC
+ip-26-0-161-103:96895:97069 [7] NCCL INFO Channel 12/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO Channel 11/0 : 59[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO Channel 12/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO Channel 05/0 : 55[7] -> 48[0] via P2P/IPC
+ip-26-0-161-78:114512:114686 [0] NCCL INFO Channel 00/0 : 0[0] -> 64[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO Channel 08/0 : 56[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32711:32884 [7] NCCL INFO Channel 13/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO Channel 12/0 : 76[4] -> 68[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO Channel 11/0 : 79[7] -> 72[0] via P2P/IPC
+ip-26-0-161-142:62888:63060 [7] NCCL INFO Channel 06/0 : 47[7] -> 40[0] via P2P/IPC
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO Channel 08/0 : 102[6] -> 101[5] via P2P/IPC
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO Channel 12/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-168-34:272154:272330 [4] NCCL INFO Channel 04/0 : 124[4] -> 116[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO Channel 11/0 : 63[7] -> 56[0] via P2P/IPC
+ip-26-0-165-202:153569:153739 [7] NCCL INFO Channel 14/0 : 95[7] -> 88[0] via P2P/IPC
+ip-26-0-161-78:114514:114689 [2] NCCL INFO Channel 02/0 : 66[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114512:114686 [0] NCCL INFO Channel 00/0 : 64[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153565:153738 [3] NCCL INFO Channel 11/0 : 107[3] -> 91[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272154:272330 [4] NCCL INFO Channel 04/0 : 116[4] -> 108[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153564:153737 [2] NCCL INFO Channel 10/0 : 106[2] -> 90[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153567:153742 [5] NCCL INFO Channel 13/0 : 109[5] -> 93[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235655:235828 [7] NCCL INFO Channel 10/0 : 111[7] -> 104[0] via P2P/IPC
+ip-26-0-161-103:96895:97069 [7] NCCL INFO Channel 13/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-161-78:114519:114692 [7] NCCL INFO Channel 11/0 : 71[7] -> 64[0] via P2P/IPC
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO Channel 12/0 : 108[4] -> 100[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO Channel 05/0 : 61[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO Channel 12/0 : 79[7] -> 72[0] via P2P/IPC
+ip-26-0-168-34:272153:272328 [3] NCCL INFO Channel 03/0 : 123[3] -> 115[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO Channel 04/0 : 92[4] -> 84[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:124162:124342 [2] NCCL INFO Channel 10/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153562:153735 [0] NCCL INFO Channel 08/0 : 104[0] -> 88[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO Channel 07/0 : 103[7] -> 96[0] via P2P/IPC
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO Channel 13/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-161-123:32711:32884 [7] NCCL INFO Channel 14/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-161-78:114514:114689 [2] NCCL INFO Channel 02/0 : 98[2] -> 66[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272154:272330 [4] NCCL INFO Channel 12/0 : 116[4] -> 108[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO Channel 02/0 : 58[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO Channel 04/0 : 84[4] -> 76[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114512:114686 [0] NCCL INFO Channel 00/0 : 96[0] -> 64[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272153:272328 [3] NCCL INFO Channel 03/0 : 115[3] -> 107[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO Channel 03/0 : 59[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153567:153742 [5] NCCL INFO Channel 13/0 : 93[5] -> 77[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153564:153737 [2] NCCL INFO Channel 10/0 : 90[2] -> 74[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153565:153738 [3] NCCL INFO Channel 11/0 : 91[3] -> 75[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO Channel 13/0 : 102[6] -> 101[5] via P2P/IPC
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO Channel 00/0 : 56[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32706:32877 [2] NCCL INFO Channel 10/0 : 42[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32709:32882 [5] NCCL INFO Channel 13/0 : 45[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32707:32878 [3] NCCL INFO Channel 11/0 : 43[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO Channel 06/0 : 55[7] -> 48[0] via P2P/IPC
+ip-26-0-165-202:153562:153735 [0] NCCL INFO Channel 08/0 : 88[0] -> 72[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235655:235828 [7] NCCL INFO Channel 11/0 : 111[7] -> 104[0] via P2P/IPC
+ip-26-0-161-123:32704:32881 [0] NCCL INFO Channel 08/0 : 40[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153569:153739 [7] NCCL INFO Channel 15/0 : 95[7] -> 88[0] via P2P/IPC
+ip-26-0-161-78:114514:114689 [2] NCCL INFO Channel 02/0 : 66[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62888:63060 [7] NCCL INFO Channel 07/0 : 47[7] -> 40[0] via P2P/IPC
+ip-26-0-168-34:272153:272328 [3] NCCL INFO Channel 11/0 : 115[3] -> 107[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO Channel 12/0 : 84[4] -> 76[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114519:114692 [7] NCCL INFO Channel 12/0 : 71[7] -> 64[0] via P2P/IPC
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO Channel 12/0 : 63[7] -> 56[0] via P2P/IPC
+ip-26-0-161-78:114512:114686 [0] NCCL INFO Channel 00/0 : 64[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96895:97069 [7] NCCL INFO Channel 14/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-161-123:32709:32882 [5] NCCL INFO Channel 13/0 : 29[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32706:32877 [2] NCCL INFO Channel 10/0 : 26[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32707:32878 [3] NCCL INFO Channel 11/0 : 27[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO Channel 09/0 : 103[7] -> 96[0] via P2P/IPC
+ip-26-0-161-123:32704:32881 [0] NCCL INFO Channel 08/0 : 24[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO Channel 13/0 : 79[7] -> 72[0] via P2P/IPC
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO Channel 00/0 : 60[4] -> 59[3] via P2P/IPC
+ip-26-0-168-30:235650:235824 [2] NCCL INFO Channel 02/0 : 114[2] -> 106[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235651:235823 [3] NCCL INFO Channel 03/0 : 115[3] -> 107[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO Channel 02/0 : 114[2] -> 98[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32711:32884 [7] NCCL INFO Channel 15/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-168-30:235655:235828 [7] NCCL INFO Channel 12/0 : 111[7] -> 104[0] via P2P/IPC
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO Channel 14/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO Channel 00/0 : 112[0] -> 96[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114519:114692 [7] NCCL INFO Channel 13/0 : 71[7] -> 64[0] via P2P/IPC
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO Channel 13/0 : 63[7] -> 56[0] via P2P/IPC
+ip-26-0-165-202:153564:153737 [2] NCCL INFO Channel 02/0 : 90[2] -> 82[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235651:235823 [3] NCCL INFO Channel 11/0 : 115[3] -> 107[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235650:235824 [2] NCCL INFO Channel 10/0 : 114[2] -> 106[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO Channel 07/0 : 55[7] -> 48[0] via P2P/IPC
+ip-26-0-161-142:62883:63058 [2] NCCL INFO Channel 02/0 : 50[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO Channel 02/0 : 82[2] -> 74[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO Channel 03/0 : 83[3] -> 75[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96563:96756 [0] NCCL INFO Channel 00/0 : 48[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114514:114689 [2] NCCL INFO Channel 10/0 : 74[2] -> 66[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114512:114686 [0] NCCL INFO Channel 08/0 : 72[0] -> 64[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO Channel 02/0 : 98[2] -> 82[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153562:153735 [0] NCCL INFO Channel 00/0 : 88[0] -> 80[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO Channel 00/0 : 96[0] -> 80[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235651:235823 [3] NCCL INFO Channel 11/0 : 107[3] -> 99[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235650:235824 [2] NCCL INFO Channel 10/0 : 106[2] -> 98[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96895:97069 [7] NCCL INFO Channel 15/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-165-202:153565:153738 [3] NCCL INFO Channel 03/0 : 91[3] -> 83[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62884:63061 [3] NCCL INFO Channel 03/0 : 51[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32706:32877 [2] NCCL INFO Channel 02/0 : 26[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32704:32881 [0] NCCL INFO Channel 00/0 : 24[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO Channel 00/0 : 80[0] -> 72[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:124160:124335 [0] NCCL INFO Channel 08/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96565:96754 [2] NCCL INFO Channel 02/0 : 50[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62881:63059 [0] NCCL INFO Channel 00/0 : 48[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO Channel 02/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO Channel 10/0 : 82[2] -> 74[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62886:63057 [5] NCCL INFO Channel 05/0 : 53[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32709:32882 [5] NCCL INFO Channel 05/0 : 29[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62888:63060 [7] NCCL INFO Channel 09/0 : 47[7] -> 40[0] via P2P/IPC
+ip-26-0-161-123:32707:32878 [3] NCCL INFO Channel 03/0 : 27[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272157:272324 [7] NCCL INFO Channel 01/0 : 119[7] -> 112[0] via P2P/IPC
+ip-26-0-161-138:96563:96756 [0] NCCL INFO Channel 00/0 : 32[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO Channel 10/0 : 103[7] -> 96[0] via P2P/IPC
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO Channel 00/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO Channel 03/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO Channel 05/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO Channel 11/0 : 83[3] -> 75[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96565:96754 [2] NCCL INFO Channel 02/0 : 34[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO Channel 10/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32708:32879 [4] NCCL INFO Channel 00/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO Channel 14/0 : 79[7] -> 72[0] via P2P/IPC
+ip-26-0-161-142:62883:63058 [2] NCCL INFO Channel 10/0 : 50[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62881:63059 [0] NCCL INFO Channel 08/0 : 48[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235655:235828 [7] NCCL INFO Channel 13/0 : 111[7] -> 104[0] via P2P/IPC
+ip-26-0-168-34:272152:272331 [2] NCCL INFO Channel 02/0 : 122[2] -> 114[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO Channel 10/0 : 74[2] -> 66[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62886:63057 [5] NCCL INFO Channel 13/0 : 53[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62884:63061 [3] NCCL INFO Channel 11/0 : 51[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO Channel 08/0 : 80[0] -> 72[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO Channel 11/0 : 75[3] -> 67[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62883:63058 [2] NCCL INFO Channel 10/0 : 42[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62881:63059 [0] NCCL INFO Channel 08/0 : 40[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO Channel 10/0 : 106[2] -> 98[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO Channel 15/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-161-78:114519:114692 [7] NCCL INFO Channel 14/0 : 71[7] -> 64[0] via P2P/IPC
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO Channel 11/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO Channel 09/0 : 55[7] -> 48[0] via P2P/IPC
+ip-26-0-165-202:153567:153742 [5] NCCL INFO Channel 05/0 : 93[5] -> 85[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235653:235825 [5] NCCL INFO Channel 05/0 : 117[5] -> 109[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272150:272327 [0] NCCL INFO Channel 00/0 : 120[0] -> 112[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO Channel 02/0 : 90[2] -> 82[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO Channel 08/0 : 72[0] -> 64[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62886:63057 [5] NCCL INFO Channel 13/0 : 45[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62884:63061 [3] NCCL INFO Channel 11/0 : 43[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235648:235826 [0] NCCL INFO Channel 00/0 : 112[0] -> 104[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272152:272331 [2] NCCL INFO Channel 02/0 : 114[2] -> 106[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO Channel 03/0 : 60[4] -> 59[3] via P2P/IPC
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO Channel 11/0 : 103[7] -> 96[0] via P2P/IPC
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO Channel 13/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO Channel 10/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO Channel 08/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO Channel 02/0 : 82[2] -> 74[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO Channel 05/0 : 85[5] -> 77[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO Channel 00/0 : 56[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO Channel 02/0 : 58[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235653:235825 [5] NCCL INFO Channel 13/0 : 117[5] -> 109[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272152:272331 [2] NCCL INFO Channel 10/0 : 114[2] -> 106[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272150:272327 [0] NCCL INFO Channel 00/0 : 112[0] -> 104[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO Channel 11/0 : 11[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96563:96756 [0] NCCL INFO Channel 08/0 : 40[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235648:235826 [0] NCCL INFO Channel 08/0 : 112[0] -> 104[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272150:272327 [0] NCCL INFO Channel 08/0 : 112[0] -> 104[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO Channel 14/0 : 63[7] -> 56[0] via P2P/IPC
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO Channel 13/0 : 13[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32708:32879 [4] NCCL INFO Channel 03/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-168-30:235653:235825 [5] NCCL INFO Channel 13/0 : 109[5] -> 101[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO Channel 10/0 : 82[2] -> 74[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO Channel 13/0 : 85[5] -> 77[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96888:97066 [0] NCCL INFO Channel 00/0 : 24[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO Channel 02/0 : 50[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO Channel 00/0 : 48[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96565:96754 [2] NCCL INFO Channel 10/0 : 42[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272157:272324 [7] NCCL INFO Channel 02/0 : 119[7] -> 112[0] via P2P/IPC
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO Channel 08/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235648:235826 [0] NCCL INFO Channel 08/0 : 104[0] -> 96[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO Channel 13/0 : 77[5] -> 69[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO Channel 15/0 : 79[7] -> 72[0] via P2P/IPC
+ip-26-0-161-142:62885:63054 [4] NCCL INFO Channel 00/0 : 44[4] -> 43[3] via P2P/IPC
+ip-26-0-168-30:235655:235828 [7] NCCL INFO Channel 14/0 : 111[7] -> 104[0] via P2P/IPC
+ip-26-0-161-103:96890:97070 [2] NCCL INFO Channel 02/0 : 26[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO Channel 10/0 : 50[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:124167:124338 [7] NCCL INFO Channel 06/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-103:96892:97064 [4] NCCL INFO Channel 00/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-161-103:96888:97066 [0] NCCL INFO Channel 00/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO Channel 08/0 : 48[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO Channel 12/0 : 103[7] -> 96[0] via P2P/IPC
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO Channel 00/0 : 52[4] -> 51[3] via P2P/IPC
+ip-26-0-161-78:114519:114692 [7] NCCL INFO Channel 15/0 : 71[7] -> 64[0] via P2P/IPC
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO Channel 10/0 : 55[7] -> 48[0] via P2P/IPC
+ip-26-0-161-103:96890:97070 [2] NCCL INFO Channel 02/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96888:97066 [0] NCCL INFO Channel 08/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235655:235828 [7] NCCL INFO Channel 15/0 : 111[7] -> 104[0] via P2P/IPC
+ip-26-0-161-103:96890:97070 [2] NCCL INFO Channel 10/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32708:32879 [4] NCCL INFO Channel 05/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO Channel 08/0 : 104[0] -> 96[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO Channel 00/0 : 78[6] -> 77[5] via P2P/IPC
+ip-26-0-168-34:272157:272324 [7] NCCL INFO Channel 03/0 : 119[7] -> 112[0] via P2P/IPC
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO Channel 05/0 : 60[4] -> 59[3] via P2P/IPC
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO Channel 00/0 : 88[0] -> 80[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO Channel 00/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-161-142:62885:63054 [4] NCCL INFO Channel 03/0 : 44[4] -> 43[3] via P2P/IPC
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO Channel 15/0 : 63[7] -> 56[0] via P2P/IPC
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO Channel 00/0 : 80[0] -> 72[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO Channel 08/0 : 80[0] -> 72[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO Channel 13/0 : 103[7] -> 96[0] via P2P/IPC
+ip-26-0-161-103:96892:97064 [4] NCCL INFO Channel 03/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO Channel 03/0 : 52[4] -> 51[3] via P2P/IPC
+ip-26-0-161-138:96570:96759 [7] NCCL INFO Channel 01/0 : 39[7] -> 32[0] via P2P/IPC
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO Channel 11/0 : 55[7] -> 48[0] via P2P/IPC
+ip-26-0-161-142:62888:63060 [7] NCCL INFO Channel 10/0 : 47[7] -> 40[0] via P2P/IPC
+ip-26-0-161-123:32708:32879 [4] NCCL INFO Channel 06/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO Channel 03/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-160-225:124167:124338 [7] NCCL INFO Channel 07/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-78:114516:114691 [4] NCCL INFO Channel 00/0 : 68[4] -> 67[3] via P2P/IPC
+ip-26-0-168-30:235652:235822 [4] NCCL INFO Channel 00/0 : 108[4] -> 107[3] via P2P/IPC
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO Channel 06/0 : 60[4] -> 59[3] via P2P/IPC
+ip-26-0-161-142:62885:63054 [4] NCCL INFO Channel 05/0 : 44[4] -> 43[3] via P2P/IPC
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO Channel 05/0 : 78[6] -> 77[5] via P2P/IPC
+ip-26-0-168-34:272157:272324 [7] NCCL INFO Channel 04/0 : 119[7] -> 112[0] via P2P/IPC
+ip-26-0-168-52:2722050:2722225 [5] NCCL INFO Channel 00/0 : 125[5] -> 124[4] via P2P/IPC
+ip-26-0-165-202:153566:153741 [4] NCCL INFO Channel 00/0 : 92[4] -> 91[3] via P2P/IPC
+ip-26-0-161-138:96570:96759 [7] NCCL INFO Channel 02/0 : 39[7] -> 32[0] via P2P/IPC
+ip-26-0-168-34:272154:272330 [4] NCCL INFO Channel 00/0 : 116[4] -> 115[3] via P2P/IPC
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO Channel 14/0 : 103[7] -> 96[0] via P2P/IPC
+ip-26-0-161-123:32708:32879 [4] NCCL INFO Channel 07/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO Channel 05/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-161-142:62888:63060 [7] NCCL INFO Channel 11/0 : 47[7] -> 40[0] via P2P/IPC
+ip-26-0-168-30:235652:235822 [4] NCCL INFO Channel 03/0 : 108[4] -> 107[3] via P2P/IPC
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO Channel 05/0 : 52[4] -> 51[3] via P2P/IPC
+ip-26-0-160-225:124167:124338 [7] NCCL INFO Channel 09/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO Channel 07/0 : 60[4] -> 59[3] via P2P/IPC
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO Channel 12/0 : 55[7] -> 48[0] via P2P/IPC
+ip-26-0-161-123:32709:32882 [5] NCCL INFO Channel 00/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO Channel 00/0 : 61[5] -> 60[4] via P2P/IPC
+ip-26-0-161-138:96570:96759 [7] NCCL INFO Channel 03/0 : 39[7] -> 32[0] via P2P/IPC
+ip-26-0-165-202:153566:153741 [4] NCCL INFO Channel 03/0 : 92[4] -> 91[3] via P2P/IPC
+ip-26-0-161-142:62885:63054 [4] NCCL INFO Channel 06/0 : 44[4] -> 43[3] via P2P/IPC
+ip-26-0-168-52:2722050:2722225 [5] NCCL INFO Channel 04/0 : 125[5] -> 124[4] via P2P/IPC
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO Channel 06/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-168-34:272157:272324 [7] NCCL INFO Channel 05/0 : 119[7] -> 112[0] via P2P/IPC
+ip-26-0-160-225:124164:124336 [4] NCCL INFO Channel 00/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-168-34:272154:272330 [4] NCCL INFO Channel 03/0 : 116[4] -> 115[3] via P2P/IPC
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO Channel 15/0 : 103[7] -> 96[0] via P2P/IPC
+ip-26-0-161-142:62888:63060 [7] NCCL INFO Channel 12/0 : 47[7] -> 40[0] via P2P/IPC
+ip-26-0-168-30:235652:235822 [4] NCCL INFO Channel 05/0 : 108[4] -> 107[3] via P2P/IPC
+ip-26-0-161-123:32709:32882 [5] NCCL INFO Channel 04/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-161-138:96570:96759 [7] NCCL INFO Channel 04/0 : 39[7] -> 32[0] via P2P/IPC
+ip-26-0-161-103:96892:97064 [4] NCCL INFO Channel 05/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO Channel 08/0 : 60[4] -> 59[3] via P2P/IPC
+ip-26-0-161-138:96567:96760 [4] NCCL INFO Channel 00/0 : 36[4] -> 35[3] via P2P/IPC
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO Channel 07/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO Channel 04/0 : 61[5] -> 60[4] via P2P/IPC
+ip-26-0-161-123:32709:32882 [5] NCCL INFO Channel 08/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-165-202:153566:153741 [4] NCCL INFO Channel 05/0 : 92[4] -> 91[3] via P2P/IPC
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO Channel 00/0 : 98[2] -> 97[1] via P2P/IPC
+ip-26-0-160-225:124167:124338 [7] NCCL INFO Channel 10/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO Channel 06/0 : 52[4] -> 51[3] via P2P/IPC
+ip-26-0-168-52:2722050:2722225 [5] NCCL INFO Channel 08/0 : 125[5] -> 124[4] via P2P/IPC
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO Channel 00/0 : 100[4] -> 99[3] via P2P/IPC
+ip-26-0-168-30:235653:235825 [5] NCCL INFO Channel 00/0 : 109[5] -> 108[4] via P2P/IPC
+ip-26-0-161-142:62886:63057 [5] NCCL INFO Channel 00/0 : 45[5] -> 44[4] via P2P/IPC
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO Channel 13/0 : 55[7] -> 48[0] via P2P/IPC
+ip-26-0-161-123:32709:32882 [5] NCCL INFO Channel 12/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-161-78:114516:114691 [4] NCCL INFO Channel 03/0 : 68[4] -> 67[3] via P2P/IPC
+ip-26-0-168-34:272157:272324 [7] NCCL INFO Channel 06/0 : 119[7] -> 112[0] via P2P/IPC
+ip-26-0-161-142:62885:63054 [4] NCCL INFO Channel 07/0 : 44[4] -> 43[3] via P2P/IPC
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO Channel 00/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-165-202:153564:153737 [2] NCCL INFO Channel 00/0 : 90[2] -> 89[1] via P2P/IPC
+ip-26-0-168-52:2722047:2722221 [2] NCCL INFO Channel 00/0 : 122[2] -> 121[1] via P2P/IPC
+ip-26-0-161-138:96570:96759 [7] NCCL INFO Channel 05/0 : 39[7] -> 32[0] via P2P/IPC
+ip-26-0-168-34:272154:272330 [4] NCCL INFO Channel 05/0 : 116[4] -> 115[3] via P2P/IPC
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO Channel 08/0 : 78[6] -> 77[5] via P2P/IPC
+ip-26-0-168-30:235652:235822 [4] NCCL INFO Channel 06/0 : 108[4] -> 107[3] via P2P/IPC
+ip-26-0-161-138:96567:96760 [4] NCCL INFO Channel 03/0 : 36[4] -> 35[3] via P2P/IPC
+ip-26-0-165-202:153567:153742 [5] NCCL INFO Channel 00/0 : 93[5] -> 92[4] via P2P/IPC
+ip-26-0-161-103:96892:97064 [4] NCCL INFO Channel 06/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-160-225:124164:124336 [4] NCCL INFO Channel 03/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO Channel 08/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-161-142:62888:63060 [7] NCCL INFO Channel 13/0 : 47[7] -> 40[0] via P2P/IPC
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO Channel 11/0 : 60[4] -> 59[3] via P2P/IPC
+ip-26-0-165-202:153566:153741 [4] NCCL INFO Channel 06/0 : 92[4] -> 91[3] via P2P/IPC
+ip-26-0-168-34:272155:272329 [5] NCCL INFO Channel 00/0 : 117[5] -> 116[4] via P2P/IPC
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO Channel 00/0 : 58[2] -> 57[1] via P2P/IPC
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO Channel 08/0 : 61[5] -> 60[4] via P2P/IPC
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO Channel 00/0 : 53[5] -> 52[4] via P2P/IPC
+ip-26-0-161-138:96568:96753 [5] NCCL INFO Channel 00/0 : 37[5] -> 36[4] via P2P/IPC
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO Channel 00/0 : 124[4] -> 123[3] via P2P/IPC
+ip-26-0-161-78:114516:114691 [4] NCCL INFO Channel 05/0 : 68[4] -> 67[3] via P2P/IPC
+ip-26-0-168-30:235653:235825 [5] NCCL INFO Channel 04/0 : 109[5] -> 108[4] via P2P/IPC
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO Channel 00/0 : 76[4] -> 75[3] via P2P/IPC
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO Channel 00/0 : 101[5] -> 100[4] via P2P/IPC
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO Channel 01/0 : 98[2] -> 97[1] via P2P/IPC
+ip-26-0-168-52:2722050:2722225 [5] NCCL INFO Channel 12/0 : 125[5] -> 124[4] via P2P/IPC
+ip-26-0-161-123:32708:32879 [4] NCCL INFO Channel 08/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-160-225:124162:124342 [2] NCCL INFO Channel 00/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO Channel 03/0 : 100[4] -> 99[3] via P2P/IPC
+ip-26-0-160-225:124167:124338 [7] NCCL INFO Channel 11/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO Channel 01/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO Channel 07/0 : 52[4] -> 51[3] via P2P/IPC
+ip-26-0-161-138:96570:96759 [7] NCCL INFO Channel 06/0 : 39[7] -> 32[0] via P2P/IPC
+ip-26-0-161-78:114514:114689 [2] NCCL INFO Channel 00/0 : 66[2] -> 65[1] via P2P/IPC
+ip-26-0-165-202:153564:153737 [2] NCCL INFO Channel 01/0 : 90[2] -> 89[1] via P2P/IPC
+ip-26-0-168-30:235650:235824 [2] NCCL INFO Channel 00/0 : 106[2] -> 105[1] via P2P/IPC
+ip-26-0-161-142:62883:63058 [2] NCCL INFO Channel 00/0 : 42[2] -> 41[1] via P2P/IPC
+ip-26-0-168-52:2722047:2722221 [2] NCCL INFO Channel 01/0 : 122[2] -> 121[1] via P2P/IPC
+ip-26-0-168-34:272157:272324 [7] NCCL INFO Channel 07/0 : 119[7] -> 112[0] via P2P/IPC
+ip-26-0-161-138:96567:96760 [4] NCCL INFO Channel 05/0 : 36[4] -> 35[3] via P2P/IPC
+ip-26-0-161-123:32708:32879 [4] NCCL INFO Channel 11/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-161-142:62886:63057 [5] NCCL INFO Channel 04/0 : 45[5] -> 44[4] via P2P/IPC
+ip-26-0-161-142:62885:63054 [4] NCCL INFO Channel 08/0 : 44[4] -> 43[3] via P2P/IPC
+ip-26-0-161-103:96892:97064 [4] NCCL INFO Channel 07/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-168-30:235652:235822 [4] NCCL INFO Channel 07/0 : 108[4] -> 107[3] via P2P/IPC
+ip-26-0-168-34:272152:272331 [2] NCCL INFO Channel 00/0 : 114[2] -> 113[1] via P2P/IPC
+ip-26-0-165-202:153567:153742 [5] NCCL INFO Channel 04/0 : 93[5] -> 92[4] via P2P/IPC
+ip-26-0-161-138:96568:96753 [5] NCCL INFO Channel 04/0 : 37[5] -> 36[4] via P2P/IPC
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO Channel 14/0 : 55[7] -> 48[0] via P2P/IPC
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO Channel 13/0 : 60[4] -> 59[3] via P2P/IPC
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO Channel 03/0 : 76[4] -> 75[3] via P2P/IPC
+ip-26-0-161-142:62888:63060 [7] NCCL INFO Channel 14/0 : 47[7] -> 40[0] via P2P/IPC
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO Channel 11/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-168-34:272154:272330 [4] NCCL INFO Channel 06/0 : 116[4] -> 115[3] via P2P/IPC
+ip-26-0-161-78:114516:114691 [4] NCCL INFO Channel 06/0 : 68[4] -> 67[3] via P2P/IPC
+ip-26-0-160-225:124164:124336 [4] NCCL INFO Channel 05/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-165-202:153566:153741 [4] NCCL INFO Channel 07/0 : 92[4] -> 91[3] via P2P/IPC
+ip-26-0-160-225:124165:124341 [5] NCCL INFO Channel 00/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO Channel 03/0 : 124[4] -> 123[3] via P2P/IPC
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO Channel 05/0 : 100[4] -> 99[3] via P2P/IPC
+ip-26-0-161-103:96893:97067 [5] NCCL INFO Channel 00/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO Channel 14/0 : 60[4] -> 59[3] via P2P/IPC
+ip-26-0-168-30:235653:235825 [5] NCCL INFO Channel 08/0 : 109[5] -> 108[4] via P2P/IPC
+ip-26-0-161-138:96570:96759 [7] NCCL INFO Channel 07/0 : 39[7] -> 32[0] via P2P/IPC
+ip-26-0-168-34:272155:272329 [5] NCCL INFO Channel 04/0 : 117[5] -> 116[4] via P2P/IPC
+ip-26-0-161-123:32708:32879 [4] NCCL INFO Channel 13/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-161-78:114514:114689 [2] NCCL INFO Channel 01/0 : 66[2] -> 65[1] via P2P/IPC
+ip-26-0-161-142:62888:63060 [7] NCCL INFO Channel 15/0 : 47[7] -> 40[0] via P2P/IPC
+ip-26-0-168-52:2722047:2722221 [2] NCCL INFO Channel 08/0 : 122[2] -> 121[1] via P2P/IPC
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO Channel 00/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-161-138:96567:96760 [4] NCCL INFO Channel 06/0 : 36[4] -> 35[3] via P2P/IPC
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO Channel 08/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO Channel 05/0 : 76[4] -> 75[3] via P2P/IPC
+ip-26-0-161-138:96568:96753 [5] NCCL INFO Channel 08/0 : 37[5] -> 36[4] via P2P/IPC
+ip-26-0-168-30:235650:235824 [2] NCCL INFO Channel 01/0 : 106[2] -> 105[1] via P2P/IPC
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO Channel 15/0 : 60[4] -> 59[3] via P2P/IPC
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO Channel 06/0 : 100[4] -> 99[3] via P2P/IPC
+ip-26-0-160-225:124162:124342 [2] NCCL INFO Channel 01/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-165-202:153564:153737 [2] NCCL INFO Channel 08/0 : 90[2] -> 89[1] via P2P/IPC
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO Channel 04/0 : 53[5] -> 52[4] via P2P/IPC
+ip-26-0-161-103:96890:97070 [2] NCCL INFO Channel 00/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-160-225:124167:124338 [7] NCCL INFO Channel 12/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-168-30:235652:235822 [4] NCCL INFO Channel 08/0 : 108[4] -> 107[3] via P2P/IPC
+ip-26-0-161-103:96892:97064 [4] NCCL INFO Channel 08/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-165-202:153567:153742 [5] NCCL INFO Channel 08/0 : 93[5] -> 92[4] via P2P/IPC
+ip-26-0-161-78:114516:114691 [4] NCCL INFO Channel 07/0 : 68[4] -> 67[3] via P2P/IPC
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO Channel 13/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO Channel 08/0 : 52[4] -> 51[3] via P2P/IPC
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO Channel 05/0 : 124[4] -> 123[3] via P2P/IPC
+ip-26-0-168-52:2722047:2722221 [2] NCCL INFO Channel 09/0 : 122[2] -> 121[1] via P2P/IPC
+ip-26-0-165-202:153566:153741 [4] NCCL INFO Channel 08/0 : 92[4] -> 91[3] via P2P/IPC
+ip-26-0-168-34:272154:272330 [4] NCCL INFO Channel 07/0 : 116[4] -> 115[3] via P2P/IPC
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO Channel 07/0 : 100[4] -> 99[3] via P2P/IPC
+ip-26-0-161-138:96570:96759 [7] NCCL INFO Channel 09/0 : 39[7] -> 32[0] via P2P/IPC
+ip-26-0-161-78:114514:114689 [2] NCCL INFO Channel 08/0 : 66[2] -> 65[1] via P2P/IPC
+ip-26-0-161-103:96893:97067 [5] NCCL INFO Channel 04/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-168-30:235653:235825 [5] NCCL INFO Channel 12/0 : 109[5] -> 108[4] via P2P/IPC
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO Channel 06/0 : 76[4] -> 75[3] via P2P/IPC
+ip-26-0-161-138:96567:96760 [4] NCCL INFO Channel 07/0 : 36[4] -> 35[3] via P2P/IPC
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO Channel 12/0 : 61[5] -> 60[4] via P2P/IPC
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO Channel 15/0 : 55[7] -> 48[0] via P2P/IPC
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO Channel 13/0 : 78[6] -> 77[5] via P2P/IPC
+ip-26-0-160-225:124164:124336 [4] NCCL INFO Channel 06/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-138:96568:96753 [5] NCCL INFO Channel 12/0 : 37[5] -> 36[4] via P2P/IPC
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO Channel 04/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-168-34:272154:272330 [4] NCCL INFO Channel 08/0 : 116[4] -> 115[3] via P2P/IPC
+ip-26-0-160-225:124165:124341 [5] NCCL INFO Channel 04/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO Channel 04/0 : 101[5] -> 100[4] via P2P/IPC
+ip-26-0-168-30:235650:235824 [2] NCCL INFO Channel 08/0 : 106[2] -> 105[1] via P2P/IPC
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO Channel 08/0 : 100[4] -> 99[3] via P2P/IPC
+ip-26-0-165-202:153564:153737 [2] NCCL INFO Channel 09/0 : 90[2] -> 89[1] via P2P/IPC
+ip-26-0-161-78:114516:114691 [4] NCCL INFO Channel 08/0 : 68[4] -> 67[3] via P2P/IPC
+ip-26-0-161-103:96890:97070 [2] NCCL INFO Channel 01/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-161-138:96565:96754 [2] NCCL INFO Channel 00/0 : 34[2] -> 33[1] via P2P/IPC
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO Channel 08/0 : 98[2] -> 97[1] via P2P/IPC
+ip-26-0-168-30:235652:235822 [4] NCCL INFO Channel 11/0 : 108[4] -> 107[3] via P2P/IPC
+ip-26-0-161-103:96892:97064 [4] NCCL INFO Channel 11/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO Channel 00/0 : 50[2] -> 49[1] via P2P/IPC
+ip-26-0-168-34:272154:272330 [4] NCCL INFO Channel 11/0 : 116[4] -> 115[3] via P2P/IPC
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO Channel 06/0 : 124[4] -> 123[3] via P2P/IPC
+ip-26-0-165-202:153567:153742 [5] NCCL INFO Channel 12/0 : 93[5] -> 92[4] via P2P/IPC
+ip-26-0-161-142:62883:63058 [2] NCCL INFO Channel 01/0 : 42[2] -> 41[1] via P2P/IPC
+ip-26-0-161-78:114514:114689 [2] NCCL INFO Channel 09/0 : 66[2] -> 65[1] via P2P/IPC
+ip-26-0-160-225:124162:124342 [2] NCCL INFO Channel 08/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-168-34:272152:272331 [2] NCCL INFO Channel 01/0 : 114[2] -> 113[1] via P2P/IPC
+ip-26-0-160-225:124167:124338 [7] NCCL INFO Channel 13/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO Channel 08/0 : 53[5] -> 52[4] via P2P/IPC
+ip-26-0-165-202:153566:153741 [4] NCCL INFO Channel 11/0 : 92[4] -> 91[3] via P2P/IPC
+ip-26-0-168-34:272154:272330 [4] NCCL INFO Channel 13/0 : 116[4] -> 115[3] via P2P/IPC
+ip-26-0-161-103:96893:97067 [5] NCCL INFO Channel 08/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO Channel 07/0 : 76[4] -> 75[3] via P2P/IPC
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO Channel 01/0 : 58[2] -> 57[1] via P2P/IPC
+ip-26-0-161-142:62883:63058 [2] NCCL INFO Channel 08/0 : 42[2] -> 41[1] via P2P/IPC
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO Channel 07/0 : 124[4] -> 123[3] via P2P/IPC
+ip-26-0-161-78:114516:114691 [4] NCCL INFO Channel 11/0 : 68[4] -> 67[3] via P2P/IPC
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO Channel 08/0 : 101[5] -> 100[4] via P2P/IPC
+ip-26-0-168-34:272152:272331 [2] NCCL INFO Channel 08/0 : 114[2] -> 113[1] via P2P/IPC
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO Channel 11/0 : 52[4] -> 51[3] via P2P/IPC
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO Channel 11/0 : 100[4] -> 99[3] via P2P/IPC
+ip-26-0-168-30:235650:235824 [2] NCCL INFO Channel 09/0 : 106[2] -> 105[1] via P2P/IPC
+ip-26-0-161-123:32706:32877 [2] NCCL INFO Channel 00/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-161-123:32708:32879 [4] NCCL INFO Channel 14/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-161-138:96565:96754 [2] NCCL INFO Channel 01/0 : 34[2] -> 33[1] via P2P/IPC
+ip-26-0-161-142:62884:63061 [3] NCCL INFO Channel 00/0 : 43[3] -> 42[2] via P2P/IPC
+ip-26-0-168-34:272154:272330 [4] NCCL INFO Channel 14/0 : 116[4] -> 115[3] via P2P/IPC
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO Channel 09/0 : 98[2] -> 97[1] via P2P/IPC
+ip-26-0-168-30:235652:235822 [4] NCCL INFO Channel 13/0 : 108[4] -> 107[3] via P2P/IPC
+ip-26-0-161-138:96567:96760 [4] NCCL INFO Channel 08/0 : 36[4] -> 35[3] via P2P/IPC
+ip-26-0-161-103:96890:97070 [2] NCCL INFO Channel 08/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-161-103:96892:97064 [4] NCCL INFO Channel 13/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-160-225:124164:124336 [4] NCCL INFO Channel 07/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO Channel 00/0 : 74[2] -> 73[1] via P2P/IPC
+ip-26-0-160-225:124165:124341 [5] NCCL INFO Channel 08/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO Channel 08/0 : 124[4] -> 123[3] via P2P/IPC
+ip-26-0-161-138:96570:96759 [7] NCCL INFO Channel 10/0 : 39[7] -> 32[0] via P2P/IPC
+ip-26-0-168-34:272152:272331 [2] NCCL INFO Channel 09/0 : 114[2] -> 113[1] via P2P/IPC
+ip-26-0-168-34:272154:272330 [4] NCCL INFO Channel 15/0 : 116[4] -> 115[3] via P2P/IPC
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO Channel 00/0 : 77[5] -> 76[4] via P2P/IPC
+ip-26-0-161-78:114516:114691 [4] NCCL INFO Channel 13/0 : 68[4] -> 67[3] via P2P/IPC
+ip-26-0-161-138:96565:96754 [2] NCCL INFO Channel 08/0 : 34[2] -> 33[1] via P2P/IPC
+ip-26-0-161-142:62885:63054 [4] NCCL INFO Channel 11/0 : 44[4] -> 43[3] via P2P/IPC
+ip-26-0-161-142:62883:63058 [2] NCCL INFO Channel 09/0 : 42[2] -> 41[1] via P2P/IPC
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO Channel 01/0 : 50[2] -> 49[1] via P2P/IPC
+ip-26-0-168-52:2722048:2722220 [3] NCCL INFO Channel 00/0 : 123[3] -> 122[2] via P2P/IPC
+ip-26-0-161-103:96893:97067 [5] NCCL INFO Channel 12/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO Channel 08/0 : 76[4] -> 75[3] via P2P/IPC
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO Channel 11/0 : 124[4] -> 123[3] via P2P/IPC
+ip-26-0-160-225:124162:124342 [2] NCCL INFO Channel 09/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-225:124167:124338 [7] NCCL INFO Channel 14/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO Channel 12/0 : 101[5] -> 100[4] via P2P/IPC
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO Channel 12/0 : 53[5] -> 52[4] via P2P/IPC
+ip-26-0-161-123:32708:32879 [4] NCCL INFO Channel 15/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-161-138:96567:96760 [4] NCCL INFO Channel 11/0 : 36[4] -> 35[3] via P2P/IPC
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO Channel 13/0 : 100[4] -> 99[3] via P2P/IPC
+ip-26-0-161-123:32707:32878 [3] NCCL INFO Channel 00/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-161-142:62884:63061 [3] NCCL INFO Channel 02/0 : 43[3] -> 42[2] via P2P/IPC
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO Channel 13/0 : 52[4] -> 51[3] via P2P/IPC
+ip-26-0-161-138:96570:96759 [7] NCCL INFO Channel 11/0 : 39[7] -> 32[0] via P2P/IPC
+ip-26-0-161-103:96890:97070 [2] NCCL INFO Channel 09/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-161-142:62886:63057 [5] NCCL INFO Channel 08/0 : 45[5] -> 44[4] via P2P/IPC
+ip-26-0-168-34:272157:272324 [7] NCCL INFO Channel 09/0 : 119[7] -> 112[0] via P2P/IPC
+ip-26-0-161-103:96892:97064 [4] NCCL INFO Channel 14/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-161-138:96565:96754 [2] NCCL INFO Channel 09/0 : 34[2] -> 33[1] via P2P/IPC
+ip-26-0-168-30:235651:235823 [3] NCCL INFO Channel 00/0 : 107[3] -> 106[2] via P2P/IPC
+ip-26-0-168-52:2722048:2722220 [3] NCCL INFO Channel 02/0 : 123[3] -> 122[2] via P2P/IPC
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO Channel 01/0 : 74[2] -> 73[1] via P2P/IPC
+ip-26-0-168-30:235652:235822 [4] NCCL INFO Channel 14/0 : 108[4] -> 107[3] via P2P/IPC
+ip-26-0-161-123:32707:32878 [3] NCCL INFO Channel 02/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-161-78:114516:114691 [4] NCCL INFO Channel 14/0 : 68[4] -> 67[3] via P2P/IPC
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO Channel 13/0 : 124[4] -> 123[3] via P2P/IPC
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO Channel 08/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO Channel 12/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO Channel 09/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO Channel 00/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-160-225:124164:124336 [4] NCCL INFO Channel 08/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-225:124165:124341 [5] NCCL INFO Channel 12/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO Channel 02/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO Channel 04/0 : 77[5] -> 76[4] via P2P/IPC
+ip-26-0-161-142:62885:63054 [4] NCCL INFO Channel 13/0 : 44[4] -> 43[3] via P2P/IPC
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO Channel 08/0 : 50[2] -> 49[1] via P2P/IPC
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO Channel 14/0 : 100[4] -> 99[3] via P2P/IPC
+ip-26-0-168-30:235651:235823 [3] NCCL INFO Channel 02/0 : 107[3] -> 106[2] via P2P/IPC
+ip-26-0-161-123:32707:32878 [3] NCCL INFO Channel 08/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-161-138:96567:96760 [4] NCCL INFO Channel 13/0 : 36[4] -> 35[3] via P2P/IPC
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO Channel 14/0 : 124[4] -> 123[3] via P2P/IPC
+ip-26-0-168-34:272155:272329 [5] NCCL INFO Channel 08/0 : 117[5] -> 116[4] via P2P/IPC
+ip-26-0-168-30:235652:235822 [4] NCCL INFO Channel 15/0 : 108[4] -> 107[3] via P2P/IPC
+ip-26-0-161-138:96570:96759 [7] NCCL INFO Channel 12/0 : 39[7] -> 32[0] via P2P/IPC
+ip-26-0-165-202:153565:153738 [3] NCCL INFO Channel 00/0 : 91[3] -> 90[2] via P2P/IPC
+ip-26-0-168-34:272157:272324 [7] NCCL INFO Channel 10/0 : 119[7] -> 112[0] via P2P/IPC
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO Channel 11/0 : 76[4] -> 75[3] via P2P/IPC
+ip-26-0-161-78:114515:114690 [3] NCCL INFO Channel 00/0 : 67[3] -> 66[2] via P2P/IPC
+ip-26-0-165-202:153566:153741 [4] NCCL INFO Channel 13/0 : 92[4] -> 91[3] via P2P/IPC
+ip-26-0-161-142:62884:63061 [3] NCCL INFO Channel 08/0 : 43[3] -> 42[2] via P2P/IPC
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO Channel 08/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO Channel 14/0 : 52[4] -> 51[3] via P2P/IPC
+ip-26-0-161-123:32707:32878 [3] NCCL INFO Channel 10/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO Channel 15/0 : 124[4] -> 123[3] via P2P/IPC
+ip-26-0-161-142:62886:63057 [5] NCCL INFO Channel 12/0 : 45[5] -> 44[4] via P2P/IPC
+ip-26-0-161-103:96892:97064 [4] NCCL INFO Channel 15/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-168-30:235651:235823 [3] NCCL INFO Channel 08/0 : 107[3] -> 106[2] via P2P/IPC
+ip-26-0-160-225:124167:124338 [7] NCCL INFO Channel 15/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO Channel 15/0 : 100[4] -> 99[3] via P2P/IPC
+ip-26-0-161-78:114516:114691 [4] NCCL INFO Channel 15/0 : 68[4] -> 67[3] via P2P/IPC
+ip-26-0-165-202:153565:153738 [3] NCCL INFO Channel 02/0 : 91[3] -> 90[2] via P2P/IPC
+ip-26-0-161-138:96567:96760 [4] NCCL INFO Channel 14/0 : 36[4] -> 35[3] via P2P/IPC
+ip-26-0-161-142:62885:63054 [4] NCCL INFO Channel 14/0 : 44[4] -> 43[3] via P2P/IPC
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO Channel 10/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO Channel 09/0 : 50[2] -> 49[1] via P2P/IPC
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO Channel 08/0 : 74[2] -> 73[1] via P2P/IPC
+ip-26-0-161-138:96570:96759 [7] NCCL INFO Channel 13/0 : 39[7] -> 32[0] via P2P/IPC
+ip-26-0-165-202:153566:153741 [4] NCCL INFO Channel 14/0 : 92[4] -> 91[3] via P2P/IPC
+ip-26-0-168-34:272155:272329 [5] NCCL INFO Channel 12/0 : 117[5] -> 116[4] via P2P/IPC
+ip-26-0-168-34:272157:272324 [7] NCCL INFO Channel 11/0 : 119[7] -> 112[0] via P2P/IPC
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO Channel 00/0 : 99[3] -> 98[2] via P2P/IPC
+ip-26-0-168-30:235651:235823 [3] NCCL INFO Channel 10/0 : 107[3] -> 106[2] via P2P/IPC
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO Channel 08/0 : 77[5] -> 76[4] via P2P/IPC
+ip-26-0-161-78:114515:114690 [3] NCCL INFO Channel 02/0 : 67[3] -> 66[2] via P2P/IPC
+ip-26-0-160-225:124164:124336 [4] NCCL INFO Channel 11/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-142:62884:63061 [3] NCCL INFO Channel 10/0 : 43[3] -> 42[2] via P2P/IPC
+ip-26-0-161-103:96891:97063 [3] NCCL INFO Channel 00/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO Channel 15/0 : 52[4] -> 51[3] via P2P/IPC
+ip-26-0-161-78:114517:114687 [5] NCCL INFO Channel 00/0 : 69[5] -> 68[4] via P2P/IPC
+ip-26-0-161-138:96567:96760 [4] NCCL INFO Channel 15/0 : 36[4] -> 35[3] via P2P/IPC
+ip-26-0-161-142:62885:63054 [4] NCCL INFO Channel 15/0 : 44[4] -> 43[3] via P2P/IPC
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO Channel 02/0 : 99[3] -> 98[2] via P2P/IPC
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO Channel 09/0 : 74[2] -> 73[1] via P2P/IPC
+ip-26-0-161-138:96570:96759 [7] NCCL INFO Channel 14/0 : 39[7] -> 32[0] via P2P/IPC
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO Channel 12/0 : 77[5] -> 76[4] via P2P/IPC
+ip-26-0-161-123:32706:32877 [2] NCCL INFO Channel 01/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-161-78:114515:114690 [3] NCCL INFO Channel 08/0 : 67[3] -> 66[2] via P2P/IPC
+ip-26-0-168-34:272157:272324 [7] NCCL INFO Channel 12/0 : 119[7] -> 112[0] via P2P/IPC
+ip-26-0-160-225:124164:124336 [4] NCCL INFO Channel 13/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-78:114517:114687 [5] NCCL INFO Channel 04/0 : 69[5] -> 68[4] via P2P/IPC
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO Channel 14/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-161-138:96570:96759 [7] NCCL INFO Channel 15/0 : 39[7] -> 32[0] via P2P/IPC
+ip-26-0-161-78:114515:114690 [3] NCCL INFO Channel 10/0 : 67[3] -> 66[2] via P2P/IPC
+ip-26-0-160-225:124164:124336 [4] NCCL INFO Channel 14/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-168-34:272157:272324 [7] NCCL INFO Channel 13/0 : 119[7] -> 112[0] via P2P/IPC
+ip-26-0-161-78:114517:114687 [5] NCCL INFO Channel 08/0 : 69[5] -> 68[4] via P2P/IPC
+ip-26-0-160-225:124163:124340 [3] NCCL INFO Channel 00/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-225:124164:124336 [4] NCCL INFO Channel 15/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO Channel 13/0 : 76[4] -> 75[3] via P2P/IPC
+ip-26-0-160-225:124163:124340 [3] NCCL INFO Channel 02/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO Channel 14/0 : 76[4] -> 75[3] via P2P/IPC
+ip-26-0-160-225:124163:124340 [3] NCCL INFO Channel 08/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO Channel 15/0 : 76[4] -> 75[3] via P2P/IPC
+ip-26-0-165-202:153566:153741 [4] NCCL INFO Channel 15/0 : 92[4] -> 91[3] via P2P/IPC
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO Channel 08/0 : 58[2] -> 57[1] via P2P/IPC
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO Channel 00/0 : 75[3] -> 74[2] via P2P/IPC
+ip-26-0-161-103:96891:97063 [3] NCCL INFO Channel 02/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO Channel 00/0 : 79[7] -> 78[6] via P2P/IPC
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO Channel 08/0 : 99[3] -> 98[2] via P2P/IPC
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO Channel 00/0 : 103[7] -> 102[6] via P2P/IPC
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO Channel 15/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-168-52:2722048:2722220 [3] NCCL INFO Channel 08/0 : 123[3] -> 122[2] via P2P/IPC
+ip-26-0-161-123:32706:32877 [2] NCCL INFO Channel 08/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-165-202:153565:153738 [3] NCCL INFO Channel 08/0 : 91[3] -> 90[2] via P2P/IPC
+ip-26-0-161-78:114517:114687 [5] NCCL INFO Channel 12/0 : 69[5] -> 68[4] via P2P/IPC
+ip-26-0-168-34:272157:272324 [7] NCCL INFO Channel 14/0 : 119[7] -> 112[0] via P2P/IPC
+ip-26-0-168-34:272153:272328 [3] NCCL INFO Channel 00/0 : 115[3] -> 114[2] via P2P/IPC
+ip-26-0-161-78:114519:114692 [7] NCCL INFO Channel 00/0 : 71[7] -> 70[6] via P2P/IPC
+ip-26-0-161-103:96891:97063 [3] NCCL INFO Channel 08/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO Channel 02/0 : 75[3] -> 74[2] via P2P/IPC
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO Channel 06/0 : 79[7] -> 78[6] via P2P/IPC
+ip-26-0-168-34:272157:272324 [7] NCCL INFO Channel 15/0 : 119[7] -> 112[0] via P2P/IPC
+ip-26-0-168-34:272153:272328 [3] NCCL INFO Channel 02/0 : 115[3] -> 114[2] via P2P/IPC
+ip-26-0-168-52:2722048:2722220 [3] NCCL INFO Channel 10/0 : 123[3] -> 122[2] via P2P/IPC
+ip-26-0-161-123:32706:32877 [2] NCCL INFO Channel 09/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO Channel 00/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO Channel 10/0 : 99[3] -> 98[2] via P2P/IPC
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO Channel 06/0 : 103[7] -> 102[6] via P2P/IPC
+ip-26-0-165-202:153565:153738 [3] NCCL INFO Channel 10/0 : 91[3] -> 90[2] via P2P/IPC
+ip-26-0-161-103:96891:97063 [3] NCCL INFO Channel 10/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO Channel 00/0 : 51[3] -> 50[2] via P2P/IPC
+ip-26-0-160-225:124163:124340 [3] NCCL INFO Channel 10/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-138:96566:96755 [3] NCCL INFO Channel 00/0 : 35[3] -> 34[2] via P2P/IPC
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO Channel 08/0 : 75[3] -> 74[2] via P2P/IPC
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO Channel 08/0 : 79[7] -> 78[6] via P2P/IPC
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO Channel 08/0 : 103[7] -> 102[6] via P2P/IPC
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO Channel 09/0 : 58[2] -> 57[1] via P2P/IPC
+ip-26-0-168-52:2722052:2722226 [7] NCCL INFO Channel 00/0 : 127[7] -> 126[6] via P2P/IPC
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO Channel 00/0 : 59[3] -> 58[2] via P2P/IPC
+ip-26-0-165-202:153569:153739 [7] NCCL INFO Channel 00/0 : 95[7] -> 94[6] via P2P/IPC
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO Channel 02/0 : 51[3] -> 50[2] via P2P/IPC
+ip-26-0-161-138:96570:96759 [7] NCCL INFO Channel 00/0 : 39[7] -> 38[6] via P2P/IPC
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO Channel 00/0 : 55[7] -> 54[6] via P2P/IPC
+ip-26-0-168-34:272153:272328 [3] NCCL INFO Channel 08/0 : 115[3] -> 114[2] via P2P/IPC
+ip-26-0-161-138:96566:96755 [3] NCCL INFO Channel 02/0 : 35[3] -> 34[2] via P2P/IPC
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO Channel 08/0 : 51[3] -> 50[2] via P2P/IPC
+ip-26-0-160-225:124167:124338 [7] NCCL INFO Channel 00/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-168-34:272157:272324 [7] NCCL INFO Channel 00/0 : 119[7] -> 118[6] via P2P/IPC
+ip-26-0-161-138:96570:96759 [7] NCCL INFO Channel 06/0 : 39[7] -> 38[6] via P2P/IPC
+ip-26-0-161-138:96566:96755 [3] NCCL INFO Channel 08/0 : 35[3] -> 34[2] via P2P/IPC
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO Channel 02/0 : 59[3] -> 58[2] via P2P/IPC
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO Channel 06/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO Channel 06/0 : 55[7] -> 54[6] via P2P/IPC
+ip-26-0-168-34:272153:272328 [3] NCCL INFO Channel 10/0 : 115[3] -> 114[2] via P2P/IPC
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO Channel 10/0 : 75[3] -> 74[2] via P2P/IPC
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO Channel 14/0 : 79[7] -> 78[6] via P2P/IPC
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO Channel 10/0 : 51[3] -> 50[2] via P2P/IPC
+ip-26-0-168-34:272157:272324 [7] NCCL INFO Channel 06/0 : 119[7] -> 118[6] via P2P/IPC
+ip-26-0-161-138:96570:96759 [7] NCCL INFO Channel 08/0 : 39[7] -> 38[6] via P2P/IPC
+ip-26-0-161-138:96566:96755 [3] NCCL INFO Channel 10/0 : 35[3] -> 34[2] via P2P/IPC
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO Channel 08/0 : 59[3] -> 58[2] via P2P/IPC
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO Channel 14/0 : 103[7] -> 102[6] via P2P/IPC
+ip-26-0-168-52:2722052:2722226 [7] NCCL INFO Channel 06/0 : 127[7] -> 126[6] via P2P/IPC
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO Channel 08/0 : 55[7] -> 54[6] via P2P/IPC
+ip-26-0-168-34:272157:272324 [7] NCCL INFO Channel 08/0 : 119[7] -> 118[6] via P2P/IPC
+ip-26-0-168-34:272157:272324 [7] NCCL INFO Channel 14/0 : 119[7] -> 118[6] via P2P/IPC
+ip-26-0-161-78:114519:114692 [7] NCCL INFO Channel 06/0 : 71[7] -> 70[6] via P2P/IPC
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO Channel 10/0 : 59[3] -> 58[2] via P2P/IPC
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO Channel 14/0 : 55[7] -> 54[6] via P2P/IPC
+ip-26-0-165-202:153569:153739 [7] NCCL INFO Channel 06/0 : 95[7] -> 94[6] via P2P/IPC
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO Channel 00/0 : 63[7] -> 62[6] via P2P/IPC
+ip-26-0-160-225:124167:124338 [7] NCCL INFO Channel 06/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-165-202:153569:153739 [7] NCCL INFO Channel 08/0 : 95[7] -> 94[6] via P2P/IPC
+ip-26-0-161-142:62888:63060 [7] NCCL INFO Channel 00/0 : 47[7] -> 46[6] via P2P/IPC
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO Channel 08/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-168-30:235655:235828 [7] NCCL INFO Channel 00/0 : 111[7] -> 110[6] via P2P/IPC
+ip-26-0-161-138:96570:96759 [7] NCCL INFO Channel 14/0 : 39[7] -> 38[6] via P2P/IPC
+ip-26-0-168-52:2722052:2722226 [7] NCCL INFO Channel 08/0 : 127[7] -> 126[6] via P2P/IPC
+ip-26-0-161-103:96895:97069 [7] NCCL INFO Channel 00/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO Channel 06/0 : 63[7] -> 62[6] via P2P/IPC
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO Channel 06/0 : 87[7] -> 80[0] via P2P/IPC
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO Channel 07/0 : 87[7] -> 80[0] via P2P/IPC
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO Channel 09/0 : 87[7] -> 80[0] via P2P/IPC
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO Channel 00/0 : 84[4] -> 83[3] via P2P/IPC
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO Channel 00/0 : 85[5] -> 84[4] via P2P/IPC
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO Channel 10/0 : 87[7] -> 80[0] via P2P/IPC
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO Channel 00/0 : 82[2] -> 81[1] via P2P/IPC
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO Channel 03/0 : 84[4] -> 83[3] via P2P/IPC
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO Channel 04/0 : 85[5] -> 84[4] via P2P/IPC
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO Channel 11/0 : 87[7] -> 80[0] via P2P/IPC
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO Channel 01/0 : 82[2] -> 81[1] via P2P/IPC
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO Channel 08/0 : 85[5] -> 84[4] via P2P/IPC
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO Channel 12/0 : 87[7] -> 80[0] via P2P/IPC
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO Channel 08/0 : 82[2] -> 81[1] via P2P/IPC
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO Channel 05/0 : 84[4] -> 83[3] via P2P/IPC
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO Channel 12/0 : 85[5] -> 84[4] via P2P/IPC
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO Channel 13/0 : 87[7] -> 80[0] via P2P/IPC
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO Channel 09/0 : 82[2] -> 81[1] via P2P/IPC
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO Channel 06/0 : 84[4] -> 83[3] via P2P/IPC
+ip-26-0-168-30:235655:235828 [7] NCCL INFO Channel 06/0 : 111[7] -> 110[6] via P2P/IPC
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO Channel 14/0 : 87[7] -> 80[0] via P2P/IPC
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO Channel 07/0 : 84[4] -> 83[3] via P2P/IPC
+ip-26-0-161-103:96895:97069 [7] NCCL INFO Channel 06/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO Channel 15/0 : 87[7] -> 80[0] via P2P/IPC
+ip-26-0-165-202:153569:153739 [7] NCCL INFO Channel 14/0 : 95[7] -> 94[6] via P2P/IPC
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO Channel 08/0 : 84[4] -> 83[3] via P2P/IPC
+ip-26-0-160-225:124167:124338 [7] NCCL INFO Channel 08/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO Channel 00/0 : 83[3] -> 82[2] via P2P/IPC
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO Channel 11/0 : 84[4] -> 83[3] via P2P/IPC
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO Channel 02/0 : 83[3] -> 82[2] via P2P/IPC
+ip-26-0-161-78:114519:114692 [7] NCCL INFO Channel 08/0 : 71[7] -> 70[6] via P2P/IPC
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO Channel 13/0 : 84[4] -> 83[3] via P2P/IPC
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO Channel 08/0 : 83[3] -> 82[2] via P2P/IPC
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO Channel 14/0 : 84[4] -> 83[3] via P2P/IPC
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO Channel 10/0 : 83[3] -> 82[2] via P2P/IPC
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO Channel 15/0 : 84[4] -> 83[3] via P2P/IPC
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO Channel 14/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO Channel 08/0 : 63[7] -> 62[6] via P2P/IPC
+ip-26-0-168-52:2722052:2722226 [7] NCCL INFO Channel 14/0 : 127[7] -> 126[6] via P2P/IPC
+ip-26-0-168-30:235655:235828 [7] NCCL INFO Channel 08/0 : 111[7] -> 110[6] via P2P/IPC
+ip-26-0-161-103:96895:97069 [7] NCCL INFO Channel 08/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-168-30:235655:235828 [7] NCCL INFO Channel 14/0 : 111[7] -> 110[6] via P2P/IPC
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO Channel 00/0 : 87[7] -> 86[6] via P2P/IPC
+ip-26-0-161-142:62888:63060 [7] NCCL INFO Channel 06/0 : 47[7] -> 46[6] via P2P/IPC
+ip-26-0-160-225:124167:124338 [7] NCCL INFO Channel 14/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-78:114519:114692 [7] NCCL INFO Channel 14/0 : 71[7] -> 70[6] via P2P/IPC
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO Channel 06/0 : 87[7] -> 86[6] via P2P/IPC
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO Channel 14/0 : 63[7] -> 62[6] via P2P/IPC
+ip-26-0-161-103:96895:97069 [7] NCCL INFO Channel 14/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-161-142:62888:63060 [7] NCCL INFO Channel 08/0 : 47[7] -> 46[6] via P2P/IPC
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO Channel 08/0 : 87[7] -> 86[6] via P2P/IPC
+ip-26-0-161-142:62888:63060 [7] NCCL INFO Channel 14/0 : 47[7] -> 46[6] via P2P/IPC
+ip-26-0-161-123:32711:32884 [7] NCCL INFO Channel 00/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO Channel 14/0 : 87[7] -> 86[6] via P2P/IPC
+ip-26-0-161-123:32711:32884 [7] NCCL INFO Channel 06/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-161-123:32711:32884 [7] NCCL INFO Channel 08/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-161-123:32711:32884 [7] NCCL INFO Channel 14/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-168-52:2722046:2722223 [1] NCCL INFO Connected all trees
+ip-26-0-168-52:2722045:2722219 [0] NCCL INFO Connected all trees
+ip-26-0-168-52:2722046:2722223 [1] NCCL INFO NVLS comm 0x9800e30 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-168-52:2722045:2722219 [0] NCCL INFO NVLS comm 0x852ebd0 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-225:124163:124340 [3] NCCL INFO Connected all trees
+ip-26-0-160-225:124163:124340 [3] NCCL INFO NVLS comm 0x94c8a80 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-225:124167:124338 [7] NCCL INFO Connected all trees
+ip-26-0-160-225:124165:124341 [5] NCCL INFO Connected all trees
+ip-26-0-160-225:124167:124338 [7] NCCL INFO NVLS comm 0x9c467b0 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-225:124165:124341 [5] NCCL INFO NVLS comm 0x8838dc0 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-225:124162:124342 [2] NCCL INFO Connected all trees
+ip-26-0-160-225:124162:124342 [2] NCCL INFO NVLS comm 0xa2aa6e0 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-168-52:2722050:2722225 [5] NCCL INFO Connected all trees
+ip-26-0-168-52:2722050:2722225 [5] NCCL INFO NVLS comm 0x86815d0 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-168-52:2722052:2722226 [7] NCCL INFO Connected all trees
+ip-26-0-168-52:2722052:2722226 [7] NCCL INFO NVLS comm 0x9675870 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-225:124164:124336 [4] NCCL INFO Connected all trees
+ip-26-0-160-225:124164:124336 [4] NCCL INFO NVLS comm 0x8f6da90 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-168-52:2722048:2722220 [3] NCCL INFO Connected all trees
+ip-26-0-168-52:2722048:2722220 [3] NCCL INFO NVLS comm 0x864def0 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-225:124161:124337 [1] NCCL INFO Connected all trees
+ip-26-0-160-225:124161:124337 [1] NCCL INFO NVLS comm 0x87faca0 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-225:124160:124335 [0] NCCL INFO Connected all trees
+ip-26-0-160-225:124160:124335 [0] NCCL INFO NVLS comm 0x921d930 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-225:124166:124339 [6] NCCL INFO Connected all trees
+ip-26-0-160-225:124166:124339 [6] NCCL INFO NVLS comm 0x975d3b0 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-168-52:2722047:2722221 [2] NCCL INFO Connected all trees
+ip-26-0-168-52:2722047:2722221 [2] NCCL INFO NVLS comm 0x94226d0 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO Connected all trees
+ip-26-0-168-52:2722051:2722224 [6] NCCL INFO Connected all trees
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO NVLS comm 0x8818910 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-168-52:2722051:2722224 [6] NCCL INFO NVLS comm 0x92559c0 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-142:62886:63057 [5] NCCL INFO Connected all trees
+ip-26-0-161-142:62886:63057 [5] NCCL INFO NVLS comm 0x9de8db0 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-142:62888:63060 [7] NCCL INFO Connected all trees
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO Connected all trees
+ip-26-0-161-142:62888:63060 [7] NCCL INFO NVLS comm 0x91600d0 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO Connected all trees
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO NVLS comm 0xa08a840 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO NVLS comm 0xa120e40 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO Connected all trees
+ip-26-0-168-30:235649:235827 [1] NCCL INFO Connected all trees
+ip-26-0-168-30:235649:235827 [1] NCCL INFO NVLS comm 0x9bd9090 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO NVLS comm 0x8b40680 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-78:114512:114686 [0] NCCL INFO Connected all trees
+ip-26-0-161-78:114512:114686 [0] NCCL INFO NVLS comm 0x9281680 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-142:62881:63059 [0] NCCL INFO Connected all trees
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO Connected all trees
+ip-26-0-161-142:62881:63059 [0] NCCL INFO NVLS comm 0xa10dc70 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-142:62887:63055 [6] NCCL INFO Connected all trees
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO NVLS comm 0x94a1240 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-142:62883:63058 [2] NCCL INFO Connected all trees
+ip-26-0-161-142:62887:63055 [6] NCCL INFO NVLS comm 0x867a2a0 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-142:62883:63058 [2] NCCL INFO NVLS comm 0x8b38920 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-123:32707:32878 [3] NCCL INFO Connected all trees
+ip-26-0-161-123:32708:32879 [4] NCCL INFO Connected all trees
+ip-26-0-168-30:235648:235826 [0] NCCL INFO Connected all trees
+ip-26-0-161-123:32707:32878 [3] NCCL INFO NVLS comm 0x887a500 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-123:32708:32879 [4] NCCL INFO NVLS comm 0xa27e140 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-168-30:235648:235826 [0] NCCL INFO NVLS comm 0x890ea90 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-123:32709:32882 [5] NCCL INFO Connected all trees
+ip-26-0-161-123:32709:32882 [5] NCCL INFO NVLS comm 0xa23fe10 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-78:114519:114692 [7] NCCL INFO Connected all trees
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO Connected all trees
+ip-26-0-168-30:235655:235828 [7] NCCL INFO Connected all trees
+ip-26-0-161-78:114519:114692 [7] NCCL INFO NVLS comm 0x84be170 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-168-30:235655:235828 [7] NCCL INFO NVLS comm 0x9447f80 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO NVLS comm 0x9ff3aa0 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO Connected all trees
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO Connected all trees
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO NVLS comm 0x8bcfca0 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO Connected all trees
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO NVLS comm 0x8793c10 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO Connected all trees
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO NVLS comm 0x9406880 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO NVLS comm 0x937b130 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-168-30:235654:235821 [6] NCCL INFO Connected all trees
+ip-26-0-161-78:114515:114690 [3] NCCL INFO Connected all trees
+ip-26-0-161-78:114513:114688 [1] NCCL INFO Connected all trees
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO Connected all trees
+ip-26-0-168-30:235654:235821 [6] NCCL INFO NVLS comm 0x994c420 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-78:114515:114690 [3] NCCL INFO NVLS comm 0x8dcd620 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-78:114513:114688 [1] NCCL INFO NVLS comm 0x9e36040 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO NVLS comm 0x90c0bf0 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-142:62882:63056 [1] NCCL INFO Connected all trees
+ip-26-0-168-30:235650:235824 [2] NCCL INFO Connected all trees
+ip-26-0-161-142:62882:63056 [1] NCCL INFO NVLS comm 0x8ed2140 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-168-30:235650:235824 [2] NCCL INFO NVLS comm 0x8ef34a0 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-142:62884:63061 [3] NCCL INFO Connected all trees
+ip-26-0-161-142:62884:63061 [3] NCCL INFO NVLS comm 0x95b82f0 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO Connected all trees
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO NVLS comm 0x982c5c0 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-78:114516:114691 [4] NCCL INFO Connected all trees
+ip-26-0-161-78:114517:114687 [5] NCCL INFO Connected all trees
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO Connected all trees
+ip-26-0-161-78:114517:114687 [5] NCCL INFO NVLS comm 0xa282530 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-78:114516:114691 [4] NCCL INFO NVLS comm 0x936c450 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO NVLS comm 0x871cb90 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO Connected all trees
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO NVLS comm 0x889c010 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO Connected all trees
+ip-26-0-168-30:235652:235822 [4] NCCL INFO Connected all trees
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO NVLS comm 0x84a7790 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-168-30:235652:235822 [4] NCCL INFO NVLS comm 0x9c80d90 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO Connected all trees
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO Connected all trees
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO NVLS comm 0x9f93670 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO NVLS comm 0x97add10 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-168-30:235653:235825 [5] NCCL INFO Connected all trees
+ip-26-0-161-142:62885:63054 [4] NCCL INFO Connected all trees
+ip-26-0-168-30:235653:235825 [5] NCCL INFO NVLS comm 0x91002f0 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-142:62885:63054 [4] NCCL INFO NVLS comm 0x9e18d20 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO Connected all trees
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO Connected all trees
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO NVLS comm 0x8629c70 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO NVLS comm 0x954a920 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO Connected all trees
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO NVLS comm 0x8917950 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO Connected all trees
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO NVLS comm 0xa46efe0 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO Connected all trees
+ip-26-0-161-78:114518:114693 [6] NCCL INFO Connected all trees
+ip-26-0-161-78:114514:114689 [2] NCCL INFO Connected all trees
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO Connected all trees
+ip-26-0-161-78:114518:114693 [6] NCCL INFO NVLS comm 0x9878640 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO NVLS comm 0x9ef1450 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-78:114514:114689 [2] NCCL INFO NVLS comm 0x9cf4bf0 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO NVLS comm 0x8690d60 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO Connected all trees
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO NVLS comm 0x8e6e9d0 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO Connected all trees
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO NVLS comm 0x8f46e20 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO Connected all trees
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO NVLS comm 0x9a07a60 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-202:153567:153742 [5] NCCL INFO Connected all trees
+ip-26-0-165-202:153567:153742 [5] NCCL INFO NVLS comm 0xa1c25f0 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO Connected all trees
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO NVLS comm 0x929efe0 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-202:153562:153735 [0] NCCL INFO Connected all trees
+ip-26-0-165-202:153562:153735 [0] NCCL INFO NVLS comm 0x948d480 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-202:153565:153738 [3] NCCL INFO Connected all trees
+ip-26-0-165-202:153565:153738 [3] NCCL INFO NVLS comm 0x9998430 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO Connected all trees
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO NVLS comm 0x90028a0 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO Connected all trees
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO NVLS comm 0x9a50ce0 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-168-34:272156:272325 [6] NCCL INFO Connected all trees
+ip-26-0-168-34:272156:272325 [6] NCCL INFO NVLS comm 0xa194790 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO Connected all trees
+ip-26-0-168-34:272155:272329 [5] NCCL INFO Connected all trees
+ip-26-0-168-34:272155:272329 [5] NCCL INFO NVLS comm 0x91a3e50 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO NVLS comm 0x9e94a10 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO Connected all trees
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO NVLS comm 0xa1c3e40 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-168-34:272157:272324 [7] NCCL INFO Connected all trees
+ip-26-0-165-202:153568:153736 [6] NCCL INFO Connected all trees
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO Connected all trees
+ip-26-0-165-202:153568:153736 [6] NCCL INFO NVLS comm 0x87ae600 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-168-34:272157:272324 [7] NCCL INFO NVLS comm 0x93fc8f0 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO NVLS comm 0x994f620 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-168-30:235651:235823 [3] NCCL INFO Connected all trees
+ip-26-0-168-30:235651:235823 [3] NCCL INFO NVLS comm 0x8ed2f20 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-168-34:272154:272330 [4] NCCL INFO Connected all trees
+ip-26-0-168-34:272150:272327 [0] NCCL INFO Connected all trees
+ip-26-0-168-34:272154:272330 [4] NCCL INFO NVLS comm 0x9279c80 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-168-34:272152:272331 [2] NCCL INFO Connected all trees
+ip-26-0-165-202:153564:153737 [2] NCCL INFO Connected all trees
+ip-26-0-168-34:272153:272328 [3] NCCL INFO Connected all trees
+ip-26-0-168-34:272150:272327 [0] NCCL INFO NVLS comm 0x9d62500 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-168-34:272152:272331 [2] NCCL INFO NVLS comm 0xa466440 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-202:153564:153737 [2] NCCL INFO NVLS comm 0x87bf990 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-168-34:272153:272328 [3] NCCL INFO NVLS comm 0x85b56b0 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO Connected all trees
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO NVLS comm 0x8798710 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-202:153563:153740 [1] NCCL INFO Connected all trees
+ip-26-0-165-202:153563:153740 [1] NCCL INFO NVLS comm 0xa3c8c40 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO Connected all trees
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO NVLS comm 0x8c25390 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO Connected all trees
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO NVLS comm 0x88a2150 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-202:153569:153739 [7] NCCL INFO Connected all trees
+ip-26-0-165-202:153569:153739 [7] NCCL INFO NVLS comm 0x87276c0 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-123:32711:32884 [7] NCCL INFO Connected all trees
+ip-26-0-161-138:96567:96760 [4] NCCL INFO Connected all trees
+ip-26-0-161-123:32711:32884 [7] NCCL INFO NVLS comm 0x9e7ffa0 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-138:96567:96760 [4] NCCL INFO NVLS comm 0x886b530 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-225:124164:124336 [4] NCCL INFO Channel 01/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153566:153741 [4] NCCL INFO Connected all trees
+ip-26-0-160-225:124164:124336 [4] NCCL INFO Channel 03/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:124164:124336 [4] NCCL INFO Channel 05/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:124164:124336 [4] NCCL INFO Channel 07/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO Connected all trees
+ip-26-0-160-225:124164:124336 [4] NCCL INFO Channel 09/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722052:2722226 [7] NCCL INFO Channel 00/0 : 119[7] -> 127[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124164:124336 [4] NCCL INFO Channel 11/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:124164:124336 [4] NCCL INFO Channel 13/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722052:2722226 [7] NCCL INFO Channel 02/0 : 119[7] -> 127[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153566:153741 [4] NCCL INFO NVLS comm 0x8a05910 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-168-52:2722052:2722226 [7] NCCL INFO Channel 04/0 : 119[7] -> 127[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124164:124336 [4] NCCL INFO Channel 15/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722052:2722226 [7] NCCL INFO Channel 06/0 : 119[7] -> 127[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722052:2722226 [7] NCCL INFO Channel 08/0 : 119[7] -> 127[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722052:2722226 [7] NCCL INFO Channel 10/0 : 119[7] -> 127[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96566:96755 [3] NCCL INFO Connected all trees
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO NVLS comm 0x9b878e0 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-168-52:2722052:2722226 [7] NCCL INFO Channel 12/0 : 119[7] -> 127[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722052:2722226 [7] NCCL INFO Channel 14/0 : 119[7] -> 127[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96566:96755 [3] NCCL INFO NVLS comm 0x9315300 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-138:96568:96753 [5] NCCL INFO Connected all trees
+ip-26-0-168-52:2722046:2722223 [1] NCCL INFO Channel 00/0 : 113[1] -> 121[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-52:2722046:2722223 [1] NCCL INFO Channel 02/0 : 113[1] -> 121[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96568:96753 [5] NCCL INFO NVLS comm 0x88d80e0 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-168-52:2722051:2722224 [6] NCCL INFO Channel 00/0 : 118[6] -> 126[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-52:2722050:2722225 [5] NCCL INFO Channel 00/0 : 117[5] -> 125[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96563:96756 [0] NCCL INFO Connected all trees
+ip-26-0-168-52:2722046:2722223 [1] NCCL INFO Channel 04/0 : 113[1] -> 121[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-52:2722051:2722224 [6] NCCL INFO Channel 02/0 : 118[6] -> 126[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-52:2722050:2722225 [5] NCCL INFO Channel 02/0 : 117[5] -> 125[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO Channel 00/0 : 116[4] -> 124[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722046:2722223 [1] NCCL INFO Channel 06/0 : 113[1] -> 121[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-52:2722048:2722220 [3] NCCL INFO Channel 00/0 : 115[3] -> 123[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-52:2722051:2722224 [6] NCCL INFO Channel 04/0 : 118[6] -> 126[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-52:2722050:2722225 [5] NCCL INFO Channel 04/0 : 117[5] -> 125[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-52:2722045:2722219 [0] NCCL INFO Channel 02/0 : 112[0] -> 120[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO Channel 02/0 : 116[4] -> 124[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722046:2722223 [1] NCCL INFO Channel 08/0 : 113[1] -> 121[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-52:2722051:2722224 [6] NCCL INFO Channel 08/0 : 118[6] -> 126[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-52:2722050:2722225 [5] NCCL INFO Channel 06/0 : 117[5] -> 125[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-52:2722048:2722220 [3] NCCL INFO Channel 02/0 : 115[3] -> 123[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-52:2722047:2722221 [2] NCCL INFO Channel 00/0 : 114[2] -> 122[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-52:2722045:2722219 [0] NCCL INFO Channel 04/0 : 112[0] -> 120[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO Channel 06/0 : 116[4] -> 124[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96563:96756 [0] NCCL INFO NVLS comm 0x99b2260 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-168-52:2722046:2722223 [1] NCCL INFO Channel 10/0 : 113[1] -> 121[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-52:2722050:2722225 [5] NCCL INFO Channel 08/0 : 117[5] -> 125[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-52:2722048:2722220 [3] NCCL INFO Channel 04/0 : 115[3] -> 123[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-52:2722051:2722224 [6] NCCL INFO Channel 10/0 : 118[6] -> 126[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-52:2722047:2722221 [2] NCCL INFO Channel 04/0 : 114[2] -> 122[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:124160:124335 [0] NCCL INFO Channel 01/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO Channel 08/0 : 116[4] -> 124[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722045:2722219 [0] NCCL INFO Channel 06/0 : 112[0] -> 120[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-52:2722046:2722223 [1] NCCL INFO Channel 12/0 : 113[1] -> 121[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-52:2722050:2722225 [5] NCCL INFO Channel 10/0 : 117[5] -> 125[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124161:124337 [1] NCCL INFO Channel 01/0 : 1[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-52:2722048:2722220 [3] NCCL INFO Channel 06/0 : 115[3] -> 123[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:124160:124335 [0] NCCL INFO Channel 03/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:124166:124339 [6] NCCL INFO Channel 01/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-52:2722051:2722224 [6] NCCL INFO Channel 12/0 : 118[6] -> 126[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-52:2722047:2722221 [2] NCCL INFO Channel 06/0 : 114[2] -> 122[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:124161:124337 [1] NCCL INFO Channel 03/0 : 1[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124160:124335 [0] NCCL INFO Channel 05/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:124167:124338 [7] NCCL INFO Channel 01/0 : 7[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722046:2722223 [1] NCCL INFO Channel 14/0 : 113[1] -> 121[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO Channel 10/0 : 116[4] -> 124[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722045:2722219 [0] NCCL INFO Channel 08/0 : 112[0] -> 120[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:124165:124341 [5] NCCL INFO Channel 01/0 : 5[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-52:2722050:2722225 [5] NCCL INFO Channel 12/0 : 117[5] -> 125[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124166:124339 [6] NCCL INFO Channel 03/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:124161:124337 [1] NCCL INFO Channel 05/0 : 1[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124160:124335 [0] NCCL INFO Channel 07/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:124167:124338 [7] NCCL INFO Channel 03/0 : 7[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722048:2722220 [3] NCCL INFO Channel 08/0 : 115[3] -> 123[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:124165:124341 [5] NCCL INFO Channel 03/0 : 5[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124166:124339 [6] NCCL INFO Channel 05/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-52:2722051:2722224 [6] NCCL INFO Channel 14/0 : 118[6] -> 126[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:124161:124337 [1] NCCL INFO Channel 07/0 : 1[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-52:2722047:2722221 [2] NCCL INFO Channel 08/0 : 114[2] -> 122[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:124160:124335 [0] NCCL INFO Channel 09/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-52:2722045:2722219 [0] NCCL INFO Channel 10/0 : 112[0] -> 120[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:124167:124338 [7] NCCL INFO Channel 05/0 : 7[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO Channel 12/0 : 116[4] -> 124[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722050:2722225 [5] NCCL INFO Channel 14/0 : 117[5] -> 125[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-52:2722048:2722220 [3] NCCL INFO Channel 10/0 : 115[3] -> 123[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:124166:124339 [6] NCCL INFO Channel 07/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:124165:124341 [5] NCCL INFO Channel 05/0 : 5[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124163:124340 [3] NCCL INFO Channel 01/0 : 3[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:124161:124337 [1] NCCL INFO Channel 11/0 : 1[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124160:124335 [0] NCCL INFO Channel 11/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:124167:124338 [7] NCCL INFO Channel 07/0 : 7[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124162:124342 [2] NCCL INFO Channel 01/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-52:2722047:2722221 [2] NCCL INFO Channel 10/0 : 114[2] -> 122[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:124166:124339 [6] NCCL INFO Channel 09/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-52:2722045:2722219 [0] NCCL INFO Channel 12/0 : 112[0] -> 120[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:124165:124341 [5] NCCL INFO Channel 07/0 : 5[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO Channel 14/0 : 116[4] -> 124[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722048:2722220 [3] NCCL INFO Channel 12/0 : 115[3] -> 123[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:124163:124340 [3] NCCL INFO Channel 03/0 : 3[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-52:2722047:2722221 [2] NCCL INFO Channel 12/0 : 114[2] -> 122[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-52:2722045:2722219 [0] NCCL INFO Channel 14/0 : 112[0] -> 120[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:124161:124337 [1] NCCL INFO Channel 13/0 : 1[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124167:124338 [7] NCCL INFO Channel 09/0 : 7[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722048:2722220 [3] NCCL INFO Channel 14/0 : 115[3] -> 123[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-52:2722047:2722221 [2] NCCL INFO Channel 14/0 : 114[2] -> 122[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:124160:124335 [0] NCCL INFO Channel 13/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:124162:124342 [2] NCCL INFO Channel 03/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:124166:124339 [6] NCCL INFO Channel 11/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:124165:124341 [5] NCCL INFO Channel 09/0 : 5[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124163:124340 [3] NCCL INFO Channel 05/0 : 3[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:124167:124338 [7] NCCL INFO Channel 11/0 : 7[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124161:124337 [1] NCCL INFO Channel 15/0 : 1[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124162:124342 [2] NCCL INFO Channel 05/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:124160:124335 [0] NCCL INFO Channel 15/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:124166:124339 [6] NCCL INFO Channel 13/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:124165:124341 [5] NCCL INFO Channel 11/0 : 5[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124163:124340 [3] NCCL INFO Channel 07/0 : 3[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:124167:124338 [7] NCCL INFO Channel 13/0 : 7[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124166:124339 [6] NCCL INFO Channel 15/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:124165:124341 [5] NCCL INFO Channel 15/0 : 5[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124163:124340 [3] NCCL INFO Channel 09/0 : 3[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:124162:124342 [2] NCCL INFO Channel 07/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:124163:124340 [3] NCCL INFO Channel 13/0 : 3[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:124162:124342 [2] NCCL INFO Channel 09/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:124163:124340 [3] NCCL INFO Channel 15/0 : 3[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:124162:124342 [2] NCCL INFO Channel 11/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:124162:124342 [2] NCCL INFO Channel 13/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:124162:124342 [2] NCCL INFO Channel 15/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272151:272326 [1] NCCL INFO Connected all trees
+ip-26-0-168-34:272151:272326 [1] NCCL INFO NVLS comm 0x9494aa0 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO Connected all trees
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO NVLS comm 0x91b2fc0 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO Connected all trees
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO Connected all trees
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO NVLS comm 0x91fcaa0 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO Connected all trees
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO Connected all trees
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO NVLS comm 0x84bbf10 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO NVLS comm 0x9298e10 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO NVLS comm 0x86bcdf0 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO Connected all trees
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO NVLS comm 0x9b32d60 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO Connected all trees
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO Connected all trees
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO Connected all trees
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO NVLS comm 0x84c9ed0 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO NVLS comm 0x9061f10 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO NVLS comm 0x8e7c100 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-138:96570:96759 [7] NCCL INFO Connected all trees
+ip-26-0-161-138:96570:96759 [7] NCCL INFO NVLS comm 0xa0c08e0 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-123:32704:32881 [0] NCCL INFO Connected all trees
+ip-26-0-161-123:32704:32881 [0] NCCL INFO NVLS comm 0xa45bb70 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-138:96565:96754 [2] NCCL INFO Connected all trees
+ip-26-0-161-138:96569:96758 [6] NCCL INFO Connected all trees
+ip-26-0-161-138:96565:96754 [2] NCCL INFO NVLS comm 0xa208380 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-138:96569:96758 [6] NCCL INFO NVLS comm 0x8e85570 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-138:96564:96757 [1] NCCL INFO Connected all trees
+ip-26-0-161-138:96564:96757 [1] NCCL INFO NVLS comm 0x96232d0 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-123:32706:32877 [2] NCCL INFO Connected all trees
+ip-26-0-161-123:32706:32877 [2] NCCL INFO NVLS comm 0x9a13240 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-123:32710:32883 [6] NCCL INFO Connected all trees
+ip-26-0-161-123:32710:32883 [6] NCCL INFO NVLS comm 0x991e950 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-142:62885:63054 [4] NCCL INFO Channel 01/0 : 36[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62885:63054 [4] NCCL INFO Channel 03/0 : 36[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62887:63055 [6] NCCL INFO Channel 01/0 : 38[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62885:63054 [4] NCCL INFO Channel 05/0 : 36[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62886:63057 [5] NCCL INFO Channel 01/0 : 37[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62887:63055 [6] NCCL INFO Channel 03/0 : 38[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62885:63054 [4] NCCL INFO Channel 07/0 : 36[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62881:63059 [0] NCCL INFO Channel 01/0 : 32[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62887:63055 [6] NCCL INFO Channel 05/0 : 38[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62886:63057 [5] NCCL INFO Channel 03/0 : 37[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62885:63054 [4] NCCL INFO Channel 09/0 : 36[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62887:63055 [6] NCCL INFO Channel 07/0 : 38[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62881:63059 [0] NCCL INFO Channel 03/0 : 32[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62885:63054 [4] NCCL INFO Channel 11/0 : 36[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62887:63055 [6] NCCL INFO Channel 09/0 : 38[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62881:63059 [0] NCCL INFO Channel 05/0 : 32[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62885:63054 [4] NCCL INFO Channel 13/0 : 36[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62887:63055 [6] NCCL INFO Channel 11/0 : 38[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62881:63059 [0] NCCL INFO Channel 07/0 : 32[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62885:63054 [4] NCCL INFO Channel 15/0 : 36[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62887:63055 [6] NCCL INFO Channel 13/0 : 38[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62881:63059 [0] NCCL INFO Channel 09/0 : 32[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62885:63054 [4] NCCL INFO Channel 00/0 : 44[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62887:63055 [6] NCCL INFO Channel 15/0 : 38[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62885:63054 [4] NCCL INFO Channel 01/0 : 44[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62887:63055 [6] NCCL INFO Channel 00/0 : 46[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62881:63059 [0] NCCL INFO Channel 11/0 : 32[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62885:63054 [4] NCCL INFO Channel 02/0 : 44[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62887:63055 [6] NCCL INFO Channel 01/0 : 46[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62881:63059 [0] NCCL INFO Channel 13/0 : 32[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62885:63054 [4] NCCL INFO Channel 03/0 : 44[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32705:32880 [1] NCCL INFO Connected all trees
+ip-26-0-161-142:62887:63055 [6] NCCL INFO Channel 02/0 : 46[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62881:63059 [0] NCCL INFO Channel 15/0 : 32[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62885:63054 [4] NCCL INFO Channel 05/0 : 44[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62887:63055 [6] NCCL INFO Channel 03/0 : 46[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62881:63059 [0] NCCL INFO Channel 01/0 : 40[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62885:63054 [4] NCCL INFO Channel 06/0 : 44[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62887:63055 [6] NCCL INFO Channel 04/0 : 46[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62881:63059 [0] NCCL INFO Channel 02/0 : 40[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62885:63054 [4] NCCL INFO Channel 07/0 : 44[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62886:63057 [5] NCCL INFO Channel 05/0 : 37[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62887:63055 [6] NCCL INFO Channel 05/0 : 46[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62881:63059 [0] NCCL INFO Channel 03/0 : 40[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62885:63054 [4] NCCL INFO Channel 08/0 : 44[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62886:63057 [5] NCCL INFO Channel 07/0 : 37[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62887:63055 [6] NCCL INFO Channel 07/0 : 46[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62881:63059 [0] NCCL INFO Channel 04/0 : 40[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62885:63054 [4] NCCL INFO Channel 09/0 : 44[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62886:63057 [5] NCCL INFO Channel 09/0 : 37[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62887:63055 [6] NCCL INFO Channel 08/0 : 46[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62881:63059 [0] NCCL INFO Channel 05/0 : 40[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62885:63054 [4] NCCL INFO Channel 10/0 : 44[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62886:63057 [5] NCCL INFO Channel 11/0 : 37[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32705:32880 [1] NCCL INFO NVLS comm 0x99eb180 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-142:62887:63055 [6] NCCL INFO Channel 09/0 : 46[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62885:63054 [4] NCCL INFO Channel 11/0 : 44[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62881:63059 [0] NCCL INFO Channel 06/0 : 40[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62886:63057 [5] NCCL INFO Channel 15/0 : 37[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62887:63055 [6] NCCL INFO Channel 10/0 : 46[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62882:63056 [1] NCCL INFO Channel 01/0 : 33[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62888:63060 [7] NCCL INFO Channel 01/0 : 39[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62883:63058 [2] NCCL INFO Channel 01/0 : 34[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62885:63054 [4] NCCL INFO Channel 13/0 : 44[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62881:63059 [0] NCCL INFO Channel 07/0 : 40[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62886:63057 [5] NCCL INFO Channel 00/0 : 45[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62887:63055 [6] NCCL INFO Channel 11/0 : 46[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62888:63060 [7] NCCL INFO Channel 03/0 : 39[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62882:63056 [1] NCCL INFO Channel 03/0 : 33[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62883:63058 [2] NCCL INFO Channel 03/0 : 34[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62885:63054 [4] NCCL INFO Channel 14/0 : 44[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62881:63059 [0] NCCL INFO Channel 09/0 : 40[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62887:63055 [6] NCCL INFO Channel 12/0 : 46[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62886:63057 [5] NCCL INFO Channel 01/0 : 45[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62888:63060 [7] NCCL INFO Channel 05/0 : 39[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62884:63061 [3] NCCL INFO Channel 01/0 : 35[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62885:63054 [4] NCCL INFO Channel 15/0 : 44[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62882:63056 [1] NCCL INFO Channel 05/0 : 33[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62883:63058 [2] NCCL INFO Channel 05/0 : 34[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62881:63059 [0] NCCL INFO Channel 10/0 : 40[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62886:63057 [5] NCCL INFO Channel 02/0 : 45[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62887:63055 [6] NCCL INFO Channel 13/0 : 46[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62888:63060 [7] NCCL INFO Channel 07/0 : 39[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62884:63061 [3] NCCL INFO Channel 03/0 : 35[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62882:63056 [1] NCCL INFO Channel 07/0 : 33[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62883:63058 [2] NCCL INFO Channel 07/0 : 34[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62881:63059 [0] NCCL INFO Channel 11/0 : 40[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62886:63057 [5] NCCL INFO Channel 03/0 : 45[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62887:63055 [6] NCCL INFO Channel 15/0 : 46[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62888:63060 [7] NCCL INFO Channel 09/0 : 39[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62884:63061 [3] NCCL INFO Channel 05/0 : 35[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62882:63056 [1] NCCL INFO Channel 11/0 : 33[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62883:63058 [2] NCCL INFO Channel 09/0 : 34[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62881:63059 [0] NCCL INFO Channel 12/0 : 40[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62886:63057 [5] NCCL INFO Channel 04/0 : 45[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62888:63060 [7] NCCL INFO Channel 11/0 : 39[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62884:63061 [3] NCCL INFO Channel 07/0 : 35[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62882:63056 [1] NCCL INFO Channel 13/0 : 33[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62881:63059 [0] NCCL INFO Channel 13/0 : 40[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62883:63058 [2] NCCL INFO Channel 11/0 : 34[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62886:63057 [5] NCCL INFO Channel 06/0 : 45[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62888:63060 [7] NCCL INFO Channel 13/0 : 39[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62882:63056 [1] NCCL INFO Channel 15/0 : 33[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62884:63061 [3] NCCL INFO Channel 09/0 : 35[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62881:63059 [0] NCCL INFO Channel 14/0 : 40[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62886:63057 [5] NCCL INFO Channel 07/0 : 45[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62883:63058 [2] NCCL INFO Channel 13/0 : 34[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62888:63060 [7] NCCL INFO Channel 00/0 : 47[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62882:63056 [1] NCCL INFO Channel 00/0 : 41[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62881:63059 [0] NCCL INFO Channel 15/0 : 40[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62884:63061 [3] NCCL INFO Channel 13/0 : 35[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62886:63057 [5] NCCL INFO Channel 08/0 : 45[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62888:63060 [7] NCCL INFO Channel 01/0 : 47[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62883:63058 [2] NCCL INFO Channel 15/0 : 34[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62882:63056 [1] NCCL INFO Channel 02/0 : 41[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62886:63057 [5] NCCL INFO Channel 09/0 : 45[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62884:63061 [3] NCCL INFO Channel 15/0 : 35[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62888:63060 [7] NCCL INFO Channel 02/0 : 47[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62883:63058 [2] NCCL INFO Channel 00/0 : 42[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62882:63056 [1] NCCL INFO Channel 03/0 : 41[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62886:63057 [5] NCCL INFO Channel 10/0 : 45[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62888:63060 [7] NCCL INFO Channel 03/0 : 47[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62884:63061 [3] NCCL INFO Channel 00/0 : 43[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62883:63058 [2] NCCL INFO Channel 01/0 : 42[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62882:63056 [1] NCCL INFO Channel 04/0 : 41[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62886:63057 [5] NCCL INFO Channel 11/0 : 45[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62888:63060 [7] NCCL INFO Channel 04/0 : 47[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62884:63061 [3] NCCL INFO Channel 01/0 : 43[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO Channel 01/0 : 102[6] -> 110[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62883:63058 [2] NCCL INFO Channel 03/0 : 42[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62886:63057 [5] NCCL INFO Channel 12/0 : 45[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62882:63056 [1] NCCL INFO Channel 05/0 : 41[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62888:63060 [7] NCCL INFO Channel 05/0 : 47[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62884:63061 [3] NCCL INFO Channel 02/0 : 43[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO Channel 03/0 : 102[6] -> 110[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO Channel 05/0 : 102[6] -> 110[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO Channel 07/0 : 102[6] -> 110[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO Channel 09/0 : 102[6] -> 110[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO Channel 11/0 : 102[6] -> 110[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO Channel 13/0 : 102[6] -> 110[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO Channel 15/0 : 102[6] -> 110[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114517:114687 [5] NCCL INFO Channel 01/0 : 69[5] -> 77[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114517:114687 [5] NCCL INFO Channel 03/0 : 69[5] -> 77[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114517:114687 [5] NCCL INFO Channel 05/0 : 69[5] -> 77[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114519:114692 [7] NCCL INFO Channel 01/0 : 71[7] -> 79[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114517:114687 [5] NCCL INFO Channel 07/0 : 69[5] -> 77[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114519:114692 [7] NCCL INFO Channel 03/0 : 71[7] -> 79[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114517:114687 [5] NCCL INFO Channel 09/0 : 69[5] -> 77[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114519:114692 [7] NCCL INFO Channel 05/0 : 71[7] -> 79[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114513:114688 [1] NCCL INFO Channel 01/0 : 65[1] -> 73[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114517:114687 [5] NCCL INFO Channel 11/0 : 69[5] -> 77[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114519:114692 [7] NCCL INFO Channel 07/0 : 71[7] -> 79[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114517:114687 [5] NCCL INFO Channel 15/0 : 69[5] -> 77[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114513:114688 [1] NCCL INFO Channel 03/0 : 65[1] -> 73[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114519:114692 [7] NCCL INFO Channel 09/0 : 71[7] -> 79[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114513:114688 [1] NCCL INFO Channel 05/0 : 65[1] -> 73[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114519:114692 [7] NCCL INFO Channel 11/0 : 71[7] -> 79[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114513:114688 [1] NCCL INFO Channel 07/0 : 65[1] -> 73[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114519:114692 [7] NCCL INFO Channel 13/0 : 71[7] -> 79[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114513:114688 [1] NCCL INFO Channel 11/0 : 65[1] -> 73[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114513:114688 [1] NCCL INFO Channel 13/0 : 65[1] -> 73[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114513:114688 [1] NCCL INFO Channel 15/0 : 65[1] -> 73[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO Channel 01/0 : 96[0] -> 104[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO Channel 01/0 : 101[5] -> 109[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO Channel 01/0 : 97[1] -> 105[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114516:114691 [4] NCCL INFO Channel 01/0 : 68[4] -> 76[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114518:114693 [6] NCCL INFO Channel 01/0 : 70[6] -> 78[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO Channel 01/0 : 100[4] -> 108[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO Channel 03/0 : 101[5] -> 109[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO Channel 03/0 : 96[0] -> 104[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO Channel 01/0 : 103[7] -> 111[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114516:114691 [4] NCCL INFO Channel 03/0 : 68[4] -> 76[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO Channel 03/0 : 97[1] -> 105[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114512:114686 [0] NCCL INFO Channel 01/0 : 64[0] -> 72[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114515:114690 [3] NCCL INFO Channel 01/0 : 67[3] -> 75[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114518:114693 [6] NCCL INFO Channel 03/0 : 70[6] -> 78[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114516:114691 [4] NCCL INFO Channel 05/0 : 68[4] -> 76[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO Channel 01/0 : 98[2] -> 106[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO Channel 03/0 : 100[4] -> 108[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO Channel 05/0 : 101[5] -> 109[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO Channel 05/0 : 96[0] -> 104[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO Channel 01/0 : 99[3] -> 107[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO Channel 03/0 : 103[7] -> 111[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO Channel 05/0 : 97[1] -> 105[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114512:114686 [0] NCCL INFO Channel 03/0 : 64[0] -> 72[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114515:114690 [3] NCCL INFO Channel 03/0 : 67[3] -> 75[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114518:114693 [6] NCCL INFO Channel 05/0 : 70[6] -> 78[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114516:114691 [4] NCCL INFO Channel 07/0 : 68[4] -> 76[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO Channel 05/0 : 100[4] -> 108[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114514:114689 [2] NCCL INFO Channel 01/0 : 66[2] -> 74[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114512:114686 [0] NCCL INFO Channel 05/0 : 64[0] -> 72[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO Channel 03/0 : 98[2] -> 106[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114515:114690 [3] NCCL INFO Channel 05/0 : 67[3] -> 75[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO Channel 07/0 : 101[5] -> 109[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO Channel 07/0 : 96[0] -> 104[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO Channel 03/0 : 99[3] -> 107[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO Channel 07/0 : 97[1] -> 105[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO Channel 05/0 : 103[7] -> 111[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114518:114693 [6] NCCL INFO Channel 07/0 : 70[6] -> 78[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114516:114691 [4] NCCL INFO Channel 09/0 : 68[4] -> 76[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114512:114686 [0] NCCL INFO Channel 07/0 : 64[0] -> 72[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114514:114689 [2] NCCL INFO Channel 03/0 : 66[2] -> 74[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO Channel 05/0 : 98[2] -> 106[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114515:114690 [3] NCCL INFO Channel 07/0 : 67[3] -> 75[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114518:114693 [6] NCCL INFO Channel 09/0 : 70[6] -> 78[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114512:114686 [0] NCCL INFO Channel 09/0 : 64[0] -> 72[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO Channel 11/0 : 97[1] -> 105[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO Channel 09/0 : 96[0] -> 104[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO Channel 09/0 : 101[5] -> 109[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO Channel 05/0 : 99[3] -> 107[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO Channel 07/0 : 103[7] -> 111[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO Channel 07/0 : 100[4] -> 108[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114516:114691 [4] NCCL INFO Channel 11/0 : 68[4] -> 76[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114515:114690 [3] NCCL INFO Channel 09/0 : 67[3] -> 75[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114514:114689 [2] NCCL INFO Channel 05/0 : 66[2] -> 74[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO Channel 07/0 : 98[2] -> 106[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114518:114693 [6] NCCL INFO Channel 11/0 : 70[6] -> 78[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO Channel 11/0 : 96[0] -> 104[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114512:114686 [0] NCCL INFO Channel 11/0 : 64[0] -> 72[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO Channel 13/0 : 97[1] -> 105[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114515:114690 [3] NCCL INFO Channel 13/0 : 67[3] -> 75[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114514:114689 [2] NCCL INFO Channel 07/0 : 66[2] -> 74[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114516:114691 [4] NCCL INFO Channel 13/0 : 68[4] -> 76[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114512:114686 [0] NCCL INFO Channel 13/0 : 64[0] -> 72[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114518:114693 [6] NCCL INFO Channel 13/0 : 70[6] -> 78[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO Channel 07/0 : 99[3] -> 107[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO Channel 11/0 : 101[5] -> 109[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO Channel 09/0 : 103[7] -> 111[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO Channel 09/0 : 100[4] -> 108[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO Channel 09/0 : 98[2] -> 106[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO Channel 13/0 : 96[0] -> 104[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114515:114690 [3] NCCL INFO Channel 15/0 : 67[3] -> 75[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO Channel 15/0 : 97[1] -> 105[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114514:114689 [2] NCCL INFO Channel 09/0 : 66[2] -> 74[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114516:114691 [4] NCCL INFO Channel 15/0 : 68[4] -> 76[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114512:114686 [0] NCCL INFO Channel 15/0 : 64[0] -> 72[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114518:114693 [6] NCCL INFO Channel 15/0 : 70[6] -> 78[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO Channel 15/0 : 101[5] -> 109[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO Channel 11/0 : 98[2] -> 106[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO Channel 09/0 : 99[3] -> 107[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO Channel 15/0 : 96[0] -> 104[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO Channel 11/0 : 100[4] -> 108[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO Channel 11/0 : 103[7] -> 111[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114514:114689 [2] NCCL INFO Channel 11/0 : 66[2] -> 74[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO Channel 13/0 : 98[2] -> 106[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO Channel 13/0 : 99[3] -> 107[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO Channel 13/0 : 103[7] -> 111[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO Channel 13/0 : 100[4] -> 108[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114514:114689 [2] NCCL INFO Channel 13/0 : 66[2] -> 74[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO Channel 15/0 : 98[2] -> 106[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO Channel 15/0 : 99[3] -> 107[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114514:114689 [2] NCCL INFO Channel 15/0 : 66[2] -> 74[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO Channel 15/0 : 100[4] -> 108[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO Channel 00/0 : 55[7] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO Channel 02/0 : 55[7] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO Channel 04/0 : 55[7] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO Channel 06/0 : 55[7] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO Channel 08/0 : 55[7] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO Channel 10/0 : 55[7] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO Channel 12/0 : 55[7] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO Channel 14/0 : 55[7] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO Channel 00/0 : 52[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO Channel 02/0 : 48[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO Channel 00/0 : 53[5] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO Channel 02/0 : 52[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO Channel 02/0 : 53[5] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO Channel 04/0 : 48[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO Channel 06/0 : 52[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO Channel 00/0 : 54[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO Channel 00/0 : 50[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO Channel 04/0 : 53[5] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO Channel 06/0 : 48[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO Channel 08/0 : 52[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO Channel 02/0 : 54[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO Channel 00/0 : 51[3] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO Channel 06/0 : 53[5] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO Channel 00/0 : 49[1] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO Channel 04/0 : 50[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO Channel 08/0 : 48[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO Channel 10/0 : 52[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO Channel 04/0 : 54[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO Channel 02/0 : 51[3] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO Channel 08/0 : 53[5] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO Channel 06/0 : 50[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO Channel 10/0 : 48[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO Channel 02/0 : 49[1] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO Channel 12/0 : 52[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO Channel 08/0 : 54[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO Channel 04/0 : 51[3] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO Channel 08/0 : 50[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO Channel 10/0 : 53[5] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO Channel 12/0 : 48[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO Channel 14/0 : 52[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO Channel 10/0 : 54[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO Channel 04/0 : 49[1] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO Channel 06/0 : 51[3] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO Channel 10/0 : 50[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO Channel 12/0 : 53[5] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO Channel 14/0 : 48[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO Channel 12/0 : 54[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO Channel 06/0 : 49[1] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO Channel 08/0 : 51[3] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO Channel 14/0 : 53[5] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO Channel 12/0 : 50[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO Channel 14/0 : 54[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO Channel 08/0 : 49[1] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO Channel 10/0 : 51[3] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO Channel 14/0 : 50[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO Channel 10/0 : 49[1] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO Channel 12/0 : 51[3] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO Channel 12/0 : 49[1] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO Channel 14/0 : 51[3] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO Channel 14/0 : 49[1] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235655:235828 [7] NCCL INFO Channel 01/0 : 103[7] -> 111[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235655:235828 [7] NCCL INFO Channel 03/0 : 103[7] -> 111[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235653:235825 [5] NCCL INFO Channel 01/0 : 101[5] -> 109[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235649:235827 [1] NCCL INFO Channel 01/0 : 97[1] -> 105[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235655:235828 [7] NCCL INFO Channel 05/0 : 103[7] -> 111[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235653:235825 [5] NCCL INFO Channel 03/0 : 101[5] -> 109[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235655:235828 [7] NCCL INFO Channel 07/0 : 103[7] -> 111[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235649:235827 [1] NCCL INFO Channel 03/0 : 97[1] -> 105[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235653:235825 [5] NCCL INFO Channel 05/0 : 101[5] -> 109[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235655:235828 [7] NCCL INFO Channel 09/0 : 103[7] -> 111[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235649:235827 [1] NCCL INFO Channel 05/0 : 97[1] -> 105[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235653:235825 [5] NCCL INFO Channel 07/0 : 101[5] -> 109[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235655:235828 [7] NCCL INFO Channel 11/0 : 103[7] -> 111[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235649:235827 [1] NCCL INFO Channel 07/0 : 97[1] -> 105[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235653:235825 [5] NCCL INFO Channel 09/0 : 101[5] -> 109[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235655:235828 [7] NCCL INFO Channel 13/0 : 103[7] -> 111[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235653:235825 [5] NCCL INFO Channel 11/0 : 101[5] -> 109[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235649:235827 [1] NCCL INFO Channel 11/0 : 97[1] -> 105[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235655:235828 [7] NCCL INFO Channel 00/0 : 111[7] -> 119[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235649:235827 [1] NCCL INFO Channel 13/0 : 97[1] -> 105[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235653:235825 [5] NCCL INFO Channel 15/0 : 101[5] -> 109[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235655:235828 [7] NCCL INFO Channel 01/0 : 111[7] -> 119[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235649:235827 [1] NCCL INFO Channel 15/0 : 97[1] -> 105[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235653:235825 [5] NCCL INFO Channel 00/0 : 109[5] -> 117[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235655:235828 [7] NCCL INFO Channel 02/0 : 111[7] -> 119[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235653:235825 [5] NCCL INFO Channel 01/0 : 109[5] -> 117[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235655:235828 [7] NCCL INFO Channel 03/0 : 111[7] -> 119[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235649:235827 [1] NCCL INFO Channel 00/0 : 105[1] -> 113[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235653:235825 [5] NCCL INFO Channel 02/0 : 109[5] -> 117[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235655:235828 [7] NCCL INFO Channel 04/0 : 111[7] -> 119[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235649:235827 [1] NCCL INFO Channel 02/0 : 105[1] -> 113[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235653:235825 [5] NCCL INFO Channel 03/0 : 109[5] -> 117[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235655:235828 [7] NCCL INFO Channel 05/0 : 111[7] -> 119[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235649:235827 [1] NCCL INFO Channel 03/0 : 105[1] -> 113[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235653:235825 [5] NCCL INFO Channel 04/0 : 109[5] -> 117[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235655:235828 [7] NCCL INFO Channel 06/0 : 111[7] -> 119[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235649:235827 [1] NCCL INFO Channel 04/0 : 105[1] -> 113[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235653:235825 [5] NCCL INFO Channel 06/0 : 109[5] -> 117[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235655:235828 [7] NCCL INFO Channel 08/0 : 111[7] -> 119[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235649:235827 [1] NCCL INFO Channel 05/0 : 105[1] -> 113[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235653:235825 [5] NCCL INFO Channel 07/0 : 109[5] -> 117[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235655:235828 [7] NCCL INFO Channel 09/0 : 111[7] -> 119[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235649:235827 [1] NCCL INFO Channel 06/0 : 105[1] -> 113[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235653:235825 [5] NCCL INFO Channel 08/0 : 109[5] -> 117[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235655:235828 [7] NCCL INFO Channel 10/0 : 111[7] -> 119[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235649:235827 [1] NCCL INFO Channel 07/0 : 105[1] -> 113[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235654:235821 [6] NCCL INFO Channel 01/0 : 102[6] -> 110[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235653:235825 [5] NCCL INFO Channel 09/0 : 109[5] -> 117[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235652:235822 [4] NCCL INFO Channel 01/0 : 100[4] -> 108[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235655:235828 [7] NCCL INFO Channel 11/0 : 111[7] -> 119[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235649:235827 [1] NCCL INFO Channel 08/0 : 105[1] -> 113[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235653:235825 [5] NCCL INFO Channel 10/0 : 109[5] -> 117[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235648:235826 [0] NCCL INFO Channel 01/0 : 96[0] -> 104[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235654:235821 [6] NCCL INFO Channel 03/0 : 102[6] -> 110[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235651:235823 [3] NCCL INFO Channel 01/0 : 99[3] -> 107[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235655:235828 [7] NCCL INFO Channel 12/0 : 111[7] -> 119[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235652:235822 [4] NCCL INFO Channel 03/0 : 100[4] -> 108[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235649:235827 [1] NCCL INFO Channel 10/0 : 105[1] -> 113[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235653:235825 [5] NCCL INFO Channel 11/0 : 109[5] -> 117[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235654:235821 [6] NCCL INFO Channel 05/0 : 102[6] -> 110[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235648:235826 [0] NCCL INFO Channel 03/0 : 96[0] -> 104[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235655:235828 [7] NCCL INFO Channel 13/0 : 111[7] -> 119[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235651:235823 [3] NCCL INFO Channel 03/0 : 99[3] -> 107[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235652:235822 [4] NCCL INFO Channel 05/0 : 100[4] -> 108[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235649:235827 [1] NCCL INFO Channel 11/0 : 105[1] -> 113[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235650:235824 [2] NCCL INFO Channel 01/0 : 98[2] -> 106[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235653:235825 [5] NCCL INFO Channel 12/0 : 109[5] -> 117[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235654:235821 [6] NCCL INFO Channel 07/0 : 102[6] -> 110[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235655:235828 [7] NCCL INFO Channel 14/0 : 111[7] -> 119[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235648:235826 [0] NCCL INFO Channel 05/0 : 96[0] -> 104[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235651:235823 [3] NCCL INFO Channel 05/0 : 99[3] -> 107[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235649:235827 [1] NCCL INFO Channel 12/0 : 105[1] -> 113[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235653:235825 [5] NCCL INFO Channel 14/0 : 109[5] -> 117[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235652:235822 [4] NCCL INFO Channel 07/0 : 100[4] -> 108[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235654:235821 [6] NCCL INFO Channel 09/0 : 102[6] -> 110[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235650:235824 [2] NCCL INFO Channel 03/0 : 98[2] -> 106[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235648:235826 [0] NCCL INFO Channel 07/0 : 96[0] -> 104[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235649:235827 [1] NCCL INFO Channel 13/0 : 105[1] -> 113[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235651:235823 [3] NCCL INFO Channel 07/0 : 99[3] -> 107[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235653:235825 [5] NCCL INFO Channel 15/0 : 109[5] -> 117[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235654:235821 [6] NCCL INFO Channel 11/0 : 102[6] -> 110[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235652:235822 [4] NCCL INFO Channel 09/0 : 100[4] -> 108[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235650:235824 [2] NCCL INFO Channel 05/0 : 98[2] -> 106[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235648:235826 [0] NCCL INFO Channel 09/0 : 96[0] -> 104[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235649:235827 [1] NCCL INFO Channel 14/0 : 105[1] -> 113[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235651:235823 [3] NCCL INFO Channel 09/0 : 99[3] -> 107[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235654:235821 [6] NCCL INFO Channel 13/0 : 102[6] -> 110[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235652:235822 [4] NCCL INFO Channel 11/0 : 100[4] -> 108[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO Channel 00/0 : 87[7] -> 103[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235650:235824 [2] NCCL INFO Channel 07/0 : 98[2] -> 106[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235648:235826 [0] NCCL INFO Channel 11/0 : 96[0] -> 104[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235649:235827 [1] NCCL INFO Channel 15/0 : 105[1] -> 113[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235651:235823 [3] NCCL INFO Channel 13/0 : 99[3] -> 107[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235654:235821 [6] NCCL INFO Channel 15/0 : 102[6] -> 110[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO Channel 02/0 : 87[7] -> 103[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235652:235822 [4] NCCL INFO Channel 13/0 : 100[4] -> 108[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235650:235824 [2] NCCL INFO Channel 09/0 : 98[2] -> 106[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO Channel 04/0 : 87[7] -> 103[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235651:235823 [3] NCCL INFO Channel 15/0 : 99[3] -> 107[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235654:235821 [6] NCCL INFO Channel 00/0 : 110[6] -> 118[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235648:235826 [0] NCCL INFO Channel 13/0 : 96[0] -> 104[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235652:235822 [4] NCCL INFO Channel 15/0 : 100[4] -> 108[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235650:235824 [2] NCCL INFO Channel 11/0 : 98[2] -> 106[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO Channel 00/0 : 85[5] -> 101[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235654:235821 [6] NCCL INFO Channel 01/0 : 110[6] -> 118[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235651:235823 [3] NCCL INFO Channel 00/0 : 107[3] -> 115[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235648:235826 [0] NCCL INFO Channel 15/0 : 96[0] -> 104[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO Channel 06/0 : 87[7] -> 103[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235652:235822 [4] NCCL INFO Channel 00/0 : 108[4] -> 116[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235650:235824 [2] NCCL INFO Channel 13/0 : 98[2] -> 106[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO Channel 02/0 : 85[5] -> 101[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235654:235821 [6] NCCL INFO Channel 02/0 : 110[6] -> 118[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO Channel 00/0 : 81[1] -> 97[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO Channel 08/0 : 87[7] -> 103[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235651:235823 [3] NCCL INFO Channel 01/0 : 107[3] -> 115[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235648:235826 [0] NCCL INFO Channel 01/0 : 104[0] -> 112[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235652:235822 [4] NCCL INFO Channel 01/0 : 108[4] -> 116[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235650:235824 [2] NCCL INFO Channel 15/0 : 98[2] -> 106[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO Channel 04/0 : 85[5] -> 101[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235654:235821 [6] NCCL INFO Channel 03/0 : 110[6] -> 118[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235648:235826 [0] NCCL INFO Channel 02/0 : 104[0] -> 112[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235651:235823 [3] NCCL INFO Channel 02/0 : 107[3] -> 115[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235650:235824 [2] NCCL INFO Channel 00/0 : 106[2] -> 114[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO Channel 02/0 : 81[1] -> 97[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO Channel 10/0 : 87[7] -> 103[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235652:235822 [4] NCCL INFO Channel 02/0 : 108[4] -> 116[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO Channel 06/0 : 85[5] -> 101[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235654:235821 [6] NCCL INFO Channel 04/0 : 110[6] -> 118[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235648:235826 [0] NCCL INFO Channel 03/0 : 104[0] -> 112[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO Channel 04/0 : 81[1] -> 97[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235651:235823 [3] NCCL INFO Channel 04/0 : 107[3] -> 115[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO Channel 12/0 : 87[7] -> 103[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235650:235824 [2] NCCL INFO Channel 01/0 : 106[2] -> 114[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO Channel 08/0 : 85[5] -> 101[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235652:235822 [4] NCCL INFO Channel 03/0 : 108[4] -> 116[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235654:235821 [6] NCCL INFO Channel 05/0 : 110[6] -> 118[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235648:235826 [0] NCCL INFO Channel 04/0 : 104[0] -> 112[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235651:235823 [3] NCCL INFO Channel 05/0 : 107[3] -> 115[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO Channel 06/0 : 81[1] -> 97[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235650:235824 [2] NCCL INFO Channel 03/0 : 106[2] -> 114[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO Channel 14/0 : 87[7] -> 103[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO Channel 10/0 : 85[5] -> 101[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235652:235822 [4] NCCL INFO Channel 05/0 : 108[4] -> 116[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235654:235821 [6] NCCL INFO Channel 07/0 : 110[6] -> 118[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235648:235826 [0] NCCL INFO Channel 05/0 : 104[0] -> 112[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235651:235823 [3] NCCL INFO Channel 06/0 : 107[3] -> 115[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235650:235824 [2] NCCL INFO Channel 04/0 : 106[2] -> 114[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO Channel 08/0 : 81[1] -> 97[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO Channel 12/0 : 85[5] -> 101[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO Channel 00/0 : 103[7] -> 119[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235652:235822 [4] NCCL INFO Channel 06/0 : 108[4] -> 116[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235654:235821 [6] NCCL INFO Channel 08/0 : 110[6] -> 118[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235648:235826 [0] NCCL INFO Channel 06/0 : 104[0] -> 112[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235651:235823 [3] NCCL INFO Channel 07/0 : 107[3] -> 115[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235650:235824 [2] NCCL INFO Channel 05/0 : 106[2] -> 114[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO Channel 10/0 : 81[1] -> 97[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO Channel 14/0 : 85[5] -> 101[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO Channel 02/0 : 103[7] -> 119[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO Channel 12/0 : 81[1] -> 97[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO Channel 04/0 : 103[7] -> 119[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO Channel 00/0 : 101[5] -> 117[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO Channel 14/0 : 81[1] -> 97[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO Channel 06/0 : 103[7] -> 119[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO Channel 02/0 : 101[5] -> 117[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO Channel 00/0 : 97[1] -> 113[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO Channel 08/0 : 103[7] -> 119[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO Channel 04/0 : 101[5] -> 117[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO Channel 02/0 : 97[1] -> 113[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO Channel 10/0 : 103[7] -> 119[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO Channel 06/0 : 101[5] -> 117[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO Channel 04/0 : 97[1] -> 113[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO Channel 12/0 : 103[7] -> 119[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO Channel 06/0 : 97[1] -> 113[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO Channel 08/0 : 101[5] -> 117[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO Channel 14/0 : 103[7] -> 119[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO Channel 08/0 : 97[1] -> 113[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO Channel 10/0 : 101[5] -> 117[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO Channel 10/0 : 97[1] -> 113[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO Channel 12/0 : 101[5] -> 117[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO Channel 12/0 : 97[1] -> 113[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO Channel 14/0 : 101[5] -> 117[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO Channel 14/0 : 97[1] -> 113[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO Channel 00/0 : 86[6] -> 102[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO Channel 00/0 : 83[3] -> 99[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO Channel 02/0 : 86[6] -> 102[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO Channel 00/0 : 84[4] -> 100[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO Channel 02/0 : 80[0] -> 96[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO Channel 02/0 : 83[3] -> 99[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO Channel 02/0 : 84[4] -> 100[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO Channel 00/0 : 82[2] -> 98[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO Channel 04/0 : 86[6] -> 102[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO Channel 04/0 : 83[3] -> 99[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO Channel 04/0 : 80[0] -> 96[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO Channel 06/0 : 84[4] -> 100[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO Channel 08/0 : 86[6] -> 102[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO Channel 04/0 : 82[2] -> 98[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO Channel 06/0 : 83[3] -> 99[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO Channel 06/0 : 80[0] -> 96[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO Channel 08/0 : 84[4] -> 100[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO Channel 10/0 : 86[6] -> 102[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO Channel 06/0 : 82[2] -> 98[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO Channel 08/0 : 83[3] -> 99[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO Channel 08/0 : 80[0] -> 96[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO Channel 12/0 : 86[6] -> 102[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO Channel 10/0 : 84[4] -> 100[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO Channel 08/0 : 82[2] -> 98[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO Channel 10/0 : 80[0] -> 96[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO Channel 10/0 : 83[3] -> 99[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO Channel 14/0 : 86[6] -> 102[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO Channel 12/0 : 84[4] -> 100[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO Channel 10/0 : 82[2] -> 98[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO Channel 12/0 : 80[0] -> 96[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO Channel 12/0 : 83[3] -> 99[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO Channel 00/0 : 102[6] -> 118[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO Channel 12/0 : 82[2] -> 98[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO Channel 14/0 : 84[4] -> 100[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO Channel 14/0 : 80[0] -> 96[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO Channel 14/0 : 83[3] -> 99[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO Channel 02/0 : 102[6] -> 118[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO Channel 14/0 : 82[2] -> 98[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO Channel 02/0 : 96[0] -> 112[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO Channel 00/0 : 100[4] -> 116[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO Channel 00/0 : 99[3] -> 115[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO Channel 04/0 : 102[6] -> 118[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO Channel 00/0 : 98[2] -> 114[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO Channel 04/0 : 96[0] -> 112[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO Channel 08/0 : 102[6] -> 118[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO Channel 02/0 : 99[3] -> 115[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO Channel 02/0 : 100[4] -> 116[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO Channel 04/0 : 98[2] -> 114[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO Channel 06/0 : 96[0] -> 112[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO Channel 10/0 : 102[6] -> 118[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO Channel 04/0 : 99[3] -> 115[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO Channel 06/0 : 98[2] -> 114[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO Channel 08/0 : 96[0] -> 112[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO Channel 06/0 : 100[4] -> 116[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO Channel 12/0 : 102[6] -> 118[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO Channel 06/0 : 99[3] -> 115[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO Channel 08/0 : 98[2] -> 114[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO Channel 10/0 : 96[0] -> 112[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO Channel 08/0 : 100[4] -> 116[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO Channel 14/0 : 102[6] -> 118[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO Channel 08/0 : 99[3] -> 115[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO Channel 12/0 : 96[0] -> 112[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO Channel 10/0 : 98[2] -> 114[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO Channel 10/0 : 100[4] -> 116[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO Channel 10/0 : 99[3] -> 115[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO Channel 14/0 : 96[0] -> 112[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO Channel 12/0 : 98[2] -> 114[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO Channel 12/0 : 100[4] -> 116[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO Channel 12/0 : 99[3] -> 115[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO Channel 14/0 : 98[2] -> 114[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO Channel 14/0 : 99[3] -> 115[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO Channel 14/0 : 100[4] -> 116[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO Channel 00/0 : 46[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO Channel 01/0 : 46[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO Channel 02/0 : 46[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO Channel 03/0 : 46[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO Channel 01/0 : 40[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO Channel 04/0 : 46[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO Channel 02/0 : 40[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO Channel 05/0 : 46[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO Channel 00/0 : 44[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO Channel 03/0 : 40[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO Channel 07/0 : 46[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO Channel 01/0 : 44[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO Channel 04/0 : 40[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO Channel 08/0 : 46[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO Channel 02/0 : 44[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO Channel 09/0 : 46[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO Channel 05/0 : 40[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO Channel 03/0 : 44[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO Channel 10/0 : 46[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO Channel 05/0 : 44[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO Channel 06/0 : 40[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO Channel 11/0 : 46[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO Channel 06/0 : 44[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO Channel 07/0 : 40[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO Channel 12/0 : 46[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO Channel 07/0 : 44[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO Channel 09/0 : 40[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO Channel 13/0 : 46[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO Channel 08/0 : 44[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO Channel 15/0 : 46[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO Channel 10/0 : 40[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO Channel 09/0 : 44[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO Channel 00/0 : 54[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO Channel 10/0 : 44[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO Channel 11/0 : 40[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO Channel 02/0 : 54[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO Channel 11/0 : 44[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO Channel 12/0 : 40[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO Channel 04/0 : 54[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO Channel 13/0 : 44[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO Channel 08/0 : 54[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO Channel 13/0 : 40[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO Channel 14/0 : 44[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO Channel 10/0 : 54[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO Channel 14/0 : 40[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO Channel 15/0 : 44[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO Channel 12/0 : 54[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO Channel 15/0 : 40[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO Channel 00/0 : 52[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO Channel 00/0 : 45[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO Channel 14/0 : 54[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO Channel 02/0 : 48[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO Channel 02/0 : 52[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO Channel 06/0 : 52[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO Channel 04/0 : 48[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO Channel 01/0 : 45[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO Channel 08/0 : 52[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO Channel 06/0 : 48[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO Channel 02/0 : 45[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO Channel 00/0 : 41[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO Channel 10/0 : 52[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO Channel 00/0 : 42[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO Channel 08/0 : 48[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO Channel 03/0 : 45[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO Channel 00/0 : 47[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO Channel 12/0 : 52[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO Channel 01/0 : 30[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO Channel 02/0 : 41[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO Channel 01/0 : 42[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO Channel 00/0 : 38[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO Channel 03/0 : 30[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO Channel 10/0 : 48[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO Channel 04/0 : 45[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO Channel 01/0 : 47[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO Channel 05/0 : 30[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO Channel 14/0 : 52[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO Channel 00/0 : 43[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO Channel 02/0 : 38[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO Channel 03/0 : 42[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO Channel 07/0 : 30[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO Channel 03/0 : 41[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO Channel 12/0 : 48[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO Channel 06/0 : 45[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO Channel 02/0 : 47[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO Channel 09/0 : 30[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO Channel 04/0 : 38[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO Channel 11/0 : 30[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO Channel 01/0 : 43[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO Channel 04/0 : 42[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO Channel 14/0 : 48[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO Channel 07/0 : 45[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO Channel 03/0 : 47[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO Channel 04/0 : 41[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO Channel 13/0 : 30[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO Channel 08/0 : 38[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO Channel 02/0 : 43[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO Channel 05/0 : 42[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO Channel 15/0 : 30[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO Channel 08/0 : 45[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO Channel 04/0 : 47[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO Channel 01/0 : 28[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO Channel 10/0 : 38[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO Channel 05/0 : 41[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO Channel 04/0 : 43[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO Channel 00/0 : 36[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO Channel 06/0 : 42[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO Channel 01/0 : 62[6] -> 94[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO Channel 09/0 : 45[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO Channel 05/0 : 47[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO Channel 12/0 : 38[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO Channel 03/0 : 28[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO Channel 06/0 : 41[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO Channel 03/0 : 62[6] -> 94[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO Channel 05/0 : 43[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO Channel 02/0 : 36[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO Channel 05/0 : 28[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO Channel 05/0 : 62[6] -> 94[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO Channel 07/0 : 42[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO Channel 10/0 : 45[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO Channel 06/0 : 47[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO Channel 14/0 : 38[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO Channel 07/0 : 28[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO Channel 07/0 : 41[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO Channel 07/0 : 62[6] -> 94[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO Channel 06/0 : 43[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO Channel 02/0 : 32[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO Channel 06/0 : 36[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO Channel 09/0 : 28[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO Channel 11/0 : 45[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO Channel 01/0 : 24[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO Channel 09/0 : 62[6] -> 94[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO Channel 08/0 : 47[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO Channel 08/0 : 42[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO Channel 08/0 : 41[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO Channel 11/0 : 28[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO Channel 08/0 : 36[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO Channel 07/0 : 43[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO Channel 12/0 : 45[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO Channel 04/0 : 32[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO Channel 11/0 : 62[6] -> 94[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO Channel 03/0 : 24[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO Channel 13/0 : 28[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO Channel 09/0 : 47[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO Channel 09/0 : 42[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO Channel 10/0 : 41[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO Channel 13/0 : 62[6] -> 94[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO Channel 05/0 : 24[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO Channel 15/0 : 28[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO Channel 10/0 : 36[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO Channel 14/0 : 45[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO Channel 15/0 : 62[6] -> 94[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO Channel 07/0 : 24[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO Channel 01/0 : 60[4] -> 92[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO Channel 08/0 : 43[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO Channel 06/0 : 32[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO Channel 10/0 : 47[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO Channel 11/0 : 42[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO Channel 11/0 : 41[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO Channel 03/0 : 60[4] -> 92[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO Channel 12/0 : 36[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO Channel 15/0 : 45[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO Channel 09/0 : 24[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153568:153736 [6] NCCL INFO Channel 00/0 : 86[6] -> 94[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO Channel 09/0 : 43[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO Channel 08/0 : 32[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO Channel 05/0 : 60[4] -> 92[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO Channel 11/0 : 47[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO Channel 11/0 : 24[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153562:153735 [0] NCCL INFO Channel 02/0 : 80[0] -> 88[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO Channel 12/0 : 42[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153568:153736 [6] NCCL INFO Channel 02/0 : 86[6] -> 94[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO Channel 07/0 : 60[4] -> 92[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO Channel 12/0 : 41[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO Channel 14/0 : 36[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO Channel 13/0 : 24[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO Channel 00/0 : 53[5] -> 61[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO Channel 10/0 : 43[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153562:153735 [0] NCCL INFO Channel 04/0 : 80[0] -> 88[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153568:153736 [6] NCCL INFO Channel 04/0 : 86[6] -> 94[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO Channel 09/0 : 60[4] -> 92[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO Channel 12/0 : 47[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO Channel 10/0 : 32[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO Channel 13/0 : 42[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO Channel 13/0 : 41[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO Channel 02/0 : 53[5] -> 61[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153562:153735 [0] NCCL INFO Channel 06/0 : 80[0] -> 88[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO Channel 15/0 : 24[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO Channel 11/0 : 60[4] -> 92[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO Channel 12/0 : 43[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO Channel 13/0 : 47[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO Channel 12/0 : 32[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153568:153736 [6] NCCL INFO Channel 08/0 : 86[6] -> 94[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO Channel 14/0 : 42[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153562:153735 [0] NCCL INFO Channel 08/0 : 80[0] -> 88[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO Channel 01/0 : 56[0] -> 88[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO Channel 13/0 : 60[4] -> 92[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO Channel 04/0 : 53[5] -> 61[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO Channel 14/0 : 41[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153568:153736 [6] NCCL INFO Channel 10/0 : 86[6] -> 94[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO Channel 13/0 : 43[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153562:153735 [0] NCCL INFO Channel 10/0 : 80[0] -> 88[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO Channel 03/0 : 56[0] -> 88[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO Channel 15/0 : 60[4] -> 92[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO Channel 14/0 : 47[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO Channel 14/0 : 32[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153568:153736 [6] NCCL INFO Channel 12/0 : 86[6] -> 94[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO Channel 15/0 : 42[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153562:153735 [0] NCCL INFO Channel 12/0 : 80[0] -> 88[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO Channel 05/0 : 56[0] -> 88[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO Channel 06/0 : 53[5] -> 61[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153568:153736 [6] NCCL INFO Channel 14/0 : 86[6] -> 94[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO Channel 07/0 : 56[0] -> 88[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153562:153735 [0] NCCL INFO Channel 14/0 : 80[0] -> 88[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO Channel 15/0 : 41[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO Channel 14/0 : 43[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO Channel 00/0 : 55[7] -> 63[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO Channel 00/0 : 50[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO Channel 08/0 : 53[5] -> 61[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO Channel 09/0 : 56[0] -> 88[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO Channel 11/0 : 56[0] -> 88[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO Channel 00/0 : 49[1] -> 57[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO Channel 15/0 : 43[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO Channel 13/0 : 56[0] -> 88[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO Channel 02/0 : 55[7] -> 63[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO Channel 10/0 : 53[5] -> 61[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO Channel 04/0 : 50[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO Channel 02/0 : 49[1] -> 57[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO Channel 00/0 : 51[3] -> 59[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO Channel 15/0 : 56[0] -> 88[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO Channel 12/0 : 53[5] -> 61[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO Channel 04/0 : 55[7] -> 63[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO Channel 06/0 : 50[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO Channel 04/0 : 49[1] -> 57[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO Channel 02/0 : 51[3] -> 59[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO Channel 14/0 : 53[5] -> 61[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO Channel 06/0 : 55[7] -> 63[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO Channel 08/0 : 50[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO Channel 06/0 : 49[1] -> 57[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO Channel 04/0 : 51[3] -> 59[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO Channel 08/0 : 55[7] -> 63[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO Channel 10/0 : 50[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO Channel 08/0 : 49[1] -> 57[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO Channel 06/0 : 51[3] -> 59[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO Channel 10/0 : 55[7] -> 63[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO Channel 12/0 : 50[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO Channel 10/0 : 49[1] -> 57[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO Channel 12/0 : 55[7] -> 63[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO Channel 08/0 : 51[3] -> 59[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO Channel 14/0 : 50[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO Channel 12/0 : 49[1] -> 57[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO Channel 14/0 : 55[7] -> 63[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO Channel 10/0 : 51[3] -> 59[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO Channel 14/0 : 49[1] -> 57[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO Channel 00/0 : 37[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO Channel 12/0 : 51[3] -> 59[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO Channel 02/0 : 37[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO Channel 01/0 : 29[5] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO Channel 14/0 : 51[3] -> 59[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153564:153737 [2] NCCL INFO Channel 00/0 : 82[2] -> 90[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO Channel 04/0 : 37[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO Channel 03/0 : 29[5] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153569:153739 [7] NCCL INFO Channel 00/0 : 87[7] -> 95[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO Channel 06/0 : 37[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153564:153737 [2] NCCL INFO Channel 04/0 : 82[2] -> 90[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO Channel 05/0 : 29[5] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO Channel 00/0 : 34[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153566:153741 [4] NCCL INFO Channel 00/0 : 84[4] -> 92[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153563:153740 [1] NCCL INFO Channel 00/0 : 81[1] -> 89[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153569:153739 [7] NCCL INFO Channel 02/0 : 87[7] -> 95[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO Channel 01/0 : 26[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO Channel 08/0 : 37[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153564:153737 [2] NCCL INFO Channel 06/0 : 82[2] -> 90[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153566:153741 [4] NCCL INFO Channel 02/0 : 84[4] -> 92[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO Channel 07/0 : 29[5] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO Channel 01/0 : 31[7] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO Channel 00/0 : 39[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO Channel 04/0 : 34[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153569:153739 [7] NCCL INFO Channel 04/0 : 87[7] -> 95[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153563:153740 [1] NCCL INFO Channel 02/0 : 81[1] -> 89[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153564:153737 [2] NCCL INFO Channel 08/0 : 82[2] -> 90[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153566:153741 [4] NCCL INFO Channel 06/0 : 84[4] -> 92[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO Channel 10/0 : 37[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO Channel 00/0 : 33[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO Channel 03/0 : 26[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO Channel 02/0 : 39[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO Channel 09/0 : 29[5] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO Channel 06/0 : 34[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO Channel 03/0 : 31[7] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153569:153739 [7] NCCL INFO Channel 06/0 : 87[7] -> 95[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153563:153740 [1] NCCL INFO Channel 04/0 : 81[1] -> 89[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO Channel 00/0 : 35[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO Channel 12/0 : 37[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO Channel 01/0 : 25[1] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO Channel 05/0 : 26[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO Channel 11/0 : 29[5] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153564:153737 [2] NCCL INFO Channel 10/0 : 82[2] -> 90[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153567:153742 [5] NCCL INFO Channel 00/0 : 85[5] -> 93[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153566:153741 [4] NCCL INFO Channel 08/0 : 84[4] -> 92[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO Channel 02/0 : 33[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO Channel 04/0 : 39[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO Channel 08/0 : 34[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO Channel 01/0 : 27[3] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO Channel 07/0 : 26[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO Channel 02/0 : 35[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO Channel 05/0 : 31[7] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO Channel 03/0 : 25[1] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153569:153739 [7] NCCL INFO Channel 08/0 : 87[7] -> 95[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153563:153740 [1] NCCL INFO Channel 06/0 : 81[1] -> 89[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO Channel 15/0 : 29[5] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153565:153738 [3] NCCL INFO Channel 00/0 : 83[3] -> 91[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153564:153737 [2] NCCL INFO Channel 12/0 : 82[2] -> 90[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO Channel 14/0 : 37[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO Channel 04/0 : 33[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO Channel 09/0 : 26[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153566:153741 [4] NCCL INFO Channel 10/0 : 84[4] -> 92[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153567:153742 [5] NCCL INFO Channel 02/0 : 85[5] -> 93[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153563:153740 [1] NCCL INFO Channel 08/0 : 81[1] -> 89[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO Channel 06/0 : 39[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO Channel 10/0 : 34[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO Channel 04/0 : 35[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO Channel 06/0 : 33[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO Channel 03/0 : 27[3] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO Channel 05/0 : 25[1] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO Channel 07/0 : 31[7] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO Channel 01/0 : 61[5] -> 93[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153569:153739 [7] NCCL INFO Channel 10/0 : 87[7] -> 95[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153565:153738 [3] NCCL INFO Channel 02/0 : 83[3] -> 91[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO Channel 08/0 : 39[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153564:153737 [2] NCCL INFO Channel 14/0 : 82[2] -> 90[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153567:153742 [5] NCCL INFO Channel 04/0 : 85[5] -> 93[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153566:153741 [4] NCCL INFO Channel 12/0 : 84[4] -> 92[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO Channel 11/0 : 26[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO Channel 05/0 : 27[3] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO Channel 07/0 : 25[1] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO Channel 12/0 : 34[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO Channel 06/0 : 35[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO Channel 08/0 : 33[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153563:153740 [1] NCCL INFO Channel 10/0 : 81[1] -> 89[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO Channel 10/0 : 39[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153569:153739 [7] NCCL INFO Channel 12/0 : 87[7] -> 95[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153565:153738 [3] NCCL INFO Channel 04/0 : 83[3] -> 91[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153566:153741 [4] NCCL INFO Channel 14/0 : 84[4] -> 92[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153567:153742 [5] NCCL INFO Channel 06/0 : 85[5] -> 93[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO Channel 03/0 : 61[5] -> 93[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO Channel 09/0 : 31[7] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153563:153740 [1] NCCL INFO Channel 12/0 : 81[1] -> 89[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO Channel 13/0 : 26[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO Channel 07/0 : 27[3] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO Channel 11/0 : 25[1] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO Channel 08/0 : 35[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO Channel 10/0 : 33[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO Channel 14/0 : 34[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO Channel 12/0 : 39[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153569:153739 [7] NCCL INFO Channel 14/0 : 87[7] -> 95[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153565:153738 [3] NCCL INFO Channel 06/0 : 83[3] -> 91[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153567:153742 [5] NCCL INFO Channel 08/0 : 85[5] -> 93[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153563:153740 [1] NCCL INFO Channel 14/0 : 81[1] -> 89[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO Channel 05/0 : 61[5] -> 93[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO Channel 15/0 : 26[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO Channel 11/0 : 31[7] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO Channel 09/0 : 27[3] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO Channel 13/0 : 25[1] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO Channel 10/0 : 35[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153565:153738 [3] NCCL INFO Channel 08/0 : 83[3] -> 91[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153567:153742 [5] NCCL INFO Channel 10/0 : 85[5] -> 93[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO Channel 12/0 : 33[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO Channel 14/0 : 39[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO Channel 01/0 : 58[2] -> 90[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153565:153738 [3] NCCL INFO Channel 10/0 : 83[3] -> 91[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153567:153742 [5] NCCL INFO Channel 12/0 : 85[5] -> 93[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO Channel 14/0 : 33[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO Channel 12/0 : 35[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO Channel 07/0 : 61[5] -> 93[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO Channel 13/0 : 31[7] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO Channel 15/0 : 25[1] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO Channel 13/0 : 27[3] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153565:153738 [3] NCCL INFO Channel 12/0 : 83[3] -> 91[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153567:153742 [5] NCCL INFO Channel 14/0 : 85[5] -> 93[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO Channel 14/0 : 35[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO Channel 03/0 : 58[2] -> 90[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153565:153738 [3] NCCL INFO Channel 14/0 : 83[3] -> 91[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO Channel 09/0 : 61[5] -> 93[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO Channel 15/0 : 27[3] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO Channel 01/0 : 63[7] -> 95[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO Channel 01/0 : 57[1] -> 89[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO Channel 05/0 : 58[2] -> 90[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO Channel 11/0 : 61[5] -> 93[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO Channel 01/0 : 59[3] -> 91[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO Channel 07/0 : 58[2] -> 90[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO Channel 03/0 : 57[1] -> 89[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO Channel 03/0 : 63[7] -> 95[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO Channel 15/0 : 61[5] -> 93[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO Channel 09/0 : 58[2] -> 90[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO Channel 03/0 : 59[3] -> 91[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO Channel 05/0 : 63[7] -> 95[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO Channel 05/0 : 57[1] -> 89[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO Channel 11/0 : 58[2] -> 90[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO Channel 05/0 : 59[3] -> 91[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO Channel 07/0 : 63[7] -> 95[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO Channel 07/0 : 57[1] -> 89[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO Channel 13/0 : 58[2] -> 90[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO Channel 07/0 : 59[3] -> 91[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO Channel 09/0 : 63[7] -> 95[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO Channel 11/0 : 57[1] -> 89[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO Channel 15/0 : 58[2] -> 90[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO Channel 09/0 : 59[3] -> 91[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO Channel 11/0 : 63[7] -> 95[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO Channel 13/0 : 57[1] -> 89[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO Channel 13/0 : 59[3] -> 91[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO Channel 13/0 : 63[7] -> 95[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO Channel 15/0 : 57[1] -> 89[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO Channel 15/0 : 59[3] -> 91[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272151:272326 [1] NCCL INFO Channel 00/0 : 105[1] -> 113[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272151:272326 [1] NCCL INFO Channel 02/0 : 105[1] -> 113[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272151:272326 [1] NCCL INFO Channel 03/0 : 105[1] -> 113[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272151:272326 [1] NCCL INFO Channel 04/0 : 105[1] -> 113[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272151:272326 [1] NCCL INFO Channel 05/0 : 105[1] -> 113[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272151:272326 [1] NCCL INFO Channel 06/0 : 105[1] -> 113[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272151:272326 [1] NCCL INFO Channel 07/0 : 105[1] -> 113[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272151:272326 [1] NCCL INFO Channel 08/0 : 105[1] -> 113[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272151:272326 [1] NCCL INFO Channel 10/0 : 105[1] -> 113[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272151:272326 [1] NCCL INFO Channel 11/0 : 105[1] -> 113[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272151:272326 [1] NCCL INFO Channel 12/0 : 105[1] -> 113[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272151:272326 [1] NCCL INFO Channel 13/0 : 105[1] -> 113[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272151:272326 [1] NCCL INFO Channel 14/0 : 105[1] -> 113[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272151:272326 [1] NCCL INFO Channel 15/0 : 105[1] -> 113[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272151:272326 [1] NCCL INFO Channel 00/0 : 113[1] -> 121[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272151:272326 [1] NCCL INFO Channel 02/0 : 113[1] -> 121[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272151:272326 [1] NCCL INFO Channel 04/0 : 113[1] -> 121[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272151:272326 [1] NCCL INFO Channel 06/0 : 113[1] -> 121[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272151:272326 [1] NCCL INFO Channel 08/0 : 113[1] -> 121[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272151:272326 [1] NCCL INFO Channel 10/0 : 113[1] -> 121[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272155:272329 [5] NCCL INFO Channel 00/0 : 109[5] -> 117[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272156:272325 [6] NCCL INFO Channel 00/0 : 110[6] -> 118[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272151:272326 [1] NCCL INFO Channel 12/0 : 113[1] -> 121[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272155:272329 [5] NCCL INFO Channel 01/0 : 109[5] -> 117[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272153:272328 [3] NCCL INFO Channel 00/0 : 107[3] -> 115[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272151:272326 [1] NCCL INFO Channel 14/0 : 113[1] -> 121[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272156:272325 [6] NCCL INFO Channel 01/0 : 110[6] -> 118[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272155:272329 [5] NCCL INFO Channel 02/0 : 109[5] -> 117[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272154:272330 [4] NCCL INFO Channel 00/0 : 108[4] -> 116[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272157:272324 [7] NCCL INFO Channel 00/0 : 111[7] -> 119[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272153:272328 [3] NCCL INFO Channel 01/0 : 107[3] -> 115[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272156:272325 [6] NCCL INFO Channel 02/0 : 110[6] -> 118[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272154:272330 [4] NCCL INFO Channel 01/0 : 108[4] -> 116[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272155:272329 [5] NCCL INFO Channel 03/0 : 109[5] -> 117[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272150:272327 [0] NCCL INFO Channel 01/0 : 104[0] -> 112[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272152:272331 [2] NCCL INFO Channel 00/0 : 106[2] -> 114[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272156:272325 [6] NCCL INFO Channel 03/0 : 110[6] -> 118[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272153:272328 [3] NCCL INFO Channel 02/0 : 107[3] -> 115[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272157:272324 [7] NCCL INFO Channel 01/0 : 111[7] -> 119[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272154:272330 [4] NCCL INFO Channel 02/0 : 108[4] -> 116[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272155:272329 [5] NCCL INFO Channel 04/0 : 109[5] -> 117[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272150:272327 [0] NCCL INFO Channel 02/0 : 104[0] -> 112[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272152:272331 [2] NCCL INFO Channel 01/0 : 106[2] -> 114[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272156:272325 [6] NCCL INFO Channel 04/0 : 110[6] -> 118[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272154:272330 [4] NCCL INFO Channel 03/0 : 108[4] -> 116[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272153:272328 [3] NCCL INFO Channel 04/0 : 107[3] -> 115[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272157:272324 [7] NCCL INFO Channel 02/0 : 111[7] -> 119[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272155:272329 [5] NCCL INFO Channel 06/0 : 109[5] -> 117[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272150:272327 [0] NCCL INFO Channel 03/0 : 104[0] -> 112[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272156:272325 [6] NCCL INFO Channel 05/0 : 110[6] -> 118[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272152:272331 [2] NCCL INFO Channel 03/0 : 106[2] -> 114[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272151:272326 [1] NCCL INFO Channel 00/0 : 97[1] -> 113[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272154:272330 [4] NCCL INFO Channel 05/0 : 108[4] -> 116[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272153:272328 [3] NCCL INFO Channel 05/0 : 107[3] -> 115[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272157:272324 [7] NCCL INFO Channel 03/0 : 111[7] -> 119[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272150:272327 [0] NCCL INFO Channel 04/0 : 104[0] -> 112[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272155:272329 [5] NCCL INFO Channel 07/0 : 109[5] -> 117[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272156:272325 [6] NCCL INFO Channel 07/0 : 110[6] -> 118[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272152:272331 [2] NCCL INFO Channel 04/0 : 106[2] -> 114[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272154:272330 [4] NCCL INFO Channel 06/0 : 108[4] -> 116[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272151:272326 [1] NCCL INFO Channel 02/0 : 97[1] -> 113[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272153:272328 [3] NCCL INFO Channel 06/0 : 107[3] -> 115[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272157:272324 [7] NCCL INFO Channel 04/0 : 111[7] -> 119[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272155:272329 [5] NCCL INFO Channel 08/0 : 109[5] -> 117[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272150:272327 [0] NCCL INFO Channel 05/0 : 104[0] -> 112[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272156:272325 [6] NCCL INFO Channel 08/0 : 110[6] -> 118[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272154:272330 [4] NCCL INFO Channel 07/0 : 108[4] -> 116[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272151:272326 [1] NCCL INFO Channel 04/0 : 97[1] -> 113[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272152:272331 [2] NCCL INFO Channel 05/0 : 106[2] -> 114[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272153:272328 [3] NCCL INFO Channel 07/0 : 107[3] -> 115[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272157:272324 [7] NCCL INFO Channel 05/0 : 111[7] -> 119[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272155:272329 [5] NCCL INFO Channel 09/0 : 109[5] -> 117[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272150:272327 [0] NCCL INFO Channel 06/0 : 104[0] -> 112[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272156:272325 [6] NCCL INFO Channel 09/0 : 110[6] -> 118[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272154:272330 [4] NCCL INFO Channel 08/0 : 108[4] -> 116[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272151:272326 [1] NCCL INFO Channel 06/0 : 97[1] -> 113[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272152:272331 [2] NCCL INFO Channel 06/0 : 106[2] -> 114[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272157:272324 [7] NCCL INFO Channel 06/0 : 111[7] -> 119[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272153:272328 [3] NCCL INFO Channel 08/0 : 107[3] -> 115[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272156:272325 [6] NCCL INFO Channel 10/0 : 110[6] -> 118[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272155:272329 [5] NCCL INFO Channel 10/0 : 109[5] -> 117[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272154:272330 [4] NCCL INFO Channel 09/0 : 108[4] -> 116[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272150:272327 [0] NCCL INFO Channel 07/0 : 104[0] -> 112[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272151:272326 [1] NCCL INFO Channel 08/0 : 97[1] -> 113[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272152:272331 [2] NCCL INFO Channel 07/0 : 106[2] -> 114[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272157:272324 [7] NCCL INFO Channel 08/0 : 111[7] -> 119[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272153:272328 [3] NCCL INFO Channel 09/0 : 107[3] -> 115[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272155:272329 [5] NCCL INFO Channel 11/0 : 109[5] -> 117[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272156:272325 [6] NCCL INFO Channel 11/0 : 110[6] -> 118[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272154:272330 [4] NCCL INFO Channel 10/0 : 108[4] -> 116[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272150:272327 [0] NCCL INFO Channel 09/0 : 104[0] -> 112[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272151:272326 [1] NCCL INFO Channel 10/0 : 97[1] -> 113[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272152:272331 [2] NCCL INFO Channel 08/0 : 106[2] -> 114[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272157:272324 [7] NCCL INFO Channel 09/0 : 111[7] -> 119[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272153:272328 [3] NCCL INFO Channel 10/0 : 107[3] -> 115[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272156:272325 [6] NCCL INFO Channel 12/0 : 110[6] -> 118[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272155:272329 [5] NCCL INFO Channel 12/0 : 109[5] -> 117[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272154:272330 [4] NCCL INFO Channel 11/0 : 108[4] -> 116[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272150:272327 [0] NCCL INFO Channel 10/0 : 104[0] -> 112[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272151:272326 [1] NCCL INFO Channel 12/0 : 97[1] -> 113[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272157:272324 [7] NCCL INFO Channel 10/0 : 111[7] -> 119[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272152:272331 [2] NCCL INFO Channel 09/0 : 106[2] -> 114[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272153:272328 [3] NCCL INFO Channel 12/0 : 107[3] -> 115[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272156:272325 [6] NCCL INFO Channel 13/0 : 110[6] -> 118[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272155:272329 [5] NCCL INFO Channel 14/0 : 109[5] -> 117[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272154:272330 [4] NCCL INFO Channel 13/0 : 108[4] -> 116[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272150:272327 [0] NCCL INFO Channel 11/0 : 104[0] -> 112[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272151:272326 [1] NCCL INFO Channel 14/0 : 97[1] -> 113[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272157:272324 [7] NCCL INFO Channel 11/0 : 111[7] -> 119[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272152:272331 [2] NCCL INFO Channel 11/0 : 106[2] -> 114[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272153:272328 [3] NCCL INFO Channel 13/0 : 107[3] -> 115[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272156:272325 [6] NCCL INFO Channel 15/0 : 110[6] -> 118[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272155:272329 [5] NCCL INFO Channel 15/0 : 109[5] -> 117[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272154:272330 [4] NCCL INFO Channel 14/0 : 108[4] -> 116[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272150:272327 [0] NCCL INFO Channel 12/0 : 104[0] -> 112[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272157:272324 [7] NCCL INFO Channel 12/0 : 111[7] -> 119[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272152:272331 [2] NCCL INFO Channel 12/0 : 106[2] -> 114[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272153:272328 [3] NCCL INFO Channel 14/0 : 107[3] -> 115[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272156:272325 [6] NCCL INFO Channel 00/0 : 118[6] -> 126[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272155:272329 [5] NCCL INFO Channel 00/0 : 117[5] -> 125[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272154:272330 [4] NCCL INFO Channel 15/0 : 108[4] -> 116[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272150:272327 [0] NCCL INFO Channel 13/0 : 104[0] -> 112[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272157:272324 [7] NCCL INFO Channel 13/0 : 111[7] -> 119[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272151:272326 [1] NCCL INFO Channel 00/0 : 113[1] -> 97[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272152:272331 [2] NCCL INFO Channel 13/0 : 106[2] -> 114[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272153:272328 [3] NCCL INFO Channel 15/0 : 107[3] -> 115[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272156:272325 [6] NCCL INFO Channel 02/0 : 118[6] -> 126[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272155:272329 [5] NCCL INFO Channel 02/0 : 117[5] -> 125[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272154:272330 [4] NCCL INFO Channel 00/0 : 116[4] -> 124[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272157:272324 [7] NCCL INFO Channel 14/0 : 111[7] -> 119[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272150:272327 [0] NCCL INFO Channel 14/0 : 104[0] -> 112[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272151:272326 [1] NCCL INFO Channel 02/0 : 113[1] -> 97[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-52:2722046:2722223 [1] NCCL INFO Channel 01/0 : 57[1] -> 121[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272152:272331 [2] NCCL INFO Channel 14/0 : 106[2] -> 114[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272153:272328 [3] NCCL INFO Channel 00/0 : 115[3] -> 123[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272156:272325 [6] NCCL INFO Channel 04/0 : 118[6] -> 126[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272155:272329 [5] NCCL INFO Channel 04/0 : 117[5] -> 125[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-52:2722046:2722223 [1] NCCL INFO Channel 03/0 : 57[1] -> 121[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272154:272330 [4] NCCL INFO Channel 02/0 : 116[4] -> 124[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272157:272324 [7] NCCL INFO Channel 00/0 : 119[7] -> 127[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272150:272327 [0] NCCL INFO Channel 15/0 : 104[0] -> 112[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272151:272326 [1] NCCL INFO Channel 04/0 : 113[1] -> 97[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-52:2722046:2722223 [1] NCCL INFO Channel 05/0 : 57[1] -> 121[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272156:272325 [6] NCCL INFO Channel 08/0 : 118[6] -> 126[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272152:272331 [2] NCCL INFO Channel 15/0 : 106[2] -> 114[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272153:272328 [3] NCCL INFO Channel 02/0 : 115[3] -> 123[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272155:272329 [5] NCCL INFO Channel 06/0 : 117[5] -> 125[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-52:2722046:2722223 [1] NCCL INFO Channel 07/0 : 57[1] -> 121[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272154:272330 [4] NCCL INFO Channel 06/0 : 116[4] -> 124[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272157:272324 [7] NCCL INFO Channel 02/0 : 119[7] -> 127[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272150:272327 [0] NCCL INFO Channel 02/0 : 112[0] -> 120[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272151:272326 [1] NCCL INFO Channel 06/0 : 113[1] -> 97[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272156:272325 [6] NCCL INFO Channel 10/0 : 118[6] -> 126[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-52:2722046:2722223 [1] NCCL INFO Channel 11/0 : 57[1] -> 121[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272152:272331 [2] NCCL INFO Channel 00/0 : 114[2] -> 122[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272153:272328 [3] NCCL INFO Channel 04/0 : 115[3] -> 123[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272155:272329 [5] NCCL INFO Channel 08/0 : 117[5] -> 125[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272154:272330 [4] NCCL INFO Channel 08/0 : 116[4] -> 124[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272157:272324 [7] NCCL INFO Channel 04/0 : 119[7] -> 127[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722046:2722223 [1] NCCL INFO Channel 13/0 : 57[1] -> 121[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272150:272327 [0] NCCL INFO Channel 04/0 : 112[0] -> 120[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272151:272326 [1] NCCL INFO Channel 08/0 : 113[1] -> 97[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272156:272325 [6] NCCL INFO Channel 12/0 : 118[6] -> 126[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272152:272331 [2] NCCL INFO Channel 04/0 : 114[2] -> 122[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272153:272328 [3] NCCL INFO Channel 06/0 : 115[3] -> 123[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272155:272329 [5] NCCL INFO Channel 10/0 : 117[5] -> 125[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272154:272330 [4] NCCL INFO Channel 10/0 : 116[4] -> 124[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722046:2722223 [1] NCCL INFO Channel 15/0 : 57[1] -> 121[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272157:272324 [7] NCCL INFO Channel 06/0 : 119[7] -> 127[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272151:272326 [1] NCCL INFO Channel 10/0 : 113[1] -> 97[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272156:272325 [6] NCCL INFO Channel 14/0 : 118[6] -> 126[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272152:272331 [2] NCCL INFO Channel 06/0 : 114[2] -> 122[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272150:272327 [0] NCCL INFO Channel 06/0 : 112[0] -> 120[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-52:2722046:2722223 [1] NCCL INFO Channel 01/0 : 121[1] -> 57[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272153:272328 [3] NCCL INFO Channel 08/0 : 115[3] -> 123[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272155:272329 [5] NCCL INFO Channel 12/0 : 117[5] -> 125[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272154:272330 [4] NCCL INFO Channel 12/0 : 116[4] -> 124[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272157:272324 [7] NCCL INFO Channel 08/0 : 119[7] -> 127[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272151:272326 [1] NCCL INFO Channel 12/0 : 113[1] -> 97[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-52:2722046:2722223 [1] NCCL INFO Channel 03/0 : 121[1] -> 57[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272152:272331 [2] NCCL INFO Channel 08/0 : 114[2] -> 122[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272150:272327 [0] NCCL INFO Channel 08/0 : 112[0] -> 120[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272153:272328 [3] NCCL INFO Channel 10/0 : 115[3] -> 123[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-52:2722046:2722223 [1] NCCL INFO Channel 05/0 : 121[1] -> 57[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272154:272330 [4] NCCL INFO Channel 14/0 : 116[4] -> 124[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272155:272329 [5] NCCL INFO Channel 14/0 : 117[5] -> 125[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272157:272324 [7] NCCL INFO Channel 10/0 : 119[7] -> 127[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272151:272326 [1] NCCL INFO Channel 14/0 : 113[1] -> 97[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272152:272331 [2] NCCL INFO Channel 10/0 : 114[2] -> 122[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272150:272327 [0] NCCL INFO Channel 10/0 : 112[0] -> 120[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-52:2722046:2722223 [1] NCCL INFO Channel 07/0 : 121[1] -> 57[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272153:272328 [3] NCCL INFO Channel 12/0 : 115[3] -> 123[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272157:272324 [7] NCCL INFO Channel 12/0 : 119[7] -> 127[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272152:272331 [2] NCCL INFO Channel 12/0 : 114[2] -> 122[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272150:272327 [0] NCCL INFO Channel 12/0 : 112[0] -> 120[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272153:272328 [3] NCCL INFO Channel 14/0 : 115[3] -> 123[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-52:2722046:2722223 [1] NCCL INFO Channel 11/0 : 121[1] -> 57[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272157:272324 [7] NCCL INFO Channel 14/0 : 119[7] -> 127[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272152:272331 [2] NCCL INFO Channel 14/0 : 114[2] -> 122[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-52:2722046:2722223 [1] NCCL INFO Channel 13/0 : 121[1] -> 57[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272150:272327 [0] NCCL INFO Channel 14/0 : 112[0] -> 120[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272156:272325 [6] NCCL INFO Channel 00/0 : 102[6] -> 118[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-52:2722051:2722224 [6] NCCL INFO Channel 01/0 : 62[6] -> 126[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272156:272325 [6] NCCL INFO Channel 02/0 : 102[6] -> 118[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-52:2722046:2722223 [1] NCCL INFO Channel 15/0 : 121[1] -> 57[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-52:2722051:2722224 [6] NCCL INFO Channel 03/0 : 62[6] -> 126[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272156:272325 [6] NCCL INFO Channel 04/0 : 102[6] -> 118[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-52:2722051:2722224 [6] NCCL INFO Channel 05/0 : 62[6] -> 126[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272155:272329 [5] NCCL INFO Channel 00/0 : 101[5] -> 117[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272154:272330 [4] NCCL INFO Channel 00/0 : 100[4] -> 116[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272156:272325 [6] NCCL INFO Channel 08/0 : 102[6] -> 118[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO Channel 01/0 : 60[4] -> 124[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272155:272329 [5] NCCL INFO Channel 02/0 : 101[5] -> 117[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272154:272330 [4] NCCL INFO Channel 02/0 : 100[4] -> 116[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272156:272325 [6] NCCL INFO Channel 10/0 : 102[6] -> 118[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272153:272328 [3] NCCL INFO Channel 00/0 : 99[3] -> 115[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-52:2722051:2722224 [6] NCCL INFO Channel 07/0 : 62[6] -> 126[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272150:272327 [0] NCCL INFO Channel 02/0 : 96[0] -> 112[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272157:272324 [7] NCCL INFO Channel 00/0 : 103[7] -> 119[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722048:2722220 [3] NCCL INFO Channel 01/0 : 59[3] -> 123[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-52:2722050:2722225 [5] NCCL INFO Channel 01/0 : 61[5] -> 125[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272152:272331 [2] NCCL INFO Channel 00/0 : 98[2] -> 114[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272155:272329 [5] NCCL INFO Channel 04/0 : 101[5] -> 117[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272154:272330 [4] NCCL INFO Channel 06/0 : 100[4] -> 116[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272156:272325 [6] NCCL INFO Channel 12/0 : 102[6] -> 118[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272153:272328 [3] NCCL INFO Channel 02/0 : 99[3] -> 115[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272150:272327 [0] NCCL INFO Channel 04/0 : 96[0] -> 112[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO Channel 03/0 : 60[4] -> 124[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272157:272324 [7] NCCL INFO Channel 02/0 : 103[7] -> 119[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722047:2722221 [2] NCCL INFO Channel 01/0 : 58[2] -> 122[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272155:272329 [5] NCCL INFO Channel 06/0 : 101[5] -> 117[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-52:2722045:2722219 [0] NCCL INFO Channel 01/0 : 56[0] -> 120[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-52:2722051:2722224 [6] NCCL INFO Channel 09/0 : 62[6] -> 126[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272152:272331 [2] NCCL INFO Channel 04/0 : 98[2] -> 114[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-52:2722052:2722226 [7] NCCL INFO Channel 01/0 : 63[7] -> 127[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722048:2722220 [3] NCCL INFO Channel 03/0 : 59[3] -> 123[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-52:2722050:2722225 [5] NCCL INFO Channel 03/0 : 61[5] -> 125[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235652:235822 [4] NCCL INFO Channel 07/0 : 108[4] -> 116[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235654:235821 [6] NCCL INFO Channel 09/0 : 110[6] -> 118[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO Channel 05/0 : 60[4] -> 124[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235648:235826 [0] NCCL INFO Channel 07/0 : 104[0] -> 112[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272156:272325 [6] NCCL INFO Channel 14/0 : 102[6] -> 118[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272154:272330 [4] NCCL INFO Channel 08/0 : 100[4] -> 116[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722047:2722221 [2] NCCL INFO Channel 03/0 : 58[2] -> 122[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235650:235824 [2] NCCL INFO Channel 06/0 : 106[2] -> 114[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235651:235823 [3] NCCL INFO Channel 08/0 : 107[3] -> 115[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235652:235822 [4] NCCL INFO Channel 08/0 : 108[4] -> 116[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272153:272328 [3] NCCL INFO Channel 04/0 : 99[3] -> 115[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272150:272327 [0] NCCL INFO Channel 06/0 : 96[0] -> 112[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-52:2722051:2722224 [6] NCCL INFO Channel 11/0 : 62[6] -> 126[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235654:235821 [6] NCCL INFO Channel 10/0 : 110[6] -> 118[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272157:272324 [7] NCCL INFO Channel 04/0 : 103[7] -> 119[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235648:235826 [0] NCCL INFO Channel 09/0 : 104[0] -> 112[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272155:272329 [5] NCCL INFO Channel 08/0 : 101[5] -> 117[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235651:235823 [3] NCCL INFO Channel 09/0 : 107[3] -> 115[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235650:235824 [2] NCCL INFO Channel 07/0 : 106[2] -> 114[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272152:272331 [2] NCCL INFO Channel 06/0 : 98[2] -> 114[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235652:235822 [4] NCCL INFO Channel 09/0 : 108[4] -> 116[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235654:235821 [6] NCCL INFO Channel 11/0 : 110[6] -> 118[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-52:2722045:2722219 [0] NCCL INFO Channel 03/0 : 56[0] -> 120[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-52:2722048:2722220 [3] NCCL INFO Channel 05/0 : 59[3] -> 123[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235648:235826 [0] NCCL INFO Channel 10/0 : 104[0] -> 112[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-52:2722052:2722226 [7] NCCL INFO Channel 03/0 : 63[7] -> 127[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722050:2722225 [5] NCCL INFO Channel 05/0 : 61[5] -> 125[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235650:235824 [2] NCCL INFO Channel 08/0 : 106[2] -> 114[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235651:235823 [3] NCCL INFO Channel 10/0 : 107[3] -> 115[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272154:272330 [4] NCCL INFO Channel 10/0 : 100[4] -> 116[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO Channel 07/0 : 60[4] -> 124[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722047:2722221 [2] NCCL INFO Channel 05/0 : 58[2] -> 122[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235652:235822 [4] NCCL INFO Channel 10/0 : 108[4] -> 116[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272153:272328 [3] NCCL INFO Channel 06/0 : 99[3] -> 115[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272157:272324 [7] NCCL INFO Channel 06/0 : 103[7] -> 119[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722051:2722224 [6] NCCL INFO Channel 13/0 : 62[6] -> 126[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235654:235821 [6] NCCL INFO Channel 12/0 : 110[6] -> 118[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235648:235826 [0] NCCL INFO Channel 11/0 : 104[0] -> 112[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272150:272327 [0] NCCL INFO Channel 08/0 : 96[0] -> 112[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235650:235824 [2] NCCL INFO Channel 09/0 : 106[2] -> 114[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272155:272329 [5] NCCL INFO Channel 10/0 : 101[5] -> 117[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235651:235823 [3] NCCL INFO Channel 12/0 : 107[3] -> 115[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235652:235822 [4] NCCL INFO Channel 11/0 : 108[4] -> 116[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235654:235821 [6] NCCL INFO Channel 13/0 : 110[6] -> 118[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235648:235826 [0] NCCL INFO Channel 12/0 : 104[0] -> 112[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235650:235824 [2] NCCL INFO Channel 11/0 : 106[2] -> 114[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-52:2722045:2722219 [0] NCCL INFO Channel 05/0 : 56[0] -> 120[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235651:235823 [3] NCCL INFO Channel 13/0 : 107[3] -> 115[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-52:2722050:2722225 [5] NCCL INFO Channel 07/0 : 61[5] -> 125[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-52:2722048:2722220 [3] NCCL INFO Channel 07/0 : 59[3] -> 123[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235652:235822 [4] NCCL INFO Channel 13/0 : 108[4] -> 116[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272152:272331 [2] NCCL INFO Channel 08/0 : 98[2] -> 114[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272154:272330 [4] NCCL INFO Channel 12/0 : 100[4] -> 116[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722052:2722226 [7] NCCL INFO Channel 05/0 : 63[7] -> 127[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235648:235826 [0] NCCL INFO Channel 13/0 : 104[0] -> 112[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235654:235821 [6] NCCL INFO Channel 15/0 : 110[6] -> 118[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235650:235824 [2] NCCL INFO Channel 12/0 : 106[2] -> 114[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235651:235823 [3] NCCL INFO Channel 14/0 : 107[3] -> 115[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235652:235822 [4] NCCL INFO Channel 14/0 : 108[4] -> 116[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235648:235826 [0] NCCL INFO Channel 14/0 : 104[0] -> 112[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272153:272328 [3] NCCL INFO Channel 08/0 : 99[3] -> 115[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235650:235824 [2] NCCL INFO Channel 13/0 : 106[2] -> 114[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272157:272324 [7] NCCL INFO Channel 08/0 : 103[7] -> 119[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722047:2722221 [2] NCCL INFO Channel 07/0 : 58[2] -> 122[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO Channel 09/0 : 60[4] -> 124[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722051:2722224 [6] NCCL INFO Channel 15/0 : 62[6] -> 126[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-52:2722045:2722219 [0] NCCL INFO Channel 07/0 : 56[0] -> 120[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235651:235823 [3] NCCL INFO Channel 15/0 : 107[3] -> 115[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272156:272325 [6] NCCL INFO Channel 00/0 : 118[6] -> 102[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-52:2722048:2722220 [3] NCCL INFO Channel 09/0 : 59[3] -> 123[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-52:2722050:2722225 [5] NCCL INFO Channel 09/0 : 61[5] -> 125[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-52:2722052:2722226 [7] NCCL INFO Channel 07/0 : 63[7] -> 127[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235652:235822 [4] NCCL INFO Channel 15/0 : 108[4] -> 116[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235648:235826 [0] NCCL INFO Channel 15/0 : 104[0] -> 112[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235650:235824 [2] NCCL INFO Channel 14/0 : 106[2] -> 114[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235650:235824 [2] NCCL INFO Channel 15/0 : 106[2] -> 114[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235649:235827 [1] NCCL INFO Channel 01/0 : 89[1] -> 105[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235649:235827 [1] NCCL INFO Channel 03/0 : 89[1] -> 105[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272150:272327 [0] NCCL INFO Channel 10/0 : 96[0] -> 112[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235649:235827 [1] NCCL INFO Channel 05/0 : 89[1] -> 105[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272155:272329 [5] NCCL INFO Channel 12/0 : 101[5] -> 117[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272154:272330 [4] NCCL INFO Channel 14/0 : 100[4] -> 116[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO Channel 11/0 : 60[4] -> 124[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235649:235827 [1] NCCL INFO Channel 07/0 : 89[1] -> 105[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272152:272331 [2] NCCL INFO Channel 10/0 : 98[2] -> 114[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-52:2722051:2722224 [6] NCCL INFO Channel 01/0 : 126[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235649:235827 [1] NCCL INFO Channel 11/0 : 89[1] -> 105[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272153:272328 [3] NCCL INFO Channel 10/0 : 99[3] -> 115[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-52:2722047:2722221 [2] NCCL INFO Channel 09/0 : 58[2] -> 122[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235649:235827 [1] NCCL INFO Channel 13/0 : 89[1] -> 105[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-52:2722045:2722219 [0] NCCL INFO Channel 09/0 : 56[0] -> 120[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235649:235827 [1] NCCL INFO Channel 15/0 : 89[1] -> 105[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235654:235821 [6] NCCL INFO Channel 01/0 : 94[6] -> 110[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235654:235821 [6] NCCL INFO Channel 03/0 : 94[6] -> 110[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235654:235821 [6] NCCL INFO Channel 05/0 : 94[6] -> 110[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235653:235825 [5] NCCL INFO Channel 01/0 : 93[5] -> 109[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272156:272325 [6] NCCL INFO Channel 02/0 : 118[6] -> 102[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235654:235821 [6] NCCL INFO Channel 07/0 : 94[6] -> 110[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272157:272324 [7] NCCL INFO Channel 10/0 : 103[7] -> 119[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722048:2722220 [3] NCCL INFO Channel 13/0 : 59[3] -> 123[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-52:2722050:2722225 [5] NCCL INFO Channel 11/0 : 61[5] -> 125[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-52:2722052:2722226 [7] NCCL INFO Channel 09/0 : 63[7] -> 127[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235653:235825 [5] NCCL INFO Channel 03/0 : 93[5] -> 109[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272150:272327 [0] NCCL INFO Channel 12/0 : 96[0] -> 112[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272155:272329 [5] NCCL INFO Channel 14/0 : 101[5] -> 117[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO Channel 13/0 : 60[4] -> 124[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235651:235823 [3] NCCL INFO Channel 01/0 : 91[3] -> 107[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272152:272331 [2] NCCL INFO Channel 12/0 : 98[2] -> 114[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272153:272328 [3] NCCL INFO Channel 12/0 : 99[3] -> 115[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-52:2722051:2722224 [6] NCCL INFO Channel 03/0 : 126[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235654:235821 [6] NCCL INFO Channel 09/0 : 94[6] -> 110[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-52:2722047:2722221 [2] NCCL INFO Channel 11/0 : 58[2] -> 122[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-52:2722045:2722219 [0] NCCL INFO Channel 11/0 : 56[0] -> 120[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235655:235828 [7] NCCL INFO Channel 01/0 : 95[7] -> 111[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235648:235826 [0] NCCL INFO Channel 01/0 : 88[0] -> 104[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235653:235825 [5] NCCL INFO Channel 05/0 : 93[5] -> 109[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235650:235824 [2] NCCL INFO Channel 01/0 : 90[2] -> 106[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235651:235823 [3] NCCL INFO Channel 03/0 : 91[3] -> 107[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235654:235821 [6] NCCL INFO Channel 11/0 : 94[6] -> 110[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272156:272325 [6] NCCL INFO Channel 04/0 : 118[6] -> 102[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-52:2722048:2722220 [3] NCCL INFO Channel 15/0 : 59[3] -> 123[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235655:235828 [7] NCCL INFO Channel 03/0 : 95[7] -> 111[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272157:272324 [7] NCCL INFO Channel 12/0 : 103[7] -> 119[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722052:2722226 [7] NCCL INFO Channel 11/0 : 63[7] -> 127[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722050:2722225 [5] NCCL INFO Channel 15/0 : 61[5] -> 125[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235648:235826 [0] NCCL INFO Channel 03/0 : 88[0] -> 104[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272150:272327 [0] NCCL INFO Channel 14/0 : 96[0] -> 112[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272152:272331 [2] NCCL INFO Channel 14/0 : 98[2] -> 114[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272153:272328 [3] NCCL INFO Channel 14/0 : 99[3] -> 115[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO Channel 15/0 : 60[4] -> 124[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235653:235825 [5] NCCL INFO Channel 07/0 : 93[5] -> 109[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272156:272325 [6] NCCL INFO Channel 08/0 : 118[6] -> 102[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-52:2722051:2722224 [6] NCCL INFO Channel 05/0 : 126[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235650:235824 [2] NCCL INFO Channel 03/0 : 90[2] -> 106[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-52:2722047:2722221 [2] NCCL INFO Channel 13/0 : 58[2] -> 122[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235654:235821 [6] NCCL INFO Channel 13/0 : 94[6] -> 110[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235651:235823 [3] NCCL INFO Channel 05/0 : 91[3] -> 107[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-52:2722045:2722219 [0] NCCL INFO Channel 13/0 : 56[0] -> 120[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235648:235826 [0] NCCL INFO Channel 05/0 : 88[0] -> 104[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235655:235828 [7] NCCL INFO Channel 05/0 : 95[7] -> 111[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235653:235825 [5] NCCL INFO Channel 09/0 : 93[5] -> 109[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235650:235824 [2] NCCL INFO Channel 05/0 : 90[2] -> 106[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272157:272324 [7] NCCL INFO Channel 14/0 : 103[7] -> 119[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235654:235821 [6] NCCL INFO Channel 15/0 : 94[6] -> 110[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235651:235823 [3] NCCL INFO Channel 07/0 : 91[3] -> 107[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272156:272325 [6] NCCL INFO Channel 10/0 : 118[6] -> 102[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235648:235826 [0] NCCL INFO Channel 07/0 : 88[0] -> 104[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235653:235825 [5] NCCL INFO Channel 11/0 : 93[5] -> 109[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235655:235828 [7] NCCL INFO Channel 07/0 : 95[7] -> 111[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO Channel 01/0 : 64[0] -> 72[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-52:2722050:2722225 [5] NCCL INFO Channel 01/0 : 125[5] -> 61[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-52:2722052:2722226 [7] NCCL INFO Channel 13/0 : 63[7] -> 127[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722048:2722220 [3] NCCL INFO Channel 01/0 : 123[3] -> 59[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-52:2722051:2722224 [6] NCCL INFO Channel 07/0 : 126[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235650:235824 [2] NCCL INFO Channel 07/0 : 90[2] -> 106[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235651:235823 [3] NCCL INFO Channel 09/0 : 91[3] -> 107[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272156:272325 [6] NCCL INFO Channel 12/0 : 118[6] -> 102[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235648:235826 [0] NCCL INFO Channel 09/0 : 88[0] -> 104[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272155:272329 [5] NCCL INFO Channel 00/0 : 117[5] -> 101[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO Channel 01/0 : 124[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO Channel 03/0 : 64[0] -> 72[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-52:2722047:2722221 [2] NCCL INFO Channel 15/0 : 58[2] -> 122[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-52:2722045:2722219 [0] NCCL INFO Channel 15/0 : 56[0] -> 120[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235653:235825 [5] NCCL INFO Channel 15/0 : 93[5] -> 109[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235655:235828 [7] NCCL INFO Channel 09/0 : 95[7] -> 111[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235650:235824 [2] NCCL INFO Channel 09/0 : 90[2] -> 106[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272150:272327 [0] NCCL INFO Channel 02/0 : 112[0] -> 96[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235648:235826 [0] NCCL INFO Channel 11/0 : 88[0] -> 104[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235651:235823 [3] NCCL INFO Channel 13/0 : 91[3] -> 107[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235655:235828 [7] NCCL INFO Channel 11/0 : 95[7] -> 111[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235652:235822 [4] NCCL INFO Channel 01/0 : 92[4] -> 108[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235650:235824 [2] NCCL INFO Channel 11/0 : 90[2] -> 106[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-52:2722052:2722226 [7] NCCL INFO Channel 01/0 : 127[7] -> 63[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235648:235826 [0] NCCL INFO Channel 13/0 : 88[0] -> 104[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235651:235823 [3] NCCL INFO Channel 15/0 : 91[3] -> 107[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-52:2722048:2722220 [3] NCCL INFO Channel 03/0 : 123[3] -> 59[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235655:235828 [7] NCCL INFO Channel 13/0 : 95[7] -> 111[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO Channel 05/0 : 64[0] -> 72[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-52:2722051:2722224 [6] NCCL INFO Channel 09/0 : 126[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO Channel 03/0 : 124[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722047:2722221 [2] NCCL INFO Channel 01/0 : 122[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235650:235824 [2] NCCL INFO Channel 13/0 : 90[2] -> 106[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235652:235822 [4] NCCL INFO Channel 03/0 : 92[4] -> 108[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235648:235826 [0] NCCL INFO Channel 15/0 : 88[0] -> 104[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272156:272325 [6] NCCL INFO Channel 14/0 : 118[6] -> 102[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-52:2722045:2722219 [0] NCCL INFO Channel 01/0 : 120[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235650:235824 [2] NCCL INFO Channel 15/0 : 90[2] -> 106[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272152:272331 [2] NCCL INFO Channel 00/0 : 114[2] -> 98[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235652:235822 [4] NCCL INFO Channel 05/0 : 92[4] -> 108[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272153:272328 [3] NCCL INFO Channel 00/0 : 115[3] -> 99[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235652:235822 [4] NCCL INFO Channel 07/0 : 92[4] -> 108[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272155:272329 [5] NCCL INFO Channel 02/0 : 117[5] -> 101[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235652:235822 [4] NCCL INFO Channel 09/0 : 92[4] -> 108[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272157:272324 [7] NCCL INFO Channel 00/0 : 119[7] -> 103[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235652:235822 [4] NCCL INFO Channel 11/0 : 92[4] -> 108[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272150:272327 [0] NCCL INFO Channel 04/0 : 112[0] -> 96[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235652:235822 [4] NCCL INFO Channel 13/0 : 92[4] -> 108[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO Channel 07/0 : 64[0] -> 72[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-52:2722050:2722225 [5] NCCL INFO Channel 03/0 : 125[5] -> 61[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-52:2722052:2722226 [7] NCCL INFO Channel 03/0 : 127[7] -> 63[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722048:2722220 [3] NCCL INFO Channel 05/0 : 123[3] -> 59[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272152:272331 [2] NCCL INFO Channel 04/0 : 114[2] -> 98[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272153:272328 [3] NCCL INFO Channel 02/0 : 115[3] -> 99[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235652:235822 [4] NCCL INFO Channel 15/0 : 92[4] -> 108[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272155:272329 [5] NCCL INFO Channel 04/0 : 117[5] -> 101[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272157:272324 [7] NCCL INFO Channel 02/0 : 119[7] -> 103[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272150:272327 [0] NCCL INFO Channel 06/0 : 112[0] -> 96[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO Channel 09/0 : 64[0] -> 72[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO Channel 05/0 : 124[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722051:2722224 [6] NCCL INFO Channel 11/0 : 126[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-52:2722045:2722219 [0] NCCL INFO Channel 03/0 : 120[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-52:2722047:2722221 [2] NCCL INFO Channel 03/0 : 122[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-52:2722050:2722225 [5] NCCL INFO Channel 05/0 : 125[5] -> 61[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-52:2722048:2722220 [3] NCCL INFO Channel 07/0 : 123[3] -> 59[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-52:2722052:2722226 [7] NCCL INFO Channel 05/0 : 127[7] -> 63[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272152:272331 [2] NCCL INFO Channel 06/0 : 114[2] -> 98[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO Channel 11/0 : 64[0] -> 72[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272153:272328 [3] NCCL INFO Channel 04/0 : 115[3] -> 99[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272155:272329 [5] NCCL INFO Channel 06/0 : 117[5] -> 101[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272157:272324 [7] NCCL INFO Channel 04/0 : 119[7] -> 103[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272150:272327 [0] NCCL INFO Channel 08/0 : 112[0] -> 96[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-52:2722051:2722224 [6] NCCL INFO Channel 13/0 : 126[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO Channel 07/0 : 124[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722045:2722219 [0] NCCL INFO Channel 05/0 : 120[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO Channel 13/0 : 64[0] -> 72[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272152:272331 [2] NCCL INFO Channel 08/0 : 114[2] -> 98[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272153:272328 [3] NCCL INFO Channel 06/0 : 115[3] -> 99[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272155:272329 [5] NCCL INFO Channel 08/0 : 117[5] -> 101[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-52:2722047:2722221 [2] NCCL INFO Channel 05/0 : 122[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-52:2722050:2722225 [5] NCCL INFO Channel 07/0 : 125[5] -> 61[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-52:2722048:2722220 [3] NCCL INFO Channel 09/0 : 123[3] -> 59[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-52:2722052:2722226 [7] NCCL INFO Channel 07/0 : 127[7] -> 63[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722051:2722224 [6] NCCL INFO Channel 15/0 : 126[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO Channel 09/0 : 124[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722045:2722219 [0] NCCL INFO Channel 07/0 : 120[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO Channel 15/0 : 64[0] -> 72[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272157:272324 [7] NCCL INFO Channel 06/0 : 119[7] -> 103[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272150:272327 [0] NCCL INFO Channel 10/0 : 112[0] -> 96[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272152:272331 [2] NCCL INFO Channel 10/0 : 114[2] -> 98[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272153:272328 [3] NCCL INFO Channel 08/0 : 115[3] -> 99[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-52:2722047:2722221 [2] NCCL INFO Channel 07/0 : 122[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-52:2722048:2722220 [3] NCCL INFO Channel 13/0 : 123[3] -> 59[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-52:2722050:2722225 [5] NCCL INFO Channel 09/0 : 125[5] -> 61[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-52:2722052:2722226 [7] NCCL INFO Channel 09/0 : 127[7] -> 63[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO Channel 11/0 : 124[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722045:2722219 [0] NCCL INFO Channel 09/0 : 120[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO Channel 01/0 : 72[0] -> 80[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272155:272329 [5] NCCL INFO Channel 10/0 : 117[5] -> 101[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272154:272330 [4] NCCL INFO Channel 00/0 : 116[4] -> 100[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272157:272324 [7] NCCL INFO Channel 08/0 : 119[7] -> 103[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272150:272327 [0] NCCL INFO Channel 12/0 : 112[0] -> 96[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-52:2722047:2722221 [2] NCCL INFO Channel 09/0 : 122[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272152:272331 [2] NCCL INFO Channel 12/0 : 114[2] -> 98[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272153:272328 [3] NCCL INFO Channel 10/0 : 115[3] -> 99[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-52:2722048:2722220 [3] NCCL INFO Channel 15/0 : 123[3] -> 59[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO Channel 02/0 : 72[0] -> 80[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272155:272329 [5] NCCL INFO Channel 12/0 : 117[5] -> 101[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272154:272330 [4] NCCL INFO Channel 02/0 : 116[4] -> 100[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722052:2722226 [7] NCCL INFO Channel 11/0 : 127[7] -> 63[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722050:2722225 [5] NCCL INFO Channel 11/0 : 125[5] -> 61[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO Channel 13/0 : 124[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722045:2722219 [0] NCCL INFO Channel 11/0 : 120[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-52:2722047:2722221 [2] NCCL INFO Channel 11/0 : 122[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272157:272324 [7] NCCL INFO Channel 10/0 : 119[7] -> 103[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272150:272327 [0] NCCL INFO Channel 14/0 : 112[0] -> 96[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO Channel 03/0 : 72[0] -> 80[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-52:2722052:2722226 [7] NCCL INFO Channel 13/0 : 127[7] -> 63[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722050:2722225 [5] NCCL INFO Channel 15/0 : 125[5] -> 61[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO Channel 15/0 : 124[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722047:2722221 [2] NCCL INFO Channel 13/0 : 122[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-52:2722045:2722219 [0] NCCL INFO Channel 13/0 : 120[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272153:272328 [3] NCCL INFO Channel 12/0 : 115[3] -> 99[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272152:272331 [2] NCCL INFO Channel 14/0 : 114[2] -> 98[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272154:272330 [4] NCCL INFO Channel 06/0 : 116[4] -> 100[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272155:272329 [5] NCCL INFO Channel 14/0 : 117[5] -> 101[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272157:272324 [7] NCCL INFO Channel 12/0 : 119[7] -> 103[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO Channel 04/0 : 72[0] -> 80[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272153:272328 [3] NCCL INFO Channel 14/0 : 115[3] -> 99[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-52:2722047:2722221 [2] NCCL INFO Channel 15/0 : 122[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO Channel 05/0 : 72[0] -> 80[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272154:272330 [4] NCCL INFO Channel 08/0 : 116[4] -> 100[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722045:2722219 [0] NCCL INFO Channel 15/0 : 120[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272157:272324 [7] NCCL INFO Channel 14/0 : 119[7] -> 103[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272154:272330 [4] NCCL INFO Channel 10/0 : 116[4] -> 100[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO Channel 06/0 : 72[0] -> 80[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272154:272330 [4] NCCL INFO Channel 12/0 : 116[4] -> 100[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272154:272330 [4] NCCL INFO Channel 14/0 : 116[4] -> 100[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO Channel 07/0 : 72[0] -> 80[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO Channel 09/0 : 72[0] -> 80[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO Channel 10/0 : 72[0] -> 80[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO Channel 11/0 : 72[0] -> 80[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO Channel 12/0 : 72[0] -> 80[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO Channel 13/0 : 72[0] -> 80[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO Channel 14/0 : 72[0] -> 80[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO Channel 15/0 : 72[0] -> 80[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO Channel 01/0 : 68[4] -> 76[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO Channel 01/0 : 70[6] -> 78[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO Channel 01/0 : 71[7] -> 79[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO Channel 03/0 : 68[4] -> 76[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO Channel 03/0 : 70[6] -> 78[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO Channel 05/0 : 68[4] -> 76[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO Channel 03/0 : 71[7] -> 79[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO Channel 05/0 : 70[6] -> 78[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO Channel 01/0 : 69[5] -> 77[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO Channel 01/0 : 65[1] -> 73[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO Channel 07/0 : 68[4] -> 76[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO Channel 05/0 : 71[7] -> 79[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO Channel 01/0 : 66[2] -> 74[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO Channel 07/0 : 70[6] -> 78[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO Channel 03/0 : 69[5] -> 77[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO Channel 09/0 : 68[4] -> 76[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO Channel 03/0 : 65[1] -> 73[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO Channel 07/0 : 71[7] -> 79[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO Channel 09/0 : 70[6] -> 78[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO Channel 01/0 : 67[3] -> 75[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO Channel 03/0 : 66[2] -> 74[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO Channel 05/0 : 69[5] -> 77[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO Channel 11/0 : 68[4] -> 76[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO Channel 05/0 : 65[1] -> 73[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO Channel 09/0 : 71[7] -> 79[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO Channel 11/0 : 70[6] -> 78[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO Channel 03/0 : 67[3] -> 75[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO Channel 05/0 : 66[2] -> 74[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO Channel 13/0 : 68[4] -> 76[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO Channel 07/0 : 69[5] -> 77[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO Channel 11/0 : 71[7] -> 79[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO Channel 07/0 : 65[1] -> 73[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO Channel 13/0 : 70[6] -> 78[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO Channel 07/0 : 66[2] -> 74[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO Channel 15/0 : 68[4] -> 76[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO Channel 05/0 : 67[3] -> 75[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO Channel 13/0 : 71[7] -> 79[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO Channel 09/0 : 69[5] -> 77[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO Channel 11/0 : 65[1] -> 73[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO Channel 15/0 : 70[6] -> 78[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO Channel 09/0 : 66[2] -> 74[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO Channel 00/0 : 76[4] -> 84[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO Channel 07/0 : 67[3] -> 75[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO Channel 00/0 : 79[7] -> 87[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO Channel 11/0 : 69[5] -> 77[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO Channel 00/0 : 78[6] -> 86[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO Channel 13/0 : 65[1] -> 73[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO Channel 01/0 : 76[4] -> 84[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO Channel 11/0 : 66[2] -> 74[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO Channel 01/0 : 79[7] -> 87[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO Channel 09/0 : 67[3] -> 75[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO Channel 01/0 : 78[6] -> 86[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO Channel 15/0 : 69[5] -> 77[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO Channel 02/0 : 76[4] -> 84[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO Channel 15/0 : 65[1] -> 73[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO Channel 13/0 : 66[2] -> 74[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO Channel 02/0 : 79[7] -> 87[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO Channel 13/0 : 67[3] -> 75[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO Channel 02/0 : 78[6] -> 86[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO Channel 00/0 : 77[5] -> 85[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO Channel 03/0 : 76[4] -> 84[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO Channel 00/0 : 73[1] -> 81[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO Channel 03/0 : 79[7] -> 87[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO Channel 15/0 : 66[2] -> 74[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO Channel 03/0 : 78[6] -> 86[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO Channel 15/0 : 67[3] -> 75[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO Channel 01/0 : 77[5] -> 85[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO Channel 05/0 : 76[4] -> 84[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO Channel 02/0 : 73[1] -> 81[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO Channel 04/0 : 79[7] -> 87[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO Channel 00/0 : 74[2] -> 82[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO Channel 04/0 : 78[6] -> 86[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO Channel 00/0 : 75[3] -> 83[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO Channel 06/0 : 76[4] -> 84[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO Channel 02/0 : 77[5] -> 85[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO Channel 03/0 : 73[1] -> 81[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO Channel 05/0 : 79[7] -> 87[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO Channel 05/0 : 78[6] -> 86[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO Channel 01/0 : 74[2] -> 82[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO Channel 01/0 : 75[3] -> 83[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO Channel 07/0 : 76[4] -> 84[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO Channel 03/0 : 77[5] -> 85[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO Channel 04/0 : 73[1] -> 81[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO Channel 06/0 : 79[7] -> 87[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO Channel 07/0 : 78[6] -> 86[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO Channel 03/0 : 74[2] -> 82[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO Channel 02/0 : 75[3] -> 83[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO Channel 08/0 : 76[4] -> 84[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO Channel 04/0 : 77[5] -> 85[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO Channel 08/0 : 79[7] -> 87[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO Channel 05/0 : 73[1] -> 81[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO Channel 08/0 : 78[6] -> 86[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO Channel 04/0 : 74[2] -> 82[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO Channel 04/0 : 75[3] -> 83[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO Channel 09/0 : 76[4] -> 84[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO Channel 06/0 : 77[5] -> 85[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO Channel 09/0 : 79[7] -> 87[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO Channel 06/0 : 73[1] -> 81[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO Channel 09/0 : 78[6] -> 86[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO Channel 05/0 : 74[2] -> 82[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO Channel 05/0 : 75[3] -> 83[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO Channel 10/0 : 76[4] -> 84[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO Channel 07/0 : 77[5] -> 85[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO Channel 10/0 : 79[7] -> 87[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO Channel 07/0 : 73[1] -> 81[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO Channel 10/0 : 78[6] -> 86[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO Channel 06/0 : 74[2] -> 82[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO Channel 11/0 : 76[4] -> 84[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO Channel 06/0 : 75[3] -> 83[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO Channel 08/0 : 77[5] -> 85[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62883:63058 [2] NCCL INFO Channel 04/0 : 42[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62886:63057 [5] NCCL INFO Channel 14/0 : 45[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62882:63056 [1] NCCL INFO Channel 06/0 : 41[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62888:63060 [7] NCCL INFO Channel 06/0 : 47[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62884:63061 [3] NCCL INFO Channel 04/0 : 43[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62883:63058 [2] NCCL INFO Channel 05/0 : 42[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62886:63057 [5] NCCL INFO Channel 15/0 : 45[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62882:63056 [1] NCCL INFO Channel 07/0 : 41[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62888:63060 [7] NCCL INFO Channel 08/0 : 47[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62883:63058 [2] NCCL INFO Channel 06/0 : 42[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62884:63061 [3] NCCL INFO Channel 05/0 : 43[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62882:63056 [1] NCCL INFO Channel 08/0 : 41[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62888:63060 [7] NCCL INFO Channel 09/0 : 47[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62883:63058 [2] NCCL INFO Channel 07/0 : 42[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62884:63061 [3] NCCL INFO Channel 06/0 : 43[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62882:63056 [1] NCCL INFO Channel 10/0 : 41[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62888:63060 [7] NCCL INFO Channel 10/0 : 47[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62883:63058 [2] NCCL INFO Channel 08/0 : 42[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62884:63061 [3] NCCL INFO Channel 07/0 : 43[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62882:63056 [1] NCCL INFO Channel 11/0 : 41[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62888:63060 [7] NCCL INFO Channel 11/0 : 47[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62883:63058 [2] NCCL INFO Channel 09/0 : 42[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62884:63061 [3] NCCL INFO Channel 08/0 : 43[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62882:63056 [1] NCCL INFO Channel 12/0 : 41[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62888:63060 [7] NCCL INFO Channel 12/0 : 47[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62883:63058 [2] NCCL INFO Channel 11/0 : 42[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62884:63061 [3] NCCL INFO Channel 09/0 : 43[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62882:63056 [1] NCCL INFO Channel 13/0 : 41[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62888:63060 [7] NCCL INFO Channel 13/0 : 47[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62883:63058 [2] NCCL INFO Channel 12/0 : 42[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62884:63061 [3] NCCL INFO Channel 10/0 : 43[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62882:63056 [1] NCCL INFO Channel 14/0 : 41[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62888:63060 [7] NCCL INFO Channel 14/0 : 47[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62883:63058 [2] NCCL INFO Channel 13/0 : 42[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62884:63061 [3] NCCL INFO Channel 12/0 : 43[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62882:63056 [1] NCCL INFO Channel 15/0 : 41[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62883:63058 [2] NCCL INFO Channel 14/0 : 42[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62884:63061 [3] NCCL INFO Channel 13/0 : 43[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62883:63058 [2] NCCL INFO Channel 15/0 : 42[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62884:63061 [3] NCCL INFO Channel 14/0 : 43[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62884:63061 [3] NCCL INFO Channel 15/0 : 43[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114516:114691 [4] NCCL INFO Channel 00/0 : 36[4] -> 68[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114512:114686 [0] NCCL INFO Channel 02/0 : 32[0] -> 64[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114516:114691 [4] NCCL INFO Channel 02/0 : 36[4] -> 68[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114519:114692 [7] NCCL INFO Channel 00/0 : 39[7] -> 71[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114512:114686 [0] NCCL INFO Channel 04/0 : 32[0] -> 64[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114516:114691 [4] NCCL INFO Channel 06/0 : 36[4] -> 68[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114515:114690 [3] NCCL INFO Channel 00/0 : 35[3] -> 67[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114519:114692 [7] NCCL INFO Channel 02/0 : 39[7] -> 71[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114512:114686 [0] NCCL INFO Channel 06/0 : 32[0] -> 64[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114516:114691 [4] NCCL INFO Channel 08/0 : 36[4] -> 68[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114515:114690 [3] NCCL INFO Channel 02/0 : 35[3] -> 67[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114512:114686 [0] NCCL INFO Channel 08/0 : 32[0] -> 64[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114519:114692 [7] NCCL INFO Channel 04/0 : 39[7] -> 71[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114516:114691 [4] NCCL INFO Channel 10/0 : 36[4] -> 68[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114512:114686 [0] NCCL INFO Channel 10/0 : 32[0] -> 64[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114515:114690 [3] NCCL INFO Channel 04/0 : 35[3] -> 67[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114516:114691 [4] NCCL INFO Channel 12/0 : 36[4] -> 68[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114519:114692 [7] NCCL INFO Channel 06/0 : 39[7] -> 71[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114512:114686 [0] NCCL INFO Channel 12/0 : 32[0] -> 64[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114516:114691 [4] NCCL INFO Channel 14/0 : 36[4] -> 68[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114515:114690 [3] NCCL INFO Channel 06/0 : 35[3] -> 67[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114519:114692 [7] NCCL INFO Channel 08/0 : 39[7] -> 71[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114512:114686 [0] NCCL INFO Channel 14/0 : 32[0] -> 64[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114516:114691 [4] NCCL INFO Channel 00/0 : 68[4] -> 100[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114515:114690 [3] NCCL INFO Channel 08/0 : 35[3] -> 67[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114512:114686 [0] NCCL INFO Channel 02/0 : 64[0] -> 96[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114519:114692 [7] NCCL INFO Channel 10/0 : 39[7] -> 71[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114516:114691 [4] NCCL INFO Channel 02/0 : 68[4] -> 100[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114512:114686 [0] NCCL INFO Channel 04/0 : 64[0] -> 96[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114515:114690 [3] NCCL INFO Channel 10/0 : 35[3] -> 67[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114514:114689 [2] NCCL INFO Channel 00/0 : 34[2] -> 66[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114519:114692 [7] NCCL INFO Channel 12/0 : 39[7] -> 71[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114516:114691 [4] NCCL INFO Channel 06/0 : 68[4] -> 100[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114518:114693 [6] NCCL INFO Channel 00/0 : 38[6] -> 70[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114512:114686 [0] NCCL INFO Channel 06/0 : 64[0] -> 96[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114515:114690 [3] NCCL INFO Channel 12/0 : 35[3] -> 67[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114516:114691 [4] NCCL INFO Channel 08/0 : 68[4] -> 100[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114514:114689 [2] NCCL INFO Channel 04/0 : 34[2] -> 66[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114512:114686 [0] NCCL INFO Channel 08/0 : 64[0] -> 96[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114519:114692 [7] NCCL INFO Channel 14/0 : 39[7] -> 71[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114518:114693 [6] NCCL INFO Channel 02/0 : 38[6] -> 70[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114515:114690 [3] NCCL INFO Channel 14/0 : 35[3] -> 67[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114516:114691 [4] NCCL INFO Channel 10/0 : 68[4] -> 100[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114512:114686 [0] NCCL INFO Channel 10/0 : 64[0] -> 96[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114514:114689 [2] NCCL INFO Channel 06/0 : 34[2] -> 66[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114518:114693 [6] NCCL INFO Channel 04/0 : 38[6] -> 70[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114519:114692 [7] NCCL INFO Channel 00/0 : 71[7] -> 103[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114515:114690 [3] NCCL INFO Channel 00/0 : 67[3] -> 99[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114516:114691 [4] NCCL INFO Channel 12/0 : 68[4] -> 100[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114512:114686 [0] NCCL INFO Channel 12/0 : 64[0] -> 96[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114514:114689 [2] NCCL INFO Channel 08/0 : 34[2] -> 66[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114518:114693 [6] NCCL INFO Channel 08/0 : 38[6] -> 70[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114515:114690 [3] NCCL INFO Channel 02/0 : 67[3] -> 99[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114519:114692 [7] NCCL INFO Channel 02/0 : 71[7] -> 103[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114512:114686 [0] NCCL INFO Channel 14/0 : 64[0] -> 96[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114516:114691 [4] NCCL INFO Channel 14/0 : 68[4] -> 100[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114517:114687 [5] NCCL INFO Channel 00/0 : 37[5] -> 69[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114514:114689 [2] NCCL INFO Channel 10/0 : 34[2] -> 66[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114518:114693 [6] NCCL INFO Channel 10/0 : 38[6] -> 70[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114513:114688 [1] NCCL INFO Channel 00/0 : 33[1] -> 65[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114515:114690 [3] NCCL INFO Channel 04/0 : 67[3] -> 99[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114519:114692 [7] NCCL INFO Channel 04/0 : 71[7] -> 103[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114517:114687 [5] NCCL INFO Channel 02/0 : 37[5] -> 69[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114518:114693 [6] NCCL INFO Channel 12/0 : 38[6] -> 70[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114513:114688 [1] NCCL INFO Channel 02/0 : 33[1] -> 65[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114514:114689 [2] NCCL INFO Channel 12/0 : 34[2] -> 66[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114515:114690 [3] NCCL INFO Channel 06/0 : 67[3] -> 99[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114519:114692 [7] NCCL INFO Channel 06/0 : 71[7] -> 103[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114517:114687 [5] NCCL INFO Channel 04/0 : 37[5] -> 69[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114518:114693 [6] NCCL INFO Channel 14/0 : 38[6] -> 70[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114513:114688 [1] NCCL INFO Channel 04/0 : 33[1] -> 65[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114515:114690 [3] NCCL INFO Channel 08/0 : 67[3] -> 99[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114514:114689 [2] NCCL INFO Channel 14/0 : 34[2] -> 66[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114519:114692 [7] NCCL INFO Channel 08/0 : 71[7] -> 103[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114517:114687 [5] NCCL INFO Channel 06/0 : 37[5] -> 69[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114518:114693 [6] NCCL INFO Channel 00/0 : 70[6] -> 102[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114513:114688 [1] NCCL INFO Channel 06/0 : 33[1] -> 65[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114515:114690 [3] NCCL INFO Channel 10/0 : 67[3] -> 99[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114514:114689 [2] NCCL INFO Channel 00/0 : 66[2] -> 98[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114519:114692 [7] NCCL INFO Channel 10/0 : 71[7] -> 103[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114517:114687 [5] NCCL INFO Channel 08/0 : 37[5] -> 69[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114518:114693 [6] NCCL INFO Channel 02/0 : 70[6] -> 102[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114513:114688 [1] NCCL INFO Channel 08/0 : 33[1] -> 65[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114515:114690 [3] NCCL INFO Channel 12/0 : 67[3] -> 99[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114514:114689 [2] NCCL INFO Channel 04/0 : 66[2] -> 98[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114519:114692 [7] NCCL INFO Channel 12/0 : 71[7] -> 103[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114517:114687 [5] NCCL INFO Channel 10/0 : 37[5] -> 69[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114513:114688 [1] NCCL INFO Channel 10/0 : 33[1] -> 65[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114518:114693 [6] NCCL INFO Channel 04/0 : 70[6] -> 102[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114515:114690 [3] NCCL INFO Channel 14/0 : 67[3] -> 99[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114514:114689 [2] NCCL INFO Channel 06/0 : 66[2] -> 98[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114519:114692 [7] NCCL INFO Channel 14/0 : 71[7] -> 103[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114517:114687 [5] NCCL INFO Channel 12/0 : 37[5] -> 69[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114513:114688 [1] NCCL INFO Channel 12/0 : 33[1] -> 65[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114514:114689 [2] NCCL INFO Channel 08/0 : 66[2] -> 98[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114518:114693 [6] NCCL INFO Channel 08/0 : 70[6] -> 102[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114517:114687 [5] NCCL INFO Channel 14/0 : 37[5] -> 69[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114513:114688 [1] NCCL INFO Channel 14/0 : 33[1] -> 65[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114514:114689 [2] NCCL INFO Channel 10/0 : 66[2] -> 98[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114518:114693 [6] NCCL INFO Channel 10/0 : 70[6] -> 102[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114517:114687 [5] NCCL INFO Channel 00/0 : 69[5] -> 101[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114513:114688 [1] NCCL INFO Channel 00/0 : 65[1] -> 97[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114514:114689 [2] NCCL INFO Channel 12/0 : 66[2] -> 98[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114518:114693 [6] NCCL INFO Channel 12/0 : 70[6] -> 102[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114517:114687 [5] NCCL INFO Channel 02/0 : 69[5] -> 101[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114513:114688 [1] NCCL INFO Channel 02/0 : 65[1] -> 97[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114514:114689 [2] NCCL INFO Channel 14/0 : 66[2] -> 98[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114518:114693 [6] NCCL INFO Channel 14/0 : 70[6] -> 102[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114517:114687 [5] NCCL INFO Channel 04/0 : 69[5] -> 101[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114513:114688 [1] NCCL INFO Channel 04/0 : 65[1] -> 97[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114517:114687 [5] NCCL INFO Channel 06/0 : 69[5] -> 101[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114513:114688 [1] NCCL INFO Channel 06/0 : 65[1] -> 97[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114517:114687 [5] NCCL INFO Channel 08/0 : 69[5] -> 101[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114513:114688 [1] NCCL INFO Channel 08/0 : 65[1] -> 97[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114517:114687 [5] NCCL INFO Channel 10/0 : 69[5] -> 101[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114513:114688 [1] NCCL INFO Channel 10/0 : 65[1] -> 97[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114517:114687 [5] NCCL INFO Channel 12/0 : 69[5] -> 101[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114513:114688 [1] NCCL INFO Channel 12/0 : 65[1] -> 97[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114517:114687 [5] NCCL INFO Channel 14/0 : 69[5] -> 101[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114513:114688 [1] NCCL INFO Channel 14/0 : 65[1] -> 97[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO Channel 00/0 : 77[5] -> 85[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO Channel 01/0 : 77[5] -> 85[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO Channel 02/0 : 77[5] -> 85[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO Channel 03/0 : 77[5] -> 85[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO Channel 04/0 : 77[5] -> 85[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO Channel 06/0 : 77[5] -> 85[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO Channel 07/0 : 77[5] -> 85[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO Channel 08/0 : 77[5] -> 85[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO Channel 09/0 : 77[5] -> 85[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO Channel 10/0 : 77[5] -> 85[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO Channel 00/0 : 79[7] -> 87[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO Channel 11/0 : 77[5] -> 85[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO Channel 01/0 : 79[7] -> 87[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO Channel 12/0 : 77[5] -> 85[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO Channel 02/0 : 79[7] -> 87[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO Channel 14/0 : 77[5] -> 85[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO Channel 03/0 : 79[7] -> 87[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO Channel 15/0 : 77[5] -> 85[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO Channel 04/0 : 79[7] -> 87[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO Channel 00/0 : 85[5] -> 93[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO Channel 05/0 : 79[7] -> 87[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO Channel 02/0 : 85[5] -> 93[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO Channel 06/0 : 79[7] -> 87[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO Channel 04/0 : 85[5] -> 93[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO Channel 00/0 : 78[6] -> 86[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO Channel 06/0 : 85[5] -> 93[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO Channel 08/0 : 79[7] -> 87[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO Channel 08/0 : 85[5] -> 93[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO Channel 09/0 : 79[7] -> 87[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO Channel 01/0 : 78[6] -> 86[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO Channel 00/0 : 76[4] -> 84[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO Channel 00/0 : 73[1] -> 81[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO Channel 10/0 : 85[5] -> 93[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO Channel 10/0 : 79[7] -> 87[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO Channel 00/0 : 75[3] -> 83[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO Channel 02/0 : 78[6] -> 86[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO Channel 01/0 : 76[4] -> 84[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO Channel 02/0 : 73[1] -> 81[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO Channel 12/0 : 85[5] -> 93[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO Channel 01/0 : 72[0] -> 80[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO Channel 11/0 : 79[7] -> 87[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO Channel 00/0 : 74[2] -> 82[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO Channel 01/0 : 75[3] -> 83[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO Channel 03/0 : 78[6] -> 86[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO Channel 02/0 : 76[4] -> 84[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO Channel 14/0 : 85[5] -> 93[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO Channel 03/0 : 73[1] -> 81[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO Channel 12/0 : 79[7] -> 87[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO Channel 02/0 : 72[0] -> 80[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO Channel 01/0 : 74[2] -> 82[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO Channel 04/0 : 78[6] -> 86[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO Channel 03/0 : 76[4] -> 84[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO Channel 02/0 : 75[3] -> 83[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO Channel 04/0 : 73[1] -> 81[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO Channel 13/0 : 79[7] -> 87[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO Channel 03/0 : 72[0] -> 80[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO Channel 03/0 : 74[2] -> 82[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO Channel 05/0 : 78[6] -> 86[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO Channel 05/0 : 76[4] -> 84[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO Channel 14/0 : 79[7] -> 87[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO Channel 05/0 : 73[1] -> 81[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153567:153742 [5] NCCL INFO Channel 01/0 : 77[5] -> 93[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153567:153742 [5] NCCL INFO Channel 03/0 : 77[5] -> 93[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153567:153742 [5] NCCL INFO Channel 05/0 : 77[5] -> 93[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153567:153742 [5] NCCL INFO Channel 07/0 : 77[5] -> 93[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153567:153742 [5] NCCL INFO Channel 09/0 : 77[5] -> 93[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153567:153742 [5] NCCL INFO Channel 11/0 : 77[5] -> 93[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153567:153742 [5] NCCL INFO Channel 15/0 : 77[5] -> 93[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153567:153742 [5] NCCL INFO Channel 01/0 : 93[5] -> 109[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153567:153742 [5] NCCL INFO Channel 03/0 : 93[5] -> 109[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153567:153742 [5] NCCL INFO Channel 05/0 : 93[5] -> 109[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153567:153742 [5] NCCL INFO Channel 07/0 : 93[5] -> 109[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153567:153742 [5] NCCL INFO Channel 09/0 : 93[5] -> 109[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153567:153742 [5] NCCL INFO Channel 11/0 : 93[5] -> 109[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO Channel 04/0 : 72[0] -> 80[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO Channel 04/0 : 75[3] -> 83[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO Channel 04/0 : 74[2] -> 82[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO Channel 06/0 : 76[4] -> 84[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO Channel 07/0 : 78[6] -> 86[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO Channel 00/0 : 87[7] -> 95[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO Channel 06/0 : 73[1] -> 81[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO Channel 05/0 : 72[0] -> 80[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153567:153742 [5] NCCL INFO Channel 15/0 : 93[5] -> 109[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO Channel 05/0 : 74[2] -> 82[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO Channel 05/0 : 75[3] -> 83[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO Channel 08/0 : 78[6] -> 86[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO Channel 07/0 : 76[4] -> 84[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO Channel 02/0 : 87[7] -> 95[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO Channel 00/0 : 85[5] -> 101[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO Channel 07/0 : 73[1] -> 81[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO Channel 06/0 : 72[0] -> 80[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO Channel 06/0 : 74[2] -> 82[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO Channel 06/0 : 75[3] -> 83[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO Channel 09/0 : 78[6] -> 86[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO Channel 08/0 : 76[4] -> 84[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO Channel 04/0 : 87[7] -> 95[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO Channel 02/0 : 85[5] -> 101[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO Channel 08/0 : 73[1] -> 81[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO Channel 07/0 : 74[2] -> 82[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153569:153739 [7] NCCL INFO Channel 01/0 : 79[7] -> 95[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO Channel 07/0 : 72[0] -> 80[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO Channel 07/0 : 75[3] -> 83[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO Channel 10/0 : 78[6] -> 86[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO Channel 04/0 : 85[5] -> 101[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO Channel 06/0 : 87[7] -> 95[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO Channel 09/0 : 76[4] -> 84[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO Channel 10/0 : 73[1] -> 81[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO Channel 08/0 : 74[2] -> 82[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO Channel 00/0 : 69[5] -> 101[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO Channel 09/0 : 72[0] -> 80[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO Channel 11/0 : 78[6] -> 86[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO Channel 08/0 : 75[3] -> 83[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO Channel 06/0 : 85[5] -> 101[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO Channel 08/0 : 87[7] -> 95[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO Channel 10/0 : 76[4] -> 84[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153569:153739 [7] NCCL INFO Channel 03/0 : 79[7] -> 95[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO Channel 11/0 : 73[1] -> 81[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO Channel 02/0 : 69[5] -> 101[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO Channel 09/0 : 74[2] -> 82[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO Channel 10/0 : 72[0] -> 80[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO Channel 12/0 : 78[6] -> 86[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO Channel 09/0 : 75[3] -> 83[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO Channel 08/0 : 85[5] -> 101[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO Channel 11/0 : 76[4] -> 84[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO Channel 10/0 : 87[7] -> 95[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO Channel 12/0 : 73[1] -> 81[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO Channel 11/0 : 74[2] -> 82[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO Channel 13/0 : 78[6] -> 86[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO Channel 10/0 : 75[3] -> 83[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO Channel 13/0 : 76[4] -> 84[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO Channel 12/0 : 87[7] -> 95[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153569:153739 [7] NCCL INFO Channel 05/0 : 79[7] -> 95[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO Channel 10/0 : 85[5] -> 101[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO Channel 04/0 : 69[5] -> 101[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO Channel 12/0 : 74[2] -> 82[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO Channel 13/0 : 73[1] -> 81[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO Channel 15/0 : 78[6] -> 86[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO Channel 12/0 : 75[3] -> 83[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO Channel 14/0 : 76[4] -> 84[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO Channel 14/0 : 87[7] -> 95[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO Channel 12/0 : 85[5] -> 101[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO Channel 14/0 : 73[1] -> 81[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO Channel 00/0 : 86[6] -> 94[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO Channel 13/0 : 74[2] -> 82[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO Channel 13/0 : 75[3] -> 83[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153569:153739 [7] NCCL INFO Channel 07/0 : 79[7] -> 95[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO Channel 15/0 : 76[4] -> 84[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO Channel 14/0 : 85[5] -> 101[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO Channel 15/0 : 73[1] -> 81[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO Channel 06/0 : 69[5] -> 101[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO Channel 02/0 : 86[6] -> 94[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO Channel 14/0 : 74[2] -> 82[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO Channel 14/0 : 75[3] -> 83[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO Channel 00/0 : 84[4] -> 92[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153567:153742 [5] NCCL INFO Channel 01/0 : 61[5] -> 93[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO Channel 00/0 : 81[1] -> 89[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235653:235825 [5] NCCL INFO Channel 01/0 : 109[5] -> 93[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO Channel 04/0 : 86[6] -> 94[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO Channel 15/0 : 74[2] -> 82[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO Channel 02/0 : 84[4] -> 92[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO Channel 15/0 : 75[3] -> 83[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO Channel 02/0 : 81[1] -> 89[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO Channel 08/0 : 69[5] -> 101[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO Channel 08/0 : 86[6] -> 94[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO Channel 00/0 : 101[5] -> 85[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153569:153739 [7] NCCL INFO Channel 09/0 : 79[7] -> 95[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153567:153742 [5] NCCL INFO Channel 03/0 : 61[5] -> 93[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235653:235825 [5] NCCL INFO Channel 03/0 : 109[5] -> 93[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO Channel 10/0 : 69[5] -> 101[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO Channel 06/0 : 84[4] -> 92[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO Channel 00/0 : 82[2] -> 90[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153569:153739 [7] NCCL INFO Channel 11/0 : 79[7] -> 95[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO Channel 00/0 : 83[3] -> 91[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO Channel 04/0 : 81[1] -> 89[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO Channel 00/0 : 87[7] -> 103[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO Channel 10/0 : 86[6] -> 94[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO Channel 02/0 : 101[5] -> 85[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153567:153742 [5] NCCL INFO Channel 05/0 : 61[5] -> 93[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO Channel 08/0 : 84[4] -> 92[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235653:235825 [5] NCCL INFO Channel 05/0 : 109[5] -> 93[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO Channel 04/0 : 82[2] -> 90[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO Channel 02/0 : 83[3] -> 91[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO Channel 11/0 : 72[0] -> 80[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO Channel 06/0 : 81[1] -> 89[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO Channel 12/0 : 69[5] -> 101[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO Channel 02/0 : 87[7] -> 103[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO Channel 12/0 : 86[6] -> 94[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO Channel 10/0 : 84[4] -> 92[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153569:153739 [7] NCCL INFO Channel 13/0 : 79[7] -> 95[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO Channel 04/0 : 101[5] -> 85[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO Channel 06/0 : 82[2] -> 90[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153567:153742 [5] NCCL INFO Channel 07/0 : 61[5] -> 93[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO Channel 04/0 : 83[3] -> 91[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO Channel 08/0 : 81[1] -> 89[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO Channel 12/0 : 72[0] -> 80[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235653:235825 [5] NCCL INFO Channel 07/0 : 109[5] -> 93[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO Channel 14/0 : 86[6] -> 94[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO Channel 12/0 : 84[4] -> 92[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO Channel 04/0 : 87[7] -> 103[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO Channel 08/0 : 82[2] -> 90[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO Channel 14/0 : 69[5] -> 101[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO Channel 06/0 : 101[5] -> 85[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO Channel 06/0 : 83[3] -> 91[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153569:153739 [7] NCCL INFO Channel 01/0 : 95[7] -> 111[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO Channel 10/0 : 81[1] -> 89[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO Channel 13/0 : 72[0] -> 80[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235653:235825 [5] NCCL INFO Channel 09/0 : 109[5] -> 93[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153567:153742 [5] NCCL INFO Channel 09/0 : 61[5] -> 93[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153569:153739 [7] NCCL INFO Channel 03/0 : 95[7] -> 111[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO Channel 14/0 : 84[4] -> 92[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO Channel 06/0 : 87[7] -> 103[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO Channel 08/0 : 101[5] -> 85[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO Channel 10/0 : 82[2] -> 90[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO Channel 12/0 : 81[1] -> 89[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO Channel 08/0 : 83[3] -> 91[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO Channel 14/0 : 72[0] -> 80[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235653:235825 [5] NCCL INFO Channel 11/0 : 109[5] -> 93[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO Channel 08/0 : 87[7] -> 103[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96568:96753 [5] NCCL INFO Channel 01/0 : 37[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO Channel 12/0 : 82[2] -> 90[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153567:153742 [5] NCCL INFO Channel 11/0 : 61[5] -> 93[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO Channel 14/0 : 81[1] -> 89[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO Channel 10/0 : 101[5] -> 85[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153569:153739 [7] NCCL INFO Channel 05/0 : 95[7] -> 111[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO Channel 15/0 : 72[0] -> 80[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO Channel 10/0 : 83[3] -> 91[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO Channel 10/0 : 87[7] -> 103[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96564:96757 [1] NCCL INFO Channel 01/0 : 33[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96568:96753 [5] NCCL INFO Channel 03/0 : 37[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235653:235825 [5] NCCL INFO Channel 15/0 : 109[5] -> 93[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153568:153736 [6] NCCL INFO Channel 01/0 : 78[6] -> 94[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153567:153742 [5] NCCL INFO Channel 15/0 : 61[5] -> 93[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153569:153739 [7] NCCL INFO Channel 07/0 : 95[7] -> 111[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO Channel 14/0 : 82[2] -> 90[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO Channel 12/0 : 101[5] -> 85[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO Channel 12/0 : 83[3] -> 91[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO Channel 12/0 : 87[7] -> 103[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO Channel 00/0 : 86[6] -> 102[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO Channel 02/0 : 80[0] -> 88[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO Channel 00/0 : 101[5] -> 69[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO Channel 14/0 : 101[5] -> 85[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO Channel 00/0 : 84[4] -> 100[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153568:153736 [6] NCCL INFO Channel 03/0 : 78[6] -> 94[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96570:96759 [7] NCCL INFO Channel 01/0 : 39[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153566:153741 [4] NCCL INFO Channel 01/0 : 76[4] -> 92[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153569:153739 [7] NCCL INFO Channel 09/0 : 95[7] -> 111[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO Channel 02/0 : 101[5] -> 69[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO Channel 14/0 : 87[7] -> 103[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO Channel 02/0 : 86[6] -> 102[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO Channel 04/0 : 80[0] -> 88[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO Channel 14/0 : 83[3] -> 91[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96564:96757 [1] NCCL INFO Channel 03/0 : 33[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96568:96753 [5] NCCL INFO Channel 05/0 : 37[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153568:153736 [6] NCCL INFO Channel 05/0 : 78[6] -> 94[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO Channel 02/0 : 84[4] -> 100[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153566:153741 [4] NCCL INFO Channel 03/0 : 76[4] -> 92[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153569:153739 [7] NCCL INFO Channel 11/0 : 95[7] -> 111[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO Channel 04/0 : 101[5] -> 69[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96570:96759 [7] NCCL INFO Channel 03/0 : 39[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO Channel 04/0 : 86[6] -> 102[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96568:96753 [5] NCCL INFO Channel 07/0 : 37[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96564:96757 [1] NCCL INFO Channel 05/0 : 33[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO Channel 06/0 : 84[4] -> 100[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153568:153736 [6] NCCL INFO Channel 07/0 : 78[6] -> 94[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO Channel 06/0 : 80[0] -> 88[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153567:153742 [5] NCCL INFO Channel 01/0 : 93[5] -> 61[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153569:153739 [7] NCCL INFO Channel 13/0 : 95[7] -> 111[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153566:153741 [4] NCCL INFO Channel 05/0 : 76[4] -> 92[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO Channel 06/0 : 101[5] -> 69[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96570:96759 [7] NCCL INFO Channel 05/0 : 39[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO Channel 00/0 : 81[1] -> 97[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO Channel 11/0 : 79[7] -> 87[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO Channel 08/0 : 86[6] -> 102[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO Channel 11/0 : 78[6] -> 86[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO Channel 08/0 : 73[1] -> 81[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO Channel 00/0 : 82[2] -> 98[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO Channel 07/0 : 74[2] -> 82[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO Channel 08/0 : 84[4] -> 100[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO Channel 13/0 : 76[4] -> 84[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153563:153740 [1] NCCL INFO Channel 01/0 : 73[1] -> 89[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO Channel 08/0 : 101[5] -> 69[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO Channel 09/0 : 77[5] -> 85[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO Channel 12/0 : 79[7] -> 87[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153568:153736 [6] NCCL INFO Channel 09/0 : 78[6] -> 94[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO Channel 07/0 : 75[3] -> 83[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96568:96753 [5] NCCL INFO Channel 09/0 : 37[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153567:153742 [5] NCCL INFO Channel 03/0 : 93[5] -> 61[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO Channel 12/0 : 78[6] -> 86[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO Channel 10/0 : 73[1] -> 81[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96564:96757 [1] NCCL INFO Channel 07/0 : 33[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO Channel 08/0 : 74[2] -> 82[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96570:96759 [7] NCCL INFO Channel 07/0 : 39[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO Channel 08/0 : 80[0] -> 88[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO Channel 14/0 : 76[4] -> 84[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO Channel 00/0 : 71[7] -> 103[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO Channel 00/0 : 103[7] -> 87[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO Channel 02/0 : 81[1] -> 97[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO Channel 10/0 : 77[5] -> 85[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO Channel 10/0 : 86[6] -> 102[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO Channel 13/0 : 79[7] -> 87[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153566:153741 [4] NCCL INFO Channel 07/0 : 76[4] -> 92[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153564:153737 [2] NCCL INFO Channel 01/0 : 74[2] -> 90[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO Channel 10/0 : 84[4] -> 100[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO Channel 08/0 : 75[3] -> 83[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153563:153740 [1] NCCL INFO Channel 03/0 : 73[1] -> 89[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO Channel 13/0 : 78[6] -> 86[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO Channel 11/0 : 73[1] -> 81[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96568:96753 [5] NCCL INFO Channel 11/0 : 37[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96564:96757 [1] NCCL INFO Channel 11/0 : 33[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153568:153736 [6] NCCL INFO Channel 11/0 : 78[6] -> 94[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153567:153742 [5] NCCL INFO Channel 05/0 : 93[5] -> 61[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO Channel 09/0 : 74[2] -> 82[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO Channel 15/0 : 76[4] -> 84[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96570:96759 [7] NCCL INFO Channel 09/0 : 39[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO Channel 10/0 : 101[5] -> 69[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO Channel 11/0 : 77[5] -> 85[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO Channel 10/0 : 80[0] -> 88[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO Channel 04/0 : 82[2] -> 98[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO Channel 14/0 : 79[7] -> 87[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO Channel 02/0 : 103[7] -> 87[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO Channel 09/0 : 75[3] -> 83[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO Channel 12/0 : 86[6] -> 102[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO Channel 15/0 : 78[6] -> 86[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153566:153741 [4] NCCL INFO Channel 09/0 : 76[4] -> 92[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO Channel 04/0 : 81[1] -> 97[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO Channel 12/0 : 84[4] -> 100[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO Channel 12/0 : 73[1] -> 81[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO Channel 11/0 : 74[2] -> 82[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO Channel 12/0 : 77[5] -> 85[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153564:153737 [2] NCCL INFO Channel 03/0 : 74[2] -> 90[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO Channel 02/0 : 71[7] -> 103[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO Channel 10/0 : 75[3] -> 83[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96568:96753 [5] NCCL INFO Channel 15/0 : 37[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96564:96757 [1] NCCL INFO Channel 13/0 : 33[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96570:96759 [7] NCCL INFO Channel 11/0 : 39[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153563:153740 [1] NCCL INFO Channel 05/0 : 73[1] -> 89[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO Channel 12/0 : 101[5] -> 69[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO Channel 13/0 : 73[1] -> 81[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153567:153742 [5] NCCL INFO Channel 07/0 : 93[5] -> 61[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO Channel 14/0 : 77[5] -> 85[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO Channel 12/0 : 74[2] -> 82[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153568:153736 [6] NCCL INFO Channel 13/0 : 78[6] -> 94[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO Channel 12/0 : 80[0] -> 88[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO Channel 12/0 : 75[3] -> 83[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO Channel 00/0 : 83[3] -> 99[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO Channel 14/0 : 73[1] -> 81[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO Channel 06/0 : 82[2] -> 98[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO Channel 04/0 : 103[7] -> 87[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO Channel 15/0 : 77[5] -> 85[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO Channel 13/0 : 75[3] -> 83[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO Channel 14/0 : 101[5] -> 69[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO Channel 13/0 : 74[2] -> 82[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO Channel 04/0 : 71[7] -> 103[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO Channel 15/0 : 73[1] -> 81[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153564:153737 [2] NCCL INFO Channel 05/0 : 74[2] -> 90[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO Channel 14/0 : 75[3] -> 83[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96570:96759 [7] NCCL INFO Channel 13/0 : 39[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96564:96757 [1] NCCL INFO Channel 15/0 : 33[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153566:153741 [4] NCCL INFO Channel 11/0 : 76[4] -> 92[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153565:153738 [3] NCCL INFO Channel 01/0 : 75[3] -> 91[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO Channel 14/0 : 86[6] -> 102[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO Channel 14/0 : 74[2] -> 82[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153567:153742 [5] NCCL INFO Channel 09/0 : 93[5] -> 61[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO Channel 14/0 : 84[4] -> 100[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO Channel 15/0 : 75[3] -> 83[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153568:153736 [6] NCCL INFO Channel 15/0 : 78[6] -> 94[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO Channel 06/0 : 81[1] -> 97[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO Channel 15/0 : 74[2] -> 82[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153563:153740 [1] NCCL INFO Channel 07/0 : 73[1] -> 89[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO Channel 14/0 : 80[0] -> 88[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO Channel 01/0 : 77[5] -> 93[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO Channel 06/0 : 71[7] -> 103[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO Channel 02/0 : 83[3] -> 99[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO Channel 03/0 : 77[5] -> 93[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235655:235828 [7] NCCL INFO Channel 01/0 : 111[7] -> 95[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO Channel 05/0 : 77[5] -> 93[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO Channel 07/0 : 77[5] -> 93[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO Channel 09/0 : 77[5] -> 93[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153569:153739 [7] NCCL INFO Channel 01/0 : 63[7] -> 95[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO Channel 11/0 : 77[5] -> 93[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153566:153741 [4] NCCL INFO Channel 13/0 : 76[4] -> 92[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO Channel 08/0 : 82[2] -> 98[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO Channel 06/0 : 103[7] -> 87[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO Channel 15/0 : 77[5] -> 93[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153564:153737 [2] NCCL INFO Channel 07/0 : 74[2] -> 90[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153565:153738 [3] NCCL INFO Channel 03/0 : 75[3] -> 91[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO Channel 08/0 : 81[1] -> 97[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO Channel 01/0 : 79[7] -> 95[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO Channel 08/0 : 71[7] -> 103[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO Channel 04/0 : 83[3] -> 99[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO Channel 03/0 : 79[7] -> 95[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO Channel 05/0 : 79[7] -> 95[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235655:235828 [7] NCCL INFO Channel 03/0 : 111[7] -> 95[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO Channel 07/0 : 79[7] -> 95[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO Channel 09/0 : 79[7] -> 95[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153567:153742 [5] NCCL INFO Channel 11/0 : 93[5] -> 61[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153568:153736 [6] NCCL INFO Channel 01/0 : 94[6] -> 110[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO Channel 01/0 : 93[5] -> 77[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96568:96753 [5] NCCL INFO Channel 00/0 : 21[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153563:153740 [1] NCCL INFO Channel 11/0 : 73[1] -> 89[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO Channel 08/0 : 103[7] -> 87[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO Channel 10/0 : 82[2] -> 98[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO Channel 10/0 : 81[1] -> 97[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO Channel 11/0 : 79[7] -> 95[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153569:153739 [7] NCCL INFO Channel 03/0 : 63[7] -> 95[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO Channel 03/0 : 93[5] -> 77[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO Channel 10/0 : 71[7] -> 103[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO Channel 13/0 : 79[7] -> 95[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235655:235828 [7] NCCL INFO Channel 05/0 : 111[7] -> 95[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO Channel 05/0 : 93[5] -> 77[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO Channel 07/0 : 93[5] -> 77[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO Channel 06/0 : 83[3] -> 99[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO Channel 09/0 : 93[5] -> 77[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153566:153741 [4] NCCL INFO Channel 15/0 : 76[4] -> 92[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO Channel 11/0 : 93[5] -> 77[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96564:96757 [1] NCCL INFO Channel 00/0 : 17[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153564:153737 [2] NCCL INFO Channel 09/0 : 74[2] -> 90[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO Channel 15/0 : 93[5] -> 77[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153565:153738 [3] NCCL INFO Channel 05/0 : 75[3] -> 91[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO Channel 01/0 : 78[6] -> 94[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153568:153736 [6] NCCL INFO Channel 03/0 : 94[6] -> 110[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO Channel 00/0 : 70[6] -> 102[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO Channel 00/0 : 68[4] -> 100[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO Channel 01/0 : 76[4] -> 92[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153567:153742 [5] NCCL INFO Channel 15/0 : 93[5] -> 61[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235655:235828 [7] NCCL INFO Channel 07/0 : 111[7] -> 95[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO Channel 00/0 : 100[4] -> 84[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO Channel 00/0 : 102[6] -> 86[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO Channel 10/0 : 103[7] -> 87[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO Channel 12/0 : 81[1] -> 97[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO Channel 03/0 : 78[6] -> 94[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153569:153739 [7] NCCL INFO Channel 05/0 : 63[7] -> 95[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153563:153740 [1] NCCL INFO Channel 13/0 : 73[1] -> 89[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO Channel 12/0 : 82[2] -> 98[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO Channel 08/0 : 83[3] -> 99[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO Channel 03/0 : 76[4] -> 92[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96568:96753 [5] NCCL INFO Channel 02/0 : 21[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO Channel 12/0 : 71[7] -> 103[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO Channel 05/0 : 78[6] -> 94[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96570:96759 [7] NCCL INFO Channel 00/0 : 23[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235655:235828 [7] NCCL INFO Channel 09/0 : 111[7] -> 95[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO Channel 05/0 : 76[4] -> 92[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153566:153741 [4] NCCL INFO Channel 01/0 : 92[4] -> 108[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO Channel 07/0 : 78[6] -> 94[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153564:153737 [2] NCCL INFO Channel 11/0 : 74[2] -> 90[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO Channel 02/0 : 100[4] -> 84[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO Channel 01/0 : 73[1] -> 89[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153568:153736 [6] NCCL INFO Channel 05/0 : 94[6] -> 110[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO Channel 02/0 : 102[6] -> 86[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO Channel 07/0 : 76[4] -> 92[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153565:153738 [3] NCCL INFO Channel 07/0 : 75[3] -> 91[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO Channel 02/0 : 70[6] -> 102[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO Channel 12/0 : 103[7] -> 87[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO Channel 01/0 : 74[2] -> 90[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153563:153740 [1] NCCL INFO Channel 15/0 : 73[1] -> 89[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153569:153739 [7] NCCL INFO Channel 07/0 : 63[7] -> 95[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO Channel 02/0 : 68[4] -> 100[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO Channel 14/0 : 71[7] -> 103[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO Channel 14/0 : 81[1] -> 97[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO Channel 09/0 : 78[6] -> 94[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96564:96757 [1] NCCL INFO Channel 02/0 : 17[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153566:153741 [4] NCCL INFO Channel 03/0 : 92[4] -> 108[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235655:235828 [7] NCCL INFO Channel 11/0 : 111[7] -> 95[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO Channel 03/0 : 73[1] -> 89[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO Channel 09/0 : 76[4] -> 92[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96568:96753 [5] NCCL INFO Channel 04/0 : 21[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO Channel 03/0 : 74[2] -> 90[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96570:96759 [7] NCCL INFO Channel 02/0 : 23[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO Channel 11/0 : 78[6] -> 94[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96564:96757 [1] NCCL INFO Channel 04/0 : 17[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO Channel 11/0 : 76[4] -> 92[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO Channel 05/0 : 73[1] -> 89[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO Channel 04/0 : 70[6] -> 102[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO Channel 05/0 : 74[2] -> 90[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO Channel 14/0 : 82[2] -> 98[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153568:153736 [6] NCCL INFO Channel 07/0 : 94[6] -> 110[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235655:235828 [7] NCCL INFO Channel 13/0 : 111[7] -> 95[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO Channel 02/0 : 80[0] -> 96[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO Channel 06/0 : 100[4] -> 84[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO Channel 10/0 : 83[3] -> 99[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153564:153737 [2] NCCL INFO Channel 13/0 : 74[2] -> 90[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO Channel 04/0 : 102[6] -> 86[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96568:96753 [5] NCCL INFO Channel 06/0 : 21[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96570:96759 [7] NCCL INFO Channel 04/0 : 23[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153562:153735 [0] NCCL INFO Channel 01/0 : 72[0] -> 88[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO Channel 14/0 : 103[7] -> 87[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96564:96757 [1] NCCL INFO Channel 06/0 : 17[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153565:153738 [3] NCCL INFO Channel 09/0 : 75[3] -> 91[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO Channel 06/0 : 68[4] -> 100[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96568:96753 [5] NCCL INFO Channel 08/0 : 21[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96570:96759 [7] NCCL INFO Channel 06/0 : 23[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153563:153740 [1] NCCL INFO Channel 01/0 : 89[1] -> 105[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO Channel 13/0 : 78[6] -> 94[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153569:153739 [7] NCCL INFO Channel 09/0 : 63[7] -> 95[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153566:153741 [4] NCCL INFO Channel 05/0 : 92[4] -> 108[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO Channel 04/0 : 80[0] -> 96[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO Channel 01/0 : 75[3] -> 91[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO Channel 12/0 : 83[3] -> 99[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO Channel 08/0 : 100[4] -> 84[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO Channel 13/0 : 76[4] -> 92[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO Channel 08/0 : 102[6] -> 86[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO Channel 07/0 : 74[2] -> 90[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO Channel 07/0 : 73[1] -> 89[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO Channel 06/0 : 80[0] -> 96[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO Channel 15/0 : 78[6] -> 94[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO Channel 08/0 : 70[6] -> 102[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO Channel 03/0 : 75[3] -> 91[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96564:96757 [1] NCCL INFO Channel 08/0 : 17[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO Channel 08/0 : 68[4] -> 100[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO Channel 15/0 : 76[4] -> 92[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96568:96753 [5] NCCL INFO Channel 10/0 : 21[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO Channel 01/0 : 95[7] -> 79[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96570:96759 [7] NCCL INFO Channel 08/0 : 23[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO Channel 11/0 : 73[1] -> 89[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96564:96757 [1] NCCL INFO Channel 10/0 : 17[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO Channel 10/0 : 100[4] -> 84[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO Channel 00/0 : 97[1] -> 81[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO Channel 09/0 : 74[2] -> 90[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO Channel 05/0 : 75[3] -> 91[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO Channel 03/0 : 95[7] -> 79[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO Channel 00/0 : 103[7] -> 71[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO Channel 13/0 : 73[1] -> 89[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO Channel 10/0 : 70[6] -> 102[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO Channel 07/0 : 75[3] -> 91[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO Channel 10/0 : 68[4] -> 100[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO Channel 11/0 : 74[2] -> 90[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO Channel 00/0 : 66[2] -> 98[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO Channel 00/0 : 65[1] -> 97[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO Channel 05/0 : 95[7] -> 79[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO Channel 14/0 : 83[3] -> 99[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO Channel 15/0 : 73[1] -> 89[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96568:96753 [5] NCCL INFO Channel 12/0 : 21[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96570:96759 [7] NCCL INFO Channel 10/0 : 23[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO Channel 10/0 : 102[6] -> 86[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO Channel 00/0 : 98[2] -> 82[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO Channel 09/0 : 75[3] -> 91[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96564:96757 [1] NCCL INFO Channel 12/0 : 17[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO Channel 08/0 : 80[0] -> 96[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO Channel 13/0 : 74[2] -> 90[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO Channel 07/0 : 95[7] -> 79[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO Channel 13/0 : 75[3] -> 91[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO Channel 02/0 : 103[7] -> 71[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO Channel 15/0 : 74[2] -> 90[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO Channel 12/0 : 70[6] -> 102[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO Channel 09/0 : 95[7] -> 79[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO Channel 12/0 : 100[4] -> 84[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO Channel 15/0 : 75[3] -> 91[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96566:96755 [3] NCCL INFO Channel 01/0 : 35[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO Channel 02/0 : 97[1] -> 81[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO Channel 12/0 : 102[6] -> 86[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO Channel 01/0 : 72[0] -> 88[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96567:96760 [4] NCCL INFO Channel 01/0 : 36[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO Channel 04/0 : 98[2] -> 82[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO Channel 10/0 : 80[0] -> 96[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO Channel 11/0 : 95[7] -> 79[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96563:96756 [0] NCCL INFO Channel 01/0 : 32[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO Channel 03/0 : 72[0] -> 88[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO Channel 13/0 : 95[7] -> 79[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO Channel 04/0 : 103[7] -> 71[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO Channel 02/0 : 65[1] -> 97[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO Channel 05/0 : 72[0] -> 88[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO Channel 12/0 : 68[4] -> 100[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO Channel 07/0 : 72[0] -> 88[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO Channel 14/0 : 100[4] -> 84[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96570:96759 [7] NCCL INFO Channel 12/0 : 23[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96568:96753 [5] NCCL INFO Channel 14/0 : 21[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96564:96757 [1] NCCL INFO Channel 14/0 : 17[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96566:96755 [3] NCCL INFO Channel 03/0 : 35[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96569:96758 [6] NCCL INFO Channel 01/0 : 38[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO Channel 09/0 : 72[0] -> 88[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO Channel 11/0 : 72[0] -> 88[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO Channel 04/0 : 66[2] -> 98[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO Channel 14/0 : 70[6] -> 102[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO Channel 14/0 : 102[6] -> 86[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO Channel 04/0 : 97[1] -> 81[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO Channel 13/0 : 72[0] -> 88[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO Channel 06/0 : 103[7] -> 71[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO Channel 12/0 : 80[0] -> 96[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO Channel 15/0 : 72[0] -> 88[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO Channel 06/0 : 98[2] -> 82[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO Channel 00/0 : 99[3] -> 83[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96567:96760 [4] NCCL INFO Channel 03/0 : 36[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96563:96756 [0] NCCL INFO Channel 03/0 : 32[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96570:96759 [7] NCCL INFO Channel 14/0 : 23[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO Channel 04/0 : 65[1] -> 97[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO Channel 14/0 : 68[4] -> 100[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO Channel 06/0 : 66[2] -> 98[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO Channel 08/0 : 103[7] -> 71[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO Channel 14/0 : 80[0] -> 96[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96568:96753 [5] NCCL INFO Channel 00/0 : 37[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96566:96755 [3] NCCL INFO Channel 05/0 : 35[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96569:96758 [6] NCCL INFO Channel 03/0 : 38[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96564:96757 [1] NCCL INFO Channel 00/0 : 33[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96565:96754 [2] NCCL INFO Channel 01/0 : 34[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96567:96760 [4] NCCL INFO Channel 05/0 : 36[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO Channel 01/0 : 94[6] -> 78[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO Channel 06/0 : 65[1] -> 97[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO Channel 00/0 : 67[3] -> 99[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO Channel 06/0 : 97[1] -> 81[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO Channel 08/0 : 98[2] -> 82[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO Channel 08/0 : 66[2] -> 98[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO Channel 02/0 : 99[3] -> 83[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO Channel 10/0 : 103[7] -> 71[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96563:96756 [0] NCCL INFO Channel 05/0 : 32[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96570:96759 [7] NCCL INFO Channel 00/0 : 39[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96568:96753 [5] NCCL INFO Channel 02/0 : 37[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO Channel 08/0 : 97[1] -> 81[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO Channel 03/0 : 94[6] -> 78[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96566:96755 [3] NCCL INFO Channel 07/0 : 35[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO Channel 00/0 : 102[6] -> 70[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO Channel 02/0 : 67[3] -> 99[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO Channel 10/0 : 98[2] -> 82[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO Channel 04/0 : 99[3] -> 83[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO Channel 08/0 : 65[1] -> 97[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO Channel 10/0 : 66[2] -> 98[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO Channel 12/0 : 103[7] -> 71[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO Channel 01/0 : 92[4] -> 76[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96564:96757 [1] NCCL INFO Channel 02/0 : 33[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96569:96758 [6] NCCL INFO Channel 05/0 : 38[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO Channel 10/0 : 97[1] -> 81[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96565:96754 [2] NCCL INFO Channel 03/0 : 34[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO Channel 12/0 : 98[2] -> 82[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96567:96760 [4] NCCL INFO Channel 07/0 : 36[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO Channel 00/0 : 100[4] -> 68[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO Channel 06/0 : 99[3] -> 83[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96570:96759 [7] NCCL INFO Channel 02/0 : 39[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96563:96756 [0] NCCL INFO Channel 07/0 : 32[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO Channel 02/0 : 102[6] -> 70[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96568:96753 [5] NCCL INFO Channel 04/0 : 37[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96566:96755 [3] NCCL INFO Channel 09/0 : 35[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO Channel 05/0 : 94[6] -> 78[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO Channel 02/0 : 96[0] -> 80[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO Channel 12/0 : 97[1] -> 81[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO Channel 04/0 : 67[3] -> 99[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO Channel 14/0 : 103[7] -> 71[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO Channel 10/0 : 65[1] -> 97[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO Channel 12/0 : 66[2] -> 98[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO Channel 04/0 : 102[6] -> 70[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO Channel 02/0 : 100[4] -> 68[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96564:96757 [1] NCCL INFO Channel 04/0 : 33[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO Channel 03/0 : 92[4] -> 76[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96569:96758 [6] NCCL INFO Channel 07/0 : 38[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO Channel 14/0 : 98[2] -> 82[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO Channel 07/0 : 94[6] -> 78[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96565:96754 [2] NCCL INFO Channel 05/0 : 34[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO Channel 08/0 : 99[3] -> 83[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO Channel 01/0 : 89[1] -> 73[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96567:96760 [4] NCCL INFO Channel 09/0 : 36[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO Channel 04/0 : 96[0] -> 80[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO Channel 06/0 : 67[3] -> 99[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO Channel 05/0 : 92[4] -> 76[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96570:96759 [7] NCCL INFO Channel 04/0 : 39[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO Channel 14/0 : 97[1] -> 81[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96563:96756 [0] NCCL INFO Channel 09/0 : 32[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96568:96753 [5] NCCL INFO Channel 06/0 : 37[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO Channel 10/0 : 99[3] -> 83[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96566:96755 [3] NCCL INFO Channel 13/0 : 35[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO Channel 12/0 : 65[1] -> 97[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO Channel 14/0 : 66[2] -> 98[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO Channel 02/0 : 64[0] -> 96[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO Channel 06/0 : 96[0] -> 80[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96564:96757 [1] NCCL INFO Channel 06/0 : 33[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96569:96758 [6] NCCL INFO Channel 09/0 : 38[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO Channel 08/0 : 102[6] -> 70[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO Channel 06/0 : 100[4] -> 68[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96565:96754 [2] NCCL INFO Channel 07/0 : 34[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO Channel 08/0 : 67[3] -> 99[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96567:96760 [4] NCCL INFO Channel 11/0 : 36[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO Channel 09/0 : 94[6] -> 78[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO Channel 03/0 : 89[1] -> 73[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96570:96759 [7] NCCL INFO Channel 06/0 : 39[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO Channel 01/0 : 90[2] -> 74[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO Channel 12/0 : 99[3] -> 83[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO Channel 08/0 : 96[0] -> 80[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO Channel 14/0 : 65[1] -> 97[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235652:235822 [4] NCCL INFO Channel 01/0 : 108[4] -> 92[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96568:96753 [5] NCCL INFO Channel 08/0 : 37[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO Channel 07/0 : 92[4] -> 76[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96563:96756 [0] NCCL INFO Channel 11/0 : 32[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO Channel 01/0 : 91[3] -> 75[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96566:96755 [3] NCCL INFO Channel 15/0 : 35[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO Channel 11/0 : 94[6] -> 78[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO Channel 05/0 : 89[1] -> 73[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96569:96758 [6] NCCL INFO Channel 11/0 : 38[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO Channel 14/0 : 99[3] -> 83[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96564:96757 [1] NCCL INFO Channel 08/0 : 33[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96565:96754 [2] NCCL INFO Channel 09/0 : 34[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO Channel 04/0 : 64[0] -> 96[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO Channel 10/0 : 102[6] -> 70[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235654:235821 [6] NCCL INFO Channel 01/0 : 110[6] -> 94[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO Channel 10/0 : 96[0] -> 80[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96567:96760 [4] NCCL INFO Channel 13/0 : 36[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO Channel 08/0 : 100[4] -> 68[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO Channel 10/0 : 67[3] -> 99[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO Channel 06/0 : 64[0] -> 96[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO Channel 03/0 : 90[2] -> 74[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO Channel 09/0 : 92[4] -> 76[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO Channel 03/0 : 91[3] -> 75[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235652:235822 [4] NCCL INFO Channel 03/0 : 108[4] -> 92[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO Channel 12/0 : 96[0] -> 80[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO Channel 13/0 : 94[6] -> 78[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235650:235824 [2] NCCL INFO Channel 01/0 : 106[2] -> 90[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96570:96759 [7] NCCL INFO Channel 08/0 : 39[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO Channel 12/0 : 102[6] -> 70[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235651:235823 [3] NCCL INFO Channel 01/0 : 107[3] -> 91[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96568:96753 [5] NCCL INFO Channel 10/0 : 37[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96563:96756 [0] NCCL INFO Channel 13/0 : 32[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO Channel 10/0 : 100[4] -> 68[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO Channel 00/0 : 98[2] -> 66[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96569:96758 [6] NCCL INFO Channel 13/0 : 38[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO Channel 12/0 : 67[3] -> 99[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96565:96754 [2] NCCL INFO Channel 11/0 : 34[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO Channel 08/0 : 64[0] -> 96[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO Channel 07/0 : 89[1] -> 73[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235654:235821 [6] NCCL INFO Channel 03/0 : 110[6] -> 94[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO Channel 14/0 : 96[0] -> 80[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO Channel 11/0 : 92[4] -> 76[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO Channel 05/0 : 90[2] -> 74[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235652:235822 [4] NCCL INFO Channel 05/0 : 108[4] -> 92[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO Channel 05/0 : 91[3] -> 75[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235650:235824 [2] NCCL INFO Channel 03/0 : 106[2] -> 90[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235651:235823 [3] NCCL INFO Channel 03/0 : 107[3] -> 91[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO Channel 15/0 : 94[6] -> 78[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235654:235821 [6] NCCL INFO Channel 05/0 : 110[6] -> 94[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96567:96760 [4] NCCL INFO Channel 15/0 : 36[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO Channel 14/0 : 102[6] -> 70[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96570:96759 [7] NCCL INFO Channel 10/0 : 39[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO Channel 00/0 : 97[1] -> 65[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96564:96757 [1] NCCL INFO Channel 10/0 : 33[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96568:96753 [5] NCCL INFO Channel 12/0 : 37[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153568:153736 [6] NCCL INFO Channel 09/0 : 94[6] -> 110[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96569:96758 [6] NCCL INFO Channel 15/0 : 38[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96563:96756 [0] NCCL INFO Channel 15/0 : 32[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153564:153737 [2] NCCL INFO Channel 15/0 : 74[2] -> 90[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235649:235827 [1] NCCL INFO Channel 01/0 : 105[1] -> 89[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO Channel 11/0 : 89[1] -> 73[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96565:96754 [2] NCCL INFO Channel 13/0 : 34[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153565:153738 [3] NCCL INFO Channel 13/0 : 75[3] -> 91[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153562:153735 [0] NCCL INFO Channel 03/0 : 72[0] -> 88[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235652:235822 [4] NCCL INFO Channel 07/0 : 108[4] -> 92[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO Channel 13/0 : 92[4] -> 76[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96570:96759 [7] NCCL INFO Channel 12/0 : 39[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153563:153740 [1] NCCL INFO Channel 03/0 : 89[1] -> 105[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235651:235823 [3] NCCL INFO Channel 05/0 : 107[3] -> 91[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO Channel 07/0 : 90[2] -> 74[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153566:153741 [4] NCCL INFO Channel 07/0 : 92[4] -> 108[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153569:153739 [7] NCCL INFO Channel 11/0 : 63[7] -> 95[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO Channel 04/0 : 98[2] -> 66[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO Channel 12/0 : 100[4] -> 68[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO Channel 07/0 : 91[3] -> 75[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153568:153736 [6] NCCL INFO Channel 11/0 : 94[6] -> 110[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO Channel 14/0 : 67[3] -> 99[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO Channel 10/0 : 64[0] -> 96[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO Channel 13/0 : 89[1] -> 73[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153564:153737 [2] NCCL INFO Channel 01/0 : 90[2] -> 106[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO Channel 02/0 : 97[1] -> 65[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96566:96755 [3] NCCL INFO Channel 00/0 : 19[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153565:153738 [3] NCCL INFO Channel 15/0 : 75[3] -> 91[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO Channel 06/0 : 98[2] -> 66[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235650:235824 [2] NCCL INFO Channel 05/0 : 106[2] -> 90[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235654:235821 [6] NCCL INFO Channel 07/0 : 110[6] -> 94[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96564:96757 [1] NCCL INFO Channel 12/0 : 33[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153563:153740 [1] NCCL INFO Channel 05/0 : 89[1] -> 105[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235652:235822 [4] NCCL INFO Channel 09/0 : 108[4] -> 92[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235649:235827 [1] NCCL INFO Channel 03/0 : 105[1] -> 89[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96568:96753 [5] NCCL INFO Channel 14/0 : 37[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153562:153735 [0] NCCL INFO Channel 05/0 : 72[0] -> 88[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153566:153741 [4] NCCL INFO Channel 09/0 : 92[4] -> 108[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235651:235823 [3] NCCL INFO Channel 07/0 : 107[3] -> 91[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96565:96754 [2] NCCL INFO Channel 15/0 : 34[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153569:153739 [7] NCCL INFO Channel 13/0 : 63[7] -> 95[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO Channel 15/0 : 92[4] -> 76[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96570:96759 [7] NCCL INFO Channel 14/0 : 39[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153568:153736 [6] NCCL INFO Channel 13/0 : 94[6] -> 110[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO Channel 09/0 : 90[2] -> 74[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO Channel 09/0 : 91[3] -> 75[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96566:96755 [3] NCCL INFO Channel 02/0 : 19[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96564:96757 [1] NCCL INFO Channel 14/0 : 33[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153564:153737 [2] NCCL INFO Channel 03/0 : 90[2] -> 106[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO Channel 14/0 : 100[4] -> 68[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO Channel 04/0 : 97[1] -> 65[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO Channel 12/0 : 64[0] -> 96[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO Channel 15/0 : 89[1] -> 73[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153565:153738 [3] NCCL INFO Channel 01/0 : 91[3] -> 107[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO Channel 08/0 : 98[2] -> 66[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153563:153740 [1] NCCL INFO Channel 07/0 : 89[1] -> 105[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235654:235821 [6] NCCL INFO Channel 09/0 : 110[6] -> 94[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153566:153741 [4] NCCL INFO Channel 11/0 : 92[4] -> 108[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235650:235824 [2] NCCL INFO Channel 07/0 : 106[2] -> 90[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153562:153735 [0] NCCL INFO Channel 07/0 : 72[0] -> 88[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235652:235822 [4] NCCL INFO Channel 11/0 : 108[4] -> 92[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96566:96755 [3] NCCL INFO Channel 04/0 : 19[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153568:153736 [6] NCCL INFO Channel 15/0 : 94[6] -> 110[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235649:235827 [1] NCCL INFO Channel 05/0 : 105[1] -> 89[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO Channel 11/0 : 90[2] -> 74[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO Channel 13/0 : 91[3] -> 75[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96567:96760 [4] NCCL INFO Channel 00/0 : 20[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153564:153737 [2] NCCL INFO Channel 05/0 : 90[2] -> 106[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO Channel 14/0 : 64[0] -> 96[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO Channel 06/0 : 97[1] -> 65[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235651:235823 [3] NCCL INFO Channel 09/0 : 107[3] -> 91[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96569:96758 [6] NCCL INFO Channel 00/0 : 22[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153565:153738 [3] NCCL INFO Channel 03/0 : 91[3] -> 107[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153563:153740 [1] NCCL INFO Channel 11/0 : 89[1] -> 105[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO Channel 10/0 : 98[2] -> 66[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153566:153741 [4] NCCL INFO Channel 13/0 : 92[4] -> 108[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO Channel 00/0 : 99[3] -> 67[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153562:153735 [0] NCCL INFO Channel 09/0 : 72[0] -> 88[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153564:153737 [2] NCCL INFO Channel 07/0 : 90[2] -> 106[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO Channel 13/0 : 90[2] -> 74[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153565:153738 [3] NCCL INFO Channel 05/0 : 91[3] -> 107[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153563:153740 [1] NCCL INFO Channel 13/0 : 89[1] -> 105[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235654:235821 [6] NCCL INFO Channel 11/0 : 110[6] -> 94[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO Channel 15/0 : 91[3] -> 75[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153566:153741 [4] NCCL INFO Channel 15/0 : 92[4] -> 108[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153569:153739 [7] NCCL INFO Channel 01/0 : 95[7] -> 63[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235650:235824 [2] NCCL INFO Channel 09/0 : 106[2] -> 90[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96566:96755 [3] NCCL INFO Channel 06/0 : 19[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153564:153737 [2] NCCL INFO Channel 09/0 : 90[2] -> 106[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235652:235822 [4] NCCL INFO Channel 13/0 : 108[4] -> 92[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96563:96756 [0] NCCL INFO Channel 02/0 : 16[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96567:96760 [4] NCCL INFO Channel 02/0 : 20[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153562:153735 [0] NCCL INFO Channel 11/0 : 72[0] -> 88[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO Channel 08/0 : 97[1] -> 65[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235649:235827 [1] NCCL INFO Channel 07/0 : 105[1] -> 89[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153565:153738 [3] NCCL INFO Channel 07/0 : 91[3] -> 107[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO Channel 12/0 : 98[2] -> 66[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235651:235823 [3] NCCL INFO Channel 13/0 : 107[3] -> 91[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153563:153740 [1] NCCL INFO Channel 15/0 : 89[1] -> 105[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153569:153739 [7] NCCL INFO Channel 03/0 : 95[7] -> 63[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO Channel 02/0 : 99[3] -> 67[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235654:235821 [6] NCCL INFO Channel 13/0 : 110[6] -> 94[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO Channel 15/0 : 90[2] -> 74[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153564:153737 [2] NCCL INFO Channel 11/0 : 90[2] -> 106[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153562:153735 [0] NCCL INFO Channel 13/0 : 72[0] -> 88[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96569:96758 [6] NCCL INFO Channel 02/0 : 22[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96565:96754 [2] NCCL INFO Channel 00/0 : 18[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153565:153738 [3] NCCL INFO Channel 09/0 : 91[3] -> 107[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96566:96755 [3] NCCL INFO Channel 08/0 : 19[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153569:153739 [7] NCCL INFO Channel 05/0 : 95[7] -> 63[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96563:96756 [0] NCCL INFO Channel 04/0 : 16[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96567:96760 [4] NCCL INFO Channel 06/0 : 20[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153564:153737 [2] NCCL INFO Channel 13/0 : 90[2] -> 106[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235650:235824 [2] NCCL INFO Channel 11/0 : 106[2] -> 90[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153562:153735 [0] NCCL INFO Channel 15/0 : 72[0] -> 88[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO Channel 10/0 : 97[1] -> 65[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235652:235822 [4] NCCL INFO Channel 15/0 : 108[4] -> 92[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153565:153738 [3] NCCL INFO Channel 13/0 : 91[3] -> 107[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO Channel 14/0 : 98[2] -> 66[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235649:235827 [1] NCCL INFO Channel 11/0 : 105[1] -> 89[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153569:153739 [7] NCCL INFO Channel 07/0 : 95[7] -> 63[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO Channel 04/0 : 99[3] -> 67[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235654:235821 [6] NCCL INFO Channel 15/0 : 110[6] -> 94[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153564:153737 [2] NCCL INFO Channel 15/0 : 90[2] -> 106[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153568:153736 [6] NCCL INFO Channel 01/0 : 62[6] -> 94[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO Channel 12/0 : 97[1] -> 65[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235651:235823 [3] NCCL INFO Channel 15/0 : 107[3] -> 91[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96569:96758 [6] NCCL INFO Channel 04/0 : 22[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96565:96754 [2] NCCL INFO Channel 04/0 : 18[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153565:153738 [3] NCCL INFO Channel 15/0 : 91[3] -> 107[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153562:153735 [0] NCCL INFO Channel 01/0 : 88[0] -> 104[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235650:235824 [2] NCCL INFO Channel 13/0 : 106[2] -> 90[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96566:96755 [3] NCCL INFO Channel 10/0 : 19[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153569:153739 [7] NCCL INFO Channel 09/0 : 95[7] -> 63[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO Channel 01/0 : 88[0] -> 72[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96563:96756 [0] NCCL INFO Channel 06/0 : 16[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96567:96760 [4] NCCL INFO Channel 08/0 : 20[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153568:153736 [6] NCCL INFO Channel 03/0 : 62[6] -> 94[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153562:153735 [0] NCCL INFO Channel 03/0 : 88[0] -> 104[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153569:153739 [7] NCCL INFO Channel 11/0 : 95[7] -> 63[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153566:153741 [4] NCCL INFO Channel 01/0 : 60[4] -> 92[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153568:153736 [6] NCCL INFO Channel 05/0 : 62[6] -> 94[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO Channel 02/0 : 96[0] -> 64[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153563:153740 [1] NCCL INFO Channel 01/0 : 57[1] -> 89[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO Channel 06/0 : 99[3] -> 67[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153569:153739 [7] NCCL INFO Channel 13/0 : 95[7] -> 63[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO Channel 14/0 : 97[1] -> 65[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153566:153741 [4] NCCL INFO Channel 03/0 : 60[4] -> 92[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153568:153736 [6] NCCL INFO Channel 07/0 : 62[6] -> 94[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153562:153735 [0] NCCL INFO Channel 05/0 : 88[0] -> 104[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO Channel 03/0 : 88[0] -> 72[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153563:153740 [1] NCCL INFO Channel 03/0 : 57[1] -> 89[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO Channel 00/0 : 55[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO Channel 00/0 : 53[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153568:153736 [6] NCCL INFO Channel 09/0 : 62[6] -> 94[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153566:153741 [4] NCCL INFO Channel 05/0 : 60[4] -> 92[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153562:153735 [0] NCCL INFO Channel 07/0 : 88[0] -> 104[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO Channel 04/0 : 96[0] -> 64[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO Channel 08/0 : 99[3] -> 67[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153563:153740 [1] NCCL INFO Channel 05/0 : 57[1] -> 89[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153565:153738 [3] NCCL INFO Channel 01/0 : 59[3] -> 91[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153564:153737 [2] NCCL INFO Channel 01/0 : 58[2] -> 90[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153568:153736 [6] NCCL INFO Channel 11/0 : 62[6] -> 94[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153566:153741 [4] NCCL INFO Channel 07/0 : 60[4] -> 92[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO Channel 05/0 : 88[0] -> 72[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153562:153735 [0] NCCL INFO Channel 09/0 : 88[0] -> 104[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO Channel 00/0 : 49[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153563:153740 [1] NCCL INFO Channel 07/0 : 57[1] -> 89[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153565:153738 [3] NCCL INFO Channel 03/0 : 59[3] -> 91[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153564:153737 [2] NCCL INFO Channel 03/0 : 58[2] -> 90[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO Channel 10/0 : 99[3] -> 67[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153568:153736 [6] NCCL INFO Channel 13/0 : 62[6] -> 94[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO Channel 06/0 : 96[0] -> 64[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153566:153741 [4] NCCL INFO Channel 09/0 : 60[4] -> 92[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153562:153735 [0] NCCL INFO Channel 11/0 : 88[0] -> 104[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153565:153738 [3] NCCL INFO Channel 05/0 : 59[3] -> 91[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153563:153740 [1] NCCL INFO Channel 11/0 : 57[1] -> 89[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO Channel 02/0 : 55[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO Channel 02/0 : 53[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153564:153737 [2] NCCL INFO Channel 05/0 : 58[2] -> 90[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO Channel 07/0 : 88[0] -> 72[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO Channel 02/0 : 49[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153568:153736 [6] NCCL INFO Channel 15/0 : 62[6] -> 94[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO Channel 12/0 : 99[3] -> 67[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO Channel 08/0 : 96[0] -> 64[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO Channel 09/0 : 88[0] -> 72[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO Channel 04/0 : 55[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO Channel 04/0 : 53[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153562:153735 [0] NCCL INFO Channel 13/0 : 88[0] -> 104[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153566:153741 [4] NCCL INFO Channel 11/0 : 60[4] -> 92[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153565:153738 [3] NCCL INFO Channel 07/0 : 59[3] -> 91[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO Channel 14/0 : 99[3] -> 67[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153564:153737 [2] NCCL INFO Channel 07/0 : 58[2] -> 90[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO Channel 10/0 : 96[0] -> 64[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153563:153740 [1] NCCL INFO Channel 13/0 : 57[1] -> 89[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153562:153735 [0] NCCL INFO Channel 15/0 : 88[0] -> 104[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153566:153741 [4] NCCL INFO Channel 13/0 : 60[4] -> 92[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153565:153738 [3] NCCL INFO Channel 09/0 : 59[3] -> 91[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153564:153737 [2] NCCL INFO Channel 09/0 : 58[2] -> 90[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153563:153740 [1] NCCL INFO Channel 15/0 : 57[1] -> 89[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO Channel 04/0 : 49[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153568:153736 [6] NCCL INFO Channel 01/0 : 94[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153566:153741 [4] NCCL INFO Channel 15/0 : 60[4] -> 92[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO Channel 12/0 : 96[0] -> 64[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO Channel 11/0 : 88[0] -> 72[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153565:153738 [3] NCCL INFO Channel 13/0 : 59[3] -> 91[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153564:153737 [2] NCCL INFO Channel 11/0 : 58[2] -> 90[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153568:153736 [6] NCCL INFO Channel 03/0 : 94[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153564:153737 [2] NCCL INFO Channel 13/0 : 58[2] -> 90[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153565:153738 [3] NCCL INFO Channel 15/0 : 59[3] -> 91[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153568:153736 [6] NCCL INFO Channel 05/0 : 94[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO Channel 06/0 : 53[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO Channel 06/0 : 55[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153563:153740 [1] NCCL INFO Channel 01/0 : 89[1] -> 57[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153568:153736 [6] NCCL INFO Channel 07/0 : 94[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153564:153737 [2] NCCL INFO Channel 15/0 : 58[2] -> 90[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO Channel 14/0 : 96[0] -> 64[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153566:153741 [4] NCCL INFO Channel 01/0 : 92[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO Channel 13/0 : 88[0] -> 72[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153566:153741 [4] NCCL INFO Channel 03/0 : 92[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153566:153741 [4] NCCL INFO Channel 05/0 : 92[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO Channel 06/0 : 49[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153566:153741 [4] NCCL INFO Channel 07/0 : 92[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO Channel 15/0 : 88[0] -> 72[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO Channel 08/0 : 53[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153566:153741 [4] NCCL INFO Channel 09/0 : 92[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO Channel 08/0 : 55[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO Channel 08/0 : 49[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153566:153741 [4] NCCL INFO Channel 11/0 : 92[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO Channel 10/0 : 53[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153566:153741 [4] NCCL INFO Channel 13/0 : 92[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO Channel 10/0 : 55[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO Channel 10/0 : 49[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153566:153741 [4] NCCL INFO Channel 15/0 : 92[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO Channel 12/0 : 53[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153563:153740 [1] NCCL INFO Channel 03/0 : 89[1] -> 57[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO Channel 12/0 : 55[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO Channel 12/0 : 49[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO Channel 14/0 : 53[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153565:153738 [3] NCCL INFO Channel 01/0 : 91[3] -> 59[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO Channel 14/0 : 55[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO Channel 14/0 : 49[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153563:153740 [1] NCCL INFO Channel 05/0 : 89[1] -> 57[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153568:153736 [6] NCCL INFO Channel 09/0 : 94[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153565:153738 [3] NCCL INFO Channel 03/0 : 91[3] -> 59[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153563:153740 [1] NCCL INFO Channel 07/0 : 89[1] -> 57[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153568:153736 [6] NCCL INFO Channel 11/0 : 94[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153565:153738 [3] NCCL INFO Channel 05/0 : 91[3] -> 59[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153563:153740 [1] NCCL INFO Channel 11/0 : 89[1] -> 57[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153568:153736 [6] NCCL INFO Channel 13/0 : 94[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153565:153738 [3] NCCL INFO Channel 07/0 : 91[3] -> 59[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153563:153740 [1] NCCL INFO Channel 13/0 : 89[1] -> 57[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153568:153736 [6] NCCL INFO Channel 15/0 : 94[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153565:153738 [3] NCCL INFO Channel 09/0 : 91[3] -> 59[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153563:153740 [1] NCCL INFO Channel 15/0 : 89[1] -> 57[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153562:153735 [0] NCCL INFO Channel 01/0 : 56[0] -> 88[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153565:153738 [3] NCCL INFO Channel 13/0 : 91[3] -> 59[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO Channel 00/0 : 51[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153565:153738 [3] NCCL INFO Channel 15/0 : 91[3] -> 59[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153564:153737 [2] NCCL INFO Channel 01/0 : 90[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153562:153735 [0] NCCL INFO Channel 03/0 : 56[0] -> 88[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153564:153737 [2] NCCL INFO Channel 03/0 : 90[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO Channel 02/0 : 51[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO Channel 00/0 : 52[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO Channel 00/0 : 54[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153562:153735 [0] NCCL INFO Channel 05/0 : 56[0] -> 88[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153564:153737 [2] NCCL INFO Channel 05/0 : 90[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO Channel 04/0 : 51[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153562:153735 [0] NCCL INFO Channel 07/0 : 56[0] -> 88[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153564:153737 [2] NCCL INFO Channel 07/0 : 90[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO Channel 02/0 : 54[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO Channel 02/0 : 52[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO Channel 02/0 : 48[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO Channel 00/0 : 50[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153562:153735 [0] NCCL INFO Channel 09/0 : 56[0] -> 88[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153564:153737 [2] NCCL INFO Channel 09/0 : 90[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO Channel 06/0 : 51[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153562:153735 [0] NCCL INFO Channel 11/0 : 56[0] -> 88[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153564:153737 [2] NCCL INFO Channel 11/0 : 90[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO Channel 04/0 : 54[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO Channel 06/0 : 52[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153562:153735 [0] NCCL INFO Channel 13/0 : 56[0] -> 88[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO Channel 04/0 : 50[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO Channel 04/0 : 48[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO Channel 08/0 : 51[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153564:153737 [2] NCCL INFO Channel 13/0 : 90[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153562:153735 [0] NCCL INFO Channel 15/0 : 56[0] -> 88[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO Channel 08/0 : 52[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153564:153737 [2] NCCL INFO Channel 15/0 : 90[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO Channel 10/0 : 51[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO Channel 06/0 : 50[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO Channel 06/0 : 48[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO Channel 10/0 : 52[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO Channel 12/0 : 51[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153562:153735 [0] NCCL INFO Channel 01/0 : 88[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO Channel 12/0 : 52[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO Channel 08/0 : 50[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO Channel 08/0 : 48[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153562:153735 [0] NCCL INFO Channel 03/0 : 88[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO Channel 14/0 : 51[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153562:153735 [0] NCCL INFO Channel 05/0 : 88[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO Channel 14/0 : 52[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO Channel 10/0 : 50[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO Channel 10/0 : 48[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153562:153735 [0] NCCL INFO Channel 07/0 : 88[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO Channel 12/0 : 50[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO Channel 12/0 : 48[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153562:153735 [0] NCCL INFO Channel 09/0 : 88[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO Channel 14/0 : 48[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153562:153735 [0] NCCL INFO Channel 11/0 : 88[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO Channel 14/0 : 50[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO Channel 08/0 : 54[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153562:153735 [0] NCCL INFO Channel 13/0 : 88[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO Channel 10/0 : 54[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153562:153735 [0] NCCL INFO Channel 15/0 : 88[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO Channel 12/0 : 54[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO Channel 14/0 : 54[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96892:97064 [4] NCCL INFO Connected all trees
+ip-26-0-161-103:96892:97064 [4] NCCL INFO NVLS comm 0x9b9ac80 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO Connected all trees
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO NVLS comm 0x9e64d60 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-103:96895:97069 [7] NCCL INFO Connected all trees
+ip-26-0-161-103:96895:97069 [7] NCCL INFO NVLS comm 0xa4213d0 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-103:96888:97066 [0] NCCL INFO Connected all trees
+ip-26-0-161-103:96888:97066 [0] NCCL INFO NVLS comm 0x8e3cea0 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-103:96891:97063 [3] NCCL INFO Connected all trees
+ip-26-0-161-103:96891:97063 [3] NCCL INFO NVLS comm 0x8c696e0 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-123:32708:32879 [4] NCCL INFO Channel 00/0 : 20[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32708:32879 [4] NCCL INFO Channel 02/0 : 20[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32708:32879 [4] NCCL INFO Channel 06/0 : 20[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32708:32879 [4] NCCL INFO Channel 08/0 : 20[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32708:32879 [4] NCCL INFO Channel 10/0 : 20[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32708:32879 [4] NCCL INFO Channel 12/0 : 20[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32708:32879 [4] NCCL INFO Channel 14/0 : 20[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO Connected all trees
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO NVLS comm 0xa3ae1e0 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-123:32704:32881 [0] NCCL INFO Channel 02/0 : 16[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32710:32883 [6] NCCL INFO Channel 00/0 : 22[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32704:32881 [0] NCCL INFO Channel 04/0 : 16[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32710:32883 [6] NCCL INFO Channel 02/0 : 22[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32704:32881 [0] NCCL INFO Channel 06/0 : 16[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32710:32883 [6] NCCL INFO Channel 04/0 : 22[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32704:32881 [0] NCCL INFO Channel 08/0 : 16[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32710:32883 [6] NCCL INFO Channel 08/0 : 22[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32704:32881 [0] NCCL INFO Channel 10/0 : 16[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32710:32883 [6] NCCL INFO Channel 10/0 : 22[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32704:32881 [0] NCCL INFO Channel 12/0 : 16[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32710:32883 [6] NCCL INFO Channel 12/0 : 22[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32704:32881 [0] NCCL INFO Channel 14/0 : 16[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32710:32883 [6] NCCL INFO Channel 14/0 : 22[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96894:97068 [6] NCCL INFO Connected all trees
+ip-26-0-161-103:96893:97067 [5] NCCL INFO Connected all trees
+ip-26-0-161-103:96894:97068 [6] NCCL INFO NVLS comm 0x8a7aba0 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-103:96893:97067 [5] NCCL INFO NVLS comm 0x86ea090 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-123:32709:32882 [5] NCCL INFO Channel 00/0 : 21[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32706:32877 [2] NCCL INFO Channel 00/0 : 18[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32707:32878 [3] NCCL INFO Channel 00/0 : 19[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32711:32884 [7] NCCL INFO Channel 00/0 : 23[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32705:32880 [1] NCCL INFO Channel 00/0 : 17[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32706:32877 [2] NCCL INFO Channel 04/0 : 18[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32707:32878 [3] NCCL INFO Channel 02/0 : 19[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32709:32882 [5] NCCL INFO Channel 02/0 : 21[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32711:32884 [7] NCCL INFO Channel 02/0 : 23[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32705:32880 [1] NCCL INFO Channel 02/0 : 17[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32707:32878 [3] NCCL INFO Channel 04/0 : 19[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32706:32877 [2] NCCL INFO Channel 06/0 : 18[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32709:32882 [5] NCCL INFO Channel 04/0 : 21[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32711:32884 [7] NCCL INFO Channel 04/0 : 23[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32705:32880 [1] NCCL INFO Channel 04/0 : 17[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32706:32877 [2] NCCL INFO Channel 08/0 : 18[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32707:32878 [3] NCCL INFO Channel 06/0 : 19[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32709:32882 [5] NCCL INFO Channel 06/0 : 21[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32711:32884 [7] NCCL INFO Channel 06/0 : 23[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32705:32880 [1] NCCL INFO Channel 06/0 : 17[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32706:32877 [2] NCCL INFO Channel 10/0 : 18[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32707:32878 [3] NCCL INFO Channel 08/0 : 19[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32709:32882 [5] NCCL INFO Channel 08/0 : 21[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32705:32880 [1] NCCL INFO Channel 08/0 : 17[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32711:32884 [7] NCCL INFO Channel 08/0 : 23[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32706:32877 [2] NCCL INFO Channel 12/0 : 18[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32707:32878 [3] NCCL INFO Channel 10/0 : 19[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32705:32880 [1] NCCL INFO Channel 10/0 : 17[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32709:32882 [5] NCCL INFO Channel 10/0 : 21[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32706:32877 [2] NCCL INFO Channel 14/0 : 18[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32711:32884 [7] NCCL INFO Channel 10/0 : 23[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32707:32878 [3] NCCL INFO Channel 12/0 : 19[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32705:32880 [1] NCCL INFO Channel 12/0 : 17[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32709:32882 [5] NCCL INFO Channel 12/0 : 21[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32711:32884 [7] NCCL INFO Channel 12/0 : 23[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32707:32878 [3] NCCL INFO Channel 14/0 : 19[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32705:32880 [1] NCCL INFO Channel 14/0 : 17[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32709:32882 [5] NCCL INFO Channel 14/0 : 21[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32711:32884 [7] NCCL INFO Channel 14/0 : 23[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO Connected all trees
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO NVLS comm 0x963e6f0 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-103:96890:97070 [2] NCCL INFO Connected all trees
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO Connected all trees
+ip-26-0-161-103:96890:97070 [2] NCCL INFO NVLS comm 0x8e18080 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO NVLS comm 0x9ad1540 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-142:62886:63057 [5] NCCL INFO Channel 01/0 : 29[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62882:63056 [1] NCCL INFO Channel 01/0 : 25[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62886:63057 [5] NCCL INFO Channel 03/0 : 29[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62888:63060 [7] NCCL INFO Channel 01/0 : 31[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62886:63057 [5] NCCL INFO Channel 05/0 : 29[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62882:63056 [1] NCCL INFO Channel 03/0 : 25[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62888:63060 [7] NCCL INFO Channel 03/0 : 31[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62888:63060 [7] NCCL INFO Channel 05/0 : 31[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62886:63057 [5] NCCL INFO Channel 07/0 : 29[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62882:63056 [1] NCCL INFO Channel 05/0 : 25[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62888:63060 [7] NCCL INFO Channel 07/0 : 31[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62882:63056 [1] NCCL INFO Channel 07/0 : 25[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62886:63057 [5] NCCL INFO Channel 09/0 : 29[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62888:63060 [7] NCCL INFO Channel 09/0 : 31[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62882:63056 [1] NCCL INFO Channel 11/0 : 25[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62886:63057 [5] NCCL INFO Channel 11/0 : 29[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62888:63060 [7] NCCL INFO Channel 11/0 : 31[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62882:63056 [1] NCCL INFO Channel 13/0 : 25[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62886:63057 [5] NCCL INFO Channel 15/0 : 29[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62888:63060 [7] NCCL INFO Channel 13/0 : 31[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62882:63056 [1] NCCL INFO Channel 15/0 : 25[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62884:63061 [3] NCCL INFO Channel 01/0 : 27[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62884:63061 [3] NCCL INFO Channel 03/0 : 27[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62884:63061 [3] NCCL INFO Channel 05/0 : 27[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62881:63059 [0] NCCL INFO Channel 01/0 : 24[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62885:63054 [4] NCCL INFO Channel 01/0 : 28[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62883:63058 [2] NCCL INFO Channel 01/0 : 26[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62884:63061 [3] NCCL INFO Channel 07/0 : 27[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62881:63059 [0] NCCL INFO Channel 03/0 : 24[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62885:63054 [4] NCCL INFO Channel 03/0 : 28[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62887:63055 [6] NCCL INFO Channel 01/0 : 30[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62883:63058 [2] NCCL INFO Channel 03/0 : 26[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62884:63061 [3] NCCL INFO Channel 09/0 : 27[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62881:63059 [0] NCCL INFO Channel 05/0 : 24[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62885:63054 [4] NCCL INFO Channel 05/0 : 28[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62887:63055 [6] NCCL INFO Channel 03/0 : 30[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62883:63058 [2] NCCL INFO Channel 05/0 : 26[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62884:63061 [3] NCCL INFO Channel 13/0 : 27[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62881:63059 [0] NCCL INFO Channel 07/0 : 24[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62887:63055 [6] NCCL INFO Channel 05/0 : 30[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62885:63054 [4] NCCL INFO Channel 07/0 : 28[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62884:63061 [3] NCCL INFO Channel 15/0 : 27[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62883:63058 [2] NCCL INFO Channel 07/0 : 26[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62881:63059 [0] NCCL INFO Channel 09/0 : 24[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62887:63055 [6] NCCL INFO Channel 07/0 : 30[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62885:63054 [4] NCCL INFO Channel 09/0 : 28[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62881:63059 [0] NCCL INFO Channel 11/0 : 24[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62883:63058 [2] NCCL INFO Channel 09/0 : 26[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62885:63054 [4] NCCL INFO Channel 11/0 : 28[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62887:63055 [6] NCCL INFO Channel 09/0 : 30[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62881:63059 [0] NCCL INFO Channel 13/0 : 24[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62883:63058 [2] NCCL INFO Channel 11/0 : 26[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62885:63054 [4] NCCL INFO Channel 13/0 : 28[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62887:63055 [6] NCCL INFO Channel 11/0 : 30[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62881:63059 [0] NCCL INFO Channel 15/0 : 24[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62883:63058 [2] NCCL INFO Channel 13/0 : 26[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62885:63054 [4] NCCL INFO Channel 15/0 : 28[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62887:63055 [6] NCCL INFO Channel 13/0 : 30[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62883:63058 [2] NCCL INFO Channel 15/0 : 26[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62887:63055 [6] NCCL INFO Channel 15/0 : 30[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96889:97065 [1] NCCL INFO Connected all trees
+ip-26-0-161-103:96889:97065 [1] NCCL INFO NVLS comm 0x87d4fe0 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-168-30:235649:235827 [1] NCCL INFO Channel 13/0 : 105[1] -> 89[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96569:96758 [6] NCCL INFO Channel 08/0 : 22[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235650:235824 [2] NCCL INFO Channel 15/0 : 106[2] -> 90[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96565:96754 [2] NCCL INFO Channel 06/0 : 18[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96566:96755 [3] NCCL INFO Channel 12/0 : 19[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235649:235827 [1] NCCL INFO Channel 15/0 : 105[1] -> 89[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96567:96760 [4] NCCL INFO Channel 10/0 : 20[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96563:96756 [0] NCCL INFO Channel 08/0 : 16[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235648:235826 [0] NCCL INFO Channel 01/0 : 104[0] -> 88[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96569:96758 [6] NCCL INFO Channel 10/0 : 22[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235648:235826 [0] NCCL INFO Channel 03/0 : 104[0] -> 88[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96565:96754 [2] NCCL INFO Channel 08/0 : 18[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96566:96755 [3] NCCL INFO Channel 14/0 : 19[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235648:235826 [0] NCCL INFO Channel 05/0 : 104[0] -> 88[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96567:96760 [4] NCCL INFO Channel 12/0 : 20[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235648:235826 [0] NCCL INFO Channel 07/0 : 104[0] -> 88[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96563:96756 [0] NCCL INFO Channel 10/0 : 16[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235648:235826 [0] NCCL INFO Channel 09/0 : 104[0] -> 88[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96569:96758 [6] NCCL INFO Channel 12/0 : 22[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235648:235826 [0] NCCL INFO Channel 11/0 : 104[0] -> 88[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96566:96755 [3] NCCL INFO Channel 00/0 : 35[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96565:96754 [2] NCCL INFO Channel 10/0 : 18[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235648:235826 [0] NCCL INFO Channel 13/0 : 104[0] -> 88[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96567:96760 [4] NCCL INFO Channel 14/0 : 20[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235648:235826 [0] NCCL INFO Channel 15/0 : 104[0] -> 88[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96569:96758 [6] NCCL INFO Channel 14/0 : 22[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96563:96756 [0] NCCL INFO Channel 12/0 : 16[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96566:96755 [3] NCCL INFO Channel 02/0 : 35[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96565:96754 [2] NCCL INFO Channel 12/0 : 18[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96567:96760 [4] NCCL INFO Channel 00/0 : 36[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96569:96758 [6] NCCL INFO Channel 00/0 : 38[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96563:96756 [0] NCCL INFO Channel 14/0 : 16[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96566:96755 [3] NCCL INFO Channel 04/0 : 35[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96565:96754 [2] NCCL INFO Channel 14/0 : 18[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96567:96760 [4] NCCL INFO Channel 02/0 : 36[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96569:96758 [6] NCCL INFO Channel 02/0 : 38[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96563:96756 [0] NCCL INFO Channel 02/0 : 32[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96566:96755 [3] NCCL INFO Channel 06/0 : 35[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96565:96754 [2] NCCL INFO Channel 00/0 : 34[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96567:96760 [4] NCCL INFO Channel 06/0 : 36[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96569:96758 [6] NCCL INFO Channel 04/0 : 38[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96563:96756 [0] NCCL INFO Channel 04/0 : 32[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96566:96755 [3] NCCL INFO Channel 08/0 : 35[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96565:96754 [2] NCCL INFO Channel 04/0 : 34[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96567:96760 [4] NCCL INFO Channel 08/0 : 36[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96569:96758 [6] NCCL INFO Channel 08/0 : 38[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96563:96756 [0] NCCL INFO Channel 06/0 : 32[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96566:96755 [3] NCCL INFO Channel 10/0 : 35[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96565:96754 [2] NCCL INFO Channel 06/0 : 34[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96567:96760 [4] NCCL INFO Channel 10/0 : 36[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96569:96758 [6] NCCL INFO Channel 10/0 : 38[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96563:96756 [0] NCCL INFO Channel 08/0 : 32[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96566:96755 [3] NCCL INFO Channel 12/0 : 35[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96565:96754 [2] NCCL INFO Channel 08/0 : 34[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96567:96760 [4] NCCL INFO Channel 12/0 : 36[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96569:96758 [6] NCCL INFO Channel 12/0 : 38[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96563:96756 [0] NCCL INFO Channel 10/0 : 32[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96566:96755 [3] NCCL INFO Channel 14/0 : 35[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96565:96754 [2] NCCL INFO Channel 10/0 : 34[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96567:96760 [4] NCCL INFO Channel 14/0 : 36[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96569:96758 [6] NCCL INFO Channel 14/0 : 38[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96563:96756 [0] NCCL INFO Channel 12/0 : 32[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96565:96754 [2] NCCL INFO Channel 12/0 : 34[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96563:96756 [0] NCCL INFO Channel 14/0 : 32[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96565:96754 [2] NCCL INFO Channel 14/0 : 34[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO Channel 01/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO Channel 01/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO Channel 03/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO Channel 01/0 : 7[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO Channel 03/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO Channel 05/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO Channel 03/0 : 7[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO Channel 01/0 : 5[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO Channel 01/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO Channel 05/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO Channel 07/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO Channel 05/0 : 7[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO Channel 03/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96895:97069 [7] NCCL INFO Channel 00/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96889:97065 [1] NCCL INFO Channel 00/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO Channel 03/0 : 5[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO Channel 07/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO Channel 01/0 : 1[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO Channel 09/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO Channel 01/0 : 3[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96893:97067 [5] NCCL INFO Channel 00/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO Channel 07/0 : 7[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO Channel 05/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO Channel 09/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO Channel 05/0 : 5[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96895:97069 [7] NCCL INFO Channel 01/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO Channel 03/0 : 1[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO Channel 11/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96889:97065 [1] NCCL INFO Channel 02/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO Channel 03/0 : 3[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO Channel 01/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO Channel 07/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96893:97067 [5] NCCL INFO Channel 01/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO Channel 09/0 : 7[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96895:97069 [7] NCCL INFO Channel 02/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO Channel 07/0 : 5[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96889:97065 [1] NCCL INFO Channel 03/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO Channel 11/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96893:97067 [5] NCCL INFO Channel 02/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO Channel 05/0 : 1[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO Channel 13/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96889:97065 [1] NCCL INFO Channel 04/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96895:97069 [7] NCCL INFO Channel 03/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO Channel 05/0 : 3[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO Channel 09/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO Channel 03/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO Channel 11/0 : 7[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO Channel 09/0 : 5[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO Channel 13/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96893:97067 [5] NCCL INFO Channel 03/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO Channel 07/0 : 1[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO Channel 15/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO Channel 07/0 : 3[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO Channel 11/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96889:97065 [1] NCCL INFO Channel 05/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96895:97069 [7] NCCL INFO Channel 04/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96893:97067 [5] NCCL INFO Channel 04/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO Channel 11/0 : 5[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO Channel 05/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO Channel 13/0 : 7[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96889:97065 [1] NCCL INFO Channel 06/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO Channel 15/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96895:97069 [7] NCCL INFO Channel 05/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96893:97067 [5] NCCL INFO Channel 06/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO Channel 11/0 : 1[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO Channel 09/0 : 3[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO Channel 00/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO Channel 13/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96889:97065 [1] NCCL INFO Channel 07/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96893:97067 [5] NCCL INFO Channel 07/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO Channel 07/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO Channel 15/0 : 5[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO Channel 01/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO Channel 00/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96895:97069 [7] NCCL INFO Channel 06/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96889:97065 [1] NCCL INFO Channel 08/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO Channel 13/0 : 1[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO Channel 13/0 : 3[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO Channel 15/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96893:97067 [5] NCCL INFO Channel 08/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO Channel 01/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96889:97065 [1] NCCL INFO Channel 10/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96895:97069 [7] NCCL INFO Channel 08/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO Channel 09/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO Channel 00/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO Channel 02/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO Channel 01/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96893:97067 [5] NCCL INFO Channel 09/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO Channel 15/0 : 3[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO Channel 15/0 : 1[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96889:97065 [1] NCCL INFO Channel 11/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO Channel 00/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96895:97069 [7] NCCL INFO Channel 09/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO Channel 02/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96893:97067 [5] NCCL INFO Channel 10/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO Channel 03/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO Channel 11/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO Channel 01/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO Channel 02/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO Channel 00/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96889:97065 [1] NCCL INFO Channel 12/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96893:97067 [5] NCCL INFO Channel 11/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO Channel 01/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO Channel 00/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96895:97069 [7] NCCL INFO Channel 10/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96889:97065 [1] NCCL INFO Channel 13/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96893:97067 [5] NCCL INFO Channel 12/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO Channel 03/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO Channel 04/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO Channel 02/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO Channel 13/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO Channel 01/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO Channel 03/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO Channel 03/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO Channel 02/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96895:97069 [7] NCCL INFO Channel 11/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96889:97065 [1] NCCL INFO Channel 14/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96893:97067 [5] NCCL INFO Channel 14/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO Channel 04/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96895:97069 [7] NCCL INFO Channel 12/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96889:97065 [1] NCCL INFO Channel 15/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96893:97067 [5] NCCL INFO Channel 15/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO Channel 03/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO Channel 02/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO Channel 04/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96895:97069 [7] NCCL INFO Channel 13/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96889:97065 [1] NCCL INFO Channel 00/0 : 17[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96893:97067 [5] NCCL INFO Channel 00/0 : 21[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO Channel 03/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO Channel 04/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO Channel 15/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO Channel 05/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO Channel 05/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96895:97069 [7] NCCL INFO Channel 14/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO Channel 04/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO Channel 04/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96889:97065 [1] NCCL INFO Channel 02/0 : 17[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96893:97067 [5] NCCL INFO Channel 02/0 : 21[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96892:97064 [4] NCCL INFO Channel 00/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96895:97069 [7] NCCL INFO Channel 00/0 : 23[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO Channel 05/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO Channel 04/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO Channel 00/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO Channel 07/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO Channel 06/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO Channel 05/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO Channel 05/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96894:97068 [6] NCCL INFO Channel 00/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO Channel 06/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96889:97065 [1] NCCL INFO Channel 04/0 : 17[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96893:97067 [5] NCCL INFO Channel 04/0 : 21[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO Channel 06/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO Channel 05/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96895:97069 [7] NCCL INFO Channel 02/0 : 23[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96892:97064 [4] NCCL INFO Channel 01/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96891:97063 [3] NCCL INFO Channel 00/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96889:97065 [1] NCCL INFO Channel 06/0 : 17[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO Channel 06/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO Channel 08/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO Channel 06/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO Channel 07/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO Channel 01/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO Channel 07/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO Channel 07/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO Channel 06/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96894:97068 [6] NCCL INFO Channel 01/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96893:97067 [5] NCCL INFO Channel 06/0 : 21[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96888:97066 [0] NCCL INFO Channel 01/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96895:97069 [7] NCCL INFO Channel 04/0 : 23[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96892:97064 [4] NCCL INFO Channel 02/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO Channel 07/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96891:97063 [3] NCCL INFO Channel 01/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO Channel 09/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96889:97065 [1] NCCL INFO Channel 08/0 : 17[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96890:97070 [2] NCCL INFO Channel 00/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96894:97068 [6] NCCL INFO Channel 02/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96893:97067 [5] NCCL INFO Channel 08/0 : 21[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96888:97066 [0] NCCL INFO Channel 02/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96895:97069 [7] NCCL INFO Channel 06/0 : 23[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO Channel 08/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO Channel 08/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO Channel 07/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO Channel 02/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO Channel 09/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO Channel 08/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO Channel 08/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96892:97064 [4] NCCL INFO Channel 03/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96889:97065 [1] NCCL INFO Channel 10/0 : 17[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96891:97063 [3] NCCL INFO Channel 02/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96890:97070 [2] NCCL INFO Channel 01/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96893:97067 [5] NCCL INFO Channel 10/0 : 21[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96894:97068 [6] NCCL INFO Channel 03/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96895:97069 [7] NCCL INFO Channel 08/0 : 23[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96888:97066 [0] NCCL INFO Channel 03/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO Channel 10/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96889:97065 [1] NCCL INFO Channel 12/0 : 17[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96892:97064 [4] NCCL INFO Channel 05/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96891:97063 [3] NCCL INFO Channel 04/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO Channel 09/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO Channel 09/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96890:97070 [2] NCCL INFO Channel 03/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO Channel 08/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO Channel 09/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO Channel 10/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO Channel 03/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO Channel 09/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO Channel 11/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96894:97068 [6] NCCL INFO Channel 04/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96893:97067 [5] NCCL INFO Channel 12/0 : 21[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96895:97069 [7] NCCL INFO Channel 10/0 : 23[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96888:97066 [0] NCCL INFO Channel 04/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO Channel 11/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96889:97065 [1] NCCL INFO Channel 14/0 : 17[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO Channel 10/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96892:97064 [4] NCCL INFO Channel 06/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO Channel 10/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96891:97063 [3] NCCL INFO Channel 05/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96890:97070 [2] NCCL INFO Channel 04/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96894:97068 [6] NCCL INFO Channel 05/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96893:97067 [5] NCCL INFO Channel 14/0 : 21[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96895:97069 [7] NCCL INFO Channel 12/0 : 23[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO Channel 10/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96888:97066 [0] NCCL INFO Channel 05/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96892:97064 [4] NCCL INFO Channel 07/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO Channel 11/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO Channel 12/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO Channel 05/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO Channel 10/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO Channel 12/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO Channel 11/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO Channel 11/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96891:97063 [3] NCCL INFO Channel 06/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96890:97070 [2] NCCL INFO Channel 05/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96895:97069 [7] NCCL INFO Channel 14/0 : 23[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96894:97068 [6] NCCL INFO Channel 07/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO Channel 12/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO Channel 13/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO Channel 12/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96892:97064 [4] NCCL INFO Channel 08/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96888:97066 [0] NCCL INFO Channel 06/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96891:97063 [3] NCCL INFO Channel 07/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96890:97070 [2] NCCL INFO Channel 06/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO Channel 13/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96894:97068 [6] NCCL INFO Channel 08/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO Channel 06/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96892:97064 [4] NCCL INFO Channel 09/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO Channel 11/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO Channel 12/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO Channel 12/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO Channel 13/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO Channel 15/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96888:97066 [0] NCCL INFO Channel 07/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96891:97063 [3] NCCL INFO Channel 08/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96890:97070 [2] NCCL INFO Channel 07/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO Channel 14/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO Channel 13/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO Channel 07/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO Channel 12/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO Channel 13/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO Channel 14/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO Channel 13/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96894:97068 [6] NCCL INFO Channel 09/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96888:97066 [0] NCCL INFO Channel 09/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96892:97064 [4] NCCL INFO Channel 10/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96891:97063 [3] NCCL INFO Channel 09/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96890:97070 [2] NCCL INFO Channel 08/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO Channel 15/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO Channel 14/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO Channel 08/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96894:97068 [6] NCCL INFO Channel 10/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96888:97066 [0] NCCL INFO Channel 10/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96891:97063 [3] NCCL INFO Channel 10/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96892:97064 [4] NCCL INFO Channel 11/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96890:97070 [2] NCCL INFO Channel 09/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO Channel 14/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO Channel 15/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO Channel 14/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO Channel 14/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO Channel 15/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO Channel 09/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO Channel 15/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32705:32880 [1] NCCL INFO Channel 01/0 : 9[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32709:32882 [5] NCCL INFO Channel 01/0 : 13[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96894:97068 [6] NCCL INFO Channel 11/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96888:97066 [0] NCCL INFO Channel 11/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO Channel 15/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO Channel 10/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96890:97070 [2] NCCL INFO Channel 11/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96892:97064 [4] NCCL INFO Channel 13/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96891:97063 [3] NCCL INFO Channel 12/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32705:32880 [1] NCCL INFO Channel 03/0 : 9[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96894:97068 [6] NCCL INFO Channel 12/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96888:97066 [0] NCCL INFO Channel 12/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO Channel 11/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:124166:124339 [6] NCCL INFO Channel 00/0 : 70[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32709:32882 [5] NCCL INFO Channel 03/0 : 13[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32711:32884 [7] NCCL INFO Channel 01/0 : 15[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96892:97064 [4] NCCL INFO Channel 14/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96890:97070 [2] NCCL INFO Channel 12/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96894:97068 [6] NCCL INFO Channel 13/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96888:97066 [0] NCCL INFO Channel 13/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96891:97063 [3] NCCL INFO Channel 13/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO Channel 13/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32705:32880 [1] NCCL INFO Channel 05/0 : 9[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124162:124342 [2] NCCL INFO Channel 00/0 : 66[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:124166:124339 [6] NCCL INFO Channel 02/0 : 70[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO Channel 14/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32709:32882 [5] NCCL INFO Channel 05/0 : 13[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32711:32884 [7] NCCL INFO Channel 03/0 : 15[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32705:32880 [1] NCCL INFO Channel 07/0 : 9[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124160:124335 [0] NCCL INFO Channel 02/0 : 64[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32709:32882 [5] NCCL INFO Channel 07/0 : 13[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124162:124342 [2] NCCL INFO Channel 04/0 : 66[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO Channel 15/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:124166:124339 [6] NCCL INFO Channel 04/0 : 70[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:124161:124337 [1] NCCL INFO Channel 00/0 : 65[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124163:124340 [3] NCCL INFO Channel 00/0 : 67[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32711:32884 [7] NCCL INFO Channel 05/0 : 15[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124160:124335 [0] NCCL INFO Channel 04/0 : 64[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO Channel 01/0 : 15[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32705:32880 [1] NCCL INFO Channel 11/0 : 9[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124162:124342 [2] NCCL INFO Channel 06/0 : 66[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:124165:124341 [5] NCCL INFO Channel 00/0 : 69[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32709:32882 [5] NCCL INFO Channel 09/0 : 13[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124166:124339 [6] NCCL INFO Channel 08/0 : 70[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO Channel 01/0 : 9[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO Channel 01/0 : 13[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO Channel 03/0 : 15[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32711:32884 [7] NCCL INFO Channel 07/0 : 15[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32705:32880 [1] NCCL INFO Channel 13/0 : 9[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32709:32882 [5] NCCL INFO Channel 11/0 : 13[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124167:124338 [7] NCCL INFO Channel 00/0 : 71[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124161:124337 [1] NCCL INFO Channel 02/0 : 65[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124163:124340 [3] NCCL INFO Channel 02/0 : 67[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:124160:124335 [0] NCCL INFO Channel 06/0 : 64[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO Channel 03/0 : 9[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO Channel 03/0 : 13[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124165:124341 [5] NCCL INFO Channel 02/0 : 69[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124162:124342 [2] NCCL INFO Channel 08/0 : 66[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:124166:124339 [6] NCCL INFO Channel 10/0 : 70[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO Channel 05/0 : 15[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32711:32884 [7] NCCL INFO Channel 09/0 : 15[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32705:32880 [1] NCCL INFO Channel 15/0 : 9[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32709:32882 [5] NCCL INFO Channel 15/0 : 13[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO Channel 05/0 : 9[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO Channel 05/0 : 13[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO Channel 07/0 : 15[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32705:32880 [1] NCCL INFO Channel 01/0 : 25[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124161:124337 [1] NCCL INFO Channel 04/0 : 65[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124167:124338 [7] NCCL INFO Channel 02/0 : 71[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124163:124340 [3] NCCL INFO Channel 04/0 : 67[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:124160:124335 [0] NCCL INFO Channel 08/0 : 64[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:124162:124342 [2] NCCL INFO Channel 10/0 : 66[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:124166:124339 [6] NCCL INFO Channel 12/0 : 70[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:124165:124341 [5] NCCL INFO Channel 04/0 : 69[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO Channel 07/0 : 9[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO Channel 07/0 : 13[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32711:32884 [7] NCCL INFO Channel 11/0 : 15[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO Channel 09/0 : 15[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32709:32882 [5] NCCL INFO Channel 01/0 : 29[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32705:32880 [1] NCCL INFO Channel 03/0 : 25[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124164:124336 [4] NCCL INFO Channel 00/0 : 68[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:124167:124338 [7] NCCL INFO Channel 04/0 : 71[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO Channel 11/0 : 9[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO Channel 09/0 : 13[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32711:32884 [7] NCCL INFO Channel 13/0 : 15[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32709:32882 [5] NCCL INFO Channel 03/0 : 29[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124163:124340 [3] NCCL INFO Channel 06/0 : 67[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:124161:124337 [1] NCCL INFO Channel 06/0 : 65[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO Channel 11/0 : 15[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32705:32880 [1] NCCL INFO Channel 05/0 : 25[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124160:124335 [0] NCCL INFO Channel 10/0 : 64[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:124162:124342 [2] NCCL INFO Channel 12/0 : 66[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:124166:124339 [6] NCCL INFO Channel 14/0 : 70[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:124165:124341 [5] NCCL INFO Channel 06/0 : 69[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO Channel 13/0 : 9[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32709:32882 [5] NCCL INFO Channel 05/0 : 29[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32711:32884 [7] NCCL INFO Channel 01/0 : 31[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32705:32880 [1] NCCL INFO Channel 07/0 : 25[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124167:124338 [7] NCCL INFO Channel 06/0 : 71[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32709:32882 [5] NCCL INFO Channel 07/0 : 29[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124163:124340 [3] NCCL INFO Channel 08/0 : 67[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO Channel 13/0 : 15[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO Channel 11/0 : 13[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124161:124337 [1] NCCL INFO Channel 08/0 : 65[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124164:124336 [4] NCCL INFO Channel 02/0 : 68[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:124160:124335 [0] NCCL INFO Channel 12/0 : 64[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:124162:124342 [2] NCCL INFO Channel 14/0 : 66[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:124166:124339 [6] NCCL INFO Channel 00/0 : 6[6] -> 70[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32711:32884 [7] NCCL INFO Channel 03/0 : 31[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32705:32880 [1] NCCL INFO Channel 11/0 : 25[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124165:124341 [5] NCCL INFO Channel 08/0 : 69[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO Channel 15/0 : 9[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32709:32882 [5] NCCL INFO Channel 09/0 : 29[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124167:124338 [7] NCCL INFO Channel 08/0 : 71[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO Channel 15/0 : 13[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124163:124340 [3] NCCL INFO Channel 10/0 : 67[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:124161:124337 [1] NCCL INFO Channel 10/0 : 65[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32705:32880 [1] NCCL INFO Channel 13/0 : 25[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32711:32884 [7] NCCL INFO Channel 05/0 : 31[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32709:32882 [5] NCCL INFO Channel 11/0 : 29[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124164:124336 [4] NCCL INFO Channel 06/0 : 68[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32705:32880 [1] NCCL INFO Channel 15/0 : 25[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124162:124342 [2] NCCL INFO Channel 00/0 : 2[2] -> 66[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:124160:124335 [0] NCCL INFO Channel 14/0 : 64[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32711:32884 [7] NCCL INFO Channel 07/0 : 31[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124166:124339 [6] NCCL INFO Channel 02/0 : 6[6] -> 70[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:124165:124341 [5] NCCL INFO Channel 10/0 : 69[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32709:32882 [5] NCCL INFO Channel 15/0 : 29[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124167:124338 [7] NCCL INFO Channel 10/0 : 71[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124161:124337 [1] NCCL INFO Channel 12/0 : 65[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124163:124340 [3] NCCL INFO Channel 12/0 : 67[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:124164:124336 [4] NCCL INFO Channel 08/0 : 68[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:124162:124342 [2] NCCL INFO Channel 04/0 : 2[2] -> 66[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32711:32884 [7] NCCL INFO Channel 09/0 : 31[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124160:124335 [0] NCCL INFO Channel 02/0 : 0[0] -> 64[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:124166:124339 [6] NCCL INFO Channel 04/0 : 6[6] -> 70[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32711:32884 [7] NCCL INFO Channel 11/0 : 31[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124165:124341 [5] NCCL INFO Channel 12/0 : 69[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124167:124338 [7] NCCL INFO Channel 12/0 : 71[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124163:124340 [3] NCCL INFO Channel 14/0 : 67[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32711:32884 [7] NCCL INFO Channel 13/0 : 31[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124164:124336 [4] NCCL INFO Channel 10/0 : 68[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:124161:124337 [1] NCCL INFO Channel 14/0 : 65[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124162:124342 [2] NCCL INFO Channel 06/0 : 2[2] -> 66[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO Channel 01/0 : 12[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:124166:124339 [6] NCCL INFO Channel 08/0 : 6[6] -> 70[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:124160:124335 [0] NCCL INFO Channel 04/0 : 0[0] -> 64[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:124165:124341 [5] NCCL INFO Channel 14/0 : 69[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124167:124338 [7] NCCL INFO Channel 14/0 : 71[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO Channel 01/0 : 25[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124163:124340 [3] NCCL INFO Channel 00/0 : 3[3] -> 67[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO Channel 01/0 : 29[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124164:124336 [4] NCCL INFO Channel 12/0 : 68[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:124161:124337 [1] NCCL INFO Channel 00/0 : 1[1] -> 65[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124162:124342 [2] NCCL INFO Channel 08/0 : 2[2] -> 66[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO Channel 03/0 : 12[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:124166:124339 [6] NCCL INFO Channel 10/0 : 6[6] -> 70[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:124160:124335 [0] NCCL INFO Channel 06/0 : 0[0] -> 64[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO Channel 03/0 : 25[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124165:124341 [5] NCCL INFO Channel 00/0 : 5[5] -> 69[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO Channel 03/0 : 29[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32705:32880 [1] NCCL INFO Channel 01/0 : 25[1] -> 57[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124167:124338 [7] NCCL INFO Channel 00/0 : 7[7] -> 71[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124163:124340 [3] NCCL INFO Channel 02/0 : 3[3] -> 67[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO Channel 01/0 : 14[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32709:32882 [5] NCCL INFO Channel 01/0 : 29[5] -> 61[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32708:32879 [4] NCCL INFO Channel 01/0 : 12[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:124164:124336 [4] NCCL INFO Channel 14/0 : 68[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO Channel 05/0 : 12[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32705:32880 [1] NCCL INFO Channel 03/0 : 25[1] -> 57[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124162:124342 [2] NCCL INFO Channel 10/0 : 2[2] -> 66[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62886:63057 [5] NCCL INFO Channel 01/0 : 45[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62882:63056 [1] NCCL INFO Channel 01/0 : 41[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO Channel 01/0 : 8[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:124161:124337 [1] NCCL INFO Channel 02/0 : 1[1] -> 65[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO Channel 05/0 : 25[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124166:124339 [6] NCCL INFO Channel 12/0 : 6[6] -> 70[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO Channel 05/0 : 29[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32709:32882 [5] NCCL INFO Channel 03/0 : 29[5] -> 61[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96564:96757 [1] NCCL INFO Channel 00/0 : 33[1] -> 65[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32708:32879 [4] NCCL INFO Channel 03/0 : 12[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:124167:124338 [7] NCCL INFO Channel 02/0 : 7[7] -> 71[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124165:124341 [5] NCCL INFO Channel 02/0 : 5[5] -> 69[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124160:124335 [0] NCCL INFO Channel 08/0 : 0[0] -> 64[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:124163:124340 [3] NCCL INFO Channel 04/0 : 3[3] -> 67[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32705:32880 [1] NCCL INFO Channel 05/0 : 25[1] -> 57[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124164:124336 [4] NCCL INFO Channel 00/0 : 4[4] -> 68[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96892:97064 [4] NCCL INFO Channel 15/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO Channel 01/0 : 10[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:124162:124342 [2] NCCL INFO Channel 12/0 : 2[2] -> 66[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:124161:124337 [1] NCCL INFO Channel 04/0 : 1[1] -> 65[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96888:97066 [0] NCCL INFO Channel 14/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96890:97070 [2] NCCL INFO Channel 13/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO Channel 03/0 : 14[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:124166:124339 [6] NCCL INFO Channel 14/0 : 6[6] -> 70[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96570:96759 [7] NCCL INFO Channel 00/0 : 39[7] -> 71[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96894:97068 [6] NCCL INFO Channel 15/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO Channel 07/0 : 12[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32710:32883 [6] NCCL INFO Channel 01/0 : 14[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96564:96757 [1] NCCL INFO Channel 02/0 : 33[1] -> 65[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96891:97063 [3] NCCL INFO Channel 14/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96892:97064 [4] NCCL INFO Channel 00/0 : 20[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO Channel 01/0 : 31[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32709:32882 [5] NCCL INFO Channel 05/0 : 29[5] -> 61[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32704:32881 [0] NCCL INFO Channel 01/0 : 8[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96888:97066 [0] NCCL INFO Channel 15/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO Channel 07/0 : 25[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124167:124338 [7] NCCL INFO Channel 04/0 : 7[7] -> 71[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96894:97068 [6] NCCL INFO Channel 00/0 : 22[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO Channel 03/0 : 8[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO Channel 07/0 : 29[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124165:124341 [5] NCCL INFO Channel 04/0 : 5[5] -> 69[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96890:97070 [2] NCCL INFO Channel 14/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32708:32879 [4] NCCL INFO Channel 05/0 : 12[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:124160:124335 [0] NCCL INFO Channel 10/0 : 0[0] -> 64[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:124163:124340 [3] NCCL INFO Channel 06/0 : 3[3] -> 67[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96570:96759 [7] NCCL INFO Channel 02/0 : 39[7] -> 71[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96892:97064 [4] NCCL INFO Channel 02/0 : 20[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32705:32880 [1] NCCL INFO Channel 07/0 : 25[1] -> 57[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32711:32884 [7] NCCL INFO Channel 01/0 : 31[7] -> 63[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124164:124336 [4] NCCL INFO Channel 02/0 : 4[4] -> 68[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96564:96757 [1] NCCL INFO Channel 04/0 : 33[1] -> 65[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96568:96753 [5] NCCL INFO Channel 00/0 : 37[5] -> 69[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96891:97063 [3] NCCL INFO Channel 15/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32710:32883 [6] NCCL INFO Channel 03/0 : 14[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:124162:124342 [2] NCCL INFO Channel 14/0 : 2[2] -> 66[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:124161:124337 [1] NCCL INFO Channel 06/0 : 1[1] -> 65[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96889:97065 [1] NCCL INFO Channel 00/0 : 17[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO Channel 01/0 : 11[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:124167:124338 [7] NCCL INFO Channel 06/0 : 7[7] -> 71[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96895:97069 [7] NCCL INFO Channel 00/0 : 23[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO Channel 03/0 : 10[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:124165:124341 [5] NCCL INFO Channel 06/0 : 5[5] -> 69[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96888:97066 [0] NCCL INFO Channel 02/0 : 16[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO Channel 05/0 : 14[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32706:32877 [2] NCCL INFO Channel 01/0 : 10[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32709:32882 [5] NCCL INFO Channel 07/0 : 29[5] -> 61[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32704:32881 [0] NCCL INFO Channel 03/0 : 8[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:124160:124335 [0] NCCL INFO Channel 12/0 : 0[0] -> 64[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96570:96759 [7] NCCL INFO Channel 04/0 : 39[7] -> 71[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96893:97067 [5] NCCL INFO Channel 00/0 : 21[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO Channel 09/0 : 12[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32708:32879 [4] NCCL INFO Channel 07/0 : 12[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32705:32880 [1] NCCL INFO Channel 11/0 : 25[1] -> 57[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32711:32884 [7] NCCL INFO Channel 03/0 : 31[7] -> 63[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124164:124336 [4] NCCL INFO Channel 06/0 : 4[4] -> 68[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96564:96757 [1] NCCL INFO Channel 06/0 : 33[1] -> 65[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96894:97068 [6] NCCL INFO Channel 02/0 : 22[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96890:97070 [2] NCCL INFO Channel 15/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO Channel 11/0 : 25[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124161:124337 [1] NCCL INFO Channel 08/0 : 1[1] -> 65[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96568:96753 [5] NCCL INFO Channel 02/0 : 37[5] -> 69[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96892:97064 [4] NCCL INFO Channel 06/0 : 20[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO Channel 03/0 : 31[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124167:124338 [7] NCCL INFO Channel 08/0 : 7[7] -> 71[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96891:97063 [3] NCCL INFO Channel 00/0 : 19[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO Channel 09/0 : 29[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32707:32878 [3] NCCL INFO Channel 01/0 : 11[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32710:32883 [6] NCCL INFO Channel 05/0 : 14[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:124165:124341 [5] NCCL INFO Channel 08/0 : 5[5] -> 69[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124160:124335 [0] NCCL INFO Channel 14/0 : 0[0] -> 64[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:124163:124340 [3] NCCL INFO Channel 08/0 : 3[3] -> 67[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:124164:124336 [4] NCCL INFO Channel 08/0 : 4[4] -> 68[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96889:97065 [1] NCCL INFO Channel 02/0 : 17[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO Channel 05/0 : 8[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96895:97069 [7] NCCL INFO Channel 02/0 : 23[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO Channel 03/0 : 11[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO Channel 05/0 : 10[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:124161:124337 [1] NCCL INFO Channel 10/0 : 1[1] -> 65[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124167:124338 [7] NCCL INFO Channel 10/0 : 7[7] -> 71[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96570:96759 [7] NCCL INFO Channel 06/0 : 39[7] -> 71[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96888:97066 [0] NCCL INFO Channel 04/0 : 16[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96568:96753 [5] NCCL INFO Channel 04/0 : 37[5] -> 69[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96564:96757 [1] NCCL INFO Channel 08/0 : 33[1] -> 65[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96890:97070 [2] NCCL INFO Channel 00/0 : 18[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96894:97068 [6] NCCL INFO Channel 04/0 : 22[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96893:97067 [5] NCCL INFO Channel 02/0 : 21[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32709:32882 [5] NCCL INFO Channel 09/0 : 29[5] -> 61[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32706:32877 [2] NCCL INFO Channel 03/0 : 10[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32704:32881 [0] NCCL INFO Channel 05/0 : 8[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32705:32880 [1] NCCL INFO Channel 13/0 : 25[1] -> 57[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32711:32884 [7] NCCL INFO Channel 05/0 : 31[7] -> 63[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124163:124340 [3] NCCL INFO Channel 10/0 : 3[3] -> 67[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96570:96759 [7] NCCL INFO Channel 08/0 : 39[7] -> 71[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96892:97064 [4] NCCL INFO Channel 08/0 : 20[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO Channel 07/0 : 14[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO Channel 11/0 : 12[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32708:32879 [4] NCCL INFO Channel 09/0 : 12[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:124164:124336 [4] NCCL INFO Channel 10/0 : 4[4] -> 68[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96891:97063 [3] NCCL INFO Channel 02/0 : 19[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO Channel 13/0 : 25[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32710:32883 [6] NCCL INFO Channel 07/0 : 14[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:124165:124341 [5] NCCL INFO Channel 10/0 : 5[5] -> 69[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96889:97065 [1] NCCL INFO Channel 04/0 : 17[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO Channel 05/0 : 31[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124161:124337 [1] NCCL INFO Channel 12/0 : 1[1] -> 65[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124167:124338 [7] NCCL INFO Channel 12/0 : 7[7] -> 71[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96895:97069 [7] NCCL INFO Channel 04/0 : 23[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96888:97066 [0] NCCL INFO Channel 06/0 : 16[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO Channel 11/0 : 29[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124163:124340 [3] NCCL INFO Channel 12/0 : 3[3] -> 67[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96568:96753 [5] NCCL INFO Channel 06/0 : 37[5] -> 69[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96564:96757 [1] NCCL INFO Channel 10/0 : 33[1] -> 65[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96890:97070 [2] NCCL INFO Channel 04/0 : 18[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO Channel 07/0 : 8[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96570:96759 [7] NCCL INFO Channel 10/0 : 39[7] -> 71[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96894:97068 [6] NCCL INFO Channel 08/0 : 22[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO Channel 05/0 : 11[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO Channel 07/0 : 10[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32707:32878 [3] NCCL INFO Channel 03/0 : 11[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:124164:124336 [4] NCCL INFO Channel 12/0 : 4[4] -> 68[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:124165:124341 [5] NCCL INFO Channel 12/0 : 5[5] -> 69[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96892:97064 [4] NCCL INFO Channel 10/0 : 20[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO Channel 09/0 : 14[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32706:32877 [2] NCCL INFO Channel 05/0 : 10[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96891:97063 [3] NCCL INFO Channel 04/0 : 19[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96889:97065 [1] NCCL INFO Channel 06/0 : 17[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO Channel 13/0 : 12[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32709:32882 [5] NCCL INFO Channel 11/0 : 29[5] -> 61[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32704:32881 [0] NCCL INFO Channel 07/0 : 8[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96893:97067 [5] NCCL INFO Channel 04/0 : 21[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO Channel 15/0 : 25[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124167:124338 [7] NCCL INFO Channel 14/0 : 7[7] -> 71[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124161:124337 [1] NCCL INFO Channel 14/0 : 1[1] -> 65[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124163:124340 [3] NCCL INFO Channel 14/0 : 3[3] -> 67[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96568:96753 [5] NCCL INFO Channel 08/0 : 37[5] -> 69[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96564:96757 [1] NCCL INFO Channel 12/0 : 33[1] -> 65[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96888:97066 [0] NCCL INFO Channel 08/0 : 16[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96895:97069 [7] NCCL INFO Channel 06/0 : 23[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO Channel 07/0 : 31[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96570:96759 [7] NCCL INFO Channel 12/0 : 39[7] -> 71[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96890:97070 [2] NCCL INFO Channel 06/0 : 18[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO Channel 15/0 : 29[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96894:97068 [6] NCCL INFO Channel 10/0 : 22[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO Channel 09/0 : 8[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32705:32880 [1] NCCL INFO Channel 15/0 : 25[1] -> 57[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32711:32884 [7] NCCL INFO Channel 07/0 : 31[7] -> 63[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124165:124341 [5] NCCL INFO Channel 14/0 : 5[5] -> 69[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124164:124336 [4] NCCL INFO Channel 14/0 : 4[4] -> 68[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96892:97064 [4] NCCL INFO Channel 12/0 : 20[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32708:32879 [4] NCCL INFO Channel 11/0 : 12[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96891:97063 [3] NCCL INFO Channel 06/0 : 19[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32710:32883 [6] NCCL INFO Channel 09/0 : 14[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32707:32878 [3] NCCL INFO Channel 05/0 : 11[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96889:97065 [1] NCCL INFO Channel 08/0 : 17[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32706:32877 [2] NCCL INFO Channel 07/0 : 10[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96568:96753 [5] NCCL INFO Channel 10/0 : 37[5] -> 69[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96893:97067 [5] NCCL INFO Channel 06/0 : 21[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96564:96757 [1] NCCL INFO Channel 14/0 : 33[1] -> 65[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96570:96759 [7] NCCL INFO Channel 14/0 : 39[7] -> 71[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96888:97066 [0] NCCL INFO Channel 10/0 : 16[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO Channel 09/0 : 10[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96890:97070 [2] NCCL INFO Channel 08/0 : 18[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO Channel 07/0 : 11[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96895:97069 [7] NCCL INFO Channel 08/0 : 23[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO Channel 11/0 : 14[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96894:97068 [6] NCCL INFO Channel 12/0 : 22[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96892:97064 [4] NCCL INFO Channel 14/0 : 20[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO Channel 15/0 : 12[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32704:32881 [0] NCCL INFO Channel 09/0 : 8[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32709:32882 [5] NCCL INFO Channel 15/0 : 29[5] -> 61[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96891:97063 [3] NCCL INFO Channel 08/0 : 19[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO Channel 09/0 : 31[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32711:32884 [7] NCCL INFO Channel 09/0 : 31[7] -> 63[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96889:97065 [1] NCCL INFO Channel 10/0 : 17[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO Channel 11/0 : 8[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32708:32879 [4] NCCL INFO Channel 13/0 : 12[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32707:32878 [3] NCCL INFO Channel 07/0 : 11[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32710:32883 [6] NCCL INFO Channel 11/0 : 14[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96893:97067 [5] NCCL INFO Channel 08/0 : 21[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO Channel 11/0 : 10[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32706:32877 [2] NCCL INFO Channel 09/0 : 10[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96568:96753 [5] NCCL INFO Channel 12/0 : 37[5] -> 69[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96888:97066 [0] NCCL INFO Channel 12/0 : 16[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO Channel 09/0 : 11[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96890:97070 [2] NCCL INFO Channel 10/0 : 18[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO Channel 13/0 : 14[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96895:97069 [7] NCCL INFO Channel 10/0 : 23[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO Channel 11/0 : 31[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96894:97068 [6] NCCL INFO Channel 14/0 : 22[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96891:97063 [3] NCCL INFO Channel 10/0 : 19[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO Channel 13/0 : 8[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96889:97065 [1] NCCL INFO Channel 12/0 : 17[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO Channel 13/0 : 10[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32704:32881 [0] NCCL INFO Channel 11/0 : 8[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96893:97067 [5] NCCL INFO Channel 10/0 : 21[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO Channel 13/0 : 11[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32711:32884 [7] NCCL INFO Channel 11/0 : 31[7] -> 63[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96888:97066 [0] NCCL INFO Channel 14/0 : 16[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO Channel 15/0 : 14[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32710:32883 [6] NCCL INFO Channel 13/0 : 14[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32706:32877 [2] NCCL INFO Channel 11/0 : 10[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96568:96753 [5] NCCL INFO Channel 14/0 : 37[5] -> 69[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96890:97070 [2] NCCL INFO Channel 12/0 : 18[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO Channel 13/0 : 31[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96567:96760 [4] NCCL INFO Channel 00/0 : 36[4] -> 68[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96895:97069 [7] NCCL INFO Channel 12/0 : 23[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96889:97065 [1] NCCL INFO Channel 14/0 : 17[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96891:97063 [3] NCCL INFO Channel 12/0 : 19[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96893:97067 [5] NCCL INFO Channel 12/0 : 21[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96890:97070 [2] NCCL INFO Channel 14/0 : 18[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32707:32878 [3] NCCL INFO Channel 09/0 : 11[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32704:32881 [0] NCCL INFO Channel 13/0 : 8[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32708:32879 [4] NCCL INFO Channel 15/0 : 12[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32705:32880 [1] NCCL INFO Channel 01/0 : 57[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96895:97069 [7] NCCL INFO Channel 14/0 : 23[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32711:32884 [7] NCCL INFO Channel 13/0 : 31[7] -> 63[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96893:97067 [5] NCCL INFO Channel 14/0 : 21[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO Channel 01/0 : 121[1] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO Channel 15/0 : 8[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32709:32882 [5] NCCL INFO Channel 01/0 : 61[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96567:96760 [4] NCCL INFO Channel 02/0 : 36[4] -> 68[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96891:97063 [3] NCCL INFO Channel 14/0 : 19[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO Channel 15/0 : 10[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96570:96759 [7] NCCL INFO Channel 00/0 : 71[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96892:97064 [4] NCCL INFO Channel 00/0 : 20[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO Channel 15/0 : 11[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96892:97064 [4] NCCL INFO Channel 02/0 : 20[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96895:97069 [7] NCCL INFO Channel 00/0 : 39[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96889:97065 [1] NCCL INFO Channel 00/0 : 33[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96894:97068 [6] NCCL INFO Channel 00/0 : 22[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO Channel 01/0 : 125[5] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32706:32877 [2] NCCL INFO Channel 13/0 : 10[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32710:32883 [6] NCCL INFO Channel 15/0 : 14[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114519:114692 [7] NCCL INFO Channel 00/0 : 7[7] -> 71[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96888:97066 [0] NCCL INFO Channel 02/0 : 16[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96892:97064 [4] NCCL INFO Channel 06/0 : 20[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32704:32881 [0] NCCL INFO Channel 15/0 : 8[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96564:96757 [1] NCCL INFO Channel 00/0 : 65[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96893:97067 [5] NCCL INFO Channel 00/0 : 37[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96567:96760 [4] NCCL INFO Channel 06/0 : 36[4] -> 68[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96895:97069 [7] NCCL INFO Channel 02/0 : 39[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96570:96759 [7] NCCL INFO Channel 02/0 : 71[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96889:97065 [1] NCCL INFO Channel 02/0 : 33[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO Channel 03/0 : 121[1] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32705:32880 [1] NCCL INFO Channel 03/0 : 57[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32708:32879 [4] NCCL INFO Channel 01/0 : 28[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114513:114688 [1] NCCL INFO Channel 00/0 : 1[1] -> 65[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32707:32878 [3] NCCL INFO Channel 13/0 : 11[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32709:32882 [5] NCCL INFO Channel 03/0 : 61[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32710:32883 [6] NCCL INFO Channel 01/0 : 30[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32706:32877 [2] NCCL INFO Channel 15/0 : 10[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32704:32881 [0] NCCL INFO Channel 01/0 : 24[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96569:96758 [6] NCCL INFO Channel 00/0 : 38[6] -> 70[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96564:96757 [1] NCCL INFO Channel 02/0 : 65[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96890:97070 [2] NCCL INFO Channel 00/0 : 18[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO Channel 03/0 : 125[5] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96563:96756 [0] NCCL INFO Channel 02/0 : 32[0] -> 64[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96567:96760 [4] NCCL INFO Channel 08/0 : 36[4] -> 68[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96894:97068 [6] NCCL INFO Channel 02/0 : 22[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO Channel 05/0 : 121[1] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96570:96759 [7] NCCL INFO Channel 04/0 : 71[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114513:114688 [1] NCCL INFO Channel 02/0 : 1[1] -> 65[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96888:97066 [0] NCCL INFO Channel 04/0 : 16[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96568:96753 [5] NCCL INFO Channel 00/0 : 69[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114519:114692 [7] NCCL INFO Channel 02/0 : 7[7] -> 71[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96892:97064 [4] NCCL INFO Channel 08/0 : 20[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32705:32880 [1] NCCL INFO Channel 05/0 : 57[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96893:97067 [5] NCCL INFO Channel 02/0 : 37[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO Channel 05/0 : 125[5] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32708:32879 [4] NCCL INFO Channel 03/0 : 28[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96889:97065 [1] NCCL INFO Channel 04/0 : 33[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96895:97069 [7] NCCL INFO Channel 04/0 : 39[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO Channel 01/0 : 127[7] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO Channel 07/0 : 121[1] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32707:32878 [3] NCCL INFO Channel 15/0 : 11[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96891:97063 [3] NCCL INFO Channel 00/0 : 19[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32709:32882 [5] NCCL INFO Channel 05/0 : 61[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96564:96757 [1] NCCL INFO Channel 04/0 : 65[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114517:114687 [5] NCCL INFO Channel 00/0 : 5[5] -> 69[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96890:97070 [2] NCCL INFO Channel 04/0 : 18[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96569:96758 [6] NCCL INFO Channel 02/0 : 38[6] -> 70[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114513:114688 [1] NCCL INFO Channel 04/0 : 1[1] -> 65[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114519:114692 [7] NCCL INFO Channel 04/0 : 7[7] -> 71[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96894:97068 [6] NCCL INFO Channel 04/0 : 22[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96565:96754 [2] NCCL INFO Channel 00/0 : 34[2] -> 66[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96888:97066 [0] NCCL INFO Channel 06/0 : 16[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO Channel 07/0 : 125[5] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96892:97064 [4] NCCL INFO Channel 10/0 : 20[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO Channel 11/0 : 121[1] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32710:32883 [6] NCCL INFO Channel 03/0 : 30[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32706:32877 [2] NCCL INFO Channel 01/0 : 26[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32704:32881 [0] NCCL INFO Channel 03/0 : 24[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32711:32884 [7] NCCL INFO Channel 01/0 : 63[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96893:97067 [5] NCCL INFO Channel 04/0 : 37[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32705:32880 [1] NCCL INFO Channel 07/0 : 57[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96889:97065 [1] NCCL INFO Channel 06/0 : 33[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32708:32879 [4] NCCL INFO Channel 05/0 : 28[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114517:114687 [5] NCCL INFO Channel 02/0 : 5[5] -> 69[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96891:97063 [3] NCCL INFO Channel 02/0 : 19[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96895:97069 [7] NCCL INFO Channel 06/0 : 39[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32707:32878 [3] NCCL INFO Channel 01/0 : 27[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32709:32882 [5] NCCL INFO Channel 07/0 : 61[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114513:114688 [1] NCCL INFO Channel 06/0 : 1[1] -> 65[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114519:114692 [7] NCCL INFO Channel 06/0 : 7[7] -> 71[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62886:63057 [5] NCCL INFO Channel 03/0 : 45[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96890:97070 [2] NCCL INFO Channel 06/0 : 18[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO Channel 03/0 : 127[7] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96894:97068 [6] NCCL INFO Channel 08/0 : 22[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96888:97066 [0] NCCL INFO Channel 08/0 : 16[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO Channel 09/0 : 125[5] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96892:97064 [4] NCCL INFO Channel 12/0 : 20[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96889:97065 [1] NCCL INFO Channel 08/0 : 33[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96893:97067 [5] NCCL INFO Channel 06/0 : 37[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32704:32881 [0] NCCL INFO Channel 05/0 : 24[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32706:32877 [2] NCCL INFO Channel 03/0 : 26[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96891:97063 [3] NCCL INFO Channel 04/0 : 19[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32711:32884 [7] NCCL INFO Channel 03/0 : 63[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32705:32880 [1] NCCL INFO Channel 11/0 : 57[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32710:32883 [6] NCCL INFO Channel 05/0 : 30[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114517:114687 [5] NCCL INFO Channel 04/0 : 5[5] -> 69[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62886:63057 [5] NCCL INFO Channel 05/0 : 45[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96890:97070 [2] NCCL INFO Channel 08/0 : 18[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32708:32879 [4] NCCL INFO Channel 07/0 : 28[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114513:114688 [1] NCCL INFO Channel 08/0 : 1[1] -> 65[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114519:114692 [7] NCCL INFO Channel 08/0 : 7[7] -> 71[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96895:97069 [7] NCCL INFO Channel 08/0 : 39[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96888:97066 [0] NCCL INFO Channel 10/0 : 16[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO Channel 13/0 : 121[1] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96894:97068 [6] NCCL INFO Channel 10/0 : 22[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO Channel 05/0 : 127[7] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96892:97064 [4] NCCL INFO Channel 14/0 : 20[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO Channel 11/0 : 125[5] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96889:97065 [1] NCCL INFO Channel 10/0 : 33[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO Channel 15/0 : 121[1] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32707:32878 [3] NCCL INFO Channel 03/0 : 27[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32709:32882 [5] NCCL INFO Channel 09/0 : 61[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62886:63057 [5] NCCL INFO Channel 07/0 : 45[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96893:97067 [5] NCCL INFO Channel 08/0 : 37[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32704:32881 [0] NCCL INFO Channel 07/0 : 24[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32706:32877 [2] NCCL INFO Channel 05/0 : 26[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114517:114687 [5] NCCL INFO Channel 06/0 : 5[5] -> 69[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96891:97063 [3] NCCL INFO Channel 06/0 : 19[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96890:97070 [2] NCCL INFO Channel 10/0 : 18[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96888:97066 [0] NCCL INFO Channel 12/0 : 16[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO Channel 07/0 : 127[7] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96895:97069 [7] NCCL INFO Channel 10/0 : 39[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO Channel 01/0 : 57[1] -> 121[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO Channel 15/0 : 125[5] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32705:32880 [1] NCCL INFO Channel 13/0 : 57[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62886:63057 [5] NCCL INFO Channel 09/0 : 45[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96894:97068 [6] NCCL INFO Channel 12/0 : 22[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32708:32879 [4] NCCL INFO Channel 09/0 : 28[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32710:32883 [6] NCCL INFO Channel 07/0 : 30[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32711:32884 [7] NCCL INFO Channel 05/0 : 63[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114513:114688 [1] NCCL INFO Channel 10/0 : 1[1] -> 65[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96889:97065 [1] NCCL INFO Channel 12/0 : 33[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114517:114687 [5] NCCL INFO Channel 08/0 : 5[5] -> 69[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96893:97067 [5] NCCL INFO Channel 10/0 : 37[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114519:114692 [7] NCCL INFO Channel 10/0 : 7[7] -> 71[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96891:97063 [3] NCCL INFO Channel 08/0 : 19[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96890:97070 [2] NCCL INFO Channel 12/0 : 18[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO Channel 09/0 : 127[7] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62882:63056 [1] NCCL INFO Channel 03/0 : 41[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96888:97066 [0] NCCL INFO Channel 14/0 : 16[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO Channel 03/0 : 57[1] -> 121[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32707:32878 [3] NCCL INFO Channel 05/0 : 27[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32709:32882 [5] NCCL INFO Channel 11/0 : 61[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96895:97069 [7] NCCL INFO Channel 12/0 : 39[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32704:32881 [0] NCCL INFO Channel 09/0 : 24[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32706:32877 [2] NCCL INFO Channel 07/0 : 26[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96894:97068 [6] NCCL INFO Channel 14/0 : 22[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32705:32880 [1] NCCL INFO Channel 15/0 : 57[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96889:97065 [1] NCCL INFO Channel 14/0 : 33[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114513:114688 [1] NCCL INFO Channel 12/0 : 1[1] -> 65[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96890:97070 [2] NCCL INFO Channel 14/0 : 18[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114517:114687 [5] NCCL INFO Channel 10/0 : 5[5] -> 69[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62886:63057 [5] NCCL INFO Channel 11/0 : 45[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96893:97067 [5] NCCL INFO Channel 12/0 : 37[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO Channel 01/0 : 61[5] -> 125[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114519:114692 [7] NCCL INFO Channel 12/0 : 7[7] -> 71[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62882:63056 [1] NCCL INFO Channel 05/0 : 41[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96892:97064 [4] NCCL INFO Channel 00/0 : 36[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO Channel 11/0 : 127[7] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96891:97063 [3] NCCL INFO Channel 10/0 : 19[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO Channel 05/0 : 57[1] -> 121[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96895:97069 [7] NCCL INFO Channel 14/0 : 39[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32710:32883 [6] NCCL INFO Channel 09/0 : 30[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96893:97067 [5] NCCL INFO Channel 14/0 : 37[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96892:97064 [4] NCCL INFO Channel 02/0 : 36[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114513:114688 [1] NCCL INFO Channel 14/0 : 1[1] -> 65[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114517:114687 [5] NCCL INFO Channel 12/0 : 5[5] -> 69[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96891:97063 [3] NCCL INFO Channel 12/0 : 19[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114519:114692 [7] NCCL INFO Channel 14/0 : 7[7] -> 71[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62886:63057 [5] NCCL INFO Channel 15/0 : 45[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96892:97064 [4] NCCL INFO Channel 06/0 : 36[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62882:63056 [1] NCCL INFO Channel 07/0 : 41[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96888:97066 [0] NCCL INFO Channel 02/0 : 32[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO Channel 03/0 : 61[5] -> 125[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32711:32884 [7] NCCL INFO Channel 07/0 : 63[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96894:97068 [6] NCCL INFO Channel 00/0 : 38[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO Channel 13/0 : 127[7] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32707:32878 [3] NCCL INFO Channel 07/0 : 27[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96891:97063 [3] NCCL INFO Channel 14/0 : 19[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO Channel 07/0 : 57[1] -> 121[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32704:32881 [0] NCCL INFO Channel 11/0 : 24[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62882:63056 [1] NCCL INFO Channel 11/0 : 41[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96892:97064 [4] NCCL INFO Channel 08/0 : 36[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32709:32882 [5] NCCL INFO Channel 15/0 : 61[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32706:32877 [2] NCCL INFO Channel 09/0 : 26[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114513:114688 [1] NCCL INFO Channel 00/0 : 65[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114517:114687 [5] NCCL INFO Channel 14/0 : 5[5] -> 69[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114519:114692 [7] NCCL INFO Channel 00/0 : 71[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62882:63056 [1] NCCL INFO Channel 13/0 : 41[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96890:97070 [2] NCCL INFO Channel 00/0 : 34[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96894:97068 [6] NCCL INFO Channel 02/0 : 38[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96888:97066 [0] NCCL INFO Channel 04/0 : 32[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO Channel 05/0 : 61[5] -> 125[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96892:97064 [4] NCCL INFO Channel 10/0 : 36[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO Channel 01/0 : 63[7] -> 127[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32710:32883 [6] NCCL INFO Channel 11/0 : 30[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96890:97070 [2] NCCL INFO Channel 04/0 : 34[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO Channel 11/0 : 57[1] -> 121[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32711:32884 [7] NCCL INFO Channel 09/0 : 63[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO Channel 07/0 : 61[5] -> 125[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32704:32881 [0] NCCL INFO Channel 13/0 : 24[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32707:32878 [3] NCCL INFO Channel 09/0 : 27[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32706:32877 [2] NCCL INFO Channel 11/0 : 26[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO Channel 03/0 : 63[7] -> 127[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32710:32883 [6] NCCL INFO Channel 13/0 : 30[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO Channel 13/0 : 57[1] -> 121[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32711:32884 [7] NCCL INFO Channel 11/0 : 63[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114513:114688 [1] NCCL INFO Channel 02/0 : 65[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114517:114687 [5] NCCL INFO Channel 00/0 : 69[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62882:63056 [1] NCCL INFO Channel 15/0 : 41[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96888:97066 [0] NCCL INFO Channel 06/0 : 32[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96894:97068 [6] NCCL INFO Channel 04/0 : 38[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96892:97064 [4] NCCL INFO Channel 12/0 : 36[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96890:97070 [2] NCCL INFO Channel 06/0 : 34[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96891:97063 [3] NCCL INFO Channel 00/0 : 35[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO Channel 09/0 : 61[5] -> 125[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32704:32881 [0] NCCL INFO Channel 15/0 : 24[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96888:97066 [0] NCCL INFO Channel 08/0 : 32[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO Channel 05/0 : 63[7] -> 127[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32707:32878 [3] NCCL INFO Channel 13/0 : 27[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32706:32877 [2] NCCL INFO Channel 13/0 : 26[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114519:114692 [7] NCCL INFO Channel 02/0 : 71[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96892:97064 [4] NCCL INFO Channel 14/0 : 36[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96894:97068 [6] NCCL INFO Channel 08/0 : 38[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO Channel 15/0 : 57[1] -> 121[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32710:32883 [6] NCCL INFO Channel 15/0 : 30[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96890:97070 [2] NCCL INFO Channel 08/0 : 34[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96891:97063 [3] NCCL INFO Channel 02/0 : 35[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO Channel 11/0 : 61[5] -> 125[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32711:32884 [7] NCCL INFO Channel 13/0 : 63[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96888:97066 [0] NCCL INFO Channel 10/0 : 32[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96894:97068 [6] NCCL INFO Channel 10/0 : 38[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96891:97063 [3] NCCL INFO Channel 04/0 : 35[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96890:97070 [2] NCCL INFO Channel 10/0 : 34[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96888:97066 [0] NCCL INFO Channel 12/0 : 32[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114513:114688 [1] NCCL INFO Channel 04/0 : 65[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96894:97068 [6] NCCL INFO Channel 12/0 : 38[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96891:97063 [3] NCCL INFO Channel 06/0 : 35[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114517:114687 [5] NCCL INFO Channel 02/0 : 69[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114519:114692 [7] NCCL INFO Channel 04/0 : 71[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96890:97070 [2] NCCL INFO Channel 12/0 : 34[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96888:97066 [0] NCCL INFO Channel 14/0 : 32[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO Channel 07/0 : 63[7] -> 127[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32707:32878 [3] NCCL INFO Channel 15/0 : 27[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32706:32877 [2] NCCL INFO Channel 15/0 : 26[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96894:97068 [6] NCCL INFO Channel 14/0 : 38[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO Channel 15/0 : 61[5] -> 125[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96891:97063 [3] NCCL INFO Channel 08/0 : 35[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96890:97070 [2] NCCL INFO Channel 14/0 : 34[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO Channel 09/0 : 63[7] -> 127[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96891:97063 [3] NCCL INFO Channel 10/0 : 35[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96564:96757 [1] NCCL INFO Channel 06/0 : 65[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114513:114688 [1] NCCL INFO Channel 06/0 : 65[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96891:97063 [3] NCCL INFO Channel 12/0 : 35[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114517:114687 [5] NCCL INFO Channel 04/0 : 69[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96891:97063 [3] NCCL INFO Channel 14/0 : 35[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114519:114692 [7] NCCL INFO Channel 06/0 : 71[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO Channel 11/0 : 63[7] -> 127[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO Channel 01/0 : 24[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96564:96757 [1] NCCL INFO Channel 08/0 : 65[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96567:96760 [4] NCCL INFO Channel 10/0 : 36[4] -> 68[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114513:114688 [1] NCCL INFO Channel 08/0 : 65[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114517:114687 [5] NCCL INFO Channel 06/0 : 69[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114519:114692 [7] NCCL INFO Channel 08/0 : 71[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62888:63060 [7] NCCL INFO Channel 01/0 : 47[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO Channel 13/0 : 63[7] -> 127[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO Channel 01/0 : 30[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO Channel 01/0 : 89[1] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96570:96759 [7] NCCL INFO Channel 06/0 : 71[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722046:2722223 [1] NCCL INFO Channel 00/0 : 121[1] -> 113[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114513:114688 [1] NCCL INFO Channel 10/0 : 65[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62888:63060 [7] NCCL INFO Channel 03/0 : 47[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO Channel 03/0 : 24[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114517:114687 [5] NCCL INFO Channel 08/0 : 69[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO Channel 01/0 : 93[5] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO Channel 01/0 : 27[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96565:96754 [2] NCCL INFO Channel 04/0 : 34[2] -> 66[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114519:114692 [7] NCCL INFO Channel 10/0 : 71[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO Channel 03/0 : 89[1] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO Channel 01/0 : 26[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96564:96757 [1] NCCL INFO Channel 10/0 : 65[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96567:96760 [4] NCCL INFO Channel 12/0 : 36[4] -> 68[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722046:2722223 [1] NCCL INFO Channel 02/0 : 121[1] -> 113[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96570:96759 [7] NCCL INFO Channel 08/0 : 71[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62888:63060 [7] NCCL INFO Channel 05/0 : 47[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO Channel 03/0 : 93[5] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32706:32877 [2] NCCL INFO Channel 01/0 : 26[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114513:114688 [1] NCCL INFO Channel 12/0 : 65[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO Channel 05/0 : 89[1] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO Channel 03/0 : 30[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114517:114687 [5] NCCL INFO Channel 10/0 : 69[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO Channel 05/0 : 24[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-52:2722050:2722225 [5] NCCL INFO Channel 00/0 : 125[5] -> 117[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96565:96754 [2] NCCL INFO Channel 06/0 : 34[2] -> 66[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114519:114692 [7] NCCL INFO Channel 12/0 : 71[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO Channel 03/0 : 27[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96567:96760 [4] NCCL INFO Channel 14/0 : 36[4] -> 68[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114513:114688 [1] NCCL INFO Channel 14/0 : 65[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO Channel 03/0 : 26[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96564:96757 [1] NCCL INFO Channel 12/0 : 65[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO Channel 01/0 : 95[7] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO Channel 05/0 : 30[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32706:32877 [2] NCCL INFO Channel 03/0 : 26[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96570:96759 [7] NCCL INFO Channel 10/0 : 71[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62888:63060 [7] NCCL INFO Channel 07/0 : 47[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO Channel 05/0 : 93[5] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96565:96754 [2] NCCL INFO Channel 08/0 : 34[2] -> 66[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO Channel 07/0 : 89[1] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-52:2722046:2722223 [1] NCCL INFO Channel 04/0 : 121[1] -> 113[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114517:114687 [5] NCCL INFO Channel 12/0 : 69[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114519:114692 [7] NCCL INFO Channel 14/0 : 71[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114517:114687 [5] NCCL INFO Channel 14/0 : 69[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO Channel 07/0 : 24[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32706:32877 [2] NCCL INFO Channel 05/0 : 26[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO Channel 05/0 : 27[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96564:96757 [1] NCCL INFO Channel 14/0 : 65[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62883:63058 [2] NCCL INFO Channel 01/0 : 42[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO Channel 05/0 : 26[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO Channel 07/0 : 30[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96570:96759 [7] NCCL INFO Channel 12/0 : 71[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO Channel 03/0 : 95[7] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO Channel 09/0 : 24[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-52:2722046:2722223 [1] NCCL INFO Channel 06/0 : 121[1] -> 113[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96565:96754 [2] NCCL INFO Channel 10/0 : 34[2] -> 66[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO Channel 07/0 : 93[5] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO Channel 07/0 : 27[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-52:2722050:2722225 [5] NCCL INFO Channel 02/0 : 125[5] -> 117[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96563:96756 [0] NCCL INFO Channel 04/0 : 32[0] -> 64[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO Channel 11/0 : 89[1] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-52:2722052:2722226 [7] NCCL INFO Channel 00/0 : 127[7] -> 119[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62888:63060 [7] NCCL INFO Channel 09/0 : 47[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62883:63058 [2] NCCL INFO Channel 03/0 : 42[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32706:32877 [2] NCCL INFO Channel 07/0 : 26[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO Channel 07/0 : 26[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO Channel 09/0 : 30[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114519:114692 [7] NCCL INFO Channel 00/0 : 103[7] -> 71[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO Channel 05/0 : 95[7] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO Channel 11/0 : 24[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO Channel 09/0 : 27[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96570:96759 [7] NCCL INFO Channel 14/0 : 71[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62888:63060 [7] NCCL INFO Channel 11/0 : 47[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO Channel 09/0 : 93[5] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32708:32879 [4] NCCL INFO Channel 11/0 : 28[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:124167:124338 [7] NCCL INFO Channel 01/0 : 15[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722046:2722223 [1] NCCL INFO Channel 08/0 : 121[1] -> 113[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96565:96754 [2] NCCL INFO Channel 12/0 : 34[2] -> 66[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62883:63058 [2] NCCL INFO Channel 05/0 : 42[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO Channel 13/0 : 89[1] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32706:32877 [2] NCCL INFO Channel 09/0 : 26[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-52:2722050:2722225 [5] NCCL INFO Channel 04/0 : 125[5] -> 117[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96563:96756 [0] NCCL INFO Channel 06/0 : 32[0] -> 64[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114513:114688 [1] NCCL INFO Channel 00/0 : 97[1] -> 65[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO Channel 09/0 : 26[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-52:2722052:2722226 [7] NCCL INFO Channel 02/0 : 127[7] -> 119[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO Channel 11/0 : 30[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO Channel 13/0 : 27[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO Channel 13/0 : 24[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62888:63060 [7] NCCL INFO Channel 13/0 : 47[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO Channel 07/0 : 95[7] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32708:32879 [4] NCCL INFO Channel 13/0 : 28[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:124167:124338 [7] NCCL INFO Channel 03/0 : 15[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62883:63058 [2] NCCL INFO Channel 07/0 : 42[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO Channel 11/0 : 93[5] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32706:32877 [2] NCCL INFO Channel 11/0 : 26[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:124165:124341 [5] NCCL INFO Channel 01/0 : 13[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96567:96760 [4] NCCL INFO Channel 00/0 : 68[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114516:114691 [4] NCCL INFO Channel 00/0 : 4[4] -> 68[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114517:114687 [5] NCCL INFO Channel 00/0 : 101[5] -> 69[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO Channel 15/0 : 89[1] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-52:2722046:2722223 [1] NCCL INFO Channel 10/0 : 121[1] -> 113[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96565:96754 [2] NCCL INFO Channel 14/0 : 34[2] -> 66[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114519:114692 [7] NCCL INFO Channel 02/0 : 103[7] -> 71[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114513:114688 [1] NCCL INFO Channel 02/0 : 97[1] -> 65[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO Channel 09/0 : 95[7] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO Channel 11/0 : 26[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-52:2722050:2722225 [5] NCCL INFO Channel 06/0 : 125[5] -> 117[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96563:96756 [0] NCCL INFO Channel 08/0 : 32[0] -> 64[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO Channel 15/0 : 93[5] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO Channel 13/0 : 30[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-52:2722052:2722226 [7] NCCL INFO Channel 04/0 : 127[7] -> 119[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96567:96760 [4] NCCL INFO Channel 02/0 : 68[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62883:63058 [2] NCCL INFO Channel 09/0 : 42[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO Channel 01/0 : 57[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO Channel 15/0 : 27[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32707:32878 [3] NCCL INFO Channel 01/0 : 27[3] -> 59[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32708:32879 [4] NCCL INFO Channel 15/0 : 28[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:124167:124338 [7] NCCL INFO Channel 05/0 : 15[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722046:2722223 [1] NCCL INFO Channel 12/0 : 121[1] -> 113[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96563:96756 [0] NCCL INFO Channel 10/0 : 32[0] -> 64[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62883:63058 [2] NCCL INFO Channel 11/0 : 42[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO Channel 15/0 : 24[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32706:32877 [2] NCCL INFO Channel 13/0 : 26[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:124165:124341 [5] NCCL INFO Channel 03/0 : 13[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-52:2722050:2722225 [5] NCCL INFO Channel 08/0 : 125[5] -> 117[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114517:114687 [5] NCCL INFO Channel 02/0 : 101[5] -> 69[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO Channel 11/0 : 95[7] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114513:114688 [1] NCCL INFO Channel 04/0 : 97[1] -> 65[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO Channel 01/0 : 61[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO Channel 03/0 : 57[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114516:114691 [4] NCCL INFO Channel 02/0 : 4[4] -> 68[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114519:114692 [7] NCCL INFO Channel 04/0 : 103[7] -> 71[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO Channel 13/0 : 95[7] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96568:96753 [5] NCCL INFO Channel 02/0 : 69[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62884:63061 [3] NCCL INFO Channel 01/0 : 43[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO Channel 13/0 : 26[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO Channel 15/0 : 30[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32707:32878 [3] NCCL INFO Channel 03/0 : 27[3] -> 59[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:124167:124338 [7] NCCL INFO Channel 07/0 : 15[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124165:124341 [5] NCCL INFO Channel 05/0 : 13[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-52:2722052:2722226 [7] NCCL INFO Channel 06/0 : 127[7] -> 119[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96567:96760 [4] NCCL INFO Channel 06/0 : 68[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62883:63058 [2] NCCL INFO Channel 13/0 : 42[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO Channel 15/0 : 26[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32706:32877 [2] NCCL INFO Channel 15/0 : 26[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-52:2722046:2722223 [1] NCCL INFO Channel 14/0 : 121[1] -> 113[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96563:96756 [0] NCCL INFO Channel 12/0 : 32[0] -> 64[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO Channel 03/0 : 61[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO Channel 05/0 : 57[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32707:32878 [3] NCCL INFO Channel 05/0 : 27[3] -> 59[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-52:2722050:2722225 [5] NCCL INFO Channel 10/0 : 125[5] -> 117[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96568:96753 [5] NCCL INFO Channel 04/0 : 69[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114517:114687 [5] NCCL INFO Channel 04/0 : 101[5] -> 69[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO Channel 01/0 : 63[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722052:2722226 [7] NCCL INFO Channel 08/0 : 127[7] -> 119[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96567:96760 [4] NCCL INFO Channel 08/0 : 68[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114513:114688 [1] NCCL INFO Channel 06/0 : 97[1] -> 65[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124167:124338 [7] NCCL INFO Channel 09/0 : 15[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722050:2722225 [5] NCCL INFO Channel 12/0 : 125[5] -> 117[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96563:96756 [0] NCCL INFO Channel 14/0 : 32[0] -> 64[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114516:114691 [4] NCCL INFO Channel 06/0 : 4[4] -> 68[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114519:114692 [7] NCCL INFO Channel 06/0 : 103[7] -> 71[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114514:114689 [2] NCCL INFO Channel 00/0 : 2[2] -> 66[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62884:63061 [3] NCCL INFO Channel 03/0 : 43[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32707:32878 [3] NCCL INFO Channel 07/0 : 27[3] -> 59[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:124165:124341 [5] NCCL INFO Channel 07/0 : 13[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96565:96754 [2] NCCL INFO Channel 00/0 : 66[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62883:63058 [2] NCCL INFO Channel 15/0 : 42[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO Channel 05/0 : 61[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO Channel 07/0 : 57[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO Channel 01/0 : 28[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32707:32878 [3] NCCL INFO Channel 09/0 : 27[3] -> 59[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62884:63061 [3] NCCL INFO Channel 05/0 : 43[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO Channel 03/0 : 63[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124167:124338 [7] NCCL INFO Channel 11/0 : 15[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114517:114687 [5] NCCL INFO Channel 06/0 : 101[5] -> 69[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124165:124341 [5] NCCL INFO Channel 09/0 : 13[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96568:96753 [5] NCCL INFO Channel 06/0 : 69[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114513:114688 [1] NCCL INFO Channel 08/0 : 97[1] -> 65[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32706:32877 [2] NCCL INFO Channel 01/0 : 58[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96567:96760 [4] NCCL INFO Channel 10/0 : 68[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114519:114692 [7] NCCL INFO Channel 08/0 : 103[7] -> 71[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62884:63061 [3] NCCL INFO Channel 07/0 : 43[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO Channel 07/0 : 61[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO Channel 03/0 : 28[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96565:96754 [2] NCCL INFO Channel 04/0 : 66[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114516:114691 [4] NCCL INFO Channel 08/0 : 4[4] -> 68[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO Channel 11/0 : 57[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO Channel 05/0 : 63[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96568:96753 [5] NCCL INFO Channel 08/0 : 69[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114514:114689 [2] NCCL INFO Channel 04/0 : 2[2] -> 66[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO Channel 01/0 : 122[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO Channel 09/0 : 61[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32707:32878 [3] NCCL INFO Channel 13/0 : 27[3] -> 59[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96567:96760 [4] NCCL INFO Channel 12/0 : 68[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32706:32877 [2] NCCL INFO Channel 03/0 : 58[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32708:32879 [4] NCCL INFO Channel 01/0 : 28[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:124167:124338 [7] NCCL INFO Channel 13/0 : 15[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96565:96754 [2] NCCL INFO Channel 06/0 : 66[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62884:63061 [3] NCCL INFO Channel 09/0 : 43[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO Channel 05/0 : 28[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:124165:124341 [5] NCCL INFO Channel 11/0 : 13[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96568:96753 [5] NCCL INFO Channel 10/0 : 69[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO Channel 13/0 : 57[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO Channel 07/0 : 63[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32710:32883 [6] NCCL INFO Channel 01/0 : 30[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32707:32878 [3] NCCL INFO Channel 15/0 : 27[3] -> 59[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96567:96760 [4] NCCL INFO Channel 14/0 : 68[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114517:114687 [5] NCCL INFO Channel 08/0 : 101[5] -> 69[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114513:114688 [1] NCCL INFO Channel 10/0 : 97[1] -> 65[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO Channel 11/0 : 61[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114519:114692 [7] NCCL INFO Channel 10/0 : 103[7] -> 71[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62887:63055 [6] NCCL INFO Channel 01/0 : 46[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO Channel 03/0 : 122[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO Channel 07/0 : 28[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:124165:124341 [5] NCCL INFO Channel 15/0 : 13[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114516:114691 [4] NCCL INFO Channel 10/0 : 4[4] -> 68[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62884:63061 [3] NCCL INFO Channel 13/0 : 43[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32706:32877 [2] NCCL INFO Channel 05/0 : 58[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114514:114689 [2] NCCL INFO Channel 06/0 : 2[2] -> 66[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32708:32879 [4] NCCL INFO Channel 03/0 : 28[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96565:96754 [2] NCCL INFO Channel 08/0 : 66[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO Channel 15/0 : 57[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO Channel 09/0 : 63[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32710:32883 [6] NCCL INFO Channel 03/0 : 30[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96563:96756 [0] NCCL INFO Channel 02/0 : 64[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO Channel 15/0 : 61[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96568:96753 [5] NCCL INFO Channel 12/0 : 69[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO Channel 05/0 : 122[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96565:96754 [2] NCCL INFO Channel 10/0 : 66[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62887:63055 [6] NCCL INFO Channel 03/0 : 46[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62884:63061 [3] NCCL INFO Channel 15/0 : 43[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96563:96756 [0] NCCL INFO Channel 04/0 : 64[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114513:114688 [1] NCCL INFO Channel 12/0 : 97[1] -> 65[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114517:114687 [5] NCCL INFO Channel 10/0 : 101[5] -> 69[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62885:63054 [4] NCCL INFO Channel 01/0 : 44[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96568:96753 [5] NCCL INFO Channel 14/0 : 69[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114512:114686 [0] NCCL INFO Channel 02/0 : 0[0] -> 64[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114519:114692 [7] NCCL INFO Channel 12/0 : 103[7] -> 71[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO Channel 09/0 : 28[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32706:32877 [2] NCCL INFO Channel 07/0 : 58[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96565:96754 [2] NCCL INFO Channel 12/0 : 66[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114516:114691 [4] NCCL INFO Channel 12/0 : 4[4] -> 68[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32708:32879 [4] NCCL INFO Channel 05/0 : 28[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32710:32883 [6] NCCL INFO Channel 05/0 : 30[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96563:96756 [0] NCCL INFO Channel 06/0 : 64[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO Channel 11/0 : 63[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32706:32877 [2] NCCL INFO Channel 09/0 : 58[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO Channel 07/0 : 122[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62887:63055 [6] NCCL INFO Channel 05/0 : 46[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO Channel 11/0 : 28[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114513:114688 [1] NCCL INFO Channel 14/0 : 97[1] -> 65[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114517:114687 [5] NCCL INFO Channel 12/0 : 101[5] -> 69[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114514:114689 [2] NCCL INFO Channel 08/0 : 2[2] -> 66[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32707:32878 [3] NCCL INFO Channel 01/0 : 59[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114512:114686 [0] NCCL INFO Channel 04/0 : 0[0] -> 64[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114519:114692 [7] NCCL INFO Channel 14/0 : 103[7] -> 71[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO Channel 13/0 : 63[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32708:32879 [4] NCCL INFO Channel 07/0 : 28[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114516:114691 [4] NCCL INFO Channel 14/0 : 4[4] -> 68[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62885:63054 [4] NCCL INFO Channel 03/0 : 44[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO Channel 09/0 : 122[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32710:32883 [6] NCCL INFO Channel 07/0 : 30[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114517:114687 [5] NCCL INFO Channel 14/0 : 101[5] -> 69[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO Channel 13/0 : 28[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62887:63055 [6] NCCL INFO Channel 07/0 : 46[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62885:63054 [4] NCCL INFO Channel 05/0 : 44[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32706:32877 [2] NCCL INFO Channel 11/0 : 58[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO Channel 01/0 : 123[3] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32707:32878 [3] NCCL INFO Channel 03/0 : 59[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32708:32879 [4] NCCL INFO Channel 09/0 : 28[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114514:114689 [2] NCCL INFO Channel 10/0 : 2[2] -> 66[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO Channel 11/0 : 122[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO Channel 15/0 : 28[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32710:32883 [6] NCCL INFO Channel 09/0 : 30[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114512:114686 [0] NCCL INFO Channel 06/0 : 0[0] -> 64[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114519:114692 [7] NCCL INFO Channel 00/0 : 71[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114513:114688 [1] NCCL INFO Channel 00/0 : 65[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32706:32877 [2] NCCL INFO Channel 13/0 : 58[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114516:114691 [4] NCCL INFO Channel 00/0 : 68[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114517:114687 [5] NCCL INFO Channel 00/0 : 69[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62885:63054 [4] NCCL INFO Channel 07/0 : 44[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO Channel 03/0 : 123[3] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153563:153740 [1] NCCL INFO Channel 01/0 : 105[1] -> 89[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153567:153742 [5] NCCL INFO Channel 01/0 : 109[5] -> 93[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO Channel 13/0 : 122[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32707:32878 [3] NCCL INFO Channel 05/0 : 59[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62881:63059 [0] NCCL INFO Channel 01/0 : 40[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114514:114689 [2] NCCL INFO Channel 12/0 : 2[2] -> 66[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO Channel 05/0 : 123[3] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32710:32883 [6] NCCL INFO Channel 11/0 : 30[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32708:32879 [4] NCCL INFO Channel 11/0 : 28[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32706:32877 [2] NCCL INFO Channel 15/0 : 58[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32704:32881 [0] NCCL INFO Channel 01/0 : 24[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62885:63054 [4] NCCL INFO Channel 09/0 : 44[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO Channel 00/0 : 57[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32705:32880 [1] NCCL INFO Channel 01/0 : 41[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153563:153740 [1] NCCL INFO Channel 03/0 : 105[1] -> 89[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO Channel 15/0 : 122[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32707:32878 [3] NCCL INFO Channel 07/0 : 59[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32709:32882 [5] NCCL INFO Channel 01/0 : 45[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114512:114686 [0] NCCL INFO Channel 08/0 : 0[0] -> 64[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114513:114688 [1] NCCL INFO Channel 02/0 : 65[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114519:114692 [7] NCCL INFO Channel 02/0 : 71[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114516:114691 [4] NCCL INFO Channel 02/0 : 68[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO Channel 00/0 : 61[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114517:114687 [5] NCCL INFO Channel 02/0 : 69[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114514:114689 [2] NCCL INFO Channel 14/0 : 2[2] -> 66[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62881:63059 [0] NCCL INFO Channel 03/0 : 40[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62885:63054 [4] NCCL INFO Channel 11/0 : 44[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32710:32883 [6] NCCL INFO Channel 13/0 : 30[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32708:32879 [4] NCCL INFO Channel 13/0 : 28[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32704:32881 [0] NCCL INFO Channel 03/0 : 24[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO Channel 07/0 : 123[3] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32705:32880 [1] NCCL INFO Channel 03/0 : 41[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153563:153740 [1] NCCL INFO Channel 05/0 : 105[1] -> 89[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO Channel 02/0 : 57[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32707:32878 [3] NCCL INFO Channel 09/0 : 59[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114513:114688 [1] NCCL INFO Channel 04/0 : 65[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO Channel 01/0 : 58[2] -> 122[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO Channel 02/0 : 61[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32709:32882 [5] NCCL INFO Channel 03/0 : 45[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114519:114692 [7] NCCL INFO Channel 04/0 : 71[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62885:63054 [4] NCCL INFO Channel 13/0 : 44[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62881:63059 [0] NCCL INFO Channel 05/0 : 40[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO Channel 00/0 : 63[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114512:114686 [0] NCCL INFO Channel 10/0 : 0[0] -> 64[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114517:114687 [5] NCCL INFO Channel 04/0 : 69[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114516:114691 [4] NCCL INFO Channel 06/0 : 68[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153563:153740 [1] NCCL INFO Channel 07/0 : 105[1] -> 89[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114514:114689 [2] NCCL INFO Channel 00/0 : 66[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32710:32883 [6] NCCL INFO Channel 15/0 : 30[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32708:32879 [4] NCCL INFO Channel 15/0 : 28[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32704:32881 [0] NCCL INFO Channel 05/0 : 24[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO Channel 04/0 : 57[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO Channel 03/0 : 58[2] -> 122[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO Channel 04/0 : 61[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO Channel 09/0 : 123[3] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO Channel 02/0 : 63[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32705:32880 [1] NCCL INFO Channel 05/0 : 41[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114513:114688 [1] NCCL INFO Channel 06/0 : 65[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32707:32878 [3] NCCL INFO Channel 13/0 : 59[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114519:114692 [7] NCCL INFO Channel 06/0 : 71[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62885:63054 [4] NCCL INFO Channel 15/0 : 44[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62881:63059 [0] NCCL INFO Channel 07/0 : 40[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32709:32882 [5] NCCL INFO Channel 05/0 : 45[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153563:153740 [1] NCCL INFO Channel 11/0 : 105[1] -> 89[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114512:114686 [0] NCCL INFO Channel 12/0 : 0[0] -> 64[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114517:114687 [5] NCCL INFO Channel 06/0 : 69[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32711:32884 [7] NCCL INFO Channel 01/0 : 47[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62881:63059 [0] NCCL INFO Channel 09/0 : 40[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114514:114689 [2] NCCL INFO Channel 04/0 : 66[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114516:114691 [4] NCCL INFO Channel 08/0 : 68[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32704:32881 [0] NCCL INFO Channel 07/0 : 24[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153563:153740 [1] NCCL INFO Channel 13/0 : 105[1] -> 89[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114513:114688 [1] NCCL INFO Channel 08/0 : 65[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32705:32880 [1] NCCL INFO Channel 07/0 : 41[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114519:114692 [7] NCCL INFO Channel 08/0 : 71[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32707:32878 [3] NCCL INFO Channel 15/0 : 59[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32709:32882 [5] NCCL INFO Channel 07/0 : 45[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32704:32881 [0] NCCL INFO Channel 09/0 : 24[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32711:32884 [7] NCCL INFO Channel 03/0 : 47[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32705:32880 [1] NCCL INFO Channel 11/0 : 41[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62881:63059 [0] NCCL INFO Channel 11/0 : 40[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114512:114686 [0] NCCL INFO Channel 14/0 : 0[0] -> 64[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153563:153740 [1] NCCL INFO Channel 15/0 : 105[1] -> 89[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114517:114687 [5] NCCL INFO Channel 08/0 : 69[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114514:114689 [2] NCCL INFO Channel 06/0 : 66[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114513:114688 [1] NCCL INFO Channel 10/0 : 65[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32709:32882 [5] NCCL INFO Channel 09/0 : 45[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114516:114691 [4] NCCL INFO Channel 10/0 : 68[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114519:114692 [7] NCCL INFO Channel 10/0 : 71[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62881:63059 [0] NCCL INFO Channel 13/0 : 40[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32711:32884 [7] NCCL INFO Channel 05/0 : 47[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32704:32881 [0] NCCL INFO Channel 11/0 : 24[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153563:153740 [1] NCCL INFO Channel 01/0 : 89[1] -> 73[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32705:32880 [1] NCCL INFO Channel 13/0 : 41[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32708:32879 [4] NCCL INFO Channel 01/0 : 60[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32710:32883 [6] NCCL INFO Channel 01/0 : 62[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114512:114686 [0] NCCL INFO Channel 02/0 : 64[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114517:114687 [5] NCCL INFO Channel 10/0 : 69[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62881:63059 [0] NCCL INFO Channel 15/0 : 40[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114513:114688 [1] NCCL INFO Channel 12/0 : 65[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114514:114689 [2] NCCL INFO Channel 08/0 : 66[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114516:114691 [4] NCCL INFO Channel 12/0 : 68[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32709:32882 [5] NCCL INFO Channel 11/0 : 45[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153563:153740 [1] NCCL INFO Channel 03/0 : 89[1] -> 73[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62887:63055 [6] NCCL INFO Channel 09/0 : 46[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114519:114692 [7] NCCL INFO Channel 12/0 : 71[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32704:32881 [0] NCCL INFO Channel 13/0 : 24[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32711:32884 [7] NCCL INFO Channel 07/0 : 47[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114512:114686 [0] NCCL INFO Channel 04/0 : 64[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32705:32880 [1] NCCL INFO Channel 15/0 : 41[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114517:114687 [5] NCCL INFO Channel 12/0 : 69[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32710:32883 [6] NCCL INFO Channel 03/0 : 62[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153563:153740 [1] NCCL INFO Channel 05/0 : 89[1] -> 73[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114515:114690 [3] NCCL INFO Channel 00/0 : 3[3] -> 67[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62887:63055 [6] NCCL INFO Channel 11/0 : 46[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32708:32879 [4] NCCL INFO Channel 03/0 : 60[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114513:114688 [1] NCCL INFO Channel 14/0 : 65[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32709:32882 [5] NCCL INFO Channel 15/0 : 45[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114514:114689 [2] NCCL INFO Channel 10/0 : 66[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114516:114691 [4] NCCL INFO Channel 14/0 : 68[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114519:114692 [7] NCCL INFO Channel 14/0 : 71[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153563:153740 [1] NCCL INFO Channel 07/0 : 89[1] -> 73[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62887:63055 [6] NCCL INFO Channel 13/0 : 46[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32704:32881 [0] NCCL INFO Channel 15/0 : 24[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32705:32880 [1] NCCL INFO Channel 01/0 : 25[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32711:32884 [7] NCCL INFO Channel 09/0 : 47[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32710:32883 [6] NCCL INFO Channel 05/0 : 62[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32708:32879 [4] NCCL INFO Channel 05/0 : 60[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32709:32882 [5] NCCL INFO Channel 01/0 : 29[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114512:114686 [0] NCCL INFO Channel 06/0 : 64[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114517:114687 [5] NCCL INFO Channel 14/0 : 69[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114514:114689 [2] NCCL INFO Channel 12/0 : 66[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153563:153740 [1] NCCL INFO Channel 11/0 : 89[1] -> 73[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114515:114690 [3] NCCL INFO Channel 02/0 : 3[3] -> 67[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62887:63055 [6] NCCL INFO Channel 15/0 : 46[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32705:32880 [1] NCCL INFO Channel 03/0 : 25[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114512:114686 [0] NCCL INFO Channel 08/0 : 64[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114514:114689 [2] NCCL INFO Channel 14/0 : 66[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32711:32884 [7] NCCL INFO Channel 11/0 : 47[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153563:153740 [1] NCCL INFO Channel 13/0 : 89[1] -> 73[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32710:32883 [6] NCCL INFO Channel 07/0 : 62[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32709:32882 [5] NCCL INFO Channel 03/0 : 29[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32708:32879 [4] NCCL INFO Channel 07/0 : 60[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32705:32880 [1] NCCL INFO Channel 05/0 : 25[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114512:114686 [0] NCCL INFO Channel 10/0 : 64[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114518:114693 [6] NCCL INFO Channel 00/0 : 6[6] -> 70[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114515:114690 [3] NCCL INFO Channel 04/0 : 3[3] -> 67[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:124161:124337 [1] NCCL INFO Channel 01/0 : 9[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114516:114691 [4] NCCL INFO Channel 00/0 : 100[4] -> 68[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32704:32881 [0] NCCL INFO Channel 01/0 : 56[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32711:32884 [7] NCCL INFO Channel 13/0 : 47[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153563:153740 [1] NCCL INFO Channel 15/0 : 89[1] -> 73[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO Channel 00/0 : 113[1] -> 97[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124161:124337 [1] NCCL INFO Channel 03/0 : 9[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114518:114693 [6] NCCL INFO Channel 02/0 : 6[6] -> 70[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32710:32883 [6] NCCL INFO Channel 09/0 : 62[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32709:32882 [5] NCCL INFO Channel 05/0 : 29[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114512:114686 [0] NCCL INFO Channel 12/0 : 64[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32708:32879 [4] NCCL INFO Channel 09/0 : 60[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32705:32880 [1] NCCL INFO Channel 07/0 : 25[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114516:114691 [4] NCCL INFO Channel 02/0 : 100[4] -> 68[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114515:114690 [3] NCCL INFO Channel 06/0 : 3[3] -> 67[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO Channel 00/0 : 119[7] -> 103[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32704:32881 [0] NCCL INFO Channel 03/0 : 56[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32711:32884 [7] NCCL INFO Channel 01/0 : 31[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124162:124342 [2] NCCL INFO Channel 01/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32710:32883 [6] NCCL INFO Channel 11/0 : 62[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:124161:124337 [1] NCCL INFO Channel 05/0 : 9[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114514:114689 [2] NCCL INFO Channel 00/0 : 98[2] -> 66[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114513:114688 [1] NCCL INFO Channel 01/0 : 73[1] -> 65[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO Channel 00/0 : 117[5] -> 101[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO Channel 02/0 : 113[1] -> 97[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114518:114693 [6] NCCL INFO Channel 04/0 : 6[6] -> 70[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO Channel 02/0 : 119[7] -> 103[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114512:114686 [0] NCCL INFO Channel 14/0 : 64[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32708:32879 [4] NCCL INFO Channel 11/0 : 60[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32705:32880 [1] NCCL INFO Channel 11/0 : 25[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32709:32882 [5] NCCL INFO Channel 07/0 : 29[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124162:124342 [2] NCCL INFO Channel 03/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32704:32881 [0] NCCL INFO Channel 05/0 : 56[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32711:32884 [7] NCCL INFO Channel 03/0 : 31[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124161:124337 [1] NCCL INFO Channel 07/0 : 9[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32710:32883 [6] NCCL INFO Channel 13/0 : 62[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114516:114691 [4] NCCL INFO Channel 06/0 : 100[4] -> 68[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114515:114690 [3] NCCL INFO Channel 08/0 : 3[3] -> 67[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114519:114692 [7] NCCL INFO Channel 01/0 : 79[7] -> 71[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO Channel 02/0 : 117[5] -> 101[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114514:114689 [2] NCCL INFO Channel 04/0 : 98[2] -> 66[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO Channel 04/0 : 113[1] -> 97[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO Channel 04/0 : 119[7] -> 103[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114517:114687 [5] NCCL INFO Channel 01/0 : 77[5] -> 69[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124162:124342 [2] NCCL INFO Channel 05/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114513:114688 [1] NCCL INFO Channel 03/0 : 73[1] -> 65[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32705:32880 [1] NCCL INFO Channel 13/0 : 25[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114518:114693 [6] NCCL INFO Channel 08/0 : 6[6] -> 70[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32709:32882 [5] NCCL INFO Channel 09/0 : 29[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124161:124337 [1] NCCL INFO Channel 11/0 : 9[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32708:32879 [4] NCCL INFO Channel 13/0 : 60[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO Channel 04/0 : 117[5] -> 101[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235649:235827 [1] NCCL INFO Channel 00/0 : 113[1] -> 105[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124162:124342 [2] NCCL INFO Channel 07/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114516:114691 [4] NCCL INFO Channel 08/0 : 100[4] -> 68[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32704:32881 [0] NCCL INFO Channel 07/0 : 56[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32711:32884 [7] NCCL INFO Channel 05/0 : 31[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153563:153740 [1] NCCL INFO Channel 00/0 : 89[1] -> 81[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114515:114690 [3] NCCL INFO Channel 10/0 : 3[3] -> 67[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO Channel 06/0 : 113[1] -> 97[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO Channel 06/0 : 119[7] -> 103[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32710:32883 [6] NCCL INFO Channel 15/0 : 62[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:124161:124337 [1] NCCL INFO Channel 13/0 : 9[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124160:124335 [0] NCCL INFO Channel 01/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114519:114692 [7] NCCL INFO Channel 03/0 : 79[7] -> 71[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO Channel 06/0 : 117[5] -> 101[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235649:235827 [1] NCCL INFO Channel 02/0 : 113[1] -> 105[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32705:32880 [1] NCCL INFO Channel 15/0 : 25[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114514:114689 [2] NCCL INFO Channel 06/0 : 98[2] -> 66[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32709:32882 [5] NCCL INFO Channel 11/0 : 29[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114513:114688 [1] NCCL INFO Channel 05/0 : 73[1] -> 65[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32708:32879 [4] NCCL INFO Channel 15/0 : 60[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:124162:124342 [2] NCCL INFO Channel 09/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153563:153740 [1] NCCL INFO Channel 02/0 : 89[1] -> 81[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114517:114687 [5] NCCL INFO Channel 03/0 : 77[5] -> 69[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114518:114693 [6] NCCL INFO Channel 10/0 : 6[6] -> 70[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO Channel 08/0 : 113[1] -> 97[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO Channel 08/0 : 119[7] -> 103[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO Channel 08/0 : 117[5] -> 101[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32704:32881 [0] NCCL INFO Channel 09/0 : 56[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32711:32884 [7] NCCL INFO Channel 07/0 : 31[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124161:124337 [1] NCCL INFO Channel 15/0 : 9[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124160:124335 [0] NCCL INFO Channel 03/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235649:235827 [1] NCCL INFO Channel 03/0 : 113[1] -> 105[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32709:32882 [5] NCCL INFO Channel 15/0 : 29[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153567:153742 [5] NCCL INFO Channel 03/0 : 109[5] -> 93[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32704:32881 [0] NCCL INFO Channel 11/0 : 56[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32711:32884 [7] NCCL INFO Channel 09/0 : 31[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124162:124342 [2] NCCL INFO Channel 11/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114516:114691 [4] NCCL INFO Channel 10/0 : 100[4] -> 68[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO Channel 10/0 : 113[1] -> 97[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114515:114690 [3] NCCL INFO Channel 12/0 : 3[3] -> 67[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62882:63056 [1] NCCL INFO Channel 00/0 : 49[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO Channel 10/0 : 117[5] -> 101[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO Channel 10/0 : 119[7] -> 103[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114512:114686 [0] NCCL INFO Channel 02/0 : 96[0] -> 64[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114519:114692 [7] NCCL INFO Channel 05/0 : 79[7] -> 71[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235649:235827 [1] NCCL INFO Channel 04/0 : 113[1] -> 105[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32704:32881 [0] NCCL INFO Channel 13/0 : 56[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32711:32884 [7] NCCL INFO Channel 11/0 : 31[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124160:124335 [0] NCCL INFO Channel 05/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153563:153740 [1] NCCL INFO Channel 04/0 : 89[1] -> 81[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114514:114689 [2] NCCL INFO Channel 08/0 : 98[2] -> 66[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114513:114688 [1] NCCL INFO Channel 07/0 : 73[1] -> 65[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124162:124342 [2] NCCL INFO Channel 13/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153567:153742 [5] NCCL INFO Channel 05/0 : 109[5] -> 93[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114517:114687 [5] NCCL INFO Channel 05/0 : 77[5] -> 69[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153563:153740 [1] NCCL INFO Channel 06/0 : 89[1] -> 81[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114518:114693 [6] NCCL INFO Channel 12/0 : 6[6] -> 70[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62882:63056 [1] NCCL INFO Channel 02/0 : 49[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO Channel 12/0 : 113[1] -> 97[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32711:32884 [7] NCCL INFO Channel 13/0 : 31[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32704:32881 [0] NCCL INFO Channel 15/0 : 56[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114516:114691 [4] NCCL INFO Channel 12/0 : 100[4] -> 68[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO Channel 12/0 : 117[5] -> 101[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO Channel 12/0 : 119[7] -> 103[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235649:235827 [1] NCCL INFO Channel 05/0 : 113[1] -> 105[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124160:124335 [0] NCCL INFO Channel 07/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114515:114690 [3] NCCL INFO Channel 14/0 : 3[3] -> 67[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153567:153742 [5] NCCL INFO Channel 07/0 : 109[5] -> 93[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114512:114686 [0] NCCL INFO Channel 04/0 : 96[0] -> 64[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32705:32880 [1] NCCL INFO Channel 00/0 : 25[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124162:124342 [2] NCCL INFO Channel 15/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153563:153740 [1] NCCL INFO Channel 08/0 : 89[1] -> 81[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114519:114692 [7] NCCL INFO Channel 07/0 : 79[7] -> 71[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62886:63057 [5] NCCL INFO Channel 00/0 : 53[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO Channel 14/0 : 113[1] -> 97[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO Channel 00/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO Channel 14/0 : 117[5] -> 101[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO Channel 14/0 : 119[7] -> 103[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235649:235827 [1] NCCL INFO Channel 06/0 : 113[1] -> 105[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32705:32880 [1] NCCL INFO Channel 02/0 : 25[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124160:124335 [0] NCCL INFO Channel 09/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153563:153740 [1] NCCL INFO Channel 10/0 : 89[1] -> 81[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114514:114689 [2] NCCL INFO Channel 10/0 : 98[2] -> 66[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62882:63056 [1] NCCL INFO Channel 03/0 : 49[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153567:153742 [5] NCCL INFO Channel 09/0 : 109[5] -> 93[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114513:114688 [1] NCCL INFO Channel 11/0 : 73[1] -> 65[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62886:63057 [5] NCCL INFO Channel 01/0 : 53[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO Channel 02/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153569:153739 [7] NCCL INFO Channel 01/0 : 111[7] -> 95[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114517:114687 [5] NCCL INFO Channel 07/0 : 77[5] -> 69[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62882:63056 [1] NCCL INFO Channel 04/0 : 49[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO Channel 00/0 : 97[1] -> 81[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32709:32882 [5] NCCL INFO Channel 00/0 : 29[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114516:114691 [4] NCCL INFO Channel 14/0 : 100[4] -> 68[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO Channel 00/0 : 101[5] -> 85[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235649:235827 [1] NCCL INFO Channel 07/0 : 113[1] -> 105[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32705:32880 [1] NCCL INFO Channel 04/0 : 25[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124160:124335 [0] NCCL INFO Channel 11/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114515:114690 [3] NCCL INFO Channel 00/0 : 67[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO Channel 00/0 : 103[7] -> 87[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO Channel 00/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153563:153740 [1] NCCL INFO Channel 12/0 : 89[1] -> 81[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114512:114686 [0] NCCL INFO Channel 06/0 : 96[0] -> 64[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62886:63057 [5] NCCL INFO Channel 02/0 : 53[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO Channel 06/0 : 57[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO Channel 03/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153567:153742 [5] NCCL INFO Channel 11/0 : 109[5] -> 93[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114518:114693 [6] NCCL INFO Channel 14/0 : 6[6] -> 70[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62882:63056 [1] NCCL INFO Channel 05/0 : 49[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO Channel 05/0 : 58[2] -> 122[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32709:32882 [5] NCCL INFO Channel 02/0 : 29[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153569:153739 [7] NCCL INFO Channel 03/0 : 111[7] -> 95[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114519:114692 [7] NCCL INFO Channel 09/0 : 79[7] -> 71[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235649:235827 [1] NCCL INFO Channel 08/0 : 113[1] -> 105[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO Channel 04/0 : 63[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32705:32880 [1] NCCL INFO Channel 06/0 : 25[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124160:124335 [0] NCCL INFO Channel 13/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114514:114689 [2] NCCL INFO Channel 12/0 : 98[2] -> 66[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO Channel 02/0 : 97[1] -> 81[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO Channel 06/0 : 61[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO Channel 13/0 : 123[3] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO Channel 01/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114513:114688 [1] NCCL INFO Channel 13/0 : 73[1] -> 65[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62886:63057 [5] NCCL INFO Channel 03/0 : 53[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62888:63060 [7] NCCL INFO Channel 00/0 : 55[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO Channel 02/0 : 101[5] -> 85[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO Channel 08/0 : 57[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO Channel 07/0 : 58[2] -> 122[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO Channel 04/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153563:153740 [1] NCCL INFO Channel 14/0 : 89[1] -> 81[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114517:114687 [5] NCCL INFO Channel 09/0 : 77[5] -> 69[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62882:63056 [1] NCCL INFO Channel 06/0 : 49[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO Channel 02/0 : 103[7] -> 87[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO Channel 06/0 : 63[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO Channel 08/0 : 61[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32709:32882 [5] NCCL INFO Channel 04/0 : 29[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153567:153742 [5] NCCL INFO Channel 15/0 : 109[5] -> 93[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114515:114690 [3] NCCL INFO Channel 02/0 : 67[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO Channel 04/0 : 97[1] -> 81[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO Channel 15/0 : 123[3] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32705:32880 [1] NCCL INFO Channel 08/0 : 25[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124164:124336 [4] NCCL INFO Channel 01/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:124160:124335 [0] NCCL INFO Channel 15/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153569:153739 [7] NCCL INFO Channel 05/0 : 111[7] -> 95[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114516:114691 [4] NCCL INFO Channel 00/0 : 68[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235649:235827 [1] NCCL INFO Channel 10/0 : 113[1] -> 105[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO Channel 09/0 : 58[2] -> 122[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO Channel 10/0 : 57[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO Channel 02/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114512:114686 [0] NCCL INFO Channel 08/0 : 96[0] -> 64[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62886:63057 [5] NCCL INFO Channel 04/0 : 53[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62888:63060 [7] NCCL INFO Channel 01/0 : 55[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO Channel 01/0 : 126[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO Channel 05/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO Channel 00/0 : 81[1] -> 73[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114519:114692 [7] NCCL INFO Channel 11/0 : 79[7] -> 71[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62882:63056 [1] NCCL INFO Channel 07/0 : 49[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO Channel 04/0 : 101[5] -> 85[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO Channel 04/0 : 103[7] -> 87[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO Channel 01/0 : 124[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO Channel 08/0 : 63[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32711:32884 [7] NCCL INFO Channel 00/0 : 31[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32709:32882 [5] NCCL INFO Channel 06/0 : 29[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124164:124336 [4] NCCL INFO Channel 03/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO Channel 06/0 : 97[1] -> 81[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235649:235827 [1] NCCL INFO Channel 11/0 : 113[1] -> 105[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO Channel 10/0 : 61[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32705:32880 [1] NCCL INFO Channel 10/0 : 25[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153567:153742 [5] NCCL INFO Channel 01/0 : 93[5] -> 77[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO Channel 06/0 : 101[5] -> 85[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO Channel 06/0 : 103[7] -> 87[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO Channel 01/0 : 59[3] -> 123[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO Channel 11/0 : 58[2] -> 122[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO Channel 00/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153569:153739 [7] NCCL INFO Channel 07/0 : 111[7] -> 95[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62886:63057 [5] NCCL INFO Channel 06/0 : 53[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO Channel 12/0 : 57[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO Channel 03/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114518:114693 [6] NCCL INFO Channel 00/0 : 70[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62888:63060 [7] NCCL INFO Channel 02/0 : 55[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO Channel 03/0 : 126[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO Channel 06/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO Channel 02/0 : 81[1] -> 73[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114514:114689 [2] NCCL INFO Channel 14/0 : 98[2] -> 66[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114513:114688 [1] NCCL INFO Channel 15/0 : 73[1] -> 65[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62882:63056 [1] NCCL INFO Channel 08/0 : 49[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO Channel 10/0 : 63[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32709:32882 [5] NCCL INFO Channel 08/0 : 29[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32711:32884 [7] NCCL INFO Channel 02/0 : 31[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32705:32880 [1] NCCL INFO Channel 12/0 : 25[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124164:124336 [4] NCCL INFO Channel 05/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114517:114687 [5] NCCL INFO Channel 11/0 : 77[5] -> 69[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO Channel 08/0 : 97[1] -> 81[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235649:235827 [1] NCCL INFO Channel 12/0 : 113[1] -> 105[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO Channel 03/0 : 124[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153567:153742 [5] NCCL INFO Channel 03/0 : 93[5] -> 77[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114515:114690 [3] NCCL INFO Channel 04/0 : 67[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO Channel 08/0 : 101[5] -> 85[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO Channel 12/0 : 61[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO Channel 01/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO Channel 03/0 : 81[1] -> 73[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153569:153739 [7] NCCL INFO Channel 09/0 : 111[7] -> 95[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114516:114691 [4] NCCL INFO Channel 02/0 : 68[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62886:63057 [5] NCCL INFO Channel 07/0 : 53[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO Channel 08/0 : 103[7] -> 87[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO Channel 13/0 : 58[2] -> 122[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO Channel 03/0 : 59[3] -> 123[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO Channel 14/0 : 57[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO Channel 07/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO Channel 04/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153567:153742 [5] NCCL INFO Channel 05/0 : 93[5] -> 77[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114512:114686 [0] NCCL INFO Channel 10/0 : 96[0] -> 64[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62888:63060 [7] NCCL INFO Channel 03/0 : 55[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62882:63056 [1] NCCL INFO Channel 10/0 : 49[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO Channel 10/0 : 97[1] -> 81[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO Channel 05/0 : 126[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:124164:124336 [4] NCCL INFO Channel 07/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114519:114692 [7] NCCL INFO Channel 13/0 : 79[7] -> 71[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62886:63057 [5] NCCL INFO Channel 08/0 : 53[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO Channel 12/0 : 63[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO Channel 04/0 : 81[1] -> 73[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96565:96754 [2] NCCL INFO Channel 14/0 : 66[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114518:114693 [6] NCCL INFO Channel 02/0 : 70[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO Channel 14/0 : 61[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO Channel 02/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96563:96756 [0] NCCL INFO Channel 08/0 : 64[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153569:153739 [7] NCCL INFO Channel 11/0 : 111[7] -> 95[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114514:114689 [2] NCCL INFO Channel 00/0 : 66[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO Channel 10/0 : 101[5] -> 85[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO Channel 05/0 : 124[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96566:96755 [3] NCCL INFO Channel 00/0 : 35[3] -> 67[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153567:153742 [5] NCCL INFO Channel 07/0 : 93[5] -> 77[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114517:114687 [5] NCCL INFO Channel 15/0 : 77[5] -> 69[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO Channel 10/0 : 103[7] -> 87[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO Channel 15/0 : 58[2] -> 122[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO Channel 05/0 : 59[3] -> 123[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96563:96756 [0] NCCL INFO Channel 10/0 : 64[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153569:153739 [7] NCCL INFO Channel 13/0 : 111[7] -> 95[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114515:114690 [3] NCCL INFO Channel 06/0 : 67[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62882:63056 [1] NCCL INFO Channel 11/0 : 49[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62888:63060 [7] NCCL INFO Channel 04/0 : 55[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO Channel 12/0 : 97[1] -> 81[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO Channel 14/0 : 63[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO Channel 07/0 : 126[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO Channel 05/0 : 81[1] -> 73[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124164:124336 [4] NCCL INFO Channel 09/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96566:96755 [3] NCCL INFO Channel 02/0 : 35[3] -> 67[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114512:114686 [0] NCCL INFO Channel 12/0 : 96[0] -> 64[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62886:63057 [5] NCCL INFO Channel 09/0 : 53[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO Channel 12/0 : 101[5] -> 85[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO Channel 07/0 : 124[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO Channel 08/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO Channel 06/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96563:96756 [0] NCCL INFO Channel 12/0 : 64[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114516:114691 [4] NCCL INFO Channel 06/0 : 68[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO Channel 12/0 : 103[7] -> 87[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO Channel 07/0 : 59[3] -> 123[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO Channel 03/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96566:96755 [3] NCCL INFO Channel 04/0 : 35[3] -> 67[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153567:153742 [5] NCCL INFO Channel 09/0 : 93[5] -> 77[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114518:114693 [6] NCCL INFO Channel 04/0 : 70[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114514:114689 [2] NCCL INFO Channel 04/0 : 66[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO Channel 14/0 : 97[1] -> 81[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO Channel 09/0 : 126[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO Channel 06/0 : 81[1] -> 73[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96563:96756 [0] NCCL INFO Channel 14/0 : 64[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153569:153739 [7] NCCL INFO Channel 01/0 : 95[7] -> 79[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114515:114690 [3] NCCL INFO Channel 08/0 : 67[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62882:63056 [1] NCCL INFO Channel 12/0 : 49[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO Channel 01/0 : 120[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO Channel 09/0 : 59[3] -> 123[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:124164:124336 [4] NCCL INFO Channel 11/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96566:96755 [3] NCCL INFO Channel 06/0 : 35[3] -> 67[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114512:114686 [0] NCCL INFO Channel 14/0 : 96[0] -> 64[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62888:63060 [7] NCCL INFO Channel 05/0 : 55[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO Channel 09/0 : 124[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO Channel 10/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO Channel 07/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96566:96755 [3] NCCL INFO Channel 08/0 : 35[3] -> 67[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114516:114691 [4] NCCL INFO Channel 08/0 : 68[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62886:63057 [5] NCCL INFO Channel 10/0 : 53[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO Channel 11/0 : 126[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO Channel 04/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96566:96755 [3] NCCL INFO Channel 10/0 : 35[3] -> 67[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114518:114693 [6] NCCL INFO Channel 08/0 : 70[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO Channel 14/0 : 101[5] -> 85[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO Channel 13/0 : 59[3] -> 123[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO Channel 07/0 : 81[1] -> 73[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96566:96755 [3] NCCL INFO Channel 12/0 : 35[3] -> 67[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153567:153742 [5] NCCL INFO Channel 11/0 : 93[5] -> 77[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114514:114689 [2] NCCL INFO Channel 06/0 : 66[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO Channel 14/0 : 103[7] -> 87[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO Channel 03/0 : 120[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:124164:124336 [4] NCCL INFO Channel 13/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96569:96758 [6] NCCL INFO Channel 04/0 : 38[6] -> 70[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153569:153739 [7] NCCL INFO Channel 03/0 : 95[7] -> 79[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114515:114690 [3] NCCL INFO Channel 10/0 : 67[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62882:63056 [1] NCCL INFO Channel 13/0 : 49[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO Channel 01/0 : 90[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO Channel 11/0 : 124[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO Channel 11/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96566:96755 [3] NCCL INFO Channel 14/0 : 35[3] -> 67[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153567:153742 [5] NCCL INFO Channel 15/0 : 93[5] -> 77[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114512:114686 [0] NCCL INFO Channel 02/0 : 64[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62888:63060 [7] NCCL INFO Channel 06/0 : 55[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62886:63057 [5] NCCL INFO Channel 11/0 : 53[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO Channel 13/0 : 126[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO Channel 08/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96569:96758 [6] NCCL INFO Channel 08/0 : 38[6] -> 70[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO Channel 15/0 : 59[3] -> 123[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO Channel 05/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO Channel 08/0 : 81[1] -> 73[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96569:96758 [6] NCCL INFO Channel 10/0 : 38[6] -> 70[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO Channel 05/0 : 120[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:124164:124336 [4] NCCL INFO Channel 15/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96569:96758 [6] NCCL INFO Channel 12/0 : 38[6] -> 70[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153569:153739 [7] NCCL INFO Channel 05/0 : 95[7] -> 79[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62882:63056 [1] NCCL INFO Channel 14/0 : 49[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO Channel 03/0 : 90[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96566:96755 [3] NCCL INFO Channel 00/0 : 67[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114516:114691 [4] NCCL INFO Channel 10/0 : 68[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62888:63060 [7] NCCL INFO Channel 08/0 : 55[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62886:63057 [5] NCCL INFO Channel 12/0 : 53[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO Channel 13/0 : 124[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO Channel 15/0 : 126[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96569:96758 [6] NCCL INFO Channel 14/0 : 38[6] -> 70[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114518:114693 [6] NCCL INFO Channel 10/0 : 70[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO Channel 05/0 : 90[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO Channel 07/0 : 120[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO Channel 12/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO Channel 10/0 : 81[1] -> 73[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96566:96755 [3] NCCL INFO Channel 02/0 : 67[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114514:114689 [2] NCCL INFO Channel 08/0 : 66[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO Channel 01/0 : 62[6] -> 126[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO Channel 15/0 : 124[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO Channel 09/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124166:124339 [6] NCCL INFO Channel 01/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96566:96755 [3] NCCL INFO Channel 04/0 : 67[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153569:153739 [7] NCCL INFO Channel 07/0 : 95[7] -> 79[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114515:114690 [3] NCCL INFO Channel 12/0 : 67[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114512:114686 [0] NCCL INFO Channel 04/0 : 64[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272151:272326 [1] NCCL INFO Channel 00/0 : 121[1] -> 113[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO Channel 07/0 : 90[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO Channel 06/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96566:96755 [3] NCCL INFO Channel 06/0 : 67[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114516:114691 [4] NCCL INFO Channel 12/0 : 68[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62882:63056 [1] NCCL INFO Channel 15/0 : 49[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO Channel 09/0 : 120[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO Channel 11/0 : 81[1] -> 73[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO Channel 00/0 : 63[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96569:96758 [6] NCCL INFO Channel 00/0 : 70[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114518:114693 [6] NCCL INFO Channel 12/0 : 70[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62888:63060 [7] NCCL INFO Channel 09/0 : 55[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62886:63057 [5] NCCL INFO Channel 14/0 : 53[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO Channel 00/0 : 116[4] -> 100[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO Channel 01/0 : 91[3] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:124166:124339 [6] NCCL INFO Channel 03/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96566:96755 [3] NCCL INFO Channel 08/0 : 67[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153569:153739 [7] NCCL INFO Channel 09/0 : 95[7] -> 79[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114514:114689 [2] NCCL INFO Channel 10/0 : 66[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62882:63056 [1] NCCL INFO Channel 01/0 : 41[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272151:272326 [1] NCCL INFO Channel 02/0 : 121[1] -> 113[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272155:272329 [5] NCCL INFO Channel 00/0 : 125[5] -> 117[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96895:97069 [7] NCCL INFO Channel 00/0 : 31[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO Channel 03/0 : 62[6] -> 126[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO Channel 01/0 : 60[4] -> 124[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO Channel 13/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96569:96758 [6] NCCL INFO Channel 02/0 : 70[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114515:114690 [3] NCCL INFO Channel 14/0 : 67[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO Channel 09/0 : 90[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO Channel 10/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO Channel 12/0 : 81[1] -> 73[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO Channel 00/0 : 61[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96566:96755 [3] NCCL INFO Channel 10/0 : 67[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114512:114686 [0] NCCL INFO Channel 06/0 : 64[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO Channel 00/0 : 114[2] -> 98[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO Channel 01/0 : 105[1] -> 97[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO Channel 11/0 : 120[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO Channel 03/0 : 91[3] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO Channel 08/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO Channel 00/0 : 89[1] -> 81[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124166:124339 [6] NCCL INFO Channel 05/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96569:96758 [6] NCCL INFO Channel 04/0 : 70[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153569:153739 [7] NCCL INFO Channel 11/0 : 95[7] -> 79[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114516:114691 [4] NCCL INFO Channel 14/0 : 68[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62888:63060 [7] NCCL INFO Channel 10/0 : 55[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96893:97067 [5] NCCL INFO Channel 00/0 : 29[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO Channel 03/0 : 60[4] -> 124[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO Channel 05/0 : 62[6] -> 126[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO Channel 11/0 : 90[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO Channel 14/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96566:96755 [3] NCCL INFO Channel 12/0 : 67[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114518:114693 [6] NCCL INFO Channel 14/0 : 70[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62886:63057 [5] NCCL INFO Channel 15/0 : 53[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272155:272329 [5] NCCL INFO Channel 02/0 : 125[5] -> 117[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96889:97065 [1] NCCL INFO Channel 00/0 : 25[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO Channel 05/0 : 91[3] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO Channel 13/0 : 120[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO Channel 13/0 : 81[1] -> 73[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO Channel 00/0 : 57[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO Channel 02/0 : 63[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96569:96758 [6] NCCL INFO Channel 08/0 : 70[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114514:114689 [2] NCCL INFO Channel 12/0 : 66[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62882:63056 [1] NCCL INFO Channel 03/0 : 41[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO Channel 02/0 : 116[4] -> 100[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96895:97069 [7] NCCL INFO Channel 02/0 : 31[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO Channel 05/0 : 60[4] -> 124[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO Channel 13/0 : 90[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO Channel 07/0 : 62[6] -> 126[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO Channel 00/0 : 95[7] -> 87[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO Channel 02/0 : 61[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124166:124339 [6] NCCL INFO Channel 07/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96564:96757 [1] NCCL INFO Channel 00/0 : 49[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96566:96755 [3] NCCL INFO Channel 14/0 : 67[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153569:153739 [7] NCCL INFO Channel 13/0 : 95[7] -> 79[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114512:114686 [0] NCCL INFO Channel 08/0 : 64[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO Channel 01/0 : 109[5] -> 101[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235649:235827 [1] NCCL INFO Channel 13/0 : 113[1] -> 105[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272151:272326 [1] NCCL INFO Channel 04/0 : 121[1] -> 113[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO Channel 07/0 : 91[3] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO Channel 11/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96569:96758 [6] NCCL INFO Channel 10/0 : 70[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153567:153742 [5] NCCL INFO Channel 00/0 : 93[5] -> 85[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114514:114689 [2] NCCL INFO Channel 14/0 : 66[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62886:63057 [5] NCCL INFO Channel 01/0 : 45[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62888:63060 [7] NCCL INFO Channel 11/0 : 55[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62882:63056 [1] NCCL INFO Channel 05/0 : 41[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO Channel 01/0 : 111[7] -> 103[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272157:272324 [7] NCCL INFO Channel 00/0 : 127[7] -> 119[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96893:97067 [5] NCCL INFO Channel 02/0 : 29[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO Channel 15/0 : 120[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO Channel 09/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO Channel 14/0 : 81[1] -> 73[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96570:96759 [7] NCCL INFO Channel 00/0 : 55[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114512:114686 [0] NCCL INFO Channel 10/0 : 64[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO Channel 04/0 : 114[2] -> 98[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO Channel 03/0 : 105[1] -> 97[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96889:97065 [1] NCCL INFO Channel 02/0 : 25[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO Channel 15/0 : 90[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO Channel 15/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO Channel 02/0 : 89[1] -> 81[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO Channel 04/0 : 63[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124166:124339 [6] NCCL INFO Channel 09/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96564:96757 [1] NCCL INFO Channel 02/0 : 49[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114512:114686 [0] NCCL INFO Channel 12/0 : 64[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62886:63057 [5] NCCL INFO Channel 03/0 : 45[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235649:235827 [1] NCCL INFO Channel 14/0 : 113[1] -> 105[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96895:97069 [7] NCCL INFO Channel 04/0 : 31[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96893:97067 [5] NCCL INFO Channel 04/0 : 29[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO Channel 09/0 : 91[3] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO Channel 09/0 : 62[6] -> 126[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO Channel 07/0 : 60[4] -> 124[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO Channel 01/0 : 56[0] -> 120[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO Channel 12/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO Channel 02/0 : 57[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96568:96753 [5] NCCL INFO Channel 00/0 : 53[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153567:153742 [5] NCCL INFO Channel 02/0 : 93[5] -> 85[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114515:114690 [3] NCCL INFO Channel 00/0 : 99[3] -> 67[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62888:63060 [7] NCCL INFO Channel 12/0 : 55[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62882:63056 [1] NCCL INFO Channel 07/0 : 41[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO Channel 06/0 : 116[4] -> 100[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272155:272329 [5] NCCL INFO Channel 04/0 : 125[5] -> 117[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272151:272326 [1] NCCL INFO Channel 06/0 : 121[1] -> 113[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO Channel 01/0 : 58[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO Channel 00/0 : 85[5] -> 77[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO Channel 04/0 : 61[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124166:124339 [6] NCCL INFO Channel 11/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96569:96758 [6] NCCL INFO Channel 12/0 : 70[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114518:114693 [6] NCCL INFO Channel 00/0 : 102[6] -> 70[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO Channel 03/0 : 111[7] -> 103[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO Channel 03/0 : 109[5] -> 101[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235649:235827 [1] NCCL INFO Channel 15/0 : 113[1] -> 105[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272157:272324 [7] NCCL INFO Channel 02/0 : 127[7] -> 119[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96889:97065 [1] NCCL INFO Channel 04/0 : 25[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO Channel 10/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO Channel 01/0 : 9[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO Channel 02/0 : 95[7] -> 87[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO Channel 04/0 : 89[1] -> 81[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO Channel 15/0 : 81[1] -> 73[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96570:96759 [7] NCCL INFO Channel 02/0 : 55[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62886:63057 [5] NCCL INFO Channel 05/0 : 45[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO Channel 05/0 : 105[1] -> 97[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO Channel 06/0 : 114[2] -> 98[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96893:97067 [5] NCCL INFO Channel 06/0 : 29[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96895:97069 [7] NCCL INFO Channel 06/0 : 31[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO Channel 13/0 : 91[3] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO Channel 14/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32707:32878 [3] NCCL INFO Channel 01/0 : 43[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO Channel 00/0 : 93[5] -> 85[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124166:124339 [6] NCCL INFO Channel 13/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96564:96757 [1] NCCL INFO Channel 04/0 : 49[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153564:153737 [2] NCCL INFO Channel 01/0 : 106[2] -> 90[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62888:63060 [7] NCCL INFO Channel 13/0 : 55[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62882:63056 [1] NCCL INFO Channel 11/0 : 41[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO Channel 02/0 : 112[0] -> 96[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235649:235827 [1] NCCL INFO Channel 01/0 : 105[1] -> 97[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272151:272326 [1] NCCL INFO Channel 08/0 : 121[1] -> 113[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272155:272329 [5] NCCL INFO Channel 06/0 : 125[5] -> 117[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96889:97065 [1] NCCL INFO Channel 06/0 : 25[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO Channel 11/0 : 62[6] -> 126[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO Channel 03/0 : 56[0] -> 120[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO Channel 09/0 : 60[4] -> 124[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO Channel 03/0 : 9[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO Channel 06/0 : 63[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96568:96753 [5] NCCL INFO Channel 02/0 : 53[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153567:153742 [5] NCCL INFO Channel 04/0 : 93[5] -> 85[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114512:114686 [0] NCCL INFO Channel 14/0 : 64[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62886:63057 [5] NCCL INFO Channel 07/0 : 45[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO Channel 08/0 : 116[4] -> 100[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272157:272324 [7] NCCL INFO Channel 04/0 : 127[7] -> 119[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO Channel 03/0 : 58[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO Channel 11/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO Channel 01/0 : 85[5] -> 77[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO Channel 04/0 : 57[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO Channel 06/0 : 61[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96569:96758 [6] NCCL INFO Channel 14/0 : 70[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153565:153738 [3] NCCL INFO Channel 01/0 : 107[3] -> 91[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114515:114690 [3] NCCL INFO Channel 02/0 : 99[3] -> 67[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235649:235827 [1] NCCL INFO Channel 03/0 : 105[1] -> 97[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO Channel 15/0 : 91[3] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO Channel 15/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32707:32878 [3] NCCL INFO Channel 03/0 : 43[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO Channel 06/0 : 89[1] -> 81[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO Channel 01/0 : 73[1] -> 65[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO Channel 08/0 : 63[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124166:124339 [6] NCCL INFO Channel 15/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96570:96759 [7] NCCL INFO Channel 04/0 : 55[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114518:114693 [6] NCCL INFO Channel 02/0 : 102[6] -> 70[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62888:63060 [7] NCCL INFO Channel 14/0 : 55[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62882:63056 [1] NCCL INFO Channel 13/0 : 41[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62886:63057 [5] NCCL INFO Channel 09/0 : 45[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96893:97067 [5] NCCL INFO Channel 08/0 : 29[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96895:97069 [7] NCCL INFO Channel 08/0 : 31[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO Channel 05/0 : 56[0] -> 120[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO Channel 05/0 : 9[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO Channel 04/0 : 95[7] -> 87[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO Channel 02/0 : 85[5] -> 77[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96564:96757 [1] NCCL INFO Channel 06/0 : 49[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114516:114691 [4] NCCL INFO Channel 01/0 : 76[4] -> 68[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO Channel 05/0 : 111[7] -> 103[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO Channel 05/0 : 109[5] -> 101[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272151:272326 [1] NCCL INFO Channel 10/0 : 121[1] -> 113[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272157:272324 [7] NCCL INFO Channel 06/0 : 127[7] -> 119[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272155:272329 [5] NCCL INFO Channel 08/0 : 125[5] -> 117[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96889:97065 [1] NCCL INFO Channel 08/0 : 25[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO Channel 13/0 : 62[6] -> 126[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32707:32878 [3] NCCL INFO Channel 05/0 : 43[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO Channel 02/0 : 93[5] -> 85[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96568:96753 [5] NCCL INFO Channel 04/0 : 53[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153564:153737 [2] NCCL INFO Channel 03/0 : 106[2] -> 90[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114515:114690 [3] NCCL INFO Channel 04/0 : 99[3] -> 67[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO Channel 07/0 : 105[1] -> 97[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO Channel 08/0 : 114[2] -> 98[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO Channel 04/0 : 112[0] -> 96[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235649:235827 [1] NCCL INFO Channel 05/0 : 105[1] -> 97[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96893:97067 [5] NCCL INFO Channel 10/0 : 29[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO Channel 05/0 : 58[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO Channel 08/0 : 89[1] -> 81[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO Channel 06/0 : 57[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96570:96759 [7] NCCL INFO Channel 06/0 : 55[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153567:153742 [5] NCCL INFO Channel 06/0 : 93[5] -> 85[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114518:114693 [6] NCCL INFO Channel 04/0 : 102[6] -> 70[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62888:63060 [7] NCCL INFO Channel 01/0 : 47[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62886:63057 [5] NCCL INFO Channel 11/0 : 45[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62882:63056 [1] NCCL INFO Channel 15/0 : 41[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO Channel 10/0 : 116[4] -> 100[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO Channel 11/0 : 60[4] -> 124[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO Channel 01/0 : 13[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO Channel 03/0 : 73[1] -> 65[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO Channel 10/0 : 63[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO Channel 08/0 : 61[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96564:96757 [1] NCCL INFO Channel 08/0 : 49[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153565:153738 [3] NCCL INFO Channel 03/0 : 107[3] -> 91[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114514:114689 [2] NCCL INFO Channel 01/0 : 74[2] -> 66[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO Channel 07/0 : 111[7] -> 103[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO Channel 07/0 : 109[5] -> 101[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272151:272326 [1] NCCL INFO Channel 12/0 : 121[1] -> 113[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO Channel 01/0 : 59[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO Channel 12/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32707:32878 [3] NCCL INFO Channel 07/0 : 43[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO Channel 03/0 : 85[5] -> 77[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO Channel 08/0 : 57[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96568:96753 [5] NCCL INFO Channel 06/0 : 53[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153567:153742 [5] NCCL INFO Channel 08/0 : 93[5] -> 85[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114515:114690 [3] NCCL INFO Channel 06/0 : 99[3] -> 67[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114516:114691 [4] NCCL INFO Channel 03/0 : 76[4] -> 68[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO Channel 11/0 : 105[1] -> 97[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235649:235827 [1] NCCL INFO Channel 07/0 : 105[1] -> 97[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272157:272324 [7] NCCL INFO Channel 08/0 : 127[7] -> 119[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272155:272329 [5] NCCL INFO Channel 10/0 : 125[5] -> 117[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96895:97069 [7] NCCL INFO Channel 10/0 : 31[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO Channel 07/0 : 56[0] -> 120[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO Channel 07/0 : 9[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO Channel 05/0 : 73[1] -> 65[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96570:96759 [7] NCCL INFO Channel 08/0 : 55[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153564:153737 [2] NCCL INFO Channel 05/0 : 106[2] -> 90[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114518:114693 [6] NCCL INFO Channel 08/0 : 102[6] -> 70[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO Channel 10/0 : 114[2] -> 98[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO Channel 06/0 : 112[0] -> 96[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96889:97065 [1] NCCL INFO Channel 10/0 : 25[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO Channel 15/0 : 62[6] -> 126[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO Channel 03/0 : 13[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO Channel 06/0 : 95[7] -> 87[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO Channel 04/0 : 93[5] -> 85[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO Channel 00/0 : 87[7] -> 79[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96568:96753 [5] NCCL INFO Channel 08/0 : 53[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96564:96757 [1] NCCL INFO Channel 10/0 : 49[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153569:153739 [7] NCCL INFO Channel 00/0 : 95[7] -> 87[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114514:114689 [2] NCCL INFO Channel 03/0 : 74[2] -> 66[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62888:63060 [7] NCCL INFO Channel 03/0 : 47[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62886:63057 [5] NCCL INFO Channel 15/0 : 45[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO Channel 12/0 : 116[4] -> 100[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96893:97067 [5] NCCL INFO Channel 12/0 : 29[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO Channel 07/0 : 58[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO Channel 11/0 : 9[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32707:32878 [3] NCCL INFO Channel 09/0 : 43[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO Channel 10/0 : 89[1] -> 81[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO Channel 12/0 : 63[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO Channel 10/0 : 61[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96570:96759 [7] NCCL INFO Channel 10/0 : 55[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114515:114690 [3] NCCL INFO Channel 08/0 : 99[3] -> 67[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235649:235827 [1] NCCL INFO Channel 11/0 : 105[1] -> 97[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272151:272326 [1] NCCL INFO Channel 14/0 : 121[1] -> 113[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96895:97069 [7] NCCL INFO Channel 12/0 : 31[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96889:97065 [1] NCCL INFO Channel 12/0 : 25[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO Channel 03/0 : 59[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO Channel 13/0 : 60[4] -> 124[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO Channel 13/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO Channel 06/0 : 93[5] -> 85[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO Channel 10/0 : 57[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96568:96753 [5] NCCL INFO Channel 10/0 : 53[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114516:114691 [4] NCCL INFO Channel 05/0 : 76[4] -> 68[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114518:114693 [6] NCCL INFO Channel 10/0 : 102[6] -> 70[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272155:272329 [5] NCCL INFO Channel 12/0 : 125[5] -> 117[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272157:272324 [7] NCCL INFO Channel 10/0 : 127[7] -> 119[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96893:97067 [5] NCCL INFO Channel 14/0 : 29[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO Channel 09/0 : 56[0] -> 120[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO Channel 05/0 : 13[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO Channel 08/0 : 95[7] -> 87[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO Channel 04/0 : 85[5] -> 77[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO Channel 14/0 : 63[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO Channel 12/0 : 61[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96564:96757 [1] NCCL INFO Channel 12/0 : 49[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96570:96759 [7] NCCL INFO Channel 12/0 : 55[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153567:153742 [5] NCCL INFO Channel 10/0 : 93[5] -> 85[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114514:114689 [2] NCCL INFO Channel 05/0 : 74[2] -> 66[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62888:63060 [7] NCCL INFO Channel 05/0 : 47[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO Channel 09/0 : 111[7] -> 103[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO Channel 09/0 : 58[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32707:32878 [3] NCCL INFO Channel 13/0 : 43[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO Channel 12/0 : 89[1] -> 81[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO Channel 07/0 : 73[1] -> 65[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96568:96753 [5] NCCL INFO Channel 12/0 : 53[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153565:153738 [3] NCCL INFO Channel 05/0 : 107[3] -> 91[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153569:153739 [7] NCCL INFO Channel 02/0 : 95[7] -> 87[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114512:114686 [0] NCCL INFO Channel 01/0 : 72[0] -> 64[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO Channel 09/0 : 109[5] -> 101[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO Channel 13/0 : 105[1] -> 97[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272151:272326 [1] NCCL INFO Channel 00/0 : 113[1] -> 105[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272157:272324 [7] NCCL INFO Channel 12/0 : 127[7] -> 119[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272155:272329 [5] NCCL INFO Channel 14/0 : 125[5] -> 117[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96889:97065 [1] NCCL INFO Channel 14/0 : 25[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96895:97069 [7] NCCL INFO Channel 14/0 : 31[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96893:97067 [5] NCCL INFO Channel 00/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO Channel 05/0 : 59[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO Channel 01/0 : 87[7] -> 79[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96570:96759 [7] NCCL INFO Channel 14/0 : 55[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153564:153737 [2] NCCL INFO Channel 07/0 : 106[2] -> 90[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114515:114690 [3] NCCL INFO Channel 10/0 : 99[3] -> 67[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO Channel 12/0 : 114[2] -> 98[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235649:235827 [1] NCCL INFO Channel 13/0 : 105[1] -> 97[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO Channel 15/0 : 60[4] -> 124[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO Channel 11/0 : 56[0] -> 120[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO Channel 13/0 : 9[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO Channel 11/0 : 73[1] -> 65[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO Channel 06/0 : 85[5] -> 77[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96564:96757 [1] NCCL INFO Channel 14/0 : 49[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153567:153742 [5] NCCL INFO Channel 12/0 : 93[5] -> 85[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114516:114691 [4] NCCL INFO Channel 07/0 : 76[4] -> 68[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62888:63060 [7] NCCL INFO Channel 07/0 : 47[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO Channel 08/0 : 112[0] -> 96[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO Channel 11/0 : 58[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO Channel 14/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32707:32878 [3] NCCL INFO Channel 15/0 : 43[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO Channel 12/0 : 57[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96568:96753 [5] NCCL INFO Channel 14/0 : 53[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114518:114693 [6] NCCL INFO Channel 12/0 : 102[6] -> 70[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO Channel 11/0 : 111[7] -> 103[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO Channel 14/0 : 116[4] -> 100[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO Channel 11/0 : 109[5] -> 101[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO Channel 15/0 : 105[1] -> 97[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272155:272329 [5] NCCL INFO Channel 00/0 : 117[5] -> 109[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96889:97065 [1] NCCL INFO Channel 00/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO Channel 07/0 : 59[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO Channel 07/0 : 13[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32705:32880 [1] NCCL INFO Channel 14/0 : 25[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO Channel 08/0 : 93[5] -> 85[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO Channel 00/0 : 55[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96570:96759 [7] NCCL INFO Channel 00/0 : 39[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114514:114689 [2] NCCL INFO Channel 07/0 : 74[2] -> 66[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO Channel 14/0 : 114[2] -> 98[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235655:235828 [7] NCCL INFO Channel 00/0 : 119[7] -> 111[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272157:272324 [7] NCCL INFO Channel 14/0 : 127[7] -> 119[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272151:272326 [1] NCCL INFO Channel 02/0 : 113[1] -> 105[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96893:97067 [5] NCCL INFO Channel 01/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO Channel 13/0 : 56[0] -> 120[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO Channel 15/0 : 9[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO Channel 10/0 : 95[7] -> 87[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO Channel 14/0 : 89[1] -> 81[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO Channel 02/0 : 87[7] -> 79[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO Channel 14/0 : 61[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96564:96757 [1] NCCL INFO Channel 00/0 : 33[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153569:153739 [7] NCCL INFO Channel 04/0 : 95[7] -> 87[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114512:114686 [0] NCCL INFO Channel 03/0 : 72[0] -> 64[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62888:63060 [7] NCCL INFO Channel 09/0 : 47[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO Channel 10/0 : 112[0] -> 96[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235649:235827 [1] NCCL INFO Channel 15/0 : 105[1] -> 97[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96895:97069 [7] NCCL INFO Channel 00/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO Channel 13/0 : 58[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO Channel 01/0 : 15[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO Channel 10/0 : 93[5] -> 85[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO Channel 13/0 : 73[1] -> 65[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO Channel 14/0 : 57[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96568:96753 [5] NCCL INFO Channel 00/0 : 37[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153565:153738 [3] NCCL INFO Channel 07/0 : 107[3] -> 91[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153564:153737 [2] NCCL INFO Channel 09/0 : 106[2] -> 90[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114515:114690 [3] NCCL INFO Channel 12/0 : 99[3] -> 67[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272155:272329 [5] NCCL INFO Channel 01/0 : 117[5] -> 109[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO Channel 01/0 : 94[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO Channel 09/0 : 13[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32706:32877 [2] NCCL INFO Channel 01/0 : 42[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO Channel 00/0 : 81[1] -> 73[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO Channel 07/0 : 85[5] -> 77[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO Channel 03/0 : 87[7] -> 79[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96570:96759 [7] NCCL INFO Channel 02/0 : 39[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153567:153742 [5] NCCL INFO Channel 14/0 : 93[5] -> 85[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62888:63060 [7] NCCL INFO Channel 11/0 : 47[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272151:272326 [1] NCCL INFO Channel 03/0 : 113[1] -> 105[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272157:272324 [7] NCCL INFO Channel 00/0 : 119[7] -> 111[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO Channel 09/0 : 59[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO Channel 15/0 : 56[0] -> 120[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO Channel 03/0 : 15[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32707:32878 [3] NCCL INFO Channel 01/0 : 27[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO Channel 12/0 : 95[7] -> 87[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO Channel 15/0 : 73[1] -> 65[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96564:96757 [1] NCCL INFO Channel 02/0 : 33[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153569:153739 [7] NCCL INFO Channel 06/0 : 95[7] -> 87[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO Channel 13/0 : 111[7] -> 103[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235655:235828 [7] NCCL INFO Channel 01/0 : 119[7] -> 111[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96889:97065 [1] NCCL INFO Channel 02/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO Channel 15/0 : 58[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO Channel 11/0 : 13[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO Channel 12/0 : 93[5] -> 85[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96568:96753 [5] NCCL INFO Channel 02/0 : 37[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153565:153738 [3] NCCL INFO Channel 09/0 : 107[3] -> 91[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO Channel 15/0 : 109[5] -> 101[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235655:235828 [7] NCCL INFO Channel 02/0 : 119[7] -> 111[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96893:97067 [5] NCCL INFO Channel 02/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96895:97069 [7] NCCL INFO Channel 01/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO Channel 03/0 : 94[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO Channel 01/0 : 55[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96570:96759 [7] NCCL INFO Channel 04/0 : 39[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153564:153737 [2] NCCL INFO Channel 11/0 : 106[2] -> 90[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114516:114691 [4] NCCL INFO Channel 09/0 : 76[4] -> 68[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO Channel 00/0 : 100[4] -> 84[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272155:272329 [5] NCCL INFO Channel 02/0 : 117[5] -> 109[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96889:97065 [1] NCCL INFO Channel 03/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO Channel 13/0 : 59[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32706:32877 [2] NCCL INFO Channel 03/0 : 42[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO Channel 00/0 : 53[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96564:96757 [1] NCCL INFO Channel 04/0 : 33[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114518:114693 [6] NCCL INFO Channel 14/0 : 102[6] -> 70[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114514:114689 [2] NCCL INFO Channel 09/0 : 74[2] -> 66[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62888:63060 [7] NCCL INFO Channel 13/0 : 47[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO Channel 00/0 : 98[2] -> 82[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272151:272326 [1] NCCL INFO Channel 04/0 : 113[1] -> 105[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272157:272324 [7] NCCL INFO Channel 01/0 : 119[7] -> 111[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO Channel 01/0 : 92[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO Channel 05/0 : 15[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32707:32878 [3] NCCL INFO Channel 03/0 : 27[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO Channel 00/0 : 49[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96568:96753 [5] NCCL INFO Channel 04/0 : 37[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114512:114686 [0] NCCL INFO Channel 05/0 : 72[0] -> 64[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO Channel 12/0 : 112[0] -> 96[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235655:235828 [7] NCCL INFO Channel 03/0 : 119[7] -> 111[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO Channel 05/0 : 94[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO Channel 15/0 : 13[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO Channel 02/0 : 81[1] -> 73[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO Channel 02/0 : 55[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96570:96759 [7] NCCL INFO Channel 06/0 : 39[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153569:153739 [7] NCCL INFO Channel 08/0 : 95[7] -> 87[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114515:114690 [3] NCCL INFO Channel 14/0 : 99[3] -> 67[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO Channel 02/0 : 100[4] -> 84[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235653:235825 [5] NCCL INFO Channel 00/0 : 117[5] -> 109[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96895:97069 [7] NCCL INFO Channel 02/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96893:97067 [5] NCCL INFO Channel 03/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO Channel 15/0 : 59[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO Channel 07/0 : 15[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO Channel 14/0 : 95[7] -> 87[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO Channel 01/0 : 53[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96564:96757 [1] NCCL INFO Channel 06/0 : 33[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153565:153738 [3] NCCL INFO Channel 13/0 : 107[3] -> 91[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114516:114691 [4] NCCL INFO Channel 11/0 : 76[4] -> 68[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO Channel 04/0 : 98[2] -> 82[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272155:272329 [5] NCCL INFO Channel 03/0 : 117[5] -> 109[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96889:97065 [1] NCCL INFO Channel 04/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO Channel 03/0 : 92[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO Channel 09/0 : 15[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO Channel 14/0 : 93[5] -> 85[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO Channel 02/0 : 49[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96568:96753 [5] NCCL INFO Channel 06/0 : 37[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153564:153737 [2] NCCL INFO Channel 13/0 : 106[2] -> 90[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114518:114693 [6] NCCL INFO Channel 00/0 : 70[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114514:114689 [2] NCCL INFO Channel 11/0 : 74[2] -> 66[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO Channel 14/0 : 112[0] -> 96[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272151:272326 [1] NCCL INFO Channel 05/0 : 113[1] -> 105[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272157:272324 [7] NCCL INFO Channel 02/0 : 119[7] -> 111[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96893:97067 [5] NCCL INFO Channel 04/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96895:97069 [7] NCCL INFO Channel 03/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO Channel 07/0 : 94[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO Channel 01/0 : 88[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO Channel 11/0 : 15[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32706:32877 [2] NCCL INFO Channel 05/0 : 42[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32707:32878 [3] NCCL INFO Channel 05/0 : 27[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO Channel 03/0 : 81[1] -> 73[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO Channel 03/0 : 55[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96570:96759 [7] NCCL INFO Channel 08/0 : 39[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153569:153739 [7] NCCL INFO Channel 10/0 : 95[7] -> 87[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114515:114690 [3] NCCL INFO Channel 00/0 : 67[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114512:114686 [0] NCCL INFO Channel 07/0 : 72[0] -> 64[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO Channel 06/0 : 100[4] -> 84[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235655:235828 [7] NCCL INFO Channel 04/0 : 119[7] -> 111[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272155:272329 [5] NCCL INFO Channel 04/0 : 117[5] -> 109[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO Channel 05/0 : 92[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO Channel 00/0 : 87[7] -> 79[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO Channel 00/0 : 85[5] -> 77[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96568:96753 [5] NCCL INFO Channel 08/0 : 37[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153565:153738 [3] NCCL INFO Channel 15/0 : 107[3] -> 91[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114516:114691 [4] NCCL INFO Channel 13/0 : 76[4] -> 68[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO Channel 06/0 : 98[2] -> 82[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96889:97065 [1] NCCL INFO Channel 05/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO Channel 09/0 : 94[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO Channel 04/0 : 81[1] -> 73[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96564:96757 [1] NCCL INFO Channel 08/0 : 33[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96570:96759 [7] NCCL INFO Channel 10/0 : 39[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153564:153737 [2] NCCL INFO Channel 15/0 : 106[2] -> 90[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153569:153739 [7] NCCL INFO Channel 12/0 : 95[7] -> 87[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114518:114693 [6] NCCL INFO Channel 02/0 : 70[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96893:97067 [5] NCCL INFO Channel 06/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO Channel 03/0 : 88[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO Channel 13/0 : 15[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32707:32878 [3] NCCL INFO Channel 07/0 : 27[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32706:32877 [2] NCCL INFO Channel 07/0 : 42[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO Channel 01/0 : 87[7] -> 79[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO Channel 01/0 : 85[5] -> 77[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO Channel 02/0 : 53[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96568:96753 [5] NCCL INFO Channel 10/0 : 37[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153565:153738 [3] NCCL INFO Channel 01/0 : 91[3] -> 75[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114514:114689 [2] NCCL INFO Channel 13/0 : 74[2] -> 66[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235653:235825 [5] NCCL INFO Channel 01/0 : 117[5] -> 109[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272151:272326 [1] NCCL INFO Channel 06/0 : 113[1] -> 105[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96895:97069 [7] NCCL INFO Channel 04/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO Channel 07/0 : 92[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32707:32878 [3] NCCL INFO Channel 09/0 : 27[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO Channel 03/0 : 49[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96570:96759 [7] NCCL INFO Channel 12/0 : 39[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114515:114690 [3] NCCL INFO Channel 02/0 : 67[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114512:114686 [0] NCCL INFO Channel 09/0 : 72[0] -> 64[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO Channel 02/0 : 96[0] -> 80[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235655:235828 [7] NCCL INFO Channel 05/0 : 119[7] -> 111[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272157:272324 [7] NCCL INFO Channel 03/0 : 119[7] -> 111[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96889:97065 [1] NCCL INFO Channel 06/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO Channel 00/0 : 58[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32706:32877 [2] NCCL INFO Channel 09/0 : 42[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO Channel 04/0 : 55[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96564:96757 [1] NCCL INFO Channel 10/0 : 33[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114516:114691 [4] NCCL INFO Channel 15/0 : 76[4] -> 68[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO Channel 08/0 : 100[4] -> 84[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272155:272329 [5] NCCL INFO Channel 06/0 : 117[5] -> 109[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO Channel 11/0 : 94[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO Channel 05/0 : 81[1] -> 73[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO Channel 03/0 : 53[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96568:96753 [5] NCCL INFO Channel 12/0 : 37[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153569:153739 [7] NCCL INFO Channel 14/0 : 95[7] -> 87[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114518:114693 [6] NCCL INFO Channel 04/0 : 70[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO Channel 08/0 : 98[2] -> 82[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO Channel 05/0 : 88[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO Channel 09/0 : 92[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO Channel 02/0 : 87[7] -> 79[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO Channel 05/0 : 55[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO Channel 04/0 : 49[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96570:96759 [7] NCCL INFO Channel 14/0 : 39[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153565:153738 [3] NCCL INFO Channel 03/0 : 91[3] -> 75[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114514:114689 [2] NCCL INFO Channel 15/0 : 74[2] -> 66[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114515:114690 [3] NCCL INFO Channel 04/0 : 67[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO Channel 04/0 : 96[0] -> 80[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235653:235825 [5] NCCL INFO Channel 02/0 : 117[5] -> 109[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96893:97067 [5] NCCL INFO Channel 07/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO Channel 04/0 : 58[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32707:32878 [3] NCCL INFO Channel 13/0 : 27[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO Channel 02/0 : 85[5] -> 77[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO Channel 04/0 : 53[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96564:96757 [1] NCCL INFO Channel 12/0 : 33[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153564:153737 [2] NCCL INFO Channel 01/0 : 90[2] -> 74[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114512:114686 [0] NCCL INFO Channel 11/0 : 72[0] -> 64[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO Channel 10/0 : 100[4] -> 84[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235655:235828 [7] NCCL INFO Channel 06/0 : 119[7] -> 111[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272151:272326 [1] NCCL INFO Channel 07/0 : 113[1] -> 105[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96895:97069 [7] NCCL INFO Channel 05/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO Channel 13/0 : 94[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32706:32877 [2] NCCL INFO Channel 11/0 : 42[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO Channel 06/0 : 81[1] -> 73[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153565:153738 [3] NCCL INFO Channel 05/0 : 91[3] -> 75[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153564:153737 [2] NCCL INFO Channel 03/0 : 90[2] -> 74[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114518:114693 [6] NCCL INFO Channel 08/0 : 70[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO Channel 10/0 : 98[2] -> 82[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272157:272324 [7] NCCL INFO Channel 04/0 : 119[7] -> 111[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96889:97065 [1] NCCL INFO Channel 07/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO Channel 03/0 : 87[7] -> 79[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO Channel 06/0 : 55[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO Channel 06/0 : 53[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO Channel 05/0 : 49[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114515:114690 [3] NCCL INFO Channel 06/0 : 67[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO Channel 06/0 : 96[0] -> 80[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272155:272329 [5] NCCL INFO Channel 07/0 : 117[5] -> 109[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96893:97067 [5] NCCL INFO Channel 08/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO Channel 03/0 : 85[5] -> 77[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO Channel 07/0 : 81[1] -> 73[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO Channel 08/0 : 55[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO Channel 07/0 : 53[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO Channel 06/0 : 49[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96568:96753 [5] NCCL INFO Channel 14/0 : 37[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114512:114686 [0] NCCL INFO Channel 13/0 : 72[0] -> 64[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114518:114693 [6] NCCL INFO Channel 10/0 : 70[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO Channel 12/0 : 100[4] -> 84[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235655:235828 [7] NCCL INFO Channel 08/0 : 119[7] -> 111[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272151:272326 [1] NCCL INFO Channel 08/0 : 113[1] -> 105[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272157:272324 [7] NCCL INFO Channel 05/0 : 119[7] -> 111[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO Channel 07/0 : 88[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO Channel 00/0 : 59[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32707:32878 [3] NCCL INFO Channel 15/0 : 27[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO Channel 04/0 : 87[7] -> 79[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96564:96757 [1] NCCL INFO Channel 14/0 : 33[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153565:153738 [3] NCCL INFO Channel 07/0 : 91[3] -> 75[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153564:153737 [2] NCCL INFO Channel 05/0 : 90[2] -> 74[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114515:114690 [3] NCCL INFO Channel 08/0 : 67[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO Channel 12/0 : 98[2] -> 82[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235653:235825 [5] NCCL INFO Channel 03/0 : 117[5] -> 109[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272155:272329 [5] NCCL INFO Channel 08/0 : 117[5] -> 109[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96895:97069 [7] NCCL INFO Channel 06/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96889:97065 [1] NCCL INFO Channel 08/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO Channel 11/0 : 92[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO Channel 06/0 : 58[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32706:32877 [2] NCCL INFO Channel 13/0 : 42[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96570:96759 [7] NCCL INFO Channel 01/0 : 47[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153565:153738 [3] NCCL INFO Channel 09/0 : 91[3] -> 75[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO Channel 08/0 : 96[0] -> 80[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235655:235828 [7] NCCL INFO Channel 09/0 : 119[7] -> 111[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96893:97067 [5] NCCL INFO Channel 09/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO Channel 15/0 : 94[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32706:32877 [2] NCCL INFO Channel 15/0 : 42[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO Channel 08/0 : 53[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96568:96753 [5] NCCL INFO Channel 01/0 : 45[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153564:153737 [2] NCCL INFO Channel 07/0 : 90[2] -> 74[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO Channel 14/0 : 100[4] -> 84[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235653:235825 [5] NCCL INFO Channel 04/0 : 117[5] -> 109[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96889:97065 [1] NCCL INFO Channel 10/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96895:97069 [7] NCCL INFO Channel 08/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO Channel 09/0 : 88[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO Channel 04/0 : 85[5] -> 77[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO Channel 08/0 : 81[1] -> 73[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO Channel 09/0 : 55[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96567:96760 [4] NCCL INFO Channel 00/0 : 52[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114512:114686 [0] NCCL INFO Channel 15/0 : 72[0] -> 64[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114518:114693 [6] NCCL INFO Channel 12/0 : 70[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272151:272326 [1] NCCL INFO Channel 10/0 : 113[1] -> 105[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272155:272329 [5] NCCL INFO Channel 09/0 : 117[5] -> 109[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272157:272324 [7] NCCL INFO Channel 06/0 : 119[7] -> 111[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96893:97067 [5] NCCL INFO Channel 10/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO Channel 02/0 : 59[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32706:32877 [2] NCCL INFO Channel 01/0 : 26[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO Channel 05/0 : 87[7] -> 79[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96570:96759 [7] NCCL INFO Channel 03/0 : 47[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114515:114690 [3] NCCL INFO Channel 10/0 : 67[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO Channel 14/0 : 98[2] -> 82[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO Channel 10/0 : 96[0] -> 80[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272155:272329 [5] NCCL INFO Channel 10/0 : 117[5] -> 109[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272157:272324 [7] NCCL INFO Channel 08/0 : 119[7] -> 111[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272151:272326 [1] NCCL INFO Channel 11/0 : 113[1] -> 105[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO Channel 13/0 : 92[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO Channel 08/0 : 58[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO Channel 06/0 : 85[5] -> 77[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO Channel 10/0 : 81[1] -> 73[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96564:96757 [1] NCCL INFO Channel 01/0 : 41[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153565:153738 [3] NCCL INFO Channel 13/0 : 91[3] -> 75[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114518:114693 [6] NCCL INFO Channel 14/0 : 70[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO Channel 12/0 : 96[0] -> 80[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235655:235828 [7] NCCL INFO Channel 10/0 : 119[7] -> 111[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96889:97065 [1] NCCL INFO Channel 11/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO Channel 01/0 : 62[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32706:32877 [2] NCCL INFO Channel 03/0 : 26[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO Channel 06/0 : 87[7] -> 79[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO Channel 07/0 : 49[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96568:96753 [5] NCCL INFO Channel 03/0 : 45[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153564:153737 [2] NCCL INFO Channel 09/0 : 90[2] -> 74[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114515:114690 [3] NCCL INFO Channel 12/0 : 67[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO Channel 14/0 : 96[0] -> 80[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235653:235825 [5] NCCL INFO Channel 06/0 : 117[5] -> 109[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96895:97069 [7] NCCL INFO Channel 09/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO Channel 11/0 : 88[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO Channel 11/0 : 81[1] -> 73[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO Channel 09/0 : 53[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96565:96754 [2] NCCL INFO Channel 00/0 : 50[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153565:153738 [3] NCCL INFO Channel 15/0 : 91[3] -> 75[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114515:114690 [3] NCCL INFO Channel 14/0 : 67[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO Channel 00/0 : 118[6] -> 102[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235655:235828 [7] NCCL INFO Channel 11/0 : 119[7] -> 111[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272152:272331 [2] NCCL INFO Channel 00/0 : 122[2] -> 114[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96893:97067 [5] NCCL INFO Channel 11/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO Channel 04/0 : 59[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO Channel 07/0 : 85[5] -> 77[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96567:96760 [4] NCCL INFO Channel 02/0 : 52[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114518:114693 [6] NCCL INFO Channel 01/0 : 78[6] -> 70[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO Channel 00/0 : 115[3] -> 99[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235653:235825 [5] NCCL INFO Channel 07/0 : 117[5] -> 109[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272154:272330 [4] NCCL INFO Channel 00/0 : 124[4] -> 116[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96889:97065 [1] NCCL INFO Channel 12/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96895:97069 [7] NCCL INFO Channel 10/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO Channel 10/0 : 58[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO Channel 15/0 : 92[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32706:32877 [2] NCCL INFO Channel 05/0 : 26[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO Channel 08/0 : 87[7] -> 79[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96570:96759 [7] NCCL INFO Channel 05/0 : 47[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO Channel 02/0 : 118[6] -> 102[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272155:272329 [5] NCCL INFO Channel 11/0 : 117[5] -> 109[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96893:97067 [5] NCCL INFO Channel 12/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO Channel 03/0 : 62[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO Channel 10/0 : 55[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO Channel 08/0 : 49[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96564:96757 [1] NCCL INFO Channel 03/0 : 41[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153564:153737 [2] NCCL INFO Channel 11/0 : 90[2] -> 74[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235655:235828 [7] NCCL INFO Channel 12/0 : 119[7] -> 111[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272157:272324 [7] NCCL INFO Channel 09/0 : 119[7] -> 111[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO Channel 13/0 : 88[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO Channel 06/0 : 59[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO Channel 00/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO Channel 10/0 : 53[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO Channel 00/0 : 58[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96568:96753 [5] NCCL INFO Channel 05/0 : 45[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153568:153736 [6] NCCL INFO Channel 01/0 : 110[6] -> 94[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114518:114693 [6] NCCL INFO Channel 03/0 : 78[6] -> 70[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235653:235825 [5] NCCL INFO Channel 08/0 : 117[5] -> 109[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272151:272326 [1] NCCL INFO Channel 12/0 : 113[1] -> 105[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO Channel 12/0 : 58[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32707:32878 [3] NCCL INFO Channel 00/0 : 27[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO Channel 08/0 : 85[5] -> 77[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO Channel 11/0 : 55[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96565:96754 [2] NCCL INFO Channel 04/0 : 50[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153564:153737 [2] NCCL INFO Channel 13/0 : 90[2] -> 74[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114515:114690 [3] NCCL INFO Channel 01/0 : 75[3] -> 67[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO Channel 01/0 : 106[2] -> 98[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235655:235828 [7] NCCL INFO Channel 13/0 : 119[7] -> 111[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272152:272331 [2] NCCL INFO Channel 04/0 : 122[2] -> 114[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96889:97065 [1] NCCL INFO Channel 13/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO Channel 05/0 : 62[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32706:32877 [2] NCCL INFO Channel 07/0 : 26[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO Channel 12/0 : 81[1] -> 73[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO Channel 09/0 : 87[7] -> 79[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96570:96759 [7] NCCL INFO Channel 07/0 : 47[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153568:153736 [6] NCCL INFO Channel 03/0 : 110[6] -> 94[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114518:114693 [6] NCCL INFO Channel 05/0 : 78[6] -> 70[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO Channel 01/0 : 108[4] -> 100[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO Channel 04/0 : 118[6] -> 102[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO Channel 03/0 : 106[2] -> 98[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO Channel 02/0 : 115[3] -> 99[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235653:235825 [5] NCCL INFO Channel 09/0 : 117[5] -> 109[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96895:97069 [7] NCCL INFO Channel 11/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96893:97067 [5] NCCL INFO Channel 14/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO Channel 08/0 : 59[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO Channel 15/0 : 88[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO Channel 01/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32710:32883 [6] NCCL INFO Channel 01/0 : 46[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO Channel 09/0 : 85[5] -> 77[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO Channel 10/0 : 87[7] -> 79[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96567:96760 [4] NCCL INFO Channel 06/0 : 52[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114515:114690 [3] NCCL INFO Channel 03/0 : 75[3] -> 67[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272155:272329 [5] NCCL INFO Channel 12/0 : 117[5] -> 109[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96890:97070 [2] NCCL INFO Channel 00/0 : 26[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO Channel 14/0 : 58[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO Channel 13/0 : 81[1] -> 73[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO Channel 00/0 : 92[4] -> 84[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO Channel 10/0 : 49[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96568:96753 [5] NCCL INFO Channel 07/0 : 45[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96564:96757 [1] NCCL INFO Channel 05/0 : 41[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114518:114693 [6] NCCL INFO Channel 07/0 : 78[6] -> 70[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO Channel 05/0 : 106[2] -> 98[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO Channel 03/0 : 108[4] -> 100[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272157:272324 [7] NCCL INFO Channel 10/0 : 119[7] -> 111[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO Channel 07/0 : 62[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO Channel 10/0 : 59[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO Channel 01/0 : 56[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO Channel 11/0 : 87[7] -> 79[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO Channel 10/0 : 85[5] -> 77[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO Channel 11/0 : 53[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO Channel 00/0 : 60[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96565:96754 [2] NCCL INFO Channel 06/0 : 50[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96563:96756 [0] NCCL INFO Channel 02/0 : 48[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153562:153735 [0] NCCL INFO Channel 01/0 : 104[0] -> 88[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153564:153737 [2] NCCL INFO Channel 15/0 : 90[2] -> 74[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114515:114690 [3] NCCL INFO Channel 05/0 : 75[3] -> 67[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO Channel 01/0 : 104[0] -> 96[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO Channel 08/0 : 118[6] -> 102[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO Channel 04/0 : 115[3] -> 99[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235655:235828 [7] NCCL INFO Channel 14/0 : 119[7] -> 111[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272154:272330 [4] NCCL INFO Channel 02/0 : 124[4] -> 116[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272151:272326 [1] NCCL INFO Channel 13/0 : 113[1] -> 105[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO Channel 09/0 : 62[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO Channel 03/0 : 56[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO Channel 02/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32707:32878 [3] NCCL INFO Channel 02/0 : 27[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO Channel 00/0 : 90[2] -> 82[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO Channel 04/0 : 58[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96570:96759 [7] NCCL INFO Channel 09/0 : 47[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153568:153736 [6] NCCL INFO Channel 05/0 : 110[6] -> 94[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235653:235825 [5] NCCL INFO Channel 10/0 : 117[5] -> 109[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272152:272331 [2] NCCL INFO Channel 06/0 : 122[2] -> 114[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96892:97064 [4] NCCL INFO Channel 00/0 : 28[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96889:97065 [1] NCCL INFO Channel 14/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO Channel 12/0 : 59[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32706:32877 [2] NCCL INFO Channel 09/0 : 26[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO Channel 12/0 : 55[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96567:96760 [4] NCCL INFO Channel 08/0 : 52[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153562:153735 [0] NCCL INFO Channel 03/0 : 104[0] -> 88[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235655:235828 [7] NCCL INFO Channel 01/0 : 111[7] -> 103[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272150:272327 [0] NCCL INFO Channel 02/0 : 120[0] -> 112[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272155:272329 [5] NCCL INFO Channel 14/0 : 117[5] -> 109[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96895:97069 [7] NCCL INFO Channel 12/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO Channel 05/0 : 56[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO Channel 11/0 : 62[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO Channel 14/0 : 59[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO Channel 04/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32710:32883 [6] NCCL INFO Channel 03/0 : 46[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96568:96753 [5] NCCL INFO Channel 09/0 : 45[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114518:114693 [6] NCCL INFO Channel 09/0 : 78[6] -> 70[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235651:235823 [3] NCCL INFO Channel 00/0 : 115[3] -> 107[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272157:272324 [7] NCCL INFO Channel 11/0 : 119[7] -> 111[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96893:97067 [5] NCCL INFO Channel 15/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO Channel 07/0 : 56[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO Channel 13/0 : 62[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32707:32878 [3] NCCL INFO Channel 04/0 : 27[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32704:32881 [0] NCCL INFO Channel 01/0 : 40[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO Channel 14/0 : 81[1] -> 73[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96564:96757 [1] NCCL INFO Channel 07/0 : 41[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114515:114690 [3] NCCL INFO Channel 07/0 : 75[3] -> 67[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272151:272326 [1] NCCL INFO Channel 14/0 : 113[1] -> 105[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96890:97070 [2] NCCL INFO Channel 04/0 : 26[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO Channel 09/0 : 56[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32706:32877 [2] NCCL INFO Channel 11/0 : 26[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO Channel 12/0 : 87[7] -> 79[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO Channel 11/0 : 49[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96565:96754 [2] NCCL INFO Channel 08/0 : 50[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153568:153736 [6] NCCL INFO Channel 07/0 : 110[6] -> 94[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114518:114693 [6] NCCL INFO Channel 11/0 : 78[6] -> 70[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235655:235828 [7] NCCL INFO Channel 03/0 : 111[7] -> 103[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235653:235825 [5] NCCL INFO Channel 11/0 : 117[5] -> 109[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272154:272330 [4] NCCL INFO Channel 06/0 : 124[4] -> 116[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96892:97064 [4] NCCL INFO Channel 02/0 : 28[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96889:97065 [1] NCCL INFO Channel 15/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO Channel 15/0 : 62[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO Channel 05/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO Channel 11/0 : 85[5] -> 77[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO Channel 02/0 : 92[4] -> 84[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO Channel 12/0 : 53[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96563:96756 [0] NCCL INFO Channel 04/0 : 48[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153565:153738 [3] NCCL INFO Channel 00/0 : 91[3] -> 83[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114515:114690 [3] NCCL INFO Channel 09/0 : 75[3] -> 67[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235651:235823 [3] NCCL INFO Channel 01/0 : 115[3] -> 107[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96895:97069 [7] NCCL INFO Channel 13/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO Channel 11/0 : 56[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO Channel 04/0 : 90[2] -> 82[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO Channel 02/0 : 60[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96570:96759 [7] NCCL INFO Channel 11/0 : 47[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153568:153736 [6] NCCL INFO Channel 09/0 : 110[6] -> 94[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153562:153735 [0] NCCL INFO Channel 05/0 : 104[0] -> 88[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114518:114693 [6] NCCL INFO Channel 13/0 : 78[6] -> 70[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235655:235828 [7] NCCL INFO Channel 05/0 : 111[7] -> 103[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96890:97070 [2] NCCL INFO Channel 06/0 : 26[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO Channel 13/0 : 56[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32710:32883 [6] NCCL INFO Channel 05/0 : 46[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO Channel 15/0 : 81[1] -> 73[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO Channel 13/0 : 55[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO Channel 06/0 : 58[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96567:96760 [4] NCCL INFO Channel 10/0 : 52[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153565:153738 [3] NCCL INFO Channel 02/0 : 91[3] -> 83[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114515:114690 [3] NCCL INFO Channel 13/0 : 75[3] -> 67[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235653:235825 [5] NCCL INFO Channel 12/0 : 117[5] -> 109[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272155:272329 [5] NCCL INFO Channel 15/0 : 117[5] -> 109[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96888:97066 [0] NCCL INFO Channel 02/0 : 24[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO Channel 01/0 : 60[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32704:32881 [0] NCCL INFO Channel 03/0 : 40[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32707:32878 [3] NCCL INFO Channel 06/0 : 27[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32706:32877 [2] NCCL INFO Channel 13/0 : 26[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO Channel 02/0 : 88[0] -> 80[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO Channel 12/0 : 49[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96568:96753 [5] NCCL INFO Channel 11/0 : 45[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235650:235824 [2] NCCL INFO Channel 00/0 : 114[2] -> 106[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272152:272331 [2] NCCL INFO Channel 08/0 : 122[2] -> 114[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32706:32877 [2] NCCL INFO Channel 15/0 : 26[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32710:32883 [6] NCCL INFO Channel 07/0 : 46[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO Channel 13/0 : 87[7] -> 79[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO Channel 14/0 : 53[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96564:96757 [1] NCCL INFO Channel 11/0 : 41[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272157:272324 [7] NCCL INFO Channel 12/0 : 119[7] -> 111[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32704:32881 [0] NCCL INFO Channel 05/0 : 40[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO Channel 12/0 : 85[5] -> 77[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO Channel 02/0 : 56[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96565:96754 [2] NCCL INFO Channel 10/0 : 50[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272150:272327 [0] NCCL INFO Channel 04/0 : 120[0] -> 112[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272151:272326 [1] NCCL INFO Channel 15/0 : 113[1] -> 105[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96892:97064 [4] NCCL INFO Channel 06/0 : 28[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96895:97069 [7] NCCL INFO Channel 14/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32707:32878 [3] NCCL INFO Channel 08/0 : 27[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO Channel 06/0 : 92[4] -> 84[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO Channel 06/0 : 60[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96570:96759 [7] NCCL INFO Channel 13/0 : 47[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153568:153736 [6] NCCL INFO Channel 11/0 : 110[6] -> 94[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153562:153735 [0] NCCL INFO Channel 07/0 : 104[0] -> 88[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235651:235823 [3] NCCL INFO Channel 02/0 : 115[3] -> 107[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272154:272330 [4] NCCL INFO Channel 08/0 : 124[4] -> 116[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96888:97066 [0] NCCL INFO Channel 04/0 : 24[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96890:97070 [2] NCCL INFO Channel 08/0 : 26[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96892:97064 [4] NCCL INFO Channel 08/0 : 28[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO Channel 15/0 : 56[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO Channel 06/0 : 90[2] -> 82[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO Channel 14/0 : 55[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96563:96756 [0] NCCL INFO Channel 06/0 : 48[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153565:153738 [3] NCCL INFO Channel 04/0 : 91[3] -> 83[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235655:235828 [7] NCCL INFO Channel 07/0 : 111[7] -> 103[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272152:272331 [2] NCCL INFO Channel 10/0 : 122[2] -> 114[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96890:97070 [2] NCCL INFO Channel 10/0 : 26[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96888:97066 [0] NCCL INFO Channel 06/0 : 24[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96892:97064 [4] NCCL INFO Channel 10/0 : 28[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO Channel 03/0 : 60[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32704:32881 [0] NCCL INFO Channel 07/0 : 40[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32707:32878 [3] NCCL INFO Channel 10/0 : 27[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96568:96753 [5] NCCL INFO Channel 15/0 : 45[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96567:96760 [4] NCCL INFO Channel 12/0 : 52[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153564:153737 [2] NCCL INFO Channel 00/0 : 90[2] -> 82[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235653:235825 [5] NCCL INFO Channel 14/0 : 117[5] -> 109[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272157:272324 [7] NCCL INFO Channel 13/0 : 119[7] -> 111[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO Channel 05/0 : 60[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32710:32883 [6] NCCL INFO Channel 09/0 : 46[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96564:96757 [1] NCCL INFO Channel 13/0 : 41[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153568:153736 [6] NCCL INFO Channel 13/0 : 110[6] -> 94[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153562:153735 [0] NCCL INFO Channel 09/0 : 104[0] -> 88[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235650:235824 [2] NCCL INFO Channel 01/0 : 114[2] -> 106[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272150:272327 [0] NCCL INFO Channel 06/0 : 120[0] -> 112[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO Channel 00/0 : 62[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO Channel 06/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO Channel 14/0 : 87[7] -> 79[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO Channel 04/0 : 88[0] -> 80[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO Channel 14/0 : 85[5] -> 77[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96565:96754 [2] NCCL INFO Channel 12/0 : 50[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153565:153738 [3] NCCL INFO Channel 06/0 : 91[3] -> 83[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62884:63061 [3] NCCL INFO Channel 00/0 : 51[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235655:235828 [7] NCCL INFO Channel 09/0 : 111[7] -> 103[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272154:272330 [4] NCCL INFO Channel 10/0 : 124[4] -> 116[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96890:97070 [2] NCCL INFO Channel 12/0 : 26[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96888:97066 [0] NCCL INFO Channel 08/0 : 24[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96892:97064 [4] NCCL INFO Channel 12/0 : 28[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO Channel 07/0 : 60[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO Channel 07/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO Channel 08/0 : 92[4] -> 84[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO Channel 08/0 : 90[2] -> 82[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO Channel 08/0 : 58[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96563:96756 [0] NCCL INFO Channel 08/0 : 48[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235653:235825 [5] NCCL INFO Channel 15/0 : 117[5] -> 109[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235651:235823 [3] NCCL INFO Channel 04/0 : 115[3] -> 107[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272157:272324 [7] NCCL INFO Channel 14/0 : 119[7] -> 111[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO Channel 02/0 : 56[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO Channel 02/0 : 62[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO Channel 09/0 : 60[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32704:32881 [0] NCCL INFO Channel 09/0 : 40[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO Channel 15/0 : 85[5] -> 77[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO Channel 13/0 : 49[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96564:96757 [1] NCCL INFO Channel 15/0 : 41[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272152:272331 [2] NCCL INFO Channel 12/0 : 122[2] -> 114[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO Channel 04/0 : 56[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO Channel 04/0 : 62[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO Channel 11/0 : 60[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32707:32878 [3] NCCL INFO Channel 12/0 : 27[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32710:32883 [6] NCCL INFO Channel 11/0 : 46[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO Channel 06/0 : 88[0] -> 80[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO Channel 15/0 : 53[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96567:96760 [4] NCCL INFO Channel 14/0 : 52[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96565:96754 [2] NCCL INFO Channel 14/0 : 50[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153564:153737 [2] NCCL INFO Channel 04/0 : 90[2] -> 82[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272150:272327 [0] NCCL INFO Channel 08/0 : 120[0] -> 112[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96888:97066 [0] NCCL INFO Channel 10/0 : 24[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO Channel 06/0 : 56[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO Channel 08/0 : 62[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO Channel 08/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO Channel 10/0 : 90[2] -> 82[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO Channel 08/0 : 60[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO Channel 04/0 : 56[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96563:96756 [0] NCCL INFO Channel 10/0 : 48[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153562:153735 [0] NCCL INFO Channel 11/0 : 104[0] -> 88[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235650:235824 [2] NCCL INFO Channel 03/0 : 114[2] -> 106[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO Channel 13/0 : 60[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32704:32881 [0] NCCL INFO Channel 11/0 : 40[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO Channel 10/0 : 92[4] -> 84[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO Channel 10/0 : 58[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96565:96754 [2] NCCL INFO Channel 00/0 : 34[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153568:153736 [6] NCCL INFO Channel 15/0 : 110[6] -> 94[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235655:235828 [7] NCCL INFO Channel 11/0 : 111[7] -> 103[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO Channel 08/0 : 56[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32707:32878 [3] NCCL INFO Channel 14/0 : 27[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO Channel 08/0 : 88[0] -> 80[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO Channel 14/0 : 49[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96567:96760 [4] NCCL INFO Channel 00/0 : 36[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153564:153737 [2] NCCL INFO Channel 06/0 : 90[2] -> 82[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153565:153738 [3] NCCL INFO Channel 08/0 : 91[3] -> 83[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62884:63061 [3] NCCL INFO Channel 01/0 : 51[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235653:235825 [5] NCCL INFO Channel 01/0 : 109[5] -> 101[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272154:272330 [4] NCCL INFO Channel 12/0 : 124[4] -> 116[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96892:97064 [4] NCCL INFO Channel 14/0 : 28[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96890:97070 [2] NCCL INFO Channel 14/0 : 26[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO Channel 10/0 : 62[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO Channel 15/0 : 60[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO Channel 09/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32710:32883 [6] NCCL INFO Channel 13/0 : 46[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO Channel 12/0 : 90[2] -> 82[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO Channel 12/0 : 92[4] -> 84[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO Channel 10/0 : 60[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96563:96756 [0] NCCL INFO Channel 12/0 : 48[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153562:153735 [0] NCCL INFO Channel 13/0 : 104[0] -> 88[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235651:235823 [3] NCCL INFO Channel 05/0 : 115[3] -> 107[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272152:272331 [2] NCCL INFO Channel 14/0 : 122[2] -> 114[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO Channel 10/0 : 56[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32704:32881 [0] NCCL INFO Channel 13/0 : 40[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO Channel 06/0 : 56[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96565:96754 [2] NCCL INFO Channel 04/0 : 34[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62883:63058 [2] NCCL INFO Channel 00/0 : 50[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235650:235824 [2] NCCL INFO Channel 04/0 : 114[2] -> 106[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272150:272327 [0] NCCL INFO Channel 10/0 : 120[0] -> 112[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272154:272330 [4] NCCL INFO Channel 14/0 : 124[4] -> 116[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO Channel 12/0 : 62[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO Channel 15/0 : 49[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO Channel 12/0 : 58[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96567:96760 [4] NCCL INFO Channel 02/0 : 36[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235655:235828 [7] NCCL INFO Channel 13/0 : 111[7] -> 103[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272152:272331 [2] NCCL INFO Channel 00/0 : 114[2] -> 106[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96888:97066 [0] NCCL INFO Channel 12/0 : 24[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO Channel 10/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO Channel 10/0 : 88[0] -> 80[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96563:96756 [0] NCCL INFO Channel 14/0 : 48[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153568:153736 [6] NCCL INFO Channel 01/0 : 94[6] -> 78[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235653:235825 [5] NCCL INFO Channel 03/0 : 109[5] -> 101[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272154:272330 [4] NCCL INFO Channel 00/0 : 116[4] -> 108[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96892:97064 [4] NCCL INFO Channel 00/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96890:97070 [2] NCCL INFO Channel 00/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32710:32883 [6] NCCL INFO Channel 15/0 : 46[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO Channel 14/0 : 92[4] -> 84[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO Channel 14/0 : 90[2] -> 82[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96565:96754 [2] NCCL INFO Channel 06/0 : 34[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153564:153737 [2] NCCL INFO Channel 08/0 : 90[2] -> 82[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62883:63058 [2] NCCL INFO Channel 01/0 : 50[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235651:235823 [3] NCCL INFO Channel 06/0 : 115[3] -> 107[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272150:272327 [0] NCCL INFO Channel 12/0 : 120[0] -> 112[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO Channel 12/0 : 56[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32704:32881 [0] NCCL INFO Channel 15/0 : 40[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO Channel 12/0 : 88[0] -> 80[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96567:96760 [4] NCCL INFO Channel 06/0 : 36[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153565:153738 [3] NCCL INFO Channel 10/0 : 91[3] -> 83[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62884:63061 [3] NCCL INFO Channel 02/0 : 51[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235650:235824 [2] NCCL INFO Channel 05/0 : 114[2] -> 106[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272152:272331 [2] NCCL INFO Channel 01/0 : 114[2] -> 106[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO Channel 14/0 : 62[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO Channel 12/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32710:32883 [6] NCCL INFO Channel 01/0 : 30[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32704:32881 [0] NCCL INFO Channel 01/0 : 24[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO Channel 00/0 : 84[4] -> 76[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO Channel 00/0 : 82[2] -> 74[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO Channel 12/0 : 60[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96563:96756 [0] NCCL INFO Channel 02/0 : 32[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153562:153735 [0] NCCL INFO Channel 15/0 : 104[0] -> 88[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272154:272330 [4] NCCL INFO Channel 01/0 : 116[4] -> 108[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96888:97066 [0] NCCL INFO Channel 14/0 : 24[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO Channel 14/0 : 56[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO Channel 14/0 : 88[0] -> 80[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO Channel 08/0 : 56[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO Channel 14/0 : 58[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96565:96754 [2] NCCL INFO Channel 08/0 : 34[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153566:153741 [4] NCCL INFO Channel 01/0 : 108[4] -> 92[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62883:63058 [2] NCCL INFO Channel 03/0 : 50[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272152:272331 [2] NCCL INFO Channel 03/0 : 114[2] -> 106[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272150:272327 [0] NCCL INFO Channel 14/0 : 120[0] -> 112[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96892:97064 [4] NCCL INFO Channel 01/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO Channel 00/0 : 60[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO Channel 14/0 : 60[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96567:96760 [4] NCCL INFO Channel 08/0 : 36[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153568:153736 [6] NCCL INFO Channel 03/0 : 94[6] -> 78[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62884:63061 [3] NCCL INFO Channel 04/0 : 51[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235653:235825 [5] NCCL INFO Channel 05/0 : 109[5] -> 101[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96890:97070 [2] NCCL INFO Channel 01/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO Channel 02/0 : 60[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO Channel 13/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32710:32883 [6] NCCL INFO Channel 03/0 : 30[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32704:32881 [0] NCCL INFO Channel 03/0 : 24[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO Channel 00/0 : 50[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96563:96756 [0] NCCL INFO Channel 04/0 : 32[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96565:96754 [2] NCCL INFO Channel 10/0 : 34[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153565:153738 [3] NCCL INFO Channel 12/0 : 91[3] -> 83[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153564:153737 [2] NCCL INFO Channel 10/0 : 90[2] -> 82[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235651:235823 [3] NCCL INFO Channel 07/0 : 115[3] -> 107[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272154:272330 [4] NCCL INFO Channel 02/0 : 116[4] -> 108[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO Channel 01/0 : 84[4] -> 76[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO Channel 10/0 : 56[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96567:96760 [4] NCCL INFO Channel 10/0 : 36[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153562:153735 [0] NCCL INFO Channel 01/0 : 88[0] -> 72[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62883:63058 [2] NCCL INFO Channel 04/0 : 50[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235650:235824 [2] NCCL INFO Channel 06/0 : 114[2] -> 106[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272152:272331 [2] NCCL INFO Channel 04/0 : 114[2] -> 106[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272150:272327 [0] NCCL INFO Channel 01/0 : 112[0] -> 104[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96888:97066 [0] NCCL INFO Channel 01/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96892:97064 [4] NCCL INFO Channel 02/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO Channel 06/0 : 60[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO Channel 14/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO Channel 01/0 : 82[2] -> 74[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO Channel 00/0 : 52[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96565:96754 [2] NCCL INFO Channel 12/0 : 34[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96563:96756 [0] NCCL INFO Channel 06/0 : 32[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153566:153741 [4] NCCL INFO Channel 03/0 : 108[4] -> 92[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62884:63061 [3] NCCL INFO Channel 05/0 : 51[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235653:235825 [5] NCCL INFO Channel 07/0 : 109[5] -> 101[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272154:272330 [4] NCCL INFO Channel 03/0 : 116[4] -> 108[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96890:97070 [2] NCCL INFO Channel 03/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO Channel 08/0 : 60[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO Channel 15/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32704:32881 [0] NCCL INFO Channel 05/0 : 24[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32710:32883 [6] NCCL INFO Channel 05/0 : 30[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO Channel 01/0 : 80[0] -> 72[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO Channel 01/0 : 50[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96567:96760 [4] NCCL INFO Channel 12/0 : 36[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153568:153736 [6] NCCL INFO Channel 05/0 : 94[6] -> 78[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62883:63058 [2] NCCL INFO Channel 05/0 : 50[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235651:235823 [3] NCCL INFO Channel 08/0 : 115[3] -> 107[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235650:235824 [2] NCCL INFO Channel 07/0 : 114[2] -> 106[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272152:272331 [2] NCCL INFO Channel 05/0 : 114[2] -> 106[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272150:272327 [0] NCCL INFO Channel 02/0 : 112[0] -> 104[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96888:97066 [0] NCCL INFO Channel 02/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO Channel 10/0 : 60[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO Channel 02/0 : 84[4] -> 76[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO Channel 12/0 : 56[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96565:96754 [2] NCCL INFO Channel 14/0 : 34[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235653:235825 [5] NCCL INFO Channel 09/0 : 109[5] -> 101[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272154:272330 [4] NCCL INFO Channel 05/0 : 116[4] -> 108[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96892:97064 [4] NCCL INFO Channel 03/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96890:97070 [2] NCCL INFO Channel 04/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32704:32881 [0] NCCL INFO Channel 07/0 : 24[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO Channel 03/0 : 82[2] -> 74[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO Channel 02/0 : 80[0] -> 72[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO Channel 01/0 : 52[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO Channel 03/0 : 50[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96563:96756 [0] NCCL INFO Channel 08/0 : 32[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235651:235823 [3] NCCL INFO Channel 09/0 : 115[3] -> 107[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272152:272331 [2] NCCL INFO Channel 06/0 : 114[2] -> 106[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO Channel 01/0 : 11[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32710:32883 [6] NCCL INFO Channel 07/0 : 30[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO Channel 03/0 : 84[4] -> 76[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO Channel 14/0 : 56[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96567:96760 [4] NCCL INFO Channel 14/0 : 36[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153565:153738 [3] NCCL INFO Channel 14/0 : 91[3] -> 83[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153564:153737 [2] NCCL INFO Channel 12/0 : 90[2] -> 82[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62884:63061 [3] NCCL INFO Channel 06/0 : 51[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235653:235825 [5] NCCL INFO Channel 11/0 : 109[5] -> 101[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235650:235824 [2] NCCL INFO Channel 08/0 : 114[2] -> 106[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272150:272327 [0] NCCL INFO Channel 03/0 : 112[0] -> 104[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO Channel 12/0 : 60[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32709:32882 [5] NCCL INFO Channel 10/0 : 29[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96563:96756 [0] NCCL INFO Channel 10/0 : 32[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153562:153735 [0] NCCL INFO Channel 03/0 : 88[0] -> 72[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62883:63058 [2] NCCL INFO Channel 06/0 : 50[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272154:272330 [4] NCCL INFO Channel 06/0 : 116[4] -> 108[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96888:97066 [0] NCCL INFO Channel 03/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO Channel 03/0 : 11[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153568:153736 [6] NCCL INFO Channel 07/0 : 94[6] -> 78[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153566:153741 [4] NCCL INFO Channel 05/0 : 108[4] -> 92[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62884:63061 [3] NCCL INFO Channel 07/0 : 51[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96890:97070 [2] NCCL INFO Channel 05/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96892:97064 [4] NCCL INFO Channel 05/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO Channel 04/0 : 82[2] -> 74[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO Channel 04/0 : 50[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153564:153737 [2] NCCL INFO Channel 14/0 : 90[2] -> 82[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153562:153735 [0] NCCL INFO Channel 05/0 : 88[0] -> 72[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235651:235823 [3] NCCL INFO Channel 10/0 : 115[3] -> 107[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96888:97066 [0] NCCL INFO Channel 04/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32704:32881 [0] NCCL INFO Channel 09/0 : 24[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO Channel 03/0 : 80[0] -> 72[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO Channel 01/0 : 48[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153568:153736 [6] NCCL INFO Channel 09/0 : 94[6] -> 78[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235653:235825 [5] NCCL INFO Channel 15/0 : 109[5] -> 101[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235650:235824 [2] NCCL INFO Channel 09/0 : 114[2] -> 106[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272152:272331 [2] NCCL INFO Channel 07/0 : 114[2] -> 106[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO Channel 05/0 : 11[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32710:32883 [6] NCCL INFO Channel 09/0 : 30[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32709:32882 [5] NCCL INFO Channel 12/0 : 29[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO Channel 05/0 : 84[4] -> 76[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO Channel 05/0 : 50[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96563:96756 [0] NCCL INFO Channel 12/0 : 32[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153566:153741 [4] NCCL INFO Channel 07/0 : 108[4] -> 92[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62883:63058 [2] NCCL INFO Channel 07/0 : 50[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235651:235823 [3] NCCL INFO Channel 12/0 : 115[3] -> 107[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272150:272327 [0] NCCL INFO Channel 04/0 : 112[0] -> 104[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO Channel 05/0 : 82[2] -> 74[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO Channel 04/0 : 80[0] -> 72[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO Channel 02/0 : 48[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96563:96756 [0] NCCL INFO Channel 14/0 : 32[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96569:96758 [6] NCCL INFO Channel 00/0 : 54[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153562:153735 [0] NCCL INFO Channel 07/0 : 88[0] -> 72[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62884:63061 [3] NCCL INFO Channel 08/0 : 51[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235650:235824 [2] NCCL INFO Channel 11/0 : 114[2] -> 106[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272154:272330 [4] NCCL INFO Channel 07/0 : 116[4] -> 108[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96892:97064 [4] NCCL INFO Channel 06/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96890:97070 [2] NCCL INFO Channel 06/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96888:97066 [0] NCCL INFO Channel 05/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO Channel 06/0 : 84[4] -> 76[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO Channel 06/0 : 50[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO Channel 03/0 : 48[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96569:96758 [6] NCCL INFO Channel 02/0 : 54[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153568:153736 [6] NCCL INFO Channel 11/0 : 94[6] -> 78[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62883:63058 [2] NCCL INFO Channel 08/0 : 50[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235651:235823 [3] NCCL INFO Channel 13/0 : 115[3] -> 107[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272152:272331 [2] NCCL INFO Channel 08/0 : 114[2] -> 106[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO Channel 07/0 : 11[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32706:32877 [2] NCCL INFO Channel 00/0 : 26[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32704:32881 [0] NCCL INFO Channel 11/0 : 24[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO Channel 05/0 : 80[0] -> 72[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO Channel 07/0 : 50[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96566:96755 [3] NCCL INFO Channel 00/0 : 51[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153566:153741 [4] NCCL INFO Channel 09/0 : 108[4] -> 92[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235650:235824 [2] NCCL INFO Channel 12/0 : 114[2] -> 106[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272150:272327 [0] NCCL INFO Channel 05/0 : 112[0] -> 104[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32710:32883 [6] NCCL INFO Channel 11/0 : 30[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32709:32882 [5] NCCL INFO Channel 14/0 : 29[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO Channel 06/0 : 82[2] -> 74[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO Channel 07/0 : 84[4] -> 76[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO Channel 04/0 : 48[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96565:96754 [2] NCCL INFO Channel 01/0 : 42[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153562:153735 [0] NCCL INFO Channel 09/0 : 88[0] -> 72[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62883:63058 [2] NCCL INFO Channel 09/0 : 50[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272154:272330 [4] NCCL INFO Channel 08/0 : 116[4] -> 108[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96890:97070 [2] NCCL INFO Channel 07/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32711:32884 [7] NCCL INFO Channel 04/0 : 31[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96569:96758 [6] NCCL INFO Channel 04/0 : 54[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153568:153736 [6] NCCL INFO Channel 13/0 : 94[6] -> 78[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62884:63061 [3] NCCL INFO Channel 09/0 : 51[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272152:272331 [2] NCCL INFO Channel 09/0 : 114[2] -> 106[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96892:97064 [4] NCCL INFO Channel 07/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96888:97066 [0] NCCL INFO Channel 06/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO Channel 00/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96566:96755 [3] NCCL INFO Channel 02/0 : 51[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153566:153741 [4] NCCL INFO Channel 11/0 : 108[4] -> 92[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62883:63058 [2] NCCL INFO Channel 11/0 : 50[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235650:235824 [2] NCCL INFO Channel 13/0 : 114[2] -> 106[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235651:235823 [3] NCCL INFO Channel 14/0 : 115[3] -> 107[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272150:272327 [0] NCCL INFO Channel 06/0 : 112[0] -> 104[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO Channel 09/0 : 11[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO Channel 06/0 : 80[0] -> 72[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO Channel 08/0 : 50[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96567:96760 [4] NCCL INFO Channel 01/0 : 44[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235650:235824 [2] NCCL INFO Channel 14/0 : 114[2] -> 106[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235651:235823 [3] NCCL INFO Channel 15/0 : 115[3] -> 107[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272154:272330 [4] NCCL INFO Channel 09/0 : 116[4] -> 108[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32704:32881 [0] NCCL INFO Channel 13/0 : 24[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32706:32877 [2] NCCL INFO Channel 04/0 : 26[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO Channel 08/0 : 84[4] -> 76[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO Channel 07/0 : 82[2] -> 74[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO Channel 05/0 : 48[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96565:96754 [2] NCCL INFO Channel 03/0 : 42[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235651:235823 [3] NCCL INFO Channel 01/0 : 107[3] -> 99[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235650:235824 [2] NCCL INFO Channel 15/0 : 114[2] -> 106[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272152:272331 [2] NCCL INFO Channel 11/0 : 114[2] -> 106[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272150:272327 [0] NCCL INFO Channel 07/0 : 112[0] -> 104[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96890:97070 [2] NCCL INFO Channel 08/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32710:32883 [6] NCCL INFO Channel 13/0 : 30[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO Channel 07/0 : 80[0] -> 72[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO Channel 09/0 : 50[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96569:96758 [6] NCCL INFO Channel 08/0 : 54[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153562:153735 [0] NCCL INFO Channel 11/0 : 88[0] -> 72[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62884:63061 [3] NCCL INFO Channel 10/0 : 51[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96892:97064 [4] NCCL INFO Channel 08/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96888:97066 [0] NCCL INFO Channel 07/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO Channel 01/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32711:32884 [7] NCCL INFO Channel 06/0 : 31[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO Channel 09/0 : 84[4] -> 76[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO Channel 06/0 : 48[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96566:96755 [3] NCCL INFO Channel 04/0 : 51[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153568:153736 [6] NCCL INFO Channel 15/0 : 94[6] -> 78[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62883:63058 [2] NCCL INFO Channel 12/0 : 50[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235651:235823 [3] NCCL INFO Channel 03/0 : 107[3] -> 99[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96891:97063 [3] NCCL INFO Channel 00/0 : 27[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96890:97070 [2] NCCL INFO Channel 09/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO Channel 13/0 : 11[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32704:32881 [0] NCCL INFO Channel 15/0 : 24[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO Channel 08/0 : 82[2] -> 74[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO Channel 11/0 : 50[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96567:96760 [4] NCCL INFO Channel 03/0 : 44[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96565:96754 [2] NCCL INFO Channel 05/0 : 42[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153566:153741 [4] NCCL INFO Channel 13/0 : 108[4] -> 92[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62884:63061 [3] NCCL INFO Channel 12/0 : 51[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235650:235824 [2] NCCL INFO Channel 01/0 : 106[2] -> 98[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272154:272330 [4] NCCL INFO Channel 10/0 : 116[4] -> 108[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO Channel 03/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32706:32877 [2] NCCL INFO Channel 06/0 : 26[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO Channel 09/0 : 80[0] -> 72[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO Channel 07/0 : 48[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96569:96758 [6] NCCL INFO Channel 10/0 : 54[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153562:153735 [0] NCCL INFO Channel 13/0 : 88[0] -> 72[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272156:272325 [6] NCCL INFO Channel 00/0 : 126[6] -> 118[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272150:272327 [0] NCCL INFO Channel 09/0 : 112[0] -> 104[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96892:97064 [4] NCCL INFO Channel 09/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96888:97066 [0] NCCL INFO Channel 09/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96566:96755 [3] NCCL INFO Channel 06/0 : 51[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96563:96756 [0] NCCL INFO Channel 01/0 : 40[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153562:153735 [0] NCCL INFO Channel 15/0 : 88[0] -> 72[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272152:272331 [2] NCCL INFO Channel 12/0 : 114[2] -> 106[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272154:272330 [4] NCCL INFO Channel 11/0 : 116[4] -> 108[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96890:97070 [2] NCCL INFO Channel 11/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96565:96754 [2] NCCL INFO Channel 07/0 : 42[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96567:96760 [4] NCCL INFO Channel 05/0 : 44[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153566:153741 [4] NCCL INFO Channel 15/0 : 108[4] -> 92[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62883:63058 [2] NCCL INFO Channel 13/0 : 50[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235654:235821 [6] NCCL INFO Channel 00/0 : 118[6] -> 110[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272150:272327 [0] NCCL INFO Channel 10/0 : 112[0] -> 104[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96891:97063 [3] NCCL INFO Channel 02/0 : 27[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32711:32884 [7] NCCL INFO Channel 08/0 : 31[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96569:96758 [6] NCCL INFO Channel 12/0 : 54[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153566:153741 [4] NCCL INFO Channel 01/0 : 92[4] -> 76[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62884:63061 [3] NCCL INFO Channel 13/0 : 51[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235651:235823 [3] NCCL INFO Channel 05/0 : 107[3] -> 99[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235650:235824 [2] NCCL INFO Channel 03/0 : 106[2] -> 98[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272154:272330 [4] NCCL INFO Channel 13/0 : 116[4] -> 108[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272156:272325 [6] NCCL INFO Channel 02/0 : 126[6] -> 118[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272152:272331 [2] NCCL INFO Channel 13/0 : 114[2] -> 106[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96888:97066 [0] NCCL INFO Channel 10/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO Channel 15/0 : 11[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32711:32884 [7] NCCL INFO Channel 10/0 : 31[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO Channel 10/0 : 84[4] -> 76[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO Channel 12/0 : 50[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96563:96756 [0] NCCL INFO Channel 03/0 : 40[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96566:96755 [3] NCCL INFO Channel 08/0 : 51[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153566:153741 [4] NCCL INFO Channel 03/0 : 92[4] -> 76[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62883:63058 [2] NCCL INFO Channel 14/0 : 50[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272153:272328 [3] NCCL INFO Channel 00/0 : 123[3] -> 115[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96892:97064 [4] NCCL INFO Channel 10/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO Channel 04/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32711:32884 [7] NCCL INFO Channel 12/0 : 31[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO Channel 00/0 : 94[6] -> 86[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO Channel 09/0 : 48[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96567:96760 [4] NCCL INFO Channel 07/0 : 44[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235654:235821 [6] NCCL INFO Channel 01/0 : 118[6] -> 110[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272150:272327 [0] NCCL INFO Channel 11/0 : 112[0] -> 104[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96890:97070 [2] NCCL INFO Channel 12/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO Channel 09/0 : 82[2] -> 74[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO Channel 13/0 : 50[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96569:96758 [6] NCCL INFO Channel 14/0 : 54[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235648:235826 [0] NCCL INFO Channel 01/0 : 112[0] -> 104[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96891:97063 [3] NCCL INFO Channel 04/0 : 27[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO Channel 10/0 : 80[0] -> 72[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO Channel 11/0 : 84[4] -> 76[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO Channel 10/0 : 48[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96565:96754 [2] NCCL INFO Channel 09/0 : 42[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153566:153741 [4] NCCL INFO Channel 05/0 : 92[4] -> 76[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62884:63061 [3] NCCL INFO Channel 14/0 : 51[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235650:235824 [2] NCCL INFO Channel 05/0 : 106[2] -> 98[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235651:235823 [3] NCCL INFO Channel 07/0 : 107[3] -> 99[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO Channel 14/0 : 60[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO Channel 05/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32711:32884 [7] NCCL INFO Channel 14/0 : 31[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO Channel 02/0 : 94[6] -> 86[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO Channel 14/0 : 50[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96563:96756 [0] NCCL INFO Channel 05/0 : 40[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96566:96755 [3] NCCL INFO Channel 10/0 : 51[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153568:153736 [6] NCCL INFO Channel 00/0 : 94[6] -> 86[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62883:63058 [2] NCCL INFO Channel 15/0 : 50[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272154:272330 [4] NCCL INFO Channel 14/0 : 116[4] -> 108[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO Channel 06/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO Channel 13/0 : 84[4] -> 76[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO Channel 11/0 : 82[2] -> 74[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO Channel 11/0 : 48[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96567:96760 [4] NCCL INFO Channel 09/0 : 44[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96569:96758 [6] NCCL INFO Channel 00/0 : 38[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153566:153741 [4] NCCL INFO Channel 07/0 : 92[4] -> 76[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62884:63061 [3] NCCL INFO Channel 15/0 : 51[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235654:235821 [6] NCCL INFO Channel 02/0 : 118[6] -> 110[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272152:272331 [2] NCCL INFO Channel 14/0 : 114[2] -> 106[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272156:272325 [6] NCCL INFO Channel 04/0 : 126[6] -> 118[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96888:97066 [0] NCCL INFO Channel 11/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO Channel 11/0 : 80[0] -> 72[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96565:96754 [2] NCCL INFO Channel 11/0 : 42[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153568:153736 [6] NCCL INFO Channel 02/0 : 94[6] -> 86[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62883:63058 [2] NCCL INFO Channel 01/0 : 42[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235648:235826 [0] NCCL INFO Channel 02/0 : 112[0] -> 104[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272150:272327 [0] NCCL INFO Channel 12/0 : 112[0] -> 104[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96892:97064 [4] NCCL INFO Channel 11/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO Channel 00/0 : 91[3] -> 83[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96563:96756 [0] NCCL INFO Channel 07/0 : 40[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96566:96755 [3] NCCL INFO Channel 12/0 : 51[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153562:153735 [0] NCCL INFO Channel 02/0 : 88[0] -> 80[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235650:235824 [2] NCCL INFO Channel 07/0 : 106[2] -> 98[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235651:235823 [3] NCCL INFO Channel 09/0 : 107[3] -> 99[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272153:272328 [3] NCCL INFO Channel 02/0 : 123[3] -> 115[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96890:97070 [2] NCCL INFO Channel 13/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO Channel 07/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96565:96754 [2] NCCL INFO Channel 13/0 : 42[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153566:153741 [4] NCCL INFO Channel 09/0 : 92[4] -> 76[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235654:235821 [6] NCCL INFO Channel 03/0 : 118[6] -> 110[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272154:272330 [4] NCCL INFO Channel 15/0 : 116[4] -> 108[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96891:97063 [3] NCCL INFO Channel 06/0 : 27[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96566:96755 [3] NCCL INFO Channel 14/0 : 51[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153568:153736 [6] NCCL INFO Channel 04/0 : 94[6] -> 86[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62883:63058 [2] NCCL INFO Channel 03/0 : 42[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272152:272331 [2] NCCL INFO Channel 15/0 : 114[2] -> 106[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96888:97066 [0] NCCL INFO Channel 12/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO Channel 04/0 : 94[6] -> 86[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO Channel 04/0 : 87[7] -> 79[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96565:96754 [2] NCCL INFO Channel 15/0 : 42[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153562:153735 [0] NCCL INFO Channel 04/0 : 88[0] -> 80[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62884:63061 [3] NCCL INFO Channel 01/0 : 43[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272150:272327 [0] NCCL INFO Channel 13/0 : 112[0] -> 104[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96892:97064 [4] NCCL INFO Channel 13/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO Channel 08/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO Channel 14/0 : 84[4] -> 76[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO Channel 08/0 : 85[5] -> 77[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96566:96755 [3] NCCL INFO Channel 00/0 : 35[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62883:63058 [2] NCCL INFO Channel 05/0 : 42[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235648:235826 [0] NCCL INFO Channel 03/0 : 112[0] -> 104[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272156:272325 [6] NCCL INFO Channel 08/0 : 126[6] -> 118[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96890:97070 [2] NCCL INFO Channel 14/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO Channel 12/0 : 80[0] -> 72[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO Channel 12/0 : 82[2] -> 74[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO Channel 05/0 : 87[7] -> 79[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96566:96755 [3] NCCL INFO Channel 02/0 : 35[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62884:63061 [3] NCCL INFO Channel 03/0 : 43[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235651:235823 [3] NCCL INFO Channel 13/0 : 107[3] -> 99[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272153:272328 [3] NCCL INFO Channel 04/0 : 123[3] -> 115[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96888:97066 [0] NCCL INFO Channel 13/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO Channel 02/0 : 91[3] -> 83[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO Channel 08/0 : 94[6] -> 86[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO Channel 09/0 : 85[5] -> 77[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96566:96755 [3] NCCL INFO Channel 04/0 : 35[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153566:153741 [4] NCCL INFO Channel 11/0 : 92[4] -> 76[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153568:153736 [6] NCCL INFO Channel 08/0 : 94[6] -> 86[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235650:235824 [2] NCCL INFO Channel 09/0 : 106[2] -> 98[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235654:235821 [6] NCCL INFO Channel 04/0 : 118[6] -> 110[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272150:272327 [0] NCCL INFO Channel 14/0 : 112[0] -> 104[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96891:97063 [3] NCCL INFO Channel 08/0 : 27[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO Channel 09/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO Channel 15/0 : 84[4] -> 76[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO Channel 06/0 : 87[7] -> 79[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96566:96755 [3] NCCL INFO Channel 06/0 : 35[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153562:153735 [0] NCCL INFO Channel 06/0 : 88[0] -> 80[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235648:235826 [0] NCCL INFO Channel 04/0 : 112[0] -> 104[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO Channel 13/0 : 80[0] -> 72[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO Channel 10/0 : 85[5] -> 77[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96566:96755 [3] NCCL INFO Channel 08/0 : 35[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153566:153741 [4] NCCL INFO Channel 13/0 : 92[4] -> 76[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153568:153736 [6] NCCL INFO Channel 10/0 : 94[6] -> 86[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62883:63058 [2] NCCL INFO Channel 07/0 : 42[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO Channel 13/0 : 82[2] -> 74[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO Channel 08/0 : 87[7] -> 79[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96566:96755 [3] NCCL INFO Channel 10/0 : 35[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153562:153735 [0] NCCL INFO Channel 08/0 : 88[0] -> 80[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62884:63061 [3] NCCL INFO Channel 05/0 : 43[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272156:272325 [6] NCCL INFO Channel 10/0 : 126[6] -> 118[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272153:272328 [3] NCCL INFO Channel 06/0 : 123[3] -> 115[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO Channel 10/0 : 94[6] -> 86[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO Channel 04/0 : 91[3] -> 83[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO Channel 11/0 : 85[5] -> 77[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96566:96755 [3] NCCL INFO Channel 12/0 : 35[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153566:153741 [4] NCCL INFO Channel 15/0 : 92[4] -> 76[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62883:63058 [2] NCCL INFO Channel 09/0 : 42[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272150:272327 [0] NCCL INFO Channel 15/0 : 112[0] -> 104[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96892:97064 [4] NCCL INFO Channel 14/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO Channel 11/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO Channel 14/0 : 80[0] -> 72[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO Channel 09/0 : 87[7] -> 79[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96566:96755 [3] NCCL INFO Channel 14/0 : 35[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153568:153736 [6] NCCL INFO Channel 12/0 : 94[6] -> 86[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272156:272325 [6] NCCL INFO Channel 12/0 : 126[6] -> 118[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272153:272328 [3] NCCL INFO Channel 08/0 : 123[3] -> 115[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96890:97070 [2] NCCL INFO Channel 15/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO Channel 14/0 : 82[2] -> 74[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO Channel 12/0 : 85[5] -> 77[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96569:96758 [6] NCCL INFO Channel 02/0 : 38[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153562:153735 [0] NCCL INFO Channel 10/0 : 88[0] -> 80[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272153:272328 [3] NCCL INFO Channel 10/0 : 123[3] -> 115[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272156:272325 [6] NCCL INFO Channel 14/0 : 126[6] -> 118[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96888:97066 [0] NCCL INFO Channel 14/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO Channel 10/0 : 87[7] -> 79[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96569:96758 [6] NCCL INFO Channel 04/0 : 38[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153568:153736 [6] NCCL INFO Channel 14/0 : 94[6] -> 86[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62884:63061 [3] NCCL INFO Channel 07/0 : 43[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96892:97064 [4] NCCL INFO Channel 15/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO Channel 12/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO Channel 14/0 : 85[5] -> 77[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96569:96758 [6] NCCL INFO Channel 08/0 : 38[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62883:63058 [2] NCCL INFO Channel 11/0 : 42[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96891:97063 [3] NCCL INFO Channel 10/0 : 27[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO Channel 13/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO Channel 12/0 : 94[6] -> 86[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO Channel 06/0 : 91[3] -> 83[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO Channel 11/0 : 87[7] -> 79[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62884:63061 [3] NCCL INFO Channel 09/0 : 43[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272156:272325 [6] NCCL INFO Channel 00/0 : 118[6] -> 110[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272153:272328 [3] NCCL INFO Channel 12/0 : 123[3] -> 115[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96888:97066 [0] NCCL INFO Channel 15/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO Channel 15/0 : 82[2] -> 74[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO Channel 15/0 : 85[5] -> 77[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96894:97068 [6] NCCL INFO Channel 00/0 : 30[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO Channel 15/0 : 80[0] -> 72[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO Channel 12/0 : 87[7] -> 79[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96569:96758 [6] NCCL INFO Channel 10/0 : 38[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153562:153735 [0] NCCL INFO Channel 12/0 : 88[0] -> 80[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96891:97063 [3] NCCL INFO Channel 12/0 : 27[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO Channel 14/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO Channel 14/0 : 94[6] -> 86[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO Channel 08/0 : 91[3] -> 83[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO Channel 01/0 : 77[5] -> 69[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96569:96758 [6] NCCL INFO Channel 12/0 : 38[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96566:96755 [3] NCCL INFO Channel 01/0 : 43[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153562:153735 [0] NCCL INFO Channel 14/0 : 88[0] -> 80[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62883:63058 [2] NCCL INFO Channel 13/0 : 42[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272156:272325 [6] NCCL INFO Channel 01/0 : 118[6] -> 110[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272153:272328 [3] NCCL INFO Channel 14/0 : 123[3] -> 115[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96894:97068 [6] NCCL INFO Channel 02/0 : 30[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO Channel 00/0 : 86[6] -> 78[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO Channel 10/0 : 91[3] -> 83[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO Channel 13/0 : 87[7] -> 79[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96569:96758 [6] NCCL INFO Channel 14/0 : 38[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96566:96755 [3] NCCL INFO Channel 03/0 : 43[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62884:63061 [3] NCCL INFO Channel 13/0 : 43[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO Channel 03/0 : 77[5] -> 69[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96566:96755 [3] NCCL INFO Channel 05/0 : 43[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62883:63058 [2] NCCL INFO Channel 15/0 : 42[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO Channel 15/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO Channel 14/0 : 87[7] -> 79[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96566:96755 [3] NCCL INFO Channel 07/0 : 43[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153566:153741 [4] NCCL INFO Channel 00/0 : 92[4] -> 84[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272156:272325 [6] NCCL INFO Channel 02/0 : 118[6] -> 110[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96891:97063 [3] NCCL INFO Channel 14/0 : 27[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96894:97068 [6] NCCL INFO Channel 04/0 : 30[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO Channel 01/0 : 86[6] -> 78[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO Channel 05/0 : 77[5] -> 69[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96566:96755 [3] NCCL INFO Channel 09/0 : 43[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272153:272328 [3] NCCL INFO Channel 00/0 : 115[3] -> 107[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96891:97063 [3] NCCL INFO Channel 00/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO Channel 12/0 : 91[3] -> 83[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO Channel 01/0 : 79[7] -> 71[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96566:96755 [3] NCCL INFO Channel 13/0 : 43[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62884:63061 [3] NCCL INFO Channel 15/0 : 43[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96894:97068 [6] NCCL INFO Channel 08/0 : 30[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO Channel 01/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO Channel 07/0 : 77[5] -> 69[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96563:96756 [0] NCCL INFO Channel 09/0 : 40[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153566:153741 [4] NCCL INFO Channel 02/0 : 92[4] -> 84[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96891:97063 [3] NCCL INFO Channel 01/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO Channel 03/0 : 79[7] -> 71[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96566:96755 [3] NCCL INFO Channel 15/0 : 43[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272156:272325 [6] NCCL INFO Channel 03/0 : 118[6] -> 110[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96894:97068 [6] NCCL INFO Channel 10/0 : 30[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO Channel 02/0 : 86[6] -> 78[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO Channel 14/0 : 91[3] -> 83[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO Channel 09/0 : 77[5] -> 69[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96569:96758 [6] NCCL INFO Channel 01/0 : 46[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272153:272328 [3] NCCL INFO Channel 01/0 : 115[3] -> 107[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96891:97063 [3] NCCL INFO Channel 02/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO Channel 05/0 : 79[7] -> 71[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124163:124340 [3] NCCL INFO Channel 01/0 : 11[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96563:96756 [0] NCCL INFO Channel 11/0 : 40[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96894:97068 [6] NCCL INFO Channel 12/0 : 30[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO Channel 03/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO Channel 11/0 : 77[5] -> 69[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96569:96758 [6] NCCL INFO Channel 03/0 : 46[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153566:153741 [4] NCCL INFO Channel 06/0 : 92[4] -> 84[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO Channel 03/0 : 86[6] -> 78[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO Channel 00/0 : 83[3] -> 75[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO Channel 07/0 : 79[7] -> 71[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124163:124340 [3] NCCL INFO Channel 03/0 : 11[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96563:96756 [0] NCCL INFO Channel 13/0 : 40[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272156:272325 [6] NCCL INFO Channel 04/0 : 118[6] -> 110[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO Channel 15/0 : 77[5] -> 69[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96569:96758 [6] NCCL INFO Channel 05/0 : 46[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272153:272328 [3] NCCL INFO Channel 02/0 : 115[3] -> 107[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96891:97063 [3] NCCL INFO Channel 04/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO Channel 05/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO Channel 09/0 : 79[7] -> 71[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96563:96756 [0] NCCL INFO Channel 15/0 : 40[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96894:97068 [6] NCCL INFO Channel 14/0 : 30[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO Channel 00/0 : 83[3] -> 75[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:124163:124340 [3] NCCL INFO Channel 05/0 : 11[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96569:96758 [6] NCCL INFO Channel 07/0 : 46[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153566:153741 [4] NCCL INFO Channel 08/0 : 92[4] -> 84[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96891:97063 [3] NCCL INFO Channel 05/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO Channel 04/0 : 86[6] -> 78[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO Channel 01/0 : 83[3] -> 75[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO Channel 11/0 : 79[7] -> 71[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96567:96760 [4] NCCL INFO Channel 11/0 : 44[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272156:272325 [6] NCCL INFO Channel 05/0 : 118[6] -> 110[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96894:97068 [6] NCCL INFO Channel 00/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO Channel 07/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO Channel 01/0 : 83[3] -> 75[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96569:96758 [6] NCCL INFO Channel 09/0 : 46[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272153:272328 [3] NCCL INFO Channel 04/0 : 115[3] -> 107[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96891:97063 [3] NCCL INFO Channel 06/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO Channel 13/0 : 79[7] -> 71[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153566:153741 [4] NCCL INFO Channel 10/0 : 92[4] -> 84[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272156:272325 [6] NCCL INFO Channel 07/0 : 118[6] -> 110[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO Channel 02/0 : 83[3] -> 75[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:124163:124340 [3] NCCL INFO Channel 07/0 : 11[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO Channel 09/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO Channel 05/0 : 86[6] -> 78[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO Channel 02/0 : 83[3] -> 75[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO Channel 00/0 : 82[2] -> 74[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96894:97068 [6] NCCL INFO Channel 01/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO Channel 04/0 : 83[3] -> 75[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96567:96760 [4] NCCL INFO Channel 13/0 : 44[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153566:153741 [4] NCCL INFO Channel 12/0 : 92[4] -> 84[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272153:272328 [3] NCCL INFO Channel 05/0 : 115[3] -> 107[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96891:97063 [3] NCCL INFO Channel 07/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO Channel 01/0 : 82[2] -> 74[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96569:96758 [6] NCCL INFO Channel 11/0 : 46[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272156:272325 [6] NCCL INFO Channel 08/0 : 118[6] -> 110[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96894:97068 [6] NCCL INFO Channel 02/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO Channel 11/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO Channel 05/0 : 83[3] -> 75[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:124163:124340 [3] NCCL INFO Channel 09/0 : 11[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96567:96760 [4] NCCL INFO Channel 15/0 : 44[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272153:272328 [3] NCCL INFO Channel 06/0 : 115[3] -> 107[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96891:97063 [3] NCCL INFO Channel 08/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO Channel 07/0 : 86[6] -> 78[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO Channel 04/0 : 83[3] -> 75[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO Channel 03/0 : 82[2] -> 74[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO Channel 06/0 : 83[3] -> 75[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96569:96758 [6] NCCL INFO Channel 13/0 : 46[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153566:153741 [4] NCCL INFO Channel 14/0 : 92[4] -> 84[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO Channel 13/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO Channel 07/0 : 83[3] -> 75[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO Channel 04/0 : 82[2] -> 74[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96569:96758 [6] NCCL INFO Channel 15/0 : 46[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO Channel 05/0 : 82[2] -> 74[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO Channel 08/0 : 83[3] -> 75[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:124163:124340 [3] NCCL INFO Channel 13/0 : 11[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272156:272325 [6] NCCL INFO Channel 09/0 : 118[6] -> 110[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96894:97068 [6] NCCL INFO Channel 03/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO Channel 06/0 : 82[2] -> 74[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO Channel 09/0 : 83[3] -> 75[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272153:272328 [3] NCCL INFO Channel 07/0 : 115[3] -> 107[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96891:97063 [3] NCCL INFO Channel 09/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO Channel 07/0 : 82[2] -> 74[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO Channel 10/0 : 83[3] -> 75[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272156:272325 [6] NCCL INFO Channel 10/0 : 118[6] -> 110[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96894:97068 [6] NCCL INFO Channel 04/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO Channel 15/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO Channel 08/0 : 86[6] -> 78[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO Channel 05/0 : 83[3] -> 75[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO Channel 12/0 : 83[3] -> 75[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO Channel 08/0 : 82[2] -> 74[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96891:97063 [3] NCCL INFO Channel 10/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO Channel 13/0 : 83[3] -> 75[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO Channel 09/0 : 82[2] -> 74[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:124163:124340 [3] NCCL INFO Channel 15/0 : 11[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO Channel 14/0 : 83[3] -> 75[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272153:272328 [3] NCCL INFO Channel 08/0 : 115[3] -> 107[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO Channel 11/0 : 82[2] -> 74[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272156:272325 [6] NCCL INFO Channel 11/0 : 118[6] -> 110[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO Channel 15/0 : 83[3] -> 75[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO Channel 12/0 : 82[2] -> 74[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96894:97068 [6] NCCL INFO Channel 05/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32708:32879 [4] NCCL INFO Channel 01/0 : 44[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO Channel 01/0 : 75[3] -> 67[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO Channel 13/0 : 82[2] -> 74[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96891:97063 [3] NCCL INFO Channel 12/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO Channel 09/0 : 86[6] -> 78[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO Channel 06/0 : 83[3] -> 75[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO Channel 03/0 : 75[3] -> 67[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO Channel 14/0 : 82[2] -> 74[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96894:97068 [6] NCCL INFO Channel 07/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO Channel 05/0 : 75[3] -> 67[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO Channel 15/0 : 82[2] -> 74[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272153:272328 [3] NCCL INFO Channel 09/0 : 115[3] -> 107[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO Channel 00/0 : 86[6] -> 78[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272156:272325 [6] NCCL INFO Channel 12/0 : 118[6] -> 110[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO Channel 07/0 : 75[3] -> 67[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO Channel 01/0 : 74[2] -> 66[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32708:32879 [4] NCCL INFO Channel 03/0 : 44[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO Channel 01/0 : 86[6] -> 78[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96891:97063 [3] NCCL INFO Channel 13/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO Channel 10/0 : 86[6] -> 78[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO Channel 07/0 : 83[3] -> 75[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO Channel 09/0 : 75[3] -> 67[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO Channel 03/0 : 74[2] -> 66[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96894:97068 [6] NCCL INFO Channel 08/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO Channel 01/0 : 80[0] -> 72[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO Channel 02/0 : 86[6] -> 78[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96891:97063 [3] NCCL INFO Channel 14/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO Channel 05/0 : 74[2] -> 66[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272153:272328 [3] NCCL INFO Channel 10/0 : 115[3] -> 107[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO Channel 13/0 : 75[3] -> 67[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272156:272325 [6] NCCL INFO Channel 13/0 : 118[6] -> 110[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO Channel 02/0 : 80[0] -> 72[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO Channel 03/0 : 86[6] -> 78[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272153:272328 [3] NCCL INFO Channel 12/0 : 115[3] -> 107[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32708:32879 [4] NCCL INFO Channel 05/0 : 44[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO Channel 07/0 : 74[2] -> 66[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96894:97068 [6] NCCL INFO Channel 09/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO Channel 11/0 : 86[6] -> 78[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO Channel 15/0 : 75[3] -> 67[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96891:97063 [3] NCCL INFO Channel 15/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO Channel 03/0 : 80[0] -> 72[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO Channel 04/0 : 86[6] -> 78[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272156:272325 [6] NCCL INFO Channel 15/0 : 118[6] -> 110[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272153:272328 [3] NCCL INFO Channel 13/0 : 115[3] -> 107[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32708:32879 [4] NCCL INFO Channel 07/0 : 44[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO Channel 08/0 : 83[3] -> 75[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96894:97068 [6] NCCL INFO Channel 10/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO Channel 09/0 : 74[2] -> 66[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO Channel 04/0 : 80[0] -> 72[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272153:272328 [3] NCCL INFO Channel 14/0 : 115[3] -> 107[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO Channel 11/0 : 74[2] -> 66[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO Channel 05/0 : 86[6] -> 78[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO Channel 12/0 : 86[6] -> 78[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO Channel 05/0 : 80[0] -> 72[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96894:97068 [6] NCCL INFO Channel 11/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32708:32879 [4] NCCL INFO Channel 09/0 : 44[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO Channel 09/0 : 83[3] -> 75[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO Channel 13/0 : 74[2] -> 66[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO Channel 07/0 : 86[6] -> 78[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO Channel 06/0 : 80[0] -> 72[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO Channel 15/0 : 74[2] -> 66[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO Channel 08/0 : 86[6] -> 78[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO Channel 07/0 : 80[0] -> 72[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96894:97068 [6] NCCL INFO Channel 12/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32710:32883 [6] NCCL INFO Channel 15/0 : 30[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO Channel 00/0 : 84[4] -> 76[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO Channel 13/0 : 86[6] -> 78[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO Channel 09/0 : 86[6] -> 78[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO Channel 10/0 : 83[3] -> 75[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO Channel 09/0 : 80[0] -> 72[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272153:272328 [3] NCCL INFO Channel 15/0 : 115[3] -> 107[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO Channel 01/0 : 84[4] -> 76[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96894:97068 [6] NCCL INFO Channel 13/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32708:32879 [4] NCCL INFO Channel 11/0 : 44[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO Channel 10/0 : 86[6] -> 78[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO Channel 15/0 : 86[6] -> 78[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO Channel 12/0 : 83[3] -> 75[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO Channel 10/0 : 80[0] -> 72[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO Channel 02/0 : 84[4] -> 76[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO Channel 11/0 : 86[6] -> 78[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32708:32879 [4] NCCL INFO Channel 13/0 : 44[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96894:97068 [6] NCCL INFO Channel 15/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO Channel 13/0 : 83[3] -> 75[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO Channel 12/0 : 86[6] -> 78[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO Channel 13/0 : 86[6] -> 78[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32708:32879 [4] NCCL INFO Channel 15/0 : 44[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO Channel 14/0 : 83[3] -> 75[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO Channel 15/0 : 86[6] -> 78[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62887:63055 [6] NCCL INFO Channel 00/0 : 54[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32708:32879 [4] NCCL INFO Channel 01/0 : 28[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO Channel 15/0 : 83[3] -> 75[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO Channel 01/0 : 78[6] -> 70[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32706:32877 [2] NCCL INFO Channel 08/0 : 26[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62887:63055 [6] NCCL INFO Channel 01/0 : 54[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32708:32879 [4] NCCL INFO Channel 03/0 : 28[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32710:32883 [6] NCCL INFO Channel 00/0 : 30[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32706:32877 [2] NCCL INFO Channel 10/0 : 26[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO Channel 03/0 : 78[6] -> 70[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO Channel 00/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62887:63055 [6] NCCL INFO Channel 02/0 : 54[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32708:32879 [4] NCCL INFO Channel 05/0 : 28[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO Channel 05/0 : 78[6] -> 70[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO Channel 01/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32706:32877 [2] NCCL INFO Channel 12/0 : 26[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32710:32883 [6] NCCL INFO Channel 02/0 : 30[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62887:63055 [6] NCCL INFO Channel 03/0 : 54[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO Channel 07/0 : 78[6] -> 70[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32708:32879 [4] NCCL INFO Channel 07/0 : 28[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO Channel 02/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62887:63055 [6] NCCL INFO Channel 04/0 : 54[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32706:32877 [2] NCCL INFO Channel 14/0 : 26[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32710:32883 [6] NCCL INFO Channel 04/0 : 30[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO Channel 09/0 : 78[6] -> 70[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62887:63055 [6] NCCL INFO Channel 05/0 : 54[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO Channel 03/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32708:32879 [4] NCCL INFO Channel 09/0 : 28[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO Channel 11/0 : 80[0] -> 72[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32710:32883 [6] NCCL INFO Channel 08/0 : 30[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62887:63055 [6] NCCL INFO Channel 07/0 : 54[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO Channel 11/0 : 78[6] -> 70[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32708:32879 [4] NCCL INFO Channel 11/0 : 28[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO Channel 12/0 : 80[0] -> 72[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32710:32883 [6] NCCL INFO Channel 10/0 : 30[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO Channel 04/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO Channel 13/0 : 78[6] -> 70[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62887:63055 [6] NCCL INFO Channel 08/0 : 54[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32708:32879 [4] NCCL INFO Channel 13/0 : 28[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO Channel 13/0 : 80[0] -> 72[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32710:32883 [6] NCCL INFO Channel 12/0 : 30[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO Channel 05/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32708:32879 [4] NCCL INFO Channel 15/0 : 28[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO Channel 15/0 : 78[6] -> 70[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32710:32883 [6] NCCL INFO Channel 14/0 : 30[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO Channel 14/0 : 80[0] -> 72[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62887:63055 [6] NCCL INFO Channel 09/0 : 54[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO Channel 07/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO Channel 15/0 : 80[0] -> 72[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62887:63055 [6] NCCL INFO Channel 10/0 : 54[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62887:63055 [6] NCCL INFO Channel 11/0 : 54[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO Channel 08/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO Channel 01/0 : 72[0] -> 64[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62887:63055 [6] NCCL INFO Channel 12/0 : 54[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62885:63054 [4] NCCL INFO Channel 00/0 : 52[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO Channel 09/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO Channel 03/0 : 72[0] -> 64[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62887:63055 [6] NCCL INFO Channel 13/0 : 54[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62885:63054 [4] NCCL INFO Channel 01/0 : 52[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO Channel 10/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32708:32879 [4] NCCL INFO Channel 00/0 : 28[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO Channel 05/0 : 72[0] -> 64[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO Channel 00/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32708:32879 [4] NCCL INFO Channel 02/0 : 28[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62887:63055 [6] NCCL INFO Channel 15/0 : 54[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62885:63054 [4] NCCL INFO Channel 02/0 : 52[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO Channel 03/0 : 84[4] -> 76[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO Channel 07/0 : 72[0] -> 64[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO Channel 11/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32708:32879 [4] NCCL INFO Channel 06/0 : 28[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO Channel 05/0 : 84[4] -> 76[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO Channel 09/0 : 72[0] -> 64[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62887:63055 [6] NCCL INFO Channel 01/0 : 46[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62885:63054 [4] NCCL INFO Channel 03/0 : 52[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO Channel 01/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32708:32879 [4] NCCL INFO Channel 08/0 : 28[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO Channel 12/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO Channel 06/0 : 84[4] -> 76[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO Channel 11/0 : 72[0] -> 64[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62887:63055 [6] NCCL INFO Channel 03/0 : 46[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO Channel 02/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32708:32879 [4] NCCL INFO Channel 10/0 : 28[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO Channel 13/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO Channel 07/0 : 84[4] -> 76[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62885:63054 [4] NCCL INFO Channel 05/0 : 52[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO Channel 13/0 : 72[0] -> 64[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO Channel 15/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32708:32879 [4] NCCL INFO Channel 12/0 : 28[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO Channel 03/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62887:63055 [6] NCCL INFO Channel 05/0 : 46[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62885:63054 [4] NCCL INFO Channel 06/0 : 52[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO Channel 08/0 : 84[4] -> 76[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO Channel 15/0 : 72[0] -> 64[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO Channel 09/0 : 84[4] -> 76[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO Channel 01/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32708:32879 [4] NCCL INFO Channel 14/0 : 28[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62887:63055 [6] NCCL INFO Channel 07/0 : 46[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62885:63054 [4] NCCL INFO Channel 07/0 : 52[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO Channel 10/0 : 84[4] -> 76[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO Channel 05/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO Channel 03/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62881:63059 [0] NCCL INFO Channel 01/0 : 48[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62887:63055 [6] NCCL INFO Channel 09/0 : 46[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO Channel 11/0 : 84[4] -> 76[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62885:63054 [4] NCCL INFO Channel 08/0 : 52[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO Channel 06/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32704:32881 [0] NCCL INFO Channel 02/0 : 24[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO Channel 05/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62887:63055 [6] NCCL INFO Channel 11/0 : 46[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62881:63059 [0] NCCL INFO Channel 02/0 : 48[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO Channel 13/0 : 84[4] -> 76[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62885:63054 [4] NCCL INFO Channel 09/0 : 52[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO Channel 07/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO Channel 01/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO Channel 07/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32704:32881 [0] NCCL INFO Channel 04/0 : 24[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62887:63055 [6] NCCL INFO Channel 13/0 : 46[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO Channel 14/0 : 84[4] -> 76[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO Channel 08/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32704:32881 [0] NCCL INFO Channel 06/0 : 24[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62881:63059 [0] NCCL INFO Channel 03/0 : 48[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62885:63054 [4] NCCL INFO Channel 10/0 : 52[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO Channel 15/0 : 84[4] -> 76[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO Channel 09/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO Channel 02/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62887:63055 [6] NCCL INFO Channel 15/0 : 46[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32704:32881 [0] NCCL INFO Channel 08/0 : 24[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO Channel 01/0 : 76[4] -> 68[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62885:63054 [4] NCCL INFO Channel 11/0 : 52[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62881:63059 [0] NCCL INFO Channel 04/0 : 48[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO Channel 09/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO Channel 11/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32704:32881 [0] NCCL INFO Channel 10/0 : 24[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO Channel 03/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO Channel 03/0 : 76[4] -> 68[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62885:63054 [4] NCCL INFO Channel 13/0 : 52[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62881:63059 [0] NCCL INFO Channel 05/0 : 48[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32704:32881 [0] NCCL INFO Channel 12/0 : 24[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO Channel 10/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO Channel 13/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO Channel 04/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO Channel 11/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO Channel 05/0 : 76[4] -> 68[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62885:63054 [4] NCCL INFO Channel 14/0 : 52[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62881:63059 [0] NCCL INFO Channel 06/0 : 48[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32704:32881 [0] NCCL INFO Channel 14/0 : 24[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO Channel 15/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO Channel 07/0 : 76[4] -> 68[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62885:63054 [4] NCCL INFO Channel 15/0 : 52[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62881:63059 [0] NCCL INFO Channel 07/0 : 48[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO Channel 05/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO Channel 13/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO Channel 09/0 : 76[4] -> 68[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62885:63054 [4] NCCL INFO Channel 01/0 : 44[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO Channel 06/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO Channel 14/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO Channel 11/0 : 76[4] -> 68[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62881:63059 [0] NCCL INFO Channel 09/0 : 48[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO Channel 15/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO Channel 13/0 : 76[4] -> 68[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62885:63054 [4] NCCL INFO Channel 03/0 : 44[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62881:63059 [0] NCCL INFO Channel 10/0 : 48[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO Channel 07/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO Channel 15/0 : 76[4] -> 68[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62885:63054 [4] NCCL INFO Channel 05/0 : 44[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62881:63059 [0] NCCL INFO Channel 11/0 : 48[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO Channel 01/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62885:63054 [4] NCCL INFO Channel 07/0 : 44[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62881:63059 [0] NCCL INFO Channel 12/0 : 48[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO Channel 09/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO Channel 03/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62885:63054 [4] NCCL INFO Channel 09/0 : 44[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO Channel 10/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO Channel 05/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62881:63059 [0] NCCL INFO Channel 13/0 : 48[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO Channel 11/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO Channel 07/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62885:63054 [4] NCCL INFO Channel 11/0 : 44[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62881:63059 [0] NCCL INFO Channel 14/0 : 48[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO Channel 12/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO Channel 09/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62885:63054 [4] NCCL INFO Channel 13/0 : 44[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62881:63059 [0] NCCL INFO Channel 15/0 : 48[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO Channel 13/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62885:63054 [4] NCCL INFO Channel 15/0 : 44[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62881:63059 [0] NCCL INFO Channel 01/0 : 40[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO Channel 11/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62881:63059 [0] NCCL INFO Channel 03/0 : 40[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO Channel 14/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO Channel 13/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62881:63059 [0] NCCL INFO Channel 05/0 : 40[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO Channel 15/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO Channel 15/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62881:63059 [0] NCCL INFO Channel 07/0 : 40[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO Channel 01/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62881:63059 [0] NCCL INFO Channel 09/0 : 40[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO Channel 03/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62881:63059 [0] NCCL INFO Channel 11/0 : 40[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62881:63059 [0] NCCL INFO Channel 13/0 : 40[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO Channel 05/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO Channel 07/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO Channel 09/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO Channel 11/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO Channel 13/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO Channel 15/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62881:63059 [0] NCCL INFO Channel 15/0 : 40[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-52:2722052:2722226 [7] NCCL INFO Channel 10/0 : 127[7] -> 119[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722050:2722225 [5] NCCL INFO Channel 14/0 : 125[5] -> 117[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-52:2722052:2722226 [7] NCCL INFO Channel 12/0 : 127[7] -> 119[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722052:2722226 [7] NCCL INFO Channel 14/0 : 127[7] -> 119[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722047:2722221 [2] NCCL INFO Channel 00/0 : 122[2] -> 114[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-52:2722047:2722221 [2] NCCL INFO Channel 04/0 : 122[2] -> 114[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-52:2722047:2722221 [2] NCCL INFO Channel 06/0 : 122[2] -> 114[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-52:2722047:2722221 [2] NCCL INFO Channel 08/0 : 122[2] -> 114[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-52:2722047:2722221 [2] NCCL INFO Channel 10/0 : 122[2] -> 114[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-52:2722048:2722220 [3] NCCL INFO Channel 00/0 : 123[3] -> 115[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-52:2722047:2722221 [2] NCCL INFO Channel 12/0 : 122[2] -> 114[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-52:2722048:2722220 [3] NCCL INFO Channel 02/0 : 123[3] -> 115[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-52:2722047:2722221 [2] NCCL INFO Channel 14/0 : 122[2] -> 114[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-52:2722048:2722220 [3] NCCL INFO Channel 04/0 : 123[3] -> 115[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-52:2722048:2722220 [3] NCCL INFO Channel 06/0 : 123[3] -> 115[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-52:2722048:2722220 [3] NCCL INFO Channel 08/0 : 123[3] -> 115[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-52:2722048:2722220 [3] NCCL INFO Channel 10/0 : 123[3] -> 115[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-52:2722048:2722220 [3] NCCL INFO Channel 12/0 : 123[3] -> 115[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-52:2722048:2722220 [3] NCCL INFO Channel 14/0 : 123[3] -> 115[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-52:2722051:2722224 [6] NCCL INFO Channel 00/0 : 126[6] -> 118[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO Channel 00/0 : 124[4] -> 116[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722051:2722224 [6] NCCL INFO Channel 02/0 : 126[6] -> 118[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-52:2722051:2722224 [6] NCCL INFO Channel 04/0 : 126[6] -> 118[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-52:2722051:2722224 [6] NCCL INFO Channel 08/0 : 126[6] -> 118[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-52:2722051:2722224 [6] NCCL INFO Channel 10/0 : 126[6] -> 118[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-52:2722051:2722224 [6] NCCL INFO Channel 12/0 : 126[6] -> 118[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-52:2722051:2722224 [6] NCCL INFO Channel 14/0 : 126[6] -> 118[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO Channel 02/0 : 124[4] -> 116[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO Channel 06/0 : 124[4] -> 116[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO Channel 08/0 : 124[4] -> 116[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO Channel 10/0 : 124[4] -> 116[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO Channel 12/0 : 124[4] -> 116[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO Channel 14/0 : 124[4] -> 116[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722045:2722219 [0] NCCL INFO Channel 02/0 : 120[0] -> 112[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-52:2722045:2722219 [0] NCCL INFO Channel 04/0 : 120[0] -> 112[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-52:2722045:2722219 [0] NCCL INFO Channel 06/0 : 120[0] -> 112[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-52:2722045:2722219 [0] NCCL INFO Channel 08/0 : 120[0] -> 112[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-52:2722045:2722219 [0] NCCL INFO Channel 10/0 : 120[0] -> 112[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-52:2722045:2722219 [0] NCCL INFO Channel 12/0 : 120[0] -> 112[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-52:2722045:2722219 [0] NCCL INFO Channel 14/0 : 120[0] -> 112[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO Channel 07/0 : 106[2] -> 98[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114518:114693 [6] NCCL INFO Channel 15/0 : 78[6] -> 70[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO Channel 03/0 : 104[0] -> 96[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO Channel 06/0 : 115[3] -> 99[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114515:114690 [3] NCCL INFO Channel 15/0 : 75[3] -> 67[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO Channel 05/0 : 108[4] -> 100[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO Channel 10/0 : 118[6] -> 102[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO Channel 09/0 : 106[2] -> 98[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO Channel 05/0 : 104[0] -> 96[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO Channel 08/0 : 115[3] -> 99[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO Channel 07/0 : 108[4] -> 100[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO Channel 12/0 : 118[6] -> 102[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO Channel 11/0 : 106[2] -> 98[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO Channel 07/0 : 104[0] -> 96[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO Channel 10/0 : 115[3] -> 99[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO Channel 14/0 : 118[6] -> 102[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO Channel 13/0 : 106[2] -> 98[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO Channel 09/0 : 108[4] -> 100[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO Channel 09/0 : 104[0] -> 96[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO Channel 12/0 : 115[3] -> 99[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO Channel 00/0 : 102[6] -> 86[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO Channel 15/0 : 106[2] -> 98[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO Channel 11/0 : 104[0] -> 96[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO Channel 14/0 : 115[3] -> 99[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO Channel 11/0 : 108[4] -> 100[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO Channel 02/0 : 102[6] -> 86[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO Channel 13/0 : 104[0] -> 96[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO Channel 00/0 : 99[3] -> 83[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO Channel 13/0 : 108[4] -> 100[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO Channel 04/0 : 102[6] -> 86[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO Channel 02/0 : 99[3] -> 83[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO Channel 15/0 : 104[0] -> 96[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO Channel 15/0 : 108[4] -> 100[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO Channel 08/0 : 102[6] -> 86[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO Channel 04/0 : 99[3] -> 83[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO Channel 10/0 : 102[6] -> 86[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO Channel 06/0 : 99[3] -> 83[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO Channel 12/0 : 102[6] -> 86[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO Channel 08/0 : 99[3] -> 83[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO Channel 14/0 : 102[6] -> 86[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO Channel 10/0 : 99[3] -> 83[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO Channel 12/0 : 99[3] -> 83[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO Channel 14/0 : 99[3] -> 83[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO Channel 01/0 : 110[6] -> 102[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO Channel 03/0 : 110[6] -> 102[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO Channel 05/0 : 110[6] -> 102[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO Channel 01/0 : 107[3] -> 99[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO Channel 07/0 : 110[6] -> 102[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO Channel 03/0 : 107[3] -> 99[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO Channel 09/0 : 110[6] -> 102[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO Channel 05/0 : 107[3] -> 99[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO Channel 11/0 : 110[6] -> 102[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO Channel 07/0 : 107[3] -> 99[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO Channel 13/0 : 110[6] -> 102[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO Channel 09/0 : 107[3] -> 99[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO Channel 15/0 : 110[6] -> 102[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO Channel 13/0 : 107[3] -> 99[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO Channel 15/0 : 107[3] -> 99[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235651:235823 [3] NCCL INFO Channel 15/0 : 107[3] -> 99[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO Channel 15/0 : 50[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235650:235824 [2] NCCL INFO Channel 11/0 : 106[2] -> 98[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235654:235821 [6] NCCL INFO Channel 05/0 : 118[6] -> 110[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO Channel 12/0 : 48[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235648:235826 [0] NCCL INFO Channel 05/0 : 112[0] -> 104[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO Channel 13/0 : 48[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235654:235821 [6] NCCL INFO Channel 07/0 : 118[6] -> 110[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO Channel 14/0 : 48[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235650:235824 [2] NCCL INFO Channel 13/0 : 106[2] -> 98[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO Channel 00/0 : 62[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235648:235826 [0] NCCL INFO Channel 06/0 : 112[0] -> 104[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO Channel 15/0 : 48[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235654:235821 [6] NCCL INFO Channel 08/0 : 118[6] -> 110[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO Channel 02/0 : 62[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235652:235822 [4] NCCL INFO Channel 00/0 : 116[4] -> 108[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235650:235824 [2] NCCL INFO Channel 15/0 : 106[2] -> 98[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO Channel 04/0 : 62[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235648:235826 [0] NCCL INFO Channel 07/0 : 112[0] -> 104[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO Channel 08/0 : 62[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235654:235821 [6] NCCL INFO Channel 09/0 : 118[6] -> 110[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO Channel 10/0 : 62[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235652:235822 [4] NCCL INFO Channel 01/0 : 116[4] -> 108[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO Channel 12/0 : 62[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235648:235826 [0] NCCL INFO Channel 09/0 : 112[0] -> 104[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235654:235821 [6] NCCL INFO Channel 10/0 : 118[6] -> 110[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO Channel 00/0 : 59[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235652:235822 [4] NCCL INFO Channel 02/0 : 116[4] -> 108[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO Channel 14/0 : 62[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235654:235821 [6] NCCL INFO Channel 11/0 : 118[6] -> 110[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO Channel 02/0 : 59[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235648:235826 [0] NCCL INFO Channel 10/0 : 112[0] -> 104[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO Channel 00/0 : 54[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235652:235822 [4] NCCL INFO Channel 03/0 : 116[4] -> 108[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO Channel 02/0 : 52[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235654:235821 [6] NCCL INFO Channel 12/0 : 118[6] -> 110[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO Channel 04/0 : 59[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235648:235826 [0] NCCL INFO Channel 11/0 : 112[0] -> 104[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235652:235822 [4] NCCL INFO Channel 05/0 : 116[4] -> 108[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO Channel 01/0 : 54[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO Channel 03/0 : 52[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235654:235821 [6] NCCL INFO Channel 13/0 : 118[6] -> 110[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO Channel 06/0 : 59[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235652:235822 [4] NCCL INFO Channel 06/0 : 116[4] -> 108[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235648:235826 [0] NCCL INFO Channel 12/0 : 112[0] -> 104[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO Channel 05/0 : 52[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO Channel 08/0 : 59[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235654:235821 [6] NCCL INFO Channel 15/0 : 118[6] -> 110[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO Channel 02/0 : 54[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235652:235822 [4] NCCL INFO Channel 07/0 : 116[4] -> 108[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235648:235826 [0] NCCL INFO Channel 13/0 : 112[0] -> 104[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO Channel 10/0 : 59[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235654:235821 [6] NCCL INFO Channel 01/0 : 110[6] -> 102[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO Channel 03/0 : 54[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235652:235822 [4] NCCL INFO Channel 08/0 : 116[4] -> 108[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO Channel 06/0 : 52[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235654:235821 [6] NCCL INFO Channel 03/0 : 110[6] -> 102[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235648:235826 [0] NCCL INFO Channel 14/0 : 112[0] -> 104[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO Channel 12/0 : 59[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235652:235822 [4] NCCL INFO Channel 09/0 : 116[4] -> 108[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO Channel 04/0 : 54[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO Channel 07/0 : 52[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235654:235821 [6] NCCL INFO Channel 05/0 : 110[6] -> 102[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO Channel 14/0 : 59[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235648:235826 [0] NCCL INFO Channel 15/0 : 112[0] -> 104[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO Channel 08/0 : 52[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235652:235822 [4] NCCL INFO Channel 10/0 : 116[4] -> 108[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO Channel 05/0 : 54[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235654:235821 [6] NCCL INFO Channel 07/0 : 110[6] -> 102[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO Channel 00/0 : 51[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO Channel 09/0 : 52[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235648:235826 [0] NCCL INFO Channel 01/0 : 104[0] -> 96[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO Channel 07/0 : 54[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235652:235822 [4] NCCL INFO Channel 11/0 : 116[4] -> 108[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO Channel 01/0 : 51[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO Channel 10/0 : 52[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO Channel 08/0 : 54[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235654:235821 [6] NCCL INFO Channel 09/0 : 110[6] -> 102[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235648:235826 [0] NCCL INFO Channel 03/0 : 104[0] -> 96[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO Channel 11/0 : 52[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO Channel 09/0 : 54[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO Channel 02/0 : 51[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235652:235822 [4] NCCL INFO Channel 13/0 : 116[4] -> 108[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO Channel 10/0 : 54[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO Channel 04/0 : 51[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235654:235821 [6] NCCL INFO Channel 11/0 : 110[6] -> 102[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO Channel 13/0 : 52[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235648:235826 [0] NCCL INFO Channel 05/0 : 104[0] -> 96[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO Channel 11/0 : 54[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235652:235822 [4] NCCL INFO Channel 14/0 : 116[4] -> 108[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO Channel 05/0 : 51[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO Channel 14/0 : 52[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235654:235821 [6] NCCL INFO Channel 13/0 : 110[6] -> 102[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO Channel 12/0 : 54[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235648:235826 [0] NCCL INFO Channel 07/0 : 104[0] -> 96[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235652:235822 [4] NCCL INFO Channel 15/0 : 116[4] -> 108[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO Channel 06/0 : 51[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235654:235821 [6] NCCL INFO Channel 15/0 : 110[6] -> 102[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO Channel 15/0 : 52[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235648:235826 [0] NCCL INFO Channel 09/0 : 104[0] -> 96[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO Channel 13/0 : 54[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235652:235822 [4] NCCL INFO Channel 01/0 : 108[4] -> 100[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO Channel 07/0 : 51[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235648:235826 [0] NCCL INFO Channel 11/0 : 104[0] -> 96[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO Channel 15/0 : 54[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235652:235822 [4] NCCL INFO Channel 03/0 : 108[4] -> 100[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO Channel 08/0 : 51[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235652:235822 [4] NCCL INFO Channel 05/0 : 108[4] -> 100[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO Channel 09/0 : 51[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235648:235826 [0] NCCL INFO Channel 13/0 : 104[0] -> 96[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO Channel 10/0 : 51[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235652:235822 [4] NCCL INFO Channel 07/0 : 108[4] -> 100[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO Channel 12/0 : 51[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235648:235826 [0] NCCL INFO Channel 15/0 : 104[0] -> 96[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO Channel 13/0 : 51[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235652:235822 [4] NCCL INFO Channel 09/0 : 108[4] -> 100[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO Channel 14/0 : 51[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235652:235822 [4] NCCL INFO Channel 11/0 : 108[4] -> 100[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO Channel 15/0 : 51[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235652:235822 [4] NCCL INFO Channel 13/0 : 108[4] -> 100[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235652:235822 [4] NCCL INFO Channel 15/0 : 108[4] -> 100[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:124161:124337 [1] NCCL INFO Connected NVLS tree
+ip-26-0-160-225:124161:124337 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-225:124161:124337 [1] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-160-225:124161:124337 [1] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-160-225:124163:124340 [3] NCCL INFO Connected NVLS tree
+ip-26-0-160-225:124163:124340 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-225:124163:124340 [3] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-160-225:124163:124340 [3] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-160-225:124167:124338 [7] NCCL INFO Connected NVLS tree
+ip-26-0-160-225:124167:124338 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-225:124167:124338 [7] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-160-225:124167:124338 [7] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-160-225:124165:124341 [5] NCCL INFO Connected NVLS tree
+ip-26-0-160-225:124165:124341 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-225:124165:124341 [5] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-160-225:124165:124341 [5] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-168-52:2722045:2722219 [0] NCCL INFO Connected NVLS tree
+ip-26-0-168-52:2722045:2722219 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-52:2722045:2722219 [0] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-168-52:2722045:2722219 [0] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO Connected NVLS tree
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO Connected NVLS tree
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-103:96895:97069 [7] NCCL INFO Connected NVLS tree
+ip-26-0-161-103:96895:97069 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-103:96895:97069 [7] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-161-103:96895:97069 [7] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-168-52:2722051:2722224 [6] NCCL INFO Connected NVLS tree
+ip-26-0-168-52:2722051:2722224 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-52:2722051:2722224 [6] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-168-52:2722051:2722224 [6] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-168-52:2722047:2722221 [2] NCCL INFO Connected NVLS tree
+ip-26-0-168-52:2722047:2722221 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-52:2722047:2722221 [2] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-168-52:2722047:2722221 [2] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO Connected NVLS tree
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO Connected NVLS tree
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO Connected NVLS tree
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-168-52:2722052:2722226 [7] NCCL INFO Connected NVLS tree
+ip-26-0-168-52:2722052:2722226 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-52:2722052:2722226 [7] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-168-52:2722052:2722226 [7] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO Connected NVLS tree
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-123:32709:32882 [5] NCCL INFO Connected NVLS tree
+ip-26-0-161-123:32709:32882 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-123:32709:32882 [5] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-161-123:32709:32882 [5] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-123:32711:32884 [7] NCCL INFO Connected NVLS tree
+ip-26-0-161-123:32711:32884 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-123:32711:32884 [7] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-161-123:32711:32884 [7] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO Connected NVLS tree
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-165-202:153565:153738 [3] NCCL INFO Connected NVLS tree
+ip-26-0-165-202:153565:153738 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-202:153565:153738 [3] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-165-202:153565:153738 [3] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO Connected NVLS tree
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-168-34:272154:272330 [4] NCCL INFO Connected NVLS tree
+ip-26-0-168-34:272154:272330 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-34:272154:272330 [4] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-168-34:272154:272330 [4] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-123:32705:32880 [1] NCCL INFO Connected NVLS tree
+ip-26-0-161-123:32705:32880 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-123:32705:32880 [1] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-161-123:32705:32880 [1] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-103:96891:97063 [3] NCCL INFO Connected NVLS tree
+ip-26-0-161-103:96891:97063 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-103:96891:97063 [3] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-161-103:96891:97063 [3] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-168-30:235655:235828 [7] NCCL INFO Connected NVLS tree
+ip-26-0-168-30:235655:235828 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-30:235655:235828 [7] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-168-30:235655:235828 [7] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-168-52:2722046:2722223 [1] NCCL INFO Connected NVLS tree
+ip-26-0-168-52:2722046:2722223 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-52:2722046:2722223 [1] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-168-52:2722046:2722223 [1] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-160-225:124164:124336 [4] NCCL INFO Connected NVLS tree
+ip-26-0-160-225:124164:124336 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-225:124164:124336 [4] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-160-225:124164:124336 [4] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO Connected NVLS tree
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO Connected NVLS tree
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-168-34:272156:272325 [6] NCCL INFO Connected NVLS tree
+ip-26-0-168-34:272156:272325 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-34:272156:272325 [6] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-168-34:272156:272325 [6] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-168-52:2722048:2722220 [3] NCCL INFO Connected NVLS tree
+ip-26-0-168-52:2722048:2722220 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-52:2722048:2722220 [3] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-168-52:2722048:2722220 [3] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-168-34:272150:272327 [0] NCCL INFO Connected NVLS tree
+ip-26-0-168-34:272150:272327 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-34:272150:272327 [0] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-168-34:272150:272327 [0] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-165-202:153563:153740 [1] NCCL INFO Connected NVLS tree
+ip-26-0-165-202:153563:153740 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-202:153563:153740 [1] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-165-202:153563:153740 [1] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO Connected NVLS tree
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-160-225:124166:124339 [6] NCCL INFO Connected NVLS tree
+ip-26-0-160-225:124166:124339 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-225:124166:124339 [6] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-160-225:124166:124339 [6] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-165-202:153567:153742 [5] NCCL INFO Connected NVLS tree
+ip-26-0-165-202:153567:153742 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-202:153567:153742 [5] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-165-202:153567:153742 [5] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-168-30:235654:235821 [6] NCCL INFO Connected NVLS tree
+ip-26-0-168-30:235654:235821 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-30:235654:235821 [6] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-168-30:235654:235821 [6] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-160-225:124162:124342 [2] NCCL INFO Connected NVLS tree
+ip-26-0-160-225:124162:124342 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-225:124162:124342 [2] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-160-225:124162:124342 [2] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-103:96894:97068 [6] NCCL INFO Connected NVLS tree
+ip-26-0-161-103:96894:97068 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-103:96894:97068 [6] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-161-103:96894:97068 [6] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-168-30:235648:235826 [0] NCCL INFO Connected NVLS tree
+ip-26-0-168-30:235648:235826 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-30:235648:235826 [0] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-168-30:235648:235826 [0] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-103:96890:97070 [2] NCCL INFO Connected NVLS tree
+ip-26-0-161-103:96890:97070 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-103:96890:97070 [2] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-161-103:96890:97070 [2] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-123:32707:32878 [3] NCCL INFO Connected NVLS tree
+ip-26-0-161-123:32707:32878 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-123:32707:32878 [3] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-161-123:32707:32878 [3] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO Connected NVLS tree
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-168-34:272152:272331 [2] NCCL INFO Connected NVLS tree
+ip-26-0-168-34:272152:272331 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-34:272152:272331 [2] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-168-34:272152:272331 [2] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-103:96892:97064 [4] NCCL INFO Connected NVLS tree
+ip-26-0-161-103:96892:97064 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-103:96892:97064 [4] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-161-103:96892:97064 [4] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO Connected NVLS tree
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO Connected NVLS tree
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-168-30:235651:235823 [3] NCCL INFO Connected NVLS tree
+ip-26-0-168-30:235651:235823 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-30:235651:235823 [3] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-168-30:235651:235823 [3] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-138:96564:96757 [1] NCCL INFO Connected NVLS tree
+ip-26-0-161-138:96564:96757 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-138:96564:96757 [1] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-161-138:96564:96757 [1] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-165-202:153562:153735 [0] NCCL INFO Connected NVLS tree
+ip-26-0-165-202:153562:153735 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-202:153562:153735 [0] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-165-202:153562:153735 [0] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO Connected NVLS tree
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-168-34:272157:272324 [7] NCCL INFO Connected NVLS tree
+ip-26-0-168-34:272157:272324 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-34:272157:272324 [7] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-168-34:272157:272324 [7] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-168-34:272153:272328 [3] NCCL INFO Connected NVLS tree
+ip-26-0-168-34:272153:272328 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-34:272153:272328 [3] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-168-34:272153:272328 [3] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-103:96893:97067 [5] NCCL INFO Connected NVLS tree
+ip-26-0-161-103:96893:97067 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-103:96893:97067 [5] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-161-103:96893:97067 [5] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-168-30:235650:235824 [2] NCCL INFO Connected NVLS tree
+ip-26-0-168-30:235650:235824 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-30:235650:235824 [2] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-168-30:235650:235824 [2] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-168-52:2722050:2722225 [5] NCCL INFO Connected NVLS tree
+ip-26-0-168-52:2722050:2722225 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-52:2722050:2722225 [5] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-168-52:2722050:2722225 [5] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO Connected NVLS tree
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO Connected NVLS tree
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO Connected NVLS tree
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-165-202:153569:153739 [7] NCCL INFO Connected NVLS tree
+ip-26-0-165-202:153569:153739 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-202:153569:153739 [7] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-165-202:153569:153739 [7] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO Connected NVLS tree
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO Connected NVLS tree
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-103:96889:97065 [1] NCCL INFO Connected NVLS tree
+ip-26-0-161-103:96889:97065 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-103:96889:97065 [1] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-161-103:96889:97065 [1] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-123:32706:32877 [2] NCCL INFO Connected NVLS tree
+ip-26-0-161-123:32706:32877 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-123:32706:32877 [2] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-161-123:32706:32877 [2] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-168-34:272151:272326 [1] NCCL INFO Connected NVLS tree
+ip-26-0-168-34:272151:272326 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-34:272151:272326 [1] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-168-34:272151:272326 [1] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-168-52:2722049:2722222 [4] NCCL INFO comm 0x8818910 rank 124 nranks 128 cudaDev 4 nvmlDev 4 busId 97000 commId 0xe67d0548a5bfb9af - Init COMPLETE
+ip-26-0-168-52:2722051:2722224 [6] NCCL INFO comm 0x92559c0 rank 126 nranks 128 cudaDev 6 nvmlDev 6 busId b9000 commId 0xe67d0548a5bfb9af - Init COMPLETE
+ip-26-0-168-52:2722052:2722226 [7] NCCL INFO comm 0x9675870 rank 127 nranks 128 cudaDev 7 nvmlDev 7 busId ca000 commId 0xe67d0548a5bfb9af - Init COMPLETE
+ip-26-0-168-52:2722046:2722223 [1] NCCL INFO comm 0x9800e30 rank 121 nranks 128 cudaDev 1 nvmlDev 1 busId 64000 commId 0xe67d0548a5bfb9af - Init COMPLETE
+ip-26-0-168-52:2722048:2722220 [3] NCCL INFO comm 0x864def0 rank 123 nranks 128 cudaDev 3 nvmlDev 3 busId 86000 commId 0xe67d0548a5bfb9af - Init COMPLETE
+ip-26-0-168-52:2722047:2722221 [2] NCCL INFO comm 0x94226d0 rank 122 nranks 128 cudaDev 2 nvmlDev 2 busId 75000 commId 0xe67d0548a5bfb9af - Init COMPLETE
+ip-26-0-168-52:2722050:2722225 [5] NCCL INFO comm 0x86815d0 rank 125 nranks 128 cudaDev 5 nvmlDev 5 busId a8000 commId 0xe67d0548a5bfb9af - Init COMPLETE
+ip-26-0-168-52:2722045:2722219 [0] NCCL INFO comm 0x852ebd0 rank 120 nranks 128 cudaDev 0 nvmlDev 0 busId 53000 commId 0xe67d0548a5bfb9af - Init COMPLETE
+ip-26-0-168-30:235652:235822 [4] NCCL INFO Connected NVLS tree
+ip-26-0-168-30:235652:235822 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-30:235652:235822 [4] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-168-30:235652:235822 [4] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-168-30:235649:235827 [1] NCCL INFO Connected NVLS tree
+ip-26-0-168-30:235649:235827 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-30:235649:235827 [1] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-168-30:235649:235827 [1] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO Connected NVLS tree
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO Connected NVLS tree
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-178:2815966:2816154 [3] NCCL INFO comm 0x94a1240 rank 59 nranks 128 cudaDev 3 nvmlDev 3 busId 86000 commId 0xe67d0548a5bfb9af - Init COMPLETE
+ip-26-0-161-178:2815964:2816155 [1] NCCL INFO comm 0x8793c10 rank 57 nranks 128 cudaDev 1 nvmlDev 1 busId 64000 commId 0xe67d0548a5bfb9af - Init COMPLETE
+ip-26-0-161-178:2815965:2816151 [2] NCCL INFO comm 0x8bcfca0 rank 58 nranks 128 cudaDev 2 nvmlDev 2 busId 75000 commId 0xe67d0548a5bfb9af - Init COMPLETE
+ip-26-0-161-178:2815969:2816157 [6] NCCL INFO comm 0x9a50ce0 rank 62 nranks 128 cudaDev 6 nvmlDev 6 busId b9000 commId 0xe67d0548a5bfb9af - Init COMPLETE
+ip-26-0-161-178:2815967:2816153 [4] NCCL INFO comm 0x8629c70 rank 60 nranks 128 cudaDev 4 nvmlDev 4 busId 97000 commId 0xe67d0548a5bfb9af - Init COMPLETE
+ip-26-0-161-178:2815963:2816152 [0] NCCL INFO comm 0x937b130 rank 56 nranks 128 cudaDev 0 nvmlDev 0 busId 53000 commId 0xe67d0548a5bfb9af - Init COMPLETE
+ip-26-0-160-225:124160:124335 [0] NCCL INFO Connected NVLS tree
+ip-26-0-160-225:124160:124335 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-225:124160:124335 [0] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-160-225:124160:124335 [0] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-103:96888:97066 [0] NCCL INFO Connected NVLS tree
+ip-26-0-161-103:96888:97066 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-103:96888:97066 [0] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-161-103:96888:97066 [0] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-178:2815968:2816150 [5] NCCL INFO comm 0xa08a840 rank 61 nranks 128 cudaDev 5 nvmlDev 5 busId a8000 commId 0xe67d0548a5bfb9af - Init COMPLETE
+ip-26-0-168-30:235653:235825 [5] NCCL INFO Connected NVLS tree
+ip-26-0-168-30:235653:235825 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-30:235653:235825 [5] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-168-30:235653:235825 [5] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-178:2815970:2816156 [7] NCCL INFO comm 0x90c0bf0 rank 63 nranks 128 cudaDev 7 nvmlDev 7 busId ca000 commId 0xe67d0548a5bfb9af - Init COMPLETE
+ip-26-0-168-34:272155:272329 [5] NCCL INFO Connected NVLS tree
+ip-26-0-168-34:272155:272329 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-34:272155:272329 [5] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-168-34:272155:272329 [5] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO Connected NVLS tree
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO Connected NVLS tree
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-165-202:153564:153737 [2] NCCL INFO Connected NVLS tree
+ip-26-0-165-202:153564:153737 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-202:153564:153737 [2] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-165-202:153564:153737 [2] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-160-225:124162:124342 [2] NCCL INFO comm 0xa2aa6e0 rank 2 nranks 128 cudaDev 2 nvmlDev 2 busId 75000 commId 0xe67d0548a5bfb9af - Init COMPLETE
+ip-26-0-160-225:124161:124337 [1] NCCL INFO comm 0x87faca0 rank 1 nranks 128 cudaDev 1 nvmlDev 1 busId 64000 commId 0xe67d0548a5bfb9af - Init COMPLETE
+ip-26-0-160-225:124160:124335 [0] NCCL INFO comm 0x921d930 rank 0 nranks 128 cudaDev 0 nvmlDev 0 busId 53000 commId 0xe67d0548a5bfb9af - Init COMPLETE
+ip-26-0-160-225:124167:124338 [7] NCCL INFO comm 0x9c467b0 rank 7 nranks 128 cudaDev 7 nvmlDev 7 busId ca000 commId 0xe67d0548a5bfb9af - Init COMPLETE
+ip-26-0-160-225:124166:124339 [6] NCCL INFO comm 0x975d3b0 rank 6 nranks 128 cudaDev 6 nvmlDev 6 busId b9000 commId 0xe67d0548a5bfb9af - Init COMPLETE
+ip-26-0-160-225:124164:124336 [4] NCCL INFO comm 0x8f6da90 rank 4 nranks 128 cudaDev 4 nvmlDev 4 busId 97000 commId 0xe67d0548a5bfb9af - Init COMPLETE
+ip-26-0-160-225:124163:124340 [3] NCCL INFO comm 0x94c8a80 rank 3 nranks 128 cudaDev 3 nvmlDev 3 busId 86000 commId 0xe67d0548a5bfb9af - Init COMPLETE
+ip-26-0-160-225:124165:124341 [5] NCCL INFO comm 0x8838dc0 rank 5 nranks 128 cudaDev 5 nvmlDev 5 busId a8000 commId 0xe67d0548a5bfb9af - Init COMPLETE
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO Connected NVLS tree
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-103:96893:97067 [5] NCCL INFO comm 0x86ea090 rank 21 nranks 128 cudaDev 5 nvmlDev 5 busId a8000 commId 0xe67d0548a5bfb9af - Init COMPLETE
+ip-26-0-161-103:96891:97063 [3] NCCL INFO comm 0x8c696e0 rank 19 nranks 128 cudaDev 3 nvmlDev 3 busId 86000 commId 0xe67d0548a5bfb9af - Init COMPLETE
+ip-26-0-161-103:96889:97065 [1] NCCL INFO comm 0x87d4fe0 rank 17 nranks 128 cudaDev 1 nvmlDev 1 busId 64000 commId 0xe67d0548a5bfb9af - Init COMPLETE
+ip-26-0-161-103:96895:97069 [7] NCCL INFO comm 0xa4213d0 rank 23 nranks 128 cudaDev 7 nvmlDev 7 busId ca000 commId 0xe67d0548a5bfb9af - Init COMPLETE
+ip-26-0-161-103:96894:97068 [6] NCCL INFO comm 0x8a7aba0 rank 22 nranks 128 cudaDev 6 nvmlDev 6 busId b9000 commId 0xe67d0548a5bfb9af - Init COMPLETE
+ip-26-0-161-103:96890:97070 [2] NCCL INFO comm 0x8e18080 rank 18 nranks 128 cudaDev 2 nvmlDev 2 busId 75000 commId 0xe67d0548a5bfb9af - Init COMPLETE
+ip-26-0-161-103:96892:97064 [4] NCCL INFO comm 0x9b9ac80 rank 20 nranks 128 cudaDev 4 nvmlDev 4 busId 97000 commId 0xe67d0548a5bfb9af - Init COMPLETE
+ip-26-0-161-103:96888:97066 [0] NCCL INFO comm 0x8e3cea0 rank 16 nranks 128 cudaDev 0 nvmlDev 0 busId 53000 commId 0xe67d0548a5bfb9af - Init COMPLETE
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO Connected NVLS tree
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-165-202:153566:153741 [4] NCCL INFO Connected NVLS tree
+ip-26-0-165-202:153566:153741 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-202:153566:153741 [4] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-165-202:153566:153741 [4] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-168-30:235649:235827 [1] NCCL INFO comm 0x9bd9090 rank 105 nranks 128 cudaDev 1 nvmlDev 1 busId 64000 commId 0xe67d0548a5bfb9af - Init COMPLETE
+ip-26-0-168-30:235653:235825 [5] NCCL INFO comm 0x91002f0 rank 109 nranks 128 cudaDev 5 nvmlDev 5 busId a8000 commId 0xe67d0548a5bfb9af - Init COMPLETE
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO Connected NVLS tree
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-168-30:235652:235822 [4] NCCL INFO comm 0x9c80d90 rank 108 nranks 128 cudaDev 4 nvmlDev 4 busId 97000 commId 0xe67d0548a5bfb9af - Init COMPLETE
+ip-26-0-168-30:235648:235826 [0] NCCL INFO comm 0x890ea90 rank 104 nranks 128 cudaDev 0 nvmlDev 0 busId 53000 commId 0xe67d0548a5bfb9af - Init COMPLETE
+ip-26-0-168-30:235654:235821 [6] NCCL INFO comm 0x994c420 rank 110 nranks 128 cudaDev 6 nvmlDev 6 busId b9000 commId 0xe67d0548a5bfb9af - Init COMPLETE
+ip-26-0-168-30:235650:235824 [2] NCCL INFO comm 0x8ef34a0 rank 106 nranks 128 cudaDev 2 nvmlDev 2 busId 75000 commId 0xe67d0548a5bfb9af - Init COMPLETE
+ip-26-0-168-30:235651:235823 [3] NCCL INFO comm 0x8ed2f20 rank 107 nranks 128 cudaDev 3 nvmlDev 3 busId 86000 commId 0xe67d0548a5bfb9af - Init COMPLETE
+ip-26-0-168-30:235655:235828 [7] NCCL INFO comm 0x9447f80 rank 111 nranks 128 cudaDev 7 nvmlDev 7 busId ca000 commId 0xe67d0548a5bfb9af - Init COMPLETE
+ip-26-0-168-34:272152:272331 [2] NCCL INFO comm 0xa466440 rank 114 nranks 128 cudaDev 2 nvmlDev 2 busId 75000 commId 0xe67d0548a5bfb9af - Init COMPLETE
+ip-26-0-168-34:272154:272330 [4] NCCL INFO comm 0x9279c80 rank 116 nranks 128 cudaDev 4 nvmlDev 4 busId 97000 commId 0xe67d0548a5bfb9af - Init COMPLETE
+ip-26-0-168-34:272150:272327 [0] NCCL INFO comm 0x9d62500 rank 112 nranks 128 cudaDev 0 nvmlDev 0 busId 53000 commId 0xe67d0548a5bfb9af - Init COMPLETE
+ip-26-0-168-34:272156:272325 [6] NCCL INFO comm 0xa194790 rank 118 nranks 128 cudaDev 6 nvmlDev 6 busId b9000 commId 0xe67d0548a5bfb9af - Init COMPLETE
+ip-26-0-168-34:272153:272328 [3] NCCL INFO comm 0x85b56b0 rank 115 nranks 128 cudaDev 3 nvmlDev 3 busId 86000 commId 0xe67d0548a5bfb9af - Init COMPLETE
+ip-26-0-168-34:272157:272324 [7] NCCL INFO comm 0x93fc8f0 rank 119 nranks 128 cudaDev 7 nvmlDev 7 busId ca000 commId 0xe67d0548a5bfb9af - Init COMPLETE
+ip-26-0-168-34:272151:272326 [1] NCCL INFO comm 0x9494aa0 rank 113 nranks 128 cudaDev 1 nvmlDev 1 busId 64000 commId 0xe67d0548a5bfb9af - Init COMPLETE
+ip-26-0-168-34:272155:272329 [5] NCCL INFO comm 0x91a3e50 rank 117 nranks 128 cudaDev 5 nvmlDev 5 busId a8000 commId 0xe67d0548a5bfb9af - Init COMPLETE
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO Connected NVLS tree
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-123:32708:32879 [4] NCCL INFO Connected NVLS tree
+ip-26-0-161-123:32708:32879 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-123:32708:32879 [4] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-161-123:32708:32879 [4] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-160-242:1364905:1365088 [3] NCCL INFO comm 0x982c5c0 rank 11 nranks 128 cudaDev 3 nvmlDev 3 busId 86000 commId 0xe67d0548a5bfb9af - Init COMPLETE
+ip-26-0-160-242:1364909:1365086 [7] NCCL INFO comm 0x9e64d60 rank 15 nranks 128 cudaDev 7 nvmlDev 7 busId ca000 commId 0xe67d0548a5bfb9af - Init COMPLETE
+ip-26-0-160-242:1364908:1365089 [6] NCCL INFO comm 0xa3ae1e0 rank 14 nranks 128 cudaDev 6 nvmlDev 6 busId b9000 commId 0xe67d0548a5bfb9af - Init COMPLETE
+ip-26-0-160-242:1364904:1365085 [2] NCCL INFO comm 0x963e6f0 rank 10 nranks 128 cudaDev 2 nvmlDev 2 busId 75000 commId 0xe67d0548a5bfb9af - Init COMPLETE
+ip-26-0-160-242:1364907:1365090 [5] NCCL INFO comm 0x8690d60 rank 13 nranks 128 cudaDev 5 nvmlDev 5 busId a8000 commId 0xe67d0548a5bfb9af - Init COMPLETE
+ip-26-0-160-242:1364903:1365084 [1] NCCL INFO comm 0x9ad1540 rank 9 nranks 128 cudaDev 1 nvmlDev 1 busId 64000 commId 0xe67d0548a5bfb9af - Init COMPLETE
+ip-26-0-160-242:1364906:1365087 [4] NCCL INFO comm 0x9f93670 rank 12 nranks 128 cudaDev 4 nvmlDev 4 busId 97000 commId 0xe67d0548a5bfb9af - Init COMPLETE
+ip-26-0-160-242:1364902:1365083 [0] NCCL INFO comm 0x97add10 rank 8 nranks 128 cudaDev 0 nvmlDev 0 busId 53000 commId 0xe67d0548a5bfb9af - Init COMPLETE
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO Connected NVLS tree
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-123:32704:32881 [0] NCCL INFO Connected NVLS tree
+ip-26-0-161-123:32704:32881 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-123:32704:32881 [0] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-161-123:32704:32881 [0] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO Connected NVLS tree
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-78:114512:114686 [0] NCCL INFO Connected NVLS tree
+ip-26-0-161-78:114512:114686 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-78:114512:114686 [0] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-161-78:114512:114686 [0] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO Connected NVLS tree
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-123:32710:32883 [6] NCCL INFO Connected NVLS tree
+ip-26-0-161-123:32710:32883 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-123:32710:32883 [6] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-161-123:32710:32883 [6] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-78:114518:114693 [6] NCCL INFO Connected NVLS tree
+ip-26-0-161-78:114518:114693 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-78:114518:114693 [6] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-161-78:114518:114693 [6] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-78:114516:114691 [4] NCCL INFO Connected NVLS tree
+ip-26-0-161-78:114516:114691 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-78:114516:114691 [4] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-161-78:114516:114691 [4] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO Connected NVLS tree
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO Connected NVLS tree
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-78:114514:114689 [2] NCCL INFO Connected NVLS tree
+ip-26-0-161-78:114514:114689 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-78:114514:114689 [2] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-161-78:114514:114689 [2] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-123:32709:32882 [5] NCCL INFO comm 0xa23fe10 rank 29 nranks 128 cudaDev 5 nvmlDev 5 busId a8000 commId 0xe67d0548a5bfb9af - Init COMPLETE
+ip-26-0-161-123:32707:32878 [3] NCCL INFO comm 0x887a500 rank 27 nranks 128 cudaDev 3 nvmlDev 3 busId 86000 commId 0xe67d0548a5bfb9af - Init COMPLETE
+ip-26-0-161-123:32706:32877 [2] NCCL INFO comm 0x9a13240 rank 26 nranks 128 cudaDev 2 nvmlDev 2 busId 75000 commId 0xe67d0548a5bfb9af - Init COMPLETE
+ip-26-0-161-123:32705:32880 [1] NCCL INFO comm 0x99eb180 rank 25 nranks 128 cudaDev 1 nvmlDev 1 busId 64000 commId 0xe67d0548a5bfb9af - Init COMPLETE
+ip-26-0-161-123:32708:32879 [4] NCCL INFO comm 0xa27e140 rank 28 nranks 128 cudaDev 4 nvmlDev 4 busId 97000 commId 0xe67d0548a5bfb9af - Init COMPLETE
+ip-26-0-161-123:32704:32881 [0] NCCL INFO comm 0xa45bb70 rank 24 nranks 128 cudaDev 0 nvmlDev 0 busId 53000 commId 0xe67d0548a5bfb9af - Init COMPLETE
+ip-26-0-161-123:32711:32884 [7] NCCL INFO comm 0x9e7ffa0 rank 31 nranks 128 cudaDev 7 nvmlDev 7 busId ca000 commId 0xe67d0548a5bfb9af - Init COMPLETE
+ip-26-0-161-123:32710:32883 [6] NCCL INFO comm 0x991e950 rank 30 nranks 128 cudaDev 6 nvmlDev 6 busId b9000 commId 0xe67d0548a5bfb9af - Init COMPLETE
+ip-26-0-161-138:96570:96759 [7] NCCL INFO Connected NVLS tree
+ip-26-0-161-138:96570:96759 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-138:96570:96759 [7] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-161-138:96570:96759 [7] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO Connected NVLS tree
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-165-164:1227469:1227636 [6] NCCL INFO comm 0x84bbf10 rank 86 nranks 128 cudaDev 6 nvmlDev 6 busId b9000 commId 0xe67d0548a5bfb9af - Init COMPLETE
+ip-26-0-165-164:1227465:1227639 [2] NCCL INFO comm 0x8c25390 rank 82 nranks 128 cudaDev 2 nvmlDev 2 busId 75000 commId 0xe67d0548a5bfb9af - Init COMPLETE
+ip-26-0-165-164:1227467:1227638 [4] NCCL INFO comm 0x91fcaa0 rank 84 nranks 128 cudaDev 4 nvmlDev 4 busId 97000 commId 0xe67d0548a5bfb9af - Init COMPLETE
+ip-26-0-165-164:1227463:1227642 [0] NCCL INFO comm 0x8798710 rank 80 nranks 128 cudaDev 0 nvmlDev 0 busId 53000 commId 0xe67d0548a5bfb9af - Init COMPLETE
+ip-26-0-165-164:1227468:1227643 [5] NCCL INFO comm 0x9b878e0 rank 85 nranks 128 cudaDev 5 nvmlDev 5 busId a8000 commId 0xe67d0548a5bfb9af - Init COMPLETE
+ip-26-0-165-164:1227466:1227641 [3] NCCL INFO comm 0x8e7c100 rank 83 nranks 128 cudaDev 3 nvmlDev 3 busId 86000 commId 0xe67d0548a5bfb9af - Init COMPLETE
+ip-26-0-165-164:1227470:1227640 [7] NCCL INFO comm 0x9298e10 rank 87 nranks 128 cudaDev 7 nvmlDev 7 busId ca000 commId 0xe67d0548a5bfb9af - Init COMPLETE
+ip-26-0-165-164:1227464:1227644 [1] NCCL INFO comm 0x9061f10 rank 81 nranks 128 cudaDev 1 nvmlDev 1 busId 64000 commId 0xe67d0548a5bfb9af - Init COMPLETE
+ip-26-0-161-78:114513:114688 [1] NCCL INFO Connected NVLS tree
+ip-26-0-161-78:114513:114688 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-78:114513:114688 [1] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-161-78:114513:114688 [1] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO Connected NVLS tree
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-78:114517:114687 [5] NCCL INFO Connected NVLS tree
+ip-26-0-161-78:114517:114687 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-78:114517:114687 [5] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-161-78:114517:114687 [5] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-78:114519:114692 [7] NCCL INFO Connected NVLS tree
+ip-26-0-161-78:114519:114692 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-78:114519:114692 [7] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-161-78:114519:114692 [7] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-165-59:1557272:1557442 [6] NCCL INFO comm 0x9406880 rank 102 nranks 128 cudaDev 6 nvmlDev 6 busId b9000 commId 0xe67d0548a5bfb9af - Init COMPLETE
+ip-26-0-165-59:1557266:1557441 [0] NCCL INFO comm 0xa120e40 rank 96 nranks 128 cudaDev 0 nvmlDev 0 busId 53000 commId 0xe67d0548a5bfb9af - Init COMPLETE
+ip-26-0-165-59:1557268:1557440 [2] NCCL INFO comm 0x8b40680 rank 98 nranks 128 cudaDev 2 nvmlDev 2 busId 75000 commId 0xe67d0548a5bfb9af - Init COMPLETE
+ip-26-0-165-59:1557271:1557438 [5] NCCL INFO comm 0x84a7790 rank 101 nranks 128 cudaDev 5 nvmlDev 5 busId a8000 commId 0xe67d0548a5bfb9af - Init COMPLETE
+ip-26-0-165-59:1557270:1557443 [4] NCCL INFO comm 0x8917950 rank 100 nranks 128 cudaDev 4 nvmlDev 4 busId 97000 commId 0xe67d0548a5bfb9af - Init COMPLETE
+ip-26-0-165-59:1557273:1557439 [7] NCCL INFO comm 0x9ff3aa0 rank 103 nranks 128 cudaDev 7 nvmlDev 7 busId ca000 commId 0xe67d0548a5bfb9af - Init COMPLETE
+ip-26-0-165-59:1557267:1557444 [1] NCCL INFO comm 0x889c010 rank 97 nranks 128 cudaDev 1 nvmlDev 1 busId 64000 commId 0xe67d0548a5bfb9af - Init COMPLETE
+ip-26-0-165-59:1557269:1557445 [3] NCCL INFO comm 0x9ef1450 rank 99 nranks 128 cudaDev 3 nvmlDev 3 busId 86000 commId 0xe67d0548a5bfb9af - Init COMPLETE
+ip-26-0-161-78:114515:114690 [3] NCCL INFO Connected NVLS tree
+ip-26-0-161-78:114515:114690 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-78:114515:114690 [3] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-161-78:114515:114690 [3] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-138:96568:96753 [5] NCCL INFO Connected NVLS tree
+ip-26-0-161-138:96568:96753 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-138:96568:96753 [5] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-161-138:96568:96753 [5] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO Connected NVLS tree
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO Connected NVLS tree
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-142:62888:63060 [7] NCCL INFO Connected NVLS tree
+ip-26-0-161-142:62888:63060 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-142:62888:63060 [7] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-161-142:62888:63060 [7] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-142:62882:63056 [1] NCCL INFO Connected NVLS tree
+ip-26-0-161-142:62882:63056 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-142:62882:63056 [1] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-161-142:62882:63056 [1] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-138:96566:96755 [3] NCCL INFO Connected NVLS tree
+ip-26-0-161-138:96566:96755 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-138:96566:96755 [3] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-161-138:96566:96755 [3] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO Connected NVLS tree
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-78:114514:114689 [2] NCCL INFO comm 0x9cf4bf0 rank 66 nranks 128 cudaDev 2 nvmlDev 2 busId 75000 commId 0xe67d0548a5bfb9af - Init COMPLETE
+ip-26-0-161-78:114516:114691 [4] NCCL INFO comm 0x936c450 rank 68 nranks 128 cudaDev 4 nvmlDev 4 busId 97000 commId 0xe67d0548a5bfb9af - Init COMPLETE
+ip-26-0-161-78:114518:114693 [6] NCCL INFO comm 0x9878640 rank 70 nranks 128 cudaDev 6 nvmlDev 6 busId b9000 commId 0xe67d0548a5bfb9af - Init COMPLETE
+ip-26-0-161-78:114519:114692 [7] NCCL INFO comm 0x84be170 rank 71 nranks 128 cudaDev 7 nvmlDev 7 busId ca000 commId 0xe67d0548a5bfb9af - Init COMPLETE
+ip-26-0-161-78:114512:114686 [0] NCCL INFO comm 0x9281680 rank 64 nranks 128 cudaDev 0 nvmlDev 0 busId 53000 commId 0xe67d0548a5bfb9af - Init COMPLETE
+ip-26-0-161-78:114517:114687 [5] NCCL INFO comm 0xa282530 rank 69 nranks 128 cudaDev 5 nvmlDev 5 busId a8000 commId 0xe67d0548a5bfb9af - Init COMPLETE
+ip-26-0-161-142:62886:63057 [5] NCCL INFO Connected NVLS tree
+ip-26-0-161-142:62886:63057 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-142:62886:63057 [5] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-161-142:62886:63057 [5] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-138:96565:96754 [2] NCCL INFO Connected NVLS tree
+ip-26-0-161-138:96565:96754 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-138:96565:96754 [2] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-161-138:96565:96754 [2] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO Connected NVLS tree
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-142:62884:63061 [3] NCCL INFO Connected NVLS tree
+ip-26-0-161-142:62884:63061 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-142:62884:63061 [3] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-161-142:62884:63061 [3] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-142:62883:63058 [2] NCCL INFO Connected NVLS tree
+ip-26-0-161-142:62883:63058 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-142:62883:63058 [2] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-161-142:62883:63058 [2] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO Connected NVLS tree
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-138:96569:96758 [6] NCCL INFO Connected NVLS tree
+ip-26-0-161-138:96569:96758 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-138:96569:96758 [6] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-161-138:96569:96758 [6] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-138:96567:96760 [4] NCCL INFO Connected NVLS tree
+ip-26-0-161-138:96567:96760 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-138:96567:96760 [4] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-161-138:96567:96760 [4] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-78:114513:114688 [1] NCCL INFO comm 0x9e36040 rank 65 nranks 128 cudaDev 1 nvmlDev 1 busId 64000 commId 0xe67d0548a5bfb9af - Init COMPLETE
+ip-26-0-161-78:114515:114690 [3] NCCL INFO comm 0x8dcd620 rank 67 nranks 128 cudaDev 3 nvmlDev 3 busId 86000 commId 0xe67d0548a5bfb9af - Init COMPLETE
+ip-26-0-161-138:96563:96756 [0] NCCL INFO Connected NVLS tree
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO Connected NVLS tree
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-138:96563:96756 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-138:96563:96756 [0] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-161-138:96563:96756 [0] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-142:62887:63055 [6] NCCL INFO Connected NVLS tree
+ip-26-0-161-142:62887:63055 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-142:62887:63055 [6] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-161-142:62887:63055 [6] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-142:62885:63054 [4] NCCL INFO Connected NVLS tree
+ip-26-0-161-142:62885:63054 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-142:62885:63054 [4] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-161-142:62885:63054 [4] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO Connected NVLS tree
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-142:62881:63059 [0] NCCL INFO Connected NVLS tree
+ip-26-0-161-142:62881:63059 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-142:62881:63059 [0] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-161-142:62881:63059 [0] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-138:96565:96754 [2] NCCL INFO comm 0xa208380 rank 34 nranks 128 cudaDev 2 nvmlDev 2 busId 75000 commId 0xe67d0548a5bfb9af - Init COMPLETE
+ip-26-0-161-138:96569:96758 [6] NCCL INFO comm 0x8e85570 rank 38 nranks 128 cudaDev 6 nvmlDev 6 busId b9000 commId 0xe67d0548a5bfb9af - Init COMPLETE
+ip-26-0-161-138:96568:96753 [5] NCCL INFO comm 0x88d80e0 rank 37 nranks 128 cudaDev 5 nvmlDev 5 busId a8000 commId 0xe67d0548a5bfb9af - Init COMPLETE
+ip-26-0-161-138:96566:96755 [3] NCCL INFO comm 0x9315300 rank 35 nranks 128 cudaDev 3 nvmlDev 3 busId 86000 commId 0xe67d0548a5bfb9af - Init COMPLETE
+ip-26-0-161-138:96563:96756 [0] NCCL INFO comm 0x99b2260 rank 32 nranks 128 cudaDev 0 nvmlDev 0 busId 53000 commId 0xe67d0548a5bfb9af - Init COMPLETE
+ip-26-0-161-138:96567:96760 [4] NCCL INFO comm 0x886b530 rank 36 nranks 128 cudaDev 4 nvmlDev 4 busId 97000 commId 0xe67d0548a5bfb9af - Init COMPLETE
+ip-26-0-161-138:96570:96759 [7] NCCL INFO comm 0xa0c08e0 rank 39 nranks 128 cudaDev 7 nvmlDev 7 busId ca000 commId 0xe67d0548a5bfb9af - Init COMPLETE
+ip-26-0-161-138:96564:96757 [1] NCCL INFO comm 0x96232d0 rank 33 nranks 128 cudaDev 1 nvmlDev 1 busId 64000 commId 0xe67d0548a5bfb9af - Init COMPLETE
+ip-26-0-161-153:2636089:2636264 [5] NCCL INFO comm 0x871cb90 rank 53 nranks 128 cudaDev 5 nvmlDev 5 busId a8000 commId 0xe67d0548a5bfb9af - Init COMPLETE
+ip-26-0-161-153:2636087:2636260 [3] NCCL INFO comm 0x88a2150 rank 51 nranks 128 cudaDev 3 nvmlDev 3 busId 86000 commId 0xe67d0548a5bfb9af - Init COMPLETE
+ip-26-0-161-153:2636090:2636258 [6] NCCL INFO comm 0x9a07a60 rank 54 nranks 128 cudaDev 6 nvmlDev 6 busId b9000 commId 0xe67d0548a5bfb9af - Init COMPLETE
+ip-26-0-161-153:2636086:2636261 [2] NCCL INFO comm 0x994f620 rank 50 nranks 128 cudaDev 2 nvmlDev 2 busId 75000 commId 0xe67d0548a5bfb9af - Init COMPLETE
+ip-26-0-161-153:2636091:2636265 [7] NCCL INFO comm 0x90028a0 rank 55 nranks 128 cudaDev 7 nvmlDev 7 busId ca000 commId 0xe67d0548a5bfb9af - Init COMPLETE
+ip-26-0-161-153:2636088:2636263 [4] NCCL INFO comm 0x929efe0 rank 52 nranks 128 cudaDev 4 nvmlDev 4 busId 97000 commId 0xe67d0548a5bfb9af - Init COMPLETE
+ip-26-0-161-153:2636084:2636262 [0] NCCL INFO comm 0x8f46e20 rank 48 nranks 128 cudaDev 0 nvmlDev 0 busId 53000 commId 0xe67d0548a5bfb9af - Init COMPLETE
+ip-26-0-161-153:2636085:2636259 [1] NCCL INFO comm 0xa1c3e40 rank 49 nranks 128 cudaDev 1 nvmlDev 1 busId 64000 commId 0xe67d0548a5bfb9af - Init COMPLETE
+ip-26-0-161-142:62886:63057 [5] NCCL INFO comm 0x9de8db0 rank 45 nranks 128 cudaDev 5 nvmlDev 5 busId a8000 commId 0xe67d0548a5bfb9af - Init COMPLETE
+ip-26-0-161-142:62888:63060 [7] NCCL INFO comm 0x91600d0 rank 47 nranks 128 cudaDev 7 nvmlDev 7 busId ca000 commId 0xe67d0548a5bfb9af - Init COMPLETE
+ip-26-0-161-142:62884:63061 [3] NCCL INFO comm 0x95b82f0 rank 43 nranks 128 cudaDev 3 nvmlDev 3 busId 86000 commId 0xe67d0548a5bfb9af - Init COMPLETE
+ip-26-0-161-142:62882:63056 [1] NCCL INFO comm 0x8ed2140 rank 41 nranks 128 cudaDev 1 nvmlDev 1 busId 64000 commId 0xe67d0548a5bfb9af - Init COMPLETE
+ip-26-0-161-142:62883:63058 [2] NCCL INFO comm 0x8b38920 rank 42 nranks 128 cudaDev 2 nvmlDev 2 busId 75000 commId 0xe67d0548a5bfb9af - Init COMPLETE
+ip-26-0-161-142:62885:63054 [4] NCCL INFO comm 0x9e18d20 rank 44 nranks 128 cudaDev 4 nvmlDev 4 busId 97000 commId 0xe67d0548a5bfb9af - Init COMPLETE
+ip-26-0-161-142:62887:63055 [6] NCCL INFO comm 0x867a2a0 rank 46 nranks 128 cudaDev 6 nvmlDev 6 busId b9000 commId 0xe67d0548a5bfb9af - Init COMPLETE
+ip-26-0-161-142:62881:63059 [0] NCCL INFO comm 0xa10dc70 rank 40 nranks 128 cudaDev 0 nvmlDev 0 busId 53000 commId 0xe67d0548a5bfb9af - Init COMPLETE
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO Connected NVLS tree
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO Connected NVLS tree
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO Connected NVLS tree
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO Connected NVLS tree
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO Connected NVLS tree
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-165-202:153568:153736 [6] NCCL INFO Connected NVLS tree
+ip-26-0-165-202:153568:153736 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-202:153568:153736 [6] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-165-202:153568:153736 [6] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO Connected NVLS tree
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO Connected NVLS tree
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-165-202:153569:153739 [7] NCCL INFO comm 0x87276c0 rank 95 nranks 128 cudaDev 7 nvmlDev 7 busId ca000 commId 0xe67d0548a5bfb9af - Init COMPLETE
+ip-26-0-165-202:153568:153736 [6] NCCL INFO comm 0x87ae600 rank 94 nranks 128 cudaDev 6 nvmlDev 6 busId b9000 commId 0xe67d0548a5bfb9af - Init COMPLETE
+ip-26-0-165-202:153562:153735 [0] NCCL INFO comm 0x948d480 rank 88 nranks 128 cudaDev 0 nvmlDev 0 busId 53000 commId 0xe67d0548a5bfb9af - Init COMPLETE
+ip-26-0-165-202:153564:153737 [2] NCCL INFO comm 0x87bf990 rank 90 nranks 128 cudaDev 2 nvmlDev 2 busId 75000 commId 0xe67d0548a5bfb9af - Init COMPLETE
+ip-26-0-165-202:153567:153742 [5] NCCL INFO comm 0xa1c25f0 rank 93 nranks 128 cudaDev 5 nvmlDev 5 busId a8000 commId 0xe67d0548a5bfb9af - Init COMPLETE
+ip-26-0-165-202:153565:153738 [3] NCCL INFO comm 0x9998430 rank 91 nranks 128 cudaDev 3 nvmlDev 3 busId 86000 commId 0xe67d0548a5bfb9af - Init COMPLETE
+ip-26-0-165-202:153563:153740 [1] NCCL INFO comm 0xa3c8c40 rank 89 nranks 128 cudaDev 1 nvmlDev 1 busId 64000 commId 0xe67d0548a5bfb9af - Init COMPLETE
+ip-26-0-165-202:153566:153741 [4] NCCL INFO comm 0x8a05910 rank 92 nranks 128 cudaDev 4 nvmlDev 4 busId 97000 commId 0xe67d0548a5bfb9af - Init COMPLETE
+ip-26-0-165-131:1630208:1630395 [3] NCCL INFO comm 0x84c9ed0 rank 75 nranks 128 cudaDev 3 nvmlDev 3 busId 86000 commId 0xe67d0548a5bfb9af - Init COMPLETE
+ip-26-0-165-131:1630212:1630402 [7] NCCL INFO comm 0x86bcdf0 rank 79 nranks 128 cudaDev 7 nvmlDev 7 busId ca000 commId 0xe67d0548a5bfb9af - Init COMPLETE
+ip-26-0-165-131:1630210:1630400 [5] NCCL INFO comm 0x9e94a10 rank 77 nranks 128 cudaDev 5 nvmlDev 5 busId a8000 commId 0xe67d0548a5bfb9af - Init COMPLETE
+ip-26-0-165-131:1630206:1630398 [1] NCCL INFO comm 0x9b32d60 rank 73 nranks 128 cudaDev 1 nvmlDev 1 busId 64000 commId 0xe67d0548a5bfb9af - Init COMPLETE
+ip-26-0-165-131:1630211:1630399 [6] NCCL INFO comm 0xa46efe0 rank 78 nranks 128 cudaDev 6 nvmlDev 6 busId b9000 commId 0xe67d0548a5bfb9af - Init COMPLETE
+ip-26-0-165-131:1630207:1630401 [2] NCCL INFO comm 0x954a920 rank 74 nranks 128 cudaDev 2 nvmlDev 2 busId 75000 commId 0xe67d0548a5bfb9af - Init COMPLETE
+ip-26-0-165-131:1630205:1630397 [0] NCCL INFO comm 0x8e6e9d0 rank 72 nranks 128 cudaDev 0 nvmlDev 0 busId 53000 commId 0xe67d0548a5bfb9af - Init COMPLETE
+ip-26-0-165-131:1630209:1630396 [4] NCCL INFO comm 0x91b2fc0 rank 76 nranks 128 cudaDev 4 nvmlDev 4 busId 97000 commId 0xe67d0548a5bfb9af - Init COMPLETE
+ip-26-0-160-225:124160:124581 [0] NCCL INFO Using network Libfabric
+ip-26-0-160-225:124166:124583 [6] NCCL INFO Using network Libfabric
+ip-26-0-160-242:1364909:1365327 [7] NCCL INFO Using network Libfabric
+ip-26-0-160-242:1364905:1365328 [3] NCCL INFO Using network Libfabric
+ip-26-0-160-242:1364907:1365329 [5] NCCL INFO Using network Libfabric
+ip-26-0-160-225:124164:124585 [4] NCCL INFO Using network Libfabric
+ip-26-0-160-225:124167:124586 [7] NCCL INFO Using network Libfabric
+ip-26-0-160-225:124162:124587 [2] NCCL INFO Using network Libfabric
+ip-26-0-165-131:1630205:1630646 [0] NCCL INFO Using network Libfabric
+ip-26-0-165-131:1630206:1630647 [1] NCCL INFO Using network Libfabric
+ip-26-0-161-153:2636090:2636505 [6] NCCL INFO Using network Libfabric
+ip-26-0-161-153:2636086:2636507 [2] NCCL INFO Using network Libfabric
+ip-26-0-160-225:124163:124588 [3] NCCL INFO Using network Libfabric
+ip-26-0-160-225:124161:124582 [1] NCCL INFO Using network Libfabric
+ip-26-0-161-123:32706:33122 [2] NCCL INFO Using network Libfabric
+ip-26-0-161-153:2636089:2636508 [5] NCCL INFO Using network Libfabric
+ip-26-0-168-34:272156:272570 [6] NCCL INFO Using network Libfabric
+ip-26-0-168-34:272151:272571 [1] NCCL INFO Using network Libfabric
+ip-26-0-161-103:96888:97306 [0] NCCL INFO Using network Libfabric
+ip-26-0-161-123:32707:33124 [3] NCCL INFO Using network Libfabric
+ip-26-0-161-123:32711:33127 [7] NCCL INFO Using network Libfabric
+ip-26-0-165-131:1630207:1630648 [2] NCCL INFO Using network Libfabric
+ip-26-0-161-153:2636085:2636509 [1] NCCL INFO Using network Libfabric
+ip-26-0-168-34:272157:272572 [7] NCCL INFO Using network Libfabric
+ip-26-0-168-34:272154:272569 [4] NCCL INFO Using network Libfabric
+ip-26-0-161-123:32709:33129 [5] NCCL INFO Using network Libfabric
+ip-26-0-161-123:32708:33126 [4] NCCL INFO Using network Libfabric
+ip-26-0-168-34:272153:272573 [3] NCCL INFO Using network Libfabric
+ip-26-0-161-103:96892:97308 [4] NCCL INFO Using network Libfabric
+ip-26-0-161-123:32704:33123 [0] NCCL INFO Using network Libfabric
+ip-26-0-168-52:2722046:2722466 [1] NCCL INFO Using network Libfabric
+ip-26-0-161-153:2636087:2636506 [3] NCCL INFO Using network Libfabric
+ip-26-0-161-138:96570:97005 [7] NCCL INFO Using network Libfabric
+ip-26-0-161-138:96563:97003 [0] NCCL INFO Using network Libfabric
+ip-26-0-161-123:32705:33128 [1] NCCL INFO Using network Libfabric
+ip-26-0-161-123:32710:33125 [6] NCCL INFO Using network Libfabric
+ip-26-0-168-34:272152:272575 [2] NCCL INFO Using network Libfabric
+ip-26-0-161-103:96895:97307 [7] NCCL INFO Using network Libfabric
+ip-26-0-168-52:2722049:2722467 [4] NCCL INFO Using network Libfabric
+ip-26-0-161-178:2815966:2816396 [3] NCCL INFO Using network Libfabric
+ip-26-0-161-178:2815968:2816394 [5] NCCL INFO Using network Libfabric
+ip-26-0-161-138:96569:97006 [6] NCCL INFO Using network Libfabric
+ip-26-0-161-153:2636088:2636510 [4] NCCL INFO Using network Libfabric
+ip-26-0-161-103:96889:97309 [1] NCCL INFO Using network Libfabric
+ip-26-0-168-52:2722048:2722469 [3] NCCL INFO Using network Libfabric
+ip-26-0-161-138:96568:97007 [5] NCCL INFO Using network Libfabric
+ip-26-0-161-142:62882:63300 [1] NCCL INFO Using network Libfabric
+ip-26-0-161-178:2815965:2816395 [2] NCCL INFO Using network Libfabric
+ip-26-0-161-138:96567:97008 [4] NCCL INFO Using network Libfabric
+ip-26-0-168-30:235648:236064 [0] NCCL INFO Using network Libfabric
+ip-26-0-161-178:2815970:2816397 [7] NCCL INFO Using network Libfabric
+ip-26-0-168-52:2722045:2722470 [0] NCCL INFO Using network Libfabric
+ip-26-0-161-78:114513:114933 [1] NCCL INFO Using network Libfabric
+ip-26-0-161-103:96893:97310 [5] NCCL INFO Using network Libfabric
+ip-26-0-161-153:2636084:2636511 [0] NCCL INFO Using network Libfabric
+ip-26-0-160-225:124165:124584 [5] NCCL INFO Using network Libfabric
+ip-26-0-168-30:235652:236066 [4] NCCL INFO Using network Libfabric
+ip-26-0-168-34:272155:272574 [5] NCCL INFO Using network Libfabric
+ip-26-0-161-178:2815969:2816398 [6] NCCL INFO Using network Libfabric
+ip-26-0-160-242:1364908:1365332 [6] NCCL INFO Using network Libfabric
+ip-26-0-168-52:2722052:2722464 [7] NCCL INFO Using network Libfabric
+ip-26-0-161-138:96564:97009 [1] NCCL INFO Using network Libfabric
+ip-26-0-161-142:62888:63302 [7] NCCL INFO Using network Libfabric
+ip-26-0-165-59:1557269:1557684 [3] NCCL INFO Using network Libfabric
+ip-26-0-165-59:1557267:1557683 [1] NCCL INFO Using network Libfabric
+ip-26-0-168-30:235653:236067 [5] NCCL INFO Using network Libfabric
+ip-26-0-168-30:235651:236068 [3] NCCL INFO Using network Libfabric
+ip-26-0-161-103:96891:97311 [3] NCCL INFO Using network Libfabric
+ip-26-0-161-178:2815967:2816399 [4] NCCL INFO Using network Libfabric
+ip-26-0-160-242:1364902:1365333 [0] NCCL INFO Using network Libfabric
+ip-26-0-168-52:2722050:2722471 [5] NCCL INFO Using network Libfabric
+ip-26-0-168-52:2722051:2722465 [6] NCCL INFO Using network Libfabric
+ip-26-0-161-142:62886:63303 [5] NCCL INFO Using network Libfabric
+ip-26-0-168-30:235655:236069 [7] NCCL INFO Using network Libfabric
+ip-26-0-161-103:96894:97312 [6] NCCL INFO Using network Libfabric
+ip-26-0-165-202:153567:153979 [5] NCCL INFO Using network Libfabric
+ip-26-0-161-78:114514:114932 [2] NCCL INFO Using network Libfabric
+ip-26-0-161-78:114516:114934 [4] NCCL INFO Using network Libfabric
+ip-26-0-161-78:114518:114937 [6] NCCL INFO Using network Libfabric
+ip-26-0-161-78:114519:114931 [7] NCCL INFO Using network Libfabric
+ip-26-0-161-142:62885:63304 [4] NCCL INFO Using network Libfabric
+ip-26-0-161-142:62881:63298 [0] NCCL INFO Using network Libfabric
+ip-26-0-161-142:62887:63305 [6] NCCL INFO Using network Libfabric
+ip-26-0-165-59:1557268:1557686 [2] NCCL INFO Using network Libfabric
+ip-26-0-165-59:1557272:1557687 [6] NCCL INFO Using network Libfabric
+ip-26-0-165-59:1557270:1557688 [4] NCCL INFO Using network Libfabric
+ip-26-0-165-131:1630211:1630649 [6] NCCL INFO Using network Libfabric
+ip-26-0-165-59:1557273:1557685 [7] NCCL INFO Using network Libfabric
+ip-26-0-161-178:2815964:2816401 [1] NCCL INFO Using network Libfabric
+ip-26-0-160-242:1364903:1365334 [1] NCCL INFO Using network Libfabric
+ip-26-0-161-78:114512:114930 [0] NCCL INFO Using network Libfabric
+ip-26-0-161-138:96566:97010 [3] NCCL INFO Using network Libfabric
+ip-26-0-161-103:96890:97313 [2] NCCL INFO Using network Libfabric
+ip-26-0-165-59:1557266:1557689 [0] NCCL INFO Using network Libfabric
+ip-26-0-161-78:114515:114936 [3] NCCL INFO Using network Libfabric
+ip-26-0-168-30:235649:236071 [1] NCCL INFO Using network Libfabric
+ip-26-0-165-164:1227465:1227881 [2] NCCL INFO Using network Libfabric
+ip-26-0-161-142:62883:63301 [2] NCCL INFO Using network Libfabric
+ip-26-0-168-34:272150:272576 [0] NCCL INFO Using network Libfabric
+ip-26-0-160-242:1364906:1365331 [4] NCCL INFO Using network Libfabric
+ip-26-0-161-138:96565:97004 [2] NCCL INFO Using network Libfabric
+ip-26-0-165-202:153564:153984 [2] NCCL INFO Using network Libfabric
+ip-26-0-168-30:235654:236070 [6] NCCL INFO Using network Libfabric
+ip-26-0-165-164:1227463:1227880 [0] NCCL INFO Using network Libfabric
+ip-26-0-165-164:1227464:1227884 [1] NCCL INFO Using network Libfabric
+ip-26-0-165-131:1630209:1630651 [4] NCCL INFO Using network Libfabric
+ip-26-0-165-131:1630212:1630652 [7] NCCL INFO Using network Libfabric
+ip-26-0-161-153:2636091:2636512 [7] NCCL INFO Using network Libfabric
+ip-26-0-168-52:2722047:2722468 [2] NCCL INFO Using network Libfabric
+ip-26-0-165-164:1227470:1227885 [7] NCCL INFO Using network Libfabric
+ip-26-0-165-164:1227467:1227886 [4] NCCL INFO Using network Libfabric
+ip-26-0-165-131:1630210:1630653 [5] NCCL INFO Using network Libfabric
+ip-26-0-165-202:153565:153985 [3] NCCL INFO Using network Libfabric
+ip-26-0-165-202:153562:153983 [0] NCCL INFO Using network Libfabric
+ip-26-0-165-164:1227468:1227883 [5] NCCL INFO Using network Libfabric
+ip-26-0-165-202:153563:153986 [1] NCCL INFO Using network Libfabric
+ip-26-0-165-59:1557271:1557690 [5] NCCL INFO Using network Libfabric
+ip-26-0-165-164:1227469:1227882 [6] NCCL INFO Using network Libfabric
+ip-26-0-161-78:114517:114935 [5] NCCL INFO Using network Libfabric
+ip-26-0-165-164:1227466:1227887 [3] NCCL INFO Using network Libfabric
+ip-26-0-161-142:62884:63299 [3] NCCL INFO Using network Libfabric
+ip-26-0-160-242:1364904:1365335 [2] NCCL INFO Using network Libfabric
+ip-26-0-165-202:153566:153980 [4] NCCL INFO Using network Libfabric
+ip-26-0-165-202:153569:153981 [7] NCCL INFO Using network Libfabric
+ip-26-0-165-202:153568:153982 [6] NCCL INFO Using network Libfabric
+ip-26-0-168-30:235650:236065 [2] NCCL INFO Using network Libfabric
+ip-26-0-161-178:2815963:2816400 [0] NCCL INFO Using network Libfabric
+ip-26-0-165-131:1630208:1630650 [3] NCCL INFO Using network Libfabric
+ip-26-0-160-225:124160:124581 [0] NCCL INFO comm 0x94a3500 rank 0 nranks 128 cudaDev 0 nvmlDev 0 busId 53000 commId 0x759f85fd5295c628 - Init START
+ip-26-0-168-52:2722052:2722464 [7] NCCL INFO comm 0x98d4300 rank 127 nranks 128 cudaDev 7 nvmlDev 7 busId ca000 commId 0x759f85fd5295c628 - Init START
+ip-26-0-168-52:2722051:2722465 [6] NCCL INFO comm 0x94b31c0 rank 126 nranks 128 cudaDev 6 nvmlDev 6 busId b9000 commId 0x759f85fd5295c628 - Init START
+ip-26-0-168-52:2722049:2722467 [4] NCCL INFO comm 0x8a76060 rank 124 nranks 128 cudaDev 4 nvmlDev 4 busId 97000 commId 0x759f85fd5295c628 - Init START
+ip-26-0-168-52:2722048:2722469 [3] NCCL INFO comm 0x88a3670 rank 123 nranks 128 cudaDev 3 nvmlDev 3 busId 86000 commId 0x759f85fd5295c628 - Init START
+ip-26-0-168-52:2722050:2722471 [5] NCCL INFO comm 0x88d9840 rank 125 nranks 128 cudaDev 5 nvmlDev 5 busId a8000 commId 0x759f85fd5295c628 - Init START
+ip-26-0-165-202:153569:153981 [7] NCCL INFO comm 0x8981bb0 rank 95 nranks 128 cudaDev 7 nvmlDev 7 busId ca000 commId 0x759f85fd5295c628 - Init START
+ip-26-0-165-202:153568:153982 [6] NCCL INFO comm 0x8a079f0 rank 94 nranks 128 cudaDev 6 nvmlDev 6 busId b9000 commId 0x759f85fd5295c628 - Init START
+ip-26-0-165-59:1557270:1557688 [4] NCCL INFO comm 0x8b767c0 rank 100 nranks 128 cudaDev 4 nvmlDev 4 busId 97000 commId 0x759f85fd5295c628 - Init START
+ip-26-0-165-59:1557271:1557690 [5] NCCL INFO comm 0x8702fb0 rank 101 nranks 128 cudaDev 5 nvmlDev 5 busId a8000 commId 0x759f85fd5295c628 - Init START
+ip-26-0-165-59:1557272:1557687 [6] NCCL INFO comm 0x9666ec0 rank 102 nranks 128 cudaDev 6 nvmlDev 6 busId b9000 commId 0x759f85fd5295c628 - Init START
+ip-26-0-165-59:1557267:1557683 [1] NCCL INFO comm 0x8af4bf0 rank 97 nranks 128 cudaDev 1 nvmlDev 1 busId 64000 commId 0x759f85fd5295c628 - Init START
+ip-26-0-165-59:1557269:1557684 [3] NCCL INFO comm 0xa1496b0 rank 99 nranks 128 cudaDev 3 nvmlDev 3 busId 86000 commId 0x759f85fd5295c628 - Init START
+ip-26-0-165-59:1557268:1557686 [2] NCCL INFO comm 0x8d9f540 rank 98 nranks 128 cudaDev 2 nvmlDev 2 busId 75000 commId 0x759f85fd5295c628 - Init START
+ip-26-0-165-164:1227470:1227885 [7] NCCL INFO comm 0x94f35b0 rank 87 nranks 128 cudaDev 7 nvmlDev 7 busId ca000 commId 0x759f85fd5295c628 - Init START
+ip-26-0-165-164:1227469:1227882 [6] NCCL INFO comm 0x871acf0 rank 86 nranks 128 cudaDev 6 nvmlDev 6 busId b9000 commId 0x759f85fd5295c628 - Init START
+ip-26-0-168-34:272157:272572 [7] NCCL INFO comm 0x965ac80 rank 119 nranks 128 cudaDev 7 nvmlDev 7 busId ca000 commId 0x759f85fd5295c628 - Init START
+ip-26-0-168-34:272156:272570 [6] NCCL INFO comm 0xa3f1b00 rank 118 nranks 128 cudaDev 6 nvmlDev 6 busId b9000 commId 0x759f85fd5295c628 - Init START
+ip-26-0-168-34:272155:272574 [5] NCCL INFO comm 0x93fcd30 rank 117 nranks 128 cudaDev 5 nvmlDev 5 busId a8000 commId 0x759f85fd5295c628 - Init START
+ip-26-0-168-34:272154:272569 [4] NCCL INFO comm 0x94d3e70 rank 116 nranks 128 cudaDev 4 nvmlDev 4 busId 97000 commId 0x759f85fd5295c628 - Init START
+ip-26-0-168-34:272153:272573 [3] NCCL INFO comm 0x880e9b0 rank 115 nranks 128 cudaDev 3 nvmlDev 3 busId 86000 commId 0x759f85fd5295c628 - Init START
+ip-26-0-168-34:272151:272571 [1] NCCL INFO comm 0x96eebf0 rank 113 nranks 128 cudaDev 1 nvmlDev 1 busId 64000 commId 0x759f85fd5295c628 - Init START
+ip-26-0-161-178:2815970:2816397 [7] NCCL INFO comm 0x931bca0 rank 63 nranks 128 cudaDev 7 nvmlDev 7 busId ca000 commId 0x759f85fd5295c628 - Init START
+ip-26-0-161-178:2815969:2816398 [6] NCCL INFO comm 0x9caa670 rank 62 nranks 128 cudaDev 6 nvmlDev 6 busId b9000 commId 0x759f85fd5295c628 - Init START
+ip-26-0-168-52:2722047:2722468 [2] NCCL INFO comm 0x967a780 rank 122 nranks 128 cudaDev 2 nvmlDev 2 busId 75000 commId 0x759f85fd5295c628 - Init START
+ip-26-0-168-52:2722046:2722466 [1] NCCL INFO comm 0x9a58530 rank 121 nranks 128 cudaDev 1 nvmlDev 1 busId 64000 commId 0x759f85fd5295c628 - Init START
+ip-26-0-165-202:153567:153979 [5] NCCL INFO comm 0xa41d470 rank 93 nranks 128 cudaDev 5 nvmlDev 5 busId a8000 commId 0x759f85fd5295c628 - Init START
+ip-26-0-168-30:235655:236069 [7] NCCL INFO comm 0x96a2a80 rank 111 nranks 128 cudaDev 7 nvmlDev 7 busId ca000 commId 0x759f85fd5295c628 - Init START
+ip-26-0-168-30:235653:236067 [5] NCCL INFO comm 0x935a0b0 rank 109 nranks 128 cudaDev 5 nvmlDev 5 busId a8000 commId 0x759f85fd5295c628 - Init START
+ip-26-0-168-30:235648:236064 [0] NCCL INFO comm 0x8b6eec0 rank 104 nranks 128 cudaDev 0 nvmlDev 0 busId 53000 commId 0x759f85fd5295c628 - Init START
+ip-26-0-168-30:235654:236070 [6] NCCL INFO comm 0x9bac4c0 rank 110 nranks 128 cudaDev 6 nvmlDev 6 busId b9000 commId 0x759f85fd5295c628 - Init START
+ip-26-0-168-30:235652:236066 [4] NCCL INFO comm 0x9edc280 rank 108 nranks 128 cudaDev 4 nvmlDev 4 busId 97000 commId 0x759f85fd5295c628 - Init START
+ip-26-0-168-34:272152:272575 [2] NCCL INFO comm 0xa6c77f0 rank 114 nranks 128 cudaDev 2 nvmlDev 2 busId 75000 commId 0x759f85fd5295c628 - Init START
+ip-26-0-168-34:272150:272576 [0] NCCL INFO comm 0x9fbd540 rank 112 nranks 128 cudaDev 0 nvmlDev 0 busId 53000 commId 0x759f85fd5295c628 - Init START
+ip-26-0-165-131:1630212:1630652 [7] NCCL INFO comm 0x891b6c0 rank 79 nranks 128 cudaDev 7 nvmlDev 7 busId ca000 commId 0x759f85fd5295c628 - Init START
+ip-26-0-165-131:1630211:1630649 [6] NCCL INFO comm 0xa6c7430 rank 78 nranks 128 cudaDev 6 nvmlDev 6 busId b9000 commId 0x759f85fd5295c628 - Init START
+ip-26-0-165-131:1630210:1630653 [5] NCCL INFO comm 0xa0edf00 rank 77 nranks 128 cudaDev 5 nvmlDev 5 busId a8000 commId 0x759f85fd5295c628 - Init START
+ip-26-0-165-131:1630209:1630651 [4] NCCL INFO comm 0x940d970 rank 76 nranks 128 cudaDev 4 nvmlDev 4 busId 97000 commId 0x759f85fd5295c628 - Init START
+ip-26-0-165-131:1630207:1630648 [2] NCCL INFO comm 0x97a3d30 rank 74 nranks 128 cudaDev 2 nvmlDev 2 busId 75000 commId 0x759f85fd5295c628 - Init START
+ip-26-0-165-131:1630208:1630650 [3] NCCL INFO comm 0x87280e0 rank 75 nranks 128 cudaDev 3 nvmlDev 3 busId 86000 commId 0x759f85fd5295c628 - Init START
+ip-26-0-161-153:2636091:2636512 [7] NCCL INFO comm 0x925c9f0 rank 55 nranks 128 cudaDev 7 nvmlDev 7 busId ca000 commId 0x759f85fd5295c628 - Init START
+ip-26-0-161-153:2636090:2636505 [6] NCCL INFO comm 0x9c681e0 rank 54 nranks 128 cudaDev 6 nvmlDev 6 busId b9000 commId 0x759f85fd5295c628 - Init START
+ip-26-0-165-202:153562:153983 [0] NCCL INFO comm 0x96ebb40 rank 88 nranks 128 cudaDev 0 nvmlDev 0 busId 53000 commId 0x759f85fd5295c628 - Init START
+ip-26-0-165-202:153564:153984 [2] NCCL INFO comm 0x8a1afc0 rank 90 nranks 128 cudaDev 2 nvmlDev 2 busId 75000 commId 0x759f85fd5295c628 - Init START
+ip-26-0-165-202:153566:153980 [4] NCCL INFO comm 0x8c5b600 rank 92 nranks 128 cudaDev 4 nvmlDev 4 busId 97000 commId 0x759f85fd5295c628 - Init START
+ip-26-0-161-78:114515:114936 [3] NCCL INFO comm 0x902dae0 rank 67 nranks 128 cudaDev 3 nvmlDev 3 busId 86000 commId 0x759f85fd5295c628 - Init START
+ip-26-0-161-78:114514:114932 [2] NCCL INFO comm 0x9f4e0f0 rank 66 nranks 128 cudaDev 2 nvmlDev 2 busId 75000 commId 0x759f85fd5295c628 - Init START
+ip-26-0-161-78:114516:114934 [4] NCCL INFO comm 0x95c5e60 rank 68 nranks 128 cudaDev 4 nvmlDev 4 busId 97000 commId 0x759f85fd5295c628 - Init START
+ip-26-0-161-78:114519:114931 [7] NCCL INFO comm 0x87170f0 rank 71 nranks 128 cudaDev 7 nvmlDev 7 busId ca000 commId 0x759f85fd5295c628 - Init START
+ip-26-0-161-78:114518:114937 [6] NCCL INFO comm 0x9acb870 rank 70 nranks 128 cudaDev 6 nvmlDev 6 busId b9000 commId 0x759f85fd5295c628 - Init START
+ip-26-0-161-78:114513:114933 [1] NCCL INFO comm 0xa08d8f0 rank 65 nranks 128 cudaDev 1 nvmlDev 1 busId 64000 commId 0x759f85fd5295c628 - Init START
+ip-26-0-161-142:62888:63302 [7] NCCL INFO comm 0x93c50f0 rank 47 nranks 128 cudaDev 7 nvmlDev 7 busId ca000 commId 0x759f85fd5295c628 - Init START
+ip-26-0-165-59:1557266:1557689 [0] NCCL INFO comm 0xa380e00 rank 96 nranks 128 cudaDev 0 nvmlDev 0 busId 53000 commId 0x759f85fd5295c628 - Init START
+ip-26-0-165-164:1227468:1227883 [5] NCCL INFO comm 0x9de3570 rank 85 nranks 128 cudaDev 5 nvmlDev 5 busId a8000 commId 0x759f85fd5295c628 - Init START
+ip-26-0-165-164:1227467:1227886 [4] NCCL INFO comm 0x945c240 rank 84 nranks 128 cudaDev 4 nvmlDev 4 busId 97000 commId 0x759f85fd5295c628 - Init START
+ip-26-0-165-202:153563:153986 [1] NCCL INFO comm 0xa623b70 rank 89 nranks 128 cudaDev 1 nvmlDev 1 busId 64000 commId 0x759f85fd5295c628 - Init START
+ip-26-0-165-202:153565:153985 [3] NCCL INFO comm 0x9bf2bb0 rank 91 nranks 128 cudaDev 3 nvmlDev 3 busId 86000 commId 0x759f85fd5295c628 - Init START
+ip-26-0-161-78:114517:114935 [5] NCCL INFO comm 0xa4db9a0 rank 69 nranks 128 cudaDev 5 nvmlDev 5 busId a8000 commId 0x759f85fd5295c628 - Init START
+ip-26-0-161-78:114512:114930 [0] NCCL INFO comm 0x94e08f0 rank 64 nranks 128 cudaDev 0 nvmlDev 0 busId 53000 commId 0x759f85fd5295c628 - Init START
+ip-26-0-165-59:1557273:1557685 [7] NCCL INFO comm 0xa24f4e0 rank 103 nranks 128 cudaDev 7 nvmlDev 7 busId ca000 commId 0x759f85fd5295c628 - Init START
+ip-26-0-168-30:235649:236071 [1] NCCL INFO comm 0x9e33a20 rank 105 nranks 128 cudaDev 1 nvmlDev 1 busId 64000 commId 0x759f85fd5295c628 - Init START
+ip-26-0-168-30:235650:236065 [2] NCCL INFO comm 0x914c6b0 rank 106 nranks 128 cudaDev 2 nvmlDev 2 busId 75000 commId 0x759f85fd5295c628 - Init START
+ip-26-0-168-30:235651:236068 [3] NCCL INFO comm 0x912b830 rank 107 nranks 128 cudaDev 3 nvmlDev 3 busId 86000 commId 0x759f85fd5295c628 - Init START
+ip-26-0-161-178:2815968:2816394 [5] NCCL INFO comm 0xa2e3d30 rank 61 nranks 128 cudaDev 5 nvmlDev 5 busId a8000 commId 0x759f85fd5295c628 - Init START
+ip-26-0-161-178:2815967:2816399 [4] NCCL INFO comm 0x88842c0 rank 60 nranks 128 cudaDev 4 nvmlDev 4 busId 97000 commId 0x759f85fd5295c628 - Init START
+ip-26-0-165-164:1227466:1227887 [3] NCCL INFO comm 0x90d62c0 rank 83 nranks 128 cudaDev 3 nvmlDev 3 busId 86000 commId 0x759f85fd5295c628 - Init START
+ip-26-0-165-164:1227465:1227881 [2] NCCL INFO comm 0x8e8ccf0 rank 82 nranks 128 cudaDev 2 nvmlDev 2 busId 75000 commId 0x759f85fd5295c628 - Init START
+ip-26-0-165-164:1227463:1227880 [0] NCCL INFO comm 0x89f8580 rank 80 nranks 128 cudaDev 0 nvmlDev 0 busId 53000 commId 0x759f85fd5295c628 - Init START
+ip-26-0-165-164:1227464:1227884 [1] NCCL INFO comm 0x92bcaf0 rank 81 nranks 128 cudaDev 1 nvmlDev 1 busId 64000 commId 0x759f85fd5295c628 - Init START
+ip-26-0-165-131:1630206:1630647 [1] NCCL INFO comm 0x9d917e0 rank 73 nranks 128 cudaDev 1 nvmlDev 1 busId 64000 commId 0x759f85fd5295c628 - Init START
+ip-26-0-165-131:1630205:1630646 [0] NCCL INFO comm 0x90c8640 rank 72 nranks 128 cudaDev 0 nvmlDev 0 busId 53000 commId 0x759f85fd5295c628 - Init START
+ip-26-0-161-153:2636089:2636508 [5] NCCL INFO comm 0x897b060 rank 53 nranks 128 cudaDev 5 nvmlDev 5 busId a8000 commId 0x759f85fd5295c628 - Init START
+ip-26-0-161-153:2636088:2636510 [4] NCCL INFO comm 0x94f8db0 rank 52 nranks 128 cudaDev 4 nvmlDev 4 busId 97000 commId 0x759f85fd5295c628 - Init START
+ip-26-0-168-52:2722045:2722470 [0] NCCL INFO comm 0x8786e80 rank 120 nranks 128 cudaDev 0 nvmlDev 0 busId 53000 commId 0x759f85fd5295c628 - Init START
+ip-26-0-161-178:2815966:2816396 [3] NCCL INFO comm 0x96fb430 rank 59 nranks 128 cudaDev 3 nvmlDev 3 busId 86000 commId 0x759f85fd5295c628 - Init START
+ip-26-0-161-178:2815965:2816395 [2] NCCL INFO comm 0x8e2eb00 rank 58 nranks 128 cudaDev 2 nvmlDev 2 busId 75000 commId 0x759f85fd5295c628 - Init START
+ip-26-0-161-178:2815964:2816401 [1] NCCL INFO comm 0x89ef570 rank 57 nranks 128 cudaDev 1 nvmlDev 1 busId 64000 commId 0x759f85fd5295c628 - Init START
+ip-26-0-161-153:2636087:2636506 [3] NCCL INFO comm 0x8b01d20 rank 51 nranks 128 cudaDev 3 nvmlDev 3 busId 86000 commId 0x759f85fd5295c628 - Init START
+ip-26-0-161-153:2636086:2636507 [2] NCCL INFO comm 0x9ba9780 rank 50 nranks 128 cudaDev 2 nvmlDev 2 busId 75000 commId 0x759f85fd5295c628 - Init START
+ip-26-0-161-153:2636085:2636509 [1] NCCL INFO comm 0xa41e5f0 rank 49 nranks 128 cudaDev 1 nvmlDev 1 busId 64000 commId 0x759f85fd5295c628 - Init START
+ip-26-0-161-153:2636084:2636511 [0] NCCL INFO comm 0x91a14c0 rank 48 nranks 128 cudaDev 0 nvmlDev 0 busId 53000 commId 0x759f85fd5295c628 - Init START
+ip-26-0-161-138:96570:97005 [7] NCCL INFO comm 0xa31fb20 rank 39 nranks 128 cudaDev 7 nvmlDev 7 busId ca000 commId 0x759f85fd5295c628 - Init START
+ip-26-0-161-138:96569:97006 [6] NCCL INFO comm 0x90dbd30 rank 38 nranks 128 cudaDev 6 nvmlDev 6 busId b9000 commId 0x759f85fd5295c628 - Init START
+ip-26-0-161-142:62887:63305 [6] NCCL INFO comm 0x88e3b40 rank 46 nranks 128 cudaDev 6 nvmlDev 6 busId b9000 commId 0x759f85fd5295c628 - Init START
+ip-26-0-161-178:2815963:2816400 [0] NCCL INFO comm 0x95d2fb0 rank 56 nranks 128 cudaDev 0 nvmlDev 0 busId 53000 commId 0x759f85fd5295c628 - Init START
+ip-26-0-161-142:62886:63303 [5] NCCL INFO comm 0xa047900 rank 45 nranks 128 cudaDev 5 nvmlDev 5 busId a8000 commId 0x759f85fd5295c628 - Init START
+ip-26-0-161-142:62885:63304 [4] NCCL INFO comm 0xa074600 rank 44 nranks 128 cudaDev 4 nvmlDev 4 busId 97000 commId 0x759f85fd5295c628 - Init START
+ip-26-0-160-242:1364903:1365334 [1] NCCL INFO comm 0x9d31280 rank 9 nranks 128 cudaDev 1 nvmlDev 1 busId 64000 commId 0x759f85fd5295c628 - Init START
+ip-26-0-160-242:1364902:1365333 [0] NCCL INFO comm 0x9a0bd70 rank 8 nranks 128 cudaDev 0 nvmlDev 0 busId 53000 commId 0x759f85fd5295c628 - Init START
+ip-26-0-160-242:1364904:1365335 [2] NCCL INFO comm 0x9899fb0 rank 10 nranks 128 cudaDev 2 nvmlDev 2 busId 75000 commId 0x759f85fd5295c628 - Init START
+ip-26-0-161-123:32710:33125 [6] NCCL INFO comm 0x9b78cf0 rank 30 nranks 128 cudaDev 6 nvmlDev 6 busId b9000 commId 0x759f85fd5295c628 - Init START
+ip-26-0-161-123:32711:33127 [7] NCCL INFO comm 0xa0d8f70 rank 31 nranks 128 cudaDev 7 nvmlDev 7 busId ca000 commId 0x759f85fd5295c628 - Init START
+ip-26-0-161-123:32709:33129 [5] NCCL INFO comm 0xa49ae70 rank 29 nranks 128 cudaDev 5 nvmlDev 5 busId a8000 commId 0x759f85fd5295c628 - Init START
+ip-26-0-161-123:32708:33126 [4] NCCL INFO comm 0xa4e8b60 rank 28 nranks 128 cudaDev 4 nvmlDev 4 busId 97000 commId 0x759f85fd5295c628 - Init START
+ip-26-0-161-123:32707:33124 [3] NCCL INFO comm 0x8ad9f30 rank 27 nranks 128 cudaDev 3 nvmlDev 3 busId 86000 commId 0x759f85fd5295c628 - Init START
+ip-26-0-160-225:124167:124586 [7] NCCL INFO comm 0x9e9e880 rank 7 nranks 128 cudaDev 7 nvmlDev 7 busId ca000 commId 0x759f85fd5295c628 - Init START
+ip-26-0-160-225:124166:124583 [6] NCCL INFO comm 0x99b4f20 rank 6 nranks 128 cudaDev 6 nvmlDev 6 busId b9000 commId 0x759f85fd5295c628 - Init START
+ip-26-0-160-225:124165:124584 [5] NCCL INFO comm 0x8a90b80 rank 5 nranks 128 cudaDev 5 nvmlDev 5 busId a8000 commId 0x759f85fd5295c628 - Init START
+ip-26-0-160-225:124161:124582 [1] NCCL INFO comm 0x8a53580 rank 1 nranks 128 cudaDev 1 nvmlDev 1 busId 64000 commId 0x759f85fd5295c628 - Init START
+ip-26-0-160-225:124164:124585 [4] NCCL INFO comm 0x91c6330 rank 4 nranks 128 cudaDev 4 nvmlDev 4 busId 97000 commId 0x759f85fd5295c628 - Init START
+ip-26-0-160-225:124163:124588 [3] NCCL INFO comm 0x9726fa0 rank 3 nranks 128 cudaDev 3 nvmlDev 3 busId 86000 commId 0x759f85fd5295c628 - Init START
+ip-26-0-161-142:62884:63299 [3] NCCL INFO comm 0x9811ff0 rank 43 nranks 128 cudaDev 3 nvmlDev 3 busId 86000 commId 0x759f85fd5295c628 - Init START
+ip-26-0-161-142:62883:63301 [2] NCCL INFO comm 0x8d92c70 rank 42 nranks 128 cudaDev 2 nvmlDev 2 busId 75000 commId 0x759f85fd5295c628 - Init START
+ip-26-0-161-142:62882:63300 [1] NCCL INFO comm 0x912d0f0 rank 41 nranks 128 cudaDev 1 nvmlDev 1 busId 64000 commId 0x759f85fd5295c628 - Init START
+ip-26-0-161-142:62881:63298 [0] NCCL INFO comm 0xa365ab0 rank 40 nranks 128 cudaDev 0 nvmlDev 0 busId 53000 commId 0x759f85fd5295c628 - Init START
+ip-26-0-160-225:124162:124587 [2] NCCL INFO comm 0xa5007c0 rank 2 nranks 128 cudaDev 2 nvmlDev 2 busId 75000 commId 0x759f85fd5295c628 - Init START
+ip-26-0-161-138:96568:97007 [5] NCCL INFO comm 0x8b312f0 rank 37 nranks 128 cudaDev 5 nvmlDev 5 busId a8000 commId 0x759f85fd5295c628 - Init START
+ip-26-0-160-242:1364905:1365328 [3] NCCL INFO comm 0x9a8b4e0 rank 11 nranks 128 cudaDev 3 nvmlDev 3 busId 86000 commId 0x759f85fd5295c628 - Init START
+ip-26-0-160-242:1364909:1365327 [7] NCCL INFO comm 0xa0bdc30 rank 15 nranks 128 cudaDev 7 nvmlDev 7 busId ca000 commId 0x759f85fd5295c628 - Init START
+ip-26-0-160-242:1364908:1365332 [6] NCCL INFO comm 0xa6080b0 rank 14 nranks 128 cudaDev 6 nvmlDev 6 busId b9000 commId 0x759f85fd5295c628 - Init START
+ip-26-0-161-123:32706:33122 [2] NCCL INFO comm 0x9c6ed80 rank 26 nranks 128 cudaDev 2 nvmlDev 2 busId 75000 commId 0x759f85fd5295c628 - Init START
+ip-26-0-161-138:96567:97008 [4] NCCL INFO comm 0x8aca3d0 rank 36 nranks 128 cudaDev 4 nvmlDev 4 busId 97000 commId 0x759f85fd5295c628 - Init START
+ip-26-0-161-138:96566:97010 [3] NCCL INFO comm 0x9576100 rank 35 nranks 128 cudaDev 3 nvmlDev 3 busId 86000 commId 0x759f85fd5295c628 - Init START
+ip-26-0-160-242:1364906:1365331 [4] NCCL INFO comm 0xa1ed2c0 rank 12 nranks 128 cudaDev 4 nvmlDev 4 busId 97000 commId 0x759f85fd5295c628 - Init START
+ip-26-0-160-242:1364907:1365329 [5] NCCL INFO comm 0x88ea1b0 rank 13 nranks 128 cudaDev 5 nvmlDev 5 busId a8000 commId 0x759f85fd5295c628 - Init START
+ip-26-0-161-123:32705:33128 [1] NCCL INFO comm 0x9c4b570 rank 25 nranks 128 cudaDev 1 nvmlDev 1 busId 64000 commId 0x759f85fd5295c628 - Init START
+ip-26-0-161-138:96565:97004 [2] NCCL INFO comm 0xa462c20 rank 34 nranks 128 cudaDev 2 nvmlDev 2 busId 75000 commId 0x759f85fd5295c628 - Init START
+ip-26-0-161-103:96893:97310 [5] NCCL INFO comm 0x8943870 rank 21 nranks 128 cudaDev 5 nvmlDev 5 busId a8000 commId 0x759f85fd5295c628 - Init START
+ip-26-0-161-103:96892:97308 [4] NCCL INFO comm 0x9df5b40 rank 20 nranks 128 cudaDev 4 nvmlDev 4 busId 97000 commId 0x759f85fd5295c628 - Init START
+ip-26-0-161-103:96895:97307 [7] NCCL INFO comm 0xa67bcf0 rank 23 nranks 128 cudaDev 7 nvmlDev 7 busId ca000 commId 0x759f85fd5295c628 - Init START
+ip-26-0-161-103:96894:97312 [6] NCCL INFO comm 0x8cd64e0 rank 22 nranks 128 cudaDev 6 nvmlDev 6 busId b9000 commId 0x759f85fd5295c628 - Init START
+ip-26-0-161-103:96891:97311 [3] NCCL INFO comm 0x8ec46b0 rank 19 nranks 128 cudaDev 3 nvmlDev 3 busId 86000 commId 0x759f85fd5295c628 - Init START
+ip-26-0-161-103:96890:97313 [2] NCCL INFO comm 0x9078a80 rank 18 nranks 128 cudaDev 2 nvmlDev 2 busId 75000 commId 0x759f85fd5295c628 - Init START
+ip-26-0-161-123:32704:33123 [0] NCCL INFO comm 0xa6bba20 rank 24 nranks 128 cudaDev 0 nvmlDev 0 busId 53000 commId 0x759f85fd5295c628 - Init START
+ip-26-0-161-138:96564:97009 [1] NCCL INFO comm 0x987dbe0 rank 33 nranks 128 cudaDev 1 nvmlDev 1 busId 64000 commId 0x759f85fd5295c628 - Init START
+ip-26-0-161-138:96563:97003 [0] NCCL INFO comm 0x9c090c0 rank 32 nranks 128 cudaDev 0 nvmlDev 0 busId 53000 commId 0x759f85fd5295c628 - Init START
+ip-26-0-161-103:96889:97309 [1] NCCL INFO comm 0x8a2fc20 rank 17 nranks 128 cudaDev 1 nvmlDev 1 busId 64000 commId 0x759f85fd5295c628 - Init START
+ip-26-0-161-103:96888:97306 [0] NCCL INFO comm 0x9097a60 rank 16 nranks 128 cudaDev 0 nvmlDev 0 busId 53000 commId 0x759f85fd5295c628 - Init START
+ip-26-0-168-52:2722052:2722464 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-52:2722050:2722471 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-30:235655:236069 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-34:272155:272574 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-34:272156:272570 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-34:272154:272569 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-52:2722051:2722465 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-52:2722049:2722467 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-30:235654:236070 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-30:235653:236067 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-52:2722048:2722469 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-30:235651:236068 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-30:235649:236071 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-30:235648:236064 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-30:235652:236066 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-34:272153:272573 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-34:272157:272572 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-52:2722047:2722468 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-30:235650:236065 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-34:272150:272576 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-34:272151:272571 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-52:2722046:2722466 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-52:2722045:2722470 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-34:272152:272575 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-202:153569:153981 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-178:2815970:2816397 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-164:1227470:1227885 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-202:153568:153982 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-202:153564:153984 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-202:153567:153979 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-78:114514:114932 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-78:114516:114934 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-78:114513:114933 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-78:114515:114936 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-78:114512:114930 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-131:1630212:1630652 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/409
+ip-26-0-165-202:153565:153985 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-202:153566:153980 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-202:153563:153986 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-164:1227464:1227884 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-59:1557272:1557687 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-59:1557271:1557690 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-164:1227469:1227882 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-164:1227466:1227887 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-78:114517:114935 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-78:114518:114937 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-78:114519:114931 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-164:1227467:1227886 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-131:1630209:1630651 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/409
+ip-26-0-165-131:1630210:1630653 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/409
+ip-26-0-165-164:1227468:1227883 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-131:1630208:1630650 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/409
+ip-26-0-165-131:1630211:1630649 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/409
+ip-26-0-165-131:1630207:1630648 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/409
+ip-26-0-165-202:153562:153983 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-164:1227463:1227880 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-164:1227465:1227881 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-131:1630206:1630647 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/409
+ip-26-0-165-131:1630205:1630646 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/409
+ip-26-0-165-59:1557270:1557688 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-59:1557273:1557685 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-59:1557269:1557684 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-59:1557268:1557686 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-59:1557266:1557689 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-59:1557267:1557683 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-160-225:124160:124581 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/541
+ip-26-0-160-225:124161:124582 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-160-225:124162:124587 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-160-225:124163:124588 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-160-225:124164:124585 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-160-225:124165:124584 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-160-225:124167:124586 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-160-242:1364902:1365333 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-160-242:1364903:1365334 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-160-225:124166:124583 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-160-242:1364904:1365335 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-160-242:1364905:1365328 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-160-242:1364906:1365331 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-160-242:1364909:1365327 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-160-242:1364907:1365329 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-160-242:1364908:1365332 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-103:96888:97306 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-103:96891:97311 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-103:96890:97313 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-103:96889:97309 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-103:96892:97308 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-103:96893:97310 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-123:32710:33125 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-123:32711:33127 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-123:32709:33129 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-123:32707:33124 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-123:32704:33123 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-138:96570:97005 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-138:96569:97006 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-138:96568:97007 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-142:62885:63304 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-142:62886:63303 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-142:62887:63305 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-142:62888:63302 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-142:62884:63299 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-103:96894:97312 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-103:96895:97307 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-178:2815969:2816398 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-178:2815968:2816394 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-123:32705:33128 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-123:32708:33126 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-153:2636084:2636511 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-153:2636085:2636509 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-138:96564:97009 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-123:32706:33122 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-138:96566:97010 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-138:96567:97008 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-138:96563:97003 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-138:96565:97004 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-142:62881:63298 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-142:62882:63300 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-142:62883:63301 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-178:2815967:2816399 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-178:2815966:2816396 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-153:2636086:2636507 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-178:2815964:2816401 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-178:2815965:2816395 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-153:2636090:2636505 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-153:2636089:2636508 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-153:2636087:2636506 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-153:2636091:2636512 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-178:2815963:2816400 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-153:2636088:2636510 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-160-242:1364906:1365331 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,fff00000,00000000
+ip-26-0-160-242:1364906:1365331 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-165-164:1227466:1227887 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-165-164:1227466:1227887 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-161-78:114518:114937 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-161-78:114518:114937 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-168-34:272150:272576 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-168-34:272150:272576 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-168-30:235654:236070 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-168-30:235654:236070 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-168-30:235655:236069 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-168-30:235655:236069 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-168-30:235650:236065 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-168-30:235650:236065 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-165-131:1630210:1630653 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-165-131:1630210:1630653 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-161-78:114515:114936 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-161-78:114515:114936 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-165-202:153567:153979 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-165-202:153567:153979 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-161-78:114519:114931 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-161-78:114519:114931 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-165-131:1630206:1630647 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-165-131:1630206:1630647 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-168-30:235651:236068 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-168-30:235651:236068 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-168-34:272152:272575 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-168-34:272152:272575 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-161-103:96894:97312 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-161-103:96894:97312 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-168-30:235649:236071 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-168-30:235649:236071 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-161-103:96888:97306 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-161-103:96888:97306 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-160-242:1364907:1365329 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,fff00000,00000000
+ip-26-0-160-242:1364907:1365329 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-161-103:96891:97311 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-161-103:96891:97311 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-165-202:153568:153982 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-165-202:153568:153982 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-168-30:235648:236064 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-168-30:235648:236064 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-165-131:1630208:1630650 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-165-131:1630208:1630650 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-168-30:235652:236066 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-168-30:235652:236066 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-161-103:96893:97310 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-161-103:96893:97310 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-168-34:272156:272570 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-168-34:272156:272570 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-161-103:96889:97309 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-161-103:96889:97309 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-165-164:1227463:1227880 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-165-164:1227463:1227880 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-168-30:235653:236067 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-168-30:235653:236067 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-160-242:1364903:1365334 [1] NCCL INFO Setting affinity for GPU 1 to 0fff,ffffffff
+ip-26-0-160-242:1364903:1365334 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-161-103:96890:97313 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-165-131:1630212:1630652 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-165-131:1630212:1630652 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-161-103:96890:97313 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-165-164:1227470:1227885 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-165-164:1227470:1227885 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-160-225:124161:124582 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-160-225:124161:124582 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-165-131:1630205:1630646 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-165-131:1630205:1630646 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-160-225:124164:124585 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-160-225:124164:124585 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-161-103:96892:97308 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-161-103:96892:97308 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-161-78:114514:114932 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-161-78:114514:114932 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-161-103:96895:97307 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-161-103:96895:97307 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-161-78:114517:114935 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-161-78:114517:114935 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-168-34:272154:272569 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-168-34:272154:272569 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-165-131:1630209:1630651 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-165-131:1630209:1630651 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-168-34:272157:272572 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-168-34:272157:272572 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-160-225:124166:124583 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-160-225:124166:124583 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-161-78:114512:114930 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-161-78:114512:114930 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-161-78:114513:114933 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-161-78:114513:114933 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-165-131:1630207:1630648 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-165-131:1630207:1630648 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-165-164:1227465:1227881 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-165-164:1227465:1227881 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-160-242:1364909:1365327 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,fff00000,00000000
+ip-26-0-165-131:1630211:1630649 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-165-131:1630211:1630649 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-160-242:1364909:1365327 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-160-225:124163:124588 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-160-225:124163:124588 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-160-242:1364905:1365328 [3] NCCL INFO Setting affinity for GPU 3 to 0fff,ffffffff
+ip-26-0-160-242:1364905:1365328 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-161-142:62881:63298 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-161-142:62881:63298 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-165-164:1227467:1227886 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-165-164:1227467:1227886 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-161-138:96568:97007 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,fff00000,00000000
+ip-26-0-161-138:96568:97007 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-161-142:62884:63299 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-161-142:62884:63299 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-165-202:153569:153981 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-165-202:153569:153981 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-161-78:114516:114934 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-161-78:114516:114934 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-160-225:124167:124586 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-160-225:124167:124586 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-165-202:153565:153985 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-165-202:153565:153985 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-165-164:1227469:1227882 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-165-164:1227469:1227882 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-160-225:124162:124587 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-160-225:124162:124587 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-161-142:62886:63303 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-161-142:62886:63303 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-168-34:272153:272573 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-161-142:62888:63302 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-161-142:62888:63302 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-168-34:272153:272573 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-168-34:272155:272574 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-168-34:272155:272574 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-168-34:272151:272571 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-168-34:272151:272571 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-165-164:1227464:1227884 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-165-164:1227464:1227884 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-160-242:1364904:1365335 [2] NCCL INFO Setting affinity for GPU 2 to 0fff,ffffffff
+ip-26-0-160-242:1364904:1365335 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-160-225:124165:124584 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-160-225:124165:124584 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-165-202:153564:153984 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-165-202:153564:153984 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-160-225:124160:124581 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-160-225:124160:124581 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-168-52:2722047:2722468 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-168-52:2722047:2722468 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-165-202:153563:153986 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-160-242:1364908:1365332 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,fff00000,00000000
+ip-26-0-160-242:1364908:1365332 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-165-202:153563:153986 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-165-202:153566:153980 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-165-202:153566:153980 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-165-164:1227468:1227883 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-165-164:1227468:1227883 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-165-202:153562:153983 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-165-202:153562:153983 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-161-138:96564:97009 [1] NCCL INFO Setting affinity for GPU 1 to 0fff,ffffffff
+ip-26-0-161-138:96564:97009 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-160-242:1364902:1365333 [0] NCCL INFO Setting affinity for GPU 0 to 0fff,ffffffff
+ip-26-0-160-242:1364902:1365333 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-161-142:62883:63301 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-161-142:62883:63301 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-161-138:96569:97006 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,fff00000,00000000
+ip-26-0-161-138:96569:97006 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-165-59:1557271:1557690 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-165-59:1557271:1557690 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-168-52:2722046:2722466 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-168-52:2722046:2722466 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-161-138:96570:97005 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,fff00000,00000000
+ip-26-0-161-138:96570:97005 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-161-142:62887:63305 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-161-142:62887:63305 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-161-142:62882:63300 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-161-142:62882:63300 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-161-142:62885:63304 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-161-142:62885:63304 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-168-52:2722048:2722469 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-168-52:2722048:2722469 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-161-138:96565:97004 [2] NCCL INFO Setting affinity for GPU 2 to 0fff,ffffffff
+ip-26-0-161-138:96565:97004 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-165-59:1557266:1557689 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-165-59:1557266:1557689 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-168-52:2722045:2722470 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-168-52:2722045:2722470 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-161-138:96566:97010 [3] NCCL INFO Setting affinity for GPU 3 to 0fff,ffffffff
+ip-26-0-161-138:96566:97010 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-161-138:96567:97008 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,fff00000,00000000
+ip-26-0-161-138:96567:97008 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-161-138:96563:97003 [0] NCCL INFO Setting affinity for GPU 0 to 0fff,ffffffff
+ip-26-0-161-138:96563:97003 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-168-52:2722052:2722464 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-168-52:2722052:2722464 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-165-59:1557272:1557687 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-165-59:1557272:1557687 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-165-59:1557267:1557683 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-165-59:1557267:1557683 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-168-52:2722050:2722471 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-168-52:2722050:2722471 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-168-52:2722049:2722467 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-168-52:2722049:2722467 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-168-52:2722051:2722465 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-168-52:2722051:2722465 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-165-59:1557269:1557684 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-165-59:1557269:1557684 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-165-59:1557270:1557688 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-165-59:1557270:1557688 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-161-123:32706:33122 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-161-123:32706:33122 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-165-59:1557268:1557686 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-165-59:1557268:1557686 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-165-59:1557273:1557685 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-165-59:1557273:1557685 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-161-178:2815969:2816398 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,fff00000,00000000
+ip-26-0-161-178:2815969:2816398 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-161-178:2815966:2816396 [3] NCCL INFO Setting affinity for GPU 3 to 0fff,ffffffff
+ip-26-0-161-178:2815966:2816396 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-161-178:2815968:2816394 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,fff00000,00000000
+ip-26-0-161-178:2815968:2816394 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-161-178:2815965:2816395 [2] NCCL INFO Setting affinity for GPU 2 to 0fff,ffffffff
+ip-26-0-161-178:2815965:2816395 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-161-178:2815964:2816401 [1] NCCL INFO Setting affinity for GPU 1 to 0fff,ffffffff
+ip-26-0-161-178:2815964:2816401 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-161-178:2815970:2816397 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,fff00000,00000000
+ip-26-0-161-178:2815970:2816397 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-161-153:2636085:2636509 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-161-153:2636085:2636509 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-161-178:2815967:2816399 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,fff00000,00000000
+ip-26-0-161-178:2815967:2816399 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-161-178:2815963:2816400 [0] NCCL INFO Setting affinity for GPU 0 to 0fff,ffffffff
+ip-26-0-161-178:2815963:2816400 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-161-123:32710:33125 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-161-123:32710:33125 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-161-123:32705:33128 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-161-123:32705:33128 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-161-123:32711:33127 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-161-123:32711:33127 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-161-153:2636087:2636506 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-161-153:2636087:2636506 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-161-123:32707:33124 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-161-123:32707:33124 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-161-123:32708:33126 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-161-123:32708:33126 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-161-123:32709:33129 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-161-123:32709:33129 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-161-153:2636088:2636510 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-161-153:2636088:2636510 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-161-123:32704:33123 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-161-123:32704:33123 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-161-153:2636090:2636505 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-161-153:2636090:2636505 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-161-153:2636091:2636512 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-161-153:2636091:2636512 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-161-153:2636086:2636507 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-161-153:2636086:2636507 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-161-153:2636089:2636508 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-161-153:2636089:2636508 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-161-153:2636084:2636511 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-161-153:2636084:2636511 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-161-153:2636089:2636508 [5] NCCL INFO Trees [0] 54/-1/-1->53->52 [1] 54/-1/-1->53->52 [2] 54/-1/-1->53->52 [3] 54/-1/-1->53->52 [4] 54/-1/-1->53->52 [5] 54/45/61->53->37 [6] -1/-1/-1->53->52 [7] 54/-1/-1->53->52 [8] 54/-1/-1->53->52 [9] 54/-1/-1->53->52 [10] 54/-1/-1->53->52 [11] 54/-1/-1->53->52 [12] 54/-1/-1->53->52 [13] 54/-1/-1->53->45 [14] -1/-1/-1->53->52 [15] 54/-1/-1->53->52
+ip-26-0-161-153:2636089:2636508 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-153:2636090:2636505 [6] NCCL INFO Trees [0] 55/-1/-1->54->53 [1] 55/-1/-1->54->53 [2] 55/-1/-1->54->53 [3] 55/-1/-1->54->53 [4] 55/-1/-1->54->53 [5] 55/-1/-1->54->53 [6] 55/46/62->54->38 [7] -1/-1/-1->54->53 [8] 55/-1/-1->54->53 [9] 55/-1/-1->54->53 [10] 55/-1/-1->54->53 [11] 55/-1/-1->54->53 [12] 55/-1/-1->54->53 [13] 55/-1/-1->54->53 [14] 55/-1/-1->54->46 [15] -1/-1/-1->54->53
+ip-26-0-161-153:2636090:2636505 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-153:2636091:2636512 [7] NCCL INFO Trees [0] -1/-1/-1->55->54 [1] 48/-1/-1->55->54 [2] 48/-1/-1->55->54 [3] 48/-1/-1->55->54 [4] 48/-1/-1->55->54 [5] 48/-1/-1->55->54 [6] 48/-1/-1->55->54 [7] 48/47/63->55->39 [8] -1/-1/-1->55->54 [9] 48/-1/-1->55->54 [10] 48/-1/-1->55->54 [11] 48/-1/-1->55->54 [12] 48/-1/-1->55->54 [13] 48/-1/-1->55->54 [14] 48/-1/-1->55->54 [15] 48/-1/-1->55->47
+ip-26-0-161-153:2636091:2636512 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-178:2815963:2816400 [0] NCCL INFO Trees [0] 57/-1/-1->56->48 [1] -1/-1/-1->56->63 [2] 57/-1/-1->56->63 [3] 57/-1/-1->56->63 [4] 57/-1/-1->56->63 [5] 57/-1/-1->56->63 [6] 57/-1/-1->56->63 [7] 57/-1/-1->56->63 [8] 57/88/24->56->120 [9] -1/-1/-1->56->63 [10] 57/-1/-1->56->63 [11] 57/-1/-1->56->63 [12] 57/-1/-1->56->63 [13] 57/-1/-1->56->63 [14] 57/-1/-1->56->63 [15] 57/-1/-1->56->63
+ip-26-0-161-178:2815963:2816400 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-178:2815965:2816395 [2] NCCL INFO Trees [0] 59/-1/-1->58->57 [1] 59/-1/-1->58->57 [2] 59/-1/-1->58->50 [3] -1/-1/-1->58->57 [4] 59/-1/-1->58->57 [5] 59/-1/-1->58->57 [6] 59/-1/-1->58->57 [7] 59/-1/-1->58->57 [8] 59/-1/-1->58->57 [9] 59/-1/-1->58->57 [10] 59/90/26->58->122 [11] -1/-1/-1->58->57 [12] 59/-1/-1->58->57 [13] 59/-1/-1->58->57 [14] 59/-1/-1->58->57 [15] 59/-1/-1->58->57
+ip-26-0-161-178:2815965:2816395 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-178:2815964:2816401 [1] NCCL INFO Trees [0] 58/-1/-1->57->56 [1] 58/-1/-1->57->49 [2] -1/-1/-1->57->56 [3] 58/-1/-1->57->56 [4] 58/-1/-1->57->56 [5] 58/-1/-1->57->56 [6] 58/-1/-1->57->56 [7] 58/-1/-1->57->56 [8] 58/-1/-1->57->56 [9] 58/89/25->57->121 [10] -1/-1/-1->57->56 [11] 58/-1/-1->57->56 [12] 58/-1/-1->57->56 [13] 58/-1/-1->57->56 [14] 58/-1/-1->57->56 [15] 58/-1/-1->57->56
+ip-26-0-161-178:2815964:2816401 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-178:2815970:2816397 [7] NCCL INFO Trees [0] -1/-1/-1->63->62 [1] 56/-1/-1->63->62 [2] 56/-1/-1->63->62 [3] 56/-1/-1->63->62 [4] 56/-1/-1->63->62 [5] 56/-1/-1->63->62 [6] 56/-1/-1->63->62 [7] 56/-1/-1->63->55 [8] -1/-1/-1->63->62 [9] 56/-1/-1->63->62 [10] 56/-1/-1->63->62 [11] 56/-1/-1->63->62 [12] 56/-1/-1->63->62 [13] 56/-1/-1->63->62 [14] 56/-1/-1->63->62 [15] 56/95/31->63->127
+ip-26-0-161-178:2815970:2816397 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-178:2815966:2816396 [3] NCCL INFO Trees [0] 60/-1/-1->59->58 [1] 60/-1/-1->59->58 [2] 60/-1/-1->59->58 [3] 60/-1/-1->59->51 [4] -1/-1/-1->59->58 [5] 60/-1/-1->59->58 [6] 60/-1/-1->59->58 [7] 60/-1/-1->59->58 [8] 60/-1/-1->59->58 [9] 60/-1/-1->59->58 [10] 60/-1/-1->59->58 [11] 60/91/27->59->123 [12] -1/-1/-1->59->58 [13] 60/-1/-1->59->58 [14] 60/-1/-1->59->58 [15] 60/-1/-1->59->58
+ip-26-0-161-178:2815967:2816399 [4] NCCL INFO Trees [0] 61/-1/-1->60->59 [1] 61/-1/-1->60->59 [2] 61/-1/-1->60->59 [3] 61/-1/-1->60->59 [4] 61/-1/-1->60->52 [5] -1/-1/-1->60->59 [6] 61/-1/-1->60->59 [7] 61/-1/-1->60->59 [8] 61/-1/-1->60->59 [9] 61/-1/-1->60->59 [10] 61/-1/-1->60->59 [11] 61/-1/-1->60->59 [12] 61/92/28->60->124 [13] -1/-1/-1->60->59 [14] 61/-1/-1->60->59 [15] 61/-1/-1->60->59
+ip-26-0-161-178:2815966:2816396 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-178:2815969:2816398 [6] NCCL INFO Trees [0] 63/-1/-1->62->61 [1] 63/-1/-1->62->61 [2] 63/-1/-1->62->61 [3] 63/-1/-1->62->61 [4] 63/-1/-1->62->61 [5] 63/-1/-1->62->61 [6] 63/-1/-1->62->54 [7] -1/-1/-1->62->61 [8] 63/-1/-1->62->61 [9] 63/-1/-1->62->61 [10] 63/-1/-1->62->61 [11] 63/-1/-1->62->61 [12] 63/-1/-1->62->61 [13] 63/-1/-1->62->61 [14] 63/94/30->62->126 [15] -1/-1/-1->62->61
+ip-26-0-161-178:2815967:2816399 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-178:2815969:2816398 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-178:2815968:2816394 [5] NCCL INFO Trees [0] 62/-1/-1->61->60 [1] 62/-1/-1->61->60 [2] 62/-1/-1->61->60 [3] 62/-1/-1->61->60 [4] 62/-1/-1->61->60 [5] 62/-1/-1->61->53 [6] -1/-1/-1->61->60 [7] 62/-1/-1->61->60 [8] 62/-1/-1->61->60 [9] 62/-1/-1->61->60 [10] 62/-1/-1->61->60 [11] 62/-1/-1->61->60 [12] 62/-1/-1->61->60 [13] 62/93/29->61->125 [14] -1/-1/-1->61->60 [15] 62/-1/-1->61->60
+ip-26-0-161-178:2815968:2816394 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-142:62888:63302 [7] NCCL INFO Trees [0] -1/-1/-1->47->46 [1] 40/-1/-1->47->46 [2] 40/-1/-1->47->46 [3] 40/-1/-1->47->46 [4] 40/-1/-1->47->46 [5] 40/-1/-1->47->46 [6] 40/-1/-1->47->46 [7] 40/-1/-1->47->55 [8] -1/-1/-1->47->46 [9] 40/-1/-1->47->46 [10] 40/-1/-1->47->46 [11] 40/-1/-1->47->46 [12] 40/-1/-1->47->46 [13] 40/-1/-1->47->46 [14] 40/-1/-1->47->46 [15] 40/55/39->47->31
+ip-26-0-161-142:62888:63302 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-225:124160:124581 [0] NCCL INFO NVLS Head  0:  0  8 16 24 32 40 48 56 64 72 80 88 96 104 112 120
+ip-26-0-160-225:124160:124581 [0] NCCL INFO NVLS Head  1:  1  9 17 25 33 41 49 57 65 73 81 89 97 105 113 121
+ip-26-0-161-153:2636086:2636507 [2] NCCL INFO Trees [0] 51/-1/-1->50->49 [1] 51/-1/-1->50->49 [2] 51/42/58->50->34 [3] -1/-1/-1->50->49 [4] 51/-1/-1->50->49 [5] 51/-1/-1->50->49 [6] 51/-1/-1->50->49 [7] 51/-1/-1->50->49 [8] 51/-1/-1->50->49 [9] 51/-1/-1->50->49 [10] 51/-1/-1->50->42 [11] -1/-1/-1->50->49 [12] 51/-1/-1->50->49 [13] 51/-1/-1->50->49 [14] 51/-1/-1->50->49 [15] 51/-1/-1->50->49
+ip-26-0-161-153:2636086:2636507 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-225:124160:124581 [0] NCCL INFO NVLS Head  2:  2 10 18 26 34 42 50 58 66 74 82 90 98 106 114 122
+ip-26-0-160-225:124160:124581 [0] NCCL INFO NVLS Head  3:  3 11 19 27 35 43 51 59 67 75 83 91 99 107 115 123
+ip-26-0-160-225:124160:124581 [0] NCCL INFO NVLS Head  4:  4 12 20 28 36 44 52 60 68 76 84 92 100 108 116 124
+ip-26-0-160-225:124160:124581 [0] NCCL INFO NVLS Head  5:  5 13 21 29 37 45 53 61 69 77 85 93 101 109 117 125
+ip-26-0-161-142:62886:63303 [5] NCCL INFO Trees [0] 46/-1/-1->45->44 [1] 46/-1/-1->45->44 [2] 46/-1/-1->45->44 [3] 46/-1/-1->45->44 [4] 46/-1/-1->45->44 [5] 46/-1/-1->45->53 [6] -1/-1/-1->45->44 [7] 46/-1/-1->45->44 [8] 46/-1/-1->45->44 [9] 46/-1/-1->45->44 [10] 46/-1/-1->45->44 [11] 46/-1/-1->45->44 [12] 46/-1/-1->45->44 [13] 46/53/37->45->29 [14] -1/-1/-1->45->44 [15] 46/-1/-1->45->44
+ip-26-0-161-142:62886:63303 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-59:1557272:1557687 [6] NCCL INFO Trees [0] 103/-1/-1->102->101 [1] 103/-1/-1->102->101 [2] 103/-1/-1->102->101 [3] 103/-1/-1->102->101 [4] 103/-1/-1->102->101 [5] 103/-1/-1->102->101 [6] 103/86/118->102->70 [7] -1/-1/-1->102->101 [8] 103/-1/-1->102->101 [9] 103/-1/-1->102->101 [10] 103/-1/-1->102->101 [11] 103/-1/-1->102->101 [12] 103/-1/-1->102->101 [13] 103/-1/-1->102->101 [14] 103/-1/-1->102->110 [15] -1/-1/-1->102->101
+ip-26-0-165-59:1557273:1557685 [7] NCCL INFO Trees [0] -1/-1/-1->103->102 [1] 96/-1/-1->103->102 [2] 96/-1/-1->103->102 [3] 96/-1/-1->103->102 [4] 96/-1/-1->103->102 [5] 96/-1/-1->103->102 [6] 96/-1/-1->103->102 [7] 96/87/119->103->71 [8] -1/-1/-1->103->102 [9] 96/-1/-1->103->102 [10] 96/-1/-1->103->102 [11] 96/-1/-1->103->102 [12] 96/-1/-1->103->102 [13] 96/-1/-1->103->102 [14] 96/-1/-1->103->102 [15] 96/-1/-1->103->111
+ip-26-0-165-59:1557272:1557687 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-59:1557273:1557685 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-168-30:235655:236069 [7] NCCL INFO Trees [0] -1/-1/-1->111->110 [1] 104/-1/-1->111->110 [2] 104/-1/-1->111->110 [3] 104/-1/-1->111->110 [4] 104/-1/-1->111->110 [5] 104/-1/-1->111->110 [6] 104/-1/-1->111->110 [7] 104/-1/-1->111->119 [8] -1/-1/-1->111->110 [9] 104/-1/-1->111->110 [10] 104/-1/-1->111->110 [11] 104/-1/-1->111->110 [12] 104/-1/-1->111->110 [13] 104/-1/-1->111->110 [14] 104/-1/-1->111->110 [15] 104/119/103->111->95
+ip-26-0-168-30:235655:236069 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-168-34:272150:272576 [0] NCCL INFO Trees [0] 113/104/120->112->96 [1] -1/-1/-1->112->119 [2] 113/-1/-1->112->119 [3] 113/-1/-1->112->119 [4] 113/-1/-1->112->119 [5] 113/-1/-1->112->119 [6] 113/-1/-1->112->119 [7] 113/-1/-1->112->119 [8] 113/-1/-1->112->104 [9] -1/-1/-1->112->119 [10] 113/-1/-1->112->119 [11] 113/-1/-1->112->119 [12] 113/-1/-1->112->119 [13] 113/-1/-1->112->119 [14] 113/-1/-1->112->119 [15] 113/-1/-1->112->119
+ip-26-0-168-34:272151:272571 [1] NCCL INFO Trees [0] 114/-1/-1->113->112 [1] 114/105/121->113->97 [2] -1/-1/-1->113->112 [3] 114/-1/-1->113->112 [4] 114/-1/-1->113->112 [5] 114/-1/-1->113->112 [6] 114/-1/-1->113->112 [7] 114/-1/-1->113->112 [8] 114/-1/-1->113->112 [9] 114/-1/-1->113->105 [10] -1/-1/-1->113->112 [11] 114/-1/-1->113->112 [12] 114/-1/-1->113->112 [13] 114/-1/-1->113->112 [14] 114/-1/-1->113->112 [15] 114/-1/-1->113->112
+ip-26-0-168-34:272150:272576 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-168-34:272151:272571 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-225:124160:124581 [0] NCCL INFO NVLS Head  6:  6 14 22 30 38 46 54 62 70 78 86 94 102 110 118 126
+ip-26-0-160-225:124160:124581 [0] NCCL INFO NVLS Head  7:  7 15 23 31 39 47 55 63 71 79 87 95 103 111 119 127
+ip-26-0-160-225:124160:124581 [0] NCCL INFO Channel 00/16 :    0   1   2   3   4   5   6   7   8   9  10  11  12  13  14  15  16  17  18  19
+ip-26-0-160-225:124160:124581 [0] NCCL INFO Channel 01/16 :    0   7   6   5   4   3   2   9   8  15  14  13  12  11  10  17  16  23  22  21
+ip-26-0-168-52:2722047:2722468 [2] NCCL INFO Trees [0] 123/-1/-1->122->121 [1] 123/-1/-1->122->121 [2] 123/-1/-1->122->114 [3] -1/-1/-1->122->121 [4] 123/-1/-1->122->121 [5] 123/-1/-1->122->121 [6] 123/-1/-1->122->121 [7] 123/-1/-1->122->121 [8] 123/-1/-1->122->121 [9] 123/-1/-1->122->121 [10] 123/58/-1->122->-1 [11] -1/-1/-1->122->121 [12] 123/-1/-1->122->121 [13] 123/-1/-1->122->121 [14] 123/-1/-1->122->121 [15] 123/-1/-1->122->121
+ip-26-0-168-52:2722047:2722468 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-138:96570:97005 [7] NCCL INFO Trees [0] -1/-1/-1->39->38 [1] 32/-1/-1->39->38 [2] 32/-1/-1->39->38 [3] 32/-1/-1->39->38 [4] 32/-1/-1->39->38 [5] 32/-1/-1->39->38 [6] 32/-1/-1->39->38 [7] 32/23/55->39->71 [8] -1/-1/-1->39->38 [9] 32/-1/-1->39->38 [10] 32/-1/-1->39->38 [11] 32/-1/-1->39->38 [12] 32/-1/-1->39->38 [13] 32/-1/-1->39->38 [14] 32/-1/-1->39->38 [15] 32/-1/-1->39->47
+ip-26-0-161-138:96570:97005 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-202:153569:153981 [7] NCCL INFO Trees [0] -1/-1/-1->95->94 [1] 88/-1/-1->95->94 [2] 88/-1/-1->95->94 [3] 88/-1/-1->95->94 [4] 88/-1/-1->95->94 [5] 88/-1/-1->95->94 [6] 88/-1/-1->95->94 [7] 88/-1/-1->95->87 [8] -1/-1/-1->95->94 [9] 88/-1/-1->95->94 [10] 88/-1/-1->95->94 [11] 88/-1/-1->95->94 [12] 88/-1/-1->95->94 [13] 88/-1/-1->95->94 [14] 88/-1/-1->95->94 [15] 88/111/79->95->63
+ip-26-0-165-202:153569:153981 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-142:62887:63305 [6] NCCL INFO Trees [0] 47/-1/-1->46->45 [1] 47/-1/-1->46->45 [2] 47/-1/-1->46->45 [3] 47/-1/-1->46->45 [4] 47/-1/-1->46->45 [5] 47/-1/-1->46->45 [6] 47/-1/-1->46->54 [7] -1/-1/-1->46->45 [8] 47/-1/-1->46->45 [9] 47/-1/-1->46->45 [10] 47/-1/-1->46->45 [11] 47/-1/-1->46->45 [12] 47/-1/-1->46->45 [13] 47/-1/-1->46->45 [14] 47/54/38->46->30 [15] -1/-1/-1->46->45
+ip-26-0-161-142:62887:63305 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-59:1557271:1557690 [5] NCCL INFO Trees [0] 102/-1/-1->101->100 [1] 102/-1/-1->101->100 [2] 102/-1/-1->101->100 [3] 102/-1/-1->101->100 [4] 102/-1/-1->101->100 [5] 102/85/117->101->69 [6] -1/-1/-1->101->100 [7] 102/-1/-1->101->100 [8] 102/-1/-1->101->100 [9] 102/-1/-1->101->100 [10] 102/-1/-1->101->100 [11] 102/-1/-1->101->100 [12] 102/-1/-1->101->100 [13] 102/-1/-1->101->109 [14] -1/-1/-1->101->100 [15] 102/-1/-1->101->100
+ip-26-0-165-59:1557271:1557690 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-168-30:235653:236067 [5] NCCL INFO Trees [0] 110/-1/-1->109->108 [1] 110/-1/-1->109->108 [2] 110/-1/-1->109->108 [3] 110/-1/-1->109->108 [4] 110/-1/-1->109->108 [5] 110/-1/-1->109->117 [6] -1/-1/-1->109->108 [7] 110/-1/-1->109->108 [8] 110/-1/-1->109->108 [9] 110/-1/-1->109->108 [10] 110/-1/-1->109->108 [11] 110/-1/-1->109->108 [12] 110/-1/-1->109->108 [13] 110/117/101->109->93 [14] -1/-1/-1->109->108 [15] 110/-1/-1->109->108
+ip-26-0-168-30:235653:236067 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-168-30:235652:236066 [4] NCCL INFO Trees [0] 109/-1/-1->108->107 [1] 109/-1/-1->108->107 [2] 109/-1/-1->108->107 [3] 109/-1/-1->108->107 [4] 109/-1/-1->108->116 [5] -1/-1/-1->108->107 [6] 109/-1/-1->108->107 [7] 109/-1/-1->108->107 [8] 109/-1/-1->108->107 [9] 109/-1/-1->108->107 [10] 109/-1/-1->108->107 [11] 109/-1/-1->108->107 [12] 109/116/100->108->92 [13] -1/-1/-1->108->107 [14] 109/-1/-1->108->107 [15] 109/-1/-1->108->107
+ip-26-0-168-30:235652:236066 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-168-34:272154:272569 [4] NCCL INFO Trees [0] 117/-1/-1->116->115 [1] 117/-1/-1->116->115 [2] 117/-1/-1->116->115 [3] 117/-1/-1->116->115 [4] 117/108/124->116->100 [5] -1/-1/-1->116->115 [6] 117/-1/-1->116->115 [7] 117/-1/-1->116->115 [8] 117/-1/-1->116->115 [9] 117/-1/-1->116->115 [10] 117/-1/-1->116->115 [11] 117/-1/-1->116->115 [12] 117/-1/-1->116->108 [13] -1/-1/-1->116->115 [14] 117/-1/-1->116->115 [15] 117/-1/-1->116->115
+ip-26-0-168-34:272154:272569 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-168-34:272153:272573 [3] NCCL INFO Trees [0] 116/-1/-1->115->114 [1] 116/-1/-1->115->114 [2] 116/-1/-1->115->114 [3] 116/107/123->115->99 [4] -1/-1/-1->115->114 [5] 116/-1/-1->115->114 [6] 116/-1/-1->115->114 [7] 116/-1/-1->115->114 [8] 116/-1/-1->115->114 [9] 116/-1/-1->115->114 [10] 116/-1/-1->115->114 [11] 116/-1/-1->115->107 [12] -1/-1/-1->115->114 [13] 116/-1/-1->115->114 [14] 116/-1/-1->115->114 [15] 116/-1/-1->115->114
+ip-26-0-161-123:32710:33125 [6] NCCL INFO Trees [0] 31/-1/-1->30->29 [1] 31/-1/-1->30->29 [2] 31/-1/-1->30->29 [3] 31/-1/-1->30->29 [4] 31/-1/-1->30->29 [5] 31/-1/-1->30->29 [6] 31/-1/-1->30->22 [7] -1/-1/-1->30->29 [8] 31/-1/-1->30->29 [9] 31/-1/-1->30->29 [10] 31/-1/-1->30->29 [11] 31/-1/-1->30->29 [12] 31/-1/-1->30->29 [13] 31/-1/-1->30->29 [14] 31/46/14->30->62 [15] -1/-1/-1->30->29
+ip-26-0-161-123:32710:33125 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-164:1227469:1227882 [6] NCCL INFO Trees [0] 87/-1/-1->86->85 [1] 87/-1/-1->86->85 [2] 87/-1/-1->86->85 [3] 87/-1/-1->86->85 [4] 87/-1/-1->86->85 [5] 87/-1/-1->86->85 [6] 87/78/94->86->102 [7] -1/-1/-1->86->85 [8] 87/-1/-1->86->85 [9] 87/-1/-1->86->85 [10] 87/-1/-1->86->85 [11] 87/-1/-1->86->85 [12] 87/-1/-1->86->85 [13] 87/-1/-1->86->85 [14] 87/-1/-1->86->78 [15] -1/-1/-1->86->85
+ip-26-0-165-164:1227469:1227882 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-153:2636087:2636506 [3] NCCL INFO Trees [0] 52/-1/-1->51->50 [1] 52/-1/-1->51->50 [2] 52/-1/-1->51->50 [3] 52/43/59->51->35 [4] -1/-1/-1->51->50 [5] 52/-1/-1->51->50 [6] 52/-1/-1->51->50 [7] 52/-1/-1->51->50 [8] 52/-1/-1->51->50 [9] 52/-1/-1->51->50 [10] 52/-1/-1->51->50 [11] 52/-1/-1->51->43 [12] -1/-1/-1->51->50 [13] 52/-1/-1->51->50 [14] 52/-1/-1->51->50 [15] 52/-1/-1->51->50
+ip-26-0-161-153:2636087:2636506 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-225:124160:124581 [0] NCCL INFO Channel 02/16 :    0   7   6   5   4   3  10   9   8  15  14  13  12  11  18  17  16  23  22  21
+ip-26-0-160-225:124160:124581 [0] NCCL INFO Channel 03/16 :    0   7   6   5   4  11  10   9   8  15  14  13  12  19  18  17  16  23  22  21
+ip-26-0-168-52:2722049:2722467 [4] NCCL INFO Trees [0] 125/-1/-1->124->123 [1] 125/-1/-1->124->123 [2] 125/-1/-1->124->123 [3] 125/-1/-1->124->123 [4] 125/-1/-1->124->116 [5] -1/-1/-1->124->123 [6] 125/-1/-1->124->123 [7] 125/-1/-1->124->123 [8] 125/-1/-1->124->123 [9] 125/-1/-1->124->123 [10] 125/-1/-1->124->123 [11] 125/-1/-1->124->123 [12] 125/60/-1->124->-1 [13] -1/-1/-1->124->123 [14] 125/-1/-1->124->123 [15] 125/-1/-1->124->123
+ip-26-0-168-52:2722051:2722465 [6] NCCL INFO Trees [0] 127/-1/-1->126->125 [1] 127/-1/-1->126->125 [2] 127/-1/-1->126->125 [3] 127/-1/-1->126->125 [4] 127/-1/-1->126->125 [5] 127/-1/-1->126->125 [6] 127/-1/-1->126->118 [7] -1/-1/-1->126->125 [8] 127/-1/-1->126->125 [9] 127/-1/-1->126->125 [10] 127/-1/-1->126->125 [11] 127/-1/-1->126->125 [12] 127/-1/-1->126->125 [13] 127/-1/-1->126->125 [14] 127/62/-1->126->-1 [15] -1/-1/-1->126->125
+ip-26-0-161-78:114516:114934 [4] NCCL INFO Trees [0] 69/-1/-1->68->67 [1] 69/-1/-1->68->67 [2] 69/-1/-1->68->67 [3] 69/-1/-1->68->67 [4] 69/36/100->68->4 [5] -1/-1/-1->68->67 [6] 69/-1/-1->68->67 [7] 69/-1/-1->68->67 [8] 69/-1/-1->68->67 [9] 69/-1/-1->68->67 [10] 69/-1/-1->68->67 [11] 69/-1/-1->68->67 [12] 69/-1/-1->68->76 [13] -1/-1/-1->68->67 [14] 69/-1/-1->68->67 [15] 69/-1/-1->68->67
+ip-26-0-161-78:114516:114934 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-142:62884:63299 [3] NCCL INFO Trees [0] 44/-1/-1->43->42 [1] 44/-1/-1->43->42 [2] 44/-1/-1->43->42 [3] 44/-1/-1->43->51 [4] -1/-1/-1->43->42 [5] 44/-1/-1->43->42 [6] 44/-1/-1->43->42 [7] 44/-1/-1->43->42 [8] 44/-1/-1->43->42 [9] 44/-1/-1->43->42 [10] 44/-1/-1->43->42 [11] 44/51/35->43->27 [12] -1/-1/-1->43->42 [13] 44/-1/-1->43->42 [14] 44/-1/-1->43->42 [15] 44/-1/-1->43->42
+ip-26-0-165-59:1557270:1557688 [4] NCCL INFO Trees [0] 101/-1/-1->100->99 [1] 101/-1/-1->100->99 [2] 101/-1/-1->100->99 [3] 101/-1/-1->100->99 [4] 101/84/116->100->68 [5] -1/-1/-1->100->99 [6] 101/-1/-1->100->99 [7] 101/-1/-1->100->99 [8] 101/-1/-1->100->99 [9] 101/-1/-1->100->99 [10] 101/-1/-1->100->99 [11] 101/-1/-1->100->99 [12] 101/-1/-1->100->108 [13] -1/-1/-1->100->99 [14] 101/-1/-1->100->99 [15] 101/-1/-1->100->99
+ip-26-0-165-59:1557270:1557688 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-168-30:235650:236065 [2] NCCL INFO Trees [0] 107/-1/-1->106->105 [1] 107/-1/-1->106->105 [2] 107/-1/-1->106->114 [3] -1/-1/-1->106->105 [4] 107/-1/-1->106->105 [5] 107/-1/-1->106->105 [6] 107/-1/-1->106->105 [7] 107/-1/-1->106->105 [8] 107/-1/-1->106->105 [9] 107/-1/-1->106->105 [10] 107/114/98->106->90 [11] -1/-1/-1->106->105 [12] 107/-1/-1->106->105 [13] 107/-1/-1->106->105 [14] 107/-1/-1->106->105 [15] 107/-1/-1->106->105
+ip-26-0-168-30:235650:236065 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-168-30:235648:236064 [0] NCCL INFO Trees [0] 105/-1/-1->104->112 [1] -1/-1/-1->104->111 [2] 105/-1/-1->104->111 [3] 105/-1/-1->104->111 [4] 105/-1/-1->104->111 [5] 105/-1/-1->104->111 [6] 105/-1/-1->104->111 [7] 105/-1/-1->104->111 [8] 105/112/96->104->88 [9] -1/-1/-1->104->111 [10] 105/-1/-1->104->111 [11] 105/-1/-1->104->111 [12] 105/-1/-1->104->111 [13] 105/-1/-1->104->111 [14] 105/-1/-1->104->111 [15] 105/-1/-1->104->111
+ip-26-0-168-30:235648:236064 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-168-34:272156:272570 [6] NCCL INFO Trees [0] 119/-1/-1->118->117 [1] 119/-1/-1->118->117 [2] 119/-1/-1->118->117 [3] 119/-1/-1->118->117 [4] 119/-1/-1->118->117 [5] 119/-1/-1->118->117 [6] 119/110/126->118->102 [7] -1/-1/-1->118->117 [8] 119/-1/-1->118->117 [9] 119/-1/-1->118->117 [10] 119/-1/-1->118->117 [11] 119/-1/-1->118->117 [12] 119/-1/-1->118->117 [13] 119/-1/-1->118->117 [14] 119/-1/-1->118->110 [15] -1/-1/-1->118->117
+ip-26-0-168-34:272153:272573 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-168-34:272156:272570 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-168-34:272152:272575 [2] NCCL INFO Trees [0] 115/-1/-1->114->113 [1] 115/-1/-1->114->113 [2] 115/106/122->114->98 [3] -1/-1/-1->114->113 [4] 115/-1/-1->114->113 [5] 115/-1/-1->114->113 [6] 115/-1/-1->114->113 [7] 115/-1/-1->114->113 [8] 115/-1/-1->114->113 [9] 115/-1/-1->114->113 [10] 115/-1/-1->114->106 [11] -1/-1/-1->114->113 [12] 115/-1/-1->114->113 [13] 115/-1/-1->114->113 [14] 115/-1/-1->114->113 [15] 115/-1/-1->114->113
+ip-26-0-161-103:96893:97310 [5] NCCL INFO Trees [0] 22/-1/-1->21->20 [1] 22/-1/-1->21->20 [2] 22/-1/-1->21->20 [3] 22/-1/-1->21->20 [4] 22/-1/-1->21->20 [5] 22/13/29->21->37 [6] -1/-1/-1->21->20 [7] 22/-1/-1->21->20 [8] 22/-1/-1->21->20 [9] 22/-1/-1->21->20 [10] 22/-1/-1->21->20 [11] 22/-1/-1->21->20 [12] 22/-1/-1->21->20 [13] 22/-1/-1->21->13 [14] -1/-1/-1->21->20 [15] 22/-1/-1->21->20
+ip-26-0-161-103:96893:97310 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-103:96895:97307 [7] NCCL INFO Trees [0] -1/-1/-1->23->22 [1] 16/-1/-1->23->22 [2] 16/-1/-1->23->22 [3] 16/-1/-1->23->22 [4] 16/-1/-1->23->22 [5] 16/-1/-1->23->22 [6] 16/-1/-1->23->22 [7] 16/15/31->23->39 [8] -1/-1/-1->23->22 [9] 16/-1/-1->23->22 [10] 16/-1/-1->23->22 [11] 16/-1/-1->23->22 [12] 16/-1/-1->23->22 [13] 16/-1/-1->23->22 [14] 16/-1/-1->23->22 [15] 16/-1/-1->23->15
+ip-26-0-160-242:1364909:1365327 [7] NCCL INFO Trees [0] -1/-1/-1->15->14 [1] 8/-1/-1->15->14 [2] 8/-1/-1->15->14 [3] 8/-1/-1->15->14 [4] 8/-1/-1->15->14 [5] 8/-1/-1->15->14 [6] 8/-1/-1->15->14 [7] 8/-1/-1->15->23 [8] -1/-1/-1->15->14 [9] 8/-1/-1->15->14 [10] 8/-1/-1->15->14 [11] 8/-1/-1->15->14 [12] 8/-1/-1->15->14 [13] 8/-1/-1->15->14 [14] 8/-1/-1->15->14 [15] 8/23/7->15->31
+ip-26-0-160-242:1364909:1365327 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-242:1364908:1365332 [6] NCCL INFO Trees [0] 15/-1/-1->14->13 [1] 15/-1/-1->14->13 [2] 15/-1/-1->14->13 [3] 15/-1/-1->14->13 [4] 15/-1/-1->14->13 [5] 15/-1/-1->14->13 [6] 15/-1/-1->14->22 [7] -1/-1/-1->14->13 [8] 15/-1/-1->14->13 [9] 15/-1/-1->14->13 [10] 15/-1/-1->14->13 [11] 15/-1/-1->14->13 [12] 15/-1/-1->14->13 [13] 15/-1/-1->14->13 [14] 15/22/6->14->30 [15] -1/-1/-1->14->13
+ip-26-0-160-242:1364908:1365332 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-123:32708:33126 [4] NCCL INFO Trees [0] 29/-1/-1->28->27 [1] 29/-1/-1->28->27 [2] 29/-1/-1->28->27 [3] 29/-1/-1->28->27 [4] 29/-1/-1->28->20 [5] -1/-1/-1->28->27 [6] 29/-1/-1->28->27 [7] 29/-1/-1->28->27 [8] 29/-1/-1->28->27 [9] 29/-1/-1->28->27 [10] 29/-1/-1->28->27 [11] 29/-1/-1->28->27 [12] 29/44/12->28->60 [13] -1/-1/-1->28->27 [14] 29/-1/-1->28->27 [15] 29/-1/-1->28->27
+ip-26-0-161-123:32708:33126 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-123:32709:33129 [5] NCCL INFO Trees [0] 30/-1/-1->29->28 [1] 30/-1/-1->29->28 [2] 30/-1/-1->29->28 [3] 30/-1/-1->29->28 [4] 30/-1/-1->29->28 [5] 30/-1/-1->29->21 [6] -1/-1/-1->29->28 [7] 30/-1/-1->29->28 [8] 30/-1/-1->29->28 [9] 30/-1/-1->29->28 [10] 30/-1/-1->29->28 [11] 30/-1/-1->29->28 [12] 30/-1/-1->29->28 [13] 30/45/13->29->61 [14] -1/-1/-1->29->28 [15] 30/-1/-1->29->28
+ip-26-0-161-123:32709:33129 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-131:1630208:1630650 [3] NCCL INFO Trees [0] 76/-1/-1->75->74 [1] 76/-1/-1->75->74 [2] 76/-1/-1->75->74 [3] 76/-1/-1->75->83 [4] -1/-1/-1->75->74 [5] 76/-1/-1->75->74 [6] 76/-1/-1->75->74 [7] 76/-1/-1->75->74 [8] 76/-1/-1->75->74 [9] 76/-1/-1->75->74 [10] 76/-1/-1->75->74 [11] 76/83/67->75->91 [12] -1/-1/-1->75->74 [13] 76/-1/-1->75->74 [14] 76/-1/-1->75->74 [15] 76/-1/-1->75->74
+ip-26-0-165-131:1630206:1630647 [1] NCCL INFO Trees [0] 74/-1/-1->73->72 [1] 74/-1/-1->73->81 [2] -1/-1/-1->73->72 [3] 74/-1/-1->73->72 [4] 74/-1/-1->73->72 [5] 74/-1/-1->73->72 [6] 74/-1/-1->73->72 [7] 74/-1/-1->73->72 [8] 74/-1/-1->73->72 [9] 74/81/65->73->89 [10] -1/-1/-1->73->72 [11] 74/-1/-1->73->72 [12] 74/-1/-1->73->72 [13] 74/-1/-1->73->72 [14] 74/-1/-1->73->72 [15] 74/-1/-1->73->72
+ip-26-0-165-131:1630208:1630650 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-131:1630206:1630647 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-153:2636088:2636510 [4] NCCL INFO Trees [0] 53/-1/-1->52->51 [1] 53/-1/-1->52->51 [2] 53/-1/-1->52->51 [3] 53/-1/-1->52->51 [4] 53/44/60->52->36 [5] -1/-1/-1->52->51 [6] 53/-1/-1->52->51 [7] 53/-1/-1->52->51 [8] 53/-1/-1->52->51 [9] 53/-1/-1->52->51 [10] 53/-1/-1->52->51 [11] 53/-1/-1->52->51 [12] 53/-1/-1->52->44 [13] -1/-1/-1->52->51 [14] 53/-1/-1->52->51 [15] 53/-1/-1->52->51
+ip-26-0-161-153:2636084:2636511 [0] NCCL INFO Trees [0] 49/40/56->48->32 [1] -1/-1/-1->48->55 [2] 49/-1/-1->48->55 [3] 49/-1/-1->48->55 [4] 49/-1/-1->48->55 [5] 49/-1/-1->48->55 [6] 49/-1/-1->48->55 [7] 49/-1/-1->48->55 [8] 49/-1/-1->48->40 [9] -1/-1/-1->48->55 [10] 49/-1/-1->48->55 [11] 49/-1/-1->48->55 [12] 49/-1/-1->48->55 [13] 49/-1/-1->48->55 [14] 49/-1/-1->48->55 [15] 49/-1/-1->48->55
+ip-26-0-161-153:2636088:2636510 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-153:2636084:2636511 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-225:124160:124581 [0] NCCL INFO Channel 04/16 :    0   7   6   5  12  11  10   9   8  15  14  13  20  19  18  17  16  23  22  21
+ip-26-0-168-52:2722052:2722464 [7] NCCL INFO Trees [0] -1/-1/-1->127->126 [1] 120/-1/-1->127->126 [2] 120/-1/-1->127->126 [3] 120/-1/-1->127->126 [4] 120/-1/-1->127->126 [5] 120/-1/-1->127->126 [6] 120/-1/-1->127->126 [7] 120/-1/-1->127->119 [8] -1/-1/-1->127->126 [9] 120/-1/-1->127->126 [10] 120/-1/-1->127->126 [11] 120/-1/-1->127->126 [12] 120/-1/-1->127->126 [13] 120/-1/-1->127->126 [14] 120/-1/-1->127->126 [15] 120/63/-1->127->-1
+ip-26-0-168-52:2722049:2722467 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-168-52:2722048:2722469 [3] NCCL INFO Trees [0] 124/-1/-1->123->122 [1] 124/-1/-1->123->122 [2] 124/-1/-1->123->122 [3] 124/-1/-1->123->115 [4] -1/-1/-1->123->122 [5] 124/-1/-1->123->122 [6] 124/-1/-1->123->122 [7] 124/-1/-1->123->122 [8] 124/-1/-1->123->122 [9] 124/-1/-1->123->122 [10] 124/-1/-1->123->122 [11] 124/59/-1->123->-1 [12] -1/-1/-1->123->122 [13] 124/-1/-1->123->122 [14] 124/-1/-1->123->122 [15] 124/-1/-1->123->122
+ip-26-0-168-52:2722052:2722464 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-138:96568:97007 [5] NCCL INFO Trees [0] 38/-1/-1->37->36 [1] 38/-1/-1->37->36 [2] 38/-1/-1->37->36 [3] 38/-1/-1->37->36 [4] 38/-1/-1->37->36 [5] 38/21/53->37->69 [6] -1/-1/-1->37->36 [7] 38/-1/-1->37->36 [8] 38/-1/-1->37->36 [9] 38/-1/-1->37->36 [10] 38/-1/-1->37->36 [11] 38/-1/-1->37->36 [12] 38/-1/-1->37->36 [13] 38/-1/-1->37->45 [14] -1/-1/-1->37->36 [15] 38/-1/-1->37->36
+ip-26-0-161-138:96569:97006 [6] NCCL INFO Trees [0] 39/-1/-1->38->37 [1] 39/-1/-1->38->37 [2] 39/-1/-1->38->37 [3] 39/-1/-1->38->37 [4] 39/-1/-1->38->37 [5] 39/-1/-1->38->37 [6] 39/22/54->38->70 [7] -1/-1/-1->38->37 [8] 39/-1/-1->38->37 [9] 39/-1/-1->38->37 [10] 39/-1/-1->38->37 [11] 39/-1/-1->38->37 [12] 39/-1/-1->38->37 [13] 39/-1/-1->38->37 [14] 39/-1/-1->38->46 [15] -1/-1/-1->38->37
+ip-26-0-161-138:96568:97007 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-138:96569:97006 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-202:153568:153982 [6] NCCL INFO Trees [0] 95/-1/-1->94->93 [1] 95/-1/-1->94->93 [2] 95/-1/-1->94->93 [3] 95/-1/-1->94->93 [4] 95/-1/-1->94->93 [5] 95/-1/-1->94->93 [6] 95/-1/-1->94->86 [7] -1/-1/-1->94->93 [8] 95/-1/-1->94->93 [9] 95/-1/-1->94->93 [10] 95/-1/-1->94->93 [11] 95/-1/-1->94->93 [12] 95/-1/-1->94->93 [13] 95/-1/-1->94->93 [14] 95/110/78->94->62 [15] -1/-1/-1->94->93
+ip-26-0-165-202:153568:153982 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-78:114519:114931 [7] NCCL INFO Trees [0] -1/-1/-1->71->70 [1] 64/-1/-1->71->70 [2] 64/-1/-1->71->70 [3] 64/-1/-1->71->70 [4] 64/-1/-1->71->70 [5] 64/-1/-1->71->70 [6] 64/-1/-1->71->70 [7] 64/39/103->71->7 [8] -1/-1/-1->71->70 [9] 64/-1/-1->71->70 [10] 64/-1/-1->71->70 [11] 64/-1/-1->71->70 [12] 64/-1/-1->71->70 [13] 64/-1/-1->71->70 [14] 64/-1/-1->71->70 [15] 64/-1/-1->71->79
+ip-26-0-161-78:114515:114936 [3] NCCL INFO Trees [0] 68/-1/-1->67->66 [1] 68/-1/-1->67->66 [2] 68/-1/-1->67->66 [3] 68/35/99->67->3 [4] -1/-1/-1->67->66 [5] 68/-1/-1->67->66 [6] 68/-1/-1->67->66 [7] 68/-1/-1->67->66 [8] 68/-1/-1->67->66 [9] 68/-1/-1->67->66 [10] 68/-1/-1->67->66 [11] 68/-1/-1->67->75 [12] -1/-1/-1->67->66 [13] 68/-1/-1->67->66 [14] 68/-1/-1->67->66 [15] 68/-1/-1->67->66
+ip-26-0-161-142:62884:63299 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-59:1557268:1557686 [2] NCCL INFO Trees [0] 99/-1/-1->98->97 [1] 99/-1/-1->98->97 [2] 99/82/114->98->66 [3] -1/-1/-1->98->97 [4] 99/-1/-1->98->97 [5] 99/-1/-1->98->97 [6] 99/-1/-1->98->97 [7] 99/-1/-1->98->97 [8] 99/-1/-1->98->97 [9] 99/-1/-1->98->97 [10] 99/-1/-1->98->106 [11] -1/-1/-1->98->97 [12] 99/-1/-1->98->97 [13] 99/-1/-1->98->97 [14] 99/-1/-1->98->97 [15] 99/-1/-1->98->97
+ip-26-0-165-59:1557266:1557689 [0] NCCL INFO Trees [0] 97/80/112->96->64 [1] -1/-1/-1->96->103 [2] 97/-1/-1->96->103 [3] 97/-1/-1->96->103 [4] 97/-1/-1->96->103 [5] 97/-1/-1->96->103 [6] 97/-1/-1->96->103 [7] 97/-1/-1->96->103 [8] 97/-1/-1->96->104 [9] -1/-1/-1->96->103 [10] 97/-1/-1->96->103 [11] 97/-1/-1->96->103 [12] 97/-1/-1->96->103 [13] 97/-1/-1->96->103 [14] 97/-1/-1->96->103 [15] 97/-1/-1->96->103
+ip-26-0-165-59:1557268:1557686 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-59:1557266:1557689 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-168-30:235651:236068 [3] NCCL INFO Trees [0] 108/-1/-1->107->106 [1] 108/-1/-1->107->106 [2] 108/-1/-1->107->106 [3] 108/-1/-1->107->115 [4] -1/-1/-1->107->106 [5] 108/-1/-1->107->106 [6] 108/-1/-1->107->106 [7] 108/-1/-1->107->106 [8] 108/-1/-1->107->106 [9] 108/-1/-1->107->106 [10] 108/-1/-1->107->106 [11] 108/115/99->107->91 [12] -1/-1/-1->107->106 [13] 108/-1/-1->107->106 [14] 108/-1/-1->107->106 [15] 108/-1/-1->107->106
+ip-26-0-168-30:235651:236068 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-168-34:272152:272575 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-103:96894:97312 [6] NCCL INFO Trees [0] 23/-1/-1->22->21 [1] 23/-1/-1->22->21 [2] 23/-1/-1->22->21 [3] 23/-1/-1->22->21 [4] 23/-1/-1->22->21 [5] 23/-1/-1->22->21 [6] 23/14/30->22->38 [7] -1/-1/-1->22->21 [8] 23/-1/-1->22->21 [9] 23/-1/-1->22->21 [10] 23/-1/-1->22->21 [11] 23/-1/-1->22->21 [12] 23/-1/-1->22->21 [13] 23/-1/-1->22->21 [14] 23/-1/-1->22->14 [15] -1/-1/-1->22->21
+ip-26-0-161-103:96895:97307 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-242:1364905:1365328 [3] NCCL INFO Trees [0] 12/-1/-1->11->10 [1] 12/-1/-1->11->10 [2] 12/-1/-1->11->10 [3] 12/-1/-1->11->19 [4] -1/-1/-1->11->10 [5] 12/-1/-1->11->10 [6] 12/-1/-1->11->10 [7] 12/-1/-1->11->10 [8] 12/-1/-1->11->10 [9] 12/-1/-1->11->10 [10] 12/-1/-1->11->10 [11] 12/19/3->11->27 [12] -1/-1/-1->11->10 [13] 12/-1/-1->11->10 [14] 12/-1/-1->11->10 [15] 12/-1/-1->11->10
+ip-26-0-160-242:1364905:1365328 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-123:32711:33127 [7] NCCL INFO Trees [0] -1/-1/-1->31->30 [1] 24/-1/-1->31->30 [2] 24/-1/-1->31->30 [3] 24/-1/-1->31->30 [4] 24/-1/-1->31->30 [5] 24/-1/-1->31->30 [6] 24/-1/-1->31->30 [7] 24/-1/-1->31->23 [8] -1/-1/-1->31->30 [9] 24/-1/-1->31->30 [10] 24/-1/-1->31->30 [11] 24/-1/-1->31->30 [12] 24/-1/-1->31->30 [13] 24/-1/-1->31->30 [14] 24/-1/-1->31->30 [15] 24/47/15->31->63
+ip-26-0-161-123:32707:33124 [3] NCCL INFO Trees [0] 28/-1/-1->27->26 [1] 28/-1/-1->27->26 [2] 28/-1/-1->27->26 [3] 28/-1/-1->27->19 [4] -1/-1/-1->27->26 [5] 28/-1/-1->27->26 [6] 28/-1/-1->27->26 [7] 28/-1/-1->27->26 [8] 28/-1/-1->27->26 [9] 28/-1/-1->27->26 [10] 28/-1/-1->27->26 [11] 28/43/11->27->59 [12] -1/-1/-1->27->26 [13] 28/-1/-1->27->26 [14] 28/-1/-1->27->26 [15] 28/-1/-1->27->26
+ip-26-0-161-123:32711:33127 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-123:32707:33124 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-164:1227468:1227883 [5] NCCL INFO Trees [0] 86/-1/-1->85->84 [1] 86/-1/-1->85->84 [2] 86/-1/-1->85->84 [3] 86/-1/-1->85->84 [4] 86/-1/-1->85->84 [5] 86/77/93->85->101 [6] -1/-1/-1->85->84 [7] 86/-1/-1->85->84 [8] 86/-1/-1->85->84 [9] 86/-1/-1->85->84 [10] 86/-1/-1->85->84 [11] 86/-1/-1->85->84 [12] 86/-1/-1->85->84 [13] 86/-1/-1->85->77 [14] -1/-1/-1->85->84 [15] 86/-1/-1->85->84
+ip-26-0-165-164:1227468:1227883 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-164:1227466:1227887 [3] NCCL INFO Trees [0] 84/-1/-1->83->82 [1] 84/-1/-1->83->82 [2] 84/-1/-1->83->82 [3] 84/75/91->83->99 [4] -1/-1/-1->83->82 [5] 84/-1/-1->83->82 [6] 84/-1/-1->83->82 [7] 84/-1/-1->83->82 [8] 84/-1/-1->83->82 [9] 84/-1/-1->83->82 [10] 84/-1/-1->83->82 [11] 84/-1/-1->83->75 [12] -1/-1/-1->83->82 [13] 84/-1/-1->83->82 [14] 84/-1/-1->83->82 [15] 84/-1/-1->83->82
+ip-26-0-165-131:1630205:1630646 [0] NCCL INFO Trees [0] 73/-1/-1->72->80 [1] -1/-1/-1->72->79 [2] 73/-1/-1->72->79 [3] 73/-1/-1->72->79 [4] 73/-1/-1->72->79 [5] 73/-1/-1->72->79 [6] 73/-1/-1->72->79 [7] 73/-1/-1->72->79 [8] 73/80/64->72->88 [9] -1/-1/-1->72->79 [10] 73/-1/-1->72->79 [11] 73/-1/-1->72->79 [12] 73/-1/-1->72->79 [13] 73/-1/-1->72->79 [14] 73/-1/-1->72->79 [15] 73/-1/-1->72->79
+ip-26-0-165-131:1630211:1630649 [6] NCCL INFO Trees [0] 79/-1/-1->78->77 [1] 79/-1/-1->78->77 [2] 79/-1/-1->78->77 [3] 79/-1/-1->78->77 [4] 79/-1/-1->78->77 [5] 79/-1/-1->78->77 [6] 79/-1/-1->78->86 [7] -1/-1/-1->78->77 [8] 79/-1/-1->78->77 [9] 79/-1/-1->78->77 [10] 79/-1/-1->78->77 [11] 79/-1/-1->78->77 [12] 79/-1/-1->78->77 [13] 79/-1/-1->78->77 [14] 79/86/70->78->94 [15] -1/-1/-1->78->77
+ip-26-0-161-153:2636085:2636509 [1] NCCL INFO Trees [0] 50/-1/-1->49->48 [1] 50/41/57->49->33 [2] -1/-1/-1->49->48 [3] 50/-1/-1->49->48 [4] 50/-1/-1->49->48 [5] 50/-1/-1->49->48 [6] 50/-1/-1->49->48 [7] 50/-1/-1->49->48 [8] 50/-1/-1->49->48 [9] 50/-1/-1->49->41 [10] -1/-1/-1->49->48 [11] 50/-1/-1->49->48 [12] 50/-1/-1->49->48 [13] 50/-1/-1->49->48 [14] 50/-1/-1->49->48 [15] 50/-1/-1->49->48
+ip-26-0-161-153:2636085:2636509 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-225:124160:124581 [0] NCCL INFO Channel 05/16 :    0   4   7   6  13  11  10   9   8  12  15  14  21  19  18  17  16  20  23  22
+ip-26-0-168-52:2722051:2722465 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-168-52:2722048:2722469 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-138:96565:97004 [2] NCCL INFO Trees [0] 35/-1/-1->34->33 [1] 35/-1/-1->34->33 [2] 35/18/50->34->66 [3] -1/-1/-1->34->33 [4] 35/-1/-1->34->33 [5] 35/-1/-1->34->33 [6] 35/-1/-1->34->33 [7] 35/-1/-1->34->33 [8] 35/-1/-1->34->33 [9] 35/-1/-1->34->33 [10] 35/-1/-1->34->42 [11] -1/-1/-1->34->33 [12] 35/-1/-1->34->33 [13] 35/-1/-1->34->33 [14] 35/-1/-1->34->33 [15] 35/-1/-1->34->33
+ip-26-0-161-138:96567:97008 [4] NCCL INFO Trees [0] 37/-1/-1->36->35 [1] 37/-1/-1->36->35 [2] 37/-1/-1->36->35 [3] 37/-1/-1->36->35 [4] 37/20/52->36->68 [5] -1/-1/-1->36->35 [6] 37/-1/-1->36->35 [7] 37/-1/-1->36->35 [8] 37/-1/-1->36->35 [9] 37/-1/-1->36->35 [10] 37/-1/-1->36->35 [11] 37/-1/-1->36->35 [12] 37/-1/-1->36->44 [13] -1/-1/-1->36->35 [14] 37/-1/-1->36->35 [15] 37/-1/-1->36->35
+ip-26-0-161-138:96565:97004 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-138:96567:97008 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-202:153566:153980 [4] NCCL INFO Trees [0] 93/-1/-1->92->91 [1] 93/-1/-1->92->91 [2] 93/-1/-1->92->91 [3] 93/-1/-1->92->91 [4] 93/-1/-1->92->84 [5] -1/-1/-1->92->91 [6] 93/-1/-1->92->91 [7] 93/-1/-1->92->91 [8] 93/-1/-1->92->91 [9] 93/-1/-1->92->91 [10] 93/-1/-1->92->91 [11] 93/-1/-1->92->91 [12] 93/108/76->92->60 [13] -1/-1/-1->92->91 [14] 93/-1/-1->92->91 [15] 93/-1/-1->92->91
+ip-26-0-165-202:153566:153980 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-78:114517:114935 [5] NCCL INFO Trees [0] 70/-1/-1->69->68 [1] 70/-1/-1->69->68 [2] 70/-1/-1->69->68 [3] 70/-1/-1->69->68 [4] 70/-1/-1->69->68 [5] 70/37/101->69->5 [6] -1/-1/-1->69->68 [7] 70/-1/-1->69->68 [8] 70/-1/-1->69->68 [9] 70/-1/-1->69->68 [10] 70/-1/-1->69->68 [11] 70/-1/-1->69->68 [12] 70/-1/-1->69->68 [13] 70/-1/-1->69->77 [14] -1/-1/-1->69->68 [15] 70/-1/-1->69->68
+ip-26-0-161-78:114519:114931 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-78:114515:114936 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-78:114517:114935 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-78:114514:114932 [2] NCCL INFO Trees [0] 67/-1/-1->66->65 [1] 67/-1/-1->66->65 [2] 67/34/98->66->2 [3] -1/-1/-1->66->65 [4] 67/-1/-1->66->65 [5] 67/-1/-1->66->65 [6] 67/-1/-1->66->65 [7] 67/-1/-1->66->65 [8] 67/-1/-1->66->65 [9] 67/-1/-1->66->65 [10] 67/-1/-1->66->74 [11] -1/-1/-1->66->65 [12] 67/-1/-1->66->65 [13] 67/-1/-1->66->65 [14] 67/-1/-1->66->65 [15] 67/-1/-1->66->65
+ip-26-0-161-142:62883:63301 [2] NCCL INFO Trees [0] 43/-1/-1->42->41 [1] 43/-1/-1->42->41 [2] 43/-1/-1->42->50 [3] -1/-1/-1->42->41 [4] 43/-1/-1->42->41 [5] 43/-1/-1->42->41 [6] 43/-1/-1->42->41 [7] 43/-1/-1->42->41 [8] 43/-1/-1->42->41 [9] 43/-1/-1->42->41 [10] 43/50/34->42->26 [11] -1/-1/-1->42->41 [12] 43/-1/-1->42->41 [13] 43/-1/-1->42->41 [14] 43/-1/-1->42->41 [15] 43/-1/-1->42->41
+ip-26-0-161-142:62883:63301 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-59:1557269:1557684 [3] NCCL INFO Trees [0] 100/-1/-1->99->98 [1] 100/-1/-1->99->98 [2] 100/-1/-1->99->98 [3] 100/83/115->99->67 [4] -1/-1/-1->99->98 [5] 100/-1/-1->99->98 [6] 100/-1/-1->99->98 [7] 100/-1/-1->99->98 [8] 100/-1/-1->99->98 [9] 100/-1/-1->99->98 [10] 100/-1/-1->99->98 [11] 100/-1/-1->99->107 [12] -1/-1/-1->99->98 [13] 100/-1/-1->99->98 [14] 100/-1/-1->99->98 [15] 100/-1/-1->99->98
+ip-26-0-168-30:235649:236071 [1] NCCL INFO Trees [0] 106/-1/-1->105->104 [1] 106/-1/-1->105->113 [2] -1/-1/-1->105->104 [3] 106/-1/-1->105->104 [4] 106/-1/-1->105->104 [5] 106/-1/-1->105->104 [6] 106/-1/-1->105->104 [7] 106/-1/-1->105->104 [8] 106/-1/-1->105->104 [9] 106/113/97->105->89 [10] -1/-1/-1->105->104 [11] 106/-1/-1->105->104 [12] 106/-1/-1->105->104 [13] 106/-1/-1->105->104 [14] 106/-1/-1->105->104 [15] 106/-1/-1->105->104
+ip-26-0-168-30:235649:236071 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-168-34:272155:272574 [5] NCCL INFO Trees [0] 118/-1/-1->117->116 [1] 118/-1/-1->117->116 [2] 118/-1/-1->117->116 [3] 118/-1/-1->117->116 [4] 118/-1/-1->117->116 [5] 118/109/125->117->101 [6] -1/-1/-1->117->116 [7] 118/-1/-1->117->116 [8] 118/-1/-1->117->116 [9] 118/-1/-1->117->116 [10] 118/-1/-1->117->116 [11] 118/-1/-1->117->116 [12] 118/-1/-1->117->116 [13] 118/-1/-1->117->109 [14] -1/-1/-1->117->116 [15] 118/-1/-1->117->116
+ip-26-0-168-34:272155:272574 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-168-34:272157:272572 [7] NCCL INFO Trees [0] -1/-1/-1->119->118 [1] 112/-1/-1->119->118 [2] 112/-1/-1->119->118 [3] 112/-1/-1->119->118 [4] 112/-1/-1->119->118 [5] 112/-1/-1->119->118 [6] 112/-1/-1->119->118 [7] 112/111/127->119->103 [8] -1/-1/-1->119->118 [9] 112/-1/-1->119->118 [10] 112/-1/-1->119->118 [11] 112/-1/-1->119->118 [12] 112/-1/-1->119->118 [13] 112/-1/-1->119->118 [14] 112/-1/-1->119->118 [15] 112/-1/-1->119->111
+ip-26-0-161-103:96892:97308 [4] NCCL INFO Trees [0] 21/-1/-1->20->19 [1] 21/-1/-1->20->19 [2] 21/-1/-1->20->19 [3] 21/-1/-1->20->19 [4] 21/12/28->20->36 [5] -1/-1/-1->20->19 [6] 21/-1/-1->20->19 [7] 21/-1/-1->20->19 [8] 21/-1/-1->20->19 [9] 21/-1/-1->20->19 [10] 21/-1/-1->20->19 [11] 21/-1/-1->20->19 [12] 21/-1/-1->20->12 [13] -1/-1/-1->20->19 [14] 21/-1/-1->20->19 [15] 21/-1/-1->20->19
+ip-26-0-161-103:96894:97312 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-103:96892:97308 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-103:96888:97306 [0] NCCL INFO Trees [0] 17/8/24->16->32 [1] -1/-1/-1->16->23 [2] 17/-1/-1->16->23 [3] 17/-1/-1->16->23 [4] 17/-1/-1->16->23 [5] 17/-1/-1->16->23 [6] 17/-1/-1->16->23 [7] 17/-1/-1->16->23 [8] 17/-1/-1->16->8 [9] -1/-1/-1->16->23 [10] 17/-1/-1->16->23 [11] 17/-1/-1->16->23 [12] 17/-1/-1->16->23 [13] 17/-1/-1->16->23 [14] 17/-1/-1->16->23 [15] 17/-1/-1->16->23
+ip-26-0-161-103:96888:97306 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-242:1364907:1365329 [5] NCCL INFO Trees [0] 14/-1/-1->13->12 [1] 14/-1/-1->13->12 [2] 14/-1/-1->13->12 [3] 14/-1/-1->13->12 [4] 14/-1/-1->13->12 [5] 14/-1/-1->13->21 [6] -1/-1/-1->13->12 [7] 14/-1/-1->13->12 [8] 14/-1/-1->13->12 [9] 14/-1/-1->13->12 [10] 14/-1/-1->13->12 [11] 14/-1/-1->13->12 [12] 14/-1/-1->13->12 [13] 14/21/5->13->29 [14] -1/-1/-1->13->12 [15] 14/-1/-1->13->12
+ip-26-0-160-242:1364907:1365329 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-123:32706:33122 [2] NCCL INFO Trees [0] 27/-1/-1->26->25 [1] 27/-1/-1->26->25 [2] 27/-1/-1->26->18 [3] -1/-1/-1->26->25 [4] 27/-1/-1->26->25 [5] 27/-1/-1->26->25 [6] 27/-1/-1->26->25 [7] 27/-1/-1->26->25 [8] 27/-1/-1->26->25 [9] 27/-1/-1->26->25 [10] 27/42/10->26->58 [11] -1/-1/-1->26->25 [12] 27/-1/-1->26->25 [13] 27/-1/-1->26->25 [14] 27/-1/-1->26->25 [15] 27/-1/-1->26->25
+ip-26-0-161-123:32706:33122 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-164:1227464:1227884 [1] NCCL INFO Trees [0] 82/-1/-1->81->80 [1] 82/73/89->81->97 [2] -1/-1/-1->81->80 [3] 82/-1/-1->81->80 [4] 82/-1/-1->81->80 [5] 82/-1/-1->81->80 [6] 82/-1/-1->81->80 [7] 82/-1/-1->81->80 [8] 82/-1/-1->81->80 [9] 82/-1/-1->81->73 [10] -1/-1/-1->81->80 [11] 82/-1/-1->81->80 [12] 82/-1/-1->81->80 [13] 82/-1/-1->81->80 [14] 82/-1/-1->81->80 [15] 82/-1/-1->81->80
+ip-26-0-165-131:1630212:1630652 [7] NCCL INFO Trees [0] -1/-1/-1->79->78 [1] 72/-1/-1->79->78 [2] 72/-1/-1->79->78 [3] 72/-1/-1->79->78 [4] 72/-1/-1->79->78 [5] 72/-1/-1->79->78 [6] 72/-1/-1->79->78 [7] 72/-1/-1->79->87 [8] -1/-1/-1->79->78 [9] 72/-1/-1->79->78 [10] 72/-1/-1->79->78 [11] 72/-1/-1->79->78 [12] 72/-1/-1->79->78 [13] 72/-1/-1->79->78 [14] 72/-1/-1->79->78 [15] 72/87/71->79->95
+ip-26-0-165-131:1630205:1630646 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-131:1630211:1630649 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-131:1630210:1630653 [5] NCCL INFO Trees [0] 78/-1/-1->77->76 [1] 78/-1/-1->77->76 [2] 78/-1/-1->77->76 [3] 78/-1/-1->77->76 [4] 78/-1/-1->77->76 [5] 78/-1/-1->77->85 [6] -1/-1/-1->77->76 [7] 78/-1/-1->77->76 [8] 78/-1/-1->77->76 [9] 78/-1/-1->77->76 [10] 78/-1/-1->77->76 [11] 78/-1/-1->77->76 [12] 78/-1/-1->77->76 [13] 78/85/69->77->93 [14] -1/-1/-1->77->76 [15] 78/-1/-1->77->76
+ip-26-0-160-225:124160:124581 [0] NCCL INFO Channel 06/16 :    0   5   4   7  14  11  10   9   8  13  12  15  22  19  18  17  16  21  20  23
+ip-26-0-160-225:124160:124581 [0] NCCL INFO Channel 07/16 :    0   6   5   4  15  11  10   9   8  14  13  12  23  19  18  17  16  22  21  20
+ip-26-0-168-52:2722050:2722471 [5] NCCL INFO Trees [0] 126/-1/-1->125->124 [1] 126/-1/-1->125->124 [2] 126/-1/-1->125->124 [3] 126/-1/-1->125->124 [4] 126/-1/-1->125->124 [5] 126/-1/-1->125->117 [6] -1/-1/-1->125->124 [7] 126/-1/-1->125->124 [8] 126/-1/-1->125->124 [9] 126/-1/-1->125->124 [10] 126/-1/-1->125->124 [11] 126/-1/-1->125->124 [12] 126/-1/-1->125->124 [13] 126/61/-1->125->-1 [14] -1/-1/-1->125->124 [15] 126/-1/-1->125->124
+ip-26-0-168-52:2722050:2722471 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-168-52:2722046:2722466 [1] NCCL INFO Trees [0] 122/-1/-1->121->120 [1] 122/-1/-1->121->113 [2] -1/-1/-1->121->120 [3] 122/-1/-1->121->120 [4] 122/-1/-1->121->120 [5] 122/-1/-1->121->120 [6] 122/-1/-1->121->120 [7] 122/-1/-1->121->120 [8] 122/-1/-1->121->120 [9] 122/57/-1->121->-1 [10] -1/-1/-1->121->120 [11] 122/-1/-1->121->120 [12] 122/-1/-1->121->120 [13] 122/-1/-1->121->120 [14] 122/-1/-1->121->120 [15] 122/-1/-1->121->120
+ip-26-0-168-52:2722046:2722466 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-138:96563:97003 [0] NCCL INFO Trees [0] 33/16/48->32->64 [1] -1/-1/-1->32->39 [2] 33/-1/-1->32->39 [3] 33/-1/-1->32->39 [4] 33/-1/-1->32->39 [5] 33/-1/-1->32->39 [6] 33/-1/-1->32->39 [7] 33/-1/-1->32->39 [8] 33/-1/-1->32->40 [9] -1/-1/-1->32->39 [10] 33/-1/-1->32->39 [11] 33/-1/-1->32->39 [12] 33/-1/-1->32->39 [13] 33/-1/-1->32->39 [14] 33/-1/-1->32->39 [15] 33/-1/-1->32->39
+ip-26-0-161-138:96563:97003 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-138:96566:97010 [3] NCCL INFO Trees [0] 36/-1/-1->35->34 [1] 36/-1/-1->35->34 [2] 36/-1/-1->35->34 [3] 36/19/51->35->67 [4] -1/-1/-1->35->34 [5] 36/-1/-1->35->34 [6] 36/-1/-1->35->34 [7] 36/-1/-1->35->34 [8] 36/-1/-1->35->34 [9] 36/-1/-1->35->34 [10] 36/-1/-1->35->34 [11] 36/-1/-1->35->43 [12] -1/-1/-1->35->34 [13] 36/-1/-1->35->34 [14] 36/-1/-1->35->34 [15] 36/-1/-1->35->34
+ip-26-0-165-202:153565:153985 [3] NCCL INFO Trees [0] 92/-1/-1->91->90 [1] 92/-1/-1->91->90 [2] 92/-1/-1->91->90 [3] 92/-1/-1->91->83 [4] -1/-1/-1->91->90 [5] 92/-1/-1->91->90 [6] 92/-1/-1->91->90 [7] 92/-1/-1->91->90 [8] 92/-1/-1->91->90 [9] 92/-1/-1->91->90 [10] 92/-1/-1->91->90 [11] 92/107/75->91->59 [12] -1/-1/-1->91->90 [13] 92/-1/-1->91->90 [14] 92/-1/-1->91->90 [15] 92/-1/-1->91->90
+ip-26-0-165-202:153565:153985 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-78:114513:114933 [1] NCCL INFO Trees [0] 66/-1/-1->65->64 [1] 66/33/97->65->1 [2] -1/-1/-1->65->64 [3] 66/-1/-1->65->64 [4] 66/-1/-1->65->64 [5] 66/-1/-1->65->64 [6] 66/-1/-1->65->64 [7] 66/-1/-1->65->64 [8] 66/-1/-1->65->64 [9] 66/-1/-1->65->73 [10] -1/-1/-1->65->64 [11] 66/-1/-1->65->64 [12] 66/-1/-1->65->64 [13] 66/-1/-1->65->64 [14] 66/-1/-1->65->64 [15] 66/-1/-1->65->64
+ip-26-0-161-78:114514:114932 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-78:114513:114933 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-78:114512:114930 [0] NCCL INFO Trees [0] 65/32/96->64->0 [1] -1/-1/-1->64->71 [2] 65/-1/-1->64->71 [3] 65/-1/-1->64->71 [4] 65/-1/-1->64->71 [5] 65/-1/-1->64->71 [6] 65/-1/-1->64->71 [7] 65/-1/-1->64->71 [8] 65/-1/-1->64->72 [9] -1/-1/-1->64->71 [10] 65/-1/-1->64->71 [11] 65/-1/-1->64->71 [12] 65/-1/-1->64->71 [13] 65/-1/-1->64->71 [14] 65/-1/-1->64->71 [15] 65/-1/-1->64->71
+ip-26-0-161-78:114512:114930 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-142:62885:63304 [4] NCCL INFO Trees [0] 45/-1/-1->44->43 [1] 45/-1/-1->44->43 [2] 45/-1/-1->44->43 [3] 45/-1/-1->44->43 [4] 45/-1/-1->44->52 [5] -1/-1/-1->44->43 [6] 45/-1/-1->44->43 [7] 45/-1/-1->44->43 [8] 45/-1/-1->44->43 [9] 45/-1/-1->44->43 [10] 45/-1/-1->44->43 [11] 45/-1/-1->44->43 [12] 45/52/36->44->28 [13] -1/-1/-1->44->43 [14] 45/-1/-1->44->43 [15] 45/-1/-1->44->43
+ip-26-0-165-59:1557269:1557684 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-59:1557267:1557683 [1] NCCL INFO Trees [0] 98/-1/-1->97->96 [1] 98/81/113->97->65 [2] -1/-1/-1->97->96 [3] 98/-1/-1->97->96 [4] 98/-1/-1->97->96 [5] 98/-1/-1->97->96 [6] 98/-1/-1->97->96 [7] 98/-1/-1->97->96 [8] 98/-1/-1->97->96 [9] 98/-1/-1->97->105 [10] -1/-1/-1->97->96 [11] 98/-1/-1->97->96 [12] 98/-1/-1->97->96 [13] 98/-1/-1->97->96 [14] 98/-1/-1->97->96 [15] 98/-1/-1->97->96
+ip-26-0-165-59:1557267:1557683 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-168-30:235654:236070 [6] NCCL INFO Trees [0] 111/-1/-1->110->109 [1] 111/-1/-1->110->109 [2] 111/-1/-1->110->109 [3] 111/-1/-1->110->109 [4] 111/-1/-1->110->109 [5] 111/-1/-1->110->109 [6] 111/-1/-1->110->118 [7] -1/-1/-1->110->109 [8] 111/-1/-1->110->109 [9] 111/-1/-1->110->109 [10] 111/-1/-1->110->109 [11] 111/-1/-1->110->109 [12] 111/-1/-1->110->109 [13] 111/-1/-1->110->109 [14] 111/118/102->110->94 [15] -1/-1/-1->110->109
+ip-26-0-168-30:235654:236070 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-168-34:272157:272572 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-103:96889:97309 [1] NCCL INFO Trees [0] 18/-1/-1->17->16 [1] 18/9/25->17->33 [2] -1/-1/-1->17->16 [3] 18/-1/-1->17->16 [4] 18/-1/-1->17->16 [5] 18/-1/-1->17->16 [6] 18/-1/-1->17->16 [7] 18/-1/-1->17->16 [8] 18/-1/-1->17->16 [9] 18/-1/-1->17->9 [10] -1/-1/-1->17->16 [11] 18/-1/-1->17->16 [12] 18/-1/-1->17->16 [13] 18/-1/-1->17->16 [14] 18/-1/-1->17->16 [15] 18/-1/-1->17->16
+ip-26-0-160-242:1364906:1365331 [4] NCCL INFO Trees [0] 13/-1/-1->12->11 [1] 13/-1/-1->12->11 [2] 13/-1/-1->12->11 [3] 13/-1/-1->12->11 [4] 13/-1/-1->12->20 [5] -1/-1/-1->12->11 [6] 13/-1/-1->12->11 [7] 13/-1/-1->12->11 [8] 13/-1/-1->12->11 [9] 13/-1/-1->12->11 [10] 13/-1/-1->12->11 [11] 13/-1/-1->12->11 [12] 13/20/4->12->28 [13] -1/-1/-1->12->11 [14] 13/-1/-1->12->11 [15] 13/-1/-1->12->11
+ip-26-0-160-242:1364902:1365333 [0] NCCL INFO Trees [0] 9/-1/-1->8->16 [1] -1/-1/-1->8->15 [2] 9/-1/-1->8->15 [3] 9/-1/-1->8->15 [4] 9/-1/-1->8->15 [5] 9/-1/-1->8->15 [6] 9/-1/-1->8->15 [7] 9/-1/-1->8->15 [8] 9/16/0->8->24 [9] -1/-1/-1->8->15 [10] 9/-1/-1->8->15 [11] 9/-1/-1->8->15 [12] 9/-1/-1->8->15 [13] 9/-1/-1->8->15 [14] 9/-1/-1->8->15 [15] 9/-1/-1->8->15
+ip-26-0-161-123:32704:33123 [0] NCCL INFO Trees [0] 25/-1/-1->24->16 [1] -1/-1/-1->24->31 [2] 25/-1/-1->24->31 [3] 25/-1/-1->24->31 [4] 25/-1/-1->24->31 [5] 25/-1/-1->24->31 [6] 25/-1/-1->24->31 [7] 25/-1/-1->24->31 [8] 25/40/8->24->56 [9] -1/-1/-1->24->31 [10] 25/-1/-1->24->31 [11] 25/-1/-1->24->31 [12] 25/-1/-1->24->31 [13] 25/-1/-1->24->31 [14] 25/-1/-1->24->31 [15] 25/-1/-1->24->31
+ip-26-0-161-123:32704:33123 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-164:1227467:1227886 [4] NCCL INFO Trees [0] 85/-1/-1->84->83 [1] 85/-1/-1->84->83 [2] 85/-1/-1->84->83 [3] 85/-1/-1->84->83 [4] 85/76/92->84->100 [5] -1/-1/-1->84->83 [6] 85/-1/-1->84->83 [7] 85/-1/-1->84->83 [8] 85/-1/-1->84->83 [9] 85/-1/-1->84->83 [10] 85/-1/-1->84->83 [11] 85/-1/-1->84->83 [12] 85/-1/-1->84->76 [13] -1/-1/-1->84->83 [14] 85/-1/-1->84->83 [15] 85/-1/-1->84->83
+ip-26-0-165-164:1227466:1227887 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-164:1227464:1227884 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-164:1227467:1227886 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-131:1630207:1630648 [2] NCCL INFO Trees [0] 75/-1/-1->74->73 [1] 75/-1/-1->74->73 [2] 75/-1/-1->74->82 [3] -1/-1/-1->74->73 [4] 75/-1/-1->74->73 [5] 75/-1/-1->74->73 [6] 75/-1/-1->74->73 [7] 75/-1/-1->74->73 [8] 75/-1/-1->74->73 [9] 75/-1/-1->74->73 [10] 75/82/66->74->90 [11] -1/-1/-1->74->73 [12] 75/-1/-1->74->73 [13] 75/-1/-1->74->73 [14] 75/-1/-1->74->73 [15] 75/-1/-1->74->73
+ip-26-0-165-131:1630212:1630652 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-131:1630210:1630653 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-131:1630207:1630648 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-225:124160:124581 [0] NCCL INFO Channel 08/16 :    0   1   2   3   4   5   6   7   8   9  10  11  12  13  14  15  16  17  18  19
+ip-26-0-160-225:124160:124581 [0] NCCL INFO Channel 09/16 :    0   7   6   5   4   3   2   9   8  15  14  13  12  11  10  17  16  23  22  21
+ip-26-0-168-52:2722045:2722470 [0] NCCL INFO Trees [0] 121/-1/-1->120->112 [1] -1/-1/-1->120->127 [2] 121/-1/-1->120->127 [3] 121/-1/-1->120->127 [4] 121/-1/-1->120->127 [5] 121/-1/-1->120->127 [6] 121/-1/-1->120->127 [7] 121/-1/-1->120->127 [8] 121/56/-1->120->-1 [9] -1/-1/-1->120->127 [10] 121/-1/-1->120->127 [11] 121/-1/-1->120->127 [12] 121/-1/-1->120->127 [13] 121/-1/-1->120->127 [14] 121/-1/-1->120->127 [15] 121/-1/-1->120->127
+ip-26-0-161-138:96566:97010 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-202:153564:153984 [2] NCCL INFO Trees [0] 91/-1/-1->90->89 [1] 91/-1/-1->90->89 [2] 91/-1/-1->90->82 [3] -1/-1/-1->90->89 [4] 91/-1/-1->90->89 [5] 91/-1/-1->90->89 [6] 91/-1/-1->90->89 [7] 91/-1/-1->90->89 [8] 91/-1/-1->90->89 [9] 91/-1/-1->90->89 [10] 91/106/74->90->58 [11] -1/-1/-1->90->89 [12] 91/-1/-1->90->89 [13] 91/-1/-1->90->89 [14] 91/-1/-1->90->89 [15] 91/-1/-1->90->89
+ip-26-0-165-202:153564:153984 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-202:153562:153983 [0] NCCL INFO Trees [0] 89/-1/-1->88->80 [1] -1/-1/-1->88->95 [2] 89/-1/-1->88->95 [3] 89/-1/-1->88->95 [4] 89/-1/-1->88->95 [5] 89/-1/-1->88->95 [6] 89/-1/-1->88->95 [7] 89/-1/-1->88->95 [8] 89/104/72->88->56 [9] -1/-1/-1->88->95 [10] 89/-1/-1->88->95 [11] 89/-1/-1->88->95 [12] 89/-1/-1->88->95 [13] 89/-1/-1->88->95 [14] 89/-1/-1->88->95 [15] 89/-1/-1->88->95
+ip-26-0-165-202:153562:153983 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-78:114518:114937 [6] NCCL INFO Trees [0] 71/-1/-1->70->69 [1] 71/-1/-1->70->69 [2] 71/-1/-1->70->69 [3] 71/-1/-1->70->69 [4] 71/-1/-1->70->69 [5] 71/-1/-1->70->69 [6] 71/38/102->70->6 [7] -1/-1/-1->70->69 [8] 71/-1/-1->70->69 [9] 71/-1/-1->70->69 [10] 71/-1/-1->70->69 [11] 71/-1/-1->70->69 [12] 71/-1/-1->70->69 [13] 71/-1/-1->70->69 [14] 71/-1/-1->70->78 [15] -1/-1/-1->70->69
+ip-26-0-161-78:114518:114937 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-142:62885:63304 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-103:96889:97309 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-103:96890:97313 [2] NCCL INFO Trees [0] 19/-1/-1->18->17 [1] 19/-1/-1->18->17 [2] 19/10/26->18->34 [3] -1/-1/-1->18->17 [4] 19/-1/-1->18->17 [5] 19/-1/-1->18->17 [6] 19/-1/-1->18->17 [7] 19/-1/-1->18->17 [8] 19/-1/-1->18->17 [9] 19/-1/-1->18->17 [10] 19/-1/-1->18->10 [11] -1/-1/-1->18->17 [12] 19/-1/-1->18->17 [13] 19/-1/-1->18->17 [14] 19/-1/-1->18->17 [15] 19/-1/-1->18->17
+ip-26-0-161-103:96890:97313 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-242:1364906:1365331 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-242:1364904:1365335 [2] NCCL INFO Trees [0] 11/-1/-1->10->9 [1] 11/-1/-1->10->9 [2] 11/-1/-1->10->18 [3] -1/-1/-1->10->9 [4] 11/-1/-1->10->9 [5] 11/-1/-1->10->9 [6] 11/-1/-1->10->9 [7] 11/-1/-1->10->9 [8] 11/-1/-1->10->9 [9] 11/-1/-1->10->9 [10] 11/18/2->10->26 [11] -1/-1/-1->10->9 [12] 11/-1/-1->10->9 [13] 11/-1/-1->10->9 [14] 11/-1/-1->10->9 [15] 11/-1/-1->10->9
+ip-26-0-160-242:1364902:1365333 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-242:1364904:1365335 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-123:32705:33128 [1] NCCL INFO Trees [0] 26/-1/-1->25->24 [1] 26/-1/-1->25->17 [2] -1/-1/-1->25->24 [3] 26/-1/-1->25->24 [4] 26/-1/-1->25->24 [5] 26/-1/-1->25->24 [6] 26/-1/-1->25->24 [7] 26/-1/-1->25->24 [8] 26/-1/-1->25->24 [9] 26/41/9->25->57 [10] -1/-1/-1->25->24 [11] 26/-1/-1->25->24 [12] 26/-1/-1->25->24 [13] 26/-1/-1->25->24 [14] 26/-1/-1->25->24 [15] 26/-1/-1->25->24
+ip-26-0-161-123:32705:33128 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-164:1227465:1227881 [2] NCCL INFO Trees [0] 83/-1/-1->82->81 [1] 83/-1/-1->82->81 [2] 83/74/90->82->98 [3] -1/-1/-1->82->81 [4] 83/-1/-1->82->81 [5] 83/-1/-1->82->81 [6] 83/-1/-1->82->81 [7] 83/-1/-1->82->81 [8] 83/-1/-1->82->81 [9] 83/-1/-1->82->81 [10] 83/-1/-1->82->74 [11] -1/-1/-1->82->81 [12] 83/-1/-1->82->81 [13] 83/-1/-1->82->81 [14] 83/-1/-1->82->81 [15] 83/-1/-1->82->81
+ip-26-0-165-164:1227465:1227881 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-164:1227470:1227885 [7] NCCL INFO Trees [0] -1/-1/-1->87->86 [1] 80/-1/-1->87->86 [2] 80/-1/-1->87->86 [3] 80/-1/-1->87->86 [4] 80/-1/-1->87->86 [5] 80/-1/-1->87->86 [6] 80/-1/-1->87->86 [7] 80/79/95->87->103 [8] -1/-1/-1->87->86 [9] 80/-1/-1->87->86 [10] 80/-1/-1->87->86 [11] 80/-1/-1->87->86 [12] 80/-1/-1->87->86 [13] 80/-1/-1->87->86 [14] 80/-1/-1->87->86 [15] 80/-1/-1->87->79
+ip-26-0-165-131:1630209:1630651 [4] NCCL INFO Trees [0] 77/-1/-1->76->75 [1] 77/-1/-1->76->75 [2] 77/-1/-1->76->75 [3] 77/-1/-1->76->75 [4] 77/-1/-1->76->84 [5] -1/-1/-1->76->75 [6] 77/-1/-1->76->75 [7] 77/-1/-1->76->75 [8] 77/-1/-1->76->75 [9] 77/-1/-1->76->75 [10] 77/-1/-1->76->75 [11] 77/-1/-1->76->75 [12] 77/84/68->76->92 [13] -1/-1/-1->76->75 [14] 77/-1/-1->76->75 [15] 77/-1/-1->76->75
+ip-26-0-165-131:1630209:1630651 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-225:124160:124581 [0] NCCL INFO Channel 10/16 :    0   7   6   5   4   3  10   9   8  15  14  13  12  11  18  17  16  23  22  21
+ip-26-0-168-52:2722045:2722470 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-138:96564:97009 [1] NCCL INFO Trees [0] 34/-1/-1->33->32 [1] 34/17/49->33->65 [2] -1/-1/-1->33->32 [3] 34/-1/-1->33->32 [4] 34/-1/-1->33->32 [5] 34/-1/-1->33->32 [6] 34/-1/-1->33->32 [7] 34/-1/-1->33->32 [8] 34/-1/-1->33->32 [9] 34/-1/-1->33->41 [10] -1/-1/-1->33->32 [11] 34/-1/-1->33->32 [12] 34/-1/-1->33->32 [13] 34/-1/-1->33->32 [14] 34/-1/-1->33->32 [15] 34/-1/-1->33->32
+ip-26-0-161-138:96564:97009 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-202:153563:153986 [1] NCCL INFO Trees [0] 90/-1/-1->89->88 [1] 90/-1/-1->89->81 [2] -1/-1/-1->89->88 [3] 90/-1/-1->89->88 [4] 90/-1/-1->89->88 [5] 90/-1/-1->89->88 [6] 90/-1/-1->89->88 [7] 90/-1/-1->89->88 [8] 90/-1/-1->89->88 [9] 90/105/73->89->57 [10] -1/-1/-1->89->88 [11] 90/-1/-1->89->88 [12] 90/-1/-1->89->88 [13] 90/-1/-1->89->88 [14] 90/-1/-1->89->88 [15] 90/-1/-1->89->88
+ip-26-0-165-202:153567:153979 [5] NCCL INFO Trees [0] 94/-1/-1->93->92 [1] 94/-1/-1->93->92 [2] 94/-1/-1->93->92 [3] 94/-1/-1->93->92 [4] 94/-1/-1->93->92 [5] 94/-1/-1->93->85 [6] -1/-1/-1->93->92 [7] 94/-1/-1->93->92 [8] 94/-1/-1->93->92 [9] 94/-1/-1->93->92 [10] 94/-1/-1->93->92 [11] 94/-1/-1->93->92 [12] 94/-1/-1->93->92 [13] 94/109/77->93->61 [14] -1/-1/-1->93->92 [15] 94/-1/-1->93->92
+ip-26-0-165-202:153563:153986 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-142:62882:63300 [1] NCCL INFO Trees [0] 42/-1/-1->41->40 [1] 42/-1/-1->41->49 [2] -1/-1/-1->41->40 [3] 42/-1/-1->41->40 [4] 42/-1/-1->41->40 [5] 42/-1/-1->41->40 [6] 42/-1/-1->41->40 [7] 42/-1/-1->41->40 [8] 42/-1/-1->41->40 [9] 42/49/33->41->25 [10] -1/-1/-1->41->40 [11] 42/-1/-1->41->40 [12] 42/-1/-1->41->40 [13] 42/-1/-1->41->40 [14] 42/-1/-1->41->40 [15] 42/-1/-1->41->40
+ip-26-0-161-142:62882:63300 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-103:96891:97311 [3] NCCL INFO Trees [0] 20/-1/-1->19->18 [1] 20/-1/-1->19->18 [2] 20/-1/-1->19->18 [3] 20/11/27->19->35 [4] -1/-1/-1->19->18 [5] 20/-1/-1->19->18 [6] 20/-1/-1->19->18 [7] 20/-1/-1->19->18 [8] 20/-1/-1->19->18 [9] 20/-1/-1->19->18 [10] 20/-1/-1->19->18 [11] 20/-1/-1->19->11 [12] -1/-1/-1->19->18 [13] 20/-1/-1->19->18 [14] 20/-1/-1->19->18 [15] 20/-1/-1->19->18
+ip-26-0-161-103:96891:97311 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-242:1364903:1365334 [1] NCCL INFO Trees [0] 10/-1/-1->9->8 [1] 10/-1/-1->9->17 [2] -1/-1/-1->9->8 [3] 10/-1/-1->9->8 [4] 10/-1/-1->9->8 [5] 10/-1/-1->9->8 [6] 10/-1/-1->9->8 [7] 10/-1/-1->9->8 [8] 10/-1/-1->9->8 [9] 10/17/1->9->25 [10] -1/-1/-1->9->8 [11] 10/-1/-1->9->8 [12] 10/-1/-1->9->8 [13] 10/-1/-1->9->8 [14] 10/-1/-1->9->8 [15] 10/-1/-1->9->8
+ip-26-0-160-242:1364903:1365334 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-164:1227470:1227885 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-225:124160:124581 [0] NCCL INFO Channel 11/16 :    0   7   6   5   4  11  10   9   8  15  14  13  12  19  18  17  16  23  22  21
+ip-26-0-165-202:153567:153979 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-142:62881:63298 [0] NCCL INFO Trees [0] 41/-1/-1->40->48 [1] -1/-1/-1->40->47 [2] 41/-1/-1->40->47 [3] 41/-1/-1->40->47 [4] 41/-1/-1->40->47 [5] 41/-1/-1->40->47 [6] 41/-1/-1->40->47 [7] 41/-1/-1->40->47 [8] 41/48/32->40->24 [9] -1/-1/-1->40->47 [10] 41/-1/-1->40->47 [11] 41/-1/-1->40->47 [12] 41/-1/-1->40->47 [13] 41/-1/-1->40->47 [14] 41/-1/-1->40->47 [15] 41/-1/-1->40->47
+ip-26-0-165-164:1227463:1227880 [0] NCCL INFO Trees [0] 81/72/88->80->96 [1] -1/-1/-1->80->87 [2] 81/-1/-1->80->87 [3] 81/-1/-1->80->87 [4] 81/-1/-1->80->87 [5] 81/-1/-1->80->87 [6] 81/-1/-1->80->87 [7] 81/-1/-1->80->87 [8] 81/-1/-1->80->72 [9] -1/-1/-1->80->87 [10] 81/-1/-1->80->87 [11] 81/-1/-1->80->87 [12] 81/-1/-1->80->87 [13] 81/-1/-1->80->87 [14] 81/-1/-1->80->87 [15] 81/-1/-1->80->87
+ip-26-0-165-164:1227463:1227880 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-225:124160:124581 [0] NCCL INFO Channel 12/16 :    0   7   6   5  12  11  10   9   8  15  14  13  20  19  18  17  16  23  22  21
+ip-26-0-160-225:124161:124582 [1] NCCL INFO NVLS Head  0:  0  8 16 24 32 40 48 56 64 72 80 88 96 104 112 120
+ip-26-0-161-142:62881:63298 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-225:124160:124581 [0] NCCL INFO Channel 13/16 :    0   4   7   6  13  11  10   9   8  12  15  14  21  19  18  17  16  20  23  22
+ip-26-0-160-225:124161:124582 [1] NCCL INFO NVLS Head  1:  1  9 17 25 33 41 49 57 65 73 81 89 97 105 113 121
+ip-26-0-160-225:124161:124582 [1] NCCL INFO NVLS Head  2:  2 10 18 26 34 42 50 58 66 74 82 90 98 106 114 122
+ip-26-0-160-225:124161:124582 [1] NCCL INFO NVLS Head  3:  3 11 19 27 35 43 51 59 67 75 83 91 99 107 115 123
+ip-26-0-160-225:124160:124581 [0] NCCL INFO Channel 14/16 :    0   5   4   7  14  11  10   9   8  13  12  15  22  19  18  17  16  21  20  23
+ip-26-0-160-225:124161:124582 [1] NCCL INFO NVLS Head  4:  4 12 20 28 36 44 52 60 68 76 84 92 100 108 116 124
+ip-26-0-160-225:124161:124582 [1] NCCL INFO NVLS Head  5:  5 13 21 29 37 45 53 61 69 77 85 93 101 109 117 125
+ip-26-0-160-225:124160:124581 [0] NCCL INFO Channel 15/16 :    0   6   5   4  15  11  10   9   8  14  13  12  23  19  18  17  16  22  21  20
+ip-26-0-160-225:124161:124582 [1] NCCL INFO NVLS Head  6:  6 14 22 30 38 46 54 62 70 78 86 94 102 110 118 126
+ip-26-0-160-225:124161:124582 [1] NCCL INFO NVLS Head  7:  7 15 23 31 39 47 55 63 71 79 87 95 103 111 119 127
+ip-26-0-160-225:124162:124587 [2] NCCL INFO NVLS Head  0:  0  8 16 24 32 40 48 56 64 72 80 88 96 104 112 120
+ip-26-0-160-225:124162:124587 [2] NCCL INFO NVLS Head  1:  1  9 17 25 33 41 49 57 65 73 81 89 97 105 113 121
+ip-26-0-160-225:124160:124581 [0] NCCL INFO Trees [0] 1/64/-1->0->-1 [1] -1/-1/-1->0->7 [2] 1/-1/-1->0->7 [3] 1/-1/-1->0->7 [4] 1/-1/-1->0->7 [5] 1/-1/-1->0->7 [6] 1/-1/-1->0->7 [7] 1/-1/-1->0->7 [8] 1/-1/-1->0->8 [9] -1/-1/-1->0->7 [10] 1/-1/-1->0->7 [11] 1/-1/-1->0->7 [12] 1/-1/-1->0->7 [13] 1/-1/-1->0->7 [14] 1/-1/-1->0->7 [15] 1/-1/-1->0->7
+ip-26-0-160-225:124162:124587 [2] NCCL INFO NVLS Head  2:  2 10 18 26 34 42 50 58 66 74 82 90 98 106 114 122
+ip-26-0-160-225:124160:124581 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-225:124162:124587 [2] NCCL INFO NVLS Head  3:  3 11 19 27 35 43 51 59 67 75 83 91 99 107 115 123
+ip-26-0-160-225:124162:124587 [2] NCCL INFO NVLS Head  4:  4 12 20 28 36 44 52 60 68 76 84 92 100 108 116 124
+ip-26-0-160-225:124162:124587 [2] NCCL INFO NVLS Head  5:  5 13 21 29 37 45 53 61 69 77 85 93 101 109 117 125
+ip-26-0-160-225:124162:124587 [2] NCCL INFO NVLS Head  6:  6 14 22 30 38 46 54 62 70 78 86 94 102 110 118 126
+ip-26-0-160-225:124162:124587 [2] NCCL INFO NVLS Head  7:  7 15 23 31 39 47 55 63 71 79 87 95 103 111 119 127
+ip-26-0-160-225:124166:124583 [6] NCCL INFO NVLS Head  0:  0  8 16 24 32 40 48 56 64 72 80 88 96 104 112 120
+ip-26-0-160-225:124164:124585 [4] NCCL INFO NVLS Head  0:  0  8 16 24 32 40 48 56 64 72 80 88 96 104 112 120
+ip-26-0-160-225:124167:124586 [7] NCCL INFO NVLS Head  0:  0  8 16 24 32 40 48 56 64 72 80 88 96 104 112 120
+ip-26-0-160-225:124166:124583 [6] NCCL INFO NVLS Head  1:  1  9 17 25 33 41 49 57 65 73 81 89 97 105 113 121
+ip-26-0-160-225:124164:124585 [4] NCCL INFO NVLS Head  1:  1  9 17 25 33 41 49 57 65 73 81 89 97 105 113 121
+ip-26-0-160-225:124167:124586 [7] NCCL INFO NVLS Head  1:  1  9 17 25 33 41 49 57 65 73 81 89 97 105 113 121
+ip-26-0-160-225:124166:124583 [6] NCCL INFO NVLS Head  2:  2 10 18 26 34 42 50 58 66 74 82 90 98 106 114 122
+ip-26-0-160-225:124164:124585 [4] NCCL INFO NVLS Head  2:  2 10 18 26 34 42 50 58 66 74 82 90 98 106 114 122
+ip-26-0-160-225:124167:124586 [7] NCCL INFO NVLS Head  2:  2 10 18 26 34 42 50 58 66 74 82 90 98 106 114 122
+ip-26-0-160-225:124166:124583 [6] NCCL INFO NVLS Head  3:  3 11 19 27 35 43 51 59 67 75 83 91 99 107 115 123
+ip-26-0-160-225:124167:124586 [7] NCCL INFO NVLS Head  3:  3 11 19 27 35 43 51 59 67 75 83 91 99 107 115 123
+ip-26-0-160-225:124164:124585 [4] NCCL INFO NVLS Head  3:  3 11 19 27 35 43 51 59 67 75 83 91 99 107 115 123
+ip-26-0-160-225:124166:124583 [6] NCCL INFO NVLS Head  4:  4 12 20 28 36 44 52 60 68 76 84 92 100 108 116 124
+ip-26-0-160-225:124167:124586 [7] NCCL INFO NVLS Head  4:  4 12 20 28 36 44 52 60 68 76 84 92 100 108 116 124
+ip-26-0-160-225:124164:124585 [4] NCCL INFO NVLS Head  4:  4 12 20 28 36 44 52 60 68 76 84 92 100 108 116 124
+ip-26-0-160-225:124165:124584 [5] NCCL INFO NVLS Head  0:  0  8 16 24 32 40 48 56 64 72 80 88 96 104 112 120
+ip-26-0-160-225:124166:124583 [6] NCCL INFO NVLS Head  5:  5 13 21 29 37 45 53 61 69 77 85 93 101 109 117 125
+ip-26-0-160-225:124167:124586 [7] NCCL INFO NVLS Head  5:  5 13 21 29 37 45 53 61 69 77 85 93 101 109 117 125
+ip-26-0-160-225:124164:124585 [4] NCCL INFO NVLS Head  5:  5 13 21 29 37 45 53 61 69 77 85 93 101 109 117 125
+ip-26-0-160-225:124166:124583 [6] NCCL INFO NVLS Head  6:  6 14 22 30 38 46 54 62 70 78 86 94 102 110 118 126
+ip-26-0-160-225:124167:124586 [7] NCCL INFO NVLS Head  6:  6 14 22 30 38 46 54 62 70 78 86 94 102 110 118 126
+ip-26-0-160-225:124165:124584 [5] NCCL INFO NVLS Head  1:  1  9 17 25 33 41 49 57 65 73 81 89 97 105 113 121
+ip-26-0-160-225:124164:124585 [4] NCCL INFO NVLS Head  6:  6 14 22 30 38 46 54 62 70 78 86 94 102 110 118 126
+ip-26-0-160-225:124166:124583 [6] NCCL INFO NVLS Head  7:  7 15 23 31 39 47 55 63 71 79 87 95 103 111 119 127
+ip-26-0-160-225:124167:124586 [7] NCCL INFO NVLS Head  7:  7 15 23 31 39 47 55 63 71 79 87 95 103 111 119 127
+ip-26-0-160-225:124164:124585 [4] NCCL INFO NVLS Head  7:  7 15 23 31 39 47 55 63 71 79 87 95 103 111 119 127
+ip-26-0-160-225:124161:124582 [1] NCCL INFO Trees [0] 2/-1/-1->1->0 [1] 2/65/-1->1->-1 [2] -1/-1/-1->1->0 [3] 2/-1/-1->1->0 [4] 2/-1/-1->1->0 [5] 2/-1/-1->1->0 [6] 2/-1/-1->1->0 [7] 2/-1/-1->1->0 [8] 2/-1/-1->1->0 [9] 2/-1/-1->1->9 [10] -1/-1/-1->1->0 [11] 2/-1/-1->1->0 [12] 2/-1/-1->1->0 [13] 2/-1/-1->1->0 [14] 2/-1/-1->1->0 [15] 2/-1/-1->1->0
+ip-26-0-160-225:124165:124584 [5] NCCL INFO NVLS Head  2:  2 10 18 26 34 42 50 58 66 74 82 90 98 106 114 122
+ip-26-0-160-225:124161:124582 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-225:124165:124584 [5] NCCL INFO NVLS Head  3:  3 11 19 27 35 43 51 59 67 75 83 91 99 107 115 123
+ip-26-0-160-225:124165:124584 [5] NCCL INFO NVLS Head  4:  4 12 20 28 36 44 52 60 68 76 84 92 100 108 116 124
+ip-26-0-160-225:124165:124584 [5] NCCL INFO NVLS Head  5:  5 13 21 29 37 45 53 61 69 77 85 93 101 109 117 125
+ip-26-0-160-225:124165:124584 [5] NCCL INFO NVLS Head  6:  6 14 22 30 38 46 54 62 70 78 86 94 102 110 118 126
+ip-26-0-160-225:124165:124584 [5] NCCL INFO NVLS Head  7:  7 15 23 31 39 47 55 63 71 79 87 95 103 111 119 127
+ip-26-0-160-225:124162:124587 [2] NCCL INFO Trees [0] 3/-1/-1->2->1 [1] 3/-1/-1->2->1 [2] 3/66/-1->2->-1 [3] -1/-1/-1->2->1 [4] 3/-1/-1->2->1 [5] 3/-1/-1->2->1 [6] 3/-1/-1->2->1 [7] 3/-1/-1->2->1 [8] 3/-1/-1->2->1 [9] 3/-1/-1->2->1 [10] 3/-1/-1->2->10 [11] -1/-1/-1->2->1 [12] 3/-1/-1->2->1 [13] 3/-1/-1->2->1 [14] 3/-1/-1->2->1 [15] 3/-1/-1->2->1
+ip-26-0-160-225:124162:124587 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-225:124167:124586 [7] NCCL INFO Trees [0] -1/-1/-1->7->6 [1] 0/-1/-1->7->6 [2] 0/-1/-1->7->6 [3] 0/-1/-1->7->6 [4] 0/-1/-1->7->6 [5] 0/-1/-1->7->6 [6] 0/-1/-1->7->6 [7] 0/71/-1->7->-1 [8] -1/-1/-1->7->6 [9] 0/-1/-1->7->6 [10] 0/-1/-1->7->6 [11] 0/-1/-1->7->6 [12] 0/-1/-1->7->6 [13] 0/-1/-1->7->6 [14] 0/-1/-1->7->6 [15] 0/-1/-1->7->15
+ip-26-0-160-225:124166:124583 [6] NCCL INFO Trees [0] 7/-1/-1->6->5 [1] 7/-1/-1->6->5 [2] 7/-1/-1->6->5 [3] 7/-1/-1->6->5 [4] 7/-1/-1->6->5 [5] 7/-1/-1->6->5 [6] 7/70/-1->6->-1 [7] -1/-1/-1->6->5 [8] 7/-1/-1->6->5 [9] 7/-1/-1->6->5 [10] 7/-1/-1->6->5 [11] 7/-1/-1->6->5 [12] 7/-1/-1->6->5 [13] 7/-1/-1->6->5 [14] 7/-1/-1->6->14 [15] -1/-1/-1->6->5
+ip-26-0-160-225:124164:124585 [4] NCCL INFO Trees [0] 5/-1/-1->4->3 [1] 5/-1/-1->4->3 [2] 5/-1/-1->4->3 [3] 5/-1/-1->4->3 [4] 5/68/-1->4->-1 [5] -1/-1/-1->4->3 [6] 5/-1/-1->4->3 [7] 5/-1/-1->4->3 [8] 5/-1/-1->4->3 [9] 5/-1/-1->4->3 [10] 5/-1/-1->4->3 [11] 5/-1/-1->4->3 [12] 5/-1/-1->4->12 [13] -1/-1/-1->4->3 [14] 5/-1/-1->4->3 [15] 5/-1/-1->4->3
+ip-26-0-160-225:124167:124586 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-225:124166:124583 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-225:124164:124585 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-225:124163:124588 [3] NCCL INFO NVLS Head  0:  0  8 16 24 32 40 48 56 64 72 80 88 96 104 112 120
+ip-26-0-160-225:124163:124588 [3] NCCL INFO NVLS Head  1:  1  9 17 25 33 41 49 57 65 73 81 89 97 105 113 121
+ip-26-0-160-225:124165:124584 [5] NCCL INFO Trees [0] 6/-1/-1->5->4 [1] 6/-1/-1->5->4 [2] 6/-1/-1->5->4 [3] 6/-1/-1->5->4 [4] 6/-1/-1->5->4 [5] 6/69/-1->5->-1 [6] -1/-1/-1->5->4 [7] 6/-1/-1->5->4 [8] 6/-1/-1->5->4 [9] 6/-1/-1->5->4 [10] 6/-1/-1->5->4 [11] 6/-1/-1->5->4 [12] 6/-1/-1->5->4 [13] 6/-1/-1->5->13 [14] -1/-1/-1->5->4 [15] 6/-1/-1->5->4
+ip-26-0-160-225:124163:124588 [3] NCCL INFO NVLS Head  2:  2 10 18 26 34 42 50 58 66 74 82 90 98 106 114 122
+ip-26-0-160-225:124165:124584 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-225:124163:124588 [3] NCCL INFO NVLS Head  3:  3 11 19 27 35 43 51 59 67 75 83 91 99 107 115 123
+ip-26-0-160-225:124163:124588 [3] NCCL INFO NVLS Head  4:  4 12 20 28 36 44 52 60 68 76 84 92 100 108 116 124
+ip-26-0-160-225:124163:124588 [3] NCCL INFO NVLS Head  5:  5 13 21 29 37 45 53 61 69 77 85 93 101 109 117 125
+ip-26-0-160-225:124163:124588 [3] NCCL INFO NVLS Head  6:  6 14 22 30 38 46 54 62 70 78 86 94 102 110 118 126
+ip-26-0-160-225:124163:124588 [3] NCCL INFO NVLS Head  7:  7 15 23 31 39 47 55 63 71 79 87 95 103 111 119 127
+ip-26-0-160-225:124163:124588 [3] NCCL INFO Trees [0] 4/-1/-1->3->2 [1] 4/-1/-1->3->2 [2] 4/-1/-1->3->2 [3] 4/67/-1->3->-1 [4] -1/-1/-1->3->2 [5] 4/-1/-1->3->2 [6] 4/-1/-1->3->2 [7] 4/-1/-1->3->2 [8] 4/-1/-1->3->2 [9] 4/-1/-1->3->2 [10] 4/-1/-1->3->2 [11] 4/-1/-1->3->11 [12] -1/-1/-1->3->2 [13] 4/-1/-1->3->2 [14] 4/-1/-1->3->2 [15] 4/-1/-1->3->2
+ip-26-0-160-225:124163:124588 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-153:2636089:2636508 [5] NCCL INFO Channel 00/0 : 53[5] -> 54[6] via P2P/IPC
+ip-26-0-161-153:2636090:2636505 [6] NCCL INFO Channel 00/0 : 54[6] -> 55[7] via P2P/IPC
+ip-26-0-161-153:2636089:2636508 [5] NCCL INFO Channel 08/0 : 53[5] -> 54[6] via P2P/IPC
+ip-26-0-161-153:2636090:2636505 [6] NCCL INFO Channel 08/0 : 54[6] -> 55[7] via P2P/IPC
+ip-26-0-168-34:272154:272569 [4] NCCL INFO Channel 00/0 : 116[4] -> 117[5] via P2P/IPC
+ip-26-0-165-164:1227469:1227882 [6] NCCL INFO Channel 00/0 : 86[6] -> 87[7] via P2P/IPC
+ip-26-0-161-103:96893:97310 [5] NCCL INFO Channel 00/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-165-202:153566:153980 [4] NCCL INFO Channel 00/0 : 92[4] -> 93[5] via P2P/IPC
+ip-26-0-161-178:2815965:2816395 [2] NCCL INFO Channel 00/0 : 58[2] -> 59[3] via P2P/IPC
+ip-26-0-161-123:32706:33122 [2] NCCL INFO Channel 00/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-161-123:32708:33126 [4] NCCL INFO Channel 00/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-168-30:235652:236066 [4] NCCL INFO Channel 00/0 : 108[4] -> 109[5] via P2P/IPC
+ip-26-0-161-142:62886:63303 [5] NCCL INFO Channel 00/0 : 45[5] -> 46[6] via P2P/IPC
+ip-26-0-161-178:2815967:2816399 [4] NCCL INFO Channel 00/0 : 60[4] -> 61[5] via P2P/IPC
+ip-26-0-161-138:96568:97007 [5] NCCL INFO Channel 00/0 : 37[5] -> 38[6] via P2P/IPC
+ip-26-0-168-52:2722051:2722465 [6] NCCL INFO Channel 00/0 : 126[6] -> 127[7] via P2P/IPC
+ip-26-0-168-52:2722049:2722467 [4] NCCL INFO Channel 00/0 : 124[4] -> 125[5] via P2P/IPC
+ip-26-0-165-164:1227468:1227883 [5] NCCL INFO Channel 00/0 : 85[5] -> 86[6] via P2P/IPC
+ip-26-0-161-103:96892:97308 [4] NCCL INFO Channel 00/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-161-138:96567:97008 [4] NCCL INFO Channel 00/0 : 36[4] -> 37[5] via P2P/IPC
+ip-26-0-165-59:1557272:1557687 [6] NCCL INFO Channel 00/0 : 102[6] -> 103[7] via P2P/IPC
+ip-26-0-168-30:235653:236067 [5] NCCL INFO Channel 00/0 : 109[5] -> 110[6] via P2P/IPC
+ip-26-0-160-242:1364904:1365335 [2] NCCL INFO Channel 00/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-161-178:2815968:2816394 [5] NCCL INFO Channel 00/0 : 61[5] -> 62[6] via P2P/IPC
+ip-26-0-160-242:1364908:1365332 [6] NCCL INFO Channel 00/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-165-164:1227467:1227886 [4] NCCL INFO Channel 00/0 : 84[4] -> 85[5] via P2P/IPC
+ip-26-0-168-34:272156:272570 [6] NCCL INFO Channel 00/0 : 118[6] -> 119[7] via P2P/IPC
+ip-26-0-165-59:1557271:1557690 [5] NCCL INFO Channel 00/0 : 101[5] -> 102[6] via P2P/IPC
+ip-26-0-160-242:1364907:1365329 [5] NCCL INFO Channel 00/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-165-202:153564:153984 [2] NCCL INFO Channel 00/0 : 90[2] -> 91[3] via P2P/IPC
+ip-26-0-161-142:62882:63300 [1] NCCL INFO Channel 00/0 : 41[1] -> 42[2] via P2P/IPC
+ip-26-0-161-142:62883:63301 [2] NCCL INFO Channel 00/0 : 42[2] -> 43[3] via P2P/IPC
+ip-26-0-168-34:272155:272574 [5] NCCL INFO Channel 00/0 : 117[5] -> 118[6] via P2P/IPC
+ip-26-0-161-178:2815969:2816398 [6] NCCL INFO Channel 00/0 : 62[6] -> 63[7] via P2P/IPC
+ip-26-0-168-52:2722046:2722466 [1] NCCL INFO Channel 00/0 : 121[1] -> 122[2] via P2P/IPC
+ip-26-0-161-178:2815966:2816396 [3] NCCL INFO Channel 00/0 : 59[3] -> 60[4] via P2P/IPC
+ip-26-0-161-178:2815964:2816401 [1] NCCL INFO Channel 00/0 : 57[1] -> 58[2] via P2P/IPC
+ip-26-0-160-242:1364906:1365331 [4] NCCL INFO Channel 00/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-161-142:62887:63305 [6] NCCL INFO Channel 00/0 : 46[6] -> 47[7] via P2P/IPC
+ip-26-0-161-78:114516:114934 [4] NCCL INFO Channel 00/0 : 68[4] -> 69[5] via P2P/IPC
+ip-26-0-161-142:62885:63304 [4] NCCL INFO Channel 00/0 : 44[4] -> 45[5] via P2P/IPC
+ip-26-0-161-78:114517:114935 [5] NCCL INFO Channel 00/0 : 69[5] -> 70[6] via P2P/IPC
+ip-26-0-168-34:272154:272569 [4] NCCL INFO Channel 08/0 : 116[4] -> 117[5] via P2P/IPC
+ip-26-0-161-178:2815965:2816395 [2] NCCL INFO Channel 08/0 : 58[2] -> 59[3] via P2P/IPC
+ip-26-0-165-164:1227469:1227882 [6] NCCL INFO Channel 08/0 : 86[6] -> 87[7] via P2P/IPC
+ip-26-0-161-138:96565:97004 [2] NCCL INFO Channel 00/0 : 34[2] -> 35[3] via P2P/IPC
+ip-26-0-165-202:153566:153980 [4] NCCL INFO Channel 08/0 : 92[4] -> 93[5] via P2P/IPC
+ip-26-0-161-78:114518:114937 [6] NCCL INFO Channel 00/0 : 70[6] -> 71[7] via P2P/IPC
+ip-26-0-161-103:96893:97310 [5] NCCL INFO Channel 08/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-160-242:1364905:1365328 [3] NCCL INFO Channel 00/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-168-52:2722050:2722471 [5] NCCL INFO Channel 00/0 : 125[5] -> 126[6] via P2P/IPC
+ip-26-0-161-123:32705:33128 [1] NCCL INFO Channel 00/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-168-52:2722047:2722468 [2] NCCL INFO Channel 00/0 : 122[2] -> 123[3] via P2P/IPC
+ip-26-0-165-202:153565:153985 [3] NCCL INFO Channel 00/0 : 91[3] -> 92[4] via P2P/IPC
+ip-26-0-168-30:235649:236071 [1] NCCL INFO Channel 00/0 : 105[1] -> 106[2] via P2P/IPC
+ip-26-0-161-138:96564:97009 [1] NCCL INFO Channel 00/0 : 33[1] -> 34[2] via P2P/IPC
+ip-26-0-161-178:2815967:2816399 [4] NCCL INFO Channel 08/0 : 60[4] -> 61[5] via P2P/IPC
+ip-26-0-168-34:272151:272571 [1] NCCL INFO Channel 00/0 : 113[1] -> 114[2] via P2P/IPC
+ip-26-0-161-123:32710:33125 [6] NCCL INFO Channel 00/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-168-52:2722048:2722469 [3] NCCL INFO Channel 00/0 : 123[3] -> 124[4] via P2P/IPC
+ip-26-0-168-30:235650:236065 [2] NCCL INFO Channel 00/0 : 106[2] -> 107[3] via P2P/IPC
+ip-26-0-161-78:114513:114933 [1] NCCL INFO Channel 00/0 : 65[1] -> 66[2] via P2P/IPC
+ip-26-0-165-202:153567:153979 [5] NCCL INFO Channel 00/0 : 93[5] -> 94[6] via P2P/IPC
+ip-26-0-161-142:62886:63303 [5] NCCL INFO Channel 08/0 : 45[5] -> 46[6] via P2P/IPC
+ip-26-0-165-59:1557270:1557688 [4] NCCL INFO Channel 00/0 : 100[4] -> 101[5] via P2P/IPC
+ip-26-0-165-59:1557268:1557686 [2] NCCL INFO Channel 00/0 : 98[2] -> 99[3] via P2P/IPC
+ip-26-0-168-30:235651:236068 [3] NCCL INFO Channel 00/0 : 107[3] -> 108[4] via P2P/IPC
+ip-26-0-168-34:272153:272573 [3] NCCL INFO Channel 00/0 : 115[3] -> 116[4] via P2P/IPC
+ip-26-0-168-30:235652:236066 [4] NCCL INFO Channel 08/0 : 108[4] -> 109[5] via P2P/IPC
+ip-26-0-161-138:96569:97006 [6] NCCL INFO Channel 00/0 : 38[6] -> 39[7] via P2P/IPC
+ip-26-0-161-78:114515:114936 [3] NCCL INFO Channel 00/0 : 67[3] -> 68[4] via P2P/IPC
+ip-26-0-160-242:1364903:1365334 [1] NCCL INFO Channel 00/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-161-153:2636086:2636507 [2] NCCL INFO Channel 00/0 : 50[2] -> 51[3] via P2P/IPC
+ip-26-0-165-59:1557269:1557684 [3] NCCL INFO Channel 00/0 : 99[3] -> 100[4] via P2P/IPC
+ip-26-0-165-59:1557267:1557683 [1] NCCL INFO Channel 00/0 : 97[1] -> 98[2] via P2P/IPC
+ip-26-0-168-34:272152:272575 [2] NCCL INFO Channel 00/0 : 114[2] -> 115[3] via P2P/IPC
+ip-26-0-161-103:96890:97313 [2] NCCL INFO Channel 00/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-161-78:114514:114932 [2] NCCL INFO Channel 00/0 : 66[2] -> 67[3] via P2P/IPC
+ip-26-0-168-30:235653:236067 [5] NCCL INFO Channel 08/0 : 109[5] -> 110[6] via P2P/IPC
+ip-26-0-161-138:96568:97007 [5] NCCL INFO Channel 08/0 : 37[5] -> 38[6] via P2P/IPC
+ip-26-0-165-202:153563:153986 [1] NCCL INFO Channel 00/0 : 89[1] -> 90[2] via P2P/IPC
+ip-26-0-160-242:1364904:1365335 [2] NCCL INFO Channel 08/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-165-164:1227464:1227884 [1] NCCL INFO Channel 00/0 : 81[1] -> 82[2] via P2P/IPC
+ip-26-0-168-52:2722049:2722467 [4] NCCL INFO Channel 08/0 : 124[4] -> 125[5] via P2P/IPC
+ip-26-0-165-59:1557272:1557687 [6] NCCL INFO Channel 08/0 : 102[6] -> 103[7] via P2P/IPC
+ip-26-0-161-178:2815968:2816394 [5] NCCL INFO Channel 08/0 : 61[5] -> 62[6] via P2P/IPC
+ip-26-0-161-153:2636085:2636509 [1] NCCL INFO Channel 00/0 : 49[1] -> 50[2] via P2P/IPC
+ip-26-0-165-202:153564:153984 [2] NCCL INFO Channel 08/0 : 90[2] -> 91[3] via P2P/IPC
+ip-26-0-161-103:96892:97308 [4] NCCL INFO Channel 08/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-160-242:1364908:1365332 [6] NCCL INFO Channel 08/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-165-164:1227468:1227883 [5] NCCL INFO Channel 08/0 : 85[5] -> 86[6] via P2P/IPC
+ip-26-0-168-52:2722051:2722465 [6] NCCL INFO Channel 08/0 : 126[6] -> 127[7] via P2P/IPC
+ip-26-0-161-138:96567:97008 [4] NCCL INFO Channel 08/0 : 36[4] -> 37[5] via P2P/IPC
+ip-26-0-168-30:235654:236070 [6] NCCL INFO Channel 00/0 : 110[6] -> 111[7] via P2P/IPC
+ip-26-0-161-103:96889:97309 [1] NCCL INFO Channel 00/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-165-59:1557271:1557690 [5] NCCL INFO Channel 08/0 : 101[5] -> 102[6] via P2P/IPC
+ip-26-0-161-178:2815966:2816396 [3] NCCL INFO Channel 08/0 : 59[3] -> 60[4] via P2P/IPC
+ip-26-0-160-242:1364907:1365329 [5] NCCL INFO Channel 08/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-165-164:1227466:1227887 [3] NCCL INFO Channel 00/0 : 83[3] -> 84[4] via P2P/IPC
+ip-26-0-161-138:96566:97010 [3] NCCL INFO Channel 00/0 : 35[3] -> 36[4] via P2P/IPC
+ip-26-0-161-103:96894:97312 [6] NCCL INFO Channel 00/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-161-153:2636088:2636510 [4] NCCL INFO Channel 00/0 : 52[4] -> 53[5] via P2P/IPC
+ip-26-0-161-178:2815969:2816398 [6] NCCL INFO Channel 08/0 : 62[6] -> 63[7] via P2P/IPC
+ip-26-0-161-142:62883:63301 [2] NCCL INFO Channel 08/0 : 42[2] -> 43[3] via P2P/IPC
+ip-26-0-161-142:62882:63300 [1] NCCL INFO Channel 08/0 : 41[1] -> 42[2] via P2P/IPC
+ip-26-0-165-164:1227465:1227881 [2] NCCL INFO Channel 00/0 : 82[2] -> 83[3] via P2P/IPC
+ip-26-0-165-131:1630211:1630649 [6] NCCL INFO Channel 00/0 : 78[6] -> 79[7] via P2P/IPC
+ip-26-0-168-52:2722046:2722466 [1] NCCL INFO Channel 08/0 : 121[1] -> 122[2] via P2P/IPC
+ip-26-0-161-103:96891:97311 [3] NCCL INFO Channel 00/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-168-34:272156:272570 [6] NCCL INFO Channel 08/0 : 118[6] -> 119[7] via P2P/IPC
+ip-26-0-161-178:2815964:2816401 [1] NCCL INFO Channel 08/0 : 57[1] -> 58[2] via P2P/IPC
+ip-26-0-165-164:1227467:1227886 [4] NCCL INFO Channel 08/0 : 84[4] -> 85[5] via P2P/IPC
+ip-26-0-160-242:1364906:1365331 [4] NCCL INFO Channel 08/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-165-202:153565:153985 [3] NCCL INFO Channel 08/0 : 91[3] -> 92[4] via P2P/IPC
+ip-26-0-168-34:272155:272574 [5] NCCL INFO Channel 08/0 : 117[5] -> 118[6] via P2P/IPC
+ip-26-0-161-142:62887:63305 [6] NCCL INFO Channel 08/0 : 46[6] -> 47[7] via P2P/IPC
+ip-26-0-161-123:32705:33128 [1] NCCL INFO Channel 08/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-161-78:114516:114934 [4] NCCL INFO Channel 08/0 : 68[4] -> 69[5] via P2P/IPC
+ip-26-0-160-242:1364905:1365328 [3] NCCL INFO Channel 08/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-168-30:235649:236071 [1] NCCL INFO Channel 08/0 : 105[1] -> 106[2] via P2P/IPC
+ip-26-0-161-142:62885:63304 [4] NCCL INFO Channel 08/0 : 44[4] -> 45[5] via P2P/IPC
+ip-26-0-168-52:2722047:2722468 [2] NCCL INFO Channel 08/0 : 122[2] -> 123[3] via P2P/IPC
+ip-26-0-161-78:114517:114935 [5] NCCL INFO Channel 08/0 : 69[5] -> 70[6] via P2P/IPC
+ip-26-0-168-52:2722050:2722471 [5] NCCL INFO Channel 08/0 : 125[5] -> 126[6] via P2P/IPC
+ip-26-0-161-138:96565:97004 [2] NCCL INFO Channel 08/0 : 34[2] -> 35[3] via P2P/IPC
+ip-26-0-165-202:153567:153979 [5] NCCL INFO Channel 08/0 : 93[5] -> 94[6] via P2P/IPC
+ip-26-0-161-78:114518:114937 [6] NCCL INFO Channel 08/0 : 70[6] -> 71[7] via P2P/IPC
+ip-26-0-161-138:96564:97009 [1] NCCL INFO Channel 08/0 : 33[1] -> 34[2] via P2P/IPC
+ip-26-0-165-202:153563:153986 [1] NCCL INFO Channel 08/0 : 89[1] -> 90[2] via P2P/IPC
+ip-26-0-165-59:1557268:1557686 [2] NCCL INFO Channel 08/0 : 98[2] -> 99[3] via P2P/IPC
+ip-26-0-161-123:32710:33125 [6] NCCL INFO Channel 08/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-165-131:1630210:1630653 [5] NCCL INFO Channel 00/0 : 77[5] -> 78[6] via P2P/IPC
+ip-26-0-168-30:235651:236068 [3] NCCL INFO Channel 08/0 : 107[3] -> 108[4] via P2P/IPC
+ip-26-0-161-153:2636087:2636506 [3] NCCL INFO Channel 00/0 : 51[3] -> 52[4] via P2P/IPC
+ip-26-0-168-30:235650:236065 [2] NCCL INFO Channel 08/0 : 106[2] -> 107[3] via P2P/IPC
+ip-26-0-161-153:2636086:2636507 [2] NCCL INFO Channel 08/0 : 50[2] -> 51[3] via P2P/IPC
+ip-26-0-165-59:1557270:1557688 [4] NCCL INFO Channel 08/0 : 100[4] -> 101[5] via P2P/IPC
+ip-26-0-168-34:272151:272571 [1] NCCL INFO Channel 08/0 : 113[1] -> 114[2] via P2P/IPC
+ip-26-0-161-138:96569:97006 [6] NCCL INFO Channel 08/0 : 38[6] -> 39[7] via P2P/IPC
+ip-26-0-161-103:96890:97313 [2] NCCL INFO Channel 08/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-160-242:1364903:1365334 [1] NCCL INFO Channel 08/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-165-131:1630208:1630650 [3] NCCL INFO Channel 00/0 : 75[3] -> 76[4] via P2P/IPC
+ip-26-0-165-131:1630206:1630647 [1] NCCL INFO Channel 00/0 : 73[1] -> 74[2] via P2P/IPC
+ip-26-0-168-34:272153:272573 [3] NCCL INFO Channel 08/0 : 115[3] -> 116[4] via P2P/IPC
+ip-26-0-168-52:2722048:2722469 [3] NCCL INFO Channel 08/0 : 123[3] -> 124[4] via P2P/IPC
+ip-26-0-165-131:1630211:1630649 [6] NCCL INFO Channel 08/0 : 78[6] -> 79[7] via P2P/IPC
+ip-26-0-161-153:2636085:2636509 [1] NCCL INFO Channel 08/0 : 49[1] -> 50[2] via P2P/IPC
+ip-26-0-161-78:114513:114933 [1] NCCL INFO Channel 08/0 : 65[1] -> 66[2] via P2P/IPC
+ip-26-0-165-59:1557267:1557683 [1] NCCL INFO Channel 08/0 : 97[1] -> 98[2] via P2P/IPC
+ip-26-0-161-78:114515:114936 [3] NCCL INFO Channel 08/0 : 67[3] -> 68[4] via P2P/IPC
+ip-26-0-165-59:1557269:1557684 [3] NCCL INFO Channel 08/0 : 99[3] -> 100[4] via P2P/IPC
+ip-26-0-168-34:272152:272575 [2] NCCL INFO Channel 08/0 : 114[2] -> 115[3] via P2P/IPC
+ip-26-0-165-164:1227464:1227884 [1] NCCL INFO Channel 08/0 : 81[1] -> 82[2] via P2P/IPC
+ip-26-0-161-78:114514:114932 [2] NCCL INFO Channel 08/0 : 66[2] -> 67[3] via P2P/IPC
+ip-26-0-161-153:2636088:2636510 [4] NCCL INFO Channel 08/0 : 52[4] -> 53[5] via P2P/IPC
+ip-26-0-168-30:235654:236070 [6] NCCL INFO Channel 08/0 : 110[6] -> 111[7] via P2P/IPC
+ip-26-0-161-103:96889:97309 [1] NCCL INFO Channel 08/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-165-164:1227465:1227881 [2] NCCL INFO Channel 08/0 : 82[2] -> 83[3] via P2P/IPC
+ip-26-0-165-164:1227466:1227887 [3] NCCL INFO Channel 08/0 : 83[3] -> 84[4] via P2P/IPC
+ip-26-0-165-202:153566:153980 [4] NCCL INFO Channel 05/0 : 92[4] -> 95[7] via P2P/IPC
+ip-26-0-161-103:96894:97312 [6] NCCL INFO Channel 08/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-161-178:2815967:2816399 [4] NCCL INFO Channel 05/0 : 60[4] -> 63[7] via P2P/IPC
+ip-26-0-161-103:96891:97311 [3] NCCL INFO Channel 08/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-161-138:96566:97010 [3] NCCL INFO Channel 08/0 : 35[3] -> 36[4] via P2P/IPC
+ip-26-0-161-123:32709:33129 [5] NCCL INFO Channel 00/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-160-242:1364906:1365331 [4] NCCL INFO Channel 05/0 : 12[4] -> 15[7] via P2P/IPC
+ip-26-0-165-131:1630210:1630653 [5] NCCL INFO Channel 08/0 : 77[5] -> 78[6] via P2P/IPC
+ip-26-0-165-202:153566:153980 [4] NCCL INFO Channel 06/0 : 92[4] -> 95[7] via P2P/IPC
+ip-26-0-168-30:235652:236066 [4] NCCL INFO Channel 05/0 : 108[4] -> 111[7] via P2P/IPC
+ip-26-0-161-123:32707:33124 [3] NCCL INFO Channel 00/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-161-153:2636087:2636506 [3] NCCL INFO Channel 08/0 : 51[3] -> 52[4] via P2P/IPC
+ip-26-0-161-178:2815967:2816399 [4] NCCL INFO Channel 06/0 : 60[4] -> 63[7] via P2P/IPC
+ip-26-0-165-202:153568:153982 [6] NCCL INFO Channel 00/0 : 94[6] -> 95[7] via P2P/IPC
+ip-26-0-168-52:2722049:2722467 [4] NCCL INFO Channel 05/0 : 124[4] -> 127[7] via P2P/IPC
+ip-26-0-168-34:272154:272569 [4] NCCL INFO Channel 05/0 : 116[4] -> 119[7] via P2P/IPC
+ip-26-0-161-138:96567:97008 [4] NCCL INFO Channel 05/0 : 36[4] -> 39[7] via P2P/IPC
+ip-26-0-160-242:1364906:1365331 [4] NCCL INFO Channel 06/0 : 12[4] -> 15[7] via P2P/IPC
+ip-26-0-161-142:62884:63299 [3] NCCL INFO Channel 00/0 : 43[3] -> 44[4] via P2P/IPC
+ip-26-0-168-30:235652:236066 [4] NCCL INFO Channel 06/0 : 108[4] -> 111[7] via P2P/IPC
+ip-26-0-165-202:153566:153980 [4] NCCL INFO Channel 13/0 : 92[4] -> 95[7] via P2P/IPC
+ip-26-0-161-178:2815967:2816399 [4] NCCL INFO Channel 13/0 : 60[4] -> 63[7] via P2P/IPC
+ip-26-0-165-131:1630207:1630648 [2] NCCL INFO Channel 00/0 : 74[2] -> 75[3] via P2P/IPC
+ip-26-0-165-164:1227467:1227886 [4] NCCL INFO Channel 05/0 : 84[4] -> 87[7] via P2P/IPC
+ip-26-0-165-59:1557270:1557688 [4] NCCL INFO Channel 05/0 : 100[4] -> 103[7] via P2P/IPC
+ip-26-0-161-78:114516:114934 [4] NCCL INFO Channel 05/0 : 68[4] -> 71[7] via P2P/IPC
+ip-26-0-168-34:272154:272569 [4] NCCL INFO Channel 06/0 : 116[4] -> 119[7] via P2P/IPC
+ip-26-0-161-103:96892:97308 [4] NCCL INFO Channel 05/0 : 20[4] -> 23[7] via P2P/IPC
+ip-26-0-161-123:32709:33129 [5] NCCL INFO Channel 08/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-165-131:1630206:1630647 [1] NCCL INFO Channel 08/0 : 73[1] -> 74[2] via P2P/IPC
+ip-26-0-165-131:1630208:1630650 [3] NCCL INFO Channel 08/0 : 75[3] -> 76[4] via P2P/IPC
+ip-26-0-161-123:32706:33122 [2] NCCL INFO Channel 08/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-161-138:96567:97008 [4] NCCL INFO Channel 06/0 : 36[4] -> 39[7] via P2P/IPC
+ip-26-0-168-52:2722049:2722467 [4] NCCL INFO Channel 06/0 : 124[4] -> 127[7] via P2P/IPC
+ip-26-0-165-202:153568:153982 [6] NCCL INFO Channel 08/0 : 94[6] -> 95[7] via P2P/IPC
+ip-26-0-168-30:235652:236066 [4] NCCL INFO Channel 13/0 : 108[4] -> 111[7] via P2P/IPC
+ip-26-0-161-178:2815967:2816399 [4] NCCL INFO Channel 14/0 : 60[4] -> 63[7] via P2P/IPC
+ip-26-0-165-131:1630209:1630651 [4] NCCL INFO Channel 00/0 : 76[4] -> 77[5] via P2P/IPC
+ip-26-0-165-202:153566:153980 [4] NCCL INFO Channel 14/0 : 92[4] -> 95[7] via P2P/IPC
+ip-26-0-161-142:62884:63299 [3] NCCL INFO Channel 08/0 : 43[3] -> 44[4] via P2P/IPC
+ip-26-0-160-242:1364906:1365331 [4] NCCL INFO Channel 13/0 : 12[4] -> 15[7] via P2P/IPC
+ip-26-0-165-59:1557270:1557688 [4] NCCL INFO Channel 06/0 : 100[4] -> 103[7] via P2P/IPC
+ip-26-0-161-153:2636088:2636510 [4] NCCL INFO Channel 05/0 : 52[4] -> 55[7] via P2P/IPC
+ip-26-0-168-34:272154:272569 [4] NCCL INFO Channel 13/0 : 116[4] -> 119[7] via P2P/IPC
+ip-26-0-165-164:1227467:1227886 [4] NCCL INFO Channel 06/0 : 84[4] -> 87[7] via P2P/IPC
+ip-26-0-161-78:114516:114934 [4] NCCL INFO Channel 06/0 : 68[4] -> 71[7] via P2P/IPC
+ip-26-0-161-103:96892:97308 [4] NCCL INFO Channel 06/0 : 20[4] -> 23[7] via P2P/IPC
+ip-26-0-161-123:32707:33124 [3] NCCL INFO Channel 08/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-161-138:96567:97008 [4] NCCL INFO Channel 13/0 : 36[4] -> 39[7] via P2P/IPC
+ip-26-0-168-52:2722049:2722467 [4] NCCL INFO Channel 13/0 : 124[4] -> 127[7] via P2P/IPC
+ip-26-0-168-30:235652:236066 [4] NCCL INFO Channel 14/0 : 108[4] -> 111[7] via P2P/IPC
+ip-26-0-160-242:1364906:1365331 [4] NCCL INFO Channel 14/0 : 12[4] -> 15[7] via P2P/IPC
+ip-26-0-168-34:272154:272569 [4] NCCL INFO Channel 14/0 : 116[4] -> 119[7] via P2P/IPC
+ip-26-0-161-153:2636088:2636510 [4] NCCL INFO Channel 06/0 : 52[4] -> 55[7] via P2P/IPC
+ip-26-0-165-59:1557270:1557688 [4] NCCL INFO Channel 13/0 : 100[4] -> 103[7] via P2P/IPC
+ip-26-0-161-78:114516:114934 [4] NCCL INFO Channel 13/0 : 68[4] -> 71[7] via P2P/IPC
+ip-26-0-161-103:96892:97308 [4] NCCL INFO Channel 13/0 : 20[4] -> 23[7] via P2P/IPC
+ip-26-0-161-138:96567:97008 [4] NCCL INFO Channel 14/0 : 36[4] -> 39[7] via P2P/IPC
+ip-26-0-165-131:1630207:1630648 [2] NCCL INFO Channel 08/0 : 74[2] -> 75[3] via P2P/IPC
+ip-26-0-168-52:2722049:2722467 [4] NCCL INFO Channel 14/0 : 124[4] -> 127[7] via P2P/IPC
+ip-26-0-161-142:62885:63304 [4] NCCL INFO Channel 05/0 : 44[4] -> 47[7] via P2P/IPC
+ip-26-0-165-59:1557270:1557688 [4] NCCL INFO Channel 14/0 : 100[4] -> 103[7] via P2P/IPC
+ip-26-0-161-78:114516:114934 [4] NCCL INFO Channel 14/0 : 68[4] -> 71[7] via P2P/IPC
+ip-26-0-161-103:96892:97308 [4] NCCL INFO Channel 14/0 : 20[4] -> 23[7] via P2P/IPC
+ip-26-0-161-153:2636088:2636510 [4] NCCL INFO Channel 13/0 : 52[4] -> 55[7] via P2P/IPC
+ip-26-0-161-123:32708:33126 [4] NCCL INFO Channel 08/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-165-131:1630209:1630651 [4] NCCL INFO Channel 08/0 : 76[4] -> 77[5] via P2P/IPC
+ip-26-0-161-142:62885:63304 [4] NCCL INFO Channel 06/0 : 44[4] -> 47[7] via P2P/IPC
+ip-26-0-161-153:2636088:2636510 [4] NCCL INFO Channel 14/0 : 52[4] -> 55[7] via P2P/IPC
+ip-26-0-161-142:62885:63304 [4] NCCL INFO Channel 13/0 : 44[4] -> 47[7] via P2P/IPC
+ip-26-0-165-131:1630209:1630651 [4] NCCL INFO Channel 05/0 : 76[4] -> 79[7] via P2P/IPC
+ip-26-0-161-142:62885:63304 [4] NCCL INFO Channel 14/0 : 44[4] -> 47[7] via P2P/IPC
+ip-26-0-165-164:1227467:1227886 [4] NCCL INFO Channel 13/0 : 84[4] -> 87[7] via P2P/IPC
+ip-26-0-165-131:1630209:1630651 [4] NCCL INFO Channel 06/0 : 76[4] -> 79[7] via P2P/IPC
+ip-26-0-165-164:1227467:1227886 [4] NCCL INFO Channel 14/0 : 84[4] -> 87[7] via P2P/IPC
+ip-26-0-165-131:1630209:1630651 [4] NCCL INFO Channel 13/0 : 76[4] -> 79[7] via P2P/IPC
+ip-26-0-161-123:32708:33126 [4] NCCL INFO Channel 05/0 : 28[4] -> 31[7] via P2P/IPC
+ip-26-0-165-131:1630209:1630651 [4] NCCL INFO Channel 14/0 : 76[4] -> 79[7] via P2P/IPC
+ip-26-0-161-123:32708:33126 [4] NCCL INFO Channel 06/0 : 28[4] -> 31[7] via P2P/IPC
+ip-26-0-161-123:32708:33126 [4] NCCL INFO Channel 13/0 : 28[4] -> 31[7] via P2P/IPC
+ip-26-0-161-123:32708:33126 [4] NCCL INFO Channel 14/0 : 28[4] -> 31[7] via P2P/IPC
+ip-26-0-160-225:124161:124582 [1] NCCL INFO Channel 00/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-225:124164:124585 [4] NCCL INFO Channel 00/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-225:124166:124583 [6] NCCL INFO Channel 00/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-225:124165:124584 [5] NCCL INFO Channel 00/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-225:124161:124582 [1] NCCL INFO Channel 08/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-225:124162:124587 [2] NCCL INFO Channel 00/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-225:124166:124583 [6] NCCL INFO Channel 08/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-225:124163:124588 [3] NCCL INFO Channel 00/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-225:124164:124585 [4] NCCL INFO Channel 08/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-225:124165:124584 [5] NCCL INFO Channel 08/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-225:124162:124587 [2] NCCL INFO Channel 08/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-225:124163:124588 [3] NCCL INFO Channel 08/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-225:124164:124585 [4] NCCL INFO Channel 05/0 : 4[4] -> 7[7] via P2P/IPC
+ip-26-0-160-225:124164:124585 [4] NCCL INFO Channel 06/0 : 4[4] -> 7[7] via P2P/IPC
+ip-26-0-160-225:124164:124585 [4] NCCL INFO Channel 13/0 : 4[4] -> 7[7] via P2P/IPC
+ip-26-0-160-225:124164:124585 [4] NCCL INFO Channel 14/0 : 4[4] -> 7[7] via P2P/IPC
+ip-26-0-165-202:153562:153983 [0] NCCL INFO Channel 00/0 : 87[7] -> 88[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153562:153983 [0] NCCL INFO Channel 08/0 : 87[7] -> 88[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153562:153983 [0] NCCL INFO Channel 00/0 : 88[0] -> 89[1] via P2P/IPC
+ip-26-0-165-202:153569:153981 [7] NCCL INFO Channel 00/0 : 95[7] -> 96[0] [send] via NET/Libfabric/0(88)/GDRDMA
+ip-26-0-161-103:96888:97306 [0] NCCL INFO Channel 00/0 : 15[7] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96888:97306 [0] NCCL INFO Channel 08/0 : 15[7] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96888:97306 [0] NCCL INFO Channel 00/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-161-103:96895:97307 [7] NCCL INFO Channel 00/0 : 23[7] -> 24[0] [send] via NET/Libfabric/0(16)/GDRDMA
+ip-26-0-165-202:153562:153983 [0] NCCL INFO Channel 08/0 : 88[0] -> 89[1] via P2P/IPC
+ip-26-0-165-202:153569:153981 [7] NCCL INFO Channel 08/0 : 95[7] -> 96[0] [send] via NET/Libfabric/0(88)/GDRDMA
+ip-26-0-161-103:96888:97306 [0] NCCL INFO Channel 08/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-161-103:96895:97307 [7] NCCL INFO Channel 08/0 : 23[7] -> 24[0] [send] via NET/Libfabric/0(16)/GDRDMA
+ip-26-0-161-142:62881:63298 [0] NCCL INFO Channel 00/0 : 39[7] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62881:63298 [0] NCCL INFO Channel 08/0 : 39[7] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272150:272576 [0] NCCL INFO Channel 00/0 : 111[7] -> 112[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62881:63298 [0] NCCL INFO Channel 00/0 : 40[0] -> 41[1] via P2P/IPC
+ip-26-0-168-34:272150:272576 [0] NCCL INFO Channel 08/0 : 111[7] -> 112[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96563:97003 [0] NCCL INFO Channel 00/0 : 31[7] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62888:63302 [7] NCCL INFO Channel 00/0 : 47[7] -> 48[0] [send] via NET/Libfabric/0(40)/GDRDMA
+ip-26-0-168-34:272150:272576 [0] NCCL INFO Channel 00/0 : 112[0] -> 113[1] via P2P/IPC
+ip-26-0-165-131:1630205:1630646 [0] NCCL INFO Channel 00/0 : 71[7] -> 72[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272157:272572 [7] NCCL INFO Channel 00/0 : 119[7] -> 120[0] [send] via NET/Libfabric/0(112)/GDRDMA
+ip-26-0-165-131:1630205:1630646 [0] NCCL INFO Channel 08/0 : 71[7] -> 72[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96563:97003 [0] NCCL INFO Channel 08/0 : 31[7] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630205:1630646 [0] NCCL INFO Channel 00/0 : 72[0] -> 73[1] via P2P/IPC
+ip-26-0-161-138:96563:97003 [0] NCCL INFO Channel 00/0 : 32[0] -> 33[1] via P2P/IPC
+ip-26-0-161-138:96570:97005 [7] NCCL INFO Channel 00/0 : 39[7] -> 40[0] [send] via NET/Libfabric/0(32)/GDRDMA
+ip-26-0-165-131:1630212:1630652 [7] NCCL INFO Channel 00/0 : 79[7] -> 80[0] [send] via NET/Libfabric/0(72)/GDRDMA
+ip-26-0-168-30:235648:236064 [0] NCCL INFO Channel 00/0 : 103[7] -> 104[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62881:63298 [0] NCCL INFO Channel 08/0 : 40[0] -> 41[1] via P2P/IPC
+ip-26-0-168-34:272150:272576 [0] NCCL INFO Channel 08/0 : 112[0] -> 113[1] via P2P/IPC
+ip-26-0-161-142:62888:63302 [7] NCCL INFO Channel 08/0 : 47[7] -> 48[0] [send] via NET/Libfabric/0(40)/GDRDMA
+ip-26-0-168-30:235648:236064 [0] NCCL INFO Channel 08/0 : 103[7] -> 104[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:124160:124581 [0] NCCL INFO Channel 00/0 : 127[7] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272157:272572 [7] NCCL INFO Channel 08/0 : 119[7] -> 120[0] [send] via NET/Libfabric/0(112)/GDRDMA
+ip-26-0-168-30:235648:236064 [0] NCCL INFO Channel 00/0 : 104[0] -> 105[1] via P2P/IPC
+ip-26-0-165-131:1630205:1630646 [0] NCCL INFO Channel 08/0 : 72[0] -> 73[1] via P2P/IPC
+ip-26-0-160-225:124160:124581 [0] NCCL INFO Channel 08/0 : 127[7] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96563:97003 [0] NCCL INFO Channel 08/0 : 32[0] -> 33[1] via P2P/IPC
+ip-26-0-161-78:114512:114930 [0] NCCL INFO Channel 00/0 : 63[7] -> 64[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235655:236069 [7] NCCL INFO Channel 00/0 : 111[7] -> 112[0] [send] via NET/Libfabric/0(104)/GDRDMA
+ip-26-0-160-225:124160:124581 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-225:124167:124586 [7] NCCL INFO Channel 00/0 : 7[7] -> 8[0] [send] via NET/Libfabric/0(0)/GDRDMA
+ip-26-0-165-131:1630212:1630652 [7] NCCL INFO Channel 08/0 : 79[7] -> 80[0] [send] via NET/Libfabric/0(72)/GDRDMA
+ip-26-0-161-138:96570:97005 [7] NCCL INFO Channel 08/0 : 39[7] -> 40[0] [send] via NET/Libfabric/0(32)/GDRDMA
+ip-26-0-161-78:114512:114930 [0] NCCL INFO Channel 08/0 : 63[7] -> 64[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114512:114930 [0] NCCL INFO Channel 00/0 : 64[0] -> 65[1] via P2P/IPC
+ip-26-0-165-164:1227463:1227880 [0] NCCL INFO Channel 00/0 : 79[7] -> 80[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114519:114931 [7] NCCL INFO Channel 00/0 : 71[7] -> 72[0] [send] via NET/Libfabric/0(64)/GDRDMA
+ip-26-0-165-164:1227463:1227880 [0] NCCL INFO Channel 08/0 : 79[7] -> 80[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227470:1227885 [7] NCCL INFO Channel 00/0 : 87[7] -> 88[0] [send] via NET/Libfabric/0(80)/GDRDMA
+ip-26-0-165-164:1227463:1227880 [0] NCCL INFO Channel 00/0 : 80[0] -> 81[1] via P2P/IPC
+ip-26-0-161-142:62881:63298 [0] NCCL INFO Channel 05/0 : 40[0] -> 44[4] via P2P/IPC
+ip-26-0-165-164:1227470:1227885 [7] NCCL INFO Channel 08/0 : 87[7] -> 88[0] [send] via NET/Libfabric/0(80)/GDRDMA
+ip-26-0-160-225:124160:124581 [0] NCCL INFO Channel 08/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-59:1557266:1557689 [0] NCCL INFO Channel 00/0 : 95[7] -> 96[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557266:1557689 [0] NCCL INFO Channel 08/0 : 95[7] -> 96[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:124167:124586 [7] NCCL INFO Channel 08/0 : 7[7] -> 8[0] [send] via NET/Libfabric/0(0)/GDRDMA
+ip-26-0-161-78:114512:114930 [0] NCCL INFO Channel 08/0 : 64[0] -> 65[1] via P2P/IPC
+ip-26-0-161-142:62881:63298 [0] NCCL INFO Channel 13/0 : 40[0] -> 44[4] via P2P/IPC
+ip-26-0-165-59:1557266:1557689 [0] NCCL INFO Channel 00/0 : 96[0] -> 97[1] via P2P/IPC
+ip-26-0-161-78:114519:114931 [7] NCCL INFO Channel 08/0 : 71[7] -> 72[0] [send] via NET/Libfabric/0(64)/GDRDMA
+ip-26-0-165-59:1557273:1557685 [7] NCCL INFO Channel 00/0 : 103[7] -> 104[0] [send] via NET/Libfabric/0(96)/GDRDMA
+ip-26-0-165-202:153562:153983 [0] NCCL INFO Channel 05/0 : 88[0] -> 92[4] via P2P/IPC
+ip-26-0-161-138:96565:97004 [2] NCCL INFO Channel 02/0 : 27[3] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235648:236064 [0] NCCL INFO Channel 08/0 : 104[0] -> 105[1] via P2P/IPC
+ip-26-0-161-138:96565:97004 [2] NCCL INFO Channel 10/0 : 27[3] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235655:236069 [7] NCCL INFO Channel 08/0 : 111[7] -> 112[0] [send] via NET/Libfabric/0(104)/GDRDMA
+ip-26-0-161-123:32704:33123 [0] NCCL INFO Channel 00/0 : 23[7] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630205:1630646 [0] NCCL INFO Channel 05/0 : 72[0] -> 76[4] via P2P/IPC
+ip-26-0-165-202:153562:153983 [0] NCCL INFO Channel 13/0 : 88[0] -> 92[4] via P2P/IPC
+ip-26-0-165-164:1227463:1227880 [0] NCCL INFO Channel 08/0 : 80[0] -> 81[1] via P2P/IPC
+ip-26-0-165-59:1557266:1557689 [0] NCCL INFO Channel 08/0 : 96[0] -> 97[1] via P2P/IPC
+ip-26-0-161-123:32711:33127 [7] NCCL INFO Channel 00/0 : 31[7] -> 32[0] [send] via NET/Libfabric/0(24)/GDRDMA
+ip-26-0-161-123:32704:33123 [0] NCCL INFO Channel 08/0 : 23[7] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557273:1557685 [7] NCCL INFO Channel 08/0 : 103[7] -> 104[0] [send] via NET/Libfabric/0(96)/GDRDMA
+ip-26-0-161-123:32704:33123 [0] NCCL INFO Channel 00/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-165-131:1630205:1630646 [0] NCCL INFO Channel 13/0 : 72[0] -> 76[4] via P2P/IPC
+ip-26-0-161-123:32711:33127 [7] NCCL INFO Channel 08/0 : 31[7] -> 32[0] [send] via NET/Libfabric/0(24)/GDRDMA
+ip-26-0-168-30:235648:236064 [0] NCCL INFO Channel 05/0 : 104[0] -> 108[4] via P2P/IPC
+ip-26-0-161-103:96890:97313 [2] NCCL INFO Channel 02/0 : 11[3] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96563:97003 [0] NCCL INFO Channel 05/0 : 32[0] -> 36[4] via P2P/IPC
+ip-26-0-161-103:96890:97313 [2] NCCL INFO Channel 10/0 : 11[3] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227463:1227880 [0] NCCL INFO Channel 05/0 : 80[0] -> 84[4] via P2P/IPC
+ip-26-0-161-123:32704:33123 [0] NCCL INFO Channel 08/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-168-30:235648:236064 [0] NCCL INFO Channel 13/0 : 104[0] -> 108[4] via P2P/IPC
+ip-26-0-165-164:1227466:1227887 [3] NCCL INFO Channel 03/0 : 76[4] -> 83[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96563:97003 [0] NCCL INFO Channel 13/0 : 32[0] -> 36[4] via P2P/IPC
+ip-26-0-165-164:1227463:1227880 [0] NCCL INFO Channel 13/0 : 80[0] -> 84[4] via P2P/IPC
+ip-26-0-165-164:1227466:1227887 [3] NCCL INFO Channel 11/0 : 76[4] -> 83[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815963:2816400 [0] NCCL INFO Channel 00/0 : 55[7] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-52:2722045:2722470 [0] NCCL INFO Channel 00/0 : 119[7] -> 120[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815963:2816400 [0] NCCL INFO Channel 08/0 : 55[7] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-52:2722045:2722470 [0] NCCL INFO Channel 08/0 : 119[7] -> 120[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-52:2722045:2722470 [0] NCCL INFO Channel 00/0 : 120[0] -> 121[1] via P2P/IPC
+ip-26-0-161-178:2815963:2816400 [0] NCCL INFO Channel 00/0 : 56[0] -> 57[1] via P2P/IPC
+ip-26-0-161-103:96891:97311 [3] NCCL INFO Channel 03/0 : 12[4] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815970:2816397 [7] NCCL INFO Channel 00/0 : 63[7] -> 64[0] [send] via NET/Libfabric/0(56)/GDRDMA
+ip-26-0-168-52:2722052:2722464 [7] NCCL INFO Channel 00/0 : 127[7] -> 0[0] [send] via NET/Libfabric/0(120)/GDRDMA
+ip-26-0-161-103:96891:97311 [3] NCCL INFO Channel 11/0 : 12[4] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227465:1227881 [2] NCCL INFO Channel 02/0 : 75[3] -> 82[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153565:153985 [3] NCCL INFO Channel 03/0 : 84[4] -> 91[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32704:33123 [0] NCCL INFO Channel 05/0 : 24[0] -> 28[4] via P2P/IPC
+ip-26-0-165-164:1227465:1227881 [2] NCCL INFO Channel 10/0 : 75[3] -> 82[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227466:1227887 [3] NCCL INFO Channel 02/0 : 83[3] -> 90[2] [send] via NET/Libfabric/2(82)/GDRDMA
+ip-26-0-165-202:153565:153985 [3] NCCL INFO Channel 11/0 : 84[4] -> 91[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96891:97311 [3] NCCL INFO Channel 02/0 : 19[3] -> 26[2] [send] via NET/Libfabric/2(18)/GDRDMA
+ip-26-0-165-164:1227466:1227887 [3] NCCL INFO Channel 10/0 : 83[3] -> 90[2] [send] via NET/Libfabric/2(82)/GDRDMA
+ip-26-0-161-178:2815963:2816400 [0] NCCL INFO Channel 08/0 : 56[0] -> 57[1] via P2P/IPC
+ip-26-0-161-103:96891:97311 [3] NCCL INFO Channel 10/0 : 19[3] -> 26[2] [send] via NET/Libfabric/2(18)/GDRDMA
+ip-26-0-168-52:2722045:2722470 [0] NCCL INFO Channel 08/0 : 120[0] -> 121[1] via P2P/IPC
+ip-26-0-161-178:2815970:2816397 [7] NCCL INFO Channel 08/0 : 63[7] -> 64[0] [send] via NET/Libfabric/0(56)/GDRDMA
+ip-26-0-161-123:32704:33123 [0] NCCL INFO Channel 13/0 : 24[0] -> 28[4] via P2P/IPC
+ip-26-0-165-202:153564:153984 [2] NCCL INFO Channel 02/0 : 83[3] -> 90[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62883:63301 [2] NCCL INFO Channel 02/0 : 35[3] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153564:153984 [2] NCCL INFO Channel 10/0 : 83[3] -> 90[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364904:1365335 [2] NCCL INFO Channel 02/0 : 3[3] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227463:1227880 [0] NCCL INFO Channel 06/0 : 80[0] -> 85[5] via P2P/IPC
+ip-26-0-165-202:153565:153985 [3] NCCL INFO Channel 02/0 : 91[3] -> 98[2] [send] via NET/Libfabric/2(90)/GDRDMA
+ip-26-0-161-142:62883:63301 [2] NCCL INFO Channel 10/0 : 35[3] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-52:2722052:2722464 [7] NCCL INFO Channel 08/0 : 127[7] -> 0[0] [send] via NET/Libfabric/0(120)/GDRDMA
+ip-26-0-160-242:1364904:1365335 [2] NCCL INFO Channel 10/0 : 3[3] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153565:153985 [3] NCCL INFO Channel 10/0 : 91[3] -> 98[2] [send] via NET/Libfabric/2(90)/GDRDMA
+ip-26-0-161-138:96566:97010 [3] NCCL INFO Channel 03/0 : 28[4] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364902:1365333 [0] NCCL INFO Channel 00/0 : 7[7] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114512:114930 [0] NCCL INFO Channel 05/0 : 64[0] -> 68[4] via P2P/IPC
+ip-26-0-161-138:96563:97003 [0] NCCL INFO Channel 06/0 : 32[0] -> 37[5] via P2P/IPC
+ip-26-0-160-242:1364909:1365327 [7] NCCL INFO Channel 00/0 : 15[7] -> 16[0] [send] via NET/Libfabric/0(8)/GDRDMA
+ip-26-0-160-242:1364902:1365333 [0] NCCL INFO Channel 08/0 : 7[7] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364902:1365333 [0] NCCL INFO Channel 00/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-168-30:235651:236068 [3] NCCL INFO Channel 03/0 : 100[4] -> 107[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227463:1227880 [0] NCCL INFO Channel 14/0 : 80[0] -> 85[5] via P2P/IPC
+ip-26-0-168-52:2722045:2722470 [0] NCCL INFO Channel 05/0 : 120[0] -> 124[4] via P2P/IPC
+ip-26-0-160-242:1364909:1365327 [7] NCCL INFO Channel 08/0 : 15[7] -> 16[0] [send] via NET/Libfabric/0(8)/GDRDMA
+ip-26-0-160-225:124160:124581 [0] NCCL INFO Channel 05/0 : 0[0] -> 4[4] via P2P/IPC
+ip-26-0-161-138:96566:97010 [3] NCCL INFO Channel 11/0 : 28[4] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557266:1557689 [0] NCCL INFO Channel 05/0 : 96[0] -> 100[4] via P2P/IPC
+ip-26-0-161-138:96563:97003 [0] NCCL INFO Channel 14/0 : 32[0] -> 37[5] via P2P/IPC
+ip-26-0-168-30:235651:236068 [3] NCCL INFO Channel 11/0 : 100[4] -> 107[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364905:1365328 [3] NCCL INFO Channel 03/0 : 4[4] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114512:114930 [0] NCCL INFO Channel 13/0 : 64[0] -> 68[4] via P2P/IPC
+ip-26-0-160-242:1364905:1365328 [3] NCCL INFO Channel 11/0 : 4[4] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:124160:124581 [0] NCCL INFO Channel 13/0 : 0[0] -> 4[4] via P2P/IPC
+ip-26-0-161-138:96566:97010 [3] NCCL INFO Channel 02/0 : 35[3] -> 42[2] [send] via NET/Libfabric/2(34)/GDRDMA
+ip-26-0-168-34:272150:272576 [0] NCCL INFO Channel 05/0 : 112[0] -> 116[4] via P2P/IPC
+ip-26-0-161-138:96566:97010 [3] NCCL INFO Channel 10/0 : 35[3] -> 42[2] [send] via NET/Libfabric/2(34)/GDRDMA
+ip-26-0-161-103:96888:97306 [0] NCCL INFO Channel 05/0 : 16[0] -> 20[4] via P2P/IPC
+ip-26-0-160-242:1364905:1365328 [3] NCCL INFO Channel 02/0 : 11[3] -> 18[2] [send] via NET/Libfabric/2(10)/GDRDMA
+ip-26-0-161-138:96563:97003 [0] NCCL INFO Channel 07/0 : 32[0] -> 38[6] via P2P/IPC
+ip-26-0-165-59:1557268:1557686 [2] NCCL INFO Channel 02/0 : 91[3] -> 98[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227463:1227880 [0] NCCL INFO Channel 07/0 : 80[0] -> 86[6] via P2P/IPC
+ip-26-0-165-131:1630207:1630648 [2] NCCL INFO Channel 02/0 : 67[3] -> 74[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364905:1365328 [3] NCCL INFO Channel 10/0 : 11[3] -> 18[2] [send] via NET/Libfabric/2(10)/GDRDMA
+ip-26-0-165-59:1557266:1557689 [0] NCCL INFO Channel 13/0 : 96[0] -> 100[4] via P2P/IPC
+ip-26-0-160-242:1364902:1365333 [0] NCCL INFO Channel 08/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-165-59:1557268:1557686 [2] NCCL INFO Channel 10/0 : 91[3] -> 98[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630207:1630648 [2] NCCL INFO Channel 10/0 : 67[3] -> 74[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272150:272576 [0] NCCL INFO Channel 13/0 : 112[0] -> 116[4] via P2P/IPC
+ip-26-0-161-138:96563:97003 [0] NCCL INFO Channel 15/0 : 32[0] -> 38[6] via P2P/IPC
+ip-26-0-168-52:2722045:2722470 [0] NCCL INFO Channel 13/0 : 120[0] -> 124[4] via P2P/IPC
+ip-26-0-161-103:96888:97306 [0] NCCL INFO Channel 13/0 : 16[0] -> 20[4] via P2P/IPC
+ip-26-0-165-131:1630207:1630648 [2] NCCL INFO Channel 01/0 : 74[2] -> 81[1] [send] via NET/Libfabric/1(73)/GDRDMA
+ip-26-0-161-142:62884:63299 [3] NCCL INFO Channel 03/0 : 36[4] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96563:97003 [0] NCCL INFO Channel 01/0 : 32[0] -> 39[7] via P2P/IPC
+ip-26-0-165-131:1630208:1630650 [3] NCCL INFO Channel 03/0 : 68[4] -> 75[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62884:63299 [3] NCCL INFO Channel 11/0 : 36[4] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630208:1630650 [3] NCCL INFO Channel 11/0 : 68[4] -> 75[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96563:97003 [0] NCCL INFO Channel 02/0 : 32[0] -> 39[7] via P2P/IPC
+ip-26-0-168-30:235650:236065 [2] NCCL INFO Channel 02/0 : 99[3] -> 106[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364902:1365333 [0] NCCL INFO Channel 05/0 : 8[0] -> 12[4] via P2P/IPC
+ip-26-0-161-142:62884:63299 [3] NCCL INFO Channel 02/0 : 43[3] -> 50[2] [send] via NET/Libfabric/2(42)/GDRDMA
+ip-26-0-165-131:1630205:1630646 [0] NCCL INFO Channel 06/0 : 72[0] -> 77[5] via P2P/IPC
+ip-26-0-168-30:235650:236065 [2] NCCL INFO Channel 10/0 : 99[3] -> 106[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96888:97306 [0] NCCL INFO Channel 06/0 : 16[0] -> 21[5] via P2P/IPC
+ip-26-0-161-142:62884:63299 [3] NCCL INFO Channel 10/0 : 43[3] -> 50[2] [send] via NET/Libfabric/2(42)/GDRDMA
+ip-26-0-165-131:1630208:1630650 [3] NCCL INFO Channel 02/0 : 75[3] -> 82[2] [send] via NET/Libfabric/2(74)/GDRDMA
+ip-26-0-168-30:235648:236064 [0] NCCL INFO Channel 06/0 : 104[0] -> 109[5] via P2P/IPC
+ip-26-0-165-164:1227463:1227880 [0] NCCL INFO Channel 15/0 : 80[0] -> 86[6] via P2P/IPC
+ip-26-0-168-30:235651:236068 [3] NCCL INFO Channel 02/0 : 107[3] -> 114[2] [send] via NET/Libfabric/2(106)/GDRDMA
+ip-26-0-165-131:1630208:1630650 [3] NCCL INFO Channel 10/0 : 75[3] -> 82[2] [send] via NET/Libfabric/2(74)/GDRDMA
+ip-26-0-168-30:235651:236068 [3] NCCL INFO Channel 10/0 : 107[3] -> 114[2] [send] via NET/Libfabric/2(106)/GDRDMA
+ip-26-0-168-34:272153:272573 [3] NCCL INFO Channel 03/0 : 108[4] -> 115[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272153:272573 [3] NCCL INFO Channel 11/0 : 108[4] -> 115[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815966:2816396 [3] NCCL INFO Channel 03/0 : 52[4] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114515:114936 [3] NCCL INFO Channel 03/0 : 60[4] -> 67[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96888:97306 [0] NCCL INFO Channel 14/0 : 16[0] -> 21[5] via P2P/IPC
+ip-26-0-160-242:1364902:1365333 [0] NCCL INFO Channel 13/0 : 8[0] -> 12[4] via P2P/IPC
+ip-26-0-165-131:1630205:1630646 [0] NCCL INFO Channel 14/0 : 72[0] -> 77[5] via P2P/IPC
+ip-26-0-168-30:235648:236064 [0] NCCL INFO Channel 14/0 : 104[0] -> 109[5] via P2P/IPC
+ip-26-0-161-178:2815966:2816396 [3] NCCL INFO Channel 11/0 : 52[4] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114515:114936 [3] NCCL INFO Channel 11/0 : 60[4] -> 67[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364904:1365335 [2] NCCL INFO Channel 01/0 : 10[2] -> 17[1] [send] via NET/Libfabric/1(9)/GDRDMA
+ip-26-0-161-178:2815965:2816395 [2] NCCL INFO Channel 02/0 : 51[3] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272152:272575 [2] NCCL INFO Channel 02/0 : 107[3] -> 114[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227463:1227880 [0] NCCL INFO Channel 01/0 : 80[0] -> 87[7] via P2P/IPC
+ip-26-0-161-123:32704:33123 [0] NCCL INFO Channel 06/0 : 24[0] -> 29[5] via P2P/IPC
+ip-26-0-161-178:2815965:2816395 [2] NCCL INFO Channel 10/0 : 51[3] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272153:272573 [3] NCCL INFO Channel 02/0 : 115[3] -> 122[2] [send] via NET/Libfabric/2(114)/GDRDMA
+ip-26-0-161-178:2815966:2816396 [3] NCCL INFO Channel 02/0 : 59[3] -> 66[2] [send] via NET/Libfabric/2(58)/GDRDMA
+ip-26-0-165-131:1630205:1630646 [0] NCCL INFO Channel 07/0 : 72[0] -> 78[6] via P2P/IPC
+ip-26-0-165-59:1557269:1557684 [3] NCCL INFO Channel 03/0 : 92[4] -> 99[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96888:97306 [0] NCCL INFO Channel 07/0 : 16[0] -> 22[6] via P2P/IPC
+ip-26-0-168-34:272152:272575 [2] NCCL INFO Channel 10/0 : 107[3] -> 114[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153562:153983 [0] NCCL INFO Channel 06/0 : 88[0] -> 93[5] via P2P/IPC
+ip-26-0-161-178:2815966:2816396 [3] NCCL INFO Channel 10/0 : 59[3] -> 66[2] [send] via NET/Libfabric/2(58)/GDRDMA
+ip-26-0-165-164:1227463:1227880 [0] NCCL INFO Channel 02/0 : 80[0] -> 87[7] via P2P/IPC
+ip-26-0-168-34:272153:272573 [3] NCCL INFO Channel 10/0 : 115[3] -> 122[2] [send] via NET/Libfabric/2(114)/GDRDMA
+ip-26-0-165-59:1557269:1557684 [3] NCCL INFO Channel 11/0 : 92[4] -> 99[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32704:33123 [0] NCCL INFO Channel 14/0 : 24[0] -> 29[5] via P2P/IPC
+ip-26-0-168-30:235648:236064 [0] NCCL INFO Channel 07/0 : 104[0] -> 110[6] via P2P/IPC
+ip-26-0-165-131:1630205:1630646 [0] NCCL INFO Channel 15/0 : 72[0] -> 78[6] via P2P/IPC
+ip-26-0-165-164:1227463:1227880 [0] NCCL INFO Channel 03/0 : 80[0] -> 87[7] via P2P/IPC
+ip-26-0-161-103:96888:97306 [0] NCCL INFO Channel 15/0 : 16[0] -> 22[6] via P2P/IPC
+ip-26-0-165-59:1557269:1557684 [3] NCCL INFO Channel 02/0 : 99[3] -> 106[2] [send] via NET/Libfabric/2(98)/GDRDMA
+ip-26-0-168-34:272150:272576 [0] NCCL INFO Channel 06/0 : 112[0] -> 117[5] via P2P/IPC
+ip-26-0-161-138:96570:97005 [7] NCCL INFO Channel 06/0 : 39[7] -> 46[6] [send] via NET/Libfabric/6(38)/GDRDMA
+ip-26-0-165-202:153562:153983 [0] NCCL INFO Channel 14/0 : 88[0] -> 93[5] via P2P/IPC
+ip-26-0-165-59:1557269:1557684 [3] NCCL INFO Channel 10/0 : 99[3] -> 106[2] [send] via NET/Libfabric/2(98)/GDRDMA
+ip-26-0-161-78:114514:114932 [2] NCCL INFO Channel 02/0 : 59[3] -> 66[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-52:2722047:2722468 [2] NCCL INFO Channel 02/0 : 115[3] -> 122[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235648:236064 [0] NCCL INFO Channel 15/0 : 104[0] -> 110[6] via P2P/IPC
+ip-26-0-161-138:96563:97003 [0] NCCL INFO Channel 03/0 : 32[0] -> 39[7] via P2P/IPC
+ip-26-0-161-78:114514:114932 [2] NCCL INFO Channel 10/0 : 59[3] -> 66[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-52:2722047:2722468 [2] NCCL INFO Channel 10/0 : 115[3] -> 122[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227463:1227880 [0] NCCL INFO Channel 04/0 : 80[0] -> 87[7] via P2P/IPC
+ip-26-0-161-123:32704:33123 [0] NCCL INFO Channel 07/0 : 24[0] -> 30[6] via P2P/IPC
+ip-26-0-161-78:114515:114936 [3] NCCL INFO Channel 02/0 : 67[3] -> 74[2] [send] via NET/Libfabric/2(66)/GDRDMA
+ip-26-0-160-242:1364902:1365333 [0] NCCL INFO Channel 06/0 : 8[0] -> 13[5] via P2P/IPC
+ip-26-0-165-131:1630205:1630646 [0] NCCL INFO Channel 01/0 : 72[0] -> 79[7] via P2P/IPC
+ip-26-0-165-59:1557268:1557686 [2] NCCL INFO Channel 01/0 : 98[2] -> 105[1] [send] via NET/Libfabric/1(97)/GDRDMA
+ip-26-0-161-78:114515:114936 [3] NCCL INFO Channel 10/0 : 67[3] -> 74[2] [send] via NET/Libfabric/2(66)/GDRDMA
+ip-26-0-168-52:2722045:2722470 [0] NCCL INFO Channel 06/0 : 120[0] -> 125[5] via P2P/IPC
+ip-26-0-161-138:96563:97003 [0] NCCL INFO Channel 04/0 : 32[0] -> 39[7] via P2P/IPC
+ip-26-0-165-59:1557268:1557686 [2] NCCL INFO Channel 09/0 : 98[2] -> 105[1] [send] via NET/Libfabric/1(97)/GDRDMA
+ip-26-0-165-164:1227463:1227880 [0] NCCL INFO Channel 09/0 : 80[0] -> 87[7] via P2P/IPC
+ip-26-0-168-52:2722047:2722468 [2] NCCL INFO Channel 01/0 : 122[2] -> 1[1] [send] via NET/Libfabric/1(121)/GDRDMA
+ip-26-0-161-153:2636084:2636511 [0] NCCL INFO Channel 00/0 : 47[7] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96888:97306 [0] NCCL INFO Channel 01/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-161-78:114514:114932 [2] NCCL INFO Channel 01/0 : 66[2] -> 73[1] [send] via NET/Libfabric/1(65)/GDRDMA
+ip-26-0-161-123:32704:33123 [0] NCCL INFO Channel 15/0 : 24[0] -> 30[6] via P2P/IPC
+ip-26-0-168-52:2722047:2722468 [2] NCCL INFO Channel 09/0 : 122[2] -> 1[1] [send] via NET/Libfabric/1(121)/GDRDMA
+ip-26-0-165-202:153562:153983 [0] NCCL INFO Channel 07/0 : 88[0] -> 94[6] via P2P/IPC
+ip-26-0-168-30:235648:236064 [0] NCCL INFO Channel 01/0 : 104[0] -> 111[7] via P2P/IPC
+ip-26-0-161-153:2636091:2636512 [7] NCCL INFO Channel 00/0 : 55[7] -> 56[0] [send] via NET/Libfabric/0(48)/GDRDMA
+ip-26-0-165-131:1630205:1630646 [0] NCCL INFO Channel 02/0 : 72[0] -> 79[7] via P2P/IPC
+ip-26-0-161-78:114514:114932 [2] NCCL INFO Channel 09/0 : 66[2] -> 73[1] [send] via NET/Libfabric/1(65)/GDRDMA
+ip-26-0-161-138:96563:97003 [0] NCCL INFO Channel 09/0 : 32[0] -> 39[7] via P2P/IPC
+ip-26-0-161-153:2636084:2636511 [0] NCCL INFO Channel 08/0 : 47[7] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636084:2636511 [0] NCCL INFO Channel 00/0 : 48[0] -> 49[1] via P2P/IPC
+ip-26-0-160-242:1364902:1365333 [0] NCCL INFO Channel 14/0 : 8[0] -> 13[5] via P2P/IPC
+ip-26-0-161-123:32706:33122 [2] NCCL INFO Channel 02/0 : 19[3] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636091:2636512 [7] NCCL INFO Channel 08/0 : 55[7] -> 56[0] [send] via NET/Libfabric/0(48)/GDRDMA
+ip-26-0-161-103:96888:97306 [0] NCCL INFO Channel 02/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-165-164:1227463:1227880 [0] NCCL INFO Channel 10/0 : 80[0] -> 87[7] via P2P/IPC
+ip-26-0-168-52:2722045:2722470 [0] NCCL INFO Channel 14/0 : 120[0] -> 125[5] via P2P/IPC
+ip-26-0-161-123:32706:33122 [2] NCCL INFO Channel 10/0 : 19[3] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235648:236064 [0] NCCL INFO Channel 02/0 : 104[0] -> 111[7] via P2P/IPC
+ip-26-0-165-131:1630205:1630646 [0] NCCL INFO Channel 03/0 : 72[0] -> 79[7] via P2P/IPC
+ip-26-0-160-225:124160:124581 [0] NCCL INFO Channel 06/0 : 0[0] -> 5[5] via P2P/IPC
+ip-26-0-165-202:153562:153983 [0] NCCL INFO Channel 15/0 : 88[0] -> 94[6] via P2P/IPC
+ip-26-0-161-78:114519:114931 [7] NCCL INFO Channel 06/0 : 71[7] -> 78[6] [send] via NET/Libfabric/6(70)/GDRDMA
+ip-26-0-161-153:2636087:2636506 [3] NCCL INFO Channel 03/0 : 44[4] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32704:33123 [0] NCCL INFO Channel 01/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-161-138:96563:97003 [0] NCCL INFO Channel 10/0 : 32[0] -> 39[7] via P2P/IPC
+ip-26-0-161-103:96888:97306 [0] NCCL INFO Channel 03/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-165-164:1227463:1227880 [0] NCCL INFO Channel 11/0 : 80[0] -> 87[7] via P2P/IPC
+ip-26-0-161-78:114519:114931 [7] NCCL INFO Channel 14/0 : 71[7] -> 78[6] [send] via NET/Libfabric/6(70)/GDRDMA
+ip-26-0-161-153:2636087:2636506 [3] NCCL INFO Channel 11/0 : 44[4] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557273:1557685 [7] NCCL INFO Channel 06/0 : 103[7] -> 110[6] [send] via NET/Libfabric/6(102)/GDRDMA
+ip-26-0-168-30:235648:236064 [0] NCCL INFO Channel 03/0 : 104[0] -> 111[7] via P2P/IPC
+ip-26-0-161-178:2815963:2816400 [0] NCCL INFO Channel 05/0 : 56[0] -> 60[4] via P2P/IPC
+ip-26-0-165-59:1557273:1557685 [7] NCCL INFO Channel 14/0 : 103[7] -> 110[6] [send] via NET/Libfabric/6(102)/GDRDMA
+ip-26-0-165-131:1630212:1630652 [7] NCCL INFO Channel 06/0 : 79[7] -> 86[6] [send] via NET/Libfabric/6(78)/GDRDMA
+ip-26-0-165-131:1630205:1630646 [0] NCCL INFO Channel 04/0 : 72[0] -> 79[7] via P2P/IPC
+ip-26-0-160-242:1364902:1365333 [0] NCCL INFO Channel 07/0 : 8[0] -> 14[6] via P2P/IPC
+ip-26-0-161-123:32704:33123 [0] NCCL INFO Channel 02/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-161-103:96888:97306 [0] NCCL INFO Channel 04/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-165-202:153562:153983 [0] NCCL INFO Channel 01/0 : 88[0] -> 95[7] via P2P/IPC
+ip-26-0-165-164:1227463:1227880 [0] NCCL INFO Channel 12/0 : 80[0] -> 87[7] via P2P/IPC
+ip-26-0-168-52:2722045:2722470 [0] NCCL INFO Channel 07/0 : 120[0] -> 126[6] via P2P/IPC
+ip-26-0-161-123:32711:33127 [7] NCCL INFO Channel 06/0 : 31[7] -> 38[6] [send] via NET/Libfabric/6(30)/GDRDMA
+ip-26-0-168-30:235648:236064 [0] NCCL INFO Channel 04/0 : 104[0] -> 111[7] via P2P/IPC
+ip-26-0-160-225:124160:124581 [0] NCCL INFO Channel 14/0 : 0[0] -> 5[5] via P2P/IPC
+ip-26-0-165-131:1630205:1630646 [0] NCCL INFO Channel 09/0 : 72[0] -> 79[7] via P2P/IPC
+ip-26-0-161-178:2815963:2816400 [0] NCCL INFO Channel 13/0 : 56[0] -> 60[4] via P2P/IPC
+ip-26-0-161-123:32704:33123 [0] NCCL INFO Channel 03/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-161-103:96888:97306 [0] NCCL INFO Channel 09/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-160-242:1364902:1365333 [0] NCCL INFO Channel 15/0 : 8[0] -> 14[6] via P2P/IPC
+ip-26-0-165-202:153562:153983 [0] NCCL INFO Channel 02/0 : 88[0] -> 95[7] via P2P/IPC
+ip-26-0-165-164:1227470:1227885 [7] NCCL INFO Channel 06/0 : 87[7] -> 94[6] [send] via NET/Libfabric/6(86)/GDRDMA
+ip-26-0-168-52:2722045:2722470 [0] NCCL INFO Channel 15/0 : 120[0] -> 126[6] via P2P/IPC
+ip-26-0-165-164:1227470:1227885 [7] NCCL INFO Channel 14/0 : 87[7] -> 94[6] [send] via NET/Libfabric/6(86)/GDRDMA
+ip-26-0-160-225:124163:124588 [3] NCCL INFO Channel 03/0 : 124[4] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235648:236064 [0] NCCL INFO Channel 09/0 : 104[0] -> 111[7] via P2P/IPC
+ip-26-0-161-123:32704:33123 [0] NCCL INFO Channel 04/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-165-131:1630205:1630646 [0] NCCL INFO Channel 10/0 : 72[0] -> 79[7] via P2P/IPC
+ip-26-0-161-153:2636084:2636511 [0] NCCL INFO Channel 08/0 : 48[0] -> 49[1] via P2P/IPC
+ip-26-0-161-153:2636086:2636507 [2] NCCL INFO Channel 02/0 : 43[3] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96888:97306 [0] NCCL INFO Channel 10/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-165-202:153562:153983 [0] NCCL INFO Channel 03/0 : 88[0] -> 95[7] via P2P/IPC
+ip-26-0-160-225:124163:124588 [3] NCCL INFO Channel 11/0 : 124[4] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235655:236069 [7] NCCL INFO Channel 06/0 : 111[7] -> 118[6] [send] via NET/Libfabric/6(110)/GDRDMA
+ip-26-0-168-34:272150:272576 [0] NCCL INFO Channel 14/0 : 112[0] -> 117[5] via P2P/IPC
+ip-26-0-161-178:2815963:2816400 [0] NCCL INFO Channel 06/0 : 56[0] -> 61[5] via P2P/IPC
+ip-26-0-160-225:124160:124581 [0] NCCL INFO Channel 07/0 : 0[0] -> 6[6] via P2P/IPC
+ip-26-0-161-153:2636086:2636507 [2] NCCL INFO Channel 10/0 : 43[3] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364902:1365333 [0] NCCL INFO Channel 01/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-161-123:32704:33123 [0] NCCL INFO Channel 09/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-165-131:1630205:1630646 [0] NCCL INFO Channel 11/0 : 72[0] -> 79[7] via P2P/IPC
+ip-26-0-168-30:235648:236064 [0] NCCL INFO Channel 10/0 : 104[0] -> 111[7] via P2P/IPC
+ip-26-0-161-103:96888:97306 [0] NCCL INFO Channel 11/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-161-153:2636087:2636506 [3] NCCL INFO Channel 02/0 : 51[3] -> 58[2] [send] via NET/Libfabric/2(50)/GDRDMA
+ip-26-0-165-202:153562:153983 [0] NCCL INFO Channel 04/0 : 88[0] -> 95[7] via P2P/IPC
+ip-26-0-168-52:2722045:2722470 [0] NCCL INFO Channel 01/0 : 120[0] -> 127[7] via P2P/IPC
+ip-26-0-161-178:2815963:2816400 [0] NCCL INFO Channel 14/0 : 56[0] -> 61[5] via P2P/IPC
+ip-26-0-161-153:2636087:2636506 [3] NCCL INFO Channel 10/0 : 51[3] -> 58[2] [send] via NET/Libfabric/2(50)/GDRDMA
+ip-26-0-160-242:1364902:1365333 [0] NCCL INFO Channel 02/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-161-123:32704:33123 [0] NCCL INFO Channel 10/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-165-131:1630205:1630646 [0] NCCL INFO Channel 12/0 : 72[0] -> 79[7] via P2P/IPC
+ip-26-0-165-202:153562:153983 [0] NCCL INFO Channel 09/0 : 88[0] -> 95[7] via P2P/IPC
+ip-26-0-168-30:235648:236064 [0] NCCL INFO Channel 11/0 : 104[0] -> 111[7] via P2P/IPC
+ip-26-0-168-34:272150:272576 [0] NCCL INFO Channel 07/0 : 112[0] -> 118[6] via P2P/IPC
+ip-26-0-161-103:96888:97306 [0] NCCL INFO Channel 12/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-168-52:2722045:2722470 [0] NCCL INFO Channel 02/0 : 120[0] -> 127[7] via P2P/IPC
+ip-26-0-160-225:124160:124581 [0] NCCL INFO Channel 15/0 : 0[0] -> 6[6] via P2P/IPC
+ip-26-0-168-52:2722048:2722469 [3] NCCL INFO Channel 03/0 : 116[4] -> 123[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32704:33123 [0] NCCL INFO Channel 11/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-165-202:153562:153983 [0] NCCL INFO Channel 10/0 : 88[0] -> 95[7] via P2P/IPC
+ip-26-0-160-242:1364902:1365333 [0] NCCL INFO Channel 03/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-168-30:235648:236064 [0] NCCL INFO Channel 12/0 : 104[0] -> 111[7] via P2P/IPC
+ip-26-0-161-178:2815963:2816400 [0] NCCL INFO Channel 07/0 : 56[0] -> 62[6] via P2P/IPC
+ip-26-0-168-52:2722048:2722469 [3] NCCL INFO Channel 11/0 : 116[4] -> 123[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272150:272576 [0] NCCL INFO Channel 15/0 : 112[0] -> 118[6] via P2P/IPC
+ip-26-0-168-52:2722045:2722470 [0] NCCL INFO Channel 03/0 : 120[0] -> 127[7] via P2P/IPC
+ip-26-0-161-138:96563:97003 [0] NCCL INFO Channel 11/0 : 32[0] -> 39[7] via P2P/IPC
+ip-26-0-161-178:2815970:2816397 [7] NCCL INFO Channel 06/0 : 63[7] -> 70[6] [send] via NET/Libfabric/6(62)/GDRDMA
+ip-26-0-161-103:96895:97307 [7] NCCL INFO Channel 06/0 : 23[7] -> 30[6] [send] via NET/Libfabric/6(22)/GDRDMA
+ip-26-0-161-123:32704:33123 [0] NCCL INFO Channel 12/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-160-242:1364902:1365333 [0] NCCL INFO Channel 04/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-161-178:2815970:2816397 [7] NCCL INFO Channel 14/0 : 63[7] -> 70[6] [send] via NET/Libfabric/6(62)/GDRDMA
+ip-26-0-161-178:2815963:2816400 [0] NCCL INFO Channel 15/0 : 56[0] -> 62[6] via P2P/IPC
+ip-26-0-161-153:2636084:2636511 [0] NCCL INFO Channel 05/0 : 48[0] -> 52[4] via P2P/IPC
+ip-26-0-161-103:96895:97307 [7] NCCL INFO Channel 14/0 : 23[7] -> 30[6] [send] via NET/Libfabric/6(22)/GDRDMA
+ip-26-0-168-52:2722048:2722469 [3] NCCL INFO Channel 02/0 : 123[3] -> 2[2] [send] via NET/Libfabric/2(122)/GDRDMA
+ip-26-0-165-59:1557266:1557689 [0] NCCL INFO Channel 06/0 : 96[0] -> 101[5] via P2P/IPC
+ip-26-0-160-225:124160:124581 [0] NCCL INFO Channel 01/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-161-138:96563:97003 [0] NCCL INFO Channel 12/0 : 32[0] -> 39[7] via P2P/IPC
+ip-26-0-168-52:2722045:2722470 [0] NCCL INFO Channel 04/0 : 120[0] -> 127[7] via P2P/IPC
+ip-26-0-168-52:2722048:2722469 [3] NCCL INFO Channel 10/0 : 123[3] -> 2[2] [send] via NET/Libfabric/2(122)/GDRDMA
+ip-26-0-160-242:1364902:1365333 [0] NCCL INFO Channel 09/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-168-52:2722045:2722470 [0] NCCL INFO Channel 09/0 : 120[0] -> 127[7] via P2P/IPC
+ip-26-0-168-34:272150:272576 [0] NCCL INFO Channel 01/0 : 112[0] -> 119[7] via P2P/IPC
+ip-26-0-160-225:124162:124587 [2] NCCL INFO Channel 02/0 : 123[3] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815963:2816400 [0] NCCL INFO Channel 01/0 : 56[0] -> 63[7] via P2P/IPC
+ip-26-0-165-202:153569:153981 [7] NCCL INFO Channel 06/0 : 95[7] -> 102[6] [send] via NET/Libfabric/6(94)/GDRDMA
+ip-26-0-161-153:2636084:2636511 [0] NCCL INFO Channel 13/0 : 48[0] -> 52[4] via P2P/IPC
+ip-26-0-160-225:124160:124581 [0] NCCL INFO Channel 02/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-165-59:1557266:1557689 [0] NCCL INFO Channel 14/0 : 96[0] -> 101[5] via P2P/IPC
+ip-26-0-160-242:1364902:1365333 [0] NCCL INFO Channel 10/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-160-225:124163:124588 [3] NCCL INFO Channel 02/0 : 3[3] -> 10[2] [send] via NET/Libfabric/2(2)/GDRDMA
+ip-26-0-160-225:124162:124587 [2] NCCL INFO Channel 10/0 : 123[3] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-52:2722045:2722470 [0] NCCL INFO Channel 10/0 : 120[0] -> 127[7] via P2P/IPC
+ip-26-0-161-78:114512:114930 [0] NCCL INFO Channel 06/0 : 64[0] -> 69[5] via P2P/IPC
+ip-26-0-168-34:272150:272576 [0] NCCL INFO Channel 02/0 : 112[0] -> 119[7] via P2P/IPC
+ip-26-0-161-123:32707:33124 [3] NCCL INFO Channel 03/0 : 20[4] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:124160:124581 [0] NCCL INFO Channel 03/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-160-225:124163:124588 [3] NCCL INFO Channel 10/0 : 3[3] -> 10[2] [send] via NET/Libfabric/2(2)/GDRDMA
+ip-26-0-160-242:1364902:1365333 [0] NCCL INFO Channel 11/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-161-123:32707:33124 [3] NCCL INFO Channel 11/0 : 20[4] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-52:2722045:2722470 [0] NCCL INFO Channel 11/0 : 120[0] -> 127[7] via P2P/IPC
+ip-26-0-161-178:2815963:2816400 [0] NCCL INFO Channel 02/0 : 56[0] -> 63[7] via P2P/IPC
+ip-26-0-165-59:1557266:1557689 [0] NCCL INFO Channel 07/0 : 96[0] -> 102[6] via P2P/IPC
+ip-26-0-168-34:272150:272576 [0] NCCL INFO Channel 03/0 : 112[0] -> 119[7] via P2P/IPC
+ip-26-0-160-225:124160:124581 [0] NCCL INFO Channel 04/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-161-78:114512:114930 [0] NCCL INFO Channel 14/0 : 64[0] -> 69[5] via P2P/IPC
+ip-26-0-161-123:32707:33124 [3] NCCL INFO Channel 02/0 : 27[3] -> 34[2] [send] via NET/Libfabric/2(26)/GDRDMA
+ip-26-0-160-242:1364902:1365333 [0] NCCL INFO Channel 12/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-165-202:153562:153983 [0] NCCL INFO Channel 11/0 : 88[0] -> 95[7] via P2P/IPC
+ip-26-0-168-52:2722045:2722470 [0] NCCL INFO Channel 12/0 : 120[0] -> 127[7] via P2P/IPC
+ip-26-0-161-123:32707:33124 [3] NCCL INFO Channel 10/0 : 27[3] -> 34[2] [send] via NET/Libfabric/2(26)/GDRDMA
+ip-26-0-161-178:2815963:2816400 [0] NCCL INFO Channel 03/0 : 56[0] -> 63[7] via P2P/IPC
+ip-26-0-168-34:272150:272576 [0] NCCL INFO Channel 04/0 : 112[0] -> 119[7] via P2P/IPC
+ip-26-0-160-225:124160:124581 [0] NCCL INFO Channel 09/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-165-59:1557266:1557689 [0] NCCL INFO Channel 15/0 : 96[0] -> 102[6] via P2P/IPC
+ip-26-0-165-202:153562:153983 [0] NCCL INFO Channel 12/0 : 88[0] -> 95[7] via P2P/IPC
+ip-26-0-168-52:2722052:2722464 [7] NCCL INFO Channel 06/0 : 127[7] -> 6[6] [send] via NET/Libfabric/6(126)/GDRDMA
+ip-26-0-161-178:2815963:2816400 [0] NCCL INFO Channel 04/0 : 56[0] -> 63[7] via P2P/IPC
+ip-26-0-160-225:124160:124581 [0] NCCL INFO Channel 10/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-161-78:114512:114930 [0] NCCL INFO Channel 07/0 : 64[0] -> 70[6] via P2P/IPC
+ip-26-0-168-34:272150:272576 [0] NCCL INFO Channel 09/0 : 112[0] -> 119[7] via P2P/IPC
+ip-26-0-168-52:2722052:2722464 [7] NCCL INFO Channel 14/0 : 127[7] -> 6[6] [send] via NET/Libfabric/6(126)/GDRDMA
+ip-26-0-161-153:2636086:2636507 [2] NCCL INFO Channel 01/0 : 50[2] -> 57[1] [send] via NET/Libfabric/1(49)/GDRDMA
+ip-26-0-161-178:2815963:2816400 [0] NCCL INFO Channel 09/0 : 56[0] -> 63[7] via P2P/IPC
+ip-26-0-160-225:124160:124581 [0] NCCL INFO Channel 11/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-168-34:272150:272576 [0] NCCL INFO Channel 10/0 : 112[0] -> 119[7] via P2P/IPC
+ip-26-0-165-59:1557266:1557689 [0] NCCL INFO Channel 01/0 : 96[0] -> 103[7] via P2P/IPC
+ip-26-0-161-178:2815963:2816400 [0] NCCL INFO Channel 10/0 : 56[0] -> 63[7] via P2P/IPC
+ip-26-0-168-34:272150:272576 [0] NCCL INFO Channel 11/0 : 112[0] -> 119[7] via P2P/IPC
+ip-26-0-160-225:124160:124581 [0] NCCL INFO Channel 12/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-165-59:1557266:1557689 [0] NCCL INFO Channel 02/0 : 96[0] -> 103[7] via P2P/IPC
+ip-26-0-161-178:2815963:2816400 [0] NCCL INFO Channel 11/0 : 56[0] -> 63[7] via P2P/IPC
+ip-26-0-160-225:124167:124586 [7] NCCL INFO Channel 06/0 : 7[7] -> 14[6] [send] via NET/Libfabric/6(6)/GDRDMA
+ip-26-0-168-34:272150:272576 [0] NCCL INFO Channel 12/0 : 112[0] -> 119[7] via P2P/IPC
+ip-26-0-161-153:2636084:2636511 [0] NCCL INFO Channel 06/0 : 48[0] -> 53[5] via P2P/IPC
+ip-26-0-165-59:1557266:1557689 [0] NCCL INFO Channel 03/0 : 96[0] -> 103[7] via P2P/IPC
+ip-26-0-161-178:2815963:2816400 [0] NCCL INFO Channel 12/0 : 56[0] -> 63[7] via P2P/IPC
+ip-26-0-168-34:272157:272572 [7] NCCL INFO Channel 06/0 : 119[7] -> 126[6] [send] via NET/Libfabric/6(118)/GDRDMA
+ip-26-0-165-59:1557266:1557689 [0] NCCL INFO Channel 04/0 : 96[0] -> 103[7] via P2P/IPC
+ip-26-0-161-153:2636084:2636511 [0] NCCL INFO Channel 14/0 : 48[0] -> 53[5] via P2P/IPC
+ip-26-0-165-59:1557266:1557689 [0] NCCL INFO Channel 09/0 : 96[0] -> 103[7] via P2P/IPC
+ip-26-0-165-59:1557266:1557689 [0] NCCL INFO Channel 10/0 : 96[0] -> 103[7] via P2P/IPC
+ip-26-0-161-153:2636084:2636511 [0] NCCL INFO Channel 07/0 : 48[0] -> 54[6] via P2P/IPC
+ip-26-0-165-59:1557266:1557689 [0] NCCL INFO Channel 11/0 : 96[0] -> 103[7] via P2P/IPC
+ip-26-0-165-59:1557266:1557689 [0] NCCL INFO Channel 12/0 : 96[0] -> 103[7] via P2P/IPC
+ip-26-0-161-153:2636084:2636511 [0] NCCL INFO Channel 15/0 : 48[0] -> 54[6] via P2P/IPC
+ip-26-0-161-142:62881:63298 [0] NCCL INFO Channel 06/0 : 40[0] -> 45[5] via P2P/IPC
+ip-26-0-161-78:114512:114930 [0] NCCL INFO Channel 15/0 : 64[0] -> 70[6] via P2P/IPC
+ip-26-0-161-142:62881:63298 [0] NCCL INFO Channel 14/0 : 40[0] -> 45[5] via P2P/IPC
+ip-26-0-161-153:2636084:2636511 [0] NCCL INFO Channel 01/0 : 48[0] -> 55[7] via P2P/IPC
+ip-26-0-161-153:2636084:2636511 [0] NCCL INFO Channel 02/0 : 48[0] -> 55[7] via P2P/IPC
+ip-26-0-161-142:62881:63298 [0] NCCL INFO Channel 07/0 : 40[0] -> 46[6] via P2P/IPC
+ip-26-0-161-153:2636084:2636511 [0] NCCL INFO Channel 03/0 : 48[0] -> 55[7] via P2P/IPC
+ip-26-0-161-142:62881:63298 [0] NCCL INFO Channel 15/0 : 40[0] -> 46[6] via P2P/IPC
+ip-26-0-161-153:2636084:2636511 [0] NCCL INFO Channel 04/0 : 48[0] -> 55[7] via P2P/IPC
+ip-26-0-161-142:62881:63298 [0] NCCL INFO Channel 01/0 : 40[0] -> 47[7] via P2P/IPC
+ip-26-0-161-153:2636084:2636511 [0] NCCL INFO Channel 09/0 : 48[0] -> 55[7] via P2P/IPC
+ip-26-0-161-78:114512:114930 [0] NCCL INFO Channel 01/0 : 64[0] -> 71[7] via P2P/IPC
+ip-26-0-161-142:62881:63298 [0] NCCL INFO Channel 02/0 : 40[0] -> 47[7] via P2P/IPC
+ip-26-0-161-153:2636084:2636511 [0] NCCL INFO Channel 10/0 : 48[0] -> 55[7] via P2P/IPC
+ip-26-0-161-78:114512:114930 [0] NCCL INFO Channel 02/0 : 64[0] -> 71[7] via P2P/IPC
+ip-26-0-161-142:62888:63302 [7] NCCL INFO Channel 06/0 : 47[7] -> 54[6] [send] via NET/Libfabric/6(46)/GDRDMA
+ip-26-0-161-142:62881:63298 [0] NCCL INFO Channel 03/0 : 40[0] -> 47[7] via P2P/IPC
+ip-26-0-161-153:2636084:2636511 [0] NCCL INFO Channel 11/0 : 48[0] -> 55[7] via P2P/IPC
+ip-26-0-161-78:114512:114930 [0] NCCL INFO Channel 03/0 : 64[0] -> 71[7] via P2P/IPC
+ip-26-0-161-153:2636091:2636512 [7] NCCL INFO Channel 06/0 : 55[7] -> 62[6] [send] via NET/Libfabric/6(54)/GDRDMA
+ip-26-0-161-142:62881:63298 [0] NCCL INFO Channel 04/0 : 40[0] -> 47[7] via P2P/IPC
+ip-26-0-161-153:2636084:2636511 [0] NCCL INFO Channel 12/0 : 48[0] -> 55[7] via P2P/IPC
+ip-26-0-161-78:114512:114930 [0] NCCL INFO Channel 04/0 : 64[0] -> 71[7] via P2P/IPC
+ip-26-0-161-142:62881:63298 [0] NCCL INFO Channel 09/0 : 40[0] -> 47[7] via P2P/IPC
+ip-26-0-161-142:62881:63298 [0] NCCL INFO Channel 10/0 : 40[0] -> 47[7] via P2P/IPC
+ip-26-0-161-142:62881:63298 [0] NCCL INFO Channel 11/0 : 40[0] -> 47[7] via P2P/IPC
+ip-26-0-161-142:62881:63298 [0] NCCL INFO Channel 12/0 : 40[0] -> 47[7] via P2P/IPC
+ip-26-0-161-78:114512:114930 [0] NCCL INFO Channel 09/0 : 64[0] -> 71[7] via P2P/IPC
+ip-26-0-161-78:114512:114930 [0] NCCL INFO Channel 10/0 : 64[0] -> 71[7] via P2P/IPC
+ip-26-0-161-78:114512:114930 [0] NCCL INFO Channel 11/0 : 64[0] -> 71[7] via P2P/IPC
+ip-26-0-161-78:114512:114930 [0] NCCL INFO Channel 12/0 : 64[0] -> 71[7] via P2P/IPC
+ip-26-0-161-103:96889:97309 [1] NCCL INFO Channel 01/0 : 10[2] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96564:97009 [1] NCCL INFO Channel 01/0 : 26[2] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96565:97004 [2] NCCL INFO Channel 01/0 : 34[2] -> 41[1] [send] via NET/Libfabric/1(33)/GDRDMA
+ip-26-0-161-103:96890:97313 [2] NCCL INFO Channel 01/0 : 18[2] -> 25[1] [send] via NET/Libfabric/1(17)/GDRDMA
+ip-26-0-161-138:96564:97009 [1] NCCL INFO Channel 09/0 : 26[2] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96889:97309 [1] NCCL INFO Channel 09/0 : 10[2] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96890:97313 [2] NCCL INFO Channel 09/0 : 18[2] -> 25[1] [send] via NET/Libfabric/1(17)/GDRDMA
+ip-26-0-161-138:96565:97004 [2] NCCL INFO Channel 09/0 : 34[2] -> 41[1] [send] via NET/Libfabric/1(33)/GDRDMA
+ip-26-0-161-142:62882:63300 [1] NCCL INFO Channel 01/0 : 34[2] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62882:63300 [1] NCCL INFO Channel 09/0 : 34[2] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62883:63301 [2] NCCL INFO Channel 01/0 : 42[2] -> 49[1] [send] via NET/Libfabric/1(41)/GDRDMA
+ip-26-0-165-202:153563:153986 [1] NCCL INFO Channel 01/0 : 82[2] -> 89[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62883:63301 [2] NCCL INFO Channel 09/0 : 42[2] -> 49[1] [send] via NET/Libfabric/1(41)/GDRDMA
+ip-26-0-165-202:153564:153984 [2] NCCL INFO Channel 01/0 : 90[2] -> 97[1] [send] via NET/Libfabric/1(89)/GDRDMA
+ip-26-0-165-202:153563:153986 [1] NCCL INFO Channel 09/0 : 82[2] -> 89[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153564:153984 [2] NCCL INFO Channel 09/0 : 90[2] -> 97[1] [send] via NET/Libfabric/1(89)/GDRDMA
+ip-26-0-165-164:1227464:1227884 [1] NCCL INFO Channel 01/0 : 74[2] -> 81[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227465:1227881 [2] NCCL INFO Channel 01/0 : 82[2] -> 89[1] [send] via NET/Libfabric/1(81)/GDRDMA
+ip-26-0-165-164:1227464:1227884 [1] NCCL INFO Channel 09/0 : 74[2] -> 81[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227465:1227881 [2] NCCL INFO Channel 09/0 : 82[2] -> 89[1] [send] via NET/Libfabric/1(81)/GDRDMA
+ip-26-0-165-131:1630206:1630647 [1] NCCL INFO Channel 01/0 : 66[2] -> 73[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630207:1630648 [2] NCCL INFO Channel 09/0 : 74[2] -> 81[1] [send] via NET/Libfabric/1(73)/GDRDMA
+ip-26-0-161-142:62882:63300 [1] NCCL INFO Channel 01/0 : 41[1] -> 40[0] via P2P/IPC
+ip-26-0-165-131:1630206:1630647 [1] NCCL INFO Channel 09/0 : 66[2] -> 73[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62882:63300 [1] NCCL INFO Channel 02/0 : 41[1] -> 40[0] via P2P/IPC
+ip-26-0-160-225:124161:124582 [1] NCCL INFO Channel 01/0 : 122[2] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124161:124582 [1] NCCL INFO Channel 09/0 : 122[2] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124162:124587 [2] NCCL INFO Channel 01/0 : 2[2] -> 9[1] [send] via NET/Libfabric/1(1)/GDRDMA
+ip-26-0-168-30:235649:236071 [1] NCCL INFO Channel 01/0 : 98[2] -> 105[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62882:63300 [1] NCCL INFO Channel 03/0 : 41[1] -> 40[0] via P2P/IPC
+ip-26-0-160-225:124162:124587 [2] NCCL INFO Channel 09/0 : 2[2] -> 9[1] [send] via NET/Libfabric/1(1)/GDRDMA
+ip-26-0-168-30:235650:236065 [2] NCCL INFO Channel 01/0 : 106[2] -> 113[1] [send] via NET/Libfabric/1(105)/GDRDMA
+ip-26-0-168-30:235649:236071 [1] NCCL INFO Channel 09/0 : 98[2] -> 105[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235650:236065 [2] NCCL INFO Channel 09/0 : 106[2] -> 113[1] [send] via NET/Libfabric/1(105)/GDRDMA
+ip-26-0-161-142:62882:63300 [1] NCCL INFO Channel 04/0 : 41[1] -> 40[0] via P2P/IPC
+ip-26-0-161-142:62882:63300 [1] NCCL INFO Channel 05/0 : 41[1] -> 40[0] via P2P/IPC
+ip-26-0-161-138:96565:97004 [2] NCCL INFO Channel 02/0 : 34[2] -> 33[1] via P2P/IPC
+ip-26-0-161-142:62882:63300 [1] NCCL INFO Channel 06/0 : 41[1] -> 40[0] via P2P/IPC
+ip-26-0-161-138:96565:97004 [2] NCCL INFO Channel 03/0 : 34[2] -> 33[1] via P2P/IPC
+ip-26-0-161-138:96565:97004 [2] NCCL INFO Channel 04/0 : 34[2] -> 33[1] via P2P/IPC
+ip-26-0-161-142:62882:63300 [1] NCCL INFO Channel 07/0 : 41[1] -> 40[0] via P2P/IPC
+ip-26-0-161-138:96565:97004 [2] NCCL INFO Channel 05/0 : 34[2] -> 33[1] via P2P/IPC
+ip-26-0-161-142:62882:63300 [1] NCCL INFO Channel 09/0 : 41[1] -> 40[0] via P2P/IPC
+ip-26-0-161-138:96565:97004 [2] NCCL INFO Channel 06/0 : 34[2] -> 33[1] via P2P/IPC
+ip-26-0-161-142:62882:63300 [1] NCCL INFO Channel 10/0 : 41[1] -> 40[0] via P2P/IPC
+ip-26-0-165-164:1227467:1227886 [4] NCCL INFO Channel 04/0 : 77[5] -> 84[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96565:97004 [2] NCCL INFO Channel 07/0 : 34[2] -> 33[1] via P2P/IPC
+ip-26-0-165-164:1227467:1227886 [4] NCCL INFO Channel 12/0 : 77[5] -> 84[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62882:63300 [1] NCCL INFO Channel 11/0 : 41[1] -> 40[0] via P2P/IPC
+ip-26-0-160-242:1364903:1365334 [1] NCCL INFO Channel 01/0 : 2[2] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96568:97007 [5] NCCL INFO Channel 05/0 : 30[6] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364904:1365335 [2] NCCL INFO Channel 09/0 : 10[2] -> 17[1] [send] via NET/Libfabric/1(9)/GDRDMA
+ip-26-0-161-138:96568:97007 [5] NCCL INFO Channel 13/0 : 30[6] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96892:97308 [4] NCCL INFO Channel 04/0 : 13[5] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227467:1227886 [4] NCCL INFO Channel 03/0 : 84[4] -> 91[3] [send] via NET/Libfabric/3(83)/GDRDMA
+ip-26-0-165-202:153563:153986 [1] NCCL INFO Channel 01/0 : 89[1] -> 88[0] via P2P/IPC
+ip-26-0-161-142:62882:63300 [1] NCCL INFO Channel 12/0 : 41[1] -> 40[0] via P2P/IPC
+ip-26-0-160-242:1364903:1365334 [1] NCCL INFO Channel 09/0 : 2[2] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96892:97308 [4] NCCL INFO Channel 12/0 : 13[5] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227467:1227886 [4] NCCL INFO Channel 11/0 : 84[4] -> 91[3] [send] via NET/Libfabric/3(83)/GDRDMA
+ip-26-0-165-164:1227468:1227883 [5] NCCL INFO Channel 05/0 : 78[6] -> 85[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153563:153986 [1] NCCL INFO Channel 02/0 : 89[1] -> 88[0] via P2P/IPC
+ip-26-0-161-142:62882:63300 [1] NCCL INFO Channel 13/0 : 41[1] -> 40[0] via P2P/IPC
+ip-26-0-165-164:1227468:1227883 [5] NCCL INFO Channel 13/0 : 78[6] -> 85[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96892:97308 [4] NCCL INFO Channel 03/0 : 20[4] -> 27[3] [send] via NET/Libfabric/3(19)/GDRDMA
+ip-26-0-161-138:96567:97008 [4] NCCL INFO Channel 04/0 : 29[5] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272151:272571 [1] NCCL INFO Channel 01/0 : 106[2] -> 113[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96567:97008 [4] NCCL INFO Channel 12/0 : 29[5] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96892:97308 [4] NCCL INFO Channel 11/0 : 20[4] -> 27[3] [send] via NET/Libfabric/3(19)/GDRDMA
+ip-26-0-165-202:153563:153986 [1] NCCL INFO Channel 03/0 : 89[1] -> 88[0] via P2P/IPC
+ip-26-0-161-142:62882:63300 [1] NCCL INFO Channel 14/0 : 41[1] -> 40[0] via P2P/IPC
+ip-26-0-161-138:96568:97007 [5] NCCL INFO Channel 04/0 : 37[5] -> 44[4] [send] via NET/Libfabric/4(36)/GDRDMA
+ip-26-0-161-138:96568:97007 [5] NCCL INFO Channel 12/0 : 37[5] -> 44[4] [send] via NET/Libfabric/4(36)/GDRDMA
+ip-26-0-165-164:1227468:1227883 [5] NCCL INFO Channel 04/0 : 85[5] -> 92[4] [send] via NET/Libfabric/4(84)/GDRDMA
+ip-26-0-161-138:96569:97006 [6] NCCL INFO Channel 06/0 : 31[7] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227468:1227883 [5] NCCL INFO Channel 12/0 : 85[5] -> 92[4] [send] via NET/Libfabric/4(84)/GDRDMA
+ip-26-0-161-138:96567:97008 [4] NCCL INFO Channel 03/0 : 36[4] -> 43[3] [send] via NET/Libfabric/3(35)/GDRDMA
+ip-26-0-165-202:153563:153986 [1] NCCL INFO Channel 04/0 : 89[1] -> 88[0] via P2P/IPC
+ip-26-0-161-142:62882:63300 [1] NCCL INFO Channel 15/0 : 41[1] -> 40[0] via P2P/IPC
+ip-26-0-161-138:96570:97005 [7] NCCL INFO Channel 14/0 : 39[7] -> 46[6] [send] via NET/Libfabric/6(38)/GDRDMA
+ip-26-0-161-138:96565:97004 [2] NCCL INFO Channel 10/0 : 34[2] -> 33[1] via P2P/IPC
+ip-26-0-161-138:96569:97006 [6] NCCL INFO Channel 14/0 : 31[7] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96567:97008 [4] NCCL INFO Channel 11/0 : 36[4] -> 43[3] [send] via NET/Libfabric/3(35)/GDRDMA
+ip-26-0-161-138:96569:97006 [6] NCCL INFO Channel 05/0 : 38[6] -> 45[5] [send] via NET/Libfabric/5(37)/GDRDMA
+ip-26-0-161-78:114514:114932 [2] NCCL INFO Channel 02/0 : 66[2] -> 65[1] via P2P/IPC
+ip-26-0-161-138:96565:97004 [2] NCCL INFO Channel 11/0 : 34[2] -> 33[1] via P2P/IPC
+ip-26-0-161-138:96569:97006 [6] NCCL INFO Channel 13/0 : 38[6] -> 45[5] [send] via NET/Libfabric/5(37)/GDRDMA
+ip-26-0-165-202:153563:153986 [1] NCCL INFO Channel 05/0 : 89[1] -> 88[0] via P2P/IPC
+ip-26-0-168-34:272152:272575 [2] NCCL INFO Channel 01/0 : 114[2] -> 121[1] [send] via NET/Libfabric/1(113)/GDRDMA
+ip-26-0-168-34:272151:272571 [1] NCCL INFO Channel 09/0 : 106[2] -> 113[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272152:272575 [2] NCCL INFO Channel 09/0 : 114[2] -> 121[1] [send] via NET/Libfabric/1(113)/GDRDMA
+ip-26-0-161-78:114514:114932 [2] NCCL INFO Channel 03/0 : 66[2] -> 65[1] via P2P/IPC
+ip-26-0-161-138:96565:97004 [2] NCCL INFO Channel 12/0 : 34[2] -> 33[1] via P2P/IPC
+ip-26-0-165-59:1557267:1557683 [1] NCCL INFO Channel 01/0 : 90[2] -> 97[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124161:124582 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-78:114514:114932 [2] NCCL INFO Channel 04/0 : 66[2] -> 65[1] via P2P/IPC
+ip-26-0-165-59:1557267:1557683 [1] NCCL INFO Channel 09/0 : 90[2] -> 97[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153563:153986 [1] NCCL INFO Channel 06/0 : 89[1] -> 88[0] via P2P/IPC
+ip-26-0-168-52:2722047:2722468 [2] NCCL INFO Channel 02/0 : 122[2] -> 121[1] via P2P/IPC
+ip-26-0-161-138:96565:97004 [2] NCCL INFO Channel 13/0 : 34[2] -> 33[1] via P2P/IPC
+ip-26-0-161-78:114514:114932 [2] NCCL INFO Channel 05/0 : 66[2] -> 65[1] via P2P/IPC
+ip-26-0-160-225:124161:124582 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-131:1630211:1630649 [6] NCCL INFO Channel 06/0 : 71[7] -> 78[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-52:2722047:2722468 [2] NCCL INFO Channel 03/0 : 122[2] -> 121[1] via P2P/IPC
+ip-26-0-165-202:153563:153986 [1] NCCL INFO Channel 07/0 : 89[1] -> 88[0] via P2P/IPC
+ip-26-0-160-225:124164:124585 [4] NCCL INFO Channel 04/0 : 125[5] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815964:2816401 [1] NCCL INFO Channel 01/0 : 50[2] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630212:1630652 [7] NCCL INFO Channel 14/0 : 79[7] -> 86[6] [send] via NET/Libfabric/6(78)/GDRDMA
+ip-26-0-160-225:124164:124585 [4] NCCL INFO Channel 12/0 : 125[5] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815964:2816401 [1] NCCL INFO Channel 09/0 : 50[2] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114514:114932 [2] NCCL INFO Channel 06/0 : 66[2] -> 65[1] via P2P/IPC
+ip-26-0-165-131:1630211:1630649 [6] NCCL INFO Channel 14/0 : 71[7] -> 78[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235649:236071 [1] NCCL INFO Channel 01/0 : 105[1] -> 104[0] via P2P/IPC
+ip-26-0-161-178:2815965:2816395 [2] NCCL INFO Channel 01/0 : 58[2] -> 65[1] [send] via NET/Libfabric/1(57)/GDRDMA
+ip-26-0-168-52:2722047:2722468 [2] NCCL INFO Channel 04/0 : 122[2] -> 121[1] via P2P/IPC
+ip-26-0-165-164:1227469:1227882 [6] NCCL INFO Channel 06/0 : 79[7] -> 86[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96565:97004 [2] NCCL INFO Channel 14/0 : 34[2] -> 33[1] via P2P/IPC
+ip-26-0-165-202:153563:153986 [1] NCCL INFO Channel 09/0 : 89[1] -> 88[0] via P2P/IPC
+ip-26-0-160-225:124161:124582 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-164:1227469:1227882 [6] NCCL INFO Channel 14/0 : 79[7] -> 86[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815965:2816395 [2] NCCL INFO Channel 09/0 : 58[2] -> 65[1] [send] via NET/Libfabric/1(57)/GDRDMA
+ip-26-0-161-78:114514:114932 [2] NCCL INFO Channel 07/0 : 66[2] -> 65[1] via P2P/IPC
+ip-26-0-160-225:124164:124585 [4] NCCL INFO Channel 03/0 : 4[4] -> 11[3] [send] via NET/Libfabric/3(3)/GDRDMA
+ip-26-0-168-30:235649:236071 [1] NCCL INFO Channel 02/0 : 105[1] -> 104[0] via P2P/IPC
+ip-26-0-165-164:1227469:1227882 [6] NCCL INFO Channel 05/0 : 86[6] -> 93[5] [send] via NET/Libfabric/5(85)/GDRDMA
+ip-26-0-160-225:124164:124585 [4] NCCL INFO Channel 11/0 : 4[4] -> 11[3] [send] via NET/Libfabric/3(3)/GDRDMA
+ip-26-0-168-52:2722047:2722468 [2] NCCL INFO Channel 05/0 : 122[2] -> 121[1] via P2P/IPC
+ip-26-0-161-123:32705:33128 [1] NCCL INFO Channel 01/0 : 18[2] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124161:124582 [1] NCCL INFO Channel 04/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-138:96565:97004 [2] NCCL INFO Channel 15/0 : 34[2] -> 33[1] via P2P/IPC
+ip-26-0-165-202:153563:153986 [1] NCCL INFO Channel 10/0 : 89[1] -> 88[0] via P2P/IPC
+ip-26-0-161-78:114514:114932 [2] NCCL INFO Channel 10/0 : 66[2] -> 65[1] via P2P/IPC
+ip-26-0-165-164:1227469:1227882 [6] NCCL INFO Channel 13/0 : 86[6] -> 93[5] [send] via NET/Libfabric/5(85)/GDRDMA
+ip-26-0-168-30:235649:236071 [1] NCCL INFO Channel 03/0 : 105[1] -> 104[0] via P2P/IPC
+ip-26-0-161-123:32705:33128 [1] NCCL INFO Channel 09/0 : 18[2] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-52:2722047:2722468 [2] NCCL INFO Channel 06/0 : 122[2] -> 121[1] via P2P/IPC
+ip-26-0-161-123:32706:33122 [2] NCCL INFO Channel 01/0 : 26[2] -> 33[1] [send] via NET/Libfabric/1(25)/GDRDMA
+ip-26-0-161-178:2815967:2816399 [4] NCCL INFO Channel 04/0 : 53[5] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32706:33122 [2] NCCL INFO Channel 09/0 : 26[2] -> 33[1] [send] via NET/Libfabric/1(25)/GDRDMA
+ip-26-0-161-78:114514:114932 [2] NCCL INFO Channel 11/0 : 66[2] -> 65[1] via P2P/IPC
+ip-26-0-165-131:1630209:1630651 [4] NCCL INFO Channel 04/0 : 69[5] -> 76[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815967:2816399 [4] NCCL INFO Channel 12/0 : 53[5] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:124161:124582 [1] NCCL INFO Channel 05/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-168-30:235649:236071 [1] NCCL INFO Channel 04/0 : 105[1] -> 104[0] via P2P/IPC
+ip-26-0-168-52:2722047:2722468 [2] NCCL INFO Channel 07/0 : 122[2] -> 121[1] via P2P/IPC
+ip-26-0-165-202:153563:153986 [1] NCCL INFO Channel 11/0 : 89[1] -> 88[0] via P2P/IPC
+ip-26-0-165-164:1227465:1227881 [2] NCCL INFO Channel 02/0 : 82[2] -> 81[1] via P2P/IPC
+ip-26-0-165-131:1630209:1630651 [4] NCCL INFO Channel 12/0 : 69[5] -> 76[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630210:1630653 [5] NCCL INFO Channel 05/0 : 70[6] -> 77[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815967:2816399 [4] NCCL INFO Channel 03/0 : 60[4] -> 67[3] [send] via NET/Libfabric/3(59)/GDRDMA
+ip-26-0-168-52:2722047:2722468 [2] NCCL INFO Channel 10/0 : 122[2] -> 121[1] via P2P/IPC
+ip-26-0-161-78:114513:114933 [1] NCCL INFO Channel 01/0 : 58[2] -> 65[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235649:236071 [1] NCCL INFO Channel 05/0 : 105[1] -> 104[0] via P2P/IPC
+ip-26-0-165-131:1630210:1630653 [5] NCCL INFO Channel 13/0 : 70[6] -> 77[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124161:124582 [1] NCCL INFO Channel 06/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-178:2815967:2816399 [4] NCCL INFO Channel 11/0 : 60[4] -> 67[3] [send] via NET/Libfabric/3(59)/GDRDMA
+ip-26-0-165-131:1630211:1630649 [6] NCCL INFO Channel 05/0 : 78[6] -> 85[5] [send] via NET/Libfabric/5(77)/GDRDMA
+ip-26-0-161-78:114513:114933 [1] NCCL INFO Channel 09/0 : 58[2] -> 65[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153563:153986 [1] NCCL INFO Channel 12/0 : 89[1] -> 88[0] via P2P/IPC
+ip-26-0-165-164:1227465:1227881 [2] NCCL INFO Channel 03/0 : 82[2] -> 81[1] via P2P/IPC
+ip-26-0-165-131:1630209:1630651 [4] NCCL INFO Channel 03/0 : 76[4] -> 83[3] [send] via NET/Libfabric/3(75)/GDRDMA
+ip-26-0-165-131:1630211:1630649 [6] NCCL INFO Channel 13/0 : 78[6] -> 85[5] [send] via NET/Libfabric/5(77)/GDRDMA
+ip-26-0-168-52:2722047:2722468 [2] NCCL INFO Channel 11/0 : 122[2] -> 121[1] via P2P/IPC
+ip-26-0-165-131:1630209:1630651 [4] NCCL INFO Channel 11/0 : 76[4] -> 83[3] [send] via NET/Libfabric/3(75)/GDRDMA
+ip-26-0-165-131:1630210:1630653 [5] NCCL INFO Channel 04/0 : 77[5] -> 84[4] [send] via NET/Libfabric/4(76)/GDRDMA
+ip-26-0-168-30:235654:236070 [6] NCCL INFO Channel 06/0 : 103[7] -> 110[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235649:236071 [1] NCCL INFO Channel 06/0 : 105[1] -> 104[0] via P2P/IPC
+ip-26-0-165-131:1630210:1630653 [5] NCCL INFO Channel 12/0 : 77[5] -> 84[4] [send] via NET/Libfabric/4(76)/GDRDMA
+ip-26-0-168-30:235655:236069 [7] NCCL INFO Channel 14/0 : 111[7] -> 118[6] [send] via NET/Libfabric/6(110)/GDRDMA
+ip-26-0-160-225:124161:124582 [1] NCCL INFO Channel 07/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-164:1227465:1227881 [2] NCCL INFO Channel 04/0 : 82[2] -> 81[1] via P2P/IPC
+ip-26-0-165-202:153566:153980 [4] NCCL INFO Channel 04/0 : 85[5] -> 92[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722047:2722468 [2] NCCL INFO Channel 12/0 : 122[2] -> 121[1] via P2P/IPC
+ip-26-0-168-30:235654:236070 [6] NCCL INFO Channel 14/0 : 103[7] -> 110[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96893:97310 [5] NCCL INFO Channel 05/0 : 14[6] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153563:153986 [1] NCCL INFO Channel 13/0 : 89[1] -> 88[0] via P2P/IPC
+ip-26-0-168-30:235649:236071 [1] NCCL INFO Channel 07/0 : 105[1] -> 104[0] via P2P/IPC
+ip-26-0-165-202:153566:153980 [4] NCCL INFO Channel 12/0 : 85[5] -> 92[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96893:97310 [5] NCCL INFO Channel 13/0 : 14[6] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153568:153982 [6] NCCL INFO Channel 06/0 : 87[7] -> 94[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364906:1365331 [4] NCCL INFO Channel 04/0 : 5[5] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272154:272569 [4] NCCL INFO Channel 04/0 : 109[5] -> 116[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227464:1227884 [1] NCCL INFO Channel 01/0 : 81[1] -> 80[0] via P2P/IPC
+ip-26-0-165-202:153569:153981 [7] NCCL INFO Channel 14/0 : 95[7] -> 102[6] [send] via NET/Libfabric/6(94)/GDRDMA
+ip-26-0-165-202:153566:153980 [4] NCCL INFO Channel 03/0 : 92[4] -> 99[3] [send] via NET/Libfabric/3(91)/GDRDMA
+ip-26-0-168-30:235649:236071 [1] NCCL INFO Channel 09/0 : 105[1] -> 104[0] via P2P/IPC
+ip-26-0-165-131:1630211:1630649 [6] NCCL INFO Channel 06/0 : 78[6] -> 75[3] via P2P/IPC
+ip-26-0-160-242:1364906:1365331 [4] NCCL INFO Channel 12/0 : 5[5] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227465:1227881 [2] NCCL INFO Channel 05/0 : 82[2] -> 81[1] via P2P/IPC
+ip-26-0-160-225:124161:124582 [1] NCCL INFO Channel 09/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-103:96893:97310 [5] NCCL INFO Channel 04/0 : 21[5] -> 28[4] [send] via NET/Libfabric/4(20)/GDRDMA
+ip-26-0-165-202:153567:153979 [5] NCCL INFO Channel 05/0 : 86[6] -> 93[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153568:153982 [6] NCCL INFO Channel 14/0 : 87[7] -> 94[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153563:153986 [1] NCCL INFO Channel 14/0 : 89[1] -> 88[0] via P2P/IPC
+ip-26-0-168-30:235652:236066 [4] NCCL INFO Channel 04/0 : 101[5] -> 108[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557268:1557686 [2] NCCL INFO Channel 02/0 : 98[2] -> 97[1] via P2P/IPC
+ip-26-0-165-202:153566:153980 [4] NCCL INFO Channel 11/0 : 92[4] -> 99[3] [send] via NET/Libfabric/3(91)/GDRDMA
+ip-26-0-161-103:96893:97310 [5] NCCL INFO Channel 12/0 : 21[5] -> 28[4] [send] via NET/Libfabric/4(20)/GDRDMA
+ip-26-0-168-30:235652:236066 [4] NCCL INFO Channel 12/0 : 101[5] -> 108[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153567:153979 [5] NCCL INFO Channel 13/0 : 86[6] -> 93[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96889:97309 [1] NCCL INFO Channel 01/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-165-164:1227464:1227884 [1] NCCL INFO Channel 02/0 : 81[1] -> 80[0] via P2P/IPC
+ip-26-0-165-202:153568:153982 [6] NCCL INFO Channel 05/0 : 94[6] -> 101[5] [send] via NET/Libfabric/5(93)/GDRDMA
+ip-26-0-160-242:1364906:1365331 [4] NCCL INFO Channel 03/0 : 12[4] -> 19[3] [send] via NET/Libfabric/3(11)/GDRDMA
+ip-26-0-165-202:153563:153986 [1] NCCL INFO Channel 15/0 : 89[1] -> 88[0] via P2P/IPC
+ip-26-0-160-225:124161:124582 [1] NCCL INFO Channel 10/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-202:153567:153979 [5] NCCL INFO Channel 04/0 : 93[5] -> 100[4] [send] via NET/Libfabric/4(92)/GDRDMA
+ip-26-0-160-242:1364906:1365331 [4] NCCL INFO Channel 11/0 : 12[4] -> 19[3] [send] via NET/Libfabric/3(11)/GDRDMA
+ip-26-0-165-202:153568:153982 [6] NCCL INFO Channel 13/0 : 94[6] -> 101[5] [send] via NET/Libfabric/5(93)/GDRDMA
+ip-26-0-165-59:1557268:1557686 [2] NCCL INFO Channel 03/0 : 98[2] -> 97[1] via P2P/IPC
+ip-26-0-168-30:235649:236071 [1] NCCL INFO Channel 10/0 : 105[1] -> 104[0] via P2P/IPC
+ip-26-0-165-202:153567:153979 [5] NCCL INFO Channel 12/0 : 93[5] -> 100[4] [send] via NET/Libfabric/4(92)/GDRDMA
+ip-26-0-168-52:2722047:2722468 [2] NCCL INFO Channel 13/0 : 122[2] -> 121[1] via P2P/IPC
+ip-26-0-168-30:235652:236066 [4] NCCL INFO Channel 03/0 : 108[4] -> 115[3] [send] via NET/Libfabric/3(107)/GDRDMA
+ip-26-0-161-103:96889:97309 [1] NCCL INFO Channel 02/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-160-242:1364908:1365332 [6] NCCL INFO Channel 06/0 : 7[7] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227468:1227883 [5] NCCL INFO Channel 05/0 : 85[5] -> 83[3] via P2P/IPC
+ip-26-0-168-30:235653:236067 [5] NCCL INFO Channel 05/0 : 102[6] -> 109[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364904:1365335 [2] NCCL INFO Channel 02/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-165-164:1227465:1227881 [2] NCCL INFO Channel 06/0 : 82[2] -> 81[1] via P2P/IPC
+ip-26-0-160-242:1364909:1365327 [7] NCCL INFO Channel 06/0 : 15[7] -> 22[6] [send] via NET/Libfabric/6(14)/GDRDMA
+ip-26-0-168-30:235652:236066 [4] NCCL INFO Channel 11/0 : 108[4] -> 115[3] [send] via NET/Libfabric/3(107)/GDRDMA
+ip-26-0-160-225:124166:124583 [6] NCCL INFO Channel 06/0 : 127[7] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364908:1365332 [6] NCCL INFO Channel 14/0 : 7[7] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235653:236067 [5] NCCL INFO Channel 13/0 : 102[6] -> 109[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235654:236070 [6] NCCL INFO Channel 05/0 : 110[6] -> 117[5] [send] via NET/Libfabric/5(109)/GDRDMA
+ip-26-0-165-59:1557268:1557686 [2] NCCL INFO Channel 04/0 : 98[2] -> 97[1] via P2P/IPC
+ip-26-0-160-242:1364909:1365327 [7] NCCL INFO Channel 14/0 : 15[7] -> 22[6] [send] via NET/Libfabric/6(14)/GDRDMA
+ip-26-0-168-52:2722047:2722468 [2] NCCL INFO Channel 14/0 : 122[2] -> 121[1] via P2P/IPC
+ip-26-0-160-225:124166:124583 [6] NCCL INFO Channel 14/0 : 127[7] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:124161:124582 [1] NCCL INFO Channel 11/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-103:96889:97309 [1] NCCL INFO Channel 03/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-160-242:1364903:1365334 [1] NCCL INFO Channel 01/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-165-164:1227464:1227884 [1] NCCL INFO Channel 03/0 : 81[1] -> 80[0] via P2P/IPC
+ip-26-0-168-30:235654:236070 [6] NCCL INFO Channel 13/0 : 110[6] -> 117[5] [send] via NET/Libfabric/5(109)/GDRDMA
+ip-26-0-160-242:1364904:1365335 [2] NCCL INFO Channel 03/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-161-123:32710:33125 [6] NCCL INFO Channel 06/0 : 23[7] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:124167:124586 [7] NCCL INFO Channel 14/0 : 7[7] -> 14[6] [send] via NET/Libfabric/6(6)/GDRDMA
+ip-26-0-160-242:1364907:1365329 [5] NCCL INFO Channel 05/0 : 6[6] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235649:236071 [1] NCCL INFO Channel 11/0 : 105[1] -> 104[0] via P2P/IPC
+ip-26-0-165-164:1227469:1227882 [6] NCCL INFO Channel 06/0 : 86[6] -> 83[3] via P2P/IPC
+ip-26-0-161-123:32708:33126 [4] NCCL INFO Channel 04/0 : 21[5] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32711:33127 [7] NCCL INFO Channel 14/0 : 31[7] -> 38[6] [send] via NET/Libfabric/6(30)/GDRDMA
+ip-26-0-168-30:235653:236067 [5] NCCL INFO Channel 04/0 : 109[5] -> 116[4] [send] via NET/Libfabric/4(108)/GDRDMA
+ip-26-0-161-123:32710:33125 [6] NCCL INFO Channel 14/0 : 23[7] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815968:2816394 [5] NCCL INFO Channel 05/0 : 54[6] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364907:1365329 [5] NCCL INFO Channel 13/0 : 6[6] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-52:2722046:2722466 [1] NCCL INFO Channel 01/0 : 114[2] -> 121[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235650:236065 [2] NCCL INFO Channel 02/0 : 106[2] -> 105[1] via P2P/IPC
+ip-26-0-168-30:235653:236067 [5] NCCL INFO Channel 12/0 : 109[5] -> 116[4] [send] via NET/Libfabric/4(108)/GDRDMA
+ip-26-0-161-123:32708:33126 [4] NCCL INFO Channel 12/0 : 21[5] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364903:1365334 [1] NCCL INFO Channel 02/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-160-242:1364908:1365332 [6] NCCL INFO Channel 05/0 : 14[6] -> 21[5] [send] via NET/Libfabric/5(13)/GDRDMA
+ip-26-0-168-52:2722047:2722468 [2] NCCL INFO Channel 15/0 : 122[2] -> 121[1] via P2P/IPC
+ip-26-0-168-52:2722049:2722467 [4] NCCL INFO Channel 04/0 : 117[5] -> 124[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557268:1557686 [2] NCCL INFO Channel 05/0 : 98[2] -> 97[1] via P2P/IPC
+ip-26-0-161-178:2815968:2816394 [5] NCCL INFO Channel 13/0 : 54[6] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364904:1365335 [2] NCCL INFO Channel 04/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-168-34:272154:272569 [4] NCCL INFO Channel 12/0 : 109[5] -> 116[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96889:97309 [1] NCCL INFO Channel 04/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-168-52:2722046:2722466 [1] NCCL INFO Channel 09/0 : 114[2] -> 121[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272151:272571 [1] NCCL INFO Channel 01/0 : 113[1] -> 112[0] via P2P/IPC
+ip-26-0-168-34:272154:272569 [4] NCCL INFO Channel 03/0 : 116[4] -> 123[3] [send] via NET/Libfabric/3(115)/GDRDMA
+ip-26-0-160-225:124161:124582 [1] NCCL INFO Channel 12/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-168-34:272151:272571 [1] NCCL INFO Channel 02/0 : 113[1] -> 112[0] via P2P/IPC
+ip-26-0-160-242:1364908:1365332 [6] NCCL INFO Channel 13/0 : 14[6] -> 21[5] [send] via NET/Libfabric/5(13)/GDRDMA
+ip-26-0-168-34:272154:272569 [4] NCCL INFO Channel 11/0 : 116[4] -> 123[3] [send] via NET/Libfabric/3(115)/GDRDMA
+ip-26-0-168-34:272151:272571 [1] NCCL INFO Channel 03/0 : 113[1] -> 112[0] via P2P/IPC
+ip-26-0-168-52:2722049:2722467 [4] NCCL INFO Channel 12/0 : 117[5] -> 124[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272151:272571 [1] NCCL INFO Channel 04/0 : 113[1] -> 112[0] via P2P/IPC
+ip-26-0-160-225:124162:124587 [2] NCCL INFO Channel 02/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-165-59:1557270:1557688 [4] NCCL INFO Channel 04/0 : 93[5] -> 100[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272151:272571 [1] NCCL INFO Channel 05/0 : 113[1] -> 112[0] via P2P/IPC
+ip-26-0-160-242:1364907:1365329 [5] NCCL INFO Channel 04/0 : 13[5] -> 20[4] [send] via NET/Libfabric/4(12)/GDRDMA
+ip-26-0-161-123:32708:33126 [4] NCCL INFO Channel 03/0 : 28[4] -> 35[3] [send] via NET/Libfabric/3(27)/GDRDMA
+ip-26-0-161-178:2815969:2816398 [6] NCCL INFO Channel 06/0 : 55[7] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235649:236071 [1] NCCL INFO Channel 12/0 : 105[1] -> 104[0] via P2P/IPC
+ip-26-0-165-59:1557270:1557688 [4] NCCL INFO Channel 12/0 : 93[5] -> 100[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636085:2636509 [1] NCCL INFO Channel 01/0 : 42[2] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364907:1365329 [5] NCCL INFO Channel 12/0 : 13[5] -> 20[4] [send] via NET/Libfabric/4(12)/GDRDMA
+ip-26-0-165-164:1227465:1227881 [2] NCCL INFO Channel 07/0 : 82[2] -> 81[1] via P2P/IPC
+ip-26-0-160-242:1364903:1365334 [1] NCCL INFO Channel 03/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-161-123:32708:33126 [4] NCCL INFO Channel 11/0 : 28[4] -> 35[3] [send] via NET/Libfabric/3(27)/GDRDMA
+ip-26-0-168-30:235650:236065 [2] NCCL INFO Channel 03/0 : 106[2] -> 105[1] via P2P/IPC
+ip-26-0-161-178:2815968:2816394 [5] NCCL INFO Channel 04/0 : 61[5] -> 68[4] [send] via NET/Libfabric/4(60)/GDRDMA
+ip-26-0-160-242:1364904:1365335 [2] NCCL INFO Channel 05/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-161-178:2815969:2816398 [6] NCCL INFO Channel 14/0 : 55[7] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96889:97309 [1] NCCL INFO Channel 05/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-161-153:2636086:2636507 [2] NCCL INFO Channel 09/0 : 50[2] -> 57[1] [send] via NET/Libfabric/1(49)/GDRDMA
+ip-26-0-161-153:2636085:2636509 [1] NCCL INFO Channel 09/0 : 42[2] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227468:1227883 [5] NCCL INFO Channel 13/0 : 85[5] -> 83[3] via P2P/IPC
+ip-26-0-160-225:124161:124582 [1] NCCL INFO Channel 13/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-59:1557268:1557686 [2] NCCL INFO Channel 06/0 : 98[2] -> 97[1] via P2P/IPC
+ip-26-0-161-178:2815968:2816394 [5] NCCL INFO Channel 12/0 : 61[5] -> 68[4] [send] via NET/Libfabric/4(60)/GDRDMA
+ip-26-0-161-178:2815969:2816398 [6] NCCL INFO Channel 05/0 : 62[6] -> 69[5] [send] via NET/Libfabric/5(61)/GDRDMA
+ip-26-0-165-164:1227464:1227884 [1] NCCL INFO Channel 04/0 : 81[1] -> 80[0] via P2P/IPC
+ip-26-0-168-52:2722050:2722471 [5] NCCL INFO Channel 05/0 : 118[6] -> 125[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364908:1365332 [6] NCCL INFO Channel 06/0 : 14[6] -> 11[3] via P2P/IPC
+ip-26-0-160-225:124162:124587 [2] NCCL INFO Channel 03/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-178:2815969:2816398 [6] NCCL INFO Channel 13/0 : 62[6] -> 69[5] [send] via NET/Libfabric/5(61)/GDRDMA
+ip-26-0-168-30:235649:236071 [1] NCCL INFO Channel 13/0 : 105[1] -> 104[0] via P2P/IPC
+ip-26-0-161-78:114514:114932 [2] NCCL INFO Channel 12/0 : 66[2] -> 65[1] via P2P/IPC
+ip-26-0-161-103:96893:97310 [5] NCCL INFO Channel 05/0 : 21[5] -> 19[3] via P2P/IPC
+ip-26-0-168-30:235650:236065 [2] NCCL INFO Channel 04/0 : 106[2] -> 105[1] via P2P/IPC
+ip-26-0-161-103:96889:97309 [1] NCCL INFO Channel 06/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-165-164:1227467:1227886 [4] NCCL INFO Channel 07/0 : 84[4] -> 95[7] [send] via NET/Libfabric/7(87)/GDRDMA
+ip-26-0-165-59:1557270:1557688 [4] NCCL INFO Channel 03/0 : 100[4] -> 107[3] [send] via NET/Libfabric/3(99)/GDRDMA
+ip-26-0-165-59:1557268:1557686 [2] NCCL INFO Channel 07/0 : 98[2] -> 97[1] via P2P/IPC
+ip-26-0-165-164:1227469:1227882 [6] NCCL INFO Channel 14/0 : 86[6] -> 83[3] via P2P/IPC
+ip-26-0-160-242:1364903:1365334 [1] NCCL INFO Channel 04/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-160-225:124165:124584 [5] NCCL INFO Channel 05/0 : 126[6] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557270:1557688 [4] NCCL INFO Channel 11/0 : 100[4] -> 107[3] [send] via NET/Libfabric/3(99)/GDRDMA
+ip-26-0-165-164:1227467:1227886 [4] NCCL INFO Channel 15/0 : 84[4] -> 95[7] [send] via NET/Libfabric/7(87)/GDRDMA
+ip-26-0-160-225:124161:124582 [1] NCCL INFO Channel 14/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-242:1364904:1365335 [2] NCCL INFO Channel 06/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-165-59:1557272:1557687 [6] NCCL INFO Channel 06/0 : 95[7] -> 102[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227465:1227881 [2] NCCL INFO Channel 10/0 : 82[2] -> 81[1] via P2P/IPC
+ip-26-0-160-225:124165:124584 [5] NCCL INFO Channel 13/0 : 126[6] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124166:124583 [6] NCCL INFO Channel 05/0 : 6[6] -> 13[5] [send] via NET/Libfabric/5(5)/GDRDMA
+ip-26-0-165-59:1557272:1557687 [6] NCCL INFO Channel 14/0 : 95[7] -> 102[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630211:1630649 [6] NCCL INFO Channel 14/0 : 78[6] -> 75[3] via P2P/IPC
+ip-26-0-160-225:124162:124587 [2] NCCL INFO Channel 04/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-78:114514:114932 [2] NCCL INFO Channel 13/0 : 66[2] -> 65[1] via P2P/IPC
+ip-26-0-168-30:235649:236071 [1] NCCL INFO Channel 14/0 : 105[1] -> 104[0] via P2P/IPC
+ip-26-0-168-52:2722050:2722471 [5] NCCL INFO Channel 13/0 : 118[6] -> 125[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153564:153984 [2] NCCL INFO Channel 02/0 : 90[2] -> 89[1] via P2P/IPC
+ip-26-0-165-164:1227464:1227884 [1] NCCL INFO Channel 05/0 : 81[1] -> 80[0] via P2P/IPC
+ip-26-0-160-225:124166:124583 [6] NCCL INFO Channel 13/0 : 6[6] -> 13[5] [send] via NET/Libfabric/5(5)/GDRDMA
+ip-26-0-168-52:2722050:2722471 [5] NCCL INFO Channel 04/0 : 125[5] -> 4[4] [send] via NET/Libfabric/4(124)/GDRDMA
+ip-26-0-165-59:1557267:1557683 [1] NCCL INFO Channel 01/0 : 97[1] -> 96[0] via P2P/IPC
+ip-26-0-160-225:124165:124584 [5] NCCL INFO Channel 04/0 : 5[5] -> 12[4] [send] via NET/Libfabric/4(4)/GDRDMA
+ip-26-0-168-52:2722050:2722471 [5] NCCL INFO Channel 12/0 : 125[5] -> 4[4] [send] via NET/Libfabric/4(124)/GDRDMA
+ip-26-0-168-30:235650:236065 [2] NCCL INFO Channel 05/0 : 106[2] -> 105[1] via P2P/IPC
+ip-26-0-165-202:153566:153980 [4] NCCL INFO Channel 07/0 : 92[4] -> 103[7] [send] via NET/Libfabric/7(95)/GDRDMA
+ip-26-0-160-225:124165:124584 [5] NCCL INFO Channel 12/0 : 5[5] -> 12[4] [send] via NET/Libfabric/4(4)/GDRDMA
+ip-26-0-161-123:32709:33129 [5] NCCL INFO Channel 05/0 : 22[6] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96889:97309 [1] NCCL INFO Channel 07/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-160-225:124161:124582 [1] NCCL INFO Channel 15/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-202:153566:153980 [4] NCCL INFO Channel 15/0 : 92[4] -> 103[7] [send] via NET/Libfabric/7(95)/GDRDMA
+ip-26-0-161-103:96893:97310 [5] NCCL INFO Channel 13/0 : 21[5] -> 19[3] via P2P/IPC
+ip-26-0-161-123:32709:33129 [5] NCCL INFO Channel 13/0 : 22[6] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557267:1557683 [1] NCCL INFO Channel 02/0 : 97[1] -> 96[0] via P2P/IPC
+ip-26-0-165-164:1227465:1227881 [2] NCCL INFO Channel 11/0 : 82[2] -> 81[1] via P2P/IPC
+ip-26-0-165-202:153567:153979 [5] NCCL INFO Channel 05/0 : 93[5] -> 91[3] via P2P/IPC
+ip-26-0-161-123:32710:33125 [6] NCCL INFO Channel 05/0 : 30[6] -> 37[5] [send] via NET/Libfabric/5(29)/GDRDMA
+ip-26-0-165-164:1227464:1227884 [1] NCCL INFO Channel 06/0 : 81[1] -> 80[0] via P2P/IPC
+ip-26-0-165-202:153564:153984 [2] NCCL INFO Channel 03/0 : 90[2] -> 89[1] via P2P/IPC
+ip-26-0-161-78:114516:114934 [4] NCCL INFO Channel 04/0 : 61[5] -> 68[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96894:97312 [6] NCCL INFO Channel 06/0 : 15[7] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364908:1365332 [6] NCCL INFO Channel 14/0 : 14[6] -> 11[3] via P2P/IPC
+ip-26-0-160-225:124162:124587 [2] NCCL INFO Channel 05/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-168-30:235649:236071 [1] NCCL INFO Channel 15/0 : 105[1] -> 104[0] via P2P/IPC
+ip-26-0-168-34:272151:272571 [1] NCCL INFO Channel 06/0 : 113[1] -> 112[0] via P2P/IPC
+ip-26-0-161-78:114516:114934 [4] NCCL INFO Channel 12/0 : 61[5] -> 68[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272151:272571 [1] NCCL INFO Channel 07/0 : 113[1] -> 112[0] via P2P/IPC
+ip-26-0-160-242:1364903:1365334 [1] NCCL INFO Channel 05/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-161-123:32710:33125 [6] NCCL INFO Channel 13/0 : 30[6] -> 37[5] [send] via NET/Libfabric/5(29)/GDRDMA
+ip-26-0-168-34:272151:272571 [1] NCCL INFO Channel 09/0 : 113[1] -> 112[0] via P2P/IPC
+ip-26-0-161-103:96894:97312 [6] NCCL INFO Channel 14/0 : 15[7] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364904:1365335 [2] NCCL INFO Channel 07/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-165-59:1557267:1557683 [1] NCCL INFO Channel 03/0 : 97[1] -> 96[0] via P2P/IPC
+ip-26-0-168-34:272151:272571 [1] NCCL INFO Channel 10/0 : 113[1] -> 112[0] via P2P/IPC
+ip-26-0-161-78:114514:114932 [2] NCCL INFO Channel 14/0 : 66[2] -> 65[1] via P2P/IPC
+ip-26-0-168-34:272156:272570 [6] NCCL INFO Channel 06/0 : 111[7] -> 118[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32709:33129 [5] NCCL INFO Channel 04/0 : 29[5] -> 36[4] [send] via NET/Libfabric/4(28)/GDRDMA
+ip-26-0-168-34:272157:272572 [7] NCCL INFO Channel 14/0 : 119[7] -> 126[6] [send] via NET/Libfabric/6(118)/GDRDMA
+ip-26-0-168-34:272156:272570 [6] NCCL INFO Channel 14/0 : 111[7] -> 118[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272151:272571 [1] NCCL INFO Channel 11/0 : 113[1] -> 112[0] via P2P/IPC
+ip-26-0-160-225:124166:124583 [6] NCCL INFO Channel 06/0 : 6[6] -> 3[3] via P2P/IPC
+ip-26-0-161-123:32709:33129 [5] NCCL INFO Channel 12/0 : 29[5] -> 36[4] [send] via NET/Libfabric/4(28)/GDRDMA
+ip-26-0-160-242:1364907:1365329 [5] NCCL INFO Channel 05/0 : 13[5] -> 11[3] via P2P/IPC
+ip-26-0-165-59:1557270:1557688 [4] NCCL INFO Channel 07/0 : 100[4] -> 111[7] [send] via NET/Libfabric/7(103)/GDRDMA
+ip-26-0-165-164:1227465:1227881 [2] NCCL INFO Channel 12/0 : 82[2] -> 81[1] via P2P/IPC
+ip-26-0-161-103:96889:97309 [1] NCCL INFO Channel 09/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-165-164:1227464:1227884 [1] NCCL INFO Channel 07/0 : 81[1] -> 80[0] via P2P/IPC
+ip-26-0-165-59:1557270:1557688 [4] NCCL INFO Channel 15/0 : 100[4] -> 111[7] [send] via NET/Libfabric/7(103)/GDRDMA
+ip-26-0-168-34:272151:272571 [1] NCCL INFO Channel 12/0 : 113[1] -> 112[0] via P2P/IPC
+ip-26-0-165-59:1557267:1557683 [1] NCCL INFO Channel 04/0 : 97[1] -> 96[0] via P2P/IPC
+ip-26-0-165-202:153564:153984 [2] NCCL INFO Channel 04/0 : 90[2] -> 89[1] via P2P/IPC
+ip-26-0-161-78:114514:114932 [2] NCCL INFO Channel 15/0 : 66[2] -> 65[1] via P2P/IPC
+ip-26-0-161-178:2815965:2816395 [2] NCCL INFO Channel 02/0 : 58[2] -> 57[1] via P2P/IPC
+ip-26-0-161-103:96894:97312 [6] NCCL INFO Channel 05/0 : 22[6] -> 29[5] [send] via NET/Libfabric/5(21)/GDRDMA
+ip-26-0-161-142:62887:63305 [6] NCCL INFO Channel 06/0 : 39[7] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:124162:124587 [2] NCCL INFO Channel 06/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-168-34:272154:272569 [4] NCCL INFO Channel 07/0 : 116[4] -> 127[7] [send] via NET/Libfabric/7(119)/GDRDMA
+ip-26-0-161-103:96894:97312 [6] NCCL INFO Channel 13/0 : 22[6] -> 29[5] [send] via NET/Libfabric/5(21)/GDRDMA
+ip-26-0-161-123:32710:33125 [6] NCCL INFO Channel 06/0 : 30[6] -> 27[3] via P2P/IPC
+ip-26-0-165-59:1557271:1557690 [5] NCCL INFO Channel 05/0 : 94[6] -> 101[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557267:1557683 [1] NCCL INFO Channel 05/0 : 97[1] -> 96[0] via P2P/IPC
+ip-26-0-168-52:2722049:2722467 [4] NCCL INFO Channel 03/0 : 124[4] -> 3[3] [send] via NET/Libfabric/3(123)/GDRDMA
+ip-26-0-161-142:62888:63302 [7] NCCL INFO Channel 14/0 : 47[7] -> 54[6] [send] via NET/Libfabric/6(46)/GDRDMA
+ip-26-0-168-34:272154:272569 [4] NCCL INFO Channel 15/0 : 116[4] -> 127[7] [send] via NET/Libfabric/7(119)/GDRDMA
+ip-26-0-161-103:96889:97309 [1] NCCL INFO Channel 10/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-165-164:1227465:1227881 [2] NCCL INFO Channel 13/0 : 82[2] -> 81[1] via P2P/IPC
+ip-26-0-165-202:153567:153979 [5] NCCL INFO Channel 13/0 : 93[5] -> 91[3] via P2P/IPC
+ip-26-0-161-78:114516:114934 [4] NCCL INFO Channel 03/0 : 68[4] -> 75[3] [send] via NET/Libfabric/3(67)/GDRDMA
+ip-26-0-160-242:1364904:1365335 [2] NCCL INFO Channel 10/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-160-225:124166:124583 [6] NCCL INFO Channel 14/0 : 6[6] -> 3[3] via P2P/IPC
+ip-26-0-161-142:62887:63305 [6] NCCL INFO Channel 14/0 : 39[7] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272151:272571 [1] NCCL INFO Channel 13/0 : 113[1] -> 112[0] via P2P/IPC
+ip-26-0-165-164:1227464:1227884 [1] NCCL INFO Channel 09/0 : 81[1] -> 80[0] via P2P/IPC
+ip-26-0-168-52:2722049:2722467 [4] NCCL INFO Channel 11/0 : 124[4] -> 3[3] [send] via NET/Libfabric/3(123)/GDRDMA
+ip-26-0-161-142:62885:63304 [4] NCCL INFO Channel 04/0 : 37[5] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557271:1557690 [5] NCCL INFO Channel 13/0 : 94[6] -> 101[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235650:236065 [2] NCCL INFO Channel 06/0 : 106[2] -> 105[1] via P2P/IPC
+ip-26-0-161-178:2815965:2816395 [2] NCCL INFO Channel 03/0 : 58[2] -> 57[1] via P2P/IPC
+ip-26-0-160-242:1364903:1365334 [1] NCCL INFO Channel 06/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-165-202:153564:153984 [2] NCCL INFO Channel 05/0 : 90[2] -> 89[1] via P2P/IPC
+ip-26-0-161-78:114516:114934 [4] NCCL INFO Channel 11/0 : 68[4] -> 75[3] [send] via NET/Libfabric/3(67)/GDRDMA
+ip-26-0-161-103:96892:97308 [4] NCCL INFO Channel 07/0 : 20[4] -> 31[7] [send] via NET/Libfabric/7(23)/GDRDMA
+ip-26-0-161-142:62885:63304 [4] NCCL INFO Channel 12/0 : 37[5] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557268:1557686 [2] NCCL INFO Channel 10/0 : 98[2] -> 97[1] via P2P/IPC
+ip-26-0-165-59:1557272:1557687 [6] NCCL INFO Channel 05/0 : 102[6] -> 109[5] [send] via NET/Libfabric/5(101)/GDRDMA
+ip-26-0-165-59:1557267:1557683 [1] NCCL INFO Channel 06/0 : 97[1] -> 96[0] via P2P/IPC
+ip-26-0-165-59:1557272:1557687 [6] NCCL INFO Channel 13/0 : 102[6] -> 109[5] [send] via NET/Libfabric/5(101)/GDRDMA
+ip-26-0-160-225:124162:124587 [2] NCCL INFO Channel 07/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-103:96889:97309 [1] NCCL INFO Channel 11/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-165-59:1557271:1557690 [5] NCCL INFO Channel 04/0 : 101[5] -> 108[4] [send] via NET/Libfabric/4(100)/GDRDMA
+ip-26-0-165-164:1227465:1227881 [2] NCCL INFO Channel 14/0 : 82[2] -> 81[1] via P2P/IPC
+ip-26-0-161-123:32705:33128 [1] NCCL INFO Channel 01/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-161-103:96890:97313 [2] NCCL INFO Channel 02/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-168-34:272151:272571 [1] NCCL INFO Channel 14/0 : 113[1] -> 112[0] via P2P/IPC
+ip-26-0-165-164:1227464:1227884 [1] NCCL INFO Channel 10/0 : 81[1] -> 80[0] via P2P/IPC
+ip-26-0-161-178:2815965:2816395 [2] NCCL INFO Channel 04/0 : 58[2] -> 57[1] via P2P/IPC
+ip-26-0-160-242:1364907:1365329 [5] NCCL INFO Channel 13/0 : 13[5] -> 11[3] via P2P/IPC
+ip-26-0-161-142:62885:63304 [4] NCCL INFO Channel 03/0 : 44[4] -> 51[3] [send] via NET/Libfabric/3(43)/GDRDMA
+ip-26-0-165-59:1557271:1557690 [5] NCCL INFO Channel 12/0 : 101[5] -> 108[4] [send] via NET/Libfabric/4(100)/GDRDMA
+ip-26-0-161-103:96894:97312 [6] NCCL INFO Channel 06/0 : 22[6] -> 19[3] via P2P/IPC
+ip-26-0-161-123:32709:33129 [5] NCCL INFO Channel 05/0 : 29[5] -> 27[3] via P2P/IPC
+ip-26-0-161-78:114513:114933 [1] NCCL INFO Channel 01/0 : 65[1] -> 64[0] via P2P/IPC
+ip-26-0-165-202:153564:153984 [2] NCCL INFO Channel 06/0 : 90[2] -> 89[1] via P2P/IPC
+ip-26-0-161-123:32710:33125 [6] NCCL INFO Channel 14/0 : 30[6] -> 27[3] via P2P/IPC
+ip-26-0-161-142:62885:63304 [4] NCCL INFO Channel 11/0 : 44[4] -> 51[3] [send] via NET/Libfabric/3(43)/GDRDMA
+ip-26-0-165-59:1557268:1557686 [2] NCCL INFO Channel 11/0 : 98[2] -> 97[1] via P2P/IPC
+ip-26-0-165-59:1557267:1557683 [1] NCCL INFO Channel 07/0 : 97[1] -> 96[0] via P2P/IPC
+ip-26-0-161-78:114513:114933 [1] NCCL INFO Channel 02/0 : 65[1] -> 64[0] via P2P/IPC
+ip-26-0-160-242:1364904:1365335 [2] NCCL INFO Channel 11/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-160-225:124162:124587 [2] NCCL INFO Channel 10/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-168-30:235650:236065 [2] NCCL INFO Channel 07/0 : 106[2] -> 105[1] via P2P/IPC
+ip-26-0-160-242:1364903:1365334 [1] NCCL INFO Channel 07/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-165-164:1227465:1227881 [2] NCCL INFO Channel 15/0 : 82[2] -> 81[1] via P2P/IPC
+ip-26-0-165-59:1557272:1557687 [6] NCCL INFO Channel 06/0 : 102[6] -> 99[3] via P2P/IPC
+ip-26-0-168-34:272151:272571 [1] NCCL INFO Channel 15/0 : 113[1] -> 112[0] via P2P/IPC
+ip-26-0-161-178:2815965:2816395 [2] NCCL INFO Channel 05/0 : 58[2] -> 57[1] via P2P/IPC
+ip-26-0-161-123:32706:33122 [2] NCCL INFO Channel 02/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-161-78:114517:114935 [5] NCCL INFO Channel 05/0 : 62[6] -> 69[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235653:236067 [5] NCCL INFO Channel 05/0 : 109[5] -> 107[3] via P2P/IPC
+ip-26-0-165-164:1227464:1227884 [1] NCCL INFO Channel 11/0 : 81[1] -> 80[0] via P2P/IPC
+ip-26-0-165-202:153568:153982 [6] NCCL INFO Channel 06/0 : 94[6] -> 91[3] via P2P/IPC
+ip-26-0-165-202:153564:153984 [2] NCCL INFO Channel 07/0 : 90[2] -> 89[1] via P2P/IPC
+ip-26-0-161-78:114513:114933 [1] NCCL INFO Channel 03/0 : 65[1] -> 64[0] via P2P/IPC
+ip-26-0-161-78:114517:114935 [5] NCCL INFO Channel 13/0 : 62[6] -> 69[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62886:63303 [5] NCCL INFO Channel 05/0 : 38[6] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62886:63303 [5] NCCL INFO Channel 13/0 : 38[6] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557271:1557690 [5] NCCL INFO Channel 05/0 : 101[5] -> 99[3] via P2P/IPC
+ip-26-0-161-178:2815965:2816395 [2] NCCL INFO Channel 06/0 : 58[2] -> 57[1] via P2P/IPC
+ip-26-0-161-123:32705:33128 [1] NCCL INFO Channel 02/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-161-142:62887:63305 [6] NCCL INFO Channel 05/0 : 46[6] -> 53[5] [send] via NET/Libfabric/5(45)/GDRDMA
+ip-26-0-161-78:114513:114933 [1] NCCL INFO Channel 04/0 : 65[1] -> 64[0] via P2P/IPC
+ip-26-0-165-59:1557267:1557683 [1] NCCL INFO Channel 09/0 : 97[1] -> 96[0] via P2P/IPC
+ip-26-0-165-59:1557268:1557686 [2] NCCL INFO Channel 12/0 : 98[2] -> 97[1] via P2P/IPC
+ip-26-0-161-103:96889:97309 [1] NCCL INFO Channel 12/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-165-164:1227464:1227884 [1] NCCL INFO Channel 12/0 : 81[1] -> 80[0] via P2P/IPC
+ip-26-0-161-78:114517:114935 [5] NCCL INFO Channel 04/0 : 69[5] -> 76[4] [send] via NET/Libfabric/4(68)/GDRDMA
+ip-26-0-161-103:96890:97313 [2] NCCL INFO Channel 03/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-161-138:96564:97009 [1] NCCL INFO Channel 01/0 : 33[1] -> 32[0] via P2P/IPC
+ip-26-0-161-78:114517:114935 [5] NCCL INFO Channel 12/0 : 69[5] -> 76[4] [send] via NET/Libfabric/4(68)/GDRDMA
+ip-26-0-160-242:1364904:1365335 [2] NCCL INFO Channel 12/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-161-142:62887:63305 [6] NCCL INFO Channel 13/0 : 46[6] -> 53[5] [send] via NET/Libfabric/5(45)/GDRDMA
+ip-26-0-168-34:272155:272574 [5] NCCL INFO Channel 05/0 : 110[6] -> 117[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124162:124587 [2] NCCL INFO Channel 11/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-242:1364906:1365331 [4] NCCL INFO Channel 07/0 : 12[4] -> 23[7] [send] via NET/Libfabric/7(15)/GDRDMA
+ip-26-0-160-242:1364903:1365334 [1] NCCL INFO Channel 09/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-161-103:96894:97312 [6] NCCL INFO Channel 14/0 : 22[6] -> 19[3] via P2P/IPC
+ip-26-0-161-142:62886:63303 [5] NCCL INFO Channel 04/0 : 45[5] -> 52[4] [send] via NET/Libfabric/4(44)/GDRDMA
+ip-26-0-168-34:272155:272574 [5] NCCL INFO Channel 13/0 : 110[6] -> 117[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114513:114933 [1] NCCL INFO Channel 05/0 : 65[1] -> 64[0] via P2P/IPC
+ip-26-0-161-123:32709:33129 [5] NCCL INFO Channel 13/0 : 29[5] -> 27[3] via P2P/IPC
+ip-26-0-161-142:62886:63303 [5] NCCL INFO Channel 12/0 : 45[5] -> 52[4] [send] via NET/Libfabric/4(44)/GDRDMA
+ip-26-0-168-34:272156:272570 [6] NCCL INFO Channel 05/0 : 118[6] -> 125[5] [send] via NET/Libfabric/5(117)/GDRDMA
+ip-26-0-161-178:2815965:2816395 [2] NCCL INFO Channel 07/0 : 58[2] -> 57[1] via P2P/IPC
+ip-26-0-161-138:96564:97009 [1] NCCL INFO Channel 02/0 : 33[1] -> 32[0] via P2P/IPC
+ip-26-0-165-202:153564:153984 [2] NCCL INFO Channel 10/0 : 90[2] -> 89[1] via P2P/IPC
+ip-26-0-165-164:1227464:1227884 [1] NCCL INFO Channel 13/0 : 81[1] -> 80[0] via P2P/IPC
+ip-26-0-161-123:32706:33122 [2] NCCL INFO Channel 03/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-165-202:153568:153982 [6] NCCL INFO Channel 14/0 : 94[6] -> 91[3] via P2P/IPC
+ip-26-0-168-34:272156:272570 [6] NCCL INFO Channel 13/0 : 118[6] -> 125[5] [send] via NET/Libfabric/5(117)/GDRDMA
+ip-26-0-168-34:272155:272574 [5] NCCL INFO Channel 04/0 : 117[5] -> 124[4] [send] via NET/Libfabric/4(116)/GDRDMA
+ip-26-0-161-78:114513:114933 [1] NCCL INFO Channel 06/0 : 65[1] -> 64[0] via P2P/IPC
+ip-26-0-165-59:1557272:1557687 [6] NCCL INFO Channel 14/0 : 102[6] -> 99[3] via P2P/IPC
+ip-26-0-161-138:96564:97009 [1] NCCL INFO Channel 03/0 : 33[1] -> 32[0] via P2P/IPC
+ip-26-0-168-34:272155:272574 [5] NCCL INFO Channel 12/0 : 117[5] -> 124[4] [send] via NET/Libfabric/4(116)/GDRDMA
+ip-26-0-161-103:96889:97309 [1] NCCL INFO Channel 13/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-161-123:32705:33128 [1] NCCL INFO Channel 03/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-165-164:1227464:1227884 [1] NCCL INFO Channel 14/0 : 81[1] -> 80[0] via P2P/IPC
+ip-26-0-160-242:1364904:1365335 [2] NCCL INFO Channel 13/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-160-225:124162:124587 [2] NCCL INFO Channel 12/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-178:2815965:2816395 [2] NCCL INFO Channel 10/0 : 58[2] -> 57[1] via P2P/IPC
+ip-26-0-161-103:96890:97313 [2] NCCL INFO Channel 04/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-160-242:1364903:1365334 [1] NCCL INFO Channel 10/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-161-138:96567:97008 [4] NCCL INFO Channel 07/0 : 36[4] -> 47[7] [send] via NET/Libfabric/7(39)/GDRDMA
+ip-26-0-161-138:96569:97006 [6] NCCL INFO Channel 06/0 : 38[6] -> 35[3] via P2P/IPC
+ip-26-0-161-78:114517:114935 [5] NCCL INFO Channel 05/0 : 69[5] -> 67[3] via P2P/IPC
+ip-26-0-165-59:1557267:1557683 [1] NCCL INFO Channel 10/0 : 97[1] -> 96[0] via P2P/IPC
+ip-26-0-161-78:114513:114933 [1] NCCL INFO Channel 07/0 : 65[1] -> 64[0] via P2P/IPC
+ip-26-0-161-123:32706:33122 [2] NCCL INFO Channel 04/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-161-138:96564:97009 [1] NCCL INFO Channel 04/0 : 33[1] -> 32[0] via P2P/IPC
+ip-26-0-165-202:153564:153984 [2] NCCL INFO Channel 11/0 : 90[2] -> 89[1] via P2P/IPC
+ip-26-0-165-59:1557271:1557690 [5] NCCL INFO Channel 13/0 : 101[5] -> 99[3] via P2P/IPC
+ip-26-0-165-59:1557268:1557686 [2] NCCL INFO Channel 13/0 : 98[2] -> 97[1] via P2P/IPC
+ip-26-0-165-164:1227464:1227884 [1] NCCL INFO Channel 15/0 : 81[1] -> 80[0] via P2P/IPC
+ip-26-0-161-178:2815965:2816395 [2] NCCL INFO Channel 11/0 : 58[2] -> 57[1] via P2P/IPC
+ip-26-0-161-103:96889:97309 [1] NCCL INFO Channel 14/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-161-123:32705:33128 [1] NCCL INFO Channel 04/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-168-34:272156:272570 [6] NCCL INFO Channel 06/0 : 118[6] -> 115[3] via P2P/IPC
+ip-26-0-168-52:2722050:2722471 [5] NCCL INFO Channel 05/0 : 125[5] -> 123[3] via P2P/IPC
+ip-26-0-161-103:96890:97313 [2] NCCL INFO Channel 05/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-160-225:124162:124587 [2] NCCL INFO Channel 13/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-168-30:235654:236070 [6] NCCL INFO Channel 06/0 : 110[6] -> 107[3] via P2P/IPC
+ip-26-0-161-123:32706:33122 [2] NCCL INFO Channel 05/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-161-138:96569:97006 [6] NCCL INFO Channel 14/0 : 38[6] -> 35[3] via P2P/IPC
+ip-26-0-160-242:1364904:1365335 [2] NCCL INFO Channel 14/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-165-202:153564:153984 [2] NCCL INFO Channel 12/0 : 90[2] -> 89[1] via P2P/IPC
+ip-26-0-161-138:96564:97009 [1] NCCL INFO Channel 05/0 : 33[1] -> 32[0] via P2P/IPC
+ip-26-0-168-34:272155:272574 [5] NCCL INFO Channel 05/0 : 117[5] -> 115[3] via P2P/IPC
+ip-26-0-160-242:1364903:1365334 [1] NCCL INFO Channel 11/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-161-142:62883:63301 [2] NCCL INFO Channel 02/0 : 42[2] -> 41[1] via P2P/IPC
+ip-26-0-161-78:114513:114933 [1] NCCL INFO Channel 09/0 : 65[1] -> 64[0] via P2P/IPC
+ip-26-0-161-178:2815965:2816395 [2] NCCL INFO Channel 12/0 : 58[2] -> 57[1] via P2P/IPC
+ip-26-0-161-153:2636088:2636510 [4] NCCL INFO Channel 04/0 : 45[5] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636085:2636509 [1] NCCL INFO Channel 01/0 : 49[1] -> 48[0] via P2P/IPC
+ip-26-0-161-78:114517:114935 [5] NCCL INFO Channel 13/0 : 69[5] -> 67[3] via P2P/IPC
+ip-26-0-165-59:1557267:1557683 [1] NCCL INFO Channel 11/0 : 97[1] -> 96[0] via P2P/IPC
+ip-26-0-165-59:1557268:1557686 [2] NCCL INFO Channel 14/0 : 98[2] -> 97[1] via P2P/IPC
+ip-26-0-160-225:124162:124587 [2] NCCL INFO Channel 14/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-103:96889:97309 [1] NCCL INFO Channel 15/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-161-123:32705:33128 [1] NCCL INFO Channel 05/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-165-202:153564:153984 [2] NCCL INFO Channel 13/0 : 90[2] -> 89[1] via P2P/IPC
+ip-26-0-161-103:96890:97313 [2] NCCL INFO Channel 06/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-161-123:32706:33122 [2] NCCL INFO Channel 06/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-161-138:96568:97007 [5] NCCL INFO Channel 05/0 : 37[5] -> 35[3] via P2P/IPC
+ip-26-0-161-153:2636088:2636510 [4] NCCL INFO Channel 12/0 : 45[5] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114513:114933 [1] NCCL INFO Channel 10/0 : 65[1] -> 64[0] via P2P/IPC
+ip-26-0-161-138:96564:97009 [1] NCCL INFO Channel 06/0 : 33[1] -> 32[0] via P2P/IPC
+ip-26-0-161-142:62883:63301 [2] NCCL INFO Channel 03/0 : 42[2] -> 41[1] via P2P/IPC
+ip-26-0-160-242:1364904:1365335 [2] NCCL INFO Channel 15/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-168-30:235653:236067 [5] NCCL INFO Channel 13/0 : 109[5] -> 107[3] via P2P/IPC
+ip-26-0-168-52:2722046:2722466 [1] NCCL INFO Channel 01/0 : 121[1] -> 120[0] via P2P/IPC
+ip-26-0-161-178:2815964:2816401 [1] NCCL INFO Channel 01/0 : 57[1] -> 56[0] via P2P/IPC
+ip-26-0-160-242:1364903:1365334 [1] NCCL INFO Channel 12/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-161-153:2636085:2636509 [1] NCCL INFO Channel 02/0 : 49[1] -> 48[0] via P2P/IPC
+ip-26-0-168-30:235654:236070 [6] NCCL INFO Channel 14/0 : 110[6] -> 107[3] via P2P/IPC
+ip-26-0-168-30:235650:236065 [2] NCCL INFO Channel 10/0 : 106[2] -> 105[1] via P2P/IPC
+ip-26-0-168-52:2722050:2722471 [5] NCCL INFO Channel 13/0 : 125[5] -> 123[3] via P2P/IPC
+ip-26-0-161-178:2815965:2816395 [2] NCCL INFO Channel 13/0 : 58[2] -> 57[1] via P2P/IPC
+ip-26-0-160-225:124162:124587 [2] NCCL INFO Channel 15/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-78:114513:114933 [1] NCCL INFO Channel 11/0 : 65[1] -> 64[0] via P2P/IPC
+ip-26-0-165-59:1557267:1557683 [1] NCCL INFO Channel 12/0 : 97[1] -> 96[0] via P2P/IPC
+ip-26-0-161-123:32705:33128 [1] NCCL INFO Channel 06/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-165-59:1557268:1557686 [2] NCCL INFO Channel 15/0 : 98[2] -> 97[1] via P2P/IPC
+ip-26-0-161-103:96890:97313 [2] NCCL INFO Channel 07/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-168-30:235652:236066 [4] NCCL INFO Channel 07/0 : 108[4] -> 119[7] [send] via NET/Libfabric/7(111)/GDRDMA
+ip-26-0-161-123:32706:33122 [2] NCCL INFO Channel 07/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-161-153:2636086:2636507 [2] NCCL INFO Channel 02/0 : 50[2] -> 49[1] via P2P/IPC
+ip-26-0-165-131:1630206:1630647 [1] NCCL INFO Channel 01/0 : 73[1] -> 72[0] via P2P/IPC
+ip-26-0-161-142:62883:63301 [2] NCCL INFO Channel 04/0 : 42[2] -> 41[1] via P2P/IPC
+ip-26-0-168-52:2722046:2722466 [1] NCCL INFO Channel 02/0 : 121[1] -> 120[0] via P2P/IPC
+ip-26-0-161-138:96564:97009 [1] NCCL INFO Channel 07/0 : 33[1] -> 32[0] via P2P/IPC
+ip-26-0-161-153:2636088:2636510 [4] NCCL INFO Channel 03/0 : 52[4] -> 59[3] [send] via NET/Libfabric/3(51)/GDRDMA
+ip-26-0-161-153:2636085:2636509 [1] NCCL INFO Channel 03/0 : 49[1] -> 48[0] via P2P/IPC
+ip-26-0-161-78:114513:114933 [1] NCCL INFO Channel 12/0 : 65[1] -> 64[0] via P2P/IPC
+ip-26-0-161-178:2815964:2816401 [1] NCCL INFO Channel 02/0 : 57[1] -> 56[0] via P2P/IPC
+ip-26-0-161-138:96568:97007 [5] NCCL INFO Channel 13/0 : 37[5] -> 35[3] via P2P/IPC
+ip-26-0-161-178:2815965:2816395 [2] NCCL INFO Channel 14/0 : 58[2] -> 57[1] via P2P/IPC
+ip-26-0-160-242:1364903:1365334 [1] NCCL INFO Channel 13/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-168-52:2722046:2722466 [1] NCCL INFO Channel 03/0 : 121[1] -> 120[0] via P2P/IPC
+ip-26-0-161-142:62883:63301 [2] NCCL INFO Channel 05/0 : 42[2] -> 41[1] via P2P/IPC
+ip-26-0-165-59:1557267:1557683 [1] NCCL INFO Channel 13/0 : 97[1] -> 96[0] via P2P/IPC
+ip-26-0-161-138:96564:97009 [1] NCCL INFO Channel 09/0 : 33[1] -> 32[0] via P2P/IPC
+ip-26-0-161-123:32705:33128 [1] NCCL INFO Channel 07/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-161-153:2636086:2636507 [2] NCCL INFO Channel 03/0 : 50[2] -> 49[1] via P2P/IPC
+ip-26-0-161-78:114513:114933 [1] NCCL INFO Channel 13/0 : 65[1] -> 64[0] via P2P/IPC
+ip-26-0-161-123:32706:33122 [2] NCCL INFO Channel 10/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-168-30:235650:236065 [2] NCCL INFO Channel 11/0 : 106[2] -> 105[1] via P2P/IPC
+ip-26-0-168-34:272152:272575 [2] NCCL INFO Channel 02/0 : 114[2] -> 113[1] via P2P/IPC
+ip-26-0-161-178:2815964:2816401 [1] NCCL INFO Channel 03/0 : 57[1] -> 56[0] via P2P/IPC
+ip-26-0-161-142:62883:63301 [2] NCCL INFO Channel 06/0 : 42[2] -> 41[1] via P2P/IPC
+ip-26-0-165-59:1557267:1557683 [1] NCCL INFO Channel 14/0 : 97[1] -> 96[0] via P2P/IPC
+ip-26-0-161-178:2815965:2816395 [2] NCCL INFO Channel 15/0 : 58[2] -> 57[1] via P2P/IPC
+ip-26-0-165-131:1630206:1630647 [1] NCCL INFO Channel 02/0 : 73[1] -> 72[0] via P2P/IPC
+ip-26-0-160-242:1364903:1365334 [1] NCCL INFO Channel 14/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-161-138:96564:97009 [1] NCCL INFO Channel 10/0 : 33[1] -> 32[0] via P2P/IPC
+ip-26-0-161-78:114513:114933 [1] NCCL INFO Channel 14/0 : 65[1] -> 64[0] via P2P/IPC
+ip-26-0-161-153:2636090:2636505 [6] NCCL INFO Channel 06/0 : 47[7] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32705:33128 [1] NCCL INFO Channel 09/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-161-142:62885:63304 [4] NCCL INFO Channel 07/0 : 44[4] -> 55[7] [send] via NET/Libfabric/7(47)/GDRDMA
+ip-26-0-165-59:1557267:1557683 [1] NCCL INFO Channel 15/0 : 97[1] -> 96[0] via P2P/IPC
+ip-26-0-161-123:32706:33122 [2] NCCL INFO Channel 11/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-161-153:2636085:2636509 [1] NCCL INFO Channel 04/0 : 49[1] -> 48[0] via P2P/IPC
+ip-26-0-168-30:235650:236065 [2] NCCL INFO Channel 12/0 : 106[2] -> 105[1] via P2P/IPC
+ip-26-0-161-138:96564:97009 [1] NCCL INFO Channel 11/0 : 33[1] -> 32[0] via P2P/IPC
+ip-26-0-161-142:62883:63301 [2] NCCL INFO Channel 07/0 : 42[2] -> 41[1] via P2P/IPC
+ip-26-0-161-78:114513:114933 [1] NCCL INFO Channel 15/0 : 65[1] -> 64[0] via P2P/IPC
+ip-26-0-161-153:2636091:2636512 [7] NCCL INFO Channel 14/0 : 55[7] -> 62[6] [send] via NET/Libfabric/6(54)/GDRDMA
+ip-26-0-161-178:2815964:2816401 [1] NCCL INFO Channel 04/0 : 57[1] -> 56[0] via P2P/IPC
+ip-26-0-161-153:2636090:2636505 [6] NCCL INFO Channel 14/0 : 47[7] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636086:2636507 [2] NCCL INFO Channel 04/0 : 50[2] -> 49[1] via P2P/IPC
+ip-26-0-165-131:1630206:1630647 [1] NCCL INFO Channel 03/0 : 73[1] -> 72[0] via P2P/IPC
+ip-26-0-161-142:62885:63304 [4] NCCL INFO Channel 15/0 : 44[4] -> 55[7] [send] via NET/Libfabric/7(47)/GDRDMA
+ip-26-0-165-202:153564:153984 [2] NCCL INFO Channel 14/0 : 90[2] -> 89[1] via P2P/IPC
+ip-26-0-160-242:1364903:1365334 [1] NCCL INFO Channel 15/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-161-142:62883:63301 [2] NCCL INFO Channel 10/0 : 42[2] -> 41[1] via P2P/IPC
+ip-26-0-161-138:96564:97009 [1] NCCL INFO Channel 12/0 : 33[1] -> 32[0] via P2P/IPC
+ip-26-0-161-178:2815964:2816401 [1] NCCL INFO Channel 05/0 : 57[1] -> 56[0] via P2P/IPC
+ip-26-0-168-30:235650:236065 [2] NCCL INFO Channel 13/0 : 106[2] -> 105[1] via P2P/IPC
+ip-26-0-161-123:32705:33128 [1] NCCL INFO Channel 10/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-161-123:32706:33122 [2] NCCL INFO Channel 12/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-161-142:62883:63301 [2] NCCL INFO Channel 11/0 : 42[2] -> 41[1] via P2P/IPC
+ip-26-0-165-131:1630206:1630647 [1] NCCL INFO Channel 04/0 : 73[1] -> 72[0] via P2P/IPC
+ip-26-0-161-153:2636085:2636509 [1] NCCL INFO Channel 05/0 : 49[1] -> 48[0] via P2P/IPC
+ip-26-0-161-178:2815964:2816401 [1] NCCL INFO Channel 06/0 : 57[1] -> 56[0] via P2P/IPC
+ip-26-0-161-78:114518:114937 [6] NCCL INFO Channel 06/0 : 63[7] -> 70[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636088:2636510 [4] NCCL INFO Channel 11/0 : 52[4] -> 59[3] [send] via NET/Libfabric/3(51)/GDRDMA
+ip-26-0-161-178:2815969:2816398 [6] NCCL INFO Channel 06/0 : 62[6] -> 59[3] via P2P/IPC
+ip-26-0-161-78:114518:114937 [6] NCCL INFO Channel 14/0 : 63[7] -> 70[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96890:97313 [2] NCCL INFO Channel 10/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-161-153:2636089:2636508 [5] NCCL INFO Channel 05/0 : 46[6] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62883:63301 [2] NCCL INFO Channel 12/0 : 42[2] -> 41[1] via P2P/IPC
+ip-26-0-165-131:1630207:1630648 [2] NCCL INFO Channel 02/0 : 74[2] -> 73[1] via P2P/IPC
+ip-26-0-161-153:2636086:2636507 [2] NCCL INFO Channel 05/0 : 50[2] -> 49[1] via P2P/IPC
+ip-26-0-168-52:2722046:2722466 [1] NCCL INFO Channel 04/0 : 121[1] -> 120[0] via P2P/IPC
+ip-26-0-161-78:114518:114937 [6] NCCL INFO Channel 05/0 : 70[6] -> 77[5] [send] via NET/Libfabric/5(69)/GDRDMA
+ip-26-0-161-178:2815964:2816401 [1] NCCL INFO Channel 07/0 : 57[1] -> 56[0] via P2P/IPC
+ip-26-0-165-131:1630206:1630647 [1] NCCL INFO Channel 05/0 : 73[1] -> 72[0] via P2P/IPC
+ip-26-0-161-153:2636089:2636508 [5] NCCL INFO Channel 13/0 : 46[6] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636090:2636505 [6] NCCL INFO Channel 05/0 : 54[6] -> 61[5] [send] via NET/Libfabric/5(53)/GDRDMA
+ip-26-0-161-78:114518:114937 [6] NCCL INFO Channel 13/0 : 70[6] -> 77[5] [send] via NET/Libfabric/5(69)/GDRDMA
+ip-26-0-161-142:62883:63301 [2] NCCL INFO Channel 13/0 : 42[2] -> 41[1] via P2P/IPC
+ip-26-0-161-123:32705:33128 [1] NCCL INFO Channel 11/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-168-52:2722046:2722466 [1] NCCL INFO Channel 05/0 : 121[1] -> 120[0] via P2P/IPC
+ip-26-0-161-153:2636085:2636509 [1] NCCL INFO Channel 06/0 : 49[1] -> 48[0] via P2P/IPC
+ip-26-0-161-142:62886:63303 [5] NCCL INFO Channel 05/0 : 45[5] -> 43[3] via P2P/IPC
+ip-26-0-161-123:32706:33122 [2] NCCL INFO Channel 13/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-161-153:2636090:2636505 [6] NCCL INFO Channel 13/0 : 54[6] -> 61[5] [send] via NET/Libfabric/5(53)/GDRDMA
+ip-26-0-161-153:2636089:2636508 [5] NCCL INFO Channel 04/0 : 53[5] -> 60[4] [send] via NET/Libfabric/4(52)/GDRDMA
+ip-26-0-161-178:2815969:2816398 [6] NCCL INFO Channel 14/0 : 62[6] -> 59[3] via P2P/IPC
+ip-26-0-161-153:2636086:2636507 [2] NCCL INFO Channel 06/0 : 50[2] -> 49[1] via P2P/IPC
+ip-26-0-165-131:1630207:1630648 [2] NCCL INFO Channel 03/0 : 74[2] -> 73[1] via P2P/IPC
+ip-26-0-168-52:2722046:2722466 [1] NCCL INFO Channel 06/0 : 121[1] -> 120[0] via P2P/IPC
+ip-26-0-161-178:2815964:2816401 [1] NCCL INFO Channel 09/0 : 57[1] -> 56[0] via P2P/IPC
+ip-26-0-161-153:2636089:2636508 [5] NCCL INFO Channel 12/0 : 53[5] -> 60[4] [send] via NET/Libfabric/4(52)/GDRDMA
+ip-26-0-161-142:62883:63301 [2] NCCL INFO Channel 14/0 : 42[2] -> 41[1] via P2P/IPC
+ip-26-0-165-131:1630206:1630647 [1] NCCL INFO Channel 06/0 : 73[1] -> 72[0] via P2P/IPC
+ip-26-0-161-153:2636085:2636509 [1] NCCL INFO Channel 07/0 : 49[1] -> 48[0] via P2P/IPC
+ip-26-0-165-202:153564:153984 [2] NCCL INFO Channel 15/0 : 90[2] -> 89[1] via P2P/IPC
+ip-26-0-168-52:2722046:2722466 [1] NCCL INFO Channel 07/0 : 121[1] -> 120[0] via P2P/IPC
+ip-26-0-161-78:114518:114937 [6] NCCL INFO Channel 06/0 : 70[6] -> 67[3] via P2P/IPC
+ip-26-0-161-123:32705:33128 [1] NCCL INFO Channel 12/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-165-131:1630210:1630653 [5] NCCL INFO Channel 05/0 : 77[5] -> 75[3] via P2P/IPC
+ip-26-0-161-123:32706:33122 [2] NCCL INFO Channel 14/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-161-142:62887:63305 [6] NCCL INFO Channel 06/0 : 46[6] -> 43[3] via P2P/IPC
+ip-26-0-161-153:2636086:2636507 [2] NCCL INFO Channel 07/0 : 50[2] -> 49[1] via P2P/IPC
+ip-26-0-161-142:62883:63301 [2] NCCL INFO Channel 15/0 : 42[2] -> 41[1] via P2P/IPC
+ip-26-0-161-178:2815968:2816394 [5] NCCL INFO Channel 05/0 : 61[5] -> 59[3] via P2P/IPC
+ip-26-0-161-138:96564:97009 [1] NCCL INFO Channel 13/0 : 33[1] -> 32[0] via P2P/IPC
+ip-26-0-168-30:235650:236065 [2] NCCL INFO Channel 14/0 : 106[2] -> 105[1] via P2P/IPC
+ip-26-0-161-153:2636090:2636505 [6] NCCL INFO Channel 06/0 : 54[6] -> 51[3] via P2P/IPC
+ip-26-0-161-142:62886:63303 [5] NCCL INFO Channel 13/0 : 45[5] -> 43[3] via P2P/IPC
+ip-26-0-168-52:2722046:2722466 [1] NCCL INFO Channel 09/0 : 121[1] -> 120[0] via P2P/IPC
+ip-26-0-161-153:2636085:2636509 [1] NCCL INFO Channel 09/0 : 49[1] -> 48[0] via P2P/IPC
+ip-26-0-161-178:2815964:2816401 [1] NCCL INFO Channel 10/0 : 57[1] -> 56[0] via P2P/IPC
+ip-26-0-165-131:1630207:1630648 [2] NCCL INFO Channel 04/0 : 74[2] -> 73[1] via P2P/IPC
+ip-26-0-161-153:2636089:2636508 [5] NCCL INFO Channel 05/0 : 53[5] -> 51[3] via P2P/IPC
+ip-26-0-165-131:1630206:1630647 [1] NCCL INFO Channel 07/0 : 73[1] -> 72[0] via P2P/IPC
+ip-26-0-161-138:96564:97009 [1] NCCL INFO Channel 14/0 : 33[1] -> 32[0] via P2P/IPC
+ip-26-0-161-123:32705:33128 [1] NCCL INFO Channel 13/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-161-123:32706:33122 [2] NCCL INFO Channel 15/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-161-142:62887:63305 [6] NCCL INFO Channel 14/0 : 46[6] -> 43[3] via P2P/IPC
+ip-26-0-161-138:96564:97009 [1] NCCL INFO Channel 15/0 : 33[1] -> 32[0] via P2P/IPC
+ip-26-0-168-34:272156:272570 [6] NCCL INFO Channel 14/0 : 118[6] -> 115[3] via P2P/IPC
+ip-26-0-168-34:272155:272574 [5] NCCL INFO Channel 13/0 : 117[5] -> 115[3] via P2P/IPC
+ip-26-0-168-34:272152:272575 [2] NCCL INFO Channel 03/0 : 114[2] -> 113[1] via P2P/IPC
+ip-26-0-168-34:272152:272575 [2] NCCL INFO Channel 04/0 : 114[2] -> 113[1] via P2P/IPC
+ip-26-0-168-34:272152:272575 [2] NCCL INFO Channel 05/0 : 114[2] -> 113[1] via P2P/IPC
+ip-26-0-161-178:2815964:2816401 [1] NCCL INFO Channel 11/0 : 57[1] -> 56[0] via P2P/IPC
+ip-26-0-168-34:272152:272575 [2] NCCL INFO Channel 06/0 : 114[2] -> 113[1] via P2P/IPC
+ip-26-0-168-34:272152:272575 [2] NCCL INFO Channel 07/0 : 114[2] -> 113[1] via P2P/IPC
+ip-26-0-168-34:272152:272575 [2] NCCL INFO Channel 10/0 : 114[2] -> 113[1] via P2P/IPC
+ip-26-0-161-178:2815968:2816394 [5] NCCL INFO Channel 13/0 : 61[5] -> 59[3] via P2P/IPC
+ip-26-0-165-131:1630210:1630653 [5] NCCL INFO Channel 13/0 : 77[5] -> 75[3] via P2P/IPC
+ip-26-0-165-131:1630207:1630648 [2] NCCL INFO Channel 05/0 : 74[2] -> 73[1] via P2P/IPC
+ip-26-0-165-131:1630206:1630647 [1] NCCL INFO Channel 09/0 : 73[1] -> 72[0] via P2P/IPC
+ip-26-0-161-123:32705:33128 [1] NCCL INFO Channel 14/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-161-153:2636086:2636507 [2] NCCL INFO Channel 10/0 : 50[2] -> 49[1] via P2P/IPC
+ip-26-0-161-153:2636090:2636505 [6] NCCL INFO Channel 14/0 : 54[6] -> 51[3] via P2P/IPC
+ip-26-0-161-153:2636085:2636509 [1] NCCL INFO Channel 10/0 : 49[1] -> 48[0] via P2P/IPC
+ip-26-0-161-178:2815964:2816401 [1] NCCL INFO Channel 12/0 : 57[1] -> 56[0] via P2P/IPC
+ip-26-0-161-153:2636089:2636508 [5] NCCL INFO Channel 13/0 : 53[5] -> 51[3] via P2P/IPC
+ip-26-0-161-123:32705:33128 [1] NCCL INFO Channel 15/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-161-178:2815964:2816401 [1] NCCL INFO Channel 13/0 : 57[1] -> 56[0] via P2P/IPC
+ip-26-0-168-52:2722046:2722466 [1] NCCL INFO Channel 10/0 : 121[1] -> 120[0] via P2P/IPC
+ip-26-0-168-30:235650:236065 [2] NCCL INFO Channel 15/0 : 106[2] -> 105[1] via P2P/IPC
+ip-26-0-168-52:2722046:2722466 [1] NCCL INFO Channel 11/0 : 121[1] -> 120[0] via P2P/IPC
+ip-26-0-168-52:2722046:2722466 [1] NCCL INFO Channel 12/0 : 121[1] -> 120[0] via P2P/IPC
+ip-26-0-168-52:2722046:2722466 [1] NCCL INFO Channel 13/0 : 121[1] -> 120[0] via P2P/IPC
+ip-26-0-168-52:2722046:2722466 [1] NCCL INFO Channel 14/0 : 121[1] -> 120[0] via P2P/IPC
+ip-26-0-165-131:1630207:1630648 [2] NCCL INFO Channel 06/0 : 74[2] -> 73[1] via P2P/IPC
+ip-26-0-165-131:1630206:1630647 [1] NCCL INFO Channel 10/0 : 73[1] -> 72[0] via P2P/IPC
+ip-26-0-161-178:2815964:2816401 [1] NCCL INFO Channel 14/0 : 57[1] -> 56[0] via P2P/IPC
+ip-26-0-161-178:2815964:2816401 [1] NCCL INFO Channel 15/0 : 57[1] -> 56[0] via P2P/IPC
+ip-26-0-161-153:2636086:2636507 [2] NCCL INFO Channel 11/0 : 50[2] -> 49[1] via P2P/IPC
+ip-26-0-161-153:2636085:2636509 [1] NCCL INFO Channel 11/0 : 49[1] -> 48[0] via P2P/IPC
+ip-26-0-165-131:1630207:1630648 [2] NCCL INFO Channel 07/0 : 74[2] -> 73[1] via P2P/IPC
+ip-26-0-165-131:1630206:1630647 [1] NCCL INFO Channel 11/0 : 73[1] -> 72[0] via P2P/IPC
+ip-26-0-168-34:272152:272575 [2] NCCL INFO Channel 11/0 : 114[2] -> 113[1] via P2P/IPC
+ip-26-0-161-153:2636086:2636507 [2] NCCL INFO Channel 12/0 : 50[2] -> 49[1] via P2P/IPC
+ip-26-0-165-131:1630207:1630648 [2] NCCL INFO Channel 10/0 : 74[2] -> 73[1] via P2P/IPC
+ip-26-0-161-153:2636085:2636509 [1] NCCL INFO Channel 12/0 : 49[1] -> 48[0] via P2P/IPC
+ip-26-0-165-131:1630206:1630647 [1] NCCL INFO Channel 12/0 : 73[1] -> 72[0] via P2P/IPC
+ip-26-0-168-52:2722046:2722466 [1] NCCL INFO Channel 15/0 : 121[1] -> 120[0] via P2P/IPC
+ip-26-0-165-131:1630207:1630648 [2] NCCL INFO Channel 11/0 : 74[2] -> 73[1] via P2P/IPC
+ip-26-0-161-153:2636086:2636507 [2] NCCL INFO Channel 13/0 : 50[2] -> 49[1] via P2P/IPC
+ip-26-0-161-153:2636085:2636509 [1] NCCL INFO Channel 13/0 : 49[1] -> 48[0] via P2P/IPC
+ip-26-0-165-131:1630206:1630647 [1] NCCL INFO Channel 13/0 : 73[1] -> 72[0] via P2P/IPC
+ip-26-0-165-131:1630207:1630648 [2] NCCL INFO Channel 12/0 : 74[2] -> 73[1] via P2P/IPC
+ip-26-0-168-34:272152:272575 [2] NCCL INFO Channel 12/0 : 114[2] -> 113[1] via P2P/IPC
+ip-26-0-165-131:1630206:1630647 [1] NCCL INFO Channel 14/0 : 73[1] -> 72[0] via P2P/IPC
+ip-26-0-161-153:2636086:2636507 [2] NCCL INFO Channel 14/0 : 50[2] -> 49[1] via P2P/IPC
+ip-26-0-161-153:2636085:2636509 [1] NCCL INFO Channel 14/0 : 49[1] -> 48[0] via P2P/IPC
+ip-26-0-161-78:114518:114937 [6] NCCL INFO Channel 14/0 : 70[6] -> 67[3] via P2P/IPC
+ip-26-0-168-34:272152:272575 [2] NCCL INFO Channel 13/0 : 114[2] -> 113[1] via P2P/IPC
+ip-26-0-165-131:1630207:1630648 [2] NCCL INFO Channel 13/0 : 74[2] -> 73[1] via P2P/IPC
+ip-26-0-165-131:1630206:1630647 [1] NCCL INFO Channel 15/0 : 73[1] -> 72[0] via P2P/IPC
+ip-26-0-161-153:2636086:2636507 [2] NCCL INFO Channel 15/0 : 50[2] -> 49[1] via P2P/IPC
+ip-26-0-161-153:2636085:2636509 [1] NCCL INFO Channel 15/0 : 49[1] -> 48[0] via P2P/IPC
+ip-26-0-165-131:1630207:1630648 [2] NCCL INFO Channel 14/0 : 74[2] -> 73[1] via P2P/IPC
+ip-26-0-165-131:1630207:1630648 [2] NCCL INFO Channel 15/0 : 74[2] -> 73[1] via P2P/IPC
+ip-26-0-168-52:2722051:2722465 [6] NCCL INFO Channel 06/0 : 119[7] -> 126[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-52:2722051:2722465 [6] NCCL INFO Channel 14/0 : 119[7] -> 126[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272152:272575 [2] NCCL INFO Channel 14/0 : 114[2] -> 113[1] via P2P/IPC
+ip-26-0-161-103:96890:97313 [2] NCCL INFO Channel 11/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-168-52:2722051:2722465 [6] NCCL INFO Channel 05/0 : 126[6] -> 5[5] [send] via NET/Libfabric/5(125)/GDRDMA
+ip-26-0-168-52:2722051:2722465 [6] NCCL INFO Channel 13/0 : 126[6] -> 5[5] [send] via NET/Libfabric/5(125)/GDRDMA
+ip-26-0-168-52:2722051:2722465 [6] NCCL INFO Channel 06/0 : 126[6] -> 123[3] via P2P/IPC
+ip-26-0-160-225:124165:124584 [5] NCCL INFO Channel 05/0 : 5[5] -> 3[3] via P2P/IPC
+ip-26-0-161-103:96890:97313 [2] NCCL INFO Channel 12/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-161-103:96890:97313 [2] NCCL INFO Channel 13/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-168-34:272152:272575 [2] NCCL INFO Channel 15/0 : 114[2] -> 113[1] via P2P/IPC
+ip-26-0-161-103:96890:97313 [2] NCCL INFO Channel 14/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-160-225:124165:124584 [5] NCCL INFO Channel 13/0 : 5[5] -> 3[3] via P2P/IPC
+ip-26-0-168-52:2722051:2722465 [6] NCCL INFO Channel 14/0 : 126[6] -> 123[3] via P2P/IPC
+ip-26-0-161-103:96890:97313 [2] NCCL INFO Channel 15/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-165-131:1630212:1630652 [7] NCCL INFO Channel 07/0 : 68[4] -> 79[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96895:97307 [7] NCCL INFO Channel 07/0 : 12[4] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96892:97308 [4] NCCL INFO Channel 15/0 : 20[4] -> 31[7] [send] via NET/Libfabric/7(23)/GDRDMA
+ip-26-0-165-131:1630212:1630652 [7] NCCL INFO Channel 15/0 : 68[4] -> 79[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630209:1630651 [4] NCCL INFO Channel 07/0 : 76[4] -> 87[7] [send] via NET/Libfabric/7(79)/GDRDMA
+ip-26-0-165-131:1630209:1630651 [4] NCCL INFO Channel 15/0 : 76[4] -> 87[7] [send] via NET/Libfabric/7(79)/GDRDMA
+ip-26-0-160-225:124167:124586 [7] NCCL INFO Channel 07/0 : 124[4] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124164:124585 [4] NCCL INFO Channel 07/0 : 4[4] -> 15[7] [send] via NET/Libfabric/7(7)/GDRDMA
+ip-26-0-160-225:124167:124586 [7] NCCL INFO Channel 15/0 : 124[4] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124164:124585 [4] NCCL INFO Channel 15/0 : 4[4] -> 15[7] [send] via NET/Libfabric/7(7)/GDRDMA
+ip-26-0-161-138:96570:97005 [7] NCCL INFO Channel 07/0 : 28[4] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153569:153981 [7] NCCL INFO Channel 07/0 : 84[4] -> 95[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96567:97008 [4] NCCL INFO Channel 15/0 : 36[4] -> 47[7] [send] via NET/Libfabric/7(39)/GDRDMA
+ip-26-0-161-138:96570:97005 [7] NCCL INFO Channel 15/0 : 28[4] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96895:97307 [7] NCCL INFO Channel 15/0 : 12[4] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153569:153981 [7] NCCL INFO Channel 15/0 : 84[4] -> 95[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227470:1227885 [7] NCCL INFO Channel 07/0 : 76[4] -> 87[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227470:1227885 [7] NCCL INFO Channel 15/0 : 76[4] -> 87[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153569:153981 [7] NCCL INFO Channel 07/0 : 95[7] -> 91[3] via P2P/IPC
+ip-26-0-165-59:1557273:1557685 [7] NCCL INFO Channel 07/0 : 92[4] -> 103[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227470:1227885 [7] NCCL INFO Channel 07/0 : 87[7] -> 83[3] via P2P/IPC
+ip-26-0-165-59:1557273:1557685 [7] NCCL INFO Channel 15/0 : 92[4] -> 103[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114519:114931 [7] NCCL INFO Channel 07/0 : 60[4] -> 71[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114519:114931 [7] NCCL INFO Channel 15/0 : 60[4] -> 71[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114516:114934 [4] NCCL INFO Channel 07/0 : 68[4] -> 79[7] [send] via NET/Libfabric/7(71)/GDRDMA
+ip-26-0-165-59:1557273:1557685 [7] NCCL INFO Channel 07/0 : 103[7] -> 99[3] via P2P/IPC
+ip-26-0-161-123:32711:33127 [7] NCCL INFO Channel 07/0 : 20[4] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114516:114934 [4] NCCL INFO Channel 15/0 : 68[4] -> 79[7] [send] via NET/Libfabric/7(71)/GDRDMA
+ip-26-0-161-123:32708:33126 [4] NCCL INFO Channel 07/0 : 28[4] -> 39[7] [send] via NET/Libfabric/7(31)/GDRDMA
+ip-26-0-161-123:32711:33127 [7] NCCL INFO Channel 15/0 : 20[4] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32708:33126 [4] NCCL INFO Channel 15/0 : 28[4] -> 39[7] [send] via NET/Libfabric/7(31)/GDRDMA
+ip-26-0-160-242:1364909:1365327 [7] NCCL INFO Channel 07/0 : 4[4] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364906:1365331 [4] NCCL INFO Channel 15/0 : 12[4] -> 23[7] [send] via NET/Libfabric/7(15)/GDRDMA
+ip-26-0-160-242:1364909:1365327 [7] NCCL INFO Channel 15/0 : 4[4] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630212:1630652 [7] NCCL INFO Channel 07/0 : 79[7] -> 75[3] via P2P/IPC
+ip-26-0-161-123:32711:33127 [7] NCCL INFO Channel 07/0 : 31[7] -> 27[3] via P2P/IPC
+ip-26-0-161-138:96570:97005 [7] NCCL INFO Channel 07/0 : 39[7] -> 35[3] via P2P/IPC
+ip-26-0-161-103:96895:97307 [7] NCCL INFO Channel 07/0 : 23[7] -> 19[3] via P2P/IPC
+ip-26-0-160-242:1364909:1365327 [7] NCCL INFO Channel 07/0 : 15[7] -> 11[3] via P2P/IPC
+ip-26-0-161-142:62888:63302 [7] NCCL INFO Channel 07/0 : 36[4] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62888:63302 [7] NCCL INFO Channel 15/0 : 36[4] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235655:236069 [7] NCCL INFO Channel 07/0 : 100[4] -> 111[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62888:63302 [7] NCCL INFO Channel 07/0 : 47[7] -> 43[3] via P2P/IPC
+ip-26-0-168-30:235655:236069 [7] NCCL INFO Channel 15/0 : 100[4] -> 111[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235652:236066 [4] NCCL INFO Channel 15/0 : 108[4] -> 119[7] [send] via NET/Libfabric/7(111)/GDRDMA
+ip-26-0-168-52:2722052:2722464 [7] NCCL INFO Channel 07/0 : 116[4] -> 127[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153569:153981 [7] NCCL INFO Channel 15/0 : 95[7] -> 91[3] via P2P/IPC
+ip-26-0-168-52:2722049:2722467 [4] NCCL INFO Channel 07/0 : 124[4] -> 7[7] [send] via NET/Libfabric/7(127)/GDRDMA
+ip-26-0-168-52:2722052:2722464 [7] NCCL INFO Channel 15/0 : 116[4] -> 127[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722049:2722467 [4] NCCL INFO Channel 15/0 : 124[4] -> 7[7] [send] via NET/Libfabric/7(127)/GDRDMA
+ip-26-0-168-30:235655:236069 [7] NCCL INFO Channel 07/0 : 111[7] -> 107[3] via P2P/IPC
+ip-26-0-165-59:1557273:1557685 [7] NCCL INFO Channel 15/0 : 103[7] -> 99[3] via P2P/IPC
+ip-26-0-160-225:124167:124586 [7] NCCL INFO Channel 07/0 : 7[7] -> 3[3] via P2P/IPC
+ip-26-0-168-52:2722052:2722464 [7] NCCL INFO Channel 07/0 : 127[7] -> 123[3] via P2P/IPC
+ip-26-0-161-178:2815970:2816397 [7] NCCL INFO Channel 07/0 : 52[4] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815967:2816399 [4] NCCL INFO Channel 07/0 : 60[4] -> 71[7] [send] via NET/Libfabric/7(63)/GDRDMA
+ip-26-0-161-178:2815970:2816397 [7] NCCL INFO Channel 15/0 : 52[4] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815967:2816399 [4] NCCL INFO Channel 15/0 : 60[4] -> 71[7] [send] via NET/Libfabric/7(63)/GDRDMA
+ip-26-0-161-78:114519:114931 [7] NCCL INFO Channel 07/0 : 71[7] -> 67[3] via P2P/IPC
+ip-26-0-165-164:1227470:1227885 [7] NCCL INFO Channel 15/0 : 87[7] -> 83[3] via P2P/IPC
+ip-26-0-161-123:32711:33127 [7] NCCL INFO Channel 15/0 : 31[7] -> 27[3] via P2P/IPC
+ip-26-0-161-138:96570:97005 [7] NCCL INFO Channel 15/0 : 39[7] -> 35[3] via P2P/IPC
+ip-26-0-168-52:2722052:2722464 [7] NCCL INFO Channel 15/0 : 127[7] -> 123[3] via P2P/IPC
+ip-26-0-165-131:1630212:1630652 [7] NCCL INFO Channel 15/0 : 79[7] -> 75[3] via P2P/IPC
+ip-26-0-160-242:1364909:1365327 [7] NCCL INFO Channel 15/0 : 15[7] -> 11[3] via P2P/IPC
+ip-26-0-165-59:1557273:1557685 [7] NCCL INFO Channel 01/0 : 103[7] -> 102[6] via P2P/IPC
+ip-26-0-165-59:1557273:1557685 [7] NCCL INFO Channel 02/0 : 103[7] -> 102[6] via P2P/IPC
+ip-26-0-165-59:1557273:1557685 [7] NCCL INFO Channel 03/0 : 103[7] -> 102[6] via P2P/IPC
+ip-26-0-168-30:235655:236069 [7] NCCL INFO Channel 15/0 : 111[7] -> 107[3] via P2P/IPC
+ip-26-0-165-59:1557273:1557685 [7] NCCL INFO Channel 04/0 : 103[7] -> 102[6] via P2P/IPC
+ip-26-0-160-225:124167:124586 [7] NCCL INFO Channel 15/0 : 7[7] -> 3[3] via P2P/IPC
+ip-26-0-165-164:1227470:1227885 [7] NCCL INFO Channel 01/0 : 87[7] -> 86[6] via P2P/IPC
+ip-26-0-168-30:235655:236069 [7] NCCL INFO Channel 01/0 : 111[7] -> 110[6] via P2P/IPC
+ip-26-0-165-59:1557273:1557685 [7] NCCL INFO Channel 05/0 : 103[7] -> 102[6] via P2P/IPC
+ip-26-0-168-30:235655:236069 [7] NCCL INFO Channel 02/0 : 111[7] -> 110[6] via P2P/IPC
+ip-26-0-165-164:1227470:1227885 [7] NCCL INFO Channel 02/0 : 87[7] -> 86[6] via P2P/IPC
+ip-26-0-165-59:1557273:1557685 [7] NCCL INFO Channel 09/0 : 103[7] -> 102[6] via P2P/IPC
+ip-26-0-168-30:235655:236069 [7] NCCL INFO Channel 03/0 : 111[7] -> 110[6] via P2P/IPC
+ip-26-0-165-202:153569:153981 [7] NCCL INFO Channel 01/0 : 95[7] -> 94[6] via P2P/IPC
+ip-26-0-161-142:62888:63302 [7] NCCL INFO Channel 15/0 : 47[7] -> 43[3] via P2P/IPC
+ip-26-0-168-52:2722052:2722464 [7] NCCL INFO Channel 01/0 : 127[7] -> 126[6] via P2P/IPC
+ip-26-0-165-164:1227470:1227885 [7] NCCL INFO Channel 03/0 : 87[7] -> 86[6] via P2P/IPC
+ip-26-0-168-30:235655:236069 [7] NCCL INFO Channel 04/0 : 111[7] -> 110[6] via P2P/IPC
+ip-26-0-165-59:1557273:1557685 [7] NCCL INFO Channel 10/0 : 103[7] -> 102[6] via P2P/IPC
+ip-26-0-161-153:2636091:2636512 [7] NCCL INFO Channel 07/0 : 44[4] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636088:2636510 [4] NCCL INFO Channel 07/0 : 52[4] -> 63[7] [send] via NET/Libfabric/7(55)/GDRDMA
+ip-26-0-161-153:2636091:2636512 [7] NCCL INFO Channel 15/0 : 44[4] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235655:236069 [7] NCCL INFO Channel 05/0 : 111[7] -> 110[6] via P2P/IPC
+ip-26-0-161-153:2636088:2636510 [4] NCCL INFO Channel 15/0 : 52[4] -> 63[7] [send] via NET/Libfabric/7(55)/GDRDMA
+ip-26-0-165-164:1227470:1227885 [7] NCCL INFO Channel 04/0 : 87[7] -> 86[6] via P2P/IPC
+ip-26-0-160-242:1364909:1365327 [7] NCCL INFO Channel 01/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-165-59:1557273:1557685 [7] NCCL INFO Channel 11/0 : 103[7] -> 102[6] via P2P/IPC
+ip-26-0-168-30:235655:236069 [7] NCCL INFO Channel 09/0 : 111[7] -> 110[6] via P2P/IPC
+ip-26-0-161-138:96570:97005 [7] NCCL INFO Channel 01/0 : 39[7] -> 38[6] via P2P/IPC
+ip-26-0-161-178:2815970:2816397 [7] NCCL INFO Channel 07/0 : 63[7] -> 59[3] via P2P/IPC
+ip-26-0-161-153:2636091:2636512 [7] NCCL INFO Channel 07/0 : 55[7] -> 51[3] via P2P/IPC
+ip-26-0-161-123:32711:33127 [7] NCCL INFO Channel 01/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-161-78:114519:114931 [7] NCCL INFO Channel 15/0 : 71[7] -> 67[3] via P2P/IPC
+ip-26-0-165-59:1557273:1557685 [7] NCCL INFO Channel 12/0 : 103[7] -> 102[6] via P2P/IPC
+ip-26-0-168-30:235655:236069 [7] NCCL INFO Channel 10/0 : 111[7] -> 110[6] via P2P/IPC
+ip-26-0-165-164:1227470:1227885 [7] NCCL INFO Channel 05/0 : 87[7] -> 86[6] via P2P/IPC
+ip-26-0-161-138:96570:97005 [7] NCCL INFO Channel 02/0 : 39[7] -> 38[6] via P2P/IPC
+ip-26-0-165-59:1557273:1557685 [7] NCCL INFO Channel 13/0 : 103[7] -> 102[6] via P2P/IPC
+ip-26-0-168-30:235655:236069 [7] NCCL INFO Channel 11/0 : 111[7] -> 110[6] via P2P/IPC
+ip-26-0-168-52:2722052:2722464 [7] NCCL INFO Channel 02/0 : 127[7] -> 126[6] via P2P/IPC
+ip-26-0-165-131:1630212:1630652 [7] NCCL INFO Channel 01/0 : 79[7] -> 78[6] via P2P/IPC
+ip-26-0-165-164:1227470:1227885 [7] NCCL INFO Channel 09/0 : 87[7] -> 86[6] via P2P/IPC
+ip-26-0-161-138:96570:97005 [7] NCCL INFO Channel 03/0 : 39[7] -> 38[6] via P2P/IPC
+ip-26-0-160-225:124167:124586 [7] NCCL INFO Channel 01/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-168-30:235655:236069 [7] NCCL INFO Channel 12/0 : 111[7] -> 110[6] via P2P/IPC
+ip-26-0-168-52:2722052:2722464 [7] NCCL INFO Channel 03/0 : 127[7] -> 126[6] via P2P/IPC
+ip-26-0-160-242:1364909:1365327 [7] NCCL INFO Channel 02/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-161-138:96570:97005 [7] NCCL INFO Channel 04/0 : 39[7] -> 38[6] via P2P/IPC
+ip-26-0-160-225:124167:124586 [7] NCCL INFO Channel 02/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-168-30:235655:236069 [7] NCCL INFO Channel 13/0 : 111[7] -> 110[6] via P2P/IPC
+ip-26-0-161-123:32711:33127 [7] NCCL INFO Channel 02/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-168-52:2722052:2722464 [7] NCCL INFO Channel 04/0 : 127[7] -> 126[6] via P2P/IPC
+ip-26-0-165-59:1557270:1557688 [4] NCCL INFO Channel 01/0 : 100[4] -> 99[3] via P2P/IPC
+ip-26-0-160-242:1364909:1365327 [7] NCCL INFO Channel 03/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-161-138:96570:97005 [7] NCCL INFO Channel 05/0 : 39[7] -> 38[6] via P2P/IPC
+ip-26-0-161-78:114519:114931 [7] NCCL INFO Channel 01/0 : 71[7] -> 70[6] via P2P/IPC
+ip-26-0-160-225:124167:124586 [7] NCCL INFO Channel 03/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-123:32711:33127 [7] NCCL INFO Channel 03/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-160-242:1364909:1365327 [7] NCCL INFO Channel 04/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-161-78:114519:114931 [7] NCCL INFO Channel 02/0 : 71[7] -> 70[6] via P2P/IPC
+ip-26-0-161-138:96570:97005 [7] NCCL INFO Channel 09/0 : 39[7] -> 38[6] via P2P/IPC
+ip-26-0-161-103:96895:97307 [7] NCCL INFO Channel 15/0 : 23[7] -> 19[3] via P2P/IPC
+ip-26-0-165-131:1630212:1630652 [7] NCCL INFO Channel 02/0 : 79[7] -> 78[6] via P2P/IPC
+ip-26-0-168-52:2722052:2722464 [7] NCCL INFO Channel 05/0 : 127[7] -> 126[6] via P2P/IPC
+ip-26-0-161-123:32711:33127 [7] NCCL INFO Channel 04/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-160-225:124167:124586 [7] NCCL INFO Channel 04/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-78:114519:114931 [7] NCCL INFO Channel 03/0 : 71[7] -> 70[6] via P2P/IPC
+ip-26-0-165-59:1557270:1557688 [4] NCCL INFO Channel 02/0 : 100[4] -> 99[3] via P2P/IPC
+ip-26-0-161-138:96570:97005 [7] NCCL INFO Channel 10/0 : 39[7] -> 38[6] via P2P/IPC
+ip-26-0-165-164:1227470:1227885 [7] NCCL INFO Channel 10/0 : 87[7] -> 86[6] via P2P/IPC
+ip-26-0-160-242:1364909:1365327 [7] NCCL INFO Channel 05/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-160-225:124167:124586 [7] NCCL INFO Channel 05/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-78:114519:114931 [7] NCCL INFO Channel 04/0 : 71[7] -> 70[6] via P2P/IPC
+ip-26-0-165-131:1630212:1630652 [7] NCCL INFO Channel 03/0 : 79[7] -> 78[6] via P2P/IPC
+ip-26-0-161-123:32711:33127 [7] NCCL INFO Channel 05/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-165-59:1557270:1557688 [4] NCCL INFO Channel 04/0 : 100[4] -> 99[3] via P2P/IPC
+ip-26-0-168-52:2722052:2722464 [7] NCCL INFO Channel 09/0 : 127[7] -> 126[6] via P2P/IPC
+ip-26-0-160-225:124167:124586 [7] NCCL INFO Channel 09/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-138:96570:97005 [7] NCCL INFO Channel 11/0 : 39[7] -> 38[6] via P2P/IPC
+ip-26-0-165-202:153569:153981 [7] NCCL INFO Channel 02/0 : 95[7] -> 94[6] via P2P/IPC
+ip-26-0-161-78:114519:114931 [7] NCCL INFO Channel 05/0 : 71[7] -> 70[6] via P2P/IPC
+ip-26-0-165-164:1227470:1227885 [7] NCCL INFO Channel 11/0 : 87[7] -> 86[6] via P2P/IPC
+ip-26-0-160-242:1364909:1365327 [7] NCCL INFO Channel 09/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-161-138:96567:97008 [4] NCCL INFO Channel 01/0 : 36[4] -> 35[3] via P2P/IPC
+ip-26-0-165-131:1630212:1630652 [7] NCCL INFO Channel 04/0 : 79[7] -> 78[6] via P2P/IPC
+ip-26-0-168-34:272157:272572 [7] NCCL INFO Channel 07/0 : 108[4] -> 119[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32711:33127 [7] NCCL INFO Channel 09/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-160-225:124167:124586 [7] NCCL INFO Channel 10/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-165-59:1557270:1557688 [4] NCCL INFO Channel 09/0 : 100[4] -> 99[3] via P2P/IPC
+ip-26-0-168-34:272157:272572 [7] NCCL INFO Channel 15/0 : 108[4] -> 119[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114519:114931 [7] NCCL INFO Channel 09/0 : 71[7] -> 70[6] via P2P/IPC
+ip-26-0-165-164:1227470:1227885 [7] NCCL INFO Channel 12/0 : 87[7] -> 86[6] via P2P/IPC
+ip-26-0-160-225:124164:124585 [4] NCCL INFO Channel 01/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-225:124167:124586 [7] NCCL INFO Channel 11/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-103:96895:97307 [7] NCCL INFO Channel 01/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-168-34:272157:272572 [7] NCCL INFO Channel 07/0 : 119[7] -> 115[3] via P2P/IPC
+ip-26-0-160-242:1364909:1365327 [7] NCCL INFO Channel 10/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-165-59:1557270:1557688 [4] NCCL INFO Channel 10/0 : 100[4] -> 99[3] via P2P/IPC
+ip-26-0-168-52:2722052:2722464 [7] NCCL INFO Channel 10/0 : 127[7] -> 126[6] via P2P/IPC
+ip-26-0-161-123:32711:33127 [7] NCCL INFO Channel 10/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-161-78:114516:114934 [4] NCCL INFO Channel 01/0 : 68[4] -> 67[3] via P2P/IPC
+ip-26-0-160-225:124164:124585 [4] NCCL INFO Channel 02/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-225:124167:124586 [7] NCCL INFO Channel 12/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-78:114519:114931 [7] NCCL INFO Channel 10/0 : 71[7] -> 70[6] via P2P/IPC
+ip-26-0-165-164:1227467:1227886 [4] NCCL INFO Channel 01/0 : 84[4] -> 83[3] via P2P/IPC
+ip-26-0-165-59:1557270:1557688 [4] NCCL INFO Channel 12/0 : 100[4] -> 99[3] via P2P/IPC
+ip-26-0-161-142:62888:63302 [7] NCCL INFO Channel 01/0 : 47[7] -> 46[6] via P2P/IPC
+ip-26-0-165-202:153569:153981 [7] NCCL INFO Channel 03/0 : 95[7] -> 94[6] via P2P/IPC
+ip-26-0-168-30:235651:236068 [3] NCCL INFO Channel 01/0 : 107[3] -> 106[2] via P2P/IPC
+ip-26-0-168-52:2722052:2722464 [7] NCCL INFO Channel 11/0 : 127[7] -> 126[6] via P2P/IPC
+ip-26-0-161-138:96570:97005 [7] NCCL INFO Channel 12/0 : 39[7] -> 38[6] via P2P/IPC
+ip-26-0-160-242:1364909:1365327 [7] NCCL INFO Channel 11/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-160-225:124164:124585 [4] NCCL INFO Channel 04/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-225:124167:124586 [7] NCCL INFO Channel 13/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-78:114516:114934 [4] NCCL INFO Channel 02/0 : 68[4] -> 67[3] via P2P/IPC
+ip-26-0-160-242:1364906:1365331 [4] NCCL INFO Channel 01/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-161-78:114519:114931 [7] NCCL INFO Channel 11/0 : 71[7] -> 70[6] via P2P/IPC
+ip-26-0-165-131:1630212:1630652 [7] NCCL INFO Channel 05/0 : 79[7] -> 78[6] via P2P/IPC
+ip-26-0-165-202:153569:153981 [7] NCCL INFO Channel 04/0 : 95[7] -> 94[6] via P2P/IPC
+ip-26-0-161-123:32711:33127 [7] NCCL INFO Channel 11/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-165-131:1630212:1630652 [7] NCCL INFO Channel 09/0 : 79[7] -> 78[6] via P2P/IPC
+ip-26-0-165-131:1630212:1630652 [7] NCCL INFO Channel 10/0 : 79[7] -> 78[6] via P2P/IPC
+ip-26-0-161-138:96567:97008 [4] NCCL INFO Channel 02/0 : 36[4] -> 35[3] via P2P/IPC
+ip-26-0-165-59:1557269:1557684 [3] NCCL INFO Channel 01/0 : 99[3] -> 98[2] via P2P/IPC
+ip-26-0-165-164:1227470:1227885 [7] NCCL INFO Channel 13/0 : 87[7] -> 86[6] via P2P/IPC
+ip-26-0-168-30:235651:236068 [3] NCCL INFO Channel 03/0 : 107[3] -> 106[2] via P2P/IPC
+ip-26-0-161-153:2636091:2636512 [7] NCCL INFO Channel 15/0 : 55[7] -> 51[3] via P2P/IPC
+ip-26-0-161-138:96570:97005 [7] NCCL INFO Channel 13/0 : 39[7] -> 38[6] via P2P/IPC
+ip-26-0-165-202:153569:153981 [7] NCCL INFO Channel 05/0 : 95[7] -> 94[6] via P2P/IPC
+ip-26-0-168-52:2722049:2722467 [4] NCCL INFO Channel 01/0 : 124[4] -> 123[3] via P2P/IPC
+ip-26-0-165-59:1557269:1557684 [3] NCCL INFO Channel 03/0 : 99[3] -> 98[2] via P2P/IPC
+ip-26-0-160-225:124164:124585 [4] NCCL INFO Channel 09/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-168-52:2722052:2722464 [7] NCCL INFO Channel 12/0 : 127[7] -> 126[6] via P2P/IPC
+ip-26-0-160-242:1364909:1365327 [7] NCCL INFO Channel 12/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-165-202:153569:153981 [7] NCCL INFO Channel 09/0 : 95[7] -> 94[6] via P2P/IPC
+ip-26-0-168-30:235654:236070 [6] NCCL INFO Channel 01/0 : 110[6] -> 109[5] via P2P/IPC
+ip-26-0-165-164:1227467:1227886 [4] NCCL INFO Channel 02/0 : 84[4] -> 83[3] via P2P/IPC
+ip-26-0-161-78:114516:114934 [4] NCCL INFO Channel 04/0 : 68[4] -> 67[3] via P2P/IPC
+ip-26-0-161-78:114519:114931 [7] NCCL INFO Channel 12/0 : 71[7] -> 70[6] via P2P/IPC
+ip-26-0-160-242:1364906:1365331 [4] NCCL INFO Channel 02/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-165-59:1557269:1557684 [3] NCCL INFO Channel 04/0 : 99[3] -> 98[2] via P2P/IPC
+ip-26-0-168-30:235651:236068 [3] NCCL INFO Channel 04/0 : 107[3] -> 106[2] via P2P/IPC
+ip-26-0-161-123:32711:33127 [7] NCCL INFO Channel 12/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-161-138:96567:97008 [4] NCCL INFO Channel 04/0 : 36[4] -> 35[3] via P2P/IPC
+ip-26-0-160-242:1364909:1365327 [7] NCCL INFO Channel 13/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-161-123:32708:33126 [4] NCCL INFO Channel 01/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-160-225:124164:124585 [4] NCCL INFO Channel 10/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-142:62888:63302 [7] NCCL INFO Channel 02/0 : 47[7] -> 46[6] via P2P/IPC
+ip-26-0-160-242:1364906:1365331 [4] NCCL INFO Channel 04/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-165-131:1630212:1630652 [7] NCCL INFO Channel 11/0 : 79[7] -> 78[6] via P2P/IPC
+ip-26-0-165-131:1630209:1630651 [4] NCCL INFO Channel 01/0 : 76[4] -> 75[3] via P2P/IPC
+ip-26-0-165-131:1630212:1630652 [7] NCCL INFO Channel 12/0 : 79[7] -> 78[6] via P2P/IPC
+ip-26-0-165-131:1630212:1630652 [7] NCCL INFO Channel 13/0 : 79[7] -> 78[6] via P2P/IPC
+ip-26-0-168-52:2722049:2722467 [4] NCCL INFO Channel 02/0 : 124[4] -> 123[3] via P2P/IPC
+ip-26-0-165-131:1630209:1630651 [4] NCCL INFO Channel 02/0 : 76[4] -> 75[3] via P2P/IPC
+ip-26-0-165-202:153569:153981 [7] NCCL INFO Channel 10/0 : 95[7] -> 94[6] via P2P/IPC
+ip-26-0-165-59:1557269:1557684 [3] NCCL INFO Channel 05/0 : 99[3] -> 98[2] via P2P/IPC
+ip-26-0-168-52:2722052:2722464 [7] NCCL INFO Channel 13/0 : 127[7] -> 126[6] via P2P/IPC
+ip-26-0-165-164:1227467:1227886 [4] NCCL INFO Channel 04/0 : 84[4] -> 83[3] via P2P/IPC
+ip-26-0-160-225:124164:124585 [4] NCCL INFO Channel 12/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-242:1364906:1365331 [4] NCCL INFO Channel 09/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-168-30:235654:236070 [6] NCCL INFO Channel 02/0 : 110[6] -> 109[5] via P2P/IPC
+ip-26-0-161-78:114519:114931 [7] NCCL INFO Channel 13/0 : 71[7] -> 70[6] via P2P/IPC
+ip-26-0-161-142:62888:63302 [7] NCCL INFO Channel 03/0 : 47[7] -> 46[6] via P2P/IPC
+ip-26-0-161-78:114516:114934 [4] NCCL INFO Channel 09/0 : 68[4] -> 67[3] via P2P/IPC
+ip-26-0-160-242:1364906:1365331 [4] NCCL INFO Channel 10/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-161-123:32711:33127 [7] NCCL INFO Channel 13/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-161-123:32708:33126 [4] NCCL INFO Channel 02/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-165-59:1557269:1557684 [3] NCCL INFO Channel 06/0 : 99[3] -> 98[2] via P2P/IPC
+ip-26-0-165-164:1227467:1227886 [4] NCCL INFO Channel 09/0 : 84[4] -> 83[3] via P2P/IPC
+ip-26-0-165-202:153569:153981 [7] NCCL INFO Channel 11/0 : 95[7] -> 94[6] via P2P/IPC
+ip-26-0-161-103:96895:97307 [7] NCCL INFO Channel 02/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-160-242:1364906:1365331 [4] NCCL INFO Channel 12/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-161-142:62888:63302 [7] NCCL INFO Channel 04/0 : 47[7] -> 46[6] via P2P/IPC
+ip-26-0-168-52:2722049:2722467 [4] NCCL INFO Channel 04/0 : 124[4] -> 123[3] via P2P/IPC
+ip-26-0-161-123:32708:33126 [4] NCCL INFO Channel 04/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-165-164:1227469:1227882 [6] NCCL INFO Channel 01/0 : 86[6] -> 85[5] via P2P/IPC
+ip-26-0-165-59:1557269:1557684 [3] NCCL INFO Channel 07/0 : 99[3] -> 98[2] via P2P/IPC
+ip-26-0-161-138:96567:97008 [4] NCCL INFO Channel 09/0 : 36[4] -> 35[3] via P2P/IPC
+ip-26-0-161-123:32708:33126 [4] NCCL INFO Channel 09/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-161-78:114516:114934 [4] NCCL INFO Channel 10/0 : 68[4] -> 67[3] via P2P/IPC
+ip-26-0-165-59:1557271:1557690 [5] NCCL INFO Channel 01/0 : 101[5] -> 100[4] via P2P/IPC
+ip-26-0-165-164:1227467:1227886 [4] NCCL INFO Channel 10/0 : 84[4] -> 83[3] via P2P/IPC
+ip-26-0-168-52:2722049:2722467 [4] NCCL INFO Channel 09/0 : 124[4] -> 123[3] via P2P/IPC
+ip-26-0-161-142:62888:63302 [7] NCCL INFO Channel 05/0 : 47[7] -> 46[6] via P2P/IPC
+ip-26-0-161-123:32708:33126 [4] NCCL INFO Channel 10/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-165-202:153569:153981 [7] NCCL INFO Channel 12/0 : 95[7] -> 94[6] via P2P/IPC
+ip-26-0-168-52:2722048:2722469 [3] NCCL INFO Channel 01/0 : 123[3] -> 122[2] via P2P/IPC
+ip-26-0-165-131:1630209:1630651 [4] NCCL INFO Channel 04/0 : 76[4] -> 75[3] via P2P/IPC
+ip-26-0-161-123:32708:33126 [4] NCCL INFO Channel 12/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-168-30:235651:236068 [3] NCCL INFO Channel 05/0 : 107[3] -> 106[2] via P2P/IPC
+ip-26-0-161-138:96567:97008 [4] NCCL INFO Channel 10/0 : 36[4] -> 35[3] via P2P/IPC
+ip-26-0-161-78:114516:114934 [4] NCCL INFO Channel 12/0 : 68[4] -> 67[3] via P2P/IPC
+ip-26-0-168-52:2722049:2722467 [4] NCCL INFO Channel 10/0 : 124[4] -> 123[3] via P2P/IPC
+ip-26-0-160-242:1364905:1365328 [3] NCCL INFO Channel 01/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-165-164:1227467:1227886 [4] NCCL INFO Channel 12/0 : 84[4] -> 83[3] via P2P/IPC
+ip-26-0-161-142:62888:63302 [7] NCCL INFO Channel 09/0 : 47[7] -> 46[6] via P2P/IPC
+ip-26-0-165-131:1630209:1630651 [4] NCCL INFO Channel 09/0 : 76[4] -> 75[3] via P2P/IPC
+ip-26-0-168-34:272157:272572 [7] NCCL INFO Channel 15/0 : 119[7] -> 115[3] via P2P/IPC
+ip-26-0-161-153:2636091:2636512 [7] NCCL INFO Channel 01/0 : 55[7] -> 54[6] via P2P/IPC
+ip-26-0-168-52:2722048:2722469 [3] NCCL INFO Channel 03/0 : 123[3] -> 122[2] via P2P/IPC
+ip-26-0-165-59:1557272:1557687 [6] NCCL INFO Channel 01/0 : 102[6] -> 101[5] via P2P/IPC
+ip-26-0-165-164:1227469:1227882 [6] NCCL INFO Channel 02/0 : 86[6] -> 85[5] via P2P/IPC
+ip-26-0-161-142:62888:63302 [7] NCCL INFO Channel 10/0 : 47[7] -> 46[6] via P2P/IPC
+ip-26-0-160-242:1364905:1365328 [3] NCCL INFO Channel 03/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-168-52:2722051:2722465 [6] NCCL INFO Channel 01/0 : 126[6] -> 125[5] via P2P/IPC
+ip-26-0-168-52:2722049:2722467 [4] NCCL INFO Channel 12/0 : 124[4] -> 123[3] via P2P/IPC
+ip-26-0-165-59:1557272:1557687 [6] NCCL INFO Channel 02/0 : 102[6] -> 101[5] via P2P/IPC
+ip-26-0-161-123:32707:33124 [3] NCCL INFO Channel 01/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-161-138:96567:97008 [4] NCCL INFO Channel 12/0 : 36[4] -> 35[3] via P2P/IPC
+ip-26-0-165-202:153569:153981 [7] NCCL INFO Channel 13/0 : 95[7] -> 94[6] via P2P/IPC
+ip-26-0-168-52:2722048:2722469 [3] NCCL INFO Channel 04/0 : 123[3] -> 122[2] via P2P/IPC
+ip-26-0-168-30:235651:236068 [3] NCCL INFO Channel 06/0 : 107[3] -> 106[2] via P2P/IPC
+ip-26-0-161-142:62888:63302 [7] NCCL INFO Channel 11/0 : 47[7] -> 46[6] via P2P/IPC
+ip-26-0-168-30:235654:236070 [6] NCCL INFO Channel 03/0 : 110[6] -> 109[5] via P2P/IPC
+ip-26-0-165-59:1557272:1557687 [6] NCCL INFO Channel 03/0 : 102[6] -> 101[5] via P2P/IPC
+ip-26-0-161-78:114515:114936 [3] NCCL INFO Channel 01/0 : 67[3] -> 66[2] via P2P/IPC
+ip-26-0-160-242:1364905:1365328 [3] NCCL INFO Channel 04/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-165-131:1630209:1630651 [4] NCCL INFO Channel 10/0 : 76[4] -> 75[3] via P2P/IPC
+ip-26-0-161-123:32707:33124 [3] NCCL INFO Channel 03/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-165-164:1227466:1227887 [3] NCCL INFO Channel 01/0 : 83[3] -> 82[2] via P2P/IPC
+ip-26-0-161-142:62888:63302 [7] NCCL INFO Channel 12/0 : 47[7] -> 46[6] via P2P/IPC
+ip-26-0-165-59:1557272:1557687 [6] NCCL INFO Channel 04/0 : 102[6] -> 101[5] via P2P/IPC
+ip-26-0-168-52:2722051:2722465 [6] NCCL INFO Channel 02/0 : 126[6] -> 125[5] via P2P/IPC
+ip-26-0-168-52:2722048:2722469 [3] NCCL INFO Channel 05/0 : 123[3] -> 122[2] via P2P/IPC
+ip-26-0-161-78:114515:114936 [3] NCCL INFO Channel 03/0 : 67[3] -> 66[2] via P2P/IPC
+ip-26-0-165-59:1557272:1557687 [6] NCCL INFO Channel 07/0 : 102[6] -> 101[5] via P2P/IPC
+ip-26-0-161-123:32707:33124 [3] NCCL INFO Channel 04/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-160-242:1364905:1365328 [3] NCCL INFO Channel 05/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-165-59:1557271:1557690 [5] NCCL INFO Channel 02/0 : 101[5] -> 100[4] via P2P/IPC
+ip-26-0-161-142:62888:63302 [7] NCCL INFO Channel 13/0 : 47[7] -> 46[6] via P2P/IPC
+ip-26-0-168-30:235651:236068 [3] NCCL INFO Channel 07/0 : 107[3] -> 106[2] via P2P/IPC
+ip-26-0-168-30:235654:236070 [6] NCCL INFO Channel 04/0 : 110[6] -> 109[5] via P2P/IPC
+ip-26-0-168-30:235653:236067 [5] NCCL INFO Channel 01/0 : 109[5] -> 108[4] via P2P/IPC
+ip-26-0-168-52:2722051:2722465 [6] NCCL INFO Channel 03/0 : 126[6] -> 125[5] via P2P/IPC
+ip-26-0-165-164:1227469:1227882 [6] NCCL INFO Channel 03/0 : 86[6] -> 85[5] via P2P/IPC
+ip-26-0-165-164:1227466:1227887 [3] NCCL INFO Channel 03/0 : 83[3] -> 82[2] via P2P/IPC
+ip-26-0-168-52:2722048:2722469 [3] NCCL INFO Channel 06/0 : 123[3] -> 122[2] via P2P/IPC
+ip-26-0-161-123:32709:33129 [5] NCCL INFO Channel 01/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-161-138:96566:97010 [3] NCCL INFO Channel 01/0 : 35[3] -> 34[2] via P2P/IPC
+ip-26-0-160-242:1364908:1365332 [6] NCCL INFO Channel 01/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-165-59:1557272:1557687 [6] NCCL INFO Channel 09/0 : 102[6] -> 101[5] via P2P/IPC
+ip-26-0-160-225:124166:124583 [6] NCCL INFO Channel 01/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-242:1364905:1365328 [3] NCCL INFO Channel 06/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-165-59:1557271:1557690 [5] NCCL INFO Channel 03/0 : 101[5] -> 100[4] via P2P/IPC
+ip-26-0-161-103:96895:97307 [7] NCCL INFO Channel 03/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-161-123:32707:33124 [3] NCCL INFO Channel 05/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-160-225:124163:124588 [3] NCCL INFO Channel 01/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-168-52:2722051:2722465 [6] NCCL INFO Channel 04/0 : 126[6] -> 125[5] via P2P/IPC
+ip-26-0-168-52:2722048:2722469 [3] NCCL INFO Channel 07/0 : 123[3] -> 122[2] via P2P/IPC
+ip-26-0-165-59:1557269:1557684 [3] NCCL INFO Channel 09/0 : 99[3] -> 98[2] via P2P/IPC
+ip-26-0-165-164:1227469:1227882 [6] NCCL INFO Channel 04/0 : 86[6] -> 85[5] via P2P/IPC
+ip-26-0-161-123:32709:33129 [5] NCCL INFO Channel 02/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-161-138:96566:97010 [3] NCCL INFO Channel 03/0 : 35[3] -> 34[2] via P2P/IPC
+ip-26-0-168-30:235651:236068 [3] NCCL INFO Channel 09/0 : 107[3] -> 106[2] via P2P/IPC
+ip-26-0-168-30:235652:236066 [4] NCCL INFO Channel 01/0 : 108[4] -> 107[3] via P2P/IPC
+ip-26-0-165-164:1227466:1227887 [3] NCCL INFO Channel 04/0 : 83[3] -> 82[2] via P2P/IPC
+ip-26-0-168-30:235654:236070 [6] NCCL INFO Channel 07/0 : 110[6] -> 109[5] via P2P/IPC
+ip-26-0-165-59:1557272:1557687 [6] NCCL INFO Channel 10/0 : 102[6] -> 101[5] via P2P/IPC
+ip-26-0-168-30:235653:236067 [5] NCCL INFO Channel 02/0 : 109[5] -> 108[4] via P2P/IPC
+ip-26-0-161-138:96569:97006 [6] NCCL INFO Channel 01/0 : 38[6] -> 37[5] via P2P/IPC
+ip-26-0-161-123:32707:33124 [3] NCCL INFO Channel 06/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-160-242:1364908:1365332 [6] NCCL INFO Channel 02/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-160-242:1364907:1365329 [5] NCCL INFO Channel 01/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-160-242:1364905:1365328 [3] NCCL INFO Channel 07/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-165-59:1557271:1557690 [5] NCCL INFO Channel 06/0 : 101[5] -> 100[4] via P2P/IPC
+ip-26-0-165-59:1557269:1557684 [3] NCCL INFO Channel 11/0 : 99[3] -> 98[2] via P2P/IPC
+ip-26-0-161-123:32710:33125 [6] NCCL INFO Channel 01/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-161-78:114515:114936 [3] NCCL INFO Channel 04/0 : 67[3] -> 66[2] via P2P/IPC
+ip-26-0-161-123:32709:33129 [5] NCCL INFO Channel 03/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-165-164:1227466:1227887 [3] NCCL INFO Channel 05/0 : 83[3] -> 82[2] via P2P/IPC
+ip-26-0-165-164:1227469:1227882 [6] NCCL INFO Channel 07/0 : 86[6] -> 85[5] via P2P/IPC
+ip-26-0-165-59:1557272:1557687 [6] NCCL INFO Channel 11/0 : 102[6] -> 101[5] via P2P/IPC
+ip-26-0-165-164:1227468:1227883 [5] NCCL INFO Channel 01/0 : 85[5] -> 84[4] via P2P/IPC
+ip-26-0-160-225:124165:124584 [5] NCCL INFO Channel 01/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-123:32707:33124 [3] NCCL INFO Channel 07/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-165-202:153566:153980 [4] NCCL INFO Channel 01/0 : 92[4] -> 91[3] via P2P/IPC
+ip-26-0-168-30:235654:236070 [6] NCCL INFO Channel 09/0 : 110[6] -> 109[5] via P2P/IPC
+ip-26-0-161-78:114515:114936 [3] NCCL INFO Channel 05/0 : 67[3] -> 66[2] via P2P/IPC
+ip-26-0-160-242:1364908:1365332 [6] NCCL INFO Channel 03/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-160-242:1364905:1365328 [3] NCCL INFO Channel 09/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-165-59:1557271:1557690 [5] NCCL INFO Channel 07/0 : 101[5] -> 100[4] via P2P/IPC
+ip-26-0-168-30:235652:236066 [4] NCCL INFO Channel 02/0 : 108[4] -> 107[3] via P2P/IPC
+ip-26-0-168-30:235653:236067 [5] NCCL INFO Channel 03/0 : 109[5] -> 108[4] via P2P/IPC
+ip-26-0-160-242:1364907:1365329 [5] NCCL INFO Channel 02/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-168-30:235651:236068 [3] NCCL INFO Channel 11/0 : 107[3] -> 106[2] via P2P/IPC
+ip-26-0-160-225:124165:124584 [5] NCCL INFO Channel 02/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-123:32710:33125 [6] NCCL INFO Channel 02/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-161-123:32709:33129 [5] NCCL INFO Channel 06/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-161-153:2636091:2636512 [7] NCCL INFO Channel 02/0 : 55[7] -> 54[6] via P2P/IPC
+ip-26-0-165-59:1557269:1557684 [3] NCCL INFO Channel 12/0 : 99[3] -> 98[2] via P2P/IPC
+ip-26-0-168-34:272157:272572 [7] NCCL INFO Channel 01/0 : 119[7] -> 118[6] via P2P/IPC
+ip-26-0-165-202:153565:153985 [3] NCCL INFO Channel 01/0 : 91[3] -> 90[2] via P2P/IPC
+ip-26-0-165-59:1557272:1557687 [6] NCCL INFO Channel 12/0 : 102[6] -> 101[5] via P2P/IPC
+ip-26-0-165-164:1227466:1227887 [3] NCCL INFO Channel 06/0 : 83[3] -> 82[2] via P2P/IPC
+ip-26-0-165-164:1227469:1227882 [6] NCCL INFO Channel 09/0 : 86[6] -> 85[5] via P2P/IPC
+ip-26-0-161-123:32707:33124 [3] NCCL INFO Channel 09/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-165-164:1227468:1227883 [5] NCCL INFO Channel 02/0 : 85[5] -> 84[4] via P2P/IPC
+ip-26-0-160-225:124165:124584 [5] NCCL INFO Channel 03/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-138:96566:97010 [3] NCCL INFO Channel 04/0 : 35[3] -> 34[2] via P2P/IPC
+ip-26-0-160-225:124166:124583 [6] NCCL INFO Channel 02/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-78:114518:114937 [6] NCCL INFO Channel 01/0 : 70[6] -> 69[5] via P2P/IPC
+ip-26-0-165-59:1557271:1557690 [5] NCCL INFO Channel 09/0 : 101[5] -> 100[4] via P2P/IPC
+ip-26-0-161-78:114515:114936 [3] NCCL INFO Channel 06/0 : 67[3] -> 66[2] via P2P/IPC
+ip-26-0-165-202:153566:153980 [4] NCCL INFO Channel 02/0 : 92[4] -> 91[3] via P2P/IPC
+ip-26-0-161-138:96569:97006 [6] NCCL INFO Channel 02/0 : 38[6] -> 37[5] via P2P/IPC
+ip-26-0-160-242:1364908:1365332 [6] NCCL INFO Channel 04/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-160-242:1364905:1365328 [3] NCCL INFO Channel 11/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-168-30:235654:236070 [6] NCCL INFO Channel 10/0 : 110[6] -> 109[5] via P2P/IPC
+ip-26-0-160-242:1364907:1365329 [5] NCCL INFO Channel 03/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-161-103:96895:97307 [7] NCCL INFO Channel 04/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-161-123:32709:33129 [5] NCCL INFO Channel 07/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-168-30:235652:236066 [4] NCCL INFO Channel 04/0 : 108[4] -> 107[3] via P2P/IPC
+ip-26-0-168-30:235653:236067 [5] NCCL INFO Channel 06/0 : 109[5] -> 108[4] via P2P/IPC
+ip-26-0-161-123:32710:33125 [6] NCCL INFO Channel 03/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-165-202:153565:153985 [3] NCCL INFO Channel 03/0 : 91[3] -> 90[2] via P2P/IPC
+ip-26-0-168-30:235651:236068 [3] NCCL INFO Channel 12/0 : 107[3] -> 106[2] via P2P/IPC
+ip-26-0-165-59:1557269:1557684 [3] NCCL INFO Channel 13/0 : 99[3] -> 98[2] via P2P/IPC
+ip-26-0-165-59:1557272:1557687 [6] NCCL INFO Channel 15/0 : 102[6] -> 101[5] via P2P/IPC
+ip-26-0-161-123:32707:33124 [3] NCCL INFO Channel 11/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-161-138:96566:97010 [3] NCCL INFO Channel 05/0 : 35[3] -> 34[2] via P2P/IPC
+ip-26-0-165-164:1227466:1227887 [3] NCCL INFO Channel 07/0 : 83[3] -> 82[2] via P2P/IPC
+ip-26-0-161-142:62885:63304 [4] NCCL INFO Channel 01/0 : 44[4] -> 43[3] via P2P/IPC
+ip-26-0-161-78:114518:114937 [6] NCCL INFO Channel 02/0 : 70[6] -> 69[5] via P2P/IPC
+ip-26-0-165-59:1557271:1557690 [5] NCCL INFO Channel 10/0 : 101[5] -> 100[4] via P2P/IPC
+ip-26-0-160-225:124163:124588 [3] NCCL INFO Channel 03/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-225:124165:124584 [5] NCCL INFO Channel 06/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-138:96569:97006 [6] NCCL INFO Channel 03/0 : 38[6] -> 37[5] via P2P/IPC
+ip-26-0-161-78:114515:114936 [3] NCCL INFO Channel 07/0 : 67[3] -> 66[2] via P2P/IPC
+ip-26-0-160-225:124166:124583 [6] NCCL INFO Channel 03/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-202:153566:153980 [4] NCCL INFO Channel 04/0 : 92[4] -> 91[3] via P2P/IPC
+ip-26-0-165-164:1227466:1227887 [3] NCCL INFO Channel 09/0 : 83[3] -> 82[2] via P2P/IPC
+ip-26-0-161-123:32709:33129 [5] NCCL INFO Channel 09/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-160-242:1364905:1365328 [3] NCCL INFO Channel 12/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-161-123:32710:33125 [6] NCCL INFO Channel 04/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-160-242:1364908:1365332 [6] NCCL INFO Channel 07/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-160-242:1364907:1365329 [5] NCCL INFO Channel 06/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-165-202:153565:153985 [3] NCCL INFO Channel 04/0 : 91[3] -> 90[2] via P2P/IPC
+ip-26-0-161-138:96566:97010 [3] NCCL INFO Channel 06/0 : 35[3] -> 34[2] via P2P/IPC
+ip-26-0-165-59:1557269:1557684 [3] NCCL INFO Channel 14/0 : 99[3] -> 98[2] via P2P/IPC
+ip-26-0-161-123:32707:33124 [3] NCCL INFO Channel 12/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-168-30:235654:236070 [6] NCCL INFO Channel 11/0 : 110[6] -> 109[5] via P2P/IPC
+ip-26-0-168-30:235653:236067 [5] NCCL INFO Channel 07/0 : 109[5] -> 108[4] via P2P/IPC
+ip-26-0-165-164:1227466:1227887 [3] NCCL INFO Channel 11/0 : 83[3] -> 82[2] via P2P/IPC
+ip-26-0-165-131:1630209:1630651 [4] NCCL INFO Channel 12/0 : 76[4] -> 75[3] via P2P/IPC
+ip-26-0-168-30:235652:236066 [4] NCCL INFO Channel 09/0 : 108[4] -> 107[3] via P2P/IPC
+ip-26-0-161-178:2815970:2816397 [7] NCCL INFO Channel 15/0 : 63[7] -> 59[3] via P2P/IPC
+ip-26-0-168-30:235651:236068 [3] NCCL INFO Channel 13/0 : 107[3] -> 106[2] via P2P/IPC
+ip-26-0-165-59:1557271:1557690 [5] NCCL INFO Channel 11/0 : 101[5] -> 100[4] via P2P/IPC
+ip-26-0-161-142:62885:63304 [4] NCCL INFO Channel 02/0 : 44[4] -> 43[3] via P2P/IPC
+ip-26-0-161-78:114518:114937 [6] NCCL INFO Channel 03/0 : 70[6] -> 69[5] via P2P/IPC
+ip-26-0-161-138:96569:97006 [6] NCCL INFO Channel 04/0 : 38[6] -> 37[5] via P2P/IPC
+ip-26-0-160-225:124165:124584 [5] NCCL INFO Channel 07/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-225:124163:124588 [3] NCCL INFO Channel 04/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-78:114517:114935 [5] NCCL INFO Channel 01/0 : 69[5] -> 68[4] via P2P/IPC
+ip-26-0-161-138:96568:97007 [5] NCCL INFO Channel 01/0 : 37[5] -> 36[4] via P2P/IPC
+ip-26-0-161-78:114515:114936 [3] NCCL INFO Channel 09/0 : 67[3] -> 66[2] via P2P/IPC
+ip-26-0-165-164:1227466:1227887 [3] NCCL INFO Channel 12/0 : 83[3] -> 82[2] via P2P/IPC
+ip-26-0-160-225:124166:124583 [6] NCCL INFO Channel 04/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-202:153568:153982 [6] NCCL INFO Channel 01/0 : 94[6] -> 93[5] via P2P/IPC
+ip-26-0-161-123:32709:33129 [5] NCCL INFO Channel 10/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-161-138:96566:97010 [3] NCCL INFO Channel 07/0 : 35[3] -> 34[2] via P2P/IPC
+ip-26-0-165-202:153566:153980 [4] NCCL INFO Channel 09/0 : 92[4] -> 91[3] via P2P/IPC
+ip-26-0-160-242:1364905:1365328 [3] NCCL INFO Channel 13/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-161-123:32710:33125 [6] NCCL INFO Channel 07/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-160-242:1364908:1365332 [6] NCCL INFO Channel 09/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-161-123:32707:33124 [3] NCCL INFO Channel 13/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-160-242:1364907:1365329 [5] NCCL INFO Channel 07/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-165-202:153565:153985 [3] NCCL INFO Channel 05/0 : 91[3] -> 90[2] via P2P/IPC
+ip-26-0-161-178:2815970:2816397 [7] NCCL INFO Channel 01/0 : 63[7] -> 62[6] via P2P/IPC
+ip-26-0-165-164:1227466:1227887 [3] NCCL INFO Channel 13/0 : 83[3] -> 82[2] via P2P/IPC
+ip-26-0-161-153:2636091:2636512 [7] NCCL INFO Channel 03/0 : 55[7] -> 54[6] via P2P/IPC
+ip-26-0-165-164:1227469:1227882 [6] NCCL INFO Channel 10/0 : 86[6] -> 85[5] via P2P/IPC
+ip-26-0-165-164:1227468:1227883 [5] NCCL INFO Channel 03/0 : 85[5] -> 84[4] via P2P/IPC
+ip-26-0-161-142:62885:63304 [4] NCCL INFO Channel 04/0 : 44[4] -> 43[3] via P2P/IPC
+ip-26-0-161-78:114518:114937 [6] NCCL INFO Channel 04/0 : 70[6] -> 69[5] via P2P/IPC
+ip-26-0-168-30:235654:236070 [6] NCCL INFO Channel 12/0 : 110[6] -> 109[5] via P2P/IPC
+ip-26-0-168-30:235653:236067 [5] NCCL INFO Channel 09/0 : 109[5] -> 108[4] via P2P/IPC
+ip-26-0-161-138:96569:97006 [6] NCCL INFO Channel 07/0 : 38[6] -> 37[5] via P2P/IPC
+ip-26-0-161-78:114517:114935 [5] NCCL INFO Channel 02/0 : 69[5] -> 68[4] via P2P/IPC
+ip-26-0-161-78:114515:114936 [3] NCCL INFO Channel 11/0 : 67[3] -> 66[2] via P2P/IPC
+ip-26-0-168-30:235652:236066 [4] NCCL INFO Channel 10/0 : 108[4] -> 107[3] via P2P/IPC
+ip-26-0-161-138:96568:97007 [5] NCCL INFO Channel 02/0 : 37[5] -> 36[4] via P2P/IPC
+ip-26-0-168-30:235651:236068 [3] NCCL INFO Channel 14/0 : 107[3] -> 106[2] via P2P/IPC
+ip-26-0-161-123:32709:33129 [5] NCCL INFO Channel 11/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-161-138:96566:97010 [3] NCCL INFO Channel 09/0 : 35[3] -> 34[2] via P2P/IPC
+ip-26-0-161-153:2636091:2636512 [7] NCCL INFO Channel 04/0 : 55[7] -> 54[6] via P2P/IPC
+ip-26-0-160-225:124163:124588 [3] NCCL INFO Channel 05/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-225:124166:124583 [6] NCCL INFO Channel 07/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-142:62884:63299 [3] NCCL INFO Channel 01/0 : 43[3] -> 42[2] via P2P/IPC
+ip-26-0-161-178:2815970:2816397 [7] NCCL INFO Channel 02/0 : 63[7] -> 62[6] via P2P/IPC
+ip-26-0-161-123:32710:33125 [6] NCCL INFO Channel 09/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-160-225:124165:124584 [5] NCCL INFO Channel 09/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-123:32707:33124 [3] NCCL INFO Channel 14/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-160-242:1364908:1365332 [6] NCCL INFO Channel 10/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-165-202:153568:153982 [6] NCCL INFO Channel 02/0 : 94[6] -> 93[5] via P2P/IPC
+ip-26-0-165-202:153566:153980 [4] NCCL INFO Channel 10/0 : 92[4] -> 91[3] via P2P/IPC
+ip-26-0-160-242:1364905:1365328 [3] NCCL INFO Channel 14/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-160-242:1364907:1365329 [5] NCCL INFO Channel 09/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-165-202:153565:153985 [3] NCCL INFO Channel 06/0 : 91[3] -> 90[2] via P2P/IPC
+ip-26-0-165-164:1227466:1227887 [3] NCCL INFO Channel 14/0 : 83[3] -> 82[2] via P2P/IPC
+ip-26-0-165-164:1227469:1227882 [6] NCCL INFO Channel 11/0 : 86[6] -> 85[5] via P2P/IPC
+ip-26-0-165-164:1227468:1227883 [5] NCCL INFO Channel 06/0 : 85[5] -> 84[4] via P2P/IPC
+ip-26-0-161-142:62885:63304 [4] NCCL INFO Channel 09/0 : 44[4] -> 43[3] via P2P/IPC
+ip-26-0-161-178:2815970:2816397 [7] NCCL INFO Channel 03/0 : 63[7] -> 62[6] via P2P/IPC
+ip-26-0-165-131:1630208:1630650 [3] NCCL INFO Channel 01/0 : 75[3] -> 74[2] via P2P/IPC
+ip-26-0-161-78:114518:114937 [6] NCCL INFO Channel 07/0 : 70[6] -> 69[5] via P2P/IPC
+ip-26-0-161-78:114517:114935 [5] NCCL INFO Channel 03/0 : 69[5] -> 68[4] via P2P/IPC
+ip-26-0-161-123:32709:33129 [5] NCCL INFO Channel 14/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-161-138:96566:97010 [3] NCCL INFO Channel 11/0 : 35[3] -> 34[2] via P2P/IPC
+ip-26-0-161-78:114515:114936 [3] NCCL INFO Channel 12/0 : 67[3] -> 66[2] via P2P/IPC
+ip-26-0-161-123:32710:33125 [6] NCCL INFO Channel 10/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-161-123:32707:33124 [3] NCCL INFO Channel 15/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-160-225:124163:124588 [3] NCCL INFO Channel 06/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-225:124166:124583 [6] NCCL INFO Channel 09/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-225:124165:124584 [5] NCCL INFO Channel 10/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-202:153567:153979 [5] NCCL INFO Channel 01/0 : 93[5] -> 92[4] via P2P/IPC
+ip-26-0-161-142:62884:63299 [3] NCCL INFO Channel 03/0 : 43[3] -> 42[2] via P2P/IPC
+ip-26-0-160-242:1364908:1365332 [6] NCCL INFO Channel 11/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-165-202:153566:153980 [4] NCCL INFO Channel 12/0 : 92[4] -> 91[3] via P2P/IPC
+ip-26-0-165-59:1557269:1557684 [3] NCCL INFO Channel 15/0 : 99[3] -> 98[2] via P2P/IPC
+ip-26-0-160-242:1364905:1365328 [3] NCCL INFO Channel 15/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-160-242:1364907:1365329 [5] NCCL INFO Channel 10/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-165-164:1227466:1227887 [3] NCCL INFO Channel 15/0 : 83[3] -> 82[2] via P2P/IPC
+ip-26-0-161-138:96566:97010 [3] NCCL INFO Channel 12/0 : 35[3] -> 34[2] via P2P/IPC
+ip-26-0-165-131:1630208:1630650 [3] NCCL INFO Channel 03/0 : 75[3] -> 74[2] via P2P/IPC
+ip-26-0-165-202:153568:153982 [6] NCCL INFO Channel 03/0 : 94[6] -> 93[5] via P2P/IPC
+ip-26-0-168-34:272157:272572 [7] NCCL INFO Channel 02/0 : 119[7] -> 118[6] via P2P/IPC
+ip-26-0-161-153:2636091:2636512 [7] NCCL INFO Channel 05/0 : 55[7] -> 54[6] via P2P/IPC
+ip-26-0-165-164:1227469:1227882 [6] NCCL INFO Channel 12/0 : 86[6] -> 85[5] via P2P/IPC
+ip-26-0-168-52:2722051:2722465 [6] NCCL INFO Channel 07/0 : 126[6] -> 125[5] via P2P/IPC
+ip-26-0-165-164:1227468:1227883 [5] NCCL INFO Channel 07/0 : 85[5] -> 84[4] via P2P/IPC
+ip-26-0-165-202:153565:153985 [3] NCCL INFO Channel 07/0 : 91[3] -> 90[2] via P2P/IPC
+ip-26-0-165-59:1557271:1557690 [5] NCCL INFO Channel 14/0 : 101[5] -> 100[4] via P2P/IPC
+ip-26-0-168-52:2722048:2722469 [3] NCCL INFO Channel 09/0 : 123[3] -> 122[2] via P2P/IPC
+ip-26-0-161-142:62887:63305 [6] NCCL INFO Channel 01/0 : 46[6] -> 45[5] via P2P/IPC
+ip-26-0-161-178:2815970:2816397 [7] NCCL INFO Channel 04/0 : 63[7] -> 62[6] via P2P/IPC
+ip-26-0-161-142:62885:63304 [4] NCCL INFO Channel 10/0 : 44[4] -> 43[3] via P2P/IPC
+ip-26-0-161-138:96566:97010 [3] NCCL INFO Channel 13/0 : 35[3] -> 34[2] via P2P/IPC
+ip-26-0-161-123:32709:33129 [5] NCCL INFO Channel 15/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-161-78:114518:114937 [6] NCCL INFO Channel 09/0 : 70[6] -> 69[5] via P2P/IPC
+ip-26-0-161-123:32710:33125 [6] NCCL INFO Channel 11/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-161-78:114517:114935 [5] NCCL INFO Channel 06/0 : 69[5] -> 68[4] via P2P/IPC
+ip-26-0-161-78:114515:114936 [3] NCCL INFO Channel 13/0 : 67[3] -> 66[2] via P2P/IPC
+ip-26-0-161-142:62884:63299 [3] NCCL INFO Channel 04/0 : 43[3] -> 42[2] via P2P/IPC
+ip-26-0-160-242:1364908:1365332 [6] NCCL INFO Channel 12/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-160-225:124166:124583 [6] NCCL INFO Channel 10/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-225:124163:124588 [3] NCCL INFO Channel 07/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-138:96566:97010 [3] NCCL INFO Channel 14/0 : 35[3] -> 34[2] via P2P/IPC
+ip-26-0-161-142:62886:63303 [5] NCCL INFO Channel 01/0 : 45[5] -> 44[4] via P2P/IPC
+ip-26-0-160-242:1364907:1365329 [5] NCCL INFO Channel 11/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-165-202:153567:153979 [5] NCCL INFO Channel 02/0 : 93[5] -> 92[4] via P2P/IPC
+ip-26-0-161-103:96895:97307 [7] NCCL INFO Channel 05/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-165-164:1227469:1227882 [6] NCCL INFO Channel 15/0 : 86[6] -> 85[5] via P2P/IPC
+ip-26-0-168-52:2722051:2722465 [6] NCCL INFO Channel 09/0 : 126[6] -> 125[5] via P2P/IPC
+ip-26-0-165-164:1227468:1227883 [5] NCCL INFO Channel 09/0 : 85[5] -> 84[4] via P2P/IPC
+ip-26-0-168-52:2722048:2722469 [3] NCCL INFO Channel 11/0 : 123[3] -> 122[2] via P2P/IPC
+ip-26-0-165-202:153568:153982 [6] NCCL INFO Channel 04/0 : 94[6] -> 93[5] via P2P/IPC
+ip-26-0-165-202:153565:153985 [3] NCCL INFO Channel 09/0 : 91[3] -> 90[2] via P2P/IPC
+ip-26-0-161-138:96566:97010 [3] NCCL INFO Channel 15/0 : 35[3] -> 34[2] via P2P/IPC
+ip-26-0-161-178:2815970:2816397 [7] NCCL INFO Channel 05/0 : 63[7] -> 62[6] via P2P/IPC
+ip-26-0-168-30:235652:236066 [4] NCCL INFO Channel 12/0 : 108[4] -> 107[3] via P2P/IPC
+ip-26-0-161-78:114518:114937 [6] NCCL INFO Channel 10/0 : 70[6] -> 69[5] via P2P/IPC
+ip-26-0-161-142:62887:63305 [6] NCCL INFO Channel 02/0 : 46[6] -> 45[5] via P2P/IPC
+ip-26-0-161-138:96569:97006 [6] NCCL INFO Channel 09/0 : 38[6] -> 37[5] via P2P/IPC
+ip-26-0-160-225:124166:124583 [6] NCCL INFO Channel 11/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-78:114517:114935 [5] NCCL INFO Channel 07/0 : 69[5] -> 68[4] via P2P/IPC
+ip-26-0-161-78:114515:114936 [3] NCCL INFO Channel 14/0 : 67[3] -> 66[2] via P2P/IPC
+ip-26-0-160-225:124163:124588 [3] NCCL INFO Channel 09/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-138:96568:97007 [5] NCCL INFO Channel 03/0 : 37[5] -> 36[4] via P2P/IPC
+ip-26-0-161-142:62885:63304 [4] NCCL INFO Channel 12/0 : 44[4] -> 43[3] via P2P/IPC
+ip-26-0-165-202:153567:153979 [5] NCCL INFO Channel 03/0 : 93[5] -> 92[4] via P2P/IPC
+ip-26-0-168-52:2722051:2722465 [6] NCCL INFO Channel 10/0 : 126[6] -> 125[5] via P2P/IPC
+ip-26-0-161-142:62884:63299 [3] NCCL INFO Channel 05/0 : 43[3] -> 42[2] via P2P/IPC
+ip-26-0-160-242:1364908:1365332 [6] NCCL INFO Channel 15/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-160-242:1364907:1365329 [5] NCCL INFO Channel 14/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-168-52:2722048:2722469 [3] NCCL INFO Channel 12/0 : 123[3] -> 122[2] via P2P/IPC
+ip-26-0-161-142:62886:63303 [5] NCCL INFO Channel 02/0 : 45[5] -> 44[4] via P2P/IPC
+ip-26-0-165-202:153565:153985 [3] NCCL INFO Channel 11/0 : 91[3] -> 90[2] via P2P/IPC
+ip-26-0-161-153:2636091:2636512 [7] NCCL INFO Channel 09/0 : 55[7] -> 54[6] via P2P/IPC
+ip-26-0-161-153:2636091:2636512 [7] NCCL INFO Channel 10/0 : 55[7] -> 54[6] via P2P/IPC
+ip-26-0-161-178:2815970:2816397 [7] NCCL INFO Channel 09/0 : 63[7] -> 62[6] via P2P/IPC
+ip-26-0-168-52:2722050:2722471 [5] NCCL INFO Channel 01/0 : 125[5] -> 124[4] via P2P/IPC
+ip-26-0-160-225:124166:124583 [6] NCCL INFO Channel 12/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-202:153568:153982 [6] NCCL INFO Channel 07/0 : 94[6] -> 93[5] via P2P/IPC
+ip-26-0-161-153:2636091:2636512 [7] NCCL INFO Channel 11/0 : 55[7] -> 54[6] via P2P/IPC
+ip-26-0-161-138:96569:97006 [6] NCCL INFO Channel 10/0 : 38[6] -> 37[5] via P2P/IPC
+ip-26-0-160-225:124163:124588 [3] NCCL INFO Channel 11/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-165-202:153567:153979 [5] NCCL INFO Channel 06/0 : 93[5] -> 92[4] via P2P/IPC
+ip-26-0-161-78:114518:114937 [6] NCCL INFO Channel 11/0 : 70[6] -> 69[5] via P2P/IPC
+ip-26-0-161-138:96568:97007 [5] NCCL INFO Channel 06/0 : 37[5] -> 36[4] via P2P/IPC
+ip-26-0-161-78:114517:114935 [5] NCCL INFO Channel 09/0 : 69[5] -> 68[4] via P2P/IPC
+ip-26-0-161-78:114515:114936 [3] NCCL INFO Channel 15/0 : 67[3] -> 66[2] via P2P/IPC
+ip-26-0-161-142:62887:63305 [6] NCCL INFO Channel 03/0 : 46[6] -> 45[5] via P2P/IPC
+ip-26-0-168-34:272157:272572 [7] NCCL INFO Channel 03/0 : 119[7] -> 118[6] via P2P/IPC
+ip-26-0-168-52:2722051:2722465 [6] NCCL INFO Channel 11/0 : 126[6] -> 125[5] via P2P/IPC
+ip-26-0-168-52:2722048:2722469 [3] NCCL INFO Channel 13/0 : 123[3] -> 122[2] via P2P/IPC
+ip-26-0-161-178:2815967:2816399 [4] NCCL INFO Channel 01/0 : 60[4] -> 59[3] via P2P/IPC
+ip-26-0-161-142:62884:63299 [3] NCCL INFO Channel 06/0 : 43[3] -> 42[2] via P2P/IPC
+ip-26-0-161-142:62886:63303 [5] NCCL INFO Channel 03/0 : 45[5] -> 44[4] via P2P/IPC
+ip-26-0-160-225:124166:124583 [6] NCCL INFO Channel 15/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-225:124163:124588 [3] NCCL INFO Channel 12/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-168-52:2722050:2722471 [5] NCCL INFO Channel 02/0 : 125[5] -> 124[4] via P2P/IPC
+ip-26-0-165-202:153565:153985 [3] NCCL INFO Channel 12/0 : 91[3] -> 90[2] via P2P/IPC
+ip-26-0-161-178:2815970:2816397 [7] NCCL INFO Channel 10/0 : 63[7] -> 62[6] via P2P/IPC
+ip-26-0-165-202:153568:153982 [6] NCCL INFO Channel 09/0 : 94[6] -> 93[5] via P2P/IPC
+ip-26-0-168-34:272157:272572 [7] NCCL INFO Channel 04/0 : 119[7] -> 118[6] via P2P/IPC
+ip-26-0-165-202:153567:153979 [5] NCCL INFO Channel 07/0 : 93[5] -> 92[4] via P2P/IPC
+ip-26-0-168-30:235651:236068 [3] NCCL INFO Channel 15/0 : 107[3] -> 106[2] via P2P/IPC
+ip-26-0-161-78:114518:114937 [6] NCCL INFO Channel 12/0 : 70[6] -> 69[5] via P2P/IPC
+ip-26-0-161-78:114517:114935 [5] NCCL INFO Channel 10/0 : 69[5] -> 68[4] via P2P/IPC
+ip-26-0-168-30:235654:236070 [6] NCCL INFO Channel 15/0 : 110[6] -> 109[5] via P2P/IPC
+ip-26-0-161-123:32710:33125 [6] NCCL INFO Channel 12/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-168-52:2722051:2722465 [6] NCCL INFO Channel 12/0 : 126[6] -> 125[5] via P2P/IPC
+ip-26-0-168-52:2722048:2722469 [3] NCCL INFO Channel 14/0 : 123[3] -> 122[2] via P2P/IPC
+ip-26-0-161-153:2636091:2636512 [7] NCCL INFO Channel 12/0 : 55[7] -> 54[6] via P2P/IPC
+ip-26-0-160-225:124163:124588 [3] NCCL INFO Channel 13/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-168-52:2722050:2722471 [5] NCCL INFO Channel 03/0 : 125[5] -> 124[4] via P2P/IPC
+ip-26-0-161-123:32710:33125 [6] NCCL INFO Channel 15/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-160-225:124165:124584 [5] NCCL INFO Channel 11/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-178:2815967:2816399 [4] NCCL INFO Channel 02/0 : 60[4] -> 59[3] via P2P/IPC
+ip-26-0-161-78:114518:114937 [6] NCCL INFO Channel 15/0 : 70[6] -> 69[5] via P2P/IPC
+ip-26-0-168-34:272157:272572 [7] NCCL INFO Channel 05/0 : 119[7] -> 118[6] via P2P/IPC
+ip-26-0-168-30:235653:236067 [5] NCCL INFO Channel 10/0 : 109[5] -> 108[4] via P2P/IPC
+ip-26-0-161-178:2815970:2816397 [7] NCCL INFO Channel 11/0 : 63[7] -> 62[6] via P2P/IPC
+ip-26-0-161-78:114517:114935 [5] NCCL INFO Channel 11/0 : 69[5] -> 68[4] via P2P/IPC
+ip-26-0-168-52:2722051:2722465 [6] NCCL INFO Channel 15/0 : 126[6] -> 125[5] via P2P/IPC
+ip-26-0-168-52:2722048:2722469 [3] NCCL INFO Channel 15/0 : 123[3] -> 122[2] via P2P/IPC
+ip-26-0-168-52:2722050:2722471 [5] NCCL INFO Channel 06/0 : 125[5] -> 124[4] via P2P/IPC
+ip-26-0-161-178:2815967:2816399 [4] NCCL INFO Channel 04/0 : 60[4] -> 59[3] via P2P/IPC
+ip-26-0-161-178:2815970:2816397 [7] NCCL INFO Channel 12/0 : 63[7] -> 62[6] via P2P/IPC
+ip-26-0-168-30:235653:236067 [5] NCCL INFO Channel 11/0 : 109[5] -> 108[4] via P2P/IPC
+ip-26-0-161-153:2636091:2636512 [7] NCCL INFO Channel 13/0 : 55[7] -> 54[6] via P2P/IPC
+ip-26-0-161-153:2636088:2636510 [4] NCCL INFO Channel 01/0 : 52[4] -> 51[3] via P2P/IPC
+ip-26-0-165-59:1557271:1557690 [5] NCCL INFO Channel 15/0 : 101[5] -> 100[4] via P2P/IPC
+ip-26-0-168-52:2722050:2722471 [5] NCCL INFO Channel 07/0 : 125[5] -> 124[4] via P2P/IPC
+ip-26-0-165-131:1630208:1630650 [3] NCCL INFO Channel 04/0 : 75[3] -> 74[2] via P2P/IPC
+ip-26-0-161-178:2815967:2816399 [4] NCCL INFO Channel 09/0 : 60[4] -> 59[3] via P2P/IPC
+ip-26-0-161-103:96895:97307 [7] NCCL INFO Channel 09/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-161-178:2815970:2816397 [7] NCCL INFO Channel 13/0 : 63[7] -> 62[6] via P2P/IPC
+ip-26-0-161-142:62886:63303 [5] NCCL INFO Channel 06/0 : 45[5] -> 44[4] via P2P/IPC
+ip-26-0-161-142:62886:63303 [5] NCCL INFO Channel 07/0 : 45[5] -> 44[4] via P2P/IPC
+ip-26-0-161-142:62886:63303 [5] NCCL INFO Channel 09/0 : 45[5] -> 44[4] via P2P/IPC
+ip-26-0-161-142:62886:63303 [5] NCCL INFO Channel 10/0 : 45[5] -> 44[4] via P2P/IPC
+ip-26-0-161-142:62886:63303 [5] NCCL INFO Channel 11/0 : 45[5] -> 44[4] via P2P/IPC
+ip-26-0-161-142:62887:63305 [6] NCCL INFO Channel 04/0 : 46[6] -> 45[5] via P2P/IPC
+ip-26-0-161-142:62886:63303 [5] NCCL INFO Channel 14/0 : 45[5] -> 44[4] via P2P/IPC
+ip-26-0-161-142:62884:63299 [3] NCCL INFO Channel 07/0 : 43[3] -> 42[2] via P2P/IPC
+ip-26-0-168-34:272157:272572 [7] NCCL INFO Channel 09/0 : 119[7] -> 118[6] via P2P/IPC
+ip-26-0-165-131:1630208:1630650 [3] NCCL INFO Channel 05/0 : 75[3] -> 74[2] via P2P/IPC
+ip-26-0-160-225:124163:124588 [3] NCCL INFO Channel 14/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-153:2636088:2636510 [4] NCCL INFO Channel 02/0 : 52[4] -> 51[3] via P2P/IPC
+ip-26-0-165-131:1630211:1630649 [6] NCCL INFO Channel 01/0 : 78[6] -> 77[5] via P2P/IPC
+ip-26-0-161-178:2815967:2816399 [4] NCCL INFO Channel 10/0 : 60[4] -> 59[3] via P2P/IPC
+ip-26-0-161-142:62887:63305 [6] NCCL INFO Channel 07/0 : 46[6] -> 45[5] via P2P/IPC
+ip-26-0-161-142:62886:63303 [5] NCCL INFO Channel 15/0 : 45[5] -> 44[4] via P2P/IPC
+ip-26-0-168-34:272154:272569 [4] NCCL INFO Channel 01/0 : 116[4] -> 115[3] via P2P/IPC
+ip-26-0-161-142:62884:63299 [3] NCCL INFO Channel 09/0 : 43[3] -> 42[2] via P2P/IPC
+ip-26-0-165-202:153565:153985 [3] NCCL INFO Channel 13/0 : 91[3] -> 90[2] via P2P/IPC
+ip-26-0-168-34:272157:272572 [7] NCCL INFO Channel 10/0 : 119[7] -> 118[6] via P2P/IPC
+ip-26-0-165-131:1630208:1630650 [3] NCCL INFO Channel 06/0 : 75[3] -> 74[2] via P2P/IPC
+ip-26-0-165-202:153565:153985 [3] NCCL INFO Channel 14/0 : 91[3] -> 90[2] via P2P/IPC
+ip-26-0-165-131:1630211:1630649 [6] NCCL INFO Channel 02/0 : 78[6] -> 77[5] via P2P/IPC
+ip-26-0-165-202:153567:153979 [5] NCCL INFO Channel 09/0 : 93[5] -> 92[4] via P2P/IPC
+ip-26-0-161-178:2815967:2816399 [4] NCCL INFO Channel 12/0 : 60[4] -> 59[3] via P2P/IPC
+ip-26-0-161-153:2636088:2636510 [4] NCCL INFO Channel 04/0 : 52[4] -> 51[3] via P2P/IPC
+ip-26-0-161-142:62887:63305 [6] NCCL INFO Channel 09/0 : 46[6] -> 45[5] via P2P/IPC
+ip-26-0-165-202:153568:153982 [6] NCCL INFO Channel 10/0 : 94[6] -> 93[5] via P2P/IPC
+ip-26-0-161-142:62884:63299 [3] NCCL INFO Channel 11/0 : 43[3] -> 42[2] via P2P/IPC
+ip-26-0-165-131:1630208:1630650 [3] NCCL INFO Channel 07/0 : 75[3] -> 74[2] via P2P/IPC
+ip-26-0-168-34:272154:272569 [4] NCCL INFO Channel 02/0 : 116[4] -> 115[3] via P2P/IPC
+ip-26-0-168-34:272157:272572 [7] NCCL INFO Channel 11/0 : 119[7] -> 118[6] via P2P/IPC
+ip-26-0-165-202:153565:153985 [3] NCCL INFO Channel 15/0 : 91[3] -> 90[2] via P2P/IPC
+ip-26-0-165-131:1630211:1630649 [6] NCCL INFO Channel 03/0 : 78[6] -> 77[5] via P2P/IPC
+ip-26-0-161-142:62887:63305 [6] NCCL INFO Channel 10/0 : 46[6] -> 45[5] via P2P/IPC
+ip-26-0-165-202:153567:153979 [5] NCCL INFO Channel 10/0 : 93[5] -> 92[4] via P2P/IPC
+ip-26-0-161-142:62884:63299 [3] NCCL INFO Channel 12/0 : 43[3] -> 42[2] via P2P/IPC
+ip-26-0-160-225:124165:124584 [5] NCCL INFO Channel 14/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-153:2636088:2636510 [4] NCCL INFO Channel 09/0 : 52[4] -> 51[3] via P2P/IPC
+ip-26-0-165-131:1630208:1630650 [3] NCCL INFO Channel 09/0 : 75[3] -> 74[2] via P2P/IPC
+ip-26-0-165-202:153568:153982 [6] NCCL INFO Channel 11/0 : 94[6] -> 93[5] via P2P/IPC
+ip-26-0-161-103:96895:97307 [7] NCCL INFO Channel 10/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-160-225:124163:124588 [3] NCCL INFO Channel 15/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-165-131:1630211:1630649 [6] NCCL INFO Channel 04/0 : 78[6] -> 77[5] via P2P/IPC
+ip-26-0-161-142:62887:63305 [6] NCCL INFO Channel 11/0 : 46[6] -> 45[5] via P2P/IPC
+ip-26-0-161-138:96569:97006 [6] NCCL INFO Channel 11/0 : 38[6] -> 37[5] via P2P/IPC
+ip-26-0-165-202:153567:153979 [5] NCCL INFO Channel 11/0 : 93[5] -> 92[4] via P2P/IPC
+ip-26-0-168-34:272154:272569 [4] NCCL INFO Channel 04/0 : 116[4] -> 115[3] via P2P/IPC
+ip-26-0-161-142:62884:63299 [3] NCCL INFO Channel 13/0 : 43[3] -> 42[2] via P2P/IPC
+ip-26-0-161-138:96568:97007 [5] NCCL INFO Channel 07/0 : 37[5] -> 36[4] via P2P/IPC
+ip-26-0-168-34:272157:272572 [7] NCCL INFO Channel 12/0 : 119[7] -> 118[6] via P2P/IPC
+ip-26-0-165-202:153568:153982 [6] NCCL INFO Channel 12/0 : 94[6] -> 93[5] via P2P/IPC
+ip-26-0-165-131:1630208:1630650 [3] NCCL INFO Channel 11/0 : 75[3] -> 74[2] via P2P/IPC
+ip-26-0-161-142:62887:63305 [6] NCCL INFO Channel 12/0 : 46[6] -> 45[5] via P2P/IPC
+ip-26-0-165-131:1630211:1630649 [6] NCCL INFO Channel 07/0 : 78[6] -> 77[5] via P2P/IPC
+ip-26-0-161-142:62884:63299 [3] NCCL INFO Channel 14/0 : 43[3] -> 42[2] via P2P/IPC
+ip-26-0-161-153:2636088:2636510 [4] NCCL INFO Channel 10/0 : 52[4] -> 51[3] via P2P/IPC
+ip-26-0-161-103:96895:97307 [7] NCCL INFO Channel 11/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-161-153:2636087:2636506 [3] NCCL INFO Channel 01/0 : 51[3] -> 50[2] via P2P/IPC
+ip-26-0-168-34:272154:272569 [4] NCCL INFO Channel 09/0 : 116[4] -> 115[3] via P2P/IPC
+ip-26-0-161-142:62887:63305 [6] NCCL INFO Channel 15/0 : 46[6] -> 45[5] via P2P/IPC
+ip-26-0-161-178:2815966:2816396 [3] NCCL INFO Channel 01/0 : 59[3] -> 58[2] via P2P/IPC
+ip-26-0-168-34:272157:272572 [7] NCCL INFO Channel 13/0 : 119[7] -> 118[6] via P2P/IPC
+ip-26-0-165-131:1630208:1630650 [3] NCCL INFO Channel 12/0 : 75[3] -> 74[2] via P2P/IPC
+ip-26-0-165-131:1630210:1630653 [5] NCCL INFO Channel 01/0 : 77[5] -> 76[4] via P2P/IPC
+ip-26-0-161-142:62884:63299 [3] NCCL INFO Channel 15/0 : 43[3] -> 42[2] via P2P/IPC
+ip-26-0-165-131:1630211:1630649 [6] NCCL INFO Channel 09/0 : 78[6] -> 77[5] via P2P/IPC
+ip-26-0-161-178:2815969:2816398 [6] NCCL INFO Channel 01/0 : 62[6] -> 61[5] via P2P/IPC
+ip-26-0-161-153:2636088:2636510 [4] NCCL INFO Channel 12/0 : 52[4] -> 51[3] via P2P/IPC
+ip-26-0-161-178:2815966:2816396 [3] NCCL INFO Channel 03/0 : 59[3] -> 58[2] via P2P/IPC
+ip-26-0-160-225:124165:124584 [5] NCCL INFO Channel 15/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-153:2636087:2636506 [3] NCCL INFO Channel 03/0 : 51[3] -> 50[2] via P2P/IPC
+ip-26-0-168-52:2722050:2722471 [5] NCCL INFO Channel 09/0 : 125[5] -> 124[4] via P2P/IPC
+ip-26-0-161-103:96892:97308 [4] NCCL INFO Channel 01/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-161-178:2815968:2816394 [5] NCCL INFO Channel 01/0 : 61[5] -> 60[4] via P2P/IPC
+ip-26-0-161-103:96895:97307 [7] NCCL INFO Channel 12/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-161-178:2815969:2816398 [6] NCCL INFO Channel 02/0 : 62[6] -> 61[5] via P2P/IPC
+ip-26-0-165-131:1630208:1630650 [3] NCCL INFO Channel 13/0 : 75[3] -> 74[2] via P2P/IPC
+ip-26-0-165-164:1227468:1227883 [5] NCCL INFO Channel 10/0 : 85[5] -> 84[4] via P2P/IPC
+ip-26-0-168-34:272154:272569 [4] NCCL INFO Channel 10/0 : 116[4] -> 115[3] via P2P/IPC
+ip-26-0-165-131:1630210:1630653 [5] NCCL INFO Channel 02/0 : 77[5] -> 76[4] via P2P/IPC
+ip-26-0-165-131:1630211:1630649 [6] NCCL INFO Channel 10/0 : 78[6] -> 77[5] via P2P/IPC
+ip-26-0-161-178:2815966:2816396 [3] NCCL INFO Channel 04/0 : 59[3] -> 58[2] via P2P/IPC
+ip-26-0-161-178:2815968:2816394 [5] NCCL INFO Channel 02/0 : 61[5] -> 60[4] via P2P/IPC
+ip-26-0-161-178:2815969:2816398 [6] NCCL INFO Channel 03/0 : 62[6] -> 61[5] via P2P/IPC
+ip-26-0-161-153:2636087:2636506 [3] NCCL INFO Channel 04/0 : 51[3] -> 50[2] via P2P/IPC
+ip-26-0-168-34:272154:272569 [4] NCCL INFO Channel 12/0 : 116[4] -> 115[3] via P2P/IPC
+ip-26-0-161-103:96892:97308 [4] NCCL INFO Channel 02/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-165-131:1630208:1630650 [3] NCCL INFO Channel 14/0 : 75[3] -> 74[2] via P2P/IPC
+ip-26-0-161-153:2636090:2636505 [6] NCCL INFO Channel 01/0 : 54[6] -> 53[5] via P2P/IPC
+ip-26-0-161-178:2815966:2816396 [3] NCCL INFO Channel 05/0 : 59[3] -> 58[2] via P2P/IPC
+ip-26-0-161-103:96895:97307 [7] NCCL INFO Channel 13/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-165-131:1630210:1630653 [5] NCCL INFO Channel 03/0 : 77[5] -> 76[4] via P2P/IPC
+ip-26-0-165-131:1630211:1630649 [6] NCCL INFO Channel 11/0 : 78[6] -> 77[5] via P2P/IPC
+ip-26-0-161-178:2815968:2816394 [5] NCCL INFO Channel 03/0 : 61[5] -> 60[4] via P2P/IPC
+ip-26-0-161-178:2815969:2816398 [6] NCCL INFO Channel 04/0 : 62[6] -> 61[5] via P2P/IPC
+ip-26-0-161-178:2815966:2816396 [3] NCCL INFO Channel 06/0 : 59[3] -> 58[2] via P2P/IPC
+ip-26-0-161-103:96892:97308 [4] NCCL INFO Channel 04/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-165-131:1630208:1630650 [3] NCCL INFO Channel 15/0 : 75[3] -> 74[2] via P2P/IPC
+ip-26-0-165-131:1630210:1630653 [5] NCCL INFO Channel 06/0 : 77[5] -> 76[4] via P2P/IPC
+ip-26-0-165-131:1630211:1630649 [6] NCCL INFO Channel 12/0 : 78[6] -> 77[5] via P2P/IPC
+ip-26-0-168-34:272153:272573 [3] NCCL INFO Channel 01/0 : 115[3] -> 114[2] via P2P/IPC
+ip-26-0-161-103:96892:97308 [4] NCCL INFO Channel 09/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-161-178:2815968:2816394 [5] NCCL INFO Channel 06/0 : 61[5] -> 60[4] via P2P/IPC
+ip-26-0-161-178:2815969:2816398 [6] NCCL INFO Channel 07/0 : 62[6] -> 61[5] via P2P/IPC
+ip-26-0-161-178:2815966:2816396 [3] NCCL INFO Channel 07/0 : 59[3] -> 58[2] via P2P/IPC
+ip-26-0-168-34:272156:272570 [6] NCCL INFO Channel 01/0 : 118[6] -> 117[5] via P2P/IPC
+ip-26-0-161-103:96892:97308 [4] NCCL INFO Channel 10/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-165-202:153567:153979 [5] NCCL INFO Channel 14/0 : 93[5] -> 92[4] via P2P/IPC
+ip-26-0-168-34:272153:272573 [3] NCCL INFO Channel 03/0 : 115[3] -> 114[2] via P2P/IPC
+ip-26-0-165-131:1630210:1630653 [5] NCCL INFO Channel 07/0 : 77[5] -> 76[4] via P2P/IPC
+ip-26-0-165-131:1630211:1630649 [6] NCCL INFO Channel 15/0 : 78[6] -> 77[5] via P2P/IPC
+ip-26-0-161-103:96892:97308 [4] NCCL INFO Channel 12/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-161-178:2815968:2816394 [5] NCCL INFO Channel 07/0 : 61[5] -> 60[4] via P2P/IPC
+ip-26-0-161-178:2815969:2816398 [6] NCCL INFO Channel 09/0 : 62[6] -> 61[5] via P2P/IPC
+ip-26-0-161-178:2815966:2816396 [3] NCCL INFO Channel 09/0 : 59[3] -> 58[2] via P2P/IPC
+ip-26-0-168-34:272156:272570 [6] NCCL INFO Channel 02/0 : 118[6] -> 117[5] via P2P/IPC
+ip-26-0-168-34:272153:272573 [3] NCCL INFO Channel 04/0 : 115[3] -> 114[2] via P2P/IPC
+ip-26-0-165-131:1630210:1630653 [5] NCCL INFO Channel 09/0 : 77[5] -> 76[4] via P2P/IPC
+ip-26-0-161-103:96891:97311 [3] NCCL INFO Channel 01/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-161-153:2636090:2636505 [6] NCCL INFO Channel 02/0 : 54[6] -> 53[5] via P2P/IPC
+ip-26-0-168-34:272156:272570 [6] NCCL INFO Channel 03/0 : 118[6] -> 117[5] via P2P/IPC
+ip-26-0-161-78:114517:114935 [5] NCCL INFO Channel 14/0 : 69[5] -> 68[4] via P2P/IPC
+ip-26-0-168-34:272153:272573 [3] NCCL INFO Channel 05/0 : 115[3] -> 114[2] via P2P/IPC
+ip-26-0-161-103:96891:97311 [3] NCCL INFO Channel 03/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-161-153:2636090:2636505 [6] NCCL INFO Channel 03/0 : 54[6] -> 53[5] via P2P/IPC
+ip-26-0-161-153:2636089:2636508 [5] NCCL INFO Channel 01/0 : 53[5] -> 52[4] via P2P/IPC
+ip-26-0-161-153:2636090:2636505 [6] NCCL INFO Channel 04/0 : 54[6] -> 53[5] via P2P/IPC
+ip-26-0-161-153:2636087:2636506 [3] NCCL INFO Channel 05/0 : 51[3] -> 50[2] via P2P/IPC
+ip-26-0-161-103:96891:97311 [3] NCCL INFO Channel 04/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-168-34:272156:272570 [6] NCCL INFO Channel 04/0 : 118[6] -> 117[5] via P2P/IPC
+ip-26-0-168-34:272153:272573 [3] NCCL INFO Channel 06/0 : 115[3] -> 114[2] via P2P/IPC
+ip-26-0-168-34:272155:272574 [5] NCCL INFO Channel 01/0 : 117[5] -> 116[4] via P2P/IPC
+ip-26-0-161-103:96894:97312 [6] NCCL INFO Channel 01/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-161-153:2636089:2636508 [5] NCCL INFO Channel 02/0 : 53[5] -> 52[4] via P2P/IPC
+ip-26-0-161-153:2636090:2636505 [6] NCCL INFO Channel 07/0 : 54[6] -> 53[5] via P2P/IPC
+ip-26-0-161-153:2636087:2636506 [3] NCCL INFO Channel 06/0 : 51[3] -> 50[2] via P2P/IPC
+ip-26-0-161-103:96891:97311 [3] NCCL INFO Channel 05/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-165-202:153567:153979 [5] NCCL INFO Channel 15/0 : 93[5] -> 92[4] via P2P/IPC
+ip-26-0-161-103:96894:97312 [6] NCCL INFO Channel 02/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-161-178:2815969:2816398 [6] NCCL INFO Channel 10/0 : 62[6] -> 61[5] via P2P/IPC
+ip-26-0-161-178:2815966:2816396 [3] NCCL INFO Channel 11/0 : 59[3] -> 58[2] via P2P/IPC
+ip-26-0-161-153:2636089:2636508 [5] NCCL INFO Channel 03/0 : 53[5] -> 52[4] via P2P/IPC
+ip-26-0-161-153:2636090:2636505 [6] NCCL INFO Channel 09/0 : 54[6] -> 53[5] via P2P/IPC
+ip-26-0-161-103:96891:97311 [3] NCCL INFO Channel 06/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-161-103:96893:97310 [5] NCCL INFO Channel 01/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-161-153:2636087:2636506 [3] NCCL INFO Channel 07/0 : 51[3] -> 50[2] via P2P/IPC
+ip-26-0-165-202:153568:153982 [6] NCCL INFO Channel 15/0 : 94[6] -> 93[5] via P2P/IPC
+ip-26-0-161-178:2815969:2816398 [6] NCCL INFO Channel 11/0 : 62[6] -> 61[5] via P2P/IPC
+ip-26-0-161-178:2815966:2816396 [3] NCCL INFO Channel 12/0 : 59[3] -> 58[2] via P2P/IPC
+ip-26-0-161-103:96894:97312 [6] NCCL INFO Channel 03/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-168-34:272156:272570 [6] NCCL INFO Channel 07/0 : 118[6] -> 117[5] via P2P/IPC
+ip-26-0-161-103:96891:97311 [3] NCCL INFO Channel 07/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-161-103:96893:97310 [5] NCCL INFO Channel 02/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-161-153:2636089:2636508 [5] NCCL INFO Channel 06/0 : 53[5] -> 52[4] via P2P/IPC
+ip-26-0-168-34:272156:272570 [6] NCCL INFO Channel 09/0 : 118[6] -> 117[5] via P2P/IPC
+ip-26-0-161-153:2636090:2636505 [6] NCCL INFO Channel 10/0 : 54[6] -> 53[5] via P2P/IPC
+ip-26-0-161-178:2815969:2816398 [6] NCCL INFO Channel 12/0 : 62[6] -> 61[5] via P2P/IPC
+ip-26-0-161-138:96568:97007 [5] NCCL INFO Channel 09/0 : 37[5] -> 36[4] via P2P/IPC
+ip-26-0-161-103:96894:97312 [6] NCCL INFO Channel 04/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-161-178:2815966:2816396 [3] NCCL INFO Channel 13/0 : 59[3] -> 58[2] via P2P/IPC
+ip-26-0-168-34:272153:272573 [3] NCCL INFO Channel 07/0 : 115[3] -> 114[2] via P2P/IPC
+ip-26-0-161-153:2636087:2636506 [3] NCCL INFO Channel 09/0 : 51[3] -> 50[2] via P2P/IPC
+ip-26-0-161-178:2815968:2816394 [5] NCCL INFO Channel 09/0 : 61[5] -> 60[4] via P2P/IPC
+ip-26-0-168-34:272156:272570 [6] NCCL INFO Channel 10/0 : 118[6] -> 117[5] via P2P/IPC
+ip-26-0-161-103:96891:97311 [3] NCCL INFO Channel 09/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-168-34:272153:272573 [3] NCCL INFO Channel 09/0 : 115[3] -> 114[2] via P2P/IPC
+ip-26-0-161-103:96893:97310 [5] NCCL INFO Channel 03/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-161-103:96894:97312 [6] NCCL INFO Channel 07/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-161-178:2815969:2816398 [6] NCCL INFO Channel 15/0 : 62[6] -> 61[5] via P2P/IPC
+ip-26-0-161-178:2815966:2816396 [3] NCCL INFO Channel 14/0 : 59[3] -> 58[2] via P2P/IPC
+ip-26-0-168-34:272155:272574 [5] NCCL INFO Channel 02/0 : 117[5] -> 116[4] via P2P/IPC
+ip-26-0-168-30:235653:236067 [5] NCCL INFO Channel 14/0 : 109[5] -> 108[4] via P2P/IPC
+ip-26-0-168-34:272156:272570 [6] NCCL INFO Channel 11/0 : 118[6] -> 117[5] via P2P/IPC
+ip-26-0-161-178:2815968:2816394 [5] NCCL INFO Channel 10/0 : 61[5] -> 60[4] via P2P/IPC
+ip-26-0-161-138:96569:97006 [6] NCCL INFO Channel 12/0 : 38[6] -> 37[5] via P2P/IPC
+ip-26-0-168-34:272153:272573 [3] NCCL INFO Channel 11/0 : 115[3] -> 114[2] via P2P/IPC
+ip-26-0-161-103:96891:97311 [3] NCCL INFO Channel 11/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-161-103:96893:97310 [5] NCCL INFO Channel 06/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-168-34:272155:272574 [5] NCCL INFO Channel 03/0 : 117[5] -> 116[4] via P2P/IPC
+ip-26-0-161-103:96894:97312 [6] NCCL INFO Channel 09/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-161-178:2815966:2816396 [3] NCCL INFO Channel 15/0 : 59[3] -> 58[2] via P2P/IPC
+ip-26-0-161-178:2815968:2816394 [5] NCCL INFO Channel 11/0 : 61[5] -> 60[4] via P2P/IPC
+ip-26-0-168-34:272156:272570 [6] NCCL INFO Channel 12/0 : 118[6] -> 117[5] via P2P/IPC
+ip-26-0-168-34:272153:272573 [3] NCCL INFO Channel 12/0 : 115[3] -> 114[2] via P2P/IPC
+ip-26-0-161-103:96891:97311 [3] NCCL INFO Channel 12/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-161-103:96893:97310 [5] NCCL INFO Channel 07/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-161-103:96894:97312 [6] NCCL INFO Channel 10/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-168-34:272155:272574 [5] NCCL INFO Channel 06/0 : 117[5] -> 116[4] via P2P/IPC
+ip-26-0-160-242:1364907:1365329 [5] NCCL INFO Channel 15/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-168-34:272156:272570 [6] NCCL INFO Channel 15/0 : 118[6] -> 117[5] via P2P/IPC
+ip-26-0-161-78:114517:114935 [5] NCCL INFO Channel 15/0 : 69[5] -> 68[4] via P2P/IPC
+ip-26-0-168-34:272153:272573 [3] NCCL INFO Channel 13/0 : 115[3] -> 114[2] via P2P/IPC
+ip-26-0-161-103:96891:97311 [3] NCCL INFO Channel 13/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-168-34:272155:272574 [5] NCCL INFO Channel 07/0 : 117[5] -> 116[4] via P2P/IPC
+ip-26-0-161-103:96893:97310 [5] NCCL INFO Channel 09/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-161-103:96894:97312 [6] NCCL INFO Channel 11/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-165-164:1227468:1227883 [5] NCCL INFO Channel 11/0 : 85[5] -> 84[4] via P2P/IPC
+ip-26-0-161-103:96891:97311 [3] NCCL INFO Channel 14/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-168-34:272153:272573 [3] NCCL INFO Channel 14/0 : 115[3] -> 114[2] via P2P/IPC
+ip-26-0-168-34:272155:272574 [5] NCCL INFO Channel 09/0 : 117[5] -> 116[4] via P2P/IPC
+ip-26-0-161-103:96893:97310 [5] NCCL INFO Channel 10/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-161-153:2636090:2636505 [6] NCCL INFO Channel 11/0 : 54[6] -> 53[5] via P2P/IPC
+ip-26-0-161-153:2636089:2636508 [5] NCCL INFO Channel 07/0 : 53[5] -> 52[4] via P2P/IPC
+ip-26-0-161-103:96894:97312 [6] NCCL INFO Channel 12/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-161-103:96891:97311 [3] NCCL INFO Channel 15/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-161-153:2636087:2636506 [3] NCCL INFO Channel 11/0 : 51[3] -> 50[2] via P2P/IPC
+ip-26-0-161-103:96893:97310 [5] NCCL INFO Channel 11/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-161-178:2815968:2816394 [5] NCCL INFO Channel 14/0 : 61[5] -> 60[4] via P2P/IPC
+ip-26-0-161-153:2636089:2636508 [5] NCCL INFO Channel 09/0 : 53[5] -> 52[4] via P2P/IPC
+ip-26-0-161-103:96894:97312 [6] NCCL INFO Channel 15/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-161-153:2636090:2636505 [6] NCCL INFO Channel 12/0 : 54[6] -> 53[5] via P2P/IPC
+ip-26-0-161-153:2636087:2636506 [3] NCCL INFO Channel 12/0 : 51[3] -> 50[2] via P2P/IPC
+ip-26-0-161-103:96893:97310 [5] NCCL INFO Channel 14/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-161-153:2636089:2636508 [5] NCCL INFO Channel 10/0 : 53[5] -> 52[4] via P2P/IPC
+ip-26-0-161-153:2636090:2636505 [6] NCCL INFO Channel 15/0 : 54[6] -> 53[5] via P2P/IPC
+ip-26-0-161-153:2636087:2636506 [3] NCCL INFO Channel 13/0 : 51[3] -> 50[2] via P2P/IPC
+ip-26-0-168-34:272153:272573 [3] NCCL INFO Channel 15/0 : 115[3] -> 114[2] via P2P/IPC
+ip-26-0-168-34:272155:272574 [5] NCCL INFO Channel 10/0 : 117[5] -> 116[4] via P2P/IPC
+ip-26-0-168-52:2722050:2722471 [5] NCCL INFO Channel 10/0 : 125[5] -> 124[4] via P2P/IPC
+ip-26-0-165-131:1630210:1630653 [5] NCCL INFO Channel 10/0 : 77[5] -> 76[4] via P2P/IPC
+ip-26-0-161-138:96568:97007 [5] NCCL INFO Channel 10/0 : 37[5] -> 36[4] via P2P/IPC
+ip-26-0-161-153:2636087:2636506 [3] NCCL INFO Channel 14/0 : 51[3] -> 50[2] via P2P/IPC
+ip-26-0-161-153:2636089:2636508 [5] NCCL INFO Channel 11/0 : 53[5] -> 52[4] via P2P/IPC
+ip-26-0-161-138:96569:97006 [6] NCCL INFO Channel 15/0 : 38[6] -> 37[5] via P2P/IPC
+ip-26-0-161-178:2815968:2816394 [5] NCCL INFO Channel 15/0 : 61[5] -> 60[4] via P2P/IPC
+ip-26-0-161-153:2636087:2636506 [3] NCCL INFO Channel 15/0 : 51[3] -> 50[2] via P2P/IPC
+ip-26-0-161-153:2636089:2636508 [5] NCCL INFO Channel 14/0 : 53[5] -> 52[4] via P2P/IPC
+ip-26-0-168-34:272155:272574 [5] NCCL INFO Channel 11/0 : 117[5] -> 116[4] via P2P/IPC
+ip-26-0-168-30:235653:236067 [5] NCCL INFO Channel 15/0 : 109[5] -> 108[4] via P2P/IPC
+ip-26-0-165-131:1630210:1630653 [5] NCCL INFO Channel 11/0 : 77[5] -> 76[4] via P2P/IPC
+ip-26-0-168-52:2722050:2722471 [5] NCCL INFO Channel 11/0 : 125[5] -> 124[4] via P2P/IPC
+ip-26-0-161-153:2636089:2636508 [5] NCCL INFO Channel 15/0 : 53[5] -> 52[4] via P2P/IPC
+ip-26-0-161-103:96893:97310 [5] NCCL INFO Channel 15/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-161-138:96568:97007 [5] NCCL INFO Channel 11/0 : 37[5] -> 36[4] via P2P/IPC
+ip-26-0-165-164:1227468:1227883 [5] NCCL INFO Channel 14/0 : 85[5] -> 84[4] via P2P/IPC
+ip-26-0-168-52:2722050:2722471 [5] NCCL INFO Channel 14/0 : 125[5] -> 124[4] via P2P/IPC
+ip-26-0-168-34:272155:272574 [5] NCCL INFO Channel 14/0 : 117[5] -> 116[4] via P2P/IPC
+ip-26-0-161-138:96568:97007 [5] NCCL INFO Channel 14/0 : 37[5] -> 36[4] via P2P/IPC
+ip-26-0-165-131:1630210:1630653 [5] NCCL INFO Channel 14/0 : 77[5] -> 76[4] via P2P/IPC
+ip-26-0-165-131:1630210:1630653 [5] NCCL INFO Channel 15/0 : 77[5] -> 76[4] via P2P/IPC
+ip-26-0-168-34:272155:272574 [5] NCCL INFO Channel 15/0 : 117[5] -> 116[4] via P2P/IPC
+ip-26-0-168-52:2722050:2722471 [5] NCCL INFO Channel 15/0 : 125[5] -> 124[4] via P2P/IPC
+ip-26-0-161-138:96568:97007 [5] NCCL INFO Channel 15/0 : 37[5] -> 36[4] via P2P/IPC
+ip-26-0-165-164:1227468:1227883 [5] NCCL INFO Channel 15/0 : 85[5] -> 84[4] via P2P/IPC
+ip-26-0-168-30:235648:236064 [0] NCCL INFO Connected all rings
+ip-26-0-168-30:235648:236064 [0] NCCL INFO Channel 02/0 : 104[0] -> 105[1] via P2P/IPC
+ip-26-0-161-142:62882:63300 [1] NCCL INFO Connected all rings
+ip-26-0-161-123:32705:33128 [1] NCCL INFO Connected all rings
+ip-26-0-161-138:96563:97003 [0] NCCL INFO Connected all rings
+ip-26-0-161-138:96563:97003 [0] NCCL INFO Channel 02/0 : 32[0] -> 33[1] via P2P/IPC
+ip-26-0-160-225:124161:124582 [1] NCCL INFO Connected all rings
+ip-26-0-160-242:1364902:1365333 [0] NCCL INFO Connected all rings
+ip-26-0-160-242:1364902:1365333 [0] NCCL INFO Channel 02/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-161-138:96563:97003 [0] NCCL INFO Channel 03/0 : 32[0] -> 33[1] via P2P/IPC
+ip-26-0-161-138:96563:97003 [0] NCCL INFO Channel 04/0 : 32[0] -> 33[1] via P2P/IPC
+ip-26-0-161-138:96563:97003 [0] NCCL INFO Channel 05/0 : 32[0] -> 33[1] via P2P/IPC
+ip-26-0-161-138:96563:97003 [0] NCCL INFO Channel 06/0 : 32[0] -> 33[1] via P2P/IPC
+ip-26-0-161-138:96563:97003 [0] NCCL INFO Channel 07/0 : 32[0] -> 33[1] via P2P/IPC
+ip-26-0-165-131:1630205:1630646 [0] NCCL INFO Connected all rings
+ip-26-0-165-131:1630205:1630646 [0] NCCL INFO Channel 02/0 : 72[0] -> 73[1] via P2P/IPC
+ip-26-0-168-30:235650:236065 [2] NCCL INFO Connected all rings
+ip-26-0-161-138:96563:97003 [0] NCCL INFO Channel 10/0 : 32[0] -> 33[1] via P2P/IPC
+ip-26-0-161-138:96565:97004 [2] NCCL INFO Connected all rings
+ip-26-0-165-202:153562:153983 [0] NCCL INFO Connected all rings
+ip-26-0-165-202:153562:153983 [0] NCCL INFO Channel 02/0 : 88[0] -> 89[1] via P2P/IPC
+ip-26-0-165-202:153563:153986 [1] NCCL INFO Connected all rings
+ip-26-0-161-142:62882:63300 [1] NCCL INFO Channel 01/0 : 41[1] -> 42[2] via P2P/IPC
+ip-26-0-161-142:62882:63300 [1] NCCL INFO Channel 03/0 : 41[1] -> 42[2] via P2P/IPC
+ip-26-0-165-164:1227466:1227887 [3] NCCL INFO Connected all rings
+ip-26-0-161-142:62882:63300 [1] NCCL INFO Channel 04/0 : 41[1] -> 42[2] via P2P/IPC
+ip-26-0-160-242:1364907:1365329 [5] NCCL INFO Connected all rings
+ip-26-0-161-142:62882:63300 [1] NCCL INFO Channel 05/0 : 41[1] -> 42[2] via P2P/IPC
+ip-26-0-161-103:96889:97309 [1] NCCL INFO Connected all rings
+ip-26-0-161-142:62882:63300 [1] NCCL INFO Channel 06/0 : 41[1] -> 42[2] via P2P/IPC
+ip-26-0-161-142:62882:63300 [1] NCCL INFO Channel 07/0 : 41[1] -> 42[2] via P2P/IPC
+ip-26-0-160-242:1364905:1365328 [3] NCCL INFO Connected all rings
+ip-26-0-161-142:62882:63300 [1] NCCL INFO Channel 09/0 : 41[1] -> 42[2] via P2P/IPC
+ip-26-0-161-142:62882:63300 [1] NCCL INFO Channel 11/0 : 41[1] -> 42[2] via P2P/IPC
+ip-26-0-161-142:62882:63300 [1] NCCL INFO Channel 12/0 : 41[1] -> 42[2] via P2P/IPC
+ip-26-0-161-138:96569:97006 [6] NCCL INFO Connected all rings
+ip-26-0-161-142:62882:63300 [1] NCCL INFO Channel 13/0 : 41[1] -> 42[2] via P2P/IPC
+ip-26-0-161-123:32708:33126 [4] NCCL INFO Connected all rings
+ip-26-0-161-142:62882:63300 [1] NCCL INFO Channel 14/0 : 41[1] -> 42[2] via P2P/IPC
+ip-26-0-161-142:62882:63300 [1] NCCL INFO Channel 15/0 : 41[1] -> 42[2] via P2P/IPC
+ip-26-0-161-142:62886:63303 [5] NCCL INFO Connected all rings
+ip-26-0-160-225:124166:124583 [6] NCCL INFO Connected all rings
+ip-26-0-165-59:1557273:1557685 [7] NCCL INFO Connected all rings
+ip-26-0-168-52:2722047:2722468 [2] NCCL INFO Connected all rings
+ip-26-0-168-34:272151:272571 [1] NCCL INFO Connected all rings
+ip-26-0-161-153:2636084:2636511 [0] NCCL INFO Connected all rings
+ip-26-0-161-153:2636084:2636511 [0] NCCL INFO Channel 02/0 : 48[0] -> 49[1] via P2P/IPC
+ip-26-0-165-131:1630208:1630650 [3] NCCL INFO Connected all rings
+ip-26-0-165-59:1557267:1557683 [1] NCCL INFO Connected all rings
+ip-26-0-161-78:114514:114932 [2] NCCL INFO Connected all rings
+ip-26-0-168-30:235649:236071 [1] NCCL INFO Connected all rings
+ip-26-0-161-78:114512:114930 [0] NCCL INFO Connected all rings
+ip-26-0-161-78:114512:114930 [0] NCCL INFO Channel 02/0 : 64[0] -> 65[1] via P2P/IPC
+ip-26-0-168-30:235648:236064 [0] NCCL INFO Channel 03/0 : 104[0] -> 105[1] via P2P/IPC
+ip-26-0-161-78:114516:114934 [4] NCCL INFO Connected all rings
+ip-26-0-165-202:153562:153983 [0] NCCL INFO Channel 03/0 : 88[0] -> 89[1] via P2P/IPC
+ip-26-0-160-225:124160:124581 [0] NCCL INFO Connected all rings
+ip-26-0-160-225:124160:124581 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-78:114512:114930 [0] NCCL INFO Channel 03/0 : 64[0] -> 65[1] via P2P/IPC
+ip-26-0-168-30:235648:236064 [0] NCCL INFO Channel 04/0 : 104[0] -> 105[1] via P2P/IPC
+ip-26-0-161-78:114512:114930 [0] NCCL INFO Channel 04/0 : 64[0] -> 65[1] via P2P/IPC
+ip-26-0-160-225:124160:124581 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-202:153562:153983 [0] NCCL INFO Channel 04/0 : 88[0] -> 89[1] via P2P/IPC
+ip-26-0-161-78:114512:114930 [0] NCCL INFO Channel 05/0 : 64[0] -> 65[1] via P2P/IPC
+ip-26-0-168-30:235648:236064 [0] NCCL INFO Channel 05/0 : 104[0] -> 105[1] via P2P/IPC
+ip-26-0-160-225:124160:124581 [0] NCCL INFO Channel 04/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-123:32704:33123 [0] NCCL INFO Connected all rings
+ip-26-0-161-123:32704:33123 [0] NCCL INFO Channel 02/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-161-153:2636086:2636507 [2] NCCL INFO Connected all rings
+ip-26-0-160-225:124160:124581 [0] NCCL INFO Channel 05/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-78:114512:114930 [0] NCCL INFO Channel 06/0 : 64[0] -> 65[1] via P2P/IPC
+ip-26-0-161-123:32704:33123 [0] NCCL INFO Channel 03/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-165-202:153562:153983 [0] NCCL INFO Channel 05/0 : 88[0] -> 89[1] via P2P/IPC
+ip-26-0-161-123:32704:33123 [0] NCCL INFO Channel 04/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-161-78:114512:114930 [0] NCCL INFO Channel 07/0 : 64[0] -> 65[1] via P2P/IPC
+ip-26-0-160-225:124160:124581 [0] NCCL INFO Channel 06/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-123:32704:33123 [0] NCCL INFO Channel 05/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-168-30:235648:236064 [0] NCCL INFO Channel 06/0 : 104[0] -> 105[1] via P2P/IPC
+ip-26-0-165-131:1630206:1630647 [1] NCCL INFO Connected all rings
+ip-26-0-161-123:32704:33123 [0] NCCL INFO Channel 06/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-161-78:114512:114930 [0] NCCL INFO Channel 10/0 : 64[0] -> 65[1] via P2P/IPC
+ip-26-0-165-59:1557268:1557686 [2] NCCL INFO Connected all rings
+ip-26-0-161-123:32704:33123 [0] NCCL INFO Channel 07/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-161-123:32704:33123 [0] NCCL INFO Channel 10/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-168-30:235648:236064 [0] NCCL INFO Channel 07/0 : 104[0] -> 105[1] via P2P/IPC
+ip-26-0-168-30:235648:236064 [0] NCCL INFO Channel 10/0 : 104[0] -> 105[1] via P2P/IPC
+ip-26-0-160-225:124160:124581 [0] NCCL INFO Channel 07/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-123:32704:33123 [0] NCCL INFO Channel 11/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-161-153:2636084:2636511 [0] NCCL INFO Channel 03/0 : 48[0] -> 49[1] via P2P/IPC
+ip-26-0-168-30:235648:236064 [0] NCCL INFO Channel 11/0 : 104[0] -> 105[1] via P2P/IPC
+ip-26-0-161-78:114512:114930 [0] NCCL INFO Channel 11/0 : 64[0] -> 65[1] via P2P/IPC
+ip-26-0-160-225:124160:124581 [0] NCCL INFO Channel 10/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-168-52:2722045:2722470 [0] NCCL INFO Connected all rings
+ip-26-0-168-52:2722045:2722470 [0] NCCL INFO Channel 02/0 : 120[0] -> 121[1] via P2P/IPC
+ip-26-0-161-123:32704:33123 [0] NCCL INFO Channel 12/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-165-59:1557267:1557683 [1] NCCL INFO Channel 01/0 : 97[1] -> 98[2] via P2P/IPC
+ip-26-0-160-225:124160:124581 [0] NCCL INFO Channel 11/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-168-30:235648:236064 [0] NCCL INFO Channel 12/0 : 104[0] -> 105[1] via P2P/IPC
+ip-26-0-168-52:2722045:2722470 [0] NCCL INFO Channel 03/0 : 120[0] -> 121[1] via P2P/IPC
+ip-26-0-161-153:2636084:2636511 [0] NCCL INFO Channel 04/0 : 48[0] -> 49[1] via P2P/IPC
+ip-26-0-161-123:32704:33123 [0] NCCL INFO Channel 13/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-165-59:1557267:1557683 [1] NCCL INFO Channel 03/0 : 97[1] -> 98[2] via P2P/IPC
+ip-26-0-168-30:235648:236064 [0] NCCL INFO Channel 13/0 : 104[0] -> 105[1] via P2P/IPC
+ip-26-0-161-153:2636084:2636511 [0] NCCL INFO Channel 05/0 : 48[0] -> 49[1] via P2P/IPC
+ip-26-0-168-52:2722045:2722470 [0] NCCL INFO Channel 04/0 : 120[0] -> 121[1] via P2P/IPC
+ip-26-0-161-123:32706:33122 [2] NCCL INFO Connected all rings
+ip-26-0-161-123:32704:33123 [0] NCCL INFO Channel 14/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-165-59:1557267:1557683 [1] NCCL INFO Channel 04/0 : 97[1] -> 98[2] via P2P/IPC
+ip-26-0-161-78:114512:114930 [0] NCCL INFO Channel 12/0 : 64[0] -> 65[1] via P2P/IPC
+ip-26-0-161-153:2636084:2636511 [0] NCCL INFO Channel 06/0 : 48[0] -> 49[1] via P2P/IPC
+ip-26-0-161-123:32704:33123 [0] NCCL INFO Channel 15/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-161-178:2815964:2816401 [1] NCCL INFO Connected all rings
+ip-26-0-168-30:235648:236064 [0] NCCL INFO Channel 14/0 : 104[0] -> 105[1] via P2P/IPC
+ip-26-0-165-59:1557267:1557683 [1] NCCL INFO Channel 05/0 : 97[1] -> 98[2] via P2P/IPC
+ip-26-0-165-164:1227465:1227881 [2] NCCL INFO Connected all rings
+ip-26-0-161-153:2636084:2636511 [0] NCCL INFO Channel 07/0 : 48[0] -> 49[1] via P2P/IPC
+ip-26-0-168-30:235648:236064 [0] NCCL INFO Channel 15/0 : 104[0] -> 105[1] via P2P/IPC
+ip-26-0-160-225:124160:124581 [0] NCCL INFO Channel 12/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-168-52:2722045:2722470 [0] NCCL INFO Channel 05/0 : 120[0] -> 121[1] via P2P/IPC
+ip-26-0-161-78:114512:114930 [0] NCCL INFO Channel 13/0 : 64[0] -> 65[1] via P2P/IPC
+ip-26-0-168-30:235649:236071 [1] NCCL INFO Channel 01/0 : 105[1] -> 106[2] via P2P/IPC
+ip-26-0-165-164:1227463:1227880 [0] NCCL INFO Connected all rings
+ip-26-0-161-153:2636084:2636511 [0] NCCL INFO Channel 10/0 : 48[0] -> 49[1] via P2P/IPC
+ip-26-0-165-164:1227463:1227880 [0] NCCL INFO Channel 02/0 : 80[0] -> 81[1] via P2P/IPC
+ip-26-0-165-59:1557267:1557683 [1] NCCL INFO Channel 06/0 : 97[1] -> 98[2] via P2P/IPC
+ip-26-0-161-78:114512:114930 [0] NCCL INFO Channel 14/0 : 64[0] -> 65[1] via P2P/IPC
+ip-26-0-160-225:124160:124581 [0] NCCL INFO Channel 13/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-153:2636084:2636511 [0] NCCL INFO Channel 11/0 : 48[0] -> 49[1] via P2P/IPC
+ip-26-0-165-59:1557267:1557683 [1] NCCL INFO Channel 07/0 : 97[1] -> 98[2] via P2P/IPC
+ip-26-0-160-242:1364902:1365333 [0] NCCL INFO Channel 03/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-168-30:235649:236071 [1] NCCL INFO Channel 03/0 : 105[1] -> 106[2] via P2P/IPC
+ip-26-0-168-30:235649:236071 [1] NCCL INFO Channel 04/0 : 105[1] -> 106[2] via P2P/IPC
+ip-26-0-165-164:1227464:1227884 [1] NCCL INFO Connected all rings
+ip-26-0-165-164:1227463:1227880 [0] NCCL INFO Channel 03/0 : 80[0] -> 81[1] via P2P/IPC
+ip-26-0-161-153:2636084:2636511 [0] NCCL INFO Channel 12/0 : 48[0] -> 49[1] via P2P/IPC
+ip-26-0-160-225:124160:124581 [0] NCCL INFO Channel 14/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-59:1557267:1557683 [1] NCCL INFO Channel 09/0 : 97[1] -> 98[2] via P2P/IPC
+ip-26-0-161-78:114512:114930 [0] NCCL INFO Channel 15/0 : 64[0] -> 65[1] via P2P/IPC
+ip-26-0-168-52:2722045:2722470 [0] NCCL INFO Channel 06/0 : 120[0] -> 121[1] via P2P/IPC
+ip-26-0-165-59:1557267:1557683 [1] NCCL INFO Channel 11/0 : 97[1] -> 98[2] via P2P/IPC
+ip-26-0-165-59:1557272:1557687 [6] NCCL INFO Connected all rings
+ip-26-0-165-164:1227463:1227880 [0] NCCL INFO Channel 04/0 : 80[0] -> 81[1] via P2P/IPC
+ip-26-0-168-30:235649:236071 [1] NCCL INFO Channel 05/0 : 105[1] -> 106[2] via P2P/IPC
+ip-26-0-161-153:2636084:2636511 [0] NCCL INFO Channel 13/0 : 48[0] -> 49[1] via P2P/IPC
+ip-26-0-160-225:124160:124581 [0] NCCL INFO Channel 15/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-168-30:235649:236071 [1] NCCL INFO Channel 06/0 : 105[1] -> 106[2] via P2P/IPC
+ip-26-0-160-242:1364902:1365333 [0] NCCL INFO Channel 04/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-161-153:2636084:2636511 [0] NCCL INFO Channel 14/0 : 48[0] -> 49[1] via P2P/IPC
+ip-26-0-168-52:2722045:2722470 [0] NCCL INFO Channel 07/0 : 120[0] -> 121[1] via P2P/IPC
+ip-26-0-165-59:1557267:1557683 [1] NCCL INFO Channel 12/0 : 97[1] -> 98[2] via P2P/IPC
+ip-26-0-165-164:1227463:1227880 [0] NCCL INFO Channel 05/0 : 80[0] -> 81[1] via P2P/IPC
+ip-26-0-161-103:96888:97306 [0] NCCL INFO Connected all rings
+ip-26-0-160-242:1364902:1365333 [0] NCCL INFO Channel 05/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-161-103:96888:97306 [0] NCCL INFO Channel 02/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-168-30:235649:236071 [1] NCCL INFO Channel 07/0 : 105[1] -> 106[2] via P2P/IPC
+ip-26-0-161-138:96564:97009 [1] NCCL INFO Connected all rings
+ip-26-0-161-153:2636084:2636511 [0] NCCL INFO Channel 15/0 : 48[0] -> 49[1] via P2P/IPC
+ip-26-0-161-123:32705:33128 [1] NCCL INFO Channel 01/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-168-30:235649:236071 [1] NCCL INFO Channel 09/0 : 105[1] -> 106[2] via P2P/IPC
+ip-26-0-165-164:1227463:1227880 [0] NCCL INFO Channel 06/0 : 80[0] -> 81[1] via P2P/IPC
+ip-26-0-168-52:2722045:2722470 [0] NCCL INFO Channel 10/0 : 120[0] -> 121[1] via P2P/IPC
+ip-26-0-161-103:96888:97306 [0] NCCL INFO Channel 03/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-168-30:235649:236071 [1] NCCL INFO Channel 11/0 : 105[1] -> 106[2] via P2P/IPC
+ip-26-0-165-59:1557271:1557690 [5] NCCL INFO Connected all rings
+ip-26-0-165-59:1557267:1557683 [1] NCCL INFO Channel 13/0 : 97[1] -> 98[2] via P2P/IPC
+ip-26-0-160-242:1364902:1365333 [0] NCCL INFO Channel 06/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-165-164:1227463:1227880 [0] NCCL INFO Channel 07/0 : 80[0] -> 81[1] via P2P/IPC
+ip-26-0-160-242:1364902:1365333 [0] NCCL INFO Channel 07/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-165-59:1557268:1557686 [2] NCCL INFO Channel 01/0 : 98[2] -> 99[3] via P2P/IPC
+ip-26-0-168-52:2722045:2722470 [0] NCCL INFO Channel 11/0 : 120[0] -> 121[1] via P2P/IPC
+ip-26-0-161-103:96888:97306 [0] NCCL INFO Channel 04/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-161-123:32709:33129 [5] NCCL INFO Connected all rings
+ip-26-0-165-59:1557267:1557683 [1] NCCL INFO Channel 14/0 : 97[1] -> 98[2] via P2P/IPC
+ip-26-0-168-30:235649:236071 [1] NCCL INFO Channel 12/0 : 105[1] -> 106[2] via P2P/IPC
+ip-26-0-161-103:96888:97306 [0] NCCL INFO Channel 05/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-168-52:2722045:2722470 [0] NCCL INFO Channel 12/0 : 120[0] -> 121[1] via P2P/IPC
+ip-26-0-165-59:1557268:1557686 [2] NCCL INFO Channel 02/0 : 98[2] -> 99[3] via P2P/IPC
+ip-26-0-168-34:272150:272576 [0] NCCL INFO Connected all rings
+ip-26-0-161-123:32705:33128 [1] NCCL INFO Channel 03/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-160-242:1364902:1365333 [0] NCCL INFO Channel 10/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-168-34:272150:272576 [0] NCCL INFO Channel 02/0 : 112[0] -> 113[1] via P2P/IPC
+ip-26-0-165-59:1557267:1557683 [1] NCCL INFO Channel 15/0 : 97[1] -> 98[2] via P2P/IPC
+ip-26-0-165-59:1557269:1557684 [3] NCCL INFO Connected all rings
+ip-26-0-165-59:1557268:1557686 [2] NCCL INFO Channel 04/0 : 98[2] -> 99[3] via P2P/IPC
+ip-26-0-168-30:235653:236067 [5] NCCL INFO Connected all rings
+ip-26-0-165-164:1227463:1227880 [0] NCCL INFO Channel 10/0 : 80[0] -> 81[1] via P2P/IPC
+ip-26-0-161-103:96888:97306 [0] NCCL INFO Channel 06/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-161-123:32705:33128 [1] NCCL INFO Channel 04/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-160-242:1364902:1365333 [0] NCCL INFO Channel 11/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-165-131:1630205:1630646 [0] NCCL INFO Channel 03/0 : 72[0] -> 73[1] via P2P/IPC
+ip-26-0-168-52:2722045:2722470 [0] NCCL INFO Channel 13/0 : 120[0] -> 121[1] via P2P/IPC
+ip-26-0-168-34:272150:272576 [0] NCCL INFO Channel 03/0 : 112[0] -> 113[1] via P2P/IPC
+ip-26-0-161-123:32705:33128 [1] NCCL INFO Channel 05/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-161-103:96888:97306 [0] NCCL INFO Channel 07/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-165-59:1557268:1557686 [2] NCCL INFO Channel 05/0 : 98[2] -> 99[3] via P2P/IPC
+ip-26-0-161-103:96894:97312 [6] NCCL INFO Connected all rings
+ip-26-0-165-164:1227463:1227880 [0] NCCL INFO Channel 11/0 : 80[0] -> 81[1] via P2P/IPC
+ip-26-0-160-242:1364902:1365333 [0] NCCL INFO Channel 12/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-161-123:32706:33122 [2] NCCL INFO Channel 01/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-161-123:32705:33128 [1] NCCL INFO Channel 06/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-160-242:1364902:1365333 [0] NCCL INFO Channel 13/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-168-52:2722045:2722470 [0] NCCL INFO Channel 14/0 : 120[0] -> 121[1] via P2P/IPC
+ip-26-0-168-34:272150:272576 [0] NCCL INFO Channel 04/0 : 112[0] -> 113[1] via P2P/IPC
+ip-26-0-165-59:1557268:1557686 [2] NCCL INFO Channel 06/0 : 98[2] -> 99[3] via P2P/IPC
+ip-26-0-165-202:153567:153979 [5] NCCL INFO Connected all rings
+ip-26-0-168-34:272150:272576 [0] NCCL INFO Channel 05/0 : 112[0] -> 113[1] via P2P/IPC
+ip-26-0-168-30:235649:236071 [1] NCCL INFO Channel 13/0 : 105[1] -> 106[2] via P2P/IPC
+ip-26-0-160-225:124161:124582 [1] NCCL INFO Channel 01/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-164:1227470:1227885 [7] NCCL INFO Connected all rings
+ip-26-0-168-52:2722045:2722470 [0] NCCL INFO Channel 15/0 : 120[0] -> 121[1] via P2P/IPC
+ip-26-0-161-123:32706:33122 [2] NCCL INFO Channel 02/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-165-59:1557268:1557686 [2] NCCL INFO Channel 07/0 : 98[2] -> 99[3] via P2P/IPC
+ip-26-0-161-123:32705:33128 [1] NCCL INFO Channel 07/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-168-34:272150:272576 [0] NCCL INFO Channel 06/0 : 112[0] -> 113[1] via P2P/IPC
+ip-26-0-161-142:62888:63302 [7] NCCL INFO Connected all rings
+ip-26-0-161-123:32706:33122 [2] NCCL INFO Channel 04/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-161-138:96563:97003 [0] NCCL INFO Channel 11/0 : 32[0] -> 33[1] via P2P/IPC
+ip-26-0-161-103:96890:97313 [2] NCCL INFO Connected all rings
+ip-26-0-165-59:1557268:1557686 [2] NCCL INFO Channel 09/0 : 98[2] -> 99[3] via P2P/IPC
+ip-26-0-161-103:96888:97306 [0] NCCL INFO Channel 10/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-165-131:1630205:1630646 [0] NCCL INFO Channel 04/0 : 72[0] -> 73[1] via P2P/IPC
+ip-26-0-168-30:235651:236068 [3] NCCL INFO Connected all rings
+ip-26-0-160-225:124161:124582 [1] NCCL INFO Channel 03/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-202:153564:153984 [2] NCCL INFO Connected all rings
+ip-26-0-168-34:272150:272576 [0] NCCL INFO Channel 07/0 : 112[0] -> 113[1] via P2P/IPC
+ip-26-0-165-131:1630205:1630646 [0] NCCL INFO Channel 05/0 : 72[0] -> 73[1] via P2P/IPC
+ip-26-0-160-225:124161:124582 [1] NCCL INFO Channel 04/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-103:96888:97306 [0] NCCL INFO Channel 11/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-165-131:1630205:1630646 [0] NCCL INFO Channel 06/0 : 72[0] -> 73[1] via P2P/IPC
+ip-26-0-165-164:1227467:1227886 [4] NCCL INFO Connected all rings
+ip-26-0-165-59:1557268:1557686 [2] NCCL INFO Channel 10/0 : 98[2] -> 99[3] via P2P/IPC
+ip-26-0-165-164:1227463:1227880 [0] NCCL INFO Channel 12/0 : 80[0] -> 81[1] via P2P/IPC
+ip-26-0-161-123:32705:33128 [1] NCCL INFO Channel 09/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-161-123:32706:33122 [2] NCCL INFO Channel 05/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-168-30:235655:236069 [7] NCCL INFO Connected all rings
+ip-26-0-161-142:62881:63298 [0] NCCL INFO Connected all rings
+ip-26-0-161-142:62881:63298 [0] NCCL INFO Channel 02/0 : 40[0] -> 41[1] via P2P/IPC
+ip-26-0-165-202:153562:153983 [0] NCCL INFO Channel 06/0 : 88[0] -> 89[1] via P2P/IPC
+ip-26-0-168-30:235649:236071 [1] NCCL INFO Channel 14/0 : 105[1] -> 106[2] via P2P/IPC
+ip-26-0-161-123:32705:33128 [1] NCCL INFO Channel 11/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-161-123:32706:33122 [2] NCCL INFO Channel 06/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-165-164:1227463:1227880 [0] NCCL INFO Channel 13/0 : 80[0] -> 81[1] via P2P/IPC
+ip-26-0-161-103:96888:97306 [0] NCCL INFO Channel 12/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-165-131:1630205:1630646 [0] NCCL INFO Channel 07/0 : 72[0] -> 73[1] via P2P/IPC
+ip-26-0-168-34:272150:272576 [0] NCCL INFO Channel 10/0 : 112[0] -> 113[1] via P2P/IPC
+ip-26-0-160-225:124161:124582 [1] NCCL INFO Channel 05/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-138:96563:97003 [0] NCCL INFO Channel 12/0 : 32[0] -> 33[1] via P2P/IPC
+ip-26-0-165-59:1557270:1557688 [4] NCCL INFO Connected all rings
+ip-26-0-161-123:32705:33128 [1] NCCL INFO Channel 12/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-160-242:1364902:1365333 [0] NCCL INFO Channel 14/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-160-242:1364908:1365332 [6] NCCL INFO Connected all rings
+ip-26-0-160-225:124161:124582 [1] NCCL INFO Channel 06/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-131:1630205:1630646 [0] NCCL INFO Channel 10/0 : 72[0] -> 73[1] via P2P/IPC
+ip-26-0-161-142:62885:63304 [4] NCCL INFO Connected all rings
+ip-26-0-165-164:1227463:1227880 [0] NCCL INFO Channel 14/0 : 80[0] -> 81[1] via P2P/IPC
+ip-26-0-161-138:96563:97003 [0] NCCL INFO Channel 13/0 : 32[0] -> 33[1] via P2P/IPC
+ip-26-0-160-242:1364902:1365333 [0] NCCL INFO Channel 15/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-165-59:1557268:1557686 [2] NCCL INFO Channel 12/0 : 98[2] -> 99[3] via P2P/IPC
+ip-26-0-168-30:235649:236071 [1] NCCL INFO Channel 15/0 : 105[1] -> 106[2] via P2P/IPC
+ip-26-0-161-142:62887:63305 [6] NCCL INFO Connected all rings
+ip-26-0-161-103:96888:97306 [0] NCCL INFO Channel 13/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-168-30:235654:236070 [6] NCCL INFO Connected all rings
+ip-26-0-161-142:62881:63298 [0] NCCL INFO Channel 03/0 : 40[0] -> 41[1] via P2P/IPC
+ip-26-0-165-131:1630205:1630646 [0] NCCL INFO Channel 11/0 : 72[0] -> 73[1] via P2P/IPC
+ip-26-0-165-164:1227465:1227881 [2] NCCL INFO Channel 01/0 : 82[2] -> 83[3] via P2P/IPC
+ip-26-0-161-78:114519:114931 [7] NCCL INFO Connected all rings
+ip-26-0-165-202:153562:153983 [0] NCCL INFO Channel 07/0 : 88[0] -> 89[1] via P2P/IPC
+ip-26-0-165-59:1557268:1557686 [2] NCCL INFO Channel 13/0 : 98[2] -> 99[3] via P2P/IPC
+ip-26-0-161-138:96563:97003 [0] NCCL INFO Channel 14/0 : 32[0] -> 33[1] via P2P/IPC
+ip-26-0-165-131:1630205:1630646 [0] NCCL INFO Channel 12/0 : 72[0] -> 73[1] via P2P/IPC
+ip-26-0-168-34:272150:272576 [0] NCCL INFO Channel 11/0 : 112[0] -> 113[1] via P2P/IPC
+ip-26-0-168-30:235648:236064 [0] NCCL INFO Channel 05/0 : 104[0] -> 111[7] via P2P/IPC
+ip-26-0-165-202:153562:153983 [0] NCCL INFO Channel 10/0 : 88[0] -> 89[1] via P2P/IPC
+ip-26-0-160-225:124161:124582 [1] NCCL INFO Channel 07/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-59:1557266:1557689 [0] NCCL INFO Connected all rings
+ip-26-0-165-59:1557266:1557689 [0] NCCL INFO Channel 02/0 : 96[0] -> 97[1] via P2P/IPC
+ip-26-0-165-164:1227468:1227883 [5] NCCL INFO Connected all rings
+ip-26-0-165-59:1557268:1557686 [2] NCCL INFO Channel 14/0 : 98[2] -> 99[3] via P2P/IPC
+ip-26-0-165-164:1227463:1227880 [0] NCCL INFO Channel 15/0 : 80[0] -> 81[1] via P2P/IPC
+ip-26-0-165-59:1557272:1557687 [6] NCCL INFO Channel 01/0 : 102[6] -> 103[7] via P2P/IPC
+ip-26-0-161-78:114515:114936 [3] NCCL INFO Connected all rings
+ip-26-0-168-52:2722048:2722469 [3] NCCL INFO Connected all rings
+ip-26-0-168-34:272150:272576 [0] NCCL INFO Channel 12/0 : 112[0] -> 113[1] via P2P/IPC
+ip-26-0-165-131:1630205:1630646 [0] NCCL INFO Channel 13/0 : 72[0] -> 73[1] via P2P/IPC
+ip-26-0-161-142:62881:63298 [0] NCCL INFO Channel 04/0 : 40[0] -> 41[1] via P2P/IPC
+ip-26-0-168-30:235648:236064 [0] NCCL INFO Channel 06/0 : 104[0] -> 111[7] via P2P/IPC
+ip-26-0-165-202:153562:153983 [0] NCCL INFO Channel 11/0 : 88[0] -> 89[1] via P2P/IPC
+ip-26-0-165-164:1227465:1227881 [2] NCCL INFO Channel 02/0 : 82[2] -> 83[3] via P2P/IPC
+ip-26-0-161-138:96566:97010 [3] NCCL INFO Connected all rings
+ip-26-0-168-52:2722051:2722465 [6] NCCL INFO Connected all rings
+ip-26-0-168-34:272150:272576 [0] NCCL INFO Channel 13/0 : 112[0] -> 113[1] via P2P/IPC
+ip-26-0-168-30:235648:236064 [0] NCCL INFO Channel 07/0 : 104[0] -> 111[7] via P2P/IPC
+ip-26-0-165-131:1630205:1630646 [0] NCCL INFO Channel 14/0 : 72[0] -> 73[1] via P2P/IPC
+ip-26-0-161-123:32706:33122 [2] NCCL INFO Channel 07/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-165-59:1557268:1557686 [2] NCCL INFO Channel 15/0 : 98[2] -> 99[3] via P2P/IPC
+ip-26-0-165-59:1557272:1557687 [6] NCCL INFO Channel 02/0 : 102[6] -> 103[7] via P2P/IPC
+ip-26-0-165-59:1557266:1557689 [0] NCCL INFO Channel 03/0 : 96[0] -> 97[1] via P2P/IPC
+ip-26-0-160-242:1364909:1365327 [7] NCCL INFO Connected all rings
+ip-26-0-165-164:1227465:1227881 [2] NCCL INFO Channel 04/0 : 82[2] -> 83[3] via P2P/IPC
+ip-26-0-161-138:96563:97003 [0] NCCL INFO Channel 15/0 : 32[0] -> 33[1] via P2P/IPC
+ip-26-0-165-202:153562:153983 [0] NCCL INFO Channel 12/0 : 88[0] -> 89[1] via P2P/IPC
+ip-26-0-165-59:1557271:1557690 [5] NCCL INFO Channel 01/0 : 101[5] -> 102[6] via P2P/IPC
+ip-26-0-161-103:96888:97306 [0] NCCL INFO Channel 14/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-168-34:272150:272576 [0] NCCL INFO Channel 14/0 : 112[0] -> 113[1] via P2P/IPC
+ip-26-0-161-142:62881:63298 [0] NCCL INFO Channel 05/0 : 40[0] -> 41[1] via P2P/IPC
+ip-26-0-168-30:235648:236064 [0] NCCL INFO Channel 13/0 : 104[0] -> 111[7] via P2P/IPC
+ip-26-0-165-131:1630205:1630646 [0] NCCL INFO Channel 15/0 : 72[0] -> 73[1] via P2P/IPC
+ip-26-0-165-59:1557272:1557687 [6] NCCL INFO Channel 03/0 : 102[6] -> 103[7] via P2P/IPC
+ip-26-0-165-59:1557266:1557689 [0] NCCL INFO Channel 04/0 : 96[0] -> 97[1] via P2P/IPC
+ip-26-0-165-202:153565:153985 [3] NCCL INFO Connected all rings
+ip-26-0-165-202:153562:153983 [0] NCCL INFO Channel 13/0 : 88[0] -> 89[1] via P2P/IPC
+ip-26-0-161-78:114514:114932 [2] NCCL INFO Channel 01/0 : 66[2] -> 67[3] via P2P/IPC
+ip-26-0-160-225:124163:124588 [3] NCCL INFO Connected all rings
+ip-26-0-161-78:114517:114935 [5] NCCL INFO Connected all rings
+ip-26-0-165-59:1557271:1557690 [5] NCCL INFO Channel 02/0 : 101[5] -> 102[6] via P2P/IPC
+ip-26-0-160-242:1364904:1365335 [2] NCCL INFO Connected all rings
+ip-26-0-161-103:96888:97306 [0] NCCL INFO Channel 15/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-160-225:124164:124585 [4] NCCL INFO Connected all rings
+ip-26-0-168-34:272150:272576 [0] NCCL INFO Channel 15/0 : 112[0] -> 113[1] via P2P/IPC
+ip-26-0-161-123:32705:33128 [1] NCCL INFO Channel 13/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-168-34:272155:272574 [5] NCCL INFO Connected all rings
+ip-26-0-161-123:32706:33122 [2] NCCL INFO Channel 09/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-161-142:62881:63298 [0] NCCL INFO Channel 06/0 : 40[0] -> 41[1] via P2P/IPC
+ip-26-0-160-225:124161:124582 [1] NCCL INFO Channel 09/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-78:114514:114932 [2] NCCL INFO Channel 02/0 : 66[2] -> 67[3] via P2P/IPC
+ip-26-0-165-164:1227469:1227882 [6] NCCL INFO Connected all rings
+ip-26-0-165-59:1557272:1557687 [6] NCCL INFO Channel 04/0 : 102[6] -> 103[7] via P2P/IPC
+ip-26-0-165-59:1557266:1557689 [0] NCCL INFO Channel 05/0 : 96[0] -> 97[1] via P2P/IPC
+ip-26-0-161-138:96564:97009 [1] NCCL INFO Channel 01/0 : 33[1] -> 34[2] via P2P/IPC
+ip-26-0-165-202:153562:153983 [0] NCCL INFO Channel 14/0 : 88[0] -> 89[1] via P2P/IPC
+ip-26-0-165-202:153569:153981 [7] NCCL INFO Connected all rings
+ip-26-0-165-164:1227464:1227884 [1] NCCL INFO Channel 01/0 : 81[1] -> 82[2] via P2P/IPC
+ip-26-0-165-164:1227465:1227881 [2] NCCL INFO Channel 05/0 : 82[2] -> 83[3] via P2P/IPC
+ip-26-0-161-142:62881:63298 [0] NCCL INFO Channel 07/0 : 40[0] -> 41[1] via P2P/IPC
+ip-26-0-165-59:1557271:1557690 [5] NCCL INFO Channel 03/0 : 101[5] -> 102[6] via P2P/IPC
+ip-26-0-161-123:32705:33128 [1] NCCL INFO Channel 14/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-161-123:32706:33122 [2] NCCL INFO Channel 10/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-160-225:124161:124582 [1] NCCL INFO Channel 11/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-78:114514:114932 [2] NCCL INFO Channel 04/0 : 66[2] -> 67[3] via P2P/IPC
+ip-26-0-165-59:1557272:1557687 [6] NCCL INFO Channel 05/0 : 102[6] -> 103[7] via P2P/IPC
+ip-26-0-161-142:62881:63298 [0] NCCL INFO Channel 10/0 : 40[0] -> 41[1] via P2P/IPC
+ip-26-0-165-59:1557269:1557684 [3] NCCL INFO Channel 01/0 : 99[3] -> 100[4] via P2P/IPC
+ip-26-0-161-138:96564:97009 [1] NCCL INFO Channel 03/0 : 33[1] -> 34[2] via P2P/IPC
+ip-26-0-165-59:1557266:1557689 [0] NCCL INFO Channel 06/0 : 96[0] -> 97[1] via P2P/IPC
+ip-26-0-165-164:1227465:1227881 [2] NCCL INFO Channel 06/0 : 82[2] -> 83[3] via P2P/IPC
+ip-26-0-165-164:1227464:1227884 [1] NCCL INFO Channel 03/0 : 81[1] -> 82[2] via P2P/IPC
+ip-26-0-165-202:153562:153983 [0] NCCL INFO Channel 15/0 : 88[0] -> 89[1] via P2P/IPC
+ip-26-0-165-59:1557271:1557690 [5] NCCL INFO Channel 04/0 : 101[5] -> 102[6] via P2P/IPC
+ip-26-0-161-123:32705:33128 [1] NCCL INFO Channel 15/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-161-78:114514:114932 [2] NCCL INFO Channel 05/0 : 66[2] -> 67[3] via P2P/IPC
+ip-26-0-161-123:32706:33122 [2] NCCL INFO Channel 12/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-161-142:62881:63298 [0] NCCL INFO Channel 11/0 : 40[0] -> 41[1] via P2P/IPC
+ip-26-0-160-225:124167:124586 [7] NCCL INFO Connected all rings
+ip-26-0-165-59:1557272:1557687 [6] NCCL INFO Channel 06/0 : 102[6] -> 103[7] via P2P/IPC
+ip-26-0-165-59:1557269:1557684 [3] NCCL INFO Channel 02/0 : 99[3] -> 100[4] via P2P/IPC
+ip-26-0-168-52:2722049:2722467 [4] NCCL INFO Connected all rings
+ip-26-0-165-164:1227465:1227881 [2] NCCL INFO Channel 07/0 : 82[2] -> 83[3] via P2P/IPC
+ip-26-0-165-59:1557266:1557689 [0] NCCL INFO Channel 07/0 : 96[0] -> 97[1] via P2P/IPC
+ip-26-0-165-164:1227464:1227884 [1] NCCL INFO Channel 04/0 : 81[1] -> 82[2] via P2P/IPC
+ip-26-0-161-123:32706:33122 [2] NCCL INFO Channel 13/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-161-123:32704:33123 [0] NCCL INFO Channel 05/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-165-59:1557271:1557690 [5] NCCL INFO Channel 05/0 : 101[5] -> 102[6] via P2P/IPC
+ip-26-0-161-78:114514:114932 [2] NCCL INFO Channel 06/0 : 66[2] -> 67[3] via P2P/IPC
+ip-26-0-160-225:124161:124582 [1] NCCL INFO Channel 12/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-103:96889:97309 [1] NCCL INFO Channel 01/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-165-164:1227465:1227881 [2] NCCL INFO Channel 09/0 : 82[2] -> 83[3] via P2P/IPC
+ip-26-0-165-59:1557269:1557684 [3] NCCL INFO Channel 03/0 : 99[3] -> 100[4] via P2P/IPC
+ip-26-0-165-164:1227464:1227884 [1] NCCL INFO Channel 05/0 : 81[1] -> 82[2] via P2P/IPC
+ip-26-0-165-59:1557272:1557687 [6] NCCL INFO Channel 09/0 : 102[6] -> 103[7] via P2P/IPC
+ip-26-0-161-138:96564:97009 [1] NCCL INFO Channel 04/0 : 33[1] -> 34[2] via P2P/IPC
+ip-26-0-165-59:1557266:1557689 [0] NCCL INFO Channel 10/0 : 96[0] -> 97[1] via P2P/IPC
+ip-26-0-165-59:1557271:1557690 [5] NCCL INFO Channel 07/0 : 101[5] -> 102[6] via P2P/IPC
+ip-26-0-168-30:235648:236064 [0] NCCL INFO Channel 14/0 : 104[0] -> 111[7] via P2P/IPC
+ip-26-0-161-123:32706:33122 [2] NCCL INFO Channel 14/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-161-123:32704:33123 [0] NCCL INFO Channel 06/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-160-242:1364907:1365329 [5] NCCL INFO Channel 01/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-160-225:124161:124582 [1] NCCL INFO Channel 13/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-202:153563:153986 [1] NCCL INFO Channel 01/0 : 89[1] -> 90[2] via P2P/IPC
+ip-26-0-161-103:96889:97309 [1] NCCL INFO Channel 03/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-160-242:1364905:1365328 [3] NCCL INFO Channel 01/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-161-138:96564:97009 [1] NCCL INFO Channel 05/0 : 33[1] -> 34[2] via P2P/IPC
+ip-26-0-160-242:1364903:1365334 [1] NCCL INFO Connected all rings
+ip-26-0-165-164:1227465:1227881 [2] NCCL INFO Channel 10/0 : 82[2] -> 83[3] via P2P/IPC
+ip-26-0-168-30:235648:236064 [0] NCCL INFO Channel 15/0 : 104[0] -> 111[7] via P2P/IPC
+ip-26-0-165-164:1227464:1227884 [1] NCCL INFO Channel 06/0 : 81[1] -> 82[2] via P2P/IPC
+ip-26-0-161-142:62881:63298 [0] NCCL INFO Channel 12/0 : 40[0] -> 41[1] via P2P/IPC
+ip-26-0-165-59:1557269:1557684 [3] NCCL INFO Channel 05/0 : 99[3] -> 100[4] via P2P/IPC
+ip-26-0-165-59:1557272:1557687 [6] NCCL INFO Channel 10/0 : 102[6] -> 103[7] via P2P/IPC
+ip-26-0-160-225:124161:124582 [1] NCCL INFO Channel 14/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-202:153563:153986 [1] NCCL INFO Channel 03/0 : 89[1] -> 90[2] via P2P/IPC
+ip-26-0-165-59:1557266:1557689 [0] NCCL INFO Channel 11/0 : 96[0] -> 97[1] via P2P/IPC
+ip-26-0-165-59:1557271:1557690 [5] NCCL INFO Channel 09/0 : 101[5] -> 102[6] via P2P/IPC
+ip-26-0-161-123:32706:33122 [2] NCCL INFO Channel 15/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-161-123:32704:33123 [0] NCCL INFO Channel 07/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-168-30:235650:236065 [2] NCCL INFO Channel 01/0 : 106[2] -> 107[3] via P2P/IPC
+ip-26-0-161-123:32711:33127 [7] NCCL INFO Connected all rings
+ip-26-0-160-225:124162:124587 [2] NCCL INFO Connected all rings
+ip-26-0-160-242:1364907:1365329 [5] NCCL INFO Channel 02/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-161-138:96564:97009 [1] NCCL INFO Channel 06/0 : 33[1] -> 34[2] via P2P/IPC
+ip-26-0-161-103:96889:97309 [1] NCCL INFO Channel 04/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-160-242:1364905:1365328 [3] NCCL INFO Channel 02/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-161-78:114518:114937 [6] NCCL INFO Connected all rings
+ip-26-0-161-78:114514:114932 [2] NCCL INFO Channel 07/0 : 66[2] -> 67[3] via P2P/IPC
+ip-26-0-165-131:1630207:1630648 [2] NCCL INFO Connected all rings
+ip-26-0-161-123:32708:33126 [4] NCCL INFO Channel 01/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-168-30:235650:236065 [2] NCCL INFO Channel 02/0 : 106[2] -> 107[3] via P2P/IPC
+ip-26-0-165-202:153564:153984 [2] NCCL INFO Channel 01/0 : 90[2] -> 91[3] via P2P/IPC
+ip-26-0-165-164:1227466:1227887 [3] NCCL INFO Channel 01/0 : 83[3] -> 84[4] via P2P/IPC
+ip-26-0-161-142:62881:63298 [0] NCCL INFO Channel 13/0 : 40[0] -> 41[1] via P2P/IPC
+ip-26-0-165-202:153563:153986 [1] NCCL INFO Channel 04/0 : 89[1] -> 90[2] via P2P/IPC
+ip-26-0-165-164:1227465:1227881 [2] NCCL INFO Channel 12/0 : 82[2] -> 83[3] via P2P/IPC
+ip-26-0-160-225:124161:124582 [1] NCCL INFO Channel 15/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-59:1557269:1557684 [3] NCCL INFO Channel 06/0 : 99[3] -> 100[4] via P2P/IPC
+ip-26-0-161-138:96564:97009 [1] NCCL INFO Channel 07/0 : 33[1] -> 34[2] via P2P/IPC
+ip-26-0-165-164:1227464:1227884 [1] NCCL INFO Channel 07/0 : 81[1] -> 82[2] via P2P/IPC
+ip-26-0-165-59:1557272:1557687 [6] NCCL INFO Channel 11/0 : 102[6] -> 103[7] via P2P/IPC
+ip-26-0-160-242:1364907:1365329 [5] NCCL INFO Channel 03/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-165-59:1557266:1557689 [0] NCCL INFO Channel 12/0 : 96[0] -> 97[1] via P2P/IPC
+ip-26-0-165-59:1557271:1557690 [5] NCCL INFO Channel 10/0 : 101[5] -> 102[6] via P2P/IPC
+ip-26-0-161-103:96889:97309 [1] NCCL INFO Channel 05/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-168-52:2722052:2722464 [7] NCCL INFO Connected all rings
+ip-26-0-161-123:32704:33123 [0] NCCL INFO Channel 13/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-168-52:2722050:2722471 [5] NCCL INFO Connected all rings
+ip-26-0-168-30:235650:236065 [2] NCCL INFO Channel 04/0 : 106[2] -> 107[3] via P2P/IPC
+ip-26-0-161-123:32705:33128 [1] NCCL INFO Channel 01/0 : 17[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364905:1365328 [3] NCCL INFO Channel 03/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-161-78:114514:114932 [2] NCCL INFO Channel 09/0 : 66[2] -> 67[3] via P2P/IPC
+ip-26-0-161-123:32708:33126 [4] NCCL INFO Channel 02/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-165-59:1557269:1557684 [3] NCCL INFO Channel 07/0 : 99[3] -> 100[4] via P2P/IPC
+ip-26-0-165-202:153564:153984 [2] NCCL INFO Channel 02/0 : 90[2] -> 91[3] via P2P/IPC
+ip-26-0-161-142:62881:63298 [0] NCCL INFO Channel 14/0 : 40[0] -> 41[1] via P2P/IPC
+ip-26-0-160-225:124160:124581 [0] NCCL INFO Channel 05/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-165-59:1557272:1557687 [6] NCCL INFO Channel 12/0 : 102[6] -> 103[7] via P2P/IPC
+ip-26-0-161-103:96889:97309 [1] NCCL INFO Channel 06/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-165-202:153563:153986 [1] NCCL INFO Channel 05/0 : 89[1] -> 90[2] via P2P/IPC
+ip-26-0-165-164:1227466:1227887 [3] NCCL INFO Channel 02/0 : 83[3] -> 84[4] via P2P/IPC
+ip-26-0-165-164:1227465:1227881 [2] NCCL INFO Channel 13/0 : 82[2] -> 83[3] via P2P/IPC
+ip-26-0-165-164:1227464:1227884 [1] NCCL INFO Channel 09/0 : 81[1] -> 82[2] via P2P/IPC
+ip-26-0-168-30:235650:236065 [2] NCCL INFO Channel 05/0 : 106[2] -> 107[3] via P2P/IPC
+ip-26-0-165-59:1557266:1557689 [0] NCCL INFO Channel 13/0 : 96[0] -> 97[1] via P2P/IPC
+ip-26-0-165-59:1557270:1557688 [4] NCCL INFO Channel 01/0 : 100[4] -> 101[5] via P2P/IPC
+ip-26-0-165-59:1557271:1557690 [5] NCCL INFO Channel 11/0 : 101[5] -> 102[6] via P2P/IPC
+ip-26-0-168-30:235653:236067 [5] NCCL INFO Channel 01/0 : 109[5] -> 110[6] via P2P/IPC
+ip-26-0-161-103:96889:97309 [1] NCCL INFO Channel 07/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-161-123:32704:33123 [0] NCCL INFO Channel 14/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-160-242:1364907:1365329 [5] NCCL INFO Channel 04/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-161-142:62881:63298 [0] NCCL INFO Channel 15/0 : 40[0] -> 41[1] via P2P/IPC
+ip-26-0-161-123:32708:33126 [4] NCCL INFO Channel 03/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-165-202:153566:153980 [4] NCCL INFO Connected all rings
+ip-26-0-165-59:1557269:1557684 [3] NCCL INFO Channel 09/0 : 99[3] -> 100[4] via P2P/IPC
+ip-26-0-165-202:153564:153984 [2] NCCL INFO Channel 04/0 : 90[2] -> 91[3] via P2P/IPC
+ip-26-0-165-202:153563:153986 [1] NCCL INFO Channel 06/0 : 89[1] -> 90[2] via P2P/IPC
+ip-26-0-161-78:114514:114932 [2] NCCL INFO Channel 10/0 : 66[2] -> 67[3] via P2P/IPC
+ip-26-0-168-52:2722047:2722468 [2] NCCL INFO Channel 01/0 : 122[2] -> 123[3] via P2P/IPC
+ip-26-0-168-30:235650:236065 [2] NCCL INFO Channel 06/0 : 106[2] -> 107[3] via P2P/IPC
+ip-26-0-165-59:1557272:1557687 [6] NCCL INFO Channel 13/0 : 102[6] -> 103[7] via P2P/IPC
+ip-26-0-160-225:124160:124581 [0] NCCL INFO Channel 06/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-165-59:1557266:1557689 [0] NCCL INFO Channel 14/0 : 96[0] -> 97[1] via P2P/IPC
+ip-26-0-168-30:235653:236067 [5] NCCL INFO Channel 02/0 : 109[5] -> 110[6] via P2P/IPC
+ip-26-0-165-164:1227465:1227881 [2] NCCL INFO Channel 14/0 : 82[2] -> 83[3] via P2P/IPC
+ip-26-0-165-164:1227466:1227887 [3] NCCL INFO Channel 03/0 : 83[3] -> 84[4] via P2P/IPC
+ip-26-0-165-164:1227464:1227884 [1] NCCL INFO Channel 11/0 : 81[1] -> 82[2] via P2P/IPC
+ip-26-0-165-59:1557271:1557690 [5] NCCL INFO Channel 12/0 : 101[5] -> 102[6] via P2P/IPC
+ip-26-0-165-59:1557270:1557688 [4] NCCL INFO Channel 02/0 : 100[4] -> 101[5] via P2P/IPC
+ip-26-0-161-103:96889:97309 [1] NCCL INFO Channel 09/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-161-123:32704:33123 [0] NCCL INFO Channel 15/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-161-142:62881:63298 [0] NCCL INFO Channel 05/0 : 40[0] -> 47[7] via P2P/IPC
+ip-26-0-161-78:114513:114933 [1] NCCL INFO Connected all rings
+ip-26-0-161-138:96564:97009 [1] NCCL INFO Channel 09/0 : 33[1] -> 34[2] via P2P/IPC
+ip-26-0-168-30:235650:236065 [2] NCCL INFO Channel 07/0 : 106[2] -> 107[3] via P2P/IPC
+ip-26-0-161-138:96567:97008 [4] NCCL INFO Connected all rings
+ip-26-0-161-103:96889:97309 [1] NCCL INFO Channel 11/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-165-202:153564:153984 [2] NCCL INFO Channel 05/0 : 90[2] -> 91[3] via P2P/IPC
+ip-26-0-165-202:153563:153986 [1] NCCL INFO Channel 07/0 : 89[1] -> 90[2] via P2P/IPC
+ip-26-0-161-78:114514:114932 [2] NCCL INFO Channel 12/0 : 66[2] -> 67[3] via P2P/IPC
+ip-26-0-165-59:1557269:1557684 [3] NCCL INFO Channel 10/0 : 99[3] -> 100[4] via P2P/IPC
+ip-26-0-165-59:1557272:1557687 [6] NCCL INFO Channel 14/0 : 102[6] -> 103[7] via P2P/IPC
+ip-26-0-160-242:1364905:1365328 [3] NCCL INFO Channel 05/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-161-123:32708:33126 [4] NCCL INFO Channel 04/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-168-30:235653:236067 [5] NCCL INFO Channel 03/0 : 109[5] -> 110[6] via P2P/IPC
+ip-26-0-160-242:1364908:1365332 [6] NCCL INFO Channel 01/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-165-59:1557266:1557689 [0] NCCL INFO Channel 15/0 : 96[0] -> 97[1] via P2P/IPC
+ip-26-0-165-59:1557270:1557688 [4] NCCL INFO Channel 03/0 : 100[4] -> 101[5] via P2P/IPC
+ip-26-0-168-52:2722047:2722468 [2] NCCL INFO Channel 02/0 : 122[2] -> 123[3] via P2P/IPC
+ip-26-0-165-59:1557271:1557690 [5] NCCL INFO Channel 13/0 : 101[5] -> 102[6] via P2P/IPC
+ip-26-0-168-30:235650:236065 [2] NCCL INFO Channel 09/0 : 106[2] -> 107[3] via P2P/IPC
+ip-26-0-160-242:1364907:1365329 [5] NCCL INFO Channel 05/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-161-142:62884:63299 [3] NCCL INFO Connected all rings
+ip-26-0-161-123:32708:33126 [4] NCCL INFO Channel 06/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-168-30:235653:236067 [5] NCCL INFO Channel 04/0 : 109[5] -> 110[6] via P2P/IPC
+ip-26-0-161-103:96889:97309 [1] NCCL INFO Channel 12/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-160-242:1364905:1365328 [3] NCCL INFO Channel 06/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-165-59:1557269:1557684 [3] NCCL INFO Channel 11/0 : 99[3] -> 100[4] via P2P/IPC
+ip-26-0-165-164:1227465:1227881 [2] NCCL INFO Channel 15/0 : 82[2] -> 83[3] via P2P/IPC
+ip-26-0-160-242:1364908:1365332 [6] NCCL INFO Channel 02/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-161-78:114514:114932 [2] NCCL INFO Channel 13/0 : 66[2] -> 67[3] via P2P/IPC
+ip-26-0-165-59:1557270:1557688 [4] NCCL INFO Channel 04/0 : 100[4] -> 101[5] via P2P/IPC
+ip-26-0-165-59:1557271:1557690 [5] NCCL INFO Channel 15/0 : 101[5] -> 102[6] via P2P/IPC
+ip-26-0-161-142:62887:63305 [6] NCCL INFO Channel 01/0 : 46[6] -> 47[7] via P2P/IPC
+ip-26-0-161-142:62883:63301 [2] NCCL INFO Connected all rings
+ip-26-0-165-164:1227466:1227887 [3] NCCL INFO Channel 05/0 : 83[3] -> 84[4] via P2P/IPC
+ip-26-0-165-131:1630210:1630653 [5] NCCL INFO Connected all rings
+ip-26-0-165-59:1557266:1557689 [0] NCCL INFO Channel 05/0 : 96[0] -> 103[7] via P2P/IPC
+ip-26-0-161-138:96570:97005 [7] NCCL INFO Connected all rings
+ip-26-0-168-52:2722047:2722468 [2] NCCL INFO Channel 04/0 : 122[2] -> 123[3] via P2P/IPC
+ip-26-0-168-30:235650:236065 [2] NCCL INFO Channel 10/0 : 106[2] -> 107[3] via P2P/IPC
+ip-26-0-165-59:1557267:1557683 [1] NCCL INFO Channel 09/0 : 97[1] -> 105[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32708:33126 [4] NCCL INFO Channel 07/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-160-242:1364907:1365329 [5] NCCL INFO Channel 07/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-168-34:272152:272575 [2] NCCL INFO Connected all rings
+ip-26-0-161-103:96893:97310 [5] NCCL INFO Connected all rings
+ip-26-0-168-30:235653:236067 [5] NCCL INFO Channel 05/0 : 109[5] -> 110[6] via P2P/IPC
+ip-26-0-161-103:96889:97309 [1] NCCL INFO Channel 13/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-161-153:2636085:2636509 [1] NCCL INFO Connected all rings
+ip-26-0-165-59:1557269:1557684 [3] NCCL INFO Channel 13/0 : 99[3] -> 100[4] via P2P/IPC
+ip-26-0-161-138:96564:97009 [1] NCCL INFO Channel 11/0 : 33[1] -> 34[2] via P2P/IPC
+ip-26-0-165-164:1227466:1227887 [3] NCCL INFO Channel 06/0 : 83[3] -> 84[4] via P2P/IPC
+ip-26-0-161-123:32708:33126 [4] NCCL INFO Channel 09/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-160-242:1364905:1365328 [3] NCCL INFO Channel 07/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-168-30:235654:236070 [6] NCCL INFO Channel 01/0 : 110[6] -> 111[7] via P2P/IPC
+ip-26-0-165-59:1557270:1557688 [4] NCCL INFO Channel 06/0 : 100[4] -> 101[5] via P2P/IPC
+ip-26-0-160-242:1364908:1365332 [6] NCCL INFO Channel 03/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-161-78:114514:114932 [2] NCCL INFO Channel 14/0 : 66[2] -> 67[3] via P2P/IPC
+ip-26-0-168-30:235650:236065 [2] NCCL INFO Channel 12/0 : 106[2] -> 107[3] via P2P/IPC
+ip-26-0-161-78:114516:114934 [4] NCCL INFO Channel 01/0 : 68[4] -> 69[5] via P2P/IPC
+ip-26-0-165-59:1557266:1557689 [0] NCCL INFO Channel 06/0 : 96[0] -> 103[7] via P2P/IPC
+ip-26-0-161-78:114515:114936 [3] NCCL INFO Channel 01/0 : 67[3] -> 68[4] via P2P/IPC
+ip-26-0-161-103:96889:97309 [1] NCCL INFO Channel 14/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-168-52:2722051:2722465 [6] NCCL INFO Channel 01/0 : 126[6] -> 127[7] via P2P/IPC
+ip-26-0-165-59:1557272:1557687 [6] NCCL INFO Channel 14/0 : 102[6] -> 110[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-52:2722047:2722468 [2] NCCL INFO Channel 05/0 : 122[2] -> 123[3] via P2P/IPC
+ip-26-0-165-164:1227464:1227884 [1] NCCL INFO Channel 12/0 : 81[1] -> 82[2] via P2P/IPC
+ip-26-0-160-242:1364907:1365329 [5] NCCL INFO Channel 09/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-161-142:62881:63298 [0] NCCL INFO Channel 06/0 : 40[0] -> 47[7] via P2P/IPC
+ip-26-0-168-30:235653:236067 [5] NCCL INFO Channel 07/0 : 109[5] -> 110[6] via P2P/IPC
+ip-26-0-165-59:1557269:1557684 [3] NCCL INFO Channel 14/0 : 99[3] -> 100[4] via P2P/IPC
+ip-26-0-161-138:96565:97004 [2] NCCL INFO Channel 01/0 : 34[2] -> 35[3] via P2P/IPC
+ip-26-0-165-164:1227466:1227887 [3] NCCL INFO Channel 07/0 : 83[3] -> 84[4] via P2P/IPC
+ip-26-0-165-164:1227468:1227883 [5] NCCL INFO Channel 01/0 : 85[5] -> 86[6] via P2P/IPC
+ip-26-0-161-138:96564:97009 [1] NCCL INFO Channel 12/0 : 33[1] -> 34[2] via P2P/IPC
+ip-26-0-165-59:1557270:1557688 [4] NCCL INFO Channel 07/0 : 100[4] -> 101[5] via P2P/IPC
+ip-26-0-161-123:32708:33126 [4] NCCL INFO Channel 10/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-168-30:235654:236070 [6] NCCL INFO Channel 02/0 : 110[6] -> 111[7] via P2P/IPC
+ip-26-0-161-123:32709:33129 [5] NCCL INFO Channel 01/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-161-142:62887:63305 [6] NCCL INFO Channel 02/0 : 46[6] -> 47[7] via P2P/IPC
+ip-26-0-165-59:1557266:1557689 [0] NCCL INFO Channel 07/0 : 96[0] -> 103[7] via P2P/IPC
+ip-26-0-160-225:124160:124581 [0] NCCL INFO Channel 07/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-168-30:235650:236065 [2] NCCL INFO Channel 13/0 : 106[2] -> 107[3] via P2P/IPC
+ip-26-0-160-242:1364905:1365328 [3] NCCL INFO Channel 09/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-160-242:1364908:1365332 [6] NCCL INFO Channel 04/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-165-164:1227464:1227884 [1] NCCL INFO Channel 13/0 : 81[1] -> 82[2] via P2P/IPC
+ip-26-0-161-78:114514:114932 [2] NCCL INFO Channel 15/0 : 66[2] -> 67[3] via P2P/IPC
+ip-26-0-161-142:62881:63298 [0] NCCL INFO Channel 07/0 : 40[0] -> 47[7] via P2P/IPC
+ip-26-0-168-30:235653:236067 [5] NCCL INFO Channel 09/0 : 109[5] -> 110[6] via P2P/IPC
+ip-26-0-161-103:96889:97309 [1] NCCL INFO Channel 15/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-161-78:114516:114934 [4] NCCL INFO Channel 02/0 : 68[4] -> 69[5] via P2P/IPC
+ip-26-0-161-78:114515:114936 [3] NCCL INFO Channel 02/0 : 67[3] -> 68[4] via P2P/IPC
+ip-26-0-161-123:32708:33126 [4] NCCL INFO Channel 11/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-165-59:1557269:1557684 [3] NCCL INFO Channel 15/0 : 99[3] -> 100[4] via P2P/IPC
+ip-26-0-161-123:32709:33129 [5] NCCL INFO Channel 02/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-165-59:1557270:1557688 [4] NCCL INFO Channel 09/0 : 100[4] -> 101[5] via P2P/IPC
+ip-26-0-168-30:235654:236070 [6] NCCL INFO Channel 03/0 : 110[6] -> 111[7] via P2P/IPC
+ip-26-0-161-142:62887:63305 [6] NCCL INFO Channel 03/0 : 46[6] -> 47[7] via P2P/IPC
+ip-26-0-160-242:1364907:1365329 [5] NCCL INFO Channel 10/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-168-30:235650:236065 [2] NCCL INFO Channel 14/0 : 106[2] -> 107[3] via P2P/IPC
+ip-26-0-165-202:153565:153985 [3] NCCL INFO Channel 01/0 : 91[3] -> 92[4] via P2P/IPC
+ip-26-0-168-52:2722051:2722465 [6] NCCL INFO Channel 02/0 : 126[6] -> 127[7] via P2P/IPC
+ip-26-0-161-138:96564:97009 [1] NCCL INFO Channel 13/0 : 33[1] -> 34[2] via P2P/IPC
+ip-26-0-161-138:96565:97004 [2] NCCL INFO Channel 02/0 : 34[2] -> 35[3] via P2P/IPC
+ip-26-0-160-242:1364904:1365335 [2] NCCL INFO Channel 01/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-168-52:2722047:2722468 [2] NCCL INFO Channel 06/0 : 122[2] -> 123[3] via P2P/IPC
+ip-26-0-165-164:1227466:1227887 [3] NCCL INFO Channel 09/0 : 83[3] -> 84[4] via P2P/IPC
+ip-26-0-165-59:1557266:1557689 [0] NCCL INFO Channel 13/0 : 96[0] -> 103[7] via P2P/IPC
+ip-26-0-168-34:272153:272573 [3] NCCL INFO Connected all rings
+ip-26-0-165-164:1227468:1227883 [5] NCCL INFO Channel 02/0 : 85[5] -> 86[6] via P2P/IPC
+ip-26-0-160-225:124160:124581 [0] NCCL INFO Channel 13/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-168-34:272151:272571 [1] NCCL INFO Channel 01/0 : 113[1] -> 114[2] via P2P/IPC
+ip-26-0-165-202:153563:153986 [1] NCCL INFO Channel 09/0 : 89[1] -> 90[2] via P2P/IPC
+ip-26-0-161-142:62881:63298 [0] NCCL INFO Channel 13/0 : 40[0] -> 47[7] via P2P/IPC
+ip-26-0-165-164:1227464:1227884 [1] NCCL INFO Channel 14/0 : 81[1] -> 82[2] via P2P/IPC
+ip-26-0-168-30:235653:236067 [5] NCCL INFO Channel 10/0 : 109[5] -> 110[6] via P2P/IPC
+ip-26-0-160-242:1364905:1365328 [3] NCCL INFO Channel 10/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-165-164:1227469:1227882 [6] NCCL INFO Channel 01/0 : 86[6] -> 87[7] via P2P/IPC
+ip-26-0-161-103:96888:97306 [0] NCCL INFO Channel 05/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-160-242:1364908:1365332 [6] NCCL INFO Channel 05/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-168-30:235654:236070 [6] NCCL INFO Channel 04/0 : 110[6] -> 111[7] via P2P/IPC
+ip-26-0-168-30:235650:236065 [2] NCCL INFO Channel 15/0 : 106[2] -> 107[3] via P2P/IPC
+ip-26-0-161-78:114517:114935 [5] NCCL INFO Channel 01/0 : 69[5] -> 70[6] via P2P/IPC
+ip-26-0-161-142:62887:63305 [6] NCCL INFO Channel 04/0 : 46[6] -> 47[7] via P2P/IPC
+ip-26-0-165-59:1557270:1557688 [4] NCCL INFO Channel 10/0 : 100[4] -> 101[5] via P2P/IPC
+ip-26-0-161-123:32708:33126 [4] NCCL INFO Channel 12/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-161-123:32709:33129 [5] NCCL INFO Channel 03/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-165-202:153565:153985 [3] NCCL INFO Channel 02/0 : 91[3] -> 92[4] via P2P/IPC
+ip-26-0-161-78:114516:114934 [4] NCCL INFO Channel 03/0 : 68[4] -> 69[5] via P2P/IPC
+ip-26-0-161-78:114515:114936 [3] NCCL INFO Channel 03/0 : 67[3] -> 68[4] via P2P/IPC
+ip-26-0-165-59:1557266:1557689 [0] NCCL INFO Channel 14/0 : 96[0] -> 103[7] via P2P/IPC
+ip-26-0-165-59:1557268:1557686 [2] NCCL INFO Channel 10/0 : 98[2] -> 106[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:124166:124583 [6] NCCL INFO Channel 01/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-168-30:235653:236067 [5] NCCL INFO Channel 11/0 : 109[5] -> 110[6] via P2P/IPC
+ip-26-0-160-242:1364907:1365329 [5] NCCL INFO Channel 11/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-161-142:62881:63298 [0] NCCL INFO Channel 14/0 : 40[0] -> 47[7] via P2P/IPC
+ip-26-0-161-142:62885:63304 [4] NCCL INFO Channel 01/0 : 44[4] -> 45[5] via P2P/IPC
+ip-26-0-165-202:153563:153986 [1] NCCL INFO Channel 11/0 : 89[1] -> 90[2] via P2P/IPC
+ip-26-0-160-242:1364904:1365335 [2] NCCL INFO Channel 02/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-165-164:1227466:1227887 [3] NCCL INFO Channel 10/0 : 83[3] -> 84[4] via P2P/IPC
+ip-26-0-165-164:1227468:1227883 [5] NCCL INFO Channel 03/0 : 85[5] -> 86[6] via P2P/IPC
+ip-26-0-161-138:96565:97004 [2] NCCL INFO Channel 04/0 : 34[2] -> 35[3] via P2P/IPC
+ip-26-0-165-59:1557270:1557688 [4] NCCL INFO Channel 11/0 : 100[4] -> 101[5] via P2P/IPC
+ip-26-0-168-52:2722051:2722465 [6] NCCL INFO Channel 03/0 : 126[6] -> 127[7] via P2P/IPC
+ip-26-0-161-138:96564:97009 [1] NCCL INFO Channel 14/0 : 33[1] -> 34[2] via P2P/IPC
+ip-26-0-165-59:1557266:1557689 [0] NCCL INFO Channel 15/0 : 96[0] -> 103[7] via P2P/IPC
+ip-26-0-160-242:1364905:1365328 [3] NCCL INFO Channel 11/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-161-123:32709:33129 [5] NCCL INFO Channel 04/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-161-123:32708:33126 [4] NCCL INFO Channel 14/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-168-30:235654:236070 [6] NCCL INFO Channel 05/0 : 110[6] -> 111[7] via P2P/IPC
+ip-26-0-168-52:2722047:2722468 [2] NCCL INFO Channel 07/0 : 122[2] -> 123[3] via P2P/IPC
+ip-26-0-165-164:1227464:1227884 [1] NCCL INFO Channel 15/0 : 81[1] -> 82[2] via P2P/IPC
+ip-26-0-161-142:62887:63305 [6] NCCL INFO Channel 05/0 : 46[6] -> 47[7] via P2P/IPC
+ip-26-0-165-164:1227469:1227882 [6] NCCL INFO Channel 02/0 : 86[6] -> 87[7] via P2P/IPC
+ip-26-0-165-202:153565:153985 [3] NCCL INFO Channel 03/0 : 91[3] -> 92[4] via P2P/IPC
+ip-26-0-168-34:272151:272571 [1] NCCL INFO Channel 03/0 : 113[1] -> 114[2] via P2P/IPC
+ip-26-0-168-30:235653:236067 [5] NCCL INFO Channel 12/0 : 109[5] -> 110[6] via P2P/IPC
+ip-26-0-161-78:114517:114935 [5] NCCL INFO Channel 02/0 : 69[5] -> 70[6] via P2P/IPC
+ip-26-0-160-225:124160:124581 [0] NCCL INFO Channel 14/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-160-242:1364908:1365332 [6] NCCL INFO Channel 06/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-161-78:114515:114936 [3] NCCL INFO Channel 05/0 : 67[3] -> 68[4] via P2P/IPC
+ip-26-0-161-78:114516:114934 [4] NCCL INFO Channel 04/0 : 68[4] -> 69[5] via P2P/IPC
+ip-26-0-168-30:235649:236071 [1] NCCL INFO Channel 09/0 : 97[1] -> 105[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557270:1557688 [4] NCCL INFO Channel 12/0 : 100[4] -> 101[5] via P2P/IPC
+ip-26-0-161-142:62881:63298 [0] NCCL INFO Channel 15/0 : 40[0] -> 47[7] via P2P/IPC
+ip-26-0-168-52:2722048:2722469 [3] NCCL INFO Channel 01/0 : 123[3] -> 124[4] via P2P/IPC
+ip-26-0-168-30:235654:236070 [6] NCCL INFO Channel 06/0 : 110[6] -> 111[7] via P2P/IPC
+ip-26-0-161-142:62885:63304 [4] NCCL INFO Channel 02/0 : 44[4] -> 45[5] via P2P/IPC
+ip-26-0-168-30:235649:236071 [1] NCCL INFO Channel 01/0 : 105[1] -> 113[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153563:153986 [1] NCCL INFO Channel 12/0 : 89[1] -> 90[2] via P2P/IPC
+ip-26-0-165-202:153564:153984 [2] NCCL INFO Channel 06/0 : 90[2] -> 91[3] via P2P/IPC
+ip-26-0-168-30:235653:236067 [5] NCCL INFO Channel 13/0 : 109[5] -> 110[6] via P2P/IPC
+ip-26-0-168-30:235649:236071 [1] NCCL INFO Channel 09/0 : 105[1] -> 113[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227466:1227887 [3] NCCL INFO Channel 11/0 : 83[3] -> 84[4] via P2P/IPC
+ip-26-0-165-59:1557270:1557688 [4] NCCL INFO Channel 14/0 : 100[4] -> 101[5] via P2P/IPC
+ip-26-0-160-225:124166:124583 [6] NCCL INFO Channel 02/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-242:1364907:1365329 [5] NCCL INFO Channel 12/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-165-164:1227468:1227883 [5] NCCL INFO Channel 04/0 : 85[5] -> 86[6] via P2P/IPC
+ip-26-0-165-59:1557273:1557685 [7] NCCL INFO Channel 15/0 : 103[7] -> 111[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557266:1557689 [0] NCCL INFO Channel 08/0 : 96[0] -> 104[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235654:236070 [6] NCCL INFO Channel 09/0 : 110[6] -> 111[7] via P2P/IPC
+ip-26-0-161-123:32709:33129 [5] NCCL INFO Channel 05/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-161-142:62887:63305 [6] NCCL INFO Channel 06/0 : 46[6] -> 47[7] via P2P/IPC
+ip-26-0-160-242:1364904:1365335 [2] NCCL INFO Channel 04/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-161-123:32708:33126 [4] NCCL INFO Channel 15/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-165-164:1227469:1227882 [6] NCCL INFO Channel 03/0 : 86[6] -> 87[7] via P2P/IPC
+ip-26-0-160-242:1364905:1365328 [3] NCCL INFO Channel 13/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-165-59:1557267:1557683 [1] NCCL INFO Channel 01/0 : 81[1] -> 97[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235653:236067 [5] NCCL INFO Channel 15/0 : 109[5] -> 110[6] via P2P/IPC
+ip-26-0-165-59:1557270:1557688 [4] NCCL INFO Channel 15/0 : 100[4] -> 101[5] via P2P/IPC
+ip-26-0-168-52:2722051:2722465 [6] NCCL INFO Channel 04/0 : 126[6] -> 127[7] via P2P/IPC
+ip-26-0-165-164:1227463:1227880 [0] NCCL INFO Channel 05/0 : 80[0] -> 87[7] via P2P/IPC
+ip-26-0-160-225:124160:124581 [0] NCCL INFO Channel 15/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-165-59:1557267:1557683 [1] NCCL INFO Channel 01/0 : 97[1] -> 113[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114517:114935 [5] NCCL INFO Channel 03/0 : 69[5] -> 70[6] via P2P/IPC
+ip-26-0-165-202:153565:153985 [3] NCCL INFO Channel 05/0 : 91[3] -> 92[4] via P2P/IPC
+ip-26-0-161-78:114516:114934 [4] NCCL INFO Channel 06/0 : 68[4] -> 69[5] via P2P/IPC
+ip-26-0-161-142:62885:63304 [4] NCCL INFO Channel 03/0 : 44[4] -> 45[5] via P2P/IPC
+ip-26-0-160-242:1364908:1365332 [6] NCCL INFO Channel 09/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-161-78:114515:114936 [3] NCCL INFO Channel 06/0 : 67[3] -> 68[4] via P2P/IPC
+ip-26-0-168-30:235654:236070 [6] NCCL INFO Channel 10/0 : 110[6] -> 111[7] via P2P/IPC
+ip-26-0-168-52:2722047:2722468 [2] NCCL INFO Channel 09/0 : 122[2] -> 123[3] via P2P/IPC
+ip-26-0-165-164:1227464:1227884 [1] NCCL INFO Channel 01/0 : 73[1] -> 81[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96891:97311 [3] NCCL INFO Connected all rings
+ip-26-0-161-123:32709:33129 [5] NCCL INFO Channel 07/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-165-164:1227464:1227884 [1] NCCL INFO Channel 09/0 : 73[1] -> 81[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62887:63305 [6] NCCL INFO Channel 09/0 : 46[6] -> 47[7] via P2P/IPC
+ip-26-0-161-103:96888:97306 [0] NCCL INFO Channel 06/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-165-164:1227466:1227887 [3] NCCL INFO Channel 13/0 : 83[3] -> 84[4] via P2P/IPC
+ip-26-0-165-164:1227468:1227883 [5] NCCL INFO Channel 05/0 : 85[5] -> 86[6] via P2P/IPC
+ip-26-0-160-225:124166:124583 [6] NCCL INFO Channel 03/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-164:1227464:1227884 [1] NCCL INFO Channel 01/0 : 81[1] -> 89[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235654:236070 [6] NCCL INFO Channel 11/0 : 110[6] -> 111[7] via P2P/IPC
+ip-26-0-165-202:153563:153986 [1] NCCL INFO Channel 13/0 : 89[1] -> 90[2] via P2P/IPC
+ip-26-0-165-202:153564:153984 [2] NCCL INFO Channel 07/0 : 90[2] -> 91[3] via P2P/IPC
+ip-26-0-165-131:1630206:1630647 [1] NCCL INFO Channel 01/0 : 73[1] -> 74[2] via P2P/IPC
+ip-26-0-165-164:1227469:1227882 [6] NCCL INFO Channel 04/0 : 86[6] -> 87[7] via P2P/IPC
+ip-26-0-160-242:1364907:1365329 [5] NCCL INFO Channel 13/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-168-52:2722048:2722469 [3] NCCL INFO Channel 02/0 : 123[3] -> 124[4] via P2P/IPC
+ip-26-0-168-30:235654:236070 [6] NCCL INFO Channel 12/0 : 110[6] -> 111[7] via P2P/IPC
+ip-26-0-160-242:1364904:1365335 [2] NCCL INFO Channel 05/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-161-123:32709:33129 [5] NCCL INFO Channel 09/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-161-142:62885:63304 [4] NCCL INFO Channel 04/0 : 44[4] -> 45[5] via P2P/IPC
+ip-26-0-161-78:114517:114935 [5] NCCL INFO Channel 04/0 : 69[5] -> 70[6] via P2P/IPC
+ip-26-0-160-242:1364905:1365328 [3] NCCL INFO Channel 14/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-165-164:1227463:1227880 [0] NCCL INFO Channel 06/0 : 80[0] -> 87[7] via P2P/IPC
+ip-26-0-161-142:62886:63303 [5] NCCL INFO Channel 01/0 : 45[5] -> 46[6] via P2P/IPC
+ip-26-0-161-78:114515:114936 [3] NCCL INFO Channel 07/0 : 67[3] -> 68[4] via P2P/IPC
+ip-26-0-161-142:62887:63305 [6] NCCL INFO Channel 10/0 : 46[6] -> 47[7] via P2P/IPC
+ip-26-0-161-138:96565:97004 [2] NCCL INFO Channel 05/0 : 34[2] -> 35[3] via P2P/IPC
+ip-26-0-168-52:2722047:2722468 [2] NCCL INFO Channel 10/0 : 122[2] -> 123[3] via P2P/IPC
+ip-26-0-161-123:32709:33129 [5] NCCL INFO Channel 10/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-168-30:235654:236070 [6] NCCL INFO Channel 13/0 : 110[6] -> 111[7] via P2P/IPC
+ip-26-0-161-78:114516:114934 [4] NCCL INFO Channel 07/0 : 68[4] -> 69[5] via P2P/IPC
+ip-26-0-165-164:1227466:1227887 [3] NCCL INFO Channel 14/0 : 83[3] -> 84[4] via P2P/IPC
+ip-26-0-161-138:96564:97009 [1] NCCL INFO Channel 15/0 : 33[1] -> 34[2] via P2P/IPC
+ip-26-0-161-103:96888:97306 [0] NCCL INFO Channel 07/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-168-34:272151:272571 [1] NCCL INFO Channel 04/0 : 113[1] -> 114[2] via P2P/IPC
+ip-26-0-160-242:1364908:1365332 [6] NCCL INFO Channel 10/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-165-131:1630209:1630651 [4] NCCL INFO Connected all rings
+ip-26-0-160-225:124164:124585 [4] NCCL INFO Channel 01/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-164:1227468:1227883 [5] NCCL INFO Channel 07/0 : 85[5] -> 86[6] via P2P/IPC
+ip-26-0-160-225:124166:124583 [6] NCCL INFO Channel 04/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-168-52:2722049:2722467 [4] NCCL INFO Channel 01/0 : 124[4] -> 125[5] via P2P/IPC
+ip-26-0-168-30:235654:236070 [6] NCCL INFO Channel 14/0 : 110[6] -> 111[7] via P2P/IPC
+ip-26-0-165-131:1630206:1630647 [1] NCCL INFO Channel 03/0 : 73[1] -> 74[2] via P2P/IPC
+ip-26-0-161-123:32709:33129 [5] NCCL INFO Channel 11/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-161-142:62885:63304 [4] NCCL INFO Channel 06/0 : 44[4] -> 45[5] via P2P/IPC
+ip-26-0-161-138:96566:97010 [3] NCCL INFO Channel 01/0 : 35[3] -> 36[4] via P2P/IPC
+ip-26-0-168-52:2722051:2722465 [6] NCCL INFO Channel 05/0 : 126[6] -> 127[7] via P2P/IPC
+ip-26-0-165-164:1227469:1227882 [6] NCCL INFO Channel 05/0 : 86[6] -> 87[7] via P2P/IPC
+ip-26-0-161-78:114517:114935 [5] NCCL INFO Channel 05/0 : 69[5] -> 70[6] via P2P/IPC
+ip-26-0-168-52:2722047:2722468 [2] NCCL INFO Channel 12/0 : 122[2] -> 123[3] via P2P/IPC
+ip-26-0-160-242:1364907:1365329 [5] NCCL INFO Channel 15/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-161-142:62886:63303 [5] NCCL INFO Channel 02/0 : 45[5] -> 46[6] via P2P/IPC
+ip-26-0-165-164:1227463:1227880 [0] NCCL INFO Channel 07/0 : 80[0] -> 87[7] via P2P/IPC
+ip-26-0-160-242:1364904:1365335 [2] NCCL INFO Channel 06/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-161-142:62887:63305 [6] NCCL INFO Channel 11/0 : 46[6] -> 47[7] via P2P/IPC
+ip-26-0-165-202:153568:153982 [6] NCCL INFO Connected all rings
+ip-26-0-161-123:32709:33129 [5] NCCL INFO Channel 12/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-161-78:114515:114936 [3] NCCL INFO Channel 09/0 : 67[3] -> 68[4] via P2P/IPC
+ip-26-0-161-138:96565:97004 [2] NCCL INFO Channel 06/0 : 34[2] -> 35[3] via P2P/IPC
+ip-26-0-160-242:1364905:1365328 [3] NCCL INFO Channel 15/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-161-78:114516:114934 [4] NCCL INFO Channel 09/0 : 68[4] -> 69[5] via P2P/IPC
+ip-26-0-168-34:272151:272571 [1] NCCL INFO Channel 05/0 : 113[1] -> 114[2] via P2P/IPC
+ip-26-0-161-103:96888:97306 [0] NCCL INFO Channel 13/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-165-164:1227466:1227887 [3] NCCL INFO Channel 15/0 : 83[3] -> 84[4] via P2P/IPC
+ip-26-0-161-138:96563:97003 [0] NCCL INFO Channel 05/0 : 32[0] -> 39[7] via P2P/IPC
+ip-26-0-165-164:1227468:1227883 [5] NCCL INFO Channel 09/0 : 85[5] -> 86[6] via P2P/IPC
+ip-26-0-160-225:124166:124583 [6] NCCL INFO Channel 05/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-225:124164:124585 [4] NCCL INFO Channel 02/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-142:62885:63304 [4] NCCL INFO Channel 07/0 : 44[4] -> 45[5] via P2P/IPC
+ip-26-0-168-30:235654:236070 [6] NCCL INFO Channel 14/0 : 102[6] -> 110[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32709:33129 [5] NCCL INFO Channel 13/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-160-242:1364908:1365332 [6] NCCL INFO Channel 11/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-165-59:1557269:1557684 [3] NCCL INFO Channel 11/0 : 99[3] -> 107[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235654:236070 [6] NCCL INFO Channel 06/0 : 110[6] -> 118[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:124162:124587 [2] NCCL INFO Channel 01/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-131:1630206:1630647 [1] NCCL INFO Channel 04/0 : 73[1] -> 74[2] via P2P/IPC
+ip-26-0-161-78:114517:114935 [5] NCCL INFO Channel 07/0 : 69[5] -> 70[6] via P2P/IPC
+ip-26-0-165-59:1557271:1557690 [5] NCCL INFO Channel 13/0 : 101[5] -> 109[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235654:236070 [6] NCCL INFO Channel 14/0 : 110[6] -> 118[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-52:2722049:2722467 [4] NCCL INFO Channel 02/0 : 124[4] -> 125[5] via P2P/IPC
+ip-26-0-165-202:153565:153985 [3] NCCL INFO Channel 06/0 : 91[3] -> 92[4] via P2P/IPC
+ip-26-0-165-59:1557270:1557688 [4] NCCL INFO Channel 12/0 : 100[4] -> 108[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62887:63305 [6] NCCL INFO Channel 12/0 : 46[6] -> 47[7] via P2P/IPC
+ip-26-0-168-52:2722051:2722465 [6] NCCL INFO Channel 06/0 : 126[6] -> 127[7] via P2P/IPC
+ip-26-0-161-142:62886:63303 [5] NCCL INFO Channel 03/0 : 45[5] -> 46[6] via P2P/IPC
+ip-26-0-165-164:1227467:1227886 [4] NCCL INFO Channel 01/0 : 84[4] -> 85[5] via P2P/IPC
+ip-26-0-168-52:2722047:2722468 [2] NCCL INFO Channel 13/0 : 122[2] -> 123[3] via P2P/IPC
+ip-26-0-161-138:96566:97010 [3] NCCL INFO Channel 02/0 : 35[3] -> 36[4] via P2P/IPC
+ip-26-0-165-164:1227469:1227882 [6] NCCL INFO Channel 06/0 : 86[6] -> 87[7] via P2P/IPC
+ip-26-0-165-164:1227463:1227880 [0] NCCL INFO Channel 13/0 : 80[0] -> 87[7] via P2P/IPC
+ip-26-0-165-202:153563:153986 [1] NCCL INFO Channel 14/0 : 89[1] -> 90[2] via P2P/IPC
+ip-26-0-165-202:153564:153984 [2] NCCL INFO Channel 09/0 : 90[2] -> 91[3] via P2P/IPC
+ip-26-0-161-123:32709:33129 [5] NCCL INFO Channel 15/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-165-59:1557272:1557687 [6] NCCL INFO Channel 06/0 : 86[6] -> 102[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364904:1365335 [2] NCCL INFO Channel 07/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-165-59:1557272:1557687 [6] NCCL INFO Channel 06/0 : 102[6] -> 118[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62885:63304 [4] NCCL INFO Channel 09/0 : 44[4] -> 45[5] via P2P/IPC
+ip-26-0-165-164:1227468:1227883 [5] NCCL INFO Channel 10/0 : 85[5] -> 86[6] via P2P/IPC
+ip-26-0-168-34:272151:272571 [1] NCCL INFO Channel 06/0 : 113[1] -> 114[2] via P2P/IPC
+ip-26-0-160-225:124164:124585 [4] NCCL INFO Channel 03/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-225:124166:124583 [6] NCCL INFO Channel 06/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-225:124162:124587 [2] NCCL INFO Channel 02/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-225:124165:124584 [5] NCCL INFO Connected all rings
+ip-26-0-161-138:96565:97004 [2] NCCL INFO Channel 07/0 : 34[2] -> 35[3] via P2P/IPC
+ip-26-0-165-131:1630206:1630647 [1] NCCL INFO Channel 05/0 : 73[1] -> 74[2] via P2P/IPC
+ip-26-0-160-242:1364908:1365332 [6] NCCL INFO Channel 12/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-161-142:62887:63305 [6] NCCL INFO Channel 13/0 : 46[6] -> 47[7] via P2P/IPC
+ip-26-0-165-131:1630211:1630649 [6] NCCL INFO Connected all rings
+ip-26-0-160-242:1364903:1365334 [1] NCCL INFO Channel 01/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-161-142:62886:63303 [5] NCCL INFO Channel 04/0 : 45[5] -> 46[6] via P2P/IPC
+ip-26-0-165-202:153565:153985 [3] NCCL INFO Channel 07/0 : 91[3] -> 92[4] via P2P/IPC
+ip-26-0-165-164:1227465:1227881 [2] NCCL INFO Channel 02/0 : 74[2] -> 82[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-52:2722049:2722467 [4] NCCL INFO Channel 03/0 : 124[4] -> 125[5] via P2P/IPC
+ip-26-0-165-164:1227465:1227881 [2] NCCL INFO Channel 10/0 : 74[2] -> 82[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227469:1227882 [6] NCCL INFO Channel 09/0 : 86[6] -> 87[7] via P2P/IPC
+ip-26-0-165-164:1227467:1227886 [4] NCCL INFO Channel 02/0 : 84[4] -> 85[5] via P2P/IPC
+ip-26-0-165-164:1227463:1227880 [0] NCCL INFO Channel 14/0 : 80[0] -> 87[7] via P2P/IPC
+ip-26-0-161-138:96563:97003 [0] NCCL INFO Channel 06/0 : 32[0] -> 39[7] via P2P/IPC
+ip-26-0-161-103:96888:97306 [0] NCCL INFO Channel 14/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-165-164:1227465:1227881 [2] NCCL INFO Channel 02/0 : 82[2] -> 90[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-52:2722051:2722465 [6] NCCL INFO Channel 09/0 : 126[6] -> 127[7] via P2P/IPC
+ip-26-0-168-30:235652:236066 [4] NCCL INFO Connected all rings
+ip-26-0-161-178:2815963:2816400 [0] NCCL INFO Connected all rings
+ip-26-0-161-138:96566:97010 [3] NCCL INFO Channel 03/0 : 35[3] -> 36[4] via P2P/IPC
+ip-26-0-168-52:2722047:2722468 [2] NCCL INFO Channel 14/0 : 122[2] -> 123[3] via P2P/IPC
+ip-26-0-160-242:1364904:1365335 [2] NCCL INFO Channel 09/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-165-164:1227468:1227883 [5] NCCL INFO Channel 11/0 : 85[5] -> 86[6] via P2P/IPC
+ip-26-0-161-142:62885:63304 [4] NCCL INFO Channel 10/0 : 44[4] -> 45[5] via P2P/IPC
+ip-26-0-161-178:2815963:2816400 [0] NCCL INFO Channel 02/0 : 56[0] -> 57[1] via P2P/IPC
+ip-26-0-160-242:1364908:1365332 [6] NCCL INFO Channel 13/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-165-202:153563:153986 [1] NCCL INFO Channel 15/0 : 89[1] -> 90[2] via P2P/IPC
+ip-26-0-160-225:124164:124585 [4] NCCL INFO Channel 04/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-242:1364903:1365334 [1] NCCL INFO Channel 03/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-165-202:153564:153984 [2] NCCL INFO Channel 10/0 : 90[2] -> 91[3] via P2P/IPC
+ip-26-0-161-78:114515:114936 [3] NCCL INFO Channel 10/0 : 67[3] -> 68[4] via P2P/IPC
+ip-26-0-161-78:114518:114937 [6] NCCL INFO Channel 01/0 : 70[6] -> 71[7] via P2P/IPC
+ip-26-0-161-138:96565:97004 [2] NCCL INFO Channel 09/0 : 34[2] -> 35[3] via P2P/IPC
+ip-26-0-168-34:272151:272571 [1] NCCL INFO Channel 07/0 : 113[1] -> 114[2] via P2P/IPC
+ip-26-0-161-78:114516:114934 [4] NCCL INFO Channel 10/0 : 68[4] -> 69[5] via P2P/IPC
+ip-26-0-161-142:62887:63305 [6] NCCL INFO Channel 14/0 : 46[6] -> 47[7] via P2P/IPC
+ip-26-0-165-131:1630206:1630647 [1] NCCL INFO Channel 06/0 : 73[1] -> 74[2] via P2P/IPC
+ip-26-0-165-164:1227467:1227886 [4] NCCL INFO Channel 03/0 : 84[4] -> 85[5] via P2P/IPC
+ip-26-0-161-142:62886:63303 [5] NCCL INFO Channel 05/0 : 45[5] -> 46[6] via P2P/IPC
+ip-26-0-165-164:1227469:1227882 [6] NCCL INFO Channel 10/0 : 86[6] -> 87[7] via P2P/IPC
+ip-26-0-165-164:1227463:1227880 [0] NCCL INFO Channel 15/0 : 80[0] -> 87[7] via P2P/IPC
+ip-26-0-160-225:124162:124587 [2] NCCL INFO Channel 04/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-242:1364904:1365335 [2] NCCL INFO Channel 10/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-165-164:1227468:1227883 [5] NCCL INFO Channel 12/0 : 85[5] -> 86[6] via P2P/IPC
+ip-26-0-161-138:96563:97003 [0] NCCL INFO Channel 07/0 : 32[0] -> 39[7] via P2P/IPC
+ip-26-0-168-52:2722050:2722471 [5] NCCL INFO Channel 01/0 : 125[5] -> 126[6] via P2P/IPC
+ip-26-0-165-202:153565:153985 [3] NCCL INFO Channel 09/0 : 91[3] -> 92[4] via P2P/IPC
+ip-26-0-168-52:2722049:2722467 [4] NCCL INFO Channel 04/0 : 124[4] -> 125[5] via P2P/IPC
+ip-26-0-168-52:2722051:2722465 [6] NCCL INFO Channel 10/0 : 126[6] -> 127[7] via P2P/IPC
+ip-26-0-160-242:1364908:1365332 [6] NCCL INFO Channel 14/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-160-225:124164:124585 [4] NCCL INFO Channel 06/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-168-52:2722046:2722466 [1] NCCL INFO Connected all rings
+ip-26-0-160-242:1364903:1365334 [1] NCCL INFO Channel 04/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-161-142:62885:63304 [4] NCCL INFO Channel 11/0 : 44[4] -> 45[5] via P2P/IPC
+ip-26-0-168-52:2722047:2722468 [2] NCCL INFO Channel 15/0 : 122[2] -> 123[3] via P2P/IPC
+ip-26-0-161-78:114515:114936 [3] NCCL INFO Channel 11/0 : 67[3] -> 68[4] via P2P/IPC
+ip-26-0-161-138:96566:97010 [3] NCCL INFO Channel 05/0 : 35[3] -> 36[4] via P2P/IPC
+ip-26-0-161-78:114518:114937 [6] NCCL INFO Channel 02/0 : 70[6] -> 71[7] via P2P/IPC
+ip-26-0-161-178:2815963:2816400 [0] NCCL INFO Channel 03/0 : 56[0] -> 57[1] via P2P/IPC
+ip-26-0-165-202:153564:153984 [2] NCCL INFO Channel 12/0 : 90[2] -> 91[3] via P2P/IPC
+ip-26-0-165-164:1227467:1227886 [4] NCCL INFO Channel 04/0 : 84[4] -> 85[5] via P2P/IPC
+ip-26-0-165-131:1630207:1630648 [2] NCCL INFO Channel 01/0 : 74[2] -> 75[3] via P2P/IPC
+ip-26-0-161-78:114516:114934 [4] NCCL INFO Channel 11/0 : 68[4] -> 69[5] via P2P/IPC
+ip-26-0-160-242:1364904:1365335 [2] NCCL INFO Channel 12/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-165-164:1227469:1227882 [6] NCCL INFO Channel 11/0 : 86[6] -> 87[7] via P2P/IPC
+ip-26-0-165-202:153562:153983 [0] NCCL INFO Channel 05/0 : 88[0] -> 95[7] via P2P/IPC
+ip-26-0-165-164:1227468:1227883 [5] NCCL INFO Channel 13/0 : 85[5] -> 86[6] via P2P/IPC
+ip-26-0-161-142:62886:63303 [5] NCCL INFO Channel 07/0 : 45[5] -> 46[6] via P2P/IPC
+ip-26-0-165-131:1630206:1630647 [1] NCCL INFO Channel 07/0 : 73[1] -> 74[2] via P2P/IPC
+ip-26-0-160-225:124162:124587 [2] NCCL INFO Channel 05/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-242:1364903:1365334 [1] NCCL INFO Channel 05/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-161-138:96565:97004 [2] NCCL INFO Channel 10/0 : 34[2] -> 35[3] via P2P/IPC
+ip-26-0-160-225:124164:124585 [4] NCCL INFO Channel 07/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-123:32710:33125 [6] NCCL INFO Connected all rings
+ip-26-0-160-242:1364904:1365335 [2] NCCL INFO Channel 13/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-161-78:114517:114935 [5] NCCL INFO Channel 09/0 : 69[5] -> 70[6] via P2P/IPC
+ip-26-0-161-142:62885:63304 [4] NCCL INFO Channel 12/0 : 44[4] -> 45[5] via P2P/IPC
+ip-26-0-161-138:96563:97003 [0] NCCL INFO Channel 13/0 : 32[0] -> 39[7] via P2P/IPC
+ip-26-0-165-202:153565:153985 [3] NCCL INFO Channel 10/0 : 91[3] -> 92[4] via P2P/IPC
+ip-26-0-165-164:1227467:1227886 [4] NCCL INFO Channel 06/0 : 84[4] -> 85[5] via P2P/IPC
+ip-26-0-165-164:1227469:1227882 [6] NCCL INFO Channel 12/0 : 86[6] -> 87[7] via P2P/IPC
+ip-26-0-168-52:2722049:2722467 [4] NCCL INFO Channel 06/0 : 124[4] -> 125[5] via P2P/IPC
+ip-26-0-168-52:2722050:2722471 [5] NCCL INFO Channel 02/0 : 125[5] -> 126[6] via P2P/IPC
+ip-26-0-165-164:1227468:1227883 [5] NCCL INFO Channel 15/0 : 85[5] -> 86[6] via P2P/IPC
+ip-26-0-161-78:114515:114936 [3] NCCL INFO Channel 13/0 : 67[3] -> 68[4] via P2P/IPC
+ip-26-0-161-78:114518:114937 [6] NCCL INFO Channel 03/0 : 70[6] -> 71[7] via P2P/IPC
+ip-26-0-168-52:2722051:2722465 [6] NCCL INFO Channel 11/0 : 126[6] -> 127[7] via P2P/IPC
+ip-26-0-161-138:96566:97010 [3] NCCL INFO Channel 06/0 : 35[3] -> 36[4] via P2P/IPC
+ip-26-0-160-242:1364903:1365334 [1] NCCL INFO Channel 06/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-161-178:2815963:2816400 [0] NCCL INFO Channel 04/0 : 56[0] -> 57[1] via P2P/IPC
+ip-26-0-160-242:1364904:1365335 [2] NCCL INFO Channel 14/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-161-78:114513:114933 [1] NCCL INFO Channel 01/0 : 65[1] -> 66[2] via P2P/IPC
+ip-26-0-160-225:124162:124587 [2] NCCL INFO Channel 06/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-78:114516:114934 [4] NCCL INFO Channel 12/0 : 68[4] -> 69[5] via P2P/IPC
+ip-26-0-165-131:1630207:1630648 [2] NCCL INFO Channel 02/0 : 74[2] -> 75[3] via P2P/IPC
+ip-26-0-161-142:62886:63303 [5] NCCL INFO Channel 09/0 : 45[5] -> 46[6] via P2P/IPC
+ip-26-0-160-225:124164:124585 [4] NCCL INFO Channel 09/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-202:153564:153984 [2] NCCL INFO Channel 13/0 : 90[2] -> 91[3] via P2P/IPC
+ip-26-0-168-34:272154:272569 [4] NCCL INFO Connected all rings
+ip-26-0-165-131:1630206:1630647 [1] NCCL INFO Channel 09/0 : 73[1] -> 74[2] via P2P/IPC
+ip-26-0-165-202:153562:153983 [0] NCCL INFO Channel 06/0 : 88[0] -> 95[7] via P2P/IPC
+ip-26-0-161-103:96892:97308 [4] NCCL INFO Connected all rings
+ip-26-0-165-164:1227467:1227886 [4] NCCL INFO Channel 07/0 : 84[4] -> 85[5] via P2P/IPC
+ip-26-0-165-164:1227469:1227882 [6] NCCL INFO Channel 13/0 : 86[6] -> 87[7] via P2P/IPC
+ip-26-0-160-242:1364903:1365334 [1] NCCL INFO Channel 07/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-160-242:1364904:1365335 [2] NCCL INFO Channel 15/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-161-138:96565:97004 [2] NCCL INFO Channel 12/0 : 34[2] -> 35[3] via P2P/IPC
+ip-26-0-168-30:235655:236069 [7] NCCL INFO Channel 15/0 : 103[7] -> 111[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62885:63304 [4] NCCL INFO Channel 14/0 : 44[4] -> 45[5] via P2P/IPC
+ip-26-0-161-78:114517:114935 [5] NCCL INFO Channel 10/0 : 69[5] -> 70[6] via P2P/IPC
+ip-26-0-160-225:124163:124588 [3] NCCL INFO Channel 01/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-168-30:235648:236064 [0] NCCL INFO Channel 08/0 : 96[0] -> 104[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235655:236069 [7] NCCL INFO Channel 07/0 : 111[7] -> 119[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272151:272571 [1] NCCL INFO Channel 09/0 : 113[1] -> 114[2] via P2P/IPC
+ip-26-0-168-52:2722049:2722467 [4] NCCL INFO Channel 07/0 : 124[4] -> 125[5] via P2P/IPC
+ip-26-0-168-52:2722050:2722471 [5] NCCL INFO Channel 03/0 : 125[5] -> 126[6] via P2P/IPC
+ip-26-0-168-30:235648:236064 [0] NCCL INFO Channel 00/0 : 104[0] -> 112[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96563:97003 [0] NCCL INFO Channel 14/0 : 32[0] -> 39[7] via P2P/IPC
+ip-26-0-165-202:153565:153985 [3] NCCL INFO Channel 11/0 : 91[3] -> 92[4] via P2P/IPC
+ip-26-0-168-52:2722051:2722465 [6] NCCL INFO Channel 12/0 : 126[6] -> 127[7] via P2P/IPC
+ip-26-0-168-30:235655:236069 [7] NCCL INFO Channel 15/0 : 111[7] -> 119[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235648:236064 [0] NCCL INFO Channel 08/0 : 104[0] -> 112[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153566:153980 [4] NCCL INFO Channel 01/0 : 92[4] -> 93[5] via P2P/IPC
+ip-26-0-165-164:1227467:1227886 [4] NCCL INFO Channel 09/0 : 84[4] -> 85[5] via P2P/IPC
+ip-26-0-161-78:114518:114937 [6] NCCL INFO Channel 04/0 : 70[6] -> 71[7] via P2P/IPC
+ip-26-0-160-225:124162:124587 [2] NCCL INFO Channel 07/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-164:1227469:1227882 [6] NCCL INFO Channel 14/0 : 86[6] -> 87[7] via P2P/IPC
+ip-26-0-161-78:114515:114936 [3] NCCL INFO Channel 14/0 : 67[3] -> 68[4] via P2P/IPC
+ip-26-0-161-153:2636089:2636508 [5] NCCL INFO Connected all rings
+ip-26-0-161-78:114513:114933 [1] NCCL INFO Channel 03/0 : 65[1] -> 66[2] via P2P/IPC
+ip-26-0-161-138:96566:97010 [3] NCCL INFO Channel 07/0 : 35[3] -> 36[4] via P2P/IPC
+ip-26-0-160-242:1364903:1365334 [1] NCCL INFO Channel 09/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-161-142:62886:63303 [5] NCCL INFO Channel 10/0 : 45[5] -> 46[6] via P2P/IPC
+ip-26-0-160-225:124164:124585 [4] NCCL INFO Channel 10/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-78:114516:114934 [4] NCCL INFO Channel 14/0 : 68[4] -> 69[5] via P2P/IPC
+ip-26-0-165-59:1557273:1557685 [7] NCCL INFO Channel 07/0 : 87[7] -> 103[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815963:2816400 [0] NCCL INFO Channel 05/0 : 56[0] -> 57[1] via P2P/IPC
+ip-26-0-165-131:1630207:1630648 [2] NCCL INFO Channel 04/0 : 74[2] -> 75[3] via P2P/IPC
+ip-26-0-165-59:1557266:1557689 [0] NCCL INFO Channel 00/0 : 80[0] -> 96[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630206:1630647 [1] NCCL INFO Channel 11/0 : 73[1] -> 74[2] via P2P/IPC
+ip-26-0-165-59:1557273:1557685 [7] NCCL INFO Channel 07/0 : 103[7] -> 119[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153564:153984 [2] NCCL INFO Channel 14/0 : 90[2] -> 91[3] via P2P/IPC
+ip-26-0-165-164:1227467:1227886 [4] NCCL INFO Channel 10/0 : 84[4] -> 85[5] via P2P/IPC
+ip-26-0-165-59:1557266:1557689 [0] NCCL INFO Channel 00/0 : 96[0] -> 112[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96888:97306 [0] NCCL INFO Channel 15/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-160-242:1364903:1365334 [1] NCCL INFO Channel 11/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-160-225:124163:124588 [3] NCCL INFO Channel 02/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-202:153562:153983 [0] NCCL INFO Channel 07/0 : 88[0] -> 95[7] via P2P/IPC
+ip-26-0-161-142:62885:63304 [4] NCCL INFO Channel 15/0 : 44[4] -> 45[5] via P2P/IPC
+ip-26-0-161-78:114517:114935 [5] NCCL INFO Channel 11/0 : 69[5] -> 70[6] via P2P/IPC
+ip-26-0-161-138:96565:97004 [2] NCCL INFO Channel 13/0 : 34[2] -> 35[3] via P2P/IPC
+ip-26-0-165-164:1227467:1227886 [4] NCCL INFO Channel 11/0 : 84[4] -> 85[5] via P2P/IPC
+ip-26-0-161-138:96569:97006 [6] NCCL INFO Channel 01/0 : 38[6] -> 39[7] via P2P/IPC
+ip-26-0-161-138:96563:97003 [0] NCCL INFO Channel 15/0 : 32[0] -> 39[7] via P2P/IPC
+ip-26-0-160-225:124162:124587 [2] NCCL INFO Channel 09/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-168-52:2722048:2722469 [3] NCCL INFO Channel 03/0 : 123[3] -> 124[4] via P2P/IPC
+ip-26-0-168-52:2722049:2722467 [4] NCCL INFO Channel 09/0 : 124[4] -> 125[5] via P2P/IPC
+ip-26-0-165-164:1227469:1227882 [6] NCCL INFO Channel 06/0 : 78[6] -> 86[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-52:2722050:2722471 [5] NCCL INFO Channel 04/0 : 125[5] -> 126[6] via P2P/IPC
+ip-26-0-168-52:2722051:2722465 [6] NCCL INFO Channel 13/0 : 126[6] -> 127[7] via P2P/IPC
+ip-26-0-160-242:1364903:1365334 [1] NCCL INFO Channel 12/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-165-164:1227469:1227882 [6] NCCL INFO Channel 14/0 : 78[6] -> 86[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272151:272571 [1] NCCL INFO Channel 11/0 : 113[1] -> 114[2] via P2P/IPC
+ip-26-0-161-78:114518:114937 [6] NCCL INFO Channel 05/0 : 70[6] -> 71[7] via P2P/IPC
+ip-26-0-165-202:153565:153985 [3] NCCL INFO Channel 13/0 : 91[3] -> 92[4] via P2P/IPC
+ip-26-0-160-225:124164:124585 [4] NCCL INFO Channel 11/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-142:62886:63303 [5] NCCL INFO Channel 11/0 : 45[5] -> 46[6] via P2P/IPC
+ip-26-0-165-164:1227469:1227882 [6] NCCL INFO Channel 06/0 : 86[6] -> 94[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114515:114936 [3] NCCL INFO Channel 15/0 : 67[3] -> 68[4] via P2P/IPC
+ip-26-0-165-202:153566:153980 [4] NCCL INFO Channel 02/0 : 92[4] -> 93[5] via P2P/IPC
+ip-26-0-161-78:114513:114933 [1] NCCL INFO Channel 04/0 : 65[1] -> 66[2] via P2P/IPC
+ip-26-0-165-164:1227467:1227886 [4] NCCL INFO Channel 12/0 : 84[4] -> 85[5] via P2P/IPC
+ip-26-0-160-242:1364903:1365334 [1] NCCL INFO Channel 13/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-161-78:114516:114934 [4] NCCL INFO Channel 15/0 : 68[4] -> 69[5] via P2P/IPC
+ip-26-0-161-178:2815963:2816400 [0] NCCL INFO Channel 06/0 : 56[0] -> 57[1] via P2P/IPC
+ip-26-0-160-225:124163:124588 [3] NCCL INFO Channel 03/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-138:96566:97010 [3] NCCL INFO Channel 09/0 : 35[3] -> 36[4] via P2P/IPC
+ip-26-0-160-242:1364903:1365334 [1] NCCL INFO Channel 14/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-165-164:1227467:1227886 [4] NCCL INFO Channel 14/0 : 84[4] -> 85[5] via P2P/IPC
+ip-26-0-165-202:153564:153984 [2] NCCL INFO Channel 15/0 : 90[2] -> 91[3] via P2P/IPC
+ip-26-0-165-131:1630207:1630648 [2] NCCL INFO Channel 05/0 : 74[2] -> 75[3] via P2P/IPC
+ip-26-0-161-78:114517:114935 [5] NCCL INFO Channel 12/0 : 69[5] -> 70[6] via P2P/IPC
+ip-26-0-160-225:124162:124587 [2] NCCL INFO Channel 10/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-168-30:235651:236068 [3] NCCL INFO Channel 01/0 : 107[3] -> 108[4] via P2P/IPC
+ip-26-0-165-131:1630206:1630647 [1] NCCL INFO Channel 12/0 : 73[1] -> 74[2] via P2P/IPC
+ip-26-0-168-52:2722049:2722467 [4] NCCL INFO Channel 10/0 : 124[4] -> 125[5] via P2P/IPC
+ip-26-0-160-225:124164:124585 [4] NCCL INFO Channel 12/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-202:153562:153983 [0] NCCL INFO Channel 13/0 : 88[0] -> 95[7] via P2P/IPC
+ip-26-0-168-34:272151:272571 [1] NCCL INFO Channel 12/0 : 113[1] -> 114[2] via P2P/IPC
+ip-26-0-168-52:2722048:2722469 [3] NCCL INFO Channel 05/0 : 123[3] -> 124[4] via P2P/IPC
+ip-26-0-168-52:2722051:2722465 [6] NCCL INFO Channel 14/0 : 126[6] -> 127[7] via P2P/IPC
+ip-26-0-168-52:2722050:2722471 [5] NCCL INFO Channel 05/0 : 125[5] -> 126[6] via P2P/IPC
+ip-26-0-161-142:62886:63303 [5] NCCL INFO Channel 12/0 : 45[5] -> 46[6] via P2P/IPC
+ip-26-0-161-138:96565:97004 [2] NCCL INFO Channel 14/0 : 34[2] -> 35[3] via P2P/IPC
+ip-26-0-161-138:96569:97006 [6] NCCL INFO Channel 02/0 : 38[6] -> 39[7] via P2P/IPC
+ip-26-0-165-202:153567:153979 [5] NCCL INFO Channel 01/0 : 93[5] -> 94[6] via P2P/IPC
+ip-26-0-165-164:1227467:1227886 [4] NCCL INFO Channel 15/0 : 84[4] -> 85[5] via P2P/IPC
+ip-26-0-161-78:114518:114937 [6] NCCL INFO Channel 06/0 : 70[6] -> 71[7] via P2P/IPC
+ip-26-0-160-225:124163:124588 [3] NCCL INFO Channel 05/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-78:114513:114933 [1] NCCL INFO Channel 05/0 : 65[1] -> 66[2] via P2P/IPC
+ip-26-0-168-30:235651:236068 [3] NCCL INFO Channel 02/0 : 107[3] -> 108[4] via P2P/IPC
+ip-26-0-160-242:1364903:1365334 [1] NCCL INFO Channel 15/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-165-202:153565:153985 [3] NCCL INFO Channel 14/0 : 91[3] -> 92[4] via P2P/IPC
+ip-26-0-168-30:235652:236066 [4] NCCL INFO Channel 01/0 : 108[4] -> 109[5] via P2P/IPC
+ip-26-0-165-202:153566:153980 [4] NCCL INFO Channel 03/0 : 92[4] -> 93[5] via P2P/IPC
+ip-26-0-161-138:96566:97010 [3] NCCL INFO Channel 10/0 : 35[3] -> 36[4] via P2P/IPC
+ip-26-0-161-123:32710:33125 [6] NCCL INFO Channel 01/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-161-78:114517:114935 [5] NCCL INFO Channel 13/0 : 69[5] -> 70[6] via P2P/IPC
+ip-26-0-161-78:114515:114936 [3] NCCL INFO Channel 11/0 : 67[3] -> 75[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62886:63303 [5] NCCL INFO Channel 13/0 : 45[5] -> 46[6] via P2P/IPC
+ip-26-0-161-153:2636087:2636506 [3] NCCL INFO Connected all rings
+ip-26-0-161-78:114518:114937 [6] NCCL INFO Channel 09/0 : 70[6] -> 71[7] via P2P/IPC
+ip-26-0-168-30:235651:236068 [3] NCCL INFO Channel 03/0 : 107[3] -> 108[4] via P2P/IPC
+ip-26-0-161-78:114513:114933 [1] NCCL INFO Channel 06/0 : 65[1] -> 66[2] via P2P/IPC
+ip-26-0-168-30:235652:236066 [4] NCCL INFO Channel 02/0 : 108[4] -> 109[5] via P2P/IPC
+ip-26-0-161-123:32710:33125 [6] NCCL INFO Channel 02/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-168-52:2722049:2722467 [4] NCCL INFO Channel 11/0 : 124[4] -> 125[5] via P2P/IPC
+ip-26-0-165-164:1227466:1227887 [3] NCCL INFO Channel 03/0 : 75[3] -> 83[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96565:97004 [2] NCCL INFO Channel 15/0 : 34[2] -> 35[3] via P2P/IPC
+ip-26-0-161-138:96569:97006 [6] NCCL INFO Channel 03/0 : 38[6] -> 39[7] via P2P/IPC
+ip-26-0-168-52:2722048:2722469 [3] NCCL INFO Channel 06/0 : 123[3] -> 124[4] via P2P/IPC
+ip-26-0-168-52:2722050:2722471 [5] NCCL INFO Channel 07/0 : 125[5] -> 126[6] via P2P/IPC
+ip-26-0-165-164:1227468:1227883 [5] NCCL INFO Channel 05/0 : 77[5] -> 85[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630207:1630648 [2] NCCL INFO Channel 06/0 : 74[2] -> 75[3] via P2P/IPC
+ip-26-0-165-164:1227466:1227887 [3] NCCL INFO Channel 11/0 : 75[3] -> 83[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153562:153983 [0] NCCL INFO Channel 14/0 : 88[0] -> 95[7] via P2P/IPC
+ip-26-0-165-164:1227467:1227886 [4] NCCL INFO Channel 04/0 : 76[4] -> 84[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815963:2816400 [0] NCCL INFO Channel 07/0 : 56[0] -> 57[1] via P2P/IPC
+ip-26-0-165-164:1227468:1227883 [5] NCCL INFO Channel 13/0 : 77[5] -> 85[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32710:33125 [6] NCCL INFO Channel 03/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-165-164:1227466:1227887 [3] NCCL INFO Channel 03/0 : 83[3] -> 91[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153567:153979 [5] NCCL INFO Channel 02/0 : 93[5] -> 94[6] via P2P/IPC
+ip-26-0-168-30:235651:236068 [3] NCCL INFO Channel 05/0 : 107[3] -> 108[4] via P2P/IPC
+ip-26-0-165-164:1227467:1227886 [4] NCCL INFO Channel 12/0 : 76[4] -> 84[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227468:1227883 [5] NCCL INFO Channel 05/0 : 85[5] -> 93[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630206:1630647 [1] NCCL INFO Channel 13/0 : 73[1] -> 74[2] via P2P/IPC
+ip-26-0-168-34:272151:272571 [1] NCCL INFO Channel 13/0 : 113[1] -> 114[2] via P2P/IPC
+ip-26-0-161-123:32707:33124 [3] NCCL INFO Connected all rings
+ip-26-0-161-138:96566:97010 [3] NCCL INFO Channel 11/0 : 35[3] -> 36[4] via P2P/IPC
+ip-26-0-165-202:153563:153986 [1] NCCL INFO Channel 01/0 : 81[1] -> 89[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114517:114935 [5] NCCL INFO Channel 15/0 : 69[5] -> 70[6] via P2P/IPC
+ip-26-0-161-78:114518:114937 [6] NCCL INFO Channel 10/0 : 70[6] -> 71[7] via P2P/IPC
+ip-26-0-161-142:62884:63299 [3] NCCL INFO Channel 01/0 : 43[3] -> 44[4] via P2P/IPC
+ip-26-0-168-30:235652:236066 [4] NCCL INFO Channel 03/0 : 108[4] -> 109[5] via P2P/IPC
+ip-26-0-165-164:1227467:1227886 [4] NCCL INFO Channel 04/0 : 84[4] -> 92[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153565:153985 [3] NCCL INFO Channel 15/0 : 91[3] -> 92[4] via P2P/IPC
+ip-26-0-161-78:114513:114933 [1] NCCL INFO Channel 07/0 : 65[1] -> 66[2] via P2P/IPC
+ip-26-0-161-142:62886:63303 [5] NCCL INFO Channel 15/0 : 45[5] -> 46[6] via P2P/IPC
+ip-26-0-165-164:1227463:1227880 [0] NCCL INFO Channel 00/0 : 72[0] -> 80[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227470:1227885 [7] NCCL INFO Channel 07/0 : 79[7] -> 87[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153566:153980 [4] NCCL INFO Channel 04/0 : 92[4] -> 93[5] via P2P/IPC
+ip-26-0-161-123:32710:33125 [6] NCCL INFO Channel 04/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-165-202:153563:153986 [1] NCCL INFO Channel 09/0 : 73[1] -> 89[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227470:1227885 [7] NCCL INFO Channel 15/0 : 79[7] -> 87[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227463:1227880 [0] NCCL INFO Channel 08/0 : 72[0] -> 80[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96569:97006 [6] NCCL INFO Channel 04/0 : 38[6] -> 39[7] via P2P/IPC
+ip-26-0-165-202:153563:153986 [1] NCCL INFO Channel 09/0 : 89[1] -> 105[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227470:1227885 [7] NCCL INFO Channel 07/0 : 87[7] -> 95[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235651:236068 [3] NCCL INFO Channel 06/0 : 107[3] -> 108[4] via P2P/IPC
+ip-26-0-165-164:1227463:1227880 [0] NCCL INFO Channel 00/0 : 80[0] -> 88[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630208:1630650 [3] NCCL INFO Channel 01/0 : 75[3] -> 76[4] via P2P/IPC
+ip-26-0-168-52:2722049:2722467 [4] NCCL INFO Channel 12/0 : 124[4] -> 125[5] via P2P/IPC
+ip-26-0-168-30:235652:236066 [4] NCCL INFO Channel 04/0 : 108[4] -> 109[5] via P2P/IPC
+ip-26-0-160-225:124162:124587 [2] NCCL INFO Channel 12/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-138:96566:97010 [3] NCCL INFO Channel 13/0 : 35[3] -> 36[4] via P2P/IPC
+ip-26-0-168-52:2722048:2722469 [3] NCCL INFO Channel 07/0 : 123[3] -> 124[4] via P2P/IPC
+ip-26-0-168-52:2722050:2722471 [5] NCCL INFO Channel 09/0 : 125[5] -> 126[6] via P2P/IPC
+ip-26-0-160-225:124166:124583 [6] NCCL INFO Channel 09/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-123:32710:33125 [6] NCCL INFO Channel 05/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-161-138:96564:97009 [1] NCCL INFO Channel 09/0 : 33[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114518:114937 [6] NCCL INFO Channel 11/0 : 70[6] -> 71[7] via P2P/IPC
+ip-26-0-161-142:62884:63299 [3] NCCL INFO Channel 02/0 : 43[3] -> 44[4] via P2P/IPC
+ip-26-0-160-242:1364906:1365331 [4] NCCL INFO Connected all rings
+ip-26-0-160-225:124164:124585 [4] NCCL INFO Channel 14/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-78:114513:114933 [1] NCCL INFO Channel 09/0 : 65[1] -> 66[2] via P2P/IPC
+ip-26-0-165-202:153562:153983 [0] NCCL INFO Channel 15/0 : 88[0] -> 95[7] via P2P/IPC
+ip-26-0-165-202:153567:153979 [5] NCCL INFO Channel 03/0 : 93[5] -> 94[6] via P2P/IPC
+ip-26-0-168-30:235651:236068 [3] NCCL INFO Channel 07/0 : 107[3] -> 108[4] via P2P/IPC
+ip-26-0-161-78:114516:114934 [4] NCCL INFO Channel 12/0 : 68[4] -> 76[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62881:63298 [0] NCCL INFO Channel 08/0 : 32[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62888:63302 [7] NCCL INFO Channel 15/0 : 39[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235652:236066 [4] NCCL INFO Channel 06/0 : 108[4] -> 109[5] via P2P/IPC
+ip-26-0-160-242:1364902:1365333 [0] NCCL INFO Channel 05/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-161-138:96569:97006 [6] NCCL INFO Channel 05/0 : 38[6] -> 39[7] via P2P/IPC
+ip-26-0-161-123:32710:33125 [6] NCCL INFO Channel 06/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-165-202:153566:153980 [4] NCCL INFO Channel 06/0 : 92[4] -> 93[5] via P2P/IPC
+ip-26-0-161-142:62888:63302 [7] NCCL INFO Channel 07/0 : 47[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62881:63298 [0] NCCL INFO Channel 00/0 : 40[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62884:63299 [3] NCCL INFO Channel 03/0 : 43[3] -> 44[4] via P2P/IPC
+ip-26-0-161-78:114518:114937 [6] NCCL INFO Channel 12/0 : 70[6] -> 71[7] via P2P/IPC
+ip-26-0-165-131:1630207:1630648 [2] NCCL INFO Channel 07/0 : 74[2] -> 75[3] via P2P/IPC
+ip-26-0-161-138:96566:97010 [3] NCCL INFO Channel 14/0 : 35[3] -> 36[4] via P2P/IPC
+ip-26-0-160-242:1364904:1365335 [2] NCCL INFO Channel 10/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153564:153984 [2] NCCL INFO Channel 02/0 : 82[2] -> 90[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62888:63302 [7] NCCL INFO Channel 15/0 : 47[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114513:114933 [1] NCCL INFO Channel 11/0 : 65[1] -> 66[2] via P2P/IPC
+ip-26-0-161-103:96893:97310 [5] NCCL INFO Channel 01/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-168-52:2722049:2722467 [4] NCCL INFO Channel 14/0 : 124[4] -> 125[5] via P2P/IPC
+ip-26-0-160-242:1364903:1365334 [1] NCCL INFO Channel 09/0 : 1[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62881:63298 [0] NCCL INFO Channel 08/0 : 40[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364904:1365335 [2] NCCL INFO Channel 02/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630206:1630647 [1] NCCL INFO Channel 14/0 : 73[1] -> 74[2] via P2P/IPC
+ip-26-0-161-142:62887:63305 [6] NCCL INFO Channel 14/0 : 38[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96569:97006 [6] NCCL INFO Channel 06/0 : 38[6] -> 39[7] via P2P/IPC
+ip-26-0-160-225:124162:124587 [2] NCCL INFO Channel 13/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-225:124166:124583 [6] NCCL INFO Channel 10/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-242:1364903:1365334 [1] NCCL INFO Channel 01/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364904:1365335 [2] NCCL INFO Channel 10/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62886:63303 [5] NCCL INFO Channel 13/0 : 37[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153564:153984 [2] NCCL INFO Channel 10/0 : 74[2] -> 90[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-52:2722048:2722469 [3] NCCL INFO Channel 09/0 : 123[3] -> 124[4] via P2P/IPC
+ip-26-0-160-242:1364903:1365334 [1] NCCL INFO Channel 09/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62887:63305 [6] NCCL INFO Channel 06/0 : 46[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32710:33125 [6] NCCL INFO Channel 09/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-161-142:62886:63303 [5] NCCL INFO Channel 05/0 : 45[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-52:2722050:2722471 [5] NCCL INFO Channel 10/0 : 125[5] -> 126[6] via P2P/IPC
+ip-26-0-165-202:153564:153984 [2] NCCL INFO Channel 10/0 : 90[2] -> 106[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96566:97010 [3] NCCL INFO Channel 15/0 : 35[3] -> 36[4] via P2P/IPC
+ip-26-0-161-142:62887:63305 [6] NCCL INFO Channel 14/0 : 46[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114518:114937 [6] NCCL INFO Channel 13/0 : 70[6] -> 71[7] via P2P/IPC
+ip-26-0-161-142:62884:63299 [3] NCCL INFO Channel 05/0 : 43[3] -> 44[4] via P2P/IPC
+ip-26-0-160-242:1364902:1365333 [0] NCCL INFO Channel 06/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-165-202:153567:153979 [5] NCCL INFO Channel 04/0 : 93[5] -> 94[6] via P2P/IPC
+ip-26-0-161-142:62886:63303 [5] NCCL INFO Channel 13/0 : 45[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124164:124585 [4] NCCL INFO Channel 15/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-78:114513:114933 [1] NCCL INFO Channel 12/0 : 65[1] -> 66[2] via P2P/IPC
+ip-26-0-165-202:153566:153980 [4] NCCL INFO Channel 07/0 : 92[4] -> 93[5] via P2P/IPC
+ip-26-0-161-138:96569:97006 [6] NCCL INFO Channel 09/0 : 38[6] -> 39[7] via P2P/IPC
+ip-26-0-161-103:96893:97310 [5] NCCL INFO Channel 02/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-165-131:1630208:1630650 [3] NCCL INFO Channel 02/0 : 75[3] -> 76[4] via P2P/IPC
+ip-26-0-161-123:32710:33125 [6] NCCL INFO Channel 10/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-161-142:62884:63299 [3] NCCL INFO Channel 06/0 : 43[3] -> 44[4] via P2P/IPC
+ip-26-0-161-78:114518:114937 [6] NCCL INFO Channel 14/0 : 70[6] -> 71[7] via P2P/IPC
+ip-26-0-161-142:62883:63301 [2] NCCL INFO Channel 01/0 : 42[2] -> 43[3] via P2P/IPC
+ip-26-0-168-30:235651:236068 [3] NCCL INFO Channel 09/0 : 107[3] -> 108[4] via P2P/IPC
+ip-26-0-160-242:1364902:1365333 [0] NCCL INFO Channel 07/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-168-52:2722049:2722467 [4] NCCL INFO Channel 15/0 : 124[4] -> 125[5] via P2P/IPC
+ip-26-0-165-202:153567:153979 [5] NCCL INFO Channel 05/0 : 93[5] -> 94[6] via P2P/IPC
+ip-26-0-160-225:124162:124587 [2] NCCL INFO Channel 14/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-78:114513:114933 [1] NCCL INFO Channel 13/0 : 65[1] -> 66[2] via P2P/IPC
+ip-26-0-168-30:235652:236066 [4] NCCL INFO Channel 07/0 : 108[4] -> 109[5] via P2P/IPC
+ip-26-0-160-225:124166:124583 [6] NCCL INFO Channel 11/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-178:2815963:2816400 [0] NCCL INFO Channel 10/0 : 56[0] -> 57[1] via P2P/IPC
+ip-26-0-168-34:272156:272570 [6] NCCL INFO Connected all rings
+ip-26-0-165-202:153566:153980 [4] NCCL INFO Channel 09/0 : 92[4] -> 93[5] via P2P/IPC
+ip-26-0-161-138:96569:97006 [6] NCCL INFO Channel 10/0 : 38[6] -> 39[7] via P2P/IPC
+ip-26-0-161-123:32710:33125 [6] NCCL INFO Channel 11/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-168-52:2722048:2722469 [3] NCCL INFO Channel 10/0 : 123[3] -> 124[4] via P2P/IPC
+ip-26-0-161-142:62884:63299 [3] NCCL INFO Channel 07/0 : 43[3] -> 44[4] via P2P/IPC
+ip-26-0-168-52:2722050:2722471 [5] NCCL INFO Channel 11/0 : 125[5] -> 126[6] via P2P/IPC
+ip-26-0-161-103:96894:97312 [6] NCCL INFO Channel 01/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-161-138:96565:97004 [2] NCCL INFO Channel 10/0 : 34[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153567:153979 [5] NCCL INFO Channel 07/0 : 93[5] -> 94[6] via P2P/IPC
+ip-26-0-161-142:62883:63301 [2] NCCL INFO Channel 02/0 : 42[2] -> 43[3] via P2P/IPC
+ip-26-0-161-78:114513:114933 [1] NCCL INFO Channel 14/0 : 65[1] -> 66[2] via P2P/IPC
+ip-26-0-168-30:235651:236068 [3] NCCL INFO Channel 10/0 : 107[3] -> 108[4] via P2P/IPC
+ip-26-0-161-103:96893:97310 [5] NCCL INFO Channel 03/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-160-242:1364902:1365333 [0] NCCL INFO Channel 13/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-161-138:96569:97006 [6] NCCL INFO Channel 11/0 : 38[6] -> 39[7] via P2P/IPC
+ip-26-0-168-34:272151:272571 [1] NCCL INFO Channel 14/0 : 113[1] -> 114[2] via P2P/IPC
+ip-26-0-165-131:1630207:1630648 [2] NCCL INFO Channel 09/0 : 74[2] -> 75[3] via P2P/IPC
+ip-26-0-168-30:235652:236066 [4] NCCL INFO Channel 09/0 : 108[4] -> 109[5] via P2P/IPC
+ip-26-0-165-131:1630206:1630647 [1] NCCL INFO Channel 15/0 : 73[1] -> 74[2] via P2P/IPC
+ip-26-0-160-225:124162:124587 [2] NCCL INFO Channel 15/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-123:32710:33125 [6] NCCL INFO Channel 12/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-160-225:124166:124583 [6] NCCL INFO Channel 12/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-78:114513:114933 [1] NCCL INFO Channel 15/0 : 65[1] -> 66[2] via P2P/IPC
+ip-26-0-161-78:114517:114935 [5] NCCL INFO Channel 13/0 : 69[5] -> 77[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96569:97006 [6] NCCL INFO Channel 12/0 : 38[6] -> 39[7] via P2P/IPC
+ip-26-0-165-202:153566:153980 [4] NCCL INFO Channel 10/0 : 92[4] -> 93[5] via P2P/IPC
+ip-26-0-161-142:62884:63299 [3] NCCL INFO Channel 09/0 : 43[3] -> 44[4] via P2P/IPC
+ip-26-0-161-78:114518:114937 [6] NCCL INFO Channel 14/0 : 70[6] -> 78[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235651:236068 [3] NCCL INFO Channel 11/0 : 107[3] -> 108[4] via P2P/IPC
+ip-26-0-161-142:62883:63301 [2] NCCL INFO Channel 04/0 : 42[2] -> 43[3] via P2P/IPC
+ip-26-0-161-103:96890:97313 [2] NCCL INFO Channel 01/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-168-30:235652:236066 [4] NCCL INFO Channel 10/0 : 108[4] -> 109[5] via P2P/IPC
+ip-26-0-165-202:153567:153979 [5] NCCL INFO Channel 09/0 : 93[5] -> 94[6] via P2P/IPC
+ip-26-0-161-178:2815970:2816397 [7] NCCL INFO Connected all rings
+ip-26-0-160-242:1364902:1365333 [0] NCCL INFO Channel 14/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-165-131:1630208:1630650 [3] NCCL INFO Channel 03/0 : 75[3] -> 76[4] via P2P/IPC
+ip-26-0-168-52:2722048:2722469 [3] NCCL INFO Channel 11/0 : 123[3] -> 124[4] via P2P/IPC
+ip-26-0-161-178:2815963:2816400 [0] NCCL INFO Channel 11/0 : 56[0] -> 57[1] via P2P/IPC
+ip-26-0-161-138:96569:97006 [6] NCCL INFO Channel 13/0 : 38[6] -> 39[7] via P2P/IPC
+ip-26-0-161-123:32710:33125 [6] NCCL INFO Channel 13/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-168-52:2722050:2722471 [5] NCCL INFO Channel 12/0 : 125[5] -> 126[6] via P2P/IPC
+ip-26-0-161-103:96894:97312 [6] NCCL INFO Channel 02/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-161-142:62884:63299 [3] NCCL INFO Channel 10/0 : 43[3] -> 44[4] via P2P/IPC
+ip-26-0-165-202:153566:153980 [4] NCCL INFO Channel 11/0 : 92[4] -> 93[5] via P2P/IPC
+ip-26-0-161-103:96893:97310 [5] NCCL INFO Channel 04/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-160-225:124163:124588 [3] NCCL INFO Channel 06/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-142:62883:63301 [2] NCCL INFO Channel 05/0 : 42[2] -> 43[3] via P2P/IPC
+ip-26-0-168-30:235651:236068 [3] NCCL INFO Channel 13/0 : 107[3] -> 108[4] via P2P/IPC
+ip-26-0-165-202:153568:153982 [6] NCCL INFO Channel 01/0 : 94[6] -> 95[7] via P2P/IPC
+ip-26-0-161-138:96569:97006 [6] NCCL INFO Channel 14/0 : 38[6] -> 39[7] via P2P/IPC
+ip-26-0-168-30:235652:236066 [4] NCCL INFO Channel 11/0 : 108[4] -> 109[5] via P2P/IPC
+ip-26-0-168-34:272151:272571 [1] NCCL INFO Channel 15/0 : 113[1] -> 114[2] via P2P/IPC
+ip-26-0-165-202:153567:153979 [5] NCCL INFO Channel 10/0 : 93[5] -> 94[6] via P2P/IPC
+ip-26-0-161-153:2636091:2636512 [7] NCCL INFO Connected all rings
+ip-26-0-160-225:124166:124583 [6] NCCL INFO Channel 13/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-242:1364902:1365333 [0] NCCL INFO Channel 15/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-161-123:32710:33125 [6] NCCL INFO Channel 14/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-161-78:114513:114933 [1] NCCL INFO Channel 09/0 : 65[1] -> 73[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114514:114932 [2] NCCL INFO Channel 10/0 : 66[2] -> 74[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272152:272575 [2] NCCL INFO Channel 01/0 : 114[2] -> 115[3] via P2P/IPC
+ip-26-0-161-142:62884:63299 [3] NCCL INFO Channel 11/0 : 43[3] -> 44[4] via P2P/IPC
+ip-26-0-161-78:114512:114930 [0] NCCL INFO Channel 05/0 : 64[0] -> 71[7] via P2P/IPC
+ip-26-0-160-225:124161:124582 [1] NCCL INFO Channel 09/0 : 1[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62883:63301 [2] NCCL INFO Channel 06/0 : 42[2] -> 43[3] via P2P/IPC
+ip-26-0-168-30:235651:236068 [3] NCCL INFO Channel 14/0 : 107[3] -> 108[4] via P2P/IPC
+ip-26-0-168-52:2722048:2722469 [3] NCCL INFO Channel 13/0 : 123[3] -> 124[4] via P2P/IPC
+ip-26-0-161-103:96890:97313 [2] NCCL INFO Channel 02/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-165-131:1630207:1630648 [2] NCCL INFO Channel 10/0 : 74[2] -> 75[3] via P2P/IPC
+ip-26-0-168-30:235652:236066 [4] NCCL INFO Channel 12/0 : 108[4] -> 109[5] via P2P/IPC
+ip-26-0-165-202:153566:153980 [4] NCCL INFO Channel 12/0 : 92[4] -> 93[5] via P2P/IPC
+ip-26-0-160-225:124163:124588 [3] NCCL INFO Channel 07/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-168-52:2722050:2722471 [5] NCCL INFO Channel 13/0 : 125[5] -> 126[6] via P2P/IPC
+ip-26-0-165-202:153568:153982 [6] NCCL INFO Channel 02/0 : 94[6] -> 95[7] via P2P/IPC
+ip-26-0-168-34:272153:272573 [3] NCCL INFO Channel 01/0 : 115[3] -> 116[4] via P2P/IPC
+ip-26-0-165-202:153567:153979 [5] NCCL INFO Channel 11/0 : 93[5] -> 94[6] via P2P/IPC
+ip-26-0-160-225:124161:124582 [1] NCCL INFO Channel 01/0 : 65[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630205:1630646 [0] NCCL INFO Channel 05/0 : 72[0] -> 79[7] via P2P/IPC
+ip-26-0-160-225:124166:124583 [6] NCCL INFO Channel 14/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-103:96894:97312 [6] NCCL INFO Channel 03/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-161-142:62884:63299 [3] NCCL INFO Channel 13/0 : 43[3] -> 44[4] via P2P/IPC
+ip-26-0-160-225:124161:124582 [1] NCCL INFO Channel 01/0 : 1[1] -> 65[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124165:124584 [5] NCCL INFO Channel 01/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-142:62883:63301 [2] NCCL INFO Channel 07/0 : 42[2] -> 43[3] via P2P/IPC
+ip-26-0-168-30:235651:236068 [3] NCCL INFO Channel 15/0 : 107[3] -> 108[4] via P2P/IPC
+ip-26-0-161-103:96893:97310 [5] NCCL INFO Channel 05/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-165-131:1630209:1630651 [4] NCCL INFO Channel 01/0 : 76[4] -> 77[5] via P2P/IPC
+ip-26-0-168-30:235652:236066 [4] NCCL INFO Channel 14/0 : 108[4] -> 109[5] via P2P/IPC
+ip-26-0-161-178:2815963:2816400 [0] NCCL INFO Channel 12/0 : 56[0] -> 57[1] via P2P/IPC
+ip-26-0-161-123:32709:33129 [5] NCCL INFO Channel 05/0 : 21[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630208:1630650 [3] NCCL INFO Channel 05/0 : 75[3] -> 76[4] via P2P/IPC
+ip-26-0-161-78:114512:114930 [0] NCCL INFO Channel 06/0 : 64[0] -> 71[7] via P2P/IPC
+ip-26-0-161-123:32710:33125 [6] NCCL INFO Channel 06/0 : 22[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:124163:124588 [3] NCCL INFO Channel 09/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-202:153566:153980 [4] NCCL INFO Channel 14/0 : 92[4] -> 93[5] via P2P/IPC
+ip-26-0-168-52:2722048:2722469 [3] NCCL INFO Channel 14/0 : 123[3] -> 124[4] via P2P/IPC
+ip-26-0-165-131:1630212:1630652 [7] NCCL INFO Connected all rings
+ip-26-0-165-202:153568:153982 [6] NCCL INFO Channel 03/0 : 94[6] -> 95[7] via P2P/IPC
+ip-26-0-165-202:153567:153979 [5] NCCL INFO Channel 12/0 : 93[5] -> 94[6] via P2P/IPC
+ip-26-0-161-142:62884:63299 [3] NCCL INFO Channel 14/0 : 43[3] -> 44[4] via P2P/IPC
+ip-26-0-168-52:2722050:2722471 [5] NCCL INFO Channel 15/0 : 125[5] -> 126[6] via P2P/IPC
+ip-26-0-161-142:62883:63301 [2] NCCL INFO Channel 09/0 : 42[2] -> 43[3] via P2P/IPC
+ip-26-0-168-34:272150:272576 [0] NCCL INFO Channel 05/0 : 112[0] -> 119[7] via P2P/IPC
+ip-26-0-168-30:235652:236066 [4] NCCL INFO Channel 15/0 : 108[4] -> 109[5] via P2P/IPC
+ip-26-0-168-34:272152:272575 [2] NCCL INFO Channel 02/0 : 114[2] -> 115[3] via P2P/IPC
+ip-26-0-160-225:124165:124584 [5] NCCL INFO Channel 02/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-168-34:272153:272573 [3] NCCL INFO Channel 02/0 : 115[3] -> 116[4] via P2P/IPC
+ip-26-0-161-103:96890:97313 [2] NCCL INFO Channel 04/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-161-123:32707:33124 [3] NCCL INFO Channel 01/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-160-225:124163:124588 [3] NCCL INFO Channel 10/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-168-52:2722046:2722466 [1] NCCL INFO Channel 01/0 : 121[1] -> 122[2] via P2P/IPC
+ip-26-0-161-78:114512:114930 [0] NCCL INFO Channel 07/0 : 64[0] -> 71[7] via P2P/IPC
+ip-26-0-168-30:235650:236065 [2] NCCL INFO Channel 10/0 : 98[2] -> 106[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62884:63299 [3] NCCL INFO Channel 15/0 : 43[3] -> 44[4] via P2P/IPC
+ip-26-0-168-52:2722048:2722469 [3] NCCL INFO Channel 15/0 : 123[3] -> 124[4] via P2P/IPC
+ip-26-0-165-202:153566:153980 [4] NCCL INFO Channel 15/0 : 92[4] -> 93[5] via P2P/IPC
+ip-26-0-168-30:235650:236065 [2] NCCL INFO Channel 02/0 : 106[2] -> 114[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96894:97312 [6] NCCL INFO Channel 04/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-161-142:62883:63301 [2] NCCL INFO Channel 10/0 : 42[2] -> 43[3] via P2P/IPC
+ip-26-0-165-131:1630207:1630648 [2] NCCL INFO Channel 12/0 : 74[2] -> 75[3] via P2P/IPC
+ip-26-0-165-202:153568:153982 [6] NCCL INFO Channel 04/0 : 94[6] -> 95[7] via P2P/IPC
+ip-26-0-165-202:153567:153979 [5] NCCL INFO Channel 13/0 : 93[5] -> 94[6] via P2P/IPC
+ip-26-0-168-30:235650:236065 [2] NCCL INFO Channel 10/0 : 106[2] -> 114[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96893:97310 [5] NCCL INFO Channel 07/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-161-78:114512:114930 [0] NCCL INFO Channel 13/0 : 64[0] -> 71[7] via P2P/IPC
+ip-26-0-168-30:235651:236068 [3] NCCL INFO Channel 11/0 : 99[3] -> 107[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32707:33124 [3] NCCL INFO Channel 02/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-160-225:124165:124584 [5] NCCL INFO Channel 03/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-168-52:2722046:2722466 [1] NCCL INFO Channel 03/0 : 121[1] -> 122[2] via P2P/IPC
+ip-26-0-168-30:235653:236067 [5] NCCL INFO Channel 13/0 : 101[5] -> 109[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235651:236068 [3] NCCL INFO Channel 03/0 : 107[3] -> 115[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815963:2816400 [0] NCCL INFO Channel 13/0 : 56[0] -> 57[1] via P2P/IPC
+ip-26-0-160-225:124163:124588 [3] NCCL INFO Channel 11/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-142:62883:63301 [2] NCCL INFO Channel 12/0 : 42[2] -> 43[3] via P2P/IPC
+ip-26-0-168-52:2722051:2722465 [6] NCCL INFO Channel 06/0 : 118[6] -> 126[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235652:236066 [4] NCCL INFO Channel 12/0 : 100[4] -> 108[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722050:2722471 [5] NCCL INFO Channel 05/0 : 117[5] -> 125[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630205:1630646 [0] NCCL INFO Channel 06/0 : 72[0] -> 79[7] via P2P/IPC
+ip-26-0-165-131:1630209:1630651 [4] NCCL INFO Channel 02/0 : 76[4] -> 77[5] via P2P/IPC
+ip-26-0-161-78:114512:114930 [0] NCCL INFO Channel 14/0 : 64[0] -> 71[7] via P2P/IPC
+ip-26-0-168-30:235651:236068 [3] NCCL INFO Channel 11/0 : 107[3] -> 115[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235652:236066 [4] NCCL INFO Channel 04/0 : 108[4] -> 116[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235653:236067 [5] NCCL INFO Channel 05/0 : 109[5] -> 117[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557268:1557686 [2] NCCL INFO Channel 02/0 : 82[2] -> 98[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32707:33124 [3] NCCL INFO Channel 03/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-165-202:153568:153982 [6] NCCL INFO Channel 05/0 : 94[6] -> 95[7] via P2P/IPC
+ip-26-0-168-30:235652:236066 [4] NCCL INFO Channel 12/0 : 108[4] -> 116[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235653:236067 [5] NCCL INFO Channel 13/0 : 109[5] -> 117[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-52:2722046:2722466 [1] NCCL INFO Channel 04/0 : 121[1] -> 122[2] via P2P/IPC
+ip-26-0-165-59:1557268:1557686 [2] NCCL INFO Channel 02/0 : 98[2] -> 114[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630208:1630650 [3] NCCL INFO Channel 06/0 : 75[3] -> 76[4] via P2P/IPC
+ip-26-0-165-202:153567:153979 [5] NCCL INFO Channel 15/0 : 93[5] -> 94[6] via P2P/IPC
+ip-26-0-161-142:62883:63301 [2] NCCL INFO Channel 13/0 : 42[2] -> 43[3] via P2P/IPC
+ip-26-0-161-142:62885:63304 [4] NCCL INFO Channel 12/0 : 36[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722049:2722467 [4] NCCL INFO Channel 04/0 : 116[4] -> 124[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272152:272575 [2] NCCL INFO Channel 04/0 : 114[2] -> 115[3] via P2P/IPC
+ip-26-0-161-78:114512:114930 [0] NCCL INFO Channel 15/0 : 64[0] -> 71[7] via P2P/IPC
+ip-26-0-165-59:1557269:1557684 [3] NCCL INFO Channel 03/0 : 83[3] -> 99[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62885:63304 [4] NCCL INFO Channel 04/0 : 44[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722048:2722469 [3] NCCL INFO Channel 03/0 : 115[3] -> 123[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272150:272576 [0] NCCL INFO Channel 06/0 : 112[0] -> 119[7] via P2P/IPC
+ip-26-0-161-142:62885:63304 [4] NCCL INFO Channel 12/0 : 44[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722046:2722466 [1] NCCL INFO Channel 05/0 : 121[1] -> 122[2] via P2P/IPC
+ip-26-0-165-59:1557269:1557684 [3] NCCL INFO Channel 03/0 : 99[3] -> 115[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32707:33124 [3] NCCL INFO Channel 05/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-165-202:153565:153985 [3] NCCL INFO Channel 03/0 : 83[3] -> 91[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62883:63301 [2] NCCL INFO Channel 14/0 : 42[2] -> 43[3] via P2P/IPC
+ip-26-0-160-225:124165:124584 [5] NCCL INFO Channel 04/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-225:124163:124588 [3] NCCL INFO Channel 13/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-59:1557270:1557688 [4] NCCL INFO Channel 04/0 : 84[4] -> 100[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557271:1557690 [5] NCCL INFO Channel 05/0 : 85[5] -> 101[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153568:153982 [6] NCCL INFO Channel 06/0 : 94[6] -> 95[7] via P2P/IPC
+ip-26-0-168-34:272153:272573 [3] NCCL INFO Channel 03/0 : 115[3] -> 116[4] via P2P/IPC
+ip-26-0-165-59:1557270:1557688 [4] NCCL INFO Channel 04/0 : 100[4] -> 116[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557271:1557690 [5] NCCL INFO Channel 05/0 : 101[5] -> 117[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815969:2816398 [6] NCCL INFO Connected all rings
+ip-26-0-168-52:2722046:2722466 [1] NCCL INFO Channel 06/0 : 121[1] -> 122[2] via P2P/IPC
+ip-26-0-161-142:62883:63301 [2] NCCL INFO Channel 15/0 : 42[2] -> 43[3] via P2P/IPC
+ip-26-0-165-202:153565:153985 [3] NCCL INFO Channel 11/0 : 75[3] -> 91[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32707:33124 [3] NCCL INFO Channel 06/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-165-202:153568:153982 [6] NCCL INFO Channel 09/0 : 94[6] -> 95[7] via P2P/IPC
+ip-26-0-161-78:114512:114930 [0] NCCL INFO Channel 08/0 : 64[0] -> 72[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114519:114931 [7] NCCL INFO Channel 15/0 : 71[7] -> 79[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153565:153985 [3] NCCL INFO Channel 11/0 : 91[3] -> 107[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153566:153980 [4] NCCL INFO Channel 04/0 : 84[4] -> 92[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722046:2722466 [1] NCCL INFO Channel 07/0 : 121[1] -> 122[2] via P2P/IPC
+ip-26-0-165-131:1630207:1630648 [2] NCCL INFO Channel 13/0 : 74[2] -> 75[3] via P2P/IPC
+ip-26-0-161-178:2815963:2816400 [0] NCCL INFO Channel 14/0 : 56[0] -> 57[1] via P2P/IPC
+ip-26-0-161-123:32707:33124 [3] NCCL INFO Channel 07/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-160-225:124165:124584 [5] NCCL INFO Channel 05/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-165-202:153568:153982 [6] NCCL INFO Channel 10/0 : 94[6] -> 95[7] via P2P/IPC
+ip-26-0-160-225:124163:124588 [3] NCCL INFO Channel 14/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-123:32711:33127 [7] NCCL INFO Channel 07/0 : 23[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153566:153980 [4] NCCL INFO Channel 12/0 : 76[4] -> 92[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722046:2722466 [1] NCCL INFO Channel 09/0 : 121[1] -> 122[2] via P2P/IPC
+ip-26-0-161-123:32704:33123 [0] NCCL INFO Channel 00/0 : 16[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153566:153980 [4] NCCL INFO Channel 12/0 : 92[4] -> 108[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32707:33124 [3] NCCL INFO Channel 09/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-165-131:1630209:1630651 [4] NCCL INFO Channel 03/0 : 76[4] -> 77[5] via P2P/IPC
+ip-26-0-165-202:153568:153982 [6] NCCL INFO Channel 11/0 : 94[6] -> 95[7] via P2P/IPC
+ip-26-0-161-142:62882:63300 [1] NCCL INFO Channel 09/0 : 33[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62884:63299 [3] NCCL INFO Channel 11/0 : 35[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96894:97312 [6] NCCL INFO Channel 05/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-165-131:1630208:1630650 [3] NCCL INFO Channel 07/0 : 75[3] -> 76[4] via P2P/IPC
+ip-26-0-165-131:1630205:1630646 [0] NCCL INFO Channel 07/0 : 72[0] -> 79[7] via P2P/IPC
+ip-26-0-168-52:2722046:2722466 [1] NCCL INFO Channel 11/0 : 121[1] -> 122[2] via P2P/IPC
+ip-26-0-161-142:62883:63301 [2] NCCL INFO Channel 10/0 : 34[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96890:97313 [2] NCCL INFO Channel 05/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-161-142:62882:63300 [1] NCCL INFO Channel 01/0 : 41[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62884:63299 [3] NCCL INFO Channel 03/0 : 43[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62883:63301 [2] NCCL INFO Channel 02/0 : 42[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32707:33124 [3] NCCL INFO Channel 10/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-161-103:96893:97310 [5] NCCL INFO Channel 09/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-161-138:96568:97007 [5] NCCL INFO Connected all rings
+ip-26-0-161-142:62882:63300 [1] NCCL INFO Channel 09/0 : 41[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153568:153982 [6] NCCL INFO Channel 12/0 : 94[6] -> 95[7] via P2P/IPC
+ip-26-0-161-142:62884:63299 [3] NCCL INFO Channel 11/0 : 43[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62883:63301 [2] NCCL INFO Channel 10/0 : 42[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:124165:124584 [5] NCCL INFO Channel 07/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-242:1364906:1365331 [4] NCCL INFO Channel 01/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-160-225:124163:124588 [3] NCCL INFO Channel 15/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-168-52:2722046:2722466 [1] NCCL INFO Channel 12/0 : 121[1] -> 122[2] via P2P/IPC
+ip-26-0-160-242:1364908:1365332 [6] NCCL INFO Channel 14/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32707:33124 [3] NCCL INFO Channel 11/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-168-34:272152:272575 [2] NCCL INFO Channel 05/0 : 114[2] -> 115[3] via P2P/IPC
+ip-26-0-160-242:1364908:1365332 [6] NCCL INFO Channel 06/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153568:153982 [6] NCCL INFO Channel 13/0 : 94[6] -> 95[7] via P2P/IPC
+ip-26-0-161-138:96564:97009 [1] NCCL INFO Channel 01/0 : 17[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96565:97004 [2] NCCL INFO Channel 02/0 : 18[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-52:2722046:2722466 [1] NCCL INFO Channel 13/0 : 121[1] -> 122[2] via P2P/IPC
+ip-26-0-160-242:1364908:1365332 [6] NCCL INFO Channel 14/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96570:97005 [7] NCCL INFO Channel 15/0 : 39[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96563:97003 [0] NCCL INFO Channel 08/0 : 32[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364906:1365331 [4] NCCL INFO Channel 02/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-161-138:96564:97009 [1] NCCL INFO Channel 01/0 : 33[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96565:97004 [2] NCCL INFO Channel 02/0 : 34[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32707:33124 [3] NCCL INFO Channel 13/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-161-103:96894:97312 [6] NCCL INFO Channel 06/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-165-202:153568:153982 [6] NCCL INFO Channel 14/0 : 94[6] -> 95[7] via P2P/IPC
+ip-26-0-165-131:1630207:1630648 [2] NCCL INFO Channel 14/0 : 74[2] -> 75[3] via P2P/IPC
+ip-26-0-168-52:2722046:2722466 [1] NCCL INFO Channel 14/0 : 121[1] -> 122[2] via P2P/IPC
+ip-26-0-161-103:96890:97313 [2] NCCL INFO Channel 06/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-161-103:96893:97310 [5] NCCL INFO Channel 10/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-160-225:124165:124584 [5] NCCL INFO Channel 09/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-138:96563:97003 [0] NCCL INFO Channel 00/0 : 16[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630211:1630649 [6] NCCL INFO Channel 01/0 : 78[6] -> 79[7] via P2P/IPC
+ip-26-0-161-123:32707:33124 [3] NCCL INFO Channel 14/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-161-138:96570:97005 [7] NCCL INFO Channel 07/0 : 23[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96563:97003 [0] NCCL INFO Channel 00/0 : 32[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96570:97005 [7] NCCL INFO Channel 07/0 : 39[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722046:2722466 [1] NCCL INFO Channel 15/0 : 121[1] -> 122[2] via P2P/IPC
+ip-26-0-160-242:1364906:1365331 [4] NCCL INFO Channel 03/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-165-131:1630208:1630650 [3] NCCL INFO Channel 09/0 : 75[3] -> 76[4] via P2P/IPC
+ip-26-0-168-34:272152:272575 [2] NCCL INFO Channel 06/0 : 114[2] -> 115[3] via P2P/IPC
+ip-26-0-160-225:124162:124587 [2] NCCL INFO Channel 10/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32707:33124 [3] NCCL INFO Channel 15/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-165-131:1630209:1630651 [4] NCCL INFO Channel 04/0 : 76[4] -> 77[5] via P2P/IPC
+ip-26-0-160-225:124165:124584 [5] NCCL INFO Channel 10/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-168-34:272153:272573 [3] NCCL INFO Channel 05/0 : 115[3] -> 116[4] via P2P/IPC
+ip-26-0-160-225:124163:124588 [3] NCCL INFO Channel 11/0 : 3[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153567:153979 [5] NCCL INFO Channel 05/0 : 85[5] -> 93[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630205:1630646 [0] NCCL INFO Channel 13/0 : 72[0] -> 79[7] via P2P/IPC
+ip-26-0-161-103:96894:97312 [6] NCCL INFO Channel 09/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-165-202:153568:153982 [6] NCCL INFO Channel 06/0 : 86[6] -> 94[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96890:97313 [2] NCCL INFO Channel 07/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-161-103:96893:97310 [5] NCCL INFO Channel 11/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-160-225:124162:124587 [2] NCCL INFO Channel 02/0 : 66[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364906:1365331 [4] NCCL INFO Channel 04/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-160-225:124165:124584 [5] NCCL INFO Channel 11/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-168-52:2722045:2722470 [0] NCCL INFO Channel 05/0 : 120[0] -> 127[7] via P2P/IPC
+ip-26-0-160-225:124162:124587 [2] NCCL INFO Channel 02/0 : 2[2] -> 66[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153567:153979 [5] NCCL INFO Channel 13/0 : 77[5] -> 93[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124167:124586 [7] NCCL INFO Channel 15/0 : 7[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124160:124581 [0] NCCL INFO Channel 08/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-52:2722047:2722468 [2] NCCL INFO Channel 02/0 : 114[2] -> 122[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153568:153982 [6] NCCL INFO Channel 14/0 : 78[6] -> 94[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153567:153979 [5] NCCL INFO Channel 13/0 : 93[5] -> 109[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32706:33122 [2] NCCL INFO Channel 02/0 : 18[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32708:33126 [4] NCCL INFO Channel 04/0 : 20[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:124165:124584 [5] NCCL INFO Channel 12/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-168-52:2722046:2722466 [1] NCCL INFO Channel 01/0 : 113[1] -> 121[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153568:153982 [6] NCCL INFO Channel 14/0 : 94[6] -> 110[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364906:1365331 [4] NCCL INFO Channel 06/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-161-123:32707:33124 [3] NCCL INFO Channel 03/0 : 19[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:124165:124584 [5] NCCL INFO Channel 13/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-165-131:1630207:1630648 [2] NCCL INFO Channel 15/0 : 74[2] -> 75[3] via P2P/IPC
+ip-26-0-168-52:2722045:2722470 [0] NCCL INFO Channel 06/0 : 120[0] -> 127[7] via P2P/IPC
+ip-26-0-165-131:1630211:1630649 [6] NCCL INFO Channel 02/0 : 78[6] -> 79[7] via P2P/IPC
+ip-26-0-160-242:1364906:1365331 [4] NCCL INFO Channel 07/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-160-225:124165:124584 [5] NCCL INFO Channel 15/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-168-52:2722045:2722470 [0] NCCL INFO Channel 07/0 : 120[0] -> 127[7] via P2P/IPC
+ip-26-0-165-131:1630208:1630650 [3] NCCL INFO Channel 10/0 : 75[3] -> 76[4] via P2P/IPC
+ip-26-0-165-131:1630209:1630651 [4] NCCL INFO Channel 06/0 : 76[4] -> 77[5] via P2P/IPC
+ip-26-0-165-131:1630205:1630646 [0] NCCL INFO Channel 14/0 : 72[0] -> 79[7] via P2P/IPC
+ip-26-0-168-52:2722045:2722470 [0] NCCL INFO Channel 13/0 : 120[0] -> 127[7] via P2P/IPC
+ip-26-0-160-242:1364906:1365331 [4] NCCL INFO Channel 09/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-161-103:96892:97308 [4] NCCL INFO Channel 01/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-161-153:2636085:2636509 [1] NCCL INFO Channel 01/0 : 49[1] -> 50[2] via P2P/IPC
+ip-26-0-160-242:1364909:1365327 [7] NCCL INFO Channel 15/0 : 7[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124164:124585 [4] NCCL INFO Channel 12/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364906:1365331 [4] NCCL INFO Channel 10/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-165-202:153569:153981 [7] NCCL INFO Channel 07/0 : 87[7] -> 95[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364909:1365327 [7] NCCL INFO Channel 07/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124166:124583 [6] NCCL INFO Channel 14/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:124165:124584 [5] NCCL INFO Channel 13/0 : 5[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153562:153983 [0] NCCL INFO Channel 00/0 : 80[0] -> 88[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364909:1365327 [7] NCCL INFO Channel 15/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364906:1365331 [4] NCCL INFO Channel 11/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-165-202:153569:153981 [7] NCCL INFO Channel 15/0 : 79[7] -> 95[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153562:153983 [0] NCCL INFO Channel 08/0 : 72[0] -> 88[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:124166:124583 [6] NCCL INFO Channel 06/0 : 70[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96567:97008 [4] NCCL INFO Channel 01/0 : 36[4] -> 37[5] via P2P/IPC
+ip-26-0-165-202:153569:153981 [7] NCCL INFO Channel 15/0 : 95[7] -> 111[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124167:124586 [7] NCCL INFO Channel 07/0 : 71[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364906:1365331 [4] NCCL INFO Channel 12/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-165-131:1630211:1630649 [6] NCCL INFO Channel 03/0 : 78[6] -> 79[7] via P2P/IPC
+ip-26-0-160-225:124166:124583 [6] NCCL INFO Channel 06/0 : 6[6] -> 70[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153562:153983 [0] NCCL INFO Channel 08/0 : 88[0] -> 104[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:124167:124586 [7] NCCL INFO Channel 07/0 : 7[7] -> 71[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815963:2816400 [0] NCCL INFO Channel 15/0 : 56[0] -> 57[1] via P2P/IPC
+ip-26-0-165-131:1630208:1630650 [3] NCCL INFO Channel 11/0 : 75[3] -> 76[4] via P2P/IPC
+ip-26-0-160-242:1364906:1365331 [4] NCCL INFO Channel 14/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-161-138:96567:97008 [4] NCCL INFO Channel 02/0 : 36[4] -> 37[5] via P2P/IPC
+ip-26-0-165-131:1630206:1630647 [1] NCCL INFO Channel 09/0 : 65[1] -> 73[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630209:1630651 [4] NCCL INFO Channel 07/0 : 76[4] -> 77[5] via P2P/IPC
+ip-26-0-165-131:1630206:1630647 [1] NCCL INFO Channel 01/0 : 73[1] -> 81[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630206:1630647 [1] NCCL INFO Channel 09/0 : 73[1] -> 81[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364906:1365331 [4] NCCL INFO Channel 15/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-161-138:96568:97007 [5] NCCL INFO Channel 01/0 : 37[5] -> 38[6] via P2P/IPC
+ip-26-0-161-138:96567:97008 [4] NCCL INFO Channel 03/0 : 36[4] -> 37[5] via P2P/IPC
+ip-26-0-168-34:272150:272576 [0] NCCL INFO Channel 07/0 : 112[0] -> 119[7] via P2P/IPC
+ip-26-0-161-78:114513:114933 [1] NCCL INFO Channel 01/0 : 33[1] -> 65[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630206:1630647 [1] NCCL INFO Channel 09/0 : 73[1] -> 89[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114513:114933 [1] NCCL INFO Channel 01/0 : 65[1] -> 97[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227464:1227884 [1] NCCL INFO Channel 01/0 : 81[1] -> 97[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96568:97007 [5] NCCL INFO Channel 02/0 : 37[5] -> 38[6] via P2P/IPC
+ip-26-0-161-138:96567:97008 [4] NCCL INFO Channel 04/0 : 36[4] -> 37[5] via P2P/IPC
+ip-26-0-165-131:1630211:1630649 [6] NCCL INFO Channel 04/0 : 78[6] -> 79[7] via P2P/IPC
+ip-26-0-168-34:272152:272575 [2] NCCL INFO Channel 07/0 : 114[2] -> 115[3] via P2P/IPC
+ip-26-0-160-242:1364905:1365328 [3] NCCL INFO Channel 11/0 : 3[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630208:1630650 [3] NCCL INFO Channel 13/0 : 75[3] -> 76[4] via P2P/IPC
+ip-26-0-165-131:1630210:1630653 [5] NCCL INFO Channel 01/0 : 77[5] -> 78[6] via P2P/IPC
+ip-26-0-160-242:1364907:1365329 [5] NCCL INFO Channel 13/0 : 5[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630206:1630647 [1] NCCL INFO Channel 09/0 : 89[1] -> 73[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227464:1227884 [1] NCCL INFO Channel 01/0 : 97[1] -> 81[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636085:2636509 [1] NCCL INFO Channel 03/0 : 49[1] -> 50[2] via P2P/IPC
+ip-26-0-160-242:1364905:1365328 [3] NCCL INFO Channel 03/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364906:1365331 [4] NCCL INFO Channel 12/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630209:1630651 [4] NCCL INFO Channel 09/0 : 76[4] -> 77[5] via P2P/IPC
+ip-26-0-160-242:1364907:1365329 [5] NCCL INFO Channel 05/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96568:97007 [5] NCCL INFO Channel 03/0 : 37[5] -> 38[6] via P2P/IPC
+ip-26-0-161-103:96894:97312 [6] NCCL INFO Channel 10/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-160-242:1364905:1365328 [3] NCCL INFO Channel 11/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364906:1365331 [4] NCCL INFO Channel 04/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364907:1365329 [5] NCCL INFO Channel 13/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96567:97008 [4] NCCL INFO Channel 06/0 : 36[4] -> 37[5] via P2P/IPC
+ip-26-0-160-242:1364906:1365331 [4] NCCL INFO Channel 12/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272150:272576 [0] NCCL INFO Channel 13/0 : 112[0] -> 119[7] via P2P/IPC
+ip-26-0-168-34:272154:272569 [4] NCCL INFO Channel 01/0 : 116[4] -> 117[5] via P2P/IPC
+ip-26-0-161-103:96893:97310 [5] NCCL INFO Channel 12/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-160-225:124165:124584 [5] NCCL INFO Channel 05/0 : 69[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124163:124588 [3] NCCL INFO Channel 03/0 : 67[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96568:97007 [5] NCCL INFO Channel 04/0 : 37[5] -> 38[6] via P2P/IPC
+ip-26-0-168-34:272152:272575 [2] NCCL INFO Channel 09/0 : 114[2] -> 115[3] via P2P/IPC
+ip-26-0-160-225:124164:124585 [4] NCCL INFO Channel 04/0 : 68[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96890:97313 [2] NCCL INFO Channel 09/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-160-225:124165:124584 [5] NCCL INFO Channel 05/0 : 5[5] -> 69[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124163:124588 [3] NCCL INFO Channel 03/0 : 3[3] -> 67[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96567:97008 [4] NCCL INFO Channel 07/0 : 36[4] -> 37[5] via P2P/IPC
+ip-26-0-160-225:124164:124585 [4] NCCL INFO Channel 04/0 : 4[4] -> 68[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630211:1630649 [6] NCCL INFO Channel 05/0 : 78[6] -> 79[7] via P2P/IPC
+ip-26-0-165-131:1630208:1630650 [3] NCCL INFO Channel 14/0 : 75[3] -> 76[4] via P2P/IPC
+ip-26-0-161-138:96568:97007 [5] NCCL INFO Channel 05/0 : 37[5] -> 38[6] via P2P/IPC
+ip-26-0-161-153:2636085:2636509 [1] NCCL INFO Channel 04/0 : 49[1] -> 50[2] via P2P/IPC
+ip-26-0-161-103:96894:97312 [6] NCCL INFO Channel 11/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-165-131:1630210:1630653 [5] NCCL INFO Channel 02/0 : 77[5] -> 78[6] via P2P/IPC
+ip-26-0-161-138:96567:97008 [4] NCCL INFO Channel 09/0 : 36[4] -> 37[5] via P2P/IPC
+ip-26-0-165-131:1630209:1630651 [4] NCCL INFO Channel 10/0 : 76[4] -> 77[5] via P2P/IPC
+ip-26-0-161-178:2815964:2816401 [1] NCCL INFO Channel 01/0 : 57[1] -> 58[2] via P2P/IPC
+ip-26-0-168-34:272150:272576 [0] NCCL INFO Channel 14/0 : 112[0] -> 119[7] via P2P/IPC
+ip-26-0-168-34:272154:272569 [4] NCCL INFO Channel 02/0 : 116[4] -> 117[5] via P2P/IPC
+ip-26-0-161-103:96892:97308 [4] NCCL INFO Channel 02/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-161-103:96893:97310 [5] NCCL INFO Channel 13/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-161-138:96568:97007 [5] NCCL INFO Channel 07/0 : 37[5] -> 38[6] via P2P/IPC
+ip-26-0-161-103:96890:97313 [2] NCCL INFO Channel 10/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-161-138:96567:97008 [4] NCCL INFO Channel 10/0 : 36[4] -> 37[5] via P2P/IPC
+ip-26-0-168-34:272152:272575 [2] NCCL INFO Channel 10/0 : 114[2] -> 115[3] via P2P/IPC
+ip-26-0-161-138:96568:97007 [5] NCCL INFO Channel 09/0 : 37[5] -> 38[6] via P2P/IPC
+ip-26-0-161-103:96894:97312 [6] NCCL INFO Channel 12/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-161-138:96567:97008 [4] NCCL INFO Channel 11/0 : 36[4] -> 37[5] via P2P/IPC
+ip-26-0-165-131:1630211:1630649 [6] NCCL INFO Channel 06/0 : 78[6] -> 79[7] via P2P/IPC
+ip-26-0-168-34:272150:272576 [0] NCCL INFO Channel 15/0 : 112[0] -> 119[7] via P2P/IPC
+ip-26-0-168-34:272154:272569 [4] NCCL INFO Channel 03/0 : 116[4] -> 117[5] via P2P/IPC
+ip-26-0-161-178:2815964:2816401 [1] NCCL INFO Channel 03/0 : 57[1] -> 58[2] via P2P/IPC
+ip-26-0-165-131:1630208:1630650 [3] NCCL INFO Channel 15/0 : 75[3] -> 76[4] via P2P/IPC
+ip-26-0-165-131:1630210:1630653 [5] NCCL INFO Channel 03/0 : 77[5] -> 78[6] via P2P/IPC
+ip-26-0-165-131:1630209:1630651 [4] NCCL INFO Channel 11/0 : 76[4] -> 77[5] via P2P/IPC
+ip-26-0-161-153:2636085:2636509 [1] NCCL INFO Channel 05/0 : 49[1] -> 50[2] via P2P/IPC
+ip-26-0-161-138:96568:97007 [5] NCCL INFO Channel 10/0 : 37[5] -> 38[6] via P2P/IPC
+ip-26-0-161-103:96892:97308 [4] NCCL INFO Channel 03/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-161-138:96567:97008 [4] NCCL INFO Channel 12/0 : 36[4] -> 37[5] via P2P/IPC
+ip-26-0-161-103:96893:97310 [5] NCCL INFO Channel 15/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-168-34:272152:272575 [2] NCCL INFO Channel 12/0 : 114[2] -> 115[3] via P2P/IPC
+ip-26-0-161-153:2636086:2636507 [2] NCCL INFO Channel 01/0 : 50[2] -> 51[3] via P2P/IPC
+ip-26-0-161-103:96890:97313 [2] NCCL INFO Channel 12/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-161-138:96568:97007 [5] NCCL INFO Channel 11/0 : 37[5] -> 38[6] via P2P/IPC
+ip-26-0-168-52:2722045:2722470 [0] NCCL INFO Channel 14/0 : 120[0] -> 127[7] via P2P/IPC
+ip-26-0-161-138:96567:97008 [4] NCCL INFO Channel 14/0 : 36[4] -> 37[5] via P2P/IPC
+ip-26-0-161-178:2815964:2816401 [1] NCCL INFO Channel 04/0 : 57[1] -> 58[2] via P2P/IPC
+ip-26-0-161-178:2815965:2816395 [2] NCCL INFO Connected all rings
+ip-26-0-168-34:272154:272569 [4] NCCL INFO Channel 04/0 : 116[4] -> 117[5] via P2P/IPC
+ip-26-0-161-103:96894:97312 [6] NCCL INFO Channel 13/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-165-131:1630211:1630649 [6] NCCL INFO Channel 09/0 : 78[6] -> 79[7] via P2P/IPC
+ip-26-0-168-34:272152:272575 [2] NCCL INFO Channel 13/0 : 114[2] -> 115[3] via P2P/IPC
+ip-26-0-168-34:272155:272574 [5] NCCL INFO Channel 01/0 : 117[5] -> 118[6] via P2P/IPC
+ip-26-0-161-138:96568:97007 [5] NCCL INFO Channel 12/0 : 37[5] -> 38[6] via P2P/IPC
+ip-26-0-168-52:2722045:2722470 [0] NCCL INFO Channel 15/0 : 120[0] -> 127[7] via P2P/IPC
+ip-26-0-161-138:96567:97008 [4] NCCL INFO Channel 15/0 : 36[4] -> 37[5] via P2P/IPC
+ip-26-0-165-131:1630209:1630651 [4] NCCL INFO Channel 12/0 : 76[4] -> 77[5] via P2P/IPC
+ip-26-0-165-131:1630210:1630653 [5] NCCL INFO Channel 04/0 : 77[5] -> 78[6] via P2P/IPC
+ip-26-0-161-103:96892:97308 [4] NCCL INFO Channel 04/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-165-131:1630207:1630648 [2] NCCL INFO Channel 10/0 : 66[2] -> 74[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96568:97007 [5] NCCL INFO Channel 13/0 : 37[5] -> 38[6] via P2P/IPC
+ip-26-0-161-103:96890:97313 [2] NCCL INFO Channel 13/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-168-34:272154:272569 [4] NCCL INFO Channel 06/0 : 116[4] -> 117[5] via P2P/IPC
+ip-26-0-165-131:1630207:1630648 [2] NCCL INFO Channel 02/0 : 74[2] -> 82[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636085:2636509 [1] NCCL INFO Channel 06/0 : 49[1] -> 50[2] via P2P/IPC
+ip-26-0-165-131:1630207:1630648 [2] NCCL INFO Channel 10/0 : 74[2] -> 82[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630211:1630649 [6] NCCL INFO Channel 10/0 : 78[6] -> 79[7] via P2P/IPC
+ip-26-0-168-52:2722052:2722464 [7] NCCL INFO Channel 07/0 : 119[7] -> 127[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272152:272575 [2] NCCL INFO Channel 14/0 : 114[2] -> 115[3] via P2P/IPC
+ip-26-0-161-153:2636086:2636507 [2] NCCL INFO Channel 02/0 : 50[2] -> 51[3] via P2P/IPC
+ip-26-0-161-138:96568:97007 [5] NCCL INFO Channel 15/0 : 37[5] -> 38[6] via P2P/IPC
+ip-26-0-161-103:96894:97312 [6] NCCL INFO Channel 14/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-168-52:2722045:2722470 [0] NCCL INFO Channel 00/0 : 112[0] -> 120[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272155:272574 [5] NCCL INFO Channel 02/0 : 117[5] -> 118[6] via P2P/IPC
+ip-26-0-161-138:96566:97010 [3] NCCL INFO Channel 11/0 : 35[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815964:2816401 [1] NCCL INFO Channel 05/0 : 57[1] -> 58[2] via P2P/IPC
+ip-26-0-165-131:1630209:1630651 [4] NCCL INFO Channel 14/0 : 76[4] -> 77[5] via P2P/IPC
+ip-26-0-165-131:1630210:1630653 [5] NCCL INFO Channel 05/0 : 77[5] -> 78[6] via P2P/IPC
+ip-26-0-165-131:1630207:1630648 [2] NCCL INFO Channel 10/0 : 74[2] -> 90[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114514:114932 [2] NCCL INFO Channel 02/0 : 34[2] -> 66[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96892:97308 [4] NCCL INFO Channel 06/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-165-164:1227465:1227881 [2] NCCL INFO Channel 02/0 : 82[2] -> 98[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96566:97010 [3] NCCL INFO Channel 03/0 : 19[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96890:97313 [2] NCCL INFO Channel 14/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-161-78:114514:114932 [2] NCCL INFO Channel 02/0 : 66[2] -> 98[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96567:97008 [4] NCCL INFO Channel 12/0 : 36[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96566:97010 [3] NCCL INFO Channel 03/0 : 35[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630207:1630648 [2] NCCL INFO Channel 10/0 : 90[2] -> 74[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227465:1227881 [2] NCCL INFO Channel 02/0 : 98[2] -> 82[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96569:97006 [6] NCCL INFO Channel 14/0 : 38[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96568:97007 [5] NCCL INFO Channel 13/0 : 37[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630211:1630649 [6] NCCL INFO Channel 11/0 : 78[6] -> 79[7] via P2P/IPC
+ip-26-0-161-138:96567:97008 [4] NCCL INFO Channel 04/0 : 20[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96567:97008 [4] NCCL INFO Channel 04/0 : 36[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630205:1630646 [0] NCCL INFO Channel 15/0 : 72[0] -> 79[7] via P2P/IPC
+ip-26-0-161-138:96568:97007 [5] NCCL INFO Channel 05/0 : 21[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96569:97006 [6] NCCL INFO Channel 06/0 : 22[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630209:1630651 [4] NCCL INFO Channel 15/0 : 76[4] -> 77[5] via P2P/IPC
+ip-26-0-161-138:96568:97007 [5] NCCL INFO Channel 05/0 : 37[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630210:1630653 [5] NCCL INFO Channel 07/0 : 77[5] -> 78[6] via P2P/IPC
+ip-26-0-161-138:96569:97006 [6] NCCL INFO Channel 06/0 : 38[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96892:97308 [4] NCCL INFO Channel 07/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-161-103:96890:97313 [2] NCCL INFO Channel 15/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-161-178:2815964:2816401 [1] NCCL INFO Channel 06/0 : 57[1] -> 58[2] via P2P/IPC
+ip-26-0-160-242:1364902:1365333 [0] NCCL INFO Channel 08/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364902:1365333 [0] NCCL INFO Channel 00/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272153:272573 [3] NCCL INFO Channel 06/0 : 115[3] -> 116[4] via P2P/IPC
+ip-26-0-161-103:96892:97308 [4] NCCL INFO Channel 09/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-165-131:1630211:1630649 [6] NCCL INFO Channel 12/0 : 78[6] -> 79[7] via P2P/IPC
+ip-26-0-168-34:272154:272569 [4] NCCL INFO Channel 07/0 : 116[4] -> 117[5] via P2P/IPC
+ip-26-0-161-153:2636088:2636510 [4] NCCL INFO Connected all rings
+ip-26-0-161-178:2815966:2816396 [3] NCCL INFO Connected all rings
+ip-26-0-160-242:1364902:1365333 [0] NCCL INFO Channel 08/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272155:272574 [5] NCCL INFO Channel 03/0 : 117[5] -> 118[6] via P2P/IPC
+ip-26-0-161-103:96891:97311 [3] NCCL INFO Channel 01/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-168-34:272152:272575 [2] NCCL INFO Channel 15/0 : 114[2] -> 115[3] via P2P/IPC
+ip-26-0-165-131:1630210:1630653 [5] NCCL INFO Channel 09/0 : 77[5] -> 78[6] via P2P/IPC
+ip-26-0-161-103:96889:97309 [1] NCCL INFO Channel 01/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96892:97308 [4] NCCL INFO Channel 10/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-161-103:96889:97309 [1] NCCL INFO Channel 09/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124160:124581 [0] NCCL INFO Channel 00/0 : 64[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96889:97309 [1] NCCL INFO Channel 01/0 : 17[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96891:97311 [3] NCCL INFO Channel 02/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-165-131:1630211:1630649 [6] NCCL INFO Channel 13/0 : 78[6] -> 79[7] via P2P/IPC
+ip-26-0-160-225:124160:124581 [0] NCCL INFO Channel 00/0 : 0[0] -> 64[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630208:1630650 [3] NCCL INFO Channel 11/0 : 67[3] -> 75[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636085:2636509 [1] NCCL INFO Channel 07/0 : 49[1] -> 50[2] via P2P/IPC
+ip-26-0-161-178:2815964:2816401 [1] NCCL INFO Channel 07/0 : 57[1] -> 58[2] via P2P/IPC
+ip-26-0-165-131:1630208:1630650 [3] NCCL INFO Channel 03/0 : 75[3] -> 83[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96892:97308 [4] NCCL INFO Channel 11/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-161-153:2636086:2636507 [2] NCCL INFO Channel 04/0 : 50[2] -> 51[3] via P2P/IPC
+ip-26-0-168-34:272153:272573 [3] NCCL INFO Channel 07/0 : 115[3] -> 116[4] via P2P/IPC
+ip-26-0-168-34:272154:272569 [4] NCCL INFO Channel 09/0 : 116[4] -> 117[5] via P2P/IPC
+ip-26-0-165-131:1630210:1630653 [5] NCCL INFO Channel 10/0 : 77[5] -> 78[6] via P2P/IPC
+ip-26-0-165-131:1630208:1630650 [3] NCCL INFO Channel 11/0 : 75[3] -> 83[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364903:1365334 [1] NCCL INFO Channel 09/0 : 9[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630211:1630649 [6] NCCL INFO Channel 14/0 : 78[6] -> 79[7] via P2P/IPC
+ip-26-0-161-103:96891:97311 [3] NCCL INFO Channel 03/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-161-103:96889:97309 [1] NCCL INFO Channel 01/0 : 17[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272155:272574 [5] NCCL INFO Channel 04/0 : 117[5] -> 118[6] via P2P/IPC
+ip-26-0-165-131:1630210:1630653 [5] NCCL INFO Channel 11/0 : 77[5] -> 78[6] via P2P/IPC
+ip-26-0-161-78:114515:114936 [3] NCCL INFO Channel 03/0 : 35[3] -> 67[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32705:33128 [1] NCCL INFO Channel 09/0 : 9[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630208:1630650 [3] NCCL INFO Channel 11/0 : 75[3] -> 91[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114515:114936 [3] NCCL INFO Channel 03/0 : 67[3] -> 99[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32705:33128 [1] NCCL INFO Channel 09/0 : 25[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96892:97308 [4] NCCL INFO Channel 12/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-165-164:1227466:1227887 [3] NCCL INFO Channel 03/0 : 83[3] -> 99[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96889:97309 [1] NCCL INFO Channel 01/0 : 33[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272151:272571 [1] NCCL INFO Channel 01/0 : 105[1] -> 113[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96891:97311 [3] NCCL INFO Channel 05/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-165-131:1630208:1630650 [3] NCCL INFO Channel 11/0 : 91[3] -> 75[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272154:272569 [4] NCCL INFO Channel 10/0 : 116[4] -> 117[5] via P2P/IPC
+ip-26-0-168-34:272153:272573 [3] NCCL INFO Channel 09/0 : 115[3] -> 116[4] via P2P/IPC
+ip-26-0-165-131:1630210:1630653 [5] NCCL INFO Channel 12/0 : 77[5] -> 78[6] via P2P/IPC
+ip-26-0-168-34:272151:272571 [1] NCCL INFO Channel 09/0 : 105[1] -> 113[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227466:1227887 [3] NCCL INFO Channel 03/0 : 99[3] -> 83[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815964:2816401 [1] NCCL INFO Channel 09/0 : 57[1] -> 58[2] via P2P/IPC
+ip-26-0-160-242:1364903:1365334 [1] NCCL INFO Channel 09/0 : 25[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272151:272571 [1] NCCL INFO Channel 01/0 : 113[1] -> 121[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96892:97308 [4] NCCL INFO Channel 14/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-161-178:2815969:2816398 [6] NCCL INFO Channel 01/0 : 62[6] -> 63[7] via P2P/IPC
+ip-26-0-161-103:96891:97311 [3] NCCL INFO Channel 06/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-165-131:1630210:1630653 [5] NCCL INFO Channel 13/0 : 77[5] -> 78[6] via P2P/IPC
+ip-26-0-161-103:96895:97307 [7] NCCL INFO Connected all rings
+ip-26-0-168-34:272155:272574 [5] NCCL INFO Channel 05/0 : 117[5] -> 118[6] via P2P/IPC
+ip-26-0-168-30:235649:236071 [1] NCCL INFO Channel 09/0 : 89[1] -> 105[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272151:272571 [1] NCCL INFO Channel 01/0 : 97[1] -> 113[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96892:97308 [4] NCCL INFO Channel 15/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-168-52:2722046:2722466 [1] NCCL INFO Channel 09/0 : 57[1] -> 121[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630210:1630653 [5] NCCL INFO Channel 15/0 : 77[5] -> 78[6] via P2P/IPC
+ip-26-0-168-34:272154:272569 [4] NCCL INFO Channel 11/0 : 116[4] -> 117[5] via P2P/IPC
+ip-26-0-168-34:272153:272573 [3] NCCL INFO Channel 10/0 : 115[3] -> 116[4] via P2P/IPC
+ip-26-0-168-52:2722046:2722466 [1] NCCL INFO Channel 09/0 : 121[1] -> 57[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235649:236071 [1] NCCL INFO Channel 09/0 : 105[1] -> 89[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272151:272571 [1] NCCL INFO Channel 01/0 : 113[1] -> 97[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96891:97311 [3] NCCL INFO Channel 07/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-165-202:153563:153986 [1] NCCL INFO Channel 09/0 : 57[1] -> 89[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272155:272574 [5] NCCL INFO Channel 07/0 : 117[5] -> 118[6] via P2P/IPC
+ip-26-0-161-178:2815964:2816401 [1] NCCL INFO Channel 11/0 : 57[1] -> 58[2] via P2P/IPC
+ip-26-0-161-178:2815969:2816398 [6] NCCL INFO Channel 02/0 : 62[6] -> 63[7] via P2P/IPC
+ip-26-0-161-103:96891:97311 [3] NCCL INFO Channel 09/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-168-34:272154:272569 [4] NCCL INFO Channel 12/0 : 116[4] -> 117[5] via P2P/IPC
+ip-26-0-168-34:272153:272573 [3] NCCL INFO Channel 11/0 : 115[3] -> 116[4] via P2P/IPC
+ip-26-0-165-131:1630209:1630651 [4] NCCL INFO Channel 12/0 : 68[4] -> 76[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96893:97310 [5] NCCL INFO Channel 05/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815968:2816394 [5] NCCL INFO Connected all rings
+ip-26-0-165-131:1630211:1630649 [6] NCCL INFO Channel 14/0 : 70[6] -> 78[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96893:97310 [5] NCCL INFO Channel 13/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630209:1630651 [4] NCCL INFO Channel 04/0 : 76[4] -> 84[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96891:97311 [3] NCCL INFO Channel 10/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-161-103:96893:97310 [5] NCCL INFO Channel 05/0 : 21[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630210:1630653 [5] NCCL INFO Channel 13/0 : 69[5] -> 77[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630211:1630649 [6] NCCL INFO Channel 06/0 : 78[6] -> 86[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630209:1630651 [4] NCCL INFO Channel 12/0 : 76[4] -> 84[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630210:1630653 [5] NCCL INFO Channel 05/0 : 77[5] -> 85[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630211:1630649 [6] NCCL INFO Channel 14/0 : 78[6] -> 86[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272155:272574 [5] NCCL INFO Channel 09/0 : 117[5] -> 118[6] via P2P/IPC
+ip-26-0-165-131:1630210:1630653 [5] NCCL INFO Channel 13/0 : 77[5] -> 85[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364907:1365329 [5] NCCL INFO Channel 13/0 : 13[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815964:2816401 [1] NCCL INFO Channel 12/0 : 57[1] -> 58[2] via P2P/IPC
+ip-26-0-161-78:114516:114934 [4] NCCL INFO Channel 04/0 : 36[4] -> 68[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815969:2816398 [6] NCCL INFO Channel 03/0 : 62[6] -> 63[7] via P2P/IPC
+ip-26-0-165-131:1630209:1630651 [4] NCCL INFO Channel 12/0 : 76[4] -> 92[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114518:114937 [6] NCCL INFO Channel 06/0 : 38[6] -> 70[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96893:97310 [5] NCCL INFO Channel 05/0 : 21[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227467:1227886 [4] NCCL INFO Channel 04/0 : 84[4] -> 100[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630211:1630649 [6] NCCL INFO Channel 14/0 : 78[6] -> 94[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114517:114935 [5] NCCL INFO Channel 05/0 : 37[5] -> 69[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114516:114934 [4] NCCL INFO Channel 04/0 : 68[4] -> 100[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114518:114937 [6] NCCL INFO Channel 06/0 : 70[6] -> 102[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32709:33129 [5] NCCL INFO Channel 13/0 : 13[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227469:1227882 [6] NCCL INFO Channel 06/0 : 86[6] -> 102[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630210:1630653 [5] NCCL INFO Channel 13/0 : 77[5] -> 93[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114517:114935 [5] NCCL INFO Channel 05/0 : 69[5] -> 101[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32709:33129 [5] NCCL INFO Channel 13/0 : 29[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630205:1630646 [0] NCCL INFO Channel 08/0 : 64[0] -> 72[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96893:97310 [5] NCCL INFO Channel 05/0 : 37[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636086:2636507 [2] NCCL INFO Channel 05/0 : 50[2] -> 51[3] via P2P/IPC
+ip-26-0-168-34:272154:272569 [4] NCCL INFO Channel 14/0 : 116[4] -> 117[5] via P2P/IPC
+ip-26-0-165-131:1630212:1630652 [7] NCCL INFO Channel 15/0 : 71[7] -> 79[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630209:1630651 [4] NCCL INFO Channel 12/0 : 92[4] -> 76[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227468:1227883 [5] NCCL INFO Channel 05/0 : 85[5] -> 101[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227467:1227886 [4] NCCL INFO Channel 04/0 : 100[4] -> 84[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630212:1630652 [7] NCCL INFO Channel 07/0 : 79[7] -> 87[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630205:1630646 [0] NCCL INFO Channel 00/0 : 72[0] -> 80[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227469:1227882 [6] NCCL INFO Channel 06/0 : 102[6] -> 86[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630210:1630653 [5] NCCL INFO Channel 13/0 : 93[5] -> 77[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630212:1630652 [7] NCCL INFO Channel 15/0 : 79[7] -> 87[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630205:1630646 [0] NCCL INFO Channel 08/0 : 72[0] -> 80[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364907:1365329 [5] NCCL INFO Channel 13/0 : 29[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815964:2816401 [1] NCCL INFO Channel 13/0 : 57[1] -> 58[2] via P2P/IPC
+ip-26-0-161-178:2815969:2816398 [6] NCCL INFO Channel 04/0 : 62[6] -> 63[7] via P2P/IPC
+ip-26-0-165-131:1630212:1630652 [7] NCCL INFO Channel 15/0 : 79[7] -> 95[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227468:1227883 [5] NCCL INFO Channel 05/0 : 101[5] -> 85[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630205:1630646 [0] NCCL INFO Channel 08/0 : 72[0] -> 88[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636086:2636507 [2] NCCL INFO Channel 06/0 : 50[2] -> 51[3] via P2P/IPC
+ip-26-0-165-131:1630212:1630652 [7] NCCL INFO Channel 15/0 : 95[7] -> 79[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630205:1630646 [0] NCCL INFO Channel 08/0 : 88[0] -> 72[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272153:272573 [3] NCCL INFO Channel 13/0 : 115[3] -> 116[4] via P2P/IPC
+ip-26-0-168-34:272157:272572 [7] NCCL INFO Connected all rings
+ip-26-0-161-178:2815969:2816398 [6] NCCL INFO Channel 05/0 : 62[6] -> 63[7] via P2P/IPC
+ip-26-0-161-178:2815964:2816401 [1] NCCL INFO Channel 14/0 : 57[1] -> 58[2] via P2P/IPC
+ip-26-0-161-153:2636086:2636507 [2] NCCL INFO Channel 07/0 : 50[2] -> 51[3] via P2P/IPC
+ip-26-0-161-103:96891:97311 [3] NCCL INFO Channel 11/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-168-34:272155:272574 [5] NCCL INFO Channel 10/0 : 117[5] -> 118[6] via P2P/IPC
+ip-26-0-161-153:2636085:2636509 [1] NCCL INFO Channel 09/0 : 49[1] -> 50[2] via P2P/IPC
+ip-26-0-161-103:96891:97311 [3] NCCL INFO Channel 13/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-161-153:2636086:2636507 [2] NCCL INFO Channel 09/0 : 50[2] -> 51[3] via P2P/IPC
+ip-26-0-165-131:1630211:1630649 [6] NCCL INFO Channel 14/0 : 94[6] -> 78[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815969:2816398 [6] NCCL INFO Channel 06/0 : 62[6] -> 63[7] via P2P/IPC
+ip-26-0-168-34:272154:272569 [4] NCCL INFO Channel 15/0 : 116[4] -> 117[5] via P2P/IPC
+ip-26-0-168-34:272155:272574 [5] NCCL INFO Channel 11/0 : 117[5] -> 118[6] via P2P/IPC
+ip-26-0-161-178:2815964:2816401 [1] NCCL INFO Channel 15/0 : 57[1] -> 58[2] via P2P/IPC
+ip-26-0-161-103:96891:97311 [3] NCCL INFO Channel 14/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-165-164:1227470:1227885 [7] NCCL INFO Channel 07/0 : 87[7] -> 103[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557267:1557683 [1] NCCL INFO Channel 01/0 : 65[1] -> 97[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114519:114931 [7] NCCL INFO Channel 07/0 : 39[7] -> 71[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96891:97311 [3] NCCL INFO Channel 15/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-161-78:114519:114931 [7] NCCL INFO Channel 07/0 : 71[7] -> 103[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227470:1227885 [7] NCCL INFO Channel 07/0 : 103[7] -> 87[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636085:2636509 [1] NCCL INFO Channel 11/0 : 49[1] -> 50[2] via P2P/IPC
+ip-26-0-165-59:1557267:1557683 [1] NCCL INFO Channel 01/0 : 97[1] -> 65[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272153:272573 [3] NCCL INFO Channel 14/0 : 115[3] -> 116[4] via P2P/IPC
+ip-26-0-161-153:2636086:2636507 [2] NCCL INFO Channel 10/0 : 50[2] -> 51[3] via P2P/IPC
+ip-26-0-168-34:272155:272574 [5] NCCL INFO Channel 12/0 : 117[5] -> 118[6] via P2P/IPC
+ip-26-0-161-178:2815967:2816399 [4] NCCL INFO Connected all rings
+ip-26-0-161-178:2815969:2816398 [6] NCCL INFO Channel 09/0 : 62[6] -> 63[7] via P2P/IPC
+ip-26-0-161-78:114512:114930 [0] NCCL INFO Channel 00/0 : 32[0] -> 64[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114512:114930 [0] NCCL INFO Channel 00/0 : 64[0] -> 96[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815963:2816400 [0] NCCL INFO Channel 05/0 : 56[0] -> 63[7] via P2P/IPC
+ip-26-0-168-34:272153:272573 [3] NCCL INFO Channel 15/0 : 115[3] -> 116[4] via P2P/IPC
+ip-26-0-168-34:272155:272574 [5] NCCL INFO Channel 13/0 : 117[5] -> 118[6] via P2P/IPC
+ip-26-0-161-153:2636085:2636509 [1] NCCL INFO Channel 12/0 : 49[1] -> 50[2] via P2P/IPC
+ip-26-0-161-153:2636086:2636507 [2] NCCL INFO Channel 12/0 : 50[2] -> 51[3] via P2P/IPC
+ip-26-0-165-164:1227463:1227880 [0] NCCL INFO Channel 00/0 : 80[0] -> 96[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815963:2816400 [0] NCCL INFO Channel 06/0 : 56[0] -> 63[7] via P2P/IPC
+ip-26-0-168-34:272155:272574 [5] NCCL INFO Channel 15/0 : 117[5] -> 118[6] via P2P/IPC
+ip-26-0-161-103:96892:97308 [4] NCCL INFO Channel 04/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96890:97313 [2] NCCL INFO Channel 02/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96891:97311 [3] NCCL INFO Channel 03/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227463:1227880 [0] NCCL INFO Channel 00/0 : 96[0] -> 80[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96892:97308 [4] NCCL INFO Channel 12/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815965:2816395 [2] NCCL INFO Channel 01/0 : 58[2] -> 59[3] via P2P/IPC
+ip-26-0-161-103:96890:97313 [2] NCCL INFO Channel 10/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96891:97311 [3] NCCL INFO Channel 11/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272152:272575 [2] NCCL INFO Channel 02/0 : 106[2] -> 114[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96891:97311 [3] NCCL INFO Channel 03/0 : 19[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96892:97308 [4] NCCL INFO Channel 04/0 : 20[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96890:97313 [2] NCCL INFO Channel 02/0 : 18[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636085:2636509 [1] NCCL INFO Channel 13/0 : 49[1] -> 50[2] via P2P/IPC
+ip-26-0-161-153:2636086:2636507 [2] NCCL INFO Channel 13/0 : 50[2] -> 51[3] via P2P/IPC
+ip-26-0-168-34:272152:272575 [2] NCCL INFO Channel 10/0 : 106[2] -> 114[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272153:272573 [3] NCCL INFO Channel 03/0 : 107[3] -> 115[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272152:272575 [2] NCCL INFO Channel 02/0 : 114[2] -> 122[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272153:272573 [3] NCCL INFO Channel 11/0 : 107[3] -> 115[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272153:272573 [3] NCCL INFO Channel 03/0 : 115[3] -> 123[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364906:1365331 [4] NCCL INFO Channel 12/0 : 12[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815963:2816400 [0] NCCL INFO Channel 07/0 : 56[0] -> 63[7] via P2P/IPC
+ip-26-0-161-178:2815969:2816398 [6] NCCL INFO Channel 10/0 : 62[6] -> 63[7] via P2P/IPC
+ip-26-0-168-34:272154:272569 [4] NCCL INFO Channel 04/0 : 108[4] -> 116[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96890:97313 [2] NCCL INFO Channel 02/0 : 18[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96892:97308 [4] NCCL INFO Channel 04/0 : 20[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364904:1365335 [2] NCCL INFO Channel 10/0 : 10[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96891:97311 [3] NCCL INFO Channel 03/0 : 19[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32706:33122 [2] NCCL INFO Channel 10/0 : 10[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32708:33126 [4] NCCL INFO Channel 12/0 : 12[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272154:272569 [4] NCCL INFO Channel 12/0 : 108[4] -> 116[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235650:236065 [2] NCCL INFO Channel 10/0 : 90[2] -> 106[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815965:2816395 [2] NCCL INFO Channel 02/0 : 58[2] -> 59[3] via P2P/IPC
+ip-26-0-161-123:32707:33124 [3] NCCL INFO Channel 11/0 : 11[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272152:272575 [2] NCCL INFO Channel 02/0 : 98[2] -> 114[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32708:33126 [4] NCCL INFO Channel 12/0 : 28[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722047:2722468 [2] NCCL INFO Channel 10/0 : 58[2] -> 122[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32706:33122 [2] NCCL INFO Channel 10/0 : 26[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272154:272569 [4] NCCL INFO Channel 04/0 : 116[4] -> 124[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272153:272573 [3] NCCL INFO Channel 03/0 : 99[3] -> 115[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96890:97313 [2] NCCL INFO Channel 02/0 : 34[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-52:2722047:2722468 [2] NCCL INFO Channel 10/0 : 122[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32707:33124 [3] NCCL INFO Channel 11/0 : 27[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96892:97308 [4] NCCL INFO Channel 04/0 : 36[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96891:97311 [3] NCCL INFO Channel 03/0 : 35[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96894:97312 [6] NCCL INFO Channel 06/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-52:2722048:2722469 [3] NCCL INFO Channel 11/0 : 59[3] -> 123[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272152:272575 [2] NCCL INFO Channel 02/0 : 114[2] -> 98[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96894:97312 [6] NCCL INFO Channel 14/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-52:2722048:2722469 [3] NCCL INFO Channel 11/0 : 123[3] -> 59[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272153:272573 [3] NCCL INFO Channel 03/0 : 115[3] -> 99[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96894:97312 [6] NCCL INFO Channel 06/0 : 22[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364904:1365335 [2] NCCL INFO Channel 10/0 : 26[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364906:1365331 [4] NCCL INFO Channel 12/0 : 28[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272154:272569 [4] NCCL INFO Channel 04/0 : 100[4] -> 116[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636085:2636509 [1] NCCL INFO Channel 14/0 : 49[1] -> 50[2] via P2P/IPC
+ip-26-0-161-153:2636086:2636507 [2] NCCL INFO Channel 14/0 : 50[2] -> 51[3] via P2P/IPC
+ip-26-0-160-242:1364905:1365328 [3] NCCL INFO Channel 11/0 : 11[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96894:97312 [6] NCCL INFO Channel 06/0 : 22[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815963:2816400 [0] NCCL INFO Channel 13/0 : 56[0] -> 63[7] via P2P/IPC
+ip-26-0-161-123:32710:33125 [6] NCCL INFO Channel 14/0 : 14[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815969:2816398 [6] NCCL INFO Channel 11/0 : 62[6] -> 63[7] via P2P/IPC
+ip-26-0-168-30:235652:236066 [4] NCCL INFO Channel 12/0 : 92[4] -> 108[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32710:33125 [6] NCCL INFO Channel 14/0 : 30[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815965:2816395 [2] NCCL INFO Channel 04/0 : 58[2] -> 59[3] via P2P/IPC
+ip-26-0-160-242:1364905:1365328 [3] NCCL INFO Channel 11/0 : 27[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96894:97312 [6] NCCL INFO Channel 06/0 : 38[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272154:272569 [4] NCCL INFO Channel 04/0 : 116[4] -> 100[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364908:1365332 [6] NCCL INFO Channel 14/0 : 14[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557268:1557686 [2] NCCL INFO Channel 02/0 : 66[2] -> 98[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557269:1557684 [3] NCCL INFO Channel 03/0 : 67[3] -> 99[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557269:1557684 [3] NCCL INFO Channel 03/0 : 99[3] -> 67[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557268:1557686 [2] NCCL INFO Channel 02/0 : 98[2] -> 66[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235651:236068 [3] NCCL INFO Channel 11/0 : 91[3] -> 107[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235652:236066 [4] NCCL INFO Channel 12/0 : 108[4] -> 92[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235650:236065 [2] NCCL INFO Channel 10/0 : 106[2] -> 90[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153566:153980 [4] NCCL INFO Channel 12/0 : 60[4] -> 92[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153564:153984 [2] NCCL INFO Channel 10/0 : 58[2] -> 90[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364908:1365332 [6] NCCL INFO Channel 14/0 : 30[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235651:236068 [3] NCCL INFO Channel 11/0 : 107[3] -> 91[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272156:272570 [6] NCCL INFO Channel 01/0 : 118[6] -> 119[7] via P2P/IPC
+ip-26-0-165-202:153565:153985 [3] NCCL INFO Channel 11/0 : 59[3] -> 91[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815963:2816400 [0] NCCL INFO Channel 14/0 : 56[0] -> 63[7] via P2P/IPC
+ip-26-0-168-34:272156:272570 [6] NCCL INFO Channel 02/0 : 118[6] -> 119[7] via P2P/IPC
+ip-26-0-161-103:96895:97307 [7] NCCL INFO Channel 07/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96888:97306 [0] NCCL INFO Channel 00/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815969:2816398 [6] NCCL INFO Channel 12/0 : 62[6] -> 63[7] via P2P/IPC
+ip-26-0-161-178:2815966:2816396 [3] NCCL INFO Channel 01/0 : 59[3] -> 60[4] via P2P/IPC
+ip-26-0-161-103:96895:97307 [7] NCCL INFO Channel 15/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815965:2816395 [2] NCCL INFO Channel 05/0 : 58[2] -> 59[3] via P2P/IPC
+ip-26-0-168-52:2722049:2722467 [4] NCCL INFO Channel 12/0 : 60[4] -> 124[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96888:97306 [0] NCCL INFO Channel 08/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-52:2722049:2722467 [4] NCCL INFO Channel 12/0 : 124[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96895:97307 [7] NCCL INFO Channel 07/0 : 23[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96888:97306 [0] NCCL INFO Channel 00/0 : 16[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272156:272570 [6] NCCL INFO Channel 03/0 : 118[6] -> 119[7] via P2P/IPC
+ip-26-0-160-242:1364909:1365327 [7] NCCL INFO Channel 15/0 : 15[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364902:1365333 [0] NCCL INFO Channel 08/0 : 8[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96888:97306 [0] NCCL INFO Channel 00/0 : 16[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96895:97307 [7] NCCL INFO Channel 07/0 : 23[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32711:33127 [7] NCCL INFO Channel 15/0 : 15[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32704:33123 [0] NCCL INFO Channel 08/0 : 8[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32711:33127 [7] NCCL INFO Channel 15/0 : 31[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272156:272570 [6] NCCL INFO Channel 04/0 : 118[6] -> 119[7] via P2P/IPC
+ip-26-0-161-123:32704:33123 [0] NCCL INFO Channel 08/0 : 24[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96895:97307 [7] NCCL INFO Channel 07/0 : 39[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96888:97306 [0] NCCL INFO Channel 00/0 : 32[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815963:2816400 [0] NCCL INFO Channel 15/0 : 56[0] -> 63[7] via P2P/IPC
+ip-26-0-161-178:2815969:2816398 [6] NCCL INFO Channel 13/0 : 62[6] -> 63[7] via P2P/IPC
+ip-26-0-161-178:2815965:2816395 [2] NCCL INFO Channel 06/0 : 58[2] -> 59[3] via P2P/IPC
+ip-26-0-161-178:2815966:2816396 [3] NCCL INFO Channel 02/0 : 59[3] -> 60[4] via P2P/IPC
+ip-26-0-168-34:272156:272570 [6] NCCL INFO Channel 05/0 : 118[6] -> 119[7] via P2P/IPC
+ip-26-0-160-242:1364909:1365327 [7] NCCL INFO Channel 15/0 : 31[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636090:2636505 [6] NCCL INFO Connected all rings
+ip-26-0-160-242:1364902:1365333 [0] NCCL INFO Channel 08/0 : 24[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636086:2636507 [2] NCCL INFO Channel 15/0 : 50[2] -> 51[3] via P2P/IPC
+ip-26-0-161-153:2636085:2636509 [1] NCCL INFO Channel 15/0 : 49[1] -> 50[2] via P2P/IPC
+ip-26-0-168-34:272156:272570 [6] NCCL INFO Channel 06/0 : 118[6] -> 119[7] via P2P/IPC
+ip-26-0-168-34:272156:272570 [6] NCCL INFO Channel 09/0 : 118[6] -> 119[7] via P2P/IPC
+ip-26-0-165-59:1557270:1557688 [4] NCCL INFO Channel 04/0 : 68[4] -> 100[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272156:272570 [6] NCCL INFO Channel 10/0 : 118[6] -> 119[7] via P2P/IPC
+ip-26-0-161-178:2815969:2816398 [6] NCCL INFO Channel 14/0 : 62[6] -> 63[7] via P2P/IPC
+ip-26-0-165-59:1557270:1557688 [4] NCCL INFO Channel 04/0 : 100[4] -> 68[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815965:2816395 [2] NCCL INFO Channel 07/0 : 58[2] -> 59[3] via P2P/IPC
+ip-26-0-161-178:2815966:2816396 [3] NCCL INFO Channel 03/0 : 59[3] -> 60[4] via P2P/IPC
+ip-26-0-168-34:272156:272570 [6] NCCL INFO Channel 11/0 : 118[6] -> 119[7] via P2P/IPC
+ip-26-0-161-153:2636087:2636506 [3] NCCL INFO Channel 01/0 : 51[3] -> 52[4] via P2P/IPC
+ip-26-0-168-34:272156:272570 [6] NCCL INFO Channel 12/0 : 118[6] -> 119[7] via P2P/IPC
+ip-26-0-161-153:2636084:2636511 [0] NCCL INFO Channel 05/0 : 48[0] -> 55[7] via P2P/IPC
+ip-26-0-161-178:2815965:2816395 [2] NCCL INFO Channel 09/0 : 58[2] -> 59[3] via P2P/IPC
+ip-26-0-161-178:2815966:2816396 [3] NCCL INFO Channel 05/0 : 59[3] -> 60[4] via P2P/IPC
+ip-26-0-168-34:272156:272570 [6] NCCL INFO Channel 13/0 : 118[6] -> 119[7] via P2P/IPC
+ip-26-0-161-153:2636085:2636509 [1] NCCL INFO Channel 01/0 : 41[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636085:2636509 [1] NCCL INFO Channel 09/0 : 41[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272156:272570 [6] NCCL INFO Channel 14/0 : 118[6] -> 119[7] via P2P/IPC
+ip-26-0-161-153:2636085:2636509 [1] NCCL INFO Channel 01/0 : 49[1] -> 57[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636087:2636506 [3] NCCL INFO Channel 02/0 : 51[3] -> 52[4] via P2P/IPC
+ip-26-0-161-178:2815968:2816394 [5] NCCL INFO Channel 01/0 : 61[5] -> 62[6] via P2P/IPC
+ip-26-0-161-142:62882:63300 [1] NCCL INFO Channel 09/0 : 25[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815965:2816395 [2] NCCL INFO Channel 10/0 : 58[2] -> 59[3] via P2P/IPC
+ip-26-0-168-34:272155:272574 [5] NCCL INFO Channel 05/0 : 109[5] -> 117[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62882:63300 [1] NCCL INFO Channel 09/0 : 41[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815966:2816396 [3] NCCL INFO Channel 06/0 : 59[3] -> 60[4] via P2P/IPC
+ip-26-0-168-34:272156:272570 [6] NCCL INFO Channel 06/0 : 110[6] -> 118[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272155:272574 [5] NCCL INFO Channel 13/0 : 109[5] -> 117[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32705:33128 [1] NCCL INFO Channel 09/0 : 25[1] -> 57[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272156:272570 [6] NCCL INFO Channel 14/0 : 110[6] -> 118[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272155:272574 [5] NCCL INFO Channel 05/0 : 117[5] -> 125[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272156:272570 [6] NCCL INFO Channel 06/0 : 118[6] -> 126[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636084:2636511 [0] NCCL INFO Channel 06/0 : 48[0] -> 55[7] via P2P/IPC
+ip-26-0-161-178:2815968:2816394 [5] NCCL INFO Channel 02/0 : 61[5] -> 62[6] via P2P/IPC
+ip-26-0-168-30:235653:236067 [5] NCCL INFO Channel 13/0 : 93[5] -> 109[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636088:2636510 [4] NCCL INFO Channel 01/0 : 52[4] -> 53[5] via P2P/IPC
+ip-26-0-161-153:2636087:2636506 [3] NCCL INFO Channel 03/0 : 51[3] -> 52[4] via P2P/IPC
+ip-26-0-168-30:235654:236070 [6] NCCL INFO Channel 14/0 : 94[6] -> 110[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272155:272574 [5] NCCL INFO Channel 05/0 : 101[5] -> 117[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-52:2722050:2722471 [5] NCCL INFO Channel 13/0 : 61[5] -> 125[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272156:272570 [6] NCCL INFO Channel 06/0 : 102[6] -> 118[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-52:2722051:2722465 [6] NCCL INFO Channel 14/0 : 62[6] -> 126[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235653:236067 [5] NCCL INFO Channel 13/0 : 109[5] -> 93[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-52:2722050:2722471 [5] NCCL INFO Channel 13/0 : 125[5] -> 61[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815965:2816395 [2] NCCL INFO Channel 12/0 : 58[2] -> 59[3] via P2P/IPC
+ip-26-0-161-178:2815966:2816396 [3] NCCL INFO Channel 07/0 : 59[3] -> 60[4] via P2P/IPC
+ip-26-0-165-202:153567:153979 [5] NCCL INFO Channel 13/0 : 61[5] -> 93[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235654:236070 [6] NCCL INFO Channel 14/0 : 110[6] -> 94[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-52:2722051:2722465 [6] NCCL INFO Channel 14/0 : 126[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272155:272574 [5] NCCL INFO Channel 05/0 : 117[5] -> 101[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153568:153982 [6] NCCL INFO Channel 14/0 : 62[6] -> 94[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272156:272570 [6] NCCL INFO Channel 06/0 : 118[6] -> 102[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557271:1557690 [5] NCCL INFO Channel 05/0 : 69[5] -> 101[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557272:1557687 [6] NCCL INFO Channel 06/0 : 70[6] -> 102[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557271:1557690 [5] NCCL INFO Channel 05/0 : 101[5] -> 69[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815968:2816394 [5] NCCL INFO Channel 03/0 : 61[5] -> 62[6] via P2P/IPC
+ip-26-0-165-59:1557272:1557687 [6] NCCL INFO Channel 06/0 : 102[6] -> 70[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636084:2636511 [0] NCCL INFO Channel 07/0 : 48[0] -> 55[7] via P2P/IPC
+ip-26-0-161-153:2636088:2636510 [4] NCCL INFO Channel 02/0 : 52[4] -> 53[5] via P2P/IPC
+ip-26-0-161-153:2636087:2636506 [3] NCCL INFO Channel 05/0 : 51[3] -> 52[4] via P2P/IPC
+ip-26-0-161-178:2815966:2816396 [3] NCCL INFO Channel 09/0 : 59[3] -> 60[4] via P2P/IPC
+ip-26-0-161-178:2815965:2816395 [2] NCCL INFO Channel 13/0 : 58[2] -> 59[3] via P2P/IPC
+ip-26-0-161-178:2815968:2816394 [5] NCCL INFO Channel 04/0 : 61[5] -> 62[6] via P2P/IPC
+ip-26-0-161-178:2815966:2816396 [3] NCCL INFO Channel 10/0 : 59[3] -> 60[4] via P2P/IPC
+ip-26-0-161-178:2815965:2816395 [2] NCCL INFO Channel 14/0 : 58[2] -> 59[3] via P2P/IPC
+ip-26-0-161-153:2636088:2636510 [4] NCCL INFO Channel 03/0 : 52[4] -> 53[5] via P2P/IPC
+ip-26-0-168-34:272157:272572 [7] NCCL INFO Channel 07/0 : 111[7] -> 119[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636087:2636506 [3] NCCL INFO Channel 06/0 : 51[3] -> 52[4] via P2P/IPC
+ip-26-0-168-34:272150:272576 [0] NCCL INFO Channel 00/0 : 104[0] -> 112[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272157:272572 [7] NCCL INFO Channel 15/0 : 111[7] -> 119[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272157:272572 [7] NCCL INFO Channel 07/0 : 119[7] -> 127[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272150:272576 [0] NCCL INFO Channel 08/0 : 104[0] -> 112[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815968:2816394 [5] NCCL INFO Channel 05/0 : 61[5] -> 62[6] via P2P/IPC
+ip-26-0-168-34:272150:272576 [0] NCCL INFO Channel 00/0 : 112[0] -> 120[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272157:272572 [7] NCCL INFO Channel 07/0 : 103[7] -> 119[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722052:2722464 [7] NCCL INFO Channel 15/0 : 63[7] -> 127[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815966:2816396 [3] NCCL INFO Channel 11/0 : 59[3] -> 60[4] via P2P/IPC
+ip-26-0-168-52:2722052:2722464 [7] NCCL INFO Channel 15/0 : 127[7] -> 63[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636088:2636510 [4] NCCL INFO Channel 04/0 : 52[4] -> 53[5] via P2P/IPC
+ip-26-0-161-178:2815965:2816395 [2] NCCL INFO Channel 15/0 : 58[2] -> 59[3] via P2P/IPC
+ip-26-0-168-34:272150:272576 [0] NCCL INFO Channel 00/0 : 96[0] -> 112[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636087:2636506 [3] NCCL INFO Channel 07/0 : 51[3] -> 52[4] via P2P/IPC
+ip-26-0-168-34:272157:272572 [7] NCCL INFO Channel 07/0 : 119[7] -> 103[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722045:2722470 [0] NCCL INFO Channel 08/0 : 56[0] -> 120[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-52:2722045:2722470 [0] NCCL INFO Channel 08/0 : 120[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815967:2816399 [4] NCCL INFO Channel 01/0 : 60[4] -> 61[5] via P2P/IPC
+ip-26-0-161-153:2636084:2636511 [0] NCCL INFO Channel 13/0 : 48[0] -> 55[7] via P2P/IPC
+ip-26-0-168-34:272150:272576 [0] NCCL INFO Channel 00/0 : 112[0] -> 96[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815968:2816394 [5] NCCL INFO Channel 07/0 : 61[5] -> 62[6] via P2P/IPC
+ip-26-0-165-59:1557273:1557685 [7] NCCL INFO Channel 07/0 : 71[7] -> 103[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235648:236064 [0] NCCL INFO Channel 08/0 : 88[0] -> 104[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636088:2636510 [4] NCCL INFO Channel 06/0 : 52[4] -> 53[5] via P2P/IPC
+ip-26-0-168-30:235648:236064 [0] NCCL INFO Channel 08/0 : 104[0] -> 88[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815966:2816396 [3] NCCL INFO Channel 13/0 : 59[3] -> 60[4] via P2P/IPC
+ip-26-0-168-30:235655:236069 [7] NCCL INFO Channel 15/0 : 95[7] -> 111[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636087:2636506 [3] NCCL INFO Channel 09/0 : 51[3] -> 52[4] via P2P/IPC
+ip-26-0-165-202:153562:153983 [0] NCCL INFO Channel 08/0 : 56[0] -> 88[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815967:2816399 [4] NCCL INFO Channel 02/0 : 60[4] -> 61[5] via P2P/IPC
+ip-26-0-165-59:1557266:1557689 [0] NCCL INFO Channel 00/0 : 64[0] -> 96[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557273:1557685 [7] NCCL INFO Channel 07/0 : 103[7] -> 71[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815968:2816394 [5] NCCL INFO Channel 09/0 : 61[5] -> 62[6] via P2P/IPC
+ip-26-0-161-178:2815964:2816401 [1] NCCL INFO Channel 01/0 : 49[1] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557266:1557689 [0] NCCL INFO Channel 00/0 : 96[0] -> 64[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235655:236069 [7] NCCL INFO Channel 15/0 : 111[7] -> 95[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815963:2816400 [0] NCCL INFO Channel 00/0 : 48[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815970:2816397 [7] NCCL INFO Channel 07/0 : 55[7] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815966:2816396 [3] NCCL INFO Channel 14/0 : 59[3] -> 60[4] via P2P/IPC
+ip-26-0-161-178:2815964:2816401 [1] NCCL INFO Channel 09/0 : 25[1] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636084:2636511 [0] NCCL INFO Channel 14/0 : 48[0] -> 55[7] via P2P/IPC
+ip-26-0-161-178:2815964:2816401 [1] NCCL INFO Channel 09/0 : 57[1] -> 89[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636085:2636509 [1] NCCL INFO Channel 01/0 : 33[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815967:2816399 [4] NCCL INFO Channel 03/0 : 60[4] -> 61[5] via P2P/IPC
+ip-26-0-161-178:2815968:2816394 [5] NCCL INFO Channel 10/0 : 61[5] -> 62[6] via P2P/IPC
+ip-26-0-161-153:2636085:2636509 [1] NCCL INFO Channel 01/0 : 49[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96564:97009 [1] NCCL INFO Channel 01/0 : 33[1] -> 65[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32705:33128 [1] NCCL INFO Channel 09/0 : 57[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636088:2636510 [4] NCCL INFO Channel 07/0 : 52[4] -> 53[5] via P2P/IPC
+ip-26-0-161-178:2815966:2816396 [3] NCCL INFO Channel 15/0 : 59[3] -> 60[4] via P2P/IPC
+ip-26-0-161-178:2815964:2816401 [1] NCCL INFO Channel 09/0 : 121[1] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815964:2816401 [1] NCCL INFO Channel 09/0 : 57[1] -> 121[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96564:97009 [1] NCCL INFO Channel 01/0 : 65[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636087:2636506 [3] NCCL INFO Channel 10/0 : 51[3] -> 52[4] via P2P/IPC
+ip-26-0-161-78:114513:114933 [1] NCCL INFO Channel 01/0 : 1[1] -> 65[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114513:114933 [1] NCCL INFO Channel 01/0 : 65[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815964:2816401 [1] NCCL INFO Channel 09/0 : 89[1] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815967:2816399 [4] NCCL INFO Channel 04/0 : 60[4] -> 61[5] via P2P/IPC
+ip-26-0-161-178:2815968:2816394 [5] NCCL INFO Channel 11/0 : 61[5] -> 62[6] via P2P/IPC
+ip-26-0-161-178:2815964:2816401 [1] NCCL INFO Channel 09/0 : 57[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-52:2722046:2722466 [1] NCCL INFO Channel 01/0 : 121[1] -> 113[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636084:2636511 [0] NCCL INFO Channel 15/0 : 48[0] -> 55[7] via P2P/IPC
+ip-26-0-161-78:114513:114933 [1] NCCL INFO Channel 01/0 : 97[1] -> 65[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124161:124582 [1] NCCL INFO Channel 09/0 : 9[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114513:114933 [1] NCCL INFO Channel 01/0 : 65[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815965:2816395 [2] NCCL INFO Channel 02/0 : 50[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815967:2816399 [4] NCCL INFO Channel 06/0 : 60[4] -> 61[5] via P2P/IPC
+ip-26-0-161-178:2815968:2816394 [5] NCCL INFO Channel 12/0 : 61[5] -> 62[6] via P2P/IPC
+ip-26-0-161-153:2636088:2636510 [4] NCCL INFO Channel 09/0 : 52[4] -> 53[5] via P2P/IPC
+ip-26-0-161-123:32705:33128 [1] NCCL INFO Channel 09/0 : 41[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32705:33128 [1] NCCL INFO Channel 09/0 : 25[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636087:2636506 [3] NCCL INFO Channel 11/0 : 51[3] -> 52[4] via P2P/IPC
+ip-26-0-161-178:2815967:2816399 [4] NCCL INFO Channel 07/0 : 60[4] -> 61[5] via P2P/IPC
+ip-26-0-161-178:2815968:2816394 [5] NCCL INFO Channel 13/0 : 61[5] -> 62[6] via P2P/IPC
+ip-26-0-161-142:62882:63300 [1] NCCL INFO Channel 01/0 : 49[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62882:63300 [1] NCCL INFO Channel 09/0 : 49[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364903:1365334 [1] NCCL INFO Channel 01/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62882:63300 [1] NCCL INFO Channel 09/0 : 41[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364903:1365334 [1] NCCL INFO Channel 09/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364903:1365334 [1] NCCL INFO Channel 09/0 : 9[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815967:2816399 [4] NCCL INFO Channel 09/0 : 60[4] -> 61[5] via P2P/IPC
+ip-26-0-165-202:153569:153981 [7] NCCL INFO Channel 15/0 : 63[7] -> 95[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815968:2816394 [5] NCCL INFO Channel 15/0 : 61[5] -> 62[6] via P2P/IPC
+ip-26-0-161-153:2636088:2636510 [4] NCCL INFO Channel 10/0 : 52[4] -> 53[5] via P2P/IPC
+ip-26-0-165-59:1557267:1557683 [1] NCCL INFO Channel 01/0 : 113[1] -> 97[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636087:2636506 [3] NCCL INFO Channel 13/0 : 51[3] -> 52[4] via P2P/IPC
+ip-26-0-161-78:114513:114933 [1] NCCL INFO Channel 09/0 : 73[1] -> 65[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557267:1557683 [1] NCCL INFO Channel 01/0 : 97[1] -> 81[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96564:97009 [1] NCCL INFO Channel 01/0 : 49[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815967:2816399 [4] NCCL INFO Channel 10/0 : 60[4] -> 61[5] via P2P/IPC
+ip-26-0-165-202:153563:153986 [1] NCCL INFO Channel 09/0 : 89[1] -> 57[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96564:97009 [1] NCCL INFO Channel 01/0 : 33[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32705:33128 [1] NCCL INFO Channel 01/0 : 25[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636089:2636508 [5] NCCL INFO Channel 01/0 : 53[5] -> 54[6] via P2P/IPC
+ip-26-0-168-34:272151:272571 [1] NCCL INFO Channel 01/0 : 121[1] -> 113[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815967:2816399 [4] NCCL INFO Channel 11/0 : 60[4] -> 61[5] via P2P/IPC
+ip-26-0-165-202:153563:153986 [1] NCCL INFO Channel 09/0 : 105[1] -> 89[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272151:272571 [1] NCCL INFO Channel 01/0 : 113[1] -> 105[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557267:1557683 [1] NCCL INFO Channel 09/0 : 105[1] -> 97[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815964:2816401 [1] NCCL INFO Channel 01/0 : 57[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636085:2636509 [1] NCCL INFO Channel 01/0 : 57[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272151:272571 [1] NCCL INFO Channel 09/0 : 113[1] -> 105[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815969:2816398 [6] NCCL INFO Channel 06/0 : 54[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153563:153986 [1] NCCL INFO Channel 09/0 : 89[1] -> 73[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636085:2636509 [1] NCCL INFO Channel 01/0 : 49[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96564:97009 [1] NCCL INFO Channel 09/0 : 41[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636085:2636509 [1] NCCL INFO Channel 09/0 : 49[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96889:97309 [1] NCCL INFO Channel 01/0 : 25[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815967:2816399 [4] NCCL INFO Channel 12/0 : 60[4] -> 61[5] via P2P/IPC
+ip-26-0-161-103:96889:97309 [1] NCCL INFO Channel 01/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636088:2636510 [4] NCCL INFO Channel 11/0 : 52[4] -> 53[5] via P2P/IPC
+ip-26-0-161-103:96889:97309 [1] NCCL INFO Channel 09/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636087:2636506 [3] NCCL INFO Channel 14/0 : 51[3] -> 52[4] via P2P/IPC
+ip-26-0-168-30:235649:236071 [1] NCCL INFO Channel 01/0 : 113[1] -> 105[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235649:236071 [1] NCCL INFO Channel 09/0 : 113[1] -> 105[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153563:153986 [1] NCCL INFO Channel 01/0 : 89[1] -> 81[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235649:236071 [1] NCCL INFO Channel 09/0 : 105[1] -> 97[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630206:1630647 [1] NCCL INFO Channel 01/0 : 81[1] -> 73[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815967:2816399 [4] NCCL INFO Channel 14/0 : 60[4] -> 61[5] via P2P/IPC
+ip-26-0-165-131:1630206:1630647 [1] NCCL INFO Channel 09/0 : 81[1] -> 73[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124161:124582 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-153:2636089:2636508 [5] NCCL INFO Channel 02/0 : 53[5] -> 54[6] via P2P/IPC
+ip-26-0-165-131:1630206:1630647 [1] NCCL INFO Channel 09/0 : 73[1] -> 65[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815967:2816399 [4] NCCL INFO Channel 15/0 : 60[4] -> 61[5] via P2P/IPC
+ip-26-0-160-225:124161:124582 [1] NCCL INFO Channel 08/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-153:2636088:2636510 [4] NCCL INFO Channel 12/0 : 52[4] -> 53[5] via P2P/IPC
+ip-26-0-161-153:2636087:2636506 [3] NCCL INFO Channel 15/0 : 51[3] -> 52[4] via P2P/IPC
+ip-26-0-161-138:96564:97009 [1] NCCL INFO Channel 00/0 : 33[1] -> 32[0] via P2P/IPC
+ip-26-0-165-164:1227464:1227884 [1] NCCL INFO Channel 01/0 : 89[1] -> 81[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815966:2816396 [3] NCCL INFO Channel 03/0 : 51[3] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636090:2636505 [6] NCCL INFO Channel 01/0 : 54[6] -> 55[7] via P2P/IPC
+ip-26-0-165-164:1227464:1227884 [1] NCCL INFO Channel 01/0 : 81[1] -> 73[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815968:2816394 [5] NCCL INFO Channel 05/0 : 53[5] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32705:33128 [1] NCCL INFO Channel 00/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-161-138:96564:97009 [1] NCCL INFO Channel 08/0 : 33[1] -> 32[0] via P2P/IPC
+ip-26-0-161-178:2815967:2816399 [4] NCCL INFO Channel 04/0 : 52[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227464:1227884 [1] NCCL INFO Channel 09/0 : 81[1] -> 73[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96889:97309 [1] NCCL INFO Channel 00/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-161-178:2815964:2816401 [1] NCCL INFO Channel 00/0 : 57[1] -> 56[0] via P2P/IPC
+ip-26-0-161-153:2636089:2636508 [5] NCCL INFO Channel 03/0 : 53[5] -> 54[6] via P2P/IPC
+ip-26-0-160-242:1364903:1365334 [1] NCCL INFO Channel 00/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-161-123:32705:33128 [1] NCCL INFO Channel 08/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-168-34:272151:272571 [1] NCCL INFO Channel 00/0 : 113[1] -> 112[0] via P2P/IPC
+ip-26-0-161-103:96889:97309 [1] NCCL INFO Channel 08/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-161-178:2815964:2816401 [1] NCCL INFO Channel 08/0 : 57[1] -> 56[0] via P2P/IPC
+ip-26-0-168-30:235649:236071 [1] NCCL INFO Channel 00/0 : 105[1] -> 104[0] via P2P/IPC
+ip-26-0-168-34:272151:272571 [1] NCCL INFO Channel 08/0 : 113[1] -> 112[0] via P2P/IPC
+ip-26-0-165-59:1557267:1557683 [1] NCCL INFO Channel 00/0 : 97[1] -> 96[0] via P2P/IPC
+ip-26-0-161-153:2636088:2636510 [4] NCCL INFO Channel 14/0 : 52[4] -> 53[5] via P2P/IPC
+ip-26-0-168-30:235649:236071 [1] NCCL INFO Channel 08/0 : 105[1] -> 104[0] via P2P/IPC
+ip-26-0-165-59:1557267:1557683 [1] NCCL INFO Channel 08/0 : 97[1] -> 96[0] via P2P/IPC
+ip-26-0-161-153:2636090:2636505 [6] NCCL INFO Channel 02/0 : 54[6] -> 55[7] via P2P/IPC
+ip-26-0-161-153:2636086:2636507 [2] NCCL INFO Channel 02/0 : 42[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636089:2636508 [5] NCCL INFO Channel 04/0 : 53[5] -> 54[6] via P2P/IPC
+ip-26-0-161-153:2636086:2636507 [2] NCCL INFO Channel 10/0 : 42[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636086:2636507 [2] NCCL INFO Channel 02/0 : 50[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62882:63300 [1] NCCL INFO Channel 00/0 : 41[1] -> 40[0] via P2P/IPC
+ip-26-0-165-131:1630206:1630647 [1] NCCL INFO Channel 00/0 : 73[1] -> 72[0] via P2P/IPC
+ip-26-0-165-202:153563:153986 [1] NCCL INFO Channel 00/0 : 89[1] -> 88[0] via P2P/IPC
+ip-26-0-161-153:2636088:2636510 [4] NCCL INFO Channel 15/0 : 52[4] -> 53[5] via P2P/IPC
+ip-26-0-165-164:1227464:1227884 [1] NCCL INFO Channel 00/0 : 81[1] -> 80[0] via P2P/IPC
+ip-26-0-161-142:62883:63301 [2] NCCL INFO Channel 10/0 : 26[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636086:2636507 [2] NCCL INFO Channel 02/0 : 34[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62882:63300 [1] NCCL INFO Channel 08/0 : 41[1] -> 40[0] via P2P/IPC
+ip-26-0-161-178:2815965:2816395 [2] NCCL INFO Channel 10/0 : 26[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636090:2636505 [6] NCCL INFO Channel 03/0 : 54[6] -> 55[7] via P2P/IPC
+ip-26-0-165-202:153563:153986 [1] NCCL INFO Channel 08/0 : 89[1] -> 88[0] via P2P/IPC
+ip-26-0-165-131:1630206:1630647 [1] NCCL INFO Channel 08/0 : 73[1] -> 72[0] via P2P/IPC
+ip-26-0-161-178:2815965:2816395 [2] NCCL INFO Channel 10/0 : 58[2] -> 90[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227464:1227884 [1] NCCL INFO Channel 08/0 : 81[1] -> 80[0] via P2P/IPC
+ip-26-0-161-142:62883:63301 [2] NCCL INFO Channel 10/0 : 42[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636089:2636508 [5] NCCL INFO Channel 05/0 : 53[5] -> 54[6] via P2P/IPC
+ip-26-0-161-153:2636086:2636507 [2] NCCL INFO Channel 02/0 : 50[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-52:2722046:2722466 [1] NCCL INFO Channel 00/0 : 121[1] -> 120[0] via P2P/IPC
+ip-26-0-160-242:1364903:1365334 [1] NCCL INFO Channel 08/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-165-202:153564:153984 [2] NCCL INFO Channel 10/0 : 90[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114513:114933 [1] NCCL INFO Channel 00/0 : 65[1] -> 64[0] via P2P/IPC
+ip-26-0-161-138:96565:97004 [2] NCCL INFO Channel 02/0 : 34[2] -> 66[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636090:2636505 [6] NCCL INFO Channel 04/0 : 54[6] -> 55[7] via P2P/IPC
+ip-26-0-161-153:2636085:2636509 [1] NCCL INFO Channel 00/0 : 49[1] -> 48[0] via P2P/IPC
+ip-26-0-161-78:114513:114933 [1] NCCL INFO Channel 08/0 : 65[1] -> 64[0] via P2P/IPC
+ip-26-0-161-153:2636089:2636508 [5] NCCL INFO Channel 07/0 : 53[5] -> 54[6] via P2P/IPC
+ip-26-0-161-138:96565:97004 [2] NCCL INFO Channel 02/0 : 66[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636087:2636506 [3] NCCL INFO Channel 03/0 : 43[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114514:114932 [2] NCCL INFO Channel 02/0 : 2[2] -> 66[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636087:2636506 [3] NCCL INFO Channel 11/0 : 43[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114514:114932 [2] NCCL INFO Channel 02/0 : 66[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636087:2636506 [3] NCCL INFO Channel 03/0 : 51[3] -> 59[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114514:114932 [2] NCCL INFO Channel 02/0 : 98[2] -> 66[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:124162:124587 [2] NCCL INFO Channel 10/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114514:114932 [2] NCCL INFO Channel 02/0 : 66[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636090:2636505 [6] NCCL INFO Channel 05/0 : 54[6] -> 55[7] via P2P/IPC
+ip-26-0-161-153:2636085:2636509 [1] NCCL INFO Channel 08/0 : 49[1] -> 48[0] via P2P/IPC
+ip-26-0-168-52:2722046:2722466 [1] NCCL INFO Channel 08/0 : 121[1] -> 120[0] via P2P/IPC
+ip-26-0-161-178:2815966:2816396 [3] NCCL INFO Channel 11/0 : 27[3] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636087:2636506 [3] NCCL INFO Channel 03/0 : 35[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636089:2636508 [5] NCCL INFO Channel 09/0 : 53[5] -> 54[6] via P2P/IPC
+ip-26-0-161-178:2815966:2816396 [3] NCCL INFO Channel 11/0 : 59[3] -> 91[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96565:97004 [2] NCCL INFO Channel 02/0 : 50[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114514:114932 [2] NCCL INFO Channel 10/0 : 74[2] -> 66[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636087:2636506 [3] NCCL INFO Channel 03/0 : 51[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32706:33122 [2] NCCL INFO Channel 10/0 : 26[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96566:97010 [3] NCCL INFO Channel 03/0 : 35[3] -> 67[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96565:97004 [2] NCCL INFO Channel 02/0 : 34[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636090:2636505 [6] NCCL INFO Channel 06/0 : 54[6] -> 55[7] via P2P/IPC
+ip-26-0-161-138:96566:97010 [3] NCCL INFO Channel 03/0 : 67[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815965:2816395 [2] NCCL INFO Channel 10/0 : 122[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32706:33122 [2] NCCL INFO Channel 10/0 : 58[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153565:153985 [3] NCCL INFO Channel 11/0 : 91[3] -> 59[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636089:2636508 [5] NCCL INFO Channel 10/0 : 53[5] -> 54[6] via P2P/IPC
+ip-26-0-161-78:114515:114936 [3] NCCL INFO Channel 03/0 : 3[3] -> 67[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815965:2816395 [2] NCCL INFO Channel 10/0 : 58[2] -> 122[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636086:2636507 [2] NCCL INFO Channel 02/0 : 58[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114515:114936 [3] NCCL INFO Channel 03/0 : 67[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96565:97004 [2] NCCL INFO Channel 10/0 : 42[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636086:2636507 [2] NCCL INFO Channel 02/0 : 50[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96890:97313 [2] NCCL INFO Channel 02/0 : 26[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636086:2636507 [2] NCCL INFO Channel 10/0 : 50[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96890:97313 [2] NCCL INFO Channel 02/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815965:2816395 [2] NCCL INFO Channel 10/0 : 90[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636090:2636505 [6] NCCL INFO Channel 09/0 : 54[6] -> 55[7] via P2P/IPC
+ip-26-0-161-78:114515:114936 [3] NCCL INFO Channel 03/0 : 99[3] -> 67[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815965:2816395 [2] NCCL INFO Channel 10/0 : 58[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-52:2722047:2722468 [2] NCCL INFO Channel 02/0 : 122[2] -> 114[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96890:97313 [2] NCCL INFO Channel 10/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636089:2636508 [5] NCCL INFO Channel 11/0 : 53[5] -> 54[6] via P2P/IPC
+ip-26-0-160-225:124163:124588 [3] NCCL INFO Channel 11/0 : 11[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114515:114936 [3] NCCL INFO Channel 03/0 : 67[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153564:153984 [2] NCCL INFO Channel 10/0 : 106[2] -> 90[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557268:1557686 [2] NCCL INFO Channel 02/0 : 114[2] -> 98[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153564:153984 [2] NCCL INFO Channel 10/0 : 90[2] -> 74[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815965:2816395 [2] NCCL INFO Channel 02/0 : 58[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636090:2636505 [6] NCCL INFO Channel 10/0 : 54[6] -> 55[7] via P2P/IPC
+ip-26-0-165-59:1557269:1557684 [3] NCCL INFO Channel 03/0 : 115[3] -> 99[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557268:1557686 [2] NCCL INFO Channel 02/0 : 98[2] -> 82[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32706:33122 [2] NCCL INFO Channel 10/0 : 42[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114515:114936 [3] NCCL INFO Channel 11/0 : 75[3] -> 67[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636089:2636508 [5] NCCL INFO Channel 12/0 : 53[5] -> 54[6] via P2P/IPC
+ip-26-0-165-59:1557269:1557684 [3] NCCL INFO Channel 03/0 : 99[3] -> 83[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32706:33122 [2] NCCL INFO Channel 10/0 : 26[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235650:236065 [2] NCCL INFO Channel 02/0 : 114[2] -> 106[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272152:272575 [2] NCCL INFO Channel 02/0 : 122[2] -> 114[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153564:153984 [2] NCCL INFO Channel 02/0 : 90[2] -> 82[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235650:236065 [2] NCCL INFO Channel 10/0 : 114[2] -> 106[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630207:1630648 [2] NCCL INFO Channel 02/0 : 82[2] -> 74[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272153:272573 [3] NCCL INFO Channel 03/0 : 123[3] -> 115[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557268:1557686 [2] NCCL INFO Channel 10/0 : 106[2] -> 98[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235650:236065 [2] NCCL INFO Channel 10/0 : 106[2] -> 98[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630207:1630648 [2] NCCL INFO Channel 10/0 : 82[2] -> 74[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32706:33122 [2] NCCL INFO Channel 02/0 : 26[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227465:1227881 [2] NCCL INFO Channel 02/0 : 90[2] -> 82[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636090:2636505 [6] NCCL INFO Channel 11/0 : 54[6] -> 55[7] via P2P/IPC
+ip-26-0-168-34:272152:272575 [2] NCCL INFO Channel 02/0 : 114[2] -> 106[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62884:63299 [3] NCCL INFO Channel 11/0 : 27[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62883:63301 [2] NCCL INFO Channel 02/0 : 50[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557269:1557684 [3] NCCL INFO Channel 11/0 : 107[3] -> 99[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364904:1365335 [2] NCCL INFO Channel 02/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630207:1630648 [2] NCCL INFO Channel 10/0 : 74[2] -> 66[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227466:1227887 [3] NCCL INFO Channel 03/0 : 91[3] -> 83[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636089:2636508 [5] NCCL INFO Channel 13/0 : 53[5] -> 54[6] via P2P/IPC
+ip-26-0-168-34:272153:272573 [3] NCCL INFO Channel 03/0 : 115[3] -> 107[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272152:272575 [2] NCCL INFO Channel 10/0 : 114[2] -> 106[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364904:1365335 [2] NCCL INFO Channel 10/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227465:1227881 [2] NCCL INFO Channel 02/0 : 82[2] -> 74[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62883:63301 [2] NCCL INFO Channel 10/0 : 50[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272153:272573 [3] NCCL INFO Channel 11/0 : 115[3] -> 107[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227466:1227887 [3] NCCL INFO Channel 03/0 : 83[3] -> 75[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227465:1227881 [2] NCCL INFO Channel 10/0 : 82[2] -> 74[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364904:1365335 [2] NCCL INFO Channel 10/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227466:1227887 [3] NCCL INFO Channel 11/0 : 83[3] -> 75[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62884:63299 [3] NCCL INFO Channel 11/0 : 43[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62883:63301 [2] NCCL INFO Channel 10/0 : 42[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32707:33124 [3] NCCL INFO Channel 11/0 : 27[3] -> 59[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636090:2636505 [6] NCCL INFO Channel 12/0 : 54[6] -> 55[7] via P2P/IPC
+ip-26-0-161-153:2636089:2636508 [5] NCCL INFO Channel 15/0 : 53[5] -> 54[6] via P2P/IPC
+ip-26-0-161-123:32707:33124 [3] NCCL INFO Channel 11/0 : 59[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815966:2816396 [3] NCCL INFO Channel 11/0 : 123[3] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815966:2816396 [3] NCCL INFO Channel 11/0 : 59[3] -> 123[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815966:2816396 [3] NCCL INFO Channel 11/0 : 91[3] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-52:2722048:2722469 [3] NCCL INFO Channel 03/0 : 123[3] -> 115[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96566:97010 [3] NCCL INFO Channel 03/0 : 51[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815966:2816396 [3] NCCL INFO Channel 11/0 : 59[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636090:2636505 [6] NCCL INFO Channel 13/0 : 54[6] -> 55[7] via P2P/IPC
+ip-26-0-161-138:96566:97010 [3] NCCL INFO Channel 03/0 : 35[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32706:33122 [2] NCCL INFO Channel 00/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-168-52:2722047:2722468 [2] NCCL INFO Channel 00/0 : 122[2] -> 121[1] via P2P/IPC
+ip-26-0-165-202:153565:153985 [3] NCCL INFO Channel 11/0 : 107[3] -> 91[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272152:272575 [2] NCCL INFO Channel 00/0 : 114[2] -> 113[1] via P2P/IPC
+ip-26-0-165-202:153564:153984 [2] NCCL INFO Channel 00/0 : 90[2] -> 89[1] via P2P/IPC
+ip-26-0-161-153:2636090:2636505 [6] NCCL INFO Channel 14/0 : 54[6] -> 55[7] via P2P/IPC
+ip-26-0-161-178:2815966:2816396 [3] NCCL INFO Channel 03/0 : 59[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153565:153985 [3] NCCL INFO Channel 11/0 : 91[3] -> 75[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32707:33124 [3] NCCL INFO Channel 11/0 : 43[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636088:2636510 [4] NCCL INFO Channel 04/0 : 44[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32707:33124 [3] NCCL INFO Channel 11/0 : 27[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32706:33122 [2] NCCL INFO Channel 01/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-161-153:2636087:2636506 [3] NCCL INFO Channel 03/0 : 59[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96566:97010 [3] NCCL INFO Channel 11/0 : 43[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557268:1557686 [2] NCCL INFO Channel 00/0 : 98[2] -> 97[1] via P2P/IPC
+ip-26-0-165-131:1630207:1630648 [2] NCCL INFO Channel 00/0 : 74[2] -> 73[1] via P2P/IPC
+ip-26-0-161-153:2636088:2636510 [4] NCCL INFO Channel 12/0 : 44[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96891:97311 [3] NCCL INFO Channel 03/0 : 27[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272152:272575 [2] NCCL INFO Channel 01/0 : 114[2] -> 113[1] via P2P/IPC
+ip-26-0-160-242:1364904:1365335 [2] NCCL INFO Channel 00/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-161-153:2636087:2636506 [3] NCCL INFO Channel 03/0 : 51[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114514:114932 [2] NCCL INFO Channel 00/0 : 66[2] -> 65[1] via P2P/IPC
+ip-26-0-161-142:62883:63301 [2] NCCL INFO Channel 00/0 : 42[2] -> 41[1] via P2P/IPC
+ip-26-0-161-153:2636088:2636510 [4] NCCL INFO Channel 04/0 : 52[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153564:153984 [2] NCCL INFO Channel 01/0 : 90[2] -> 89[1] via P2P/IPC
+ip-26-0-161-103:96890:97313 [2] NCCL INFO Channel 00/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-160-225:124162:124587 [2] NCCL INFO Channel 00/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-153:2636087:2636506 [3] NCCL INFO Channel 11/0 : 51[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235651:236068 [3] NCCL INFO Channel 03/0 : 115[3] -> 107[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96891:97311 [3] NCCL INFO Channel 03/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96891:97311 [3] NCCL INFO Channel 11/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557268:1557686 [2] NCCL INFO Channel 01/0 : 98[2] -> 97[1] via P2P/IPC
+ip-26-0-161-123:32706:33122 [2] NCCL INFO Channel 08/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-168-30:235651:236068 [3] NCCL INFO Channel 11/0 : 115[3] -> 107[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96565:97004 [2] NCCL INFO Channel 00/0 : 34[2] -> 33[1] via P2P/IPC
+ip-26-0-161-142:62884:63299 [3] NCCL INFO Channel 03/0 : 51[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630207:1630648 [2] NCCL INFO Channel 01/0 : 74[2] -> 73[1] via P2P/IPC
+ip-26-0-160-242:1364904:1365335 [2] NCCL INFO Channel 01/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-168-34:272152:272575 [2] NCCL INFO Channel 08/0 : 114[2] -> 113[1] via P2P/IPC
+ip-26-0-161-78:114514:114932 [2] NCCL INFO Channel 01/0 : 66[2] -> 65[1] via P2P/IPC
+ip-26-0-168-30:235651:236068 [3] NCCL INFO Channel 11/0 : 107[3] -> 99[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153564:153984 [2] NCCL INFO Channel 08/0 : 90[2] -> 89[1] via P2P/IPC
+ip-26-0-161-142:62883:63301 [2] NCCL INFO Channel 01/0 : 42[2] -> 41[1] via P2P/IPC
+ip-26-0-161-123:32707:33124 [3] NCCL INFO Channel 03/0 : 27[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96890:97313 [2] NCCL INFO Channel 01/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-160-225:124162:124587 [2] NCCL INFO Channel 01/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-142:62884:63299 [3] NCCL INFO Channel 11/0 : 51[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364905:1365328 [3] NCCL INFO Channel 03/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62885:63304 [4] NCCL INFO Channel 12/0 : 28[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364905:1365328 [3] NCCL INFO Channel 11/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96565:97004 [2] NCCL INFO Channel 01/0 : 34[2] -> 33[1] via P2P/IPC
+ip-26-0-165-59:1557268:1557686 [2] NCCL INFO Channel 08/0 : 98[2] -> 97[1] via P2P/IPC
+ip-26-0-161-142:62884:63299 [3] NCCL INFO Channel 11/0 : 43[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815967:2816399 [4] NCCL INFO Channel 12/0 : 28[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32706:33122 [2] NCCL INFO Channel 09/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-161-153:2636088:2636510 [4] NCCL INFO Channel 04/0 : 36[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630207:1630648 [2] NCCL INFO Channel 08/0 : 74[2] -> 73[1] via P2P/IPC
+ip-26-0-161-78:114514:114932 [2] NCCL INFO Channel 08/0 : 66[2] -> 65[1] via P2P/IPC
+ip-26-0-168-34:272152:272575 [2] NCCL INFO Channel 09/0 : 114[2] -> 113[1] via P2P/IPC
+ip-26-0-161-178:2815967:2816399 [4] NCCL INFO Channel 12/0 : 60[4] -> 92[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364905:1365328 [3] NCCL INFO Channel 11/0 : 11[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62883:63301 [2] NCCL INFO Channel 08/0 : 42[2] -> 41[1] via P2P/IPC
+ip-26-0-161-103:96890:97313 [2] NCCL INFO Channel 08/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-161-123:32708:33126 [4] NCCL INFO Channel 12/0 : 28[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:124162:124587 [2] NCCL INFO Channel 08/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-142:62885:63304 [4] NCCL INFO Channel 12/0 : 44[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722047:2722468 [2] NCCL INFO Channel 01/0 : 122[2] -> 121[1] via P2P/IPC
+ip-26-0-161-138:96565:97004 [2] NCCL INFO Channel 08/0 : 34[2] -> 33[1] via P2P/IPC
+ip-26-0-165-59:1557268:1557686 [2] NCCL INFO Channel 09/0 : 98[2] -> 97[1] via P2P/IPC
+ip-26-0-161-153:2636088:2636510 [4] NCCL INFO Channel 04/0 : 52[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630207:1630648 [2] NCCL INFO Channel 09/0 : 74[2] -> 73[1] via P2P/IPC
+ip-26-0-161-78:114514:114932 [2] NCCL INFO Channel 09/0 : 66[2] -> 65[1] via P2P/IPC
+ip-26-0-161-138:96567:97008 [4] NCCL INFO Channel 04/0 : 36[4] -> 68[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153565:153985 [3] NCCL INFO Channel 03/0 : 91[3] -> 83[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630208:1630650 [3] NCCL INFO Channel 03/0 : 83[3] -> 75[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:124162:124587 [2] NCCL INFO Channel 09/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-123:32708:33126 [4] NCCL INFO Channel 12/0 : 60[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62883:63301 [2] NCCL INFO Channel 09/0 : 42[2] -> 41[1] via P2P/IPC
+ip-26-0-161-103:96890:97313 [2] NCCL INFO Channel 09/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-161-178:2815967:2816399 [4] NCCL INFO Channel 12/0 : 124[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630208:1630650 [3] NCCL INFO Channel 11/0 : 83[3] -> 75[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815967:2816399 [4] NCCL INFO Channel 12/0 : 60[4] -> 124[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630208:1630650 [3] NCCL INFO Channel 11/0 : 75[3] -> 67[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-52:2722047:2722468 [2] NCCL INFO Channel 08/0 : 122[2] -> 121[1] via P2P/IPC
+ip-26-0-161-138:96565:97004 [2] NCCL INFO Channel 09/0 : 34[2] -> 33[1] via P2P/IPC
+ip-26-0-161-138:96567:97008 [4] NCCL INFO Channel 04/0 : 68[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114516:114934 [4] NCCL INFO Channel 04/0 : 4[4] -> 68[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235650:236065 [2] NCCL INFO Channel 00/0 : 106[2] -> 105[1] via P2P/IPC
+ip-26-0-161-78:114516:114934 [4] NCCL INFO Channel 04/0 : 68[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815967:2816399 [4] NCCL INFO Channel 12/0 : 92[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815967:2816399 [4] NCCL INFO Channel 12/0 : 60[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722049:2722467 [4] NCCL INFO Channel 04/0 : 124[4] -> 116[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722047:2722468 [2] NCCL INFO Channel 09/0 : 122[2] -> 121[1] via P2P/IPC
+ip-26-0-168-30:235650:236065 [2] NCCL INFO Channel 01/0 : 106[2] -> 105[1] via P2P/IPC
+ip-26-0-161-78:114516:114934 [4] NCCL INFO Channel 04/0 : 100[4] -> 68[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364904:1365335 [2] NCCL INFO Channel 08/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-160-225:124164:124585 [4] NCCL INFO Channel 12/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235650:236065 [2] NCCL INFO Channel 08/0 : 106[2] -> 105[1] via P2P/IPC
+ip-26-0-161-178:2815965:2816395 [2] NCCL INFO Channel 00/0 : 58[2] -> 57[1] via P2P/IPC
+ip-26-0-161-123:32708:33126 [4] NCCL INFO Channel 12/0 : 44[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32708:33126 [4] NCCL INFO Channel 12/0 : 28[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227465:1227881 [2] NCCL INFO Channel 00/0 : 82[2] -> 81[1] via P2P/IPC
+ip-26-0-168-30:235650:236065 [2] NCCL INFO Channel 09/0 : 106[2] -> 105[1] via P2P/IPC
+ip-26-0-161-178:2815965:2816395 [2] NCCL INFO Channel 01/0 : 58[2] -> 57[1] via P2P/IPC
+ip-26-0-161-142:62885:63304 [4] NCCL INFO Channel 04/0 : 52[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227465:1227881 [2] NCCL INFO Channel 01/0 : 82[2] -> 81[1] via P2P/IPC
+ip-26-0-161-142:62885:63304 [4] NCCL INFO Channel 12/0 : 52[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32708:33126 [4] NCCL INFO Channel 04/0 : 28[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62885:63304 [4] NCCL INFO Channel 12/0 : 44[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815965:2816395 [2] NCCL INFO Channel 08/0 : 58[2] -> 57[1] via P2P/IPC
+ip-26-0-165-164:1227465:1227881 [2] NCCL INFO Channel 08/0 : 82[2] -> 81[1] via P2P/IPC
+ip-26-0-160-242:1364904:1365335 [2] NCCL INFO Channel 09/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-165-202:153564:153984 [2] NCCL INFO Channel 09/0 : 90[2] -> 89[1] via P2P/IPC
+ip-26-0-161-153:2636089:2636508 [5] NCCL INFO Channel 05/0 : 45[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153566:153980 [4] NCCL INFO Channel 12/0 : 92[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636089:2636508 [5] NCCL INFO Channel 13/0 : 45[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636090:2636505 [6] NCCL INFO Channel 06/0 : 46[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815965:2816395 [2] NCCL INFO Channel 09/0 : 58[2] -> 57[1] via P2P/IPC
+ip-26-0-165-164:1227465:1227881 [2] NCCL INFO Channel 09/0 : 82[2] -> 81[1] via P2P/IPC
+ip-26-0-161-153:2636089:2636508 [5] NCCL INFO Channel 05/0 : 53[5] -> 61[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636090:2636505 [6] NCCL INFO Channel 14/0 : 46[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153566:153980 [4] NCCL INFO Channel 12/0 : 108[4] -> 92[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815967:2816399 [4] NCCL INFO Channel 04/0 : 60[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364906:1365331 [4] NCCL INFO Channel 04/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364906:1365331 [4] NCCL INFO Channel 12/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153566:153980 [4] NCCL INFO Channel 12/0 : 92[4] -> 76[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32707:33124 [3] NCCL INFO Channel 00/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-160-242:1364906:1365331 [4] NCCL INFO Channel 12/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557269:1557684 [3] NCCL INFO Channel 00/0 : 99[3] -> 98[2] via P2P/IPC
+ip-26-0-161-142:62886:63303 [5] NCCL INFO Channel 13/0 : 29[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-52:2722048:2722469 [3] NCCL INFO Channel 00/0 : 123[3] -> 122[2] via P2P/IPC
+ip-26-0-161-142:62884:63299 [3] NCCL INFO Channel 00/0 : 43[3] -> 42[2] via P2P/IPC
+ip-26-0-161-123:32707:33124 [3] NCCL INFO Channel 02/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-161-142:62887:63305 [6] NCCL INFO Channel 14/0 : 30[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815968:2816394 [5] NCCL INFO Channel 13/0 : 29[5] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96566:97010 [3] NCCL INFO Channel 00/0 : 35[3] -> 34[2] via P2P/IPC
+ip-26-0-165-59:1557269:1557684 [3] NCCL INFO Channel 02/0 : 99[3] -> 98[2] via P2P/IPC
+ip-26-0-165-202:153566:153980 [4] NCCL INFO Channel 04/0 : 92[4] -> 84[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62886:63303 [5] NCCL INFO Channel 13/0 : 45[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815968:2816394 [5] NCCL INFO Channel 13/0 : 61[5] -> 93[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630209:1630651 [4] NCCL INFO Channel 04/0 : 84[4] -> 76[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815969:2816398 [6] NCCL INFO Channel 14/0 : 30[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32709:33129 [5] NCCL INFO Channel 13/0 : 29[5] -> 61[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153565:153985 [3] NCCL INFO Channel 00/0 : 91[3] -> 90[2] via P2P/IPC
+ip-26-0-165-131:1630209:1630651 [4] NCCL INFO Channel 12/0 : 84[4] -> 76[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62884:63299 [3] NCCL INFO Channel 02/0 : 43[3] -> 42[2] via P2P/IPC
+ip-26-0-161-178:2815969:2816398 [6] NCCL INFO Channel 14/0 : 62[6] -> 94[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32707:33124 [3] NCCL INFO Channel 08/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-160-225:124163:124588 [3] NCCL INFO Channel 00/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-142:62887:63305 [6] NCCL INFO Channel 14/0 : 46[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630209:1630651 [4] NCCL INFO Channel 12/0 : 76[4] -> 68[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96568:97007 [5] NCCL INFO Channel 05/0 : 37[5] -> 69[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557269:1557684 [3] NCCL INFO Channel 08/0 : 99[3] -> 98[2] via P2P/IPC
+ip-26-0-161-123:32710:33125 [6] NCCL INFO Channel 14/0 : 30[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96566:97010 [3] NCCL INFO Channel 02/0 : 35[3] -> 34[2] via P2P/IPC
+ip-26-0-165-202:153565:153985 [3] NCCL INFO Channel 02/0 : 91[3] -> 90[2] via P2P/IPC
+ip-26-0-161-123:32709:33129 [5] NCCL INFO Channel 13/0 : 61[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62884:63299 [3] NCCL INFO Channel 08/0 : 43[3] -> 42[2] via P2P/IPC
+ip-26-0-161-123:32707:33124 [3] NCCL INFO Channel 10/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-161-178:2815968:2816394 [5] NCCL INFO Channel 13/0 : 125[5] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630208:1630650 [3] NCCL INFO Channel 00/0 : 75[3] -> 74[2] via P2P/IPC
+ip-26-0-165-202:153567:153979 [5] NCCL INFO Channel 13/0 : 93[5] -> 61[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557269:1557684 [3] NCCL INFO Channel 10/0 : 99[3] -> 98[2] via P2P/IPC
+ip-26-0-161-123:32710:33125 [6] NCCL INFO Channel 14/0 : 62[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96568:97007 [5] NCCL INFO Channel 05/0 : 69[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815968:2816394 [5] NCCL INFO Channel 13/0 : 61[5] -> 125[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124163:124588 [3] NCCL INFO Channel 02/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-78:114517:114935 [5] NCCL INFO Channel 05/0 : 5[5] -> 69[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96566:97010 [3] NCCL INFO Channel 08/0 : 35[3] -> 34[2] via P2P/IPC
+ip-26-0-161-178:2815969:2816398 [6] NCCL INFO Channel 14/0 : 126[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153565:153985 [3] NCCL INFO Channel 08/0 : 91[3] -> 90[2] via P2P/IPC
+ip-26-0-165-202:153568:153982 [6] NCCL INFO Channel 14/0 : 94[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114516:114934 [4] NCCL INFO Channel 04/0 : 68[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62884:63299 [3] NCCL INFO Channel 10/0 : 43[3] -> 42[2] via P2P/IPC
+ip-26-0-161-78:114517:114935 [5] NCCL INFO Channel 05/0 : 69[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815969:2816398 [6] NCCL INFO Channel 14/0 : 62[6] -> 126[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636090:2636505 [6] NCCL INFO Channel 06/0 : 54[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636086:2636507 [2] NCCL INFO Channel 00/0 : 50[2] -> 49[1] via P2P/IPC
+ip-26-0-161-153:2636089:2636508 [5] NCCL INFO Channel 05/0 : 37[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636090:2636505 [6] NCCL INFO Channel 06/0 : 38[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636089:2636508 [5] NCCL INFO Channel 05/0 : 53[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636090:2636505 [6] NCCL INFO Channel 06/0 : 54[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815968:2816394 [5] NCCL INFO Channel 13/0 : 93[5] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96566:97010 [3] NCCL INFO Channel 10/0 : 35[3] -> 34[2] via P2P/IPC
+ip-26-0-165-202:153565:153985 [3] NCCL INFO Channel 10/0 : 91[3] -> 90[2] via P2P/IPC
+ip-26-0-161-178:2815968:2816394 [5] NCCL INFO Channel 13/0 : 61[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96569:97006 [6] NCCL INFO Channel 06/0 : 38[6] -> 70[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114517:114935 [5] NCCL INFO Channel 05/0 : 101[5] -> 69[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124163:124588 [3] NCCL INFO Channel 08/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-165-59:1557270:1557688 [4] NCCL INFO Channel 04/0 : 116[4] -> 100[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:124165:124584 [5] NCCL INFO Channel 13/0 : 13[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815969:2816398 [6] NCCL INFO Channel 14/0 : 94[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114517:114935 [5] NCCL INFO Channel 05/0 : 69[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557270:1557688 [4] NCCL INFO Channel 04/0 : 100[4] -> 84[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815969:2816398 [6] NCCL INFO Channel 14/0 : 62[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636086:2636507 [2] NCCL INFO Channel 01/0 : 50[2] -> 49[1] via P2P/IPC
+ip-26-0-161-138:96567:97008 [4] NCCL INFO Channel 04/0 : 52[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114516:114934 [4] NCCL INFO Channel 12/0 : 76[4] -> 68[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96891:97311 [3] NCCL INFO Channel 00/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-161-138:96569:97006 [6] NCCL INFO Channel 06/0 : 70[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114518:114937 [6] NCCL INFO Channel 06/0 : 6[6] -> 70[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153567:153979 [5] NCCL INFO Channel 13/0 : 109[5] -> 93[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96567:97008 [4] NCCL INFO Channel 04/0 : 36[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114518:114937 [6] NCCL INFO Channel 06/0 : 70[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153567:153979 [5] NCCL INFO Channel 13/0 : 93[5] -> 77[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815968:2816394 [5] NCCL INFO Channel 05/0 : 61[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32709:33129 [5] NCCL INFO Channel 13/0 : 45[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124163:124588 [3] NCCL INFO Channel 10/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-165-59:1557271:1557690 [5] NCCL INFO Channel 05/0 : 117[5] -> 101[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96891:97311 [3] NCCL INFO Channel 02/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-168-34:272154:272569 [4] NCCL INFO Channel 04/0 : 124[4] -> 116[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153568:153982 [6] NCCL INFO Channel 14/0 : 110[6] -> 94[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32709:33129 [5] NCCL INFO Channel 13/0 : 29[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557270:1557688 [4] NCCL INFO Channel 12/0 : 108[4] -> 100[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557271:1557690 [5] NCCL INFO Channel 05/0 : 101[5] -> 85[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272154:272569 [4] NCCL INFO Channel 04/0 : 116[4] -> 108[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96568:97007 [5] NCCL INFO Channel 05/0 : 53[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153568:153982 [6] NCCL INFO Channel 14/0 : 94[6] -> 78[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235652:236066 [4] NCCL INFO Channel 04/0 : 116[4] -> 108[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227467:1227886 [4] NCCL INFO Channel 04/0 : 92[4] -> 84[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114517:114935 [5] NCCL INFO Channel 13/0 : 77[5] -> 69[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272154:272569 [4] NCCL INFO Channel 12/0 : 116[4] -> 108[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815969:2816398 [6] NCCL INFO Channel 06/0 : 62[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32710:33125 [6] NCCL INFO Channel 14/0 : 46[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636088:2636510 [4] NCCL INFO Channel 04/0 : 60[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96568:97007 [5] NCCL INFO Channel 05/0 : 37[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235652:236066 [4] NCCL INFO Channel 12/0 : 116[4] -> 108[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227467:1227886 [4] NCCL INFO Channel 04/0 : 84[4] -> 76[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114518:114937 [6] NCCL INFO Channel 06/0 : 102[6] -> 70[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96891:97311 [3] NCCL INFO Channel 08/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-161-123:32710:33125 [6] NCCL INFO Channel 14/0 : 30[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:124166:124583 [6] NCCL INFO Channel 14/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636086:2636507 [2] NCCL INFO Channel 08/0 : 50[2] -> 49[1] via P2P/IPC
+ip-26-0-161-138:96567:97008 [4] NCCL INFO Channel 12/0 : 44[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235653:236067 [5] NCCL INFO Channel 05/0 : 117[5] -> 109[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815966:2816396 [3] NCCL INFO Channel 00/0 : 59[3] -> 58[2] via P2P/IPC
+ip-26-0-165-164:1227466:1227887 [3] NCCL INFO Channel 00/0 : 83[3] -> 82[2] via P2P/IPC
+ip-26-0-161-153:2636088:2636510 [4] NCCL INFO Channel 04/0 : 52[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722051:2722465 [6] NCCL INFO Channel 06/0 : 126[6] -> 118[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114518:114937 [6] NCCL INFO Channel 06/0 : 70[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235652:236066 [4] NCCL INFO Channel 12/0 : 108[4] -> 100[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153567:153979 [5] NCCL INFO Channel 05/0 : 93[5] -> 85[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636088:2636510 [4] NCCL INFO Channel 12/0 : 52[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227467:1227886 [4] NCCL INFO Channel 12/0 : 84[4] -> 76[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:124164:124585 [4] NCCL INFO Channel 00/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-168-30:235653:236067 [5] NCCL INFO Channel 13/0 : 117[5] -> 109[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630210:1630653 [5] NCCL INFO Channel 05/0 : 85[5] -> 77[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272155:272574 [5] NCCL INFO Channel 05/0 : 125[5] -> 117[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235654:236070 [6] NCCL INFO Channel 06/0 : 118[6] -> 110[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32709:33129 [5] NCCL INFO Channel 05/0 : 29[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153568:153982 [6] NCCL INFO Channel 06/0 : 94[6] -> 86[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235653:236067 [5] NCCL INFO Channel 13/0 : 109[5] -> 101[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630210:1630653 [5] NCCL INFO Channel 13/0 : 85[5] -> 77[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636089:2636508 [5] NCCL INFO Channel 05/0 : 61[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557271:1557690 [5] NCCL INFO Channel 13/0 : 109[5] -> 101[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272155:272574 [5] NCCL INFO Channel 05/0 : 117[5] -> 109[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227466:1227887 [3] NCCL INFO Channel 02/0 : 83[3] -> 82[2] via P2P/IPC
+ip-26-0-161-178:2815966:2816396 [3] NCCL INFO Channel 02/0 : 59[3] -> 58[2] via P2P/IPC
+ip-26-0-168-30:235654:236070 [6] NCCL INFO Channel 14/0 : 118[6] -> 110[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630211:1630649 [6] NCCL INFO Channel 06/0 : 86[6] -> 78[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557272:1557687 [6] NCCL INFO Channel 06/0 : 118[6] -> 102[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96891:97311 [3] NCCL INFO Channel 10/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-165-131:1630210:1630653 [5] NCCL INFO Channel 13/0 : 77[5] -> 69[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636089:2636508 [5] NCCL INFO Channel 05/0 : 53[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272155:272574 [5] NCCL INFO Channel 13/0 : 117[5] -> 109[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364907:1365329 [5] NCCL INFO Channel 05/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32710:33125 [6] NCCL INFO Channel 06/0 : 30[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227468:1227883 [5] NCCL INFO Channel 05/0 : 93[5] -> 85[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96568:97007 [5] NCCL INFO Channel 13/0 : 45[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114518:114937 [6] NCCL INFO Channel 14/0 : 78[6] -> 70[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557272:1557687 [6] NCCL INFO Channel 06/0 : 102[6] -> 86[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235654:236070 [6] NCCL INFO Channel 14/0 : 110[6] -> 102[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96893:97310 [5] NCCL INFO Channel 05/0 : 29[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630211:1630649 [6] NCCL INFO Channel 14/0 : 86[6] -> 78[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636089:2636508 [5] NCCL INFO Channel 13/0 : 53[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364908:1365332 [6] NCCL INFO Channel 06/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227468:1227883 [5] NCCL INFO Channel 05/0 : 85[5] -> 77[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364907:1365329 [5] NCCL INFO Channel 13/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96569:97006 [6] NCCL INFO Channel 06/0 : 54[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96893:97310 [5] NCCL INFO Channel 05/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630211:1630649 [6] NCCL INFO Channel 14/0 : 78[6] -> 70[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:124164:124585 [4] NCCL INFO Channel 03/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-165-164:1227468:1227883 [5] NCCL INFO Channel 13/0 : 85[5] -> 77[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364908:1365332 [6] NCCL INFO Channel 14/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96569:97006 [6] NCCL INFO Channel 06/0 : 38[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364907:1365329 [5] NCCL INFO Channel 13/0 : 13[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630208:1630650 [3] NCCL INFO Channel 02/0 : 75[3] -> 74[2] via P2P/IPC
+ip-26-0-161-153:2636086:2636507 [2] NCCL INFO Channel 09/0 : 50[2] -> 49[1] via P2P/IPC
+ip-26-0-161-103:96893:97310 [5] NCCL INFO Channel 13/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227466:1227887 [3] NCCL INFO Channel 08/0 : 83[3] -> 82[2] via P2P/IPC
+ip-26-0-168-30:235651:236068 [3] NCCL INFO Channel 00/0 : 107[3] -> 106[2] via P2P/IPC
+ip-26-0-160-242:1364908:1365332 [6] NCCL INFO Channel 14/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272156:272570 [6] NCCL INFO Channel 06/0 : 126[6] -> 118[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815966:2816396 [3] NCCL INFO Channel 08/0 : 59[3] -> 58[2] via P2P/IPC
+ip-26-0-168-34:272156:272570 [6] NCCL INFO Channel 06/0 : 118[6] -> 110[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557272:1557687 [6] NCCL INFO Channel 14/0 : 110[6] -> 102[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630208:1630650 [3] NCCL INFO Channel 08/0 : 75[3] -> 74[2] via P2P/IPC
+ip-26-0-161-153:2636090:2636505 [6] NCCL INFO Channel 06/0 : 62[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227469:1227882 [6] NCCL INFO Channel 06/0 : 94[6] -> 86[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272156:272570 [6] NCCL INFO Channel 14/0 : 118[6] -> 110[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636090:2636505 [6] NCCL INFO Channel 06/0 : 54[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227469:1227882 [6] NCCL INFO Channel 06/0 : 86[6] -> 78[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96569:97006 [6] NCCL INFO Channel 14/0 : 46[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-52:2722048:2722469 [3] NCCL INFO Channel 02/0 : 123[3] -> 122[2] via P2P/IPC
+ip-26-0-161-103:96894:97312 [6] NCCL INFO Channel 06/0 : 30[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636090:2636505 [6] NCCL INFO Channel 14/0 : 54[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227469:1227882 [6] NCCL INFO Channel 14/0 : 86[6] -> 78[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96894:97312 [6] NCCL INFO Channel 06/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235651:236068 [3] NCCL INFO Channel 02/0 : 107[3] -> 106[2] via P2P/IPC
+ip-26-0-161-103:96894:97312 [6] NCCL INFO Channel 14/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-52:2722050:2722471 [5] NCCL INFO Channel 05/0 : 125[5] -> 117[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114516:114934 [4] NCCL INFO Channel 00/0 : 68[4] -> 67[3] via P2P/IPC
+ip-26-0-161-178:2815966:2816396 [3] NCCL INFO Channel 10/0 : 59[3] -> 58[2] via P2P/IPC
+ip-26-0-161-103:96892:97308 [4] NCCL INFO Channel 04/0 : 28[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272153:272573 [3] NCCL INFO Channel 00/0 : 115[3] -> 114[2] via P2P/IPC
+ip-26-0-161-103:96892:97308 [4] NCCL INFO Channel 04/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630208:1630650 [3] NCCL INFO Channel 10/0 : 75[3] -> 74[2] via P2P/IPC
+ip-26-0-168-52:2722048:2722469 [3] NCCL INFO Channel 08/0 : 123[3] -> 122[2] via P2P/IPC
+ip-26-0-161-103:96892:97308 [4] NCCL INFO Channel 12/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153566:153980 [4] NCCL INFO Channel 00/0 : 92[4] -> 91[3] via P2P/IPC
+ip-26-0-168-30:235651:236068 [3] NCCL INFO Channel 08/0 : 107[3] -> 106[2] via P2P/IPC
+ip-26-0-161-142:62886:63303 [5] NCCL INFO Channel 05/0 : 53[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62887:63305 [6] NCCL INFO Channel 06/0 : 54[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62886:63303 [5] NCCL INFO Channel 13/0 : 53[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114516:114934 [4] NCCL INFO Channel 03/0 : 68[4] -> 67[3] via P2P/IPC
+ip-26-0-161-142:62887:63305 [6] NCCL INFO Channel 14/0 : 54[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62886:63303 [5] NCCL INFO Channel 13/0 : 45[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62887:63305 [6] NCCL INFO Channel 14/0 : 46[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272153:272573 [3] NCCL INFO Channel 02/0 : 115[3] -> 114[2] via P2P/IPC
+ip-26-0-165-202:153566:153980 [4] NCCL INFO Channel 03/0 : 92[4] -> 91[3] via P2P/IPC
+ip-26-0-161-78:114515:114936 [3] NCCL INFO Channel 00/0 : 67[3] -> 66[2] via P2P/IPC
+ip-26-0-168-52:2722048:2722469 [3] NCCL INFO Channel 10/0 : 123[3] -> 122[2] via P2P/IPC
+ip-26-0-168-30:235651:236068 [3] NCCL INFO Channel 10/0 : 107[3] -> 106[2] via P2P/IPC
+ip-26-0-165-202:153567:153979 [5] NCCL INFO Channel 00/0 : 93[5] -> 92[4] via P2P/IPC
+ip-26-0-165-164:1227466:1227887 [3] NCCL INFO Channel 10/0 : 83[3] -> 82[2] via P2P/IPC
+ip-26-0-168-30:235652:236066 [4] NCCL INFO Channel 00/0 : 108[4] -> 107[3] via P2P/IPC
+ip-26-0-168-34:272153:272573 [3] NCCL INFO Channel 08/0 : 115[3] -> 114[2] via P2P/IPC
+ip-26-0-165-202:153566:153980 [4] NCCL INFO Channel 05/0 : 92[4] -> 91[3] via P2P/IPC
+ip-26-0-165-202:153567:153979 [5] NCCL INFO Channel 04/0 : 93[5] -> 92[4] via P2P/IPC
+ip-26-0-160-242:1364905:1365328 [3] NCCL INFO Channel 00/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-161-123:32708:33126 [4] NCCL INFO Channel 00/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-165-202:153566:153980 [4] NCCL INFO Channel 06/0 : 92[4] -> 91[3] via P2P/IPC
+ip-26-0-168-30:235653:236067 [5] NCCL INFO Channel 00/0 : 109[5] -> 108[4] via P2P/IPC
+ip-26-0-165-202:153568:153982 [6] NCCL INFO Channel 00/0 : 94[6] -> 93[5] via P2P/IPC
+ip-26-0-168-30:235652:236066 [4] NCCL INFO Channel 03/0 : 108[4] -> 107[3] via P2P/IPC
+ip-26-0-168-34:272153:272573 [3] NCCL INFO Channel 10/0 : 115[3] -> 114[2] via P2P/IPC
+ip-26-0-165-202:153567:153979 [5] NCCL INFO Channel 08/0 : 93[5] -> 92[4] via P2P/IPC
+ip-26-0-168-34:272154:272569 [4] NCCL INFO Channel 00/0 : 116[4] -> 115[3] via P2P/IPC
+ip-26-0-160-242:1364905:1365328 [3] NCCL INFO Channel 02/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-161-123:32708:33126 [4] NCCL INFO Channel 03/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-161-153:2636088:2636510 [4] NCCL INFO Channel 00/0 : 52[4] -> 51[3] via P2P/IPC
+ip-26-0-161-78:114516:114934 [4] NCCL INFO Channel 05/0 : 68[4] -> 67[3] via P2P/IPC
+ip-26-0-161-123:32708:33126 [4] NCCL INFO Channel 05/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-168-34:272156:272570 [6] NCCL INFO Channel 00/0 : 118[6] -> 117[5] via P2P/IPC
+ip-26-0-160-225:124164:124585 [4] NCCL INFO Channel 05/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-153:2636091:2636512 [7] NCCL INFO Channel 07/0 : 47[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815967:2816399 [4] NCCL INFO Channel 00/0 : 60[4] -> 59[3] via P2P/IPC
+ip-26-0-160-242:1364905:1365328 [3] NCCL INFO Channel 08/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-161-153:2636084:2636511 [0] NCCL INFO Channel 00/0 : 40[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636091:2636512 [7] NCCL INFO Channel 15/0 : 47[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364907:1365329 [5] NCCL INFO Channel 00/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-168-34:272154:272569 [4] NCCL INFO Channel 03/0 : 116[4] -> 115[3] via P2P/IPC
+ip-26-0-161-153:2636084:2636511 [0] NCCL INFO Channel 08/0 : 40[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364906:1365331 [4] NCCL INFO Channel 00/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-161-153:2636091:2636512 [7] NCCL INFO Channel 07/0 : 55[7] -> 63[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636084:2636511 [0] NCCL INFO Channel 00/0 : 48[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636088:2636510 [4] NCCL INFO Channel 03/0 : 52[4] -> 51[3] via P2P/IPC
+ip-26-0-161-78:114515:114936 [3] NCCL INFO Channel 02/0 : 67[3] -> 66[2] via P2P/IPC
+ip-26-0-161-78:114516:114934 [4] NCCL INFO Channel 06/0 : 68[4] -> 67[3] via P2P/IPC
+ip-26-0-161-142:62888:63302 [7] NCCL INFO Channel 15/0 : 31[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114517:114935 [5] NCCL INFO Channel 00/0 : 69[5] -> 68[4] via P2P/IPC
+ip-26-0-161-78:114518:114937 [6] NCCL INFO Channel 00/0 : 70[6] -> 69[5] via P2P/IPC
+ip-26-0-161-153:2636091:2636512 [7] NCCL INFO Channel 07/0 : 39[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815970:2816397 [7] NCCL INFO Channel 15/0 : 31[7] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636084:2636511 [0] NCCL INFO Channel 00/0 : 32[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815963:2816400 [0] NCCL INFO Channel 08/0 : 24[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62888:63302 [7] NCCL INFO Channel 15/0 : 47[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636087:2636506 [3] NCCL INFO Channel 00/0 : 51[3] -> 50[2] via P2P/IPC
+ip-26-0-161-178:2815970:2816397 [7] NCCL INFO Channel 15/0 : 63[7] -> 95[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364907:1365329 [5] NCCL INFO Channel 04/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-160-242:1364905:1365328 [3] NCCL INFO Channel 10/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-161-123:32711:33127 [7] NCCL INFO Channel 15/0 : 31[7] -> 63[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815963:2816400 [0] NCCL INFO Channel 08/0 : 56[0] -> 88[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364906:1365331 [4] NCCL INFO Channel 03/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-161-153:2636091:2636512 [7] NCCL INFO Channel 07/0 : 55[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722051:2722465 [6] NCCL INFO Channel 00/0 : 126[6] -> 125[5] via P2P/IPC
+ip-26-0-161-78:114515:114936 [3] NCCL INFO Channel 08/0 : 67[3] -> 66[2] via P2P/IPC
+ip-26-0-161-78:114516:114934 [4] NCCL INFO Channel 07/0 : 68[4] -> 67[3] via P2P/IPC
+ip-26-0-161-153:2636084:2636511 [0] NCCL INFO Channel 00/0 : 48[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96563:97003 [0] NCCL INFO Channel 00/0 : 32[0] -> 64[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636088:2636510 [4] NCCL INFO Channel 05/0 : 52[4] -> 51[3] via P2P/IPC
+ip-26-0-161-138:96570:97005 [7] NCCL INFO Channel 07/0 : 39[7] -> 71[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32711:33127 [7] NCCL INFO Channel 15/0 : 63[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815970:2816397 [7] NCCL INFO Channel 15/0 : 127[7] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62881:63298 [0] NCCL INFO Channel 08/0 : 24[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153569:153981 [7] NCCL INFO Channel 15/0 : 95[7] -> 63[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815970:2816397 [7] NCCL INFO Channel 15/0 : 63[7] -> 127[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96563:97003 [0] NCCL INFO Channel 00/0 : 64[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114516:114934 [4] NCCL INFO Channel 08/0 : 68[4] -> 67[3] via P2P/IPC
+ip-26-0-161-142:62881:63298 [0] NCCL INFO Channel 08/0 : 40[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32708:33126 [4] NCCL INFO Channel 06/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-161-153:2636087:2636506 [3] NCCL INFO Channel 02/0 : 51[3] -> 50[2] via P2P/IPC
+ip-26-0-161-178:2815970:2816397 [7] NCCL INFO Channel 15/0 : 95[7] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32704:33123 [0] NCCL INFO Channel 08/0 : 24[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114516:114934 [4] NCCL INFO Channel 11/0 : 68[4] -> 67[3] via P2P/IPC
+ip-26-0-161-178:2815970:2816397 [7] NCCL INFO Channel 15/0 : 63[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364907:1365329 [5] NCCL INFO Channel 08/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-168-52:2722052:2722464 [7] NCCL INFO Channel 07/0 : 127[7] -> 119[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153566:153980 [4] NCCL INFO Channel 07/0 : 92[4] -> 91[3] via P2P/IPC
+ip-26-0-161-153:2636089:2636508 [5] NCCL INFO Channel 00/0 : 53[5] -> 52[4] via P2P/IPC
+ip-26-0-161-153:2636090:2636505 [6] NCCL INFO Channel 00/0 : 54[6] -> 53[5] via P2P/IPC
+ip-26-0-161-123:32704:33123 [0] NCCL INFO Channel 08/0 : 56[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815963:2816400 [0] NCCL INFO Channel 08/0 : 120[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636088:2636510 [4] NCCL INFO Channel 06/0 : 52[4] -> 51[3] via P2P/IPC
+ip-26-0-161-78:114516:114934 [4] NCCL INFO Channel 13/0 : 68[4] -> 67[3] via P2P/IPC
+ip-26-0-161-178:2815963:2816400 [0] NCCL INFO Channel 08/0 : 56[0] -> 120[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153569:153981 [7] NCCL INFO Channel 15/0 : 111[7] -> 95[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153569:153981 [7] NCCL INFO Channel 15/0 : 95[7] -> 79[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815970:2816397 [7] NCCL INFO Channel 07/0 : 63[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32711:33127 [7] NCCL INFO Channel 15/0 : 47[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153562:153983 [0] NCCL INFO Channel 08/0 : 88[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114516:114934 [4] NCCL INFO Channel 14/0 : 68[4] -> 67[3] via P2P/IPC
+ip-26-0-161-178:2815963:2816400 [0] NCCL INFO Channel 08/0 : 88[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32711:33127 [7] NCCL INFO Channel 15/0 : 31[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722049:2722467 [4] NCCL INFO Channel 00/0 : 124[4] -> 123[3] via P2P/IPC
+ip-26-0-161-178:2815963:2816400 [0] NCCL INFO Channel 08/0 : 56[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-52:2722045:2722470 [0] NCCL INFO Channel 00/0 : 120[0] -> 112[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235655:236069 [7] NCCL INFO Channel 07/0 : 119[7] -> 111[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235653:236067 [5] NCCL INFO Channel 04/0 : 109[5] -> 108[4] via P2P/IPC
+ip-26-0-161-153:2636087:2636506 [3] NCCL INFO Channel 08/0 : 51[3] -> 50[2] via P2P/IPC
+ip-26-0-161-78:114516:114934 [4] NCCL INFO Channel 15/0 : 68[4] -> 67[3] via P2P/IPC
+ip-26-0-165-202:153569:153981 [7] NCCL INFO Channel 07/0 : 95[7] -> 87[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235652:236066 [4] NCCL INFO Channel 05/0 : 108[4] -> 107[3] via P2P/IPC
+ip-26-0-168-34:272155:272574 [5] NCCL INFO Channel 00/0 : 117[5] -> 116[4] via P2P/IPC
+ip-26-0-168-52:2722050:2722471 [5] NCCL INFO Channel 00/0 : 125[5] -> 124[4] via P2P/IPC
+ip-26-0-168-30:235655:236069 [7] NCCL INFO Channel 15/0 : 119[7] -> 111[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272156:272570 [6] NCCL INFO Channel 05/0 : 118[6] -> 117[5] via P2P/IPC
+ip-26-0-165-164:1227467:1227886 [4] NCCL INFO Channel 00/0 : 84[4] -> 83[3] via P2P/IPC
+ip-26-0-165-202:153562:153983 [0] NCCL INFO Channel 08/0 : 104[0] -> 88[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114518:114937 [6] NCCL INFO Channel 05/0 : 70[6] -> 69[5] via P2P/IPC
+ip-26-0-168-30:235655:236069 [7] NCCL INFO Channel 15/0 : 111[7] -> 103[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364906:1365331 [4] NCCL INFO Channel 05/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-160-242:1364907:1365329 [5] NCCL INFO Channel 12/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-160-242:1364906:1365331 [4] NCCL INFO Channel 06/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-165-131:1630212:1630652 [7] NCCL INFO Channel 07/0 : 87[7] -> 79[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636089:2636508 [5] NCCL INFO Channel 04/0 : 53[5] -> 52[4] via P2P/IPC
+ip-26-0-161-153:2636090:2636505 [6] NCCL INFO Channel 05/0 : 54[6] -> 53[5] via P2P/IPC
+ip-26-0-161-153:2636088:2636510 [4] NCCL INFO Channel 07/0 : 52[4] -> 51[3] via P2P/IPC
+ip-26-0-161-178:2815963:2816400 [0] NCCL INFO Channel 00/0 : 56[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32704:33123 [0] NCCL INFO Channel 08/0 : 40[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153562:153983 [0] NCCL INFO Channel 08/0 : 88[0] -> 72[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630212:1630652 [7] NCCL INFO Channel 15/0 : 87[7] -> 79[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32704:33123 [0] NCCL INFO Channel 08/0 : 24[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630212:1630652 [7] NCCL INFO Channel 15/0 : 79[7] -> 71[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96570:97005 [7] NCCL INFO Channel 07/0 : 71[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114519:114931 [7] NCCL INFO Channel 07/0 : 7[7] -> 71[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815968:2816394 [5] NCCL INFO Channel 00/0 : 61[5] -> 60[4] via P2P/IPC
+ip-26-0-161-78:114519:114931 [7] NCCL INFO Channel 07/0 : 71[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815967:2816399 [4] NCCL INFO Channel 03/0 : 60[4] -> 59[3] via P2P/IPC
+ip-26-0-168-30:235648:236064 [0] NCCL INFO Channel 00/0 : 112[0] -> 104[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62881:63298 [0] NCCL INFO Channel 00/0 : 48[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235648:236064 [0] NCCL INFO Channel 08/0 : 112[0] -> 104[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153562:153983 [0] NCCL INFO Channel 00/0 : 88[0] -> 80[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636087:2636506 [3] NCCL INFO Channel 10/0 : 51[3] -> 50[2] via P2P/IPC
+ip-26-0-161-142:62881:63298 [0] NCCL INFO Channel 08/0 : 48[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235648:236064 [0] NCCL INFO Channel 08/0 : 104[0] -> 96[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32704:33123 [0] NCCL INFO Channel 00/0 : 24[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630205:1630646 [0] NCCL INFO Channel 00/0 : 80[0] -> 72[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62881:63298 [0] NCCL INFO Channel 08/0 : 40[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364902:1365333 [0] NCCL INFO Channel 00/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630205:1630646 [0] NCCL INFO Channel 08/0 : 80[0] -> 72[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114519:114931 [7] NCCL INFO Channel 07/0 : 103[7] -> 71[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124167:124586 [7] NCCL INFO Channel 15/0 : 15[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630205:1630646 [0] NCCL INFO Channel 08/0 : 72[0] -> 64[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364902:1365333 [0] NCCL INFO Channel 08/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114512:114930 [0] NCCL INFO Channel 00/0 : 0[0] -> 64[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114517:114935 [5] NCCL INFO Channel 04/0 : 69[5] -> 68[4] via P2P/IPC
+ip-26-0-161-78:114512:114930 [0] NCCL INFO Channel 00/0 : 64[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114519:114931 [7] NCCL INFO Channel 07/0 : 71[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364902:1365333 [0] NCCL INFO Channel 08/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153568:153982 [6] NCCL INFO Channel 05/0 : 94[6] -> 93[5] via P2P/IPC
+ip-26-0-161-153:2636089:2636508 [5] NCCL INFO Channel 08/0 : 53[5] -> 52[4] via P2P/IPC
+ip-26-0-161-142:62888:63302 [7] NCCL INFO Channel 07/0 : 55[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636090:2636505 [6] NCCL INFO Channel 08/0 : 54[6] -> 53[5] via P2P/IPC
+ip-26-0-161-153:2636088:2636510 [4] NCCL INFO Channel 08/0 : 52[4] -> 51[3] via P2P/IPC
+ip-26-0-161-142:62888:63302 [7] NCCL INFO Channel 15/0 : 55[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32711:33127 [7] NCCL INFO Channel 07/0 : 31[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114512:114930 [0] NCCL INFO Channel 00/0 : 96[0] -> 64[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62888:63302 [7] NCCL INFO Channel 15/0 : 47[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124160:124581 [0] NCCL INFO Channel 08/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364909:1365327 [7] NCCL INFO Channel 07/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557273:1557685 [7] NCCL INFO Channel 07/0 : 119[7] -> 103[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114512:114930 [0] NCCL INFO Channel 00/0 : 64[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364909:1365327 [7] NCCL INFO Channel 15/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114519:114931 [7] NCCL INFO Channel 15/0 : 79[7] -> 71[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96570:97005 [7] NCCL INFO Channel 07/0 : 55[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364909:1365327 [7] NCCL INFO Channel 15/0 : 15[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96570:97005 [7] NCCL INFO Channel 07/0 : 39[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114519:114931 [7] NCCL INFO Channel 01/0 : 71[7] -> 64[0] via P2P/IPC
+ip-26-0-168-34:272154:272569 [4] NCCL INFO Channel 05/0 : 116[4] -> 115[3] via P2P/IPC
+ip-26-0-168-52:2722051:2722465 [6] NCCL INFO Channel 05/0 : 126[6] -> 125[5] via P2P/IPC
+ip-26-0-165-59:1557266:1557689 [0] NCCL INFO Channel 00/0 : 112[0] -> 96[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235654:236070 [6] NCCL INFO Channel 00/0 : 110[6] -> 109[5] via P2P/IPC
+ip-26-0-160-242:1364906:1365331 [4] NCCL INFO Channel 07/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-160-242:1364908:1365332 [6] NCCL INFO Channel 00/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-168-52:2722049:2722467 [4] NCCL INFO Channel 03/0 : 124[4] -> 123[3] via P2P/IPC
+ip-26-0-161-78:114512:114930 [0] NCCL INFO Channel 08/0 : 72[0] -> 64[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557266:1557689 [0] NCCL INFO Channel 00/0 : 96[0] -> 80[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:124164:124585 [4] NCCL INFO Channel 06/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-138:96563:97003 [0] NCCL INFO Channel 00/0 : 48[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:124167:124586 [7] NCCL INFO Channel 01/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-78:114518:114937 [6] NCCL INFO Channel 08/0 : 70[6] -> 69[5] via P2P/IPC
+ip-26-0-161-78:114515:114936 [3] NCCL INFO Channel 10/0 : 67[3] -> 66[2] via P2P/IPC
+ip-26-0-161-153:2636091:2636512 [7] NCCL INFO Channel 07/0 : 63[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96563:97003 [0] NCCL INFO Channel 00/0 : 32[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235653:236067 [5] NCCL INFO Channel 08/0 : 109[5] -> 108[4] via P2P/IPC
+ip-26-0-168-30:235652:236066 [4] NCCL INFO Channel 06/0 : 108[4] -> 107[3] via P2P/IPC
+ip-26-0-161-153:2636089:2636508 [5] NCCL INFO Channel 12/0 : 53[5] -> 52[4] via P2P/IPC
+ip-26-0-161-153:2636090:2636505 [6] NCCL INFO Channel 13/0 : 54[6] -> 53[5] via P2P/IPC
+ip-26-0-161-138:96570:97005 [7] NCCL INFO Channel 15/0 : 47[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636091:2636512 [7] NCCL INFO Channel 07/0 : 55[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96895:97307 [7] NCCL INFO Channel 07/0 : 31[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636088:2636510 [4] NCCL INFO Channel 11/0 : 52[4] -> 51[3] via P2P/IPC
+ip-26-0-161-153:2636091:2636512 [7] NCCL INFO Channel 15/0 : 55[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96895:97307 [7] NCCL INFO Channel 07/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32708:33126 [4] NCCL INFO Channel 07/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-168-34:272150:272576 [0] NCCL INFO Channel 00/0 : 120[0] -> 112[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557266:1557689 [0] NCCL INFO Channel 08/0 : 104[0] -> 96[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96895:97307 [7] NCCL INFO Channel 15/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636084:2636511 [0] NCCL INFO Channel 00/0 : 56[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815970:2816397 [7] NCCL INFO Channel 01/0 : 63[7] -> 56[0] via P2P/IPC
+ip-26-0-165-164:1227463:1227880 [0] NCCL INFO Channel 00/0 : 88[0] -> 80[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272150:272576 [0] NCCL INFO Channel 00/0 : 112[0] -> 104[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636084:2636511 [0] NCCL INFO Channel 00/0 : 48[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96563:97003 [0] NCCL INFO Channel 08/0 : 40[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114517:114935 [5] NCCL INFO Channel 08/0 : 69[5] -> 68[4] via P2P/IPC
+ip-26-0-161-78:114519:114931 [7] NCCL INFO Channel 02/0 : 71[7] -> 64[0] via P2P/IPC
+ip-26-0-168-34:272150:272576 [0] NCCL INFO Channel 08/0 : 112[0] -> 104[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227463:1227880 [0] NCCL INFO Channel 00/0 : 80[0] -> 72[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96888:97306 [0] NCCL INFO Channel 00/0 : 24[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96570:97005 [7] NCCL INFO Channel 01/0 : 39[7] -> 32[0] via P2P/IPC
+ip-26-0-161-142:62888:63302 [7] NCCL INFO Channel 01/0 : 47[7] -> 40[0] via P2P/IPC
+ip-26-0-161-178:2815968:2816394 [5] NCCL INFO Channel 04/0 : 61[5] -> 60[4] via P2P/IPC
+ip-26-0-161-153:2636091:2636512 [7] NCCL INFO Channel 01/0 : 55[7] -> 48[0] via P2P/IPC
+ip-26-0-161-153:2636084:2636511 [0] NCCL INFO Channel 08/0 : 48[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32711:33127 [7] NCCL INFO Channel 01/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-165-164:1227463:1227880 [0] NCCL INFO Channel 08/0 : 80[0] -> 72[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96888:97306 [0] NCCL INFO Channel 00/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815967:2816399 [4] NCCL INFO Channel 05/0 : 60[4] -> 59[3] via P2P/IPC
+ip-26-0-161-103:96888:97306 [0] NCCL INFO Channel 08/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364909:1365327 [7] NCCL INFO Channel 01/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-161-103:96895:97307 [7] NCCL INFO Channel 01/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-165-59:1557273:1557685 [7] NCCL INFO Channel 07/0 : 103[7] -> 87[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153566:153980 [4] NCCL INFO Channel 08/0 : 92[4] -> 91[3] via P2P/IPC
+ip-26-0-168-52:2722050:2722471 [5] NCCL INFO Channel 04/0 : 125[5] -> 124[4] via P2P/IPC
+ip-26-0-165-131:1630209:1630651 [4] NCCL INFO Channel 00/0 : 76[4] -> 75[3] via P2P/IPC
+ip-26-0-165-131:1630210:1630653 [5] NCCL INFO Channel 00/0 : 77[5] -> 76[4] via P2P/IPC
+ip-26-0-168-34:272155:272574 [5] NCCL INFO Channel 04/0 : 117[5] -> 116[4] via P2P/IPC
+ip-26-0-168-34:272156:272570 [6] NCCL INFO Channel 08/0 : 118[6] -> 117[5] via P2P/IPC
+ip-26-0-168-34:272157:272572 [7] NCCL INFO Channel 07/0 : 127[7] -> 119[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557273:1557685 [7] NCCL INFO Channel 15/0 : 111[7] -> 103[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272157:272572 [7] NCCL INFO Channel 07/0 : 119[7] -> 111[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227470:1227885 [7] NCCL INFO Channel 07/0 : 95[7] -> 87[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272157:272572 [7] NCCL INFO Channel 15/0 : 119[7] -> 111[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227470:1227885 [7] NCCL INFO Channel 07/0 : 87[7] -> 79[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557273:1557685 [7] NCCL INFO Channel 01/0 : 103[7] -> 96[0] via P2P/IPC
+ip-26-0-165-164:1227470:1227885 [7] NCCL INFO Channel 15/0 : 87[7] -> 79[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722052:2722464 [7] NCCL INFO Channel 01/0 : 127[7] -> 120[0] via P2P/IPC
+ip-26-0-168-34:272157:272572 [7] NCCL INFO Channel 01/0 : 119[7] -> 112[0] via P2P/IPC
+ip-26-0-165-202:153569:153981 [7] NCCL INFO Channel 01/0 : 95[7] -> 88[0] via P2P/IPC
+ip-26-0-168-30:235655:236069 [7] NCCL INFO Channel 01/0 : 111[7] -> 104[0] via P2P/IPC
+ip-26-0-165-164:1227470:1227885 [7] NCCL INFO Channel 01/0 : 87[7] -> 80[0] via P2P/IPC
+ip-26-0-165-131:1630212:1630652 [7] NCCL INFO Channel 01/0 : 79[7] -> 72[0] via P2P/IPC
+ip-26-0-161-78:114518:114937 [6] NCCL INFO Channel 13/0 : 70[6] -> 69[5] via P2P/IPC
+ip-26-0-165-202:153568:153982 [6] NCCL INFO Channel 08/0 : 94[6] -> 93[5] via P2P/IPC
+ip-26-0-161-142:62888:63302 [7] NCCL INFO Channel 02/0 : 47[7] -> 40[0] via P2P/IPC
+ip-26-0-165-202:153569:153981 [7] NCCL INFO Channel 02/0 : 95[7] -> 88[0] via P2P/IPC
+ip-26-0-168-34:272154:272569 [4] NCCL INFO Channel 06/0 : 116[4] -> 115[3] via P2P/IPC
+ip-26-0-168-30:235653:236067 [5] NCCL INFO Channel 12/0 : 109[5] -> 108[4] via P2P/IPC
+ip-26-0-168-30:235652:236066 [4] NCCL INFO Channel 07/0 : 108[4] -> 107[3] via P2P/IPC
+ip-26-0-168-30:235654:236070 [6] NCCL INFO Channel 05/0 : 110[6] -> 109[5] via P2P/IPC
+ip-26-0-160-225:124164:124585 [4] NCCL INFO Channel 07/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-165-131:1630211:1630649 [6] NCCL INFO Channel 00/0 : 78[6] -> 77[5] via P2P/IPC
+ip-26-0-160-225:124167:124586 [7] NCCL INFO Channel 02/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-131:1630210:1630653 [5] NCCL INFO Channel 04/0 : 77[5] -> 76[4] via P2P/IPC
+ip-26-0-165-131:1630209:1630651 [4] NCCL INFO Channel 03/0 : 76[4] -> 75[3] via P2P/IPC
+ip-26-0-161-123:32708:33126 [4] NCCL INFO Channel 08/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-161-153:2636088:2636510 [4] NCCL INFO Channel 13/0 : 52[4] -> 51[3] via P2P/IPC
+ip-26-0-161-153:2636091:2636512 [7] NCCL INFO Channel 02/0 : 55[7] -> 48[0] via P2P/IPC
+ip-26-0-161-78:114519:114931 [7] NCCL INFO Channel 03/0 : 71[7] -> 64[0] via P2P/IPC
+ip-26-0-165-59:1557273:1557685 [7] NCCL INFO Channel 02/0 : 103[7] -> 96[0] via P2P/IPC
+ip-26-0-161-78:114517:114935 [5] NCCL INFO Channel 12/0 : 69[5] -> 68[4] via P2P/IPC
+ip-26-0-160-242:1364906:1365331 [4] NCCL INFO Channel 08/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-165-59:1557271:1557690 [5] NCCL INFO Channel 00/0 : 101[5] -> 100[4] via P2P/IPC
+ip-26-0-160-242:1364908:1365332 [6] NCCL INFO Channel 05/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-160-242:1364909:1365327 [7] NCCL INFO Channel 02/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-161-142:62885:63304 [4] NCCL INFO Channel 00/0 : 44[4] -> 43[3] via P2P/IPC
+ip-26-0-168-30:235655:236069 [7] NCCL INFO Channel 02/0 : 111[7] -> 104[0] via P2P/IPC
+ip-26-0-165-164:1227469:1227882 [6] NCCL INFO Channel 00/0 : 86[6] -> 85[5] via P2P/IPC
+ip-26-0-168-52:2722051:2722465 [6] NCCL INFO Channel 08/0 : 126[6] -> 125[5] via P2P/IPC
+ip-26-0-168-52:2722049:2722467 [4] NCCL INFO Channel 05/0 : 124[4] -> 123[3] via P2P/IPC
+ip-26-0-165-59:1557270:1557688 [4] NCCL INFO Channel 00/0 : 100[4] -> 99[3] via P2P/IPC
+ip-26-0-165-131:1630212:1630652 [7] NCCL INFO Channel 02/0 : 79[7] -> 72[0] via P2P/IPC
+ip-26-0-161-123:32711:33127 [7] NCCL INFO Channel 02/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-161-178:2815970:2816397 [7] NCCL INFO Channel 02/0 : 63[7] -> 56[0] via P2P/IPC
+ip-26-0-161-178:2815969:2816398 [6] NCCL INFO Channel 00/0 : 62[6] -> 61[5] via P2P/IPC
+ip-26-0-161-178:2815968:2816394 [5] NCCL INFO Channel 08/0 : 61[5] -> 60[4] via P2P/IPC
+ip-26-0-168-52:2722050:2722471 [5] NCCL INFO Channel 08/0 : 125[5] -> 124[4] via P2P/IPC
+ip-26-0-161-178:2815967:2816399 [4] NCCL INFO Channel 06/0 : 60[4] -> 59[3] via P2P/IPC
+ip-26-0-161-138:96570:97005 [7] NCCL INFO Channel 02/0 : 39[7] -> 32[0] via P2P/IPC
+ip-26-0-161-142:62888:63302 [7] NCCL INFO Channel 03/0 : 47[7] -> 40[0] via P2P/IPC
+ip-26-0-165-59:1557272:1557687 [6] NCCL INFO Channel 00/0 : 102[6] -> 101[5] via P2P/IPC
+ip-26-0-161-78:114519:114931 [7] NCCL INFO Channel 04/0 : 71[7] -> 64[0] via P2P/IPC
+ip-26-0-168-34:272156:272570 [6] NCCL INFO Channel 13/0 : 118[6] -> 117[5] via P2P/IPC
+ip-26-0-165-202:153567:153979 [5] NCCL INFO Channel 12/0 : 93[5] -> 92[4] via P2P/IPC
+ip-26-0-165-164:1227470:1227885 [7] NCCL INFO Channel 02/0 : 87[7] -> 80[0] via P2P/IPC
+ip-26-0-161-142:62886:63303 [5] NCCL INFO Channel 00/0 : 45[5] -> 44[4] via P2P/IPC
+ip-26-0-160-242:1364909:1365327 [7] NCCL INFO Channel 03/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-168-30:235652:236066 [4] NCCL INFO Channel 08/0 : 108[4] -> 107[3] via P2P/IPC
+ip-26-0-160-242:1364906:1365331 [4] NCCL INFO Channel 11/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-168-30:235654:236070 [6] NCCL INFO Channel 08/0 : 110[6] -> 109[5] via P2P/IPC
+ip-26-0-160-225:124164:124585 [4] NCCL INFO Channel 08/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-168-34:272155:272574 [5] NCCL INFO Channel 08/0 : 117[5] -> 116[4] via P2P/IPC
+ip-26-0-160-225:124165:124584 [5] NCCL INFO Channel 00/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-225:124166:124583 [6] NCCL INFO Channel 00/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-131:1630211:1630649 [6] NCCL INFO Channel 05/0 : 78[6] -> 77[5] via P2P/IPC
+ip-26-0-160-225:124167:124586 [7] NCCL INFO Channel 03/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-168-34:272157:272572 [7] NCCL INFO Channel 02/0 : 119[7] -> 112[0] via P2P/IPC
+ip-26-0-160-242:1364909:1365327 [7] NCCL INFO Channel 04/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-165-131:1630210:1630653 [5] NCCL INFO Channel 08/0 : 77[5] -> 76[4] via P2P/IPC
+ip-26-0-165-131:1630209:1630651 [4] NCCL INFO Channel 05/0 : 76[4] -> 75[3] via P2P/IPC
+ip-26-0-168-52:2722052:2722464 [7] NCCL INFO Channel 02/0 : 127[7] -> 120[0] via P2P/IPC
+ip-26-0-165-59:1557273:1557685 [7] NCCL INFO Channel 03/0 : 103[7] -> 96[0] via P2P/IPC
+ip-26-0-161-103:96893:97310 [5] NCCL INFO Channel 00/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-161-103:96892:97308 [4] NCCL INFO Channel 00/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-165-59:1557271:1557690 [5] NCCL INFO Channel 04/0 : 101[5] -> 100[4] via P2P/IPC
+ip-26-0-161-142:62887:63305 [6] NCCL INFO Channel 00/0 : 46[6] -> 45[5] via P2P/IPC
+ip-26-0-165-59:1557270:1557688 [4] NCCL INFO Channel 03/0 : 100[4] -> 99[3] via P2P/IPC
+ip-26-0-161-142:62885:63304 [4] NCCL INFO Channel 03/0 : 44[4] -> 43[3] via P2P/IPC
+ip-26-0-161-123:32709:33129 [5] NCCL INFO Channel 00/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-161-123:32708:33126 [4] NCCL INFO Channel 11/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-165-202:153566:153980 [4] NCCL INFO Channel 11/0 : 92[4] -> 91[3] via P2P/IPC
+ip-26-0-165-164:1227469:1227882 [6] NCCL INFO Channel 05/0 : 86[6] -> 85[5] via P2P/IPC
+ip-26-0-165-164:1227468:1227883 [5] NCCL INFO Channel 00/0 : 85[5] -> 84[4] via P2P/IPC
+ip-26-0-168-30:235655:236069 [7] NCCL INFO Channel 03/0 : 111[7] -> 104[0] via P2P/IPC
+ip-26-0-168-34:272154:272569 [4] NCCL INFO Channel 07/0 : 116[4] -> 115[3] via P2P/IPC
+ip-26-0-161-123:32711:33127 [7] NCCL INFO Channel 03/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-160-242:1364906:1365331 [4] NCCL INFO Channel 13/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-161-123:32710:33125 [6] NCCL INFO Channel 00/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-168-30:235652:236066 [4] NCCL INFO Channel 11/0 : 108[4] -> 107[3] via P2P/IPC
+ip-26-0-161-123:32709:33129 [5] NCCL INFO Channel 04/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-161-153:2636088:2636510 [4] NCCL INFO Channel 14/0 : 52[4] -> 51[3] via P2P/IPC
+ip-26-0-161-153:2636091:2636512 [7] NCCL INFO Channel 03/0 : 55[7] -> 48[0] via P2P/IPC
+ip-26-0-161-123:32708:33126 [4] NCCL INFO Channel 13/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-168-52:2722051:2722465 [6] NCCL INFO Channel 13/0 : 126[6] -> 125[5] via P2P/IPC
+ip-26-0-168-52:2722049:2722467 [4] NCCL INFO Channel 06/0 : 124[4] -> 123[3] via P2P/IPC
+ip-26-0-161-178:2815970:2816397 [7] NCCL INFO Channel 03/0 : 63[7] -> 56[0] via P2P/IPC
+ip-26-0-161-78:114519:114931 [7] NCCL INFO Channel 05/0 : 71[7] -> 64[0] via P2P/IPC
+ip-26-0-161-138:96570:97005 [7] NCCL INFO Channel 03/0 : 39[7] -> 32[0] via P2P/IPC
+ip-26-0-161-178:2815968:2816394 [5] NCCL INFO Channel 12/0 : 61[5] -> 60[4] via P2P/IPC
+ip-26-0-161-178:2815967:2816399 [4] NCCL INFO Channel 07/0 : 60[4] -> 59[3] via P2P/IPC
+ip-26-0-168-52:2722050:2722471 [5] NCCL INFO Channel 12/0 : 125[5] -> 124[4] via P2P/IPC
+ip-26-0-161-178:2815969:2816398 [6] NCCL INFO Channel 05/0 : 62[6] -> 61[5] via P2P/IPC
+ip-26-0-165-59:1557272:1557687 [6] NCCL INFO Channel 05/0 : 102[6] -> 101[5] via P2P/IPC
+ip-26-0-161-142:62888:63302 [7] NCCL INFO Channel 04/0 : 47[7] -> 40[0] via P2P/IPC
+ip-26-0-165-164:1227467:1227886 [4] NCCL INFO Channel 03/0 : 84[4] -> 83[3] via P2P/IPC
+ip-26-0-165-164:1227470:1227885 [7] NCCL INFO Channel 03/0 : 87[7] -> 80[0] via P2P/IPC
+ip-26-0-165-202:153568:153982 [6] NCCL INFO Channel 13/0 : 94[6] -> 93[5] via P2P/IPC
+ip-26-0-168-34:272155:272574 [5] NCCL INFO Channel 12/0 : 117[5] -> 116[4] via P2P/IPC
+ip-26-0-168-34:272157:272572 [7] NCCL INFO Channel 03/0 : 119[7] -> 112[0] via P2P/IPC
+ip-26-0-168-30:235654:236070 [6] NCCL INFO Channel 13/0 : 110[6] -> 109[5] via P2P/IPC
+ip-26-0-165-131:1630212:1630652 [7] NCCL INFO Channel 03/0 : 79[7] -> 72[0] via P2P/IPC
+ip-26-0-161-142:62886:63303 [5] NCCL INFO Channel 04/0 : 45[5] -> 44[4] via P2P/IPC
+ip-26-0-168-52:2722052:2722464 [7] NCCL INFO Channel 03/0 : 127[7] -> 120[0] via P2P/IPC
+ip-26-0-161-142:62887:63305 [6] NCCL INFO Channel 05/0 : 46[6] -> 45[5] via P2P/IPC
+ip-26-0-160-242:1364909:1365327 [7] NCCL INFO Channel 05/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-161-138:96569:97006 [6] NCCL INFO Channel 00/0 : 38[6] -> 37[5] via P2P/IPC
+ip-26-0-165-59:1557273:1557685 [7] NCCL INFO Channel 04/0 : 103[7] -> 96[0] via P2P/IPC
+ip-26-0-165-59:1557271:1557690 [5] NCCL INFO Channel 08/0 : 101[5] -> 100[4] via P2P/IPC
+ip-26-0-165-202:153569:153981 [7] NCCL INFO Channel 03/0 : 95[7] -> 88[0] via P2P/IPC
+ip-26-0-165-59:1557270:1557688 [4] NCCL INFO Channel 05/0 : 100[4] -> 99[3] via P2P/IPC
+ip-26-0-161-178:2815970:2816397 [7] NCCL INFO Channel 04/0 : 63[7] -> 56[0] via P2P/IPC
+ip-26-0-168-52:2722049:2722467 [4] NCCL INFO Channel 07/0 : 124[4] -> 123[3] via P2P/IPC
+ip-26-0-161-138:96570:97005 [7] NCCL INFO Channel 04/0 : 39[7] -> 32[0] via P2P/IPC
+ip-26-0-160-242:1364906:1365331 [4] NCCL INFO Channel 14/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-165-164:1227469:1227882 [6] NCCL INFO Channel 08/0 : 86[6] -> 85[5] via P2P/IPC
+ip-26-0-161-153:2636091:2636512 [7] NCCL INFO Channel 04/0 : 55[7] -> 48[0] via P2P/IPC
+ip-26-0-168-34:272154:272569 [4] NCCL INFO Channel 08/0 : 116[4] -> 115[3] via P2P/IPC
+ip-26-0-165-131:1630210:1630653 [5] NCCL INFO Channel 12/0 : 77[5] -> 76[4] via P2P/IPC
+ip-26-0-160-225:124164:124585 [4] NCCL INFO Channel 11/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-165-131:1630209:1630651 [4] NCCL INFO Channel 06/0 : 76[4] -> 75[3] via P2P/IPC
+ip-26-0-160-225:124165:124584 [5] NCCL INFO Channel 04/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-225:124166:124583 [6] NCCL INFO Channel 05/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-131:1630211:1630649 [6] NCCL INFO Channel 08/0 : 78[6] -> 77[5] via P2P/IPC
+ip-26-0-161-142:62885:63304 [4] NCCL INFO Channel 05/0 : 44[4] -> 43[3] via P2P/IPC
+ip-26-0-165-164:1227467:1227886 [4] NCCL INFO Channel 05/0 : 84[4] -> 83[3] via P2P/IPC
+ip-26-0-161-142:62888:63302 [7] NCCL INFO Channel 05/0 : 47[7] -> 40[0] via P2P/IPC
+ip-26-0-165-164:1227467:1227886 [4] NCCL INFO Channel 06/0 : 84[4] -> 83[3] via P2P/IPC
+ip-26-0-168-30:235655:236069 [7] NCCL INFO Channel 04/0 : 111[7] -> 104[0] via P2P/IPC
+ip-26-0-161-138:96567:97008 [4] NCCL INFO Channel 00/0 : 36[4] -> 35[3] via P2P/IPC
+ip-26-0-161-138:96568:97007 [5] NCCL INFO Channel 00/0 : 37[5] -> 36[4] via P2P/IPC
+ip-26-0-165-164:1227468:1227883 [5] NCCL INFO Channel 04/0 : 85[5] -> 84[4] via P2P/IPC
+ip-26-0-165-164:1227467:1227886 [4] NCCL INFO Channel 07/0 : 84[4] -> 83[3] via P2P/IPC
+ip-26-0-161-142:62886:63303 [5] NCCL INFO Channel 08/0 : 45[5] -> 44[4] via P2P/IPC
+ip-26-0-160-225:124167:124586 [7] NCCL INFO Channel 04/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-142:62885:63304 [4] NCCL INFO Channel 06/0 : 44[4] -> 43[3] via P2P/IPC
+ip-26-0-168-30:235652:236066 [4] NCCL INFO Channel 13/0 : 108[4] -> 107[3] via P2P/IPC
+ip-26-0-160-242:1364909:1365327 [7] NCCL INFO Channel 06/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-165-164:1227469:1227882 [6] NCCL INFO Channel 13/0 : 86[6] -> 85[5] via P2P/IPC
+ip-26-0-168-52:2722052:2722464 [7] NCCL INFO Channel 04/0 : 127[7] -> 120[0] via P2P/IPC
+ip-26-0-161-103:96893:97310 [5] NCCL INFO Channel 04/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-161-103:96892:97308 [4] NCCL INFO Channel 03/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-161-103:96894:97312 [6] NCCL INFO Channel 00/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-161-142:62888:63302 [7] NCCL INFO Channel 06/0 : 47[7] -> 40[0] via P2P/IPC
+ip-26-0-161-103:96895:97307 [7] NCCL INFO Channel 02/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-161-153:2636088:2636510 [4] NCCL INFO Channel 15/0 : 52[4] -> 51[3] via P2P/IPC
+ip-26-0-161-142:62886:63303 [5] NCCL INFO Channel 12/0 : 45[5] -> 44[4] via P2P/IPC
+ip-26-0-161-142:62885:63304 [4] NCCL INFO Channel 07/0 : 44[4] -> 43[3] via P2P/IPC
+ip-26-0-161-178:2815967:2816399 [4] NCCL INFO Channel 08/0 : 60[4] -> 59[3] via P2P/IPC
+ip-26-0-165-164:1227470:1227885 [7] NCCL INFO Channel 04/0 : 87[7] -> 80[0] via P2P/IPC
+ip-26-0-165-59:1557272:1557687 [6] NCCL INFO Channel 08/0 : 102[6] -> 101[5] via P2P/IPC
+ip-26-0-160-242:1364906:1365331 [4] NCCL INFO Channel 15/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-168-52:2722049:2722467 [4] NCCL INFO Channel 08/0 : 124[4] -> 123[3] via P2P/IPC
+ip-26-0-168-30:235655:236069 [7] NCCL INFO Channel 05/0 : 111[7] -> 104[0] via P2P/IPC
+ip-26-0-165-164:1227468:1227883 [5] NCCL INFO Channel 08/0 : 85[5] -> 84[4] via P2P/IPC
+ip-26-0-165-164:1227467:1227886 [4] NCCL INFO Channel 08/0 : 84[4] -> 83[3] via P2P/IPC
+ip-26-0-165-202:153569:153981 [7] NCCL INFO Channel 04/0 : 95[7] -> 88[0] via P2P/IPC
+ip-26-0-168-34:272157:272572 [7] NCCL INFO Channel 04/0 : 119[7] -> 112[0] via P2P/IPC
+ip-26-0-161-178:2815969:2816398 [6] NCCL INFO Channel 08/0 : 62[6] -> 61[5] via P2P/IPC
+ip-26-0-161-142:62888:63302 [7] NCCL INFO Channel 07/0 : 47[7] -> 40[0] via P2P/IPC
+ip-26-0-165-131:1630209:1630651 [4] NCCL INFO Channel 07/0 : 76[4] -> 75[3] via P2P/IPC
+ip-26-0-165-131:1630211:1630649 [6] NCCL INFO Channel 13/0 : 78[6] -> 77[5] via P2P/IPC
+ip-26-0-161-153:2636091:2636512 [7] NCCL INFO Channel 05/0 : 55[7] -> 48[0] via P2P/IPC
+ip-26-0-161-138:96570:97005 [7] NCCL INFO Channel 05/0 : 39[7] -> 32[0] via P2P/IPC
+ip-26-0-161-138:96567:97008 [4] NCCL INFO Channel 03/0 : 36[4] -> 35[3] via P2P/IPC
+ip-26-0-161-178:2815970:2816397 [7] NCCL INFO Channel 05/0 : 63[7] -> 56[0] via P2P/IPC
+ip-26-0-161-138:96569:97006 [6] NCCL INFO Channel 05/0 : 38[6] -> 37[5] via P2P/IPC
+ip-26-0-160-225:124165:124584 [5] NCCL INFO Channel 08/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-168-30:235652:236066 [4] NCCL INFO Channel 14/0 : 108[4] -> 107[3] via P2P/IPC
+ip-26-0-160-242:1364909:1365327 [7] NCCL INFO Channel 07/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-165-59:1557273:1557685 [7] NCCL INFO Channel 05/0 : 103[7] -> 96[0] via P2P/IPC
+ip-26-0-165-59:1557271:1557690 [5] NCCL INFO Channel 12/0 : 101[5] -> 100[4] via P2P/IPC
+ip-26-0-168-34:272154:272569 [4] NCCL INFO Channel 11/0 : 116[4] -> 115[3] via P2P/IPC
+ip-26-0-165-59:1557270:1557688 [4] NCCL INFO Channel 06/0 : 100[4] -> 99[3] via P2P/IPC
+ip-26-0-160-242:1364908:1365332 [6] NCCL INFO Channel 08/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-165-164:1227470:1227885 [7] NCCL INFO Channel 05/0 : 87[7] -> 80[0] via P2P/IPC
+ip-26-0-161-103:96893:97310 [5] NCCL INFO Channel 08/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-161-103:96892:97308 [4] NCCL INFO Channel 05/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-161-103:96894:97312 [6] NCCL INFO Channel 05/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-161-103:96895:97307 [7] NCCL INFO Channel 03/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-161-138:96568:97007 [5] NCCL INFO Channel 04/0 : 37[5] -> 36[4] via P2P/IPC
+ip-26-0-161-178:2815967:2816399 [4] NCCL INFO Channel 11/0 : 60[4] -> 59[3] via P2P/IPC
+ip-26-0-161-178:2815969:2816398 [6] NCCL INFO Channel 13/0 : 62[6] -> 61[5] via P2P/IPC
+ip-26-0-168-30:235655:236069 [7] NCCL INFO Channel 06/0 : 111[7] -> 104[0] via P2P/IPC
+ip-26-0-165-59:1557272:1557687 [6] NCCL INFO Channel 13/0 : 102[6] -> 101[5] via P2P/IPC
+ip-26-0-165-202:153566:153980 [4] NCCL INFO Channel 13/0 : 92[4] -> 91[3] via P2P/IPC
+ip-26-0-168-52:2722052:2722464 [7] NCCL INFO Channel 05/0 : 127[7] -> 120[0] via P2P/IPC
+ip-26-0-165-202:153569:153981 [7] NCCL INFO Channel 05/0 : 95[7] -> 88[0] via P2P/IPC
+ip-26-0-161-142:62885:63304 [4] NCCL INFO Channel 08/0 : 44[4] -> 43[3] via P2P/IPC
+ip-26-0-165-164:1227468:1227883 [5] NCCL INFO Channel 12/0 : 85[5] -> 84[4] via P2P/IPC
+ip-26-0-165-164:1227467:1227886 [4] NCCL INFO Channel 11/0 : 84[4] -> 83[3] via P2P/IPC
+ip-26-0-161-142:62887:63305 [6] NCCL INFO Channel 08/0 : 46[6] -> 45[5] via P2P/IPC
+ip-26-0-161-142:62885:63304 [4] NCCL INFO Channel 11/0 : 44[4] -> 43[3] via P2P/IPC
+ip-26-0-160-225:124167:124586 [7] NCCL INFO Channel 05/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-225:124164:124585 [4] NCCL INFO Channel 13/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-153:2636091:2636512 [7] NCCL INFO Channel 06/0 : 55[7] -> 48[0] via P2P/IPC
+ip-26-0-165-131:1630209:1630651 [4] NCCL INFO Channel 08/0 : 76[4] -> 75[3] via P2P/IPC
+ip-26-0-165-131:1630212:1630652 [7] NCCL INFO Channel 04/0 : 79[7] -> 72[0] via P2P/IPC
+ip-26-0-161-138:96570:97005 [7] NCCL INFO Channel 06/0 : 39[7] -> 32[0] via P2P/IPC
+ip-26-0-161-178:2815970:2816397 [7] NCCL INFO Channel 06/0 : 63[7] -> 56[0] via P2P/IPC
+ip-26-0-161-138:96567:97008 [4] NCCL INFO Channel 05/0 : 36[4] -> 35[3] via P2P/IPC
+ip-26-0-161-138:96569:97006 [6] NCCL INFO Channel 08/0 : 38[6] -> 37[5] via P2P/IPC
+ip-26-0-161-138:96568:97007 [5] NCCL INFO Channel 08/0 : 37[5] -> 36[4] via P2P/IPC
+ip-26-0-168-30:235652:236066 [4] NCCL INFO Channel 15/0 : 108[4] -> 107[3] via P2P/IPC
+ip-26-0-165-59:1557273:1557685 [7] NCCL INFO Channel 06/0 : 103[7] -> 96[0] via P2P/IPC
+ip-26-0-165-59:1557270:1557688 [4] NCCL INFO Channel 07/0 : 100[4] -> 99[3] via P2P/IPC
+ip-26-0-168-34:272154:272569 [4] NCCL INFO Channel 13/0 : 116[4] -> 115[3] via P2P/IPC
+ip-26-0-165-164:1227470:1227885 [7] NCCL INFO Channel 06/0 : 87[7] -> 80[0] via P2P/IPC
+ip-26-0-168-52:2722049:2722467 [4] NCCL INFO Channel 11/0 : 124[4] -> 123[3] via P2P/IPC
+ip-26-0-161-178:2815967:2816399 [4] NCCL INFO Channel 13/0 : 60[4] -> 59[3] via P2P/IPC
+ip-26-0-161-123:32710:33125 [6] NCCL INFO Channel 05/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-168-34:272157:272572 [7] NCCL INFO Channel 05/0 : 119[7] -> 112[0] via P2P/IPC
+ip-26-0-168-30:235655:236069 [7] NCCL INFO Channel 07/0 : 111[7] -> 104[0] via P2P/IPC
+ip-26-0-161-103:96893:97310 [5] NCCL INFO Channel 12/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-161-103:96892:97308 [4] NCCL INFO Channel 06/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-161-103:96894:97312 [6] NCCL INFO Channel 08/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-161-103:96895:97307 [7] NCCL INFO Channel 04/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-165-202:153566:153980 [4] NCCL INFO Channel 14/0 : 92[4] -> 91[3] via P2P/IPC
+ip-26-0-165-202:153569:153981 [7] NCCL INFO Channel 06/0 : 95[7] -> 88[0] via P2P/IPC
+ip-26-0-160-242:1364909:1365327 [7] NCCL INFO Channel 09/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-160-242:1364908:1365332 [6] NCCL INFO Channel 13/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-165-164:1227467:1227886 [4] NCCL INFO Channel 13/0 : 84[4] -> 83[3] via P2P/IPC
+ip-26-0-160-225:124166:124583 [6] NCCL INFO Channel 08/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-225:124164:124585 [4] NCCL INFO Channel 14/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-78:114519:114931 [7] NCCL INFO Channel 06/0 : 71[7] -> 64[0] via P2P/IPC
+ip-26-0-161-123:32709:33129 [5] NCCL INFO Channel 08/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-160-225:124165:124584 [5] NCCL INFO Channel 12/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-123:32708:33126 [4] NCCL INFO Channel 14/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-161-138:96570:97005 [7] NCCL INFO Channel 07/0 : 39[7] -> 32[0] via P2P/IPC
+ip-26-0-160-225:124166:124583 [6] NCCL INFO Channel 13/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-168-52:2722049:2722467 [4] NCCL INFO Channel 13/0 : 124[4] -> 123[3] via P2P/IPC
+ip-26-0-161-138:96567:97008 [4] NCCL INFO Channel 06/0 : 36[4] -> 35[3] via P2P/IPC
+ip-26-0-160-225:124164:124585 [4] NCCL INFO Channel 15/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-138:96568:97007 [5] NCCL INFO Channel 12/0 : 37[5] -> 36[4] via P2P/IPC
+ip-26-0-161-178:2815970:2816397 [7] NCCL INFO Channel 07/0 : 63[7] -> 56[0] via P2P/IPC
+ip-26-0-168-34:272154:272569 [4] NCCL INFO Channel 14/0 : 116[4] -> 115[3] via P2P/IPC
+ip-26-0-161-138:96569:97006 [6] NCCL INFO Channel 13/0 : 38[6] -> 37[5] via P2P/IPC
+ip-26-0-168-34:272157:272572 [7] NCCL INFO Channel 06/0 : 119[7] -> 112[0] via P2P/IPC
+ip-26-0-165-164:1227470:1227885 [7] NCCL INFO Channel 07/0 : 87[7] -> 80[0] via P2P/IPC
+ip-26-0-165-59:1557273:1557685 [7] NCCL INFO Channel 07/0 : 103[7] -> 96[0] via P2P/IPC
+ip-26-0-165-59:1557270:1557688 [4] NCCL INFO Channel 08/0 : 100[4] -> 99[3] via P2P/IPC
+ip-26-0-161-103:96892:97308 [4] NCCL INFO Channel 07/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-168-52:2722052:2722464 [7] NCCL INFO Channel 06/0 : 127[7] -> 120[0] via P2P/IPC
+ip-26-0-161-178:2815967:2816399 [4] NCCL INFO Channel 14/0 : 60[4] -> 59[3] via P2P/IPC
+ip-26-0-165-131:1630209:1630651 [4] NCCL INFO Channel 11/0 : 76[4] -> 75[3] via P2P/IPC
+ip-26-0-165-131:1630212:1630652 [7] NCCL INFO Channel 05/0 : 79[7] -> 72[0] via P2P/IPC
+ip-26-0-161-123:32710:33125 [6] NCCL INFO Channel 08/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-165-202:153566:153980 [4] NCCL INFO Channel 15/0 : 92[4] -> 91[3] via P2P/IPC
+ip-26-0-165-202:153569:153981 [7] NCCL INFO Channel 07/0 : 95[7] -> 88[0] via P2P/IPC
+ip-26-0-161-103:96894:97312 [6] NCCL INFO Channel 13/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-161-103:96895:97307 [7] NCCL INFO Channel 05/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-165-164:1227467:1227886 [4] NCCL INFO Channel 14/0 : 84[4] -> 83[3] via P2P/IPC
+ip-26-0-161-142:62888:63302 [7] NCCL INFO Channel 09/0 : 47[7] -> 40[0] via P2P/IPC
+ip-26-0-161-142:62887:63305 [6] NCCL INFO Channel 13/0 : 46[6] -> 45[5] via P2P/IPC
+ip-26-0-165-59:1557273:1557685 [7] NCCL INFO Channel 09/0 : 103[7] -> 96[0] via P2P/IPC
+ip-26-0-161-142:62885:63304 [4] NCCL INFO Channel 13/0 : 44[4] -> 43[3] via P2P/IPC
+ip-26-0-161-178:2815970:2816397 [7] NCCL INFO Channel 09/0 : 63[7] -> 56[0] via P2P/IPC
+ip-26-0-161-142:62888:63302 [7] NCCL INFO Channel 10/0 : 47[7] -> 40[0] via P2P/IPC
+ip-26-0-161-123:32709:33129 [5] NCCL INFO Channel 12/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-161-123:32708:33126 [4] NCCL INFO Channel 15/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-161-142:62885:63304 [4] NCCL INFO Channel 14/0 : 44[4] -> 43[3] via P2P/IPC
+ip-26-0-168-34:272154:272569 [4] NCCL INFO Channel 15/0 : 116[4] -> 115[3] via P2P/IPC
+ip-26-0-161-142:62888:63302 [7] NCCL INFO Channel 11/0 : 47[7] -> 40[0] via P2P/IPC
+ip-26-0-168-34:272157:272572 [7] NCCL INFO Channel 07/0 : 119[7] -> 112[0] via P2P/IPC
+ip-26-0-165-164:1227470:1227885 [7] NCCL INFO Channel 09/0 : 87[7] -> 80[0] via P2P/IPC
+ip-26-0-168-52:2722049:2722467 [4] NCCL INFO Channel 14/0 : 124[4] -> 123[3] via P2P/IPC
+ip-26-0-161-103:96892:97308 [4] NCCL INFO Channel 08/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-165-59:1557270:1557688 [4] NCCL INFO Channel 11/0 : 100[4] -> 99[3] via P2P/IPC
+ip-26-0-168-52:2722052:2722464 [7] NCCL INFO Channel 07/0 : 127[7] -> 120[0] via P2P/IPC
+ip-26-0-161-103:96895:97307 [7] NCCL INFO Channel 06/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-161-138:96567:97008 [4] NCCL INFO Channel 07/0 : 36[4] -> 35[3] via P2P/IPC
+ip-26-0-161-138:96570:97005 [7] NCCL INFO Channel 09/0 : 39[7] -> 32[0] via P2P/IPC
+ip-26-0-161-178:2815967:2816399 [4] NCCL INFO Channel 15/0 : 60[4] -> 59[3] via P2P/IPC
+ip-26-0-165-131:1630209:1630651 [4] NCCL INFO Channel 13/0 : 76[4] -> 75[3] via P2P/IPC
+ip-26-0-165-131:1630212:1630652 [7] NCCL INFO Channel 06/0 : 79[7] -> 72[0] via P2P/IPC
+ip-26-0-161-123:32710:33125 [6] NCCL INFO Channel 13/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-165-202:153569:153981 [7] NCCL INFO Channel 09/0 : 95[7] -> 88[0] via P2P/IPC
+ip-26-0-165-164:1227467:1227886 [4] NCCL INFO Channel 15/0 : 84[4] -> 83[3] via P2P/IPC
+ip-26-0-161-142:62885:63304 [4] NCCL INFO Channel 15/0 : 44[4] -> 43[3] via P2P/IPC
+ip-26-0-160-225:124167:124586 [7] NCCL INFO Channel 06/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-164:1227470:1227885 [7] NCCL INFO Channel 10/0 : 87[7] -> 80[0] via P2P/IPC
+ip-26-0-165-59:1557273:1557685 [7] NCCL INFO Channel 10/0 : 103[7] -> 96[0] via P2P/IPC
+ip-26-0-165-59:1557270:1557688 [4] NCCL INFO Channel 13/0 : 100[4] -> 99[3] via P2P/IPC
+ip-26-0-161-178:2815970:2816397 [7] NCCL INFO Channel 10/0 : 63[7] -> 56[0] via P2P/IPC
+ip-26-0-161-138:96567:97008 [4] NCCL INFO Channel 08/0 : 36[4] -> 35[3] via P2P/IPC
+ip-26-0-161-123:32711:33127 [7] NCCL INFO Channel 04/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-161-142:62888:63302 [7] NCCL INFO Channel 12/0 : 47[7] -> 40[0] via P2P/IPC
+ip-26-0-168-34:272157:272572 [7] NCCL INFO Channel 09/0 : 119[7] -> 112[0] via P2P/IPC
+ip-26-0-168-30:235655:236069 [7] NCCL INFO Channel 09/0 : 111[7] -> 104[0] via P2P/IPC
+ip-26-0-168-30:235655:236069 [7] NCCL INFO Channel 10/0 : 111[7] -> 104[0] via P2P/IPC
+ip-26-0-168-52:2722052:2722464 [7] NCCL INFO Channel 09/0 : 127[7] -> 120[0] via P2P/IPC
+ip-26-0-161-103:96895:97307 [7] NCCL INFO Channel 07/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-168-30:235655:236069 [7] NCCL INFO Channel 11/0 : 111[7] -> 104[0] via P2P/IPC
+ip-26-0-161-138:96570:97005 [7] NCCL INFO Channel 10/0 : 39[7] -> 32[0] via P2P/IPC
+ip-26-0-168-30:235655:236069 [7] NCCL INFO Channel 12/0 : 111[7] -> 104[0] via P2P/IPC
+ip-26-0-165-131:1630209:1630651 [4] NCCL INFO Channel 14/0 : 76[4] -> 75[3] via P2P/IPC
+ip-26-0-165-131:1630212:1630652 [7] NCCL INFO Channel 07/0 : 79[7] -> 72[0] via P2P/IPC
+ip-26-0-168-30:235655:236069 [7] NCCL INFO Channel 13/0 : 111[7] -> 104[0] via P2P/IPC
+ip-26-0-165-202:153569:153981 [7] NCCL INFO Channel 10/0 : 95[7] -> 88[0] via P2P/IPC
+ip-26-0-168-30:235655:236069 [7] NCCL INFO Channel 14/0 : 111[7] -> 104[0] via P2P/IPC
+ip-26-0-168-30:235655:236069 [7] NCCL INFO Channel 15/0 : 111[7] -> 104[0] via P2P/IPC
+ip-26-0-165-164:1227470:1227885 [7] NCCL INFO Channel 11/0 : 87[7] -> 80[0] via P2P/IPC
+ip-26-0-165-59:1557273:1557685 [7] NCCL INFO Channel 11/0 : 103[7] -> 96[0] via P2P/IPC
+ip-26-0-165-59:1557270:1557688 [4] NCCL INFO Channel 14/0 : 100[4] -> 99[3] via P2P/IPC
+ip-26-0-161-138:96567:97008 [4] NCCL INFO Channel 11/0 : 36[4] -> 35[3] via P2P/IPC
+ip-26-0-161-178:2815970:2816397 [7] NCCL INFO Channel 11/0 : 63[7] -> 56[0] via P2P/IPC
+ip-26-0-161-78:114519:114931 [7] NCCL INFO Channel 07/0 : 71[7] -> 64[0] via P2P/IPC
+ip-26-0-160-242:1364909:1365327 [7] NCCL INFO Channel 10/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-161-142:62888:63302 [7] NCCL INFO Channel 13/0 : 47[7] -> 40[0] via P2P/IPC
+ip-26-0-168-52:2722052:2722464 [7] NCCL INFO Channel 10/0 : 127[7] -> 120[0] via P2P/IPC
+ip-26-0-161-138:96570:97005 [7] NCCL INFO Channel 11/0 : 39[7] -> 32[0] via P2P/IPC
+ip-26-0-165-131:1630209:1630651 [4] NCCL INFO Channel 15/0 : 76[4] -> 75[3] via P2P/IPC
+ip-26-0-165-131:1630212:1630652 [7] NCCL INFO Channel 09/0 : 79[7] -> 72[0] via P2P/IPC
+ip-26-0-161-123:32711:33127 [7] NCCL INFO Channel 05/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-161-103:96895:97307 [7] NCCL INFO Channel 09/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-165-59:1557270:1557688 [4] NCCL INFO Channel 15/0 : 100[4] -> 99[3] via P2P/IPC
+ip-26-0-165-59:1557273:1557685 [7] NCCL INFO Channel 12/0 : 103[7] -> 96[0] via P2P/IPC
+ip-26-0-161-138:96567:97008 [4] NCCL INFO Channel 13/0 : 36[4] -> 35[3] via P2P/IPC
+ip-26-0-161-138:96570:97005 [7] NCCL INFO Channel 12/0 : 39[7] -> 32[0] via P2P/IPC
+ip-26-0-168-52:2722052:2722464 [7] NCCL INFO Channel 11/0 : 127[7] -> 120[0] via P2P/IPC
+ip-26-0-161-103:96892:97308 [4] NCCL INFO Channel 11/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-165-131:1630212:1630652 [7] NCCL INFO Channel 10/0 : 79[7] -> 72[0] via P2P/IPC
+ip-26-0-161-103:96895:97307 [7] NCCL INFO Channel 10/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-165-59:1557273:1557685 [7] NCCL INFO Channel 13/0 : 103[7] -> 96[0] via P2P/IPC
+ip-26-0-161-138:96567:97008 [4] NCCL INFO Channel 14/0 : 36[4] -> 35[3] via P2P/IPC
+ip-26-0-161-153:2636091:2636512 [7] NCCL INFO Channel 07/0 : 55[7] -> 48[0] via P2P/IPC
+ip-26-0-161-138:96570:97005 [7] NCCL INFO Channel 13/0 : 39[7] -> 32[0] via P2P/IPC
+ip-26-0-165-164:1227470:1227885 [7] NCCL INFO Channel 12/0 : 87[7] -> 80[0] via P2P/IPC
+ip-26-0-165-164:1227470:1227885 [7] NCCL INFO Channel 13/0 : 87[7] -> 80[0] via P2P/IPC
+ip-26-0-168-30:235655:236069 [7] NCCL INFO Channel 00/0 : 111[7] -> 110[6] via P2P/IPC
+ip-26-0-168-52:2722049:2722467 [4] NCCL INFO Channel 15/0 : 124[4] -> 123[3] via P2P/IPC
+ip-26-0-168-52:2722052:2722464 [7] NCCL INFO Channel 12/0 : 127[7] -> 120[0] via P2P/IPC
+ip-26-0-161-103:96892:97308 [4] NCCL INFO Channel 13/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-161-103:96895:97307 [7] NCCL INFO Channel 11/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-161-138:96567:97008 [4] NCCL INFO Channel 15/0 : 36[4] -> 35[3] via P2P/IPC
+ip-26-0-161-138:96570:97005 [7] NCCL INFO Channel 14/0 : 39[7] -> 32[0] via P2P/IPC
+ip-26-0-165-59:1557273:1557685 [7] NCCL INFO Channel 14/0 : 103[7] -> 96[0] via P2P/IPC
+ip-26-0-168-52:2722052:2722464 [7] NCCL INFO Channel 13/0 : 127[7] -> 120[0] via P2P/IPC
+ip-26-0-161-103:96892:97308 [4] NCCL INFO Channel 14/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-161-123:32711:33127 [7] NCCL INFO Channel 06/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-161-103:96895:97307 [7] NCCL INFO Channel 12/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-160-242:1364909:1365327 [7] NCCL INFO Channel 11/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-161-138:96570:97005 [7] NCCL INFO Channel 15/0 : 39[7] -> 32[0] via P2P/IPC
+ip-26-0-168-30:235655:236069 [7] NCCL INFO Channel 06/0 : 111[7] -> 110[6] via P2P/IPC
+ip-26-0-165-131:1630212:1630652 [7] NCCL INFO Channel 11/0 : 79[7] -> 72[0] via P2P/IPC
+ip-26-0-161-142:62888:63302 [7] NCCL INFO Channel 14/0 : 47[7] -> 40[0] via P2P/IPC
+ip-26-0-161-142:62888:63302 [7] NCCL INFO Channel 15/0 : 47[7] -> 40[0] via P2P/IPC
+ip-26-0-165-164:1227470:1227885 [7] NCCL INFO Channel 14/0 : 87[7] -> 80[0] via P2P/IPC
+ip-26-0-161-123:32711:33127 [7] NCCL INFO Channel 07/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-161-103:96895:97307 [7] NCCL INFO Channel 13/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-161-103:96892:97308 [4] NCCL INFO Channel 15/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-165-202:153569:153981 [7] NCCL INFO Channel 11/0 : 95[7] -> 88[0] via P2P/IPC
+ip-26-0-161-123:32711:33127 [7] NCCL INFO Channel 09/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-161-123:32711:33127 [7] NCCL INFO Channel 10/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-161-123:32711:33127 [7] NCCL INFO Channel 11/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-160-225:124167:124586 [7] NCCL INFO Channel 07/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-123:32711:33127 [7] NCCL INFO Channel 12/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-161-123:32711:33127 [7] NCCL INFO Channel 13/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-168-52:2722052:2722464 [7] NCCL INFO Channel 14/0 : 127[7] -> 120[0] via P2P/IPC
+ip-26-0-161-123:32711:33127 [7] NCCL INFO Channel 14/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-161-103:96895:97307 [7] NCCL INFO Channel 14/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-161-123:32711:33127 [7] NCCL INFO Channel 15/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-161-153:2636091:2636512 [7] NCCL INFO Channel 09/0 : 55[7] -> 48[0] via P2P/IPC
+ip-26-0-168-30:235655:236069 [7] NCCL INFO Channel 08/0 : 111[7] -> 110[6] via P2P/IPC
+ip-26-0-161-178:2815970:2816397 [7] NCCL INFO Channel 12/0 : 63[7] -> 56[0] via P2P/IPC
+ip-26-0-168-34:272157:272572 [7] NCCL INFO Channel 10/0 : 119[7] -> 112[0] via P2P/IPC
+ip-26-0-165-131:1630212:1630652 [7] NCCL INFO Channel 12/0 : 79[7] -> 72[0] via P2P/IPC
+ip-26-0-160-242:1364909:1365327 [7] NCCL INFO Channel 12/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-160-225:124167:124586 [7] NCCL INFO Channel 09/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-59:1557273:1557685 [7] NCCL INFO Channel 15/0 : 103[7] -> 96[0] via P2P/IPC
+ip-26-0-161-138:96570:97005 [7] NCCL INFO Channel 00/0 : 39[7] -> 38[6] via P2P/IPC
+ip-26-0-168-30:235655:236069 [7] NCCL INFO Channel 14/0 : 111[7] -> 110[6] via P2P/IPC
+ip-26-0-161-78:114519:114931 [7] NCCL INFO Channel 09/0 : 71[7] -> 64[0] via P2P/IPC
+ip-26-0-160-225:124167:124586 [7] NCCL INFO Channel 10/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-103:96895:97307 [7] NCCL INFO Channel 15/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-161-142:62888:63302 [7] NCCL INFO Channel 00/0 : 47[7] -> 46[6] via P2P/IPC
+ip-26-0-168-52:2722052:2722464 [7] NCCL INFO Channel 15/0 : 127[7] -> 120[0] via P2P/IPC
+ip-26-0-165-202:153569:153981 [7] NCCL INFO Channel 12/0 : 95[7] -> 88[0] via P2P/IPC
+ip-26-0-165-164:1227470:1227885 [7] NCCL INFO Channel 15/0 : 87[7] -> 80[0] via P2P/IPC
+ip-26-0-160-242:1364909:1365327 [7] NCCL INFO Channel 13/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-161-178:2815970:2816397 [7] NCCL INFO Channel 13/0 : 63[7] -> 56[0] via P2P/IPC
+ip-26-0-161-138:96570:97005 [7] NCCL INFO Channel 06/0 : 39[7] -> 38[6] via P2P/IPC
+ip-26-0-161-78:114519:114931 [7] NCCL INFO Channel 10/0 : 71[7] -> 64[0] via P2P/IPC
+ip-26-0-168-52:2722052:2722464 [7] NCCL INFO Channel 00/0 : 127[7] -> 126[6] via P2P/IPC
+ip-26-0-165-131:1630212:1630652 [7] NCCL INFO Channel 13/0 : 79[7] -> 72[0] via P2P/IPC
+ip-26-0-160-225:124167:124586 [7] NCCL INFO Channel 11/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-202:153569:153981 [7] NCCL INFO Channel 13/0 : 95[7] -> 88[0] via P2P/IPC
+ip-26-0-161-142:62888:63302 [7] NCCL INFO Channel 06/0 : 47[7] -> 46[6] via P2P/IPC
+ip-26-0-161-178:2815970:2816397 [7] NCCL INFO Channel 14/0 : 63[7] -> 56[0] via P2P/IPC
+ip-26-0-165-59:1557273:1557685 [7] NCCL INFO Channel 00/0 : 103[7] -> 102[6] via P2P/IPC
+ip-26-0-165-131:1630212:1630652 [7] NCCL INFO Channel 14/0 : 79[7] -> 72[0] via P2P/IPC
+ip-26-0-168-34:272157:272572 [7] NCCL INFO Channel 11/0 : 119[7] -> 112[0] via P2P/IPC
+ip-26-0-165-164:1227470:1227885 [7] NCCL INFO Channel 00/0 : 87[7] -> 86[6] via P2P/IPC
+ip-26-0-161-142:62888:63302 [7] NCCL INFO Channel 08/0 : 47[7] -> 46[6] via P2P/IPC
+ip-26-0-160-242:1364909:1365327 [7] NCCL INFO Channel 14/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-161-153:2636091:2636512 [7] NCCL INFO Channel 10/0 : 55[7] -> 48[0] via P2P/IPC
+ip-26-0-161-78:114519:114931 [7] NCCL INFO Channel 11/0 : 71[7] -> 64[0] via P2P/IPC
+ip-26-0-161-178:2815970:2816397 [7] NCCL INFO Channel 15/0 : 63[7] -> 56[0] via P2P/IPC
+ip-26-0-160-225:124167:124586 [7] NCCL INFO Channel 12/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-103:96895:97307 [7] NCCL INFO Channel 00/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-165-202:153569:153981 [7] NCCL INFO Channel 14/0 : 95[7] -> 88[0] via P2P/IPC
+ip-26-0-168-52:2722052:2722464 [7] NCCL INFO Channel 06/0 : 127[7] -> 126[6] via P2P/IPC
+ip-26-0-160-225:124167:124586 [7] NCCL INFO Channel 13/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-59:1557273:1557685 [7] NCCL INFO Channel 06/0 : 103[7] -> 102[6] via P2P/IPC
+ip-26-0-160-242:1364909:1365327 [7] NCCL INFO Channel 15/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-161-103:96895:97307 [7] NCCL INFO Channel 06/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-161-138:96570:97005 [7] NCCL INFO Channel 08/0 : 39[7] -> 38[6] via P2P/IPC
+ip-26-0-161-142:62888:63302 [7] NCCL INFO Channel 14/0 : 47[7] -> 46[6] via P2P/IPC
+ip-26-0-165-131:1630212:1630652 [7] NCCL INFO Channel 15/0 : 79[7] -> 72[0] via P2P/IPC
+ip-26-0-161-78:114519:114931 [7] NCCL INFO Channel 12/0 : 71[7] -> 64[0] via P2P/IPC
+ip-26-0-161-153:2636091:2636512 [7] NCCL INFO Channel 11/0 : 55[7] -> 48[0] via P2P/IPC
+ip-26-0-160-225:124167:124586 [7] NCCL INFO Channel 14/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-202:153569:153981 [7] NCCL INFO Channel 15/0 : 95[7] -> 88[0] via P2P/IPC
+ip-26-0-165-164:1227470:1227885 [7] NCCL INFO Channel 06/0 : 87[7] -> 86[6] via P2P/IPC
+ip-26-0-161-78:114519:114931 [7] NCCL INFO Channel 13/0 : 71[7] -> 64[0] via P2P/IPC
+ip-26-0-165-59:1557273:1557685 [7] NCCL INFO Channel 08/0 : 103[7] -> 102[6] via P2P/IPC
+ip-26-0-168-52:2722052:2722464 [7] NCCL INFO Channel 08/0 : 127[7] -> 126[6] via P2P/IPC
+ip-26-0-161-103:96895:97307 [7] NCCL INFO Channel 08/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-165-164:1227470:1227885 [7] NCCL INFO Channel 08/0 : 87[7] -> 86[6] via P2P/IPC
+ip-26-0-161-138:96570:97005 [7] NCCL INFO Channel 14/0 : 39[7] -> 38[6] via P2P/IPC
+ip-26-0-160-225:124167:124586 [7] NCCL INFO Channel 15/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-153:2636091:2636512 [7] NCCL INFO Channel 12/0 : 55[7] -> 48[0] via P2P/IPC
+ip-26-0-168-34:272157:272572 [7] NCCL INFO Channel 12/0 : 119[7] -> 112[0] via P2P/IPC
+ip-26-0-165-131:1630212:1630652 [7] NCCL INFO Channel 00/0 : 79[7] -> 78[6] via P2P/IPC
+ip-26-0-165-59:1557273:1557685 [7] NCCL INFO Channel 14/0 : 103[7] -> 102[6] via P2P/IPC
+ip-26-0-161-103:96895:97307 [7] NCCL INFO Channel 14/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-165-164:1227470:1227885 [7] NCCL INFO Channel 14/0 : 87[7] -> 86[6] via P2P/IPC
+ip-26-0-168-34:272157:272572 [7] NCCL INFO Channel 13/0 : 119[7] -> 112[0] via P2P/IPC
+ip-26-0-168-52:2722052:2722464 [7] NCCL INFO Channel 14/0 : 127[7] -> 126[6] via P2P/IPC
+ip-26-0-161-153:2636091:2636512 [7] NCCL INFO Channel 13/0 : 55[7] -> 48[0] via P2P/IPC
+ip-26-0-161-123:32711:33127 [7] NCCL INFO Channel 00/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-168-34:272157:272572 [7] NCCL INFO Channel 14/0 : 119[7] -> 112[0] via P2P/IPC
+ip-26-0-165-131:1630212:1630652 [7] NCCL INFO Channel 06/0 : 79[7] -> 78[6] via P2P/IPC
+ip-26-0-160-242:1364909:1365327 [7] NCCL INFO Channel 00/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-161-78:114519:114931 [7] NCCL INFO Channel 14/0 : 71[7] -> 64[0] via P2P/IPC
+ip-26-0-161-153:2636091:2636512 [7] NCCL INFO Channel 14/0 : 55[7] -> 48[0] via P2P/IPC
+ip-26-0-165-131:1630212:1630652 [7] NCCL INFO Channel 08/0 : 79[7] -> 78[6] via P2P/IPC
+ip-26-0-161-178:2815970:2816397 [7] NCCL INFO Channel 00/0 : 63[7] -> 62[6] via P2P/IPC
+ip-26-0-160-225:124167:124586 [7] NCCL INFO Channel 00/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-165-202:153569:153981 [7] NCCL INFO Channel 00/0 : 95[7] -> 94[6] via P2P/IPC
+ip-26-0-168-34:272157:272572 [7] NCCL INFO Channel 15/0 : 119[7] -> 112[0] via P2P/IPC
+ip-26-0-165-131:1630212:1630652 [7] NCCL INFO Channel 14/0 : 79[7] -> 78[6] via P2P/IPC
+ip-26-0-161-153:2636091:2636512 [7] NCCL INFO Channel 15/0 : 55[7] -> 48[0] via P2P/IPC
+ip-26-0-161-178:2815970:2816397 [7] NCCL INFO Channel 06/0 : 63[7] -> 62[6] via P2P/IPC
+ip-26-0-160-225:124167:124586 [7] NCCL INFO Channel 06/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-242:1364909:1365327 [7] NCCL INFO Channel 06/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-161-78:114519:114931 [7] NCCL INFO Channel 15/0 : 71[7] -> 64[0] via P2P/IPC
+ip-26-0-161-178:2815970:2816397 [7] NCCL INFO Channel 08/0 : 63[7] -> 62[6] via P2P/IPC
+ip-26-0-165-202:153569:153981 [7] NCCL INFO Channel 06/0 : 95[7] -> 94[6] via P2P/IPC
+ip-26-0-168-34:272157:272572 [7] NCCL INFO Channel 00/0 : 119[7] -> 118[6] via P2P/IPC
+ip-26-0-161-123:32711:33127 [7] NCCL INFO Channel 06/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-168-34:272157:272572 [7] NCCL INFO Channel 06/0 : 119[7] -> 118[6] via P2P/IPC
+ip-26-0-165-202:153569:153981 [7] NCCL INFO Channel 08/0 : 95[7] -> 94[6] via P2P/IPC
+ip-26-0-168-34:272157:272572 [7] NCCL INFO Channel 08/0 : 119[7] -> 118[6] via P2P/IPC
+ip-26-0-160-242:1364909:1365327 [7] NCCL INFO Channel 08/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-161-78:114519:114931 [7] NCCL INFO Channel 00/0 : 71[7] -> 70[6] via P2P/IPC
+ip-26-0-161-123:32711:33127 [7] NCCL INFO Channel 08/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-161-178:2815970:2816397 [7] NCCL INFO Channel 14/0 : 63[7] -> 62[6] via P2P/IPC
+ip-26-0-160-242:1364909:1365327 [7] NCCL INFO Channel 14/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-161-78:114519:114931 [7] NCCL INFO Channel 06/0 : 71[7] -> 70[6] via P2P/IPC
+ip-26-0-161-153:2636091:2636512 [7] NCCL INFO Channel 00/0 : 55[7] -> 54[6] via P2P/IPC
+ip-26-0-165-202:153569:153981 [7] NCCL INFO Channel 14/0 : 95[7] -> 94[6] via P2P/IPC
+ip-26-0-160-225:124167:124586 [7] NCCL INFO Channel 08/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-168-34:272157:272572 [7] NCCL INFO Channel 14/0 : 119[7] -> 118[6] via P2P/IPC
+ip-26-0-161-123:32711:33127 [7] NCCL INFO Channel 14/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-161-153:2636091:2636512 [7] NCCL INFO Channel 06/0 : 55[7] -> 54[6] via P2P/IPC
+ip-26-0-161-153:2636091:2636512 [7] NCCL INFO Channel 08/0 : 55[7] -> 54[6] via P2P/IPC
+ip-26-0-160-225:124167:124586 [7] NCCL INFO Channel 14/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-78:114519:114931 [7] NCCL INFO Channel 08/0 : 71[7] -> 70[6] via P2P/IPC
+ip-26-0-161-153:2636091:2636512 [7] NCCL INFO Channel 14/0 : 55[7] -> 54[6] via P2P/IPC
+ip-26-0-161-78:114519:114931 [7] NCCL INFO Channel 14/0 : 71[7] -> 70[6] via P2P/IPC
+ip-26-0-168-52:2722047:2722468 [2] NCCL INFO Connected all trees
+ip-26-0-168-52:2722047:2722468 [2] NCCL INFO NVLS comm 0x967a780 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-225:124162:124587 [2] NCCL INFO Connected all trees
+ip-26-0-160-225:124162:124587 [2] NCCL INFO NVLS comm 0xa5007c0 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-225:124165:124584 [5] NCCL INFO Connected all trees
+ip-26-0-160-225:124165:124584 [5] NCCL INFO NVLS comm 0x8a90b80 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-225:124166:124583 [6] NCCL INFO Connected all trees
+ip-26-0-160-225:124166:124583 [6] NCCL INFO NVLS comm 0x99b4f20 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-225:124161:124582 [1] NCCL INFO Connected all trees
+ip-26-0-168-52:2722046:2722466 [1] NCCL INFO Connected all trees
+ip-26-0-160-225:124161:124582 [1] NCCL INFO NVLS comm 0x8a53580 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-168-52:2722046:2722466 [1] NCCL INFO NVLS comm 0x9a58530 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-225:124160:124581 [0] NCCL INFO Connected all trees
+ip-26-0-160-225:124160:124581 [0] NCCL INFO NVLS comm 0x94a3500 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-168-52:2722050:2722471 [5] NCCL INFO Connected all trees
+ip-26-0-168-52:2722050:2722471 [5] NCCL INFO NVLS comm 0x88d9840 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-168-52:2722051:2722465 [6] NCCL INFO Connected all trees
+ip-26-0-168-52:2722051:2722465 [6] NCCL INFO NVLS comm 0x94b31c0 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-168-52:2722049:2722467 [4] NCCL INFO Connected all trees
+ip-26-0-168-52:2722049:2722467 [4] NCCL INFO NVLS comm 0x8a76060 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-225:124164:124585 [4] NCCL INFO Connected all trees
+ip-26-0-160-225:124164:124585 [4] NCCL INFO NVLS comm 0x91c6330 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-168-52:2722045:2722470 [0] NCCL INFO Connected all trees
+ip-26-0-160-225:124163:124588 [3] NCCL INFO Connected all trees
+ip-26-0-168-52:2722045:2722470 [0] NCCL INFO NVLS comm 0x8786e80 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-225:124163:124588 [3] NCCL INFO NVLS comm 0x9726fa0 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-168-52:2722052:2722464 [7] NCCL INFO Connected all trees
+ip-26-0-168-52:2722052:2722464 [7] NCCL INFO NVLS comm 0x98d4300 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-103:96890:97313 [2] NCCL INFO Connected all trees
+ip-26-0-161-103:96890:97313 [2] NCCL INFO NVLS comm 0x9078a80 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-59:1557268:1557686 [2] NCCL INFO Connected all trees
+ip-26-0-165-59:1557268:1557686 [2] NCCL INFO NVLS comm 0x8d9f540 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-168-30:235650:236065 [2] NCCL INFO Connected all trees
+ip-26-0-168-30:235650:236065 [2] NCCL INFO NVLS comm 0x914c6b0 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-168-30:235652:236066 [4] NCCL INFO Connected all trees
+ip-26-0-168-30:235652:236066 [4] NCCL INFO NVLS comm 0x9edc280 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-59:1557271:1557690 [5] NCCL INFO Connected all trees
+ip-26-0-165-59:1557271:1557690 [5] NCCL INFO NVLS comm 0x8702fb0 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-178:2815965:2816395 [2] NCCL INFO Connected all trees
+ip-26-0-161-178:2815965:2816395 [2] NCCL INFO NVLS comm 0x8e2eb00 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-168-52:2722048:2722469 [3] NCCL INFO Connected all trees
+ip-26-0-161-78:114514:114932 [2] NCCL INFO Connected all trees
+ip-26-0-168-52:2722048:2722469 [3] NCCL INFO NVLS comm 0x88a3670 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-78:114514:114932 [2] NCCL INFO NVLS comm 0x9f4e0f0 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-123:32708:33126 [4] NCCL INFO Connected all trees
+ip-26-0-161-123:32706:33122 [2] NCCL INFO Connected all trees
+ip-26-0-161-123:32708:33126 [4] NCCL INFO NVLS comm 0xa4e8b60 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-142:62883:63301 [2] NCCL INFO Connected all trees
+ip-26-0-161-123:32706:33122 [2] NCCL INFO NVLS comm 0x9c6ed80 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-142:62883:63301 [2] NCCL INFO NVLS comm 0x8d92c70 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-78:114516:114934 [4] NCCL INFO Connected all trees
+ip-26-0-160-242:1364904:1365335 [2] NCCL INFO Connected all trees
+ip-26-0-161-138:96565:97004 [2] NCCL INFO Connected all trees
+ip-26-0-168-34:272152:272575 [2] NCCL INFO Connected all trees
+ip-26-0-161-78:114516:114934 [4] NCCL INFO NVLS comm 0x95c5e60 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-242:1364904:1365335 [2] NCCL INFO NVLS comm 0x9899fb0 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-138:96565:97004 [2] NCCL INFO NVLS comm 0xa462c20 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-168-30:235649:236071 [1] NCCL INFO Connected all trees
+ip-26-0-168-34:272152:272575 [2] NCCL INFO NVLS comm 0xa6c77f0 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-59:1557266:1557689 [0] NCCL INFO Connected all trees
+ip-26-0-168-30:235649:236071 [1] NCCL INFO NVLS comm 0x9e33a20 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-78:114517:114935 [5] NCCL INFO Connected all trees
+ip-26-0-165-164:1227465:1227881 [2] NCCL INFO Connected all trees
+ip-26-0-165-59:1557273:1557685 [7] NCCL INFO Connected all trees
+ip-26-0-165-59:1557266:1557689 [0] NCCL INFO NVLS comm 0xa380e00 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-78:114517:114935 [5] NCCL INFO NVLS comm 0xa4db9a0 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-164:1227465:1227881 [2] NCCL INFO NVLS comm 0x8e8ccf0 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-59:1557273:1557685 [7] NCCL INFO NVLS comm 0xa24f4e0 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-168-30:235653:236067 [5] NCCL INFO Connected all trees
+ip-26-0-160-225:124167:124586 [7] NCCL INFO Connected all trees
+ip-26-0-168-30:235653:236067 [5] NCCL INFO NVLS comm 0x935a0b0 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-225:124167:124586 [7] NCCL INFO NVLS comm 0x9e9e880 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-168-30:235651:236068 [3] NCCL INFO Connected all trees
+ip-26-0-168-30:235651:236068 [3] NCCL INFO NVLS comm 0x912b830 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-142:62885:63304 [4] NCCL INFO Connected all trees
+ip-26-0-161-142:62885:63304 [4] NCCL INFO NVLS comm 0xa074600 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-168-30:235655:236069 [7] NCCL INFO Connected all trees
+ip-26-0-168-30:235655:236069 [7] NCCL INFO NVLS comm 0x96a2a80 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-123:32707:33124 [3] NCCL INFO Connected all trees
+ip-26-0-161-123:32707:33124 [3] NCCL INFO NVLS comm 0x8ad9f30 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-142:62884:63299 [3] NCCL INFO Connected all trees
+ip-26-0-161-142:62884:63299 [3] NCCL INFO NVLS comm 0x9811ff0 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-78:114515:114936 [3] NCCL INFO Connected all trees
+ip-26-0-161-78:114515:114936 [3] NCCL INFO NVLS comm 0x902dae0 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-103:96892:97308 [4] NCCL INFO Connected all trees
+ip-26-0-161-103:96892:97308 [4] NCCL INFO NVLS comm 0x9df5b40 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-242:1364906:1365331 [4] NCCL INFO Connected all trees
+ip-26-0-165-164:1227467:1227886 [4] NCCL INFO Connected all trees
+ip-26-0-160-242:1364906:1365331 [4] NCCL INFO NVLS comm 0xa1ed2c0 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-164:1227467:1227886 [4] NCCL INFO NVLS comm 0x945c240 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-142:62886:63303 [5] NCCL INFO Connected all trees
+ip-26-0-161-103:96891:97311 [3] NCCL INFO Connected all trees
+ip-26-0-161-142:62888:63302 [7] NCCL INFO Connected all trees
+ip-26-0-161-142:62887:63305 [6] NCCL INFO Connected all trees
+ip-26-0-165-164:1227468:1227883 [5] NCCL INFO Connected all trees
+ip-26-0-165-131:1630207:1630648 [2] NCCL INFO Connected all trees
+ip-26-0-165-202:153564:153984 [2] NCCL INFO Connected all trees
+ip-26-0-161-142:62886:63303 [5] NCCL INFO NVLS comm 0xa047900 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-142:62888:63302 [7] NCCL INFO NVLS comm 0x93c50f0 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-103:96891:97311 [3] NCCL INFO NVLS comm 0x8ec46b0 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-202:153564:153984 [2] NCCL INFO NVLS comm 0x8a1afc0 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-142:62887:63305 [6] NCCL INFO NVLS comm 0x88e3b40 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-164:1227468:1227883 [5] NCCL INFO NVLS comm 0x9de3570 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-131:1630207:1630648 [2] NCCL INFO NVLS comm 0x97a3d30 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-168-30:235648:236064 [0] NCCL INFO Connected all trees
+ip-26-0-161-103:96888:97306 [0] NCCL INFO Connected all trees
+ip-26-0-168-30:235648:236064 [0] NCCL INFO NVLS comm 0x8b6eec0 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-103:96888:97306 [0] NCCL INFO NVLS comm 0x9097a60 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-164:1227466:1227887 [3] NCCL INFO Connected all trees
+ip-26-0-161-103:96893:97310 [5] NCCL INFO Connected all trees
+ip-26-0-161-103:96893:97310 [5] NCCL INFO NVLS comm 0x8943870 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-164:1227466:1227887 [3] NCCL INFO NVLS comm 0x90d62c0 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-123:32709:33129 [5] NCCL INFO Connected all trees
+ip-26-0-165-59:1557270:1557688 [4] NCCL INFO Connected all trees
+ip-26-0-161-123:32709:33129 [5] NCCL INFO NVLS comm 0xa49ae70 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-242:1364907:1365329 [5] NCCL INFO Connected all trees
+ip-26-0-161-178:2815966:2816396 [3] NCCL INFO Connected all trees
+ip-26-0-165-59:1557270:1557688 [4] NCCL INFO NVLS comm 0x8b767c0 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-242:1364907:1365329 [5] NCCL INFO NVLS comm 0x88ea1b0 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-242:1364905:1365328 [3] NCCL INFO Connected all trees
+ip-26-0-161-178:2815966:2816396 [3] NCCL INFO NVLS comm 0x96fb430 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-78:114519:114931 [7] NCCL INFO Connected all trees
+ip-26-0-160-242:1364905:1365328 [3] NCCL INFO NVLS comm 0x9a8b4e0 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-78:114512:114930 [0] NCCL INFO Connected all trees
+ip-26-0-161-78:114519:114931 [7] NCCL INFO NVLS comm 0x87170f0 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-78:114512:114930 [0] NCCL INFO NVLS comm 0x94e08f0 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-164:1227470:1227885 [7] NCCL INFO Connected all trees
+ip-26-0-165-164:1227470:1227885 [7] NCCL INFO NVLS comm 0x94f35b0 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-59:1557267:1557683 [1] NCCL INFO Connected all trees
+ip-26-0-165-59:1557267:1557683 [1] NCCL INFO NVLS comm 0x8af4bf0 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-103:96895:97307 [7] NCCL INFO Connected all trees
+ip-26-0-160-242:1364902:1365333 [0] NCCL INFO Connected all trees
+ip-26-0-161-103:96895:97307 [7] NCCL INFO NVLS comm 0xa67bcf0 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-59:1557269:1557684 [3] NCCL INFO Connected all trees
+ip-26-0-165-131:1630210:1630653 [5] NCCL INFO Connected all trees
+ip-26-0-161-153:2636086:2636507 [2] NCCL INFO Connected all trees
+ip-26-0-160-242:1364902:1365333 [0] NCCL INFO NVLS comm 0x9a0bd70 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-59:1557269:1557684 [3] NCCL INFO NVLS comm 0xa1496b0 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-131:1630210:1630653 [5] NCCL INFO NVLS comm 0xa0edf00 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-153:2636086:2636507 [2] NCCL INFO NVLS comm 0x9ba9780 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-242:1364908:1365332 [6] NCCL INFO Connected all trees
+ip-26-0-161-123:32711:33127 [7] NCCL INFO Connected all trees
+ip-26-0-160-242:1364903:1365334 [1] NCCL INFO Connected all trees
+ip-26-0-160-242:1364908:1365332 [6] NCCL INFO NVLS comm 0xa6080b0 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-123:32711:33127 [7] NCCL INFO NVLS comm 0xa0d8f70 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-242:1364903:1365334 [1] NCCL INFO NVLS comm 0x9d31280 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-123:32704:33123 [0] NCCL INFO Connected all trees
+ip-26-0-161-123:32704:33123 [0] NCCL INFO NVLS comm 0xa6bba20 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-168-34:272151:272571 [1] NCCL INFO Connected all trees
+ip-26-0-165-131:1630212:1630652 [7] NCCL INFO Connected all trees
+ip-26-0-161-138:96563:97003 [0] NCCL INFO Connected all trees
+ip-26-0-168-34:272154:272569 [4] NCCL INFO Connected all trees
+ip-26-0-168-34:272153:272573 [3] NCCL INFO Connected all trees
+ip-26-0-165-131:1630212:1630652 [7] NCCL INFO NVLS comm 0x891b6c0 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-168-34:272151:272571 [1] NCCL INFO NVLS comm 0x96eebf0 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-138:96563:97003 [0] NCCL INFO NVLS comm 0x9c090c0 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-168-34:272154:272569 [4] NCCL INFO NVLS comm 0x94d3e70 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-164:1227463:1227880 [0] NCCL INFO Connected all trees
+ip-26-0-165-131:1630209:1630651 [4] NCCL INFO Connected all trees
+ip-26-0-168-34:272153:272573 [3] NCCL INFO NVLS comm 0x880e9b0 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-78:114518:114937 [6] NCCL INFO Connected all trees
+ip-26-0-161-178:2815967:2816399 [4] NCCL INFO Connected all trees
+ip-26-0-165-164:1227463:1227880 [0] NCCL INFO NVLS comm 0x89f8580 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-131:1630209:1630651 [4] NCCL INFO NVLS comm 0x940d970 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-59:1557272:1557687 [6] NCCL INFO Connected all trees
+ip-26-0-161-178:2815967:2816399 [4] NCCL INFO NVLS comm 0x88842c0 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-78:114518:114937 [6] NCCL INFO NVLS comm 0x9acb870 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-78:114513:114933 [1] NCCL INFO Connected all trees
+ip-26-0-165-59:1557272:1557687 [6] NCCL INFO NVLS comm 0x9666ec0 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-164:1227464:1227884 [1] NCCL INFO Connected all trees
+ip-26-0-161-178:2815970:2816397 [7] NCCL INFO Connected all trees
+ip-26-0-168-34:272155:272574 [5] NCCL INFO Connected all trees
+ip-26-0-161-78:114513:114933 [1] NCCL INFO NVLS comm 0xa08d8f0 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-178:2815970:2816397 [7] NCCL INFO NVLS comm 0x931bca0 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-164:1227464:1227884 [1] NCCL INFO NVLS comm 0x92bcaf0 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-168-34:272155:272574 [5] NCCL INFO NVLS comm 0x93fcd30 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-103:96889:97309 [1] NCCL INFO Connected all trees
+ip-26-0-161-103:96889:97309 [1] NCCL INFO NVLS comm 0x8a2fc20 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-131:1630206:1630647 [1] NCCL INFO Connected all trees
+ip-26-0-165-202:153566:153980 [4] NCCL INFO Connected all trees
+ip-26-0-165-131:1630206:1630647 [1] NCCL INFO NVLS comm 0x9d917e0 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-202:153566:153980 [4] NCCL INFO NVLS comm 0x8c5b600 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-123:32710:33125 [6] NCCL INFO Connected all trees
+ip-26-0-161-123:32710:33125 [6] NCCL INFO NVLS comm 0x9b78cf0 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-178:2815968:2816394 [5] NCCL INFO Connected all trees
+ip-26-0-160-242:1364909:1365327 [7] NCCL INFO Connected all trees
+ip-26-0-161-153:2636087:2636506 [3] NCCL INFO Connected all trees
+ip-26-0-161-138:96566:97010 [3] NCCL INFO Connected all trees
+ip-26-0-161-178:2815968:2816394 [5] NCCL INFO NVLS comm 0xa2e3d30 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-138:96564:97009 [1] NCCL INFO Connected all trees
+ip-26-0-161-142:62881:63298 [0] NCCL INFO Connected all trees
+ip-26-0-160-242:1364909:1365327 [7] NCCL INFO NVLS comm 0xa0bdc30 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-142:62882:63300 [1] NCCL INFO Connected all trees
+ip-26-0-161-103:96894:97312 [6] NCCL INFO Connected all trees
+ip-26-0-161-153:2636088:2636510 [4] NCCL INFO Connected all trees
+ip-26-0-168-34:272150:272576 [0] NCCL INFO Connected all trees
+ip-26-0-161-153:2636087:2636506 [3] NCCL INFO NVLS comm 0x8b01d20 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-138:96566:97010 [3] NCCL INFO NVLS comm 0x9576100 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-142:62881:63298 [0] NCCL INFO NVLS comm 0xa365ab0 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-202:153567:153979 [5] NCCL INFO Connected all trees
+ip-26-0-161-138:96564:97009 [1] NCCL INFO NVLS comm 0x987dbe0 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-142:62882:63300 [1] NCCL INFO NVLS comm 0x912d0f0 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-103:96894:97312 [6] NCCL INFO NVLS comm 0x8cd64e0 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-131:1630205:1630646 [0] NCCL INFO Connected all trees
+ip-26-0-168-34:272150:272576 [0] NCCL INFO NVLS comm 0x9fbd540 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-153:2636088:2636510 [4] NCCL INFO NVLS comm 0x94f8db0 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-202:153567:153979 [5] NCCL INFO NVLS comm 0xa41d470 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-131:1630205:1630646 [0] NCCL INFO NVLS comm 0x90c8640 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-123:32705:33128 [1] NCCL INFO Connected all trees
+ip-26-0-161-123:32705:33128 [1] NCCL INFO NVLS comm 0x9c4b570 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-178:2815963:2816400 [0] NCCL INFO Connected all trees
+ip-26-0-161-178:2815963:2816400 [0] NCCL INFO NVLS comm 0x95d2fb0 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-164:1227469:1227882 [6] NCCL INFO Connected all trees
+ip-26-0-165-202:153562:153983 [0] NCCL INFO Connected all trees
+ip-26-0-168-30:235654:236070 [6] NCCL INFO Connected all trees
+ip-26-0-165-164:1227469:1227882 [6] NCCL INFO NVLS comm 0x871acf0 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-202:153562:153983 [0] NCCL INFO NVLS comm 0x96ebb40 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-168-30:235654:236070 [6] NCCL INFO NVLS comm 0x9bac4c0 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-202:153565:153985 [3] NCCL INFO Connected all trees
+ip-26-0-161-138:96568:97007 [5] NCCL INFO Connected all trees
+ip-26-0-165-202:153565:153985 [3] NCCL INFO NVLS comm 0x9bf2bb0 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-138:96568:97007 [5] NCCL INFO NVLS comm 0x8b312f0 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-131:1630211:1630649 [6] NCCL INFO Connected all trees
+ip-26-0-161-138:96570:97005 [7] NCCL INFO Connected all trees
+ip-26-0-165-131:1630211:1630649 [6] NCCL INFO NVLS comm 0xa6c7430 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-138:96570:97005 [7] NCCL INFO NVLS comm 0xa31fb20 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-138:96569:97006 [6] NCCL INFO Connected all trees
+ip-26-0-161-153:2636089:2636508 [5] NCCL INFO Connected all trees
+ip-26-0-161-138:96569:97006 [6] NCCL INFO NVLS comm 0x90dbd30 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-153:2636089:2636508 [5] NCCL INFO NVLS comm 0x897b060 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-153:2636091:2636512 [7] NCCL INFO Connected all trees
+ip-26-0-161-153:2636091:2636512 [7] NCCL INFO NVLS comm 0x925c9f0 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-178:2815964:2816401 [1] NCCL INFO Connected all trees
+ip-26-0-165-131:1630208:1630650 [3] NCCL INFO Connected all trees
+ip-26-0-161-178:2815964:2816401 [1] NCCL INFO NVLS comm 0x89ef570 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-131:1630208:1630650 [3] NCCL INFO NVLS comm 0x87280e0 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-168-34:272156:272570 [6] NCCL INFO Connected all trees
+ip-26-0-161-138:96567:97008 [4] NCCL INFO Connected all trees
+ip-26-0-168-34:272156:272570 [6] NCCL INFO NVLS comm 0xa3f1b00 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-138:96567:97008 [4] NCCL INFO NVLS comm 0x8aca3d0 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-153:2636085:2636509 [1] NCCL INFO Connected all trees
+ip-26-0-161-153:2636084:2636511 [0] NCCL INFO Connected all trees
+ip-26-0-161-153:2636085:2636509 [1] NCCL INFO NVLS comm 0xa41e5f0 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-153:2636090:2636505 [6] NCCL INFO Connected all trees
+ip-26-0-161-153:2636084:2636511 [0] NCCL INFO NVLS comm 0x91a14c0 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-153:2636090:2636505 [6] NCCL INFO NVLS comm 0x9c681e0 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-202:153568:153982 [6] NCCL INFO Connected all trees
+ip-26-0-165-202:153568:153982 [6] NCCL INFO NVLS comm 0x8a079f0 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-168-34:272157:272572 [7] NCCL INFO Connected all trees
+ip-26-0-168-34:272157:272572 [7] NCCL INFO NVLS comm 0x965ac80 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-178:2815969:2816398 [6] NCCL INFO Connected all trees
+ip-26-0-161-178:2815969:2816398 [6] NCCL INFO NVLS comm 0x9caa670 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-202:153563:153986 [1] NCCL INFO Connected all trees
+ip-26-0-165-202:153569:153981 [7] NCCL INFO Connected all trees
+ip-26-0-165-202:153563:153986 [1] NCCL INFO NVLS comm 0xa623b70 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-202:153569:153981 [7] NCCL INFO NVLS comm 0x8981bb0 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-168-52:2722050:2722471 [5] NCCL INFO Channel 00/0 : 117[5] -> 125[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-52:2722050:2722471 [5] NCCL INFO Channel 02/0 : 117[5] -> 125[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-52:2722052:2722464 [7] NCCL INFO Channel 00/0 : 119[7] -> 127[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722052:2722464 [7] NCCL INFO Channel 02/0 : 119[7] -> 127[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722050:2722471 [5] NCCL INFO Channel 04/0 : 117[5] -> 125[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-52:2722048:2722469 [3] NCCL INFO Channel 00/0 : 115[3] -> 123[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-52:2722046:2722466 [1] NCCL INFO Channel 00/0 : 113[1] -> 121[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-52:2722049:2722467 [4] NCCL INFO Channel 00/0 : 116[4] -> 124[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722052:2722464 [7] NCCL INFO Channel 04/0 : 119[7] -> 127[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722051:2722465 [6] NCCL INFO Channel 00/0 : 118[6] -> 126[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-52:2722050:2722471 [5] NCCL INFO Channel 06/0 : 117[5] -> 125[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-52:2722048:2722469 [3] NCCL INFO Channel 02/0 : 115[3] -> 123[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-52:2722046:2722466 [1] NCCL INFO Channel 02/0 : 113[1] -> 121[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-52:2722052:2722464 [7] NCCL INFO Channel 06/0 : 119[7] -> 127[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722049:2722467 [4] NCCL INFO Channel 02/0 : 116[4] -> 124[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722051:2722465 [6] NCCL INFO Channel 02/0 : 118[6] -> 126[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-52:2722050:2722471 [5] NCCL INFO Channel 08/0 : 117[5] -> 125[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-52:2722048:2722469 [3] NCCL INFO Channel 04/0 : 115[3] -> 123[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-52:2722046:2722466 [1] NCCL INFO Channel 04/0 : 113[1] -> 121[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-52:2722047:2722468 [2] NCCL INFO Channel 00/0 : 114[2] -> 122[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-52:2722051:2722465 [6] NCCL INFO Channel 04/0 : 118[6] -> 126[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-52:2722049:2722467 [4] NCCL INFO Channel 06/0 : 116[4] -> 124[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722052:2722464 [7] NCCL INFO Channel 08/0 : 119[7] -> 127[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722050:2722471 [5] NCCL INFO Channel 10/0 : 117[5] -> 125[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-52:2722048:2722469 [3] NCCL INFO Channel 06/0 : 115[3] -> 123[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-52:2722046:2722466 [1] NCCL INFO Channel 06/0 : 113[1] -> 121[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-52:2722047:2722468 [2] NCCL INFO Channel 04/0 : 114[2] -> 122[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-52:2722045:2722470 [0] NCCL INFO Channel 02/0 : 112[0] -> 120[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-52:2722051:2722465 [6] NCCL INFO Channel 08/0 : 118[6] -> 126[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:124166:124583 [6] NCCL INFO Channel 01/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:124160:124581 [0] NCCL INFO Channel 01/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:124166:124583 [6] NCCL INFO Channel 03/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:124160:124581 [0] NCCL INFO Channel 03/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:124166:124583 [6] NCCL INFO Channel 05/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:124164:124585 [4] NCCL INFO Channel 01/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:124160:124581 [0] NCCL INFO Channel 05/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:124166:124583 [6] NCCL INFO Channel 07/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:124162:124587 [2] NCCL INFO Channel 01/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:124164:124585 [4] NCCL INFO Channel 03/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:124160:124581 [0] NCCL INFO Channel 07/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:124166:124583 [6] NCCL INFO Channel 09/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:124162:124587 [2] NCCL INFO Channel 03/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:124164:124585 [4] NCCL INFO Channel 05/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:124160:124581 [0] NCCL INFO Channel 09/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:124166:124583 [6] NCCL INFO Channel 11/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:124162:124587 [2] NCCL INFO Channel 05/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:124161:124582 [1] NCCL INFO Channel 01/0 : 1[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124164:124585 [4] NCCL INFO Channel 07/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:124167:124586 [7] NCCL INFO Channel 01/0 : 7[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124160:124581 [0] NCCL INFO Channel 11/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:124166:124583 [6] NCCL INFO Channel 13/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:124162:124587 [2] NCCL INFO Channel 07/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:124161:124582 [1] NCCL INFO Channel 03/0 : 1[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124164:124585 [4] NCCL INFO Channel 09/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:124167:124586 [7] NCCL INFO Channel 03/0 : 7[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124160:124581 [0] NCCL INFO Channel 13/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:124166:124583 [6] NCCL INFO Channel 15/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:124162:124587 [2] NCCL INFO Channel 09/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:124161:124582 [1] NCCL INFO Channel 05/0 : 1[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124165:124584 [5] NCCL INFO Channel 01/0 : 5[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124164:124585 [4] NCCL INFO Channel 11/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:124163:124588 [3] NCCL INFO Channel 01/0 : 3[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:124160:124581 [0] NCCL INFO Channel 15/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:124167:124586 [7] NCCL INFO Channel 05/0 : 7[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124162:124587 [2] NCCL INFO Channel 11/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:124161:124582 [1] NCCL INFO Channel 07/0 : 1[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124165:124584 [5] NCCL INFO Channel 03/0 : 5[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124164:124585 [4] NCCL INFO Channel 13/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:124163:124588 [3] NCCL INFO Channel 03/0 : 3[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:124167:124586 [7] NCCL INFO Channel 07/0 : 7[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124162:124587 [2] NCCL INFO Channel 13/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:124165:124584 [5] NCCL INFO Channel 05/0 : 5[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124164:124585 [4] NCCL INFO Channel 15/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:124161:124582 [1] NCCL INFO Channel 11/0 : 1[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124163:124588 [3] NCCL INFO Channel 05/0 : 3[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:124167:124586 [7] NCCL INFO Channel 09/0 : 7[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124162:124587 [2] NCCL INFO Channel 15/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:124165:124584 [5] NCCL INFO Channel 07/0 : 5[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124161:124582 [1] NCCL INFO Channel 13/0 : 1[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124167:124586 [7] NCCL INFO Channel 11/0 : 7[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124163:124588 [3] NCCL INFO Channel 07/0 : 3[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:124161:124582 [1] NCCL INFO Channel 15/0 : 1[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124167:124586 [7] NCCL INFO Channel 13/0 : 7[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124165:124584 [5] NCCL INFO Channel 09/0 : 5[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124163:124588 [3] NCCL INFO Channel 09/0 : 3[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:124165:124584 [5] NCCL INFO Channel 11/0 : 5[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124163:124588 [3] NCCL INFO Channel 13/0 : 3[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:124163:124588 [3] NCCL INFO Channel 15/0 : 3[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:124165:124584 [5] NCCL INFO Channel 15/0 : 5[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557273:1557685 [7] NCCL INFO Channel 01/0 : 103[7] -> 111[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557269:1557684 [3] NCCL INFO Channel 01/0 : 99[3] -> 107[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557272:1557687 [6] NCCL INFO Channel 01/0 : 102[6] -> 110[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557273:1557685 [7] NCCL INFO Channel 03/0 : 103[7] -> 111[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557269:1557684 [3] NCCL INFO Channel 03/0 : 99[3] -> 107[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557272:1557687 [6] NCCL INFO Channel 03/0 : 102[6] -> 110[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557270:1557688 [4] NCCL INFO Channel 01/0 : 100[4] -> 108[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557273:1557685 [7] NCCL INFO Channel 05/0 : 103[7] -> 111[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557267:1557683 [1] NCCL INFO Channel 01/0 : 97[1] -> 105[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557272:1557687 [6] NCCL INFO Channel 05/0 : 102[6] -> 110[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557271:1557690 [5] NCCL INFO Channel 01/0 : 101[5] -> 109[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557269:1557684 [3] NCCL INFO Channel 05/0 : 99[3] -> 107[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557270:1557688 [4] NCCL INFO Channel 03/0 : 100[4] -> 108[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557273:1557685 [7] NCCL INFO Channel 07/0 : 103[7] -> 111[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557266:1557689 [0] NCCL INFO Channel 01/0 : 96[0] -> 104[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557271:1557690 [5] NCCL INFO Channel 03/0 : 101[5] -> 109[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114519:114931 [7] NCCL INFO Channel 01/0 : 71[7] -> 79[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557272:1557687 [6] NCCL INFO Channel 07/0 : 102[6] -> 110[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557267:1557683 [1] NCCL INFO Channel 03/0 : 97[1] -> 105[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114519:114931 [7] NCCL INFO Channel 03/0 : 71[7] -> 79[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557273:1557685 [7] NCCL INFO Channel 09/0 : 103[7] -> 111[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557270:1557688 [4] NCCL INFO Channel 05/0 : 100[4] -> 108[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557269:1557684 [3] NCCL INFO Channel 07/0 : 99[3] -> 107[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557268:1557686 [2] NCCL INFO Channel 01/0 : 98[2] -> 106[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557271:1557690 [5] NCCL INFO Channel 05/0 : 101[5] -> 109[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114519:114931 [7] NCCL INFO Channel 05/0 : 71[7] -> 79[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557266:1557689 [0] NCCL INFO Channel 03/0 : 96[0] -> 104[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557272:1557687 [6] NCCL INFO Channel 09/0 : 102[6] -> 110[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557267:1557683 [1] NCCL INFO Channel 05/0 : 97[1] -> 105[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557273:1557685 [7] NCCL INFO Channel 11/0 : 103[7] -> 111[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114517:114935 [5] NCCL INFO Channel 01/0 : 69[5] -> 77[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114513:114933 [1] NCCL INFO Channel 01/0 : 65[1] -> 73[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114518:114937 [6] NCCL INFO Channel 01/0 : 70[6] -> 78[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557270:1557688 [4] NCCL INFO Channel 07/0 : 100[4] -> 108[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557269:1557684 [3] NCCL INFO Channel 09/0 : 99[3] -> 107[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557271:1557690 [5] NCCL INFO Channel 07/0 : 101[5] -> 109[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557268:1557686 [2] NCCL INFO Channel 03/0 : 98[2] -> 106[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557266:1557689 [0] NCCL INFO Channel 05/0 : 96[0] -> 104[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114519:114931 [7] NCCL INFO Channel 07/0 : 71[7] -> 79[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557272:1557687 [6] NCCL INFO Channel 11/0 : 102[6] -> 110[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114517:114935 [5] NCCL INFO Channel 03/0 : 69[5] -> 77[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557267:1557683 [1] NCCL INFO Channel 07/0 : 97[1] -> 105[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114518:114937 [6] NCCL INFO Channel 03/0 : 70[6] -> 78[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114519:114931 [7] NCCL INFO Channel 09/0 : 71[7] -> 79[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114513:114933 [1] NCCL INFO Channel 03/0 : 65[1] -> 73[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557273:1557685 [7] NCCL INFO Channel 13/0 : 103[7] -> 111[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114517:114935 [5] NCCL INFO Channel 05/0 : 69[5] -> 77[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557269:1557684 [3] NCCL INFO Channel 13/0 : 99[3] -> 107[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557271:1557690 [5] NCCL INFO Channel 09/0 : 101[5] -> 109[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557270:1557688 [4] NCCL INFO Channel 09/0 : 100[4] -> 108[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557268:1557686 [2] NCCL INFO Channel 05/0 : 98[2] -> 106[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557266:1557689 [0] NCCL INFO Channel 07/0 : 96[0] -> 104[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114512:114930 [0] NCCL INFO Channel 01/0 : 64[0] -> 72[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557267:1557683 [1] NCCL INFO Channel 11/0 : 97[1] -> 105[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557272:1557687 [6] NCCL INFO Channel 13/0 : 102[6] -> 110[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114513:114933 [1] NCCL INFO Channel 05/0 : 65[1] -> 73[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114518:114937 [6] NCCL INFO Channel 05/0 : 70[6] -> 78[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114519:114931 [7] NCCL INFO Channel 11/0 : 71[7] -> 79[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114517:114935 [5] NCCL INFO Channel 07/0 : 69[5] -> 77[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114512:114930 [0] NCCL INFO Channel 03/0 : 64[0] -> 72[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557269:1557684 [3] NCCL INFO Channel 15/0 : 99[3] -> 107[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557271:1557690 [5] NCCL INFO Channel 11/0 : 101[5] -> 109[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557266:1557689 [0] NCCL INFO Channel 09/0 : 96[0] -> 104[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557270:1557688 [4] NCCL INFO Channel 11/0 : 100[4] -> 108[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557268:1557686 [2] NCCL INFO Channel 07/0 : 98[2] -> 106[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557267:1557683 [1] NCCL INFO Channel 13/0 : 97[1] -> 105[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557272:1557687 [6] NCCL INFO Channel 15/0 : 102[6] -> 110[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114514:114932 [2] NCCL INFO Channel 01/0 : 66[2] -> 74[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557266:1557689 [0] NCCL INFO Channel 11/0 : 96[0] -> 104[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557271:1557690 [5] NCCL INFO Channel 15/0 : 101[5] -> 109[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557270:1557688 [4] NCCL INFO Channel 13/0 : 100[4] -> 108[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114518:114937 [6] NCCL INFO Channel 07/0 : 70[6] -> 78[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114519:114931 [7] NCCL INFO Channel 13/0 : 71[7] -> 79[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114515:114936 [3] NCCL INFO Channel 01/0 : 67[3] -> 75[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114517:114935 [5] NCCL INFO Channel 09/0 : 69[5] -> 77[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114512:114930 [0] NCCL INFO Channel 05/0 : 64[0] -> 72[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114513:114933 [1] NCCL INFO Channel 07/0 : 65[1] -> 73[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557268:1557686 [2] NCCL INFO Channel 09/0 : 98[2] -> 106[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557267:1557683 [1] NCCL INFO Channel 15/0 : 97[1] -> 105[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114514:114932 [2] NCCL INFO Channel 03/0 : 66[2] -> 74[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114516:114934 [4] NCCL INFO Channel 01/0 : 68[4] -> 76[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557266:1557689 [0] NCCL INFO Channel 13/0 : 96[0] -> 104[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114518:114937 [6] NCCL INFO Channel 09/0 : 70[6] -> 78[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557270:1557688 [4] NCCL INFO Channel 15/0 : 100[4] -> 108[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557268:1557686 [2] NCCL INFO Channel 11/0 : 98[2] -> 106[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114517:114935 [5] NCCL INFO Channel 11/0 : 69[5] -> 77[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114513:114933 [1] NCCL INFO Channel 11/0 : 65[1] -> 73[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114515:114936 [3] NCCL INFO Channel 03/0 : 67[3] -> 75[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114512:114930 [0] NCCL INFO Channel 07/0 : 64[0] -> 72[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114514:114932 [2] NCCL INFO Channel 05/0 : 66[2] -> 74[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114516:114934 [4] NCCL INFO Channel 03/0 : 68[4] -> 76[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114518:114937 [6] NCCL INFO Channel 11/0 : 70[6] -> 78[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557266:1557689 [0] NCCL INFO Channel 15/0 : 96[0] -> 104[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557268:1557686 [2] NCCL INFO Channel 13/0 : 98[2] -> 106[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557268:1557686 [2] NCCL INFO Channel 15/0 : 98[2] -> 106[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96893:97310 [5] NCCL INFO Channel 00/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96893:97310 [5] NCCL INFO Channel 01/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96895:97307 [7] NCCL INFO Channel 00/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96889:97309 [1] NCCL INFO Channel 00/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96891:97311 [3] NCCL INFO Channel 00/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96893:97310 [5] NCCL INFO Channel 02/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96889:97309 [1] NCCL INFO Channel 02/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96895:97307 [7] NCCL INFO Channel 01/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96893:97310 [5] NCCL INFO Channel 03/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96891:97311 [3] NCCL INFO Channel 01/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96892:97308 [4] NCCL INFO Channel 00/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96889:97309 [1] NCCL INFO Channel 03/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96893:97310 [5] NCCL INFO Channel 04/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96888:97306 [0] NCCL INFO Channel 01/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96895:97307 [7] NCCL INFO Channel 02/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96891:97311 [3] NCCL INFO Channel 02/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96892:97308 [4] NCCL INFO Channel 01/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96889:97309 [1] NCCL INFO Channel 04/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364908:1365332 [6] NCCL INFO Channel 01/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96893:97310 [5] NCCL INFO Channel 06/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96888:97306 [0] NCCL INFO Channel 02/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96894:97312 [6] NCCL INFO Channel 00/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96895:97307 [7] NCCL INFO Channel 03/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96891:97311 [3] NCCL INFO Channel 04/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96892:97308 [4] NCCL INFO Channel 02/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96889:97309 [1] NCCL INFO Channel 05/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96890:97313 [2] NCCL INFO Channel 00/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364906:1365331 [4] NCCL INFO Channel 01/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96893:97310 [5] NCCL INFO Channel 07/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96888:97306 [0] NCCL INFO Channel 03/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364908:1365332 [6] NCCL INFO Channel 03/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96895:97307 [7] NCCL INFO Channel 04/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364906:1365331 [4] NCCL INFO Channel 03/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96894:97312 [6] NCCL INFO Channel 01/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96892:97308 [4] NCCL INFO Channel 03/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96889:97309 [1] NCCL INFO Channel 06/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96891:97311 [3] NCCL INFO Channel 05/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96890:97313 [2] NCCL INFO Channel 01/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96893:97310 [5] NCCL INFO Channel 08/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364902:1365333 [0] NCCL INFO Channel 01/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96888:97306 [0] NCCL INFO Channel 04/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96895:97307 [7] NCCL INFO Channel 05/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364908:1365332 [6] NCCL INFO Channel 05/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364906:1365331 [4] NCCL INFO Channel 05/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96889:97309 [1] NCCL INFO Channel 07/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96890:97313 [2] NCCL INFO Channel 03/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96891:97311 [3] NCCL INFO Channel 06/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96894:97312 [6] NCCL INFO Channel 02/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96892:97308 [4] NCCL INFO Channel 05/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364902:1365333 [0] NCCL INFO Channel 03/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364908:1365332 [6] NCCL INFO Channel 07/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364904:1365335 [2] NCCL INFO Channel 01/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364906:1365331 [4] NCCL INFO Channel 07/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96888:97306 [0] NCCL INFO Channel 05/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96893:97310 [5] NCCL INFO Channel 09/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96895:97307 [7] NCCL INFO Channel 06/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96889:97309 [1] NCCL INFO Channel 08/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364909:1365327 [7] NCCL INFO Channel 01/0 : 7[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364902:1365333 [0] NCCL INFO Channel 05/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364907:1365329 [5] NCCL INFO Channel 01/0 : 5[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96892:97308 [4] NCCL INFO Channel 06/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96891:97311 [3] NCCL INFO Channel 07/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364904:1365335 [2] NCCL INFO Channel 03/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364908:1365332 [6] NCCL INFO Channel 09/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364906:1365331 [4] NCCL INFO Channel 09/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364909:1365327 [7] NCCL INFO Channel 03/0 : 7[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364902:1365333 [0] NCCL INFO Channel 07/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96890:97313 [2] NCCL INFO Channel 04/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96894:97312 [6] NCCL INFO Channel 03/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96888:97306 [0] NCCL INFO Channel 06/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96889:97309 [1] NCCL INFO Channel 10/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364907:1365329 [5] NCCL INFO Channel 03/0 : 5[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364905:1365328 [3] NCCL INFO Channel 01/0 : 3[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96893:97310 [5] NCCL INFO Channel 10/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96895:97307 [7] NCCL INFO Channel 08/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364904:1365335 [2] NCCL INFO Channel 05/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364908:1365332 [6] NCCL INFO Channel 11/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364906:1365331 [4] NCCL INFO Channel 11/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364902:1365333 [0] NCCL INFO Channel 09/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96891:97311 [3] NCCL INFO Channel 08/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364909:1365327 [7] NCCL INFO Channel 05/0 : 7[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96892:97308 [4] NCCL INFO Channel 07/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96888:97306 [0] NCCL INFO Channel 07/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364905:1365328 [3] NCCL INFO Channel 03/0 : 3[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364907:1365329 [5] NCCL INFO Channel 05/0 : 5[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96889:97309 [1] NCCL INFO Channel 11/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96890:97313 [2] NCCL INFO Channel 05/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96893:97310 [5] NCCL INFO Channel 11/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96894:97312 [6] NCCL INFO Channel 04/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96895:97307 [7] NCCL INFO Channel 09/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96891:97311 [3] NCCL INFO Channel 09/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364904:1365335 [2] NCCL INFO Channel 07/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364903:1365334 [1] NCCL INFO Channel 01/0 : 1[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364906:1365331 [4] NCCL INFO Channel 13/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364908:1365332 [6] NCCL INFO Channel 13/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364902:1365333 [0] NCCL INFO Channel 11/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364909:1365327 [7] NCCL INFO Channel 07/0 : 7[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96892:97308 [4] NCCL INFO Channel 08/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96888:97306 [0] NCCL INFO Channel 09/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364905:1365328 [3] NCCL INFO Channel 05/0 : 3[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96889:97309 [1] NCCL INFO Channel 12/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364907:1365329 [5] NCCL INFO Channel 07/0 : 5[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96890:97313 [2] NCCL INFO Channel 06/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364904:1365335 [2] NCCL INFO Channel 09/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96895:97307 [7] NCCL INFO Channel 10/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96893:97310 [5] NCCL INFO Channel 12/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364906:1365331 [4] NCCL INFO Channel 15/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96891:97311 [3] NCCL INFO Channel 10/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364908:1365332 [6] NCCL INFO Channel 15/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32710:33125 [6] NCCL INFO Channel 00/0 : 22[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96888:97306 [0] NCCL INFO Channel 10/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96894:97312 [6] NCCL INFO Channel 05/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364903:1365334 [1] NCCL INFO Channel 03/0 : 1[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96892:97308 [4] NCCL INFO Channel 09/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96889:97309 [1] NCCL INFO Channel 13/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96890:97313 [2] NCCL INFO Channel 07/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32710:33125 [6] NCCL INFO Channel 02/0 : 22[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364902:1365333 [0] NCCL INFO Channel 13/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364905:1365328 [3] NCCL INFO Channel 07/0 : 3[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364909:1365327 [7] NCCL INFO Channel 09/0 : 7[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364907:1365329 [5] NCCL INFO Channel 09/0 : 5[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364904:1365335 [2] NCCL INFO Channel 11/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96895:97307 [7] NCCL INFO Channel 11/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364906:1365331 [4] NCCL INFO Channel 00/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96893:97310 [5] NCCL INFO Channel 14/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364908:1365332 [6] NCCL INFO Channel 00/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32710:33125 [6] NCCL INFO Channel 04/0 : 22[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96891:97311 [3] NCCL INFO Channel 12/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96888:97306 [0] NCCL INFO Channel 11/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96894:97312 [6] NCCL INFO Channel 07/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96889:97309 [1] NCCL INFO Channel 14/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96890:97313 [2] NCCL INFO Channel 08/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96892:97308 [4] NCCL INFO Channel 10/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364903:1365334 [1] NCCL INFO Channel 05/0 : 1[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364902:1365333 [0] NCCL INFO Channel 15/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364905:1365328 [3] NCCL INFO Channel 09/0 : 3[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364909:1365327 [7] NCCL INFO Channel 11/0 : 7[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32706:33122 [2] NCCL INFO Channel 00/0 : 18[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96895:97307 [7] NCCL INFO Channel 12/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364907:1365329 [5] NCCL INFO Channel 11/0 : 5[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96893:97310 [5] NCCL INFO Channel 15/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364904:1365335 [2] NCCL INFO Channel 13/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96891:97311 [3] NCCL INFO Channel 13/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96888:97306 [0] NCCL INFO Channel 12/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364906:1365331 [4] NCCL INFO Channel 01/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364908:1365332 [6] NCCL INFO Channel 01/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32710:33125 [6] NCCL INFO Channel 08/0 : 22[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32708:33126 [4] NCCL INFO Channel 00/0 : 20[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32707:33124 [3] NCCL INFO Channel 00/0 : 19[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96894:97312 [6] NCCL INFO Channel 08/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96889:97309 [1] NCCL INFO Channel 15/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96892:97308 [4] NCCL INFO Channel 11/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96895:97307 [7] NCCL INFO Channel 13/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96893:97310 [5] NCCL INFO Channel 00/0 : 21[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96890:97313 [2] NCCL INFO Channel 09/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364903:1365334 [1] NCCL INFO Channel 07/0 : 1[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96888:97306 [0] NCCL INFO Channel 13/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96891:97311 [3] NCCL INFO Channel 14/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364902:1365333 [0] NCCL INFO Channel 01/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364905:1365328 [3] NCCL INFO Channel 13/0 : 3[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32704:33123 [0] NCCL INFO Channel 02/0 : 16[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32706:33122 [2] NCCL INFO Channel 04/0 : 18[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364909:1365327 [7] NCCL INFO Channel 13/0 : 7[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32710:33125 [6] NCCL INFO Channel 10/0 : 22[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364904:1365335 [2] NCCL INFO Channel 15/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364906:1365331 [4] NCCL INFO Channel 02/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96889:97309 [1] NCCL INFO Channel 00/0 : 17[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364907:1365329 [5] NCCL INFO Channel 15/0 : 5[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364908:1365332 [6] NCCL INFO Channel 02/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96894:97312 [6] NCCL INFO Channel 09/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364902:1365333 [0] NCCL INFO Channel 02/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364903:1365334 [1] NCCL INFO Channel 11/0 : 1[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96892:97308 [4] NCCL INFO Channel 13/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364905:1365328 [3] NCCL INFO Channel 15/0 : 3[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32707:33124 [3] NCCL INFO Channel 02/0 : 19[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32708:33126 [4] NCCL INFO Channel 02/0 : 20[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32709:33129 [5] NCCL INFO Channel 00/0 : 21[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32705:33128 [1] NCCL INFO Channel 00/0 : 17[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96893:97310 [5] NCCL INFO Channel 02/0 : 21[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96890:97313 [2] NCCL INFO Channel 11/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96888:97306 [0] NCCL INFO Channel 14/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96891:97311 [3] NCCL INFO Channel 15/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96895:97307 [7] NCCL INFO Channel 14/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96889:97309 [1] NCCL INFO Channel 02/0 : 17[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364909:1365327 [7] NCCL INFO Channel 00/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32706:33122 [2] NCCL INFO Channel 06/0 : 18[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96894:97312 [6] NCCL INFO Channel 10/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364904:1365335 [2] NCCL INFO Channel 00/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32704:33123 [0] NCCL INFO Channel 04/0 : 16[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96892:97308 [4] NCCL INFO Channel 14/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364906:1365331 [4] NCCL INFO Channel 03/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364908:1365332 [6] NCCL INFO Channel 03/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364907:1365329 [5] NCCL INFO Channel 00/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364905:1365328 [3] NCCL INFO Channel 00/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364903:1365334 [1] NCCL INFO Channel 13/0 : 1[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32710:33125 [6] NCCL INFO Channel 12/0 : 22[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32707:33124 [3] NCCL INFO Channel 04/0 : 19[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32711:33127 [7] NCCL INFO Channel 00/0 : 23[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96890:97313 [2] NCCL INFO Channel 12/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32708:33126 [4] NCCL INFO Channel 06/0 : 20[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96888:97306 [0] NCCL INFO Channel 15/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32709:33129 [5] NCCL INFO Channel 02/0 : 21[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96889:97309 [1] NCCL INFO Channel 04/0 : 17[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364909:1365327 [7] NCCL INFO Channel 01/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364902:1365333 [0] NCCL INFO Channel 03/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364904:1365335 [2] NCCL INFO Channel 01/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364906:1365331 [4] NCCL INFO Channel 05/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364908:1365332 [6] NCCL INFO Channel 04/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364907:1365329 [5] NCCL INFO Channel 01/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364905:1365328 [3] NCCL INFO Channel 01/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32705:33128 [1] NCCL INFO Channel 02/0 : 17[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32706:33122 [2] NCCL INFO Channel 08/0 : 18[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96895:97307 [7] NCCL INFO Channel 00/0 : 23[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96891:97311 [3] NCCL INFO Channel 00/0 : 19[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96893:97310 [5] NCCL INFO Channel 04/0 : 21[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96894:97312 [6] NCCL INFO Channel 11/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364903:1365334 [1] NCCL INFO Channel 15/0 : 1[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32710:33125 [6] NCCL INFO Channel 14/0 : 22[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96892:97308 [4] NCCL INFO Channel 15/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32707:33124 [3] NCCL INFO Channel 06/0 : 19[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96890:97313 [2] NCCL INFO Channel 13/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32704:33123 [0] NCCL INFO Channel 06/0 : 16[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32708:33126 [4] NCCL INFO Channel 08/0 : 20[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32711:33127 [7] NCCL INFO Channel 02/0 : 23[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96888:97306 [0] NCCL INFO Channel 02/0 : 16[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96889:97309 [1] NCCL INFO Channel 06/0 : 17[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32709:33129 [5] NCCL INFO Channel 04/0 : 21[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364902:1365333 [0] NCCL INFO Channel 04/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364904:1365335 [2] NCCL INFO Channel 03/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364909:1365327 [7] NCCL INFO Channel 02/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364906:1365331 [4] NCCL INFO Channel 06/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364908:1365332 [6] NCCL INFO Channel 05/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364907:1365329 [5] NCCL INFO Channel 02/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364905:1365328 [3] NCCL INFO Channel 02/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364903:1365334 [1] NCCL INFO Channel 00/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96895:97307 [7] NCCL INFO Channel 02/0 : 23[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364902:1365333 [0] NCCL INFO Channel 05/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96891:97311 [3] NCCL INFO Channel 02/0 : 19[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96893:97310 [5] NCCL INFO Channel 06/0 : 21[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364904:1365335 [2] NCCL INFO Channel 04/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32706:33122 [2] NCCL INFO Channel 10/0 : 18[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32705:33128 [1] NCCL INFO Channel 04/0 : 17[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96894:97312 [6] NCCL INFO Channel 12/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32707:33124 [3] NCCL INFO Channel 08/0 : 19[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96892:97308 [4] NCCL INFO Channel 00/0 : 20[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32709:33129 [5] NCCL INFO Channel 06/0 : 21[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96890:97313 [2] NCCL INFO Channel 14/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96889:97309 [1] NCCL INFO Channel 08/0 : 17[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32711:33127 [7] NCCL INFO Channel 04/0 : 23[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32708:33126 [4] NCCL INFO Channel 10/0 : 20[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96888:97306 [0] NCCL INFO Channel 04/0 : 16[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96895:97307 [7] NCCL INFO Channel 04/0 : 23[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364906:1365331 [4] NCCL INFO Channel 07/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364909:1365327 [7] NCCL INFO Channel 03/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364908:1365332 [6] NCCL INFO Channel 07/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364905:1365328 [3] NCCL INFO Channel 04/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364907:1365329 [5] NCCL INFO Channel 03/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364903:1365334 [1] NCCL INFO Channel 02/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364902:1365333 [0] NCCL INFO Channel 06/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364904:1365335 [2] NCCL INFO Channel 05/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32704:33123 [0] NCCL INFO Channel 08/0 : 16[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96891:97311 [3] NCCL INFO Channel 04/0 : 19[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32706:33122 [2] NCCL INFO Channel 12/0 : 18[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32705:33128 [1] NCCL INFO Channel 06/0 : 17[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32707:33124 [3] NCCL INFO Channel 10/0 : 19[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96893:97310 [5] NCCL INFO Channel 08/0 : 21[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32709:33129 [5] NCCL INFO Channel 08/0 : 21[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96892:97308 [4] NCCL INFO Channel 02/0 : 20[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96894:97312 [6] NCCL INFO Channel 13/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32711:33127 [7] NCCL INFO Channel 06/0 : 23[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96889:97309 [1] NCCL INFO Channel 10/0 : 17[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96890:97313 [2] NCCL INFO Channel 15/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32708:33126 [4] NCCL INFO Channel 12/0 : 20[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96895:97307 [7] NCCL INFO Channel 06/0 : 23[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96888:97306 [0] NCCL INFO Channel 06/0 : 16[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96891:97311 [3] NCCL INFO Channel 06/0 : 19[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32704:33123 [0] NCCL INFO Channel 10/0 : 16[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364906:1365331 [4] NCCL INFO Channel 08/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364909:1365327 [7] NCCL INFO Channel 04/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364908:1365332 [6] NCCL INFO Channel 08/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96893:97310 [5] NCCL INFO Channel 10/0 : 21[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96892:97308 [4] NCCL INFO Channel 06/0 : 20[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32706:33122 [2] NCCL INFO Channel 14/0 : 18[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96889:97309 [1] NCCL INFO Channel 12/0 : 17[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96890:97313 [2] NCCL INFO Channel 00/0 : 18[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32705:33128 [1] NCCL INFO Channel 08/0 : 17[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96894:97312 [6] NCCL INFO Channel 15/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364905:1365328 [3] NCCL INFO Channel 05/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364903:1365334 [1] NCCL INFO Channel 03/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32707:33124 [3] NCCL INFO Channel 12/0 : 19[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62884:63299 [3] NCCL INFO Channel 01/0 : 35[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96895:97307 [7] NCCL INFO Channel 08/0 : 23[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364904:1365335 [2] NCCL INFO Channel 06/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32709:33129 [5] NCCL INFO Channel 10/0 : 21[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96888:97306 [0] NCCL INFO Channel 08/0 : 16[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364907:1365329 [5] NCCL INFO Channel 04/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32711:33127 [7] NCCL INFO Channel 08/0 : 23[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364902:1365333 [0] NCCL INFO Channel 07/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364906:1365331 [4] NCCL INFO Channel 09/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32708:33126 [4] NCCL INFO Channel 14/0 : 20[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364909:1365327 [7] NCCL INFO Channel 05/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364908:1365332 [6] NCCL INFO Channel 09/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32704:33123 [0] NCCL INFO Channel 12/0 : 16[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32705:33128 [1] NCCL INFO Channel 10/0 : 17[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96891:97311 [3] NCCL INFO Channel 08/0 : 19[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96893:97310 [5] NCCL INFO Channel 12/0 : 21[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96892:97308 [4] NCCL INFO Channel 08/0 : 20[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62884:63299 [3] NCCL INFO Channel 03/0 : 35[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96889:97309 [1] NCCL INFO Channel 14/0 : 17[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96890:97313 [2] NCCL INFO Channel 04/0 : 18[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96895:97307 [7] NCCL INFO Channel 10/0 : 23[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364905:1365328 [3] NCCL INFO Channel 06/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96894:97312 [6] NCCL INFO Channel 00/0 : 22[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364903:1365334 [1] NCCL INFO Channel 04/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364904:1365335 [2] NCCL INFO Channel 07/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96888:97306 [0] NCCL INFO Channel 10/0 : 16[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364907:1365329 [5] NCCL INFO Channel 06/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32709:33129 [5] NCCL INFO Channel 12/0 : 21[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32707:33124 [3] NCCL INFO Channel 14/0 : 19[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364902:1365333 [0] NCCL INFO Channel 09/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32711:33127 [7] NCCL INFO Channel 10/0 : 23[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62886:63303 [5] NCCL INFO Channel 01/0 : 37[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364906:1365331 [4] NCCL INFO Channel 10/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32704:33123 [0] NCCL INFO Channel 14/0 : 16[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62888:63302 [7] NCCL INFO Channel 01/0 : 39[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364908:1365332 [6] NCCL INFO Channel 10/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364909:1365327 [7] NCCL INFO Channel 06/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32705:33128 [1] NCCL INFO Channel 12/0 : 17[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32709:33129 [5] NCCL INFO Channel 14/0 : 21[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96893:97310 [5] NCCL INFO Channel 14/0 : 21[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96890:97313 [2] NCCL INFO Channel 06/0 : 18[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96891:97311 [3] NCCL INFO Channel 10/0 : 19[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96892:97308 [4] NCCL INFO Channel 10/0 : 20[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32711:33127 [7] NCCL INFO Channel 12/0 : 23[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96895:97307 [7] NCCL INFO Channel 12/0 : 23[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32705:33128 [1] NCCL INFO Channel 14/0 : 17[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96888:97306 [0] NCCL INFO Channel 12/0 : 16[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96894:97312 [6] NCCL INFO Channel 02/0 : 22[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32711:33127 [7] NCCL INFO Channel 14/0 : 23[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62884:63299 [3] NCCL INFO Channel 05/0 : 35[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62886:63303 [5] NCCL INFO Channel 03/0 : 37[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364905:1365328 [3] NCCL INFO Channel 07/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62888:63302 [7] NCCL INFO Channel 03/0 : 39[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364903:1365334 [1] NCCL INFO Channel 05/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364904:1365335 [2] NCCL INFO Channel 08/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364907:1365329 [5] NCCL INFO Channel 07/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96890:97313 [2] NCCL INFO Channel 08/0 : 18[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364906:1365331 [4] NCCL INFO Channel 11/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96892:97308 [4] NCCL INFO Channel 12/0 : 20[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364902:1365333 [0] NCCL INFO Channel 10/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96895:97307 [7] NCCL INFO Channel 14/0 : 23[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96891:97311 [3] NCCL INFO Channel 12/0 : 19[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96894:97312 [6] NCCL INFO Channel 04/0 : 22[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96888:97306 [0] NCCL INFO Channel 14/0 : 16[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62886:63303 [5] NCCL INFO Channel 05/0 : 37[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62884:63299 [3] NCCL INFO Channel 07/0 : 35[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62888:63302 [7] NCCL INFO Channel 05/0 : 39[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96890:97313 [2] NCCL INFO Channel 10/0 : 18[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364909:1365327 [7] NCCL INFO Channel 08/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364908:1365332 [6] NCCL INFO Channel 11/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364905:1365328 [3] NCCL INFO Channel 08/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364904:1365335 [2] NCCL INFO Channel 09/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364903:1365334 [1] NCCL INFO Channel 06/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364907:1365329 [5] NCCL INFO Channel 08/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62882:63300 [1] NCCL INFO Channel 01/0 : 33[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96892:97308 [4] NCCL INFO Channel 14/0 : 20[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364906:1365331 [4] NCCL INFO Channel 13/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96891:97311 [3] NCCL INFO Channel 14/0 : 19[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364902:1365333 [0] NCCL INFO Channel 11/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96894:97312 [6] NCCL INFO Channel 08/0 : 22[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32705:33128 [1] NCCL INFO Channel 01/0 : 9[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96890:97313 [2] NCCL INFO Channel 12/0 : 18[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96894:97312 [6] NCCL INFO Channel 10/0 : 22[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62886:63303 [5] NCCL INFO Channel 07/0 : 37[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62888:63302 [7] NCCL INFO Channel 07/0 : 39[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62884:63299 [3] NCCL INFO Channel 09/0 : 35[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364908:1365332 [6] NCCL INFO Channel 12/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62882:63300 [1] NCCL INFO Channel 03/0 : 33[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62881:63298 [0] NCCL INFO Channel 01/0 : 32[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364909:1365327 [7] NCCL INFO Channel 09/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364905:1365328 [3] NCCL INFO Channel 09/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32709:33129 [5] NCCL INFO Channel 01/0 : 13[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96890:97313 [2] NCCL INFO Channel 14/0 : 18[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364907:1365329 [5] NCCL INFO Channel 09/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364906:1365331 [4] NCCL INFO Channel 14/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364904:1365335 [2] NCCL INFO Channel 11/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96894:97312 [6] NCCL INFO Channel 12/0 : 22[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364903:1365334 [1] NCCL INFO Channel 07/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96894:97312 [6] NCCL INFO Channel 14/0 : 22[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364902:1365333 [0] NCCL INFO Channel 12/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62886:63303 [5] NCCL INFO Channel 09/0 : 37[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364908:1365332 [6] NCCL INFO Channel 13/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32705:33128 [1] NCCL INFO Channel 03/0 : 9[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32709:33129 [5] NCCL INFO Channel 03/0 : 13[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62887:63305 [6] NCCL INFO Channel 01/0 : 38[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62888:63302 [7] NCCL INFO Channel 09/0 : 39[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364909:1365327 [7] NCCL INFO Channel 10/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364905:1365328 [3] NCCL INFO Channel 10/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62885:63304 [4] NCCL INFO Channel 01/0 : 36[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364906:1365331 [4] NCCL INFO Channel 15/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364907:1365329 [5] NCCL INFO Channel 10/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32711:33127 [7] NCCL INFO Channel 01/0 : 15[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62883:63301 [2] NCCL INFO Channel 01/0 : 34[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364904:1365335 [2] NCCL INFO Channel 12/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62882:63300 [1] NCCL INFO Channel 05/0 : 33[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62881:63298 [0] NCCL INFO Channel 03/0 : 32[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364903:1365334 [1] NCCL INFO Channel 08/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62884:63299 [3] NCCL INFO Channel 13/0 : 35[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364902:1365333 [0] NCCL INFO Channel 13/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62886:63303 [5] NCCL INFO Channel 11/0 : 37[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364908:1365332 [6] NCCL INFO Channel 15/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364909:1365327 [7] NCCL INFO Channel 11/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364905:1365328 [3] NCCL INFO Channel 12/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32709:33129 [5] NCCL INFO Channel 05/0 : 13[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32705:33128 [1] NCCL INFO Channel 05/0 : 9[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62887:63305 [6] NCCL INFO Channel 03/0 : 38[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62888:63302 [7] NCCL INFO Channel 11/0 : 39[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364904:1365335 [2] NCCL INFO Channel 13/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364907:1365329 [5] NCCL INFO Channel 11/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364903:1365334 [1] NCCL INFO Channel 10/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32707:33124 [3] NCCL INFO Channel 01/0 : 11[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32708:33126 [4] NCCL INFO Channel 01/0 : 12[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364902:1365333 [0] NCCL INFO Channel 14/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32704:33123 [0] NCCL INFO Channel 01/0 : 8[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364905:1365328 [3] NCCL INFO Channel 13/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32711:33127 [7] NCCL INFO Channel 03/0 : 15[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124164:124585 [4] NCCL INFO Channel 00/0 : 68[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62881:63298 [0] NCCL INFO Channel 05/0 : 32[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62883:63301 [2] NCCL INFO Channel 03/0 : 34[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62885:63304 [4] NCCL INFO Channel 03/0 : 36[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62882:63300 [1] NCCL INFO Channel 07/0 : 33[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62886:63303 [5] NCCL INFO Channel 15/0 : 37[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364909:1365327 [7] NCCL INFO Channel 12/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62884:63299 [3] NCCL INFO Channel 15/0 : 35[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364904:1365335 [2] NCCL INFO Channel 14/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62887:63305 [6] NCCL INFO Channel 05/0 : 38[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62888:63302 [7] NCCL INFO Channel 13/0 : 39[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364907:1365329 [5] NCCL INFO Channel 12/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364902:1365333 [0] NCCL INFO Channel 15/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62881:63298 [0] NCCL INFO Channel 07/0 : 32[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32709:33129 [5] NCCL INFO Channel 07/0 : 13[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32705:33128 [1] NCCL INFO Channel 07/0 : 9[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124164:124585 [4] NCCL INFO Channel 02/0 : 68[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32706:33122 [2] NCCL INFO Channel 01/0 : 10[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:124166:124583 [6] NCCL INFO Channel 00/0 : 70[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62883:63301 [2] NCCL INFO Channel 05/0 : 34[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364905:1365328 [3] NCCL INFO Channel 14/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96892:97308 [4] NCCL INFO Channel 00/0 : 20[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364903:1365334 [1] NCCL INFO Channel 11/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364909:1365327 [7] NCCL INFO Channel 13/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364904:1365335 [2] NCCL INFO Channel 15/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32707:33124 [3] NCCL INFO Channel 03/0 : 11[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32708:33126 [4] NCCL INFO Channel 03/0 : 12[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:124164:124585 [4] NCCL INFO Channel 06/0 : 68[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62886:63303 [5] NCCL INFO Channel 00/0 : 45[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62885:63304 [4] NCCL INFO Channel 05/0 : 36[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96892:97308 [4] NCCL INFO Channel 02/0 : 20[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364907:1365329 [5] NCCL INFO Channel 14/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32711:33127 [7] NCCL INFO Channel 05/0 : 15[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32709:33129 [5] NCCL INFO Channel 09/0 : 13[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124166:124583 [6] NCCL INFO Channel 02/0 : 70[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62882:63300 [1] NCCL INFO Channel 11/0 : 33[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364905:1365328 [3] NCCL INFO Channel 15/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32705:33128 [1] NCCL INFO Channel 11/0 : 9[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62888:63302 [7] NCCL INFO Channel 00/0 : 47[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364909:1365327 [7] NCCL INFO Channel 14/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32704:33123 [0] NCCL INFO Channel 03/0 : 8[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32706:33122 [2] NCCL INFO Channel 03/0 : 10[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62887:63305 [6] NCCL INFO Channel 07/0 : 38[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62884:63299 [3] NCCL INFO Channel 00/0 : 43[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62881:63298 [0] NCCL INFO Channel 09/0 : 32[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96894:97312 [6] NCCL INFO Channel 00/0 : 22[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62883:63301 [2] NCCL INFO Channel 07/0 : 34[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96892:97308 [4] NCCL INFO Channel 06/0 : 20[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:124164:124585 [4] NCCL INFO Channel 08/0 : 68[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364906:1365331 [4] NCCL INFO Channel 01/0 : 12[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364903:1365334 [1] NCCL INFO Channel 12/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364907:1365329 [5] NCCL INFO Channel 15/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32710:33125 [6] NCCL INFO Channel 01/0 : 14[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:124166:124583 [6] NCCL INFO Channel 04/0 : 70[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364903:1365334 [1] NCCL INFO Channel 13/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32707:33124 [3] NCCL INFO Channel 05/0 : 11[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32708:33126 [4] NCCL INFO Channel 05/0 : 12[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:124162:124587 [2] NCCL INFO Channel 00/0 : 66[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:124160:124581 [0] NCCL INFO Channel 02/0 : 64[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62886:63303 [5] NCCL INFO Channel 01/0 : 45[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96894:97312 [6] NCCL INFO Channel 02/0 : 22[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364906:1365331 [4] NCCL INFO Channel 03/0 : 12[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364908:1365332 [6] NCCL INFO Channel 01/0 : 14[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32709:33129 [5] NCCL INFO Channel 11/0 : 13[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124167:124586 [7] NCCL INFO Channel 00/0 : 71[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62885:63304 [4] NCCL INFO Channel 07/0 : 36[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62882:63300 [1] NCCL INFO Channel 13/0 : 33[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96892:97308 [4] NCCL INFO Channel 08/0 : 20[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364903:1365334 [1] NCCL INFO Channel 14/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32711:33127 [7] NCCL INFO Channel 07/0 : 15[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32705:33128 [1] NCCL INFO Channel 13/0 : 9[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62888:63302 [7] NCCL INFO Channel 01/0 : 47[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32706:33122 [2] NCCL INFO Channel 05/0 : 10[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62881:63298 [0] NCCL INFO Channel 11/0 : 32[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32704:33123 [0] NCCL INFO Channel 05/0 : 8[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:124164:124585 [4] NCCL INFO Channel 10/0 : 68[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:124166:124583 [6] NCCL INFO Channel 08/0 : 70[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62884:63299 [3] NCCL INFO Channel 01/0 : 43[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62887:63305 [6] NCCL INFO Channel 09/0 : 38[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62886:63303 [5] NCCL INFO Channel 02/0 : 45[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62883:63301 [2] NCCL INFO Channel 09/0 : 34[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96894:97312 [6] NCCL INFO Channel 04/0 : 22[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96892:97308 [4] NCCL INFO Channel 10/0 : 20[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364906:1365331 [4] NCCL INFO Channel 05/0 : 12[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32710:33125 [6] NCCL INFO Channel 03/0 : 14[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:124163:124588 [3] NCCL INFO Channel 00/0 : 67[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364908:1365332 [6] NCCL INFO Channel 03/0 : 14[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32707:33124 [3] NCCL INFO Channel 07/0 : 11[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32708:33126 [4] NCCL INFO Channel 07/0 : 12[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364903:1365334 [1] NCCL INFO Channel 15/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32709:33129 [5] NCCL INFO Channel 15/0 : 13[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124162:124587 [2] NCCL INFO Channel 04/0 : 66[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62882:63300 [1] NCCL INFO Channel 15/0 : 33[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62885:63304 [4] NCCL INFO Channel 09/0 : 36[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62888:63302 [7] NCCL INFO Channel 02/0 : 47[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364906:1365331 [4] NCCL INFO Channel 07/0 : 12[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32711:33127 [7] NCCL INFO Channel 09/0 : 15[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124167:124586 [7] NCCL INFO Channel 02/0 : 71[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62881:63298 [0] NCCL INFO Channel 13/0 : 32[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96888:97306 [0] NCCL INFO Channel 02/0 : 16[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364908:1365332 [6] NCCL INFO Channel 05/0 : 14[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32705:33128 [1] NCCL INFO Channel 15/0 : 9[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124160:124581 [0] NCCL INFO Channel 04/0 : 64[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:124165:124584 [5] NCCL INFO Channel 00/0 : 69[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62884:63299 [3] NCCL INFO Channel 02/0 : 43[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62886:63303 [5] NCCL INFO Channel 03/0 : 45[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96894:97312 [6] NCCL INFO Channel 08/0 : 22[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32706:33122 [2] NCCL INFO Channel 07/0 : 10[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:124164:124585 [4] NCCL INFO Channel 12/0 : 68[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:124166:124583 [6] NCCL INFO Channel 10/0 : 70[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62887:63305 [6] NCCL INFO Channel 11/0 : 38[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96891:97311 [3] NCCL INFO Channel 00/0 : 19[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32710:33125 [6] NCCL INFO Channel 05/0 : 14[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:124163:124588 [3] NCCL INFO Channel 02/0 : 67[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62883:63301 [2] NCCL INFO Channel 11/0 : 34[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96895:97307 [7] NCCL INFO Channel 00/0 : 23[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96890:97313 [2] NCCL INFO Channel 00/0 : 18[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:124162:124587 [2] NCCL INFO Channel 06/0 : 66[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:124167:124586 [7] NCCL INFO Channel 04/0 : 71[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96892:97308 [4] NCCL INFO Channel 12/0 : 20[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364904:1365335 [2] NCCL INFO Channel 01/0 : 10[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:124165:124584 [5] NCCL INFO Channel 02/0 : 69[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124164:124585 [4] NCCL INFO Channel 14/0 : 68[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:124166:124583 [6] NCCL INFO Channel 12/0 : 70[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96888:97306 [0] NCCL INFO Channel 04/0 : 16[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364902:1365333 [0] NCCL INFO Channel 01/0 : 8[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32707:33124 [3] NCCL INFO Channel 09/0 : 11[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62888:63302 [7] NCCL INFO Channel 03/0 : 47[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364905:1365328 [3] NCCL INFO Channel 01/0 : 11[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32708:33126 [4] NCCL INFO Channel 09/0 : 12[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32709:33129 [5] NCCL INFO Channel 01/0 : 29[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32704:33123 [0] NCCL INFO Channel 07/0 : 8[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:124163:124588 [3] NCCL INFO Channel 04/0 : 67[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:124162:124587 [2] NCCL INFO Channel 08/0 : 66[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:124160:124581 [0] NCCL INFO Channel 06/0 : 64[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62885:63304 [4] NCCL INFO Channel 11/0 : 36[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62882:63300 [1] NCCL INFO Channel 00/0 : 41[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364909:1365327 [7] NCCL INFO Channel 01/0 : 15[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32711:33127 [7] NCCL INFO Channel 11/0 : 15[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32705:33128 [1] NCCL INFO Channel 01/0 : 25[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32706:33122 [2] NCCL INFO Channel 09/0 : 10[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:124167:124586 [7] NCCL INFO Channel 06/0 : 71[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62881:63298 [0] NCCL INFO Channel 15/0 : 32[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96894:97312 [6] NCCL INFO Channel 10/0 : 22[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364906:1365331 [4] NCCL INFO Channel 09/0 : 12[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32710:33125 [6] NCCL INFO Channel 07/0 : 14[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62886:63303 [5] NCCL INFO Channel 04/0 : 45[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96893:97310 [5] NCCL INFO Channel 00/0 : 21[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364908:1365332 [6] NCCL INFO Channel 07/0 : 14[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32707:33124 [3] NCCL INFO Channel 13/0 : 11[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62884:63299 [3] NCCL INFO Channel 04/0 : 43[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96891:97311 [3] NCCL INFO Channel 02/0 : 19[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364904:1365335 [2] NCCL INFO Channel 03/0 : 10[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32709:33129 [5] NCCL INFO Channel 03/0 : 29[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124161:124582 [1] NCCL INFO Channel 00/0 : 65[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124164:124585 [4] NCCL INFO Channel 00/0 : 4[4] -> 68[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:124165:124584 [5] NCCL INFO Channel 04/0 : 69[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124166:124583 [6] NCCL INFO Channel 14/0 : 70[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62887:63305 [6] NCCL INFO Channel 13/0 : 38[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96890:97313 [2] NCCL INFO Channel 04/0 : 18[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364907:1365329 [5] NCCL INFO Channel 01/0 : 13[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32704:33123 [0] NCCL INFO Channel 09/0 : 8[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62883:63301 [2] NCCL INFO Channel 13/0 : 34[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96895:97307 [7] NCCL INFO Channel 02/0 : 23[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96892:97308 [4] NCCL INFO Channel 14/0 : 20[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96888:97306 [0] NCCL INFO Channel 06/0 : 16[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364902:1365333 [0] NCCL INFO Channel 03/0 : 8[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32708:33126 [4] NCCL INFO Channel 11/0 : 12[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:124162:124587 [2] NCCL INFO Channel 10/0 : 66[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62888:63302 [7] NCCL INFO Channel 04/0 : 47[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96894:97312 [6] NCCL INFO Channel 12/0 : 22[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364905:1365328 [3] NCCL INFO Channel 03/0 : 11[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:124163:124588 [3] NCCL INFO Channel 06/0 : 67[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96893:97310 [5] NCCL INFO Channel 02/0 : 21[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96891:97311 [3] NCCL INFO Channel 04/0 : 19[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:124160:124581 [0] NCCL INFO Channel 08/0 : 64[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:124167:124586 [7] NCCL INFO Channel 08/0 : 71[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32705:33128 [1] NCCL INFO Channel 03/0 : 25[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32706:33122 [2] NCCL INFO Channel 11/0 : 10[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32711:33127 [7] NCCL INFO Channel 13/0 : 15[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62882:63300 [1] NCCL INFO Channel 02/0 : 41[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32710:33125 [6] NCCL INFO Channel 09/0 : 14[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62885:63304 [4] NCCL INFO Channel 13/0 : 36[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364906:1365331 [4] NCCL INFO Channel 11/0 : 12[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364909:1365327 [7] NCCL INFO Channel 03/0 : 15[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364908:1365332 [6] NCCL INFO Channel 09/0 : 14[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32707:33124 [3] NCCL INFO Channel 15/0 : 11[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:124166:124583 [6] NCCL INFO Channel 00/0 : 6[6] -> 70[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62881:63298 [0] NCCL INFO Channel 01/0 : 40[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96890:97313 [2] NCCL INFO Channel 06/0 : 18[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364904:1365335 [2] NCCL INFO Channel 05/0 : 10[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32709:33129 [5] NCCL INFO Channel 05/0 : 29[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124161:124582 [1] NCCL INFO Channel 02/0 : 65[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124164:124585 [4] NCCL INFO Channel 02/0 : 4[4] -> 68[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62886:63303 [5] NCCL INFO Channel 06/0 : 45[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96888:97306 [0] NCCL INFO Channel 08/0 : 16[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96895:97307 [7] NCCL INFO Channel 04/0 : 23[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364907:1365329 [5] NCCL INFO Channel 03/0 : 13[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32708:33126 [4] NCCL INFO Channel 13/0 : 12[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32704:33123 [0] NCCL INFO Channel 11/0 : 8[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:124165:124584 [5] NCCL INFO Channel 06/0 : 69[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62884:63299 [3] NCCL INFO Channel 05/0 : 43[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96894:97312 [6] NCCL INFO Channel 14/0 : 22[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96889:97309 [1] NCCL INFO Channel 00/0 : 17[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364902:1365333 [0] NCCL INFO Channel 05/0 : 8[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32705:33128 [1] NCCL INFO Channel 05/0 : 25[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124162:124587 [2] NCCL INFO Channel 12/0 : 66[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62887:63305 [6] NCCL INFO Channel 15/0 : 38[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96891:97311 [3] NCCL INFO Channel 06/0 : 19[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96890:97313 [2] NCCL INFO Channel 08/0 : 18[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364905:1365328 [3] NCCL INFO Channel 05/0 : 11[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32706:33122 [2] NCCL INFO Channel 13/0 : 10[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:124163:124588 [3] NCCL INFO Channel 08/0 : 67[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62888:63302 [7] NCCL INFO Channel 05/0 : 47[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62883:63301 [2] NCCL INFO Channel 15/0 : 34[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96893:97310 [5] NCCL INFO Channel 04/0 : 21[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364906:1365331 [4] NCCL INFO Channel 13/0 : 12[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32711:33127 [7] NCCL INFO Channel 01/0 : 31[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124167:124586 [7] NCCL INFO Channel 10/0 : 71[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62882:63300 [1] NCCL INFO Channel 03/0 : 41[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364904:1365335 [2] NCCL INFO Channel 07/0 : 10[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:124160:124581 [0] NCCL INFO Channel 10/0 : 64[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62885:63304 [4] NCCL INFO Channel 15/0 : 36[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62881:63298 [0] NCCL INFO Channel 02/0 : 40[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364909:1365327 [7] NCCL INFO Channel 05/0 : 15[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364908:1365332 [6] NCCL INFO Channel 11/0 : 14[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:124166:124583 [6] NCCL INFO Channel 02/0 : 6[6] -> 70[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62886:63303 [5] NCCL INFO Channel 07/0 : 45[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364903:1365334 [1] NCCL INFO Channel 01/0 : 9[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364907:1365329 [5] NCCL INFO Channel 05/0 : 13[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364902:1365333 [0] NCCL INFO Channel 07/0 : 8[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364905:1365328 [3] NCCL INFO Channel 07/0 : 11[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:124164:124585 [4] NCCL INFO Channel 06/0 : 4[4] -> 68[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:124161:124582 [1] NCCL INFO Channel 04/0 : 65[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96888:97306 [0] NCCL INFO Channel 10/0 : 16[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364906:1365331 [4] NCCL INFO Channel 15/0 : 12[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32707:33124 [3] NCCL INFO Channel 01/0 : 27[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32710:33125 [6] NCCL INFO Channel 11/0 : 14[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96889:97309 [1] NCCL INFO Channel 02/0 : 17[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96895:97307 [7] NCCL INFO Channel 06/0 : 23[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96890:97313 [2] NCCL INFO Channel 10/0 : 18[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96891:97311 [3] NCCL INFO Channel 08/0 : 19[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32709:33129 [5] NCCL INFO Channel 07/0 : 29[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227470:1227885 [7] NCCL INFO Channel 00/0 : 79[7] -> 87[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124162:124587 [2] NCCL INFO Channel 14/0 : 66[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:124165:124584 [5] NCCL INFO Channel 08/0 : 69[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62884:63299 [3] NCCL INFO Channel 06/0 : 43[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62887:63305 [6] NCCL INFO Channel 00/0 : 46[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96893:97310 [5] NCCL INFO Channel 06/0 : 21[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32708:33126 [4] NCCL INFO Channel 15/0 : 12[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:124163:124588 [3] NCCL INFO Channel 10/0 : 67[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62888:63302 [7] NCCL INFO Channel 06/0 : 47[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96888:97306 [0] NCCL INFO Channel 12/0 : 16[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364904:1365335 [2] NCCL INFO Channel 09/0 : 10[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32704:33123 [0] NCCL INFO Channel 13/0 : 8[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32705:33128 [1] NCCL INFO Channel 07/0 : 25[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124167:124586 [7] NCCL INFO Channel 12/0 : 71[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124166:124583 [6] NCCL INFO Channel 04/0 : 6[6] -> 70[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62883:63301 [2] NCCL INFO Channel 00/0 : 42[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96889:97309 [1] NCCL INFO Channel 04/0 : 17[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364908:1365332 [6] NCCL INFO Channel 13/0 : 14[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364909:1365327 [7] NCCL INFO Channel 07/0 : 15[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32706:33122 [2] NCCL INFO Channel 15/0 : 10[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:124160:124581 [0] NCCL INFO Channel 12/0 : 64[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62882:63300 [1] NCCL INFO Channel 04/0 : 41[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364903:1365334 [1] NCCL INFO Channel 03/0 : 9[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364902:1365333 [0] NCCL INFO Channel 09/0 : 8[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32711:33127 [7] NCCL INFO Channel 03/0 : 31[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227470:1227885 [7] NCCL INFO Channel 01/0 : 79[7] -> 87[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124164:124585 [4] NCCL INFO Channel 08/0 : 4[4] -> 68[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:124161:124582 [1] NCCL INFO Channel 06/0 : 65[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62881:63298 [0] NCCL INFO Channel 03/0 : 40[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364905:1365328 [3] NCCL INFO Channel 09/0 : 11[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32707:33124 [3] NCCL INFO Channel 03/0 : 27[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32709:33129 [5] NCCL INFO Channel 09/0 : 29[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124162:124587 [2] NCCL INFO Channel 00/0 : 2[2] -> 66[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62885:63304 [4] NCCL INFO Channel 00/0 : 44[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62886:63303 [5] NCCL INFO Channel 08/0 : 45[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364907:1365329 [5] NCCL INFO Channel 07/0 : 13[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32710:33125 [6] NCCL INFO Channel 13/0 : 14[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62884:63299 [3] NCCL INFO Channel 07/0 : 43[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62887:63305 [6] NCCL INFO Channel 01/0 : 46[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364904:1365335 [2] NCCL INFO Channel 11/0 : 10[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32708:33126 [4] NCCL INFO Channel 01/0 : 28[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:124165:124584 [5] NCCL INFO Channel 10/0 : 69[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62888:63302 [7] NCCL INFO Channel 08/0 : 47[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364908:1365332 [6] NCCL INFO Channel 15/0 : 14[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32705:33128 [1] NCCL INFO Channel 11/0 : 25[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62883:63301 [2] NCCL INFO Channel 01/0 : 42[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364909:1365327 [7] NCCL INFO Channel 09/0 : 15[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32706:33122 [2] NCCL INFO Channel 01/0 : 26[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32704:33123 [0] NCCL INFO Channel 15/0 : 8[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:124163:124588 [3] NCCL INFO Channel 12/0 : 67[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62882:63300 [1] NCCL INFO Channel 05/0 : 41[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364902:1365333 [0] NCCL INFO Channel 11/0 : 8[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364903:1365334 [1] NCCL INFO Channel 05/0 : 9[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227464:1227884 [1] NCCL INFO Channel 00/0 : 73[1] -> 81[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227470:1227885 [7] NCCL INFO Channel 02/0 : 79[7] -> 87[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364905:1365328 [3] NCCL INFO Channel 13/0 : 11[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364907:1365329 [5] NCCL INFO Channel 09/0 : 13[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364904:1365335 [2] NCCL INFO Channel 13/0 : 10[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:124167:124586 [7] NCCL INFO Channel 14/0 : 71[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124166:124583 [6] NCCL INFO Channel 08/0 : 6[6] -> 70[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364909:1365327 [7] NCCL INFO Channel 11/0 : 15[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32711:33127 [7] NCCL INFO Channel 05/0 : 31[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124160:124581 [0] NCCL INFO Channel 14/0 : 64[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:124164:124585 [4] NCCL INFO Channel 10/0 : 4[4] -> 68[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32707:33124 [3] NCCL INFO Channel 05/0 : 27[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32709:33129 [5] NCCL INFO Channel 11/0 : 29[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124162:124587 [2] NCCL INFO Channel 04/0 : 2[2] -> 66[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62881:63298 [0] NCCL INFO Channel 04/0 : 40[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62886:63303 [5] NCCL INFO Channel 09/0 : 45[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62885:63304 [4] NCCL INFO Channel 01/0 : 44[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32710:33125 [6] NCCL INFO Channel 15/0 : 14[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227466:1227887 [3] NCCL INFO Channel 00/0 : 75[3] -> 83[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:124161:124582 [1] NCCL INFO Channel 08/0 : 65[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62887:63305 [6] NCCL INFO Channel 02/0 : 46[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62888:63302 [7] NCCL INFO Channel 09/0 : 47[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62884:63299 [3] NCCL INFO Channel 08/0 : 43[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32708:33126 [4] NCCL INFO Channel 03/0 : 28[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32705:33128 [1] NCCL INFO Channel 13/0 : 25[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227470:1227885 [7] NCCL INFO Channel 03/0 : 79[7] -> 87[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124165:124584 [5] NCCL INFO Channel 12/0 : 69[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62883:63301 [2] NCCL INFO Channel 03/0 : 42[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32706:33122 [2] NCCL INFO Channel 03/0 : 26[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227464:1227884 [1] NCCL INFO Channel 02/0 : 73[1] -> 81[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124163:124588 [3] NCCL INFO Channel 14/0 : 67[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62882:63300 [1] NCCL INFO Channel 06/0 : 41[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364903:1365334 [1] NCCL INFO Channel 07/0 : 9[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32704:33123 [0] NCCL INFO Channel 01/0 : 24[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32711:33127 [7] NCCL INFO Channel 07/0 : 31[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62885:63304 [4] NCCL INFO Channel 02/0 : 44[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62886:63303 [5] NCCL INFO Channel 10/0 : 45[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62881:63298 [0] NCCL INFO Channel 05/0 : 40[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364902:1365333 [0] NCCL INFO Channel 13/0 : 8[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32707:33124 [3] NCCL INFO Channel 07/0 : 27[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:124166:124583 [6] NCCL INFO Channel 10/0 : 6[6] -> 70[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62888:63302 [7] NCCL INFO Channel 10/0 : 47[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364904:1365335 [2] NCCL INFO Channel 15/0 : 10[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32709:33129 [5] NCCL INFO Channel 15/0 : 29[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124167:124586 [7] NCCL INFO Channel 00/0 : 7[7] -> 71[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62887:63305 [6] NCCL INFO Channel 03/0 : 46[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364907:1365329 [5] NCCL INFO Channel 11/0 : 13[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32710:33125 [6] NCCL INFO Channel 01/0 : 30[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227468:1227883 [5] NCCL INFO Channel 00/0 : 77[5] -> 85[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227466:1227887 [3] NCCL INFO Channel 01/0 : 75[3] -> 83[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:124164:124585 [4] NCCL INFO Channel 12/0 : 4[4] -> 68[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62884:63299 [3] NCCL INFO Channel 09/0 : 43[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364909:1365327 [7] NCCL INFO Channel 13/0 : 15[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32705:33128 [1] NCCL INFO Channel 15/0 : 25[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227470:1227885 [7] NCCL INFO Channel 04/0 : 79[7] -> 87[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124160:124581 [0] NCCL INFO Channel 02/0 : 0[0] -> 64[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62883:63301 [2] NCCL INFO Channel 04/0 : 42[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364905:1365328 [3] NCCL INFO Channel 15/0 : 11[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32708:33126 [4] NCCL INFO Channel 05/0 : 28[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32706:33122 [2] NCCL INFO Channel 05/0 : 26[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:124162:124587 [2] NCCL INFO Channel 06/0 : 2[2] -> 66[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:124165:124584 [5] NCCL INFO Channel 14/0 : 69[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364903:1365334 [1] NCCL INFO Channel 11/0 : 9[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32704:33123 [0] NCCL INFO Channel 03/0 : 24[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:124161:124582 [1] NCCL INFO Channel 10/0 : 65[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124163:124588 [3] NCCL INFO Channel 00/0 : 3[3] -> 67[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364902:1365333 [0] NCCL INFO Channel 15/0 : 8[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32711:33127 [7] NCCL INFO Channel 09/0 : 31[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124166:124583 [6] NCCL INFO Channel 12/0 : 6[6] -> 70[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364907:1365329 [5] NCCL INFO Channel 15/0 : 13[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32707:33124 [3] NCCL INFO Channel 09/0 : 27[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227469:1227882 [6] NCCL INFO Channel 00/0 : 78[6] -> 86[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:124167:124586 [7] NCCL INFO Channel 02/0 : 7[7] -> 71[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62882:63300 [1] NCCL INFO Channel 07/0 : 41[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364903:1365334 [1] NCCL INFO Channel 13/0 : 9[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227464:1227884 [1] NCCL INFO Channel 03/0 : 73[1] -> 81[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124164:124585 [4] NCCL INFO Channel 14/0 : 4[4] -> 68[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62886:63303 [5] NCCL INFO Channel 11/0 : 45[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62881:63298 [0] NCCL INFO Channel 06/0 : 40[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227463:1227880 [0] NCCL INFO Channel 01/0 : 72[0] -> 80[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227466:1227887 [3] NCCL INFO Channel 02/0 : 75[3] -> 83[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62885:63304 [4] NCCL INFO Channel 03/0 : 44[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32710:33125 [6] NCCL INFO Channel 03/0 : 30[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227470:1227885 [7] NCCL INFO Channel 05/0 : 79[7] -> 87[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124160:124581 [0] NCCL INFO Channel 04/0 : 0[0] -> 64[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62888:63302 [7] NCCL INFO Channel 11/0 : 47[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32708:33126 [4] NCCL INFO Channel 07/0 : 28[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227468:1227883 [5] NCCL INFO Channel 01/0 : 77[5] -> 85[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124165:124584 [5] NCCL INFO Channel 00/0 : 5[5] -> 69[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62887:63305 [6] NCCL INFO Channel 04/0 : 46[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364903:1365334 [1] NCCL INFO Channel 15/0 : 9[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32706:33122 [2] NCCL INFO Channel 07/0 : 26[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:124162:124587 [2] NCCL INFO Channel 08/0 : 2[2] -> 66[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62884:63299 [3] NCCL INFO Channel 10/0 : 43[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62883:63301 [2] NCCL INFO Channel 05/0 : 42[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364907:1365329 [5] NCCL INFO Channel 01/0 : 29[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32711:33127 [7] NCCL INFO Channel 11/0 : 31[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124163:124588 [3] NCCL INFO Channel 02/0 : 3[3] -> 67[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62882:63300 [1] NCCL INFO Channel 08/0 : 41[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32707:33124 [3] NCCL INFO Channel 13/0 : 27[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:124166:124583 [6] NCCL INFO Channel 14/0 : 6[6] -> 70[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:124161:124582 [1] NCCL INFO Channel 12/0 : 65[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62885:63304 [4] NCCL INFO Channel 05/0 : 44[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62886:63303 [5] NCCL INFO Channel 12/0 : 45[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62881:63298 [0] NCCL INFO Channel 07/0 : 40[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32704:33123 [0] NCCL INFO Channel 05/0 : 24[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32710:33125 [6] NCCL INFO Channel 05/0 : 30[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227469:1227882 [6] NCCL INFO Channel 01/0 : 78[6] -> 86[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:124167:124586 [7] NCCL INFO Channel 04/0 : 7[7] -> 71[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62888:63302 [7] NCCL INFO Channel 12/0 : 47[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62887:63305 [6] NCCL INFO Channel 05/0 : 46[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32708:33126 [4] NCCL INFO Channel 09/0 : 28[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227467:1227886 [4] NCCL INFO Channel 00/0 : 76[4] -> 84[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:124160:124581 [0] NCCL INFO Channel 06/0 : 0[0] -> 64[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62884:63299 [3] NCCL INFO Channel 12/0 : 43[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364903:1365334 [1] NCCL INFO Channel 01/0 : 25[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32706:33122 [2] NCCL INFO Channel 09/0 : 26[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227464:1227884 [1] NCCL INFO Channel 04/0 : 73[1] -> 81[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364907:1365329 [5] NCCL INFO Channel 03/0 : 29[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32711:33127 [7] NCCL INFO Channel 13/0 : 31[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227470:1227885 [7] NCCL INFO Channel 06/0 : 79[7] -> 87[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124165:124584 [5] NCCL INFO Channel 02/0 : 5[5] -> 69[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364903:1365334 [1] NCCL INFO Channel 03/0 : 25[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32707:33124 [3] NCCL INFO Channel 15/0 : 27[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227466:1227887 [3] NCCL INFO Channel 04/0 : 75[3] -> 83[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:124163:124588 [3] NCCL INFO Channel 04/0 : 3[3] -> 67[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:124162:124587 [2] NCCL INFO Channel 10/0 : 2[2] -> 66[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32704:33123 [0] NCCL INFO Channel 07/0 : 24[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32708:33126 [4] NCCL INFO Channel 11/0 : 28[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227468:1227883 [5] NCCL INFO Channel 02/0 : 77[5] -> 85[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124167:124586 [7] NCCL INFO Channel 06/0 : 7[7] -> 71[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62883:63301 [2] NCCL INFO Channel 06/0 : 42[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32710:33125 [6] NCCL INFO Channel 07/0 : 30[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227469:1227882 [6] NCCL INFO Channel 02/0 : 78[6] -> 86[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227465:1227881 [2] NCCL INFO Channel 00/0 : 74[2] -> 82[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227463:1227880 [0] NCCL INFO Channel 02/0 : 72[0] -> 80[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:124161:124582 [1] NCCL INFO Channel 14/0 : 65[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124160:124581 [0] NCCL INFO Channel 08/0 : 0[0] -> 64[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62882:63300 [1] NCCL INFO Channel 10/0 : 41[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364907:1365329 [5] NCCL INFO Channel 05/0 : 29[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32706:33122 [2] NCCL INFO Channel 11/0 : 26[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:124165:124584 [5] NCCL INFO Channel 04/0 : 5[5] -> 69[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62886:63303 [5] NCCL INFO Channel 14/0 : 45[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62881:63298 [0] NCCL INFO Channel 09/0 : 40[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364905:1365328 [3] NCCL INFO Channel 01/0 : 27[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364903:1365334 [1] NCCL INFO Channel 05/0 : 25[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124163:124588 [3] NCCL INFO Channel 06/0 : 3[3] -> 67[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62888:63302 [7] NCCL INFO Channel 13/0 : 47[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124167:124586 [7] NCCL INFO Channel 08/0 : 7[7] -> 71[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62885:63304 [4] NCCL INFO Channel 06/0 : 44[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62887:63305 [6] NCCL INFO Channel 07/0 : 46[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227467:1227886 [4] NCCL INFO Channel 01/0 : 76[4] -> 84[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:124160:124581 [0] NCCL INFO Channel 10/0 : 0[0] -> 64[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62884:63299 [3] NCCL INFO Channel 13/0 : 43[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62882:63300 [1] NCCL INFO Channel 11/0 : 41[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32708:33126 [4] NCCL INFO Channel 13/0 : 28[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227464:1227884 [1] NCCL INFO Channel 05/0 : 73[1] -> 81[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124161:124582 [1] NCCL INFO Channel 00/0 : 1[1] -> 65[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62883:63301 [2] NCCL INFO Channel 07/0 : 42[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364909:1365327 [7] NCCL INFO Channel 01/0 : 31[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364907:1365329 [5] NCCL INFO Channel 07/0 : 29[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32710:33125 [6] NCCL INFO Channel 09/0 : 30[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32704:33123 [0] NCCL INFO Channel 09/0 : 24[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32706:33122 [2] NCCL INFO Channel 13/0 : 26[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227470:1227885 [7] NCCL INFO Channel 08/0 : 79[7] -> 87[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124165:124584 [5] NCCL INFO Channel 06/0 : 5[5] -> 69[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124162:124587 [2] NCCL INFO Channel 12/0 : 2[2] -> 66[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62886:63303 [5] NCCL INFO Channel 15/0 : 45[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62881:63298 [0] NCCL INFO Channel 10/0 : 40[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364903:1365334 [1] NCCL INFO Channel 07/0 : 25[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32704:33123 [0] NCCL INFO Channel 11/0 : 24[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32708:33126 [4] NCCL INFO Channel 15/0 : 28[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32710:33125 [6] NCCL INFO Channel 11/0 : 30[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227468:1227883 [5] NCCL INFO Channel 03/0 : 77[5] -> 85[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124163:124588 [3] NCCL INFO Channel 08/0 : 3[3] -> 67[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62888:63302 [7] NCCL INFO Channel 14/0 : 47[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364905:1365328 [3] NCCL INFO Channel 03/0 : 27[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32706:33122 [2] NCCL INFO Channel 15/0 : 26[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227466:1227887 [3] NCCL INFO Channel 05/0 : 75[3] -> 83[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:124167:124586 [7] NCCL INFO Channel 10/0 : 7[7] -> 71[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62885:63304 [4] NCCL INFO Channel 07/0 : 44[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62887:63305 [6] NCCL INFO Channel 08/0 : 46[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227469:1227882 [6] NCCL INFO Channel 03/0 : 78[6] -> 86[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:124160:124581 [0] NCCL INFO Channel 12/0 : 0[0] -> 64[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:124161:124582 [1] NCCL INFO Channel 02/0 : 1[1] -> 65[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124165:124584 [5] NCCL INFO Channel 08/0 : 5[5] -> 69[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32710:33125 [6] NCCL INFO Channel 13/0 : 30[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:124162:124587 [2] NCCL INFO Channel 14/0 : 2[2] -> 66[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:124163:124588 [3] NCCL INFO Channel 10/0 : 3[3] -> 67[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32704:33123 [0] NCCL INFO Channel 13/0 : 24[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227465:1227881 [2] NCCL INFO Channel 01/0 : 74[2] -> 82[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227463:1227880 [0] NCCL INFO Channel 03/0 : 72[0] -> 80[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227467:1227886 [4] NCCL INFO Channel 02/0 : 76[4] -> 84[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227470:1227885 [7] NCCL INFO Channel 09/0 : 79[7] -> 87[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227464:1227884 [1] NCCL INFO Channel 06/0 : 73[1] -> 81[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124167:124586 [7] NCCL INFO Channel 12/0 : 7[7] -> 71[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32710:33125 [6] NCCL INFO Channel 15/0 : 30[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227468:1227883 [5] NCCL INFO Channel 04/0 : 77[5] -> 85[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124160:124581 [0] NCCL INFO Channel 14/0 : 0[0] -> 64[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:124161:124582 [1] NCCL INFO Channel 04/0 : 1[1] -> 65[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364907:1365329 [5] NCCL INFO Channel 09/0 : 29[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364909:1365327 [7] NCCL INFO Channel 03/0 : 31[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227466:1227887 [3] NCCL INFO Channel 06/0 : 75[3] -> 83[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:124165:124584 [5] NCCL INFO Channel 10/0 : 5[5] -> 69[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364903:1365334 [1] NCCL INFO Channel 11/0 : 25[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227469:1227882 [6] NCCL INFO Channel 04/0 : 78[6] -> 86[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:124163:124588 [3] NCCL INFO Channel 12/0 : 3[3] -> 67[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364905:1365328 [3] NCCL INFO Channel 05/0 : 27[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32704:33123 [0] NCCL INFO Channel 15/0 : 24[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227465:1227881 [2] NCCL INFO Channel 03/0 : 74[2] -> 82[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227470:1227885 [7] NCCL INFO Channel 10/0 : 79[7] -> 87[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124167:124586 [7] NCCL INFO Channel 14/0 : 7[7] -> 71[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227463:1227880 [0] NCCL INFO Channel 04/0 : 72[0] -> 80[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:124161:124582 [1] NCCL INFO Channel 06/0 : 1[1] -> 65[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124165:124584 [5] NCCL INFO Channel 12/0 : 5[5] -> 69[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364907:1365329 [5] NCCL INFO Channel 11/0 : 29[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364909:1365327 [7] NCCL INFO Channel 05/0 : 31[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124163:124588 [3] NCCL INFO Channel 14/0 : 3[3] -> 67[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364903:1365334 [1] NCCL INFO Channel 13/0 : 25[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124161:124582 [1] NCCL INFO Channel 08/0 : 1[1] -> 65[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364905:1365328 [3] NCCL INFO Channel 07/0 : 27[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227464:1227884 [1] NCCL INFO Channel 07/0 : 73[1] -> 81[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124165:124584 [5] NCCL INFO Channel 14/0 : 5[5] -> 69[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364906:1365331 [4] NCCL INFO Channel 01/0 : 28[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227467:1227886 [4] NCCL INFO Channel 03/0 : 76[4] -> 84[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227468:1227883 [5] NCCL INFO Channel 06/0 : 77[5] -> 85[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227466:1227887 [3] NCCL INFO Channel 07/0 : 75[3] -> 83[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364907:1365329 [5] NCCL INFO Channel 15/0 : 29[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227469:1227882 [6] NCCL INFO Channel 05/0 : 78[6] -> 86[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364909:1365327 [7] NCCL INFO Channel 07/0 : 31[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227470:1227885 [7] NCCL INFO Channel 11/0 : 79[7] -> 87[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364903:1365334 [1] NCCL INFO Channel 15/0 : 25[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227465:1227881 [2] NCCL INFO Channel 04/0 : 74[2] -> 82[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227463:1227880 [0] NCCL INFO Channel 05/0 : 72[0] -> 80[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:124161:124582 [1] NCCL INFO Channel 10/0 : 1[1] -> 65[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124161:124582 [1] NCCL INFO Channel 12/0 : 1[1] -> 65[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364904:1365335 [2] NCCL INFO Channel 01/0 : 26[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227464:1227884 [1] NCCL INFO Channel 08/0 : 73[1] -> 81[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227467:1227886 [4] NCCL INFO Channel 05/0 : 76[4] -> 84[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364906:1365331 [4] NCCL INFO Channel 03/0 : 28[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364905:1365328 [3] NCCL INFO Channel 09/0 : 27[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227468:1227883 [5] NCCL INFO Channel 07/0 : 77[5] -> 85[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227469:1227882 [6] NCCL INFO Channel 07/0 : 78[6] -> 86[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227466:1227887 [3] NCCL INFO Channel 08/0 : 75[3] -> 83[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364909:1365327 [7] NCCL INFO Channel 09/0 : 31[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227470:1227885 [7] NCCL INFO Channel 12/0 : 79[7] -> 87[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364908:1365332 [6] NCCL INFO Channel 01/0 : 30[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227463:1227880 [0] NCCL INFO Channel 06/0 : 72[0] -> 80[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227465:1227881 [2] NCCL INFO Channel 05/0 : 74[2] -> 82[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364904:1365335 [2] NCCL INFO Channel 03/0 : 26[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227464:1227884 [1] NCCL INFO Channel 10/0 : 73[1] -> 81[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124161:124582 [1] NCCL INFO Channel 14/0 : 1[1] -> 65[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235648:236064 [0] NCCL INFO Channel 01/0 : 96[0] -> 104[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364905:1365328 [3] NCCL INFO Channel 13/0 : 27[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364902:1365333 [0] NCCL INFO Channel 01/0 : 24[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364906:1365331 [4] NCCL INFO Channel 05/0 : 28[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227468:1227883 [5] NCCL INFO Channel 08/0 : 77[5] -> 85[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364909:1365327 [7] NCCL INFO Channel 11/0 : 31[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227469:1227882 [6] NCCL INFO Channel 08/0 : 78[6] -> 86[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227467:1227886 [4] NCCL INFO Channel 06/0 : 76[4] -> 84[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227470:1227885 [7] NCCL INFO Channel 13/0 : 79[7] -> 87[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364904:1365335 [2] NCCL INFO Channel 05/0 : 26[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364908:1365332 [6] NCCL INFO Channel 03/0 : 30[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227466:1227887 [3] NCCL INFO Channel 09/0 : 75[3] -> 83[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227463:1227880 [0] NCCL INFO Channel 07/0 : 72[0] -> 80[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227464:1227884 [1] NCCL INFO Channel 11/0 : 73[1] -> 81[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227465:1227881 [2] NCCL INFO Channel 06/0 : 74[2] -> 82[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235648:236064 [0] NCCL INFO Channel 03/0 : 96[0] -> 104[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227468:1227883 [5] NCCL INFO Channel 09/0 : 77[5] -> 85[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364905:1365328 [3] NCCL INFO Channel 15/0 : 27[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364902:1365333 [0] NCCL INFO Channel 03/0 : 24[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364906:1365331 [4] NCCL INFO Channel 07/0 : 28[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364909:1365327 [7] NCCL INFO Channel 13/0 : 31[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364904:1365335 [2] NCCL INFO Channel 07/0 : 26[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364908:1365332 [6] NCCL INFO Channel 05/0 : 30[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364902:1365333 [0] NCCL INFO Channel 05/0 : 24[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235648:236064 [0] NCCL INFO Channel 05/0 : 96[0] -> 104[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364906:1365331 [4] NCCL INFO Channel 09/0 : 28[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235648:236064 [0] NCCL INFO Channel 07/0 : 96[0] -> 104[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364904:1365335 [2] NCCL INFO Channel 09/0 : 26[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364908:1365332 [6] NCCL INFO Channel 07/0 : 30[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364902:1365333 [0] NCCL INFO Channel 07/0 : 24[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364906:1365331 [4] NCCL INFO Channel 11/0 : 28[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235650:236065 [2] NCCL INFO Channel 01/0 : 98[2] -> 106[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364904:1365335 [2] NCCL INFO Channel 11/0 : 26[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364908:1365332 [6] NCCL INFO Channel 09/0 : 30[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364902:1365333 [0] NCCL INFO Channel 09/0 : 24[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364906:1365331 [4] NCCL INFO Channel 13/0 : 28[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235648:236064 [0] NCCL INFO Channel 09/0 : 96[0] -> 104[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235652:236066 [4] NCCL INFO Channel 01/0 : 100[4] -> 108[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235654:236070 [6] NCCL INFO Channel 01/0 : 102[6] -> 110[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364904:1365335 [2] NCCL INFO Channel 13/0 : 26[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364908:1365332 [6] NCCL INFO Channel 11/0 : 30[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364906:1365331 [4] NCCL INFO Channel 15/0 : 28[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235650:236065 [2] NCCL INFO Channel 03/0 : 98[2] -> 106[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235648:236064 [0] NCCL INFO Channel 11/0 : 96[0] -> 104[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235652:236066 [4] NCCL INFO Channel 03/0 : 100[4] -> 108[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364902:1365333 [0] NCCL INFO Channel 11/0 : 24[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364904:1365335 [2] NCCL INFO Channel 15/0 : 26[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364908:1365332 [6] NCCL INFO Channel 13/0 : 30[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235654:236070 [6] NCCL INFO Channel 03/0 : 102[6] -> 110[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235650:236065 [2] NCCL INFO Channel 05/0 : 98[2] -> 106[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235648:236064 [0] NCCL INFO Channel 13/0 : 96[0] -> 104[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364908:1365332 [6] NCCL INFO Channel 15/0 : 30[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364902:1365333 [0] NCCL INFO Channel 13/0 : 24[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235652:236066 [4] NCCL INFO Channel 05/0 : 100[4] -> 108[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235655:236069 [7] NCCL INFO Channel 01/0 : 103[7] -> 111[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235654:236070 [6] NCCL INFO Channel 05/0 : 102[6] -> 110[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235650:236065 [2] NCCL INFO Channel 07/0 : 98[2] -> 106[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235648:236064 [0] NCCL INFO Channel 15/0 : 96[0] -> 104[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364902:1365333 [0] NCCL INFO Channel 15/0 : 24[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235651:236068 [3] NCCL INFO Channel 01/0 : 99[3] -> 107[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235652:236066 [4] NCCL INFO Channel 07/0 : 100[4] -> 108[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235653:236067 [5] NCCL INFO Channel 01/0 : 101[5] -> 109[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235655:236069 [7] NCCL INFO Channel 03/0 : 103[7] -> 111[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235648:236064 [0] NCCL INFO Channel 01/0 : 104[0] -> 112[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235654:236070 [6] NCCL INFO Channel 07/0 : 102[6] -> 110[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235650:236065 [2] NCCL INFO Channel 09/0 : 98[2] -> 106[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235652:236066 [4] NCCL INFO Channel 09/0 : 100[4] -> 108[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235651:236068 [3] NCCL INFO Channel 03/0 : 99[3] -> 107[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235648:236064 [0] NCCL INFO Channel 02/0 : 104[0] -> 112[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235653:236067 [5] NCCL INFO Channel 03/0 : 101[5] -> 109[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235655:236069 [7] NCCL INFO Channel 05/0 : 103[7] -> 111[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235650:236065 [2] NCCL INFO Channel 11/0 : 98[2] -> 106[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235654:236070 [6] NCCL INFO Channel 09/0 : 102[6] -> 110[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235652:236066 [4] NCCL INFO Channel 11/0 : 100[4] -> 108[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235649:236071 [1] NCCL INFO Channel 01/0 : 97[1] -> 105[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235651:236068 [3] NCCL INFO Channel 05/0 : 99[3] -> 107[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235648:236064 [0] NCCL INFO Channel 03/0 : 104[0] -> 112[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235653:236067 [5] NCCL INFO Channel 05/0 : 101[5] -> 109[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235652:236066 [4] NCCL INFO Channel 13/0 : 100[4] -> 108[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235650:236065 [2] NCCL INFO Channel 13/0 : 98[2] -> 106[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235655:236069 [7] NCCL INFO Channel 07/0 : 103[7] -> 111[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235654:236070 [6] NCCL INFO Channel 11/0 : 102[6] -> 110[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235648:236064 [0] NCCL INFO Channel 04/0 : 104[0] -> 112[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235651:236068 [3] NCCL INFO Channel 07/0 : 99[3] -> 107[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235649:236071 [1] NCCL INFO Channel 03/0 : 97[1] -> 105[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235652:236066 [4] NCCL INFO Channel 15/0 : 100[4] -> 108[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235653:236067 [5] NCCL INFO Channel 07/0 : 101[5] -> 109[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235650:236065 [2] NCCL INFO Channel 15/0 : 98[2] -> 106[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235655:236069 [7] NCCL INFO Channel 09/0 : 103[7] -> 111[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235648:236064 [0] NCCL INFO Channel 05/0 : 104[0] -> 112[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235654:236070 [6] NCCL INFO Channel 13/0 : 102[6] -> 110[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235651:236068 [3] NCCL INFO Channel 09/0 : 99[3] -> 107[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235649:236071 [1] NCCL INFO Channel 05/0 : 97[1] -> 105[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227467:1227886 [4] NCCL INFO Channel 07/0 : 76[4] -> 84[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227470:1227885 [7] NCCL INFO Channel 14/0 : 79[7] -> 87[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227469:1227882 [6] NCCL INFO Channel 09/0 : 78[6] -> 86[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235652:236066 [4] NCCL INFO Channel 00/0 : 108[4] -> 116[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235653:236067 [5] NCCL INFO Channel 09/0 : 101[5] -> 109[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227466:1227887 [3] NCCL INFO Channel 10/0 : 75[3] -> 83[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227463:1227880 [0] NCCL INFO Channel 09/0 : 72[0] -> 80[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227468:1227883 [5] NCCL INFO Channel 10/0 : 77[5] -> 85[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227464:1227884 [1] NCCL INFO Channel 12/0 : 73[1] -> 81[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227465:1227881 [2] NCCL INFO Channel 07/0 : 74[2] -> 82[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227467:1227886 [4] NCCL INFO Channel 08/0 : 76[4] -> 84[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227470:1227885 [7] NCCL INFO Channel 00/0 : 87[7] -> 95[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227469:1227882 [6] NCCL INFO Channel 10/0 : 78[6] -> 86[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227466:1227887 [3] NCCL INFO Channel 12/0 : 75[3] -> 83[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235650:236065 [2] NCCL INFO Channel 00/0 : 106[2] -> 114[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227463:1227880 [0] NCCL INFO Channel 10/0 : 72[0] -> 80[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235655:236069 [7] NCCL INFO Channel 11/0 : 103[7] -> 111[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235648:236064 [0] NCCL INFO Channel 06/0 : 104[0] -> 112[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227464:1227884 [1] NCCL INFO Channel 13/0 : 73[1] -> 81[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227468:1227883 [5] NCCL INFO Channel 11/0 : 77[5] -> 85[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235654:236070 [6] NCCL INFO Channel 15/0 : 102[6] -> 110[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235649:236071 [1] NCCL INFO Channel 07/0 : 97[1] -> 105[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235651:236068 [3] NCCL INFO Channel 13/0 : 99[3] -> 107[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227467:1227886 [4] NCCL INFO Channel 09/0 : 76[4] -> 84[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227470:1227885 [7] NCCL INFO Channel 02/0 : 87[7] -> 95[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227465:1227881 [2] NCCL INFO Channel 08/0 : 74[2] -> 82[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227469:1227882 [6] NCCL INFO Channel 11/0 : 78[6] -> 86[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227466:1227887 [3] NCCL INFO Channel 13/0 : 75[3] -> 83[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227463:1227880 [0] NCCL INFO Channel 11/0 : 72[0] -> 80[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227468:1227883 [5] NCCL INFO Channel 12/0 : 77[5] -> 85[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227464:1227884 [1] NCCL INFO Channel 14/0 : 73[1] -> 81[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227467:1227886 [4] NCCL INFO Channel 10/0 : 76[4] -> 84[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227470:1227885 [7] NCCL INFO Channel 04/0 : 87[7] -> 95[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227469:1227882 [6] NCCL INFO Channel 12/0 : 78[6] -> 86[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227466:1227887 [3] NCCL INFO Channel 14/0 : 75[3] -> 83[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227463:1227880 [0] NCCL INFO Channel 12/0 : 72[0] -> 80[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227465:1227881 [2] NCCL INFO Channel 09/0 : 74[2] -> 82[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227468:1227883 [5] NCCL INFO Channel 14/0 : 77[5] -> 85[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227464:1227884 [1] NCCL INFO Channel 15/0 : 73[1] -> 81[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235652:236066 [4] NCCL INFO Channel 01/0 : 108[4] -> 116[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235653:236067 [5] NCCL INFO Channel 11/0 : 101[5] -> 109[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227470:1227885 [7] NCCL INFO Channel 06/0 : 87[7] -> 95[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227467:1227886 [4] NCCL INFO Channel 11/0 : 76[4] -> 84[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235650:236065 [2] NCCL INFO Channel 01/0 : 106[2] -> 114[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227469:1227882 [6] NCCL INFO Channel 13/0 : 78[6] -> 86[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227466:1227887 [3] NCCL INFO Channel 15/0 : 75[3] -> 83[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227464:1227884 [1] NCCL INFO Channel 00/0 : 81[1] -> 89[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227463:1227880 [0] NCCL INFO Channel 13/0 : 72[0] -> 80[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227468:1227883 [5] NCCL INFO Channel 15/0 : 77[5] -> 85[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227465:1227881 [2] NCCL INFO Channel 11/0 : 74[2] -> 82[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227470:1227885 [7] NCCL INFO Channel 08/0 : 87[7] -> 95[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227467:1227886 [4] NCCL INFO Channel 13/0 : 76[4] -> 84[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227469:1227882 [6] NCCL INFO Channel 15/0 : 78[6] -> 86[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227466:1227887 [3] NCCL INFO Channel 00/0 : 83[3] -> 91[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235648:236064 [0] NCCL INFO Channel 07/0 : 104[0] -> 112[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227464:1227884 [1] NCCL INFO Channel 02/0 : 81[1] -> 89[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227468:1227883 [5] NCCL INFO Channel 00/0 : 85[5] -> 93[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227470:1227885 [7] NCCL INFO Channel 10/0 : 87[7] -> 95[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227463:1227880 [0] NCCL INFO Channel 14/0 : 72[0] -> 80[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227467:1227886 [4] NCCL INFO Channel 14/0 : 76[4] -> 84[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227465:1227881 [2] NCCL INFO Channel 12/0 : 74[2] -> 82[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227469:1227882 [6] NCCL INFO Channel 00/0 : 86[6] -> 94[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227466:1227887 [3] NCCL INFO Channel 02/0 : 83[3] -> 91[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227464:1227884 [1] NCCL INFO Channel 04/0 : 81[1] -> 89[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235654:236070 [6] NCCL INFO Channel 00/0 : 110[6] -> 118[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235655:236069 [7] NCCL INFO Channel 13/0 : 103[7] -> 111[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227468:1227883 [5] NCCL INFO Channel 02/0 : 85[5] -> 93[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235651:236068 [3] NCCL INFO Channel 15/0 : 99[3] -> 107[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235649:236071 [1] NCCL INFO Channel 11/0 : 97[1] -> 105[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227470:1227885 [7] NCCL INFO Channel 12/0 : 87[7] -> 95[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235652:236066 [4] NCCL INFO Channel 02/0 : 108[4] -> 116[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235653:236067 [5] NCCL INFO Channel 15/0 : 101[5] -> 109[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235650:236065 [2] NCCL INFO Channel 03/0 : 106[2] -> 114[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227463:1227880 [0] NCCL INFO Channel 15/0 : 72[0] -> 80[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227469:1227882 [6] NCCL INFO Channel 02/0 : 86[6] -> 94[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227465:1227881 [2] NCCL INFO Channel 13/0 : 74[2] -> 82[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235648:236064 [0] NCCL INFO Channel 09/0 : 104[0] -> 112[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227467:1227886 [4] NCCL INFO Channel 15/0 : 76[4] -> 84[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227466:1227887 [3] NCCL INFO Channel 04/0 : 83[3] -> 91[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227464:1227884 [1] NCCL INFO Channel 06/0 : 81[1] -> 89[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227468:1227883 [5] NCCL INFO Channel 04/0 : 85[5] -> 93[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227470:1227885 [7] NCCL INFO Channel 14/0 : 87[7] -> 95[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227469:1227882 [6] NCCL INFO Channel 04/0 : 86[6] -> 94[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227463:1227880 [0] NCCL INFO Channel 02/0 : 80[0] -> 88[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235655:236069 [7] NCCL INFO Channel 00/0 : 111[7] -> 119[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227467:1227886 [4] NCCL INFO Channel 00/0 : 84[4] -> 92[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227468:1227883 [5] NCCL INFO Channel 06/0 : 85[5] -> 93[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227466:1227887 [3] NCCL INFO Channel 06/0 : 83[3] -> 91[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227465:1227881 [2] NCCL INFO Channel 14/0 : 74[2] -> 82[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227464:1227884 [1] NCCL INFO Channel 08/0 : 81[1] -> 89[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227469:1227882 [6] NCCL INFO Channel 08/0 : 86[6] -> 94[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227463:1227880 [0] NCCL INFO Channel 04/0 : 80[0] -> 88[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227467:1227886 [4] NCCL INFO Channel 02/0 : 84[4] -> 92[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227468:1227883 [5] NCCL INFO Channel 08/0 : 85[5] -> 93[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227464:1227884 [1] NCCL INFO Channel 10/0 : 81[1] -> 89[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227466:1227887 [3] NCCL INFO Channel 08/0 : 83[3] -> 91[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227465:1227881 [2] NCCL INFO Channel 15/0 : 74[2] -> 82[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227469:1227882 [6] NCCL INFO Channel 10/0 : 86[6] -> 94[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227463:1227880 [0] NCCL INFO Channel 06/0 : 80[0] -> 88[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235654:236070 [6] NCCL INFO Channel 01/0 : 110[6] -> 118[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235649:236071 [1] NCCL INFO Channel 13/0 : 97[1] -> 105[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235651:236068 [3] NCCL INFO Channel 00/0 : 107[3] -> 115[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235653:236067 [5] NCCL INFO Channel 00/0 : 109[5] -> 117[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235648:236064 [0] NCCL INFO Channel 10/0 : 104[0] -> 112[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235650:236065 [2] NCCL INFO Channel 04/0 : 106[2] -> 114[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235652:236066 [4] NCCL INFO Channel 03/0 : 108[4] -> 116[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235655:236069 [7] NCCL INFO Channel 01/0 : 111[7] -> 119[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227467:1227886 [4] NCCL INFO Channel 06/0 : 84[4] -> 92[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227468:1227883 [5] NCCL INFO Channel 10/0 : 85[5] -> 93[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227466:1227887 [3] NCCL INFO Channel 10/0 : 83[3] -> 91[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227464:1227884 [1] NCCL INFO Channel 12/0 : 81[1] -> 89[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227469:1227882 [6] NCCL INFO Channel 12/0 : 86[6] -> 94[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227465:1227881 [2] NCCL INFO Channel 00/0 : 82[2] -> 90[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227463:1227880 [0] NCCL INFO Channel 08/0 : 80[0] -> 88[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235649:236071 [1] NCCL INFO Channel 15/0 : 97[1] -> 105[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227467:1227886 [4] NCCL INFO Channel 08/0 : 84[4] -> 92[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227468:1227883 [5] NCCL INFO Channel 12/0 : 85[5] -> 93[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227466:1227887 [3] NCCL INFO Channel 12/0 : 83[3] -> 91[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227469:1227882 [6] NCCL INFO Channel 14/0 : 86[6] -> 94[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227465:1227881 [2] NCCL INFO Channel 04/0 : 82[2] -> 90[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227464:1227884 [1] NCCL INFO Channel 14/0 : 81[1] -> 89[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227467:1227886 [4] NCCL INFO Channel 10/0 : 84[4] -> 92[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227463:1227880 [0] NCCL INFO Channel 10/0 : 80[0] -> 88[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227468:1227883 [5] NCCL INFO Channel 14/0 : 85[5] -> 93[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235654:236070 [6] NCCL INFO Channel 02/0 : 110[6] -> 118[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235651:236068 [3] NCCL INFO Channel 01/0 : 107[3] -> 115[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227466:1227887 [3] NCCL INFO Channel 14/0 : 83[3] -> 91[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235653:236067 [5] NCCL INFO Channel 01/0 : 109[5] -> 117[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235655:236069 [7] NCCL INFO Channel 02/0 : 111[7] -> 119[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227463:1227880 [0] NCCL INFO Channel 12/0 : 80[0] -> 88[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227467:1227886 [4] NCCL INFO Channel 12/0 : 84[4] -> 92[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227465:1227881 [2] NCCL INFO Channel 06/0 : 82[2] -> 90[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235650:236065 [2] NCCL INFO Channel 05/0 : 106[2] -> 114[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227463:1227880 [0] NCCL INFO Channel 14/0 : 80[0] -> 88[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235652:236066 [4] NCCL INFO Channel 05/0 : 108[4] -> 116[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235648:236064 [0] NCCL INFO Channel 11/0 : 104[0] -> 112[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227467:1227886 [4] NCCL INFO Channel 14/0 : 84[4] -> 92[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235649:236071 [1] NCCL INFO Channel 00/0 : 105[1] -> 113[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227465:1227881 [2] NCCL INFO Channel 08/0 : 82[2] -> 90[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235651:236068 [3] NCCL INFO Channel 02/0 : 107[3] -> 115[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235654:236070 [6] NCCL INFO Channel 03/0 : 110[6] -> 118[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227465:1227881 [2] NCCL INFO Channel 10/0 : 82[2] -> 90[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227465:1227881 [2] NCCL INFO Channel 12/0 : 82[2] -> 90[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235653:236067 [5] NCCL INFO Channel 02/0 : 109[5] -> 117[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235655:236069 [7] NCCL INFO Channel 03/0 : 111[7] -> 119[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235652:236066 [4] NCCL INFO Channel 06/0 : 108[4] -> 116[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235650:236065 [2] NCCL INFO Channel 06/0 : 106[2] -> 114[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235648:236064 [0] NCCL INFO Channel 12/0 : 104[0] -> 112[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235649:236071 [1] NCCL INFO Channel 02/0 : 105[1] -> 113[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227465:1227881 [2] NCCL INFO Channel 14/0 : 82[2] -> 90[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235654:236070 [6] NCCL INFO Channel 04/0 : 110[6] -> 118[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235651:236068 [3] NCCL INFO Channel 04/0 : 107[3] -> 115[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235653:236067 [5] NCCL INFO Channel 03/0 : 109[5] -> 117[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235655:236069 [7] NCCL INFO Channel 04/0 : 111[7] -> 119[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235648:236064 [0] NCCL INFO Channel 13/0 : 104[0] -> 112[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235652:236066 [4] NCCL INFO Channel 07/0 : 108[4] -> 116[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235650:236065 [2] NCCL INFO Channel 07/0 : 106[2] -> 114[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235649:236071 [1] NCCL INFO Channel 03/0 : 105[1] -> 113[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235654:236070 [6] NCCL INFO Channel 05/0 : 110[6] -> 118[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235653:236067 [5] NCCL INFO Channel 04/0 : 109[5] -> 117[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235651:236068 [3] NCCL INFO Channel 05/0 : 107[3] -> 115[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235655:236069 [7] NCCL INFO Channel 05/0 : 111[7] -> 119[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235648:236064 [0] NCCL INFO Channel 14/0 : 104[0] -> 112[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235652:236066 [4] NCCL INFO Channel 08/0 : 108[4] -> 116[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235650:236065 [2] NCCL INFO Channel 08/0 : 106[2] -> 114[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235649:236071 [1] NCCL INFO Channel 04/0 : 105[1] -> 113[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235653:236067 [5] NCCL INFO Channel 06/0 : 109[5] -> 117[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235654:236070 [6] NCCL INFO Channel 07/0 : 110[6] -> 118[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235651:236068 [3] NCCL INFO Channel 06/0 : 107[3] -> 115[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235655:236069 [7] NCCL INFO Channel 06/0 : 111[7] -> 119[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235648:236064 [0] NCCL INFO Channel 15/0 : 104[0] -> 112[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235652:236066 [4] NCCL INFO Channel 09/0 : 108[4] -> 116[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235650:236065 [2] NCCL INFO Channel 09/0 : 106[2] -> 114[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235649:236071 [1] NCCL INFO Channel 05/0 : 105[1] -> 113[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235653:236067 [5] NCCL INFO Channel 07/0 : 109[5] -> 117[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235654:236070 [6] NCCL INFO Channel 08/0 : 110[6] -> 118[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235655:236069 [7] NCCL INFO Channel 08/0 : 111[7] -> 119[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235651:236068 [3] NCCL INFO Channel 07/0 : 107[3] -> 115[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235652:236066 [4] NCCL INFO Channel 10/0 : 108[4] -> 116[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235650:236065 [2] NCCL INFO Channel 11/0 : 106[2] -> 114[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235653:236067 [5] NCCL INFO Channel 08/0 : 109[5] -> 117[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235649:236071 [1] NCCL INFO Channel 06/0 : 105[1] -> 113[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235654:236070 [6] NCCL INFO Channel 09/0 : 110[6] -> 118[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235655:236069 [7] NCCL INFO Channel 09/0 : 111[7] -> 119[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235652:236066 [4] NCCL INFO Channel 11/0 : 108[4] -> 116[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235651:236068 [3] NCCL INFO Channel 08/0 : 107[3] -> 115[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235653:236067 [5] NCCL INFO Channel 09/0 : 109[5] -> 117[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235650:236065 [2] NCCL INFO Channel 12/0 : 106[2] -> 114[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557266:1557689 [0] NCCL INFO Channel 02/0 : 80[0] -> 96[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235654:236070 [6] NCCL INFO Channel 10/0 : 110[6] -> 118[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235649:236071 [1] NCCL INFO Channel 07/0 : 105[1] -> 113[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235655:236069 [7] NCCL INFO Channel 10/0 : 111[7] -> 119[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557266:1557689 [0] NCCL INFO Channel 04/0 : 80[0] -> 96[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235652:236066 [4] NCCL INFO Channel 13/0 : 108[4] -> 116[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235651:236068 [3] NCCL INFO Channel 09/0 : 107[3] -> 115[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557266:1557689 [0] NCCL INFO Channel 06/0 : 80[0] -> 96[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235653:236067 [5] NCCL INFO Channel 10/0 : 109[5] -> 117[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235650:236065 [2] NCCL INFO Channel 13/0 : 106[2] -> 114[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235654:236070 [6] NCCL INFO Channel 11/0 : 110[6] -> 118[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235649:236071 [1] NCCL INFO Channel 08/0 : 105[1] -> 113[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235655:236069 [7] NCCL INFO Channel 11/0 : 111[7] -> 119[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235652:236066 [4] NCCL INFO Channel 14/0 : 108[4] -> 116[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557266:1557689 [0] NCCL INFO Channel 08/0 : 80[0] -> 96[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235651:236068 [3] NCCL INFO Channel 10/0 : 107[3] -> 115[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235653:236067 [5] NCCL INFO Channel 11/0 : 109[5] -> 117[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235650:236065 [2] NCCL INFO Channel 14/0 : 106[2] -> 114[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557266:1557689 [0] NCCL INFO Channel 10/0 : 80[0] -> 96[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235655:236069 [7] NCCL INFO Channel 12/0 : 111[7] -> 119[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235654:236070 [6] NCCL INFO Channel 12/0 : 110[6] -> 118[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235649:236071 [1] NCCL INFO Channel 10/0 : 105[1] -> 113[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235652:236066 [4] NCCL INFO Channel 15/0 : 108[4] -> 116[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235651:236068 [3] NCCL INFO Channel 12/0 : 107[3] -> 115[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235653:236067 [5] NCCL INFO Channel 12/0 : 109[5] -> 117[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235650:236065 [2] NCCL INFO Channel 15/0 : 106[2] -> 114[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557266:1557689 [0] NCCL INFO Channel 12/0 : 80[0] -> 96[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235655:236069 [7] NCCL INFO Channel 13/0 : 111[7] -> 119[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235654:236070 [6] NCCL INFO Channel 13/0 : 110[6] -> 118[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235649:236071 [1] NCCL INFO Channel 11/0 : 105[1] -> 113[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235651:236068 [3] NCCL INFO Channel 13/0 : 107[3] -> 115[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235653:236067 [5] NCCL INFO Channel 14/0 : 109[5] -> 117[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557266:1557689 [0] NCCL INFO Channel 14/0 : 80[0] -> 96[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235655:236069 [7] NCCL INFO Channel 14/0 : 111[7] -> 119[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557266:1557689 [0] NCCL INFO Channel 02/0 : 96[0] -> 112[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235654:236070 [6] NCCL INFO Channel 15/0 : 110[6] -> 118[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235649:236071 [1] NCCL INFO Channel 12/0 : 105[1] -> 113[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235653:236067 [5] NCCL INFO Channel 15/0 : 109[5] -> 117[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235651:236068 [3] NCCL INFO Channel 14/0 : 107[3] -> 115[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-52:2722050:2722471 [5] NCCL INFO Channel 12/0 : 117[5] -> 125[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-52:2722049:2722467 [4] NCCL INFO Channel 08/0 : 116[4] -> 124[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557270:1557688 [4] NCCL INFO Channel 00/0 : 84[4] -> 100[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557266:1557689 [0] NCCL INFO Channel 04/0 : 96[0] -> 112[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235649:236071 [1] NCCL INFO Channel 13/0 : 105[1] -> 113[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-52:2722048:2722469 [3] NCCL INFO Channel 08/0 : 115[3] -> 123[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-52:2722052:2722464 [7] NCCL INFO Channel 10/0 : 119[7] -> 127[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722046:2722466 [1] NCCL INFO Channel 08/0 : 113[1] -> 121[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235651:236068 [3] NCCL INFO Channel 15/0 : 107[3] -> 115[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-52:2722047:2722468 [2] NCCL INFO Channel 06/0 : 114[2] -> 122[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-52:2722051:2722465 [6] NCCL INFO Channel 10/0 : 118[6] -> 126[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-52:2722045:2722470 [0] NCCL INFO Channel 04/0 : 112[0] -> 120[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-52:2722050:2722471 [5] NCCL INFO Channel 14/0 : 117[5] -> 125[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-52:2722049:2722467 [4] NCCL INFO Channel 10/0 : 116[4] -> 124[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722048:2722469 [3] NCCL INFO Channel 10/0 : 115[3] -> 123[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557268:1557686 [2] NCCL INFO Channel 00/0 : 82[2] -> 98[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-52:2722046:2722466 [1] NCCL INFO Channel 10/0 : 113[1] -> 121[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-52:2722052:2722464 [7] NCCL INFO Channel 12/0 : 119[7] -> 127[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557270:1557688 [4] NCCL INFO Channel 02/0 : 84[4] -> 100[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235649:236071 [1] NCCL INFO Channel 14/0 : 105[1] -> 113[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-52:2722047:2722468 [2] NCCL INFO Channel 08/0 : 114[2] -> 122[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-52:2722045:2722470 [0] NCCL INFO Channel 06/0 : 112[0] -> 120[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235649:236071 [1] NCCL INFO Channel 15/0 : 105[1] -> 113[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-52:2722051:2722465 [6] NCCL INFO Channel 12/0 : 118[6] -> 126[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-52:2722048:2722469 [3] NCCL INFO Channel 12/0 : 115[3] -> 123[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-52:2722049:2722467 [4] NCCL INFO Channel 12/0 : 116[4] -> 124[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722046:2722466 [1] NCCL INFO Channel 12/0 : 113[1] -> 121[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557268:1557686 [2] NCCL INFO Channel 04/0 : 82[2] -> 98[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-52:2722052:2722464 [7] NCCL INFO Channel 14/0 : 119[7] -> 127[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722047:2722468 [2] NCCL INFO Channel 10/0 : 114[2] -> 122[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557266:1557689 [0] NCCL INFO Channel 06/0 : 96[0] -> 112[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-52:2722051:2722465 [6] NCCL INFO Channel 14/0 : 118[6] -> 126[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-52:2722045:2722470 [0] NCCL INFO Channel 08/0 : 112[0] -> 120[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-52:2722048:2722469 [3] NCCL INFO Channel 14/0 : 115[3] -> 123[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-52:2722046:2722466 [1] NCCL INFO Channel 14/0 : 113[1] -> 121[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-52:2722049:2722467 [4] NCCL INFO Channel 14/0 : 116[4] -> 124[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722047:2722468 [2] NCCL INFO Channel 12/0 : 114[2] -> 122[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557273:1557685 [7] NCCL INFO Channel 00/0 : 87[7] -> 103[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722045:2722470 [0] NCCL INFO Channel 10/0 : 112[0] -> 120[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557270:1557688 [4] NCCL INFO Channel 06/0 : 84[4] -> 100[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722047:2722468 [2] NCCL INFO Channel 14/0 : 114[2] -> 122[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557268:1557686 [2] NCCL INFO Channel 06/0 : 82[2] -> 98[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-52:2722045:2722470 [0] NCCL INFO Channel 12/0 : 112[0] -> 120[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557272:1557687 [6] NCCL INFO Channel 00/0 : 86[6] -> 102[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-52:2722045:2722470 [0] NCCL INFO Channel 14/0 : 112[0] -> 120[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557266:1557689 [0] NCCL INFO Channel 08/0 : 96[0] -> 112[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557271:1557690 [5] NCCL INFO Channel 00/0 : 85[5] -> 101[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557273:1557685 [7] NCCL INFO Channel 02/0 : 87[7] -> 103[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557269:1557684 [3] NCCL INFO Channel 00/0 : 83[3] -> 99[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557270:1557688 [4] NCCL INFO Channel 08/0 : 84[4] -> 100[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557268:1557686 [2] NCCL INFO Channel 08/0 : 82[2] -> 98[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557272:1557687 [6] NCCL INFO Channel 02/0 : 86[6] -> 102[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557266:1557689 [0] NCCL INFO Channel 10/0 : 96[0] -> 112[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557271:1557690 [5] NCCL INFO Channel 02/0 : 85[5] -> 101[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557273:1557685 [7] NCCL INFO Channel 04/0 : 87[7] -> 103[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557269:1557684 [3] NCCL INFO Channel 02/0 : 83[3] -> 99[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557270:1557688 [4] NCCL INFO Channel 10/0 : 84[4] -> 100[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557267:1557683 [1] NCCL INFO Channel 00/0 : 81[1] -> 97[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557266:1557689 [0] NCCL INFO Channel 12/0 : 96[0] -> 112[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557268:1557686 [2] NCCL INFO Channel 10/0 : 82[2] -> 98[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557272:1557687 [6] NCCL INFO Channel 04/0 : 86[6] -> 102[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557271:1557690 [5] NCCL INFO Channel 04/0 : 85[5] -> 101[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557273:1557685 [7] NCCL INFO Channel 06/0 : 87[7] -> 103[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557269:1557684 [3] NCCL INFO Channel 04/0 : 83[3] -> 99[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557270:1557688 [4] NCCL INFO Channel 12/0 : 84[4] -> 100[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557267:1557683 [1] NCCL INFO Channel 02/0 : 81[1] -> 97[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557266:1557689 [0] NCCL INFO Channel 14/0 : 96[0] -> 112[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557268:1557686 [2] NCCL INFO Channel 12/0 : 82[2] -> 98[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557271:1557690 [5] NCCL INFO Channel 06/0 : 85[5] -> 101[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557273:1557685 [7] NCCL INFO Channel 08/0 : 87[7] -> 103[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557272:1557687 [6] NCCL INFO Channel 08/0 : 86[6] -> 102[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557269:1557684 [3] NCCL INFO Channel 06/0 : 83[3] -> 99[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557270:1557688 [4] NCCL INFO Channel 14/0 : 84[4] -> 100[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557267:1557683 [1] NCCL INFO Channel 04/0 : 81[1] -> 97[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557268:1557686 [2] NCCL INFO Channel 14/0 : 82[2] -> 98[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557271:1557690 [5] NCCL INFO Channel 08/0 : 85[5] -> 101[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557273:1557685 [7] NCCL INFO Channel 10/0 : 87[7] -> 103[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557272:1557687 [6] NCCL INFO Channel 10/0 : 86[6] -> 102[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557269:1557684 [3] NCCL INFO Channel 08/0 : 83[3] -> 99[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557270:1557688 [4] NCCL INFO Channel 00/0 : 100[4] -> 116[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557268:1557686 [2] NCCL INFO Channel 00/0 : 98[2] -> 114[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557267:1557683 [1] NCCL INFO Channel 06/0 : 81[1] -> 97[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557271:1557690 [5] NCCL INFO Channel 10/0 : 85[5] -> 101[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557273:1557685 [7] NCCL INFO Channel 12/0 : 87[7] -> 103[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557272:1557687 [6] NCCL INFO Channel 12/0 : 86[6] -> 102[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557269:1557684 [3] NCCL INFO Channel 10/0 : 83[3] -> 99[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557270:1557688 [4] NCCL INFO Channel 02/0 : 100[4] -> 116[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557268:1557686 [2] NCCL INFO Channel 04/0 : 98[2] -> 114[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557271:1557690 [5] NCCL INFO Channel 12/0 : 85[5] -> 101[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557267:1557683 [1] NCCL INFO Channel 08/0 : 81[1] -> 97[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557273:1557685 [7] NCCL INFO Channel 14/0 : 87[7] -> 103[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557272:1557687 [6] NCCL INFO Channel 14/0 : 86[6] -> 102[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557269:1557684 [3] NCCL INFO Channel 12/0 : 83[3] -> 99[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557270:1557688 [4] NCCL INFO Channel 06/0 : 100[4] -> 116[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557268:1557686 [2] NCCL INFO Channel 06/0 : 98[2] -> 114[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557271:1557690 [5] NCCL INFO Channel 14/0 : 85[5] -> 101[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557267:1557683 [1] NCCL INFO Channel 10/0 : 81[1] -> 97[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557273:1557685 [7] NCCL INFO Channel 00/0 : 103[7] -> 119[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557272:1557687 [6] NCCL INFO Channel 00/0 : 102[6] -> 118[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557269:1557684 [3] NCCL INFO Channel 14/0 : 83[3] -> 99[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557270:1557688 [4] NCCL INFO Channel 08/0 : 100[4] -> 116[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557268:1557686 [2] NCCL INFO Channel 08/0 : 98[2] -> 114[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557271:1557690 [5] NCCL INFO Channel 00/0 : 101[5] -> 117[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557267:1557683 [1] NCCL INFO Channel 12/0 : 81[1] -> 97[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557273:1557685 [7] NCCL INFO Channel 02/0 : 103[7] -> 119[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557272:1557687 [6] NCCL INFO Channel 02/0 : 102[6] -> 118[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557269:1557684 [3] NCCL INFO Channel 00/0 : 99[3] -> 115[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557268:1557686 [2] NCCL INFO Channel 10/0 : 98[2] -> 114[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557270:1557688 [4] NCCL INFO Channel 10/0 : 100[4] -> 116[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557271:1557690 [5] NCCL INFO Channel 02/0 : 101[5] -> 117[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557267:1557683 [1] NCCL INFO Channel 14/0 : 81[1] -> 97[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557273:1557685 [7] NCCL INFO Channel 04/0 : 103[7] -> 119[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557272:1557687 [6] NCCL INFO Channel 04/0 : 102[6] -> 118[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557268:1557686 [2] NCCL INFO Channel 12/0 : 98[2] -> 114[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557269:1557684 [3] NCCL INFO Channel 02/0 : 99[3] -> 115[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557270:1557688 [4] NCCL INFO Channel 12/0 : 100[4] -> 116[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557271:1557690 [5] NCCL INFO Channel 04/0 : 101[5] -> 117[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557273:1557685 [7] NCCL INFO Channel 06/0 : 103[7] -> 119[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557267:1557683 [1] NCCL INFO Channel 00/0 : 97[1] -> 113[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557272:1557687 [6] NCCL INFO Channel 08/0 : 102[6] -> 118[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557268:1557686 [2] NCCL INFO Channel 14/0 : 98[2] -> 114[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557269:1557684 [3] NCCL INFO Channel 04/0 : 99[3] -> 115[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557270:1557688 [4] NCCL INFO Channel 14/0 : 100[4] -> 116[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557271:1557690 [5] NCCL INFO Channel 06/0 : 101[5] -> 117[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557273:1557685 [7] NCCL INFO Channel 08/0 : 103[7] -> 119[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557272:1557687 [6] NCCL INFO Channel 10/0 : 102[6] -> 118[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557267:1557683 [1] NCCL INFO Channel 02/0 : 97[1] -> 113[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557269:1557684 [3] NCCL INFO Channel 06/0 : 99[3] -> 115[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557271:1557690 [5] NCCL INFO Channel 08/0 : 101[5] -> 117[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557273:1557685 [7] NCCL INFO Channel 10/0 : 103[7] -> 119[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557272:1557687 [6] NCCL INFO Channel 12/0 : 102[6] -> 118[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557267:1557683 [1] NCCL INFO Channel 04/0 : 97[1] -> 113[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557269:1557684 [3] NCCL INFO Channel 08/0 : 99[3] -> 115[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557273:1557685 [7] NCCL INFO Channel 12/0 : 103[7] -> 119[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557272:1557687 [6] NCCL INFO Channel 14/0 : 102[6] -> 118[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557269:1557684 [3] NCCL INFO Channel 10/0 : 99[3] -> 115[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557273:1557685 [7] NCCL INFO Channel 14/0 : 103[7] -> 119[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557267:1557683 [1] NCCL INFO Channel 06/0 : 97[1] -> 113[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557269:1557684 [3] NCCL INFO Channel 12/0 : 99[3] -> 115[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557267:1557683 [1] NCCL INFO Channel 08/0 : 97[1] -> 113[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557271:1557690 [5] NCCL INFO Channel 10/0 : 101[5] -> 117[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557269:1557684 [3] NCCL INFO Channel 14/0 : 99[3] -> 115[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557267:1557683 [1] NCCL INFO Channel 10/0 : 97[1] -> 113[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557271:1557690 [5] NCCL INFO Channel 12/0 : 101[5] -> 117[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557267:1557683 [1] NCCL INFO Channel 12/0 : 97[1] -> 113[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557271:1557690 [5] NCCL INFO Channel 14/0 : 101[5] -> 117[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557267:1557683 [1] NCCL INFO Channel 14/0 : 97[1] -> 113[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96564:97009 [1] NCCL INFO Channel 01/0 : 33[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96564:97009 [1] NCCL INFO Channel 03/0 : 33[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96568:97007 [5] NCCL INFO Channel 01/0 : 37[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96564:97009 [1] NCCL INFO Channel 05/0 : 33[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96568:97007 [5] NCCL INFO Channel 03/0 : 37[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96564:97009 [1] NCCL INFO Channel 07/0 : 33[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96566:97010 [3] NCCL INFO Channel 01/0 : 35[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96570:97005 [7] NCCL INFO Channel 01/0 : 39[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96568:97007 [5] NCCL INFO Channel 05/0 : 37[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96564:97009 [1] NCCL INFO Channel 11/0 : 33[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96570:97005 [7] NCCL INFO Channel 03/0 : 39[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96566:97010 [3] NCCL INFO Channel 03/0 : 35[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96568:97007 [5] NCCL INFO Channel 07/0 : 37[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630206:1630647 [1] NCCL INFO Channel 01/0 : 65[1] -> 73[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96564:97009 [1] NCCL INFO Channel 13/0 : 33[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96567:97008 [4] NCCL INFO Channel 01/0 : 36[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630212:1630652 [7] NCCL INFO Channel 01/0 : 71[7] -> 79[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96570:97005 [7] NCCL INFO Channel 05/0 : 39[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630208:1630650 [3] NCCL INFO Channel 01/0 : 67[3] -> 75[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96568:97007 [5] NCCL INFO Channel 09/0 : 37[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630210:1630653 [5] NCCL INFO Channel 01/0 : 69[5] -> 77[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630206:1630647 [1] NCCL INFO Channel 03/0 : 65[1] -> 73[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96563:97003 [0] NCCL INFO Channel 01/0 : 32[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96569:97006 [6] NCCL INFO Channel 01/0 : 38[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630212:1630652 [7] NCCL INFO Channel 03/0 : 71[7] -> 79[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96567:97008 [4] NCCL INFO Channel 03/0 : 36[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630208:1630650 [3] NCCL INFO Channel 03/0 : 67[3] -> 75[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96566:97010 [3] NCCL INFO Channel 05/0 : 35[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96564:97009 [1] NCCL INFO Channel 15/0 : 33[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96570:97005 [7] NCCL INFO Channel 07/0 : 39[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96568:97007 [5] NCCL INFO Channel 11/0 : 37[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630210:1630653 [5] NCCL INFO Channel 03/0 : 69[5] -> 77[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630206:1630647 [1] NCCL INFO Channel 05/0 : 65[1] -> 73[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630212:1630652 [7] NCCL INFO Channel 05/0 : 71[7] -> 79[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96563:97003 [0] NCCL INFO Channel 03/0 : 32[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630208:1630650 [3] NCCL INFO Channel 05/0 : 67[3] -> 75[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96569:97006 [6] NCCL INFO Channel 03/0 : 38[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630210:1630653 [5] NCCL INFO Channel 05/0 : 69[5] -> 77[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96566:97010 [3] NCCL INFO Channel 07/0 : 35[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96567:97008 [4] NCCL INFO Channel 05/0 : 36[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96570:97005 [7] NCCL INFO Channel 09/0 : 39[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630206:1630647 [1] NCCL INFO Channel 07/0 : 65[1] -> 73[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96568:97007 [5] NCCL INFO Channel 15/0 : 37[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630212:1630652 [7] NCCL INFO Channel 07/0 : 71[7] -> 79[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96565:97004 [2] NCCL INFO Channel 01/0 : 34[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96563:97003 [0] NCCL INFO Channel 05/0 : 32[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96569:97006 [6] NCCL INFO Channel 05/0 : 38[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630208:1630650 [3] NCCL INFO Channel 07/0 : 67[3] -> 75[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630205:1630646 [0] NCCL INFO Channel 01/0 : 64[0] -> 72[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630211:1630649 [6] NCCL INFO Channel 01/0 : 70[6] -> 78[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630206:1630647 [1] NCCL INFO Channel 11/0 : 65[1] -> 73[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630210:1630653 [5] NCCL INFO Channel 07/0 : 69[5] -> 77[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96570:97005 [7] NCCL INFO Channel 11/0 : 39[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630212:1630652 [7] NCCL INFO Channel 09/0 : 71[7] -> 79[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96567:97008 [4] NCCL INFO Channel 07/0 : 36[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630208:1630650 [3] NCCL INFO Channel 09/0 : 67[3] -> 75[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96566:97010 [3] NCCL INFO Channel 09/0 : 35[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96563:97003 [0] NCCL INFO Channel 07/0 : 32[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96565:97004 [2] NCCL INFO Channel 03/0 : 34[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96569:97006 [6] NCCL INFO Channel 07/0 : 38[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96564:97009 [1] NCCL INFO Channel 00/0 : 17[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96570:97005 [7] NCCL INFO Channel 13/0 : 39[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96567:97008 [4] NCCL INFO Channel 09/0 : 36[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630205:1630646 [0] NCCL INFO Channel 03/0 : 64[0] -> 72[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630209:1630651 [4] NCCL INFO Channel 01/0 : 68[4] -> 76[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630211:1630649 [6] NCCL INFO Channel 03/0 : 70[6] -> 78[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630206:1630647 [1] NCCL INFO Channel 13/0 : 65[1] -> 73[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630210:1630653 [5] NCCL INFO Channel 09/0 : 69[5] -> 77[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630212:1630652 [7] NCCL INFO Channel 11/0 : 71[7] -> 79[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630208:1630650 [3] NCCL INFO Channel 13/0 : 67[3] -> 75[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630205:1630646 [0] NCCL INFO Channel 05/0 : 64[0] -> 72[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96566:97010 [3] NCCL INFO Channel 13/0 : 35[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96563:97003 [0] NCCL INFO Channel 09/0 : 32[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96569:97006 [6] NCCL INFO Channel 09/0 : 38[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96565:97004 [2] NCCL INFO Channel 05/0 : 34[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96564:97009 [1] NCCL INFO Channel 02/0 : 17[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96567:97008 [4] NCCL INFO Channel 11/0 : 36[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630209:1630651 [4] NCCL INFO Channel 03/0 : 68[4] -> 76[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96568:97007 [5] NCCL INFO Channel 00/0 : 21[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630211:1630649 [6] NCCL INFO Channel 05/0 : 70[6] -> 78[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96563:97003 [0] NCCL INFO Channel 11/0 : 32[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96566:97010 [3] NCCL INFO Channel 15/0 : 35[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96569:97006 [6] NCCL INFO Channel 11/0 : 38[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630206:1630647 [1] NCCL INFO Channel 15/0 : 65[1] -> 73[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96565:97004 [2] NCCL INFO Channel 07/0 : 34[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630212:1630652 [7] NCCL INFO Channel 13/0 : 71[7] -> 79[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630210:1630653 [5] NCCL INFO Channel 11/0 : 69[5] -> 77[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630208:1630650 [3] NCCL INFO Channel 15/0 : 67[3] -> 75[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96564:97009 [1] NCCL INFO Channel 04/0 : 17[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630205:1630646 [0] NCCL INFO Channel 07/0 : 64[0] -> 72[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630209:1630651 [4] NCCL INFO Channel 05/0 : 68[4] -> 76[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96567:97008 [4] NCCL INFO Channel 13/0 : 36[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96568:97007 [5] NCCL INFO Channel 02/0 : 21[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630211:1630649 [6] NCCL INFO Channel 07/0 : 70[6] -> 78[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630206:1630647 [1] NCCL INFO Channel 00/0 : 73[1] -> 81[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630207:1630648 [2] NCCL INFO Channel 01/0 : 66[2] -> 74[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96563:97003 [0] NCCL INFO Channel 13/0 : 32[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96569:97006 [6] NCCL INFO Channel 13/0 : 38[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96570:97005 [7] NCCL INFO Channel 00/0 : 23[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96565:97004 [2] NCCL INFO Channel 09/0 : 34[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96567:97008 [4] NCCL INFO Channel 15/0 : 36[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96568:97007 [5] NCCL INFO Channel 04/0 : 21[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630212:1630652 [7] NCCL INFO Channel 00/0 : 79[7] -> 87[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96564:97009 [1] NCCL INFO Channel 06/0 : 17[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630210:1630653 [5] NCCL INFO Channel 15/0 : 69[5] -> 77[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630208:1630650 [3] NCCL INFO Channel 00/0 : 75[3] -> 83[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630205:1630646 [0] NCCL INFO Channel 09/0 : 64[0] -> 72[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630209:1630651 [4] NCCL INFO Channel 07/0 : 68[4] -> 76[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96569:97006 [6] NCCL INFO Channel 15/0 : 38[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96563:97003 [0] NCCL INFO Channel 15/0 : 32[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96570:97005 [7] NCCL INFO Channel 02/0 : 23[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630211:1630649 [6] NCCL INFO Channel 09/0 : 70[6] -> 78[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630206:1630647 [1] NCCL INFO Channel 02/0 : 73[1] -> 81[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630207:1630648 [2] NCCL INFO Channel 03/0 : 66[2] -> 74[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96566:97010 [3] NCCL INFO Channel 00/0 : 19[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96565:97004 [2] NCCL INFO Channel 11/0 : 34[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630212:1630652 [7] NCCL INFO Channel 01/0 : 79[7] -> 87[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96568:97007 [5] NCCL INFO Channel 06/0 : 21[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96564:97009 [1] NCCL INFO Channel 08/0 : 17[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96570:97005 [7] NCCL INFO Channel 04/0 : 23[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630210:1630653 [5] NCCL INFO Channel 00/0 : 77[5] -> 85[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630208:1630650 [3] NCCL INFO Channel 01/0 : 75[3] -> 83[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630209:1630651 [4] NCCL INFO Channel 09/0 : 68[4] -> 76[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96565:97004 [2] NCCL INFO Channel 13/0 : 34[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96566:97010 [3] NCCL INFO Channel 02/0 : 19[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96568:97007 [5] NCCL INFO Channel 08/0 : 21[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96564:97009 [1] NCCL INFO Channel 10/0 : 17[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96570:97005 [7] NCCL INFO Channel 06/0 : 23[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96567:97008 [4] NCCL INFO Channel 00/0 : 20[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630211:1630649 [6] NCCL INFO Channel 11/0 : 70[6] -> 78[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630205:1630646 [0] NCCL INFO Channel 11/0 : 64[0] -> 72[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96565:97004 [2] NCCL INFO Channel 15/0 : 34[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630206:1630647 [1] NCCL INFO Channel 03/0 : 73[1] -> 81[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630212:1630652 [7] NCCL INFO Channel 02/0 : 79[7] -> 87[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630207:1630648 [2] NCCL INFO Channel 05/0 : 66[2] -> 74[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630210:1630653 [5] NCCL INFO Channel 01/0 : 77[5] -> 85[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96568:97007 [5] NCCL INFO Channel 10/0 : 21[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96566:97010 [3] NCCL INFO Channel 04/0 : 19[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630209:1630651 [4] NCCL INFO Channel 11/0 : 68[4] -> 76[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96569:97006 [6] NCCL INFO Channel 00/0 : 22[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96570:97005 [7] NCCL INFO Channel 08/0 : 23[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630211:1630649 [6] NCCL INFO Channel 13/0 : 70[6] -> 78[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96564:97009 [1] NCCL INFO Channel 12/0 : 17[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96567:97008 [4] NCCL INFO Channel 02/0 : 20[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630205:1630646 [0] NCCL INFO Channel 13/0 : 64[0] -> 72[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630212:1630652 [7] NCCL INFO Channel 03/0 : 79[7] -> 87[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630206:1630647 [1] NCCL INFO Channel 04/0 : 73[1] -> 81[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96563:97003 [0] NCCL INFO Channel 02/0 : 16[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630208:1630650 [3] NCCL INFO Channel 02/0 : 75[3] -> 83[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630210:1630653 [5] NCCL INFO Channel 02/0 : 77[5] -> 85[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96568:97007 [5] NCCL INFO Channel 12/0 : 21[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630207:1630648 [2] NCCL INFO Channel 07/0 : 66[2] -> 74[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96566:97010 [3] NCCL INFO Channel 06/0 : 19[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630209:1630651 [4] NCCL INFO Channel 13/0 : 68[4] -> 76[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630211:1630649 [6] NCCL INFO Channel 15/0 : 70[6] -> 78[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96570:97005 [7] NCCL INFO Channel 10/0 : 23[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96567:97008 [4] NCCL INFO Channel 06/0 : 20[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96569:97006 [6] NCCL INFO Channel 02/0 : 22[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96564:97009 [1] NCCL INFO Channel 14/0 : 17[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96563:97003 [0] NCCL INFO Channel 04/0 : 16[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630205:1630646 [0] NCCL INFO Channel 15/0 : 64[0] -> 72[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96565:97004 [2] NCCL INFO Channel 00/0 : 18[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630212:1630652 [7] NCCL INFO Channel 04/0 : 79[7] -> 87[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96568:97007 [5] NCCL INFO Channel 14/0 : 21[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630210:1630653 [5] NCCL INFO Channel 03/0 : 77[5] -> 85[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630206:1630647 [1] NCCL INFO Channel 05/0 : 73[1] -> 81[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96566:97010 [3] NCCL INFO Channel 08/0 : 19[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630208:1630650 [3] NCCL INFO Channel 04/0 : 75[3] -> 83[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96570:97005 [7] NCCL INFO Channel 12/0 : 23[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630211:1630649 [6] NCCL INFO Channel 00/0 : 78[6] -> 86[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630207:1630648 [2] NCCL INFO Channel 09/0 : 66[2] -> 74[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630209:1630651 [4] NCCL INFO Channel 15/0 : 68[4] -> 76[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630205:1630646 [0] NCCL INFO Channel 01/0 : 72[0] -> 80[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630212:1630652 [7] NCCL INFO Channel 05/0 : 79[7] -> 87[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96567:97008 [4] NCCL INFO Channel 08/0 : 20[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96569:97006 [6] NCCL INFO Channel 04/0 : 22[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96563:97003 [0] NCCL INFO Channel 06/0 : 16[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96568:97007 [5] NCCL INFO Channel 00/0 : 37[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630210:1630653 [5] NCCL INFO Channel 04/0 : 77[5] -> 85[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96564:97009 [1] NCCL INFO Channel 00/0 : 33[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630206:1630647 [1] NCCL INFO Channel 06/0 : 73[1] -> 81[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96565:97004 [2] NCCL INFO Channel 04/0 : 18[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96570:97005 [7] NCCL INFO Channel 14/0 : 23[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630208:1630650 [3] NCCL INFO Channel 05/0 : 75[3] -> 83[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96566:97010 [3] NCCL INFO Channel 10/0 : 19[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630211:1630649 [6] NCCL INFO Channel 01/0 : 78[6] -> 86[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630209:1630651 [4] NCCL INFO Channel 00/0 : 76[4] -> 84[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630207:1630648 [2] NCCL INFO Channel 11/0 : 66[2] -> 74[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96567:97008 [4] NCCL INFO Channel 10/0 : 20[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630212:1630652 [7] NCCL INFO Channel 06/0 : 79[7] -> 87[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630210:1630653 [5] NCCL INFO Channel 06/0 : 77[5] -> 85[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630205:1630646 [0] NCCL INFO Channel 02/0 : 72[0] -> 80[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96569:97006 [6] NCCL INFO Channel 08/0 : 22[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630206:1630647 [1] NCCL INFO Channel 07/0 : 73[1] -> 81[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630208:1630650 [3] NCCL INFO Channel 06/0 : 75[3] -> 83[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96563:97003 [0] NCCL INFO Channel 08/0 : 16[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96568:97007 [5] NCCL INFO Channel 02/0 : 37[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96570:97005 [7] NCCL INFO Channel 00/0 : 39[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96564:97009 [1] NCCL INFO Channel 02/0 : 33[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96567:97008 [4] NCCL INFO Channel 12/0 : 20[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630209:1630651 [4] NCCL INFO Channel 01/0 : 76[4] -> 84[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96566:97010 [3] NCCL INFO Channel 12/0 : 19[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630207:1630648 [2] NCCL INFO Channel 13/0 : 66[2] -> 74[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630212:1630652 [7] NCCL INFO Channel 08/0 : 79[7] -> 87[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630210:1630653 [5] NCCL INFO Channel 07/0 : 77[5] -> 85[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96565:97004 [2] NCCL INFO Channel 06/0 : 18[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96569:97006 [6] NCCL INFO Channel 10/0 : 22[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630211:1630649 [6] NCCL INFO Channel 02/0 : 78[6] -> 86[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630206:1630647 [1] NCCL INFO Channel 08/0 : 73[1] -> 81[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630205:1630646 [0] NCCL INFO Channel 03/0 : 72[0] -> 80[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630208:1630650 [3] NCCL INFO Channel 07/0 : 75[3] -> 83[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630209:1630651 [4] NCCL INFO Channel 02/0 : 76[4] -> 84[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630207:1630648 [2] NCCL INFO Channel 15/0 : 66[2] -> 74[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96563:97003 [0] NCCL INFO Channel 10/0 : 16[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96568:97007 [5] NCCL INFO Channel 04/0 : 37[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630212:1630652 [7] NCCL INFO Channel 09/0 : 79[7] -> 87[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630210:1630653 [5] NCCL INFO Channel 08/0 : 77[5] -> 85[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630211:1630649 [6] NCCL INFO Channel 03/0 : 78[6] -> 86[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96570:97005 [7] NCCL INFO Channel 02/0 : 39[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96567:97008 [4] NCCL INFO Channel 14/0 : 20[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96564:97009 [1] NCCL INFO Channel 04/0 : 33[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96569:97006 [6] NCCL INFO Channel 12/0 : 22[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96566:97010 [3] NCCL INFO Channel 14/0 : 19[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96565:97004 [2] NCCL INFO Channel 08/0 : 18[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630205:1630646 [0] NCCL INFO Channel 04/0 : 72[0] -> 80[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630209:1630651 [4] NCCL INFO Channel 03/0 : 76[4] -> 84[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630208:1630650 [3] NCCL INFO Channel 08/0 : 75[3] -> 83[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630206:1630647 [1] NCCL INFO Channel 10/0 : 73[1] -> 81[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630207:1630648 [2] NCCL INFO Channel 00/0 : 74[2] -> 82[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96563:97003 [0] NCCL INFO Channel 12/0 : 16[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630212:1630652 [7] NCCL INFO Channel 10/0 : 79[7] -> 87[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96568:97007 [5] NCCL INFO Channel 06/0 : 37[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96570:97005 [7] NCCL INFO Channel 04/0 : 39[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630210:1630653 [5] NCCL INFO Channel 09/0 : 77[5] -> 85[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630211:1630649 [6] NCCL INFO Channel 04/0 : 78[6] -> 86[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96567:97008 [4] NCCL INFO Channel 00/0 : 36[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630208:1630650 [3] NCCL INFO Channel 09/0 : 75[3] -> 83[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96564:97009 [1] NCCL INFO Channel 06/0 : 33[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96569:97006 [6] NCCL INFO Channel 14/0 : 22[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96563:97003 [0] NCCL INFO Channel 14/0 : 16[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96566:97010 [3] NCCL INFO Channel 00/0 : 35[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96565:97004 [2] NCCL INFO Channel 10/0 : 18[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96570:97005 [7] NCCL INFO Channel 06/0 : 39[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96568:97007 [5] NCCL INFO Channel 08/0 : 37[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630209:1630651 [4] NCCL INFO Channel 05/0 : 76[4] -> 84[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630205:1630646 [0] NCCL INFO Channel 05/0 : 72[0] -> 80[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630206:1630647 [1] NCCL INFO Channel 11/0 : 73[1] -> 81[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630207:1630648 [2] NCCL INFO Channel 01/0 : 74[2] -> 82[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630212:1630652 [7] NCCL INFO Channel 11/0 : 79[7] -> 87[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630210:1630653 [5] NCCL INFO Channel 10/0 : 77[5] -> 85[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96567:97008 [4] NCCL INFO Channel 02/0 : 36[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630211:1630649 [6] NCCL INFO Channel 05/0 : 78[6] -> 86[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96564:97009 [1] NCCL INFO Channel 08/0 : 33[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630208:1630650 [3] NCCL INFO Channel 10/0 : 75[3] -> 83[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96569:97006 [6] NCCL INFO Channel 00/0 : 38[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96570:97005 [7] NCCL INFO Channel 08/0 : 39[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630209:1630651 [4] NCCL INFO Channel 06/0 : 76[4] -> 84[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96563:97003 [0] NCCL INFO Channel 02/0 : 32[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630205:1630646 [0] NCCL INFO Channel 06/0 : 72[0] -> 80[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96568:97007 [5] NCCL INFO Channel 10/0 : 37[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630206:1630647 [1] NCCL INFO Channel 12/0 : 73[1] -> 81[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630207:1630648 [2] NCCL INFO Channel 03/0 : 74[2] -> 82[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96567:97008 [4] NCCL INFO Channel 06/0 : 36[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630210:1630653 [5] NCCL INFO Channel 11/0 : 77[5] -> 85[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96565:97004 [2] NCCL INFO Channel 12/0 : 18[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96566:97010 [3] NCCL INFO Channel 02/0 : 35[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630211:1630649 [6] NCCL INFO Channel 07/0 : 78[6] -> 86[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630212:1630652 [7] NCCL INFO Channel 12/0 : 79[7] -> 87[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96569:97006 [6] NCCL INFO Channel 02/0 : 38[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96564:97009 [1] NCCL INFO Channel 10/0 : 33[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96570:97005 [7] NCCL INFO Channel 10/0 : 39[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630209:1630651 [4] NCCL INFO Channel 07/0 : 76[4] -> 84[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630208:1630650 [3] NCCL INFO Channel 12/0 : 75[3] -> 83[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630205:1630646 [0] NCCL INFO Channel 07/0 : 72[0] -> 80[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630206:1630647 [1] NCCL INFO Channel 13/0 : 73[1] -> 81[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630207:1630648 [2] NCCL INFO Channel 04/0 : 74[2] -> 82[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630210:1630653 [5] NCCL INFO Channel 12/0 : 77[5] -> 85[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96563:97003 [0] NCCL INFO Channel 04/0 : 32[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96568:97007 [5] NCCL INFO Channel 12/0 : 37[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96567:97008 [4] NCCL INFO Channel 08/0 : 36[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96565:97004 [2] NCCL INFO Channel 14/0 : 18[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96569:97006 [6] NCCL INFO Channel 04/0 : 38[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96564:97009 [1] NCCL INFO Channel 12/0 : 33[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96566:97010 [3] NCCL INFO Channel 04/0 : 35[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96570:97005 [7] NCCL INFO Channel 12/0 : 39[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630211:1630649 [6] NCCL INFO Channel 08/0 : 78[6] -> 86[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630212:1630652 [7] NCCL INFO Channel 13/0 : 79[7] -> 87[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630209:1630651 [4] NCCL INFO Channel 08/0 : 76[4] -> 84[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630208:1630650 [3] NCCL INFO Channel 13/0 : 75[3] -> 83[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630205:1630646 [0] NCCL INFO Channel 09/0 : 72[0] -> 80[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630206:1630647 [1] NCCL INFO Channel 14/0 : 73[1] -> 81[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96563:97003 [0] NCCL INFO Channel 06/0 : 32[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96568:97007 [5] NCCL INFO Channel 14/0 : 37[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630207:1630648 [2] NCCL INFO Channel 05/0 : 74[2] -> 82[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630210:1630653 [5] NCCL INFO Channel 14/0 : 77[5] -> 85[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96567:97008 [4] NCCL INFO Channel 10/0 : 36[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630211:1630649 [6] NCCL INFO Channel 09/0 : 78[6] -> 86[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630212:1630652 [7] NCCL INFO Channel 14/0 : 79[7] -> 87[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96569:97006 [6] NCCL INFO Channel 08/0 : 38[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630209:1630651 [4] NCCL INFO Channel 09/0 : 76[4] -> 84[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96565:97004 [2] NCCL INFO Channel 00/0 : 34[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96570:97005 [7] NCCL INFO Channel 14/0 : 39[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630206:1630647 [1] NCCL INFO Channel 15/0 : 73[1] -> 81[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630205:1630646 [0] NCCL INFO Channel 10/0 : 72[0] -> 80[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630208:1630650 [3] NCCL INFO Channel 14/0 : 75[3] -> 83[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96564:97009 [1] NCCL INFO Channel 14/0 : 33[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96566:97010 [3] NCCL INFO Channel 06/0 : 35[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630207:1630648 [2] NCCL INFO Channel 06/0 : 74[2] -> 82[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630210:1630653 [5] NCCL INFO Channel 15/0 : 77[5] -> 85[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96563:97003 [0] NCCL INFO Channel 08/0 : 32[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96567:97008 [4] NCCL INFO Channel 12/0 : 36[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630211:1630649 [6] NCCL INFO Channel 10/0 : 78[6] -> 86[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630209:1630651 [4] NCCL INFO Channel 10/0 : 76[4] -> 84[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630208:1630650 [3] NCCL INFO Channel 15/0 : 75[3] -> 83[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630205:1630646 [0] NCCL INFO Channel 11/0 : 72[0] -> 80[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630207:1630648 [2] NCCL INFO Channel 07/0 : 74[2] -> 82[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630211:1630649 [6] NCCL INFO Channel 11/0 : 78[6] -> 86[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96569:97006 [6] NCCL INFO Channel 10/0 : 38[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96565:97004 [2] NCCL INFO Channel 04/0 : 34[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96563:97003 [0] NCCL INFO Channel 10/0 : 32[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96567:97008 [4] NCCL INFO Channel 14/0 : 36[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96566:97010 [3] NCCL INFO Channel 08/0 : 35[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630209:1630651 [4] NCCL INFO Channel 11/0 : 76[4] -> 84[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96569:97006 [6] NCCL INFO Channel 12/0 : 38[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630205:1630646 [0] NCCL INFO Channel 12/0 : 72[0] -> 80[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630211:1630649 [6] NCCL INFO Channel 12/0 : 78[6] -> 86[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630207:1630648 [2] NCCL INFO Channel 08/0 : 74[2] -> 82[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630209:1630651 [4] NCCL INFO Channel 13/0 : 76[4] -> 84[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96565:97004 [2] NCCL INFO Channel 06/0 : 34[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96563:97003 [0] NCCL INFO Channel 12/0 : 32[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96569:97006 [6] NCCL INFO Channel 14/0 : 38[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96566:97010 [3] NCCL INFO Channel 10/0 : 35[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96565:97004 [2] NCCL INFO Channel 08/0 : 34[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630205:1630646 [0] NCCL INFO Channel 13/0 : 72[0] -> 80[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96563:97003 [0] NCCL INFO Channel 14/0 : 32[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630211:1630649 [6] NCCL INFO Channel 13/0 : 78[6] -> 86[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630209:1630651 [4] NCCL INFO Channel 14/0 : 76[4] -> 84[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630207:1630648 [2] NCCL INFO Channel 09/0 : 74[2] -> 82[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630212:1630652 [7] NCCL INFO Channel 01/0 : 79[7] -> 95[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630205:1630646 [0] NCCL INFO Channel 14/0 : 72[0] -> 80[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630211:1630649 [6] NCCL INFO Channel 15/0 : 78[6] -> 86[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630210:1630653 [5] NCCL INFO Channel 01/0 : 77[5] -> 93[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630206:1630647 [1] NCCL INFO Channel 01/0 : 73[1] -> 89[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630209:1630651 [4] NCCL INFO Channel 15/0 : 76[4] -> 84[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630207:1630648 [2] NCCL INFO Channel 11/0 : 74[2] -> 82[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630212:1630652 [7] NCCL INFO Channel 03/0 : 79[7] -> 95[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630205:1630646 [0] NCCL INFO Channel 15/0 : 72[0] -> 80[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630208:1630650 [3] NCCL INFO Channel 01/0 : 75[3] -> 91[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630210:1630653 [5] NCCL INFO Channel 03/0 : 77[5] -> 93[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630206:1630647 [1] NCCL INFO Channel 03/0 : 73[1] -> 89[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630207:1630648 [2] NCCL INFO Channel 12/0 : 74[2] -> 82[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630212:1630652 [7] NCCL INFO Channel 05/0 : 79[7] -> 95[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630208:1630650 [3] NCCL INFO Channel 03/0 : 75[3] -> 91[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630210:1630653 [5] NCCL INFO Channel 05/0 : 77[5] -> 93[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630206:1630647 [1] NCCL INFO Channel 05/0 : 73[1] -> 89[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630207:1630648 [2] NCCL INFO Channel 13/0 : 74[2] -> 82[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630212:1630652 [7] NCCL INFO Channel 07/0 : 79[7] -> 95[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630208:1630650 [3] NCCL INFO Channel 05/0 : 75[3] -> 91[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630210:1630653 [5] NCCL INFO Channel 07/0 : 77[5] -> 93[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630206:1630647 [1] NCCL INFO Channel 07/0 : 73[1] -> 89[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630207:1630648 [2] NCCL INFO Channel 14/0 : 74[2] -> 82[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630212:1630652 [7] NCCL INFO Channel 09/0 : 79[7] -> 95[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630210:1630653 [5] NCCL INFO Channel 09/0 : 77[5] -> 93[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630206:1630647 [1] NCCL INFO Channel 11/0 : 73[1] -> 89[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630208:1630650 [3] NCCL INFO Channel 07/0 : 75[3] -> 91[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630211:1630649 [6] NCCL INFO Channel 01/0 : 78[6] -> 94[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630207:1630648 [2] NCCL INFO Channel 15/0 : 74[2] -> 82[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630212:1630652 [7] NCCL INFO Channel 11/0 : 79[7] -> 95[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630210:1630653 [5] NCCL INFO Channel 11/0 : 77[5] -> 93[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630206:1630647 [1] NCCL INFO Channel 13/0 : 73[1] -> 89[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630209:1630651 [4] NCCL INFO Channel 01/0 : 76[4] -> 92[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630211:1630649 [6] NCCL INFO Channel 03/0 : 78[6] -> 94[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630205:1630646 [0] NCCL INFO Channel 01/0 : 72[0] -> 88[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630212:1630652 [7] NCCL INFO Channel 13/0 : 79[7] -> 95[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630210:1630653 [5] NCCL INFO Channel 15/0 : 77[5] -> 93[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630208:1630650 [3] NCCL INFO Channel 09/0 : 75[3] -> 91[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630206:1630647 [1] NCCL INFO Channel 15/0 : 73[1] -> 89[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630209:1630651 [4] NCCL INFO Channel 03/0 : 76[4] -> 92[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630211:1630649 [6] NCCL INFO Channel 05/0 : 78[6] -> 94[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630208:1630650 [3] NCCL INFO Channel 13/0 : 75[3] -> 91[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630205:1630646 [0] NCCL INFO Channel 03/0 : 72[0] -> 88[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630209:1630651 [4] NCCL INFO Channel 05/0 : 76[4] -> 92[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630211:1630649 [6] NCCL INFO Channel 07/0 : 78[6] -> 94[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630208:1630650 [3] NCCL INFO Channel 15/0 : 75[3] -> 91[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630205:1630646 [0] NCCL INFO Channel 05/0 : 72[0] -> 88[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630209:1630651 [4] NCCL INFO Channel 07/0 : 76[4] -> 92[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630211:1630649 [6] NCCL INFO Channel 09/0 : 78[6] -> 94[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630207:1630648 [2] NCCL INFO Channel 01/0 : 74[2] -> 90[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630205:1630646 [0] NCCL INFO Channel 07/0 : 72[0] -> 88[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630209:1630651 [4] NCCL INFO Channel 09/0 : 76[4] -> 92[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96890:97313 [2] NCCL INFO Channel 12/0 : 18[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96891:97311 [3] NCCL INFO Channel 10/0 : 19[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96895:97307 [7] NCCL INFO Channel 08/0 : 23[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96888:97306 [0] NCCL INFO Channel 14/0 : 16[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96893:97310 [5] NCCL INFO Channel 08/0 : 21[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114517:114935 [5] NCCL INFO Channel 15/0 : 69[5] -> 77[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96889:97309 [1] NCCL INFO Channel 06/0 : 17[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630211:1630649 [6] NCCL INFO Channel 11/0 : 78[6] -> 94[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114513:114933 [1] NCCL INFO Channel 13/0 : 65[1] -> 73[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96890:97313 [2] NCCL INFO Channel 14/0 : 18[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96891:97311 [3] NCCL INFO Channel 12/0 : 19[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630207:1630648 [2] NCCL INFO Channel 03/0 : 74[2] -> 90[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114515:114936 [3] NCCL INFO Channel 05/0 : 67[3] -> 75[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96895:97307 [7] NCCL INFO Channel 10/0 : 23[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630209:1630651 [4] NCCL INFO Channel 11/0 : 76[4] -> 92[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630205:1630646 [0] NCCL INFO Channel 09/0 : 72[0] -> 88[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114514:114932 [2] NCCL INFO Channel 07/0 : 66[2] -> 74[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114518:114937 [6] NCCL INFO Channel 13/0 : 70[6] -> 78[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114512:114930 [0] NCCL INFO Channel 09/0 : 64[0] -> 72[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96893:97310 [5] NCCL INFO Channel 10/0 : 21[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114516:114934 [4] NCCL INFO Channel 05/0 : 68[4] -> 76[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96889:97309 [1] NCCL INFO Channel 08/0 : 17[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114513:114933 [1] NCCL INFO Channel 15/0 : 65[1] -> 73[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96891:97311 [3] NCCL INFO Channel 14/0 : 19[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114515:114936 [3] NCCL INFO Channel 07/0 : 67[3] -> 75[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96895:97307 [7] NCCL INFO Channel 12/0 : 23[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114514:114932 [2] NCCL INFO Channel 09/0 : 66[2] -> 74[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114518:114937 [6] NCCL INFO Channel 15/0 : 70[6] -> 78[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96889:97309 [1] NCCL INFO Channel 10/0 : 17[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96893:97310 [5] NCCL INFO Channel 12/0 : 21[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630211:1630649 [6] NCCL INFO Channel 13/0 : 78[6] -> 94[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114512:114930 [0] NCCL INFO Channel 11/0 : 64[0] -> 72[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96895:97307 [7] NCCL INFO Channel 14/0 : 23[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630207:1630648 [2] NCCL INFO Channel 05/0 : 74[2] -> 90[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114516:114934 [4] NCCL INFO Channel 07/0 : 68[4] -> 76[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114515:114936 [3] NCCL INFO Channel 09/0 : 67[3] -> 75[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96889:97309 [1] NCCL INFO Channel 12/0 : 17[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96893:97310 [5] NCCL INFO Channel 14/0 : 21[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630209:1630651 [4] NCCL INFO Channel 13/0 : 76[4] -> 92[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114514:114932 [2] NCCL INFO Channel 11/0 : 66[2] -> 74[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114512:114930 [0] NCCL INFO Channel 13/0 : 64[0] -> 72[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96889:97309 [1] NCCL INFO Channel 14/0 : 17[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114515:114936 [3] NCCL INFO Channel 13/0 : 67[3] -> 75[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96893:97310 [5] NCCL INFO Channel 00/0 : 37[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114514:114932 [2] NCCL INFO Channel 13/0 : 66[2] -> 74[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96889:97309 [1] NCCL INFO Channel 00/0 : 33[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96895:97307 [7] NCCL INFO Channel 00/0 : 39[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114512:114930 [0] NCCL INFO Channel 15/0 : 64[0] -> 72[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96893:97310 [5] NCCL INFO Channel 02/0 : 37[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114515:114936 [3] NCCL INFO Channel 15/0 : 67[3] -> 75[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96889:97309 [1] NCCL INFO Channel 02/0 : 33[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630205:1630646 [0] NCCL INFO Channel 11/0 : 72[0] -> 88[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630211:1630649 [6] NCCL INFO Channel 15/0 : 78[6] -> 94[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630207:1630648 [2] NCCL INFO Channel 07/0 : 74[2] -> 90[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114516:114934 [4] NCCL INFO Channel 09/0 : 68[4] -> 76[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96895:97307 [7] NCCL INFO Channel 02/0 : 39[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96893:97310 [5] NCCL INFO Channel 04/0 : 37[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630209:1630651 [4] NCCL INFO Channel 15/0 : 76[4] -> 92[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114514:114932 [2] NCCL INFO Channel 15/0 : 66[2] -> 74[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96892:97308 [4] NCCL INFO Channel 00/0 : 36[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114516:114934 [4] NCCL INFO Channel 11/0 : 68[4] -> 76[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96889:97309 [1] NCCL INFO Channel 04/0 : 33[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114516:114934 [4] NCCL INFO Channel 13/0 : 68[4] -> 76[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96895:97307 [7] NCCL INFO Channel 04/0 : 39[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114516:114934 [4] NCCL INFO Channel 15/0 : 68[4] -> 76[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96893:97310 [5] NCCL INFO Channel 06/0 : 37[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96892:97308 [4] NCCL INFO Channel 02/0 : 36[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114519:114931 [7] NCCL INFO Channel 00/0 : 39[7] -> 71[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96894:97312 [6] NCCL INFO Channel 00/0 : 38[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630207:1630648 [2] NCCL INFO Channel 09/0 : 74[2] -> 90[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630205:1630646 [0] NCCL INFO Channel 13/0 : 72[0] -> 88[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114513:114933 [1] NCCL INFO Channel 00/0 : 33[1] -> 65[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114519:114931 [7] NCCL INFO Channel 02/0 : 39[7] -> 71[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114517:114935 [5] NCCL INFO Channel 00/0 : 37[5] -> 69[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96889:97309 [1] NCCL INFO Channel 06/0 : 33[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114515:114936 [3] NCCL INFO Channel 00/0 : 35[3] -> 67[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96888:97306 [0] NCCL INFO Channel 02/0 : 32[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96895:97307 [7] NCCL INFO Channel 06/0 : 39[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114519:114931 [7] NCCL INFO Channel 04/0 : 39[7] -> 71[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114513:114933 [1] NCCL INFO Channel 02/0 : 33[1] -> 65[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114517:114935 [5] NCCL INFO Channel 02/0 : 37[5] -> 69[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96893:97310 [5] NCCL INFO Channel 08/0 : 37[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114515:114936 [3] NCCL INFO Channel 02/0 : 35[3] -> 67[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96892:97308 [4] NCCL INFO Channel 06/0 : 36[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96894:97312 [6] NCCL INFO Channel 02/0 : 38[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114519:114931 [7] NCCL INFO Channel 06/0 : 39[7] -> 71[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96889:97309 [1] NCCL INFO Channel 08/0 : 33[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630207:1630648 [2] NCCL INFO Channel 11/0 : 74[2] -> 90[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630205:1630646 [0] NCCL INFO Channel 15/0 : 72[0] -> 88[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114513:114933 [1] NCCL INFO Channel 04/0 : 33[1] -> 65[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96888:97306 [0] NCCL INFO Channel 04/0 : 32[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114517:114935 [5] NCCL INFO Channel 04/0 : 37[5] -> 69[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96891:97311 [3] NCCL INFO Channel 00/0 : 35[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114519:114931 [7] NCCL INFO Channel 08/0 : 39[7] -> 71[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96895:97307 [7] NCCL INFO Channel 08/0 : 39[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114513:114933 [1] NCCL INFO Channel 06/0 : 33[1] -> 65[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114517:114935 [5] NCCL INFO Channel 06/0 : 37[5] -> 69[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96893:97310 [5] NCCL INFO Channel 10/0 : 37[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96892:97308 [4] NCCL INFO Channel 08/0 : 36[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96889:97309 [1] NCCL INFO Channel 10/0 : 33[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96894:97312 [6] NCCL INFO Channel 04/0 : 38[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114515:114936 [3] NCCL INFO Channel 04/0 : 35[3] -> 67[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96890:97313 [2] NCCL INFO Channel 00/0 : 34[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630207:1630648 [2] NCCL INFO Channel 13/0 : 74[2] -> 90[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114519:114931 [7] NCCL INFO Channel 10/0 : 39[7] -> 71[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96888:97306 [0] NCCL INFO Channel 06/0 : 32[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114513:114933 [1] NCCL INFO Channel 08/0 : 33[1] -> 65[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96891:97311 [3] NCCL INFO Channel 02/0 : 35[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114517:114935 [5] NCCL INFO Channel 08/0 : 37[5] -> 69[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114518:114937 [6] NCCL INFO Channel 00/0 : 38[6] -> 70[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96895:97307 [7] NCCL INFO Channel 10/0 : 39[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114515:114936 [3] NCCL INFO Channel 06/0 : 35[3] -> 67[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114519:114931 [7] NCCL INFO Channel 12/0 : 39[7] -> 71[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96892:97308 [4] NCCL INFO Channel 10/0 : 36[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114512:114930 [0] NCCL INFO Channel 02/0 : 32[0] -> 64[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96889:97309 [1] NCCL INFO Channel 12/0 : 33[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96893:97310 [5] NCCL INFO Channel 12/0 : 37[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96894:97312 [6] NCCL INFO Channel 08/0 : 38[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630207:1630648 [2] NCCL INFO Channel 15/0 : 74[2] -> 90[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114513:114933 [1] NCCL INFO Channel 10/0 : 33[1] -> 65[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114516:114934 [4] NCCL INFO Channel 00/0 : 36[4] -> 68[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96890:97313 [2] NCCL INFO Channel 04/0 : 34[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114517:114935 [5] NCCL INFO Channel 10/0 : 37[5] -> 69[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114518:114937 [6] NCCL INFO Channel 02/0 : 38[6] -> 70[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114519:114931 [7] NCCL INFO Channel 14/0 : 39[7] -> 71[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114515:114936 [3] NCCL INFO Channel 08/0 : 35[3] -> 67[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96891:97311 [3] NCCL INFO Channel 04/0 : 35[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96888:97306 [0] NCCL INFO Channel 08/0 : 32[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96895:97307 [7] NCCL INFO Channel 12/0 : 39[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114512:114930 [0] NCCL INFO Channel 04/0 : 32[0] -> 64[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96892:97308 [4] NCCL INFO Channel 12/0 : 36[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114513:114933 [1] NCCL INFO Channel 12/0 : 33[1] -> 65[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114516:114934 [4] NCCL INFO Channel 02/0 : 36[4] -> 68[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96889:97309 [1] NCCL INFO Channel 14/0 : 33[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114517:114935 [5] NCCL INFO Channel 12/0 : 37[5] -> 69[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96893:97310 [5] NCCL INFO Channel 14/0 : 37[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114515:114936 [3] NCCL INFO Channel 10/0 : 35[3] -> 67[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96894:97312 [6] NCCL INFO Channel 10/0 : 38[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114518:114937 [6] NCCL INFO Channel 04/0 : 38[6] -> 70[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114512:114930 [0] NCCL INFO Channel 06/0 : 32[0] -> 64[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114519:114931 [7] NCCL INFO Channel 00/0 : 71[7] -> 103[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96890:97313 [2] NCCL INFO Channel 06/0 : 34[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96891:97311 [3] NCCL INFO Channel 06/0 : 35[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114513:114933 [1] NCCL INFO Channel 14/0 : 33[1] -> 65[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96888:97306 [0] NCCL INFO Channel 10/0 : 32[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114516:114934 [4] NCCL INFO Channel 06/0 : 36[4] -> 68[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96895:97307 [7] NCCL INFO Channel 14/0 : 39[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96892:97308 [4] NCCL INFO Channel 14/0 : 36[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114517:114935 [5] NCCL INFO Channel 14/0 : 37[5] -> 69[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96894:97312 [6] NCCL INFO Channel 12/0 : 38[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114515:114936 [3] NCCL INFO Channel 12/0 : 35[3] -> 67[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96890:97313 [2] NCCL INFO Channel 08/0 : 34[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96891:97311 [3] NCCL INFO Channel 08/0 : 35[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96888:97306 [0] NCCL INFO Channel 12/0 : 32[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114512:114930 [0] NCCL INFO Channel 08/0 : 32[0] -> 64[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96894:97312 [6] NCCL INFO Channel 14/0 : 38[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114518:114937 [6] NCCL INFO Channel 08/0 : 38[6] -> 70[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114513:114933 [1] NCCL INFO Channel 00/0 : 65[1] -> 97[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114514:114932 [2] NCCL INFO Channel 00/0 : 34[2] -> 66[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96890:97313 [2] NCCL INFO Channel 10/0 : 34[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96891:97311 [3] NCCL INFO Channel 10/0 : 35[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96888:97306 [0] NCCL INFO Channel 14/0 : 32[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114519:114931 [7] NCCL INFO Channel 02/0 : 71[7] -> 103[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96891:97311 [3] NCCL INFO Channel 12/0 : 35[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114517:114935 [5] NCCL INFO Channel 00/0 : 69[5] -> 101[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114516:114934 [4] NCCL INFO Channel 08/0 : 36[4] -> 68[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114515:114936 [3] NCCL INFO Channel 14/0 : 35[3] -> 67[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96890:97313 [2] NCCL INFO Channel 12/0 : 34[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114512:114930 [0] NCCL INFO Channel 10/0 : 32[0] -> 64[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96891:97311 [3] NCCL INFO Channel 14/0 : 35[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114513:114933 [1] NCCL INFO Channel 02/0 : 65[1] -> 97[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96890:97313 [2] NCCL INFO Channel 14/0 : 34[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114518:114937 [6] NCCL INFO Channel 10/0 : 38[6] -> 70[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114519:114931 [7] NCCL INFO Channel 04/0 : 71[7] -> 103[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114515:114936 [3] NCCL INFO Channel 00/0 : 67[3] -> 99[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114514:114932 [2] NCCL INFO Channel 04/0 : 34[2] -> 66[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114516:114934 [4] NCCL INFO Channel 10/0 : 36[4] -> 68[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114517:114935 [5] NCCL INFO Channel 02/0 : 69[5] -> 101[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114513:114933 [1] NCCL INFO Channel 04/0 : 65[1] -> 97[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114512:114930 [0] NCCL INFO Channel 12/0 : 32[0] -> 64[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114519:114931 [7] NCCL INFO Channel 06/0 : 71[7] -> 103[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114518:114937 [6] NCCL INFO Channel 12/0 : 38[6] -> 70[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114515:114936 [3] NCCL INFO Channel 02/0 : 67[3] -> 99[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272156:272570 [6] NCCL INFO Channel 00/0 : 110[6] -> 118[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114514:114932 [2] NCCL INFO Channel 06/0 : 34[2] -> 66[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114513:114933 [1] NCCL INFO Channel 06/0 : 65[1] -> 97[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114516:114934 [4] NCCL INFO Channel 12/0 : 36[4] -> 68[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114517:114935 [5] NCCL INFO Channel 04/0 : 69[5] -> 101[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114512:114930 [0] NCCL INFO Channel 14/0 : 32[0] -> 64[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114515:114936 [3] NCCL INFO Channel 04/0 : 67[3] -> 99[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114519:114931 [7] NCCL INFO Channel 08/0 : 71[7] -> 103[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114517:114935 [5] NCCL INFO Channel 06/0 : 69[5] -> 101[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114514:114932 [2] NCCL INFO Channel 08/0 : 34[2] -> 66[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114519:114931 [7] NCCL INFO Channel 10/0 : 71[7] -> 103[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272154:272569 [4] NCCL INFO Channel 00/0 : 108[4] -> 116[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114515:114936 [3] NCCL INFO Channel 06/0 : 67[3] -> 99[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114518:114937 [6] NCCL INFO Channel 14/0 : 38[6] -> 70[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114514:114932 [2] NCCL INFO Channel 10/0 : 34[2] -> 66[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114519:114931 [7] NCCL INFO Channel 12/0 : 71[7] -> 103[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114517:114935 [5] NCCL INFO Channel 08/0 : 69[5] -> 101[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114515:114936 [3] NCCL INFO Channel 08/0 : 67[3] -> 99[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114518:114937 [6] NCCL INFO Channel 00/0 : 70[6] -> 102[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114515:114936 [3] NCCL INFO Channel 10/0 : 67[3] -> 99[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114514:114932 [2] NCCL INFO Channel 12/0 : 34[2] -> 66[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114516:114934 [4] NCCL INFO Channel 14/0 : 36[4] -> 68[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114517:114935 [5] NCCL INFO Channel 10/0 : 69[5] -> 101[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114519:114931 [7] NCCL INFO Channel 14/0 : 71[7] -> 103[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272156:272570 [6] NCCL INFO Channel 01/0 : 110[6] -> 118[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114518:114937 [6] NCCL INFO Channel 02/0 : 70[6] -> 102[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272154:272569 [4] NCCL INFO Channel 01/0 : 108[4] -> 116[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114515:114936 [3] NCCL INFO Channel 12/0 : 67[3] -> 99[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114512:114930 [0] NCCL INFO Channel 02/0 : 64[0] -> 96[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114514:114932 [2] NCCL INFO Channel 14/0 : 34[2] -> 66[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114517:114935 [5] NCCL INFO Channel 12/0 : 69[5] -> 101[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114516:114934 [4] NCCL INFO Channel 00/0 : 68[4] -> 100[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114518:114937 [6] NCCL INFO Channel 04/0 : 70[6] -> 102[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114515:114936 [3] NCCL INFO Channel 14/0 : 67[3] -> 99[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114512:114930 [0] NCCL INFO Channel 04/0 : 64[0] -> 96[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114516:114934 [4] NCCL INFO Channel 02/0 : 68[4] -> 100[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272156:272570 [6] NCCL INFO Channel 02/0 : 110[6] -> 118[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272150:272576 [0] NCCL INFO Channel 01/0 : 104[0] -> 112[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272154:272569 [4] NCCL INFO Channel 02/0 : 108[4] -> 116[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114514:114932 [2] NCCL INFO Channel 00/0 : 66[2] -> 98[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114517:114935 [5] NCCL INFO Channel 14/0 : 69[5] -> 101[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114518:114937 [6] NCCL INFO Channel 08/0 : 70[6] -> 102[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114513:114933 [1] NCCL INFO Channel 08/0 : 65[1] -> 97[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114512:114930 [0] NCCL INFO Channel 06/0 : 64[0] -> 96[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114514:114932 [2] NCCL INFO Channel 04/0 : 66[2] -> 98[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114516:114934 [4] NCCL INFO Channel 06/0 : 68[4] -> 100[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114518:114937 [6] NCCL INFO Channel 10/0 : 70[6] -> 102[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272156:272570 [6] NCCL INFO Channel 03/0 : 110[6] -> 118[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114512:114930 [0] NCCL INFO Channel 08/0 : 64[0] -> 96[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114513:114933 [1] NCCL INFO Channel 10/0 : 65[1] -> 97[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272150:272576 [0] NCCL INFO Channel 02/0 : 104[0] -> 112[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114514:114932 [2] NCCL INFO Channel 06/0 : 66[2] -> 98[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272154:272569 [4] NCCL INFO Channel 03/0 : 108[4] -> 116[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114516:114934 [4] NCCL INFO Channel 08/0 : 68[4] -> 100[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114512:114930 [0] NCCL INFO Channel 10/0 : 64[0] -> 96[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114513:114933 [1] NCCL INFO Channel 12/0 : 65[1] -> 97[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114518:114937 [6] NCCL INFO Channel 12/0 : 70[6] -> 102[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114514:114932 [2] NCCL INFO Channel 08/0 : 66[2] -> 98[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272155:272574 [5] NCCL INFO Channel 00/0 : 109[5] -> 117[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272156:272570 [6] NCCL INFO Channel 04/0 : 110[6] -> 118[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272150:272576 [0] NCCL INFO Channel 03/0 : 104[0] -> 112[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272154:272569 [4] NCCL INFO Channel 05/0 : 108[4] -> 116[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114516:114934 [4] NCCL INFO Channel 10/0 : 68[4] -> 100[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114512:114930 [0] NCCL INFO Channel 12/0 : 64[0] -> 96[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114513:114933 [1] NCCL INFO Channel 14/0 : 65[1] -> 97[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114518:114937 [6] NCCL INFO Channel 14/0 : 70[6] -> 102[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114514:114932 [2] NCCL INFO Channel 10/0 : 66[2] -> 98[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114516:114934 [4] NCCL INFO Channel 12/0 : 68[4] -> 100[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114512:114930 [0] NCCL INFO Channel 14/0 : 64[0] -> 96[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114514:114932 [2] NCCL INFO Channel 12/0 : 66[2] -> 98[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272155:272574 [5] NCCL INFO Channel 01/0 : 109[5] -> 117[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272157:272572 [7] NCCL INFO Channel 00/0 : 111[7] -> 119[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272156:272570 [6] NCCL INFO Channel 05/0 : 110[6] -> 118[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114516:114934 [4] NCCL INFO Channel 14/0 : 68[4] -> 100[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114514:114932 [2] NCCL INFO Channel 14/0 : 66[2] -> 98[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272154:272569 [4] NCCL INFO Channel 06/0 : 108[4] -> 116[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272150:272576 [0] NCCL INFO Channel 04/0 : 104[0] -> 112[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272151:272571 [1] NCCL INFO Channel 00/0 : 105[1] -> 113[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272155:272574 [5] NCCL INFO Channel 02/0 : 109[5] -> 117[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272157:272572 [7] NCCL INFO Channel 01/0 : 111[7] -> 119[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272156:272570 [6] NCCL INFO Channel 07/0 : 110[6] -> 118[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272154:272569 [4] NCCL INFO Channel 07/0 : 108[4] -> 116[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272150:272576 [0] NCCL INFO Channel 05/0 : 104[0] -> 112[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272153:272573 [3] NCCL INFO Channel 00/0 : 107[3] -> 115[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272155:272574 [5] NCCL INFO Channel 03/0 : 109[5] -> 117[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272157:272572 [7] NCCL INFO Channel 02/0 : 111[7] -> 119[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272156:272570 [6] NCCL INFO Channel 08/0 : 110[6] -> 118[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272154:272569 [4] NCCL INFO Channel 08/0 : 108[4] -> 116[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272151:272571 [1] NCCL INFO Channel 02/0 : 105[1] -> 113[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272150:272576 [0] NCCL INFO Channel 06/0 : 104[0] -> 112[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272155:272574 [5] NCCL INFO Channel 04/0 : 109[5] -> 117[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272157:272572 [7] NCCL INFO Channel 03/0 : 111[7] -> 119[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272153:272573 [3] NCCL INFO Channel 01/0 : 107[3] -> 115[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272156:272570 [6] NCCL INFO Channel 09/0 : 110[6] -> 118[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272152:272575 [2] NCCL INFO Channel 00/0 : 106[2] -> 114[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272154:272569 [4] NCCL INFO Channel 09/0 : 108[4] -> 116[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272150:272576 [0] NCCL INFO Channel 07/0 : 104[0] -> 112[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272151:272571 [1] NCCL INFO Channel 03/0 : 105[1] -> 113[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272155:272574 [5] NCCL INFO Channel 06/0 : 109[5] -> 117[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272157:272572 [7] NCCL INFO Channel 04/0 : 111[7] -> 119[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272153:272573 [3] NCCL INFO Channel 02/0 : 107[3] -> 115[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272156:272570 [6] NCCL INFO Channel 10/0 : 110[6] -> 118[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272154:272569 [4] NCCL INFO Channel 10/0 : 108[4] -> 116[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272152:272575 [2] NCCL INFO Channel 01/0 : 106[2] -> 114[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272150:272576 [0] NCCL INFO Channel 09/0 : 104[0] -> 112[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272157:272572 [7] NCCL INFO Channel 05/0 : 111[7] -> 119[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272155:272574 [5] NCCL INFO Channel 07/0 : 109[5] -> 117[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272151:272571 [1] NCCL INFO Channel 04/0 : 105[1] -> 113[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272153:272573 [3] NCCL INFO Channel 04/0 : 107[3] -> 115[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272156:272570 [6] NCCL INFO Channel 11/0 : 110[6] -> 118[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272154:272569 [4] NCCL INFO Channel 11/0 : 108[4] -> 116[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272152:272575 [2] NCCL INFO Channel 03/0 : 106[2] -> 114[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272150:272576 [0] NCCL INFO Channel 10/0 : 104[0] -> 112[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272157:272572 [7] NCCL INFO Channel 06/0 : 111[7] -> 119[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153565:153985 [3] NCCL INFO Channel 00/0 : 83[3] -> 91[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272151:272571 [1] NCCL INFO Channel 05/0 : 105[1] -> 113[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272155:272574 [5] NCCL INFO Channel 08/0 : 109[5] -> 117[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272153:272573 [3] NCCL INFO Channel 05/0 : 107[3] -> 115[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272156:272570 [6] NCCL INFO Channel 12/0 : 110[6] -> 118[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272154:272569 [4] NCCL INFO Channel 13/0 : 108[4] -> 116[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272157:272572 [7] NCCL INFO Channel 08/0 : 111[7] -> 119[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272152:272575 [2] NCCL INFO Channel 04/0 : 106[2] -> 114[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272150:272576 [0] NCCL INFO Channel 11/0 : 104[0] -> 112[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272151:272571 [1] NCCL INFO Channel 06/0 : 105[1] -> 113[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153565:153985 [3] NCCL INFO Channel 02/0 : 83[3] -> 91[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272155:272574 [5] NCCL INFO Channel 09/0 : 109[5] -> 117[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272153:272573 [3] NCCL INFO Channel 06/0 : 107[3] -> 115[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272156:272570 [6] NCCL INFO Channel 13/0 : 110[6] -> 118[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153569:153981 [7] NCCL INFO Channel 00/0 : 87[7] -> 95[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272154:272569 [4] NCCL INFO Channel 14/0 : 108[4] -> 116[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272157:272572 [7] NCCL INFO Channel 09/0 : 111[7] -> 119[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272150:272576 [0] NCCL INFO Channel 12/0 : 104[0] -> 112[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153565:153985 [3] NCCL INFO Channel 04/0 : 83[3] -> 91[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153569:153981 [7] NCCL INFO Channel 02/0 : 87[7] -> 95[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272152:272575 [2] NCCL INFO Channel 05/0 : 106[2] -> 114[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272151:272571 [1] NCCL INFO Channel 07/0 : 105[1] -> 113[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272155:272574 [5] NCCL INFO Channel 10/0 : 109[5] -> 117[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272153:272573 [3] NCCL INFO Channel 07/0 : 107[3] -> 115[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272156:272570 [6] NCCL INFO Channel 15/0 : 110[6] -> 118[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272154:272569 [4] NCCL INFO Channel 15/0 : 108[4] -> 116[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153567:153979 [5] NCCL INFO Channel 00/0 : 85[5] -> 93[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272157:272572 [7] NCCL INFO Channel 10/0 : 111[7] -> 119[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153568:153982 [6] NCCL INFO Channel 00/0 : 86[6] -> 94[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153565:153985 [3] NCCL INFO Channel 06/0 : 83[3] -> 91[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272150:272576 [0] NCCL INFO Channel 13/0 : 104[0] -> 112[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153569:153981 [7] NCCL INFO Channel 04/0 : 87[7] -> 95[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272152:272575 [2] NCCL INFO Channel 06/0 : 106[2] -> 114[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153564:153984 [2] NCCL INFO Channel 00/0 : 82[2] -> 90[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153562:153983 [0] NCCL INFO Channel 02/0 : 80[0] -> 88[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272155:272574 [5] NCCL INFO Channel 11/0 : 109[5] -> 117[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153567:153979 [5] NCCL INFO Channel 02/0 : 85[5] -> 93[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272151:272571 [1] NCCL INFO Channel 08/0 : 105[1] -> 113[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272156:272570 [6] NCCL INFO Channel 00/0 : 118[6] -> 126[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272154:272569 [4] NCCL INFO Channel 00/0 : 116[4] -> 124[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272153:272573 [3] NCCL INFO Channel 08/0 : 107[3] -> 115[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272157:272572 [7] NCCL INFO Channel 11/0 : 111[7] -> 119[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272150:272576 [0] NCCL INFO Channel 14/0 : 104[0] -> 112[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272152:272575 [2] NCCL INFO Channel 07/0 : 106[2] -> 114[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272155:272574 [5] NCCL INFO Channel 12/0 : 109[5] -> 117[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272151:272571 [1] NCCL INFO Channel 10/0 : 105[1] -> 113[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153568:153982 [6] NCCL INFO Channel 02/0 : 86[6] -> 94[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153565:153985 [3] NCCL INFO Channel 08/0 : 83[3] -> 91[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153566:153980 [4] NCCL INFO Channel 00/0 : 84[4] -> 92[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153569:153981 [7] NCCL INFO Channel 06/0 : 87[7] -> 95[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153564:153984 [2] NCCL INFO Channel 04/0 : 82[2] -> 90[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272156:272570 [6] NCCL INFO Channel 02/0 : 118[6] -> 126[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272154:272569 [4] NCCL INFO Channel 02/0 : 116[4] -> 124[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272157:272572 [7] NCCL INFO Channel 12/0 : 111[7] -> 119[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153567:153979 [5] NCCL INFO Channel 04/0 : 85[5] -> 93[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272153:272573 [3] NCCL INFO Channel 09/0 : 107[3] -> 115[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272150:272576 [0] NCCL INFO Channel 15/0 : 104[0] -> 112[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272152:272575 [2] NCCL INFO Channel 08/0 : 106[2] -> 114[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272155:272574 [5] NCCL INFO Channel 14/0 : 109[5] -> 117[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272151:272571 [1] NCCL INFO Channel 11/0 : 105[1] -> 113[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272156:272570 [6] NCCL INFO Channel 04/0 : 118[6] -> 126[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272154:272569 [4] NCCL INFO Channel 06/0 : 116[4] -> 124[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153565:153985 [3] NCCL INFO Channel 10/0 : 83[3] -> 91[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153562:153983 [0] NCCL INFO Channel 04/0 : 80[0] -> 88[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153568:153982 [6] NCCL INFO Channel 04/0 : 86[6] -> 94[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153566:153980 [4] NCCL INFO Channel 02/0 : 84[4] -> 92[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153564:153984 [2] NCCL INFO Channel 06/0 : 82[2] -> 90[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153569:153981 [7] NCCL INFO Channel 08/0 : 87[7] -> 95[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272157:272572 [7] NCCL INFO Channel 13/0 : 111[7] -> 119[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272153:272573 [3] NCCL INFO Channel 10/0 : 107[3] -> 115[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153567:153979 [5] NCCL INFO Channel 06/0 : 85[5] -> 93[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153563:153986 [1] NCCL INFO Channel 00/0 : 81[1] -> 89[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153565:153985 [3] NCCL INFO Channel 12/0 : 83[3] -> 91[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272150:272576 [0] NCCL INFO Channel 02/0 : 112[0] -> 120[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272152:272575 [2] NCCL INFO Channel 09/0 : 106[2] -> 114[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272155:272574 [5] NCCL INFO Channel 15/0 : 109[5] -> 117[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272151:272571 [1] NCCL INFO Channel 12/0 : 105[1] -> 113[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272154:272569 [4] NCCL INFO Channel 08/0 : 116[4] -> 124[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272156:272570 [6] NCCL INFO Channel 08/0 : 118[6] -> 126[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272157:272572 [7] NCCL INFO Channel 14/0 : 111[7] -> 119[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272153:272573 [3] NCCL INFO Channel 12/0 : 107[3] -> 115[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153562:153983 [0] NCCL INFO Channel 06/0 : 80[0] -> 88[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272152:272575 [2] NCCL INFO Channel 11/0 : 106[2] -> 114[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153568:153982 [6] NCCL INFO Channel 08/0 : 86[6] -> 94[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153564:153984 [2] NCCL INFO Channel 08/0 : 82[2] -> 90[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153566:153980 [4] NCCL INFO Channel 06/0 : 84[4] -> 92[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153569:153981 [7] NCCL INFO Channel 10/0 : 87[7] -> 95[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272155:272574 [5] NCCL INFO Channel 00/0 : 117[5] -> 125[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272150:272576 [0] NCCL INFO Channel 04/0 : 112[0] -> 120[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272151:272571 [1] NCCL INFO Channel 13/0 : 105[1] -> 113[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272154:272569 [4] NCCL INFO Channel 10/0 : 116[4] -> 124[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272156:272570 [6] NCCL INFO Channel 10/0 : 118[6] -> 126[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153563:153986 [1] NCCL INFO Channel 02/0 : 81[1] -> 89[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153567:153979 [5] NCCL INFO Channel 08/0 : 85[5] -> 93[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153565:153985 [3] NCCL INFO Channel 14/0 : 83[3] -> 91[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272157:272572 [7] NCCL INFO Channel 00/0 : 119[7] -> 127[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153562:153983 [0] NCCL INFO Channel 08/0 : 80[0] -> 88[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272153:272573 [3] NCCL INFO Channel 13/0 : 107[3] -> 115[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153568:153982 [6] NCCL INFO Channel 10/0 : 86[6] -> 94[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272152:272575 [2] NCCL INFO Channel 12/0 : 106[2] -> 114[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272150:272576 [0] NCCL INFO Channel 06/0 : 112[0] -> 120[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272155:272574 [5] NCCL INFO Channel 02/0 : 117[5] -> 125[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272151:272571 [1] NCCL INFO Channel 14/0 : 105[1] -> 113[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272156:272570 [6] NCCL INFO Channel 12/0 : 118[6] -> 126[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272154:272569 [4] NCCL INFO Channel 12/0 : 116[4] -> 124[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272157:272572 [7] NCCL INFO Channel 02/0 : 119[7] -> 127[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272153:272573 [3] NCCL INFO Channel 14/0 : 107[3] -> 115[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272152:272575 [2] NCCL INFO Channel 13/0 : 106[2] -> 114[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272150:272576 [0] NCCL INFO Channel 08/0 : 112[0] -> 120[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272155:272574 [5] NCCL INFO Channel 04/0 : 117[5] -> 125[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272151:272571 [1] NCCL INFO Channel 15/0 : 105[1] -> 113[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272156:272570 [6] NCCL INFO Channel 14/0 : 118[6] -> 126[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227466:1227887 [3] NCCL INFO Channel 00/0 : 83[3] -> 99[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272154:272569 [4] NCCL INFO Channel 14/0 : 116[4] -> 124[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272157:272572 [7] NCCL INFO Channel 04/0 : 119[7] -> 127[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227466:1227887 [3] NCCL INFO Channel 02/0 : 83[3] -> 99[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272153:272573 [3] NCCL INFO Channel 15/0 : 107[3] -> 115[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272152:272575 [2] NCCL INFO Channel 14/0 : 106[2] -> 114[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272150:272576 [0] NCCL INFO Channel 10/0 : 112[0] -> 120[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272155:272574 [5] NCCL INFO Channel 06/0 : 117[5] -> 125[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272151:272571 [1] NCCL INFO Channel 00/0 : 113[1] -> 121[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272157:272572 [7] NCCL INFO Channel 06/0 : 119[7] -> 127[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227466:1227887 [3] NCCL INFO Channel 04/0 : 83[3] -> 99[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272153:272573 [3] NCCL INFO Channel 00/0 : 115[3] -> 123[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272152:272575 [2] NCCL INFO Channel 15/0 : 106[2] -> 114[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272150:272576 [0] NCCL INFO Channel 12/0 : 112[0] -> 120[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272155:272574 [5] NCCL INFO Channel 08/0 : 117[5] -> 125[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272151:272571 [1] NCCL INFO Channel 02/0 : 113[1] -> 121[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227466:1227887 [3] NCCL INFO Channel 06/0 : 83[3] -> 99[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272157:272572 [7] NCCL INFO Channel 08/0 : 119[7] -> 127[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227470:1227885 [7] NCCL INFO Channel 00/0 : 87[7] -> 103[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272153:272573 [3] NCCL INFO Channel 02/0 : 115[3] -> 123[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272152:272575 [2] NCCL INFO Channel 00/0 : 114[2] -> 122[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235654:236070 [6] NCCL INFO Channel 01/0 : 94[6] -> 110[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235652:236066 [4] NCCL INFO Channel 01/0 : 92[4] -> 108[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272150:272576 [0] NCCL INFO Channel 14/0 : 112[0] -> 120[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272151:272571 [1] NCCL INFO Channel 04/0 : 113[1] -> 121[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272155:272574 [5] NCCL INFO Channel 10/0 : 117[5] -> 125[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272157:272572 [7] NCCL INFO Channel 10/0 : 119[7] -> 127[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227466:1227887 [3] NCCL INFO Channel 08/0 : 83[3] -> 99[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227469:1227882 [6] NCCL INFO Channel 00/0 : 86[6] -> 102[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227470:1227885 [7] NCCL INFO Channel 02/0 : 87[7] -> 103[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235654:236070 [6] NCCL INFO Channel 03/0 : 94[6] -> 110[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235652:236066 [4] NCCL INFO Channel 03/0 : 92[4] -> 108[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272153:272573 [3] NCCL INFO Channel 04/0 : 115[3] -> 123[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272152:272575 [2] NCCL INFO Channel 04/0 : 114[2] -> 122[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272151:272571 [1] NCCL INFO Channel 06/0 : 113[1] -> 121[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227468:1227883 [5] NCCL INFO Channel 00/0 : 85[5] -> 101[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227465:1227881 [2] NCCL INFO Channel 00/0 : 82[2] -> 98[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272157:272572 [7] NCCL INFO Channel 12/0 : 119[7] -> 127[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272155:272574 [5] NCCL INFO Channel 12/0 : 117[5] -> 125[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227463:1227880 [0] NCCL INFO Channel 02/0 : 80[0] -> 96[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227466:1227887 [3] NCCL INFO Channel 10/0 : 83[3] -> 99[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227469:1227882 [6] NCCL INFO Channel 02/0 : 86[6] -> 102[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227470:1227885 [7] NCCL INFO Channel 04/0 : 87[7] -> 103[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227467:1227886 [4] NCCL INFO Channel 00/0 : 84[4] -> 100[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235652:236066 [4] NCCL INFO Channel 05/0 : 92[4] -> 108[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235654:236070 [6] NCCL INFO Channel 05/0 : 94[6] -> 110[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272153:272573 [3] NCCL INFO Channel 06/0 : 115[3] -> 123[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272152:272575 [2] NCCL INFO Channel 06/0 : 114[2] -> 122[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272156:272570 [6] NCCL INFO Channel 00/0 : 102[6] -> 118[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272154:272569 [4] NCCL INFO Channel 00/0 : 100[4] -> 116[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227468:1227883 [5] NCCL INFO Channel 02/0 : 85[5] -> 101[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272151:272571 [1] NCCL INFO Channel 08/0 : 113[1] -> 121[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227465:1227881 [2] NCCL INFO Channel 04/0 : 82[2] -> 98[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272157:272572 [7] NCCL INFO Channel 14/0 : 119[7] -> 127[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272155:272574 [5] NCCL INFO Channel 14/0 : 117[5] -> 125[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235648:236064 [0] NCCL INFO Channel 01/0 : 88[0] -> 104[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235652:236066 [4] NCCL INFO Channel 07/0 : 92[4] -> 108[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235654:236070 [6] NCCL INFO Channel 07/0 : 94[6] -> 110[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227463:1227880 [0] NCCL INFO Channel 04/0 : 80[0] -> 96[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272153:272573 [3] NCCL INFO Channel 08/0 : 115[3] -> 123[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272152:272575 [2] NCCL INFO Channel 08/0 : 114[2] -> 122[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272156:272570 [6] NCCL INFO Channel 02/0 : 102[6] -> 118[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272154:272569 [4] NCCL INFO Channel 02/0 : 100[4] -> 116[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227466:1227887 [3] NCCL INFO Channel 12/0 : 83[3] -> 99[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272151:272571 [1] NCCL INFO Channel 10/0 : 113[1] -> 121[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227469:1227882 [6] NCCL INFO Channel 04/0 : 86[6] -> 102[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227470:1227885 [7] NCCL INFO Channel 06/0 : 87[7] -> 103[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227467:1227886 [4] NCCL INFO Channel 02/0 : 84[4] -> 100[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235648:236064 [0] NCCL INFO Channel 03/0 : 88[0] -> 104[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227468:1227883 [5] NCCL INFO Channel 04/0 : 85[5] -> 101[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272153:272573 [3] NCCL INFO Channel 10/0 : 115[3] -> 123[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272152:272575 [2] NCCL INFO Channel 10/0 : 114[2] -> 122[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235652:236066 [4] NCCL INFO Channel 09/0 : 92[4] -> 108[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235654:236070 [6] NCCL INFO Channel 09/0 : 94[6] -> 110[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227465:1227881 [2] NCCL INFO Channel 06/0 : 82[2] -> 98[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227463:1227880 [0] NCCL INFO Channel 06/0 : 80[0] -> 96[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227466:1227887 [3] NCCL INFO Channel 14/0 : 83[3] -> 99[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227469:1227882 [6] NCCL INFO Channel 08/0 : 86[6] -> 102[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227470:1227885 [7] NCCL INFO Channel 08/0 : 87[7] -> 103[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227467:1227886 [4] NCCL INFO Channel 06/0 : 84[4] -> 100[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272156:272570 [6] NCCL INFO Channel 04/0 : 102[6] -> 118[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272154:272569 [4] NCCL INFO Channel 06/0 : 100[4] -> 116[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272151:272571 [1] NCCL INFO Channel 12/0 : 113[1] -> 121[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227468:1227883 [5] NCCL INFO Channel 06/0 : 85[5] -> 101[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272150:272576 [0] NCCL INFO Channel 02/0 : 96[0] -> 112[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227463:1227880 [0] NCCL INFO Channel 08/0 : 80[0] -> 96[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227465:1227881 [2] NCCL INFO Channel 08/0 : 82[2] -> 98[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235648:236064 [0] NCCL INFO Channel 05/0 : 88[0] -> 104[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272153:272573 [3] NCCL INFO Channel 12/0 : 115[3] -> 123[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235652:236066 [4] NCCL INFO Channel 11/0 : 92[4] -> 108[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235654:236070 [6] NCCL INFO Channel 11/0 : 94[6] -> 110[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227469:1227882 [6] NCCL INFO Channel 10/0 : 86[6] -> 102[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227470:1227885 [7] NCCL INFO Channel 10/0 : 87[7] -> 103[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272152:272575 [2] NCCL INFO Channel 12/0 : 114[2] -> 122[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227467:1227886 [4] NCCL INFO Channel 08/0 : 84[4] -> 100[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272151:272571 [1] NCCL INFO Channel 14/0 : 113[1] -> 121[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272156:272570 [6] NCCL INFO Channel 08/0 : 102[6] -> 118[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272154:272569 [4] NCCL INFO Channel 08/0 : 100[4] -> 116[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227468:1227883 [5] NCCL INFO Channel 08/0 : 85[5] -> 101[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227464:1227884 [1] NCCL INFO Channel 00/0 : 81[1] -> 97[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235648:236064 [0] NCCL INFO Channel 07/0 : 88[0] -> 104[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272150:272576 [0] NCCL INFO Channel 04/0 : 96[0] -> 112[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272153:272573 [3] NCCL INFO Channel 14/0 : 115[3] -> 123[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235653:236067 [5] NCCL INFO Channel 01/0 : 93[5] -> 109[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235652:236066 [4] NCCL INFO Channel 13/0 : 92[4] -> 108[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235655:236069 [7] NCCL INFO Channel 01/0 : 95[7] -> 111[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235654:236070 [6] NCCL INFO Channel 13/0 : 94[6] -> 110[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272152:272575 [2] NCCL INFO Channel 14/0 : 114[2] -> 122[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227465:1227881 [2] NCCL INFO Channel 10/0 : 82[2] -> 98[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636089:2636508 [5] NCCL INFO Channel 00/0 : 45[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235648:236064 [0] NCCL INFO Channel 09/0 : 88[0] -> 104[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227463:1227880 [0] NCCL INFO Channel 10/0 : 80[0] -> 96[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235653:236067 [5] NCCL INFO Channel 03/0 : 93[5] -> 109[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227470:1227885 [7] NCCL INFO Channel 12/0 : 87[7] -> 103[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227469:1227882 [6] NCCL INFO Channel 12/0 : 86[6] -> 102[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235652:236066 [4] NCCL INFO Channel 15/0 : 92[4] -> 108[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272155:272574 [5] NCCL INFO Channel 00/0 : 101[5] -> 117[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272154:272569 [4] NCCL INFO Channel 10/0 : 100[4] -> 116[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272156:272570 [6] NCCL INFO Channel 10/0 : 102[6] -> 118[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227466:1227887 [3] NCCL INFO Channel 00/0 : 99[3] -> 83[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235655:236069 [7] NCCL INFO Channel 03/0 : 95[7] -> 111[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235654:236070 [6] NCCL INFO Channel 15/0 : 94[6] -> 110[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272150:272576 [0] NCCL INFO Channel 06/0 : 96[0] -> 112[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227467:1227886 [4] NCCL INFO Channel 10/0 : 84[4] -> 100[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272157:272572 [7] NCCL INFO Channel 00/0 : 103[7] -> 119[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636089:2636508 [5] NCCL INFO Channel 01/0 : 45[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227468:1227883 [5] NCCL INFO Channel 10/0 : 85[5] -> 101[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227464:1227884 [1] NCCL INFO Channel 02/0 : 81[1] -> 97[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227465:1227881 [2] NCCL INFO Channel 12/0 : 82[2] -> 98[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227463:1227880 [0] NCCL INFO Channel 12/0 : 80[0] -> 96[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272155:272574 [5] NCCL INFO Channel 02/0 : 101[5] -> 117[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227469:1227882 [6] NCCL INFO Channel 14/0 : 86[6] -> 102[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272154:272569 [4] NCCL INFO Channel 12/0 : 100[4] -> 116[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272156:272570 [6] NCCL INFO Channel 12/0 : 102[6] -> 118[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227470:1227885 [7] NCCL INFO Channel 14/0 : 87[7] -> 103[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636085:2636509 [1] NCCL INFO Channel 00/0 : 41[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272150:272576 [0] NCCL INFO Channel 08/0 : 96[0] -> 112[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272157:272572 [7] NCCL INFO Channel 02/0 : 103[7] -> 119[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227467:1227886 [4] NCCL INFO Channel 12/0 : 84[4] -> 100[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227466:1227887 [3] NCCL INFO Channel 02/0 : 99[3] -> 83[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636087:2636506 [3] NCCL INFO Channel 00/0 : 43[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227468:1227883 [5] NCCL INFO Channel 12/0 : 85[5] -> 101[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227464:1227884 [1] NCCL INFO Channel 04/0 : 81[1] -> 97[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227463:1227880 [0] NCCL INFO Channel 14/0 : 80[0] -> 96[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227465:1227881 [2] NCCL INFO Channel 14/0 : 82[2] -> 98[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636089:2636508 [5] NCCL INFO Channel 02/0 : 45[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636085:2636509 [1] NCCL INFO Channel 02/0 : 41[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636087:2636506 [3] NCCL INFO Channel 01/0 : 43[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153564:153984 [2] NCCL INFO Channel 10/0 : 82[2] -> 90[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153569:153981 [7] NCCL INFO Channel 12/0 : 87[7] -> 95[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153566:153980 [4] NCCL INFO Channel 08/0 : 84[4] -> 92[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153567:153979 [5] NCCL INFO Channel 10/0 : 85[5] -> 93[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153568:153982 [6] NCCL INFO Channel 12/0 : 86[6] -> 94[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153562:153983 [0] NCCL INFO Channel 10/0 : 80[0] -> 88[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153563:153986 [1] NCCL INFO Channel 04/0 : 81[1] -> 89[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227467:1227886 [4] NCCL INFO Channel 14/0 : 84[4] -> 100[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153569:153981 [7] NCCL INFO Channel 14/0 : 87[7] -> 95[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153566:153980 [4] NCCL INFO Channel 10/0 : 84[4] -> 92[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227468:1227883 [5] NCCL INFO Channel 14/0 : 85[5] -> 101[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227466:1227887 [3] NCCL INFO Channel 04/0 : 99[3] -> 83[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227464:1227884 [1] NCCL INFO Channel 06/0 : 81[1] -> 97[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153564:153984 [2] NCCL INFO Channel 12/0 : 82[2] -> 90[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153567:153979 [5] NCCL INFO Channel 12/0 : 85[5] -> 93[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153565:153985 [3] NCCL INFO Channel 01/0 : 75[3] -> 91[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636089:2636508 [5] NCCL INFO Channel 03/0 : 45[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153568:153982 [6] NCCL INFO Channel 14/0 : 86[6] -> 94[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153562:153983 [0] NCCL INFO Channel 12/0 : 80[0] -> 88[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153563:153986 [1] NCCL INFO Channel 06/0 : 81[1] -> 89[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636085:2636509 [1] NCCL INFO Channel 03/0 : 41[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636087:2636506 [3] NCCL INFO Channel 02/0 : 43[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153566:153980 [4] NCCL INFO Channel 12/0 : 84[4] -> 92[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153564:153984 [2] NCCL INFO Channel 14/0 : 82[2] -> 90[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153567:153979 [5] NCCL INFO Channel 14/0 : 85[5] -> 93[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227464:1227884 [1] NCCL INFO Channel 08/0 : 81[1] -> 97[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153562:153983 [0] NCCL INFO Channel 14/0 : 80[0] -> 88[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153565:153985 [3] NCCL INFO Channel 03/0 : 75[3] -> 91[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227466:1227887 [3] NCCL INFO Channel 06/0 : 99[3] -> 83[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153566:153980 [4] NCCL INFO Channel 14/0 : 84[4] -> 92[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153563:153986 [1] NCCL INFO Channel 08/0 : 81[1] -> 89[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227469:1227882 [6] NCCL INFO Channel 00/0 : 102[6] -> 86[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227470:1227885 [7] NCCL INFO Channel 00/0 : 103[7] -> 87[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153565:153985 [3] NCCL INFO Channel 05/0 : 75[3] -> 91[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153563:153986 [1] NCCL INFO Channel 10/0 : 81[1] -> 89[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153569:153981 [7] NCCL INFO Channel 01/0 : 79[7] -> 95[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227465:1227881 [2] NCCL INFO Channel 00/0 : 98[2] -> 82[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636089:2636508 [5] NCCL INFO Channel 04/0 : 45[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636090:2636505 [6] NCCL INFO Channel 00/0 : 46[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153568:153982 [6] NCCL INFO Channel 01/0 : 78[6] -> 94[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227463:1227880 [0] NCCL INFO Channel 02/0 : 96[0] -> 80[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636084:2636511 [0] NCCL INFO Channel 01/0 : 40[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153567:153979 [5] NCCL INFO Channel 01/0 : 77[5] -> 93[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636085:2636509 [1] NCCL INFO Channel 04/0 : 41[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153564:153984 [2] NCCL INFO Channel 01/0 : 74[2] -> 90[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153563:153986 [1] NCCL INFO Channel 12/0 : 81[1] -> 89[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153565:153985 [3] NCCL INFO Channel 07/0 : 75[3] -> 91[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153562:153983 [0] NCCL INFO Channel 01/0 : 72[0] -> 88[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153569:153981 [7] NCCL INFO Channel 03/0 : 79[7] -> 95[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227464:1227884 [1] NCCL INFO Channel 10/0 : 81[1] -> 97[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153568:153982 [6] NCCL INFO Channel 03/0 : 78[6] -> 94[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636087:2636506 [3] NCCL INFO Channel 04/0 : 43[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153567:153979 [5] NCCL INFO Channel 03/0 : 77[5] -> 93[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153566:153980 [4] NCCL INFO Channel 01/0 : 76[4] -> 92[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636091:2636512 [7] NCCL INFO Channel 00/0 : 47[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153564:153984 [2] NCCL INFO Channel 03/0 : 74[2] -> 90[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153563:153986 [1] NCCL INFO Channel 14/0 : 81[1] -> 89[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557270:1557688 [4] NCCL INFO Channel 00/0 : 68[4] -> 100[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557272:1557687 [6] NCCL INFO Channel 00/0 : 70[6] -> 102[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227470:1227885 [7] NCCL INFO Channel 02/0 : 103[7] -> 87[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227469:1227882 [6] NCCL INFO Channel 02/0 : 102[6] -> 86[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227466:1227887 [3] NCCL INFO Channel 08/0 : 99[3] -> 83[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153565:153985 [3] NCCL INFO Channel 09/0 : 75[3] -> 91[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153562:153983 [0] NCCL INFO Channel 03/0 : 72[0] -> 88[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227467:1227886 [4] NCCL INFO Channel 00/0 : 100[4] -> 84[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636090:2636505 [6] NCCL INFO Channel 01/0 : 46[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153569:153981 [7] NCCL INFO Channel 05/0 : 79[7] -> 95[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227465:1227881 [2] NCCL INFO Channel 04/0 : 98[2] -> 82[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227468:1227883 [5] NCCL INFO Channel 00/0 : 101[5] -> 85[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636085:2636509 [1] NCCL INFO Channel 05/0 : 41[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153568:153982 [6] NCCL INFO Channel 05/0 : 78[6] -> 94[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153564:153984 [2] NCCL INFO Channel 05/0 : 74[2] -> 90[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227463:1227880 [0] NCCL INFO Channel 04/0 : 96[0] -> 80[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227464:1227884 [1] NCCL INFO Channel 12/0 : 81[1] -> 97[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636089:2636508 [5] NCCL INFO Channel 06/0 : 45[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636084:2636511 [0] NCCL INFO Channel 02/0 : 40[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153566:153980 [4] NCCL INFO Channel 03/0 : 76[4] -> 92[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153567:153979 [5] NCCL INFO Channel 05/0 : 77[5] -> 93[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557270:1557688 [4] NCCL INFO Channel 02/0 : 68[4] -> 100[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557272:1557687 [6] NCCL INFO Channel 02/0 : 70[6] -> 102[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153565:153985 [3] NCCL INFO Channel 13/0 : 75[3] -> 91[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153562:153983 [0] NCCL INFO Channel 05/0 : 72[0] -> 88[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153569:153981 [7] NCCL INFO Channel 07/0 : 79[7] -> 95[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153568:153982 [6] NCCL INFO Channel 07/0 : 78[6] -> 94[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153566:153980 [4] NCCL INFO Channel 05/0 : 76[4] -> 92[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227470:1227885 [7] NCCL INFO Channel 04/0 : 103[7] -> 87[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636087:2636506 [3] NCCL INFO Channel 05/0 : 43[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636091:2636512 [7] NCCL INFO Channel 01/0 : 47[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153567:153979 [5] NCCL INFO Channel 07/0 : 77[5] -> 93[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153564:153984 [2] NCCL INFO Channel 07/0 : 74[2] -> 90[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227469:1227882 [6] NCCL INFO Channel 04/0 : 102[6] -> 86[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227467:1227886 [4] NCCL INFO Channel 02/0 : 100[4] -> 84[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636090:2636505 [6] NCCL INFO Channel 02/0 : 46[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153565:153985 [3] NCCL INFO Channel 15/0 : 75[3] -> 91[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227466:1227887 [3] NCCL INFO Channel 10/0 : 99[3] -> 83[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153562:153983 [0] NCCL INFO Channel 07/0 : 72[0] -> 88[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557272:1557687 [6] NCCL INFO Channel 04/0 : 70[6] -> 102[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557270:1557688 [4] NCCL INFO Channel 06/0 : 68[4] -> 100[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235648:236064 [0] NCCL INFO Channel 11/0 : 88[0] -> 104[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235653:236067 [5] NCCL INFO Channel 05/0 : 93[5] -> 109[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272155:272574 [5] NCCL INFO Channel 04/0 : 101[5] -> 117[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815967:2816399 [4] NCCL INFO Channel 00/0 : 52[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630208:1630650 [3] NCCL INFO Channel 01/0 : 91[3] -> 75[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636089:2636508 [5] NCCL INFO Channel 07/0 : 45[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-52:2722051:2722465 [6] NCCL INFO Channel 01/0 : 62[6] -> 126[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153563:153986 [1] NCCL INFO Channel 01/0 : 73[1] -> 89[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557272:1557687 [6] NCCL INFO Channel 08/0 : 70[6] -> 102[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235655:236069 [7] NCCL INFO Channel 05/0 : 95[7] -> 111[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272154:272569 [4] NCCL INFO Channel 14/0 : 100[4] -> 116[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227465:1227881 [2] NCCL INFO Channel 06/0 : 98[2] -> 82[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227468:1227883 [5] NCCL INFO Channel 02/0 : 101[5] -> 85[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227464:1227884 [1] NCCL INFO Channel 14/0 : 81[1] -> 97[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227463:1227880 [0] NCCL INFO Channel 06/0 : 96[0] -> 80[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227470:1227885 [7] NCCL INFO Channel 06/0 : 103[7] -> 87[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636085:2636509 [1] NCCL INFO Channel 06/0 : 41[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-52:2722051:2722465 [6] NCCL INFO Channel 03/0 : 62[6] -> 126[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153569:153981 [7] NCCL INFO Channel 09/0 : 79[7] -> 95[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557270:1557688 [4] NCCL INFO Channel 08/0 : 68[4] -> 100[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235648:236064 [0] NCCL INFO Channel 13/0 : 88[0] -> 104[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272156:272570 [6] NCCL INFO Channel 14/0 : 102[6] -> 118[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227469:1227882 [6] NCCL INFO Channel 08/0 : 102[6] -> 86[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636088:2636510 [4] NCCL INFO Channel 00/0 : 44[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636084:2636511 [0] NCCL INFO Channel 03/0 : 40[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-52:2722051:2722465 [6] NCCL INFO Channel 05/0 : 62[6] -> 126[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153568:153982 [6] NCCL INFO Channel 09/0 : 78[6] -> 94[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557266:1557689 [0] NCCL INFO Channel 02/0 : 64[0] -> 96[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235653:236067 [5] NCCL INFO Channel 07/0 : 93[5] -> 109[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235649:236071 [1] NCCL INFO Channel 01/0 : 89[1] -> 105[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272150:272576 [0] NCCL INFO Channel 10/0 : 96[0] -> 112[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272157:272572 [7] NCCL INFO Channel 04/0 : 103[7] -> 119[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272155:272574 [5] NCCL INFO Channel 06/0 : 101[5] -> 117[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815969:2816398 [6] NCCL INFO Channel 00/0 : 54[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815967:2816399 [4] NCCL INFO Channel 02/0 : 52[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227467:1227886 [4] NCCL INFO Channel 06/0 : 100[4] -> 84[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630208:1630650 [3] NCCL INFO Channel 03/0 : 91[3] -> 75[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636091:2636512 [7] NCCL INFO Channel 02/0 : 47[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722051:2722465 [6] NCCL INFO Channel 07/0 : 62[6] -> 126[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153567:153979 [5] NCCL INFO Channel 09/0 : 77[5] -> 93[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153566:153980 [4] NCCL INFO Channel 07/0 : 76[4] -> 92[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153565:153985 [3] NCCL INFO Channel 01/0 : 91[3] -> 107[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153564:153984 [2] NCCL INFO Channel 09/0 : 74[2] -> 90[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557272:1557687 [6] NCCL INFO Channel 10/0 : 70[6] -> 102[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235655:236069 [7] NCCL INFO Channel 07/0 : 95[7] -> 111[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272157:272572 [7] NCCL INFO Channel 06/0 : 103[7] -> 119[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272150:272576 [0] NCCL INFO Channel 12/0 : 96[0] -> 112[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815965:2816395 [2] NCCL INFO Channel 00/0 : 50[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227466:1227887 [3] NCCL INFO Channel 12/0 : 99[3] -> 83[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630208:1630650 [3] NCCL INFO Channel 05/0 : 91[3] -> 75[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636090:2636505 [6] NCCL INFO Channel 03/0 : 46[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636087:2636506 [3] NCCL INFO Channel 06/0 : 43[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-52:2722051:2722465 [6] NCCL INFO Channel 09/0 : 62[6] -> 126[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153562:153983 [0] NCCL INFO Channel 09/0 : 72[0] -> 88[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557270:1557688 [4] NCCL INFO Channel 10/0 : 68[4] -> 100[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235651:236068 [3] NCCL INFO Channel 01/0 : 91[3] -> 107[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272155:272574 [5] NCCL INFO Channel 08/0 : 101[5] -> 117[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815963:2816400 [0] NCCL INFO Channel 02/0 : 48[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227470:1227885 [7] NCCL INFO Channel 08/0 : 103[7] -> 87[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227468:1227883 [5] NCCL INFO Channel 04/0 : 101[5] -> 85[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630208:1630650 [3] NCCL INFO Channel 07/0 : 91[3] -> 75[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636089:2636508 [5] NCCL INFO Channel 08/0 : 45[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636086:2636507 [2] NCCL INFO Channel 00/0 : 42[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636085:2636509 [1] NCCL INFO Channel 07/0 : 41[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-52:2722051:2722465 [6] NCCL INFO Channel 11/0 : 62[6] -> 126[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153563:153986 [1] NCCL INFO Channel 03/0 : 73[1] -> 89[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557266:1557689 [0] NCCL INFO Channel 04/0 : 64[0] -> 96[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235648:236064 [0] NCCL INFO Channel 15/0 : 88[0] -> 104[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272152:272575 [2] NCCL INFO Channel 00/0 : 98[2] -> 114[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815967:2816399 [4] NCCL INFO Channel 06/0 : 52[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815969:2816398 [6] NCCL INFO Channel 02/0 : 54[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227465:1227881 [2] NCCL INFO Channel 08/0 : 98[2] -> 82[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227463:1227880 [0] NCCL INFO Channel 08/0 : 96[0] -> 80[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630212:1630652 [7] NCCL INFO Channel 01/0 : 95[7] -> 79[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636084:2636511 [0] NCCL INFO Channel 04/0 : 40[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636091:2636512 [7] NCCL INFO Channel 03/0 : 47[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722051:2722465 [6] NCCL INFO Channel 13/0 : 62[6] -> 126[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153569:153981 [7] NCCL INFO Channel 11/0 : 79[7] -> 95[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557272:1557687 [6] NCCL INFO Channel 12/0 : 70[6] -> 102[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235653:236067 [5] NCCL INFO Channel 09/0 : 93[5] -> 109[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272153:272573 [3] NCCL INFO Channel 00/0 : 99[3] -> 115[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815965:2816395 [2] NCCL INFO Channel 04/0 : 50[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227469:1227882 [6] NCCL INFO Channel 10/0 : 102[6] -> 86[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630210:1630653 [5] NCCL INFO Channel 01/0 : 93[5] -> 77[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630208:1630650 [3] NCCL INFO Channel 09/0 : 91[3] -> 75[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636088:2636510 [4] NCCL INFO Channel 01/0 : 44[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636090:2636505 [6] NCCL INFO Channel 04/0 : 46[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-52:2722050:2722471 [5] NCCL INFO Channel 01/0 : 61[5] -> 125[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153568:153982 [6] NCCL INFO Channel 11/0 : 78[6] -> 94[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557270:1557688 [4] NCCL INFO Channel 12/0 : 68[4] -> 100[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557271:1557690 [5] NCCL INFO Channel 00/0 : 69[5] -> 101[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235650:236065 [2] NCCL INFO Channel 01/0 : 90[2] -> 106[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235649:236071 [1] NCCL INFO Channel 03/0 : 89[1] -> 105[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235655:236069 [7] NCCL INFO Channel 09/0 : 95[7] -> 111[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272157:272572 [7] NCCL INFO Channel 08/0 : 103[7] -> 119[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815963:2816400 [0] NCCL INFO Channel 04/0 : 48[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227467:1227886 [4] NCCL INFO Channel 08/0 : 100[4] -> 84[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630212:1630652 [7] NCCL INFO Channel 03/0 : 95[7] -> 79[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636089:2636508 [5] NCCL INFO Channel 09/0 : 45[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-52:2722051:2722465 [6] NCCL INFO Channel 15/0 : 62[6] -> 126[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153565:153985 [3] NCCL INFO Channel 03/0 : 91[3] -> 107[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153564:153984 [2] NCCL INFO Channel 11/0 : 74[2] -> 90[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153566:153980 [4] NCCL INFO Channel 09/0 : 76[4] -> 92[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153567:153979 [5] NCCL INFO Channel 11/0 : 77[5] -> 93[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153562:153983 [0] NCCL INFO Channel 11/0 : 72[0] -> 88[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557272:1557687 [6] NCCL INFO Channel 14/0 : 70[6] -> 102[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557266:1557689 [0] NCCL INFO Channel 06/0 : 64[0] -> 96[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235651:236068 [3] NCCL INFO Channel 03/0 : 91[3] -> 107[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235653:236067 [5] NCCL INFO Channel 11/0 : 93[5] -> 109[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272155:272574 [5] NCCL INFO Channel 10/0 : 101[5] -> 117[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272156:272570 [6] NCCL INFO Channel 00/0 : 118[6] -> 102[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272152:272575 [2] NCCL INFO Channel 04/0 : 98[2] -> 114[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272154:272569 [4] NCCL INFO Channel 00/0 : 116[4] -> 100[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815967:2816399 [4] NCCL INFO Channel 08/0 : 52[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227466:1227887 [3] NCCL INFO Channel 14/0 : 99[3] -> 83[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630210:1630653 [5] NCCL INFO Channel 03/0 : 93[5] -> 77[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636087:2636506 [3] NCCL INFO Channel 07/0 : 43[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636085:2636509 [1] NCCL INFO Channel 08/0 : 41[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-52:2722050:2722471 [5] NCCL INFO Channel 03/0 : 61[5] -> 125[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153563:153986 [1] NCCL INFO Channel 05/0 : 73[1] -> 89[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557270:1557688 [4] NCCL INFO Channel 14/0 : 68[4] -> 100[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235655:236069 [7] NCCL INFO Channel 11/0 : 95[7] -> 111[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235650:236065 [2] NCCL INFO Channel 03/0 : 90[2] -> 106[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272150:272576 [0] NCCL INFO Channel 14/0 : 96[0] -> 112[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815969:2816398 [6] NCCL INFO Channel 04/0 : 54[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227470:1227885 [7] NCCL INFO Channel 10/0 : 103[7] -> 87[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227464:1227884 [1] NCCL INFO Channel 00/0 : 97[1] -> 81[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630208:1630650 [3] NCCL INFO Channel 13/0 : 91[3] -> 75[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636086:2636507 [2] NCCL INFO Channel 01/0 : 42[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636091:2636512 [7] NCCL INFO Channel 04/0 : 47[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722051:2722465 [6] NCCL INFO Channel 01/0 : 126[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153568:153982 [6] NCCL INFO Channel 13/0 : 78[6] -> 94[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153569:153981 [7] NCCL INFO Channel 13/0 : 79[7] -> 95[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557271:1557690 [5] NCCL INFO Channel 02/0 : 69[5] -> 101[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235649:236071 [1] NCCL INFO Channel 05/0 : 89[1] -> 105[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272153:272573 [3] NCCL INFO Channel 02/0 : 99[3] -> 115[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272157:272572 [7] NCCL INFO Channel 10/0 : 103[7] -> 119[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815968:2816394 [5] NCCL INFO Channel 00/0 : 53[5] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227468:1227883 [5] NCCL INFO Channel 06/0 : 101[5] -> 85[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630211:1630649 [6] NCCL INFO Channel 01/0 : 94[6] -> 78[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636084:2636511 [0] NCCL INFO Channel 05/0 : 40[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-52:2722050:2722471 [5] NCCL INFO Channel 05/0 : 61[5] -> 125[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153564:153984 [2] NCCL INFO Channel 13/0 : 74[2] -> 90[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153565:153985 [3] NCCL INFO Channel 05/0 : 91[3] -> 107[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557273:1557685 [7] NCCL INFO Channel 00/0 : 71[7] -> 103[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235653:236067 [5] NCCL INFO Channel 15/0 : 93[5] -> 109[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272155:272574 [5] NCCL INFO Channel 12/0 : 101[5] -> 117[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815963:2816400 [0] NCCL INFO Channel 06/0 : 48[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815965:2816395 [2] NCCL INFO Channel 06/0 : 50[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227465:1227881 [2] NCCL INFO Channel 10/0 : 98[2] -> 82[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630212:1630652 [7] NCCL INFO Channel 05/0 : 95[7] -> 79[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636088:2636510 [4] NCCL INFO Channel 02/0 : 44[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722051:2722465 [6] NCCL INFO Channel 03/0 : 126[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153567:153979 [5] NCCL INFO Channel 15/0 : 77[5] -> 93[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153566:153980 [4] NCCL INFO Channel 11/0 : 76[4] -> 92[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153562:153983 [0] NCCL INFO Channel 13/0 : 72[0] -> 88[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153563:153986 [1] NCCL INFO Channel 07/0 : 73[1] -> 89[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153569:153981 [7] NCCL INFO Channel 01/0 : 95[7] -> 111[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557266:1557689 [0] NCCL INFO Channel 08/0 : 64[0] -> 96[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235651:236068 [3] NCCL INFO Channel 05/0 : 91[3] -> 107[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272156:272570 [6] NCCL INFO Channel 02/0 : 118[6] -> 102[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815970:2816397 [7] NCCL INFO Channel 00/0 : 55[7] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227469:1227882 [6] NCCL INFO Channel 12/0 : 102[6] -> 86[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630210:1630653 [5] NCCL INFO Channel 05/0 : 93[5] -> 77[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636089:2636508 [5] NCCL INFO Channel 10/0 : 45[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-52:2722050:2722471 [5] NCCL INFO Channel 07/0 : 61[5] -> 125[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153568:153982 [6] NCCL INFO Channel 15/0 : 78[6] -> 94[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557271:1557690 [5] NCCL INFO Channel 04/0 : 69[5] -> 101[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235655:236069 [7] NCCL INFO Channel 13/0 : 95[7] -> 111[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272154:272569 [4] NCCL INFO Channel 02/0 : 116[4] -> 100[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815967:2816399 [4] NCCL INFO Channel 10/0 : 52[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227463:1227880 [0] NCCL INFO Channel 10/0 : 96[0] -> 80[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227467:1227886 [4] NCCL INFO Channel 10/0 : 100[4] -> 84[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630208:1630650 [3] NCCL INFO Channel 15/0 : 91[3] -> 75[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636090:2636505 [6] NCCL INFO Channel 05/0 : 46[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-52:2722051:2722465 [6] NCCL INFO Channel 05/0 : 126[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153564:153984 [2] NCCL INFO Channel 15/0 : 74[2] -> 90[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153565:153985 [3] NCCL INFO Channel 07/0 : 91[3] -> 107[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557273:1557685 [7] NCCL INFO Channel 02/0 : 71[7] -> 103[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235650:236065 [2] NCCL INFO Channel 05/0 : 90[2] -> 106[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272152:272575 [2] NCCL INFO Channel 06/0 : 98[2] -> 114[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272157:272572 [7] NCCL INFO Channel 12/0 : 103[7] -> 119[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272153:272573 [3] NCCL INFO Channel 04/0 : 99[3] -> 115[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815969:2816398 [6] NCCL INFO Channel 08/0 : 54[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227470:1227885 [7] NCCL INFO Channel 12/0 : 103[7] -> 87[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630207:1630648 [2] NCCL INFO Channel 01/0 : 90[2] -> 74[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630211:1630649 [6] NCCL INFO Channel 03/0 : 94[6] -> 78[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630212:1630652 [7] NCCL INFO Channel 07/0 : 95[7] -> 79[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630210:1630653 [5] NCCL INFO Channel 07/0 : 93[5] -> 77[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636085:2636509 [1] NCCL INFO Channel 10/0 : 41[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636087:2636506 [3] NCCL INFO Channel 08/0 : 43[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636091:2636512 [7] NCCL INFO Channel 05/0 : 47[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722046:2722466 [1] NCCL INFO Channel 01/0 : 57[1] -> 121[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153567:153979 [5] NCCL INFO Channel 01/0 : 93[5] -> 109[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557266:1557689 [0] NCCL INFO Channel 10/0 : 64[0] -> 96[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235649:236071 [1] NCCL INFO Channel 07/0 : 89[1] -> 105[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235651:236068 [3] NCCL INFO Channel 07/0 : 91[3] -> 107[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272155:272574 [5] NCCL INFO Channel 14/0 : 101[5] -> 117[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815968:2816394 [5] NCCL INFO Channel 02/0 : 53[5] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227464:1227884 [1] NCCL INFO Channel 02/0 : 97[1] -> 81[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227468:1227883 [5] NCCL INFO Channel 08/0 : 101[5] -> 85[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630207:1630648 [2] NCCL INFO Channel 03/0 : 90[2] -> 74[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636088:2636510 [4] NCCL INFO Channel 03/0 : 44[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636086:2636507 [2] NCCL INFO Channel 03/0 : 42[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636090:2636505 [6] NCCL INFO Channel 07/0 : 46[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636089:2636508 [5] NCCL INFO Channel 11/0 : 45[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636084:2636511 [0] NCCL INFO Channel 06/0 : 40[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-52:2722051:2722465 [6] NCCL INFO Channel 07/0 : 126[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-52:2722050:2722471 [5] NCCL INFO Channel 09/0 : 61[5] -> 125[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153562:153983 [0] NCCL INFO Channel 15/0 : 72[0] -> 88[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557271:1557690 [5] NCCL INFO Channel 06/0 : 69[5] -> 101[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235649:236071 [1] NCCL INFO Channel 11/0 : 89[1] -> 105[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235651:236068 [3] NCCL INFO Channel 09/0 : 91[3] -> 107[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272156:272570 [6] NCCL INFO Channel 04/0 : 118[6] -> 102[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815963:2816400 [0] NCCL INFO Channel 08/0 : 48[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815965:2816395 [2] NCCL INFO Channel 08/0 : 50[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227465:1227881 [2] NCCL INFO Channel 12/0 : 98[2] -> 82[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630211:1630649 [6] NCCL INFO Channel 05/0 : 94[6] -> 78[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630212:1630652 [7] NCCL INFO Channel 09/0 : 95[7] -> 79[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636085:2636509 [1] NCCL INFO Channel 11/0 : 41[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-52:2722046:2722466 [1] NCCL INFO Channel 03/0 : 57[1] -> 121[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153568:153982 [6] NCCL INFO Channel 01/0 : 94[6] -> 110[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153566:153980 [4] NCCL INFO Channel 13/0 : 76[4] -> 92[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557272:1557687 [6] NCCL INFO Channel 00/0 : 102[6] -> 70[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557273:1557685 [7] NCCL INFO Channel 04/0 : 71[7] -> 103[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235650:236065 [2] NCCL INFO Channel 07/0 : 90[2] -> 106[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272154:272569 [4] NCCL INFO Channel 06/0 : 116[4] -> 100[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815970:2816397 [7] NCCL INFO Channel 02/0 : 55[7] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227469:1227882 [6] NCCL INFO Channel 14/0 : 102[6] -> 86[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227463:1227880 [0] NCCL INFO Channel 12/0 : 96[0] -> 80[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630210:1630653 [5] NCCL INFO Channel 09/0 : 93[5] -> 77[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636091:2636512 [7] NCCL INFO Channel 06/0 : 47[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722051:2722465 [6] NCCL INFO Channel 09/0 : 126[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153569:153981 [7] NCCL INFO Channel 03/0 : 95[7] -> 111[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557266:1557689 [0] NCCL INFO Channel 12/0 : 64[0] -> 96[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235651:236068 [3] NCCL INFO Channel 13/0 : 91[3] -> 107[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272152:272575 [2] NCCL INFO Channel 08/0 : 98[2] -> 114[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272157:272572 [7] NCCL INFO Channel 14/0 : 103[7] -> 119[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815967:2816399 [4] NCCL INFO Channel 12/0 : 52[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815969:2816398 [6] NCCL INFO Channel 10/0 : 54[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227467:1227886 [4] NCCL INFO Channel 12/0 : 100[4] -> 84[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227470:1227885 [7] NCCL INFO Channel 14/0 : 103[7] -> 87[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630205:1630646 [0] NCCL INFO Channel 01/0 : 88[0] -> 72[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636087:2636506 [3] NCCL INFO Channel 09/0 : 43[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-52:2722050:2722471 [5] NCCL INFO Channel 11/0 : 61[5] -> 125[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-52:2722047:2722468 [2] NCCL INFO Channel 01/0 : 58[2] -> 122[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153563:153986 [1] NCCL INFO Channel 11/0 : 73[1] -> 89[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153564:153984 [2] NCCL INFO Channel 01/0 : 90[2] -> 106[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557271:1557690 [5] NCCL INFO Channel 08/0 : 69[5] -> 101[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235649:236071 [1] NCCL INFO Channel 13/0 : 89[1] -> 105[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272150:272576 [0] NCCL INFO Channel 02/0 : 112[0] -> 96[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272153:272573 [3] NCCL INFO Channel 06/0 : 99[3] -> 115[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815968:2816394 [5] NCCL INFO Channel 04/0 : 53[5] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227468:1227883 [5] NCCL INFO Channel 10/0 : 101[5] -> 85[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227464:1227884 [1] NCCL INFO Channel 04/0 : 97[1] -> 81[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630207:1630648 [2] NCCL INFO Channel 05/0 : 90[2] -> 74[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630209:1630651 [4] NCCL INFO Channel 01/0 : 92[4] -> 76[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636088:2636510 [4] NCCL INFO Channel 05/0 : 44[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722046:2722466 [1] NCCL INFO Channel 05/0 : 57[1] -> 121[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153565:153985 [3] NCCL INFO Channel 09/0 : 91[3] -> 107[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557270:1557688 [4] NCCL INFO Channel 00/0 : 100[4] -> 68[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235650:236065 [2] NCCL INFO Channel 09/0 : 90[2] -> 106[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272156:272570 [6] NCCL INFO Channel 08/0 : 118[6] -> 102[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815965:2816395 [2] NCCL INFO Channel 10/0 : 50[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815963:2816400 [0] NCCL INFO Channel 10/0 : 48[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227465:1227881 [2] NCCL INFO Channel 14/0 : 98[2] -> 82[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227467:1227886 [4] NCCL INFO Channel 14/0 : 100[4] -> 84[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227463:1227880 [0] NCCL INFO Channel 14/0 : 96[0] -> 80[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630206:1630647 [1] NCCL INFO Channel 01/0 : 89[1] -> 73[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636089:2636508 [5] NCCL INFO Channel 12/0 : 45[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-52:2722051:2722465 [6] NCCL INFO Channel 11/0 : 126[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153567:153979 [5] NCCL INFO Channel 03/0 : 93[5] -> 109[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557268:1557686 [2] NCCL INFO Channel 00/0 : 66[2] -> 98[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235649:236071 [1] NCCL INFO Channel 15/0 : 89[1] -> 105[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235651:236068 [3] NCCL INFO Channel 15/0 : 91[3] -> 107[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272154:272569 [4] NCCL INFO Channel 08/0 : 116[4] -> 100[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815970:2816397 [7] NCCL INFO Channel 04/0 : 55[7] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815966:2816396 [3] NCCL INFO Channel 00/0 : 51[3] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227468:1227883 [5] NCCL INFO Channel 12/0 : 101[5] -> 85[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227464:1227884 [1] NCCL INFO Channel 06/0 : 97[1] -> 81[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630211:1630649 [6] NCCL INFO Channel 07/0 : 94[6] -> 78[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636090:2636505 [6] NCCL INFO Channel 08/0 : 46[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-52:2722047:2722468 [2] NCCL INFO Channel 03/0 : 58[2] -> 122[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-52:2722046:2722466 [1] NCCL INFO Channel 07/0 : 57[1] -> 121[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-52:2722050:2722471 [5] NCCL INFO Channel 15/0 : 61[5] -> 125[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-52:2722051:2722465 [6] NCCL INFO Channel 13/0 : 126[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153562:153983 [0] NCCL INFO Channel 01/0 : 88[0] -> 104[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557272:1557687 [6] NCCL INFO Channel 02/0 : 102[6] -> 70[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235650:236065 [2] NCCL INFO Channel 11/0 : 90[2] -> 106[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272152:272575 [2] NCCL INFO Channel 10/0 : 98[2] -> 114[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272150:272576 [0] NCCL INFO Channel 04/0 : 112[0] -> 96[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272153:272573 [3] NCCL INFO Channel 08/0 : 99[3] -> 115[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815969:2816398 [6] NCCL INFO Channel 12/0 : 54[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227468:1227883 [5] NCCL INFO Channel 14/0 : 101[5] -> 85[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630210:1630653 [5] NCCL INFO Channel 11/0 : 93[5] -> 77[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636084:2636511 [0] NCCL INFO Channel 07/0 : 40[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636086:2636507 [2] NCCL INFO Channel 04/0 : 42[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636085:2636509 [1] NCCL INFO Channel 12/0 : 41[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-52:2722047:2722468 [2] NCCL INFO Channel 05/0 : 58[2] -> 122[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153568:153982 [6] NCCL INFO Channel 03/0 : 94[6] -> 110[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557266:1557689 [0] NCCL INFO Channel 14/0 : 64[0] -> 96[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557273:1557685 [7] NCCL INFO Channel 06/0 : 71[7] -> 103[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235650:236065 [2] NCCL INFO Channel 13/0 : 90[2] -> 106[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272156:272570 [6] NCCL INFO Channel 10/0 : 118[6] -> 102[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815964:2816401 [1] NCCL INFO Channel 00/0 : 49[1] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815967:2816399 [4] NCCL INFO Channel 14/0 : 52[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815968:2816394 [5] NCCL INFO Channel 06/0 : 53[5] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227464:1227884 [1] NCCL INFO Channel 08/0 : 97[1] -> 81[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630212:1630652 [7] NCCL INFO Channel 11/0 : 95[7] -> 79[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636091:2636512 [7] NCCL INFO Channel 08/0 : 47[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636087:2636506 [3] NCCL INFO Channel 10/0 : 43[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-52:2722051:2722465 [6] NCCL INFO Channel 15/0 : 126[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153569:153981 [7] NCCL INFO Channel 05/0 : 95[7] -> 111[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557270:1557688 [4] NCCL INFO Channel 02/0 : 100[4] -> 68[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557271:1557690 [5] NCCL INFO Channel 10/0 : 69[5] -> 101[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235650:236065 [2] NCCL INFO Channel 15/0 : 90[2] -> 106[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272154:272569 [4] NCCL INFO Channel 10/0 : 116[4] -> 100[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815965:2816395 [2] NCCL INFO Channel 12/0 : 50[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227464:1227884 [1] NCCL INFO Channel 10/0 : 97[1] -> 81[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630205:1630646 [0] NCCL INFO Channel 03/0 : 88[0] -> 72[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636088:2636510 [4] NCCL INFO Channel 06/0 : 44[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722046:2722466 [1] NCCL INFO Channel 11/0 : 57[1] -> 121[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153563:153986 [1] NCCL INFO Channel 13/0 : 73[1] -> 89[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153564:153984 [2] NCCL INFO Channel 03/0 : 90[2] -> 106[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153566:153980 [4] NCCL INFO Channel 15/0 : 76[4] -> 92[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153565:153985 [3] NCCL INFO Channel 13/0 : 91[3] -> 107[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557268:1557686 [2] NCCL INFO Channel 04/0 : 66[2] -> 98[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557272:1557687 [6] NCCL INFO Channel 04/0 : 102[6] -> 70[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235651:236068 [3] NCCL INFO Channel 01/0 : 107[3] -> 91[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272155:272574 [5] NCCL INFO Channel 00/0 : 117[5] -> 101[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815963:2816400 [0] NCCL INFO Channel 12/0 : 48[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227464:1227884 [1] NCCL INFO Channel 12/0 : 97[1] -> 81[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630207:1630648 [2] NCCL INFO Channel 07/0 : 90[2] -> 74[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630209:1630651 [4] NCCL INFO Channel 03/0 : 92[4] -> 76[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636089:2636508 [5] NCCL INFO Channel 14/0 : 45[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636090:2636505 [6] NCCL INFO Channel 09/0 : 46[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-52:2722050:2722471 [5] NCCL INFO Channel 01/0 : 125[5] -> 61[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153567:153979 [5] NCCL INFO Channel 05/0 : 93[5] -> 109[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153562:153983 [0] NCCL INFO Channel 03/0 : 88[0] -> 104[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557273:1557685 [7] NCCL INFO Channel 08/0 : 71[7] -> 103[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235651:236068 [3] NCCL INFO Channel 03/0 : 107[3] -> 91[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272150:272576 [0] NCCL INFO Channel 06/0 : 112[0] -> 96[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815970:2816397 [7] NCCL INFO Channel 06/0 : 55[7] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815966:2816396 [3] NCCL INFO Channel 02/0 : 51[3] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227464:1227884 [1] NCCL INFO Channel 14/0 : 97[1] -> 81[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630206:1630647 [1] NCCL INFO Channel 03/0 : 89[1] -> 73[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636086:2636507 [2] NCCL INFO Channel 05/0 : 42[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636085:2636509 [1] NCCL INFO Channel 13/0 : 41[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-52:2722047:2722468 [2] NCCL INFO Channel 07/0 : 58[2] -> 122[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153568:153982 [6] NCCL INFO Channel 05/0 : 94[6] -> 110[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557271:1557690 [5] NCCL INFO Channel 12/0 : 69[5] -> 101[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557270:1557688 [4] NCCL INFO Channel 06/0 : 100[4] -> 68[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235651:236068 [3] NCCL INFO Channel 05/0 : 107[3] -> 91[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272152:272575 [2] NCCL INFO Channel 12/0 : 98[2] -> 114[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815969:2816398 [6] NCCL INFO Channel 14/0 : 54[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630210:1630653 [5] NCCL INFO Channel 15/0 : 93[5] -> 77[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630211:1630649 [6] NCCL INFO Channel 09/0 : 94[6] -> 78[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636084:2636511 [0] NCCL INFO Channel 09/0 : 40[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-52:2722046:2722466 [1] NCCL INFO Channel 13/0 : 57[1] -> 121[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-52:2722050:2722471 [5] NCCL INFO Channel 03/0 : 125[5] -> 61[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153569:153981 [7] NCCL INFO Channel 07/0 : 95[7] -> 111[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557272:1557687 [6] NCCL INFO Channel 08/0 : 102[6] -> 70[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557268:1557686 [2] NCCL INFO Channel 06/0 : 66[2] -> 98[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235651:236068 [3] NCCL INFO Channel 07/0 : 107[3] -> 91[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272156:272570 [6] NCCL INFO Channel 12/0 : 118[6] -> 102[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272153:272573 [3] NCCL INFO Channel 10/0 : 99[3] -> 115[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815964:2816401 [1] NCCL INFO Channel 02/0 : 49[1] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630212:1630652 [7] NCCL INFO Channel 13/0 : 95[7] -> 79[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636091:2636512 [7] NCCL INFO Channel 09/0 : 47[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722047:2722468 [2] NCCL INFO Channel 09/0 : 58[2] -> 122[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153564:153984 [2] NCCL INFO Channel 05/0 : 90[2] -> 106[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557269:1557684 [3] NCCL INFO Channel 00/0 : 67[3] -> 99[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235651:236068 [3] NCCL INFO Channel 09/0 : 107[3] -> 91[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272154:272569 [4] NCCL INFO Channel 12/0 : 116[4] -> 100[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815965:2816395 [2] NCCL INFO Channel 14/0 : 50[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630205:1630646 [0] NCCL INFO Channel 05/0 : 88[0] -> 72[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636087:2636506 [3] NCCL INFO Channel 12/0 : 43[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636088:2636510 [4] NCCL INFO Channel 07/0 : 44[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636089:2636508 [5] NCCL INFO Channel 15/0 : 45[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636090:2636505 [6] NCCL INFO Channel 10/0 : 46[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-52:2722046:2722466 [1] NCCL INFO Channel 15/0 : 57[1] -> 121[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-52:2722050:2722471 [5] NCCL INFO Channel 05/0 : 125[5] -> 61[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153565:153985 [3] NCCL INFO Channel 15/0 : 91[3] -> 107[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557273:1557685 [7] NCCL INFO Channel 10/0 : 71[7] -> 103[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235655:236069 [7] NCCL INFO Channel 01/0 : 111[7] -> 95[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272155:272574 [5] NCCL INFO Channel 02/0 : 117[5] -> 101[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815963:2816400 [0] NCCL INFO Channel 14/0 : 48[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815968:2816394 [5] NCCL INFO Channel 08/0 : 53[5] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630207:1630648 [2] NCCL INFO Channel 09/0 : 90[2] -> 74[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636086:2636507 [2] NCCL INFO Channel 06/0 : 42[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636085:2636509 [1] NCCL INFO Channel 14/0 : 41[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-52:2722047:2722468 [2] NCCL INFO Channel 11/0 : 58[2] -> 122[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153563:153986 [1] NCCL INFO Channel 15/0 : 73[1] -> 89[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153566:153980 [4] NCCL INFO Channel 01/0 : 92[4] -> 108[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557270:1557688 [4] NCCL INFO Channel 08/0 : 100[4] -> 68[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557271:1557690 [5] NCCL INFO Channel 14/0 : 69[5] -> 101[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235651:236068 [3] NCCL INFO Channel 13/0 : 107[3] -> 91[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272157:272572 [7] NCCL INFO Channel 00/0 : 119[7] -> 103[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815966:2816396 [3] NCCL INFO Channel 04/0 : 51[3] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815970:2816397 [7] NCCL INFO Channel 08/0 : 55[7] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630209:1630651 [4] NCCL INFO Channel 05/0 : 92[4] -> 76[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636091:2636512 [7] NCCL INFO Channel 10/0 : 47[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636084:2636511 [0] NCCL INFO Channel 10/0 : 40[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-52:2722046:2722466 [1] NCCL INFO Channel 01/0 : 121[1] -> 57[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-52:2722050:2722471 [5] NCCL INFO Channel 07/0 : 125[5] -> 61[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153562:153983 [0] NCCL INFO Channel 05/0 : 88[0] -> 104[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153567:153979 [5] NCCL INFO Channel 07/0 : 93[5] -> 109[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153568:153982 [6] NCCL INFO Channel 07/0 : 94[6] -> 110[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557266:1557689 [0] NCCL INFO Channel 02/0 : 96[0] -> 64[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235653:236067 [5] NCCL INFO Channel 01/0 : 109[5] -> 93[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272150:272576 [0] NCCL INFO Channel 08/0 : 112[0] -> 96[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815964:2816401 [1] NCCL INFO Channel 04/0 : 49[1] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630206:1630647 [1] NCCL INFO Channel 05/0 : 89[1] -> 73[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630211:1630649 [6] NCCL INFO Channel 11/0 : 94[6] -> 78[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636089:2636508 [5] NCCL INFO Channel 00/0 : 53[5] -> 61[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636088:2636510 [4] NCCL INFO Channel 08/0 : 44[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722047:2722468 [2] NCCL INFO Channel 13/0 : 58[2] -> 122[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153569:153981 [7] NCCL INFO Channel 09/0 : 95[7] -> 111[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557272:1557687 [6] NCCL INFO Channel 10/0 : 102[6] -> 70[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235655:236069 [7] NCCL INFO Channel 03/0 : 111[7] -> 95[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272152:272575 [2] NCCL INFO Channel 14/0 : 98[2] -> 114[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815968:2816394 [5] NCCL INFO Channel 10/0 : 53[5] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630205:1630646 [0] NCCL INFO Channel 07/0 : 88[0] -> 72[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630207:1630648 [2] NCCL INFO Channel 11/0 : 90[2] -> 74[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636090:2636505 [6] NCCL INFO Channel 11/0 : 46[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-52:2722050:2722471 [5] NCCL INFO Channel 09/0 : 125[5] -> 61[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-52:2722046:2722466 [1] NCCL INFO Channel 03/0 : 121[1] -> 57[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153564:153984 [2] NCCL INFO Channel 07/0 : 90[2] -> 106[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557269:1557684 [3] NCCL INFO Channel 02/0 : 67[3] -> 99[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557268:1557686 [2] NCCL INFO Channel 08/0 : 66[2] -> 98[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235651:236068 [3] NCCL INFO Channel 15/0 : 107[3] -> 91[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272156:272570 [6] NCCL INFO Channel 14/0 : 118[6] -> 102[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815970:2816397 [7] NCCL INFO Channel 10/0 : 55[7] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815966:2816396 [3] NCCL INFO Channel 06/0 : 51[3] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630211:1630649 [6] NCCL INFO Channel 13/0 : 94[6] -> 78[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630209:1630651 [4] NCCL INFO Channel 07/0 : 92[4] -> 76[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636087:2636506 [3] NCCL INFO Channel 13/0 : 43[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636085:2636509 [1] NCCL INFO Channel 15/0 : 41[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-52:2722047:2722468 [2] NCCL INFO Channel 15/0 : 58[2] -> 122[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153566:153980 [4] NCCL INFO Channel 03/0 : 92[4] -> 108[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557273:1557685 [7] NCCL INFO Channel 12/0 : 71[7] -> 103[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557270:1557688 [4] NCCL INFO Channel 10/0 : 100[4] -> 68[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235653:236067 [5] NCCL INFO Channel 03/0 : 109[5] -> 93[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272153:272573 [3] NCCL INFO Channel 12/0 : 99[3] -> 115[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815964:2816401 [1] NCCL INFO Channel 06/0 : 49[1] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630206:1630647 [1] NCCL INFO Channel 07/0 : 89[1] -> 73[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636086:2636507 [2] NCCL INFO Channel 07/0 : 42[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-52:2722050:2722471 [5] NCCL INFO Channel 11/0 : 125[5] -> 61[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-52:2722046:2722466 [1] NCCL INFO Channel 05/0 : 121[1] -> 57[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153562:153983 [0] NCCL INFO Channel 07/0 : 88[0] -> 104[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153567:153979 [5] NCCL INFO Channel 09/0 : 93[5] -> 109[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153568:153982 [6] NCCL INFO Channel 09/0 : 94[6] -> 110[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153563:153986 [1] NCCL INFO Channel 01/0 : 89[1] -> 105[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153569:153981 [7] NCCL INFO Channel 11/0 : 95[7] -> 111[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153564:153984 [2] NCCL INFO Channel 09/0 : 90[2] -> 106[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557266:1557689 [0] NCCL INFO Channel 04/0 : 96[0] -> 64[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235655:236069 [7] NCCL INFO Channel 05/0 : 111[7] -> 95[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272154:272569 [4] NCCL INFO Channel 14/0 : 116[4] -> 100[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815968:2816394 [5] NCCL INFO Channel 12/0 : 53[5] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630205:1630646 [0] NCCL INFO Channel 09/0 : 88[0] -> 72[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630207:1630648 [2] NCCL INFO Channel 13/0 : 90[2] -> 74[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636091:2636512 [7] NCCL INFO Channel 11/0 : 47[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636084:2636511 [0] NCCL INFO Channel 11/0 : 40[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-52:2722047:2722468 [2] NCCL INFO Channel 01/0 : 122[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153566:153980 [4] NCCL INFO Channel 05/0 : 92[4] -> 108[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557272:1557687 [6] NCCL INFO Channel 12/0 : 102[6] -> 70[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235654:236070 [6] NCCL INFO Channel 01/0 : 110[6] -> 94[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272155:272574 [5] NCCL INFO Channel 04/0 : 117[5] -> 101[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815970:2816397 [7] NCCL INFO Channel 12/0 : 55[7] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630211:1630649 [6] NCCL INFO Channel 15/0 : 94[6] -> 78[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636089:2636508 [5] NCCL INFO Channel 02/0 : 53[5] -> 61[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-52:2722049:2722467 [4] NCCL INFO Channel 01/0 : 60[4] -> 124[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153567:153979 [5] NCCL INFO Channel 11/0 : 93[5] -> 109[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557268:1557686 [2] NCCL INFO Channel 10/0 : 66[2] -> 98[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557270:1557688 [4] NCCL INFO Channel 12/0 : 100[4] -> 68[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557273:1557685 [7] NCCL INFO Channel 14/0 : 71[7] -> 103[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235650:236065 [2] NCCL INFO Channel 01/0 : 106[2] -> 90[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272157:272572 [7] NCCL INFO Channel 02/0 : 119[7] -> 103[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272150:272576 [0] NCCL INFO Channel 10/0 : 112[0] -> 96[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815966:2816396 [3] NCCL INFO Channel 08/0 : 51[3] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630209:1630651 [4] NCCL INFO Channel 09/0 : 92[4] -> 76[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636090:2636505 [6] NCCL INFO Channel 12/0 : 46[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636088:2636510 [4] NCCL INFO Channel 09/0 : 44[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722046:2722466 [1] NCCL INFO Channel 07/0 : 121[1] -> 57[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153563:153986 [1] NCCL INFO Channel 03/0 : 89[1] -> 105[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153568:153982 [6] NCCL INFO Channel 11/0 : 94[6] -> 110[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557269:1557684 [3] NCCL INFO Channel 04/0 : 67[3] -> 99[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557266:1557689 [0] NCCL INFO Channel 06/0 : 96[0] -> 64[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235653:236067 [5] NCCL INFO Channel 05/0 : 109[5] -> 93[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272153:272573 [3] NCCL INFO Channel 14/0 : 99[3] -> 115[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815964:2816401 [1] NCCL INFO Channel 08/0 : 49[1] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630207:1630648 [2] NCCL INFO Channel 15/0 : 90[2] -> 74[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630205:1630646 [0] NCCL INFO Channel 11/0 : 88[0] -> 72[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636087:2636506 [3] NCCL INFO Channel 14/0 : 43[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-52:2722050:2722471 [5] NCCL INFO Channel 15/0 : 125[5] -> 61[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153564:153984 [2] NCCL INFO Channel 11/0 : 90[2] -> 106[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153569:153981 [7] NCCL INFO Channel 13/0 : 95[7] -> 111[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153562:153983 [0] NCCL INFO Channel 09/0 : 88[0] -> 104[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153566:153980 [4] NCCL INFO Channel 07/0 : 92[4] -> 108[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557272:1557687 [6] NCCL INFO Channel 14/0 : 102[6] -> 70[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235655:236069 [7] NCCL INFO Channel 07/0 : 111[7] -> 95[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272151:272571 [1] NCCL INFO Channel 00/0 : 97[1] -> 113[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272155:272574 [5] NCCL INFO Channel 06/0 : 117[5] -> 101[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815968:2816394 [5] NCCL INFO Channel 14/0 : 53[5] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630206:1630647 [1] NCCL INFO Channel 11/0 : 89[1] -> 73[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636085:2636509 [1] NCCL INFO Channel 00/0 : 49[1] -> 57[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-52:2722047:2722468 [2] NCCL INFO Channel 03/0 : 122[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153565:153985 [3] NCCL INFO Channel 01/0 : 59[3] -> 91[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153567:153979 [5] NCCL INFO Channel 15/0 : 93[5] -> 109[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557268:1557686 [2] NCCL INFO Channel 12/0 : 66[2] -> 98[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557271:1557690 [5] NCCL INFO Channel 00/0 : 101[5] -> 69[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557270:1557688 [4] NCCL INFO Channel 14/0 : 100[4] -> 68[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235654:236070 [6] NCCL INFO Channel 03/0 : 110[6] -> 94[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272157:272572 [7] NCCL INFO Channel 04/0 : 119[7] -> 103[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272150:272576 [0] NCCL INFO Channel 12/0 : 112[0] -> 96[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815970:2816397 [7] NCCL INFO Channel 14/0 : 55[7] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630209:1630651 [4] NCCL INFO Channel 11/0 : 92[4] -> 76[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636091:2636512 [7] NCCL INFO Channel 12/0 : 47[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722049:2722467 [4] NCCL INFO Channel 03/0 : 60[4] -> 124[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153568:153982 [6] NCCL INFO Channel 13/0 : 94[6] -> 110[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557266:1557689 [0] NCCL INFO Channel 08/0 : 96[0] -> 64[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557269:1557684 [3] NCCL INFO Channel 06/0 : 67[3] -> 99[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235653:236067 [5] NCCL INFO Channel 07/0 : 109[5] -> 93[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272152:272575 [2] NCCL INFO Channel 00/0 : 114[2] -> 98[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815966:2816396 [3] NCCL INFO Channel 10/0 : 51[3] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630205:1630646 [0] NCCL INFO Channel 13/0 : 88[0] -> 72[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630206:1630647 [1] NCCL INFO Channel 13/0 : 89[1] -> 73[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636089:2636508 [5] NCCL INFO Channel 04/0 : 53[5] -> 61[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-52:2722046:2722466 [1] NCCL INFO Channel 11/0 : 121[1] -> 57[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153563:153986 [1] NCCL INFO Channel 05/0 : 89[1] -> 105[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557271:1557690 [5] NCCL INFO Channel 02/0 : 101[5] -> 69[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557268:1557686 [2] NCCL INFO Channel 14/0 : 66[2] -> 98[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235655:236069 [7] NCCL INFO Channel 09/0 : 111[7] -> 95[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272151:272571 [1] NCCL INFO Channel 02/0 : 97[1] -> 113[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815964:2816401 [1] NCCL INFO Channel 10/0 : 49[1] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636084:2636511 [0] NCCL INFO Channel 12/0 : 40[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-52:2722047:2722468 [2] NCCL INFO Channel 05/0 : 122[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153564:153984 [2] NCCL INFO Channel 13/0 : 90[2] -> 106[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153562:153983 [0] NCCL INFO Channel 11/0 : 88[0] -> 104[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557266:1557689 [0] NCCL INFO Channel 10/0 : 96[0] -> 64[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235650:236065 [2] NCCL INFO Channel 03/0 : 106[2] -> 90[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272155:272574 [5] NCCL INFO Channel 08/0 : 117[5] -> 101[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815966:2816396 [3] NCCL INFO Channel 12/0 : 51[3] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636086:2636507 [2] NCCL INFO Channel 08/0 : 42[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636090:2636505 [6] NCCL INFO Channel 13/0 : 46[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636088:2636510 [4] NCCL INFO Channel 10/0 : 44[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722049:2722467 [4] NCCL INFO Channel 05/0 : 60[4] -> 124[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153566:153980 [4] NCCL INFO Channel 09/0 : 92[4] -> 108[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557273:1557685 [7] NCCL INFO Channel 00/0 : 103[7] -> 71[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557269:1557684 [3] NCCL INFO Channel 08/0 : 67[3] -> 99[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235654:236070 [6] NCCL INFO Channel 05/0 : 110[6] -> 94[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272150:272576 [0] NCCL INFO Channel 14/0 : 112[0] -> 96[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272157:272572 [7] NCCL INFO Channel 06/0 : 119[7] -> 103[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815964:2816401 [1] NCCL INFO Channel 12/0 : 49[1] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636087:2636506 [3] NCCL INFO Channel 15/0 : 43[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636085:2636509 [1] NCCL INFO Channel 02/0 : 49[1] -> 57[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-52:2722047:2722468 [2] NCCL INFO Channel 07/0 : 122[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153565:153985 [3] NCCL INFO Channel 03/0 : 59[3] -> 91[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557271:1557690 [5] NCCL INFO Channel 04/0 : 101[5] -> 69[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235653:236067 [5] NCCL INFO Channel 09/0 : 109[5] -> 93[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272152:272575 [2] NCCL INFO Channel 04/0 : 114[2] -> 98[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815966:2816396 [3] NCCL INFO Channel 14/0 : 51[3] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636089:2636508 [5] NCCL INFO Channel 06/0 : 53[5] -> 61[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636091:2636512 [7] NCCL INFO Channel 13/0 : 47[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722049:2722467 [4] NCCL INFO Channel 07/0 : 60[4] -> 124[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153568:153982 [6] NCCL INFO Channel 15/0 : 94[6] -> 110[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557266:1557689 [0] NCCL INFO Channel 12/0 : 96[0] -> 64[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235655:236069 [7] NCCL INFO Channel 11/0 : 111[7] -> 95[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272155:272574 [5] NCCL INFO Channel 10/0 : 117[5] -> 101[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272151:272571 [1] NCCL INFO Channel 04/0 : 97[1] -> 113[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815964:2816401 [1] NCCL INFO Channel 14/0 : 49[1] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-52:2722048:2722469 [3] NCCL INFO Channel 01/0 : 59[3] -> 123[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153563:153986 [1] NCCL INFO Channel 07/0 : 89[1] -> 105[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557267:1557683 [1] NCCL INFO Channel 00/0 : 65[1] -> 97[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235654:236070 [6] NCCL INFO Channel 07/0 : 110[6] -> 94[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272153:272573 [3] NCCL INFO Channel 00/0 : 115[3] -> 99[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272157:272572 [7] NCCL INFO Channel 08/0 : 119[7] -> 103[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722047:2722468 [2] NCCL INFO Channel 09/0 : 122[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153564:153984 [2] NCCL INFO Channel 15/0 : 90[2] -> 106[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557273:1557685 [7] NCCL INFO Channel 02/0 : 103[7] -> 71[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235650:236065 [2] NCCL INFO Channel 05/0 : 106[2] -> 90[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235648:236064 [0] NCCL INFO Channel 01/0 : 104[0] -> 88[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272152:272575 [2] NCCL INFO Channel 06/0 : 114[2] -> 98[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-52:2722045:2722470 [0] NCCL INFO Channel 01/0 : 56[0] -> 120[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153566:153980 [4] NCCL INFO Channel 11/0 : 92[4] -> 108[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153565:153985 [3] NCCL INFO Channel 05/0 : 59[3] -> 91[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557269:1557684 [3] NCCL INFO Channel 10/0 : 67[3] -> 99[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557271:1557690 [5] NCCL INFO Channel 06/0 : 101[5] -> 69[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235649:236071 [1] NCCL INFO Channel 01/0 : 105[1] -> 89[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235652:236066 [4] NCCL INFO Channel 01/0 : 108[4] -> 92[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235653:236067 [5] NCCL INFO Channel 11/0 : 109[5] -> 93[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272155:272574 [5] NCCL INFO Channel 12/0 : 117[5] -> 101[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272151:272571 [1] NCCL INFO Channel 06/0 : 97[1] -> 113[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636084:2636511 [0] NCCL INFO Channel 13/0 : 40[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-52:2722052:2722464 [7] NCCL INFO Channel 01/0 : 63[7] -> 127[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722049:2722467 [4] NCCL INFO Channel 09/0 : 60[4] -> 124[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153562:153983 [0] NCCL INFO Channel 13/0 : 88[0] -> 104[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153563:153986 [1] NCCL INFO Channel 11/0 : 89[1] -> 105[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557267:1557683 [1] NCCL INFO Channel 02/0 : 65[1] -> 97[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235655:236069 [7] NCCL INFO Channel 13/0 : 111[7] -> 95[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272153:272573 [3] NCCL INFO Channel 02/0 : 115[3] -> 99[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272157:272572 [7] NCCL INFO Channel 10/0 : 119[7] -> 103[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636088:2636510 [4] NCCL INFO Channel 11/0 : 44[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722048:2722469 [3] NCCL INFO Channel 03/0 : 59[3] -> 123[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153566:153980 [4] NCCL INFO Channel 13/0 : 92[4] -> 108[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153565:153985 [3] NCCL INFO Channel 07/0 : 59[3] -> 91[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557268:1557686 [2] NCCL INFO Channel 00/0 : 98[2] -> 66[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557273:1557685 [7] NCCL INFO Channel 04/0 : 103[7] -> 71[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235654:236070 [6] NCCL INFO Channel 09/0 : 110[6] -> 94[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272152:272575 [2] NCCL INFO Channel 08/0 : 114[2] -> 98[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815968:2816394 [5] NCCL INFO Channel 01/0 : 29[5] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636090:2636505 [6] NCCL INFO Channel 15/0 : 46[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-52:2722047:2722468 [2] NCCL INFO Channel 11/0 : 122[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153563:153986 [1] NCCL INFO Channel 13/0 : 89[1] -> 105[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557269:1557684 [3] NCCL INFO Channel 12/0 : 67[3] -> 99[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557271:1557690 [5] NCCL INFO Channel 08/0 : 101[5] -> 69[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557267:1557683 [1] NCCL INFO Channel 04/0 : 65[1] -> 97[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235650:236065 [2] NCCL INFO Channel 07/0 : 106[2] -> 90[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235648:236064 [0] NCCL INFO Channel 03/0 : 104[0] -> 88[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272155:272574 [5] NCCL INFO Channel 14/0 : 117[5] -> 101[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815968:2816394 [5] NCCL INFO Channel 03/0 : 29[5] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636086:2636507 [2] NCCL INFO Channel 09/0 : 42[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-52:2722049:2722467 [4] NCCL INFO Channel 11/0 : 60[4] -> 124[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153562:153983 [0] NCCL INFO Channel 15/0 : 88[0] -> 104[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153569:153981 [7] NCCL INFO Channel 01/0 : 63[7] -> 95[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557268:1557686 [2] NCCL INFO Channel 04/0 : 98[2] -> 66[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235653:236067 [5] NCCL INFO Channel 15/0 : 109[5] -> 93[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235652:236066 [4] NCCL INFO Channel 03/0 : 108[4] -> 92[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235649:236071 [1] NCCL INFO Channel 03/0 : 105[1] -> 89[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235654:236070 [6] NCCL INFO Channel 11/0 : 110[6] -> 94[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272151:272571 [1] NCCL INFO Channel 08/0 : 97[1] -> 113[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636089:2636508 [5] NCCL INFO Channel 08/0 : 53[5] -> 61[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636087:2636506 [3] NCCL INFO Channel 00/0 : 51[3] -> 59[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636085:2636509 [1] NCCL INFO Channel 04/0 : 49[1] -> 57[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636091:2636512 [7] NCCL INFO Channel 14/0 : 47[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722045:2722470 [0] NCCL INFO Channel 03/0 : 56[0] -> 120[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-52:2722052:2722464 [7] NCCL INFO Channel 03/0 : 63[7] -> 127[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153566:153980 [4] NCCL INFO Channel 15/0 : 92[4] -> 108[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557273:1557685 [7] NCCL INFO Channel 06/0 : 103[7] -> 71[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235650:236065 [2] NCCL INFO Channel 09/0 : 106[2] -> 90[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235648:236064 [0] NCCL INFO Channel 05/0 : 104[0] -> 88[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272153:272573 [3] NCCL INFO Channel 04/0 : 115[3] -> 99[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272157:272572 [7] NCCL INFO Channel 12/0 : 119[7] -> 103[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636084:2636511 [0] NCCL INFO Channel 14/0 : 40[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-52:2722047:2722468 [2] NCCL INFO Channel 13/0 : 122[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153565:153985 [3] NCCL INFO Channel 09/0 : 59[3] -> 91[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235652:236066 [4] NCCL INFO Channel 05/0 : 108[4] -> 92[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815968:2816394 [5] NCCL INFO Channel 05/0 : 29[5] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636088:2636510 [4] NCCL INFO Channel 13/0 : 44[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722048:2722469 [3] NCCL INFO Channel 05/0 : 59[3] -> 123[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153563:153986 [1] NCCL INFO Channel 15/0 : 89[1] -> 105[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153567:153979 [5] NCCL INFO Channel 01/0 : 61[5] -> 93[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235649:236071 [1] NCCL INFO Channel 05/0 : 105[1] -> 89[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235654:236070 [6] NCCL INFO Channel 13/0 : 110[6] -> 94[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636090:2636505 [6] NCCL INFO Channel 00/0 : 54[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-52:2722049:2722467 [4] NCCL INFO Channel 13/0 : 60[4] -> 124[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153569:153981 [7] NCCL INFO Channel 03/0 : 63[7] -> 95[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235648:236064 [0] NCCL INFO Channel 07/0 : 104[0] -> 88[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235650:236065 [2] NCCL INFO Channel 11/0 : 106[2] -> 90[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636089:2636508 [5] NCCL INFO Channel 10/0 : 53[5] -> 61[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-52:2722047:2722468 [2] NCCL INFO Channel 15/0 : 122[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-52:2722045:2722470 [0] NCCL INFO Channel 05/0 : 56[0] -> 120[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153565:153985 [3] NCCL INFO Channel 13/0 : 59[3] -> 91[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557269:1557684 [3] NCCL INFO Channel 14/0 : 67[3] -> 99[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557267:1557683 [1] NCCL INFO Channel 06/0 : 65[1] -> 97[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557271:1557690 [5] NCCL INFO Channel 10/0 : 101[5] -> 69[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272152:272575 [2] NCCL INFO Channel 10/0 : 114[2] -> 98[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815968:2816394 [5] NCCL INFO Channel 07/0 : 29[5] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636086:2636507 [2] NCCL INFO Channel 11/0 : 42[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-52:2722048:2722469 [3] NCCL INFO Channel 07/0 : 59[3] -> 123[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-52:2722052:2722464 [7] NCCL INFO Channel 05/0 : 63[7] -> 127[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722049:2722467 [4] NCCL INFO Channel 15/0 : 60[4] -> 124[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153569:153981 [7] NCCL INFO Channel 05/0 : 63[7] -> 95[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153567:153979 [5] NCCL INFO Channel 03/0 : 61[5] -> 93[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557268:1557686 [2] NCCL INFO Channel 06/0 : 98[2] -> 66[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557273:1557685 [7] NCCL INFO Channel 08/0 : 103[7] -> 71[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272151:272571 [1] NCCL INFO Channel 10/0 : 97[1] -> 113[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636085:2636509 [1] NCCL INFO Channel 06/0 : 49[1] -> 57[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636091:2636512 [7] NCCL INFO Channel 00/0 : 55[7] -> 63[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153564:153984 [2] NCCL INFO Channel 01/0 : 58[2] -> 90[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153568:153982 [6] NCCL INFO Channel 01/0 : 62[6] -> 94[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557266:1557689 [0] NCCL INFO Channel 14/0 : 96[0] -> 64[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235652:236066 [4] NCCL INFO Channel 07/0 : 108[4] -> 92[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272153:272573 [3] NCCL INFO Channel 06/0 : 115[3] -> 99[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272157:272572 [7] NCCL INFO Channel 14/0 : 119[7] -> 103[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636087:2636506 [3] NCCL INFO Channel 02/0 : 51[3] -> 59[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153565:153985 [3] NCCL INFO Channel 15/0 : 59[3] -> 91[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557267:1557683 [1] NCCL INFO Channel 08/0 : 65[1] -> 97[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557271:1557690 [5] NCCL INFO Channel 12/0 : 101[5] -> 69[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235654:236070 [6] NCCL INFO Channel 15/0 : 110[6] -> 94[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235649:236071 [1] NCCL INFO Channel 07/0 : 105[1] -> 89[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272152:272575 [2] NCCL INFO Channel 12/0 : 114[2] -> 98[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636084:2636511 [0] NCCL INFO Channel 15/0 : 40[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636088:2636510 [4] NCCL INFO Channel 14/0 : 44[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636090:2636505 [6] NCCL INFO Channel 02/0 : 54[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153569:153981 [7] NCCL INFO Channel 07/0 : 63[7] -> 95[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153567:153979 [5] NCCL INFO Channel 05/0 : 61[5] -> 93[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557273:1557685 [7] NCCL INFO Channel 10/0 : 103[7] -> 71[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235648:236064 [0] NCCL INFO Channel 09/0 : 104[0] -> 88[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235650:236065 [2] NCCL INFO Channel 13/0 : 106[2] -> 90[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272151:272571 [1] NCCL INFO Channel 12/0 : 97[1] -> 113[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815968:2816394 [5] NCCL INFO Channel 09/0 : 29[5] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636089:2636508 [5] NCCL INFO Channel 12/0 : 53[5] -> 61[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153564:153984 [2] NCCL INFO Channel 03/0 : 58[2] -> 90[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153568:153982 [6] NCCL INFO Channel 03/0 : 62[6] -> 94[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557268:1557686 [2] NCCL INFO Channel 08/0 : 98[2] -> 66[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235652:236066 [4] NCCL INFO Channel 09/0 : 108[4] -> 92[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272153:272573 [3] NCCL INFO Channel 08/0 : 115[3] -> 99[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636086:2636507 [2] NCCL INFO Channel 12/0 : 42[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636091:2636512 [7] NCCL INFO Channel 02/0 : 55[7] -> 63[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722052:2722464 [7] NCCL INFO Channel 07/0 : 63[7] -> 127[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722049:2722467 [4] NCCL INFO Channel 01/0 : 124[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722048:2722469 [3] NCCL INFO Channel 09/0 : 59[3] -> 123[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153569:153981 [7] NCCL INFO Channel 09/0 : 63[7] -> 95[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153567:153979 [5] NCCL INFO Channel 07/0 : 61[5] -> 93[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557271:1557690 [5] NCCL INFO Channel 14/0 : 101[5] -> 69[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557267:1557683 [1] NCCL INFO Channel 10/0 : 65[1] -> 97[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235649:236071 [1] NCCL INFO Channel 11/0 : 105[1] -> 89[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272152:272575 [2] NCCL INFO Channel 14/0 : 114[2] -> 98[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636087:2636506 [3] NCCL INFO Channel 04/0 : 51[3] -> 59[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636085:2636509 [1] NCCL INFO Channel 08/0 : 49[1] -> 57[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636089:2636508 [5] NCCL INFO Channel 14/0 : 53[5] -> 61[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636090:2636505 [6] NCCL INFO Channel 04/0 : 54[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636084:2636511 [0] NCCL INFO Channel 02/0 : 48[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636088:2636510 [4] NCCL INFO Channel 15/0 : 44[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722045:2722470 [0] NCCL INFO Channel 07/0 : 56[0] -> 120[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153562:153983 [0] NCCL INFO Channel 01/0 : 56[0] -> 88[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557273:1557685 [7] NCCL INFO Channel 12/0 : 103[7] -> 71[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557268:1557686 [2] NCCL INFO Channel 10/0 : 98[2] -> 66[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235648:236064 [0] NCCL INFO Channel 11/0 : 104[0] -> 88[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235650:236065 [2] NCCL INFO Channel 15/0 : 106[2] -> 90[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272151:272571 [1] NCCL INFO Channel 14/0 : 97[1] -> 113[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636091:2636512 [7] NCCL INFO Channel 04/0 : 55[7] -> 63[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636086:2636507 [2] NCCL INFO Channel 13/0 : 42[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-52:2722049:2722467 [4] NCCL INFO Channel 03/0 : 124[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153564:153984 [2] NCCL INFO Channel 05/0 : 58[2] -> 90[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557269:1557684 [3] NCCL INFO Channel 00/0 : 99[3] -> 67[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235652:236066 [4] NCCL INFO Channel 11/0 : 108[4] -> 92[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272153:272573 [3] NCCL INFO Channel 10/0 : 115[3] -> 99[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815968:2816394 [5] NCCL INFO Channel 11/0 : 29[5] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636087:2636506 [3] NCCL INFO Channel 06/0 : 51[3] -> 59[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636085:2636509 [1] NCCL INFO Channel 10/0 : 49[1] -> 57[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-52:2722052:2722464 [7] NCCL INFO Channel 09/0 : 63[7] -> 127[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153568:153982 [6] NCCL INFO Channel 05/0 : 62[6] -> 94[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557267:1557683 [1] NCCL INFO Channel 12/0 : 65[1] -> 97[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235648:236064 [0] NCCL INFO Channel 13/0 : 104[0] -> 88[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272153:272573 [3] NCCL INFO Channel 12/0 : 115[3] -> 99[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636090:2636505 [6] NCCL INFO Channel 08/0 : 54[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636088:2636510 [4] NCCL INFO Channel 00/0 : 52[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722045:2722470 [0] NCCL INFO Channel 09/0 : 56[0] -> 120[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-52:2722048:2722469 [3] NCCL INFO Channel 13/0 : 59[3] -> 123[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153566:153980 [4] NCCL INFO Channel 01/0 : 60[4] -> 92[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557273:1557685 [7] NCCL INFO Channel 14/0 : 103[7] -> 71[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557268:1557686 [2] NCCL INFO Channel 12/0 : 98[2] -> 66[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235649:236071 [1] NCCL INFO Channel 13/0 : 105[1] -> 89[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235652:236066 [4] NCCL INFO Channel 13/0 : 108[4] -> 92[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272153:272573 [3] NCCL INFO Channel 14/0 : 115[3] -> 99[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636084:2636511 [0] NCCL INFO Channel 04/0 : 48[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-52:2722049:2722467 [4] NCCL INFO Channel 05/0 : 124[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153569:153981 [7] NCCL INFO Channel 11/0 : 63[7] -> 95[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557269:1557684 [3] NCCL INFO Channel 02/0 : 99[3] -> 67[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235648:236064 [0] NCCL INFO Channel 15/0 : 104[0] -> 88[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272151:272571 [1] NCCL INFO Channel 00/0 : 113[1] -> 97[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636091:2636512 [7] NCCL INFO Channel 06/0 : 55[7] -> 63[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636086:2636507 [2] NCCL INFO Channel 14/0 : 42[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-52:2722052:2722464 [7] NCCL INFO Channel 11/0 : 63[7] -> 127[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153567:153979 [5] NCCL INFO Channel 09/0 : 61[5] -> 93[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153563:153986 [1] NCCL INFO Channel 01/0 : 57[1] -> 89[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557267:1557683 [1] NCCL INFO Channel 14/0 : 65[1] -> 97[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235652:236066 [4] NCCL INFO Channel 15/0 : 108[4] -> 92[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235649:236071 [1] NCCL INFO Channel 15/0 : 105[1] -> 89[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272151:272571 [1] NCCL INFO Channel 02/0 : 113[1] -> 97[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815968:2816394 [5] NCCL INFO Channel 15/0 : 29[5] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636087:2636506 [3] NCCL INFO Channel 08/0 : 51[3] -> 59[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-52:2722048:2722469 [3] NCCL INFO Channel 15/0 : 59[3] -> 123[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153562:153983 [0] NCCL INFO Channel 03/0 : 56[0] -> 88[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153564:153984 [2] NCCL INFO Channel 07/0 : 58[2] -> 90[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557268:1557686 [2] NCCL INFO Channel 14/0 : 98[2] -> 66[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272151:272571 [1] NCCL INFO Channel 04/0 : 113[1] -> 97[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636090:2636505 [6] NCCL INFO Channel 10/0 : 54[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636085:2636509 [1] NCCL INFO Channel 12/0 : 49[1] -> 57[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-52:2722045:2722470 [0] NCCL INFO Channel 11/0 : 56[0] -> 120[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-52:2722049:2722467 [4] NCCL INFO Channel 07/0 : 124[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153568:153982 [6] NCCL INFO Channel 07/0 : 62[6] -> 94[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153566:153980 [4] NCCL INFO Channel 03/0 : 60[4] -> 92[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557269:1557684 [3] NCCL INFO Channel 04/0 : 99[3] -> 67[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272151:272571 [1] NCCL INFO Channel 06/0 : 113[1] -> 97[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636088:2636510 [4] NCCL INFO Channel 02/0 : 52[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722052:2722464 [7] NCCL INFO Channel 13/0 : 63[7] -> 127[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153569:153981 [7] NCCL INFO Channel 13/0 : 63[7] -> 95[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557269:1557684 [3] NCCL INFO Channel 06/0 : 99[3] -> 67[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272151:272571 [1] NCCL INFO Channel 08/0 : 113[1] -> 97[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636091:2636512 [7] NCCL INFO Channel 08/0 : 55[7] -> 63[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722048:2722469 [3] NCCL INFO Channel 01/0 : 123[3] -> 59[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153567:153979 [5] NCCL INFO Channel 11/0 : 61[5] -> 93[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557269:1557684 [3] NCCL INFO Channel 08/0 : 99[3] -> 67[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272151:272571 [1] NCCL INFO Channel 10/0 : 113[1] -> 97[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636084:2636511 [0] NCCL INFO Channel 06/0 : 48[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-52:2722049:2722467 [4] NCCL INFO Channel 09/0 : 124[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153563:153986 [1] NCCL INFO Channel 03/0 : 57[1] -> 89[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557267:1557683 [1] NCCL INFO Channel 00/0 : 97[1] -> 65[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272151:272571 [1] NCCL INFO Channel 12/0 : 113[1] -> 97[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815964:2816401 [1] NCCL INFO Channel 01/0 : 25[1] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636087:2636506 [3] NCCL INFO Channel 10/0 : 51[3] -> 59[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636086:2636507 [2] NCCL INFO Channel 15/0 : 42[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-52:2722045:2722470 [0] NCCL INFO Channel 13/0 : 56[0] -> 120[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-52:2722052:2722464 [7] NCCL INFO Channel 01/0 : 127[7] -> 63[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153562:153983 [0] NCCL INFO Channel 05/0 : 56[0] -> 88[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557269:1557684 [3] NCCL INFO Channel 10/0 : 99[3] -> 67[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272151:272571 [1] NCCL INFO Channel 14/0 : 113[1] -> 97[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815968:2816394 [5] NCCL INFO Channel 01/0 : 61[5] -> 93[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636090:2636505 [6] NCCL INFO Channel 12/0 : 54[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-52:2722048:2722469 [3] NCCL INFO Channel 03/0 : 123[3] -> 59[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-52:2722049:2722467 [4] NCCL INFO Channel 11/0 : 124[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153564:153984 [2] NCCL INFO Channel 09/0 : 58[2] -> 90[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557267:1557683 [1] NCCL INFO Channel 02/0 : 97[1] -> 65[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-52:2722052:2722464 [7] NCCL INFO Channel 03/0 : 127[7] -> 63[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153566:153980 [4] NCCL INFO Channel 05/0 : 60[4] -> 92[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153568:153982 [6] NCCL INFO Channel 09/0 : 62[6] -> 94[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557269:1557684 [3] NCCL INFO Channel 12/0 : 99[3] -> 67[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-52:2722045:2722470 [0] NCCL INFO Channel 15/0 : 56[0] -> 120[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153567:153979 [5] NCCL INFO Channel 15/0 : 61[5] -> 93[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557267:1557683 [1] NCCL INFO Channel 04/0 : 97[1] -> 65[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-52:2722049:2722467 [4] NCCL INFO Channel 13/0 : 124[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722048:2722469 [3] NCCL INFO Channel 05/0 : 123[3] -> 59[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153563:153986 [1] NCCL INFO Channel 05/0 : 57[1] -> 89[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557269:1557684 [3] NCCL INFO Channel 14/0 : 99[3] -> 67[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815969:2816398 [6] NCCL INFO Channel 01/0 : 30[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-52:2722045:2722470 [0] NCCL INFO Channel 01/0 : 120[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-52:2722052:2722464 [7] NCCL INFO Channel 05/0 : 127[7] -> 63[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153562:153983 [0] NCCL INFO Channel 07/0 : 56[0] -> 88[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557267:1557683 [1] NCCL INFO Channel 06/0 : 97[1] -> 65[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815964:2816401 [1] NCCL INFO Channel 03/0 : 25[1] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815968:2816394 [5] NCCL INFO Channel 03/0 : 61[5] -> 93[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636085:2636509 [1] NCCL INFO Channel 14/0 : 49[1] -> 57[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636088:2636510 [4] NCCL INFO Channel 06/0 : 52[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722049:2722467 [4] NCCL INFO Channel 15/0 : 124[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153564:153984 [2] NCCL INFO Channel 11/0 : 58[2] -> 90[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557267:1557683 [1] NCCL INFO Channel 08/0 : 97[1] -> 65[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636089:2636508 [5] NCCL INFO Channel 00/0 : 37[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-52:2722048:2722469 [3] NCCL INFO Channel 07/0 : 123[3] -> 59[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153566:153980 [4] NCCL INFO Channel 07/0 : 60[4] -> 92[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153568:153982 [6] NCCL INFO Channel 11/0 : 62[6] -> 94[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557267:1557683 [1] NCCL INFO Channel 10/0 : 97[1] -> 65[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636091:2636512 [7] NCCL INFO Channel 10/0 : 55[7] -> 63[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722045:2722470 [0] NCCL INFO Channel 03/0 : 120[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-52:2722052:2722464 [7] NCCL INFO Channel 07/0 : 127[7] -> 63[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153563:153986 [1] NCCL INFO Channel 07/0 : 57[1] -> 89[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557267:1557683 [1] NCCL INFO Channel 12/0 : 97[1] -> 65[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636084:2636511 [0] NCCL INFO Channel 08/0 : 48[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636087:2636506 [3] NCCL INFO Channel 12/0 : 51[3] -> 59[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-52:2722048:2722469 [3] NCCL INFO Channel 09/0 : 123[3] -> 59[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153562:153983 [0] NCCL INFO Channel 09/0 : 56[0] -> 88[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815966:2816396 [3] NCCL INFO Channel 01/0 : 27[3] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32709:33129 [5] NCCL INFO Channel 01/0 : 29[5] -> 61[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636086:2636507 [2] NCCL INFO Channel 00/0 : 50[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636090:2636505 [6] NCCL INFO Channel 14/0 : 54[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-52:2722045:2722470 [0] NCCL INFO Channel 05/0 : 120[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-52:2722052:2722464 [7] NCCL INFO Channel 09/0 : 127[7] -> 63[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153564:153984 [2] NCCL INFO Channel 13/0 : 58[2] -> 90[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815969:2816398 [6] NCCL INFO Channel 03/0 : 30[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636088:2636510 [4] NCCL INFO Channel 08/0 : 52[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636089:2636508 [5] NCCL INFO Channel 02/0 : 37[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636091:2636512 [7] NCCL INFO Channel 12/0 : 55[7] -> 63[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722048:2722469 [3] NCCL INFO Channel 13/0 : 123[3] -> 59[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153566:153980 [4] NCCL INFO Channel 09/0 : 60[4] -> 92[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153568:153982 [6] NCCL INFO Channel 13/0 : 62[6] -> 94[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153563:153986 [1] NCCL INFO Channel 11/0 : 57[1] -> 89[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815968:2816394 [5] NCCL INFO Channel 05/0 : 61[5] -> 93[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815964:2816401 [1] NCCL INFO Channel 05/0 : 25[1] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636084:2636511 [0] NCCL INFO Channel 10/0 : 48[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636087:2636506 [3] NCCL INFO Channel 14/0 : 51[3] -> 59[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-52:2722045:2722470 [0] NCCL INFO Channel 07/0 : 120[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153562:153983 [0] NCCL INFO Channel 11/0 : 56[0] -> 88[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153564:153984 [2] NCCL INFO Channel 15/0 : 58[2] -> 90[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557267:1557683 [1] NCCL INFO Channel 14/0 : 97[1] -> 65[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636086:2636507 [2] NCCL INFO Channel 04/0 : 50[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-52:2722052:2722464 [7] NCCL INFO Channel 11/0 : 127[7] -> 63[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32709:33129 [5] NCCL INFO Channel 03/0 : 29[5] -> 61[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636091:2636512 [7] NCCL INFO Channel 14/0 : 55[7] -> 63[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636088:2636510 [4] NCCL INFO Channel 10/0 : 52[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636089:2636508 [5] NCCL INFO Channel 04/0 : 37[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-52:2722048:2722469 [3] NCCL INFO Channel 15/0 : 123[3] -> 59[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636084:2636511 [0] NCCL INFO Channel 12/0 : 48[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-52:2722045:2722470 [0] NCCL INFO Channel 09/0 : 120[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-52:2722052:2722464 [7] NCCL INFO Channel 13/0 : 127[7] -> 63[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815966:2816396 [3] NCCL INFO Channel 03/0 : 27[3] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636089:2636508 [5] NCCL INFO Channel 06/0 : 37[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636088:2636510 [4] NCCL INFO Channel 12/0 : 52[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636086:2636507 [2] NCCL INFO Channel 06/0 : 50[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-52:2722045:2722470 [0] NCCL INFO Channel 11/0 : 120[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153568:153982 [6] NCCL INFO Channel 15/0 : 62[6] -> 94[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153566:153980 [4] NCCL INFO Channel 11/0 : 60[4] -> 92[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815970:2816397 [7] NCCL INFO Channel 01/0 : 31[7] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636084:2636511 [0] NCCL INFO Channel 14/0 : 48[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-52:2722045:2722470 [0] NCCL INFO Channel 13/0 : 120[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153562:153983 [0] NCCL INFO Channel 13/0 : 56[0] -> 88[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153563:153986 [1] NCCL INFO Channel 13/0 : 57[1] -> 89[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636085:2636509 [1] NCCL INFO Channel 00/0 : 33[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-52:2722045:2722470 [0] NCCL INFO Channel 15/0 : 120[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153566:153980 [4] NCCL INFO Channel 13/0 : 60[4] -> 92[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32709:33129 [5] NCCL INFO Channel 05/0 : 29[5] -> 61[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636089:2636508 [5] NCCL INFO Channel 08/0 : 37[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636088:2636510 [4] NCCL INFO Channel 14/0 : 52[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636086:2636507 [2] NCCL INFO Channel 08/0 : 50[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153562:153983 [0] NCCL INFO Channel 15/0 : 56[0] -> 88[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153563:153986 [1] NCCL INFO Channel 15/0 : 57[1] -> 89[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636085:2636509 [1] NCCL INFO Channel 02/0 : 33[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153566:153980 [4] NCCL INFO Channel 15/0 : 60[4] -> 92[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815968:2816394 [5] NCCL INFO Channel 07/0 : 61[5] -> 93[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636090:2636505 [6] NCCL INFO Channel 00/0 : 38[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815969:2816398 [6] NCCL INFO Channel 05/0 : 30[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-52:2722046:2722466 [1] NCCL INFO Channel 13/0 : 121[1] -> 57[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815964:2816401 [1] NCCL INFO Channel 07/0 : 25[1] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815966:2816396 [3] NCCL INFO Channel 05/0 : 27[3] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-52:2722046:2722466 [1] NCCL INFO Channel 15/0 : 121[1] -> 57[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815963:2816400 [0] NCCL INFO Channel 01/0 : 24[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815970:2816397 [7] NCCL INFO Channel 03/0 : 31[7] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32709:33129 [5] NCCL INFO Channel 07/0 : 29[5] -> 61[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815968:2816394 [5] NCCL INFO Channel 09/0 : 61[5] -> 93[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636089:2636508 [5] NCCL INFO Channel 10/0 : 37[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636087:2636506 [3] NCCL INFO Channel 00/0 : 35[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636086:2636507 [2] NCCL INFO Channel 10/0 : 50[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636090:2636505 [6] NCCL INFO Channel 02/0 : 38[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636085:2636509 [1] NCCL INFO Channel 04/0 : 33[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32709:33129 [5] NCCL INFO Channel 09/0 : 29[5] -> 61[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636091:2636512 [7] NCCL INFO Channel 00/0 : 39[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815969:2816398 [6] NCCL INFO Channel 07/0 : 30[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636089:2636508 [5] NCCL INFO Channel 12/0 : 37[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815966:2816396 [3] NCCL INFO Channel 07/0 : 27[3] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636087:2636506 [3] NCCL INFO Channel 02/0 : 35[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815964:2816401 [1] NCCL INFO Channel 11/0 : 25[1] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636090:2636505 [6] NCCL INFO Channel 04/0 : 38[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636086:2636507 [2] NCCL INFO Channel 12/0 : 50[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636085:2636509 [1] NCCL INFO Channel 06/0 : 33[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815963:2816400 [0] NCCL INFO Channel 03/0 : 24[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636091:2636512 [7] NCCL INFO Channel 02/0 : 39[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636084:2636511 [0] NCCL INFO Channel 02/0 : 32[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636089:2636508 [5] NCCL INFO Channel 14/0 : 37[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815968:2816394 [5] NCCL INFO Channel 11/0 : 61[5] -> 93[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815970:2816397 [7] NCCL INFO Channel 05/0 : 31[7] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32709:33129 [5] NCCL INFO Channel 11/0 : 29[5] -> 61[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636087:2636506 [3] NCCL INFO Channel 04/0 : 35[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636088:2636510 [4] NCCL INFO Channel 00/0 : 36[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815966:2816396 [3] NCCL INFO Channel 09/0 : 27[3] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815967:2816399 [4] NCCL INFO Channel 01/0 : 28[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32709:33129 [5] NCCL INFO Channel 15/0 : 29[5] -> 61[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636090:2636505 [6] NCCL INFO Channel 08/0 : 38[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815969:2816398 [6] NCCL INFO Channel 09/0 : 30[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636086:2636507 [2] NCCL INFO Channel 14/0 : 50[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636085:2636509 [1] NCCL INFO Channel 08/0 : 33[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815964:2816401 [1] NCCL INFO Channel 13/0 : 25[1] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815963:2816400 [0] NCCL INFO Channel 05/0 : 24[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815968:2816394 [5] NCCL INFO Channel 15/0 : 61[5] -> 93[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636084:2636511 [0] NCCL INFO Channel 04/0 : 32[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636091:2636512 [7] NCCL INFO Channel 04/0 : 39[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815970:2816397 [7] NCCL INFO Channel 07/0 : 31[7] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815966:2816396 [3] NCCL INFO Channel 13/0 : 27[3] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62884:63299 [3] NCCL INFO Channel 14/0 : 43[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636088:2636510 [4] NCCL INFO Channel 02/0 : 36[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636090:2636505 [6] NCCL INFO Channel 10/0 : 38[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636087:2636506 [3] NCCL INFO Channel 06/0 : 35[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62882:63300 [1] NCCL INFO Channel 12/0 : 41[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62881:63298 [0] NCCL INFO Channel 11/0 : 40[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62883:63301 [2] NCCL INFO Channel 08/0 : 42[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636089:2636508 [5] NCCL INFO Channel 00/0 : 53[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62885:63304 [4] NCCL INFO Channel 08/0 : 44[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62887:63305 [6] NCCL INFO Channel 09/0 : 46[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815969:2816398 [6] NCCL INFO Channel 11/0 : 30[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815967:2816399 [4] NCCL INFO Channel 03/0 : 28[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636085:2636509 [1] NCCL INFO Channel 10/0 : 33[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62884:63299 [3] NCCL INFO Channel 15/0 : 43[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62881:63298 [0] NCCL INFO Channel 12/0 : 40[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62882:63300 [1] NCCL INFO Channel 13/0 : 41[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815963:2816400 [0] NCCL INFO Channel 07/0 : 24[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815964:2816401 [1] NCCL INFO Channel 15/0 : 25[1] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62883:63301 [2] NCCL INFO Channel 09/0 : 42[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62885:63304 [4] NCCL INFO Channel 09/0 : 44[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62887:63305 [6] NCCL INFO Channel 10/0 : 46[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815966:2816396 [3] NCCL INFO Channel 15/0 : 27[3] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62881:63298 [0] NCCL INFO Channel 13/0 : 40[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62882:63300 [1] NCCL INFO Channel 14/0 : 41[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815970:2816397 [7] NCCL INFO Channel 09/0 : 31[7] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32705:33128 [1] NCCL INFO Channel 01/0 : 25[1] -> 57[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62883:63301 [2] NCCL INFO Channel 11/0 : 42[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815969:2816398 [6] NCCL INFO Channel 13/0 : 30[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62887:63305 [6] NCCL INFO Channel 11/0 : 46[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62885:63304 [4] NCCL INFO Channel 10/0 : 44[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636091:2636512 [7] NCCL INFO Channel 06/0 : 39[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62882:63300 [1] NCCL INFO Channel 15/0 : 41[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62881:63298 [0] NCCL INFO Channel 14/0 : 40[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636084:2636511 [0] NCCL INFO Channel 06/0 : 32[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62883:63301 [2] NCCL INFO Channel 12/0 : 42[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636088:2636510 [4] NCCL INFO Channel 06/0 : 36[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62887:63305 [6] NCCL INFO Channel 12/0 : 46[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62885:63304 [4] NCCL INFO Channel 11/0 : 44[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636090:2636505 [6] NCCL INFO Channel 12/0 : 38[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62881:63298 [0] NCCL INFO Channel 15/0 : 40[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815967:2816399 [4] NCCL INFO Channel 05/0 : 28[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815963:2816400 [0] NCCL INFO Channel 09/0 : 24[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636089:2636508 [5] NCCL INFO Channel 02/0 : 53[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62883:63301 [2] NCCL INFO Channel 13/0 : 42[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815964:2816401 [1] NCCL INFO Channel 01/0 : 57[1] -> 89[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62887:63305 [6] NCCL INFO Channel 13/0 : 46[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62885:63304 [4] NCCL INFO Channel 13/0 : 44[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815966:2816396 [3] NCCL INFO Channel 01/0 : 59[3] -> 91[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62883:63301 [2] NCCL INFO Channel 14/0 : 42[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815970:2816397 [7] NCCL INFO Channel 11/0 : 31[7] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32705:33128 [1] NCCL INFO Channel 03/0 : 25[1] -> 57[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32709:33129 [5] NCCL INFO Channel 01/0 : 61[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62887:63305 [6] NCCL INFO Channel 15/0 : 46[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62885:63304 [4] NCCL INFO Channel 14/0 : 44[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815969:2816398 [6] NCCL INFO Channel 15/0 : 30[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62883:63301 [2] NCCL INFO Channel 15/0 : 42[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636087:2636506 [3] NCCL INFO Channel 08/0 : 35[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636085:2636509 [1] NCCL INFO Channel 12/0 : 33[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62885:63304 [4] NCCL INFO Channel 15/0 : 44[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636091:2636512 [7] NCCL INFO Channel 08/0 : 39[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62886:63303 [5] NCCL INFO Channel 01/0 : 29[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636084:2636511 [0] NCCL INFO Channel 08/0 : 32[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62886:63303 [5] NCCL INFO Channel 03/0 : 29[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636088:2636510 [4] NCCL INFO Channel 08/0 : 36[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62886:63303 [5] NCCL INFO Channel 05/0 : 29[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32709:33129 [5] NCCL INFO Channel 03/0 : 61[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32705:33128 [1] NCCL INFO Channel 05/0 : 25[1] -> 57[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636090:2636505 [6] NCCL INFO Channel 14/0 : 38[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636089:2636508 [5] NCCL INFO Channel 04/0 : 53[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153567:153979 [5] NCCL INFO Channel 01/0 : 93[5] -> 61[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62886:63303 [5] NCCL INFO Channel 07/0 : 29[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815967:2816399 [4] NCCL INFO Channel 07/0 : 28[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815963:2816400 [0] NCCL INFO Channel 11/0 : 24[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815964:2816401 [1] NCCL INFO Channel 03/0 : 57[1] -> 89[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815966:2816396 [3] NCCL INFO Channel 03/0 : 59[3] -> 91[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62886:63303 [5] NCCL INFO Channel 09/0 : 29[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815969:2816398 [6] NCCL INFO Channel 01/0 : 62[6] -> 94[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815970:2816397 [7] NCCL INFO Channel 13/0 : 31[7] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62886:63303 [5] NCCL INFO Channel 11/0 : 29[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815968:2816394 [5] NCCL INFO Channel 01/0 : 125[5] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62882:63300 [1] NCCL INFO Channel 01/0 : 25[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62886:63303 [5] NCCL INFO Channel 15/0 : 29[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815965:2816395 [2] NCCL INFO Channel 01/0 : 26[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62887:63305 [6] NCCL INFO Channel 01/0 : 30[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62882:63300 [1] NCCL INFO Channel 03/0 : 25[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32709:33129 [5] NCCL INFO Channel 05/0 : 61[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636085:2636509 [1] NCCL INFO Channel 14/0 : 33[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153567:153979 [5] NCCL INFO Channel 03/0 : 93[5] -> 61[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62884:63299 [3] NCCL INFO Channel 01/0 : 27[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62887:63305 [6] NCCL INFO Channel 03/0 : 30[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636087:2636506 [3] NCCL INFO Channel 10/0 : 35[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636091:2636512 [7] NCCL INFO Channel 10/0 : 39[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62882:63300 [1] NCCL INFO Channel 05/0 : 25[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636084:2636511 [0] NCCL INFO Channel 10/0 : 32[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62888:63302 [7] NCCL INFO Channel 01/0 : 31[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815966:2816396 [3] NCCL INFO Channel 05/0 : 59[3] -> 91[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636088:2636510 [4] NCCL INFO Channel 10/0 : 36[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636089:2636508 [5] NCCL INFO Channel 06/0 : 53[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636086:2636507 [2] NCCL INFO Channel 00/0 : 34[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62886:63303 [5] NCCL INFO Channel 01/0 : 45[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62882:63300 [1] NCCL INFO Channel 07/0 : 25[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815967:2816399 [4] NCCL INFO Channel 09/0 : 28[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815963:2816400 [0] NCCL INFO Channel 13/0 : 24[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815964:2816401 [1] NCCL INFO Channel 05/0 : 57[1] -> 89[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32705:33128 [1] NCCL INFO Channel 07/0 : 25[1] -> 57[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62884:63299 [3] NCCL INFO Channel 03/0 : 27[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62887:63305 [6] NCCL INFO Channel 05/0 : 30[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815969:2816398 [6] NCCL INFO Channel 03/0 : 62[6] -> 94[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32707:33124 [3] NCCL INFO Channel 01/0 : 27[3] -> 59[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153567:153979 [5] NCCL INFO Channel 05/0 : 93[5] -> 61[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62888:63302 [7] NCCL INFO Channel 03/0 : 31[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62881:63298 [0] NCCL INFO Channel 01/0 : 24[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815970:2816397 [7] NCCL INFO Channel 01/0 : 63[7] -> 95[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62886:63303 [5] NCCL INFO Channel 03/0 : 45[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815968:2816394 [5] NCCL INFO Channel 03/0 : 125[5] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636087:2636506 [3] NCCL INFO Channel 12/0 : 35[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636091:2636512 [7] NCCL INFO Channel 12/0 : 39[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62882:63300 [1] NCCL INFO Channel 11/0 : 25[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815965:2816395 [2] NCCL INFO Channel 03/0 : 26[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636084:2636511 [0] NCCL INFO Channel 12/0 : 32[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62887:63305 [6] NCCL INFO Channel 07/0 : 30[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62884:63299 [3] NCCL INFO Channel 05/0 : 27[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62888:63302 [7] NCCL INFO Channel 05/0 : 31[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32710:33125 [6] NCCL INFO Channel 01/0 : 30[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153567:153979 [5] NCCL INFO Channel 07/0 : 93[5] -> 61[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62881:63298 [0] NCCL INFO Channel 03/0 : 24[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815966:2816396 [3] NCCL INFO Channel 07/0 : 59[3] -> 91[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32709:33129 [5] NCCL INFO Channel 07/0 : 61[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636088:2636510 [4] NCCL INFO Channel 12/0 : 36[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636089:2636508 [5] NCCL INFO Channel 08/0 : 53[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62886:63303 [5] NCCL INFO Channel 05/0 : 45[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815967:2816399 [4] NCCL INFO Channel 11/0 : 28[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32711:33127 [7] NCCL INFO Channel 01/0 : 31[7] -> 63[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62885:63304 [4] NCCL INFO Channel 01/0 : 28[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62882:63300 [1] NCCL INFO Channel 13/0 : 25[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32707:33124 [3] NCCL INFO Channel 03/0 : 27[3] -> 59[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62887:63305 [6] NCCL INFO Channel 09/0 : 30[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62884:63299 [3] NCCL INFO Channel 07/0 : 27[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153567:153979 [5] NCCL INFO Channel 09/0 : 93[5] -> 61[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62888:63302 [7] NCCL INFO Channel 07/0 : 31[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62886:63303 [5] NCCL INFO Channel 07/0 : 45[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636090:2636505 [6] NCCL INFO Channel 00/0 : 54[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636086:2636507 [2] NCCL INFO Channel 04/0 : 34[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62881:63298 [0] NCCL INFO Channel 05/0 : 24[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815963:2816400 [0] NCCL INFO Channel 15/0 : 24[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815969:2816398 [6] NCCL INFO Channel 05/0 : 62[6] -> 94[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636091:2636512 [7] NCCL INFO Channel 14/0 : 39[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62885:63304 [4] NCCL INFO Channel 03/0 : 28[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62882:63300 [1] NCCL INFO Channel 15/0 : 25[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815970:2816397 [7] NCCL INFO Channel 03/0 : 63[7] -> 95[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32709:33129 [5] NCCL INFO Channel 09/0 : 61[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636085:2636509 [1] NCCL INFO Channel 00/0 : 49[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636087:2636506 [3] NCCL INFO Channel 14/0 : 35[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62887:63305 [6] NCCL INFO Channel 11/0 : 30[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62884:63299 [3] NCCL INFO Channel 09/0 : 27[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815964:2816401 [1] NCCL INFO Channel 07/0 : 57[1] -> 89[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815968:2816394 [5] NCCL INFO Channel 05/0 : 125[5] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32705:33128 [1] NCCL INFO Channel 11/0 : 25[1] -> 57[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153567:153979 [5] NCCL INFO Channel 11/0 : 93[5] -> 61[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62886:63303 [5] NCCL INFO Channel 09/0 : 45[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815966:2816396 [3] NCCL INFO Channel 09/0 : 59[3] -> 91[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815965:2816395 [2] NCCL INFO Channel 05/0 : 26[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32710:33125 [6] NCCL INFO Channel 03/0 : 30[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32711:33127 [7] NCCL INFO Channel 03/0 : 31[7] -> 63[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62888:63302 [7] NCCL INFO Channel 09/0 : 31[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62881:63298 [0] NCCL INFO Channel 07/0 : 24[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815967:2816399 [4] NCCL INFO Channel 13/0 : 28[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32707:33124 [3] NCCL INFO Channel 05/0 : 27[3] -> 59[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62885:63304 [4] NCCL INFO Channel 05/0 : 28[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32709:33129 [5] NCCL INFO Channel 11/0 : 61[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62884:63299 [3] NCCL INFO Channel 13/0 : 27[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62886:63303 [5] NCCL INFO Channel 11/0 : 45[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636089:2636508 [5] NCCL INFO Channel 10/0 : 53[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636088:2636510 [4] NCCL INFO Channel 14/0 : 36[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62887:63305 [6] NCCL INFO Channel 13/0 : 30[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636084:2636511 [0] NCCL INFO Channel 14/0 : 32[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62881:63298 [0] NCCL INFO Channel 09/0 : 24[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62888:63302 [7] NCCL INFO Channel 11/0 : 31[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62885:63304 [4] NCCL INFO Channel 07/0 : 28[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636090:2636505 [6] NCCL INFO Channel 02/0 : 54[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636086:2636507 [2] NCCL INFO Channel 06/0 : 34[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153567:153979 [5] NCCL INFO Channel 15/0 : 93[5] -> 61[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62882:63300 [1] NCCL INFO Channel 01/0 : 41[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815963:2816400 [0] NCCL INFO Channel 01/0 : 56[0] -> 88[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32710:33125 [6] NCCL INFO Channel 05/0 : 30[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32704:33123 [0] NCCL INFO Channel 01/0 : 24[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32709:33129 [5] NCCL INFO Channel 15/0 : 61[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32711:33127 [7] NCCL INFO Channel 05/0 : 31[7] -> 63[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636085:2636509 [1] NCCL INFO Channel 02/0 : 49[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62886:63303 [5] NCCL INFO Channel 15/0 : 45[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815969:2816398 [6] NCCL INFO Channel 07/0 : 62[6] -> 94[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32707:33124 [3] NCCL INFO Channel 07/0 : 27[3] -> 59[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815970:2816397 [7] NCCL INFO Channel 05/0 : 63[7] -> 95[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32705:33128 [1] NCCL INFO Channel 13/0 : 25[1] -> 57[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32708:33126 [4] NCCL INFO Channel 01/0 : 28[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815964:2816401 [1] NCCL INFO Channel 11/0 : 57[1] -> 89[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815966:2816396 [3] NCCL INFO Channel 13/0 : 59[3] -> 91[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636089:2636508 [5] NCCL INFO Channel 12/0 : 53[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62884:63299 [3] NCCL INFO Channel 15/0 : 27[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636090:2636505 [6] NCCL INFO Channel 04/0 : 54[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62887:63305 [6] NCCL INFO Channel 15/0 : 30[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32710:33125 [6] NCCL INFO Channel 07/0 : 30[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32707:33124 [3] NCCL INFO Channel 09/0 : 27[3] -> 59[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636085:2636509 [1] NCCL INFO Channel 04/0 : 49[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62881:63298 [0] NCCL INFO Channel 11/0 : 24[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62888:63302 [7] NCCL INFO Channel 13/0 : 31[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62885:63304 [4] NCCL INFO Channel 09/0 : 28[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62882:63300 [1] NCCL INFO Channel 03/0 : 41[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636089:2636508 [5] NCCL INFO Channel 14/0 : 53[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62883:63301 [2] NCCL INFO Channel 01/0 : 26[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815968:2816394 [5] NCCL INFO Channel 07/0 : 125[5] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815965:2816395 [2] NCCL INFO Channel 07/0 : 26[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62881:63298 [0] NCCL INFO Channel 13/0 : 24[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815967:2816399 [4] NCCL INFO Channel 15/0 : 28[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32711:33127 [7] NCCL INFO Channel 07/0 : 31[7] -> 63[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62885:63304 [4] NCCL INFO Channel 11/0 : 28[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815963:2816400 [0] NCCL INFO Channel 03/0 : 56[0] -> 88[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815969:2816398 [6] NCCL INFO Channel 09/0 : 62[6] -> 94[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32704:33123 [0] NCCL INFO Channel 03/0 : 24[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62882:63300 [1] NCCL INFO Channel 05/0 : 41[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62883:63301 [2] NCCL INFO Channel 03/0 : 26[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815970:2816397 [7] NCCL INFO Channel 07/0 : 63[7] -> 95[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32708:33126 [4] NCCL INFO Channel 03/0 : 28[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32705:33128 [1] NCCL INFO Channel 15/0 : 25[1] -> 57[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636090:2636505 [6] NCCL INFO Channel 08/0 : 54[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62881:63298 [0] NCCL INFO Channel 15/0 : 24[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62885:63304 [4] NCCL INFO Channel 13/0 : 28[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815966:2816396 [3] NCCL INFO Channel 15/0 : 59[3] -> 91[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636085:2636509 [1] NCCL INFO Channel 06/0 : 49[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62883:63301 [2] NCCL INFO Channel 05/0 : 26[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62882:63300 [1] NCCL INFO Channel 07/0 : 41[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815964:2816401 [1] NCCL INFO Channel 13/0 : 57[1] -> 89[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636086:2636507 [2] NCCL INFO Channel 08/0 : 34[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62884:63299 [3] NCCL INFO Channel 01/0 : 43[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62887:63305 [6] NCCL INFO Channel 01/0 : 46[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62888:63302 [7] NCCL INFO Channel 01/0 : 47[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636087:2636506 [3] NCCL INFO Channel 00/0 : 51[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62885:63304 [4] NCCL INFO Channel 15/0 : 28[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62883:63301 [2] NCCL INFO Channel 07/0 : 26[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32707:33124 [3] NCCL INFO Channel 13/0 : 27[3] -> 59[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62882:63300 [1] NCCL INFO Channel 11/0 : 41[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32710:33125 [6] NCCL INFO Channel 09/0 : 30[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62887:63305 [6] NCCL INFO Channel 03/0 : 46[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62888:63302 [7] NCCL INFO Channel 03/0 : 47[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62884:63299 [3] NCCL INFO Channel 03/0 : 43[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815968:2816394 [5] NCCL INFO Channel 09/0 : 125[5] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32711:33127 [7] NCCL INFO Channel 09/0 : 31[7] -> 63[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636084:2636511 [0] NCCL INFO Channel 02/0 : 48[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636090:2636505 [6] NCCL INFO Channel 10/0 : 54[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62882:63300 [1] NCCL INFO Channel 13/0 : 41[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62883:63301 [2] NCCL INFO Channel 09/0 : 26[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815967:2816399 [4] NCCL INFO Channel 01/0 : 60[4] -> 92[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815963:2816400 [0] NCCL INFO Channel 05/0 : 56[0] -> 88[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815965:2816395 [2] NCCL INFO Channel 09/0 : 26[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32704:33123 [0] NCCL INFO Channel 05/0 : 24[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32708:33126 [4] NCCL INFO Channel 05/0 : 28[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636088:2636510 [4] NCCL INFO Channel 00/0 : 52[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636085:2636509 [1] NCCL INFO Channel 08/0 : 49[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636086:2636507 [2] NCCL INFO Channel 10/0 : 34[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62881:63298 [0] NCCL INFO Channel 01/0 : 40[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815969:2816398 [6] NCCL INFO Channel 11/0 : 62[6] -> 94[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815970:2816397 [7] NCCL INFO Channel 09/0 : 63[7] -> 95[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636087:2636506 [3] NCCL INFO Channel 02/0 : 51[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62887:63305 [6] NCCL INFO Channel 05/0 : 46[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62888:63302 [7] NCCL INFO Channel 05/0 : 47[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62884:63299 [3] NCCL INFO Channel 05/0 : 43[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815964:2816401 [1] NCCL INFO Channel 15/0 : 57[1] -> 89[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62882:63300 [1] NCCL INFO Channel 15/0 : 41[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62883:63301 [2] NCCL INFO Channel 11/0 : 26[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815968:2816394 [5] NCCL INFO Channel 11/0 : 125[5] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62885:63304 [4] NCCL INFO Channel 01/0 : 44[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815963:2816400 [0] NCCL INFO Channel 07/0 : 56[0] -> 88[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32707:33124 [3] NCCL INFO Channel 15/0 : 27[3] -> 59[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62881:63298 [0] NCCL INFO Channel 03/0 : 40[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32710:33125 [6] NCCL INFO Channel 11/0 : 30[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62887:63305 [6] NCCL INFO Channel 07/0 : 46[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32711:33127 [7] NCCL INFO Channel 11/0 : 31[7] -> 63[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62888:63302 [7] NCCL INFO Channel 07/0 : 47[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32708:33126 [4] NCCL INFO Channel 07/0 : 28[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636090:2636505 [6] NCCL INFO Channel 12/0 : 54[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636088:2636510 [4] NCCL INFO Channel 02/0 : 52[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62884:63299 [3] NCCL INFO Channel 07/0 : 43[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32704:33123 [0] NCCL INFO Channel 07/0 : 24[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62883:63301 [2] NCCL INFO Channel 13/0 : 26[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62885:63304 [4] NCCL INFO Channel 03/0 : 44[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62881:63298 [0] NCCL INFO Channel 05/0 : 40[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815965:2816395 [2] NCCL INFO Channel 11/0 : 26[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815967:2816399 [4] NCCL INFO Channel 03/0 : 60[4] -> 92[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62887:63305 [6] NCCL INFO Channel 09/0 : 46[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815970:2816397 [7] NCCL INFO Channel 11/0 : 63[7] -> 95[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815969:2816398 [6] NCCL INFO Channel 13/0 : 62[6] -> 94[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62888:63302 [7] NCCL INFO Channel 09/0 : 47[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815968:2816394 [5] NCCL INFO Channel 15/0 : 125[5] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636084:2636511 [0] NCCL INFO Channel 04/0 : 48[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636085:2636509 [1] NCCL INFO Channel 10/0 : 49[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636086:2636507 [2] NCCL INFO Channel 12/0 : 34[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815963:2816400 [0] NCCL INFO Channel 09/0 : 56[0] -> 88[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32706:33122 [2] NCCL INFO Channel 01/0 : 26[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636087:2636506 [3] NCCL INFO Channel 04/0 : 51[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815967:2816399 [4] NCCL INFO Channel 05/0 : 60[4] -> 92[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32710:33125 [6] NCCL INFO Channel 13/0 : 30[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636090:2636505 [6] NCCL INFO Channel 14/0 : 54[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153565:153985 [3] NCCL INFO Channel 01/0 : 91[3] -> 59[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815969:2816398 [6] NCCL INFO Channel 15/0 : 62[6] -> 94[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32711:33127 [7] NCCL INFO Channel 13/0 : 31[7] -> 63[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62884:63299 [3] NCCL INFO Channel 09/0 : 43[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815965:2816395 [2] NCCL INFO Channel 13/0 : 26[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815970:2816397 [7] NCCL INFO Channel 13/0 : 63[7] -> 95[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32708:33126 [4] NCCL INFO Channel 09/0 : 28[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32704:33123 [0] NCCL INFO Channel 09/0 : 24[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62883:63301 [2] NCCL INFO Channel 15/0 : 26[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815968:2816394 [5] NCCL INFO Channel 01/0 : 61[5] -> 125[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32706:33122 [2] NCCL INFO Channel 03/0 : 26[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32710:33125 [6] NCCL INFO Channel 15/0 : 30[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62885:63304 [4] NCCL INFO Channel 05/0 : 44[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62881:63298 [0] NCCL INFO Channel 07/0 : 40[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815963:2816400 [0] NCCL INFO Channel 11/0 : 56[0] -> 88[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62887:63305 [6] NCCL INFO Channel 11/0 : 46[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636088:2636510 [4] NCCL INFO Channel 06/0 : 52[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153565:153985 [3] NCCL INFO Channel 03/0 : 91[3] -> 59[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62888:63302 [7] NCCL INFO Channel 11/0 : 47[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636085:2636509 [1] NCCL INFO Channel 12/0 : 49[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636084:2636511 [0] NCCL INFO Channel 06/0 : 48[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636086:2636507 [2] NCCL INFO Channel 14/0 : 34[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636087:2636506 [3] NCCL INFO Channel 06/0 : 51[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62884:63299 [3] NCCL INFO Channel 13/0 : 43[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32705:33128 [1] NCCL INFO Channel 01/0 : 57[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62885:63304 [4] NCCL INFO Channel 07/0 : 44[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815967:2816399 [4] NCCL INFO Channel 07/0 : 60[4] -> 92[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32708:33126 [4] NCCL INFO Channel 11/0 : 28[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62881:63298 [0] NCCL INFO Channel 09/0 : 40[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62887:63305 [6] NCCL INFO Channel 13/0 : 46[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62888:63302 [7] NCCL INFO Channel 13/0 : 47[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62884:63299 [3] NCCL INFO Channel 15/0 : 43[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815968:2816394 [5] NCCL INFO Channel 03/0 : 61[5] -> 125[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815963:2816400 [0] NCCL INFO Channel 13/0 : 56[0] -> 88[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32704:33123 [0] NCCL INFO Channel 11/0 : 24[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32707:33124 [3] NCCL INFO Channel 01/0 : 59[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62885:63304 [4] NCCL INFO Channel 09/0 : 44[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815967:2816399 [4] NCCL INFO Channel 09/0 : 60[4] -> 92[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815965:2816395 [2] NCCL INFO Channel 15/0 : 26[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32706:33122 [2] NCCL INFO Channel 05/0 : 26[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636088:2636510 [4] NCCL INFO Channel 08/0 : 52[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153565:153985 [3] NCCL INFO Channel 05/0 : 91[3] -> 59[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62887:63305 [6] NCCL INFO Channel 15/0 : 46[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62881:63298 [0] NCCL INFO Channel 11/0 : 40[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815966:2816396 [3] NCCL INFO Channel 01/0 : 123[3] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32705:33128 [1] NCCL INFO Channel 03/0 : 57[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32708:33126 [4] NCCL INFO Channel 13/0 : 28[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636084:2636511 [0] NCCL INFO Channel 08/0 : 48[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62885:63304 [4] NCCL INFO Channel 11/0 : 44[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62883:63301 [2] NCCL INFO Channel 01/0 : 42[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815964:2816401 [1] NCCL INFO Channel 01/0 : 121[1] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62881:63298 [0] NCCL INFO Channel 13/0 : 40[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62885:63304 [4] NCCL INFO Channel 13/0 : 44[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153563:153986 [1] NCCL INFO Channel 01/0 : 89[1] -> 57[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114513:114933 [1] NCCL INFO Channel 00/0 : 1[1] -> 65[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62883:63301 [2] NCCL INFO Channel 03/0 : 42[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32707:33124 [3] NCCL INFO Channel 03/0 : 59[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636085:2636509 [1] NCCL INFO Channel 14/0 : 49[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62881:63298 [0] NCCL INFO Channel 15/0 : 40[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815963:2816400 [0] NCCL INFO Channel 15/0 : 56[0] -> 88[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32704:33123 [0] NCCL INFO Channel 13/0 : 24[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636088:2636510 [4] NCCL INFO Channel 10/0 : 52[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636087:2636506 [3] NCCL INFO Channel 08/0 : 51[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62885:63304 [4] NCCL INFO Channel 15/0 : 44[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815968:2816394 [5] NCCL INFO Channel 05/0 : 61[5] -> 125[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32706:33122 [2] NCCL INFO Channel 07/0 : 26[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153563:153986 [1] NCCL INFO Channel 03/0 : 89[1] -> 57[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153565:153985 [3] NCCL INFO Channel 07/0 : 91[3] -> 59[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815965:2816395 [2] NCCL INFO Channel 01/0 : 58[2] -> 90[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32708:33126 [4] NCCL INFO Channel 15/0 : 28[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114513:114933 [1] NCCL INFO Channel 02/0 : 1[1] -> 65[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815967:2816399 [4] NCCL INFO Channel 11/0 : 60[4] -> 92[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32705:33128 [1] NCCL INFO Channel 05/0 : 57[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815966:2816396 [3] NCCL INFO Channel 03/0 : 123[3] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32704:33123 [0] NCCL INFO Channel 15/0 : 24[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32707:33124 [3] NCCL INFO Channel 05/0 : 59[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636084:2636511 [0] NCCL INFO Channel 10/0 : 48[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636091:2636512 [7] NCCL INFO Channel 00/0 : 55[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62883:63301 [2] NCCL INFO Channel 05/0 : 42[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815964:2816401 [1] NCCL INFO Channel 03/0 : 121[1] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636088:2636510 [4] NCCL INFO Channel 12/0 : 52[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153563:153986 [1] NCCL INFO Channel 05/0 : 89[1] -> 57[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62883:63301 [2] NCCL INFO Channel 07/0 : 42[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815968:2816394 [5] NCCL INFO Channel 07/0 : 61[5] -> 125[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114513:114933 [1] NCCL INFO Channel 04/0 : 1[1] -> 65[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62883:63301 [2] NCCL INFO Channel 09/0 : 42[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32706:33122 [2] NCCL INFO Channel 09/0 : 26[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62883:63301 [2] NCCL INFO Channel 11/0 : 42[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32711:33127 [7] NCCL INFO Channel 01/0 : 63[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32710:33125 [6] NCCL INFO Channel 01/0 : 62[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32705:33128 [1] NCCL INFO Channel 07/0 : 57[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636087:2636506 [3] NCCL INFO Channel 10/0 : 51[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636086:2636507 [2] NCCL INFO Channel 00/0 : 50[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636091:2636512 [7] NCCL INFO Channel 02/0 : 55[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153565:153985 [3] NCCL INFO Channel 09/0 : 91[3] -> 59[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32707:33124 [3] NCCL INFO Channel 07/0 : 59[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636084:2636511 [0] NCCL INFO Channel 12/0 : 48[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153568:153982 [6] NCCL INFO Channel 01/0 : 94[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153569:153981 [7] NCCL INFO Channel 01/0 : 95[7] -> 63[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815965:2816395 [2] NCCL INFO Channel 03/0 : 58[2] -> 90[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32706:33122 [2] NCCL INFO Channel 11/0 : 26[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636088:2636510 [4] NCCL INFO Channel 14/0 : 52[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114513:114933 [1] NCCL INFO Channel 06/0 : 1[1] -> 65[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62883:63301 [2] NCCL INFO Channel 13/0 : 42[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815966:2816396 [3] NCCL INFO Channel 05/0 : 123[3] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32711:33127 [7] NCCL INFO Channel 03/0 : 63[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62883:63301 [2] NCCL INFO Channel 15/0 : 42[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815967:2816399 [4] NCCL INFO Channel 13/0 : 60[4] -> 92[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32710:33125 [6] NCCL INFO Channel 03/0 : 62[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32705:33128 [1] NCCL INFO Channel 11/0 : 57[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153563:153986 [1] NCCL INFO Channel 07/0 : 89[1] -> 57[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815964:2816401 [1] NCCL INFO Channel 05/0 : 121[1] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153565:153985 [3] NCCL INFO Channel 13/0 : 91[3] -> 59[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815969:2816398 [6] NCCL INFO Channel 01/0 : 126[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153569:153981 [7] NCCL INFO Channel 03/0 : 95[7] -> 63[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153568:153982 [6] NCCL INFO Channel 03/0 : 94[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815968:2816394 [5] NCCL INFO Channel 09/0 : 61[5] -> 125[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636086:2636507 [2] NCCL INFO Channel 04/0 : 50[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114513:114933 [1] NCCL INFO Channel 08/0 : 1[1] -> 65[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815965:2816395 [2] NCCL INFO Channel 05/0 : 58[2] -> 90[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815970:2816397 [7] NCCL INFO Channel 01/0 : 127[7] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32706:33122 [2] NCCL INFO Channel 13/0 : 26[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32707:33124 [3] NCCL INFO Channel 09/0 : 59[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815967:2816399 [4] NCCL INFO Channel 15/0 : 60[4] -> 92[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32711:33127 [7] NCCL INFO Channel 05/0 : 63[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32705:33128 [1] NCCL INFO Channel 13/0 : 57[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153563:153986 [1] NCCL INFO Channel 11/0 : 89[1] -> 57[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32710:33125 [6] NCCL INFO Channel 05/0 : 62[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636091:2636512 [7] NCCL INFO Channel 04/0 : 55[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636087:2636506 [3] NCCL INFO Channel 12/0 : 51[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636084:2636511 [0] NCCL INFO Channel 14/0 : 48[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153565:153985 [3] NCCL INFO Channel 15/0 : 91[3] -> 59[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114516:114934 [4] NCCL INFO Channel 00/0 : 4[4] -> 68[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32704:33123 [0] NCCL INFO Channel 01/0 : 56[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636086:2636507 [2] NCCL INFO Channel 06/0 : 50[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815966:2816396 [3] NCCL INFO Channel 07/0 : 123[3] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32707:33124 [3] NCCL INFO Channel 13/0 : 59[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32706:33122 [2] NCCL INFO Channel 15/0 : 26[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636091:2636512 [7] NCCL INFO Channel 06/0 : 55[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815964:2816401 [1] NCCL INFO Channel 07/0 : 121[1] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32711:33127 [7] NCCL INFO Channel 07/0 : 63[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815969:2816398 [6] NCCL INFO Channel 03/0 : 126[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815968:2816394 [5] NCCL INFO Channel 11/0 : 61[5] -> 125[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153562:153983 [0] NCCL INFO Channel 01/0 : 88[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153569:153981 [7] NCCL INFO Channel 05/0 : 95[7] -> 63[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114513:114933 [1] NCCL INFO Channel 10/0 : 1[1] -> 65[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815965:2816395 [2] NCCL INFO Channel 07/0 : 58[2] -> 90[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153568:153982 [6] NCCL INFO Channel 05/0 : 94[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114516:114934 [4] NCCL INFO Channel 02/0 : 4[4] -> 68[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815970:2816397 [7] NCCL INFO Channel 03/0 : 127[7] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153563:153986 [1] NCCL INFO Channel 13/0 : 89[1] -> 57[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815966:2816396 [3] NCCL INFO Channel 09/0 : 123[3] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636087:2636506 [3] NCCL INFO Channel 14/0 : 51[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636086:2636507 [2] NCCL INFO Channel 08/0 : 50[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153562:153983 [0] NCCL INFO Channel 03/0 : 88[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815963:2816400 [0] NCCL INFO Channel 01/0 : 120[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636091:2636512 [7] NCCL INFO Channel 08/0 : 55[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815964:2816401 [1] NCCL INFO Channel 11/0 : 121[1] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114513:114933 [1] NCCL INFO Channel 12/0 : 1[1] -> 65[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815969:2816398 [6] NCCL INFO Channel 05/0 : 126[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815968:2816394 [5] NCCL INFO Channel 15/0 : 61[5] -> 125[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114516:114934 [4] NCCL INFO Channel 06/0 : 4[4] -> 68[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153563:153986 [1] NCCL INFO Channel 15/0 : 89[1] -> 57[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153569:153981 [7] NCCL INFO Channel 07/0 : 95[7] -> 63[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153568:153982 [6] NCCL INFO Channel 07/0 : 94[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636086:2636507 [2] NCCL INFO Channel 10/0 : 50[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96566:97010 [3] NCCL INFO Channel 12/0 : 35[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153562:153983 [0] NCCL INFO Channel 05/0 : 88[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636091:2636512 [7] NCCL INFO Channel 10/0 : 55[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96565:97004 [2] NCCL INFO Channel 10/0 : 34[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815965:2816395 [2] NCCL INFO Channel 09/0 : 58[2] -> 90[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636086:2636507 [2] NCCL INFO Channel 12/0 : 50[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96566:97010 [3] NCCL INFO Channel 14/0 : 35[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815970:2816397 [7] NCCL INFO Channel 05/0 : 127[7] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96565:97004 [2] NCCL INFO Channel 12/0 : 34[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114513:114933 [1] NCCL INFO Channel 14/0 : 1[1] -> 65[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815966:2816396 [3] NCCL INFO Channel 13/0 : 123[3] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815963:2816400 [0] NCCL INFO Channel 03/0 : 120[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96565:97004 [2] NCCL INFO Channel 14/0 : 34[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153569:153981 [7] NCCL INFO Channel 09/0 : 95[7] -> 63[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153568:153982 [6] NCCL INFO Channel 09/0 : 94[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114516:114934 [4] NCCL INFO Channel 08/0 : 4[4] -> 68[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815964:2816401 [1] NCCL INFO Channel 13/0 : 121[1] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96568:97007 [5] NCCL INFO Channel 00/0 : 37[5] -> 69[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153562:153983 [0] NCCL INFO Channel 07/0 : 88[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815969:2816398 [6] NCCL INFO Channel 07/0 : 126[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636091:2636512 [7] NCCL INFO Channel 12/0 : 55[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96568:97007 [5] NCCL INFO Channel 02/0 : 37[5] -> 69[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153566:153980 [4] NCCL INFO Channel 01/0 : 92[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815970:2816397 [7] NCCL INFO Channel 07/0 : 127[7] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636086:2636507 [2] NCCL INFO Channel 14/0 : 50[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96568:97007 [5] NCCL INFO Channel 04/0 : 37[5] -> 69[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96568:97007 [5] NCCL INFO Channel 06/0 : 37[5] -> 69[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114513:114933 [1] NCCL INFO Channel 00/0 : 65[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-52:2722050:2722471 [5] NCCL INFO Channel 00/0 : 125[5] -> 117[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96568:97007 [5] NCCL INFO Channel 08/0 : 37[5] -> 69[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114516:114934 [4] NCCL INFO Channel 10/0 : 4[4] -> 68[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96568:97007 [5] NCCL INFO Channel 10/0 : 37[5] -> 69[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153569:153981 [7] NCCL INFO Channel 11/0 : 95[7] -> 63[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815966:2816396 [3] NCCL INFO Channel 15/0 : 123[3] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636091:2636512 [7] NCCL INFO Channel 14/0 : 55[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96568:97007 [5] NCCL INFO Channel 12/0 : 37[5] -> 69[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153568:153982 [6] NCCL INFO Channel 11/0 : 94[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815963:2816400 [0] NCCL INFO Channel 05/0 : 120[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96568:97007 [5] NCCL INFO Channel 14/0 : 37[5] -> 69[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153562:153983 [0] NCCL INFO Channel 09/0 : 88[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815969:2816398 [6] NCCL INFO Channel 09/0 : 126[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815964:2816401 [1] NCCL INFO Channel 15/0 : 121[1] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-52:2722050:2722471 [5] NCCL INFO Channel 02/0 : 125[5] -> 117[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96569:97006 [6] NCCL INFO Channel 00/0 : 38[6] -> 70[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153566:153980 [4] NCCL INFO Channel 03/0 : 92[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114513:114933 [1] NCCL INFO Channel 02/0 : 65[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815965:2816395 [2] NCCL INFO Channel 11/0 : 58[2] -> 90[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96564:97009 [1] NCCL INFO Channel 00/0 : 33[1] -> 65[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114516:114934 [4] NCCL INFO Channel 12/0 : 4[4] -> 68[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815967:2816399 [4] NCCL INFO Channel 01/0 : 124[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96569:97006 [6] NCCL INFO Channel 02/0 : 38[6] -> 70[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114513:114933 [1] NCCL INFO Channel 04/0 : 65[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815968:2816394 [5] NCCL INFO Channel 01/0 : 93[5] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96564:97009 [1] NCCL INFO Channel 02/0 : 33[1] -> 65[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815970:2816397 [7] NCCL INFO Channel 09/0 : 127[7] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96569:97006 [6] NCCL INFO Channel 04/0 : 38[6] -> 70[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815966:2816396 [3] NCCL INFO Channel 01/0 : 59[3] -> 123[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-52:2722050:2722471 [5] NCCL INFO Channel 04/0 : 125[5] -> 117[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96564:97009 [1] NCCL INFO Channel 04/0 : 33[1] -> 65[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153569:153981 [7] NCCL INFO Channel 13/0 : 95[7] -> 63[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815963:2816400 [0] NCCL INFO Channel 07/0 : 120[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96569:97006 [6] NCCL INFO Channel 08/0 : 38[6] -> 70[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153568:153982 [6] NCCL INFO Channel 13/0 : 94[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153562:153983 [0] NCCL INFO Channel 11/0 : 88[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96564:97009 [1] NCCL INFO Channel 06/0 : 33[1] -> 65[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153566:153980 [4] NCCL INFO Channel 05/0 : 92[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114516:114934 [4] NCCL INFO Channel 14/0 : 4[4] -> 68[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96569:97006 [6] NCCL INFO Channel 10/0 : 38[6] -> 70[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114513:114933 [1] NCCL INFO Channel 06/0 : 65[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-52:2722050:2722471 [5] NCCL INFO Channel 06/0 : 125[5] -> 117[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96564:97009 [1] NCCL INFO Channel 08/0 : 33[1] -> 65[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114516:114934 [4] NCCL INFO Channel 00/0 : 68[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815969:2816398 [6] NCCL INFO Channel 11/0 : 126[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96569:97006 [6] NCCL INFO Channel 12/0 : 38[6] -> 70[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815964:2816401 [1] NCCL INFO Channel 01/0 : 57[1] -> 121[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96566:97010 [3] NCCL INFO Channel 00/0 : 35[3] -> 67[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815967:2816399 [4] NCCL INFO Channel 03/0 : 124[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815965:2816395 [2] NCCL INFO Channel 13/0 : 58[2] -> 90[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96567:97008 [4] NCCL INFO Channel 00/0 : 36[4] -> 68[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153562:153983 [0] NCCL INFO Channel 13/0 : 88[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153568:153982 [6] NCCL INFO Channel 15/0 : 94[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153566:153980 [4] NCCL INFO Channel 07/0 : 92[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815968:2816394 [5] NCCL INFO Channel 03/0 : 93[5] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815966:2816396 [3] NCCL INFO Channel 03/0 : 59[3] -> 123[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96569:97006 [6] NCCL INFO Channel 14/0 : 38[6] -> 70[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96564:97009 [1] NCCL INFO Channel 10/0 : 33[1] -> 65[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96563:97003 [0] NCCL INFO Channel 02/0 : 32[0] -> 64[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153562:153983 [0] NCCL INFO Channel 15/0 : 88[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815970:2816397 [7] NCCL INFO Channel 11/0 : 127[7] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722050:2722471 [5] NCCL INFO Channel 08/0 : 125[5] -> 117[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96566:97010 [3] NCCL INFO Channel 02/0 : 35[3] -> 67[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114513:114933 [1] NCCL INFO Channel 08/0 : 65[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815963:2816400 [0] NCCL INFO Channel 09/0 : 120[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96567:97008 [4] NCCL INFO Channel 02/0 : 36[4] -> 68[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114516:114934 [4] NCCL INFO Channel 02/0 : 68[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815964:2816401 [1] NCCL INFO Channel 03/0 : 57[1] -> 121[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96563:97003 [0] NCCL INFO Channel 04/0 : 32[0] -> 64[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96564:97009 [1] NCCL INFO Channel 12/0 : 33[1] -> 65[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153566:153980 [4] NCCL INFO Channel 09/0 : 92[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815969:2816398 [6] NCCL INFO Channel 13/0 : 126[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96566:97010 [3] NCCL INFO Channel 04/0 : 35[3] -> 67[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153566:153980 [4] NCCL INFO Channel 11/0 : 92[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815965:2816395 [2] NCCL INFO Channel 15/0 : 58[2] -> 90[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-52:2722050:2722471 [5] NCCL INFO Channel 10/0 : 125[5] -> 117[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96567:97008 [4] NCCL INFO Channel 06/0 : 36[4] -> 68[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96564:97009 [1] NCCL INFO Channel 14/0 : 33[1] -> 65[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96563:97003 [0] NCCL INFO Channel 06/0 : 32[0] -> 64[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114513:114933 [1] NCCL INFO Channel 10/0 : 65[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96566:97010 [3] NCCL INFO Channel 06/0 : 35[3] -> 67[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114516:114934 [4] NCCL INFO Channel 06/0 : 68[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96567:97008 [4] NCCL INFO Channel 08/0 : 36[4] -> 68[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114517:114935 [5] NCCL INFO Channel 00/0 : 5[5] -> 69[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96569:97006 [6] NCCL INFO Channel 00/0 : 70[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153566:153980 [4] NCCL INFO Channel 13/0 : 92[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114513:114933 [1] NCCL INFO Channel 12/0 : 65[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815967:2816399 [4] NCCL INFO Channel 05/0 : 124[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815966:2816396 [3] NCCL INFO Channel 05/0 : 59[3] -> 123[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-52:2722050:2722471 [5] NCCL INFO Channel 12/0 : 125[5] -> 117[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96566:97010 [3] NCCL INFO Channel 08/0 : 35[3] -> 67[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96563:97003 [0] NCCL INFO Channel 08/0 : 32[0] -> 64[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96567:97008 [4] NCCL INFO Channel 10/0 : 36[4] -> 68[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114516:114934 [4] NCCL INFO Channel 08/0 : 68[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815968:2816394 [5] NCCL INFO Channel 05/0 : 93[5] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96569:97006 [6] NCCL INFO Channel 02/0 : 70[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815963:2816400 [0] NCCL INFO Channel 11/0 : 120[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815970:2816397 [7] NCCL INFO Channel 13/0 : 127[7] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96563:97003 [0] NCCL INFO Channel 10/0 : 32[0] -> 64[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96567:97008 [4] NCCL INFO Channel 12/0 : 36[4] -> 68[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815969:2816398 [6] NCCL INFO Channel 15/0 : 126[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96568:97007 [5] NCCL INFO Channel 00/0 : 69[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815964:2816401 [1] NCCL INFO Channel 05/0 : 57[1] -> 121[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96569:97006 [6] NCCL INFO Channel 04/0 : 70[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153566:153980 [4] NCCL INFO Channel 15/0 : 92[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815967:2816399 [4] NCCL INFO Channel 07/0 : 124[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96566:97010 [3] NCCL INFO Channel 10/0 : 35[3] -> 67[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114513:114933 [1] NCCL INFO Channel 14/0 : 65[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815966:2816396 [3] NCCL INFO Channel 07/0 : 59[3] -> 123[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-52:2722050:2722471 [5] NCCL INFO Channel 14/0 : 125[5] -> 117[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96564:97009 [1] NCCL INFO Channel 00/0 : 65[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114517:114935 [5] NCCL INFO Channel 02/0 : 5[5] -> 69[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114516:114934 [4] NCCL INFO Channel 10/0 : 68[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815963:2816400 [0] NCCL INFO Channel 13/0 : 120[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96563:97003 [0] NCCL INFO Channel 12/0 : 32[0] -> 64[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96567:97008 [4] NCCL INFO Channel 14/0 : 36[4] -> 68[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815968:2816394 [5] NCCL INFO Channel 07/0 : 93[5] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815970:2816397 [7] NCCL INFO Channel 01/0 : 63[7] -> 127[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96569:97006 [6] NCCL INFO Channel 08/0 : 70[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96570:97005 [7] NCCL INFO Channel 00/0 : 39[7] -> 71[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96568:97007 [5] NCCL INFO Channel 02/0 : 69[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815969:2816398 [6] NCCL INFO Channel 01/0 : 62[6] -> 126[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815964:2816401 [1] NCCL INFO Channel 07/0 : 57[1] -> 121[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96566:97010 [3] NCCL INFO Channel 12/0 : 35[3] -> 67[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96564:97009 [1] NCCL INFO Channel 02/0 : 65[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96563:97003 [0] NCCL INFO Channel 14/0 : 32[0] -> 64[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114516:114934 [4] NCCL INFO Channel 12/0 : 68[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96565:97004 [2] NCCL INFO Channel 00/0 : 34[2] -> 66[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114517:114935 [5] NCCL INFO Channel 04/0 : 5[5] -> 69[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96569:97006 [6] NCCL INFO Channel 10/0 : 70[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96568:97007 [5] NCCL INFO Channel 04/0 : 69[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815967:2816399 [4] NCCL INFO Channel 09/0 : 124[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96570:97005 [7] NCCL INFO Channel 02/0 : 39[7] -> 71[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815966:2816396 [3] NCCL INFO Channel 09/0 : 59[3] -> 123[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96566:97010 [3] NCCL INFO Channel 14/0 : 35[3] -> 67[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96564:97009 [1] NCCL INFO Channel 04/0 : 65[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815963:2816400 [0] NCCL INFO Channel 15/0 : 120[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96565:97004 [2] NCCL INFO Channel 04/0 : 34[2] -> 66[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153564:153984 [2] NCCL INFO Channel 01/0 : 90[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815968:2816394 [5] NCCL INFO Channel 09/0 : 93[5] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96567:97008 [4] NCCL INFO Channel 00/0 : 68[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114516:114934 [4] NCCL INFO Channel 14/0 : 68[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815970:2816397 [7] NCCL INFO Channel 03/0 : 63[7] -> 127[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96569:97006 [6] NCCL INFO Channel 12/0 : 70[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815964:2816401 [1] NCCL INFO Channel 11/0 : 57[1] -> 121[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815969:2816398 [6] NCCL INFO Channel 03/0 : 62[6] -> 126[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96568:97007 [5] NCCL INFO Channel 06/0 : 69[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96570:97005 [7] NCCL INFO Channel 04/0 : 39[7] -> 71[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815967:2816399 [4] NCCL INFO Channel 11/0 : 124[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96563:97003 [0] NCCL INFO Channel 02/0 : 64[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815966:2816396 [3] NCCL INFO Channel 13/0 : 59[3] -> 123[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96564:97009 [1] NCCL INFO Channel 06/0 : 65[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153564:153984 [2] NCCL INFO Channel 03/0 : 90[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815963:2816400 [0] NCCL INFO Channel 01/0 : 56[0] -> 120[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:124161:124582 [1] NCCL INFO Channel 01/0 : 9[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96567:97008 [4] NCCL INFO Channel 02/0 : 68[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114514:114932 [2] NCCL INFO Channel 00/0 : 2[2] -> 66[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114517:114935 [5] NCCL INFO Channel 06/0 : 5[5] -> 69[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815965:2816395 [2] NCCL INFO Channel 01/0 : 122[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96565:97004 [2] NCCL INFO Channel 06/0 : 34[2] -> 66[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114513:114933 [1] NCCL INFO Channel 00/0 : 97[1] -> 65[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96569:97006 [6] NCCL INFO Channel 14/0 : 70[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153564:153984 [2] NCCL INFO Channel 05/0 : 90[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:124161:124582 [1] NCCL INFO Channel 03/0 : 9[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96568:97007 [5] NCCL INFO Channel 08/0 : 69[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96570:97005 [7] NCCL INFO Channel 06/0 : 39[7] -> 71[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815968:2816394 [5] NCCL INFO Channel 11/0 : 93[5] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96563:97003 [0] NCCL INFO Channel 04/0 : 64[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96566:97010 [3] NCCL INFO Channel 00/0 : 67[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114514:114932 [2] NCCL INFO Channel 04/0 : 2[2] -> 66[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815970:2816397 [7] NCCL INFO Channel 05/0 : 63[7] -> 127[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96564:97009 [1] NCCL INFO Channel 08/0 : 65[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96567:97008 [4] NCCL INFO Channel 06/0 : 68[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114517:114935 [5] NCCL INFO Channel 08/0 : 5[5] -> 69[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815969:2816398 [6] NCCL INFO Channel 05/0 : 62[6] -> 126[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96565:97004 [2] NCCL INFO Channel 08/0 : 34[2] -> 66[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153564:153984 [2] NCCL INFO Channel 07/0 : 90[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114513:114933 [1] NCCL INFO Channel 02/0 : 97[1] -> 65[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815964:2816401 [1] NCCL INFO Channel 13/0 : 57[1] -> 121[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124161:124582 [1] NCCL INFO Channel 05/0 : 9[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96568:97007 [5] NCCL INFO Channel 10/0 : 69[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114514:114932 [2] NCCL INFO Channel 06/0 : 2[2] -> 66[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815966:2816396 [3] NCCL INFO Channel 15/0 : 59[3] -> 123[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815963:2816400 [0] NCCL INFO Channel 03/0 : 56[0] -> 120[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96570:97005 [7] NCCL INFO Channel 08/0 : 39[7] -> 71[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815967:2816399 [4] NCCL INFO Channel 13/0 : 124[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815965:2816395 [2] NCCL INFO Channel 03/0 : 122[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:124164:124585 [4] NCCL INFO Channel 01/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815968:2816394 [5] NCCL INFO Channel 15/0 : 93[5] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153564:153984 [2] NCCL INFO Channel 09/0 : 90[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815970:2816397 [7] NCCL INFO Channel 07/0 : 63[7] -> 127[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114517:114935 [5] NCCL INFO Channel 10/0 : 5[5] -> 69[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815964:2816401 [1] NCCL INFO Channel 15/0 : 57[1] -> 121[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124161:124582 [1] NCCL INFO Channel 07/0 : 9[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96563:97003 [0] NCCL INFO Channel 06/0 : 64[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114513:114933 [1] NCCL INFO Channel 04/0 : 97[1] -> 65[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815969:2816398 [6] NCCL INFO Channel 07/0 : 62[6] -> 126[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96564:97009 [1] NCCL INFO Channel 10/0 : 65[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96567:97008 [4] NCCL INFO Channel 08/0 : 68[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114516:114934 [4] NCCL INFO Channel 00/0 : 100[4] -> 68[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114514:114932 [2] NCCL INFO Channel 08/0 : 2[2] -> 66[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96565:97004 [2] NCCL INFO Channel 10/0 : 34[2] -> 66[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153564:153984 [2] NCCL INFO Channel 11/0 : 90[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114517:114935 [5] NCCL INFO Channel 12/0 : 5[5] -> 69[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114513:114933 [1] NCCL INFO Channel 06/0 : 97[1] -> 65[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124164:124585 [4] NCCL INFO Channel 03/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96566:97010 [3] NCCL INFO Channel 02/0 : 67[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96570:97005 [7] NCCL INFO Channel 10/0 : 39[7] -> 71[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815963:2816400 [0] NCCL INFO Channel 05/0 : 56[0] -> 120[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:124161:124582 [1] NCCL INFO Channel 11/0 : 9[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-52:2722048:2722469 [3] NCCL INFO Channel 00/0 : 123[3] -> 115[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96568:97007 [5] NCCL INFO Channel 12/0 : 69[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815965:2816395 [2] NCCL INFO Channel 05/0 : 122[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815968:2816394 [5] NCCL INFO Channel 01/0 : 61[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815967:2816399 [4] NCCL INFO Channel 15/0 : 124[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96563:97003 [0] NCCL INFO Channel 08/0 : 64[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153564:153984 [2] NCCL INFO Channel 13/0 : 90[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114514:114932 [2] NCCL INFO Channel 10/0 : 2[2] -> 66[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114516:114934 [4] NCCL INFO Channel 02/0 : 100[4] -> 68[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815970:2816397 [7] NCCL INFO Channel 09/0 : 63[7] -> 127[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124164:124585 [4] NCCL INFO Channel 05/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96564:97009 [1] NCCL INFO Channel 12/0 : 65[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114513:114933 [1] NCCL INFO Channel 08/0 : 97[1] -> 65[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114517:114935 [5] NCCL INFO Channel 14/0 : 5[5] -> 69[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815969:2816398 [6] NCCL INFO Channel 09/0 : 62[6] -> 126[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:124161:124582 [1] NCCL INFO Channel 13/0 : 9[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-52:2722046:2722466 [1] NCCL INFO Channel 00/0 : 121[1] -> 113[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96567:97008 [4] NCCL INFO Channel 10/0 : 68[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114514:114932 [2] NCCL INFO Channel 12/0 : 2[2] -> 66[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114516:114934 [4] NCCL INFO Channel 06/0 : 100[4] -> 68[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815963:2816400 [0] NCCL INFO Channel 07/0 : 56[0] -> 120[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-52:2722048:2722469 [3] NCCL INFO Channel 02/0 : 123[3] -> 115[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96570:97005 [7] NCCL INFO Channel 12/0 : 39[7] -> 71[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96568:97007 [5] NCCL INFO Channel 14/0 : 69[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96566:97010 [3] NCCL INFO Channel 04/0 : 67[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153564:153984 [2] NCCL INFO Channel 15/0 : 90[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815968:2816394 [5] NCCL INFO Channel 03/0 : 61[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124164:124585 [4] NCCL INFO Channel 07/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96565:97004 [2] NCCL INFO Channel 12/0 : 34[2] -> 66[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96563:97003 [0] NCCL INFO Channel 10/0 : 64[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815965:2816395 [2] NCCL INFO Channel 07/0 : 122[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96564:97009 [1] NCCL INFO Channel 14/0 : 65[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114513:114933 [1] NCCL INFO Channel 10/0 : 97[1] -> 65[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114517:114935 [5] NCCL INFO Channel 00/0 : 69[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815967:2816399 [4] NCCL INFO Channel 01/0 : 60[4] -> 124[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722046:2722466 [1] NCCL INFO Channel 02/0 : 121[1] -> 113[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96567:97008 [4] NCCL INFO Channel 12/0 : 68[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114514:114932 [2] NCCL INFO Channel 14/0 : 2[2] -> 66[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114516:114934 [4] NCCL INFO Channel 08/0 : 100[4] -> 68[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815969:2816398 [6] NCCL INFO Channel 11/0 : 62[6] -> 126[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815970:2816397 [7] NCCL INFO Channel 11/0 : 63[7] -> 127[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815966:2816396 [3] NCCL INFO Channel 01/0 : 91[3] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815963:2816400 [0] NCCL INFO Channel 09/0 : 56[0] -> 120[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-52:2722048:2722469 [3] NCCL INFO Channel 04/0 : 123[3] -> 115[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96570:97005 [7] NCCL INFO Channel 14/0 : 39[7] -> 71[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114513:114933 [1] NCCL INFO Channel 12/0 : 97[1] -> 65[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114517:114935 [5] NCCL INFO Channel 02/0 : 69[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815964:2816401 [1] NCCL INFO Channel 01/0 : 89[1] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815968:2816394 [5] NCCL INFO Channel 05/0 : 61[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96565:97004 [2] NCCL INFO Channel 14/0 : 34[2] -> 66[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96566:97010 [3] NCCL INFO Channel 06/0 : 67[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96567:97008 [4] NCCL INFO Channel 14/0 : 68[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96563:97003 [0] NCCL INFO Channel 12/0 : 64[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96566:97010 [3] NCCL INFO Channel 08/0 : 67[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-52:2722046:2722466 [1] NCCL INFO Channel 04/0 : 121[1] -> 113[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96563:97003 [0] NCCL INFO Channel 14/0 : 64[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114514:114932 [2] NCCL INFO Channel 00/0 : 66[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114516:114934 [4] NCCL INFO Channel 10/0 : 100[4] -> 68[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815965:2816395 [2] NCCL INFO Channel 09/0 : 122[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-52:2722048:2722469 [3] NCCL INFO Channel 06/0 : 123[3] -> 115[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96566:97010 [3] NCCL INFO Channel 10/0 : 67[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114513:114933 [1] NCCL INFO Channel 14/0 : 97[1] -> 65[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114517:114935 [5] NCCL INFO Channel 04/0 : 69[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815967:2816399 [4] NCCL INFO Channel 03/0 : 60[4] -> 124[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96570:97005 [7] NCCL INFO Channel 00/0 : 71[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815963:2816400 [0] NCCL INFO Channel 11/0 : 56[0] -> 120[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815969:2816398 [6] NCCL INFO Channel 13/0 : 62[6] -> 126[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96566:97010 [3] NCCL INFO Channel 12/0 : 67[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815966:2816396 [3] NCCL INFO Channel 03/0 : 91[3] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96565:97004 [2] NCCL INFO Channel 00/0 : 66[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815970:2816397 [7] NCCL INFO Channel 13/0 : 63[7] -> 127[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96570:97005 [7] NCCL INFO Channel 02/0 : 71[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815964:2816401 [1] NCCL INFO Channel 03/0 : 89[1] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815968:2816394 [5] NCCL INFO Channel 07/0 : 61[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815965:2816395 [2] NCCL INFO Channel 11/0 : 122[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-52:2722046:2722466 [1] NCCL INFO Channel 06/0 : 121[1] -> 113[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96566:97010 [3] NCCL INFO Channel 14/0 : 67[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114514:114932 [2] NCCL INFO Channel 04/0 : 66[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815967:2816399 [4] NCCL INFO Channel 05/0 : 60[4] -> 124[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722048:2722469 [3] NCCL INFO Channel 08/0 : 123[3] -> 115[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96565:97004 [2] NCCL INFO Channel 04/0 : 66[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114516:114934 [4] NCCL INFO Channel 12/0 : 100[4] -> 68[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114512:114930 [0] NCCL INFO Channel 02/0 : 0[0] -> 64[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815963:2816400 [0] NCCL INFO Channel 13/0 : 56[0] -> 120[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-52:2722046:2722466 [1] NCCL INFO Channel 08/0 : 121[1] -> 113[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96570:97005 [7] NCCL INFO Channel 04/0 : 71[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114513:114933 [1] NCCL INFO Channel 00/0 : 65[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815969:2816398 [6] NCCL INFO Channel 15/0 : 62[6] -> 126[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815966:2816396 [3] NCCL INFO Channel 05/0 : 91[3] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96565:97004 [2] NCCL INFO Channel 06/0 : 66[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114517:114935 [5] NCCL INFO Channel 06/0 : 69[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96570:97005 [7] NCCL INFO Channel 06/0 : 71[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114514:114932 [2] NCCL INFO Channel 06/0 : 66[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96565:97004 [2] NCCL INFO Channel 08/0 : 66[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-52:2722048:2722469 [3] NCCL INFO Channel 10/0 : 123[3] -> 115[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96570:97005 [7] NCCL INFO Channel 08/0 : 71[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722046:2722466 [1] NCCL INFO Channel 10/0 : 121[1] -> 113[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96565:97004 [2] NCCL INFO Channel 10/0 : 66[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815968:2816394 [5] NCCL INFO Channel 09/0 : 61[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96570:97005 [7] NCCL INFO Channel 10/0 : 71[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815965:2816395 [2] NCCL INFO Channel 13/0 : 122[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96565:97004 [2] NCCL INFO Channel 12/0 : 66[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96570:97005 [7] NCCL INFO Channel 12/0 : 71[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114516:114934 [4] NCCL INFO Channel 14/0 : 100[4] -> 68[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815963:2816400 [0] NCCL INFO Channel 15/0 : 56[0] -> 120[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815964:2816401 [1] NCCL INFO Channel 05/0 : 89[1] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815967:2816399 [4] NCCL INFO Channel 07/0 : 60[4] -> 124[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114513:114933 [1] NCCL INFO Channel 02/0 : 65[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815966:2816396 [3] NCCL INFO Channel 07/0 : 91[3] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114512:114930 [0] NCCL INFO Channel 04/0 : 0[0] -> 64[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815970:2816397 [7] NCCL INFO Channel 01/0 : 95[7] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722052:2722464 [7] NCCL INFO Channel 00/0 : 127[7] -> 119[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114517:114935 [5] NCCL INFO Channel 08/0 : 69[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815968:2816394 [5] NCCL INFO Channel 11/0 : 61[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-52:2722046:2722466 [1] NCCL INFO Channel 12/0 : 121[1] -> 113[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114514:114932 [2] NCCL INFO Channel 08/0 : 66[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815965:2816395 [2] NCCL INFO Channel 15/0 : 122[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96565:97004 [2] NCCL INFO Channel 14/0 : 66[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96570:97005 [7] NCCL INFO Channel 14/0 : 71[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114516:114934 [4] NCCL INFO Channel 00/0 : 68[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114518:114937 [6] NCCL INFO Channel 00/0 : 6[6] -> 70[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815964:2816401 [1] NCCL INFO Channel 07/0 : 89[1] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815967:2816399 [4] NCCL INFO Channel 09/0 : 60[4] -> 124[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722048:2722469 [3] NCCL INFO Channel 12/0 : 123[3] -> 115[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-52:2722052:2722464 [7] NCCL INFO Channel 02/0 : 127[7] -> 119[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722051:2722465 [6] NCCL INFO Channel 00/0 : 126[6] -> 118[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-52:2722046:2722466 [1] NCCL INFO Channel 14/0 : 121[1] -> 113[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114513:114933 [1] NCCL INFO Channel 04/0 : 65[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114515:114936 [3] NCCL INFO Channel 00/0 : 3[3] -> 67[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114514:114932 [2] NCCL INFO Channel 10/0 : 66[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114512:114930 [0] NCCL INFO Channel 06/0 : 0[0] -> 64[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114517:114935 [5] NCCL INFO Channel 10/0 : 69[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114516:114934 [4] NCCL INFO Channel 02/0 : 68[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815966:2816396 [3] NCCL INFO Channel 09/0 : 91[3] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815970:2816397 [7] NCCL INFO Channel 03/0 : 95[7] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114513:114933 [1] NCCL INFO Channel 06/0 : 65[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815969:2816398 [6] NCCL INFO Channel 01/0 : 94[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815968:2816394 [5] NCCL INFO Channel 15/0 : 61[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-52:2722048:2722469 [3] NCCL INFO Channel 14/0 : 123[3] -> 115[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815965:2816395 [2] NCCL INFO Channel 01/0 : 58[2] -> 122[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-52:2722052:2722464 [7] NCCL INFO Channel 04/0 : 127[7] -> 119[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815967:2816399 [4] NCCL INFO Channel 11/0 : 60[4] -> 124[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722051:2722465 [6] NCCL INFO Channel 02/0 : 126[6] -> 118[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-52:2722045:2722470 [0] NCCL INFO Channel 02/0 : 120[0] -> 112[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815964:2816401 [1] NCCL INFO Channel 11/0 : 89[1] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114514:114932 [2] NCCL INFO Channel 12/0 : 66[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815966:2816396 [3] NCCL INFO Channel 13/0 : 91[3] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114519:114931 [7] NCCL INFO Channel 00/0 : 7[7] -> 71[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114512:114930 [0] NCCL INFO Channel 08/0 : 0[0] -> 64[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815963:2816400 [0] NCCL INFO Channel 01/0 : 88[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114513:114933 [1] NCCL INFO Channel 08/0 : 65[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114517:114935 [5] NCCL INFO Channel 12/0 : 69[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114516:114934 [4] NCCL INFO Channel 06/0 : 68[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114515:114936 [3] NCCL INFO Channel 02/0 : 3[3] -> 67[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815970:2816397 [7] NCCL INFO Channel 05/0 : 95[7] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815969:2816398 [6] NCCL INFO Channel 03/0 : 94[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153567:153979 [5] NCCL INFO Channel 01/0 : 109[5] -> 93[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815965:2816395 [2] NCCL INFO Channel 03/0 : 58[2] -> 122[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-52:2722052:2722464 [7] NCCL INFO Channel 06/0 : 127[7] -> 119[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815967:2816399 [4] NCCL INFO Channel 13/0 : 60[4] -> 124[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722051:2722465 [6] NCCL INFO Channel 04/0 : 126[6] -> 118[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114514:114932 [2] NCCL INFO Channel 14/0 : 66[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815964:2816401 [1] NCCL INFO Channel 13/0 : 89[1] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-52:2722045:2722470 [0] NCCL INFO Channel 04/0 : 120[0] -> 112[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153567:153979 [5] NCCL INFO Channel 03/0 : 109[5] -> 93[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114513:114933 [1] NCCL INFO Channel 10/0 : 65[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114512:114930 [0] NCCL INFO Channel 10/0 : 0[0] -> 64[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114516:114934 [4] NCCL INFO Channel 08/0 : 68[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114517:114935 [5] NCCL INFO Channel 14/0 : 69[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815966:2816396 [3] NCCL INFO Channel 15/0 : 91[3] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-52:2722052:2722464 [7] NCCL INFO Channel 08/0 : 127[7] -> 119[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114515:114936 [3] NCCL INFO Channel 04/0 : 3[3] -> 67[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815963:2816400 [0] NCCL INFO Channel 03/0 : 88[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-52:2722051:2722465 [6] NCCL INFO Channel 08/0 : 126[6] -> 118[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114519:114931 [7] NCCL INFO Channel 02/0 : 7[7] -> 71[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815970:2816397 [7] NCCL INFO Channel 07/0 : 95[7] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815965:2816395 [2] NCCL INFO Channel 05/0 : 58[2] -> 122[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-52:2722045:2722470 [0] NCCL INFO Channel 06/0 : 120[0] -> 112[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153567:153979 [5] NCCL INFO Channel 05/0 : 109[5] -> 93[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815969:2816398 [6] NCCL INFO Channel 05/0 : 94[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-52:2722052:2722464 [7] NCCL INFO Channel 10/0 : 127[7] -> 119[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815967:2816399 [4] NCCL INFO Channel 15/0 : 60[4] -> 124[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722051:2722465 [6] NCCL INFO Channel 10/0 : 126[6] -> 118[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815964:2816401 [1] NCCL INFO Channel 15/0 : 89[1] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114513:114933 [1] NCCL INFO Channel 12/0 : 65[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815963:2816400 [0] NCCL INFO Channel 05/0 : 88[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114516:114934 [4] NCCL INFO Channel 10/0 : 68[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114512:114930 [0] NCCL INFO Channel 12/0 : 0[0] -> 64[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815966:2816396 [3] NCCL INFO Channel 01/0 : 59[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815965:2816395 [2] NCCL INFO Channel 07/0 : 58[2] -> 122[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153567:153979 [5] NCCL INFO Channel 07/0 : 109[5] -> 93[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-52:2722045:2722470 [0] NCCL INFO Channel 08/0 : 120[0] -> 112[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-52:2722052:2722464 [7] NCCL INFO Channel 12/0 : 127[7] -> 119[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722051:2722465 [6] NCCL INFO Channel 12/0 : 126[6] -> 118[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114513:114933 [1] NCCL INFO Channel 14/0 : 65[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114515:114936 [3] NCCL INFO Channel 06/0 : 3[3] -> 67[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114519:114931 [7] NCCL INFO Channel 04/0 : 7[7] -> 71[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815969:2816398 [6] NCCL INFO Channel 07/0 : 94[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815970:2816397 [7] NCCL INFO Channel 09/0 : 95[7] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153567:153979 [5] NCCL INFO Channel 09/0 : 109[5] -> 93[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815964:2816401 [1] NCCL INFO Channel 01/0 : 57[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815963:2816400 [0] NCCL INFO Channel 07/0 : 88[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815966:2816396 [3] NCCL INFO Channel 03/0 : 59[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815968:2816394 [5] NCCL INFO Channel 00/0 : 61[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-52:2722045:2722470 [0] NCCL INFO Channel 10/0 : 120[0] -> 112[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-52:2722052:2722464 [7] NCCL INFO Channel 14/0 : 127[7] -> 119[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815965:2816395 [2] NCCL INFO Channel 09/0 : 58[2] -> 122[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-52:2722051:2722465 [6] NCCL INFO Channel 14/0 : 126[6] -> 118[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114516:114934 [4] NCCL INFO Channel 12/0 : 68[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114512:114930 [0] NCCL INFO Channel 14/0 : 0[0] -> 64[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114514:114932 [2] NCCL INFO Channel 00/0 : 98[2] -> 66[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815969:2816398 [6] NCCL INFO Channel 09/0 : 94[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-52:2722049:2722467 [4] NCCL INFO Channel 00/0 : 124[4] -> 116[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114515:114936 [3] NCCL INFO Channel 08/0 : 3[3] -> 67[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114517:114935 [5] NCCL INFO Channel 00/0 : 101[5] -> 69[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815964:2816401 [1] NCCL INFO Channel 03/0 : 57[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815970:2816397 [7] NCCL INFO Channel 11/0 : 95[7] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153567:153979 [5] NCCL INFO Channel 11/0 : 109[5] -> 93[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815967:2816399 [4] NCCL INFO Channel 01/0 : 92[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722045:2722470 [0] NCCL INFO Channel 12/0 : 120[0] -> 112[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114519:114931 [7] NCCL INFO Channel 06/0 : 7[7] -> 71[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114516:114934 [4] NCCL INFO Channel 14/0 : 68[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114512:114930 [0] NCCL INFO Channel 02/0 : 64[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114514:114932 [2] NCCL INFO Channel 04/0 : 98[2] -> 66[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557267:1557683 [1] NCCL INFO Channel 00/0 : 113[1] -> 97[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815965:2816395 [2] NCCL INFO Channel 11/0 : 58[2] -> 122[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815966:2816396 [3] NCCL INFO Channel 05/0 : 59[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:124165:124584 [5] NCCL INFO Channel 01/0 : 13[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815968:2816394 [5] NCCL INFO Channel 02/0 : 61[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-52:2722045:2722470 [0] NCCL INFO Channel 14/0 : 120[0] -> 112[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153567:153979 [5] NCCL INFO Channel 15/0 : 109[5] -> 93[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114515:114936 [3] NCCL INFO Channel 10/0 : 3[3] -> 67[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114517:114935 [5] NCCL INFO Channel 02/0 : 101[5] -> 69[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815963:2816400 [0] NCCL INFO Channel 09/0 : 88[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815969:2816398 [6] NCCL INFO Channel 11/0 : 94[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114514:114932 [2] NCCL INFO Channel 06/0 : 98[2] -> 66[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815964:2816401 [1] NCCL INFO Channel 05/0 : 57[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114512:114930 [0] NCCL INFO Channel 04/0 : 64[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114519:114931 [7] NCCL INFO Channel 08/0 : 7[7] -> 71[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815970:2816397 [7] NCCL INFO Channel 13/0 : 95[7] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124165:124584 [5] NCCL INFO Channel 03/0 : 13[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557267:1557683 [1] NCCL INFO Channel 02/0 : 113[1] -> 97[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815967:2816399 [4] NCCL INFO Channel 03/0 : 92[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153567:153979 [5] NCCL INFO Channel 01/0 : 93[5] -> 77[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815965:2816395 [2] NCCL INFO Channel 13/0 : 58[2] -> 122[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815966:2816396 [3] NCCL INFO Channel 07/0 : 59[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815968:2816394 [5] NCCL INFO Channel 04/0 : 61[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96564:97009 [1] NCCL INFO Channel 00/0 : 49[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114517:114935 [5] NCCL INFO Channel 04/0 : 101[5] -> 69[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124165:124584 [5] NCCL INFO Channel 05/0 : 13[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153567:153979 [5] NCCL INFO Channel 03/0 : 93[5] -> 77[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114515:114936 [3] NCCL INFO Channel 12/0 : 3[3] -> 67[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114514:114932 [2] NCCL INFO Channel 08/0 : 98[2] -> 66[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114512:114930 [0] NCCL INFO Channel 06/0 : 64[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114513:114933 [1] NCCL INFO Channel 01/0 : 73[1] -> 65[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815963:2816400 [0] NCCL INFO Channel 11/0 : 88[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815969:2816398 [6] NCCL INFO Channel 13/0 : 94[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557267:1557683 [1] NCCL INFO Channel 04/0 : 113[1] -> 97[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557270:1557688 [4] NCCL INFO Channel 00/0 : 116[4] -> 100[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815964:2816401 [1] NCCL INFO Channel 07/0 : 57[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815970:2816397 [7] NCCL INFO Channel 01/0 : 63[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815967:2816399 [4] NCCL INFO Channel 05/0 : 92[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:124165:124584 [5] NCCL INFO Channel 07/0 : 13[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96564:97009 [1] NCCL INFO Channel 02/0 : 49[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153567:153979 [5] NCCL INFO Channel 05/0 : 93[5] -> 77[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815965:2816395 [2] NCCL INFO Channel 15/0 : 58[2] -> 122[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114517:114935 [5] NCCL INFO Channel 06/0 : 101[5] -> 69[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114519:114931 [7] NCCL INFO Channel 10/0 : 7[7] -> 71[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815966:2816396 [3] NCCL INFO Channel 09/0 : 59[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114514:114932 [2] NCCL INFO Channel 10/0 : 98[2] -> 66[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815968:2816394 [5] NCCL INFO Channel 06/0 : 61[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114515:114936 [3] NCCL INFO Channel 14/0 : 3[3] -> 67[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815969:2816398 [6] NCCL INFO Channel 15/0 : 94[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114512:114930 [0] NCCL INFO Channel 08/0 : 64[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815963:2816400 [0] NCCL INFO Channel 13/0 : 88[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96567:97008 [4] NCCL INFO Channel 00/0 : 52[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153567:153979 [5] NCCL INFO Channel 07/0 : 93[5] -> 77[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114513:114933 [1] NCCL INFO Channel 03/0 : 73[1] -> 65[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557267:1557683 [1] NCCL INFO Channel 06/0 : 113[1] -> 97[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557270:1557688 [4] NCCL INFO Channel 02/0 : 116[4] -> 100[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815964:2816401 [1] NCCL INFO Channel 11/0 : 57[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124165:124584 [5] NCCL INFO Channel 09/0 : 13[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815970:2816397 [7] NCCL INFO Channel 03/0 : 63[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722047:2722468 [2] NCCL INFO Channel 00/0 : 122[2] -> 114[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815967:2816399 [4] NCCL INFO Channel 07/0 : 92[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96564:97009 [1] NCCL INFO Channel 04/0 : 49[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153567:153979 [5] NCCL INFO Channel 09/0 : 93[5] -> 77[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114517:114935 [5] NCCL INFO Channel 08/0 : 101[5] -> 69[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557267:1557683 [1] NCCL INFO Channel 08/0 : 113[1] -> 97[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815966:2816396 [3] NCCL INFO Channel 13/0 : 59[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:124165:124584 [5] NCCL INFO Channel 11/0 : 13[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114514:114932 [2] NCCL INFO Channel 12/0 : 98[2] -> 66[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815968:2816394 [5] NCCL INFO Channel 08/0 : 61[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-52:2722047:2722468 [2] NCCL INFO Channel 04/0 : 122[2] -> 114[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114515:114936 [3] NCCL INFO Channel 00/0 : 67[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114512:114930 [0] NCCL INFO Channel 10/0 : 64[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96567:97008 [4] NCCL INFO Channel 02/0 : 52[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153567:153979 [5] NCCL INFO Channel 11/0 : 93[5] -> 77[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114519:114931 [7] NCCL INFO Channel 12/0 : 7[7] -> 71[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114516:114934 [4] NCCL INFO Channel 01/0 : 76[4] -> 68[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557270:1557688 [4] NCCL INFO Channel 06/0 : 116[4] -> 100[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96564:97009 [1] NCCL INFO Channel 06/0 : 49[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114513:114933 [1] NCCL INFO Channel 05/0 : 73[1] -> 65[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124165:124584 [5] NCCL INFO Channel 15/0 : 13[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114517:114935 [5] NCCL INFO Channel 10/0 : 101[5] -> 69[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114514:114932 [2] NCCL INFO Channel 14/0 : 98[2] -> 66[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815969:2816398 [6] NCCL INFO Channel 01/0 : 62[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815963:2816400 [0] NCCL INFO Channel 15/0 : 88[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-52:2722047:2722468 [2] NCCL INFO Channel 06/0 : 122[2] -> 114[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114512:114930 [0] NCCL INFO Channel 12/0 : 64[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114515:114936 [3] NCCL INFO Channel 02/0 : 67[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557267:1557683 [1] NCCL INFO Channel 10/0 : 113[1] -> 97[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815964:2816401 [1] NCCL INFO Channel 13/0 : 57[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96567:97008 [4] NCCL INFO Channel 06/0 : 52[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153567:153979 [5] NCCL INFO Channel 15/0 : 93[5] -> 77[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557270:1557688 [4] NCCL INFO Channel 08/0 : 116[4] -> 100[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815970:2816397 [7] NCCL INFO Channel 05/0 : 63[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96564:97009 [1] NCCL INFO Channel 08/0 : 49[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815967:2816399 [4] NCCL INFO Channel 09/0 : 92[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815966:2816396 [3] NCCL INFO Channel 15/0 : 59[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-52:2722047:2722468 [2] NCCL INFO Channel 08/0 : 122[2] -> 114[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114516:114934 [4] NCCL INFO Channel 03/0 : 76[4] -> 68[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114513:114933 [1] NCCL INFO Channel 07/0 : 73[1] -> 65[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815965:2816395 [2] NCCL INFO Channel 01/0 : 90[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815968:2816394 [5] NCCL INFO Channel 10/0 : 61[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153565:153985 [3] NCCL INFO Channel 01/0 : 107[3] -> 91[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114519:114931 [7] NCCL INFO Channel 14/0 : 7[7] -> 71[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557267:1557683 [1] NCCL INFO Channel 12/0 : 113[1] -> 97[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815969:2816398 [6] NCCL INFO Channel 03/0 : 62[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815963:2816400 [0] NCCL INFO Channel 01/0 : 56[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96567:97008 [4] NCCL INFO Channel 08/0 : 52[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114514:114932 [2] NCCL INFO Channel 00/0 : 66[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557270:1557688 [4] NCCL INFO Channel 10/0 : 116[4] -> 100[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815964:2816401 [1] NCCL INFO Channel 15/0 : 57[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96564:97009 [1] NCCL INFO Channel 10/0 : 49[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114512:114930 [0] NCCL INFO Channel 14/0 : 64[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114515:114936 [3] NCCL INFO Channel 04/0 : 67[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815970:2816397 [7] NCCL INFO Channel 07/0 : 63[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114517:114935 [5] NCCL INFO Channel 12/0 : 101[5] -> 69[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815967:2816399 [4] NCCL INFO Channel 11/0 : 92[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722047:2722468 [2] NCCL INFO Channel 10/0 : 122[2] -> 114[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153563:153986 [1] NCCL INFO Channel 01/0 : 105[1] -> 89[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114513:114933 [1] NCCL INFO Channel 11/0 : 73[1] -> 65[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114516:114934 [4] NCCL INFO Channel 05/0 : 76[4] -> 68[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815965:2816395 [2] NCCL INFO Channel 03/0 : 90[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114514:114932 [2] NCCL INFO Channel 04/0 : 66[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114519:114931 [7] NCCL INFO Channel 00/0 : 71[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557267:1557683 [1] NCCL INFO Channel 14/0 : 113[1] -> 97[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815968:2816394 [5] NCCL INFO Channel 12/0 : 61[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96567:97008 [4] NCCL INFO Channel 10/0 : 52[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815963:2816400 [0] NCCL INFO Channel 03/0 : 56[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96564:97009 [1] NCCL INFO Channel 12/0 : 49[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815969:2816398 [6] NCCL INFO Channel 05/0 : 62[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-52:2722047:2722468 [2] NCCL INFO Channel 12/0 : 122[2] -> 114[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153565:153985 [3] NCCL INFO Channel 03/0 : 107[3] -> 91[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815970:2816397 [7] NCCL INFO Channel 09/0 : 63[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153563:153986 [1] NCCL INFO Channel 03/0 : 105[1] -> 89[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114515:114936 [3] NCCL INFO Channel 06/0 : 67[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114517:114935 [5] NCCL INFO Channel 14/0 : 101[5] -> 69[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96567:97008 [4] NCCL INFO Channel 12/0 : 52[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114513:114933 [1] NCCL INFO Channel 13/0 : 73[1] -> 65[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114516:114934 [4] NCCL INFO Channel 07/0 : 76[4] -> 68[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96564:97009 [1] NCCL INFO Channel 14/0 : 49[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114514:114932 [2] NCCL INFO Channel 06/0 : 66[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235653:236067 [5] NCCL INFO Channel 00/0 : 117[5] -> 109[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153565:153985 [3] NCCL INFO Channel 05/0 : 107[3] -> 91[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114519:114931 [7] NCCL INFO Channel 02/0 : 71[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815967:2816399 [4] NCCL INFO Channel 13/0 : 92[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722047:2722468 [2] NCCL INFO Channel 14/0 : 122[2] -> 114[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114515:114936 [3] NCCL INFO Channel 08/0 : 67[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114517:114935 [5] NCCL INFO Channel 00/0 : 69[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815965:2816395 [2] NCCL INFO Channel 05/0 : 90[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815968:2816394 [5] NCCL INFO Channel 14/0 : 61[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114513:114933 [1] NCCL INFO Channel 15/0 : 73[1] -> 65[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114516:114934 [4] NCCL INFO Channel 09/0 : 76[4] -> 68[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114514:114932 [2] NCCL INFO Channel 08/0 : 66[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815963:2816400 [0] NCCL INFO Channel 05/0 : 56[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96567:97008 [4] NCCL INFO Channel 14/0 : 52[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235653:236067 [5] NCCL INFO Channel 01/0 : 117[5] -> 109[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815969:2816398 [6] NCCL INFO Channel 07/0 : 62[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96564:97009 [1] NCCL INFO Channel 00/0 : 33[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153563:153986 [1] NCCL INFO Channel 05/0 : 105[1] -> 89[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815970:2816397 [7] NCCL INFO Channel 11/0 : 63[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153565:153985 [3] NCCL INFO Channel 07/0 : 107[3] -> 91[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62886:63303 [5] NCCL INFO Channel 00/0 : 53[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815967:2816399 [4] NCCL INFO Channel 15/0 : 92[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815965:2816395 [2] NCCL INFO Channel 07/0 : 90[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153567:153979 [5] NCCL INFO Channel 00/0 : 93[5] -> 85[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114519:114931 [7] NCCL INFO Channel 04/0 : 71[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114517:114935 [5] NCCL INFO Channel 02/0 : 69[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235653:236067 [5] NCCL INFO Channel 02/0 : 117[5] -> 109[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815963:2816400 [0] NCCL INFO Channel 07/0 : 56[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96567:97008 [4] NCCL INFO Channel 00/0 : 36[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96564:97009 [1] NCCL INFO Channel 02/0 : 33[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114514:114932 [2] NCCL INFO Channel 10/0 : 66[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815969:2816398 [6] NCCL INFO Channel 09/0 : 62[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114515:114936 [3] NCCL INFO Channel 10/0 : 67[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114516:114934 [4] NCCL INFO Channel 11/0 : 76[4] -> 68[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62886:63303 [5] NCCL INFO Channel 01/0 : 53[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815970:2816397 [7] NCCL INFO Channel 13/0 : 63[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114517:114935 [5] NCCL INFO Channel 04/0 : 69[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815967:2816399 [4] NCCL INFO Channel 01/0 : 60[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153563:153986 [1] NCCL INFO Channel 07/0 : 105[1] -> 89[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114519:114931 [7] NCCL INFO Channel 06/0 : 71[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235653:236067 [5] NCCL INFO Channel 03/0 : 117[5] -> 109[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815966:2816396 [3] NCCL INFO Channel 00/0 : 59[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815963:2816400 [0] NCCL INFO Channel 09/0 : 56[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96567:97008 [4] NCCL INFO Channel 02/0 : 36[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96564:97009 [1] NCCL INFO Channel 04/0 : 33[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153567:153979 [5] NCCL INFO Channel 02/0 : 93[5] -> 85[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62886:63303 [5] NCCL INFO Channel 02/0 : 53[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815965:2816395 [2] NCCL INFO Channel 09/0 : 90[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364907:1365329 [5] NCCL INFO Channel 00/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815969:2816398 [6] NCCL INFO Channel 11/0 : 62[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815964:2816401 [1] NCCL INFO Channel 00/0 : 57[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114514:114932 [2] NCCL INFO Channel 12/0 : 66[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235653:236067 [5] NCCL INFO Channel 04/0 : 117[5] -> 109[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815967:2816399 [4] NCCL INFO Channel 03/0 : 60[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114515:114936 [3] NCCL INFO Channel 12/0 : 67[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96564:97009 [1] NCCL INFO Channel 06/0 : 33[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153565:153985 [3] NCCL INFO Channel 09/0 : 107[3] -> 91[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114516:114934 [4] NCCL INFO Channel 13/0 : 76[4] -> 68[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364907:1365329 [5] NCCL INFO Channel 01/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153563:153986 [1] NCCL INFO Channel 11/0 : 105[1] -> 89[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153569:153981 [7] NCCL INFO Channel 01/0 : 111[7] -> 95[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114517:114935 [5] NCCL INFO Channel 06/0 : 69[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62886:63303 [5] NCCL INFO Channel 03/0 : 53[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153567:153979 [5] NCCL INFO Channel 04/0 : 93[5] -> 85[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114519:114931 [7] NCCL INFO Channel 08/0 : 71[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235653:236067 [5] NCCL INFO Channel 06/0 : 117[5] -> 109[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815965:2816395 [2] NCCL INFO Channel 11/0 : 90[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815963:2816400 [0] NCCL INFO Channel 11/0 : 56[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815966:2816396 [3] NCCL INFO Channel 02/0 : 59[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96567:97008 [4] NCCL INFO Channel 06/0 : 36[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153565:153985 [3] NCCL INFO Channel 13/0 : 107[3] -> 91[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114514:114932 [2] NCCL INFO Channel 14/0 : 66[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815969:2816398 [6] NCCL INFO Channel 13/0 : 62[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96564:97009 [1] NCCL INFO Channel 08/0 : 33[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114515:114936 [3] NCCL INFO Channel 14/0 : 67[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815967:2816399 [4] NCCL INFO Channel 05/0 : 60[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364907:1365329 [5] NCCL INFO Channel 02/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114516:114934 [4] NCCL INFO Channel 15/0 : 76[4] -> 68[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114517:114935 [5] NCCL INFO Channel 08/0 : 69[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62886:63303 [5] NCCL INFO Channel 04/0 : 53[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815964:2816401 [1] NCCL INFO Channel 02/0 : 57[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114519:114931 [7] NCCL INFO Channel 10/0 : 71[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235653:236067 [5] NCCL INFO Channel 07/0 : 117[5] -> 109[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815963:2816400 [0] NCCL INFO Channel 13/0 : 56[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153569:153981 [7] NCCL INFO Channel 03/0 : 111[7] -> 95[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815965:2816395 [2] NCCL INFO Channel 13/0 : 90[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815966:2816396 [3] NCCL INFO Channel 04/0 : 59[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96567:97008 [4] NCCL INFO Channel 08/0 : 36[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153567:153979 [5] NCCL INFO Channel 06/0 : 93[5] -> 85[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815967:2816399 [4] NCCL INFO Channel 07/0 : 60[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815969:2816398 [6] NCCL INFO Channel 15/0 : 62[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96564:97009 [1] NCCL INFO Channel 10/0 : 33[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153563:153986 [1] NCCL INFO Channel 13/0 : 105[1] -> 89[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62886:63303 [5] NCCL INFO Channel 06/0 : 53[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815964:2816401 [1] NCCL INFO Channel 04/0 : 57[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364907:1365329 [5] NCCL INFO Channel 03/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96567:97008 [4] NCCL INFO Channel 10/0 : 36[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153565:153985 [3] NCCL INFO Channel 15/0 : 107[3] -> 91[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114517:114935 [5] NCCL INFO Channel 10/0 : 69[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235653:236067 [5] NCCL INFO Channel 08/0 : 117[5] -> 109[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815963:2816400 [0] NCCL INFO Channel 15/0 : 56[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:124163:124588 [3] NCCL INFO Channel 01/0 : 11[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153569:153981 [7] NCCL INFO Channel 05/0 : 111[7] -> 95[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114519:114931 [7] NCCL INFO Channel 12/0 : 71[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815966:2816396 [3] NCCL INFO Channel 06/0 : 59[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815965:2816395 [2] NCCL INFO Channel 15/0 : 90[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153567:153979 [5] NCCL INFO Channel 08/0 : 93[5] -> 85[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114517:114935 [5] NCCL INFO Channel 12/0 : 69[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62886:63303 [5] NCCL INFO Channel 07/0 : 53[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815970:2816397 [7] NCCL INFO Channel 00/0 : 63[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96564:97009 [1] NCCL INFO Channel 12/0 : 33[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114515:114936 [3] NCCL INFO Channel 00/0 : 99[3] -> 67[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114517:114935 [5] NCCL INFO Channel 14/0 : 69[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114519:114931 [7] NCCL INFO Channel 14/0 : 71[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235653:236067 [5] NCCL INFO Channel 09/0 : 117[5] -> 109[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815964:2816401 [1] NCCL INFO Channel 06/0 : 57[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815967:2816399 [4] NCCL INFO Channel 09/0 : 60[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364907:1365329 [5] NCCL INFO Channel 04/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124163:124588 [3] NCCL INFO Channel 03/0 : 11[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96567:97008 [4] NCCL INFO Channel 12/0 : 36[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153565:153985 [3] NCCL INFO Channel 01/0 : 91[3] -> 75[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153563:153986 [1] NCCL INFO Channel 15/0 : 105[1] -> 89[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62886:63303 [5] NCCL INFO Channel 08/0 : 53[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153569:153981 [7] NCCL INFO Channel 07/0 : 111[7] -> 95[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114515:114936 [3] NCCL INFO Channel 02/0 : 99[3] -> 67[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235653:236067 [5] NCCL INFO Channel 10/0 : 117[5] -> 109[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815966:2816396 [3] NCCL INFO Channel 08/0 : 59[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815965:2816395 [2] NCCL INFO Channel 01/0 : 58[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364907:1365329 [5] NCCL INFO Channel 06/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153568:153982 [6] NCCL INFO Channel 01/0 : 110[6] -> 94[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153567:153979 [5] NCCL INFO Channel 10/0 : 93[5] -> 85[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114514:114932 [2] NCCL INFO Channel 01/0 : 74[2] -> 66[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815970:2816397 [7] NCCL INFO Channel 02/0 : 63[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124163:124588 [3] NCCL INFO Channel 05/0 : 11[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96564:97009 [1] NCCL INFO Channel 14/0 : 33[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153565:153985 [3] NCCL INFO Channel 03/0 : 91[3] -> 75[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114518:114937 [6] NCCL INFO Channel 02/0 : 6[6] -> 70[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62886:63303 [5] NCCL INFO Channel 09/0 : 53[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815967:2816399 [4] NCCL INFO Channel 11/0 : 60[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815964:2816401 [1] NCCL INFO Channel 08/0 : 57[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96567:97008 [4] NCCL INFO Channel 14/0 : 36[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153563:153986 [1] NCCL INFO Channel 01/0 : 89[1] -> 73[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153562:153983 [0] NCCL INFO Channel 01/0 : 104[0] -> 88[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114515:114936 [3] NCCL INFO Channel 04/0 : 99[3] -> 67[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235653:236067 [5] NCCL INFO Channel 11/0 : 117[5] -> 109[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815966:2816396 [3] NCCL INFO Channel 10/0 : 59[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364907:1365329 [5] NCCL INFO Channel 07/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124163:124588 [3] NCCL INFO Channel 07/0 : 11[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96565:97004 [2] NCCL INFO Channel 00/0 : 50[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153569:153981 [7] NCCL INFO Channel 09/0 : 111[7] -> 95[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62886:63303 [5] NCCL INFO Channel 10/0 : 53[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815965:2816395 [2] NCCL INFO Channel 03/0 : 58[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153567:153979 [5] NCCL INFO Channel 12/0 : 93[5] -> 85[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153568:153982 [6] NCCL INFO Channel 03/0 : 110[6] -> 94[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815970:2816397 [7] NCCL INFO Channel 04/0 : 63[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114514:114932 [2] NCCL INFO Channel 03/0 : 74[2] -> 66[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235653:236067 [5] NCCL INFO Channel 12/0 : 117[5] -> 109[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815967:2816399 [4] NCCL INFO Channel 13/0 : 60[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815964:2816401 [1] NCCL INFO Channel 10/0 : 57[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32705:33128 [1] NCCL INFO Channel 15/0 : 57[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32710:33125 [6] NCCL INFO Channel 07/0 : 62[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114518:114937 [6] NCCL INFO Channel 04/0 : 6[6] -> 70[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114512:114930 [0] NCCL INFO Channel 02/0 : 96[0] -> 64[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815966:2816396 [3] NCCL INFO Channel 12/0 : 59[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364907:1365329 [5] NCCL INFO Channel 08/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32707:33124 [3] NCCL INFO Channel 15/0 : 59[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32704:33123 [0] NCCL INFO Channel 03/0 : 56[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32711:33127 [7] NCCL INFO Channel 09/0 : 63[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124163:124588 [3] NCCL INFO Channel 09/0 : 11[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96565:97004 [2] NCCL INFO Channel 04/0 : 50[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114515:114936 [3] NCCL INFO Channel 06/0 : 99[3] -> 67[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62886:63303 [5] NCCL INFO Channel 11/0 : 53[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815965:2816395 [2] NCCL INFO Channel 05/0 : 58[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32710:33125 [6] NCCL INFO Channel 09/0 : 62[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96565:97004 [2] NCCL INFO Channel 06/0 : 50[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153565:153985 [3] NCCL INFO Channel 05/0 : 91[3] -> 75[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114514:114932 [2] NCCL INFO Channel 05/0 : 74[2] -> 66[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235653:236067 [5] NCCL INFO Channel 14/0 : 117[5] -> 109[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815969:2816398 [6] NCCL INFO Channel 00/0 : 62[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32708:33126 [4] NCCL INFO Channel 01/0 : 60[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636085:2636509 [1] NCCL INFO Channel 00/0 : 57[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153563:153986 [1] NCCL INFO Channel 03/0 : 89[1] -> 73[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114512:114930 [0] NCCL INFO Channel 04/0 : 96[0] -> 64[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114518:114937 [6] NCCL INFO Channel 08/0 : 6[6] -> 70[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364907:1365329 [5] NCCL INFO Channel 09/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32711:33127 [7] NCCL INFO Channel 11/0 : 63[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124163:124588 [3] NCCL INFO Channel 13/0 : 11[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153562:153983 [0] NCCL INFO Channel 03/0 : 104[0] -> 88[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62886:63303 [5] NCCL INFO Channel 12/0 : 53[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32704:33123 [0] NCCL INFO Channel 05/0 : 56[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32710:33125 [6] NCCL INFO Channel 11/0 : 62[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96568:97007 [5] NCCL INFO Channel 00/0 : 53[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96565:97004 [2] NCCL INFO Channel 08/0 : 50[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153567:153979 [5] NCCL INFO Channel 14/0 : 93[5] -> 85[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153569:153981 [7] NCCL INFO Channel 11/0 : 111[7] -> 95[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235653:236067 [5] NCCL INFO Channel 15/0 : 117[5] -> 109[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815963:2816400 [0] NCCL INFO Channel 02/0 : 56[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32708:33126 [4] NCCL INFO Channel 03/0 : 60[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636088:2636510 [4] NCCL INFO Channel 00/0 : 60[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96564:97009 [1] NCCL INFO Channel 01/0 : 41[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153565:153985 [3] NCCL INFO Channel 07/0 : 91[3] -> 75[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153568:153982 [6] NCCL INFO Channel 05/0 : 110[6] -> 94[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114515:114936 [3] NCCL INFO Channel 08/0 : 99[3] -> 67[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815970:2816397 [7] NCCL INFO Channel 06/0 : 63[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364907:1365329 [5] NCCL INFO Channel 10/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32711:33127 [7] NCCL INFO Channel 13/0 : 63[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636085:2636509 [1] NCCL INFO Channel 02/0 : 57[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124163:124588 [3] NCCL INFO Channel 15/0 : 11[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153563:153986 [1] NCCL INFO Channel 05/0 : 89[1] -> 73[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114514:114932 [2] NCCL INFO Channel 07/0 : 74[2] -> 66[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62886:63303 [5] NCCL INFO Channel 14/0 : 53[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235653:236067 [5] NCCL INFO Channel 01/0 : 109[5] -> 101[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96889:97309 [1] NCCL INFO Channel 00/0 : 25[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815967:2816399 [4] NCCL INFO Channel 15/0 : 60[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32710:33125 [6] NCCL INFO Channel 13/0 : 62[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32704:33123 [0] NCCL INFO Channel 07/0 : 56[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153562:153983 [0] NCCL INFO Channel 05/0 : 104[0] -> 88[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114519:114931 [7] NCCL INFO Channel 00/0 : 103[7] -> 71[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815966:2816396 [3] NCCL INFO Channel 14/0 : 59[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815964:2816401 [1] NCCL INFO Channel 12/0 : 57[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364907:1365329 [5] NCCL INFO Channel 11/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32708:33126 [4] NCCL INFO Channel 05/0 : 60[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96568:97007 [5] NCCL INFO Channel 02/0 : 53[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153569:153981 [7] NCCL INFO Channel 13/0 : 111[7] -> 95[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114517:114935 [5] NCCL INFO Channel 01/0 : 77[5] -> 69[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815965:2816395 [2] NCCL INFO Channel 07/0 : 58[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32710:33125 [6] NCCL INFO Channel 15/0 : 62[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636088:2636510 [4] NCCL INFO Channel 02/0 : 60[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96565:97004 [2] NCCL INFO Channel 10/0 : 50[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96567:97008 [4] NCCL INFO Channel 01/0 : 44[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153565:153985 [3] NCCL INFO Channel 09/0 : 91[3] -> 75[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153563:153986 [1] NCCL INFO Channel 07/0 : 89[1] -> 73[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153568:153982 [6] NCCL INFO Channel 07/0 : 110[6] -> 94[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114518:114937 [6] NCCL INFO Channel 10/0 : 6[6] -> 70[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62886:63303 [5] NCCL INFO Channel 15/0 : 53[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235653:236067 [5] NCCL INFO Channel 03/0 : 109[5] -> 101[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96892:97308 [4] NCCL INFO Channel 00/0 : 28[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815969:2816398 [6] NCCL INFO Channel 02/0 : 62[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32704:33123 [0] NCCL INFO Channel 09/0 : 56[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32708:33126 [4] NCCL INFO Channel 07/0 : 60[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636085:2636509 [1] NCCL INFO Channel 04/0 : 57[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96564:97009 [1] NCCL INFO Channel 03/0 : 41[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153562:153983 [0] NCCL INFO Channel 07/0 : 104[0] -> 88[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114515:114936 [3] NCCL INFO Channel 10/0 : 99[3] -> 67[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114512:114930 [0] NCCL INFO Channel 06/0 : 96[0] -> 64[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114514:114932 [2] NCCL INFO Channel 09/0 : 74[2] -> 66[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272154:272569 [4] NCCL INFO Channel 00/0 : 124[4] -> 116[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96889:97309 [1] NCCL INFO Channel 02/0 : 25[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815963:2816400 [0] NCCL INFO Channel 04/0 : 56[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364907:1365329 [5] NCCL INFO Channel 12/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32704:33123 [0] NCCL INFO Channel 11/0 : 56[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32708:33126 [4] NCCL INFO Channel 09/0 : 60[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636088:2636510 [4] NCCL INFO Channel 06/0 : 60[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722049:2722467 [4] NCCL INFO Channel 02/0 : 124[4] -> 116[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114517:114935 [5] NCCL INFO Channel 03/0 : 77[5] -> 69[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114519:114931 [7] NCCL INFO Channel 02/0 : 103[7] -> 71[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62886:63303 [5] NCCL INFO Channel 01/0 : 45[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235653:236067 [5] NCCL INFO Channel 05/0 : 109[5] -> 101[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815970:2816397 [7] NCCL INFO Channel 08/0 : 63[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32704:33123 [0] NCCL INFO Channel 13/0 : 56[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96568:97007 [5] NCCL INFO Channel 04/0 : 53[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114518:114937 [6] NCCL INFO Channel 12/0 : 6[6] -> 70[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272154:272569 [4] NCCL INFO Channel 02/0 : 124[4] -> 116[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815969:2816398 [6] NCCL INFO Channel 04/0 : 62[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32708:33126 [4] NCCL INFO Channel 11/0 : 60[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96567:97008 [4] NCCL INFO Channel 03/0 : 44[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96565:97004 [2] NCCL INFO Channel 12/0 : 50[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153569:153981 [7] NCCL INFO Channel 01/0 : 95[7] -> 79[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96892:97308 [4] NCCL INFO Channel 02/0 : 28[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815963:2816400 [0] NCCL INFO Channel 06/0 : 56[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364907:1365329 [5] NCCL INFO Channel 14/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32704:33123 [0] NCCL INFO Channel 15/0 : 56[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636085:2636509 [1] NCCL INFO Channel 06/0 : 57[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-52:2722049:2722467 [4] NCCL INFO Channel 06/0 : 124[4] -> 116[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96564:97009 [1] NCCL INFO Channel 05/0 : 41[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153563:153986 [1] NCCL INFO Channel 11/0 : 89[1] -> 73[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62886:63303 [5] NCCL INFO Channel 03/0 : 45[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235653:236067 [5] NCCL INFO Channel 07/0 : 109[5] -> 101[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96889:97309 [1] NCCL INFO Channel 04/0 : 25[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815964:2816401 [1] NCCL INFO Channel 14/0 : 57[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32708:33126 [4] NCCL INFO Channel 13/0 : 60[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227467:1227886 [4] NCCL INFO Channel 00/0 : 92[4] -> 84[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636088:2636510 [4] NCCL INFO Channel 08/0 : 60[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96568:97007 [5] NCCL INFO Channel 06/0 : 53[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153565:153985 [3] NCCL INFO Channel 13/0 : 91[3] -> 75[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153562:153983 [0] NCCL INFO Channel 09/0 : 104[0] -> 88[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114512:114930 [0] NCCL INFO Channel 08/0 : 96[0] -> 64[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114515:114936 [3] NCCL INFO Channel 12/0 : 99[3] -> 67[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114514:114932 [2] NCCL INFO Channel 11/0 : 74[2] -> 66[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235653:236067 [5] NCCL INFO Channel 09/0 : 109[5] -> 101[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272154:272569 [4] NCCL INFO Channel 06/0 : 124[4] -> 116[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96892:97308 [4] NCCL INFO Channel 06/0 : 28[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815970:2816397 [7] NCCL INFO Channel 10/0 : 63[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364907:1365329 [5] NCCL INFO Channel 15/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32708:33126 [4] NCCL INFO Channel 15/0 : 60[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636085:2636509 [1] NCCL INFO Channel 08/0 : 57[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-52:2722049:2722467 [4] NCCL INFO Channel 08/0 : 124[4] -> 116[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153568:153982 [6] NCCL INFO Channel 09/0 : 110[6] -> 94[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114517:114935 [5] NCCL INFO Channel 05/0 : 77[5] -> 69[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62886:63303 [5] NCCL INFO Channel 05/0 : 45[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815969:2816398 [6] NCCL INFO Channel 08/0 : 62[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32706:33122 [2] NCCL INFO Channel 01/0 : 58[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227467:1227886 [4] NCCL INFO Channel 02/0 : 92[4] -> 84[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636088:2636510 [4] NCCL INFO Channel 10/0 : 60[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153566:153980 [4] NCCL INFO Channel 01/0 : 108[4] -> 92[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153569:153981 [7] NCCL INFO Channel 03/0 : 95[7] -> 79[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153563:153986 [1] NCCL INFO Channel 13/0 : 89[1] -> 73[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114518:114937 [6] NCCL INFO Channel 14/0 : 6[6] -> 70[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62886:63303 [5] NCCL INFO Channel 07/0 : 45[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235653:236067 [5] NCCL INFO Channel 11/0 : 109[5] -> 101[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272154:272569 [4] NCCL INFO Channel 08/0 : 124[4] -> 116[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96889:97309 [1] NCCL INFO Channel 06/0 : 25[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815963:2816400 [0] NCCL INFO Channel 08/0 : 56[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32706:33122 [2] NCCL INFO Channel 03/0 : 58[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96565:97004 [2] NCCL INFO Channel 14/0 : 50[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153565:153985 [3] NCCL INFO Channel 15/0 : 91[3] -> 75[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114519:114931 [7] NCCL INFO Channel 04/0 : 103[7] -> 71[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235651:236068 [3] NCCL INFO Channel 00/0 : 115[3] -> 107[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96892:97308 [4] NCCL INFO Channel 08/0 : 28[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815970:2816397 [7] NCCL INFO Channel 12/0 : 63[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364907:1365329 [5] NCCL INFO Channel 01/0 : 13[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32706:33122 [2] NCCL INFO Channel 05/0 : 58[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-52:2722049:2722467 [4] NCCL INFO Channel 10/0 : 124[4] -> 116[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96567:97008 [4] NCCL INFO Channel 05/0 : 44[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153562:153983 [0] NCCL INFO Channel 11/0 : 104[0] -> 88[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114515:114936 [3] NCCL INFO Channel 14/0 : 99[3] -> 67[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114512:114930 [0] NCCL INFO Channel 10/0 : 96[0] -> 64[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114514:114932 [2] NCCL INFO Channel 13/0 : 74[2] -> 66[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62886:63303 [5] NCCL INFO Channel 09/0 : 45[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96889:97309 [1] NCCL INFO Channel 08/0 : 25[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815969:2816398 [6] NCCL INFO Channel 10/0 : 62[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32706:33122 [2] NCCL INFO Channel 07/0 : 58[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227467:1227886 [4] NCCL INFO Channel 06/0 : 92[4] -> 84[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636085:2636509 [1] NCCL INFO Channel 10/0 : 57[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96564:97009 [1] NCCL INFO Channel 07/0 : 41[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153568:153982 [6] NCCL INFO Channel 11/0 : 110[6] -> 94[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114519:114931 [7] NCCL INFO Channel 06/0 : 103[7] -> 71[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114515:114936 [3] NCCL INFO Channel 00/0 : 67[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114512:114930 [0] NCCL INFO Channel 12/0 : 96[0] -> 64[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235649:236071 [1] NCCL INFO Channel 00/0 : 113[1] -> 105[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96892:97308 [4] NCCL INFO Channel 10/0 : 28[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32706:33122 [2] NCCL INFO Channel 09/0 : 58[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636088:2636510 [4] NCCL INFO Channel 12/0 : 60[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96568:97007 [5] NCCL INFO Channel 08/0 : 53[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153563:153986 [1] NCCL INFO Channel 15/0 : 89[1] -> 73[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153569:153981 [7] NCCL INFO Channel 05/0 : 95[7] -> 79[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153566:153980 [4] NCCL INFO Channel 03/0 : 108[4] -> 92[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114515:114936 [3] NCCL INFO Channel 02/0 : 67[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114512:114930 [0] NCCL INFO Channel 14/0 : 96[0] -> 64[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62886:63303 [5] NCCL INFO Channel 11/0 : 45[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235653:236067 [5] NCCL INFO Channel 15/0 : 109[5] -> 101[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272154:272569 [4] NCCL INFO Channel 10/0 : 124[4] -> 116[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815963:2816400 [0] NCCL INFO Channel 10/0 : 56[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364907:1365329 [5] NCCL INFO Channel 03/0 : 13[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32706:33122 [2] NCCL INFO Channel 11/0 : 58[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227467:1227886 [4] NCCL INFO Channel 08/0 : 92[4] -> 84[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636085:2636509 [1] NCCL INFO Channel 12/0 : 57[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-52:2722049:2722467 [4] NCCL INFO Channel 12/0 : 124[4] -> 116[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96565:97004 [2] NCCL INFO Channel 00/0 : 34[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235651:236068 [3] NCCL INFO Channel 01/0 : 115[3] -> 107[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96889:97309 [1] NCCL INFO Channel 10/0 : 25[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815970:2816397 [7] NCCL INFO Channel 14/0 : 63[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32706:33122 [2] NCCL INFO Channel 13/0 : 58[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636088:2636510 [4] NCCL INFO Channel 14/0 : 60[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722049:2722467 [4] NCCL INFO Channel 14/0 : 124[4] -> 116[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96567:97008 [4] NCCL INFO Channel 07/0 : 44[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153568:153982 [6] NCCL INFO Channel 13/0 : 110[6] -> 94[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153562:153983 [0] NCCL INFO Channel 13/0 : 104[0] -> 88[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114519:114931 [7] NCCL INFO Channel 08/0 : 103[7] -> 71[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62882:63300 [1] NCCL INFO Channel 00/0 : 49[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62886:63303 [5] NCCL INFO Channel 15/0 : 45[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62884:63299 [3] NCCL INFO Channel 00/0 : 51[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272154:272569 [4] NCCL INFO Channel 12/0 : 124[4] -> 116[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96892:97308 [4] NCCL INFO Channel 12/0 : 28[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815969:2816398 [6] NCCL INFO Channel 12/0 : 62[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32706:33122 [2] NCCL INFO Channel 15/0 : 58[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636085:2636509 [1] NCCL INFO Channel 14/0 : 57[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96568:97007 [5] NCCL INFO Channel 10/0 : 53[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96564:97009 [1] NCCL INFO Channel 11/0 : 41[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153569:153981 [7] NCCL INFO Channel 07/0 : 95[7] -> 79[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153566:153980 [4] NCCL INFO Channel 05/0 : 108[4] -> 92[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114512:114930 [0] NCCL INFO Channel 02/0 : 64[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114515:114936 [3] NCCL INFO Channel 04/0 : 67[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235649:236071 [1] NCCL INFO Channel 02/0 : 113[1] -> 105[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272154:272569 [4] NCCL INFO Channel 14/0 : 124[4] -> 116[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96889:97309 [1] NCCL INFO Channel 12/0 : 25[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815963:2816400 [0] NCCL INFO Channel 12/0 : 56[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364907:1365329 [5] NCCL INFO Channel 05/0 : 13[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32709:33129 [5] NCCL INFO Channel 01/0 : 45[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227467:1227886 [4] NCCL INFO Channel 10/0 : 92[4] -> 84[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153568:153982 [6] NCCL INFO Channel 15/0 : 110[6] -> 94[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153562:153983 [0] NCCL INFO Channel 15/0 : 104[0] -> 88[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114519:114931 [7] NCCL INFO Channel 10/0 : 103[7] -> 71[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235651:236068 [3] NCCL INFO Channel 02/0 : 115[3] -> 107[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96892:97308 [4] NCCL INFO Channel 14/0 : 28[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815967:2816399 [4] NCCL INFO Channel 00/0 : 60[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32709:33129 [5] NCCL INFO Channel 03/0 : 45[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636088:2636510 [4] NCCL INFO Channel 00/0 : 52[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153569:153981 [7] NCCL INFO Channel 09/0 : 95[7] -> 79[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153566:153980 [4] NCCL INFO Channel 07/0 : 108[4] -> 92[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114512:114930 [0] NCCL INFO Channel 04/0 : 64[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114515:114936 [3] NCCL INFO Channel 06/0 : 67[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235649:236071 [1] NCCL INFO Channel 03/0 : 113[1] -> 105[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815969:2816398 [6] NCCL INFO Channel 14/0 : 62[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32709:33129 [5] NCCL INFO Channel 05/0 : 45[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636085:2636509 [1] NCCL INFO Channel 00/0 : 49[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96565:97004 [2] NCCL INFO Channel 04/0 : 34[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153568:153982 [6] NCCL INFO Channel 01/0 : 94[6] -> 78[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153562:153983 [0] NCCL INFO Channel 01/0 : 88[0] -> 72[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114519:114931 [7] NCCL INFO Channel 12/0 : 103[7] -> 71[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235651:236068 [3] NCCL INFO Channel 04/0 : 115[3] -> 107[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272154:272569 [4] NCCL INFO Channel 00/0 : 116[4] -> 108[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815963:2816400 [0] NCCL INFO Channel 14/0 : 56[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364907:1365329 [5] NCCL INFO Channel 07/0 : 13[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32709:33129 [5] NCCL INFO Channel 07/0 : 45[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227467:1227886 [4] NCCL INFO Channel 12/0 : 92[4] -> 84[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636088:2636510 [4] NCCL INFO Channel 01/0 : 52[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96567:97008 [4] NCCL INFO Channel 09/0 : 44[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153569:153981 [7] NCCL INFO Channel 11/0 : 95[7] -> 79[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114515:114936 [3] NCCL INFO Channel 08/0 : 67[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114512:114930 [0] NCCL INFO Channel 06/0 : 64[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62882:63300 [1] NCCL INFO Channel 02/0 : 49[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62884:63299 [3] NCCL INFO Channel 01/0 : 51[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96889:97309 [1] NCCL INFO Channel 14/0 : 25[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96892:97308 [4] NCCL INFO Channel 00/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815967:2816399 [4] NCCL INFO Channel 02/0 : 60[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364907:1365329 [5] NCCL INFO Channel 09/0 : 13[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32709:33129 [5] NCCL INFO Channel 09/0 : 45[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636085:2636509 [1] NCCL INFO Channel 02/0 : 49[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96568:97007 [5] NCCL INFO Channel 12/0 : 53[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153566:153980 [4] NCCL INFO Channel 09/0 : 108[4] -> 92[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114519:114931 [7] NCCL INFO Channel 14/0 : 103[7] -> 71[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114515:114936 [3] NCCL INFO Channel 10/0 : 67[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235649:236071 [1] NCCL INFO Channel 04/0 : 113[1] -> 105[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96892:97308 [4] NCCL INFO Channel 01/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815967:2816399 [4] NCCL INFO Channel 06/0 : 60[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364903:1365334 [1] NCCL INFO Channel 00/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32709:33129 [5] NCCL INFO Channel 11/0 : 45[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636088:2636510 [4] NCCL INFO Channel 02/0 : 52[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96564:97009 [1] NCCL INFO Channel 13/0 : 41[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153565:153985 [3] NCCL INFO Channel 00/0 : 91[3] -> 83[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114512:114930 [0] NCCL INFO Channel 08/0 : 64[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235651:236068 [3] NCCL INFO Channel 05/0 : 115[3] -> 107[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272154:272569 [4] NCCL INFO Channel 01/0 : 116[4] -> 108[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96889:97309 [1] NCCL INFO Channel 00/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815967:2816399 [4] NCCL INFO Channel 08/0 : 60[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32709:33129 [5] NCCL INFO Channel 15/0 : 45[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227467:1227886 [4] NCCL INFO Channel 14/0 : 92[4] -> 84[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636085:2636509 [1] NCCL INFO Channel 03/0 : 49[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96565:97004 [2] NCCL INFO Channel 06/0 : 34[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153568:153982 [6] NCCL INFO Channel 03/0 : 94[6] -> 78[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62884:63299 [3] NCCL INFO Channel 02/0 : 51[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62882:63300 [1] NCCL INFO Channel 03/0 : 49[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235649:236071 [1] NCCL INFO Channel 05/0 : 113[1] -> 105[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815967:2816399 [4] NCCL INFO Channel 10/0 : 60[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364905:1365328 [3] NCCL INFO Channel 00/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32709:33129 [5] NCCL INFO Channel 01/0 : 29[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96567:97008 [4] NCCL INFO Channel 11/0 : 44[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153562:153983 [0] NCCL INFO Channel 03/0 : 88[0] -> 72[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114519:114931 [7] NCCL INFO Channel 00/0 : 71[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272154:272569 [4] NCCL INFO Channel 02/0 : 116[4] -> 108[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96889:97309 [1] NCCL INFO Channel 02/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96892:97308 [4] NCCL INFO Channel 02/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815967:2816399 [4] NCCL INFO Channel 12/0 : 60[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364907:1365329 [5] NCCL INFO Channel 11/0 : 13[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32709:33129 [5] NCCL INFO Channel 03/0 : 29[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227467:1227886 [4] NCCL INFO Channel 00/0 : 84[4] -> 76[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96568:97007 [5] NCCL INFO Channel 14/0 : 53[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153569:153981 [7] NCCL INFO Channel 13/0 : 95[7] -> 79[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153563:153986 [1] NCCL INFO Channel 00/0 : 89[1] -> 81[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114512:114930 [0] NCCL INFO Channel 10/0 : 64[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114515:114936 [3] NCCL INFO Channel 12/0 : 67[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62882:63300 [1] NCCL INFO Channel 04/0 : 49[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62884:63299 [3] NCCL INFO Channel 04/0 : 51[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815967:2816399 [4] NCCL INFO Channel 14/0 : 60[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364903:1365334 [1] NCCL INFO Channel 02/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364905:1365328 [3] NCCL INFO Channel 01/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32709:33129 [5] NCCL INFO Channel 05/0 : 29[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636088:2636510 [4] NCCL INFO Channel 03/0 : 52[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96564:97009 [1] NCCL INFO Channel 15/0 : 41[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96565:97004 [2] NCCL INFO Channel 08/0 : 34[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153566:153980 [4] NCCL INFO Channel 11/0 : 108[4] -> 92[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153565:153985 [3] NCCL INFO Channel 02/0 : 91[3] -> 83[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114518:114937 [6] NCCL INFO Channel 00/0 : 70[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557270:1557688 [4] NCCL INFO Channel 12/0 : 116[4] -> 100[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235651:236068 [3] NCCL INFO Channel 06/0 : 115[3] -> 107[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272154:272569 [4] NCCL INFO Channel 03/0 : 116[4] -> 108[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364907:1365329 [5] NCCL INFO Channel 15/0 : 13[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32709:33129 [5] NCCL INFO Channel 07/0 : 29[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636085:2636509 [1] NCCL INFO Channel 04/0 : 49[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96567:97008 [4] NCCL INFO Channel 13/0 : 44[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114512:114930 [0] NCCL INFO Channel 12/0 : 64[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114515:114936 [3] NCCL INFO Channel 14/0 : 67[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114519:114931 [7] NCCL INFO Channel 02/0 : 71[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62882:63300 [1] NCCL INFO Channel 05/0 : 49[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557270:1557688 [4] NCCL INFO Channel 14/0 : 116[4] -> 100[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235649:236071 [1] NCCL INFO Channel 06/0 : 113[1] -> 105[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96889:97309 [1] NCCL INFO Channel 03/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96892:97308 [4] NCCL INFO Channel 03/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815965:2816395 [2] NCCL INFO Channel 09/0 : 58[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32709:33129 [5] NCCL INFO Channel 09/0 : 29[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227467:1227886 [4] NCCL INFO Channel 01/0 : 84[4] -> 76[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636088:2636510 [4] NCCL INFO Channel 05/0 : 52[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96568:97007 [5] NCCL INFO Channel 00/0 : 37[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114518:114937 [6] NCCL INFO Channel 02/0 : 70[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62884:63299 [3] NCCL INFO Channel 05/0 : 51[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557270:1557688 [4] NCCL INFO Channel 00/0 : 100[4] -> 84[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235655:236069 [7] NCCL INFO Channel 00/0 : 119[7] -> 111[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272151:272571 [1] NCCL INFO Channel 00/0 : 121[1] -> 113[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815965:2816395 [2] NCCL INFO Channel 11/0 : 58[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32709:33129 [5] NCCL INFO Channel 11/0 : 29[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636085:2636509 [1] NCCL INFO Channel 05/0 : 49[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96565:97004 [2] NCCL INFO Channel 10/0 : 34[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96567:97008 [4] NCCL INFO Channel 15/0 : 44[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153568:153982 [6] NCCL INFO Channel 05/0 : 94[6] -> 78[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557270:1557688 [4] NCCL INFO Channel 02/0 : 100[4] -> 84[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235651:236068 [3] NCCL INFO Channel 07/0 : 115[3] -> 107[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272154:272569 [4] NCCL INFO Channel 05/0 : 116[4] -> 108[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815965:2816395 [2] NCCL INFO Channel 13/0 : 58[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364903:1365334 [1] NCCL INFO Channel 03/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364905:1365328 [3] NCCL INFO Channel 02/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32709:33129 [5] NCCL INFO Channel 15/0 : 29[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636088:2636510 [4] NCCL INFO Channel 06/0 : 52[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96568:97007 [5] NCCL INFO Channel 02/0 : 37[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153562:153983 [0] NCCL INFO Channel 05/0 : 88[0] -> 72[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557270:1557688 [4] NCCL INFO Channel 06/0 : 100[4] -> 84[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235649:236071 [1] NCCL INFO Channel 07/0 : 113[1] -> 105[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96889:97309 [1] NCCL INFO Channel 04/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364903:1365334 [1] NCCL INFO Channel 04/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32707:33124 [3] NCCL INFO Channel 01/0 : 43[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227467:1227886 [4] NCCL INFO Channel 02/0 : 84[4] -> 76[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636085:2636509 [1] NCCL INFO Channel 06/0 : 49[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153563:153986 [1] NCCL INFO Channel 02/0 : 89[1] -> 81[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114512:114930 [0] NCCL INFO Channel 14/0 : 64[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114519:114931 [7] NCCL INFO Channel 04/0 : 71[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114518:114937 [6] NCCL INFO Channel 04/0 : 70[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62888:63302 [7] NCCL INFO Channel 00/0 : 55[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557270:1557688 [4] NCCL INFO Channel 08/0 : 100[4] -> 84[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96892:97308 [4] NCCL INFO Channel 05/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364905:1365328 [3] NCCL INFO Channel 04/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32705:33128 [1] NCCL INFO Channel 01/0 : 41[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227467:1227886 [4] NCCL INFO Channel 03/0 : 84[4] -> 76[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153565:153985 [3] NCCL INFO Channel 04/0 : 91[3] -> 83[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153566:153980 [4] NCCL INFO Channel 13/0 : 108[4] -> 92[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114519:114931 [7] NCCL INFO Channel 06/0 : 71[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114518:114937 [6] NCCL INFO Channel 08/0 : 70[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62884:63299 [3] NCCL INFO Channel 06/0 : 51[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62882:63300 [1] NCCL INFO Channel 06/0 : 49[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557270:1557688 [4] NCCL INFO Channel 10/0 : 100[4] -> 84[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272151:272571 [1] NCCL INFO Channel 02/0 : 121[1] -> 113[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815965:2816395 [2] NCCL INFO Channel 15/0 : 58[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32707:33124 [3] NCCL INFO Channel 03/0 : 43[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96565:97004 [2] NCCL INFO Channel 12/0 : 34[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153568:153982 [6] NCCL INFO Channel 07/0 : 94[6] -> 78[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153562:153983 [0] NCCL INFO Channel 07/0 : 88[0] -> 72[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114514:114932 [2] NCCL INFO Channel 15/0 : 74[2] -> 66[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557270:1557688 [4] NCCL INFO Channel 12/0 : 100[4] -> 84[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235655:236069 [7] NCCL INFO Channel 01/0 : 119[7] -> 111[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272154:272569 [4] NCCL INFO Channel 06/0 : 116[4] -> 108[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32705:33128 [1] NCCL INFO Channel 03/0 : 41[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96568:97007 [5] NCCL INFO Channel 04/0 : 37[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153563:153986 [1] NCCL INFO Channel 04/0 : 89[1] -> 81[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114519:114931 [7] NCCL INFO Channel 08/0 : 71[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114518:114937 [6] NCCL INFO Channel 10/0 : 70[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557270:1557688 [4] NCCL INFO Channel 14/0 : 100[4] -> 84[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235651:236068 [3] NCCL INFO Channel 08/0 : 115[3] -> 107[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235649:236071 [1] NCCL INFO Channel 08/0 : 113[1] -> 105[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96889:97309 [1] NCCL INFO Channel 05/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364903:1365334 [1] NCCL INFO Channel 05/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32707:33124 [3] NCCL INFO Channel 05/0 : 43[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227464:1227884 [1] NCCL INFO Channel 00/0 : 89[1] -> 81[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96565:97004 [2] NCCL INFO Channel 14/0 : 34[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153565:153985 [3] NCCL INFO Channel 06/0 : 91[3] -> 83[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62888:63302 [7] NCCL INFO Channel 01/0 : 55[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557267:1557683 [1] NCCL INFO Channel 00/0 : 97[1] -> 81[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557271:1557690 [5] NCCL INFO Channel 00/0 : 117[5] -> 101[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235655:236069 [7] NCCL INFO Channel 02/0 : 119[7] -> 111[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96892:97308 [4] NCCL INFO Channel 06/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364905:1365328 [3] NCCL INFO Channel 05/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32705:33128 [1] NCCL INFO Channel 05/0 : 41[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227467:1227886 [4] NCCL INFO Channel 05/0 : 84[4] -> 76[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96568:97007 [5] NCCL INFO Channel 06/0 : 37[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153566:153980 [4] NCCL INFO Channel 15/0 : 108[4] -> 92[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62882:63300 [1] NCCL INFO Channel 07/0 : 49[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62884:63299 [3] NCCL INFO Channel 07/0 : 51[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557271:1557690 [5] NCCL INFO Channel 02/0 : 117[5] -> 101[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235649:236071 [1] NCCL INFO Channel 10/0 : 113[1] -> 105[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235651:236068 [3] NCCL INFO Channel 09/0 : 115[3] -> 107[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272151:272571 [1] NCCL INFO Channel 04/0 : 121[1] -> 113[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96889:97309 [1] NCCL INFO Channel 06/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364903:1365334 [1] NCCL INFO Channel 06/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32707:33124 [3] NCCL INFO Channel 07/0 : 43[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32709:33129 [5] NCCL INFO Channel 00/0 : 29[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96568:97007 [5] NCCL INFO Channel 08/0 : 37[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153562:153983 [0] NCCL INFO Channel 09/0 : 88[0] -> 72[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153568:153982 [6] NCCL INFO Channel 09/0 : 94[6] -> 78[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153563:153986 [1] NCCL INFO Channel 06/0 : 89[1] -> 81[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557267:1557683 [1] NCCL INFO Channel 02/0 : 97[1] -> 81[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235655:236069 [7] NCCL INFO Channel 03/0 : 119[7] -> 111[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272154:272569 [4] NCCL INFO Channel 07/0 : 116[4] -> 108[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364909:1365327 [7] NCCL INFO Channel 00/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32705:33128 [1] NCCL INFO Channel 07/0 : 41[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96568:97007 [5] NCCL INFO Channel 10/0 : 37[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153565:153985 [3] NCCL INFO Channel 08/0 : 91[3] -> 83[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114518:114937 [6] NCCL INFO Channel 12/0 : 70[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114519:114931 [7] NCCL INFO Channel 10/0 : 71[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557271:1557690 [5] NCCL INFO Channel 04/0 : 117[5] -> 101[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557270:1557688 [4] NCCL INFO Channel 01/0 : 108[4] -> 100[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272151:272571 [1] NCCL INFO Channel 06/0 : 121[1] -> 113[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364905:1365328 [3] NCCL INFO Channel 06/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32709:33129 [5] NCCL INFO Channel 02/0 : 29[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227464:1227884 [1] NCCL INFO Channel 02/0 : 89[1] -> 81[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96568:97007 [5] NCCL INFO Channel 12/0 : 37[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153566:153980 [4] NCCL INFO Channel 01/0 : 92[4] -> 76[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114515:114936 [3] NCCL INFO Channel 01/0 : 75[3] -> 67[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62888:63302 [7] NCCL INFO Channel 02/0 : 55[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557267:1557683 [1] NCCL INFO Channel 04/0 : 97[1] -> 81[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272154:272569 [4] NCCL INFO Channel 08/0 : 116[4] -> 108[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96892:97308 [4] NCCL INFO Channel 07/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96890:97313 [2] NCCL INFO Channel 00/0 : 26[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32707:33124 [3] NCCL INFO Channel 09/0 : 43[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32705:33128 [1] NCCL INFO Channel 11/0 : 41[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227467:1227886 [4] NCCL INFO Channel 06/0 : 84[4] -> 76[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96568:97007 [5] NCCL INFO Channel 14/0 : 37[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153562:153983 [0] NCCL INFO Channel 11/0 : 88[0] -> 72[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114519:114931 [7] NCCL INFO Channel 12/0 : 71[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114518:114937 [6] NCCL INFO Channel 14/0 : 70[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62882:63300 [1] NCCL INFO Channel 08/0 : 49[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62884:63299 [3] NCCL INFO Channel 08/0 : 51[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557270:1557688 [4] NCCL INFO Channel 03/0 : 108[4] -> 100[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557271:1557690 [5] NCCL INFO Channel 06/0 : 117[5] -> 101[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557267:1557683 [1] NCCL INFO Channel 06/0 : 97[1] -> 81[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235651:236068 [3] NCCL INFO Channel 10/0 : 115[3] -> 107[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235649:236071 [1] NCCL INFO Channel 11/0 : 113[1] -> 105[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96889:97309 [1] NCCL INFO Channel 07/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32711:33127 [7] NCCL INFO Channel 01/0 : 47[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153563:153986 [1] NCCL INFO Channel 08/0 : 89[1] -> 81[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114515:114936 [3] NCCL INFO Channel 03/0 : 75[3] -> 67[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62888:63302 [7] NCCL INFO Channel 03/0 : 55[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557267:1557683 [1] NCCL INFO Channel 08/0 : 97[1] -> 81[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557271:1557690 [5] NCCL INFO Channel 08/0 : 117[5] -> 101[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557270:1557688 [4] NCCL INFO Channel 05/0 : 108[4] -> 100[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235655:236069 [7] NCCL INFO Channel 04/0 : 119[7] -> 111[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96892:97308 [4] NCCL INFO Channel 08/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96890:97313 [2] NCCL INFO Channel 04/0 : 26[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364903:1365334 [1] NCCL INFO Channel 07/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32709:33129 [5] NCCL INFO Channel 04/0 : 29[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153568:153982 [6] NCCL INFO Channel 11/0 : 94[6] -> 78[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153565:153985 [3] NCCL INFO Channel 10/0 : 91[3] -> 83[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114519:114931 [7] NCCL INFO Channel 14/0 : 71[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114512:114930 [0] NCCL INFO Channel 01/0 : 72[0] -> 64[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557267:1557683 [1] NCCL INFO Channel 10/0 : 97[1] -> 81[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235649:236071 [1] NCCL INFO Channel 12/0 : 113[1] -> 105[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235651:236068 [3] NCCL INFO Channel 12/0 : 115[3] -> 107[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272151:272571 [1] NCCL INFO Channel 08/0 : 121[1] -> 113[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96889:97309 [1] NCCL INFO Channel 08/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815965:2816395 [2] NCCL INFO Channel 00/0 : 58[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364909:1365327 [7] NCCL INFO Channel 01/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364905:1365328 [3] NCCL INFO Channel 07/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32705:33128 [1] NCCL INFO Channel 13/0 : 41[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227464:1227884 [1] NCCL INFO Channel 04/0 : 89[1] -> 81[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227467:1227886 [4] NCCL INFO Channel 07/0 : 84[4] -> 76[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96565:97004 [2] NCCL INFO Channel 01/0 : 42[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96566:97010 [3] NCCL INFO Channel 00/0 : 51[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153569:153981 [7] NCCL INFO Channel 00/0 : 95[7] -> 87[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153566:153980 [4] NCCL INFO Channel 03/0 : 92[4] -> 76[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62882:63300 [1] NCCL INFO Channel 10/0 : 49[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62884:63299 [3] NCCL INFO Channel 09/0 : 51[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557270:1557688 [4] NCCL INFO Channel 07/0 : 108[4] -> 100[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557271:1557690 [5] NCCL INFO Channel 10/0 : 117[5] -> 101[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235655:236069 [7] NCCL INFO Channel 05/0 : 119[7] -> 111[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272154:272569 [4] NCCL INFO Channel 09/0 : 116[4] -> 108[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364903:1365334 [1] NCCL INFO Channel 08/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32711:33127 [7] NCCL INFO Channel 03/0 : 47[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32707:33124 [3] NCCL INFO Channel 13/0 : 43[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227467:1227886 [4] NCCL INFO Channel 08/0 : 84[4] -> 76[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96563:97003 [0] NCCL INFO Channel 02/0 : 48[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96566:97010 [3] NCCL INFO Channel 02/0 : 51[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153562:153983 [0] NCCL INFO Channel 13/0 : 88[0] -> 72[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62888:63302 [7] NCCL INFO Channel 04/0 : 55[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557267:1557683 [1] NCCL INFO Channel 12/0 : 97[1] -> 81[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235649:236071 [1] NCCL INFO Channel 13/0 : 113[1] -> 105[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272151:272571 [1] NCCL INFO Channel 10/0 : 121[1] -> 113[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96892:97308 [4] NCCL INFO Channel 09/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96890:97313 [2] NCCL INFO Channel 06/0 : 26[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815965:2816395 [2] NCCL INFO Channel 04/0 : 58[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364905:1365328 [3] NCCL INFO Channel 08/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32709:33129 [5] NCCL INFO Channel 06/0 : 29[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96565:97004 [2] NCCL INFO Channel 03/0 : 42[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153563:153986 [1] NCCL INFO Channel 10/0 : 89[1] -> 81[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114515:114936 [3] NCCL INFO Channel 05/0 : 75[3] -> 67[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62884:63299 [3] NCCL INFO Channel 10/0 : 51[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557270:1557688 [4] NCCL INFO Channel 09/0 : 108[4] -> 100[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272154:272569 [4] NCCL INFO Channel 10/0 : 116[4] -> 108[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96889:97309 [1] NCCL INFO Channel 10/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32705:33128 [1] NCCL INFO Channel 15/0 : 41[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96566:97010 [3] NCCL INFO Channel 04/0 : 51[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96565:97004 [2] NCCL INFO Channel 05/0 : 42[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96563:97003 [0] NCCL INFO Channel 04/0 : 48[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114512:114930 [0] NCCL INFO Channel 03/0 : 72[0] -> 64[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557271:1557690 [5] NCCL INFO Channel 12/0 : 117[5] -> 101[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96892:97308 [4] NCCL INFO Channel 10/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32707:33124 [3] NCCL INFO Channel 15/0 : 43[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227464:1227884 [1] NCCL INFO Channel 06/0 : 89[1] -> 81[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114515:114936 [3] NCCL INFO Channel 07/0 : 75[3] -> 67[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557267:1557683 [1] NCCL INFO Channel 14/0 : 97[1] -> 81[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96890:97313 [2] NCCL INFO Channel 08/0 : 26[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96889:97309 [1] NCCL INFO Channel 11/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815965:2816395 [2] NCCL INFO Channel 06/0 : 58[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364903:1365334 [1] NCCL INFO Channel 10/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364909:1365327 [7] NCCL INFO Channel 02/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32709:33129 [5] NCCL INFO Channel 08/0 : 29[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32711:33127 [7] NCCL INFO Channel 05/0 : 47[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227467:1227886 [4] NCCL INFO Channel 09/0 : 84[4] -> 76[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153565:153985 [3] NCCL INFO Channel 12/0 : 91[3] -> 83[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114512:114930 [0] NCCL INFO Channel 05/0 : 72[0] -> 64[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62882:63300 [1] NCCL INFO Channel 11/0 : 49[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557270:1557688 [4] NCCL INFO Channel 11/0 : 108[4] -> 100[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272151:272571 [1] NCCL INFO Channel 12/0 : 121[1] -> 113[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272154:272569 [4] NCCL INFO Channel 11/0 : 116[4] -> 108[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364905:1365328 [3] NCCL INFO Channel 09/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32705:33128 [1] NCCL INFO Channel 01/0 : 25[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227464:1227884 [1] NCCL INFO Channel 08/0 : 89[1] -> 81[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96565:97004 [2] NCCL INFO Channel 07/0 : 42[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96566:97010 [3] NCCL INFO Channel 06/0 : 51[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153568:153982 [6] NCCL INFO Channel 13/0 : 94[6] -> 78[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114518:114937 [6] NCCL INFO Channel 00/0 : 102[6] -> 70[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62888:63302 [7] NCCL INFO Channel 05/0 : 55[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557271:1557690 [5] NCCL INFO Channel 14/0 : 117[5] -> 101[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272154:272569 [4] NCCL INFO Channel 13/0 : 116[4] -> 108[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272151:272571 [1] NCCL INFO Channel 14/0 : 121[1] -> 113[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364903:1365334 [1] NCCL INFO Channel 11/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32707:33124 [3] NCCL INFO Channel 01/0 : 27[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227467:1227886 [4] NCCL INFO Channel 10/0 : 84[4] -> 76[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96568:97007 [5] NCCL INFO Channel 01/0 : 45[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96563:97003 [0] NCCL INFO Channel 06/0 : 48[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153566:153980 [4] NCCL INFO Channel 05/0 : 92[4] -> 76[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153562:153983 [0] NCCL INFO Channel 15/0 : 88[0] -> 72[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153569:153981 [7] NCCL INFO Channel 02/0 : 95[7] -> 87[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114515:114936 [3] NCCL INFO Channel 09/0 : 75[3] -> 67[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62884:63299 [3] NCCL INFO Channel 12/0 : 51[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557270:1557688 [4] NCCL INFO Channel 13/0 : 108[4] -> 100[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815965:2816395 [2] NCCL INFO Channel 08/0 : 58[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364909:1365327 [7] NCCL INFO Channel 03/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32709:33129 [5] NCCL INFO Channel 10/0 : 29[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96565:97004 [2] NCCL INFO Channel 09/0 : 42[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96566:97010 [3] NCCL INFO Channel 08/0 : 51[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153563:153986 [1] NCCL INFO Channel 12/0 : 89[1] -> 81[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62882:63300 [1] NCCL INFO Channel 12/0 : 49[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62888:63302 [7] NCCL INFO Channel 06/0 : 55[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557271:1557690 [5] NCCL INFO Channel 00/0 : 101[5] -> 85[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364905:1365328 [3] NCCL INFO Channel 10/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32705:33128 [1] NCCL INFO Channel 03/0 : 25[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96568:97007 [5] NCCL INFO Channel 03/0 : 45[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153565:153985 [3] NCCL INFO Channel 14/0 : 91[3] -> 83[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557270:1557688 [4] NCCL INFO Channel 15/0 : 108[4] -> 100[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272154:272569 [4] NCCL INFO Channel 14/0 : 116[4] -> 108[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364903:1365334 [1] NCCL INFO Channel 12/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32704:33123 [0] NCCL INFO Channel 01/0 : 40[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32710:33125 [6] NCCL INFO Channel 01/0 : 46[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227464:1227884 [1] NCCL INFO Channel 10/0 : 89[1] -> 81[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153568:153982 [6] NCCL INFO Channel 15/0 : 94[6] -> 78[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557271:1557690 [5] NCCL INFO Channel 02/0 : 101[5] -> 85[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272151:272571 [1] NCCL INFO Channel 00/0 : 113[1] -> 105[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364909:1365327 [7] NCCL INFO Channel 04/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32711:33127 [7] NCCL INFO Channel 07/0 : 47[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227467:1227886 [4] NCCL INFO Channel 11/0 : 84[4] -> 76[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153566:153980 [4] NCCL INFO Channel 07/0 : 92[4] -> 76[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153569:153981 [7] NCCL INFO Channel 04/0 : 95[7] -> 87[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153563:153986 [1] NCCL INFO Channel 14/0 : 89[1] -> 81[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114512:114930 [0] NCCL INFO Channel 07/0 : 72[0] -> 64[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62884:63299 [3] NCCL INFO Channel 13/0 : 51[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815965:2816395 [2] NCCL INFO Channel 10/0 : 58[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32707:33124 [3] NCCL INFO Channel 03/0 : 27[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227464:1227884 [1] NCCL INFO Channel 12/0 : 89[1] -> 81[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96566:97010 [3] NCCL INFO Channel 10/0 : 51[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96565:97004 [2] NCCL INFO Channel 11/0 : 42[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153566:153980 [4] NCCL INFO Channel 09/0 : 92[4] -> 76[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153569:153981 [7] NCCL INFO Channel 06/0 : 95[7] -> 87[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114518:114937 [6] NCCL INFO Channel 02/0 : 102[6] -> 70[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62882:63300 [1] NCCL INFO Channel 13/0 : 49[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62888:63302 [7] NCCL INFO Channel 08/0 : 55[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32705:33128 [1] NCCL INFO Channel 05/0 : 25[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32709:33129 [5] NCCL INFO Channel 12/0 : 29[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227467:1227886 [4] NCCL INFO Channel 13/0 : 84[4] -> 76[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96563:97003 [0] NCCL INFO Channel 08/0 : 48[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153569:153981 [7] NCCL INFO Channel 08/0 : 95[7] -> 87[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153566:153980 [4] NCCL INFO Channel 11/0 : 92[4] -> 76[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114515:114936 [3] NCCL INFO Channel 13/0 : 75[3] -> 67[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62884:63299 [3] NCCL INFO Channel 14/0 : 51[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364905:1365328 [3] NCCL INFO Channel 12/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32704:33123 [0] NCCL INFO Channel 03/0 : 40[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96568:97007 [5] NCCL INFO Channel 05/0 : 45[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96570:97005 [7] NCCL INFO Channel 00/0 : 55[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153564:153984 [2] NCCL INFO Channel 01/0 : 106[2] -> 90[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114512:114930 [0] NCCL INFO Channel 09/0 : 72[0] -> 64[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62882:63300 [1] NCCL INFO Channel 14/0 : 49[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62888:63302 [7] NCCL INFO Channel 09/0 : 55[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557271:1557690 [5] NCCL INFO Channel 04/0 : 101[5] -> 85[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272154:272569 [4] NCCL INFO Channel 15/0 : 116[4] -> 108[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364903:1365334 [1] NCCL INFO Channel 13/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32710:33125 [6] NCCL INFO Channel 03/0 : 46[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32711:33127 [7] NCCL INFO Channel 09/0 : 47[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96565:97004 [2] NCCL INFO Channel 13/0 : 42[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153562:153983 [0] NCCL INFO Channel 02/0 : 88[0] -> 80[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114518:114937 [6] NCCL INFO Channel 04/0 : 102[6] -> 70[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557269:1557684 [3] NCCL INFO Channel 00/0 : 115[3] -> 99[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557271:1557690 [5] NCCL INFO Channel 06/0 : 101[5] -> 85[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272151:272571 [1] NCCL INFO Channel 02/0 : 113[1] -> 105[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815965:2816395 [2] NCCL INFO Channel 12/0 : 58[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364909:1365327 [7] NCCL INFO Channel 05/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630209:1630651 [4] NCCL INFO Channel 13/0 : 92[4] -> 76[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96566:97010 [3] NCCL INFO Channel 12/0 : 51[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153566:153980 [4] NCCL INFO Channel 13/0 : 92[4] -> 76[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153569:153981 [7] NCCL INFO Channel 10/0 : 95[7] -> 87[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114515:114936 [3] NCCL INFO Channel 15/0 : 75[3] -> 67[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557267:1557683 [1] NCCL INFO Channel 01/0 : 105[1] -> 97[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272155:272574 [5] NCCL INFO Channel 00/0 : 125[5] -> 117[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364905:1365328 [3] NCCL INFO Channel 13/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227467:1227886 [4] NCCL INFO Channel 14/0 : 84[4] -> 76[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227464:1227884 [1] NCCL INFO Channel 14/0 : 89[1] -> 81[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630205:1630646 [0] NCCL INFO Channel 15/0 : 88[0] -> 72[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630206:1630647 [1] NCCL INFO Channel 15/0 : 89[1] -> 73[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96563:97003 [0] NCCL INFO Channel 10/0 : 48[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153564:153984 [2] NCCL INFO Channel 03/0 : 106[2] -> 90[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114519:114931 [7] NCCL INFO Channel 01/0 : 79[7] -> 71[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62884:63299 [3] NCCL INFO Channel 15/0 : 51[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557268:1557686 [2] NCCL INFO Channel 00/0 : 114[2] -> 98[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272151:272571 [1] NCCL INFO Channel 03/0 : 113[1] -> 105[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364903:1365334 [1] NCCL INFO Channel 14/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32707:33124 [3] NCCL INFO Channel 05/0 : 27[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630209:1630651 [4] NCCL INFO Channel 15/0 : 92[4] -> 76[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96568:97007 [5] NCCL INFO Channel 07/0 : 45[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153562:153983 [0] NCCL INFO Channel 04/0 : 88[0] -> 80[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153568:153982 [6] NCCL INFO Channel 00/0 : 94[6] -> 86[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114512:114930 [0] NCCL INFO Channel 11/0 : 72[0] -> 64[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114518:114937 [6] NCCL INFO Channel 08/0 : 102[6] -> 70[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62882:63300 [1] NCCL INFO Channel 15/0 : 49[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62888:63302 [7] NCCL INFO Channel 10/0 : 55[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557271:1557690 [5] NCCL INFO Channel 08/0 : 101[5] -> 85[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557269:1557684 [3] NCCL INFO Channel 02/0 : 115[3] -> 99[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557267:1557683 [1] NCCL INFO Channel 03/0 : 105[1] -> 97[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815965:2816395 [2] NCCL INFO Channel 14/0 : 58[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364909:1365327 [7] NCCL INFO Channel 06/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32705:33128 [1] NCCL INFO Channel 07/0 : 25[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32709:33129 [5] NCCL INFO Channel 14/0 : 29[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630210:1630653 [5] NCCL INFO Channel 00/0 : 85[5] -> 77[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153566:153980 [4] NCCL INFO Channel 15/0 : 92[4] -> 76[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153569:153981 [7] NCCL INFO Channel 12/0 : 95[7] -> 87[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62884:63299 [3] NCCL INFO Channel 01/0 : 43[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557266:1557689 [0] NCCL INFO Channel 02/0 : 112[0] -> 96[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557268:1557686 [2] NCCL INFO Channel 04/0 : 114[2] -> 98[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364905:1365328 [3] NCCL INFO Channel 14/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32711:33127 [7] NCCL INFO Channel 11/0 : 47[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227467:1227886 [4] NCCL INFO Channel 15/0 : 84[4] -> 76[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630210:1630653 [5] NCCL INFO Channel 01/0 : 85[5] -> 77[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557271:1557690 [5] NCCL INFO Channel 10/0 : 101[5] -> 85[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557267:1557683 [1] NCCL INFO Channel 05/0 : 105[1] -> 97[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557269:1557684 [3] NCCL INFO Channel 04/0 : 115[3] -> 99[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272155:272574 [5] NCCL INFO Channel 02/0 : 125[5] -> 117[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272151:272571 [1] NCCL INFO Channel 04/0 : 113[1] -> 105[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364903:1365334 [1] NCCL INFO Channel 15/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32704:33123 [0] NCCL INFO Channel 05/0 : 40[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32710:33125 [6] NCCL INFO Channel 05/0 : 46[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32707:33124 [3] NCCL INFO Channel 07/0 : 27[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227464:1227884 [1] NCCL INFO Channel 00/0 : 81[1] -> 73[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630210:1630653 [5] NCCL INFO Channel 02/0 : 85[5] -> 77[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96570:97005 [7] NCCL INFO Channel 02/0 : 55[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114519:114931 [7] NCCL INFO Channel 03/0 : 79[7] -> 71[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557266:1557689 [0] NCCL INFO Channel 04/0 : 112[0] -> 96[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557268:1557686 [2] NCCL INFO Channel 06/0 : 114[2] -> 98[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272151:272571 [1] NCCL INFO Channel 05/0 : 113[1] -> 105[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364909:1365327 [7] NCCL INFO Channel 08/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32705:33128 [1] NCCL INFO Channel 11/0 : 25[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227468:1227883 [5] NCCL INFO Channel 00/0 : 93[5] -> 85[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630210:1630653 [5] NCCL INFO Channel 03/0 : 85[5] -> 77[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96565:97004 [2] NCCL INFO Channel 15/0 : 42[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153564:153984 [2] NCCL INFO Channel 05/0 : 106[2] -> 90[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114512:114930 [0] NCCL INFO Channel 13/0 : 72[0] -> 64[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114518:114937 [6] NCCL INFO Channel 10/0 : 102[6] -> 70[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62882:63300 [1] NCCL INFO Channel 01/0 : 41[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62888:63302 [7] NCCL INFO Channel 11/0 : 55[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557271:1557690 [5] NCCL INFO Channel 12/0 : 101[5] -> 85[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272155:272574 [5] NCCL INFO Channel 04/0 : 125[5] -> 117[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32711:33127 [7] NCCL INFO Channel 13/0 : 47[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630210:1630653 [5] NCCL INFO Channel 04/0 : 85[5] -> 77[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96566:97010 [3] NCCL INFO Channel 14/0 : 51[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96563:97003 [0] NCCL INFO Channel 12/0 : 48[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153562:153983 [0] NCCL INFO Channel 06/0 : 88[0] -> 80[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114519:114931 [7] NCCL INFO Channel 05/0 : 79[7] -> 71[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62884:63299 [3] NCCL INFO Channel 03/0 : 43[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557267:1557683 [1] NCCL INFO Channel 07/0 : 105[1] -> 97[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32707:33124 [3] NCCL INFO Channel 09/0 : 27[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32704:33123 [0] NCCL INFO Channel 07/0 : 40[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630210:1630653 [5] NCCL INFO Channel 06/0 : 85[5] -> 77[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96568:97007 [5] NCCL INFO Channel 09/0 : 45[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153568:153982 [6] NCCL INFO Channel 02/0 : 94[6] -> 86[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114512:114930 [0] NCCL INFO Channel 15/0 : 72[0] -> 64[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114518:114937 [6] NCCL INFO Channel 12/0 : 102[6] -> 70[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557269:1557684 [3] NCCL INFO Channel 06/0 : 115[3] -> 99[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364905:1365328 [3] NCCL INFO Channel 15/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32710:33125 [6] NCCL INFO Channel 07/0 : 46[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227464:1227884 [1] NCCL INFO Channel 02/0 : 81[1] -> 73[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630210:1630653 [5] NCCL INFO Channel 07/0 : 85[5] -> 77[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96570:97005 [7] NCCL INFO Channel 04/0 : 55[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153569:153981 [7] NCCL INFO Channel 14/0 : 95[7] -> 87[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114519:114931 [7] NCCL INFO Channel 07/0 : 79[7] -> 71[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557266:1557689 [0] NCCL INFO Channel 06/0 : 112[0] -> 96[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557268:1557686 [2] NCCL INFO Channel 08/0 : 114[2] -> 98[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272151:272571 [1] NCCL INFO Channel 06/0 : 113[1] -> 105[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364903:1365334 [1] NCCL INFO Channel 01/0 : 9[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32705:33128 [1] NCCL INFO Channel 13/0 : 25[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227468:1227883 [5] NCCL INFO Channel 02/0 : 93[5] -> 85[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630210:1630653 [5] NCCL INFO Channel 08/0 : 85[5] -> 77[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96566:97010 [3] NCCL INFO Channel 00/0 : 35[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153564:153984 [2] NCCL INFO Channel 07/0 : 106[2] -> 90[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153562:153983 [0] NCCL INFO Channel 08/0 : 88[0] -> 80[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153568:153982 [6] NCCL INFO Channel 04/0 : 94[6] -> 86[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114518:114937 [6] NCCL INFO Channel 14/0 : 102[6] -> 70[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62888:63302 [7] NCCL INFO Channel 12/0 : 55[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62882:63300 [1] NCCL INFO Channel 03/0 : 41[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62884:63299 [3] NCCL INFO Channel 05/0 : 43[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557271:1557690 [5] NCCL INFO Channel 14/0 : 101[5] -> 85[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272155:272574 [5] NCCL INFO Channel 06/0 : 125[5] -> 117[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364909:1365327 [7] NCCL INFO Channel 09/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32708:33126 [4] NCCL INFO Channel 01/0 : 44[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227464:1227884 [1] NCCL INFO Channel 03/0 : 81[1] -> 73[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630210:1630653 [5] NCCL INFO Channel 09/0 : 85[5] -> 77[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96568:97007 [5] NCCL INFO Channel 11/0 : 45[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153562:153983 [0] NCCL INFO Channel 10/0 : 88[0] -> 80[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114519:114931 [7] NCCL INFO Channel 09/0 : 79[7] -> 71[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557267:1557683 [1] NCCL INFO Channel 11/0 : 105[1] -> 97[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364905:1365328 [3] NCCL INFO Channel 01/0 : 11[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32711:33127 [7] NCCL INFO Channel 01/0 : 31[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630210:1630653 [5] NCCL INFO Channel 10/0 : 85[5] -> 77[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96570:97005 [7] NCCL INFO Channel 06/0 : 55[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153568:153982 [6] NCCL INFO Channel 08/0 : 94[6] -> 86[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153564:153984 [2] NCCL INFO Channel 09/0 : 106[2] -> 90[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557269:1557684 [3] NCCL INFO Channel 08/0 : 115[3] -> 99[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364903:1365334 [1] NCCL INFO Channel 03/0 : 9[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32707:33124 [3] NCCL INFO Channel 13/0 : 27[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630210:1630653 [5] NCCL INFO Channel 11/0 : 85[5] -> 77[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96563:97003 [0] NCCL INFO Channel 14/0 : 48[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153562:153983 [0] NCCL INFO Channel 12/0 : 88[0] -> 80[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62887:63305 [6] NCCL INFO Channel 00/0 : 54[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557268:1557686 [2] NCCL INFO Channel 10/0 : 114[2] -> 98[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557266:1557689 [0] NCCL INFO Channel 08/0 : 112[0] -> 96[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557273:1557685 [7] NCCL INFO Channel 00/0 : 119[7] -> 103[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272151:272571 [1] NCCL INFO Channel 07/0 : 113[1] -> 105[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364909:1365327 [7] NCCL INFO Channel 10/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32705:33128 [1] NCCL INFO Channel 15/0 : 25[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32704:33123 [0] NCCL INFO Channel 09/0 : 40[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630210:1630653 [5] NCCL INFO Channel 12/0 : 85[5] -> 77[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153568:153982 [6] NCCL INFO Channel 10/0 : 94[6] -> 86[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557267:1557683 [1] NCCL INFO Channel 13/0 : 105[1] -> 97[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272155:272574 [5] NCCL INFO Channel 08/0 : 125[5] -> 117[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364905:1365328 [3] NCCL INFO Channel 03/0 : 11[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32710:33125 [6] NCCL INFO Channel 09/0 : 46[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227468:1227883 [5] NCCL INFO Channel 04/0 : 93[5] -> 85[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630210:1630653 [5] NCCL INFO Channel 14/0 : 85[5] -> 77[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153564:153984 [2] NCCL INFO Channel 11/0 : 106[2] -> 90[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114518:114937 [6] NCCL INFO Channel 00/0 : 70[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557269:1557684 [3] NCCL INFO Channel 10/0 : 115[3] -> 99[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272151:272571 [1] NCCL INFO Channel 08/0 : 113[1] -> 105[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364903:1365334 [1] NCCL INFO Channel 05/0 : 9[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32708:33126 [4] NCCL INFO Channel 03/0 : 44[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227464:1227884 [1] NCCL INFO Channel 04/0 : 81[1] -> 73[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630210:1630653 [5] NCCL INFO Channel 15/0 : 85[5] -> 77[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153562:153983 [0] NCCL INFO Channel 14/0 : 88[0] -> 80[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114519:114931 [7] NCCL INFO Channel 11/0 : 79[7] -> 71[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62888:63302 [7] NCCL INFO Channel 13/0 : 55[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62882:63300 [1] NCCL INFO Channel 05/0 : 41[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62884:63299 [3] NCCL INFO Channel 07/0 : 43[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557268:1557686 [2] NCCL INFO Channel 12/0 : 114[2] -> 98[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364909:1365327 [7] NCCL INFO Channel 11/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32711:33127 [7] NCCL INFO Channel 03/0 : 31[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32707:33124 [3] NCCL INFO Channel 15/0 : 27[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227468:1227883 [5] NCCL INFO Channel 06/0 : 93[5] -> 85[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630210:1630653 [5] NCCL INFO Channel 01/0 : 77[5] -> 69[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153566:153980 [4] NCCL INFO Channel 00/0 : 92[4] -> 84[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114518:114937 [6] NCCL INFO Channel 02/0 : 70[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62887:63305 [6] NCCL INFO Channel 01/0 : 54[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557266:1557689 [0] NCCL INFO Channel 10/0 : 112[0] -> 96[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32704:33123 [0] NCCL INFO Channel 11/0 : 40[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630210:1630653 [5] NCCL INFO Channel 03/0 : 77[5] -> 69[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153568:153982 [6] NCCL INFO Channel 12/0 : 94[6] -> 86[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114519:114931 [7] NCCL INFO Channel 13/0 : 79[7] -> 71[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557267:1557683 [1] NCCL INFO Channel 15/0 : 105[1] -> 97[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272155:272574 [5] NCCL INFO Channel 10/0 : 125[5] -> 117[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32710:33125 [6] NCCL INFO Channel 11/0 : 46[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630210:1630653 [5] NCCL INFO Channel 05/0 : 77[5] -> 69[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96566:97010 [3] NCCL INFO Channel 02/0 : 35[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153564:153984 [2] NCCL INFO Channel 13/0 : 106[2] -> 90[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114518:114937 [6] NCCL INFO Channel 04/0 : 70[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557273:1557685 [7] NCCL INFO Channel 02/0 : 119[7] -> 103[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272151:272571 [1] NCCL INFO Channel 10/0 : 113[1] -> 105[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364905:1365328 [3] NCCL INFO Channel 05/0 : 11[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364903:1365334 [1] NCCL INFO Channel 07/0 : 9[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32711:33127 [7] NCCL INFO Channel 05/0 : 31[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227464:1227884 [1] NCCL INFO Channel 05/0 : 81[1] -> 73[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630210:1630653 [5] NCCL INFO Channel 07/0 : 77[5] -> 69[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96568:97007 [5] NCCL INFO Channel 15/0 : 45[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62882:63300 [1] NCCL INFO Channel 07/0 : 41[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62888:63302 [7] NCCL INFO Channel 14/0 : 55[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62884:63299 [3] NCCL INFO Channel 09/0 : 43[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557269:1557684 [3] NCCL INFO Channel 12/0 : 115[3] -> 99[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364909:1365327 [7] NCCL INFO Channel 12/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32708:33126 [4] NCCL INFO Channel 05/0 : 44[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227464:1227884 [1] NCCL INFO Channel 06/0 : 81[1] -> 73[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630208:1630650 [3] NCCL INFO Channel 00/0 : 83[3] -> 75[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96570:97005 [7] NCCL INFO Channel 08/0 : 55[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62887:63305 [6] NCCL INFO Channel 02/0 : 54[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557268:1557686 [2] NCCL INFO Channel 14/0 : 114[2] -> 98[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364905:1365328 [3] NCCL INFO Channel 07/0 : 11[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364903:1365334 [1] NCCL INFO Channel 11/0 : 9[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32704:33123 [0] NCCL INFO Channel 13/0 : 40[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630210:1630653 [5] NCCL INFO Channel 09/0 : 77[5] -> 69[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96563:97003 [0] NCCL INFO Channel 02/0 : 32[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153566:153980 [4] NCCL INFO Channel 02/0 : 92[4] -> 84[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114518:114937 [6] NCCL INFO Channel 08/0 : 70[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557266:1557689 [0] NCCL INFO Channel 12/0 : 112[0] -> 96[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272155:272574 [5] NCCL INFO Channel 12/0 : 125[5] -> 117[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364909:1365327 [7] NCCL INFO Channel 13/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32710:33125 [6] NCCL INFO Channel 13/0 : 46[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630206:1630647 [1] NCCL INFO Channel 00/0 : 81[1] -> 73[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96566:97010 [3] NCCL INFO Channel 04/0 : 35[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153568:153982 [6] NCCL INFO Channel 14/0 : 94[6] -> 86[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114518:114937 [6] NCCL INFO Channel 10/0 : 70[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272151:272571 [1] NCCL INFO Channel 11/0 : 113[1] -> 105[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364903:1365334 [1] NCCL INFO Channel 13/0 : 9[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364905:1365328 [3] NCCL INFO Channel 09/0 : 11[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32711:33127 [7] NCCL INFO Channel 07/0 : 31[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227464:1227884 [1] NCCL INFO Channel 07/0 : 81[1] -> 73[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630208:1630650 [3] NCCL INFO Channel 01/0 : 83[3] -> 75[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96570:97005 [7] NCCL INFO Channel 10/0 : 55[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153564:153984 [2] NCCL INFO Channel 15/0 : 106[2] -> 90[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153566:153980 [4] NCCL INFO Channel 06/0 : 92[4] -> 84[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114518:114937 [6] NCCL INFO Channel 12/0 : 70[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62881:63298 [0] NCCL INFO Channel 01/0 : 48[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272155:272574 [5] NCCL INFO Channel 14/0 : 125[5] -> 117[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364908:1365332 [6] NCCL INFO Channel 00/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364909:1365327 [7] NCCL INFO Channel 14/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32708:33126 [4] NCCL INFO Channel 07/0 : 44[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630210:1630653 [5] NCCL INFO Channel 11/0 : 77[5] -> 69[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96563:97003 [0] NCCL INFO Channel 04/0 : 32[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96566:97010 [3] NCCL INFO Channel 06/0 : 35[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153566:153980 [4] NCCL INFO Channel 08/0 : 92[4] -> 84[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153564:153984 [2] NCCL INFO Channel 01/0 : 90[2] -> 74[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62882:63300 [1] NCCL INFO Channel 11/0 : 41[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557269:1557684 [3] NCCL INFO Channel 14/0 : 115[3] -> 99[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32704:33123 [0] NCCL INFO Channel 15/0 : 40[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630206:1630647 [1] NCCL INFO Channel 02/0 : 81[1] -> 73[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153564:153984 [2] NCCL INFO Channel 03/0 : 90[2] -> 74[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153566:153980 [4] NCCL INFO Channel 10/0 : 92[4] -> 84[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62888:63302 [7] NCCL INFO Channel 01/0 : 47[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62884:63299 [3] NCCL INFO Channel 13/0 : 43[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557268:1557686 [2] NCCL INFO Channel 00/0 : 98[2] -> 82[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557273:1557685 [7] NCCL INFO Channel 04/0 : 119[7] -> 103[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32710:33125 [6] NCCL INFO Channel 15/0 : 46[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227464:1227884 [1] NCCL INFO Channel 08/0 : 81[1] -> 73[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630208:1630650 [3] NCCL INFO Channel 02/0 : 83[3] -> 75[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630210:1630653 [5] NCCL INFO Channel 15/0 : 77[5] -> 69[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114518:114937 [6] NCCL INFO Channel 14/0 : 70[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557266:1557689 [0] NCCL INFO Channel 14/0 : 112[0] -> 96[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557271:1557690 [5] NCCL INFO Channel 01/0 : 109[5] -> 101[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272151:272571 [1] NCCL INFO Channel 12/0 : 113[1] -> 105[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364905:1365328 [3] NCCL INFO Channel 13/0 : 11[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32711:33127 [7] NCCL INFO Channel 09/0 : 31[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630206:1630647 [1] NCCL INFO Channel 03/0 : 81[1] -> 73[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557269:1557684 [3] NCCL INFO Channel 00/0 : 99[3] -> 83[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557268:1557686 [2] NCCL INFO Channel 04/0 : 98[2] -> 82[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272155:272574 [5] NCCL INFO Channel 00/0 : 117[5] -> 109[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364909:1365327 [7] NCCL INFO Channel 01/0 : 15[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32708:33126 [4] NCCL INFO Channel 09/0 : 44[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630208:1630650 [3] NCCL INFO Channel 04/0 : 83[3] -> 75[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96570:97005 [7] NCCL INFO Channel 12/0 : 55[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96563:97003 [0] NCCL INFO Channel 06/0 : 32[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153564:153984 [2] NCCL INFO Channel 05/0 : 90[2] -> 74[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153566:153980 [4] NCCL INFO Channel 12/0 : 92[4] -> 84[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62887:63305 [6] NCCL INFO Channel 03/0 : 54[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62881:63298 [0] NCCL INFO Channel 02/0 : 48[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557273:1557685 [7] NCCL INFO Channel 06/0 : 119[7] -> 103[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272151:272571 [1] NCCL INFO Channel 13/0 : 113[1] -> 105[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364908:1365332 [6] NCCL INFO Channel 01/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364903:1365334 [1] NCCL INFO Channel 15/0 : 9[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32704:33123 [0] NCCL INFO Channel 01/0 : 24[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227464:1227884 [1] NCCL INFO Channel 10/0 : 81[1] -> 73[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630206:1630647 [1] NCCL INFO Channel 04/0 : 81[1] -> 73[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96566:97010 [3] NCCL INFO Channel 08/0 : 35[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153564:153984 [2] NCCL INFO Channel 07/0 : 90[2] -> 74[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153566:153980 [4] NCCL INFO Channel 14/0 : 92[4] -> 84[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62882:63300 [1] NCCL INFO Channel 13/0 : 41[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557266:1557689 [0] NCCL INFO Channel 02/0 : 96[0] -> 80[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364905:1365328 [3] NCCL INFO Channel 15/0 : 11[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32710:33125 [6] NCCL INFO Channel 01/0 : 30[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630208:1630650 [3] NCCL INFO Channel 05/0 : 83[3] -> 75[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96570:97005 [7] NCCL INFO Channel 14/0 : 55[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96563:97003 [0] NCCL INFO Channel 08/0 : 32[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96566:97010 [3] NCCL INFO Channel 10/0 : 35[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62888:63302 [7] NCCL INFO Channel 03/0 : 47[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557271:1557690 [5] NCCL INFO Channel 03/0 : 109[5] -> 101[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364909:1365327 [7] NCCL INFO Channel 03/0 : 15[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32711:33127 [7] NCCL INFO Channel 11/0 : 31[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630206:1630647 [1] NCCL INFO Channel 05/0 : 81[1] -> 73[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96570:97005 [7] NCCL INFO Channel 00/0 : 39[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62884:63299 [3] NCCL INFO Channel 15/0 : 43[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557269:1557684 [3] NCCL INFO Channel 02/0 : 99[3] -> 83[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557268:1557686 [2] NCCL INFO Channel 06/0 : 98[2] -> 82[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557273:1557685 [7] NCCL INFO Channel 08/0 : 119[7] -> 103[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364908:1365332 [6] NCCL INFO Channel 02/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32708:33126 [4] NCCL INFO Channel 11/0 : 44[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630208:1630650 [3] NCCL INFO Channel 06/0 : 83[3] -> 75[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96563:97003 [0] NCCL INFO Channel 10/0 : 32[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96566:97010 [3] NCCL INFO Channel 12/0 : 35[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557266:1557689 [0] NCCL INFO Channel 04/0 : 96[0] -> 80[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272155:272574 [5] NCCL INFO Channel 01/0 : 117[5] -> 109[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32705:33128 [1] NCCL INFO Channel 00/0 : 25[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32704:33123 [0] NCCL INFO Channel 03/0 : 24[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227464:1227884 [1] NCCL INFO Channel 11/0 : 81[1] -> 73[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630206:1630647 [1] NCCL INFO Channel 06/0 : 81[1] -> 73[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153564:153984 [2] NCCL INFO Channel 09/0 : 90[2] -> 74[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557271:1557690 [5] NCCL INFO Channel 05/0 : 109[5] -> 101[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272151:272571 [1] NCCL INFO Channel 14/0 : 113[1] -> 105[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32710:33125 [6] NCCL INFO Channel 03/0 : 30[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32711:33127 [7] NCCL INFO Channel 13/0 : 31[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630212:1630652 [7] NCCL INFO Channel 00/0 : 87[7] -> 79[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630208:1630650 [3] NCCL INFO Channel 07/0 : 83[3] -> 75[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153564:153984 [2] NCCL INFO Channel 11/0 : 90[2] -> 74[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62881:63298 [0] NCCL INFO Channel 03/0 : 48[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62887:63305 [6] NCCL INFO Channel 04/0 : 54[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557268:1557686 [2] NCCL INFO Channel 08/0 : 98[2] -> 82[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557269:1557684 [3] NCCL INFO Channel 04/0 : 99[3] -> 83[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272155:272574 [5] NCCL INFO Channel 02/0 : 117[5] -> 109[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32707:33124 [3] NCCL INFO Channel 00/0 : 27[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630206:1630647 [1] NCCL INFO Channel 07/0 : 81[1] -> 73[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62882:63300 [1] NCCL INFO Channel 15/0 : 41[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557273:1557685 [7] NCCL INFO Channel 10/0 : 119[7] -> 103[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272151:272571 [1] NCCL INFO Channel 15/0 : 113[1] -> 105[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364909:1365327 [7] NCCL INFO Channel 05/0 : 15[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32708:33126 [4] NCCL INFO Channel 13/0 : 44[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630208:1630650 [3] NCCL INFO Channel 08/0 : 83[3] -> 75[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630212:1630652 [7] NCCL INFO Channel 01/0 : 87[7] -> 79[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630206:1630647 [1] NCCL INFO Channel 08/0 : 81[1] -> 73[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96570:97005 [7] NCCL INFO Channel 02/0 : 39[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62885:63304 [4] NCCL INFO Channel 00/0 : 52[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557266:1557689 [0] NCCL INFO Channel 06/0 : 96[0] -> 80[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364908:1365332 [6] NCCL INFO Channel 03/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32704:33123 [0] NCCL INFO Channel 05/0 : 24[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32705:33128 [1] NCCL INFO Channel 02/0 : 25[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227464:1227884 [1] NCCL INFO Channel 12/0 : 81[1] -> 73[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630208:1630650 [3] NCCL INFO Channel 09/0 : 83[3] -> 75[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96563:97003 [0] NCCL INFO Channel 12/0 : 32[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96566:97010 [3] NCCL INFO Channel 14/0 : 35[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153564:153984 [2] NCCL INFO Channel 13/0 : 90[2] -> 74[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62888:63302 [7] NCCL INFO Channel 05/0 : 47[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557271:1557690 [5] NCCL INFO Channel 07/0 : 109[5] -> 101[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272155:272574 [5] NCCL INFO Channel 03/0 : 117[5] -> 109[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364909:1365327 [7] NCCL INFO Channel 07/0 : 15[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32710:33125 [6] NCCL INFO Channel 05/0 : 30[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630206:1630647 [1] NCCL INFO Channel 10/0 : 81[1] -> 73[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96570:97005 [7] NCCL INFO Channel 04/0 : 39[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62881:63298 [0] NCCL INFO Channel 04/0 : 48[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557268:1557686 [2] NCCL INFO Channel 10/0 : 98[2] -> 82[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272152:272575 [2] NCCL INFO Channel 00/0 : 122[2] -> 114[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364908:1365332 [6] NCCL INFO Channel 04/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32707:33124 [3] NCCL INFO Channel 02/0 : 27[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630212:1630652 [7] NCCL INFO Channel 02/0 : 87[7] -> 79[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96563:97003 [0] NCCL INFO Channel 14/0 : 32[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62887:63305 [6] NCCL INFO Channel 05/0 : 54[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557269:1557684 [3] NCCL INFO Channel 06/0 : 99[3] -> 83[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272155:272574 [5] NCCL INFO Channel 04/0 : 117[5] -> 109[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364902:1365333 [0] NCCL INFO Channel 01/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364909:1365327 [7] NCCL INFO Channel 09/0 : 15[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32708:33126 [4] NCCL INFO Channel 15/0 : 44[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227464:1227884 [1] NCCL INFO Channel 13/0 : 81[1] -> 73[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630208:1630650 [3] NCCL INFO Channel 10/0 : 83[3] -> 75[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153564:153984 [2] NCCL INFO Channel 15/0 : 90[2] -> 74[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114518:114937 [6] NCCL INFO Channel 01/0 : 78[6] -> 70[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62888:63302 [7] NCCL INFO Channel 07/0 : 47[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557273:1557685 [7] NCCL INFO Channel 12/0 : 119[7] -> 103[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557266:1557689 [0] NCCL INFO Channel 08/0 : 96[0] -> 80[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557271:1557690 [5] NCCL INFO Channel 09/0 : 109[5] -> 101[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32704:33123 [0] NCCL INFO Channel 07/0 : 24[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32705:33128 [1] NCCL INFO Channel 04/0 : 25[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630206:1630647 [1] NCCL INFO Channel 11/0 : 81[1] -> 73[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630212:1630652 [7] NCCL INFO Channel 03/0 : 87[7] -> 79[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557268:1557686 [2] NCCL INFO Channel 12/0 : 98[2] -> 82[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557269:1557684 [3] NCCL INFO Channel 08/0 : 99[3] -> 83[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272152:272575 [2] NCCL INFO Channel 04/0 : 122[2] -> 114[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32707:33124 [3] NCCL INFO Channel 04/0 : 27[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32710:33125 [6] NCCL INFO Channel 07/0 : 30[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630208:1630650 [3] NCCL INFO Channel 12/0 : 83[3] -> 75[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96570:97005 [7] NCCL INFO Channel 06/0 : 39[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557266:1557689 [0] NCCL INFO Channel 10/0 : 96[0] -> 80[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557273:1557685 [7] NCCL INFO Channel 14/0 : 119[7] -> 103[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235651:236068 [3] NCCL INFO Channel 13/0 : 115[3] -> 107[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272155:272574 [5] NCCL INFO Channel 06/0 : 117[5] -> 109[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364908:1365332 [6] NCCL INFO Channel 05/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32704:33123 [0] NCCL INFO Channel 09/0 : 24[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227464:1227884 [1] NCCL INFO Channel 14/0 : 81[1] -> 73[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630206:1630647 [1] NCCL INFO Channel 12/0 : 81[1] -> 73[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96570:97005 [7] NCCL INFO Channel 08/0 : 39[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114518:114937 [6] NCCL INFO Channel 03/0 : 78[6] -> 70[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62885:63304 [4] NCCL INFO Channel 01/0 : 52[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557271:1557690 [5] NCCL INFO Channel 11/0 : 109[5] -> 101[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557268:1557686 [2] NCCL INFO Channel 14/0 : 98[2] -> 82[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557269:1557684 [3] NCCL INFO Channel 10/0 : 99[3] -> 83[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235648:236064 [0] NCCL INFO Channel 01/0 : 112[0] -> 104[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235655:236069 [7] NCCL INFO Channel 06/0 : 119[7] -> 111[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272152:272575 [2] NCCL INFO Channel 06/0 : 122[2] -> 114[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364906:1365331 [4] NCCL INFO Channel 00/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32708:33126 [4] NCCL INFO Channel 01/0 : 28[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227464:1227884 [1] NCCL INFO Channel 15/0 : 81[1] -> 73[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630212:1630652 [7] NCCL INFO Channel 04/0 : 87[7] -> 79[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96570:97005 [7] NCCL INFO Channel 10/0 : 39[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62881:63298 [0] NCCL INFO Channel 05/0 : 48[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235654:236070 [6] NCCL INFO Channel 00/0 : 118[6] -> 110[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235649:236071 [1] NCCL INFO Channel 14/0 : 113[1] -> 105[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364909:1365327 [7] NCCL INFO Channel 11/0 : 15[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32705:33128 [1] NCCL INFO Channel 06/0 : 25[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630208:1630650 [3] NCCL INFO Channel 13/0 : 83[3] -> 75[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:124160:124581 [0] NCCL INFO Channel 01/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:124162:124587 [2] NCCL INFO Channel 01/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62887:63305 [6] NCCL INFO Channel 07/0 : 54[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557266:1557689 [0] NCCL INFO Channel 12/0 : 96[0] -> 80[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235651:236068 [3] NCCL INFO Channel 14/0 : 115[3] -> 107[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364902:1365333 [0] NCCL INFO Channel 02/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32707:33124 [3] NCCL INFO Channel 06/0 : 27[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32710:33125 [6] NCCL INFO Channel 09/0 : 30[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630206:1630647 [1] NCCL INFO Channel 13/0 : 81[1] -> 73[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114518:114937 [6] NCCL INFO Channel 05/0 : 78[6] -> 70[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62888:63302 [7] NCCL INFO Channel 09/0 : 47[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557273:1557685 [7] NCCL INFO Channel 00/0 : 103[7] -> 87[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235655:236069 [7] NCCL INFO Channel 08/0 : 119[7] -> 111[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235648:236064 [0] NCCL INFO Channel 02/0 : 112[0] -> 104[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272155:272574 [5] NCCL INFO Channel 07/0 : 117[5] -> 109[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364908:1365332 [6] NCCL INFO Channel 07/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32704:33123 [0] NCCL INFO Channel 11/0 : 24[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227468:1227883 [5] NCCL INFO Channel 08/0 : 93[5] -> 85[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630205:1630646 [0] NCCL INFO Channel 01/0 : 80[0] -> 72[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630212:1630652 [7] NCCL INFO Channel 05/0 : 87[7] -> 79[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96569:97006 [6] NCCL INFO Channel 00/0 : 54[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62881:63298 [0] NCCL INFO Channel 06/0 : 48[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557271:1557690 [5] NCCL INFO Channel 15/0 : 109[5] -> 101[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235649:236071 [1] NCCL INFO Channel 15/0 : 113[1] -> 105[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272152:272575 [2] NCCL INFO Channel 08/0 : 122[2] -> 114[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272150:272576 [0] NCCL INFO Channel 02/0 : 120[0] -> 112[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364906:1365331 [4] NCCL INFO Channel 01/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364909:1365327 [7] NCCL INFO Channel 13/0 : 15[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32705:33128 [1] NCCL INFO Channel 08/0 : 25[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630208:1630650 [3] NCCL INFO Channel 14/0 : 83[3] -> 75[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96570:97005 [7] NCCL INFO Channel 12/0 : 39[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62885:63304 [4] NCCL INFO Channel 02/0 : 52[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557269:1557684 [3] NCCL INFO Channel 12/0 : 99[3] -> 83[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235654:236070 [6] NCCL INFO Channel 01/0 : 118[6] -> 110[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272153:272573 [3] NCCL INFO Channel 00/0 : 123[3] -> 115[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32711:33127 [7] NCCL INFO Channel 00/0 : 31[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32708:33126 [4] NCCL INFO Channel 03/0 : 28[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32707:33124 [3] NCCL INFO Channel 08/0 : 27[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630206:1630647 [1] NCCL INFO Channel 14/0 : 81[1] -> 73[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124162:124587 [2] NCCL INFO Channel 03/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:124160:124581 [0] NCCL INFO Channel 03/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96569:97006 [6] NCCL INFO Channel 02/0 : 54[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114518:114937 [6] NCCL INFO Channel 07/0 : 78[6] -> 70[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62888:63302 [7] NCCL INFO Channel 11/0 : 47[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62887:63305 [6] NCCL INFO Channel 08/0 : 54[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557273:1557685 [7] NCCL INFO Channel 02/0 : 103[7] -> 87[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235651:236068 [3] NCCL INFO Channel 15/0 : 115[3] -> 107[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272155:272574 [5] NCCL INFO Channel 08/0 : 117[5] -> 109[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364902:1365333 [0] NCCL INFO Channel 03/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364908:1365332 [6] NCCL INFO Channel 08/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32710:33125 [6] NCCL INFO Channel 11/0 : 30[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227468:1227883 [5] NCCL INFO Channel 10/0 : 93[5] -> 85[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630211:1630649 [6] NCCL INFO Channel 00/0 : 86[6] -> 78[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:124162:124587 [2] NCCL INFO Channel 05/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96566:97010 [3] NCCL INFO Channel 01/0 : 43[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62881:63298 [0] NCCL INFO Channel 07/0 : 48[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557266:1557689 [0] NCCL INFO Channel 14/0 : 96[0] -> 80[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557269:1557684 [3] NCCL INFO Channel 14/0 : 99[3] -> 83[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235655:236069 [7] NCCL INFO Channel 09/0 : 119[7] -> 111[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235648:236064 [0] NCCL INFO Channel 03/0 : 112[0] -> 104[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364906:1365331 [4] NCCL INFO Channel 02/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32704:33123 [0] NCCL INFO Channel 13/0 : 24[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227468:1227883 [5] NCCL INFO Channel 12/0 : 93[5] -> 85[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630212:1630652 [7] NCCL INFO Channel 06/0 : 87[7] -> 79[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630205:1630646 [0] NCCL INFO Channel 02/0 : 80[0] -> 72[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630208:1630650 [3] NCCL INFO Channel 15/0 : 83[3] -> 75[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:124160:124581 [0] NCCL INFO Channel 05/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96570:97005 [7] NCCL INFO Channel 14/0 : 39[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62885:63304 [4] NCCL INFO Channel 03/0 : 52[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557273:1557685 [7] NCCL INFO Channel 04/0 : 103[7] -> 87[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235649:236071 [1] NCCL INFO Channel 01/0 : 105[1] -> 97[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272152:272575 [2] NCCL INFO Channel 10/0 : 122[2] -> 114[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364902:1365333 [0] NCCL INFO Channel 04/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32705:33128 [1] NCCL INFO Channel 10/0 : 25[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630206:1630647 [1] NCCL INFO Channel 15/0 : 81[1] -> 73[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630211:1630649 [6] NCCL INFO Channel 01/0 : 86[6] -> 78[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:124162:124587 [2] NCCL INFO Channel 07/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114518:114937 [6] NCCL INFO Channel 09/0 : 78[6] -> 70[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62888:63302 [7] NCCL INFO Channel 13/0 : 47[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557272:1557687 [6] NCCL INFO Channel 00/0 : 118[6] -> 102[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235654:236070 [6] NCCL INFO Channel 02/0 : 118[6] -> 110[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235651:236068 [3] NCCL INFO Channel 01/0 : 107[3] -> 99[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272150:272576 [0] NCCL INFO Channel 04/0 : 120[0] -> 112[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272155:272574 [5] NCCL INFO Channel 09/0 : 117[5] -> 109[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272153:272573 [3] NCCL INFO Channel 02/0 : 123[3] -> 115[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364908:1365332 [6] NCCL INFO Channel 09/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32705:33128 [1] NCCL INFO Channel 12/0 : 25[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630212:1630652 [7] NCCL INFO Channel 08/0 : 87[7] -> 79[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124160:124581 [0] NCCL INFO Channel 07/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114518:114937 [6] NCCL INFO Channel 11/0 : 78[6] -> 70[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557273:1557685 [7] NCCL INFO Channel 06/0 : 103[7] -> 87[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235655:236069 [7] NCCL INFO Channel 10/0 : 119[7] -> 111[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272152:272575 [2] NCCL INFO Channel 12/0 : 122[2] -> 114[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364906:1365331 [4] NCCL INFO Channel 03/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32705:33128 [1] NCCL INFO Channel 14/0 : 25[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630205:1630646 [0] NCCL INFO Channel 03/0 : 80[0] -> 72[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630208:1630650 [3] NCCL INFO Channel 01/0 : 75[3] -> 67[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96569:97006 [6] NCCL INFO Channel 04/0 : 54[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557272:1557687 [6] NCCL INFO Channel 02/0 : 118[6] -> 102[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235649:236071 [1] NCCL INFO Channel 03/0 : 105[1] -> 97[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235648:236064 [0] NCCL INFO Channel 04/0 : 112[0] -> 104[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272155:272574 [5] NCCL INFO Channel 10/0 : 117[5] -> 109[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272150:272576 [0] NCCL INFO Channel 06/0 : 120[0] -> 112[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272153:272573 [3] NCCL INFO Channel 04/0 : 123[3] -> 115[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364902:1365333 [0] NCCL INFO Channel 05/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32711:33127 [7] NCCL INFO Channel 02/0 : 31[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227468:1227883 [5] NCCL INFO Channel 14/0 : 93[5] -> 85[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630211:1630649 [6] NCCL INFO Channel 02/0 : 86[6] -> 78[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630206:1630647 [1] NCCL INFO Channel 01/0 : 73[1] -> 65[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124160:124581 [0] NCCL INFO Channel 09/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:124162:124587 [2] NCCL INFO Channel 09/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96563:97003 [0] NCCL INFO Channel 01/0 : 40[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62887:63305 [6] NCCL INFO Channel 09/0 : 54[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557273:1557685 [7] NCCL INFO Channel 08/0 : 103[7] -> 87[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235654:236070 [6] NCCL INFO Channel 03/0 : 118[6] -> 110[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235651:236068 [3] NCCL INFO Channel 03/0 : 107[3] -> 99[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272152:272575 [2] NCCL INFO Channel 14/0 : 122[2] -> 114[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364908:1365332 [6] NCCL INFO Channel 10/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32710:33125 [6] NCCL INFO Channel 13/0 : 30[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227468:1227883 [5] NCCL INFO Channel 00/0 : 85[5] -> 77[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630212:1630652 [7] NCCL INFO Channel 09/0 : 87[7] -> 79[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96566:97010 [3] NCCL INFO Channel 03/0 : 43[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114518:114937 [6] NCCL INFO Channel 13/0 : 78[6] -> 70[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62881:63298 [0] NCCL INFO Channel 09/0 : 48[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557272:1557687 [6] NCCL INFO Channel 04/0 : 118[6] -> 102[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235655:236069 [7] NCCL INFO Channel 11/0 : 119[7] -> 111[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364906:1365331 [4] NCCL INFO Channel 05/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32711:33127 [7] NCCL INFO Channel 04/0 : 31[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630205:1630646 [0] NCCL INFO Channel 04/0 : 80[0] -> 72[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:124166:124583 [6] NCCL INFO Channel 01/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96563:97003 [0] NCCL INFO Channel 03/0 : 40[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96569:97006 [6] NCCL INFO Channel 08/0 : 54[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96566:97010 [3] NCCL INFO Channel 05/0 : 43[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153564:153984 [2] NCCL INFO Channel 00/0 : 90[2] -> 82[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62885:63304 [4] NCCL INFO Channel 05/0 : 52[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557273:1557685 [7] NCCL INFO Channel 10/0 : 103[7] -> 87[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235649:236071 [1] NCCL INFO Channel 05/0 : 105[1] -> 97[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272155:272574 [5] NCCL INFO Channel 11/0 : 117[5] -> 109[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272150:272576 [0] NCCL INFO Channel 08/0 : 120[0] -> 112[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272152:272575 [2] NCCL INFO Channel 00/0 : 114[2] -> 106[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32710:33125 [6] NCCL INFO Channel 15/0 : 30[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227468:1227883 [5] NCCL INFO Channel 01/0 : 85[5] -> 77[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630208:1630650 [3] NCCL INFO Channel 03/0 : 75[3] -> 67[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630206:1630647 [1] NCCL INFO Channel 03/0 : 73[1] -> 65[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630211:1630649 [6] NCCL INFO Channel 03/0 : 86[6] -> 78[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:124160:124581 [0] NCCL INFO Channel 11/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:124162:124587 [2] NCCL INFO Channel 11/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96563:97003 [0] NCCL INFO Channel 05/0 : 40[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62887:63305 [6] NCCL INFO Channel 10/0 : 54[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557272:1557687 [6] NCCL INFO Channel 08/0 : 118[6] -> 102[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235648:236064 [0] NCCL INFO Channel 05/0 : 112[0] -> 104[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272153:272573 [3] NCCL INFO Channel 06/0 : 123[3] -> 115[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364908:1365332 [6] NCCL INFO Channel 11/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32711:33127 [7] NCCL INFO Channel 06/0 : 31[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227468:1227883 [5] NCCL INFO Channel 02/0 : 85[5] -> 77[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630212:1630652 [7] NCCL INFO Channel 10/0 : 87[7] -> 79[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124166:124583 [6] NCCL INFO Channel 03/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114518:114937 [6] NCCL INFO Channel 15/0 : 78[6] -> 70[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62881:63298 [0] NCCL INFO Channel 10/0 : 48[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62885:63304 [4] NCCL INFO Channel 06/0 : 52[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235654:236070 [6] NCCL INFO Channel 04/0 : 118[6] -> 110[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235651:236068 [3] NCCL INFO Channel 05/0 : 107[3] -> 99[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235655:236069 [7] NCCL INFO Channel 12/0 : 119[7] -> 111[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364902:1365333 [0] NCCL INFO Channel 06/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62887:63305 [6] NCCL INFO Channel 11/0 : 54[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557266:1557689 [0] NCCL INFO Channel 01/0 : 104[0] -> 96[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557273:1557685 [7] NCCL INFO Channel 12/0 : 103[7] -> 87[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62881:63298 [0] NCCL INFO Channel 11/0 : 48[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62885:63304 [4] NCCL INFO Channel 07/0 : 52[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557269:1557684 [3] NCCL INFO Channel 01/0 : 107[3] -> 99[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62887:63305 [6] NCCL INFO Channel 12/0 : 54[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557272:1557687 [6] NCCL INFO Channel 10/0 : 118[6] -> 102[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62881:63298 [0] NCCL INFO Channel 12/0 : 48[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557266:1557689 [0] NCCL INFO Channel 03/0 : 104[0] -> 96[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557273:1557685 [7] NCCL INFO Channel 14/0 : 103[7] -> 87[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557269:1557684 [3] NCCL INFO Channel 03/0 : 107[3] -> 99[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62885:63304 [4] NCCL INFO Channel 08/0 : 52[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557272:1557687 [6] NCCL INFO Channel 12/0 : 118[6] -> 102[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62887:63305 [6] NCCL INFO Channel 13/0 : 54[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557266:1557689 [0] NCCL INFO Channel 05/0 : 104[0] -> 96[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557269:1557684 [3] NCCL INFO Channel 05/0 : 107[3] -> 99[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557272:1557687 [6] NCCL INFO Channel 14/0 : 118[6] -> 102[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235649:236071 [1] NCCL INFO Channel 07/0 : 105[1] -> 97[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235652:236066 [4] NCCL INFO Channel 00/0 : 116[4] -> 108[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235648:236064 [0] NCCL INFO Channel 06/0 : 112[0] -> 104[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557266:1557689 [0] NCCL INFO Channel 07/0 : 104[0] -> 96[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235651:236068 [3] NCCL INFO Channel 07/0 : 107[3] -> 99[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557269:1557684 [3] NCCL INFO Channel 07/0 : 107[3] -> 99[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557272:1557687 [6] NCCL INFO Channel 00/0 : 102[6] -> 86[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235655:236069 [7] NCCL INFO Channel 13/0 : 119[7] -> 111[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62881:63298 [0] NCCL INFO Channel 13/0 : 48[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557266:1557689 [0] NCCL INFO Channel 09/0 : 104[0] -> 96[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62885:63304 [4] NCCL INFO Channel 09/0 : 52[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62887:63305 [6] NCCL INFO Channel 15/0 : 54[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557272:1557687 [6] NCCL INFO Channel 02/0 : 102[6] -> 86[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557266:1557689 [0] NCCL INFO Channel 11/0 : 104[0] -> 96[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557269:1557684 [3] NCCL INFO Channel 09/0 : 107[3] -> 99[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62881:63298 [0] NCCL INFO Channel 14/0 : 48[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557273:1557685 [7] NCCL INFO Channel 01/0 : 111[7] -> 103[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62885:63304 [4] NCCL INFO Channel 10/0 : 52[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62887:63305 [6] NCCL INFO Channel 01/0 : 46[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235654:236070 [6] NCCL INFO Channel 05/0 : 118[6] -> 110[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62881:63298 [0] NCCL INFO Channel 15/0 : 48[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235649:236071 [1] NCCL INFO Channel 11/0 : 105[1] -> 97[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235648:236064 [0] NCCL INFO Channel 07/0 : 112[0] -> 104[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235652:236066 [4] NCCL INFO Channel 01/0 : 116[4] -> 108[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62887:63305 [6] NCCL INFO Channel 03/0 : 46[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62885:63304 [4] NCCL INFO Channel 11/0 : 52[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235651:236068 [3] NCCL INFO Channel 09/0 : 107[3] -> 99[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62881:63298 [0] NCCL INFO Channel 01/0 : 40[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557272:1557687 [6] NCCL INFO Channel 04/0 : 102[6] -> 86[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62887:63305 [6] NCCL INFO Channel 05/0 : 46[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557266:1557689 [0] NCCL INFO Channel 13/0 : 104[0] -> 96[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557269:1557684 [3] NCCL INFO Channel 13/0 : 107[3] -> 99[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557273:1557685 [7] NCCL INFO Channel 03/0 : 111[7] -> 103[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62885:63304 [4] NCCL INFO Channel 13/0 : 52[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62881:63298 [0] NCCL INFO Channel 03/0 : 40[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557272:1557687 [6] NCCL INFO Channel 08/0 : 102[6] -> 86[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557269:1557684 [3] NCCL INFO Channel 15/0 : 107[3] -> 99[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557273:1557685 [7] NCCL INFO Channel 05/0 : 111[7] -> 103[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235655:236069 [7] NCCL INFO Channel 14/0 : 119[7] -> 111[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235654:236070 [6] NCCL INFO Channel 07/0 : 118[6] -> 110[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557266:1557689 [0] NCCL INFO Channel 15/0 : 104[0] -> 96[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235649:236071 [1] NCCL INFO Channel 13/0 : 105[1] -> 97[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557272:1557687 [6] NCCL INFO Channel 10/0 : 102[6] -> 86[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235648:236064 [0] NCCL INFO Channel 09/0 : 112[0] -> 104[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62887:63305 [6] NCCL INFO Channel 07/0 : 46[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557268:1557686 [2] NCCL INFO Channel 01/0 : 106[2] -> 98[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557273:1557685 [7] NCCL INFO Channel 07/0 : 111[7] -> 103[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62881:63298 [0] NCCL INFO Channel 05/0 : 40[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62885:63304 [4] NCCL INFO Channel 14/0 : 52[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557272:1557687 [6] NCCL INFO Channel 12/0 : 102[6] -> 86[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62883:63301 [2] NCCL INFO Channel 00/0 : 50[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62887:63305 [6] NCCL INFO Channel 09/0 : 46[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235652:236066 [4] NCCL INFO Channel 02/0 : 116[4] -> 108[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62881:63298 [0] NCCL INFO Channel 07/0 : 40[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62885:63304 [4] NCCL INFO Channel 15/0 : 52[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235651:236068 [3] NCCL INFO Channel 13/0 : 107[3] -> 99[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62887:63305 [6] NCCL INFO Channel 11/0 : 46[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62883:63301 [2] NCCL INFO Channel 01/0 : 50[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557268:1557686 [2] NCCL INFO Channel 03/0 : 106[2] -> 98[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557273:1557685 [7] NCCL INFO Channel 09/0 : 111[7] -> 103[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62881:63298 [0] NCCL INFO Channel 09/0 : 40[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557272:1557687 [6] NCCL INFO Channel 14/0 : 102[6] -> 86[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62885:63304 [4] NCCL INFO Channel 01/0 : 44[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557268:1557686 [2] NCCL INFO Channel 05/0 : 106[2] -> 98[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557273:1557685 [7] NCCL INFO Channel 11/0 : 111[7] -> 103[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235655:236069 [7] NCCL INFO Channel 01/0 : 111[7] -> 103[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272157:272572 [7] NCCL INFO Channel 00/0 : 127[7] -> 119[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235654:236070 [6] NCCL INFO Channel 08/0 : 118[6] -> 110[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272157:272572 [7] NCCL INFO Channel 02/0 : 127[7] -> 119[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62887:63305 [6] NCCL INFO Channel 13/0 : 46[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62883:63301 [2] NCCL INFO Channel 03/0 : 50[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557268:1557686 [2] NCCL INFO Channel 07/0 : 106[2] -> 98[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62881:63298 [0] NCCL INFO Channel 11/0 : 40[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62885:63304 [4] NCCL INFO Channel 03/0 : 44[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557273:1557685 [7] NCCL INFO Channel 13/0 : 111[7] -> 103[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62887:63305 [6] NCCL INFO Channel 15/0 : 46[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557268:1557686 [2] NCCL INFO Channel 09/0 : 106[2] -> 98[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62883:63301 [2] NCCL INFO Channel 04/0 : 50[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235649:236071 [1] NCCL INFO Channel 15/0 : 105[1] -> 97[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62881:63298 [0] NCCL INFO Channel 13/0 : 40[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62885:63304 [4] NCCL INFO Channel 05/0 : 44[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235648:236064 [0] NCCL INFO Channel 10/0 : 112[0] -> 104[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235652:236066 [4] NCCL INFO Channel 03/0 : 116[4] -> 108[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272157:272572 [7] NCCL INFO Channel 04/0 : 127[7] -> 119[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235655:236069 [7] NCCL INFO Channel 03/0 : 111[7] -> 103[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235651:236068 [3] NCCL INFO Channel 15/0 : 107[3] -> 99[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272157:272572 [7] NCCL INFO Channel 06/0 : 127[7] -> 119[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235654:236070 [6] NCCL INFO Channel 09/0 : 118[6] -> 110[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272157:272572 [7] NCCL INFO Channel 08/0 : 127[7] -> 119[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62883:63301 [2] NCCL INFO Channel 05/0 : 50[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235648:236064 [0] NCCL INFO Channel 11/0 : 112[0] -> 104[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62881:63298 [0] NCCL INFO Channel 15/0 : 40[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62885:63304 [4] NCCL INFO Channel 07/0 : 44[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557272:1557687 [6] NCCL INFO Channel 01/0 : 110[6] -> 102[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62883:63301 [2] NCCL INFO Channel 06/0 : 50[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235652:236066 [4] NCCL INFO Channel 05/0 : 116[4] -> 108[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235655:236069 [7] NCCL INFO Channel 05/0 : 111[7] -> 103[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272157:272572 [7] NCCL INFO Channel 10/0 : 127[7] -> 119[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557268:1557686 [2] NCCL INFO Channel 11/0 : 106[2] -> 98[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235654:236070 [6] NCCL INFO Channel 10/0 : 118[6] -> 110[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235648:236064 [0] NCCL INFO Channel 12/0 : 112[0] -> 104[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62885:63304 [4] NCCL INFO Channel 09/0 : 44[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557272:1557687 [6] NCCL INFO Channel 03/0 : 110[6] -> 102[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62883:63301 [2] NCCL INFO Channel 07/0 : 50[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235652:236066 [4] NCCL INFO Channel 06/0 : 116[4] -> 108[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62885:63304 [4] NCCL INFO Channel 11/0 : 44[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235655:236069 [7] NCCL INFO Channel 07/0 : 111[7] -> 103[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235654:236070 [6] NCCL INFO Channel 11/0 : 118[6] -> 110[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235648:236064 [0] NCCL INFO Channel 13/0 : 112[0] -> 104[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272157:272572 [7] NCCL INFO Channel 12/0 : 127[7] -> 119[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557268:1557686 [2] NCCL INFO Channel 13/0 : 106[2] -> 98[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235652:236066 [4] NCCL INFO Channel 07/0 : 116[4] -> 108[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62883:63301 [2] NCCL INFO Channel 08/0 : 50[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557272:1557687 [6] NCCL INFO Channel 05/0 : 110[6] -> 102[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62885:63304 [4] NCCL INFO Channel 13/0 : 44[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235655:236069 [7] NCCL INFO Channel 09/0 : 111[7] -> 103[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235648:236064 [0] NCCL INFO Channel 14/0 : 112[0] -> 104[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272155:272574 [5] NCCL INFO Channel 12/0 : 117[5] -> 109[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557268:1557686 [2] NCCL INFO Channel 15/0 : 106[2] -> 98[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235654:236070 [6] NCCL INFO Channel 12/0 : 118[6] -> 110[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235652:236066 [4] NCCL INFO Channel 08/0 : 116[4] -> 108[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235655:236069 [7] NCCL INFO Channel 11/0 : 111[7] -> 103[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62883:63301 [2] NCCL INFO Channel 09/0 : 50[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557272:1557687 [6] NCCL INFO Channel 07/0 : 110[6] -> 102[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62885:63304 [4] NCCL INFO Channel 15/0 : 44[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235648:236064 [0] NCCL INFO Channel 15/0 : 112[0] -> 104[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557272:1557687 [6] NCCL INFO Channel 09/0 : 110[6] -> 102[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235654:236070 [6] NCCL INFO Channel 13/0 : 118[6] -> 110[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235652:236066 [4] NCCL INFO Channel 09/0 : 116[4] -> 108[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235655:236069 [7] NCCL INFO Channel 13/0 : 111[7] -> 103[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62883:63301 [2] NCCL INFO Channel 11/0 : 50[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235648:236064 [0] NCCL INFO Channel 01/0 : 104[0] -> 96[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62883:63301 [2] NCCL INFO Channel 12/0 : 50[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235654:236070 [6] NCCL INFO Channel 15/0 : 118[6] -> 110[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235652:236066 [4] NCCL INFO Channel 10/0 : 116[4] -> 108[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557272:1557687 [6] NCCL INFO Channel 11/0 : 110[6] -> 102[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235648:236064 [0] NCCL INFO Channel 03/0 : 104[0] -> 96[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557272:1557687 [6] NCCL INFO Channel 13/0 : 110[6] -> 102[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235654:236070 [6] NCCL INFO Channel 01/0 : 110[6] -> 102[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235652:236066 [4] NCCL INFO Channel 11/0 : 116[4] -> 108[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235648:236064 [0] NCCL INFO Channel 05/0 : 104[0] -> 96[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272157:272572 [7] NCCL INFO Channel 14/0 : 127[7] -> 119[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62883:63301 [2] NCCL INFO Channel 13/0 : 50[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272155:272574 [5] NCCL INFO Channel 14/0 : 117[5] -> 109[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557272:1557687 [6] NCCL INFO Channel 15/0 : 110[6] -> 102[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272157:272572 [7] NCCL INFO Channel 00/0 : 119[7] -> 111[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235654:236070 [6] NCCL INFO Channel 03/0 : 110[6] -> 102[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272155:272574 [5] NCCL INFO Channel 15/0 : 117[5] -> 109[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62883:63301 [2] NCCL INFO Channel 14/0 : 50[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235652:236066 [4] NCCL INFO Channel 13/0 : 116[4] -> 108[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235648:236064 [0] NCCL INFO Channel 07/0 : 104[0] -> 96[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235650:236065 [2] NCCL INFO Channel 00/0 : 114[2] -> 106[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272156:272570 [6] NCCL INFO Channel 00/0 : 126[6] -> 118[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235654:236070 [6] NCCL INFO Channel 05/0 : 110[6] -> 102[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96893:97310 [5] NCCL INFO Channel 00/0 : 29[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96892:97308 [4] NCCL INFO Channel 11/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235648:236064 [0] NCCL INFO Channel 09/0 : 104[0] -> 96[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235652:236066 [4] NCCL INFO Channel 14/0 : 116[4] -> 108[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96890:97313 [2] NCCL INFO Channel 10/0 : 26[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96889:97309 [1] NCCL INFO Channel 12/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62883:63301 [2] NCCL INFO Channel 15/0 : 50[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235650:236065 [2] NCCL INFO Channel 01/0 : 114[2] -> 106[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235654:236070 [6] NCCL INFO Channel 07/0 : 110[6] -> 102[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272157:272572 [7] NCCL INFO Channel 01/0 : 119[7] -> 111[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235652:236066 [4] NCCL INFO Channel 15/0 : 116[4] -> 108[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235648:236064 [0] NCCL INFO Channel 11/0 : 104[0] -> 96[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272156:272570 [6] NCCL INFO Channel 02/0 : 126[6] -> 118[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235654:236070 [6] NCCL INFO Channel 09/0 : 110[6] -> 102[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272157:272572 [7] NCCL INFO Channel 02/0 : 119[7] -> 111[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62883:63301 [2] NCCL INFO Channel 01/0 : 42[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235650:236065 [2] NCCL INFO Channel 03/0 : 114[2] -> 106[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235648:236064 [0] NCCL INFO Channel 13/0 : 104[0] -> 96[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272157:272572 [7] NCCL INFO Channel 03/0 : 119[7] -> 111[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272156:272570 [6] NCCL INFO Channel 04/0 : 126[6] -> 118[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235652:236066 [4] NCCL INFO Channel 01/0 : 108[4] -> 100[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272157:272572 [7] NCCL INFO Channel 04/0 : 119[7] -> 111[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235654:236070 [6] NCCL INFO Channel 11/0 : 110[6] -> 102[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272152:272575 [2] NCCL INFO Channel 01/0 : 114[2] -> 106[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272156:272570 [6] NCCL INFO Channel 08/0 : 126[6] -> 118[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272153:272573 [3] NCCL INFO Channel 08/0 : 123[3] -> 115[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62883:63301 [2] NCCL INFO Channel 03/0 : 42[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235650:236065 [2] NCCL INFO Channel 04/0 : 114[2] -> 106[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235648:236064 [0] NCCL INFO Channel 15/0 : 104[0] -> 96[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272157:272572 [7] NCCL INFO Channel 05/0 : 119[7] -> 111[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235652:236066 [4] NCCL INFO Channel 03/0 : 108[4] -> 100[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272156:272570 [6] NCCL INFO Channel 10/0 : 126[6] -> 118[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272153:272573 [3] NCCL INFO Channel 10/0 : 123[3] -> 115[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272152:272575 [2] NCCL INFO Channel 03/0 : 114[2] -> 106[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62883:63301 [2] NCCL INFO Channel 05/0 : 42[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235654:236070 [6] NCCL INFO Channel 13/0 : 110[6] -> 102[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235650:236065 [2] NCCL INFO Channel 05/0 : 114[2] -> 106[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272157:272572 [7] NCCL INFO Channel 06/0 : 119[7] -> 111[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235652:236066 [4] NCCL INFO Channel 05/0 : 108[4] -> 100[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235654:236070 [6] NCCL INFO Channel 15/0 : 110[6] -> 102[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272153:272573 [3] NCCL INFO Channel 12/0 : 123[3] -> 115[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235650:236065 [2] NCCL INFO Channel 06/0 : 114[2] -> 106[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272152:272575 [2] NCCL INFO Channel 04/0 : 114[2] -> 106[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272156:272570 [6] NCCL INFO Channel 12/0 : 126[6] -> 118[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62883:63301 [2] NCCL INFO Channel 07/0 : 42[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235652:236066 [4] NCCL INFO Channel 07/0 : 108[4] -> 100[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235650:236065 [2] NCCL INFO Channel 07/0 : 114[2] -> 106[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272157:272572 [7] NCCL INFO Channel 08/0 : 119[7] -> 111[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235652:236066 [4] NCCL INFO Channel 09/0 : 108[4] -> 100[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272150:272576 [0] NCCL INFO Channel 10/0 : 120[0] -> 112[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62883:63301 [2] NCCL INFO Channel 09/0 : 42[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235650:236065 [2] NCCL INFO Channel 08/0 : 114[2] -> 106[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235652:236066 [4] NCCL INFO Channel 11/0 : 108[4] -> 100[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272153:272573 [3] NCCL INFO Channel 14/0 : 123[3] -> 115[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235650:236065 [2] NCCL INFO Channel 09/0 : 114[2] -> 106[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272152:272575 [2] NCCL INFO Channel 05/0 : 114[2] -> 106[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235652:236066 [4] NCCL INFO Channel 13/0 : 108[4] -> 100[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272157:272572 [7] NCCL INFO Channel 09/0 : 119[7] -> 111[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272156:272570 [6] NCCL INFO Channel 14/0 : 126[6] -> 118[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62883:63301 [2] NCCL INFO Channel 11/0 : 42[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235650:236065 [2] NCCL INFO Channel 11/0 : 114[2] -> 106[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235652:236066 [4] NCCL INFO Channel 15/0 : 108[4] -> 100[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272153:272573 [3] NCCL INFO Channel 00/0 : 115[3] -> 107[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272150:272576 [0] NCCL INFO Channel 12/0 : 120[0] -> 112[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235650:236065 [2] NCCL INFO Channel 12/0 : 114[2] -> 106[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272152:272575 [2] NCCL INFO Channel 06/0 : 114[2] -> 106[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62883:63301 [2] NCCL INFO Channel 13/0 : 42[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235650:236065 [2] NCCL INFO Channel 13/0 : 114[2] -> 106[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235650:236065 [2] NCCL INFO Channel 14/0 : 114[2] -> 106[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96892:97308 [4] NCCL INFO Channel 13/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235650:236065 [2] NCCL INFO Channel 15/0 : 114[2] -> 106[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272157:272572 [7] NCCL INFO Channel 10/0 : 119[7] -> 111[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272156:272570 [6] NCCL INFO Channel 00/0 : 118[6] -> 110[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235650:236065 [2] NCCL INFO Channel 01/0 : 106[2] -> 98[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272153:272573 [3] NCCL INFO Channel 01/0 : 115[3] -> 107[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272150:272576 [0] NCCL INFO Channel 14/0 : 120[0] -> 112[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62883:63301 [2] NCCL INFO Channel 15/0 : 42[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235650:236065 [2] NCCL INFO Channel 03/0 : 106[2] -> 98[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235650:236065 [2] NCCL INFO Channel 05/0 : 106[2] -> 98[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272152:272575 [2] NCCL INFO Channel 07/0 : 114[2] -> 106[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235650:236065 [2] NCCL INFO Channel 07/0 : 106[2] -> 98[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272157:272572 [7] NCCL INFO Channel 11/0 : 119[7] -> 111[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272156:272570 [6] NCCL INFO Channel 01/0 : 118[6] -> 110[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235650:236065 [2] NCCL INFO Channel 09/0 : 106[2] -> 98[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272150:272576 [0] NCCL INFO Channel 01/0 : 112[0] -> 104[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272152:272575 [2] NCCL INFO Channel 08/0 : 114[2] -> 106[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272153:272573 [3] NCCL INFO Channel 02/0 : 115[3] -> 107[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235650:236065 [2] NCCL INFO Channel 11/0 : 106[2] -> 98[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272157:272572 [7] NCCL INFO Channel 12/0 : 119[7] -> 111[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272156:272570 [6] NCCL INFO Channel 02/0 : 118[6] -> 110[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235650:236065 [2] NCCL INFO Channel 13/0 : 106[2] -> 98[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272153:272573 [3] NCCL INFO Channel 04/0 : 115[3] -> 107[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272150:272576 [0] NCCL INFO Channel 02/0 : 112[0] -> 104[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235650:236065 [2] NCCL INFO Channel 15/0 : 106[2] -> 98[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272152:272575 [2] NCCL INFO Channel 09/0 : 114[2] -> 106[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272157:272572 [7] NCCL INFO Channel 13/0 : 119[7] -> 111[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272156:272570 [6] NCCL INFO Channel 03/0 : 118[6] -> 110[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96893:97310 [5] NCCL INFO Channel 02/0 : 29[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272150:272576 [0] NCCL INFO Channel 03/0 : 112[0] -> 104[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272153:272573 [3] NCCL INFO Channel 05/0 : 115[3] -> 107[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272152:272575 [2] NCCL INFO Channel 11/0 : 114[2] -> 106[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96889:97309 [1] NCCL INFO Channel 13/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96890:97313 [2] NCCL INFO Channel 12/0 : 26[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272157:272572 [7] NCCL INFO Channel 14/0 : 119[7] -> 111[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272156:272570 [6] NCCL INFO Channel 04/0 : 118[6] -> 110[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96892:97308 [4] NCCL INFO Channel 14/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272153:272573 [3] NCCL INFO Channel 06/0 : 115[3] -> 107[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272152:272575 [2] NCCL INFO Channel 12/0 : 114[2] -> 106[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96893:97310 [5] NCCL INFO Channel 04/0 : 29[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272150:272576 [0] NCCL INFO Channel 04/0 : 112[0] -> 104[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272156:272570 [6] NCCL INFO Channel 05/0 : 118[6] -> 110[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96889:97309 [1] NCCL INFO Channel 14/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96890:97313 [2] NCCL INFO Channel 14/0 : 26[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272153:272573 [3] NCCL INFO Channel 07/0 : 115[3] -> 107[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272152:272575 [2] NCCL INFO Channel 13/0 : 114[2] -> 106[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96892:97308 [4] NCCL INFO Channel 15/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96893:97310 [5] NCCL INFO Channel 06/0 : 29[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364906:1365331 [4] NCCL INFO Channel 06/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96889:97309 [1] NCCL INFO Channel 15/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96890:97313 [2] NCCL INFO Channel 00/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364908:1365332 [6] NCCL INFO Channel 12/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272156:272570 [6] NCCL INFO Channel 07/0 : 118[6] -> 110[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272150:272576 [0] NCCL INFO Channel 05/0 : 112[0] -> 104[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96893:97310 [5] NCCL INFO Channel 08/0 : 29[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272153:272573 [3] NCCL INFO Channel 08/0 : 115[3] -> 107[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272152:272575 [2] NCCL INFO Channel 14/0 : 114[2] -> 106[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96890:97313 [2] NCCL INFO Channel 01/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272150:272576 [0] NCCL INFO Channel 06/0 : 112[0] -> 104[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272156:272570 [6] NCCL INFO Channel 08/0 : 118[6] -> 110[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96893:97310 [5] NCCL INFO Channel 10/0 : 29[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272153:272573 [3] NCCL INFO Channel 09/0 : 115[3] -> 107[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272152:272575 [2] NCCL INFO Channel 15/0 : 114[2] -> 106[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96890:97313 [2] NCCL INFO Channel 03/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272150:272576 [0] NCCL INFO Channel 07/0 : 112[0] -> 104[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272156:272570 [6] NCCL INFO Channel 09/0 : 118[6] -> 110[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96893:97310 [5] NCCL INFO Channel 12/0 : 29[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272153:272573 [3] NCCL INFO Channel 10/0 : 115[3] -> 107[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96890:97313 [2] NCCL INFO Channel 04/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272150:272576 [0] NCCL INFO Channel 09/0 : 112[0] -> 104[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96893:97310 [5] NCCL INFO Channel 14/0 : 29[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272153:272573 [3] NCCL INFO Channel 12/0 : 115[3] -> 107[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272156:272570 [6] NCCL INFO Channel 10/0 : 118[6] -> 110[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96890:97313 [2] NCCL INFO Channel 05/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96893:97310 [5] NCCL INFO Channel 00/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364902:1365333 [0] NCCL INFO Channel 07/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96890:97313 [2] NCCL INFO Channel 06/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364906:1365331 [4] NCCL INFO Channel 07/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272150:272576 [0] NCCL INFO Channel 10/0 : 112[0] -> 104[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96893:97310 [5] NCCL INFO Channel 01/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272156:272570 [6] NCCL INFO Channel 11/0 : 118[6] -> 110[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272153:272573 [3] NCCL INFO Channel 13/0 : 115[3] -> 107[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96890:97313 [2] NCCL INFO Channel 07/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272150:272576 [0] NCCL INFO Channel 11/0 : 112[0] -> 104[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96893:97310 [5] NCCL INFO Channel 02/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96890:97313 [2] NCCL INFO Channel 08/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364908:1365332 [6] NCCL INFO Channel 13/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96893:97310 [5] NCCL INFO Channel 03/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364902:1365333 [0] NCCL INFO Channel 09/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96890:97313 [2] NCCL INFO Channel 09/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364906:1365331 [4] NCCL INFO Channel 08/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272156:272570 [6] NCCL INFO Channel 12/0 : 118[6] -> 110[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272153:272573 [3] NCCL INFO Channel 14/0 : 115[3] -> 107[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272150:272576 [0] NCCL INFO Channel 12/0 : 112[0] -> 104[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96893:97310 [5] NCCL INFO Channel 04/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272153:272573 [3] NCCL INFO Channel 15/0 : 115[3] -> 107[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272156:272570 [6] NCCL INFO Channel 13/0 : 118[6] -> 110[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96890:97313 [2] NCCL INFO Channel 11/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272150:272576 [0] NCCL INFO Channel 13/0 : 112[0] -> 104[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96893:97310 [5] NCCL INFO Channel 06/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96890:97313 [2] NCCL INFO Channel 12/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364908:1365332 [6] NCCL INFO Channel 15/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96893:97310 [5] NCCL INFO Channel 07/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32708:33126 [4] NCCL INFO Channel 05/0 : 28[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272156:272570 [6] NCCL INFO Channel 15/0 : 118[6] -> 110[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272150:272576 [0] NCCL INFO Channel 14/0 : 112[0] -> 104[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96890:97313 [2] NCCL INFO Channel 13/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96893:97310 [5] NCCL INFO Channel 08/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364902:1365333 [0] NCCL INFO Channel 10/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96890:97313 [2] NCCL INFO Channel 14/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364906:1365331 [4] NCCL INFO Channel 09/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272150:272576 [0] NCCL INFO Channel 15/0 : 112[0] -> 104[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96893:97310 [5] NCCL INFO Channel 09/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96890:97313 [2] NCCL INFO Channel 15/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364908:1365332 [6] NCCL INFO Channel 01/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96893:97310 [5] NCCL INFO Channel 10/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364902:1365333 [0] NCCL INFO Channel 11/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96893:97310 [5] NCCL INFO Channel 11/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364906:1365331 [4] NCCL INFO Channel 10/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364908:1365332 [6] NCCL INFO Channel 03/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96893:97310 [5] NCCL INFO Channel 12/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364902:1365333 [0] NCCL INFO Channel 12/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96893:97310 [5] NCCL INFO Channel 14/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364908:1365332 [6] NCCL INFO Channel 05/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96891:97311 [3] NCCL INFO Channel 00/0 : 27[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364906:1365331 [4] NCCL INFO Channel 11/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96888:97306 [0] NCCL INFO Channel 02/0 : 24[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96893:97310 [5] NCCL INFO Channel 15/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364902:1365333 [0] NCCL INFO Channel 13/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96891:97311 [3] NCCL INFO Channel 02/0 : 27[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364908:1365332 [6] NCCL INFO Channel 07/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96888:97306 [0] NCCL INFO Channel 04/0 : 24[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364906:1365331 [4] NCCL INFO Channel 13/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96888:97306 [0] NCCL INFO Channel 06/0 : 24[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96891:97311 [3] NCCL INFO Channel 04/0 : 27[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364908:1365332 [6] NCCL INFO Channel 09/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96888:97306 [0] NCCL INFO Channel 08/0 : 24[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364902:1365333 [0] NCCL INFO Channel 14/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96891:97311 [3] NCCL INFO Channel 06/0 : 27[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364906:1365331 [4] NCCL INFO Channel 14/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96888:97306 [0] NCCL INFO Channel 10/0 : 24[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96895:97307 [7] NCCL INFO Channel 00/0 : 31[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364908:1365332 [6] NCCL INFO Channel 11/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96891:97311 [3] NCCL INFO Channel 08/0 : 27[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364904:1365335 [2] NCCL INFO Channel 00/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96888:97306 [0] NCCL INFO Channel 12/0 : 24[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364906:1365331 [4] NCCL INFO Channel 15/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364902:1365333 [0] NCCL INFO Channel 15/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96891:97311 [3] NCCL INFO Channel 10/0 : 27[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96895:97307 [7] NCCL INFO Channel 02/0 : 31[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364908:1365332 [6] NCCL INFO Channel 13/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96888:97306 [0] NCCL INFO Channel 14/0 : 24[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364904:1365335 [2] NCCL INFO Channel 01/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96891:97311 [3] NCCL INFO Channel 12/0 : 27[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96895:97307 [7] NCCL INFO Channel 04/0 : 31[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364904:1365335 [2] NCCL INFO Channel 03/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96888:97306 [0] NCCL INFO Channel 01/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364908:1365332 [6] NCCL INFO Channel 15/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96891:97311 [3] NCCL INFO Channel 14/0 : 27[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364904:1365335 [2] NCCL INFO Channel 04/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96895:97307 [7] NCCL INFO Channel 06/0 : 31[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364904:1365335 [2] NCCL INFO Channel 05/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96888:97306 [0] NCCL INFO Channel 02/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364904:1365335 [2] NCCL INFO Channel 06/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96891:97311 [3] NCCL INFO Channel 00/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364904:1365335 [2] NCCL INFO Channel 07/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96895:97307 [7] NCCL INFO Channel 08/0 : 31[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364906:1365331 [4] NCCL INFO Channel 01/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96888:97306 [0] NCCL INFO Channel 03/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364904:1365335 [2] NCCL INFO Channel 08/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96891:97311 [3] NCCL INFO Channel 01/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364902:1365333 [0] NCCL INFO Channel 01/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96895:97307 [7] NCCL INFO Channel 10/0 : 31[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364906:1365331 [4] NCCL INFO Channel 03/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96888:97306 [0] NCCL INFO Channel 04/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364904:1365335 [2] NCCL INFO Channel 09/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96891:97311 [3] NCCL INFO Channel 02/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364902:1365333 [0] NCCL INFO Channel 03/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96888:97306 [0] NCCL INFO Channel 05/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96895:97307 [7] NCCL INFO Channel 12/0 : 31[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364906:1365331 [4] NCCL INFO Channel 05/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96891:97311 [3] NCCL INFO Channel 04/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364904:1365335 [2] NCCL INFO Channel 11/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364902:1365333 [0] NCCL INFO Channel 05/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32711:33127 [7] NCCL INFO Channel 08/0 : 31[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364906:1365331 [4] NCCL INFO Channel 07/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32708:33126 [4] NCCL INFO Channel 07/0 : 28[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364904:1365335 [2] NCCL INFO Channel 12/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364902:1365333 [0] NCCL INFO Channel 07/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32711:33127 [7] NCCL INFO Channel 10/0 : 31[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96888:97306 [0] NCCL INFO Channel 06/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364906:1365331 [4] NCCL INFO Channel 09/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96895:97307 [7] NCCL INFO Channel 14/0 : 31[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364902:1365333 [0] NCCL INFO Channel 09/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96891:97311 [3] NCCL INFO Channel 05/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364904:1365335 [2] NCCL INFO Channel 13/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96888:97306 [0] NCCL INFO Channel 07/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364906:1365331 [4] NCCL INFO Channel 11/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96895:97307 [7] NCCL INFO Channel 00/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96891:97311 [3] NCCL INFO Channel 06/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364902:1365333 [0] NCCL INFO Channel 11/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364904:1365335 [2] NCCL INFO Channel 14/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96888:97306 [0] NCCL INFO Channel 09/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364906:1365331 [4] NCCL INFO Channel 13/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96891:97311 [3] NCCL INFO Channel 07/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96895:97307 [7] NCCL INFO Channel 01/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364904:1365335 [2] NCCL INFO Channel 15/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96888:97306 [0] NCCL INFO Channel 10/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364902:1365333 [0] NCCL INFO Channel 13/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96895:97307 [7] NCCL INFO Channel 02/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96891:97311 [3] NCCL INFO Channel 08/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96888:97306 [0] NCCL INFO Channel 11/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364906:1365331 [4] NCCL INFO Channel 15/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96894:97312 [6] NCCL INFO Channel 00/0 : 30[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364904:1365335 [2] NCCL INFO Channel 01/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364902:1365333 [0] NCCL INFO Channel 15/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32708:33126 [4] NCCL INFO Channel 09/0 : 28[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364904:1365335 [2] NCCL INFO Channel 03/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32704:33123 [0] NCCL INFO Channel 15/0 : 24[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364904:1365335 [2] NCCL INFO Channel 05/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32711:33127 [7] NCCL INFO Channel 12/0 : 31[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364904:1365335 [2] NCCL INFO Channel 07/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32708:33126 [4] NCCL INFO Channel 11/0 : 28[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364904:1365335 [2] NCCL INFO Channel 09/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32711:33127 [7] NCCL INFO Channel 14/0 : 31[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364904:1365335 [2] NCCL INFO Channel 11/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32708:33126 [4] NCCL INFO Channel 13/0 : 28[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364904:1365335 [2] NCCL INFO Channel 13/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32707:33124 [3] NCCL INFO Channel 10/0 : 27[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364904:1365335 [2] NCCL INFO Channel 15/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32710:33125 [6] NCCL INFO Channel 00/0 : 30[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32708:33126 [4] NCCL INFO Channel 15/0 : 28[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227468:1227883 [5] NCCL INFO Channel 03/0 : 85[5] -> 77[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32707:33124 [3] NCCL INFO Channel 12/0 : 27[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32710:33125 [6] NCCL INFO Channel 02/0 : 30[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227468:1227883 [5] NCCL INFO Channel 04/0 : 85[5] -> 77[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32706:33122 [2] NCCL INFO Channel 01/0 : 42[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227468:1227883 [5] NCCL INFO Channel 06/0 : 85[5] -> 77[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32707:33124 [3] NCCL INFO Channel 14/0 : 27[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227468:1227883 [5] NCCL INFO Channel 07/0 : 85[5] -> 77[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32710:33125 [6] NCCL INFO Channel 04/0 : 30[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227463:1227880 [0] NCCL INFO Channel 02/0 : 88[0] -> 80[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32706:33122 [2] NCCL INFO Channel 03/0 : 42[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227470:1227885 [7] NCCL INFO Channel 00/0 : 95[7] -> 87[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96888:97306 [0] NCCL INFO Channel 12/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32704:33123 [0] NCCL INFO Channel 02/0 : 24[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32710:33125 [6] NCCL INFO Channel 08/0 : 30[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227468:1227883 [5] NCCL INFO Channel 08/0 : 85[5] -> 77[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32706:33122 [2] NCCL INFO Channel 05/0 : 42[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227463:1227880 [0] NCCL INFO Channel 04/0 : 88[0] -> 80[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32704:33123 [0] NCCL INFO Channel 04/0 : 24[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227470:1227885 [7] NCCL INFO Channel 02/0 : 95[7] -> 87[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32710:33125 [6] NCCL INFO Channel 10/0 : 30[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227468:1227883 [5] NCCL INFO Channel 09/0 : 85[5] -> 77[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96888:97306 [0] NCCL INFO Channel 13/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32706:33122 [2] NCCL INFO Channel 07/0 : 42[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32704:33123 [0] NCCL INFO Channel 06/0 : 24[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227463:1227880 [0] NCCL INFO Channel 06/0 : 88[0] -> 80[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32710:33125 [6] NCCL INFO Channel 12/0 : 30[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227470:1227885 [7] NCCL INFO Channel 04/0 : 95[7] -> 87[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227468:1227883 [5] NCCL INFO Channel 10/0 : 85[5] -> 77[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96894:97312 [6] NCCL INFO Channel 02/0 : 30[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32706:33122 [2] NCCL INFO Channel 09/0 : 42[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96888:97306 [0] NCCL INFO Channel 14/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32708:33126 [4] NCCL INFO Channel 00/0 : 28[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32704:33123 [0] NCCL INFO Channel 08/0 : 24[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32710:33125 [6] NCCL INFO Channel 14/0 : 30[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227463:1227880 [0] NCCL INFO Channel 08/0 : 88[0] -> 80[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32706:33122 [2] NCCL INFO Channel 11/0 : 42[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227468:1227883 [5] NCCL INFO Channel 11/0 : 85[5] -> 77[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227470:1227885 [7] NCCL INFO Channel 06/0 : 95[7] -> 87[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32708:33126 [4] NCCL INFO Channel 02/0 : 28[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32704:33123 [0] NCCL INFO Channel 10/0 : 24[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227463:1227880 [0] NCCL INFO Channel 10/0 : 88[0] -> 80[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96894:97312 [6] NCCL INFO Channel 04/0 : 30[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32706:33122 [2] NCCL INFO Channel 13/0 : 42[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32704:33123 [0] NCCL INFO Channel 12/0 : 24[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227468:1227883 [5] NCCL INFO Channel 12/0 : 85[5] -> 77[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227470:1227885 [7] NCCL INFO Channel 08/0 : 95[7] -> 87[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32708:33126 [4] NCCL INFO Channel 06/0 : 28[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227463:1227880 [0] NCCL INFO Channel 12/0 : 88[0] -> 80[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32706:33122 [2] NCCL INFO Channel 15/0 : 42[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227468:1227883 [5] NCCL INFO Channel 14/0 : 85[5] -> 77[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96888:97306 [0] NCCL INFO Channel 15/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32704:33123 [0] NCCL INFO Channel 14/0 : 24[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96894:97312 [6] NCCL INFO Channel 08/0 : 30[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32708:33126 [4] NCCL INFO Channel 08/0 : 28[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32706:33122 [2] NCCL INFO Channel 01/0 : 26[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227470:1227885 [7] NCCL INFO Channel 10/0 : 95[7] -> 87[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32708:33126 [4] NCCL INFO Channel 10/0 : 28[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227463:1227880 [0] NCCL INFO Channel 14/0 : 88[0] -> 80[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32706:33122 [2] NCCL INFO Channel 03/0 : 26[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227468:1227883 [5] NCCL INFO Channel 15/0 : 85[5] -> 77[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227469:1227882 [6] NCCL INFO Channel 00/0 : 94[6] -> 86[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96894:97312 [6] NCCL INFO Channel 10/0 : 30[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32708:33126 [4] NCCL INFO Channel 12/0 : 28[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32706:33122 [2] NCCL INFO Channel 05/0 : 26[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227470:1227885 [7] NCCL INFO Channel 12/0 : 95[7] -> 87[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32708:33126 [4] NCCL INFO Channel 14/0 : 28[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32706:33122 [2] NCCL INFO Channel 07/0 : 26[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227463:1227880 [0] NCCL INFO Channel 01/0 : 80[0] -> 72[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96894:97312 [6] NCCL INFO Channel 12/0 : 30[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32706:33122 [2] NCCL INFO Channel 09/0 : 26[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32706:33122 [2] NCCL INFO Channel 11/0 : 26[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227469:1227882 [6] NCCL INFO Channel 02/0 : 94[6] -> 86[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227470:1227885 [7] NCCL INFO Channel 14/0 : 95[7] -> 87[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32706:33122 [2] NCCL INFO Channel 13/0 : 26[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227463:1227880 [0] NCCL INFO Channel 02/0 : 80[0] -> 72[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32706:33122 [2] NCCL INFO Channel 15/0 : 26[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227469:1227882 [6] NCCL INFO Channel 04/0 : 94[6] -> 86[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227470:1227885 [7] NCCL INFO Channel 00/0 : 87[7] -> 79[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96894:97312 [6] NCCL INFO Channel 14/0 : 30[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32706:33122 [2] NCCL INFO Channel 00/0 : 26[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32706:33122 [2] NCCL INFO Channel 04/0 : 26[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227463:1227880 [0] NCCL INFO Channel 03/0 : 80[0] -> 72[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32706:33122 [2] NCCL INFO Channel 06/0 : 26[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227466:1227887 [3] NCCL INFO Channel 00/0 : 91[3] -> 83[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32706:33122 [2] NCCL INFO Channel 08/0 : 26[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227470:1227885 [7] NCCL INFO Channel 01/0 : 87[7] -> 79[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32706:33122 [2] NCCL INFO Channel 10/0 : 26[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227463:1227880 [0] NCCL INFO Channel 04/0 : 80[0] -> 72[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96894:97312 [6] NCCL INFO Channel 00/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32706:33122 [2] NCCL INFO Channel 12/0 : 26[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32706:33122 [2] NCCL INFO Channel 14/0 : 26[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227469:1227882 [6] NCCL INFO Channel 08/0 : 94[6] -> 86[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227466:1227887 [3] NCCL INFO Channel 02/0 : 91[3] -> 83[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227470:1227885 [7] NCCL INFO Channel 02/0 : 87[7] -> 79[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630205:1630646 [0] NCCL INFO Channel 05/0 : 80[0] -> 72[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96894:97312 [6] NCCL INFO Channel 01/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227463:1227880 [0] NCCL INFO Channel 05/0 : 80[0] -> 72[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227466:1227887 [3] NCCL INFO Channel 04/0 : 91[3] -> 83[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227469:1227882 [6] NCCL INFO Channel 10/0 : 94[6] -> 86[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630208:1630650 [3] NCCL INFO Channel 05/0 : 75[3] -> 67[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630211:1630649 [6] NCCL INFO Channel 04/0 : 86[6] -> 78[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227465:1227881 [2] NCCL INFO Channel 00/0 : 90[2] -> 82[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630206:1630647 [1] NCCL INFO Channel 05/0 : 73[1] -> 65[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630212:1630652 [7] NCCL INFO Channel 11/0 : 87[7] -> 79[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227470:1227885 [7] NCCL INFO Channel 03/0 : 87[7] -> 79[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630205:1630646 [0] NCCL INFO Channel 06/0 : 80[0] -> 72[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630209:1630651 [4] NCCL INFO Channel 00/0 : 84[4] -> 76[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96894:97312 [6] NCCL INFO Channel 02/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227463:1227880 [0] NCCL INFO Channel 06/0 : 80[0] -> 72[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227466:1227887 [3] NCCL INFO Channel 06/0 : 91[3] -> 83[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630208:1630650 [3] NCCL INFO Channel 07/0 : 75[3] -> 67[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630211:1630649 [6] NCCL INFO Channel 05/0 : 86[6] -> 78[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227469:1227882 [6] NCCL INFO Channel 12/0 : 94[6] -> 86[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630206:1630647 [1] NCCL INFO Channel 07/0 : 73[1] -> 65[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630212:1630652 [7] NCCL INFO Channel 12/0 : 87[7] -> 79[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227465:1227881 [2] NCCL INFO Channel 04/0 : 90[2] -> 82[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630205:1630646 [0] NCCL INFO Channel 07/0 : 80[0] -> 72[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630209:1630651 [4] NCCL INFO Channel 01/0 : 84[4] -> 76[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96894:97312 [6] NCCL INFO Channel 03/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227470:1227885 [7] NCCL INFO Channel 04/0 : 87[7] -> 79[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227463:1227880 [0] NCCL INFO Channel 07/0 : 80[0] -> 72[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227466:1227887 [3] NCCL INFO Channel 08/0 : 91[3] -> 83[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227469:1227882 [6] NCCL INFO Channel 14/0 : 94[6] -> 86[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630211:1630649 [6] NCCL INFO Channel 07/0 : 86[6] -> 78[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630208:1630650 [3] NCCL INFO Channel 09/0 : 75[3] -> 67[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227465:1227881 [2] NCCL INFO Channel 06/0 : 90[2] -> 82[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630212:1630652 [7] NCCL INFO Channel 13/0 : 87[7] -> 79[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630206:1630647 [1] NCCL INFO Channel 11/0 : 73[1] -> 65[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227470:1227885 [7] NCCL INFO Channel 05/0 : 87[7] -> 79[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630205:1630646 [0] NCCL INFO Channel 09/0 : 80[0] -> 72[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227463:1227880 [0] NCCL INFO Channel 09/0 : 80[0] -> 72[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630209:1630651 [4] NCCL INFO Channel 02/0 : 84[4] -> 76[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630211:1630649 [6] NCCL INFO Channel 08/0 : 86[6] -> 78[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630208:1630650 [3] NCCL INFO Channel 13/0 : 75[3] -> 67[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96894:97312 [6] NCCL INFO Channel 04/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227469:1227882 [6] NCCL INFO Channel 00/0 : 86[6] -> 78[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227466:1227887 [3] NCCL INFO Channel 10/0 : 91[3] -> 83[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227465:1227881 [2] NCCL INFO Channel 08/0 : 90[2] -> 82[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630212:1630652 [7] NCCL INFO Channel 14/0 : 87[7] -> 79[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630206:1630647 [1] NCCL INFO Channel 13/0 : 73[1] -> 65[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227470:1227885 [7] NCCL INFO Channel 06/0 : 87[7] -> 79[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630205:1630646 [0] NCCL INFO Channel 10/0 : 80[0] -> 72[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630211:1630649 [6] NCCL INFO Channel 09/0 : 86[6] -> 78[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227463:1227880 [0] NCCL INFO Channel 10/0 : 80[0] -> 72[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630209:1630651 [4] NCCL INFO Channel 03/0 : 84[4] -> 76[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630208:1630650 [3] NCCL INFO Channel 15/0 : 75[3] -> 67[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227469:1227882 [6] NCCL INFO Channel 01/0 : 86[6] -> 78[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630212:1630652 [7] NCCL INFO Channel 01/0 : 79[7] -> 71[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630206:1630647 [1] NCCL INFO Channel 15/0 : 73[1] -> 65[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96894:97312 [6] NCCL INFO Channel 05/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227466:1227887 [3] NCCL INFO Channel 12/0 : 91[3] -> 83[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227470:1227885 [7] NCCL INFO Channel 08/0 : 87[7] -> 79[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227465:1227881 [2] NCCL INFO Channel 10/0 : 90[2] -> 82[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630211:1630649 [6] NCCL INFO Channel 10/0 : 86[6] -> 78[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630205:1630646 [0] NCCL INFO Channel 11/0 : 80[0] -> 72[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227463:1227880 [0] NCCL INFO Channel 11/0 : 80[0] -> 72[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630209:1630651 [4] NCCL INFO Channel 05/0 : 84[4] -> 76[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630212:1630652 [7] NCCL INFO Channel 03/0 : 79[7] -> 71[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227469:1227882 [6] NCCL INFO Channel 02/0 : 86[6] -> 78[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630211:1630649 [6] NCCL INFO Channel 11/0 : 86[6] -> 78[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96891:97311 [3] NCCL INFO Channel 09/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227466:1227887 [3] NCCL INFO Channel 14/0 : 91[3] -> 83[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227470:1227885 [7] NCCL INFO Channel 09/0 : 87[7] -> 79[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630209:1630651 [4] NCCL INFO Channel 06/0 : 84[4] -> 76[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630212:1630652 [7] NCCL INFO Channel 05/0 : 79[7] -> 71[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630205:1630646 [0] NCCL INFO Channel 12/0 : 80[0] -> 72[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227463:1227880 [0] NCCL INFO Channel 12/0 : 80[0] -> 72[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630211:1630649 [6] NCCL INFO Channel 12/0 : 86[6] -> 78[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96894:97312 [6] NCCL INFO Channel 07/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227465:1227881 [2] NCCL INFO Channel 12/0 : 90[2] -> 82[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227469:1227882 [6] NCCL INFO Channel 03/0 : 86[6] -> 78[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630209:1630651 [4] NCCL INFO Channel 07/0 : 84[4] -> 76[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96891:97311 [3] NCCL INFO Channel 10/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227466:1227887 [3] NCCL INFO Channel 00/0 : 83[3] -> 75[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96895:97307 [7] NCCL INFO Channel 03/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96894:97312 [6] NCCL INFO Channel 08/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227470:1227885 [7] NCCL INFO Channel 10/0 : 87[7] -> 79[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227463:1227880 [0] NCCL INFO Channel 13/0 : 80[0] -> 72[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630212:1630652 [7] NCCL INFO Channel 07/0 : 79[7] -> 71[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227469:1227882 [6] NCCL INFO Channel 04/0 : 86[6] -> 78[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227465:1227881 [2] NCCL INFO Channel 14/0 : 90[2] -> 82[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630205:1630646 [0] NCCL INFO Channel 13/0 : 80[0] -> 72[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630211:1630649 [6] NCCL INFO Channel 13/0 : 86[6] -> 78[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227466:1227887 [3] NCCL INFO Channel 01/0 : 83[3] -> 75[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227470:1227885 [7] NCCL INFO Channel 11/0 : 87[7] -> 79[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630209:1630651 [4] NCCL INFO Channel 08/0 : 84[4] -> 76[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96891:97311 [3] NCCL INFO Channel 12/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227463:1227880 [0] NCCL INFO Channel 14/0 : 80[0] -> 72[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96894:97312 [6] NCCL INFO Channel 09/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96895:97307 [7] NCCL INFO Channel 04/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630212:1630652 [7] NCCL INFO Channel 09/0 : 79[7] -> 71[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630211:1630649 [6] NCCL INFO Channel 15/0 : 86[6] -> 78[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630205:1630646 [0] NCCL INFO Channel 14/0 : 80[0] -> 72[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636088:2636510 [4] NCCL INFO Channel 07/0 : 52[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227469:1227882 [6] NCCL INFO Channel 05/0 : 86[6] -> 78[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630209:1630651 [4] NCCL INFO Channel 09/0 : 84[4] -> 76[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630212:1630652 [7] NCCL INFO Channel 11/0 : 79[7] -> 71[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96891:97311 [3] NCCL INFO Channel 13/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227465:1227881 [2] NCCL INFO Channel 00/0 : 82[2] -> 74[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227470:1227885 [7] NCCL INFO Channel 12/0 : 87[7] -> 79[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227463:1227880 [0] NCCL INFO Channel 15/0 : 80[0] -> 72[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227466:1227887 [3] NCCL INFO Channel 02/0 : 83[3] -> 75[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630211:1630649 [6] NCCL INFO Channel 01/0 : 78[6] -> 70[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630205:1630646 [0] NCCL INFO Channel 15/0 : 80[0] -> 72[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227469:1227882 [6] NCCL INFO Channel 07/0 : 86[6] -> 78[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227465:1227881 [2] NCCL INFO Channel 01/0 : 82[2] -> 74[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630212:1630652 [7] NCCL INFO Channel 13/0 : 79[7] -> 71[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630209:1630651 [4] NCCL INFO Channel 10/0 : 84[4] -> 76[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227470:1227885 [7] NCCL INFO Channel 13/0 : 87[7] -> 79[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630211:1630649 [6] NCCL INFO Channel 03/0 : 78[6] -> 70[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630205:1630646 [0] NCCL INFO Channel 01/0 : 72[0] -> 64[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96894:97312 [6] NCCL INFO Channel 10/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96895:97307 [7] NCCL INFO Channel 05/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227466:1227887 [3] NCCL INFO Channel 04/0 : 83[3] -> 75[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96891:97311 [3] NCCL INFO Channel 14/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227465:1227881 [2] NCCL INFO Channel 03/0 : 82[2] -> 74[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227469:1227882 [6] NCCL INFO Channel 08/0 : 86[6] -> 78[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96894:97312 [6] NCCL INFO Channel 11/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227470:1227885 [7] NCCL INFO Channel 14/0 : 87[7] -> 79[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227466:1227887 [3] NCCL INFO Channel 05/0 : 83[3] -> 75[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227465:1227881 [2] NCCL INFO Channel 04/0 : 82[2] -> 74[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227469:1227882 [6] NCCL INFO Channel 09/0 : 86[6] -> 78[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630209:1630651 [4] NCCL INFO Channel 11/0 : 84[4] -> 76[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227466:1227887 [3] NCCL INFO Channel 06/0 : 83[3] -> 75[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630211:1630649 [6] NCCL INFO Channel 05/0 : 78[6] -> 70[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630205:1630646 [0] NCCL INFO Channel 03/0 : 72[0] -> 64[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96895:97307 [7] NCCL INFO Channel 06/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96891:97311 [3] NCCL INFO Channel 15/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227469:1227882 [6] NCCL INFO Channel 10/0 : 86[6] -> 78[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227465:1227881 [2] NCCL INFO Channel 05/0 : 82[2] -> 74[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227466:1227887 [3] NCCL INFO Channel 07/0 : 83[3] -> 75[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96894:97312 [6] NCCL INFO Channel 12/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227465:1227881 [2] NCCL INFO Channel 06/0 : 82[2] -> 74[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227469:1227882 [6] NCCL INFO Channel 11/0 : 86[6] -> 78[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227466:1227887 [3] NCCL INFO Channel 08/0 : 83[3] -> 75[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630209:1630651 [4] NCCL INFO Channel 13/0 : 84[4] -> 76[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227465:1227881 [2] NCCL INFO Channel 07/0 : 82[2] -> 74[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630211:1630649 [6] NCCL INFO Channel 07/0 : 78[6] -> 70[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227469:1227882 [6] NCCL INFO Channel 12/0 : 86[6] -> 78[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227466:1227887 [3] NCCL INFO Channel 09/0 : 83[3] -> 75[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630205:1630646 [0] NCCL INFO Channel 05/0 : 72[0] -> 64[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630209:1630651 [4] NCCL INFO Channel 14/0 : 84[4] -> 76[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96895:97307 [7] NCCL INFO Channel 08/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227465:1227881 [2] NCCL INFO Channel 08/0 : 82[2] -> 74[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96894:97312 [6] NCCL INFO Channel 13/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227469:1227882 [6] NCCL INFO Channel 13/0 : 86[6] -> 78[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227466:1227887 [3] NCCL INFO Channel 10/0 : 83[3] -> 75[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227465:1227881 [2] NCCL INFO Channel 09/0 : 82[2] -> 74[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630211:1630649 [6] NCCL INFO Channel 09/0 : 78[6] -> 70[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227466:1227887 [3] NCCL INFO Channel 12/0 : 83[3] -> 75[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227469:1227882 [6] NCCL INFO Channel 15/0 : 86[6] -> 78[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630209:1630651 [4] NCCL INFO Channel 15/0 : 84[4] -> 76[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630205:1630646 [0] NCCL INFO Channel 07/0 : 72[0] -> 64[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227465:1227881 [2] NCCL INFO Channel 11/0 : 82[2] -> 74[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630211:1630649 [6] NCCL INFO Channel 11/0 : 78[6] -> 70[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96895:97307 [7] NCCL INFO Channel 09/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227466:1227887 [3] NCCL INFO Channel 13/0 : 83[3] -> 75[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96894:97312 [6] NCCL INFO Channel 15/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227465:1227881 [2] NCCL INFO Channel 12/0 : 82[2] -> 74[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96895:97307 [7] NCCL INFO Channel 10/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227466:1227887 [3] NCCL INFO Channel 14/0 : 83[3] -> 75[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227465:1227881 [2] NCCL INFO Channel 13/0 : 82[2] -> 74[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630205:1630646 [0] NCCL INFO Channel 09/0 : 72[0] -> 64[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227466:1227887 [3] NCCL INFO Channel 15/0 : 83[3] -> 75[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630209:1630651 [4] NCCL INFO Channel 01/0 : 76[4] -> 68[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227465:1227881 [2] NCCL INFO Channel 14/0 : 82[2] -> 74[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630211:1630649 [6] NCCL INFO Channel 13/0 : 78[6] -> 70[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227465:1227881 [2] NCCL INFO Channel 15/0 : 82[2] -> 74[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630207:1630648 [2] NCCL INFO Channel 00/0 : 82[2] -> 74[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96895:97307 [7] NCCL INFO Channel 11/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630205:1630646 [0] NCCL INFO Channel 11/0 : 72[0] -> 64[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96895:97307 [7] NCCL INFO Channel 12/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630209:1630651 [4] NCCL INFO Channel 03/0 : 76[4] -> 68[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630211:1630649 [6] NCCL INFO Channel 15/0 : 78[6] -> 70[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636085:2636509 [1] NCCL INFO Channel 07/0 : 49[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630207:1630648 [2] NCCL INFO Channel 01/0 : 82[2] -> 74[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636088:2636510 [4] NCCL INFO Channel 08/0 : 52[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96895:97307 [7] NCCL INFO Channel 13/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630205:1630646 [0] NCCL INFO Channel 13/0 : 72[0] -> 64[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630209:1630651 [4] NCCL INFO Channel 05/0 : 76[4] -> 68[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636085:2636509 [1] NCCL INFO Channel 08/0 : 49[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630207:1630648 [2] NCCL INFO Channel 03/0 : 82[2] -> 74[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630205:1630646 [0] NCCL INFO Channel 15/0 : 72[0] -> 64[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636086:2636507 [2] NCCL INFO Channel 00/0 : 58[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636088:2636510 [4] NCCL INFO Channel 09/0 : 52[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630209:1630651 [4] NCCL INFO Channel 07/0 : 76[4] -> 68[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636085:2636509 [1] NCCL INFO Channel 10/0 : 49[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96895:97307 [7] NCCL INFO Channel 14/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630207:1630648 [2] NCCL INFO Channel 04/0 : 82[2] -> 74[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630209:1630651 [4] NCCL INFO Channel 09/0 : 76[4] -> 68[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636088:2636510 [4] NCCL INFO Channel 10/0 : 52[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630207:1630648 [2] NCCL INFO Channel 05/0 : 82[2] -> 74[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636085:2636509 [1] NCCL INFO Channel 11/0 : 49[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636086:2636507 [2] NCCL INFO Channel 04/0 : 58[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630209:1630651 [4] NCCL INFO Channel 11/0 : 76[4] -> 68[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636088:2636510 [4] NCCL INFO Channel 11/0 : 52[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630207:1630648 [2] NCCL INFO Channel 06/0 : 82[2] -> 74[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630209:1630651 [4] NCCL INFO Channel 13/0 : 76[4] -> 68[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636085:2636509 [1] NCCL INFO Channel 12/0 : 49[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636086:2636507 [2] NCCL INFO Channel 06/0 : 58[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630207:1630648 [2] NCCL INFO Channel 07/0 : 82[2] -> 74[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630209:1630651 [4] NCCL INFO Channel 15/0 : 76[4] -> 68[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636088:2636510 [4] NCCL INFO Channel 13/0 : 52[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630207:1630648 [2] NCCL INFO Channel 08/0 : 82[2] -> 74[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636089:2636508 [5] NCCL INFO Channel 00/0 : 61[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636085:2636509 [1] NCCL INFO Channel 13/0 : 49[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630207:1630648 [2] NCCL INFO Channel 09/0 : 82[2] -> 74[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636086:2636507 [2] NCCL INFO Channel 08/0 : 58[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630207:1630648 [2] NCCL INFO Channel 11/0 : 82[2] -> 74[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636088:2636510 [4] NCCL INFO Channel 14/0 : 52[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630207:1630648 [2] NCCL INFO Channel 12/0 : 82[2] -> 74[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636085:2636509 [1] NCCL INFO Channel 14/0 : 49[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636089:2636508 [5] NCCL INFO Channel 02/0 : 61[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636088:2636510 [4] NCCL INFO Channel 15/0 : 52[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630207:1630648 [2] NCCL INFO Channel 13/0 : 82[2] -> 74[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636086:2636507 [2] NCCL INFO Channel 10/0 : 58[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630207:1630648 [2] NCCL INFO Channel 14/0 : 82[2] -> 74[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636085:2636509 [1] NCCL INFO Channel 15/0 : 49[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630207:1630648 [2] NCCL INFO Channel 15/0 : 82[2] -> 74[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636089:2636508 [5] NCCL INFO Channel 04/0 : 61[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636086:2636507 [2] NCCL INFO Channel 12/0 : 58[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630207:1630648 [2] NCCL INFO Channel 01/0 : 74[2] -> 66[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636086:2636507 [2] NCCL INFO Channel 14/0 : 58[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636089:2636508 [5] NCCL INFO Channel 06/0 : 61[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630207:1630648 [2] NCCL INFO Channel 03/0 : 74[2] -> 66[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636086:2636507 [2] NCCL INFO Channel 00/0 : 50[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630207:1630648 [2] NCCL INFO Channel 05/0 : 74[2] -> 66[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636089:2636508 [5] NCCL INFO Channel 08/0 : 61[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630207:1630648 [2] NCCL INFO Channel 07/0 : 74[2] -> 66[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636086:2636507 [2] NCCL INFO Channel 01/0 : 50[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630207:1630648 [2] NCCL INFO Channel 09/0 : 74[2] -> 66[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636089:2636508 [5] NCCL INFO Channel 10/0 : 61[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630207:1630648 [2] NCCL INFO Channel 11/0 : 74[2] -> 66[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636086:2636507 [2] NCCL INFO Channel 03/0 : 50[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630207:1630648 [2] NCCL INFO Channel 13/0 : 74[2] -> 66[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636089:2636508 [5] NCCL INFO Channel 12/0 : 61[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630207:1630648 [2] NCCL INFO Channel 15/0 : 74[2] -> 66[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636086:2636507 [2] NCCL INFO Channel 04/0 : 50[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636089:2636508 [5] NCCL INFO Channel 14/0 : 61[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124160:124581 [0] NCCL INFO Channel 13/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:124162:124587 [2] NCCL INFO Channel 13/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636086:2636507 [2] NCCL INFO Channel 05/0 : 50[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:124166:124583 [6] NCCL INFO Channel 05/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636089:2636508 [5] NCCL INFO Channel 00/0 : 53[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124160:124581 [0] NCCL INFO Channel 15/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:124162:124587 [2] NCCL INFO Channel 15/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636086:2636507 [2] NCCL INFO Channel 06/0 : 50[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:124166:124583 [6] NCCL INFO Channel 07/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636089:2636508 [5] NCCL INFO Channel 01/0 : 53[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124166:124583 [6] NCCL INFO Channel 09/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636086:2636507 [2] NCCL INFO Channel 07/0 : 50[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:124166:124583 [6] NCCL INFO Channel 11/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636089:2636508 [5] NCCL INFO Channel 02/0 : 53[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124161:124582 [1] NCCL INFO Channel 15/0 : 9[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636086:2636507 [2] NCCL INFO Channel 08/0 : 50[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:124166:124583 [6] NCCL INFO Channel 13/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636089:2636508 [5] NCCL INFO Channel 03/0 : 53[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124166:124583 [6] NCCL INFO Channel 15/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636086:2636507 [2] NCCL INFO Channel 09/0 : 50[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:124164:124585 [4] NCCL INFO Channel 09/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636089:2636508 [5] NCCL INFO Channel 04/0 : 53[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124164:124585 [4] NCCL INFO Channel 11/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636086:2636507 [2] NCCL INFO Channel 11/0 : 50[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:124164:124585 [4] NCCL INFO Channel 13/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636089:2636508 [5] NCCL INFO Channel 06/0 : 53[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124164:124585 [4] NCCL INFO Channel 15/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636086:2636507 [2] NCCL INFO Channel 12/0 : 50[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:124167:124586 [7] NCCL INFO Channel 01/0 : 15[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636089:2636508 [5] NCCL INFO Channel 07/0 : 53[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124167:124586 [7] NCCL INFO Channel 03/0 : 15[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636086:2636507 [2] NCCL INFO Channel 13/0 : 50[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:124167:124586 [7] NCCL INFO Channel 05/0 : 15[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636089:2636508 [5] NCCL INFO Channel 08/0 : 53[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124167:124586 [7] NCCL INFO Channel 07/0 : 15[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636086:2636507 [2] NCCL INFO Channel 14/0 : 50[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:124167:124586 [7] NCCL INFO Channel 09/0 : 15[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636089:2636508 [5] NCCL INFO Channel 09/0 : 53[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124167:124586 [7] NCCL INFO Channel 11/0 : 15[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636086:2636507 [2] NCCL INFO Channel 15/0 : 50[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:124167:124586 [7] NCCL INFO Channel 13/0 : 15[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636089:2636508 [5] NCCL INFO Channel 10/0 : 53[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636087:2636506 [3] NCCL INFO Channel 00/0 : 59[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96569:97006 [6] NCCL INFO Channel 10/0 : 54[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96566:97010 [3] NCCL INFO Channel 07/0 : 43[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636084:2636511 [0] NCCL INFO Channel 02/0 : 56[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96563:97003 [0] NCCL INFO Channel 07/0 : 40[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636089:2636508 [5] NCCL INFO Channel 11/0 : 53[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96569:97006 [6] NCCL INFO Channel 12/0 : 54[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636087:2636506 [3] NCCL INFO Channel 02/0 : 59[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636084:2636511 [0] NCCL INFO Channel 04/0 : 56[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96566:97010 [3] NCCL INFO Channel 09/0 : 43[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96570:97005 [7] NCCL INFO Channel 01/0 : 47[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636089:2636508 [5] NCCL INFO Channel 12/0 : 53[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96563:97003 [0] NCCL INFO Channel 09/0 : 40[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636087:2636506 [3] NCCL INFO Channel 04/0 : 59[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96569:97006 [6] NCCL INFO Channel 14/0 : 54[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636089:2636508 [5] NCCL INFO Channel 14/0 : 53[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636084:2636511 [0] NCCL INFO Channel 06/0 : 56[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96566:97010 [3] NCCL INFO Channel 13/0 : 43[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636087:2636506 [3] NCCL INFO Channel 06/0 : 59[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96570:97005 [7] NCCL INFO Channel 03/0 : 47[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636089:2636508 [5] NCCL INFO Channel 15/0 : 53[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636084:2636511 [0] NCCL INFO Channel 08/0 : 56[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96563:97003 [0] NCCL INFO Channel 11/0 : 40[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636087:2636506 [3] NCCL INFO Channel 08/0 : 59[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636091:2636512 [7] NCCL INFO Channel 00/0 : 63[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96569:97006 [6] NCCL INFO Channel 00/0 : 38[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636084:2636511 [0] NCCL INFO Channel 10/0 : 56[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153564:153984 [2] NCCL INFO Channel 04/0 : 90[2] -> 82[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636087:2636506 [3] NCCL INFO Channel 10/0 : 59[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114517:114935 [5] NCCL INFO Channel 07/0 : 77[5] -> 69[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636091:2636512 [7] NCCL INFO Channel 02/0 : 63[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636084:2636511 [0] NCCL INFO Channel 12/0 : 56[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636087:2636506 [3] NCCL INFO Channel 12/0 : 59[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96570:97005 [7] NCCL INFO Channel 05/0 : 47[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636091:2636512 [7] NCCL INFO Channel 04/0 : 63[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636084:2636511 [0] NCCL INFO Channel 14/0 : 56[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96566:97010 [3] NCCL INFO Channel 15/0 : 43[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636087:2636506 [3] NCCL INFO Channel 14/0 : 59[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96563:97003 [0] NCCL INFO Channel 13/0 : 40[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96569:97006 [6] NCCL INFO Channel 02/0 : 38[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96570:97005 [7] NCCL INFO Channel 07/0 : 47[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153564:153984 [2] NCCL INFO Channel 06/0 : 90[2] -> 82[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96569:97006 [6] NCCL INFO Channel 04/0 : 38[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153564:153984 [2] NCCL INFO Channel 08/0 : 90[2] -> 82[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96563:97003 [0] NCCL INFO Channel 15/0 : 40[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153564:153984 [2] NCCL INFO Channel 10/0 : 90[2] -> 82[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96570:97005 [7] NCCL INFO Channel 09/0 : 47[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153564:153984 [2] NCCL INFO Channel 12/0 : 90[2] -> 82[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636091:2636512 [7] NCCL INFO Channel 06/0 : 63[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96569:97006 [6] NCCL INFO Channel 08/0 : 38[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636084:2636511 [0] NCCL INFO Channel 01/0 : 48[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636087:2636506 [3] NCCL INFO Channel 00/0 : 51[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96570:97005 [7] NCCL INFO Channel 11/0 : 47[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636091:2636512 [7] NCCL INFO Channel 08/0 : 63[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96569:97006 [6] NCCL INFO Channel 10/0 : 38[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636084:2636511 [0] NCCL INFO Channel 02/0 : 48[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636087:2636506 [3] NCCL INFO Channel 01/0 : 51[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96570:97005 [7] NCCL INFO Channel 13/0 : 47[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636091:2636512 [7] NCCL INFO Channel 10/0 : 63[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96569:97006 [6] NCCL INFO Channel 12/0 : 38[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636084:2636511 [0] NCCL INFO Channel 03/0 : 48[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636087:2636506 [3] NCCL INFO Channel 02/0 : 51[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96569:97006 [6] NCCL INFO Channel 14/0 : 38[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636091:2636512 [7] NCCL INFO Channel 12/0 : 63[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96569:97006 [6] NCCL INFO Channel 01/0 : 46[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636087:2636506 [3] NCCL INFO Channel 04/0 : 51[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636084:2636511 [0] NCCL INFO Channel 04/0 : 48[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636091:2636512 [7] NCCL INFO Channel 14/0 : 63[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96569:97006 [6] NCCL INFO Channel 03/0 : 46[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636084:2636511 [0] NCCL INFO Channel 05/0 : 48[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636091:2636512 [7] NCCL INFO Channel 00/0 : 55[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636087:2636506 [3] NCCL INFO Channel 05/0 : 51[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96569:97006 [6] NCCL INFO Channel 05/0 : 46[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636084:2636511 [0] NCCL INFO Channel 06/0 : 48[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96569:97006 [6] NCCL INFO Channel 07/0 : 46[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636091:2636512 [7] NCCL INFO Channel 01/0 : 55[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636087:2636506 [3] NCCL INFO Channel 06/0 : 51[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96569:97006 [6] NCCL INFO Channel 09/0 : 46[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636090:2636505 [6] NCCL INFO Channel 00/0 : 62[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96569:97006 [6] NCCL INFO Channel 11/0 : 46[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636084:2636511 [0] NCCL INFO Channel 07/0 : 48[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636091:2636512 [7] NCCL INFO Channel 02/0 : 55[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96569:97006 [6] NCCL INFO Channel 13/0 : 46[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636090:2636505 [6] NCCL INFO Channel 02/0 : 62[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96569:97006 [6] NCCL INFO Channel 15/0 : 46[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636087:2636506 [3] NCCL INFO Channel 07/0 : 51[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153564:153984 [2] NCCL INFO Channel 14/0 : 90[2] -> 82[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636091:2636512 [7] NCCL INFO Channel 03/0 : 55[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114517:114935 [5] NCCL INFO Channel 09/0 : 77[5] -> 69[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636084:2636511 [0] NCCL INFO Channel 09/0 : 48[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114517:114935 [5] NCCL INFO Channel 11/0 : 77[5] -> 69[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636090:2636505 [6] NCCL INFO Channel 04/0 : 62[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636087:2636506 [3] NCCL INFO Channel 08/0 : 51[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114517:114935 [5] NCCL INFO Channel 15/0 : 77[5] -> 69[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636091:2636512 [7] NCCL INFO Channel 04/0 : 55[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636084:2636511 [0] NCCL INFO Channel 10/0 : 48[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636087:2636506 [3] NCCL INFO Channel 09/0 : 51[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636090:2636505 [6] NCCL INFO Channel 08/0 : 62[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636091:2636512 [7] NCCL INFO Channel 05/0 : 55[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636084:2636511 [0] NCCL INFO Channel 11/0 : 48[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636087:2636506 [3] NCCL INFO Channel 10/0 : 51[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636090:2636505 [6] NCCL INFO Channel 10/0 : 62[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636091:2636512 [7] NCCL INFO Channel 06/0 : 55[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636084:2636511 [0] NCCL INFO Channel 12/0 : 48[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636087:2636506 [3] NCCL INFO Channel 12/0 : 51[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636091:2636512 [7] NCCL INFO Channel 08/0 : 55[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636090:2636505 [6] NCCL INFO Channel 12/0 : 62[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636084:2636511 [0] NCCL INFO Channel 13/0 : 48[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636087:2636506 [3] NCCL INFO Channel 13/0 : 51[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636091:2636512 [7] NCCL INFO Channel 09/0 : 55[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636090:2636505 [6] NCCL INFO Channel 14/0 : 62[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636084:2636511 [0] NCCL INFO Channel 14/0 : 48[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636087:2636506 [3] NCCL INFO Channel 14/0 : 51[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636091:2636512 [7] NCCL INFO Channel 10/0 : 55[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636090:2636505 [6] NCCL INFO Channel 00/0 : 54[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636084:2636511 [0] NCCL INFO Channel 15/0 : 48[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636091:2636512 [7] NCCL INFO Channel 11/0 : 55[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636087:2636506 [3] NCCL INFO Channel 15/0 : 51[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636090:2636505 [6] NCCL INFO Channel 01/0 : 54[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636091:2636512 [7] NCCL INFO Channel 12/0 : 55[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636090:2636505 [6] NCCL INFO Channel 02/0 : 54[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636091:2636512 [7] NCCL INFO Channel 13/0 : 55[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636090:2636505 [6] NCCL INFO Channel 03/0 : 54[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636091:2636512 [7] NCCL INFO Channel 14/0 : 55[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636090:2636505 [6] NCCL INFO Channel 04/0 : 54[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636090:2636505 [6] NCCL INFO Channel 05/0 : 54[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636090:2636505 [6] NCCL INFO Channel 07/0 : 54[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636090:2636505 [6] NCCL INFO Channel 08/0 : 54[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636090:2636505 [6] NCCL INFO Channel 09/0 : 54[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636090:2636505 [6] NCCL INFO Channel 10/0 : 54[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636090:2636505 [6] NCCL INFO Channel 11/0 : 54[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636090:2636505 [6] NCCL INFO Channel 12/0 : 54[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636090:2636505 [6] NCCL INFO Channel 13/0 : 54[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636090:2636505 [6] NCCL INFO Channel 15/0 : 54[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:124163:124588 [3] NCCL INFO Connected NVLS tree
+ip-26-0-160-225:124163:124588 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-225:124163:124588 [3] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-160-225:124163:124588 [3] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-160-225:124165:124584 [5] NCCL INFO Connected NVLS tree
+ip-26-0-160-225:124165:124584 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-225:124165:124584 [5] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-160-225:124165:124584 [5] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-168-52:2722046:2722466 [1] NCCL INFO Connected NVLS tree
+ip-26-0-168-52:2722046:2722466 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-52:2722046:2722466 [1] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-168-52:2722046:2722466 [1] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-168-52:2722052:2722464 [7] NCCL INFO Connected NVLS tree
+ip-26-0-168-52:2722052:2722464 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-52:2722052:2722464 [7] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-168-52:2722052:2722464 [7] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-165-59:1557268:1557686 [2] NCCL INFO Connected NVLS tree
+ip-26-0-165-59:1557268:1557686 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-59:1557268:1557686 [2] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-165-59:1557268:1557686 [2] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-168-52:2722047:2722468 [2] NCCL INFO Connected NVLS tree
+ip-26-0-168-52:2722047:2722468 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-52:2722047:2722468 [2] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-168-52:2722047:2722468 [2] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-103:96891:97311 [3] NCCL INFO Connected NVLS tree
+ip-26-0-161-103:96891:97311 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-103:96891:97311 [3] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-161-103:96891:97311 [3] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-160-225:124167:124586 [7] NCCL INFO Connected NVLS tree
+ip-26-0-160-225:124167:124586 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-225:124167:124586 [7] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-160-225:124167:124586 [7] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-168-52:2722049:2722467 [4] NCCL INFO Connected NVLS tree
+ip-26-0-168-52:2722049:2722467 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-52:2722049:2722467 [4] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-168-52:2722049:2722467 [4] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-178:2815966:2816396 [3] NCCL INFO Connected NVLS tree
+ip-26-0-161-178:2815966:2816396 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-178:2815966:2816396 [3] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-161-178:2815966:2816396 [3] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-153:2636086:2636507 [2] NCCL INFO Connected NVLS tree
+ip-26-0-161-153:2636086:2636507 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-153:2636086:2636507 [2] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-161-153:2636086:2636507 [2] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-153:2636090:2636505 [6] NCCL INFO Connected NVLS tree
+ip-26-0-161-153:2636090:2636505 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-153:2636090:2636505 [6] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-161-153:2636090:2636505 [6] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-168-52:2722045:2722470 [0] NCCL INFO Connected NVLS tree
+ip-26-0-168-52:2722045:2722470 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-52:2722045:2722470 [0] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-168-52:2722045:2722470 [0] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-168-30:235655:236069 [7] NCCL INFO Connected NVLS tree
+ip-26-0-168-30:235655:236069 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-30:235655:236069 [7] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-168-30:235655:236069 [7] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-103:96893:97310 [5] NCCL INFO Connected NVLS tree
+ip-26-0-161-103:96893:97310 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-103:96893:97310 [5] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-161-103:96893:97310 [5] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-142:62882:63300 [1] NCCL INFO Connected NVLS tree
+ip-26-0-161-142:62882:63300 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-142:62882:63300 [1] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-161-142:62882:63300 [1] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-168-52:2722051:2722465 [6] NCCL INFO Connected NVLS tree
+ip-26-0-168-52:2722051:2722465 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-52:2722051:2722465 [6] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-168-52:2722051:2722465 [6] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-165-164:1227467:1227886 [4] NCCL INFO Connected NVLS tree
+ip-26-0-165-164:1227467:1227886 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-164:1227467:1227886 [4] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-165-164:1227467:1227886 [4] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-138:96565:97004 [2] NCCL INFO Connected NVLS tree
+ip-26-0-161-138:96565:97004 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-138:96565:97004 [2] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-161-138:96565:97004 [2] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-178:2815970:2816397 [7] NCCL INFO Connected NVLS tree
+ip-26-0-161-178:2815970:2816397 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-178:2815970:2816397 [7] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-161-178:2815970:2816397 [7] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-160-225:124161:124582 [1] NCCL INFO Connected NVLS tree
+ip-26-0-160-225:124161:124582 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-225:124161:124582 [1] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-160-225:124161:124582 [1] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-123:32711:33127 [7] NCCL INFO Connected NVLS tree
+ip-26-0-161-123:32711:33127 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-123:32711:33127 [7] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-161-123:32711:33127 [7] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-103:96889:97309 [1] NCCL INFO Connected NVLS tree
+ip-26-0-161-103:96889:97309 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-103:96889:97309 [1] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-161-103:96889:97309 [1] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-168-30:235649:236071 [1] NCCL INFO Connected NVLS tree
+ip-26-0-168-30:235649:236071 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-30:235649:236071 [1] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-168-30:235649:236071 [1] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-78:114512:114930 [0] NCCL INFO Connected NVLS tree
+ip-26-0-161-78:114512:114930 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-78:114512:114930 [0] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-161-78:114512:114930 [0] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-153:2636091:2636512 [7] NCCL INFO Connected NVLS tree
+ip-26-0-161-153:2636091:2636512 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-153:2636091:2636512 [7] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-161-153:2636091:2636512 [7] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-178:2815968:2816394 [5] NCCL INFO Connected NVLS tree
+ip-26-0-161-178:2815968:2816394 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-178:2815968:2816394 [5] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-161-178:2815968:2816394 [5] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-160-242:1364909:1365327 [7] NCCL INFO Connected NVLS tree
+ip-26-0-160-242:1364909:1365327 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-242:1364909:1365327 [7] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-160-242:1364909:1365327 [7] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-160-225:124164:124585 [4] NCCL INFO Connected NVLS tree
+ip-26-0-160-225:124164:124585 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-225:124164:124585 [4] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-160-225:124164:124585 [4] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-178:2815964:2816401 [1] NCCL INFO Connected NVLS tree
+ip-26-0-161-178:2815964:2816401 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-178:2815964:2816401 [1] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-161-178:2815964:2816401 [1] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-160-225:124162:124587 [2] NCCL INFO Connected NVLS tree
+ip-26-0-160-225:124162:124587 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-225:124162:124587 [2] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-160-225:124162:124587 [2] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-168-52:2722048:2722469 [3] NCCL INFO Connected NVLS tree
+ip-26-0-168-52:2722048:2722469 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-52:2722048:2722469 [3] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-168-52:2722048:2722469 [3] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-138:96567:97008 [4] NCCL INFO Connected NVLS tree
+ip-26-0-161-138:96567:97008 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-138:96567:97008 [4] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-161-138:96567:97008 [4] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-103:96892:97308 [4] NCCL INFO Connected NVLS tree
+ip-26-0-161-103:96892:97308 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-103:96892:97308 [4] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-161-103:96892:97308 [4] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-78:114518:114937 [6] NCCL INFO Connected NVLS tree
+ip-26-0-165-164:1227463:1227880 [0] NCCL INFO Connected NVLS tree
+ip-26-0-161-78:114518:114937 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-78:114518:114937 [6] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-161-78:114518:114937 [6] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-165-164:1227463:1227880 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-164:1227463:1227880 [0] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-165-164:1227463:1227880 [0] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-165-164:1227465:1227881 [2] NCCL INFO Connected NVLS tree
+ip-26-0-165-164:1227465:1227881 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-164:1227465:1227881 [2] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-165-164:1227465:1227881 [2] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-160-242:1364903:1365334 [1] NCCL INFO Connected NVLS tree
+ip-26-0-160-242:1364903:1365334 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-242:1364903:1365334 [1] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-160-242:1364903:1365334 [1] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-165-202:153567:153979 [5] NCCL INFO Connected NVLS tree
+ip-26-0-165-202:153567:153979 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-202:153567:153979 [5] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-165-202:153567:153979 [5] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-78:114516:114934 [4] NCCL INFO Connected NVLS tree
+ip-26-0-161-78:114516:114934 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-78:114516:114934 [4] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-161-78:114516:114934 [4] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-123:32705:33128 [1] NCCL INFO Connected NVLS tree
+ip-26-0-161-123:32705:33128 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-123:32705:33128 [1] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-161-123:32705:33128 [1] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-78:114514:114932 [2] NCCL INFO Connected NVLS tree
+ip-26-0-161-78:114514:114932 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-78:114514:114932 [2] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-161-78:114514:114932 [2] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-103:96895:97307 [7] NCCL INFO Connected NVLS tree
+ip-26-0-161-103:96895:97307 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-103:96895:97307 [7] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-161-103:96895:97307 [7] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-168-34:272150:272576 [0] NCCL INFO Connected NVLS tree
+ip-26-0-168-34:272150:272576 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-34:272150:272576 [0] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-168-34:272150:272576 [0] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-168-30:235648:236064 [0] NCCL INFO Connected NVLS tree
+ip-26-0-168-30:235648:236064 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-30:235648:236064 [0] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-168-30:235648:236064 [0] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-153:2636085:2636509 [1] NCCL INFO Connected NVLS tree
+ip-26-0-161-153:2636085:2636509 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-153:2636085:2636509 [1] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-161-153:2636085:2636509 [1] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-178:2815965:2816395 [2] NCCL INFO Connected NVLS tree
+ip-26-0-161-178:2815965:2816395 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-178:2815965:2816395 [2] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-161-178:2815965:2816395 [2] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-153:2636089:2636508 [5] NCCL INFO Connected NVLS tree
+ip-26-0-161-153:2636089:2636508 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-153:2636089:2636508 [5] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-161-153:2636089:2636508 [5] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-103:96894:97312 [6] NCCL INFO Connected NVLS tree
+ip-26-0-161-103:96894:97312 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-103:96894:97312 [6] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-161-103:96894:97312 [6] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-160-225:124166:124583 [6] NCCL INFO Connected NVLS tree
+ip-26-0-160-225:124166:124583 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-225:124166:124583 [6] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-160-225:124166:124583 [6] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-123:32707:33124 [3] NCCL INFO Connected NVLS tree
+ip-26-0-161-123:32707:33124 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-123:32707:33124 [3] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-161-123:32707:33124 [3] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-168-30:235654:236070 [6] NCCL INFO Connected NVLS tree
+ip-26-0-168-30:235654:236070 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-30:235654:236070 [6] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-168-30:235654:236070 [6] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-168-34:272156:272570 [6] NCCL INFO Connected NVLS tree
+ip-26-0-168-34:272156:272570 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-34:272156:272570 [6] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-168-34:272156:272570 [6] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-142:62887:63305 [6] NCCL INFO Connected NVLS tree
+ip-26-0-161-142:62887:63305 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-142:62887:63305 [6] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-161-142:62887:63305 [6] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-160-242:1364907:1365329 [5] NCCL INFO Connected NVLS tree
+ip-26-0-160-242:1364907:1365329 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-242:1364907:1365329 [5] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-160-242:1364907:1365329 [5] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-165-202:153569:153981 [7] NCCL INFO Connected NVLS tree
+ip-26-0-165-202:153569:153981 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-202:153569:153981 [7] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-165-202:153569:153981 [7] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-160-242:1364904:1365335 [2] NCCL INFO Connected NVLS tree
+ip-26-0-160-242:1364904:1365335 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-242:1364904:1365335 [2] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-160-242:1364904:1365335 [2] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-142:62885:63304 [4] NCCL INFO Connected NVLS tree
+ip-26-0-161-142:62885:63304 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-142:62885:63304 [4] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-161-142:62885:63304 [4] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-168-34:272157:272572 [7] NCCL INFO Connected NVLS tree
+ip-26-0-168-34:272157:272572 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-34:272157:272572 [7] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-168-34:272157:272572 [7] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-142:62881:63298 [0] NCCL INFO Connected NVLS tree
+ip-26-0-161-142:62881:63298 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-142:62881:63298 [0] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-161-142:62881:63298 [0] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-153:2636087:2636506 [3] NCCL INFO Connected NVLS tree
+ip-26-0-161-153:2636087:2636506 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-153:2636087:2636506 [3] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-161-153:2636087:2636506 [3] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-165-131:1630207:1630648 [2] NCCL INFO Connected NVLS tree
+ip-26-0-165-131:1630207:1630648 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-131:1630207:1630648 [2] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-165-131:1630207:1630648 [2] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-165-202:153565:153985 [3] NCCL INFO Connected NVLS tree
+ip-26-0-165-202:153565:153985 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-202:153563:153986 [1] NCCL INFO Connected NVLS tree
+ip-26-0-165-202:153565:153985 [3] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-165-202:153565:153985 [3] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-165-202:153563:153986 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-202:153563:153986 [1] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-165-202:153563:153986 [1] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-138:96570:97005 [7] NCCL INFO Connected NVLS tree
+ip-26-0-161-178:2815969:2816398 [6] NCCL INFO Connected NVLS tree
+ip-26-0-161-138:96570:97005 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-138:96570:97005 [7] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-161-138:96570:97005 [7] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-178:2815969:2816398 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-178:2815969:2816398 [6] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-161-178:2815969:2816398 [6] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-78:114513:114933 [1] NCCL INFO Connected NVLS tree
+ip-26-0-161-78:114513:114933 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-78:114513:114933 [1] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-161-78:114513:114933 [1] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-165-59:1557270:1557688 [4] NCCL INFO Connected NVLS tree
+ip-26-0-161-138:96563:97003 [0] NCCL INFO Connected NVLS tree
+ip-26-0-165-59:1557270:1557688 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-59:1557270:1557688 [4] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-165-59:1557270:1557688 [4] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-138:96563:97003 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-138:96563:97003 [0] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-161-138:96563:97003 [0] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-160-242:1364906:1365331 [4] NCCL INFO Connected NVLS tree
+ip-26-0-160-242:1364906:1365331 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-242:1364906:1365331 [4] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-160-242:1364906:1365331 [4] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-165-164:1227466:1227887 [3] NCCL INFO Connected NVLS tree
+ip-26-0-165-164:1227466:1227887 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-164:1227466:1227887 [3] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-165-164:1227466:1227887 [3] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-160-242:1364908:1365332 [6] NCCL INFO Connected NVLS tree
+ip-26-0-160-242:1364908:1365332 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-242:1364908:1365332 [6] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-160-242:1364908:1365332 [6] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-78:114517:114935 [5] NCCL INFO Connected NVLS tree
+ip-26-0-161-78:114517:114935 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-78:114517:114935 [5] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-161-78:114517:114935 [5] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-160-242:1364905:1365328 [3] NCCL INFO Connected NVLS tree
+ip-26-0-160-242:1364905:1365328 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-242:1364905:1365328 [3] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-160-242:1364905:1365328 [3] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-123:32706:33122 [2] NCCL INFO Connected NVLS tree
+ip-26-0-161-123:32706:33122 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-123:32706:33122 [2] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-161-123:32706:33122 [2] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-168-52:2722050:2722471 [5] NCCL INFO Connected NVLS tree
+ip-26-0-168-52:2722050:2722471 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-142:62886:63303 [5] NCCL INFO Connected NVLS tree
+ip-26-0-161-142:62886:63303 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-142:62886:63303 [5] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-168-52:2722050:2722471 [5] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-168-52:2722050:2722471 [5] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-142:62886:63303 [5] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-165-164:1227470:1227885 [7] NCCL INFO Connected NVLS tree
+ip-26-0-165-164:1227470:1227885 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-164:1227470:1227885 [7] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-165-164:1227470:1227885 [7] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-178:2815967:2816399 [4] NCCL INFO Connected NVLS tree
+ip-26-0-161-178:2815967:2816399 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-178:2815967:2816399 [4] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-161-178:2815967:2816399 [4] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-168-34:272153:272573 [3] NCCL INFO Connected NVLS tree
+ip-26-0-168-34:272153:272573 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-34:272153:272573 [3] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-168-34:272153:272573 [3] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-168-34:272151:272571 [1] NCCL INFO Connected NVLS tree
+ip-26-0-168-34:272151:272571 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-34:272151:272571 [1] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-168-34:272151:272571 [1] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-165-202:153564:153984 [2] NCCL INFO Connected NVLS tree
+ip-26-0-165-202:153564:153984 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-202:153564:153984 [2] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-165-202:153564:153984 [2] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-103:96890:97313 [2] NCCL INFO Connected NVLS tree
+ip-26-0-161-103:96890:97313 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-103:96890:97313 [2] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-161-103:96890:97313 [2] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-168-30:235651:236068 [3] NCCL INFO Connected NVLS tree
+ip-26-0-168-30:235651:236068 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-30:235651:236068 [3] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-168-30:235651:236068 [3] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-142:62883:63301 [2] NCCL INFO Connected NVLS tree
+ip-26-0-161-142:62883:63301 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-142:62883:63301 [2] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-161-142:62883:63301 [2] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-78:114519:114931 [7] NCCL INFO Connected NVLS tree
+ip-26-0-161-78:114519:114931 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-78:114519:114931 [7] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-161-78:114519:114931 [7] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-168-52:2722051:2722465 [6] NCCL INFO comm 0x94b31c0 rank 126 nranks 128 cudaDev 6 nvmlDev 6 busId b9000 commId 0x759f85fd5295c628 - Init COMPLETE
+ip-26-0-168-52:2722048:2722469 [3] NCCL INFO comm 0x88a3670 rank 123 nranks 128 cudaDev 3 nvmlDev 3 busId 86000 commId 0x759f85fd5295c628 - Init COMPLETE
+ip-26-0-168-52:2722047:2722468 [2] NCCL INFO comm 0x967a780 rank 122 nranks 128 cudaDev 2 nvmlDev 2 busId 75000 commId 0x759f85fd5295c628 - Init COMPLETE
+ip-26-0-168-52:2722050:2722471 [5] NCCL INFO comm 0x88d9840 rank 125 nranks 128 cudaDev 5 nvmlDev 5 busId a8000 commId 0x759f85fd5295c628 - Init COMPLETE
+ip-26-0-168-52:2722049:2722467 [4] NCCL INFO comm 0x8a76060 rank 124 nranks 128 cudaDev 4 nvmlDev 4 busId 97000 commId 0x759f85fd5295c628 - Init COMPLETE
+ip-26-0-168-52:2722052:2722464 [7] NCCL INFO comm 0x98d4300 rank 127 nranks 128 cudaDev 7 nvmlDev 7 busId ca000 commId 0x759f85fd5295c628 - Init COMPLETE
+ip-26-0-168-52:2722046:2722466 [1] NCCL INFO comm 0x9a58530 rank 121 nranks 128 cudaDev 1 nvmlDev 1 busId 64000 commId 0x759f85fd5295c628 - Init COMPLETE
+ip-26-0-168-52:2722045:2722470 [0] NCCL INFO comm 0x8786e80 rank 120 nranks 128 cudaDev 0 nvmlDev 0 busId 53000 commId 0x759f85fd5295c628 - Init COMPLETE
+ip-26-0-165-131:1630212:1630652 [7] NCCL INFO Connected NVLS tree
+ip-26-0-165-131:1630212:1630652 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-131:1630212:1630652 [7] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-165-131:1630212:1630652 [7] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-178:2815963:2816400 [0] NCCL INFO Connected NVLS tree
+ip-26-0-161-178:2815963:2816400 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-178:2815963:2816400 [0] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-161-178:2815963:2816400 [0] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-165-59:1557266:1557689 [0] NCCL INFO Connected NVLS tree
+ip-26-0-168-34:272152:272575 [2] NCCL INFO Connected NVLS tree
+ip-26-0-165-59:1557266:1557689 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-59:1557266:1557689 [0] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-165-59:1557266:1557689 [0] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-168-34:272152:272575 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-34:272152:272575 [2] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-168-34:272152:272575 [2] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-142:62884:63299 [3] NCCL INFO Connected NVLS tree
+ip-26-0-161-142:62884:63299 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-142:62884:63299 [3] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-161-142:62884:63299 [3] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-165-202:153562:153983 [0] NCCL INFO Connected NVLS tree
+ip-26-0-165-202:153562:153983 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-202:153562:153983 [0] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-165-202:153562:153983 [0] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-168-30:235653:236067 [5] NCCL INFO Connected NVLS tree
+ip-26-0-168-30:235653:236067 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-30:235653:236067 [5] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-168-30:235653:236067 [5] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-165-164:1227469:1227882 [6] NCCL INFO Connected NVLS tree
+ip-26-0-165-164:1227469:1227882 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-164:1227469:1227882 [6] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-165-164:1227469:1227882 [6] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-138:96564:97009 [1] NCCL INFO Connected NVLS tree
+ip-26-0-161-138:96564:97009 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-138:96564:97009 [1] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-161-138:96564:97009 [1] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-123:32704:33123 [0] NCCL INFO Connected NVLS tree
+ip-26-0-161-123:32704:33123 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-123:32704:33123 [0] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-161-123:32704:33123 [0] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-168-34:272154:272569 [4] NCCL INFO Connected NVLS tree
+ip-26-0-168-34:272154:272569 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-34:272154:272569 [4] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-168-34:272154:272569 [4] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-165-164:1227464:1227884 [1] NCCL INFO Connected NVLS tree
+ip-26-0-165-164:1227464:1227884 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-164:1227464:1227884 [1] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-165-164:1227464:1227884 [1] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-153:2636084:2636511 [0] NCCL INFO Connected NVLS tree
+ip-26-0-161-153:2636084:2636511 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-153:2636084:2636511 [0] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-161-153:2636084:2636511 [0] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-142:62888:63302 [7] NCCL INFO Connected NVLS tree
+ip-26-0-161-142:62888:63302 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-142:62888:63302 [7] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-161-142:62888:63302 [7] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-153:2636088:2636510 [4] NCCL INFO Connected NVLS tree
+ip-26-0-161-153:2636088:2636510 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-153:2636088:2636510 [4] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-161-153:2636088:2636510 [4] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-165-164:1227468:1227883 [5] NCCL INFO Connected NVLS tree
+ip-26-0-165-164:1227468:1227883 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-164:1227468:1227883 [5] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-165-164:1227468:1227883 [5] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-168-34:272155:272574 [5] NCCL INFO Connected NVLS tree
+ip-26-0-168-34:272155:272574 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-34:272155:272574 [5] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-168-34:272155:272574 [5] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-165-131:1630211:1630649 [6] NCCL INFO Connected NVLS tree
+ip-26-0-165-131:1630211:1630649 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-131:1630211:1630649 [6] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-165-131:1630211:1630649 [6] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-168-30:235652:236066 [4] NCCL INFO Connected NVLS tree
+ip-26-0-168-30:235652:236066 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-30:235652:236066 [4] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-168-30:235652:236066 [4] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-165-131:1630206:1630647 [1] NCCL INFO Connected NVLS tree
+ip-26-0-165-131:1630206:1630647 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-131:1630206:1630647 [1] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-165-131:1630206:1630647 [1] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-160-225:124160:124581 [0] NCCL INFO Connected NVLS tree
+ip-26-0-160-225:124160:124581 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-30:235650:236065 [2] NCCL INFO Connected NVLS tree
+ip-26-0-160-225:124160:124581 [0] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-160-225:124160:124581 [0] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-168-30:235650:236065 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-30:235650:236065 [2] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-168-30:235650:236065 [2] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-165-131:1630209:1630651 [4] NCCL INFO Connected NVLS tree
+ip-26-0-165-131:1630209:1630651 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-131:1630209:1630651 [4] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-165-131:1630209:1630651 [4] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-123:32709:33129 [5] NCCL INFO Connected NVLS tree
+ip-26-0-161-123:32709:33129 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-123:32709:33129 [5] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-161-123:32709:33129 [5] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-178:2815970:2816397 [7] NCCL INFO comm 0x931bca0 rank 63 nranks 128 cudaDev 7 nvmlDev 7 busId ca000 commId 0x759f85fd5295c628 - Init COMPLETE
+ip-26-0-161-178:2815968:2816394 [5] NCCL INFO comm 0xa2e3d30 rank 61 nranks 128 cudaDev 5 nvmlDev 5 busId a8000 commId 0x759f85fd5295c628 - Init COMPLETE
+ip-26-0-161-178:2815964:2816401 [1] NCCL INFO comm 0x89ef570 rank 57 nranks 128 cudaDev 1 nvmlDev 1 busId 64000 commId 0x759f85fd5295c628 - Init COMPLETE
+ip-26-0-161-178:2815965:2816395 [2] NCCL INFO comm 0x8e2eb00 rank 58 nranks 128 cudaDev 2 nvmlDev 2 busId 75000 commId 0x759f85fd5295c628 - Init COMPLETE
+ip-26-0-161-178:2815969:2816398 [6] NCCL INFO comm 0x9caa670 rank 62 nranks 128 cudaDev 6 nvmlDev 6 busId b9000 commId 0x759f85fd5295c628 - Init COMPLETE
+ip-26-0-161-178:2815966:2816396 [3] NCCL INFO comm 0x96fb430 rank 59 nranks 128 cudaDev 3 nvmlDev 3 busId 86000 commId 0x759f85fd5295c628 - Init COMPLETE
+ip-26-0-161-178:2815967:2816399 [4] NCCL INFO comm 0x88842c0 rank 60 nranks 128 cudaDev 4 nvmlDev 4 busId 97000 commId 0x759f85fd5295c628 - Init COMPLETE
+ip-26-0-161-178:2815963:2816400 [0] NCCL INFO comm 0x95d2fb0 rank 56 nranks 128 cudaDev 0 nvmlDev 0 busId 53000 commId 0x759f85fd5295c628 - Init COMPLETE
+ip-26-0-160-225:124167:124586 [7] NCCL INFO comm 0x9e9e880 rank 7 nranks 128 cudaDev 7 nvmlDev 7 busId ca000 commId 0x759f85fd5295c628 - Init COMPLETE
+ip-26-0-160-225:124162:124587 [2] NCCL INFO comm 0xa5007c0 rank 2 nranks 128 cudaDev 2 nvmlDev 2 busId 75000 commId 0x759f85fd5295c628 - Init COMPLETE
+ip-26-0-160-225:124161:124582 [1] NCCL INFO comm 0x8a53580 rank 1 nranks 128 cudaDev 1 nvmlDev 1 busId 64000 commId 0x759f85fd5295c628 - Init COMPLETE
+ip-26-0-160-225:124165:124584 [5] NCCL INFO comm 0x8a90b80 rank 5 nranks 128 cudaDev 5 nvmlDev 5 busId a8000 commId 0x759f85fd5295c628 - Init COMPLETE
+ip-26-0-160-225:124166:124583 [6] NCCL INFO comm 0x99b4f20 rank 6 nranks 128 cudaDev 6 nvmlDev 6 busId b9000 commId 0x759f85fd5295c628 - Init COMPLETE
+ip-26-0-160-225:124160:124581 [0] NCCL INFO comm 0x94a3500 rank 0 nranks 128 cudaDev 0 nvmlDev 0 busId 53000 commId 0x759f85fd5295c628 - Init COMPLETE
+ip-26-0-160-225:124164:124585 [4] NCCL INFO comm 0x91c6330 rank 4 nranks 128 cudaDev 4 nvmlDev 4 busId 97000 commId 0x759f85fd5295c628 - Init COMPLETE
+ip-26-0-160-225:124163:124588 [3] NCCL INFO comm 0x9726fa0 rank 3 nranks 128 cudaDev 3 nvmlDev 3 busId 86000 commId 0x759f85fd5295c628 - Init COMPLETE
+ip-26-0-161-103:96888:97306 [0] NCCL INFO Connected NVLS tree
+ip-26-0-161-103:96888:97306 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-103:96888:97306 [0] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-161-103:96888:97306 [0] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-165-164:1227465:1227881 [2] NCCL INFO comm 0x8e8ccf0 rank 82 nranks 128 cudaDev 2 nvmlDev 2 busId 75000 commId 0x759f85fd5295c628 - Init COMPLETE
+ip-26-0-165-164:1227463:1227880 [0] NCCL INFO comm 0x89f8580 rank 80 nranks 128 cudaDev 0 nvmlDev 0 busId 53000 commId 0x759f85fd5295c628 - Init COMPLETE
+ip-26-0-165-164:1227469:1227882 [6] NCCL INFO comm 0x871acf0 rank 86 nranks 128 cudaDev 6 nvmlDev 6 busId b9000 commId 0x759f85fd5295c628 - Init COMPLETE
+ip-26-0-165-164:1227466:1227887 [3] NCCL INFO comm 0x90d62c0 rank 83 nranks 128 cudaDev 3 nvmlDev 3 busId 86000 commId 0x759f85fd5295c628 - Init COMPLETE
+ip-26-0-165-164:1227467:1227886 [4] NCCL INFO comm 0x945c240 rank 84 nranks 128 cudaDev 4 nvmlDev 4 busId 97000 commId 0x759f85fd5295c628 - Init COMPLETE
+ip-26-0-165-164:1227464:1227884 [1] NCCL INFO comm 0x92bcaf0 rank 81 nranks 128 cudaDev 1 nvmlDev 1 busId 64000 commId 0x759f85fd5295c628 - Init COMPLETE
+ip-26-0-165-164:1227468:1227883 [5] NCCL INFO comm 0x9de3570 rank 85 nranks 128 cudaDev 5 nvmlDev 5 busId a8000 commId 0x759f85fd5295c628 - Init COMPLETE
+ip-26-0-165-164:1227470:1227885 [7] NCCL INFO comm 0x94f35b0 rank 87 nranks 128 cudaDev 7 nvmlDev 7 busId ca000 commId 0x759f85fd5295c628 - Init COMPLETE
+ip-26-0-165-131:1630205:1630646 [0] NCCL INFO Connected NVLS tree
+ip-26-0-165-131:1630210:1630653 [5] NCCL INFO Connected NVLS tree
+ip-26-0-165-131:1630205:1630646 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-131:1630210:1630653 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-131:1630205:1630646 [0] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-165-131:1630210:1630653 [5] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-165-131:1630205:1630646 [0] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-165-131:1630210:1630653 [5] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-165-59:1557267:1557683 [1] NCCL INFO Connected NVLS tree
+ip-26-0-165-59:1557267:1557683 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-59:1557267:1557683 [1] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-165-59:1557267:1557683 [1] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-142:62883:63301 [2] NCCL INFO comm 0x8d92c70 rank 42 nranks 128 cudaDev 2 nvmlDev 2 busId 75000 commId 0x759f85fd5295c628 - Init COMPLETE
+ip-26-0-161-142:62881:63298 [0] NCCL INFO comm 0xa365ab0 rank 40 nranks 128 cudaDev 0 nvmlDev 0 busId 53000 commId 0x759f85fd5295c628 - Init COMPLETE
+ip-26-0-161-142:62885:63304 [4] NCCL INFO comm 0xa074600 rank 44 nranks 128 cudaDev 4 nvmlDev 4 busId 97000 commId 0x759f85fd5295c628 - Init COMPLETE
+ip-26-0-161-142:62884:63299 [3] NCCL INFO comm 0x9811ff0 rank 43 nranks 128 cudaDev 3 nvmlDev 3 busId 86000 commId 0x759f85fd5295c628 - Init COMPLETE
+ip-26-0-161-142:62887:63305 [6] NCCL INFO comm 0x88e3b40 rank 46 nranks 128 cudaDev 6 nvmlDev 6 busId b9000 commId 0x759f85fd5295c628 - Init COMPLETE
+ip-26-0-161-142:62886:63303 [5] NCCL INFO comm 0xa047900 rank 45 nranks 128 cudaDev 5 nvmlDev 5 busId a8000 commId 0x759f85fd5295c628 - Init COMPLETE
+ip-26-0-161-142:62888:63302 [7] NCCL INFO comm 0x93c50f0 rank 47 nranks 128 cudaDev 7 nvmlDev 7 busId ca000 commId 0x759f85fd5295c628 - Init COMPLETE
+ip-26-0-161-142:62882:63300 [1] NCCL INFO comm 0x912d0f0 rank 41 nranks 128 cudaDev 1 nvmlDev 1 busId 64000 commId 0x759f85fd5295c628 - Init COMPLETE
+ip-26-0-161-153:2636087:2636506 [3] NCCL INFO comm 0x8b01d20 rank 51 nranks 128 cudaDev 3 nvmlDev 3 busId 86000 commId 0x759f85fd5295c628 - Init COMPLETE
+ip-26-0-161-153:2636089:2636508 [5] NCCL INFO comm 0x897b060 rank 53 nranks 128 cudaDev 5 nvmlDev 5 busId a8000 commId 0x759f85fd5295c628 - Init COMPLETE
+ip-26-0-161-153:2636091:2636512 [7] NCCL INFO comm 0x925c9f0 rank 55 nranks 128 cudaDev 7 nvmlDev 7 busId ca000 commId 0x759f85fd5295c628 - Init COMPLETE
+ip-26-0-161-153:2636085:2636509 [1] NCCL INFO comm 0xa41e5f0 rank 49 nranks 128 cudaDev 1 nvmlDev 1 busId 64000 commId 0x759f85fd5295c628 - Init COMPLETE
+ip-26-0-161-153:2636086:2636507 [2] NCCL INFO comm 0x9ba9780 rank 50 nranks 128 cudaDev 2 nvmlDev 2 busId 75000 commId 0x759f85fd5295c628 - Init COMPLETE
+ip-26-0-161-153:2636088:2636510 [4] NCCL INFO comm 0x94f8db0 rank 52 nranks 128 cudaDev 4 nvmlDev 4 busId 97000 commId 0x759f85fd5295c628 - Init COMPLETE
+ip-26-0-161-153:2636090:2636505 [6] NCCL INFO comm 0x9c681e0 rank 54 nranks 128 cudaDev 6 nvmlDev 6 busId b9000 commId 0x759f85fd5295c628 - Init COMPLETE
+ip-26-0-161-153:2636084:2636511 [0] NCCL INFO comm 0x91a14c0 rank 48 nranks 128 cudaDev 0 nvmlDev 0 busId 53000 commId 0x759f85fd5295c628 - Init COMPLETE
+ip-26-0-168-34:272154:272569 [4] NCCL INFO comm 0x94d3e70 rank 116 nranks 128 cudaDev 4 nvmlDev 4 busId 97000 commId 0x759f85fd5295c628 - Init COMPLETE
+ip-26-0-168-34:272153:272573 [3] NCCL INFO comm 0x880e9b0 rank 115 nranks 128 cudaDev 3 nvmlDev 3 busId 86000 commId 0x759f85fd5295c628 - Init COMPLETE
+ip-26-0-168-34:272151:272571 [1] NCCL INFO comm 0x96eebf0 rank 113 nranks 128 cudaDev 1 nvmlDev 1 busId 64000 commId 0x759f85fd5295c628 - Init COMPLETE
+ip-26-0-168-34:272156:272570 [6] NCCL INFO comm 0xa3f1b00 rank 118 nranks 128 cudaDev 6 nvmlDev 6 busId b9000 commId 0x759f85fd5295c628 - Init COMPLETE
+ip-26-0-168-34:272157:272572 [7] NCCL INFO comm 0x965ac80 rank 119 nranks 128 cudaDev 7 nvmlDev 7 busId ca000 commId 0x759f85fd5295c628 - Init COMPLETE
+ip-26-0-168-34:272152:272575 [2] NCCL INFO comm 0xa6c77f0 rank 114 nranks 128 cudaDev 2 nvmlDev 2 busId 75000 commId 0x759f85fd5295c628 - Init COMPLETE
+ip-26-0-168-34:272155:272574 [5] NCCL INFO comm 0x93fcd30 rank 117 nranks 128 cudaDev 5 nvmlDev 5 busId a8000 commId 0x759f85fd5295c628 - Init COMPLETE
+ip-26-0-168-34:272150:272576 [0] NCCL INFO comm 0x9fbd540 rank 112 nranks 128 cudaDev 0 nvmlDev 0 busId 53000 commId 0x759f85fd5295c628 - Init COMPLETE
+ip-26-0-161-123:32708:33126 [4] NCCL INFO Connected NVLS tree
+ip-26-0-161-123:32708:33126 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-123:32708:33126 [4] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-161-123:32708:33126 [4] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-168-30:235655:236069 [7] NCCL INFO comm 0x96a2a80 rank 111 nranks 128 cudaDev 7 nvmlDev 7 busId ca000 commId 0x759f85fd5295c628 - Init COMPLETE
+ip-26-0-168-30:235651:236068 [3] NCCL INFO comm 0x912b830 rank 107 nranks 128 cudaDev 3 nvmlDev 3 busId 86000 commId 0x759f85fd5295c628 - Init COMPLETE
+ip-26-0-168-30:235653:236067 [5] NCCL INFO comm 0x935a0b0 rank 109 nranks 128 cudaDev 5 nvmlDev 5 busId a8000 commId 0x759f85fd5295c628 - Init COMPLETE
+ip-26-0-168-30:235649:236071 [1] NCCL INFO comm 0x9e33a20 rank 105 nranks 128 cudaDev 1 nvmlDev 1 busId 64000 commId 0x759f85fd5295c628 - Init COMPLETE
+ip-26-0-168-30:235650:236065 [2] NCCL INFO comm 0x914c6b0 rank 106 nranks 128 cudaDev 2 nvmlDev 2 busId 75000 commId 0x759f85fd5295c628 - Init COMPLETE
+ip-26-0-168-30:235648:236064 [0] NCCL INFO comm 0x8b6eec0 rank 104 nranks 128 cudaDev 0 nvmlDev 0 busId 53000 commId 0x759f85fd5295c628 - Init COMPLETE
+ip-26-0-168-30:235654:236070 [6] NCCL INFO comm 0x9bac4c0 rank 110 nranks 128 cudaDev 6 nvmlDev 6 busId b9000 commId 0x759f85fd5295c628 - Init COMPLETE
+ip-26-0-168-30:235652:236066 [4] NCCL INFO comm 0x9edc280 rank 108 nranks 128 cudaDev 4 nvmlDev 4 busId 97000 commId 0x759f85fd5295c628 - Init COMPLETE
+ip-26-0-165-202:153566:153980 [4] NCCL INFO Connected NVLS tree
+ip-26-0-165-202:153566:153980 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-202:153566:153980 [4] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-165-202:153566:153980 [4] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-165-59:1557272:1557687 [6] NCCL INFO Connected NVLS tree
+ip-26-0-165-59:1557272:1557687 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-59:1557272:1557687 [6] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-165-59:1557272:1557687 [6] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-138:96569:97006 [6] NCCL INFO Connected NVLS tree
+ip-26-0-161-138:96569:97006 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-138:96569:97006 [6] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-161-138:96569:97006 [6] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-103:96895:97307 [7] NCCL INFO comm 0xa67bcf0 rank 23 nranks 128 cudaDev 7 nvmlDev 7 busId ca000 commId 0x759f85fd5295c628 - Init COMPLETE
+ip-26-0-161-103:96893:97310 [5] NCCL INFO comm 0x8943870 rank 21 nranks 128 cudaDev 5 nvmlDev 5 busId a8000 commId 0x759f85fd5295c628 - Init COMPLETE
+ip-26-0-161-103:96891:97311 [3] NCCL INFO comm 0x8ec46b0 rank 19 nranks 128 cudaDev 3 nvmlDev 3 busId 86000 commId 0x759f85fd5295c628 - Init COMPLETE
+ip-26-0-161-103:96894:97312 [6] NCCL INFO comm 0x8cd64e0 rank 22 nranks 128 cudaDev 6 nvmlDev 6 busId b9000 commId 0x759f85fd5295c628 - Init COMPLETE
+ip-26-0-161-103:96890:97313 [2] NCCL INFO comm 0x9078a80 rank 18 nranks 128 cudaDev 2 nvmlDev 2 busId 75000 commId 0x759f85fd5295c628 - Init COMPLETE
+ip-26-0-161-103:96892:97308 [4] NCCL INFO comm 0x9df5b40 rank 20 nranks 128 cudaDev 4 nvmlDev 4 busId 97000 commId 0x759f85fd5295c628 - Init COMPLETE
+ip-26-0-161-103:96888:97306 [0] NCCL INFO comm 0x9097a60 rank 16 nranks 128 cudaDev 0 nvmlDev 0 busId 53000 commId 0x759f85fd5295c628 - Init COMPLETE
+ip-26-0-161-103:96889:97309 [1] NCCL INFO comm 0x8a2fc20 rank 17 nranks 128 cudaDev 1 nvmlDev 1 busId 64000 commId 0x759f85fd5295c628 - Init COMPLETE
+ip-26-0-160-242:1364902:1365333 [0] NCCL INFO Connected NVLS tree
+ip-26-0-160-242:1364902:1365333 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-242:1364902:1365333 [0] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-160-242:1364902:1365333 [0] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-165-131:1630208:1630650 [3] NCCL INFO Connected NVLS tree
+ip-26-0-165-131:1630208:1630650 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-131:1630208:1630650 [3] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-165-131:1630208:1630650 [3] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-165-59:1557273:1557685 [7] NCCL INFO Connected NVLS tree
+ip-26-0-165-59:1557273:1557685 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-59:1557273:1557685 [7] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-165-59:1557273:1557685 [7] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-165-202:153568:153982 [6] NCCL INFO Connected NVLS tree
+ip-26-0-165-202:153568:153982 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-202:153568:153982 [6] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-165-202:153568:153982 [6] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-165-59:1557271:1557690 [5] NCCL INFO Connected NVLS tree
+ip-26-0-165-59:1557271:1557690 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-59:1557271:1557690 [5] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-165-59:1557271:1557690 [5] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-160-242:1364909:1365327 [7] NCCL INFO comm 0xa0bdc30 rank 15 nranks 128 cudaDev 7 nvmlDev 7 busId ca000 commId 0x759f85fd5295c628 - Init COMPLETE
+ip-26-0-160-242:1364905:1365328 [3] NCCL INFO comm 0x9a8b4e0 rank 11 nranks 128 cudaDev 3 nvmlDev 3 busId 86000 commId 0x759f85fd5295c628 - Init COMPLETE
+ip-26-0-160-242:1364907:1365329 [5] NCCL INFO comm 0x88ea1b0 rank 13 nranks 128 cudaDev 5 nvmlDev 5 busId a8000 commId 0x759f85fd5295c628 - Init COMPLETE
+ip-26-0-160-242:1364903:1365334 [1] NCCL INFO comm 0x9d31280 rank 9 nranks 128 cudaDev 1 nvmlDev 1 busId 64000 commId 0x759f85fd5295c628 - Init COMPLETE
+ip-26-0-160-242:1364908:1365332 [6] NCCL INFO comm 0xa6080b0 rank 14 nranks 128 cudaDev 6 nvmlDev 6 busId b9000 commId 0x759f85fd5295c628 - Init COMPLETE
+ip-26-0-160-242:1364902:1365333 [0] NCCL INFO comm 0x9a0bd70 rank 8 nranks 128 cudaDev 0 nvmlDev 0 busId 53000 commId 0x759f85fd5295c628 - Init COMPLETE
+ip-26-0-160-242:1364906:1365331 [4] NCCL INFO comm 0xa1ed2c0 rank 12 nranks 128 cudaDev 4 nvmlDev 4 busId 97000 commId 0x759f85fd5295c628 - Init COMPLETE
+ip-26-0-160-242:1364904:1365335 [2] NCCL INFO comm 0x9899fb0 rank 10 nranks 128 cudaDev 2 nvmlDev 2 busId 75000 commId 0x759f85fd5295c628 - Init COMPLETE
+ip-26-0-165-59:1557269:1557684 [3] NCCL INFO Connected NVLS tree
+ip-26-0-165-59:1557269:1557684 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-59:1557269:1557684 [3] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-165-59:1557269:1557684 [3] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-165-202:153566:153980 [4] NCCL INFO comm 0x8c5b600 rank 92 nranks 128 cudaDev 4 nvmlDev 4 busId 97000 commId 0x759f85fd5295c628 - Init COMPLETE
+ip-26-0-165-202:153562:153983 [0] NCCL INFO comm 0x96ebb40 rank 88 nranks 128 cudaDev 0 nvmlDev 0 busId 53000 commId 0x759f85fd5295c628 - Init COMPLETE
+ip-26-0-165-202:153567:153979 [5] NCCL INFO comm 0xa41d470 rank 93 nranks 128 cudaDev 5 nvmlDev 5 busId a8000 commId 0x759f85fd5295c628 - Init COMPLETE
+ip-26-0-165-202:153564:153984 [2] NCCL INFO comm 0x8a1afc0 rank 90 nranks 128 cudaDev 2 nvmlDev 2 busId 75000 commId 0x759f85fd5295c628 - Init COMPLETE
+ip-26-0-165-202:153565:153985 [3] NCCL INFO comm 0x9bf2bb0 rank 91 nranks 128 cudaDev 3 nvmlDev 3 busId 86000 commId 0x759f85fd5295c628 - Init COMPLETE
+ip-26-0-165-202:153568:153982 [6] NCCL INFO comm 0x8a079f0 rank 94 nranks 128 cudaDev 6 nvmlDev 6 busId b9000 commId 0x759f85fd5295c628 - Init COMPLETE
+ip-26-0-165-202:153563:153986 [1] NCCL INFO comm 0xa623b70 rank 89 nranks 128 cudaDev 1 nvmlDev 1 busId 64000 commId 0x759f85fd5295c628 - Init COMPLETE
+ip-26-0-165-202:153569:153981 [7] NCCL INFO comm 0x8981bb0 rank 95 nranks 128 cudaDev 7 nvmlDev 7 busId ca000 commId 0x759f85fd5295c628 - Init COMPLETE
+ip-26-0-161-123:32710:33125 [6] NCCL INFO Connected NVLS tree
+ip-26-0-161-123:32710:33125 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-123:32710:33125 [6] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-161-123:32710:33125 [6] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-78:114515:114936 [3] NCCL INFO Connected NVLS tree
+ip-26-0-161-78:114515:114936 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-78:114515:114936 [3] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-161-78:114515:114936 [3] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-138:96566:97010 [3] NCCL INFO Connected NVLS tree
+ip-26-0-161-138:96566:97010 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-138:96566:97010 [3] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-161-138:96566:97010 [3] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-165-131:1630208:1630650 [3] NCCL INFO comm 0x87280e0 rank 75 nranks 128 cudaDev 3 nvmlDev 3 busId 86000 commId 0x759f85fd5295c628 - Init COMPLETE
+ip-26-0-165-131:1630210:1630653 [5] NCCL INFO comm 0xa0edf00 rank 77 nranks 128 cudaDev 5 nvmlDev 5 busId a8000 commId 0x759f85fd5295c628 - Init COMPLETE
+ip-26-0-165-131:1630212:1630652 [7] NCCL INFO comm 0x891b6c0 rank 79 nranks 128 cudaDev 7 nvmlDev 7 busId ca000 commId 0x759f85fd5295c628 - Init COMPLETE
+ip-26-0-165-131:1630206:1630647 [1] NCCL INFO comm 0x9d917e0 rank 73 nranks 128 cudaDev 1 nvmlDev 1 busId 64000 commId 0x759f85fd5295c628 - Init COMPLETE
+ip-26-0-165-131:1630205:1630646 [0] NCCL INFO comm 0x90c8640 rank 72 nranks 128 cudaDev 0 nvmlDev 0 busId 53000 commId 0x759f85fd5295c628 - Init COMPLETE
+ip-26-0-165-131:1630211:1630649 [6] NCCL INFO comm 0xa6c7430 rank 78 nranks 128 cudaDev 6 nvmlDev 6 busId b9000 commId 0x759f85fd5295c628 - Init COMPLETE
+ip-26-0-165-131:1630209:1630651 [4] NCCL INFO comm 0x940d970 rank 76 nranks 128 cudaDev 4 nvmlDev 4 busId 97000 commId 0x759f85fd5295c628 - Init COMPLETE
+ip-26-0-165-131:1630207:1630648 [2] NCCL INFO comm 0x97a3d30 rank 74 nranks 128 cudaDev 2 nvmlDev 2 busId 75000 commId 0x759f85fd5295c628 - Init COMPLETE
+ip-26-0-161-78:114515:114936 [3] NCCL INFO comm 0x902dae0 rank 67 nranks 128 cudaDev 3 nvmlDev 3 busId 86000 commId 0x759f85fd5295c628 - Init COMPLETE
+ip-26-0-161-78:114519:114931 [7] NCCL INFO comm 0x87170f0 rank 71 nranks 128 cudaDev 7 nvmlDev 7 busId ca000 commId 0x759f85fd5295c628 - Init COMPLETE
+ip-26-0-161-78:114516:114934 [4] NCCL INFO comm 0x95c5e60 rank 68 nranks 128 cudaDev 4 nvmlDev 4 busId 97000 commId 0x759f85fd5295c628 - Init COMPLETE
+ip-26-0-161-78:114512:114930 [0] NCCL INFO comm 0x94e08f0 rank 64 nranks 128 cudaDev 0 nvmlDev 0 busId 53000 commId 0x759f85fd5295c628 - Init COMPLETE
+ip-26-0-161-78:114513:114933 [1] NCCL INFO comm 0xa08d8f0 rank 65 nranks 128 cudaDev 1 nvmlDev 1 busId 64000 commId 0x759f85fd5295c628 - Init COMPLETE
+ip-26-0-161-78:114517:114935 [5] NCCL INFO comm 0xa4db9a0 rank 69 nranks 128 cudaDev 5 nvmlDev 5 busId a8000 commId 0x759f85fd5295c628 - Init COMPLETE
+ip-26-0-161-78:114518:114937 [6] NCCL INFO comm 0x9acb870 rank 70 nranks 128 cudaDev 6 nvmlDev 6 busId b9000 commId 0x759f85fd5295c628 - Init COMPLETE
+ip-26-0-161-78:114514:114932 [2] NCCL INFO comm 0x9f4e0f0 rank 66 nranks 128 cudaDev 2 nvmlDev 2 busId 75000 commId 0x759f85fd5295c628 - Init COMPLETE
+ip-26-0-161-123:32710:33125 [6] NCCL INFO comm 0x9b78cf0 rank 30 nranks 128 cudaDev 6 nvmlDev 6 busId b9000 commId 0x759f85fd5295c628 - Init COMPLETE
+ip-26-0-161-123:32706:33122 [2] NCCL INFO comm 0x9c6ed80 rank 26 nranks 128 cudaDev 2 nvmlDev 2 busId 75000 commId 0x759f85fd5295c628 - Init COMPLETE
+ip-26-0-161-123:32709:33129 [5] NCCL INFO comm 0xa49ae70 rank 29 nranks 128 cudaDev 5 nvmlDev 5 busId a8000 commId 0x759f85fd5295c628 - Init COMPLETE
+ip-26-0-161-123:32707:33124 [3] NCCL INFO comm 0x8ad9f30 rank 27 nranks 128 cudaDev 3 nvmlDev 3 busId 86000 commId 0x759f85fd5295c628 - Init COMPLETE
+ip-26-0-161-123:32708:33126 [4] NCCL INFO comm 0xa4e8b60 rank 28 nranks 128 cudaDev 4 nvmlDev 4 busId 97000 commId 0x759f85fd5295c628 - Init COMPLETE
+ip-26-0-161-123:32705:33128 [1] NCCL INFO comm 0x9c4b570 rank 25 nranks 128 cudaDev 1 nvmlDev 1 busId 64000 commId 0x759f85fd5295c628 - Init COMPLETE
+ip-26-0-161-123:32711:33127 [7] NCCL INFO comm 0xa0d8f70 rank 31 nranks 128 cudaDev 7 nvmlDev 7 busId ca000 commId 0x759f85fd5295c628 - Init COMPLETE
+ip-26-0-161-123:32704:33123 [0] NCCL INFO comm 0xa6bba20 rank 24 nranks 128 cudaDev 0 nvmlDev 0 busId 53000 commId 0x759f85fd5295c628 - Init COMPLETE
+ip-26-0-165-59:1557270:1557688 [4] NCCL INFO comm 0x8b767c0 rank 100 nranks 128 cudaDev 4 nvmlDev 4 busId 97000 commId 0x759f85fd5295c628 - Init COMPLETE
+ip-26-0-165-59:1557272:1557687 [6] NCCL INFO comm 0x9666ec0 rank 102 nranks 128 cudaDev 6 nvmlDev 6 busId b9000 commId 0x759f85fd5295c628 - Init COMPLETE
+ip-26-0-165-59:1557267:1557683 [1] NCCL INFO comm 0x8af4bf0 rank 97 nranks 128 cudaDev 1 nvmlDev 1 busId 64000 commId 0x759f85fd5295c628 - Init COMPLETE
+ip-26-0-165-59:1557273:1557685 [7] NCCL INFO comm 0xa24f4e0 rank 103 nranks 128 cudaDev 7 nvmlDev 7 busId ca000 commId 0x759f85fd5295c628 - Init COMPLETE
+ip-26-0-165-59:1557268:1557686 [2] NCCL INFO comm 0x8d9f540 rank 98 nranks 128 cudaDev 2 nvmlDev 2 busId 75000 commId 0x759f85fd5295c628 - Init COMPLETE
+ip-26-0-165-59:1557266:1557689 [0] NCCL INFO comm 0xa380e00 rank 96 nranks 128 cudaDev 0 nvmlDev 0 busId 53000 commId 0x759f85fd5295c628 - Init COMPLETE
+ip-26-0-165-59:1557271:1557690 [5] NCCL INFO comm 0x8702fb0 rank 101 nranks 128 cudaDev 5 nvmlDev 5 busId a8000 commId 0x759f85fd5295c628 - Init COMPLETE
+ip-26-0-165-59:1557269:1557684 [3] NCCL INFO comm 0xa1496b0 rank 99 nranks 128 cudaDev 3 nvmlDev 3 busId 86000 commId 0x759f85fd5295c628 - Init COMPLETE
+ip-26-0-161-138:96568:97007 [5] NCCL INFO Connected NVLS tree
+ip-26-0-161-138:96568:97007 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-138:96568:97007 [5] NCCL INFO threadThresholds 8/8/64 | 1024/8/64 | 512 | 512
+ip-26-0-161-138:96568:97007 [5] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-138:96567:97008 [4] NCCL INFO comm 0x8aca3d0 rank 36 nranks 128 cudaDev 4 nvmlDev 4 busId 97000 commId 0x759f85fd5295c628 - Init COMPLETE
+ip-26-0-161-138:96565:97004 [2] NCCL INFO comm 0xa462c20 rank 34 nranks 128 cudaDev 2 nvmlDev 2 busId 75000 commId 0x759f85fd5295c628 - Init COMPLETE
+ip-26-0-161-138:96569:97006 [6] NCCL INFO comm 0x90dbd30 rank 38 nranks 128 cudaDev 6 nvmlDev 6 busId b9000 commId 0x759f85fd5295c628 - Init COMPLETE
+ip-26-0-161-138:96563:97003 [0] NCCL INFO comm 0x9c090c0 rank 32 nranks 128 cudaDev 0 nvmlDev 0 busId 53000 commId 0x759f85fd5295c628 - Init COMPLETE
+ip-26-0-161-138:96568:97007 [5] NCCL INFO comm 0x8b312f0 rank 37 nranks 128 cudaDev 5 nvmlDev 5 busId a8000 commId 0x759f85fd5295c628 - Init COMPLETE
+ip-26-0-161-138:96564:97009 [1] NCCL INFO comm 0x987dbe0 rank 33 nranks 128 cudaDev 1 nvmlDev 1 busId 64000 commId 0x759f85fd5295c628 - Init COMPLETE
+ip-26-0-161-138:96570:97005 [7] NCCL INFO comm 0xa31fb20 rank 39 nranks 128 cudaDev 7 nvmlDev 7 busId ca000 commId 0x759f85fd5295c628 - Init COMPLETE
+ip-26-0-161-138:96566:97010 [3] NCCL INFO comm 0x9576100 rank 35 nranks 128 cudaDev 3 nvmlDev 3 busId 86000 commId 0x759f85fd5295c628 - Init COMPLETE
+ip-26-0-160-225:124161:124621 [1] NCCL INFO Using network Libfabric
+ip-26-0-160-225:124160:124620 [0] NCCL INFO Using network Libfabric
+ip-26-0-160-225:124164:124623 [4] NCCL INFO Using network Libfabric
+ip-26-0-160-225:124163:124626 [3] NCCL INFO Using network Libfabric
+ip-26-0-160-225:124165:124625 [5] NCCL INFO Using network Libfabric
+ip-26-0-160-225:124167:124624 [7] NCCL INFO Using network Libfabric
+ip-26-0-160-225:124166:124622 [6] NCCL INFO Using network Libfabric
+ip-26-0-160-225:124162:124627 [2] NCCL INFO Using network Libfabric
+ip-26-0-160-225:124166:124622 [6] NCCL INFO comm 0x99cf4f0 rank 6 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0xf16fff8ce090ba3b - Init START
+ip-26-0-160-225:124167:124624 [7] NCCL INFO comm 0x9eb7fb0 rank 7 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0xf16fff8ce090ba3b - Init START
+ip-26-0-160-225:124160:124620 [0] NCCL INFO comm 0x94b61c0 rank 0 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0xf16fff8ce090ba3b - Init START
+ip-26-0-160-225:124164:124623 [4] NCCL INFO comm 0x91e0850 rank 4 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0xf16fff8ce090ba3b - Init START
+ip-26-0-160-225:124163:124626 [3] NCCL INFO comm 0x972ae70 rank 3 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0xf16fff8ce090ba3b - Init START
+ip-26-0-160-225:124161:124621 [1] NCCL INFO comm 0x8a5caa0 rank 1 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0xf16fff8ce090ba3b - Init START
+ip-26-0-160-225:124165:124625 [5] NCCL INFO comm 0x8aa9a50 rank 5 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0xf16fff8ce090ba3b - Init START
+ip-26-0-160-225:124162:124627 [2] NCCL INFO comm 0xa51c750 rank 2 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0xf16fff8ce090ba3b - Init START
+ip-26-0-160-225:124166:124622 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-160-225:124165:124625 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-160-225:124162:124627 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-160-225:124163:124626 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-160-225:124161:124621 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-160-225:124160:124620 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/541
+ip-26-0-160-225:124164:124623 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-160-225:124167:124624 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+NCCL version 2.18.5+cuda12.2
+ip-26-0-161-78:114512:114968 [0] NCCL INFO Using network Libfabric
+ip-26-0-161-78:114519:114970 [7] NCCL INFO Using network Libfabric
+ip-26-0-161-78:114515:114972 [3] NCCL INFO Using network Libfabric
+ip-26-0-161-78:114517:114969 [5] NCCL INFO Using network Libfabric
+ip-26-0-161-78:114518:114974 [6] NCCL INFO Using network Libfabric
+ip-26-0-161-78:114516:114975 [4] NCCL INFO Using network Libfabric
+ip-26-0-161-78:114514:114971 [2] NCCL INFO Using network Libfabric
+ip-26-0-161-78:114513:114973 [1] NCCL INFO Using network Libfabric
+ip-26-0-161-78:114513:114973 [1] NCCL INFO comm 0xa094be0 rank 1 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0x2bcb5da8b3cfa2cd - Init START
+ip-26-0-161-78:114519:114970 [7] NCCL INFO comm 0x8721f20 rank 7 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0x2bcb5da8b3cfa2cd - Init START
+ip-26-0-161-78:114512:114968 [0] NCCL INFO comm 0x94f6090 rank 0 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0x2bcb5da8b3cfa2cd - Init START
+ip-26-0-161-78:114518:114974 [6] NCCL INFO comm 0x9ad7200 rank 6 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0x2bcb5da8b3cfa2cd - Init START
+ip-26-0-161-78:114517:114969 [5] NCCL INFO comm 0xa4e6130 rank 5 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0x2bcb5da8b3cfa2cd - Init START
+ip-26-0-161-78:114514:114971 [2] NCCL INFO comm 0x9f67ee0 rank 2 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0x2bcb5da8b3cfa2cd - Init START
+ip-26-0-161-78:114516:114975 [4] NCCL INFO comm 0x95def10 rank 4 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0x2bcb5da8b3cfa2cd - Init START
+ip-26-0-161-78:114515:114972 [3] NCCL INFO comm 0x9042190 rank 3 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0x2bcb5da8b3cfa2cd - Init START
+ip-26-0-161-78:114519:114970 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-78:114512:114968 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-78:114513:114973 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-78:114518:114974 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-78:114514:114971 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-78:114517:114969 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-78:114516:114975 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-78:114515:114972 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+ip-26-0-161-138:96569:97049 [6] NCCL INFO Using network Libfabric
+ip-26-0-161-138:96564:97046 [1] NCCL INFO Using network Libfabric
+ip-26-0-161-138:96566:97047 [3] NCCL INFO Using network Libfabric
+ip-26-0-161-138:96567:97050 [4] NCCL INFO Using network Libfabric
+ip-26-0-161-138:96565:97044 [2] NCCL INFO Using network Libfabric
+ip-26-0-161-138:96563:97045 [0] NCCL INFO Using network Libfabric
+ip-26-0-161-138:96568:97048 [5] NCCL INFO Using network Libfabric
+ip-26-0-165-59:1557266:1557721 [0] NCCL INFO Using network Libfabric
+ip-26-0-161-138:96570:97051 [7] NCCL INFO Using network Libfabric
+ip-26-0-165-59:1557272:1557722 [6] NCCL INFO Using network Libfabric
+ip-26-0-165-59:1557271:1557726 [5] NCCL INFO Using network Libfabric
+ip-26-0-165-59:1557273:1557723 [7] NCCL INFO Using network Libfabric
+ip-26-0-165-59:1557270:1557725 [4] NCCL INFO Using network Libfabric
+ip-26-0-165-59:1557267:1557727 [1] NCCL INFO Using network Libfabric
+ip-26-0-165-59:1557269:1557724 [3] NCCL INFO Using network Libfabric
+ip-26-0-165-59:1557268:1557728 [2] NCCL INFO Using network Libfabric
+ip-26-0-161-138:96567:97050 [4] NCCL INFO comm 0x8adfa60 rank 4 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0xe65dda020e3348b0 - Init START
+ip-26-0-161-138:96568:97048 [5] NCCL INFO comm 0x8b4ae50 rank 5 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0xe65dda020e3348b0 - Init START
+ip-26-0-161-138:96565:97044 [2] NCCL INFO comm 0xa46bac0 rank 2 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0xe65dda020e3348b0 - Init START
+ip-26-0-161-138:96566:97047 [3] NCCL INFO comm 0x9589d40 rank 3 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0xe65dda020e3348b0 - Init START
+ip-26-0-161-138:96570:97051 [7] NCCL INFO comm 0xa334e80 rank 7 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0xe65dda020e3348b0 - Init START
+ip-26-0-161-138:96569:97049 [6] NCCL INFO comm 0x90f9380 rank 6 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0xe65dda020e3348b0 - Init START
+ip-26-0-161-138:96564:97046 [1] NCCL INFO comm 0x9897100 rank 1 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0xe65dda020e3348b0 - Init START
+ip-26-0-161-138:96563:97045 [0] NCCL INFO comm 0x9c16200 rank 0 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0xe65dda020e3348b0 - Init START
+ip-26-0-161-138:96570:97051 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-138:96569:97049 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-138:96564:97046 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-138:96568:97048 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-138:96567:97050 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-138:96563:97045 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-59:1557267:1557727 [1] NCCL INFO comm 0x8aff890 rank 1 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0x1bf22b79fd53b3bb - Init START
+ip-26-0-165-59:1557270:1557725 [4] NCCL INFO comm 0x8b8b740 rank 4 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0x1bf22b79fd53b3bb - Init START
+ip-26-0-165-59:1557268:1557728 [2] NCCL INFO comm 0x8db6590 rank 2 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0x1bf22b79fd53b3bb - Init START
+ip-26-0-165-59:1557271:1557726 [5] NCCL INFO comm 0x871d350 rank 5 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0x1bf22b79fd53b3bb - Init START
+ip-26-0-165-59:1557272:1557722 [6] NCCL INFO comm 0x966b0a0 rank 6 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0x1bf22b79fd53b3bb - Init START
+ip-26-0-165-59:1557273:1557723 [7] NCCL INFO comm 0xa268900 rank 7 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0x1bf22b79fd53b3bb - Init START
+ip-26-0-161-138:96565:97044 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-138:96566:97047 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-59:1557269:1557724 [3] NCCL INFO comm 0xa155f80 rank 3 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0x1bf22b79fd53b3bb - Init START
+ip-26-0-165-59:1557266:1557721 [0] NCCL INFO comm 0xa38bfc0 rank 0 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0x1bf22b79fd53b3bb - Init START
+ip-26-0-165-59:1557268:1557728 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-59:1557269:1557724 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-59:1557267:1557727 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-59:1557270:1557725 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-59:1557266:1557721 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-59:1557272:1557722 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-59:1557273:1557723 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-59:1557271:1557726 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+NCCL version 2.18.5+cuda12.2
+ip-26-0-161-103:96888:97344 [0] NCCL INFO Using network Libfabric
+NCCL version 2.18.5+cuda12.2
+ip-26-0-161-103:96892:97345 [4] NCCL INFO Using network Libfabric
+ip-26-0-161-103:96893:97346 [5] NCCL INFO Using network Libfabric
+ip-26-0-161-103:96895:97351 [7] NCCL INFO Using network Libfabric
+ip-26-0-161-103:96894:97349 [6] NCCL INFO Using network Libfabric
+ip-26-0-161-103:96889:97348 [1] NCCL INFO Using network Libfabric
+ip-26-0-161-103:96891:97347 [3] NCCL INFO Using network Libfabric
+ip-26-0-161-103:96890:97350 [2] NCCL INFO Using network Libfabric
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+ip-26-0-160-242:1364902:1365366 [0] NCCL INFO Using network Libfabric
+ip-26-0-160-242:1364905:1365372 [3] NCCL INFO Using network Libfabric
+ip-26-0-160-242:1364908:1365368 [6] NCCL INFO Using network Libfabric
+ip-26-0-160-242:1364909:1365373 [7] NCCL INFO Using network Libfabric
+ip-26-0-160-242:1364906:1365370 [4] NCCL INFO Using network Libfabric
+ip-26-0-160-242:1364903:1365371 [1] NCCL INFO Using network Libfabric
+ip-26-0-160-242:1364904:1365369 [2] NCCL INFO Using network Libfabric
+ip-26-0-161-103:96892:97345 [4] NCCL INFO comm 0x9dff040 rank 4 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0x6ac0a43191d4b189 - Init START
+ip-26-0-161-103:96895:97351 [7] NCCL INFO comm 0xa694e10 rank 7 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0x6ac0a43191d4b189 - Init START
+ip-26-0-161-103:96889:97348 [1] NCCL INFO comm 0x8a48ee0 rank 1 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0x6ac0a43191d4b189 - Init START
+ip-26-0-161-103:96888:97344 [0] NCCL INFO comm 0x90a1500 rank 0 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0x6ac0a43191d4b189 - Init START
+ip-26-0-161-103:96894:97349 [6] NCCL INFO comm 0x8cf0450 rank 6 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0x6ac0a43191d4b189 - Init START
+ip-26-0-161-103:96890:97350 [2] NCCL INFO comm 0x907c3f0 rank 2 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0x6ac0a43191d4b189 - Init START
+ip-26-0-161-103:96891:97347 [3] NCCL INFO comm 0x8eddb20 rank 3 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0x6ac0a43191d4b189 - Init START
+ip-26-0-161-103:96893:97346 [5] NCCL INFO comm 0x894f3a0 rank 5 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0x6ac0a43191d4b189 - Init START
+ip-26-0-161-153:2636084:2636546 [0] NCCL INFO Using network Libfabric
+ip-26-0-161-103:96891:97347 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-103:96892:97345 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-103:96893:97346 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-103:96894:97349 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-103:96890:97350 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-103:96889:97348 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-103:96895:97351 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-103:96888:97344 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-153:2636090:2636547 [6] NCCL INFO Using network Libfabric
+ip-26-0-161-153:2636088:2636549 [4] NCCL INFO Using network Libfabric
+ip-26-0-161-153:2636087:2636550 [3] NCCL INFO Using network Libfabric
+ip-26-0-161-153:2636091:2636548 [7] NCCL INFO Using network Libfabric
+ip-26-0-165-164:1227469:1227920 [6] NCCL INFO Using network Libfabric
+ip-26-0-161-153:2636086:2636552 [2] NCCL INFO Using network Libfabric
+ip-26-0-161-153:2636089:2636551 [5] NCCL INFO Using network Libfabric
+ip-26-0-160-242:1364907:1365367 [5] NCCL INFO Using network Libfabric
+ip-26-0-161-153:2636085:2636553 [1] NCCL INFO Using network Libfabric
+ip-26-0-165-164:1227463:1227919 [0] NCCL INFO Using network Libfabric
+ip-26-0-165-164:1227468:1227922 [5] NCCL INFO Using network Libfabric
+ip-26-0-165-164:1227470:1227923 [7] NCCL INFO Using network Libfabric
+ip-26-0-165-164:1227465:1227924 [2] NCCL INFO Using network Libfabric
+ip-26-0-165-164:1227466:1227921 [3] NCCL INFO Using network Libfabric
+ip-26-0-165-164:1227467:1227925 [4] NCCL INFO Using network Libfabric
+ip-26-0-165-164:1227464:1227926 [1] NCCL INFO Using network Libfabric
+ip-26-0-160-242:1364903:1365371 [1] NCCL INFO comm 0x9d45a90 rank 1 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0x7f823661d2a228a2 - Init START
+ip-26-0-160-242:1364902:1365366 [0] NCCL INFO comm 0x9a12810 rank 0 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0x7f823661d2a228a2 - Init START
+ip-26-0-160-242:1364904:1365369 [2] NCCL INFO comm 0x98a3450 rank 2 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0x7f823661d2a228a2 - Init START
+ip-26-0-160-242:1364906:1365370 [4] NCCL INFO comm 0xa1f5e10 rank 4 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0x7f823661d2a228a2 - Init START
+ip-26-0-160-242:1364905:1365372 [3] NCCL INFO comm 0x9a91f10 rank 3 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0x7f823661d2a228a2 - Init START
+ip-26-0-160-242:1364909:1365373 [7] NCCL INFO comm 0xa0d8c00 rank 7 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0x7f823661d2a228a2 - Init START
+ip-26-0-160-242:1364907:1365367 [5] NCCL INFO comm 0x88ef400 rank 5 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0x7f823661d2a228a2 - Init START
+ip-26-0-160-242:1364908:1365368 [6] NCCL INFO comm 0xa611850 rank 6 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0x7f823661d2a228a2 - Init START
+ip-26-0-161-153:2636087:2636550 [3] NCCL INFO comm 0x8b16c80 rank 3 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0x7a5c769d1a9a12a1 - Init START
+ip-26-0-161-153:2636086:2636552 [2] NCCL INFO comm 0x9bc2e10 rank 2 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0x7a5c769d1a9a12a1 - Init START
+ip-26-0-161-153:2636085:2636553 [1] NCCL INFO comm 0xa427240 rank 1 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0x7a5c769d1a9a12a1 - Init START
+ip-26-0-161-153:2636090:2636547 [6] NCCL INFO comm 0x9c7bbd0 rank 6 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0x7a5c769d1a9a12a1 - Init START
+ip-26-0-160-242:1364909:1365373 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-160-242:1364906:1365370 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-160-242:1364904:1365369 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-160-242:1364903:1365371 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-160-242:1364907:1365367 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-160-242:1364902:1365366 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-153:2636091:2636548 [7] NCCL INFO comm 0x9275a00 rank 7 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0x7a5c769d1a9a12a1 - Init START
+ip-26-0-161-153:2636084:2636546 [0] NCCL INFO comm 0x91aa590 rank 0 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0x7a5c769d1a9a12a1 - Init START
+ip-26-0-161-153:2636088:2636549 [4] NCCL INFO comm 0x9514760 rank 4 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0x7a5c769d1a9a12a1 - Init START
+ip-26-0-161-153:2636089:2636551 [5] NCCL INFO comm 0x8990aa0 rank 5 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0x7a5c769d1a9a12a1 - Init START
+ip-26-0-160-242:1364905:1365372 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-160-242:1364908:1365368 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-164:1227466:1227921 [3] NCCL INFO comm 0x90f2670 rank 3 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0xc0a121b95733704f - Init START
+ip-26-0-165-164:1227467:1227925 [4] NCCL INFO comm 0x94610b0 rank 4 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0xc0a121b95733704f - Init START
+ip-26-0-165-164:1227468:1227922 [5] NCCL INFO comm 0x9dec4e0 rank 5 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0xc0a121b95733704f - Init START
+ip-26-0-165-164:1227465:1227924 [2] NCCL INFO comm 0x8e85d80 rank 2 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0xc0a121b95733704f - Init START
+ip-26-0-165-164:1227463:1227919 [0] NCCL INFO comm 0x8a0d780 rank 0 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0xc0a121b95733704f - Init START
+ip-26-0-165-164:1227470:1227923 [7] NCCL INFO comm 0x94f7a70 rank 7 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0xc0a121b95733704f - Init START
+ip-26-0-165-164:1227464:1227926 [1] NCCL INFO comm 0x92d57b0 rank 1 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0xc0a121b95733704f - Init START
+ip-26-0-165-164:1227469:1227920 [6] NCCL INFO comm 0x8720770 rank 6 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0xc0a121b95733704f - Init START
+ip-26-0-165-164:1227468:1227922 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-164:1227469:1227920 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-164:1227466:1227921 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-164:1227465:1227924 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-164:1227467:1227925 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-164:1227470:1227923 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-164:1227463:1227919 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-164:1227464:1227926 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+ip-26-0-161-178:2815963:2816433 [0] NCCL INFO Using network Libfabric
+ip-26-0-161-123:32704:33159 [0] NCCL INFO Using network Libfabric
+ip-26-0-165-202:153562:154017 [0] NCCL INFO Using network Libfabric
+ip-26-0-161-123:32706:33161 [2] NCCL INFO Using network Libfabric
+ip-26-0-161-123:32710:33160 [6] NCCL INFO Using network Libfabric
+ip-26-0-168-52:2722049:2722504 [4] NCCL INFO Using network Libfabric
+ip-26-0-168-34:272150:272606 [0] NCCL INFO Using network Libfabric
+ip-26-0-161-123:32708:33162 [4] NCCL INFO Using network Libfabric
+ip-26-0-161-178:2815965:2816435 [2] NCCL INFO Using network Libfabric
+ip-26-0-168-52:2722045:2722502 [0] NCCL INFO Using network Libfabric
+ip-26-0-168-52:2722047:2722507 [2] NCCL INFO Using network Libfabric
+ip-26-0-161-123:32709:33164 [5] NCCL INFO Using network Libfabric
+ip-26-0-161-178:2815970:2816437 [7] NCCL INFO Using network Libfabric
+ip-26-0-168-52:2722051:2722503 [6] NCCL INFO Using network Libfabric
+ip-26-0-161-123:32707:33163 [3] NCCL INFO Using network Libfabric
+ip-26-0-161-178:2815969:2816436 [6] NCCL INFO Using network Libfabric
+ip-26-0-161-178:2815967:2816438 [4] NCCL INFO Using network Libfabric
+ip-26-0-161-178:2815968:2816434 [5] NCCL INFO Using network Libfabric
+ip-26-0-161-123:32711:33165 [7] NCCL INFO Using network Libfabric
+ip-26-0-168-34:272156:272607 [6] NCCL INFO Using network Libfabric
+ip-26-0-168-52:2722048:2722505 [3] NCCL INFO Using network Libfabric
+ip-26-0-161-178:2815966:2816439 [3] NCCL INFO Using network Libfabric
+ip-26-0-168-34:272154:272608 [4] NCCL INFO Using network Libfabric
+ip-26-0-165-202:153564:154018 [2] NCCL INFO Using network Libfabric
+ip-26-0-161-178:2815964:2816440 [1] NCCL INFO Using network Libfabric
+ip-26-0-168-52:2722052:2722506 [7] NCCL INFO Using network Libfabric
+ip-26-0-161-123:32705:33166 [1] NCCL INFO Using network Libfabric
+ip-26-0-165-202:153568:154019 [6] NCCL INFO Using network Libfabric
+ip-26-0-168-34:272152:272610 [2] NCCL INFO Using network Libfabric
+ip-26-0-168-52:2722046:2722509 [1] NCCL INFO Using network Libfabric
+ip-26-0-168-34:272151:272611 [1] NCCL INFO Using network Libfabric
+ip-26-0-168-34:272157:272613 [7] NCCL INFO Using network Libfabric
+ip-26-0-168-52:2722050:2722508 [5] NCCL INFO Using network Libfabric
+ip-26-0-168-34:272153:272612 [3] NCCL INFO Using network Libfabric
+ip-26-0-165-202:153567:154023 [5] NCCL INFO Using network Libfabric
+ip-26-0-165-202:153565:154022 [3] NCCL INFO Using network Libfabric
+ip-26-0-168-34:272155:272609 [5] NCCL INFO Using network Libfabric
+ip-26-0-165-202:153563:154024 [1] NCCL INFO Using network Libfabric
+ip-26-0-165-202:153566:154021 [4] NCCL INFO Using network Libfabric
+ip-26-0-161-178:2815968:2816434 [5] NCCL INFO comm 0xa2fead0 rank 5 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0x6a396e45aa98de50 - Init START
+ip-26-0-161-178:2815967:2816438 [4] NCCL INFO comm 0x888d050 rank 4 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0x6a396e45aa98de50 - Init START
+ip-26-0-161-178:2815969:2816436 [6] NCCL INFO comm 0x9cc60a0 rank 6 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0x6a396e45aa98de50 - Init START
+ip-26-0-161-178:2815964:2816440 [1] NCCL INFO comm 0x89f39e0 rank 1 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0x6a396e45aa98de50 - Init START
+ip-26-0-161-178:2815970:2816437 [7] NCCL INFO comm 0x9335470 rank 7 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0x6a396e45aa98de50 - Init START
+ip-26-0-161-178:2815965:2816435 [2] NCCL INFO comm 0x8e44410 rank 2 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0x6a396e45aa98de50 - Init START
+ip-26-0-161-178:2815963:2816433 [0] NCCL INFO comm 0x95efcc0 rank 0 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0x6a396e45aa98de50 - Init START
+ip-26-0-161-178:2815966:2816439 [3] NCCL INFO comm 0x9715940 rank 3 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0x6a396e45aa98de50 - Init START
+ip-26-0-161-123:32710:33160 [6] NCCL INFO comm 0x9b7e830 rank 6 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0x738064f7309df2d9 - Init START
+ip-26-0-161-123:32704:33159 [0] NCCL INFO comm 0xa6c2260 rank 0 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0x738064f7309df2d9 - Init START
+ip-26-0-161-123:32711:33165 [7] NCCL INFO comm 0xa0ea330 rank 7 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0x738064f7309df2d9 - Init START
+ip-26-0-161-123:32709:33164 [5] NCCL INFO comm 0xa4a44d0 rank 5 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0x738064f7309df2d9 - Init START
+ip-26-0-161-123:32705:33166 [1] NCCL INFO comm 0x9c5f0d0 rank 1 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0x738064f7309df2d9 - Init START
+ip-26-0-161-123:32706:33161 [2] NCCL INFO comm 0x9c86c40 rank 2 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0x738064f7309df2d9 - Init START
+ip-26-0-161-123:32707:33163 [3] NCCL INFO comm 0x8aede90 rank 3 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0x738064f7309df2d9 - Init START
+ip-26-0-161-123:32708:33162 [4] NCCL INFO comm 0xa4ddf70 rank 4 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0x738064f7309df2d9 - Init START
+ip-26-0-168-52:2722052:2722506 [7] NCCL INFO comm 0x98d7670 rank 7 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0xd177a32f3dbed555 - Init START
+ip-26-0-168-52:2722051:2722503 [6] NCCL INFO comm 0x94c7540 rank 6 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0xd177a32f3dbed555 - Init START
+ip-26-0-168-52:2722050:2722508 [5] NCCL INFO comm 0x88f2340 rank 5 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0xd177a32f3dbed555 - Init START
+ip-26-0-168-52:2722045:2722502 [0] NCCL INFO comm 0x8798320 rank 0 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0xd177a32f3dbed555 - Init START
+ip-26-0-168-52:2722049:2722504 [4] NCCL INFO comm 0x8a7b070 rank 4 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0xd177a32f3dbed555 - Init START
+ip-26-0-168-52:2722046:2722509 [1] NCCL INFO comm 0x9a62ea0 rank 1 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0xd177a32f3dbed555 - Init START
+ip-26-0-168-52:2722048:2722505 [3] NCCL INFO comm 0x88af800 rank 3 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0xd177a32f3dbed555 - Init START
+ip-26-0-168-52:2722047:2722507 [2] NCCL INFO comm 0x96845c0 rank 2 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0xd177a32f3dbed555 - Init START
+ip-26-0-161-178:2815967:2816438 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-178:2815968:2816434 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-178:2815965:2816435 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-178:2815970:2816437 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-178:2815964:2816440 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-178:2815969:2816436 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-178:2815966:2816439 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-178:2815963:2816433 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-123:32704:33159 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-123:32710:33160 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-123:32706:33161 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-123:32709:33164 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-123:32711:33165 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-123:32707:33163 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-123:32708:33162 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-123:32705:33166 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-34:272153:272612 [3] NCCL INFO comm 0x88144f0 rank 3 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0x84900e3ce4980498 - Init START
+ip-26-0-168-34:272154:272608 [4] NCCL INFO comm 0x94dc980 rank 4 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0x84900e3ce4980498 - Init START
+ip-26-0-168-34:272152:272610 [2] NCCL INFO comm 0xa6db2e0 rank 2 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0x84900e3ce4980498 - Init START
+ip-26-0-168-34:272155:272609 [5] NCCL INFO comm 0x9418790 rank 5 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0x84900e3ce4980498 - Init START
+ip-26-0-168-34:272156:272607 [6] NCCL INFO comm 0xa3fa270 rank 6 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0x84900e3ce4980498 - Init START
+ip-26-0-168-34:272151:272611 [1] NCCL INFO comm 0x9709190 rank 1 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0x84900e3ce4980498 - Init START
+ip-26-0-168-34:272150:272606 [0] NCCL INFO comm 0x9fd74c0 rank 0 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0x84900e3ce4980498 - Init START
+ip-26-0-168-52:2722046:2722509 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-52:2722047:2722507 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-34:272157:272613 [7] NCCL INFO comm 0x9663cf0 rank 7 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0x84900e3ce4980498 - Init START
+ip-26-0-168-52:2722052:2722506 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-52:2722049:2722504 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-52:2722045:2722502 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-52:2722051:2722503 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-52:2722050:2722508 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-52:2722048:2722505 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-34:272156:272607 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-34:272152:272610 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-34:272153:272612 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-34:272151:272611 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-34:272155:272609 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-34:272154:272608 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-34:272150:272606 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-34:272157:272613 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-202:153569:154020 [7] NCCL INFO Using network Libfabric
+NCCL version 2.18.5+cuda12.2
+ip-26-0-165-202:153569:154020 [7] NCCL INFO comm 0x898b390 rank 7 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0x74595ea6be2a949 - Init START
+ip-26-0-165-202:153562:154017 [0] NCCL INFO comm 0x96f2310 rank 0 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0x74595ea6be2a949 - Init START
+ip-26-0-165-202:153563:154024 [1] NCCL INFO comm 0xa62e5f0 rank 1 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0x74595ea6be2a949 - Init START
+ip-26-0-165-202:153564:154018 [2] NCCL INFO comm 0x8a1eea0 rank 2 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0x74595ea6be2a949 - Init START
+ip-26-0-165-202:153565:154022 [3] NCCL INFO comm 0x9c0d0c0 rank 3 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0x74595ea6be2a949 - Init START
+ip-26-0-165-202:153567:154023 [5] NCCL INFO comm 0xa426920 rank 5 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0x74595ea6be2a949 - Init START
+ip-26-0-165-202:153566:154021 [4] NCCL INFO comm 0x8c7b650 rank 4 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0x74595ea6be2a949 - Init START
+ip-26-0-165-202:153568:154019 [6] NCCL INFO comm 0x8a11410 rank 6 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0x74595ea6be2a949 - Init START
+ip-26-0-165-202:153564:154018 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-202:153562:154017 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-202:153563:154024 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-202:153565:154022 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-202:153569:154020 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-202:153568:154019 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-202:153566:154021 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-202:153567:154023 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-131:1630205:1630687 [0] NCCL INFO Using network Libfabric
+ip-26-0-165-131:1630210:1630689 [5] NCCL INFO Using network Libfabric
+ip-26-0-165-131:1630212:1630690 [7] NCCL INFO Using network Libfabric
+ip-26-0-165-131:1630211:1630688 [6] NCCL INFO Using network Libfabric
+ip-26-0-165-131:1630208:1630691 [3] NCCL INFO Using network Libfabric
+ip-26-0-165-131:1630206:1630692 [1] NCCL INFO Using network Libfabric
+ip-26-0-165-131:1630209:1630693 [4] NCCL INFO Using network Libfabric
+ip-26-0-165-131:1630207:1630694 [2] NCCL INFO Using network Libfabric
+ip-26-0-165-131:1630212:1630690 [7] NCCL INFO comm 0x8930b50 rank 7 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0x90a426b68b3ef643 - Init START
+ip-26-0-165-131:1630206:1630692 [1] NCCL INFO comm 0x9d97500 rank 1 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0x90a426b68b3ef643 - Init START
+ip-26-0-165-131:1630211:1630688 [6] NCCL INFO comm 0xa6d1fa0 rank 6 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0x90a426b68b3ef643 - Init START
+ip-26-0-165-131:1630207:1630694 [2] NCCL INFO comm 0x97a9230 rank 2 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0x90a426b68b3ef643 - Init START
+ip-26-0-165-131:1630209:1630693 [4] NCCL INFO comm 0x9416000 rank 4 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0x90a426b68b3ef643 - Init START
+ip-26-0-165-131:1630205:1630687 [0] NCCL INFO comm 0x90d2dc0 rank 0 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0x90a426b68b3ef643 - Init START
+ip-26-0-165-131:1630208:1630691 [3] NCCL INFO comm 0x873de50 rank 3 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0x90a426b68b3ef643 - Init START
+ip-26-0-165-131:1630210:1630689 [5] NCCL INFO comm 0xa1085c0 rank 5 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0x90a426b68b3ef643 - Init START
+NCCL version 2.18.5+cuda12.2
+ip-26-0-165-131:1630206:1630692 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/409
+ip-26-0-165-131:1630212:1630690 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/409
+ip-26-0-165-131:1630205:1630687 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/409
+ip-26-0-165-131:1630211:1630688 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/409
+ip-26-0-165-131:1630209:1630693 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/409
+ip-26-0-165-131:1630208:1630691 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/409
+ip-26-0-165-131:1630207:1630694 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/409
+ip-26-0-165-131:1630210:1630689 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/409
+NCCL version 2.18.5+cuda12.2
+ip-26-0-161-142:62881:63337 [0] NCCL INFO Using network Libfabric
+ip-26-0-161-142:62888:63340 [7] NCCL INFO Using network Libfabric
+ip-26-0-161-142:62885:63342 [4] NCCL INFO Using network Libfabric
+ip-26-0-161-142:62886:63341 [5] NCCL INFO Using network Libfabric
+ip-26-0-161-142:62883:63338 [2] NCCL INFO Using network Libfabric
+ip-26-0-168-30:235648:236102 [0] NCCL INFO Using network Libfabric
+ip-26-0-168-30:235652:236105 [4] NCCL INFO Using network Libfabric
+ip-26-0-168-30:235650:236106 [2] NCCL INFO Using network Libfabric
+ip-26-0-161-142:62882:63343 [1] NCCL INFO Using network Libfabric
+ip-26-0-168-30:235653:236103 [5] NCCL INFO Using network Libfabric
+ip-26-0-168-30:235655:236104 [7] NCCL INFO Using network Libfabric
+ip-26-0-161-142:62887:63344 [6] NCCL INFO Using network Libfabric
+ip-26-0-168-30:235651:236108 [3] NCCL INFO Using network Libfabric
+ip-26-0-168-30:235654:236107 [6] NCCL INFO Using network Libfabric
+ip-26-0-168-30:235649:236109 [1] NCCL INFO Using network Libfabric
+ip-26-0-161-142:62884:63339 [3] NCCL INFO Using network Libfabric
+ip-26-0-168-30:235654:236107 [6] NCCL INFO comm 0x9bb0ad0 rank 6 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0xd7d26cee33c1eceb - Init START
+ip-26-0-168-30:235653:236103 [5] NCCL INFO comm 0x93640f0 rank 5 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0xd7d26cee33c1eceb - Init START
+ip-26-0-168-30:235655:236104 [7] NCCL INFO comm 0x96bac20 rank 7 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0xd7d26cee33c1eceb - Init START
+ip-26-0-168-30:235648:236102 [0] NCCL INFO comm 0x8b68a40 rank 0 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0xd7d26cee33c1eceb - Init START
+ip-26-0-168-30:235652:236105 [4] NCCL INFO comm 0x9ee4a80 rank 4 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0xd7d26cee33c1eceb - Init START
+ip-26-0-168-30:235649:236109 [1] NCCL INFO comm 0x9e4d2f0 rank 1 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0xd7d26cee33c1eceb - Init START
+ip-26-0-168-30:235650:236106 [2] NCCL INFO comm 0x9168a20 rank 2 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0xd7d26cee33c1eceb - Init START
+ip-26-0-168-30:235651:236108 [3] NCCL INFO comm 0x9147400 rank 3 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0xd7d26cee33c1eceb - Init START
+ip-26-0-161-142:62882:63343 [1] NCCL INFO comm 0x9135480 rank 1 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0x66d95c3778073495 - Init START
+ip-26-0-161-142:62884:63339 [3] NCCL INFO comm 0x982c860 rank 3 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0x66d95c3778073495 - Init START
+ip-26-0-161-142:62888:63340 [7] NCCL INFO comm 0x93bef90 rank 7 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0x66d95c3778073495 - Init START
+ip-26-0-161-142:62881:63337 [0] NCCL INFO comm 0xa371a00 rank 0 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0x66d95c3778073495 - Init START
+ip-26-0-161-142:62883:63338 [2] NCCL INFO comm 0x8d9b840 rank 2 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0x66d95c3778073495 - Init START
+ip-26-0-161-142:62887:63344 [6] NCCL INFO comm 0x88e06e0 rank 6 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0x66d95c3778073495 - Init START
+ip-26-0-161-142:62885:63342 [4] NCCL INFO comm 0xa08eb80 rank 4 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0x66d95c3778073495 - Init START
+ip-26-0-161-142:62886:63341 [5] NCCL INFO comm 0xa05d130 rank 5 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0x66d95c3778073495 - Init START
+ip-26-0-168-30:235655:236104 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-30:235652:236105 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-30:235654:236107 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-30:235648:236102 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-30:235649:236109 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-30:235650:236106 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-30:235651:236108 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-30:235653:236103 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-142:62888:63340 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-142:62881:63337 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-142:62882:63343 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-142:62887:63344 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-142:62884:63339 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-142:62883:63338 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-142:62886:63341 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-142:62885:63342 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-153:2636087:2636550 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-153:2636086:2636552 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-153:2636085:2636553 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-153:2636090:2636547 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-153:2636089:2636551 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-153:2636084:2636546 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-153:2636088:2636549 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-153:2636091:2636548 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-202:153568:154019 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-165-202:153568:154019 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-165-59:1557266:1557721 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-165-59:1557266:1557721 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-161-123:32711:33165 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-161-123:32711:33165 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-165-202:153569:154020 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-165-202:153569:154020 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-161-123:32705:33166 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-161-123:32705:33166 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-161-103:96895:97351 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-161-103:96895:97351 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-161-78:114515:114972 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-161-78:114515:114972 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-161-142:62887:63344 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-161-142:62887:63344 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-161-123:32710:33160 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-161-123:32710:33160 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-165-202:153564:154018 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-165-202:153564:154018 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-165-59:1557269:1557724 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-165-59:1557269:1557724 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-165-59:1557267:1557727 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-165-59:1557267:1557727 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-161-178:2815963:2816433 [0] NCCL INFO Setting affinity for GPU 0 to 0fff,ffffffff
+ip-26-0-161-178:2815963:2816433 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-165-59:1557271:1557726 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-165-59:1557271:1557726 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-161-138:96568:97048 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,fff00000,00000000
+ip-26-0-161-138:96568:97048 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-161-123:32704:33159 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-161-123:32704:33159 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-161-178:2815970:2816437 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,fff00000,00000000
+ip-26-0-161-178:2815970:2816437 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-168-30:235655:236104 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-168-30:235655:236104 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-168-34:272150:272606 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-168-34:272150:272606 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-161-178:2815964:2816440 [1] NCCL INFO Setting affinity for GPU 1 to 0fff,ffffffff
+ip-26-0-161-178:2815964:2816440 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-165-59:1557272:1557722 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-165-59:1557272:1557722 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-161-178:2815967:2816438 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,fff00000,00000000
+ip-26-0-161-178:2815967:2816438 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-168-52:2722046:2722509 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-168-52:2722046:2722509 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-161-138:96569:97049 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,fff00000,00000000
+ip-26-0-161-138:96569:97049 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-161-123:32708:33162 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-161-123:32708:33162 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-165-202:153565:154022 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-165-202:153565:154022 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-165-59:1557270:1557725 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-165-59:1557270:1557725 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-165-202:153567:154023 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-165-202:153567:154023 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-165-202:153563:154024 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-165-202:153563:154024 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-161-138:96564:97046 [1] NCCL INFO Setting affinity for GPU 1 to 0fff,ffffffff
+ip-26-0-161-138:96564:97046 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-160-225:124160:124620 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-160-225:124160:124620 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-168-34:272156:272607 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-168-34:272156:272607 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-165-202:153566:154021 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-165-202:153566:154021 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-161-138:96565:97044 [2] NCCL INFO Setting affinity for GPU 2 to 0fff,ffffffff
+ip-26-0-161-138:96565:97044 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-165-59:1557273:1557723 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-165-59:1557273:1557723 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-161-142:62883:63338 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-161-142:62883:63338 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-161-123:32709:33164 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-161-123:32709:33164 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-161-123:32707:33163 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-161-123:32707:33163 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-168-52:2722047:2722507 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-168-52:2722047:2722507 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-161-123:32706:33161 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-161-123:32706:33161 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-161-123:32706:33161 [2] NCCL INFO Trees [0] 3/-1/-1->2->1 [1] 3/-1/-1->2->1 [2] 3/-1/-1->2->1 [3] 3/-1/-1->2->1 [4] 3/-1/-1->2->1 [5] 3/-1/-1->2->1 [6] 3/-1/-1->2->1 [7] 3/-1/-1->2->1 [8] 3/-1/-1->2->1 [9] 3/-1/-1->2->1 [10] 3/-1/-1->2->1 [11] 3/-1/-1->2->1 [12] 3/-1/-1->2->1 [13] 3/-1/-1->2->1 [14] 3/-1/-1->2->1 [15] 3/-1/-1->2->1 [16] 3/-1/-1->2->1 [17] 3/-1/-1->2->1 [18] 3/-1/-1->2->1 [19] 3/-1/-1->2->1 [20] 3/-1/-1->2->1 [21] 3/-1/-1->2->1 [22] 3/-1/-1->2->1 [23] 3/-1/-1->2->1
+ip-26-0-161-123:32706:33161 [2] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-161-123:32704:33159 [0] NCCL INFO Channel 00/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-123:32704:33159 [0] NCCL INFO Channel 01/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-123:32704:33159 [0] NCCL INFO Channel 02/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-123:32704:33159 [0] NCCL INFO Channel 03/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-123:32704:33159 [0] NCCL INFO Channel 04/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-123:32710:33160 [6] NCCL INFO Trees [0] 7/-1/-1->6->5 [1] 7/-1/-1->6->5 [2] 7/-1/-1->6->5 [3] 7/-1/-1->6->5 [4] 7/-1/-1->6->5 [5] 7/-1/-1->6->5 [6] 7/-1/-1->6->5 [7] 7/-1/-1->6->5 [8] 7/-1/-1->6->5 [9] 7/-1/-1->6->5 [10] 7/-1/-1->6->5 [11] 7/-1/-1->6->5 [12] 7/-1/-1->6->5 [13] 7/-1/-1->6->5 [14] 7/-1/-1->6->5 [15] 7/-1/-1->6->5 [16] 7/-1/-1->6->5 [17] 7/-1/-1->6->5 [18] 7/-1/-1->6->5 [19] 7/-1/-1->6->5 [20] 7/-1/-1->6->5 [21] 7/-1/-1->6->5 [22] 7/-1/-1->6->5 [23] 7/-1/-1->6->5
+ip-26-0-161-123:32704:33159 [0] NCCL INFO Channel 05/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-123:32705:33166 [1] NCCL INFO Trees [0] 2/-1/-1->1->0 [1] 2/-1/-1->1->0 [2] 2/-1/-1->1->0 [3] 2/-1/-1->1->0 [4] 2/-1/-1->1->0 [5] 2/-1/-1->1->0 [6] 2/-1/-1->1->0 [7] 2/-1/-1->1->0 [8] 2/-1/-1->1->0 [9] 2/-1/-1->1->0 [10] 2/-1/-1->1->0 [11] 2/-1/-1->1->0 [12] 2/-1/-1->1->0 [13] 2/-1/-1->1->0 [14] 2/-1/-1->1->0 [15] 2/-1/-1->1->0 [16] 2/-1/-1->1->0 [17] 2/-1/-1->1->0 [18] 2/-1/-1->1->0 [19] 2/-1/-1->1->0 [20] 2/-1/-1->1->0 [21] 2/-1/-1->1->0 [22] 2/-1/-1->1->0 [23] 2/-1/-1->1->0
+ip-26-0-161-123:32704:33159 [0] NCCL INFO Channel 06/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-123:32710:33160 [6] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-161-123:32704:33159 [0] NCCL INFO Channel 07/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-123:32705:33166 [1] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-161-123:32704:33159 [0] NCCL INFO Channel 08/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-123:32709:33164 [5] NCCL INFO Trees [0] 6/-1/-1->5->4 [1] 6/-1/-1->5->4 [2] 6/-1/-1->5->4 [3] 6/-1/-1->5->4 [4] 6/-1/-1->5->4 [5] 6/-1/-1->5->4 [6] 6/-1/-1->5->4 [7] 6/-1/-1->5->4 [8] 6/-1/-1->5->4 [9] 6/-1/-1->5->4 [10] 6/-1/-1->5->4 [11] 6/-1/-1->5->4 [12] 6/-1/-1->5->4 [13] 6/-1/-1->5->4 [14] 6/-1/-1->5->4 [15] 6/-1/-1->5->4 [16] 6/-1/-1->5->4 [17] 6/-1/-1->5->4 [18] 6/-1/-1->5->4 [19] 6/-1/-1->5->4 [20] 6/-1/-1->5->4 [21] 6/-1/-1->5->4 [22] 6/-1/-1->5->4 [23] 6/-1/-1->5->4
+ip-26-0-161-123:32707:33163 [3] NCCL INFO Trees [0] 4/-1/-1->3->2 [1] 4/-1/-1->3->2 [2] 4/-1/-1->3->2 [3] 4/-1/-1->3->2 [4] 4/-1/-1->3->2 [5] 4/-1/-1->3->2 [6] 4/-1/-1->3->2 [7] 4/-1/-1->3->2 [8] 4/-1/-1->3->2 [9] 4/-1/-1->3->2 [10] 4/-1/-1->3->2 [11] 4/-1/-1->3->2 [12] 4/-1/-1->3->2 [13] 4/-1/-1->3->2 [14] 4/-1/-1->3->2 [15] 4/-1/-1->3->2 [16] 4/-1/-1->3->2 [17] 4/-1/-1->3->2 [18] 4/-1/-1->3->2 [19] 4/-1/-1->3->2 [20] 4/-1/-1->3->2 [21] 4/-1/-1->3->2 [22] 4/-1/-1->3->2 [23] 4/-1/-1->3->2
+ip-26-0-161-123:32704:33159 [0] NCCL INFO Channel 09/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-123:32708:33162 [4] NCCL INFO Trees [0] 5/-1/-1->4->3 [1] 5/-1/-1->4->3 [2] 5/-1/-1->4->3 [3] 5/-1/-1->4->3 [4] 5/-1/-1->4->3 [5] 5/-1/-1->4->3 [6] 5/-1/-1->4->3 [7] 5/-1/-1->4->3 [8] 5/-1/-1->4->3 [9] 5/-1/-1->4->3 [10] 5/-1/-1->4->3 [11] 5/-1/-1->4->3 [12] 5/-1/-1->4->3 [13] 5/-1/-1->4->3 [14] 5/-1/-1->4->3 [15] 5/-1/-1->4->3 [16] 5/-1/-1->4->3 [17] 5/-1/-1->4->3 [18] 5/-1/-1->4->3 [19] 5/-1/-1->4->3 [20] 5/-1/-1->4->3 [21] 5/-1/-1->4->3 [22] 5/-1/-1->4->3 [23] 5/-1/-1->4->3
+ip-26-0-161-123:32709:33164 [5] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-161-123:32704:33159 [0] NCCL INFO Channel 10/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-123:32707:33163 [3] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-161-123:32708:33162 [4] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-161-123:32704:33159 [0] NCCL INFO Channel 11/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-123:32711:33165 [7] NCCL INFO Trees [0] -1/-1/-1->7->6 [1] -1/-1/-1->7->6 [2] -1/-1/-1->7->6 [3] -1/-1/-1->7->6 [4] -1/-1/-1->7->6 [5] -1/-1/-1->7->6 [6] -1/-1/-1->7->6 [7] -1/-1/-1->7->6 [8] -1/-1/-1->7->6 [9] -1/-1/-1->7->6 [10] -1/-1/-1->7->6 [11] -1/-1/-1->7->6 [12] -1/-1/-1->7->6 [13] -1/-1/-1->7->6 [14] -1/-1/-1->7->6 [15] -1/-1/-1->7->6 [16] -1/-1/-1->7->6 [17] -1/-1/-1->7->6 [18] -1/-1/-1->7->6 [19] -1/-1/-1->7->6 [20] -1/-1/-1->7->6 [21] -1/-1/-1->7->6 [22] -1/-1/-1->7->6 [23] -1/-1/-1->7->6
+ip-26-0-161-123:32704:33159 [0] NCCL INFO Channel 12/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-123:32704:33159 [0] NCCL INFO Channel 13/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-123:32704:33159 [0] NCCL INFO Channel 14/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-123:32704:33159 [0] NCCL INFO Channel 15/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-123:32704:33159 [0] NCCL INFO Channel 16/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-123:32711:33165 [7] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-161-123:32704:33159 [0] NCCL INFO Channel 17/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-123:32704:33159 [0] NCCL INFO Channel 18/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-123:32704:33159 [0] NCCL INFO Channel 19/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-123:32704:33159 [0] NCCL INFO Channel 20/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-123:32704:33159 [0] NCCL INFO Channel 21/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-123:32704:33159 [0] NCCL INFO Channel 22/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-123:32704:33159 [0] NCCL INFO Channel 23/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-123:32704:33159 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] 1/-1/-1->0->-1 [7] 1/-1/-1->0->-1 [8] 1/-1/-1->0->-1 [9] 1/-1/-1->0->-1 [10] 1/-1/-1->0->-1 [11] 1/-1/-1->0->-1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] 1/-1/-1->0->-1 [19] 1/-1/-1->0->-1 [20] 1/-1/-1->0->-1 [21] 1/-1/-1->0->-1 [22] 1/-1/-1->0->-1 [23] 1/-1/-1->0->-1
+ip-26-0-161-123:32704:33159 [0] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-165-202:153562:154017 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-165-202:153562:154017 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-165-202:153562:154017 [0] NCCL INFO Channel 00/24 :    0   1   2   3   4   5   6   7
+ip-26-0-165-202:153562:154017 [0] NCCL INFO Channel 01/24 :    0   1   2   3   4   5   6   7
+ip-26-0-165-202:153562:154017 [0] NCCL INFO Channel 02/24 :    0   1   2   3   4   5   6   7
+ip-26-0-165-202:153562:154017 [0] NCCL INFO Channel 03/24 :    0   1   2   3   4   5   6   7
+ip-26-0-165-202:153562:154017 [0] NCCL INFO Channel 04/24 :    0   1   2   3   4   5   6   7
+ip-26-0-165-202:153562:154017 [0] NCCL INFO Channel 05/24 :    0   1   2   3   4   5   6   7
+ip-26-0-165-202:153562:154017 [0] NCCL INFO Channel 06/24 :    0   1   2   3   4   5   6   7
+ip-26-0-165-202:153562:154017 [0] NCCL INFO Channel 07/24 :    0   1   2   3   4   5   6   7
+ip-26-0-165-202:153562:154017 [0] NCCL INFO Channel 08/24 :    0   1   2   3   4   5   6   7
+ip-26-0-165-202:153562:154017 [0] NCCL INFO Channel 09/24 :    0   1   2   3   4   5   6   7
+ip-26-0-165-202:153562:154017 [0] NCCL INFO Channel 10/24 :    0   1   2   3   4   5   6   7
+ip-26-0-165-202:153562:154017 [0] NCCL INFO Channel 11/24 :    0   1   2   3   4   5   6   7
+ip-26-0-165-202:153562:154017 [0] NCCL INFO Channel 12/24 :    0   1   2   3   4   5   6   7
+ip-26-0-165-202:153562:154017 [0] NCCL INFO Channel 13/24 :    0   1   2   3   4   5   6   7
+ip-26-0-165-202:153562:154017 [0] NCCL INFO Channel 14/24 :    0   1   2   3   4   5   6   7
+ip-26-0-165-202:153562:154017 [0] NCCL INFO Channel 15/24 :    0   1   2   3   4   5   6   7
+ip-26-0-165-202:153562:154017 [0] NCCL INFO Channel 16/24 :    0   1   2   3   4   5   6   7
+ip-26-0-165-202:153562:154017 [0] NCCL INFO Channel 17/24 :    0   1   2   3   4   5   6   7
+ip-26-0-165-202:153562:154017 [0] NCCL INFO Channel 18/24 :    0   1   2   3   4   5   6   7
+ip-26-0-165-202:153562:154017 [0] NCCL INFO Channel 19/24 :    0   1   2   3   4   5   6   7
+ip-26-0-165-202:153562:154017 [0] NCCL INFO Channel 20/24 :    0   1   2   3   4   5   6   7
+ip-26-0-165-202:153562:154017 [0] NCCL INFO Channel 21/24 :    0   1   2   3   4   5   6   7
+ip-26-0-165-202:153562:154017 [0] NCCL INFO Channel 22/24 :    0   1   2   3   4   5   6   7
+ip-26-0-165-202:153562:154017 [0] NCCL INFO Channel 23/24 :    0   1   2   3   4   5   6   7
+ip-26-0-165-202:153563:154024 [1] NCCL INFO Trees [0] 2/-1/-1->1->0 [1] 2/-1/-1->1->0 [2] 2/-1/-1->1->0 [3] 2/-1/-1->1->0 [4] 2/-1/-1->1->0 [5] 2/-1/-1->1->0 [6] 2/-1/-1->1->0 [7] 2/-1/-1->1->0 [8] 2/-1/-1->1->0 [9] 2/-1/-1->1->0 [10] 2/-1/-1->1->0 [11] 2/-1/-1->1->0 [12] 2/-1/-1->1->0 [13] 2/-1/-1->1->0 [14] 2/-1/-1->1->0 [15] 2/-1/-1->1->0 [16] 2/-1/-1->1->0 [17] 2/-1/-1->1->0 [18] 2/-1/-1->1->0 [19] 2/-1/-1->1->0 [20] 2/-1/-1->1->0 [21] 2/-1/-1->1->0 [22] 2/-1/-1->1->0 [23] 2/-1/-1->1->0
+ip-26-0-165-202:153569:154020 [7] NCCL INFO Trees [0] -1/-1/-1->7->6 [1] -1/-1/-1->7->6 [2] -1/-1/-1->7->6 [3] -1/-1/-1->7->6 [4] -1/-1/-1->7->6 [5] -1/-1/-1->7->6 [6] -1/-1/-1->7->6 [7] -1/-1/-1->7->6 [8] -1/-1/-1->7->6 [9] -1/-1/-1->7->6 [10] -1/-1/-1->7->6 [11] -1/-1/-1->7->6 [12] -1/-1/-1->7->6 [13] -1/-1/-1->7->6 [14] -1/-1/-1->7->6 [15] -1/-1/-1->7->6 [16] -1/-1/-1->7->6 [17] -1/-1/-1->7->6 [18] -1/-1/-1->7->6 [19] -1/-1/-1->7->6 [20] -1/-1/-1->7->6 [21] -1/-1/-1->7->6 [22] -1/-1/-1->7->6 [23] -1/-1/-1->7->6
+ip-26-0-165-202:153568:154019 [6] NCCL INFO Trees [0] 7/-1/-1->6->5 [1] 7/-1/-1->6->5 [2] 7/-1/-1->6->5 [3] 7/-1/-1->6->5 [4] 7/-1/-1->6->5 [5] 7/-1/-1->6->5 [6] 7/-1/-1->6->5 [7] 7/-1/-1->6->5 [8] 7/-1/-1->6->5 [9] 7/-1/-1->6->5 [10] 7/-1/-1->6->5 [11] 7/-1/-1->6->5 [12] 7/-1/-1->6->5 [13] 7/-1/-1->6->5 [14] 7/-1/-1->6->5 [15] 7/-1/-1->6->5 [16] 7/-1/-1->6->5 [17] 7/-1/-1->6->5 [18] 7/-1/-1->6->5 [19] 7/-1/-1->6->5 [20] 7/-1/-1->6->5 [21] 7/-1/-1->6->5 [22] 7/-1/-1->6->5 [23] 7/-1/-1->6->5
+ip-26-0-165-202:153564:154018 [2] NCCL INFO Trees [0] 3/-1/-1->2->1 [1] 3/-1/-1->2->1 [2] 3/-1/-1->2->1 [3] 3/-1/-1->2->1 [4] 3/-1/-1->2->1 [5] 3/-1/-1->2->1 [6] 3/-1/-1->2->1 [7] 3/-1/-1->2->1 [8] 3/-1/-1->2->1 [9] 3/-1/-1->2->1 [10] 3/-1/-1->2->1 [11] 3/-1/-1->2->1 [12] 3/-1/-1->2->1 [13] 3/-1/-1->2->1 [14] 3/-1/-1->2->1 [15] 3/-1/-1->2->1 [16] 3/-1/-1->2->1 [17] 3/-1/-1->2->1 [18] 3/-1/-1->2->1 [19] 3/-1/-1->2->1 [20] 3/-1/-1->2->1 [21] 3/-1/-1->2->1 [22] 3/-1/-1->2->1 [23] 3/-1/-1->2->1
+ip-26-0-165-202:153569:154020 [7] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-165-202:153565:154022 [3] NCCL INFO Trees [0] 4/-1/-1->3->2 [1] 4/-1/-1->3->2 [2] 4/-1/-1->3->2 [3] 4/-1/-1->3->2 [4] 4/-1/-1->3->2 [5] 4/-1/-1->3->2 [6] 4/-1/-1->3->2 [7] 4/-1/-1->3->2 [8] 4/-1/-1->3->2 [9] 4/-1/-1->3->2 [10] 4/-1/-1->3->2 [11] 4/-1/-1->3->2 [12] 4/-1/-1->3->2 [13] 4/-1/-1->3->2 [14] 4/-1/-1->3->2 [15] 4/-1/-1->3->2 [16] 4/-1/-1->3->2 [17] 4/-1/-1->3->2 [18] 4/-1/-1->3->2 [19] 4/-1/-1->3->2 [20] 4/-1/-1->3->2 [21] 4/-1/-1->3->2 [22] 4/-1/-1->3->2 [23] 4/-1/-1->3->2
+ip-26-0-165-202:153563:154024 [1] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-165-202:153568:154019 [6] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-165-202:153562:154017 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] 1/-1/-1->0->-1 [7] 1/-1/-1->0->-1 [8] 1/-1/-1->0->-1 [9] 1/-1/-1->0->-1 [10] 1/-1/-1->0->-1 [11] 1/-1/-1->0->-1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] 1/-1/-1->0->-1 [19] 1/-1/-1->0->-1 [20] 1/-1/-1->0->-1 [21] 1/-1/-1->0->-1 [22] 1/-1/-1->0->-1 [23] 1/-1/-1->0->-1
+ip-26-0-165-202:153564:154018 [2] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-165-202:153566:154021 [4] NCCL INFO Trees [0] 5/-1/-1->4->3 [1] 5/-1/-1->4->3 [2] 5/-1/-1->4->3 [3] 5/-1/-1->4->3 [4] 5/-1/-1->4->3 [5] 5/-1/-1->4->3 [6] 5/-1/-1->4->3 [7] 5/-1/-1->4->3 [8] 5/-1/-1->4->3 [9] 5/-1/-1->4->3 [10] 5/-1/-1->4->3 [11] 5/-1/-1->4->3 [12] 5/-1/-1->4->3 [13] 5/-1/-1->4->3 [14] 5/-1/-1->4->3 [15] 5/-1/-1->4->3 [16] 5/-1/-1->4->3 [17] 5/-1/-1->4->3 [18] 5/-1/-1->4->3 [19] 5/-1/-1->4->3 [20] 5/-1/-1->4->3 [21] 5/-1/-1->4->3 [22] 5/-1/-1->4->3 [23] 5/-1/-1->4->3
+ip-26-0-165-202:153562:154017 [0] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-165-202:153567:154023 [5] NCCL INFO Trees [0] 6/-1/-1->5->4 [1] 6/-1/-1->5->4 [2] 6/-1/-1->5->4 [3] 6/-1/-1->5->4 [4] 6/-1/-1->5->4 [5] 6/-1/-1->5->4 [6] 6/-1/-1->5->4 [7] 6/-1/-1->5->4 [8] 6/-1/-1->5->4 [9] 6/-1/-1->5->4 [10] 6/-1/-1->5->4 [11] 6/-1/-1->5->4 [12] 6/-1/-1->5->4 [13] 6/-1/-1->5->4 [14] 6/-1/-1->5->4 [15] 6/-1/-1->5->4 [16] 6/-1/-1->5->4 [17] 6/-1/-1->5->4 [18] 6/-1/-1->5->4 [19] 6/-1/-1->5->4 [20] 6/-1/-1->5->4 [21] 6/-1/-1->5->4 [22] 6/-1/-1->5->4 [23] 6/-1/-1->5->4
+ip-26-0-165-202:153565:154022 [3] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-165-202:153566:154021 [4] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-165-202:153567:154023 [5] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-161-178:2815966:2816439 [3] NCCL INFO Setting affinity for GPU 3 to 0fff,ffffffff
+ip-26-0-161-178:2815966:2816439 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-161-78:114512:114968 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-161-78:114512:114968 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-165-59:1557268:1557728 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-165-59:1557268:1557728 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-165-59:1557272:1557722 [6] NCCL INFO Trees [0] 7/-1/-1->6->5 [1] 7/-1/-1->6->5 [2] 7/-1/-1->6->5 [3] 7/-1/-1->6->5 [4] 7/-1/-1->6->5 [5] 7/-1/-1->6->5 [6] 7/-1/-1->6->5 [7] 7/-1/-1->6->5 [8] 7/-1/-1->6->5 [9] 7/-1/-1->6->5 [10] 7/-1/-1->6->5 [11] 7/-1/-1->6->5 [12] 7/-1/-1->6->5 [13] 7/-1/-1->6->5 [14] 7/-1/-1->6->5 [15] 7/-1/-1->6->5 [16] 7/-1/-1->6->5 [17] 7/-1/-1->6->5 [18] 7/-1/-1->6->5 [19] 7/-1/-1->6->5 [20] 7/-1/-1->6->5 [21] 7/-1/-1->6->5 [22] 7/-1/-1->6->5 [23] 7/-1/-1->6->5
+ip-26-0-165-59:1557271:1557726 [5] NCCL INFO Trees [0] 6/-1/-1->5->4 [1] 6/-1/-1->5->4 [2] 6/-1/-1->5->4 [3] 6/-1/-1->5->4 [4] 6/-1/-1->5->4 [5] 6/-1/-1->5->4 [6] 6/-1/-1->5->4 [7] 6/-1/-1->5->4 [8] 6/-1/-1->5->4 [9] 6/-1/-1->5->4 [10] 6/-1/-1->5->4 [11] 6/-1/-1->5->4 [12] 6/-1/-1->5->4 [13] 6/-1/-1->5->4 [14] 6/-1/-1->5->4 [15] 6/-1/-1->5->4 [16] 6/-1/-1->5->4 [17] 6/-1/-1->5->4 [18] 6/-1/-1->5->4 [19] 6/-1/-1->5->4 [20] 6/-1/-1->5->4 [21] 6/-1/-1->5->4 [22] 6/-1/-1->5->4 [23] 6/-1/-1->5->4
+ip-26-0-165-59:1557272:1557722 [6] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-165-59:1557271:1557726 [5] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-165-59:1557266:1557721 [0] NCCL INFO Channel 00/24 :    0   1   2   3   4   5   6   7
+ip-26-0-165-59:1557273:1557723 [7] NCCL INFO Trees [0] -1/-1/-1->7->6 [1] -1/-1/-1->7->6 [2] -1/-1/-1->7->6 [3] -1/-1/-1->7->6 [4] -1/-1/-1->7->6 [5] -1/-1/-1->7->6 [6] -1/-1/-1->7->6 [7] -1/-1/-1->7->6 [8] -1/-1/-1->7->6 [9] -1/-1/-1->7->6 [10] -1/-1/-1->7->6 [11] -1/-1/-1->7->6 [12] -1/-1/-1->7->6 [13] -1/-1/-1->7->6 [14] -1/-1/-1->7->6 [15] -1/-1/-1->7->6 [16] -1/-1/-1->7->6 [17] -1/-1/-1->7->6 [18] -1/-1/-1->7->6 [19] -1/-1/-1->7->6 [20] -1/-1/-1->7->6 [21] -1/-1/-1->7->6 [22] -1/-1/-1->7->6 [23] -1/-1/-1->7->6
+ip-26-0-165-59:1557266:1557721 [0] NCCL INFO Channel 01/24 :    0   1   2   3   4   5   6   7
+ip-26-0-165-59:1557273:1557723 [7] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-165-59:1557269:1557724 [3] NCCL INFO Trees [0] 4/-1/-1->3->2 [1] 4/-1/-1->3->2 [2] 4/-1/-1->3->2 [3] 4/-1/-1->3->2 [4] 4/-1/-1->3->2 [5] 4/-1/-1->3->2 [6] 4/-1/-1->3->2 [7] 4/-1/-1->3->2 [8] 4/-1/-1->3->2 [9] 4/-1/-1->3->2 [10] 4/-1/-1->3->2 [11] 4/-1/-1->3->2 [12] 4/-1/-1->3->2 [13] 4/-1/-1->3->2 [14] 4/-1/-1->3->2 [15] 4/-1/-1->3->2 [16] 4/-1/-1->3->2 [17] 4/-1/-1->3->2 [18] 4/-1/-1->3->2 [19] 4/-1/-1->3->2 [20] 4/-1/-1->3->2 [21] 4/-1/-1->3->2 [22] 4/-1/-1->3->2 [23] 4/-1/-1->3->2
+ip-26-0-165-59:1557270:1557725 [4] NCCL INFO Trees [0] 5/-1/-1->4->3 [1] 5/-1/-1->4->3 [2] 5/-1/-1->4->3 [3] 5/-1/-1->4->3 [4] 5/-1/-1->4->3 [5] 5/-1/-1->4->3 [6] 5/-1/-1->4->3 [7] 5/-1/-1->4->3 [8] 5/-1/-1->4->3 [9] 5/-1/-1->4->3 [10] 5/-1/-1->4->3 [11] 5/-1/-1->4->3 [12] 5/-1/-1->4->3 [13] 5/-1/-1->4->3 [14] 5/-1/-1->4->3 [15] 5/-1/-1->4->3 [16] 5/-1/-1->4->3 [17] 5/-1/-1->4->3 [18] 5/-1/-1->4->3 [19] 5/-1/-1->4->3 [20] 5/-1/-1->4->3 [21] 5/-1/-1->4->3 [22] 5/-1/-1->4->3 [23] 5/-1/-1->4->3
+ip-26-0-165-59:1557266:1557721 [0] NCCL INFO Channel 02/24 :    0   1   2   3   4   5   6   7
+ip-26-0-165-59:1557268:1557728 [2] NCCL INFO Trees [0] 3/-1/-1->2->1 [1] 3/-1/-1->2->1 [2] 3/-1/-1->2->1 [3] 3/-1/-1->2->1 [4] 3/-1/-1->2->1 [5] 3/-1/-1->2->1 [6] 3/-1/-1->2->1 [7] 3/-1/-1->2->1 [8] 3/-1/-1->2->1 [9] 3/-1/-1->2->1 [10] 3/-1/-1->2->1 [11] 3/-1/-1->2->1 [12] 3/-1/-1->2->1 [13] 3/-1/-1->2->1 [14] 3/-1/-1->2->1 [15] 3/-1/-1->2->1 [16] 3/-1/-1->2->1 [17] 3/-1/-1->2->1 [18] 3/-1/-1->2->1 [19] 3/-1/-1->2->1 [20] 3/-1/-1->2->1 [21] 3/-1/-1->2->1 [22] 3/-1/-1->2->1 [23] 3/-1/-1->2->1
+ip-26-0-165-59:1557267:1557727 [1] NCCL INFO Trees [0] 2/-1/-1->1->0 [1] 2/-1/-1->1->0 [2] 2/-1/-1->1->0 [3] 2/-1/-1->1->0 [4] 2/-1/-1->1->0 [5] 2/-1/-1->1->0 [6] 2/-1/-1->1->0 [7] 2/-1/-1->1->0 [8] 2/-1/-1->1->0 [9] 2/-1/-1->1->0 [10] 2/-1/-1->1->0 [11] 2/-1/-1->1->0 [12] 2/-1/-1->1->0 [13] 2/-1/-1->1->0 [14] 2/-1/-1->1->0 [15] 2/-1/-1->1->0 [16] 2/-1/-1->1->0 [17] 2/-1/-1->1->0 [18] 2/-1/-1->1->0 [19] 2/-1/-1->1->0 [20] 2/-1/-1->1->0 [21] 2/-1/-1->1->0 [22] 2/-1/-1->1->0 [23] 2/-1/-1->1->0
+ip-26-0-165-59:1557269:1557724 [3] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-165-59:1557266:1557721 [0] NCCL INFO Channel 03/24 :    0   1   2   3   4   5   6   7
+ip-26-0-165-59:1557268:1557728 [2] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-165-59:1557270:1557725 [4] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-165-59:1557266:1557721 [0] NCCL INFO Channel 04/24 :    0   1   2   3   4   5   6   7
+ip-26-0-165-59:1557267:1557727 [1] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-165-59:1557266:1557721 [0] NCCL INFO Channel 05/24 :    0   1   2   3   4   5   6   7
+ip-26-0-165-59:1557266:1557721 [0] NCCL INFO Channel 06/24 :    0   1   2   3   4   5   6   7
+ip-26-0-165-59:1557266:1557721 [0] NCCL INFO Channel 07/24 :    0   1   2   3   4   5   6   7
+ip-26-0-165-59:1557266:1557721 [0] NCCL INFO Channel 08/24 :    0   1   2   3   4   5   6   7
+ip-26-0-165-59:1557266:1557721 [0] NCCL INFO Channel 09/24 :    0   1   2   3   4   5   6   7
+ip-26-0-165-59:1557266:1557721 [0] NCCL INFO Channel 10/24 :    0   1   2   3   4   5   6   7
+ip-26-0-165-59:1557266:1557721 [0] NCCL INFO Channel 11/24 :    0   1   2   3   4   5   6   7
+ip-26-0-165-59:1557266:1557721 [0] NCCL INFO Channel 12/24 :    0   1   2   3   4   5   6   7
+ip-26-0-165-59:1557266:1557721 [0] NCCL INFO Channel 13/24 :    0   1   2   3   4   5   6   7
+ip-26-0-165-59:1557266:1557721 [0] NCCL INFO Channel 14/24 :    0   1   2   3   4   5   6   7
+ip-26-0-165-59:1557266:1557721 [0] NCCL INFO Channel 15/24 :    0   1   2   3   4   5   6   7
+ip-26-0-165-59:1557266:1557721 [0] NCCL INFO Channel 16/24 :    0   1   2   3   4   5   6   7
+ip-26-0-165-59:1557266:1557721 [0] NCCL INFO Channel 17/24 :    0   1   2   3   4   5   6   7
+ip-26-0-165-59:1557266:1557721 [0] NCCL INFO Channel 18/24 :    0   1   2   3   4   5   6   7
+ip-26-0-165-59:1557266:1557721 [0] NCCL INFO Channel 19/24 :    0   1   2   3   4   5   6   7
+ip-26-0-165-59:1557266:1557721 [0] NCCL INFO Channel 20/24 :    0   1   2   3   4   5   6   7
+ip-26-0-165-59:1557266:1557721 [0] NCCL INFO Channel 21/24 :    0   1   2   3   4   5   6   7
+ip-26-0-165-59:1557266:1557721 [0] NCCL INFO Channel 22/24 :    0   1   2   3   4   5   6   7
+ip-26-0-165-59:1557266:1557721 [0] NCCL INFO Channel 23/24 :    0   1   2   3   4   5   6   7
+ip-26-0-165-59:1557266:1557721 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] 1/-1/-1->0->-1 [7] 1/-1/-1->0->-1 [8] 1/-1/-1->0->-1 [9] 1/-1/-1->0->-1 [10] 1/-1/-1->0->-1 [11] 1/-1/-1->0->-1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] 1/-1/-1->0->-1 [19] 1/-1/-1->0->-1 [20] 1/-1/-1->0->-1 [21] 1/-1/-1->0->-1 [22] 1/-1/-1->0->-1 [23] 1/-1/-1->0->-1
+ip-26-0-165-59:1557266:1557721 [0] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-161-78:114519:114970 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-161-78:114519:114970 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-161-178:2815965:2816435 [2] NCCL INFO Setting affinity for GPU 2 to 0fff,ffffffff
+ip-26-0-161-178:2815965:2816435 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-161-78:114516:114975 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-161-78:114516:114975 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-161-138:96563:97045 [0] NCCL INFO Setting affinity for GPU 0 to 0fff,ffffffff
+ip-26-0-161-138:96563:97045 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-161-178:2815969:2816436 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,fff00000,00000000
+ip-26-0-161-178:2815969:2816436 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-165-202:153568:154019 [6] NCCL INFO Channel 00/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-202:153568:154019 [6] NCCL INFO Channel 01/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-202:153568:154019 [6] NCCL INFO Channel 02/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-142:62885:63342 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-165-202:153563:154024 [1] NCCL INFO Channel 00/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-202:153568:154019 [6] NCCL INFO Channel 03/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-123:32708:33162 [4] NCCL INFO Channel 00/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-142:62885:63342 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-161-123:32706:33161 [2] NCCL INFO Channel 00/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-123:32704:33159 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-123:32711:33165 [7] NCCL INFO Channel 00/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-202:153566:154021 [4] NCCL INFO Channel 00/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-123:32707:33163 [3] NCCL INFO Channel 00/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-202:153563:154024 [1] NCCL INFO Channel 01/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-78:114513:114973 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-161-123:32709:33164 [5] NCCL INFO Channel 00/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-123:32710:33160 [6] NCCL INFO Channel 00/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-123:32708:33162 [4] NCCL INFO Channel 01/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-202:153568:154019 [6] NCCL INFO Channel 04/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-202:153567:154023 [5] NCCL INFO Channel 00/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-123:32706:33161 [2] NCCL INFO Channel 01/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-123:32704:33159 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-123:32711:33165 [7] NCCL INFO Channel 01/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-202:153566:154021 [4] NCCL INFO Channel 01/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-123:32707:33163 [3] NCCL INFO Channel 01/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-202:153563:154024 [1] NCCL INFO Channel 02/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-123:32709:33164 [5] NCCL INFO Channel 01/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-78:114513:114973 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-161-123:32710:33160 [6] NCCL INFO Channel 01/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-123:32708:33162 [4] NCCL INFO Channel 02/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-202:153568:154019 [6] NCCL INFO Channel 05/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-123:32706:33161 [2] NCCL INFO Channel 02/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-202:153567:154023 [5] NCCL INFO Channel 01/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-123:32704:33159 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-123:32711:33165 [7] NCCL INFO Channel 02/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-123:32707:33163 [3] NCCL INFO Channel 02/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-202:153566:154021 [4] NCCL INFO Channel 02/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-202:153564:154018 [2] NCCL INFO Channel 00/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-123:32709:33164 [5] NCCL INFO Channel 02/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-123:32710:33160 [6] NCCL INFO Channel 02/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-202:153563:154024 [1] NCCL INFO Channel 03/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-123:32706:33161 [2] NCCL INFO Channel 03/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-202:153568:154019 [6] NCCL INFO Channel 06/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-202:153567:154023 [5] NCCL INFO Channel 02/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-123:32704:33159 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-178:2815968:2816434 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,fff00000,00000000
+ip-26-0-161-178:2815968:2816434 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-161-123:32705:33166 [1] NCCL INFO Channel 00/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-123:32709:33164 [5] NCCL INFO Channel 03/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-123:32710:33160 [6] NCCL INFO Channel 03/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-123:32706:33161 [2] NCCL INFO Channel 04/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-178:2815968:2816434 [5] NCCL INFO Trees [0] 6/-1/-1->5->4 [1] 6/-1/-1->5->4 [2] 6/-1/-1->5->4 [3] 6/-1/-1->5->4 [4] 6/-1/-1->5->4 [5] 6/-1/-1->5->4 [6] 6/-1/-1->5->4 [7] 6/-1/-1->5->4 [8] 6/-1/-1->5->4 [9] 6/-1/-1->5->4 [10] 6/-1/-1->5->4 [11] 6/-1/-1->5->4 [12] 6/-1/-1->5->4 [13] 6/-1/-1->5->4 [14] 6/-1/-1->5->4 [15] 6/-1/-1->5->4 [16] 6/-1/-1->5->4 [17] 6/-1/-1->5->4 [18] 6/-1/-1->5->4 [19] 6/-1/-1->5->4 [20] 6/-1/-1->5->4 [21] 6/-1/-1->5->4 [22] 6/-1/-1->5->4 [23] 6/-1/-1->5->4
+ip-26-0-161-178:2815968:2816434 [5] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-161-178:2815969:2816436 [6] NCCL INFO Trees [0] 7/-1/-1->6->5 [1] 7/-1/-1->6->5 [2] 7/-1/-1->6->5 [3] 7/-1/-1->6->5 [4] 7/-1/-1->6->5 [5] 7/-1/-1->6->5 [6] 7/-1/-1->6->5 [7] 7/-1/-1->6->5 [8] 7/-1/-1->6->5 [9] 7/-1/-1->6->5 [10] 7/-1/-1->6->5 [11] 7/-1/-1->6->5 [12] 7/-1/-1->6->5 [13] 7/-1/-1->6->5 [14] 7/-1/-1->6->5 [15] 7/-1/-1->6->5 [16] 7/-1/-1->6->5 [17] 7/-1/-1->6->5 [18] 7/-1/-1->6->5 [19] 7/-1/-1->6->5 [20] 7/-1/-1->6->5 [21] 7/-1/-1->6->5 [22] 7/-1/-1->6->5 [23] 7/-1/-1->6->5
+ip-26-0-161-178:2815969:2816436 [6] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-161-178:2815963:2816433 [0] NCCL INFO Channel 00/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-178:2815967:2816438 [4] NCCL INFO Trees [0] 5/-1/-1->4->3 [1] 5/-1/-1->4->3 [2] 5/-1/-1->4->3 [3] 5/-1/-1->4->3 [4] 5/-1/-1->4->3 [5] 5/-1/-1->4->3 [6] 5/-1/-1->4->3 [7] 5/-1/-1->4->3 [8] 5/-1/-1->4->3 [9] 5/-1/-1->4->3 [10] 5/-1/-1->4->3 [11] 5/-1/-1->4->3 [12] 5/-1/-1->4->3 [13] 5/-1/-1->4->3 [14] 5/-1/-1->4->3 [15] 5/-1/-1->4->3 [16] 5/-1/-1->4->3 [17] 5/-1/-1->4->3 [18] 5/-1/-1->4->3 [19] 5/-1/-1->4->3 [20] 5/-1/-1->4->3 [21] 5/-1/-1->4->3 [22] 5/-1/-1->4->3 [23] 5/-1/-1->4->3
+ip-26-0-161-178:2815963:2816433 [0] NCCL INFO Channel 01/24 :    0   1   2   3   4   5   6   7
+ip-26-0-165-202:153566:154021 [4] NCCL INFO Channel 03/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-178:2815967:2816438 [4] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-161-178:2815963:2816433 [0] NCCL INFO Channel 02/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-178:2815963:2816433 [0] NCCL INFO Channel 03/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-178:2815970:2816437 [7] NCCL INFO Trees [0] -1/-1/-1->7->6 [1] -1/-1/-1->7->6 [2] -1/-1/-1->7->6 [3] -1/-1/-1->7->6 [4] -1/-1/-1->7->6 [5] -1/-1/-1->7->6 [6] -1/-1/-1->7->6 [7] -1/-1/-1->7->6 [8] -1/-1/-1->7->6 [9] -1/-1/-1->7->6 [10] -1/-1/-1->7->6 [11] -1/-1/-1->7->6 [12] -1/-1/-1->7->6 [13] -1/-1/-1->7->6 [14] -1/-1/-1->7->6 [15] -1/-1/-1->7->6 [16] -1/-1/-1->7->6 [17] -1/-1/-1->7->6 [18] -1/-1/-1->7->6 [19] -1/-1/-1->7->6 [20] -1/-1/-1->7->6 [21] -1/-1/-1->7->6 [22] -1/-1/-1->7->6 [23] -1/-1/-1->7->6
+ip-26-0-161-178:2815965:2816435 [2] NCCL INFO Trees [0] 3/-1/-1->2->1 [1] 3/-1/-1->2->1 [2] 3/-1/-1->2->1 [3] 3/-1/-1->2->1 [4] 3/-1/-1->2->1 [5] 3/-1/-1->2->1 [6] 3/-1/-1->2->1 [7] 3/-1/-1->2->1 [8] 3/-1/-1->2->1 [9] 3/-1/-1->2->1 [10] 3/-1/-1->2->1 [11] 3/-1/-1->2->1 [12] 3/-1/-1->2->1 [13] 3/-1/-1->2->1 [14] 3/-1/-1->2->1 [15] 3/-1/-1->2->1 [16] 3/-1/-1->2->1 [17] 3/-1/-1->2->1 [18] 3/-1/-1->2->1 [19] 3/-1/-1->2->1 [20] 3/-1/-1->2->1 [21] 3/-1/-1->2->1 [22] 3/-1/-1->2->1 [23] 3/-1/-1->2->1
+ip-26-0-161-178:2815963:2816433 [0] NCCL INFO Channel 04/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-178:2815970:2816437 [7] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-161-178:2815963:2816433 [0] NCCL INFO Channel 05/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-178:2815965:2816435 [2] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-161-178:2815964:2816440 [1] NCCL INFO Trees [0] 2/-1/-1->1->0 [1] 2/-1/-1->1->0 [2] 2/-1/-1->1->0 [3] 2/-1/-1->1->0 [4] 2/-1/-1->1->0 [5] 2/-1/-1->1->0 [6] 2/-1/-1->1->0 [7] 2/-1/-1->1->0 [8] 2/-1/-1->1->0 [9] 2/-1/-1->1->0 [10] 2/-1/-1->1->0 [11] 2/-1/-1->1->0 [12] 2/-1/-1->1->0 [13] 2/-1/-1->1->0 [14] 2/-1/-1->1->0 [15] 2/-1/-1->1->0 [16] 2/-1/-1->1->0 [17] 2/-1/-1->1->0 [18] 2/-1/-1->1->0 [19] 2/-1/-1->1->0 [20] 2/-1/-1->1->0 [21] 2/-1/-1->1->0 [22] 2/-1/-1->1->0 [23] 2/-1/-1->1->0
+ip-26-0-161-178:2815963:2816433 [0] NCCL INFO Channel 06/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-178:2815966:2816439 [3] NCCL INFO Trees [0] 4/-1/-1->3->2 [1] 4/-1/-1->3->2 [2] 4/-1/-1->3->2 [3] 4/-1/-1->3->2 [4] 4/-1/-1->3->2 [5] 4/-1/-1->3->2 [6] 4/-1/-1->3->2 [7] 4/-1/-1->3->2 [8] 4/-1/-1->3->2 [9] 4/-1/-1->3->2 [10] 4/-1/-1->3->2 [11] 4/-1/-1->3->2 [12] 4/-1/-1->3->2 [13] 4/-1/-1->3->2 [14] 4/-1/-1->3->2 [15] 4/-1/-1->3->2 [16] 4/-1/-1->3->2 [17] 4/-1/-1->3->2 [18] 4/-1/-1->3->2 [19] 4/-1/-1->3->2 [20] 4/-1/-1->3->2 [21] 4/-1/-1->3->2 [22] 4/-1/-1->3->2 [23] 4/-1/-1->3->2
+ip-26-0-161-178:2815963:2816433 [0] NCCL INFO Channel 07/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-178:2815963:2816433 [0] NCCL INFO Channel 08/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-178:2815963:2816433 [0] NCCL INFO Channel 09/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-178:2815963:2816433 [0] NCCL INFO Channel 10/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-178:2815963:2816433 [0] NCCL INFO Channel 11/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-178:2815963:2816433 [0] NCCL INFO Channel 12/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-178:2815963:2816433 [0] NCCL INFO Channel 13/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-178:2815963:2816433 [0] NCCL INFO Channel 14/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-178:2815964:2816440 [1] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-161-178:2815963:2816433 [0] NCCL INFO Channel 15/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-178:2815963:2816433 [0] NCCL INFO Channel 16/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-178:2815963:2816433 [0] NCCL INFO Channel 17/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-178:2815963:2816433 [0] NCCL INFO Channel 18/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-178:2815963:2816433 [0] NCCL INFO Channel 19/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-178:2815963:2816433 [0] NCCL INFO Channel 20/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-178:2815963:2816433 [0] NCCL INFO Channel 21/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-178:2815963:2816433 [0] NCCL INFO Channel 22/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-178:2815963:2816433 [0] NCCL INFO Channel 23/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-178:2815963:2816433 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] 1/-1/-1->0->-1 [7] 1/-1/-1->0->-1 [8] 1/-1/-1->0->-1 [9] 1/-1/-1->0->-1 [10] 1/-1/-1->0->-1 [11] 1/-1/-1->0->-1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] 1/-1/-1->0->-1 [19] 1/-1/-1->0->-1 [20] 1/-1/-1->0->-1 [21] 1/-1/-1->0->-1 [22] 1/-1/-1->0->-1 [23] 1/-1/-1->0->-1
+ip-26-0-161-178:2815966:2816439 [3] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-161-178:2815963:2816433 [0] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-165-202:153564:154018 [2] NCCL INFO Channel 01/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-202:153563:154024 [1] NCCL INFO Channel 04/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-123:32704:33159 [0] NCCL INFO Channel 04/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-202:153568:154019 [6] NCCL INFO Channel 07/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-59:1557267:1557727 [1] NCCL INFO Channel 00/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-202:153567:154023 [5] NCCL INFO Channel 03/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-123:32707:33163 [3] NCCL INFO Channel 03/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-123:32709:33164 [5] NCCL INFO Channel 04/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-123:32705:33166 [1] NCCL INFO Channel 01/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-123:32706:33161 [2] NCCL INFO Channel 05/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-123:32710:33160 [6] NCCL INFO Channel 04/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-123:32704:33159 [0] NCCL INFO Channel 05/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-59:1557272:1557722 [6] NCCL INFO Channel 00/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-202:153566:154021 [4] NCCL INFO Channel 04/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-59:1557268:1557728 [2] NCCL INFO Channel 00/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-202:153564:154018 [2] NCCL INFO Channel 02/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-202:153563:154024 [1] NCCL INFO Channel 05/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-59:1557267:1557727 [1] NCCL INFO Channel 01/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-59:1557269:1557724 [3] NCCL INFO Channel 00/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-202:153562:154017 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-202:153568:154019 [6] NCCL INFO Channel 08/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-202:153567:154023 [5] NCCL INFO Channel 04/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-123:32707:33163 [3] NCCL INFO Channel 04/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-123:32709:33164 [5] NCCL INFO Channel 05/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-123:32706:33161 [2] NCCL INFO Channel 06/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-123:32710:33160 [6] NCCL INFO Channel 05/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-123:32705:33166 [1] NCCL INFO Channel 02/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-59:1557272:1557722 [6] NCCL INFO Channel 01/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-123:32704:33159 [0] NCCL INFO Channel 06/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-59:1557268:1557728 [2] NCCL INFO Channel 01/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-59:1557267:1557727 [1] NCCL INFO Channel 02/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-59:1557269:1557724 [3] NCCL INFO Channel 01/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-202:153566:154021 [4] NCCL INFO Channel 05/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-202:153564:154018 [2] NCCL INFO Channel 03/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-202:153563:154024 [1] NCCL INFO Channel 06/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-202:153562:154017 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-123:32708:33162 [4] NCCL INFO Channel 03/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-202:153567:154023 [5] NCCL INFO Channel 05/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-165-59:1557272:1557722 [6] NCCL INFO Channel 02/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-202:153568:154019 [6] NCCL INFO Channel 09/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-59:1557270:1557725 [4] NCCL INFO Channel 00/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-123:32711:33165 [7] NCCL INFO Channel 03/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-123:32707:33163 [3] NCCL INFO Channel 05/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-59:1557267:1557727 [1] NCCL INFO Channel 03/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-123:32709:33164 [5] NCCL INFO Channel 06/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-123:32706:33161 [2] NCCL INFO Channel 07/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-123:32710:33160 [6] NCCL INFO Channel 06/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-123:32705:33166 [1] NCCL INFO Channel 03/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-123:32704:33159 [0] NCCL INFO Channel 07/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-202:153565:154022 [3] NCCL INFO Channel 00/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-59:1557272:1557722 [6] NCCL INFO Channel 03/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-59:1557267:1557727 [1] NCCL INFO Channel 04/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-202:153566:154021 [4] NCCL INFO Channel 06/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-59:1557269:1557724 [3] NCCL INFO Channel 02/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-202:153564:154018 [2] NCCL INFO Channel 04/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-202:153563:154024 [1] NCCL INFO Channel 07/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-123:32708:33162 [4] NCCL INFO Channel 04/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-202:153562:154017 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-202:153567:154023 [5] NCCL INFO Channel 06/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-165-202:153568:154019 [6] NCCL INFO Channel 10/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-59:1557272:1557722 [6] NCCL INFO Channel 04/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-59:1557267:1557727 [1] NCCL INFO Channel 05/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-59:1557269:1557724 [3] NCCL INFO Channel 03/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-123:32707:33163 [3] NCCL INFO Channel 06/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-123:32711:33165 [7] NCCL INFO Channel 04/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-123:32709:33164 [5] NCCL INFO Channel 07/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-123:32706:33161 [2] NCCL INFO Channel 08/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-123:32705:33166 [1] NCCL INFO Channel 04/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-123:32710:33160 [6] NCCL INFO Channel 07/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-123:32704:33159 [0] NCCL INFO Channel 08/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-202:153565:154022 [3] NCCL INFO Channel 01/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-202:153566:154021 [4] NCCL INFO Channel 07/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-202:153564:154018 [2] NCCL INFO Channel 05/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-202:153563:154024 [1] NCCL INFO Channel 08/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-123:32708:33162 [4] NCCL INFO Channel 05/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-59:1557270:1557725 [4] NCCL INFO Channel 01/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-59:1557272:1557722 [6] NCCL INFO Channel 05/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-59:1557267:1557727 [1] NCCL INFO Channel 06/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-202:153562:154017 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-59:1557269:1557724 [3] NCCL INFO Channel 04/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-59:1557268:1557728 [2] NCCL INFO Channel 02/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-202:153567:154023 [5] NCCL INFO Channel 07/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-165-202:153568:154019 [6] NCCL INFO Channel 11/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-123:32711:33165 [7] NCCL INFO Channel 05/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-123:32707:33163 [3] NCCL INFO Channel 07/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-123:32706:33161 [2] NCCL INFO Channel 09/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-123:32709:33164 [5] NCCL INFO Channel 08/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-123:32705:33166 [1] NCCL INFO Channel 05/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-123:32710:33160 [6] NCCL INFO Channel 08/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-123:32704:33159 [0] NCCL INFO Channel 09/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-202:153565:154022 [3] NCCL INFO Channel 02/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-123:32708:33162 [4] NCCL INFO Channel 06/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-202:153566:154021 [4] NCCL INFO Channel 08/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-202:153564:154018 [2] NCCL INFO Channel 06/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-59:1557272:1557722 [6] NCCL INFO Channel 06/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-59:1557267:1557727 [1] NCCL INFO Channel 07/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-59:1557270:1557725 [4] NCCL INFO Channel 02/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-202:153563:154024 [1] NCCL INFO Channel 09/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-59:1557269:1557724 [3] NCCL INFO Channel 05/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-59:1557268:1557728 [2] NCCL INFO Channel 03/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-202:153562:154017 [0] NCCL INFO Channel 04/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-202:153567:154023 [5] NCCL INFO Channel 08/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-165-202:153568:154019 [6] NCCL INFO Channel 12/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-59:1557266:1557721 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-123:32711:33165 [7] NCCL INFO Channel 06/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-123:32707:33163 [3] NCCL INFO Channel 08/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-123:32706:33161 [2] NCCL INFO Channel 10/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-123:32709:33164 [5] NCCL INFO Channel 09/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-123:32705:33166 [1] NCCL INFO Channel 06/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-123:32710:33160 [6] NCCL INFO Channel 09/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-123:32704:33159 [0] NCCL INFO Channel 10/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-202:153565:154022 [3] NCCL INFO Channel 03/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-123:32708:33162 [4] NCCL INFO Channel 07/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-202:153566:154021 [4] NCCL INFO Channel 09/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-202:153564:154018 [2] NCCL INFO Channel 07/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-202:153563:154024 [1] NCCL INFO Channel 10/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-202:153567:154023 [5] NCCL INFO Channel 09/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-165-202:153562:154017 [0] NCCL INFO Channel 05/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-202:153568:154019 [6] NCCL INFO Channel 13/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-59:1557270:1557725 [4] NCCL INFO Channel 03/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-59:1557269:1557724 [3] NCCL INFO Channel 06/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-123:32711:33165 [7] NCCL INFO Channel 07/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-59:1557272:1557722 [6] NCCL INFO Channel 07/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-123:32707:33163 [3] NCCL INFO Channel 09/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-123:32709:33164 [5] NCCL INFO Channel 10/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-123:32706:33161 [2] NCCL INFO Channel 11/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-123:32705:33166 [1] NCCL INFO Channel 07/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-123:32704:33159 [0] NCCL INFO Channel 11/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-202:153565:154022 [3] NCCL INFO Channel 04/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-123:32710:33160 [6] NCCL INFO Channel 10/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-123:32708:33162 [4] NCCL INFO Channel 08/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-202:153566:154021 [4] NCCL INFO Channel 10/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-202:153564:154018 [2] NCCL INFO Channel 08/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-59:1557266:1557721 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-202:153563:154024 [1] NCCL INFO Channel 11/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-202:153567:154023 [5] NCCL INFO Channel 10/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-165-202:153569:154020 [7] NCCL INFO Channel 00/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-202:153562:154017 [0] NCCL INFO Channel 06/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-202:153568:154019 [6] NCCL INFO Channel 14/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-59:1557270:1557725 [4] NCCL INFO Channel 04/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-59:1557269:1557724 [3] NCCL INFO Channel 07/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-59:1557272:1557722 [6] NCCL INFO Channel 08/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-59:1557273:1557723 [7] NCCL INFO Channel 00/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-59:1557266:1557721 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-59:1557268:1557728 [2] NCCL INFO Channel 04/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-123:32707:33163 [3] NCCL INFO Channel 10/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-123:32709:33164 [5] NCCL INFO Channel 11/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-123:32711:33165 [7] NCCL INFO Channel 08/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-123:32706:33161 [2] NCCL INFO Channel 12/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-202:153565:154022 [3] NCCL INFO Channel 05/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-123:32705:33166 [1] NCCL INFO Channel 08/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-123:32704:33159 [0] NCCL INFO Channel 12/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-123:32710:33160 [6] NCCL INFO Channel 11/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-123:32708:33162 [4] NCCL INFO Channel 09/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-202:153566:154021 [4] NCCL INFO Channel 11/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-202:153564:154018 [2] NCCL INFO Channel 09/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-59:1557270:1557725 [4] NCCL INFO Channel 05/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-59:1557269:1557724 [3] NCCL INFO Channel 08/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-202:153563:154024 [1] NCCL INFO Channel 12/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-59:1557273:1557723 [7] NCCL INFO Channel 01/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-59:1557272:1557722 [6] NCCL INFO Channel 09/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-202:153569:154020 [7] NCCL INFO Channel 01/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-202:153567:154023 [5] NCCL INFO Channel 11/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-165-202:153562:154017 [0] NCCL INFO Channel 07/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-202:153568:154019 [6] NCCL INFO Channel 15/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-59:1557266:1557721 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-59:1557268:1557728 [2] NCCL INFO Channel 05/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-202:153565:154022 [3] NCCL INFO Channel 06/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-123:32707:33163 [3] NCCL INFO Channel 11/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-59:1557270:1557725 [4] NCCL INFO Channel 06/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-123:32711:33165 [7] NCCL INFO Channel 09/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-123:32709:33164 [5] NCCL INFO Channel 12/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-123:32706:33161 [2] NCCL INFO Channel 13/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-59:1557269:1557724 [3] NCCL INFO Channel 09/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-59:1557273:1557723 [7] NCCL INFO Channel 02/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-59:1557272:1557722 [6] NCCL INFO Channel 10/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-123:32705:33166 [1] NCCL INFO Channel 09/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-123:32704:33159 [0] NCCL INFO Channel 13/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-123:32710:33160 [6] NCCL INFO Channel 12/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-123:32708:33162 [4] NCCL INFO Channel 10/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-202:153564:154018 [2] NCCL INFO Channel 10/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-202:153566:154021 [4] NCCL INFO Channel 12/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-59:1557266:1557721 [0] NCCL INFO Channel 04/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-59:1557268:1557728 [2] NCCL INFO Channel 06/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-202:153563:154024 [1] NCCL INFO Channel 13/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-202:153569:154020 [7] NCCL INFO Channel 02/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-202:153567:154023 [5] NCCL INFO Channel 12/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-165-202:153562:154017 [0] NCCL INFO Channel 08/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-202:153568:154019 [6] NCCL INFO Channel 16/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-59:1557270:1557725 [4] NCCL INFO Channel 07/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-59:1557269:1557724 [3] NCCL INFO Channel 10/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-59:1557273:1557723 [7] NCCL INFO Channel 03/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-59:1557272:1557722 [6] NCCL INFO Channel 11/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-202:153565:154022 [3] NCCL INFO Channel 07/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-59:1557266:1557721 [0] NCCL INFO Channel 05/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-59:1557268:1557728 [2] NCCL INFO Channel 07/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-123:32707:33163 [3] NCCL INFO Channel 12/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-123:32709:33164 [5] NCCL INFO Channel 13/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-123:32711:33165 [7] NCCL INFO Channel 10/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-123:32706:33161 [2] NCCL INFO Channel 14/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-123:32705:33166 [1] NCCL INFO Channel 10/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-123:32704:33159 [0] NCCL INFO Channel 14/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-123:32710:33160 [6] NCCL INFO Channel 13/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-123:32708:33162 [4] NCCL INFO Channel 11/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-202:153564:154018 [2] NCCL INFO Channel 11/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-202:153566:154021 [4] NCCL INFO Channel 13/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-202:153563:154024 [1] NCCL INFO Channel 14/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-202:153569:154020 [7] NCCL INFO Channel 03/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-202:153567:154023 [5] NCCL INFO Channel 13/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-165-202:153562:154017 [0] NCCL INFO Channel 09/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-59:1557270:1557725 [4] NCCL INFO Channel 08/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-59:1557273:1557723 [7] NCCL INFO Channel 04/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-202:153568:154019 [6] NCCL INFO Channel 17/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-59:1557269:1557724 [3] NCCL INFO Channel 11/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-59:1557272:1557722 [6] NCCL INFO Channel 12/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-59:1557266:1557721 [0] NCCL INFO Channel 06/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-59:1557268:1557728 [2] NCCL INFO Channel 08/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-202:153565:154022 [3] NCCL INFO Channel 08/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-123:32707:33163 [3] NCCL INFO Channel 13/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-123:32709:33164 [5] NCCL INFO Channel 14/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-123:32711:33165 [7] NCCL INFO Channel 11/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-123:32706:33161 [2] NCCL INFO Channel 15/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-123:32705:33166 [1] NCCL INFO Channel 11/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-123:32704:33159 [0] NCCL INFO Channel 15/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-123:32710:33160 [6] NCCL INFO Channel 14/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-59:1557267:1557727 [1] NCCL INFO Channel 08/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-123:32708:33162 [4] NCCL INFO Channel 12/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-202:153564:154018 [2] NCCL INFO Channel 12/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-202:153566:154021 [4] NCCL INFO Channel 14/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-202:153563:154024 [1] NCCL INFO Channel 15/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-59:1557270:1557725 [4] NCCL INFO Channel 09/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-59:1557273:1557723 [7] NCCL INFO Channel 05/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-202:153569:154020 [7] NCCL INFO Channel 04/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-202:153567:154023 [5] NCCL INFO Channel 14/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-165-59:1557272:1557722 [6] NCCL INFO Channel 13/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-59:1557269:1557724 [3] NCCL INFO Channel 12/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-202:153562:154017 [0] NCCL INFO Channel 10/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-59:1557266:1557721 [0] NCCL INFO Channel 07/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-202:153568:154019 [6] NCCL INFO Channel 18/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-59:1557268:1557728 [2] NCCL INFO Channel 09/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-202:153565:154022 [3] NCCL INFO Channel 09/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-59:1557267:1557727 [1] NCCL INFO Channel 09/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-202:153564:154018 [2] NCCL INFO Channel 13/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-202:153566:154021 [4] NCCL INFO Channel 15/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-202:153563:154024 [1] NCCL INFO Channel 16/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-202:153569:154020 [7] NCCL INFO Channel 05/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-202:153567:154023 [5] NCCL INFO Channel 15/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-165-59:1557270:1557725 [4] NCCL INFO Channel 10/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-59:1557273:1557723 [7] NCCL INFO Channel 06/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-59:1557272:1557722 [6] NCCL INFO Channel 14/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-59:1557269:1557724 [3] NCCL INFO Channel 13/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-59:1557266:1557721 [0] NCCL INFO Channel 08/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-59:1557268:1557728 [2] NCCL INFO Channel 10/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-202:153562:154017 [0] NCCL INFO Channel 11/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-202:153568:154019 [6] NCCL INFO Channel 19/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-123:32708:33162 [4] NCCL INFO Channel 13/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-202:153565:154022 [3] NCCL INFO Channel 10/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-123:32707:33163 [3] NCCL INFO Channel 14/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-123:32709:33164 [5] NCCL INFO Channel 15/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-123:32711:33165 [7] NCCL INFO Channel 12/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-123:32706:33161 [2] NCCL INFO Channel 16/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-59:1557267:1557727 [1] NCCL INFO Channel 10/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-202:153566:154021 [4] NCCL INFO Channel 16/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-202:153564:154018 [2] NCCL INFO Channel 14/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-202:153563:154024 [1] NCCL INFO Channel 17/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-202:153569:154020 [7] NCCL INFO Channel 06/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-59:1557273:1557723 [7] NCCL INFO Channel 07/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-202:153567:154023 [5] NCCL INFO Channel 16/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-165-59:1557270:1557725 [4] NCCL INFO Channel 11/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-59:1557272:1557722 [6] NCCL INFO Channel 15/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-59:1557269:1557724 [3] NCCL INFO Channel 14/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-59:1557266:1557721 [0] NCCL INFO Channel 09/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-59:1557268:1557728 [2] NCCL INFO Channel 11/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-202:153562:154017 [0] NCCL INFO Channel 12/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-123:32708:33162 [4] NCCL INFO Channel 14/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-123:32707:33163 [3] NCCL INFO Channel 15/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-123:32709:33164 [5] NCCL INFO Channel 16/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-123:32711:33165 [7] NCCL INFO Channel 13/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-123:32706:33161 [2] NCCL INFO Channel 17/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-202:153568:154019 [6] NCCL INFO Channel 20/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-202:153565:154022 [3] NCCL INFO Channel 11/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-123:32705:33166 [1] NCCL INFO Channel 12/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-202:153566:154021 [4] NCCL INFO Channel 17/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-202:153564:154018 [2] NCCL INFO Channel 15/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-202:153563:154024 [1] NCCL INFO Channel 18/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-59:1557267:1557727 [1] NCCL INFO Channel 11/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-202:153569:154020 [7] NCCL INFO Channel 07/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-59:1557273:1557723 [7] NCCL INFO Channel 08/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-202:153567:154023 [5] NCCL INFO Channel 17/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-165-59:1557270:1557725 [4] NCCL INFO Channel 12/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-59:1557272:1557722 [6] NCCL INFO Channel 16/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-59:1557269:1557724 [3] NCCL INFO Channel 15/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-59:1557266:1557721 [0] NCCL INFO Channel 10/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-59:1557268:1557728 [2] NCCL INFO Channel 12/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-202:153562:154017 [0] NCCL INFO Channel 13/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-202:153568:154019 [6] NCCL INFO Channel 21/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-202:153565:154022 [3] NCCL INFO Channel 12/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-123:32707:33163 [3] NCCL INFO Channel 16/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-123:32704:33159 [0] NCCL INFO Channel 16/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-123:32710:33160 [6] NCCL INFO Channel 15/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-123:32709:33164 [5] NCCL INFO Channel 17/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-123:32707:33163 [3] NCCL INFO Channel 17/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-202:153564:154018 [2] NCCL INFO Channel 16/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-202:153566:154021 [4] NCCL INFO Channel 18/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-202:153563:154024 [1] NCCL INFO Channel 19/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-123:32704:33159 [0] NCCL INFO Channel 17/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-123:32709:33164 [5] NCCL INFO Channel 18/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-165-59:1557267:1557727 [1] NCCL INFO Channel 12/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-202:153569:154020 [7] NCCL INFO Channel 08/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-202:153567:154023 [5] NCCL INFO Channel 18/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-123:32710:33160 [6] NCCL INFO Channel 16/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-59:1557273:1557723 [7] NCCL INFO Channel 09/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-59:1557270:1557725 [4] NCCL INFO Channel 13/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-59:1557272:1557722 [6] NCCL INFO Channel 17/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-59:1557269:1557724 [3] NCCL INFO Channel 16/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-59:1557266:1557721 [0] NCCL INFO Channel 11/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-59:1557268:1557728 [2] NCCL INFO Channel 13/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-123:32711:33165 [7] NCCL INFO Channel 14/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-123:32706:33161 [2] NCCL INFO Channel 18/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-202:153562:154017 [0] NCCL INFO Channel 14/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-202:153565:154022 [3] NCCL INFO Channel 13/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-202:153568:154019 [6] NCCL INFO Channel 22/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-202:153564:154018 [2] NCCL INFO Channel 17/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-59:1557267:1557727 [1] NCCL INFO Channel 13/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-202:153566:154021 [4] NCCL INFO Channel 19/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-202:153569:154020 [7] NCCL INFO Channel 09/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-59:1557273:1557723 [7] NCCL INFO Channel 10/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-202:153563:154024 [1] NCCL INFO Channel 20/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-202:153567:154023 [5] NCCL INFO Channel 19/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-165-59:1557270:1557725 [4] NCCL INFO Channel 14/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-59:1557272:1557722 [6] NCCL INFO Channel 18/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-59:1557269:1557724 [3] NCCL INFO Channel 17/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-59:1557268:1557728 [2] NCCL INFO Channel 14/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-59:1557266:1557721 [0] NCCL INFO Channel 12/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-202:153562:154017 [0] NCCL INFO Channel 15/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-202:153565:154022 [3] NCCL INFO Channel 14/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-202:153568:154019 [6] NCCL INFO Channel 23/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-168-34:272155:272609 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-168-34:272155:272609 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-165-202:153564:154018 [2] NCCL INFO Channel 18/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-59:1557267:1557727 [1] NCCL INFO Channel 14/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-59:1557273:1557723 [7] NCCL INFO Channel 11/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-202:153566:154021 [4] NCCL INFO Channel 20/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-202:153569:154020 [7] NCCL INFO Channel 10/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-202:153563:154024 [1] NCCL INFO Channel 21/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-59:1557270:1557725 [4] NCCL INFO Channel 15/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-59:1557272:1557722 [6] NCCL INFO Channel 19/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-59:1557269:1557724 [3] NCCL INFO Channel 18/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-59:1557266:1557721 [0] NCCL INFO Channel 13/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-202:153567:154023 [5] NCCL INFO Channel 20/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-165-59:1557268:1557728 [2] NCCL INFO Channel 15/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-202:153562:154017 [0] NCCL INFO Channel 16/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-202:153565:154022 [3] NCCL INFO Channel 15/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-123:32708:33162 [4] NCCL INFO Channel 15/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-123:32705:33166 [1] NCCL INFO Channel 13/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-123:32704:33159 [0] NCCL INFO Channel 18/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-202:153564:154018 [2] NCCL INFO Channel 19/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-202:153566:154021 [4] NCCL INFO Channel 21/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-202:153569:154020 [7] NCCL INFO Channel 11/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-123:32711:33165 [7] NCCL INFO Channel 15/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-123:32708:33162 [4] NCCL INFO Channel 16/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-123:32705:33166 [1] NCCL INFO Channel 14/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-202:153563:154024 [1] NCCL INFO Channel 22/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-59:1557267:1557727 [1] NCCL INFO Channel 15/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-59:1557273:1557723 [7] NCCL INFO Channel 12/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-202:153567:154023 [5] NCCL INFO Channel 21/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-165-59:1557270:1557725 [4] NCCL INFO Channel 16/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-59:1557272:1557722 [6] NCCL INFO Channel 20/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-59:1557266:1557721 [0] NCCL INFO Channel 14/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-59:1557269:1557724 [3] NCCL INFO Channel 19/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-59:1557268:1557728 [2] NCCL INFO Channel 16/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-202:153562:154017 [0] NCCL INFO Channel 17/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-202:153565:154022 [3] NCCL INFO Channel 16/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-123:32704:33159 [0] NCCL INFO Channel 19/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-123:32709:33164 [5] NCCL INFO Channel 19/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-123:32705:33166 [1] NCCL INFO Channel 15/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-123:32711:33165 [7] NCCL INFO Channel 16/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-123:32708:33162 [4] NCCL INFO Channel 17/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-202:153564:154018 [2] NCCL INFO Channel 20/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-202:153569:154020 [7] NCCL INFO Channel 12/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-202:153566:154021 [4] NCCL INFO Channel 22/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-202:153563:154024 [1] NCCL INFO Channel 23/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-202:153567:154023 [5] NCCL INFO Channel 22/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-165-59:1557267:1557727 [1] NCCL INFO Channel 16/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-59:1557273:1557723 [7] NCCL INFO Channel 13/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-202:153562:154017 [0] NCCL INFO Channel 18/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-123:32704:33159 [0] NCCL INFO Channel 20/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-59:1557270:1557725 [4] NCCL INFO Channel 17/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-59:1557266:1557721 [0] NCCL INFO Channel 15/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-59:1557272:1557722 [6] NCCL INFO Channel 21/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-59:1557268:1557728 [2] NCCL INFO Channel 17/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-59:1557269:1557724 [3] NCCL INFO Channel 20/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-202:153565:154022 [3] NCCL INFO Channel 17/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-123:32709:33164 [5] NCCL INFO Channel 20/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-123:32705:33166 [1] NCCL INFO Channel 16/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-123:32711:33165 [7] NCCL INFO Channel 17/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-123:32708:33162 [4] NCCL INFO Channel 18/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-178:2815965:2816435 [2] NCCL INFO Channel 00/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-178:2815968:2816434 [5] NCCL INFO Channel 00/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-123:32704:33159 [0] NCCL INFO Channel 21/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-123:32709:33164 [5] NCCL INFO Channel 21/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-123:32705:33166 [1] NCCL INFO Channel 17/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-123:32711:33165 [7] NCCL INFO Channel 18/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-123:32708:33162 [4] NCCL INFO Channel 19/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-59:1557273:1557723 [7] NCCL INFO Channel 14/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-59:1557267:1557727 [1] NCCL INFO Channel 17/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-59:1557270:1557725 [4] NCCL INFO Channel 18/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-59:1557266:1557721 [0] NCCL INFO Channel 16/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-59:1557272:1557722 [6] NCCL INFO Channel 22/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-59:1557268:1557728 [2] NCCL INFO Channel 18/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-59:1557269:1557724 [3] NCCL INFO Channel 21/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-202:153569:154020 [7] NCCL INFO Channel 13/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-202:153566:154021 [4] NCCL INFO Channel 23/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-59:1557271:1557726 [5] NCCL INFO Channel 00/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-178:2815965:2816435 [2] NCCL INFO Channel 01/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-123:32704:33159 [0] NCCL INFO Channel 22/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-123:32709:33164 [5] NCCL INFO Channel 22/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-123:32705:33166 [1] NCCL INFO Channel 18/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-202:153562:154017 [0] NCCL INFO Channel 19/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-123:32711:33165 [7] NCCL INFO Channel 19/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-178:2815968:2816434 [5] NCCL INFO Channel 01/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-123:32708:33162 [4] NCCL INFO Channel 20/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-202:153569:154020 [7] NCCL INFO Channel 14/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-202:153562:154017 [0] NCCL INFO Channel 20/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-202:153565:154022 [3] NCCL INFO Channel 18/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-202:153569:154020 [7] NCCL INFO Channel 15/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-123:32710:33160 [6] NCCL INFO Channel 17/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-59:1557273:1557723 [7] NCCL INFO Channel 15/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-59:1557267:1557727 [1] NCCL INFO Channel 18/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-59:1557270:1557725 [4] NCCL INFO Channel 19/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-59:1557266:1557721 [0] NCCL INFO Channel 17/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-59:1557272:1557722 [6] NCCL INFO Channel 23/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-59:1557269:1557724 [3] NCCL INFO Channel 22/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-59:1557268:1557728 [2] NCCL INFO Channel 19/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-178:2815965:2816435 [2] NCCL INFO Channel 02/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-123:32707:33163 [3] NCCL INFO Channel 18/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-202:153567:154023 [5] NCCL INFO Channel 23/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-165-59:1557271:1557726 [5] NCCL INFO Channel 01/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-178:2815968:2816434 [5] NCCL INFO Channel 02/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-123:32704:33159 [0] NCCL INFO Channel 23/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-202:153562:154017 [0] NCCL INFO Channel 21/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-202:153565:154022 [3] NCCL INFO Channel 19/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-123:32709:33164 [5] NCCL INFO Channel 23/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-165-202:153569:154020 [7] NCCL INFO Channel 16/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-123:32705:33166 [1] NCCL INFO Channel 19/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-123:32706:33161 [2] NCCL INFO Channel 19/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-123:32711:33165 [7] NCCL INFO Channel 20/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-123:32708:33162 [4] NCCL INFO Channel 21/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-202:153564:154018 [2] NCCL INFO Channel 21/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-202:153562:154017 [0] NCCL INFO Channel 22/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-202:153565:154022 [3] NCCL INFO Channel 20/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-202:153569:154020 [7] NCCL INFO Channel 17/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-178:2815965:2816435 [2] NCCL INFO Channel 03/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-123:32710:33160 [6] NCCL INFO Channel 18/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-59:1557273:1557723 [7] NCCL INFO Channel 16/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-59:1557270:1557725 [4] NCCL INFO Channel 20/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-59:1557267:1557727 [1] NCCL INFO Channel 19/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-59:1557266:1557721 [0] NCCL INFO Channel 18/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-59:1557269:1557724 [3] NCCL INFO Channel 23/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-59:1557268:1557728 [2] NCCL INFO Channel 20/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-123:32707:33163 [3] NCCL INFO Channel 19/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-178:2815968:2816434 [5] NCCL INFO Channel 03/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-165-59:1557271:1557726 [5] NCCL INFO Channel 02/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-123:32705:33166 [1] NCCL INFO Channel 20/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-123:32706:33161 [2] NCCL INFO Channel 20/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-123:32711:33165 [7] NCCL INFO Channel 21/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-123:32708:33162 [4] NCCL INFO Channel 22/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-178:2815969:2816436 [6] NCCL INFO Channel 00/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-178:2815967:2816438 [4] NCCL INFO Channel 00/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-123:32710:33160 [6] NCCL INFO Channel 19/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-123:32707:33163 [3] NCCL INFO Channel 20/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-178:2815965:2816435 [2] NCCL INFO Channel 04/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-59:1557273:1557723 [7] NCCL INFO Channel 17/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-59:1557267:1557727 [1] NCCL INFO Channel 20/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-59:1557270:1557725 [4] NCCL INFO Channel 21/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-59:1557266:1557721 [0] NCCL INFO Channel 19/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-59:1557268:1557728 [2] NCCL INFO Channel 21/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-178:2815968:2816434 [5] NCCL INFO Channel 04/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-165-202:153564:154018 [2] NCCL INFO Channel 22/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-59:1557271:1557726 [5] NCCL INFO Channel 03/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-123:32706:33161 [2] NCCL INFO Channel 21/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-123:32705:33166 [1] NCCL INFO Channel 21/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-123:32711:33165 [7] NCCL INFO Channel 22/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-123:32708:33162 [4] NCCL INFO Channel 23/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-178:2815969:2816436 [6] NCCL INFO Channel 01/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-178:2815967:2816438 [4] NCCL INFO Channel 01/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-123:32710:33160 [6] NCCL INFO Channel 20/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-178:2815963:2816433 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-123:32707:33163 [3] NCCL INFO Channel 21/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-59:1557273:1557723 [7] NCCL INFO Channel 18/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-59:1557267:1557727 [1] NCCL INFO Channel 21/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-59:1557270:1557725 [4] NCCL INFO Channel 22/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-59:1557266:1557721 [0] NCCL INFO Channel 20/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-59:1557268:1557728 [2] NCCL INFO Channel 22/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-178:2815965:2816435 [2] NCCL INFO Channel 05/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-178:2815968:2816434 [5] NCCL INFO Channel 05/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-165-59:1557271:1557726 [5] NCCL INFO Channel 04/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-123:32706:33161 [2] NCCL INFO Channel 22/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-123:32711:33165 [7] NCCL INFO Channel 23/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-123:32705:33166 [1] NCCL INFO Channel 22/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-202:153565:154022 [3] NCCL INFO Channel 21/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-202:153562:154017 [0] NCCL INFO Channel 23/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-202:153569:154020 [7] NCCL INFO Channel 18/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-123:32710:33160 [6] NCCL INFO Channel 21/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-123:32707:33163 [3] NCCL INFO Channel 22/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-202:153564:154018 [2] NCCL INFO Channel 23/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-178:2815969:2816436 [6] NCCL INFO Channel 02/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-178:2815967:2816438 [4] NCCL INFO Channel 02/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-59:1557273:1557723 [7] NCCL INFO Channel 19/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-59:1557270:1557725 [4] NCCL INFO Channel 23/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-59:1557267:1557727 [1] NCCL INFO Channel 22/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-178:2815963:2816433 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-59:1557268:1557728 [2] NCCL INFO Channel 23/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-59:1557266:1557721 [0] NCCL INFO Channel 21/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-123:32706:33161 [2] NCCL INFO Channel 23/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-123:32705:33166 [1] NCCL INFO Channel 23/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-59:1557271:1557726 [5] NCCL INFO Channel 05/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-178:2815965:2816435 [2] NCCL INFO Channel 06/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-178:2815968:2816434 [5] NCCL INFO Channel 06/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-123:32707:33163 [3] NCCL INFO Channel 23/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-123:32710:33160 [6] NCCL INFO Channel 22/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-59:1557273:1557723 [7] NCCL INFO Channel 20/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-59:1557267:1557727 [1] NCCL INFO Channel 23/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-59:1557266:1557721 [0] NCCL INFO Channel 22/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-178:2815969:2816436 [6] NCCL INFO Channel 03/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-178:2815967:2816438 [4] NCCL INFO Channel 03/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-59:1557271:1557726 [5] NCCL INFO Channel 06/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-178:2815963:2816433 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-202:153565:154022 [3] NCCL INFO Channel 22/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-202:153569:154020 [7] NCCL INFO Channel 19/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-178:2815965:2816435 [2] NCCL INFO Channel 07/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-178:2815968:2816434 [5] NCCL INFO Channel 07/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-123:32710:33160 [6] NCCL INFO Channel 23/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-178:2815969:2816436 [6] NCCL INFO Channel 04/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-178:2815967:2816438 [4] NCCL INFO Channel 04/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-178:2815963:2816433 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-178:2815965:2816435 [2] NCCL INFO Channel 08/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-59:1557273:1557723 [7] NCCL INFO Channel 21/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-59:1557266:1557721 [0] NCCL INFO Channel 23/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-178:2815968:2816434 [5] NCCL INFO Channel 08/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-165-59:1557271:1557726 [5] NCCL INFO Channel 07/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-178:2815969:2816436 [6] NCCL INFO Channel 05/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-178:2815967:2816438 [4] NCCL INFO Channel 05/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-178:2815963:2816433 [0] NCCL INFO Channel 04/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-202:153565:154022 [3] NCCL INFO Channel 23/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-202:153569:154020 [7] NCCL INFO Channel 20/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-178:2815965:2816435 [2] NCCL INFO Channel 09/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-178:2815970:2816437 [7] NCCL INFO Channel 00/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-178:2815968:2816434 [5] NCCL INFO Channel 09/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-242:1364904:1365369 [2] NCCL INFO Setting affinity for GPU 2 to 0fff,ffffffff
+ip-26-0-160-242:1364904:1365369 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-161-178:2815969:2816436 [6] NCCL INFO Channel 06/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-178:2815967:2816438 [4] NCCL INFO Channel 06/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-178:2815964:2816440 [1] NCCL INFO Channel 00/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-178:2815963:2816433 [0] NCCL INFO Channel 05/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-78:114517:114969 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-161-78:114517:114969 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-161-178:2815965:2816435 [2] NCCL INFO Channel 10/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-178:2815970:2816437 [7] NCCL INFO Channel 01/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-178:2815968:2816434 [5] NCCL INFO Channel 10/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-178:2815969:2816436 [6] NCCL INFO Channel 07/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-178:2815967:2816438 [4] NCCL INFO Channel 07/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-178:2815964:2816440 [1] NCCL INFO Channel 01/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-178:2815963:2816433 [0] NCCL INFO Channel 06/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-178:2815965:2816435 [2] NCCL INFO Channel 11/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-178:2815970:2816437 [7] NCCL INFO Channel 02/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-178:2815968:2816434 [5] NCCL INFO Channel 11/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-178:2815969:2816436 [6] NCCL INFO Channel 08/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-178:2815967:2816438 [4] NCCL INFO Channel 08/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-178:2815964:2816440 [1] NCCL INFO Channel 02/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-178:2815963:2816433 [0] NCCL INFO Channel 07/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-178:2815965:2816435 [2] NCCL INFO Channel 12/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-178:2815970:2816437 [7] NCCL INFO Channel 03/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-202:153569:154020 [7] NCCL INFO Channel 21/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-178:2815968:2816434 [5] NCCL INFO Channel 12/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-225:124163:124626 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-160-225:124163:124626 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-161-178:2815966:2816439 [3] NCCL INFO Channel 00/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-178:2815969:2816436 [6] NCCL INFO Channel 09/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-178:2815967:2816438 [4] NCCL INFO Channel 09/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-178:2815964:2816440 [1] NCCL INFO Channel 03/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-178:2815963:2816433 [0] NCCL INFO Channel 08/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-178:2815965:2816435 [2] NCCL INFO Channel 13/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-178:2815970:2816437 [7] NCCL INFO Channel 04/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-178:2815968:2816434 [5] NCCL INFO Channel 13/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-165-59:1557273:1557723 [7] NCCL INFO Channel 22/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-178:2815966:2816439 [3] NCCL INFO Channel 01/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-178:2815969:2816436 [6] NCCL INFO Channel 10/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-178:2815964:2816440 [1] NCCL INFO Channel 04/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-178:2815967:2816438 [4] NCCL INFO Channel 10/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-178:2815963:2816433 [0] NCCL INFO Channel 09/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-78:114514:114971 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-161-178:2815965:2816435 [2] NCCL INFO Channel 14/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-78:114514:114971 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-161-178:2815970:2816437 [7] NCCL INFO Channel 05/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-178:2815968:2816434 [5] NCCL INFO Channel 14/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-165-59:1557271:1557726 [5] NCCL INFO Channel 08/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-178:2815966:2816439 [3] NCCL INFO Channel 02/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-178:2815964:2816440 [1] NCCL INFO Channel 05/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-178:2815969:2816436 [6] NCCL INFO Channel 11/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-178:2815967:2816438 [4] NCCL INFO Channel 11/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-178:2815963:2816433 [0] NCCL INFO Channel 10/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-178:2815965:2816435 [2] NCCL INFO Channel 15/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-178:2815970:2816437 [7] NCCL INFO Channel 06/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-178:2815968:2816434 [5] NCCL INFO Channel 15/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-178:2815966:2816439 [3] NCCL INFO Channel 03/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-178:2815964:2816440 [1] NCCL INFO Channel 06/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-178:2815969:2816436 [6] NCCL INFO Channel 12/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-178:2815967:2816438 [4] NCCL INFO Channel 12/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-178:2815963:2816433 [0] NCCL INFO Channel 11/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-178:2815965:2816435 [2] NCCL INFO Channel 16/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-178:2815970:2816437 [7] NCCL INFO Channel 07/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-178:2815968:2816434 [5] NCCL INFO Channel 16/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-78:114518:114974 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-161-78:114518:114974 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-161-78:114518:114974 [6] NCCL INFO Trees [0] 7/-1/-1->6->5 [1] 7/-1/-1->6->5 [2] 7/-1/-1->6->5 [3] 7/-1/-1->6->5 [4] 7/-1/-1->6->5 [5] 7/-1/-1->6->5 [6] 7/-1/-1->6->5 [7] 7/-1/-1->6->5 [8] 7/-1/-1->6->5 [9] 7/-1/-1->6->5 [10] 7/-1/-1->6->5 [11] 7/-1/-1->6->5 [12] 7/-1/-1->6->5 [13] 7/-1/-1->6->5 [14] 7/-1/-1->6->5 [15] 7/-1/-1->6->5 [16] 7/-1/-1->6->5 [17] 7/-1/-1->6->5 [18] 7/-1/-1->6->5 [19] 7/-1/-1->6->5 [20] 7/-1/-1->6->5 [21] 7/-1/-1->6->5 [22] 7/-1/-1->6->5 [23] 7/-1/-1->6->5
+ip-26-0-161-78:114518:114974 [6] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-161-178:2815966:2816439 [3] NCCL INFO Channel 04/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-178:2815964:2816440 [1] NCCL INFO Channel 07/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-178:2815969:2816436 [6] NCCL INFO Channel 13/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-78:114519:114970 [7] NCCL INFO Trees [0] -1/-1/-1->7->6 [1] -1/-1/-1->7->6 [2] -1/-1/-1->7->6 [3] -1/-1/-1->7->6 [4] -1/-1/-1->7->6 [5] -1/-1/-1->7->6 [6] -1/-1/-1->7->6 [7] -1/-1/-1->7->6 [8] -1/-1/-1->7->6 [9] -1/-1/-1->7->6 [10] -1/-1/-1->7->6 [11] -1/-1/-1->7->6 [12] -1/-1/-1->7->6 [13] -1/-1/-1->7->6 [14] -1/-1/-1->7->6 [15] -1/-1/-1->7->6 [16] -1/-1/-1->7->6 [17] -1/-1/-1->7->6 [18] -1/-1/-1->7->6 [19] -1/-1/-1->7->6 [20] -1/-1/-1->7->6 [21] -1/-1/-1->7->6 [22] -1/-1/-1->7->6 [23] -1/-1/-1->7->6
+ip-26-0-161-78:114519:114970 [7] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-161-78:114512:114968 [0] NCCL INFO Channel 00/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-178:2815967:2816438 [4] NCCL INFO Channel 13/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-78:114512:114968 [0] NCCL INFO Channel 01/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-78:114514:114971 [2] NCCL INFO Trees [0] 3/-1/-1->2->1 [1] 3/-1/-1->2->1 [2] 3/-1/-1->2->1 [3] 3/-1/-1->2->1 [4] 3/-1/-1->2->1 [5] 3/-1/-1->2->1 [6] 3/-1/-1->2->1 [7] 3/-1/-1->2->1 [8] 3/-1/-1->2->1 [9] 3/-1/-1->2->1 [10] 3/-1/-1->2->1 [11] 3/-1/-1->2->1 [12] 3/-1/-1->2->1 [13] 3/-1/-1->2->1 [14] 3/-1/-1->2->1 [15] 3/-1/-1->2->1 [16] 3/-1/-1->2->1 [17] 3/-1/-1->2->1 [18] 3/-1/-1->2->1 [19] 3/-1/-1->2->1 [20] 3/-1/-1->2->1 [21] 3/-1/-1->2->1 [22] 3/-1/-1->2->1 [23] 3/-1/-1->2->1
+ip-26-0-161-78:114512:114968 [0] NCCL INFO Channel 02/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-178:2815963:2816433 [0] NCCL INFO Channel 12/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-78:114513:114973 [1] NCCL INFO Trees [0] 2/-1/-1->1->0 [1] 2/-1/-1->1->0 [2] 2/-1/-1->1->0 [3] 2/-1/-1->1->0 [4] 2/-1/-1->1->0 [5] 2/-1/-1->1->0 [6] 2/-1/-1->1->0 [7] 2/-1/-1->1->0 [8] 2/-1/-1->1->0 [9] 2/-1/-1->1->0 [10] 2/-1/-1->1->0 [11] 2/-1/-1->1->0 [12] 2/-1/-1->1->0 [13] 2/-1/-1->1->0 [14] 2/-1/-1->1->0 [15] 2/-1/-1->1->0 [16] 2/-1/-1->1->0 [17] 2/-1/-1->1->0 [18] 2/-1/-1->1->0 [19] 2/-1/-1->1->0 [20] 2/-1/-1->1->0 [21] 2/-1/-1->1->0 [22] 2/-1/-1->1->0 [23] 2/-1/-1->1->0
+ip-26-0-161-78:114514:114971 [2] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-161-78:114512:114968 [0] NCCL INFO Channel 03/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-78:114516:114975 [4] NCCL INFO Trees [0] 5/-1/-1->4->3 [1] 5/-1/-1->4->3 [2] 5/-1/-1->4->3 [3] 5/-1/-1->4->3 [4] 5/-1/-1->4->3 [5] 5/-1/-1->4->3 [6] 5/-1/-1->4->3 [7] 5/-1/-1->4->3 [8] 5/-1/-1->4->3 [9] 5/-1/-1->4->3 [10] 5/-1/-1->4->3 [11] 5/-1/-1->4->3 [12] 5/-1/-1->4->3 [13] 5/-1/-1->4->3 [14] 5/-1/-1->4->3 [15] 5/-1/-1->4->3 [16] 5/-1/-1->4->3 [17] 5/-1/-1->4->3 [18] 5/-1/-1->4->3 [19] 5/-1/-1->4->3 [20] 5/-1/-1->4->3 [21] 5/-1/-1->4->3 [22] 5/-1/-1->4->3 [23] 5/-1/-1->4->3
+ip-26-0-161-78:114517:114969 [5] NCCL INFO Trees [0] 6/-1/-1->5->4 [1] 6/-1/-1->5->4 [2] 6/-1/-1->5->4 [3] 6/-1/-1->5->4 [4] 6/-1/-1->5->4 [5] 6/-1/-1->5->4 [6] 6/-1/-1->5->4 [7] 6/-1/-1->5->4 [8] 6/-1/-1->5->4 [9] 6/-1/-1->5->4 [10] 6/-1/-1->5->4 [11] 6/-1/-1->5->4 [12] 6/-1/-1->5->4 [13] 6/-1/-1->5->4 [14] 6/-1/-1->5->4 [15] 6/-1/-1->5->4 [16] 6/-1/-1->5->4 [17] 6/-1/-1->5->4 [18] 6/-1/-1->5->4 [19] 6/-1/-1->5->4 [20] 6/-1/-1->5->4 [21] 6/-1/-1->5->4 [22] 6/-1/-1->5->4 [23] 6/-1/-1->5->4
+ip-26-0-161-78:114512:114968 [0] NCCL INFO Channel 04/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-78:114513:114973 [1] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-161-78:114516:114975 [4] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-161-78:114517:114969 [5] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-161-78:114512:114968 [0] NCCL INFO Channel 05/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-78:114512:114968 [0] NCCL INFO Channel 06/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-78:114512:114968 [0] NCCL INFO Channel 07/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-78:114512:114968 [0] NCCL INFO Channel 08/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-78:114512:114968 [0] NCCL INFO Channel 09/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-78:114512:114968 [0] NCCL INFO Channel 10/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-78:114512:114968 [0] NCCL INFO Channel 11/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-78:114515:114972 [3] NCCL INFO Trees [0] 4/-1/-1->3->2 [1] 4/-1/-1->3->2 [2] 4/-1/-1->3->2 [3] 4/-1/-1->3->2 [4] 4/-1/-1->3->2 [5] 4/-1/-1->3->2 [6] 4/-1/-1->3->2 [7] 4/-1/-1->3->2 [8] 4/-1/-1->3->2 [9] 4/-1/-1->3->2 [10] 4/-1/-1->3->2 [11] 4/-1/-1->3->2 [12] 4/-1/-1->3->2 [13] 4/-1/-1->3->2 [14] 4/-1/-1->3->2 [15] 4/-1/-1->3->2 [16] 4/-1/-1->3->2 [17] 4/-1/-1->3->2 [18] 4/-1/-1->3->2 [19] 4/-1/-1->3->2 [20] 4/-1/-1->3->2 [21] 4/-1/-1->3->2 [22] 4/-1/-1->3->2 [23] 4/-1/-1->3->2
+ip-26-0-161-78:114512:114968 [0] NCCL INFO Channel 12/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-78:114512:114968 [0] NCCL INFO Channel 13/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-78:114512:114968 [0] NCCL INFO Channel 14/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-78:114512:114968 [0] NCCL INFO Channel 15/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-78:114512:114968 [0] NCCL INFO Channel 16/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-78:114512:114968 [0] NCCL INFO Channel 17/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-78:114515:114972 [3] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-161-78:114512:114968 [0] NCCL INFO Channel 18/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-78:114512:114968 [0] NCCL INFO Channel 19/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-78:114512:114968 [0] NCCL INFO Channel 20/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-78:114512:114968 [0] NCCL INFO Channel 21/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-78:114512:114968 [0] NCCL INFO Channel 22/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-78:114512:114968 [0] NCCL INFO Channel 23/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-178:2815965:2816435 [2] NCCL INFO Channel 17/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-78:114512:114968 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] 1/-1/-1->0->-1 [7] 1/-1/-1->0->-1 [8] 1/-1/-1->0->-1 [9] 1/-1/-1->0->-1 [10] 1/-1/-1->0->-1 [11] 1/-1/-1->0->-1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] 1/-1/-1->0->-1 [19] 1/-1/-1->0->-1 [20] 1/-1/-1->0->-1 [21] 1/-1/-1->0->-1 [22] 1/-1/-1->0->-1 [23] 1/-1/-1->0->-1
+ip-26-0-161-78:114512:114968 [0] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-161-178:2815968:2816434 [5] NCCL INFO Channel 17/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-178:2815970:2816437 [7] NCCL INFO Channel 08/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-59:1557273:1557723 [7] NCCL INFO Channel 23/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-138:96566:97047 [3] NCCL INFO Setting affinity for GPU 3 to 0fff,ffffffff
+ip-26-0-161-178:2815964:2816440 [1] NCCL INFO Channel 08/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-138:96566:97047 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-161-178:2815966:2816439 [3] NCCL INFO Channel 05/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-178:2815969:2816436 [6] NCCL INFO Channel 14/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-178:2815967:2816438 [4] NCCL INFO Channel 14/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-178:2815963:2816433 [0] NCCL INFO Channel 13/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-202:153569:154020 [7] NCCL INFO Channel 22/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-178:2815965:2816435 [2] NCCL INFO Channel 18/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-178:2815968:2816434 [5] NCCL INFO Channel 18/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-178:2815970:2816437 [7] NCCL INFO Channel 09/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-59:1557271:1557726 [5] NCCL INFO Channel 09/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-178:2815964:2816440 [1] NCCL INFO Channel 09/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-178:2815966:2816439 [3] NCCL INFO Channel 06/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-178:2815969:2816436 [6] NCCL INFO Channel 15/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-178:2815967:2816438 [4] NCCL INFO Channel 15/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-178:2815963:2816433 [0] NCCL INFO Channel 14/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-178:2815965:2816435 [2] NCCL INFO Channel 19/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-178:2815968:2816434 [5] NCCL INFO Channel 19/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-178:2815970:2816437 [7] NCCL INFO Channel 10/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-178:2815964:2816440 [1] NCCL INFO Channel 10/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-178:2815969:2816436 [6] NCCL INFO Channel 16/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-178:2815966:2816439 [3] NCCL INFO Channel 07/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-178:2815967:2816438 [4] NCCL INFO Channel 16/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-178:2815963:2816433 [0] NCCL INFO Channel 15/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-178:2815965:2816435 [2] NCCL INFO Channel 20/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-178:2815968:2816434 [5] NCCL INFO Channel 20/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-178:2815970:2816437 [7] NCCL INFO Channel 11/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-59:1557271:1557726 [5] NCCL INFO Channel 10/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-178:2815964:2816440 [1] NCCL INFO Channel 11/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-178:2815969:2816436 [6] NCCL INFO Channel 17/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-178:2815966:2816439 [3] NCCL INFO Channel 08/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-178:2815967:2816438 [4] NCCL INFO Channel 17/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-178:2815963:2816433 [0] NCCL INFO Channel 16/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-178:2815965:2816435 [2] NCCL INFO Channel 21/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-178:2815968:2816434 [5] NCCL INFO Channel 21/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-178:2815970:2816437 [7] NCCL INFO Channel 12/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-168-34:272157:272613 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-168-34:272157:272613 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-161-178:2815964:2816440 [1] NCCL INFO Channel 12/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-178:2815969:2816436 [6] NCCL INFO Channel 18/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-178:2815966:2816439 [3] NCCL INFO Channel 09/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-178:2815967:2816438 [4] NCCL INFO Channel 18/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-178:2815963:2816433 [0] NCCL INFO Channel 17/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-178:2815965:2816435 [2] NCCL INFO Channel 22/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-178:2815970:2816437 [7] NCCL INFO Channel 13/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-178:2815968:2816434 [5] NCCL INFO Channel 22/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-178:2815964:2816440 [1] NCCL INFO Channel 13/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-178:2815969:2816436 [6] NCCL INFO Channel 19/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-178:2815966:2816439 [3] NCCL INFO Channel 10/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-178:2815967:2816438 [4] NCCL INFO Channel 19/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-178:2815965:2816435 [2] NCCL INFO Channel 23/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-178:2815963:2816433 [0] NCCL INFO Channel 18/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-178:2815970:2816437 [7] NCCL INFO Channel 14/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-178:2815968:2816434 [5] NCCL INFO Channel 23/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-168-52:2722052:2722506 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-168-52:2722052:2722506 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-161-178:2815964:2816440 [1] NCCL INFO Channel 14/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-178:2815969:2816436 [6] NCCL INFO Channel 20/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-178:2815966:2816439 [3] NCCL INFO Channel 11/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-178:2815967:2816438 [4] NCCL INFO Channel 20/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-178:2815963:2816433 [0] NCCL INFO Channel 19/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-178:2815970:2816437 [7] NCCL INFO Channel 15/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-178:2815964:2816440 [1] NCCL INFO Channel 15/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-178:2815969:2816436 [6] NCCL INFO Channel 21/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-178:2815966:2816439 [3] NCCL INFO Channel 12/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-178:2815963:2816433 [0] NCCL INFO Channel 20/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-178:2815967:2816438 [4] NCCL INFO Channel 21/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-178:2815970:2816437 [7] NCCL INFO Channel 16/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-178:2815964:2816440 [1] NCCL INFO Channel 16/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-178:2815966:2816439 [3] NCCL INFO Channel 13/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-178:2815963:2816433 [0] NCCL INFO Channel 21/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-178:2815969:2816436 [6] NCCL INFO Channel 22/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-178:2815967:2816438 [4] NCCL INFO Channel 22/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-178:2815970:2816437 [7] NCCL INFO Channel 17/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-178:2815964:2816440 [1] NCCL INFO Channel 17/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-178:2815966:2816439 [3] NCCL INFO Channel 14/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-178:2815969:2816436 [6] NCCL INFO Channel 23/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-178:2815967:2816438 [4] NCCL INFO Channel 23/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-178:2815963:2816433 [0] NCCL INFO Channel 22/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-178:2815970:2816437 [7] NCCL INFO Channel 18/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-178:2815964:2816440 [1] NCCL INFO Channel 18/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-178:2815966:2816439 [3] NCCL INFO Channel 15/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-178:2815963:2816433 [0] NCCL INFO Channel 23/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-178:2815970:2816437 [7] NCCL INFO Channel 19/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-178:2815966:2816439 [3] NCCL INFO Channel 16/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-178:2815964:2816440 [1] NCCL INFO Channel 19/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-178:2815970:2816437 [7] NCCL INFO Channel 20/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-168-52:2722051:2722503 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-168-52:2722051:2722503 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-161-178:2815966:2816439 [3] NCCL INFO Channel 17/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-178:2815964:2816440 [1] NCCL INFO Channel 20/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-178:2815970:2816437 [7] NCCL INFO Channel 21/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-202:153569:154020 [7] NCCL INFO Channel 23/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-178:2815966:2816439 [3] NCCL INFO Channel 18/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-178:2815964:2816440 [1] NCCL INFO Channel 21/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-178:2815970:2816437 [7] NCCL INFO Channel 22/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-242:1364906:1365370 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,fff00000,00000000
+ip-26-0-160-242:1364906:1365370 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-161-178:2815966:2816439 [3] NCCL INFO Channel 19/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-178:2815964:2816440 [1] NCCL INFO Channel 22/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-178:2815970:2816437 [7] NCCL INFO Channel 23/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-59:1557271:1557726 [5] NCCL INFO Channel 11/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-153:2636091:2636548 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-161-153:2636091:2636548 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-161-178:2815966:2816439 [3] NCCL INFO Channel 20/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-178:2815964:2816440 [1] NCCL INFO Channel 23/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-78:114518:114974 [6] NCCL INFO Channel 00/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-78:114519:114970 [7] NCCL INFO Channel 00/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-78:114516:114975 [4] NCCL INFO Channel 00/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-78:114514:114971 [2] NCCL INFO Channel 00/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-78:114513:114973 [1] NCCL INFO Channel 00/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-78:114518:114974 [6] NCCL INFO Channel 01/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-78:114514:114971 [2] NCCL INFO Channel 01/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-78:114519:114970 [7] NCCL INFO Channel 01/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-78:114516:114975 [4] NCCL INFO Channel 01/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-168-52:2722049:2722504 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-168-52:2722049:2722504 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-161-78:114513:114973 [1] NCCL INFO Channel 01/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-78:114518:114974 [6] NCCL INFO Channel 02/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-78:114513:114973 [1] NCCL INFO Channel 02/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-78:114519:114970 [7] NCCL INFO Channel 02/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-78:114518:114974 [6] NCCL INFO Channel 03/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-138:96567:97050 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,fff00000,00000000
+ip-26-0-161-138:96567:97050 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-161-178:2815966:2816439 [3] NCCL INFO Channel 21/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-78:114514:114971 [2] NCCL INFO Channel 02/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-78:114519:114970 [7] NCCL INFO Channel 03/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-78:114513:114973 [1] NCCL INFO Channel 03/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-78:114518:114974 [6] NCCL INFO Channel 04/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-78:114514:114971 [2] NCCL INFO Channel 03/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-78:114513:114973 [1] NCCL INFO Channel 04/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-78:114519:114970 [7] NCCL INFO Channel 04/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-78:114518:114974 [6] NCCL INFO Channel 05/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-142:62886:63341 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-161-142:62886:63341 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-161-78:114514:114971 [2] NCCL INFO Channel 04/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-78:114519:114970 [7] NCCL INFO Channel 05/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-78:114513:114973 [1] NCCL INFO Channel 05/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-78:114518:114974 [6] NCCL INFO Channel 06/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-78:114516:114975 [4] NCCL INFO Channel 02/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-78:114514:114971 [2] NCCL INFO Channel 05/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-78:114519:114970 [7] NCCL INFO Channel 06/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-78:114513:114973 [1] NCCL INFO Channel 06/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-242:1364908:1365368 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,fff00000,00000000
+ip-26-0-161-78:114518:114974 [6] NCCL INFO Channel 07/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-242:1364908:1365368 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-161-78:114516:114975 [4] NCCL INFO Channel 03/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-78:114514:114971 [2] NCCL INFO Channel 06/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-78:114518:114974 [6] NCCL INFO Channel 08/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-78:114519:114970 [7] NCCL INFO Channel 07/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-78:114513:114973 [1] NCCL INFO Channel 07/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-78:114515:114972 [3] NCCL INFO Channel 00/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-138:96570:97051 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,fff00000,00000000
+ip-26-0-161-138:96570:97051 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-161-138:96570:97051 [7] NCCL INFO Trees [0] -1/-1/-1->7->6 [1] -1/-1/-1->7->6 [2] -1/-1/-1->7->6 [3] -1/-1/-1->7->6 [4] -1/-1/-1->7->6 [5] -1/-1/-1->7->6 [6] -1/-1/-1->7->6 [7] -1/-1/-1->7->6 [8] -1/-1/-1->7->6 [9] -1/-1/-1->7->6 [10] -1/-1/-1->7->6 [11] -1/-1/-1->7->6 [12] -1/-1/-1->7->6 [13] -1/-1/-1->7->6 [14] -1/-1/-1->7->6 [15] -1/-1/-1->7->6 [16] -1/-1/-1->7->6 [17] -1/-1/-1->7->6 [18] -1/-1/-1->7->6 [19] -1/-1/-1->7->6 [20] -1/-1/-1->7->6 [21] -1/-1/-1->7->6 [22] -1/-1/-1->7->6 [23] -1/-1/-1->7->6
+ip-26-0-161-138:96570:97051 [7] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-161-78:114516:114975 [4] NCCL INFO Channel 04/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-138:96568:97048 [5] NCCL INFO Trees [0] 6/-1/-1->5->4 [1] 6/-1/-1->5->4 [2] 6/-1/-1->5->4 [3] 6/-1/-1->5->4 [4] 6/-1/-1->5->4 [5] 6/-1/-1->5->4 [6] 6/-1/-1->5->4 [7] 6/-1/-1->5->4 [8] 6/-1/-1->5->4 [9] 6/-1/-1->5->4 [10] 6/-1/-1->5->4 [11] 6/-1/-1->5->4 [12] 6/-1/-1->5->4 [13] 6/-1/-1->5->4 [14] 6/-1/-1->5->4 [15] 6/-1/-1->5->4 [16] 6/-1/-1->5->4 [17] 6/-1/-1->5->4 [18] 6/-1/-1->5->4 [19] 6/-1/-1->5->4 [20] 6/-1/-1->5->4 [21] 6/-1/-1->5->4 [22] 6/-1/-1->5->4 [23] 6/-1/-1->5->4
+ip-26-0-161-138:96563:97045 [0] NCCL INFO Channel 00/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-138:96568:97048 [5] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-161-138:96563:97045 [0] NCCL INFO Channel 01/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-138:96567:97050 [4] NCCL INFO Trees [0] 5/-1/-1->4->3 [1] 5/-1/-1->4->3 [2] 5/-1/-1->4->3 [3] 5/-1/-1->4->3 [4] 5/-1/-1->4->3 [5] 5/-1/-1->4->3 [6] 5/-1/-1->4->3 [7] 5/-1/-1->4->3 [8] 5/-1/-1->4->3 [9] 5/-1/-1->4->3 [10] 5/-1/-1->4->3 [11] 5/-1/-1->4->3 [12] 5/-1/-1->4->3 [13] 5/-1/-1->4->3 [14] 5/-1/-1->4->3 [15] 5/-1/-1->4->3 [16] 5/-1/-1->4->3 [17] 5/-1/-1->4->3 [18] 5/-1/-1->4->3 [19] 5/-1/-1->4->3 [20] 5/-1/-1->4->3 [21] 5/-1/-1->4->3 [22] 5/-1/-1->4->3 [23] 5/-1/-1->4->3
+ip-26-0-161-138:96563:97045 [0] NCCL INFO Channel 02/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-138:96565:97044 [2] NCCL INFO Trees [0] 3/-1/-1->2->1 [1] 3/-1/-1->2->1 [2] 3/-1/-1->2->1 [3] 3/-1/-1->2->1 [4] 3/-1/-1->2->1 [5] 3/-1/-1->2->1 [6] 3/-1/-1->2->1 [7] 3/-1/-1->2->1 [8] 3/-1/-1->2->1 [9] 3/-1/-1->2->1 [10] 3/-1/-1->2->1 [11] 3/-1/-1->2->1 [12] 3/-1/-1->2->1 [13] 3/-1/-1->2->1 [14] 3/-1/-1->2->1 [15] 3/-1/-1->2->1 [16] 3/-1/-1->2->1 [17] 3/-1/-1->2->1 [18] 3/-1/-1->2->1 [19] 3/-1/-1->2->1 [20] 3/-1/-1->2->1 [21] 3/-1/-1->2->1 [22] 3/-1/-1->2->1 [23] 3/-1/-1->2->1
+ip-26-0-161-138:96567:97050 [4] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-161-138:96566:97047 [3] NCCL INFO Trees [0] 4/-1/-1->3->2 [1] 4/-1/-1->3->2 [2] 4/-1/-1->3->2 [3] 4/-1/-1->3->2 [4] 4/-1/-1->3->2 [5] 4/-1/-1->3->2 [6] 4/-1/-1->3->2 [7] 4/-1/-1->3->2 [8] 4/-1/-1->3->2 [9] 4/-1/-1->3->2 [10] 4/-1/-1->3->2 [11] 4/-1/-1->3->2 [12] 4/-1/-1->3->2 [13] 4/-1/-1->3->2 [14] 4/-1/-1->3->2 [15] 4/-1/-1->3->2 [16] 4/-1/-1->3->2 [17] 4/-1/-1->3->2 [18] 4/-1/-1->3->2 [19] 4/-1/-1->3->2 [20] 4/-1/-1->3->2 [21] 4/-1/-1->3->2 [22] 4/-1/-1->3->2 [23] 4/-1/-1->3->2
+ip-26-0-161-138:96563:97045 [0] NCCL INFO Channel 03/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-138:96565:97044 [2] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-161-138:96563:97045 [0] NCCL INFO Channel 04/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-138:96569:97049 [6] NCCL INFO Trees [0] 7/-1/-1->6->5 [1] 7/-1/-1->6->5 [2] 7/-1/-1->6->5 [3] 7/-1/-1->6->5 [4] 7/-1/-1->6->5 [5] 7/-1/-1->6->5 [6] 7/-1/-1->6->5 [7] 7/-1/-1->6->5 [8] 7/-1/-1->6->5 [9] 7/-1/-1->6->5 [10] 7/-1/-1->6->5 [11] 7/-1/-1->6->5 [12] 7/-1/-1->6->5 [13] 7/-1/-1->6->5 [14] 7/-1/-1->6->5 [15] 7/-1/-1->6->5 [16] 7/-1/-1->6->5 [17] 7/-1/-1->6->5 [18] 7/-1/-1->6->5 [19] 7/-1/-1->6->5 [20] 7/-1/-1->6->5 [21] 7/-1/-1->6->5 [22] 7/-1/-1->6->5 [23] 7/-1/-1->6->5
+ip-26-0-161-138:96566:97047 [3] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-161-138:96564:97046 [1] NCCL INFO Trees [0] 2/-1/-1->1->0 [1] 2/-1/-1->1->0 [2] 2/-1/-1->1->0 [3] 2/-1/-1->1->0 [4] 2/-1/-1->1->0 [5] 2/-1/-1->1->0 [6] 2/-1/-1->1->0 [7] 2/-1/-1->1->0 [8] 2/-1/-1->1->0 [9] 2/-1/-1->1->0 [10] 2/-1/-1->1->0 [11] 2/-1/-1->1->0 [12] 2/-1/-1->1->0 [13] 2/-1/-1->1->0 [14] 2/-1/-1->1->0 [15] 2/-1/-1->1->0 [16] 2/-1/-1->1->0 [17] 2/-1/-1->1->0 [18] 2/-1/-1->1->0 [19] 2/-1/-1->1->0 [20] 2/-1/-1->1->0 [21] 2/-1/-1->1->0 [22] 2/-1/-1->1->0 [23] 2/-1/-1->1->0
+ip-26-0-161-138:96563:97045 [0] NCCL INFO Channel 05/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-138:96563:97045 [0] NCCL INFO Channel 06/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-138:96563:97045 [0] NCCL INFO Channel 07/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-138:96564:97046 [1] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-161-138:96563:97045 [0] NCCL INFO Channel 08/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-138:96569:97049 [6] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-161-138:96563:97045 [0] NCCL INFO Channel 09/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-138:96563:97045 [0] NCCL INFO Channel 10/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-138:96563:97045 [0] NCCL INFO Channel 11/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-138:96563:97045 [0] NCCL INFO Channel 12/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-138:96563:97045 [0] NCCL INFO Channel 13/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-138:96563:97045 [0] NCCL INFO Channel 14/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-138:96563:97045 [0] NCCL INFO Channel 15/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-138:96563:97045 [0] NCCL INFO Channel 16/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-138:96563:97045 [0] NCCL INFO Channel 17/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-138:96563:97045 [0] NCCL INFO Channel 18/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-138:96563:97045 [0] NCCL INFO Channel 19/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-138:96563:97045 [0] NCCL INFO Channel 20/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-138:96563:97045 [0] NCCL INFO Channel 21/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-138:96563:97045 [0] NCCL INFO Channel 22/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-138:96563:97045 [0] NCCL INFO Channel 23/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-138:96563:97045 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] 1/-1/-1->0->-1 [7] 1/-1/-1->0->-1 [8] 1/-1/-1->0->-1 [9] 1/-1/-1->0->-1 [10] 1/-1/-1->0->-1 [11] 1/-1/-1->0->-1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] 1/-1/-1->0->-1 [19] 1/-1/-1->0->-1 [20] 1/-1/-1->0->-1 [21] 1/-1/-1->0->-1 [22] 1/-1/-1->0->-1 [23] 1/-1/-1->0->-1
+ip-26-0-161-138:96563:97045 [0] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-161-78:114514:114971 [2] NCCL INFO Channel 07/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-78:114519:114970 [7] NCCL INFO Channel 08/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-78:114518:114974 [6] NCCL INFO Channel 09/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-78:114513:114973 [1] NCCL INFO Channel 08/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-78:114515:114972 [3] NCCL INFO Channel 01/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-78:114516:114975 [4] NCCL INFO Channel 05/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-78:114514:114971 [2] NCCL INFO Channel 08/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-78:114519:114970 [7] NCCL INFO Channel 09/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-78:114518:114974 [6] NCCL INFO Channel 10/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-78:114513:114973 [1] NCCL INFO Channel 09/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-78:114515:114972 [3] NCCL INFO Channel 02/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-78:114512:114968 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-78:114516:114975 [4] NCCL INFO Channel 06/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-78:114514:114971 [2] NCCL INFO Channel 09/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-78:114519:114970 [7] NCCL INFO Channel 10/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-78:114513:114973 [1] NCCL INFO Channel 10/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-78:114518:114974 [6] NCCL INFO Channel 11/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-78:114515:114972 [3] NCCL INFO Channel 03/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-78:114512:114968 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-168-34:272153:272612 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-161-78:114516:114975 [4] NCCL INFO Channel 07/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-168-34:272153:272612 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-160-242:1364903:1365371 [1] NCCL INFO Setting affinity for GPU 1 to 0fff,ffffffff
+ip-26-0-161-78:114514:114971 [2] NCCL INFO Channel 10/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-78:114519:114970 [7] NCCL INFO Channel 11/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-78:114513:114973 [1] NCCL INFO Channel 11/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-78:114518:114974 [6] NCCL INFO Channel 12/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-78:114515:114972 [3] NCCL INFO Channel 04/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-78:114512:114968 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-78:114517:114969 [5] NCCL INFO Channel 00/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-78:114516:114975 [4] NCCL INFO Channel 08/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-242:1364903:1365371 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-161-78:114514:114971 [2] NCCL INFO Channel 11/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-78:114519:114970 [7] NCCL INFO Channel 12/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-78:114513:114973 [1] NCCL INFO Channel 12/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-78:114518:114974 [6] NCCL INFO Channel 13/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-78:114515:114972 [3] NCCL INFO Channel 05/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-78:114512:114968 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-78:114517:114969 [5] NCCL INFO Channel 01/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-78:114516:114975 [4] NCCL INFO Channel 09/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-168-34:272154:272608 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-168-34:272154:272608 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-168-34:272151:272611 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-161-78:114514:114971 [2] NCCL INFO Channel 12/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-78:114519:114970 [7] NCCL INFO Channel 13/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-168-34:272151:272611 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-161-78:114513:114973 [1] NCCL INFO Channel 13/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-78:114518:114974 [6] NCCL INFO Channel 14/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-78:114512:114968 [0] NCCL INFO Channel 04/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-78:114515:114972 [3] NCCL INFO Channel 06/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-78:114516:114975 [4] NCCL INFO Channel 10/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-78:114517:114969 [5] NCCL INFO Channel 02/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-165-59:1557271:1557726 [5] NCCL INFO Channel 12/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-78:114514:114971 [2] NCCL INFO Channel 13/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-78:114519:114970 [7] NCCL INFO Channel 14/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-78:114513:114973 [1] NCCL INFO Channel 14/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-78:114518:114974 [6] NCCL INFO Channel 15/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-78:114515:114972 [3] NCCL INFO Channel 07/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-78:114512:114968 [0] NCCL INFO Channel 05/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-78:114517:114969 [5] NCCL INFO Channel 03/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-78:114516:114975 [4] NCCL INFO Channel 11/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-78:114514:114971 [2] NCCL INFO Channel 14/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-78:114519:114970 [7] NCCL INFO Channel 15/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-78:114513:114973 [1] NCCL INFO Channel 15/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-78:114518:114974 [6] NCCL INFO Channel 16/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-78:114515:114972 [3] NCCL INFO Channel 08/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-78:114517:114969 [5] NCCL INFO Channel 04/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-78:114512:114968 [0] NCCL INFO Channel 06/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-78:114516:114975 [4] NCCL INFO Channel 12/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-178:2815966:2816439 [3] NCCL INFO Channel 22/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-78:114514:114971 [2] NCCL INFO Channel 15/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-78:114519:114970 [7] NCCL INFO Channel 16/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-168-34:272152:272610 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-160-242:1364905:1365372 [3] NCCL INFO Setting affinity for GPU 3 to 0fff,ffffffff
+ip-26-0-161-78:114513:114973 [1] NCCL INFO Channel 16/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-78:114515:114972 [3] NCCL INFO Channel 09/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-78:114518:114974 [6] NCCL INFO Channel 17/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-78:114517:114969 [5] NCCL INFO Channel 05/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-78:114512:114968 [0] NCCL INFO Channel 07/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-78:114516:114975 [4] NCCL INFO Channel 13/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-168-34:272152:272610 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-160-242:1364905:1365372 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-168-34:272150:272606 [0] NCCL INFO Channel 00/24 :    0   1   2   3   4   5   6   7
+ip-26-0-168-34:272150:272606 [0] NCCL INFO Channel 01/24 :    0   1   2   3   4   5   6   7
+ip-26-0-168-34:272151:272611 [1] NCCL INFO Trees [0] 2/-1/-1->1->0 [1] 2/-1/-1->1->0 [2] 2/-1/-1->1->0 [3] 2/-1/-1->1->0 [4] 2/-1/-1->1->0 [5] 2/-1/-1->1->0 [6] 2/-1/-1->1->0 [7] 2/-1/-1->1->0 [8] 2/-1/-1->1->0 [9] 2/-1/-1->1->0 [10] 2/-1/-1->1->0 [11] 2/-1/-1->1->0 [12] 2/-1/-1->1->0 [13] 2/-1/-1->1->0 [14] 2/-1/-1->1->0 [15] 2/-1/-1->1->0 [16] 2/-1/-1->1->0 [17] 2/-1/-1->1->0 [18] 2/-1/-1->1->0 [19] 2/-1/-1->1->0 [20] 2/-1/-1->1->0 [21] 2/-1/-1->1->0 [22] 2/-1/-1->1->0 [23] 2/-1/-1->1->0
+ip-26-0-168-34:272152:272610 [2] NCCL INFO Trees [0] 3/-1/-1->2->1 [1] 3/-1/-1->2->1 [2] 3/-1/-1->2->1 [3] 3/-1/-1->2->1 [4] 3/-1/-1->2->1 [5] 3/-1/-1->2->1 [6] 3/-1/-1->2->1 [7] 3/-1/-1->2->1 [8] 3/-1/-1->2->1 [9] 3/-1/-1->2->1 [10] 3/-1/-1->2->1 [11] 3/-1/-1->2->1 [12] 3/-1/-1->2->1 [13] 3/-1/-1->2->1 [14] 3/-1/-1->2->1 [15] 3/-1/-1->2->1 [16] 3/-1/-1->2->1 [17] 3/-1/-1->2->1 [18] 3/-1/-1->2->1 [19] 3/-1/-1->2->1 [20] 3/-1/-1->2->1 [21] 3/-1/-1->2->1 [22] 3/-1/-1->2->1 [23] 3/-1/-1->2->1
+ip-26-0-168-34:272150:272606 [0] NCCL INFO Channel 02/24 :    0   1   2   3   4   5   6   7
+ip-26-0-168-34:272150:272606 [0] NCCL INFO Channel 03/24 :    0   1   2   3   4   5   6   7
+ip-26-0-168-34:272151:272611 [1] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-168-34:272150:272606 [0] NCCL INFO Channel 04/24 :    0   1   2   3   4   5   6   7
+ip-26-0-168-34:272152:272610 [2] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-168-34:272150:272606 [0] NCCL INFO Channel 05/24 :    0   1   2   3   4   5   6   7
+ip-26-0-168-34:272150:272606 [0] NCCL INFO Channel 06/24 :    0   1   2   3   4   5   6   7
+ip-26-0-168-34:272150:272606 [0] NCCL INFO Channel 07/24 :    0   1   2   3   4   5   6   7
+ip-26-0-168-34:272150:272606 [0] NCCL INFO Channel 08/24 :    0   1   2   3   4   5   6   7
+ip-26-0-168-34:272150:272606 [0] NCCL INFO Channel 09/24 :    0   1   2   3   4   5   6   7
+ip-26-0-168-34:272157:272613 [7] NCCL INFO Trees [0] -1/-1/-1->7->6 [1] -1/-1/-1->7->6 [2] -1/-1/-1->7->6 [3] -1/-1/-1->7->6 [4] -1/-1/-1->7->6 [5] -1/-1/-1->7->6 [6] -1/-1/-1->7->6 [7] -1/-1/-1->7->6 [8] -1/-1/-1->7->6 [9] -1/-1/-1->7->6 [10] -1/-1/-1->7->6 [11] -1/-1/-1->7->6 [12] -1/-1/-1->7->6 [13] -1/-1/-1->7->6 [14] -1/-1/-1->7->6 [15] -1/-1/-1->7->6 [16] -1/-1/-1->7->6 [17] -1/-1/-1->7->6 [18] -1/-1/-1->7->6 [19] -1/-1/-1->7->6 [20] -1/-1/-1->7->6 [21] -1/-1/-1->7->6 [22] -1/-1/-1->7->6 [23] -1/-1/-1->7->6
+ip-26-0-168-34:272150:272606 [0] NCCL INFO Channel 10/24 :    0   1   2   3   4   5   6   7
+ip-26-0-168-34:272150:272606 [0] NCCL INFO Channel 11/24 :    0   1   2   3   4   5   6   7
+ip-26-0-168-34:272157:272613 [7] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-168-34:272150:272606 [0] NCCL INFO Channel 12/24 :    0   1   2   3   4   5   6   7
+ip-26-0-168-34:272156:272607 [6] NCCL INFO Trees [0] 7/-1/-1->6->5 [1] 7/-1/-1->6->5 [2] 7/-1/-1->6->5 [3] 7/-1/-1->6->5 [4] 7/-1/-1->6->5 [5] 7/-1/-1->6->5 [6] 7/-1/-1->6->5 [7] 7/-1/-1->6->5 [8] 7/-1/-1->6->5 [9] 7/-1/-1->6->5 [10] 7/-1/-1->6->5 [11] 7/-1/-1->6->5 [12] 7/-1/-1->6->5 [13] 7/-1/-1->6->5 [14] 7/-1/-1->6->5 [15] 7/-1/-1->6->5 [16] 7/-1/-1->6->5 [17] 7/-1/-1->6->5 [18] 7/-1/-1->6->5 [19] 7/-1/-1->6->5 [20] 7/-1/-1->6->5 [21] 7/-1/-1->6->5 [22] 7/-1/-1->6->5 [23] 7/-1/-1->6->5
+ip-26-0-168-34:272150:272606 [0] NCCL INFO Channel 13/24 :    0   1   2   3   4   5   6   7
+ip-26-0-168-34:272153:272612 [3] NCCL INFO Trees [0] 4/-1/-1->3->2 [1] 4/-1/-1->3->2 [2] 4/-1/-1->3->2 [3] 4/-1/-1->3->2 [4] 4/-1/-1->3->2 [5] 4/-1/-1->3->2 [6] 4/-1/-1->3->2 [7] 4/-1/-1->3->2 [8] 4/-1/-1->3->2 [9] 4/-1/-1->3->2 [10] 4/-1/-1->3->2 [11] 4/-1/-1->3->2 [12] 4/-1/-1->3->2 [13] 4/-1/-1->3->2 [14] 4/-1/-1->3->2 [15] 4/-1/-1->3->2 [16] 4/-1/-1->3->2 [17] 4/-1/-1->3->2 [18] 4/-1/-1->3->2 [19] 4/-1/-1->3->2 [20] 4/-1/-1->3->2 [21] 4/-1/-1->3->2 [22] 4/-1/-1->3->2 [23] 4/-1/-1->3->2
+ip-26-0-168-34:272150:272606 [0] NCCL INFO Channel 14/24 :    0   1   2   3   4   5   6   7
+ip-26-0-168-34:272156:272607 [6] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-168-34:272150:272606 [0] NCCL INFO Channel 15/24 :    0   1   2   3   4   5   6   7
+ip-26-0-168-34:272153:272612 [3] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-168-34:272150:272606 [0] NCCL INFO Channel 16/24 :    0   1   2   3   4   5   6   7
+ip-26-0-168-34:272150:272606 [0] NCCL INFO Channel 17/24 :    0   1   2   3   4   5   6   7
+ip-26-0-168-34:272150:272606 [0] NCCL INFO Channel 18/24 :    0   1   2   3   4   5   6   7
+ip-26-0-168-34:272150:272606 [0] NCCL INFO Channel 19/24 :    0   1   2   3   4   5   6   7
+ip-26-0-168-34:272150:272606 [0] NCCL INFO Channel 20/24 :    0   1   2   3   4   5   6   7
+ip-26-0-168-34:272155:272609 [5] NCCL INFO Trees [0] 6/-1/-1->5->4 [1] 6/-1/-1->5->4 [2] 6/-1/-1->5->4 [3] 6/-1/-1->5->4 [4] 6/-1/-1->5->4 [5] 6/-1/-1->5->4 [6] 6/-1/-1->5->4 [7] 6/-1/-1->5->4 [8] 6/-1/-1->5->4 [9] 6/-1/-1->5->4 [10] 6/-1/-1->5->4 [11] 6/-1/-1->5->4 [12] 6/-1/-1->5->4 [13] 6/-1/-1->5->4 [14] 6/-1/-1->5->4 [15] 6/-1/-1->5->4 [16] 6/-1/-1->5->4 [17] 6/-1/-1->5->4 [18] 6/-1/-1->5->4 [19] 6/-1/-1->5->4 [20] 6/-1/-1->5->4 [21] 6/-1/-1->5->4 [22] 6/-1/-1->5->4 [23] 6/-1/-1->5->4
+ip-26-0-168-34:272154:272608 [4] NCCL INFO Trees [0] 5/-1/-1->4->3 [1] 5/-1/-1->4->3 [2] 5/-1/-1->4->3 [3] 5/-1/-1->4->3 [4] 5/-1/-1->4->3 [5] 5/-1/-1->4->3 [6] 5/-1/-1->4->3 [7] 5/-1/-1->4->3 [8] 5/-1/-1->4->3 [9] 5/-1/-1->4->3 [10] 5/-1/-1->4->3 [11] 5/-1/-1->4->3 [12] 5/-1/-1->4->3 [13] 5/-1/-1->4->3 [14] 5/-1/-1->4->3 [15] 5/-1/-1->4->3 [16] 5/-1/-1->4->3 [17] 5/-1/-1->4->3 [18] 5/-1/-1->4->3 [19] 5/-1/-1->4->3 [20] 5/-1/-1->4->3 [21] 5/-1/-1->4->3 [22] 5/-1/-1->4->3 [23] 5/-1/-1->4->3
+ip-26-0-168-34:272150:272606 [0] NCCL INFO Channel 21/24 :    0   1   2   3   4   5   6   7
+ip-26-0-168-34:272154:272608 [4] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-168-34:272150:272606 [0] NCCL INFO Channel 22/24 :    0   1   2   3   4   5   6   7
+ip-26-0-168-34:272155:272609 [5] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-168-34:272150:272606 [0] NCCL INFO Channel 23/24 :    0   1   2   3   4   5   6   7
+ip-26-0-168-34:272150:272606 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] 1/-1/-1->0->-1 [7] 1/-1/-1->0->-1 [8] 1/-1/-1->0->-1 [9] 1/-1/-1->0->-1 [10] 1/-1/-1->0->-1 [11] 1/-1/-1->0->-1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] 1/-1/-1->0->-1 [19] 1/-1/-1->0->-1 [20] 1/-1/-1->0->-1 [21] 1/-1/-1->0->-1 [22] 1/-1/-1->0->-1 [23] 1/-1/-1->0->-1
+ip-26-0-168-34:272150:272606 [0] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-161-78:114514:114971 [2] NCCL INFO Channel 16/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-78:114519:114970 [7] NCCL INFO Channel 17/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-78:114513:114973 [1] NCCL INFO Channel 17/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-78:114515:114972 [3] NCCL INFO Channel 10/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-78:114517:114969 [5] NCCL INFO Channel 06/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-78:114512:114968 [0] NCCL INFO Channel 08/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-78:114518:114974 [6] NCCL INFO Channel 18/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-78:114516:114975 [4] NCCL INFO Channel 14/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-78:114514:114971 [2] NCCL INFO Channel 17/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-78:114519:114970 [7] NCCL INFO Channel 18/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-78:114513:114973 [1] NCCL INFO Channel 18/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-78:114515:114972 [3] NCCL INFO Channel 11/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-78:114517:114969 [5] NCCL INFO Channel 07/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-78:114516:114975 [4] NCCL INFO Channel 15/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-78:114512:114968 [0] NCCL INFO Channel 09/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-78:114518:114974 [6] NCCL INFO Channel 19/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-78:114514:114971 [2] NCCL INFO Channel 18/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-78:114519:114970 [7] NCCL INFO Channel 19/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-78:114513:114973 [1] NCCL INFO Channel 19/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-78:114515:114972 [3] NCCL INFO Channel 12/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-78:114517:114969 [5] NCCL INFO Channel 08/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-78:114516:114975 [4] NCCL INFO Channel 16/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-78:114512:114968 [0] NCCL INFO Channel 10/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-78:114518:114974 [6] NCCL INFO Channel 20/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-78:114514:114971 [2] NCCL INFO Channel 19/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-78:114519:114970 [7] NCCL INFO Channel 20/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-78:114513:114973 [1] NCCL INFO Channel 20/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-78:114517:114969 [5] NCCL INFO Channel 09/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-78:114515:114972 [3] NCCL INFO Channel 13/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-78:114516:114975 [4] NCCL INFO Channel 17/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-78:114518:114974 [6] NCCL INFO Channel 21/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-78:114512:114968 [0] NCCL INFO Channel 11/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-178:2815966:2816439 [3] NCCL INFO Channel 23/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-78:114514:114971 [2] NCCL INFO Channel 20/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-78:114513:114973 [1] NCCL INFO Channel 21/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-78:114519:114970 [7] NCCL INFO Channel 21/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-78:114517:114969 [5] NCCL INFO Channel 10/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-78:114515:114972 [3] NCCL INFO Channel 14/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-78:114516:114975 [4] NCCL INFO Channel 18/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-78:114518:114974 [6] NCCL INFO Channel 22/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-78:114512:114968 [0] NCCL INFO Channel 12/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-78:114514:114971 [2] NCCL INFO Channel 21/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-168-30:235648:236102 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-168-30:235648:236102 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-161-78:114519:114970 [7] NCCL INFO Channel 22/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-78:114513:114973 [1] NCCL INFO Channel 22/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-103:96893:97346 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-161-78:114517:114969 [5] NCCL INFO Channel 11/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-78:114516:114975 [4] NCCL INFO Channel 19/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-78:114515:114972 [3] NCCL INFO Channel 15/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-78:114512:114968 [0] NCCL INFO Channel 13/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-78:114518:114974 [6] NCCL INFO Channel 23/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-78:114514:114971 [2] NCCL INFO Channel 22/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-103:96893:97346 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-161-78:114519:114970 [7] NCCL INFO Channel 23/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-78:114513:114973 [1] NCCL INFO Channel 23/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-153:2636085:2636553 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-161-153:2636085:2636553 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-168-52:2722048:2722505 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-161-78:114517:114969 [5] NCCL INFO Channel 12/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-168-52:2722048:2722505 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-161-78:114516:114975 [4] NCCL INFO Channel 20/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-142:62881:63337 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-161-142:62881:63337 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-161-78:114516:114975 [4] NCCL INFO Channel 21/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-78:114512:114968 [0] NCCL INFO Channel 14/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-242:1364902:1365366 [0] NCCL INFO Setting affinity for GPU 0 to 0fff,ffffffff
+ip-26-0-161-103:96892:97345 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-161-103:96892:97345 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-161-78:114515:114972 [3] NCCL INFO Channel 16/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-242:1364902:1365366 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-165-59:1557271:1557726 [5] NCCL INFO Channel 13/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-78:114516:114975 [4] NCCL INFO Channel 22/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-78:114517:114969 [5] NCCL INFO Channel 13/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-78:114514:114971 [2] NCCL INFO Channel 23/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-78:114512:114968 [0] NCCL INFO Channel 15/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-142:62884:63339 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-161-142:62884:63339 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-168-52:2722045:2722502 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-168-52:2722045:2722502 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-161-138:96564:97046 [1] NCCL INFO Channel 00/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-138:96563:97045 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-138:96568:97048 [5] NCCL INFO Channel 00/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-138:96564:97046 [1] NCCL INFO Channel 01/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-138:96570:97051 [7] NCCL INFO Channel 00/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-138:96563:97045 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-138:96568:97048 [5] NCCL INFO Channel 01/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-138:96564:97046 [1] NCCL INFO Channel 02/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-138:96570:97051 [7] NCCL INFO Channel 01/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-138:96563:97045 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-138:96568:97048 [5] NCCL INFO Channel 02/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-138:96564:97046 [1] NCCL INFO Channel 03/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-138:96570:97051 [7] NCCL INFO Channel 02/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-242:1364907:1365367 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,fff00000,00000000
+ip-26-0-161-103:96889:97348 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-161-103:96889:97348 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-161-138:96566:97047 [3] NCCL INFO Channel 00/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-138:96563:97045 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-138:96568:97048 [5] NCCL INFO Channel 03/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-138:96564:97046 [1] NCCL INFO Channel 04/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-138:96570:97051 [7] NCCL INFO Channel 03/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-168-52:2722050:2722508 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-168-52:2722050:2722508 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-160-242:1364907:1365367 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-168-52:2722045:2722502 [0] NCCL INFO Channel 00/24 :    0   1   2   3   4   5   6   7
+ip-26-0-168-52:2722052:2722506 [7] NCCL INFO Trees [0] -1/-1/-1->7->6 [1] -1/-1/-1->7->6 [2] -1/-1/-1->7->6 [3] -1/-1/-1->7->6 [4] -1/-1/-1->7->6 [5] -1/-1/-1->7->6 [6] -1/-1/-1->7->6 [7] -1/-1/-1->7->6 [8] -1/-1/-1->7->6 [9] -1/-1/-1->7->6 [10] -1/-1/-1->7->6 [11] -1/-1/-1->7->6 [12] -1/-1/-1->7->6 [13] -1/-1/-1->7->6 [14] -1/-1/-1->7->6 [15] -1/-1/-1->7->6 [16] -1/-1/-1->7->6 [17] -1/-1/-1->7->6 [18] -1/-1/-1->7->6 [19] -1/-1/-1->7->6 [20] -1/-1/-1->7->6 [21] -1/-1/-1->7->6 [22] -1/-1/-1->7->6 [23] -1/-1/-1->7->6
+ip-26-0-161-138:96566:97047 [3] NCCL INFO Channel 01/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-168-52:2722051:2722503 [6] NCCL INFO Trees [0] 7/-1/-1->6->5 [1] 7/-1/-1->6->5 [2] 7/-1/-1->6->5 [3] 7/-1/-1->6->5 [4] 7/-1/-1->6->5 [5] 7/-1/-1->6->5 [6] 7/-1/-1->6->5 [7] 7/-1/-1->6->5 [8] 7/-1/-1->6->5 [9] 7/-1/-1->6->5 [10] 7/-1/-1->6->5 [11] 7/-1/-1->6->5 [12] 7/-1/-1->6->5 [13] 7/-1/-1->6->5 [14] 7/-1/-1->6->5 [15] 7/-1/-1->6->5 [16] 7/-1/-1->6->5 [17] 7/-1/-1->6->5 [18] 7/-1/-1->6->5 [19] 7/-1/-1->6->5 [20] 7/-1/-1->6->5 [21] 7/-1/-1->6->5 [22] 7/-1/-1->6->5 [23] 7/-1/-1->6->5
+ip-26-0-168-52:2722045:2722502 [0] NCCL INFO Channel 01/24 :    0   1   2   3   4   5   6   7
+ip-26-0-168-52:2722052:2722506 [7] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-168-52:2722051:2722503 [6] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-168-52:2722045:2722502 [0] NCCL INFO Channel 02/24 :    0   1   2   3   4   5   6   7
+ip-26-0-168-52:2722050:2722508 [5] NCCL INFO Trees [0] 6/-1/-1->5->4 [1] 6/-1/-1->5->4 [2] 6/-1/-1->5->4 [3] 6/-1/-1->5->4 [4] 6/-1/-1->5->4 [5] 6/-1/-1->5->4 [6] 6/-1/-1->5->4 [7] 6/-1/-1->5->4 [8] 6/-1/-1->5->4 [9] 6/-1/-1->5->4 [10] 6/-1/-1->5->4 [11] 6/-1/-1->5->4 [12] 6/-1/-1->5->4 [13] 6/-1/-1->5->4 [14] 6/-1/-1->5->4 [15] 6/-1/-1->5->4 [16] 6/-1/-1->5->4 [17] 6/-1/-1->5->4 [18] 6/-1/-1->5->4 [19] 6/-1/-1->5->4 [20] 6/-1/-1->5->4 [21] 6/-1/-1->5->4 [22] 6/-1/-1->5->4 [23] 6/-1/-1->5->4
+ip-26-0-168-52:2722045:2722502 [0] NCCL INFO Channel 03/24 :    0   1   2   3   4   5   6   7
+ip-26-0-168-52:2722045:2722502 [0] NCCL INFO Channel 04/24 :    0   1   2   3   4   5   6   7
+ip-26-0-168-52:2722049:2722504 [4] NCCL INFO Trees [0] 5/-1/-1->4->3 [1] 5/-1/-1->4->3 [2] 5/-1/-1->4->3 [3] 5/-1/-1->4->3 [4] 5/-1/-1->4->3 [5] 5/-1/-1->4->3 [6] 5/-1/-1->4->3 [7] 5/-1/-1->4->3 [8] 5/-1/-1->4->3 [9] 5/-1/-1->4->3 [10] 5/-1/-1->4->3 [11] 5/-1/-1->4->3 [12] 5/-1/-1->4->3 [13] 5/-1/-1->4->3 [14] 5/-1/-1->4->3 [15] 5/-1/-1->4->3 [16] 5/-1/-1->4->3 [17] 5/-1/-1->4->3 [18] 5/-1/-1->4->3 [19] 5/-1/-1->4->3 [20] 5/-1/-1->4->3 [21] 5/-1/-1->4->3 [22] 5/-1/-1->4->3 [23] 5/-1/-1->4->3
+ip-26-0-168-52:2722050:2722508 [5] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-168-52:2722045:2722502 [0] NCCL INFO Channel 05/24 :    0   1   2   3   4   5   6   7
+ip-26-0-168-52:2722049:2722504 [4] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-168-52:2722045:2722502 [0] NCCL INFO Channel 06/24 :    0   1   2   3   4   5   6   7
+ip-26-0-168-52:2722048:2722505 [3] NCCL INFO Trees [0] 4/-1/-1->3->2 [1] 4/-1/-1->3->2 [2] 4/-1/-1->3->2 [3] 4/-1/-1->3->2 [4] 4/-1/-1->3->2 [5] 4/-1/-1->3->2 [6] 4/-1/-1->3->2 [7] 4/-1/-1->3->2 [8] 4/-1/-1->3->2 [9] 4/-1/-1->3->2 [10] 4/-1/-1->3->2 [11] 4/-1/-1->3->2 [12] 4/-1/-1->3->2 [13] 4/-1/-1->3->2 [14] 4/-1/-1->3->2 [15] 4/-1/-1->3->2 [16] 4/-1/-1->3->2 [17] 4/-1/-1->3->2 [18] 4/-1/-1->3->2 [19] 4/-1/-1->3->2 [20] 4/-1/-1->3->2 [21] 4/-1/-1->3->2 [22] 4/-1/-1->3->2 [23] 4/-1/-1->3->2
+ip-26-0-168-52:2722045:2722502 [0] NCCL INFO Channel 07/24 :    0   1   2   3   4   5   6   7
+ip-26-0-168-52:2722045:2722502 [0] NCCL INFO Channel 08/24 :    0   1   2   3   4   5   6   7
+ip-26-0-168-52:2722046:2722509 [1] NCCL INFO Trees [0] 2/-1/-1->1->0 [1] 2/-1/-1->1->0 [2] 2/-1/-1->1->0 [3] 2/-1/-1->1->0 [4] 2/-1/-1->1->0 [5] 2/-1/-1->1->0 [6] 2/-1/-1->1->0 [7] 2/-1/-1->1->0 [8] 2/-1/-1->1->0 [9] 2/-1/-1->1->0 [10] 2/-1/-1->1->0 [11] 2/-1/-1->1->0 [12] 2/-1/-1->1->0 [13] 2/-1/-1->1->0 [14] 2/-1/-1->1->0 [15] 2/-1/-1->1->0 [16] 2/-1/-1->1->0 [17] 2/-1/-1->1->0 [18] 2/-1/-1->1->0 [19] 2/-1/-1->1->0 [20] 2/-1/-1->1->0 [21] 2/-1/-1->1->0 [22] 2/-1/-1->1->0 [23] 2/-1/-1->1->0
+ip-26-0-168-52:2722047:2722507 [2] NCCL INFO Trees [0] 3/-1/-1->2->1 [1] 3/-1/-1->2->1 [2] 3/-1/-1->2->1 [3] 3/-1/-1->2->1 [4] 3/-1/-1->2->1 [5] 3/-1/-1->2->1 [6] 3/-1/-1->2->1 [7] 3/-1/-1->2->1 [8] 3/-1/-1->2->1 [9] 3/-1/-1->2->1 [10] 3/-1/-1->2->1 [11] 3/-1/-1->2->1 [12] 3/-1/-1->2->1 [13] 3/-1/-1->2->1 [14] 3/-1/-1->2->1 [15] 3/-1/-1->2->1 [16] 3/-1/-1->2->1 [17] 3/-1/-1->2->1 [18] 3/-1/-1->2->1 [19] 3/-1/-1->2->1 [20] 3/-1/-1->2->1 [21] 3/-1/-1->2->1 [22] 3/-1/-1->2->1 [23] 3/-1/-1->2->1
+ip-26-0-168-52:2722048:2722505 [3] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-168-52:2722045:2722502 [0] NCCL INFO Channel 09/24 :    0   1   2   3   4   5   6   7
+ip-26-0-168-52:2722046:2722509 [1] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-168-52:2722045:2722502 [0] NCCL INFO Channel 10/24 :    0   1   2   3   4   5   6   7
+ip-26-0-168-52:2722047:2722507 [2] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-168-52:2722045:2722502 [0] NCCL INFO Channel 11/24 :    0   1   2   3   4   5   6   7
+ip-26-0-168-52:2722045:2722502 [0] NCCL INFO Channel 12/24 :    0   1   2   3   4   5   6   7
+ip-26-0-168-52:2722045:2722502 [0] NCCL INFO Channel 13/24 :    0   1   2   3   4   5   6   7
+ip-26-0-168-52:2722045:2722502 [0] NCCL INFO Channel 14/24 :    0   1   2   3   4   5   6   7
+ip-26-0-168-52:2722045:2722502 [0] NCCL INFO Channel 15/24 :    0   1   2   3   4   5   6   7
+ip-26-0-168-52:2722045:2722502 [0] NCCL INFO Channel 16/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-138:96563:97045 [0] NCCL INFO Channel 04/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-168-52:2722045:2722502 [0] NCCL INFO Channel 17/24 :    0   1   2   3   4   5   6   7
+ip-26-0-168-52:2722045:2722502 [0] NCCL INFO Channel 18/24 :    0   1   2   3   4   5   6   7
+ip-26-0-168-52:2722045:2722502 [0] NCCL INFO Channel 19/24 :    0   1   2   3   4   5   6   7
+ip-26-0-168-52:2722045:2722502 [0] NCCL INFO Channel 20/24 :    0   1   2   3   4   5   6   7
+ip-26-0-168-52:2722045:2722502 [0] NCCL INFO Channel 21/24 :    0   1   2   3   4   5   6   7
+ip-26-0-168-52:2722045:2722502 [0] NCCL INFO Channel 22/24 :    0   1   2   3   4   5   6   7
+ip-26-0-168-52:2722045:2722502 [0] NCCL INFO Channel 23/24 :    0   1   2   3   4   5   6   7
+ip-26-0-168-52:2722045:2722502 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] 1/-1/-1->0->-1 [7] 1/-1/-1->0->-1 [8] 1/-1/-1->0->-1 [9] 1/-1/-1->0->-1 [10] 1/-1/-1->0->-1 [11] 1/-1/-1->0->-1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] 1/-1/-1->0->-1 [19] 1/-1/-1->0->-1 [20] 1/-1/-1->0->-1 [21] 1/-1/-1->0->-1 [22] 1/-1/-1->0->-1 [23] 1/-1/-1->0->-1
+ip-26-0-168-52:2722045:2722502 [0] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-161-138:96568:97048 [5] NCCL INFO Channel 04/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-138:96564:97046 [1] NCCL INFO Channel 05/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-138:96570:97051 [7] NCCL INFO Channel 04/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-138:96566:97047 [3] NCCL INFO Channel 02/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-138:96563:97045 [0] NCCL INFO Channel 05/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-138:96568:97048 [5] NCCL INFO Channel 05/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-138:96564:97046 [1] NCCL INFO Channel 06/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-138:96570:97051 [7] NCCL INFO Channel 05/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-138:96565:97044 [2] NCCL INFO Channel 00/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-138:96566:97047 [3] NCCL INFO Channel 03/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-138:96569:97049 [6] NCCL INFO Channel 00/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-138:96563:97045 [0] NCCL INFO Channel 06/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-138:96568:97048 [5] NCCL INFO Channel 06/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-138:96564:97046 [1] NCCL INFO Channel 07/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-138:96570:97051 [7] NCCL INFO Channel 06/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-138:96565:97044 [2] NCCL INFO Channel 01/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-138:96566:97047 [3] NCCL INFO Channel 04/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-138:96569:97049 [6] NCCL INFO Channel 01/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-138:96563:97045 [0] NCCL INFO Channel 07/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-59:1557271:1557726 [5] NCCL INFO Channel 14/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-138:96568:97048 [5] NCCL INFO Channel 07/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-138:96564:97046 [1] NCCL INFO Channel 08/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-138:96570:97051 [7] NCCL INFO Channel 07/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-138:96565:97044 [2] NCCL INFO Channel 02/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-138:96566:97047 [3] NCCL INFO Channel 05/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-138:96569:97049 [6] NCCL INFO Channel 02/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-138:96563:97045 [0] NCCL INFO Channel 08/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-138:96568:97048 [5] NCCL INFO Channel 08/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-138:96564:97046 [1] NCCL INFO Channel 09/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-138:96570:97051 [7] NCCL INFO Channel 08/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-138:96565:97044 [2] NCCL INFO Channel 03/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-138:96569:97049 [6] NCCL INFO Channel 03/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-138:96563:97045 [0] NCCL INFO Channel 09/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-138:96568:97048 [5] NCCL INFO Channel 09/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-138:96564:97046 [1] NCCL INFO Channel 10/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-138:96570:97051 [7] NCCL INFO Channel 09/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-138:96565:97044 [2] NCCL INFO Channel 04/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-59:1557271:1557726 [5] NCCL INFO Channel 15/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-138:96569:97049 [6] NCCL INFO Channel 04/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-138:96563:97045 [0] NCCL INFO Channel 10/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-138:96568:97048 [5] NCCL INFO Channel 10/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-142:62882:63343 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-161-138:96570:97051 [7] NCCL INFO Channel 10/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-138:96564:97046 [1] NCCL INFO Channel 11/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-138:96565:97044 [2] NCCL INFO Channel 05/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-242:1364909:1365373 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,fff00000,00000000
+ip-26-0-161-142:62882:63343 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-160-242:1364909:1365373 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-160-242:1364909:1365373 [7] NCCL INFO Trees [0] -1/-1/-1->7->6 [1] -1/-1/-1->7->6 [2] -1/-1/-1->7->6 [3] -1/-1/-1->7->6 [4] -1/-1/-1->7->6 [5] -1/-1/-1->7->6 [6] -1/-1/-1->7->6 [7] -1/-1/-1->7->6 [8] -1/-1/-1->7->6 [9] -1/-1/-1->7->6 [10] -1/-1/-1->7->6 [11] -1/-1/-1->7->6 [12] -1/-1/-1->7->6 [13] -1/-1/-1->7->6 [14] -1/-1/-1->7->6 [15] -1/-1/-1->7->6 [16] -1/-1/-1->7->6 [17] -1/-1/-1->7->6 [18] -1/-1/-1->7->6 [19] -1/-1/-1->7->6 [20] -1/-1/-1->7->6 [21] -1/-1/-1->7->6 [22] -1/-1/-1->7->6 [23] -1/-1/-1->7->6
+ip-26-0-160-242:1364909:1365373 [7] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-160-242:1364902:1365366 [0] NCCL INFO Channel 00/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-242:1364902:1365366 [0] NCCL INFO Channel 01/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-242:1364902:1365366 [0] NCCL INFO Channel 02/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-242:1364902:1365366 [0] NCCL INFO Channel 03/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-242:1364902:1365366 [0] NCCL INFO Channel 04/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-242:1364902:1365366 [0] NCCL INFO Channel 05/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-242:1364902:1365366 [0] NCCL INFO Channel 06/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-242:1364907:1365367 [5] NCCL INFO Trees [0] 6/-1/-1->5->4 [1] 6/-1/-1->5->4 [2] 6/-1/-1->5->4 [3] 6/-1/-1->5->4 [4] 6/-1/-1->5->4 [5] 6/-1/-1->5->4 [6] 6/-1/-1->5->4 [7] 6/-1/-1->5->4 [8] 6/-1/-1->5->4 [9] 6/-1/-1->5->4 [10] 6/-1/-1->5->4 [11] 6/-1/-1->5->4 [12] 6/-1/-1->5->4 [13] 6/-1/-1->5->4 [14] 6/-1/-1->5->4 [15] 6/-1/-1->5->4 [16] 6/-1/-1->5->4 [17] 6/-1/-1->5->4 [18] 6/-1/-1->5->4 [19] 6/-1/-1->5->4 [20] 6/-1/-1->5->4 [21] 6/-1/-1->5->4 [22] 6/-1/-1->5->4 [23] 6/-1/-1->5->4
+ip-26-0-160-242:1364902:1365366 [0] NCCL INFO Channel 07/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-242:1364905:1365372 [3] NCCL INFO Trees [0] 4/-1/-1->3->2 [1] 4/-1/-1->3->2 [2] 4/-1/-1->3->2 [3] 4/-1/-1->3->2 [4] 4/-1/-1->3->2 [5] 4/-1/-1->3->2 [6] 4/-1/-1->3->2 [7] 4/-1/-1->3->2 [8] 4/-1/-1->3->2 [9] 4/-1/-1->3->2 [10] 4/-1/-1->3->2 [11] 4/-1/-1->3->2 [12] 4/-1/-1->3->2 [13] 4/-1/-1->3->2 [14] 4/-1/-1->3->2 [15] 4/-1/-1->3->2 [16] 4/-1/-1->3->2 [17] 4/-1/-1->3->2 [18] 4/-1/-1->3->2 [19] 4/-1/-1->3->2 [20] 4/-1/-1->3->2 [21] 4/-1/-1->3->2 [22] 4/-1/-1->3->2 [23] 4/-1/-1->3->2
+ip-26-0-160-242:1364902:1365366 [0] NCCL INFO Channel 08/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-242:1364907:1365367 [5] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-160-242:1364905:1365372 [3] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-160-242:1364903:1365371 [1] NCCL INFO Trees [0] 2/-1/-1->1->0 [1] 2/-1/-1->1->0 [2] 2/-1/-1->1->0 [3] 2/-1/-1->1->0 [4] 2/-1/-1->1->0 [5] 2/-1/-1->1->0 [6] 2/-1/-1->1->0 [7] 2/-1/-1->1->0 [8] 2/-1/-1->1->0 [9] 2/-1/-1->1->0 [10] 2/-1/-1->1->0 [11] 2/-1/-1->1->0 [12] 2/-1/-1->1->0 [13] 2/-1/-1->1->0 [14] 2/-1/-1->1->0 [15] 2/-1/-1->1->0 [16] 2/-1/-1->1->0 [17] 2/-1/-1->1->0 [18] 2/-1/-1->1->0 [19] 2/-1/-1->1->0 [20] 2/-1/-1->1->0 [21] 2/-1/-1->1->0 [22] 2/-1/-1->1->0 [23] 2/-1/-1->1->0
+ip-26-0-160-242:1364908:1365368 [6] NCCL INFO Trees [0] 7/-1/-1->6->5 [1] 7/-1/-1->6->5 [2] 7/-1/-1->6->5 [3] 7/-1/-1->6->5 [4] 7/-1/-1->6->5 [5] 7/-1/-1->6->5 [6] 7/-1/-1->6->5 [7] 7/-1/-1->6->5 [8] 7/-1/-1->6->5 [9] 7/-1/-1->6->5 [10] 7/-1/-1->6->5 [11] 7/-1/-1->6->5 [12] 7/-1/-1->6->5 [13] 7/-1/-1->6->5 [14] 7/-1/-1->6->5 [15] 7/-1/-1->6->5 [16] 7/-1/-1->6->5 [17] 7/-1/-1->6->5 [18] 7/-1/-1->6->5 [19] 7/-1/-1->6->5 [20] 7/-1/-1->6->5 [21] 7/-1/-1->6->5 [22] 7/-1/-1->6->5 [23] 7/-1/-1->6->5
+ip-26-0-160-242:1364902:1365366 [0] NCCL INFO Channel 09/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-242:1364904:1365369 [2] NCCL INFO Trees [0] 3/-1/-1->2->1 [1] 3/-1/-1->2->1 [2] 3/-1/-1->2->1 [3] 3/-1/-1->2->1 [4] 3/-1/-1->2->1 [5] 3/-1/-1->2->1 [6] 3/-1/-1->2->1 [7] 3/-1/-1->2->1 [8] 3/-1/-1->2->1 [9] 3/-1/-1->2->1 [10] 3/-1/-1->2->1 [11] 3/-1/-1->2->1 [12] 3/-1/-1->2->1 [13] 3/-1/-1->2->1 [14] 3/-1/-1->2->1 [15] 3/-1/-1->2->1 [16] 3/-1/-1->2->1 [17] 3/-1/-1->2->1 [18] 3/-1/-1->2->1 [19] 3/-1/-1->2->1 [20] 3/-1/-1->2->1 [21] 3/-1/-1->2->1 [22] 3/-1/-1->2->1 [23] 3/-1/-1->2->1
+ip-26-0-160-242:1364902:1365366 [0] NCCL INFO Channel 10/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-242:1364903:1365371 [1] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-160-242:1364902:1365366 [0] NCCL INFO Channel 11/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-242:1364902:1365366 [0] NCCL INFO Channel 12/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-242:1364904:1365369 [2] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-160-242:1364902:1365366 [0] NCCL INFO Channel 13/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-242:1364906:1365370 [4] NCCL INFO Trees [0] 5/-1/-1->4->3 [1] 5/-1/-1->4->3 [2] 5/-1/-1->4->3 [3] 5/-1/-1->4->3 [4] 5/-1/-1->4->3 [5] 5/-1/-1->4->3 [6] 5/-1/-1->4->3 [7] 5/-1/-1->4->3 [8] 5/-1/-1->4->3 [9] 5/-1/-1->4->3 [10] 5/-1/-1->4->3 [11] 5/-1/-1->4->3 [12] 5/-1/-1->4->3 [13] 5/-1/-1->4->3 [14] 5/-1/-1->4->3 [15] 5/-1/-1->4->3 [16] 5/-1/-1->4->3 [17] 5/-1/-1->4->3 [18] 5/-1/-1->4->3 [19] 5/-1/-1->4->3 [20] 5/-1/-1->4->3 [21] 5/-1/-1->4->3 [22] 5/-1/-1->4->3 [23] 5/-1/-1->4->3
+ip-26-0-160-242:1364902:1365366 [0] NCCL INFO Channel 14/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-242:1364902:1365366 [0] NCCL INFO Channel 15/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-242:1364902:1365366 [0] NCCL INFO Channel 16/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-242:1364902:1365366 [0] NCCL INFO Channel 17/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-242:1364902:1365366 [0] NCCL INFO Channel 18/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-242:1364906:1365370 [4] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-160-242:1364902:1365366 [0] NCCL INFO Channel 19/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-242:1364902:1365366 [0] NCCL INFO Channel 20/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-242:1364902:1365366 [0] NCCL INFO Channel 21/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-242:1364902:1365366 [0] NCCL INFO Channel 22/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-242:1364902:1365366 [0] NCCL INFO Channel 23/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-242:1364908:1365368 [6] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-160-242:1364902:1365366 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] 1/-1/-1->0->-1 [7] 1/-1/-1->0->-1 [8] 1/-1/-1->0->-1 [9] 1/-1/-1->0->-1 [10] 1/-1/-1->0->-1 [11] 1/-1/-1->0->-1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] 1/-1/-1->0->-1 [19] 1/-1/-1->0->-1 [20] 1/-1/-1->0->-1 [21] 1/-1/-1->0->-1 [22] 1/-1/-1->0->-1 [23] 1/-1/-1->0->-1
+ip-26-0-160-242:1364902:1365366 [0] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-161-138:96569:97049 [6] NCCL INFO Channel 05/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-168-34:272151:272611 [1] NCCL INFO Channel 00/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-138:96563:97045 [0] NCCL INFO Channel 11/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-138:96568:97048 [5] NCCL INFO Channel 11/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-138:96570:97051 [7] NCCL INFO Channel 11/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-138:96564:97046 [1] NCCL INFO Channel 12/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-138:96565:97044 [2] NCCL INFO Channel 06/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-168-34:272156:272607 [6] NCCL INFO Channel 00/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-168-34:272155:272609 [5] NCCL INFO Channel 00/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-168-34:272152:272610 [2] NCCL INFO Channel 00/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-168-34:272151:272611 [1] NCCL INFO Channel 01/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-138:96569:97049 [6] NCCL INFO Channel 06/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-168-34:272156:272607 [6] NCCL INFO Channel 01/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-168-34:272155:272609 [5] NCCL INFO Channel 01/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-138:96568:97048 [5] NCCL INFO Channel 12/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-138:96563:97045 [0] NCCL INFO Channel 12/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-168-34:272152:272610 [2] NCCL INFO Channel 01/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-138:96570:97051 [7] NCCL INFO Channel 12/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-138:96564:97046 [1] NCCL INFO Channel 13/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-138:96565:97044 [2] NCCL INFO Channel 07/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-168-34:272151:272611 [1] NCCL INFO Channel 02/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-168-34:272156:272607 [6] NCCL INFO Channel 02/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-168-34:272155:272609 [5] NCCL INFO Channel 02/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-138:96569:97049 [6] NCCL INFO Channel 07/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-168-34:272152:272610 [2] NCCL INFO Channel 02/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-168-34:272151:272611 [1] NCCL INFO Channel 03/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-168-34:272156:272607 [6] NCCL INFO Channel 03/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-168-34:272155:272609 [5] NCCL INFO Channel 03/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-138:96568:97048 [5] NCCL INFO Channel 13/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-138:96563:97045 [0] NCCL INFO Channel 13/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-138:96570:97051 [7] NCCL INFO Channel 13/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-138:96564:97046 [1] NCCL INFO Channel 14/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-138:96565:97044 [2] NCCL INFO Channel 08/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-138:96566:97047 [3] NCCL INFO Channel 06/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-168-34:272152:272610 [2] NCCL INFO Channel 03/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-168-34:272151:272611 [1] NCCL INFO Channel 04/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-138:96569:97049 [6] NCCL INFO Channel 08/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-142:62888:63340 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-168-34:272155:272609 [5] NCCL INFO Channel 04/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-168-34:272156:272607 [6] NCCL INFO Channel 04/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-142:62888:63340 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-161-142:62881:63337 [0] NCCL INFO Channel 00/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-142:62881:63337 [0] NCCL INFO Channel 01/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-142:62881:63337 [0] NCCL INFO Channel 02/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-142:62881:63337 [0] NCCL INFO Channel 03/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-142:62881:63337 [0] NCCL INFO Channel 04/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-142:62882:63343 [1] NCCL INFO Trees [0] 2/-1/-1->1->0 [1] 2/-1/-1->1->0 [2] 2/-1/-1->1->0 [3] 2/-1/-1->1->0 [4] 2/-1/-1->1->0 [5] 2/-1/-1->1->0 [6] 2/-1/-1->1->0 [7] 2/-1/-1->1->0 [8] 2/-1/-1->1->0 [9] 2/-1/-1->1->0 [10] 2/-1/-1->1->0 [11] 2/-1/-1->1->0 [12] 2/-1/-1->1->0 [13] 2/-1/-1->1->0 [14] 2/-1/-1->1->0 [15] 2/-1/-1->1->0 [16] 2/-1/-1->1->0 [17] 2/-1/-1->1->0 [18] 2/-1/-1->1->0 [19] 2/-1/-1->1->0 [20] 2/-1/-1->1->0 [21] 2/-1/-1->1->0 [22] 2/-1/-1->1->0 [23] 2/-1/-1->1->0
+ip-26-0-161-142:62888:63340 [7] NCCL INFO Trees [0] -1/-1/-1->7->6 [1] -1/-1/-1->7->6 [2] -1/-1/-1->7->6 [3] -1/-1/-1->7->6 [4] -1/-1/-1->7->6 [5] -1/-1/-1->7->6 [6] -1/-1/-1->7->6 [7] -1/-1/-1->7->6 [8] -1/-1/-1->7->6 [9] -1/-1/-1->7->6 [10] -1/-1/-1->7->6 [11] -1/-1/-1->7->6 [12] -1/-1/-1->7->6 [13] -1/-1/-1->7->6 [14] -1/-1/-1->7->6 [15] -1/-1/-1->7->6 [16] -1/-1/-1->7->6 [17] -1/-1/-1->7->6 [18] -1/-1/-1->7->6 [19] -1/-1/-1->7->6 [20] -1/-1/-1->7->6 [21] -1/-1/-1->7->6 [22] -1/-1/-1->7->6 [23] -1/-1/-1->7->6
+ip-26-0-161-142:62881:63337 [0] NCCL INFO Channel 05/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-142:62882:63343 [1] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-161-142:62888:63340 [7] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-161-142:62881:63337 [0] NCCL INFO Channel 06/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-142:62881:63337 [0] NCCL INFO Channel 07/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-142:62881:63337 [0] NCCL INFO Channel 08/24 :    0   1   2   3   4   5   6   7
+ip-26-0-168-34:272152:272610 [2] NCCL INFO Channel 04/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-142:62887:63344 [6] NCCL INFO Trees [0] 7/-1/-1->6->5 [1] 7/-1/-1->6->5 [2] 7/-1/-1->6->5 [3] 7/-1/-1->6->5 [4] 7/-1/-1->6->5 [5] 7/-1/-1->6->5 [6] 7/-1/-1->6->5 [7] 7/-1/-1->6->5 [8] 7/-1/-1->6->5 [9] 7/-1/-1->6->5 [10] 7/-1/-1->6->5 [11] 7/-1/-1->6->5 [12] 7/-1/-1->6->5 [13] 7/-1/-1->6->5 [14] 7/-1/-1->6->5 [15] 7/-1/-1->6->5 [16] 7/-1/-1->6->5 [17] 7/-1/-1->6->5 [18] 7/-1/-1->6->5 [19] 7/-1/-1->6->5 [20] 7/-1/-1->6->5 [21] 7/-1/-1->6->5 [22] 7/-1/-1->6->5 [23] 7/-1/-1->6->5
+ip-26-0-161-142:62881:63337 [0] NCCL INFO Channel 09/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-142:62883:63338 [2] NCCL INFO Trees [0] 3/-1/-1->2->1 [1] 3/-1/-1->2->1 [2] 3/-1/-1->2->1 [3] 3/-1/-1->2->1 [4] 3/-1/-1->2->1 [5] 3/-1/-1->2->1 [6] 3/-1/-1->2->1 [7] 3/-1/-1->2->1 [8] 3/-1/-1->2->1 [9] 3/-1/-1->2->1 [10] 3/-1/-1->2->1 [11] 3/-1/-1->2->1 [12] 3/-1/-1->2->1 [13] 3/-1/-1->2->1 [14] 3/-1/-1->2->1 [15] 3/-1/-1->2->1 [16] 3/-1/-1->2->1 [17] 3/-1/-1->2->1 [18] 3/-1/-1->2->1 [19] 3/-1/-1->2->1 [20] 3/-1/-1->2->1 [21] 3/-1/-1->2->1 [22] 3/-1/-1->2->1 [23] 3/-1/-1->2->1
+ip-26-0-161-142:62884:63339 [3] NCCL INFO Trees [0] 4/-1/-1->3->2 [1] 4/-1/-1->3->2 [2] 4/-1/-1->3->2 [3] 4/-1/-1->3->2 [4] 4/-1/-1->3->2 [5] 4/-1/-1->3->2 [6] 4/-1/-1->3->2 [7] 4/-1/-1->3->2 [8] 4/-1/-1->3->2 [9] 4/-1/-1->3->2 [10] 4/-1/-1->3->2 [11] 4/-1/-1->3->2 [12] 4/-1/-1->3->2 [13] 4/-1/-1->3->2 [14] 4/-1/-1->3->2 [15] 4/-1/-1->3->2 [16] 4/-1/-1->3->2 [17] 4/-1/-1->3->2 [18] 4/-1/-1->3->2 [19] 4/-1/-1->3->2 [20] 4/-1/-1->3->2 [21] 4/-1/-1->3->2 [22] 4/-1/-1->3->2 [23] 4/-1/-1->3->2
+ip-26-0-161-142:62887:63344 [6] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-161-142:62881:63337 [0] NCCL INFO Channel 10/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-142:62883:63338 [2] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-161-142:62881:63337 [0] NCCL INFO Channel 11/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-142:62884:63339 [3] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-161-142:62881:63337 [0] NCCL INFO Channel 12/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-142:62886:63341 [5] NCCL INFO Trees [0] 6/-1/-1->5->4 [1] 6/-1/-1->5->4 [2] 6/-1/-1->5->4 [3] 6/-1/-1->5->4 [4] 6/-1/-1->5->4 [5] 6/-1/-1->5->4 [6] 6/-1/-1->5->4 [7] 6/-1/-1->5->4 [8] 6/-1/-1->5->4 [9] 6/-1/-1->5->4 [10] 6/-1/-1->5->4 [11] 6/-1/-1->5->4 [12] 6/-1/-1->5->4 [13] 6/-1/-1->5->4 [14] 6/-1/-1->5->4 [15] 6/-1/-1->5->4 [16] 6/-1/-1->5->4 [17] 6/-1/-1->5->4 [18] 6/-1/-1->5->4 [19] 6/-1/-1->5->4 [20] 6/-1/-1->5->4 [21] 6/-1/-1->5->4 [22] 6/-1/-1->5->4 [23] 6/-1/-1->5->4
+ip-26-0-168-34:272151:272611 [1] NCCL INFO Channel 05/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-142:62881:63337 [0] NCCL INFO Channel 13/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-142:62881:63337 [0] NCCL INFO Channel 14/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-142:62886:63341 [5] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-161-142:62881:63337 [0] NCCL INFO Channel 15/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-142:62881:63337 [0] NCCL INFO Channel 16/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-142:62881:63337 [0] NCCL INFO Channel 17/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-142:62881:63337 [0] NCCL INFO Channel 18/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-142:62881:63337 [0] NCCL INFO Channel 19/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-142:62881:63337 [0] NCCL INFO Channel 20/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-142:62881:63337 [0] NCCL INFO Channel 21/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-142:62881:63337 [0] NCCL INFO Channel 22/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-142:62881:63337 [0] NCCL INFO Channel 23/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-142:62885:63342 [4] NCCL INFO Trees [0] 5/-1/-1->4->3 [1] 5/-1/-1->4->3 [2] 5/-1/-1->4->3 [3] 5/-1/-1->4->3 [4] 5/-1/-1->4->3 [5] 5/-1/-1->4->3 [6] 5/-1/-1->4->3 [7] 5/-1/-1->4->3 [8] 5/-1/-1->4->3 [9] 5/-1/-1->4->3 [10] 5/-1/-1->4->3 [11] 5/-1/-1->4->3 [12] 5/-1/-1->4->3 [13] 5/-1/-1->4->3 [14] 5/-1/-1->4->3 [15] 5/-1/-1->4->3 [16] 5/-1/-1->4->3 [17] 5/-1/-1->4->3 [18] 5/-1/-1->4->3 [19] 5/-1/-1->4->3 [20] 5/-1/-1->4->3 [21] 5/-1/-1->4->3 [22] 5/-1/-1->4->3 [23] 5/-1/-1->4->3
+ip-26-0-161-142:62881:63337 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] 1/-1/-1->0->-1 [7] 1/-1/-1->0->-1 [8] 1/-1/-1->0->-1 [9] 1/-1/-1->0->-1 [10] 1/-1/-1->0->-1 [11] 1/-1/-1->0->-1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] 1/-1/-1->0->-1 [19] 1/-1/-1->0->-1 [20] 1/-1/-1->0->-1 [21] 1/-1/-1->0->-1 [22] 1/-1/-1->0->-1 [23] 1/-1/-1->0->-1
+ip-26-0-161-142:62881:63337 [0] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-161-142:62885:63342 [4] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-161-138:96568:97048 [5] NCCL INFO Channel 14/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-138:96563:97045 [0] NCCL INFO Channel 14/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-138:96570:97051 [7] NCCL INFO Channel 14/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-138:96565:97044 [2] NCCL INFO Channel 09/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-138:96564:97046 [1] NCCL INFO Channel 15/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-138:96566:97047 [3] NCCL INFO Channel 07/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-168-34:272156:272607 [6] NCCL INFO Channel 05/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-138:96569:97049 [6] NCCL INFO Channel 09/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-168-34:272155:272609 [5] NCCL INFO Channel 05/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-168-34:272152:272610 [2] NCCL INFO Channel 05/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-168-34:272151:272611 [1] NCCL INFO Channel 06/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-138:96568:97048 [5] NCCL INFO Channel 15/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-138:96563:97045 [0] NCCL INFO Channel 15/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-138:96565:97044 [2] NCCL INFO Channel 10/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-138:96570:97051 [7] NCCL INFO Channel 15/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-138:96564:97046 [1] NCCL INFO Channel 16/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-168-34:272156:272607 [6] NCCL INFO Channel 06/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-138:96566:97047 [3] NCCL INFO Channel 08/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-168-34:272155:272609 [5] NCCL INFO Channel 06/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-138:96569:97049 [6] NCCL INFO Channel 10/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-168-34:272152:272610 [2] NCCL INFO Channel 06/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-168-34:272151:272611 [1] NCCL INFO Channel 07/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-168-34:272156:272607 [6] NCCL INFO Channel 07/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-168-34:272155:272609 [5] NCCL INFO Channel 07/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-138:96568:97048 [5] NCCL INFO Channel 16/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-138:96563:97045 [0] NCCL INFO Channel 16/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-138:96565:97044 [2] NCCL INFO Channel 11/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-138:96570:97051 [7] NCCL INFO Channel 16/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-138:96564:97046 [1] NCCL INFO Channel 17/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-138:96569:97049 [6] NCCL INFO Channel 11/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-138:96566:97047 [3] NCCL INFO Channel 09/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-168-34:272154:272608 [4] NCCL INFO Channel 00/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-168-34:272152:272610 [2] NCCL INFO Channel 07/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-168-34:272151:272611 [1] NCCL INFO Channel 08/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-168-34:272155:272609 [5] NCCL INFO Channel 08/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-138:96563:97045 [0] NCCL INFO Channel 17/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-138:96568:97048 [5] NCCL INFO Channel 17/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-138:96565:97044 [2] NCCL INFO Channel 12/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-138:96570:97051 [7] NCCL INFO Channel 17/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-138:96564:97046 [1] NCCL INFO Channel 18/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-138:96569:97049 [6] NCCL INFO Channel 12/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-138:96566:97047 [3] NCCL INFO Channel 10/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-168-34:272153:272612 [3] NCCL INFO Channel 00/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-168-34:272150:272606 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-168-34:272152:272610 [2] NCCL INFO Channel 08/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-168-34:272154:272608 [4] NCCL INFO Channel 01/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-168-34:272151:272611 [1] NCCL INFO Channel 09/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-168-34:272157:272613 [7] NCCL INFO Channel 00/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-168-34:272155:272609 [5] NCCL INFO Channel 09/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-138:96568:97048 [5] NCCL INFO Channel 18/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-138:96563:97045 [0] NCCL INFO Channel 18/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-138:96565:97044 [2] NCCL INFO Channel 13/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-138:96570:97051 [7] NCCL INFO Channel 18/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-138:96564:97046 [1] NCCL INFO Channel 19/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-138:96569:97049 [6] NCCL INFO Channel 13/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-168-34:272153:272612 [3] NCCL INFO Channel 01/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-138:96566:97047 [3] NCCL INFO Channel 11/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-168-34:272152:272610 [2] NCCL INFO Channel 09/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-168-34:272150:272606 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-168-34:272154:272608 [4] NCCL INFO Channel 02/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-59:1557271:1557726 [5] NCCL INFO Channel 16/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-168-34:272151:272611 [1] NCCL INFO Channel 10/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-168-34:272157:272613 [7] NCCL INFO Channel 01/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-168-34:272155:272609 [5] NCCL INFO Channel 10/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-138:96568:97048 [5] NCCL INFO Channel 19/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-138:96563:97045 [0] NCCL INFO Channel 19/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-138:96565:97044 [2] NCCL INFO Channel 14/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-138:96570:97051 [7] NCCL INFO Channel 19/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-168-34:272153:272612 [3] NCCL INFO Channel 02/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-138:96564:97046 [1] NCCL INFO Channel 20/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-138:96569:97049 [6] NCCL INFO Channel 14/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-168-34:272152:272610 [2] NCCL INFO Channel 10/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-168-34:272150:272606 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-168-34:272154:272608 [4] NCCL INFO Channel 03/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-168-34:272157:272613 [7] NCCL INFO Channel 02/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-168-34:272151:272611 [1] NCCL INFO Channel 11/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-138:96566:97047 [3] NCCL INFO Channel 12/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-168-34:272155:272609 [5] NCCL INFO Channel 11/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-138:96567:97050 [4] NCCL INFO Channel 00/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-168-34:272153:272612 [3] NCCL INFO Channel 03/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-168-34:272150:272606 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-138:96568:97048 [5] NCCL INFO Channel 20/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-138:96565:97044 [2] NCCL INFO Channel 15/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-138:96563:97045 [0] NCCL INFO Channel 20/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-168-34:272152:272610 [2] NCCL INFO Channel 11/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-168-34:272154:272608 [4] NCCL INFO Channel 04/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-138:96570:97051 [7] NCCL INFO Channel 20/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-138:96569:97049 [6] NCCL INFO Channel 15/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-168-34:272157:272613 [7] NCCL INFO Channel 03/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-168-34:272151:272611 [1] NCCL INFO Channel 12/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-138:96564:97046 [1] NCCL INFO Channel 21/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-138:96566:97047 [3] NCCL INFO Channel 13/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-168-34:272155:272609 [5] NCCL INFO Channel 12/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-138:96567:97050 [4] NCCL INFO Channel 01/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-168-34:272153:272612 [3] NCCL INFO Channel 04/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-168-34:272150:272606 [0] NCCL INFO Channel 04/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-168-34:272152:272610 [2] NCCL INFO Channel 12/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-168-34:272154:272608 [4] NCCL INFO Channel 05/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-168-34:272157:272613 [7] NCCL INFO Channel 04/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-168-34:272151:272611 [1] NCCL INFO Channel 13/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-138:96565:97044 [2] NCCL INFO Channel 16/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-138:96568:97048 [5] NCCL INFO Channel 21/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-138:96563:97045 [0] NCCL INFO Channel 21/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-138:96570:97051 [7] NCCL INFO Channel 21/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-138:96569:97049 [6] NCCL INFO Channel 16/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-138:96564:97046 [1] NCCL INFO Channel 22/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-138:96566:97047 [3] NCCL INFO Channel 14/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-168-34:272155:272609 [5] NCCL INFO Channel 13/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-138:96567:97050 [4] NCCL INFO Channel 02/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-168-34:272153:272612 [3] NCCL INFO Channel 05/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-168-34:272150:272606 [0] NCCL INFO Channel 05/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-168-34:272154:272608 [4] NCCL INFO Channel 06/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-168-34:272157:272613 [7] NCCL INFO Channel 05/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-168-34:272152:272610 [2] NCCL INFO Channel 13/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-168-34:272151:272611 [1] NCCL INFO Channel 14/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-78:114517:114969 [5] NCCL INFO Channel 14/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-78:114516:114975 [4] NCCL INFO Channel 23/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-78:114512:114968 [0] NCCL INFO Channel 16/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-138:96565:97044 [2] NCCL INFO Channel 17/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-138:96568:97048 [5] NCCL INFO Channel 22/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-138:96563:97045 [0] NCCL INFO Channel 22/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-138:96570:97051 [7] NCCL INFO Channel 22/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-138:96569:97049 [6] NCCL INFO Channel 17/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-138:96564:97046 [1] NCCL INFO Channel 23/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-138:96566:97047 [3] NCCL INFO Channel 15/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-168-34:272156:272607 [6] NCCL INFO Channel 08/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-168-34:272155:272609 [5] NCCL INFO Channel 14/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-225:124162:124627 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-161-138:96567:97050 [4] NCCL INFO Channel 03/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-78:114515:114972 [3] NCCL INFO Channel 17/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-168-34:272153:272612 [3] NCCL INFO Channel 06/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-225:124162:124627 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-168-34:272150:272606 [0] NCCL INFO Channel 06/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-168-34:272154:272608 [4] NCCL INFO Channel 07/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-168-34:272157:272613 [7] NCCL INFO Channel 06/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-168-34:272152:272610 [2] NCCL INFO Channel 14/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-168-34:272151:272611 [1] NCCL INFO Channel 15/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-138:96565:97044 [2] NCCL INFO Channel 18/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-138:96568:97048 [5] NCCL INFO Channel 23/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-138:96570:97051 [7] NCCL INFO Channel 23/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-138:96563:97045 [0] NCCL INFO Channel 23/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-138:96569:97049 [6] NCCL INFO Channel 18/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-78:114517:114969 [5] NCCL INFO Channel 15/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-138:96566:97047 [3] NCCL INFO Channel 16/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-78:114512:114968 [0] NCCL INFO Channel 17/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-78:114515:114972 [3] NCCL INFO Channel 18/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-138:96565:97044 [2] NCCL INFO Channel 19/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-168-34:272155:272609 [5] NCCL INFO Channel 15/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-138:96569:97049 [6] NCCL INFO Channel 19/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-138:96567:97050 [4] NCCL INFO Channel 04/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-168-34:272156:272607 [6] NCCL INFO Channel 09/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-138:96566:97047 [3] NCCL INFO Channel 17/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-168-34:272153:272612 [3] NCCL INFO Channel 07/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-78:114517:114969 [5] NCCL INFO Channel 16/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-168-34:272152:272610 [2] NCCL INFO Channel 15/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-78:114512:114968 [0] NCCL INFO Channel 18/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-78:114515:114972 [3] NCCL INFO Channel 19/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-168-34:272151:272611 [1] NCCL INFO Channel 16/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-168-34:272155:272609 [5] NCCL INFO Channel 16/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-168-34:272156:272607 [6] NCCL INFO Channel 10/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-168-34:272153:272612 [3] NCCL INFO Channel 08/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-168-34:272152:272610 [2] NCCL INFO Channel 16/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-138:96565:97044 [2] NCCL INFO Channel 20/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-168-34:272150:272606 [0] NCCL INFO Channel 07/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-168-34:272156:272607 [6] NCCL INFO Channel 11/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-168-34:272153:272612 [3] NCCL INFO Channel 09/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-168-34:272155:272609 [5] NCCL INFO Channel 17/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-168-34:272151:272611 [1] NCCL INFO Channel 17/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-168-34:272152:272610 [2] NCCL INFO Channel 17/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-153:2636089:2636551 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-161-153:2636089:2636551 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-168-52:2722050:2722508 [5] NCCL INFO Channel 00/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-168-34:272155:272609 [5] NCCL INFO Channel 18/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-168-34:272154:272608 [4] NCCL INFO Channel 08/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-168-34:272157:272613 [7] NCCL INFO Channel 07/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-138:96569:97049 [6] NCCL INFO Channel 20/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-168-34:272151:272611 [1] NCCL INFO Channel 18/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-168-52:2722050:2722508 [5] NCCL INFO Channel 01/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-168-34:272155:272609 [5] NCCL INFO Channel 19/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-168-34:272154:272608 [4] NCCL INFO Channel 09/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-153:2636084:2636546 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-168-52:2722046:2722509 [1] NCCL INFO Channel 00/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-168-52:2722050:2722508 [5] NCCL INFO Channel 02/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-168-52:2722051:2722503 [6] NCCL INFO Channel 00/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-168-52:2722045:2722502 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-168-52:2722046:2722509 [1] NCCL INFO Channel 01/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-168-52:2722048:2722505 [3] NCCL INFO Channel 00/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-168-34:272157:272613 [7] NCCL INFO Channel 08/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-168-52:2722050:2722508 [5] NCCL INFO Channel 03/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-165-59:1557271:1557726 [5] NCCL INFO Channel 17/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-168-34:272153:272612 [3] NCCL INFO Channel 10/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-153:2636084:2636546 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-168-34:272152:272610 [2] NCCL INFO Channel 18/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-168-52:2722051:2722503 [6] NCCL INFO Channel 01/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-168-34:272156:272607 [6] NCCL INFO Channel 12/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-168-34:272157:272613 [7] NCCL INFO Channel 09/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-225:124165:124625 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-160-225:124165:124625 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-168-52:2722045:2722502 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-168-52:2722046:2722509 [1] NCCL INFO Channel 02/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-168-34:272153:272612 [3] NCCL INFO Channel 11/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-168-52:2722048:2722505 [3] NCCL INFO Channel 01/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-168-52:2722050:2722508 [5] NCCL INFO Channel 04/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-168-34:272152:272610 [2] NCCL INFO Channel 19/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-168-34:272156:272607 [6] NCCL INFO Channel 13/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-168-34:272157:272613 [7] NCCL INFO Channel 10/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-168-52:2722051:2722503 [6] NCCL INFO Channel 02/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-138:96567:97050 [4] NCCL INFO Channel 05/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-168-34:272153:272612 [3] NCCL INFO Channel 12/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-168-34:272152:272610 [2] NCCL INFO Channel 20/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-168-34:272157:272613 [7] NCCL INFO Channel 11/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-168-34:272156:272607 [6] NCCL INFO Channel 14/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-168-52:2722046:2722509 [1] NCCL INFO Channel 03/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-168-52:2722045:2722502 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-168-52:2722050:2722508 [5] NCCL INFO Channel 05/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-168-52:2722048:2722505 [3] NCCL INFO Channel 02/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-168-34:272153:272612 [3] NCCL INFO Channel 13/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-168-34:272152:272610 [2] NCCL INFO Channel 21/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-168-52:2722051:2722503 [6] NCCL INFO Channel 03/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-138:96566:97047 [3] NCCL INFO Channel 18/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-168-52:2722046:2722509 [1] NCCL INFO Channel 04/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-168-52:2722045:2722502 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-168-52:2722050:2722508 [5] NCCL INFO Channel 06/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-168-52:2722048:2722505 [3] NCCL INFO Channel 03/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-168-34:272150:272606 [0] NCCL INFO Channel 08/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-168-34:272157:272613 [7] NCCL INFO Channel 12/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-168-52:2722051:2722503 [6] NCCL INFO Channel 04/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-168-34:272151:272611 [1] NCCL INFO Channel 19/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-138:96567:97050 [4] NCCL INFO Channel 06/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-138:96566:97047 [3] NCCL INFO Channel 19/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-168-52:2722046:2722509 [1] NCCL INFO Channel 05/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-168-52:2722045:2722502 [0] NCCL INFO Channel 04/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-168-52:2722050:2722508 [5] NCCL INFO Channel 07/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-168-52:2722048:2722505 [3] NCCL INFO Channel 04/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-168-52:2722051:2722503 [6] NCCL INFO Channel 05/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-153:2636087:2636550 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-161-138:96565:97044 [2] NCCL INFO Channel 21/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-153:2636087:2636550 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-168-52:2722046:2722509 [1] NCCL INFO Channel 06/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-168-52:2722045:2722502 [0] NCCL INFO Channel 05/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-168-52:2722050:2722508 [5] NCCL INFO Channel 08/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-168-52:2722048:2722505 [3] NCCL INFO Channel 05/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-168-52:2722052:2722506 [7] NCCL INFO Channel 00/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-168-52:2722047:2722507 [2] NCCL INFO Channel 00/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-168-52:2722051:2722503 [6] NCCL INFO Channel 06/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-153:2636090:2636547 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-161-153:2636090:2636547 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-161-142:62885:63342 [4] NCCL INFO Channel 00/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-168-34:272154:272608 [4] NCCL INFO Channel 10/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-168-34:272156:272607 [6] NCCL INFO Channel 15/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-168-52:2722046:2722509 [1] NCCL INFO Channel 07/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-168-34:272153:272612 [3] NCCL INFO Channel 14/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-168-34:272152:272610 [2] NCCL INFO Channel 22/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-168-52:2722045:2722502 [0] NCCL INFO Channel 06/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-168-52:2722049:2722504 [4] NCCL INFO Channel 00/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-168-52:2722050:2722508 [5] NCCL INFO Channel 09/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-168-52:2722048:2722505 [3] NCCL INFO Channel 06/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-168-34:272154:272608 [4] NCCL INFO Channel 11/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-168-52:2722052:2722506 [7] NCCL INFO Channel 01/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-168-34:272156:272607 [6] NCCL INFO Channel 16/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-142:62886:63341 [5] NCCL INFO Channel 00/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-168-52:2722047:2722507 [2] NCCL INFO Channel 01/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-142:62885:63342 [4] NCCL INFO Channel 01/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-168-52:2722051:2722503 [6] NCCL INFO Channel 07/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-168-34:272153:272612 [3] NCCL INFO Channel 15/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-168-34:272152:272610 [2] NCCL INFO Channel 23/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-138:96567:97050 [4] NCCL INFO Channel 07/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-168-34:272154:272608 [4] NCCL INFO Channel 12/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-153:2636086:2636552 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-168-34:272156:272607 [6] NCCL INFO Channel 17/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-153:2636086:2636552 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-168-34:272155:272609 [5] NCCL INFO Channel 20/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-168-34:272153:272612 [3] NCCL INFO Channel 16/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-168-52:2722046:2722509 [1] NCCL INFO Channel 08/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-168-52:2722045:2722502 [0] NCCL INFO Channel 07/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-168-52:2722050:2722508 [5] NCCL INFO Channel 10/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-168-52:2722049:2722504 [4] NCCL INFO Channel 01/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-142:62886:63341 [5] NCCL INFO Channel 01/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-142:62885:63342 [4] NCCL INFO Channel 02/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-168-52:2722048:2722505 [3] NCCL INFO Channel 07/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-168-52:2722052:2722506 [7] NCCL INFO Channel 02/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-168-34:272154:272608 [4] NCCL INFO Channel 13/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-168-52:2722047:2722507 [2] NCCL INFO Channel 02/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-168-34:272156:272607 [6] NCCL INFO Channel 18/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-168-52:2722051:2722503 [6] NCCL INFO Channel 08/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-168-34:272155:272609 [5] NCCL INFO Channel 21/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-138:96565:97044 [2] NCCL INFO Channel 22/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-168-34:272153:272612 [3] NCCL INFO Channel 17/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-168-34:272154:272608 [4] NCCL INFO Channel 14/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-168-34:272156:272607 [6] NCCL INFO Channel 19/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-168-34:272157:272613 [7] NCCL INFO Channel 13/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-242:1364903:1365371 [1] NCCL INFO Channel 00/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-168-34:272155:272609 [5] NCCL INFO Channel 22/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-168-52:2722046:2722509 [1] NCCL INFO Channel 09/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-168-52:2722045:2722502 [0] NCCL INFO Channel 08/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-168-52:2722050:2722508 [5] NCCL INFO Channel 11/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-168-52:2722049:2722504 [4] NCCL INFO Channel 02/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-168-52:2722048:2722505 [3] NCCL INFO Channel 08/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-168-34:272153:272612 [3] NCCL INFO Channel 18/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-168-52:2722052:2722506 [7] NCCL INFO Channel 03/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-168-34:272150:272606 [0] NCCL INFO Channel 09/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-168-52:2722047:2722507 [2] NCCL INFO Channel 03/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-168-52:2722051:2722503 [6] NCCL INFO Channel 09/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-242:1364906:1365370 [4] NCCL INFO Channel 00/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-168-34:272154:272608 [4] NCCL INFO Channel 15/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-168-34:272156:272607 [6] NCCL INFO Channel 20/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-168-34:272151:272611 [1] NCCL INFO Channel 20/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-242:1364905:1365372 [3] NCCL INFO Channel 00/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-168-34:272157:272613 [7] NCCL INFO Channel 14/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-242:1364903:1365371 [1] NCCL INFO Channel 01/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-168-34:272155:272609 [5] NCCL INFO Channel 23/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-168-52:2722046:2722509 [1] NCCL INFO Channel 10/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-168-34:272153:272612 [3] NCCL INFO Channel 19/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-138:96569:97049 [6] NCCL INFO Channel 21/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-168-34:272150:272606 [0] NCCL INFO Channel 10/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-168-34:272154:272608 [4] NCCL INFO Channel 16/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-242:1364906:1365370 [4] NCCL INFO Channel 01/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-138:96566:97047 [3] NCCL INFO Channel 20/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-168-34:272156:272607 [6] NCCL INFO Channel 21/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-142:62886:63341 [5] NCCL INFO Channel 02/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-168-34:272151:272611 [1] NCCL INFO Channel 21/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-242:1364902:1365366 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-168-52:2722047:2722507 [2] NCCL INFO Channel 04/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-168-34:272157:272613 [7] NCCL INFO Channel 15/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-242:1364905:1365372 [3] NCCL INFO Channel 01/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-168-52:2722048:2722505 [3] NCCL INFO Channel 09/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-168-52:2722051:2722503 [6] NCCL INFO Channel 10/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-168-52:2722052:2722506 [7] NCCL INFO Channel 04/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-168-52:2722045:2722502 [0] NCCL INFO Channel 09/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-242:1364903:1365371 [1] NCCL INFO Channel 02/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-153:2636088:2636549 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-168-34:272153:272612 [3] NCCL INFO Channel 20/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-168-34:272150:272606 [0] NCCL INFO Channel 11/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-168-52:2722047:2722507 [2] NCCL INFO Channel 05/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-142:62885:63342 [4] NCCL INFO Channel 03/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-142:62882:63343 [1] NCCL INFO Channel 00/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-142:62886:63341 [5] NCCL INFO Channel 03/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-153:2636088:2636549 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-168-34:272154:272608 [4] NCCL INFO Channel 17/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-168-34:272156:272607 [6] NCCL INFO Channel 22/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-168-34:272151:272611 [1] NCCL INFO Channel 22/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-168-34:272157:272613 [7] NCCL INFO Channel 16/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-242:1364906:1365370 [4] NCCL INFO Channel 02/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-168-52:2722048:2722505 [3] NCCL INFO Channel 10/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-168-52:2722052:2722506 [7] NCCL INFO Channel 05/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-242:1364902:1365366 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-168-52:2722051:2722503 [6] NCCL INFO Channel 11/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-168-52:2722045:2722502 [0] NCCL INFO Channel 10/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-168-52:2722050:2722508 [5] NCCL INFO Channel 12/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-153:2636090:2636547 [6] NCCL INFO Trees [0] 7/-1/-1->6->5 [1] 7/-1/-1->6->5 [2] 7/-1/-1->6->5 [3] 7/-1/-1->6->5 [4] 7/-1/-1->6->5 [5] 7/-1/-1->6->5 [6] 7/-1/-1->6->5 [7] 7/-1/-1->6->5 [8] 7/-1/-1->6->5 [9] 7/-1/-1->6->5 [10] 7/-1/-1->6->5 [11] 7/-1/-1->6->5 [12] 7/-1/-1->6->5 [13] 7/-1/-1->6->5 [14] 7/-1/-1->6->5 [15] 7/-1/-1->6->5 [16] 7/-1/-1->6->5 [17] 7/-1/-1->6->5 [18] 7/-1/-1->6->5 [19] 7/-1/-1->6->5 [20] 7/-1/-1->6->5 [21] 7/-1/-1->6->5 [22] 7/-1/-1->6->5 [23] 7/-1/-1->6->5
+ip-26-0-161-153:2636089:2636551 [5] NCCL INFO Trees [0] 6/-1/-1->5->4 [1] 6/-1/-1->5->4 [2] 6/-1/-1->5->4 [3] 6/-1/-1->5->4 [4] 6/-1/-1->5->4 [5] 6/-1/-1->5->4 [6] 6/-1/-1->5->4 [7] 6/-1/-1->5->4 [8] 6/-1/-1->5->4 [9] 6/-1/-1->5->4 [10] 6/-1/-1->5->4 [11] 6/-1/-1->5->4 [12] 6/-1/-1->5->4 [13] 6/-1/-1->5->4 [14] 6/-1/-1->5->4 [15] 6/-1/-1->5->4 [16] 6/-1/-1->5->4 [17] 6/-1/-1->5->4 [18] 6/-1/-1->5->4 [19] 6/-1/-1->5->4 [20] 6/-1/-1->5->4 [21] 6/-1/-1->5->4 [22] 6/-1/-1->5->4 [23] 6/-1/-1->5->4
+ip-26-0-161-153:2636090:2636547 [6] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-161-153:2636087:2636550 [3] NCCL INFO Trees [0] 4/-1/-1->3->2 [1] 4/-1/-1->3->2 [2] 4/-1/-1->3->2 [3] 4/-1/-1->3->2 [4] 4/-1/-1->3->2 [5] 4/-1/-1->3->2 [6] 4/-1/-1->3->2 [7] 4/-1/-1->3->2 [8] 4/-1/-1->3->2 [9] 4/-1/-1->3->2 [10] 4/-1/-1->3->2 [11] 4/-1/-1->3->2 [12] 4/-1/-1->3->2 [13] 4/-1/-1->3->2 [14] 4/-1/-1->3->2 [15] 4/-1/-1->3->2 [16] 4/-1/-1->3->2 [17] 4/-1/-1->3->2 [18] 4/-1/-1->3->2 [19] 4/-1/-1->3->2 [20] 4/-1/-1->3->2 [21] 4/-1/-1->3->2 [22] 4/-1/-1->3->2 [23] 4/-1/-1->3->2
+ip-26-0-161-153:2636089:2636551 [5] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-161-153:2636088:2636549 [4] NCCL INFO Trees [0] 5/-1/-1->4->3 [1] 5/-1/-1->4->3 [2] 5/-1/-1->4->3 [3] 5/-1/-1->4->3 [4] 5/-1/-1->4->3 [5] 5/-1/-1->4->3 [6] 5/-1/-1->4->3 [7] 5/-1/-1->4->3 [8] 5/-1/-1->4->3 [9] 5/-1/-1->4->3 [10] 5/-1/-1->4->3 [11] 5/-1/-1->4->3 [12] 5/-1/-1->4->3 [13] 5/-1/-1->4->3 [14] 5/-1/-1->4->3 [15] 5/-1/-1->4->3 [16] 5/-1/-1->4->3 [17] 5/-1/-1->4->3 [18] 5/-1/-1->4->3 [19] 5/-1/-1->4->3 [20] 5/-1/-1->4->3 [21] 5/-1/-1->4->3 [22] 5/-1/-1->4->3 [23] 5/-1/-1->4->3
+ip-26-0-161-153:2636087:2636550 [3] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-161-153:2636086:2636552 [2] NCCL INFO Trees [0] 3/-1/-1->2->1 [1] 3/-1/-1->2->1 [2] 3/-1/-1->2->1 [3] 3/-1/-1->2->1 [4] 3/-1/-1->2->1 [5] 3/-1/-1->2->1 [6] 3/-1/-1->2->1 [7] 3/-1/-1->2->1 [8] 3/-1/-1->2->1 [9] 3/-1/-1->2->1 [10] 3/-1/-1->2->1 [11] 3/-1/-1->2->1 [12] 3/-1/-1->2->1 [13] 3/-1/-1->2->1 [14] 3/-1/-1->2->1 [15] 3/-1/-1->2->1 [16] 3/-1/-1->2->1 [17] 3/-1/-1->2->1 [18] 3/-1/-1->2->1 [19] 3/-1/-1->2->1 [20] 3/-1/-1->2->1 [21] 3/-1/-1->2->1 [22] 3/-1/-1->2->1 [23] 3/-1/-1->2->1
+ip-26-0-161-153:2636085:2636553 [1] NCCL INFO Trees [0] 2/-1/-1->1->0 [1] 2/-1/-1->1->0 [2] 2/-1/-1->1->0 [3] 2/-1/-1->1->0 [4] 2/-1/-1->1->0 [5] 2/-1/-1->1->0 [6] 2/-1/-1->1->0 [7] 2/-1/-1->1->0 [8] 2/-1/-1->1->0 [9] 2/-1/-1->1->0 [10] 2/-1/-1->1->0 [11] 2/-1/-1->1->0 [12] 2/-1/-1->1->0 [13] 2/-1/-1->1->0 [14] 2/-1/-1->1->0 [15] 2/-1/-1->1->0 [16] 2/-1/-1->1->0 [17] 2/-1/-1->1->0 [18] 2/-1/-1->1->0 [19] 2/-1/-1->1->0 [20] 2/-1/-1->1->0 [21] 2/-1/-1->1->0 [22] 2/-1/-1->1->0 [23] 2/-1/-1->1->0
+ip-26-0-161-153:2636088:2636549 [4] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-161-153:2636084:2636546 [0] NCCL INFO Channel 00/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-153:2636086:2636552 [2] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-161-153:2636084:2636546 [0] NCCL INFO Channel 01/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-153:2636085:2636553 [1] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-161-153:2636084:2636546 [0] NCCL INFO Channel 02/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-153:2636084:2636546 [0] NCCL INFO Channel 03/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-153:2636084:2636546 [0] NCCL INFO Channel 04/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-153:2636084:2636546 [0] NCCL INFO Channel 05/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-153:2636084:2636546 [0] NCCL INFO Channel 06/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-153:2636084:2636546 [0] NCCL INFO Channel 07/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-153:2636084:2636546 [0] NCCL INFO Channel 08/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-153:2636091:2636548 [7] NCCL INFO Trees [0] -1/-1/-1->7->6 [1] -1/-1/-1->7->6 [2] -1/-1/-1->7->6 [3] -1/-1/-1->7->6 [4] -1/-1/-1->7->6 [5] -1/-1/-1->7->6 [6] -1/-1/-1->7->6 [7] -1/-1/-1->7->6 [8] -1/-1/-1->7->6 [9] -1/-1/-1->7->6 [10] -1/-1/-1->7->6 [11] -1/-1/-1->7->6 [12] -1/-1/-1->7->6 [13] -1/-1/-1->7->6 [14] -1/-1/-1->7->6 [15] -1/-1/-1->7->6 [16] -1/-1/-1->7->6 [17] -1/-1/-1->7->6 [18] -1/-1/-1->7->6 [19] -1/-1/-1->7->6 [20] -1/-1/-1->7->6 [21] -1/-1/-1->7->6 [22] -1/-1/-1->7->6 [23] -1/-1/-1->7->6
+ip-26-0-161-153:2636084:2636546 [0] NCCL INFO Channel 09/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-153:2636084:2636546 [0] NCCL INFO Channel 10/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-153:2636084:2636546 [0] NCCL INFO Channel 11/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-153:2636084:2636546 [0] NCCL INFO Channel 12/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-153:2636091:2636548 [7] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-161-153:2636084:2636546 [0] NCCL INFO Channel 13/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-153:2636084:2636546 [0] NCCL INFO Channel 14/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-153:2636084:2636546 [0] NCCL INFO Channel 15/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-153:2636084:2636546 [0] NCCL INFO Channel 16/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-153:2636084:2636546 [0] NCCL INFO Channel 17/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-153:2636084:2636546 [0] NCCL INFO Channel 18/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-153:2636084:2636546 [0] NCCL INFO Channel 19/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-153:2636084:2636546 [0] NCCL INFO Channel 20/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-153:2636084:2636546 [0] NCCL INFO Channel 21/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-153:2636084:2636546 [0] NCCL INFO Channel 22/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-153:2636084:2636546 [0] NCCL INFO Channel 23/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-153:2636084:2636546 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] 1/-1/-1->0->-1 [7] 1/-1/-1->0->-1 [8] 1/-1/-1->0->-1 [9] 1/-1/-1->0->-1 [10] 1/-1/-1->0->-1 [11] 1/-1/-1->0->-1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] 1/-1/-1->0->-1 [19] 1/-1/-1->0->-1 [20] 1/-1/-1->0->-1 [21] 1/-1/-1->0->-1 [22] 1/-1/-1->0->-1 [23] 1/-1/-1->0->-1
+ip-26-0-161-153:2636084:2636546 [0] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-168-52:2722047:2722507 [2] NCCL INFO Channel 06/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-242:1364905:1365372 [3] NCCL INFO Channel 02/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-168-34:272150:272606 [0] NCCL INFO Channel 12/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-168-34:272153:272612 [3] NCCL INFO Channel 21/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-242:1364903:1365371 [1] NCCL INFO Channel 03/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-242:1364907:1365367 [5] NCCL INFO Channel 00/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-168-34:272154:272608 [4] NCCL INFO Channel 18/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-168-34:272156:272607 [6] NCCL INFO Channel 23/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-138:96567:97050 [4] NCCL INFO Channel 08/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-168-34:272151:272611 [1] NCCL INFO Channel 23/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-168-34:272157:272613 [7] NCCL INFO Channel 17/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-168-52:2722052:2722506 [7] NCCL INFO Channel 06/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-168-52:2722048:2722505 [3] NCCL INFO Channel 11/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-168-52:2722051:2722503 [6] NCCL INFO Channel 12/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-168-52:2722045:2722502 [0] NCCL INFO Channel 11/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-168-52:2722050:2722508 [5] NCCL INFO Channel 13/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-168-52:2722049:2722504 [4] NCCL INFO Channel 03/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-242:1364906:1365370 [4] NCCL INFO Channel 03/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-142:62882:63343 [1] NCCL INFO Channel 01/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-168-52:2722047:2722507 [2] NCCL INFO Channel 07/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-168-34:272150:272606 [0] NCCL INFO Channel 13/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-168-34:272153:272612 [3] NCCL INFO Channel 22/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-242:1364902:1365366 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-242:1364908:1365368 [6] NCCL INFO Channel 00/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-142:62883:63338 [2] NCCL INFO Channel 00/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-142:62886:63341 [5] NCCL INFO Channel 04/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-242:1364909:1365373 [7] NCCL INFO Channel 00/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-242:1364905:1365372 [3] NCCL INFO Channel 03/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-142:62882:63343 [1] NCCL INFO Channel 02/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-168-34:272154:272608 [4] NCCL INFO Channel 19/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-242:1364903:1365371 [1] NCCL INFO Channel 04/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-168-34:272157:272613 [7] NCCL INFO Channel 18/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-242:1364907:1365367 [5] NCCL INFO Channel 01/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-168-34:272150:272606 [0] NCCL INFO Channel 14/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-168-34:272153:272612 [3] NCCL INFO Channel 23/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-138:96569:97049 [6] NCCL INFO Channel 22/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-168-52:2722046:2722509 [1] NCCL INFO Channel 11/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-168-52:2722052:2722506 [7] NCCL INFO Channel 07/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-168-52:2722051:2722503 [6] NCCL INFO Channel 13/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-168-34:272154:272608 [4] NCCL INFO Channel 20/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-168-52:2722048:2722505 [3] NCCL INFO Channel 12/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-168-52:2722045:2722502 [0] NCCL INFO Channel 12/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-168-52:2722050:2722508 [5] NCCL INFO Channel 14/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-242:1364906:1365370 [4] NCCL INFO Channel 04/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-168-52:2722049:2722504 [4] NCCL INFO Channel 04/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-168-34:272157:272613 [7] NCCL INFO Channel 19/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-168-52:2722047:2722507 [2] NCCL INFO Channel 08/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-242:1364902:1365366 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-168-34:272150:272606 [0] NCCL INFO Channel 15/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-59:1557271:1557726 [5] NCCL INFO Channel 18/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-242:1364908:1365368 [6] NCCL INFO Channel 01/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-142:62882:63343 [1] NCCL INFO Channel 03/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-168-34:272154:272608 [4] NCCL INFO Channel 21/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-168-34:272157:272613 [7] NCCL INFO Channel 20/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-242:1364909:1365373 [7] NCCL INFO Channel 01/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-142:62885:63342 [4] NCCL INFO Channel 04/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-242:1364905:1365372 [3] NCCL INFO Channel 04/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-242:1364903:1365371 [1] NCCL INFO Channel 05/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-242:1364907:1365367 [5] NCCL INFO Channel 02/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-138:96567:97050 [4] NCCL INFO Channel 09/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-142:62882:63343 [1] NCCL INFO Channel 04/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-142:62885:63342 [4] NCCL INFO Channel 05/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-168-52:2722046:2722509 [1] NCCL INFO Channel 12/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-168-52:2722052:2722506 [7] NCCL INFO Channel 08/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-168-52:2722051:2722503 [6] NCCL INFO Channel 14/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-242:1364906:1365370 [4] NCCL INFO Channel 05/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-168-52:2722045:2722502 [0] NCCL INFO Channel 13/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-168-52:2722048:2722505 [3] NCCL INFO Channel 13/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-168-52:2722050:2722508 [5] NCCL INFO Channel 15/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-168-52:2722049:2722504 [4] NCCL INFO Channel 05/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-168-52:2722047:2722507 [2] NCCL INFO Channel 09/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-242:1364902:1365366 [0] NCCL INFO Channel 04/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-142:62882:63343 [1] NCCL INFO Channel 05/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-142:62887:63344 [6] NCCL INFO Channel 00/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-142:62885:63342 [4] NCCL INFO Channel 06/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-242:1364908:1365368 [6] NCCL INFO Channel 02/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-242:1364909:1365373 [7] NCCL INFO Channel 02/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-168-34:272150:272606 [0] NCCL INFO Channel 16/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-242:1364905:1365372 [3] NCCL INFO Channel 05/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-242:1364903:1365371 [1] NCCL INFO Channel 06/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-142:62883:63338 [2] NCCL INFO Channel 01/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-242:1364907:1365367 [5] NCCL INFO Channel 03/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-142:62882:63343 [1] NCCL INFO Channel 06/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-142:62887:63344 [6] NCCL INFO Channel 01/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-138:96569:97049 [6] NCCL INFO Channel 23/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-142:62885:63342 [4] NCCL INFO Channel 07/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-168-52:2722046:2722509 [1] NCCL INFO Channel 13/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-242:1364906:1365370 [4] NCCL INFO Channel 06/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-168-52:2722052:2722506 [7] NCCL INFO Channel 09/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-168-52:2722051:2722503 [6] NCCL INFO Channel 15/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-168-52:2722049:2722504 [4] NCCL INFO Channel 06/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-168-52:2722045:2722502 [0] NCCL INFO Channel 14/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-168-52:2722048:2722505 [3] NCCL INFO Channel 14/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-168-52:2722050:2722508 [5] NCCL INFO Channel 16/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-142:62883:63338 [2] NCCL INFO Channel 02/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-168-52:2722047:2722507 [2] NCCL INFO Channel 10/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-242:1364902:1365366 [0] NCCL INFO Channel 05/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-242:1364904:1365369 [2] NCCL INFO Channel 00/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-242:1364908:1365368 [6] NCCL INFO Channel 03/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-242:1364909:1365373 [7] NCCL INFO Channel 03/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-168-34:272157:272613 [7] NCCL INFO Channel 21/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-168-34:272154:272608 [4] NCCL INFO Channel 22/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-142:62882:63343 [1] NCCL INFO Channel 07/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-242:1364903:1365371 [1] NCCL INFO Channel 07/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-242:1364905:1365372 [3] NCCL INFO Channel 06/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-142:62887:63344 [6] NCCL INFO Channel 02/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-142:62885:63342 [4] NCCL INFO Channel 08/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-242:1364907:1365367 [5] NCCL INFO Channel 04/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-138:96567:97050 [4] NCCL INFO Channel 10/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-142:62883:63338 [2] NCCL INFO Channel 03/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-168-52:2722046:2722509 [1] NCCL INFO Channel 14/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-168-52:2722052:2722506 [7] NCCL INFO Channel 10/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-168-52:2722051:2722503 [6] NCCL INFO Channel 16/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-168-52:2722049:2722504 [4] NCCL INFO Channel 07/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-168-52:2722045:2722502 [0] NCCL INFO Channel 15/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-142:62882:63343 [1] NCCL INFO Channel 08/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-168-52:2722048:2722505 [3] NCCL INFO Channel 15/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-168-52:2722050:2722508 [5] NCCL INFO Channel 17/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-242:1364906:1365370 [4] NCCL INFO Channel 07/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-142:62885:63342 [4] NCCL INFO Channel 09/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-142:62887:63344 [6] NCCL INFO Channel 03/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-168-52:2722047:2722507 [2] NCCL INFO Channel 11/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-242:1364902:1365366 [0] NCCL INFO Channel 06/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-142:62883:63338 [2] NCCL INFO Channel 04/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-242:1364904:1365369 [2] NCCL INFO Channel 01/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-242:1364908:1365368 [6] NCCL INFO Channel 04/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-242:1364909:1365373 [7] NCCL INFO Channel 04/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-242:1364903:1365371 [1] NCCL INFO Channel 08/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-242:1364905:1365372 [3] NCCL INFO Channel 07/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-242:1364907:1365367 [5] NCCL INFO Channel 05/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-138:96566:97047 [3] NCCL INFO Channel 21/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-138:96565:97044 [2] NCCL INFO Channel 23/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-142:62886:63341 [5] NCCL INFO Channel 05/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-142:62882:63343 [1] NCCL INFO Channel 09/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-142:62887:63344 [6] NCCL INFO Channel 04/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-142:62885:63342 [4] NCCL INFO Channel 10/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-168-52:2722046:2722509 [1] NCCL INFO Channel 15/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-168-52:2722052:2722506 [7] NCCL INFO Channel 11/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-168-52:2722049:2722504 [4] NCCL INFO Channel 08/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-142:62883:63338 [2] NCCL INFO Channel 05/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-168-52:2722051:2722503 [6] NCCL INFO Channel 17/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-168-52:2722050:2722508 [5] NCCL INFO Channel 18/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-168-52:2722048:2722505 [3] NCCL INFO Channel 16/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-168-52:2722045:2722502 [0] NCCL INFO Channel 16/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-168-52:2722047:2722507 [2] NCCL INFO Channel 12/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-168-34:272150:272606 [0] NCCL INFO Channel 17/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-242:1364906:1365370 [4] NCCL INFO Channel 08/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-242:1364902:1365366 [0] NCCL INFO Channel 07/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-242:1364904:1365369 [2] NCCL INFO Channel 02/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-242:1364908:1365368 [6] NCCL INFO Channel 05/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-242:1364909:1365373 [7] NCCL INFO Channel 05/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-242:1364903:1365371 [1] NCCL INFO Channel 09/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-242:1364905:1365372 [3] NCCL INFO Channel 08/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-242:1364907:1365367 [5] NCCL INFO Channel 06/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-142:62886:63341 [5] NCCL INFO Channel 06/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-142:62882:63343 [1] NCCL INFO Channel 10/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-142:62887:63344 [6] NCCL INFO Channel 05/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-142:62885:63342 [4] NCCL INFO Channel 11/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-142:62883:63338 [2] NCCL INFO Channel 06/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-168-34:272154:272608 [4] NCCL INFO Channel 23/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-142:62888:63340 [7] NCCL INFO Channel 00/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-242:1364906:1365370 [4] NCCL INFO Channel 09/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-142:62884:63339 [3] NCCL INFO Channel 00/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-168-52:2722046:2722509 [1] NCCL INFO Channel 16/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-168-52:2722047:2722507 [2] NCCL INFO Channel 13/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-242:1364902:1365366 [0] NCCL INFO Channel 08/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-242:1364904:1365369 [2] NCCL INFO Channel 03/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-242:1364908:1365368 [6] NCCL INFO Channel 06/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-242:1364909:1365373 [7] NCCL INFO Channel 06/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-78:114512:114968 [0] NCCL INFO Channel 19/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-78:114517:114969 [5] NCCL INFO Channel 17/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-242:1364903:1365371 [1] NCCL INFO Channel 10/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-242:1364905:1365372 [3] NCCL INFO Channel 09/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-168-52:2722049:2722504 [4] NCCL INFO Channel 09/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-168-52:2722052:2722506 [7] NCCL INFO Channel 12/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-242:1364907:1365367 [5] NCCL INFO Channel 07/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-168-52:2722046:2722509 [1] NCCL INFO Channel 17/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-168-52:2722047:2722507 [2] NCCL INFO Channel 14/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-142:62886:63341 [5] NCCL INFO Channel 07/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-142:62882:63343 [1] NCCL INFO Channel 11/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-142:62887:63344 [6] NCCL INFO Channel 06/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-142:62885:63342 [4] NCCL INFO Channel 12/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-168-52:2722049:2722504 [4] NCCL INFO Channel 10/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-168-34:272157:272613 [7] NCCL INFO Channel 22/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-142:62883:63338 [2] NCCL INFO Channel 07/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-168-52:2722052:2722506 [7] NCCL INFO Channel 13/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-168-34:272150:272606 [0] NCCL INFO Channel 18/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-142:62888:63340 [7] NCCL INFO Channel 01/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-142:62881:63337 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-78:114515:114972 [3] NCCL INFO Channel 20/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-242:1364906:1365370 [4] NCCL INFO Channel 10/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-142:62884:63339 [3] NCCL INFO Channel 01/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-168-52:2722046:2722509 [1] NCCL INFO Channel 18/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-168-52:2722047:2722507 [2] NCCL INFO Channel 15/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-242:1364902:1365366 [0] NCCL INFO Channel 09/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-242:1364904:1365369 [2] NCCL INFO Channel 04/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-242:1364908:1365368 [6] NCCL INFO Channel 07/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-242:1364909:1365373 [7] NCCL INFO Channel 07/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-168-52:2722049:2722504 [4] NCCL INFO Channel 11/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-242:1364903:1365371 [1] NCCL INFO Channel 11/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-168-52:2722052:2722506 [7] NCCL INFO Channel 14/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-242:1364905:1365372 [3] NCCL INFO Channel 10/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-78:114512:114968 [0] NCCL INFO Channel 20/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-242:1364907:1365367 [5] NCCL INFO Channel 08/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-78:114517:114969 [5] NCCL INFO Channel 18/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-168-52:2722046:2722509 [1] NCCL INFO Channel 19/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-142:62886:63341 [5] NCCL INFO Channel 08/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-168-52:2722047:2722507 [2] NCCL INFO Channel 16/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-142:62887:63344 [6] NCCL INFO Channel 07/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-142:62882:63343 [1] NCCL INFO Channel 12/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-142:62885:63342 [4] NCCL INFO Channel 13/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-78:114515:114972 [3] NCCL INFO Channel 21/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-168-52:2722049:2722504 [4] NCCL INFO Channel 12/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-142:62883:63338 [2] NCCL INFO Channel 08/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-168-52:2722052:2722506 [7] NCCL INFO Channel 15/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-142:62888:63340 [7] NCCL INFO Channel 02/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-142:62881:63337 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-142:62884:63339 [3] NCCL INFO Channel 02/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-242:1364906:1365370 [4] NCCL INFO Channel 11/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-138:96566:97047 [3] NCCL INFO Channel 22/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-242:1364902:1365366 [0] NCCL INFO Channel 10/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-242:1364904:1365369 [2] NCCL INFO Channel 05/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-242:1364908:1365368 [6] NCCL INFO Channel 08/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-242:1364909:1365373 [7] NCCL INFO Channel 08/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-242:1364903:1365371 [1] NCCL INFO Channel 12/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-242:1364905:1365372 [3] NCCL INFO Channel 11/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-78:114517:114969 [5] NCCL INFO Channel 19/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-78:114512:114968 [0] NCCL INFO Channel 21/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-242:1364907:1365367 [5] NCCL INFO Channel 09/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-142:62886:63341 [5] NCCL INFO Channel 09/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-142:62887:63344 [6] NCCL INFO Channel 08/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-142:62882:63343 [1] NCCL INFO Channel 13/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-142:62885:63342 [4] NCCL INFO Channel 14/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-168-52:2722047:2722507 [2] NCCL INFO Channel 17/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-225:124161:124621 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-160-225:124161:124621 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-161-142:62883:63338 [2] NCCL INFO Channel 09/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-168-52:2722050:2722508 [5] NCCL INFO Channel 19/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-168-52:2722051:2722503 [6] NCCL INFO Channel 18/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-142:62888:63340 [7] NCCL INFO Channel 03/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-142:62881:63337 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-142:62884:63339 [3] NCCL INFO Channel 03/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-242:1364906:1365370 [4] NCCL INFO Channel 12/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-242:1364902:1365366 [0] NCCL INFO Channel 11/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-242:1364904:1365369 [2] NCCL INFO Channel 06/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-242:1364908:1365368 [6] NCCL INFO Channel 09/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-168-52:2722047:2722507 [2] NCCL INFO Channel 18/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-168-52:2722045:2722502 [0] NCCL INFO Channel 17/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-242:1364909:1365373 [7] NCCL INFO Channel 09/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-242:1364903:1365371 [1] NCCL INFO Channel 13/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-242:1364905:1365372 [3] NCCL INFO Channel 12/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-242:1364907:1365367 [5] NCCL INFO Channel 10/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-142:62886:63341 [5] NCCL INFO Channel 10/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-142:62887:63344 [6] NCCL INFO Channel 09/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-142:62882:63343 [1] NCCL INFO Channel 14/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-142:62885:63342 [4] NCCL INFO Channel 15/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-142:62883:63338 [2] NCCL INFO Channel 10/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-142:62888:63340 [7] NCCL INFO Channel 04/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-142:62881:63337 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-142:62884:63339 [3] NCCL INFO Channel 04/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-242:1364906:1365370 [4] NCCL INFO Channel 13/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-242:1364902:1365366 [0] NCCL INFO Channel 12/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-168-52:2722050:2722508 [5] NCCL INFO Channel 20/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-242:1364904:1365369 [2] NCCL INFO Channel 07/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-242:1364909:1365373 [7] NCCL INFO Channel 10/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-242:1364908:1365368 [6] NCCL INFO Channel 10/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-168-52:2722046:2722509 [1] NCCL INFO Channel 20/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-242:1364905:1365372 [3] NCCL INFO Channel 13/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-242:1364903:1365371 [1] NCCL INFO Channel 14/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-242:1364907:1365367 [5] NCCL INFO Channel 11/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-168-52:2722052:2722506 [7] NCCL INFO Channel 16/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-168-52:2722049:2722504 [4] NCCL INFO Channel 13/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-168-34:272157:272613 [7] NCCL INFO Channel 23/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-168-34:272150:272606 [0] NCCL INFO Channel 19/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-142:62887:63344 [6] NCCL INFO Channel 10/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-142:62886:63341 [5] NCCL INFO Channel 11/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-142:62882:63343 [1] NCCL INFO Channel 15/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-168-52:2722050:2722508 [5] NCCL INFO Channel 21/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-165-59:1557271:1557726 [5] NCCL INFO Channel 19/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-142:62885:63342 [4] NCCL INFO Channel 16/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-168-52:2722048:2722505 [3] NCCL INFO Channel 17/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-168-52:2722046:2722509 [1] NCCL INFO Channel 21/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-142:62883:63338 [2] NCCL INFO Channel 11/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-142:62888:63340 [7] NCCL INFO Channel 05/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-138:96567:97050 [4] NCCL INFO Channel 11/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-142:62881:63337 [0] NCCL INFO Channel 04/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-142:62884:63339 [3] NCCL INFO Channel 05/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-242:1364906:1365370 [4] NCCL INFO Channel 14/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-242:1364902:1365366 [0] NCCL INFO Channel 13/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-242:1364904:1365369 [2] NCCL INFO Channel 08/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-242:1364908:1365368 [6] NCCL INFO Channel 11/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-242:1364909:1365373 [7] NCCL INFO Channel 11/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-242:1364905:1365372 [3] NCCL INFO Channel 14/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-242:1364903:1365371 [1] NCCL INFO Channel 15/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-242:1364907:1365367 [5] NCCL INFO Channel 12/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-168-52:2722051:2722503 [6] NCCL INFO Channel 19/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-168-52:2722049:2722504 [4] NCCL INFO Channel 14/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-142:62887:63344 [6] NCCL INFO Channel 11/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-142:62886:63341 [5] NCCL INFO Channel 12/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-142:62882:63343 [1] NCCL INFO Channel 16/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-142:62885:63342 [4] NCCL INFO Channel 17/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-142:62888:63340 [7] NCCL INFO Channel 06/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-142:62883:63338 [2] NCCL INFO Channel 12/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-142:62881:63337 [0] NCCL INFO Channel 05/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-168-52:2722047:2722507 [2] NCCL INFO Channel 19/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-168-52:2722045:2722502 [0] NCCL INFO Channel 18/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-142:62884:63339 [3] NCCL INFO Channel 06/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-168-52:2722051:2722503 [6] NCCL INFO Channel 20/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-168-52:2722049:2722504 [4] NCCL INFO Channel 15/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-242:1364906:1365370 [4] NCCL INFO Channel 15/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-242:1364902:1365366 [0] NCCL INFO Channel 14/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-242:1364904:1365369 [2] NCCL INFO Channel 09/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-242:1364909:1365373 [7] NCCL INFO Channel 12/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-242:1364908:1365368 [6] NCCL INFO Channel 12/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-242:1364903:1365371 [1] NCCL INFO Channel 16/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-242:1364905:1365372 [3] NCCL INFO Channel 15/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-242:1364907:1365367 [5] NCCL INFO Channel 13/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-168-52:2722047:2722507 [2] NCCL INFO Channel 20/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-168-52:2722045:2722502 [0] NCCL INFO Channel 19/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-138:96566:97047 [3] NCCL INFO Channel 23/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-168-52:2722051:2722503 [6] NCCL INFO Channel 21/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-168-52:2722049:2722504 [4] NCCL INFO Channel 16/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-142:62887:63344 [6] NCCL INFO Channel 12/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-142:62886:63341 [5] NCCL INFO Channel 13/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-142:62885:63342 [4] NCCL INFO Channel 18/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-142:62882:63343 [1] NCCL INFO Channel 17/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-168-52:2722047:2722507 [2] NCCL INFO Channel 21/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-168-52:2722045:2722502 [0] NCCL INFO Channel 20/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-142:62888:63340 [7] NCCL INFO Channel 07/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-142:62883:63338 [2] NCCL INFO Channel 13/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-142:62881:63337 [0] NCCL INFO Channel 06/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-142:62884:63339 [3] NCCL INFO Channel 07/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-168-52:2722051:2722503 [6] NCCL INFO Channel 22/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-168-52:2722049:2722504 [4] NCCL INFO Channel 17/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-242:1364906:1365370 [4] NCCL INFO Channel 16/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-242:1364902:1365366 [0] NCCL INFO Channel 15/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-242:1364904:1365369 [2] NCCL INFO Channel 10/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-242:1364909:1365373 [7] NCCL INFO Channel 13/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-242:1364908:1365368 [6] NCCL INFO Channel 13/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-242:1364903:1365371 [1] NCCL INFO Channel 17/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-168-52:2722047:2722507 [2] NCCL INFO Channel 22/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-242:1364905:1365372 [3] NCCL INFO Channel 16/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-242:1364907:1365367 [5] NCCL INFO Channel 14/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-168-52:2722045:2722502 [0] NCCL INFO Channel 21/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-168-52:2722051:2722503 [6] NCCL INFO Channel 23/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-168-52:2722049:2722504 [4] NCCL INFO Channel 18/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-142:62887:63344 [6] NCCL INFO Channel 13/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-142:62886:63341 [5] NCCL INFO Channel 14/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-168-52:2722047:2722507 [2] NCCL INFO Channel 23/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-142:62885:63342 [4] NCCL INFO Channel 19/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-142:62882:63343 [1] NCCL INFO Channel 18/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-168-52:2722045:2722502 [0] NCCL INFO Channel 22/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-142:62888:63340 [7] NCCL INFO Channel 08/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-142:62883:63338 [2] NCCL INFO Channel 14/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-142:62884:63339 [3] NCCL INFO Channel 08/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-142:62881:63337 [0] NCCL INFO Channel 07/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-168-52:2722049:2722504 [4] NCCL INFO Channel 19/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-242:1364906:1365370 [4] NCCL INFO Channel 17/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-168-52:2722045:2722502 [0] NCCL INFO Channel 23/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-242:1364902:1365366 [0] NCCL INFO Channel 16/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-242:1364904:1365369 [2] NCCL INFO Channel 11/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-242:1364909:1365373 [7] NCCL INFO Channel 14/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-242:1364908:1365368 [6] NCCL INFO Channel 14/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-242:1364903:1365371 [1] NCCL INFO Channel 18/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-242:1364907:1365367 [5] NCCL INFO Channel 15/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-242:1364905:1365372 [3] NCCL INFO Channel 17/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-142:62887:63344 [6] NCCL INFO Channel 14/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-142:62886:63341 [5] NCCL INFO Channel 15/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-142:62885:63342 [4] NCCL INFO Channel 20/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-142:62882:63343 [1] NCCL INFO Channel 19/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-142:62888:63340 [7] NCCL INFO Channel 09/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-142:62884:63339 [3] NCCL INFO Channel 09/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-142:62881:63337 [0] NCCL INFO Channel 08/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-142:62883:63338 [2] NCCL INFO Channel 15/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-168-34:272150:272606 [0] NCCL INFO Channel 20/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-168-52:2722052:2722506 [7] NCCL INFO Channel 17/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-168-52:2722049:2722504 [4] NCCL INFO Channel 20/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-242:1364906:1365370 [4] NCCL INFO Channel 18/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-242:1364902:1365366 [0] NCCL INFO Channel 17/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-242:1364904:1365369 [2] NCCL INFO Channel 12/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-242:1364909:1365373 [7] NCCL INFO Channel 15/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-242:1364908:1365368 [6] NCCL INFO Channel 15/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-242:1364903:1365371 [1] NCCL INFO Channel 19/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-242:1364907:1365367 [5] NCCL INFO Channel 16/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-168-52:2722050:2722508 [5] NCCL INFO Channel 22/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-242:1364905:1365372 [3] NCCL INFO Channel 18/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-168-52:2722046:2722509 [1] NCCL INFO Channel 22/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-168-52:2722052:2722506 [7] NCCL INFO Channel 18/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-168-52:2722049:2722504 [4] NCCL INFO Channel 21/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-142:62887:63344 [6] NCCL INFO Channel 15/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-142:62886:63341 [5] NCCL INFO Channel 16/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-142:62885:63342 [4] NCCL INFO Channel 21/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-142:62882:63343 [1] NCCL INFO Channel 20/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-142:62888:63340 [7] NCCL INFO Channel 10/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-142:62884:63339 [3] NCCL INFO Channel 10/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-142:62881:63337 [0] NCCL INFO Channel 09/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-142:62883:63338 [2] NCCL INFO Channel 16/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-168-30:235654:236107 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-168-52:2722050:2722508 [5] NCCL INFO Channel 23/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-168-52:2722046:2722509 [1] NCCL INFO Channel 23/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-168-30:235654:236107 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-168-52:2722052:2722506 [7] NCCL INFO Channel 19/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-168-52:2722049:2722504 [4] NCCL INFO Channel 22/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-242:1364906:1365370 [4] NCCL INFO Channel 19/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-242:1364902:1365366 [0] NCCL INFO Channel 18/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-242:1364904:1365369 [2] NCCL INFO Channel 13/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-242:1364909:1365373 [7] NCCL INFO Channel 16/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-242:1364908:1365368 [6] NCCL INFO Channel 16/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-242:1364903:1365371 [1] NCCL INFO Channel 20/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-242:1364907:1365367 [5] NCCL INFO Channel 17/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-242:1364905:1365372 [3] NCCL INFO Channel 19/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-168-52:2722052:2722506 [7] NCCL INFO Channel 20/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-168-52:2722049:2722504 [4] NCCL INFO Channel 23/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-142:62887:63344 [6] NCCL INFO Channel 16/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-142:62886:63341 [5] NCCL INFO Channel 17/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-142:62885:63342 [4] NCCL INFO Channel 22/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-142:62888:63340 [7] NCCL INFO Channel 11/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-142:62882:63343 [1] NCCL INFO Channel 21/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-138:96567:97050 [4] NCCL INFO Channel 12/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-142:62884:63339 [3] NCCL INFO Channel 11/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-142:62883:63338 [2] NCCL INFO Channel 17/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-142:62881:63337 [0] NCCL INFO Channel 10/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-168-52:2722048:2722505 [3] NCCL INFO Channel 18/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-168-52:2722052:2722506 [7] NCCL INFO Channel 21/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-242:1364906:1365370 [4] NCCL INFO Channel 20/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-242:1364902:1365366 [0] NCCL INFO Channel 19/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-242:1364904:1365369 [2] NCCL INFO Channel 14/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-242:1364909:1365373 [7] NCCL INFO Channel 17/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-242:1364908:1365368 [6] NCCL INFO Channel 17/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-242:1364903:1365371 [1] NCCL INFO Channel 21/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-242:1364907:1365367 [5] NCCL INFO Channel 18/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-242:1364905:1365372 [3] NCCL INFO Channel 20/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-142:62887:63344 [6] NCCL INFO Channel 17/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-142:62886:63341 [5] NCCL INFO Channel 18/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-142:62885:63342 [4] NCCL INFO Channel 23/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-142:62888:63340 [7] NCCL INFO Channel 12/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-142:62882:63343 [1] NCCL INFO Channel 22/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-142:62884:63339 [3] NCCL INFO Channel 12/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-142:62883:63338 [2] NCCL INFO Channel 18/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-142:62881:63337 [0] NCCL INFO Channel 11/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-242:1364906:1365370 [4] NCCL INFO Channel 21/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-242:1364902:1365366 [0] NCCL INFO Channel 20/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-242:1364903:1365371 [1] NCCL INFO Channel 22/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-242:1364908:1365368 [6] NCCL INFO Channel 18/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-242:1364904:1365369 [2] NCCL INFO Channel 15/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-242:1364909:1365373 [7] NCCL INFO Channel 18/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-242:1364907:1365367 [5] NCCL INFO Channel 19/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-242:1364905:1365372 [3] NCCL INFO Channel 21/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-142:62887:63344 [6] NCCL INFO Channel 18/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-142:62886:63341 [5] NCCL INFO Channel 19/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-142:62888:63340 [7] NCCL INFO Channel 13/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-142:62882:63343 [1] NCCL INFO Channel 23/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-142:62884:63339 [3] NCCL INFO Channel 13/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-142:62883:63338 [2] NCCL INFO Channel 19/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-142:62881:63337 [0] NCCL INFO Channel 12/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-168-52:2722048:2722505 [3] NCCL INFO Channel 19/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-168-52:2722052:2722506 [7] NCCL INFO Channel 22/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-242:1364906:1365370 [4] NCCL INFO Channel 22/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-242:1364902:1365366 [0] NCCL INFO Channel 21/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-242:1364903:1365371 [1] NCCL INFO Channel 23/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-242:1364908:1365368 [6] NCCL INFO Channel 19/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-242:1364904:1365369 [2] NCCL INFO Channel 16/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-242:1364909:1365373 [7] NCCL INFO Channel 19/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-242:1364907:1365367 [5] NCCL INFO Channel 20/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-242:1364905:1365372 [3] NCCL INFO Channel 22/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-142:62887:63344 [6] NCCL INFO Channel 19/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-142:62886:63341 [5] NCCL INFO Channel 20/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-142:62888:63340 [7] NCCL INFO Channel 14/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-142:62884:63339 [3] NCCL INFO Channel 14/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-142:62881:63337 [0] NCCL INFO Channel 13/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-142:62883:63338 [2] NCCL INFO Channel 20/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-164:1227466:1227921 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-165-164:1227466:1227921 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-160-242:1364906:1365370 [4] NCCL INFO Channel 23/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-242:1364902:1365366 [0] NCCL INFO Channel 22/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-242:1364908:1365368 [6] NCCL INFO Channel 20/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-242:1364904:1365369 [2] NCCL INFO Channel 17/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-242:1364909:1365373 [7] NCCL INFO Channel 20/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-242:1364907:1365367 [5] NCCL INFO Channel 21/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-142:62887:63344 [6] NCCL INFO Channel 20/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-142:62888:63340 [7] NCCL INFO Channel 15/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-142:62884:63339 [3] NCCL INFO Channel 15/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-142:62881:63337 [0] NCCL INFO Channel 14/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-142:62886:63341 [5] NCCL INFO Channel 21/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-142:62883:63338 [2] NCCL INFO Channel 21/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-242:1364905:1365372 [3] NCCL INFO Channel 23/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-59:1557271:1557726 [5] NCCL INFO Channel 20/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-142:62887:63344 [6] NCCL INFO Channel 21/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-142:62888:63340 [7] NCCL INFO Channel 16/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-242:1364902:1365366 [0] NCCL INFO Channel 23/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-142:62884:63339 [3] NCCL INFO Channel 16/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-142:62881:63337 [0] NCCL INFO Channel 15/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-142:62886:63341 [5] NCCL INFO Channel 22/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-142:62883:63338 [2] NCCL INFO Channel 22/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-242:1364904:1365369 [2] NCCL INFO Channel 18/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-242:1364908:1365368 [6] NCCL INFO Channel 21/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-242:1364907:1365367 [5] NCCL INFO Channel 22/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-242:1364909:1365373 [7] NCCL INFO Channel 21/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-242:1364904:1365369 [2] NCCL INFO Channel 19/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-242:1364907:1365367 [5] NCCL INFO Channel 23/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-242:1364909:1365373 [7] NCCL INFO Channel 22/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-242:1364908:1365368 [6] NCCL INFO Channel 22/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-142:62887:63344 [6] NCCL INFO Channel 22/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-142:62888:63340 [7] NCCL INFO Channel 17/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-142:62881:63337 [0] NCCL INFO Channel 16/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-142:62884:63339 [3] NCCL INFO Channel 17/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-142:62886:63341 [5] NCCL INFO Channel 23/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-142:62883:63338 [2] NCCL INFO Channel 23/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-242:1364904:1365369 [2] NCCL INFO Channel 20/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-168-52:2722048:2722505 [3] NCCL INFO Channel 20/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-168-52:2722052:2722506 [7] NCCL INFO Channel 23/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-242:1364909:1365373 [7] NCCL INFO Channel 23/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-242:1364908:1365368 [6] NCCL INFO Channel 23/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-142:62887:63344 [6] NCCL INFO Channel 23/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-142:62888:63340 [7] NCCL INFO Channel 18/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-142:62881:63337 [0] NCCL INFO Channel 17/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-142:62884:63339 [3] NCCL INFO Channel 18/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-142:62888:63340 [7] NCCL INFO Channel 19/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-142:62884:63339 [3] NCCL INFO Channel 19/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-142:62881:63337 [0] NCCL INFO Channel 18/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-242:1364904:1365369 [2] NCCL INFO Channel 21/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-103:96891:97347 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-161-103:96891:97347 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-161-142:62888:63340 [7] NCCL INFO Channel 20/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-142:62884:63339 [3] NCCL INFO Channel 20/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-142:62881:63337 [0] NCCL INFO Channel 19/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-225:124164:124623 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-160-225:124164:124623 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-165-164:1227463:1227919 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-165-164:1227463:1227919 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-161-138:96567:97050 [4] NCCL INFO Channel 13/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-153:2636086:2636552 [2] NCCL INFO Channel 00/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-153:2636089:2636551 [5] NCCL INFO Channel 00/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-168-34:272150:272606 [0] NCCL INFO Channel 21/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-168-52:2722048:2722505 [3] NCCL INFO Channel 21/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-153:2636086:2636552 [2] NCCL INFO Channel 01/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-153:2636085:2636553 [1] NCCL INFO Channel 00/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-153:2636089:2636551 [5] NCCL INFO Channel 01/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-165-59:1557271:1557726 [5] NCCL INFO Channel 21/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-142:62888:63340 [7] NCCL INFO Channel 21/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-142:62881:63337 [0] NCCL INFO Channel 20/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-142:62884:63339 [3] NCCL INFO Channel 21/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-153:2636085:2636553 [1] NCCL INFO Channel 01/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-153:2636086:2636552 [2] NCCL INFO Channel 02/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-153:2636089:2636551 [5] NCCL INFO Channel 02/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-153:2636088:2636549 [4] NCCL INFO Channel 00/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-153:2636085:2636553 [1] NCCL INFO Channel 02/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-78:114515:114972 [3] NCCL INFO Channel 22/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-153:2636086:2636552 [2] NCCL INFO Channel 03/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-153:2636089:2636551 [5] NCCL INFO Channel 03/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-153:2636088:2636549 [4] NCCL INFO Channel 01/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-142:62888:63340 [7] NCCL INFO Channel 22/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-142:62881:63337 [0] NCCL INFO Channel 21/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-142:62884:63339 [3] NCCL INFO Channel 22/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-153:2636086:2636552 [2] NCCL INFO Channel 04/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-168-30:235649:236109 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-168-30:235649:236109 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-160-242:1364904:1365369 [2] NCCL INFO Channel 22/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-153:2636089:2636551 [5] NCCL INFO Channel 04/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-78:114517:114969 [5] NCCL INFO Channel 20/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-78:114512:114968 [0] NCCL INFO Channel 22/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-153:2636087:2636550 [3] NCCL INFO Channel 00/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-153:2636088:2636549 [4] NCCL INFO Channel 02/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-59:1557271:1557726 [5] NCCL INFO Channel 22/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-153:2636085:2636553 [1] NCCL INFO Channel 03/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-153:2636086:2636552 [2] NCCL INFO Channel 05/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-142:62888:63340 [7] NCCL INFO Channel 23/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-153:2636089:2636551 [5] NCCL INFO Channel 05/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-142:62881:63337 [0] NCCL INFO Channel 22/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-138:96567:97050 [4] NCCL INFO Channel 14/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-168-34:272150:272606 [0] NCCL INFO Channel 22/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-142:62884:63339 [3] NCCL INFO Channel 23/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-142:62881:63337 [0] NCCL INFO Channel 23/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-153:2636087:2636550 [3] NCCL INFO Channel 01/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-153:2636088:2636549 [4] NCCL INFO Channel 03/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-78:114515:114972 [3] NCCL INFO Channel 23/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-78:114517:114969 [5] NCCL INFO Channel 21/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-153:2636085:2636553 [1] NCCL INFO Channel 04/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-168-52:2722048:2722505 [3] NCCL INFO Channel 22/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-153:2636086:2636552 [2] NCCL INFO Channel 06/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-153:2636089:2636551 [5] NCCL INFO Channel 06/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-153:2636087:2636550 [3] NCCL INFO Channel 02/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-153:2636088:2636549 [4] NCCL INFO Channel 04/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-225:124167:124624 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-160-225:124167:124624 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-161-153:2636085:2636553 [1] NCCL INFO Channel 05/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-153:2636086:2636552 [2] NCCL INFO Channel 07/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-153:2636089:2636551 [5] NCCL INFO Channel 07/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-153:2636087:2636550 [3] NCCL INFO Channel 03/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-153:2636088:2636549 [4] NCCL INFO Channel 05/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-168-34:272150:272606 [0] NCCL INFO Channel 23/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-153:2636085:2636553 [1] NCCL INFO Channel 06/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-153:2636086:2636552 [2] NCCL INFO Channel 08/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-153:2636089:2636551 [5] NCCL INFO Channel 08/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-153:2636090:2636547 [6] NCCL INFO Channel 00/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-153:2636084:2636546 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-153:2636087:2636550 [3] NCCL INFO Channel 04/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-153:2636088:2636549 [4] NCCL INFO Channel 06/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-153:2636085:2636553 [1] NCCL INFO Channel 07/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-153:2636086:2636552 [2] NCCL INFO Channel 09/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-225:124166:124622 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-160-225:124166:124622 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-161-153:2636089:2636551 [5] NCCL INFO Channel 09/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-153:2636090:2636547 [6] NCCL INFO Channel 01/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-225:124166:124622 [6] NCCL INFO Trees [0] 7/-1/-1->6->5 [1] 7/-1/-1->6->5 [2] 7/-1/-1->6->5 [3] 7/-1/-1->6->5 [4] 7/-1/-1->6->5 [5] 7/-1/-1->6->5 [6] 7/-1/-1->6->5 [7] 7/-1/-1->6->5 [8] 7/-1/-1->6->5 [9] 7/-1/-1->6->5 [10] 7/-1/-1->6->5 [11] 7/-1/-1->6->5 [12] 7/-1/-1->6->5 [13] 7/-1/-1->6->5 [14] 7/-1/-1->6->5 [15] 7/-1/-1->6->5 [16] 7/-1/-1->6->5 [17] 7/-1/-1->6->5 [18] 7/-1/-1->6->5 [19] 7/-1/-1->6->5 [20] 7/-1/-1->6->5 [21] 7/-1/-1->6->5 [22] 7/-1/-1->6->5 [23] 7/-1/-1->6->5
+ip-26-0-160-225:124166:124622 [6] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-160-225:124165:124625 [5] NCCL INFO Trees [0] 6/-1/-1->5->4 [1] 6/-1/-1->5->4 [2] 6/-1/-1->5->4 [3] 6/-1/-1->5->4 [4] 6/-1/-1->5->4 [5] 6/-1/-1->5->4 [6] 6/-1/-1->5->4 [7] 6/-1/-1->5->4 [8] 6/-1/-1->5->4 [9] 6/-1/-1->5->4 [10] 6/-1/-1->5->4 [11] 6/-1/-1->5->4 [12] 6/-1/-1->5->4 [13] 6/-1/-1->5->4 [14] 6/-1/-1->5->4 [15] 6/-1/-1->5->4 [16] 6/-1/-1->5->4 [17] 6/-1/-1->5->4 [18] 6/-1/-1->5->4 [19] 6/-1/-1->5->4 [20] 6/-1/-1->5->4 [21] 6/-1/-1->5->4 [22] 6/-1/-1->5->4 [23] 6/-1/-1->5->4
+ip-26-0-160-225:124165:124625 [5] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-160-225:124164:124623 [4] NCCL INFO Trees [0] 5/-1/-1->4->3 [1] 5/-1/-1->4->3 [2] 5/-1/-1->4->3 [3] 5/-1/-1->4->3 [4] 5/-1/-1->4->3 [5] 5/-1/-1->4->3 [6] 5/-1/-1->4->3 [7] 5/-1/-1->4->3 [8] 5/-1/-1->4->3 [9] 5/-1/-1->4->3 [10] 5/-1/-1->4->3 [11] 5/-1/-1->4->3 [12] 5/-1/-1->4->3 [13] 5/-1/-1->4->3 [14] 5/-1/-1->4->3 [15] 5/-1/-1->4->3 [16] 5/-1/-1->4->3 [17] 5/-1/-1->4->3 [18] 5/-1/-1->4->3 [19] 5/-1/-1->4->3 [20] 5/-1/-1->4->3 [21] 5/-1/-1->4->3 [22] 5/-1/-1->4->3 [23] 5/-1/-1->4->3
+ip-26-0-160-225:124164:124623 [4] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-160-225:124162:124627 [2] NCCL INFO Trees [0] 3/-1/-1->2->1 [1] 3/-1/-1->2->1 [2] 3/-1/-1->2->1 [3] 3/-1/-1->2->1 [4] 3/-1/-1->2->1 [5] 3/-1/-1->2->1 [6] 3/-1/-1->2->1 [7] 3/-1/-1->2->1 [8] 3/-1/-1->2->1 [9] 3/-1/-1->2->1 [10] 3/-1/-1->2->1 [11] 3/-1/-1->2->1 [12] 3/-1/-1->2->1 [13] 3/-1/-1->2->1 [14] 3/-1/-1->2->1 [15] 3/-1/-1->2->1 [16] 3/-1/-1->2->1 [17] 3/-1/-1->2->1 [18] 3/-1/-1->2->1 [19] 3/-1/-1->2->1 [20] 3/-1/-1->2->1 [21] 3/-1/-1->2->1 [22] 3/-1/-1->2->1 [23] 3/-1/-1->2->1
+ip-26-0-160-225:124163:124626 [3] NCCL INFO Trees [0] 4/-1/-1->3->2 [1] 4/-1/-1->3->2 [2] 4/-1/-1->3->2 [3] 4/-1/-1->3->2 [4] 4/-1/-1->3->2 [5] 4/-1/-1->3->2 [6] 4/-1/-1->3->2 [7] 4/-1/-1->3->2 [8] 4/-1/-1->3->2 [9] 4/-1/-1->3->2 [10] 4/-1/-1->3->2 [11] 4/-1/-1->3->2 [12] 4/-1/-1->3->2 [13] 4/-1/-1->3->2 [14] 4/-1/-1->3->2 [15] 4/-1/-1->3->2 [16] 4/-1/-1->3->2 [17] 4/-1/-1->3->2 [18] 4/-1/-1->3->2 [19] 4/-1/-1->3->2 [20] 4/-1/-1->3->2 [21] 4/-1/-1->3->2 [22] 4/-1/-1->3->2 [23] 4/-1/-1->3->2
+ip-26-0-160-225:124167:124624 [7] NCCL INFO Trees [0] -1/-1/-1->7->6 [1] -1/-1/-1->7->6 [2] -1/-1/-1->7->6 [3] -1/-1/-1->7->6 [4] -1/-1/-1->7->6 [5] -1/-1/-1->7->6 [6] -1/-1/-1->7->6 [7] -1/-1/-1->7->6 [8] -1/-1/-1->7->6 [9] -1/-1/-1->7->6 [10] -1/-1/-1->7->6 [11] -1/-1/-1->7->6 [12] -1/-1/-1->7->6 [13] -1/-1/-1->7->6 [14] -1/-1/-1->7->6 [15] -1/-1/-1->7->6 [16] -1/-1/-1->7->6 [17] -1/-1/-1->7->6 [18] -1/-1/-1->7->6 [19] -1/-1/-1->7->6 [20] -1/-1/-1->7->6 [21] -1/-1/-1->7->6 [22] -1/-1/-1->7->6 [23] -1/-1/-1->7->6
+ip-26-0-160-225:124161:124621 [1] NCCL INFO Trees [0] 2/-1/-1->1->0 [1] 2/-1/-1->1->0 [2] 2/-1/-1->1->0 [3] 2/-1/-1->1->0 [4] 2/-1/-1->1->0 [5] 2/-1/-1->1->0 [6] 2/-1/-1->1->0 [7] 2/-1/-1->1->0 [8] 2/-1/-1->1->0 [9] 2/-1/-1->1->0 [10] 2/-1/-1->1->0 [11] 2/-1/-1->1->0 [12] 2/-1/-1->1->0 [13] 2/-1/-1->1->0 [14] 2/-1/-1->1->0 [15] 2/-1/-1->1->0 [16] 2/-1/-1->1->0 [17] 2/-1/-1->1->0 [18] 2/-1/-1->1->0 [19] 2/-1/-1->1->0 [20] 2/-1/-1->1->0 [21] 2/-1/-1->1->0 [22] 2/-1/-1->1->0 [23] 2/-1/-1->1->0
+ip-26-0-160-225:124162:124627 [2] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-160-225:124167:124624 [7] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-161-153:2636084:2636546 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-153:2636091:2636548 [7] NCCL INFO Channel 00/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-225:124161:124621 [1] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-160-225:124163:124626 [3] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-160-225:124160:124620 [0] NCCL INFO Channel 00/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-225:124160:124620 [0] NCCL INFO Channel 01/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-225:124160:124620 [0] NCCL INFO Channel 02/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-225:124160:124620 [0] NCCL INFO Channel 03/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-225:124160:124620 [0] NCCL INFO Channel 04/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-225:124160:124620 [0] NCCL INFO Channel 05/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-225:124160:124620 [0] NCCL INFO Channel 06/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-225:124160:124620 [0] NCCL INFO Channel 07/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-225:124160:124620 [0] NCCL INFO Channel 08/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-225:124160:124620 [0] NCCL INFO Channel 09/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-225:124160:124620 [0] NCCL INFO Channel 10/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-225:124160:124620 [0] NCCL INFO Channel 11/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-225:124160:124620 [0] NCCL INFO Channel 12/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-225:124160:124620 [0] NCCL INFO Channel 13/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-225:124160:124620 [0] NCCL INFO Channel 14/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-225:124160:124620 [0] NCCL INFO Channel 15/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-225:124160:124620 [0] NCCL INFO Channel 16/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-225:124160:124620 [0] NCCL INFO Channel 17/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-225:124160:124620 [0] NCCL INFO Channel 18/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-225:124160:124620 [0] NCCL INFO Channel 19/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-225:124160:124620 [0] NCCL INFO Channel 20/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-225:124160:124620 [0] NCCL INFO Channel 21/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-225:124160:124620 [0] NCCL INFO Channel 22/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-225:124160:124620 [0] NCCL INFO Channel 23/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-153:2636088:2636549 [4] NCCL INFO Channel 07/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-225:124160:124620 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] 1/-1/-1->0->-1 [7] 1/-1/-1->0->-1 [8] 1/-1/-1->0->-1 [9] 1/-1/-1->0->-1 [10] 1/-1/-1->0->-1 [11] 1/-1/-1->0->-1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] 1/-1/-1->0->-1 [19] 1/-1/-1->0->-1 [20] 1/-1/-1->0->-1 [21] 1/-1/-1->0->-1 [22] 1/-1/-1->0->-1 [23] 1/-1/-1->0->-1
+ip-26-0-160-225:124160:124620 [0] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-161-153:2636087:2636550 [3] NCCL INFO Channel 05/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-153:2636085:2636553 [1] NCCL INFO Channel 08/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-153:2636086:2636552 [2] NCCL INFO Channel 10/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-153:2636089:2636551 [5] NCCL INFO Channel 10/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-153:2636090:2636547 [6] NCCL INFO Channel 02/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-153:2636084:2636546 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-153:2636091:2636548 [7] NCCL INFO Channel 01/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-153:2636088:2636549 [4] NCCL INFO Channel 08/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-153:2636087:2636550 [3] NCCL INFO Channel 06/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-153:2636085:2636553 [1] NCCL INFO Channel 09/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-153:2636086:2636552 [2] NCCL INFO Channel 11/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-153:2636089:2636551 [5] NCCL INFO Channel 11/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-153:2636090:2636547 [6] NCCL INFO Channel 03/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-153:2636084:2636546 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-153:2636091:2636548 [7] NCCL INFO Channel 02/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-153:2636088:2636549 [4] NCCL INFO Channel 09/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-153:2636087:2636550 [3] NCCL INFO Channel 07/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-59:1557271:1557726 [5] NCCL INFO Channel 23/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-153:2636085:2636553 [1] NCCL INFO Channel 10/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-153:2636086:2636552 [2] NCCL INFO Channel 12/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-153:2636089:2636551 [5] NCCL INFO Channel 12/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-153:2636090:2636547 [6] NCCL INFO Channel 04/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-153:2636084:2636546 [0] NCCL INFO Channel 04/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-153:2636088:2636549 [4] NCCL INFO Channel 10/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-153:2636091:2636548 [7] NCCL INFO Channel 03/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-153:2636087:2636550 [3] NCCL INFO Channel 08/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-153:2636085:2636553 [1] NCCL INFO Channel 11/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-153:2636086:2636552 [2] NCCL INFO Channel 13/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-153:2636089:2636551 [5] NCCL INFO Channel 13/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-153:2636090:2636547 [6] NCCL INFO Channel 05/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-153:2636084:2636546 [0] NCCL INFO Channel 05/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-153:2636091:2636548 [7] NCCL INFO Channel 04/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-153:2636088:2636549 [4] NCCL INFO Channel 11/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-153:2636087:2636550 [3] NCCL INFO Channel 09/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-131:1630206:1630692 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-165-131:1630206:1630692 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-161-138:96567:97050 [4] NCCL INFO Channel 15/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-153:2636085:2636553 [1] NCCL INFO Channel 12/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-153:2636086:2636552 [2] NCCL INFO Channel 14/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-153:2636089:2636551 [5] NCCL INFO Channel 14/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-153:2636090:2636547 [6] NCCL INFO Channel 06/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-153:2636084:2636546 [0] NCCL INFO Channel 06/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-153:2636088:2636549 [4] NCCL INFO Channel 12/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-153:2636091:2636548 [7] NCCL INFO Channel 05/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-153:2636087:2636550 [3] NCCL INFO Channel 10/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-153:2636085:2636553 [1] NCCL INFO Channel 13/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-153:2636086:2636552 [2] NCCL INFO Channel 15/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-153:2636089:2636551 [5] NCCL INFO Channel 15/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-153:2636084:2636546 [0] NCCL INFO Channel 07/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-153:2636090:2636547 [6] NCCL INFO Channel 07/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-153:2636088:2636549 [4] NCCL INFO Channel 13/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-153:2636091:2636548 [7] NCCL INFO Channel 06/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-153:2636087:2636550 [3] NCCL INFO Channel 11/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-242:1364904:1365369 [2] NCCL INFO Channel 23/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-153:2636085:2636553 [1] NCCL INFO Channel 14/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-153:2636086:2636552 [2] NCCL INFO Channel 16/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-153:2636089:2636551 [5] NCCL INFO Channel 16/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-153:2636084:2636546 [0] NCCL INFO Channel 08/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-153:2636090:2636547 [6] NCCL INFO Channel 08/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-153:2636088:2636549 [4] NCCL INFO Channel 14/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-153:2636091:2636548 [7] NCCL INFO Channel 07/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-153:2636087:2636550 [3] NCCL INFO Channel 12/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-153:2636086:2636552 [2] NCCL INFO Channel 17/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-153:2636085:2636553 [1] NCCL INFO Channel 15/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-153:2636084:2636546 [0] NCCL INFO Channel 09/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-153:2636089:2636551 [5] NCCL INFO Channel 17/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-153:2636090:2636547 [6] NCCL INFO Channel 09/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-153:2636088:2636549 [4] NCCL INFO Channel 15/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-153:2636091:2636548 [7] NCCL INFO Channel 08/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-153:2636087:2636550 [3] NCCL INFO Channel 13/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-168-52:2722048:2722505 [3] NCCL INFO Channel 23/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-153:2636086:2636552 [2] NCCL INFO Channel 18/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-153:2636085:2636553 [1] NCCL INFO Channel 16/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-153:2636084:2636546 [0] NCCL INFO Channel 10/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-153:2636089:2636551 [5] NCCL INFO Channel 18/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-153:2636090:2636547 [6] NCCL INFO Channel 10/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-153:2636088:2636549 [4] NCCL INFO Channel 16/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-153:2636091:2636548 [7] NCCL INFO Channel 09/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-153:2636087:2636550 [3] NCCL INFO Channel 14/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-153:2636086:2636552 [2] NCCL INFO Channel 19/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-153:2636085:2636553 [1] NCCL INFO Channel 17/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-153:2636084:2636546 [0] NCCL INFO Channel 11/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-153:2636089:2636551 [5] NCCL INFO Channel 19/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-153:2636090:2636547 [6] NCCL INFO Channel 11/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-153:2636088:2636549 [4] NCCL INFO Channel 17/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-153:2636091:2636548 [7] NCCL INFO Channel 10/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-153:2636087:2636550 [3] NCCL INFO Channel 15/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-153:2636085:2636553 [1] NCCL INFO Channel 18/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-153:2636084:2636546 [0] NCCL INFO Channel 12/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-153:2636086:2636552 [2] NCCL INFO Channel 20/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-153:2636090:2636547 [6] NCCL INFO Channel 12/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-153:2636089:2636551 [5] NCCL INFO Channel 20/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-153:2636088:2636549 [4] NCCL INFO Channel 18/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-153:2636091:2636548 [7] NCCL INFO Channel 11/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-153:2636087:2636550 [3] NCCL INFO Channel 16/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-168-30:235652:236105 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-168-30:235652:236105 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-161-153:2636085:2636553 [1] NCCL INFO Channel 19/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-153:2636084:2636546 [0] NCCL INFO Channel 13/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-153:2636086:2636552 [2] NCCL INFO Channel 21/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-153:2636089:2636551 [5] NCCL INFO Channel 21/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-153:2636090:2636547 [6] NCCL INFO Channel 13/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-153:2636088:2636549 [4] NCCL INFO Channel 19/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-153:2636091:2636548 [7] NCCL INFO Channel 12/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-153:2636087:2636550 [3] NCCL INFO Channel 17/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-153:2636084:2636546 [0] NCCL INFO Channel 14/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-153:2636089:2636551 [5] NCCL INFO Channel 22/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-153:2636086:2636552 [2] NCCL INFO Channel 22/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-153:2636090:2636547 [6] NCCL INFO Channel 14/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-153:2636088:2636549 [4] NCCL INFO Channel 20/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-153:2636085:2636553 [1] NCCL INFO Channel 20/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-153:2636091:2636548 [7] NCCL INFO Channel 13/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-153:2636087:2636550 [3] NCCL INFO Channel 18/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-153:2636084:2636546 [0] NCCL INFO Channel 15/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-153:2636089:2636551 [5] NCCL INFO Channel 23/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-138:96567:97050 [4] NCCL INFO Channel 16/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-153:2636086:2636552 [2] NCCL INFO Channel 23/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-153:2636090:2636547 [6] NCCL INFO Channel 15/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-153:2636088:2636549 [4] NCCL INFO Channel 21/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-153:2636085:2636553 [1] NCCL INFO Channel 21/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-153:2636091:2636548 [7] NCCL INFO Channel 14/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-153:2636087:2636550 [3] NCCL INFO Channel 19/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-153:2636084:2636546 [0] NCCL INFO Channel 16/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-153:2636090:2636547 [6] NCCL INFO Channel 16/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-153:2636088:2636549 [4] NCCL INFO Channel 22/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-153:2636091:2636548 [7] NCCL INFO Channel 15/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-153:2636087:2636550 [3] NCCL INFO Channel 20/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-168-30:235650:236106 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-168-30:235650:236106 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-161-153:2636085:2636553 [1] NCCL INFO Channel 22/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-153:2636084:2636546 [0] NCCL INFO Channel 17/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-153:2636090:2636547 [6] NCCL INFO Channel 17/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-153:2636088:2636549 [4] NCCL INFO Channel 23/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-153:2636091:2636548 [7] NCCL INFO Channel 16/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-153:2636087:2636550 [3] NCCL INFO Channel 21/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-153:2636085:2636553 [1] NCCL INFO Channel 23/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-153:2636084:2636546 [0] NCCL INFO Channel 18/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-153:2636090:2636547 [6] NCCL INFO Channel 18/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-78:114512:114968 [0] NCCL INFO Channel 23/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-78:114517:114969 [5] NCCL INFO Channel 22/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-153:2636091:2636548 [7] NCCL INFO Channel 17/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-153:2636087:2636550 [3] NCCL INFO Channel 22/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-153:2636084:2636546 [0] NCCL INFO Channel 19/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-153:2636090:2636547 [6] NCCL INFO Channel 19/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-131:1630205:1630687 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-165-131:1630205:1630687 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-161-153:2636091:2636548 [7] NCCL INFO Channel 18/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-153:2636087:2636550 [3] NCCL INFO Channel 23/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-153:2636084:2636546 [0] NCCL INFO Channel 20/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-78:114517:114969 [5] NCCL INFO Channel 23/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-153:2636090:2636547 [6] NCCL INFO Channel 20/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-153:2636091:2636548 [7] NCCL INFO Channel 19/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-225:124165:124625 [5] NCCL INFO Channel 00/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-225:124166:124622 [6] NCCL INFO Channel 00/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-153:2636084:2636546 [0] NCCL INFO Channel 21/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-153:2636090:2636547 [6] NCCL INFO Channel 21/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-225:124167:124624 [7] NCCL INFO Channel 00/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-225:124165:124625 [5] NCCL INFO Channel 01/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-225:124162:124627 [2] NCCL INFO Channel 00/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-225:124166:124622 [6] NCCL INFO Channel 01/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-225:124167:124624 [7] NCCL INFO Channel 01/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-225:124165:124625 [5] NCCL INFO Channel 02/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-225:124162:124627 [2] NCCL INFO Channel 01/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-225:124166:124622 [6] NCCL INFO Channel 02/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-225:124167:124624 [7] NCCL INFO Channel 02/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-225:124165:124625 [5] NCCL INFO Channel 03/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-225:124162:124627 [2] NCCL INFO Channel 02/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-153:2636091:2636548 [7] NCCL INFO Channel 20/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-225:124166:124622 [6] NCCL INFO Channel 03/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-168-30:235653:236103 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-168-30:235653:236103 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-160-225:124167:124624 [7] NCCL INFO Channel 03/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-225:124165:124625 [5] NCCL INFO Channel 04/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-225:124164:124623 [4] NCCL INFO Channel 00/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-225:124166:124622 [6] NCCL INFO Channel 04/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-225:124167:124624 [7] NCCL INFO Channel 04/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-225:124165:124625 [5] NCCL INFO Channel 05/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-225:124166:124622 [6] NCCL INFO Channel 05/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-225:124164:124623 [4] NCCL INFO Channel 01/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-168-30:235651:236108 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-160-225:124167:124624 [7] NCCL INFO Channel 05/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-168-30:235651:236108 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-160-225:124165:124625 [5] NCCL INFO Channel 06/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-168-30:235648:236102 [0] NCCL INFO Channel 00/24 :    0   1   2   3   4   5   6   7
+ip-26-0-168-30:235648:236102 [0] NCCL INFO Channel 01/24 :    0   1   2   3   4   5   6   7
+ip-26-0-168-30:235655:236104 [7] NCCL INFO Trees [0] -1/-1/-1->7->6 [1] -1/-1/-1->7->6 [2] -1/-1/-1->7->6 [3] -1/-1/-1->7->6 [4] -1/-1/-1->7->6 [5] -1/-1/-1->7->6 [6] -1/-1/-1->7->6 [7] -1/-1/-1->7->6 [8] -1/-1/-1->7->6 [9] -1/-1/-1->7->6 [10] -1/-1/-1->7->6 [11] -1/-1/-1->7->6 [12] -1/-1/-1->7->6 [13] -1/-1/-1->7->6 [14] -1/-1/-1->7->6 [15] -1/-1/-1->7->6 [16] -1/-1/-1->7->6 [17] -1/-1/-1->7->6 [18] -1/-1/-1->7->6 [19] -1/-1/-1->7->6 [20] -1/-1/-1->7->6 [21] -1/-1/-1->7->6 [22] -1/-1/-1->7->6 [23] -1/-1/-1->7->6
+ip-26-0-168-30:235648:236102 [0] NCCL INFO Channel 02/24 :    0   1   2   3   4   5   6   7
+ip-26-0-168-30:235650:236106 [2] NCCL INFO Trees [0] 3/-1/-1->2->1 [1] 3/-1/-1->2->1 [2] 3/-1/-1->2->1 [3] 3/-1/-1->2->1 [4] 3/-1/-1->2->1 [5] 3/-1/-1->2->1 [6] 3/-1/-1->2->1 [7] 3/-1/-1->2->1 [8] 3/-1/-1->2->1 [9] 3/-1/-1->2->1 [10] 3/-1/-1->2->1 [11] 3/-1/-1->2->1 [12] 3/-1/-1->2->1 [13] 3/-1/-1->2->1 [14] 3/-1/-1->2->1 [15] 3/-1/-1->2->1 [16] 3/-1/-1->2->1 [17] 3/-1/-1->2->1 [18] 3/-1/-1->2->1 [19] 3/-1/-1->2->1 [20] 3/-1/-1->2->1 [21] 3/-1/-1->2->1 [22] 3/-1/-1->2->1 [23] 3/-1/-1->2->1
+ip-26-0-168-30:235654:236107 [6] NCCL INFO Trees [0] 7/-1/-1->6->5 [1] 7/-1/-1->6->5 [2] 7/-1/-1->6->5 [3] 7/-1/-1->6->5 [4] 7/-1/-1->6->5 [5] 7/-1/-1->6->5 [6] 7/-1/-1->6->5 [7] 7/-1/-1->6->5 [8] 7/-1/-1->6->5 [9] 7/-1/-1->6->5 [10] 7/-1/-1->6->5 [11] 7/-1/-1->6->5 [12] 7/-1/-1->6->5 [13] 7/-1/-1->6->5 [14] 7/-1/-1->6->5 [15] 7/-1/-1->6->5 [16] 7/-1/-1->6->5 [17] 7/-1/-1->6->5 [18] 7/-1/-1->6->5 [19] 7/-1/-1->6->5 [20] 7/-1/-1->6->5 [21] 7/-1/-1->6->5 [22] 7/-1/-1->6->5 [23] 7/-1/-1->6->5
+ip-26-0-168-30:235655:236104 [7] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-168-30:235653:236103 [5] NCCL INFO Trees [0] 6/-1/-1->5->4 [1] 6/-1/-1->5->4 [2] 6/-1/-1->5->4 [3] 6/-1/-1->5->4 [4] 6/-1/-1->5->4 [5] 6/-1/-1->5->4 [6] 6/-1/-1->5->4 [7] 6/-1/-1->5->4 [8] 6/-1/-1->5->4 [9] 6/-1/-1->5->4 [10] 6/-1/-1->5->4 [11] 6/-1/-1->5->4 [12] 6/-1/-1->5->4 [13] 6/-1/-1->5->4 [14] 6/-1/-1->5->4 [15] 6/-1/-1->5->4 [16] 6/-1/-1->5->4 [17] 6/-1/-1->5->4 [18] 6/-1/-1->5->4 [19] 6/-1/-1->5->4 [20] 6/-1/-1->5->4 [21] 6/-1/-1->5->4 [22] 6/-1/-1->5->4 [23] 6/-1/-1->5->4
+ip-26-0-168-30:235652:236105 [4] NCCL INFO Trees [0] 5/-1/-1->4->3 [1] 5/-1/-1->4->3 [2] 5/-1/-1->4->3 [3] 5/-1/-1->4->3 [4] 5/-1/-1->4->3 [5] 5/-1/-1->4->3 [6] 5/-1/-1->4->3 [7] 5/-1/-1->4->3 [8] 5/-1/-1->4->3 [9] 5/-1/-1->4->3 [10] 5/-1/-1->4->3 [11] 5/-1/-1->4->3 [12] 5/-1/-1->4->3 [13] 5/-1/-1->4->3 [14] 5/-1/-1->4->3 [15] 5/-1/-1->4->3 [16] 5/-1/-1->4->3 [17] 5/-1/-1->4->3 [18] 5/-1/-1->4->3 [19] 5/-1/-1->4->3 [20] 5/-1/-1->4->3 [21] 5/-1/-1->4->3 [22] 5/-1/-1->4->3 [23] 5/-1/-1->4->3
+ip-26-0-168-30:235648:236102 [0] NCCL INFO Channel 03/24 :    0   1   2   3   4   5   6   7
+ip-26-0-168-30:235650:236106 [2] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-168-30:235653:236103 [5] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-168-30:235654:236107 [6] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-161-153:2636084:2636546 [0] NCCL INFO Channel 22/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-168-30:235648:236102 [0] NCCL INFO Channel 04/24 :    0   1   2   3   4   5   6   7
+ip-26-0-168-30:235652:236105 [4] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-168-30:235651:236108 [3] NCCL INFO Trees [0] 4/-1/-1->3->2 [1] 4/-1/-1->3->2 [2] 4/-1/-1->3->2 [3] 4/-1/-1->3->2 [4] 4/-1/-1->3->2 [5] 4/-1/-1->3->2 [6] 4/-1/-1->3->2 [7] 4/-1/-1->3->2 [8] 4/-1/-1->3->2 [9] 4/-1/-1->3->2 [10] 4/-1/-1->3->2 [11] 4/-1/-1->3->2 [12] 4/-1/-1->3->2 [13] 4/-1/-1->3->2 [14] 4/-1/-1->3->2 [15] 4/-1/-1->3->2 [16] 4/-1/-1->3->2 [17] 4/-1/-1->3->2 [18] 4/-1/-1->3->2 [19] 4/-1/-1->3->2 [20] 4/-1/-1->3->2 [21] 4/-1/-1->3->2 [22] 4/-1/-1->3->2 [23] 4/-1/-1->3->2
+ip-26-0-168-30:235648:236102 [0] NCCL INFO Channel 05/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-225:124166:124622 [6] NCCL INFO Channel 06/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-168-30:235649:236109 [1] NCCL INFO Trees [0] 2/-1/-1->1->0 [1] 2/-1/-1->1->0 [2] 2/-1/-1->1->0 [3] 2/-1/-1->1->0 [4] 2/-1/-1->1->0 [5] 2/-1/-1->1->0 [6] 2/-1/-1->1->0 [7] 2/-1/-1->1->0 [8] 2/-1/-1->1->0 [9] 2/-1/-1->1->0 [10] 2/-1/-1->1->0 [11] 2/-1/-1->1->0 [12] 2/-1/-1->1->0 [13] 2/-1/-1->1->0 [14] 2/-1/-1->1->0 [15] 2/-1/-1->1->0 [16] 2/-1/-1->1->0 [17] 2/-1/-1->1->0 [18] 2/-1/-1->1->0 [19] 2/-1/-1->1->0 [20] 2/-1/-1->1->0 [21] 2/-1/-1->1->0 [22] 2/-1/-1->1->0 [23] 2/-1/-1->1->0
+ip-26-0-168-30:235648:236102 [0] NCCL INFO Channel 06/24 :    0   1   2   3   4   5   6   7
+ip-26-0-168-30:235651:236108 [3] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-168-30:235648:236102 [0] NCCL INFO Channel 07/24 :    0   1   2   3   4   5   6   7
+ip-26-0-168-30:235648:236102 [0] NCCL INFO Channel 08/24 :    0   1   2   3   4   5   6   7
+ip-26-0-168-30:235648:236102 [0] NCCL INFO Channel 09/24 :    0   1   2   3   4   5   6   7
+ip-26-0-168-30:235648:236102 [0] NCCL INFO Channel 10/24 :    0   1   2   3   4   5   6   7
+ip-26-0-168-30:235648:236102 [0] NCCL INFO Channel 11/24 :    0   1   2   3   4   5   6   7
+ip-26-0-168-30:235648:236102 [0] NCCL INFO Channel 12/24 :    0   1   2   3   4   5   6   7
+ip-26-0-168-30:235648:236102 [0] NCCL INFO Channel 13/24 :    0   1   2   3   4   5   6   7
+ip-26-0-168-30:235648:236102 [0] NCCL INFO Channel 14/24 :    0   1   2   3   4   5   6   7
+ip-26-0-168-30:235648:236102 [0] NCCL INFO Channel 15/24 :    0   1   2   3   4   5   6   7
+ip-26-0-168-30:235648:236102 [0] NCCL INFO Channel 16/24 :    0   1   2   3   4   5   6   7
+ip-26-0-168-30:235648:236102 [0] NCCL INFO Channel 17/24 :    0   1   2   3   4   5   6   7
+ip-26-0-168-30:235648:236102 [0] NCCL INFO Channel 18/24 :    0   1   2   3   4   5   6   7
+ip-26-0-168-30:235648:236102 [0] NCCL INFO Channel 19/24 :    0   1   2   3   4   5   6   7
+ip-26-0-168-30:235648:236102 [0] NCCL INFO Channel 20/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-225:124164:124623 [4] NCCL INFO Channel 02/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-168-30:235648:236102 [0] NCCL INFO Channel 21/24 :    0   1   2   3   4   5   6   7
+ip-26-0-168-30:235648:236102 [0] NCCL INFO Channel 22/24 :    0   1   2   3   4   5   6   7
+ip-26-0-168-30:235648:236102 [0] NCCL INFO Channel 23/24 :    0   1   2   3   4   5   6   7
+ip-26-0-168-30:235648:236102 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] 1/-1/-1->0->-1 [7] 1/-1/-1->0->-1 [8] 1/-1/-1->0->-1 [9] 1/-1/-1->0->-1 [10] 1/-1/-1->0->-1 [11] 1/-1/-1->0->-1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] 1/-1/-1->0->-1 [19] 1/-1/-1->0->-1 [20] 1/-1/-1->0->-1 [21] 1/-1/-1->0->-1 [22] 1/-1/-1->0->-1 [23] 1/-1/-1->0->-1
+ip-26-0-168-30:235649:236109 [1] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-168-30:235648:236102 [0] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-161-153:2636090:2636547 [6] NCCL INFO Channel 22/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-138:96567:97050 [4] NCCL INFO Channel 17/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-225:124167:124624 [7] NCCL INFO Channel 06/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-225:124165:124625 [5] NCCL INFO Channel 07/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-225:124166:124622 [6] NCCL INFO Channel 07/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-225:124164:124623 [4] NCCL INFO Channel 03/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-153:2636091:2636548 [7] NCCL INFO Channel 21/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-225:124167:124624 [7] NCCL INFO Channel 07/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-225:124165:124625 [5] NCCL INFO Channel 08/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-225:124166:124622 [6] NCCL INFO Channel 08/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-225:124164:124623 [4] NCCL INFO Channel 04/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-225:124167:124624 [7] NCCL INFO Channel 08/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-225:124165:124625 [5] NCCL INFO Channel 09/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-225:124164:124623 [4] NCCL INFO Channel 05/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-225:124166:124622 [6] NCCL INFO Channel 09/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-225:124161:124621 [1] NCCL INFO Channel 00/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-164:1227465:1227924 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-160-225:124160:124620 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-164:1227465:1227924 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-161-103:96888:97344 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-161-103:96888:97344 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-160-225:124167:124624 [7] NCCL INFO Channel 09/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-225:124163:124626 [3] NCCL INFO Channel 00/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-225:124165:124625 [5] NCCL INFO Channel 10/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-165-164:1227467:1227925 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-165-164:1227467:1227925 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-160-225:124162:124627 [2] NCCL INFO Channel 03/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-103:96894:97349 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-160-225:124164:124623 [4] NCCL INFO Channel 06/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-225:124166:124622 [6] NCCL INFO Channel 10/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-225:124161:124621 [1] NCCL INFO Channel 01/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-103:96894:97349 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-161-153:2636084:2636546 [0] NCCL INFO Channel 23/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-225:124160:124620 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-225:124165:124625 [5] NCCL INFO Channel 11/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-225:124167:124624 [7] NCCL INFO Channel 10/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-225:124162:124627 [2] NCCL INFO Channel 04/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-225:124160:124620 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-103:96890:97350 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-160-225:124165:124625 [5] NCCL INFO Channel 12/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-103:96890:97350 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-160-225:124167:124624 [7] NCCL INFO Channel 11/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-103:96890:97350 [2] NCCL INFO Trees [0] 3/-1/-1->2->1 [1] 3/-1/-1->2->1 [2] 3/-1/-1->2->1 [3] 3/-1/-1->2->1 [4] 3/-1/-1->2->1 [5] 3/-1/-1->2->1 [6] 3/-1/-1->2->1 [7] 3/-1/-1->2->1 [8] 3/-1/-1->2->1 [9] 3/-1/-1->2->1 [10] 3/-1/-1->2->1 [11] 3/-1/-1->2->1 [12] 3/-1/-1->2->1 [13] 3/-1/-1->2->1 [14] 3/-1/-1->2->1 [15] 3/-1/-1->2->1 [16] 3/-1/-1->2->1 [17] 3/-1/-1->2->1 [18] 3/-1/-1->2->1 [19] 3/-1/-1->2->1 [20] 3/-1/-1->2->1 [21] 3/-1/-1->2->1 [22] 3/-1/-1->2->1 [23] 3/-1/-1->2->1
+ip-26-0-161-103:96890:97350 [2] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-161-103:96888:97344 [0] NCCL INFO Channel 00/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-103:96888:97344 [0] NCCL INFO Channel 01/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-103:96888:97344 [0] NCCL INFO Channel 02/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-103:96888:97344 [0] NCCL INFO Channel 03/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-103:96888:97344 [0] NCCL INFO Channel 04/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-103:96888:97344 [0] NCCL INFO Channel 05/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-103:96888:97344 [0] NCCL INFO Channel 06/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-103:96888:97344 [0] NCCL INFO Channel 07/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-103:96888:97344 [0] NCCL INFO Channel 08/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-103:96889:97348 [1] NCCL INFO Trees [0] 2/-1/-1->1->0 [1] 2/-1/-1->1->0 [2] 2/-1/-1->1->0 [3] 2/-1/-1->1->0 [4] 2/-1/-1->1->0 [5] 2/-1/-1->1->0 [6] 2/-1/-1->1->0 [7] 2/-1/-1->1->0 [8] 2/-1/-1->1->0 [9] 2/-1/-1->1->0 [10] 2/-1/-1->1->0 [11] 2/-1/-1->1->0 [12] 2/-1/-1->1->0 [13] 2/-1/-1->1->0 [14] 2/-1/-1->1->0 [15] 2/-1/-1->1->0 [16] 2/-1/-1->1->0 [17] 2/-1/-1->1->0 [18] 2/-1/-1->1->0 [19] 2/-1/-1->1->0 [20] 2/-1/-1->1->0 [21] 2/-1/-1->1->0 [22] 2/-1/-1->1->0 [23] 2/-1/-1->1->0
+ip-26-0-161-103:96888:97344 [0] NCCL INFO Channel 09/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-103:96892:97345 [4] NCCL INFO Trees [0] 5/-1/-1->4->3 [1] 5/-1/-1->4->3 [2] 5/-1/-1->4->3 [3] 5/-1/-1->4->3 [4] 5/-1/-1->4->3 [5] 5/-1/-1->4->3 [6] 5/-1/-1->4->3 [7] 5/-1/-1->4->3 [8] 5/-1/-1->4->3 [9] 5/-1/-1->4->3 [10] 5/-1/-1->4->3 [11] 5/-1/-1->4->3 [12] 5/-1/-1->4->3 [13] 5/-1/-1->4->3 [14] 5/-1/-1->4->3 [15] 5/-1/-1->4->3 [16] 5/-1/-1->4->3 [17] 5/-1/-1->4->3 [18] 5/-1/-1->4->3 [19] 5/-1/-1->4->3 [20] 5/-1/-1->4->3 [21] 5/-1/-1->4->3 [22] 5/-1/-1->4->3 [23] 5/-1/-1->4->3
+ip-26-0-161-103:96888:97344 [0] NCCL INFO Channel 10/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-103:96891:97347 [3] NCCL INFO Trees [0] 4/-1/-1->3->2 [1] 4/-1/-1->3->2 [2] 4/-1/-1->3->2 [3] 4/-1/-1->3->2 [4] 4/-1/-1->3->2 [5] 4/-1/-1->3->2 [6] 4/-1/-1->3->2 [7] 4/-1/-1->3->2 [8] 4/-1/-1->3->2 [9] 4/-1/-1->3->2 [10] 4/-1/-1->3->2 [11] 4/-1/-1->3->2 [12] 4/-1/-1->3->2 [13] 4/-1/-1->3->2 [14] 4/-1/-1->3->2 [15] 4/-1/-1->3->2 [16] 4/-1/-1->3->2 [17] 4/-1/-1->3->2 [18] 4/-1/-1->3->2 [19] 4/-1/-1->3->2 [20] 4/-1/-1->3->2 [21] 4/-1/-1->3->2 [22] 4/-1/-1->3->2 [23] 4/-1/-1->3->2
+ip-26-0-161-103:96888:97344 [0] NCCL INFO Channel 11/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-103:96893:97346 [5] NCCL INFO Trees [0] 6/-1/-1->5->4 [1] 6/-1/-1->5->4 [2] 6/-1/-1->5->4 [3] 6/-1/-1->5->4 [4] 6/-1/-1->5->4 [5] 6/-1/-1->5->4 [6] 6/-1/-1->5->4 [7] 6/-1/-1->5->4 [8] 6/-1/-1->5->4 [9] 6/-1/-1->5->4 [10] 6/-1/-1->5->4 [11] 6/-1/-1->5->4 [12] 6/-1/-1->5->4 [13] 6/-1/-1->5->4 [14] 6/-1/-1->5->4 [15] 6/-1/-1->5->4 [16] 6/-1/-1->5->4 [17] 6/-1/-1->5->4 [18] 6/-1/-1->5->4 [19] 6/-1/-1->5->4 [20] 6/-1/-1->5->4 [21] 6/-1/-1->5->4 [22] 6/-1/-1->5->4 [23] 6/-1/-1->5->4
+ip-26-0-161-103:96892:97345 [4] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-161-103:96888:97344 [0] NCCL INFO Channel 12/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-103:96888:97344 [0] NCCL INFO Channel 13/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-103:96888:97344 [0] NCCL INFO Channel 14/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-103:96893:97346 [5] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-161-103:96888:97344 [0] NCCL INFO Channel 15/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-103:96895:97351 [7] NCCL INFO Trees [0] -1/-1/-1->7->6 [1] -1/-1/-1->7->6 [2] -1/-1/-1->7->6 [3] -1/-1/-1->7->6 [4] -1/-1/-1->7->6 [5] -1/-1/-1->7->6 [6] -1/-1/-1->7->6 [7] -1/-1/-1->7->6 [8] -1/-1/-1->7->6 [9] -1/-1/-1->7->6 [10] -1/-1/-1->7->6 [11] -1/-1/-1->7->6 [12] -1/-1/-1->7->6 [13] -1/-1/-1->7->6 [14] -1/-1/-1->7->6 [15] -1/-1/-1->7->6 [16] -1/-1/-1->7->6 [17] -1/-1/-1->7->6 [18] -1/-1/-1->7->6 [19] -1/-1/-1->7->6 [20] -1/-1/-1->7->6 [21] -1/-1/-1->7->6 [22] -1/-1/-1->7->6 [23] -1/-1/-1->7->6
+ip-26-0-161-103:96888:97344 [0] NCCL INFO Channel 16/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-103:96891:97347 [3] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-161-103:96888:97344 [0] NCCL INFO Channel 17/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-103:96889:97348 [1] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-161-103:96888:97344 [0] NCCL INFO Channel 18/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-103:96888:97344 [0] NCCL INFO Channel 19/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-103:96888:97344 [0] NCCL INFO Channel 20/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-103:96888:97344 [0] NCCL INFO Channel 21/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-103:96888:97344 [0] NCCL INFO Channel 22/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-103:96888:97344 [0] NCCL INFO Channel 23/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-103:96888:97344 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] 1/-1/-1->0->-1 [7] 1/-1/-1->0->-1 [8] 1/-1/-1->0->-1 [9] 1/-1/-1->0->-1 [10] 1/-1/-1->0->-1 [11] 1/-1/-1->0->-1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] 1/-1/-1->0->-1 [19] 1/-1/-1->0->-1 [20] 1/-1/-1->0->-1 [21] 1/-1/-1->0->-1 [22] 1/-1/-1->0->-1 [23] 1/-1/-1->0->-1
+ip-26-0-161-103:96888:97344 [0] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-161-103:96895:97351 [7] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-161-153:2636090:2636547 [6] NCCL INFO Channel 23/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-103:96894:97349 [6] NCCL INFO Trees [0] 7/-1/-1->6->5 [1] 7/-1/-1->6->5 [2] 7/-1/-1->6->5 [3] 7/-1/-1->6->5 [4] 7/-1/-1->6->5 [5] 7/-1/-1->6->5 [6] 7/-1/-1->6->5 [7] 7/-1/-1->6->5 [8] 7/-1/-1->6->5 [9] 7/-1/-1->6->5 [10] 7/-1/-1->6->5 [11] 7/-1/-1->6->5 [12] 7/-1/-1->6->5 [13] 7/-1/-1->6->5 [14] 7/-1/-1->6->5 [15] 7/-1/-1->6->5 [16] 7/-1/-1->6->5 [17] 7/-1/-1->6->5 [18] 7/-1/-1->6->5 [19] 7/-1/-1->6->5 [20] 7/-1/-1->6->5 [21] 7/-1/-1->6->5 [22] 7/-1/-1->6->5 [23] 7/-1/-1->6->5
+ip-26-0-161-103:96894:97349 [6] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-160-225:124162:124627 [2] NCCL INFO Channel 05/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-153:2636091:2636548 [7] NCCL INFO Channel 22/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-225:124164:124623 [4] NCCL INFO Channel 07/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-225:124166:124622 [6] NCCL INFO Channel 11/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-225:124162:124627 [2] NCCL INFO Channel 06/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-131:1630207:1630694 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-165-131:1630207:1630694 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-160-225:124164:124623 [4] NCCL INFO Channel 08/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-225:124161:124621 [1] NCCL INFO Channel 02/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-225:124166:124622 [6] NCCL INFO Channel 12/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-225:124162:124627 [2] NCCL INFO Channel 07/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-225:124163:124626 [3] NCCL INFO Channel 01/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-225:124164:124623 [4] NCCL INFO Channel 09/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-225:124161:124621 [1] NCCL INFO Channel 03/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-225:124163:124626 [3] NCCL INFO Channel 02/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-225:124160:124620 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-225:124165:124625 [5] NCCL INFO Channel 13/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-225:124167:124624 [7] NCCL INFO Channel 12/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-225:124163:124626 [3] NCCL INFO Channel 03/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-225:124160:124620 [0] NCCL INFO Channel 04/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-225:124165:124625 [5] NCCL INFO Channel 14/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-165-131:1630209:1630693 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-165-131:1630209:1630693 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-160-225:124166:124622 [6] NCCL INFO Channel 13/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-153:2636091:2636548 [7] NCCL INFO Channel 23/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-225:124161:124621 [1] NCCL INFO Channel 04/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-225:124163:124626 [3] NCCL INFO Channel 04/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-225:124164:124623 [4] NCCL INFO Channel 10/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-225:124167:124624 [7] NCCL INFO Channel 13/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-225:124165:124625 [5] NCCL INFO Channel 15/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-138:96567:97050 [4] NCCL INFO Channel 18/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-225:124161:124621 [1] NCCL INFO Channel 05/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-225:124163:124626 [3] NCCL INFO Channel 05/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-225:124164:124623 [4] NCCL INFO Channel 11/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-225:124167:124624 [7] NCCL INFO Channel 14/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-225:124165:124625 [5] NCCL INFO Channel 16/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-225:124161:124621 [1] NCCL INFO Channel 06/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-225:124163:124626 [3] NCCL INFO Channel 06/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-225:124164:124623 [4] NCCL INFO Channel 12/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-225:124167:124624 [7] NCCL INFO Channel 15/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-225:124165:124625 [5] NCCL INFO Channel 17/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-225:124162:124627 [2] NCCL INFO Channel 08/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-225:124161:124621 [1] NCCL INFO Channel 07/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-225:124163:124626 [3] NCCL INFO Channel 07/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-225:124164:124623 [4] NCCL INFO Channel 13/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-225:124160:124620 [0] NCCL INFO Channel 05/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-225:124167:124624 [7] NCCL INFO Channel 16/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-225:124165:124625 [5] NCCL INFO Channel 18/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-225:124166:124622 [6] NCCL INFO Channel 14/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-225:124162:124627 [2] NCCL INFO Channel 09/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-225:124161:124621 [1] NCCL INFO Channel 08/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-225:124164:124623 [4] NCCL INFO Channel 14/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-225:124160:124620 [0] NCCL INFO Channel 06/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-225:124163:124626 [3] NCCL INFO Channel 08/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-225:124167:124624 [7] NCCL INFO Channel 17/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-225:124165:124625 [5] NCCL INFO Channel 19/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-225:124166:124622 [6] NCCL INFO Channel 15/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-225:124162:124627 [2] NCCL INFO Channel 10/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-225:124161:124621 [1] NCCL INFO Channel 09/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-225:124164:124623 [4] NCCL INFO Channel 15/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-225:124160:124620 [0] NCCL INFO Channel 07/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-225:124167:124624 [7] NCCL INFO Channel 18/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-225:124163:124626 [3] NCCL INFO Channel 09/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-225:124165:124625 [5] NCCL INFO Channel 20/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-225:124166:124622 [6] NCCL INFO Channel 16/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-225:124162:124627 [2] NCCL INFO Channel 11/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-225:124161:124621 [1] NCCL INFO Channel 10/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-225:124164:124623 [4] NCCL INFO Channel 16/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-225:124160:124620 [0] NCCL INFO Channel 08/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-225:124163:124626 [3] NCCL INFO Channel 10/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-225:124167:124624 [7] NCCL INFO Channel 19/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-225:124165:124625 [5] NCCL INFO Channel 21/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-225:124162:124627 [2] NCCL INFO Channel 12/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-225:124166:124622 [6] NCCL INFO Channel 17/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-225:124161:124621 [1] NCCL INFO Channel 11/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-225:124164:124623 [4] NCCL INFO Channel 17/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-225:124160:124620 [0] NCCL INFO Channel 09/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-225:124163:124626 [3] NCCL INFO Channel 11/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-225:124167:124624 [7] NCCL INFO Channel 20/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-225:124165:124625 [5] NCCL INFO Channel 22/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-225:124162:124627 [2] NCCL INFO Channel 13/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-225:124161:124621 [1] NCCL INFO Channel 12/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-225:124166:124622 [6] NCCL INFO Channel 18/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-225:124164:124623 [4] NCCL INFO Channel 18/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-225:124160:124620 [0] NCCL INFO Channel 10/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-225:124163:124626 [3] NCCL INFO Channel 12/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-225:124167:124624 [7] NCCL INFO Channel 21/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-225:124165:124625 [5] NCCL INFO Channel 23/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-225:124162:124627 [2] NCCL INFO Channel 14/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-225:124166:124622 [6] NCCL INFO Channel 19/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-225:124161:124621 [1] NCCL INFO Channel 13/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-168-30:235654:236107 [6] NCCL INFO Channel 00/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-168-30:235653:236103 [5] NCCL INFO Channel 00/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-168-30:235651:236108 [3] NCCL INFO Channel 00/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-225:124164:124623 [4] NCCL INFO Channel 19/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-225:124160:124620 [0] NCCL INFO Channel 11/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-225:124163:124626 [3] NCCL INFO Channel 13/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-168-30:235652:236105 [4] NCCL INFO Channel 00/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-225:124167:124624 [7] NCCL INFO Channel 22/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-168-30:235650:236106 [2] NCCL INFO Channel 00/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-168-30:235654:236107 [6] NCCL INFO Channel 01/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-225:124162:124627 [2] NCCL INFO Channel 15/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-168-30:235653:236103 [5] NCCL INFO Channel 01/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-168-30:235651:236108 [3] NCCL INFO Channel 01/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-225:124166:124622 [6] NCCL INFO Channel 20/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-225:124161:124621 [1] NCCL INFO Channel 14/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-168-30:235652:236105 [4] NCCL INFO Channel 01/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-168-30:235650:236106 [2] NCCL INFO Channel 01/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-225:124164:124623 [4] NCCL INFO Channel 20/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-225:124160:124620 [0] NCCL INFO Channel 12/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-225:124163:124626 [3] NCCL INFO Channel 14/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-225:124167:124624 [7] NCCL INFO Channel 23/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-168-30:235654:236107 [6] NCCL INFO Channel 02/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-168-30:235653:236103 [5] NCCL INFO Channel 02/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-168-30:235651:236108 [3] NCCL INFO Channel 02/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-168-30:235652:236105 [4] NCCL INFO Channel 02/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-225:124162:124627 [2] NCCL INFO Channel 16/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-225:124161:124621 [1] NCCL INFO Channel 15/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-225:124166:124622 [6] NCCL INFO Channel 21/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-168-30:235650:236106 [2] NCCL INFO Channel 02/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-225:124160:124620 [0] NCCL INFO Channel 13/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-225:124163:124626 [3] NCCL INFO Channel 15/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-225:124164:124623 [4] NCCL INFO Channel 21/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-168-30:235654:236107 [6] NCCL INFO Channel 03/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-168-30:235653:236103 [5] NCCL INFO Channel 03/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-168-30:235651:236108 [3] NCCL INFO Channel 03/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-225:124162:124627 [2] NCCL INFO Channel 17/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-225:124161:124621 [1] NCCL INFO Channel 16/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-168-30:235652:236105 [4] NCCL INFO Channel 03/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-225:124166:124622 [6] NCCL INFO Channel 22/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-225:124160:124620 [0] NCCL INFO Channel 14/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-168-30:235650:236106 [2] NCCL INFO Channel 03/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-225:124163:124626 [3] NCCL INFO Channel 16/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-225:124164:124623 [4] NCCL INFO Channel 22/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-168-30:235654:236107 [6] NCCL INFO Channel 04/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-168-30:235653:236103 [5] NCCL INFO Channel 04/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-225:124162:124627 [2] NCCL INFO Channel 18/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-225:124161:124621 [1] NCCL INFO Channel 17/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-168-30:235651:236108 [3] NCCL INFO Channel 04/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-225:124166:124622 [6] NCCL INFO Channel 23/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-168-30:235652:236105 [4] NCCL INFO Channel 04/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-225:124160:124620 [0] NCCL INFO Channel 15/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-225:124163:124626 [3] NCCL INFO Channel 17/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-225:124164:124623 [4] NCCL INFO Channel 23/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-168-30:235650:236106 [2] NCCL INFO Channel 04/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-168-30:235654:236107 [6] NCCL INFO Channel 05/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-225:124162:124627 [2] NCCL INFO Channel 19/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-225:124161:124621 [1] NCCL INFO Channel 18/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-138:96567:97050 [4] NCCL INFO Channel 19/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-168-30:235653:236103 [5] NCCL INFO Channel 05/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-168-30:235655:236104 [7] NCCL INFO Channel 00/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-168-30:235648:236102 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-168-30:235651:236108 [3] NCCL INFO Channel 05/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-168-30:235652:236105 [4] NCCL INFO Channel 05/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-225:124163:124626 [3] NCCL INFO Channel 18/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-225:124160:124620 [0] NCCL INFO Channel 16/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-168-30:235650:236106 [2] NCCL INFO Channel 05/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-168-30:235649:236109 [1] NCCL INFO Channel 00/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-168-30:235654:236107 [6] NCCL INFO Channel 06/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-168-30:235653:236103 [5] NCCL INFO Channel 06/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-168-30:235655:236104 [7] NCCL INFO Channel 01/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-168-30:235648:236102 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-168-30:235651:236108 [3] NCCL INFO Channel 06/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-168-30:235652:236105 [4] NCCL INFO Channel 06/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-225:124162:124627 [2] NCCL INFO Channel 20/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-225:124161:124621 [1] NCCL INFO Channel 19/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-168-30:235650:236106 [2] NCCL INFO Channel 06/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-168-30:235649:236109 [1] NCCL INFO Channel 01/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-168-30:235654:236107 [6] NCCL INFO Channel 07/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-168-30:235653:236103 [5] NCCL INFO Channel 07/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-168-30:235655:236104 [7] NCCL INFO Channel 02/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-168-30:235648:236102 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-168-30:235651:236108 [3] NCCL INFO Channel 07/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-168-30:235652:236105 [4] NCCL INFO Channel 07/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-168-30:235650:236106 [2] NCCL INFO Channel 07/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-168-30:235649:236109 [1] NCCL INFO Channel 02/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-168-30:235654:236107 [6] NCCL INFO Channel 08/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-168-30:235653:236103 [5] NCCL INFO Channel 08/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-168-30:235655:236104 [7] NCCL INFO Channel 03/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-168-30:235651:236108 [3] NCCL INFO Channel 08/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-168-30:235648:236102 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-168-30:235652:236105 [4] NCCL INFO Channel 08/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-225:124163:124626 [3] NCCL INFO Channel 19/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-225:124160:124620 [0] NCCL INFO Channel 17/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-168-30:235650:236106 [2] NCCL INFO Channel 08/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-168-30:235649:236109 [1] NCCL INFO Channel 03/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-168-30:235654:236107 [6] NCCL INFO Channel 09/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-168-30:235653:236103 [5] NCCL INFO Channel 09/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-168-30:235655:236104 [7] NCCL INFO Channel 04/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-138:96567:97050 [4] NCCL INFO Channel 20/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-168-30:235651:236108 [3] NCCL INFO Channel 09/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-168-30:235652:236105 [4] NCCL INFO Channel 09/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-168-30:235648:236102 [0] NCCL INFO Channel 04/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-168-30:235650:236106 [2] NCCL INFO Channel 09/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-225:124162:124627 [2] NCCL INFO Channel 21/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-168-30:235649:236109 [1] NCCL INFO Channel 04/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-168-30:235654:236107 [6] NCCL INFO Channel 10/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-168-30:235655:236104 [7] NCCL INFO Channel 05/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-168-30:235653:236103 [5] NCCL INFO Channel 10/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-168-30:235651:236108 [3] NCCL INFO Channel 10/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-168-30:235652:236105 [4] NCCL INFO Channel 10/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-168-30:235648:236102 [0] NCCL INFO Channel 05/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-168-30:235650:236106 [2] NCCL INFO Channel 10/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-225:124161:124621 [1] NCCL INFO Channel 20/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-164:1227470:1227923 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-165-131:1630208:1630691 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-165-131:1630208:1630691 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-165-164:1227470:1227923 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-168-30:235654:236107 [6] NCCL INFO Channel 11/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-168-30:235649:236109 [1] NCCL INFO Channel 05/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-168-30:235655:236104 [7] NCCL INFO Channel 06/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-168-30:235653:236103 [5] NCCL INFO Channel 11/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-168-30:235648:236102 [0] NCCL INFO Channel 06/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-168-30:235651:236108 [3] NCCL INFO Channel 11/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-168-30:235652:236105 [4] NCCL INFO Channel 11/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-168-30:235650:236106 [2] NCCL INFO Channel 11/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-225:124160:124620 [0] NCCL INFO Channel 18/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-225:124163:124626 [3] NCCL INFO Channel 20/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-168-30:235649:236109 [1] NCCL INFO Channel 06/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-168-30:235651:236108 [3] NCCL INFO Channel 12/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-225:124162:124627 [2] NCCL INFO Channel 22/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-168-30:235652:236105 [4] NCCL INFO Channel 12/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-168-30:235649:236109 [1] NCCL INFO Channel 07/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-168-30:235651:236108 [3] NCCL INFO Channel 13/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-138:96567:97050 [4] NCCL INFO Channel 21/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-168-30:235655:236104 [7] NCCL INFO Channel 07/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-168-30:235652:236105 [4] NCCL INFO Channel 13/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-168-30:235654:236107 [6] NCCL INFO Channel 12/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-168-30:235653:236103 [5] NCCL INFO Channel 12/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-168-30:235655:236104 [7] NCCL INFO Channel 08/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-168-30:235652:236105 [4] NCCL INFO Channel 14/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-59:1557266:1557721 [0] NCCL INFO Connected all rings
+ip-26-0-160-225:124161:124621 [1] NCCL INFO Channel 21/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-168-30:235654:236107 [6] NCCL INFO Channel 13/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-168-30:235653:236103 [5] NCCL INFO Channel 13/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-168-30:235655:236104 [7] NCCL INFO Channel 09/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-168-30:235652:236105 [4] NCCL INFO Channel 15/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-168-30:235654:236107 [6] NCCL INFO Channel 14/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-225:124160:124620 [0] NCCL INFO Channel 19/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-225:124163:124626 [3] NCCL INFO Channel 21/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-168-30:235652:236105 [4] NCCL INFO Channel 16/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-168-30:235648:236102 [0] NCCL INFO Channel 07/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-168-30:235650:236106 [2] NCCL INFO Channel 12/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-168-30:235652:236105 [4] NCCL INFO Channel 17/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-225:124162:124627 [2] NCCL INFO Channel 23/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-168-30:235653:236103 [5] NCCL INFO Channel 14/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-225:124161:124621 [1] NCCL INFO Channel 22/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-168-30:235655:236104 [7] NCCL INFO Channel 10/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-168-30:235648:236102 [0] NCCL INFO Channel 08/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-168-30:235650:236106 [2] NCCL INFO Channel 13/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-168-30:235654:236107 [6] NCCL INFO Channel 15/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-164:1227469:1227920 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-165-164:1227469:1227920 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-160-225:124163:124626 [3] NCCL INFO Channel 22/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-168-30:235655:236104 [7] NCCL INFO Channel 11/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-225:124160:124620 [0] NCCL INFO Channel 20/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-168-30:235648:236102 [0] NCCL INFO Channel 09/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-168-30:235650:236106 [2] NCCL INFO Channel 14/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-168-30:235654:236107 [6] NCCL INFO Channel 16/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-168-30:235655:236104 [7] NCCL INFO Channel 12/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-168-30:235648:236102 [0] NCCL INFO Channel 10/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-168-30:235650:236106 [2] NCCL INFO Channel 15/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-168-30:235654:236107 [6] NCCL INFO Channel 17/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-168-30:235655:236104 [7] NCCL INFO Channel 13/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-168-30:235648:236102 [0] NCCL INFO Channel 11/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-168-30:235650:236106 [2] NCCL INFO Channel 16/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-168-30:235654:236107 [6] NCCL INFO Channel 18/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-168-30:235655:236104 [7] NCCL INFO Channel 14/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-168-30:235648:236102 [0] NCCL INFO Channel 12/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-168-30:235650:236106 [2] NCCL INFO Channel 17/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-168-30:235654:236107 [6] NCCL INFO Channel 19/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-168-30:235655:236104 [7] NCCL INFO Channel 15/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-168-30:235648:236102 [0] NCCL INFO Channel 13/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-168-30:235650:236106 [2] NCCL INFO Channel 18/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-168-30:235654:236107 [6] NCCL INFO Channel 20/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-168-30:235655:236104 [7] NCCL INFO Channel 16/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-168-30:235648:236102 [0] NCCL INFO Channel 14/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-168-30:235651:236108 [3] NCCL INFO Channel 14/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-225:124161:124621 [1] NCCL INFO Channel 23/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-225:124163:124626 [3] NCCL INFO Channel 23/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-168-30:235650:236106 [2] NCCL INFO Channel 19/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-168-30:235654:236107 [6] NCCL INFO Channel 21/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-168-30:235655:236104 [7] NCCL INFO Channel 17/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-168-30:235649:236109 [1] NCCL INFO Channel 08/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-168-30:235648:236102 [0] NCCL INFO Channel 15/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-168-30:235651:236108 [3] NCCL INFO Channel 15/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-225:124160:124620 [0] NCCL INFO Channel 21/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-168-30:235652:236105 [4] NCCL INFO Channel 18/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-168-30:235650:236106 [2] NCCL INFO Channel 20/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-168-30:235654:236107 [6] NCCL INFO Channel 22/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-168-30:235653:236103 [5] NCCL INFO Channel 15/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-168-30:235655:236104 [7] NCCL INFO Channel 18/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-168-30:235649:236109 [1] NCCL INFO Channel 09/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-168-30:235648:236102 [0] NCCL INFO Channel 16/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-168-30:235651:236108 [3] NCCL INFO Channel 16/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-168-30:235652:236105 [4] NCCL INFO Channel 19/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-168-30:235650:236106 [2] NCCL INFO Channel 21/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-168-30:235653:236103 [5] NCCL INFO Channel 16/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-168-30:235655:236104 [7] NCCL INFO Channel 19/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-168-30:235654:236107 [6] NCCL INFO Channel 23/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-168-30:235649:236109 [1] NCCL INFO Channel 10/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-168-30:235648:236102 [0] NCCL INFO Channel 17/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-168-30:235651:236108 [3] NCCL INFO Channel 17/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-168-30:235652:236105 [4] NCCL INFO Channel 20/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-168-30:235650:236106 [2] NCCL INFO Channel 22/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-168-30:235653:236103 [5] NCCL INFO Channel 17/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-168-30:235655:236104 [7] NCCL INFO Channel 20/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-168-30:235649:236109 [1] NCCL INFO Channel 11/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-168-30:235648:236102 [0] NCCL INFO Channel 18/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-168-30:235651:236108 [3] NCCL INFO Channel 18/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-168-30:235650:236106 [2] NCCL INFO Channel 23/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-168-30:235652:236105 [4] NCCL INFO Channel 21/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-168-30:235653:236103 [5] NCCL INFO Channel 18/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-168-30:235655:236104 [7] NCCL INFO Channel 21/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-168-30:235649:236109 [1] NCCL INFO Channel 12/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-168-30:235648:236102 [0] NCCL INFO Channel 19/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-168-30:235651:236108 [3] NCCL INFO Channel 19/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-225:124160:124620 [0] NCCL INFO Channel 22/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-168-30:235649:236109 [1] NCCL INFO Channel 13/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-168-30:235652:236105 [4] NCCL INFO Channel 22/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-168-30:235653:236103 [5] NCCL INFO Channel 19/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-168-30:235655:236104 [7] NCCL INFO Channel 22/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-168-30:235648:236102 [0] NCCL INFO Channel 20/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-168-30:235651:236108 [3] NCCL INFO Channel 20/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-138:96567:97050 [4] NCCL INFO Channel 22/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-168-30:235649:236109 [1] NCCL INFO Channel 14/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-168-30:235652:236105 [4] NCCL INFO Channel 23/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-168-30:235653:236103 [5] NCCL INFO Channel 20/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-168-30:235655:236104 [7] NCCL INFO Channel 23/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-168-30:235648:236102 [0] NCCL INFO Channel 21/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-168-30:235651:236108 [3] NCCL INFO Channel 21/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-168-30:235649:236109 [1] NCCL INFO Channel 15/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-168-30:235653:236103 [5] NCCL INFO Channel 21/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-168-30:235648:236102 [0] NCCL INFO Channel 22/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-168-30:235651:236108 [3] NCCL INFO Channel 22/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-168-30:235649:236109 [1] NCCL INFO Channel 16/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-168-30:235653:236103 [5] NCCL INFO Channel 22/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-168-30:235651:236108 [3] NCCL INFO Channel 23/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-168-30:235648:236102 [0] NCCL INFO Channel 23/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-168-30:235649:236109 [1] NCCL INFO Channel 17/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-168-30:235653:236103 [5] NCCL INFO Channel 23/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-168-30:235649:236109 [1] NCCL INFO Channel 18/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-168-30:235649:236109 [1] NCCL INFO Channel 19/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-225:124160:124620 [0] NCCL INFO Channel 23/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-131:1630212:1630690 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-165-131:1630212:1630690 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-161-138:96567:97050 [4] NCCL INFO Channel 23/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-131:1630210:1630689 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-165-131:1630210:1630689 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-168-30:235649:236109 [1] NCCL INFO Channel 20/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-131:1630211:1630688 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-165-131:1630211:1630688 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-165-131:1630211:1630688 [6] NCCL INFO Trees [0] 7/-1/-1->6->5 [1] 7/-1/-1->6->5 [2] 7/-1/-1->6->5 [3] 7/-1/-1->6->5 [4] 7/-1/-1->6->5 [5] 7/-1/-1->6->5 [6] 7/-1/-1->6->5 [7] 7/-1/-1->6->5 [8] 7/-1/-1->6->5 [9] 7/-1/-1->6->5 [10] 7/-1/-1->6->5 [11] 7/-1/-1->6->5 [12] 7/-1/-1->6->5 [13] 7/-1/-1->6->5 [14] 7/-1/-1->6->5 [15] 7/-1/-1->6->5 [16] 7/-1/-1->6->5 [17] 7/-1/-1->6->5 [18] 7/-1/-1->6->5 [19] 7/-1/-1->6->5 [20] 7/-1/-1->6->5 [21] 7/-1/-1->6->5 [22] 7/-1/-1->6->5 [23] 7/-1/-1->6->5
+ip-26-0-165-131:1630211:1630688 [6] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-165-131:1630210:1630689 [5] NCCL INFO Trees [0] 6/-1/-1->5->4 [1] 6/-1/-1->5->4 [2] 6/-1/-1->5->4 [3] 6/-1/-1->5->4 [4] 6/-1/-1->5->4 [5] 6/-1/-1->5->4 [6] 6/-1/-1->5->4 [7] 6/-1/-1->5->4 [8] 6/-1/-1->5->4 [9] 6/-1/-1->5->4 [10] 6/-1/-1->5->4 [11] 6/-1/-1->5->4 [12] 6/-1/-1->5->4 [13] 6/-1/-1->5->4 [14] 6/-1/-1->5->4 [15] 6/-1/-1->5->4 [16] 6/-1/-1->5->4 [17] 6/-1/-1->5->4 [18] 6/-1/-1->5->4 [19] 6/-1/-1->5->4 [20] 6/-1/-1->5->4 [21] 6/-1/-1->5->4 [22] 6/-1/-1->5->4 [23] 6/-1/-1->5->4
+ip-26-0-165-131:1630210:1630689 [5] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-165-59:1557267:1557727 [1] NCCL INFO Connected all rings
+ip-26-0-168-30:235649:236109 [1] NCCL INFO Channel 21/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-168-30:235649:236109 [1] NCCL INFO Channel 22/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-168-30:235649:236109 [1] NCCL INFO Channel 23/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-103:96891:97347 [3] NCCL INFO Channel 00/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-103:96895:97351 [7] NCCL INFO Channel 00/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-131:1630208:1630691 [3] NCCL INFO Trees [0] 4/-1/-1->3->2 [1] 4/-1/-1->3->2 [2] 4/-1/-1->3->2 [3] 4/-1/-1->3->2 [4] 4/-1/-1->3->2 [5] 4/-1/-1->3->2 [6] 4/-1/-1->3->2 [7] 4/-1/-1->3->2 [8] 4/-1/-1->3->2 [9] 4/-1/-1->3->2 [10] 4/-1/-1->3->2 [11] 4/-1/-1->3->2 [12] 4/-1/-1->3->2 [13] 4/-1/-1->3->2 [14] 4/-1/-1->3->2 [15] 4/-1/-1->3->2 [16] 4/-1/-1->3->2 [17] 4/-1/-1->3->2 [18] 4/-1/-1->3->2 [19] 4/-1/-1->3->2 [20] 4/-1/-1->3->2 [21] 4/-1/-1->3->2 [22] 4/-1/-1->3->2 [23] 4/-1/-1->3->2
+ip-26-0-165-131:1630209:1630693 [4] NCCL INFO Trees [0] 5/-1/-1->4->3 [1] 5/-1/-1->4->3 [2] 5/-1/-1->4->3 [3] 5/-1/-1->4->3 [4] 5/-1/-1->4->3 [5] 5/-1/-1->4->3 [6] 5/-1/-1->4->3 [7] 5/-1/-1->4->3 [8] 5/-1/-1->4->3 [9] 5/-1/-1->4->3 [10] 5/-1/-1->4->3 [11] 5/-1/-1->4->3 [12] 5/-1/-1->4->3 [13] 5/-1/-1->4->3 [14] 5/-1/-1->4->3 [15] 5/-1/-1->4->3 [16] 5/-1/-1->4->3 [17] 5/-1/-1->4->3 [18] 5/-1/-1->4->3 [19] 5/-1/-1->4->3 [20] 5/-1/-1->4->3 [21] 5/-1/-1->4->3 [22] 5/-1/-1->4->3 [23] 5/-1/-1->4->3
+ip-26-0-165-131:1630208:1630691 [3] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-165-131:1630209:1630693 [4] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-165-131:1630205:1630687 [0] NCCL INFO Channel 00/24 :    0   1   2   3   4   5   6   7
+ip-26-0-165-131:1630205:1630687 [0] NCCL INFO Channel 01/24 :    0   1   2   3   4   5   6   7
+ip-26-0-165-131:1630205:1630687 [0] NCCL INFO Channel 02/24 :    0   1   2   3   4   5   6   7
+ip-26-0-165-131:1630212:1630690 [7] NCCL INFO Trees [0] -1/-1/-1->7->6 [1] -1/-1/-1->7->6 [2] -1/-1/-1->7->6 [3] -1/-1/-1->7->6 [4] -1/-1/-1->7->6 [5] -1/-1/-1->7->6 [6] -1/-1/-1->7->6 [7] -1/-1/-1->7->6 [8] -1/-1/-1->7->6 [9] -1/-1/-1->7->6 [10] -1/-1/-1->7->6 [11] -1/-1/-1->7->6 [12] -1/-1/-1->7->6 [13] -1/-1/-1->7->6 [14] -1/-1/-1->7->6 [15] -1/-1/-1->7->6 [16] -1/-1/-1->7->6 [17] -1/-1/-1->7->6 [18] -1/-1/-1->7->6 [19] -1/-1/-1->7->6 [20] -1/-1/-1->7->6 [21] -1/-1/-1->7->6 [22] -1/-1/-1->7->6 [23] -1/-1/-1->7->6
+ip-26-0-165-131:1630205:1630687 [0] NCCL INFO Channel 03/24 :    0   1   2   3   4   5   6   7
+ip-26-0-165-131:1630207:1630694 [2] NCCL INFO Trees [0] 3/-1/-1->2->1 [1] 3/-1/-1->2->1 [2] 3/-1/-1->2->1 [3] 3/-1/-1->2->1 [4] 3/-1/-1->2->1 [5] 3/-1/-1->2->1 [6] 3/-1/-1->2->1 [7] 3/-1/-1->2->1 [8] 3/-1/-1->2->1 [9] 3/-1/-1->2->1 [10] 3/-1/-1->2->1 [11] 3/-1/-1->2->1 [12] 3/-1/-1->2->1 [13] 3/-1/-1->2->1 [14] 3/-1/-1->2->1 [15] 3/-1/-1->2->1 [16] 3/-1/-1->2->1 [17] 3/-1/-1->2->1 [18] 3/-1/-1->2->1 [19] 3/-1/-1->2->1 [20] 3/-1/-1->2->1 [21] 3/-1/-1->2->1 [22] 3/-1/-1->2->1 [23] 3/-1/-1->2->1
+ip-26-0-165-131:1630212:1630690 [7] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-165-131:1630205:1630687 [0] NCCL INFO Channel 04/24 :    0   1   2   3   4   5   6   7
+ip-26-0-165-131:1630206:1630692 [1] NCCL INFO Trees [0] 2/-1/-1->1->0 [1] 2/-1/-1->1->0 [2] 2/-1/-1->1->0 [3] 2/-1/-1->1->0 [4] 2/-1/-1->1->0 [5] 2/-1/-1->1->0 [6] 2/-1/-1->1->0 [7] 2/-1/-1->1->0 [8] 2/-1/-1->1->0 [9] 2/-1/-1->1->0 [10] 2/-1/-1->1->0 [11] 2/-1/-1->1->0 [12] 2/-1/-1->1->0 [13] 2/-1/-1->1->0 [14] 2/-1/-1->1->0 [15] 2/-1/-1->1->0 [16] 2/-1/-1->1->0 [17] 2/-1/-1->1->0 [18] 2/-1/-1->1->0 [19] 2/-1/-1->1->0 [20] 2/-1/-1->1->0 [21] 2/-1/-1->1->0 [22] 2/-1/-1->1->0 [23] 2/-1/-1->1->0
+ip-26-0-165-131:1630207:1630694 [2] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-165-131:1630205:1630687 [0] NCCL INFO Channel 05/24 :    0   1   2   3   4   5   6   7
+ip-26-0-165-131:1630206:1630692 [1] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-165-131:1630205:1630687 [0] NCCL INFO Channel 06/24 :    0   1   2   3   4   5   6   7
+ip-26-0-165-131:1630205:1630687 [0] NCCL INFO Channel 07/24 :    0   1   2   3   4   5   6   7
+ip-26-0-165-131:1630205:1630687 [0] NCCL INFO Channel 08/24 :    0   1   2   3   4   5   6   7
+ip-26-0-165-131:1630205:1630687 [0] NCCL INFO Channel 09/24 :    0   1   2   3   4   5   6   7
+ip-26-0-165-131:1630205:1630687 [0] NCCL INFO Channel 10/24 :    0   1   2   3   4   5   6   7
+ip-26-0-165-131:1630205:1630687 [0] NCCL INFO Channel 11/24 :    0   1   2   3   4   5   6   7
+ip-26-0-165-131:1630205:1630687 [0] NCCL INFO Channel 12/24 :    0   1   2   3   4   5   6   7
+ip-26-0-165-131:1630205:1630687 [0] NCCL INFO Channel 13/24 :    0   1   2   3   4   5   6   7
+ip-26-0-165-131:1630205:1630687 [0] NCCL INFO Channel 14/24 :    0   1   2   3   4   5   6   7
+ip-26-0-165-131:1630205:1630687 [0] NCCL INFO Channel 15/24 :    0   1   2   3   4   5   6   7
+ip-26-0-165-131:1630205:1630687 [0] NCCL INFO Channel 16/24 :    0   1   2   3   4   5   6   7
+ip-26-0-165-131:1630205:1630687 [0] NCCL INFO Channel 17/24 :    0   1   2   3   4   5   6   7
+ip-26-0-165-131:1630205:1630687 [0] NCCL INFO Channel 18/24 :    0   1   2   3   4   5   6   7
+ip-26-0-165-131:1630205:1630687 [0] NCCL INFO Channel 19/24 :    0   1   2   3   4   5   6   7
+ip-26-0-165-131:1630205:1630687 [0] NCCL INFO Channel 20/24 :    0   1   2   3   4   5   6   7
+ip-26-0-165-131:1630205:1630687 [0] NCCL INFO Channel 21/24 :    0   1   2   3   4   5   6   7
+ip-26-0-165-131:1630205:1630687 [0] NCCL INFO Channel 22/24 :    0   1   2   3   4   5   6   7
+ip-26-0-165-131:1630205:1630687 [0] NCCL INFO Channel 23/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-103:96893:97346 [5] NCCL INFO Channel 00/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-165-131:1630205:1630687 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] 1/-1/-1->0->-1 [7] 1/-1/-1->0->-1 [8] 1/-1/-1->0->-1 [9] 1/-1/-1->0->-1 [10] 1/-1/-1->0->-1 [11] 1/-1/-1->0->-1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] 1/-1/-1->0->-1 [19] 1/-1/-1->0->-1 [20] 1/-1/-1->0->-1 [21] 1/-1/-1->0->-1 [22] 1/-1/-1->0->-1 [23] 1/-1/-1->0->-1
+ip-26-0-165-131:1630205:1630687 [0] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-165-131:1630212:1630690 [7] NCCL INFO Channel 00/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-131:1630212:1630690 [7] NCCL INFO Channel 01/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-131:1630208:1630691 [3] NCCL INFO Channel 00/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-131:1630208:1630691 [3] NCCL INFO Channel 01/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-131:1630208:1630691 [3] NCCL INFO Channel 02/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-131:1630208:1630691 [3] NCCL INFO Channel 03/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-131:1630205:1630687 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-103:96891:97347 [3] NCCL INFO Channel 01/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-131:1630208:1630691 [3] NCCL INFO Channel 04/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-131:1630205:1630687 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-131:1630208:1630691 [3] NCCL INFO Channel 05/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-131:1630209:1630693 [4] NCCL INFO Channel 00/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-131:1630205:1630687 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-131:1630207:1630694 [2] NCCL INFO Channel 00/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-131:1630208:1630691 [3] NCCL INFO Channel 06/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-131:1630206:1630692 [1] NCCL INFO Channel 00/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-103:96895:97351 [7] NCCL INFO Channel 01/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-131:1630209:1630693 [4] NCCL INFO Channel 01/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-131:1630205:1630687 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-131:1630212:1630690 [7] NCCL INFO Channel 02/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-131:1630207:1630694 [2] NCCL INFO Channel 01/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-131:1630208:1630691 [3] NCCL INFO Channel 07/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-131:1630206:1630692 [1] NCCL INFO Channel 01/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-131:1630211:1630688 [6] NCCL INFO Channel 00/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-131:1630209:1630693 [4] NCCL INFO Channel 02/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-103:96890:97350 [2] NCCL INFO Channel 00/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-103:96893:97346 [5] NCCL INFO Channel 01/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-103:96891:97347 [3] NCCL INFO Channel 02/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-103:96895:97351 [7] NCCL INFO Channel 02/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-103:96890:97350 [2] NCCL INFO Channel 01/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-164:1227464:1227926 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-165-164:1227464:1227926 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-161-103:96893:97346 [5] NCCL INFO Channel 02/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-103:96891:97347 [3] NCCL INFO Channel 03/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-103:96895:97351 [7] NCCL INFO Channel 03/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-103:96890:97350 [2] NCCL INFO Channel 02/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-103:96891:97347 [3] NCCL INFO Channel 04/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-103:96895:97351 [7] NCCL INFO Channel 04/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-103:96892:97345 [4] NCCL INFO Channel 00/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-103:96890:97350 [2] NCCL INFO Channel 03/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-103:96891:97347 [3] NCCL INFO Channel 05/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-103:96893:97346 [5] NCCL INFO Channel 03/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-103:96895:97351 [7] NCCL INFO Channel 05/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-103:96892:97345 [4] NCCL INFO Channel 01/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-103:96890:97350 [2] NCCL INFO Channel 04/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-103:96891:97347 [3] NCCL INFO Channel 06/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-103:96893:97346 [5] NCCL INFO Channel 04/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-103:96895:97351 [7] NCCL INFO Channel 06/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-103:96892:97345 [4] NCCL INFO Channel 02/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-103:96890:97350 [2] NCCL INFO Channel 05/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-103:96891:97347 [3] NCCL INFO Channel 07/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-103:96893:97346 [5] NCCL INFO Channel 05/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-103:96888:97344 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-103:96894:97349 [6] NCCL INFO Channel 00/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-164:1227468:1227922 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-165-164:1227468:1227922 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-161-103:96895:97351 [7] NCCL INFO Channel 07/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-103:96892:97345 [4] NCCL INFO Channel 03/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-103:96890:97350 [2] NCCL INFO Channel 06/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-164:1227468:1227922 [5] NCCL INFO Trees [0] 6/-1/-1->5->4 [1] 6/-1/-1->5->4 [2] 6/-1/-1->5->4 [3] 6/-1/-1->5->4 [4] 6/-1/-1->5->4 [5] 6/-1/-1->5->4 [6] 6/-1/-1->5->4 [7] 6/-1/-1->5->4 [8] 6/-1/-1->5->4 [9] 6/-1/-1->5->4 [10] 6/-1/-1->5->4 [11] 6/-1/-1->5->4 [12] 6/-1/-1->5->4 [13] 6/-1/-1->5->4 [14] 6/-1/-1->5->4 [15] 6/-1/-1->5->4 [16] 6/-1/-1->5->4 [17] 6/-1/-1->5->4 [18] 6/-1/-1->5->4 [19] 6/-1/-1->5->4 [20] 6/-1/-1->5->4 [21] 6/-1/-1->5->4 [22] 6/-1/-1->5->4 [23] 6/-1/-1->5->4
+ip-26-0-165-164:1227468:1227922 [5] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-165-164:1227469:1227920 [6] NCCL INFO Trees [0] 7/-1/-1->6->5 [1] 7/-1/-1->6->5 [2] 7/-1/-1->6->5 [3] 7/-1/-1->6->5 [4] 7/-1/-1->6->5 [5] 7/-1/-1->6->5 [6] 7/-1/-1->6->5 [7] 7/-1/-1->6->5 [8] 7/-1/-1->6->5 [9] 7/-1/-1->6->5 [10] 7/-1/-1->6->5 [11] 7/-1/-1->6->5 [12] 7/-1/-1->6->5 [13] 7/-1/-1->6->5 [14] 7/-1/-1->6->5 [15] 7/-1/-1->6->5 [16] 7/-1/-1->6->5 [17] 7/-1/-1->6->5 [18] 7/-1/-1->6->5 [19] 7/-1/-1->6->5 [20] 7/-1/-1->6->5 [21] 7/-1/-1->6->5 [22] 7/-1/-1->6->5 [23] 7/-1/-1->6->5
+ip-26-0-165-164:1227469:1227920 [6] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-165-164:1227466:1227921 [3] NCCL INFO Trees [0] 4/-1/-1->3->2 [1] 4/-1/-1->3->2 [2] 4/-1/-1->3->2 [3] 4/-1/-1->3->2 [4] 4/-1/-1->3->2 [5] 4/-1/-1->3->2 [6] 4/-1/-1->3->2 [7] 4/-1/-1->3->2 [8] 4/-1/-1->3->2 [9] 4/-1/-1->3->2 [10] 4/-1/-1->3->2 [11] 4/-1/-1->3->2 [12] 4/-1/-1->3->2 [13] 4/-1/-1->3->2 [14] 4/-1/-1->3->2 [15] 4/-1/-1->3->2 [16] 4/-1/-1->3->2 [17] 4/-1/-1->3->2 [18] 4/-1/-1->3->2 [19] 4/-1/-1->3->2 [20] 4/-1/-1->3->2 [21] 4/-1/-1->3->2 [22] 4/-1/-1->3->2 [23] 4/-1/-1->3->2
+ip-26-0-165-164:1227463:1227919 [0] NCCL INFO Channel 00/24 :    0   1   2   3   4   5   6   7
+ip-26-0-165-164:1227466:1227921 [3] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-165-164:1227463:1227919 [0] NCCL INFO Channel 01/24 :    0   1   2   3   4   5   6   7
+ip-26-0-165-164:1227463:1227919 [0] NCCL INFO Channel 02/24 :    0   1   2   3   4   5   6   7
+ip-26-0-165-164:1227463:1227919 [0] NCCL INFO Channel 03/24 :    0   1   2   3   4   5   6   7
+ip-26-0-165-164:1227463:1227919 [0] NCCL INFO Channel 04/24 :    0   1   2   3   4   5   6   7
+ip-26-0-165-164:1227470:1227923 [7] NCCL INFO Trees [0] -1/-1/-1->7->6 [1] -1/-1/-1->7->6 [2] -1/-1/-1->7->6 [3] -1/-1/-1->7->6 [4] -1/-1/-1->7->6 [5] -1/-1/-1->7->6 [6] -1/-1/-1->7->6 [7] -1/-1/-1->7->6 [8] -1/-1/-1->7->6 [9] -1/-1/-1->7->6 [10] -1/-1/-1->7->6 [11] -1/-1/-1->7->6 [12] -1/-1/-1->7->6 [13] -1/-1/-1->7->6 [14] -1/-1/-1->7->6 [15] -1/-1/-1->7->6 [16] -1/-1/-1->7->6 [17] -1/-1/-1->7->6 [18] -1/-1/-1->7->6 [19] -1/-1/-1->7->6 [20] -1/-1/-1->7->6 [21] -1/-1/-1->7->6 [22] -1/-1/-1->7->6 [23] -1/-1/-1->7->6
+ip-26-0-165-164:1227464:1227926 [1] NCCL INFO Trees [0] 2/-1/-1->1->0 [1] 2/-1/-1->1->0 [2] 2/-1/-1->1->0 [3] 2/-1/-1->1->0 [4] 2/-1/-1->1->0 [5] 2/-1/-1->1->0 [6] 2/-1/-1->1->0 [7] 2/-1/-1->1->0 [8] 2/-1/-1->1->0 [9] 2/-1/-1->1->0 [10] 2/-1/-1->1->0 [11] 2/-1/-1->1->0 [12] 2/-1/-1->1->0 [13] 2/-1/-1->1->0 [14] 2/-1/-1->1->0 [15] 2/-1/-1->1->0 [16] 2/-1/-1->1->0 [17] 2/-1/-1->1->0 [18] 2/-1/-1->1->0 [19] 2/-1/-1->1->0 [20] 2/-1/-1->1->0 [21] 2/-1/-1->1->0 [22] 2/-1/-1->1->0 [23] 2/-1/-1->1->0
+ip-26-0-165-164:1227463:1227919 [0] NCCL INFO Channel 05/24 :    0   1   2   3   4   5   6   7
+ip-26-0-165-164:1227465:1227924 [2] NCCL INFO Trees [0] 3/-1/-1->2->1 [1] 3/-1/-1->2->1 [2] 3/-1/-1->2->1 [3] 3/-1/-1->2->1 [4] 3/-1/-1->2->1 [5] 3/-1/-1->2->1 [6] 3/-1/-1->2->1 [7] 3/-1/-1->2->1 [8] 3/-1/-1->2->1 [9] 3/-1/-1->2->1 [10] 3/-1/-1->2->1 [11] 3/-1/-1->2->1 [12] 3/-1/-1->2->1 [13] 3/-1/-1->2->1 [14] 3/-1/-1->2->1 [15] 3/-1/-1->2->1 [16] 3/-1/-1->2->1 [17] 3/-1/-1->2->1 [18] 3/-1/-1->2->1 [19] 3/-1/-1->2->1 [20] 3/-1/-1->2->1 [21] 3/-1/-1->2->1 [22] 3/-1/-1->2->1 [23] 3/-1/-1->2->1
+ip-26-0-165-164:1227467:1227925 [4] NCCL INFO Trees [0] 5/-1/-1->4->3 [1] 5/-1/-1->4->3 [2] 5/-1/-1->4->3 [3] 5/-1/-1->4->3 [4] 5/-1/-1->4->3 [5] 5/-1/-1->4->3 [6] 5/-1/-1->4->3 [7] 5/-1/-1->4->3 [8] 5/-1/-1->4->3 [9] 5/-1/-1->4->3 [10] 5/-1/-1->4->3 [11] 5/-1/-1->4->3 [12] 5/-1/-1->4->3 [13] 5/-1/-1->4->3 [14] 5/-1/-1->4->3 [15] 5/-1/-1->4->3 [16] 5/-1/-1->4->3 [17] 5/-1/-1->4->3 [18] 5/-1/-1->4->3 [19] 5/-1/-1->4->3 [20] 5/-1/-1->4->3 [21] 5/-1/-1->4->3 [22] 5/-1/-1->4->3 [23] 5/-1/-1->4->3
+ip-26-0-165-164:1227463:1227919 [0] NCCL INFO Channel 06/24 :    0   1   2   3   4   5   6   7
+ip-26-0-165-164:1227464:1227926 [1] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-165-164:1227463:1227919 [0] NCCL INFO Channel 07/24 :    0   1   2   3   4   5   6   7
+ip-26-0-165-164:1227470:1227923 [7] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-165-164:1227467:1227925 [4] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-165-164:1227465:1227924 [2] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-165-164:1227463:1227919 [0] NCCL INFO Channel 08/24 :    0   1   2   3   4   5   6   7
+ip-26-0-165-164:1227463:1227919 [0] NCCL INFO Channel 09/24 :    0   1   2   3   4   5   6   7
+ip-26-0-165-164:1227463:1227919 [0] NCCL INFO Channel 10/24 :    0   1   2   3   4   5   6   7
+ip-26-0-165-164:1227463:1227919 [0] NCCL INFO Channel 11/24 :    0   1   2   3   4   5   6   7
+ip-26-0-165-164:1227463:1227919 [0] NCCL INFO Channel 12/24 :    0   1   2   3   4   5   6   7
+ip-26-0-165-164:1227463:1227919 [0] NCCL INFO Channel 13/24 :    0   1   2   3   4   5   6   7
+ip-26-0-165-164:1227463:1227919 [0] NCCL INFO Channel 14/24 :    0   1   2   3   4   5   6   7
+ip-26-0-165-164:1227463:1227919 [0] NCCL INFO Channel 15/24 :    0   1   2   3   4   5   6   7
+ip-26-0-165-164:1227463:1227919 [0] NCCL INFO Channel 16/24 :    0   1   2   3   4   5   6   7
+ip-26-0-165-164:1227463:1227919 [0] NCCL INFO Channel 17/24 :    0   1   2   3   4   5   6   7
+ip-26-0-161-103:96891:97347 [3] NCCL INFO Channel 08/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-164:1227463:1227919 [0] NCCL INFO Channel 18/24 :    0   1   2   3   4   5   6   7
+ip-26-0-165-164:1227463:1227919 [0] NCCL INFO Channel 19/24 :    0   1   2   3   4   5   6   7
+ip-26-0-165-164:1227463:1227919 [0] NCCL INFO Channel 20/24 :    0   1   2   3   4   5   6   7
+ip-26-0-165-164:1227463:1227919 [0] NCCL INFO Channel 21/24 :    0   1   2   3   4   5   6   7
+ip-26-0-165-164:1227463:1227919 [0] NCCL INFO Channel 22/24 :    0   1   2   3   4   5   6   7
+ip-26-0-165-164:1227463:1227919 [0] NCCL INFO Channel 23/24 :    0   1   2   3   4   5   6   7
+ip-26-0-165-164:1227463:1227919 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] 1/-1/-1->0->-1 [7] 1/-1/-1->0->-1 [8] 1/-1/-1->0->-1 [9] 1/-1/-1->0->-1 [10] 1/-1/-1->0->-1 [11] 1/-1/-1->0->-1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] 1/-1/-1->0->-1 [19] 1/-1/-1->0->-1 [20] 1/-1/-1->0->-1 [21] 1/-1/-1->0->-1 [22] 1/-1/-1->0->-1 [23] 1/-1/-1->0->-1
+ip-26-0-165-164:1227463:1227919 [0] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-161-103:96890:97350 [2] NCCL INFO Channel 07/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-103:96888:97344 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-103:96893:97346 [5] NCCL INFO Channel 06/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-103:96895:97351 [7] NCCL INFO Channel 08/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-103:96889:97348 [1] NCCL INFO Channel 00/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-103:96888:97344 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-103:96893:97346 [5] NCCL INFO Channel 07/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-103:96890:97350 [2] NCCL INFO Channel 08/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-103:96895:97351 [7] NCCL INFO Channel 09/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-103:96889:97348 [1] NCCL INFO Channel 01/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-103:96888:97344 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-103:96893:97346 [5] NCCL INFO Channel 08/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-103:96890:97350 [2] NCCL INFO Channel 09/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-103:96895:97351 [7] NCCL INFO Channel 10/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-103:96889:97348 [1] NCCL INFO Channel 02/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-103:96893:97346 [5] NCCL INFO Channel 09/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-103:96888:97344 [0] NCCL INFO Channel 04/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-103:96890:97350 [2] NCCL INFO Channel 10/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-103:96895:97351 [7] NCCL INFO Channel 11/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-103:96889:97348 [1] NCCL INFO Channel 03/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-103:96893:97346 [5] NCCL INFO Channel 10/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-103:96888:97344 [0] NCCL INFO Channel 05/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-103:96894:97349 [6] NCCL INFO Channel 01/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-103:96890:97350 [2] NCCL INFO Channel 11/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-103:96895:97351 [7] NCCL INFO Channel 12/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-103:96892:97345 [4] NCCL INFO Channel 04/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-138:96563:97045 [0] NCCL INFO Connected all rings
+ip-26-0-161-103:96891:97347 [3] NCCL INFO Channel 09/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-103:96889:97348 [1] NCCL INFO Channel 04/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-103:96893:97346 [5] NCCL INFO Channel 11/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-103:96888:97344 [0] NCCL INFO Channel 06/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-103:96890:97350 [2] NCCL INFO Channel 12/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-103:96894:97349 [6] NCCL INFO Channel 02/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-103:96895:97351 [7] NCCL INFO Channel 13/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-103:96892:97345 [4] NCCL INFO Channel 05/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-103:96891:97347 [3] NCCL INFO Channel 10/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-103:96889:97348 [1] NCCL INFO Channel 05/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-103:96893:97346 [5] NCCL INFO Channel 12/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-103:96888:97344 [0] NCCL INFO Channel 07/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-103:96890:97350 [2] NCCL INFO Channel 13/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-103:96894:97349 [6] NCCL INFO Channel 03/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-103:96895:97351 [7] NCCL INFO Channel 14/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-103:96892:97345 [4] NCCL INFO Channel 06/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-103:96891:97347 [3] NCCL INFO Channel 11/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-103:96889:97348 [1] NCCL INFO Channel 06/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-103:96893:97346 [5] NCCL INFO Channel 13/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-103:96888:97344 [0] NCCL INFO Channel 08/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-103:96890:97350 [2] NCCL INFO Channel 14/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-103:96894:97349 [6] NCCL INFO Channel 04/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-103:96895:97351 [7] NCCL INFO Channel 15/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-103:96892:97345 [4] NCCL INFO Channel 07/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-103:96891:97347 [3] NCCL INFO Channel 12/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-103:96889:97348 [1] NCCL INFO Channel 07/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-103:96893:97346 [5] NCCL INFO Channel 14/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-103:96888:97344 [0] NCCL INFO Channel 09/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-103:96890:97350 [2] NCCL INFO Channel 15/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-103:96894:97349 [6] NCCL INFO Channel 05/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-131:1630205:1630687 [0] NCCL INFO Channel 04/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-131:1630212:1630690 [7] NCCL INFO Channel 03/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-131:1630207:1630694 [2] NCCL INFO Channel 02/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-131:1630208:1630691 [3] NCCL INFO Channel 08/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-131:1630206:1630692 [1] NCCL INFO Channel 02/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-131:1630211:1630688 [6] NCCL INFO Channel 01/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-131:1630209:1630693 [4] NCCL INFO Channel 03/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-131:1630205:1630687 [0] NCCL INFO Channel 05/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-131:1630212:1630690 [7] NCCL INFO Channel 04/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-131:1630207:1630694 [2] NCCL INFO Channel 03/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-131:1630208:1630691 [3] NCCL INFO Channel 09/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-131:1630206:1630692 [1] NCCL INFO Channel 03/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-131:1630211:1630688 [6] NCCL INFO Channel 02/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-131:1630209:1630693 [4] NCCL INFO Channel 04/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-131:1630205:1630687 [0] NCCL INFO Channel 06/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-131:1630212:1630690 [7] NCCL INFO Channel 05/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-103:96892:97345 [4] NCCL INFO Channel 08/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-131:1630207:1630694 [2] NCCL INFO Channel 04/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-131:1630208:1630691 [3] NCCL INFO Channel 10/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-131:1630206:1630692 [1] NCCL INFO Channel 04/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-131:1630211:1630688 [6] NCCL INFO Channel 03/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-131:1630209:1630693 [4] NCCL INFO Channel 05/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-131:1630205:1630687 [0] NCCL INFO Channel 07/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-131:1630212:1630690 [7] NCCL INFO Channel 06/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-131:1630207:1630694 [2] NCCL INFO Channel 05/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-131:1630208:1630691 [3] NCCL INFO Channel 11/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-131:1630206:1630692 [1] NCCL INFO Channel 05/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-131:1630211:1630688 [6] NCCL INFO Channel 04/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-131:1630209:1630693 [4] NCCL INFO Channel 06/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-131:1630212:1630690 [7] NCCL INFO Channel 07/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-131:1630205:1630687 [0] NCCL INFO Channel 08/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-131:1630207:1630694 [2] NCCL INFO Channel 06/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-103:96895:97351 [7] NCCL INFO Channel 16/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-131:1630208:1630691 [3] NCCL INFO Channel 12/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-131:1630206:1630692 [1] NCCL INFO Channel 06/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-131:1630211:1630688 [6] NCCL INFO Channel 05/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-131:1630209:1630693 [4] NCCL INFO Channel 07/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-131:1630210:1630689 [5] NCCL INFO Channel 00/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-165-131:1630212:1630690 [7] NCCL INFO Channel 08/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-131:1630205:1630687 [0] NCCL INFO Channel 09/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-131:1630207:1630694 [2] NCCL INFO Channel 07/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-131:1630208:1630691 [3] NCCL INFO Channel 13/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-131:1630206:1630692 [1] NCCL INFO Channel 07/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-131:1630211:1630688 [6] NCCL INFO Channel 06/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-131:1630209:1630693 [4] NCCL INFO Channel 08/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-131:1630210:1630689 [5] NCCL INFO Channel 01/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-165-131:1630212:1630690 [7] NCCL INFO Channel 09/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-131:1630205:1630687 [0] NCCL INFO Channel 10/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-131:1630207:1630694 [2] NCCL INFO Channel 08/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-131:1630208:1630691 [3] NCCL INFO Channel 14/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-131:1630206:1630692 [1] NCCL INFO Channel 08/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-131:1630211:1630688 [6] NCCL INFO Channel 07/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-131:1630209:1630693 [4] NCCL INFO Channel 09/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-131:1630210:1630689 [5] NCCL INFO Channel 02/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-165-131:1630212:1630690 [7] NCCL INFO Channel 10/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-131:1630205:1630687 [0] NCCL INFO Channel 11/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-131:1630207:1630694 [2] NCCL INFO Channel 09/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-131:1630208:1630691 [3] NCCL INFO Channel 15/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-131:1630206:1630692 [1] NCCL INFO Channel 09/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-131:1630211:1630688 [6] NCCL INFO Channel 08/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-131:1630209:1630693 [4] NCCL INFO Channel 10/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-131:1630210:1630689 [5] NCCL INFO Channel 03/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-165-131:1630212:1630690 [7] NCCL INFO Channel 11/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-131:1630205:1630687 [0] NCCL INFO Channel 12/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-131:1630207:1630694 [2] NCCL INFO Channel 10/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-131:1630208:1630691 [3] NCCL INFO Channel 16/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-131:1630206:1630692 [1] NCCL INFO Channel 10/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-131:1630211:1630688 [6] NCCL INFO Channel 09/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-131:1630209:1630693 [4] NCCL INFO Channel 11/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-131:1630210:1630689 [5] NCCL INFO Channel 04/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-165-131:1630212:1630690 [7] NCCL INFO Channel 12/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-131:1630205:1630687 [0] NCCL INFO Channel 13/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-131:1630207:1630694 [2] NCCL INFO Channel 11/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-131:1630206:1630692 [1] NCCL INFO Channel 11/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-131:1630208:1630691 [3] NCCL INFO Channel 17/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-131:1630211:1630688 [6] NCCL INFO Channel 10/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-131:1630209:1630693 [4] NCCL INFO Channel 12/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-131:1630210:1630689 [5] NCCL INFO Channel 05/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-165-131:1630212:1630690 [7] NCCL INFO Channel 13/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-131:1630205:1630687 [0] NCCL INFO Channel 14/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-131:1630207:1630694 [2] NCCL INFO Channel 12/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-131:1630206:1630692 [1] NCCL INFO Channel 12/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-131:1630208:1630691 [3] NCCL INFO Channel 18/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-131:1630211:1630688 [6] NCCL INFO Channel 11/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-131:1630209:1630693 [4] NCCL INFO Channel 13/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-131:1630210:1630689 [5] NCCL INFO Channel 06/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-165-131:1630212:1630690 [7] NCCL INFO Channel 14/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-131:1630205:1630687 [0] NCCL INFO Channel 15/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-131:1630207:1630694 [2] NCCL INFO Channel 13/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-131:1630206:1630692 [1] NCCL INFO Channel 13/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-131:1630208:1630691 [3] NCCL INFO Channel 19/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-131:1630211:1630688 [6] NCCL INFO Channel 12/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-103:96891:97347 [3] NCCL INFO Channel 13/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-131:1630209:1630693 [4] NCCL INFO Channel 14/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-131:1630212:1630690 [7] NCCL INFO Channel 15/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-131:1630210:1630689 [5] NCCL INFO Channel 07/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-165-131:1630205:1630687 [0] NCCL INFO Channel 16/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-131:1630207:1630694 [2] NCCL INFO Channel 14/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-131:1630208:1630691 [3] NCCL INFO Channel 20/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-131:1630211:1630688 [6] NCCL INFO Channel 13/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-131:1630207:1630694 [2] NCCL INFO Channel 15/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-103:96889:97348 [1] NCCL INFO Channel 08/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-131:1630212:1630690 [7] NCCL INFO Channel 16/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-103:96893:97346 [5] NCCL INFO Channel 15/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-103:96888:97344 [0] NCCL INFO Channel 10/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-103:96890:97350 [2] NCCL INFO Channel 16/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-131:1630208:1630691 [3] NCCL INFO Channel 21/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-103:96894:97349 [6] NCCL INFO Channel 06/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-131:1630211:1630688 [6] NCCL INFO Channel 14/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-131:1630207:1630694 [2] NCCL INFO Channel 16/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-103:96892:97345 [4] NCCL INFO Channel 09/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-131:1630212:1630690 [7] NCCL INFO Channel 17/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-103:96895:97351 [7] NCCL INFO Channel 17/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-131:1630208:1630691 [3] NCCL INFO Channel 22/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-103:96891:97347 [3] NCCL INFO Channel 14/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-103:96889:97348 [1] NCCL INFO Channel 09/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-103:96893:97346 [5] NCCL INFO Channel 16/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-103:96888:97344 [0] NCCL INFO Channel 11/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-103:96890:97350 [2] NCCL INFO Channel 17/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-103:96894:97349 [6] NCCL INFO Channel 07/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-131:1630209:1630693 [4] NCCL INFO Channel 15/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-103:96892:97345 [4] NCCL INFO Channel 10/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-103:96895:97351 [7] NCCL INFO Channel 18/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-103:96891:97347 [3] NCCL INFO Channel 15/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-103:96889:97348 [1] NCCL INFO Channel 10/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-131:1630206:1630692 [1] NCCL INFO Channel 14/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-131:1630212:1630690 [7] NCCL INFO Channel 18/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-131:1630210:1630689 [5] NCCL INFO Channel 08/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-165-131:1630205:1630687 [0] NCCL INFO Channel 17/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-131:1630209:1630693 [4] NCCL INFO Channel 16/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-103:96893:97346 [5] NCCL INFO Channel 17/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-165-131:1630207:1630694 [2] NCCL INFO Channel 17/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-103:96890:97350 [2] NCCL INFO Channel 18/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-103:96888:97344 [0] NCCL INFO Channel 12/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-103:96894:97349 [6] NCCL INFO Channel 08/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-103:96892:97345 [4] NCCL INFO Channel 11/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-103:96895:97351 [7] NCCL INFO Channel 19/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-103:96891:97347 [3] NCCL INFO Channel 16/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-103:96889:97348 [1] NCCL INFO Channel 11/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-103:96893:97346 [5] NCCL INFO Channel 18/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-165-131:1630210:1630689 [5] NCCL INFO Channel 09/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-165-131:1630211:1630688 [6] NCCL INFO Channel 15/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-103:96890:97350 [2] NCCL INFO Channel 19/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-131:1630208:1630691 [3] NCCL INFO Channel 23/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-103:96888:97344 [0] NCCL INFO Channel 13/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-103:96894:97349 [6] NCCL INFO Channel 09/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-103:96892:97345 [4] NCCL INFO Channel 12/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-131:1630205:1630687 [0] NCCL INFO Channel 18/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-131:1630210:1630689 [5] NCCL INFO Channel 10/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-165-131:1630211:1630688 [6] NCCL INFO Channel 16/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-103:96895:97351 [7] NCCL INFO Channel 20/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-103:96891:97347 [3] NCCL INFO Channel 17/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-103:96889:97348 [1] NCCL INFO Channel 12/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-131:1630205:1630687 [0] NCCL INFO Channel 19/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-131:1630210:1630689 [5] NCCL INFO Channel 11/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-165-131:1630211:1630688 [6] NCCL INFO Channel 17/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-103:96893:97346 [5] NCCL INFO Channel 19/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-103:96890:97350 [2] NCCL INFO Channel 20/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-103:96888:97344 [0] NCCL INFO Channel 14/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-103:96894:97349 [6] NCCL INFO Channel 10/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-131:1630206:1630692 [1] NCCL INFO Channel 15/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-131:1630205:1630687 [0] NCCL INFO Channel 20/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-103:96892:97345 [4] NCCL INFO Channel 13/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-131:1630210:1630689 [5] NCCL INFO Channel 12/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-165-131:1630211:1630688 [6] NCCL INFO Channel 18/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-103:96895:97351 [7] NCCL INFO Channel 21/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-103:96891:97347 [3] NCCL INFO Channel 18/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-131:1630205:1630687 [0] NCCL INFO Channel 21/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-103:96889:97348 [1] NCCL INFO Channel 13/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-131:1630206:1630692 [1] NCCL INFO Channel 16/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-103:96893:97346 [5] NCCL INFO Channel 20/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-103:96890:97350 [2] NCCL INFO Channel 21/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-131:1630210:1630689 [5] NCCL INFO Channel 13/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-165-131:1630211:1630688 [6] NCCL INFO Channel 19/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-103:96888:97344 [0] NCCL INFO Channel 15/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-103:96894:97349 [6] NCCL INFO Channel 11/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-103:96892:97345 [4] NCCL INFO Channel 14/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-131:1630205:1630687 [0] NCCL INFO Channel 22/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-131:1630206:1630692 [1] NCCL INFO Channel 17/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-131:1630210:1630689 [5] NCCL INFO Channel 14/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-165-131:1630211:1630688 [6] NCCL INFO Channel 20/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-103:96895:97351 [7] NCCL INFO Channel 22/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-103:96891:97347 [3] NCCL INFO Channel 19/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-131:1630205:1630687 [0] NCCL INFO Channel 23/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-103:96889:97348 [1] NCCL INFO Channel 14/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-131:1630206:1630692 [1] NCCL INFO Channel 18/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-103:96893:97346 [5] NCCL INFO Channel 21/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-103:96890:97350 [2] NCCL INFO Channel 22/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-131:1630210:1630689 [5] NCCL INFO Channel 15/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-165-131:1630211:1630688 [6] NCCL INFO Channel 21/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-103:96888:97344 [0] NCCL INFO Channel 16/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-103:96894:97349 [6] NCCL INFO Channel 12/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-103:96892:97345 [4] NCCL INFO Channel 15/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-131:1630206:1630692 [1] NCCL INFO Channel 19/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-103:96895:97351 [7] NCCL INFO Channel 23/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-103:96891:97347 [3] NCCL INFO Channel 20/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-131:1630210:1630689 [5] NCCL INFO Channel 16/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-165-131:1630212:1630690 [7] NCCL INFO Channel 19/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-131:1630211:1630688 [6] NCCL INFO Channel 22/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-131:1630209:1630693 [4] NCCL INFO Channel 17/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-103:96889:97348 [1] NCCL INFO Channel 15/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-103:96893:97346 [5] NCCL INFO Channel 22/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-103:96890:97350 [2] NCCL INFO Channel 23/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-131:1630207:1630694 [2] NCCL INFO Channel 18/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-103:96888:97344 [0] NCCL INFO Channel 17/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-103:96894:97349 [6] NCCL INFO Channel 13/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-131:1630206:1630692 [1] NCCL INFO Channel 20/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-103:96892:97345 [4] NCCL INFO Channel 16/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-131:1630210:1630689 [5] NCCL INFO Channel 17/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-165-131:1630211:1630688 [6] NCCL INFO Channel 23/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-131:1630212:1630690 [7] NCCL INFO Channel 20/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-131:1630209:1630693 [4] NCCL INFO Channel 18/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-131:1630207:1630694 [2] NCCL INFO Channel 19/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-103:96891:97347 [3] NCCL INFO Channel 21/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-131:1630206:1630692 [1] NCCL INFO Channel 21/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-103:96889:97348 [1] NCCL INFO Channel 16/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-103:96888:97344 [0] NCCL INFO Channel 18/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-103:96893:97346 [5] NCCL INFO Channel 23/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-103:96894:97349 [6] NCCL INFO Channel 14/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-103:96892:97345 [4] NCCL INFO Channel 17/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-131:1630210:1630689 [5] NCCL INFO Channel 18/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-165-131:1630212:1630690 [7] NCCL INFO Channel 21/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-131:1630209:1630693 [4] NCCL INFO Channel 19/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-131:1630207:1630694 [2] NCCL INFO Channel 20/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-131:1630206:1630692 [1] NCCL INFO Channel 22/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-103:96891:97347 [3] NCCL INFO Channel 22/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-103:96889:97348 [1] NCCL INFO Channel 17/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-103:96888:97344 [0] NCCL INFO Channel 19/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-103:96894:97349 [6] NCCL INFO Channel 15/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-103:96892:97345 [4] NCCL INFO Channel 18/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-131:1630210:1630689 [5] NCCL INFO Channel 19/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-165-131:1630212:1630690 [7] NCCL INFO Channel 22/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-131:1630209:1630693 [4] NCCL INFO Channel 20/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-131:1630207:1630694 [2] NCCL INFO Channel 21/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-131:1630206:1630692 [1] NCCL INFO Channel 23/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-103:96891:97347 [3] NCCL INFO Channel 23/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-103:96889:97348 [1] NCCL INFO Channel 18/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-103:96888:97344 [0] NCCL INFO Channel 20/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-103:96894:97349 [6] NCCL INFO Channel 16/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-103:96892:97345 [4] NCCL INFO Channel 19/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-164:1227466:1227921 [3] NCCL INFO Channel 00/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-164:1227469:1227920 [6] NCCL INFO Channel 00/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-164:1227465:1227924 [2] NCCL INFO Channel 00/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-131:1630210:1630689 [5] NCCL INFO Channel 20/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-103:96889:97348 [1] NCCL INFO Channel 19/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-103:96888:97344 [0] NCCL INFO Channel 21/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-164:1227468:1227922 [5] NCCL INFO Channel 00/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-103:96892:97345 [4] NCCL INFO Channel 20/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-103:96894:97349 [6] NCCL INFO Channel 17/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-164:1227463:1227919 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-131:1630212:1630690 [7] NCCL INFO Channel 23/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-131:1630209:1630693 [4] NCCL INFO Channel 21/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-103:96889:97348 [1] NCCL INFO Channel 20/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-131:1630207:1630694 [2] NCCL INFO Channel 22/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-164:1227466:1227921 [3] NCCL INFO Channel 01/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-103:96894:97349 [6] NCCL INFO Channel 18/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-164:1227465:1227924 [2] NCCL INFO Channel 01/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-103:96892:97345 [4] NCCL INFO Channel 21/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-103:96888:97344 [0] NCCL INFO Channel 22/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-164:1227469:1227920 [6] NCCL INFO Channel 01/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-59:1557273:1557723 [7] NCCL INFO Connected all rings
+ip-26-0-165-59:1557273:1557723 [7] NCCL INFO Channel 00/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-165-131:1630210:1630689 [5] NCCL INFO Channel 21/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-165-164:1227468:1227922 [5] NCCL INFO Channel 01/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-165-164:1227463:1227919 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-131:1630209:1630693 [4] NCCL INFO Channel 22/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-131:1630207:1630694 [2] NCCL INFO Channel 23/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-103:96889:97348 [1] NCCL INFO Channel 21/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-103:96894:97349 [6] NCCL INFO Channel 19/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-103:96892:97345 [4] NCCL INFO Channel 22/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-131:1630210:1630689 [5] NCCL INFO Channel 22/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-103:96888:97344 [0] NCCL INFO Channel 23/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-164:1227466:1227921 [3] NCCL INFO Channel 02/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-164:1227465:1227924 [2] NCCL INFO Channel 02/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-164:1227469:1227920 [6] NCCL INFO Channel 02/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-103:96889:97348 [1] NCCL INFO Channel 22/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-103:96894:97349 [6] NCCL INFO Channel 20/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-103:96892:97345 [4] NCCL INFO Channel 23/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-164:1227468:1227922 [5] NCCL INFO Channel 02/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-165-164:1227463:1227919 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-103:96889:97348 [1] NCCL INFO Channel 23/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-103:96894:97349 [6] NCCL INFO Channel 21/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-59:1557268:1557728 [2] NCCL INFO Connected all rings
+ip-26-0-161-103:96894:97349 [6] NCCL INFO Channel 22/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-164:1227466:1227921 [3] NCCL INFO Channel 03/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-164:1227465:1227924 [2] NCCL INFO Channel 03/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-164:1227469:1227920 [6] NCCL INFO Channel 03/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-164:1227463:1227919 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-164:1227468:1227922 [5] NCCL INFO Channel 03/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-165-131:1630209:1630693 [4] NCCL INFO Channel 23/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-164:1227466:1227921 [3] NCCL INFO Channel 04/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-164:1227465:1227924 [2] NCCL INFO Channel 04/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-164:1227469:1227920 [6] NCCL INFO Channel 04/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-164:1227463:1227919 [0] NCCL INFO Channel 04/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-164:1227468:1227922 [5] NCCL INFO Channel 04/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-165-164:1227466:1227921 [3] NCCL INFO Channel 05/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-164:1227465:1227924 [2] NCCL INFO Channel 05/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-164:1227469:1227920 [6] NCCL INFO Channel 05/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-131:1630210:1630689 [5] NCCL INFO Channel 23/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-165-164:1227463:1227919 [0] NCCL INFO Channel 05/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-164:1227468:1227922 [5] NCCL INFO Channel 05/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-165-164:1227466:1227921 [3] NCCL INFO Channel 06/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-164:1227465:1227924 [2] NCCL INFO Channel 06/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-164:1227469:1227920 [6] NCCL INFO Channel 06/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-164:1227463:1227919 [0] NCCL INFO Channel 06/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-164:1227468:1227922 [5] NCCL INFO Channel 06/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-165-164:1227466:1227921 [3] NCCL INFO Channel 07/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-164:1227465:1227924 [2] NCCL INFO Channel 07/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-164:1227469:1227920 [6] NCCL INFO Channel 07/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-59:1557269:1557724 [3] NCCL INFO Connected all rings
+ip-26-0-165-164:1227463:1227919 [0] NCCL INFO Channel 07/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-164:1227468:1227922 [5] NCCL INFO Channel 07/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-165-164:1227470:1227923 [7] NCCL INFO Channel 00/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-164:1227467:1227925 [4] NCCL INFO Channel 00/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-164:1227466:1227921 [3] NCCL INFO Channel 08/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-164:1227465:1227924 [2] NCCL INFO Channel 08/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-164:1227469:1227920 [6] NCCL INFO Channel 08/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-164:1227464:1227926 [1] NCCL INFO Channel 00/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-164:1227463:1227919 [0] NCCL INFO Channel 08/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-164:1227468:1227922 [5] NCCL INFO Channel 08/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-103:96894:97349 [6] NCCL INFO Channel 23/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-164:1227470:1227923 [7] NCCL INFO Channel 01/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-164:1227467:1227925 [4] NCCL INFO Channel 01/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-59:1557273:1557723 [7] NCCL INFO Channel 01/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-165-164:1227466:1227921 [3] NCCL INFO Channel 09/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-164:1227465:1227924 [2] NCCL INFO Channel 09/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-164:1227464:1227926 [1] NCCL INFO Channel 01/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-164:1227469:1227920 [6] NCCL INFO Channel 09/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-164:1227463:1227919 [0] NCCL INFO Channel 09/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-164:1227468:1227922 [5] NCCL INFO Channel 09/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-165-164:1227470:1227923 [7] NCCL INFO Channel 02/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-164:1227467:1227925 [4] NCCL INFO Channel 02/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-164:1227466:1227921 [3] NCCL INFO Channel 10/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-164:1227465:1227924 [2] NCCL INFO Channel 10/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-164:1227464:1227926 [1] NCCL INFO Channel 02/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-164:1227469:1227920 [6] NCCL INFO Channel 10/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-164:1227463:1227919 [0] NCCL INFO Channel 10/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-164:1227468:1227922 [5] NCCL INFO Channel 10/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-165-164:1227470:1227923 [7] NCCL INFO Channel 03/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-164:1227467:1227925 [4] NCCL INFO Channel 03/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-164:1227466:1227921 [3] NCCL INFO Channel 11/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-164:1227465:1227924 [2] NCCL INFO Channel 11/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-164:1227464:1227926 [1] NCCL INFO Channel 03/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-164:1227469:1227920 [6] NCCL INFO Channel 11/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-164:1227463:1227919 [0] NCCL INFO Channel 11/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-164:1227470:1227923 [7] NCCL INFO Channel 04/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-164:1227468:1227922 [5] NCCL INFO Channel 11/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-165-164:1227467:1227925 [4] NCCL INFO Channel 04/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-164:1227466:1227921 [3] NCCL INFO Channel 12/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-164:1227465:1227924 [2] NCCL INFO Channel 12/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-164:1227469:1227920 [6] NCCL INFO Channel 12/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-164:1227464:1227926 [1] NCCL INFO Channel 04/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-164:1227463:1227919 [0] NCCL INFO Channel 12/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-164:1227470:1227923 [7] NCCL INFO Channel 05/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-164:1227467:1227925 [4] NCCL INFO Channel 05/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-164:1227468:1227922 [5] NCCL INFO Channel 12/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-165-164:1227466:1227921 [3] NCCL INFO Channel 13/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-164:1227465:1227924 [2] NCCL INFO Channel 13/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-164:1227469:1227920 [6] NCCL INFO Channel 13/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-164:1227464:1227926 [1] NCCL INFO Channel 05/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-164:1227463:1227919 [0] NCCL INFO Channel 13/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-164:1227470:1227923 [7] NCCL INFO Channel 06/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-164:1227468:1227922 [5] NCCL INFO Channel 13/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-165-164:1227467:1227925 [4] NCCL INFO Channel 06/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-164:1227466:1227921 [3] NCCL INFO Channel 14/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-164:1227465:1227924 [2] NCCL INFO Channel 14/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-164:1227469:1227920 [6] NCCL INFO Channel 14/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-164:1227464:1227926 [1] NCCL INFO Channel 06/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-164:1227463:1227919 [0] NCCL INFO Channel 14/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-164:1227470:1227923 [7] NCCL INFO Channel 07/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-164:1227468:1227922 [5] NCCL INFO Channel 14/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-165-164:1227467:1227925 [4] NCCL INFO Channel 07/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-164:1227466:1227921 [3] NCCL INFO Channel 15/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-164:1227465:1227924 [2] NCCL INFO Channel 15/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-164:1227469:1227920 [6] NCCL INFO Channel 15/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-164:1227464:1227926 [1] NCCL INFO Channel 07/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-164:1227463:1227919 [0] NCCL INFO Channel 15/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-164:1227470:1227923 [7] NCCL INFO Channel 08/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-164:1227468:1227922 [5] NCCL INFO Channel 15/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-165-164:1227467:1227925 [4] NCCL INFO Channel 08/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-164:1227466:1227921 [3] NCCL INFO Channel 16/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-164:1227465:1227924 [2] NCCL INFO Channel 16/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-164:1227464:1227926 [1] NCCL INFO Channel 08/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-164:1227469:1227920 [6] NCCL INFO Channel 16/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-164:1227463:1227919 [0] NCCL INFO Channel 16/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-59:1557273:1557723 [7] NCCL INFO Channel 02/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-165-164:1227470:1227923 [7] NCCL INFO Channel 09/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-164:1227468:1227922 [5] NCCL INFO Channel 16/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-165-164:1227467:1227925 [4] NCCL INFO Channel 09/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-164:1227466:1227921 [3] NCCL INFO Channel 17/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-164:1227465:1227924 [2] NCCL INFO Channel 17/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-164:1227464:1227926 [1] NCCL INFO Channel 09/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-164:1227469:1227920 [6] NCCL INFO Channel 17/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-164:1227463:1227919 [0] NCCL INFO Channel 17/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-164:1227470:1227923 [7] NCCL INFO Channel 10/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-164:1227468:1227922 [5] NCCL INFO Channel 17/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-165-164:1227467:1227925 [4] NCCL INFO Channel 10/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-164:1227466:1227921 [3] NCCL INFO Channel 18/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-164:1227465:1227924 [2] NCCL INFO Channel 18/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-164:1227464:1227926 [1] NCCL INFO Channel 10/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-164:1227469:1227920 [6] NCCL INFO Channel 18/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-164:1227463:1227919 [0] NCCL INFO Channel 18/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-164:1227470:1227923 [7] NCCL INFO Channel 11/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-164:1227468:1227922 [5] NCCL INFO Channel 18/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-165-164:1227467:1227925 [4] NCCL INFO Channel 11/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-164:1227466:1227921 [3] NCCL INFO Channel 19/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-164:1227465:1227924 [2] NCCL INFO Channel 19/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-164:1227464:1227926 [1] NCCL INFO Channel 11/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-164:1227469:1227920 [6] NCCL INFO Channel 19/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-164:1227463:1227919 [0] NCCL INFO Channel 19/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-164:1227470:1227923 [7] NCCL INFO Channel 12/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-164:1227468:1227922 [5] NCCL INFO Channel 19/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-165-164:1227467:1227925 [4] NCCL INFO Channel 12/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-164:1227466:1227921 [3] NCCL INFO Channel 20/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-164:1227465:1227924 [2] NCCL INFO Channel 20/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-164:1227464:1227926 [1] NCCL INFO Channel 12/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-164:1227469:1227920 [6] NCCL INFO Channel 20/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-59:1557273:1557723 [7] NCCL INFO Channel 03/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-178:2815968:2816434 [5] NCCL INFO Connected all rings
+ip-26-0-165-164:1227463:1227919 [0] NCCL INFO Channel 20/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-164:1227470:1227923 [7] NCCL INFO Channel 13/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-164:1227468:1227922 [5] NCCL INFO Channel 20/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-165-164:1227467:1227925 [4] NCCL INFO Channel 13/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-164:1227466:1227921 [3] NCCL INFO Channel 21/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-164:1227465:1227924 [2] NCCL INFO Channel 21/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-164:1227464:1227926 [1] NCCL INFO Channel 13/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-164:1227469:1227920 [6] NCCL INFO Channel 21/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-164:1227463:1227919 [0] NCCL INFO Channel 21/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-164:1227470:1227923 [7] NCCL INFO Channel 14/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-164:1227468:1227922 [5] NCCL INFO Channel 21/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-165-164:1227467:1227925 [4] NCCL INFO Channel 14/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-164:1227466:1227921 [3] NCCL INFO Channel 22/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-164:1227465:1227924 [2] NCCL INFO Channel 22/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-164:1227464:1227926 [1] NCCL INFO Channel 14/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-164:1227469:1227920 [6] NCCL INFO Channel 22/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-164:1227463:1227919 [0] NCCL INFO Channel 22/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-164:1227470:1227923 [7] NCCL INFO Channel 15/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-164:1227468:1227922 [5] NCCL INFO Channel 22/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-165-164:1227466:1227921 [3] NCCL INFO Channel 23/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-164:1227467:1227925 [4] NCCL INFO Channel 15/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-164:1227465:1227924 [2] NCCL INFO Channel 23/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-164:1227464:1227926 [1] NCCL INFO Channel 15/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-164:1227469:1227920 [6] NCCL INFO Channel 23/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-164:1227463:1227919 [0] NCCL INFO Channel 23/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-164:1227470:1227923 [7] NCCL INFO Channel 16/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-164:1227468:1227922 [5] NCCL INFO Channel 23/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-165-164:1227467:1227925 [4] NCCL INFO Channel 16/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-164:1227464:1227926 [1] NCCL INFO Channel 16/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-164:1227470:1227923 [7] NCCL INFO Channel 17/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-164:1227467:1227925 [4] NCCL INFO Channel 17/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-164:1227464:1227926 [1] NCCL INFO Channel 17/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-164:1227470:1227923 [7] NCCL INFO Channel 18/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-164:1227467:1227925 [4] NCCL INFO Channel 18/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-164:1227464:1227926 [1] NCCL INFO Channel 18/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-164:1227470:1227923 [7] NCCL INFO Channel 19/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-164:1227467:1227925 [4] NCCL INFO Channel 19/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-164:1227464:1227926 [1] NCCL INFO Channel 19/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-164:1227470:1227923 [7] NCCL INFO Channel 20/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-164:1227467:1227925 [4] NCCL INFO Channel 20/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-164:1227464:1227926 [1] NCCL INFO Channel 20/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-164:1227470:1227923 [7] NCCL INFO Channel 21/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-59:1557273:1557723 [7] NCCL INFO Channel 04/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-165-164:1227467:1227925 [4] NCCL INFO Channel 21/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-164:1227464:1227926 [1] NCCL INFO Channel 21/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-164:1227470:1227923 [7] NCCL INFO Channel 22/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-164:1227467:1227925 [4] NCCL INFO Channel 22/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-164:1227464:1227926 [1] NCCL INFO Channel 22/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-164:1227470:1227923 [7] NCCL INFO Channel 23/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-164:1227467:1227925 [4] NCCL INFO Channel 23/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-164:1227464:1227926 [1] NCCL INFO Channel 23/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-59:1557273:1557723 [7] NCCL INFO Channel 05/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-165-59:1557273:1557723 [7] NCCL INFO Channel 06/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-165-59:1557273:1557723 [7] NCCL INFO Channel 07/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-165-59:1557273:1557723 [7] NCCL INFO Channel 08/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-165-59:1557273:1557723 [7] NCCL INFO Channel 09/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-123:32704:33159 [0] NCCL INFO Connected all rings
+ip-26-0-165-59:1557273:1557723 [7] NCCL INFO Channel 10/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-138:96570:97051 [7] NCCL INFO Connected all rings
+ip-26-0-161-138:96570:97051 [7] NCCL INFO Channel 00/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-165-59:1557273:1557723 [7] NCCL INFO Channel 11/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-123:32707:33163 [3] NCCL INFO Connected all rings
+ip-26-0-165-59:1557273:1557723 [7] NCCL INFO Channel 12/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-178:2815970:2816437 [7] NCCL INFO Connected all rings
+ip-26-0-161-178:2815970:2816437 [7] NCCL INFO Channel 00/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-138:96570:97051 [7] NCCL INFO Channel 01/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-123:32708:33162 [4] NCCL INFO Connected all rings
+ip-26-0-165-59:1557273:1557723 [7] NCCL INFO Channel 13/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-123:32705:33166 [1] NCCL INFO Connected all rings
+ip-26-0-165-59:1557273:1557723 [7] NCCL INFO Channel 14/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-178:2815964:2816440 [1] NCCL INFO Connected all rings
+ip-26-0-161-138:96570:97051 [7] NCCL INFO Channel 02/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-165-59:1557273:1557723 [7] NCCL INFO Channel 15/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-178:2815963:2816433 [0] NCCL INFO Connected all rings
+ip-26-0-161-123:32710:33160 [6] NCCL INFO Connected all rings
+ip-26-0-161-123:32706:33161 [2] NCCL INFO Connected all rings
+ip-26-0-165-59:1557273:1557723 [7] NCCL INFO Channel 16/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-138:96570:97051 [7] NCCL INFO Channel 03/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-123:32709:33164 [5] NCCL INFO Connected all rings
+ip-26-0-161-178:2815970:2816437 [7] NCCL INFO Channel 01/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-178:2815969:2816436 [6] NCCL INFO Connected all rings
+ip-26-0-161-178:2815970:2816437 [7] NCCL INFO Channel 02/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-123:32711:33165 [7] NCCL INFO Connected all rings
+ip-26-0-161-123:32711:33165 [7] NCCL INFO Channel 00/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-165-59:1557273:1557723 [7] NCCL INFO Channel 17/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-123:32711:33165 [7] NCCL INFO Channel 01/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-165-59:1557270:1557725 [4] NCCL INFO Connected all rings
+ip-26-0-161-123:32711:33165 [7] NCCL INFO Channel 02/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-123:32711:33165 [7] NCCL INFO Channel 03/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-178:2815970:2816437 [7] NCCL INFO Channel 03/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-123:32711:33165 [7] NCCL INFO Channel 04/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-123:32711:33165 [7] NCCL INFO Channel 05/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-123:32711:33165 [7] NCCL INFO Channel 06/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-138:96565:97044 [2] NCCL INFO Connected all rings
+ip-26-0-168-34:272152:272610 [2] NCCL INFO Connected all rings
+ip-26-0-161-123:32711:33165 [7] NCCL INFO Channel 07/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-165-59:1557273:1557723 [7] NCCL INFO Channel 18/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-138:96570:97051 [7] NCCL INFO Channel 04/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-123:32711:33165 [7] NCCL INFO Channel 08/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-165-59:1557272:1557722 [6] NCCL INFO Connected all rings
+ip-26-0-161-123:32711:33165 [7] NCCL INFO Channel 09/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-178:2815970:2816437 [7] NCCL INFO Channel 04/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-123:32711:33165 [7] NCCL INFO Channel 10/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-123:32711:33165 [7] NCCL INFO Channel 11/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-123:32711:33165 [7] NCCL INFO Channel 12/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-123:32711:33165 [7] NCCL INFO Channel 13/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-165-59:1557273:1557723 [7] NCCL INFO Channel 19/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-178:2815970:2816437 [7] NCCL INFO Channel 05/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-123:32711:33165 [7] NCCL INFO Channel 14/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-165-59:1557271:1557726 [5] NCCL INFO Connected all rings
+ip-26-0-165-59:1557267:1557727 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-59:1557273:1557723 [7] NCCL INFO Channel 20/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-123:32711:33165 [7] NCCL INFO Channel 15/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-165-59:1557267:1557727 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-59:1557273:1557723 [7] NCCL INFO Channel 21/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-123:32711:33165 [7] NCCL INFO Channel 16/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-165-59:1557267:1557727 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-59:1557273:1557723 [7] NCCL INFO Channel 22/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-123:32711:33165 [7] NCCL INFO Channel 17/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-165-59:1557267:1557727 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-59:1557273:1557723 [7] NCCL INFO Channel 23/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-123:32711:33165 [7] NCCL INFO Channel 18/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-165-59:1557267:1557727 [1] NCCL INFO Channel 04/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-59:1557269:1557724 [3] NCCL INFO Channel 00/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-165-59:1557267:1557727 [1] NCCL INFO Channel 05/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-123:32711:33165 [7] NCCL INFO Channel 19/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-165-59:1557269:1557724 [3] NCCL INFO Channel 01/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-165-59:1557267:1557727 [1] NCCL INFO Channel 06/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-123:32711:33165 [7] NCCL INFO Channel 20/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-165-59:1557268:1557728 [2] NCCL INFO Channel 00/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-178:2815966:2816439 [3] NCCL INFO Connected all rings
+ip-26-0-165-59:1557269:1557724 [3] NCCL INFO Channel 02/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-178:2815970:2816437 [7] NCCL INFO Channel 06/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-165-59:1557267:1557727 [1] NCCL INFO Channel 07/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-123:32711:33165 [7] NCCL INFO Channel 21/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-123:32705:33166 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-59:1557268:1557728 [2] NCCL INFO Channel 01/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-165-59:1557269:1557724 [3] NCCL INFO Channel 03/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-165-59:1557267:1557727 [1] NCCL INFO Channel 08/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-123:32708:33162 [4] NCCL INFO Channel 00/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-123:32711:33165 [7] NCCL INFO Channel 22/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-123:32705:33166 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-59:1557268:1557728 [2] NCCL INFO Channel 02/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-165-59:1557269:1557724 [3] NCCL INFO Channel 04/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-165-59:1557267:1557727 [1] NCCL INFO Channel 09/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-138:96564:97046 [1] NCCL INFO Connected all rings
+ip-26-0-161-123:32708:33162 [4] NCCL INFO Channel 01/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-123:32711:33165 [7] NCCL INFO Channel 23/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-165-59:1557268:1557728 [2] NCCL INFO Channel 03/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-123:32705:33166 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-59:1557269:1557724 [3] NCCL INFO Channel 05/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-165-59:1557267:1557727 [1] NCCL INFO Channel 10/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-123:32709:33164 [5] NCCL INFO Channel 00/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-138:96570:97051 [7] NCCL INFO Channel 05/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-165-59:1557268:1557728 [2] NCCL INFO Channel 04/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-123:32708:33162 [4] NCCL INFO Channel 02/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-165-59:1557269:1557724 [3] NCCL INFO Channel 06/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-123:32705:33166 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-59:1557267:1557727 [1] NCCL INFO Channel 11/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-123:32709:33164 [5] NCCL INFO Channel 01/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-59:1557268:1557728 [2] NCCL INFO Channel 05/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-165-59:1557269:1557724 [3] NCCL INFO Channel 07/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-123:32708:33162 [4] NCCL INFO Channel 03/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-123:32705:33166 [1] NCCL INFO Channel 04/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-59:1557267:1557727 [1] NCCL INFO Channel 12/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-178:2815967:2816438 [4] NCCL INFO Connected all rings
+ip-26-0-161-178:2815970:2816437 [7] NCCL INFO Channel 07/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-123:32709:33164 [5] NCCL INFO Channel 02/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-59:1557268:1557728 [2] NCCL INFO Channel 06/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-123:32707:33163 [3] NCCL INFO Channel 00/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-165-59:1557269:1557724 [3] NCCL INFO Channel 08/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-123:32708:33162 [4] NCCL INFO Channel 04/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-123:32705:33166 [1] NCCL INFO Channel 05/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-59:1557267:1557727 [1] NCCL INFO Channel 13/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-123:32706:33161 [2] NCCL INFO Channel 00/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-165-59:1557268:1557728 [2] NCCL INFO Channel 07/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-165-59:1557269:1557724 [3] NCCL INFO Channel 09/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-123:32709:33164 [5] NCCL INFO Channel 03/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-123:32707:33163 [3] NCCL INFO Channel 01/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-123:32710:33160 [6] NCCL INFO Channel 00/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-123:32708:33162 [4] NCCL INFO Channel 05/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-123:32705:33166 [1] NCCL INFO Channel 06/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-178:2815965:2816435 [2] NCCL INFO Connected all rings
+ip-26-0-165-59:1557267:1557727 [1] NCCL INFO Channel 14/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-59:1557268:1557728 [2] NCCL INFO Channel 08/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-165-59:1557269:1557724 [3] NCCL INFO Channel 10/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-123:32706:33161 [2] NCCL INFO Channel 01/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-178:2815970:2816437 [7] NCCL INFO Channel 08/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-123:32709:33164 [5] NCCL INFO Channel 04/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-123:32707:33163 [3] NCCL INFO Channel 02/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-123:32710:33160 [6] NCCL INFO Channel 01/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-123:32708:33162 [4] NCCL INFO Channel 06/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-123:32705:33166 [1] NCCL INFO Channel 07/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-59:1557267:1557727 [1] NCCL INFO Channel 15/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-59:1557268:1557728 [2] NCCL INFO Channel 09/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-165-59:1557269:1557724 [3] NCCL INFO Channel 11/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-138:96570:97051 [7] NCCL INFO Channel 06/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-123:32706:33161 [2] NCCL INFO Channel 02/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-123:32709:33164 [5] NCCL INFO Channel 05/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-123:32710:33160 [6] NCCL INFO Channel 02/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-123:32707:33163 [3] NCCL INFO Channel 03/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-178:2815970:2816437 [7] NCCL INFO Channel 09/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-123:32708:33162 [4] NCCL INFO Channel 07/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-165-59:1557267:1557727 [1] NCCL INFO Channel 16/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-123:32705:33166 [1] NCCL INFO Channel 08/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-59:1557269:1557724 [3] NCCL INFO Channel 12/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-165-59:1557268:1557728 [2] NCCL INFO Channel 10/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-123:32706:33161 [2] NCCL INFO Channel 03/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-123:32709:33164 [5] NCCL INFO Channel 06/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-123:32710:33160 [6] NCCL INFO Channel 03/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-123:32707:33163 [3] NCCL INFO Channel 04/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-178:2815970:2816437 [7] NCCL INFO Channel 10/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-123:32708:33162 [4] NCCL INFO Channel 08/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-165-59:1557267:1557727 [1] NCCL INFO Channel 17/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-123:32705:33166 [1] NCCL INFO Channel 09/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-59:1557269:1557724 [3] NCCL INFO Channel 13/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-165-59:1557268:1557728 [2] NCCL INFO Channel 11/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-123:32706:33161 [2] NCCL INFO Channel 04/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-123:32710:33160 [6] NCCL INFO Channel 04/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-123:32709:33164 [5] NCCL INFO Channel 07/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-123:32707:33163 [3] NCCL INFO Channel 05/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-165-59:1557267:1557727 [1] NCCL INFO Channel 18/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-123:32708:33162 [4] NCCL INFO Channel 09/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-178:2815970:2816437 [7] NCCL INFO Channel 11/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-123:32705:33166 [1] NCCL INFO Channel 10/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-59:1557268:1557728 [2] NCCL INFO Channel 12/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-165-59:1557269:1557724 [3] NCCL INFO Channel 14/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-165-59:1557267:1557727 [1] NCCL INFO Channel 19/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-123:32706:33161 [2] NCCL INFO Channel 05/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-123:32710:33160 [6] NCCL INFO Channel 05/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-123:32709:33164 [5] NCCL INFO Channel 08/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-123:32707:33163 [3] NCCL INFO Channel 06/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-123:32708:33162 [4] NCCL INFO Channel 10/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-123:32705:33166 [1] NCCL INFO Channel 11/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-59:1557269:1557724 [3] NCCL INFO Channel 15/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-178:2815970:2816437 [7] NCCL INFO Channel 12/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-165-59:1557268:1557728 [2] NCCL INFO Channel 13/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-123:32706:33161 [2] NCCL INFO Channel 06/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-123:32710:33160 [6] NCCL INFO Channel 06/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-123:32709:33164 [5] NCCL INFO Channel 09/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-123:32707:33163 [3] NCCL INFO Channel 07/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-123:32708:33162 [4] NCCL INFO Channel 11/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-123:32705:33166 [1] NCCL INFO Channel 12/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-178:2815970:2816437 [7] NCCL INFO Channel 13/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-165-59:1557269:1557724 [3] NCCL INFO Channel 16/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-165-59:1557268:1557728 [2] NCCL INFO Channel 14/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-178:2815970:2816437 [7] NCCL INFO Channel 14/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-123:32706:33161 [2] NCCL INFO Channel 07/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-123:32710:33160 [6] NCCL INFO Channel 07/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-123:32707:33163 [3] NCCL INFO Channel 08/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-123:32709:33164 [5] NCCL INFO Channel 10/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-123:32708:33162 [4] NCCL INFO Channel 12/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-165-59:1557267:1557727 [1] NCCL INFO Channel 20/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-123:32705:33166 [1] NCCL INFO Channel 13/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-178:2815970:2816437 [7] NCCL INFO Channel 15/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-165-59:1557269:1557724 [3] NCCL INFO Channel 17/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-165-59:1557268:1557728 [2] NCCL INFO Channel 15/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-165-59:1557267:1557727 [1] NCCL INFO Channel 21/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-123:32706:33161 [2] NCCL INFO Channel 08/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-178:2815970:2816437 [7] NCCL INFO Channel 16/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-123:32707:33163 [3] NCCL INFO Channel 09/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-123:32710:33160 [6] NCCL INFO Channel 08/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-123:32708:33162 [4] NCCL INFO Channel 13/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-123:32709:33164 [5] NCCL INFO Channel 11/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-123:32705:33166 [1] NCCL INFO Channel 14/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-59:1557269:1557724 [3] NCCL INFO Channel 18/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-165-59:1557268:1557728 [2] NCCL INFO Channel 16/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-178:2815970:2816437 [7] NCCL INFO Channel 17/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-165-59:1557267:1557727 [1] NCCL INFO Channel 22/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-59:1557270:1557725 [4] NCCL INFO Channel 00/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-123:32706:33161 [2] NCCL INFO Channel 09/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-123:32707:33163 [3] NCCL INFO Channel 10/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-178:2815970:2816437 [7] NCCL INFO Channel 18/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-123:32710:33160 [6] NCCL INFO Channel 09/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-123:32708:33162 [4] NCCL INFO Channel 14/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-123:32709:33164 [5] NCCL INFO Channel 12/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-59:1557269:1557724 [3] NCCL INFO Channel 19/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-165-59:1557268:1557728 [2] NCCL INFO Channel 17/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-123:32705:33166 [1] NCCL INFO Channel 15/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-59:1557270:1557725 [4] NCCL INFO Channel 01/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-165-59:1557272:1557722 [6] NCCL INFO Channel 00/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-178:2815970:2816437 [7] NCCL INFO Channel 19/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-165-59:1557269:1557724 [3] NCCL INFO Channel 20/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-165-59:1557268:1557728 [2] NCCL INFO Channel 18/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-123:32706:33161 [2] NCCL INFO Channel 10/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-123:32708:33162 [4] NCCL INFO Channel 15/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-123:32707:33163 [3] NCCL INFO Channel 11/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-123:32709:33164 [5] NCCL INFO Channel 13/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-123:32710:33160 [6] NCCL INFO Channel 10/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-123:32705:33166 [1] NCCL INFO Channel 16/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-59:1557272:1557722 [6] NCCL INFO Channel 01/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-59:1557270:1557725 [4] NCCL INFO Channel 02/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-178:2815970:2816437 [7] NCCL INFO Channel 20/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-165-59:1557269:1557724 [3] NCCL INFO Channel 21/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-165-59:1557268:1557728 [2] NCCL INFO Channel 19/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-123:32706:33161 [2] NCCL INFO Channel 11/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-123:32708:33162 [4] NCCL INFO Channel 16/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-123:32707:33163 [3] NCCL INFO Channel 12/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-123:32709:33164 [5] NCCL INFO Channel 14/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-123:32710:33160 [6] NCCL INFO Channel 11/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-59:1557272:1557722 [6] NCCL INFO Channel 02/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-59:1557270:1557725 [4] NCCL INFO Channel 03/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-178:2815970:2816437 [7] NCCL INFO Channel 21/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-123:32705:33166 [1] NCCL INFO Channel 17/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-59:1557269:1557724 [3] NCCL INFO Channel 22/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-178:2815968:2816434 [5] NCCL INFO Channel 00/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-59:1557268:1557728 [2] NCCL INFO Channel 20/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-178:2815970:2816437 [7] NCCL INFO Channel 22/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-165-59:1557272:1557722 [6] NCCL INFO Channel 03/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-59:1557270:1557725 [4] NCCL INFO Channel 04/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-123:32706:33161 [2] NCCL INFO Channel 12/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-123:32708:33162 [4] NCCL INFO Channel 17/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-123:32710:33160 [6] NCCL INFO Channel 12/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-123:32709:33164 [5] NCCL INFO Channel 15/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-123:32707:33163 [3] NCCL INFO Channel 13/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-178:2815968:2816434 [5] NCCL INFO Channel 01/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-123:32705:33166 [1] NCCL INFO Channel 18/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-59:1557269:1557724 [3] NCCL INFO Channel 23/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-165-59:1557268:1557728 [2] NCCL INFO Channel 21/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-178:2815970:2816437 [7] NCCL INFO Channel 23/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-165-59:1557272:1557722 [6] NCCL INFO Channel 04/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-59:1557271:1557726 [5] NCCL INFO Channel 00/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-59:1557270:1557725 [4] NCCL INFO Channel 05/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-178:2815968:2816434 [5] NCCL INFO Channel 02/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-123:32706:33161 [2] NCCL INFO Channel 13/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-123:32708:33162 [4] NCCL INFO Channel 18/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-123:32710:33160 [6] NCCL INFO Channel 13/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-123:32709:33164 [5] NCCL INFO Channel 16/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-123:32707:33163 [3] NCCL INFO Channel 14/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-123:32705:33166 [1] NCCL INFO Channel 19/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-59:1557268:1557728 [2] NCCL INFO Channel 22/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-165-59:1557272:1557722 [6] NCCL INFO Channel 05/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-59:1557271:1557726 [5] NCCL INFO Channel 01/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-59:1557270:1557725 [4] NCCL INFO Channel 06/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-165-59:1557267:1557727 [1] NCCL INFO Channel 23/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-59:1557268:1557728 [2] NCCL INFO Channel 23/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-123:32708:33162 [4] NCCL INFO Channel 19/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-123:32706:33161 [2] NCCL INFO Channel 14/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-123:32709:33164 [5] NCCL INFO Channel 17/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-123:32710:33160 [6] NCCL INFO Channel 14/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-123:32707:33163 [3] NCCL INFO Channel 15/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-123:32705:33166 [1] NCCL INFO Channel 20/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-59:1557272:1557722 [6] NCCL INFO Channel 06/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-59:1557271:1557726 [5] NCCL INFO Channel 02/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-59:1557270:1557725 [4] NCCL INFO Channel 07/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-178:2815968:2816434 [5] NCCL INFO Channel 03/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-59:1557272:1557722 [6] NCCL INFO Channel 07/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-59:1557271:1557726 [5] NCCL INFO Channel 03/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-59:1557270:1557725 [4] NCCL INFO Channel 08/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-138:96569:97049 [6] NCCL INFO Connected all rings
+ip-26-0-161-123:32708:33162 [4] NCCL INFO Channel 20/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-123:32706:33161 [2] NCCL INFO Channel 15/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-178:2815968:2816434 [5] NCCL INFO Channel 04/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-123:32709:33164 [5] NCCL INFO Channel 18/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-123:32707:33163 [3] NCCL INFO Channel 16/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-123:32710:33160 [6] NCCL INFO Channel 15/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-123:32705:33166 [1] NCCL INFO Channel 21/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-138:96570:97051 [7] NCCL INFO Channel 07/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-123:32706:33161 [2] NCCL INFO Channel 16/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-123:32709:33164 [5] NCCL INFO Channel 19/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-123:32708:33162 [4] NCCL INFO Channel 21/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-123:32710:33160 [6] NCCL INFO Channel 16/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-123:32707:33163 [3] NCCL INFO Channel 17/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-123:32705:33166 [1] NCCL INFO Channel 22/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-123:32709:33164 [5] NCCL INFO Channel 20/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-123:32706:33161 [2] NCCL INFO Channel 17/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-123:32708:33162 [4] NCCL INFO Channel 22/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-123:32710:33160 [6] NCCL INFO Channel 17/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-123:32707:33163 [3] NCCL INFO Channel 18/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-123:32705:33166 [1] NCCL INFO Channel 23/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-59:1557272:1557722 [6] NCCL INFO Channel 08/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-59:1557271:1557726 [5] NCCL INFO Channel 04/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-59:1557270:1557725 [4] NCCL INFO Channel 09/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-123:32709:33164 [5] NCCL INFO Channel 21/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-123:32706:33161 [2] NCCL INFO Channel 18/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-123:32708:33162 [4] NCCL INFO Channel 23/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-123:32710:33160 [6] NCCL INFO Channel 18/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-123:32707:33163 [3] NCCL INFO Channel 19/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-178:2815968:2816434 [5] NCCL INFO Channel 05/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-178:2815966:2816439 [3] NCCL INFO Channel 00/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-123:32709:33164 [5] NCCL INFO Channel 22/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-123:32706:33161 [2] NCCL INFO Channel 19/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-123:32710:33160 [6] NCCL INFO Channel 19/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-178:2815968:2816434 [5] NCCL INFO Channel 06/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-123:32707:33163 [3] NCCL INFO Channel 20/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-178:2815969:2816436 [6] NCCL INFO Channel 00/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-123:32710:33160 [6] NCCL INFO Channel 20/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-123:32706:33161 [2] NCCL INFO Channel 20/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-123:32709:33164 [5] NCCL INFO Channel 23/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-123:32707:33163 [3] NCCL INFO Channel 21/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-178:2815966:2816439 [3] NCCL INFO Channel 01/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-178:2815968:2816434 [5] NCCL INFO Channel 07/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-123:32710:33160 [6] NCCL INFO Channel 21/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-123:32706:33161 [2] NCCL INFO Channel 21/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-123:32707:33163 [3] NCCL INFO Channel 22/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-178:2815969:2816436 [6] NCCL INFO Channel 01/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-178:2815966:2816439 [3] NCCL INFO Channel 02/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-178:2815968:2816434 [5] NCCL INFO Channel 08/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-123:32710:33160 [6] NCCL INFO Channel 22/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-59:1557271:1557726 [5] NCCL INFO Channel 05/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-123:32706:33161 [2] NCCL INFO Channel 22/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-123:32707:33163 [3] NCCL INFO Channel 23/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-165-59:1557271:1557726 [5] NCCL INFO Channel 06/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-178:2815969:2816436 [6] NCCL INFO Channel 02/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-59:1557271:1557726 [5] NCCL INFO Channel 07/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-123:32706:33161 [2] NCCL INFO Channel 23/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-123:32710:33160 [6] NCCL INFO Channel 23/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-59:1557271:1557726 [5] NCCL INFO Channel 08/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-138:96570:97051 [7] NCCL INFO Channel 08/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-178:2815968:2816434 [5] NCCL INFO Channel 09/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-178:2815964:2816440 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-178:2815968:2816434 [5] NCCL INFO Channel 10/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-178:2815969:2816436 [6] NCCL INFO Channel 03/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-178:2815964:2816440 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-178:2815968:2816434 [5] NCCL INFO Channel 11/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-178:2815969:2816436 [6] NCCL INFO Channel 04/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-178:2815964:2816440 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-178:2815968:2816434 [5] NCCL INFO Channel 12/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-178:2815966:2816439 [3] NCCL INFO Channel 03/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-178:2815964:2816440 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-178:2815969:2816436 [6] NCCL INFO Channel 05/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-178:2815969:2816436 [6] NCCL INFO Channel 06/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-178:2815968:2816434 [5] NCCL INFO Channel 13/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-178:2815969:2816436 [6] NCCL INFO Channel 07/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-138:96570:97051 [7] NCCL INFO Channel 09/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-178:2815966:2816439 [3] NCCL INFO Channel 04/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-178:2815968:2816434 [5] NCCL INFO Channel 14/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-178:2815969:2816436 [6] NCCL INFO Channel 08/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-59:1557271:1557726 [5] NCCL INFO Channel 09/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-178:2815966:2816439 [3] NCCL INFO Channel 05/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-178:2815968:2816434 [5] NCCL INFO Channel 15/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-178:2815969:2816436 [6] NCCL INFO Channel 09/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-178:2815966:2816439 [3] NCCL INFO Channel 06/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-178:2815968:2816434 [5] NCCL INFO Channel 16/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-178:2815964:2816440 [1] NCCL INFO Channel 04/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-178:2815969:2816436 [6] NCCL INFO Channel 10/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-242:1364906:1365370 [4] NCCL INFO Connected all rings
+ip-26-0-165-59:1557272:1557722 [6] NCCL INFO Channel 09/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-59:1557270:1557725 [4] NCCL INFO Channel 10/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-178:2815966:2816439 [3] NCCL INFO Channel 07/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-178:2815964:2816440 [1] NCCL INFO Channel 05/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-178:2815965:2816435 [2] NCCL INFO Channel 00/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-178:2815968:2816434 [5] NCCL INFO Channel 17/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-178:2815969:2816436 [6] NCCL INFO Channel 11/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-178:2815966:2816439 [3] NCCL INFO Channel 08/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-178:2815964:2816440 [1] NCCL INFO Channel 06/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-178:2815965:2816435 [2] NCCL INFO Channel 01/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-178:2815968:2816434 [5] NCCL INFO Channel 18/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-178:2815969:2816436 [6] NCCL INFO Channel 12/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-59:1557271:1557726 [5] NCCL INFO Channel 10/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-178:2815966:2816439 [3] NCCL INFO Channel 09/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-178:2815964:2816440 [1] NCCL INFO Channel 07/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-178:2815968:2816434 [5] NCCL INFO Channel 19/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-178:2815965:2816435 [2] NCCL INFO Channel 02/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-178:2815969:2816436 [6] NCCL INFO Channel 13/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-178:2815966:2816439 [3] NCCL INFO Channel 10/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-178:2815964:2816440 [1] NCCL INFO Channel 08/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-178:2815965:2816435 [2] NCCL INFO Channel 03/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-178:2815968:2816434 [5] NCCL INFO Channel 20/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-178:2815969:2816436 [6] NCCL INFO Channel 14/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-138:96570:97051 [7] NCCL INFO Channel 10/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-178:2815966:2816439 [3] NCCL INFO Channel 11/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-178:2815964:2816440 [1] NCCL INFO Channel 09/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-178:2815965:2816435 [2] NCCL INFO Channel 04/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-178:2815968:2816434 [5] NCCL INFO Channel 21/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-178:2815969:2816436 [6] NCCL INFO Channel 15/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-59:1557270:1557725 [4] NCCL INFO Channel 11/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-178:2815966:2816439 [3] NCCL INFO Channel 12/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-178:2815964:2816440 [1] NCCL INFO Channel 10/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-178:2815965:2816435 [2] NCCL INFO Channel 05/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-178:2815968:2816434 [5] NCCL INFO Channel 22/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-178:2815969:2816436 [6] NCCL INFO Channel 16/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-178:2815964:2816440 [1] NCCL INFO Channel 11/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-178:2815965:2816435 [2] NCCL INFO Channel 06/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-178:2815966:2816439 [3] NCCL INFO Channel 13/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-178:2815968:2816434 [5] NCCL INFO Channel 23/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-178:2815967:2816438 [4] NCCL INFO Channel 00/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-178:2815969:2816436 [6] NCCL INFO Channel 17/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-59:1557272:1557722 [6] NCCL INFO Channel 10/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-178:2815964:2816440 [1] NCCL INFO Channel 12/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-178:2815966:2816439 [3] NCCL INFO Channel 14/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-178:2815965:2816435 [2] NCCL INFO Channel 07/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-178:2815967:2816438 [4] NCCL INFO Channel 01/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-178:2815969:2816436 [6] NCCL INFO Channel 18/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-59:1557271:1557726 [5] NCCL INFO Channel 11/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-178:2815964:2816440 [1] NCCL INFO Channel 13/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-178:2815965:2816435 [2] NCCL INFO Channel 08/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-178:2815966:2816439 [3] NCCL INFO Channel 15/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-178:2815967:2816438 [4] NCCL INFO Channel 02/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-178:2815964:2816440 [1] NCCL INFO Channel 14/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-178:2815965:2816435 [2] NCCL INFO Channel 09/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-178:2815966:2816439 [3] NCCL INFO Channel 16/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-165-59:1557270:1557725 [4] NCCL INFO Channel 12/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-165-59:1557271:1557726 [5] NCCL INFO Channel 12/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-178:2815967:2816438 [4] NCCL INFO Channel 03/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-178:2815964:2816440 [1] NCCL INFO Channel 15/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-178:2815965:2816435 [2] NCCL INFO Channel 10/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-178:2815966:2816439 [3] NCCL INFO Channel 17/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-178:2815967:2816438 [4] NCCL INFO Channel 04/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-178:2815964:2816440 [1] NCCL INFO Channel 16/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-178:2815965:2816435 [2] NCCL INFO Channel 11/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-178:2815966:2816439 [3] NCCL INFO Channel 18/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-178:2815967:2816438 [4] NCCL INFO Channel 05/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-178:2815965:2816435 [2] NCCL INFO Channel 12/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-178:2815964:2816440 [1] NCCL INFO Channel 17/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-138:96570:97051 [7] NCCL INFO Channel 11/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-178:2815966:2816439 [3] NCCL INFO Channel 19/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-178:2815967:2816438 [4] NCCL INFO Channel 06/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-178:2815965:2816435 [2] NCCL INFO Channel 13/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-178:2815964:2816440 [1] NCCL INFO Channel 18/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-178:2815966:2816439 [3] NCCL INFO Channel 20/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-178:2815967:2816438 [4] NCCL INFO Channel 07/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-178:2815964:2816440 [1] NCCL INFO Channel 19/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-178:2815965:2816435 [2] NCCL INFO Channel 14/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-178:2815969:2816436 [6] NCCL INFO Channel 19/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-178:2815966:2816439 [3] NCCL INFO Channel 21/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-178:2815967:2816438 [4] NCCL INFO Channel 08/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-178:2815964:2816440 [1] NCCL INFO Channel 20/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-178:2815965:2816435 [2] NCCL INFO Channel 15/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-178:2815969:2816436 [6] NCCL INFO Channel 20/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-178:2815966:2816439 [3] NCCL INFO Channel 22/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-178:2815967:2816438 [4] NCCL INFO Channel 09/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-178:2815964:2816440 [1] NCCL INFO Channel 21/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-178:2815965:2816435 [2] NCCL INFO Channel 16/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-165-59:1557270:1557725 [4] NCCL INFO Channel 13/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-178:2815969:2816436 [6] NCCL INFO Channel 21/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-59:1557271:1557726 [5] NCCL INFO Channel 13/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-178:2815967:2816438 [4] NCCL INFO Channel 10/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-178:2815966:2816439 [3] NCCL INFO Channel 23/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-178:2815964:2816440 [1] NCCL INFO Channel 22/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-178:2815965:2816435 [2] NCCL INFO Channel 17/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-178:2815969:2816436 [6] NCCL INFO Channel 22/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-178:2815967:2816438 [4] NCCL INFO Channel 11/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-178:2815964:2816440 [1] NCCL INFO Channel 23/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-178:2815965:2816435 [2] NCCL INFO Channel 18/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-178:2815969:2816436 [6] NCCL INFO Channel 23/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-178:2815967:2816438 [4] NCCL INFO Channel 12/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-178:2815965:2816435 [2] NCCL INFO Channel 19/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-165-59:1557272:1557722 [6] NCCL INFO Channel 11/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-178:2815967:2816438 [4] NCCL INFO Channel 13/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-178:2815965:2816435 [2] NCCL INFO Channel 20/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-138:96570:97051 [7] NCCL INFO Channel 12/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-165-59:1557270:1557725 [4] NCCL INFO Channel 14/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-165-59:1557271:1557726 [5] NCCL INFO Channel 14/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-138:96570:97051 [7] NCCL INFO Channel 13/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-165-59:1557272:1557722 [6] NCCL INFO Channel 12/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-178:2815967:2816438 [4] NCCL INFO Channel 14/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-178:2815965:2816435 [2] NCCL INFO Channel 21/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-142:62886:63341 [5] NCCL INFO Connected all rings
+ip-26-0-161-138:96570:97051 [7] NCCL INFO Channel 14/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-165-59:1557271:1557726 [5] NCCL INFO Channel 15/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-59:1557270:1557725 [4] NCCL INFO Channel 15/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-165-59:1557272:1557722 [6] NCCL INFO Channel 13/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-178:2815967:2816438 [4] NCCL INFO Channel 15/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-178:2815965:2816435 [2] NCCL INFO Channel 22/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-138:96570:97051 [7] NCCL INFO Channel 15/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-168-34:272154:272608 [4] NCCL INFO Connected all rings
+ip-26-0-165-59:1557271:1557726 [5] NCCL INFO Channel 16/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-59:1557272:1557722 [6] NCCL INFO Channel 14/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-59:1557270:1557725 [4] NCCL INFO Channel 16/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-178:2815967:2816438 [4] NCCL INFO Channel 16/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-165-202:153567:154023 [5] NCCL INFO Connected all rings
+ip-26-0-161-178:2815965:2816435 [2] NCCL INFO Channel 23/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-168-34:272153:272612 [3] NCCL INFO Connected all rings
+ip-26-0-161-138:96566:97047 [3] NCCL INFO Connected all rings
+ip-26-0-168-52:2722046:2722509 [1] NCCL INFO Connected all rings
+ip-26-0-161-178:2815967:2816438 [4] NCCL INFO Channel 17/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-225:124165:124625 [5] NCCL INFO Connected all rings
+ip-26-0-161-138:96570:97051 [7] NCCL INFO Channel 16/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-138:96568:97048 [5] NCCL INFO Connected all rings
+ip-26-0-168-52:2722050:2722508 [5] NCCL INFO Connected all rings
+ip-26-0-168-52:2722045:2722502 [0] NCCL INFO Connected all rings
+ip-26-0-165-59:1557271:1557726 [5] NCCL INFO Channel 17/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-59:1557272:1557722 [6] NCCL INFO Channel 15/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-59:1557270:1557725 [4] NCCL INFO Channel 17/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-178:2815967:2816438 [4] NCCL INFO Channel 18/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-138:96570:97051 [7] NCCL INFO Channel 17/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-168-34:272150:272606 [0] NCCL INFO Connected all rings
+ip-26-0-161-138:96570:97051 [7] NCCL INFO Channel 18/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-142:62881:63337 [0] NCCL INFO Connected all rings
+ip-26-0-168-34:272155:272609 [5] NCCL INFO Connected all rings
+ip-26-0-165-59:1557271:1557726 [5] NCCL INFO Channel 18/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-138:96570:97051 [7] NCCL INFO Channel 19/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-142:62887:63344 [6] NCCL INFO Connected all rings
+ip-26-0-165-59:1557272:1557722 [6] NCCL INFO Channel 16/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-59:1557270:1557725 [4] NCCL INFO Channel 18/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-225:124166:124622 [6] NCCL INFO Connected all rings
+ip-26-0-160-242:1364907:1365367 [5] NCCL INFO Connected all rings
+ip-26-0-168-52:2722049:2722504 [4] NCCL INFO Connected all rings
+ip-26-0-160-242:1364902:1365366 [0] NCCL INFO Connected all rings
+ip-26-0-165-59:1557271:1557726 [5] NCCL INFO Channel 19/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-168-34:272151:272611 [1] NCCL INFO Connected all rings
+ip-26-0-165-59:1557272:1557722 [6] NCCL INFO Channel 17/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-138:96567:97050 [4] NCCL INFO Connected all rings
+ip-26-0-161-178:2815967:2816438 [4] NCCL INFO Channel 19/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-138:96570:97051 [7] NCCL INFO Channel 20/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-168-34:272156:272607 [6] NCCL INFO Connected all rings
+ip-26-0-165-59:1557270:1557725 [4] NCCL INFO Channel 19/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-138:96570:97051 [7] NCCL INFO Channel 21/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-138:96570:97051 [7] NCCL INFO Channel 22/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-138:96570:97051 [7] NCCL INFO Channel 23/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-165-59:1557272:1557722 [6] NCCL INFO Channel 18/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-242:1364909:1365373 [7] NCCL INFO Connected all rings
+ip-26-0-160-242:1364909:1365373 [7] NCCL INFO Channel 00/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-138:96564:97046 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-138:96564:97046 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-168-52:2722048:2722505 [3] NCCL INFO Connected all rings
+ip-26-0-161-138:96564:97046 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-59:1557272:1557722 [6] NCCL INFO Channel 19/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-138:96564:97046 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-59:1557266:1557721 [0] NCCL INFO Connected all trees
+ip-26-0-161-138:96565:97044 [2] NCCL INFO Channel 00/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-165-59:1557266:1557721 [0] NCCL INFO NVLS comm 0xa38bfc0 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-138:96564:97046 [1] NCCL INFO Channel 04/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-138:96565:97044 [2] NCCL INFO Channel 01/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-138:96564:97046 [1] NCCL INFO Channel 05/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-138:96565:97044 [2] NCCL INFO Channel 02/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-242:1364908:1365368 [6] NCCL INFO Connected all rings
+ip-26-0-168-34:272157:272613 [7] NCCL INFO Connected all rings
+ip-26-0-168-34:272157:272613 [7] NCCL INFO Channel 00/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-138:96564:97046 [1] NCCL INFO Channel 06/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-138:96565:97044 [2] NCCL INFO Channel 03/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-165-59:1557271:1557726 [5] NCCL INFO Channel 20/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-168-34:272157:272613 [7] NCCL INFO Channel 01/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-138:96569:97049 [6] NCCL INFO Channel 00/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-138:96564:97046 [1] NCCL INFO Channel 07/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-138:96565:97044 [2] NCCL INFO Channel 04/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-165-59:1557272:1557722 [6] NCCL INFO Channel 20/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-59:1557270:1557725 [4] NCCL INFO Channel 20/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-242:1364909:1365373 [7] NCCL INFO Channel 01/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-138:96569:97049 [6] NCCL INFO Channel 01/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-138:96564:97046 [1] NCCL INFO Channel 08/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-138:96565:97044 [2] NCCL INFO Channel 05/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-168-34:272157:272613 [7] NCCL INFO Channel 02/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-138:96569:97049 [6] NCCL INFO Channel 02/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-138:96565:97044 [2] NCCL INFO Channel 06/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-168-34:272157:272613 [7] NCCL INFO Channel 03/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-142:62888:63340 [7] NCCL INFO Connected all rings
+ip-26-0-161-138:96569:97049 [6] NCCL INFO Channel 03/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-142:62888:63340 [7] NCCL INFO Channel 00/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-138:96565:97044 [2] NCCL INFO Channel 07/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-242:1364904:1365369 [2] NCCL INFO Connected all rings
+ip-26-0-161-138:96569:97049 [6] NCCL INFO Channel 04/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-168-34:272157:272613 [7] NCCL INFO Channel 04/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-138:96565:97044 [2] NCCL INFO Channel 08/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-138:96569:97049 [6] NCCL INFO Channel 05/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-178:2815967:2816438 [4] NCCL INFO Channel 20/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-138:96565:97044 [2] NCCL INFO Channel 09/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-168-34:272157:272613 [7] NCCL INFO Channel 05/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-138:96569:97049 [6] NCCL INFO Channel 06/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-59:1557271:1557726 [5] NCCL INFO Channel 21/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-138:96565:97044 [2] NCCL INFO Channel 10/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-138:96568:97048 [5] NCCL INFO Channel 00/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-138:96566:97047 [3] NCCL INFO Channel 00/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-138:96569:97049 [6] NCCL INFO Channel 07/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-168-34:272157:272613 [7] NCCL INFO Channel 06/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-138:96565:97044 [2] NCCL INFO Channel 11/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-138:96568:97048 [5] NCCL INFO Channel 01/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-138:96566:97047 [3] NCCL INFO Channel 01/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-165-59:1557272:1557722 [6] NCCL INFO Channel 21/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-59:1557270:1557725 [4] NCCL INFO Channel 21/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-138:96569:97049 [6] NCCL INFO Channel 08/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-168-52:2722052:2722506 [7] NCCL INFO Connected all rings
+ip-26-0-161-138:96565:97044 [2] NCCL INFO Channel 12/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-138:96568:97048 [5] NCCL INFO Channel 02/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-168-52:2722052:2722506 [7] NCCL INFO Channel 00/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-168-34:272157:272613 [7] NCCL INFO Channel 07/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-138:96566:97047 [3] NCCL INFO Channel 02/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-138:96569:97049 [6] NCCL INFO Channel 09/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-242:1364909:1365373 [7] NCCL INFO Channel 02/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-138:96568:97048 [5] NCCL INFO Channel 03/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-168-52:2722052:2722506 [7] NCCL INFO Channel 01/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-138:96565:97044 [2] NCCL INFO Channel 13/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-138:96566:97047 [3] NCCL INFO Channel 03/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-168-34:272157:272613 [7] NCCL INFO Channel 08/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-138:96569:97049 [6] NCCL INFO Channel 10/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-168-52:2722052:2722506 [7] NCCL INFO Channel 02/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-138:96568:97048 [5] NCCL INFO Channel 04/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-138:96565:97044 [2] NCCL INFO Channel 14/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-242:1364903:1365371 [1] NCCL INFO Connected all rings
+ip-26-0-161-138:96564:97046 [1] NCCL INFO Channel 09/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-138:96569:97049 [6] NCCL INFO Channel 11/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-168-52:2722052:2722506 [7] NCCL INFO Channel 03/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-168-34:272157:272613 [7] NCCL INFO Channel 09/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-138:96568:97048 [5] NCCL INFO Channel 05/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-138:96565:97044 [2] NCCL INFO Channel 15/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-142:62882:63343 [1] NCCL INFO Connected all rings
+ip-26-0-161-138:96564:97046 [1] NCCL INFO Channel 10/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-138:96569:97049 [6] NCCL INFO Channel 12/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-138:96568:97048 [5] NCCL INFO Channel 06/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-59:1557271:1557726 [5] NCCL INFO Channel 22/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-138:96565:97044 [2] NCCL INFO Channel 16/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-165-59:1557272:1557722 [6] NCCL INFO Channel 22/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-168-34:272157:272613 [7] NCCL INFO Channel 10/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-138:96566:97047 [3] NCCL INFO Channel 04/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-242:1364909:1365373 [7] NCCL INFO Channel 03/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-138:96564:97046 [1] NCCL INFO Channel 11/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-138:96569:97049 [6] NCCL INFO Channel 13/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-142:62888:63340 [7] NCCL INFO Channel 01/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-138:96568:97048 [5] NCCL INFO Channel 07/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-138:96565:97044 [2] NCCL INFO Channel 17/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-165-59:1557270:1557725 [4] NCCL INFO Channel 22/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-138:96566:97047 [3] NCCL INFO Channel 05/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-168-34:272157:272613 [7] NCCL INFO Channel 11/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-138:96564:97046 [1] NCCL INFO Channel 12/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-138:96569:97049 [6] NCCL INFO Channel 14/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-138:96568:97048 [5] NCCL INFO Channel 08/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-138:96565:97044 [2] NCCL INFO Channel 18/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-138:96566:97047 [3] NCCL INFO Channel 06/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-138:96564:97046 [1] NCCL INFO Channel 13/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-168-34:272157:272613 [7] NCCL INFO Channel 12/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-138:96569:97049 [6] NCCL INFO Channel 15/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-168-34:272152:272610 [2] NCCL INFO Channel 00/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-138:96568:97048 [5] NCCL INFO Channel 09/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-138:96565:97044 [2] NCCL INFO Channel 19/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-138:96566:97047 [3] NCCL INFO Channel 07/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-138:96564:97046 [1] NCCL INFO Channel 14/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-242:1364909:1365373 [7] NCCL INFO Channel 04/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-138:96569:97049 [6] NCCL INFO Channel 16/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-168-34:272157:272613 [7] NCCL INFO Channel 13/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-165-59:1557271:1557726 [5] NCCL INFO Channel 23/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-138:96565:97044 [2] NCCL INFO Channel 20/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-138:96568:97048 [5] NCCL INFO Channel 10/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-168-34:272152:272610 [2] NCCL INFO Channel 01/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-138:96566:97047 [3] NCCL INFO Channel 08/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-138:96564:97046 [1] NCCL INFO Channel 15/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-138:96567:97050 [4] NCCL INFO Channel 00/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-138:96569:97049 [6] NCCL INFO Channel 17/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-142:62885:63342 [4] NCCL INFO Connected all rings
+ip-26-0-165-59:1557272:1557722 [6] NCCL INFO Channel 23/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-138:96565:97044 [2] NCCL INFO Channel 21/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-165-59:1557270:1557725 [4] NCCL INFO Channel 23/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-138:96568:97048 [5] NCCL INFO Channel 11/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-138:96566:97047 [3] NCCL INFO Channel 09/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-138:96564:97046 [1] NCCL INFO Channel 16/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-168-34:272157:272613 [7] NCCL INFO Channel 14/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-138:96567:97050 [4] NCCL INFO Channel 01/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-168-52:2722047:2722507 [2] NCCL INFO Connected all rings
+ip-26-0-161-138:96569:97049 [6] NCCL INFO Channel 18/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-178:2815967:2816438 [4] NCCL INFO Channel 21/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-168-34:272157:272613 [7] NCCL INFO Channel 15/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-138:96565:97044 [2] NCCL INFO Channel 22/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-138:96568:97048 [5] NCCL INFO Channel 12/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-138:96566:97047 [3] NCCL INFO Channel 10/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-138:96564:97046 [1] NCCL INFO Channel 17/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-138:96567:97050 [4] NCCL INFO Channel 02/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-138:96569:97049 [6] NCCL INFO Channel 19/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-242:1364909:1365373 [7] NCCL INFO Channel 05/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-168-34:272157:272613 [7] NCCL INFO Channel 16/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-142:62888:63340 [7] NCCL INFO Channel 02/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-138:96565:97044 [2] NCCL INFO Channel 23/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-138:96568:97048 [5] NCCL INFO Channel 13/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-138:96566:97047 [3] NCCL INFO Channel 11/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-138:96564:97046 [1] NCCL INFO Channel 18/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-168-34:272157:272613 [7] NCCL INFO Channel 17/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-138:96569:97049 [6] NCCL INFO Channel 20/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-138:96567:97050 [4] NCCL INFO Channel 03/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-168-34:272157:272613 [7] NCCL INFO Channel 18/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-168-52:2722052:2722506 [7] NCCL INFO Channel 04/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-138:96568:97048 [5] NCCL INFO Channel 14/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-138:96566:97047 [3] NCCL INFO Channel 12/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-138:96564:97046 [1] NCCL INFO Channel 19/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-138:96569:97049 [6] NCCL INFO Channel 21/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-138:96567:97050 [4] NCCL INFO Channel 04/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-142:62884:63339 [3] NCCL INFO Connected all rings
+ip-26-0-168-34:272154:272608 [4] NCCL INFO Channel 00/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-242:1364909:1365373 [7] NCCL INFO Channel 06/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-138:96568:97048 [5] NCCL INFO Channel 15/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-138:96566:97047 [3] NCCL INFO Channel 13/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-138:96564:97046 [1] NCCL INFO Channel 20/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-168-34:272157:272613 [7] NCCL INFO Channel 19/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-138:96569:97049 [6] NCCL INFO Channel 22/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-138:96567:97050 [4] NCCL INFO Channel 05/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-168-52:2722051:2722503 [6] NCCL INFO Connected all rings
+ip-26-0-168-34:272152:272610 [2] NCCL INFO Channel 02/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-138:96568:97048 [5] NCCL INFO Channel 16/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-168-34:272154:272608 [4] NCCL INFO Channel 01/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-138:96566:97047 [3] NCCL INFO Channel 14/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-138:96564:97046 [1] NCCL INFO Channel 21/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-138:96569:97049 [6] NCCL INFO Channel 23/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-138:96567:97050 [4] NCCL INFO Channel 06/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-168-52:2722052:2722506 [7] NCCL INFO Channel 05/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-138:96568:97048 [5] NCCL INFO Channel 17/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-168-34:272157:272613 [7] NCCL INFO Channel 20/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-168-34:272152:272610 [2] NCCL INFO Channel 03/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-138:96566:97047 [3] NCCL INFO Channel 15/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-168-34:272154:272608 [4] NCCL INFO Channel 02/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-138:96564:97046 [1] NCCL INFO Channel 22/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-138:96567:97050 [4] NCCL INFO Channel 07/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-138:96568:97048 [5] NCCL INFO Channel 18/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-142:62883:63338 [2] NCCL INFO Connected all rings
+ip-26-0-168-52:2722052:2722506 [7] NCCL INFO Channel 06/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-242:1364905:1365372 [3] NCCL INFO Connected all rings
+ip-26-0-168-34:272157:272613 [7] NCCL INFO Channel 21/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-242:1364909:1365373 [7] NCCL INFO Channel 07/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-168-34:272152:272610 [2] NCCL INFO Channel 04/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-142:62888:63340 [7] NCCL INFO Channel 03/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-168-34:272154:272608 [4] NCCL INFO Channel 03/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-142:62888:63340 [7] NCCL INFO Channel 04/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-168-52:2722052:2722506 [7] NCCL INFO Channel 07/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-138:96566:97047 [3] NCCL INFO Channel 16/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-138:96567:97050 [4] NCCL INFO Channel 08/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-138:96564:97046 [1] NCCL INFO Channel 23/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-138:96568:97048 [5] NCCL INFO Channel 19/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-168-34:272157:272613 [7] NCCL INFO Channel 22/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-168-34:272155:272609 [5] NCCL INFO Channel 00/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-242:1364909:1365373 [7] NCCL INFO Channel 08/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-142:62888:63340 [7] NCCL INFO Channel 05/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-168-34:272152:272610 [2] NCCL INFO Channel 05/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-168-34:272154:272608 [4] NCCL INFO Channel 04/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-178:2815967:2816438 [4] NCCL INFO Channel 22/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-168-52:2722052:2722506 [7] NCCL INFO Channel 08/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-142:62888:63340 [7] NCCL INFO Channel 06/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-168-34:272151:272611 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-168-34:272157:272613 [7] NCCL INFO Channel 23/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-168-34:272152:272610 [2] NCCL INFO Channel 06/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-168-34:272155:272609 [5] NCCL INFO Channel 01/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-242:1364909:1365373 [7] NCCL INFO Channel 09/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-168-34:272154:272608 [4] NCCL INFO Channel 05/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-142:62888:63340 [7] NCCL INFO Channel 07/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-168-52:2722052:2722506 [7] NCCL INFO Channel 09/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-168-34:272151:272611 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-142:62888:63340 [7] NCCL INFO Channel 08/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-168-34:272152:272610 [2] NCCL INFO Channel 07/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-242:1364909:1365373 [7] NCCL INFO Channel 10/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-168-34:272155:272609 [5] NCCL INFO Channel 02/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-168-34:272154:272608 [4] NCCL INFO Channel 06/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-138:96566:97047 [3] NCCL INFO Channel 17/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-138:96567:97050 [4] NCCL INFO Channel 09/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-138:96568:97048 [5] NCCL INFO Channel 20/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-168-52:2722052:2722506 [7] NCCL INFO Channel 10/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-142:62888:63340 [7] NCCL INFO Channel 09/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-168-34:272151:272611 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-168-34:272152:272610 [2] NCCL INFO Channel 08/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-168-34:272155:272609 [5] NCCL INFO Channel 03/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-168-34:272154:272608 [4] NCCL INFO Channel 07/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-168-34:272153:272612 [3] NCCL INFO Channel 00/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-242:1364909:1365373 [7] NCCL INFO Channel 11/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-168-52:2722052:2722506 [7] NCCL INFO Channel 11/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-168-34:272151:272611 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-168-34:272152:272610 [2] NCCL INFO Channel 09/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-168-34:272155:272609 [5] NCCL INFO Channel 04/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-168-34:272154:272608 [4] NCCL INFO Channel 08/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-168-34:272153:272612 [3] NCCL INFO Channel 01/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-242:1364909:1365373 [7] NCCL INFO Channel 12/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-168-34:272151:272611 [1] NCCL INFO Channel 04/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-168-34:272152:272610 [2] NCCL INFO Channel 10/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-168-34:272155:272609 [5] NCCL INFO Channel 05/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-168-52:2722052:2722506 [7] NCCL INFO Channel 12/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-168-34:272154:272608 [4] NCCL INFO Channel 09/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-168-34:272153:272612 [3] NCCL INFO Channel 02/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-153:2636085:2636553 [1] NCCL INFO Connected all rings
+ip-26-0-168-34:272156:272607 [6] NCCL INFO Channel 00/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-242:1364909:1365373 [7] NCCL INFO Channel 13/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-168-34:272151:272611 [1] NCCL INFO Channel 05/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-168-34:272152:272610 [2] NCCL INFO Channel 11/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-168-34:272154:272608 [4] NCCL INFO Channel 10/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-168-34:272155:272609 [5] NCCL INFO Channel 06/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-168-52:2722052:2722506 [7] NCCL INFO Channel 13/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-168-34:272153:272612 [3] NCCL INFO Channel 03/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-138:96566:97047 [3] NCCL INFO Channel 18/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-138:96566:97047 [3] NCCL INFO Channel 19/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-168-34:272156:272607 [6] NCCL INFO Channel 01/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-168-34:272151:272611 [1] NCCL INFO Channel 06/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-242:1364909:1365373 [7] NCCL INFO Channel 14/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-168-34:272152:272610 [2] NCCL INFO Channel 12/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-138:96566:97047 [3] NCCL INFO Channel 20/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-168-34:272155:272609 [5] NCCL INFO Channel 07/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-168-34:272154:272608 [4] NCCL INFO Channel 11/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-168-52:2722052:2722506 [7] NCCL INFO Channel 14/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-168-34:272153:272612 [3] NCCL INFO Channel 04/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-138:96566:97047 [3] NCCL INFO Channel 21/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-138:96566:97047 [3] NCCL INFO Channel 22/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-138:96566:97047 [3] NCCL INFO Channel 23/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-168-34:272151:272611 [1] NCCL INFO Channel 07/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-168-34:272156:272607 [6] NCCL INFO Channel 02/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-168-34:272152:272610 [2] NCCL INFO Channel 13/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-242:1364909:1365373 [7] NCCL INFO Channel 15/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-168-34:272155:272609 [5] NCCL INFO Channel 08/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-168-34:272154:272608 [4] NCCL INFO Channel 12/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-168-34:272153:272612 [3] NCCL INFO Channel 05/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-168-52:2722052:2722506 [7] NCCL INFO Channel 15/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-168-34:272151:272611 [1] NCCL INFO Channel 08/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-168-34:272156:272607 [6] NCCL INFO Channel 03/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-168-34:272152:272610 [2] NCCL INFO Channel 14/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-168-34:272155:272609 [5] NCCL INFO Channel 09/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-168-34:272154:272608 [4] NCCL INFO Channel 13/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-168-34:272153:272612 [3] NCCL INFO Channel 06/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-242:1364909:1365373 [7] NCCL INFO Channel 16/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-168-52:2722052:2722506 [7] NCCL INFO Channel 16/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-142:62888:63340 [7] NCCL INFO Channel 10/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-168-34:272151:272611 [1] NCCL INFO Channel 09/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-168-34:272156:272607 [6] NCCL INFO Channel 04/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-168-34:272152:272610 [2] NCCL INFO Channel 15/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-168-34:272155:272609 [5] NCCL INFO Channel 10/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-168-34:272153:272612 [3] NCCL INFO Channel 07/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-168-34:272154:272608 [4] NCCL INFO Channel 14/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-242:1364909:1365373 [7] NCCL INFO Channel 17/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-168-52:2722052:2722506 [7] NCCL INFO Channel 17/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-142:62888:63340 [7] NCCL INFO Channel 11/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-168-34:272151:272611 [1] NCCL INFO Channel 10/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-168-34:272156:272607 [6] NCCL INFO Channel 05/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-168-34:272152:272610 [2] NCCL INFO Channel 16/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-168-34:272155:272609 [5] NCCL INFO Channel 11/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-168-34:272153:272612 [3] NCCL INFO Channel 08/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-168-34:272154:272608 [4] NCCL INFO Channel 15/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-153:2636087:2636550 [3] NCCL INFO Connected all rings
+ip-26-0-161-138:96567:97050 [4] NCCL INFO Channel 10/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-138:96568:97048 [5] NCCL INFO Channel 21/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-242:1364909:1365373 [7] NCCL INFO Channel 18/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-168-52:2722052:2722506 [7] NCCL INFO Channel 18/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-165-202:153563:154024 [1] NCCL INFO Connected all rings
+ip-26-0-161-142:62888:63340 [7] NCCL INFO Channel 12/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-168-34:272151:272611 [1] NCCL INFO Channel 11/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-168-34:272156:272607 [6] NCCL INFO Channel 06/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-168-34:272155:272609 [5] NCCL INFO Channel 12/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-168-34:272152:272610 [2] NCCL INFO Channel 17/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-168-34:272154:272608 [4] NCCL INFO Channel 16/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-168-34:272153:272612 [3] NCCL INFO Channel 09/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-242:1364909:1365373 [7] NCCL INFO Channel 19/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-168-52:2722052:2722506 [7] NCCL INFO Channel 19/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-168-52:2722046:2722509 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-202:153565:154022 [3] NCCL INFO Connected all rings
+ip-26-0-168-34:272151:272611 [1] NCCL INFO Channel 12/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-168-34:272156:272607 [6] NCCL INFO Channel 07/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-142:62888:63340 [7] NCCL INFO Channel 13/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-168-34:272155:272609 [5] NCCL INFO Channel 13/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-178:2815967:2816438 [4] NCCL INFO Channel 23/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-168-34:272154:272608 [4] NCCL INFO Channel 17/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-168-34:272153:272612 [3] NCCL INFO Channel 10/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-168-34:272152:272610 [2] NCCL INFO Channel 18/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-168-52:2722052:2722506 [7] NCCL INFO Channel 20/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-168-52:2722046:2722509 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-242:1364909:1365373 [7] NCCL INFO Channel 20/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-242:1364906:1365370 [4] NCCL INFO Channel 00/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-168-34:272151:272611 [1] NCCL INFO Channel 13/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-168-34:272156:272607 [6] NCCL INFO Channel 08/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-142:62888:63340 [7] NCCL INFO Channel 14/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-168-34:272155:272609 [5] NCCL INFO Channel 14/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-168-34:272154:272608 [4] NCCL INFO Channel 18/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-168-34:272153:272612 [3] NCCL INFO Channel 11/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-168-34:272152:272610 [2] NCCL INFO Channel 19/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-168-52:2722046:2722509 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-168-52:2722052:2722506 [7] NCCL INFO Channel 21/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-242:1364909:1365373 [7] NCCL INFO Channel 21/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-242:1364906:1365370 [4] NCCL INFO Channel 01/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-168-34:272151:272611 [1] NCCL INFO Channel 14/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-168-34:272156:272607 [6] NCCL INFO Channel 09/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-168-34:272154:272608 [4] NCCL INFO Channel 19/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-168-34:272155:272609 [5] NCCL INFO Channel 15/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-142:62888:63340 [7] NCCL INFO Channel 15/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-168-34:272153:272612 [3] NCCL INFO Channel 12/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-168-34:272152:272610 [2] NCCL INFO Channel 20/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-168-52:2722046:2722509 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-168-52:2722052:2722506 [7] NCCL INFO Channel 22/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-168-52:2722049:2722504 [4] NCCL INFO Channel 00/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-242:1364909:1365373 [7] NCCL INFO Channel 22/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-168-34:272151:272611 [1] NCCL INFO Channel 15/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-242:1364906:1365370 [4] NCCL INFO Channel 02/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-168-34:272154:272608 [4] NCCL INFO Channel 20/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-168-34:272156:272607 [6] NCCL INFO Channel 10/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-168-34:272153:272612 [3] NCCL INFO Channel 13/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-168-34:272155:272609 [5] NCCL INFO Channel 16/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-142:62888:63340 [7] NCCL INFO Channel 16/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-168-34:272152:272610 [2] NCCL INFO Channel 21/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-168-52:2722046:2722509 [1] NCCL INFO Channel 04/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-168-52:2722048:2722505 [3] NCCL INFO Channel 00/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-168-52:2722049:2722504 [4] NCCL INFO Channel 01/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-168-52:2722052:2722506 [7] NCCL INFO Channel 23/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-138:96567:97050 [4] NCCL INFO Channel 11/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-138:96568:97048 [5] NCCL INFO Channel 22/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-242:1364909:1365373 [7] NCCL INFO Channel 23/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-242:1364907:1365367 [5] NCCL INFO Channel 00/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-168-34:272151:272611 [1] NCCL INFO Channel 16/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-168-34:272156:272607 [6] NCCL INFO Channel 11/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-168-34:272154:272608 [4] NCCL INFO Channel 21/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-168-34:272155:272609 [5] NCCL INFO Channel 17/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-168-34:272153:272612 [3] NCCL INFO Channel 14/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-242:1364906:1365370 [4] NCCL INFO Channel 03/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-138:96567:97050 [4] NCCL INFO Channel 12/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-142:62888:63340 [7] NCCL INFO Channel 17/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-168-34:272152:272610 [2] NCCL INFO Channel 22/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-142:62887:63344 [6] NCCL INFO Channel 00/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-138:96567:97050 [4] NCCL INFO Channel 13/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-168-52:2722046:2722509 [1] NCCL INFO Channel 05/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-168-52:2722048:2722505 [3] NCCL INFO Channel 01/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-168-52:2722049:2722504 [4] NCCL INFO Channel 02/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-138:96567:97050 [4] NCCL INFO Channel 14/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-138:96567:97050 [4] NCCL INFO Channel 15/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-168-34:272151:272611 [1] NCCL INFO Channel 17/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-153:2636086:2636552 [2] NCCL INFO Connected all rings
+ip-26-0-160-242:1364907:1365367 [5] NCCL INFO Channel 01/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-168-34:272156:272607 [6] NCCL INFO Channel 12/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-168-34:272154:272608 [4] NCCL INFO Channel 22/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-168-34:272153:272612 [3] NCCL INFO Channel 15/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-168-34:272155:272609 [5] NCCL INFO Channel 18/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-142:62888:63340 [7] NCCL INFO Channel 18/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-242:1364906:1365370 [4] NCCL INFO Channel 04/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-142:62887:63344 [6] NCCL INFO Channel 01/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-168-34:272152:272610 [2] NCCL INFO Channel 23/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-168-52:2722046:2722509 [1] NCCL INFO Channel 06/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-168-52:2722048:2722505 [3] NCCL INFO Channel 02/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-168-52:2722049:2722504 [4] NCCL INFO Channel 03/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-242:1364908:1365368 [6] NCCL INFO Channel 00/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-168-30:235652:236105 [4] NCCL INFO Connected all rings
+ip-26-0-168-52:2722050:2722508 [5] NCCL INFO Channel 00/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-142:62882:63343 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-242:1364907:1365367 [5] NCCL INFO Channel 02/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-168-34:272151:272611 [1] NCCL INFO Channel 18/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-168-34:272156:272607 [6] NCCL INFO Channel 13/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-242:1364904:1365369 [2] NCCL INFO Channel 00/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-242:1364906:1365370 [4] NCCL INFO Channel 05/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-142:62888:63340 [7] NCCL INFO Channel 19/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-168-34:272154:272608 [4] NCCL INFO Channel 23/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-168-34:272153:272612 [3] NCCL INFO Channel 16/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-168-34:272155:272609 [5] NCCL INFO Channel 19/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-142:62887:63344 [6] NCCL INFO Channel 02/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-168-52:2722046:2722509 [1] NCCL INFO Channel 07/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-168-52:2722048:2722505 [3] NCCL INFO Channel 03/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-168-52:2722049:2722504 [4] NCCL INFO Channel 04/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-242:1364908:1365368 [6] NCCL INFO Channel 01/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-168-52:2722047:2722507 [2] NCCL INFO Channel 00/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-168-52:2722050:2722508 [5] NCCL INFO Channel 01/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-242:1364907:1365367 [5] NCCL INFO Channel 03/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-168-34:272151:272611 [1] NCCL INFO Channel 19/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-168-34:272156:272607 [6] NCCL INFO Channel 14/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-168-34:272153:272612 [3] NCCL INFO Channel 17/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-168-34:272155:272609 [5] NCCL INFO Channel 20/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-225:124167:124624 [7] NCCL INFO Connected all rings
+ip-26-0-160-225:124167:124624 [7] NCCL INFO Channel 00/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-168-52:2722046:2722509 [1] NCCL INFO Channel 08/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-168-52:2722048:2722505 [3] NCCL INFO Channel 04/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-168-52:2722049:2722504 [4] NCCL INFO Channel 05/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-142:62882:63343 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-168-52:2722050:2722508 [5] NCCL INFO Channel 02/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-168-52:2722047:2722507 [2] NCCL INFO Channel 01/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-168-34:272151:272611 [1] NCCL INFO Channel 20/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-168-34:272156:272607 [6] NCCL INFO Channel 15/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-168-34:272153:272612 [3] NCCL INFO Channel 18/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-168-34:272155:272609 [5] NCCL INFO Channel 21/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-142:62888:63340 [7] NCCL INFO Channel 20/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-242:1364904:1365369 [2] NCCL INFO Channel 01/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-142:62882:63343 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-242:1364906:1365370 [4] NCCL INFO Channel 06/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-242:1364903:1365371 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-168-52:2722046:2722509 [1] NCCL INFO Channel 09/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-168-52:2722048:2722505 [3] NCCL INFO Channel 05/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-168-52:2722049:2722504 [4] NCCL INFO Channel 06/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-153:2636088:2636549 [4] NCCL INFO Connected all rings
+ip-26-0-168-52:2722051:2722503 [6] NCCL INFO Channel 00/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-168-34:272156:272607 [6] NCCL INFO Channel 16/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-168-34:272151:272611 [1] NCCL INFO Channel 21/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-168-34:272153:272612 [3] NCCL INFO Channel 19/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-142:62888:63340 [7] NCCL INFO Channel 21/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-168-52:2722050:2722508 [5] NCCL INFO Channel 03/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-168-34:272155:272609 [5] NCCL INFO Channel 22/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-142:62882:63343 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-168-52:2722047:2722507 [2] NCCL INFO Channel 02/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-242:1364907:1365367 [5] NCCL INFO Channel 04/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-242:1364904:1365369 [2] NCCL INFO Channel 02/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-242:1364906:1365370 [4] NCCL INFO Channel 07/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-242:1364903:1365371 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-168-52:2722046:2722509 [1] NCCL INFO Channel 10/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-168-52:2722048:2722505 [3] NCCL INFO Channel 06/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-168-34:272153:272612 [3] NCCL INFO Channel 20/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-168-34:272156:272607 [6] NCCL INFO Channel 17/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-168-34:272151:272611 [1] NCCL INFO Channel 22/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-242:1364908:1365368 [6] NCCL INFO Channel 02/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-168-52:2722049:2722504 [4] NCCL INFO Channel 07/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-168-52:2722051:2722503 [6] NCCL INFO Channel 01/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-168-34:272155:272609 [5] NCCL INFO Channel 23/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-168-52:2722050:2722508 [5] NCCL INFO Channel 04/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-168-52:2722047:2722507 [2] NCCL INFO Channel 03/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-168-34:272153:272612 [3] NCCL INFO Channel 21/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-168-34:272156:272607 [6] NCCL INFO Channel 18/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-168-34:272151:272611 [1] NCCL INFO Channel 23/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-168-52:2722048:2722505 [3] NCCL INFO Channel 07/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-168-52:2722046:2722509 [1] NCCL INFO Channel 11/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-168-52:2722051:2722503 [6] NCCL INFO Channel 02/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-202:153562:154017 [0] NCCL INFO Connected all rings
+ip-26-0-168-52:2722049:2722504 [4] NCCL INFO Channel 08/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-168-52:2722050:2722508 [5] NCCL INFO Channel 05/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-168-34:272153:272612 [3] NCCL INFO Channel 22/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-168-52:2722047:2722507 [2] NCCL INFO Channel 04/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-168-34:272156:272607 [6] NCCL INFO Channel 19/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-242:1364907:1365367 [5] NCCL INFO Channel 05/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-242:1364906:1365370 [4] NCCL INFO Channel 08/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-242:1364908:1365368 [6] NCCL INFO Channel 03/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-168-52:2722048:2722505 [3] NCCL INFO Channel 08/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-168-52:2722046:2722509 [1] NCCL INFO Channel 12/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-168-52:2722051:2722503 [6] NCCL INFO Channel 03/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-168-52:2722049:2722504 [4] NCCL INFO Channel 09/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-142:62887:63344 [6] NCCL INFO Channel 03/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-242:1364907:1365367 [5] NCCL INFO Channel 06/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-142:62888:63340 [7] NCCL INFO Channel 22/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-168-52:2722050:2722508 [5] NCCL INFO Channel 06/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-168-52:2722047:2722507 [2] NCCL INFO Channel 05/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-242:1364906:1365370 [4] NCCL INFO Channel 09/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-142:62882:63343 [1] NCCL INFO Channel 04/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-242:1364908:1365368 [6] NCCL INFO Channel 04/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-242:1364907:1365367 [5] NCCL INFO Channel 07/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-242:1364906:1365370 [4] NCCL INFO Channel 10/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-142:62887:63344 [6] NCCL INFO Channel 04/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-142:62888:63340 [7] NCCL INFO Channel 23/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-168-34:272153:272612 [3] NCCL INFO Channel 23/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-168-34:272156:272607 [6] NCCL INFO Channel 20/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-242:1364903:1365371 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-168-52:2722048:2722505 [3] NCCL INFO Channel 09/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-168-52:2722046:2722509 [1] NCCL INFO Channel 13/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-142:62882:63343 [1] NCCL INFO Channel 05/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-168-52:2722051:2722503 [6] NCCL INFO Channel 04/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-168-52:2722049:2722504 [4] NCCL INFO Channel 10/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-142:62884:63339 [3] NCCL INFO Channel 00/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-168-52:2722050:2722508 [5] NCCL INFO Channel 07/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-168-52:2722047:2722507 [2] NCCL INFO Channel 06/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-242:1364908:1365368 [6] NCCL INFO Channel 05/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-242:1364907:1365367 [5] NCCL INFO Channel 08/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-242:1364906:1365370 [4] NCCL INFO Channel 11/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-142:62887:63344 [6] NCCL INFO Channel 05/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-242:1364903:1365371 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-142:62882:63343 [1] NCCL INFO Channel 06/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-142:62884:63339 [3] NCCL INFO Channel 01/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-168-52:2722048:2722505 [3] NCCL INFO Channel 10/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-168-52:2722046:2722509 [1] NCCL INFO Channel 14/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-168-52:2722051:2722503 [6] NCCL INFO Channel 05/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-168-52:2722049:2722504 [4] NCCL INFO Channel 11/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-168-52:2722050:2722508 [5] NCCL INFO Channel 08/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-168-52:2722047:2722507 [2] NCCL INFO Channel 07/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-142:62887:63344 [6] NCCL INFO Channel 06/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-142:62882:63343 [1] NCCL INFO Channel 07/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-142:62884:63339 [3] NCCL INFO Channel 02/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-242:1364904:1365369 [2] NCCL INFO Channel 03/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-242:1364905:1365372 [3] NCCL INFO Channel 00/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-242:1364907:1365367 [5] NCCL INFO Channel 09/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-168-52:2722048:2722505 [3] NCCL INFO Channel 11/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-168-52:2722046:2722509 [1] NCCL INFO Channel 15/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-168-52:2722051:2722503 [6] NCCL INFO Channel 06/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-168-52:2722049:2722504 [4] NCCL INFO Channel 12/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-168-52:2722050:2722508 [5] NCCL INFO Channel 09/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-153:2636089:2636551 [5] NCCL INFO Connected all rings
+ip-26-0-168-52:2722047:2722507 [2] NCCL INFO Channel 08/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-142:62887:63344 [6] NCCL INFO Channel 07/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-142:62882:63343 [1] NCCL INFO Channel 08/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-142:62884:63339 [3] NCCL INFO Channel 03/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-225:124163:124626 [3] NCCL INFO Connected all rings
+ip-26-0-168-52:2722048:2722505 [3] NCCL INFO Channel 12/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-168-52:2722046:2722509 [1] NCCL INFO Channel 16/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-242:1364908:1365368 [6] NCCL INFO Channel 06/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-168-52:2722051:2722503 [6] NCCL INFO Channel 07/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-168-52:2722049:2722504 [4] NCCL INFO Channel 13/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-168-52:2722050:2722508 [5] NCCL INFO Channel 10/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-168-52:2722047:2722507 [2] NCCL INFO Channel 09/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-142:62887:63344 [6] NCCL INFO Channel 08/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-142:62882:63343 [1] NCCL INFO Channel 09/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-142:62884:63339 [3] NCCL INFO Channel 04/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-168-34:272156:272607 [6] NCCL INFO Channel 21/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-168-52:2722048:2722505 [3] NCCL INFO Channel 13/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-168-52:2722046:2722509 [1] NCCL INFO Channel 17/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-168-52:2722049:2722504 [4] NCCL INFO Channel 14/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-168-52:2722050:2722508 [5] NCCL INFO Channel 11/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-168-52:2722051:2722503 [6] NCCL INFO Channel 08/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-242:1364904:1365369 [2] NCCL INFO Channel 04/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-242:1364906:1365370 [4] NCCL INFO Channel 12/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-142:62887:63344 [6] NCCL INFO Channel 09/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-142:62882:63343 [1] NCCL INFO Channel 10/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-168-52:2722047:2722507 [2] NCCL INFO Channel 10/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-142:62884:63339 [3] NCCL INFO Channel 05/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-142:62886:63341 [5] NCCL INFO Channel 00/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-242:1364903:1365371 [1] NCCL INFO Channel 04/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-242:1364907:1365367 [5] NCCL INFO Channel 10/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-153:2636091:2636548 [7] NCCL INFO Connected all rings
+ip-26-0-160-242:1364904:1365369 [2] NCCL INFO Channel 05/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-242:1364906:1365370 [4] NCCL INFO Channel 13/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-153:2636091:2636548 [7] NCCL INFO Channel 00/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-242:1364903:1365371 [1] NCCL INFO Channel 05/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-168-52:2722048:2722505 [3] NCCL INFO Channel 14/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-168-52:2722046:2722509 [1] NCCL INFO Channel 18/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-168-52:2722050:2722508 [5] NCCL INFO Channel 12/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-168-52:2722049:2722504 [4] NCCL INFO Channel 15/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-168-52:2722051:2722503 [6] NCCL INFO Channel 09/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-142:62887:63344 [6] NCCL INFO Channel 10/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-142:62882:63343 [1] NCCL INFO Channel 11/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-142:62883:63338 [2] NCCL INFO Channel 00/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-168-52:2722047:2722507 [2] NCCL INFO Channel 11/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-142:62884:63339 [3] NCCL INFO Channel 06/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-142:62886:63341 [5] NCCL INFO Channel 01/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-242:1364905:1365372 [3] NCCL INFO Channel 01/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-242:1364904:1365369 [2] NCCL INFO Channel 06/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-242:1364907:1365367 [5] NCCL INFO Channel 11/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-242:1364906:1365370 [4] NCCL INFO Channel 14/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-242:1364903:1365371 [1] NCCL INFO Channel 06/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-168-52:2722048:2722505 [3] NCCL INFO Channel 15/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-168-52:2722046:2722509 [1] NCCL INFO Channel 19/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-168-52:2722050:2722508 [5] NCCL INFO Channel 13/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-168-52:2722049:2722504 [4] NCCL INFO Channel 16/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-168-52:2722051:2722503 [6] NCCL INFO Channel 10/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-168-52:2722047:2722507 [2] NCCL INFO Channel 12/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-142:62882:63343 [1] NCCL INFO Channel 12/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-142:62883:63338 [2] NCCL INFO Channel 01/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-142:62887:63344 [6] NCCL INFO Channel 11/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-142:62884:63339 [3] NCCL INFO Channel 07/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-142:62886:63341 [5] NCCL INFO Channel 02/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-242:1364905:1365372 [3] NCCL INFO Channel 02/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-225:124167:124624 [7] NCCL INFO Channel 01/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-242:1364904:1365369 [2] NCCL INFO Channel 07/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-242:1364907:1365367 [5] NCCL INFO Channel 12/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-242:1364906:1365370 [4] NCCL INFO Channel 15/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-242:1364903:1365371 [1] NCCL INFO Channel 07/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-138:96567:97050 [4] NCCL INFO Channel 16/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-168-52:2722048:2722505 [3] NCCL INFO Channel 16/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-138:96568:97048 [5] NCCL INFO Channel 23/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-242:1364905:1365372 [3] NCCL INFO Channel 03/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-168-52:2722046:2722509 [1] NCCL INFO Channel 20/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-168-52:2722050:2722508 [5] NCCL INFO Channel 14/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-168-52:2722049:2722504 [4] NCCL INFO Channel 17/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-168-52:2722051:2722503 [6] NCCL INFO Channel 11/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-168-52:2722047:2722507 [2] NCCL INFO Channel 13/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-142:62882:63343 [1] NCCL INFO Channel 13/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-142:62883:63338 [2] NCCL INFO Channel 02/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-142:62884:63339 [3] NCCL INFO Channel 08/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-142:62887:63344 [6] NCCL INFO Channel 12/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-142:62886:63341 [5] NCCL INFO Channel 03/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-242:1364904:1365369 [2] NCCL INFO Channel 08/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-242:1364906:1365370 [4] NCCL INFO Channel 16/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-242:1364907:1365367 [5] NCCL INFO Channel 13/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-168-30:235653:236103 [5] NCCL INFO Connected all rings
+ip-26-0-160-242:1364903:1365371 [1] NCCL INFO Channel 08/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-242:1364905:1365372 [3] NCCL INFO Channel 04/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-168-52:2722048:2722505 [3] NCCL INFO Channel 17/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-168-52:2722050:2722508 [5] NCCL INFO Channel 15/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-168-52:2722046:2722509 [1] NCCL INFO Channel 21/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-168-52:2722051:2722503 [6] NCCL INFO Channel 12/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-168-52:2722049:2722504 [4] NCCL INFO Channel 18/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-168-52:2722047:2722507 [2] NCCL INFO Channel 14/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-142:62882:63343 [1] NCCL INFO Channel 14/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-142:62883:63338 [2] NCCL INFO Channel 03/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-142:62884:63339 [3] NCCL INFO Channel 09/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-142:62887:63344 [6] NCCL INFO Channel 13/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-242:1364904:1365369 [2] NCCL INFO Channel 09/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-242:1364903:1365371 [1] NCCL INFO Channel 09/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-242:1364906:1365370 [4] NCCL INFO Channel 17/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-142:62882:63343 [1] NCCL INFO Channel 15/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-142:62883:63338 [2] NCCL INFO Channel 04/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-242:1364907:1365367 [5] NCCL INFO Channel 14/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-142:62884:63339 [3] NCCL INFO Channel 10/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-168-52:2722046:2722509 [1] NCCL INFO Channel 22/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-168-52:2722049:2722504 [4] NCCL INFO Channel 19/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-142:62887:63344 [6] NCCL INFO Channel 14/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-242:1364903:1365371 [1] NCCL INFO Channel 10/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-242:1364904:1365369 [2] NCCL INFO Channel 10/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-242:1364906:1365370 [4] NCCL INFO Channel 18/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-242:1364907:1365367 [5] NCCL INFO Channel 15/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-168-52:2722046:2722509 [1] NCCL INFO Channel 23/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-168-52:2722047:2722507 [2] NCCL INFO Channel 15/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-142:62882:63343 [1] NCCL INFO Channel 16/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-242:1364908:1365368 [6] NCCL INFO Channel 07/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-168-52:2722049:2722504 [4] NCCL INFO Channel 20/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-142:62883:63338 [2] NCCL INFO Channel 05/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-142:62884:63339 [3] NCCL INFO Channel 11/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-153:2636091:2636548 [7] NCCL INFO Channel 01/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-142:62887:63344 [6] NCCL INFO Channel 15/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-142:62885:63342 [4] NCCL INFO Channel 00/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-168-52:2722048:2722505 [3] NCCL INFO Channel 18/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-168-52:2722050:2722508 [5] NCCL INFO Channel 16/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-242:1364903:1365371 [1] NCCL INFO Channel 11/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-242:1364904:1365369 [2] NCCL INFO Channel 11/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-242:1364906:1365370 [4] NCCL INFO Channel 19/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-242:1364907:1365367 [5] NCCL INFO Channel 16/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-142:62882:63343 [1] NCCL INFO Channel 17/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-242:1364908:1365368 [6] NCCL INFO Channel 08/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-168-52:2722047:2722507 [2] NCCL INFO Channel 16/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-142:62883:63338 [2] NCCL INFO Channel 06/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-142:62884:63339 [3] NCCL INFO Channel 12/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-168-52:2722049:2722504 [4] NCCL INFO Channel 21/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-142:62887:63344 [6] NCCL INFO Channel 16/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-168-52:2722048:2722505 [3] NCCL INFO Channel 19/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-168-52:2722050:2722508 [5] NCCL INFO Channel 17/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-142:62885:63342 [4] NCCL INFO Channel 01/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-242:1364903:1365371 [1] NCCL INFO Channel 12/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-153:2636090:2636547 [6] NCCL INFO Connected all rings
+ip-26-0-160-242:1364904:1365369 [2] NCCL INFO Channel 12/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-242:1364906:1365370 [4] NCCL INFO Channel 20/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-242:1364907:1365367 [5] NCCL INFO Channel 17/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-242:1364908:1365368 [6] NCCL INFO Channel 09/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-142:62882:63343 [1] NCCL INFO Channel 18/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-142:62883:63338 [2] NCCL INFO Channel 07/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-142:62884:63339 [3] NCCL INFO Channel 13/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-142:62887:63344 [6] NCCL INFO Channel 17/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-142:62886:63341 [5] NCCL INFO Channel 04/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-142:62885:63342 [4] NCCL INFO Channel 02/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-242:1364903:1365371 [1] NCCL INFO Channel 13/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-242:1364904:1365369 [2] NCCL INFO Channel 13/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-242:1364906:1365370 [4] NCCL INFO Channel 21/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-242:1364907:1365367 [5] NCCL INFO Channel 18/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-242:1364908:1365368 [6] NCCL INFO Channel 10/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-142:62883:63338 [2] NCCL INFO Channel 08/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-168-52:2722047:2722507 [2] NCCL INFO Channel 17/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-168-52:2722049:2722504 [4] NCCL INFO Channel 22/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-142:62882:63343 [1] NCCL INFO Channel 19/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-142:62884:63339 [3] NCCL INFO Channel 14/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-142:62887:63344 [6] NCCL INFO Channel 18/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-142:62886:63341 [5] NCCL INFO Channel 05/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-142:62885:63342 [4] NCCL INFO Channel 03/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-242:1364905:1365372 [3] NCCL INFO Channel 05/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-242:1364903:1365371 [1] NCCL INFO Channel 14/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-242:1364904:1365369 [2] NCCL INFO Channel 14/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-242:1364906:1365370 [4] NCCL INFO Channel 22/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-242:1364907:1365367 [5] NCCL INFO Channel 19/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-153:2636091:2636548 [7] NCCL INFO Channel 02/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-242:1364908:1365368 [6] NCCL INFO Channel 11/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-142:62883:63338 [2] NCCL INFO Channel 09/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-142:62882:63343 [1] NCCL INFO Channel 20/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-142:62884:63339 [3] NCCL INFO Channel 15/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-168-52:2722048:2722505 [3] NCCL INFO Channel 20/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-168-52:2722050:2722508 [5] NCCL INFO Channel 18/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-142:62887:63344 [6] NCCL INFO Channel 19/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-142:62886:63341 [5] NCCL INFO Channel 06/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-142:62885:63342 [4] NCCL INFO Channel 04/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-225:124167:124624 [7] NCCL INFO Channel 02/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-242:1364903:1365371 [1] NCCL INFO Channel 15/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-242:1364905:1365372 [3] NCCL INFO Channel 06/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-242:1364904:1365369 [2] NCCL INFO Channel 15/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-242:1364907:1365367 [5] NCCL INFO Channel 20/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-242:1364906:1365370 [4] NCCL INFO Channel 23/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-242:1364908:1365368 [6] NCCL INFO Channel 12/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-142:62883:63338 [2] NCCL INFO Channel 10/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-142:62882:63343 [1] NCCL INFO Channel 21/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-142:62884:63339 [3] NCCL INFO Channel 16/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-142:62887:63344 [6] NCCL INFO Channel 20/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-142:62886:63341 [5] NCCL INFO Channel 07/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-142:62885:63342 [4] NCCL INFO Channel 05/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-242:1364903:1365371 [1] NCCL INFO Channel 16/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-242:1364904:1365369 [2] NCCL INFO Channel 16/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-242:1364905:1365372 [3] NCCL INFO Channel 07/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-242:1364907:1365367 [5] NCCL INFO Channel 21/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-242:1364908:1365368 [6] NCCL INFO Channel 13/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-138:96567:97050 [4] NCCL INFO Channel 17/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-142:62883:63338 [2] NCCL INFO Channel 11/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-142:62882:63343 [1] NCCL INFO Channel 22/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-142:62884:63339 [3] NCCL INFO Channel 17/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-142:62887:63344 [6] NCCL INFO Channel 21/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-142:62886:63341 [5] NCCL INFO Channel 08/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-142:62885:63342 [4] NCCL INFO Channel 06/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-165-202:153564:154018 [2] NCCL INFO Connected all rings
+ip-26-0-160-242:1364903:1365371 [1] NCCL INFO Channel 17/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-242:1364904:1365369 [2] NCCL INFO Channel 17/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-242:1364905:1365372 [3] NCCL INFO Channel 08/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-242:1364907:1365367 [5] NCCL INFO Channel 22/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-242:1364908:1365368 [6] NCCL INFO Channel 14/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-142:62883:63338 [2] NCCL INFO Channel 12/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-142:62884:63339 [3] NCCL INFO Channel 18/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-142:62882:63343 [1] NCCL INFO Channel 23/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-153:2636091:2636548 [7] NCCL INFO Channel 03/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-142:62887:63344 [6] NCCL INFO Channel 22/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-142:62886:63341 [5] NCCL INFO Channel 09/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-142:62885:63342 [4] NCCL INFO Channel 07/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-242:1364903:1365371 [1] NCCL INFO Channel 18/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-242:1364904:1365369 [2] NCCL INFO Channel 18/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-242:1364907:1365367 [5] NCCL INFO Channel 23/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-242:1364905:1365372 [3] NCCL INFO Channel 09/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-242:1364908:1365368 [6] NCCL INFO Channel 15/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-142:62883:63338 [2] NCCL INFO Channel 13/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-142:62884:63339 [3] NCCL INFO Channel 19/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-142:62887:63344 [6] NCCL INFO Channel 23/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-242:1364903:1365371 [1] NCCL INFO Channel 19/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-142:62885:63342 [4] NCCL INFO Channel 08/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-142:62886:63341 [5] NCCL INFO Channel 10/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-242:1364904:1365369 [2] NCCL INFO Channel 19/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-242:1364905:1365372 [3] NCCL INFO Channel 10/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-242:1364908:1365368 [6] NCCL INFO Channel 16/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-242:1364903:1365371 [1] NCCL INFO Channel 20/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-242:1364904:1365369 [2] NCCL INFO Channel 20/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-242:1364905:1365372 [3] NCCL INFO Channel 11/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-242:1364908:1365368 [6] NCCL INFO Channel 17/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-242:1364903:1365371 [1] NCCL INFO Channel 21/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-242:1364904:1365369 [2] NCCL INFO Channel 21/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-242:1364905:1365372 [3] NCCL INFO Channel 12/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-242:1364908:1365368 [6] NCCL INFO Channel 18/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-142:62883:63338 [2] NCCL INFO Channel 14/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-142:62884:63339 [3] NCCL INFO Channel 20/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-153:2636084:2636546 [0] NCCL INFO Connected all rings
+ip-26-0-160-242:1364903:1365371 [1] NCCL INFO Channel 22/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-242:1364904:1365369 [2] NCCL INFO Channel 22/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-242:1364905:1365372 [3] NCCL INFO Channel 13/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-242:1364908:1365368 [6] NCCL INFO Channel 19/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-153:2636091:2636548 [7] NCCL INFO Channel 04/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-242:1364903:1365371 [1] NCCL INFO Channel 23/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-242:1364904:1365369 [2] NCCL INFO Channel 23/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-242:1364905:1365372 [3] NCCL INFO Channel 14/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-242:1364908:1365368 [6] NCCL INFO Channel 20/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-142:62885:63342 [4] NCCL INFO Channel 09/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-142:62886:63341 [5] NCCL INFO Channel 11/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-178:2815963:2816433 [0] NCCL INFO Connected all trees
+ip-26-0-160-242:1364905:1365372 [3] NCCL INFO Channel 15/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-242:1364908:1365368 [6] NCCL INFO Channel 21/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-153:2636091:2636548 [7] NCCL INFO Channel 05/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-178:2815963:2816433 [0] NCCL INFO NVLS comm 0x95efcc0 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-168-34:272156:272607 [6] NCCL INFO Channel 22/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-142:62883:63338 [2] NCCL INFO Channel 15/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-142:62884:63339 [3] NCCL INFO Channel 21/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-242:1364905:1365372 [3] NCCL INFO Channel 16/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-242:1364908:1365368 [6] NCCL INFO Channel 22/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-153:2636091:2636548 [7] NCCL INFO Channel 06/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-168-52:2722048:2722505 [3] NCCL INFO Channel 21/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-225:124167:124624 [7] NCCL INFO Channel 03/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-142:62886:63341 [5] NCCL INFO Channel 12/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-153:2636091:2636548 [7] NCCL INFO Channel 07/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-142:62885:63342 [4] NCCL INFO Channel 10/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-168-52:2722049:2722504 [4] NCCL INFO Channel 23/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-242:1364905:1365372 [3] NCCL INFO Channel 17/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-242:1364908:1365368 [6] NCCL INFO Channel 23/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-168-52:2722047:2722507 [2] NCCL INFO Channel 18/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-168-52:2722050:2722508 [5] NCCL INFO Channel 19/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-168-52:2722048:2722505 [3] NCCL INFO Channel 22/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-153:2636091:2636548 [7] NCCL INFO Channel 08/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-142:62883:63338 [2] NCCL INFO Channel 16/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-168-52:2722051:2722503 [6] NCCL INFO Channel 13/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-153:2636091:2636548 [7] NCCL INFO Channel 09/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-142:62886:63341 [5] NCCL INFO Channel 13/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-142:62885:63342 [4] NCCL INFO Channel 11/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-138:96567:97050 [4] NCCL INFO Channel 18/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-142:62884:63339 [3] NCCL INFO Channel 22/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-153:2636091:2636548 [7] NCCL INFO Channel 10/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-142:62883:63338 [2] NCCL INFO Channel 17/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-168-52:2722047:2722507 [2] NCCL INFO Channel 19/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-168-52:2722050:2722508 [5] NCCL INFO Channel 20/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-168-52:2722048:2722505 [3] NCCL INFO Channel 23/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-168-52:2722051:2722503 [6] NCCL INFO Channel 14/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-168-34:272156:272607 [6] NCCL INFO Channel 23/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-142:62886:63341 [5] NCCL INFO Channel 14/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-142:62885:63342 [4] NCCL INFO Channel 12/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-142:62884:63339 [3] NCCL INFO Channel 23/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-242:1364905:1365372 [3] NCCL INFO Channel 18/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-142:62883:63338 [2] NCCL INFO Channel 18/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-142:62886:63341 [5] NCCL INFO Channel 15/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-142:62885:63342 [4] NCCL INFO Channel 13/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-153:2636091:2636548 [7] NCCL INFO Channel 11/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-142:62883:63338 [2] NCCL INFO Channel 19/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-142:62886:63341 [5] NCCL INFO Channel 16/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-142:62885:63342 [4] NCCL INFO Channel 14/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-168-30:235651:236108 [3] NCCL INFO Connected all rings
+ip-26-0-161-153:2636091:2636548 [7] NCCL INFO Channel 12/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-225:124167:124624 [7] NCCL INFO Channel 04/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-142:62883:63338 [2] NCCL INFO Channel 20/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-168-52:2722050:2722508 [5] NCCL INFO Channel 21/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-168-52:2722047:2722507 [2] NCCL INFO Channel 20/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-168-52:2722051:2722503 [6] NCCL INFO Channel 15/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-153:2636091:2636548 [7] NCCL INFO Channel 13/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-142:62885:63342 [4] NCCL INFO Channel 15/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-142:62886:63341 [5] NCCL INFO Channel 17/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-168-52:2722050:2722508 [5] NCCL INFO Channel 22/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-225:124162:124627 [2] NCCL INFO Connected all rings
+ip-26-0-160-225:124167:124624 [7] NCCL INFO Channel 05/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-153:2636091:2636548 [7] NCCL INFO Channel 14/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-168-52:2722047:2722507 [2] NCCL INFO Channel 21/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-153:2636091:2636548 [7] NCCL INFO Channel 15/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-168-52:2722051:2722503 [6] NCCL INFO Channel 16/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-142:62883:63338 [2] NCCL INFO Channel 21/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-153:2636091:2636548 [7] NCCL INFO Channel 16/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-168-52:2722051:2722503 [6] NCCL INFO Channel 17/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-153:2636091:2636548 [7] NCCL INFO Channel 17/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-142:62885:63342 [4] NCCL INFO Channel 16/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-142:62886:63341 [5] NCCL INFO Channel 18/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-242:1364905:1365372 [3] NCCL INFO Channel 19/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-153:2636091:2636548 [7] NCCL INFO Channel 18/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-168-52:2722047:2722507 [2] NCCL INFO Channel 22/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-168-52:2722050:2722508 [5] NCCL INFO Channel 23/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-138:96567:97050 [4] NCCL INFO Channel 19/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-142:62886:63341 [5] NCCL INFO Channel 19/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-153:2636091:2636548 [7] NCCL INFO Channel 19/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-142:62883:63338 [2] NCCL INFO Channel 22/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-168-52:2722047:2722507 [2] NCCL INFO Channel 23/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-153:2636091:2636548 [7] NCCL INFO Channel 20/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-142:62885:63342 [4] NCCL INFO Channel 17/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-142:62886:63341 [5] NCCL INFO Channel 20/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-153:2636091:2636548 [7] NCCL INFO Channel 21/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-153:2636091:2636548 [7] NCCL INFO Channel 22/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-153:2636085:2636553 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-168-30:235655:236104 [7] NCCL INFO Connected all rings
+ip-26-0-160-225:124167:124624 [7] NCCL INFO Channel 06/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-168-30:235655:236104 [7] NCCL INFO Channel 00/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-142:62883:63338 [2] NCCL INFO Channel 23/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-153:2636086:2636552 [2] NCCL INFO Channel 00/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-153:2636087:2636550 [3] NCCL INFO Channel 00/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-153:2636088:2636549 [4] NCCL INFO Channel 00/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-142:62885:63342 [4] NCCL INFO Channel 18/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-153:2636091:2636548 [7] NCCL INFO Channel 23/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-153:2636087:2636550 [3] NCCL INFO Channel 01/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-153:2636085:2636553 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-153:2636088:2636549 [4] NCCL INFO Channel 01/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-142:62886:63341 [5] NCCL INFO Channel 21/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-153:2636089:2636551 [5] NCCL INFO Channel 00/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-138:96567:97050 [4] NCCL INFO Channel 20/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-242:1364905:1365372 [3] NCCL INFO Channel 20/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-153:2636087:2636550 [3] NCCL INFO Channel 02/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-178:2815970:2816437 [7] NCCL INFO Connected all trees
+ip-26-0-168-30:235654:236107 [6] NCCL INFO Connected all rings
+ip-26-0-161-178:2815970:2816437 [7] NCCL INFO NVLS comm 0x9335470 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-225:124167:124624 [7] NCCL INFO Channel 07/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-153:2636086:2636552 [2] NCCL INFO Channel 01/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-165-202:153566:154021 [4] NCCL INFO Connected all rings
+ip-26-0-160-225:124164:124623 [4] NCCL INFO Connected all rings
+ip-26-0-165-202:153568:154019 [6] NCCL INFO Connected all rings
+ip-26-0-168-30:235655:236104 [7] NCCL INFO Channel 01/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-225:124161:124621 [1] NCCL INFO Connected all rings
+ip-26-0-161-153:2636088:2636549 [4] NCCL INFO Channel 02/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-153:2636089:2636551 [5] NCCL INFO Channel 01/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-153:2636087:2636550 [3] NCCL INFO Channel 03/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-153:2636088:2636549 [4] NCCL INFO Channel 03/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-153:2636090:2636547 [6] NCCL INFO Channel 00/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-153:2636085:2636553 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-138:96567:97050 [4] NCCL INFO Channel 21/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-153:2636086:2636552 [2] NCCL INFO Channel 02/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-153:2636090:2636547 [6] NCCL INFO Channel 01/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-153:2636085:2636553 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-225:124167:124624 [7] NCCL INFO Channel 08/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-242:1364905:1365372 [3] NCCL INFO Channel 21/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-153:2636086:2636552 [2] NCCL INFO Channel 03/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-153:2636089:2636551 [5] NCCL INFO Channel 02/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-225:124160:124620 [0] NCCL INFO Connected all rings
+ip-26-0-161-153:2636088:2636549 [4] NCCL INFO Channel 04/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-153:2636087:2636550 [3] NCCL INFO Channel 04/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-153:2636086:2636552 [2] NCCL INFO Channel 04/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-153:2636089:2636551 [5] NCCL INFO Channel 03/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-225:124167:124624 [7] NCCL INFO Channel 09/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-153:2636088:2636549 [4] NCCL INFO Channel 05/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-153:2636087:2636550 [3] NCCL INFO Channel 05/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-153:2636086:2636552 [2] NCCL INFO Channel 05/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-153:2636089:2636551 [5] NCCL INFO Channel 04/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-153:2636088:2636549 [4] NCCL INFO Channel 06/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-153:2636087:2636550 [3] NCCL INFO Channel 06/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-142:62885:63342 [4] NCCL INFO Channel 19/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-225:124167:124624 [7] NCCL INFO Channel 10/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-153:2636086:2636552 [2] NCCL INFO Channel 06/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-153:2636089:2636551 [5] NCCL INFO Channel 05/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-202:153569:154020 [7] NCCL INFO Connected all rings
+ip-26-0-165-202:153569:154020 [7] NCCL INFO Channel 00/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-153:2636088:2636549 [4] NCCL INFO Channel 07/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-153:2636087:2636550 [3] NCCL INFO Channel 07/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-153:2636086:2636552 [2] NCCL INFO Channel 07/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-153:2636089:2636551 [5] NCCL INFO Channel 06/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-142:62886:63341 [5] NCCL INFO Channel 22/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-142:62885:63342 [4] NCCL INFO Channel 20/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-225:124167:124624 [7] NCCL INFO Channel 11/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-153:2636088:2636549 [4] NCCL INFO Channel 08/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-153:2636087:2636550 [3] NCCL INFO Channel 08/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-153:2636086:2636552 [2] NCCL INFO Channel 08/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-153:2636089:2636551 [5] NCCL INFO Channel 07/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-142:62885:63342 [4] NCCL INFO Channel 21/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-142:62885:63342 [4] NCCL INFO Channel 22/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-153:2636088:2636549 [4] NCCL INFO Channel 09/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-153:2636087:2636550 [3] NCCL INFO Channel 09/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-225:124167:124624 [7] NCCL INFO Channel 12/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-138:96567:97050 [4] NCCL INFO Channel 22/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-153:2636086:2636552 [2] NCCL INFO Channel 09/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-153:2636090:2636547 [6] NCCL INFO Channel 02/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-168-52:2722045:2722502 [0] NCCL INFO Connected all trees
+ip-26-0-161-153:2636089:2636551 [5] NCCL INFO Channel 08/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-153:2636085:2636553 [1] NCCL INFO Channel 04/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-168-52:2722045:2722502 [0] NCCL INFO NVLS comm 0x8798320 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-153:2636088:2636549 [4] NCCL INFO Channel 10/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-225:124167:124624 [7] NCCL INFO Channel 13/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-153:2636087:2636550 [3] NCCL INFO Channel 10/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-168-52:2722051:2722503 [6] NCCL INFO Channel 18/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-242:1364905:1365372 [3] NCCL INFO Channel 22/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-153:2636086:2636552 [2] NCCL INFO Channel 10/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-153:2636090:2636547 [6] NCCL INFO Channel 03/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-153:2636089:2636551 [5] NCCL INFO Channel 09/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-153:2636085:2636553 [1] NCCL INFO Channel 05/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-225:124167:124624 [7] NCCL INFO Channel 14/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-153:2636088:2636549 [4] NCCL INFO Channel 11/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-168-30:235655:236104 [7] NCCL INFO Channel 02/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-153:2636087:2636550 [3] NCCL INFO Channel 11/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-165-202:153569:154020 [7] NCCL INFO Channel 01/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-153:2636086:2636552 [2] NCCL INFO Channel 11/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-153:2636090:2636547 [6] NCCL INFO Channel 04/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-153:2636089:2636551 [5] NCCL INFO Channel 10/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-225:124167:124624 [7] NCCL INFO Channel 15/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-153:2636085:2636553 [1] NCCL INFO Channel 06/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-153:2636088:2636549 [4] NCCL INFO Channel 12/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-165-202:153569:154020 [7] NCCL INFO Channel 02/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-153:2636087:2636550 [3] NCCL INFO Channel 12/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-153:2636086:2636552 [2] NCCL INFO Channel 12/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-153:2636090:2636547 [6] NCCL INFO Channel 05/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-153:2636089:2636551 [5] NCCL INFO Channel 11/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-225:124167:124624 [7] NCCL INFO Channel 16/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-153:2636085:2636553 [1] NCCL INFO Channel 07/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-202:153569:154020 [7] NCCL INFO Channel 03/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-153:2636088:2636549 [4] NCCL INFO Channel 13/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-153:2636087:2636550 [3] NCCL INFO Channel 13/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-153:2636089:2636551 [5] NCCL INFO Channel 12/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-168-52:2722051:2722503 [6] NCCL INFO Channel 19/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-153:2636086:2636552 [2] NCCL INFO Channel 13/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-153:2636090:2636547 [6] NCCL INFO Channel 06/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-225:124167:124624 [7] NCCL INFO Channel 17/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-142:62886:63341 [5] NCCL INFO Channel 23/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-153:2636085:2636553 [1] NCCL INFO Channel 08/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-202:153569:154020 [7] NCCL INFO Channel 04/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-142:62885:63342 [4] NCCL INFO Channel 23/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-153:2636088:2636549 [4] NCCL INFO Channel 14/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-153:2636087:2636550 [3] NCCL INFO Channel 14/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-153:2636089:2636551 [5] NCCL INFO Channel 13/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-153:2636086:2636552 [2] NCCL INFO Channel 14/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-153:2636090:2636547 [6] NCCL INFO Channel 07/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-78:114514:114971 [2] NCCL INFO Connected all rings
+ip-26-0-160-225:124167:124624 [7] NCCL INFO Channel 18/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-165-202:153569:154020 [7] NCCL INFO Channel 05/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-153:2636085:2636553 [1] NCCL INFO Channel 09/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-153:2636088:2636549 [4] NCCL INFO Channel 15/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-153:2636087:2636550 [3] NCCL INFO Channel 15/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-153:2636089:2636551 [5] NCCL INFO Channel 14/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-153:2636086:2636552 [2] NCCL INFO Channel 15/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-153:2636090:2636547 [6] NCCL INFO Channel 08/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-225:124165:124625 [5] NCCL INFO Channel 00/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-225:124167:124624 [7] NCCL INFO Channel 19/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-165-202:153569:154020 [7] NCCL INFO Channel 06/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-153:2636085:2636553 [1] NCCL INFO Channel 10/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-153:2636088:2636549 [4] NCCL INFO Channel 16/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-153:2636087:2636550 [3] NCCL INFO Channel 16/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-168-30:235650:236106 [2] NCCL INFO Connected all rings
+ip-26-0-161-153:2636089:2636551 [5] NCCL INFO Channel 15/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-153:2636086:2636552 [2] NCCL INFO Channel 16/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-153:2636090:2636547 [6] NCCL INFO Channel 09/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-225:124165:124625 [5] NCCL INFO Channel 01/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-225:124167:124624 [7] NCCL INFO Channel 20/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-165-202:153569:154020 [7] NCCL INFO Channel 07/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-153:2636085:2636553 [1] NCCL INFO Channel 11/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-153:2636088:2636549 [4] NCCL INFO Channel 17/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-153:2636087:2636550 [3] NCCL INFO Channel 17/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-153:2636089:2636551 [5] NCCL INFO Channel 16/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-153:2636086:2636552 [2] NCCL INFO Channel 17/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-153:2636090:2636547 [6] NCCL INFO Channel 10/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-138:96567:97050 [4] NCCL INFO Channel 23/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-165-202:153569:154020 [7] NCCL INFO Channel 08/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-225:124165:124625 [5] NCCL INFO Channel 02/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-225:124167:124624 [7] NCCL INFO Channel 21/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-153:2636085:2636553 [1] NCCL INFO Channel 12/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-153:2636088:2636549 [4] NCCL INFO Channel 18/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-153:2636087:2636550 [3] NCCL INFO Channel 18/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-168-30:235648:236102 [0] NCCL INFO Connected all rings
+ip-26-0-160-225:124166:124622 [6] NCCL INFO Channel 00/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-153:2636089:2636551 [5] NCCL INFO Channel 17/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-153:2636086:2636552 [2] NCCL INFO Channel 18/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-153:2636090:2636547 [6] NCCL INFO Channel 11/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-202:153569:154020 [7] NCCL INFO Channel 09/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-138:96570:97051 [7] NCCL INFO Connected all trees
+ip-26-0-160-225:124165:124625 [5] NCCL INFO Channel 03/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-225:124167:124624 [7] NCCL INFO Channel 22/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-242:1364905:1365372 [3] NCCL INFO Channel 23/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-153:2636085:2636553 [1] NCCL INFO Channel 13/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-153:2636088:2636549 [4] NCCL INFO Channel 19/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-153:2636087:2636550 [3] NCCL INFO Channel 19/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-138:96570:97051 [7] NCCL INFO NVLS comm 0xa334e80 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-168-30:235655:236104 [7] NCCL INFO Channel 03/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-153:2636089:2636551 [5] NCCL INFO Channel 18/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-225:124166:124622 [6] NCCL INFO Channel 01/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-153:2636086:2636552 [2] NCCL INFO Channel 19/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-153:2636090:2636547 [6] NCCL INFO Channel 12/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-202:153569:154020 [7] NCCL INFO Channel 10/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-225:124167:124624 [7] NCCL INFO Channel 23/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-225:124165:124625 [5] NCCL INFO Channel 04/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-153:2636085:2636553 [1] NCCL INFO Channel 14/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-153:2636088:2636549 [4] NCCL INFO Channel 20/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-153:2636087:2636550 [3] NCCL INFO Channel 20/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-153:2636089:2636551 [5] NCCL INFO Channel 19/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-153:2636086:2636552 [2] NCCL INFO Channel 20/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-225:124166:124622 [6] NCCL INFO Channel 02/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-202:153569:154020 [7] NCCL INFO Channel 11/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-153:2636090:2636547 [6] NCCL INFO Channel 13/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-168-52:2722051:2722503 [6] NCCL INFO Channel 20/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-153:2636085:2636553 [1] NCCL INFO Channel 15/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-153:2636088:2636549 [4] NCCL INFO Channel 21/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-225:124165:124625 [5] NCCL INFO Channel 05/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-153:2636087:2636550 [3] NCCL INFO Channel 21/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-165-202:153569:154020 [7] NCCL INFO Channel 12/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-153:2636089:2636551 [5] NCCL INFO Channel 20/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-153:2636086:2636552 [2] NCCL INFO Channel 21/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-225:124166:124622 [6] NCCL INFO Channel 03/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-153:2636090:2636547 [6] NCCL INFO Channel 14/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-168-30:235649:236109 [1] NCCL INFO Connected all rings
+ip-26-0-160-225:124163:124626 [3] NCCL INFO Channel 00/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-225:124165:124625 [5] NCCL INFO Channel 06/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-168-30:235655:236104 [7] NCCL INFO Channel 04/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-153:2636085:2636553 [1] NCCL INFO Channel 16/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-153:2636088:2636549 [4] NCCL INFO Channel 22/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-153:2636087:2636550 [3] NCCL INFO Channel 22/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-165-202:153569:154020 [7] NCCL INFO Channel 13/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-225:124166:124622 [6] NCCL INFO Channel 04/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-153:2636089:2636551 [5] NCCL INFO Channel 21/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-153:2636086:2636552 [2] NCCL INFO Channel 22/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-153:2636090:2636547 [6] NCCL INFO Channel 15/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-225:124165:124625 [5] NCCL INFO Channel 07/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-225:124163:124626 [3] NCCL INFO Channel 01/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-168-30:235655:236104 [7] NCCL INFO Channel 05/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-153:2636085:2636553 [1] NCCL INFO Channel 17/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-153:2636088:2636549 [4] NCCL INFO Channel 23/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-153:2636087:2636550 [3] NCCL INFO Channel 23/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-165-202:153569:154020 [7] NCCL INFO Channel 14/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-225:124166:124622 [6] NCCL INFO Channel 05/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-202:153567:154023 [5] NCCL INFO Channel 00/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-153:2636089:2636551 [5] NCCL INFO Channel 22/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-153:2636086:2636552 [2] NCCL INFO Channel 23/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-153:2636090:2636547 [6] NCCL INFO Channel 16/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-225:124165:124625 [5] NCCL INFO Channel 08/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-225:124163:124626 [3] NCCL INFO Channel 02/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-168-30:235655:236104 [7] NCCL INFO Channel 06/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-165-202:153565:154022 [3] NCCL INFO Channel 00/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-153:2636085:2636553 [1] NCCL INFO Channel 18/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-225:124166:124622 [6] NCCL INFO Channel 06/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-202:153569:154020 [7] NCCL INFO Channel 15/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-153:2636089:2636551 [5] NCCL INFO Channel 23/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-153:2636090:2636547 [6] NCCL INFO Channel 17/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-202:153567:154023 [5] NCCL INFO Channel 01/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-153:2636085:2636553 [1] NCCL INFO Channel 19/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-225:124165:124625 [5] NCCL INFO Channel 09/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-168-30:235655:236104 [7] NCCL INFO Channel 07/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-225:124163:124626 [3] NCCL INFO Channel 03/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-165-202:153565:154022 [3] NCCL INFO Channel 01/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-168-52:2722051:2722503 [6] NCCL INFO Channel 21/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-123:32711:33165 [7] NCCL INFO Connected all trees
+ip-26-0-160-225:124166:124622 [6] NCCL INFO Channel 07/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-225:124162:124627 [2] NCCL INFO Channel 00/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-165-202:153569:154020 [7] NCCL INFO Channel 16/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-123:32711:33165 [7] NCCL INFO NVLS comm 0xa0ea330 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-202:153567:154023 [5] NCCL INFO Channel 02/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-225:124165:124625 [5] NCCL INFO Channel 10/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-168-30:235655:236104 [7] NCCL INFO Channel 08/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-225:124163:124626 [3] NCCL INFO Channel 04/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-165-202:153565:154022 [3] NCCL INFO Channel 02/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-153:2636090:2636547 [6] NCCL INFO Channel 18/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-153:2636085:2636553 [1] NCCL INFO Channel 20/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-202:153563:154024 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-225:124166:124622 [6] NCCL INFO Channel 08/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-202:153569:154020 [7] NCCL INFO Channel 17/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-225:124162:124627 [2] NCCL INFO Channel 01/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-165-202:153567:154023 [5] NCCL INFO Channel 03/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-225:124165:124625 [5] NCCL INFO Channel 11/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-225:124163:124626 [3] NCCL INFO Channel 05/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-168-30:235655:236104 [7] NCCL INFO Channel 09/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-165-202:153565:154022 [3] NCCL INFO Channel 03/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-165-202:153563:154024 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-202:153569:154020 [7] NCCL INFO Channel 18/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-225:124166:124622 [6] NCCL INFO Channel 09/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-225:124162:124627 [2] NCCL INFO Channel 02/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-165-202:153567:154023 [5] NCCL INFO Channel 04/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-202:153566:154021 [4] NCCL INFO Channel 00/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-168-30:235655:236104 [7] NCCL INFO Channel 10/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-225:124165:124625 [5] NCCL INFO Channel 12/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-225:124163:124626 [3] NCCL INFO Channel 06/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-225:124164:124623 [4] NCCL INFO Channel 00/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-225:124161:124621 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-202:153563:154024 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-202:153569:154020 [7] NCCL INFO Channel 19/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-225:124166:124622 [6] NCCL INFO Channel 10/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-202:153567:154023 [5] NCCL INFO Channel 05/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-202:153568:154019 [6] NCCL INFO Channel 00/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-202:153564:154018 [2] NCCL INFO Channel 00/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-225:124162:124627 [2] NCCL INFO Channel 03/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-168-30:235655:236104 [7] NCCL INFO Channel 11/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-165-202:153566:154021 [4] NCCL INFO Channel 01/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-225:124163:124626 [3] NCCL INFO Channel 07/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-225:124165:124625 [5] NCCL INFO Channel 13/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-202:153563:154024 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-225:124164:124623 [4] NCCL INFO Channel 01/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-165-202:153569:154020 [7] NCCL INFO Channel 20/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-225:124161:124621 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-202:153567:154023 [5] NCCL INFO Channel 06/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-153:2636085:2636553 [1] NCCL INFO Channel 21/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-202:153568:154019 [6] NCCL INFO Channel 01/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-153:2636090:2636547 [6] NCCL INFO Channel 19/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-202:153564:154018 [2] NCCL INFO Channel 01/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-165-202:153566:154021 [4] NCCL INFO Channel 02/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-168-30:235655:236104 [7] NCCL INFO Channel 12/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-225:124166:124622 [6] NCCL INFO Channel 11/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-225:124162:124627 [2] NCCL INFO Channel 04/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-225:124163:124626 [3] NCCL INFO Channel 08/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-165-202:153563:154024 [1] NCCL INFO Channel 04/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-225:124165:124625 [5] NCCL INFO Channel 14/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-202:153569:154020 [7] NCCL INFO Channel 21/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-225:124164:124623 [4] NCCL INFO Channel 02/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-225:124161:124621 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-202:153567:154023 [5] NCCL INFO Channel 07/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-202:153568:154019 [6] NCCL INFO Channel 02/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-202:153564:154018 [2] NCCL INFO Channel 02/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-165-202:153566:154021 [4] NCCL INFO Channel 03/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-225:124166:124622 [6] NCCL INFO Channel 12/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-168-30:235655:236104 [7] NCCL INFO Channel 13/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-165-202:153563:154024 [1] NCCL INFO Channel 05/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-225:124162:124627 [2] NCCL INFO Channel 05/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-225:124163:124626 [3] NCCL INFO Channel 09/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-165-202:153569:154020 [7] NCCL INFO Channel 22/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-225:124165:124625 [5] NCCL INFO Channel 15/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-225:124164:124623 [4] NCCL INFO Channel 03/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-225:124161:124621 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-202:153564:154018 [2] NCCL INFO Channel 03/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-165-202:153565:154022 [3] NCCL INFO Channel 04/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-165-202:153567:154023 [5] NCCL INFO Channel 08/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-202:153568:154019 [6] NCCL INFO Channel 03/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-225:124166:124622 [6] NCCL INFO Channel 13/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-202:153566:154021 [4] NCCL INFO Channel 04/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-168-30:235655:236104 [7] NCCL INFO Channel 14/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-165-202:153563:154024 [1] NCCL INFO Channel 06/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-225:124163:124626 [3] NCCL INFO Channel 10/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-225:124162:124627 [2] NCCL INFO Channel 06/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-225:124165:124625 [5] NCCL INFO Channel 16/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-202:153569:154020 [7] NCCL INFO Channel 23/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-225:124164:124623 [4] NCCL INFO Channel 04/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-225:124161:124621 [1] NCCL INFO Channel 04/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-202:153565:154022 [3] NCCL INFO Channel 05/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-165-202:153564:154018 [2] NCCL INFO Channel 04/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-225:124166:124622 [6] NCCL INFO Channel 14/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-202:153567:154023 [5] NCCL INFO Channel 09/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-202:153568:154019 [6] NCCL INFO Channel 04/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-202:153563:154024 [1] NCCL INFO Channel 07/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-168-30:235655:236104 [7] NCCL INFO Channel 15/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-165-202:153566:154021 [4] NCCL INFO Channel 05/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-225:124163:124626 [3] NCCL INFO Channel 11/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-225:124162:124627 [2] NCCL INFO Channel 07/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-225:124165:124625 [5] NCCL INFO Channel 17/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-168-52:2722051:2722503 [6] NCCL INFO Channel 22/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-225:124164:124623 [4] NCCL INFO Channel 05/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-225:124161:124621 [1] NCCL INFO Channel 05/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-202:153565:154022 [3] NCCL INFO Channel 06/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-165-202:153564:154018 [2] NCCL INFO Channel 05/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-165-202:153567:154023 [5] NCCL INFO Channel 10/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-202:153563:154024 [1] NCCL INFO Channel 08/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-202:153568:154019 [6] NCCL INFO Channel 05/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-225:124166:124622 [6] NCCL INFO Channel 15/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-202:153566:154021 [4] NCCL INFO Channel 06/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-168-30:235655:236104 [7] NCCL INFO Channel 16/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-225:124162:124627 [2] NCCL INFO Channel 08/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-225:124163:124626 [3] NCCL INFO Channel 12/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-225:124165:124625 [5] NCCL INFO Channel 18/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-225:124164:124623 [4] NCCL INFO Channel 06/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-153:2636085:2636553 [1] NCCL INFO Channel 22/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-153:2636090:2636547 [6] NCCL INFO Channel 20/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-225:124161:124621 [1] NCCL INFO Channel 06/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-202:153565:154022 [3] NCCL INFO Channel 07/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-165-202:153564:154018 [2] NCCL INFO Channel 06/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-165-202:153567:154023 [5] NCCL INFO Channel 11/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-202:153568:154019 [6] NCCL INFO Channel 06/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-202:153563:154024 [1] NCCL INFO Channel 09/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-225:124166:124622 [6] NCCL INFO Channel 16/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-202:153566:154021 [4] NCCL INFO Channel 07/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-168-30:235655:236104 [7] NCCL INFO Channel 17/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-225:124162:124627 [2] NCCL INFO Channel 09/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-225:124163:124626 [3] NCCL INFO Channel 13/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-225:124165:124625 [5] NCCL INFO Channel 19/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-225:124164:124623 [4] NCCL INFO Channel 07/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-225:124161:124621 [1] NCCL INFO Channel 07/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-202:153565:154022 [3] NCCL INFO Channel 08/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-165-202:153564:154018 [2] NCCL INFO Channel 07/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-165-202:153567:154023 [5] NCCL INFO Channel 12/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-202:153568:154019 [6] NCCL INFO Channel 07/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-202:153563:154024 [1] NCCL INFO Channel 10/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-225:124166:124622 [6] NCCL INFO Channel 17/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-202:153566:154021 [4] NCCL INFO Channel 08/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-168-30:235655:236104 [7] NCCL INFO Channel 18/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-225:124162:124627 [2] NCCL INFO Channel 10/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-225:124163:124626 [3] NCCL INFO Channel 14/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-225:124165:124625 [5] NCCL INFO Channel 20/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-225:124164:124623 [4] NCCL INFO Channel 08/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-225:124161:124621 [1] NCCL INFO Channel 08/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-202:153564:154018 [2] NCCL INFO Channel 08/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-165-202:153565:154022 [3] NCCL INFO Channel 09/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-165-202:153567:154023 [5] NCCL INFO Channel 13/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-202:153568:154019 [6] NCCL INFO Channel 08/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-202:153563:154024 [1] NCCL INFO Channel 11/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-202:153566:154021 [4] NCCL INFO Channel 09/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-225:124166:124622 [6] NCCL INFO Channel 18/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-168-30:235655:236104 [7] NCCL INFO Channel 19/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-225:124162:124627 [2] NCCL INFO Channel 11/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-225:124163:124626 [3] NCCL INFO Channel 15/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-225:124165:124625 [5] NCCL INFO Channel 21/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-78:114518:114974 [6] NCCL INFO Connected all rings
+ip-26-0-160-225:124164:124623 [4] NCCL INFO Channel 09/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-225:124161:124621 [1] NCCL INFO Channel 09/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-202:153564:154018 [2] NCCL INFO Channel 09/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-165-202:153565:154022 [3] NCCL INFO Channel 10/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-165-202:153567:154023 [5] NCCL INFO Channel 14/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-202:153568:154019 [6] NCCL INFO Channel 09/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-202:153563:154024 [1] NCCL INFO Channel 12/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-202:153566:154021 [4] NCCL INFO Channel 10/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-225:124166:124622 [6] NCCL INFO Channel 19/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-168-30:235655:236104 [7] NCCL INFO Channel 20/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-225:124162:124627 [2] NCCL INFO Channel 12/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-225:124163:124626 [3] NCCL INFO Channel 16/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-225:124164:124623 [4] NCCL INFO Channel 10/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-225:124165:124625 [5] NCCL INFO Channel 22/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-168-30:235652:236105 [4] NCCL INFO Channel 00/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-225:124161:124621 [1] NCCL INFO Channel 10/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-202:153564:154018 [2] NCCL INFO Channel 10/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-165-202:153565:154022 [3] NCCL INFO Channel 11/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-165-202:153563:154024 [1] NCCL INFO Channel 13/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-202:153568:154019 [6] NCCL INFO Channel 10/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-202:153567:154023 [5] NCCL INFO Channel 15/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-202:153566:154021 [4] NCCL INFO Channel 11/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-168-30:235653:236103 [5] NCCL INFO Channel 00/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-225:124166:124622 [6] NCCL INFO Channel 20/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-153:2636085:2636553 [1] NCCL INFO Channel 23/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-153:2636090:2636547 [6] NCCL INFO Channel 21/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-225:124162:124627 [2] NCCL INFO Channel 13/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-225:124164:124623 [4] NCCL INFO Channel 11/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-168-30:235655:236104 [7] NCCL INFO Channel 21/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-225:124165:124625 [5] NCCL INFO Channel 23/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-225:124163:124626 [3] NCCL INFO Channel 17/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-168-30:235652:236105 [4] NCCL INFO Channel 01/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-225:124161:124621 [1] NCCL INFO Channel 11/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-202:153565:154022 [3] NCCL INFO Channel 12/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-165-202:153564:154018 [2] NCCL INFO Channel 11/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-165-202:153567:154023 [5] NCCL INFO Channel 16/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-202:153568:154019 [6] NCCL INFO Channel 11/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-202:153563:154024 [1] NCCL INFO Channel 14/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-202:153566:154021 [4] NCCL INFO Channel 12/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-168-30:235653:236103 [5] NCCL INFO Channel 01/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-225:124166:124622 [6] NCCL INFO Channel 21/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-225:124162:124627 [2] NCCL INFO Channel 14/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-225:124164:124623 [4] NCCL INFO Channel 12/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-225:124163:124626 [3] NCCL INFO Channel 18/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-168-30:235655:236104 [7] NCCL INFO Channel 22/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-225:124161:124621 [1] NCCL INFO Channel 12/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-168-30:235652:236105 [4] NCCL INFO Channel 02/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-165-202:153564:154018 [2] NCCL INFO Channel 12/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-165-202:153565:154022 [3] NCCL INFO Channel 13/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-165-202:153568:154019 [6] NCCL INFO Channel 12/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-225:124166:124622 [6] NCCL INFO Channel 22/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-202:153567:154023 [5] NCCL INFO Channel 17/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-202:153563:154024 [1] NCCL INFO Channel 15/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-202:153566:154021 [4] NCCL INFO Channel 13/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-168-30:235651:236108 [3] NCCL INFO Channel 00/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-168-30:235653:236103 [5] NCCL INFO Channel 02/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-225:124162:124627 [2] NCCL INFO Channel 15/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-225:124164:124623 [4] NCCL INFO Channel 13/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-225:124163:124626 [3] NCCL INFO Channel 19/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-225:124161:124621 [1] NCCL INFO Channel 13/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-225:124166:124622 [6] NCCL INFO Channel 23/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-202:153564:154018 [2] NCCL INFO Channel 13/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-168-30:235655:236104 [7] NCCL INFO Channel 23/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-165-202:153565:154022 [3] NCCL INFO Channel 14/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-165-202:153567:154023 [5] NCCL INFO Channel 18/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-202:153568:154019 [6] NCCL INFO Channel 13/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-202:153563:154024 [1] NCCL INFO Channel 16/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-202:153566:154021 [4] NCCL INFO Channel 14/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-168-30:235652:236105 [4] NCCL INFO Channel 03/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-168-30:235651:236108 [3] NCCL INFO Channel 01/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-225:124162:124627 [2] NCCL INFO Channel 16/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-225:124164:124623 [4] NCCL INFO Channel 14/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-225:124163:124626 [3] NCCL INFO Channel 20/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-168-30:235653:236103 [5] NCCL INFO Channel 03/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-225:124161:124621 [1] NCCL INFO Channel 14/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-168-52:2722051:2722503 [6] NCCL INFO Channel 23/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-202:153565:154022 [3] NCCL INFO Channel 15/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-165-202:153564:154018 [2] NCCL INFO Channel 14/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-165-202:153567:154023 [5] NCCL INFO Channel 19/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-202:153568:154019 [6] NCCL INFO Channel 14/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-202:153563:154024 [1] NCCL INFO Channel 17/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-202:153566:154021 [4] NCCL INFO Channel 15/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-165-59:1557267:1557727 [1] NCCL INFO Connected all trees
+ip-26-0-161-153:2636090:2636547 [6] NCCL INFO Channel 22/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-225:124162:124627 [2] NCCL INFO Channel 17/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-225:124163:124626 [3] NCCL INFO Channel 21/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-168-30:235652:236105 [4] NCCL INFO Channel 04/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-168-30:235651:236108 [3] NCCL INFO Channel 02/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-168-30:235653:236103 [5] NCCL INFO Channel 04/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-59:1557267:1557727 [1] NCCL INFO NVLS comm 0x8aff890 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-78:114519:114970 [7] NCCL INFO Connected all rings
+ip-26-0-165-202:153565:154022 [3] NCCL INFO Channel 16/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-165-202:153567:154023 [5] NCCL INFO Channel 20/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-78:114519:114970 [7] NCCL INFO Channel 00/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-165-202:153564:154018 [2] NCCL INFO Channel 15/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-165-202:153563:154024 [1] NCCL INFO Channel 18/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-202:153568:154019 [6] NCCL INFO Channel 15/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-202:153566:154021 [4] NCCL INFO Channel 16/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-168-30:235654:236107 [6] NCCL INFO Channel 00/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-225:124164:124623 [4] NCCL INFO Channel 15/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-168-30:235652:236105 [4] NCCL INFO Channel 05/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-168-30:235651:236108 [3] NCCL INFO Channel 03/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-123:32704:33159 [0] NCCL INFO Connected all trees
+ip-26-0-168-30:235653:236103 [5] NCCL INFO Channel 05/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-123:32704:33159 [0] NCCL INFO NVLS comm 0xa6c2260 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-202:153565:154022 [3] NCCL INFO Channel 17/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-165-202:153567:154023 [5] NCCL INFO Channel 21/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-202:153564:154018 [2] NCCL INFO Channel 16/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-165-202:153563:154024 [1] NCCL INFO Channel 19/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-202:153568:154019 [6] NCCL INFO Channel 16/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-202:153566:154021 [4] NCCL INFO Channel 17/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-225:124161:124621 [1] NCCL INFO Channel 15/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-168-30:235650:236106 [2] NCCL INFO Channel 00/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-168-30:235651:236108 [3] NCCL INFO Channel 04/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-165-202:153565:154022 [3] NCCL INFO Channel 18/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-165-202:153564:154018 [2] NCCL INFO Channel 17/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-168-30:235653:236103 [5] NCCL INFO Channel 06/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-202:153567:154023 [5] NCCL INFO Channel 22/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-202:153563:154024 [1] NCCL INFO Channel 20/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-202:153568:154019 [6] NCCL INFO Channel 17/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-202:153566:154021 [4] NCCL INFO Channel 18/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-168-30:235652:236105 [4] NCCL INFO Channel 06/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-168-30:235650:236106 [2] NCCL INFO Channel 01/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-168-30:235651:236108 [3] NCCL INFO Channel 05/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-168-30:235653:236103 [5] NCCL INFO Channel 07/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-168-30:235652:236105 [4] NCCL INFO Channel 07/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-165-202:153565:154022 [3] NCCL INFO Channel 19/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-165-202:153564:154018 [2] NCCL INFO Channel 18/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-165-202:153567:154023 [5] NCCL INFO Channel 23/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-202:153563:154024 [1] NCCL INFO Channel 21/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-202:153568:154019 [6] NCCL INFO Channel 18/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-202:153566:154021 [4] NCCL INFO Channel 19/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-168-30:235650:236106 [2] NCCL INFO Channel 02/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-168-30:235651:236108 [3] NCCL INFO Channel 06/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-168-30:235653:236103 [5] NCCL INFO Channel 08/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-168-30:235654:236107 [6] NCCL INFO Channel 01/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-168-30:235652:236105 [4] NCCL INFO Channel 08/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-165-202:153565:154022 [3] NCCL INFO Channel 20/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-165-202:153563:154024 [1] NCCL INFO Channel 22/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-202:153564:154018 [2] NCCL INFO Channel 19/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-165-202:153568:154019 [6] NCCL INFO Channel 19/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-202:153566:154021 [4] NCCL INFO Channel 20/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-168-30:235650:236106 [2] NCCL INFO Channel 03/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-225:124162:124627 [2] NCCL INFO Channel 18/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-225:124163:124626 [3] NCCL INFO Channel 22/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-168-30:235651:236108 [3] NCCL INFO Channel 07/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-168-30:235653:236103 [5] NCCL INFO Channel 09/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-168-30:235654:236107 [6] NCCL INFO Channel 02/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-225:124164:124623 [4] NCCL INFO Channel 16/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-168-30:235652:236105 [4] NCCL INFO Channel 09/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-168-30:235650:236106 [2] NCCL INFO Channel 04/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-168-30:235654:236107 [6] NCCL INFO Channel 03/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-168-30:235651:236108 [3] NCCL INFO Channel 08/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-168-30:235653:236103 [5] NCCL INFO Channel 10/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-168-30:235652:236105 [4] NCCL INFO Channel 10/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-225:124161:124621 [1] NCCL INFO Channel 16/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-153:2636090:2636547 [6] NCCL INFO Channel 23/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-168-30:235650:236106 [2] NCCL INFO Channel 05/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-168-30:235649:236109 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-168-30:235654:236107 [6] NCCL INFO Channel 04/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-168-30:235653:236103 [5] NCCL INFO Channel 11/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-168-30:235651:236108 [3] NCCL INFO Channel 09/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-168-30:235652:236105 [4] NCCL INFO Channel 11/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-225:124163:124626 [3] NCCL INFO Channel 23/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-168-30:235650:236106 [2] NCCL INFO Channel 06/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-168-30:235653:236103 [5] NCCL INFO Channel 12/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-168-30:235654:236107 [6] NCCL INFO Channel 05/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-168-30:235651:236108 [3] NCCL INFO Channel 10/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-168-30:235652:236105 [4] NCCL INFO Channel 12/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-165-202:153568:154019 [6] NCCL INFO Channel 20/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-168-30:235649:236109 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-202:153566:154021 [4] NCCL INFO Channel 21/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-165-202:153565:154022 [3] NCCL INFO Channel 21/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-168-30:235650:236106 [2] NCCL INFO Channel 07/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-168-30:235654:236107 [6] NCCL INFO Channel 06/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-202:153564:154018 [2] NCCL INFO Channel 20/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-168-30:235651:236108 [3] NCCL INFO Channel 11/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-168-30:235653:236103 [5] NCCL INFO Channel 13/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-168-30:235652:236105 [4] NCCL INFO Channel 13/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-165-202:153568:154019 [6] NCCL INFO Channel 21/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-202:153566:154021 [4] NCCL INFO Channel 22/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-168-30:235649:236109 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-202:153565:154022 [3] NCCL INFO Channel 22/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-165-202:153564:154018 [2] NCCL INFO Channel 21/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-103:96892:97345 [4] NCCL INFO Connected all rings
+ip-26-0-165-202:153568:154019 [6] NCCL INFO Channel 22/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-202:153566:154021 [4] NCCL INFO Channel 23/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-168-30:235650:236106 [2] NCCL INFO Channel 08/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-168-30:235654:236107 [6] NCCL INFO Channel 07/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-168-30:235651:236108 [3] NCCL INFO Channel 12/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-168-30:235653:236103 [5] NCCL INFO Channel 14/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-168-30:235652:236105 [4] NCCL INFO Channel 14/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-165-202:153563:154024 [1] NCCL INFO Channel 23/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-168-30:235649:236109 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-202:153565:154022 [3] NCCL INFO Channel 23/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-165-202:153564:154018 [2] NCCL INFO Channel 22/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-165-202:153568:154019 [6] NCCL INFO Channel 23/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-168-30:235650:236106 [2] NCCL INFO Channel 09/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-165-202:153564:154018 [2] NCCL INFO Channel 23/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-168-30:235654:236107 [6] NCCL INFO Channel 08/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-168-30:235651:236108 [3] NCCL INFO Channel 13/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-168-30:235652:236105 [4] NCCL INFO Channel 15/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-168-30:235653:236103 [5] NCCL INFO Channel 15/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-78:114516:114975 [4] NCCL INFO Connected all rings
+ip-26-0-168-30:235649:236109 [1] NCCL INFO Channel 04/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-168-30:235650:236106 [2] NCCL INFO Channel 10/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-168-30:235653:236103 [5] NCCL INFO Channel 16/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-168-30:235651:236108 [3] NCCL INFO Channel 14/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-168-30:235652:236105 [4] NCCL INFO Channel 16/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-168-30:235649:236109 [1] NCCL INFO Channel 05/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-168-30:235654:236107 [6] NCCL INFO Channel 09/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-168-30:235650:236106 [2] NCCL INFO Channel 11/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-168-30:235651:236108 [3] NCCL INFO Channel 15/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-168-30:235652:236105 [4] NCCL INFO Channel 17/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-168-30:235653:236103 [5] NCCL INFO Channel 17/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-168-30:235649:236109 [1] NCCL INFO Channel 06/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-168-30:235654:236107 [6] NCCL INFO Channel 10/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-168-30:235650:236106 [2] NCCL INFO Channel 12/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-168-30:235651:236108 [3] NCCL INFO Channel 16/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-168-30:235652:236105 [4] NCCL INFO Channel 18/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-168-30:235653:236103 [5] NCCL INFO Channel 18/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-168-30:235649:236109 [1] NCCL INFO Channel 07/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-168-30:235654:236107 [6] NCCL INFO Channel 11/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-78:114512:114968 [0] NCCL INFO Connected all rings
+ip-26-0-168-30:235650:236106 [2] NCCL INFO Channel 13/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-168-30:235651:236108 [3] NCCL INFO Channel 17/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-168-30:235652:236105 [4] NCCL INFO Channel 19/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-168-30:235653:236103 [5] NCCL INFO Channel 19/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-168-30:235649:236109 [1] NCCL INFO Channel 08/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-168-30:235654:236107 [6] NCCL INFO Channel 12/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-225:124162:124627 [2] NCCL INFO Channel 19/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-168-30:235650:236106 [2] NCCL INFO Channel 14/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-168-30:235651:236108 [3] NCCL INFO Channel 18/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-103:96889:97348 [1] NCCL INFO Connected all rings
+ip-26-0-168-30:235652:236105 [4] NCCL INFO Channel 20/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-168-30:235653:236103 [5] NCCL INFO Channel 20/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-168-30:235649:236109 [1] NCCL INFO Channel 09/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-168-30:235654:236107 [6] NCCL INFO Channel 13/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-168-30:235651:236108 [3] NCCL INFO Channel 19/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-168-30:235650:236106 [2] NCCL INFO Channel 15/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-168-30:235652:236105 [4] NCCL INFO Channel 21/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-168-30:235649:236109 [1] NCCL INFO Channel 10/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-168-30:235653:236103 [5] NCCL INFO Channel 21/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-168-30:235654:236107 [6] NCCL INFO Channel 14/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-225:124164:124623 [4] NCCL INFO Channel 17/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-225:124161:124621 [1] NCCL INFO Channel 17/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-168-30:235650:236106 [2] NCCL INFO Channel 16/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-168-30:235651:236108 [3] NCCL INFO Channel 20/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-168-30:235652:236105 [4] NCCL INFO Channel 22/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-168-30:235649:236109 [1] NCCL INFO Channel 11/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-168-30:235653:236103 [5] NCCL INFO Channel 22/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-78:114513:114973 [1] NCCL INFO Connected all rings
+ip-26-0-168-30:235654:236107 [6] NCCL INFO Channel 15/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-142:62888:63340 [7] NCCL INFO Connected all trees
+ip-26-0-168-30:235650:236106 [2] NCCL INFO Channel 17/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-168-30:235651:236108 [3] NCCL INFO Channel 21/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-168-30:235652:236105 [4] NCCL INFO Channel 23/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-168-30:235649:236109 [1] NCCL INFO Channel 12/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-131:1630205:1630687 [0] NCCL INFO Connected all rings
+ip-26-0-168-30:235653:236103 [5] NCCL INFO Channel 23/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-168-30:235654:236107 [6] NCCL INFO Channel 16/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-142:62888:63340 [7] NCCL INFO NVLS comm 0x93bef90 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-168-30:235650:236106 [2] NCCL INFO Channel 18/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-168-30:235651:236108 [3] NCCL INFO Channel 22/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-168-30:235649:236109 [1] NCCL INFO Channel 13/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-168-30:235654:236107 [6] NCCL INFO Channel 17/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-103:96890:97350 [2] NCCL INFO Connected all rings
+ip-26-0-161-78:114517:114969 [5] NCCL INFO Connected all rings
+ip-26-0-160-225:124162:124627 [2] NCCL INFO Channel 20/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-78:114519:114970 [7] NCCL INFO Channel 01/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-168-30:235651:236108 [3] NCCL INFO Channel 23/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-168-30:235650:236106 [2] NCCL INFO Channel 19/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-168-30:235649:236109 [1] NCCL INFO Channel 14/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-168-30:235654:236107 [6] NCCL INFO Channel 18/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-168-30:235650:236106 [2] NCCL INFO Channel 20/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-168-30:235649:236109 [1] NCCL INFO Channel 15/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-168-30:235654:236107 [6] NCCL INFO Channel 19/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-168-30:235650:236106 [2] NCCL INFO Channel 21/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-168-30:235649:236109 [1] NCCL INFO Channel 16/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-168-30:235654:236107 [6] NCCL INFO Channel 20/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-103:96888:97344 [0] NCCL INFO Connected all rings
+ip-26-0-161-78:114519:114970 [7] NCCL INFO Channel 02/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-225:124161:124621 [1] NCCL INFO Channel 18/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-225:124164:124623 [4] NCCL INFO Channel 18/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-225:124162:124627 [2] NCCL INFO Channel 21/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-78:114515:114972 [3] NCCL INFO Connected all rings
+ip-26-0-168-30:235650:236106 [2] NCCL INFO Channel 22/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-168-30:235649:236109 [1] NCCL INFO Channel 17/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-168-30:235654:236107 [6] NCCL INFO Channel 21/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-225:124161:124621 [1] NCCL INFO Channel 19/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-225:124164:124623 [4] NCCL INFO Channel 19/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-225:124162:124627 [2] NCCL INFO Channel 22/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-78:114519:114970 [7] NCCL INFO Channel 03/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-78:114519:114970 [7] NCCL INFO Channel 04/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-168-30:235650:236106 [2] NCCL INFO Channel 23/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-168-30:235649:236109 [1] NCCL INFO Channel 18/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-168-30:235654:236107 [6] NCCL INFO Channel 22/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-78:114519:114970 [7] NCCL INFO Channel 05/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-78:114519:114970 [7] NCCL INFO Channel 06/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-225:124161:124621 [1] NCCL INFO Channel 20/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-78:114519:114970 [7] NCCL INFO Channel 07/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-165-131:1630207:1630694 [2] NCCL INFO Connected all rings
+ip-26-0-168-30:235649:236109 [1] NCCL INFO Channel 19/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-168-30:235654:236107 [6] NCCL INFO Channel 23/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-78:114519:114970 [7] NCCL INFO Channel 08/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-225:124162:124627 [2] NCCL INFO Channel 23/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-225:124164:124623 [4] NCCL INFO Channel 20/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-103:96891:97347 [3] NCCL INFO Connected all rings
+ip-26-0-161-78:114519:114970 [7] NCCL INFO Channel 09/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-138:96563:97045 [0] NCCL INFO Connected all trees
+ip-26-0-161-138:96563:97045 [0] NCCL INFO NVLS comm 0x9c16200 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-78:114519:114970 [7] NCCL INFO Channel 10/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-78:114519:114970 [7] NCCL INFO Channel 11/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-168-30:235649:236109 [1] NCCL INFO Channel 20/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-225:124161:124621 [1] NCCL INFO Channel 21/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-59:1557268:1557728 [2] NCCL INFO Connected all trees
+ip-26-0-161-78:114519:114970 [7] NCCL INFO Channel 12/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-165-59:1557268:1557728 [2] NCCL INFO NVLS comm 0x8db6590 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-78:114519:114970 [7] NCCL INFO Channel 13/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-165-131:1630206:1630692 [1] NCCL INFO Connected all rings
+ip-26-0-165-131:1630212:1630690 [7] NCCL INFO Connected all rings
+ip-26-0-165-131:1630212:1630690 [7] NCCL INFO Channel 00/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-78:114519:114970 [7] NCCL INFO Channel 14/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-168-30:235649:236109 [1] NCCL INFO Channel 21/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-168-30:235649:236109 [1] NCCL INFO Channel 22/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-78:114519:114970 [7] NCCL INFO Channel 15/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-168-30:235649:236109 [1] NCCL INFO Channel 23/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-78:114519:114970 [7] NCCL INFO Channel 16/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-78:114519:114970 [7] NCCL INFO Channel 17/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-78:114519:114970 [7] NCCL INFO Channel 18/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-78:114519:114970 [7] NCCL INFO Channel 19/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-103:96893:97346 [5] NCCL INFO Connected all rings
+ip-26-0-161-78:114519:114970 [7] NCCL INFO Channel 20/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-225:124164:124623 [4] NCCL INFO Channel 21/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-225:124161:124621 [1] NCCL INFO Channel 22/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-142:62881:63337 [0] NCCL INFO Connected all trees
+ip-26-0-161-142:62881:63337 [0] NCCL INFO NVLS comm 0xa371a00 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-78:114514:114971 [2] NCCL INFO Channel 00/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-78:114519:114970 [7] NCCL INFO Channel 21/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-165-131:1630212:1630690 [7] NCCL INFO Channel 01/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-78:114514:114971 [2] NCCL INFO Channel 01/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-78:114519:114970 [7] NCCL INFO Channel 22/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-78:114514:114971 [2] NCCL INFO Channel 02/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-78:114519:114970 [7] NCCL INFO Channel 23/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-225:124164:124623 [4] NCCL INFO Channel 22/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-165-59:1557273:1557723 [7] NCCL INFO Connected all trees
+ip-26-0-165-59:1557273:1557723 [7] NCCL INFO NVLS comm 0xa268900 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-78:114514:114971 [2] NCCL INFO Channel 03/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-225:124161:124621 [1] NCCL INFO Channel 23/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-78:114518:114974 [6] NCCL INFO Channel 00/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-78:114514:114971 [2] NCCL INFO Channel 04/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-78:114518:114974 [6] NCCL INFO Channel 01/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-78:114517:114969 [5] NCCL INFO Channel 00/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-78:114514:114971 [2] NCCL INFO Channel 05/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-78:114518:114974 [6] NCCL INFO Channel 02/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-78:114517:114969 [5] NCCL INFO Channel 01/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-225:124164:124623 [4] NCCL INFO Channel 23/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-78:114515:114972 [3] NCCL INFO Channel 00/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-78:114514:114971 [2] NCCL INFO Channel 06/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-78:114518:114974 [6] NCCL INFO Channel 03/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-78:114517:114969 [5] NCCL INFO Channel 02/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-103:96894:97349 [6] NCCL INFO Connected all rings
+ip-26-0-161-78:114515:114972 [3] NCCL INFO Channel 01/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-78:114514:114971 [2] NCCL INFO Channel 07/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-78:114518:114974 [6] NCCL INFO Channel 04/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-78:114517:114969 [5] NCCL INFO Channel 03/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-78:114513:114973 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-78:114515:114972 [3] NCCL INFO Channel 02/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-78:114514:114971 [2] NCCL INFO Channel 08/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-78:114518:114974 [6] NCCL INFO Channel 05/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-78:114517:114969 [5] NCCL INFO Channel 04/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-78:114516:114975 [4] NCCL INFO Channel 00/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-78:114513:114973 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-78:114515:114972 [3] NCCL INFO Channel 03/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-78:114514:114971 [2] NCCL INFO Channel 09/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-78:114518:114974 [6] NCCL INFO Channel 06/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-78:114517:114969 [5] NCCL INFO Channel 05/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-78:114516:114975 [4] NCCL INFO Channel 01/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-165-164:1227469:1227920 [6] NCCL INFO Connected all rings
+ip-26-0-161-78:114513:114973 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-103:96895:97351 [7] NCCL INFO Connected all rings
+ip-26-0-161-78:114515:114972 [3] NCCL INFO Channel 04/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-103:96895:97351 [7] NCCL INFO Channel 00/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-78:114518:114974 [6] NCCL INFO Channel 07/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-78:114514:114971 [2] NCCL INFO Channel 10/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-78:114517:114969 [5] NCCL INFO Channel 06/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-78:114516:114975 [4] NCCL INFO Channel 02/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-78:114513:114973 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-78:114515:114972 [3] NCCL INFO Channel 05/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-78:114518:114974 [6] NCCL INFO Channel 08/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-78:114514:114971 [2] NCCL INFO Channel 11/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-103:96895:97351 [7] NCCL INFO Channel 01/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-78:114517:114969 [5] NCCL INFO Channel 07/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-78:114516:114975 [4] NCCL INFO Channel 03/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-78:114513:114973 [1] NCCL INFO Channel 04/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-131:1630211:1630688 [6] NCCL INFO Connected all rings
+ip-26-0-161-78:114515:114972 [3] NCCL INFO Channel 06/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-78:114518:114974 [6] NCCL INFO Channel 09/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-78:114514:114971 [2] NCCL INFO Channel 12/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-78:114517:114969 [5] NCCL INFO Channel 08/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-78:114516:114975 [4] NCCL INFO Channel 04/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-103:96895:97351 [7] NCCL INFO Channel 02/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-78:114513:114973 [1] NCCL INFO Channel 05/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-78:114515:114972 [3] NCCL INFO Channel 07/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-78:114518:114974 [6] NCCL INFO Channel 10/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-78:114514:114971 [2] NCCL INFO Channel 13/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-78:114517:114969 [5] NCCL INFO Channel 09/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-78:114516:114975 [4] NCCL INFO Channel 05/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-103:96895:97351 [7] NCCL INFO Channel 03/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-78:114513:114973 [1] NCCL INFO Channel 06/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-78:114515:114972 [3] NCCL INFO Channel 08/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-78:114518:114974 [6] NCCL INFO Channel 11/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-78:114514:114971 [2] NCCL INFO Channel 14/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-78:114517:114969 [5] NCCL INFO Channel 10/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-78:114516:114975 [4] NCCL INFO Channel 06/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-103:96895:97351 [7] NCCL INFO Channel 04/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-78:114513:114973 [1] NCCL INFO Channel 07/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-78:114518:114974 [6] NCCL INFO Channel 12/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-78:114515:114972 [3] NCCL INFO Channel 09/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-78:114514:114971 [2] NCCL INFO Channel 15/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-78:114517:114969 [5] NCCL INFO Channel 11/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-78:114516:114975 [4] NCCL INFO Channel 07/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-165-131:1630210:1630689 [5] NCCL INFO Connected all rings
+ip-26-0-161-103:96895:97351 [7] NCCL INFO Channel 05/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-78:114513:114973 [1] NCCL INFO Channel 08/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-78:114514:114971 [2] NCCL INFO Channel 16/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-78:114518:114974 [6] NCCL INFO Channel 13/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-78:114517:114969 [5] NCCL INFO Channel 12/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-78:114515:114972 [3] NCCL INFO Channel 10/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-78:114516:114975 [4] NCCL INFO Channel 08/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-103:96895:97351 [7] NCCL INFO Channel 06/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-78:114513:114973 [1] NCCL INFO Channel 09/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-78:114518:114974 [6] NCCL INFO Channel 14/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-78:114514:114971 [2] NCCL INFO Channel 17/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-78:114517:114969 [5] NCCL INFO Channel 13/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-78:114515:114972 [3] NCCL INFO Channel 11/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-78:114516:114975 [4] NCCL INFO Channel 09/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-165-164:1227466:1227921 [3] NCCL INFO Connected all rings
+ip-26-0-168-34:272150:272606 [0] NCCL INFO Connected all trees
+ip-26-0-161-103:96895:97351 [7] NCCL INFO Channel 07/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-165-131:1630212:1630690 [7] NCCL INFO Channel 02/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-78:114513:114973 [1] NCCL INFO Channel 10/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-78:114514:114971 [2] NCCL INFO Channel 18/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-78:114518:114974 [6] NCCL INFO Channel 15/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-78:114515:114972 [3] NCCL INFO Channel 12/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-78:114517:114969 [5] NCCL INFO Channel 14/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-78:114516:114975 [4] NCCL INFO Channel 10/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-168-34:272150:272606 [0] NCCL INFO NVLS comm 0x9fd74c0 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-103:96895:97351 [7] NCCL INFO Channel 08/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-78:114514:114971 [2] NCCL INFO Channel 19/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-78:114513:114973 [1] NCCL INFO Channel 11/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-78:114517:114969 [5] NCCL INFO Channel 15/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-78:114516:114975 [4] NCCL INFO Channel 11/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-78:114518:114974 [6] NCCL INFO Channel 16/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-103:96895:97351 [7] NCCL INFO Channel 09/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-165-131:1630208:1630691 [3] NCCL INFO Connected all rings
+ip-26-0-161-103:96895:97351 [7] NCCL INFO Channel 10/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-78:114515:114972 [3] NCCL INFO Channel 13/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-165-131:1630212:1630690 [7] NCCL INFO Channel 03/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-78:114514:114971 [2] NCCL INFO Channel 20/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-78:114517:114969 [5] NCCL INFO Channel 16/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-78:114513:114973 [1] NCCL INFO Channel 12/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-78:114515:114972 [3] NCCL INFO Channel 14/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-165-131:1630212:1630690 [7] NCCL INFO Channel 04/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-78:114517:114969 [5] NCCL INFO Channel 17/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-78:114514:114971 [2] NCCL INFO Channel 21/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-103:96895:97351 [7] NCCL INFO Channel 11/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-78:114513:114973 [1] NCCL INFO Channel 13/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-78:114515:114972 [3] NCCL INFO Channel 15/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-165-131:1630212:1630690 [7] NCCL INFO Channel 05/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-165-164:1227464:1227926 [1] NCCL INFO Connected all rings
+ip-26-0-161-78:114517:114969 [5] NCCL INFO Channel 18/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-78:114514:114971 [2] NCCL INFO Channel 22/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-78:114513:114973 [1] NCCL INFO Channel 14/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-78:114515:114972 [3] NCCL INFO Channel 16/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-165-131:1630212:1630690 [7] NCCL INFO Channel 06/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-103:96895:97351 [7] NCCL INFO Channel 12/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-78:114517:114969 [5] NCCL INFO Channel 19/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-78:114514:114971 [2] NCCL INFO Channel 23/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-78:114513:114973 [1] NCCL INFO Channel 15/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-131:1630212:1630690 [7] NCCL INFO Channel 07/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-78:114515:114972 [3] NCCL INFO Channel 17/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-78:114517:114969 [5] NCCL INFO Channel 20/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-78:114513:114973 [1] NCCL INFO Channel 16/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-131:1630212:1630690 [7] NCCL INFO Channel 08/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-103:96895:97351 [7] NCCL INFO Channel 13/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-78:114515:114972 [3] NCCL INFO Channel 18/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-78:114516:114975 [4] NCCL INFO Channel 12/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-78:114517:114969 [5] NCCL INFO Channel 21/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-78:114513:114973 [1] NCCL INFO Channel 17/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-78:114515:114972 [3] NCCL INFO Channel 19/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-78:114517:114969 [5] NCCL INFO Channel 22/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-78:114516:114975 [4] NCCL INFO Channel 13/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-103:96895:97351 [7] NCCL INFO Channel 14/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-78:114513:114973 [1] NCCL INFO Channel 18/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-78:114515:114972 [3] NCCL INFO Channel 20/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-165-131:1630209:1630693 [4] NCCL INFO Connected all rings
+ip-26-0-161-78:114517:114969 [5] NCCL INFO Channel 23/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-78:114516:114975 [4] NCCL INFO Channel 14/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-78:114518:114974 [6] NCCL INFO Channel 17/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-78:114513:114973 [1] NCCL INFO Channel 19/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-78:114515:114972 [3] NCCL INFO Channel 21/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-165-131:1630212:1630690 [7] NCCL INFO Channel 09/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-103:96895:97351 [7] NCCL INFO Channel 15/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-78:114516:114975 [4] NCCL INFO Channel 15/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-78:114518:114974 [6] NCCL INFO Channel 18/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-78:114513:114973 [1] NCCL INFO Channel 20/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-78:114515:114972 [3] NCCL INFO Channel 22/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-78:114516:114975 [4] NCCL INFO Channel 16/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-78:114518:114974 [6] NCCL INFO Channel 19/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-103:96895:97351 [7] NCCL INFO Channel 16/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-78:114513:114973 [1] NCCL INFO Channel 21/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-131:1630212:1630690 [7] NCCL INFO Channel 10/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-78:114515:114972 [3] NCCL INFO Channel 23/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-78:114516:114975 [4] NCCL INFO Channel 17/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-103:96890:97350 [2] NCCL INFO Channel 00/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-78:114518:114974 [6] NCCL INFO Channel 20/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-78:114513:114973 [1] NCCL INFO Channel 22/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-78:114516:114975 [4] NCCL INFO Channel 18/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-103:96895:97351 [7] NCCL INFO Channel 17/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-78:114518:114974 [6] NCCL INFO Channel 21/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-78:114513:114973 [1] NCCL INFO Channel 23/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-103:96890:97350 [2] NCCL INFO Channel 01/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-78:114516:114975 [4] NCCL INFO Channel 19/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-165-131:1630212:1630690 [7] NCCL INFO Channel 11/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-103:96889:97348 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-78:114518:114974 [6] NCCL INFO Channel 22/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-164:1227468:1227922 [5] NCCL INFO Connected all rings
+ip-26-0-161-78:114516:114975 [4] NCCL INFO Channel 20/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-103:96895:97351 [7] NCCL INFO Channel 18/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-103:96890:97350 [2] NCCL INFO Channel 02/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-103:96889:97348 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-131:1630212:1630690 [7] NCCL INFO Channel 12/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-103:96895:97351 [7] NCCL INFO Channel 19/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-103:96891:97347 [3] NCCL INFO Channel 00/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-103:96892:97345 [4] NCCL INFO Channel 00/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-103:96890:97350 [2] NCCL INFO Channel 03/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-103:96889:97348 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-131:1630212:1630690 [7] NCCL INFO Channel 13/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-165-164:1227467:1227925 [4] NCCL INFO Connected all rings
+ip-26-0-161-103:96895:97351 [7] NCCL INFO Channel 20/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-78:114518:114974 [6] NCCL INFO Channel 23/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-103:96892:97345 [4] NCCL INFO Channel 01/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-103:96891:97347 [3] NCCL INFO Channel 01/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-103:96890:97350 [2] NCCL INFO Channel 04/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-103:96889:97348 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-131:1630212:1630690 [7] NCCL INFO Channel 14/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-165-164:1227465:1227924 [2] NCCL INFO Connected all rings
+ip-26-0-161-103:96893:97346 [5] NCCL INFO Channel 00/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-78:114516:114975 [4] NCCL INFO Channel 21/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-103:96895:97351 [7] NCCL INFO Channel 21/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-103:96891:97347 [3] NCCL INFO Channel 02/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-103:96890:97350 [2] NCCL INFO Channel 05/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-103:96889:97348 [1] NCCL INFO Channel 04/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-103:96892:97345 [4] NCCL INFO Channel 02/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-165-164:1227470:1227923 [7] NCCL INFO Connected all rings
+ip-26-0-165-164:1227470:1227923 [7] NCCL INFO Channel 00/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-103:96893:97346 [5] NCCL INFO Channel 01/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-103:96891:97347 [3] NCCL INFO Channel 03/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-103:96895:97351 [7] NCCL INFO Channel 22/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-103:96890:97350 [2] NCCL INFO Channel 06/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-103:96889:97348 [1] NCCL INFO Channel 05/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-103:96892:97345 [4] NCCL INFO Channel 03/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-103:96893:97346 [5] NCCL INFO Channel 02/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-103:96891:97347 [3] NCCL INFO Channel 04/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-103:96895:97351 [7] NCCL INFO Channel 23/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-103:96890:97350 [2] NCCL INFO Channel 07/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-103:96892:97345 [4] NCCL INFO Channel 04/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-103:96889:97348 [1] NCCL INFO Channel 06/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-103:96894:97349 [6] NCCL INFO Channel 00/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-103:96893:97346 [5] NCCL INFO Channel 03/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-103:96891:97347 [3] NCCL INFO Channel 05/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-103:96890:97350 [2] NCCL INFO Channel 08/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-103:96892:97345 [4] NCCL INFO Channel 05/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-103:96889:97348 [1] NCCL INFO Channel 07/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-103:96894:97349 [6] NCCL INFO Channel 01/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-164:1227463:1227919 [0] NCCL INFO Connected all rings
+ip-26-0-165-164:1227470:1227923 [7] NCCL INFO Channel 01/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-165-131:1630212:1630690 [7] NCCL INFO Channel 15/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-103:96893:97346 [5] NCCL INFO Channel 04/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-131:1630212:1630690 [7] NCCL INFO Channel 16/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-103:96891:97347 [3] NCCL INFO Channel 06/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-103:96889:97348 [1] NCCL INFO Channel 08/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-103:96890:97350 [2] NCCL INFO Channel 09/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-103:96894:97349 [6] NCCL INFO Channel 02/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-103:96892:97345 [4] NCCL INFO Channel 06/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-165-131:1630212:1630690 [7] NCCL INFO Channel 17/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-165-164:1227470:1227923 [7] NCCL INFO Channel 02/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-103:96893:97346 [5] NCCL INFO Channel 05/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-103:96891:97347 [3] NCCL INFO Channel 07/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-103:96889:97348 [1] NCCL INFO Channel 09/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-103:96890:97350 [2] NCCL INFO Channel 10/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-103:96894:97349 [6] NCCL INFO Channel 03/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-103:96892:97345 [4] NCCL INFO Channel 07/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-165-131:1630212:1630690 [7] NCCL INFO Channel 18/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-103:96893:97346 [5] NCCL INFO Channel 06/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-164:1227470:1227923 [7] NCCL INFO Channel 03/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-165-131:1630212:1630690 [7] NCCL INFO Channel 19/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-103:96891:97347 [3] NCCL INFO Channel 08/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-103:96889:97348 [1] NCCL INFO Channel 10/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-103:96890:97350 [2] NCCL INFO Channel 11/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-103:96894:97349 [6] NCCL INFO Channel 04/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-103:96892:97345 [4] NCCL INFO Channel 08/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-103:96893:97346 [5] NCCL INFO Channel 07/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-164:1227470:1227923 [7] NCCL INFO Channel 04/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-165-131:1630212:1630690 [7] NCCL INFO Channel 20/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-178:2815969:2816436 [6] NCCL INFO Connected all trees
+ip-26-0-161-103:96889:97348 [1] NCCL INFO Channel 11/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-103:96891:97347 [3] NCCL INFO Channel 09/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-103:96890:97350 [2] NCCL INFO Channel 12/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-103:96892:97345 [4] NCCL INFO Channel 09/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-103:96894:97349 [6] NCCL INFO Channel 05/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-103:96893:97346 [5] NCCL INFO Channel 08/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-178:2815969:2816436 [6] NCCL INFO NVLS comm 0x9cc60a0 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-131:1630212:1630690 [7] NCCL INFO Channel 21/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-165-164:1227470:1227923 [7] NCCL INFO Channel 05/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-103:96889:97348 [1] NCCL INFO Channel 12/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-103:96891:97347 [3] NCCL INFO Channel 10/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-103:96890:97350 [2] NCCL INFO Channel 13/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-103:96894:97349 [6] NCCL INFO Channel 06/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-103:96892:97345 [4] NCCL INFO Channel 10/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-103:96893:97346 [5] NCCL INFO Channel 09/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-78:114516:114975 [4] NCCL INFO Channel 22/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-165-131:1630207:1630694 [2] NCCL INFO Channel 00/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-165-131:1630212:1630690 [7] NCCL INFO Channel 22/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-103:96889:97348 [1] NCCL INFO Channel 13/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-103:96891:97347 [3] NCCL INFO Channel 11/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-103:96890:97350 [2] NCCL INFO Channel 14/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-103:96894:97349 [6] NCCL INFO Channel 07/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-103:96892:97345 [4] NCCL INFO Channel 11/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-103:96893:97346 [5] NCCL INFO Channel 10/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-131:1630206:1630692 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-131:1630207:1630694 [2] NCCL INFO Channel 01/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-165-164:1227470:1227923 [7] NCCL INFO Channel 06/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-165-131:1630212:1630690 [7] NCCL INFO Channel 23/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-103:96889:97348 [1] NCCL INFO Channel 14/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-103:96891:97347 [3] NCCL INFO Channel 12/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-103:96890:97350 [2] NCCL INFO Channel 15/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-103:96894:97349 [6] NCCL INFO Channel 08/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-103:96892:97345 [4] NCCL INFO Channel 12/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-103:96893:97346 [5] NCCL INFO Channel 11/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-131:1630206:1630692 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-164:1227470:1227923 [7] NCCL INFO Channel 07/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-165-131:1630207:1630694 [2] NCCL INFO Channel 02/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-165-164:1227470:1227923 [7] NCCL INFO Channel 08/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-103:96889:97348 [1] NCCL INFO Channel 15/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-103:96891:97347 [3] NCCL INFO Channel 13/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-103:96890:97350 [2] NCCL INFO Channel 16/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-103:96894:97349 [6] NCCL INFO Channel 09/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-103:96892:97345 [4] NCCL INFO Channel 13/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-103:96893:97346 [5] NCCL INFO Channel 12/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-131:1630206:1630692 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-131:1630207:1630694 [2] NCCL INFO Channel 03/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-165-164:1227470:1227923 [7] NCCL INFO Channel 09/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-165-164:1227470:1227923 [7] NCCL INFO Channel 10/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-165-131:1630207:1630694 [2] NCCL INFO Channel 04/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-103:96889:97348 [1] NCCL INFO Channel 16/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-103:96891:97347 [3] NCCL INFO Channel 14/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-165-131:1630206:1630692 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-103:96892:97345 [4] NCCL INFO Channel 14/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-103:96890:97350 [2] NCCL INFO Channel 17/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-103:96894:97349 [6] NCCL INFO Channel 10/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-103:96893:97346 [5] NCCL INFO Channel 13/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-164:1227470:1227923 [7] NCCL INFO Channel 11/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-165-131:1630210:1630689 [5] NCCL INFO Channel 00/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-131:1630207:1630694 [2] NCCL INFO Channel 05/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-165-164:1227470:1227923 [7] NCCL INFO Channel 12/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-165-131:1630206:1630692 [1] NCCL INFO Channel 04/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-78:114516:114975 [4] NCCL INFO Channel 23/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-103:96889:97348 [1] NCCL INFO Channel 17/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-103:96891:97347 [3] NCCL INFO Channel 15/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-103:96892:97345 [4] NCCL INFO Channel 15/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-103:96894:97349 [6] NCCL INFO Channel 11/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-103:96890:97350 [2] NCCL INFO Channel 18/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-103:96893:97346 [5] NCCL INFO Channel 14/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-164:1227470:1227923 [7] NCCL INFO Channel 13/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-165-131:1630210:1630689 [5] NCCL INFO Channel 01/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-131:1630207:1630694 [2] NCCL INFO Channel 06/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-165-131:1630206:1630692 [1] NCCL INFO Channel 05/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-164:1227470:1227923 [7] NCCL INFO Channel 14/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-103:96889:97348 [1] NCCL INFO Channel 18/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-103:96891:97347 [3] NCCL INFO Channel 16/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-103:96892:97345 [4] NCCL INFO Channel 16/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-103:96894:97349 [6] NCCL INFO Channel 12/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-103:96890:97350 [2] NCCL INFO Channel 19/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-103:96893:97346 [5] NCCL INFO Channel 15/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-164:1227470:1227923 [7] NCCL INFO Channel 15/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-165-131:1630210:1630689 [5] NCCL INFO Channel 02/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-131:1630207:1630694 [2] NCCL INFO Channel 07/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-165-131:1630206:1630692 [1] NCCL INFO Channel 06/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-225:124167:124624 [7] NCCL INFO Connected all trees
+ip-26-0-161-103:96889:97348 [1] NCCL INFO Channel 19/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-103:96891:97347 [3] NCCL INFO Channel 17/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-103:96892:97345 [4] NCCL INFO Channel 17/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-103:96894:97349 [6] NCCL INFO Channel 13/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-103:96890:97350 [2] NCCL INFO Channel 20/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-103:96893:97346 [5] NCCL INFO Channel 16/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-225:124167:124624 [7] NCCL INFO NVLS comm 0x9eb7fb0 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-164:1227470:1227923 [7] NCCL INFO Channel 16/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-165-131:1630210:1630689 [5] NCCL INFO Channel 03/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-131:1630206:1630692 [1] NCCL INFO Channel 07/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-131:1630207:1630694 [2] NCCL INFO Channel 08/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-103:96889:97348 [1] NCCL INFO Channel 20/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-103:96891:97347 [3] NCCL INFO Channel 18/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-103:96894:97349 [6] NCCL INFO Channel 14/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-103:96892:97345 [4] NCCL INFO Channel 18/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-103:96890:97350 [2] NCCL INFO Channel 21/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-103:96893:97346 [5] NCCL INFO Channel 17/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-164:1227470:1227923 [7] NCCL INFO Channel 17/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-165-131:1630210:1630689 [5] NCCL INFO Channel 04/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-131:1630206:1630692 [1] NCCL INFO Channel 08/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-131:1630207:1630694 [2] NCCL INFO Channel 09/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-165-131:1630211:1630688 [6] NCCL INFO Channel 00/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-103:96889:97348 [1] NCCL INFO Channel 21/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-103:96891:97347 [3] NCCL INFO Channel 19/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-165-131:1630207:1630694 [2] NCCL INFO Channel 10/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-103:96894:97349 [6] NCCL INFO Channel 15/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-103:96892:97345 [4] NCCL INFO Channel 19/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-103:96890:97350 [2] NCCL INFO Channel 22/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-103:96893:97346 [5] NCCL INFO Channel 18/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-131:1630210:1630689 [5] NCCL INFO Channel 05/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-131:1630211:1630688 [6] NCCL INFO Channel 01/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-164:1227470:1227923 [7] NCCL INFO Channel 18/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-165-131:1630207:1630694 [2] NCCL INFO Channel 11/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-165-131:1630206:1630692 [1] NCCL INFO Channel 09/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-131:1630210:1630689 [5] NCCL INFO Channel 06/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-131:1630211:1630688 [6] NCCL INFO Channel 02/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-103:96889:97348 [1] NCCL INFO Channel 22/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-103:96891:97347 [3] NCCL INFO Channel 20/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-103:96894:97349 [6] NCCL INFO Channel 16/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-103:96892:97345 [4] NCCL INFO Channel 20/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-103:96890:97350 [2] NCCL INFO Channel 23/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-103:96893:97346 [5] NCCL INFO Channel 19/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-131:1630207:1630694 [2] NCCL INFO Channel 12/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-165-131:1630206:1630692 [1] NCCL INFO Channel 10/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-164:1227470:1227923 [7] NCCL INFO Channel 19/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-165-131:1630210:1630689 [5] NCCL INFO Channel 07/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-131:1630211:1630688 [6] NCCL INFO Channel 03/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-103:96889:97348 [1] NCCL INFO Channel 23/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-131:1630207:1630694 [2] NCCL INFO Channel 13/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-165-131:1630206:1630692 [1] NCCL INFO Channel 11/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-103:96891:97347 [3] NCCL INFO Channel 21/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-103:96892:97345 [4] NCCL INFO Channel 21/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-103:96894:97349 [6] NCCL INFO Channel 17/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-103:96893:97346 [5] NCCL INFO Channel 20/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-242:1364902:1365366 [0] NCCL INFO Connected all trees
+ip-26-0-165-131:1630211:1630688 [6] NCCL INFO Channel 04/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-131:1630210:1630689 [5] NCCL INFO Channel 08/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-164:1227470:1227923 [7] NCCL INFO Channel 20/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-165-131:1630207:1630694 [2] NCCL INFO Channel 14/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-165-131:1630206:1630692 [1] NCCL INFO Channel 12/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-242:1364902:1365366 [0] NCCL INFO NVLS comm 0x9a12810 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-103:96892:97345 [4] NCCL INFO Channel 22/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-103:96891:97347 [3] NCCL INFO Channel 22/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-103:96894:97349 [6] NCCL INFO Channel 18/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-131:1630211:1630688 [6] NCCL INFO Channel 05/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-131:1630210:1630689 [5] NCCL INFO Channel 09/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-103:96893:97346 [5] NCCL INFO Channel 21/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-131:1630207:1630694 [2] NCCL INFO Channel 15/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-165-131:1630206:1630692 [1] NCCL INFO Channel 13/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-131:1630211:1630688 [6] NCCL INFO Channel 06/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-131:1630210:1630689 [5] NCCL INFO Channel 10/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-164:1227470:1227923 [7] NCCL INFO Channel 21/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-165-131:1630206:1630692 [1] NCCL INFO Channel 14/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-131:1630207:1630694 [2] NCCL INFO Channel 16/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-165-131:1630211:1630688 [6] NCCL INFO Channel 07/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-131:1630210:1630689 [5] NCCL INFO Channel 11/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-131:1630206:1630692 [1] NCCL INFO Channel 15/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-131:1630207:1630694 [2] NCCL INFO Channel 17/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-165-164:1227470:1227923 [7] NCCL INFO Channel 22/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-165-164:1227464:1227926 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-131:1630211:1630688 [6] NCCL INFO Channel 08/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-131:1630210:1630689 [5] NCCL INFO Channel 12/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-164:1227468:1227922 [5] NCCL INFO Channel 00/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-131:1630206:1630692 [1] NCCL INFO Channel 16/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-103:96892:97345 [4] NCCL INFO Channel 23/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-165-131:1630207:1630694 [2] NCCL INFO Channel 18/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-103:96891:97347 [3] NCCL INFO Channel 23/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-103:96894:97349 [6] NCCL INFO Channel 19/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-103:96893:97346 [5] NCCL INFO Channel 22/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-164:1227470:1227923 [7] NCCL INFO Channel 23/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-165-164:1227464:1227926 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-164:1227467:1227925 [4] NCCL INFO Channel 00/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-165-131:1630211:1630688 [6] NCCL INFO Channel 09/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-131:1630210:1630689 [5] NCCL INFO Channel 13/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-164:1227468:1227922 [5] NCCL INFO Channel 01/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-131:1630206:1630692 [1] NCCL INFO Channel 17/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-131:1630207:1630694 [2] NCCL INFO Channel 19/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-165-164:1227464:1227926 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-131:1630211:1630688 [6] NCCL INFO Channel 10/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-131:1630209:1630693 [4] NCCL INFO Channel 00/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-165-131:1630210:1630689 [5] NCCL INFO Channel 14/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-164:1227467:1227925 [4] NCCL INFO Channel 01/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-165-131:1630206:1630692 [1] NCCL INFO Channel 18/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-131:1630207:1630694 [2] NCCL INFO Channel 20/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-165-131:1630208:1630691 [3] NCCL INFO Channel 00/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-165-131:1630209:1630693 [4] NCCL INFO Channel 01/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-165-131:1630210:1630689 [5] NCCL INFO Channel 15/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-164:1227468:1227922 [5] NCCL INFO Channel 02/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-131:1630206:1630692 [1] NCCL INFO Channel 19/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-131:1630207:1630694 [2] NCCL INFO Channel 21/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-165-131:1630208:1630691 [3] NCCL INFO Channel 01/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-165-164:1227467:1227925 [4] NCCL INFO Channel 02/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-165-164:1227468:1227922 [5] NCCL INFO Channel 03/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-131:1630209:1630693 [4] NCCL INFO Channel 02/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-165-131:1630210:1630689 [5] NCCL INFO Channel 16/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-164:1227464:1227926 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-131:1630206:1630692 [1] NCCL INFO Channel 20/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-131:1630211:1630688 [6] NCCL INFO Channel 11/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-131:1630207:1630694 [2] NCCL INFO Channel 22/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-165-131:1630208:1630691 [3] NCCL INFO Channel 02/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-165-131:1630209:1630693 [4] NCCL INFO Channel 03/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-165-131:1630210:1630689 [5] NCCL INFO Channel 17/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-131:1630211:1630688 [6] NCCL INFO Channel 12/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-131:1630206:1630692 [1] NCCL INFO Channel 21/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-131:1630207:1630694 [2] NCCL INFO Channel 23/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-165-131:1630208:1630691 [3] NCCL INFO Channel 03/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-165-164:1227467:1227925 [4] NCCL INFO Channel 03/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-165-131:1630209:1630693 [4] NCCL INFO Channel 04/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-165-131:1630210:1630689 [5] NCCL INFO Channel 18/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-131:1630211:1630688 [6] NCCL INFO Channel 13/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-131:1630206:1630692 [1] NCCL INFO Channel 22/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-131:1630208:1630691 [3] NCCL INFO Channel 04/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-165-131:1630209:1630693 [4] NCCL INFO Channel 05/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-165-131:1630210:1630689 [5] NCCL INFO Channel 19/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-131:1630211:1630688 [6] NCCL INFO Channel 14/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-131:1630206:1630692 [1] NCCL INFO Channel 23/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-131:1630208:1630691 [3] NCCL INFO Channel 05/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-165-131:1630209:1630693 [4] NCCL INFO Channel 06/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-165-131:1630210:1630689 [5] NCCL INFO Channel 20/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-131:1630211:1630688 [6] NCCL INFO Channel 15/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-103:96894:97349 [6] NCCL INFO Channel 20/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-131:1630208:1630691 [3] NCCL INFO Channel 06/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-103:96893:97346 [5] NCCL INFO Channel 23/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-131:1630209:1630693 [4] NCCL INFO Channel 07/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-165-131:1630210:1630689 [5] NCCL INFO Channel 21/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-164:1227468:1227922 [5] NCCL INFO Channel 04/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-131:1630208:1630691 [3] NCCL INFO Channel 07/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-165-164:1227464:1227926 [1] NCCL INFO Channel 04/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-164:1227468:1227922 [5] NCCL INFO Channel 05/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-103:96894:97349 [6] NCCL INFO Channel 21/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-164:1227464:1227926 [1] NCCL INFO Channel 05/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-164:1227468:1227922 [5] NCCL INFO Channel 06/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-164:1227464:1227926 [1] NCCL INFO Channel 06/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-164:1227468:1227922 [5] NCCL INFO Channel 07/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-164:1227464:1227926 [1] NCCL INFO Channel 07/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-164:1227468:1227922 [5] NCCL INFO Channel 08/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-164:1227467:1227925 [4] NCCL INFO Channel 04/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-165-164:1227464:1227926 [1] NCCL INFO Channel 08/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-131:1630211:1630688 [6] NCCL INFO Channel 16/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-164:1227468:1227922 [5] NCCL INFO Channel 09/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-164:1227467:1227925 [4] NCCL INFO Channel 05/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-165-164:1227464:1227926 [1] NCCL INFO Channel 09/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-103:96894:97349 [6] NCCL INFO Channel 22/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-131:1630208:1630691 [3] NCCL INFO Channel 08/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-165-164:1227468:1227922 [5] NCCL INFO Channel 10/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-164:1227467:1227925 [4] NCCL INFO Channel 06/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-165-164:1227464:1227926 [1] NCCL INFO Channel 10/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-164:1227468:1227922 [5] NCCL INFO Channel 11/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-164:1227467:1227925 [4] NCCL INFO Channel 07/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-165-131:1630210:1630689 [5] NCCL INFO Channel 22/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-164:1227464:1227926 [1] NCCL INFO Channel 11/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-131:1630209:1630693 [4] NCCL INFO Channel 08/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-165-164:1227466:1227921 [3] NCCL INFO Channel 00/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-165-164:1227468:1227922 [5] NCCL INFO Channel 12/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-164:1227467:1227925 [4] NCCL INFO Channel 08/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-165-164:1227464:1227926 [1] NCCL INFO Channel 12/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-164:1227465:1227924 [2] NCCL INFO Channel 00/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-165-164:1227466:1227921 [3] NCCL INFO Channel 01/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-165-164:1227468:1227922 [5] NCCL INFO Channel 13/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-164:1227467:1227925 [4] NCCL INFO Channel 09/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-165-164:1227464:1227926 [1] NCCL INFO Channel 13/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-164:1227465:1227924 [2] NCCL INFO Channel 01/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-165-131:1630211:1630688 [6] NCCL INFO Channel 17/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-131:1630208:1630691 [3] NCCL INFO Channel 09/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-165-164:1227466:1227921 [3] NCCL INFO Channel 02/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-165-164:1227468:1227922 [5] NCCL INFO Channel 14/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-164:1227469:1227920 [6] NCCL INFO Channel 00/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-164:1227467:1227925 [4] NCCL INFO Channel 10/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-165-164:1227464:1227926 [1] NCCL INFO Channel 14/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-164:1227465:1227924 [2] NCCL INFO Channel 02/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-165-164:1227466:1227921 [3] NCCL INFO Channel 03/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-165-164:1227468:1227922 [5] NCCL INFO Channel 15/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-164:1227469:1227920 [6] NCCL INFO Channel 01/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-242:1364909:1365373 [7] NCCL INFO Connected all trees
+ip-26-0-165-164:1227467:1227925 [4] NCCL INFO Channel 11/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-165-164:1227464:1227926 [1] NCCL INFO Channel 15/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-164:1227465:1227924 [2] NCCL INFO Channel 03/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-242:1364909:1365373 [7] NCCL INFO NVLS comm 0xa0d8c00 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-164:1227468:1227922 [5] NCCL INFO Channel 16/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-164:1227466:1227921 [3] NCCL INFO Channel 04/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-165-164:1227469:1227920 [6] NCCL INFO Channel 02/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-164:1227464:1227926 [1] NCCL INFO Channel 16/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-164:1227465:1227924 [2] NCCL INFO Channel 04/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-165-164:1227467:1227925 [4] NCCL INFO Channel 12/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-165-131:1630210:1630689 [5] NCCL INFO Channel 23/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-131:1630209:1630693 [4] NCCL INFO Channel 09/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-165-131:1630211:1630688 [6] NCCL INFO Channel 18/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-164:1227468:1227922 [5] NCCL INFO Channel 17/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-164:1227466:1227921 [3] NCCL INFO Channel 05/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-165-164:1227469:1227920 [6] NCCL INFO Channel 03/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-164:1227464:1227926 [1] NCCL INFO Channel 17/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-164:1227465:1227924 [2] NCCL INFO Channel 05/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-165-164:1227467:1227925 [4] NCCL INFO Channel 13/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-123:32708:33162 [4] NCCL INFO Connected all trees
+ip-26-0-165-131:1630208:1630691 [3] NCCL INFO Channel 10/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-165-164:1227468:1227922 [5] NCCL INFO Channel 18/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-164:1227466:1227921 [3] NCCL INFO Channel 06/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-123:32708:33162 [4] NCCL INFO NVLS comm 0xa4ddf70 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-164:1227469:1227920 [6] NCCL INFO Channel 04/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-164:1227464:1227926 [1] NCCL INFO Channel 18/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-164:1227465:1227924 [2] NCCL INFO Channel 06/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-165-164:1227467:1227925 [4] NCCL INFO Channel 14/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-165-131:1630209:1630693 [4] NCCL INFO Channel 10/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-165-164:1227468:1227922 [5] NCCL INFO Channel 19/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-164:1227466:1227921 [3] NCCL INFO Channel 07/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-165-164:1227469:1227920 [6] NCCL INFO Channel 05/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-164:1227464:1227926 [1] NCCL INFO Channel 19/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-164:1227465:1227924 [2] NCCL INFO Channel 07/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-165-164:1227467:1227925 [4] NCCL INFO Channel 15/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-165-131:1630211:1630688 [6] NCCL INFO Channel 19/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-131:1630208:1630691 [3] NCCL INFO Channel 11/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-165-164:1227468:1227922 [5] NCCL INFO Channel 20/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-164:1227466:1227921 [3] NCCL INFO Channel 08/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-165-164:1227464:1227926 [1] NCCL INFO Channel 20/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-164:1227469:1227920 [6] NCCL INFO Channel 06/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-164:1227465:1227924 [2] NCCL INFO Channel 08/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-165-164:1227467:1227925 [4] NCCL INFO Channel 16/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-165-164:1227468:1227922 [5] NCCL INFO Channel 21/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-131:1630209:1630693 [4] NCCL INFO Channel 11/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-165-164:1227466:1227921 [3] NCCL INFO Channel 09/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-165-164:1227464:1227926 [1] NCCL INFO Channel 21/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-164:1227465:1227924 [2] NCCL INFO Channel 09/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-165-164:1227469:1227920 [6] NCCL INFO Channel 07/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-164:1227467:1227925 [4] NCCL INFO Channel 17/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-165-164:1227468:1227922 [5] NCCL INFO Channel 22/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-103:96894:97349 [6] NCCL INFO Channel 23/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-131:1630211:1630688 [6] NCCL INFO Channel 20/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-131:1630208:1630691 [3] NCCL INFO Channel 12/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-165-164:1227466:1227921 [3] NCCL INFO Channel 10/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-165-164:1227464:1227926 [1] NCCL INFO Channel 22/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-164:1227465:1227924 [2] NCCL INFO Channel 10/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-165-164:1227469:1227920 [6] NCCL INFO Channel 08/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-164:1227467:1227925 [4] NCCL INFO Channel 18/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-165-164:1227468:1227922 [5] NCCL INFO Channel 23/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-164:1227464:1227926 [1] NCCL INFO Channel 23/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-164:1227466:1227921 [3] NCCL INFO Channel 11/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-165-164:1227465:1227924 [2] NCCL INFO Channel 11/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-165-164:1227469:1227920 [6] NCCL INFO Channel 09/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-164:1227467:1227925 [4] NCCL INFO Channel 19/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-165-131:1630209:1630693 [4] NCCL INFO Channel 12/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-165-164:1227466:1227921 [3] NCCL INFO Channel 12/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-165-164:1227465:1227924 [2] NCCL INFO Channel 12/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-165-164:1227467:1227925 [4] NCCL INFO Channel 20/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-165-164:1227469:1227920 [6] NCCL INFO Channel 10/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-164:1227466:1227921 [3] NCCL INFO Channel 13/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-165-131:1630211:1630688 [6] NCCL INFO Channel 21/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-164:1227465:1227924 [2] NCCL INFO Channel 13/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-165-131:1630208:1630691 [3] NCCL INFO Channel 13/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-165-164:1227467:1227925 [4] NCCL INFO Channel 21/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-165-164:1227469:1227920 [6] NCCL INFO Channel 11/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-131:1630209:1630693 [4] NCCL INFO Channel 13/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-165-164:1227466:1227921 [3] NCCL INFO Channel 14/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-165-164:1227467:1227925 [4] NCCL INFO Channel 22/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-165-164:1227465:1227924 [2] NCCL INFO Channel 14/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-165-164:1227469:1227920 [6] NCCL INFO Channel 12/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-164:1227465:1227924 [2] NCCL INFO Channel 15/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-165-131:1630209:1630693 [4] NCCL INFO Channel 14/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-165-164:1227469:1227920 [6] NCCL INFO Channel 13/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-164:1227465:1227924 [2] NCCL INFO Channel 16/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-168-34:272157:272613 [7] NCCL INFO Connected all trees
+ip-26-0-168-34:272157:272613 [7] NCCL INFO NVLS comm 0x9663cf0 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-131:1630211:1630688 [6] NCCL INFO Channel 22/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-131:1630208:1630691 [3] NCCL INFO Channel 14/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-165-131:1630209:1630693 [4] NCCL INFO Channel 15/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-165-164:1227466:1227921 [3] NCCL INFO Channel 15/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-165-164:1227467:1227925 [4] NCCL INFO Channel 23/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-138:96565:97044 [2] NCCL INFO Connected all trees
+ip-26-0-161-138:96565:97044 [2] NCCL INFO NVLS comm 0xa46bac0 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-131:1630211:1630688 [6] NCCL INFO Channel 23/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-164:1227466:1227921 [3] NCCL INFO Channel 16/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-165-131:1630209:1630693 [4] NCCL INFO Channel 16/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-165-164:1227466:1227921 [3] NCCL INFO Channel 17/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-165-164:1227465:1227924 [2] NCCL INFO Channel 17/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-165-164:1227469:1227920 [6] NCCL INFO Channel 14/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-164:1227466:1227921 [3] NCCL INFO Channel 18/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-123:32709:33164 [5] NCCL INFO Connected all trees
+ip-26-0-161-123:32709:33164 [5] NCCL INFO NVLS comm 0xa4a44d0 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-164:1227465:1227924 [2] NCCL INFO Channel 18/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-165-164:1227469:1227920 [6] NCCL INFO Channel 15/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-164:1227466:1227921 [3] NCCL INFO Channel 19/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-165-164:1227469:1227920 [6] NCCL INFO Channel 16/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-131:1630209:1630693 [4] NCCL INFO Channel 17/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-165-164:1227465:1227924 [2] NCCL INFO Channel 19/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-165-164:1227466:1227921 [3] NCCL INFO Channel 20/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-165-164:1227465:1227924 [2] NCCL INFO Channel 20/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-178:2815964:2816440 [1] NCCL INFO Connected all trees
+ip-26-0-165-164:1227469:1227920 [6] NCCL INFO Channel 17/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-178:2815964:2816440 [1] NCCL INFO NVLS comm 0x89f39e0 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-164:1227466:1227921 [3] NCCL INFO Channel 21/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-165-164:1227465:1227924 [2] NCCL INFO Channel 21/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-165-59:1557269:1557724 [3] NCCL INFO Connected all trees
+ip-26-0-165-164:1227469:1227920 [6] NCCL INFO Channel 18/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-59:1557269:1557724 [3] NCCL INFO NVLS comm 0xa155f80 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-59:1557272:1557722 [6] NCCL INFO Connected all trees
+ip-26-0-165-59:1557272:1557722 [6] NCCL INFO NVLS comm 0x966b0a0 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-164:1227466:1227921 [3] NCCL INFO Channel 22/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-165-164:1227465:1227924 [2] NCCL INFO Channel 22/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-165-164:1227469:1227920 [6] NCCL INFO Channel 19/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-123:32707:33163 [3] NCCL INFO Connected all trees
+ip-26-0-161-123:32707:33163 [3] NCCL INFO NVLS comm 0x8aede90 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-164:1227466:1227921 [3] NCCL INFO Channel 23/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-165-164:1227465:1227924 [2] NCCL INFO Channel 23/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-165-164:1227469:1227920 [6] NCCL INFO Channel 20/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-131:1630208:1630691 [3] NCCL INFO Channel 15/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-165-59:1557270:1557725 [4] NCCL INFO Connected all trees
+ip-26-0-165-164:1227463:1227919 [0] NCCL INFO Connected all trees
+ip-26-0-165-59:1557270:1557725 [4] NCCL INFO NVLS comm 0x8b8b740 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-164:1227463:1227919 [0] NCCL INFO NVLS comm 0x8a0d780 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-164:1227469:1227920 [6] NCCL INFO Channel 21/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-123:32710:33160 [6] NCCL INFO Connected all trees
+ip-26-0-161-123:32710:33160 [6] NCCL INFO NVLS comm 0x9b7e830 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-131:1630209:1630693 [4] NCCL INFO Channel 18/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-165-164:1227469:1227920 [6] NCCL INFO Channel 22/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-123:32705:33166 [1] NCCL INFO Connected all trees
+ip-26-0-161-178:2815968:2816434 [5] NCCL INFO Connected all trees
+ip-26-0-161-123:32705:33166 [1] NCCL INFO NVLS comm 0x9c5f0d0 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-131:1630208:1630691 [3] NCCL INFO Channel 16/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-178:2815968:2816434 [5] NCCL INFO NVLS comm 0xa2fead0 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-138:96564:97046 [1] NCCL INFO Connected all trees
+ip-26-0-161-138:96564:97046 [1] NCCL INFO NVLS comm 0x9897100 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-123:32706:33161 [2] NCCL INFO Connected all trees
+ip-26-0-161-123:32706:33161 [2] NCCL INFO NVLS comm 0x9c86c40 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-131:1630209:1630693 [4] NCCL INFO Channel 19/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-165-131:1630208:1630691 [3] NCCL INFO Channel 17/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-165-59:1557271:1557726 [5] NCCL INFO Connected all trees
+ip-26-0-165-164:1227469:1227920 [6] NCCL INFO Channel 23/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-59:1557271:1557726 [5] NCCL INFO NVLS comm 0x871d350 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-178:2815965:2816435 [2] NCCL INFO Connected all trees
+ip-26-0-161-178:2815965:2816435 [2] NCCL INFO NVLS comm 0x8e44410 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-131:1630209:1630693 [4] NCCL INFO Channel 20/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-168-52:2722052:2722506 [7] NCCL INFO Connected all trees
+ip-26-0-168-52:2722052:2722506 [7] NCCL INFO NVLS comm 0x98d7670 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-178:2815966:2816439 [3] NCCL INFO Connected all trees
+ip-26-0-161-178:2815966:2816439 [3] NCCL INFO NVLS comm 0x9715940 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-131:1630208:1630691 [3] NCCL INFO Channel 18/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-165-131:1630209:1630693 [4] NCCL INFO Channel 21/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-165-131:1630208:1630691 [3] NCCL INFO Channel 19/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-178:2815967:2816438 [4] NCCL INFO Connected all trees
+ip-26-0-161-178:2815967:2816438 [4] NCCL INFO NVLS comm 0x888d050 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-131:1630209:1630693 [4] NCCL INFO Channel 22/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-165-131:1630208:1630691 [3] NCCL INFO Channel 20/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-165-131:1630205:1630687 [0] NCCL INFO Connected all trees
+ip-26-0-165-131:1630205:1630687 [0] NCCL INFO NVLS comm 0x90d2dc0 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-153:2636084:2636546 [0] NCCL INFO Connected all trees
+ip-26-0-165-202:153569:154020 [7] NCCL INFO Connected all trees
+ip-26-0-165-202:153569:154020 [7] NCCL INFO NVLS comm 0x898b390 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-153:2636084:2636546 [0] NCCL INFO NVLS comm 0x91aa590 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-131:1630209:1630693 [4] NCCL INFO Channel 23/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-165-131:1630208:1630691 [3] NCCL INFO Channel 21/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-165-131:1630208:1630691 [3] NCCL INFO Channel 22/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-165-131:1630208:1630691 [3] NCCL INFO Channel 23/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-168-52:2722047:2722507 [2] NCCL INFO Connected all trees
+ip-26-0-168-52:2722047:2722507 [2] NCCL INFO NVLS comm 0x96845c0 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-242:1364903:1365371 [1] NCCL INFO Connected all trees
+ip-26-0-160-242:1364903:1365371 [1] NCCL INFO NVLS comm 0x9d45a90 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-103:96888:97344 [0] NCCL INFO Connected all trees
+ip-26-0-161-103:96888:97344 [0] NCCL INFO NVLS comm 0x90a1500 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-123:32709:33164 [5] NCCL INFO Connected NVLS tree
+ip-26-0-161-123:32709:33164 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-123:32709:33164 [5] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-161-123:32709:33164 [5] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-161-123:32711:33165 [7] NCCL INFO Connected NVLS tree
+ip-26-0-161-123:32711:33165 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-123:32711:33165 [7] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-161-123:32711:33165 [7] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-161-123:32707:33163 [3] NCCL INFO Connected NVLS tree
+ip-26-0-161-123:32707:33163 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-123:32707:33163 [3] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-161-123:32707:33163 [3] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-161-123:32710:33160 [6] NCCL INFO Connected NVLS tree
+ip-26-0-161-123:32710:33160 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-123:32710:33160 [6] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-161-123:32705:33166 [1] NCCL INFO Connected NVLS tree
+ip-26-0-161-123:32710:33160 [6] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-161-123:32705:33166 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-123:32705:33166 [1] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-161-123:32705:33166 [1] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-161-123:32708:33162 [4] NCCL INFO Connected NVLS tree
+ip-26-0-161-123:32708:33162 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-123:32708:33162 [4] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-161-123:32708:33162 [4] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-161-123:32706:33161 [2] NCCL INFO Connected NVLS tree
+ip-26-0-161-123:32706:33161 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-123:32706:33161 [2] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-161-123:32706:33161 [2] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-161-123:32704:33159 [0] NCCL INFO Connected NVLS tree
+ip-26-0-161-123:32704:33159 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-123:32704:33159 [0] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-161-123:32704:33159 [0] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-165-59:1557271:1557726 [5] NCCL INFO Connected NVLS tree
+ip-26-0-165-59:1557271:1557726 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-59:1557271:1557726 [5] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-165-59:1557271:1557726 [5] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-165-59:1557269:1557724 [3] NCCL INFO Connected NVLS tree
+ip-26-0-165-59:1557269:1557724 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-59:1557269:1557724 [3] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-165-59:1557269:1557724 [3] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-165-59:1557273:1557723 [7] NCCL INFO Connected NVLS tree
+ip-26-0-165-59:1557273:1557723 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-59:1557273:1557723 [7] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-165-59:1557273:1557723 [7] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-165-59:1557267:1557727 [1] NCCL INFO Connected NVLS tree
+ip-26-0-165-59:1557267:1557727 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-59:1557267:1557727 [1] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-165-59:1557267:1557727 [1] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-165-59:1557272:1557722 [6] NCCL INFO Connected NVLS tree
+ip-26-0-165-59:1557272:1557722 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-59:1557272:1557722 [6] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-165-59:1557272:1557722 [6] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-161-138:96567:97050 [4] NCCL INFO Connected all trees
+ip-26-0-165-59:1557268:1557728 [2] NCCL INFO Connected NVLS tree
+ip-26-0-165-59:1557270:1557725 [4] NCCL INFO Connected NVLS tree
+ip-26-0-165-59:1557268:1557728 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-59:1557268:1557728 [2] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-165-59:1557268:1557728 [2] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-165-59:1557270:1557725 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-59:1557270:1557725 [4] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-165-59:1557270:1557725 [4] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-165-59:1557266:1557721 [0] NCCL INFO Connected NVLS tree
+ip-26-0-165-59:1557266:1557721 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-59:1557266:1557721 [0] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-165-59:1557266:1557721 [0] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-161-153:2636091:2636548 [7] NCCL INFO Connected all trees
+ip-26-0-161-138:96567:97050 [4] NCCL INFO NVLS comm 0x8adfa60 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-153:2636091:2636548 [7] NCCL INFO NVLS comm 0x9275a00 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-202:153562:154017 [0] NCCL INFO Connected all trees
+ip-26-0-165-202:153562:154017 [0] NCCL INFO NVLS comm 0x96f2310 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-142:62884:63339 [3] NCCL INFO Connected all trees
+ip-26-0-161-78:114512:114968 [0] NCCL INFO Connected all trees
+ip-26-0-161-78:114512:114968 [0] NCCL INFO NVLS comm 0x94f6090 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-142:62884:63339 [3] NCCL INFO NVLS comm 0x982c860 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-168-30:235655:236104 [7] NCCL INFO Connected all trees
+ip-26-0-168-30:235655:236104 [7] NCCL INFO NVLS comm 0x96bac20 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-168-30:235648:236102 [0] NCCL INFO Connected all trees
+ip-26-0-168-30:235648:236102 [0] NCCL INFO NVLS comm 0x8b68a40 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-123:32709:33164 [5] NCCL INFO comm 0xa4a44d0 rank 5 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0x738064f7309df2d9 - Init COMPLETE
+ip-26-0-161-123:32705:33166 [1] NCCL INFO comm 0x9c5f0d0 rank 1 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0x738064f7309df2d9 - Init COMPLETE
+ip-26-0-161-123:32707:33163 [3] NCCL INFO comm 0x8aede90 rank 3 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0x738064f7309df2d9 - Init COMPLETE
+ip-26-0-161-123:32711:33165 [7] NCCL INFO comm 0xa0ea330 rank 7 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0x738064f7309df2d9 - Init COMPLETE
+ip-26-0-161-123:32706:33161 [2] NCCL INFO comm 0x9c86c40 rank 2 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0x738064f7309df2d9 - Init COMPLETE
+ip-26-0-161-123:32710:33160 [6] NCCL INFO comm 0x9b7e830 rank 6 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0x738064f7309df2d9 - Init COMPLETE
+ip-26-0-161-123:32704:33159 [0] NCCL INFO comm 0xa6c2260 rank 0 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0x738064f7309df2d9 - Init COMPLETE
+ip-26-0-161-123:32708:33162 [4] NCCL INFO comm 0xa4ddf70 rank 4 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0x738064f7309df2d9 - Init COMPLETE
+ip-26-0-168-34:272152:272610 [2] NCCL INFO Connected all trees
+ip-26-0-168-34:272152:272610 [2] NCCL INFO NVLS comm 0xa6db2e0 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-168-34:272151:272611 [1] NCCL INFO Connected all trees
+ip-26-0-168-34:272151:272611 [1] NCCL INFO NVLS comm 0x9709190 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-138:96569:97049 [6] NCCL INFO Connected all trees
+ip-26-0-161-138:96569:97049 [6] NCCL INFO NVLS comm 0x90f9380 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-59:1557271:1557726 [5] NCCL INFO comm 0x871d350 rank 5 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0x1bf22b79fd53b3bb - Init COMPLETE
+ip-26-0-165-59:1557267:1557727 [1] NCCL INFO comm 0x8aff890 rank 1 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0x1bf22b79fd53b3bb - Init COMPLETE
+ip-26-0-165-59:1557269:1557724 [3] NCCL INFO comm 0xa155f80 rank 3 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0x1bf22b79fd53b3bb - Init COMPLETE
+ip-26-0-165-59:1557273:1557723 [7] NCCL INFO comm 0xa268900 rank 7 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0x1bf22b79fd53b3bb - Init COMPLETE
+ip-26-0-165-59:1557272:1557722 [6] NCCL INFO comm 0x966b0a0 rank 6 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0x1bf22b79fd53b3bb - Init COMPLETE
+ip-26-0-165-59:1557266:1557721 [0] NCCL INFO comm 0xa38bfc0 rank 0 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0x1bf22b79fd53b3bb - Init COMPLETE
+ip-26-0-165-59:1557268:1557728 [2] NCCL INFO comm 0x8db6590 rank 2 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0x1bf22b79fd53b3bb - Init COMPLETE
+ip-26-0-165-59:1557270:1557725 [4] NCCL INFO comm 0x8b8b740 rank 4 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0x1bf22b79fd53b3bb - Init COMPLETE
+ip-26-0-161-178:2815970:2816437 [7] NCCL INFO Connected NVLS tree
+ip-26-0-161-178:2815970:2816437 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-178:2815970:2816437 [7] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-161-178:2815970:2816437 [7] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-161-178:2815968:2816434 [5] NCCL INFO Connected NVLS tree
+ip-26-0-161-178:2815966:2816439 [3] NCCL INFO Connected NVLS tree
+ip-26-0-161-178:2815968:2816434 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-178:2815968:2816434 [5] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-161-178:2815966:2816439 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-178:2815968:2816434 [5] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-161-178:2815966:2816439 [3] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-161-178:2815966:2816439 [3] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-161-178:2815965:2816435 [2] NCCL INFO Connected NVLS tree
+ip-26-0-161-178:2815965:2816435 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-178:2815965:2816435 [2] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-161-178:2815965:2816435 [2] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-161-178:2815964:2816440 [1] NCCL INFO Connected NVLS tree
+ip-26-0-161-178:2815964:2816440 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-178:2815964:2816440 [1] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-161-178:2815967:2816438 [4] NCCL INFO Connected NVLS tree
+ip-26-0-161-178:2815964:2816440 [1] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-161-178:2815967:2816438 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-178:2815967:2816438 [4] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-161-178:2815967:2816438 [4] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-161-178:2815963:2816433 [0] NCCL INFO Connected NVLS tree
+ip-26-0-161-178:2815963:2816433 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-178:2815963:2816433 [0] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-161-178:2815963:2816433 [0] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-161-178:2815969:2816436 [6] NCCL INFO Connected NVLS tree
+ip-26-0-161-178:2815969:2816436 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-178:2815969:2816436 [6] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-161-178:2815969:2816436 [6] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-161-138:96568:97048 [5] NCCL INFO Connected all trees
+ip-26-0-160-242:1364908:1365368 [6] NCCL INFO Connected all trees
+ip-26-0-161-138:96568:97048 [5] NCCL INFO NVLS comm 0x8b4ae50 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-242:1364908:1365368 [6] NCCL INFO NVLS comm 0xa611850 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-225:124160:124620 [0] NCCL INFO Connected all trees
+ip-26-0-160-225:124160:124620 [0] NCCL INFO NVLS comm 0x94b61c0 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-138:96566:97047 [3] NCCL INFO Connected all trees
+ip-26-0-161-138:96566:97047 [3] NCCL INFO NVLS comm 0x9589d40 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-168-52:2722049:2722504 [4] NCCL INFO Connected all trees
+ip-26-0-168-52:2722049:2722504 [4] NCCL INFO NVLS comm 0x8a7b070 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-242:1364907:1365367 [5] NCCL INFO Connected all trees
+ip-26-0-160-242:1364907:1365367 [5] NCCL INFO NVLS comm 0x88ef400 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-131:1630212:1630690 [7] NCCL INFO Connected all trees
+ip-26-0-165-131:1630212:1630690 [7] NCCL INFO NVLS comm 0x8930b50 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-78:114519:114970 [7] NCCL INFO Connected all trees
+ip-26-0-161-78:114519:114970 [7] NCCL INFO NVLS comm 0x8721f20 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-168-34:272156:272607 [6] NCCL INFO Connected all trees
+ip-26-0-168-34:272156:272607 [6] NCCL INFO NVLS comm 0xa3fa270 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-178:2815970:2816437 [7] NCCL INFO comm 0x9335470 rank 7 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0x6a396e45aa98de50 - Init COMPLETE
+ip-26-0-161-178:2815966:2816439 [3] NCCL INFO comm 0x9715940 rank 3 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0x6a396e45aa98de50 - Init COMPLETE
+ip-26-0-161-178:2815968:2816434 [5] NCCL INFO comm 0xa2fead0 rank 5 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0x6a396e45aa98de50 - Init COMPLETE
+ip-26-0-161-178:2815964:2816440 [1] NCCL INFO comm 0x89f39e0 rank 1 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0x6a396e45aa98de50 - Init COMPLETE
+ip-26-0-161-178:2815965:2816435 [2] NCCL INFO comm 0x8e44410 rank 2 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0x6a396e45aa98de50 - Init COMPLETE
+ip-26-0-161-178:2815969:2816436 [6] NCCL INFO comm 0x9cc60a0 rank 6 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0x6a396e45aa98de50 - Init COMPLETE
+ip-26-0-161-178:2815967:2816438 [4] NCCL INFO comm 0x888d050 rank 4 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0x6a396e45aa98de50 - Init COMPLETE
+ip-26-0-161-178:2815963:2816433 [0] NCCL INFO comm 0x95efcc0 rank 0 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0x6a396e45aa98de50 - Init COMPLETE
+ip-26-0-168-52:2722048:2722505 [3] NCCL INFO Connected all trees
+ip-26-0-168-52:2722048:2722505 [3] NCCL INFO NVLS comm 0x88af800 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-168-34:272154:272608 [4] NCCL INFO Connected all trees
+ip-26-0-168-34:272154:272608 [4] NCCL INFO NVLS comm 0x94dc980 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-142:62886:63341 [5] NCCL INFO Connected all trees
+ip-26-0-161-142:62886:63341 [5] NCCL INFO NVLS comm 0xa05d130 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-168-34:272153:272612 [3] NCCL INFO Connected all trees
+ip-26-0-168-34:272153:272612 [3] NCCL INFO NVLS comm 0x88144f0 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-142:62882:63343 [1] NCCL INFO Connected all trees
+ip-26-0-161-142:62882:63343 [1] NCCL INFO NVLS comm 0x9135480 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-168-34:272155:272609 [5] NCCL INFO Connected all trees
+ip-26-0-168-34:272155:272609 [5] NCCL INFO NVLS comm 0x9418790 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-142:62883:63338 [2] NCCL INFO Connected all trees
+ip-26-0-160-225:124166:124622 [6] NCCL INFO Connected all trees
+ip-26-0-161-142:62883:63338 [2] NCCL INFO NVLS comm 0x8d9b840 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-225:124166:124622 [6] NCCL INFO NVLS comm 0x99cf4f0 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-242:1364904:1365369 [2] NCCL INFO Connected all trees
+ip-26-0-160-242:1364904:1365369 [2] NCCL INFO NVLS comm 0x98a3450 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-242:1364905:1365372 [3] NCCL INFO Connected all trees
+ip-26-0-168-52:2722046:2722509 [1] NCCL INFO Connected all trees
+ip-26-0-160-242:1364905:1365372 [3] NCCL INFO NVLS comm 0x9a91f10 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-168-52:2722046:2722509 [1] NCCL INFO NVLS comm 0x9a62ea0 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-242:1364906:1365370 [4] NCCL INFO Connected all trees
+ip-26-0-168-52:2722051:2722503 [6] NCCL INFO Connected all trees
+ip-26-0-160-242:1364906:1365370 [4] NCCL INFO NVLS comm 0xa1f5e10 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-168-52:2722051:2722503 [6] NCCL INFO NVLS comm 0x94c7540 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-142:62885:63342 [4] NCCL INFO Connected all trees
+ip-26-0-161-142:62885:63342 [4] NCCL INFO NVLS comm 0xa08eb80 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-142:62887:63344 [6] NCCL INFO Connected all trees
+ip-26-0-161-142:62887:63344 [6] NCCL INFO NVLS comm 0x88e06e0 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-168-52:2722050:2722508 [5] NCCL INFO Connected all trees
+ip-26-0-168-52:2722050:2722508 [5] NCCL INFO NVLS comm 0x88f2340 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-138:96568:97048 [5] NCCL INFO Connected NVLS tree
+ip-26-0-161-138:96568:97048 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-138:96568:97048 [5] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-161-138:96568:97048 [5] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-161-138:96566:97047 [3] NCCL INFO Connected NVLS tree
+ip-26-0-161-138:96566:97047 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-138:96566:97047 [3] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-161-138:96566:97047 [3] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-161-138:96565:97044 [2] NCCL INFO Connected NVLS tree
+ip-26-0-161-138:96565:97044 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-138:96565:97044 [2] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-161-138:96565:97044 [2] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-161-138:96567:97050 [4] NCCL INFO Connected NVLS tree
+ip-26-0-161-138:96569:97049 [6] NCCL INFO Connected NVLS tree
+ip-26-0-161-138:96567:97050 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-138:96569:97049 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-138:96567:97050 [4] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-161-138:96567:97050 [4] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-161-138:96569:97049 [6] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-161-138:96569:97049 [6] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-161-138:96570:97051 [7] NCCL INFO Connected NVLS tree
+ip-26-0-161-138:96570:97051 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-138:96570:97051 [7] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-161-138:96570:97051 [7] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-161-138:96563:97045 [0] NCCL INFO Connected NVLS tree
+ip-26-0-161-138:96563:97045 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-138:96564:97046 [1] NCCL INFO Connected NVLS tree
+ip-26-0-161-138:96563:97045 [0] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-161-138:96563:97045 [0] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-161-138:96564:97046 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-138:96564:97046 [1] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-161-138:96564:97046 [1] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-161-153:2636087:2636550 [3] NCCL INFO Connected all trees
+ip-26-0-161-153:2636087:2636550 [3] NCCL INFO NVLS comm 0x8b16c80 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-138:96565:97044 [2] NCCL INFO comm 0xa46bac0 rank 2 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0xe65dda020e3348b0 - Init COMPLETE
+ip-26-0-161-138:96569:97049 [6] NCCL INFO comm 0x90f9380 rank 6 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0xe65dda020e3348b0 - Init COMPLETE
+ip-26-0-161-138:96563:97045 [0] NCCL INFO comm 0x9c16200 rank 0 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0xe65dda020e3348b0 - Init COMPLETE
+ip-26-0-161-138:96567:97050 [4] NCCL INFO comm 0x8adfa60 rank 4 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0xe65dda020e3348b0 - Init COMPLETE
+ip-26-0-161-138:96568:97048 [5] NCCL INFO comm 0x8b4ae50 rank 5 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0xe65dda020e3348b0 - Init COMPLETE
+ip-26-0-161-138:96564:97046 [1] NCCL INFO comm 0x9897100 rank 1 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0xe65dda020e3348b0 - Init COMPLETE
+ip-26-0-161-138:96566:97047 [3] NCCL INFO comm 0x9589d40 rank 3 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0xe65dda020e3348b0 - Init COMPLETE
+ip-26-0-161-138:96570:97051 [7] NCCL INFO comm 0xa334e80 rank 7 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0xe65dda020e3348b0 - Init COMPLETE
+ip-26-0-161-153:2636088:2636549 [4] NCCL INFO Connected all trees
+ip-26-0-161-153:2636088:2636549 [4] NCCL INFO NVLS comm 0x9514760 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-103:96895:97351 [7] NCCL INFO Connected all trees
+ip-26-0-161-103:96895:97351 [7] NCCL INFO NVLS comm 0xa694e10 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-131:1630206:1630692 [1] NCCL INFO Connected all trees
+ip-26-0-165-131:1630206:1630692 [1] NCCL INFO NVLS comm 0x9d97500 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-168-34:272156:272607 [6] NCCL INFO Connected NVLS tree
+ip-26-0-168-34:272156:272607 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-34:272156:272607 [6] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-168-34:272156:272607 [6] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-161-153:2636086:2636552 [2] NCCL INFO Connected all trees
+ip-26-0-168-34:272155:272609 [5] NCCL INFO Connected NVLS tree
+ip-26-0-168-34:272155:272609 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-34:272155:272609 [5] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-168-34:272155:272609 [5] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-168-34:272154:272608 [4] NCCL INFO Connected NVLS tree
+ip-26-0-168-34:272154:272608 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-34:272154:272608 [4] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-168-34:272154:272608 [4] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-168-34:272151:272611 [1] NCCL INFO Connected NVLS tree
+ip-26-0-168-34:272151:272611 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-34:272151:272611 [1] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-168-34:272151:272611 [1] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-168-34:272152:272610 [2] NCCL INFO Connected NVLS tree
+ip-26-0-168-34:272152:272610 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-34:272152:272610 [2] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-168-34:272152:272610 [2] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-168-34:272157:272613 [7] NCCL INFO Connected NVLS tree
+ip-26-0-168-34:272153:272612 [3] NCCL INFO Connected NVLS tree
+ip-26-0-168-34:272157:272613 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-34:272157:272613 [7] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-168-34:272157:272613 [7] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-168-34:272153:272612 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-34:272153:272612 [3] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-168-34:272153:272612 [3] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-168-34:272150:272606 [0] NCCL INFO Connected NVLS tree
+ip-26-0-168-34:272150:272606 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-34:272150:272606 [0] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-168-34:272150:272606 [0] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-161-153:2636086:2636552 [2] NCCL INFO NVLS comm 0x9bc2e10 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-168-30:235652:236105 [4] NCCL INFO Connected all trees
+ip-26-0-168-30:235652:236105 [4] NCCL INFO NVLS comm 0x9ee4a80 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-225:124162:124627 [2] NCCL INFO Connected all trees
+ip-26-0-160-225:124162:124627 [2] NCCL INFO NVLS comm 0xa51c750 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-168-30:235653:236103 [5] NCCL INFO Connected all trees
+ip-26-0-168-30:235653:236103 [5] NCCL INFO NVLS comm 0x93640f0 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-242:1364907:1365367 [5] NCCL INFO Connected NVLS tree
+ip-26-0-160-242:1364907:1365367 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-242:1364907:1365367 [5] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-160-242:1364907:1365367 [5] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-160-242:1364903:1365371 [1] NCCL INFO Connected NVLS tree
+ip-26-0-160-242:1364909:1365373 [7] NCCL INFO Connected NVLS tree
+ip-26-0-160-242:1364909:1365373 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-242:1364903:1365371 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-242:1364909:1365373 [7] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-160-242:1364903:1365371 [1] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-160-242:1364909:1365373 [7] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-160-242:1364903:1365371 [1] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-160-242:1364905:1365372 [3] NCCL INFO Connected NVLS tree
+ip-26-0-160-242:1364905:1365372 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-242:1364905:1365372 [3] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-160-242:1364905:1365372 [3] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-160-242:1364904:1365369 [2] NCCL INFO Connected NVLS tree
+ip-26-0-160-242:1364904:1365369 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-242:1364904:1365369 [2] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-160-242:1364904:1365369 [2] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-160-242:1364908:1365368 [6] NCCL INFO Connected NVLS tree
+ip-26-0-160-242:1364908:1365368 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-242:1364906:1365370 [4] NCCL INFO Connected NVLS tree
+ip-26-0-160-242:1364908:1365368 [6] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-160-242:1364908:1365368 [6] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-160-242:1364906:1365370 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-30:235651:236108 [3] NCCL INFO Connected all trees
+ip-26-0-160-242:1364906:1365370 [4] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-160-242:1364906:1365370 [4] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-160-242:1364902:1365366 [0] NCCL INFO Connected NVLS tree
+ip-26-0-160-242:1364902:1365366 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-242:1364902:1365366 [0] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-160-242:1364902:1365366 [0] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-168-30:235651:236108 [3] NCCL INFO NVLS comm 0x9147400 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-153:2636090:2636547 [6] NCCL INFO Connected all trees
+ip-26-0-161-153:2636090:2636547 [6] NCCL INFO NVLS comm 0x9c7bbd0 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-164:1227470:1227923 [7] NCCL INFO Connected all trees
+ip-26-0-165-164:1227470:1227923 [7] NCCL INFO NVLS comm 0x94f7a70 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-153:2636085:2636553 [1] NCCL INFO Connected all trees
+ip-26-0-161-153:2636085:2636553 [1] NCCL INFO NVLS comm 0xa427240 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-168-34:272154:272608 [4] NCCL INFO comm 0x94dc980 rank 4 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0x84900e3ce4980498 - Init COMPLETE
+ip-26-0-168-34:272156:272607 [6] NCCL INFO comm 0xa3fa270 rank 6 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0x84900e3ce4980498 - Init COMPLETE
+ip-26-0-168-34:272152:272610 [2] NCCL INFO comm 0xa6db2e0 rank 2 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0x84900e3ce4980498 - Init COMPLETE
+ip-26-0-168-34:272150:272606 [0] NCCL INFO comm 0x9fd74c0 rank 0 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0x84900e3ce4980498 - Init COMPLETE
+ip-26-0-168-34:272155:272609 [5] NCCL INFO comm 0x9418790 rank 5 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0x84900e3ce4980498 - Init COMPLETE
+ip-26-0-168-34:272151:272611 [1] NCCL INFO comm 0x9709190 rank 1 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0x84900e3ce4980498 - Init COMPLETE
+ip-26-0-168-34:272153:272612 [3] NCCL INFO comm 0x88144f0 rank 3 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0x84900e3ce4980498 - Init COMPLETE
+ip-26-0-168-34:272157:272613 [7] NCCL INFO comm 0x9663cf0 rank 7 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0x84900e3ce4980498 - Init COMPLETE
+ip-26-0-165-202:153564:154018 [2] NCCL INFO Connected all trees
+ip-26-0-165-202:153564:154018 [2] NCCL INFO NVLS comm 0x8a1eea0 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-168-30:235654:236107 [6] NCCL INFO Connected all trees
+ip-26-0-168-52:2722046:2722509 [1] NCCL INFO Connected NVLS tree
+ip-26-0-168-52:2722046:2722509 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-52:2722048:2722505 [3] NCCL INFO Connected NVLS tree
+ip-26-0-168-52:2722046:2722509 [1] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-168-52:2722046:2722509 [1] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-168-52:2722048:2722505 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-52:2722048:2722505 [3] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-168-52:2722048:2722505 [3] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-168-52:2722050:2722508 [5] NCCL INFO Connected NVLS tree
+ip-26-0-168-52:2722050:2722508 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-52:2722050:2722508 [5] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-168-52:2722050:2722508 [5] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-168-52:2722051:2722503 [6] NCCL INFO Connected NVLS tree
+ip-26-0-168-52:2722051:2722503 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-52:2722051:2722503 [6] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-168-30:235654:236107 [6] NCCL INFO NVLS comm 0x9bb0ad0 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-168-52:2722051:2722503 [6] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-168-52:2722047:2722507 [2] NCCL INFO Connected NVLS tree
+ip-26-0-168-52:2722047:2722507 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-52:2722047:2722507 [2] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-168-52:2722047:2722507 [2] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-168-52:2722049:2722504 [4] NCCL INFO Connected NVLS tree
+ip-26-0-168-52:2722049:2722504 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-52:2722049:2722504 [4] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-168-52:2722049:2722504 [4] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-168-52:2722052:2722506 [7] NCCL INFO Connected NVLS tree
+ip-26-0-168-52:2722052:2722506 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-52:2722052:2722506 [7] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-168-52:2722052:2722506 [7] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-168-52:2722045:2722502 [0] NCCL INFO Connected NVLS tree
+ip-26-0-168-52:2722045:2722502 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-52:2722045:2722502 [0] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-168-52:2722045:2722502 [0] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-161-153:2636089:2636551 [5] NCCL INFO Connected all trees
+ip-26-0-161-153:2636089:2636551 [5] NCCL INFO NVLS comm 0x8990aa0 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-242:1364908:1365368 [6] NCCL INFO comm 0xa611850 rank 6 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0x7f823661d2a228a2 - Init COMPLETE
+ip-26-0-160-242:1364902:1365366 [0] NCCL INFO comm 0x9a12810 rank 0 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0x7f823661d2a228a2 - Init COMPLETE
+ip-26-0-160-242:1364904:1365369 [2] NCCL INFO comm 0x98a3450 rank 2 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0x7f823661d2a228a2 - Init COMPLETE
+ip-26-0-160-242:1364903:1365371 [1] NCCL INFO comm 0x9d45a90 rank 1 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0x7f823661d2a228a2 - Init COMPLETE
+ip-26-0-160-242:1364909:1365373 [7] NCCL INFO comm 0xa0d8c00 rank 7 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0x7f823661d2a228a2 - Init COMPLETE
+ip-26-0-160-242:1364907:1365367 [5] NCCL INFO comm 0x88ef400 rank 5 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0x7f823661d2a228a2 - Init COMPLETE
+ip-26-0-160-242:1364906:1365370 [4] NCCL INFO comm 0xa1f5e10 rank 4 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0x7f823661d2a228a2 - Init COMPLETE
+ip-26-0-160-242:1364905:1365372 [3] NCCL INFO comm 0x9a91f10 rank 3 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0x7f823661d2a228a2 - Init COMPLETE
+ip-26-0-160-225:124165:124625 [5] NCCL INFO Connected all trees
+ip-26-0-160-225:124165:124625 [5] NCCL INFO NVLS comm 0x8aa9a50 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-168-52:2722050:2722508 [5] NCCL INFO comm 0x88f2340 rank 5 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0xd177a32f3dbed555 - Init COMPLETE
+ip-26-0-168-52:2722052:2722506 [7] NCCL INFO comm 0x98d7670 rank 7 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0xd177a32f3dbed555 - Init COMPLETE
+ip-26-0-168-52:2722046:2722509 [1] NCCL INFO comm 0x9a62ea0 rank 1 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0xd177a32f3dbed555 - Init COMPLETE
+ip-26-0-168-52:2722048:2722505 [3] NCCL INFO comm 0x88af800 rank 3 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0xd177a32f3dbed555 - Init COMPLETE
+ip-26-0-168-52:2722045:2722502 [0] NCCL INFO comm 0x8798320 rank 0 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0xd177a32f3dbed555 - Init COMPLETE
+ip-26-0-168-52:2722051:2722503 [6] NCCL INFO comm 0x94c7540 rank 6 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0xd177a32f3dbed555 - Init COMPLETE
+ip-26-0-168-52:2722049:2722504 [4] NCCL INFO comm 0x8a7b070 rank 4 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0xd177a32f3dbed555 - Init COMPLETE
+ip-26-0-168-52:2722047:2722507 [2] NCCL INFO comm 0x96845c0 rank 2 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0xd177a32f3dbed555 - Init COMPLETE
+ip-26-0-165-202:153565:154022 [3] NCCL INFO Connected all trees
+ip-26-0-165-202:153565:154022 [3] NCCL INFO NVLS comm 0x9c0d0c0 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-168-30:235649:236109 [1] NCCL INFO Connected all trees
+ip-26-0-168-30:235649:236109 [1] NCCL INFO NVLS comm 0x9e4d2f0 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-168-30:235650:236106 [2] NCCL INFO Connected all trees
+ip-26-0-165-202:153568:154019 [6] NCCL INFO Connected all trees
+ip-26-0-168-30:235650:236106 [2] NCCL INFO NVLS comm 0x9168a20 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-202:153568:154019 [6] NCCL INFO NVLS comm 0x8a11410 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-202:153567:154023 [5] NCCL INFO Connected all trees
+ip-26-0-160-225:124161:124621 [1] NCCL INFO Connected all trees
+ip-26-0-160-225:124161:124621 [1] NCCL INFO NVLS comm 0x8a5caa0 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-202:153567:154023 [5] NCCL INFO NVLS comm 0xa426920 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-225:124163:124626 [3] NCCL INFO Connected all trees
+ip-26-0-160-225:124163:124626 [3] NCCL INFO NVLS comm 0x972ae70 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-202:153566:154021 [4] NCCL INFO Connected all trees
+ip-26-0-165-202:153566:154021 [4] NCCL INFO NVLS comm 0x8c7b650 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-225:124164:124623 [4] NCCL INFO Connected all trees
+ip-26-0-160-225:124164:124623 [4] NCCL INFO NVLS comm 0x91e0850 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-103:96889:97348 [1] NCCL INFO Connected all trees
+ip-26-0-161-103:96889:97348 [1] NCCL INFO NVLS comm 0x8a48ee0 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-202:153563:154024 [1] NCCL INFO Connected all trees
+ip-26-0-165-202:153563:154024 [1] NCCL INFO NVLS comm 0xa62e5f0 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-142:62886:63341 [5] NCCL INFO Connected NVLS tree
+ip-26-0-161-142:62886:63341 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-142:62886:63341 [5] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-161-142:62886:63341 [5] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-161-142:62888:63340 [7] NCCL INFO Connected NVLS tree
+ip-26-0-161-142:62888:63340 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-142:62888:63340 [7] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-161-142:62888:63340 [7] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-161-142:62883:63338 [2] NCCL INFO Connected NVLS tree
+ip-26-0-161-142:62884:63339 [3] NCCL INFO Connected NVLS tree
+ip-26-0-161-142:62883:63338 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-142:62883:63338 [2] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-161-142:62883:63338 [2] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-161-142:62884:63339 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-142:62887:63344 [6] NCCL INFO Connected NVLS tree
+ip-26-0-161-142:62884:63339 [3] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-161-142:62884:63339 [3] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-161-142:62887:63344 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-142:62887:63344 [6] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-161-142:62887:63344 [6] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-161-142:62885:63342 [4] NCCL INFO Connected NVLS tree
+ip-26-0-161-142:62885:63342 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-142:62885:63342 [4] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-161-142:62885:63342 [4] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-161-142:62882:63343 [1] NCCL INFO Connected NVLS tree
+ip-26-0-161-142:62882:63343 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-142:62882:63343 [1] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-161-142:62882:63343 [1] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-161-142:62881:63337 [0] NCCL INFO Connected NVLS tree
+ip-26-0-161-142:62881:63337 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-142:62881:63337 [0] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-161-142:62881:63337 [0] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-161-78:114514:114971 [2] NCCL INFO Connected all trees
+ip-26-0-161-78:114514:114971 [2] NCCL INFO NVLS comm 0x9f67ee0 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-131:1630211:1630688 [6] NCCL INFO Connected all trees
+ip-26-0-165-131:1630211:1630688 [6] NCCL INFO NVLS comm 0xa6d1fa0 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-164:1227465:1227924 [2] NCCL INFO Connected all trees
+ip-26-0-165-164:1227465:1227924 [2] NCCL INFO NVLS comm 0x8e85d80 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-153:2636090:2636547 [6] NCCL INFO Connected NVLS tree
+ip-26-0-161-153:2636090:2636547 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-153:2636090:2636547 [6] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-161-153:2636090:2636547 [6] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-161-153:2636084:2636546 [0] NCCL INFO Connected NVLS tree
+ip-26-0-161-153:2636084:2636546 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-153:2636084:2636546 [0] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-161-153:2636084:2636546 [0] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-161-153:2636089:2636551 [5] NCCL INFO Connected NVLS tree
+ip-26-0-161-153:2636089:2636551 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-153:2636089:2636551 [5] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-161-153:2636089:2636551 [5] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-161-153:2636088:2636549 [4] NCCL INFO Connected NVLS tree
+ip-26-0-161-153:2636088:2636549 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-153:2636088:2636549 [4] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-161-153:2636088:2636549 [4] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-161-153:2636085:2636553 [1] NCCL INFO Connected NVLS tree
+ip-26-0-161-153:2636085:2636553 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-153:2636086:2636552 [2] NCCL INFO Connected NVLS tree
+ip-26-0-161-153:2636085:2636553 [1] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-161-153:2636085:2636553 [1] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-161-153:2636087:2636550 [3] NCCL INFO Connected NVLS tree
+ip-26-0-161-153:2636086:2636552 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-153:2636087:2636550 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-153:2636086:2636552 [2] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-161-153:2636086:2636552 [2] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-161-153:2636087:2636550 [3] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-161-153:2636087:2636550 [3] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-161-153:2636091:2636548 [7] NCCL INFO Connected NVLS tree
+ip-26-0-161-153:2636091:2636548 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-153:2636091:2636548 [7] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-161-153:2636091:2636548 [7] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-161-103:96891:97347 [3] NCCL INFO Connected all trees
+ip-26-0-161-103:96891:97347 [3] NCCL INFO NVLS comm 0x8eddb20 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-103:96892:97345 [4] NCCL INFO Connected all trees
+ip-26-0-161-103:96892:97345 [4] NCCL INFO NVLS comm 0x9dff040 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-78:114517:114969 [5] NCCL INFO Connected all trees
+ip-26-0-161-78:114517:114969 [5] NCCL INFO NVLS comm 0xa4e6130 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-168-30:235653:236103 [5] NCCL INFO Connected NVLS tree
+ip-26-0-168-30:235653:236103 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-30:235653:236103 [5] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-168-30:235653:236103 [5] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-168-30:235655:236104 [7] NCCL INFO Connected NVLS tree
+ip-26-0-168-30:235655:236104 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-30:235655:236104 [7] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-168-30:235655:236104 [7] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-168-30:235652:236105 [4] NCCL INFO Connected NVLS tree
+ip-26-0-168-30:235652:236105 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-30:235648:236102 [0] NCCL INFO Connected NVLS tree
+ip-26-0-168-30:235652:236105 [4] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-168-30:235652:236105 [4] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-168-30:235651:236108 [3] NCCL INFO Connected NVLS tree
+ip-26-0-168-30:235648:236102 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-30:235651:236108 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-30:235651:236108 [3] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-168-30:235651:236108 [3] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-168-30:235648:236102 [0] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-168-30:235649:236109 [1] NCCL INFO Connected NVLS tree
+ip-26-0-168-30:235648:236102 [0] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-168-30:235649:236109 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-30:235649:236109 [1] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-168-30:235649:236109 [1] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-168-30:235650:236106 [2] NCCL INFO Connected NVLS tree
+ip-26-0-168-30:235650:236106 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-30:235650:236106 [2] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-168-30:235650:236106 [2] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-161-103:96890:97350 [2] NCCL INFO Connected all trees
+ip-26-0-168-30:235654:236107 [6] NCCL INFO Connected NVLS tree
+ip-26-0-168-30:235654:236107 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-30:235654:236107 [6] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-168-30:235654:236107 [6] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-161-103:96890:97350 [2] NCCL INFO NVLS comm 0x907c3f0 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-78:114513:114973 [1] NCCL INFO Connected all trees
+ip-26-0-161-78:114513:114973 [1] NCCL INFO NVLS comm 0xa094be0 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-153:2636091:2636548 [7] NCCL INFO comm 0x9275a00 rank 7 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0x7a5c769d1a9a12a1 - Init COMPLETE
+ip-26-0-161-153:2636089:2636551 [5] NCCL INFO comm 0x8990aa0 rank 5 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0x7a5c769d1a9a12a1 - Init COMPLETE
+ip-26-0-161-153:2636087:2636550 [3] NCCL INFO comm 0x8b16c80 rank 3 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0x7a5c769d1a9a12a1 - Init COMPLETE
+ip-26-0-161-153:2636088:2636549 [4] NCCL INFO comm 0x9514760 rank 4 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0x7a5c769d1a9a12a1 - Init COMPLETE
+ip-26-0-161-153:2636090:2636547 [6] NCCL INFO comm 0x9c7bbd0 rank 6 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0x7a5c769d1a9a12a1 - Init COMPLETE
+ip-26-0-161-153:2636085:2636553 [1] NCCL INFO comm 0xa427240 rank 1 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0x7a5c769d1a9a12a1 - Init COMPLETE
+ip-26-0-161-153:2636084:2636546 [0] NCCL INFO comm 0x91aa590 rank 0 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0x7a5c769d1a9a12a1 - Init COMPLETE
+ip-26-0-161-153:2636086:2636552 [2] NCCL INFO comm 0x9bc2e10 rank 2 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0x7a5c769d1a9a12a1 - Init COMPLETE
+ip-26-0-161-78:114518:114974 [6] NCCL INFO Connected all trees
+ip-26-0-161-78:114518:114974 [6] NCCL INFO NVLS comm 0x9ad7200 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-103:96894:97349 [6] NCCL INFO Connected all trees
+ip-26-0-161-103:96894:97349 [6] NCCL INFO NVLS comm 0x8cf0450 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-225:124165:124625 [5] NCCL INFO Connected NVLS tree
+ip-26-0-160-225:124165:124625 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-225:124165:124625 [5] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-160-225:124165:124625 [5] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-160-225:124167:124624 [7] NCCL INFO Connected NVLS tree
+ip-26-0-160-225:124163:124626 [3] NCCL INFO Connected NVLS tree
+ip-26-0-160-225:124164:124623 [4] NCCL INFO Connected NVLS tree
+ip-26-0-160-225:124164:124623 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-225:124167:124624 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-225:124164:124623 [4] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-160-225:124164:124623 [4] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-160-225:124163:124626 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-225:124167:124624 [7] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-160-225:124167:124624 [7] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-160-225:124163:124626 [3] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-160-225:124163:124626 [3] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-160-225:124160:124620 [0] NCCL INFO Connected NVLS tree
+ip-26-0-160-225:124160:124620 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-225:124161:124621 [1] NCCL INFO Connected NVLS tree
+ip-26-0-160-225:124161:124621 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-225:124161:124621 [1] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-160-225:124160:124620 [0] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-160-225:124161:124621 [1] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-160-225:124160:124620 [0] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-160-225:124166:124622 [6] NCCL INFO Connected NVLS tree
+ip-26-0-160-225:124166:124622 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-225:124166:124622 [6] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-160-225:124166:124622 [6] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-160-225:124162:124627 [2] NCCL INFO Connected NVLS tree
+ip-26-0-160-225:124162:124627 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-225:124162:124627 [2] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-160-225:124162:124627 [2] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-161-78:114515:114972 [3] NCCL INFO Connected all trees
+ip-26-0-161-78:114515:114972 [3] NCCL INFO NVLS comm 0x9042190 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-103:96893:97346 [5] NCCL INFO Connected all trees
+ip-26-0-161-103:96893:97346 [5] NCCL INFO NVLS comm 0x894f3a0 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-131:1630208:1630691 [3] NCCL INFO Connected all trees
+ip-26-0-165-131:1630208:1630691 [3] NCCL INFO NVLS comm 0x873de50 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-78:114516:114975 [4] NCCL INFO Connected all trees
+ip-26-0-161-78:114516:114975 [4] NCCL INFO NVLS comm 0x95def10 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-164:1227464:1227926 [1] NCCL INFO Connected all trees
+ip-26-0-165-164:1227464:1227926 [1] NCCL INFO NVLS comm 0x92d57b0 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-164:1227466:1227921 [3] NCCL INFO Connected all trees
+ip-26-0-165-164:1227466:1227921 [3] NCCL INFO NVLS comm 0x90f2670 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-168-30:235652:236105 [4] NCCL INFO comm 0x9ee4a80 rank 4 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0xd7d26cee33c1eceb - Init COMPLETE
+ip-26-0-168-30:235654:236107 [6] NCCL INFO comm 0x9bb0ad0 rank 6 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0xd7d26cee33c1eceb - Init COMPLETE
+ip-26-0-168-30:235648:236102 [0] NCCL INFO comm 0x8b68a40 rank 0 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0xd7d26cee33c1eceb - Init COMPLETE
+ip-26-0-168-30:235650:236106 [2] NCCL INFO comm 0x9168a20 rank 2 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0xd7d26cee33c1eceb - Init COMPLETE
+ip-26-0-168-30:235653:236103 [5] NCCL INFO comm 0x93640f0 rank 5 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0xd7d26cee33c1eceb - Init COMPLETE
+ip-26-0-168-30:235649:236109 [1] NCCL INFO comm 0x9e4d2f0 rank 1 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0xd7d26cee33c1eceb - Init COMPLETE
+ip-26-0-168-30:235651:236108 [3] NCCL INFO comm 0x9147400 rank 3 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0xd7d26cee33c1eceb - Init COMPLETE
+ip-26-0-168-30:235655:236104 [7] NCCL INFO comm 0x96bac20 rank 7 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0xd7d26cee33c1eceb - Init COMPLETE
+ip-26-0-165-202:153568:154019 [6] NCCL INFO Connected NVLS tree
+ip-26-0-165-202:153568:154019 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-202:153568:154019 [6] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-165-202:153568:154019 [6] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-165-202:153564:154018 [2] NCCL INFO Connected NVLS tree
+ip-26-0-165-202:153564:154018 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-202:153564:154018 [2] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-165-202:153564:154018 [2] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-165-202:153566:154021 [4] NCCL INFO Connected NVLS tree
+ip-26-0-165-202:153566:154021 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-202:153566:154021 [4] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-165-202:153566:154021 [4] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-165-202:153562:154017 [0] NCCL INFO Connected NVLS tree
+ip-26-0-165-202:153562:154017 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-202:153562:154017 [0] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-165-202:153562:154017 [0] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-165-202:153567:154023 [5] NCCL INFO Connected NVLS tree
+ip-26-0-165-202:153567:154023 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-202:153567:154023 [5] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-165-202:153567:154023 [5] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-165-202:153569:154020 [7] NCCL INFO Connected NVLS tree
+ip-26-0-165-202:153569:154020 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-202:153569:154020 [7] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-165-202:153569:154020 [7] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-165-202:153565:154022 [3] NCCL INFO Connected NVLS tree
+ip-26-0-165-202:153565:154022 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-202:153563:154024 [1] NCCL INFO Connected NVLS tree
+ip-26-0-165-202:153565:154022 [3] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-165-202:153565:154022 [3] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-165-202:153563:154024 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-202:153563:154024 [1] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-165-202:153563:154024 [1] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-165-131:1630210:1630689 [5] NCCL INFO Connected all trees
+ip-26-0-165-164:1227467:1227925 [4] NCCL INFO Connected all trees
+ip-26-0-165-131:1630210:1630689 [5] NCCL INFO NVLS comm 0xa1085c0 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-164:1227467:1227925 [4] NCCL INFO NVLS comm 0x94610b0 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-225:124167:124624 [7] NCCL INFO comm 0x9eb7fb0 rank 7 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0xf16fff8ce090ba3b - Init COMPLETE
+ip-26-0-160-225:124161:124621 [1] NCCL INFO comm 0x8a5caa0 rank 1 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0xf16fff8ce090ba3b - Init COMPLETE
+ip-26-0-160-225:124165:124625 [5] NCCL INFO comm 0x8aa9a50 rank 5 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0xf16fff8ce090ba3b - Init COMPLETE
+ip-26-0-160-225:124163:124626 [3] NCCL INFO comm 0x972ae70 rank 3 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0xf16fff8ce090ba3b - Init COMPLETE
+ip-26-0-160-225:124160:124620 [0] NCCL INFO comm 0x94b61c0 rank 0 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0xf16fff8ce090ba3b - Init COMPLETE
+ip-26-0-160-225:124164:124623 [4] NCCL INFO comm 0x91e0850 rank 4 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0xf16fff8ce090ba3b - Init COMPLETE
+ip-26-0-160-225:124166:124622 [6] NCCL INFO comm 0x99cf4f0 rank 6 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0xf16fff8ce090ba3b - Init COMPLETE
+ip-26-0-160-225:124162:124627 [2] NCCL INFO comm 0xa51c750 rank 2 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0xf16fff8ce090ba3b - Init COMPLETE
+ip-26-0-165-131:1630209:1630693 [4] NCCL INFO Connected all trees
+ip-26-0-165-131:1630209:1630693 [4] NCCL INFO NVLS comm 0x9416000 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-131:1630207:1630694 [2] NCCL INFO Connected all trees
+ip-26-0-165-164:1227468:1227922 [5] NCCL INFO Connected all trees
+ip-26-0-165-164:1227468:1227922 [5] NCCL INFO NVLS comm 0x9dec4e0 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-131:1630207:1630694 [2] NCCL INFO NVLS comm 0x97a9230 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-202:153564:154018 [2] NCCL INFO comm 0x8a1eea0 rank 2 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0x74595ea6be2a949 - Init COMPLETE
+ip-26-0-165-202:153568:154019 [6] NCCL INFO comm 0x8a11410 rank 6 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0x74595ea6be2a949 - Init COMPLETE
+ip-26-0-165-202:153562:154017 [0] NCCL INFO comm 0x96f2310 rank 0 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0x74595ea6be2a949 - Init COMPLETE
+ip-26-0-165-202:153563:154024 [1] NCCL INFO comm 0xa62e5f0 rank 1 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0x74595ea6be2a949 - Init COMPLETE
+ip-26-0-165-202:153567:154023 [5] NCCL INFO comm 0xa426920 rank 5 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0x74595ea6be2a949 - Init COMPLETE
+ip-26-0-165-202:153566:154021 [4] NCCL INFO comm 0x8c7b650 rank 4 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0x74595ea6be2a949 - Init COMPLETE
+ip-26-0-165-202:153565:154022 [3] NCCL INFO comm 0x9c0d0c0 rank 3 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0x74595ea6be2a949 - Init COMPLETE
+ip-26-0-165-202:153569:154020 [7] NCCL INFO comm 0x898b390 rank 7 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0x74595ea6be2a949 - Init COMPLETE
+ip-26-0-165-164:1227469:1227920 [6] NCCL INFO Connected all trees
+ip-26-0-165-164:1227469:1227920 [6] NCCL INFO NVLS comm 0x8720770 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-142:62886:63341 [5] NCCL INFO comm 0xa05d130 rank 5 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0x66d95c3778073495 - Init COMPLETE
+ip-26-0-161-142:62882:63343 [1] NCCL INFO comm 0x9135480 rank 1 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0x66d95c3778073495 - Init COMPLETE
+ip-26-0-161-142:62883:63338 [2] NCCL INFO comm 0x8d9b840 rank 2 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0x66d95c3778073495 - Init COMPLETE
+ip-26-0-161-142:62888:63340 [7] NCCL INFO comm 0x93bef90 rank 7 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0x66d95c3778073495 - Init COMPLETE
+ip-26-0-161-142:62884:63339 [3] NCCL INFO comm 0x982c860 rank 3 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0x66d95c3778073495 - Init COMPLETE
+ip-26-0-161-142:62887:63344 [6] NCCL INFO comm 0x88e06e0 rank 6 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0x66d95c3778073495 - Init COMPLETE
+ip-26-0-161-142:62881:63337 [0] NCCL INFO comm 0xa371a00 rank 0 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0x66d95c3778073495 - Init COMPLETE
+ip-26-0-161-142:62885:63342 [4] NCCL INFO comm 0xa08eb80 rank 4 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0x66d95c3778073495 - Init COMPLETE
+ip-26-0-161-103:96892:97345 [4] NCCL INFO Connected NVLS tree
+ip-26-0-161-103:96892:97345 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-103:96892:97345 [4] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-161-103:96892:97345 [4] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-161-103:96894:97349 [6] NCCL INFO Connected NVLS tree
+ip-26-0-161-103:96894:97349 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-103:96894:97349 [6] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-161-103:96894:97349 [6] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-161-103:96891:97347 [3] NCCL INFO Connected NVLS tree
+ip-26-0-161-103:96891:97347 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-103:96891:97347 [3] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-161-103:96891:97347 [3] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-161-103:96890:97350 [2] NCCL INFO Connected NVLS tree
+ip-26-0-161-103:96893:97346 [5] NCCL INFO Connected NVLS tree
+ip-26-0-161-103:96890:97350 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-103:96893:97346 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-103:96890:97350 [2] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-161-103:96890:97350 [2] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-161-103:96893:97346 [5] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-161-103:96893:97346 [5] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-161-103:96889:97348 [1] NCCL INFO Connected NVLS tree
+ip-26-0-161-103:96895:97351 [7] NCCL INFO Connected NVLS tree
+ip-26-0-161-103:96889:97348 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-103:96889:97348 [1] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-161-103:96895:97351 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-103:96889:97348 [1] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-161-103:96895:97351 [7] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-161-103:96895:97351 [7] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-161-78:114519:114970 [7] NCCL INFO Connected NVLS tree
+ip-26-0-161-78:114519:114970 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-78:114519:114970 [7] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-161-78:114519:114970 [7] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-161-78:114516:114975 [4] NCCL INFO Connected NVLS tree
+ip-26-0-161-78:114516:114975 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-78:114516:114975 [4] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-161-78:114516:114975 [4] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-161-78:114517:114969 [5] NCCL INFO Connected NVLS tree
+ip-26-0-161-78:114517:114969 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-78:114517:114969 [5] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-161-78:114517:114969 [5] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-161-78:114513:114973 [1] NCCL INFO Connected NVLS tree
+ip-26-0-161-78:114515:114972 [3] NCCL INFO Connected NVLS tree
+ip-26-0-161-78:114513:114973 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-78:114513:114973 [1] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-161-78:114513:114973 [1] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-161-78:114515:114972 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-78:114515:114972 [3] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-161-78:114515:114972 [3] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-161-78:114514:114971 [2] NCCL INFO Connected NVLS tree
+ip-26-0-161-78:114514:114971 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-78:114514:114971 [2] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-161-78:114514:114971 [2] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-161-78:114518:114974 [6] NCCL INFO Connected NVLS tree
+ip-26-0-161-78:114518:114974 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-78:114512:114968 [0] NCCL INFO Connected NVLS tree
+ip-26-0-161-78:114512:114968 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-78:114518:114974 [6] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-161-78:114518:114974 [6] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-161-78:114512:114968 [0] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-161-78:114512:114968 [0] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-161-78:114519:114970 [7] NCCL INFO comm 0x8721f20 rank 7 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0x2bcb5da8b3cfa2cd - Init COMPLETE
+ip-26-0-161-78:114517:114969 [5] NCCL INFO comm 0xa4e6130 rank 5 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0x2bcb5da8b3cfa2cd - Init COMPLETE
+ip-26-0-161-78:114515:114972 [3] NCCL INFO comm 0x9042190 rank 3 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0x2bcb5da8b3cfa2cd - Init COMPLETE
+ip-26-0-161-78:114518:114974 [6] NCCL INFO comm 0x9ad7200 rank 6 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0x2bcb5da8b3cfa2cd - Init COMPLETE
+ip-26-0-161-78:114513:114973 [1] NCCL INFO comm 0xa094be0 rank 1 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0x2bcb5da8b3cfa2cd - Init COMPLETE
+ip-26-0-161-78:114514:114971 [2] NCCL INFO comm 0x9f67ee0 rank 2 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0x2bcb5da8b3cfa2cd - Init COMPLETE
+ip-26-0-161-78:114516:114975 [4] NCCL INFO comm 0x95def10 rank 4 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0x2bcb5da8b3cfa2cd - Init COMPLETE
+ip-26-0-161-78:114512:114968 [0] NCCL INFO comm 0x94f6090 rank 0 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0x2bcb5da8b3cfa2cd - Init COMPLETE
+ip-26-0-165-131:1630210:1630689 [5] NCCL INFO Connected NVLS tree
+ip-26-0-165-131:1630210:1630689 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-131:1630210:1630689 [5] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-165-131:1630210:1630689 [5] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-165-131:1630206:1630692 [1] NCCL INFO Connected NVLS tree
+ip-26-0-165-131:1630206:1630692 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-131:1630206:1630692 [1] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-165-131:1630206:1630692 [1] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-165-131:1630208:1630691 [3] NCCL INFO Connected NVLS tree
+ip-26-0-165-131:1630208:1630691 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-131:1630208:1630691 [3] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-165-131:1630208:1630691 [3] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-165-131:1630212:1630690 [7] NCCL INFO Connected NVLS tree
+ip-26-0-165-131:1630212:1630690 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-131:1630212:1630690 [7] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-165-131:1630212:1630690 [7] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-165-131:1630205:1630687 [0] NCCL INFO Connected NVLS tree
+ip-26-0-165-131:1630205:1630687 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-131:1630205:1630687 [0] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-165-131:1630205:1630687 [0] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-165-131:1630209:1630693 [4] NCCL INFO Connected NVLS tree
+ip-26-0-165-131:1630209:1630693 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-131:1630209:1630693 [4] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-165-131:1630209:1630693 [4] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-165-131:1630211:1630688 [6] NCCL INFO Connected NVLS tree
+ip-26-0-165-131:1630211:1630688 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-131:1630207:1630694 [2] NCCL INFO Connected NVLS tree
+ip-26-0-165-131:1630211:1630688 [6] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-165-131:1630211:1630688 [6] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-165-131:1630207:1630694 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-131:1630207:1630694 [2] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-165-131:1630207:1630694 [2] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-165-164:1227470:1227923 [7] NCCL INFO Connected NVLS tree
+ip-26-0-165-164:1227470:1227923 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-164:1227470:1227923 [7] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-165-164:1227470:1227923 [7] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-165-164:1227468:1227922 [5] NCCL INFO Connected NVLS tree
+ip-26-0-165-164:1227468:1227922 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-164:1227468:1227922 [5] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-165-164:1227468:1227922 [5] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-165-164:1227469:1227920 [6] NCCL INFO Connected NVLS tree
+ip-26-0-165-164:1227469:1227920 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-164:1227469:1227920 [6] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-165-164:1227469:1227920 [6] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-165-164:1227466:1227921 [3] NCCL INFO Connected NVLS tree
+ip-26-0-165-164:1227466:1227921 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-164:1227466:1227921 [3] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-165-164:1227466:1227921 [3] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-165-164:1227464:1227926 [1] NCCL INFO Connected NVLS tree
+ip-26-0-165-164:1227467:1227925 [4] NCCL INFO Connected NVLS tree
+ip-26-0-165-164:1227464:1227926 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-164:1227464:1227926 [1] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-165-164:1227467:1227925 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-164:1227464:1227926 [1] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-165-164:1227463:1227919 [0] NCCL INFO Connected NVLS tree
+ip-26-0-165-164:1227463:1227919 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-164:1227467:1227925 [4] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-165-164:1227467:1227925 [4] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-165-164:1227463:1227919 [0] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-165-164:1227463:1227919 [0] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-165-164:1227465:1227924 [2] NCCL INFO Connected NVLS tree
+ip-26-0-165-164:1227465:1227924 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-164:1227465:1227924 [2] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-165-164:1227465:1227924 [2] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-161-103:96888:97344 [0] NCCL INFO Connected NVLS tree
+ip-26-0-161-103:96888:97344 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-103:96888:97344 [0] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-161-103:96888:97344 [0] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-161-103:96890:97350 [2] NCCL INFO comm 0x907c3f0 rank 2 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0x6ac0a43191d4b189 - Init COMPLETE
+ip-26-0-161-103:96894:97349 [6] NCCL INFO comm 0x8cf0450 rank 6 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0x6ac0a43191d4b189 - Init COMPLETE
+ip-26-0-161-103:96892:97345 [4] NCCL INFO comm 0x9dff040 rank 4 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0x6ac0a43191d4b189 - Init COMPLETE
+ip-26-0-161-103:96888:97344 [0] NCCL INFO comm 0x90a1500 rank 0 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0x6ac0a43191d4b189 - Init COMPLETE
+ip-26-0-161-103:96889:97348 [1] NCCL INFO comm 0x8a48ee0 rank 1 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0x6ac0a43191d4b189 - Init COMPLETE
+ip-26-0-161-103:96895:97351 [7] NCCL INFO comm 0xa694e10 rank 7 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0x6ac0a43191d4b189 - Init COMPLETE
+ip-26-0-161-103:96893:97346 [5] NCCL INFO comm 0x894f3a0 rank 5 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0x6ac0a43191d4b189 - Init COMPLETE
+ip-26-0-161-103:96891:97347 [3] NCCL INFO comm 0x8eddb20 rank 3 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0x6ac0a43191d4b189 - Init COMPLETE
+ip-26-0-165-131:1630212:1630690 [7] NCCL INFO comm 0x8930b50 rank 7 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0x90a426b68b3ef643 - Init COMPLETE
+ip-26-0-165-131:1630211:1630688 [6] NCCL INFO comm 0xa6d1fa0 rank 6 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0x90a426b68b3ef643 - Init COMPLETE
+ip-26-0-165-131:1630208:1630691 [3] NCCL INFO comm 0x873de50 rank 3 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0x90a426b68b3ef643 - Init COMPLETE
+ip-26-0-165-131:1630207:1630694 [2] NCCL INFO comm 0x97a9230 rank 2 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0x90a426b68b3ef643 - Init COMPLETE
+ip-26-0-165-131:1630205:1630687 [0] NCCL INFO comm 0x90d2dc0 rank 0 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0x90a426b68b3ef643 - Init COMPLETE
+ip-26-0-165-131:1630209:1630693 [4] NCCL INFO comm 0x9416000 rank 4 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0x90a426b68b3ef643 - Init COMPLETE
+ip-26-0-165-131:1630210:1630689 [5] NCCL INFO comm 0xa1085c0 rank 5 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0x90a426b68b3ef643 - Init COMPLETE
+ip-26-0-165-131:1630206:1630692 [1] NCCL INFO comm 0x9d97500 rank 1 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0x90a426b68b3ef643 - Init COMPLETE
+ip-26-0-165-164:1227463:1227919 [0] NCCL INFO comm 0x8a0d780 rank 0 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0xc0a121b95733704f - Init COMPLETE
+ip-26-0-165-164:1227469:1227920 [6] NCCL INFO comm 0x8720770 rank 6 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0xc0a121b95733704f - Init COMPLETE
+ip-26-0-165-164:1227465:1227924 [2] NCCL INFO comm 0x8e85d80 rank 2 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0xc0a121b95733704f - Init COMPLETE
+ip-26-0-165-164:1227467:1227925 [4] NCCL INFO comm 0x94610b0 rank 4 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0xc0a121b95733704f - Init COMPLETE
+ip-26-0-165-164:1227466:1227921 [3] NCCL INFO comm 0x90f2670 rank 3 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0xc0a121b95733704f - Init COMPLETE
+ip-26-0-165-164:1227470:1227923 [7] NCCL INFO comm 0x94f7a70 rank 7 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0xc0a121b95733704f - Init COMPLETE
+ip-26-0-165-164:1227468:1227922 [5] NCCL INFO comm 0x9dec4e0 rank 5 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0xc0a121b95733704f - Init COMPLETE
+ip-26-0-165-164:1227464:1227926 [1] NCCL INFO comm 0x92d57b0 rank 1 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0xc0a121b95733704f - Init COMPLETE
+12/28/2024 11:55:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: Config:
+12/28/2024 11:55:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: Config(general=GeneralArgs(project='debug',
+12/28/2024 11:55:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                            run='469G_dp2_tp32_pp2_acc2_mbs64_seq4096_zero1_tpmodeRED_vocab131k',
+12/28/2024 11:55:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                            seed=42,
+12/28/2024 11:55:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                            step=None,
+12/28/2024 11:55:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                            consumed_train_samples=None,
+12/28/2024 11:55:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                            benchmark_csv_path=PosixPath('benchmark/results/bench_final2.csv'),
+12/28/2024 11:55:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                            ignore_sanity_checks=True),
+12/28/2024 11:55:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:        parallelism=ParallelismArgs(dp=2,
+12/28/2024 11:55:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                    pp=2,
+12/28/2024 11:55:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                    tp=32,
+12/28/2024 11:55:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                    pp_engine=<nanotron.parallel.pipeline_parallel.engine.OneForwardOneBackwardPipelineEngine object at 0x7fb23d10cb80>,
+12/28/2024 11:55:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                    tp_mode=<TensorParallelLinearMode.REDUCE_SCATTER: 2>,
+12/28/2024 11:55:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                    tp_linear_async_communication=True,
+12/28/2024 11:55:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                    recompute_layer=False,
+12/28/2024 11:55:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                    tp_recompute_allgather=True,
+12/28/2024 11:55:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                    expert_parallel_size=1),
+12/28/2024 11:55:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:        model=ModelArgs(model_config=LlamaConfig(bos_token_id=0,
+12/28/2024 11:55:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 eos_token_id=0,
+12/28/2024 11:55:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 hidden_act='silu',
+12/28/2024 11:55:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 hidden_size=16384,
+12/28/2024 11:55:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 initializer_range=0.02,
+12/28/2024 11:55:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 intermediate_size=53248,
+12/28/2024 11:55:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 is_llama_config=True,
+12/28/2024 11:55:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 max_position_embeddings=4096,
+12/28/2024 11:55:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 num_attention_heads=128,
+12/28/2024 11:55:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 num_hidden_layers=126,
+12/28/2024 11:55:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 num_key_value_heads=128,
+12/28/2024 11:55:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 pad_token_id=None,
+12/28/2024 11:55:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 pretraining_tp=1,
+12/28/2024 11:55:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 rms_norm_eps=1e-05,
+12/28/2024 11:55:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 rope_scaling=None,
+12/28/2024 11:55:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 rope_theta=10000.0,
+12/28/2024 11:55:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 rope_interleaved=False,
+12/28/2024 11:55:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 tie_word_embeddings=False,
+12/28/2024 11:55:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 use_cache=True,
+12/28/2024 11:55:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 vocab_size=131072),
+12/28/2024 11:55:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                        init_method=RandomInit(std=0.02),
+12/28/2024 11:55:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                        dtype=torch.bfloat16,
+12/28/2024 11:55:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                        make_vocab_size_divisible_by=1,
+12/28/2024 11:55:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                        ddp_bucket_cap_mb=25),
+12/28/2024 11:55:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:        tokenizer=TokenizerArgs(tokenizer_name_or_path='robot-test/dummy-tokenizer-wordlevel',
+12/28/2024 11:55:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                tokenizer_revision=None,
+12/28/2024 11:55:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                tokenizer_max_length=None),
+12/28/2024 11:55:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:        checkpoints=CheckpointsArgs(checkpoints_path=PosixPath('checkpoints'),
+12/28/2024 11:55:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                    checkpoint_interval=10000,
+12/28/2024 11:55:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                    save_initial_state=False,
+12/28/2024 11:55:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                    save_final_state=False,
+12/28/2024 11:55:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                    resume_checkpoint_path=None,
+12/28/2024 11:55:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                    checkpoints_path_is_shared_file_system=False),
+12/28/2024 11:55:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:        logging=LoggingArgs(log_level='info',
+12/28/2024 11:55:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                            log_level_replica='info',
+12/28/2024 11:55:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                            iteration_step_info_interval=1),
+12/28/2024 11:55:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:        tokens=TokensArgs(sequence_length=4096,
+12/28/2024 11:55:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                          train_steps=100,
+12/28/2024 11:55:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                          micro_batch_size=64,
+12/28/2024 11:55:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                          batch_accumulation_per_replica=2,
+12/28/2024 11:55:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                          val_check_interval=100,
+12/28/2024 11:55:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                          limit_val_batches=0,
+12/28/2024 11:55:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                          limit_test_batches=0),
+12/28/2024 11:55:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:        optimizer=OptimizerArgs(optimizer_factory=AdamWOptimizerArgs(adam_eps=1e-08,
+12/28/2024 11:55:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                                     adam_beta1=0.9,
+12/28/2024 11:55:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                                     adam_beta2=0.95,
+12/28/2024 11:55:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                                     torch_adam_is_fused=True,
+12/28/2024 11:55:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                                     name='adamW'),
+12/28/2024 11:55:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                zero_stage=1,
+12/28/2024 11:55:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                weight_decay=0.01,
+12/28/2024 11:55:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                clip_grad=1.0,
+12/28/2024 11:55:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                accumulate_grad_in_fp32=True,
+12/28/2024 11:55:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                learning_rate_scheduler=LRSchedulerArgs(learning_rate=0.0003,
+12/28/2024 11:55:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                                        lr_warmup_steps=2,
+12/28/2024 11:55:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                                        lr_warmup_style='linear',
+12/28/2024 11:55:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                                        lr_decay_style='cosine',
+12/28/2024 11:55:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                                        lr_decay_steps=13,
+12/28/2024 11:55:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                                        lr_decay_starting_step=None,
+12/28/2024 11:55:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                                        min_decay_lr=1e-05)),
+12/28/2024 11:55:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:        data_stages=[DatasetStageArgs(name='Stable Training Stage',
+12/28/2024 11:55:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                      start_training_step=1,
+12/28/2024 11:55:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                      data=DataArgs(dataset=None,
+12/28/2024 11:55:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                    seed=42,
+12/28/2024 11:55:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                    num_loading_workers=1))],
+12/28/2024 11:55:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:        profiler=None,
+12/28/2024 11:55:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:        lighteval=None,
+12/28/2024 11:55:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:        s3_upload=None)
+12/28/2024 11:55:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: Model Config:
+12/28/2024 11:55:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: LlamaConfig(bos_token_id=0,
+12/28/2024 11:55:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             eos_token_id=0,
+12/28/2024 11:55:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             hidden_act='silu',
+12/28/2024 11:55:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             hidden_size=16384,
+12/28/2024 11:55:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             initializer_range=0.02,
+12/28/2024 11:55:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             intermediate_size=53248,
+12/28/2024 11:55:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             is_llama_config=True,
+12/28/2024 11:55:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             max_position_embeddings=4096,
+12/28/2024 11:55:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             num_attention_heads=128,
+12/28/2024 11:55:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             num_hidden_layers=126,
+12/28/2024 11:55:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             num_key_value_heads=128,
+12/28/2024 11:55:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             pad_token_id=None,
+12/28/2024 11:55:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             pretraining_tp=1,
+12/28/2024 11:55:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             rms_norm_eps=1e-05,
+12/28/2024 11:55:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             rope_scaling=None,
+12/28/2024 11:55:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             rope_theta=10000.0,
+12/28/2024 11:55:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             rope_interleaved=False,
+12/28/2024 11:55:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             tie_word_embeddings=False,
+12/28/2024 11:55:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             use_cache=True,
+12/28/2024 11:55:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             vocab_size=131072)
+12/28/2024 11:55:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: Building model..
+12/28/2024 11:55:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: Initialize RoPE Theta = 10000.0
+12/28/2024 11:55:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: Setting PP block ranks...
+ip-26-0-165-59:1557266:1557759 [0] NCCL INFO Using network Libfabric
+ip-26-0-168-30:235652:236147 [4] NCCL INFO Using network Libfabric
+ip-26-0-168-30:235655:236142 [7] NCCL INFO Using network Libfabric
+ip-26-0-168-52:2722047:2722542 [2] NCCL INFO Using network Libfabric
+ip-26-0-168-30:235649:236143 [1] NCCL INFO Using network Libfabric
+ip-26-0-168-30:235648:236145 [0] NCCL INFO Using network Libfabric
+ip-26-0-165-59:1557273:1557760 [7] NCCL INFO Using network Libfabric
+ip-26-0-168-30:235651:236141 [3] NCCL INFO Using network Libfabric
+ip-26-0-168-30:235653:236140 [5] NCCL INFO Using network Libfabric
+ip-26-0-168-30:235650:236144 [2] NCCL INFO Using network Libfabric
+ip-26-0-168-34:272151:272646 [1] NCCL INFO Using network Libfabric
+ip-26-0-165-59:1557270:1557763 [4] NCCL INFO Using network Libfabric
+ip-26-0-165-59:1557271:1557762 [5] NCCL INFO Using network Libfabric
+ip-26-0-165-59:1557267:1557761 [1] NCCL INFO Using network Libfabric
+ip-26-0-168-34:272156:272648 [6] NCCL INFO Using network Libfabric
+ip-26-0-168-34:272157:272649 [7] NCCL INFO Using network Libfabric
+ip-26-0-168-52:2722048:2722545 [3] NCCL INFO Using network Libfabric
+ip-26-0-168-52:2722046:2722541 [1] NCCL INFO Using network Libfabric
+ip-26-0-168-52:2722051:2722543 [6] NCCL INFO Using network Libfabric
+ip-26-0-165-59:1557272:1557766 [6] NCCL INFO Using network Libfabric
+ip-26-0-168-34:272150:272645 [0] NCCL INFO Using network Libfabric
+ip-26-0-168-34:272152:272647 [2] NCCL INFO Using network Libfabric
+ip-26-0-165-59:1557268:1557764 [2] NCCL INFO Using network Libfabric
+ip-26-0-168-52:2722052:2722546 [7] NCCL INFO Using network Libfabric
+ip-26-0-165-59:1557269:1557765 [3] NCCL INFO Using network Libfabric
+ip-26-0-168-34:272154:272650 [4] NCCL INFO Using network Libfabric
+ip-26-0-168-30:235654:236146 [6] NCCL INFO Using network Libfabric
+ip-26-0-168-34:272155:272652 [5] NCCL INFO Using network Libfabric
+ip-26-0-168-34:272153:272651 [3] NCCL INFO Using network Libfabric
+ip-26-0-160-225:124160:124658 [0] NCCL INFO Using network Libfabric
+ip-26-0-168-52:2722045:2722544 [0] NCCL INFO Using network Libfabric
+ip-26-0-168-52:2722049:2722547 [4] NCCL INFO Using network Libfabric
+ip-26-0-160-225:124161:124659 [1] NCCL INFO Using network Libfabric
+ip-26-0-161-123:32706:33199 [2] NCCL INFO Using network Libfabric
+ip-26-0-168-52:2722050:2722548 [5] NCCL INFO Using network Libfabric
+ip-26-0-161-123:32708:33198 [4] NCCL INFO Using network Libfabric
+ip-26-0-161-123:32707:33200 [3] NCCL INFO Using network Libfabric
+ip-26-0-161-123:32705:33203 [1] NCCL INFO Using network Libfabric
+ip-26-0-161-123:32710:33202 [6] NCCL INFO Using network Libfabric
+ip-26-0-161-123:32711:33204 [7] NCCL INFO Using network Libfabric
+ip-26-0-161-123:32709:33201 [5] NCCL INFO Using network Libfabric
+ip-26-0-160-225:124162:124660 [2] NCCL INFO Using network Libfabric
+ip-26-0-161-103:96894:97381 [6] NCCL INFO Using network Libfabric
+ip-26-0-161-103:96895:97382 [7] NCCL INFO Using network Libfabric
+ip-26-0-161-123:32704:33197 [0] NCCL INFO Using network Libfabric
+ip-26-0-160-242:1364905:1365404 [3] NCCL INFO Using network Libfabric
+ip-26-0-160-242:1364907:1365403 [5] NCCL INFO Using network Libfabric
+ip-26-0-160-225:124163:124661 [3] NCCL INFO Using network Libfabric
+ip-26-0-160-225:124164:124663 [4] NCCL INFO Using network Libfabric
+ip-26-0-161-103:96890:97386 [2] NCCL INFO Using network Libfabric
+ip-26-0-161-103:96888:97387 [0] NCCL INFO Using network Libfabric
+ip-26-0-160-242:1364909:1365405 [7] NCCL INFO Using network Libfabric
+ip-26-0-160-225:124167:124664 [7] NCCL INFO Using network Libfabric
+ip-26-0-161-103:96889:97388 [1] NCCL INFO Using network Libfabric
+ip-26-0-160-242:1364908:1365406 [6] NCCL INFO Using network Libfabric
+ip-26-0-160-225:124165:124662 [5] NCCL INFO Using network Libfabric
+ip-26-0-161-103:96891:97385 [3] NCCL INFO Using network Libfabric
+ip-26-0-161-103:96893:97384 [5] NCCL INFO Using network Libfabric
+ip-26-0-160-242:1364906:1365407 [4] NCCL INFO Using network Libfabric
+ip-26-0-160-242:1364904:1365408 [2] NCCL INFO Using network Libfabric
+ip-26-0-160-225:124166:124665 [6] NCCL INFO Using network Libfabric
+ip-26-0-160-242:1364903:1365409 [1] NCCL INFO Using network Libfabric
+ip-26-0-161-138:96563:97086 [0] NCCL INFO Using network Libfabric
+ip-26-0-161-138:96568:97091 [5] NCCL INFO Using network Libfabric
+ip-26-0-160-242:1364902:1365410 [0] NCCL INFO Using network Libfabric
+ip-26-0-161-138:96564:97088 [1] NCCL INFO Using network Libfabric
+ip-26-0-161-178:2815970:2816470 [7] NCCL INFO Using network Libfabric
+ip-26-0-161-178:2815969:2816471 [6] NCCL INFO Using network Libfabric
+ip-26-0-161-138:96567:97090 [4] NCCL INFO Using network Libfabric
+ip-26-0-161-178:2815968:2816472 [5] NCCL INFO Using network Libfabric
+ip-26-0-161-138:96565:97089 [2] NCCL INFO Using network Libfabric
+ip-26-0-161-178:2815967:2816473 [4] NCCL INFO Using network Libfabric
+ip-26-0-161-138:96569:97092 [6] NCCL INFO Using network Libfabric
+ip-26-0-161-138:96570:97093 [7] NCCL INFO Using network Libfabric
+ip-26-0-161-178:2815966:2816474 [3] NCCL INFO Using network Libfabric
+ip-26-0-161-138:96566:97087 [3] NCCL INFO Using network Libfabric
+ip-26-0-161-178:2815965:2816475 [2] NCCL INFO Using network Libfabric
+ip-26-0-161-142:62883:63373 [2] NCCL INFO Using network Libfabric
+ip-26-0-161-142:62882:63374 [1] NCCL INFO Using network Libfabric
+ip-26-0-161-142:62885:63377 [4] NCCL INFO Using network Libfabric
+ip-26-0-161-142:62884:63376 [3] NCCL INFO Using network Libfabric
+ip-26-0-161-178:2815964:2816476 [1] NCCL INFO Using network Libfabric
+ip-26-0-161-142:62881:63375 [0] NCCL INFO Using network Libfabric
+ip-26-0-161-142:62886:63378 [5] NCCL INFO Using network Libfabric
+ip-26-0-161-142:62887:63379 [6] NCCL INFO Using network Libfabric
+ip-26-0-161-153:2636090:2636598 [6] NCCL INFO Using network Libfabric
+ip-26-0-161-153:2636089:2636600 [5] NCCL INFO Using network Libfabric
+ip-26-0-161-78:114512:115007 [0] NCCL INFO Using network Libfabric
+ip-26-0-161-153:2636086:2636602 [2] NCCL INFO Using network Libfabric
+ip-26-0-161-153:2636088:2636601 [4] NCCL INFO Using network Libfabric
+ip-26-0-161-78:114513:115008 [1] NCCL INFO Using network Libfabric
+ip-26-0-165-164:1227470:1227958 [7] NCCL INFO Using network Libfabric
+ip-26-0-161-142:62888:63380 [7] NCCL INFO Using network Libfabric
+ip-26-0-165-202:153564:154057 [2] NCCL INFO Using network Libfabric
+ip-26-0-165-202:153569:154055 [7] NCCL INFO Using network Libfabric
+ip-26-0-165-202:153568:154056 [6] NCCL INFO Using network Libfabric
+ip-26-0-161-78:114515:115009 [3] NCCL INFO Using network Libfabric
+ip-26-0-161-178:2815963:2816477 [0] NCCL INFO Using network Libfabric
+ip-26-0-165-202:153567:154059 [5] NCCL INFO Using network Libfabric
+ip-26-0-161-153:2636087:2636603 [3] NCCL INFO Using network Libfabric
+ip-26-0-161-78:114516:115010 [4] NCCL INFO Using network Libfabric
+ip-26-0-165-202:153563:154060 [1] NCCL INFO Using network Libfabric
+ip-26-0-165-202:153565:154058 [3] NCCL INFO Using network Libfabric
+ip-26-0-165-202:153562:154061 [0] NCCL INFO Using network Libfabric
+ip-26-0-161-153:2636091:2636599 [7] NCCL INFO Using network Libfabric
+ip-26-0-161-78:114514:115011 [2] NCCL INFO Using network Libfabric
+ip-26-0-165-131:1630205:1630727 [0] NCCL INFO Using network Libfabric
+ip-26-0-165-202:153566:154062 [4] NCCL INFO Using network Libfabric
+ip-26-0-161-78:114519:115014 [7] NCCL INFO Using network Libfabric
+ip-26-0-165-131:1630206:1630728 [1] NCCL INFO Using network Libfabric
+ip-26-0-161-153:2636085:2636604 [1] NCCL INFO Using network Libfabric
+ip-26-0-165-164:1227468:1227959 [5] NCCL INFO Using network Libfabric
+ip-26-0-161-78:114518:115012 [6] NCCL INFO Using network Libfabric
+ip-26-0-161-78:114517:115013 [5] NCCL INFO Using network Libfabric
+ip-26-0-161-153:2636084:2636605 [0] NCCL INFO Using network Libfabric
+ip-26-0-165-164:1227469:1227957 [6] NCCL INFO Using network Libfabric
+ip-26-0-165-131:1630208:1630729 [3] NCCL INFO Using network Libfabric
+ip-26-0-165-131:1630209:1630730 [4] NCCL INFO Using network Libfabric
+ip-26-0-161-103:96892:97383 [4] NCCL INFO Using network Libfabric
+ip-26-0-165-164:1227465:1227962 [2] NCCL INFO Using network Libfabric
+ip-26-0-165-164:1227467:1227960 [4] NCCL INFO Using network Libfabric
+ip-26-0-165-131:1630211:1630733 [6] NCCL INFO Using network Libfabric
+ip-26-0-165-131:1630207:1630731 [2] NCCL INFO Using network Libfabric
+ip-26-0-165-164:1227464:1227963 [1] NCCL INFO Using network Libfabric
+ip-26-0-165-131:1630210:1630732 [5] NCCL INFO Using network Libfabric
+ip-26-0-165-164:1227466:1227961 [3] NCCL INFO Using network Libfabric
+ip-26-0-165-131:1630212:1630734 [7] NCCL INFO Using network Libfabric
+ip-26-0-165-164:1227463:1227964 [0] NCCL INFO Using network Libfabric
+ip-26-0-168-52:2722052:2722546 [7] NCCL INFO comm 0xb6b79f0 rank 31 nranks 32 cudaDev 7 nvmlDev 7 busId ca000 commId 0xe6d0f49db1af694c - Init START
+ip-26-0-168-52:2722051:2722543 [6] NCCL INFO comm 0xb298c00 rank 30 nranks 32 cudaDev 6 nvmlDev 6 busId b9000 commId 0xe6d0f49db1af694c - Init START
+ip-26-0-165-59:1557267:1557761 [1] NCCL INFO comm 0x9e76b70 rank 1 nranks 32 cudaDev 1 nvmlDev 1 busId 64000 commId 0xe6d0f49db1af694c - Init START
+ip-26-0-165-59:1557268:1557764 [2] NCCL INFO comm 0xab81880 rank 2 nranks 32 cudaDev 2 nvmlDev 2 busId 75000 commId 0xe6d0f49db1af694c - Init START
+ip-26-0-165-59:1557272:1557766 [6] NCCL INFO comm 0xa9e2270 rank 6 nranks 32 cudaDev 6 nvmlDev 6 busId b9000 commId 0xe6d0f49db1af694c - Init START
+ip-26-0-165-59:1557273:1557760 [7] NCCL INFO comm 0xb5cc7e0 rank 7 nranks 32 cudaDev 7 nvmlDev 7 busId ca000 commId 0xe6d0f49db1af694c - Init START
+ip-26-0-165-59:1557269:1557765 [3] NCCL INFO comm 0xb4cbdc0 rank 3 nranks 32 cudaDev 3 nvmlDev 3 busId 86000 commId 0xe6d0f49db1af694c - Init START
+ip-26-0-165-59:1557271:1557762 [5] NCCL INFO comm 0xa4e7e00 rank 5 nranks 32 cudaDev 5 nvmlDev 5 busId a8000 commId 0xe6d0f49db1af694c - Init START
+ip-26-0-168-34:272155:272652 [5] NCCL INFO comm 0xa77e610 rank 21 nranks 32 cudaDev 5 nvmlDev 5 busId a8000 commId 0xe6d0f49db1af694c - Init START
+ip-26-0-168-34:272156:272648 [6] NCCL INFO comm 0xb76eb00 rank 22 nranks 32 cudaDev 6 nvmlDev 6 busId b9000 commId 0xe6d0f49db1af694c - Init START
+ip-26-0-168-34:272157:272649 [7] NCCL INFO comm 0xb441140 rank 23 nranks 32 cudaDev 7 nvmlDev 7 busId ca000 commId 0xe6d0f49db1af694c - Init START
+ip-26-0-168-34:272153:272651 [3] NCCL INFO comm 0x9b8ea10 rank 19 nranks 32 cudaDev 3 nvmlDev 3 busId 86000 commId 0xe6d0f49db1af694c - Init START
+ip-26-0-168-34:272154:272650 [4] NCCL INFO comm 0xa853b80 rank 20 nranks 32 cudaDev 4 nvmlDev 4 busId 97000 commId 0xe6d0f49db1af694c - Init START
+ip-26-0-168-34:272152:272647 [2] NCCL INFO comm 0xba40780 rank 18 nranks 32 cudaDev 2 nvmlDev 2 busId 75000 commId 0xe6d0f49db1af694c - Init START
+ip-26-0-168-52:2722050:2722548 [5] NCCL INFO comm 0xa6c0c80 rank 29 nranks 32 cudaDev 5 nvmlDev 5 busId a8000 commId 0xe6d0f49db1af694c - Init START
+ip-26-0-168-52:2722049:2722547 [4] NCCL INFO comm 0xa856380 rank 28 nranks 32 cudaDev 4 nvmlDev 4 busId 97000 commId 0xe6d0f49db1af694c - Init START
+ip-26-0-168-52:2722048:2722545 [3] NCCL INFO comm 0xa691040 rank 27 nranks 32 cudaDev 3 nvmlDev 3 busId 86000 commId 0xe6d0f49db1af694c - Init START
+ip-26-0-165-59:1557270:1557763 [4] NCCL INFO comm 0xa956c80 rank 4 nranks 32 cudaDev 4 nvmlDev 4 busId 97000 commId 0xe6d0f49db1af694c - Init START
+ip-26-0-165-59:1557266:1557759 [0] NCCL INFO comm 0xc14cf00 rank 0 nranks 32 cudaDev 0 nvmlDev 0 busId 53000 commId 0xe6d0f49db1af694c - Init START
+ip-26-0-168-30:235648:236145 [0] NCCL INFO comm 0xa954a00 rank 8 nranks 32 cudaDev 0 nvmlDev 0 busId 53000 commId 0xe6d0f49db1af694c - Init START
+ip-26-0-168-30:235654:236146 [6] NCCL INFO comm 0xb97f190 rank 14 nranks 32 cudaDev 6 nvmlDev 6 busId b9000 commId 0xe6d0f49db1af694c - Init START
+ip-26-0-168-30:235653:236140 [5] NCCL INFO comm 0xa6da390 rank 13 nranks 32 cudaDev 5 nvmlDev 5 busId a8000 commId 0xe6d0f49db1af694c - Init START
+ip-26-0-168-30:235650:236144 [2] NCCL INFO comm 0xa4cc2d0 rank 10 nranks 32 cudaDev 2 nvmlDev 2 busId 75000 commId 0xe6d0f49db1af694c - Init START
+ip-26-0-168-30:235652:236147 [4] NCCL INFO comm 0xb25b2a0 rank 12 nranks 32 cudaDev 4 nvmlDev 4 busId 97000 commId 0xe6d0f49db1af694c - Init START
+ip-26-0-168-30:235655:236142 [7] NCCL INFO comm 0xb48d9f0 rank 15 nranks 32 cudaDev 7 nvmlDev 7 busId ca000 commId 0xe6d0f49db1af694c - Init START
+ip-26-0-168-52:2722047:2722542 [2] NCCL INFO comm 0xa9fbb80 rank 26 nranks 32 cudaDev 2 nvmlDev 2 busId 75000 commId 0xe6d0f49db1af694c - Init START
+ip-26-0-168-34:272151:272646 [1] NCCL INFO comm 0xb4d5a80 rank 17 nranks 32 cudaDev 1 nvmlDev 1 busId 64000 commId 0xe6d0f49db1af694c - Init START
+ip-26-0-168-34:272150:272645 [0] NCCL INFO comm 0xb33a540 rank 16 nranks 32 cudaDev 0 nvmlDev 0 busId 53000 commId 0xe6d0f49db1af694c - Init START
+ip-26-0-168-30:235649:236143 [1] NCCL INFO comm 0xb1b3870 rank 9 nranks 32 cudaDev 1 nvmlDev 1 busId 64000 commId 0xe6d0f49db1af694c - Init START
+ip-26-0-168-30:235651:236141 [3] NCCL INFO comm 0xa4ad700 rank 11 nranks 32 cudaDev 3 nvmlDev 3 busId 86000 commId 0xe6d0f49db1af694c - Init START
+ip-26-0-168-52:2722046:2722541 [1] NCCL INFO comm 0xadda970 rank 25 nranks 32 cudaDev 1 nvmlDev 1 busId 64000 commId 0xe6d0f49db1af694c - Init START
+ip-26-0-168-52:2722045:2722544 [0] NCCL INFO comm 0x9b08d30 rank 24 nranks 32 cudaDev 0 nvmlDev 0 busId 53000 commId 0xe6d0f49db1af694c - Init START
+ip-26-0-168-34:272155:272652 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-34:272154:272650 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-30:235655:236142 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-30:235654:236146 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-34:272157:272649 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-34:272156:272648 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-52:2722045:2722544 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-52:2722046:2722541 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-52:2722047:2722542 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-52:2722049:2722547 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-52:2722052:2722546 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-52:2722050:2722548 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-52:2722048:2722545 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-52:2722051:2722543 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-59:1557268:1557764 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-59:1557267:1557761 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-59:1557266:1557759 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-59:1557272:1557766 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-59:1557271:1557762 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-59:1557269:1557765 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-59:1557273:1557760 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-59:1557270:1557763 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-30:235652:236147 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-30:235653:236140 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-34:272153:272651 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-34:272152:272647 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-34:272151:272646 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-34:272150:272645 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-30:235648:236145 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-30:235649:236143 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-30:235651:236141 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-30:235650:236144 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-123:32706:33199 [2] NCCL INFO comm 0xbc2d3c0 rank 26 nranks 32 cudaDev 2 nvmlDev 2 busId 75000 commId 0x6ad24daef1fe15ce - Init START
+ip-26-0-161-123:32704:33197 [0] NCCL INFO comm 0xbc119e0 rank 24 nranks 32 cudaDev 0 nvmlDev 0 busId 53000 commId 0x6ad24daef1fe15ce - Init START
+ip-26-0-161-123:32705:33203 [1] NCCL INFO comm 0xb19e430 rank 25 nranks 32 cudaDev 1 nvmlDev 1 busId 64000 commId 0x6ad24daef1fe15ce - Init START
+ip-26-0-161-103:96895:97382 [7] NCCL INFO comm 0xbbd4280 rank 23 nranks 32 cudaDev 7 nvmlDev 7 busId ca000 commId 0x6ad24daef1fe15ce - Init START
+ip-26-0-161-103:96894:97381 [6] NCCL INFO comm 0xa2306a0 rank 22 nranks 32 cudaDev 6 nvmlDev 6 busId b9000 commId 0x6ad24daef1fe15ce - Init START
+ip-26-0-160-225:124167:124664 [7] NCCL INFO comm 0xb3f9260 rank 7 nranks 32 cudaDev 7 nvmlDev 7 busId ca000 commId 0x6ad24daef1fe15ce - Init START
+ip-26-0-160-242:1364909:1365405 [7] NCCL INFO comm 0xb61ae00 rank 15 nranks 32 cudaDev 7 nvmlDev 7 busId ca000 commId 0x6ad24daef1fe15ce - Init START
+ip-26-0-160-242:1364908:1365406 [6] NCCL INFO comm 0xbb61720 rank 14 nranks 32 cudaDev 6 nvmlDev 6 busId b9000 commId 0x6ad24daef1fe15ce - Init START
+ip-26-0-160-225:124166:124665 [6] NCCL INFO comm 0xaf0f490 rank 6 nranks 32 cudaDev 6 nvmlDev 6 busId b9000 commId 0x6ad24daef1fe15ce - Init START
+ip-26-0-160-225:124162:124660 [2] NCCL INFO comm 0xba5d2d0 rank 2 nranks 32 cudaDev 2 nvmlDev 2 busId 75000 commId 0x6ad24daef1fe15ce - Init START
+ip-26-0-160-225:124163:124661 [3] NCCL INFO comm 0xb6dea80 rank 3 nranks 32 cudaDev 3 nvmlDev 3 busId 86000 commId 0x6ad24daef1fe15ce - Init START
+ip-26-0-160-225:124165:124662 [5] NCCL INFO comm 0x9fea760 rank 5 nranks 32 cudaDev 5 nvmlDev 5 busId a8000 commId 0x6ad24daef1fe15ce - Init START
+ip-26-0-160-225:124164:124663 [4] NCCL INFO comm 0xa721840 rank 4 nranks 32 cudaDev 4 nvmlDev 4 busId 97000 commId 0x6ad24daef1fe15ce - Init START
+ip-26-0-161-103:96893:97384 [5] NCCL INFO comm 0x9e9dd50 rank 21 nranks 32 cudaDev 5 nvmlDev 5 busId a8000 commId 0x6ad24daef1fe15ce - Init START
+ip-26-0-161-103:96892:97383 [4] NCCL INFO comm 0xb351080 rank 20 nranks 32 cudaDev 4 nvmlDev 4 busId 97000 commId 0x6ad24daef1fe15ce - Init START
+ip-26-0-161-103:96891:97385 [3] NCCL INFO comm 0xae813d0 rank 19 nranks 32 cudaDev 3 nvmlDev 3 busId 86000 commId 0x6ad24daef1fe15ce - Init START
+ip-26-0-161-103:96890:97386 [2] NCCL INFO comm 0xa5cc060 rank 18 nranks 32 cudaDev 2 nvmlDev 2 busId 75000 commId 0x6ad24daef1fe15ce - Init START
+ip-26-0-161-103:96889:97388 [1] NCCL INFO comm 0x9f87fe0 rank 17 nranks 32 cudaDev 1 nvmlDev 1 busId 64000 commId 0x6ad24daef1fe15ce - Init START
+ip-26-0-161-123:32707:33200 [3] NCCL INFO comm 0xa02c190 rank 27 nranks 32 cudaDev 3 nvmlDev 3 busId 86000 commId 0x6ad24daef1fe15ce - Init START
+ip-26-0-161-123:32709:33201 [5] NCCL INFO comm 0xb9f5040 rank 29 nranks 32 cudaDev 5 nvmlDev 5 busId a8000 commId 0x6ad24daef1fe15ce - Init START
+ip-26-0-161-123:32710:33202 [6] NCCL INFO comm 0xb0d3b90 rank 30 nranks 32 cudaDev 6 nvmlDev 6 busId b9000 commId 0x6ad24daef1fe15ce - Init START
+ip-26-0-161-123:32711:33204 [7] NCCL INFO comm 0xc08b660 rank 31 nranks 32 cudaDev 7 nvmlDev 7 busId ca000 commId 0x6ad24daef1fe15ce - Init START
+ip-26-0-161-123:32708:33198 [4] NCCL INFO comm 0xba333f0 rank 28 nranks 32 cudaDev 4 nvmlDev 4 busId 97000 commId 0x6ad24daef1fe15ce - Init START
+ip-26-0-160-225:124161:124659 [1] NCCL INFO comm 0x9fafb60 rank 1 nranks 32 cudaDev 1 nvmlDev 1 busId 64000 commId 0x6ad24daef1fe15ce - Init START
+ip-26-0-160-225:124160:124658 [0] NCCL INFO comm 0xb447d40 rank 0 nranks 32 cudaDev 0 nvmlDev 0 busId 53000 commId 0x6ad24daef1fe15ce - Init START
+ip-26-0-161-103:96888:97387 [0] NCCL INFO comm 0xa5f0fe0 rank 16 nranks 32 cudaDev 0 nvmlDev 0 busId 53000 commId 0x6ad24daef1fe15ce - Init START
+ip-26-0-160-242:1364907:1365403 [5] NCCL INFO comm 0x9e43b50 rank 13 nranks 32 cudaDev 5 nvmlDev 5 busId a8000 commId 0x6ad24daef1fe15ce - Init START
+ip-26-0-160-242:1364905:1365404 [3] NCCL INFO comm 0xafe0340 rank 11 nranks 32 cudaDev 3 nvmlDev 3 busId 86000 commId 0x6ad24daef1fe15ce - Init START
+ip-26-0-160-242:1364906:1365407 [4] NCCL INFO comm 0xb742ff0 rank 12 nranks 32 cudaDev 4 nvmlDev 4 busId 97000 commId 0x6ad24daef1fe15ce - Init START
+ip-26-0-160-242:1364902:1365410 [0] NCCL INFO comm 0xaf66250 rank 8 nranks 32 cudaDev 0 nvmlDev 0 busId 53000 commId 0x6ad24daef1fe15ce - Init START
+ip-26-0-160-242:1364904:1365408 [2] NCCL INFO comm 0xb849c50 rank 10 nranks 32 cudaDev 2 nvmlDev 2 busId 75000 commId 0x6ad24daef1fe15ce - Init START
+ip-26-0-160-242:1364903:1365409 [1] NCCL INFO comm 0xbcea600 rank 9 nranks 32 cudaDev 1 nvmlDev 1 busId 64000 commId 0x6ad24daef1fe15ce - Init START
+ip-26-0-165-202:153569:154055 [7] NCCL INFO comm 0x9d016a0 rank 31 nranks 32 cudaDev 7 nvmlDev 7 busId ca000 commId 0x968ef8c3ebb85844 - Init START
+ip-26-0-161-78:114513:115008 [1] NCCL INFO comm 0xb40d9a0 rank 1 nranks 32 cudaDev 1 nvmlDev 1 busId 64000 commId 0x968ef8c3ebb85844 - Init START
+ip-26-0-161-78:114512:115007 [0] NCCL INFO comm 0xa85e740 rank 0 nranks 32 cudaDev 0 nvmlDev 0 busId 53000 commId 0x968ef8c3ebb85844 - Init START
+ip-26-0-161-78:114516:115010 [4] NCCL INFO comm 0xb3ab6d0 rank 4 nranks 32 cudaDev 4 nvmlDev 4 busId 97000 commId 0x968ef8c3ebb85844 - Init START
+ip-26-0-161-78:114517:115013 [5] NCCL INFO comm 0xb85b760 rank 5 nranks 32 cudaDev 5 nvmlDev 5 busId a8000 commId 0x968ef8c3ebb85844 - Init START
+ip-26-0-161-78:114515:115009 [3] NCCL INFO comm 0xa3a81c0 rank 3 nranks 32 cudaDev 3 nvmlDev 3 busId 86000 commId 0x968ef8c3ebb85844 - Init START
+ip-26-0-161-78:114514:115011 [2] NCCL INFO comm 0xb2d0ac0 rank 2 nranks 32 cudaDev 2 nvmlDev 2 busId 75000 commId 0x968ef8c3ebb85844 - Init START
+ip-26-0-165-164:1227470:1227958 [7] NCCL INFO comm 0xb2dbd00 rank 23 nranks 32 cudaDev 7 nvmlDev 7 busId ca000 commId 0x968ef8c3ebb85844 - Init START
+ip-26-0-165-164:1227469:1227957 [6] NCCL INFO comm 0x9a95350 rank 22 nranks 32 cudaDev 6 nvmlDev 6 busId b9000 commId 0x968ef8c3ebb85844 - Init START
+ip-26-0-165-202:153568:154056 [6] NCCL INFO comm 0xa7ee300 rank 30 nranks 32 cudaDev 6 nvmlDev 6 busId b9000 commId 0x968ef8c3ebb85844 - Init START
+ip-26-0-165-202:153567:154059 [5] NCCL INFO comm 0xc2061c0 rank 29 nranks 32 cudaDev 5 nvmlDev 5 busId a8000 commId 0x968ef8c3ebb85844 - Init START
+ip-26-0-165-202:153565:154058 [3] NCCL INFO comm 0xb9d80c0 rank 27 nranks 32 cudaDev 3 nvmlDev 3 busId 86000 commId 0x968ef8c3ebb85844 - Init START
+ip-26-0-165-202:153566:154062 [4] NCCL INFO comm 0xaa4e780 rank 28 nranks 32 cudaDev 4 nvmlDev 4 busId 97000 commId 0x968ef8c3ebb85844 - Init START
+ip-26-0-165-202:153564:154057 [2] NCCL INFO comm 0x9d98af0 rank 26 nranks 32 cudaDev 2 nvmlDev 2 busId 75000 commId 0x968ef8c3ebb85844 - Init START
+ip-26-0-165-202:153563:154060 [1] NCCL INFO comm 0xb9a2ae0 rank 25 nranks 32 cudaDev 1 nvmlDev 1 busId 64000 commId 0x968ef8c3ebb85844 - Init START
+ip-26-0-165-202:153562:154061 [0] NCCL INFO comm 0xaa68a60 rank 24 nranks 32 cudaDev 0 nvmlDev 0 busId 53000 commId 0x968ef8c3ebb85844 - Init START
+ip-26-0-165-131:1630212:1630734 [7] NCCL INFO comm 0x9c95060 rank 15 nranks 32 cudaDev 7 nvmlDev 7 busId ca000 commId 0x968ef8c3ebb85844 - Init START
+ip-26-0-165-131:1630211:1630733 [6] NCCL INFO comm 0xba47990 rank 14 nranks 32 cudaDev 6 nvmlDev 6 busId b9000 commId 0x968ef8c3ebb85844 - Init START
+ip-26-0-165-131:1630209:1630730 [4] NCCL INFO comm 0xb1f6980 rank 12 nranks 32 cudaDev 4 nvmlDev 4 busId 97000 commId 0x968ef8c3ebb85844 - Init START
+ip-26-0-165-131:1630210:1630732 [5] NCCL INFO comm 0xb46d6d0 rank 13 nranks 32 cudaDev 5 nvmlDev 5 busId a8000 commId 0x968ef8c3ebb85844 - Init START
+ip-26-0-165-131:1630208:1630729 [3] NCCL INFO comm 0x9aa21d0 rank 11 nranks 32 cudaDev 3 nvmlDev 3 busId 86000 commId 0x968ef8c3ebb85844 - Init START
+ip-26-0-165-131:1630207:1630731 [2] NCCL INFO comm 0xab227f0 rank 10 nranks 32 cudaDev 2 nvmlDev 2 busId 75000 commId 0x968ef8c3ebb85844 - Init START
+ip-26-0-165-164:1227468:1227959 [5] NCCL INFO comm 0xb1614c0 rank 21 nranks 32 cudaDev 5 nvmlDev 5 busId a8000 commId 0x968ef8c3ebb85844 - Init START
+ip-26-0-165-164:1227467:1227960 [4] NCCL INFO comm 0xa7d56c0 rank 20 nranks 32 cudaDev 4 nvmlDev 4 busId 97000 commId 0x968ef8c3ebb85844 - Init START
+ip-26-0-161-78:114518:115012 [6] NCCL INFO comm 0xb8bc080 rank 6 nranks 32 cudaDev 6 nvmlDev 6 busId b9000 commId 0x968ef8c3ebb85844 - Init START
+ip-26-0-161-78:114519:115014 [7] NCCL INFO comm 0xa5040c0 rank 7 nranks 32 cudaDev 7 nvmlDev 7 busId ca000 commId 0x968ef8c3ebb85844 - Init START
+ip-26-0-165-164:1227466:1227961 [3] NCCL INFO comm 0xaebd080 rank 19 nranks 32 cudaDev 3 nvmlDev 3 busId 86000 commId 0x968ef8c3ebb85844 - Init START
+ip-26-0-165-164:1227465:1227962 [2] NCCL INFO comm 0xac59150 rank 18 nranks 32 cudaDev 2 nvmlDev 2 busId 75000 commId 0x968ef8c3ebb85844 - Init START
+ip-26-0-165-131:1630206:1630728 [1] NCCL INFO comm 0xbb76e40 rank 9 nranks 32 cudaDev 1 nvmlDev 1 busId 64000 commId 0x968ef8c3ebb85844 - Init START
+ip-26-0-165-164:1227464:1227963 [1] NCCL INFO comm 0xb0a70f0 rank 17 nranks 32 cudaDev 1 nvmlDev 1 busId 64000 commId 0x968ef8c3ebb85844 - Init START
+ip-26-0-165-164:1227463:1227964 [0] NCCL INFO comm 0xa7d40d0 rank 16 nranks 32 cudaDev 0 nvmlDev 0 busId 53000 commId 0x968ef8c3ebb85844 - Init START
+ip-26-0-165-131:1630205:1630727 [0] NCCL INFO comm 0xaeb3380 rank 8 nranks 32 cudaDev 0 nvmlDev 0 busId 53000 commId 0x968ef8c3ebb85844 - Init START
+ip-26-0-160-225:124167:124664 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-160-225:124166:124665 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-160-225:124165:124662 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-160-225:124164:124663 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-160-242:1364904:1365408 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-160-242:1364905:1365404 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-123:32706:33199 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-160-225:124163:124661 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-160-225:124162:124660 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-103:96890:97386 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-103:96895:97382 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-103:96891:97385 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-103:96892:97383 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-103:96894:97381 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-103:96893:97384 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-103:96889:97388 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-160-225:124161:124659 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-160-225:124160:124658 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/541
+ip-26-0-161-103:96888:97387 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-123:32705:33203 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-160-242:1364908:1365406 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-123:32704:33197 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-123:32711:33204 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-123:32710:33202 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-123:32709:33201 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-160-242:1364903:1365409 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-160-242:1364909:1365405 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-160-242:1364907:1365403 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-160-242:1364906:1365407 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-123:32707:33200 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-160-242:1364902:1365410 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-123:32708:33198 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-78:114517:115013 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-78:114516:115010 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-202:153569:154055 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-78:114513:115008 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-78:114515:115009 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-78:114514:115011 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-164:1227470:1227958 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-164:1227467:1227960 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-164:1227468:1227959 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-164:1227469:1227957 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-164:1227465:1227962 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-131:1630211:1630733 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/409
+ip-26-0-165-131:1630208:1630729 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/409
+ip-26-0-165-131:1630212:1630734 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/409
+ip-26-0-165-131:1630210:1630732 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/409
+ip-26-0-165-131:1630206:1630728 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/409
+ip-26-0-165-131:1630209:1630730 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/409
+ip-26-0-161-78:114519:115014 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-78:114512:115007 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-202:153568:154056 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-202:153567:154059 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-78:114518:115012 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-164:1227464:1227963 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-164:1227463:1227964 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-164:1227466:1227961 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-131:1630205:1630727 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/409
+ip-26-0-165-131:1630207:1630731 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/409
+ip-26-0-165-202:153565:154058 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-202:153564:154057 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-202:153562:154061 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-202:153566:154062 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-202:153563:154060 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-142:62888:63380 [7] NCCL INFO comm 0xa913bd0 rank 15 nranks 32 cudaDev 7 nvmlDev 7 busId ca000 commId 0xc9fdc2638087cf9c - Init START
+ip-26-0-161-153:2636086:2636602 [2] NCCL INFO comm 0xb110fc0 rank 18 nranks 32 cudaDev 2 nvmlDev 2 busId 75000 commId 0xc9fdc2638087cf9c - Init START
+ip-26-0-161-153:2636084:2636605 [0] NCCL INFO comm 0xb1686c0 rank 16 nranks 32 cudaDev 0 nvmlDev 0 busId 53000 commId 0xc9fdc2638087cf9c - Init START
+ip-26-0-161-153:2636085:2636604 [1] NCCL INFO comm 0xb985620 rank 17 nranks 32 cudaDev 1 nvmlDev 1 busId 64000 commId 0xc9fdc2638087cf9c - Init START
+ip-26-0-161-153:2636087:2636603 [3] NCCL INFO comm 0xa065880 rank 19 nranks 32 cudaDev 3 nvmlDev 3 busId 86000 commId 0xc9fdc2638087cf9c - Init START
+ip-26-0-161-138:96570:97093 [7] NCCL INFO comm 0xb873da0 rank 7 nranks 32 cudaDev 7 nvmlDev 7 busId ca000 commId 0xc9fdc2638087cf9c - Init START
+ip-26-0-161-138:96569:97092 [6] NCCL INFO comm 0xa638af0 rank 6 nranks 32 cudaDev 6 nvmlDev 6 busId b9000 commId 0xc9fdc2638087cf9c - Init START
+ip-26-0-161-142:62887:63379 [6] NCCL INFO comm 0x9e2e830 rank 14 nranks 32 cudaDev 6 nvmlDev 6 busId b9000 commId 0xc9fdc2638087cf9c - Init START
+ip-26-0-161-142:62885:63377 [4] NCCL INFO comm 0xb5cbf30 rank 12 nranks 32 cudaDev 4 nvmlDev 4 busId 97000 commId 0xc9fdc2638087cf9c - Init START
+ip-26-0-161-142:62886:63378 [5] NCCL INFO comm 0xb59e210 rank 13 nranks 32 cudaDev 5 nvmlDev 5 busId a8000 commId 0xc9fdc2638087cf9c - Init START
+ip-26-0-161-142:62884:63376 [3] NCCL INFO comm 0xad6a3d0 rank 11 nranks 32 cudaDev 3 nvmlDev 3 busId 86000 commId 0xc9fdc2638087cf9c - Init START
+ip-26-0-161-142:62881:63375 [0] NCCL INFO comm 0xb8c2040 rank 8 nranks 32 cudaDev 0 nvmlDev 0 busId 53000 commId 0xc9fdc2638087cf9c - Init START
+ip-26-0-161-142:62883:63373 [2] NCCL INFO comm 0xa2edd90 rank 10 nranks 32 cudaDev 2 nvmlDev 2 busId 75000 commId 0xc9fdc2638087cf9c - Init START
+ip-26-0-161-142:62882:63374 [1] NCCL INFO comm 0xa6869b0 rank 9 nranks 32 cudaDev 1 nvmlDev 1 busId 64000 commId 0xc9fdc2638087cf9c - Init START
+ip-26-0-161-178:2815970:2816470 [7] NCCL INFO comm 0xa873aa0 rank 31 nranks 32 cudaDev 7 nvmlDev 7 busId ca000 commId 0xc9fdc2638087cf9c - Init START
+ip-26-0-161-178:2815969:2816471 [6] NCCL INFO comm 0xb206db0 rank 30 nranks 32 cudaDev 6 nvmlDev 6 busId b9000 commId 0xc9fdc2638087cf9c - Init START
+ip-26-0-161-153:2636088:2636601 [4] NCCL INFO comm 0xaa62b50 rank 20 nranks 32 cudaDev 4 nvmlDev 4 busId 97000 commId 0xc9fdc2638087cf9c - Init START
+ip-26-0-161-138:96568:97091 [5] NCCL INFO comm 0xa08c730 rank 5 nranks 32 cudaDev 5 nvmlDev 5 busId a8000 commId 0xc9fdc2638087cf9c - Init START
+ip-26-0-161-138:96567:97090 [4] NCCL INFO comm 0xa01fba0 rank 4 nranks 32 cudaDev 4 nvmlDev 4 busId 97000 commId 0xc9fdc2638087cf9c - Init START
+ip-26-0-161-178:2815968:2816472 [5] NCCL INFO comm 0xb83d840 rank 29 nranks 32 cudaDev 5 nvmlDev 5 busId a8000 commId 0xc9fdc2638087cf9c - Init START
+ip-26-0-161-138:96566:97087 [3] NCCL INFO comm 0xb531f70 rank 3 nranks 32 cudaDev 3 nvmlDev 3 busId 86000 commId 0xc9fdc2638087cf9c - Init START
+ip-26-0-161-178:2815967:2816473 [4] NCCL INFO comm 0xa848d80 rank 28 nranks 32 cudaDev 4 nvmlDev 4 busId 97000 commId 0xc9fdc2638087cf9c - Init START
+ip-26-0-161-138:96565:97089 [2] NCCL INFO comm 0xb9b8c10 rank 2 nranks 32 cudaDev 2 nvmlDev 2 busId 75000 commId 0xc9fdc2638087cf9c - Init START
+ip-26-0-161-138:96564:97088 [1] NCCL INFO comm 0xadd6e40 rank 1 nranks 32 cudaDev 1 nvmlDev 1 busId 64000 commId 0xc9fdc2638087cf9c - Init START
+ip-26-0-161-178:2815966:2816474 [3] NCCL INFO comm 0xac53fa0 rank 27 nranks 32 cudaDev 3 nvmlDev 3 busId 86000 commId 0xc9fdc2638087cf9c - Init START
+ip-26-0-161-138:96563:97086 [0] NCCL INFO comm 0xb166280 rank 0 nranks 32 cudaDev 0 nvmlDev 0 busId 53000 commId 0xc9fdc2638087cf9c - Init START
+ip-26-0-161-178:2815964:2816476 [1] NCCL INFO comm 0x9f47930 rank 25 nranks 32 cudaDev 1 nvmlDev 1 busId 64000 commId 0xc9fdc2638087cf9c - Init START
+ip-26-0-161-178:2815965:2816475 [2] NCCL INFO comm 0xa383240 rank 26 nranks 32 cudaDev 2 nvmlDev 2 busId 75000 commId 0xc9fdc2638087cf9c - Init START
+ip-26-0-161-178:2815963:2816477 [0] NCCL INFO comm 0xab2e7e0 rank 24 nranks 32 cudaDev 0 nvmlDev 0 busId 53000 commId 0xc9fdc2638087cf9c - Init START
+ip-26-0-161-153:2636089:2636600 [5] NCCL INFO comm 0x9edfc90 rank 21 nranks 32 cudaDev 5 nvmlDev 5 busId a8000 commId 0xc9fdc2638087cf9c - Init START
+ip-26-0-161-153:2636090:2636598 [6] NCCL INFO comm 0xb1c8e30 rank 22 nranks 32 cudaDev 6 nvmlDev 6 busId b9000 commId 0xc9fdc2638087cf9c - Init START
+ip-26-0-161-153:2636091:2636599 [7] NCCL INFO comm 0xa7c32e0 rank 23 nranks 32 cudaDev 7 nvmlDev 7 busId ca000 commId 0xc9fdc2638087cf9c - Init START
+ip-26-0-161-153:2636087:2636603 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-142:62888:63380 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-153:2636086:2636602 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-138:96570:97093 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-138:96569:97092 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-142:62887:63379 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-153:2636085:2636604 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-142:62886:63378 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-153:2636084:2636605 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-178:2815970:2816470 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-178:2815968:2816472 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-138:96568:97091 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-138:96567:97090 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-142:62885:63377 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-142:62884:63376 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-142:62883:63373 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-142:62882:63374 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-153:2636089:2636600 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-153:2636090:2636598 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-138:96565:97089 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-138:96566:97087 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-142:62881:63375 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-178:2815967:2816473 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-178:2815969:2816471 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-153:2636088:2636601 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-138:96564:97088 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-138:96563:97086 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-178:2815966:2816474 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-153:2636091:2636599 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-178:2815964:2816476 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-178:2815965:2816475 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-178:2815963:2816477 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-164:1227467:1227960 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-165-164:1227467:1227960 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-160-225:124162:124660 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-160-225:124162:124660 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-165-202:153564:154057 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-165-202:153564:154057 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-165-202:153567:154059 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-165-202:153567:154059 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-168-34:272150:272645 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-168-34:272150:272645 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-168-34:272151:272646 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-168-34:272151:272646 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-165-131:1630207:1630731 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-165-131:1630207:1630731 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-161-138:96569:97092 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,fff00000,00000000
+ip-26-0-161-138:96569:97092 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-161-138:96568:97091 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,fff00000,00000000
+ip-26-0-161-138:96568:97091 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-160-225:124164:124663 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-160-225:124164:124663 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-165-202:153565:154058 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-165-202:153565:154058 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-165-202:153563:154060 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-165-202:153563:154060 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-165-164:1227465:1227962 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-165-164:1227465:1227962 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-160-225:124160:124658 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-160-225:124160:124658 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-161-78:114515:115009 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-161-78:114515:115009 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-160-225:124167:124664 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-160-225:124167:124664 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-168-34:272157:272649 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-168-34:272157:272649 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-160-225:124165:124662 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-160-225:124165:124662 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-165-164:1227463:1227964 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-165-164:1227463:1227964 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-160-225:124163:124661 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-160-225:124163:124661 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-160-225:124166:124665 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-160-225:124166:124665 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-168-30:235650:236144 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-168-30:235650:236144 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-161-138:96565:97089 [2] NCCL INFO Setting affinity for GPU 2 to 0fff,ffffffff
+ip-26-0-161-138:96565:97089 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-168-34:272156:272648 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-168-34:272156:272648 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-160-225:124161:124659 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-160-225:124161:124659 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-161-138:96570:97093 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,fff00000,00000000
+ip-26-0-161-138:96570:97093 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-161-138:96566:97087 [3] NCCL INFO Setting affinity for GPU 3 to 0fff,ffffffff
+ip-26-0-165-59:1557268:1557764 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-161-138:96566:97087 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-165-59:1557268:1557764 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-168-34:272152:272647 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-168-34:272152:272647 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-168-30:235655:236142 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-168-30:235655:236142 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-161-138:96563:97086 [0] NCCL INFO Setting affinity for GPU 0 to 0fff,ffffffff
+ip-26-0-161-138:96563:97086 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-165-131:1630208:1630729 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-165-131:1630208:1630729 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-161-138:96567:97090 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,fff00000,00000000
+ip-26-0-161-138:96567:97090 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-165-202:153562:154061 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-165-202:153562:154061 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-168-34:272155:272652 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-168-34:272155:272652 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-161-138:96564:97088 [1] NCCL INFO Setting affinity for GPU 1 to 0fff,ffffffff
+ip-26-0-161-138:96564:97088 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-165-202:153569:154055 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-165-202:153569:154055 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-168-34:272154:272650 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-168-34:272154:272650 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-168-52:2722051:2722543 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-168-52:2722051:2722543 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-168-34:272153:272651 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-168-34:272153:272651 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-165-202:153568:154056 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-165-202:153568:154056 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-165-131:1630212:1630734 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-165-131:1630212:1630734 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-165-131:1630206:1630728 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-165-131:1630206:1630728 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-165-131:1630211:1630733 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-165-131:1630211:1630733 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-165-164:1227470:1227958 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-165-164:1227470:1227958 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-165-131:1630210:1630732 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-165-131:1630210:1630732 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-168-30:235654:236146 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-168-30:235654:236146 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-165-202:153566:154062 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-165-202:153566:154062 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-161-178:2815963:2816477 [0] NCCL INFO Setting affinity for GPU 0 to 0fff,ffffffff
+ip-26-0-161-178:2815963:2816477 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-165-164:1227468:1227959 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-165-164:1227468:1227959 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-165-59:1557267:1557761 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-165-59:1557267:1557761 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-165-131:1630205:1630727 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-165-131:1630205:1630727 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-161-103:96894:97381 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-161-103:96894:97381 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-165-164:1227464:1227963 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-165-164:1227464:1227963 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-165-131:1630209:1630730 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-165-131:1630209:1630730 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-168-30:235648:236145 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-168-30:235648:236145 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-168-30:235649:236143 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-168-30:235649:236143 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-165-59:1557269:1557765 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-165-59:1557269:1557765 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-165-164:1227469:1227957 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-165-164:1227469:1227957 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-168-30:235653:236140 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-168-30:235653:236140 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-165-164:1227466:1227961 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-165-164:1227466:1227961 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-161-103:96892:97383 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-161-103:96892:97383 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-168-30:235652:236147 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-168-30:235652:236147 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-168-30:235651:236141 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-161-103:96888:97387 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-161-103:96888:97387 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-168-30:235651:236141 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-161-123:32707:33200 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-161-123:32707:33200 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-165-59:1557273:1557760 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-165-59:1557273:1557760 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-165-59:1557270:1557763 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-165-59:1557270:1557763 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-161-78:114513:115008 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-161-78:114513:115008 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-161-78:114514:115011 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-161-78:114514:115011 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-161-123:32704:33197 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-161-123:32704:33197 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-161-142:62881:63375 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-161-142:62881:63375 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-161-123:32709:33201 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-161-123:32709:33201 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-165-59:1557266:1557759 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-165-59:1557266:1557759 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-165-59:1557271:1557762 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-165-59:1557271:1557762 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-165-59:1557272:1557766 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-165-59:1557272:1557766 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-161-103:96889:97388 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-161-103:96889:97388 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-161-103:96891:97385 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-161-103:96891:97385 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-161-123:32706:33199 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-161-123:32706:33199 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-160-242:1364907:1365403 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,fff00000,00000000
+ip-26-0-160-242:1364907:1365403 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-161-178:2815970:2816470 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,fff00000,00000000
+ip-26-0-161-178:2815970:2816470 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-161-123:32705:33203 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-161-123:32705:33203 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-161-78:114517:115013 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-161-78:114517:115013 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-161-123:32710:33202 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-161-123:32710:33202 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-161-178:2815965:2816475 [2] NCCL INFO Setting affinity for GPU 2 to 0fff,ffffffff
+ip-26-0-161-178:2815965:2816475 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-161-142:62882:63374 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-161-142:62882:63374 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-161-123:32711:33204 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-161-123:32711:33204 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-161-103:96890:97386 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-161-103:96890:97386 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-161-78:114512:115007 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-161-78:114512:115007 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-161-103:96895:97382 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-161-103:96895:97382 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-161-103:96893:97384 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-161-103:96893:97384 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-161-123:32708:33198 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-161-123:32708:33198 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-161-78:114516:115010 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-161-78:114516:115010 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-161-178:2815969:2816471 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,fff00000,00000000
+ip-26-0-161-178:2815969:2816471 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-161-142:62887:63379 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-161-142:62887:63379 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-161-178:2815966:2816474 [3] NCCL INFO Setting affinity for GPU 3 to 0fff,ffffffff
+ip-26-0-161-178:2815966:2816474 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-161-178:2815967:2816473 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,fff00000,00000000
+ip-26-0-161-178:2815967:2816473 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-160-242:1364908:1365406 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,fff00000,00000000
+ip-26-0-160-242:1364908:1365406 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-168-52:2722046:2722541 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-168-52:2722046:2722541 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-161-142:62886:63378 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-161-142:62886:63378 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-168-52:2722049:2722547 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-168-52:2722049:2722547 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-161-142:62883:63373 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-161-142:62883:63373 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-161-78:114518:115012 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-161-78:114518:115012 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-168-52:2722052:2722546 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-168-52:2722052:2722546 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-168-52:2722045:2722544 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-168-52:2722045:2722544 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-161-178:2815964:2816476 [1] NCCL INFO Setting affinity for GPU 1 to 0fff,ffffffff
+ip-26-0-161-178:2815964:2816476 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-161-178:2815968:2816472 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,fff00000,00000000
+ip-26-0-161-178:2815968:2816472 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-161-78:114519:115014 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-161-78:114519:115014 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-161-142:62884:63376 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-161-142:62884:63376 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-160-242:1364902:1365410 [0] NCCL INFO Setting affinity for GPU 0 to 0fff,ffffffff
+ip-26-0-160-242:1364902:1365410 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-168-52:2722047:2722542 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-168-52:2722047:2722542 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-161-142:62888:63380 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-161-142:62888:63380 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-161-142:62885:63377 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-161-142:62885:63377 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-160-242:1364905:1365404 [3] NCCL INFO Setting affinity for GPU 3 to 0fff,ffffffff
+ip-26-0-160-242:1364905:1365404 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-160-242:1364903:1365409 [1] NCCL INFO Setting affinity for GPU 1 to 0fff,ffffffff
+ip-26-0-168-52:2722048:2722545 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-168-52:2722048:2722545 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-160-242:1364903:1365409 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-168-52:2722050:2722548 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-168-52:2722050:2722548 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-160-242:1364909:1365405 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,fff00000,00000000
+ip-26-0-160-242:1364909:1365405 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-160-242:1364906:1365407 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,fff00000,00000000
+ip-26-0-160-242:1364906:1365407 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-160-242:1364904:1365408 [2] NCCL INFO Setting affinity for GPU 2 to 0fff,ffffffff
+ip-26-0-160-242:1364904:1365408 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-161-153:2636087:2636603 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-161-153:2636087:2636603 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-161-78:114519:115014 [7] NCCL INFO NVLS Head  0:  0  8 16 24
+ip-26-0-161-78:114519:115014 [7] NCCL INFO NVLS Head  1:  1  9 17 25
+ip-26-0-161-78:114519:115014 [7] NCCL INFO NVLS Head  2:  2 10 18 26
+ip-26-0-161-78:114519:115014 [7] NCCL INFO NVLS Head  3:  3 11 19 27
+ip-26-0-161-78:114519:115014 [7] NCCL INFO NVLS Head  4:  4 12 20 28
+ip-26-0-161-78:114519:115014 [7] NCCL INFO NVLS Head  5:  5 13 21 29
+ip-26-0-161-78:114519:115014 [7] NCCL INFO NVLS Head  6:  6 14 22 30
+ip-26-0-161-78:114519:115014 [7] NCCL INFO NVLS Head  7:  7 15 23 31
+ip-26-0-161-78:114519:115014 [7] NCCL INFO Trees [0] -1/-1/-1->7->6 [1] 0/-1/-1->7->6 [2] 0/-1/-1->7->6 [3] 0/-1/-1->7->6 [4] 0/-1/-1->7->6 [5] 0/-1/-1->7->6 [6] 0/-1/-1->7->6 [7] 0/23/-1->7->-1 [8] -1/-1/-1->7->6 [9] 0/-1/-1->7->6 [10] 0/-1/-1->7->6 [11] 0/-1/-1->7->6 [12] 0/-1/-1->7->6 [13] 0/-1/-1->7->6 [14] 0/-1/-1->7->6 [15] 0/-1/-1->7->15
+ip-26-0-161-78:114519:115014 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-131:1630205:1630727 [0] NCCL INFO Trees [0] 9/-1/-1->8->16 [1] -1/-1/-1->8->15 [2] 9/-1/-1->8->15 [3] 9/-1/-1->8->15 [4] 9/-1/-1->8->15 [5] 9/-1/-1->8->15 [6] 9/-1/-1->8->15 [7] 9/-1/-1->8->15 [8] 9/16/0->8->24 [9] -1/-1/-1->8->15 [10] 9/-1/-1->8->15 [11] 9/-1/-1->8->15 [12] 9/-1/-1->8->15 [13] 9/-1/-1->8->15 [14] 9/-1/-1->8->15 [15] 9/-1/-1->8->15
+ip-26-0-165-131:1630205:1630727 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-131:1630206:1630728 [1] NCCL INFO Trees [0] 10/-1/-1->9->8 [1] 10/-1/-1->9->17 [2] -1/-1/-1->9->8 [3] 10/-1/-1->9->8 [4] 10/-1/-1->9->8 [5] 10/-1/-1->9->8 [6] 10/-1/-1->9->8 [7] 10/-1/-1->9->8 [8] 10/-1/-1->9->8 [9] 10/17/1->9->25 [10] -1/-1/-1->9->8 [11] 10/-1/-1->9->8 [12] 10/-1/-1->9->8 [13] 10/-1/-1->9->8 [14] 10/-1/-1->9->8 [15] 10/-1/-1->9->8
+ip-26-0-165-131:1630206:1630728 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-131:1630207:1630731 [2] NCCL INFO Trees [0] 11/-1/-1->10->9 [1] 11/-1/-1->10->9 [2] 11/-1/-1->10->18 [3] -1/-1/-1->10->9 [4] 11/-1/-1->10->9 [5] 11/-1/-1->10->9 [6] 11/-1/-1->10->9 [7] 11/-1/-1->10->9 [8] 11/-1/-1->10->9 [9] 11/-1/-1->10->9 [10] 11/18/2->10->26 [11] -1/-1/-1->10->9 [12] 11/-1/-1->10->9 [13] 11/-1/-1->10->9 [14] 11/-1/-1->10->9 [15] 11/-1/-1->10->9
+ip-26-0-165-131:1630207:1630731 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-131:1630208:1630729 [3] NCCL INFO Trees [0] 12/-1/-1->11->10 [1] 12/-1/-1->11->10 [2] 12/-1/-1->11->10 [3] 12/-1/-1->11->19 [4] -1/-1/-1->11->10 [5] 12/-1/-1->11->10 [6] 12/-1/-1->11->10 [7] 12/-1/-1->11->10 [8] 12/-1/-1->11->10 [9] 12/-1/-1->11->10 [10] 12/-1/-1->11->10 [11] 12/19/3->11->27 [12] -1/-1/-1->11->10 [13] 12/-1/-1->11->10 [14] 12/-1/-1->11->10 [15] 12/-1/-1->11->10
+ip-26-0-165-131:1630208:1630729 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-131:1630209:1630730 [4] NCCL INFO Trees [0] 13/-1/-1->12->11 [1] 13/-1/-1->12->11 [2] 13/-1/-1->12->11 [3] 13/-1/-1->12->11 [4] 13/-1/-1->12->20 [5] -1/-1/-1->12->11 [6] 13/-1/-1->12->11 [7] 13/-1/-1->12->11 [8] 13/-1/-1->12->11 [9] 13/-1/-1->12->11 [10] 13/-1/-1->12->11 [11] 13/-1/-1->12->11 [12] 13/20/4->12->28 [13] -1/-1/-1->12->11 [14] 13/-1/-1->12->11 [15] 13/-1/-1->12->11
+ip-26-0-165-131:1630209:1630730 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-131:1630210:1630732 [5] NCCL INFO Trees [0] 14/-1/-1->13->12 [1] 14/-1/-1->13->12 [2] 14/-1/-1->13->12 [3] 14/-1/-1->13->12 [4] 14/-1/-1->13->12 [5] 14/-1/-1->13->21 [6] -1/-1/-1->13->12 [7] 14/-1/-1->13->12 [8] 14/-1/-1->13->12 [9] 14/-1/-1->13->12 [10] 14/-1/-1->13->12 [11] 14/-1/-1->13->12 [12] 14/-1/-1->13->12 [13] 14/21/5->13->29 [14] -1/-1/-1->13->12 [15] 14/-1/-1->13->12
+ip-26-0-165-131:1630210:1630732 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-131:1630211:1630733 [6] NCCL INFO Trees [0] 15/-1/-1->14->13 [1] 15/-1/-1->14->13 [2] 15/-1/-1->14->13 [3] 15/-1/-1->14->13 [4] 15/-1/-1->14->13 [5] 15/-1/-1->14->13 [6] 15/-1/-1->14->22 [7] -1/-1/-1->14->13 [8] 15/-1/-1->14->13 [9] 15/-1/-1->14->13 [10] 15/-1/-1->14->13 [11] 15/-1/-1->14->13 [12] 15/-1/-1->14->13 [13] 15/-1/-1->14->13 [14] 15/22/6->14->30 [15] -1/-1/-1->14->13
+ip-26-0-165-131:1630211:1630733 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-131:1630212:1630734 [7] NCCL INFO Trees [0] -1/-1/-1->15->14 [1] 8/-1/-1->15->14 [2] 8/-1/-1->15->14 [3] 8/-1/-1->15->14 [4] 8/-1/-1->15->14 [5] 8/-1/-1->15->14 [6] 8/-1/-1->15->14 [7] 8/-1/-1->15->23 [8] -1/-1/-1->15->14 [9] 8/-1/-1->15->14 [10] 8/-1/-1->15->14 [11] 8/-1/-1->15->14 [12] 8/-1/-1->15->14 [13] 8/-1/-1->15->14 [14] 8/-1/-1->15->14 [15] 8/23/7->15->31
+ip-26-0-165-131:1630212:1630734 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-164:1227463:1227964 [0] NCCL INFO Trees [0] 17/8/24->16->0 [1] -1/-1/-1->16->23 [2] 17/-1/-1->16->23 [3] 17/-1/-1->16->23 [4] 17/-1/-1->16->23 [5] 17/-1/-1->16->23 [6] 17/-1/-1->16->23 [7] 17/-1/-1->16->23 [8] 17/-1/-1->16->8 [9] -1/-1/-1->16->23 [10] 17/-1/-1->16->23 [11] 17/-1/-1->16->23 [12] 17/-1/-1->16->23 [13] 17/-1/-1->16->23 [14] 17/-1/-1->16->23 [15] 17/-1/-1->16->23
+ip-26-0-165-164:1227463:1227964 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-164:1227464:1227963 [1] NCCL INFO Trees [0] 18/-1/-1->17->16 [1] 18/9/25->17->1 [2] -1/-1/-1->17->16 [3] 18/-1/-1->17->16 [4] 18/-1/-1->17->16 [5] 18/-1/-1->17->16 [6] 18/-1/-1->17->16 [7] 18/-1/-1->17->16 [8] 18/-1/-1->17->16 [9] 18/-1/-1->17->9 [10] -1/-1/-1->17->16 [11] 18/-1/-1->17->16 [12] 18/-1/-1->17->16 [13] 18/-1/-1->17->16 [14] 18/-1/-1->17->16 [15] 18/-1/-1->17->16
+ip-26-0-165-164:1227464:1227963 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-164:1227470:1227958 [7] NCCL INFO Trees [0] -1/-1/-1->23->22 [1] 16/-1/-1->23->22 [2] 16/-1/-1->23->22 [3] 16/-1/-1->23->22 [4] 16/-1/-1->23->22 [5] 16/-1/-1->23->22 [6] 16/-1/-1->23->22 [7] 16/15/31->23->7 [8] -1/-1/-1->23->22 [9] 16/-1/-1->23->22 [10] 16/-1/-1->23->22 [11] 16/-1/-1->23->22 [12] 16/-1/-1->23->22 [13] 16/-1/-1->23->22 [14] 16/-1/-1->23->22 [15] 16/-1/-1->23->15
+ip-26-0-165-164:1227469:1227957 [6] NCCL INFO Trees [0] 23/-1/-1->22->21 [1] 23/-1/-1->22->21 [2] 23/-1/-1->22->21 [3] 23/-1/-1->22->21 [4] 23/-1/-1->22->21 [5] 23/-1/-1->22->21 [6] 23/14/30->22->6 [7] -1/-1/-1->22->21 [8] 23/-1/-1->22->21 [9] 23/-1/-1->22->21 [10] 23/-1/-1->22->21 [11] 23/-1/-1->22->21 [12] 23/-1/-1->22->21 [13] 23/-1/-1->22->21 [14] 23/-1/-1->22->14 [15] -1/-1/-1->22->21
+ip-26-0-165-202:153564:154057 [2] NCCL INFO Trees [0] 27/-1/-1->26->25 [1] 27/-1/-1->26->25 [2] 27/-1/-1->26->18 [3] -1/-1/-1->26->25 [4] 27/-1/-1->26->25 [5] 27/-1/-1->26->25 [6] 27/-1/-1->26->25 [7] 27/-1/-1->26->25 [8] 27/-1/-1->26->25 [9] 27/-1/-1->26->25 [10] 27/10/-1->26->-1 [11] -1/-1/-1->26->25 [12] 27/-1/-1->26->25 [13] 27/-1/-1->26->25 [14] 27/-1/-1->26->25 [15] 27/-1/-1->26->25
+ip-26-0-165-202:153564:154057 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-202:153562:154061 [0] NCCL INFO Trees [0] 25/-1/-1->24->16 [1] -1/-1/-1->24->31 [2] 25/-1/-1->24->31 [3] 25/-1/-1->24->31 [4] 25/-1/-1->24->31 [5] 25/-1/-1->24->31 [6] 25/-1/-1->24->31 [7] 25/-1/-1->24->31 [8] 25/8/-1->24->-1 [9] -1/-1/-1->24->31 [10] 25/-1/-1->24->31 [11] 25/-1/-1->24->31 [12] 25/-1/-1->24->31 [13] 25/-1/-1->24->31 [14] 25/-1/-1->24->31 [15] 25/-1/-1->24->31
+ip-26-0-165-202:153562:154061 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-78:114512:115007 [0] NCCL INFO NVLS Head  0:  0  8 16 24
+ip-26-0-161-78:114512:115007 [0] NCCL INFO NVLS Head  1:  1  9 17 25
+ip-26-0-161-78:114512:115007 [0] NCCL INFO NVLS Head  2:  2 10 18 26
+ip-26-0-161-78:114512:115007 [0] NCCL INFO NVLS Head  3:  3 11 19 27
+ip-26-0-161-78:114512:115007 [0] NCCL INFO NVLS Head  4:  4 12 20 28
+ip-26-0-161-78:114512:115007 [0] NCCL INFO NVLS Head  5:  5 13 21 29
+ip-26-0-165-164:1227468:1227959 [5] NCCL INFO Trees [0] 22/-1/-1->21->20 [1] 22/-1/-1->21->20 [2] 22/-1/-1->21->20 [3] 22/-1/-1->21->20 [4] 22/-1/-1->21->20 [5] 22/13/29->21->5 [6] -1/-1/-1->21->20 [7] 22/-1/-1->21->20 [8] 22/-1/-1->21->20 [9] 22/-1/-1->21->20 [10] 22/-1/-1->21->20 [11] 22/-1/-1->21->20 [12] 22/-1/-1->21->20 [13] 22/-1/-1->21->13 [14] -1/-1/-1->21->20 [15] 22/-1/-1->21->20
+ip-26-0-165-164:1227470:1227958 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-164:1227469:1227957 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-164:1227468:1227959 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-164:1227466:1227961 [3] NCCL INFO Trees [0] 20/-1/-1->19->18 [1] 20/-1/-1->19->18 [2] 20/-1/-1->19->18 [3] 20/11/27->19->3 [4] -1/-1/-1->19->18 [5] 20/-1/-1->19->18 [6] 20/-1/-1->19->18 [7] 20/-1/-1->19->18 [8] 20/-1/-1->19->18 [9] 20/-1/-1->19->18 [10] 20/-1/-1->19->18 [11] 20/-1/-1->19->11 [12] -1/-1/-1->19->18 [13] 20/-1/-1->19->18 [14] 20/-1/-1->19->18 [15] 20/-1/-1->19->18
+ip-26-0-165-164:1227466:1227961 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-202:153563:154060 [1] NCCL INFO Trees [0] 26/-1/-1->25->24 [1] 26/-1/-1->25->17 [2] -1/-1/-1->25->24 [3] 26/-1/-1->25->24 [4] 26/-1/-1->25->24 [5] 26/-1/-1->25->24 [6] 26/-1/-1->25->24 [7] 26/-1/-1->25->24 [8] 26/-1/-1->25->24 [9] 26/9/-1->25->-1 [10] -1/-1/-1->25->24 [11] 26/-1/-1->25->24 [12] 26/-1/-1->25->24 [13] 26/-1/-1->25->24 [14] 26/-1/-1->25->24 [15] 26/-1/-1->25->24
+ip-26-0-165-202:153563:154060 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-78:114512:115007 [0] NCCL INFO NVLS Head  6:  6 14 22 30
+ip-26-0-161-78:114512:115007 [0] NCCL INFO NVLS Head  7:  7 15 23 31
+ip-26-0-165-164:1227465:1227962 [2] NCCL INFO Trees [0] 19/-1/-1->18->17 [1] 19/-1/-1->18->17 [2] 19/10/26->18->2 [3] -1/-1/-1->18->17 [4] 19/-1/-1->18->17 [5] 19/-1/-1->18->17 [6] 19/-1/-1->18->17 [7] 19/-1/-1->18->17 [8] 19/-1/-1->18->17 [9] 19/-1/-1->18->17 [10] 19/-1/-1->18->10 [11] -1/-1/-1->18->17 [12] 19/-1/-1->18->17 [13] 19/-1/-1->18->17 [14] 19/-1/-1->18->17 [15] 19/-1/-1->18->17
+ip-26-0-165-164:1227465:1227962 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-164:1227467:1227960 [4] NCCL INFO Trees [0] 21/-1/-1->20->19 [1] 21/-1/-1->20->19 [2] 21/-1/-1->20->19 [3] 21/-1/-1->20->19 [4] 21/12/28->20->4 [5] -1/-1/-1->20->19 [6] 21/-1/-1->20->19 [7] 21/-1/-1->20->19 [8] 21/-1/-1->20->19 [9] 21/-1/-1->20->19 [10] 21/-1/-1->20->19 [11] 21/-1/-1->20->19 [12] 21/-1/-1->20->12 [13] -1/-1/-1->20->19 [14] 21/-1/-1->20->19 [15] 21/-1/-1->20->19
+ip-26-0-165-164:1227467:1227960 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-202:153569:154055 [7] NCCL INFO Trees [0] -1/-1/-1->31->30 [1] 24/-1/-1->31->30 [2] 24/-1/-1->31->30 [3] 24/-1/-1->31->30 [4] 24/-1/-1->31->30 [5] 24/-1/-1->31->30 [6] 24/-1/-1->31->30 [7] 24/-1/-1->31->23 [8] -1/-1/-1->31->30 [9] 24/-1/-1->31->30 [10] 24/-1/-1->31->30 [11] 24/-1/-1->31->30 [12] 24/-1/-1->31->30 [13] 24/-1/-1->31->30 [14] 24/-1/-1->31->30 [15] 24/15/-1->31->-1
+ip-26-0-165-202:153569:154055 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-202:153565:154058 [3] NCCL INFO Trees [0] 28/-1/-1->27->26 [1] 28/-1/-1->27->26 [2] 28/-1/-1->27->26 [3] 28/-1/-1->27->19 [4] -1/-1/-1->27->26 [5] 28/-1/-1->27->26 [6] 28/-1/-1->27->26 [7] 28/-1/-1->27->26 [8] 28/-1/-1->27->26 [9] 28/-1/-1->27->26 [10] 28/-1/-1->27->26 [11] 28/11/-1->27->-1 [12] -1/-1/-1->27->26 [13] 28/-1/-1->27->26 [14] 28/-1/-1->27->26 [15] 28/-1/-1->27->26
+ip-26-0-165-202:153565:154058 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-78:114512:115007 [0] NCCL INFO Channel 00/16 :    0   1   2   3   4   5   6   7   8   9  10  11  12  13  14  15  16  17  18  19
+ip-26-0-161-78:114517:115013 [5] NCCL INFO NVLS Head  0:  0  8 16 24
+ip-26-0-161-78:114518:115012 [6] NCCL INFO NVLS Head  0:  0  8 16 24
+ip-26-0-161-78:114517:115013 [5] NCCL INFO NVLS Head  1:  1  9 17 25
+ip-26-0-161-78:114512:115007 [0] NCCL INFO Channel 01/16 :    0   7   6   5   4   3   2   9   8  15  14  13  12  11  10  17  16  23  22  21
+ip-26-0-161-78:114517:115013 [5] NCCL INFO NVLS Head  2:  2 10 18 26
+ip-26-0-161-78:114514:115011 [2] NCCL INFO NVLS Head  0:  0  8 16 24
+ip-26-0-161-78:114518:115012 [6] NCCL INFO NVLS Head  1:  1  9 17 25
+ip-26-0-161-78:114517:115013 [5] NCCL INFO NVLS Head  3:  3 11 19 27
+ip-26-0-161-78:114512:115007 [0] NCCL INFO Channel 02/16 :    0   7   6   5   4   3  10   9   8  15  14  13  12  11  18  17  16  23  22  21
+ip-26-0-161-78:114518:115012 [6] NCCL INFO NVLS Head  2:  2 10 18 26
+ip-26-0-165-202:153567:154059 [5] NCCL INFO Trees [0] 30/-1/-1->29->28 [1] 30/-1/-1->29->28 [2] 30/-1/-1->29->28 [3] 30/-1/-1->29->28 [4] 30/-1/-1->29->28 [5] 30/-1/-1->29->21 [6] -1/-1/-1->29->28 [7] 30/-1/-1->29->28 [8] 30/-1/-1->29->28 [9] 30/-1/-1->29->28 [10] 30/-1/-1->29->28 [11] 30/-1/-1->29->28 [12] 30/-1/-1->29->28 [13] 30/13/-1->29->-1 [14] -1/-1/-1->29->28 [15] 30/-1/-1->29->28
+ip-26-0-165-202:153566:154062 [4] NCCL INFO Trees [0] 29/-1/-1->28->27 [1] 29/-1/-1->28->27 [2] 29/-1/-1->28->27 [3] 29/-1/-1->28->27 [4] 29/-1/-1->28->20 [5] -1/-1/-1->28->27 [6] 29/-1/-1->28->27 [7] 29/-1/-1->28->27 [8] 29/-1/-1->28->27 [9] 29/-1/-1->28->27 [10] 29/-1/-1->28->27 [11] 29/-1/-1->28->27 [12] 29/12/-1->28->-1 [13] -1/-1/-1->28->27 [14] 29/-1/-1->28->27 [15] 29/-1/-1->28->27
+ip-26-0-165-202:153567:154059 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-202:153566:154062 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-202:153568:154056 [6] NCCL INFO Trees [0] 31/-1/-1->30->29 [1] 31/-1/-1->30->29 [2] 31/-1/-1->30->29 [3] 31/-1/-1->30->29 [4] 31/-1/-1->30->29 [5] 31/-1/-1->30->29 [6] 31/-1/-1->30->22 [7] -1/-1/-1->30->29 [8] 31/-1/-1->30->29 [9] 31/-1/-1->30->29 [10] 31/-1/-1->30->29 [11] 31/-1/-1->30->29 [12] 31/-1/-1->30->29 [13] 31/-1/-1->30->29 [14] 31/14/-1->30->-1 [15] -1/-1/-1->30->29
+ip-26-0-165-202:153568:154056 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-78:114516:115010 [4] NCCL INFO NVLS Head  0:  0  8 16 24
+ip-26-0-161-78:114517:115013 [5] NCCL INFO NVLS Head  4:  4 12 20 28
+ip-26-0-161-78:114512:115007 [0] NCCL INFO Channel 03/16 :    0   7   6   5   4  11  10   9   8  15  14  13  12  19  18  17  16  23  22  21
+ip-26-0-161-78:114515:115009 [3] NCCL INFO NVLS Head  0:  0  8 16 24
+ip-26-0-161-78:114514:115011 [2] NCCL INFO NVLS Head  1:  1  9 17 25
+ip-26-0-161-78:114518:115012 [6] NCCL INFO NVLS Head  3:  3 11 19 27
+ip-26-0-161-78:114514:115011 [2] NCCL INFO NVLS Head  2:  2 10 18 26
+ip-26-0-161-78:114518:115012 [6] NCCL INFO NVLS Head  4:  4 12 20 28
+ip-26-0-161-78:114512:115007 [0] NCCL INFO Channel 04/16 :    0   7   6   5  12  11  10   9   8  15  14  13  20  19  18  17  16  23  22  21
+ip-26-0-161-78:114516:115010 [4] NCCL INFO NVLS Head  1:  1  9 17 25
+ip-26-0-161-78:114517:115013 [5] NCCL INFO NVLS Head  5:  5 13 21 29
+ip-26-0-161-78:114518:115012 [6] NCCL INFO NVLS Head  5:  5 13 21 29
+ip-26-0-161-78:114513:115008 [1] NCCL INFO NVLS Head  0:  0  8 16 24
+ip-26-0-161-78:114512:115007 [0] NCCL INFO Channel 05/16 :    0   4   7   6  13  11  10   9   8  12  15  14  21  19  18  17  16  20  23  22
+ip-26-0-161-78:114514:115011 [2] NCCL INFO NVLS Head  3:  3 11 19 27
+ip-26-0-161-78:114515:115009 [3] NCCL INFO NVLS Head  1:  1  9 17 25
+ip-26-0-161-78:114517:115013 [5] NCCL INFO NVLS Head  6:  6 14 22 30
+ip-26-0-161-78:114516:115010 [4] NCCL INFO NVLS Head  2:  2 10 18 26
+ip-26-0-161-78:114517:115013 [5] NCCL INFO NVLS Head  7:  7 15 23 31
+ip-26-0-161-78:114518:115012 [6] NCCL INFO NVLS Head  6:  6 14 22 30
+ip-26-0-161-78:114516:115010 [4] NCCL INFO NVLS Head  3:  3 11 19 27
+ip-26-0-161-78:114514:115011 [2] NCCL INFO NVLS Head  4:  4 12 20 28
+ip-26-0-161-78:114513:115008 [1] NCCL INFO NVLS Head  1:  1  9 17 25
+ip-26-0-161-78:114515:115009 [3] NCCL INFO NVLS Head  2:  2 10 18 26
+ip-26-0-161-78:114514:115011 [2] NCCL INFO NVLS Head  5:  5 13 21 29
+ip-26-0-161-78:114512:115007 [0] NCCL INFO Channel 06/16 :    0   5   4   7  14  11  10   9   8  13  12  15  22  19  18  17  16  21  20  23
+ip-26-0-161-78:114518:115012 [6] NCCL INFO NVLS Head  7:  7 15 23 31
+ip-26-0-161-78:114516:115010 [4] NCCL INFO NVLS Head  4:  4 12 20 28
+ip-26-0-161-78:114513:115008 [1] NCCL INFO NVLS Head  2:  2 10 18 26
+ip-26-0-161-78:114515:115009 [3] NCCL INFO NVLS Head  3:  3 11 19 27
+ip-26-0-161-78:114516:115010 [4] NCCL INFO NVLS Head  5:  5 13 21 29
+ip-26-0-161-78:114514:115011 [2] NCCL INFO NVLS Head  6:  6 14 22 30
+ip-26-0-161-78:114512:115007 [0] NCCL INFO Channel 07/16 :    0   6   5   4  15  11  10   9   8  14  13  12  23  19  18  17  16  22  21  20
+ip-26-0-161-78:114516:115010 [4] NCCL INFO NVLS Head  6:  6 14 22 30
+ip-26-0-161-78:114513:115008 [1] NCCL INFO NVLS Head  3:  3 11 19 27
+ip-26-0-161-78:114516:115010 [4] NCCL INFO NVLS Head  7:  7 15 23 31
+ip-26-0-161-78:114515:115009 [3] NCCL INFO NVLS Head  4:  4 12 20 28
+ip-26-0-161-78:114514:115011 [2] NCCL INFO NVLS Head  7:  7 15 23 31
+ip-26-0-161-78:114517:115013 [5] NCCL INFO Trees [0] 6/-1/-1->5->4 [1] 6/-1/-1->5->4 [2] 6/-1/-1->5->4 [3] 6/-1/-1->5->4 [4] 6/-1/-1->5->4 [5] 6/21/-1->5->-1 [6] -1/-1/-1->5->4 [7] 6/-1/-1->5->4 [8] 6/-1/-1->5->4 [9] 6/-1/-1->5->4 [10] 6/-1/-1->5->4 [11] 6/-1/-1->5->4 [12] 6/-1/-1->5->4 [13] 6/-1/-1->5->13 [14] -1/-1/-1->5->4 [15] 6/-1/-1->5->4
+ip-26-0-161-78:114512:115007 [0] NCCL INFO Channel 08/16 :    0   1   2   3   4   5   6   7   8   9  10  11  12  13  14  15  16  17  18  19
+ip-26-0-161-78:114513:115008 [1] NCCL INFO NVLS Head  4:  4 12 20 28
+ip-26-0-161-78:114517:115013 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-78:114515:115009 [3] NCCL INFO NVLS Head  5:  5 13 21 29
+ip-26-0-161-78:114512:115007 [0] NCCL INFO Channel 09/16 :    0   7   6   5   4   3   2   9   8  15  14  13  12  11  10  17  16  23  22  21
+ip-26-0-161-78:114513:115008 [1] NCCL INFO NVLS Head  5:  5 13 21 29
+ip-26-0-161-78:114518:115012 [6] NCCL INFO Trees [0] 7/-1/-1->6->5 [1] 7/-1/-1->6->5 [2] 7/-1/-1->6->5 [3] 7/-1/-1->6->5 [4] 7/-1/-1->6->5 [5] 7/-1/-1->6->5 [6] 7/22/-1->6->-1 [7] -1/-1/-1->6->5 [8] 7/-1/-1->6->5 [9] 7/-1/-1->6->5 [10] 7/-1/-1->6->5 [11] 7/-1/-1->6->5 [12] 7/-1/-1->6->5 [13] 7/-1/-1->6->5 [14] 7/-1/-1->6->14 [15] -1/-1/-1->6->5
+ip-26-0-161-78:114512:115007 [0] NCCL INFO Channel 10/16 :    0   7   6   5   4   3  10   9   8  15  14  13  12  11  18  17  16  23  22  21
+ip-26-0-161-78:114515:115009 [3] NCCL INFO NVLS Head  6:  6 14 22 30
+ip-26-0-161-78:114513:115008 [1] NCCL INFO NVLS Head  6:  6 14 22 30
+ip-26-0-161-78:114518:115012 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-78:114515:115009 [3] NCCL INFO NVLS Head  7:  7 15 23 31
+ip-26-0-161-78:114512:115007 [0] NCCL INFO Channel 11/16 :    0   7   6   5   4  11  10   9   8  15  14  13  12  19  18  17  16  23  22  21
+ip-26-0-161-78:114513:115008 [1] NCCL INFO NVLS Head  7:  7 15 23 31
+ip-26-0-161-78:114514:115011 [2] NCCL INFO Trees [0] 3/-1/-1->2->1 [1] 3/-1/-1->2->1 [2] 3/18/-1->2->-1 [3] -1/-1/-1->2->1 [4] 3/-1/-1->2->1 [5] 3/-1/-1->2->1 [6] 3/-1/-1->2->1 [7] 3/-1/-1->2->1 [8] 3/-1/-1->2->1 [9] 3/-1/-1->2->1 [10] 3/-1/-1->2->10 [11] -1/-1/-1->2->1 [12] 3/-1/-1->2->1 [13] 3/-1/-1->2->1 [14] 3/-1/-1->2->1 [15] 3/-1/-1->2->1
+ip-26-0-161-78:114516:115010 [4] NCCL INFO Trees [0] 5/-1/-1->4->3 [1] 5/-1/-1->4->3 [2] 5/-1/-1->4->3 [3] 5/-1/-1->4->3 [4] 5/20/-1->4->-1 [5] -1/-1/-1->4->3 [6] 5/-1/-1->4->3 [7] 5/-1/-1->4->3 [8] 5/-1/-1->4->3 [9] 5/-1/-1->4->3 [10] 5/-1/-1->4->3 [11] 5/-1/-1->4->3 [12] 5/-1/-1->4->12 [13] -1/-1/-1->4->3 [14] 5/-1/-1->4->3 [15] 5/-1/-1->4->3
+ip-26-0-161-78:114512:115007 [0] NCCL INFO Channel 12/16 :    0   7   6   5  12  11  10   9   8  15  14  13  20  19  18  17  16  23  22  21
+ip-26-0-161-78:114514:115011 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-78:114516:115010 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-78:114512:115007 [0] NCCL INFO Channel 13/16 :    0   4   7   6  13  11  10   9   8  12  15  14  21  19  18  17  16  20  23  22
+ip-26-0-161-78:114512:115007 [0] NCCL INFO Channel 14/16 :    0   5   4   7  14  11  10   9   8  13  12  15  22  19  18  17  16  21  20  23
+ip-26-0-161-78:114512:115007 [0] NCCL INFO Channel 15/16 :    0   6   5   4  15  11  10   9   8  14  13  12  23  19  18  17  16  22  21  20
+ip-26-0-161-78:114515:115009 [3] NCCL INFO Trees [0] 4/-1/-1->3->2 [1] 4/-1/-1->3->2 [2] 4/-1/-1->3->2 [3] 4/19/-1->3->-1 [4] -1/-1/-1->3->2 [5] 4/-1/-1->3->2 [6] 4/-1/-1->3->2 [7] 4/-1/-1->3->2 [8] 4/-1/-1->3->2 [9] 4/-1/-1->3->2 [10] 4/-1/-1->3->2 [11] 4/-1/-1->3->11 [12] -1/-1/-1->3->2 [13] 4/-1/-1->3->2 [14] 4/-1/-1->3->2 [15] 4/-1/-1->3->2
+ip-26-0-161-78:114513:115008 [1] NCCL INFO Trees [0] 2/-1/-1->1->0 [1] 2/17/-1->1->-1 [2] -1/-1/-1->1->0 [3] 2/-1/-1->1->0 [4] 2/-1/-1->1->0 [5] 2/-1/-1->1->0 [6] 2/-1/-1->1->0 [7] 2/-1/-1->1->0 [8] 2/-1/-1->1->0 [9] 2/-1/-1->1->9 [10] -1/-1/-1->1->0 [11] 2/-1/-1->1->0 [12] 2/-1/-1->1->0 [13] 2/-1/-1->1->0 [14] 2/-1/-1->1->0 [15] 2/-1/-1->1->0
+ip-26-0-161-78:114515:115009 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-78:114513:115008 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-78:114512:115007 [0] NCCL INFO Trees [0] 1/16/-1->0->-1 [1] -1/-1/-1->0->7 [2] 1/-1/-1->0->7 [3] 1/-1/-1->0->7 [4] 1/-1/-1->0->7 [5] 1/-1/-1->0->7 [6] 1/-1/-1->0->7 [7] 1/-1/-1->0->7 [8] 1/-1/-1->0->8 [9] -1/-1/-1->0->7 [10] 1/-1/-1->0->7 [11] 1/-1/-1->0->7 [12] 1/-1/-1->0->7 [13] 1/-1/-1->0->7 [14] 1/-1/-1->0->7 [15] 1/-1/-1->0->7
+ip-26-0-161-78:114512:115007 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-78:114519:115014 [7] NCCL INFO Channel 00/0 : 7[7] -> 8[0] [send] via NET/Libfabric/0(0)/GDRDMA
+ip-26-0-161-78:114519:115014 [7] NCCL INFO Channel 08/0 : 7[7] -> 8[0] [send] via NET/Libfabric/0(0)/GDRDMA
+ip-26-0-165-131:1630206:1630728 [1] NCCL INFO Channel 00/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-165-131:1630208:1630729 [3] NCCL INFO Channel 00/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-165-131:1630207:1630731 [2] NCCL INFO Channel 00/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-165-131:1630206:1630728 [1] NCCL INFO Channel 08/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-165-131:1630209:1630730 [4] NCCL INFO Channel 00/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-165-164:1227469:1227957 [6] NCCL INFO Channel 00/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-161-78:114517:115013 [5] NCCL INFO Channel 00/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-165-202:153564:154057 [2] NCCL INFO Channel 00/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-161-78:114516:115010 [4] NCCL INFO Channel 00/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-164:1227468:1227959 [5] NCCL INFO Channel 00/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-165-131:1630207:1630731 [2] NCCL INFO Channel 08/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-165-131:1630208:1630729 [3] NCCL INFO Channel 08/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-165-202:153566:154062 [4] NCCL INFO Channel 00/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-165-131:1630211:1630733 [6] NCCL INFO Channel 00/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-165-164:1227469:1227957 [6] NCCL INFO Channel 08/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-165-131:1630210:1630732 [5] NCCL INFO Channel 00/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-161-78:114514:115011 [2] NCCL INFO Channel 00/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-164:1227466:1227961 [3] NCCL INFO Channel 00/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-161-78:114518:115012 [6] NCCL INFO Channel 00/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-78:114517:115013 [5] NCCL INFO Channel 08/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-78:114513:115008 [1] NCCL INFO Channel 00/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-78:114515:115009 [3] NCCL INFO Channel 00/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-78:114516:115010 [4] NCCL INFO Channel 08/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-131:1630209:1630730 [4] NCCL INFO Channel 08/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-165-202:153564:154057 [2] NCCL INFO Channel 08/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-165-202:153563:154060 [1] NCCL INFO Channel 00/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-165-202:153567:154059 [5] NCCL INFO Channel 00/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-165-202:153568:154056 [6] NCCL INFO Channel 00/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-165-164:1227464:1227963 [1] NCCL INFO Channel 00/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-165-202:153566:154062 [4] NCCL INFO Channel 08/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-165-131:1630211:1630733 [6] NCCL INFO Channel 08/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-165-164:1227465:1227962 [2] NCCL INFO Channel 00/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-165-164:1227467:1227960 [4] NCCL INFO Channel 00/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-165-164:1227468:1227959 [5] NCCL INFO Channel 08/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-165-131:1630210:1630732 [5] NCCL INFO Channel 08/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-165-202:153565:154058 [3] NCCL INFO Channel 00/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-165-164:1227466:1227961 [3] NCCL INFO Channel 08/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-161-78:114514:115011 [2] NCCL INFO Channel 08/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-153:2636089:2636600 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-161-153:2636089:2636600 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-161-78:114518:115012 [6] NCCL INFO Channel 08/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-131:1630209:1630730 [4] NCCL INFO Channel 05/0 : 12[4] -> 15[7] via P2P/IPC
+ip-26-0-161-78:114515:115009 [3] NCCL INFO Channel 08/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-202:153563:154060 [1] NCCL INFO Channel 08/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-165-202:153568:154056 [6] NCCL INFO Channel 08/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-165-202:153567:154059 [5] NCCL INFO Channel 08/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-165-164:1227467:1227960 [4] NCCL INFO Channel 08/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-165-164:1227464:1227963 [1] NCCL INFO Channel 08/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-165-131:1630209:1630730 [4] NCCL INFO Channel 06/0 : 12[4] -> 15[7] via P2P/IPC
+ip-26-0-165-202:153565:154058 [3] NCCL INFO Channel 08/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-165-164:1227465:1227962 [2] NCCL INFO Channel 08/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-161-78:114516:115010 [4] NCCL INFO Channel 05/0 : 4[4] -> 7[7] via P2P/IPC
+ip-26-0-165-131:1630209:1630730 [4] NCCL INFO Channel 13/0 : 12[4] -> 15[7] via P2P/IPC
+ip-26-0-165-131:1630209:1630730 [4] NCCL INFO Channel 14/0 : 12[4] -> 15[7] via P2P/IPC
+ip-26-0-161-78:114516:115010 [4] NCCL INFO Channel 06/0 : 4[4] -> 7[7] via P2P/IPC
+ip-26-0-165-164:1227467:1227960 [4] NCCL INFO Channel 05/0 : 20[4] -> 23[7] via P2P/IPC
+ip-26-0-161-78:114513:115008 [1] NCCL INFO Channel 08/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-202:153566:154062 [4] NCCL INFO Channel 05/0 : 28[4] -> 31[7] via P2P/IPC
+ip-26-0-165-164:1227467:1227960 [4] NCCL INFO Channel 06/0 : 20[4] -> 23[7] via P2P/IPC
+ip-26-0-165-202:153566:154062 [4] NCCL INFO Channel 06/0 : 28[4] -> 31[7] via P2P/IPC
+ip-26-0-161-78:114516:115010 [4] NCCL INFO Channel 13/0 : 4[4] -> 7[7] via P2P/IPC
+ip-26-0-165-164:1227467:1227960 [4] NCCL INFO Channel 13/0 : 20[4] -> 23[7] via P2P/IPC
+ip-26-0-161-78:114516:115010 [4] NCCL INFO Channel 14/0 : 4[4] -> 7[7] via P2P/IPC
+ip-26-0-165-202:153566:154062 [4] NCCL INFO Channel 13/0 : 28[4] -> 31[7] via P2P/IPC
+ip-26-0-165-164:1227467:1227960 [4] NCCL INFO Channel 14/0 : 20[4] -> 23[7] via P2P/IPC
+ip-26-0-165-202:153566:154062 [4] NCCL INFO Channel 14/0 : 28[4] -> 31[7] via P2P/IPC
+ip-26-0-161-153:2636090:2636598 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-161-153:2636090:2636598 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-161-153:2636085:2636604 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-161-153:2636085:2636604 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-161-153:2636091:2636599 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-161-153:2636091:2636599 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-161-153:2636088:2636601 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-161-153:2636088:2636601 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-168-52:2722050:2722548 [5] NCCL INFO Trees [0] 30/-1/-1->29->28 [1] 30/-1/-1->29->28 [2] 30/-1/-1->29->28 [3] 30/-1/-1->29->28 [4] 30/-1/-1->29->28 [5] 30/-1/-1->29->21 [6] -1/-1/-1->29->28 [7] 30/-1/-1->29->28 [8] 30/-1/-1->29->28 [9] 30/-1/-1->29->28 [10] 30/-1/-1->29->28 [11] 30/-1/-1->29->28 [12] 30/-1/-1->29->28 [13] 30/13/-1->29->-1 [14] -1/-1/-1->29->28 [15] 30/-1/-1->29->28
+ip-26-0-168-52:2722050:2722548 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-168-52:2722051:2722543 [6] NCCL INFO Trees [0] 31/-1/-1->30->29 [1] 31/-1/-1->30->29 [2] 31/-1/-1->30->29 [3] 31/-1/-1->30->29 [4] 31/-1/-1->30->29 [5] 31/-1/-1->30->29 [6] 31/-1/-1->30->22 [7] -1/-1/-1->30->29 [8] 31/-1/-1->30->29 [9] 31/-1/-1->30->29 [10] 31/-1/-1->30->29 [11] 31/-1/-1->30->29 [12] 31/-1/-1->30->29 [13] 31/-1/-1->30->29 [14] 31/14/-1->30->-1 [15] -1/-1/-1->30->29
+ip-26-0-168-52:2722051:2722543 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-168-52:2722052:2722546 [7] NCCL INFO Trees [0] -1/-1/-1->31->30 [1] 24/-1/-1->31->30 [2] 24/-1/-1->31->30 [3] 24/-1/-1->31->30 [4] 24/-1/-1->31->30 [5] 24/-1/-1->31->30 [6] 24/-1/-1->31->30 [7] 24/-1/-1->31->23 [8] -1/-1/-1->31->30 [9] 24/-1/-1->31->30 [10] 24/-1/-1->31->30 [11] 24/-1/-1->31->30 [12] 24/-1/-1->31->30 [13] 24/-1/-1->31->30 [14] 24/-1/-1->31->30 [15] 24/15/-1->31->-1
+ip-26-0-168-52:2722052:2722546 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-59:1557266:1557759 [0] NCCL INFO NVLS Head  0:  0  8 16 24
+ip-26-0-165-59:1557266:1557759 [0] NCCL INFO NVLS Head  1:  1  9 17 25
+ip-26-0-165-59:1557266:1557759 [0] NCCL INFO NVLS Head  2:  2 10 18 26
+ip-26-0-165-59:1557266:1557759 [0] NCCL INFO NVLS Head  3:  3 11 19 27
+ip-26-0-165-59:1557266:1557759 [0] NCCL INFO NVLS Head  4:  4 12 20 28
+ip-26-0-165-59:1557266:1557759 [0] NCCL INFO NVLS Head  5:  5 13 21 29
+ip-26-0-165-59:1557266:1557759 [0] NCCL INFO NVLS Head  6:  6 14 22 30
+ip-26-0-165-59:1557266:1557759 [0] NCCL INFO NVLS Head  7:  7 15 23 31
+ip-26-0-165-59:1557266:1557759 [0] NCCL INFO Channel 00/16 :    0   1   2   3   4   5   6   7   8   9  10  11  12  13  14  15  16  17  18  19
+ip-26-0-165-59:1557266:1557759 [0] NCCL INFO Channel 01/16 :    0   7   6   5   4   3   2   9   8  15  14  13  12  11  10  17  16  23  22  21
+ip-26-0-165-59:1557266:1557759 [0] NCCL INFO Channel 02/16 :    0   7   6   5   4   3  10   9   8  15  14  13  12  11  18  17  16  23  22  21
+ip-26-0-165-59:1557266:1557759 [0] NCCL INFO Channel 03/16 :    0   7   6   5   4  11  10   9   8  15  14  13  12  19  18  17  16  23  22  21
+ip-26-0-165-59:1557266:1557759 [0] NCCL INFO Channel 04/16 :    0   7   6   5  12  11  10   9   8  15  14  13  20  19  18  17  16  23  22  21
+ip-26-0-165-59:1557266:1557759 [0] NCCL INFO Channel 05/16 :    0   4   7   6  13  11  10   9   8  12  15  14  21  19  18  17  16  20  23  22
+ip-26-0-165-59:1557266:1557759 [0] NCCL INFO Channel 06/16 :    0   5   4   7  14  11  10   9   8  13  12  15  22  19  18  17  16  21  20  23
+ip-26-0-165-59:1557266:1557759 [0] NCCL INFO Channel 07/16 :    0   6   5   4  15  11  10   9   8  14  13  12  23  19  18  17  16  22  21  20
+ip-26-0-165-59:1557266:1557759 [0] NCCL INFO Channel 08/16 :    0   1   2   3   4   5   6   7   8   9  10  11  12  13  14  15  16  17  18  19
+ip-26-0-165-59:1557266:1557759 [0] NCCL INFO Channel 09/16 :    0   7   6   5   4   3   2   9   8  15  14  13  12  11  10  17  16  23  22  21
+ip-26-0-165-59:1557266:1557759 [0] NCCL INFO Channel 10/16 :    0   7   6   5   4   3  10   9   8  15  14  13  12  11  18  17  16  23  22  21
+ip-26-0-165-59:1557266:1557759 [0] NCCL INFO Channel 11/16 :    0   7   6   5   4  11  10   9   8  15  14  13  12  19  18  17  16  23  22  21
+ip-26-0-165-59:1557266:1557759 [0] NCCL INFO Channel 12/16 :    0   7   6   5  12  11  10   9   8  15  14  13  20  19  18  17  16  23  22  21
+ip-26-0-165-59:1557267:1557761 [1] NCCL INFO NVLS Head  0:  0  8 16 24
+ip-26-0-165-59:1557266:1557759 [0] NCCL INFO Channel 13/16 :    0   4   7   6  13  11  10   9   8  12  15  14  21  19  18  17  16  20  23  22
+ip-26-0-165-59:1557267:1557761 [1] NCCL INFO NVLS Head  1:  1  9 17 25
+ip-26-0-165-59:1557267:1557761 [1] NCCL INFO NVLS Head  2:  2 10 18 26
+ip-26-0-165-59:1557266:1557759 [0] NCCL INFO Channel 14/16 :    0   5   4   7  14  11  10   9   8  13  12  15  22  19  18  17  16  21  20  23
+ip-26-0-165-59:1557267:1557761 [1] NCCL INFO NVLS Head  3:  3 11 19 27
+ip-26-0-165-59:1557267:1557761 [1] NCCL INFO NVLS Head  4:  4 12 20 28
+ip-26-0-165-59:1557266:1557759 [0] NCCL INFO Channel 15/16 :    0   6   5   4  15  11  10   9   8  14  13  12  23  19  18  17  16  22  21  20
+ip-26-0-165-59:1557267:1557761 [1] NCCL INFO NVLS Head  5:  5 13 21 29
+ip-26-0-165-59:1557267:1557761 [1] NCCL INFO NVLS Head  6:  6 14 22 30
+ip-26-0-165-59:1557267:1557761 [1] NCCL INFO NVLS Head  7:  7 15 23 31
+ip-26-0-165-59:1557266:1557759 [0] NCCL INFO Trees [0] 1/16/-1->0->-1 [1] -1/-1/-1->0->7 [2] 1/-1/-1->0->7 [3] 1/-1/-1->0->7 [4] 1/-1/-1->0->7 [5] 1/-1/-1->0->7 [6] 1/-1/-1->0->7 [7] 1/-1/-1->0->7 [8] 1/-1/-1->0->8 [9] -1/-1/-1->0->7 [10] 1/-1/-1->0->7 [11] 1/-1/-1->0->7 [12] 1/-1/-1->0->7 [13] 1/-1/-1->0->7 [14] 1/-1/-1->0->7 [15] 1/-1/-1->0->7
+ip-26-0-165-59:1557266:1557759 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-59:1557267:1557761 [1] NCCL INFO Trees [0] 2/-1/-1->1->0 [1] 2/17/-1->1->-1 [2] -1/-1/-1->1->0 [3] 2/-1/-1->1->0 [4] 2/-1/-1->1->0 [5] 2/-1/-1->1->0 [6] 2/-1/-1->1->0 [7] 2/-1/-1->1->0 [8] 2/-1/-1->1->0 [9] 2/-1/-1->1->9 [10] -1/-1/-1->1->0 [11] 2/-1/-1->1->0 [12] 2/-1/-1->1->0 [13] 2/-1/-1->1->0 [14] 2/-1/-1->1->0 [15] 2/-1/-1->1->0
+ip-26-0-168-30:235655:236142 [7] NCCL INFO Trees [0] -1/-1/-1->15->14 [1] 8/-1/-1->15->14 [2] 8/-1/-1->15->14 [3] 8/-1/-1->15->14 [4] 8/-1/-1->15->14 [5] 8/-1/-1->15->14 [6] 8/-1/-1->15->14 [7] 8/-1/-1->15->23 [8] -1/-1/-1->15->14 [9] 8/-1/-1->15->14 [10] 8/-1/-1->15->14 [11] 8/-1/-1->15->14 [12] 8/-1/-1->15->14 [13] 8/-1/-1->15->14 [14] 8/-1/-1->15->14 [15] 8/23/7->15->31
+ip-26-0-168-30:235650:236144 [2] NCCL INFO Trees [0] 11/-1/-1->10->9 [1] 11/-1/-1->10->9 [2] 11/-1/-1->10->18 [3] -1/-1/-1->10->9 [4] 11/-1/-1->10->9 [5] 11/-1/-1->10->9 [6] 11/-1/-1->10->9 [7] 11/-1/-1->10->9 [8] 11/-1/-1->10->9 [9] 11/-1/-1->10->9 [10] 11/18/2->10->26 [11] -1/-1/-1->10->9 [12] 11/-1/-1->10->9 [13] 11/-1/-1->10->9 [14] 11/-1/-1->10->9 [15] 11/-1/-1->10->9
+ip-26-0-168-34:272156:272648 [6] NCCL INFO Trees [0] 23/-1/-1->22->21 [1] 23/-1/-1->22->21 [2] 23/-1/-1->22->21 [3] 23/-1/-1->22->21 [4] 23/-1/-1->22->21 [5] 23/-1/-1->22->21 [6] 23/14/30->22->6 [7] -1/-1/-1->22->21 [8] 23/-1/-1->22->21 [9] 23/-1/-1->22->21 [10] 23/-1/-1->22->21 [11] 23/-1/-1->22->21 [12] 23/-1/-1->22->21 [13] 23/-1/-1->22->21 [14] 23/-1/-1->22->14 [15] -1/-1/-1->22->21
+ip-26-0-168-34:272156:272648 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-168-34:272154:272650 [4] NCCL INFO Trees [0] 21/-1/-1->20->19 [1] 21/-1/-1->20->19 [2] 21/-1/-1->20->19 [3] 21/-1/-1->20->19 [4] 21/12/28->20->4 [5] -1/-1/-1->20->19 [6] 21/-1/-1->20->19 [7] 21/-1/-1->20->19 [8] 21/-1/-1->20->19 [9] 21/-1/-1->20->19 [10] 21/-1/-1->20->19 [11] 21/-1/-1->20->19 [12] 21/-1/-1->20->12 [13] -1/-1/-1->20->19 [14] 21/-1/-1->20->19 [15] 21/-1/-1->20->19
+ip-26-0-168-52:2722046:2722541 [1] NCCL INFO Trees [0] 26/-1/-1->25->24 [1] 26/-1/-1->25->17 [2] -1/-1/-1->25->24 [3] 26/-1/-1->25->24 [4] 26/-1/-1->25->24 [5] 26/-1/-1->25->24 [6] 26/-1/-1->25->24 [7] 26/-1/-1->25->24 [8] 26/-1/-1->25->24 [9] 26/9/-1->25->-1 [10] -1/-1/-1->25->24 [11] 26/-1/-1->25->24 [12] 26/-1/-1->25->24 [13] 26/-1/-1->25->24 [14] 26/-1/-1->25->24 [15] 26/-1/-1->25->24
+ip-26-0-168-52:2722046:2722541 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-59:1557267:1557761 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-59:1557273:1557760 [7] NCCL INFO NVLS Head  0:  0  8 16 24
+ip-26-0-165-59:1557273:1557760 [7] NCCL INFO NVLS Head  1:  1  9 17 25
+ip-26-0-165-59:1557273:1557760 [7] NCCL INFO NVLS Head  2:  2 10 18 26
+ip-26-0-168-30:235648:236145 [0] NCCL INFO Trees [0] 9/-1/-1->8->16 [1] -1/-1/-1->8->15 [2] 9/-1/-1->8->15 [3] 9/-1/-1->8->15 [4] 9/-1/-1->8->15 [5] 9/-1/-1->8->15 [6] 9/-1/-1->8->15 [7] 9/-1/-1->8->15 [8] 9/16/0->8->24 [9] -1/-1/-1->8->15 [10] 9/-1/-1->8->15 [11] 9/-1/-1->8->15 [12] 9/-1/-1->8->15 [13] 9/-1/-1->8->15 [14] 9/-1/-1->8->15 [15] 9/-1/-1->8->15
+ip-26-0-168-30:235655:236142 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-168-30:235650:236144 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-168-30:235648:236145 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-168-30:235649:236143 [1] NCCL INFO Trees [0] 10/-1/-1->9->8 [1] 10/-1/-1->9->17 [2] -1/-1/-1->9->8 [3] 10/-1/-1->9->8 [4] 10/-1/-1->9->8 [5] 10/-1/-1->9->8 [6] 10/-1/-1->9->8 [7] 10/-1/-1->9->8 [8] 10/-1/-1->9->8 [9] 10/17/1->9->25 [10] -1/-1/-1->9->8 [11] 10/-1/-1->9->8 [12] 10/-1/-1->9->8 [13] 10/-1/-1->9->8 [14] 10/-1/-1->9->8 [15] 10/-1/-1->9->8
+ip-26-0-168-34:272154:272650 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-168-34:272155:272652 [5] NCCL INFO Trees [0] 22/-1/-1->21->20 [1] 22/-1/-1->21->20 [2] 22/-1/-1->21->20 [3] 22/-1/-1->21->20 [4] 22/-1/-1->21->20 [5] 22/13/29->21->5 [6] -1/-1/-1->21->20 [7] 22/-1/-1->21->20 [8] 22/-1/-1->21->20 [9] 22/-1/-1->21->20 [10] 22/-1/-1->21->20 [11] 22/-1/-1->21->20 [12] 22/-1/-1->21->20 [13] 22/-1/-1->21->13 [14] -1/-1/-1->21->20 [15] 22/-1/-1->21->20
+ip-26-0-168-34:272155:272652 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-59:1557273:1557760 [7] NCCL INFO NVLS Head  3:  3 11 19 27
+ip-26-0-165-59:1557272:1557766 [6] NCCL INFO NVLS Head  0:  0  8 16 24
+ip-26-0-165-59:1557273:1557760 [7] NCCL INFO NVLS Head  4:  4 12 20 28
+ip-26-0-165-59:1557271:1557762 [5] NCCL INFO NVLS Head  0:  0  8 16 24
+ip-26-0-165-59:1557273:1557760 [7] NCCL INFO NVLS Head  5:  5 13 21 29
+ip-26-0-168-30:235651:236141 [3] NCCL INFO Trees [0] 12/-1/-1->11->10 [1] 12/-1/-1->11->10 [2] 12/-1/-1->11->10 [3] 12/-1/-1->11->19 [4] -1/-1/-1->11->10 [5] 12/-1/-1->11->10 [6] 12/-1/-1->11->10 [7] 12/-1/-1->11->10 [8] 12/-1/-1->11->10 [9] 12/-1/-1->11->10 [10] 12/-1/-1->11->10 [11] 12/19/3->11->27 [12] -1/-1/-1->11->10 [13] 12/-1/-1->11->10 [14] 12/-1/-1->11->10 [15] 12/-1/-1->11->10
+ip-26-0-168-34:272153:272651 [3] NCCL INFO Trees [0] 20/-1/-1->19->18 [1] 20/-1/-1->19->18 [2] 20/-1/-1->19->18 [3] 20/11/27->19->3 [4] -1/-1/-1->19->18 [5] 20/-1/-1->19->18 [6] 20/-1/-1->19->18 [7] 20/-1/-1->19->18 [8] 20/-1/-1->19->18 [9] 20/-1/-1->19->18 [10] 20/-1/-1->19->18 [11] 20/-1/-1->19->11 [12] -1/-1/-1->19->18 [13] 20/-1/-1->19->18 [14] 20/-1/-1->19->18 [15] 20/-1/-1->19->18
+ip-26-0-168-34:272150:272645 [0] NCCL INFO Trees [0] 17/8/24->16->0 [1] -1/-1/-1->16->23 [2] 17/-1/-1->16->23 [3] 17/-1/-1->16->23 [4] 17/-1/-1->16->23 [5] 17/-1/-1->16->23 [6] 17/-1/-1->16->23 [7] 17/-1/-1->16->23 [8] 17/-1/-1->16->8 [9] -1/-1/-1->16->23 [10] 17/-1/-1->16->23 [11] 17/-1/-1->16->23 [12] 17/-1/-1->16->23 [13] 17/-1/-1->16->23 [14] 17/-1/-1->16->23 [15] 17/-1/-1->16->23
+ip-26-0-168-34:272153:272651 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-59:1557272:1557766 [6] NCCL INFO NVLS Head  1:  1  9 17 25
+ip-26-0-165-59:1557271:1557762 [5] NCCL INFO NVLS Head  1:  1  9 17 25
+ip-26-0-165-59:1557273:1557760 [7] NCCL INFO NVLS Head  6:  6 14 22 30
+ip-26-0-165-59:1557268:1557764 [2] NCCL INFO NVLS Head  0:  0  8 16 24
+ip-26-0-165-59:1557272:1557766 [6] NCCL INFO NVLS Head  2:  2 10 18 26
+ip-26-0-165-59:1557271:1557762 [5] NCCL INFO NVLS Head  2:  2 10 18 26
+ip-26-0-165-59:1557273:1557760 [7] NCCL INFO NVLS Head  7:  7 15 23 31
+ip-26-0-165-59:1557272:1557766 [6] NCCL INFO NVLS Head  3:  3 11 19 27
+ip-26-0-165-59:1557270:1557763 [4] NCCL INFO NVLS Head  0:  0  8 16 24
+ip-26-0-168-30:235649:236143 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-168-30:235651:236141 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-168-30:235653:236140 [5] NCCL INFO Trees [0] 14/-1/-1->13->12 [1] 14/-1/-1->13->12 [2] 14/-1/-1->13->12 [3] 14/-1/-1->13->12 [4] 14/-1/-1->13->12 [5] 14/-1/-1->13->21 [6] -1/-1/-1->13->12 [7] 14/-1/-1->13->12 [8] 14/-1/-1->13->12 [9] 14/-1/-1->13->12 [10] 14/-1/-1->13->12 [11] 14/-1/-1->13->12 [12] 14/-1/-1->13->12 [13] 14/21/5->13->29 [14] -1/-1/-1->13->12 [15] 14/-1/-1->13->12
+ip-26-0-168-30:235653:236140 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-168-30:235652:236147 [4] NCCL INFO Trees [0] 13/-1/-1->12->11 [1] 13/-1/-1->12->11 [2] 13/-1/-1->12->11 [3] 13/-1/-1->12->11 [4] 13/-1/-1->12->20 [5] -1/-1/-1->12->11 [6] 13/-1/-1->12->11 [7] 13/-1/-1->12->11 [8] 13/-1/-1->12->11 [9] 13/-1/-1->12->11 [10] 13/-1/-1->12->11 [11] 13/-1/-1->12->11 [12] 13/20/4->12->28 [13] -1/-1/-1->12->11 [14] 13/-1/-1->12->11 [15] 13/-1/-1->12->11
+ip-26-0-168-34:272150:272645 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-168-34:272157:272649 [7] NCCL INFO Trees [0] -1/-1/-1->23->22 [1] 16/-1/-1->23->22 [2] 16/-1/-1->23->22 [3] 16/-1/-1->23->22 [4] 16/-1/-1->23->22 [5] 16/-1/-1->23->22 [6] 16/-1/-1->23->22 [7] 16/15/31->23->7 [8] -1/-1/-1->23->22 [9] 16/-1/-1->23->22 [10] 16/-1/-1->23->22 [11] 16/-1/-1->23->22 [12] 16/-1/-1->23->22 [13] 16/-1/-1->23->22 [14] 16/-1/-1->23->22 [15] 16/-1/-1->23->15
+ip-26-0-168-34:272157:272649 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-59:1557271:1557762 [5] NCCL INFO NVLS Head  3:  3 11 19 27
+ip-26-0-165-59:1557268:1557764 [2] NCCL INFO NVLS Head  1:  1  9 17 25
+ip-26-0-165-59:1557272:1557766 [6] NCCL INFO NVLS Head  4:  4 12 20 28
+ip-26-0-165-59:1557270:1557763 [4] NCCL INFO NVLS Head  1:  1  9 17 25
+ip-26-0-165-59:1557268:1557764 [2] NCCL INFO NVLS Head  2:  2 10 18 26
+ip-26-0-165-59:1557271:1557762 [5] NCCL INFO NVLS Head  4:  4 12 20 28
+ip-26-0-165-59:1557272:1557766 [6] NCCL INFO NVLS Head  5:  5 13 21 29
+ip-26-0-165-59:1557270:1557763 [4] NCCL INFO NVLS Head  2:  2 10 18 26
+ip-26-0-165-59:1557268:1557764 [2] NCCL INFO NVLS Head  3:  3 11 19 27
+ip-26-0-165-59:1557271:1557762 [5] NCCL INFO NVLS Head  5:  5 13 21 29
+ip-26-0-165-59:1557272:1557766 [6] NCCL INFO NVLS Head  6:  6 14 22 30
+ip-26-0-165-59:1557270:1557763 [4] NCCL INFO NVLS Head  3:  3 11 19 27
+ip-26-0-165-59:1557268:1557764 [2] NCCL INFO NVLS Head  4:  4 12 20 28
+ip-26-0-165-59:1557271:1557762 [5] NCCL INFO NVLS Head  6:  6 14 22 30
+ip-26-0-168-30:235652:236147 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-168-34:272151:272646 [1] NCCL INFO Trees [0] 18/-1/-1->17->16 [1] 18/9/25->17->1 [2] -1/-1/-1->17->16 [3] 18/-1/-1->17->16 [4] 18/-1/-1->17->16 [5] 18/-1/-1->17->16 [6] 18/-1/-1->17->16 [7] 18/-1/-1->17->16 [8] 18/-1/-1->17->16 [9] 18/-1/-1->17->9 [10] -1/-1/-1->17->16 [11] 18/-1/-1->17->16 [12] 18/-1/-1->17->16 [13] 18/-1/-1->17->16 [14] 18/-1/-1->17->16 [15] 18/-1/-1->17->16
+ip-26-0-168-52:2722045:2722544 [0] NCCL INFO Trees [0] 25/-1/-1->24->16 [1] -1/-1/-1->24->31 [2] 25/-1/-1->24->31 [3] 25/-1/-1->24->31 [4] 25/-1/-1->24->31 [5] 25/-1/-1->24->31 [6] 25/-1/-1->24->31 [7] 25/-1/-1->24->31 [8] 25/8/-1->24->-1 [9] -1/-1/-1->24->31 [10] 25/-1/-1->24->31 [11] 25/-1/-1->24->31 [12] 25/-1/-1->24->31 [13] 25/-1/-1->24->31 [14] 25/-1/-1->24->31 [15] 25/-1/-1->24->31
+ip-26-0-168-52:2722045:2722544 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-59:1557272:1557766 [6] NCCL INFO NVLS Head  7:  7 15 23 31
+ip-26-0-165-59:1557270:1557763 [4] NCCL INFO NVLS Head  4:  4 12 20 28
+ip-26-0-165-59:1557273:1557760 [7] NCCL INFO Trees [0] -1/-1/-1->7->6 [1] 0/-1/-1->7->6 [2] 0/-1/-1->7->6 [3] 0/-1/-1->7->6 [4] 0/-1/-1->7->6 [5] 0/-1/-1->7->6 [6] 0/-1/-1->7->6 [7] 0/23/-1->7->-1 [8] -1/-1/-1->7->6 [9] 0/-1/-1->7->6 [10] 0/-1/-1->7->6 [11] 0/-1/-1->7->6 [12] 0/-1/-1->7->6 [13] 0/-1/-1->7->6 [14] 0/-1/-1->7->6 [15] 0/-1/-1->7->15
+ip-26-0-168-30:235654:236146 [6] NCCL INFO Trees [0] 15/-1/-1->14->13 [1] 15/-1/-1->14->13 [2] 15/-1/-1->14->13 [3] 15/-1/-1->14->13 [4] 15/-1/-1->14->13 [5] 15/-1/-1->14->13 [6] 15/-1/-1->14->22 [7] -1/-1/-1->14->13 [8] 15/-1/-1->14->13 [9] 15/-1/-1->14->13 [10] 15/-1/-1->14->13 [11] 15/-1/-1->14->13 [12] 15/-1/-1->14->13 [13] 15/-1/-1->14->13 [14] 15/22/6->14->30 [15] -1/-1/-1->14->13
+ip-26-0-168-30:235654:236146 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-168-34:272151:272646 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-168-52:2722047:2722542 [2] NCCL INFO Trees [0] 27/-1/-1->26->25 [1] 27/-1/-1->26->25 [2] 27/-1/-1->26->18 [3] -1/-1/-1->26->25 [4] 27/-1/-1->26->25 [5] 27/-1/-1->26->25 [6] 27/-1/-1->26->25 [7] 27/-1/-1->26->25 [8] 27/-1/-1->26->25 [9] 27/-1/-1->26->25 [10] 27/10/-1->26->-1 [11] -1/-1/-1->26->25 [12] 27/-1/-1->26->25 [13] 27/-1/-1->26->25 [14] 27/-1/-1->26->25 [15] 27/-1/-1->26->25
+ip-26-0-168-52:2722048:2722545 [3] NCCL INFO Trees [0] 28/-1/-1->27->26 [1] 28/-1/-1->27->26 [2] 28/-1/-1->27->26 [3] 28/-1/-1->27->19 [4] -1/-1/-1->27->26 [5] 28/-1/-1->27->26 [6] 28/-1/-1->27->26 [7] 28/-1/-1->27->26 [8] 28/-1/-1->27->26 [9] 28/-1/-1->27->26 [10] 28/-1/-1->27->26 [11] 28/11/-1->27->-1 [12] -1/-1/-1->27->26 [13] 28/-1/-1->27->26 [14] 28/-1/-1->27->26 [15] 28/-1/-1->27->26
+ip-26-0-168-52:2722047:2722542 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-168-52:2722048:2722545 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-59:1557268:1557764 [2] NCCL INFO NVLS Head  5:  5 13 21 29
+ip-26-0-165-59:1557271:1557762 [5] NCCL INFO NVLS Head  7:  7 15 23 31
+ip-26-0-165-59:1557270:1557763 [4] NCCL INFO NVLS Head  5:  5 13 21 29
+ip-26-0-165-59:1557273:1557760 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-59:1557268:1557764 [2] NCCL INFO NVLS Head  6:  6 14 22 30
+ip-26-0-165-59:1557270:1557763 [4] NCCL INFO NVLS Head  6:  6 14 22 30
+ip-26-0-165-59:1557268:1557764 [2] NCCL INFO NVLS Head  7:  7 15 23 31
+ip-26-0-165-59:1557270:1557763 [4] NCCL INFO NVLS Head  7:  7 15 23 31
+ip-26-0-168-34:272152:272647 [2] NCCL INFO Trees [0] 19/-1/-1->18->17 [1] 19/-1/-1->18->17 [2] 19/10/26->18->2 [3] -1/-1/-1->18->17 [4] 19/-1/-1->18->17 [5] 19/-1/-1->18->17 [6] 19/-1/-1->18->17 [7] 19/-1/-1->18->17 [8] 19/-1/-1->18->17 [9] 19/-1/-1->18->17 [10] 19/-1/-1->18->10 [11] -1/-1/-1->18->17 [12] 19/-1/-1->18->17 [13] 19/-1/-1->18->17 [14] 19/-1/-1->18->17 [15] 19/-1/-1->18->17
+ip-26-0-168-34:272152:272647 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-168-52:2722049:2722547 [4] NCCL INFO Trees [0] 29/-1/-1->28->27 [1] 29/-1/-1->28->27 [2] 29/-1/-1->28->27 [3] 29/-1/-1->28->27 [4] 29/-1/-1->28->20 [5] -1/-1/-1->28->27 [6] 29/-1/-1->28->27 [7] 29/-1/-1->28->27 [8] 29/-1/-1->28->27 [9] 29/-1/-1->28->27 [10] 29/-1/-1->28->27 [11] 29/-1/-1->28->27 [12] 29/12/-1->28->-1 [13] -1/-1/-1->28->27 [14] 29/-1/-1->28->27 [15] 29/-1/-1->28->27
+ip-26-0-165-59:1557272:1557766 [6] NCCL INFO Trees [0] 7/-1/-1->6->5 [1] 7/-1/-1->6->5 [2] 7/-1/-1->6->5 [3] 7/-1/-1->6->5 [4] 7/-1/-1->6->5 [5] 7/-1/-1->6->5 [6] 7/22/-1->6->-1 [7] -1/-1/-1->6->5 [8] 7/-1/-1->6->5 [9] 7/-1/-1->6->5 [10] 7/-1/-1->6->5 [11] 7/-1/-1->6->5 [12] 7/-1/-1->6->5 [13] 7/-1/-1->6->5 [14] 7/-1/-1->6->14 [15] -1/-1/-1->6->5
+ip-26-0-165-59:1557272:1557766 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-59:1557271:1557762 [5] NCCL INFO Trees [0] 6/-1/-1->5->4 [1] 6/-1/-1->5->4 [2] 6/-1/-1->5->4 [3] 6/-1/-1->5->4 [4] 6/-1/-1->5->4 [5] 6/21/-1->5->-1 [6] -1/-1/-1->5->4 [7] 6/-1/-1->5->4 [8] 6/-1/-1->5->4 [9] 6/-1/-1->5->4 [10] 6/-1/-1->5->4 [11] 6/-1/-1->5->4 [12] 6/-1/-1->5->4 [13] 6/-1/-1->5->13 [14] -1/-1/-1->5->4 [15] 6/-1/-1->5->4
+ip-26-0-168-52:2722049:2722547 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-59:1557271:1557762 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-59:1557268:1557764 [2] NCCL INFO Trees [0] 3/-1/-1->2->1 [1] 3/-1/-1->2->1 [2] 3/18/-1->2->-1 [3] -1/-1/-1->2->1 [4] 3/-1/-1->2->1 [5] 3/-1/-1->2->1 [6] 3/-1/-1->2->1 [7] 3/-1/-1->2->1 [8] 3/-1/-1->2->1 [9] 3/-1/-1->2->1 [10] 3/-1/-1->2->10 [11] -1/-1/-1->2->1 [12] 3/-1/-1->2->1 [13] 3/-1/-1->2->1 [14] 3/-1/-1->2->1 [15] 3/-1/-1->2->1
+ip-26-0-165-59:1557270:1557763 [4] NCCL INFO Trees [0] 5/-1/-1->4->3 [1] 5/-1/-1->4->3 [2] 5/-1/-1->4->3 [3] 5/-1/-1->4->3 [4] 5/20/-1->4->-1 [5] -1/-1/-1->4->3 [6] 5/-1/-1->4->3 [7] 5/-1/-1->4->3 [8] 5/-1/-1->4->3 [9] 5/-1/-1->4->3 [10] 5/-1/-1->4->3 [11] 5/-1/-1->4->3 [12] 5/-1/-1->4->12 [13] -1/-1/-1->4->3 [14] 5/-1/-1->4->3 [15] 5/-1/-1->4->3
+ip-26-0-165-59:1557270:1557763 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-59:1557268:1557764 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-59:1557269:1557765 [3] NCCL INFO NVLS Head  0:  0  8 16 24
+ip-26-0-165-59:1557269:1557765 [3] NCCL INFO NVLS Head  1:  1  9 17 25
+ip-26-0-165-59:1557269:1557765 [3] NCCL INFO NVLS Head  2:  2 10 18 26
+ip-26-0-165-59:1557269:1557765 [3] NCCL INFO NVLS Head  3:  3 11 19 27
+ip-26-0-165-59:1557269:1557765 [3] NCCL INFO NVLS Head  4:  4 12 20 28
+ip-26-0-165-59:1557269:1557765 [3] NCCL INFO NVLS Head  5:  5 13 21 29
+ip-26-0-165-59:1557269:1557765 [3] NCCL INFO NVLS Head  6:  6 14 22 30
+ip-26-0-165-59:1557269:1557765 [3] NCCL INFO NVLS Head  7:  7 15 23 31
+ip-26-0-165-59:1557269:1557765 [3] NCCL INFO Trees [0] 4/-1/-1->3->2 [1] 4/-1/-1->3->2 [2] 4/-1/-1->3->2 [3] 4/19/-1->3->-1 [4] -1/-1/-1->3->2 [5] 4/-1/-1->3->2 [6] 4/-1/-1->3->2 [7] 4/-1/-1->3->2 [8] 4/-1/-1->3->2 [9] 4/-1/-1->3->2 [10] 4/-1/-1->3->2 [11] 4/-1/-1->3->11 [12] -1/-1/-1->3->2 [13] 4/-1/-1->3->2 [14] 4/-1/-1->3->2 [15] 4/-1/-1->3->2
+ip-26-0-165-59:1557269:1557765 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-153:2636084:2636605 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-161-153:2636084:2636605 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-165-59:1557271:1557762 [5] NCCL INFO Channel 00/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-168-34:272154:272650 [4] NCCL INFO Channel 00/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-168-34:272156:272648 [6] NCCL INFO Channel 00/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-168-30:235652:236147 [4] NCCL INFO Channel 00/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-165-59:1557271:1557762 [5] NCCL INFO Channel 08/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-165-59:1557272:1557766 [6] NCCL INFO Channel 00/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-168-34:272155:272652 [5] NCCL INFO Channel 00/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-165-59:1557267:1557761 [1] NCCL INFO Channel 00/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-168-34:272154:272650 [4] NCCL INFO Channel 08/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-168-34:272153:272651 [3] NCCL INFO Channel 00/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-165-131:1630205:1630727 [0] NCCL INFO Channel 00/0 : 7[7] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557268:1557764 [2] NCCL INFO Channel 00/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-59:1557270:1557763 [4] NCCL INFO Channel 00/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-131:1630205:1630727 [0] NCCL INFO Channel 08/0 : 7[7] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557269:1557765 [3] NCCL INFO Channel 00/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-131:1630205:1630727 [0] NCCL INFO Channel 00/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-168-52:2722051:2722543 [6] NCCL INFO Channel 00/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-168-30:235650:236144 [2] NCCL INFO Channel 00/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-165-131:1630212:1630734 [7] NCCL INFO Channel 00/0 : 15[7] -> 16[0] [send] via NET/Libfabric/0(8)/GDRDMA
+ip-26-0-168-34:272156:272648 [6] NCCL INFO Channel 08/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-168-30:235654:236146 [6] NCCL INFO Channel 00/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-168-34:272151:272646 [1] NCCL INFO Channel 00/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-168-34:272152:272647 [2] NCCL INFO Channel 00/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-165-202:153562:154061 [0] NCCL INFO Channel 00/0 : 23[7] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235649:236143 [1] NCCL INFO Channel 00/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-165-202:153562:154061 [0] NCCL INFO Channel 08/0 : 23[7] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272155:272652 [5] NCCL INFO Channel 08/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-165-202:153562:154061 [0] NCCL INFO Channel 00/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-165-59:1557272:1557766 [6] NCCL INFO Channel 08/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-168-52:2722046:2722541 [1] NCCL INFO Channel 00/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-165-202:153569:154055 [7] NCCL INFO Channel 00/0 : 31[7] -> 0[0] [send] via NET/Libfabric/0(24)/GDRDMA
+ip-26-0-165-131:1630205:1630727 [0] NCCL INFO Channel 08/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-165-131:1630212:1630734 [7] NCCL INFO Channel 08/0 : 15[7] -> 16[0] [send] via NET/Libfabric/0(8)/GDRDMA
+ip-26-0-168-30:235651:236141 [3] NCCL INFO Channel 00/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-168-30:235653:236140 [5] NCCL INFO Channel 00/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-165-59:1557267:1557761 [1] NCCL INFO Channel 08/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-202:153562:154061 [0] NCCL INFO Channel 08/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-165-202:153569:154055 [7] NCCL INFO Channel 08/0 : 31[7] -> 0[0] [send] via NET/Libfabric/0(24)/GDRDMA
+ip-26-0-168-52:2722047:2722542 [2] NCCL INFO Channel 00/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-168-34:272153:272651 [3] NCCL INFO Channel 08/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-161-78:114512:115007 [0] NCCL INFO Channel 00/0 : 31[7] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557270:1557763 [4] NCCL INFO Channel 08/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-59:1557268:1557764 [2] NCCL INFO Channel 08/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-59:1557269:1557765 [3] NCCL INFO Channel 08/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-168-52:2722048:2722545 [3] NCCL INFO Channel 00/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-168-52:2722051:2722543 [6] NCCL INFO Channel 08/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-161-78:114512:115007 [0] NCCL INFO Channel 08/0 : 31[7] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235650:236144 [2] NCCL INFO Channel 08/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-168-30:235654:236146 [6] NCCL INFO Channel 08/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-161-78:114512:115007 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-168-34:272152:272647 [2] NCCL INFO Channel 08/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-165-131:1630205:1630727 [0] NCCL INFO Channel 05/0 : 8[0] -> 12[4] via P2P/IPC
+ip-26-0-168-30:235649:236143 [1] NCCL INFO Channel 08/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-168-52:2722050:2722548 [5] NCCL INFO Channel 00/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-168-52:2722049:2722547 [4] NCCL INFO Channel 00/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-165-164:1227463:1227964 [0] NCCL INFO Channel 00/0 : 15[7] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-52:2722046:2722541 [1] NCCL INFO Channel 08/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-165-164:1227463:1227964 [0] NCCL INFO Channel 08/0 : 15[7] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227463:1227964 [0] NCCL INFO Channel 00/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-161-78:114512:115007 [0] NCCL INFO Channel 08/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-168-34:272151:272646 [1] NCCL INFO Channel 08/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-165-164:1227470:1227958 [7] NCCL INFO Channel 00/0 : 23[7] -> 24[0] [send] via NET/Libfabric/0(16)/GDRDMA
+ip-26-0-168-30:235653:236140 [5] NCCL INFO Channel 08/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-168-34:272154:272650 [4] NCCL INFO Channel 05/0 : 20[4] -> 23[7] via P2P/IPC
+ip-26-0-165-131:1630205:1630727 [0] NCCL INFO Channel 13/0 : 8[0] -> 12[4] via P2P/IPC
+ip-26-0-168-30:235651:236141 [3] NCCL INFO Channel 08/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-165-59:1557270:1557763 [4] NCCL INFO Channel 05/0 : 4[4] -> 7[7] via P2P/IPC
+ip-26-0-165-164:1227463:1227964 [0] NCCL INFO Channel 08/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-168-52:2722047:2722542 [2] NCCL INFO Channel 08/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-165-164:1227470:1227958 [7] NCCL INFO Channel 08/0 : 23[7] -> 24[0] [send] via NET/Libfabric/0(16)/GDRDMA
+ip-26-0-168-34:272154:272650 [4] NCCL INFO Channel 06/0 : 20[4] -> 23[7] via P2P/IPC
+ip-26-0-168-52:2722048:2722545 [3] NCCL INFO Channel 08/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-165-59:1557270:1557763 [4] NCCL INFO Channel 06/0 : 4[4] -> 7[7] via P2P/IPC
+ip-26-0-161-78:114512:115007 [0] NCCL INFO Channel 05/0 : 0[0] -> 4[4] via P2P/IPC
+ip-26-0-168-30:235652:236147 [4] NCCL INFO Channel 08/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-165-202:153562:154061 [0] NCCL INFO Channel 05/0 : 24[0] -> 28[4] via P2P/IPC
+ip-26-0-161-78:114512:115007 [0] NCCL INFO Channel 13/0 : 0[0] -> 4[4] via P2P/IPC
+ip-26-0-168-52:2722050:2722548 [5] NCCL INFO Channel 08/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-168-52:2722049:2722547 [4] NCCL INFO Channel 08/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-168-34:272154:272650 [4] NCCL INFO Channel 13/0 : 20[4] -> 23[7] via P2P/IPC
+ip-26-0-165-59:1557270:1557763 [4] NCCL INFO Channel 13/0 : 4[4] -> 7[7] via P2P/IPC
+ip-26-0-165-164:1227463:1227964 [0] NCCL INFO Channel 05/0 : 16[0] -> 20[4] via P2P/IPC
+ip-26-0-165-202:153562:154061 [0] NCCL INFO Channel 13/0 : 24[0] -> 28[4] via P2P/IPC
+ip-26-0-168-34:272154:272650 [4] NCCL INFO Channel 14/0 : 20[4] -> 23[7] via P2P/IPC
+ip-26-0-168-30:235652:236147 [4] NCCL INFO Channel 05/0 : 12[4] -> 15[7] via P2P/IPC
+ip-26-0-165-59:1557270:1557763 [4] NCCL INFO Channel 14/0 : 4[4] -> 7[7] via P2P/IPC
+ip-26-0-168-30:235652:236147 [4] NCCL INFO Channel 06/0 : 12[4] -> 15[7] via P2P/IPC
+ip-26-0-168-52:2722049:2722547 [4] NCCL INFO Channel 05/0 : 28[4] -> 31[7] via P2P/IPC
+ip-26-0-165-164:1227466:1227961 [3] NCCL INFO Channel 03/0 : 12[4] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227466:1227961 [3] NCCL INFO Channel 11/0 : 12[4] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235652:236147 [4] NCCL INFO Channel 13/0 : 12[4] -> 15[7] via P2P/IPC
+ip-26-0-165-164:1227463:1227964 [0] NCCL INFO Channel 13/0 : 16[0] -> 20[4] via P2P/IPC
+ip-26-0-168-52:2722049:2722547 [4] NCCL INFO Channel 06/0 : 28[4] -> 31[7] via P2P/IPC
+ip-26-0-168-30:235652:236147 [4] NCCL INFO Channel 14/0 : 12[4] -> 15[7] via P2P/IPC
+ip-26-0-168-52:2722049:2722547 [4] NCCL INFO Channel 13/0 : 28[4] -> 31[7] via P2P/IPC
+ip-26-0-168-52:2722049:2722547 [4] NCCL INFO Channel 14/0 : 28[4] -> 31[7] via P2P/IPC
+ip-26-0-165-131:1630207:1630731 [2] NCCL INFO Channel 02/0 : 3[3] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630207:1630731 [2] NCCL INFO Channel 10/0 : 3[3] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630208:1630729 [3] NCCL INFO Channel 03/0 : 4[4] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630208:1630729 [3] NCCL INFO Channel 11/0 : 4[4] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630208:1630729 [3] NCCL INFO Channel 02/0 : 11[3] -> 18[2] [send] via NET/Libfabric/2(10)/GDRDMA
+ip-26-0-165-164:1227465:1227962 [2] NCCL INFO Channel 02/0 : 11[3] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630208:1630729 [3] NCCL INFO Channel 10/0 : 11[3] -> 18[2] [send] via NET/Libfabric/2(10)/GDRDMA
+ip-26-0-165-164:1227465:1227962 [2] NCCL INFO Channel 10/0 : 11[3] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227466:1227961 [3] NCCL INFO Channel 02/0 : 19[3] -> 26[2] [send] via NET/Libfabric/2(18)/GDRDMA
+ip-26-0-161-78:114512:115007 [0] NCCL INFO Channel 06/0 : 0[0] -> 5[5] via P2P/IPC
+ip-26-0-165-164:1227466:1227961 [3] NCCL INFO Channel 10/0 : 19[3] -> 26[2] [send] via NET/Libfabric/2(18)/GDRDMA
+ip-26-0-165-131:1630205:1630727 [0] NCCL INFO Channel 06/0 : 8[0] -> 13[5] via P2P/IPC
+ip-26-0-165-202:153564:154057 [2] NCCL INFO Channel 02/0 : 19[3] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114515:115009 [3] NCCL INFO Channel 03/0 : 28[4] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153564:154057 [2] NCCL INFO Channel 10/0 : 19[3] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114515:115009 [3] NCCL INFO Channel 11/0 : 28[4] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114512:115007 [0] NCCL INFO Channel 14/0 : 0[0] -> 5[5] via P2P/IPC
+ip-26-0-165-131:1630205:1630727 [0] NCCL INFO Channel 14/0 : 8[0] -> 13[5] via P2P/IPC
+ip-26-0-165-202:153562:154061 [0] NCCL INFO Channel 06/0 : 24[0] -> 29[5] via P2P/IPC
+ip-26-0-165-164:1227463:1227964 [0] NCCL INFO Channel 06/0 : 16[0] -> 21[5] via P2P/IPC
+ip-26-0-161-78:114512:115007 [0] NCCL INFO Channel 07/0 : 0[0] -> 6[6] via P2P/IPC
+ip-26-0-165-131:1630205:1630727 [0] NCCL INFO Channel 07/0 : 8[0] -> 14[6] via P2P/IPC
+ip-26-0-165-202:153565:154058 [3] NCCL INFO Channel 03/0 : 20[4] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153562:154061 [0] NCCL INFO Channel 14/0 : 24[0] -> 29[5] via P2P/IPC
+ip-26-0-165-202:153565:154058 [3] NCCL INFO Channel 11/0 : 20[4] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153565:154058 [3] NCCL INFO Channel 02/0 : 27[3] -> 2[2] [send] via NET/Libfabric/2(26)/GDRDMA
+ip-26-0-165-131:1630205:1630727 [0] NCCL INFO Channel 15/0 : 8[0] -> 14[6] via P2P/IPC
+ip-26-0-161-78:114512:115007 [0] NCCL INFO Channel 15/0 : 0[0] -> 6[6] via P2P/IPC
+ip-26-0-165-202:153565:154058 [3] NCCL INFO Channel 10/0 : 27[3] -> 2[2] [send] via NET/Libfabric/2(26)/GDRDMA
+ip-26-0-165-202:153562:154061 [0] NCCL INFO Channel 07/0 : 24[0] -> 30[6] via P2P/IPC
+ip-26-0-165-202:153569:154055 [7] NCCL INFO Channel 06/0 : 31[7] -> 6[6] [send] via NET/Libfabric/6(30)/GDRDMA
+ip-26-0-165-202:153569:154055 [7] NCCL INFO Channel 14/0 : 31[7] -> 6[6] [send] via NET/Libfabric/6(30)/GDRDMA
+ip-26-0-165-131:1630205:1630727 [0] NCCL INFO Channel 01/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-161-78:114512:115007 [0] NCCL INFO Channel 01/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-165-164:1227463:1227964 [0] NCCL INFO Channel 14/0 : 16[0] -> 21[5] via P2P/IPC
+ip-26-0-165-131:1630205:1630727 [0] NCCL INFO Channel 02/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-161-78:114512:115007 [0] NCCL INFO Channel 02/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-165-131:1630205:1630727 [0] NCCL INFO Channel 03/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-165-131:1630212:1630734 [7] NCCL INFO Channel 06/0 : 15[7] -> 22[6] [send] via NET/Libfabric/6(14)/GDRDMA
+ip-26-0-161-78:114512:115007 [0] NCCL INFO Channel 03/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-165-202:153562:154061 [0] NCCL INFO Channel 15/0 : 24[0] -> 30[6] via P2P/IPC
+ip-26-0-165-131:1630205:1630727 [0] NCCL INFO Channel 04/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-161-78:114512:115007 [0] NCCL INFO Channel 04/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-165-131:1630205:1630727 [0] NCCL INFO Channel 09/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-165-202:153562:154061 [0] NCCL INFO Channel 01/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-161-78:114512:115007 [0] NCCL INFO Channel 09/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-165-131:1630205:1630727 [0] NCCL INFO Channel 10/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-165-202:153562:154061 [0] NCCL INFO Channel 02/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-161-78:114512:115007 [0] NCCL INFO Channel 10/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-165-131:1630205:1630727 [0] NCCL INFO Channel 11/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-165-202:153562:154061 [0] NCCL INFO Channel 03/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-161-78:114512:115007 [0] NCCL INFO Channel 11/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-165-131:1630205:1630727 [0] NCCL INFO Channel 12/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-165-202:153562:154061 [0] NCCL INFO Channel 04/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-161-78:114512:115007 [0] NCCL INFO Channel 12/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-165-202:153562:154061 [0] NCCL INFO Channel 09/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-165-164:1227463:1227964 [0] NCCL INFO Channel 07/0 : 16[0] -> 22[6] via P2P/IPC
+ip-26-0-165-202:153562:154061 [0] NCCL INFO Channel 10/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-165-202:153562:154061 [0] NCCL INFO Channel 11/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-165-164:1227463:1227964 [0] NCCL INFO Channel 15/0 : 16[0] -> 22[6] via P2P/IPC
+ip-26-0-165-202:153562:154061 [0] NCCL INFO Channel 12/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-161-78:114514:115011 [2] NCCL INFO Channel 02/0 : 27[3] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114514:115011 [2] NCCL INFO Channel 10/0 : 27[3] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114515:115009 [3] NCCL INFO Channel 02/0 : 3[3] -> 10[2] [send] via NET/Libfabric/2(2)/GDRDMA
+ip-26-0-165-164:1227463:1227964 [0] NCCL INFO Channel 01/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-161-78:114515:115009 [3] NCCL INFO Channel 10/0 : 3[3] -> 10[2] [send] via NET/Libfabric/2(2)/GDRDMA
+ip-26-0-165-164:1227463:1227964 [0] NCCL INFO Channel 02/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-165-164:1227463:1227964 [0] NCCL INFO Channel 03/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-165-164:1227463:1227964 [0] NCCL INFO Channel 04/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-160-242:1364904:1365408 [2] NCCL INFO Trees [0] 11/-1/-1->10->9 [1] 11/-1/-1->10->9 [2] 11/-1/-1->10->18 [3] -1/-1/-1->10->9 [4] 11/-1/-1->10->9 [5] 11/-1/-1->10->9 [6] 11/-1/-1->10->9 [7] 11/-1/-1->10->9 [8] 11/-1/-1->10->9 [9] 11/-1/-1->10->9 [10] 11/18/2->10->26 [11] -1/-1/-1->10->9 [12] 11/-1/-1->10->9 [13] 11/-1/-1->10->9 [14] 11/-1/-1->10->9 [15] 11/-1/-1->10->9
+ip-26-0-160-242:1364904:1365408 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-242:1364905:1365404 [3] NCCL INFO Trees [0] 12/-1/-1->11->10 [1] 12/-1/-1->11->10 [2] 12/-1/-1->11->10 [3] 12/-1/-1->11->19 [4] -1/-1/-1->11->10 [5] 12/-1/-1->11->10 [6] 12/-1/-1->11->10 [7] 12/-1/-1->11->10 [8] 12/-1/-1->11->10 [9] 12/-1/-1->11->10 [10] 12/-1/-1->11->10 [11] 12/19/3->11->27 [12] -1/-1/-1->11->10 [13] 12/-1/-1->11->10 [14] 12/-1/-1->11->10 [15] 12/-1/-1->11->10
+ip-26-0-160-242:1364905:1365404 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-225:124167:124664 [7] NCCL INFO NVLS Head  0:  0  8 16 24
+ip-26-0-160-225:124167:124664 [7] NCCL INFO NVLS Head  1:  1  9 17 25
+ip-26-0-160-225:124167:124664 [7] NCCL INFO NVLS Head  2:  2 10 18 26
+ip-26-0-160-225:124167:124664 [7] NCCL INFO NVLS Head  3:  3 11 19 27
+ip-26-0-160-225:124167:124664 [7] NCCL INFO NVLS Head  4:  4 12 20 28
+ip-26-0-160-225:124167:124664 [7] NCCL INFO NVLS Head  5:  5 13 21 29
+ip-26-0-160-225:124167:124664 [7] NCCL INFO NVLS Head  6:  6 14 22 30
+ip-26-0-161-103:96893:97384 [5] NCCL INFO Trees [0] 22/-1/-1->21->20 [1] 22/-1/-1->21->20 [2] 22/-1/-1->21->20 [3] 22/-1/-1->21->20 [4] 22/-1/-1->21->20 [5] 22/13/29->21->5 [6] -1/-1/-1->21->20 [7] 22/-1/-1->21->20 [8] 22/-1/-1->21->20 [9] 22/-1/-1->21->20 [10] 22/-1/-1->21->20 [11] 22/-1/-1->21->20 [12] 22/-1/-1->21->20 [13] 22/-1/-1->21->13 [14] -1/-1/-1->21->20 [15] 22/-1/-1->21->20
+ip-26-0-161-103:96893:97384 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-103:96894:97381 [6] NCCL INFO Trees [0] 23/-1/-1->22->21 [1] 23/-1/-1->22->21 [2] 23/-1/-1->22->21 [3] 23/-1/-1->22->21 [4] 23/-1/-1->22->21 [5] 23/-1/-1->22->21 [6] 23/14/30->22->6 [7] -1/-1/-1->22->21 [8] 23/-1/-1->22->21 [9] 23/-1/-1->22->21 [10] 23/-1/-1->22->21 [11] 23/-1/-1->22->21 [12] 23/-1/-1->22->21 [13] 23/-1/-1->22->21 [14] 23/-1/-1->22->14 [15] -1/-1/-1->22->21
+ip-26-0-160-242:1364906:1365407 [4] NCCL INFO Trees [0] 13/-1/-1->12->11 [1] 13/-1/-1->12->11 [2] 13/-1/-1->12->11 [3] 13/-1/-1->12->11 [4] 13/-1/-1->12->20 [5] -1/-1/-1->12->11 [6] 13/-1/-1->12->11 [7] 13/-1/-1->12->11 [8] 13/-1/-1->12->11 [9] 13/-1/-1->12->11 [10] 13/-1/-1->12->11 [11] 13/-1/-1->12->11 [12] 13/20/4->12->28 [13] -1/-1/-1->12->11 [14] 13/-1/-1->12->11 [15] 13/-1/-1->12->11
+ip-26-0-160-242:1364906:1365407 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-103:96894:97381 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-103:96895:97382 [7] NCCL INFO Trees [0] -1/-1/-1->23->22 [1] 16/-1/-1->23->22 [2] 16/-1/-1->23->22 [3] 16/-1/-1->23->22 [4] 16/-1/-1->23->22 [5] 16/-1/-1->23->22 [6] 16/-1/-1->23->22 [7] 16/15/31->23->7 [8] -1/-1/-1->23->22 [9] 16/-1/-1->23->22 [10] 16/-1/-1->23->22 [11] 16/-1/-1->23->22 [12] 16/-1/-1->23->22 [13] 16/-1/-1->23->22 [14] 16/-1/-1->23->22 [15] 16/-1/-1->23->15
+ip-26-0-161-103:96895:97382 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-123:32706:33199 [2] NCCL INFO Trees [0] 27/-1/-1->26->25 [1] 27/-1/-1->26->25 [2] 27/-1/-1->26->18 [3] -1/-1/-1->26->25 [4] 27/-1/-1->26->25 [5] 27/-1/-1->26->25 [6] 27/-1/-1->26->25 [7] 27/-1/-1->26->25 [8] 27/-1/-1->26->25 [9] 27/-1/-1->26->25 [10] 27/10/-1->26->-1 [11] -1/-1/-1->26->25 [12] 27/-1/-1->26->25 [13] 27/-1/-1->26->25 [14] 27/-1/-1->26->25 [15] 27/-1/-1->26->25
+ip-26-0-161-123:32705:33203 [1] NCCL INFO Trees [0] 26/-1/-1->25->24 [1] 26/-1/-1->25->17 [2] -1/-1/-1->25->24 [3] 26/-1/-1->25->24 [4] 26/-1/-1->25->24 [5] 26/-1/-1->25->24 [6] 26/-1/-1->25->24 [7] 26/-1/-1->25->24 [8] 26/-1/-1->25->24 [9] 26/9/-1->25->-1 [10] -1/-1/-1->25->24 [11] 26/-1/-1->25->24 [12] 26/-1/-1->25->24 [13] 26/-1/-1->25->24 [14] 26/-1/-1->25->24 [15] 26/-1/-1->25->24
+ip-26-0-161-123:32706:33199 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-123:32705:33203 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-164:1227463:1227964 [0] NCCL INFO Channel 09/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-160-225:124167:124664 [7] NCCL INFO NVLS Head  7:  7 15 23 31
+ip-26-0-160-225:124166:124665 [6] NCCL INFO NVLS Head  0:  0  8 16 24
+ip-26-0-160-225:124166:124665 [6] NCCL INFO NVLS Head  1:  1  9 17 25
+ip-26-0-160-225:124166:124665 [6] NCCL INFO NVLS Head  2:  2 10 18 26
+ip-26-0-160-225:124166:124665 [6] NCCL INFO NVLS Head  3:  3 11 19 27
+ip-26-0-160-225:124166:124665 [6] NCCL INFO NVLS Head  4:  4 12 20 28
+ip-26-0-160-225:124166:124665 [6] NCCL INFO NVLS Head  5:  5 13 21 29
+ip-26-0-160-225:124166:124665 [6] NCCL INFO NVLS Head  6:  6 14 22 30
+ip-26-0-160-225:124164:124663 [4] NCCL INFO NVLS Head  0:  0  8 16 24
+ip-26-0-160-225:124166:124665 [6] NCCL INFO NVLS Head  7:  7 15 23 31
+ip-26-0-161-103:96892:97383 [4] NCCL INFO Trees [0] 21/-1/-1->20->19 [1] 21/-1/-1->20->19 [2] 21/-1/-1->20->19 [3] 21/-1/-1->20->19 [4] 21/12/28->20->4 [5] -1/-1/-1->20->19 [6] 21/-1/-1->20->19 [7] 21/-1/-1->20->19 [8] 21/-1/-1->20->19 [9] 21/-1/-1->20->19 [10] 21/-1/-1->20->19 [11] 21/-1/-1->20->19 [12] 21/-1/-1->20->12 [13] -1/-1/-1->20->19 [14] 21/-1/-1->20->19 [15] 21/-1/-1->20->19
+ip-26-0-161-103:96892:97383 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-103:96889:97388 [1] NCCL INFO Trees [0] 18/-1/-1->17->16 [1] 18/9/25->17->1 [2] -1/-1/-1->17->16 [3] 18/-1/-1->17->16 [4] 18/-1/-1->17->16 [5] 18/-1/-1->17->16 [6] 18/-1/-1->17->16 [7] 18/-1/-1->17->16 [8] 18/-1/-1->17->16 [9] 18/-1/-1->17->9 [10] -1/-1/-1->17->16 [11] 18/-1/-1->17->16 [12] 18/-1/-1->17->16 [13] 18/-1/-1->17->16 [14] 18/-1/-1->17->16 [15] 18/-1/-1->17->16
+ip-26-0-161-123:32711:33204 [7] NCCL INFO Trees [0] -1/-1/-1->31->30 [1] 24/-1/-1->31->30 [2] 24/-1/-1->31->30 [3] 24/-1/-1->31->30 [4] 24/-1/-1->31->30 [5] 24/-1/-1->31->30 [6] 24/-1/-1->31->30 [7] 24/-1/-1->31->23 [8] -1/-1/-1->31->30 [9] 24/-1/-1->31->30 [10] 24/-1/-1->31->30 [11] 24/-1/-1->31->30 [12] 24/-1/-1->31->30 [13] 24/-1/-1->31->30 [14] 24/-1/-1->31->30 [15] 24/15/-1->31->-1
+ip-26-0-160-225:124167:124664 [7] NCCL INFO Trees [0] -1/-1/-1->7->6 [1] 0/-1/-1->7->6 [2] 0/-1/-1->7->6 [3] 0/-1/-1->7->6 [4] 0/-1/-1->7->6 [5] 0/-1/-1->7->6 [6] 0/-1/-1->7->6 [7] 0/23/-1->7->-1 [8] -1/-1/-1->7->6 [9] 0/-1/-1->7->6 [10] 0/-1/-1->7->6 [11] 0/-1/-1->7->6 [12] 0/-1/-1->7->6 [13] 0/-1/-1->7->6 [14] 0/-1/-1->7->6 [15] 0/-1/-1->7->15
+ip-26-0-160-225:124164:124663 [4] NCCL INFO NVLS Head  1:  1  9 17 25
+ip-26-0-160-225:124164:124663 [4] NCCL INFO NVLS Head  2:  2 10 18 26
+ip-26-0-160-225:124167:124664 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-225:124164:124663 [4] NCCL INFO NVLS Head  3:  3 11 19 27
+ip-26-0-160-225:124164:124663 [4] NCCL INFO NVLS Head  4:  4 12 20 28
+ip-26-0-160-225:124165:124662 [5] NCCL INFO NVLS Head  0:  0  8 16 24
+ip-26-0-160-225:124164:124663 [4] NCCL INFO NVLS Head  5:  5 13 21 29
+ip-26-0-160-225:124164:124663 [4] NCCL INFO NVLS Head  6:  6 14 22 30
+ip-26-0-160-225:124165:124662 [5] NCCL INFO NVLS Head  1:  1  9 17 25
+ip-26-0-161-103:96889:97388 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-103:96890:97386 [2] NCCL INFO Trees [0] 19/-1/-1->18->17 [1] 19/-1/-1->18->17 [2] 19/10/26->18->2 [3] -1/-1/-1->18->17 [4] 19/-1/-1->18->17 [5] 19/-1/-1->18->17 [6] 19/-1/-1->18->17 [7] 19/-1/-1->18->17 [8] 19/-1/-1->18->17 [9] 19/-1/-1->18->17 [10] 19/-1/-1->18->10 [11] -1/-1/-1->18->17 [12] 19/-1/-1->18->17 [13] 19/-1/-1->18->17 [14] 19/-1/-1->18->17 [15] 19/-1/-1->18->17
+ip-26-0-161-103:96890:97386 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-242:1364902:1365410 [0] NCCL INFO Trees [0] 9/-1/-1->8->16 [1] -1/-1/-1->8->15 [2] 9/-1/-1->8->15 [3] 9/-1/-1->8->15 [4] 9/-1/-1->8->15 [5] 9/-1/-1->8->15 [6] 9/-1/-1->8->15 [7] 9/-1/-1->8->15 [8] 9/16/0->8->24 [9] -1/-1/-1->8->15 [10] 9/-1/-1->8->15 [11] 9/-1/-1->8->15 [12] 9/-1/-1->8->15 [13] 9/-1/-1->8->15 [14] 9/-1/-1->8->15 [15] 9/-1/-1->8->15
+ip-26-0-160-242:1364909:1365405 [7] NCCL INFO Trees [0] -1/-1/-1->15->14 [1] 8/-1/-1->15->14 [2] 8/-1/-1->15->14 [3] 8/-1/-1->15->14 [4] 8/-1/-1->15->14 [5] 8/-1/-1->15->14 [6] 8/-1/-1->15->14 [7] 8/-1/-1->15->23 [8] -1/-1/-1->15->14 [9] 8/-1/-1->15->14 [10] 8/-1/-1->15->14 [11] 8/-1/-1->15->14 [12] 8/-1/-1->15->14 [13] 8/-1/-1->15->14 [14] 8/-1/-1->15->14 [15] 8/23/7->15->31
+ip-26-0-161-123:32711:33204 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-123:32709:33201 [5] NCCL INFO Trees [0] 30/-1/-1->29->28 [1] 30/-1/-1->29->28 [2] 30/-1/-1->29->28 [3] 30/-1/-1->29->28 [4] 30/-1/-1->29->28 [5] 30/-1/-1->29->21 [6] -1/-1/-1->29->28 [7] 30/-1/-1->29->28 [8] 30/-1/-1->29->28 [9] 30/-1/-1->29->28 [10] 30/-1/-1->29->28 [11] 30/-1/-1->29->28 [12] 30/-1/-1->29->28 [13] 30/13/-1->29->-1 [14] -1/-1/-1->29->28 [15] 30/-1/-1->29->28
+ip-26-0-161-123:32710:33202 [6] NCCL INFO Trees [0] 31/-1/-1->30->29 [1] 31/-1/-1->30->29 [2] 31/-1/-1->30->29 [3] 31/-1/-1->30->29 [4] 31/-1/-1->30->29 [5] 31/-1/-1->30->29 [6] 31/-1/-1->30->22 [7] -1/-1/-1->30->29 [8] 31/-1/-1->30->29 [9] 31/-1/-1->30->29 [10] 31/-1/-1->30->29 [11] 31/-1/-1->30->29 [12] 31/-1/-1->30->29 [13] 31/-1/-1->30->29 [14] 31/14/-1->30->-1 [15] -1/-1/-1->30->29
+ip-26-0-160-225:124164:124663 [4] NCCL INFO NVLS Head  7:  7 15 23 31
+ip-26-0-160-225:124165:124662 [5] NCCL INFO NVLS Head  2:  2 10 18 26
+ip-26-0-160-225:124165:124662 [5] NCCL INFO NVLS Head  3:  3 11 19 27
+ip-26-0-160-225:124165:124662 [5] NCCL INFO NVLS Head  4:  4 12 20 28
+ip-26-0-160-225:124165:124662 [5] NCCL INFO NVLS Head  5:  5 13 21 29
+ip-26-0-160-225:124165:124662 [5] NCCL INFO NVLS Head  6:  6 14 22 30
+ip-26-0-160-225:124166:124665 [6] NCCL INFO Trees [0] 7/-1/-1->6->5 [1] 7/-1/-1->6->5 [2] 7/-1/-1->6->5 [3] 7/-1/-1->6->5 [4] 7/-1/-1->6->5 [5] 7/-1/-1->6->5 [6] 7/22/-1->6->-1 [7] -1/-1/-1->6->5 [8] 7/-1/-1->6->5 [9] 7/-1/-1->6->5 [10] 7/-1/-1->6->5 [11] 7/-1/-1->6->5 [12] 7/-1/-1->6->5 [13] 7/-1/-1->6->5 [14] 7/-1/-1->6->14 [15] -1/-1/-1->6->5
+ip-26-0-160-225:124165:124662 [5] NCCL INFO NVLS Head  7:  7 15 23 31
+ip-26-0-160-225:124163:124661 [3] NCCL INFO NVLS Head  0:  0  8 16 24
+ip-26-0-160-225:124166:124665 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-103:96888:97387 [0] NCCL INFO Trees [0] 17/8/24->16->0 [1] -1/-1/-1->16->23 [2] 17/-1/-1->16->23 [3] 17/-1/-1->16->23 [4] 17/-1/-1->16->23 [5] 17/-1/-1->16->23 [6] 17/-1/-1->16->23 [7] 17/-1/-1->16->23 [8] 17/-1/-1->16->8 [9] -1/-1/-1->16->23 [10] 17/-1/-1->16->23 [11] 17/-1/-1->16->23 [12] 17/-1/-1->16->23 [13] 17/-1/-1->16->23 [14] 17/-1/-1->16->23 [15] 17/-1/-1->16->23
+ip-26-0-161-103:96891:97385 [3] NCCL INFO Trees [0] 20/-1/-1->19->18 [1] 20/-1/-1->19->18 [2] 20/-1/-1->19->18 [3] 20/11/27->19->3 [4] -1/-1/-1->19->18 [5] 20/-1/-1->19->18 [6] 20/-1/-1->19->18 [7] 20/-1/-1->19->18 [8] 20/-1/-1->19->18 [9] 20/-1/-1->19->18 [10] 20/-1/-1->19->18 [11] 20/-1/-1->19->11 [12] -1/-1/-1->19->18 [13] 20/-1/-1->19->18 [14] 20/-1/-1->19->18 [15] 20/-1/-1->19->18
+ip-26-0-161-103:96888:97387 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-103:96891:97385 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-242:1364902:1365410 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-242:1364909:1365405 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-242:1364908:1365406 [6] NCCL INFO Trees [0] 15/-1/-1->14->13 [1] 15/-1/-1->14->13 [2] 15/-1/-1->14->13 [3] 15/-1/-1->14->13 [4] 15/-1/-1->14->13 [5] 15/-1/-1->14->13 [6] 15/-1/-1->14->22 [7] -1/-1/-1->14->13 [8] 15/-1/-1->14->13 [9] 15/-1/-1->14->13 [10] 15/-1/-1->14->13 [11] 15/-1/-1->14->13 [12] 15/-1/-1->14->13 [13] 15/-1/-1->14->13 [14] 15/22/6->14->30 [15] -1/-1/-1->14->13
+ip-26-0-160-242:1364908:1365406 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-123:32707:33200 [3] NCCL INFO Trees [0] 28/-1/-1->27->26 [1] 28/-1/-1->27->26 [2] 28/-1/-1->27->26 [3] 28/-1/-1->27->19 [4] -1/-1/-1->27->26 [5] 28/-1/-1->27->26 [6] 28/-1/-1->27->26 [7] 28/-1/-1->27->26 [8] 28/-1/-1->27->26 [9] 28/-1/-1->27->26 [10] 28/-1/-1->27->26 [11] 28/11/-1->27->-1 [12] -1/-1/-1->27->26 [13] 28/-1/-1->27->26 [14] 28/-1/-1->27->26 [15] 28/-1/-1->27->26
+ip-26-0-161-123:32709:33201 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-123:32708:33198 [4] NCCL INFO Trees [0] 29/-1/-1->28->27 [1] 29/-1/-1->28->27 [2] 29/-1/-1->28->27 [3] 29/-1/-1->28->27 [4] 29/-1/-1->28->20 [5] -1/-1/-1->28->27 [6] 29/-1/-1->28->27 [7] 29/-1/-1->28->27 [8] 29/-1/-1->28->27 [9] 29/-1/-1->28->27 [10] 29/-1/-1->28->27 [11] 29/-1/-1->28->27 [12] 29/12/-1->28->-1 [13] -1/-1/-1->28->27 [14] 29/-1/-1->28->27 [15] 29/-1/-1->28->27
+ip-26-0-161-123:32710:33202 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-123:32707:33200 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-225:124163:124661 [3] NCCL INFO NVLS Head  1:  1  9 17 25
+ip-26-0-160-225:124163:124661 [3] NCCL INFO NVLS Head  2:  2 10 18 26
+ip-26-0-160-225:124164:124663 [4] NCCL INFO Trees [0] 5/-1/-1->4->3 [1] 5/-1/-1->4->3 [2] 5/-1/-1->4->3 [3] 5/-1/-1->4->3 [4] 5/20/-1->4->-1 [5] -1/-1/-1->4->3 [6] 5/-1/-1->4->3 [7] 5/-1/-1->4->3 [8] 5/-1/-1->4->3 [9] 5/-1/-1->4->3 [10] 5/-1/-1->4->3 [11] 5/-1/-1->4->3 [12] 5/-1/-1->4->12 [13] -1/-1/-1->4->3 [14] 5/-1/-1->4->3 [15] 5/-1/-1->4->3
+ip-26-0-160-225:124163:124661 [3] NCCL INFO NVLS Head  3:  3 11 19 27
+ip-26-0-160-242:1364903:1365409 [1] NCCL INFO Trees [0] 10/-1/-1->9->8 [1] 10/-1/-1->9->17 [2] -1/-1/-1->9->8 [3] 10/-1/-1->9->8 [4] 10/-1/-1->9->8 [5] 10/-1/-1->9->8 [6] 10/-1/-1->9->8 [7] 10/-1/-1->9->8 [8] 10/-1/-1->9->8 [9] 10/17/1->9->25 [10] -1/-1/-1->9->8 [11] 10/-1/-1->9->8 [12] 10/-1/-1->9->8 [13] 10/-1/-1->9->8 [14] 10/-1/-1->9->8 [15] 10/-1/-1->9->8
+ip-26-0-161-123:32708:33198 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-123:32704:33197 [0] NCCL INFO Trees [0] 25/-1/-1->24->16 [1] -1/-1/-1->24->31 [2] 25/-1/-1->24->31 [3] 25/-1/-1->24->31 [4] 25/-1/-1->24->31 [5] 25/-1/-1->24->31 [6] 25/-1/-1->24->31 [7] 25/-1/-1->24->31 [8] 25/8/-1->24->-1 [9] -1/-1/-1->24->31 [10] 25/-1/-1->24->31 [11] 25/-1/-1->24->31 [12] 25/-1/-1->24->31 [13] 25/-1/-1->24->31 [14] 25/-1/-1->24->31 [15] 25/-1/-1->24->31
+ip-26-0-161-123:32704:33197 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-225:124162:124660 [2] NCCL INFO NVLS Head  0:  0  8 16 24
+ip-26-0-160-225:124164:124663 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-225:124163:124661 [3] NCCL INFO NVLS Head  4:  4 12 20 28
+ip-26-0-160-225:124162:124660 [2] NCCL INFO NVLS Head  1:  1  9 17 25
+ip-26-0-160-225:124163:124661 [3] NCCL INFO NVLS Head  5:  5 13 21 29
+ip-26-0-160-225:124165:124662 [5] NCCL INFO Trees [0] 6/-1/-1->5->4 [1] 6/-1/-1->5->4 [2] 6/-1/-1->5->4 [3] 6/-1/-1->5->4 [4] 6/-1/-1->5->4 [5] 6/21/-1->5->-1 [6] -1/-1/-1->5->4 [7] 6/-1/-1->5->4 [8] 6/-1/-1->5->4 [9] 6/-1/-1->5->4 [10] 6/-1/-1->5->4 [11] 6/-1/-1->5->4 [12] 6/-1/-1->5->4 [13] 6/-1/-1->5->13 [14] -1/-1/-1->5->4 [15] 6/-1/-1->5->4
+ip-26-0-160-225:124162:124660 [2] NCCL INFO NVLS Head  2:  2 10 18 26
+ip-26-0-160-225:124161:124659 [1] NCCL INFO NVLS Head  0:  0  8 16 24
+ip-26-0-160-225:124163:124661 [3] NCCL INFO NVLS Head  6:  6 14 22 30
+ip-26-0-160-225:124162:124660 [2] NCCL INFO NVLS Head  3:  3 11 19 27
+ip-26-0-160-242:1364903:1365409 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-225:124160:124658 [0] NCCL INFO NVLS Head  0:  0  8 16 24
+ip-26-0-160-225:124163:124661 [3] NCCL INFO NVLS Head  7:  7 15 23 31
+ip-26-0-160-225:124162:124660 [2] NCCL INFO NVLS Head  4:  4 12 20 28
+ip-26-0-160-225:124165:124662 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-225:124161:124659 [1] NCCL INFO NVLS Head  1:  1  9 17 25
+ip-26-0-160-225:124160:124658 [0] NCCL INFO NVLS Head  1:  1  9 17 25
+ip-26-0-160-225:124162:124660 [2] NCCL INFO NVLS Head  5:  5 13 21 29
+ip-26-0-160-225:124161:124659 [1] NCCL INFO NVLS Head  2:  2 10 18 26
+ip-26-0-160-225:124162:124660 [2] NCCL INFO NVLS Head  6:  6 14 22 30
+ip-26-0-160-225:124160:124658 [0] NCCL INFO NVLS Head  2:  2 10 18 26
+ip-26-0-160-225:124161:124659 [1] NCCL INFO NVLS Head  3:  3 11 19 27
+ip-26-0-160-225:124162:124660 [2] NCCL INFO NVLS Head  7:  7 15 23 31
+ip-26-0-160-225:124160:124658 [0] NCCL INFO NVLS Head  3:  3 11 19 27
+ip-26-0-160-225:124161:124659 [1] NCCL INFO NVLS Head  4:  4 12 20 28
+ip-26-0-160-242:1364907:1365403 [5] NCCL INFO Trees [0] 14/-1/-1->13->12 [1] 14/-1/-1->13->12 [2] 14/-1/-1->13->12 [3] 14/-1/-1->13->12 [4] 14/-1/-1->13->12 [5] 14/-1/-1->13->21 [6] -1/-1/-1->13->12 [7] 14/-1/-1->13->12 [8] 14/-1/-1->13->12 [9] 14/-1/-1->13->12 [10] 14/-1/-1->13->12 [11] 14/-1/-1->13->12 [12] 14/-1/-1->13->12 [13] 14/21/5->13->29 [14] -1/-1/-1->13->12 [15] 14/-1/-1->13->12
+ip-26-0-165-164:1227463:1227964 [0] NCCL INFO Channel 10/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-160-225:124160:124658 [0] NCCL INFO NVLS Head  4:  4 12 20 28
+ip-26-0-160-225:124161:124659 [1] NCCL INFO NVLS Head  5:  5 13 21 29
+ip-26-0-160-225:124160:124658 [0] NCCL INFO NVLS Head  5:  5 13 21 29
+ip-26-0-160-225:124161:124659 [1] NCCL INFO NVLS Head  6:  6 14 22 30
+ip-26-0-160-225:124160:124658 [0] NCCL INFO NVLS Head  6:  6 14 22 30
+ip-26-0-160-225:124161:124659 [1] NCCL INFO NVLS Head  7:  7 15 23 31
+ip-26-0-160-225:124160:124658 [0] NCCL INFO NVLS Head  7:  7 15 23 31
+ip-26-0-160-242:1364907:1365403 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-225:124163:124661 [3] NCCL INFO Trees [0] 4/-1/-1->3->2 [1] 4/-1/-1->3->2 [2] 4/-1/-1->3->2 [3] 4/19/-1->3->-1 [4] -1/-1/-1->3->2 [5] 4/-1/-1->3->2 [6] 4/-1/-1->3->2 [7] 4/-1/-1->3->2 [8] 4/-1/-1->3->2 [9] 4/-1/-1->3->2 [10] 4/-1/-1->3->2 [11] 4/-1/-1->3->11 [12] -1/-1/-1->3->2 [13] 4/-1/-1->3->2 [14] 4/-1/-1->3->2 [15] 4/-1/-1->3->2
+ip-26-0-160-225:124160:124658 [0] NCCL INFO Channel 00/16 :    0   1   2   3   4   5   6   7   8   9  10  11  12  13  14  15  16  17  18  19
+ip-26-0-160-225:124163:124661 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-225:124160:124658 [0] NCCL INFO Channel 01/16 :    0   7   6   5   4   3   2   9   8  15  14  13  12  11  10  17  16  23  22  21
+ip-26-0-160-225:124162:124660 [2] NCCL INFO Trees [0] 3/-1/-1->2->1 [1] 3/-1/-1->2->1 [2] 3/18/-1->2->-1 [3] -1/-1/-1->2->1 [4] 3/-1/-1->2->1 [5] 3/-1/-1->2->1 [6] 3/-1/-1->2->1 [7] 3/-1/-1->2->1 [8] 3/-1/-1->2->1 [9] 3/-1/-1->2->1 [10] 3/-1/-1->2->10 [11] -1/-1/-1->2->1 [12] 3/-1/-1->2->1 [13] 3/-1/-1->2->1 [14] 3/-1/-1->2->1 [15] 3/-1/-1->2->1
+ip-26-0-160-225:124162:124660 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-225:124160:124658 [0] NCCL INFO Channel 02/16 :    0   7   6   5   4   3  10   9   8  15  14  13  12  11  18  17  16  23  22  21
+ip-26-0-160-225:124161:124659 [1] NCCL INFO Trees [0] 2/-1/-1->1->0 [1] 2/17/-1->1->-1 [2] -1/-1/-1->1->0 [3] 2/-1/-1->1->0 [4] 2/-1/-1->1->0 [5] 2/-1/-1->1->0 [6] 2/-1/-1->1->0 [7] 2/-1/-1->1->0 [8] 2/-1/-1->1->0 [9] 2/-1/-1->1->9 [10] -1/-1/-1->1->0 [11] 2/-1/-1->1->0 [12] 2/-1/-1->1->0 [13] 2/-1/-1->1->0 [14] 2/-1/-1->1->0 [15] 2/-1/-1->1->0
+ip-26-0-160-225:124161:124659 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-225:124160:124658 [0] NCCL INFO Channel 03/16 :    0   7   6   5   4  11  10   9   8  15  14  13  12  19  18  17  16  23  22  21
+ip-26-0-160-225:124160:124658 [0] NCCL INFO Channel 04/16 :    0   7   6   5  12  11  10   9   8  15  14  13  20  19  18  17  16  23  22  21
+ip-26-0-160-225:124160:124658 [0] NCCL INFO Channel 05/16 :    0   4   7   6  13  11  10   9   8  12  15  14  21  19  18  17  16  20  23  22
+ip-26-0-160-225:124160:124658 [0] NCCL INFO Channel 06/16 :    0   5   4   7  14  11  10   9   8  13  12  15  22  19  18  17  16  21  20  23
+ip-26-0-160-225:124160:124658 [0] NCCL INFO Channel 07/16 :    0   6   5   4  15  11  10   9   8  14  13  12  23  19  18  17  16  22  21  20
+ip-26-0-160-225:124160:124658 [0] NCCL INFO Channel 08/16 :    0   1   2   3   4   5   6   7   8   9  10  11  12  13  14  15  16  17  18  19
+ip-26-0-160-225:124160:124658 [0] NCCL INFO Channel 09/16 :    0   7   6   5   4   3   2   9   8  15  14  13  12  11  10  17  16  23  22  21
+ip-26-0-160-225:124160:124658 [0] NCCL INFO Channel 10/16 :    0   7   6   5   4   3  10   9   8  15  14  13  12  11  18  17  16  23  22  21
+ip-26-0-160-225:124160:124658 [0] NCCL INFO Channel 11/16 :    0   7   6   5   4  11  10   9   8  15  14  13  12  19  18  17  16  23  22  21
+ip-26-0-160-225:124160:124658 [0] NCCL INFO Channel 12/16 :    0   7   6   5  12  11  10   9   8  15  14  13  20  19  18  17  16  23  22  21
+ip-26-0-160-225:124160:124658 [0] NCCL INFO Channel 13/16 :    0   4   7   6  13  11  10   9   8  12  15  14  21  19  18  17  16  20  23  22
+ip-26-0-160-225:124160:124658 [0] NCCL INFO Channel 14/16 :    0   5   4   7  14  11  10   9   8  13  12  15  22  19  18  17  16  21  20  23
+ip-26-0-160-225:124160:124658 [0] NCCL INFO Channel 15/16 :    0   6   5   4  15  11  10   9   8  14  13  12  23  19  18  17  16  22  21  20
+ip-26-0-160-225:124160:124658 [0] NCCL INFO Trees [0] 1/16/-1->0->-1 [1] -1/-1/-1->0->7 [2] 1/-1/-1->0->7 [3] 1/-1/-1->0->7 [4] 1/-1/-1->0->7 [5] 1/-1/-1->0->7 [6] 1/-1/-1->0->7 [7] 1/-1/-1->0->7 [8] 1/-1/-1->0->8 [9] -1/-1/-1->0->7 [10] 1/-1/-1->0->7 [11] 1/-1/-1->0->7 [12] 1/-1/-1->0->7 [13] 1/-1/-1->0->7 [14] 1/-1/-1->0->7 [15] 1/-1/-1->0->7
+ip-26-0-160-225:124160:124658 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-164:1227470:1227958 [7] NCCL INFO Channel 06/0 : 23[7] -> 30[6] [send] via NET/Libfabric/6(22)/GDRDMA
+ip-26-0-165-164:1227463:1227964 [0] NCCL INFO Channel 11/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-165-164:1227463:1227964 [0] NCCL INFO Channel 12/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-160-242:1364904:1365408 [2] NCCL INFO Channel 00/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-161-123:32708:33198 [4] NCCL INFO Channel 00/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-160-242:1364908:1365406 [6] NCCL INFO Channel 00/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-160-242:1364904:1365408 [2] NCCL INFO Channel 08/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-161-103:96893:97384 [5] NCCL INFO Channel 00/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-161-103:96892:97383 [4] NCCL INFO Channel 00/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-161-123:32707:33200 [3] NCCL INFO Channel 00/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-161-153:2636086:2636602 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-160-242:1364905:1365404 [3] NCCL INFO Channel 00/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-161-123:32708:33198 [4] NCCL INFO Channel 08/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-160-242:1364908:1365406 [6] NCCL INFO Channel 08/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-161-123:32706:33199 [2] NCCL INFO Channel 00/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-161-103:96889:97388 [1] NCCL INFO Channel 00/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-161-123:32709:33201 [5] NCCL INFO Channel 00/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-161-123:32705:33203 [1] NCCL INFO Channel 00/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-161-123:32710:33202 [6] NCCL INFO Channel 00/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-160-242:1364906:1365407 [4] NCCL INFO Channel 00/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-160-242:1364907:1365403 [5] NCCL INFO Channel 00/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-160-242:1364903:1365409 [1] NCCL INFO Channel 00/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-161-103:96894:97381 [6] NCCL INFO Channel 00/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-161-103:96893:97384 [5] NCCL INFO Channel 08/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-161-103:96890:97386 [2] NCCL INFO Channel 00/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-161-103:96891:97385 [3] NCCL INFO Channel 00/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-161-103:96892:97383 [4] NCCL INFO Channel 08/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-161-153:2636086:2636602 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-160-242:1364905:1365404 [3] NCCL INFO Channel 08/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-161-123:32707:33200 [3] NCCL INFO Channel 08/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-161-103:96889:97388 [1] NCCL INFO Channel 08/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-160-242:1364906:1365407 [4] NCCL INFO Channel 08/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-160-242:1364907:1365403 [5] NCCL INFO Channel 08/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-160-242:1364903:1365409 [1] NCCL INFO Channel 08/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-161-123:32706:33199 [2] NCCL INFO Channel 08/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-161-123:32709:33201 [5] NCCL INFO Channel 08/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-161-123:32705:33203 [1] NCCL INFO Channel 08/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-161-123:32710:33202 [6] NCCL INFO Channel 08/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-161-103:96890:97386 [2] NCCL INFO Channel 08/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-161-103:96894:97381 [6] NCCL INFO Channel 08/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-161-103:96891:97385 [3] NCCL INFO Channel 08/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-160-242:1364906:1365407 [4] NCCL INFO Channel 05/0 : 12[4] -> 15[7] via P2P/IPC
+ip-26-0-161-123:32708:33198 [4] NCCL INFO Channel 05/0 : 28[4] -> 31[7] via P2P/IPC
+ip-26-0-160-242:1364906:1365407 [4] NCCL INFO Channel 06/0 : 12[4] -> 15[7] via P2P/IPC
+ip-26-0-161-103:96892:97383 [4] NCCL INFO Channel 05/0 : 20[4] -> 23[7] via P2P/IPC
+ip-26-0-161-123:32708:33198 [4] NCCL INFO Channel 06/0 : 28[4] -> 31[7] via P2P/IPC
+ip-26-0-160-242:1364906:1365407 [4] NCCL INFO Channel 13/0 : 12[4] -> 15[7] via P2P/IPC
+ip-26-0-161-103:96892:97383 [4] NCCL INFO Channel 06/0 : 20[4] -> 23[7] via P2P/IPC
+ip-26-0-160-242:1364906:1365407 [4] NCCL INFO Channel 14/0 : 12[4] -> 15[7] via P2P/IPC
+ip-26-0-161-123:32708:33198 [4] NCCL INFO Channel 13/0 : 28[4] -> 31[7] via P2P/IPC
+ip-26-0-161-103:96892:97383 [4] NCCL INFO Channel 13/0 : 20[4] -> 23[7] via P2P/IPC
+ip-26-0-161-123:32708:33198 [4] NCCL INFO Channel 14/0 : 28[4] -> 31[7] via P2P/IPC
+ip-26-0-161-103:96892:97383 [4] NCCL INFO Channel 14/0 : 20[4] -> 23[7] via P2P/IPC
+ip-26-0-160-225:124161:124659 [1] NCCL INFO Channel 00/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-225:124164:124663 [4] NCCL INFO Channel 00/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-225:124166:124665 [6] NCCL INFO Channel 00/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-225:124161:124659 [1] NCCL INFO Channel 08/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-225:124165:124662 [5] NCCL INFO Channel 00/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-225:124162:124660 [2] NCCL INFO Channel 00/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-225:124164:124663 [4] NCCL INFO Channel 08/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-225:124163:124661 [3] NCCL INFO Channel 00/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-225:124166:124665 [6] NCCL INFO Channel 08/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-225:124165:124662 [5] NCCL INFO Channel 08/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-225:124162:124660 [2] NCCL INFO Channel 08/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-225:124163:124661 [3] NCCL INFO Channel 08/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-225:124164:124663 [4] NCCL INFO Channel 05/0 : 4[4] -> 7[7] via P2P/IPC
+ip-26-0-160-225:124164:124663 [4] NCCL INFO Channel 06/0 : 4[4] -> 7[7] via P2P/IPC
+ip-26-0-160-225:124164:124663 [4] NCCL INFO Channel 13/0 : 4[4] -> 7[7] via P2P/IPC
+ip-26-0-160-225:124164:124663 [4] NCCL INFO Channel 14/0 : 4[4] -> 7[7] via P2P/IPC
+ip-26-0-165-59:1557266:1557759 [0] NCCL INFO Channel 00/0 : 31[7] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557266:1557759 [0] NCCL INFO Channel 08/0 : 31[7] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557266:1557759 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-59:1557273:1557760 [7] NCCL INFO Channel 00/0 : 7[7] -> 8[0] [send] via NET/Libfabric/0(0)/GDRDMA
+ip-26-0-168-34:272150:272645 [0] NCCL INFO Channel 00/0 : 15[7] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272150:272645 [0] NCCL INFO Channel 08/0 : 15[7] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272150:272645 [0] NCCL INFO Channel 00/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-168-34:272157:272649 [7] NCCL INFO Channel 00/0 : 23[7] -> 24[0] [send] via NET/Libfabric/0(16)/GDRDMA
+ip-26-0-168-30:235648:236145 [0] NCCL INFO Channel 00/0 : 7[7] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235655:236142 [7] NCCL INFO Channel 00/0 : 15[7] -> 16[0] [send] via NET/Libfabric/0(8)/GDRDMA
+ip-26-0-168-30:235648:236145 [0] NCCL INFO Channel 08/0 : 7[7] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235648:236145 [0] NCCL INFO Channel 00/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-168-30:235655:236142 [7] NCCL INFO Channel 08/0 : 15[7] -> 16[0] [send] via NET/Libfabric/0(8)/GDRDMA
+ip-26-0-168-34:272150:272645 [0] NCCL INFO Channel 08/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-168-34:272157:272649 [7] NCCL INFO Channel 08/0 : 23[7] -> 24[0] [send] via NET/Libfabric/0(16)/GDRDMA
+ip-26-0-168-30:235648:236145 [0] NCCL INFO Channel 08/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-168-34:272150:272645 [0] NCCL INFO Channel 05/0 : 16[0] -> 20[4] via P2P/IPC
+ip-26-0-168-34:272150:272645 [0] NCCL INFO Channel 13/0 : 16[0] -> 20[4] via P2P/IPC
+ip-26-0-165-131:1630206:1630728 [1] NCCL INFO Channel 01/0 : 2[2] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-52:2722045:2722544 [0] NCCL INFO Channel 00/0 : 23[7] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630206:1630728 [1] NCCL INFO Channel 09/0 : 2[2] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-52:2722052:2722546 [7] NCCL INFO Channel 00/0 : 31[7] -> 0[0] [send] via NET/Libfabric/0(24)/GDRDMA
+ip-26-0-165-131:1630207:1630731 [2] NCCL INFO Channel 01/0 : 10[2] -> 17[1] [send] via NET/Libfabric/1(9)/GDRDMA
+ip-26-0-168-52:2722045:2722544 [0] NCCL INFO Channel 08/0 : 23[7] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-52:2722045:2722544 [0] NCCL INFO Channel 00/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-165-131:1630207:1630731 [2] NCCL INFO Channel 09/0 : 10[2] -> 17[1] [send] via NET/Libfabric/1(9)/GDRDMA
+ip-26-0-168-52:2722052:2722546 [7] NCCL INFO Channel 08/0 : 31[7] -> 0[0] [send] via NET/Libfabric/0(24)/GDRDMA
+ip-26-0-165-59:1557268:1557764 [2] NCCL INFO Channel 02/0 : 27[3] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557268:1557764 [2] NCCL INFO Channel 10/0 : 27[3] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-52:2722045:2722544 [0] NCCL INFO Channel 08/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-165-164:1227464:1227963 [1] NCCL INFO Channel 01/0 : 10[2] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227464:1227963 [1] NCCL INFO Channel 09/0 : 10[2] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114513:115008 [1] NCCL INFO Channel 01/0 : 26[2] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227465:1227962 [2] NCCL INFO Channel 01/0 : 18[2] -> 25[1] [send] via NET/Libfabric/1(17)/GDRDMA
+ip-26-0-161-78:114513:115008 [1] NCCL INFO Channel 09/0 : 26[2] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227465:1227962 [2] NCCL INFO Channel 09/0 : 18[2] -> 25[1] [send] via NET/Libfabric/1(17)/GDRDMA
+ip-26-0-161-78:114514:115011 [2] NCCL INFO Channel 01/0 : 2[2] -> 9[1] [send] via NET/Libfabric/1(1)/GDRDMA
+ip-26-0-161-78:114514:115011 [2] NCCL INFO Channel 09/0 : 2[2] -> 9[1] [send] via NET/Libfabric/1(1)/GDRDMA
+ip-26-0-168-34:272153:272651 [3] NCCL INFO Channel 03/0 : 12[4] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272153:272651 [3] NCCL INFO Channel 11/0 : 12[4] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557268:1557764 [2] NCCL INFO Channel 01/0 : 2[2] -> 9[1] [send] via NET/Libfabric/1(1)/GDRDMA
+ip-26-0-165-59:1557268:1557764 [2] NCCL INFO Channel 09/0 : 2[2] -> 9[1] [send] via NET/Libfabric/1(1)/GDRDMA
+ip-26-0-165-59:1557269:1557765 [3] NCCL INFO Channel 03/0 : 28[4] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557269:1557765 [3] NCCL INFO Channel 11/0 : 28[4] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557269:1557765 [3] NCCL INFO Channel 02/0 : 3[3] -> 10[2] [send] via NET/Libfabric/2(2)/GDRDMA
+ip-26-0-165-59:1557269:1557765 [3] NCCL INFO Channel 10/0 : 3[3] -> 10[2] [send] via NET/Libfabric/2(2)/GDRDMA
+ip-26-0-168-52:2722045:2722544 [0] NCCL INFO Channel 05/0 : 24[0] -> 28[4] via P2P/IPC
+ip-26-0-168-30:235655:236142 [7] NCCL INFO Channel 06/0 : 15[7] -> 22[6] [send] via NET/Libfabric/6(14)/GDRDMA
+ip-26-0-168-30:235655:236142 [7] NCCL INFO Channel 14/0 : 15[7] -> 22[6] [send] via NET/Libfabric/6(14)/GDRDMA
+ip-26-0-168-30:235650:236144 [2] NCCL INFO Channel 02/0 : 3[3] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235650:236144 [2] NCCL INFO Channel 10/0 : 3[3] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-52:2722045:2722544 [0] NCCL INFO Channel 13/0 : 24[0] -> 28[4] via P2P/IPC
+ip-26-0-168-30:235651:236141 [3] NCCL INFO Channel 03/0 : 4[4] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235651:236141 [3] NCCL INFO Channel 11/0 : 4[4] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272152:272647 [2] NCCL INFO Channel 02/0 : 11[3] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272152:272647 [2] NCCL INFO Channel 10/0 : 11[3] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235651:236141 [3] NCCL INFO Channel 02/0 : 11[3] -> 18[2] [send] via NET/Libfabric/2(10)/GDRDMA
+ip-26-0-168-34:272153:272651 [3] NCCL INFO Channel 02/0 : 19[3] -> 26[2] [send] via NET/Libfabric/2(18)/GDRDMA
+ip-26-0-168-30:235651:236141 [3] NCCL INFO Channel 10/0 : 11[3] -> 18[2] [send] via NET/Libfabric/2(10)/GDRDMA
+ip-26-0-168-34:272153:272651 [3] NCCL INFO Channel 10/0 : 19[3] -> 26[2] [send] via NET/Libfabric/2(18)/GDRDMA
+ip-26-0-168-52:2722047:2722542 [2] NCCL INFO Channel 02/0 : 19[3] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227464:1227963 [1] NCCL INFO Channel 01/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-168-52:2722047:2722542 [2] NCCL INFO Channel 10/0 : 19[3] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227464:1227963 [1] NCCL INFO Channel 02/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-161-78:114514:115011 [2] NCCL INFO Channel 02/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-165-164:1227464:1227963 [1] NCCL INFO Channel 03/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-168-52:2722048:2722545 [3] NCCL INFO Channel 03/0 : 20[4] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153563:154060 [1] NCCL INFO Channel 01/0 : 18[2] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114514:115011 [2] NCCL INFO Channel 03/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-168-52:2722048:2722545 [3] NCCL INFO Channel 11/0 : 20[4] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153564:154057 [2] NCCL INFO Channel 01/0 : 26[2] -> 1[1] [send] via NET/Libfabric/1(25)/GDRDMA
+ip-26-0-165-202:153563:154060 [1] NCCL INFO Channel 09/0 : 18[2] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227464:1227963 [1] NCCL INFO Channel 04/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-165-202:153564:154057 [2] NCCL INFO Channel 09/0 : 26[2] -> 1[1] [send] via NET/Libfabric/1(25)/GDRDMA
+ip-26-0-161-78:114514:115011 [2] NCCL INFO Channel 04/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-168-52:2722048:2722545 [3] NCCL INFO Channel 02/0 : 27[3] -> 2[2] [send] via NET/Libfabric/2(26)/GDRDMA
+ip-26-0-161-78:114516:115010 [4] NCCL INFO Channel 04/0 : 29[5] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722048:2722545 [3] NCCL INFO Channel 10/0 : 27[3] -> 2[2] [send] via NET/Libfabric/2(26)/GDRDMA
+ip-26-0-161-78:114516:115010 [4] NCCL INFO Channel 12/0 : 29[5] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227464:1227963 [1] NCCL INFO Channel 05/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-161-78:114514:115011 [2] NCCL INFO Channel 05/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-78:114516:115010 [4] NCCL INFO Channel 03/0 : 4[4] -> 11[3] [send] via NET/Libfabric/3(3)/GDRDMA
+ip-26-0-161-78:114516:115010 [4] NCCL INFO Channel 11/0 : 4[4] -> 11[3] [send] via NET/Libfabric/3(3)/GDRDMA
+ip-26-0-168-34:272150:272645 [0] NCCL INFO Channel 06/0 : 16[0] -> 21[5] via P2P/IPC
+ip-26-0-161-78:114514:115011 [2] NCCL INFO Channel 06/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-165-131:1630209:1630730 [4] NCCL INFO Channel 04/0 : 5[5] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227464:1227963 [1] NCCL INFO Channel 06/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-165-131:1630209:1630730 [4] NCCL INFO Channel 12/0 : 5[5] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114517:115013 [5] NCCL INFO Channel 05/0 : 30[6] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227467:1227960 [4] NCCL INFO Channel 04/0 : 13[5] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114517:115013 [5] NCCL INFO Channel 13/0 : 30[6] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227467:1227960 [4] NCCL INFO Channel 12/0 : 13[5] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114514:115011 [2] NCCL INFO Channel 07/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-168-34:272150:272645 [0] NCCL INFO Channel 14/0 : 16[0] -> 21[5] via P2P/IPC
+ip-26-0-161-78:114517:115013 [5] NCCL INFO Channel 04/0 : 5[5] -> 12[4] [send] via NET/Libfabric/4(4)/GDRDMA
+ip-26-0-165-164:1227464:1227963 [1] NCCL INFO Channel 07/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-161-78:114517:115013 [5] NCCL INFO Channel 12/0 : 5[5] -> 12[4] [send] via NET/Libfabric/4(4)/GDRDMA
+ip-26-0-165-164:1227467:1227960 [4] NCCL INFO Channel 03/0 : 20[4] -> 27[3] [send] via NET/Libfabric/3(19)/GDRDMA
+ip-26-0-161-78:114514:115011 [2] NCCL INFO Channel 10/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-165-164:1227467:1227960 [4] NCCL INFO Channel 11/0 : 20[4] -> 27[3] [send] via NET/Libfabric/3(19)/GDRDMA
+ip-26-0-168-34:272150:272645 [0] NCCL INFO Channel 07/0 : 16[0] -> 22[6] via P2P/IPC
+ip-26-0-165-164:1227464:1227963 [1] NCCL INFO Channel 09/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-165-131:1630211:1630733 [6] NCCL INFO Channel 06/0 : 7[7] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114514:115011 [2] NCCL INFO Channel 11/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-165-131:1630209:1630730 [4] NCCL INFO Channel 03/0 : 12[4] -> 19[3] [send] via NET/Libfabric/3(11)/GDRDMA
+ip-26-0-165-131:1630212:1630734 [7] NCCL INFO Channel 14/0 : 15[7] -> 22[6] [send] via NET/Libfabric/6(14)/GDRDMA
+ip-26-0-165-131:1630211:1630733 [6] NCCL INFO Channel 14/0 : 7[7] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630210:1630732 [5] NCCL INFO Channel 05/0 : 6[6] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630209:1630730 [4] NCCL INFO Channel 11/0 : 12[4] -> 19[3] [send] via NET/Libfabric/3(11)/GDRDMA
+ip-26-0-165-202:153567:154059 [5] NCCL INFO Channel 05/0 : 22[6] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272150:272645 [0] NCCL INFO Channel 15/0 : 16[0] -> 22[6] via P2P/IPC
+ip-26-0-165-202:153567:154059 [5] NCCL INFO Channel 13/0 : 22[6] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630210:1630732 [5] NCCL INFO Channel 13/0 : 6[6] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227464:1227963 [1] NCCL INFO Channel 10/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-161-78:114514:115011 [2] NCCL INFO Channel 12/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-165-131:1630211:1630733 [6] NCCL INFO Channel 05/0 : 14[6] -> 21[5] [send] via NET/Libfabric/5(13)/GDRDMA
+ip-26-0-165-131:1630211:1630733 [6] NCCL INFO Channel 13/0 : 14[6] -> 21[5] [send] via NET/Libfabric/5(13)/GDRDMA
+ip-26-0-165-131:1630210:1630732 [5] NCCL INFO Channel 04/0 : 13[5] -> 20[4] [send] via NET/Libfabric/4(12)/GDRDMA
+ip-26-0-165-131:1630210:1630732 [5] NCCL INFO Channel 12/0 : 13[5] -> 20[4] [send] via NET/Libfabric/4(12)/GDRDMA
+ip-26-0-161-78:114514:115011 [2] NCCL INFO Channel 13/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-165-164:1227464:1227963 [1] NCCL INFO Channel 11/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-168-34:272150:272645 [0] NCCL INFO Channel 01/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-165-202:153568:154056 [6] NCCL INFO Channel 06/0 : 23[7] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153568:154056 [6] NCCL INFO Channel 14/0 : 23[7] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114514:115011 [2] NCCL INFO Channel 14/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-168-34:272150:272645 [0] NCCL INFO Channel 02/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-165-202:153568:154056 [6] NCCL INFO Channel 05/0 : 30[6] -> 5[5] [send] via NET/Libfabric/5(29)/GDRDMA
+ip-26-0-165-202:153568:154056 [6] NCCL INFO Channel 13/0 : 30[6] -> 5[5] [send] via NET/Libfabric/5(29)/GDRDMA
+ip-26-0-168-34:272150:272645 [0] NCCL INFO Channel 03/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-165-164:1227464:1227963 [1] NCCL INFO Channel 12/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-161-78:114514:115011 [2] NCCL INFO Channel 15/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-78:114517:115013 [5] NCCL INFO Channel 05/0 : 5[5] -> 3[3] via P2P/IPC
+ip-26-0-168-34:272150:272645 [0] NCCL INFO Channel 04/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-161-78:114518:115012 [6] NCCL INFO Channel 06/0 : 31[7] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114519:115014 [7] NCCL INFO Channel 06/0 : 7[7] -> 14[6] [send] via NET/Libfabric/6(6)/GDRDMA
+ip-26-0-161-78:114518:115012 [6] NCCL INFO Channel 14/0 : 31[7] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227464:1227963 [1] NCCL INFO Channel 13/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-168-34:272150:272645 [0] NCCL INFO Channel 09/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-165-131:1630207:1630731 [2] NCCL INFO Channel 02/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-161-78:114519:115014 [7] NCCL INFO Channel 14/0 : 7[7] -> 14[6] [send] via NET/Libfabric/6(6)/GDRDMA
+ip-26-0-168-34:272157:272649 [7] NCCL INFO Channel 06/0 : 23[7] -> 30[6] [send] via NET/Libfabric/6(22)/GDRDMA
+ip-26-0-165-202:153566:154062 [4] NCCL INFO Channel 04/0 : 21[5] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153566:154062 [4] NCCL INFO Channel 12/0 : 21[5] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153567:154059 [5] NCCL INFO Channel 04/0 : 29[5] -> 4[4] [send] via NET/Libfabric/4(28)/GDRDMA
+ip-26-0-168-34:272150:272645 [0] NCCL INFO Channel 10/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-161-78:114517:115013 [5] NCCL INFO Channel 13/0 : 5[5] -> 3[3] via P2P/IPC
+ip-26-0-165-164:1227464:1227963 [1] NCCL INFO Channel 14/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-165-202:153567:154059 [5] NCCL INFO Channel 12/0 : 29[5] -> 4[4] [send] via NET/Libfabric/4(28)/GDRDMA
+ip-26-0-165-202:153566:154062 [4] NCCL INFO Channel 03/0 : 28[4] -> 3[3] [send] via NET/Libfabric/3(27)/GDRDMA
+ip-26-0-165-131:1630209:1630730 [4] NCCL INFO Channel 07/0 : 12[4] -> 23[7] [send] via NET/Libfabric/7(15)/GDRDMA
+ip-26-0-165-131:1630209:1630730 [4] NCCL INFO Channel 15/0 : 12[4] -> 23[7] [send] via NET/Libfabric/7(15)/GDRDMA
+ip-26-0-165-202:153566:154062 [4] NCCL INFO Channel 11/0 : 28[4] -> 3[3] [send] via NET/Libfabric/3(27)/GDRDMA
+ip-26-0-168-34:272150:272645 [0] NCCL INFO Channel 11/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-165-164:1227467:1227960 [4] NCCL INFO Channel 07/0 : 20[4] -> 31[7] [send] via NET/Libfabric/7(23)/GDRDMA
+ip-26-0-165-164:1227467:1227960 [4] NCCL INFO Channel 15/0 : 20[4] -> 31[7] [send] via NET/Libfabric/7(23)/GDRDMA
+ip-26-0-161-78:114518:115012 [6] NCCL INFO Channel 05/0 : 6[6] -> 13[5] [send] via NET/Libfabric/5(5)/GDRDMA
+ip-26-0-168-34:272150:272645 [0] NCCL INFO Channel 12/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-165-164:1227464:1227963 [1] NCCL INFO Channel 15/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-161-78:114518:115012 [6] NCCL INFO Channel 13/0 : 6[6] -> 13[5] [send] via NET/Libfabric/5(5)/GDRDMA
+ip-26-0-165-131:1630206:1630728 [1] NCCL INFO Channel 01/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-165-202:153564:154057 [2] NCCL INFO Channel 02/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-161-78:114518:115012 [6] NCCL INFO Channel 06/0 : 6[6] -> 3[3] via P2P/IPC
+ip-26-0-165-131:1630210:1630732 [5] NCCL INFO Channel 05/0 : 13[5] -> 11[3] via P2P/IPC
+ip-26-0-165-202:153563:154060 [1] NCCL INFO Channel 01/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-165-131:1630207:1630731 [2] NCCL INFO Channel 03/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-165-164:1227469:1227957 [6] NCCL INFO Channel 06/0 : 15[7] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227470:1227958 [7] NCCL INFO Channel 14/0 : 23[7] -> 30[6] [send] via NET/Libfabric/6(22)/GDRDMA
+ip-26-0-165-164:1227469:1227957 [6] NCCL INFO Channel 14/0 : 15[7] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227465:1227962 [2] NCCL INFO Channel 02/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-165-131:1630206:1630728 [1] NCCL INFO Channel 02/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-165-202:153564:154057 [2] NCCL INFO Channel 03/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-165-202:153563:154060 [1] NCCL INFO Channel 02/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-165-202:153568:154056 [6] NCCL INFO Channel 06/0 : 30[6] -> 27[3] via P2P/IPC
+ip-26-0-165-164:1227465:1227962 [2] NCCL INFO Channel 03/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-161-78:114518:115012 [6] NCCL INFO Channel 14/0 : 6[6] -> 3[3] via P2P/IPC
+ip-26-0-165-164:1227465:1227962 [2] NCCL INFO Channel 04/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-161-78:114513:115008 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-164:1227468:1227959 [5] NCCL INFO Channel 05/0 : 14[6] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227469:1227957 [6] NCCL INFO Channel 05/0 : 22[6] -> 29[5] [send] via NET/Libfabric/5(21)/GDRDMA
+ip-26-0-165-164:1227468:1227959 [5] NCCL INFO Channel 13/0 : 14[6] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227465:1227962 [2] NCCL INFO Channel 05/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-165-164:1227469:1227957 [6] NCCL INFO Channel 13/0 : 22[6] -> 29[5] [send] via NET/Libfabric/5(21)/GDRDMA
+ip-26-0-165-202:153564:154057 [2] NCCL INFO Channel 04/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-165-202:153563:154060 [1] NCCL INFO Channel 03/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-165-202:153568:154056 [6] NCCL INFO Channel 14/0 : 30[6] -> 27[3] via P2P/IPC
+ip-26-0-165-164:1227468:1227959 [5] NCCL INFO Channel 04/0 : 21[5] -> 28[4] [send] via NET/Libfabric/4(20)/GDRDMA
+ip-26-0-165-164:1227468:1227959 [5] NCCL INFO Channel 12/0 : 21[5] -> 28[4] [send] via NET/Libfabric/4(20)/GDRDMA
+ip-26-0-165-164:1227465:1227962 [2] NCCL INFO Channel 06/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-161-78:114513:115008 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-131:1630207:1630731 [2] NCCL INFO Channel 04/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-165-131:1630211:1630733 [6] NCCL INFO Channel 06/0 : 14[6] -> 11[3] via P2P/IPC
+ip-26-0-165-164:1227465:1227962 [2] NCCL INFO Channel 07/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-165-164:1227469:1227957 [6] NCCL INFO Channel 06/0 : 22[6] -> 19[3] via P2P/IPC
+ip-26-0-165-202:153567:154059 [5] NCCL INFO Channel 05/0 : 29[5] -> 27[3] via P2P/IPC
+ip-26-0-165-164:1227468:1227959 [5] NCCL INFO Channel 05/0 : 21[5] -> 19[3] via P2P/IPC
+ip-26-0-161-78:114513:115008 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-202:153564:154057 [2] NCCL INFO Channel 05/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-165-202:153563:154060 [1] NCCL INFO Channel 04/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-165-131:1630210:1630732 [5] NCCL INFO Channel 13/0 : 13[5] -> 11[3] via P2P/IPC
+ip-26-0-161-78:114513:115008 [1] NCCL INFO Channel 04/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-131:1630206:1630728 [1] NCCL INFO Channel 03/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-165-131:1630207:1630731 [2] NCCL INFO Channel 05/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-165-202:153567:154059 [5] NCCL INFO Channel 13/0 : 29[5] -> 27[3] via P2P/IPC
+ip-26-0-165-202:153564:154057 [2] NCCL INFO Channel 06/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-165-164:1227465:1227962 [2] NCCL INFO Channel 10/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-165-202:153563:154060 [1] NCCL INFO Channel 05/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-165-131:1630211:1630733 [6] NCCL INFO Channel 14/0 : 14[6] -> 11[3] via P2P/IPC
+ip-26-0-165-164:1227469:1227957 [6] NCCL INFO Channel 14/0 : 22[6] -> 19[3] via P2P/IPC
+ip-26-0-161-78:114513:115008 [1] NCCL INFO Channel 05/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-164:1227468:1227959 [5] NCCL INFO Channel 13/0 : 21[5] -> 19[3] via P2P/IPC
+ip-26-0-161-78:114513:115008 [1] NCCL INFO Channel 06/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-131:1630206:1630728 [1] NCCL INFO Channel 04/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-165-202:153564:154057 [2] NCCL INFO Channel 07/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-165-131:1630207:1630731 [2] NCCL INFO Channel 06/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-165-202:153563:154060 [1] NCCL INFO Channel 06/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-165-164:1227465:1227962 [2] NCCL INFO Channel 11/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-161-78:114513:115008 [1] NCCL INFO Channel 07/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-202:153564:154057 [2] NCCL INFO Channel 10/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-161-103:96888:97387 [0] NCCL INFO Channel 00/0 : 15[7] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227465:1227962 [2] NCCL INFO Channel 12/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-165-202:153563:154060 [1] NCCL INFO Channel 07/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-161-103:96888:97387 [0] NCCL INFO Channel 08/0 : 15[7] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96888:97387 [0] NCCL INFO Channel 00/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-165-131:1630206:1630728 [1] NCCL INFO Channel 05/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-161-78:114513:115008 [1] NCCL INFO Channel 09/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-103:96895:97382 [7] NCCL INFO Channel 00/0 : 23[7] -> 24[0] [send] via NET/Libfabric/0(16)/GDRDMA
+ip-26-0-165-131:1630207:1630731 [2] NCCL INFO Channel 07/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-165-164:1227465:1227962 [2] NCCL INFO Channel 13/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-165-202:153564:154057 [2] NCCL INFO Channel 11/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-161-103:96888:97387 [0] NCCL INFO Channel 08/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-165-202:153563:154060 [1] NCCL INFO Channel 09/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-161-78:114513:115008 [1] NCCL INFO Channel 10/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-103:96895:97382 [7] NCCL INFO Channel 08/0 : 23[7] -> 24[0] [send] via NET/Libfabric/0(16)/GDRDMA
+ip-26-0-160-225:124160:124658 [0] NCCL INFO Channel 00/0 : 31[7] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:124160:124658 [0] NCCL INFO Channel 08/0 : 31[7] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:124160:124658 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-131:1630206:1630728 [1] NCCL INFO Channel 06/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-160-225:124167:124664 [7] NCCL INFO Channel 00/0 : 7[7] -> 8[0] [send] via NET/Libfabric/0(0)/GDRDMA
+ip-26-0-165-131:1630207:1630731 [2] NCCL INFO Channel 10/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-165-164:1227465:1227962 [2] NCCL INFO Channel 14/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-165-202:153564:154057 [2] NCCL INFO Channel 12/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-161-78:114513:115008 [1] NCCL INFO Channel 11/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-202:153563:154060 [1] NCCL INFO Channel 10/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-165-164:1227465:1227962 [2] NCCL INFO Channel 15/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-160-225:124160:124658 [0] NCCL INFO Channel 08/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-131:1630206:1630728 [1] NCCL INFO Channel 07/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-160-225:124167:124664 [7] NCCL INFO Channel 08/0 : 7[7] -> 8[0] [send] via NET/Libfabric/0(0)/GDRDMA
+ip-26-0-165-202:153564:154057 [2] NCCL INFO Channel 13/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-165-131:1630207:1630731 [2] NCCL INFO Channel 11/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-165-202:153563:154060 [1] NCCL INFO Channel 11/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-161-123:32704:33197 [0] NCCL INFO Channel 00/0 : 23[7] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32704:33197 [0] NCCL INFO Channel 08/0 : 23[7] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32704:33197 [0] NCCL INFO Channel 00/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-161-78:114513:115008 [1] NCCL INFO Channel 12/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-123:32711:33204 [7] NCCL INFO Channel 00/0 : 31[7] -> 0[0] [send] via NET/Libfabric/0(24)/GDRDMA
+ip-26-0-165-202:153564:154057 [2] NCCL INFO Channel 14/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-165-202:153563:154060 [1] NCCL INFO Channel 12/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-165-131:1630206:1630728 [1] NCCL INFO Channel 09/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-165-131:1630207:1630731 [2] NCCL INFO Channel 12/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-161-78:114513:115008 [1] NCCL INFO Channel 13/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-123:32704:33197 [0] NCCL INFO Channel 08/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-165-202:153564:154057 [2] NCCL INFO Channel 15/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-161-123:32711:33204 [7] NCCL INFO Channel 08/0 : 31[7] -> 0[0] [send] via NET/Libfabric/0(24)/GDRDMA
+ip-26-0-165-202:153563:154060 [1] NCCL INFO Channel 13/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-161-78:114513:115008 [1] NCCL INFO Channel 14/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-131:1630206:1630728 [1] NCCL INFO Channel 10/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-160-225:124160:124658 [0] NCCL INFO Channel 05/0 : 0[0] -> 4[4] via P2P/IPC
+ip-26-0-165-131:1630207:1630731 [2] NCCL INFO Channel 13/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-161-78:114513:115008 [1] NCCL INFO Channel 15/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-202:153563:154060 [1] NCCL INFO Channel 14/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-161-123:32704:33197 [0] NCCL INFO Channel 05/0 : 24[0] -> 28[4] via P2P/IPC
+ip-26-0-160-225:124160:124658 [0] NCCL INFO Channel 13/0 : 0[0] -> 4[4] via P2P/IPC
+ip-26-0-165-202:153563:154060 [1] NCCL INFO Channel 15/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-165-131:1630207:1630731 [2] NCCL INFO Channel 14/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-161-123:32704:33197 [0] NCCL INFO Channel 13/0 : 24[0] -> 28[4] via P2P/IPC
+ip-26-0-160-242:1364902:1365410 [0] NCCL INFO Channel 00/0 : 7[7] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364902:1365410 [0] NCCL INFO Channel 08/0 : 7[7] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630206:1630728 [1] NCCL INFO Channel 11/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-160-242:1364902:1365410 [0] NCCL INFO Channel 00/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-160-242:1364909:1365405 [7] NCCL INFO Channel 00/0 : 15[7] -> 16[0] [send] via NET/Libfabric/0(8)/GDRDMA
+ip-26-0-165-131:1630207:1630731 [2] NCCL INFO Channel 15/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-161-103:96890:97386 [2] NCCL INFO Channel 02/0 : 11[3] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96890:97386 [2] NCCL INFO Channel 10/0 : 11[3] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630206:1630728 [1] NCCL INFO Channel 12/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-160-242:1364902:1365410 [0] NCCL INFO Channel 08/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-160-242:1364909:1365405 [7] NCCL INFO Channel 08/0 : 15[7] -> 16[0] [send] via NET/Libfabric/0(8)/GDRDMA
+ip-26-0-165-131:1630206:1630728 [1] NCCL INFO Channel 13/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-161-103:96888:97387 [0] NCCL INFO Channel 05/0 : 16[0] -> 20[4] via P2P/IPC
+ip-26-0-165-131:1630206:1630728 [1] NCCL INFO Channel 14/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-165-59:1557266:1557759 [0] NCCL INFO Channel 08/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-59:1557273:1557760 [7] NCCL INFO Channel 08/0 : 7[7] -> 8[0] [send] via NET/Libfabric/0(0)/GDRDMA
+ip-26-0-160-242:1364902:1365410 [0] NCCL INFO Channel 05/0 : 8[0] -> 12[4] via P2P/IPC
+ip-26-0-165-131:1630206:1630728 [1] NCCL INFO Channel 15/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-160-225:124163:124661 [3] NCCL INFO Channel 03/0 : 28[4] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:124163:124661 [3] NCCL INFO Channel 11/0 : 28[4] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235648:236145 [0] NCCL INFO Channel 05/0 : 8[0] -> 12[4] via P2P/IPC
+ip-26-0-161-103:96888:97387 [0] NCCL INFO Channel 13/0 : 16[0] -> 20[4] via P2P/IPC
+ip-26-0-160-242:1364902:1365410 [0] NCCL INFO Channel 13/0 : 8[0] -> 12[4] via P2P/IPC
+ip-26-0-160-242:1364905:1365404 [3] NCCL INFO Channel 03/0 : 4[4] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364905:1365404 [3] NCCL INFO Channel 11/0 : 4[4] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235648:236145 [0] NCCL INFO Channel 13/0 : 8[0] -> 12[4] via P2P/IPC
+ip-26-0-161-123:32704:33197 [0] NCCL INFO Channel 06/0 : 24[0] -> 29[5] via P2P/IPC
+ip-26-0-160-242:1364904:1365408 [2] NCCL INFO Channel 02/0 : 3[3] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96888:97387 [0] NCCL INFO Channel 06/0 : 16[0] -> 21[5] via P2P/IPC
+ip-26-0-160-242:1364904:1365408 [2] NCCL INFO Channel 10/0 : 3[3] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364905:1365404 [3] NCCL INFO Channel 02/0 : 11[3] -> 18[2] [send] via NET/Libfabric/2(10)/GDRDMA
+ip-26-0-161-123:32704:33197 [0] NCCL INFO Channel 14/0 : 24[0] -> 29[5] via P2P/IPC
+ip-26-0-160-242:1364905:1365404 [3] NCCL INFO Channel 10/0 : 11[3] -> 18[2] [send] via NET/Libfabric/2(10)/GDRDMA
+ip-26-0-168-30:235648:236145 [0] NCCL INFO Channel 06/0 : 8[0] -> 13[5] via P2P/IPC
+ip-26-0-161-103:96891:97385 [3] NCCL INFO Channel 03/0 : 12[4] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96888:97387 [0] NCCL INFO Channel 14/0 : 16[0] -> 21[5] via P2P/IPC
+ip-26-0-161-103:96891:97385 [3] NCCL INFO Channel 11/0 : 12[4] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235648:236145 [0] NCCL INFO Channel 14/0 : 8[0] -> 13[5] via P2P/IPC
+ip-26-0-161-103:96891:97385 [3] NCCL INFO Channel 02/0 : 19[3] -> 26[2] [send] via NET/Libfabric/2(18)/GDRDMA
+ip-26-0-161-103:96891:97385 [3] NCCL INFO Channel 10/0 : 19[3] -> 26[2] [send] via NET/Libfabric/2(18)/GDRDMA
+ip-26-0-161-123:32704:33197 [0] NCCL INFO Channel 07/0 : 24[0] -> 30[6] via P2P/IPC
+ip-26-0-161-103:96888:97387 [0] NCCL INFO Channel 07/0 : 16[0] -> 22[6] via P2P/IPC
+ip-26-0-168-30:235648:236145 [0] NCCL INFO Channel 07/0 : 8[0] -> 14[6] via P2P/IPC
+ip-26-0-161-123:32704:33197 [0] NCCL INFO Channel 15/0 : 24[0] -> 30[6] via P2P/IPC
+ip-26-0-161-103:96888:97387 [0] NCCL INFO Channel 15/0 : 16[0] -> 22[6] via P2P/IPC
+ip-26-0-160-242:1364902:1365410 [0] NCCL INFO Channel 06/0 : 8[0] -> 13[5] via P2P/IPC
+ip-26-0-168-30:235648:236145 [0] NCCL INFO Channel 15/0 : 8[0] -> 14[6] via P2P/IPC
+ip-26-0-160-225:124162:124660 [2] NCCL INFO Channel 02/0 : 27[3] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:124162:124660 [2] NCCL INFO Channel 10/0 : 27[3] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:124163:124661 [3] NCCL INFO Channel 02/0 : 3[3] -> 10[2] [send] via NET/Libfabric/2(2)/GDRDMA
+ip-26-0-160-225:124163:124661 [3] NCCL INFO Channel 10/0 : 3[3] -> 10[2] [send] via NET/Libfabric/2(2)/GDRDMA
+ip-26-0-161-123:32704:33197 [0] NCCL INFO Channel 01/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-168-30:235648:236145 [0] NCCL INFO Channel 01/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-160-242:1364902:1365410 [0] NCCL INFO Channel 14/0 : 8[0] -> 13[5] via P2P/IPC
+ip-26-0-161-103:96888:97387 [0] NCCL INFO Channel 01/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-161-123:32704:33197 [0] NCCL INFO Channel 02/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-168-30:235648:236145 [0] NCCL INFO Channel 02/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-161-103:96888:97387 [0] NCCL INFO Channel 02/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-160-225:124160:124658 [0] NCCL INFO Channel 06/0 : 0[0] -> 5[5] via P2P/IPC
+ip-26-0-161-123:32704:33197 [0] NCCL INFO Channel 03/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-160-242:1364902:1365410 [0] NCCL INFO Channel 07/0 : 8[0] -> 14[6] via P2P/IPC
+ip-26-0-168-30:235648:236145 [0] NCCL INFO Channel 03/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-161-103:96888:97387 [0] NCCL INFO Channel 03/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-160-225:124160:124658 [0] NCCL INFO Channel 14/0 : 0[0] -> 5[5] via P2P/IPC
+ip-26-0-161-123:32704:33197 [0] NCCL INFO Channel 04/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-161-103:96888:97387 [0] NCCL INFO Channel 04/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-168-30:235648:236145 [0] NCCL INFO Channel 04/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-160-242:1364902:1365410 [0] NCCL INFO Channel 15/0 : 8[0] -> 14[6] via P2P/IPC
+ip-26-0-161-123:32704:33197 [0] NCCL INFO Channel 09/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-168-30:235648:236145 [0] NCCL INFO Channel 09/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-161-103:96888:97387 [0] NCCL INFO Channel 09/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-160-225:124160:124658 [0] NCCL INFO Channel 07/0 : 0[0] -> 6[6] via P2P/IPC
+ip-26-0-161-123:32706:33199 [2] NCCL INFO Channel 02/0 : 19[3] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96895:97382 [7] NCCL INFO Channel 06/0 : 23[7] -> 30[6] [send] via NET/Libfabric/6(22)/GDRDMA
+ip-26-0-161-123:32707:33200 [3] NCCL INFO Channel 03/0 : 20[4] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32711:33204 [7] NCCL INFO Channel 06/0 : 31[7] -> 6[6] [send] via NET/Libfabric/6(30)/GDRDMA
+ip-26-0-161-103:96895:97382 [7] NCCL INFO Channel 14/0 : 23[7] -> 30[6] [send] via NET/Libfabric/6(22)/GDRDMA
+ip-26-0-161-123:32706:33199 [2] NCCL INFO Channel 10/0 : 19[3] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32707:33200 [3] NCCL INFO Channel 11/0 : 20[4] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235648:236145 [0] NCCL INFO Channel 10/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-160-242:1364902:1365410 [0] NCCL INFO Channel 01/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-161-123:32704:33197 [0] NCCL INFO Channel 10/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-161-103:96888:97387 [0] NCCL INFO Channel 10/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-161-123:32707:33200 [3] NCCL INFO Channel 02/0 : 27[3] -> 2[2] [send] via NET/Libfabric/2(26)/GDRDMA
+ip-26-0-160-225:124160:124658 [0] NCCL INFO Channel 15/0 : 0[0] -> 6[6] via P2P/IPC
+ip-26-0-161-123:32707:33200 [3] NCCL INFO Channel 10/0 : 27[3] -> 2[2] [send] via NET/Libfabric/2(26)/GDRDMA
+ip-26-0-168-30:235648:236145 [0] NCCL INFO Channel 11/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-160-242:1364902:1365410 [0] NCCL INFO Channel 02/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-161-103:96888:97387 [0] NCCL INFO Channel 11/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-161-123:32704:33197 [0] NCCL INFO Channel 11/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-160-225:124160:124658 [0] NCCL INFO Channel 01/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-160-242:1364902:1365410 [0] NCCL INFO Channel 03/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-161-103:96888:97387 [0] NCCL INFO Channel 12/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-161-123:32704:33197 [0] NCCL INFO Channel 12/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-160-225:124167:124664 [7] NCCL INFO Channel 06/0 : 7[7] -> 14[6] [send] via NET/Libfabric/6(6)/GDRDMA
+ip-26-0-160-225:124160:124658 [0] NCCL INFO Channel 02/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-160-242:1364902:1365410 [0] NCCL INFO Channel 04/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-160-225:124160:124658 [0] NCCL INFO Channel 03/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-160-242:1364902:1365410 [0] NCCL INFO Channel 09/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-160-225:124160:124658 [0] NCCL INFO Channel 04/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-160-242:1364902:1365410 [0] NCCL INFO Channel 10/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-160-225:124160:124658 [0] NCCL INFO Channel 09/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-160-242:1364902:1365410 [0] NCCL INFO Channel 11/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-168-30:235648:236145 [0] NCCL INFO Channel 12/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-160-242:1364902:1365410 [0] NCCL INFO Channel 12/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-160-225:124160:124658 [0] NCCL INFO Channel 10/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-160-225:124160:124658 [0] NCCL INFO Channel 11/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-160-225:124160:124658 [0] NCCL INFO Channel 12/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-165-59:1557266:1557759 [0] NCCL INFO Channel 05/0 : 0[0] -> 4[4] via P2P/IPC
+ip-26-0-165-59:1557266:1557759 [0] NCCL INFO Channel 13/0 : 0[0] -> 4[4] via P2P/IPC
+ip-26-0-168-52:2722045:2722544 [0] NCCL INFO Channel 06/0 : 24[0] -> 29[5] via P2P/IPC
+ip-26-0-168-52:2722045:2722544 [0] NCCL INFO Channel 14/0 : 24[0] -> 29[5] via P2P/IPC
+ip-26-0-168-52:2722045:2722544 [0] NCCL INFO Channel 07/0 : 24[0] -> 30[6] via P2P/IPC
+ip-26-0-168-52:2722045:2722544 [0] NCCL INFO Channel 15/0 : 24[0] -> 30[6] via P2P/IPC
+ip-26-0-168-52:2722045:2722544 [0] NCCL INFO Channel 01/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-168-52:2722045:2722544 [0] NCCL INFO Channel 02/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-168-52:2722045:2722544 [0] NCCL INFO Channel 03/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-168-52:2722045:2722544 [0] NCCL INFO Channel 04/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-168-52:2722045:2722544 [0] NCCL INFO Channel 09/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-168-52:2722045:2722544 [0] NCCL INFO Channel 10/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-168-52:2722045:2722544 [0] NCCL INFO Channel 11/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-168-52:2722045:2722544 [0] NCCL INFO Channel 12/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-168-52:2722052:2722546 [7] NCCL INFO Channel 06/0 : 31[7] -> 6[6] [send] via NET/Libfabric/6(30)/GDRDMA
+ip-26-0-168-30:235649:236143 [1] NCCL INFO Channel 01/0 : 2[2] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235649:236143 [1] NCCL INFO Channel 09/0 : 2[2] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235650:236144 [2] NCCL INFO Channel 01/0 : 10[2] -> 17[1] [send] via NET/Libfabric/1(9)/GDRDMA
+ip-26-0-168-30:235650:236144 [2] NCCL INFO Channel 09/0 : 10[2] -> 17[1] [send] via NET/Libfabric/1(9)/GDRDMA
+ip-26-0-168-34:272151:272646 [1] NCCL INFO Channel 01/0 : 10[2] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272151:272646 [1] NCCL INFO Channel 09/0 : 10[2] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272152:272647 [2] NCCL INFO Channel 01/0 : 18[2] -> 25[1] [send] via NET/Libfabric/1(17)/GDRDMA
+ip-26-0-168-34:272152:272647 [2] NCCL INFO Channel 09/0 : 18[2] -> 25[1] [send] via NET/Libfabric/1(17)/GDRDMA
+ip-26-0-168-52:2722046:2722541 [1] NCCL INFO Channel 01/0 : 18[2] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-52:2722047:2722542 [2] NCCL INFO Channel 01/0 : 26[2] -> 1[1] [send] via NET/Libfabric/1(25)/GDRDMA
+ip-26-0-168-52:2722046:2722541 [1] NCCL INFO Channel 09/0 : 18[2] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-52:2722047:2722542 [2] NCCL INFO Channel 09/0 : 26[2] -> 1[1] [send] via NET/Libfabric/1(25)/GDRDMA
+ip-26-0-168-30:235649:236143 [1] NCCL INFO Channel 01/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-168-30:235649:236143 [1] NCCL INFO Channel 02/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-168-30:235649:236143 [1] NCCL INFO Channel 03/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-168-34:272151:272646 [1] NCCL INFO Channel 01/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-168-30:235650:236144 [2] NCCL INFO Channel 02/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-168-52:2722046:2722541 [1] NCCL INFO Channel 01/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-168-30:235649:236143 [1] NCCL INFO Channel 04/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-168-34:272151:272646 [1] NCCL INFO Channel 02/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-168-52:2722046:2722541 [1] NCCL INFO Channel 02/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-168-30:235650:236144 [2] NCCL INFO Channel 03/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-168-30:235649:236143 [1] NCCL INFO Channel 05/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-168-52:2722046:2722541 [1] NCCL INFO Channel 03/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-168-34:272151:272646 [1] NCCL INFO Channel 03/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-168-52:2722046:2722541 [1] NCCL INFO Channel 04/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-168-30:235650:236144 [2] NCCL INFO Channel 04/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-168-30:235649:236143 [1] NCCL INFO Channel 06/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-168-34:272154:272650 [4] NCCL INFO Channel 04/0 : 13[5] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722046:2722541 [1] NCCL INFO Channel 05/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-168-34:272151:272646 [1] NCCL INFO Channel 04/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-168-34:272154:272650 [4] NCCL INFO Channel 12/0 : 13[5] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272154:272650 [4] NCCL INFO Channel 03/0 : 20[4] -> 27[3] [send] via NET/Libfabric/3(19)/GDRDMA
+ip-26-0-168-52:2722046:2722541 [1] NCCL INFO Channel 06/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-168-30:235650:236144 [2] NCCL INFO Channel 05/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-168-34:272154:272650 [4] NCCL INFO Channel 11/0 : 20[4] -> 27[3] [send] via NET/Libfabric/3(19)/GDRDMA
+ip-26-0-168-30:235649:236143 [1] NCCL INFO Channel 07/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-168-52:2722046:2722541 [1] NCCL INFO Channel 07/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-168-30:235650:236144 [2] NCCL INFO Channel 06/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-168-52:2722046:2722541 [1] NCCL INFO Channel 09/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-168-30:235649:236143 [1] NCCL INFO Channel 09/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-168-34:272151:272646 [1] NCCL INFO Channel 05/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-168-34:272152:272647 [2] NCCL INFO Channel 02/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-168-30:235650:236144 [2] NCCL INFO Channel 07/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-168-52:2722046:2722541 [1] NCCL INFO Channel 10/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-168-30:235649:236143 [1] NCCL INFO Channel 10/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-165-164:1227470:1227958 [7] NCCL INFO Channel 07/0 : 12[4] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722046:2722541 [1] NCCL INFO Channel 11/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-165-164:1227470:1227958 [7] NCCL INFO Channel 15/0 : 12[4] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272151:272646 [1] NCCL INFO Channel 06/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-168-34:272152:272647 [2] NCCL INFO Channel 03/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-168-30:235650:236144 [2] NCCL INFO Channel 10/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-165-131:1630212:1630734 [7] NCCL INFO Channel 07/0 : 4[4] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235649:236143 [1] NCCL INFO Channel 11/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-165-131:1630212:1630734 [7] NCCL INFO Channel 15/0 : 4[4] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722046:2722541 [1] NCCL INFO Channel 12/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-165-164:1227470:1227958 [7] NCCL INFO Channel 07/0 : 23[7] -> 19[3] via P2P/IPC
+ip-26-0-168-34:272156:272648 [6] NCCL INFO Channel 06/0 : 15[7] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272157:272649 [7] NCCL INFO Channel 14/0 : 23[7] -> 30[6] [send] via NET/Libfabric/6(22)/GDRDMA
+ip-26-0-168-30:235650:236144 [2] NCCL INFO Channel 11/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-168-34:272156:272648 [6] NCCL INFO Channel 14/0 : 15[7] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235649:236143 [1] NCCL INFO Channel 12/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-168-52:2722046:2722541 [1] NCCL INFO Channel 13/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-168-52:2722046:2722541 [1] NCCL INFO Channel 14/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-168-30:235650:236144 [2] NCCL INFO Channel 12/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-168-30:235649:236143 [1] NCCL INFO Channel 13/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-168-34:272155:272652 [5] NCCL INFO Channel 05/0 : 14[6] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-52:2722046:2722541 [1] NCCL INFO Channel 15/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-161-78:114519:115014 [7] NCCL INFO Channel 07/0 : 28[4] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272156:272648 [6] NCCL INFO Channel 05/0 : 22[6] -> 29[5] [send] via NET/Libfabric/5(21)/GDRDMA
+ip-26-0-161-78:114516:115010 [4] NCCL INFO Channel 07/0 : 4[4] -> 15[7] [send] via NET/Libfabric/7(7)/GDRDMA
+ip-26-0-168-34:272155:272652 [5] NCCL INFO Channel 13/0 : 14[6] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114519:115014 [7] NCCL INFO Channel 15/0 : 28[4] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272156:272648 [6] NCCL INFO Channel 13/0 : 22[6] -> 29[5] [send] via NET/Libfabric/5(21)/GDRDMA
+ip-26-0-161-78:114516:115010 [4] NCCL INFO Channel 15/0 : 4[4] -> 15[7] [send] via NET/Libfabric/7(7)/GDRDMA
+ip-26-0-168-30:235650:236144 [2] NCCL INFO Channel 13/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-168-30:235649:236143 [1] NCCL INFO Channel 14/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-168-34:272155:272652 [5] NCCL INFO Channel 04/0 : 21[5] -> 28[4] [send] via NET/Libfabric/4(20)/GDRDMA
+ip-26-0-168-34:272152:272647 [2] NCCL INFO Channel 04/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-168-34:272151:272646 [1] NCCL INFO Channel 07/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-168-34:272155:272652 [5] NCCL INFO Channel 12/0 : 21[5] -> 28[4] [send] via NET/Libfabric/4(20)/GDRDMA
+ip-26-0-165-131:1630212:1630734 [7] NCCL INFO Channel 07/0 : 15[7] -> 11[3] via P2P/IPC
+ip-26-0-168-30:235650:236144 [2] NCCL INFO Channel 14/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-168-30:235649:236143 [1] NCCL INFO Channel 15/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-168-34:272152:272647 [2] NCCL INFO Channel 05/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-168-34:272151:272646 [1] NCCL INFO Channel 09/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-165-202:153569:154055 [7] NCCL INFO Channel 07/0 : 20[4] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153569:154055 [7] NCCL INFO Channel 15/0 : 20[4] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153566:154062 [4] NCCL INFO Channel 07/0 : 28[4] -> 7[7] [send] via NET/Libfabric/7(31)/GDRDMA
+ip-26-0-165-202:153566:154062 [4] NCCL INFO Channel 15/0 : 28[4] -> 7[7] [send] via NET/Libfabric/7(31)/GDRDMA
+ip-26-0-168-30:235650:236144 [2] NCCL INFO Channel 15/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-165-202:153569:154055 [7] NCCL INFO Channel 07/0 : 31[7] -> 27[3] via P2P/IPC
+ip-26-0-161-78:114519:115014 [7] NCCL INFO Channel 07/0 : 7[7] -> 3[3] via P2P/IPC
+ip-26-0-168-34:272152:272647 [2] NCCL INFO Channel 06/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-168-34:272151:272646 [1] NCCL INFO Channel 10/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-168-34:272152:272647 [2] NCCL INFO Channel 07/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-168-34:272151:272646 [1] NCCL INFO Channel 11/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-168-34:272152:272647 [2] NCCL INFO Channel 10/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-168-34:272151:272646 [1] NCCL INFO Channel 12/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-165-131:1630212:1630734 [7] NCCL INFO Channel 15/0 : 15[7] -> 11[3] via P2P/IPC
+ip-26-0-165-164:1227470:1227958 [7] NCCL INFO Channel 15/0 : 23[7] -> 19[3] via P2P/IPC
+ip-26-0-168-34:272152:272647 [2] NCCL INFO Channel 11/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-168-34:272151:272646 [1] NCCL INFO Channel 13/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-161-78:114519:115014 [7] NCCL INFO Channel 15/0 : 7[7] -> 3[3] via P2P/IPC
+ip-26-0-168-34:272152:272647 [2] NCCL INFO Channel 12/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-168-34:272151:272646 [1] NCCL INFO Channel 14/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-161-103:96889:97388 [1] NCCL INFO Channel 01/0 : 10[2] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96889:97388 [1] NCCL INFO Channel 09/0 : 10[2] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96890:97386 [2] NCCL INFO Channel 01/0 : 18[2] -> 25[1] [send] via NET/Libfabric/1(17)/GDRDMA
+ip-26-0-161-103:96890:97386 [2] NCCL INFO Channel 09/0 : 18[2] -> 25[1] [send] via NET/Libfabric/1(17)/GDRDMA
+ip-26-0-168-34:272152:272647 [2] NCCL INFO Channel 13/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-168-34:272151:272646 [1] NCCL INFO Channel 15/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-168-34:272152:272647 [2] NCCL INFO Channel 14/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-165-131:1630212:1630734 [7] NCCL INFO Channel 01/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-168-34:272152:272647 [2] NCCL INFO Channel 15/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-160-225:124161:124659 [1] NCCL INFO Channel 01/0 : 26[2] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124161:124659 [1] NCCL INFO Channel 09/0 : 26[2] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124162:124660 [2] NCCL INFO Channel 01/0 : 2[2] -> 9[1] [send] via NET/Libfabric/1(1)/GDRDMA
+ip-26-0-160-225:124162:124660 [2] NCCL INFO Channel 09/0 : 2[2] -> 9[1] [send] via NET/Libfabric/1(1)/GDRDMA
+ip-26-0-160-242:1364903:1365409 [1] NCCL INFO Channel 01/0 : 2[2] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364904:1365408 [2] NCCL INFO Channel 01/0 : 10[2] -> 17[1] [send] via NET/Libfabric/1(9)/GDRDMA
+ip-26-0-160-242:1364903:1365409 [1] NCCL INFO Channel 09/0 : 2[2] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364904:1365408 [2] NCCL INFO Channel 09/0 : 10[2] -> 17[1] [send] via NET/Libfabric/1(9)/GDRDMA
+ip-26-0-161-78:114519:115014 [7] NCCL INFO Channel 01/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-103:96892:97383 [4] NCCL INFO Channel 04/0 : 13[5] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96892:97383 [4] NCCL INFO Channel 12/0 : 13[5] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227470:1227958 [7] NCCL INFO Channel 01/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-161-103:96892:97383 [4] NCCL INFO Channel 03/0 : 20[4] -> 27[3] [send] via NET/Libfabric/3(19)/GDRDMA
+ip-26-0-161-103:96892:97383 [4] NCCL INFO Channel 11/0 : 20[4] -> 27[3] [send] via NET/Libfabric/3(19)/GDRDMA
+ip-26-0-165-202:153569:154055 [7] NCCL INFO Channel 15/0 : 31[7] -> 27[3] via P2P/IPC
+ip-26-0-165-131:1630212:1630734 [7] NCCL INFO Channel 02/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-161-123:32705:33203 [1] NCCL INFO Channel 01/0 : 18[2] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32705:33203 [1] NCCL INFO Channel 09/0 : 18[2] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32706:33199 [2] NCCL INFO Channel 01/0 : 26[2] -> 1[1] [send] via NET/Libfabric/1(25)/GDRDMA
+ip-26-0-161-123:32706:33199 [2] NCCL INFO Channel 09/0 : 26[2] -> 1[1] [send] via NET/Libfabric/1(25)/GDRDMA
+ip-26-0-160-225:124164:124663 [4] NCCL INFO Channel 04/0 : 29[5] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:124164:124663 [4] NCCL INFO Channel 12/0 : 29[5] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114519:115014 [7] NCCL INFO Channel 02/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-225:124164:124663 [4] NCCL INFO Channel 03/0 : 4[4] -> 11[3] [send] via NET/Libfabric/3(3)/GDRDMA
+ip-26-0-161-103:96893:97384 [5] NCCL INFO Channel 05/0 : 14[6] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124164:124663 [4] NCCL INFO Channel 11/0 : 4[4] -> 11[3] [send] via NET/Libfabric/3(3)/GDRDMA
+ip-26-0-161-103:96893:97384 [5] NCCL INFO Channel 13/0 : 14[6] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227470:1227958 [7] NCCL INFO Channel 02/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-161-78:114519:115014 [7] NCCL INFO Channel 03/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-103:96893:97384 [5] NCCL INFO Channel 04/0 : 21[5] -> 28[4] [send] via NET/Libfabric/4(20)/GDRDMA
+ip-26-0-165-131:1630212:1630734 [7] NCCL INFO Channel 03/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-161-103:96893:97384 [5] NCCL INFO Channel 12/0 : 21[5] -> 28[4] [send] via NET/Libfabric/4(20)/GDRDMA
+ip-26-0-160-225:124166:124665 [6] NCCL INFO Channel 06/0 : 31[7] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:124166:124665 [6] NCCL INFO Channel 14/0 : 31[7] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114519:115014 [7] NCCL INFO Channel 04/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-225:124167:124664 [7] NCCL INFO Channel 14/0 : 7[7] -> 14[6] [send] via NET/Libfabric/6(6)/GDRDMA
+ip-26-0-165-202:153569:154055 [7] NCCL INFO Channel 01/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-168-30:235654:236146 [6] NCCL INFO Channel 06/0 : 7[7] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:124162:124660 [2] NCCL INFO Channel 02/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-168-30:235654:236146 [6] NCCL INFO Channel 14/0 : 7[7] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630212:1630734 [7] NCCL INFO Channel 04/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-160-225:124162:124660 [2] NCCL INFO Channel 03/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-78:114519:115014 [7] NCCL INFO Channel 05/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-123:32708:33198 [4] NCCL INFO Channel 04/0 : 21[5] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32708:33198 [4] NCCL INFO Channel 12/0 : 21[5] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235653:236140 [5] NCCL INFO Channel 05/0 : 6[6] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235653:236140 [5] NCCL INFO Channel 13/0 : 6[6] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235654:236146 [6] NCCL INFO Channel 05/0 : 14[6] -> 21[5] [send] via NET/Libfabric/5(13)/GDRDMA
+ip-26-0-160-225:124162:124660 [2] NCCL INFO Channel 04/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-123:32708:33198 [4] NCCL INFO Channel 03/0 : 28[4] -> 3[3] [send] via NET/Libfabric/3(27)/GDRDMA
+ip-26-0-168-30:235654:236146 [6] NCCL INFO Channel 13/0 : 14[6] -> 21[5] [send] via NET/Libfabric/5(13)/GDRDMA
+ip-26-0-161-123:32708:33198 [4] NCCL INFO Channel 11/0 : 28[4] -> 3[3] [send] via NET/Libfabric/3(27)/GDRDMA
+ip-26-0-161-78:114519:115014 [7] NCCL INFO Channel 09/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-225:124165:124662 [5] NCCL INFO Channel 05/0 : 30[6] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124166:124665 [6] NCCL INFO Channel 05/0 : 6[6] -> 13[5] [send] via NET/Libfabric/5(5)/GDRDMA
+ip-26-0-165-202:153569:154055 [7] NCCL INFO Channel 02/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-160-225:124165:124662 [5] NCCL INFO Channel 13/0 : 30[6] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124162:124660 [2] NCCL INFO Channel 05/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-225:124166:124665 [6] NCCL INFO Channel 13/0 : 6[6] -> 13[5] [send] via NET/Libfabric/5(5)/GDRDMA
+ip-26-0-160-242:1364906:1365407 [4] NCCL INFO Channel 04/0 : 5[5] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96894:97381 [6] NCCL INFO Channel 06/0 : 15[7] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364907:1365403 [5] NCCL INFO Channel 05/0 : 6[6] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227470:1227958 [7] NCCL INFO Channel 03/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-161-103:96894:97381 [6] NCCL INFO Channel 14/0 : 15[7] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:124165:124662 [5] NCCL INFO Channel 04/0 : 5[5] -> 12[4] [send] via NET/Libfabric/4(4)/GDRDMA
+ip-26-0-160-242:1364906:1365407 [4] NCCL INFO Channel 12/0 : 5[5] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364907:1365403 [5] NCCL INFO Channel 13/0 : 6[6] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124165:124662 [5] NCCL INFO Channel 12/0 : 5[5] -> 12[4] [send] via NET/Libfabric/4(4)/GDRDMA
+ip-26-0-165-202:153569:154055 [7] NCCL INFO Channel 03/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-161-123:32710:33202 [6] NCCL INFO Channel 06/0 : 23[7] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114519:115014 [7] NCCL INFO Channel 10/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-103:96894:97381 [6] NCCL INFO Channel 05/0 : 22[6] -> 29[5] [send] via NET/Libfabric/5(21)/GDRDMA
+ip-26-0-160-242:1364907:1365403 [5] NCCL INFO Channel 04/0 : 13[5] -> 20[4] [send] via NET/Libfabric/4(12)/GDRDMA
+ip-26-0-161-123:32711:33204 [7] NCCL INFO Channel 14/0 : 31[7] -> 6[6] [send] via NET/Libfabric/6(30)/GDRDMA
+ip-26-0-161-103:96894:97381 [6] NCCL INFO Channel 13/0 : 22[6] -> 29[5] [send] via NET/Libfabric/5(21)/GDRDMA
+ip-26-0-160-242:1364907:1365403 [5] NCCL INFO Channel 12/0 : 13[5] -> 20[4] [send] via NET/Libfabric/4(12)/GDRDMA
+ip-26-0-161-123:32710:33202 [6] NCCL INFO Channel 14/0 : 23[7] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364906:1365407 [4] NCCL INFO Channel 03/0 : 12[4] -> 19[3] [send] via NET/Libfabric/3(11)/GDRDMA
+ip-26-0-160-242:1364906:1365407 [4] NCCL INFO Channel 11/0 : 12[4] -> 19[3] [send] via NET/Libfabric/3(11)/GDRDMA
+ip-26-0-160-225:124166:124665 [6] NCCL INFO Channel 06/0 : 6[6] -> 3[3] via P2P/IPC
+ip-26-0-161-78:114516:115010 [4] NCCL INFO Channel 01/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-165-164:1227470:1227958 [7] NCCL INFO Channel 04/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-165-202:153569:154055 [7] NCCL INFO Channel 04/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-161-78:114519:115014 [7] NCCL INFO Channel 11/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-242:1364907:1365403 [5] NCCL INFO Channel 05/0 : 13[5] -> 11[3] via P2P/IPC
+ip-26-0-160-225:124162:124660 [2] NCCL INFO Channel 06/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-168-30:235652:236147 [4] NCCL INFO Channel 04/0 : 5[5] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364908:1365406 [6] NCCL INFO Channel 06/0 : 7[7] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235652:236147 [4] NCCL INFO Channel 12/0 : 5[5] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32709:33201 [5] NCCL INFO Channel 05/0 : 22[6] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364909:1365405 [7] NCCL INFO Channel 06/0 : 15[7] -> 22[6] [send] via NET/Libfabric/6(14)/GDRDMA
+ip-26-0-168-30:235653:236140 [5] NCCL INFO Channel 04/0 : 13[5] -> 20[4] [send] via NET/Libfabric/4(12)/GDRDMA
+ip-26-0-160-242:1364908:1365406 [6] NCCL INFO Channel 14/0 : 7[7] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32709:33201 [5] NCCL INFO Channel 13/0 : 22[6] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364909:1365405 [7] NCCL INFO Channel 14/0 : 15[7] -> 22[6] [send] via NET/Libfabric/6(14)/GDRDMA
+ip-26-0-161-123:32710:33202 [6] NCCL INFO Channel 05/0 : 30[6] -> 5[5] [send] via NET/Libfabric/5(29)/GDRDMA
+ip-26-0-161-78:114516:115010 [4] NCCL INFO Channel 02/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-78:114519:115014 [7] NCCL INFO Channel 12/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-123:32705:33203 [1] NCCL INFO Channel 01/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-168-30:235653:236140 [5] NCCL INFO Channel 12/0 : 13[5] -> 20[4] [send] via NET/Libfabric/4(12)/GDRDMA
+ip-26-0-161-123:32710:33202 [6] NCCL INFO Channel 13/0 : 30[6] -> 5[5] [send] via NET/Libfabric/5(29)/GDRDMA
+ip-26-0-161-123:32709:33201 [5] NCCL INFO Channel 04/0 : 29[5] -> 4[4] [send] via NET/Libfabric/4(28)/GDRDMA
+ip-26-0-168-30:235652:236147 [4] NCCL INFO Channel 03/0 : 12[4] -> 19[3] [send] via NET/Libfabric/3(11)/GDRDMA
+ip-26-0-165-164:1227470:1227958 [7] NCCL INFO Channel 05/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-161-123:32709:33201 [5] NCCL INFO Channel 12/0 : 29[5] -> 4[4] [send] via NET/Libfabric/4(28)/GDRDMA
+ip-26-0-168-30:235652:236147 [4] NCCL INFO Channel 11/0 : 12[4] -> 19[3] [send] via NET/Libfabric/3(11)/GDRDMA
+ip-26-0-165-202:153569:154055 [7] NCCL INFO Channel 05/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-160-242:1364907:1365403 [5] NCCL INFO Channel 13/0 : 13[5] -> 11[3] via P2P/IPC
+ip-26-0-161-123:32708:33198 [4] NCCL INFO Channel 07/0 : 28[4] -> 7[7] [send] via NET/Libfabric/7(31)/GDRDMA
+ip-26-0-160-225:124165:124662 [5] NCCL INFO Channel 05/0 : 5[5] -> 3[3] via P2P/IPC
+ip-26-0-161-123:32710:33202 [6] NCCL INFO Channel 06/0 : 30[6] -> 27[3] via P2P/IPC
+ip-26-0-161-103:96894:97381 [6] NCCL INFO Channel 06/0 : 22[6] -> 19[3] via P2P/IPC
+ip-26-0-160-225:124166:124665 [6] NCCL INFO Channel 14/0 : 6[6] -> 3[3] via P2P/IPC
+ip-26-0-161-123:32706:33199 [2] NCCL INFO Channel 02/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-161-123:32709:33201 [5] NCCL INFO Channel 05/0 : 29[5] -> 27[3] via P2P/IPC
+ip-26-0-160-225:124162:124660 [2] NCCL INFO Channel 07/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-123:32705:33203 [1] NCCL INFO Channel 02/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-161-78:114519:115014 [7] NCCL INFO Channel 13/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-78:114516:115010 [4] NCCL INFO Channel 04/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-165-164:1227470:1227958 [7] NCCL INFO Channel 09/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-165-202:153569:154055 [7] NCCL INFO Channel 09/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-160-242:1364908:1365406 [6] NCCL INFO Channel 05/0 : 14[6] -> 21[5] [send] via NET/Libfabric/5(13)/GDRDMA
+ip-26-0-160-242:1364908:1365406 [6] NCCL INFO Channel 13/0 : 14[6] -> 21[5] [send] via NET/Libfabric/5(13)/GDRDMA
+ip-26-0-161-103:96892:97383 [4] NCCL INFO Channel 07/0 : 20[4] -> 31[7] [send] via NET/Libfabric/7(23)/GDRDMA
+ip-26-0-160-242:1364906:1365407 [4] NCCL INFO Channel 07/0 : 12[4] -> 23[7] [send] via NET/Libfabric/7(15)/GDRDMA
+ip-26-0-165-131:1630212:1630734 [7] NCCL INFO Channel 05/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-161-103:96892:97383 [4] NCCL INFO Channel 15/0 : 20[4] -> 31[7] [send] via NET/Libfabric/7(23)/GDRDMA
+ip-26-0-160-242:1364908:1365406 [6] NCCL INFO Channel 06/0 : 14[6] -> 11[3] via P2P/IPC
+ip-26-0-165-164:1227470:1227958 [7] NCCL INFO Channel 10/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-160-225:124162:124660 [2] NCCL INFO Channel 10/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-225:124165:124662 [5] NCCL INFO Channel 13/0 : 5[5] -> 3[3] via P2P/IPC
+ip-26-0-161-123:32706:33199 [2] NCCL INFO Channel 03/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-161-123:32710:33202 [6] NCCL INFO Channel 14/0 : 30[6] -> 27[3] via P2P/IPC
+ip-26-0-161-103:96893:97384 [5] NCCL INFO Channel 05/0 : 21[5] -> 19[3] via P2P/IPC
+ip-26-0-165-202:153569:154055 [7] NCCL INFO Channel 10/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-161-123:32705:33203 [1] NCCL INFO Channel 03/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-161-123:32709:33201 [5] NCCL INFO Channel 13/0 : 29[5] -> 27[3] via P2P/IPC
+ip-26-0-161-78:114516:115010 [4] NCCL INFO Channel 09/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-165-131:1630212:1630734 [7] NCCL INFO Channel 09/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-161-103:96894:97381 [6] NCCL INFO Channel 14/0 : 22[6] -> 19[3] via P2P/IPC
+ip-26-0-160-225:124164:124663 [4] NCCL INFO Channel 07/0 : 4[4] -> 15[7] [send] via NET/Libfabric/7(7)/GDRDMA
+ip-26-0-165-202:153569:154055 [7] NCCL INFO Channel 11/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-165-164:1227470:1227958 [7] NCCL INFO Channel 11/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-161-153:2636086:2636602 [2] NCCL INFO Trees [0] 19/-1/-1->18->17 [1] 19/-1/-1->18->17 [2] 19/10/26->18->2 [3] -1/-1/-1->18->17 [4] 19/-1/-1->18->17 [5] 19/-1/-1->18->17 [6] 19/-1/-1->18->17 [7] 19/-1/-1->18->17 [8] 19/-1/-1->18->17 [9] 19/-1/-1->18->17 [10] 19/-1/-1->18->10 [11] -1/-1/-1->18->17 [12] 19/-1/-1->18->17 [13] 19/-1/-1->18->17 [14] 19/-1/-1->18->17 [15] 19/-1/-1->18->17
+ip-26-0-161-153:2636086:2636602 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-78:114516:115010 [4] NCCL INFO Channel 10/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-225:124162:124660 [2] NCCL INFO Channel 11/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-165-131:1630212:1630734 [7] NCCL INFO Channel 10/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-161-153:2636087:2636603 [3] NCCL INFO Trees [0] 20/-1/-1->19->18 [1] 20/-1/-1->19->18 [2] 20/-1/-1->19->18 [3] 20/11/27->19->3 [4] -1/-1/-1->19->18 [5] 20/-1/-1->19->18 [6] 20/-1/-1->19->18 [7] 20/-1/-1->19->18 [8] 20/-1/-1->19->18 [9] 20/-1/-1->19->18 [10] 20/-1/-1->19->18 [11] 20/-1/-1->19->11 [12] -1/-1/-1->19->18 [13] 20/-1/-1->19->18 [14] 20/-1/-1->19->18 [15] 20/-1/-1->19->18
+ip-26-0-161-153:2636087:2636603 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-142:62888:63380 [7] NCCL INFO Trees [0] -1/-1/-1->15->14 [1] 8/-1/-1->15->14 [2] 8/-1/-1->15->14 [3] 8/-1/-1->15->14 [4] 8/-1/-1->15->14 [5] 8/-1/-1->15->14 [6] 8/-1/-1->15->14 [7] 8/-1/-1->15->23 [8] -1/-1/-1->15->14 [9] 8/-1/-1->15->14 [10] 8/-1/-1->15->14 [11] 8/-1/-1->15->14 [12] 8/-1/-1->15->14 [13] 8/-1/-1->15->14 [14] 8/-1/-1->15->14 [15] 8/23/7->15->31
+ip-26-0-161-142:62888:63380 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-142:62887:63379 [6] NCCL INFO Trees [0] 15/-1/-1->14->13 [1] 15/-1/-1->14->13 [2] 15/-1/-1->14->13 [3] 15/-1/-1->14->13 [4] 15/-1/-1->14->13 [5] 15/-1/-1->14->13 [6] 15/-1/-1->14->22 [7] -1/-1/-1->14->13 [8] 15/-1/-1->14->13 [9] 15/-1/-1->14->13 [10] 15/-1/-1->14->13 [11] 15/-1/-1->14->13 [12] 15/-1/-1->14->13 [13] 15/-1/-1->14->13 [14] 15/22/6->14->30 [15] -1/-1/-1->14->13
+ip-26-0-161-153:2636085:2636604 [1] NCCL INFO Trees [0] 18/-1/-1->17->16 [1] 18/9/25->17->1 [2] -1/-1/-1->17->16 [3] 18/-1/-1->17->16 [4] 18/-1/-1->17->16 [5] 18/-1/-1->17->16 [6] 18/-1/-1->17->16 [7] 18/-1/-1->17->16 [8] 18/-1/-1->17->16 [9] 18/-1/-1->17->9 [10] -1/-1/-1->17->16 [11] 18/-1/-1->17->16 [12] 18/-1/-1->17->16 [13] 18/-1/-1->17->16 [14] 18/-1/-1->17->16 [15] 18/-1/-1->17->16
+ip-26-0-161-153:2636085:2636604 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-142:62887:63379 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-142:62883:63373 [2] NCCL INFO Trees [0] 11/-1/-1->10->9 [1] 11/-1/-1->10->9 [2] 11/-1/-1->10->18 [3] -1/-1/-1->10->9 [4] 11/-1/-1->10->9 [5] 11/-1/-1->10->9 [6] 11/-1/-1->10->9 [7] 11/-1/-1->10->9 [8] 11/-1/-1->10->9 [9] 11/-1/-1->10->9 [10] 11/18/2->10->26 [11] -1/-1/-1->10->9 [12] 11/-1/-1->10->9 [13] 11/-1/-1->10->9 [14] 11/-1/-1->10->9 [15] 11/-1/-1->10->9
+ip-26-0-161-142:62883:63373 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-103:96893:97384 [5] NCCL INFO Channel 13/0 : 21[5] -> 19[3] via P2P/IPC
+ip-26-0-161-178:2815970:2816470 [7] NCCL INFO Trees [0] -1/-1/-1->31->30 [1] 24/-1/-1->31->30 [2] 24/-1/-1->31->30 [3] 24/-1/-1->31->30 [4] 24/-1/-1->31->30 [5] 24/-1/-1->31->30 [6] 24/-1/-1->31->30 [7] 24/-1/-1->31->23 [8] -1/-1/-1->31->30 [9] 24/-1/-1->31->30 [10] 24/-1/-1->31->30 [11] 24/-1/-1->31->30 [12] 24/-1/-1->31->30 [13] 24/-1/-1->31->30 [14] 24/-1/-1->31->30 [15] 24/15/-1->31->-1
+ip-26-0-161-178:2815970:2816470 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-138:96567:97090 [4] NCCL INFO NVLS Head  0:  0  8 16 24
+ip-26-0-161-138:96568:97091 [5] NCCL INFO NVLS Head  0:  0  8 16 24
+ip-26-0-161-138:96566:97087 [3] NCCL INFO NVLS Head  0:  0  8 16 24
+ip-26-0-161-138:96567:97090 [4] NCCL INFO NVLS Head  1:  1  9 17 25
+ip-26-0-161-138:96568:97091 [5] NCCL INFO NVLS Head  1:  1  9 17 25
+ip-26-0-161-138:96567:97090 [4] NCCL INFO NVLS Head  2:  2 10 18 26
+ip-26-0-161-138:96568:97091 [5] NCCL INFO NVLS Head  2:  2 10 18 26
+ip-26-0-161-138:96566:97087 [3] NCCL INFO NVLS Head  1:  1  9 17 25
+ip-26-0-161-138:96567:97090 [4] NCCL INFO NVLS Head  3:  3 11 19 27
+ip-26-0-161-138:96568:97091 [5] NCCL INFO NVLS Head  3:  3 11 19 27
+ip-26-0-161-138:96567:97090 [4] NCCL INFO NVLS Head  4:  4 12 20 28
+ip-26-0-161-138:96566:97087 [3] NCCL INFO NVLS Head  2:  2 10 18 26
+ip-26-0-161-138:96568:97091 [5] NCCL INFO NVLS Head  4:  4 12 20 28
+ip-26-0-161-138:96570:97093 [7] NCCL INFO NVLS Head  0:  0  8 16 24
+ip-26-0-161-142:62885:63377 [4] NCCL INFO Trees [0] 13/-1/-1->12->11 [1] 13/-1/-1->12->11 [2] 13/-1/-1->12->11 [3] 13/-1/-1->12->11 [4] 13/-1/-1->12->20 [5] -1/-1/-1->12->11 [6] 13/-1/-1->12->11 [7] 13/-1/-1->12->11 [8] 13/-1/-1->12->11 [9] 13/-1/-1->12->11 [10] 13/-1/-1->12->11 [11] 13/-1/-1->12->11 [12] 13/20/4->12->28 [13] -1/-1/-1->12->11 [14] 13/-1/-1->12->11 [15] 13/-1/-1->12->11
+ip-26-0-161-142:62886:63378 [5] NCCL INFO Trees [0] 14/-1/-1->13->12 [1] 14/-1/-1->13->12 [2] 14/-1/-1->13->12 [3] 14/-1/-1->13->12 [4] 14/-1/-1->13->12 [5] 14/-1/-1->13->21 [6] -1/-1/-1->13->12 [7] 14/-1/-1->13->12 [8] 14/-1/-1->13->12 [9] 14/-1/-1->13->12 [10] 14/-1/-1->13->12 [11] 14/-1/-1->13->12 [12] 14/-1/-1->13->12 [13] 14/21/5->13->29 [14] -1/-1/-1->13->12 [15] 14/-1/-1->13->12
+ip-26-0-161-123:32706:33199 [2] NCCL INFO Channel 04/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-161-142:62884:63376 [3] NCCL INFO Trees [0] 12/-1/-1->11->10 [1] 12/-1/-1->11->10 [2] 12/-1/-1->11->10 [3] 12/-1/-1->11->19 [4] -1/-1/-1->11->10 [5] 12/-1/-1->11->10 [6] 12/-1/-1->11->10 [7] 12/-1/-1->11->10 [8] 12/-1/-1->11->10 [9] 12/-1/-1->11->10 [10] 12/-1/-1->11->10 [11] 12/19/3->11->27 [12] -1/-1/-1->11->10 [13] 12/-1/-1->11->10 [14] 12/-1/-1->11->10 [15] 12/-1/-1->11->10
+ip-26-0-161-142:62885:63377 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-142:62886:63378 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-142:62884:63376 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-142:62882:63374 [1] NCCL INFO Trees [0] 10/-1/-1->9->8 [1] 10/-1/-1->9->17 [2] -1/-1/-1->9->8 [3] 10/-1/-1->9->8 [4] 10/-1/-1->9->8 [5] 10/-1/-1->9->8 [6] 10/-1/-1->9->8 [7] 10/-1/-1->9->8 [8] 10/-1/-1->9->8 [9] 10/17/1->9->25 [10] -1/-1/-1->9->8 [11] 10/-1/-1->9->8 [12] 10/-1/-1->9->8 [13] 10/-1/-1->9->8 [14] 10/-1/-1->9->8 [15] 10/-1/-1->9->8
+ip-26-0-160-242:1364908:1365406 [6] NCCL INFO Channel 14/0 : 14[6] -> 11[3] via P2P/IPC
+ip-26-0-161-153:2636084:2636605 [0] NCCL INFO Trees [0] 17/8/24->16->0 [1] -1/-1/-1->16->23 [2] 17/-1/-1->16->23 [3] 17/-1/-1->16->23 [4] 17/-1/-1->16->23 [5] 17/-1/-1->16->23 [6] 17/-1/-1->16->23 [7] 17/-1/-1->16->23 [8] 17/-1/-1->16->8 [9] -1/-1/-1->16->23 [10] 17/-1/-1->16->23 [11] 17/-1/-1->16->23 [12] 17/-1/-1->16->23 [13] 17/-1/-1->16->23 [14] 17/-1/-1->16->23 [15] 17/-1/-1->16->23
+ip-26-0-161-142:62882:63374 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-178:2815968:2816472 [5] NCCL INFO Trees [0] 30/-1/-1->29->28 [1] 30/-1/-1->29->28 [2] 30/-1/-1->29->28 [3] 30/-1/-1->29->28 [4] 30/-1/-1->29->28 [5] 30/-1/-1->29->21 [6] -1/-1/-1->29->28 [7] 30/-1/-1->29->28 [8] 30/-1/-1->29->28 [9] 30/-1/-1->29->28 [10] 30/-1/-1->29->28 [11] 30/-1/-1->29->28 [12] 30/-1/-1->29->28 [13] 30/13/-1->29->-1 [14] -1/-1/-1->29->28 [15] 30/-1/-1->29->28
+ip-26-0-161-178:2815968:2816472 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-178:2815969:2816471 [6] NCCL INFO Trees [0] 31/-1/-1->30->29 [1] 31/-1/-1->30->29 [2] 31/-1/-1->30->29 [3] 31/-1/-1->30->29 [4] 31/-1/-1->30->29 [5] 31/-1/-1->30->29 [6] 31/-1/-1->30->22 [7] -1/-1/-1->30->29 [8] 31/-1/-1->30->29 [9] 31/-1/-1->30->29 [10] 31/-1/-1->30->29 [11] 31/-1/-1->30->29 [12] 31/-1/-1->30->29 [13] 31/-1/-1->30->29 [14] 31/14/-1->30->-1 [15] -1/-1/-1->30->29
+ip-26-0-161-178:2815969:2816471 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-153:2636084:2636605 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-153:2636090:2636598 [6] NCCL INFO Trees [0] 23/-1/-1->22->21 [1] 23/-1/-1->22->21 [2] 23/-1/-1->22->21 [3] 23/-1/-1->22->21 [4] 23/-1/-1->22->21 [5] 23/-1/-1->22->21 [6] 23/14/30->22->6 [7] -1/-1/-1->22->21 [8] 23/-1/-1->22->21 [9] 23/-1/-1->22->21 [10] 23/-1/-1->22->21 [11] 23/-1/-1->22->21 [12] 23/-1/-1->22->21 [13] 23/-1/-1->22->21 [14] 23/-1/-1->22->14 [15] -1/-1/-1->22->21
+ip-26-0-161-153:2636089:2636600 [5] NCCL INFO Trees [0] 22/-1/-1->21->20 [1] 22/-1/-1->21->20 [2] 22/-1/-1->21->20 [3] 22/-1/-1->21->20 [4] 22/-1/-1->21->20 [5] 22/13/29->21->5 [6] -1/-1/-1->21->20 [7] 22/-1/-1->21->20 [8] 22/-1/-1->21->20 [9] 22/-1/-1->21->20 [10] 22/-1/-1->21->20 [11] 22/-1/-1->21->20 [12] 22/-1/-1->21->20 [13] 22/-1/-1->21->13 [14] -1/-1/-1->21->20 [15] 22/-1/-1->21->20
+ip-26-0-161-153:2636090:2636598 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-153:2636089:2636600 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-142:62881:63375 [0] NCCL INFO Trees [0] 9/-1/-1->8->16 [1] -1/-1/-1->8->15 [2] 9/-1/-1->8->15 [3] 9/-1/-1->8->15 [4] 9/-1/-1->8->15 [5] 9/-1/-1->8->15 [6] 9/-1/-1->8->15 [7] 9/-1/-1->8->15 [8] 9/16/0->8->24 [9] -1/-1/-1->8->15 [10] 9/-1/-1->8->15 [11] 9/-1/-1->8->15 [12] 9/-1/-1->8->15 [13] 9/-1/-1->8->15 [14] 9/-1/-1->8->15 [15] 9/-1/-1->8->15
+ip-26-0-161-178:2815967:2816473 [4] NCCL INFO Trees [0] 29/-1/-1->28->27 [1] 29/-1/-1->28->27 [2] 29/-1/-1->28->27 [3] 29/-1/-1->28->27 [4] 29/-1/-1->28->20 [5] -1/-1/-1->28->27 [6] 29/-1/-1->28->27 [7] 29/-1/-1->28->27 [8] 29/-1/-1->28->27 [9] 29/-1/-1->28->27 [10] 29/-1/-1->28->27 [11] 29/-1/-1->28->27 [12] 29/12/-1->28->-1 [13] -1/-1/-1->28->27 [14] 29/-1/-1->28->27 [15] 29/-1/-1->28->27
+ip-26-0-161-178:2815967:2816473 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-178:2815966:2816474 [3] NCCL INFO Trees [0] 28/-1/-1->27->26 [1] 28/-1/-1->27->26 [2] 28/-1/-1->27->26 [3] 28/-1/-1->27->19 [4] -1/-1/-1->27->26 [5] 28/-1/-1->27->26 [6] 28/-1/-1->27->26 [7] 28/-1/-1->27->26 [8] 28/-1/-1->27->26 [9] 28/-1/-1->27->26 [10] 28/-1/-1->27->26 [11] 28/11/-1->27->-1 [12] -1/-1/-1->27->26 [13] 28/-1/-1->27->26 [14] 28/-1/-1->27->26 [15] 28/-1/-1->27->26
+ip-26-0-161-153:2636088:2636601 [4] NCCL INFO Trees [0] 21/-1/-1->20->19 [1] 21/-1/-1->20->19 [2] 21/-1/-1->20->19 [3] 21/-1/-1->20->19 [4] 21/12/28->20->4 [5] -1/-1/-1->20->19 [6] 21/-1/-1->20->19 [7] 21/-1/-1->20->19 [8] 21/-1/-1->20->19 [9] 21/-1/-1->20->19 [10] 21/-1/-1->20->19 [11] 21/-1/-1->20->19 [12] 21/-1/-1->20->12 [13] -1/-1/-1->20->19 [14] 21/-1/-1->20->19 [15] 21/-1/-1->20->19
+ip-26-0-161-153:2636088:2636601 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-153:2636091:2636599 [7] NCCL INFO Trees [0] -1/-1/-1->23->22 [1] 16/-1/-1->23->22 [2] 16/-1/-1->23->22 [3] 16/-1/-1->23->22 [4] 16/-1/-1->23->22 [5] 16/-1/-1->23->22 [6] 16/-1/-1->23->22 [7] 16/15/31->23->7 [8] -1/-1/-1->23->22 [9] 16/-1/-1->23->22 [10] 16/-1/-1->23->22 [11] 16/-1/-1->23->22 [12] 16/-1/-1->23->22 [13] 16/-1/-1->23->22 [14] 16/-1/-1->23->22 [15] 16/-1/-1->23->15
+ip-26-0-161-138:96567:97090 [4] NCCL INFO NVLS Head  5:  5 13 21 29
+ip-26-0-161-138:96566:97087 [3] NCCL INFO NVLS Head  3:  3 11 19 27
+ip-26-0-161-138:96570:97093 [7] NCCL INFO NVLS Head  1:  1  9 17 25
+ip-26-0-161-138:96568:97091 [5] NCCL INFO NVLS Head  5:  5 13 21 29
+ip-26-0-161-138:96570:97093 [7] NCCL INFO NVLS Head  2:  2 10 18 26
+ip-26-0-161-138:96565:97089 [2] NCCL INFO NVLS Head  0:  0  8 16 24
+ip-26-0-161-138:96567:97090 [4] NCCL INFO NVLS Head  6:  6 14 22 30
+ip-26-0-161-138:96566:97087 [3] NCCL INFO NVLS Head  4:  4 12 20 28
+ip-26-0-161-138:96568:97091 [5] NCCL INFO NVLS Head  6:  6 14 22 30
+ip-26-0-161-138:96565:97089 [2] NCCL INFO NVLS Head  1:  1  9 17 25
+ip-26-0-161-138:96566:97087 [3] NCCL INFO NVLS Head  5:  5 13 21 29
+ip-26-0-161-138:96568:97091 [5] NCCL INFO NVLS Head  7:  7 15 23 31
+ip-26-0-161-138:96570:97093 [7] NCCL INFO NVLS Head  3:  3 11 19 27
+ip-26-0-161-138:96567:97090 [4] NCCL INFO NVLS Head  7:  7 15 23 31
+ip-26-0-161-138:96565:97089 [2] NCCL INFO NVLS Head  2:  2 10 18 26
+ip-26-0-161-142:62881:63375 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-178:2815965:2816475 [2] NCCL INFO Trees [0] 27/-1/-1->26->25 [1] 27/-1/-1->26->25 [2] 27/-1/-1->26->18 [3] -1/-1/-1->26->25 [4] 27/-1/-1->26->25 [5] 27/-1/-1->26->25 [6] 27/-1/-1->26->25 [7] 27/-1/-1->26->25 [8] 27/-1/-1->26->25 [9] 27/-1/-1->26->25 [10] 27/10/-1->26->-1 [11] -1/-1/-1->26->25 [12] 27/-1/-1->26->25 [13] 27/-1/-1->26->25 [14] 27/-1/-1->26->25 [15] 27/-1/-1->26->25
+ip-26-0-161-178:2815964:2816476 [1] NCCL INFO Trees [0] 26/-1/-1->25->24 [1] 26/-1/-1->25->17 [2] -1/-1/-1->25->24 [3] 26/-1/-1->25->24 [4] 26/-1/-1->25->24 [5] 26/-1/-1->25->24 [6] 26/-1/-1->25->24 [7] 26/-1/-1->25->24 [8] 26/-1/-1->25->24 [9] 26/9/-1->25->-1 [10] -1/-1/-1->25->24 [11] 26/-1/-1->25->24 [12] 26/-1/-1->25->24 [13] 26/-1/-1->25->24 [14] 26/-1/-1->25->24 [15] 26/-1/-1->25->24
+ip-26-0-161-123:32705:33203 [1] NCCL INFO Channel 04/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-161-153:2636091:2636599 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-138:96566:97087 [3] NCCL INFO NVLS Head  6:  6 14 22 30
+ip-26-0-165-202:153566:154062 [4] NCCL INFO Channel 01/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-161-178:2815966:2816474 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-178:2815965:2816475 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-178:2815964:2816476 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-178:2815963:2816477 [0] NCCL INFO Trees [0] 25/-1/-1->24->16 [1] -1/-1/-1->24->31 [2] 25/-1/-1->24->31 [3] 25/-1/-1->24->31 [4] 25/-1/-1->24->31 [5] 25/-1/-1->24->31 [6] 25/-1/-1->24->31 [7] 25/-1/-1->24->31 [8] 25/8/-1->24->-1 [9] -1/-1/-1->24->31 [10] 25/-1/-1->24->31 [11] 25/-1/-1->24->31 [12] 25/-1/-1->24->31 [13] 25/-1/-1->24->31 [14] 25/-1/-1->24->31 [15] 25/-1/-1->24->31
+ip-26-0-161-138:96570:97093 [7] NCCL INFO NVLS Head  4:  4 12 20 28
+ip-26-0-161-138:96565:97089 [2] NCCL INFO NVLS Head  3:  3 11 19 27
+ip-26-0-161-138:96566:97087 [3] NCCL INFO NVLS Head  7:  7 15 23 31
+ip-26-0-161-138:96563:97086 [0] NCCL INFO NVLS Head  0:  0  8 16 24
+ip-26-0-161-138:96565:97089 [2] NCCL INFO NVLS Head  4:  4 12 20 28
+ip-26-0-161-138:96570:97093 [7] NCCL INFO NVLS Head  5:  5 13 21 29
+ip-26-0-161-138:96565:97089 [2] NCCL INFO NVLS Head  5:  5 13 21 29
+ip-26-0-161-138:96563:97086 [0] NCCL INFO NVLS Head  1:  1  9 17 25
+ip-26-0-161-138:96570:97093 [7] NCCL INFO NVLS Head  6:  6 14 22 30
+ip-26-0-161-138:96565:97089 [2] NCCL INFO NVLS Head  6:  6 14 22 30
+ip-26-0-161-138:96563:97086 [0] NCCL INFO NVLS Head  2:  2 10 18 26
+ip-26-0-161-138:96570:97093 [7] NCCL INFO NVLS Head  7:  7 15 23 31
+ip-26-0-161-138:96565:97089 [2] NCCL INFO NVLS Head  7:  7 15 23 31
+ip-26-0-161-138:96563:97086 [0] NCCL INFO NVLS Head  3:  3 11 19 27
+ip-26-0-161-178:2815963:2816477 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-138:96568:97091 [5] NCCL INFO Trees [0] 6/-1/-1->5->4 [1] 6/-1/-1->5->4 [2] 6/-1/-1->5->4 [3] 6/-1/-1->5->4 [4] 6/-1/-1->5->4 [5] 6/21/-1->5->-1 [6] -1/-1/-1->5->4 [7] 6/-1/-1->5->4 [8] 6/-1/-1->5->4 [9] 6/-1/-1->5->4 [10] 6/-1/-1->5->4 [11] 6/-1/-1->5->4 [12] 6/-1/-1->5->4 [13] 6/-1/-1->5->13 [14] -1/-1/-1->5->4 [15] 6/-1/-1->5->4
+ip-26-0-161-138:96567:97090 [4] NCCL INFO Trees [0] 5/-1/-1->4->3 [1] 5/-1/-1->4->3 [2] 5/-1/-1->4->3 [3] 5/-1/-1->4->3 [4] 5/20/-1->4->-1 [5] -1/-1/-1->4->3 [6] 5/-1/-1->4->3 [7] 5/-1/-1->4->3 [8] 5/-1/-1->4->3 [9] 5/-1/-1->4->3 [10] 5/-1/-1->4->3 [11] 5/-1/-1->4->3 [12] 5/-1/-1->4->12 [13] -1/-1/-1->4->3 [14] 5/-1/-1->4->3 [15] 5/-1/-1->4->3
+ip-26-0-161-138:96563:97086 [0] NCCL INFO NVLS Head  4:  4 12 20 28
+ip-26-0-161-138:96568:97091 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-138:96567:97090 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-138:96563:97086 [0] NCCL INFO NVLS Head  5:  5 13 21 29
+ip-26-0-161-138:96563:97086 [0] NCCL INFO NVLS Head  6:  6 14 22 30
+ip-26-0-161-138:96563:97086 [0] NCCL INFO NVLS Head  7:  7 15 23 31
+ip-26-0-161-138:96566:97087 [3] NCCL INFO Trees [0] 4/-1/-1->3->2 [1] 4/-1/-1->3->2 [2] 4/-1/-1->3->2 [3] 4/19/-1->3->-1 [4] -1/-1/-1->3->2 [5] 4/-1/-1->3->2 [6] 4/-1/-1->3->2 [7] 4/-1/-1->3->2 [8] 4/-1/-1->3->2 [9] 4/-1/-1->3->2 [10] 4/-1/-1->3->2 [11] 4/-1/-1->3->11 [12] -1/-1/-1->3->2 [13] 4/-1/-1->3->2 [14] 4/-1/-1->3->2 [15] 4/-1/-1->3->2
+ip-26-0-161-138:96566:97087 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-138:96563:97086 [0] NCCL INFO Channel 00/16 :    0   1   2   3   4   5   6   7   8   9  10  11  12  13  14  15  16  17  18  19
+ip-26-0-161-138:96570:97093 [7] NCCL INFO Trees [0] -1/-1/-1->7->6 [1] 0/-1/-1->7->6 [2] 0/-1/-1->7->6 [3] 0/-1/-1->7->6 [4] 0/-1/-1->7->6 [5] 0/-1/-1->7->6 [6] 0/-1/-1->7->6 [7] 0/23/-1->7->-1 [8] -1/-1/-1->7->6 [9] 0/-1/-1->7->6 [10] 0/-1/-1->7->6 [11] 0/-1/-1->7->6 [12] 0/-1/-1->7->6 [13] 0/-1/-1->7->6 [14] 0/-1/-1->7->6 [15] 0/-1/-1->7->15
+ip-26-0-161-138:96565:97089 [2] NCCL INFO Trees [0] 3/-1/-1->2->1 [1] 3/-1/-1->2->1 [2] 3/18/-1->2->-1 [3] -1/-1/-1->2->1 [4] 3/-1/-1->2->1 [5] 3/-1/-1->2->1 [6] 3/-1/-1->2->1 [7] 3/-1/-1->2->1 [8] 3/-1/-1->2->1 [9] 3/-1/-1->2->1 [10] 3/-1/-1->2->10 [11] -1/-1/-1->2->1 [12] 3/-1/-1->2->1 [13] 3/-1/-1->2->1 [14] 3/-1/-1->2->1 [15] 3/-1/-1->2->1
+ip-26-0-161-138:96563:97086 [0] NCCL INFO Channel 01/16 :    0   7   6   5   4   3   2   9   8  15  14  13  12  11  10  17  16  23  22  21
+ip-26-0-161-138:96570:97093 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-138:96565:97089 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-138:96563:97086 [0] NCCL INFO Channel 02/16 :    0   7   6   5   4   3  10   9   8  15  14  13  12  11  18  17  16  23  22  21
+ip-26-0-161-138:96563:97086 [0] NCCL INFO Channel 03/16 :    0   7   6   5   4  11  10   9   8  15  14  13  12  19  18  17  16  23  22  21
+ip-26-0-161-138:96563:97086 [0] NCCL INFO Channel 04/16 :    0   7   6   5  12  11  10   9   8  15  14  13  20  19  18  17  16  23  22  21
+ip-26-0-161-138:96563:97086 [0] NCCL INFO Channel 05/16 :    0   4   7   6  13  11  10   9   8  12  15  14  21  19  18  17  16  20  23  22
+ip-26-0-161-138:96563:97086 [0] NCCL INFO Channel 06/16 :    0   5   4   7  14  11  10   9   8  13  12  15  22  19  18  17  16  21  20  23
+ip-26-0-161-138:96563:97086 [0] NCCL INFO Channel 07/16 :    0   6   5   4  15  11  10   9   8  14  13  12  23  19  18  17  16  22  21  20
+ip-26-0-161-138:96563:97086 [0] NCCL INFO Channel 08/16 :    0   1   2   3   4   5   6   7   8   9  10  11  12  13  14  15  16  17  18  19
+ip-26-0-161-138:96569:97092 [6] NCCL INFO NVLS Head  0:  0  8 16 24
+ip-26-0-161-138:96563:97086 [0] NCCL INFO Channel 09/16 :    0   7   6   5   4   3   2   9   8  15  14  13  12  11  10  17  16  23  22  21
+ip-26-0-161-138:96569:97092 [6] NCCL INFO NVLS Head  1:  1  9 17 25
+ip-26-0-160-225:124162:124660 [2] NCCL INFO Channel 12/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-138:96563:97086 [0] NCCL INFO Channel 10/16 :    0   7   6   5   4   3  10   9   8  15  14  13  12  11  18  17  16  23  22  21
+ip-26-0-161-138:96569:97092 [6] NCCL INFO NVLS Head  2:  2 10 18 26
+ip-26-0-161-138:96569:97092 [6] NCCL INFO NVLS Head  3:  3 11 19 27
+ip-26-0-161-138:96563:97086 [0] NCCL INFO Channel 11/16 :    0   7   6   5   4  11  10   9   8  15  14  13  12  19  18  17  16  23  22  21
+ip-26-0-161-78:114516:115010 [4] NCCL INFO Channel 12/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-138:96569:97092 [6] NCCL INFO NVLS Head  4:  4 12 20 28
+ip-26-0-161-138:96569:97092 [6] NCCL INFO NVLS Head  5:  5 13 21 29
+ip-26-0-161-138:96563:97086 [0] NCCL INFO Channel 12/16 :    0   7   6   5  12  11  10   9   8  15  14  13  20  19  18  17  16  23  22  21
+ip-26-0-161-138:96569:97092 [6] NCCL INFO NVLS Head  6:  6 14 22 30
+ip-26-0-161-138:96569:97092 [6] NCCL INFO NVLS Head  7:  7 15 23 31
+ip-26-0-161-138:96563:97086 [0] NCCL INFO Channel 13/16 :    0   4   7   6  13  11  10   9   8  12  15  14  21  19  18  17  16  20  23  22
+ip-26-0-161-138:96563:97086 [0] NCCL INFO Channel 14/16 :    0   5   4   7  14  11  10   9   8  13  12  15  22  19  18  17  16  21  20  23
+ip-26-0-161-138:96563:97086 [0] NCCL INFO Channel 15/16 :    0   6   5   4  15  11  10   9   8  14  13  12  23  19  18  17  16  22  21  20
+ip-26-0-161-138:96563:97086 [0] NCCL INFO Trees [0] 1/16/-1->0->-1 [1] -1/-1/-1->0->7 [2] 1/-1/-1->0->7 [3] 1/-1/-1->0->7 [4] 1/-1/-1->0->7 [5] 1/-1/-1->0->7 [6] 1/-1/-1->0->7 [7] 1/-1/-1->0->7 [8] 1/-1/-1->0->8 [9] -1/-1/-1->0->7 [10] 1/-1/-1->0->7 [11] 1/-1/-1->0->7 [12] 1/-1/-1->0->7 [13] 1/-1/-1->0->7 [14] 1/-1/-1->0->7 [15] 1/-1/-1->0->7
+ip-26-0-161-138:96563:97086 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-138:96569:97092 [6] NCCL INFO Trees [0] 7/-1/-1->6->5 [1] 7/-1/-1->6->5 [2] 7/-1/-1->6->5 [3] 7/-1/-1->6->5 [4] 7/-1/-1->6->5 [5] 7/-1/-1->6->5 [6] 7/22/-1->6->-1 [7] -1/-1/-1->6->5 [8] 7/-1/-1->6->5 [9] 7/-1/-1->6->5 [10] 7/-1/-1->6->5 [11] 7/-1/-1->6->5 [12] 7/-1/-1->6->5 [13] 7/-1/-1->6->5 [14] 7/-1/-1->6->14 [15] -1/-1/-1->6->5
+ip-26-0-161-138:96569:97092 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-202:153569:154055 [7] NCCL INFO Channel 12/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-161-138:96564:97088 [1] NCCL INFO NVLS Head  0:  0  8 16 24
+ip-26-0-161-138:96564:97088 [1] NCCL INFO NVLS Head  1:  1  9 17 25
+ip-26-0-161-138:96564:97088 [1] NCCL INFO NVLS Head  2:  2 10 18 26
+ip-26-0-161-138:96564:97088 [1] NCCL INFO NVLS Head  3:  3 11 19 27
+ip-26-0-161-138:96564:97088 [1] NCCL INFO NVLS Head  4:  4 12 20 28
+ip-26-0-161-138:96564:97088 [1] NCCL INFO NVLS Head  5:  5 13 21 29
+ip-26-0-161-138:96564:97088 [1] NCCL INFO NVLS Head  6:  6 14 22 30
+ip-26-0-161-138:96564:97088 [1] NCCL INFO NVLS Head  7:  7 15 23 31
+ip-26-0-168-34:272154:272650 [4] NCCL INFO Channel 07/0 : 20[4] -> 31[7] [send] via NET/Libfabric/7(23)/GDRDMA
+ip-26-0-161-138:96564:97088 [1] NCCL INFO Trees [0] 2/-1/-1->1->0 [1] 2/17/-1->1->-1 [2] -1/-1/-1->1->0 [3] 2/-1/-1->1->0 [4] 2/-1/-1->1->0 [5] 2/-1/-1->1->0 [6] 2/-1/-1->1->0 [7] 2/-1/-1->1->0 [8] 2/-1/-1->1->0 [9] 2/-1/-1->1->9 [10] -1/-1/-1->1->0 [11] 2/-1/-1->1->0 [12] 2/-1/-1->1->0 [13] 2/-1/-1->1->0 [14] 2/-1/-1->1->0 [15] 2/-1/-1->1->0
+ip-26-0-161-138:96564:97088 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-168-34:272154:272650 [4] NCCL INFO Channel 15/0 : 20[4] -> 31[7] [send] via NET/Libfabric/7(23)/GDRDMA
+ip-26-0-165-164:1227470:1227958 [7] NCCL INFO Channel 12/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-161-78:114515:115009 [3] NCCL INFO Channel 01/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-225:124161:124659 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-123:32706:33199 [2] NCCL INFO Channel 05/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-160-225:124162:124660 [2] NCCL INFO Channel 13/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-123:32705:33203 [1] NCCL INFO Channel 05/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-165-131:1630212:1630734 [7] NCCL INFO Channel 11/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-161-78:114515:115009 [3] NCCL INFO Channel 03/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-165-202:153566:154062 [4] NCCL INFO Channel 02/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-160-225:124161:124659 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-225:124162:124660 [2] NCCL INFO Channel 14/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-165-164:1227470:1227958 [7] NCCL INFO Channel 13/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-165-202:153569:154055 [7] NCCL INFO Channel 13/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-161-123:32706:33199 [2] NCCL INFO Channel 06/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-161-123:32705:33203 [1] NCCL INFO Channel 06/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-161-78:114515:115009 [3] NCCL INFO Channel 04/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-103:96889:97388 [1] NCCL INFO Channel 01/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-160-225:124161:124659 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-225:124162:124660 [2] NCCL INFO Channel 15/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-242:1364903:1365409 [1] NCCL INFO Channel 01/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-161-78:114515:115009 [3] NCCL INFO Channel 05/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-103:96889:97388 [1] NCCL INFO Channel 02/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-165-202:153566:154062 [4] NCCL INFO Channel 04/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-165-164:1227467:1227960 [4] NCCL INFO Channel 01/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-161-123:32706:33199 [2] NCCL INFO Channel 07/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-161-123:32705:33203 [1] NCCL INFO Channel 07/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-160-225:124161:124659 [1] NCCL INFO Channel 04/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-78:114515:115009 [3] NCCL INFO Channel 06/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-103:96889:97388 [1] NCCL INFO Channel 03/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-160-242:1364903:1365409 [1] NCCL INFO Channel 02/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-165-164:1227467:1227960 [4] NCCL INFO Channel 02/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-160-242:1364904:1365408 [2] NCCL INFO Channel 02/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-160-225:124161:124659 [1] NCCL INFO Channel 05/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-202:153566:154062 [4] NCCL INFO Channel 09/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-165-164:1227467:1227960 [4] NCCL INFO Channel 04/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-161-123:32705:33203 [1] NCCL INFO Channel 09/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-161-123:32706:33199 [2] NCCL INFO Channel 10/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-165-131:1630212:1630734 [7] NCCL INFO Channel 12/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-161-78:114517:115013 [5] NCCL INFO Channel 01/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-103:96889:97388 [1] NCCL INFO Channel 04/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-161-78:114515:115009 [3] NCCL INFO Channel 07/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-165-164:1227466:1227961 [3] NCCL INFO Channel 01/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-165-202:153566:154062 [4] NCCL INFO Channel 10/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-160-242:1364903:1365409 [1] NCCL INFO Channel 03/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-160-242:1364904:1365408 [2] NCCL INFO Channel 03/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-160-225:124161:124659 [1] NCCL INFO Channel 06/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-164:1227467:1227960 [4] NCCL INFO Channel 09/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-161-103:96889:97388 [1] NCCL INFO Channel 05/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-161-103:96890:97386 [2] NCCL INFO Channel 02/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-165-202:153566:154062 [4] NCCL INFO Channel 12/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-161-123:32705:33203 [1] NCCL INFO Channel 10/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-161-123:32706:33199 [2] NCCL INFO Channel 11/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-160-225:124161:124659 [1] NCCL INFO Channel 07/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-242:1364903:1365409 [1] NCCL INFO Channel 04/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-160-242:1364904:1365408 [2] NCCL INFO Channel 04/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-165-164:1227466:1227961 [3] NCCL INFO Channel 03/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-161-103:96889:97388 [1] NCCL INFO Channel 06/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-160-225:124161:124659 [1] NCCL INFO Channel 09/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-164:1227467:1227960 [4] NCCL INFO Channel 10/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-161-123:32705:33203 [1] NCCL INFO Channel 11/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-165-131:1630212:1630734 [7] NCCL INFO Channel 13/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-165-202:153565:154058 [3] NCCL INFO Channel 01/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-165-202:153568:154056 [6] NCCL INFO Channel 01/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-161-123:32706:33199 [2] NCCL INFO Channel 12/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-161-103:96889:97388 [1] NCCL INFO Channel 07/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-160-242:1364903:1365409 [1] NCCL INFO Channel 05/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-160-242:1364904:1365408 [2] NCCL INFO Channel 05/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-160-225:124161:124659 [1] NCCL INFO Channel 10/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-78:114518:115012 [6] NCCL INFO Channel 01/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-103:96890:97386 [2] NCCL INFO Channel 03/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-165-164:1227466:1227961 [3] NCCL INFO Channel 04/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-161-103:96889:97388 [1] NCCL INFO Channel 09/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-161-78:114518:115012 [6] NCCL INFO Channel 02/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-225:124161:124659 [1] NCCL INFO Channel 11/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-242:1364903:1365409 [1] NCCL INFO Channel 06/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-160-242:1364904:1365408 [2] NCCL INFO Channel 06/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-165-202:153568:154056 [6] NCCL INFO Channel 02/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-161-123:32705:33203 [1] NCCL INFO Channel 12/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-161-123:32706:33199 [2] NCCL INFO Channel 13/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-161-78:114517:115013 [5] NCCL INFO Channel 02/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-202:153565:154058 [3] NCCL INFO Channel 03/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-165-164:1227467:1227960 [4] NCCL INFO Channel 12/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-161-103:96890:97386 [2] NCCL INFO Channel 04/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-161-103:96889:97388 [1] NCCL INFO Channel 10/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-161-178:2815968:2816472 [5] NCCL INFO Channel 00/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-161-78:114515:115009 [3] NCCL INFO Channel 09/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-165-131:1630208:1630729 [3] NCCL INFO Channel 01/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-160-225:124161:124659 [1] NCCL INFO Channel 12/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-131:1630209:1630730 [4] NCCL INFO Channel 01/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-165-164:1227469:1227957 [6] NCCL INFO Channel 01/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-161-78:114518:115012 [6] NCCL INFO Channel 03/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-242:1364903:1365409 [1] NCCL INFO Channel 07/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-161-138:96568:97091 [5] NCCL INFO Channel 00/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-165-164:1227466:1227961 [3] NCCL INFO Channel 05/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-161-78:114517:115013 [5] NCCL INFO Channel 03/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-242:1364904:1365408 [2] NCCL INFO Channel 07/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-160-225:124161:124659 [1] NCCL INFO Channel 13/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-202:153568:154056 [6] NCCL INFO Channel 03/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-161-123:32705:33203 [1] NCCL INFO Channel 13/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-165-202:153565:154058 [3] NCCL INFO Channel 04/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-161-123:32706:33199 [2] NCCL INFO Channel 14/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-161-142:62886:63378 [5] NCCL INFO Channel 00/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-161-78:114515:115009 [3] NCCL INFO Channel 11/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-225:124161:124659 [1] NCCL INFO Channel 14/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-138:96567:97090 [4] NCCL INFO Channel 00/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-242:1364903:1365409 [1] NCCL INFO Channel 09/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-165-164:1227469:1227957 [6] NCCL INFO Channel 02/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-161-138:96565:97089 [2] NCCL INFO Channel 00/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-242:1364904:1365408 [2] NCCL INFO Channel 10/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-165-164:1227466:1227961 [3] NCCL INFO Channel 06/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-165-131:1630209:1630730 [4] NCCL INFO Channel 02/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-161-78:114518:115012 [6] NCCL INFO Channel 04/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-225:124161:124659 [1] NCCL INFO Channel 15/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-78:114517:115013 [5] NCCL INFO Channel 06/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-123:32705:33203 [1] NCCL INFO Channel 14/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-161-123:32706:33199 [2] NCCL INFO Channel 15/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-165-202:153568:154056 [6] NCCL INFO Channel 04/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-165-202:153565:154058 [3] NCCL INFO Channel 05/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-161-178:2815968:2816472 [5] NCCL INFO Channel 08/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-161-153:2636089:2636600 [5] NCCL INFO Channel 00/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-161-142:62885:63377 [4] NCCL INFO Channel 00/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-161-78:114515:115009 [3] NCCL INFO Channel 12/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-103:96890:97386 [2] NCCL INFO Channel 05/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-161-142:62887:63379 [6] NCCL INFO Channel 00/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-161-138:96566:97087 [3] NCCL INFO Channel 00/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-103:96889:97388 [1] NCCL INFO Channel 11/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-165-164:1227469:1227957 [6] NCCL INFO Channel 03/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-161-142:62883:63373 [2] NCCL INFO Channel 00/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-160-242:1364903:1365409 [1] NCCL INFO Channel 10/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-165-131:1630209:1630730 [4] NCCL INFO Channel 04/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-165-164:1227466:1227961 [3] NCCL INFO Channel 07/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-160-242:1364904:1365408 [2] NCCL INFO Channel 11/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-161-123:32705:33203 [1] NCCL INFO Channel 15/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-161-138:96568:97091 [5] NCCL INFO Channel 08/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-142:62884:63376 [3] NCCL INFO Channel 00/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-161-142:62882:63374 [1] NCCL INFO Channel 00/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-161-142:62886:63378 [5] NCCL INFO Channel 08/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-161-78:114518:115012 [6] NCCL INFO Channel 07/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-153:2636090:2636598 [6] NCCL INFO Channel 00/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-161-178:2815965:2816475 [2] NCCL INFO Channel 00/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-165-202:153568:154056 [6] NCCL INFO Channel 07/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-165-202:153565:154058 [3] NCCL INFO Channel 06/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-161-78:114517:115013 [5] NCCL INFO Channel 07/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-202:153567:154059 [5] NCCL INFO Channel 01/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-161-178:2815967:2816473 [4] NCCL INFO Channel 00/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-161-178:2815969:2816471 [6] NCCL INFO Channel 00/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-161-78:114515:115009 [3] NCCL INFO Channel 13/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-103:96890:97386 [2] NCCL INFO Channel 06/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-161-178:2815966:2816474 [3] NCCL INFO Channel 00/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-161-103:96889:97388 [1] NCCL INFO Channel 12/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-165-164:1227468:1227959 [5] NCCL INFO Channel 01/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-161-178:2815964:2816476 [1] NCCL INFO Channel 00/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-165-164:1227469:1227957 [6] NCCL INFO Channel 04/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-161-138:96567:97090 [4] NCCL INFO Channel 08/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-242:1364903:1365409 [1] NCCL INFO Channel 11/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-165-131:1630209:1630730 [4] NCCL INFO Channel 09/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-161-138:96569:97092 [6] NCCL INFO Channel 00/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-242:1364904:1365408 [2] NCCL INFO Channel 12/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-165-164:1227466:1227961 [3] NCCL INFO Channel 09/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-161-138:96565:97089 [2] NCCL INFO Channel 08/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-138:96564:97088 [1] NCCL INFO Channel 00/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-131:1630208:1630729 [3] NCCL INFO Channel 03/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-161-103:96890:97386 [2] NCCL INFO Channel 07/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-161-103:96889:97388 [1] NCCL INFO Channel 13/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-161-78:114517:115013 [5] NCCL INFO Channel 09/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-202:153568:154056 [6] NCCL INFO Channel 09/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-165-202:153565:154058 [3] NCCL INFO Channel 07/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-161-78:114515:115009 [3] NCCL INFO Channel 14/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-165-59:1557266:1557759 [0] NCCL INFO Channel 06/0 : 0[0] -> 5[5] via P2P/IPC
+ip-26-0-165-202:153567:154059 [5] NCCL INFO Channel 02/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-165-131:1630211:1630733 [6] NCCL INFO Channel 01/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-165-164:1227468:1227959 [5] NCCL INFO Channel 02/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-165-164:1227469:1227957 [6] NCCL INFO Channel 07/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-160-242:1364903:1365409 [1] NCCL INFO Channel 12/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-161-153:2636086:2636602 [2] NCCL INFO Channel 00/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-161-153:2636089:2636600 [5] NCCL INFO Channel 08/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-165-164:1227466:1227961 [3] NCCL INFO Channel 11/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-161-138:96566:97087 [3] NCCL INFO Channel 08/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-242:1364904:1365408 [2] NCCL INFO Channel 13/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-161-142:62885:63377 [4] NCCL INFO Channel 08/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-161-142:62887:63379 [6] NCCL INFO Channel 08/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-161-103:96890:97386 [2] NCCL INFO Channel 10/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-161-142:62883:63373 [2] NCCL INFO Channel 08/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-161-142:62882:63374 [1] NCCL INFO Channel 08/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-161-78:114517:115013 [5] NCCL INFO Channel 10/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-103:96889:97388 [1] NCCL INFO Channel 14/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-161-178:2815965:2816475 [2] NCCL INFO Channel 08/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-161-178:2815967:2816473 [4] NCCL INFO Channel 08/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-161-78:114515:115009 [3] NCCL INFO Channel 15/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-142:62884:63376 [3] NCCL INFO Channel 08/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-161-178:2815969:2816471 [6] NCCL INFO Channel 08/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-161-178:2815966:2816474 [3] NCCL INFO Channel 08/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-161-178:2815964:2816476 [1] NCCL INFO Channel 08/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-161-78:114518:115012 [6] NCCL INFO Channel 09/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-202:153568:154056 [6] NCCL INFO Channel 10/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-160-242:1364903:1365409 [1] NCCL INFO Channel 13/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-165-202:153565:154058 [3] NCCL INFO Channel 09/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-165-202:153567:154059 [5] NCCL INFO Channel 03/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-160-242:1364904:1365408 [2] NCCL INFO Channel 14/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-161-153:2636090:2636598 [6] NCCL INFO Channel 08/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-161-138:96569:97092 [6] NCCL INFO Channel 08/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-103:96890:97386 [2] NCCL INFO Channel 11/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-165-164:1227469:1227957 [6] NCCL INFO Channel 09/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-165-131:1630209:1630730 [4] NCCL INFO Channel 10/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-165-131:1630208:1630729 [3] NCCL INFO Channel 04/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-161-138:96564:97088 [1] NCCL INFO Channel 08/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-103:96889:97388 [1] NCCL INFO Channel 15/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-165-164:1227468:1227959 [5] NCCL INFO Channel 03/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-165-164:1227466:1227961 [3] NCCL INFO Channel 12/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-165-131:1630211:1630733 [6] NCCL INFO Channel 02/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-161-153:2636087:2636603 [3] NCCL INFO Channel 00/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-161-138:96567:97090 [4] NCCL INFO Channel 05/0 : 4[4] -> 7[7] via P2P/IPC
+ip-26-0-160-242:1364903:1365409 [1] NCCL INFO Channel 14/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-161-78:114517:115013 [5] NCCL INFO Channel 11/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-242:1364904:1365408 [2] NCCL INFO Channel 15/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-161-78:114518:115012 [6] NCCL INFO Channel 10/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-202:153568:154056 [6] NCCL INFO Channel 11/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-165-202:153565:154058 [3] NCCL INFO Channel 11/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-161-103:96890:97386 [2] NCCL INFO Channel 12/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-165-202:153567:154059 [5] NCCL INFO Channel 06/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-161-153:2636085:2636604 [1] NCCL INFO Channel 00/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-161-153:2636088:2636601 [4] NCCL INFO Channel 00/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-161-138:96567:97090 [4] NCCL INFO Channel 06/0 : 4[4] -> 7[7] via P2P/IPC
+ip-26-0-161-142:62885:63377 [4] NCCL INFO Channel 05/0 : 12[4] -> 15[7] via P2P/IPC
+ip-26-0-160-242:1364903:1365409 [1] NCCL INFO Channel 15/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-165-131:1630208:1630729 [3] NCCL INFO Channel 05/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-165-131:1630209:1630730 [4] NCCL INFO Channel 12/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-161-153:2636086:2636602 [2] NCCL INFO Channel 08/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-161-178:2815967:2816473 [4] NCCL INFO Channel 05/0 : 28[4] -> 31[7] via P2P/IPC
+ip-26-0-165-131:1630211:1630733 [6] NCCL INFO Channel 03/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-161-138:96567:97090 [4] NCCL INFO Channel 13/0 : 4[4] -> 7[7] via P2P/IPC
+ip-26-0-165-202:153568:154056 [6] NCCL INFO Channel 12/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-165-202:153565:154058 [3] NCCL INFO Channel 12/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-161-142:62885:63377 [4] NCCL INFO Channel 06/0 : 12[4] -> 15[7] via P2P/IPC
+ip-26-0-161-178:2815967:2816473 [4] NCCL INFO Channel 06/0 : 28[4] -> 31[7] via P2P/IPC
+ip-26-0-161-153:2636087:2636603 [3] NCCL INFO Channel 08/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-161-142:62885:63377 [4] NCCL INFO Channel 13/0 : 12[4] -> 15[7] via P2P/IPC
+ip-26-0-161-138:96567:97090 [4] NCCL INFO Channel 14/0 : 4[4] -> 7[7] via P2P/IPC
+ip-26-0-165-131:1630208:1630729 [3] NCCL INFO Channel 06/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-165-131:1630211:1630733 [6] NCCL INFO Channel 04/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-161-142:62885:63377 [4] NCCL INFO Channel 14/0 : 12[4] -> 15[7] via P2P/IPC
+ip-26-0-161-178:2815967:2816473 [4] NCCL INFO Channel 13/0 : 28[4] -> 31[7] via P2P/IPC
+ip-26-0-161-153:2636085:2636604 [1] NCCL INFO Channel 08/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-161-153:2636088:2636601 [4] NCCL INFO Channel 08/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-165-164:1227469:1227957 [6] NCCL INFO Channel 10/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-161-178:2815967:2816473 [4] NCCL INFO Channel 14/0 : 28[4] -> 31[7] via P2P/IPC
+ip-26-0-165-131:1630208:1630729 [3] NCCL INFO Channel 07/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-165-131:1630211:1630733 [6] NCCL INFO Channel 07/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-161-103:96890:97386 [2] NCCL INFO Channel 13/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-165-164:1227468:1227959 [5] NCCL INFO Channel 06/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-165-164:1227466:1227961 [3] NCCL INFO Channel 13/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-165-202:153568:154056 [6] NCCL INFO Channel 15/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-165-164:1227469:1227957 [6] NCCL INFO Channel 11/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-165-202:153567:154059 [5] NCCL INFO Channel 07/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-165-131:1630208:1630729 [3] NCCL INFO Channel 09/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-165-131:1630211:1630733 [6] NCCL INFO Channel 09/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-161-153:2636088:2636601 [4] NCCL INFO Channel 05/0 : 20[4] -> 23[7] via P2P/IPC
+ip-26-0-165-164:1227468:1227959 [5] NCCL INFO Channel 07/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-165-164:1227466:1227961 [3] NCCL INFO Channel 14/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-165-164:1227469:1227957 [6] NCCL INFO Channel 12/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-161-153:2636088:2636601 [4] NCCL INFO Channel 06/0 : 20[4] -> 23[7] via P2P/IPC
+ip-26-0-165-202:153565:154058 [3] NCCL INFO Channel 13/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-165-202:153567:154059 [5] NCCL INFO Channel 09/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-165-131:1630208:1630729 [3] NCCL INFO Channel 11/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-165-131:1630211:1630733 [6] NCCL INFO Channel 10/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-165-131:1630210:1630732 [5] NCCL INFO Channel 01/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-161-153:2636088:2636601 [4] NCCL INFO Channel 13/0 : 20[4] -> 23[7] via P2P/IPC
+ip-26-0-161-103:96890:97386 [2] NCCL INFO Channel 14/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-165-164:1227468:1227959 [5] NCCL INFO Channel 09/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-165-164:1227466:1227961 [3] NCCL INFO Channel 15/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-165-164:1227469:1227957 [6] NCCL INFO Channel 15/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-161-153:2636088:2636601 [4] NCCL INFO Channel 14/0 : 20[4] -> 23[7] via P2P/IPC
+ip-26-0-161-103:96890:97386 [2] NCCL INFO Channel 15/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-165-131:1630208:1630729 [3] NCCL INFO Channel 12/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-165-131:1630211:1630733 [6] NCCL INFO Channel 11/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-165-131:1630210:1630732 [5] NCCL INFO Channel 02/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-165-131:1630208:1630729 [3] NCCL INFO Channel 13/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-165-202:153565:154058 [3] NCCL INFO Channel 14/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-165-202:153567:154059 [5] NCCL INFO Channel 10/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-165-131:1630210:1630732 [5] NCCL INFO Channel 03/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-165-131:1630208:1630729 [3] NCCL INFO Channel 14/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-165-164:1227468:1227959 [5] NCCL INFO Channel 10/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-165-131:1630210:1630732 [5] NCCL INFO Channel 06/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-165-131:1630208:1630729 [3] NCCL INFO Channel 15/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-165-131:1630211:1630733 [6] NCCL INFO Channel 12/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-165-131:1630210:1630732 [5] NCCL INFO Channel 07/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-165-164:1227468:1227959 [5] NCCL INFO Channel 11/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-165-131:1630211:1630733 [6] NCCL INFO Channel 15/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-165-131:1630210:1630732 [5] NCCL INFO Channel 09/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-165-59:1557267:1557761 [1] NCCL INFO Channel 01/0 : 26[2] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557267:1557761 [1] NCCL INFO Channel 09/0 : 26[2] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630210:1630732 [5] NCCL INFO Channel 10/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-165-202:153565:154058 [3] NCCL INFO Channel 15/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-161-78:114517:115013 [5] NCCL INFO Channel 14/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-202:153567:154059 [5] NCCL INFO Channel 11/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-161-78:114518:115012 [6] NCCL INFO Channel 11/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-168-52:2722047:2722542 [2] NCCL INFO Channel 02/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-168-52:2722047:2722542 [2] NCCL INFO Channel 03/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-165-202:153567:154059 [5] NCCL INFO Channel 14/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-168-52:2722047:2722542 [2] NCCL INFO Channel 04/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-168-52:2722047:2722542 [2] NCCL INFO Channel 05/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-168-52:2722047:2722542 [2] NCCL INFO Channel 06/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-165-164:1227468:1227959 [5] NCCL INFO Channel 14/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-168-52:2722047:2722542 [2] NCCL INFO Channel 07/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-168-52:2722047:2722542 [2] NCCL INFO Channel 10/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-168-52:2722049:2722547 [4] NCCL INFO Channel 04/0 : 21[5] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722049:2722547 [4] NCCL INFO Channel 12/0 : 21[5] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722049:2722547 [4] NCCL INFO Channel 03/0 : 28[4] -> 3[3] [send] via NET/Libfabric/3(27)/GDRDMA
+ip-26-0-168-52:2722049:2722547 [4] NCCL INFO Channel 11/0 : 28[4] -> 3[3] [send] via NET/Libfabric/3(27)/GDRDMA
+ip-26-0-168-52:2722050:2722548 [5] NCCL INFO Channel 05/0 : 22[6] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-52:2722050:2722548 [5] NCCL INFO Channel 13/0 : 22[6] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272155:272652 [5] NCCL INFO Channel 05/0 : 21[5] -> 19[3] via P2P/IPC
+ip-26-0-168-52:2722050:2722548 [5] NCCL INFO Channel 04/0 : 29[5] -> 4[4] [send] via NET/Libfabric/4(28)/GDRDMA
+ip-26-0-168-52:2722050:2722548 [5] NCCL INFO Channel 12/0 : 29[5] -> 4[4] [send] via NET/Libfabric/4(28)/GDRDMA
+ip-26-0-165-131:1630210:1630732 [5] NCCL INFO Channel 11/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-168-34:272156:272648 [6] NCCL INFO Channel 06/0 : 22[6] -> 19[3] via P2P/IPC
+ip-26-0-161-78:114517:115013 [5] NCCL INFO Channel 15/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-168-52:2722051:2722543 [6] NCCL INFO Channel 06/0 : 23[7] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-52:2722047:2722542 [2] NCCL INFO Channel 11/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-161-78:114518:115012 [6] NCCL INFO Channel 12/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-168-52:2722052:2722546 [7] NCCL INFO Channel 14/0 : 31[7] -> 6[6] [send] via NET/Libfabric/6(30)/GDRDMA
+ip-26-0-168-52:2722051:2722543 [6] NCCL INFO Channel 14/0 : 23[7] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-52:2722047:2722542 [2] NCCL INFO Channel 12/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-168-52:2722051:2722543 [6] NCCL INFO Channel 05/0 : 30[6] -> 5[5] [send] via NET/Libfabric/5(29)/GDRDMA
+ip-26-0-168-52:2722051:2722543 [6] NCCL INFO Channel 13/0 : 30[6] -> 5[5] [send] via NET/Libfabric/5(29)/GDRDMA
+ip-26-0-168-52:2722049:2722547 [4] NCCL INFO Channel 07/0 : 28[4] -> 7[7] [send] via NET/Libfabric/7(31)/GDRDMA
+ip-26-0-168-52:2722049:2722547 [4] NCCL INFO Channel 15/0 : 28[4] -> 7[7] [send] via NET/Libfabric/7(31)/GDRDMA
+ip-26-0-168-52:2722047:2722542 [2] NCCL INFO Channel 13/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-168-52:2722047:2722542 [2] NCCL INFO Channel 14/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-168-52:2722047:2722542 [2] NCCL INFO Channel 15/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-168-34:272155:272652 [5] NCCL INFO Channel 13/0 : 21[5] -> 19[3] via P2P/IPC
+ip-26-0-168-34:272156:272648 [6] NCCL INFO Channel 14/0 : 22[6] -> 19[3] via P2P/IPC
+ip-26-0-165-131:1630210:1630732 [5] NCCL INFO Channel 14/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-165-202:153567:154059 [5] NCCL INFO Channel 15/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-161-78:114518:115012 [6] NCCL INFO Channel 15/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-164:1227468:1227959 [5] NCCL INFO Channel 15/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-165-131:1630210:1630732 [5] NCCL INFO Channel 15/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-165-59:1557266:1557759 [0] NCCL INFO Channel 14/0 : 0[0] -> 5[5] via P2P/IPC
+ip-26-0-165-59:1557266:1557759 [0] NCCL INFO Channel 07/0 : 0[0] -> 6[6] via P2P/IPC
+ip-26-0-165-59:1557266:1557759 [0] NCCL INFO Channel 15/0 : 0[0] -> 6[6] via P2P/IPC
+ip-26-0-168-30:235655:236142 [7] NCCL INFO Channel 07/0 : 4[4] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235655:236142 [7] NCCL INFO Channel 15/0 : 4[4] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124167:124664 [7] NCCL INFO Channel 07/0 : 28[4] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124164:124663 [4] NCCL INFO Channel 15/0 : 4[4] -> 15[7] [send] via NET/Libfabric/7(7)/GDRDMA
+ip-26-0-160-225:124167:124664 [7] NCCL INFO Channel 15/0 : 28[4] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96895:97382 [7] NCCL INFO Channel 07/0 : 12[4] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96895:97382 [7] NCCL INFO Channel 15/0 : 12[4] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364909:1365405 [7] NCCL INFO Channel 07/0 : 4[4] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364906:1365407 [4] NCCL INFO Channel 15/0 : 12[4] -> 23[7] [send] via NET/Libfabric/7(15)/GDRDMA
+ip-26-0-160-242:1364909:1365405 [7] NCCL INFO Channel 15/0 : 4[4] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96895:97382 [7] NCCL INFO Channel 07/0 : 23[7] -> 19[3] via P2P/IPC
+ip-26-0-160-242:1364909:1365405 [7] NCCL INFO Channel 07/0 : 15[7] -> 11[3] via P2P/IPC
+ip-26-0-161-123:32711:33204 [7] NCCL INFO Channel 07/0 : 20[4] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32708:33198 [4] NCCL INFO Channel 15/0 : 28[4] -> 7[7] [send] via NET/Libfabric/7(31)/GDRDMA
+ip-26-0-165-59:1557266:1557759 [0] NCCL INFO Channel 01/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-161-123:32711:33204 [7] NCCL INFO Channel 15/0 : 20[4] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124167:124664 [7] NCCL INFO Channel 07/0 : 7[7] -> 3[3] via P2P/IPC
+ip-26-0-161-123:32711:33204 [7] NCCL INFO Channel 07/0 : 31[7] -> 27[3] via P2P/IPC
+ip-26-0-165-59:1557273:1557760 [7] NCCL INFO Channel 06/0 : 7[7] -> 14[6] [send] via NET/Libfabric/6(6)/GDRDMA
+ip-26-0-161-103:96895:97382 [7] NCCL INFO Channel 15/0 : 23[7] -> 19[3] via P2P/IPC
+ip-26-0-161-142:62881:63375 [0] NCCL INFO Channel 00/0 : 7[7] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62881:63375 [0] NCCL INFO Channel 08/0 : 7[7] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62881:63375 [0] NCCL INFO Channel 00/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-161-142:62888:63380 [7] NCCL INFO Channel 00/0 : 15[7] -> 16[0] [send] via NET/Libfabric/0(8)/GDRDMA
+ip-26-0-161-138:96563:97086 [0] NCCL INFO Channel 00/0 : 31[7] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96563:97086 [0] NCCL INFO Channel 08/0 : 31[7] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96563:97086 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-138:96570:97093 [7] NCCL INFO Channel 00/0 : 7[7] -> 8[0] [send] via NET/Libfabric/0(0)/GDRDMA
+ip-26-0-161-142:62881:63375 [0] NCCL INFO Channel 08/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-161-142:62888:63380 [7] NCCL INFO Channel 08/0 : 15[7] -> 16[0] [send] via NET/Libfabric/0(8)/GDRDMA
+ip-26-0-161-138:96563:97086 [0] NCCL INFO Channel 08/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-138:96570:97093 [7] NCCL INFO Channel 08/0 : 7[7] -> 8[0] [send] via NET/Libfabric/0(0)/GDRDMA
+ip-26-0-161-142:62881:63375 [0] NCCL INFO Channel 05/0 : 8[0] -> 12[4] via P2P/IPC
+ip-26-0-161-142:62881:63375 [0] NCCL INFO Channel 13/0 : 8[0] -> 12[4] via P2P/IPC
+ip-26-0-161-178:2815963:2816477 [0] NCCL INFO Channel 00/0 : 23[7] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815963:2816477 [0] NCCL INFO Channel 08/0 : 23[7] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815963:2816477 [0] NCCL INFO Channel 00/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-161-178:2815970:2816470 [7] NCCL INFO Channel 00/0 : 31[7] -> 0[0] [send] via NET/Libfabric/0(24)/GDRDMA
+ip-26-0-161-123:32711:33204 [7] NCCL INFO Channel 15/0 : 31[7] -> 27[3] via P2P/IPC
+ip-26-0-161-178:2815963:2816477 [0] NCCL INFO Channel 08/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-161-178:2815970:2816470 [7] NCCL INFO Channel 08/0 : 31[7] -> 0[0] [send] via NET/Libfabric/0(24)/GDRDMA
+ip-26-0-165-59:1557270:1557763 [4] NCCL INFO Channel 04/0 : 29[5] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557270:1557763 [4] NCCL INFO Channel 12/0 : 29[5] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96563:97086 [0] NCCL INFO Channel 05/0 : 0[0] -> 4[4] via P2P/IPC
+ip-26-0-165-59:1557270:1557763 [4] NCCL INFO Channel 03/0 : 4[4] -> 11[3] [send] via NET/Libfabric/3(3)/GDRDMA
+ip-26-0-165-59:1557270:1557763 [4] NCCL INFO Channel 11/0 : 4[4] -> 11[3] [send] via NET/Libfabric/3(3)/GDRDMA
+ip-26-0-161-138:96563:97086 [0] NCCL INFO Channel 13/0 : 0[0] -> 4[4] via P2P/IPC
+ip-26-0-168-52:2722050:2722548 [5] NCCL INFO Channel 05/0 : 29[5] -> 27[3] via P2P/IPC
+ip-26-0-161-138:96565:97089 [2] NCCL INFO Channel 02/0 : 27[3] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96565:97089 [2] NCCL INFO Channel 10/0 : 27[3] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96895:97382 [7] NCCL INFO Channel 01/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-161-138:96563:97086 [0] NCCL INFO Channel 06/0 : 0[0] -> 5[5] via P2P/IPC
+ip-26-0-161-138:96563:97086 [0] NCCL INFO Channel 14/0 : 0[0] -> 5[5] via P2P/IPC
+ip-26-0-161-123:32711:33204 [7] NCCL INFO Channel 01/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-161-142:62883:63373 [2] NCCL INFO Channel 02/0 : 3[3] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96566:97087 [3] NCCL INFO Channel 03/0 : 28[4] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62883:63373 [2] NCCL INFO Channel 10/0 : 3[3] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96563:97086 [0] NCCL INFO Channel 07/0 : 0[0] -> 6[6] via P2P/IPC
+ip-26-0-161-138:96566:97087 [3] NCCL INFO Channel 11/0 : 28[4] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96566:97087 [3] NCCL INFO Channel 02/0 : 3[3] -> 10[2] [send] via NET/Libfabric/2(2)/GDRDMA
+ip-26-0-161-138:96566:97087 [3] NCCL INFO Channel 10/0 : 3[3] -> 10[2] [send] via NET/Libfabric/2(2)/GDRDMA
+ip-26-0-165-59:1557270:1557763 [4] NCCL INFO Channel 07/0 : 4[4] -> 15[7] [send] via NET/Libfabric/7(7)/GDRDMA
+ip-26-0-165-59:1557270:1557763 [4] NCCL INFO Channel 15/0 : 4[4] -> 15[7] [send] via NET/Libfabric/7(7)/GDRDMA
+ip-26-0-161-138:96563:97086 [0] NCCL INFO Channel 15/0 : 0[0] -> 6[6] via P2P/IPC
+ip-26-0-161-103:96895:97382 [7] NCCL INFO Channel 02/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-161-142:62884:63376 [3] NCCL INFO Channel 03/0 : 4[4] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235655:236142 [7] NCCL INFO Channel 07/0 : 15[7] -> 11[3] via P2P/IPC
+ip-26-0-161-142:62884:63376 [3] NCCL INFO Channel 11/0 : 4[4] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96563:97086 [0] NCCL INFO Channel 01/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-161-142:62884:63376 [3] NCCL INFO Channel 02/0 : 11[3] -> 18[2] [send] via NET/Libfabric/2(10)/GDRDMA
+ip-26-0-161-142:62884:63376 [3] NCCL INFO Channel 10/0 : 11[3] -> 18[2] [send] via NET/Libfabric/2(10)/GDRDMA
+ip-26-0-161-103:96895:97382 [7] NCCL INFO Channel 03/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-161-138:96563:97086 [0] NCCL INFO Channel 02/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-168-30:235655:236142 [7] NCCL INFO Channel 15/0 : 15[7] -> 11[3] via P2P/IPC
+ip-26-0-165-59:1557266:1557759 [0] NCCL INFO Channel 02/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-161-138:96563:97086 [0] NCCL INFO Channel 03/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-161-103:96895:97382 [7] NCCL INFO Channel 04/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-168-30:235655:236142 [7] NCCL INFO Channel 01/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-161-138:96563:97086 [0] NCCL INFO Channel 04/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-161-178:2815965:2816475 [2] NCCL INFO Channel 02/0 : 19[3] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815965:2816475 [2] NCCL INFO Channel 10/0 : 19[3] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96563:97086 [0] NCCL INFO Channel 09/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-161-103:96895:97382 [7] NCCL INFO Channel 05/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-168-30:235655:236142 [7] NCCL INFO Channel 02/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-160-242:1364909:1365405 [7] NCCL INFO Channel 15/0 : 15[7] -> 11[3] via P2P/IPC
+ip-26-0-161-178:2815965:2816475 [2] NCCL INFO Channel 01/0 : 26[2] -> 1[1] [send] via NET/Libfabric/1(25)/GDRDMA
+ip-26-0-161-138:96563:97086 [0] NCCL INFO Channel 10/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-161-178:2815965:2816475 [2] NCCL INFO Channel 09/0 : 26[2] -> 1[1] [send] via NET/Libfabric/1(25)/GDRDMA
+ip-26-0-160-225:124167:124664 [7] NCCL INFO Channel 15/0 : 7[7] -> 3[3] via P2P/IPC
+ip-26-0-168-30:235655:236142 [7] NCCL INFO Channel 03/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-161-123:32711:33204 [7] NCCL INFO Channel 02/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-161-138:96563:97086 [0] NCCL INFO Channel 11/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-161-138:96570:97093 [7] NCCL INFO Channel 06/0 : 7[7] -> 14[6] [send] via NET/Libfabric/6(6)/GDRDMA
+ip-26-0-161-138:96570:97093 [7] NCCL INFO Channel 14/0 : 7[7] -> 14[6] [send] via NET/Libfabric/6(6)/GDRDMA
+ip-26-0-161-178:2815966:2816474 [3] NCCL INFO Channel 03/0 : 20[4] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235655:236142 [7] NCCL INFO Channel 04/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-161-178:2815966:2816474 [3] NCCL INFO Channel 11/0 : 20[4] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96563:97086 [0] NCCL INFO Channel 12/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-168-52:2722050:2722548 [5] NCCL INFO Channel 13/0 : 29[5] -> 27[3] via P2P/IPC
+ip-26-0-161-178:2815970:2816470 [7] NCCL INFO Channel 06/0 : 31[7] -> 6[6] [send] via NET/Libfabric/6(30)/GDRDMA
+ip-26-0-160-225:124167:124664 [7] NCCL INFO Channel 01/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-178:2815966:2816474 [3] NCCL INFO Channel 02/0 : 27[3] -> 2[2] [send] via NET/Libfabric/2(26)/GDRDMA
+ip-26-0-168-30:235655:236142 [7] NCCL INFO Channel 05/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-161-178:2815970:2816470 [7] NCCL INFO Channel 14/0 : 31[7] -> 6[6] [send] via NET/Libfabric/6(30)/GDRDMA
+ip-26-0-161-178:2815966:2816474 [3] NCCL INFO Channel 10/0 : 27[3] -> 2[2] [send] via NET/Libfabric/2(26)/GDRDMA
+ip-26-0-161-103:96895:97382 [7] NCCL INFO Channel 09/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-168-30:235655:236142 [7] NCCL INFO Channel 09/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-160-225:124167:124664 [7] NCCL INFO Channel 02/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-168-30:235655:236142 [7] NCCL INFO Channel 10/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-168-30:235655:236142 [7] NCCL INFO Channel 11/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-161-123:32711:33204 [7] NCCL INFO Channel 03/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-161-103:96895:97382 [7] NCCL INFO Channel 10/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-160-225:124167:124664 [7] NCCL INFO Channel 03/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-168-30:235655:236142 [7] NCCL INFO Channel 12/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-160-225:124167:124664 [7] NCCL INFO Channel 04/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-168-30:235655:236142 [7] NCCL INFO Channel 13/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-161-123:32711:33204 [7] NCCL INFO Channel 04/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-161-153:2636084:2636605 [0] NCCL INFO Channel 00/0 : 15[7] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:124167:124664 [7] NCCL INFO Channel 05/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-153:2636084:2636605 [0] NCCL INFO Channel 08/0 : 15[7] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636084:2636605 [0] NCCL INFO Channel 00/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-161-153:2636091:2636599 [7] NCCL INFO Channel 00/0 : 23[7] -> 24[0] [send] via NET/Libfabric/0(16)/GDRDMA
+ip-26-0-161-123:32711:33204 [7] NCCL INFO Channel 05/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-160-225:124167:124664 [7] NCCL INFO Channel 09/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-225:124167:124664 [7] NCCL INFO Channel 10/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-103:96895:97382 [7] NCCL INFO Channel 11/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-161-123:32711:33204 [7] NCCL INFO Channel 09/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-161-153:2636084:2636605 [0] NCCL INFO Channel 08/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-160-242:1364909:1365405 [7] NCCL INFO Channel 01/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-160-225:124164:124663 [4] NCCL INFO Channel 01/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-225:124167:124664 [7] NCCL INFO Channel 11/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-153:2636091:2636599 [7] NCCL INFO Channel 08/0 : 23[7] -> 24[0] [send] via NET/Libfabric/0(16)/GDRDMA
+ip-26-0-161-153:2636084:2636605 [0] NCCL INFO Channel 05/0 : 16[0] -> 20[4] via P2P/IPC
+ip-26-0-161-123:32711:33204 [7] NCCL INFO Channel 10/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-161-178:2815963:2816477 [0] NCCL INFO Channel 05/0 : 24[0] -> 28[4] via P2P/IPC
+ip-26-0-160-225:124164:124663 [4] NCCL INFO Channel 02/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-225:124167:124664 [7] NCCL INFO Channel 12/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-178:2815963:2816477 [0] NCCL INFO Channel 13/0 : 24[0] -> 28[4] via P2P/IPC
+ip-26-0-161-153:2636086:2636602 [2] NCCL INFO Channel 02/0 : 11[3] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636086:2636602 [2] NCCL INFO Channel 10/0 : 11[3] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96895:97382 [7] NCCL INFO Channel 12/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-161-123:32708:33198 [4] NCCL INFO Channel 01/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-161-153:2636084:2636605 [0] NCCL INFO Channel 13/0 : 16[0] -> 20[4] via P2P/IPC
+ip-26-0-161-178:2815963:2816477 [0] NCCL INFO Channel 06/0 : 24[0] -> 29[5] via P2P/IPC
+ip-26-0-161-153:2636087:2636603 [3] NCCL INFO Channel 03/0 : 12[4] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32711:33204 [7] NCCL INFO Channel 11/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-160-225:124164:124663 [4] NCCL INFO Channel 04/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-225:124167:124664 [7] NCCL INFO Channel 13/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-178:2815963:2816477 [0] NCCL INFO Channel 14/0 : 24[0] -> 29[5] via P2P/IPC
+ip-26-0-161-153:2636087:2636603 [3] NCCL INFO Channel 11/0 : 12[4] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636087:2636603 [3] NCCL INFO Channel 02/0 : 19[3] -> 26[2] [send] via NET/Libfabric/2(18)/GDRDMA
+ip-26-0-161-153:2636087:2636603 [3] NCCL INFO Channel 10/0 : 19[3] -> 26[2] [send] via NET/Libfabric/2(18)/GDRDMA
+ip-26-0-161-178:2815963:2816477 [0] NCCL INFO Channel 07/0 : 24[0] -> 30[6] via P2P/IPC
+ip-26-0-161-123:32708:33198 [4] NCCL INFO Channel 02/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-161-123:32711:33204 [7] NCCL INFO Channel 12/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-161-178:2815963:2816477 [0] NCCL INFO Channel 15/0 : 24[0] -> 30[6] via P2P/IPC
+ip-26-0-161-103:96895:97382 [7] NCCL INFO Channel 13/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-160-225:124164:124663 [4] NCCL INFO Channel 09/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-178:2815963:2816477 [0] NCCL INFO Channel 01/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-161-103:96892:97383 [4] NCCL INFO Channel 01/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-161-123:32708:33198 [4] NCCL INFO Channel 04/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-161-178:2815963:2816477 [0] NCCL INFO Channel 02/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-161-123:32711:33204 [7] NCCL INFO Channel 13/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-160-225:124164:124663 [4] NCCL INFO Channel 10/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-142:62881:63375 [0] NCCL INFO Channel 06/0 : 8[0] -> 13[5] via P2P/IPC
+ip-26-0-161-178:2815963:2816477 [0] NCCL INFO Channel 03/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-161-103:96892:97383 [4] NCCL INFO Channel 02/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-161-178:2815963:2816477 [0] NCCL INFO Channel 04/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-161-153:2636084:2636605 [0] NCCL INFO Channel 06/0 : 16[0] -> 21[5] via P2P/IPC
+ip-26-0-161-123:32708:33198 [4] NCCL INFO Channel 09/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-160-225:124164:124663 [4] NCCL INFO Channel 12/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-178:2815963:2816477 [0] NCCL INFO Channel 09/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-161-142:62881:63375 [0] NCCL INFO Channel 14/0 : 8[0] -> 13[5] via P2P/IPC
+ip-26-0-161-153:2636084:2636605 [0] NCCL INFO Channel 14/0 : 16[0] -> 21[5] via P2P/IPC
+ip-26-0-161-178:2815963:2816477 [0] NCCL INFO Channel 10/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-160-242:1364909:1365405 [7] NCCL INFO Channel 02/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-161-123:32708:33198 [4] NCCL INFO Channel 10/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-161-142:62881:63375 [0] NCCL INFO Channel 07/0 : 8[0] -> 14[6] via P2P/IPC
+ip-26-0-161-178:2815963:2816477 [0] NCCL INFO Channel 11/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-161-153:2636084:2636605 [0] NCCL INFO Channel 07/0 : 16[0] -> 22[6] via P2P/IPC
+ip-26-0-161-123:32707:33200 [3] NCCL INFO Channel 01/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-161-103:96892:97383 [4] NCCL INFO Channel 04/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-161-142:62881:63375 [0] NCCL INFO Channel 15/0 : 8[0] -> 14[6] via P2P/IPC
+ip-26-0-161-178:2815963:2816477 [0] NCCL INFO Channel 12/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-160-242:1364909:1365405 [7] NCCL INFO Channel 03/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-161-153:2636084:2636605 [0] NCCL INFO Channel 15/0 : 16[0] -> 22[6] via P2P/IPC
+ip-26-0-161-123:32708:33198 [4] NCCL INFO Channel 12/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-161-123:32707:33200 [3] NCCL INFO Channel 03/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-161-142:62881:63375 [0] NCCL INFO Channel 01/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-161-103:96892:97383 [4] NCCL INFO Channel 09/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-160-242:1364909:1365405 [7] NCCL INFO Channel 04/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-161-153:2636084:2636605 [0] NCCL INFO Channel 01/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-161-103:96892:97383 [4] NCCL INFO Channel 10/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-161-153:2636084:2636605 [0] NCCL INFO Channel 02/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-161-103:96891:97385 [3] NCCL INFO Channel 01/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-161-153:2636084:2636605 [0] NCCL INFO Channel 03/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-160-225:124166:124665 [6] NCCL INFO Channel 01/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-242:1364909:1365405 [7] NCCL INFO Channel 05/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-160-225:124163:124661 [3] NCCL INFO Channel 01/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-142:62881:63375 [0] NCCL INFO Channel 02/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-161-103:96892:97383 [4] NCCL INFO Channel 12/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-161-153:2636084:2636605 [0] NCCL INFO Channel 04/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-161-142:62881:63375 [0] NCCL INFO Channel 03/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-161-103:96891:97385 [3] NCCL INFO Channel 03/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-160-225:124166:124665 [6] NCCL INFO Channel 02/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-225:124165:124662 [5] NCCL INFO Channel 01/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-153:2636084:2636605 [0] NCCL INFO Channel 09/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-160-242:1364909:1365405 [7] NCCL INFO Channel 09/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-160-225:124163:124661 [3] NCCL INFO Channel 03/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-153:2636084:2636605 [0] NCCL INFO Channel 10/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-161-103:96893:97384 [5] NCCL INFO Channel 01/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-161-153:2636091:2636599 [7] NCCL INFO Channel 06/0 : 23[7] -> 30[6] [send] via NET/Libfabric/6(22)/GDRDMA
+ip-26-0-161-153:2636091:2636599 [7] NCCL INFO Channel 14/0 : 23[7] -> 30[6] [send] via NET/Libfabric/6(22)/GDRDMA
+ip-26-0-161-103:96891:97385 [3] NCCL INFO Channel 04/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-161-153:2636084:2636605 [0] NCCL INFO Channel 11/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-160-225:124165:124662 [5] NCCL INFO Channel 02/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-123:32707:33200 [3] NCCL INFO Channel 04/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-160-225:124166:124665 [6] NCCL INFO Channel 03/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-225:124163:124661 [3] NCCL INFO Channel 04/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-153:2636084:2636605 [0] NCCL INFO Channel 12/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-161-123:32709:33201 [5] NCCL INFO Channel 01/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-161-103:96893:97384 [5] NCCL INFO Channel 02/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-160-242:1364909:1365405 [7] NCCL INFO Channel 10/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-161-103:96894:97381 [6] NCCL INFO Channel 01/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-160-242:1364906:1365407 [4] NCCL INFO Channel 01/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-161-103:96891:97385 [3] NCCL INFO Channel 05/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-161-123:32707:33200 [3] NCCL INFO Channel 05/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-161-142:62881:63375 [0] NCCL INFO Channel 04/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-161-123:32709:33201 [5] NCCL INFO Channel 02/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-160-225:124165:124662 [5] NCCL INFO Channel 03/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-225:124166:124665 [6] NCCL INFO Channel 04/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-225:124163:124661 [3] NCCL INFO Channel 05/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-142:62881:63375 [0] NCCL INFO Channel 09/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-160-242:1364909:1365405 [7] NCCL INFO Channel 11/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-161-103:96893:97384 [5] NCCL INFO Channel 03/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-161-103:96894:97381 [6] NCCL INFO Channel 02/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-160-242:1364906:1365407 [4] NCCL INFO Channel 02/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-161-123:32707:33200 [3] NCCL INFO Channel 06/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-161-103:96891:97385 [3] NCCL INFO Channel 06/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-161-142:62881:63375 [0] NCCL INFO Channel 10/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-161-123:32709:33201 [5] NCCL INFO Channel 03/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-160-225:124163:124661 [3] NCCL INFO Channel 06/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-225:124166:124665 [6] NCCL INFO Channel 07/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-225:124165:124662 [5] NCCL INFO Channel 06/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-142:62881:63375 [0] NCCL INFO Channel 11/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-161-123:32707:33200 [3] NCCL INFO Channel 07/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-161-103:96893:97384 [5] NCCL INFO Channel 06/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-161-103:96894:97381 [6] NCCL INFO Channel 03/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-161-123:32710:33202 [6] NCCL INFO Channel 01/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-161-123:32709:33201 [5] NCCL INFO Channel 06/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-160-242:1364909:1365405 [7] NCCL INFO Channel 12/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-161-142:62881:63375 [0] NCCL INFO Channel 12/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-161-103:96891:97385 [3] NCCL INFO Channel 07/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-160-242:1364906:1365407 [4] NCCL INFO Channel 04/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-160-225:124163:124661 [3] NCCL INFO Channel 07/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-225:124166:124665 [6] NCCL INFO Channel 09/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-225:124165:124662 [5] NCCL INFO Channel 07/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-123:32707:33200 [3] NCCL INFO Channel 09/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-161-123:32709:33201 [5] NCCL INFO Channel 07/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-161-103:96893:97384 [5] NCCL INFO Channel 07/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-161-103:96894:97381 [6] NCCL INFO Channel 04/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-161-123:32710:33202 [6] NCCL INFO Channel 02/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-161-103:96891:97385 [3] NCCL INFO Channel 09/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-160-242:1364909:1365405 [7] NCCL INFO Channel 13/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-160-242:1364906:1365407 [4] NCCL INFO Channel 09/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-160-225:124163:124661 [3] NCCL INFO Channel 09/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-225:124166:124665 [6] NCCL INFO Channel 10/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-225:124165:124662 [5] NCCL INFO Channel 09/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-168-34:272157:272649 [7] NCCL INFO Channel 07/0 : 12[4] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96893:97384 [5] NCCL INFO Channel 09/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-168-34:272157:272649 [7] NCCL INFO Channel 15/0 : 12[4] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32707:33200 [3] NCCL INFO Channel 11/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-161-123:32709:33201 [5] NCCL INFO Channel 09/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-161-103:96894:97381 [6] NCCL INFO Channel 07/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-161-103:96891:97385 [3] NCCL INFO Channel 11/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-161-123:32710:33202 [6] NCCL INFO Channel 03/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-160-225:124163:124661 [3] NCCL INFO Channel 11/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-225:124166:124665 [6] NCCL INFO Channel 11/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-242:1364906:1365407 [4] NCCL INFO Channel 10/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-160-225:124165:124662 [5] NCCL INFO Channel 10/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-123:32707:33200 [3] NCCL INFO Channel 12/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-161-123:32709:33201 [5] NCCL INFO Channel 10/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-161-123:32710:33202 [6] NCCL INFO Channel 04/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-161-103:96893:97384 [5] NCCL INFO Channel 10/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-161-103:96894:97381 [6] NCCL INFO Channel 09/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-160-225:124163:124661 [3] NCCL INFO Channel 12/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-225:124166:124665 [6] NCCL INFO Channel 12/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-103:96891:97385 [3] NCCL INFO Channel 12/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-160-225:124165:124662 [5] NCCL INFO Channel 11/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-242:1364906:1365407 [4] NCCL INFO Channel 12/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-160-242:1364905:1365404 [3] NCCL INFO Channel 01/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-161-123:32707:33200 [3] NCCL INFO Channel 13/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-161-123:32709:33201 [5] NCCL INFO Channel 11/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-161-123:32710:33202 [6] NCCL INFO Channel 07/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-161-103:96891:97385 [3] NCCL INFO Channel 13/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-160-242:1364905:1365404 [3] NCCL INFO Channel 03/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-161-103:96894:97381 [6] NCCL INFO Channel 10/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-161-103:96891:97385 [3] NCCL INFO Channel 14/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-161-123:32709:33201 [5] NCCL INFO Channel 14/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-161-103:96894:97381 [6] NCCL INFO Channel 11/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-160-225:124163:124661 [3] NCCL INFO Channel 13/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-242:1364905:1365404 [3] NCCL INFO Channel 04/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-161-123:32709:33201 [5] NCCL INFO Channel 15/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-161-103:96891:97385 [3] NCCL INFO Channel 15/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-160-242:1364908:1365406 [6] NCCL INFO Channel 01/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-160-225:124163:124661 [3] NCCL INFO Channel 14/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-103:96894:97381 [6] NCCL INFO Channel 12/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-160-225:124163:124661 [3] NCCL INFO Channel 15/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-242:1364905:1365404 [3] NCCL INFO Channel 05/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-160-242:1364908:1365406 [6] NCCL INFO Channel 02/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-161-103:96893:97384 [5] NCCL INFO Channel 11/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-161-103:96894:97381 [6] NCCL INFO Channel 15/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-160-225:124166:124665 [6] NCCL INFO Channel 15/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-123:32707:33200 [3] NCCL INFO Channel 14/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-160-242:1364905:1365404 [3] NCCL INFO Channel 06/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-161-123:32710:33202 [6] NCCL INFO Channel 09/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-160-242:1364907:1365403 [5] NCCL INFO Channel 01/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-160-242:1364908:1365406 [6] NCCL INFO Channel 03/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-161-123:32707:33200 [3] NCCL INFO Channel 15/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-161-123:32710:33202 [6] NCCL INFO Channel 10/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-160-242:1364905:1365404 [3] NCCL INFO Channel 07/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-160-242:1364907:1365403 [5] NCCL INFO Channel 02/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-160-242:1364908:1365406 [6] NCCL INFO Channel 04/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-161-123:32710:33202 [6] NCCL INFO Channel 11/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-160-225:124165:124662 [5] NCCL INFO Channel 14/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-242:1364907:1365403 [5] NCCL INFO Channel 03/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-160-242:1364905:1365404 [3] NCCL INFO Channel 09/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-160-242:1364908:1365406 [6] NCCL INFO Channel 07/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-161-123:32710:33202 [6] NCCL INFO Channel 12/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-161-123:32710:33202 [6] NCCL INFO Channel 15/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-160-242:1364907:1365403 [5] NCCL INFO Channel 06/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-160-242:1364905:1365404 [3] NCCL INFO Channel 11/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-160-242:1364908:1365406 [6] NCCL INFO Channel 09/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-160-242:1364905:1365404 [3] NCCL INFO Channel 12/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-160-242:1364907:1365403 [5] NCCL INFO Channel 07/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-160-242:1364908:1365406 [6] NCCL INFO Channel 10/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-161-103:96893:97384 [5] NCCL INFO Channel 14/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-160-242:1364905:1365404 [3] NCCL INFO Channel 13/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-160-242:1364907:1365403 [5] NCCL INFO Channel 09/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-160-242:1364908:1365406 [6] NCCL INFO Channel 11/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-160-242:1364908:1365406 [6] NCCL INFO Channel 12/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-160-242:1364908:1365406 [6] NCCL INFO Channel 15/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-161-103:96893:97384 [5] NCCL INFO Channel 15/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-160-242:1364905:1365404 [3] NCCL INFO Channel 14/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-160-242:1364907:1365403 [5] NCCL INFO Channel 10/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-160-225:124165:124662 [5] NCCL INFO Channel 15/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-242:1364905:1365404 [3] NCCL INFO Channel 15/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-160-242:1364907:1365403 [5] NCCL INFO Channel 11/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-165-59:1557266:1557759 [0] NCCL INFO Channel 03/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-165-59:1557266:1557759 [0] NCCL INFO Channel 04/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-160-242:1364907:1365403 [5] NCCL INFO Channel 14/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-165-59:1557271:1557762 [5] NCCL INFO Channel 05/0 : 30[6] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557271:1557762 [5] NCCL INFO Channel 13/0 : 30[6] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364907:1365403 [5] NCCL INFO Channel 15/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-165-59:1557272:1557766 [6] NCCL INFO Channel 06/0 : 31[7] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557273:1557760 [7] NCCL INFO Channel 14/0 : 7[7] -> 14[6] [send] via NET/Libfabric/6(6)/GDRDMA
+ip-26-0-165-59:1557272:1557766 [6] NCCL INFO Channel 14/0 : 31[7] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557272:1557766 [6] NCCL INFO Channel 05/0 : 6[6] -> 13[5] [send] via NET/Libfabric/5(5)/GDRDMA
+ip-26-0-165-59:1557272:1557766 [6] NCCL INFO Channel 13/0 : 6[6] -> 13[5] [send] via NET/Libfabric/5(5)/GDRDMA
+ip-26-0-168-30:235654:236146 [6] NCCL INFO Channel 06/0 : 14[6] -> 11[3] via P2P/IPC
+ip-26-0-165-59:1557272:1557766 [6] NCCL INFO Channel 06/0 : 6[6] -> 3[3] via P2P/IPC
+ip-26-0-165-59:1557268:1557764 [2] NCCL INFO Channel 02/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-165-59:1557272:1557766 [6] NCCL INFO Channel 14/0 : 6[6] -> 3[3] via P2P/IPC
+ip-26-0-165-59:1557266:1557759 [0] NCCL INFO Channel 09/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-168-30:235653:236140 [5] NCCL INFO Channel 05/0 : 13[5] -> 11[3] via P2P/IPC
+ip-26-0-168-30:235654:236146 [6] NCCL INFO Channel 14/0 : 14[6] -> 11[3] via P2P/IPC
+ip-26-0-165-59:1557268:1557764 [2] NCCL INFO Channel 03/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-138:96564:97088 [1] NCCL INFO Channel 01/0 : 26[2] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235653:236140 [5] NCCL INFO Channel 13/0 : 13[5] -> 11[3] via P2P/IPC
+ip-26-0-161-138:96564:97088 [1] NCCL INFO Channel 09/0 : 26[2] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557266:1557759 [0] NCCL INFO Channel 10/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-161-138:96565:97089 [2] NCCL INFO Channel 01/0 : 2[2] -> 9[1] [send] via NET/Libfabric/1(1)/GDRDMA
+ip-26-0-165-59:1557271:1557762 [5] NCCL INFO Channel 04/0 : 5[5] -> 12[4] [send] via NET/Libfabric/4(4)/GDRDMA
+ip-26-0-161-138:96565:97089 [2] NCCL INFO Channel 09/0 : 2[2] -> 9[1] [send] via NET/Libfabric/1(1)/GDRDMA
+ip-26-0-161-142:62882:63374 [1] NCCL INFO Channel 01/0 : 2[2] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557268:1557764 [2] NCCL INFO Channel 04/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-142:62882:63374 [1] NCCL INFO Channel 09/0 : 2[2] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62883:63373 [2] NCCL INFO Channel 01/0 : 10[2] -> 17[1] [send] via NET/Libfabric/1(9)/GDRDMA
+ip-26-0-161-142:62883:63373 [2] NCCL INFO Channel 09/0 : 10[2] -> 17[1] [send] via NET/Libfabric/1(9)/GDRDMA
+ip-26-0-165-59:1557266:1557759 [0] NCCL INFO Channel 11/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-165-59:1557271:1557762 [5] NCCL INFO Channel 12/0 : 5[5] -> 12[4] [send] via NET/Libfabric/4(4)/GDRDMA
+ip-26-0-165-59:1557271:1557762 [5] NCCL INFO Channel 05/0 : 5[5] -> 3[3] via P2P/IPC
+ip-26-0-168-30:235652:236147 [4] NCCL INFO Channel 07/0 : 12[4] -> 23[7] [send] via NET/Libfabric/7(15)/GDRDMA
+ip-26-0-168-52:2722051:2722543 [6] NCCL INFO Channel 06/0 : 30[6] -> 27[3] via P2P/IPC
+ip-26-0-168-30:235652:236147 [4] NCCL INFO Channel 15/0 : 12[4] -> 23[7] [send] via NET/Libfabric/7(15)/GDRDMA
+ip-26-0-165-59:1557268:1557764 [2] NCCL INFO Channel 05/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-168-34:272157:272649 [7] NCCL INFO Channel 07/0 : 23[7] -> 19[3] via P2P/IPC
+ip-26-0-168-52:2722051:2722543 [6] NCCL INFO Channel 14/0 : 30[6] -> 27[3] via P2P/IPC
+ip-26-0-168-34:272157:272649 [7] NCCL INFO Channel 15/0 : 23[7] -> 19[3] via P2P/IPC
+ip-26-0-165-59:1557266:1557759 [0] NCCL INFO Channel 12/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-165-59:1557267:1557761 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-168-34:272157:272649 [7] NCCL INFO Channel 01/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-165-59:1557268:1557764 [2] NCCL INFO Channel 06/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-165-59:1557271:1557762 [5] NCCL INFO Channel 13/0 : 5[5] -> 3[3] via P2P/IPC
+ip-26-0-161-142:62882:63374 [1] NCCL INFO Channel 01/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-161-142:62882:63374 [1] NCCL INFO Channel 02/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-161-142:62882:63374 [1] NCCL INFO Channel 03/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-161-142:62882:63374 [1] NCCL INFO Channel 04/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-161-138:96564:97088 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-59:1557267:1557761 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-142:62882:63374 [1] NCCL INFO Channel 05/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-161-138:96565:97089 [2] NCCL INFO Channel 02/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-168-34:272157:272649 [7] NCCL INFO Channel 02/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-161-138:96564:97088 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-142:62882:63374 [1] NCCL INFO Channel 06/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-168-34:272157:272649 [7] NCCL INFO Channel 03/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-161-142:62882:63374 [1] NCCL INFO Channel 07/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-161-138:96565:97089 [2] NCCL INFO Channel 03/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-138:96564:97088 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-168-34:272157:272649 [7] NCCL INFO Channel 04/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-161-142:62882:63374 [1] NCCL INFO Channel 09/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-168-30:235654:236146 [6] NCCL INFO Channel 01/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-161-138:96565:97089 [2] NCCL INFO Channel 04/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-142:62882:63374 [1] NCCL INFO Channel 10/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-161-138:96564:97088 [1] NCCL INFO Channel 04/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-168-30:235654:236146 [6] NCCL INFO Channel 02/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-168-30:235651:236141 [3] NCCL INFO Channel 01/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-168-34:272157:272649 [7] NCCL INFO Channel 05/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-161-142:62882:63374 [1] NCCL INFO Channel 11/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-161-138:96565:97089 [2] NCCL INFO Channel 05/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-138:96564:97088 [1] NCCL INFO Channel 05/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-142:62882:63374 [1] NCCL INFO Channel 12/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-161-178:2815965:2816475 [2] NCCL INFO Channel 02/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-168-30:235654:236146 [6] NCCL INFO Channel 03/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-168-34:272157:272649 [7] NCCL INFO Channel 09/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-161-142:62882:63374 [1] NCCL INFO Channel 13/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-168-30:235651:236141 [3] NCCL INFO Channel 03/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-161-138:96565:97089 [2] NCCL INFO Channel 06/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-178:2815965:2816475 [2] NCCL INFO Channel 03/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-161-138:96564:97088 [1] NCCL INFO Channel 06/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-142:62882:63374 [1] NCCL INFO Channel 14/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-161-178:2815965:2816475 [2] NCCL INFO Channel 04/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-168-34:272157:272649 [7] NCCL INFO Channel 10/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-161-138:96565:97089 [2] NCCL INFO Channel 07/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-142:62882:63374 [1] NCCL INFO Channel 15/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-161-138:96564:97088 [1] NCCL INFO Channel 07/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-168-30:235651:236141 [3] NCCL INFO Channel 04/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-168-30:235654:236146 [6] NCCL INFO Channel 04/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-161-178:2815965:2816475 [2] NCCL INFO Channel 05/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-161-178:2815965:2816475 [2] NCCL INFO Channel 06/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-168-34:272157:272649 [7] NCCL INFO Channel 11/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-161-138:96565:97089 [2] NCCL INFO Channel 10/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-138:96564:97088 [1] NCCL INFO Channel 09/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-178:2815965:2816475 [2] NCCL INFO Channel 07/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-168-30:235654:236146 [6] NCCL INFO Channel 07/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-161-138:96565:97089 [2] NCCL INFO Channel 11/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-168-30:235651:236141 [3] NCCL INFO Channel 05/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-168-34:272157:272649 [7] NCCL INFO Channel 12/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-161-178:2815965:2816475 [2] NCCL INFO Channel 10/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-161-138:96564:97088 [1] NCCL INFO Channel 10/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-178:2815965:2816475 [2] NCCL INFO Channel 11/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-161-138:96565:97089 [2] NCCL INFO Channel 12/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-178:2815964:2816476 [1] NCCL INFO Channel 01/0 : 18[2] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96564:97088 [1] NCCL INFO Channel 11/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-178:2815965:2816475 [2] NCCL INFO Channel 12/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-161-178:2815964:2816476 [1] NCCL INFO Channel 09/0 : 18[2] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235653:236140 [5] NCCL INFO Channel 01/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-168-30:235654:236146 [6] NCCL INFO Channel 09/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-168-30:235651:236141 [3] NCCL INFO Channel 06/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-168-34:272157:272649 [7] NCCL INFO Channel 13/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-161-178:2815965:2816475 [2] NCCL INFO Channel 13/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-161-138:96565:97089 [2] NCCL INFO Channel 13/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-138:96564:97088 [1] NCCL INFO Channel 12/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-138:96569:97092 [6] NCCL INFO Channel 06/0 : 31[7] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815965:2816475 [2] NCCL INFO Channel 14/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-161-138:96569:97092 [6] NCCL INFO Channel 14/0 : 31[7] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96568:97091 [5] NCCL INFO Channel 05/0 : 30[6] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96568:97091 [5] NCCL INFO Channel 13/0 : 30[6] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96564:97088 [1] NCCL INFO Channel 13/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-178:2815965:2816475 [2] NCCL INFO Channel 15/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-161-138:96565:97089 [2] NCCL INFO Channel 14/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-138:96569:97092 [6] NCCL INFO Channel 05/0 : 6[6] -> 13[5] [send] via NET/Libfabric/5(5)/GDRDMA
+ip-26-0-168-30:235653:236140 [5] NCCL INFO Channel 02/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-168-30:235654:236146 [6] NCCL INFO Channel 10/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-168-30:235651:236141 [3] NCCL INFO Channel 07/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-168-30:235652:236147 [4] NCCL INFO Channel 01/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-161-138:96569:97092 [6] NCCL INFO Channel 13/0 : 6[6] -> 13[5] [send] via NET/Libfabric/5(5)/GDRDMA
+ip-26-0-161-138:96564:97088 [1] NCCL INFO Channel 14/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-138:96565:97089 [2] NCCL INFO Channel 15/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-138:96564:97088 [1] NCCL INFO Channel 15/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-168-30:235654:236146 [6] NCCL INFO Channel 11/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-168-30:235653:236140 [5] NCCL INFO Channel 03/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-168-30:235651:236141 [3] NCCL INFO Channel 09/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-168-30:235652:236147 [4] NCCL INFO Channel 02/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-168-34:272153:272651 [3] NCCL INFO Channel 01/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-161-138:96567:97090 [4] NCCL INFO Channel 04/0 : 29[5] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96567:97090 [4] NCCL INFO Channel 12/0 : 29[5] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96568:97091 [5] NCCL INFO Channel 04/0 : 5[5] -> 12[4] [send] via NET/Libfabric/4(4)/GDRDMA
+ip-26-0-161-138:96568:97091 [5] NCCL INFO Channel 12/0 : 5[5] -> 12[4] [send] via NET/Libfabric/4(4)/GDRDMA
+ip-26-0-161-138:96567:97090 [4] NCCL INFO Channel 03/0 : 4[4] -> 11[3] [send] via NET/Libfabric/3(3)/GDRDMA
+ip-26-0-161-138:96567:97090 [4] NCCL INFO Channel 11/0 : 4[4] -> 11[3] [send] via NET/Libfabric/3(3)/GDRDMA
+ip-26-0-168-34:272153:272651 [3] NCCL INFO Channel 03/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-168-30:235654:236146 [6] NCCL INFO Channel 12/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-168-30:235653:236140 [5] NCCL INFO Channel 06/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-168-30:235651:236141 [3] NCCL INFO Channel 11/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-168-34:272156:272648 [6] NCCL INFO Channel 01/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-168-34:272153:272651 [3] NCCL INFO Channel 04/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-168-34:272156:272648 [6] NCCL INFO Channel 02/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-168-34:272153:272651 [3] NCCL INFO Channel 05/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-168-34:272155:272652 [5] NCCL INFO Channel 01/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-168-30:235652:236147 [4] NCCL INFO Channel 04/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-161-178:2815969:2816471 [6] NCCL INFO Channel 06/0 : 23[7] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815969:2816471 [6] NCCL INFO Channel 14/0 : 23[7] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272156:272648 [6] NCCL INFO Channel 03/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-168-34:272153:272651 [3] NCCL INFO Channel 06/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-168-34:272155:272652 [5] NCCL INFO Channel 02/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-161-142:62885:63377 [4] NCCL INFO Channel 04/0 : 5[5] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815967:2816473 [4] NCCL INFO Channel 04/0 : 21[5] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62885:63377 [4] NCCL INFO Channel 12/0 : 5[5] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235653:236140 [5] NCCL INFO Channel 07/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-168-30:235652:236147 [4] NCCL INFO Channel 09/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-161-178:2815967:2816473 [4] NCCL INFO Channel 12/0 : 21[5] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235651:236141 [3] NCCL INFO Channel 12/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-161-178:2815968:2816472 [5] NCCL INFO Channel 05/0 : 22[6] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62885:63377 [4] NCCL INFO Channel 03/0 : 12[4] -> 19[3] [send] via NET/Libfabric/3(11)/GDRDMA
+ip-26-0-161-178:2815969:2816471 [6] NCCL INFO Channel 05/0 : 30[6] -> 5[5] [send] via NET/Libfabric/5(29)/GDRDMA
+ip-26-0-161-178:2815968:2816472 [5] NCCL INFO Channel 13/0 : 22[6] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235654:236146 [6] NCCL INFO Channel 15/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-161-178:2815967:2816473 [4] NCCL INFO Channel 03/0 : 28[4] -> 3[3] [send] via NET/Libfabric/3(27)/GDRDMA
+ip-26-0-161-142:62885:63377 [4] NCCL INFO Channel 11/0 : 12[4] -> 19[3] [send] via NET/Libfabric/3(11)/GDRDMA
+ip-26-0-168-34:272153:272651 [3] NCCL INFO Channel 07/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-168-34:272156:272648 [6] NCCL INFO Channel 04/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-168-34:272155:272652 [5] NCCL INFO Channel 03/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-161-178:2815969:2816471 [6] NCCL INFO Channel 13/0 : 30[6] -> 5[5] [send] via NET/Libfabric/5(29)/GDRDMA
+ip-26-0-161-178:2815967:2816473 [4] NCCL INFO Channel 11/0 : 28[4] -> 3[3] [send] via NET/Libfabric/3(27)/GDRDMA
+ip-26-0-161-178:2815968:2816472 [5] NCCL INFO Channel 04/0 : 29[5] -> 4[4] [send] via NET/Libfabric/4(28)/GDRDMA
+ip-26-0-161-178:2815968:2816472 [5] NCCL INFO Channel 12/0 : 29[5] -> 4[4] [send] via NET/Libfabric/4(28)/GDRDMA
+ip-26-0-161-138:96568:97091 [5] NCCL INFO Channel 05/0 : 5[5] -> 3[3] via P2P/IPC
+ip-26-0-161-178:2815969:2816471 [6] NCCL INFO Channel 06/0 : 30[6] -> 27[3] via P2P/IPC
+ip-26-0-168-30:235652:236147 [4] NCCL INFO Channel 10/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-168-30:235653:236140 [5] NCCL INFO Channel 09/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-168-30:235651:236141 [3] NCCL INFO Channel 13/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-168-34:272153:272651 [3] NCCL INFO Channel 09/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-168-34:272156:272648 [6] NCCL INFO Channel 07/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-168-34:272155:272652 [5] NCCL INFO Channel 06/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-161-138:96568:97091 [5] NCCL INFO Channel 13/0 : 5[5] -> 3[3] via P2P/IPC
+ip-26-0-161-178:2815969:2816471 [6] NCCL INFO Channel 14/0 : 30[6] -> 27[3] via P2P/IPC
+ip-26-0-161-142:62887:63379 [6] NCCL INFO Channel 06/0 : 7[7] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62888:63380 [7] NCCL INFO Channel 06/0 : 15[7] -> 22[6] [send] via NET/Libfabric/6(14)/GDRDMA
+ip-26-0-161-142:62887:63379 [6] NCCL INFO Channel 14/0 : 7[7] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62888:63380 [7] NCCL INFO Channel 14/0 : 15[7] -> 22[6] [send] via NET/Libfabric/6(14)/GDRDMA
+ip-26-0-161-142:62886:63378 [5] NCCL INFO Channel 05/0 : 6[6] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272153:272651 [3] NCCL INFO Channel 11/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-168-34:272156:272648 [6] NCCL INFO Channel 09/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-168-34:272155:272652 [5] NCCL INFO Channel 07/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-161-142:62886:63378 [5] NCCL INFO Channel 13/0 : 6[6] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62887:63379 [6] NCCL INFO Channel 05/0 : 14[6] -> 21[5] [send] via NET/Libfabric/5(13)/GDRDMA
+ip-26-0-161-142:62887:63379 [6] NCCL INFO Channel 13/0 : 14[6] -> 21[5] [send] via NET/Libfabric/5(13)/GDRDMA
+ip-26-0-161-142:62886:63378 [5] NCCL INFO Channel 04/0 : 13[5] -> 20[4] [send] via NET/Libfabric/4(12)/GDRDMA
+ip-26-0-161-142:62886:63378 [5] NCCL INFO Channel 12/0 : 13[5] -> 20[4] [send] via NET/Libfabric/4(12)/GDRDMA
+ip-26-0-161-142:62885:63377 [4] NCCL INFO Channel 07/0 : 12[4] -> 23[7] [send] via NET/Libfabric/7(15)/GDRDMA
+ip-26-0-161-138:96569:97092 [6] NCCL INFO Channel 06/0 : 6[6] -> 3[3] via P2P/IPC
+ip-26-0-168-34:272156:272648 [6] NCCL INFO Channel 10/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-168-34:272153:272651 [3] NCCL INFO Channel 12/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-161-142:62885:63377 [4] NCCL INFO Channel 15/0 : 12[4] -> 23[7] [send] via NET/Libfabric/7(15)/GDRDMA
+ip-26-0-168-34:272155:272652 [5] NCCL INFO Channel 09/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-161-138:96567:97090 [4] NCCL INFO Channel 07/0 : 4[4] -> 15[7] [send] via NET/Libfabric/7(7)/GDRDMA
+ip-26-0-168-30:235652:236147 [4] NCCL INFO Channel 12/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-168-30:235653:236140 [5] NCCL INFO Channel 10/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-168-30:235651:236141 [3] NCCL INFO Channel 14/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-161-153:2636085:2636604 [1] NCCL INFO Channel 01/0 : 10[2] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636086:2636602 [2] NCCL INFO Channel 01/0 : 18[2] -> 25[1] [send] via NET/Libfabric/1(17)/GDRDMA
+ip-26-0-161-153:2636085:2636604 [1] NCCL INFO Channel 09/0 : 10[2] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636086:2636602 [2] NCCL INFO Channel 09/0 : 18[2] -> 25[1] [send] via NET/Libfabric/1(17)/GDRDMA
+ip-26-0-168-34:272153:272651 [3] NCCL INFO Channel 13/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-168-34:272155:272652 [5] NCCL INFO Channel 10/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-168-34:272156:272648 [6] NCCL INFO Channel 11/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-168-34:272153:272651 [3] NCCL INFO Channel 14/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-168-34:272155:272652 [5] NCCL INFO Channel 11/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-168-34:272156:272648 [6] NCCL INFO Channel 12/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-168-34:272153:272651 [3] NCCL INFO Channel 15/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-168-34:272155:272652 [5] NCCL INFO Channel 14/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-168-34:272156:272648 [6] NCCL INFO Channel 15/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-161-178:2815964:2816476 [1] NCCL INFO Channel 01/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-168-30:235653:236140 [5] NCCL INFO Channel 11/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-161-138:96569:97092 [6] NCCL INFO Channel 14/0 : 6[6] -> 3[3] via P2P/IPC
+ip-26-0-168-30:235651:236141 [3] NCCL INFO Channel 15/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-161-178:2815964:2816476 [1] NCCL INFO Channel 02/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-168-34:272155:272652 [5] NCCL INFO Channel 15/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-161-178:2815964:2816476 [1] NCCL INFO Channel 03/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-161-178:2815964:2816476 [1] NCCL INFO Channel 04/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-161-178:2815964:2816476 [1] NCCL INFO Channel 05/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-161-178:2815964:2816476 [1] NCCL INFO Channel 06/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-161-153:2636088:2636601 [4] NCCL INFO Channel 04/0 : 13[5] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636088:2636601 [4] NCCL INFO Channel 12/0 : 13[5] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815964:2816476 [1] NCCL INFO Channel 07/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-161-153:2636088:2636601 [4] NCCL INFO Channel 03/0 : 20[4] -> 27[3] [send] via NET/Libfabric/3(19)/GDRDMA
+ip-26-0-161-178:2815964:2816476 [1] NCCL INFO Channel 09/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-161-153:2636086:2636602 [2] NCCL INFO Channel 02/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-161-153:2636088:2636601 [4] NCCL INFO Channel 11/0 : 20[4] -> 27[3] [send] via NET/Libfabric/3(19)/GDRDMA
+ip-26-0-161-153:2636085:2636604 [1] NCCL INFO Channel 01/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-161-178:2815964:2816476 [1] NCCL INFO Channel 10/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-161-142:62886:63378 [5] NCCL INFO Channel 05/0 : 13[5] -> 11[3] via P2P/IPC
+ip-26-0-161-178:2815964:2816476 [1] NCCL INFO Channel 11/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-168-30:235653:236140 [5] NCCL INFO Channel 14/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-161-153:2636086:2636602 [2] NCCL INFO Channel 03/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-161-153:2636085:2636604 [1] NCCL INFO Channel 02/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-161-178:2815964:2816476 [1] NCCL INFO Channel 12/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-161-153:2636086:2636602 [2] NCCL INFO Channel 04/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-161-153:2636085:2636604 [1] NCCL INFO Channel 03/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-161-178:2815964:2816476 [1] NCCL INFO Channel 13/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-161-153:2636090:2636598 [6] NCCL INFO Channel 06/0 : 15[7] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636090:2636598 [6] NCCL INFO Channel 14/0 : 15[7] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815964:2816476 [1] NCCL INFO Channel 14/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-161-153:2636086:2636602 [2] NCCL INFO Channel 05/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-161-153:2636085:2636604 [1] NCCL INFO Channel 04/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-161-178:2815964:2816476 [1] NCCL INFO Channel 15/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-161-153:2636089:2636600 [5] NCCL INFO Channel 05/0 : 14[6] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636089:2636600 [5] NCCL INFO Channel 13/0 : 14[6] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636090:2636598 [6] NCCL INFO Channel 05/0 : 22[6] -> 29[5] [send] via NET/Libfabric/5(21)/GDRDMA
+ip-26-0-161-153:2636086:2636602 [2] NCCL INFO Channel 06/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-161-153:2636085:2636604 [1] NCCL INFO Channel 05/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-161-153:2636090:2636598 [6] NCCL INFO Channel 13/0 : 22[6] -> 29[5] [send] via NET/Libfabric/5(21)/GDRDMA
+ip-26-0-161-153:2636089:2636600 [5] NCCL INFO Channel 04/0 : 21[5] -> 28[4] [send] via NET/Libfabric/4(20)/GDRDMA
+ip-26-0-161-153:2636089:2636600 [5] NCCL INFO Channel 12/0 : 21[5] -> 28[4] [send] via NET/Libfabric/4(20)/GDRDMA
+ip-26-0-161-178:2815968:2816472 [5] NCCL INFO Channel 05/0 : 29[5] -> 27[3] via P2P/IPC
+ip-26-0-161-153:2636090:2636598 [6] NCCL INFO Channel 06/0 : 22[6] -> 19[3] via P2P/IPC
+ip-26-0-161-142:62887:63379 [6] NCCL INFO Channel 06/0 : 14[6] -> 11[3] via P2P/IPC
+ip-26-0-161-153:2636089:2636600 [5] NCCL INFO Channel 05/0 : 21[5] -> 19[3] via P2P/IPC
+ip-26-0-161-142:62886:63378 [5] NCCL INFO Channel 13/0 : 13[5] -> 11[3] via P2P/IPC
+ip-26-0-161-142:62883:63373 [2] NCCL INFO Channel 02/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-161-153:2636086:2636602 [2] NCCL INFO Channel 07/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-161-153:2636085:2636604 [1] NCCL INFO Channel 06/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-161-153:2636090:2636598 [6] NCCL INFO Channel 14/0 : 22[6] -> 19[3] via P2P/IPC
+ip-26-0-161-153:2636089:2636600 [5] NCCL INFO Channel 13/0 : 21[5] -> 19[3] via P2P/IPC
+ip-26-0-161-153:2636086:2636602 [2] NCCL INFO Channel 10/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-161-153:2636085:2636604 [1] NCCL INFO Channel 07/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-161-142:62883:63373 [2] NCCL INFO Channel 03/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-161-142:62887:63379 [6] NCCL INFO Channel 14/0 : 14[6] -> 11[3] via P2P/IPC
+ip-26-0-161-153:2636086:2636602 [2] NCCL INFO Channel 11/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-161-153:2636085:2636604 [1] NCCL INFO Channel 09/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-168-30:235653:236140 [5] NCCL INFO Channel 15/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-161-153:2636086:2636602 [2] NCCL INFO Channel 12/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-161-153:2636085:2636604 [1] NCCL INFO Channel 10/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-165-59:1557268:1557764 [2] NCCL INFO Channel 07/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-153:2636086:2636602 [2] NCCL INFO Channel 13/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-161-153:2636085:2636604 [1] NCCL INFO Channel 11/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-161-178:2815968:2816472 [5] NCCL INFO Channel 13/0 : 29[5] -> 27[3] via P2P/IPC
+ip-26-0-161-153:2636086:2636602 [2] NCCL INFO Channel 14/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-161-153:2636085:2636604 [1] NCCL INFO Channel 12/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-161-153:2636086:2636602 [2] NCCL INFO Channel 15/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-161-153:2636085:2636604 [1] NCCL INFO Channel 13/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-165-59:1557267:1557761 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-59:1557268:1557764 [2] NCCL INFO Channel 10/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-153:2636085:2636604 [1] NCCL INFO Channel 14/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-161-153:2636085:2636604 [1] NCCL INFO Channel 15/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-165-59:1557267:1557761 [1] NCCL INFO Channel 04/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-59:1557270:1557763 [4] NCCL INFO Channel 01/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-142:62883:63373 [2] NCCL INFO Channel 04/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-165-59:1557268:1557764 [2] NCCL INFO Channel 11/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-165-59:1557267:1557761 [1] NCCL INFO Channel 05/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-59:1557270:1557763 [4] NCCL INFO Channel 02/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-165-59:1557268:1557764 [2] NCCL INFO Channel 12/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-165-59:1557267:1557761 [1] NCCL INFO Channel 06/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-59:1557270:1557763 [4] NCCL INFO Channel 04/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-165-59:1557268:1557764 [2] NCCL INFO Channel 13/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-165-59:1557267:1557761 [1] NCCL INFO Channel 07/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-59:1557270:1557763 [4] NCCL INFO Channel 09/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-165-59:1557268:1557764 [2] NCCL INFO Channel 14/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-165-59:1557267:1557761 [1] NCCL INFO Channel 09/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-59:1557270:1557763 [4] NCCL INFO Channel 10/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-165-59:1557268:1557764 [2] NCCL INFO Channel 15/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-165-59:1557267:1557761 [1] NCCL INFO Channel 10/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-59:1557270:1557763 [4] NCCL INFO Channel 12/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-165-59:1557267:1557761 [1] NCCL INFO Channel 11/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-59:1557267:1557761 [1] NCCL INFO Channel 12/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-59:1557267:1557761 [1] NCCL INFO Channel 13/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-59:1557267:1557761 [1] NCCL INFO Channel 14/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-59:1557267:1557761 [1] NCCL INFO Channel 15/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-142:62883:63373 [2] NCCL INFO Channel 05/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-161-142:62883:63373 [2] NCCL INFO Channel 06/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-161-142:62883:63373 [2] NCCL INFO Channel 07/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-161-142:62883:63373 [2] NCCL INFO Channel 10/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-161-142:62883:63373 [2] NCCL INFO Channel 11/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-161-142:62883:63373 [2] NCCL INFO Channel 12/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-161-142:62883:63373 [2] NCCL INFO Channel 13/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-161-142:62883:63373 [2] NCCL INFO Channel 14/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-161-142:62883:63373 [2] NCCL INFO Channel 15/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-168-52:2722052:2722546 [7] NCCL INFO Channel 07/0 : 20[4] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722052:2722546 [7] NCCL INFO Channel 15/0 : 20[4] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722052:2722546 [7] NCCL INFO Channel 07/0 : 31[7] -> 27[3] via P2P/IPC
+ip-26-0-168-52:2722052:2722546 [7] NCCL INFO Channel 15/0 : 31[7] -> 27[3] via P2P/IPC
+ip-26-0-168-52:2722052:2722546 [7] NCCL INFO Channel 01/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-168-52:2722052:2722546 [7] NCCL INFO Channel 02/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-168-52:2722052:2722546 [7] NCCL INFO Channel 03/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-168-52:2722052:2722546 [7] NCCL INFO Channel 04/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-168-52:2722052:2722546 [7] NCCL INFO Channel 05/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-168-52:2722052:2722546 [7] NCCL INFO Channel 09/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-168-52:2722052:2722546 [7] NCCL INFO Channel 10/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-165-59:1557273:1557760 [7] NCCL INFO Channel 07/0 : 28[4] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557273:1557760 [7] NCCL INFO Channel 15/0 : 28[4] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557273:1557760 [7] NCCL INFO Channel 07/0 : 7[7] -> 3[3] via P2P/IPC
+ip-26-0-168-52:2722052:2722546 [7] NCCL INFO Channel 11/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-168-52:2722052:2722546 [7] NCCL INFO Channel 12/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-168-52:2722052:2722546 [7] NCCL INFO Channel 13/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-165-59:1557273:1557760 [7] NCCL INFO Channel 15/0 : 7[7] -> 3[3] via P2P/IPC
+ip-26-0-161-178:2815970:2816470 [7] NCCL INFO Channel 07/0 : 20[4] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815967:2816473 [4] NCCL INFO Channel 07/0 : 28[4] -> 7[7] [send] via NET/Libfabric/7(31)/GDRDMA
+ip-26-0-161-178:2815970:2816470 [7] NCCL INFO Channel 15/0 : 20[4] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815967:2816473 [4] NCCL INFO Channel 15/0 : 28[4] -> 7[7] [send] via NET/Libfabric/7(31)/GDRDMA
+ip-26-0-168-52:2722048:2722545 [3] NCCL INFO Channel 01/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-168-52:2722048:2722545 [3] NCCL INFO Channel 03/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-165-59:1557273:1557760 [7] NCCL INFO Channel 01/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-168-52:2722051:2722543 [6] NCCL INFO Channel 01/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-168-52:2722048:2722545 [3] NCCL INFO Channel 04/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-168-52:2722045:2722544 [0] NCCL INFO Connected all rings
+ip-26-0-168-52:2722045:2722544 [0] NCCL INFO Channel 02/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-168-52:2722051:2722543 [6] NCCL INFO Channel 02/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-168-52:2722050:2722548 [5] NCCL INFO Channel 01/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-168-52:2722048:2722545 [3] NCCL INFO Channel 05/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-168-52:2722045:2722544 [0] NCCL INFO Channel 03/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-168-52:2722051:2722543 [6] NCCL INFO Channel 03/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-168-52:2722050:2722548 [5] NCCL INFO Channel 02/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-161-138:96570:97093 [7] NCCL INFO Channel 07/0 : 28[4] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722045:2722544 [0] NCCL INFO Channel 04/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-161-138:96567:97090 [4] NCCL INFO Channel 15/0 : 4[4] -> 15[7] [send] via NET/Libfabric/7(7)/GDRDMA
+ip-26-0-168-52:2722049:2722547 [4] NCCL INFO Channel 01/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-161-138:96570:97093 [7] NCCL INFO Channel 15/0 : 28[4] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96570:97093 [7] NCCL INFO Channel 07/0 : 7[7] -> 3[3] via P2P/IPC
+ip-26-0-165-59:1557273:1557760 [7] NCCL INFO Channel 02/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-168-52:2722050:2722548 [5] NCCL INFO Channel 03/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-168-52:2722045:2722544 [0] NCCL INFO Channel 05/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-168-52:2722049:2722547 [4] NCCL INFO Channel 02/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-168-52:2722048:2722545 [3] NCCL INFO Channel 06/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-168-52:2722051:2722543 [6] NCCL INFO Channel 04/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-168-52:2722050:2722548 [5] NCCL INFO Channel 06/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-168-52:2722049:2722547 [4] NCCL INFO Channel 04/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-168-52:2722045:2722544 [0] NCCL INFO Channel 06/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-168-52:2722048:2722545 [3] NCCL INFO Channel 07/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-168-52:2722051:2722543 [6] NCCL INFO Channel 07/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-161-153:2636091:2636599 [7] NCCL INFO Channel 07/0 : 12[4] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636091:2636599 [7] NCCL INFO Channel 15/0 : 12[4] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636088:2636601 [4] NCCL INFO Channel 07/0 : 20[4] -> 31[7] [send] via NET/Libfabric/7(23)/GDRDMA
+ip-26-0-161-153:2636088:2636601 [4] NCCL INFO Channel 15/0 : 20[4] -> 31[7] [send] via NET/Libfabric/7(23)/GDRDMA
+ip-26-0-168-52:2722050:2722548 [5] NCCL INFO Channel 07/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-161-153:2636091:2636599 [7] NCCL INFO Channel 07/0 : 23[7] -> 19[3] via P2P/IPC
+ip-26-0-168-52:2722049:2722547 [4] NCCL INFO Channel 09/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-161-178:2815970:2816470 [7] NCCL INFO Channel 07/0 : 31[7] -> 27[3] via P2P/IPC
+ip-26-0-168-52:2722045:2722544 [0] NCCL INFO Channel 07/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-168-52:2722048:2722545 [3] NCCL INFO Channel 09/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-168-52:2722051:2722543 [6] NCCL INFO Channel 09/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-168-52:2722050:2722548 [5] NCCL INFO Channel 09/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-168-52:2722049:2722547 [4] NCCL INFO Channel 10/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-168-52:2722045:2722544 [0] NCCL INFO Channel 10/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-168-52:2722048:2722545 [3] NCCL INFO Channel 11/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-168-52:2722051:2722543 [6] NCCL INFO Channel 10/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-165-59:1557273:1557760 [7] NCCL INFO Channel 03/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-168-52:2722050:2722548 [5] NCCL INFO Channel 10/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-168-52:2722049:2722547 [4] NCCL INFO Channel 12/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-168-52:2722045:2722544 [0] NCCL INFO Channel 11/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-168-52:2722048:2722545 [3] NCCL INFO Channel 12/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-168-52:2722051:2722543 [6] NCCL INFO Channel 11/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-165-59:1557273:1557760 [7] NCCL INFO Channel 04/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-168-52:2722048:2722545 [3] NCCL INFO Channel 13/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-168-52:2722045:2722544 [0] NCCL INFO Channel 12/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-168-52:2722051:2722543 [6] NCCL INFO Channel 12/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-168-52:2722050:2722548 [5] NCCL INFO Channel 11/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-161-178:2815970:2816470 [7] NCCL INFO Channel 15/0 : 31[7] -> 27[3] via P2P/IPC
+ip-26-0-161-153:2636091:2636599 [7] NCCL INFO Channel 15/0 : 23[7] -> 19[3] via P2P/IPC
+ip-26-0-168-52:2722048:2722545 [3] NCCL INFO Channel 14/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-168-52:2722045:2722544 [0] NCCL INFO Channel 13/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-168-52:2722051:2722543 [6] NCCL INFO Channel 15/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-168-52:2722050:2722548 [5] NCCL INFO Channel 14/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-161-142:62888:63380 [7] NCCL INFO Channel 07/0 : 4[4] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62888:63380 [7] NCCL INFO Channel 15/0 : 4[4] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62888:63380 [7] NCCL INFO Channel 07/0 : 15[7] -> 11[3] via P2P/IPC
+ip-26-0-168-52:2722045:2722544 [0] NCCL INFO Channel 14/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-168-52:2722048:2722545 [3] NCCL INFO Channel 15/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-165-59:1557273:1557760 [7] NCCL INFO Channel 05/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-168-52:2722050:2722548 [5] NCCL INFO Channel 15/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-161-138:96570:97093 [7] NCCL INFO Channel 15/0 : 7[7] -> 3[3] via P2P/IPC
+ip-26-0-161-142:62888:63380 [7] NCCL INFO Channel 15/0 : 15[7] -> 11[3] via P2P/IPC
+ip-26-0-161-138:96570:97093 [7] NCCL INFO Channel 01/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-153:2636091:2636599 [7] NCCL INFO Channel 01/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-161-178:2815970:2816470 [7] NCCL INFO Channel 01/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-168-52:2722045:2722544 [0] NCCL INFO Channel 15/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-161-142:62888:63380 [7] NCCL INFO Channel 01/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-165-59:1557273:1557760 [7] NCCL INFO Channel 09/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-142:62888:63380 [7] NCCL INFO Channel 02/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-161-138:96570:97093 [7] NCCL INFO Channel 02/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-178:2815970:2816470 [7] NCCL INFO Channel 02/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-161-153:2636091:2636599 [7] NCCL INFO Channel 02/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-165-59:1557273:1557760 [7] NCCL INFO Channel 10/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-142:62888:63380 [7] NCCL INFO Channel 03/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-161-138:96570:97093 [7] NCCL INFO Channel 03/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-178:2815970:2816470 [7] NCCL INFO Channel 03/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-161-153:2636091:2636599 [7] NCCL INFO Channel 03/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-161-178:2815970:2816470 [7] NCCL INFO Channel 04/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-161-142:62888:63380 [7] NCCL INFO Channel 04/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-161-178:2815970:2816470 [7] NCCL INFO Channel 05/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-161-142:62888:63380 [7] NCCL INFO Channel 05/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-161-142:62885:63377 [4] NCCL INFO Channel 01/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-165-59:1557273:1557760 [7] NCCL INFO Channel 11/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-178:2815970:2816470 [7] NCCL INFO Channel 09/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-161-142:62888:63380 [7] NCCL INFO Channel 09/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-161-142:62885:63377 [4] NCCL INFO Channel 02/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-161-153:2636091:2636599 [7] NCCL INFO Channel 04/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-161-138:96570:97093 [7] NCCL INFO Channel 04/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-178:2815970:2816470 [7] NCCL INFO Channel 10/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-161-142:62888:63380 [7] NCCL INFO Channel 10/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-161-142:62885:63377 [4] NCCL INFO Channel 04/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-161-178:2815967:2816473 [4] NCCL INFO Channel 01/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-161-178:2815970:2816470 [7] NCCL INFO Channel 11/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-161-142:62888:63380 [7] NCCL INFO Channel 11/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-161-142:62885:63377 [4] NCCL INFO Channel 09/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-161-178:2815967:2816473 [4] NCCL INFO Channel 02/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-161-142:62888:63380 [7] NCCL INFO Channel 12/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-161-178:2815970:2816470 [7] NCCL INFO Channel 12/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-161-142:62885:63377 [4] NCCL INFO Channel 10/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-161-178:2815967:2816473 [4] NCCL INFO Channel 04/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-161-138:96570:97093 [7] NCCL INFO Channel 05/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-153:2636091:2636599 [7] NCCL INFO Channel 05/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-161-142:62888:63380 [7] NCCL INFO Channel 13/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-161-178:2815970:2816470 [7] NCCL INFO Channel 13/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-161-142:62885:63377 [4] NCCL INFO Channel 12/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-161-178:2815967:2816473 [4] NCCL INFO Channel 09/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-165-59:1557273:1557760 [7] NCCL INFO Channel 12/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-142:62884:63376 [3] NCCL INFO Channel 01/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-161-178:2815967:2816473 [4] NCCL INFO Channel 10/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-161-178:2815966:2816474 [3] NCCL INFO Channel 01/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-161-142:62884:63376 [3] NCCL INFO Channel 03/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-161-178:2815967:2816473 [4] NCCL INFO Channel 12/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-161-178:2815966:2816474 [3] NCCL INFO Channel 03/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-161-153:2636091:2636599 [7] NCCL INFO Channel 09/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-161-142:62884:63376 [3] NCCL INFO Channel 04/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-161-138:96570:97093 [7] NCCL INFO Channel 09/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-178:2815966:2816474 [3] NCCL INFO Channel 04/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-161-142:62887:63379 [6] NCCL INFO Channel 01/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-161-142:62884:63376 [3] NCCL INFO Channel 05/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-161-178:2815969:2816471 [6] NCCL INFO Channel 01/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-161-153:2636091:2636599 [7] NCCL INFO Channel 10/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-165-59:1557273:1557760 [7] NCCL INFO Channel 13/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-178:2815966:2816474 [3] NCCL INFO Channel 05/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-161-142:62884:63376 [3] NCCL INFO Channel 06/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-161-178:2815969:2816471 [6] NCCL INFO Channel 02/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-161-142:62887:63379 [6] NCCL INFO Channel 02/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-161-142:62886:63378 [5] NCCL INFO Channel 01/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-161-153:2636091:2636599 [7] NCCL INFO Channel 11/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-161-178:2815968:2816472 [5] NCCL INFO Channel 01/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-161-178:2815966:2816474 [3] NCCL INFO Channel 06/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-161-178:2815969:2816471 [6] NCCL INFO Channel 03/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-161-138:96570:97093 [7] NCCL INFO Channel 10/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-165-59:1557269:1557765 [3] NCCL INFO Channel 01/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-142:62884:63376 [3] NCCL INFO Channel 07/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-161-178:2815966:2816474 [3] NCCL INFO Channel 07/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-161-178:2815968:2816472 [5] NCCL INFO Channel 02/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-161-178:2815969:2816471 [6] NCCL INFO Channel 04/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-161-138:96570:97093 [7] NCCL INFO Channel 11/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-153:2636091:2636599 [7] NCCL INFO Channel 12/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-161-78:114512:115007 [0] NCCL INFO Connected all rings
+ip-26-0-161-153:2636088:2636601 [4] NCCL INFO Channel 01/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-161-78:114512:115007 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-142:62884:63376 [3] NCCL INFO Channel 09/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-165-59:1557269:1557765 [3] NCCL INFO Channel 03/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-165-59:1557272:1557766 [6] NCCL INFO Channel 01/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-178:2815968:2816472 [5] NCCL INFO Channel 03/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-161-178:2815966:2816474 [3] NCCL INFO Channel 09/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-161-178:2815969:2816471 [6] NCCL INFO Channel 07/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-161-142:62887:63379 [6] NCCL INFO Channel 03/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-161-142:62884:63376 [3] NCCL INFO Channel 11/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-165-59:1557269:1557765 [3] NCCL INFO Channel 04/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-142:62886:63378 [5] NCCL INFO Channel 02/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-165-59:1557272:1557766 [6] NCCL INFO Channel 02/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-153:2636091:2636599 [7] NCCL INFO Channel 13/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-161-153:2636088:2636601 [4] NCCL INFO Channel 02/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-161-178:2815968:2816472 [5] NCCL INFO Channel 06/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-161-178:2815966:2816474 [3] NCCL INFO Channel 11/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-161-178:2815969:2816471 [6] NCCL INFO Channel 09/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-165-59:1557269:1557765 [3] NCCL INFO Channel 05/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-142:62887:63379 [6] NCCL INFO Channel 04/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-161-142:62884:63376 [3] NCCL INFO Channel 12/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-165-59:1557272:1557766 [6] NCCL INFO Channel 03/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-142:62886:63378 [5] NCCL INFO Channel 03/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-165-59:1557271:1557762 [5] NCCL INFO Channel 01/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-59:1557269:1557765 [3] NCCL INFO Channel 06/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-153:2636088:2636601 [4] NCCL INFO Channel 04/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-161-153:2636087:2636603 [3] NCCL INFO Channel 01/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-161-78:114512:115007 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-142:62887:63379 [6] NCCL INFO Channel 07/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-161-142:62884:63376 [3] NCCL INFO Channel 13/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-161-142:62886:63378 [5] NCCL INFO Channel 06/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-165-59:1557272:1557766 [6] NCCL INFO Channel 04/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-138:96570:97093 [7] NCCL INFO Channel 12/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-165-59:1557271:1557762 [5] NCCL INFO Channel 02/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-59:1557269:1557765 [3] NCCL INFO Channel 07/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-165-59:1557272:1557766 [6] NCCL INFO Channel 07/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-142:62887:63379 [6] NCCL INFO Channel 09/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-161-142:62884:63376 [3] NCCL INFO Channel 14/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-161-142:62886:63378 [5] NCCL INFO Channel 07/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-161-153:2636088:2636601 [4] NCCL INFO Channel 09/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-161-153:2636087:2636603 [3] NCCL INFO Channel 03/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-161-178:2815968:2816472 [5] NCCL INFO Channel 07/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-161-178:2815969:2816471 [6] NCCL INFO Channel 10/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-161-178:2815966:2816474 [3] NCCL INFO Channel 12/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-165-59:1557269:1557765 [3] NCCL INFO Channel 09/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-165-59:1557271:1557762 [5] NCCL INFO Channel 03/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-142:62887:63379 [6] NCCL INFO Channel 10/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-161-142:62884:63376 [3] NCCL INFO Channel 15/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-165-59:1557272:1557766 [6] NCCL INFO Channel 09/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-142:62886:63378 [5] NCCL INFO Channel 09/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-161-138:96570:97093 [7] NCCL INFO Channel 13/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-138:96567:97090 [4] NCCL INFO Channel 01/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-178:2815968:2816472 [5] NCCL INFO Channel 09/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-161-178:2815969:2816471 [6] NCCL INFO Channel 11/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-161-178:2815966:2816474 [3] NCCL INFO Channel 13/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-161-153:2636090:2636598 [6] NCCL INFO Channel 01/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-161-142:62887:63379 [6] NCCL INFO Channel 11/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-165-59:1557269:1557765 [3] NCCL INFO Channel 11/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-142:62886:63378 [5] NCCL INFO Channel 10/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-165-59:1557271:1557762 [5] NCCL INFO Channel 06/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-59:1557272:1557766 [6] NCCL INFO Channel 10/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-153:2636090:2636598 [6] NCCL INFO Channel 02/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-161-153:2636088:2636601 [4] NCCL INFO Channel 10/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-161-78:114512:115007 [0] NCCL INFO Channel 04/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-178:2815968:2816472 [5] NCCL INFO Channel 10/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-161-178:2815969:2816471 [6] NCCL INFO Channel 12/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-161-178:2815966:2816474 [3] NCCL INFO Channel 14/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-161-153:2636090:2636598 [6] NCCL INFO Channel 03/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-165-59:1557269:1557765 [3] NCCL INFO Channel 12/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-165-59:1557271:1557762 [5] NCCL INFO Channel 07/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-59:1557272:1557766 [6] NCCL INFO Channel 11/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-138:96567:97090 [4] NCCL INFO Channel 02/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-153:2636088:2636601 [4] NCCL INFO Channel 12/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-161-153:2636090:2636598 [6] NCCL INFO Channel 04/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-161-178:2815969:2816471 [6] NCCL INFO Channel 15/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-161-178:2815968:2816472 [5] NCCL INFO Channel 11/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-161-178:2815966:2816474 [3] NCCL INFO Channel 15/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-168-34:272154:272650 [4] NCCL INFO Channel 01/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-161-138:96567:97090 [4] NCCL INFO Channel 04/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-165-59:1557271:1557762 [5] NCCL INFO Channel 09/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-153:2636090:2636598 [6] NCCL INFO Channel 07/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-165-59:1557269:1557765 [3] NCCL INFO Channel 13/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-165-59:1557272:1557766 [6] NCCL INFO Channel 12/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-153:2636090:2636598 [6] NCCL INFO Channel 09/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-161-153:2636087:2636603 [3] NCCL INFO Channel 04/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-161-178:2815968:2816472 [5] NCCL INFO Channel 14/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-161-153:2636089:2636600 [5] NCCL INFO Channel 01/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-165-59:1557269:1557765 [3] NCCL INFO Channel 14/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-165-59:1557271:1557762 [5] NCCL INFO Channel 10/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-59:1557272:1557766 [6] NCCL INFO Channel 15/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-153:2636090:2636598 [6] NCCL INFO Channel 10/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-161-78:114512:115007 [0] NCCL INFO Channel 05/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-153:2636087:2636603 [3] NCCL INFO Channel 05/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-161-138:96567:97090 [4] NCCL INFO Channel 09/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-153:2636089:2636600 [5] NCCL INFO Channel 02/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-165-59:1557269:1557765 [3] NCCL INFO Channel 15/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-138:96566:97087 [3] NCCL INFO Channel 01/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-138:96567:97090 [4] NCCL INFO Channel 10/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-165-59:1557271:1557762 [5] NCCL INFO Channel 11/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-153:2636090:2636598 [6] NCCL INFO Channel 11/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-161-153:2636087:2636603 [3] NCCL INFO Channel 06/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-165-202:153562:154061 [0] NCCL INFO Connected all rings
+ip-26-0-165-202:153562:154061 [0] NCCL INFO Channel 02/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-161-153:2636089:2636600 [5] NCCL INFO Channel 03/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-161-138:96566:97087 [3] NCCL INFO Channel 03/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-138:96567:97090 [4] NCCL INFO Channel 12/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-153:2636090:2636598 [6] NCCL INFO Channel 12/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-161-153:2636087:2636603 [3] NCCL INFO Channel 07/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-168-34:272154:272650 [4] NCCL INFO Channel 02/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-161-153:2636089:2636600 [5] NCCL INFO Channel 06/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-161-138:96566:97087 [3] NCCL INFO Channel 04/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-165-59:1557271:1557762 [5] NCCL INFO Channel 14/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-78:114512:115007 [0] NCCL INFO Channel 06/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-153:2636090:2636598 [6] NCCL INFO Channel 15/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-161-153:2636087:2636603 [3] NCCL INFO Channel 09/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-161-153:2636089:2636600 [5] NCCL INFO Channel 07/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-161-78:114512:115007 [0] NCCL INFO Channel 07/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-138:96569:97092 [6] NCCL INFO Channel 01/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-142:62887:63379 [6] NCCL INFO Channel 12/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-161-153:2636087:2636603 [3] NCCL INFO Channel 11/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-161-153:2636089:2636600 [5] NCCL INFO Channel 09/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-161-138:96566:97087 [3] NCCL INFO Channel 05/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-142:62887:63379 [6] NCCL INFO Channel 15/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-161-138:96569:97092 [6] NCCL INFO Channel 02/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-142:62886:63378 [5] NCCL INFO Channel 11/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-161-78:114512:115007 [0] NCCL INFO Channel 10/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-138:96566:97087 [3] NCCL INFO Channel 06/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-138:96569:97092 [6] NCCL INFO Channel 03/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-78:114512:115007 [0] NCCL INFO Channel 11/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-138:96566:97087 [3] NCCL INFO Channel 07/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-138:96568:97091 [5] NCCL INFO Channel 01/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-138:96569:97092 [6] NCCL INFO Channel 04/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-178:2815968:2816472 [5] NCCL INFO Channel 15/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-161-153:2636089:2636600 [5] NCCL INFO Channel 10/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-161-153:2636087:2636603 [3] NCCL INFO Channel 12/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-161-142:62886:63378 [5] NCCL INFO Channel 14/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-161-138:96566:97087 [3] NCCL INFO Channel 09/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-138:96568:97091 [5] NCCL INFO Channel 02/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-138:96569:97092 [6] NCCL INFO Channel 07/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-168-34:272154:272650 [4] NCCL INFO Channel 04/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-161-78:114512:115007 [0] NCCL INFO Channel 12/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-59:1557271:1557762 [5] NCCL INFO Channel 15/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-138:96566:97087 [3] NCCL INFO Channel 11/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-138:96568:97091 [5] NCCL INFO Channel 03/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-138:96569:97092 [6] NCCL INFO Channel 09/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-202:153563:154060 [1] NCCL INFO Connected all rings
+ip-26-0-161-138:96566:97087 [3] NCCL INFO Channel 12/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-138:96568:97091 [5] NCCL INFO Channel 06/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-138:96569:97092 [6] NCCL INFO Channel 10/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-78:114512:115007 [0] NCCL INFO Channel 13/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-138:96566:97087 [3] NCCL INFO Channel 13/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-138:96569:97092 [6] NCCL INFO Channel 11/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-153:2636087:2636603 [3] NCCL INFO Channel 13/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-161-78:114512:115007 [0] NCCL INFO Channel 14/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-138:96566:97087 [3] NCCL INFO Channel 14/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-138:96569:97092 [6] NCCL INFO Channel 12/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-168-34:272154:272650 [4] NCCL INFO Channel 09/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-161-138:96566:97087 [3] NCCL INFO Channel 15/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-138:96569:97092 [6] NCCL INFO Channel 15/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-138:96568:97091 [5] NCCL INFO Channel 07/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-78:114512:115007 [0] NCCL INFO Channel 15/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-138:96568:97091 [5] NCCL INFO Channel 09/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-142:62886:63378 [5] NCCL INFO Channel 15/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-161-153:2636087:2636603 [3] NCCL INFO Channel 14/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-161-153:2636089:2636600 [5] NCCL INFO Channel 11/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-168-34:272154:272650 [4] NCCL INFO Channel 10/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-168-34:272154:272650 [4] NCCL INFO Channel 12/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-161-153:2636087:2636603 [3] NCCL INFO Channel 15/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-161-153:2636089:2636600 [5] NCCL INFO Channel 14/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-161-138:96568:97091 [5] NCCL INFO Channel 10/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-153:2636089:2636600 [5] NCCL INFO Channel 15/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-161-138:96568:97091 [5] NCCL INFO Channel 11/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-138:96568:97091 [5] NCCL INFO Channel 14/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-138:96568:97091 [5] NCCL INFO Channel 15/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-202:153568:154056 [6] NCCL INFO Connected all rings
+ip-26-0-161-78:114513:115008 [1] NCCL INFO Connected all rings
+ip-26-0-165-131:1630206:1630728 [1] NCCL INFO Connected all rings
+ip-26-0-161-78:114517:115013 [5] NCCL INFO Connected all rings
+ip-26-0-161-78:114514:115011 [2] NCCL INFO Connected all rings
+ip-26-0-165-164:1227465:1227962 [2] NCCL INFO Connected all rings
+ip-26-0-161-78:114513:115008 [1] NCCL INFO Channel 01/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-78:114513:115008 [1] NCCL INFO Channel 03/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-78:114513:115008 [1] NCCL INFO Channel 04/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-78:114513:115008 [1] NCCL INFO Channel 05/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-78:114513:115008 [1] NCCL INFO Channel 06/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-202:153562:154061 [0] NCCL INFO Channel 03/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-161-78:114513:115008 [1] NCCL INFO Channel 07/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-78:114513:115008 [1] NCCL INFO Channel 09/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-78:114515:115009 [3] NCCL INFO Connected all rings
+ip-26-0-161-78:114513:115008 [1] NCCL INFO Channel 11/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-78:114518:115012 [6] NCCL INFO Connected all rings
+ip-26-0-161-78:114513:115008 [1] NCCL INFO Channel 12/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-202:153562:154061 [0] NCCL INFO Channel 04/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-161-78:114514:115011 [2] NCCL INFO Channel 01/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-78:114513:115008 [1] NCCL INFO Channel 13/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-78:114517:115013 [5] NCCL INFO Channel 01/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-165-202:153564:154057 [2] NCCL INFO Connected all rings
+ip-26-0-161-78:114513:115008 [1] NCCL INFO Channel 14/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-78:114514:115011 [2] NCCL INFO Channel 02/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-78:114517:115013 [5] NCCL INFO Channel 02/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-78:114513:115008 [1] NCCL INFO Channel 15/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-78:114517:115013 [5] NCCL INFO Channel 03/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-78:114514:115011 [2] NCCL INFO Channel 04/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-202:153562:154061 [0] NCCL INFO Channel 05/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-165-164:1227463:1227964 [0] NCCL INFO Connected all rings
+ip-26-0-165-164:1227463:1227964 [0] NCCL INFO Channel 02/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-161-78:114517:115013 [5] NCCL INFO Channel 04/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-78:114514:115011 [2] NCCL INFO Channel 05/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-78:114512:115007 [0] NCCL INFO Channel 05/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-165-202:153562:154061 [0] NCCL INFO Channel 06/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-165-164:1227463:1227964 [0] NCCL INFO Channel 03/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-161-78:114517:115013 [5] NCCL INFO Channel 05/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-78:114514:115011 [2] NCCL INFO Channel 06/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-202:153562:154061 [0] NCCL INFO Channel 07/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-161-78:114517:115013 [5] NCCL INFO Channel 07/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-165-164:1227463:1227964 [0] NCCL INFO Channel 04/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-161-78:114514:115011 [2] NCCL INFO Channel 07/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-78:114517:115013 [5] NCCL INFO Channel 09/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-78:114514:115011 [2] NCCL INFO Channel 09/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-164:1227463:1227964 [0] NCCL INFO Channel 05/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-165-202:153562:154061 [0] NCCL INFO Channel 10/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-165-131:1630208:1630729 [3] NCCL INFO Connected all rings
+ip-26-0-161-78:114517:115013 [5] NCCL INFO Channel 10/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-78:114514:115011 [2] NCCL INFO Channel 10/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-164:1227463:1227964 [0] NCCL INFO Channel 06/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-165-202:153562:154061 [0] NCCL INFO Channel 11/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-161-78:114517:115013 [5] NCCL INFO Channel 11/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-78:114514:115011 [2] NCCL INFO Channel 12/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-164:1227463:1227964 [0] NCCL INFO Channel 07/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-165-202:153562:154061 [0] NCCL INFO Channel 12/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-161-78:114515:115009 [3] NCCL INFO Channel 01/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-78:114517:115013 [5] NCCL INFO Channel 12/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-78:114514:115011 [2] NCCL INFO Channel 13/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-78:114515:115009 [3] NCCL INFO Channel 02/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-202:153562:154061 [0] NCCL INFO Channel 13/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-161-78:114517:115013 [5] NCCL INFO Channel 13/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-78:114514:115011 [2] NCCL INFO Channel 14/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-164:1227463:1227964 [0] NCCL INFO Channel 10/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-165-202:153567:154059 [5] NCCL INFO Connected all rings
+ip-26-0-165-202:153562:154061 [0] NCCL INFO Channel 14/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-161-78:114515:115009 [3] NCCL INFO Channel 03/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-78:114517:115013 [5] NCCL INFO Channel 15/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-78:114514:115011 [2] NCCL INFO Channel 15/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-202:153562:154061 [0] NCCL INFO Channel 15/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-165-164:1227463:1227964 [0] NCCL INFO Channel 11/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-161-78:114515:115009 [3] NCCL INFO Channel 05/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-78:114513:115008 [1] NCCL INFO Channel 09/0 : 1[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227463:1227964 [0] NCCL INFO Channel 12/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-161-78:114515:115009 [3] NCCL INFO Channel 06/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-78:114515:115009 [3] NCCL INFO Channel 07/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-78:114515:115009 [3] NCCL INFO Channel 09/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-164:1227463:1227964 [0] NCCL INFO Channel 13/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-165-202:153565:154058 [3] NCCL INFO Connected all rings
+ip-26-0-165-164:1227463:1227964 [0] NCCL INFO Channel 14/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-165-164:1227463:1227964 [0] NCCL INFO Channel 15/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-161-78:114515:115009 [3] NCCL INFO Channel 10/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-78:114512:115007 [0] NCCL INFO Channel 06/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-161-78:114515:115009 [3] NCCL INFO Channel 11/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-78:114512:115007 [0] NCCL INFO Channel 07/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-161-78:114515:115009 [3] NCCL INFO Channel 13/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-78:114512:115007 [0] NCCL INFO Channel 13/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-165-202:153563:154060 [1] NCCL INFO Channel 01/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-165-202:153563:154060 [1] NCCL INFO Channel 03/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-161-78:114515:115009 [3] NCCL INFO Channel 14/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-78:114512:115007 [0] NCCL INFO Channel 14/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-161-123:32705:33203 [1] NCCL INFO Connected all rings
+ip-26-0-165-164:1227470:1227958 [7] NCCL INFO Connected all rings
+ip-26-0-165-202:153569:154055 [7] NCCL INFO Connected all rings
+ip-26-0-161-78:114515:115009 [3] NCCL INFO Channel 15/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-78:114512:115007 [0] NCCL INFO Channel 15/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-165-202:153563:154060 [1] NCCL INFO Channel 04/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-161-78:114514:115011 [2] NCCL INFO Channel 10/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227468:1227959 [5] NCCL INFO Connected all rings
+ip-26-0-165-202:153563:154060 [1] NCCL INFO Channel 05/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-165-164:1227469:1227957 [6] NCCL INFO Connected all rings
+ip-26-0-165-164:1227467:1227960 [4] NCCL INFO Connected all rings
+ip-26-0-165-202:153563:154060 [1] NCCL INFO Channel 06/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-165-202:153563:154060 [1] NCCL INFO Channel 07/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-165-202:153563:154060 [1] NCCL INFO Channel 09/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-165-202:153567:154059 [5] NCCL INFO Channel 01/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-165-202:153563:154060 [1] NCCL INFO Channel 11/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-165-202:153564:154057 [2] NCCL INFO Channel 01/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-165-202:153567:154059 [5] NCCL INFO Channel 02/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-165-202:153563:154060 [1] NCCL INFO Channel 12/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-165-202:153564:154057 [2] NCCL INFO Channel 02/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-165-202:153567:154059 [5] NCCL INFO Channel 03/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-165-202:153564:154057 [2] NCCL INFO Channel 04/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-165-202:153563:154060 [1] NCCL INFO Channel 13/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-165-202:153567:154059 [5] NCCL INFO Channel 04/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-161-78:114519:115014 [7] NCCL INFO Connected all rings
+ip-26-0-165-202:153564:154057 [2] NCCL INFO Channel 05/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-165-202:153563:154060 [1] NCCL INFO Channel 14/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-165-202:153567:154059 [5] NCCL INFO Channel 05/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-165-202:153564:154057 [2] NCCL INFO Channel 06/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-165-202:153563:154060 [1] NCCL INFO Channel 15/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-165-164:1227468:1227959 [5] NCCL INFO Channel 01/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-165-202:153566:154062 [4] NCCL INFO Connected all rings
+ip-26-0-165-202:153564:154057 [2] NCCL INFO Channel 07/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-161-78:114516:115010 [4] NCCL INFO Connected all rings
+ip-26-0-165-202:153567:154059 [5] NCCL INFO Channel 07/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-165-164:1227468:1227959 [5] NCCL INFO Channel 02/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-165-202:153562:154061 [0] NCCL INFO Channel 05/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-165-164:1227469:1227957 [6] NCCL INFO Channel 01/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-165-164:1227468:1227959 [5] NCCL INFO Channel 03/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-165-202:153562:154061 [0] NCCL INFO Channel 06/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-161-78:114518:115012 [6] NCCL INFO Channel 01/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-164:1227469:1227957 [6] NCCL INFO Channel 02/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-165-164:1227464:1227963 [1] NCCL INFO Connected all rings
+ip-26-0-161-78:114518:115012 [6] NCCL INFO Channel 02/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-202:153562:154061 [0] NCCL INFO Channel 07/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-165-164:1227468:1227959 [5] NCCL INFO Channel 04/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-165-164:1227469:1227957 [6] NCCL INFO Channel 03/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-161-78:114518:115012 [6] NCCL INFO Channel 03/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-202:153564:154057 [2] NCCL INFO Channel 09/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-161-78:114518:115012 [6] NCCL INFO Channel 04/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-202:153562:154061 [0] NCCL INFO Channel 13/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-165-164:1227468:1227959 [5] NCCL INFO Channel 05/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-165-164:1227469:1227957 [6] NCCL INFO Channel 04/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-161-78:114518:115012 [6] NCCL INFO Channel 05/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-202:153564:154057 [2] NCCL INFO Channel 10/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-165-131:1630205:1630727 [0] NCCL INFO Connected all rings
+ip-26-0-165-131:1630205:1630727 [0] NCCL INFO Channel 02/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-161-78:114518:115012 [6] NCCL INFO Channel 06/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-202:153562:154061 [0] NCCL INFO Channel 14/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-161-103:96891:97385 [3] NCCL INFO Connected all rings
+ip-26-0-165-202:153568:154056 [6] NCCL INFO Channel 01/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-165-164:1227468:1227959 [5] NCCL INFO Channel 07/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-165-164:1227469:1227957 [6] NCCL INFO Channel 05/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-165-202:153564:154057 [2] NCCL INFO Channel 12/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-165-164:1227467:1227960 [4] NCCL INFO Channel 01/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-160-242:1364903:1365409 [1] NCCL INFO Connected all rings
+ip-26-0-161-78:114518:115012 [6] NCCL INFO Channel 09/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-202:153567:154059 [5] NCCL INFO Channel 09/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-161-78:114518:115012 [6] NCCL INFO Channel 10/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-164:1227465:1227962 [2] NCCL INFO Channel 01/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-165-202:153562:154061 [0] NCCL INFO Channel 15/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-165-202:153568:154056 [6] NCCL INFO Channel 02/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-165-131:1630205:1630727 [0] NCCL INFO Channel 03/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-165-202:153564:154057 [2] NCCL INFO Channel 13/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-165-164:1227468:1227959 [5] NCCL INFO Channel 09/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-165-164:1227469:1227957 [6] NCCL INFO Channel 06/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-161-78:114518:115012 [6] NCCL INFO Channel 11/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-164:1227467:1227960 [4] NCCL INFO Channel 02/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-161-78:114516:115010 [4] NCCL INFO Channel 01/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-78:114518:115012 [6] NCCL INFO Channel 12/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-131:1630205:1630727 [0] NCCL INFO Channel 04/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-165-202:153567:154059 [5] NCCL INFO Channel 10/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-165-131:1630207:1630731 [2] NCCL INFO Connected all rings
+ip-26-0-161-78:114516:115010 [4] NCCL INFO Channel 02/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-164:1227465:1227962 [2] NCCL INFO Channel 02/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-161-78:114518:115012 [6] NCCL INFO Channel 13/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-202:153568:154056 [6] NCCL INFO Channel 03/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-165-202:153564:154057 [2] NCCL INFO Channel 14/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-165-164:1227468:1227959 [5] NCCL INFO Channel 10/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-165-164:1227469:1227957 [6] NCCL INFO Channel 09/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-165-164:1227467:1227960 [4] NCCL INFO Channel 03/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-161-78:114516:115010 [4] NCCL INFO Channel 03/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-131:1630205:1630727 [0] NCCL INFO Channel 05/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-161-78:114518:115012 [6] NCCL INFO Channel 14/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-202:153567:154059 [5] NCCL INFO Channel 11/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-161-103:96894:97381 [6] NCCL INFO Connected all rings
+ip-26-0-165-202:153564:154057 [2] NCCL INFO Channel 15/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-165-202:153568:154056 [6] NCCL INFO Channel 04/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-161-123:32704:33197 [0] NCCL INFO Connected all rings
+ip-26-0-161-123:32704:33197 [0] NCCL INFO Channel 02/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-161-78:114516:115010 [4] NCCL INFO Channel 04/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-164:1227465:1227962 [2] NCCL INFO Channel 04/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-165-164:1227468:1227959 [5] NCCL INFO Channel 11/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-165-164:1227469:1227957 [6] NCCL INFO Channel 10/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-161-78:114516:115010 [4] NCCL INFO Channel 06/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-202:153567:154059 [5] NCCL INFO Channel 12/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-161-78:114518:115012 [6] NCCL INFO Channel 14/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153568:154056 [6] NCCL INFO Channel 05/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-161-78:114516:115010 [4] NCCL INFO Channel 07/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-164:1227465:1227962 [2] NCCL INFO Channel 05/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-165-202:153563:154060 [1] NCCL INFO Channel 01/0 : 17[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153567:154059 [5] NCCL INFO Channel 13/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-165-164:1227468:1227959 [5] NCCL INFO Channel 12/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-165-164:1227469:1227957 [6] NCCL INFO Channel 11/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-161-78:114516:115010 [4] NCCL INFO Channel 09/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-202:153568:154056 [6] NCCL INFO Channel 06/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-161-78:114516:115010 [4] NCCL INFO Channel 10/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-131:1630205:1630727 [0] NCCL INFO Channel 06/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-165-202:153567:154059 [5] NCCL INFO Channel 15/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-165-164:1227465:1227962 [2] NCCL INFO Channel 06/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-165-164:1227468:1227959 [5] NCCL INFO Channel 13/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-165-164:1227469:1227957 [6] NCCL INFO Channel 12/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-161-78:114516:115010 [4] NCCL INFO Channel 11/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-202:153568:154056 [6] NCCL INFO Channel 09/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-161-78:114516:115010 [4] NCCL INFO Channel 12/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-202:153565:154058 [3] NCCL INFO Channel 01/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-165-202:153568:154056 [6] NCCL INFO Channel 10/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-161-78:114516:115010 [4] NCCL INFO Channel 14/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-202:153566:154062 [4] NCCL INFO Channel 01/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-161-78:114512:115007 [0] NCCL INFO Channel 08/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114519:115014 [7] NCCL INFO Channel 15/0 : 7[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227468:1227959 [5] NCCL INFO Channel 15/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-161-78:114516:115010 [4] NCCL INFO Channel 15/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-164:1227467:1227960 [4] NCCL INFO Channel 04/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-165-164:1227465:1227962 [2] NCCL INFO Channel 07/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-165-164:1227469:1227957 [6] NCCL INFO Channel 13/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-165-202:153565:154058 [3] NCCL INFO Channel 02/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-165-202:153568:154056 [6] NCCL INFO Channel 11/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-165-202:153566:154062 [4] NCCL INFO Channel 02/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-161-78:114515:115009 [3] NCCL INFO Channel 11/0 : 3[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114517:115013 [5] NCCL INFO Channel 13/0 : 5[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114516:115010 [4] NCCL INFO Channel 12/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153565:154058 [3] NCCL INFO Channel 03/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-165-202:153568:154056 [6] NCCL INFO Channel 12/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-165-202:153566:154062 [4] NCCL INFO Channel 03/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-165-164:1227465:1227962 [2] NCCL INFO Channel 09/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-165-164:1227469:1227957 [6] NCCL INFO Channel 14/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-165-164:1227467:1227960 [4] NCCL INFO Channel 06/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-165-202:153565:154058 [3] NCCL INFO Channel 05/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-165-202:153568:154056 [6] NCCL INFO Channel 13/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-165-202:153566:154062 [4] NCCL INFO Channel 04/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-165-164:1227465:1227962 [2] NCCL INFO Channel 10/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-165-202:153565:154058 [3] NCCL INFO Channel 06/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-165-202:153568:154056 [6] NCCL INFO Channel 14/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-165-164:1227467:1227960 [4] NCCL INFO Channel 07/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-165-202:153566:154062 [4] NCCL INFO Channel 06/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-165-164:1227469:1227957 [6] NCCL INFO Channel 06/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227469:1227957 [6] NCCL INFO Channel 14/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227469:1227957 [6] NCCL INFO Channel 06/0 : 22[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153565:154058 [3] NCCL INFO Channel 07/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-165-202:153566:154062 [4] NCCL INFO Channel 07/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-165-164:1227465:1227962 [2] NCCL INFO Channel 12/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-165-164:1227467:1227960 [4] NCCL INFO Channel 09/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-165-202:153568:154056 [6] NCCL INFO Channel 06/0 : 22[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153565:154058 [3] NCCL INFO Channel 09/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-165-202:153568:154056 [6] NCCL INFO Channel 14/0 : 14[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153566:154062 [4] NCCL INFO Channel 09/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-165-131:1630205:1630727 [0] NCCL INFO Channel 07/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-165-202:153568:154056 [6] NCCL INFO Channel 14/0 : 30[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227465:1227962 [2] NCCL INFO Channel 13/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-165-164:1227467:1227960 [4] NCCL INFO Channel 10/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-165-164:1227464:1227963 [1] NCCL INFO Channel 01/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-165-131:1630206:1630728 [1] NCCL INFO Channel 01/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-165-202:153565:154058 [3] NCCL INFO Channel 10/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-160-225:124163:124661 [3] NCCL INFO Connected all rings
+ip-26-0-165-202:153566:154062 [4] NCCL INFO Channel 10/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-165-202:153565:154058 [3] NCCL INFO Channel 11/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-165-202:153566:154062 [4] NCCL INFO Channel 11/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-165-131:1630206:1630728 [1] NCCL INFO Channel 03/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-165-202:153565:154058 [3] NCCL INFO Channel 13/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-165-202:153566:154062 [4] NCCL INFO Channel 12/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-165-131:1630206:1630728 [1] NCCL INFO Channel 04/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-165-164:1227467:1227960 [4] NCCL INFO Channel 11/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-165-164:1227465:1227962 [2] NCCL INFO Channel 14/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-165-202:153565:154058 [3] NCCL INFO Channel 14/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-165-202:153566:154062 [4] NCCL INFO Channel 14/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-165-164:1227464:1227963 [1] NCCL INFO Channel 03/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-160-242:1364908:1365406 [6] NCCL INFO Connected all rings
+ip-26-0-165-164:1227467:1227960 [4] NCCL INFO Channel 12/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-165-164:1227465:1227962 [2] NCCL INFO Channel 15/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-165-131:1630206:1630728 [1] NCCL INFO Channel 05/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-165-202:153565:154058 [3] NCCL INFO Channel 15/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-165-202:153566:154062 [4] NCCL INFO Channel 15/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-165-202:153569:154055 [7] NCCL INFO Channel 07/0 : 23[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153562:154061 [0] NCCL INFO Channel 00/0 : 16[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227464:1227963 [1] NCCL INFO Channel 04/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-165-164:1227467:1227960 [4] NCCL INFO Channel 14/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-165-202:153567:154059 [5] NCCL INFO Channel 05/0 : 21[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153564:154057 [2] NCCL INFO Channel 02/0 : 18[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153566:154062 [4] NCCL INFO Channel 04/0 : 20[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630206:1630728 [1] NCCL INFO Channel 06/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-165-202:153565:154058 [3] NCCL INFO Channel 03/0 : 19[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227464:1227963 [1] NCCL INFO Channel 05/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-165-131:1630205:1630727 [0] NCCL INFO Channel 10/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-165-164:1227467:1227960 [4] NCCL INFO Channel 15/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-165-131:1630206:1630728 [1] NCCL INFO Channel 07/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-165-131:1630205:1630727 [0] NCCL INFO Channel 11/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-165-164:1227464:1227963 [1] NCCL INFO Channel 06/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-165-131:1630206:1630728 [1] NCCL INFO Channel 09/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-165-164:1227464:1227963 [1] NCCL INFO Channel 07/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-165-131:1630205:1630727 [0] NCCL INFO Channel 12/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-165-164:1227468:1227959 [5] NCCL INFO Channel 05/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227468:1227959 [5] NCCL INFO Channel 13/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227464:1227963 [1] NCCL INFO Channel 09/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-165-164:1227468:1227959 [5] NCCL INFO Channel 05/0 : 21[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227464:1227963 [1] NCCL INFO Channel 11/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-165-202:153567:154059 [5] NCCL INFO Channel 13/0 : 13[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153567:154059 [5] NCCL INFO Channel 13/0 : 29[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630206:1630728 [1] NCCL INFO Channel 11/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-165-164:1227464:1227963 [1] NCCL INFO Channel 12/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-165-164:1227464:1227963 [1] NCCL INFO Channel 13/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-165-131:1630205:1630727 [0] NCCL INFO Channel 13/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-165-164:1227464:1227963 [1] NCCL INFO Channel 14/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-165-131:1630206:1630728 [1] NCCL INFO Channel 12/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-165-131:1630205:1630727 [0] NCCL INFO Channel 14/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-165-131:1630206:1630728 [1] NCCL INFO Channel 13/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-165-131:1630205:1630727 [0] NCCL INFO Channel 15/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-165-131:1630212:1630734 [7] NCCL INFO Connected all rings
+ip-26-0-165-164:1227464:1227963 [1] NCCL INFO Channel 15/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-165-131:1630206:1630728 [1] NCCL INFO Channel 14/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-165-164:1227463:1227964 [0] NCCL INFO Channel 05/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-165-164:1227464:1227963 [1] NCCL INFO Channel 01/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227464:1227963 [1] NCCL INFO Channel 09/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227464:1227963 [1] NCCL INFO Channel 01/0 : 17[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227463:1227964 [0] NCCL INFO Channel 06/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-165-164:1227463:1227964 [0] NCCL INFO Channel 07/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-165-131:1630210:1630732 [5] NCCL INFO Connected all rings
+ip-26-0-165-164:1227463:1227964 [0] NCCL INFO Channel 13/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-165-164:1227463:1227964 [0] NCCL INFO Channel 14/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-165-131:1630207:1630731 [2] NCCL INFO Channel 01/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-165-164:1227466:1227961 [3] NCCL INFO Connected all rings
+ip-26-0-165-164:1227463:1227964 [0] NCCL INFO Channel 15/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-165-131:1630206:1630728 [1] NCCL INFO Channel 15/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-161-103:96888:97387 [0] NCCL INFO Connected all rings
+ip-26-0-161-103:96888:97387 [0] NCCL INFO Channel 02/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-165-131:1630207:1630731 [2] NCCL INFO Channel 02/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-165-164:1227470:1227958 [7] NCCL INFO Channel 07/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227463:1227964 [0] NCCL INFO Channel 00/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630205:1630727 [0] NCCL INFO Channel 05/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-165-164:1227463:1227964 [0] NCCL INFO Channel 08/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227463:1227964 [0] NCCL INFO Channel 00/0 : 16[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153563:154060 [1] NCCL INFO Channel 09/0 : 9[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153563:154060 [1] NCCL INFO Channel 09/0 : 25[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153562:154061 [0] NCCL INFO Channel 08/0 : 8[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153562:154061 [0] NCCL INFO Channel 08/0 : 24[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630207:1630731 [2] NCCL INFO Channel 04/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-165-131:1630205:1630727 [0] NCCL INFO Channel 06/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-161-123:32704:33197 [0] NCCL INFO Channel 03/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-165-131:1630207:1630731 [2] NCCL INFO Channel 05/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-165-164:1227470:1227958 [7] NCCL INFO Channel 15/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227470:1227958 [7] NCCL INFO Channel 07/0 : 23[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630205:1630727 [0] NCCL INFO Channel 07/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-161-103:96888:97387 [0] NCCL INFO Channel 03/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-165-131:1630207:1630731 [2] NCCL INFO Channel 06/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-165-131:1630211:1630733 [6] NCCL INFO Connected all rings
+ip-26-0-165-164:1227466:1227961 [3] NCCL INFO Channel 01/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-165-131:1630205:1630727 [0] NCCL INFO Channel 13/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-165-164:1227466:1227961 [3] NCCL INFO Channel 02/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-161-103:96888:97387 [0] NCCL INFO Channel 04/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-165-164:1227466:1227961 [3] NCCL INFO Channel 03/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-161-123:32704:33197 [0] NCCL INFO Channel 04/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-165-131:1630207:1630731 [2] NCCL INFO Channel 07/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-165-164:1227466:1227961 [3] NCCL INFO Channel 05/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-165-164:1227466:1227961 [3] NCCL INFO Channel 06/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-165-202:153569:154055 [7] NCCL INFO Channel 15/0 : 15[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153569:154055 [7] NCCL INFO Channel 15/0 : 31[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227466:1227961 [3] NCCL INFO Channel 07/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-165-131:1630205:1630727 [0] NCCL INFO Channel 14/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-161-103:96888:97387 [0] NCCL INFO Channel 05/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-165-164:1227466:1227961 [3] NCCL INFO Channel 09/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-161-123:32704:33197 [0] NCCL INFO Channel 05/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-165-164:1227466:1227961 [3] NCCL INFO Channel 10/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-165-164:1227466:1227961 [3] NCCL INFO Channel 11/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-165-164:1227466:1227961 [3] NCCL INFO Channel 13/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-161-103:96888:97387 [0] NCCL INFO Channel 06/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-165-164:1227466:1227961 [3] NCCL INFO Channel 14/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-161-123:32706:33199 [2] NCCL INFO Connected all rings
+ip-26-0-165-131:1630209:1630730 [4] NCCL INFO Connected all rings
+ip-26-0-165-131:1630207:1630731 [2] NCCL INFO Channel 09/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-165-164:1227466:1227961 [3] NCCL INFO Channel 15/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-160-225:124160:124658 [0] NCCL INFO Connected all rings
+ip-26-0-160-225:124160:124658 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-164:1227465:1227962 [2] NCCL INFO Channel 02/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630205:1630727 [0] NCCL INFO Channel 15/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-165-164:1227465:1227962 [2] NCCL INFO Channel 10/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227466:1227961 [3] NCCL INFO Channel 03/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227467:1227960 [4] NCCL INFO Channel 04/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227465:1227962 [2] NCCL INFO Channel 02/0 : 18[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227466:1227961 [3] NCCL INFO Channel 11/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227467:1227960 [4] NCCL INFO Channel 12/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96888:97387 [0] NCCL INFO Channel 07/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-165-164:1227466:1227961 [3] NCCL INFO Channel 03/0 : 19[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227467:1227960 [4] NCCL INFO Channel 04/0 : 20[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32704:33197 [0] NCCL INFO Channel 06/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-165-131:1630207:1630731 [2] NCCL INFO Channel 10/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-165-202:153565:154058 [3] NCCL INFO Channel 11/0 : 11[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153565:154058 [3] NCCL INFO Channel 11/0 : 27[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:124160:124658 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-202:153566:154062 [4] NCCL INFO Channel 12/0 : 12[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96888:97387 [0] NCCL INFO Channel 10/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-165-202:153566:154062 [4] NCCL INFO Channel 12/0 : 28[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153564:154057 [2] NCCL INFO Channel 10/0 : 10[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630207:1630731 [2] NCCL INFO Channel 12/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-165-202:153564:154057 [2] NCCL INFO Channel 10/0 : 26[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32704:33197 [0] NCCL INFO Channel 07/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-161-103:96888:97387 [0] NCCL INFO Channel 11/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-165-131:1630207:1630731 [2] NCCL INFO Channel 13/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-161-123:32704:33197 [0] NCCL INFO Channel 10/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-161-103:96888:97387 [0] NCCL INFO Channel 12/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-160-225:124160:124658 [0] NCCL INFO Channel 04/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-131:1630207:1630731 [2] NCCL INFO Channel 14/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-160-242:1364902:1365410 [0] NCCL INFO Connected all rings
+ip-26-0-160-242:1364902:1365410 [0] NCCL INFO Channel 02/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-165-131:1630208:1630729 [3] NCCL INFO Channel 01/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-161-123:32704:33197 [0] NCCL INFO Channel 11/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-165-131:1630207:1630731 [2] NCCL INFO Channel 15/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-165-131:1630208:1630729 [3] NCCL INFO Channel 02/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-165-131:1630210:1630732 [5] NCCL INFO Channel 01/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-161-123:32704:33197 [0] NCCL INFO Channel 12/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-160-225:124160:124658 [0] NCCL INFO Channel 05/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-103:96888:97387 [0] NCCL INFO Channel 13/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-160-242:1364902:1365410 [0] NCCL INFO Channel 03/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-165-131:1630208:1630729 [3] NCCL INFO Channel 03/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-165-131:1630210:1630732 [5] NCCL INFO Channel 02/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-161-103:96888:97387 [0] NCCL INFO Channel 14/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-165-131:1630206:1630728 [1] NCCL INFO Channel 09/0 : 1[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630206:1630728 [1] NCCL INFO Channel 01/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630206:1630728 [1] NCCL INFO Channel 09/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32704:33197 [0] NCCL INFO Channel 13/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-161-78:114513:115008 [1] NCCL INFO Channel 01/0 : 17[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630206:1630728 [1] NCCL INFO Channel 09/0 : 25[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114513:115008 [1] NCCL INFO Channel 01/0 : 1[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630208:1630729 [3] NCCL INFO Channel 05/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-165-164:1227464:1227963 [1] NCCL INFO Channel 01/0 : 1[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630206:1630728 [1] NCCL INFO Channel 09/0 : 9[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630210:1630732 [5] NCCL INFO Channel 03/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-165-164:1227464:1227963 [1] NCCL INFO Channel 01/0 : 17[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630206:1630728 [1] NCCL INFO Channel 01/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630206:1630728 [1] NCCL INFO Channel 09/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153563:154060 [1] NCCL INFO Channel 01/0 : 25[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96888:97387 [0] NCCL INFO Channel 15/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-165-164:1227464:1227963 [1] NCCL INFO Channel 01/0 : 25[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630206:1630728 [1] NCCL INFO Channel 09/0 : 9[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114513:115008 [1] NCCL INFO Channel 09/0 : 9[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227464:1227963 [1] NCCL INFO Channel 01/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364902:1365410 [0] NCCL INFO Channel 04/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-165-131:1630208:1630729 [3] NCCL INFO Channel 06/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-165-131:1630210:1630732 [5] NCCL INFO Channel 04/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-160-225:124160:124658 [0] NCCL INFO Channel 06/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-168-34:272150:272645 [0] NCCL INFO Connected all rings
+ip-26-0-160-242:1364902:1365410 [0] NCCL INFO Channel 05/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-168-34:272150:272645 [0] NCCL INFO Channel 02/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-165-164:1227464:1227963 [1] NCCL INFO Channel 09/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124160:124658 [0] NCCL INFO Channel 07/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-78:114513:115008 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-131:1630210:1630732 [5] NCCL INFO Channel 05/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-165-131:1630208:1630729 [3] NCCL INFO Channel 07/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-160-225:124161:124659 [1] NCCL INFO Connected all rings
+ip-26-0-161-78:114513:115008 [1] NCCL INFO Channel 08/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-123:32704:33197 [0] NCCL INFO Channel 14/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-160-242:1364902:1365410 [0] NCCL INFO Channel 06/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-165-164:1227464:1227963 [1] NCCL INFO Channel 00/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-165-131:1630210:1630732 [5] NCCL INFO Channel 07/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-165-131:1630208:1630729 [3] NCCL INFO Channel 09/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-165-131:1630211:1630733 [6] NCCL INFO Channel 01/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-165-164:1227464:1227963 [1] NCCL INFO Channel 08/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-165-202:153563:154060 [1] NCCL INFO Channel 00/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-161-123:32704:33197 [0] NCCL INFO Channel 15/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-160-225:124160:124658 [0] NCCL INFO Channel 10/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-202:153563:154060 [1] NCCL INFO Channel 08/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-160-242:1364902:1365410 [0] NCCL INFO Channel 07/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-165-131:1630210:1630732 [5] NCCL INFO Channel 09/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-165-131:1630208:1630729 [3] NCCL INFO Channel 10/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-165-131:1630211:1630733 [6] NCCL INFO Channel 02/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-160-242:1364902:1365410 [0] NCCL INFO Channel 10/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-161-123:32705:33203 [1] NCCL INFO Channel 01/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-160-225:124160:124658 [0] NCCL INFO Channel 11/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-131:1630210:1630732 [5] NCCL INFO Channel 10/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-165-131:1630209:1630730 [4] NCCL INFO Channel 01/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-165-131:1630208:1630729 [3] NCCL INFO Channel 11/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-165-131:1630211:1630733 [6] NCCL INFO Channel 03/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-160-225:124160:124658 [0] NCCL INFO Channel 12/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-123:32705:33203 [1] NCCL INFO Channel 03/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-160-242:1364902:1365410 [0] NCCL INFO Channel 11/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-165-131:1630206:1630728 [1] NCCL INFO Channel 00/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-165-131:1630209:1630730 [4] NCCL INFO Channel 02/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-165-131:1630208:1630729 [3] NCCL INFO Channel 13/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-165-131:1630210:1630732 [5] NCCL INFO Channel 11/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-165-131:1630211:1630733 [6] NCCL INFO Channel 04/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-161-123:32705:33203 [1] NCCL INFO Channel 04/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-160-242:1364902:1365410 [0] NCCL INFO Channel 12/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-161-103:96892:97383 [4] NCCL INFO Connected all rings
+ip-26-0-160-225:124160:124658 [0] NCCL INFO Channel 13/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-131:1630209:1630730 [4] NCCL INFO Channel 03/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-165-131:1630206:1630728 [1] NCCL INFO Channel 08/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-165-131:1630208:1630729 [3] NCCL INFO Channel 14/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-165-131:1630210:1630732 [5] NCCL INFO Channel 12/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-165-131:1630211:1630733 [6] NCCL INFO Channel 05/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-161-123:32705:33203 [1] NCCL INFO Channel 05/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-160-242:1364902:1365410 [0] NCCL INFO Channel 13/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-161-123:32709:33201 [5] NCCL INFO Connected all rings
+ip-26-0-160-225:124160:124658 [0] NCCL INFO Channel 14/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-103:96893:97384 [5] NCCL INFO Connected all rings
+ip-26-0-161-123:32705:33203 [1] NCCL INFO Channel 06/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-165-131:1630210:1630732 [5] NCCL INFO Channel 13/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-165-131:1630209:1630730 [4] NCCL INFO Channel 04/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-165-131:1630208:1630729 [3] NCCL INFO Channel 15/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-165-131:1630211:1630733 [6] NCCL INFO Channel 06/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-168-34:272151:272646 [1] NCCL INFO Connected all rings
+ip-26-0-160-225:124160:124658 [0] NCCL INFO Channel 15/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-242:1364902:1365410 [0] NCCL INFO Channel 14/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-161-123:32705:33203 [1] NCCL INFO Channel 07/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-160-225:124162:124660 [2] NCCL INFO Connected all rings
+ip-26-0-165-131:1630210:1630732 [5] NCCL INFO Channel 15/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-165-131:1630209:1630730 [4] NCCL INFO Channel 06/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-165-131:1630211:1630733 [6] NCCL INFO Channel 09/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-161-103:96890:97386 [2] NCCL INFO Connected all rings
+ip-26-0-165-131:1630207:1630731 [2] NCCL INFO Channel 10/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32711:33204 [7] NCCL INFO Connected all rings
+ip-26-0-165-131:1630207:1630731 [2] NCCL INFO Channel 02/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630207:1630731 [2] NCCL INFO Channel 10/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32705:33203 [1] NCCL INFO Channel 09/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-160-242:1364902:1365410 [0] NCCL INFO Channel 15/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-161-78:114514:115011 [2] NCCL INFO Channel 02/0 : 18[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630209:1630730 [4] NCCL INFO Channel 07/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-161-78:114514:115011 [2] NCCL INFO Channel 02/0 : 2[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227465:1227962 [2] NCCL INFO Channel 02/0 : 2[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630207:1630731 [2] NCCL INFO Channel 10/0 : 26[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630211:1630733 [6] NCCL INFO Channel 10/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-165-131:1630207:1630731 [2] NCCL INFO Channel 10/0 : 10[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227465:1227962 [2] NCCL INFO Channel 02/0 : 18[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630207:1630731 [2] NCCL INFO Channel 02/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227465:1227962 [2] NCCL INFO Channel 02/0 : 26[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630207:1630731 [2] NCCL INFO Channel 10/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153564:154057 [2] NCCL INFO Channel 02/0 : 26[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114514:115011 [2] NCCL INFO Channel 10/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227465:1227962 [2] NCCL INFO Channel 02/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630207:1630731 [2] NCCL INFO Channel 10/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630209:1630730 [4] NCCL INFO Channel 09/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-165-164:1227465:1227962 [2] NCCL INFO Channel 10/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630211:1630733 [6] NCCL INFO Channel 11/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-161-123:32705:33203 [1] NCCL INFO Channel 11/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-168-52:2722046:2722541 [1] NCCL INFO Connected all rings
+ip-26-0-165-131:1630209:1630730 [4] NCCL INFO Channel 10/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-165-131:1630211:1630733 [6] NCCL INFO Channel 12/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-161-123:32705:33203 [1] NCCL INFO Channel 12/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-161-123:32706:33199 [2] NCCL INFO Channel 01/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-165-131:1630209:1630730 [4] NCCL INFO Channel 11/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-165-131:1630211:1630733 [6] NCCL INFO Channel 13/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-165-202:153564:154057 [2] NCCL INFO Channel 00/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-160-225:124165:124662 [5] NCCL INFO Connected all rings
+ip-26-0-165-164:1227465:1227962 [2] NCCL INFO Channel 00/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-165-202:153564:154057 [2] NCCL INFO Channel 01/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-165-164:1227465:1227962 [2] NCCL INFO Channel 01/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-161-123:32705:33203 [1] NCCL INFO Channel 13/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-165-131:1630209:1630730 [4] NCCL INFO Channel 12/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-165-131:1630211:1630733 [6] NCCL INFO Channel 14/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-161-123:32706:33199 [2] NCCL INFO Channel 02/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-165-202:153564:154057 [2] NCCL INFO Channel 08/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-165-164:1227465:1227962 [2] NCCL INFO Channel 08/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-161-78:114514:115011 [2] NCCL INFO Channel 00/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-165-202:153564:154057 [2] NCCL INFO Channel 09/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-165-164:1227465:1227962 [2] NCCL INFO Channel 09/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-165-131:1630209:1630730 [4] NCCL INFO Channel 14/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-165-131:1630211:1630733 [6] NCCL INFO Channel 14/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630211:1630733 [6] NCCL INFO Channel 06/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630211:1630733 [6] NCCL INFO Channel 14/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630209:1630730 [4] NCCL INFO Channel 15/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-160-242:1364904:1365408 [2] NCCL INFO Connected all rings
+ip-26-0-161-123:32705:33203 [1] NCCL INFO Channel 14/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-161-123:32706:33199 [2] NCCL INFO Channel 04/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-165-131:1630211:1630733 [6] NCCL INFO Channel 14/0 : 30[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630211:1630733 [6] NCCL INFO Channel 14/0 : 14[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227469:1227957 [6] NCCL INFO Channel 06/0 : 6[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227469:1227957 [6] NCCL INFO Channel 06/0 : 22[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630211:1630733 [6] NCCL INFO Channel 06/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630208:1630729 [3] NCCL INFO Channel 11/0 : 3[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153568:154056 [6] NCCL INFO Channel 06/0 : 30[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630211:1630733 [6] NCCL INFO Channel 14/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630210:1630732 [5] NCCL INFO Channel 13/0 : 5[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630207:1630731 [2] NCCL INFO Channel 00/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-165-131:1630208:1630729 [3] NCCL INFO Channel 03/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630211:1630733 [6] NCCL INFO Channel 14/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630210:1630732 [5] NCCL INFO Channel 05/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630209:1630730 [4] NCCL INFO Channel 12/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630208:1630729 [3] NCCL INFO Channel 11/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630210:1630732 [5] NCCL INFO Channel 13/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630209:1630730 [4] NCCL INFO Channel 04/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630209:1630730 [4] NCCL INFO Channel 12/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630207:1630731 [2] NCCL INFO Channel 01/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-160-225:124161:124659 [1] NCCL INFO Channel 01/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-131:1630208:1630729 [3] NCCL INFO Channel 11/0 : 27[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630208:1630729 [3] NCCL INFO Channel 11/0 : 11[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227466:1227961 [3] NCCL INFO Channel 03/0 : 3[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630210:1630732 [5] NCCL INFO Channel 13/0 : 29[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227468:1227959 [5] NCCL INFO Channel 05/0 : 5[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630209:1630730 [4] NCCL INFO Channel 12/0 : 28[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227466:1227961 [3] NCCL INFO Channel 03/0 : 19[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630210:1630732 [5] NCCL INFO Channel 13/0 : 13[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630209:1630730 [4] NCCL INFO Channel 12/0 : 12[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227468:1227959 [5] NCCL INFO Channel 05/0 : 21[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630208:1630729 [3] NCCL INFO Channel 03/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114514:115011 [2] NCCL INFO Channel 01/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-165-131:1630207:1630731 [2] NCCL INFO Channel 08/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-161-78:114517:115013 [5] NCCL INFO Channel 05/0 : 21[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153565:154058 [3] NCCL INFO Channel 03/0 : 27[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630208:1630729 [3] NCCL INFO Channel 11/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114517:115013 [5] NCCL INFO Channel 05/0 : 5[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630210:1630732 [5] NCCL INFO Channel 05/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32705:33203 [1] NCCL INFO Channel 15/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-165-131:1630208:1630729 [3] NCCL INFO Channel 11/0 : 11[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153567:154059 [5] NCCL INFO Channel 05/0 : 29[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630209:1630730 [4] NCCL INFO Channel 04/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32706:33199 [2] NCCL INFO Channel 05/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-165-202:153566:154062 [4] NCCL INFO Channel 04/0 : 28[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630210:1630732 [5] NCCL INFO Channel 13/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630209:1630730 [4] NCCL INFO Channel 12/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630210:1630732 [5] NCCL INFO Channel 13/0 : 13[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630209:1630730 [4] NCCL INFO Channel 12/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114514:115011 [2] NCCL INFO Channel 08/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-165-164:1227468:1227959 [5] NCCL INFO Channel 05/0 : 29[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630207:1630731 [2] NCCL INFO Channel 09/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-161-78:114517:115013 [5] NCCL INFO Channel 13/0 : 13[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114515:115009 [3] NCCL INFO Channel 03/0 : 19[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227468:1227959 [5] NCCL INFO Channel 05/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114515:115009 [3] NCCL INFO Channel 03/0 : 3[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227468:1227959 [5] NCCL INFO Channel 13/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114516:115010 [4] NCCL INFO Channel 04/0 : 20[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:124161:124659 [1] NCCL INFO Channel 03/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-78:114516:115010 [4] NCCL INFO Channel 04/0 : 4[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114514:115011 [2] NCCL INFO Channel 09/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-165-131:1630212:1630734 [7] NCCL INFO Channel 15/0 : 7[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114515:115009 [3] NCCL INFO Channel 11/0 : 11[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114518:115012 [6] NCCL INFO Channel 06/0 : 22[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227466:1227961 [3] NCCL INFO Channel 03/0 : 27[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630212:1630734 [7] NCCL INFO Channel 07/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227466:1227961 [3] NCCL INFO Channel 03/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114518:115012 [6] NCCL INFO Channel 06/0 : 6[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32706:33199 [2] NCCL INFO Channel 06/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-165-164:1227466:1227961 [3] NCCL INFO Channel 11/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630212:1630734 [7] NCCL INFO Channel 15/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630205:1630727 [0] NCCL INFO Channel 08/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32704:33197 [0] NCCL INFO Channel 05/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-165-131:1630205:1630727 [0] NCCL INFO Channel 00/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364905:1365404 [3] NCCL INFO Connected all rings
+ip-26-0-165-131:1630205:1630727 [0] NCCL INFO Channel 08/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114518:115012 [6] NCCL INFO Channel 14/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:124161:124659 [1] NCCL INFO Channel 04/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-78:114519:115014 [7] NCCL INFO Channel 07/0 : 23[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227467:1227960 [4] NCCL INFO Channel 04/0 : 4[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227469:1227957 [6] NCCL INFO Channel 06/0 : 30[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630212:1630734 [7] NCCL INFO Channel 15/0 : 31[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114519:115014 [7] NCCL INFO Channel 07/0 : 7[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227467:1227960 [4] NCCL INFO Channel 04/0 : 20[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630212:1630734 [7] NCCL INFO Channel 15/0 : 15[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114512:115007 [0] NCCL INFO Channel 00/0 : 16[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227469:1227957 [6] NCCL INFO Channel 06/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630205:1630727 [0] NCCL INFO Channel 08/0 : 24[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227470:1227958 [7] NCCL INFO Channel 07/0 : 7[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114512:115007 [0] NCCL INFO Channel 00/0 : 0[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630205:1630727 [0] NCCL INFO Channel 08/0 : 8[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227469:1227957 [6] NCCL INFO Channel 14/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227463:1227964 [0] NCCL INFO Channel 00/0 : 0[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227470:1227958 [7] NCCL INFO Channel 07/0 : 23[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227467:1227960 [4] NCCL INFO Channel 04/0 : 28[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227463:1227964 [0] NCCL INFO Channel 00/0 : 16[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630212:1630734 [7] NCCL INFO Channel 07/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96891:97385 [3] NCCL INFO Channel 01/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-165-164:1227467:1227960 [4] NCCL INFO Channel 04/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630212:1630734 [7] NCCL INFO Channel 15/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630205:1630727 [0] NCCL INFO Channel 00/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227470:1227958 [7] NCCL INFO Channel 07/0 : 31[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114519:115014 [7] NCCL INFO Channel 15/0 : 15[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630212:1630734 [7] NCCL INFO Channel 15/0 : 15[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227467:1227960 [4] NCCL INFO Channel 12/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630205:1630727 [0] NCCL INFO Channel 08/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227470:1227958 [7] NCCL INFO Channel 07/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630205:1630727 [0] NCCL INFO Channel 08/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227470:1227958 [7] NCCL INFO Channel 15/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32706:33199 [2] NCCL INFO Channel 07/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-161-123:32704:33197 [0] NCCL INFO Channel 06/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-161-78:114519:115014 [7] NCCL INFO Channel 01/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-202:153562:154061 [0] NCCL INFO Channel 00/0 : 24[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630212:1630734 [7] NCCL INFO Channel 01/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-165-202:153567:154059 [5] NCCL INFO Channel 00/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-161-78:114519:115014 [7] NCCL INFO Channel 02/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-123:32707:33200 [3] NCCL INFO Connected all rings
+ip-26-0-165-131:1630212:1630734 [7] NCCL INFO Channel 02/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-161-78:114512:115007 [0] NCCL INFO Channel 08/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96892:97383 [4] NCCL INFO Channel 01/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-165-164:1227463:1227964 [0] NCCL INFO Channel 00/0 : 24[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114516:115010 [4] NCCL INFO Channel 12/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96891:97385 [3] NCCL INFO Channel 02/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-165-164:1227463:1227964 [0] NCCL INFO Channel 00/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:124161:124659 [1] NCCL INFO Channel 05/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-164:1227463:1227964 [0] NCCL INFO Channel 08/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32706:33199 [2] NCCL INFO Channel 09/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-165-202:153569:154055 [7] NCCL INFO Channel 07/0 : 31[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32704:33197 [0] NCCL INFO Channel 07/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-165-202:153567:154059 [5] NCCL INFO Channel 04/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-165-131:1630212:1630734 [7] NCCL INFO Channel 03/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-165-202:153569:154055 [7] NCCL INFO Channel 01/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-165-164:1227470:1227958 [7] NCCL INFO Channel 01/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-165-164:1227469:1227957 [6] NCCL INFO Channel 00/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-161-78:114519:115014 [7] NCCL INFO Channel 03/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-242:1364909:1365405 [7] NCCL INFO Connected all rings
+ip-26-0-165-202:153568:154056 [6] NCCL INFO Channel 00/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-161-103:96891:97385 [3] NCCL INFO Channel 03/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-161-123:32706:33199 [2] NCCL INFO Channel 10/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-161-123:32704:33197 [0] NCCL INFO Channel 13/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-160-225:124166:124665 [6] NCCL INFO Connected all rings
+ip-26-0-165-131:1630211:1630733 [6] NCCL INFO Channel 00/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-161-123:32709:33201 [5] NCCL INFO Channel 01/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-165-131:1630212:1630734 [7] NCCL INFO Channel 04/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-161-103:96893:97384 [5] NCCL INFO Channel 01/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-165-164:1227470:1227958 [7] NCCL INFO Channel 02/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-165-202:153567:154059 [5] NCCL INFO Channel 08/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-161-78:114519:115014 [7] NCCL INFO Channel 04/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-164:1227469:1227957 [6] NCCL INFO Channel 05/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-160-225:124161:124659 [1] NCCL INFO Channel 06/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-103:96891:97385 [3] NCCL INFO Channel 05/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-161-103:96892:97383 [4] NCCL INFO Channel 02/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-161-103:96893:97384 [5] NCCL INFO Channel 02/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-165-202:153568:154056 [6] NCCL INFO Channel 05/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-161-78:114517:115013 [5] NCCL INFO Channel 00/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-131:1630211:1630733 [6] NCCL INFO Channel 05/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-161-123:32706:33199 [2] NCCL INFO Channel 12/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-165-164:1227470:1227958 [7] NCCL INFO Channel 03/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-161-123:32704:33197 [0] NCCL INFO Channel 14/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-161-123:32709:33201 [5] NCCL INFO Channel 02/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-165-164:1227469:1227957 [6] NCCL INFO Channel 08/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-165-164:1227467:1227960 [4] NCCL INFO Channel 00/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-165-202:153567:154059 [5] NCCL INFO Channel 12/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-165-131:1630212:1630734 [7] NCCL INFO Channel 05/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-161-78:114518:115012 [6] NCCL INFO Channel 00/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-202:153568:154056 [6] NCCL INFO Channel 08/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-161-123:32710:33202 [6] NCCL INFO Connected all rings
+ip-26-0-165-202:153569:154055 [7] NCCL INFO Channel 02/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-160-225:124161:124659 [1] NCCL INFO Channel 07/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-103:96891:97385 [3] NCCL INFO Channel 06/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-161-103:96892:97383 [4] NCCL INFO Channel 03/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-161-103:96890:97386 [2] NCCL INFO Channel 01/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-161-103:96893:97384 [5] NCCL INFO Channel 03/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-161-78:114517:115013 [5] NCCL INFO Channel 04/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-164:1227470:1227958 [7] NCCL INFO Channel 04/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-165-131:1630211:1630733 [6] NCCL INFO Channel 08/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-161-78:114519:115014 [7] NCCL INFO Channel 05/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-78:114518:115012 [6] NCCL INFO Channel 05/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-164:1227469:1227957 [6] NCCL INFO Channel 13/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-165-164:1227467:1227960 [4] NCCL INFO Channel 03/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-165-131:1630212:1630734 [7] NCCL INFO Channel 06/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-160-225:124167:124664 [7] NCCL INFO Connected all rings
+ip-26-0-160-225:124161:124659 [1] NCCL INFO Channel 09/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-131:1630209:1630730 [4] NCCL INFO Channel 00/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-160-242:1364903:1365409 [1] NCCL INFO Channel 01/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-161-78:114516:115010 [4] NCCL INFO Channel 00/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-123:32709:33201 [5] NCCL INFO Channel 03/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-161-78:114517:115013 [5] NCCL INFO Channel 08/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-78:114518:115012 [6] NCCL INFO Channel 08/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-202:153569:154055 [7] NCCL INFO Channel 03/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-161-103:96891:97385 [3] NCCL INFO Channel 07/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-165-202:153566:154062 [4] NCCL INFO Channel 00/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-161-123:32704:33197 [0] NCCL INFO Channel 15/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-161-103:96892:97383 [4] NCCL INFO Channel 04/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-161-103:96893:97384 [5] NCCL INFO Channel 04/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-165-131:1630211:1630733 [6] NCCL INFO Channel 13/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-161-103:96890:97386 [2] NCCL INFO Channel 02/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-161-78:114519:115014 [7] NCCL INFO Channel 06/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-164:1227468:1227959 [5] NCCL INFO Channel 00/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-165-202:153568:154056 [6] NCCL INFO Channel 13/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-160-225:124161:124659 [1] NCCL INFO Channel 11/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-123:32706:33199 [2] NCCL INFO Channel 13/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-165-131:1630209:1630730 [4] NCCL INFO Channel 03/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-160-242:1364903:1365409 [1] NCCL INFO Channel 03/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-165-164:1227468:1227959 [5] NCCL INFO Channel 04/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-161-103:96891:97385 [3] NCCL INFO Channel 09/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-161-103:96893:97384 [5] NCCL INFO Channel 05/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-161-123:32709:33201 [5] NCCL INFO Channel 04/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-165-131:1630210:1630732 [5] NCCL INFO Channel 00/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-165-202:153569:154055 [7] NCCL INFO Channel 04/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-165-164:1227468:1227959 [5] NCCL INFO Channel 08/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-161-78:114516:115010 [4] NCCL INFO Channel 03/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-165-164:1227470:1227958 [7] NCCL INFO Channel 05/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-161-78:114518:115012 [6] NCCL INFO Channel 13/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-78:114517:115013 [5] NCCL INFO Channel 12/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-164:1227467:1227960 [4] NCCL INFO Channel 05/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-165-131:1630209:1630730 [4] NCCL INFO Channel 05/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-161-78:114519:115014 [7] NCCL INFO Channel 07/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-202:153569:154055 [7] NCCL INFO Channel 05/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-161-123:32706:33199 [2] NCCL INFO Channel 14/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-165-131:1630210:1630732 [5] NCCL INFO Channel 04/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-161-103:96892:97383 [4] NCCL INFO Channel 06/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-165-202:153569:154055 [7] NCCL INFO Channel 06/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-161-103:96891:97385 [3] NCCL INFO Channel 10/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-161-103:96893:97384 [5] NCCL INFO Channel 07/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-161-123:32709:33201 [5] NCCL INFO Channel 05/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-160-225:124164:124663 [4] NCCL INFO Connected all rings
+ip-26-0-165-131:1630209:1630730 [4] NCCL INFO Channel 06/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-165-164:1227468:1227959 [5] NCCL INFO Channel 12/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-165-164:1227470:1227958 [7] NCCL INFO Channel 06/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-165-202:153569:154055 [7] NCCL INFO Channel 07/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-165-164:1227467:1227960 [4] NCCL INFO Channel 06/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-165-131:1630210:1630732 [5] NCCL INFO Channel 08/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-161-123:32706:33199 [2] NCCL INFO Channel 15/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-160-242:1364903:1365409 [1] NCCL INFO Channel 04/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-165-202:153566:154062 [4] NCCL INFO Channel 03/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-161-103:96894:97381 [6] NCCL INFO Channel 01/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-161-78:114516:115010 [4] NCCL INFO Channel 05/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-78:114515:115009 [3] NCCL INFO Channel 00/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-78:114519:115014 [7] NCCL INFO Channel 09/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-131:1630209:1630730 [4] NCCL INFO Channel 07/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-161-123:32709:33201 [5] NCCL INFO Channel 07/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-165-131:1630210:1630732 [5] NCCL INFO Channel 12/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-161-103:96892:97383 [4] NCCL INFO Channel 07/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-161-103:96893:97384 [5] NCCL INFO Channel 09/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-161-103:96891:97385 [3] NCCL INFO Channel 11/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-165-202:153569:154055 [7] NCCL INFO Channel 09/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-160-242:1364903:1365409 [1] NCCL INFO Channel 05/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-165-131:1630212:1630734 [7] NCCL INFO Channel 07/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-165-164:1227470:1227958 [7] NCCL INFO Channel 07/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-165-202:153566:154062 [4] NCCL INFO Channel 05/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-165-164:1227467:1227960 [4] NCCL INFO Channel 07/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-161-123:32705:33203 [1] NCCL INFO Channel 01/0 : 17[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32709:33201 [5] NCCL INFO Channel 09/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-161-78:114516:115010 [4] NCCL INFO Channel 06/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-78:114515:115009 [3] NCCL INFO Channel 02/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-78:114519:115014 [7] NCCL INFO Channel 10/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-131:1630209:1630730 [4] NCCL INFO Channel 08/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-161-103:96894:97381 [6] NCCL INFO Channel 02/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-165-202:153569:154055 [7] NCCL INFO Channel 10/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-165-131:1630212:1630734 [7] NCCL INFO Channel 09/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-165-202:153566:154062 [4] NCCL INFO Channel 06/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-160-242:1364903:1365409 [1] NCCL INFO Channel 06/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-165-164:1227470:1227958 [7] NCCL INFO Channel 09/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-165-164:1227467:1227960 [4] NCCL INFO Channel 08/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-161-103:96890:97386 [2] NCCL INFO Channel 04/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-161-103:96891:97385 [3] NCCL INFO Channel 13/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-161-103:96892:97383 [4] NCCL INFO Channel 09/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-161-123:32709:33201 [5] NCCL INFO Channel 10/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-161-103:96893:97384 [5] NCCL INFO Channel 10/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-165-131:1630209:1630730 [4] NCCL INFO Channel 11/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-161-78:114515:115009 [3] NCCL INFO Channel 08/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-78:114516:115010 [4] NCCL INFO Channel 07/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-165-131:1630212:1630734 [7] NCCL INFO Channel 10/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-161-78:114519:115014 [7] NCCL INFO Channel 11/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-202:153569:154055 [7] NCCL INFO Channel 11/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-165-202:153566:154062 [4] NCCL INFO Channel 07/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-165-164:1227470:1227958 [7] NCCL INFO Channel 10/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-165-164:1227467:1227960 [4] NCCL INFO Channel 11/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-161-123:32709:33201 [5] NCCL INFO Channel 11/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-161-103:96894:97381 [6] NCCL INFO Channel 03/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-160-242:1364903:1365409 [1] NCCL INFO Channel 07/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-165-131:1630209:1630730 [4] NCCL INFO Channel 13/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-165-202:153569:154055 [7] NCCL INFO Channel 12/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-165-131:1630212:1630734 [7] NCCL INFO Channel 11/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-165-202:153566:154062 [4] NCCL INFO Channel 08/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-161-103:96890:97386 [2] NCCL INFO Channel 05/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-161-103:96891:97385 [3] NCCL INFO Channel 14/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-161-103:96892:97383 [4] NCCL INFO Channel 10/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-161-103:96893:97384 [5] NCCL INFO Channel 11/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-161-123:32709:33201 [5] NCCL INFO Channel 12/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-165-131:1630208:1630729 [3] NCCL INFO Channel 00/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-165-164:1227467:1227960 [4] NCCL INFO Channel 13/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-165-164:1227470:1227958 [7] NCCL INFO Channel 11/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-161-103:96894:97381 [6] NCCL INFO Channel 04/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-165-131:1630209:1630730 [4] NCCL INFO Channel 14/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-165-202:153569:154055 [7] NCCL INFO Channel 13/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-165-202:153566:154062 [4] NCCL INFO Channel 11/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-161-123:32709:33201 [5] NCCL INFO Channel 13/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-160-242:1364903:1365409 [1] NCCL INFO Channel 09/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-165-164:1227470:1227958 [7] NCCL INFO Channel 12/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-165-164:1227467:1227960 [4] NCCL INFO Channel 14/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-165-164:1227466:1227961 [3] NCCL INFO Channel 00/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-165-131:1630208:1630729 [3] NCCL INFO Channel 02/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-165-202:153565:154058 [3] NCCL INFO Channel 00/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-161-78:114519:115014 [7] NCCL INFO Channel 12/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-131:1630212:1630734 [7] NCCL INFO Channel 12/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-161-103:96890:97386 [2] NCCL INFO Channel 06/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-161-103:96891:97385 [3] NCCL INFO Channel 15/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-161-123:32709:33201 [5] NCCL INFO Channel 15/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-161-103:96892:97383 [4] NCCL INFO Channel 11/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-161-103:96893:97384 [5] NCCL INFO Channel 12/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-165-202:153569:154055 [7] NCCL INFO Channel 14/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-165-202:153566:154062 [4] NCCL INFO Channel 13/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-165-131:1630209:1630730 [4] NCCL INFO Channel 15/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-165-164:1227470:1227958 [7] NCCL INFO Channel 13/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-165-164:1227467:1227960 [4] NCCL INFO Channel 15/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-165-164:1227466:1227961 [3] NCCL INFO Channel 02/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-165-202:153565:154058 [3] NCCL INFO Channel 02/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-165-131:1630208:1630729 [3] NCCL INFO Channel 08/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-161-103:96894:97381 [6] NCCL INFO Channel 05/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-165-131:1630212:1630734 [7] NCCL INFO Channel 13/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-161-78:114516:115010 [4] NCCL INFO Channel 08/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-78:114515:115009 [3] NCCL INFO Channel 10/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-123:32707:33200 [3] NCCL INFO Channel 01/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-165-202:153569:154055 [7] NCCL INFO Channel 15/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-165-202:153566:154062 [4] NCCL INFO Channel 14/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-160-242:1364907:1365403 [5] NCCL INFO Connected all rings
+ip-26-0-160-225:124161:124659 [1] NCCL INFO Channel 12/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-242:1364903:1365409 [1] NCCL INFO Channel 11/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-161-103:96890:97386 [2] NCCL INFO Channel 07/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-160-242:1364908:1365406 [6] NCCL INFO Channel 01/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-160-242:1364904:1365408 [2] NCCL INFO Channel 01/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-165-202:153565:154058 [3] NCCL INFO Channel 08/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-165-131:1630208:1630729 [3] NCCL INFO Channel 10/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-160-242:1364905:1365404 [3] NCCL INFO Channel 01/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-161-103:96892:97383 [4] NCCL INFO Channel 12/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-161-103:96893:97384 [5] NCCL INFO Channel 13/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-165-131:1630212:1630734 [7] NCCL INFO Channel 14/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-161-103:96894:97381 [6] NCCL INFO Channel 06/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-165-202:153566:154062 [4] NCCL INFO Channel 15/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-160-225:124161:124659 [1] NCCL INFO Channel 13/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-103:96889:97388 [1] NCCL INFO Connected all rings
+ip-26-0-165-202:153565:154058 [3] NCCL INFO Channel 10/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-161-123:32708:33198 [4] NCCL INFO Connected all rings
+ip-26-0-165-164:1227466:1227961 [3] NCCL INFO Channel 08/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-165-164:1227470:1227958 [7] NCCL INFO Channel 14/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-161-103:96890:97386 [2] NCCL INFO Channel 09/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-161-78:114519:115014 [7] NCCL INFO Channel 13/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-225:124161:124659 [1] NCCL INFO Channel 14/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-123:32707:33200 [3] NCCL INFO Channel 02/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-161-103:96892:97383 [4] NCCL INFO Channel 14/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-161-103:96893:97384 [5] NCCL INFO Channel 15/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-160-242:1364903:1365409 [1] NCCL INFO Channel 12/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-160-242:1364908:1365406 [6] NCCL INFO Channel 02/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-160-242:1364904:1365408 [2] NCCL INFO Channel 02/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-161-103:96894:97381 [6] NCCL INFO Channel 09/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-160-242:1364905:1365404 [3] NCCL INFO Channel 02/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-161-78:114516:115010 [4] NCCL INFO Channel 11/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-123:32707:33200 [3] NCCL INFO Channel 03/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-160-225:124161:124659 [1] NCCL INFO Channel 15/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-202:153569:154055 [7] NCCL INFO Channel 00/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-161-103:96890:97386 [2] NCCL INFO Channel 10/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-161-103:96892:97383 [4] NCCL INFO Channel 15/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-160-242:1364903:1365409 [1] NCCL INFO Channel 13/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-160-242:1364905:1365404 [3] NCCL INFO Channel 03/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-160-225:124162:124660 [2] NCCL INFO Channel 01/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-123:32707:33200 [3] NCCL INFO Channel 05/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-161-103:96894:97381 [6] NCCL INFO Channel 10/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-160-242:1364908:1365406 [6] NCCL INFO Channel 03/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-160-242:1364904:1365408 [2] NCCL INFO Channel 04/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-161-123:32710:33202 [6] NCCL INFO Channel 01/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-165-131:1630212:1630734 [7] NCCL INFO Channel 15/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-161-78:114516:115010 [4] NCCL INFO Channel 13/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-225:124160:124658 [0] NCCL INFO Channel 05/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-161-123:32707:33200 [3] NCCL INFO Channel 06/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-161-103:96890:97386 [2] NCCL INFO Channel 12/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-161-103:96894:97381 [6] NCCL INFO Channel 11/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-160-225:124162:124660 [2] NCCL INFO Channel 02/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-242:1364905:1365404 [3] NCCL INFO Channel 05/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-160-242:1364903:1365409 [1] NCCL INFO Channel 14/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-165-164:1227470:1227958 [7] NCCL INFO Channel 15/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-161-123:32710:33202 [6] NCCL INFO Channel 02/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-161-103:96892:97383 [4] NCCL INFO Channel 04/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364908:1365406 [6] NCCL INFO Channel 04/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-161-103:96892:97383 [4] NCCL INFO Channel 12/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364904:1365408 [2] NCCL INFO Channel 05/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-161-103:96892:97383 [4] NCCL INFO Channel 04/0 : 20[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227466:1227961 [3] NCCL INFO Channel 10/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-161-123:32707:33200 [3] NCCL INFO Channel 07/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-161-103:96890:97386 [2] NCCL INFO Channel 13/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-161-103:96894:97381 [6] NCCL INFO Channel 12/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-161-123:32710:33202 [6] NCCL INFO Channel 03/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-160-225:124160:124658 [0] NCCL INFO Channel 06/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-161-78:114519:115014 [7] NCCL INFO Channel 14/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-123:32707:33200 [3] NCCL INFO Channel 09/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-161-103:96890:97386 [2] NCCL INFO Channel 14/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-161-103:96894:97381 [6] NCCL INFO Channel 13/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-160-225:124162:124660 [2] NCCL INFO Channel 04/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-242:1364903:1365409 [1] NCCL INFO Channel 15/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-160-242:1364905:1365404 [3] NCCL INFO Channel 06/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-161-123:32710:33202 [6] NCCL INFO Channel 04/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-161-103:96890:97386 [2] NCCL INFO Channel 15/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-161-123:32707:33200 [3] NCCL INFO Channel 10/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-161-103:96894:97381 [6] NCCL INFO Channel 14/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-160-225:124160:124658 [0] NCCL INFO Channel 07/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-160-242:1364908:1365406 [6] NCCL INFO Channel 05/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-161-123:32710:33202 [6] NCCL INFO Channel 05/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-160-242:1364905:1365404 [3] NCCL INFO Channel 07/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-161-123:32707:33200 [3] NCCL INFO Channel 11/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-160-225:124162:124660 [2] NCCL INFO Channel 05/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-164:1227470:1227958 [7] NCCL INFO Channel 00/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-161-103:96893:97384 [5] NCCL INFO Channel 05/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96891:97385 [3] NCCL INFO Channel 03/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96893:97384 [5] NCCL INFO Channel 13/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364908:1365406 [6] NCCL INFO Channel 06/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-161-103:96891:97385 [3] NCCL INFO Channel 11/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364905:1365404 [3] NCCL INFO Channel 09/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-161-103:96893:97384 [5] NCCL INFO Channel 05/0 : 21[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96891:97385 [3] NCCL INFO Channel 03/0 : 19[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114516:115010 [4] NCCL INFO Channel 14/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-123:32710:33202 [6] NCCL INFO Channel 06/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-161-123:32707:33200 [3] NCCL INFO Channel 13/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-160-225:124160:124658 [0] NCCL INFO Channel 13/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-160-242:1364904:1365408 [2] NCCL INFO Channel 06/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-160-242:1364908:1365406 [6] NCCL INFO Channel 09/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-161-123:32710:33202 [6] NCCL INFO Channel 09/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-160-242:1364905:1365404 [3] NCCL INFO Channel 10/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-161-123:32707:33200 [3] NCCL INFO Channel 14/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-160-225:124162:124660 [2] NCCL INFO Channel 06/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-103:96889:97388 [1] NCCL INFO Channel 01/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-160-225:124165:124662 [5] NCCL INFO Channel 01/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-242:1364904:1365408 [2] NCCL INFO Channel 07/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-161-103:96889:97388 [1] NCCL INFO Channel 03/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-160-225:124160:124658 [0] NCCL INFO Channel 14/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-161-123:32710:33202 [6] NCCL INFO Channel 10/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-161-123:32707:33200 [3] NCCL INFO Channel 15/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-161-103:96889:97388 [1] NCCL INFO Channel 04/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-160-242:1364908:1365406 [6] NCCL INFO Channel 10/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-160-242:1364905:1365404 [3] NCCL INFO Channel 11/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-161-103:96889:97388 [1] NCCL INFO Channel 05/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-160-225:124162:124660 [2] NCCL INFO Channel 07/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-123:32710:33202 [6] NCCL INFO Channel 11/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-160-225:124165:124662 [5] NCCL INFO Channel 02/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-123:32706:33199 [2] NCCL INFO Channel 02/0 : 18[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:124160:124658 [0] NCCL INFO Channel 15/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-161-123:32710:33202 [6] NCCL INFO Channel 12/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-161-123:32708:33198 [4] NCCL INFO Channel 01/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-160-242:1364902:1365410 [0] NCCL INFO Channel 05/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-161-123:32710:33202 [6] NCCL INFO Channel 13/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-161-123:32708:33198 [4] NCCL INFO Channel 02/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-160-225:124162:124660 [2] NCCL INFO Channel 09/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-225:124165:124662 [5] NCCL INFO Channel 03/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-242:1364905:1365404 [3] NCCL INFO Channel 13/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-160-242:1364904:1365408 [2] NCCL INFO Channel 09/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-161-103:96889:97388 [1] NCCL INFO Channel 06/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-161-123:32710:33202 [6] NCCL INFO Channel 14/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-160-242:1364902:1365410 [0] NCCL INFO Channel 06/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-160-225:124164:124663 [4] NCCL INFO Channel 01/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-123:32708:33198 [4] NCCL INFO Channel 03/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-161-103:96889:97388 [1] NCCL INFO Channel 07/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-161-123:32708:33198 [4] NCCL INFO Channel 04/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-161-103:96889:97388 [1] NCCL INFO Channel 09/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-160-242:1364908:1365406 [6] NCCL INFO Channel 11/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-161-123:32708:33198 [4] NCCL INFO Channel 06/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-160-225:124162:124660 [2] NCCL INFO Channel 10/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-123:32710:33202 [6] NCCL INFO Channel 06/0 : 22[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:124165:124662 [5] NCCL INFO Channel 04/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-225:124164:124663 [4] NCCL INFO Channel 02/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-103:96889:97388 [1] NCCL INFO Channel 11/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-160-225:124166:124665 [6] NCCL INFO Channel 01/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-78:114516:115010 [4] NCCL INFO Channel 15/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-242:1364905:1365404 [3] NCCL INFO Channel 14/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-160-242:1364904:1365408 [2] NCCL INFO Channel 10/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-161-123:32708:33198 [4] NCCL INFO Channel 07/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-165-202:153569:154055 [7] NCCL INFO Channel 06/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-161-78:114519:115014 [7] NCCL INFO Channel 15/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-242:1364902:1365410 [0] NCCL INFO Channel 07/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-161-103:96889:97388 [1] NCCL INFO Channel 12/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-168-34:272155:272652 [5] NCCL INFO Connected all rings
+ip-26-0-161-123:32708:33198 [4] NCCL INFO Channel 09/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-161-103:96889:97388 [1] NCCL INFO Channel 13/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-161-123:32708:33198 [4] NCCL INFO Channel 10/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-160-225:124162:124660 [2] NCCL INFO Channel 12/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-242:1364908:1365406 [6] NCCL INFO Channel 12/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-160-225:124165:124662 [5] NCCL INFO Channel 05/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-225:124164:124663 [4] NCCL INFO Channel 03/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-103:96889:97388 [1] NCCL INFO Channel 14/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-160-225:124166:124665 [6] NCCL INFO Channel 02/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-123:32708:33198 [4] NCCL INFO Channel 11/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-160-242:1364905:1365404 [3] NCCL INFO Channel 15/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-160-242:1364904:1365408 [2] NCCL INFO Channel 12/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-160-225:124163:124661 [3] NCCL INFO Channel 01/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-103:96889:97388 [1] NCCL INFO Channel 15/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-161-123:32708:33198 [4] NCCL INFO Channel 12/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-168-34:272150:272645 [0] NCCL INFO Channel 03/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-161-103:96888:97387 [0] NCCL INFO Channel 05/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-160-242:1364908:1365406 [6] NCCL INFO Channel 13/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-161-123:32708:33198 [4] NCCL INFO Channel 14/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-160-225:124162:124660 [2] NCCL INFO Channel 13/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-103:96890:97386 [2] NCCL INFO Channel 02/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227470:1227958 [7] NCCL INFO Channel 06/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-161-103:96889:97388 [1] NCCL INFO Channel 01/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124164:124663 [4] NCCL INFO Channel 04/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-225:124165:124662 [5] NCCL INFO Channel 07/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-103:96890:97386 [2] NCCL INFO Channel 10/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96889:97388 [1] NCCL INFO Channel 09/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364904:1365408 [2] NCCL INFO Channel 13/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-161-103:96890:97386 [2] NCCL INFO Channel 02/0 : 18[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364902:1365410 [0] NCCL INFO Channel 13/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-160-225:124166:124665 [6] NCCL INFO Channel 03/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-103:96889:97388 [1] NCCL INFO Channel 01/0 : 17[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96888:97387 [0] NCCL INFO Channel 06/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-161-123:32708:33198 [4] NCCL INFO Channel 15/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-160-225:124163:124661 [3] NCCL INFO Channel 02/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-123:32711:33204 [7] NCCL INFO Channel 07/0 : 23[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96888:97387 [0] NCCL INFO Channel 07/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-161-123:32706:33199 [2] NCCL INFO Channel 10/0 : 10[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153569:154055 [7] NCCL INFO Channel 08/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-161-123:32706:33199 [2] NCCL INFO Channel 10/0 : 26[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364908:1365406 [6] NCCL INFO Channel 14/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-160-242:1364907:1365403 [5] NCCL INFO Channel 01/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-161-123:32707:33200 [3] NCCL INFO Channel 03/0 : 19[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364904:1365408 [2] NCCL INFO Channel 14/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-161-103:96888:97387 [0] NCCL INFO Channel 13/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-161-123:32709:33201 [5] NCCL INFO Channel 05/0 : 21[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32708:33198 [4] NCCL INFO Channel 04/0 : 20[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32707:33200 [3] NCCL INFO Channel 11/0 : 11[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:124162:124660 [2] NCCL INFO Channel 14/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-225:124164:124663 [4] NCCL INFO Channel 06/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-242:1364902:1365410 [0] NCCL INFO Channel 14/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-160-225:124165:124662 [5] NCCL INFO Channel 09/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-225:124166:124665 [6] NCCL INFO Channel 04/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-123:32707:33200 [3] NCCL INFO Channel 11/0 : 27[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96888:97387 [0] NCCL INFO Channel 14/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-161-123:32708:33198 [4] NCCL INFO Channel 12/0 : 12[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32709:33201 [5] NCCL INFO Channel 13/0 : 13[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32708:33198 [4] NCCL INFO Channel 12/0 : 28[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32709:33201 [5] NCCL INFO Channel 13/0 : 29[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124163:124661 [3] NCCL INFO Channel 03/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-168-34:272150:272645 [0] NCCL INFO Channel 04/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-161-123:32704:33197 [0] NCCL INFO Channel 00/0 : 16[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364907:1365403 [5] NCCL INFO Channel 02/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-160-242:1364904:1365408 [2] NCCL INFO Channel 15/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-160-242:1364902:1365410 [0] NCCL INFO Channel 15/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-168-34:272150:272645 [0] NCCL INFO Channel 05/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-160-225:124164:124663 [4] NCCL INFO Channel 07/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-225:124162:124660 [2] NCCL INFO Channel 15/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-225:124166:124665 [6] NCCL INFO Channel 05/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-225:124165:124662 [5] NCCL INFO Channel 10/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-225:124163:124661 [3] NCCL INFO Channel 05/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-103:96888:97387 [0] NCCL INFO Channel 15/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-160-242:1364907:1365403 [5] NCCL INFO Channel 03/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-168-34:272150:272645 [0] NCCL INFO Channel 06/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-161-103:96895:97382 [7] NCCL INFO Connected all rings
+ip-26-0-160-242:1364903:1365409 [1] NCCL INFO Channel 09/0 : 1[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364907:1365403 [5] NCCL INFO Channel 04/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-160-242:1364904:1365408 [2] NCCL INFO Channel 10/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364903:1365409 [1] NCCL INFO Channel 01/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364904:1365408 [2] NCCL INFO Channel 02/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364903:1365409 [1] NCCL INFO Channel 09/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364904:1365408 [2] NCCL INFO Channel 10/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:124164:124663 [4] NCCL INFO Channel 09/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-225:124166:124665 [6] NCCL INFO Channel 06/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-225:124165:124662 [5] NCCL INFO Channel 11/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-168-34:272150:272645 [0] NCCL INFO Channel 07/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-160-225:124163:124661 [3] NCCL INFO Channel 06/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-103:96889:97388 [1] NCCL INFO Channel 01/0 : 1[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364907:1365403 [5] NCCL INFO Channel 05/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-161-103:96890:97386 [2] NCCL INFO Channel 02/0 : 2[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96889:97388 [1] NCCL INFO Channel 01/0 : 17[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124161:124659 [1] NCCL INFO Channel 09/0 : 1[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96890:97386 [2] NCCL INFO Channel 02/0 : 18[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:124161:124659 [1] NCCL INFO Channel 01/0 : 17[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124161:124659 [1] NCCL INFO Channel 01/0 : 1[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364903:1365409 [1] NCCL INFO Channel 09/0 : 25[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364903:1365409 [1] NCCL INFO Channel 09/0 : 9[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272150:272645 [0] NCCL INFO Channel 10/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-160-225:124161:124659 [1] NCCL INFO Channel 09/0 : 9[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96889:97388 [1] NCCL INFO Channel 01/0 : 25[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96889:97388 [1] NCCL INFO Channel 01/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124164:124663 [4] NCCL INFO Channel 10/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-225:124166:124665 [6] NCCL INFO Channel 09/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-225:124165:124662 [5] NCCL INFO Channel 12/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-225:124163:124661 [3] NCCL INFO Channel 07/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-103:96889:97388 [1] NCCL INFO Channel 09/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364906:1365407 [4] NCCL INFO Connected all rings
+ip-26-0-168-34:272150:272645 [0] NCCL INFO Channel 11/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-161-123:32705:33203 [1] NCCL INFO Channel 09/0 : 9[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32705:33203 [1] NCCL INFO Channel 09/0 : 25[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364907:1365403 [5] NCCL INFO Channel 07/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-168-30:235649:236143 [1] NCCL INFO Connected all rings
+ip-26-0-161-123:32705:33203 [1] NCCL INFO Channel 01/0 : 25[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364903:1365409 [1] NCCL INFO Channel 01/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124165:124662 [5] NCCL INFO Channel 13/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-168-34:272150:272645 [0] NCCL INFO Channel 12/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-160-242:1364903:1365409 [1] NCCL INFO Channel 09/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124164:124663 [4] NCCL INFO Channel 11/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-225:124166:124665 [6] NCCL INFO Channel 10/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-225:124163:124661 [3] NCCL INFO Channel 09/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-242:1364903:1365409 [1] NCCL INFO Channel 09/0 : 9[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272150:272645 [0] NCCL INFO Channel 13/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-160-225:124164:124663 [4] NCCL INFO Channel 12/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-225:124165:124662 [5] NCCL INFO Channel 15/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-225:124166:124665 [6] NCCL INFO Channel 11/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-225:124163:124661 [3] NCCL INFO Channel 10/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-242:1364907:1365403 [5] NCCL INFO Channel 09/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-161-123:32705:33203 [1] NCCL INFO Channel 00/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-161-123:32705:33203 [1] NCCL INFO Channel 08/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-160-242:1364907:1365403 [5] NCCL INFO Channel 10/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-160-225:124164:124663 [4] NCCL INFO Channel 14/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-225:124166:124665 [6] NCCL INFO Channel 12/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-225:124163:124661 [3] NCCL INFO Channel 11/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-168-34:272152:272647 [2] NCCL INFO Connected all rings
+ip-26-0-168-34:272150:272645 [0] NCCL INFO Channel 14/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-160-225:124166:124665 [6] NCCL INFO Channel 13/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-225:124164:124663 [4] NCCL INFO Channel 15/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-225:124163:124661 [3] NCCL INFO Channel 13/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-242:1364907:1365403 [5] NCCL INFO Channel 11/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-168-34:272150:272645 [0] NCCL INFO Channel 15/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-165-164:1227470:1227958 [7] NCCL INFO Channel 08/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-161-103:96889:97388 [1] NCCL INFO Channel 00/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-160-225:124166:124665 [6] NCCL INFO Channel 14/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-225:124163:124661 [3] NCCL INFO Channel 14/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-242:1364907:1365403 [5] NCCL INFO Channel 12/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-160-225:124163:124661 [3] NCCL INFO Channel 15/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-225:124161:124659 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-242:1364907:1365403 [5] NCCL INFO Channel 13/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-161-103:96889:97388 [1] NCCL INFO Channel 08/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-160-225:124165:124662 [5] NCCL INFO Channel 13/0 : 5[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272155:272652 [5] NCCL INFO Channel 01/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-168-30:235649:236143 [1] NCCL INFO Channel 01/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-161-78:114519:115014 [7] NCCL INFO Channel 00/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-225:124166:124665 [6] NCCL INFO Channel 14/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:124161:124659 [1] NCCL INFO Channel 08/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-242:1364907:1365403 [5] NCCL INFO Channel 15/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-168-30:235649:236143 [1] NCCL INFO Channel 03/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-160-225:124162:124660 [2] NCCL INFO Channel 10/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272155:272652 [5] NCCL INFO Channel 02/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-160-225:124164:124663 [4] NCCL INFO Channel 12/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:124163:124661 [3] NCCL INFO Channel 11/0 : 3[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:124162:124660 [2] NCCL INFO Channel 02/0 : 18[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153569:154055 [7] NCCL INFO Channel 14/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-161-103:96894:97381 [6] NCCL INFO Channel 06/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364904:1365408 [2] NCCL INFO Channel 10/0 : 26[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:124162:124660 [2] NCCL INFO Channel 02/0 : 2[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235649:236143 [1] NCCL INFO Channel 04/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-161-103:96894:97381 [6] NCCL INFO Channel 14/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364904:1365408 [2] NCCL INFO Channel 10/0 : 10[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96894:97381 [6] NCCL INFO Channel 06/0 : 22[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:124162:124660 [2] NCCL INFO Channel 10/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96890:97386 [2] NCCL INFO Channel 02/0 : 26[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364904:1365408 [2] NCCL INFO Channel 02/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235649:236143 [1] NCCL INFO Channel 05/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-161-103:96890:97386 [2] NCCL INFO Channel 02/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364903:1365409 [1] NCCL INFO Channel 00/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-160-242:1364904:1365408 [2] NCCL INFO Channel 10/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96890:97386 [2] NCCL INFO Channel 10/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364904:1365408 [2] NCCL INFO Channel 10/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32710:33202 [6] NCCL INFO Channel 14/0 : 14[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32710:33202 [6] NCCL INFO Channel 14/0 : 30[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364903:1365409 [1] NCCL INFO Channel 08/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-168-30:235649:236143 [1] NCCL INFO Channel 06/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-168-34:272155:272652 [5] NCCL INFO Channel 03/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-160-225:124167:124664 [7] NCCL INFO Channel 15/0 : 7[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124160:124658 [0] NCCL INFO Channel 08/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235649:236143 [1] NCCL INFO Channel 07/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-161-103:96895:97382 [7] NCCL INFO Channel 07/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96888:97387 [0] NCCL INFO Channel 00/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96895:97382 [7] NCCL INFO Channel 15/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272155:272652 [5] NCCL INFO Channel 04/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-161-103:96888:97387 [0] NCCL INFO Channel 08/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96895:97382 [7] NCCL INFO Channel 07/0 : 23[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96888:97387 [0] NCCL INFO Channel 00/0 : 16[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235649:236143 [1] NCCL INFO Channel 09/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-161-123:32706:33199 [2] NCCL INFO Channel 02/0 : 26[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235649:236143 [1] NCCL INFO Channel 11/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-161-123:32711:33204 [7] NCCL INFO Channel 15/0 : 15[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124162:124660 [2] NCCL INFO Channel 00/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-242:1364908:1365406 [6] NCCL INFO Channel 14/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32711:33204 [7] NCCL INFO Channel 15/0 : 31[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364908:1365406 [6] NCCL INFO Channel 06/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364908:1365406 [6] NCCL INFO Channel 14/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32704:33197 [0] NCCL INFO Channel 08/0 : 8[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32704:33197 [0] NCCL INFO Channel 08/0 : 24[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272155:272652 [5] NCCL INFO Channel 05/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-160-225:124162:124660 [2] NCCL INFO Channel 01/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-225:124166:124665 [6] NCCL INFO Channel 06/0 : 22[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364906:1365407 [4] NCCL INFO Channel 01/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-168-30:235649:236143 [1] NCCL INFO Channel 12/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-160-225:124166:124665 [6] NCCL INFO Channel 06/0 : 6[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364908:1365406 [6] NCCL INFO Channel 14/0 : 30[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364908:1365406 [6] NCCL INFO Channel 14/0 : 14[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:124162:124660 [2] NCCL INFO Channel 08/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-168-34:272155:272652 [5] NCCL INFO Channel 07/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-168-30:235649:236143 [1] NCCL INFO Channel 13/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-160-242:1364908:1365406 [6] NCCL INFO Channel 06/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364904:1365408 [2] NCCL INFO Channel 00/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-160-225:124162:124660 [2] NCCL INFO Channel 09/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-123:32710:33202 [6] NCCL INFO Channel 06/0 : 30[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364908:1365406 [6] NCCL INFO Channel 14/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364908:1365406 [6] NCCL INFO Channel 14/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364906:1365407 [4] NCCL INFO Channel 02/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-161-123:32706:33199 [2] NCCL INFO Channel 00/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-161-123:32706:33199 [2] NCCL INFO Channel 01/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-160-242:1364904:1365408 [2] NCCL INFO Channel 01/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-168-34:272155:272652 [5] NCCL INFO Channel 09/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-161-103:96894:97381 [6] NCCL INFO Channel 06/0 : 6[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235649:236143 [1] NCCL INFO Channel 14/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-161-103:96894:97381 [6] NCCL INFO Channel 06/0 : 22[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32706:33199 [2] NCCL INFO Channel 08/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-161-103:96890:97386 [2] NCCL INFO Channel 00/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-160-242:1364906:1365407 [4] NCCL INFO Channel 03/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-160-225:124166:124665 [6] NCCL INFO Channel 14/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235649:236143 [1] NCCL INFO Channel 15/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-161-103:96894:97381 [6] NCCL INFO Channel 06/0 : 30[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114519:115014 [7] NCCL INFO Channel 06/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-103:96894:97381 [6] NCCL INFO Channel 06/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96894:97381 [6] NCCL INFO Channel 14/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32706:33199 [2] NCCL INFO Channel 09/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-160-242:1364904:1365408 [2] NCCL INFO Channel 08/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-161-103:96890:97386 [2] NCCL INFO Channel 01/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-168-34:272155:272652 [5] NCCL INFO Channel 10/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-160-242:1364906:1365407 [4] NCCL INFO Channel 04/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-161-103:96890:97386 [2] NCCL INFO Channel 08/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-168-34:272151:272646 [1] NCCL INFO Channel 01/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-160-242:1364904:1365408 [2] NCCL INFO Channel 09/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-168-34:272155:272652 [5] NCCL INFO Channel 11/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-160-225:124166:124665 [6] NCCL INFO Channel 00/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-168-30:235655:236142 [7] NCCL INFO Connected all rings
+ip-26-0-160-242:1364906:1365407 [4] NCCL INFO Channel 06/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-161-123:32710:33202 [6] NCCL INFO Channel 00/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-168-34:272151:272646 [1] NCCL INFO Channel 03/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-160-225:124166:124665 [6] NCCL INFO Channel 05/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-103:96890:97386 [2] NCCL INFO Channel 09/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-161-123:32710:33202 [6] NCCL INFO Channel 05/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-168-34:272152:272647 [2] NCCL INFO Channel 01/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-160-225:124166:124665 [6] NCCL INFO Channel 08/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-168-52:2722046:2722541 [1] NCCL INFO Channel 01/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-168-34:272155:272652 [5] NCCL INFO Channel 12/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-160-242:1364906:1365407 [4] NCCL INFO Channel 07/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-168-34:272151:272646 [1] NCCL INFO Channel 04/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-161-123:32710:33202 [6] NCCL INFO Channel 08/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-160-225:124166:124665 [6] NCCL INFO Channel 13/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-242:1364909:1365405 [7] NCCL INFO Channel 15/0 : 7[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364902:1365410 [0] NCCL INFO Channel 08/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364909:1365405 [7] NCCL INFO Channel 07/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364902:1365410 [0] NCCL INFO Channel 00/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364909:1365405 [7] NCCL INFO Channel 15/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364906:1365407 [4] NCCL INFO Channel 09/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-161-123:32710:33202 [6] NCCL INFO Channel 13/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-160-242:1364902:1365410 [0] NCCL INFO Channel 08/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:124160:124658 [0] NCCL INFO Channel 00/0 : 16[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272152:272647 [2] NCCL INFO Channel 02/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-161-103:96894:97381 [6] NCCL INFO Channel 00/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-168-52:2722046:2722541 [1] NCCL INFO Channel 03/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-160-242:1364902:1365410 [0] NCCL INFO Channel 08/0 : 24[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:124160:124658 [0] NCCL INFO Channel 00/0 : 0[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272151:272646 [1] NCCL INFO Channel 05/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-168-34:272155:272652 [5] NCCL INFO Channel 13/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-160-242:1364909:1365405 [7] NCCL INFO Channel 15/0 : 31[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364906:1365407 [4] NCCL INFO Channel 10/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-161-103:96888:97387 [0] NCCL INFO Channel 00/0 : 0[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96895:97382 [7] NCCL INFO Channel 07/0 : 7[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364902:1365410 [0] NCCL INFO Channel 08/0 : 8[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364908:1365406 [6] NCCL INFO Channel 00/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-160-242:1364909:1365405 [7] NCCL INFO Channel 15/0 : 15[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96895:97382 [7] NCCL INFO Channel 07/0 : 23[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96888:97387 [0] NCCL INFO Channel 00/0 : 16[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96894:97381 [6] NCCL INFO Channel 05/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-160-242:1364902:1365410 [0] NCCL INFO Channel 00/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364909:1365405 [7] NCCL INFO Channel 07/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364902:1365410 [0] NCCL INFO Channel 08/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:124160:124658 [0] NCCL INFO Channel 08/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96888:97387 [0] NCCL INFO Channel 00/0 : 24[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364909:1365405 [7] NCCL INFO Channel 15/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364902:1365410 [0] NCCL INFO Channel 08/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96888:97387 [0] NCCL INFO Channel 00/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364909:1365405 [7] NCCL INFO Channel 15/0 : 15[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364906:1365407 [4] NCCL INFO Channel 11/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-161-103:96888:97387 [0] NCCL INFO Channel 08/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364908:1365406 [6] NCCL INFO Channel 05/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-161-103:96894:97381 [6] NCCL INFO Channel 08/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-160-225:124167:124664 [7] NCCL INFO Channel 07/0 : 23[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124167:124664 [7] NCCL INFO Channel 07/0 : 7[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272152:272647 [2] NCCL INFO Channel 04/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-168-34:272151:272646 [1] NCCL INFO Channel 06/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-161-103:96894:97381 [6] NCCL INFO Channel 13/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-161-123:32711:33204 [7] NCCL INFO Channel 07/0 : 31[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272155:272652 [5] NCCL INFO Channel 15/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-160-225:124167:124664 [7] NCCL INFO Channel 15/0 : 15[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96895:97382 [7] NCCL INFO Channel 07/0 : 31[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722046:2722541 [1] NCCL INFO Channel 04/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-161-103:96895:97382 [7] NCCL INFO Channel 07/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364906:1365407 [4] NCCL INFO Channel 12/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-160-225:124167:124664 [7] NCCL INFO Channel 01/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-103:96895:97382 [7] NCCL INFO Channel 15/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364908:1365406 [6] NCCL INFO Channel 08/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-161-123:32711:33204 [7] NCCL INFO Channel 01/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-161-103:96895:97382 [7] NCCL INFO Channel 01/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-160-242:1364909:1365405 [7] NCCL INFO Channel 01/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-160-225:124167:124664 [7] NCCL INFO Channel 02/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-168-52:2722046:2722541 [1] NCCL INFO Channel 05/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-168-34:272152:272647 [2] NCCL INFO Channel 05/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-168-34:272151:272646 [1] NCCL INFO Channel 07/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-161-123:32711:33204 [7] NCCL INFO Channel 02/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-161-103:96895:97382 [7] NCCL INFO Channel 02/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-160-242:1364906:1365407 [4] NCCL INFO Channel 14/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-160-242:1364908:1365406 [6] NCCL INFO Channel 13/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-160-225:124167:124664 [7] NCCL INFO Channel 03/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-103:96895:97382 [7] NCCL INFO Channel 03/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-161-123:32711:33204 [7] NCCL INFO Channel 03/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-168-34:272152:272647 [2] NCCL INFO Channel 06/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-160-242:1364909:1365405 [7] NCCL INFO Channel 02/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-168-34:272151:272646 [1] NCCL INFO Channel 09/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-161-103:96895:97382 [7] NCCL INFO Channel 04/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-161-123:32711:33204 [7] NCCL INFO Channel 04/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-160-225:124167:124664 [7] NCCL INFO Channel 04/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-123:32711:33204 [7] NCCL INFO Channel 05/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-160-242:1364906:1365407 [4] NCCL INFO Channel 15/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-161-123:32704:33197 [0] NCCL INFO Channel 00/0 : 24[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364909:1365405 [7] NCCL INFO Channel 03/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-168-34:272152:272647 [2] NCCL INFO Channel 07/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-168-34:272151:272646 [1] NCCL INFO Channel 11/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-161-123:32711:33204 [7] NCCL INFO Channel 06/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-168-30:235652:236147 [4] NCCL INFO Connected all rings
+ip-26-0-160-242:1364909:1365405 [7] NCCL INFO Channel 04/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-168-34:272152:272647 [2] NCCL INFO Channel 09/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-168-34:272151:272646 [1] NCCL INFO Channel 12/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-160-242:1364906:1365407 [4] NCCL INFO Channel 12/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364907:1365403 [5] NCCL INFO Channel 13/0 : 5[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32711:33204 [7] NCCL INFO Channel 07/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-160-242:1364907:1365403 [5] NCCL INFO Channel 05/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364906:1365407 [4] NCCL INFO Channel 04/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364909:1365405 [7] NCCL INFO Channel 05/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-160-242:1364907:1365403 [5] NCCL INFO Channel 13/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364906:1365407 [4] NCCL INFO Channel 12/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96895:97382 [7] NCCL INFO Channel 05/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-160-242:1364905:1365404 [3] NCCL INFO Channel 11/0 : 3[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364905:1365404 [3] NCCL INFO Channel 03/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364905:1365404 [3] NCCL INFO Channel 11/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364909:1365405 [7] NCCL INFO Channel 06/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-160-225:124164:124663 [4] NCCL INFO Channel 04/0 : 20[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:124165:124662 [5] NCCL INFO Channel 05/0 : 21[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32711:33204 [7] NCCL INFO Channel 09/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-161-103:96895:97382 [7] NCCL INFO Channel 06/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-160-225:124165:124662 [5] NCCL INFO Channel 05/0 : 5[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124164:124663 [4] NCCL INFO Channel 04/0 : 4[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364906:1365407 [4] NCCL INFO Channel 12/0 : 28[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272152:272647 [2] NCCL INFO Channel 10/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-161-103:96892:97383 [4] NCCL INFO Channel 04/0 : 4[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96893:97384 [5] NCCL INFO Channel 05/0 : 5[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364907:1365403 [5] NCCL INFO Channel 13/0 : 29[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124163:124661 [3] NCCL INFO Channel 03/0 : 19[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272151:272646 [1] NCCL INFO Channel 13/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-161-103:96892:97383 [4] NCCL INFO Channel 04/0 : 20[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364906:1365407 [4] NCCL INFO Channel 12/0 : 12[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364907:1365403 [5] NCCL INFO Channel 13/0 : 13[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364909:1365405 [7] NCCL INFO Channel 07/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-160-225:124163:124661 [3] NCCL INFO Channel 03/0 : 3[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96893:97384 [5] NCCL INFO Channel 05/0 : 21[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364905:1365404 [3] NCCL INFO Channel 11/0 : 27[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96891:97385 [3] NCCL INFO Channel 03/0 : 3[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364905:1365404 [3] NCCL INFO Channel 11/0 : 11[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96895:97382 [7] NCCL INFO Channel 07/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-160-242:1364906:1365407 [4] NCCL INFO Channel 04/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364907:1365403 [5] NCCL INFO Channel 05/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32709:33201 [5] NCCL INFO Channel 05/0 : 29[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32708:33198 [4] NCCL INFO Channel 04/0 : 28[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:124165:124662 [5] NCCL INFO Channel 13/0 : 13[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124164:124663 [4] NCCL INFO Channel 12/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96892:97383 [4] NCCL INFO Channel 04/0 : 28[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364906:1365407 [4] NCCL INFO Channel 12/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364907:1365403 [5] NCCL INFO Channel 13/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32711:33204 [7] NCCL INFO Channel 10/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-161-103:96891:97385 [3] NCCL INFO Channel 03/0 : 19[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96893:97384 [5] NCCL INFO Channel 05/0 : 29[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364906:1365407 [4] NCCL INFO Channel 12/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364907:1365403 [5] NCCL INFO Channel 13/0 : 13[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96892:97383 [4] NCCL INFO Channel 04/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96893:97384 [5] NCCL INFO Channel 05/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364905:1365404 [3] NCCL INFO Channel 03/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272152:272647 [2] NCCL INFO Channel 12/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-168-34:272151:272646 [1] NCCL INFO Channel 14/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-161-103:96892:97383 [4] NCCL INFO Channel 12/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32707:33200 [3] NCCL INFO Channel 03/0 : 27[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96893:97384 [5] NCCL INFO Channel 13/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364909:1365405 [7] NCCL INFO Channel 09/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-160-242:1364905:1365404 [3] NCCL INFO Channel 11/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:124163:124661 [3] NCCL INFO Channel 11/0 : 11[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96891:97385 [3] NCCL INFO Channel 03/0 : 27[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364905:1365404 [3] NCCL INFO Channel 11/0 : 11[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96895:97382 [7] NCCL INFO Channel 09/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-161-103:96891:97385 [3] NCCL INFO Channel 03/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96891:97385 [3] NCCL INFO Channel 11/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32711:33204 [7] NCCL INFO Channel 11/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-160-242:1364909:1365405 [7] NCCL INFO Channel 10/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-168-34:272152:272647 [2] NCCL INFO Channel 13/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-168-52:2722046:2722541 [1] NCCL INFO Channel 06/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-168-34:272151:272646 [1] NCCL INFO Channel 15/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-161-103:96895:97382 [7] NCCL INFO Channel 10/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-161-78:114519:115014 [7] NCCL INFO Channel 08/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-225:124167:124664 [7] NCCL INFO Channel 05/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-168-52:2722046:2722541 [1] NCCL INFO Channel 07/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-161-123:32711:33204 [7] NCCL INFO Channel 12/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-168-34:272152:272647 [2] NCCL INFO Channel 14/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-160-242:1364909:1365405 [7] NCCL INFO Channel 11/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-168-34:272150:272645 [0] NCCL INFO Channel 05/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-161-103:96895:97382 [7] NCCL INFO Channel 11/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-168-34:272152:272647 [2] NCCL INFO Channel 15/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-168-30:235652:236147 [4] NCCL INFO Channel 01/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-161-123:32711:33204 [7] NCCL INFO Channel 13/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-168-34:272150:272645 [0] NCCL INFO Channel 06/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-160-225:124167:124664 [7] NCCL INFO Channel 06/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-242:1364909:1365405 [7] NCCL INFO Channel 12/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-168-30:235652:236147 [4] NCCL INFO Channel 02/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-161-103:96895:97382 [7] NCCL INFO Channel 12/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-168-34:272150:272645 [0] NCCL INFO Channel 07/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-168-34:272151:272646 [1] NCCL INFO Channel 01/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235652:236147 [4] NCCL INFO Channel 03/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-168-34:272151:272646 [1] NCCL INFO Channel 09/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272151:272646 [1] NCCL INFO Channel 01/0 : 17[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-52:2722046:2722541 [1] NCCL INFO Channel 09/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-168-34:272150:272645 [0] NCCL INFO Channel 13/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-161-103:96895:97382 [7] NCCL INFO Channel 13/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-161-123:32711:33204 [7] NCCL INFO Channel 14/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-168-30:235652:236147 [4] NCCL INFO Channel 04/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-160-225:124167:124664 [7] NCCL INFO Channel 07/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-242:1364909:1365405 [7] NCCL INFO Channel 13/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-161-103:96893:97384 [5] NCCL INFO Channel 00/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-168-34:272150:272645 [0] NCCL INFO Channel 14/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-161-103:96895:97382 [7] NCCL INFO Channel 14/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-168-30:235652:236147 [4] NCCL INFO Channel 06/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-168-52:2722046:2722541 [1] NCCL INFO Channel 11/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-160-225:124165:124662 [5] NCCL INFO Channel 00/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-225:124164:124663 [4] NCCL INFO Channel 00/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-123:32711:33204 [7] NCCL INFO Channel 15/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-168-34:272150:272645 [0] NCCL INFO Channel 15/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-160-242:1364907:1365403 [5] NCCL INFO Channel 00/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-168-30:235652:236147 [4] NCCL INFO Channel 07/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-161-103:96893:97384 [5] NCCL INFO Channel 04/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-161-103:96895:97382 [7] NCCL INFO Channel 15/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-168-30:235652:236147 [4] NCCL INFO Channel 09/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-160-242:1364906:1365407 [4] NCCL INFO Channel 00/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-160-225:124167:124664 [7] NCCL INFO Channel 09/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-242:1364909:1365405 [7] NCCL INFO Channel 14/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-161-123:32709:33201 [5] NCCL INFO Channel 00/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-168-30:235652:236147 [4] NCCL INFO Channel 10/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-161-103:96892:97383 [4] NCCL INFO Channel 00/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-160-225:124164:124663 [4] NCCL INFO Channel 03/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-123:32708:33198 [4] NCCL INFO Channel 00/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-160-225:124165:124662 [5] NCCL INFO Channel 04/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-168-30:235652:236147 [4] NCCL INFO Channel 11/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-160-242:1364907:1365403 [5] NCCL INFO Channel 04/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-168-52:2722046:2722541 [1] NCCL INFO Channel 12/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-168-30:235652:236147 [4] NCCL INFO Channel 12/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-161-103:96893:97384 [5] NCCL INFO Channel 08/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-161-123:32709:33201 [5] NCCL INFO Channel 04/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-160-225:124167:124664 [7] NCCL INFO Channel 10/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-242:1364906:1365407 [4] NCCL INFO Channel 03/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-168-30:235652:236147 [4] NCCL INFO Channel 14/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-160-242:1364909:1365405 [7] NCCL INFO Channel 15/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-161-103:96892:97383 [4] NCCL INFO Channel 03/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-161-123:32708:33198 [4] NCCL INFO Channel 03/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-160-225:124164:124663 [4] NCCL INFO Channel 05/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-225:124165:124662 [5] NCCL INFO Channel 08/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-168-30:235652:236147 [4] NCCL INFO Channel 15/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-161-103:96893:97384 [5] NCCL INFO Channel 12/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-161-123:32709:33201 [5] NCCL INFO Channel 08/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-160-225:124167:124664 [7] NCCL INFO Channel 11/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-103:96892:97383 [4] NCCL INFO Channel 05/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-161-123:32708:33198 [4] NCCL INFO Channel 05/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-160-242:1364906:1365407 [4] NCCL INFO Channel 05/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-160-225:124164:124663 [4] NCCL INFO Channel 06/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-225:124165:124662 [5] NCCL INFO Channel 12/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-242:1364906:1365407 [4] NCCL INFO Channel 06/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-168-52:2722046:2722541 [1] NCCL INFO Channel 13/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-161-103:96892:97383 [4] NCCL INFO Channel 06/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-161-123:32709:33201 [5] NCCL INFO Channel 12/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-165-164:1227470:1227958 [7] NCCL INFO Channel 14/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-161-123:32708:33198 [4] NCCL INFO Channel 06/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-160-225:124167:124664 [7] NCCL INFO Channel 12/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-242:1364906:1365407 [4] NCCL INFO Channel 07/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-160-242:1364907:1365403 [5] NCCL INFO Channel 08/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-160-225:124164:124663 [4] NCCL INFO Channel 07/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-103:96892:97383 [4] NCCL INFO Channel 07/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-161-123:32708:33198 [4] NCCL INFO Channel 07/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-160-225:124167:124664 [7] NCCL INFO Channel 13/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-242:1364906:1365407 [4] NCCL INFO Channel 08/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-160-225:124164:124663 [4] NCCL INFO Channel 08/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-242:1364907:1365403 [5] NCCL INFO Channel 12/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-161-103:96892:97383 [4] NCCL INFO Channel 08/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-161-123:32708:33198 [4] NCCL INFO Channel 08/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-160-242:1364905:1365404 [3] NCCL INFO Channel 00/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-168-52:2722046:2722541 [1] NCCL INFO Channel 14/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-160-225:124167:124664 [7] NCCL INFO Channel 14/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-123:32708:33198 [4] NCCL INFO Channel 11/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-160-225:124164:124663 [4] NCCL INFO Channel 11/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-103:96892:97383 [4] NCCL INFO Channel 11/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-160-242:1364906:1365407 [4] NCCL INFO Channel 11/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-168-52:2722046:2722541 [1] NCCL INFO Channel 15/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-160-242:1364905:1365404 [3] NCCL INFO Channel 02/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-161-123:32708:33198 [4] NCCL INFO Channel 13/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-161-103:96892:97383 [4] NCCL INFO Channel 13/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-161-123:32707:33200 [3] NCCL INFO Channel 00/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-160-225:124167:124664 [7] NCCL INFO Channel 15/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-225:124163:124661 [3] NCCL INFO Channel 00/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-225:124164:124663 [4] NCCL INFO Channel 13/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-242:1364906:1365407 [4] NCCL INFO Channel 13/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-161-103:96895:97382 [7] NCCL INFO Channel 00/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-160-242:1364906:1365407 [4] NCCL INFO Channel 14/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-161-123:32708:33198 [4] NCCL INFO Channel 14/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-161-123:32707:33200 [3] NCCL INFO Channel 02/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-161-103:96891:97385 [3] NCCL INFO Channel 00/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-161-103:96892:97383 [4] NCCL INFO Channel 14/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-161-123:32711:33204 [7] NCCL INFO Channel 00/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-160-225:124163:124661 [3] NCCL INFO Channel 02/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-225:124167:124664 [7] NCCL INFO Channel 00/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-242:1364909:1365405 [7] NCCL INFO Channel 00/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-160-242:1364906:1365407 [4] NCCL INFO Channel 15/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-161-103:96895:97382 [7] NCCL INFO Channel 06/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-161-103:96891:97385 [3] NCCL INFO Channel 02/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-161-103:96892:97383 [4] NCCL INFO Channel 15/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-161-123:32707:33200 [3] NCCL INFO Channel 08/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-160-225:124163:124661 [3] NCCL INFO Channel 08/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-225:124167:124664 [7] NCCL INFO Channel 06/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-123:32708:33198 [4] NCCL INFO Channel 15/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-160-242:1364909:1365405 [7] NCCL INFO Channel 06/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-168-52:2722045:2722544 [0] NCCL INFO Channel 05/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-161-123:32711:33204 [7] NCCL INFO Channel 06/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-161-103:96895:97382 [7] NCCL INFO Channel 08/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-161-103:96891:97385 [3] NCCL INFO Channel 08/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-160-225:124164:124663 [4] NCCL INFO Channel 14/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-225:124163:124661 [3] NCCL INFO Channel 10/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-225:124167:124664 [7] NCCL INFO Channel 08/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-123:32707:33200 [3] NCCL INFO Channel 10/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-161-123:32711:33204 [7] NCCL INFO Channel 08/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-160-242:1364905:1365404 [3] NCCL INFO Channel 08/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-160-242:1364909:1365405 [7] NCCL INFO Channel 08/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-161-103:96895:97382 [7] NCCL INFO Channel 14/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-168-52:2722045:2722544 [0] NCCL INFO Channel 06/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-160-225:124164:124663 [4] NCCL INFO Channel 15/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-225:124167:124664 [7] NCCL INFO Channel 14/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-242:1364905:1365404 [3] NCCL INFO Channel 10/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-168-52:2722045:2722544 [0] NCCL INFO Channel 07/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-161-123:32711:33204 [7] NCCL INFO Channel 14/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-168-52:2722045:2722544 [0] NCCL INFO Channel 13/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-161-103:96891:97385 [3] NCCL INFO Channel 10/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-160-242:1364909:1365405 [7] NCCL INFO Channel 14/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-161-78:114519:115014 [7] NCCL INFO Channel 14/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-168-52:2722045:2722544 [0] NCCL INFO Channel 14/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-168-52:2722045:2722544 [0] NCCL INFO Channel 15/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-165-59:1557266:1557759 [0] NCCL INFO Connected all rings
+ip-26-0-165-59:1557266:1557759 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-59:1557266:1557759 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-59:1557266:1557759 [0] NCCL INFO Channel 04/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-59:1557266:1557759 [0] NCCL INFO Channel 05/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-59:1557266:1557759 [0] NCCL INFO Channel 06/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-59:1557266:1557759 [0] NCCL INFO Channel 07/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-59:1557266:1557759 [0] NCCL INFO Channel 10/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-168-52:2722049:2722547 [4] NCCL INFO Connected all rings
+ip-26-0-168-34:272156:272648 [6] NCCL INFO Connected all rings
+ip-26-0-168-52:2722049:2722547 [4] NCCL INFO Channel 01/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-168-52:2722049:2722547 [4] NCCL INFO Channel 02/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-168-52:2722049:2722547 [4] NCCL INFO Channel 03/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-168-52:2722049:2722547 [4] NCCL INFO Channel 04/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-168-52:2722049:2722547 [4] NCCL INFO Channel 06/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-161-178:2815963:2816477 [0] NCCL INFO Connected all rings
+ip-26-0-161-178:2815963:2816477 [0] NCCL INFO Channel 02/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-168-52:2722049:2722547 [4] NCCL INFO Channel 07/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-168-52:2722049:2722547 [4] NCCL INFO Channel 09/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-165-59:1557268:1557764 [2] NCCL INFO Connected all rings
+ip-26-0-168-52:2722049:2722547 [4] NCCL INFO Channel 10/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-168-52:2722049:2722547 [4] NCCL INFO Channel 11/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-165-59:1557271:1557762 [5] NCCL INFO Connected all rings
+ip-26-0-165-131:1630212:1630734 [7] NCCL INFO Channel 00/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-161-142:62883:63373 [2] NCCL INFO Connected all rings
+ip-26-0-168-52:2722049:2722547 [4] NCCL INFO Channel 12/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-168-30:235650:236144 [2] NCCL INFO Connected all rings
+ip-26-0-165-131:1630212:1630734 [7] NCCL INFO Channel 06/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-168-34:272157:272649 [7] NCCL INFO Connected all rings
+ip-26-0-168-30:235654:236146 [6] NCCL INFO Connected all rings
+ip-26-0-161-142:62883:63373 [2] NCCL INFO Channel 01/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-168-30:235650:236144 [2] NCCL INFO Channel 01/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-161-142:62883:63373 [2] NCCL INFO Channel 02/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-168-30:235650:236144 [2] NCCL INFO Channel 02/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-161-142:62883:63373 [2] NCCL INFO Channel 04/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-168-30:235650:236144 [2] NCCL INFO Channel 04/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-168-30:235648:236145 [0] NCCL INFO Connected all rings
+ip-26-0-168-30:235648:236145 [0] NCCL INFO Channel 02/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-168-30:235650:236144 [2] NCCL INFO Channel 05/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-161-142:62883:63373 [2] NCCL INFO Channel 05/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-168-30:235648:236145 [0] NCCL INFO Channel 03/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-168-30:235650:236144 [2] NCCL INFO Channel 06/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-168-34:272156:272648 [6] NCCL INFO Channel 01/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-161-153:2636085:2636604 [1] NCCL INFO Connected all rings
+ip-26-0-161-142:62883:63373 [2] NCCL INFO Channel 06/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-168-30:235648:236145 [0] NCCL INFO Channel 04/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-168-30:235650:236144 [2] NCCL INFO Channel 07/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-168-34:272153:272651 [3] NCCL INFO Connected all rings
+ip-26-0-168-34:272156:272648 [6] NCCL INFO Channel 02/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-161-142:62883:63373 [2] NCCL INFO Channel 07/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-168-30:235648:236145 [0] NCCL INFO Channel 05/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-168-30:235650:236144 [2] NCCL INFO Channel 09/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-168-34:272156:272648 [6] NCCL INFO Channel 03/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-168-34:272156:272648 [6] NCCL INFO Channel 04/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-168-30:235648:236145 [0] NCCL INFO Channel 06/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-168-30:235650:236144 [2] NCCL INFO Channel 10/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-168-34:272156:272648 [6] NCCL INFO Channel 05/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-161-142:62883:63373 [2] NCCL INFO Channel 09/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-168-30:235653:236140 [5] NCCL INFO Connected all rings
+ip-26-0-168-34:272156:272648 [6] NCCL INFO Channel 06/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-168-30:235648:236145 [0] NCCL INFO Channel 07/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-168-30:235650:236144 [2] NCCL INFO Channel 12/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-161-142:62883:63373 [2] NCCL INFO Channel 10/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-168-34:272156:272648 [6] NCCL INFO Channel 09/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-168-34:272156:272648 [6] NCCL INFO Channel 10/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-168-30:235648:236145 [0] NCCL INFO Channel 10/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-168-30:235650:236144 [2] NCCL INFO Channel 13/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-161-142:62883:63373 [2] NCCL INFO Channel 12/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-168-30:235650:236144 [2] NCCL INFO Channel 14/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-168-30:235648:236145 [0] NCCL INFO Channel 11/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-168-30:235654:236146 [6] NCCL INFO Channel 01/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-165-59:1557266:1557759 [0] NCCL INFO Channel 11/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-168-34:272156:272648 [6] NCCL INFO Channel 11/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-168-30:235650:236144 [2] NCCL INFO Channel 15/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-168-30:235648:236145 [0] NCCL INFO Channel 12/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-168-34:272156:272648 [6] NCCL INFO Channel 12/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-168-30:235654:236146 [6] NCCL INFO Channel 02/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-161-142:62883:63373 [2] NCCL INFO Channel 13/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-168-34:272156:272648 [6] NCCL INFO Channel 13/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-168-30:235648:236145 [0] NCCL INFO Channel 13/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-168-30:235654:236146 [6] NCCL INFO Channel 03/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-168-30:235648:236145 [0] NCCL INFO Channel 14/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-168-30:235654:236146 [6] NCCL INFO Channel 04/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-168-34:272154:272650 [4] NCCL INFO Connected all rings
+ip-26-0-168-34:272156:272648 [6] NCCL INFO Channel 14/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-168-30:235648:236145 [0] NCCL INFO Channel 15/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-168-30:235654:236146 [6] NCCL INFO Channel 05/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-168-30:235654:236146 [6] NCCL INFO Channel 06/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-168-30:235648:236145 [0] NCCL INFO Channel 05/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-168-30:235649:236143 [1] NCCL INFO Channel 09/0 : 1[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235649:236143 [1] NCCL INFO Channel 01/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235649:236143 [1] NCCL INFO Channel 09/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235654:236146 [6] NCCL INFO Channel 09/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-168-30:235648:236145 [0] NCCL INFO Channel 06/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-165-59:1557266:1557759 [0] NCCL INFO Channel 12/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-59:1557266:1557759 [0] NCCL INFO Channel 13/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-168-30:235654:236146 [6] NCCL INFO Channel 10/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-168-30:235648:236145 [0] NCCL INFO Channel 07/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-161-178:2815965:2816475 [2] NCCL INFO Connected all rings
+ip-26-0-168-30:235648:236145 [0] NCCL INFO Channel 13/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-168-30:235654:236146 [6] NCCL INFO Channel 11/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-161-142:62883:63373 [2] NCCL INFO Channel 14/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-168-30:235648:236145 [0] NCCL INFO Channel 14/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-168-30:235654:236146 [6] NCCL INFO Channel 12/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-168-52:2722051:2722543 [6] NCCL INFO Connected all rings
+ip-26-0-168-30:235653:236140 [5] NCCL INFO Channel 01/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-168-30:235648:236145 [0] NCCL INFO Channel 15/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-165-59:1557266:1557759 [0] NCCL INFO Channel 14/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-168-30:235654:236146 [6] NCCL INFO Channel 13/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-161-142:62883:63373 [2] NCCL INFO Channel 15/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-168-30:235653:236140 [5] NCCL INFO Channel 02/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-165-131:1630212:1630734 [7] NCCL INFO Channel 08/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-168-34:272153:272651 [3] NCCL INFO Channel 01/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-168-30:235654:236146 [6] NCCL INFO Channel 14/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-168-30:235653:236140 [5] NCCL INFO Channel 03/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-168-34:272153:272651 [3] NCCL INFO Channel 02/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-165-59:1557266:1557759 [0] NCCL INFO Channel 15/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-168-30:235653:236140 [5] NCCL INFO Channel 04/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-168-34:272153:272651 [3] NCCL INFO Channel 03/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-168-30:235653:236140 [5] NCCL INFO Channel 05/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-168-30:235653:236140 [5] NCCL INFO Channel 07/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-168-34:272153:272651 [3] NCCL INFO Channel 05/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-168-34:272154:272650 [4] NCCL INFO Channel 01/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-168-30:235653:236140 [5] NCCL INFO Channel 09/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-168-34:272156:272648 [6] NCCL INFO Channel 06/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272156:272648 [6] NCCL INFO Channel 14/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272156:272648 [6] NCCL INFO Channel 06/0 : 22[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235653:236140 [5] NCCL INFO Channel 10/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-168-34:272153:272651 [3] NCCL INFO Channel 06/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-168-34:272154:272650 [4] NCCL INFO Channel 02/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-168-30:235653:236140 [5] NCCL INFO Channel 11/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-168-34:272153:272651 [3] NCCL INFO Channel 07/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-168-34:272154:272650 [4] NCCL INFO Channel 03/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-168-30:235653:236140 [5] NCCL INFO Channel 12/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-168-34:272154:272650 [4] NCCL INFO Channel 04/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-168-34:272153:272651 [3] NCCL INFO Channel 09/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-168-30:235653:236140 [5] NCCL INFO Channel 13/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-161-138:96563:97086 [0] NCCL INFO Connected all rings
+ip-26-0-161-138:96563:97086 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-168-30:235653:236140 [5] NCCL INFO Channel 15/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-168-34:272154:272650 [4] NCCL INFO Channel 06/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-168-34:272153:272651 [3] NCCL INFO Channel 10/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-161-153:2636091:2636599 [7] NCCL INFO Connected all rings
+ip-26-0-168-34:272154:272650 [4] NCCL INFO Channel 07/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-168-34:272153:272651 [3] NCCL INFO Channel 11/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-161-142:62882:63374 [1] NCCL INFO Connected all rings
+ip-26-0-168-30:235654:236146 [6] NCCL INFO Channel 14/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235653:236140 [5] NCCL INFO Channel 13/0 : 5[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235654:236146 [6] NCCL INFO Channel 06/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235653:236140 [5] NCCL INFO Channel 05/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235654:236146 [6] NCCL INFO Channel 14/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235653:236140 [5] NCCL INFO Channel 13/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235655:236142 [7] NCCL INFO Channel 15/0 : 7[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235648:236145 [0] NCCL INFO Channel 08/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235655:236142 [7] NCCL INFO Channel 07/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235648:236145 [0] NCCL INFO Channel 00/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235655:236142 [7] NCCL INFO Channel 15/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235648:236145 [0] NCCL INFO Channel 08/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272154:272650 [4] NCCL INFO Channel 09/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-168-34:272153:272651 [3] NCCL INFO Channel 13/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-168-34:272157:272649 [7] NCCL INFO Channel 07/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272157:272649 [7] NCCL INFO Channel 15/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272157:272649 [7] NCCL INFO Channel 07/0 : 23[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272154:272650 [4] NCCL INFO Channel 10/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-168-34:272153:272651 [3] NCCL INFO Channel 14/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-168-34:272154:272650 [4] NCCL INFO Channel 11/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-168-34:272153:272651 [3] NCCL INFO Channel 15/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-161-142:62881:63375 [0] NCCL INFO Connected all rings
+ip-26-0-168-34:272154:272650 [4] NCCL INFO Channel 12/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-161-142:62881:63375 [0] NCCL INFO Channel 02/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-168-34:272150:272645 [0] NCCL INFO Channel 00/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272154:272650 [4] NCCL INFO Channel 14/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-168-34:272152:272647 [2] NCCL INFO Channel 02/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272150:272645 [0] NCCL INFO Channel 08/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272152:272647 [2] NCCL INFO Channel 10/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272152:272647 [2] NCCL INFO Channel 02/0 : 18[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272150:272645 [0] NCCL INFO Channel 00/0 : 16[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272154:272650 [4] NCCL INFO Channel 15/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-168-34:272153:272651 [3] NCCL INFO Channel 03/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62881:63375 [0] NCCL INFO Channel 03/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-168-34:272154:272650 [4] NCCL INFO Channel 04/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272153:272651 [3] NCCL INFO Channel 11/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272155:272652 [5] NCCL INFO Channel 05/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272154:272650 [4] NCCL INFO Channel 12/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272153:272651 [3] NCCL INFO Channel 03/0 : 19[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272155:272652 [5] NCCL INFO Channel 13/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272154:272650 [4] NCCL INFO Channel 04/0 : 20[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272155:272652 [5] NCCL INFO Channel 05/0 : 21[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62881:63375 [0] NCCL INFO Channel 04/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-168-30:235651:236141 [3] NCCL INFO Connected all rings
+ip-26-0-161-178:2815963:2816477 [0] NCCL INFO Channel 03/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-161-153:2636087:2636603 [3] NCCL INFO Connected all rings
+ip-26-0-161-142:62881:63375 [0] NCCL INFO Channel 05/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-161-142:62881:63375 [0] NCCL INFO Channel 06/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-161-178:2815963:2816477 [0] NCCL INFO Channel 04/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-161-142:62881:63375 [0] NCCL INFO Channel 07/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-165-59:1557269:1557765 [3] NCCL INFO Connected all rings
+ip-26-0-168-52:2722047:2722542 [2] NCCL INFO Connected all rings
+ip-26-0-161-178:2815963:2816477 [0] NCCL INFO Channel 05/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-161-178:2815963:2816477 [0] NCCL INFO Channel 06/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-161-142:62881:63375 [0] NCCL INFO Channel 10/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-168-30:235651:236141 [3] NCCL INFO Channel 01/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-168-52:2722048:2722545 [3] NCCL INFO Connected all rings
+ip-26-0-168-30:235651:236141 [3] NCCL INFO Channel 02/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-168-30:235651:236141 [3] NCCL INFO Channel 03/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-161-142:62881:63375 [0] NCCL INFO Channel 11/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-168-30:235651:236141 [3] NCCL INFO Channel 05/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-168-52:2722051:2722543 [6] NCCL INFO Channel 01/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-168-30:235651:236141 [3] NCCL INFO Channel 06/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-161-178:2815963:2816477 [0] NCCL INFO Channel 07/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-168-30:235651:236141 [3] NCCL INFO Channel 07/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-168-30:235651:236141 [3] NCCL INFO Channel 09/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-168-52:2722051:2722543 [6] NCCL INFO Channel 02/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-165-131:1630212:1630734 [7] NCCL INFO Channel 14/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-168-30:235651:236141 [3] NCCL INFO Channel 10/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-161-178:2815963:2816477 [0] NCCL INFO Channel 10/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-168-30:235651:236141 [3] NCCL INFO Channel 11/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-161-142:62885:63377 [4] NCCL INFO Connected all rings
+ip-26-0-168-52:2722052:2722546 [7] NCCL INFO Connected all rings
+ip-26-0-168-30:235651:236141 [3] NCCL INFO Channel 13/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-161-142:62881:63375 [0] NCCL INFO Channel 12/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-168-30:235651:236141 [3] NCCL INFO Channel 14/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-165-59:1557272:1557766 [6] NCCL INFO Connected all rings
+ip-26-0-161-178:2815963:2816477 [0] NCCL INFO Channel 11/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-168-30:235651:236141 [3] NCCL INFO Channel 15/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-168-52:2722051:2722543 [6] NCCL INFO Channel 03/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-168-30:235650:236144 [2] NCCL INFO Channel 10/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235651:236141 [3] NCCL INFO Channel 11/0 : 3[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235650:236144 [2] NCCL INFO Channel 02/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62881:63375 [0] NCCL INFO Channel 13/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-168-30:235651:236141 [3] NCCL INFO Channel 03/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235650:236144 [2] NCCL INFO Channel 10/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235652:236147 [4] NCCL INFO Channel 12/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235651:236141 [3] NCCL INFO Channel 11/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235652:236147 [4] NCCL INFO Channel 04/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235652:236147 [4] NCCL INFO Channel 12/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62881:63375 [0] NCCL INFO Channel 14/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-161-178:2815963:2816477 [0] NCCL INFO Channel 12/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-168-52:2722051:2722543 [6] NCCL INFO Channel 04/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-161-178:2815963:2816477 [0] NCCL INFO Channel 13/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-161-142:62881:63375 [0] NCCL INFO Channel 15/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-161-142:62882:63374 [1] NCCL INFO Channel 01/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-168-52:2722051:2722543 [6] NCCL INFO Channel 05/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-168-52:2722049:2722547 [4] NCCL INFO Channel 14/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-161-142:62882:63374 [1] NCCL INFO Channel 03/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-165-59:1557273:1557760 [7] NCCL INFO Connected all rings
+ip-26-0-161-138:96564:97088 [1] NCCL INFO Connected all rings
+ip-26-0-161-178:2815963:2816477 [0] NCCL INFO Channel 14/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-168-52:2722050:2722548 [5] NCCL INFO Connected all rings
+ip-26-0-161-142:62882:63374 [1] NCCL INFO Channel 04/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-161-142:62887:63379 [6] NCCL INFO Connected all rings
+ip-26-0-165-59:1557267:1557761 [1] NCCL INFO Connected all rings
+ip-26-0-168-52:2722051:2722543 [6] NCCL INFO Channel 06/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-168-52:2722049:2722547 [4] NCCL INFO Channel 15/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-161-142:62882:63374 [1] NCCL INFO Channel 05/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-161-178:2815963:2816477 [0] NCCL INFO Channel 15/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-161-142:62882:63374 [1] NCCL INFO Channel 06/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-168-52:2722051:2722543 [6] NCCL INFO Channel 09/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-161-142:62882:63374 [1] NCCL INFO Channel 07/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-161-178:2815968:2816472 [5] NCCL INFO Connected all rings
+ip-26-0-161-178:2815970:2816470 [7] NCCL INFO Connected all rings
+ip-26-0-161-178:2815969:2816471 [6] NCCL INFO Connected all rings
+ip-26-0-168-52:2722047:2722542 [2] NCCL INFO Channel 01/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-168-52:2722051:2722543 [6] NCCL INFO Channel 10/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-168-52:2722047:2722542 [2] NCCL INFO Channel 02/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-168-52:2722051:2722543 [6] NCCL INFO Channel 11/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-161-138:96563:97086 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-142:62882:63374 [1] NCCL INFO Channel 09/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-168-52:2722047:2722542 [2] NCCL INFO Channel 04/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-168-52:2722051:2722543 [6] NCCL INFO Channel 12/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-161-142:62882:63374 [1] NCCL INFO Channel 11/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-165-59:1557268:1557764 [2] NCCL INFO Channel 01/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-168-52:2722048:2722545 [3] NCCL INFO Channel 01/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-168-52:2722047:2722542 [2] NCCL INFO Channel 05/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-168-52:2722051:2722543 [6] NCCL INFO Channel 13/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-161-142:62882:63374 [1] NCCL INFO Channel 12/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-168-52:2722048:2722545 [3] NCCL INFO Channel 02/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-161-178:2815966:2816474 [3] NCCL INFO Connected all rings
+ip-26-0-161-138:96563:97086 [0] NCCL INFO Channel 04/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-168-52:2722047:2722542 [2] NCCL INFO Channel 06/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-168-52:2722051:2722543 [6] NCCL INFO Channel 14/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-161-142:62882:63374 [1] NCCL INFO Channel 13/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-168-52:2722048:2722545 [3] NCCL INFO Channel 03/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-161-178:2815967:2816473 [4] NCCL INFO Connected all rings
+ip-26-0-165-59:1557268:1557764 [2] NCCL INFO Channel 02/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-138:96563:97086 [0] NCCL INFO Channel 05/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-168-52:2722047:2722542 [2] NCCL INFO Channel 07/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-168-52:2722048:2722545 [3] NCCL INFO Channel 05/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-161-142:62886:63378 [5] NCCL INFO Connected all rings
+ip-26-0-165-59:1557268:1557764 [2] NCCL INFO Channel 04/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-168-52:2722047:2722542 [2] NCCL INFO Channel 09/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-161-142:62882:63374 [1] NCCL INFO Channel 14/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-168-52:2722048:2722545 [3] NCCL INFO Channel 06/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-161-138:96563:97086 [0] NCCL INFO Channel 06/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-142:62882:63374 [1] NCCL INFO Channel 15/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-168-52:2722047:2722542 [2] NCCL INFO Channel 10/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-168-52:2722048:2722545 [3] NCCL INFO Channel 07/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-165-59:1557268:1557764 [2] NCCL INFO Channel 05/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-138:96563:97086 [0] NCCL INFO Channel 07/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-59:1557271:1557762 [5] NCCL INFO Channel 01/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-168-52:2722047:2722542 [2] NCCL INFO Channel 12/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-168-52:2722048:2722545 [3] NCCL INFO Channel 09/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-161-142:62881:63375 [0] NCCL INFO Channel 05/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-161-142:62882:63374 [1] NCCL INFO Channel 09/0 : 1[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62882:63374 [1] NCCL INFO Channel 01/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62882:63374 [1] NCCL INFO Channel 09/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557268:1557764 [2] NCCL INFO Channel 06/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-168-52:2722047:2722542 [2] NCCL INFO Channel 13/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-168-52:2722048:2722545 [3] NCCL INFO Channel 10/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-165-59:1557271:1557762 [5] NCCL INFO Channel 02/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-138:96563:97086 [0] NCCL INFO Channel 10/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-142:62881:63375 [0] NCCL INFO Channel 06/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-161-138:96565:97089 [2] NCCL INFO Connected all rings
+ip-26-0-168-52:2722048:2722545 [3] NCCL INFO Channel 11/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-168-52:2722047:2722542 [2] NCCL INFO Channel 14/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-168-52:2722050:2722548 [5] NCCL INFO Channel 01/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-165-59:1557268:1557764 [2] NCCL INFO Channel 07/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-142:62881:63375 [0] NCCL INFO Channel 07/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-165-59:1557272:1557766 [6] NCCL INFO Channel 01/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-59:1557271:1557762 [5] NCCL INFO Channel 03/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-138:96563:97086 [0] NCCL INFO Channel 11/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-168-52:2722048:2722545 [3] NCCL INFO Channel 13/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-168-52:2722047:2722542 [2] NCCL INFO Channel 15/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-168-52:2722050:2722548 [5] NCCL INFO Channel 02/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-161-138:96563:97086 [0] NCCL INFO Channel 12/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-59:1557268:1557764 [2] NCCL INFO Channel 09/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-59:1557272:1557766 [6] NCCL INFO Channel 02/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-59:1557271:1557762 [5] NCCL INFO Channel 04/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-168-52:2722048:2722545 [3] NCCL INFO Channel 14/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-168-52:2722050:2722548 [5] NCCL INFO Channel 03/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-168-52:2722052:2722546 [7] NCCL INFO Channel 07/0 : 23[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722045:2722544 [0] NCCL INFO Channel 00/0 : 16[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-52:2722046:2722541 [1] NCCL INFO Channel 01/0 : 17[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815965:2816475 [2] NCCL INFO Channel 01/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-168-52:2722048:2722545 [3] NCCL INFO Channel 15/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-161-142:62881:63375 [0] NCCL INFO Channel 13/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-168-52:2722050:2722548 [5] NCCL INFO Channel 04/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-168-52:2722052:2722546 [7] NCCL INFO Channel 15/0 : 15[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722045:2722544 [0] NCCL INFO Channel 08/0 : 8[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272157:272649 [7] NCCL INFO Channel 07/0 : 7[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722052:2722546 [7] NCCL INFO Channel 15/0 : 31[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722046:2722541 [1] NCCL INFO Channel 09/0 : 9[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-52:2722045:2722544 [0] NCCL INFO Channel 08/0 : 24[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272150:272645 [0] NCCL INFO Channel 00/0 : 0[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-52:2722046:2722541 [1] NCCL INFO Channel 09/0 : 25[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272157:272649 [7] NCCL INFO Channel 07/0 : 23[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272151:272646 [1] NCCL INFO Channel 01/0 : 1[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272150:272645 [0] NCCL INFO Channel 00/0 : 16[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272151:272646 [1] NCCL INFO Channel 01/0 : 17[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-52:2722050:2722548 [5] NCCL INFO Channel 05/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-165-59:1557272:1557766 [6] NCCL INFO Channel 03/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-59:1557271:1557762 [5] NCCL INFO Channel 05/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-168-52:2722047:2722542 [2] NCCL INFO Channel 02/0 : 18[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96563:97086 [0] NCCL INFO Channel 13/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-168-52:2722050:2722548 [5] NCCL INFO Channel 07/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-161-142:62881:63375 [0] NCCL INFO Channel 14/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-168-52:2722048:2722545 [3] NCCL INFO Channel 03/0 : 19[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557270:1557763 [4] NCCL INFO Connected all rings
+ip-26-0-168-52:2722047:2722542 [2] NCCL INFO Channel 10/0 : 10[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-52:2722047:2722542 [2] NCCL INFO Channel 10/0 : 26[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272152:272647 [2] NCCL INFO Channel 02/0 : 2[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557268:1557764 [2] NCCL INFO Channel 10/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-168-52:2722048:2722545 [3] NCCL INFO Channel 11/0 : 11[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272152:272647 [2] NCCL INFO Channel 02/0 : 18[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-52:2722050:2722548 [5] NCCL INFO Channel 09/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-168-34:272153:272651 [3] NCCL INFO Channel 03/0 : 3[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-52:2722048:2722545 [3] NCCL INFO Channel 11/0 : 27[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272153:272651 [3] NCCL INFO Channel 03/0 : 19[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815965:2816475 [2] NCCL INFO Channel 02/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-168-52:2722050:2722548 [5] NCCL INFO Channel 10/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-161-142:62881:63375 [0] NCCL INFO Channel 15/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-165-59:1557272:1557766 [6] NCCL INFO Channel 04/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-168-52:2722050:2722548 [5] NCCL INFO Channel 11/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-165-59:1557267:1557761 [1] NCCL INFO Channel 01/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-59:1557271:1557762 [5] NCCL INFO Channel 07/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-178:2815965:2816475 [2] NCCL INFO Channel 04/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-168-52:2722050:2722548 [5] NCCL INFO Channel 12/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-161-138:96564:97088 [1] NCCL INFO Channel 01/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-59:1557268:1557764 [2] NCCL INFO Channel 12/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-168-52:2722050:2722548 [5] NCCL INFO Channel 13/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-161-138:96564:97088 [1] NCCL INFO Channel 03/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-168-52:2722050:2722548 [5] NCCL INFO Channel 15/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-161-153:2636084:2636605 [0] NCCL INFO Connected all rings
+ip-26-0-161-138:96563:97086 [0] NCCL INFO Channel 14/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-59:1557272:1557766 [6] NCCL INFO Channel 05/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-153:2636084:2636605 [0] NCCL INFO Channel 02/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-165-59:1557271:1557762 [5] NCCL INFO Channel 09/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-165-59:1557267:1557761 [1] NCCL INFO Channel 03/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-142:62885:63377 [4] NCCL INFO Channel 01/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-168-52:2722049:2722547 [4] NCCL INFO Channel 04/0 : 20[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722051:2722543 [6] NCCL INFO Channel 06/0 : 22[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-52:2722050:2722548 [5] NCCL INFO Channel 05/0 : 21[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62886:63378 [5] NCCL INFO Channel 01/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-161-142:62885:63377 [4] NCCL INFO Channel 02/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-168-52:2722049:2722547 [4] NCCL INFO Channel 12/0 : 12[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557268:1557764 [2] NCCL INFO Channel 13/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-168-34:272154:272650 [4] NCCL INFO Channel 04/0 : 4[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722051:2722543 [6] NCCL INFO Channel 14/0 : 14[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272156:272648 [6] NCCL INFO Channel 06/0 : 6[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-52:2722050:2722548 [5] NCCL INFO Channel 13/0 : 13[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-52:2722049:2722547 [4] NCCL INFO Channel 12/0 : 28[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722051:2722543 [6] NCCL INFO Channel 14/0 : 30[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272155:272652 [5] NCCL INFO Channel 05/0 : 5[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272154:272650 [4] NCCL INFO Channel 04/0 : 20[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96564:97088 [1] NCCL INFO Channel 04/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-168-34:272156:272648 [6] NCCL INFO Channel 06/0 : 22[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-52:2722050:2722548 [5] NCCL INFO Channel 13/0 : 29[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272155:272652 [5] NCCL INFO Channel 05/0 : 21[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96563:97086 [0] NCCL INFO Channel 15/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-161-142:62886:63378 [5] NCCL INFO Channel 02/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-161-142:62885:63377 [4] NCCL INFO Channel 03/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-165-59:1557272:1557766 [6] NCCL INFO Channel 06/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-59:1557271:1557762 [5] NCCL INFO Channel 10/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-165-59:1557267:1557761 [1] NCCL INFO Channel 04/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-59:1557268:1557764 [2] NCCL INFO Channel 14/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-178:2815965:2816475 [2] NCCL INFO Channel 05/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-161-142:62886:63378 [5] NCCL INFO Channel 03/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-161-138:96564:97088 [1] NCCL INFO Channel 05/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-142:62885:63377 [4] NCCL INFO Channel 04/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-161-138:96566:97087 [3] NCCL INFO Connected all rings
+ip-26-0-161-178:2815965:2816475 [2] NCCL INFO Channel 06/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-165-59:1557272:1557766 [6] NCCL INFO Channel 09/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-59:1557271:1557762 [5] NCCL INFO Channel 11/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-165-59:1557267:1557761 [1] NCCL INFO Channel 05/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-59:1557268:1557764 [2] NCCL INFO Channel 15/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-59:1557269:1557765 [3] NCCL INFO Channel 01/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-138:96564:97088 [1] NCCL INFO Channel 06/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-142:62886:63378 [5] NCCL INFO Channel 04/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-161-178:2815968:2816472 [5] NCCL INFO Channel 01/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-161-138:96569:97092 [6] NCCL INFO Connected all rings
+ip-26-0-161-153:2636084:2636605 [0] NCCL INFO Channel 03/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-161-178:2815965:2816475 [2] NCCL INFO Channel 07/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-161-138:96564:97088 [1] NCCL INFO Channel 07/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-178:2815968:2816472 [5] NCCL INFO Channel 02/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-165-59:1557272:1557766 [6] NCCL INFO Channel 10/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-59:1557271:1557762 [5] NCCL INFO Channel 12/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-142:62885:63377 [4] NCCL INFO Channel 06/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-165-59:1557267:1557761 [1] NCCL INFO Channel 06/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-59:1557269:1557765 [3] NCCL INFO Channel 02/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-142:62888:63380 [7] NCCL INFO Connected all rings
+ip-26-0-161-178:2815965:2816475 [2] NCCL INFO Channel 09/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-161-153:2636084:2636605 [0] NCCL INFO Channel 04/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-161-138:96564:97088 [1] NCCL INFO Channel 09/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-153:2636084:2636605 [0] NCCL INFO Channel 05/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-161-142:62886:63378 [5] NCCL INFO Channel 05/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-161-142:62885:63377 [4] NCCL INFO Channel 07/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-161-178:2815968:2816472 [5] NCCL INFO Channel 03/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-165-59:1557272:1557766 [6] NCCL INFO Channel 11/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-59:1557271:1557762 [5] NCCL INFO Channel 13/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-165-59:1557267:1557761 [1] NCCL INFO Channel 07/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-59:1557269:1557765 [3] NCCL INFO Channel 03/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-138:96567:97090 [4] NCCL INFO Connected all rings
+ip-26-0-161-138:96564:97088 [1] NCCL INFO Channel 11/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-178:2815965:2816475 [2] NCCL INFO Channel 10/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-161-138:96568:97091 [5] NCCL INFO Connected all rings
+ip-26-0-161-142:62886:63378 [5] NCCL INFO Channel 07/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-161-153:2636084:2636605 [0] NCCL INFO Channel 06/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-161-142:62885:63377 [4] NCCL INFO Channel 09/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-161-178:2815968:2816472 [5] NCCL INFO Channel 04/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-165-59:1557272:1557766 [6] NCCL INFO Channel 12/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-59:1557271:1557762 [5] NCCL INFO Channel 15/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-165-59:1557267:1557761 [1] NCCL INFO Channel 09/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-59:1557269:1557765 [3] NCCL INFO Channel 05/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-178:2815965:2816475 [2] NCCL INFO Channel 12/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-161-142:62886:63378 [5] NCCL INFO Channel 09/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-161-142:62885:63377 [4] NCCL INFO Channel 10/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-161-153:2636084:2636605 [0] NCCL INFO Channel 07/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-161-138:96564:97088 [1] NCCL INFO Channel 12/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-178:2815968:2816472 [5] NCCL INFO Channel 05/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-165-59:1557272:1557766 [6] NCCL INFO Channel 13/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-59:1557267:1557761 [1] NCCL INFO Channel 11/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-59:1557269:1557765 [3] NCCL INFO Channel 06/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-178:2815966:2816474 [3] NCCL INFO Channel 01/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-161-142:62886:63378 [5] NCCL INFO Channel 10/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-161-142:62885:63377 [4] NCCL INFO Channel 11/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-161-153:2636084:2636605 [0] NCCL INFO Channel 10/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-161-178:2815965:2816475 [2] NCCL INFO Channel 13/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-165-59:1557272:1557766 [6] NCCL INFO Channel 14/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-59:1557267:1557761 [1] NCCL INFO Channel 12/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-59:1557269:1557765 [3] NCCL INFO Channel 07/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-178:2815968:2816472 [5] NCCL INFO Channel 07/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-161-142:62885:63377 [4] NCCL INFO Channel 12/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-161-138:96564:97088 [1] NCCL INFO Channel 13/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-142:62886:63378 [5] NCCL INFO Channel 11/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-161-153:2636084:2636605 [0] NCCL INFO Channel 11/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-161-178:2815969:2816471 [6] NCCL INFO Channel 01/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-161-178:2815965:2816475 [2] NCCL INFO Channel 14/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-161-138:96570:97093 [7] NCCL INFO Connected all rings
+ip-26-0-165-59:1557267:1557761 [1] NCCL INFO Channel 13/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-59:1557269:1557765 [3] NCCL INFO Channel 09/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-142:62885:63377 [4] NCCL INFO Channel 14/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-161-178:2815968:2816472 [5] NCCL INFO Channel 09/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-161-142:62886:63378 [5] NCCL INFO Channel 12/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-165-59:1557272:1557766 [6] NCCL INFO Channel 14/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96564:97088 [1] NCCL INFO Channel 14/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-161-153:2636084:2636605 [0] NCCL INFO Channel 12/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-165-59:1557272:1557766 [6] NCCL INFO Channel 06/0 : 22[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557272:1557766 [6] NCCL INFO Channel 06/0 : 6[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235654:236146 [6] NCCL INFO Channel 14/0 : 30[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235654:236146 [6] NCCL INFO Channel 14/0 : 14[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815965:2816475 [2] NCCL INFO Channel 15/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-161-178:2815969:2816471 [6] NCCL INFO Channel 02/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-165-59:1557269:1557765 [3] NCCL INFO Channel 10/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-59:1557267:1557761 [1] NCCL INFO Channel 14/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-59:1557272:1557766 [6] NCCL INFO Channel 14/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272156:272648 [6] NCCL INFO Channel 06/0 : 30[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235654:236146 [6] NCCL INFO Channel 06/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272156:272648 [6] NCCL INFO Channel 06/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-52:2722051:2722543 [6] NCCL INFO Channel 06/0 : 30[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62885:63377 [4] NCCL INFO Channel 15/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-168-34:272156:272648 [6] NCCL INFO Channel 14/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235654:236146 [6] NCCL INFO Channel 14/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815968:2816472 [5] NCCL INFO Channel 10/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-161-142:62886:63378 [5] NCCL INFO Channel 13/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-168-30:235654:236146 [6] NCCL INFO Channel 14/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815966:2816474 [3] NCCL INFO Channel 02/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-161-138:96564:97088 [1] NCCL INFO Channel 15/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-59:1557267:1557761 [1] NCCL INFO Channel 15/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-59:1557269:1557765 [3] NCCL INFO Channel 11/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-59:1557270:1557763 [4] NCCL INFO Channel 01/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-142:62886:63378 [5] NCCL INFO Channel 15/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-161-178:2815969:2816471 [6] NCCL INFO Channel 03/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-161-178:2815967:2816473 [4] NCCL INFO Channel 01/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-161-153:2636084:2636605 [0] NCCL INFO Channel 13/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-161-178:2815968:2816472 [5] NCCL INFO Channel 11/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-161-138:96563:97086 [0] NCCL INFO Channel 05/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-161-178:2815966:2816474 [3] NCCL INFO Channel 03/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-165-59:1557269:1557765 [3] NCCL INFO Channel 13/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-59:1557270:1557763 [4] NCCL INFO Channel 02/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-59:1557266:1557759 [0] NCCL INFO Channel 05/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-168-52:2722051:2722543 [6] NCCL INFO Channel 00/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-168-34:272156:272648 [6] NCCL INFO Channel 00/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-165-59:1557267:1557761 [1] NCCL INFO Channel 09/0 : 1[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-52:2722051:2722543 [6] NCCL INFO Channel 05/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-161-178:2815969:2816471 [6] NCCL INFO Channel 04/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-165-59:1557267:1557761 [1] NCCL INFO Channel 01/0 : 17[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272156:272648 [6] NCCL INFO Channel 05/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-165-59:1557267:1557761 [1] NCCL INFO Channel 01/0 : 1[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815967:2816473 [4] NCCL INFO Channel 02/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-168-30:235649:236143 [1] NCCL INFO Channel 09/0 : 25[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235649:236143 [1] NCCL INFO Channel 09/0 : 9[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-52:2722051:2722543 [6] NCCL INFO Channel 08/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-161-178:2815968:2816472 [5] NCCL INFO Channel 12/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-161-178:2815966:2816474 [3] NCCL INFO Channel 05/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-161-142:62884:63376 [3] NCCL INFO Connected all rings
+ip-26-0-165-59:1557267:1557761 [1] NCCL INFO Channel 09/0 : 9[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272156:272648 [6] NCCL INFO Channel 08/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-168-34:272151:272646 [1] NCCL INFO Channel 01/0 : 25[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235649:236143 [1] NCCL INFO Channel 01/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272151:272646 [1] NCCL INFO Channel 01/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-52:2722046:2722541 [1] NCCL INFO Channel 01/0 : 25[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-52:2722051:2722543 [6] NCCL INFO Channel 13/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-168-30:235649:236143 [1] NCCL INFO Channel 09/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272151:272646 [1] NCCL INFO Channel 09/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235649:236143 [1] NCCL INFO Channel 09/0 : 9[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272156:272648 [6] NCCL INFO Channel 13/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-161-142:62887:63379 [6] NCCL INFO Channel 01/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-161-178:2815969:2816471 [6] NCCL INFO Channel 05/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-161-178:2815967:2816473 [4] NCCL INFO Channel 03/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-161-178:2815968:2816472 [5] NCCL INFO Channel 13/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-161-153:2636084:2636605 [0] NCCL INFO Channel 14/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-161-178:2815966:2816474 [3] NCCL INFO Channel 06/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-161-142:62887:63379 [6] NCCL INFO Channel 02/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-165-59:1557269:1557765 [3] NCCL INFO Channel 14/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-168-52:2722046:2722541 [1] NCCL INFO Channel 00/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-161-142:62887:63379 [6] NCCL INFO Channel 03/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-168-34:272151:272646 [1] NCCL INFO Channel 00/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-168-30:235649:236143 [1] NCCL INFO Channel 00/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-161-178:2815969:2816471 [6] NCCL INFO Channel 06/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-168-52:2722046:2722541 [1] NCCL INFO Channel 08/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-165-59:1557266:1557759 [0] NCCL INFO Channel 06/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-161-142:62887:63379 [6] NCCL INFO Channel 04/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-161-178:2815967:2816473 [4] NCCL INFO Channel 04/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-168-34:272151:272646 [1] NCCL INFO Channel 08/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-165-59:1557270:1557763 [4] NCCL INFO Channel 03/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-178:2815968:2816472 [5] NCCL INFO Channel 15/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-161-178:2815966:2816474 [3] NCCL INFO Channel 07/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-168-30:235649:236143 [1] NCCL INFO Channel 08/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-161-138:96565:97089 [2] NCCL INFO Channel 01/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-153:2636089:2636600 [5] NCCL INFO Connected all rings
+ip-26-0-161-153:2636084:2636605 [0] NCCL INFO Channel 15/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-161-142:62887:63379 [6] NCCL INFO Channel 05/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-161-138:96569:97092 [6] NCCL INFO Channel 01/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-142:62887:63379 [6] NCCL INFO Channel 06/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-161-178:2815969:2816471 [6] NCCL INFO Channel 09/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-161-178:2815967:2816473 [4] NCCL INFO Channel 06/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-161-142:62887:63379 [6] NCCL INFO Channel 09/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-161-138:96565:97089 [2] NCCL INFO Channel 02/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-178:2815966:2816474 [3] NCCL INFO Channel 09/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-165-59:1557266:1557759 [0] NCCL INFO Channel 07/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-161-138:96569:97092 [6] NCCL INFO Channel 02/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-178:2815969:2816471 [6] NCCL INFO Channel 10/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-161-142:62887:63379 [6] NCCL INFO Channel 10/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-161-153:2636085:2636604 [1] NCCL INFO Channel 01/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-161-142:62887:63379 [6] NCCL INFO Channel 11/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-161-178:2815967:2816473 [4] NCCL INFO Channel 07/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-161-178:2815966:2816474 [3] NCCL INFO Channel 10/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-165-59:1557266:1557759 [0] NCCL INFO Channel 13/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-165-59:1557272:1557766 [6] NCCL INFO Channel 00/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-168-30:235654:236146 [6] NCCL INFO Channel 00/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-161-178:2815969:2816471 [6] NCCL INFO Channel 11/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-161-138:96565:97089 [2] NCCL INFO Channel 04/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-142:62887:63379 [6] NCCL INFO Channel 12/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-168-30:235654:236146 [6] NCCL INFO Channel 05/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-161-178:2815967:2816473 [4] NCCL INFO Channel 09/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-161-138:96569:97092 [6] NCCL INFO Channel 03/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-168-30:235654:236146 [6] NCCL INFO Channel 08/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-161-178:2815966:2816474 [3] NCCL INFO Channel 11/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-165-59:1557266:1557759 [0] NCCL INFO Channel 14/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-161-138:96568:97091 [5] NCCL INFO Channel 01/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-165-59:1557272:1557766 [6] NCCL INFO Channel 05/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-142:62887:63379 [6] NCCL INFO Channel 13/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-161-178:2815969:2816471 [6] NCCL INFO Channel 12/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-168-30:235654:236146 [6] NCCL INFO Channel 13/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-161-142:62887:63379 [6] NCCL INFO Channel 14/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-161-142:62884:63376 [3] NCCL INFO Channel 01/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-161-138:96563:97086 [0] NCCL INFO Channel 06/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-161-138:96567:97090 [4] NCCL INFO Channel 01/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-59:1557270:1557763 [4] NCCL INFO Channel 04/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-178:2815967:2816473 [4] NCCL INFO Channel 10/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-165-59:1557266:1557759 [0] NCCL INFO Channel 15/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-165-59:1557272:1557766 [6] NCCL INFO Channel 08/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-59:1557269:1557765 [3] NCCL INFO Channel 15/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-178:2815966:2816474 [3] NCCL INFO Channel 13/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-161-138:96569:97092 [6] NCCL INFO Channel 04/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-59:1557267:1557761 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-142:62884:63376 [3] NCCL INFO Channel 02/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-161-178:2815969:2816471 [6] NCCL INFO Channel 13/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-161-142:62886:63378 [5] NCCL INFO Channel 13/0 : 5[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62884:63376 [3] NCCL INFO Channel 03/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-161-142:62886:63378 [5] NCCL INFO Channel 05/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62887:63379 [6] NCCL INFO Channel 14/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96563:97086 [0] NCCL INFO Channel 07/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-161-142:62886:63378 [5] NCCL INFO Channel 13/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62887:63379 [6] NCCL INFO Channel 06/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96567:97090 [4] NCCL INFO Channel 02/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-142:62887:63379 [6] NCCL INFO Channel 14/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815967:2816473 [4] NCCL INFO Channel 11/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-161-142:62884:63376 [3] NCCL INFO Channel 05/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-161-153:2636085:2636604 [1] NCCL INFO Channel 03/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-161-178:2815966:2816474 [3] NCCL INFO Channel 14/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-161-178:2815969:2816471 [6] NCCL INFO Channel 14/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-161-138:96569:97092 [6] NCCL INFO Channel 05/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-59:1557270:1557763 [4] NCCL INFO Channel 06/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-59:1557272:1557766 [6] NCCL INFO Channel 13/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-59:1557267:1557761 [1] NCCL INFO Channel 08/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-142:62884:63376 [3] NCCL INFO Channel 06/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-165-59:1557273:1557760 [7] NCCL INFO Channel 15/0 : 7[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557268:1557764 [2] NCCL INFO Channel 10/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557266:1557759 [0] NCCL INFO Channel 08/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96563:97086 [0] NCCL INFO Channel 13/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-161-138:96567:97090 [4] NCCL INFO Channel 03/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-178:2815967:2816473 [4] NCCL INFO Channel 12/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-161-178:2815966:2816474 [3] NCCL INFO Channel 15/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-165-59:1557268:1557764 [2] NCCL INFO Channel 02/0 : 18[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62884:63376 [3] NCCL INFO Channel 07/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-165-59:1557266:1557759 [0] NCCL INFO Channel 00/0 : 16[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557273:1557760 [7] NCCL INFO Channel 07/0 : 23[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235648:236145 [0] NCCL INFO Channel 08/0 : 24[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96569:97092 [6] NCCL INFO Channel 06/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-59:1557268:1557764 [2] NCCL INFO Channel 02/0 : 2[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557266:1557759 [0] NCCL INFO Channel 00/0 : 0[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557273:1557760 [7] NCCL INFO Channel 07/0 : 7[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235655:236142 [7] NCCL INFO Channel 15/0 : 31[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235650:236144 [2] NCCL INFO Channel 10/0 : 26[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235648:236145 [0] NCCL INFO Channel 08/0 : 8[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557270:1557763 [4] NCCL INFO Channel 07/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-168-30:235655:236142 [7] NCCL INFO Channel 15/0 : 15[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235650:236144 [2] NCCL INFO Channel 10/0 : 10[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815964:2816476 [1] NCCL INFO Connected all rings
+ip-26-0-165-59:1557268:1557764 [2] NCCL INFO Channel 10/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557266:1557759 [0] NCCL INFO Channel 08/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815969:2816471 [6] NCCL INFO Channel 06/0 : 22[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557273:1557760 [7] NCCL INFO Channel 15/0 : 15[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235648:236145 [0] NCCL INFO Channel 00/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272150:272645 [0] NCCL INFO Channel 00/0 : 24[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235650:236144 [2] NCCL INFO Channel 02/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235655:236142 [7] NCCL INFO Channel 07/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62884:63376 [3] NCCL INFO Channel 09/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-168-34:272150:272645 [0] NCCL INFO Channel 00/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636085:2636604 [1] NCCL INFO Channel 04/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-168-52:2722047:2722542 [2] NCCL INFO Channel 02/0 : 26[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235648:236145 [0] NCCL INFO Channel 08/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557270:1557763 [4] NCCL INFO Channel 09/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-168-34:272150:272645 [0] NCCL INFO Channel 08/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-52:2722045:2722544 [0] NCCL INFO Channel 00/0 : 24[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-52:2722052:2722546 [7] NCCL INFO Channel 07/0 : 31[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235650:236144 [2] NCCL INFO Channel 10/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235648:236145 [0] NCCL INFO Channel 08/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235655:236142 [7] NCCL INFO Channel 15/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235650:236144 [2] NCCL INFO Channel 10/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235655:236142 [7] NCCL INFO Channel 15/0 : 15[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96566:97087 [3] NCCL INFO Channel 01/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-178:2815967:2816473 [4] NCCL INFO Channel 14/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-161-138:96563:97086 [0] NCCL INFO Channel 14/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-165-59:1557270:1557763 [4] NCCL INFO Channel 10/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-142:62884:63376 [3] NCCL INFO Channel 10/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-165-59:1557273:1557760 [7] NCCL INFO Channel 01/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-138:96567:97090 [4] NCCL INFO Channel 04/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-168-34:272157:272649 [7] NCCL INFO Channel 07/0 : 31[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272157:272649 [7] NCCL INFO Channel 07/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96569:97092 [6] NCCL INFO Channel 09/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-168-34:272157:272649 [7] NCCL INFO Channel 15/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815967:2816473 [4] NCCL INFO Channel 15/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-161-142:62884:63376 [3] NCCL INFO Channel 11/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-168-52:2722052:2722546 [7] NCCL INFO Channel 01/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-168-34:272152:272647 [2] NCCL INFO Channel 02/0 : 26[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272157:272649 [7] NCCL INFO Channel 01/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-161-142:62888:63380 [7] NCCL INFO Channel 15/0 : 7[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272152:272647 [2] NCCL INFO Channel 02/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557270:1557763 [4] NCCL INFO Channel 11/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-153:2636085:2636604 [1] NCCL INFO Channel 05/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-161-138:96566:97087 [3] NCCL INFO Channel 02/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-142:62881:63375 [0] NCCL INFO Channel 08/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272152:272647 [2] NCCL INFO Channel 10/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62888:63380 [7] NCCL INFO Channel 07/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62884:63376 [3] NCCL INFO Channel 13/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-161-138:96563:97086 [0] NCCL INFO Channel 15/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-161-142:62881:63375 [0] NCCL INFO Channel 00/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62888:63380 [7] NCCL INFO Channel 15/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62881:63375 [0] NCCL INFO Channel 08/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-52:2722052:2722546 [7] NCCL INFO Channel 02/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-165-59:1557273:1557760 [7] NCCL INFO Channel 02/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-138:96567:97090 [4] NCCL INFO Channel 06/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-138:96569:97092 [6] NCCL INFO Channel 10/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-168-34:272157:272649 [7] NCCL INFO Channel 02/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-161-142:62884:63376 [3] NCCL INFO Channel 14/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-161-178:2815966:2816474 [3] NCCL INFO Channel 03/0 : 19[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815968:2816472 [5] NCCL INFO Channel 05/0 : 21[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815967:2816473 [4] NCCL INFO Channel 04/0 : 20[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62884:63376 [3] NCCL INFO Channel 15/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-165-59:1557270:1557763 [4] NCCL INFO Channel 12/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-153:2636085:2636604 [1] NCCL INFO Channel 06/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-168-34:272157:272649 [7] NCCL INFO Channel 03/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-161-138:96566:97087 [3] NCCL INFO Channel 03/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-59:1557273:1557760 [7] NCCL INFO Channel 03/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-138:96567:97090 [4] NCCL INFO Channel 07/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-168-52:2722047:2722542 [2] NCCL INFO Channel 00/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-161-138:96569:97092 [6] NCCL INFO Channel 11/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-138:96565:97089 [2] NCCL INFO Channel 05/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-142:62883:63373 [2] NCCL INFO Channel 10/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272157:272649 [7] NCCL INFO Channel 04/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-161-142:62885:63377 [4] NCCL INFO Channel 12/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62883:63373 [2] NCCL INFO Channel 02/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62884:63376 [3] NCCL INFO Channel 11/0 : 3[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62885:63377 [4] NCCL INFO Channel 04/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62883:63373 [2] NCCL INFO Channel 10/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62884:63376 [3] NCCL INFO Channel 03/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62885:63377 [4] NCCL INFO Channel 12/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62884:63376 [3] NCCL INFO Channel 11/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557270:1557763 [4] NCCL INFO Channel 14/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-59:1557273:1557760 [7] NCCL INFO Channel 04/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-168-52:2722047:2722542 [2] NCCL INFO Channel 01/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-168-30:235655:236142 [7] NCCL INFO Channel 01/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-168-52:2722052:2722546 [7] NCCL INFO Channel 03/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-161-138:96566:97087 [3] NCCL INFO Channel 05/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-138:96567:97090 [4] NCCL INFO Channel 09/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-138:96569:97092 [6] NCCL INFO Channel 12/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-168-52:2722047:2722542 [2] NCCL INFO Channel 08/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-161-138:96565:97089 [2] NCCL INFO Channel 06/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-153:2636085:2636604 [1] NCCL INFO Channel 07/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-168-30:235655:236142 [7] NCCL INFO Channel 02/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-165-59:1557268:1557764 [2] NCCL INFO Channel 00/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-165-59:1557273:1557760 [7] NCCL INFO Channel 05/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-59:1557270:1557763 [4] NCCL INFO Channel 15/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-168-52:2722047:2722542 [2] NCCL INFO Channel 09/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-168-52:2722052:2722546 [7] NCCL INFO Channel 04/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-161-138:96566:97087 [3] NCCL INFO Channel 06/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-168-34:272157:272649 [7] NCCL INFO Channel 05/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-161-138:96568:97091 [5] NCCL INFO Channel 02/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-138:96567:97090 [4] NCCL INFO Channel 10/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-138:96569:97092 [6] NCCL INFO Channel 13/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-168-30:235655:236142 [7] NCCL INFO Channel 03/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-161-138:96565:97089 [2] NCCL INFO Channel 07/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-168-52:2722052:2722546 [7] NCCL INFO Channel 05/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-165-59:1557268:1557764 [2] NCCL INFO Channel 01/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-165-59:1557273:1557760 [7] NCCL INFO Channel 06/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-59:1557269:1557765 [3] NCCL INFO Channel 11/0 : 3[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-52:2722052:2722546 [7] NCCL INFO Channel 06/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-161-178:2815964:2816476 [1] NCCL INFO Channel 01/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-168-30:235655:236142 [7] NCCL INFO Channel 04/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-165-59:1557271:1557762 [5] NCCL INFO Channel 13/0 : 5[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235650:236144 [2] NCCL INFO Channel 00/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-165-59:1557270:1557763 [4] NCCL INFO Channel 12/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557269:1557765 [3] NCCL INFO Channel 03/0 : 19[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557269:1557765 [3] NCCL INFO Channel 03/0 : 3[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815964:2816476 [1] NCCL INFO Channel 03/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-165-59:1557271:1557762 [5] NCCL INFO Channel 05/0 : 21[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235653:236140 [5] NCCL INFO Channel 13/0 : 29[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557268:1557764 [2] NCCL INFO Channel 08/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-165-59:1557273:1557760 [7] NCCL INFO Channel 07/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-59:1557270:1557763 [4] NCCL INFO Channel 04/0 : 20[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235652:236147 [4] NCCL INFO Channel 12/0 : 28[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557271:1557762 [5] NCCL INFO Channel 05/0 : 5[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-52:2722052:2722546 [7] NCCL INFO Channel 07/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-161-138:96566:97087 [3] NCCL INFO Channel 07/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-168-30:235653:236140 [5] NCCL INFO Channel 13/0 : 13[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235652:236147 [4] NCCL INFO Channel 12/0 : 12[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557270:1557763 [4] NCCL INFO Channel 04/0 : 4[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96568:97091 [5] NCCL INFO Channel 03/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-138:96567:97090 [4] NCCL INFO Channel 11/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-138:96569:97092 [6] NCCL INFO Channel 14/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-161-138:96565:97089 [2] NCCL INFO Channel 09/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-59:1557269:1557765 [3] NCCL INFO Channel 11/0 : 11[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272153:272651 [3] NCCL INFO Channel 03/0 : 27[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557271:1557762 [5] NCCL INFO Channel 13/0 : 13[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815964:2816476 [1] NCCL INFO Channel 04/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-168-30:235653:236140 [5] NCCL INFO Channel 05/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272153:272651 [3] NCCL INFO Channel 03/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557270:1557763 [4] NCCL INFO Channel 12/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235650:236144 [2] NCCL INFO Channel 01/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-168-34:272155:272652 [5] NCCL INFO Channel 05/0 : 29[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-52:2722050:2722548 [5] NCCL INFO Channel 05/0 : 29[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235655:236142 [7] NCCL INFO Channel 05/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-161-153:2636088:2636601 [4] NCCL INFO Connected all rings
+ip-26-0-168-30:235652:236147 [4] NCCL INFO Channel 04/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272153:272651 [3] NCCL INFO Channel 11/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272155:272652 [5] NCCL INFO Channel 05/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-52:2722049:2722547 [4] NCCL INFO Channel 04/0 : 28[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235653:236140 [5] NCCL INFO Channel 13/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636085:2636604 [1] NCCL INFO Channel 09/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-168-30:235652:236147 [4] NCCL INFO Channel 12/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235653:236140 [5] NCCL INFO Channel 13/0 : 13[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272155:272652 [5] NCCL INFO Channel 13/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-52:2722052:2722546 [7] NCCL INFO Channel 09/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-168-30:235652:236147 [4] NCCL INFO Channel 12/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815964:2816476 [1] NCCL INFO Channel 05/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-165-59:1557268:1557764 [2] NCCL INFO Channel 09/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-168-30:235651:236141 [3] NCCL INFO Channel 11/0 : 27[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235651:236141 [3] NCCL INFO Channel 11/0 : 11[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815964:2816476 [1] NCCL INFO Channel 06/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-168-34:272154:272650 [4] NCCL INFO Channel 04/0 : 28[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235650:236144 [2] NCCL INFO Channel 08/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-168-30:235655:236142 [7] NCCL INFO Channel 06/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-168-52:2722052:2722546 [7] NCCL INFO Channel 10/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-168-30:235651:236141 [3] NCCL INFO Channel 03/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272154:272650 [4] NCCL INFO Channel 04/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722048:2722545 [3] NCCL INFO Channel 03/0 : 27[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235651:236141 [3] NCCL INFO Channel 11/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272154:272650 [4] NCCL INFO Channel 12/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815964:2816476 [1] NCCL INFO Channel 07/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-165-59:1557273:1557760 [7] NCCL INFO Channel 09/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-168-30:235651:236141 [3] NCCL INFO Channel 11/0 : 11[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96566:97087 [3] NCCL INFO Channel 09/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-138:96568:97091 [5] NCCL INFO Channel 04/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-138:96567:97090 [4] NCCL INFO Channel 12/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-138:96565:97089 [2] NCCL INFO Channel 10/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-178:2815964:2816476 [1] NCCL INFO Channel 09/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-161-153:2636085:2636604 [1] NCCL INFO Channel 11/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-168-52:2722052:2722546 [7] NCCL INFO Channel 11/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-168-30:235650:236144 [2] NCCL INFO Channel 09/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-161-178:2815964:2816476 [1] NCCL INFO Channel 11/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-165-59:1557273:1557760 [7] NCCL INFO Channel 10/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-168-30:235655:236142 [7] NCCL INFO Channel 07/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-161-178:2815964:2816476 [1] NCCL INFO Channel 12/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-161-138:96568:97091 [5] NCCL INFO Channel 05/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-138:96566:97087 [3] NCCL INFO Channel 10/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-138:96567:97090 [4] NCCL INFO Channel 14/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-138:96565:97089 [2] NCCL INFO Channel 12/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-161-178:2815964:2816476 [1] NCCL INFO Channel 13/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-165-59:1557273:1557760 [7] NCCL INFO Channel 11/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-178:2815964:2816476 [1] NCCL INFO Channel 14/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-168-30:235655:236142 [7] NCCL INFO Channel 09/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-161-153:2636085:2636604 [1] NCCL INFO Channel 12/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-161-178:2815964:2816476 [1] NCCL INFO Channel 15/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-168-52:2722050:2722548 [5] NCCL INFO Channel 00/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-161-138:96568:97091 [5] NCCL INFO Channel 07/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-178:2815963:2816477 [0] NCCL INFO Channel 05/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-161-138:96567:97090 [4] NCCL INFO Channel 15/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-161-138:96566:97087 [3] NCCL INFO Channel 11/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-138:96565:97089 [2] NCCL INFO Channel 13/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-59:1557273:1557760 [7] NCCL INFO Channel 12/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-59:1557271:1557762 [5] NCCL INFO Channel 00/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-178:2815965:2816475 [2] NCCL INFO Channel 02/0 : 18[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815964:2816476 [1] NCCL INFO Channel 01/0 : 17[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235655:236142 [7] NCCL INFO Channel 10/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-161-178:2815963:2816477 [0] NCCL INFO Channel 06/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-168-52:2722049:2722547 [4] NCCL INFO Channel 00/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-161-178:2815963:2816477 [0] NCCL INFO Channel 07/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-165-59:1557270:1557763 [4] NCCL INFO Channel 00/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-153:2636085:2636604 [1] NCCL INFO Channel 13/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-168-52:2722049:2722547 [4] NCCL INFO Channel 03/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-168-30:235653:236140 [5] NCCL INFO Channel 00/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-161-138:96568:97091 [5] NCCL INFO Channel 09/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-178:2815963:2816477 [0] NCCL INFO Channel 13/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-161-138:96566:97087 [3] NCCL INFO Channel 13/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-138:96565:97089 [2] NCCL INFO Channel 14/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-59:1557271:1557762 [5] NCCL INFO Channel 04/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-168-30:235655:236142 [7] NCCL INFO Channel 11/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-168-52:2722050:2722548 [5] NCCL INFO Channel 04/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-161-178:2815963:2816477 [0] NCCL INFO Channel 14/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-168-52:2722052:2722546 [7] NCCL INFO Channel 12/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-168-52:2722049:2722547 [4] NCCL INFO Channel 05/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-165-59:1557273:1557760 [7] NCCL INFO Channel 13/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-59:1557270:1557763 [4] NCCL INFO Channel 03/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-178:2815963:2816477 [0] NCCL INFO Channel 15/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-161-153:2636089:2636600 [5] NCCL INFO Channel 01/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-161-138:96568:97091 [5] NCCL INFO Channel 10/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-165-59:1557271:1557762 [5] NCCL INFO Channel 08/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-168-30:235653:236140 [5] NCCL INFO Channel 04/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-161-138:96566:97087 [3] NCCL INFO Channel 14/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-161-138:96565:97089 [2] NCCL INFO Channel 15/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-168-30:235652:236147 [4] NCCL INFO Channel 00/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-168-30:235655:236142 [7] NCCL INFO Channel 12/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-161-153:2636090:2636598 [6] NCCL INFO Connected all rings
+ip-26-0-161-153:2636085:2636604 [1] NCCL INFO Channel 14/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-168-52:2722050:2722548 [5] NCCL INFO Channel 08/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-168-52:2722052:2722546 [7] NCCL INFO Channel 13/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-161-178:2815963:2816477 [0] NCCL INFO Channel 00/0 : 16[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-52:2722049:2722547 [4] NCCL INFO Channel 06/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-161-178:2815970:2816470 [7] NCCL INFO Channel 07/0 : 23[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557270:1557763 [4] NCCL INFO Channel 05/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-165-59:1557273:1557760 [7] NCCL INFO Channel 14/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-138:96568:97091 [5] NCCL INFO Channel 11/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-138:96566:97087 [3] NCCL INFO Channel 15/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-59:1557271:1557762 [5] NCCL INFO Channel 12/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-168-30:235653:236140 [5] NCCL INFO Channel 08/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-168-30:235652:236147 [4] NCCL INFO Channel 03/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-168-30:235655:236142 [7] NCCL INFO Channel 13/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-161-153:2636089:2636600 [5] NCCL INFO Channel 02/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-168-52:2722050:2722548 [5] NCCL INFO Channel 12/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-161-138:96564:97088 [1] NCCL INFO Channel 09/0 : 1[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-52:2722052:2722546 [7] NCCL INFO Channel 14/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-168-52:2722049:2722547 [4] NCCL INFO Channel 07/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-161-138:96564:97088 [1] NCCL INFO Channel 01/0 : 17[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636085:2636604 [1] NCCL INFO Channel 15/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-161-138:96564:97088 [1] NCCL INFO Channel 01/0 : 1[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96568:97091 [5] NCCL INFO Channel 12/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-165-59:1557270:1557763 [4] NCCL INFO Channel 06/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-165-59:1557273:1557760 [7] NCCL INFO Channel 15/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-138:96570:97093 [7] NCCL INFO Channel 15/0 : 7[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235653:236140 [5] NCCL INFO Channel 12/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-161-138:96565:97089 [2] NCCL INFO Channel 10/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96563:97086 [0] NCCL INFO Channel 08/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235652:236147 [4] NCCL INFO Channel 05/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-161-138:96568:97091 [5] NCCL INFO Channel 13/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-168-30:235655:236142 [7] NCCL INFO Channel 14/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-161-138:96566:97087 [3] NCCL INFO Channel 11/0 : 3[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-52:2722052:2722546 [7] NCCL INFO Channel 15/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-161-138:96570:97093 [7] NCCL INFO Channel 07/0 : 23[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96563:97086 [0] NCCL INFO Channel 00/0 : 16[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96565:97089 [2] NCCL INFO Channel 02/0 : 18[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96566:97087 [3] NCCL INFO Channel 03/0 : 19[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96568:97091 [5] NCCL INFO Channel 15/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-161-138:96570:97093 [7] NCCL INFO Channel 07/0 : 7[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96563:97086 [0] NCCL INFO Channel 00/0 : 0[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96565:97089 [2] NCCL INFO Channel 02/0 : 2[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96566:97087 [3] NCCL INFO Channel 03/0 : 3[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636089:2636600 [5] NCCL INFO Channel 03/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-168-52:2722052:2722546 [7] NCCL INFO Channel 00/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-168-30:235652:236147 [4] NCCL INFO Channel 06/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-165-59:1557270:1557763 [4] NCCL INFO Channel 07/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-168-30:235655:236142 [7] NCCL INFO Channel 15/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-168-52:2722052:2722546 [7] NCCL INFO Channel 06/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-161-138:96567:97090 [4] NCCL INFO Channel 12/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96569:97092 [6] NCCL INFO Channel 14/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96568:97091 [5] NCCL INFO Channel 13/0 : 5[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557270:1557763 [4] NCCL INFO Channel 08/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-153:2636084:2636605 [0] NCCL INFO Channel 05/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-161-138:96567:97090 [4] NCCL INFO Channel 04/0 : 20[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722049:2722547 [4] NCCL INFO Channel 08/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-161-138:96569:97092 [6] NCCL INFO Channel 06/0 : 22[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96567:97090 [4] NCCL INFO Channel 04/0 : 4[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96568:97091 [5] NCCL INFO Channel 05/0 : 21[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235652:236147 [4] NCCL INFO Channel 07/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-161-138:96569:97092 [6] NCCL INFO Channel 06/0 : 6[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557270:1557763 [4] NCCL INFO Channel 11/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-138:96568:97091 [5] NCCL INFO Channel 05/0 : 5[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-52:2722052:2722546 [7] NCCL INFO Channel 08/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-168-30:235655:236142 [7] NCCL INFO Channel 00/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-168-52:2722049:2722547 [4] NCCL INFO Channel 11/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-165-59:1557270:1557763 [4] NCCL INFO Channel 13/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-153:2636089:2636600 [5] NCCL INFO Channel 04/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-168-30:235652:236147 [4] NCCL INFO Channel 08/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-168-52:2722052:2722546 [7] NCCL INFO Channel 14/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-165-59:1557270:1557763 [4] NCCL INFO Channel 14/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-168-52:2722049:2722547 [4] NCCL INFO Channel 13/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-161-153:2636084:2636605 [0] NCCL INFO Channel 06/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-168-30:235655:236142 [7] NCCL INFO Channel 06/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-165-59:1557270:1557763 [4] NCCL INFO Channel 15/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-168-30:235652:236147 [4] NCCL INFO Channel 11/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-161-153:2636089:2636600 [5] NCCL INFO Channel 05/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-165-59:1557269:1557765 [3] NCCL INFO Channel 00/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-168-30:235655:236142 [7] NCCL INFO Channel 08/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-165-59:1557269:1557765 [3] NCCL INFO Channel 02/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-168-30:235652:236147 [4] NCCL INFO Channel 13/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-165-59:1557273:1557760 [7] NCCL INFO Channel 00/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-153:2636084:2636605 [0] NCCL INFO Channel 07/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-165-59:1557269:1557765 [3] NCCL INFO Channel 08/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-165-59:1557273:1557760 [7] NCCL INFO Channel 06/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-153:2636089:2636600 [5] NCCL INFO Channel 07/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-161-153:2636084:2636605 [0] NCCL INFO Channel 13/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-165-59:1557273:1557760 [7] NCCL INFO Channel 08/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-165-59:1557269:1557765 [3] NCCL INFO Channel 10/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-153:2636089:2636600 [5] NCCL INFO Channel 09/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-161-153:2636087:2636603 [3] NCCL INFO Channel 01/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-161-153:2636084:2636605 [0] NCCL INFO Channel 14/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-168-30:235655:236142 [7] NCCL INFO Channel 14/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-168-30:235651:236141 [3] NCCL INFO Channel 00/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-168-30:235652:236147 [4] NCCL INFO Channel 14/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-165-59:1557273:1557760 [7] NCCL INFO Channel 14/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-153:2636089:2636600 [5] NCCL INFO Channel 10/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-168-52:2722049:2722547 [4] NCCL INFO Channel 14/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-161-153:2636087:2636603 [3] NCCL INFO Channel 02/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-161-153:2636088:2636601 [4] NCCL INFO Channel 01/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-161-153:2636087:2636603 [3] NCCL INFO Channel 03/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-161-153:2636084:2636605 [0] NCCL INFO Channel 15/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-161-153:2636088:2636601 [4] NCCL INFO Channel 02/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-161-153:2636087:2636603 [3] NCCL INFO Channel 05/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-161-153:2636088:2636601 [4] NCCL INFO Channel 03/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-161-153:2636089:2636600 [5] NCCL INFO Channel 11/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-161-153:2636087:2636603 [3] NCCL INFO Channel 06/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-161-153:2636086:2636602 [2] NCCL INFO Connected all rings
+ip-26-0-161-153:2636088:2636601 [4] NCCL INFO Channel 04/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-161-153:2636089:2636600 [5] NCCL INFO Channel 12/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-161-153:2636087:2636603 [3] NCCL INFO Channel 07/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-161-153:2636088:2636601 [4] NCCL INFO Channel 06/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-161-153:2636089:2636600 [5] NCCL INFO Channel 13/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-168-52:2722049:2722547 [4] NCCL INFO Channel 15/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-161-153:2636087:2636603 [3] NCCL INFO Channel 09/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-161-153:2636090:2636598 [6] NCCL INFO Channel 01/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-168-30:235651:236141 [3] NCCL INFO Channel 02/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-168-30:235652:236147 [4] NCCL INFO Channel 15/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-161-153:2636089:2636600 [5] NCCL INFO Channel 15/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-161-153:2636088:2636601 [4] NCCL INFO Channel 07/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-161-153:2636087:2636603 [3] NCCL INFO Channel 10/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-161-153:2636090:2636598 [6] NCCL INFO Channel 02/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-161-153:2636088:2636601 [4] NCCL INFO Channel 09/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-161-153:2636087:2636603 [3] NCCL INFO Channel 11/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-161-153:2636090:2636598 [6] NCCL INFO Channel 03/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-161-153:2636088:2636601 [4] NCCL INFO Channel 10/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-161-153:2636087:2636603 [3] NCCL INFO Channel 13/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-161-153:2636090:2636598 [6] NCCL INFO Channel 04/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-161-153:2636088:2636601 [4] NCCL INFO Channel 11/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-161-153:2636087:2636603 [3] NCCL INFO Channel 14/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-161-153:2636090:2636598 [6] NCCL INFO Channel 05/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-161-153:2636088:2636601 [4] NCCL INFO Channel 12/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-161-153:2636087:2636603 [3] NCCL INFO Channel 15/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-161-153:2636090:2636598 [6] NCCL INFO Channel 06/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-161-153:2636088:2636601 [4] NCCL INFO Channel 14/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-161-153:2636090:2636598 [6] NCCL INFO Channel 09/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-161-153:2636088:2636601 [4] NCCL INFO Channel 15/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-161-153:2636090:2636598 [6] NCCL INFO Channel 10/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-161-153:2636090:2636598 [6] NCCL INFO Channel 11/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-161-153:2636088:2636601 [4] NCCL INFO Channel 04/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636088:2636601 [4] NCCL INFO Channel 12/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636090:2636598 [6] NCCL INFO Channel 12/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-161-153:2636088:2636601 [4] NCCL INFO Channel 04/0 : 20[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636090:2636598 [6] NCCL INFO Channel 13/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-161-153:2636088:2636601 [4] NCCL INFO Channel 04/0 : 4[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815967:2816473 [4] NCCL INFO Channel 12/0 : 12[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636088:2636601 [4] NCCL INFO Channel 04/0 : 20[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815967:2816473 [4] NCCL INFO Channel 12/0 : 28[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636088:2636601 [4] NCCL INFO Channel 04/0 : 28[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636086:2636602 [2] NCCL INFO Channel 01/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-161-153:2636088:2636601 [4] NCCL INFO Channel 04/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636090:2636598 [6] NCCL INFO Channel 14/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-161-153:2636088:2636601 [4] NCCL INFO Channel 12/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96567:97090 [4] NCCL INFO Channel 12/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62885:63377 [4] NCCL INFO Channel 12/0 : 28[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62885:63377 [4] NCCL INFO Channel 12/0 : 12[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636086:2636602 [2] NCCL INFO Channel 02/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-161-142:62885:63377 [4] NCCL INFO Channel 04/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815967:2816473 [4] NCCL INFO Channel 04/0 : 28[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62885:63377 [4] NCCL INFO Channel 12/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235651:236141 [3] NCCL INFO Channel 08/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-161-142:62885:63377 [4] NCCL INFO Channel 12/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636086:2636602 [2] NCCL INFO Channel 04/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-161-153:2636089:2636600 [5] NCCL INFO Channel 05/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636089:2636600 [5] NCCL INFO Channel 13/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636090:2636598 [6] NCCL INFO Channel 06/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636089:2636600 [5] NCCL INFO Channel 05/0 : 21[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636090:2636598 [6] NCCL INFO Channel 14/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636090:2636598 [6] NCCL INFO Channel 06/0 : 22[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62886:63378 [5] NCCL INFO Channel 13/0 : 29[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62886:63378 [5] NCCL INFO Channel 13/0 : 13[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636089:2636600 [5] NCCL INFO Channel 05/0 : 5[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62887:63379 [6] NCCL INFO Channel 14/0 : 30[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815968:2816472 [5] NCCL INFO Channel 13/0 : 13[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636086:2636602 [2] NCCL INFO Channel 05/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-161-142:62887:63379 [6] NCCL INFO Channel 14/0 : 14[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636089:2636600 [5] NCCL INFO Channel 05/0 : 21[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636090:2636598 [6] NCCL INFO Channel 06/0 : 6[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636090:2636598 [6] NCCL INFO Channel 06/0 : 22[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636089:2636600 [5] NCCL INFO Channel 05/0 : 29[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96568:97091 [5] NCCL INFO Channel 13/0 : 13[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636089:2636600 [5] NCCL INFO Channel 05/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815967:2816473 [4] NCCL INFO Channel 00/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-161-153:2636090:2636598 [6] NCCL INFO Channel 06/0 : 30[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96569:97092 [6] NCCL INFO Channel 14/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636089:2636600 [5] NCCL INFO Channel 13/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636090:2636598 [6] NCCL INFO Channel 06/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636090:2636598 [6] NCCL INFO Channel 14/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96567:97090 [4] NCCL INFO Channel 00/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-178:2815967:2816473 [4] NCCL INFO Channel 03/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-161-153:2636086:2636602 [2] NCCL INFO Channel 06/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-161-138:96567:97090 [4] NCCL INFO Channel 03/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-178:2815967:2816473 [4] NCCL INFO Channel 05/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-161-138:96567:97090 [4] NCCL INFO Channel 05/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-178:2815969:2816471 [6] NCCL INFO Channel 14/0 : 14[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636086:2636602 [2] NCCL INFO Channel 07/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-161-178:2815967:2816473 [4] NCCL INFO Channel 06/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-161-178:2815969:2816471 [6] NCCL INFO Channel 14/0 : 30[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96567:97090 [4] NCCL INFO Channel 06/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-178:2815969:2816471 [6] NCCL INFO Channel 06/0 : 30[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62887:63379 [6] NCCL INFO Channel 06/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815967:2816473 [4] NCCL INFO Channel 07/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-161-142:62887:63379 [6] NCCL INFO Channel 14/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96567:97090 [4] NCCL INFO Channel 07/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-142:62887:63379 [6] NCCL INFO Channel 14/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815967:2816473 [4] NCCL INFO Channel 08/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-161-153:2636086:2636602 [2] NCCL INFO Channel 09/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-161-138:96567:97090 [4] NCCL INFO Channel 08/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-153:2636088:2636601 [4] NCCL INFO Channel 00/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-161-142:62885:63377 [4] NCCL INFO Channel 00/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-161-138:96567:97090 [4] NCCL INFO Channel 11/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-178:2815967:2816473 [4] NCCL INFO Channel 11/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-161-142:62885:63377 [4] NCCL INFO Channel 03/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-168-34:272152:272647 [2] NCCL INFO Channel 00/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-161-138:96567:97090 [4] NCCL INFO Channel 13/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-178:2815967:2816473 [4] NCCL INFO Channel 13/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-161-153:2636086:2636602 [2] NCCL INFO Channel 10/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-161-178:2815968:2816472 [5] NCCL INFO Channel 13/0 : 29[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636088:2636601 [4] NCCL INFO Channel 03/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-161-142:62885:63377 [4] NCCL INFO Channel 05/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-168-34:272157:272649 [7] NCCL INFO Channel 06/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-161-138:96567:97090 [4] NCCL INFO Channel 14/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-178:2815968:2816472 [5] NCCL INFO Channel 05/0 : 29[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62886:63378 [5] NCCL INFO Channel 05/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815967:2816473 [4] NCCL INFO Channel 14/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-161-142:62886:63378 [5] NCCL INFO Channel 13/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62886:63378 [5] NCCL INFO Channel 13/0 : 13[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62885:63377 [4] NCCL INFO Channel 06/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-161-138:96567:97090 [4] NCCL INFO Channel 15/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-161-142:62885:63377 [4] NCCL INFO Channel 07/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-161-142:62887:63379 [6] NCCL INFO Channel 00/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-168-34:272152:272647 [2] NCCL INFO Channel 01/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-161-153:2636086:2636602 [2] NCCL INFO Channel 12/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-161-153:2636088:2636601 [4] NCCL INFO Channel 05/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-161-138:96569:97092 [6] NCCL INFO Channel 00/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-168-34:272157:272649 [7] NCCL INFO Channel 07/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-161-142:62885:63377 [4] NCCL INFO Channel 08/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-161-142:62887:63379 [6] NCCL INFO Channel 05/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-161-138:96569:97092 [6] NCCL INFO Channel 05/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-178:2815969:2816471 [6] NCCL INFO Channel 00/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-161-153:2636086:2636602 [2] NCCL INFO Channel 13/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-168-34:272152:272647 [2] NCCL INFO Channel 08/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-161-138:96569:97092 [6] NCCL INFO Channel 08/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-178:2815969:2816471 [6] NCCL INFO Channel 05/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-161-153:2636088:2636601 [4] NCCL INFO Channel 06/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-161-142:62885:63377 [4] NCCL INFO Channel 11/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-161-142:62887:63379 [6] NCCL INFO Channel 08/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-168-34:272157:272649 [7] NCCL INFO Channel 09/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-161-178:2815969:2816471 [6] NCCL INFO Channel 08/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-161-138:96569:97092 [6] NCCL INFO Channel 13/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-161-153:2636091:2636599 [7] NCCL INFO Channel 07/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96568:97091 [5] NCCL INFO Channel 00/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-153:2636091:2636599 [7] NCCL INFO Channel 15/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636091:2636599 [7] NCCL INFO Channel 07/0 : 23[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62885:63377 [4] NCCL INFO Channel 13/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-161-178:2815969:2816471 [6] NCCL INFO Channel 13/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-161-142:62887:63379 [6] NCCL INFO Channel 13/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-161-153:2636086:2636602 [2] NCCL INFO Channel 14/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-161-153:2636090:2636598 [6] NCCL INFO Channel 00/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-161-153:2636088:2636601 [4] NCCL INFO Channel 07/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-161-138:96568:97091 [5] NCCL INFO Channel 04/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-142:62888:63380 [7] NCCL INFO Channel 15/0 : 31[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62886:63378 [5] NCCL INFO Channel 00/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-168-34:272152:272647 [2] NCCL INFO Channel 09/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-161-142:62888:63380 [7] NCCL INFO Channel 15/0 : 15[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636091:2636599 [7] NCCL INFO Channel 07/0 : 7[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272155:272652 [5] NCCL INFO Channel 00/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-161-153:2636091:2636599 [7] NCCL INFO Channel 07/0 : 23[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96568:97091 [5] NCCL INFO Channel 08/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-142:62885:63377 [4] NCCL INFO Channel 14/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-168-34:272157:272649 [7] NCCL INFO Channel 10/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-161-178:2815968:2816472 [5] NCCL INFO Channel 00/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-161-178:2815967:2816473 [4] NCCL INFO Channel 15/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-161-153:2636091:2636599 [7] NCCL INFO Channel 07/0 : 31[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96570:97093 [7] NCCL INFO Channel 15/0 : 15[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815970:2816470 [7] NCCL INFO Channel 15/0 : 15[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636091:2636599 [7] NCCL INFO Channel 07/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96568:97091 [5] NCCL INFO Channel 12/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-161-142:62886:63378 [5] NCCL INFO Channel 04/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-161-178:2815970:2816470 [7] NCCL INFO Channel 15/0 : 31[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636091:2636599 [7] NCCL INFO Channel 15/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62885:63377 [4] NCCL INFO Channel 15/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-161-178:2815970:2816470 [7] NCCL INFO Channel 07/0 : 31[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636086:2636602 [2] NCCL INFO Channel 15/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-161-153:2636090:2636598 [6] NCCL INFO Channel 05/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-161-142:62888:63380 [7] NCCL INFO Channel 07/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636088:2636601 [4] NCCL INFO Channel 08/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-161-142:62888:63380 [7] NCCL INFO Channel 15/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815968:2816472 [5] NCCL INFO Channel 04/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-161-142:62886:63378 [5] NCCL INFO Channel 08/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-168-34:272154:272650 [4] NCCL INFO Channel 00/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-161-178:2815970:2816470 [7] NCCL INFO Channel 01/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-161-142:62888:63380 [7] NCCL INFO Channel 15/0 : 15[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636089:2636600 [5] NCCL INFO Channel 00/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-161-153:2636091:2636599 [7] NCCL INFO Channel 01/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-161-142:62886:63378 [5] NCCL INFO Channel 12/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-168-34:272155:272652 [5] NCCL INFO Channel 04/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-161-138:96570:97093 [7] NCCL INFO Channel 01/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-142:62888:63380 [7] NCCL INFO Channel 01/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-168-34:272157:272649 [7] NCCL INFO Channel 11/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-168-52:2722048:2722545 [3] NCCL INFO Channel 00/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-161-178:2815968:2816472 [5] NCCL INFO Channel 08/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-161-178:2815970:2816470 [7] NCCL INFO Channel 02/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-168-30:235651:236141 [3] NCCL INFO Channel 10/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-161-178:2815968:2816472 [5] NCCL INFO Channel 12/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-161-178:2815970:2816470 [7] NCCL INFO Channel 03/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-168-34:272154:272650 [4] NCCL INFO Channel 03/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-161-153:2636090:2636598 [6] NCCL INFO Channel 08/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-168-34:272155:272652 [5] NCCL INFO Channel 08/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-168-34:272157:272649 [7] NCCL INFO Channel 12/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-161-153:2636088:2636601 [4] NCCL INFO Channel 11/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-161-153:2636089:2636600 [5] NCCL INFO Channel 04/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-161-153:2636084:2636605 [0] NCCL INFO Channel 00/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636084:2636605 [0] NCCL INFO Channel 08/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636091:2636599 [7] NCCL INFO Channel 02/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-161-153:2636084:2636605 [0] NCCL INFO Channel 00/0 : 16[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636085:2636604 [1] NCCL INFO Channel 01/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815970:2816470 [7] NCCL INFO Channel 04/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-161-153:2636087:2636603 [3] NCCL INFO Channel 03/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636085:2636604 [1] NCCL INFO Channel 09/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636086:2636602 [2] NCCL INFO Channel 02/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636087:2636603 [3] NCCL INFO Channel 11/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636086:2636602 [2] NCCL INFO Channel 10/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636085:2636604 [1] NCCL INFO Channel 01/0 : 17[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636087:2636603 [3] NCCL INFO Channel 03/0 : 19[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636086:2636602 [2] NCCL INFO Channel 02/0 : 18[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636084:2636605 [0] NCCL INFO Channel 00/0 : 0[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815963:2816477 [0] NCCL INFO Channel 08/0 : 8[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636084:2636605 [0] NCCL INFO Channel 00/0 : 16[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815963:2816477 [0] NCCL INFO Channel 08/0 : 24[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636085:2636604 [1] NCCL INFO Channel 01/0 : 1[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815964:2816476 [1] NCCL INFO Channel 09/0 : 9[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62888:63380 [7] NCCL INFO Channel 02/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-168-34:272154:272650 [4] NCCL INFO Channel 05/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-161-153:2636086:2636602 [2] NCCL INFO Channel 02/0 : 2[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636084:2636605 [0] NCCL INFO Channel 00/0 : 24[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815965:2816475 [2] NCCL INFO Channel 10/0 : 10[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96563:97086 [0] NCCL INFO Channel 08/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272155:272652 [5] NCCL INFO Channel 12/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-161-153:2636085:2636604 [1] NCCL INFO Channel 01/0 : 17[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636090:2636598 [6] NCCL INFO Channel 13/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-161-178:2815964:2816476 [1] NCCL INFO Channel 09/0 : 25[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636086:2636602 [2] NCCL INFO Channel 02/0 : 18[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636084:2636605 [0] NCCL INFO Channel 00/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636089:2636600 [5] NCCL INFO Channel 08/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-161-153:2636088:2636601 [4] NCCL INFO Channel 13/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-168-34:272157:272649 [7] NCCL INFO Channel 13/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-161-178:2815965:2816475 [2] NCCL INFO Channel 10/0 : 26[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636091:2636599 [7] NCCL INFO Channel 03/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-161-142:62882:63374 [1] NCCL INFO Channel 09/0 : 25[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636084:2636605 [0] NCCL INFO Channel 08/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62882:63374 [1] NCCL INFO Channel 09/0 : 9[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636085:2636604 [1] NCCL INFO Channel 01/0 : 25[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96564:97088 [1] NCCL INFO Channel 09/0 : 9[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62888:63380 [7] NCCL INFO Channel 03/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-161-153:2636086:2636602 [2] NCCL INFO Channel 02/0 : 26[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96565:97089 [2] NCCL INFO Channel 10/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636085:2636604 [1] NCCL INFO Channel 01/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636086:2636602 [2] NCCL INFO Channel 02/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96570:97093 [7] NCCL INFO Channel 02/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-153:2636085:2636604 [1] NCCL INFO Channel 09/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636086:2636602 [2] NCCL INFO Channel 10/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62882:63374 [1] NCCL INFO Channel 01/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815964:2816476 [1] NCCL INFO Channel 01/0 : 25[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62888:63380 [7] NCCL INFO Channel 04/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-161-142:62882:63374 [1] NCCL INFO Channel 09/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62882:63374 [1] NCCL INFO Channel 09/0 : 9[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62881:63375 [0] NCCL INFO Channel 08/0 : 24[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815970:2816470 [7] NCCL INFO Channel 05/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-161-142:62883:63373 [2] NCCL INFO Channel 10/0 : 26[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62881:63375 [0] NCCL INFO Channel 08/0 : 8[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62883:63373 [2] NCCL INFO Channel 10/0 : 10[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62888:63380 [7] NCCL INFO Channel 05/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-168-34:272154:272650 [4] NCCL INFO Channel 06/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-161-142:62881:63375 [0] NCCL INFO Channel 00/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636087:2636603 [3] NCCL INFO Channel 03/0 : 3[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815963:2816477 [0] NCCL INFO Channel 00/0 : 24[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272157:272649 [7] NCCL INFO Channel 14/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-161-142:62883:63373 [2] NCCL INFO Channel 02/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62881:63375 [0] NCCL INFO Channel 08/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815966:2816474 [3] NCCL INFO Channel 11/0 : 11[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815965:2816475 [2] NCCL INFO Channel 02/0 : 26[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636087:2636603 [3] NCCL INFO Channel 03/0 : 19[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62883:63373 [2] NCCL INFO Channel 10/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62881:63375 [0] NCCL INFO Channel 08/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815966:2816474 [3] NCCL INFO Channel 11/0 : 27[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636089:2636600 [5] NCCL INFO Channel 12/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-161-153:2636088:2636601 [4] NCCL INFO Channel 14/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-161-153:2636091:2636599 [7] NCCL INFO Channel 04/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-161-142:62884:63376 [3] NCCL INFO Channel 11/0 : 27[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62883:63373 [2] NCCL INFO Channel 10/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62884:63376 [3] NCCL INFO Channel 11/0 : 11[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636087:2636603 [3] NCCL INFO Channel 03/0 : 27[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96566:97087 [3] NCCL INFO Channel 11/0 : 11[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636087:2636603 [3] NCCL INFO Channel 03/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636087:2636603 [3] NCCL INFO Channel 11/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62884:63376 [3] NCCL INFO Channel 03/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815966:2816474 [3] NCCL INFO Channel 03/0 : 27[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62884:63376 [3] NCCL INFO Channel 11/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272154:272650 [4] NCCL INFO Channel 07/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-161-142:62884:63376 [3] NCCL INFO Channel 11/0 : 11[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272157:272649 [7] NCCL INFO Channel 15/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-168-34:272154:272650 [4] NCCL INFO Channel 08/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-168-34:272157:272649 [7] NCCL INFO Channel 00/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-161-153:2636088:2636601 [4] NCCL INFO Channel 15/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-161-153:2636091:2636599 [7] NCCL INFO Channel 05/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-161-142:62888:63380 [7] NCCL INFO Channel 06/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-168-34:272154:272650 [4] NCCL INFO Channel 11/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-168-34:272153:272651 [3] NCCL INFO Channel 00/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-168-34:272157:272649 [7] NCCL INFO Channel 06/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-161-138:96570:97093 [7] NCCL INFO Channel 03/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-178:2815970:2816470 [7] NCCL INFO Channel 06/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-168-34:272154:272650 [4] NCCL INFO Channel 13/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-161-153:2636091:2636599 [7] NCCL INFO Channel 06/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-168-34:272157:272649 [7] NCCL INFO Channel 08/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-168-34:272153:272651 [3] NCCL INFO Channel 02/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-161-138:96570:97093 [7] NCCL INFO Channel 04/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-178:2815970:2816470 [7] NCCL INFO Channel 07/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-161-142:62888:63380 [7] NCCL INFO Channel 07/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-161-153:2636091:2636599 [7] NCCL INFO Channel 07/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-168-34:272154:272650 [4] NCCL INFO Channel 14/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-168-34:272153:272651 [3] NCCL INFO Channel 08/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-168-34:272157:272649 [7] NCCL INFO Channel 14/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-161-138:96565:97089 [2] NCCL INFO Channel 00/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-138:96570:97093 [7] NCCL INFO Channel 05/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-142:62888:63380 [7] NCCL INFO Channel 09/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-161-178:2815970:2816470 [7] NCCL INFO Channel 09/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-168-52:2722048:2722545 [3] NCCL INFO Channel 02/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-161-153:2636091:2636599 [7] NCCL INFO Channel 09/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-161-153:2636086:2636602 [2] NCCL INFO Channel 00/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-161-153:2636085:2636604 [1] NCCL INFO Channel 00/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-161-178:2815970:2816470 [7] NCCL INFO Channel 10/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-161-178:2815964:2816476 [1] NCCL INFO Channel 00/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-168-34:272154:272650 [4] NCCL INFO Channel 15/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-161-153:2636091:2636599 [7] NCCL INFO Channel 10/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-161-153:2636091:2636599 [7] NCCL INFO Channel 11/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-168-34:272153:272651 [3] NCCL INFO Channel 10/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-161-142:62888:63380 [7] NCCL INFO Channel 10/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-161-142:62882:63374 [1] NCCL INFO Channel 00/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-161-178:2815965:2816475 [2] NCCL INFO Channel 00/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-161-142:62883:63373 [2] NCCL INFO Channel 00/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-161-138:96565:97089 [2] NCCL INFO Channel 01/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-138:96570:97093 [7] NCCL INFO Channel 06/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-178:2815970:2816470 [7] NCCL INFO Channel 11/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-161-178:2815964:2816476 [1] NCCL INFO Channel 08/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-161-142:62888:63380 [7] NCCL INFO Channel 11/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-161-142:62882:63374 [1] NCCL INFO Channel 08/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-161-178:2815965:2816475 [2] NCCL INFO Channel 01/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-168-52:2722048:2722545 [3] NCCL INFO Channel 08/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-161-153:2636086:2636602 [2] NCCL INFO Channel 01/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-161-138:96564:97088 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-138:96570:97093 [7] NCCL INFO Channel 07/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-153:2636085:2636604 [1] NCCL INFO Channel 08/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-161-142:62883:63373 [2] NCCL INFO Channel 01/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-161-153:2636086:2636602 [2] NCCL INFO Channel 08/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-161-153:2636091:2636599 [7] NCCL INFO Channel 12/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-161-178:2815965:2816475 [2] NCCL INFO Channel 08/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-161-138:96565:97089 [2] NCCL INFO Channel 08/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-178:2815970:2816470 [7] NCCL INFO Channel 12/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-161-142:62888:63380 [7] NCCL INFO Channel 12/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-161-153:2636086:2636602 [2] NCCL INFO Channel 09/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-161-178:2815965:2816475 [2] NCCL INFO Channel 09/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-161-153:2636091:2636599 [7] NCCL INFO Channel 13/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-161-178:2815970:2816470 [7] NCCL INFO Channel 13/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-161-142:62883:63373 [2] NCCL INFO Channel 08/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-161-138:96564:97088 [1] NCCL INFO Channel 08/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-161-138:96570:97093 [7] NCCL INFO Channel 09/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-138:96565:97089 [2] NCCL INFO Channel 09/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-161-142:62888:63380 [7] NCCL INFO Channel 13/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-161-178:2815970:2816470 [7] NCCL INFO Channel 14/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-161-153:2636091:2636599 [7] NCCL INFO Channel 14/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-161-142:62883:63373 [2] NCCL INFO Channel 09/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-161-138:96570:97093 [7] NCCL INFO Channel 10/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-142:62888:63380 [7] NCCL INFO Channel 14/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-161-153:2636091:2636599 [7] NCCL INFO Channel 15/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-168-52:2722048:2722545 [3] NCCL INFO Channel 10/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-161-138:96570:97093 [7] NCCL INFO Channel 11/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-142:62888:63380 [7] NCCL INFO Channel 15/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-161-178:2815970:2816470 [7] NCCL INFO Channel 15/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-161-153:2636087:2636603 [3] NCCL INFO Channel 00/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-161-153:2636087:2636603 [3] NCCL INFO Channel 02/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-161-138:96570:97093 [7] NCCL INFO Channel 12/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-178:2815966:2816474 [3] NCCL INFO Channel 00/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-161-153:2636087:2636603 [3] NCCL INFO Channel 08/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-161-138:96570:97093 [7] NCCL INFO Channel 13/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-142:62884:63376 [3] NCCL INFO Channel 00/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-161-178:2815966:2816474 [3] NCCL INFO Channel 02/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-161-142:62884:63376 [3] NCCL INFO Channel 02/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-161-153:2636087:2636603 [3] NCCL INFO Channel 10/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-161-138:96570:97093 [7] NCCL INFO Channel 14/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-178:2815966:2816474 [3] NCCL INFO Channel 08/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-161-153:2636091:2636599 [7] NCCL INFO Channel 00/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-161-142:62884:63376 [3] NCCL INFO Channel 08/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-161-178:2815966:2816474 [3] NCCL INFO Channel 10/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-161-138:96570:97093 [7] NCCL INFO Channel 15/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-161-153:2636091:2636599 [7] NCCL INFO Channel 06/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-161-142:62884:63376 [3] NCCL INFO Channel 10/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-161-178:2815970:2816470 [7] NCCL INFO Channel 00/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-161-138:96566:97087 [3] NCCL INFO Channel 00/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-142:62888:63380 [7] NCCL INFO Channel 00/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-161-138:96566:97087 [3] NCCL INFO Channel 02/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-153:2636091:2636599 [7] NCCL INFO Channel 08/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-161-138:96566:97087 [3] NCCL INFO Channel 08/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-138:96570:97093 [7] NCCL INFO Channel 00/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-178:2815970:2816470 [7] NCCL INFO Channel 06/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-161-138:96566:97087 [3] NCCL INFO Channel 10/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-161-138:96570:97093 [7] NCCL INFO Channel 06/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-138:96570:97093 [7] NCCL INFO Channel 08/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-142:62888:63380 [7] NCCL INFO Channel 06/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-161-142:62888:63380 [7] NCCL INFO Channel 08/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-161-142:62888:63380 [7] NCCL INFO Channel 14/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-161-138:96570:97093 [7] NCCL INFO Channel 14/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-161-178:2815970:2816470 [7] NCCL INFO Channel 08/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-161-153:2636091:2636599 [7] NCCL INFO Channel 14/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-161-178:2815970:2816470 [7] NCCL INFO Channel 14/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-161-78:114515:115009 [3] NCCL INFO Connected all trees
+ip-26-0-161-78:114515:115009 [3] NCCL INFO NVLS comm 0xa3a81c0 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-202:153562:154061 [0] NCCL INFO Connected all trees
+ip-26-0-165-202:153562:154061 [0] NCCL INFO NVLS comm 0xaa68a60 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-202:153567:154059 [5] NCCL INFO Connected all trees
+ip-26-0-165-202:153567:154059 [5] NCCL INFO NVLS comm 0xc2061c0 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-123:32705:33203 [1] NCCL INFO Connected all trees
+ip-26-0-161-123:32706:33199 [2] NCCL INFO Connected all trees
+ip-26-0-161-123:32705:33203 [1] NCCL INFO NVLS comm 0xb19e430 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-123:32706:33199 [2] NCCL INFO NVLS comm 0xbc2d3c0 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-202:153568:154056 [6] NCCL INFO Connected all trees
+ip-26-0-165-202:153566:154062 [4] NCCL INFO Connected all trees
+ip-26-0-165-202:153568:154056 [6] NCCL INFO NVLS comm 0xa7ee300 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-202:153566:154062 [4] NCCL INFO NVLS comm 0xaa4e780 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-225:124160:124658 [0] NCCL INFO Connected all trees
+ip-26-0-161-123:32704:33197 [0] NCCL INFO Connected all trees
+ip-26-0-160-225:124160:124658 [0] NCCL INFO NVLS comm 0xb447d40 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-123:32704:33197 [0] NCCL INFO NVLS comm 0xbc119e0 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-202:153563:154060 [1] NCCL INFO Connected all trees
+ip-26-0-165-202:153563:154060 [1] NCCL INFO NVLS comm 0xb9a2ae0 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-202:153569:154055 [7] NCCL INFO Connected all trees
+ip-26-0-165-202:153569:154055 [7] NCCL INFO NVLS comm 0x9d016a0 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-123:32708:33198 [4] NCCL INFO Connected all trees
+ip-26-0-161-123:32708:33198 [4] NCCL INFO NVLS comm 0xba333f0 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-123:32710:33202 [6] NCCL INFO Connected all trees
+ip-26-0-161-123:32710:33202 [6] NCCL INFO NVLS comm 0xb0d3b90 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-202:153565:154058 [3] NCCL INFO Connected all trees
+ip-26-0-165-202:153565:154058 [3] NCCL INFO NVLS comm 0xb9d80c0 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-202:153564:154057 [2] NCCL INFO Connected all trees
+ip-26-0-165-202:153564:154057 [2] NCCL INFO NVLS comm 0x9d98af0 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-225:124167:124664 [7] NCCL INFO Connected all trees
+ip-26-0-161-78:114517:115013 [5] NCCL INFO Connected all trees
+ip-26-0-160-225:124167:124664 [7] NCCL INFO NVLS comm 0xb3f9260 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-78:114517:115013 [5] NCCL INFO NVLS comm 0xb85b760 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-123:32711:33204 [7] NCCL INFO Connected all trees
+ip-26-0-161-123:32711:33204 [7] NCCL INFO NVLS comm 0xc08b660 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-225:124163:124661 [3] NCCL INFO Connected all trees
+ip-26-0-160-225:124163:124661 [3] NCCL INFO NVLS comm 0xb6dea80 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-123:32709:33201 [5] NCCL INFO Connected all trees
+ip-26-0-161-123:32707:33200 [3] NCCL INFO Connected all trees
+ip-26-0-161-123:32709:33201 [5] NCCL INFO NVLS comm 0xb9f5040 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-123:32707:33200 [3] NCCL INFO NVLS comm 0xa02c190 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-225:124165:124662 [5] NCCL INFO Connected all trees
+ip-26-0-160-225:124165:124662 [5] NCCL INFO NVLS comm 0x9fea760 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-225:124164:124663 [4] NCCL INFO Connected all trees
+ip-26-0-160-225:124164:124663 [4] NCCL INFO NVLS comm 0xa721840 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-225:124166:124665 [6] NCCL INFO Connected all trees
+ip-26-0-160-225:124166:124665 [6] NCCL INFO NVLS comm 0xaf0f490 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-78:114516:115010 [4] NCCL INFO Connected all trees
+ip-26-0-161-78:114516:115010 [4] NCCL INFO NVLS comm 0xb3ab6d0 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-78:114512:115007 [0] NCCL INFO Connected all trees
+ip-26-0-161-78:114513:115008 [1] NCCL INFO Connected all trees
+ip-26-0-161-78:114512:115007 [0] NCCL INFO NVLS comm 0xa85e740 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-78:114513:115008 [1] NCCL INFO NVLS comm 0xb40d9a0 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-164:1227468:1227959 [5] NCCL INFO Connected all trees
+ip-26-0-165-164:1227468:1227959 [5] NCCL INFO NVLS comm 0xb1614c0 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-164:1227467:1227960 [4] NCCL INFO Connected all trees
+ip-26-0-165-164:1227467:1227960 [4] NCCL INFO NVLS comm 0xa7d56c0 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-164:1227464:1227963 [1] NCCL INFO Connected all trees
+ip-26-0-161-78:114518:115012 [6] NCCL INFO Connected all trees
+ip-26-0-165-164:1227464:1227963 [1] NCCL INFO NVLS comm 0xb0a70f0 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-78:114518:115012 [6] NCCL INFO NVLS comm 0xb8bc080 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-164:1227465:1227962 [2] NCCL INFO Connected all trees
+ip-26-0-165-164:1227465:1227962 [2] NCCL INFO NVLS comm 0xac59150 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-164:1227463:1227964 [0] NCCL INFO Connected all trees
+ip-26-0-165-164:1227463:1227964 [0] NCCL INFO NVLS comm 0xa7d40d0 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-78:114519:115014 [7] NCCL INFO Connected all trees
+ip-26-0-161-78:114519:115014 [7] NCCL INFO NVLS comm 0xa5040c0 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-78:114514:115011 [2] NCCL INFO Connected all trees
+ip-26-0-161-78:114514:115011 [2] NCCL INFO NVLS comm 0xb2d0ac0 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-164:1227469:1227957 [6] NCCL INFO Connected all trees
+ip-26-0-165-164:1227469:1227957 [6] NCCL INFO NVLS comm 0x9a95350 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-225:124162:124660 [2] NCCL INFO Connected all trees
+ip-26-0-160-225:124162:124660 [2] NCCL INFO NVLS comm 0xba5d2d0 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-164:1227470:1227958 [7] NCCL INFO Connected all trees
+ip-26-0-165-202:153566:154062 [4] NCCL INFO Channel 00/0 : 20[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153562:154061 [0] NCCL INFO Channel 02/0 : 16[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153564:154057 [2] NCCL INFO Channel 00/0 : 18[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153566:154062 [4] NCCL INFO Channel 02/0 : 20[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153562:154061 [0] NCCL INFO Channel 04/0 : 16[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153564:154057 [2] NCCL INFO Channel 04/0 : 18[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153566:154062 [4] NCCL INFO Channel 06/0 : 20[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227470:1227958 [7] NCCL INFO NVLS comm 0xb2dbd00 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-202:153568:154056 [6] NCCL INFO Channel 00/0 : 22[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153562:154061 [0] NCCL INFO Channel 06/0 : 16[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153564:154057 [2] NCCL INFO Channel 06/0 : 18[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153566:154062 [4] NCCL INFO Channel 08/0 : 20[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153568:154056 [6] NCCL INFO Channel 02/0 : 22[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153562:154061 [0] NCCL INFO Channel 08/0 : 16[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153564:154057 [2] NCCL INFO Channel 08/0 : 18[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153566:154062 [4] NCCL INFO Channel 10/0 : 20[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153568:154056 [6] NCCL INFO Channel 04/0 : 22[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153569:154055 [7] NCCL INFO Channel 00/0 : 23[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153562:154061 [0] NCCL INFO Channel 10/0 : 16[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153566:154062 [4] NCCL INFO Channel 12/0 : 20[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153564:154057 [2] NCCL INFO Channel 10/0 : 18[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153568:154056 [6] NCCL INFO Channel 08/0 : 22[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153569:154055 [7] NCCL INFO Channel 02/0 : 23[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153565:154058 [3] NCCL INFO Channel 00/0 : 19[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153562:154061 [0] NCCL INFO Channel 12/0 : 16[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153564:154057 [2] NCCL INFO Channel 12/0 : 18[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153566:154062 [4] NCCL INFO Channel 14/0 : 20[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153568:154056 [6] NCCL INFO Channel 10/0 : 22[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153563:154060 [1] NCCL INFO Channel 00/0 : 17[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153569:154055 [7] NCCL INFO Channel 04/0 : 23[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153565:154058 [3] NCCL INFO Channel 02/0 : 19[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153562:154061 [0] NCCL INFO Channel 14/0 : 16[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153564:154057 [2] NCCL INFO Channel 14/0 : 18[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153568:154056 [6] NCCL INFO Channel 12/0 : 22[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153567:154059 [5] NCCL INFO Channel 00/0 : 21[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153569:154055 [7] NCCL INFO Channel 06/0 : 23[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153563:154060 [1] NCCL INFO Channel 02/0 : 17[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153565:154058 [3] NCCL INFO Channel 04/0 : 19[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153568:154056 [6] NCCL INFO Channel 14/0 : 22[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153567:154059 [5] NCCL INFO Channel 02/0 : 21[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153569:154055 [7] NCCL INFO Channel 08/0 : 23[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153563:154060 [1] NCCL INFO Channel 04/0 : 17[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153565:154058 [3] NCCL INFO Channel 06/0 : 19[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153567:154059 [5] NCCL INFO Channel 04/0 : 21[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153569:154055 [7] NCCL INFO Channel 10/0 : 23[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153563:154060 [1] NCCL INFO Channel 06/0 : 17[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153565:154058 [3] NCCL INFO Channel 08/0 : 19[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153569:154055 [7] NCCL INFO Channel 12/0 : 23[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153567:154059 [5] NCCL INFO Channel 06/0 : 21[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153565:154058 [3] NCCL INFO Channel 10/0 : 19[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153563:154060 [1] NCCL INFO Channel 08/0 : 17[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153569:154055 [7] NCCL INFO Channel 14/0 : 23[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153567:154059 [5] NCCL INFO Channel 08/0 : 21[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153565:154058 [3] NCCL INFO Channel 12/0 : 19[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153563:154060 [1] NCCL INFO Channel 10/0 : 17[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153567:154059 [5] NCCL INFO Channel 10/0 : 21[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153563:154060 [1] NCCL INFO Channel 12/0 : 17[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153565:154058 [3] NCCL INFO Channel 14/0 : 19[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153567:154059 [5] NCCL INFO Channel 12/0 : 21[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153563:154060 [1] NCCL INFO Channel 14/0 : 17[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153567:154059 [5] NCCL INFO Channel 14/0 : 21[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124161:124659 [1] NCCL INFO Connected all trees
+ip-26-0-160-225:124161:124659 [1] NCCL INFO NVLS comm 0x9fafb60 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-242:1364909:1365405 [7] NCCL INFO Connected all trees
+ip-26-0-160-242:1364909:1365405 [7] NCCL INFO NVLS comm 0xb61ae00 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-164:1227466:1227961 [3] NCCL INFO Connected all trees
+ip-26-0-165-164:1227466:1227961 [3] NCCL INFO NVLS comm 0xaebd080 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-242:1364908:1365406 [6] NCCL INFO Connected all trees
+ip-26-0-160-242:1364908:1365406 [6] NCCL INFO NVLS comm 0xbb61720 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-242:1364902:1365410 [0] NCCL INFO Connected all trees
+ip-26-0-160-242:1364902:1365410 [0] NCCL INFO NVLS comm 0xaf66250 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-242:1364904:1365408 [2] NCCL INFO Connected all trees
+ip-26-0-160-242:1364904:1365408 [2] NCCL INFO NVLS comm 0xb849c50 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-242:1364906:1365407 [4] NCCL INFO Connected all trees
+ip-26-0-161-123:32709:33201 [5] NCCL INFO Channel 00/0 : 21[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32709:33201 [5] NCCL INFO Channel 02/0 : 21[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364906:1365407 [4] NCCL INFO NVLS comm 0xb742ff0 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-123:32709:33201 [5] NCCL INFO Channel 04/0 : 21[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32707:33200 [3] NCCL INFO Channel 00/0 : 19[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32711:33204 [7] NCCL INFO Channel 00/0 : 23[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32709:33201 [5] NCCL INFO Channel 06/0 : 21[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32711:33204 [7] NCCL INFO Channel 02/0 : 23[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32709:33201 [5] NCCL INFO Channel 08/0 : 21[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32707:33200 [3] NCCL INFO Channel 02/0 : 19[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32711:33204 [7] NCCL INFO Channel 04/0 : 23[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630207:1630731 [2] NCCL INFO Connected all trees
+ip-26-0-161-123:32709:33201 [5] NCCL INFO Channel 10/0 : 21[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32707:33200 [3] NCCL INFO Channel 04/0 : 19[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32710:33202 [6] NCCL INFO Channel 00/0 : 22[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32706:33199 [2] NCCL INFO Channel 00/0 : 18[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32711:33204 [7] NCCL INFO Channel 06/0 : 23[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32708:33198 [4] NCCL INFO Channel 00/0 : 20[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32709:33201 [5] NCCL INFO Channel 12/0 : 21[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32707:33200 [3] NCCL INFO Channel 06/0 : 19[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32710:33202 [6] NCCL INFO Channel 02/0 : 22[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32706:33199 [2] NCCL INFO Channel 04/0 : 18[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32711:33204 [7] NCCL INFO Channel 08/0 : 23[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32708:33198 [4] NCCL INFO Channel 02/0 : 20[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32705:33203 [1] NCCL INFO Channel 00/0 : 17[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32709:33201 [5] NCCL INFO Channel 14/0 : 21[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32707:33200 [3] NCCL INFO Channel 08/0 : 19[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32710:33202 [6] NCCL INFO Channel 04/0 : 22[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96895:97382 [7] NCCL INFO Connected all trees
+ip-26-0-161-123:32704:33197 [0] NCCL INFO Channel 02/0 : 16[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32711:33204 [7] NCCL INFO Channel 10/0 : 23[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32706:33199 [2] NCCL INFO Channel 06/0 : 18[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32708:33198 [4] NCCL INFO Channel 06/0 : 20[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32707:33200 [3] NCCL INFO Channel 10/0 : 19[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32705:33203 [1] NCCL INFO Channel 02/0 : 17[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32710:33202 [6] NCCL INFO Channel 08/0 : 22[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32711:33204 [7] NCCL INFO Channel 12/0 : 23[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32706:33199 [2] NCCL INFO Channel 08/0 : 18[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32704:33197 [0] NCCL INFO Channel 04/0 : 16[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32708:33198 [4] NCCL INFO Channel 08/0 : 20[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32707:33200 [3] NCCL INFO Channel 12/0 : 19[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32705:33203 [1] NCCL INFO Channel 04/0 : 17[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32711:33204 [7] NCCL INFO Channel 14/0 : 23[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32710:33202 [6] NCCL INFO Channel 10/0 : 22[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32704:33197 [0] NCCL INFO Channel 06/0 : 16[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32706:33199 [2] NCCL INFO Channel 10/0 : 18[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32708:33198 [4] NCCL INFO Channel 10/0 : 20[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32707:33200 [3] NCCL INFO Channel 14/0 : 19[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32705:33203 [1] NCCL INFO Channel 06/0 : 17[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32710:33202 [6] NCCL INFO Channel 12/0 : 22[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32706:33199 [2] NCCL INFO Channel 12/0 : 18[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32704:33197 [0] NCCL INFO Channel 08/0 : 16[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32705:33203 [1] NCCL INFO Channel 08/0 : 17[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32708:33198 [4] NCCL INFO Channel 12/0 : 20[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32710:33202 [6] NCCL INFO Channel 14/0 : 22[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32706:33199 [2] NCCL INFO Channel 14/0 : 18[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630207:1630731 [2] NCCL INFO NVLS comm 0xab227f0 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-123:32705:33203 [1] NCCL INFO Channel 10/0 : 17[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32708:33198 [4] NCCL INFO Channel 14/0 : 20[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32704:33197 [0] NCCL INFO Channel 10/0 : 16[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32705:33203 [1] NCCL INFO Channel 12/0 : 17[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32704:33197 [0] NCCL INFO Channel 12/0 : 16[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96895:97382 [7] NCCL INFO NVLS comm 0xbbd4280 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-123:32705:33203 [1] NCCL INFO Channel 14/0 : 17[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32704:33197 [0] NCCL INFO Channel 14/0 : 16[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630210:1630732 [5] NCCL INFO Connected all trees
+ip-26-0-165-131:1630210:1630732 [5] NCCL INFO NVLS comm 0xb46d6d0 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-242:1364903:1365409 [1] NCCL INFO Connected all trees
+ip-26-0-165-131:1630208:1630729 [3] NCCL INFO Connected all trees
+ip-26-0-165-131:1630208:1630729 [3] NCCL INFO NVLS comm 0x9aa21d0 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-242:1364903:1365409 [1] NCCL INFO NVLS comm 0xbcea600 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-131:1630209:1630730 [4] NCCL INFO Connected all trees
+ip-26-0-165-131:1630206:1630728 [1] NCCL INFO Connected all trees
+ip-26-0-165-131:1630206:1630728 [1] NCCL INFO NVLS comm 0xbb76e40 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-131:1630209:1630730 [4] NCCL INFO NVLS comm 0xb1f6980 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-131:1630205:1630727 [0] NCCL INFO Connected all trees
+ip-26-0-165-131:1630205:1630727 [0] NCCL INFO NVLS comm 0xaeb3380 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-103:96892:97383 [4] NCCL INFO Connected all trees
+ip-26-0-161-103:96892:97383 [4] NCCL INFO NVLS comm 0xb351080 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-103:96888:97387 [0] NCCL INFO Connected all trees
+ip-26-0-161-103:96888:97387 [0] NCCL INFO NVLS comm 0xa5f0fe0 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-103:96891:97385 [3] NCCL INFO Connected all trees
+ip-26-0-161-103:96891:97385 [3] NCCL INFO NVLS comm 0xae813d0 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-103:96893:97384 [5] NCCL INFO Connected all trees
+ip-26-0-161-103:96890:97386 [2] NCCL INFO Connected all trees
+ip-26-0-161-103:96893:97384 [5] NCCL INFO NVLS comm 0x9e9dd50 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-103:96890:97386 [2] NCCL INFO NVLS comm 0xa5cc060 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-103:96889:97388 [1] NCCL INFO Connected all trees
+ip-26-0-161-103:96889:97388 [1] NCCL INFO NVLS comm 0x9f87fe0 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-242:1364905:1365404 [3] NCCL INFO Connected all trees
+ip-26-0-160-242:1364905:1365404 [3] NCCL INFO NVLS comm 0xafe0340 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-131:1630212:1630734 [7] NCCL INFO Connected all trees
+ip-26-0-165-131:1630211:1630733 [6] NCCL INFO Connected all trees
+ip-26-0-165-131:1630212:1630734 [7] NCCL INFO NVLS comm 0x9c95060 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-131:1630211:1630733 [6] NCCL INFO NVLS comm 0xba47990 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-103:96894:97381 [6] NCCL INFO Connected all trees
+ip-26-0-161-103:96894:97381 [6] NCCL INFO NVLS comm 0xa2306a0 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-242:1364907:1365403 [5] NCCL INFO Connected all trees
+ip-26-0-160-242:1364907:1365403 [5] NCCL INFO NVLS comm 0x9e43b50 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-59:1557270:1557763 [4] NCCL INFO Connected all trees
+ip-26-0-165-59:1557270:1557763 [4] NCCL INFO NVLS comm 0xa956c80 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-59:1557271:1557762 [5] NCCL INFO Connected all trees
+ip-26-0-165-59:1557271:1557762 [5] NCCL INFO NVLS comm 0xa4e7e00 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-78:114519:115014 [7] NCCL INFO Channel 01/0 : 7[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114519:115014 [7] NCCL INFO Channel 03/0 : 7[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114519:115014 [7] NCCL INFO Channel 05/0 : 7[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114513:115008 [1] NCCL INFO Channel 01/0 : 1[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114519:115014 [7] NCCL INFO Channel 07/0 : 7[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114513:115008 [1] NCCL INFO Channel 03/0 : 1[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114519:115014 [7] NCCL INFO Channel 09/0 : 7[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114513:115008 [1] NCCL INFO Channel 05/0 : 1[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114512:115007 [0] NCCL INFO Channel 01/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114518:115012 [6] NCCL INFO Channel 01/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114519:115014 [7] NCCL INFO Channel 11/0 : 7[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114517:115013 [5] NCCL INFO Channel 01/0 : 5[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114513:115008 [1] NCCL INFO Channel 07/0 : 1[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114512:115007 [0] NCCL INFO Channel 03/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114516:115010 [4] NCCL INFO Channel 01/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114519:115014 [7] NCCL INFO Channel 13/0 : 7[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114518:115012 [6] NCCL INFO Channel 03/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114514:115011 [2] NCCL INFO Channel 01/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114517:115013 [5] NCCL INFO Channel 03/0 : 5[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114513:115008 [1] NCCL INFO Channel 11/0 : 1[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114512:115007 [0] NCCL INFO Channel 05/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114516:115010 [4] NCCL INFO Channel 03/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114515:115009 [3] NCCL INFO Channel 01/0 : 3[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114518:115012 [6] NCCL INFO Channel 05/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114514:115011 [2] NCCL INFO Channel 03/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114513:115008 [1] NCCL INFO Channel 13/0 : 1[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114517:115013 [5] NCCL INFO Channel 05/0 : 5[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114512:115007 [0] NCCL INFO Channel 07/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114515:115009 [3] NCCL INFO Channel 03/0 : 3[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114516:115010 [4] NCCL INFO Channel 05/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114518:115012 [6] NCCL INFO Channel 07/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114514:115011 [2] NCCL INFO Channel 05/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114513:115008 [1] NCCL INFO Channel 15/0 : 1[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114512:115007 [0] NCCL INFO Channel 09/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114517:115013 [5] NCCL INFO Channel 07/0 : 5[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114515:115009 [3] NCCL INFO Channel 05/0 : 3[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114514:115011 [2] NCCL INFO Channel 07/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114518:115012 [6] NCCL INFO Channel 09/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114516:115010 [4] NCCL INFO Channel 07/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114512:115007 [0] NCCL INFO Channel 11/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114517:115013 [5] NCCL INFO Channel 09/0 : 5[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114515:115009 [3] NCCL INFO Channel 07/0 : 3[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114514:115011 [2] NCCL INFO Channel 09/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114518:115012 [6] NCCL INFO Channel 11/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114516:115010 [4] NCCL INFO Channel 09/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114512:115007 [0] NCCL INFO Channel 13/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114517:115013 [5] NCCL INFO Channel 11/0 : 5[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114514:115011 [2] NCCL INFO Channel 11/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114515:115009 [3] NCCL INFO Channel 09/0 : 3[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114518:115012 [6] NCCL INFO Channel 13/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114512:115007 [0] NCCL INFO Channel 15/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114516:115010 [4] NCCL INFO Channel 11/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114517:115013 [5] NCCL INFO Channel 15/0 : 5[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114514:115011 [2] NCCL INFO Channel 13/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114515:115009 [3] NCCL INFO Channel 13/0 : 3[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114518:115012 [6] NCCL INFO Channel 15/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114516:115010 [4] NCCL INFO Channel 13/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114514:115011 [2] NCCL INFO Channel 15/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114515:115009 [3] NCCL INFO Channel 15/0 : 3[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114516:115010 [4] NCCL INFO Channel 15/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:124164:124663 [4] NCCL INFO Channel 01/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:124166:124665 [6] NCCL INFO Channel 01/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:124164:124663 [4] NCCL INFO Channel 03/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:124160:124658 [0] NCCL INFO Channel 01/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:124166:124665 [6] NCCL INFO Channel 03/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:124164:124663 [4] NCCL INFO Channel 05/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:124160:124658 [0] NCCL INFO Channel 03/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:124166:124665 [6] NCCL INFO Channel 05/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:124164:124663 [4] NCCL INFO Channel 07/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:124166:124665 [6] NCCL INFO Channel 07/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:124160:124658 [0] NCCL INFO Channel 05/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:124164:124663 [4] NCCL INFO Channel 09/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:124162:124660 [2] NCCL INFO Channel 01/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:124161:124659 [1] NCCL INFO Channel 01/0 : 1[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124166:124665 [6] NCCL INFO Channel 09/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:124165:124662 [5] NCCL INFO Channel 01/0 : 5[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124164:124663 [4] NCCL INFO Channel 11/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:124160:124658 [0] NCCL INFO Channel 07/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:124162:124660 [2] NCCL INFO Channel 03/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:124161:124659 [1] NCCL INFO Channel 03/0 : 1[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124167:124664 [7] NCCL INFO Channel 01/0 : 7[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124166:124665 [6] NCCL INFO Channel 11/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:124164:124663 [4] NCCL INFO Channel 13/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:124163:124661 [3] NCCL INFO Channel 01/0 : 3[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:124165:124662 [5] NCCL INFO Channel 03/0 : 5[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124160:124658 [0] NCCL INFO Channel 09/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:124161:124659 [1] NCCL INFO Channel 05/0 : 1[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124162:124660 [2] NCCL INFO Channel 05/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:124166:124665 [6] NCCL INFO Channel 13/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:124167:124664 [7] NCCL INFO Channel 03/0 : 7[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124164:124663 [4] NCCL INFO Channel 15/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:124163:124661 [3] NCCL INFO Channel 03/0 : 3[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:124165:124662 [5] NCCL INFO Channel 05/0 : 5[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124160:124658 [0] NCCL INFO Channel 11/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:124161:124659 [1] NCCL INFO Channel 07/0 : 1[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124166:124665 [6] NCCL INFO Channel 15/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:124162:124660 [2] NCCL INFO Channel 07/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:124167:124664 [7] NCCL INFO Channel 05/0 : 7[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124161:124659 [1] NCCL INFO Channel 11/0 : 1[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124163:124661 [3] NCCL INFO Channel 05/0 : 3[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:124165:124662 [5] NCCL INFO Channel 07/0 : 5[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124167:124664 [7] NCCL INFO Channel 07/0 : 7[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124160:124658 [0] NCCL INFO Channel 13/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:124162:124660 [2] NCCL INFO Channel 09/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:124161:124659 [1] NCCL INFO Channel 13/0 : 1[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124163:124661 [3] NCCL INFO Channel 07/0 : 3[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:124165:124662 [5] NCCL INFO Channel 09/0 : 5[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124167:124664 [7] NCCL INFO Channel 09/0 : 7[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124162:124660 [2] NCCL INFO Channel 11/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:124160:124658 [0] NCCL INFO Channel 15/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:124161:124659 [1] NCCL INFO Channel 15/0 : 1[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124163:124661 [3] NCCL INFO Channel 09/0 : 3[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:124167:124664 [7] NCCL INFO Channel 11/0 : 7[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124165:124662 [5] NCCL INFO Channel 11/0 : 5[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124162:124660 [2] NCCL INFO Channel 13/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:124163:124661 [3] NCCL INFO Channel 13/0 : 3[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:124167:124664 [7] NCCL INFO Channel 13/0 : 7[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124165:124662 [5] NCCL INFO Channel 15/0 : 5[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124163:124661 [3] NCCL INFO Channel 15/0 : 3[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:124162:124660 [2] NCCL INFO Channel 15/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227469:1227957 [6] NCCL INFO Channel 00/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227469:1227957 [6] NCCL INFO Channel 01/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227463:1227964 [0] NCCL INFO Channel 01/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227469:1227957 [6] NCCL INFO Channel 02/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227465:1227962 [2] NCCL INFO Channel 00/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227463:1227964 [0] NCCL INFO Channel 02/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227469:1227957 [6] NCCL INFO Channel 03/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227465:1227962 [2] NCCL INFO Channel 01/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227463:1227964 [0] NCCL INFO Channel 03/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227467:1227960 [4] NCCL INFO Channel 00/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227469:1227957 [6] NCCL INFO Channel 04/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227465:1227962 [2] NCCL INFO Channel 03/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227463:1227964 [0] NCCL INFO Channel 04/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227468:1227959 [5] NCCL INFO Channel 00/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227469:1227957 [6] NCCL INFO Channel 05/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227467:1227960 [4] NCCL INFO Channel 01/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227463:1227964 [0] NCCL INFO Channel 05/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227465:1227962 [2] NCCL INFO Channel 04/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227468:1227959 [5] NCCL INFO Channel 01/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227469:1227957 [6] NCCL INFO Channel 07/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227464:1227963 [1] NCCL INFO Channel 00/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227467:1227960 [4] NCCL INFO Channel 02/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227470:1227958 [7] NCCL INFO Channel 00/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227463:1227964 [0] NCCL INFO Channel 06/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227465:1227962 [2] NCCL INFO Channel 05/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227468:1227959 [5] NCCL INFO Channel 02/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227466:1227961 [3] NCCL INFO Channel 00/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227469:1227957 [6] NCCL INFO Channel 08/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227467:1227960 [4] NCCL INFO Channel 03/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227464:1227963 [1] NCCL INFO Channel 02/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227463:1227964 [0] NCCL INFO Channel 07/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227470:1227958 [7] NCCL INFO Channel 01/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227465:1227962 [2] NCCL INFO Channel 06/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227468:1227959 [5] NCCL INFO Channel 03/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227469:1227957 [6] NCCL INFO Channel 09/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227466:1227961 [3] NCCL INFO Channel 01/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227467:1227960 [4] NCCL INFO Channel 05/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227463:1227964 [0] NCCL INFO Channel 09/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227464:1227963 [1] NCCL INFO Channel 03/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227470:1227958 [7] NCCL INFO Channel 02/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227468:1227959 [5] NCCL INFO Channel 04/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227465:1227962 [2] NCCL INFO Channel 07/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227469:1227957 [6] NCCL INFO Channel 10/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227467:1227960 [4] NCCL INFO Channel 06/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227466:1227961 [3] NCCL INFO Channel 02/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227463:1227964 [0] NCCL INFO Channel 10/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227464:1227963 [1] NCCL INFO Channel 04/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227470:1227958 [7] NCCL INFO Channel 03/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227468:1227959 [5] NCCL INFO Channel 06/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227469:1227957 [6] NCCL INFO Channel 11/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227465:1227962 [2] NCCL INFO Channel 08/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227467:1227960 [4] NCCL INFO Channel 07/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227463:1227964 [0] NCCL INFO Channel 11/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227464:1227963 [1] NCCL INFO Channel 05/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227466:1227961 [3] NCCL INFO Channel 04/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227470:1227958 [7] NCCL INFO Channel 04/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227468:1227959 [5] NCCL INFO Channel 07/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227469:1227957 [6] NCCL INFO Channel 12/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227465:1227962 [2] NCCL INFO Channel 09/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227467:1227960 [4] NCCL INFO Channel 08/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227463:1227964 [0] NCCL INFO Channel 12/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227464:1227963 [1] NCCL INFO Channel 06/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227470:1227958 [7] NCCL INFO Channel 05/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227469:1227957 [6] NCCL INFO Channel 13/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227466:1227961 [3] NCCL INFO Channel 05/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227468:1227959 [5] NCCL INFO Channel 08/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227465:1227962 [2] NCCL INFO Channel 11/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227463:1227964 [0] NCCL INFO Channel 13/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227467:1227960 [4] NCCL INFO Channel 09/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227464:1227963 [1] NCCL INFO Channel 07/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227470:1227958 [7] NCCL INFO Channel 06/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227469:1227957 [6] NCCL INFO Channel 15/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227466:1227961 [3] NCCL INFO Channel 06/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227468:1227959 [5] NCCL INFO Channel 09/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227465:1227962 [2] NCCL INFO Channel 12/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227463:1227964 [0] NCCL INFO Channel 14/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227467:1227960 [4] NCCL INFO Channel 10/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227464:1227963 [1] NCCL INFO Channel 08/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227470:1227958 [7] NCCL INFO Channel 08/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227469:1227957 [6] NCCL INFO Channel 00/0 : 22[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227466:1227961 [3] NCCL INFO Channel 07/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227468:1227959 [5] NCCL INFO Channel 10/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227465:1227962 [2] NCCL INFO Channel 13/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227467:1227960 [4] NCCL INFO Channel 11/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227463:1227964 [0] NCCL INFO Channel 15/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227464:1227963 [1] NCCL INFO Channel 10/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227470:1227958 [7] NCCL INFO Channel 09/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227469:1227957 [6] NCCL INFO Channel 02/0 : 22[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227468:1227959 [5] NCCL INFO Channel 11/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227466:1227961 [3] NCCL INFO Channel 08/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227465:1227962 [2] NCCL INFO Channel 14/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227463:1227964 [0] NCCL INFO Channel 02/0 : 16[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227467:1227960 [4] NCCL INFO Channel 13/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227470:1227958 [7] NCCL INFO Channel 10/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227464:1227963 [1] NCCL INFO Channel 11/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227469:1227957 [6] NCCL INFO Channel 04/0 : 22[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227468:1227959 [5] NCCL INFO Channel 12/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227465:1227962 [2] NCCL INFO Channel 15/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227466:1227961 [3] NCCL INFO Channel 09/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227467:1227960 [4] NCCL INFO Channel 14/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227463:1227964 [0] NCCL INFO Channel 04/0 : 16[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227470:1227958 [7] NCCL INFO Channel 11/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227469:1227957 [6] NCCL INFO Channel 08/0 : 22[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227464:1227963 [1] NCCL INFO Channel 12/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227468:1227959 [5] NCCL INFO Channel 14/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227466:1227961 [3] NCCL INFO Channel 10/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227465:1227962 [2] NCCL INFO Channel 00/0 : 18[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227467:1227960 [4] NCCL INFO Channel 15/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227463:1227964 [0] NCCL INFO Channel 06/0 : 16[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227469:1227957 [6] NCCL INFO Channel 10/0 : 22[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227470:1227958 [7] NCCL INFO Channel 12/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227464:1227963 [1] NCCL INFO Channel 13/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227465:1227962 [2] NCCL INFO Channel 04/0 : 18[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227468:1227959 [5] NCCL INFO Channel 15/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227463:1227964 [0] NCCL INFO Channel 08/0 : 16[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227467:1227960 [4] NCCL INFO Channel 00/0 : 20[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227466:1227961 [3] NCCL INFO Channel 12/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227469:1227957 [6] NCCL INFO Channel 12/0 : 22[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227470:1227958 [7] NCCL INFO Channel 13/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227464:1227963 [1] NCCL INFO Channel 14/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227465:1227962 [2] NCCL INFO Channel 06/0 : 18[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227463:1227964 [0] NCCL INFO Channel 10/0 : 16[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227468:1227959 [5] NCCL INFO Channel 00/0 : 21[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227467:1227960 [4] NCCL INFO Channel 02/0 : 20[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227469:1227957 [6] NCCL INFO Channel 14/0 : 22[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227470:1227958 [7] NCCL INFO Channel 14/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227464:1227963 [1] NCCL INFO Channel 15/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227465:1227962 [2] NCCL INFO Channel 08/0 : 18[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227466:1227961 [3] NCCL INFO Channel 13/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227463:1227964 [0] NCCL INFO Channel 12/0 : 16[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227468:1227959 [5] NCCL INFO Channel 02/0 : 21[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227467:1227960 [4] NCCL INFO Channel 06/0 : 20[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227464:1227963 [1] NCCL INFO Channel 00/0 : 17[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227470:1227958 [7] NCCL INFO Channel 00/0 : 23[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227465:1227962 [2] NCCL INFO Channel 10/0 : 18[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227466:1227961 [3] NCCL INFO Channel 14/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227463:1227964 [0] NCCL INFO Channel 14/0 : 16[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227468:1227959 [5] NCCL INFO Channel 04/0 : 21[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227467:1227960 [4] NCCL INFO Channel 08/0 : 20[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227464:1227963 [1] NCCL INFO Channel 02/0 : 17[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227470:1227958 [7] NCCL INFO Channel 02/0 : 23[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227465:1227962 [2] NCCL INFO Channel 12/0 : 18[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227466:1227961 [3] NCCL INFO Channel 15/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227467:1227960 [4] NCCL INFO Channel 10/0 : 20[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227468:1227959 [5] NCCL INFO Channel 06/0 : 21[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227464:1227963 [1] NCCL INFO Channel 04/0 : 17[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227470:1227958 [7] NCCL INFO Channel 04/0 : 23[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227465:1227962 [2] NCCL INFO Channel 14/0 : 18[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227466:1227961 [3] NCCL INFO Channel 00/0 : 19[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227467:1227960 [4] NCCL INFO Channel 12/0 : 20[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227468:1227959 [5] NCCL INFO Channel 08/0 : 21[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227464:1227963 [1] NCCL INFO Channel 06/0 : 17[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227470:1227958 [7] NCCL INFO Channel 06/0 : 23[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153568:154056 [6] NCCL INFO Channel 01/0 : 14[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227466:1227961 [3] NCCL INFO Channel 02/0 : 19[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227467:1227960 [4] NCCL INFO Channel 14/0 : 20[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227464:1227963 [1] NCCL INFO Channel 08/0 : 17[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227470:1227958 [7] NCCL INFO Channel 08/0 : 23[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227468:1227959 [5] NCCL INFO Channel 10/0 : 21[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227466:1227961 [3] NCCL INFO Channel 04/0 : 19[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153568:154056 [6] NCCL INFO Channel 03/0 : 14[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227470:1227958 [7] NCCL INFO Channel 10/0 : 23[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153568:154056 [6] NCCL INFO Channel 05/0 : 14[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227464:1227963 [1] NCCL INFO Channel 10/0 : 17[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227468:1227959 [5] NCCL INFO Channel 12/0 : 21[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227466:1227961 [3] NCCL INFO Channel 06/0 : 19[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227470:1227958 [7] NCCL INFO Channel 12/0 : 23[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153562:154061 [0] NCCL INFO Channel 01/0 : 8[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227468:1227959 [5] NCCL INFO Channel 14/0 : 21[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227466:1227961 [3] NCCL INFO Channel 08/0 : 19[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153568:154056 [6] NCCL INFO Channel 07/0 : 14[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227464:1227963 [1] NCCL INFO Channel 12/0 : 17[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227470:1227958 [7] NCCL INFO Channel 14/0 : 23[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227466:1227961 [3] NCCL INFO Channel 10/0 : 19[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153562:154061 [0] NCCL INFO Channel 03/0 : 8[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227464:1227963 [1] NCCL INFO Channel 14/0 : 17[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153568:154056 [6] NCCL INFO Channel 09/0 : 14[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153562:154061 [0] NCCL INFO Channel 05/0 : 8[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227466:1227961 [3] NCCL INFO Channel 12/0 : 19[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153564:154057 [2] NCCL INFO Channel 01/0 : 10[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153568:154056 [6] NCCL INFO Channel 11/0 : 14[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227466:1227961 [3] NCCL INFO Channel 14/0 : 19[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153562:154061 [0] NCCL INFO Channel 07/0 : 8[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153566:154062 [4] NCCL INFO Channel 01/0 : 12[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153568:154056 [6] NCCL INFO Channel 13/0 : 14[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153564:154057 [2] NCCL INFO Channel 03/0 : 10[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153562:154061 [0] NCCL INFO Channel 09/0 : 8[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153568:154056 [6] NCCL INFO Channel 15/0 : 14[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153566:154062 [4] NCCL INFO Channel 03/0 : 12[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153564:154057 [2] NCCL INFO Channel 05/0 : 10[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153562:154061 [0] NCCL INFO Channel 11/0 : 8[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153568:154056 [6] NCCL INFO Channel 01/0 : 30[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153566:154062 [4] NCCL INFO Channel 05/0 : 12[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153564:154057 [2] NCCL INFO Channel 07/0 : 10[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153567:154059 [5] NCCL INFO Channel 01/0 : 13[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153568:154056 [6] NCCL INFO Channel 03/0 : 30[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153562:154061 [0] NCCL INFO Channel 13/0 : 8[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153566:154062 [4] NCCL INFO Channel 07/0 : 12[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153566:154062 [4] NCCL INFO Channel 09/0 : 12[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153566:154062 [4] NCCL INFO Channel 11/0 : 12[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153566:154062 [4] NCCL INFO Channel 13/0 : 12[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153566:154062 [4] NCCL INFO Channel 15/0 : 12[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153566:154062 [4] NCCL INFO Channel 01/0 : 28[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153566:154062 [4] NCCL INFO Channel 03/0 : 28[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153566:154062 [4] NCCL INFO Channel 05/0 : 28[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153566:154062 [4] NCCL INFO Channel 07/0 : 28[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153566:154062 [4] NCCL INFO Channel 09/0 : 28[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153566:154062 [4] NCCL INFO Channel 11/0 : 28[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153562:154061 [0] NCCL INFO Channel 15/0 : 8[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153566:154062 [4] NCCL INFO Channel 13/0 : 28[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153562:154061 [0] NCCL INFO Channel 01/0 : 24[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153566:154062 [4] NCCL INFO Channel 15/0 : 28[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153562:154061 [0] NCCL INFO Channel 03/0 : 24[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153562:154061 [0] NCCL INFO Channel 05/0 : 24[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153568:154056 [6] NCCL INFO Channel 05/0 : 30[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153562:154061 [0] NCCL INFO Channel 07/0 : 24[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153567:154059 [5] NCCL INFO Channel 03/0 : 13[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153562:154061 [0] NCCL INFO Channel 09/0 : 24[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153568:154056 [6] NCCL INFO Channel 07/0 : 30[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153562:154061 [0] NCCL INFO Channel 11/0 : 24[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153567:154059 [5] NCCL INFO Channel 05/0 : 13[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153568:154056 [6] NCCL INFO Channel 09/0 : 30[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153562:154061 [0] NCCL INFO Channel 13/0 : 24[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153568:154056 [6] NCCL INFO Channel 11/0 : 30[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153567:154059 [5] NCCL INFO Channel 07/0 : 13[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153562:154061 [0] NCCL INFO Channel 15/0 : 24[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153567:154059 [5] NCCL INFO Channel 09/0 : 13[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153568:154056 [6] NCCL INFO Channel 13/0 : 30[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153567:154059 [5] NCCL INFO Channel 11/0 : 13[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153568:154056 [6] NCCL INFO Channel 15/0 : 30[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153567:154059 [5] NCCL INFO Channel 15/0 : 13[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153567:154059 [5] NCCL INFO Channel 01/0 : 29[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153564:154057 [2] NCCL INFO Channel 09/0 : 10[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153567:154059 [5] NCCL INFO Channel 03/0 : 29[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153564:154057 [2] NCCL INFO Channel 11/0 : 10[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153567:154059 [5] NCCL INFO Channel 05/0 : 29[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153564:154057 [2] NCCL INFO Channel 13/0 : 10[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153567:154059 [5] NCCL INFO Channel 07/0 : 29[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153564:154057 [2] NCCL INFO Channel 15/0 : 10[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153567:154059 [5] NCCL INFO Channel 09/0 : 29[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153564:154057 [2] NCCL INFO Channel 01/0 : 26[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153563:154060 [1] NCCL INFO Channel 01/0 : 9[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153567:154059 [5] NCCL INFO Channel 11/0 : 29[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153569:154055 [7] NCCL INFO Channel 01/0 : 15[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153564:154057 [2] NCCL INFO Channel 03/0 : 26[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153567:154059 [5] NCCL INFO Channel 15/0 : 29[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153563:154060 [1] NCCL INFO Channel 03/0 : 9[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153569:154055 [7] NCCL INFO Channel 03/0 : 15[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153564:154057 [2] NCCL INFO Channel 05/0 : 26[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153565:154058 [3] NCCL INFO Channel 01/0 : 11[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153563:154060 [1] NCCL INFO Channel 05/0 : 9[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153569:154055 [7] NCCL INFO Channel 05/0 : 15[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153565:154058 [3] NCCL INFO Channel 03/0 : 11[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153564:154057 [2] NCCL INFO Channel 07/0 : 26[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153563:154060 [1] NCCL INFO Channel 07/0 : 9[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153569:154055 [7] NCCL INFO Channel 07/0 : 15[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153565:154058 [3] NCCL INFO Channel 05/0 : 11[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153563:154060 [1] NCCL INFO Channel 11/0 : 9[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153564:154057 [2] NCCL INFO Channel 09/0 : 26[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153569:154055 [7] NCCL INFO Channel 09/0 : 15[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153565:154058 [3] NCCL INFO Channel 07/0 : 11[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153564:154057 [2] NCCL INFO Channel 11/0 : 26[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153563:154060 [1] NCCL INFO Channel 13/0 : 9[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153569:154055 [7] NCCL INFO Channel 11/0 : 15[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153565:154058 [3] NCCL INFO Channel 09/0 : 11[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153564:154057 [2] NCCL INFO Channel 13/0 : 26[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153563:154060 [1] NCCL INFO Channel 15/0 : 9[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153565:154058 [3] NCCL INFO Channel 13/0 : 11[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153569:154055 [7] NCCL INFO Channel 13/0 : 15[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153564:154057 [2] NCCL INFO Channel 15/0 : 26[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153563:154060 [1] NCCL INFO Channel 01/0 : 25[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153565:154058 [3] NCCL INFO Channel 15/0 : 11[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153563:154060 [1] NCCL INFO Channel 03/0 : 25[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153569:154055 [7] NCCL INFO Channel 01/0 : 31[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153565:154058 [3] NCCL INFO Channel 01/0 : 27[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153563:154060 [1] NCCL INFO Channel 05/0 : 25[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153569:154055 [7] NCCL INFO Channel 03/0 : 31[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153565:154058 [3] NCCL INFO Channel 03/0 : 27[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153563:154060 [1] NCCL INFO Channel 07/0 : 25[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153565:154058 [3] NCCL INFO Channel 05/0 : 27[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153569:154055 [7] NCCL INFO Channel 05/0 : 31[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153563:154060 [1] NCCL INFO Channel 11/0 : 25[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153565:154058 [3] NCCL INFO Channel 07/0 : 27[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153569:154055 [7] NCCL INFO Channel 07/0 : 31[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153563:154060 [1] NCCL INFO Channel 13/0 : 25[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153565:154058 [3] NCCL INFO Channel 09/0 : 27[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153569:154055 [7] NCCL INFO Channel 09/0 : 31[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153563:154060 [1] NCCL INFO Channel 15/0 : 25[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153565:154058 [3] NCCL INFO Channel 13/0 : 27[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153569:154055 [7] NCCL INFO Channel 11/0 : 31[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153565:154058 [3] NCCL INFO Channel 15/0 : 27[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153569:154055 [7] NCCL INFO Channel 13/0 : 31[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96564:97088 [1] NCCL INFO Connected all trees
+ip-26-0-161-138:96564:97088 [1] NCCL INFO NVLS comm 0xadd6e40 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-138:96568:97091 [5] NCCL INFO Connected all trees
+ip-26-0-161-138:96568:97091 [5] NCCL INFO NVLS comm 0xa08c730 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-131:1630206:1630728 [1] NCCL INFO Channel 01/0 : 1[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630210:1630732 [5] NCCL INFO Channel 01/0 : 5[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630212:1630734 [7] NCCL INFO Channel 01/0 : 7[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630206:1630728 [1] NCCL INFO Channel 03/0 : 1[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630210:1630732 [5] NCCL INFO Channel 03/0 : 5[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630212:1630734 [7] NCCL INFO Channel 03/0 : 7[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630210:1630732 [5] NCCL INFO Channel 05/0 : 5[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630208:1630729 [3] NCCL INFO Channel 01/0 : 3[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630206:1630728 [1] NCCL INFO Channel 05/0 : 1[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630205:1630727 [0] NCCL INFO Channel 01/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630212:1630734 [7] NCCL INFO Channel 05/0 : 7[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630211:1630733 [6] NCCL INFO Channel 01/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630209:1630730 [4] NCCL INFO Channel 01/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630210:1630732 [5] NCCL INFO Channel 07/0 : 5[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630208:1630729 [3] NCCL INFO Channel 03/0 : 3[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630206:1630728 [1] NCCL INFO Channel 07/0 : 1[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630205:1630727 [0] NCCL INFO Channel 03/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630211:1630733 [6] NCCL INFO Channel 03/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630212:1630734 [7] NCCL INFO Channel 07/0 : 7[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630209:1630730 [4] NCCL INFO Channel 03/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630210:1630732 [5] NCCL INFO Channel 09/0 : 5[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630205:1630727 [0] NCCL INFO Channel 05/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630206:1630728 [1] NCCL INFO Channel 11/0 : 1[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630211:1630733 [6] NCCL INFO Channel 05/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630212:1630734 [7] NCCL INFO Channel 09/0 : 7[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630208:1630729 [3] NCCL INFO Channel 05/0 : 3[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630209:1630730 [4] NCCL INFO Channel 05/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630210:1630732 [5] NCCL INFO Channel 11/0 : 5[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630207:1630731 [2] NCCL INFO Channel 01/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630205:1630727 [0] NCCL INFO Channel 07/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630206:1630728 [1] NCCL INFO Channel 13/0 : 1[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630211:1630733 [6] NCCL INFO Channel 07/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630208:1630729 [3] NCCL INFO Channel 07/0 : 3[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630212:1630734 [7] NCCL INFO Channel 11/0 : 7[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630207:1630731 [2] NCCL INFO Channel 03/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630210:1630732 [5] NCCL INFO Channel 15/0 : 5[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630209:1630730 [4] NCCL INFO Channel 07/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630205:1630727 [0] NCCL INFO Channel 09/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630206:1630728 [1] NCCL INFO Channel 15/0 : 1[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630211:1630733 [6] NCCL INFO Channel 09/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630208:1630729 [3] NCCL INFO Channel 09/0 : 3[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630212:1630734 [7] NCCL INFO Channel 13/0 : 7[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630210:1630732 [5] NCCL INFO Channel 00/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630209:1630730 [4] NCCL INFO Channel 09/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630205:1630727 [0] NCCL INFO Channel 11/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630206:1630728 [1] NCCL INFO Channel 00/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630211:1630733 [6] NCCL INFO Channel 11/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630207:1630731 [2] NCCL INFO Channel 05/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630208:1630729 [3] NCCL INFO Channel 13/0 : 3[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630212:1630734 [7] NCCL INFO Channel 00/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630210:1630732 [5] NCCL INFO Channel 01/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630209:1630730 [4] NCCL INFO Channel 11/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630205:1630727 [0] NCCL INFO Channel 13/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630206:1630728 [1] NCCL INFO Channel 02/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630211:1630733 [6] NCCL INFO Channel 13/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630212:1630734 [7] NCCL INFO Channel 01/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630207:1630731 [2] NCCL INFO Channel 07/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630210:1630732 [5] NCCL INFO Channel 02/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630208:1630729 [3] NCCL INFO Channel 15/0 : 3[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630209:1630730 [4] NCCL INFO Channel 13/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630205:1630727 [0] NCCL INFO Channel 15/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630206:1630728 [1] NCCL INFO Channel 03/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630211:1630733 [6] NCCL INFO Channel 15/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630212:1630734 [7] NCCL INFO Channel 02/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630210:1630732 [5] NCCL INFO Channel 03/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630209:1630730 [4] NCCL INFO Channel 15/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630208:1630729 [3] NCCL INFO Channel 00/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630207:1630731 [2] NCCL INFO Channel 09/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630205:1630727 [0] NCCL INFO Channel 01/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630206:1630728 [1] NCCL INFO Channel 04/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630212:1630734 [7] NCCL INFO Channel 03/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630211:1630733 [6] NCCL INFO Channel 00/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630210:1630732 [5] NCCL INFO Channel 04/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630209:1630730 [4] NCCL INFO Channel 00/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630207:1630731 [2] NCCL INFO Channel 11/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630205:1630727 [0] NCCL INFO Channel 02/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630208:1630729 [3] NCCL INFO Channel 01/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630206:1630728 [1] NCCL INFO Channel 05/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630212:1630734 [7] NCCL INFO Channel 04/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630211:1630733 [6] NCCL INFO Channel 01/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630210:1630732 [5] NCCL INFO Channel 06/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630209:1630730 [4] NCCL INFO Channel 01/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630207:1630731 [2] NCCL INFO Channel 13/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630205:1630727 [0] NCCL INFO Channel 03/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630208:1630729 [3] NCCL INFO Channel 02/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630206:1630728 [1] NCCL INFO Channel 06/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630212:1630734 [7] NCCL INFO Channel 05/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630210:1630732 [5] NCCL INFO Channel 07/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630209:1630730 [4] NCCL INFO Channel 02/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630211:1630733 [6] NCCL INFO Channel 02/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630207:1630731 [2] NCCL INFO Channel 15/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630205:1630727 [0] NCCL INFO Channel 04/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630208:1630729 [3] NCCL INFO Channel 04/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630206:1630728 [1] NCCL INFO Channel 07/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630212:1630734 [7] NCCL INFO Channel 06/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630210:1630732 [5] NCCL INFO Channel 08/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630209:1630730 [4] NCCL INFO Channel 03/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630211:1630733 [6] NCCL INFO Channel 03/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630207:1630731 [2] NCCL INFO Channel 00/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630205:1630727 [0] NCCL INFO Channel 05/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630208:1630729 [3] NCCL INFO Channel 05/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630206:1630728 [1] NCCL INFO Channel 08/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630212:1630734 [7] NCCL INFO Channel 08/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630209:1630730 [4] NCCL INFO Channel 05/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630211:1630733 [6] NCCL INFO Channel 04/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630210:1630732 [5] NCCL INFO Channel 09/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630207:1630731 [2] NCCL INFO Channel 01/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630206:1630728 [1] NCCL INFO Channel 10/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630208:1630729 [3] NCCL INFO Channel 06/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630205:1630727 [0] NCCL INFO Channel 06/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630212:1630734 [7] NCCL INFO Channel 09/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630209:1630730 [4] NCCL INFO Channel 06/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630210:1630732 [5] NCCL INFO Channel 10/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630211:1630733 [6] NCCL INFO Channel 05/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630207:1630731 [2] NCCL INFO Channel 03/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630206:1630728 [1] NCCL INFO Channel 11/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630212:1630734 [7] NCCL INFO Channel 10/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630208:1630729 [3] NCCL INFO Channel 07/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630205:1630727 [0] NCCL INFO Channel 07/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630209:1630730 [4] NCCL INFO Channel 07/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630210:1630732 [5] NCCL INFO Channel 11/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630211:1630733 [6] NCCL INFO Channel 07/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630207:1630731 [2] NCCL INFO Channel 04/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630206:1630728 [1] NCCL INFO Channel 12/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630212:1630734 [7] NCCL INFO Channel 11/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630209:1630730 [4] NCCL INFO Channel 08/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630208:1630729 [3] NCCL INFO Channel 08/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630205:1630727 [0] NCCL INFO Channel 09/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630210:1630732 [5] NCCL INFO Channel 12/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630211:1630733 [6] NCCL INFO Channel 08/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630207:1630731 [2] NCCL INFO Channel 05/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630206:1630728 [1] NCCL INFO Channel 13/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630212:1630734 [7] NCCL INFO Channel 12/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630209:1630730 [4] NCCL INFO Channel 09/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630210:1630732 [5] NCCL INFO Channel 14/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630208:1630729 [3] NCCL INFO Channel 09/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630211:1630733 [6] NCCL INFO Channel 09/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630205:1630727 [0] NCCL INFO Channel 10/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630207:1630731 [2] NCCL INFO Channel 06/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630206:1630728 [1] NCCL INFO Channel 14/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630212:1630734 [7] NCCL INFO Channel 13/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630209:1630730 [4] NCCL INFO Channel 10/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630210:1630732 [5] NCCL INFO Channel 15/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630208:1630729 [3] NCCL INFO Channel 10/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630211:1630733 [6] NCCL INFO Channel 10/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630205:1630727 [0] NCCL INFO Channel 11/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630209:1630730 [4] NCCL INFO Channel 11/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630207:1630731 [2] NCCL INFO Channel 07/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630206:1630728 [1] NCCL INFO Channel 15/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630212:1630734 [7] NCCL INFO Channel 14/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630208:1630729 [3] NCCL INFO Channel 12/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630211:1630733 [6] NCCL INFO Channel 11/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96890:97386 [2] NCCL INFO Channel 00/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630209:1630730 [4] NCCL INFO Channel 13/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630207:1630731 [2] NCCL INFO Channel 08/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630205:1630727 [0] NCCL INFO Channel 12/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630211:1630733 [6] NCCL INFO Channel 12/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114517:115013 [5] NCCL INFO Channel 00/0 : 21[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630208:1630729 [3] NCCL INFO Channel 13/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630209:1630730 [4] NCCL INFO Channel 14/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96894:97381 [6] NCCL INFO Channel 00/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96890:97386 [2] NCCL INFO Channel 01/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630205:1630727 [0] NCCL INFO Channel 13/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114517:115013 [5] NCCL INFO Channel 02/0 : 21[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630207:1630731 [2] NCCL INFO Channel 09/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630211:1630733 [6] NCCL INFO Channel 13/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96890:97386 [2] NCCL INFO Channel 03/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96894:97381 [6] NCCL INFO Channel 01/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630209:1630730 [4] NCCL INFO Channel 15/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114513:115008 [1] NCCL INFO Channel 00/0 : 17[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114519:115014 [7] NCCL INFO Channel 00/0 : 23[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227468:1227959 [5] NCCL INFO Channel 00/0 : 5[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630208:1630729 [3] NCCL INFO Channel 14/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630205:1630727 [0] NCCL INFO Channel 14/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96893:97384 [5] NCCL INFO Channel 00/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114517:115013 [5] NCCL INFO Channel 04/0 : 21[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227468:1227959 [5] NCCL INFO Channel 02/0 : 5[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630210:1630732 [5] NCCL INFO Channel 01/0 : 29[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630211:1630733 [6] NCCL INFO Channel 15/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630207:1630731 [2] NCCL INFO Channel 11/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630208:1630729 [3] NCCL INFO Channel 15/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630205:1630727 [0] NCCL INFO Channel 15/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630206:1630728 [1] NCCL INFO Channel 01/0 : 25[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96890:97386 [2] NCCL INFO Channel 04/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96894:97381 [6] NCCL INFO Channel 02/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114513:115008 [1] NCCL INFO Channel 02/0 : 17[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227464:1227963 [1] NCCL INFO Channel 00/0 : 1[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114519:115014 [7] NCCL INFO Channel 02/0 : 23[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114517:115013 [5] NCCL INFO Channel 06/0 : 21[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227470:1227958 [7] NCCL INFO Channel 00/0 : 7[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630210:1630732 [5] NCCL INFO Channel 03/0 : 29[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630207:1630731 [2] NCCL INFO Channel 12/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96888:97387 [0] NCCL INFO Channel 01/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630212:1630734 [7] NCCL INFO Channel 01/0 : 31[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96893:97384 [5] NCCL INFO Channel 01/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630206:1630728 [1] NCCL INFO Channel 03/0 : 25[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630210:1630732 [5] NCCL INFO Channel 05/0 : 29[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114513:115008 [1] NCCL INFO Channel 04/0 : 17[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227468:1227959 [5] NCCL INFO Channel 04/0 : 5[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114517:115013 [5] NCCL INFO Channel 08/0 : 21[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114519:115014 [7] NCCL INFO Channel 04/0 : 23[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227464:1227963 [1] NCCL INFO Channel 02/0 : 1[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96890:97386 [2] NCCL INFO Channel 05/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96894:97381 [6] NCCL INFO Channel 03/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96892:97383 [4] NCCL INFO Channel 00/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630212:1630734 [7] NCCL INFO Channel 03/0 : 31[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630207:1630731 [2] NCCL INFO Channel 13/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630206:1630728 [1] NCCL INFO Channel 05/0 : 25[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630210:1630732 [5] NCCL INFO Channel 07/0 : 29[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114513:115008 [1] NCCL INFO Channel 06/0 : 17[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114516:115010 [4] NCCL INFO Channel 00/0 : 20[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114517:115013 [5] NCCL INFO Channel 10/0 : 21[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96895:97382 [7] NCCL INFO Channel 00/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227470:1227958 [7] NCCL INFO Channel 02/0 : 7[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96893:97384 [5] NCCL INFO Channel 02/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96888:97387 [0] NCCL INFO Channel 02/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227468:1227959 [5] NCCL INFO Channel 06/0 : 5[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630212:1630734 [7] NCCL INFO Channel 05/0 : 31[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96890:97386 [2] NCCL INFO Channel 06/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96891:97385 [3] NCCL INFO Channel 00/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227464:1227963 [1] NCCL INFO Channel 04/0 : 1[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227470:1227958 [7] NCCL INFO Channel 04/0 : 7[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630206:1630728 [1] NCCL INFO Channel 07/0 : 25[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630210:1630732 [5] NCCL INFO Channel 09/0 : 29[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114519:115014 [7] NCCL INFO Channel 06/0 : 23[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114515:115009 [3] NCCL INFO Channel 00/0 : 19[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114513:115008 [1] NCCL INFO Channel 08/0 : 17[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114518:115012 [6] NCCL INFO Channel 00/0 : 22[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96894:97381 [6] NCCL INFO Channel 04/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227468:1227959 [5] NCCL INFO Channel 08/0 : 5[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96889:97388 [1] NCCL INFO Channel 00/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96892:97383 [4] NCCL INFO Channel 01/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227464:1227963 [1] NCCL INFO Channel 06/0 : 1[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630207:1630731 [2] NCCL INFO Channel 14/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630209:1630730 [4] NCCL INFO Channel 01/0 : 28[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96895:97382 [7] NCCL INFO Channel 01/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227470:1227958 [7] NCCL INFO Channel 06/0 : 7[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630212:1630734 [7] NCCL INFO Channel 07/0 : 31[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227467:1227960 [4] NCCL INFO Channel 00/0 : 4[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114517:115013 [5] NCCL INFO Channel 12/0 : 21[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114519:115014 [7] NCCL INFO Channel 08/0 : 23[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114516:115010 [4] NCCL INFO Channel 02/0 : 20[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96890:97386 [2] NCCL INFO Channel 07/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630208:1630729 [3] NCCL INFO Channel 01/0 : 27[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114512:115007 [0] NCCL INFO Channel 02/0 : 16[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114513:115008 [1] NCCL INFO Channel 10/0 : 17[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114515:115009 [3] NCCL INFO Channel 02/0 : 19[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96888:97387 [0] NCCL INFO Channel 03/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96893:97384 [5] NCCL INFO Channel 03/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96891:97385 [3] NCCL INFO Channel 01/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227468:1227959 [5] NCCL INFO Channel 10/0 : 5[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630210:1630732 [5] NCCL INFO Channel 11/0 : 29[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114518:115012 [6] NCCL INFO Channel 02/0 : 22[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114517:115013 [5] NCCL INFO Channel 14/0 : 21[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96894:97381 [6] NCCL INFO Channel 05/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227464:1227963 [1] NCCL INFO Channel 08/0 : 1[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630206:1630728 [1] NCCL INFO Channel 11/0 : 25[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630205:1630727 [0] NCCL INFO Channel 01/0 : 24[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630209:1630730 [4] NCCL INFO Channel 03/0 : 28[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114519:115014 [7] NCCL INFO Channel 10/0 : 23[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114516:115010 [4] NCCL INFO Channel 06/0 : 20[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114513:115008 [1] NCCL INFO Channel 12/0 : 17[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96889:97388 [1] NCCL INFO Channel 02/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96892:97383 [4] NCCL INFO Channel 02/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96895:97382 [7] NCCL INFO Channel 02/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227463:1227964 [0] NCCL INFO Channel 02/0 : 0[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227469:1227957 [6] NCCL INFO Channel 00/0 : 6[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227470:1227958 [7] NCCL INFO Channel 08/0 : 7[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114512:115007 [0] NCCL INFO Channel 04/0 : 16[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96890:97386 [2] NCCL INFO Channel 08/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227467:1227960 [4] NCCL INFO Channel 02/0 : 4[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227468:1227959 [5] NCCL INFO Channel 12/0 : 5[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630211:1630733 [6] NCCL INFO Channel 01/0 : 30[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227466:1227961 [3] NCCL INFO Channel 00/0 : 3[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227464:1227963 [1] NCCL INFO Channel 10/0 : 1[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630212:1630734 [7] NCCL INFO Channel 09/0 : 31[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630207:1630731 [2] NCCL INFO Channel 15/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630208:1630729 [3] NCCL INFO Channel 03/0 : 27[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630210:1630732 [5] NCCL INFO Channel 15/0 : 29[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630206:1630728 [1] NCCL INFO Channel 13/0 : 25[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114517:115013 [5] NCCL INFO Channel 00/0 : 5[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114518:115012 [6] NCCL INFO Channel 04/0 : 22[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114519:115014 [7] NCCL INFO Channel 12/0 : 23[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96888:97387 [0] NCCL INFO Channel 04/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96891:97385 [3] NCCL INFO Channel 02/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114515:115009 [3] NCCL INFO Channel 04/0 : 19[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114516:115010 [4] NCCL INFO Channel 08/0 : 20[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96893:97384 [5] NCCL INFO Channel 04/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114513:115008 [1] NCCL INFO Channel 14/0 : 17[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96894:97381 [6] NCCL INFO Channel 07/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227463:1227964 [0] NCCL INFO Channel 04/0 : 0[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227470:1227958 [7] NCCL INFO Channel 10/0 : 7[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630205:1630727 [0] NCCL INFO Channel 03/0 : 24[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630209:1630730 [4] NCCL INFO Channel 05/0 : 28[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630211:1630733 [6] NCCL INFO Channel 03/0 : 30[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630212:1630734 [7] NCCL INFO Channel 11/0 : 31[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630210:1630732 [5] NCCL INFO Channel 01/0 : 13[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630206:1630728 [1] NCCL INFO Channel 15/0 : 25[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630208:1630729 [3] NCCL INFO Channel 05/0 : 27[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114512:115007 [0] NCCL INFO Channel 06/0 : 16[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96892:97383 [4] NCCL INFO Channel 03/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96890:97386 [2] NCCL INFO Channel 09/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96895:97382 [7] NCCL INFO Channel 03/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96889:97388 [1] NCCL INFO Channel 03/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114517:115013 [5] NCCL INFO Channel 02/0 : 5[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114519:115014 [7] NCCL INFO Channel 14/0 : 23[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96888:97387 [0] NCCL INFO Channel 05/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96891:97385 [3] NCCL INFO Channel 04/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96893:97384 [5] NCCL INFO Channel 06/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227469:1227957 [6] NCCL INFO Channel 02/0 : 6[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96894:97381 [6] NCCL INFO Channel 08/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227467:1227960 [4] NCCL INFO Channel 06/0 : 4[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227464:1227963 [1] NCCL INFO Channel 12/0 : 1[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227468:1227959 [5] NCCL INFO Channel 14/0 : 5[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630209:1630730 [4] NCCL INFO Channel 07/0 : 28[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227466:1227961 [3] NCCL INFO Channel 02/0 : 3[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630211:1630733 [6] NCCL INFO Channel 05/0 : 30[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630205:1630727 [0] NCCL INFO Channel 05/0 : 24[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114518:115012 [6] NCCL INFO Channel 08/0 : 22[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114513:115008 [1] NCCL INFO Channel 00/0 : 1[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630212:1630734 [7] NCCL INFO Channel 13/0 : 31[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114516:115010 [4] NCCL INFO Channel 10/0 : 20[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114515:115009 [3] NCCL INFO Channel 06/0 : 19[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96890:97386 [2] NCCL INFO Channel 11/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96892:97383 [4] NCCL INFO Channel 05/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96889:97388 [1] NCCL INFO Channel 04/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96895:97382 [7] NCCL INFO Channel 04/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227470:1227958 [7] NCCL INFO Channel 12/0 : 7[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227463:1227964 [0] NCCL INFO Channel 06/0 : 0[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630210:1630732 [5] NCCL INFO Channel 03/0 : 13[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630208:1630729 [3] NCCL INFO Channel 07/0 : 27[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630206:1630728 [1] NCCL INFO Channel 01/0 : 9[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114512:115007 [0] NCCL INFO Channel 08/0 : 16[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630209:1630730 [4] NCCL INFO Channel 09/0 : 28[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114517:115013 [5] NCCL INFO Channel 04/0 : 5[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630211:1630733 [6] NCCL INFO Channel 07/0 : 30[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114519:115014 [7] NCCL INFO Channel 00/0 : 7[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96888:97387 [0] NCCL INFO Channel 06/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114513:115008 [1] NCCL INFO Channel 02/0 : 1[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96891:97385 [3] NCCL INFO Channel 05/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96893:97384 [5] NCCL INFO Channel 07/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96894:97381 [6] NCCL INFO Channel 09/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96890:97386 [2] NCCL INFO Channel 12/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227467:1227960 [4] NCCL INFO Channel 08/0 : 4[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96892:97383 [4] NCCL INFO Channel 06/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227469:1227957 [6] NCCL INFO Channel 04/0 : 6[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96895:97382 [7] NCCL INFO Channel 05/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227468:1227959 [5] NCCL INFO Channel 00/0 : 21[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227464:1227963 [1] NCCL INFO Channel 14/0 : 1[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227470:1227958 [7] NCCL INFO Channel 14/0 : 7[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630205:1630727 [0] NCCL INFO Channel 07/0 : 24[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630212:1630734 [7] NCCL INFO Channel 01/0 : 15[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114514:115011 [2] NCCL INFO Channel 00/0 : 18[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114518:115012 [6] NCCL INFO Channel 10/0 : 22[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227466:1227961 [3] NCCL INFO Channel 04/0 : 3[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630210:1630732 [5] NCCL INFO Channel 05/0 : 13[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630207:1630731 [2] NCCL INFO Channel 01/0 : 26[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114516:115010 [4] NCCL INFO Channel 12/0 : 20[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114515:115009 [3] NCCL INFO Channel 08/0 : 19[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227463:1227964 [0] NCCL INFO Channel 08/0 : 0[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630206:1630728 [1] NCCL INFO Channel 03/0 : 9[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630208:1630729 [3] NCCL INFO Channel 09/0 : 27[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630209:1630730 [4] NCCL INFO Channel 11/0 : 28[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114512:115007 [0] NCCL INFO Channel 10/0 : 16[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630211:1630733 [6] NCCL INFO Channel 09/0 : 30[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114517:115013 [5] NCCL INFO Channel 06/0 : 5[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96889:97388 [1] NCCL INFO Channel 05/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96888:97387 [0] NCCL INFO Channel 07/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114519:115014 [7] NCCL INFO Channel 02/0 : 7[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96891:97385 [3] NCCL INFO Channel 06/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114513:115008 [1] NCCL INFO Channel 04/0 : 1[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96890:97386 [2] NCCL INFO Channel 13/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96894:97381 [6] NCCL INFO Channel 10/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96893:97384 [5] NCCL INFO Channel 08/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227465:1227962 [2] NCCL INFO Channel 00/0 : 2[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227467:1227960 [4] NCCL INFO Channel 10/0 : 4[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227469:1227957 [6] NCCL INFO Channel 08/0 : 6[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227468:1227959 [5] NCCL INFO Channel 02/0 : 21[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630205:1630727 [0] NCCL INFO Channel 09/0 : 24[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630212:1630734 [7] NCCL INFO Channel 03/0 : 15[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227470:1227958 [7] NCCL INFO Channel 00/0 : 23[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227464:1227963 [1] NCCL INFO Channel 00/0 : 17[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630210:1630732 [5] NCCL INFO Channel 07/0 : 13[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227466:1227961 [3] NCCL INFO Channel 06/0 : 3[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114515:115009 [3] NCCL INFO Channel 10/0 : 19[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114518:115012 [6] NCCL INFO Channel 12/0 : 22[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114514:115011 [2] NCCL INFO Channel 04/0 : 18[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96895:97382 [7] NCCL INFO Channel 06/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96892:97383 [4] NCCL INFO Channel 07/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114512:115007 [0] NCCL INFO Channel 12/0 : 16[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96888:97387 [0] NCCL INFO Channel 09/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96889:97388 [1] NCCL INFO Channel 06/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630206:1630728 [1] NCCL INFO Channel 05/0 : 9[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630209:1630730 [4] NCCL INFO Channel 13/0 : 28[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114517:115013 [5] NCCL INFO Channel 08/0 : 5[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114513:115008 [1] NCCL INFO Channel 06/0 : 1[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114519:115014 [7] NCCL INFO Channel 04/0 : 7[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96891:97385 [3] NCCL INFO Channel 07/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630207:1630731 [2] NCCL INFO Channel 03/0 : 26[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630211:1630733 [6] NCCL INFO Channel 11/0 : 30[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114516:115010 [4] NCCL INFO Channel 14/0 : 20[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96890:97386 [2] NCCL INFO Channel 14/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227463:1227964 [0] NCCL INFO Channel 10/0 : 0[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630208:1630729 [3] NCCL INFO Channel 13/0 : 27[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227465:1227962 [2] NCCL INFO Channel 04/0 : 2[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364906:1365407 [4] NCCL INFO Channel 01/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227467:1227960 [4] NCCL INFO Channel 12/0 : 4[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227469:1227957 [6] NCCL INFO Channel 10/0 : 6[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227468:1227959 [5] NCCL INFO Channel 04/0 : 21[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114515:115009 [3] NCCL INFO Channel 12/0 : 19[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227470:1227958 [7] NCCL INFO Channel 02/0 : 23[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227464:1227963 [1] NCCL INFO Channel 02/0 : 17[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96894:97381 [6] NCCL INFO Channel 11/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630205:1630727 [0] NCCL INFO Channel 11/0 : 24[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630212:1630734 [7] NCCL INFO Channel 05/0 : 15[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630210:1630732 [5] NCCL INFO Channel 09/0 : 13[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96893:97384 [5] NCCL INFO Channel 09/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630206:1630728 [1] NCCL INFO Channel 07/0 : 9[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96895:97382 [7] NCCL INFO Channel 08/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630209:1630730 [4] NCCL INFO Channel 15/0 : 28[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96892:97383 [4] NCCL INFO Channel 08/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114514:115011 [2] NCCL INFO Channel 06/0 : 18[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114518:115012 [6] NCCL INFO Channel 14/0 : 22[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114513:115008 [1] NCCL INFO Channel 08/0 : 1[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96888:97387 [0] NCCL INFO Channel 10/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96889:97388 [1] NCCL INFO Channel 07/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96891:97385 [3] NCCL INFO Channel 08/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114512:115007 [0] NCCL INFO Channel 14/0 : 16[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114517:115013 [5] NCCL INFO Channel 10/0 : 5[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96890:97386 [2] NCCL INFO Channel 15/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364906:1365407 [4] NCCL INFO Channel 03/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114519:115014 [7] NCCL INFO Channel 06/0 : 7[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114516:115010 [4] NCCL INFO Channel 00/0 : 4[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96894:97381 [6] NCCL INFO Channel 12/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114515:115009 [3] NCCL INFO Channel 14/0 : 19[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96893:97384 [5] NCCL INFO Channel 10/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630211:1630733 [6] NCCL INFO Channel 13/0 : 30[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630207:1630731 [2] NCCL INFO Channel 05/0 : 26[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630208:1630729 [3] NCCL INFO Channel 15/0 : 27[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630205:1630727 [0] NCCL INFO Channel 13/0 : 24[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630210:1630732 [5] NCCL INFO Channel 11/0 : 13[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630212:1630734 [7] NCCL INFO Channel 07/0 : 15[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630206:1630728 [1] NCCL INFO Channel 11/0 : 9[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630209:1630730 [4] NCCL INFO Channel 01/0 : 12[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364904:1365408 [2] NCCL INFO Channel 01/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114513:115008 [1] NCCL INFO Channel 10/0 : 1[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96895:97382 [7] NCCL INFO Channel 09/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96892:97383 [4] NCCL INFO Channel 09/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96888:97387 [0] NCCL INFO Channel 11/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96891:97385 [3] NCCL INFO Channel 09/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96889:97388 [1] NCCL INFO Channel 08/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96890:97386 [2] NCCL INFO Channel 00/0 : 18[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364906:1365407 [4] NCCL INFO Channel 05/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96894:97381 [6] NCCL INFO Channel 13/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364904:1365408 [2] NCCL INFO Channel 03/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114512:115007 [0] NCCL INFO Channel 02/0 : 0[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96893:97384 [5] NCCL INFO Channel 11/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630211:1630733 [6] NCCL INFO Channel 15/0 : 30[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114518:115012 [6] NCCL INFO Channel 00/0 : 6[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114516:115010 [4] NCCL INFO Channel 02/0 : 4[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96895:97382 [7] NCCL INFO Channel 10/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96892:97383 [4] NCCL INFO Channel 10/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114519:115014 [7] NCCL INFO Channel 08/0 : 7[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114517:115013 [5] NCCL INFO Channel 12/0 : 5[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114514:115011 [2] NCCL INFO Channel 08/0 : 18[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114515:115009 [3] NCCL INFO Channel 00/0 : 3[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96888:97387 [0] NCCL INFO Channel 12/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364909:1365405 [7] NCCL INFO Channel 01/0 : 7[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630207:1630731 [2] NCCL INFO Channel 07/0 : 26[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114513:115008 [1] NCCL INFO Channel 12/0 : 1[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630208:1630729 [3] NCCL INFO Channel 01/0 : 11[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96891:97385 [3] NCCL INFO Channel 10/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630210:1630732 [5] NCCL INFO Channel 15/0 : 13[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630205:1630727 [0] NCCL INFO Channel 15/0 : 24[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630212:1630734 [7] NCCL INFO Channel 09/0 : 15[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96890:97386 [2] NCCL INFO Channel 04/0 : 18[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630209:1630730 [4] NCCL INFO Channel 03/0 : 12[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630206:1630728 [1] NCCL INFO Channel 13/0 : 9[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630211:1630733 [6] NCCL INFO Channel 01/0 : 14[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96889:97388 [1] NCCL INFO Channel 10/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630207:1630731 [2] NCCL INFO Channel 09/0 : 26[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96894:97381 [6] NCCL INFO Channel 15/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96893:97384 [5] NCCL INFO Channel 12/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364906:1365407 [4] NCCL INFO Channel 07/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96888:97387 [0] NCCL INFO Channel 13/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96895:97382 [7] NCCL INFO Channel 11/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96892:97383 [4] NCCL INFO Channel 11/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96891:97385 [3] NCCL INFO Channel 12/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364905:1365404 [3] NCCL INFO Channel 01/0 : 3[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114512:115007 [0] NCCL INFO Channel 04/0 : 0[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364904:1365408 [2] NCCL INFO Channel 05/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364908:1365406 [6] NCCL INFO Channel 01/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364909:1365405 [7] NCCL INFO Channel 03/0 : 7[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630208:1630729 [3] NCCL INFO Channel 03/0 : 11[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96890:97386 [2] NCCL INFO Channel 06/0 : 18[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114518:115012 [6] NCCL INFO Channel 02/0 : 6[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96889:97388 [1] NCCL INFO Channel 11/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114515:115009 [3] NCCL INFO Channel 02/0 : 3[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114517:115013 [5] NCCL INFO Channel 14/0 : 5[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114516:115010 [4] NCCL INFO Channel 06/0 : 4[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114519:115014 [7] NCCL INFO Channel 10/0 : 7[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114513:115008 [1] NCCL INFO Channel 14/0 : 1[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96894:97381 [6] NCCL INFO Channel 00/0 : 22[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96893:97384 [5] NCCL INFO Channel 14/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114514:115011 [2] NCCL INFO Channel 10/0 : 18[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364906:1365407 [4] NCCL INFO Channel 09/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630205:1630727 [0] NCCL INFO Channel 01/0 : 8[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364905:1365404 [3] NCCL INFO Channel 03/0 : 3[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364904:1365408 [2] NCCL INFO Channel 07/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630212:1630734 [7] NCCL INFO Channel 11/0 : 15[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364902:1365410 [0] NCCL INFO Channel 01/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630209:1630730 [4] NCCL INFO Channel 05/0 : 12[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630211:1630733 [6] NCCL INFO Channel 03/0 : 14[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630206:1630728 [1] NCCL INFO Channel 15/0 : 9[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364908:1365406 [6] NCCL INFO Channel 03/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364909:1365405 [7] NCCL INFO Channel 05/0 : 7[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630207:1630731 [2] NCCL INFO Channel 11/0 : 26[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630208:1630729 [3] NCCL INFO Channel 05/0 : 11[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96888:97387 [0] NCCL INFO Channel 14/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114512:115007 [0] NCCL INFO Channel 06/0 : 0[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96895:97382 [7] NCCL INFO Channel 12/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114518:115012 [6] NCCL INFO Channel 04/0 : 6[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96890:97386 [2] NCCL INFO Channel 08/0 : 18[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96891:97385 [3] NCCL INFO Channel 13/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96892:97383 [4] NCCL INFO Channel 13/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96889:97388 [1] NCCL INFO Channel 12/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96894:97381 [6] NCCL INFO Channel 02/0 : 22[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96893:97384 [5] NCCL INFO Channel 15/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364903:1365409 [1] NCCL INFO Channel 01/0 : 1[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364906:1365407 [4] NCCL INFO Channel 11/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630205:1630727 [0] NCCL INFO Channel 03/0 : 8[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630212:1630734 [7] NCCL INFO Channel 13/0 : 15[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153567:154059 [5] NCCL INFO Channel 00/0 : 29[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114515:115009 [3] NCCL INFO Channel 04/0 : 3[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630209:1630730 [4] NCCL INFO Channel 07/0 : 12[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114519:115014 [7] NCCL INFO Channel 12/0 : 7[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114514:115011 [2] NCCL INFO Channel 12/0 : 18[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96888:97387 [0] NCCL INFO Channel 15/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114516:115010 [4] NCCL INFO Channel 08/0 : 4[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96895:97382 [7] NCCL INFO Channel 13/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364905:1365404 [3] NCCL INFO Channel 05/0 : 3[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114518:115012 [6] NCCL INFO Channel 08/0 : 6[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96890:97386 [2] NCCL INFO Channel 10/0 : 18[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364902:1365410 [0] NCCL INFO Channel 03/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114512:115007 [0] NCCL INFO Channel 08/0 : 0[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364904:1365408 [2] NCCL INFO Channel 09/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630211:1630733 [6] NCCL INFO Channel 05/0 : 14[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630210:1630732 [5] NCCL INFO Channel 00/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630208:1630729 [3] NCCL INFO Channel 07/0 : 11[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630207:1630731 [2] NCCL INFO Channel 13/0 : 26[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630205:1630727 [0] NCCL INFO Channel 05/0 : 8[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153567:154059 [5] NCCL INFO Channel 02/0 : 29[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364909:1365405 [7] NCCL INFO Channel 07/0 : 7[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364907:1365403 [5] NCCL INFO Channel 01/0 : 5[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630209:1630730 [4] NCCL INFO Channel 09/0 : 12[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364908:1365406 [6] NCCL INFO Channel 05/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96891:97385 [3] NCCL INFO Channel 14/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364903:1365409 [1] NCCL INFO Channel 03/0 : 1[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364906:1365407 [4] NCCL INFO Channel 13/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96892:97383 [4] NCCL INFO Channel 14/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114515:115009 [3] NCCL INFO Channel 06/0 : 3[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96893:97384 [5] NCCL INFO Channel 00/0 : 21[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96894:97381 [6] NCCL INFO Channel 04/0 : 22[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96888:97387 [0] NCCL INFO Channel 02/0 : 16[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630206:1630728 [1] NCCL INFO Channel 00/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153567:154059 [5] NCCL INFO Channel 04/0 : 29[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114519:115014 [7] NCCL INFO Channel 14/0 : 7[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96889:97388 [1] NCCL INFO Channel 13/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114514:115011 [2] NCCL INFO Channel 14/0 : 18[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96895:97382 [7] NCCL INFO Channel 14/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96890:97386 [2] NCCL INFO Channel 12/0 : 18[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364905:1365404 [3] NCCL INFO Channel 07/0 : 3[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114516:115010 [4] NCCL INFO Channel 10/0 : 4[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96891:97385 [3] NCCL INFO Channel 15/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364902:1365410 [0] NCCL INFO Channel 05/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114512:115007 [0] NCCL INFO Channel 10/0 : 0[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364904:1365408 [2] NCCL INFO Channel 11/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114518:115012 [6] NCCL INFO Channel 10/0 : 6[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364907:1365403 [5] NCCL INFO Channel 03/0 : 5[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364909:1365405 [7] NCCL INFO Channel 09/0 : 7[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364908:1365406 [6] NCCL INFO Channel 07/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630210:1630732 [5] NCCL INFO Channel 01/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153563:154060 [1] NCCL INFO Channel 00/0 : 25[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114515:115009 [3] NCCL INFO Channel 08/0 : 3[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630211:1630733 [6] NCCL INFO Channel 07/0 : 14[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630207:1630731 [2] NCCL INFO Channel 15/0 : 26[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630208:1630729 [3] NCCL INFO Channel 09/0 : 11[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153567:154059 [5] NCCL INFO Channel 06/0 : 29[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630205:1630727 [0] NCCL INFO Channel 07/0 : 8[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96892:97383 [4] NCCL INFO Channel 15/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630209:1630730 [4] NCCL INFO Channel 11/0 : 12[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96893:97384 [5] NCCL INFO Channel 02/0 : 21[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630206:1630728 [1] NCCL INFO Channel 02/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96888:97387 [0] NCCL INFO Channel 04/0 : 16[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96894:97381 [6] NCCL INFO Channel 08/0 : 22[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364903:1365409 [1] NCCL INFO Channel 05/0 : 1[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364906:1365407 [4] NCCL INFO Channel 15/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630212:1630734 [7] NCCL INFO Channel 00/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114514:115011 [2] NCCL INFO Channel 00/0 : 2[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364905:1365404 [3] NCCL INFO Channel 09/0 : 3[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153569:154055 [7] NCCL INFO Channel 00/0 : 31[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114512:115007 [0] NCCL INFO Channel 12/0 : 0[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114516:115010 [4] NCCL INFO Channel 12/0 : 4[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364902:1365410 [0] NCCL INFO Channel 07/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114518:115012 [6] NCCL INFO Channel 12/0 : 6[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364904:1365408 [2] NCCL INFO Channel 13/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114515:115009 [3] NCCL INFO Channel 10/0 : 3[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114517:115013 [5] NCCL INFO Channel 01/0 : 13[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96890:97386 [2] NCCL INFO Channel 14/0 : 18[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96889:97388 [1] NCCL INFO Channel 14/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96891:97385 [3] NCCL INFO Channel 00/0 : 19[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364907:1365403 [5] NCCL INFO Channel 05/0 : 5[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630210:1630732 [5] NCCL INFO Channel 02/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114514:115011 [2] NCCL INFO Channel 04/0 : 2[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96895:97382 [7] NCCL INFO Channel 00/0 : 23[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96892:97383 [4] NCCL INFO Channel 00/0 : 20[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630211:1630733 [6] NCCL INFO Channel 09/0 : 14[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153567:154059 [5] NCCL INFO Channel 08/0 : 29[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153563:154060 [1] NCCL INFO Channel 02/0 : 25[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96893:97384 [5] NCCL INFO Channel 04/0 : 21[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630208:1630729 [3] NCCL INFO Channel 13/0 : 11[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630207:1630731 [2] NCCL INFO Channel 01/0 : 10[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630205:1630727 [0] NCCL INFO Channel 09/0 : 8[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153569:154055 [7] NCCL INFO Channel 02/0 : 31[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96888:97387 [0] NCCL INFO Channel 06/0 : 16[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96894:97381 [6] NCCL INFO Channel 10/0 : 22[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630209:1630730 [4] NCCL INFO Channel 13/0 : 12[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364908:1365406 [6] NCCL INFO Channel 09/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364909:1365405 [7] NCCL INFO Channel 11/0 : 7[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114516:115010 [4] NCCL INFO Channel 14/0 : 4[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364906:1365407 [4] NCCL INFO Channel 00/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364905:1365404 [3] NCCL INFO Channel 13/0 : 3[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364902:1365410 [0] NCCL INFO Channel 09/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114512:115007 [0] NCCL INFO Channel 14/0 : 0[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114518:115012 [6] NCCL INFO Channel 14/0 : 6[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96891:97385 [3] NCCL INFO Channel 02/0 : 19[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96895:97382 [7] NCCL INFO Channel 02/0 : 23[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364904:1365408 [2] NCCL INFO Channel 15/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153567:154059 [5] NCCL INFO Channel 10/0 : 29[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153563:154060 [1] NCCL INFO Channel 04/0 : 25[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114513:115008 [1] NCCL INFO Channel 01/0 : 9[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114515:115009 [3] NCCL INFO Channel 12/0 : 3[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114517:115013 [5] NCCL INFO Channel 03/0 : 13[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96889:97388 [1] NCCL INFO Channel 15/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96892:97383 [4] NCCL INFO Channel 02/0 : 20[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364903:1365409 [1] NCCL INFO Channel 07/0 : 1[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364907:1365403 [5] NCCL INFO Channel 07/0 : 5[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630212:1630734 [7] NCCL INFO Channel 01/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630206:1630728 [1] NCCL INFO Channel 03/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630210:1630732 [5] NCCL INFO Channel 03/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153569:154055 [7] NCCL INFO Channel 04/0 : 31[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114514:115011 [2] NCCL INFO Channel 06/0 : 2[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114519:115014 [7] NCCL INFO Channel 01/0 : 15[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96893:97384 [5] NCCL INFO Channel 06/0 : 21[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364909:1365405 [7] NCCL INFO Channel 13/0 : 7[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364908:1365406 [6] NCCL INFO Channel 11/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630211:1630733 [6] NCCL INFO Channel 11/0 : 14[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96888:97387 [0] NCCL INFO Channel 08/0 : 16[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96894:97381 [6] NCCL INFO Channel 12/0 : 22[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153567:154059 [5] NCCL INFO Channel 12/0 : 29[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364906:1365407 [4] NCCL INFO Channel 01/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114515:115009 [3] NCCL INFO Channel 14/0 : 3[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114514:115011 [2] NCCL INFO Channel 08/0 : 2[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96891:97385 [3] NCCL INFO Channel 04/0 : 19[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96895:97382 [7] NCCL INFO Channel 04/0 : 23[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364902:1365410 [0] NCCL INFO Channel 11/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630208:1630729 [3] NCCL INFO Channel 15/0 : 11[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630207:1630731 [2] NCCL INFO Channel 03/0 : 10[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364904:1365408 [2] NCCL INFO Channel 00/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364905:1365404 [3] NCCL INFO Channel 15/0 : 3[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630205:1630727 [0] NCCL INFO Channel 11/0 : 8[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153563:154060 [1] NCCL INFO Channel 06/0 : 25[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364907:1365403 [5] NCCL INFO Channel 09/0 : 5[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630209:1630730 [4] NCCL INFO Channel 15/0 : 12[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114517:115013 [5] NCCL INFO Channel 05/0 : 13[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114519:115014 [7] NCCL INFO Channel 03/0 : 15[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630206:1630728 [1] NCCL INFO Channel 04/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630211:1630733 [6] NCCL INFO Channel 13/0 : 14[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630212:1630734 [7] NCCL INFO Channel 02/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630210:1630732 [5] NCCL INFO Channel 04/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114513:115008 [1] NCCL INFO Channel 03/0 : 9[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96892:97383 [4] NCCL INFO Channel 06/0 : 20[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630207:1630731 [2] NCCL INFO Channel 05/0 : 10[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96893:97384 [5] NCCL INFO Channel 08/0 : 21[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630205:1630727 [0] NCCL INFO Channel 13/0 : 8[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96888:97387 [0] NCCL INFO Channel 10/0 : 16[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96889:97388 [1] NCCL INFO Channel 00/0 : 17[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96894:97381 [6] NCCL INFO Channel 14/0 : 22[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153567:154059 [5] NCCL INFO Channel 14/0 : 29[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96891:97385 [3] NCCL INFO Channel 06/0 : 19[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96895:97382 [7] NCCL INFO Channel 06/0 : 23[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364909:1365405 [7] NCCL INFO Channel 00/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364908:1365406 [6] NCCL INFO Channel 13/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114514:115011 [2] NCCL INFO Channel 10/0 : 2[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364903:1365409 [1] NCCL INFO Channel 11/0 : 1[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364906:1365407 [4] NCCL INFO Channel 02/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364902:1365410 [0] NCCL INFO Channel 13/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364904:1365408 [2] NCCL INFO Channel 01/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227463:1227964 [0] NCCL INFO Channel 12/0 : 0[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227467:1227960 [4] NCCL INFO Channel 14/0 : 4[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153563:154060 [1] NCCL INFO Channel 08/0 : 25[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364905:1365404 [3] NCCL INFO Channel 00/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227466:1227961 [3] NCCL INFO Channel 08/0 : 3[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630211:1630733 [6] NCCL INFO Channel 15/0 : 14[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630206:1630728 [1] NCCL INFO Channel 05/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96892:97383 [4] NCCL INFO Channel 08/0 : 20[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364907:1365403 [5] NCCL INFO Channel 11/0 : 5[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227465:1227962 [2] NCCL INFO Channel 06/0 : 2[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630212:1630734 [7] NCCL INFO Channel 03/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630210:1630732 [5] NCCL INFO Channel 06/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96893:97384 [5] NCCL INFO Channel 10/0 : 21[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96888:97387 [0] NCCL INFO Channel 12/0 : 16[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364909:1365405 [7] NCCL INFO Channel 01/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227468:1227959 [5] NCCL INFO Channel 06/0 : 21[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227469:1227957 [6] NCCL INFO Channel 12/0 : 6[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630205:1630727 [0] NCCL INFO Channel 15/0 : 8[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96889:97388 [1] NCCL INFO Channel 02/0 : 17[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227470:1227958 [7] NCCL INFO Channel 04/0 : 23[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630208:1630729 [3] NCCL INFO Channel 00/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630207:1630731 [2] NCCL INFO Channel 07/0 : 10[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114517:115013 [5] NCCL INFO Channel 07/0 : 13[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227464:1227963 [1] NCCL INFO Channel 04/0 : 17[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630209:1630730 [4] NCCL INFO Channel 00/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153563:154060 [1] NCCL INFO Channel 10/0 : 25[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114519:115014 [7] NCCL INFO Channel 05/0 : 15[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364908:1365406 [6] NCCL INFO Channel 15/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227467:1227960 [4] NCCL INFO Channel 00/0 : 20[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153565:154058 [3] NCCL INFO Channel 00/0 : 27[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114513:115008 [1] NCCL INFO Channel 05/0 : 9[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114512:115007 [0] NCCL INFO Channel 01/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114518:115012 [6] NCCL INFO Channel 01/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96891:97385 [3] NCCL INFO Channel 08/0 : 19[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96895:97382 [7] NCCL INFO Channel 08/0 : 23[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364902:1365410 [0] NCCL INFO Channel 15/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364906:1365407 [4] NCCL INFO Channel 03/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227463:1227964 [0] NCCL INFO Channel 14/0 : 0[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114514:115011 [2] NCCL INFO Channel 12/0 : 2[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96892:97383 [4] NCCL INFO Channel 10/0 : 20[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364904:1365408 [2] NCCL INFO Channel 03/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227466:1227961 [3] NCCL INFO Channel 10/0 : 3[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96888:97387 [0] NCCL INFO Channel 14/0 : 16[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96893:97384 [5] NCCL INFO Channel 12/0 : 21[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96889:97388 [1] NCCL INFO Channel 04/0 : 17[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364905:1365404 [3] NCCL INFO Channel 01/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364903:1365409 [1] NCCL INFO Channel 13/0 : 1[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227465:1227962 [2] NCCL INFO Channel 08/0 : 2[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364909:1365405 [7] NCCL INFO Channel 02/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227469:1227957 [6] NCCL INFO Channel 14/0 : 6[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630206:1630728 [1] NCCL INFO Channel 06/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153563:154060 [1] NCCL INFO Channel 12/0 : 25[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153566:154062 [4] NCCL INFO Channel 00/0 : 28[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364907:1365403 [5] NCCL INFO Channel 15/0 : 5[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364908:1365406 [6] NCCL INFO Channel 00/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227468:1227959 [5] NCCL INFO Channel 08/0 : 21[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630210:1630732 [5] NCCL INFO Channel 07/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630212:1630734 [7] NCCL INFO Channel 04/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153565:154058 [3] NCCL INFO Channel 02/0 : 27[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227470:1227958 [7] NCCL INFO Channel 06/0 : 23[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630208:1630729 [3] NCCL INFO Channel 01/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114517:115013 [5] NCCL INFO Channel 09/0 : 13[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32706:33199 [2] NCCL INFO Channel 01/0 : 10[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227464:1227963 [1] NCCL INFO Channel 06/0 : 17[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630207:1630731 [2] NCCL INFO Channel 09/0 : 10[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630209:1630730 [4] NCCL INFO Channel 01/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114519:115014 [7] NCCL INFO Channel 07/0 : 15[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96895:97382 [7] NCCL INFO Channel 10/0 : 23[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227467:1227960 [4] NCCL INFO Channel 02/0 : 20[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630206:1630728 [1] NCCL INFO Channel 07/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114512:115007 [0] NCCL INFO Channel 03/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96892:97383 [4] NCCL INFO Channel 12/0 : 20[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364906:1365407 [4] NCCL INFO Channel 05/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227463:1227964 [0] NCCL INFO Channel 02/0 : 16[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630212:1630734 [7] NCCL INFO Channel 05/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630210:1630732 [5] NCCL INFO Channel 08/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114513:115008 [1] NCCL INFO Channel 07/0 : 9[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114518:115012 [6] NCCL INFO Channel 03/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96891:97385 [3] NCCL INFO Channel 10/0 : 19[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364902:1365410 [0] NCCL INFO Channel 01/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227466:1227961 [3] NCCL INFO Channel 12/0 : 3[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227465:1227962 [2] NCCL INFO Channel 10/0 : 2[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153566:154062 [4] NCCL INFO Channel 02/0 : 28[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153563:154060 [1] NCCL INFO Channel 14/0 : 25[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114514:115011 [2] NCCL INFO Channel 14/0 : 2[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96893:97384 [5] NCCL INFO Channel 14/0 : 21[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364904:1365408 [2] NCCL INFO Channel 04/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227468:1227959 [5] NCCL INFO Channel 10/0 : 21[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227470:1227958 [7] NCCL INFO Channel 08/0 : 23[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227469:1227957 [6] NCCL INFO Channel 00/0 : 22[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114516:115010 [4] NCCL INFO Channel 01/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114517:115013 [5] NCCL INFO Channel 11/0 : 13[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96889:97388 [1] NCCL INFO Channel 06/0 : 17[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364905:1365404 [3] NCCL INFO Channel 02/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227464:1227963 [1] NCCL INFO Channel 08/0 : 17[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227467:1227960 [4] NCCL INFO Channel 06/0 : 20[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114519:115014 [7] NCCL INFO Channel 09/0 : 15[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364903:1365409 [1] NCCL INFO Channel 15/0 : 1[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227463:1227964 [0] NCCL INFO Channel 04/0 : 16[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227466:1227961 [3] NCCL INFO Channel 14/0 : 3[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227465:1227962 [2] NCCL INFO Channel 12/0 : 2[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630211:1630733 [6] NCCL INFO Channel 00/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364908:1365406 [6] NCCL INFO Channel 01/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364909:1365405 [7] NCCL INFO Channel 03/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364907:1365403 [5] NCCL INFO Channel 00/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32706:33199 [2] NCCL INFO Channel 03/0 : 10[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227468:1227959 [5] NCCL INFO Channel 12/0 : 21[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630207:1630731 [2] NCCL INFO Channel 11/0 : 10[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630205:1630727 [0] NCCL INFO Channel 01/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630208:1630729 [3] NCCL INFO Channel 02/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153565:154058 [3] NCCL INFO Channel 04/0 : 27[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364906:1365407 [4] NCCL INFO Channel 06/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227469:1227957 [6] NCCL INFO Channel 02/0 : 22[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227470:1227958 [7] NCCL INFO Channel 10/0 : 23[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630209:1630730 [4] NCCL INFO Channel 02/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630206:1630728 [1] NCCL INFO Channel 08/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153568:154056 [6] NCCL INFO Channel 00/0 : 30[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153562:154061 [0] NCCL INFO Channel 02/0 : 24[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96895:97382 [7] NCCL INFO Channel 12/0 : 23[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227464:1227963 [1] NCCL INFO Channel 10/0 : 17[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153566:154062 [4] NCCL INFO Channel 06/0 : 28[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114512:115007 [0] NCCL INFO Channel 05/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96892:97383 [4] NCCL INFO Channel 14/0 : 20[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227463:1227964 [0] NCCL INFO Channel 06/0 : 16[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227465:1227962 [2] NCCL INFO Channel 14/0 : 2[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153565:154058 [3] NCCL INFO Channel 06/0 : 27[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114513:115008 [1] NCCL INFO Channel 11/0 : 9[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96891:97385 [3] NCCL INFO Channel 12/0 : 19[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96889:97388 [1] NCCL INFO Channel 08/0 : 17[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227468:1227959 [5] NCCL INFO Channel 14/0 : 21[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227466:1227961 [3] NCCL INFO Channel 00/0 : 19[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114518:115012 [6] NCCL INFO Channel 05/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96895:97382 [7] NCCL INFO Channel 14/0 : 23[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364902:1365410 [0] NCCL INFO Channel 02/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227470:1227958 [7] NCCL INFO Channel 12/0 : 23[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227469:1227957 [6] NCCL INFO Channel 04/0 : 22[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227464:1227963 [1] NCCL INFO Channel 12/0 : 17[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114515:115009 [3] NCCL INFO Channel 01/0 : 11[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114519:115014 [7] NCCL INFO Channel 11/0 : 15[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114517:115013 [5] NCCL INFO Channel 15/0 : 13[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364904:1365408 [2] NCCL INFO Channel 05/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227463:1227964 [0] NCCL INFO Channel 08/0 : 16[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630210:1630732 [5] NCCL INFO Channel 09/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630212:1630734 [7] NCCL INFO Channel 06/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114516:115010 [4] NCCL INFO Channel 03/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364905:1365404 [3] NCCL INFO Channel 04/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227465:1227962 [2] NCCL INFO Channel 00/0 : 18[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227466:1227961 [3] NCCL INFO Channel 02/0 : 19[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630211:1630733 [6] NCCL INFO Channel 01/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114512:115007 [0] NCCL INFO Channel 07/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364903:1365409 [1] NCCL INFO Channel 00/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32706:33199 [2] NCCL INFO Channel 05/0 : 10[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227470:1227958 [7] NCCL INFO Channel 14/0 : 23[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153568:154056 [6] NCCL INFO Channel 02/0 : 30[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153562:154061 [0] NCCL INFO Channel 04/0 : 24[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114513:115008 [1] NCCL INFO Channel 13/0 : 9[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96889:97388 [1] NCCL INFO Channel 10/0 : 17[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364908:1365406 [6] NCCL INFO Channel 02/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227469:1227957 [6] NCCL INFO Channel 08/0 : 22[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153566:154062 [4] NCCL INFO Channel 08/0 : 28[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114518:115012 [6] NCCL INFO Channel 07/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96891:97385 [3] NCCL INFO Channel 14/0 : 19[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364907:1365403 [5] NCCL INFO Channel 01/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364909:1365405 [7] NCCL INFO Channel 04/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227464:1227963 [1] NCCL INFO Channel 14/0 : 17[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630205:1630727 [0] NCCL INFO Channel 02/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630207:1630731 [2] NCCL INFO Channel 13/0 : 10[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153565:154058 [3] NCCL INFO Channel 08/0 : 27[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96889:97388 [1] NCCL INFO Channel 12/0 : 17[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364906:1365407 [4] NCCL INFO Channel 07/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227463:1227964 [0] NCCL INFO Channel 10/0 : 16[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630208:1630729 [3] NCCL INFO Channel 04/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630209:1630730 [4] NCCL INFO Channel 03/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364902:1365410 [0] NCCL INFO Channel 03/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364904:1365408 [2] NCCL INFO Channel 06/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227465:1227962 [2] NCCL INFO Channel 04/0 : 18[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227466:1227961 [3] NCCL INFO Channel 04/0 : 19[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630206:1630728 [1] NCCL INFO Channel 10/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364905:1365404 [3] NCCL INFO Channel 05/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227469:1227957 [6] NCCL INFO Channel 10/0 : 22[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630210:1630732 [5] NCCL INFO Channel 10/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364903:1365409 [1] NCCL INFO Channel 02/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227463:1227964 [0] NCCL INFO Channel 12/0 : 16[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227467:1227960 [4] NCCL INFO Channel 08/0 : 20[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227468:1227959 [5] NCCL INFO Channel 00/0 : 29[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630211:1630733 [6] NCCL INFO Channel 02/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630212:1630734 [7] NCCL INFO Channel 08/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114515:115009 [3] NCCL INFO Channel 03/0 : 11[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96889:97388 [1] NCCL INFO Channel 14/0 : 17[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32710:33202 [6] NCCL INFO Channel 01/0 : 14[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32706:33199 [2] NCCL INFO Channel 07/0 : 10[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227466:1227961 [3] NCCL INFO Channel 06/0 : 19[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153562:154061 [0] NCCL INFO Channel 06/0 : 24[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153568:154056 [6] NCCL INFO Channel 04/0 : 30[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114519:115014 [7] NCCL INFO Channel 13/0 : 15[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114514:115011 [2] NCCL INFO Channel 01/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114512:115007 [0] NCCL INFO Channel 09/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227465:1227962 [2] NCCL INFO Channel 06/0 : 18[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227469:1227957 [6] NCCL INFO Channel 12/0 : 22[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153566:154062 [4] NCCL INFO Channel 10/0 : 28[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153565:154058 [3] NCCL INFO Channel 10/0 : 27[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114513:115008 [1] NCCL INFO Channel 15/0 : 9[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114516:115010 [4] NCCL INFO Channel 05/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227463:1227964 [0] NCCL INFO Channel 14/0 : 16[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227467:1227960 [4] NCCL INFO Channel 10/0 : 20[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227468:1227959 [5] NCCL INFO Channel 02/0 : 29[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153562:154061 [0] NCCL INFO Channel 08/0 : 24[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114518:115012 [6] NCCL INFO Channel 09/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364908:1365406 [6] NCCL INFO Channel 03/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227466:1227961 [3] NCCL INFO Channel 08/0 : 19[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630205:1630727 [0] NCCL INFO Channel 03/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114515:115009 [3] NCCL INFO Channel 05/0 : 11[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364909:1365405 [7] NCCL INFO Channel 05/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227469:1227957 [6] NCCL INFO Channel 14/0 : 22[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630207:1630731 [2] NCCL INFO Channel 15/0 : 10[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364907:1365403 [5] NCCL INFO Channel 02/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32706:33199 [2] NCCL INFO Channel 09/0 : 10[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32710:33202 [6] NCCL INFO Channel 03/0 : 14[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227465:1227962 [2] NCCL INFO Channel 08/0 : 18[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630209:1630730 [4] NCCL INFO Channel 05/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630208:1630729 [3] NCCL INFO Channel 05/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364902:1365410 [0] NCCL INFO Channel 04/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364906:1365407 [4] NCCL INFO Channel 08/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227464:1227963 [1] NCCL INFO Channel 00/0 : 25[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227470:1227958 [7] NCCL INFO Channel 00/0 : 31[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153566:154062 [4] NCCL INFO Channel 12/0 : 28[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153568:154056 [6] NCCL INFO Channel 08/0 : 30[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153565:154058 [3] NCCL INFO Channel 12/0 : 27[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364904:1365408 [2] NCCL INFO Channel 07/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364903:1365409 [1] NCCL INFO Channel 03/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227467:1227960 [4] NCCL INFO Channel 12/0 : 20[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227468:1227959 [5] NCCL INFO Channel 04/0 : 29[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227466:1227961 [3] NCCL INFO Channel 10/0 : 19[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153562:154061 [0] NCCL INFO Channel 10/0 : 24[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114514:115011 [2] NCCL INFO Channel 03/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364905:1365404 [3] NCCL INFO Channel 06/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227465:1227962 [2] NCCL INFO Channel 10/0 : 18[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114512:115007 [0] NCCL INFO Channel 11/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364908:1365406 [6] NCCL INFO Channel 04/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32704:33197 [0] NCCL INFO Channel 01/0 : 8[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32709:33201 [5] NCCL INFO Channel 01/0 : 13[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227470:1227958 [7] NCCL INFO Channel 02/0 : 31[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114516:115010 [4] NCCL INFO Channel 07/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114518:115012 [6] NCCL INFO Channel 11/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114515:115009 [3] NCCL INFO Channel 07/0 : 11[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364907:1365403 [5] NCCL INFO Channel 03/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364909:1365405 [7] NCCL INFO Channel 06/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227464:1227963 [1] NCCL INFO Channel 02/0 : 25[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630206:1630728 [1] NCCL INFO Channel 11/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630210:1630732 [5] NCCL INFO Channel 11/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630211:1630733 [6] NCCL INFO Channel 03/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630205:1630727 [0] NCCL INFO Channel 04/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227467:1227960 [4] NCCL INFO Channel 14/0 : 20[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227466:1227961 [3] NCCL INFO Channel 12/0 : 19[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227468:1227959 [5] NCCL INFO Channel 06/0 : 29[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630212:1630734 [7] NCCL INFO Channel 09/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153566:154062 [4] NCCL INFO Channel 14/0 : 28[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32706:33199 [2] NCCL INFO Channel 11/0 : 10[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227465:1227962 [2] NCCL INFO Channel 12/0 : 18[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630209:1630730 [4] NCCL INFO Channel 06/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153568:154056 [6] NCCL INFO Channel 10/0 : 30[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153565:154058 [3] NCCL INFO Channel 14/0 : 27[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32710:33202 [6] NCCL INFO Channel 05/0 : 14[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227470:1227958 [7] NCCL INFO Channel 04/0 : 31[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153562:154061 [0] NCCL INFO Channel 12/0 : 24[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114514:115011 [2] NCCL INFO Channel 05/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32708:33198 [4] NCCL INFO Channel 01/0 : 12[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227464:1227963 [1] NCCL INFO Channel 04/0 : 25[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114512:115007 [0] NCCL INFO Channel 13/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364902:1365410 [0] NCCL INFO Channel 05/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227466:1227961 [3] NCCL INFO Channel 14/0 : 19[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114516:115010 [4] NCCL INFO Channel 09/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114518:115012 [6] NCCL INFO Channel 13/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364903:1365409 [1] NCCL INFO Channel 04/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364904:1365408 [2] NCCL INFO Channel 08/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227468:1227959 [5] NCCL INFO Channel 08/0 : 29[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114515:115009 [3] NCCL INFO Channel 09/0 : 11[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364906:1365407 [4] NCCL INFO Channel 09/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227465:1227962 [2] NCCL INFO Channel 14/0 : 18[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227469:1227957 [6] NCCL INFO Channel 00/0 : 30[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630206:1630728 [1] NCCL INFO Channel 12/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364905:1365404 [3] NCCL INFO Channel 07/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364908:1365406 [6] NCCL INFO Channel 05/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227470:1227958 [7] NCCL INFO Channel 06/0 : 31[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227463:1227964 [0] NCCL INFO Channel 02/0 : 24[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630210:1630732 [5] NCCL INFO Channel 12/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630211:1630733 [6] NCCL INFO Channel 04/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153568:154056 [6] NCCL INFO Channel 12/0 : 30[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364907:1365403 [5] NCCL INFO Channel 04/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364909:1365405 [7] NCCL INFO Channel 08/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32709:33201 [5] NCCL INFO Channel 03/0 : 13[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32704:33197 [0] NCCL INFO Channel 03/0 : 8[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227464:1227963 [1] NCCL INFO Channel 06/0 : 25[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630205:1630727 [0] NCCL INFO Channel 05/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630212:1630734 [7] NCCL INFO Channel 10/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153562:154061 [0] NCCL INFO Channel 14/0 : 24[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364902:1365410 [0] NCCL INFO Channel 06/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32706:33199 [2] NCCL INFO Channel 13/0 : 10[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32711:33204 [7] NCCL INFO Channel 01/0 : 15[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630208:1630729 [3] NCCL INFO Channel 06/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114512:115007 [0] NCCL INFO Channel 15/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364904:1365408 [2] NCCL INFO Channel 09/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630209:1630730 [4] NCCL INFO Channel 07/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630207:1630731 [2] NCCL INFO Channel 00/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227468:1227959 [5] NCCL INFO Channel 10/0 : 29[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227469:1227957 [6] NCCL INFO Channel 02/0 : 30[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153568:154056 [6] NCCL INFO Channel 14/0 : 30[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153564:154057 [2] NCCL INFO Channel 00/0 : 26[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32707:33200 [3] NCCL INFO Channel 01/0 : 11[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227467:1227960 [4] NCCL INFO Channel 00/0 : 28[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227470:1227958 [7] NCCL INFO Channel 08/0 : 31[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32708:33198 [4] NCCL INFO Channel 03/0 : 12[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32710:33202 [6] NCCL INFO Channel 07/0 : 14[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227464:1227963 [1] NCCL INFO Channel 08/0 : 25[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227463:1227964 [0] NCCL INFO Channel 04/0 : 24[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364903:1365409 [1] NCCL INFO Channel 05/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32709:33201 [5] NCCL INFO Channel 05/0 : 13[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227468:1227959 [5] NCCL INFO Channel 12/0 : 29[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364906:1365407 [4] NCCL INFO Channel 10/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32704:33197 [0] NCCL INFO Channel 05/0 : 8[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227469:1227957 [6] NCCL INFO Channel 04/0 : 30[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364905:1365404 [3] NCCL INFO Channel 08/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227467:1227960 [4] NCCL INFO Channel 02/0 : 28[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227470:1227958 [7] NCCL INFO Channel 10/0 : 31[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630206:1630728 [1] NCCL INFO Channel 13/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153564:154057 [2] NCCL INFO Channel 04/0 : 26[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364908:1365406 [6] NCCL INFO Channel 07/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364902:1365410 [0] NCCL INFO Channel 07/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227464:1227963 [1] NCCL INFO Channel 10/0 : 25[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227463:1227964 [0] NCCL INFO Channel 06/0 : 24[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630210:1630732 [5] NCCL INFO Channel 14/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364907:1365403 [5] NCCL INFO Channel 06/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227468:1227959 [5] NCCL INFO Channel 14/0 : 29[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630211:1630733 [6] NCCL INFO Channel 05/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364909:1365405 [7] NCCL INFO Channel 09/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227469:1227957 [6] NCCL INFO Channel 08/0 : 30[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227466:1227961 [3] NCCL INFO Channel 00/0 : 27[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630209:1630730 [4] NCCL INFO Channel 08/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630212:1630734 [7] NCCL INFO Channel 11/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630205:1630727 [0] NCCL INFO Channel 06/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364904:1365408 [2] NCCL INFO Channel 11/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364903:1365409 [1] NCCL INFO Channel 06/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32706:33199 [2] NCCL INFO Channel 15/0 : 10[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227467:1227960 [4] NCCL INFO Channel 06/0 : 28[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227470:1227958 [7] NCCL INFO Channel 12/0 : 31[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153564:154057 [2] NCCL INFO Channel 06/0 : 26[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364906:1365407 [4] NCCL INFO Channel 11/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32711:33204 [7] NCCL INFO Channel 03/0 : 15[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227464:1227963 [1] NCCL INFO Channel 12/0 : 25[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227463:1227964 [0] NCCL INFO Channel 08/0 : 24[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227468:1227959 [5] NCCL INFO Channel 00/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227469:1227957 [6] NCCL INFO Channel 10/0 : 30[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630208:1630729 [3] NCCL INFO Channel 07/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630207:1630731 [2] NCCL INFO Channel 01/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227466:1227961 [3] NCCL INFO Channel 02/0 : 27[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630206:1630728 [1] NCCL INFO Channel 14/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153564:154057 [2] NCCL INFO Channel 08/0 : 26[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364905:1365404 [3] NCCL INFO Channel 09/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32707:33200 [3] NCCL INFO Channel 03/0 : 11[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32708:33198 [4] NCCL INFO Channel 05/0 : 12[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227467:1227960 [4] NCCL INFO Channel 08/0 : 28[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227470:1227958 [7] NCCL INFO Channel 14/0 : 31[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630210:1630732 [5] NCCL INFO Channel 15/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364908:1365406 [6] NCCL INFO Channel 08/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32709:33201 [5] NCCL INFO Channel 07/0 : 13[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32704:33197 [0] NCCL INFO Channel 07/0 : 8[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32710:33202 [6] NCCL INFO Channel 09/0 : 14[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32705:33203 [1] NCCL INFO Channel 01/0 : 9[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227465:1227962 [2] NCCL INFO Channel 00/0 : 26[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630211:1630733 [6] NCCL INFO Channel 07/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364902:1365410 [0] NCCL INFO Channel 09/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364909:1365405 [7] NCCL INFO Channel 10/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32706:33199 [2] NCCL INFO Channel 01/0 : 26[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227464:1227963 [1] NCCL INFO Channel 14/0 : 25[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364907:1365403 [5] NCCL INFO Channel 07/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32711:33204 [7] NCCL INFO Channel 05/0 : 15[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227468:1227959 [5] NCCL INFO Channel 01/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227463:1227964 [0] NCCL INFO Channel 10/0 : 24[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153564:154057 [2] NCCL INFO Channel 10/0 : 26[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364904:1365408 [2] NCCL INFO Channel 12/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364903:1365409 [1] NCCL INFO Channel 07/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364906:1365407 [4] NCCL INFO Channel 13/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32707:33200 [3] NCCL INFO Channel 05/0 : 11[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227469:1227957 [6] NCCL INFO Channel 12/0 : 30[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364905:1365404 [3] NCCL INFO Channel 10/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227466:1227961 [3] NCCL INFO Channel 04/0 : 27[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227467:1227960 [4] NCCL INFO Channel 10/0 : 28[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227470:1227958 [7] NCCL INFO Channel 00/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630209:1630730 [4] NCCL INFO Channel 09/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227465:1227962 [2] NCCL INFO Channel 04/0 : 26[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630212:1630734 [7] NCCL INFO Channel 12/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153564:154057 [2] NCCL INFO Channel 12/0 : 26[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227464:1227963 [1] NCCL INFO Channel 00/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630205:1630727 [0] NCCL INFO Channel 07/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630208:1630729 [3] NCCL INFO Channel 08/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630207:1630731 [2] NCCL INFO Channel 03/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32708:33198 [4] NCCL INFO Channel 07/0 : 12[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32709:33201 [5] NCCL INFO Channel 09/0 : 13[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32704:33197 [0] NCCL INFO Channel 09/0 : 8[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32705:33203 [1] NCCL INFO Channel 03/0 : 9[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227463:1227964 [0] NCCL INFO Channel 12/0 : 24[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630210:1630732 [5] NCCL INFO Channel 01/0 : 13[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630206:1630728 [1] NCCL INFO Channel 15/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630211:1630733 [6] NCCL INFO Channel 08/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32706:33199 [2] NCCL INFO Channel 03/0 : 26[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227468:1227959 [5] NCCL INFO Channel 02/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227469:1227957 [6] NCCL INFO Channel 14/0 : 30[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630209:1630730 [4] NCCL INFO Channel 10/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364908:1365406 [6] NCCL INFO Channel 09/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32710:33202 [6] NCCL INFO Channel 11/0 : 14[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32707:33200 [3] NCCL INFO Channel 07/0 : 11[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32711:33204 [7] NCCL INFO Channel 07/0 : 15[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227466:1227961 [3] NCCL INFO Channel 06/0 : 27[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630212:1630734 [7] NCCL INFO Channel 13/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364902:1365410 [0] NCCL INFO Channel 10/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227467:1227960 [4] NCCL INFO Channel 12/0 : 28[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227470:1227958 [7] NCCL INFO Channel 01/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364909:1365405 [7] NCCL INFO Channel 11/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227465:1227962 [2] NCCL INFO Channel 06/0 : 26[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153564:154057 [2] NCCL INFO Channel 14/0 : 26[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364907:1365403 [5] NCCL INFO Channel 08/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32709:33201 [5] NCCL INFO Channel 11/0 : 13[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227464:1227963 [1] NCCL INFO Channel 02/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227463:1227964 [0] NCCL INFO Channel 14/0 : 24[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630208:1630729 [3] NCCL INFO Channel 09/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227468:1227959 [5] NCCL INFO Channel 03/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227469:1227957 [6] NCCL INFO Channel 00/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227466:1227961 [3] NCCL INFO Channel 08/0 : 27[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32708:33198 [4] NCCL INFO Channel 09/0 : 12[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227470:1227958 [7] NCCL INFO Channel 02/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364903:1365409 [1] NCCL INFO Channel 08/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364904:1365408 [2] NCCL INFO Channel 13/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32704:33197 [0] NCCL INFO Channel 11/0 : 8[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227467:1227960 [4] NCCL INFO Channel 14/0 : 28[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364905:1365404 [3] NCCL INFO Channel 12/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227465:1227962 [2] NCCL INFO Channel 08/0 : 26[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630210:1630732 [5] NCCL INFO Channel 03/0 : 13[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630206:1630728 [1] NCCL INFO Channel 01/0 : 9[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630205:1630727 [0] NCCL INFO Channel 09/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630211:1630733 [6] NCCL INFO Channel 09/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364906:1365407 [4] NCCL INFO Channel 14/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227464:1227963 [1] NCCL INFO Channel 03/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630207:1630731 [2] NCCL INFO Channel 04/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630212:1630734 [7] NCCL INFO Channel 14/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630209:1630730 [4] NCCL INFO Channel 11/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364908:1365406 [6] NCCL INFO Channel 10/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364902:1365410 [0] NCCL INFO Channel 11/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227463:1227964 [0] NCCL INFO Channel 01/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227469:1227957 [6] NCCL INFO Channel 01/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227468:1227959 [5] NCCL INFO Channel 04/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364909:1365405 [7] NCCL INFO Channel 12/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227466:1227961 [3] NCCL INFO Channel 10/0 : 27[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364907:1365403 [5] NCCL INFO Channel 09/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32705:33203 [1] NCCL INFO Channel 05/0 : 9[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32706:33199 [2] NCCL INFO Channel 05/0 : 26[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114516:115010 [4] NCCL INFO Channel 11/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32710:33202 [6] NCCL INFO Channel 13/0 : 14[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32707:33200 [3] NCCL INFO Channel 09/0 : 11[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32711:33204 [7] NCCL INFO Channel 09/0 : 15[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630208:1630729 [3] NCCL INFO Channel 10/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32709:33201 [5] NCCL INFO Channel 15/0 : 13[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227470:1227958 [7] NCCL INFO Channel 03/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364903:1365409 [1] NCCL INFO Channel 10/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364904:1365408 [2] NCCL INFO Channel 14/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227467:1227960 [4] NCCL INFO Channel 00/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227465:1227962 [2] NCCL INFO Channel 10/0 : 26[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114518:115012 [6] NCCL INFO Channel 15/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364905:1365404 [3] NCCL INFO Channel 13/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227464:1227963 [1] NCCL INFO Channel 04/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630206:1630728 [1] NCCL INFO Channel 03/0 : 9[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630205:1630727 [0] NCCL INFO Channel 10/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630210:1630732 [5] NCCL INFO Channel 05/0 : 13[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114516:115010 [4] NCCL INFO Channel 13/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364902:1365410 [0] NCCL INFO Channel 12/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364906:1365407 [4] NCCL INFO Channel 15/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364908:1365406 [6] NCCL INFO Channel 11/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32704:33197 [0] NCCL INFO Channel 13/0 : 8[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32708:33198 [4] NCCL INFO Channel 11/0 : 12[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227463:1227964 [0] NCCL INFO Channel 02/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364909:1365405 [7] NCCL INFO Channel 13/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32706:33199 [2] NCCL INFO Channel 07/0 : 26[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227469:1227957 [6] NCCL INFO Channel 02/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364907:1365403 [5] NCCL INFO Channel 10/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32705:33203 [1] NCCL INFO Channel 07/0 : 9[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227468:1227959 [5] NCCL INFO Channel 06/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364904:1365408 [2] NCCL INFO Channel 15/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364903:1365409 [1] NCCL INFO Channel 11/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227470:1227958 [7] NCCL INFO Channel 04/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227466:1227961 [3] NCCL INFO Channel 12/0 : 27[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364905:1365404 [3] NCCL INFO Channel 14/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227467:1227960 [4] NCCL INFO Channel 01/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630211:1630733 [6] NCCL INFO Channel 10/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630209:1630730 [4] NCCL INFO Channel 13/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227465:1227962 [2] NCCL INFO Channel 12/0 : 26[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630212:1630734 [7] NCCL INFO Channel 01/0 : 15[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114516:115010 [4] NCCL INFO Channel 15/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32707:33200 [3] NCCL INFO Channel 13/0 : 11[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227464:1227963 [1] NCCL INFO Channel 05/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630207:1630731 [2] NCCL INFO Channel 05/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630208:1630729 [3] NCCL INFO Channel 12/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153569:154055 [7] NCCL INFO Channel 06/0 : 31[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32710:33202 [6] NCCL INFO Channel 15/0 : 14[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227469:1227957 [6] NCCL INFO Channel 03/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227463:1227964 [0] NCCL INFO Channel 03/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227468:1227959 [5] NCCL INFO Channel 07/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32709:33201 [5] NCCL INFO Channel 01/0 : 29[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227470:1227958 [7] NCCL INFO Channel 05/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364908:1365406 [6] NCCL INFO Channel 12/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32711:33204 [7] NCCL INFO Channel 11/0 : 15[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227466:1227961 [3] NCCL INFO Channel 14/0 : 27[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364902:1365410 [0] NCCL INFO Channel 13/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32704:33197 [0] NCCL INFO Channel 15/0 : 8[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32706:33199 [2] NCCL INFO Channel 09/0 : 26[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32708:33198 [4] NCCL INFO Channel 13/0 : 12[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227467:1227960 [4] NCCL INFO Channel 02/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227465:1227962 [2] NCCL INFO Channel 14/0 : 26[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364909:1365405 [7] NCCL INFO Channel 14/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227464:1227963 [1] NCCL INFO Channel 06/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630210:1630732 [5] NCCL INFO Channel 07/0 : 13[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364907:1365403 [5] NCCL INFO Channel 11/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227469:1227957 [6] NCCL INFO Channel 04/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630205:1630727 [0] NCCL INFO Channel 11/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153569:154055 [7] NCCL INFO Channel 08/0 : 31[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364903:1365409 [1] NCCL INFO Channel 12/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364905:1365404 [3] NCCL INFO Channel 15/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227463:1227964 [0] NCCL INFO Channel 04/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364908:1365406 [6] NCCL INFO Channel 13/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32705:33203 [1] NCCL INFO Channel 11/0 : 9[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32707:33200 [3] NCCL INFO Channel 15/0 : 11[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227468:1227959 [5] NCCL INFO Channel 08/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364902:1365410 [0] NCCL INFO Channel 14/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32709:33201 [5] NCCL INFO Channel 03/0 : 29[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32710:33202 [6] NCCL INFO Channel 01/0 : 30[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227470:1227958 [7] NCCL INFO Channel 06/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32711:33204 [7] NCCL INFO Channel 13/0 : 15[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227466:1227961 [3] NCCL INFO Channel 00/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32704:33197 [0] NCCL INFO Channel 01/0 : 24[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227467:1227960 [4] NCCL INFO Channel 03/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227465:1227962 [2] NCCL INFO Channel 00/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630209:1630730 [4] NCCL INFO Channel 14/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630211:1630733 [6] NCCL INFO Channel 11/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630206:1630728 [1] NCCL INFO Channel 05/0 : 9[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630212:1630734 [7] NCCL INFO Channel 03/0 : 15[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153569:154055 [7] NCCL INFO Channel 10/0 : 31[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32706:33199 [2] NCCL INFO Channel 11/0 : 26[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227464:1227963 [1] NCCL INFO Channel 07/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227469:1227957 [6] NCCL INFO Channel 05/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630208:1630729 [3] NCCL INFO Channel 13/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364907:1365403 [5] NCCL INFO Channel 12/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364903:1365409 [1] NCCL INFO Channel 13/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227468:1227959 [5] NCCL INFO Channel 09/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630207:1630731 [2] NCCL INFO Channel 06/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630210:1630732 [5] NCCL INFO Channel 09/0 : 13[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364908:1365406 [6] NCCL INFO Channel 15/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630205:1630727 [0] NCCL INFO Channel 12/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153569:154055 [7] NCCL INFO Channel 12/0 : 31[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364902:1365410 [0] NCCL INFO Channel 15/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32708:33198 [4] NCCL INFO Channel 15/0 : 12[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364907:1365403 [5] NCCL INFO Channel 14/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364903:1365409 [1] NCCL INFO Channel 14/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32707:33200 [3] NCCL INFO Channel 01/0 : 27[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32705:33203 [1] NCCL INFO Channel 13/0 : 9[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32709:33201 [5] NCCL INFO Channel 05/0 : 29[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32710:33202 [6] NCCL INFO Channel 03/0 : 30[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32711:33204 [7] NCCL INFO Channel 01/0 : 31[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227463:1227964 [0] NCCL INFO Channel 05/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227470:1227958 [7] NCCL INFO Channel 08/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153569:154055 [7] NCCL INFO Channel 14/0 : 31[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96892:97383 [4] NCCL INFO Channel 00/0 : 4[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32704:33197 [0] NCCL INFO Channel 03/0 : 24[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32706:33199 [2] NCCL INFO Channel 13/0 : 26[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227466:1227961 [3] NCCL INFO Channel 01/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227467:1227960 [4] NCCL INFO Channel 05/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630209:1630730 [4] NCCL INFO Channel 15/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364907:1365403 [5] NCCL INFO Channel 15/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32708:33198 [4] NCCL INFO Channel 01/0 : 28[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32707:33200 [3] NCCL INFO Channel 03/0 : 27[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227469:1227957 [6] NCCL INFO Channel 07/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227464:1227963 [1] NCCL INFO Channel 08/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630212:1630734 [7] NCCL INFO Channel 05/0 : 15[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630206:1630728 [1] NCCL INFO Channel 07/0 : 9[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124162:124660 [2] NCCL INFO Channel 00/0 : 18[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364903:1365409 [1] NCCL INFO Channel 15/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227465:1227962 [2] NCCL INFO Channel 01/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630211:1630733 [6] NCCL INFO Channel 12/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364906:1365407 [4] NCCL INFO Channel 01/0 : 28[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227468:1227959 [5] NCCL INFO Channel 10/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630207:1630731 [2] NCCL INFO Channel 07/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630208:1630729 [3] NCCL INFO Channel 14/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630210:1630732 [5] NCCL INFO Channel 11/0 : 13[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96890:97386 [2] NCCL INFO Channel 00/0 : 2[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364904:1365408 [2] NCCL INFO Channel 01/0 : 26[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227470:1227958 [7] NCCL INFO Channel 09/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227463:1227964 [0] NCCL INFO Channel 06/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630205:1630727 [0] NCCL INFO Channel 13/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96892:97383 [4] NCCL INFO Channel 02/0 : 4[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32709:33201 [5] NCCL INFO Channel 07/0 : 29[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227466:1227961 [3] NCCL INFO Channel 02/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227467:1227960 [4] NCCL INFO Channel 06/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32704:33197 [0] NCCL INFO Channel 05/0 : 24[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227469:1227957 [6] NCCL INFO Channel 08/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227464:1227963 [1] NCCL INFO Channel 10/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227465:1227962 [2] NCCL INFO Channel 03/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364904:1365408 [2] NCCL INFO Channel 03/0 : 26[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364906:1365407 [4] NCCL INFO Channel 03/0 : 28[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227468:1227959 [5] NCCL INFO Channel 11/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630209:1630730 [4] NCCL INFO Channel 01/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364909:1365405 [7] NCCL INFO Channel 01/0 : 31[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364905:1365404 [3] NCCL INFO Channel 01/0 : 27[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227470:1227958 [7] NCCL INFO Channel 10/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227463:1227964 [0] NCCL INFO Channel 07/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630212:1630734 [7] NCCL INFO Channel 07/0 : 15[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124162:124660 [2] NCCL INFO Channel 04/0 : 18[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96895:97382 [7] NCCL INFO Channel 00/0 : 7[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96890:97386 [2] NCCL INFO Channel 04/0 : 2[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32709:33201 [5] NCCL INFO Channel 09/0 : 29[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227466:1227961 [3] NCCL INFO Channel 04/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630206:1630728 [1] NCCL INFO Channel 11/0 : 9[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96892:97383 [4] NCCL INFO Channel 06/0 : 4[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32704:33197 [0] NCCL INFO Channel 07/0 : 24[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227467:1227960 [4] NCCL INFO Channel 07/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630211:1630733 [6] NCCL INFO Channel 13/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364904:1365408 [2] NCCL INFO Channel 05/0 : 26[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227469:1227957 [6] NCCL INFO Channel 09/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227465:1227962 [2] NCCL INFO Channel 04/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227464:1227963 [1] NCCL INFO Channel 11/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96891:97385 [3] NCCL INFO Channel 00/0 : 3[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32709:33201 [5] NCCL INFO Channel 11/0 : 29[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630210:1630732 [5] NCCL INFO Channel 15/0 : 13[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124162:124660 [2] NCCL INFO Channel 06/0 : 18[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96895:97382 [7] NCCL INFO Channel 02/0 : 7[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364906:1365407 [4] NCCL INFO Channel 05/0 : 28[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32704:33197 [0] NCCL INFO Channel 09/0 : 24[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630208:1630729 [3] NCCL INFO Channel 15/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630205:1630727 [0] NCCL INFO Channel 14/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630209:1630730 [4] NCCL INFO Channel 03/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630207:1630731 [2] NCCL INFO Channel 08/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364909:1365405 [7] NCCL INFO Channel 03/0 : 31[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32709:33201 [5] NCCL INFO Channel 15/0 : 29[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227468:1227959 [5] NCCL INFO Channel 12/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227470:1227958 [7] NCCL INFO Channel 11/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630212:1630734 [7] NCCL INFO Channel 09/0 : 15[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630206:1630728 [1] NCCL INFO Channel 13/0 : 9[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364905:1365404 [3] NCCL INFO Channel 03/0 : 27[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364902:1365410 [0] NCCL INFO Channel 01/0 : 24[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364908:1365406 [6] NCCL INFO Channel 01/0 : 30[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227463:1227964 [0] NCCL INFO Channel 09/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630211:1630733 [6] NCCL INFO Channel 15/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227467:1227960 [4] NCCL INFO Channel 08/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227469:1227957 [6] NCCL INFO Channel 10/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96890:97386 [2] NCCL INFO Channel 06/0 : 2[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227466:1227961 [3] NCCL INFO Channel 05/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96892:97383 [4] NCCL INFO Channel 08/0 : 4[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32704:33197 [0] NCCL INFO Channel 11/0 : 24[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227465:1227962 [2] NCCL INFO Channel 05/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:124162:124660 [2] NCCL INFO Channel 08/0 : 18[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96891:97385 [3] NCCL INFO Channel 02/0 : 3[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96888:97387 [0] NCCL INFO Channel 02/0 : 0[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96894:97381 [6] NCCL INFO Channel 00/0 : 6[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364904:1365408 [2] NCCL INFO Channel 07/0 : 26[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227464:1227963 [1] NCCL INFO Channel 12/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630209:1630730 [4] NCCL INFO Channel 05/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96895:97382 [7] NCCL INFO Channel 04/0 : 7[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364906:1365407 [4] NCCL INFO Channel 07/0 : 28[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227470:1227958 [7] NCCL INFO Channel 12/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227468:1227959 [5] NCCL INFO Channel 14/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630208:1630729 [3] NCCL INFO Channel 01/0 : 11[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630207:1630731 [2] NCCL INFO Channel 09/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630205:1630727 [0] NCCL INFO Channel 15/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96890:97386 [2] NCCL INFO Channel 08/0 : 2[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364909:1365405 [7] NCCL INFO Channel 05/0 : 31[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227467:1227960 [4] NCCL INFO Channel 09/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630212:1630734 [7] NCCL INFO Channel 11/0 : 15[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630206:1630728 [1] NCCL INFO Channel 15/0 : 9[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630211:1630733 [6] NCCL INFO Channel 01/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32704:33197 [0] NCCL INFO Channel 13/0 : 24[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227469:1227957 [6] NCCL INFO Channel 11/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630209:1630730 [4] NCCL INFO Channel 07/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227463:1227964 [0] NCCL INFO Channel 10/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227466:1227961 [3] NCCL INFO Channel 06/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114515:115009 [3] NCCL INFO Channel 13/0 : 11[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96892:97383 [4] NCCL INFO Channel 10/0 : 4[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227465:1227962 [2] NCCL INFO Channel 06/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227464:1227963 [1] NCCL INFO Channel 13/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124162:124660 [2] NCCL INFO Channel 10/0 : 18[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96891:97385 [3] NCCL INFO Channel 04/0 : 3[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364902:1365410 [0] NCCL INFO Channel 03/0 : 24[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32704:33197 [0] NCCL INFO Channel 15/0 : 24[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227470:1227958 [7] NCCL INFO Channel 13/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227468:1227959 [5] NCCL INFO Channel 15/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364905:1365404 [3] NCCL INFO Channel 05/0 : 27[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227467:1227960 [4] NCCL INFO Channel 10/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630207:1630731 [2] NCCL INFO Channel 11/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:124167:124664 [7] NCCL INFO Channel 00/0 : 23[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364904:1365408 [2] NCCL INFO Channel 09/0 : 26[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364908:1365406 [6] NCCL INFO Channel 03/0 : 30[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227469:1227957 [6] NCCL INFO Channel 12/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630205:1630727 [0] NCCL INFO Channel 01/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114515:115009 [3] NCCL INFO Channel 15/0 : 11[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227463:1227964 [0] NCCL INFO Channel 11/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227466:1227961 [3] NCCL INFO Channel 07/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630208:1630729 [3] NCCL INFO Channel 03/0 : 11[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:124162:124660 [2] NCCL INFO Channel 12/0 : 18[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96894:97381 [6] NCCL INFO Channel 02/0 : 6[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96888:97387 [0] NCCL INFO Channel 04/0 : 0[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96893:97384 [5] NCCL INFO Channel 00/0 : 5[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96895:97382 [7] NCCL INFO Channel 06/0 : 7[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227465:1227962 [2] NCCL INFO Channel 07/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227464:1227963 [1] NCCL INFO Channel 14/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227470:1227958 [7] NCCL INFO Channel 14/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630212:1630734 [7] NCCL INFO Channel 13/0 : 15[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96890:97386 [2] NCCL INFO Channel 10/0 : 2[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630211:1630733 [6] NCCL INFO Channel 03/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630209:1630730 [4] NCCL INFO Channel 09/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96889:97388 [1] NCCL INFO Channel 00/0 : 1[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96892:97383 [4] NCCL INFO Channel 12/0 : 4[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364907:1365403 [5] NCCL INFO Channel 01/0 : 29[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364903:1365409 [1] NCCL INFO Channel 01/0 : 25[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364906:1365407 [4] NCCL INFO Channel 09/0 : 28[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:124167:124664 [7] NCCL INFO Channel 02/0 : 23[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96891:97385 [3] NCCL INFO Channel 06/0 : 3[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364909:1365405 [7] NCCL INFO Channel 07/0 : 31[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364902:1365410 [0] NCCL INFO Channel 05/0 : 24[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364905:1365404 [3] NCCL INFO Channel 07/0 : 27[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227467:1227960 [4] NCCL INFO Channel 11/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227469:1227957 [6] NCCL INFO Channel 13/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:124162:124660 [2] NCCL INFO Channel 14/0 : 18[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227463:1227964 [0] NCCL INFO Channel 12/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227466:1227961 [3] NCCL INFO Channel 08/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630205:1630727 [0] NCCL INFO Channel 03/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227465:1227962 [2] NCCL INFO Channel 08/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630208:1630729 [3] NCCL INFO Channel 05/0 : 11[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96894:97381 [6] NCCL INFO Channel 04/0 : 6[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227464:1227963 [1] NCCL INFO Channel 15/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630209:1630730 [4] NCCL INFO Channel 11/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:124167:124664 [7] NCCL INFO Channel 04/0 : 23[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364904:1365408 [2] NCCL INFO Channel 11/0 : 26[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227467:1227960 [4] NCCL INFO Channel 13/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630211:1630733 [6] NCCL INFO Channel 05/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364908:1365406 [6] NCCL INFO Channel 05/0 : 30[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227469:1227957 [6] NCCL INFO Channel 15/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630207:1630731 [2] NCCL INFO Channel 12/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364903:1365409 [1] NCCL INFO Channel 03/0 : 25[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227466:1227961 [3] NCCL INFO Channel 09/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:124162:124660 [2] NCCL INFO Channel 00/0 : 2[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96888:97387 [0] NCCL INFO Channel 06/0 : 0[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364907:1365403 [5] NCCL INFO Channel 03/0 : 29[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364906:1365407 [4] NCCL INFO Channel 11/0 : 28[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364902:1365410 [0] NCCL INFO Channel 07/0 : 24[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227463:1227964 [0] NCCL INFO Channel 13/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227467:1227960 [4] NCCL INFO Channel 14/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96893:97384 [5] NCCL INFO Channel 02/0 : 5[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96895:97382 [7] NCCL INFO Channel 08/0 : 7[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96890:97386 [2] NCCL INFO Channel 12/0 : 2[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364909:1365405 [7] NCCL INFO Channel 09/0 : 31[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227465:1227962 [2] NCCL INFO Channel 09/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:124167:124664 [7] NCCL INFO Channel 06/0 : 23[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96892:97383 [4] NCCL INFO Channel 14/0 : 4[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364905:1365404 [3] NCCL INFO Channel 09/0 : 27[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227466:1227961 [3] NCCL INFO Channel 10/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630205:1630727 [0] NCCL INFO Channel 05/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630208:1630729 [3] NCCL INFO Channel 07/0 : 11[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96891:97385 [3] NCCL INFO Channel 08/0 : 3[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227467:1227960 [4] NCCL INFO Channel 15/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630209:1630730 [4] NCCL INFO Channel 13/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:124166:124665 [6] NCCL INFO Channel 00/0 : 22[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96889:97388 [1] NCCL INFO Channel 02/0 : 1[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227463:1227964 [0] NCCL INFO Channel 14/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630211:1630733 [6] NCCL INFO Channel 07/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:124162:124660 [2] NCCL INFO Channel 04/0 : 2[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96894:97381 [6] NCCL INFO Channel 08/0 : 6[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227465:1227962 [2] NCCL INFO Channel 11/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227466:1227961 [3] NCCL INFO Channel 12/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630207:1630731 [2] NCCL INFO Channel 13/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364904:1365408 [2] NCCL INFO Channel 13/0 : 26[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630208:1630729 [3] NCCL INFO Channel 09/0 : 11[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:124167:124664 [7] NCCL INFO Channel 08/0 : 23[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364908:1365406 [6] NCCL INFO Channel 07/0 : 30[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630205:1630727 [0] NCCL INFO Channel 07/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364903:1365409 [1] NCCL INFO Channel 05/0 : 25[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630209:1630730 [4] NCCL INFO Channel 15/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:124162:124660 [2] NCCL INFO Channel 06/0 : 2[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227463:1227964 [0] NCCL INFO Channel 15/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227465:1227962 [2] NCCL INFO Channel 12/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227466:1227961 [3] NCCL INFO Channel 13/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96895:97382 [7] NCCL INFO Channel 10/0 : 7[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227466:1227961 [3] NCCL INFO Channel 14/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227465:1227962 [2] NCCL INFO Channel 13/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:124166:124665 [6] NCCL INFO Channel 02/0 : 22[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96893:97384 [5] NCCL INFO Channel 04/0 : 5[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124167:124664 [7] NCCL INFO Channel 10/0 : 23[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96890:97386 [2] NCCL INFO Channel 14/0 : 2[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96888:97387 [0] NCCL INFO Channel 08/0 : 0[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96891:97385 [3] NCCL INFO Channel 10/0 : 3[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96892:97383 [4] NCCL INFO Channel 00/0 : 20[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364902:1365410 [0] NCCL INFO Channel 09/0 : 24[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96889:97388 [1] NCCL INFO Channel 04/0 : 1[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96894:97381 [6] NCCL INFO Channel 10/0 : 6[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364906:1365407 [4] NCCL INFO Channel 13/0 : 28[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630211:1630733 [6] NCCL INFO Channel 09/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364909:1365405 [7] NCCL INFO Channel 11/0 : 31[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364907:1365403 [5] NCCL INFO Channel 05/0 : 29[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364905:1365404 [3] NCCL INFO Channel 13/0 : 27[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227466:1227961 [3] NCCL INFO Channel 15/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227465:1227962 [2] NCCL INFO Channel 14/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630207:1630731 [2] NCCL INFO Channel 14/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:124162:124660 [2] NCCL INFO Channel 08/0 : 2[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364904:1365408 [2] NCCL INFO Channel 15/0 : 26[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630208:1630729 [3] NCCL INFO Channel 13/0 : 11[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364903:1365409 [1] NCCL INFO Channel 07/0 : 25[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364908:1365406 [6] NCCL INFO Channel 09/0 : 30[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630211:1630733 [6] NCCL INFO Channel 11/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630207:1630731 [2] NCCL INFO Channel 15/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96895:97382 [7] NCCL INFO Channel 12/0 : 7[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364902:1365410 [0] NCCL INFO Channel 11/0 : 24[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630208:1630729 [3] NCCL INFO Channel 15/0 : 11[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96893:97384 [5] NCCL INFO Channel 06/0 : 5[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630205:1630727 [0] NCCL INFO Channel 09/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:124167:124664 [7] NCCL INFO Channel 12/0 : 23[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124166:124665 [6] NCCL INFO Channel 04/0 : 22[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:124162:124660 [2] NCCL INFO Channel 10/0 : 2[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364906:1365407 [4] NCCL INFO Channel 15/0 : 28[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96890:97386 [2] NCCL INFO Channel 00/0 : 18[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364909:1365405 [7] NCCL INFO Channel 13/0 : 31[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124167:124664 [7] NCCL INFO Channel 14/0 : 23[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96888:97387 [0] NCCL INFO Channel 10/0 : 0[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364904:1365408 [2] NCCL INFO Channel 01/0 : 10[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630211:1630733 [6] NCCL INFO Channel 13/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630207:1630731 [2] NCCL INFO Channel 01/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:124162:124660 [2] NCCL INFO Channel 12/0 : 2[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96891:97385 [3] NCCL INFO Channel 12/0 : 3[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364905:1365404 [3] NCCL INFO Channel 15/0 : 27[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364907:1365403 [5] NCCL INFO Channel 07/0 : 29[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630205:1630727 [0] NCCL INFO Channel 11/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:124166:124665 [6] NCCL INFO Channel 08/0 : 22[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96892:97383 [4] NCCL INFO Channel 02/0 : 20[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364903:1365409 [1] NCCL INFO Channel 11/0 : 25[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96894:97381 [6] NCCL INFO Channel 12/0 : 6[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96889:97388 [1] NCCL INFO Channel 06/0 : 1[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96895:97382 [7] NCCL INFO Channel 14/0 : 7[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96893:97384 [5] NCCL INFO Channel 08/0 : 5[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364908:1365406 [6] NCCL INFO Channel 11/0 : 30[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:124167:124664 [7] NCCL INFO Channel 00/0 : 7[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364902:1365410 [0] NCCL INFO Channel 13/0 : 24[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:124162:124660 [2] NCCL INFO Channel 14/0 : 2[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364906:1365407 [4] NCCL INFO Channel 01/0 : 12[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630211:1630733 [6] NCCL INFO Channel 15/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630207:1630731 [2] NCCL INFO Channel 03/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:124166:124665 [6] NCCL INFO Channel 10/0 : 22[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96890:97386 [2] NCCL INFO Channel 04/0 : 18[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630205:1630727 [0] NCCL INFO Channel 13/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96888:97387 [0] NCCL INFO Channel 12/0 : 0[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:124167:124664 [7] NCCL INFO Channel 02/0 : 7[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96891:97385 [3] NCCL INFO Channel 14/0 : 3[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364904:1365408 [2] NCCL INFO Channel 03/0 : 10[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96892:97383 [4] NCCL INFO Channel 06/0 : 20[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364909:1365405 [7] NCCL INFO Channel 01/0 : 15[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96895:97382 [7] NCCL INFO Channel 00/0 : 23[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364905:1365404 [3] NCCL INFO Channel 01/0 : 11[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364907:1365403 [5] NCCL INFO Channel 09/0 : 29[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364903:1365409 [1] NCCL INFO Channel 13/0 : 25[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630207:1630731 [2] NCCL INFO Channel 05/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364908:1365406 [6] NCCL INFO Channel 13/0 : 30[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364902:1365410 [0] NCCL INFO Channel 15/0 : 24[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364906:1365407 [4] NCCL INFO Channel 03/0 : 12[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227465:1227962 [2] NCCL INFO Channel 15/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630205:1630727 [0] NCCL INFO Channel 15/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:124166:124665 [6] NCCL INFO Channel 12/0 : 22[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:124167:124664 [7] NCCL INFO Channel 04/0 : 7[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96894:97381 [6] NCCL INFO Channel 14/0 : 6[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96893:97384 [5] NCCL INFO Channel 10/0 : 5[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96890:97386 [2] NCCL INFO Channel 06/0 : 18[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96889:97388 [1] NCCL INFO Channel 08/0 : 1[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364904:1365408 [2] NCCL INFO Channel 05/0 : 10[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96888:97387 [0] NCCL INFO Channel 14/0 : 0[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364905:1365404 [3] NCCL INFO Channel 03/0 : 11[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364909:1365405 [7] NCCL INFO Channel 03/0 : 15[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630207:1630731 [2] NCCL INFO Channel 07/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96891:97385 [3] NCCL INFO Channel 00/0 : 19[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96892:97383 [4] NCCL INFO Channel 08/0 : 20[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364903:1365409 [1] NCCL INFO Channel 15/0 : 25[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630207:1630731 [2] NCCL INFO Channel 09/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:124166:124665 [6] NCCL INFO Channel 14/0 : 22[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96895:97382 [7] NCCL INFO Channel 02/0 : 23[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96894:97381 [6] NCCL INFO Channel 00/0 : 22[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:124167:124664 [7] NCCL INFO Channel 06/0 : 7[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124160:124658 [0] NCCL INFO Channel 02/0 : 16[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364907:1365403 [5] NCCL INFO Channel 11/0 : 29[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364906:1365407 [4] NCCL INFO Channel 05/0 : 12[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630207:1630731 [2] NCCL INFO Channel 11/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364908:1365406 [6] NCCL INFO Channel 15/0 : 30[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:124166:124665 [6] NCCL INFO Channel 00/0 : 6[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96890:97386 [2] NCCL INFO Channel 08/0 : 18[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364904:1365408 [2] NCCL INFO Channel 07/0 : 10[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364902:1365410 [0] NCCL INFO Channel 01/0 : 8[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:124167:124664 [7] NCCL INFO Channel 08/0 : 7[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96893:97384 [5] NCCL INFO Channel 12/0 : 5[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96889:97388 [1] NCCL INFO Channel 10/0 : 1[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96888:97387 [0] NCCL INFO Channel 02/0 : 16[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96891:97385 [3] NCCL INFO Channel 02/0 : 19[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96892:97383 [4] NCCL INFO Channel 10/0 : 20[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96895:97382 [7] NCCL INFO Channel 04/0 : 23[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96894:97381 [6] NCCL INFO Channel 02/0 : 22[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630207:1630731 [2] NCCL INFO Channel 13/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:124160:124658 [0] NCCL INFO Channel 04/0 : 16[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364905:1365404 [3] NCCL INFO Channel 05/0 : 11[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364909:1365405 [7] NCCL INFO Channel 05/0 : 15[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124166:124665 [6] NCCL INFO Channel 02/0 : 6[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364903:1365409 [1] NCCL INFO Channel 01/0 : 9[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124167:124664 [7] NCCL INFO Channel 10/0 : 7[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364906:1365407 [4] NCCL INFO Channel 07/0 : 12[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96890:97386 [2] NCCL INFO Channel 10/0 : 18[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364907:1365403 [5] NCCL INFO Channel 15/0 : 29[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630207:1630731 [2] NCCL INFO Channel 15/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96893:97384 [5] NCCL INFO Channel 14/0 : 5[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364908:1365406 [6] NCCL INFO Channel 01/0 : 14[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:124160:124658 [0] NCCL INFO Channel 06/0 : 16[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96888:97387 [0] NCCL INFO Channel 04/0 : 16[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:124166:124665 [6] NCCL INFO Channel 04/0 : 6[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:124167:124664 [7] NCCL INFO Channel 12/0 : 7[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364902:1365410 [0] NCCL INFO Channel 03/0 : 8[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364904:1365408 [2] NCCL INFO Channel 09/0 : 10[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364905:1365404 [3] NCCL INFO Channel 07/0 : 11[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364909:1365405 [7] NCCL INFO Channel 07/0 : 15[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96891:97385 [3] NCCL INFO Channel 04/0 : 19[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96892:97383 [4] NCCL INFO Channel 12/0 : 20[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364903:1365409 [1] NCCL INFO Channel 03/0 : 9[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96895:97382 [7] NCCL INFO Channel 06/0 : 23[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96894:97381 [6] NCCL INFO Channel 04/0 : 22[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96889:97388 [1] NCCL INFO Channel 12/0 : 1[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96890:97386 [2] NCCL INFO Channel 12/0 : 18[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364906:1365407 [4] NCCL INFO Channel 09/0 : 12[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96893:97384 [5] NCCL INFO Channel 00/0 : 21[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364907:1365403 [5] NCCL INFO Channel 01/0 : 13[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124160:124658 [0] NCCL INFO Channel 08/0 : 16[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:124166:124665 [6] NCCL INFO Channel 08/0 : 6[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96888:97387 [0] NCCL INFO Channel 06/0 : 16[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364908:1365406 [6] NCCL INFO Channel 03/0 : 14[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:124167:124664 [7] NCCL INFO Channel 14/0 : 7[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364902:1365410 [0] NCCL INFO Channel 05/0 : 8[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96891:97385 [3] NCCL INFO Channel 06/0 : 19[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:124166:124665 [6] NCCL INFO Channel 10/0 : 6[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96892:97383 [4] NCCL INFO Channel 14/0 : 20[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364904:1365408 [2] NCCL INFO Channel 11/0 : 10[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:124160:124658 [0] NCCL INFO Channel 10/0 : 16[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364905:1365404 [3] NCCL INFO Channel 09/0 : 11[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364903:1365409 [1] NCCL INFO Channel 05/0 : 9[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364909:1365405 [7] NCCL INFO Channel 09/0 : 15[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364906:1365407 [4] NCCL INFO Channel 11/0 : 12[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364907:1365403 [5] NCCL INFO Channel 03/0 : 13[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364908:1365406 [6] NCCL INFO Channel 05/0 : 14[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:124166:124665 [6] NCCL INFO Channel 12/0 : 6[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96895:97382 [7] NCCL INFO Channel 08/0 : 23[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96890:97386 [2] NCCL INFO Channel 14/0 : 18[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96894:97381 [6] NCCL INFO Channel 08/0 : 22[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96889:97388 [1] NCCL INFO Channel 14/0 : 1[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96888:97387 [0] NCCL INFO Channel 08/0 : 16[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364902:1365410 [0] NCCL INFO Channel 07/0 : 8[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:124160:124658 [0] NCCL INFO Channel 12/0 : 16[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96893:97384 [5] NCCL INFO Channel 02/0 : 21[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364904:1365408 [2] NCCL INFO Channel 13/0 : 10[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96891:97385 [3] NCCL INFO Channel 08/0 : 19[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:124166:124665 [6] NCCL INFO Channel 14/0 : 6[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:124160:124658 [0] NCCL INFO Channel 14/0 : 16[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96895:97382 [7] NCCL INFO Channel 10/0 : 23[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364903:1365409 [1] NCCL INFO Channel 07/0 : 9[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96894:97381 [6] NCCL INFO Channel 10/0 : 22[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364905:1365404 [3] NCCL INFO Channel 13/0 : 11[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96888:97387 [0] NCCL INFO Channel 10/0 : 16[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364909:1365405 [7] NCCL INFO Channel 11/0 : 15[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96893:97384 [5] NCCL INFO Channel 04/0 : 21[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96891:97385 [3] NCCL INFO Channel 10/0 : 19[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364906:1365407 [4] NCCL INFO Channel 13/0 : 12[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96889:97388 [1] NCCL INFO Channel 00/0 : 17[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364907:1365403 [5] NCCL INFO Channel 05/0 : 13[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364904:1365408 [2] NCCL INFO Channel 15/0 : 10[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364902:1365410 [0] NCCL INFO Channel 09/0 : 8[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364908:1365406 [6] NCCL INFO Channel 07/0 : 14[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364903:1365409 [1] NCCL INFO Channel 11/0 : 9[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124160:124658 [0] NCCL INFO Channel 02/0 : 0[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364905:1365404 [3] NCCL INFO Channel 15/0 : 11[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:124162:124660 [2] NCCL INFO Channel 01/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364909:1365405 [7] NCCL INFO Channel 13/0 : 15[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364906:1365407 [4] NCCL INFO Channel 15/0 : 12[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96895:97382 [7] NCCL INFO Channel 12/0 : 23[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96894:97381 [6] NCCL INFO Channel 12/0 : 22[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:124160:124658 [0] NCCL INFO Channel 04/0 : 0[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:124162:124660 [2] NCCL INFO Channel 03/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364907:1365403 [5] NCCL INFO Channel 07/0 : 13[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364902:1365410 [0] NCCL INFO Channel 11/0 : 8[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364908:1365406 [6] NCCL INFO Channel 09/0 : 14[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96891:97385 [3] NCCL INFO Channel 12/0 : 19[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96888:97387 [0] NCCL INFO Channel 12/0 : 16[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96893:97384 [5] NCCL INFO Channel 06/0 : 21[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364903:1365409 [1] NCCL INFO Channel 13/0 : 9[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96890:97386 [2] NCCL INFO Channel 00/0 : 26[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364907:1365403 [5] NCCL INFO Channel 09/0 : 13[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124165:124662 [5] NCCL INFO Channel 00/0 : 21[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96895:97382 [7] NCCL INFO Channel 14/0 : 23[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364908:1365406 [6] NCCL INFO Channel 11/0 : 14[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96894:97381 [6] NCCL INFO Channel 14/0 : 22[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364902:1365410 [0] NCCL INFO Channel 13/0 : 8[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364903:1365409 [1] NCCL INFO Channel 15/0 : 9[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96889:97388 [1] NCCL INFO Channel 02/0 : 17[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364907:1365403 [5] NCCL INFO Channel 11/0 : 13[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124160:124658 [0] NCCL INFO Channel 06/0 : 0[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:124162:124660 [2] NCCL INFO Channel 05/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96891:97385 [3] NCCL INFO Channel 14/0 : 19[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96888:97387 [0] NCCL INFO Channel 14/0 : 16[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364908:1365406 [6] NCCL INFO Channel 13/0 : 14[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32705:33203 [1] NCCL INFO Channel 15/0 : 9[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96890:97386 [2] NCCL INFO Channel 04/0 : 26[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96893:97384 [5] NCCL INFO Channel 08/0 : 21[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364902:1365410 [0] NCCL INFO Channel 15/0 : 8[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:124165:124662 [5] NCCL INFO Channel 02/0 : 21[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96889:97388 [1] NCCL INFO Channel 04/0 : 17[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364907:1365403 [5] NCCL INFO Channel 15/0 : 13[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124162:124660 [2] NCCL INFO Channel 07/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364908:1365406 [6] NCCL INFO Channel 15/0 : 14[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:124161:124659 [1] NCCL INFO Channel 00/0 : 17[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32708:33198 [4] NCCL INFO Channel 03/0 : 28[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:124160:124658 [0] NCCL INFO Channel 08/0 : 0[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96890:97386 [2] NCCL INFO Channel 06/0 : 26[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96893:97384 [5] NCCL INFO Channel 10/0 : 21[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124166:124665 [6] NCCL INFO Channel 01/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:124167:124664 [7] NCCL INFO Channel 01/0 : 15[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124162:124660 [2] NCCL INFO Channel 09/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32711:33204 [7] NCCL INFO Channel 03/0 : 31[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124165:124662 [5] NCCL INFO Channel 04/0 : 21[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32708:33198 [4] NCCL INFO Channel 05/0 : 28[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96894:97381 [6] NCCL INFO Channel 00/0 : 30[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364902:1365410 [0] NCCL INFO Channel 01/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96895:97382 [7] NCCL INFO Channel 00/0 : 31[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96889:97388 [1] NCCL INFO Channel 06/0 : 17[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124160:124658 [0] NCCL INFO Channel 10/0 : 0[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:124166:124665 [6] NCCL INFO Channel 03/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96890:97386 [2] NCCL INFO Channel 08/0 : 26[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96893:97384 [5] NCCL INFO Channel 12/0 : 21[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32705:33203 [1] NCCL INFO Channel 01/0 : 25[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124161:124659 [1] NCCL INFO Channel 02/0 : 17[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96894:97381 [6] NCCL INFO Channel 02/0 : 30[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364907:1365403 [5] NCCL INFO Channel 00/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32711:33204 [7] NCCL INFO Channel 05/0 : 31[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32708:33198 [4] NCCL INFO Channel 07/0 : 28[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:124167:124664 [7] NCCL INFO Channel 03/0 : 15[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124162:124660 [2] NCCL INFO Channel 11/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:124165:124662 [5] NCCL INFO Channel 06/0 : 21[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124164:124663 [4] NCCL INFO Channel 00/0 : 20[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96895:97382 [7] NCCL INFO Channel 02/0 : 31[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364902:1365410 [0] NCCL INFO Channel 02/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364907:1365403 [5] NCCL INFO Channel 01/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96890:97386 [2] NCCL INFO Channel 10/0 : 26[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96889:97388 [1] NCCL INFO Channel 08/0 : 17[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96893:97384 [5] NCCL INFO Channel 14/0 : 21[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32709:33201 [5] NCCL INFO Channel 00/0 : 29[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32704:33197 [0] NCCL INFO Channel 02/0 : 24[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:124166:124665 [6] NCCL INFO Channel 05/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96894:97381 [6] NCCL INFO Channel 04/0 : 30[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32706:33199 [2] NCCL INFO Channel 15/0 : 26[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32711:33204 [7] NCCL INFO Channel 07/0 : 31[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124160:124658 [0] NCCL INFO Channel 12/0 : 0[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96895:97382 [7] NCCL INFO Channel 04/0 : 31[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124167:124664 [7] NCCL INFO Channel 05/0 : 15[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364902:1365410 [0] NCCL INFO Channel 03/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364907:1365403 [5] NCCL INFO Channel 02/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124161:124659 [1] NCCL INFO Channel 04/0 : 17[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124162:124660 [2] NCCL INFO Channel 13/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32705:33203 [1] NCCL INFO Channel 03/0 : 25[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32708:33198 [4] NCCL INFO Channel 09/0 : 28[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:124165:124662 [5] NCCL INFO Channel 08/0 : 21[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32709:33201 [5] NCCL INFO Channel 02/0 : 29[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32704:33197 [0] NCCL INFO Channel 04/0 : 24[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:124164:124663 [4] NCCL INFO Channel 02/0 : 20[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:124166:124665 [6] NCCL INFO Channel 07/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96890:97386 [2] NCCL INFO Channel 12/0 : 26[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32711:33204 [7] NCCL INFO Channel 09/0 : 31[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124167:124664 [7] NCCL INFO Channel 07/0 : 15[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124160:124658 [0] NCCL INFO Channel 14/0 : 0[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96889:97388 [1] NCCL INFO Channel 10/0 : 17[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32708:33198 [4] NCCL INFO Channel 11/0 : 28[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:124162:124660 [2] NCCL INFO Channel 15/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96894:97381 [6] NCCL INFO Channel 08/0 : 30[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96895:97382 [7] NCCL INFO Channel 06/0 : 31[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124161:124659 [1] NCCL INFO Channel 06/0 : 17[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96890:97386 [2] NCCL INFO Channel 14/0 : 26[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364902:1365410 [0] NCCL INFO Channel 04/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:124165:124662 [5] NCCL INFO Channel 10/0 : 21[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32705:33203 [1] NCCL INFO Channel 05/0 : 25[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124166:124665 [6] NCCL INFO Channel 09/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32704:33197 [0] NCCL INFO Channel 06/0 : 24[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32709:33201 [5] NCCL INFO Channel 04/0 : 29[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32711:33204 [7] NCCL INFO Channel 11/0 : 31[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32708:33198 [4] NCCL INFO Channel 13/0 : 28[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:124164:124663 [4] NCCL INFO Channel 06/0 : 20[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:124167:124664 [7] NCCL INFO Channel 09/0 : 15[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364907:1365403 [5] NCCL INFO Channel 03/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124161:124659 [1] NCCL INFO Channel 08/0 : 17[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32706:33199 [2] NCCL INFO Channel 00/0 : 26[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32710:33202 [6] NCCL INFO Channel 05/0 : 30[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32705:33203 [1] NCCL INFO Channel 07/0 : 25[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124166:124665 [6] NCCL INFO Channel 11/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96889:97388 [1] NCCL INFO Channel 12/0 : 17[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364902:1365410 [0] NCCL INFO Channel 05/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:124165:124662 [5] NCCL INFO Channel 12/0 : 21[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96895:97382 [7] NCCL INFO Channel 08/0 : 31[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96894:97381 [6] NCCL INFO Channel 10/0 : 30[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364904:1365408 [2] NCCL INFO Channel 00/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364907:1365403 [5] NCCL INFO Channel 04/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32707:33200 [3] NCCL INFO Channel 05/0 : 27[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32704:33197 [0] NCCL INFO Channel 08/0 : 24[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:124167:124664 [7] NCCL INFO Channel 11/0 : 15[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124164:124663 [4] NCCL INFO Channel 08/0 : 20[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96890:97386 [2] NCCL INFO Channel 00/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32709:33201 [5] NCCL INFO Channel 06/0 : 29[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32708:33198 [4] NCCL INFO Channel 15/0 : 28[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32711:33204 [7] NCCL INFO Channel 13/0 : 31[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124161:124659 [1] NCCL INFO Channel 10/0 : 17[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96889:97388 [1] NCCL INFO Channel 14/0 : 17[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124166:124665 [6] NCCL INFO Channel 13/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364902:1365410 [0] NCCL INFO Channel 06/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:124165:124662 [5] NCCL INFO Channel 14/0 : 21[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364907:1365403 [5] NCCL INFO Channel 06/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364904:1365408 [2] NCCL INFO Channel 01/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32705:33203 [1] NCCL INFO Channel 11/0 : 25[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32710:33202 [6] NCCL INFO Channel 07/0 : 30[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:124167:124664 [7] NCCL INFO Channel 13/0 : 15[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364902:1365410 [0] NCCL INFO Channel 07/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32706:33199 [2] NCCL INFO Channel 04/0 : 26[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:124164:124663 [4] NCCL INFO Channel 10/0 : 20[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96895:97382 [7] NCCL INFO Channel 10/0 : 31[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96890:97386 [2] NCCL INFO Channel 01/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96894:97381 [6] NCCL INFO Channel 12/0 : 30[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:124166:124665 [6] NCCL INFO Channel 15/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:124161:124659 [1] NCCL INFO Channel 12/0 : 17[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124160:124658 [0] NCCL INFO Channel 01/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96895:97382 [7] NCCL INFO Channel 12/0 : 31[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96890:97386 [2] NCCL INFO Channel 03/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96894:97381 [6] NCCL INFO Channel 14/0 : 30[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364907:1365403 [5] NCCL INFO Channel 07/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364904:1365408 [2] NCCL INFO Channel 03/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364902:1365410 [0] NCCL INFO Channel 09/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32704:33197 [0] NCCL INFO Channel 10/0 : 24[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32709:33201 [5] NCCL INFO Channel 08/0 : 29[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124165:124662 [5] NCCL INFO Channel 00/0 : 5[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32707:33200 [3] NCCL INFO Channel 07/0 : 27[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32710:33202 [6] NCCL INFO Channel 09/0 : 30[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96895:97382 [7] NCCL INFO Channel 14/0 : 31[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96890:97386 [2] NCCL INFO Channel 04/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96894:97381 [6] NCCL INFO Channel 00/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96888:97387 [0] NCCL INFO Channel 02/0 : 24[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364907:1365403 [5] NCCL INFO Channel 08/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124164:124663 [4] NCCL INFO Channel 12/0 : 20[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:124161:124659 [1] NCCL INFO Channel 14/0 : 17[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96895:97382 [7] NCCL INFO Channel 00/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32706:33199 [2] NCCL INFO Channel 06/0 : 26[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32705:33203 [1] NCCL INFO Channel 13/0 : 25[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124160:124658 [0] NCCL INFO Channel 03/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:124165:124662 [5] NCCL INFO Channel 02/0 : 5[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96890:97386 [2] NCCL INFO Channel 05/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32704:33197 [0] NCCL INFO Channel 12/0 : 24[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32709:33201 [5] NCCL INFO Channel 10/0 : 29[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32710:33202 [6] NCCL INFO Channel 11/0 : 30[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364904:1365408 [2] NCCL INFO Channel 04/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364902:1365410 [0] NCCL INFO Channel 10/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364909:1365405 [7] NCCL INFO Channel 00/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364906:1365407 [4] NCCL INFO Channel 00/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96894:97381 [6] NCCL INFO Channel 01/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96888:97387 [0] NCCL INFO Channel 04/0 : 24[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364907:1365403 [5] NCCL INFO Channel 09/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96895:97382 [7] NCCL INFO Channel 01/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32708:33198 [4] NCCL INFO Channel 00/0 : 28[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32707:33200 [3] NCCL INFO Channel 09/0 : 27[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32711:33204 [7] NCCL INFO Channel 00/0 : 31[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96890:97386 [2] NCCL INFO Channel 06/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32705:33203 [1] NCCL INFO Channel 15/0 : 25[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32706:33199 [2] NCCL INFO Channel 08/0 : 26[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:124164:124663 [4] NCCL INFO Channel 14/0 : 20[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:124161:124659 [1] NCCL INFO Channel 00/0 : 1[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124165:124662 [5] NCCL INFO Channel 04/0 : 5[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124160:124658 [0] NCCL INFO Channel 05/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364904:1365408 [2] NCCL INFO Channel 05/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364909:1365405 [7] NCCL INFO Channel 01/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96894:97381 [6] NCCL INFO Channel 02/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96888:97387 [0] NCCL INFO Channel 06/0 : 24[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364906:1365407 [4] NCCL INFO Channel 01/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32704:33197 [0] NCCL INFO Channel 14/0 : 24[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32709:33201 [5] NCCL INFO Channel 12/0 : 29[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32710:33202 [6] NCCL INFO Channel 13/0 : 30[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96895:97382 [7] NCCL INFO Channel 02/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96890:97386 [2] NCCL INFO Channel 07/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:124161:124659 [1] NCCL INFO Channel 02/0 : 1[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364907:1365403 [5] NCCL INFO Channel 10/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364904:1365408 [2] NCCL INFO Channel 06/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32707:33200 [3] NCCL INFO Channel 13/0 : 27[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32708:33198 [4] NCCL INFO Channel 02/0 : 28[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32711:33204 [7] NCCL INFO Channel 02/0 : 31[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124164:124663 [4] NCCL INFO Channel 00/0 : 4[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:124165:124662 [5] NCCL INFO Channel 06/0 : 5[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364902:1365410 [0] NCCL INFO Channel 11/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32706:33199 [2] NCCL INFO Channel 10/0 : 26[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32709:33201 [5] NCCL INFO Channel 14/0 : 29[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32710:33202 [6] NCCL INFO Channel 15/0 : 30[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:124160:124658 [0] NCCL INFO Channel 07/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96888:97387 [0] NCCL INFO Channel 08/0 : 24[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96894:97381 [6] NCCL INFO Channel 03/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364909:1365405 [7] NCCL INFO Channel 02/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96895:97382 [7] NCCL INFO Channel 03/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364906:1365407 [4] NCCL INFO Channel 02/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96890:97386 [2] NCCL INFO Channel 08/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364907:1365403 [5] NCCL INFO Channel 11/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364904:1365408 [2] NCCL INFO Channel 07/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32707:33200 [3] NCCL INFO Channel 15/0 : 27[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32711:33204 [7] NCCL INFO Channel 04/0 : 31[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32708:33198 [4] NCCL INFO Channel 06/0 : 28[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32706:33199 [2] NCCL INFO Channel 12/0 : 26[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:124165:124662 [5] NCCL INFO Channel 08/0 : 5[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124161:124659 [1] NCCL INFO Channel 04/0 : 1[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96888:97387 [0] NCCL INFO Channel 10/0 : 24[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364902:1365410 [0] NCCL INFO Channel 12/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:124164:124663 [4] NCCL INFO Channel 02/0 : 4[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96894:97381 [6] NCCL INFO Channel 04/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96895:97382 [7] NCCL INFO Channel 04/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96890:97386 [2] NCCL INFO Channel 09/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364909:1365405 [7] NCCL INFO Channel 03/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32711:33204 [7] NCCL INFO Channel 06/0 : 31[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32708:33198 [4] NCCL INFO Channel 08/0 : 28[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:124160:124658 [0] NCCL INFO Channel 09/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364906:1365407 [4] NCCL INFO Channel 03/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32706:33199 [2] NCCL INFO Channel 14/0 : 26[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:124163:124661 [3] NCCL INFO Channel 00/0 : 19[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364907:1365403 [5] NCCL INFO Channel 12/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364904:1365408 [2] NCCL INFO Channel 08/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364902:1365410 [0] NCCL INFO Channel 13/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96888:97387 [0] NCCL INFO Channel 12/0 : 24[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:124165:124662 [5] NCCL INFO Channel 10/0 : 5[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32705:33203 [1] NCCL INFO Channel 00/0 : 25[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124161:124659 [1] NCCL INFO Channel 06/0 : 1[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364909:1365405 [7] NCCL INFO Channel 04/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124164:124663 [4] NCCL INFO Channel 06/0 : 4[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364906:1365407 [4] NCCL INFO Channel 05/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364903:1365409 [1] NCCL INFO Channel 00/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124160:124658 [0] NCCL INFO Channel 11/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:124163:124661 [3] NCCL INFO Channel 02/0 : 19[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96895:97382 [7] NCCL INFO Channel 05/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96890:97386 [2] NCCL INFO Channel 11/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96894:97381 [6] NCCL INFO Channel 05/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364907:1365403 [5] NCCL INFO Channel 14/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96888:97387 [0] NCCL INFO Channel 14/0 : 24[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364902:1365410 [0] NCCL INFO Channel 14/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32708:33198 [4] NCCL INFO Channel 10/0 : 28[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32711:33204 [7] NCCL INFO Channel 08/0 : 31[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124165:124662 [5] NCCL INFO Channel 12/0 : 5[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364904:1365408 [2] NCCL INFO Channel 09/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32710:33202 [6] NCCL INFO Channel 00/0 : 30[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:124161:124659 [1] NCCL INFO Channel 08/0 : 1[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364909:1365405 [7] NCCL INFO Channel 05/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32707:33200 [3] NCCL INFO Channel 00/0 : 27[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32705:33203 [1] NCCL INFO Channel 02/0 : 25[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96895:97382 [7] NCCL INFO Channel 06/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124164:124663 [4] NCCL INFO Channel 08/0 : 4[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:124160:124658 [0] NCCL INFO Channel 13/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:124163:124661 [3] NCCL INFO Channel 04/0 : 19[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:124165:124662 [5] NCCL INFO Channel 14/0 : 5[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364903:1365409 [1] NCCL INFO Channel 02/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364906:1365407 [4] NCCL INFO Channel 06/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32708:33198 [4] NCCL INFO Channel 12/0 : 28[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:124161:124659 [1] NCCL INFO Channel 10/0 : 1[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96890:97386 [2] NCCL INFO Channel 12/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96888:97387 [0] NCCL INFO Channel 01/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96894:97381 [6] NCCL INFO Channel 07/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364908:1365406 [6] NCCL INFO Channel 00/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32711:33204 [7] NCCL INFO Channel 10/0 : 31[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124164:124663 [4] NCCL INFO Channel 10/0 : 4[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364905:1365404 [3] NCCL INFO Channel 00/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364907:1365403 [5] NCCL INFO Channel 15/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32710:33202 [6] NCCL INFO Channel 02/0 : 30[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364902:1365410 [0] NCCL INFO Channel 15/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32705:33203 [1] NCCL INFO Channel 04/0 : 25[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124160:124658 [0] NCCL INFO Channel 15/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364904:1365408 [2] NCCL INFO Channel 11/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32707:33200 [3] NCCL INFO Channel 02/0 : 27[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:124163:124661 [3] NCCL INFO Channel 06/0 : 19[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96895:97382 [7] NCCL INFO Channel 08/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364909:1365405 [7] NCCL INFO Channel 06/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124161:124659 [1] NCCL INFO Channel 12/0 : 1[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96888:97387 [0] NCCL INFO Channel 02/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96890:97386 [2] NCCL INFO Channel 13/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96894:97381 [6] NCCL INFO Channel 08/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:124164:124663 [4] NCCL INFO Channel 12/0 : 4[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32708:33198 [4] NCCL INFO Channel 14/0 : 28[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32711:33204 [7] NCCL INFO Channel 12/0 : 31[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32710:33202 [6] NCCL INFO Channel 04/0 : 30[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:124161:124659 [1] NCCL INFO Channel 14/0 : 1[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96895:97382 [7] NCCL INFO Channel 09/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364906:1365407 [4] NCCL INFO Channel 07/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364903:1365409 [1] NCCL INFO Channel 03/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32705:33203 [1] NCCL INFO Channel 06/0 : 25[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124163:124661 [3] NCCL INFO Channel 08/0 : 19[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364908:1365406 [6] NCCL INFO Channel 01/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32707:33200 [3] NCCL INFO Channel 04/0 : 27[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:124164:124663 [4] NCCL INFO Channel 14/0 : 4[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364905:1365404 [3] NCCL INFO Channel 01/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32711:33204 [7] NCCL INFO Channel 14/0 : 31[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96890:97386 [2] NCCL INFO Channel 14/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96888:97387 [0] NCCL INFO Channel 03/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96895:97382 [7] NCCL INFO Channel 10/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96894:97381 [6] NCCL INFO Channel 09/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364902:1365410 [0] NCCL INFO Channel 01/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32710:33202 [6] NCCL INFO Channel 08/0 : 30[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364904:1365408 [2] NCCL INFO Channel 12/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364907:1365403 [5] NCCL INFO Channel 01/0 : 13[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124165:124662 [5] NCCL INFO Channel 01/0 : 13[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124163:124661 [3] NCCL INFO Channel 10/0 : 19[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32705:33203 [1] NCCL INFO Channel 08/0 : 25[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32707:33200 [3] NCCL INFO Channel 06/0 : 27[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96890:97386 [2] NCCL INFO Channel 15/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364909:1365405 [7] NCCL INFO Channel 08/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32710:33202 [6] NCCL INFO Channel 10/0 : 30[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364908:1365406 [6] NCCL INFO Channel 02/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32705:33203 [1] NCCL INFO Channel 10/0 : 25[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32707:33200 [3] NCCL INFO Channel 08/0 : 27[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:124165:124662 [5] NCCL INFO Channel 03/0 : 13[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364903:1365409 [1] NCCL INFO Channel 04/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364906:1365407 [4] NCCL INFO Channel 08/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32710:33202 [6] NCCL INFO Channel 12/0 : 30[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364905:1365404 [3] NCCL INFO Channel 02/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364902:1365410 [0] NCCL INFO Channel 03/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364904:1365408 [2] NCCL INFO Channel 13/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96895:97382 [7] NCCL INFO Channel 11/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96894:97381 [6] NCCL INFO Channel 10/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96893:97384 [5] NCCL INFO Channel 00/0 : 29[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364907:1365403 [5] NCCL INFO Channel 03/0 : 13[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124163:124661 [3] NCCL INFO Channel 12/0 : 19[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96888:97387 [0] NCCL INFO Channel 04/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32707:33200 [3] NCCL INFO Channel 10/0 : 27[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32705:33203 [1] NCCL INFO Channel 12/0 : 25[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32710:33202 [6] NCCL INFO Channel 14/0 : 30[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:124165:124662 [5] NCCL INFO Channel 05/0 : 13[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124161:124659 [1] NCCL INFO Channel 01/0 : 9[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32707:33200 [3] NCCL INFO Channel 12/0 : 27[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32705:33203 [1] NCCL INFO Channel 14/0 : 25[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364909:1365405 [7] NCCL INFO Channel 09/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96895:97382 [7] NCCL INFO Channel 12/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364908:1365406 [6] NCCL INFO Channel 03/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:124164:124663 [4] NCCL INFO Channel 01/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96894:97381 [6] NCCL INFO Channel 11/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364903:1365409 [1] NCCL INFO Channel 05/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364902:1365410 [0] NCCL INFO Channel 05/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96893:97384 [5] NCCL INFO Channel 02/0 : 29[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364906:1365407 [4] NCCL INFO Channel 09/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364905:1365404 [3] NCCL INFO Channel 04/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96888:97387 [0] NCCL INFO Channel 05/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364907:1365403 [5] NCCL INFO Channel 05/0 : 13[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364904:1365408 [2] NCCL INFO Channel 14/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32707:33200 [3] NCCL INFO Channel 14/0 : 27[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:124163:124661 [3] NCCL INFO Channel 14/0 : 19[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364909:1365405 [7] NCCL INFO Channel 10/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124165:124662 [5] NCCL INFO Channel 07/0 : 13[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364908:1365406 [6] NCCL INFO Channel 04/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:124161:124659 [1] NCCL INFO Channel 03/0 : 9[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124164:124663 [4] NCCL INFO Channel 03/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96889:97388 [1] NCCL INFO Channel 00/0 : 25[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96895:97382 [7] NCCL INFO Channel 13/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96894:97381 [6] NCCL INFO Channel 12/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96892:97383 [4] NCCL INFO Channel 00/0 : 28[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96893:97384 [5] NCCL INFO Channel 04/0 : 29[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96888:97387 [0] NCCL INFO Channel 06/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:124163:124661 [3] NCCL INFO Channel 00/0 : 3[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364903:1365409 [1] NCCL INFO Channel 06/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124165:124662 [5] NCCL INFO Channel 09/0 : 13[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364902:1365410 [0] NCCL INFO Channel 07/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364906:1365407 [4] NCCL INFO Channel 10/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96895:97382 [7] NCCL INFO Channel 14/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364905:1365404 [3] NCCL INFO Channel 05/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:124161:124659 [1] NCCL INFO Channel 05/0 : 9[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124164:124663 [4] NCCL INFO Channel 05/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96889:97388 [1] NCCL INFO Channel 02/0 : 25[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364904:1365408 [2] NCCL INFO Channel 15/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96894:97381 [6] NCCL INFO Channel 13/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364907:1365403 [5] NCCL INFO Channel 07/0 : 13[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364909:1365405 [7] NCCL INFO Channel 11/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124163:124661 [3] NCCL INFO Channel 02/0 : 3[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364908:1365406 [6] NCCL INFO Channel 05/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:124165:124662 [5] NCCL INFO Channel 11/0 : 13[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364903:1365409 [1] NCCL INFO Channel 07/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364902:1365410 [0] NCCL INFO Channel 09/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96888:97387 [0] NCCL INFO Channel 07/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96892:97383 [4] NCCL INFO Channel 02/0 : 28[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96893:97384 [5] NCCL INFO Channel 06/0 : 29[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124164:124663 [4] NCCL INFO Channel 07/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:124161:124659 [1] NCCL INFO Channel 07/0 : 9[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96889:97388 [1] NCCL INFO Channel 04/0 : 25[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124163:124661 [3] NCCL INFO Channel 04/0 : 3[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96894:97381 [6] NCCL INFO Channel 15/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364905:1365404 [3] NCCL INFO Channel 06/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364906:1365407 [4] NCCL INFO Channel 11/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:124165:124662 [5] NCCL INFO Channel 15/0 : 13[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364904:1365408 [2] NCCL INFO Channel 01/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364907:1365403 [5] NCCL INFO Channel 09/0 : 13[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364909:1365405 [7] NCCL INFO Channel 12/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96888:97387 [0] NCCL INFO Channel 09/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364903:1365409 [1] NCCL INFO Channel 08/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364908:1365406 [6] NCCL INFO Channel 07/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:124164:124663 [4] NCCL INFO Channel 09/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96893:97384 [5] NCCL INFO Channel 08/0 : 29[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96892:97383 [4] NCCL INFO Channel 06/0 : 28[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364902:1365410 [0] NCCL INFO Channel 11/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:124161:124659 [1] NCCL INFO Channel 11/0 : 9[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96889:97388 [1] NCCL INFO Channel 06/0 : 25[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364905:1365404 [3] NCCL INFO Channel 07/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:124163:124661 [3] NCCL INFO Channel 06/0 : 3[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96888:97387 [0] NCCL INFO Channel 10/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364906:1365407 [4] NCCL INFO Channel 13/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:124164:124663 [4] NCCL INFO Channel 11/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96893:97384 [5] NCCL INFO Channel 10/0 : 29[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96892:97383 [4] NCCL INFO Channel 08/0 : 28[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364904:1365408 [2] NCCL INFO Channel 03/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:124163:124661 [3] NCCL INFO Channel 08/0 : 3[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:124161:124659 [1] NCCL INFO Channel 13/0 : 9[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96889:97388 [1] NCCL INFO Channel 08/0 : 25[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364907:1365403 [5] NCCL INFO Channel 11/0 : 13[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96888:97387 [0] NCCL INFO Channel 11/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364909:1365405 [7] NCCL INFO Channel 13/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364903:1365409 [1] NCCL INFO Channel 10/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364902:1365410 [0] NCCL INFO Channel 13/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:124164:124663 [4] NCCL INFO Channel 13/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364908:1365406 [6] NCCL INFO Channel 08/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364905:1365404 [3] NCCL INFO Channel 08/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:124163:124661 [3] NCCL INFO Channel 10/0 : 3[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:124161:124659 [1] NCCL INFO Channel 15/0 : 9[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364906:1365407 [4] NCCL INFO Channel 14/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364904:1365408 [2] NCCL INFO Channel 05/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96893:97384 [5] NCCL INFO Channel 12/0 : 29[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364907:1365403 [5] NCCL INFO Channel 15/0 : 13[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96892:97383 [4] NCCL INFO Channel 10/0 : 28[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364909:1365405 [7] NCCL INFO Channel 14/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96888:97387 [0] NCCL INFO Channel 12/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:124164:124663 [4] NCCL INFO Channel 15/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96889:97388 [1] NCCL INFO Channel 10/0 : 25[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96893:97384 [5] NCCL INFO Channel 14/0 : 29[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124163:124661 [3] NCCL INFO Channel 12/0 : 3[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96892:97383 [4] NCCL INFO Channel 12/0 : 28[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96888:97387 [0] NCCL INFO Channel 13/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364902:1365410 [0] NCCL INFO Channel 15/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364903:1365409 [1] NCCL INFO Channel 11/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364908:1365406 [6] NCCL INFO Channel 09/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364904:1365408 [2] NCCL INFO Channel 07/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364906:1365407 [4] NCCL INFO Channel 15/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364909:1365405 [7] NCCL INFO Channel 01/0 : 15[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364905:1365404 [3] NCCL INFO Channel 09/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364903:1365409 [1] NCCL INFO Channel 12/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364908:1365406 [6] NCCL INFO Channel 10/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:124163:124661 [3] NCCL INFO Channel 14/0 : 3[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96889:97388 [1] NCCL INFO Channel 12/0 : 25[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96893:97384 [5] NCCL INFO Channel 00/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96888:97387 [0] NCCL INFO Channel 14/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96892:97383 [4] NCCL INFO Channel 14/0 : 28[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364909:1365405 [7] NCCL INFO Channel 03/0 : 15[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364904:1365408 [2] NCCL INFO Channel 09/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364906:1365407 [4] NCCL INFO Channel 01/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364903:1365409 [1] NCCL INFO Channel 13/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364905:1365404 [3] NCCL INFO Channel 10/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364908:1365406 [6] NCCL INFO Channel 11/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96889:97388 [1] NCCL INFO Channel 14/0 : 25[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96893:97384 [5] NCCL INFO Channel 01/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96888:97387 [0] NCCL INFO Channel 15/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96892:97383 [4] NCCL INFO Channel 00/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364909:1365405 [7] NCCL INFO Channel 05/0 : 15[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96893:97384 [5] NCCL INFO Channel 02/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96889:97388 [1] NCCL INFO Channel 00/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96892:97383 [4] NCCL INFO Channel 01/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96893:97384 [5] NCCL INFO Channel 03/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96889:97388 [1] NCCL INFO Channel 02/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364904:1365408 [2] NCCL INFO Channel 11/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364906:1365407 [4] NCCL INFO Channel 03/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364903:1365409 [1] NCCL INFO Channel 14/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364905:1365404 [3] NCCL INFO Channel 12/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364908:1365406 [6] NCCL INFO Channel 12/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:124163:124661 [3] NCCL INFO Channel 01/0 : 11[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96892:97383 [4] NCCL INFO Channel 02/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96893:97384 [5] NCCL INFO Channel 04/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364909:1365405 [7] NCCL INFO Channel 07/0 : 15[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96889:97388 [1] NCCL INFO Channel 03/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364904:1365408 [2] NCCL INFO Channel 13/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:124163:124661 [3] NCCL INFO Channel 03/0 : 11[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96891:97385 [3] NCCL INFO Channel 00/0 : 27[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96892:97383 [4] NCCL INFO Channel 03/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364906:1365407 [4] NCCL INFO Channel 05/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96893:97384 [5] NCCL INFO Channel 06/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364903:1365409 [1] NCCL INFO Channel 15/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124163:124661 [3] NCCL INFO Channel 05/0 : 11[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96889:97388 [1] NCCL INFO Channel 04/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364905:1365404 [3] NCCL INFO Channel 13/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364908:1365406 [6] NCCL INFO Channel 13/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364909:1365405 [7] NCCL INFO Channel 09/0 : 15[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124163:124661 [3] NCCL INFO Channel 07/0 : 11[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96891:97385 [3] NCCL INFO Channel 02/0 : 27[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96892:97383 [4] NCCL INFO Channel 05/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364906:1365407 [4] NCCL INFO Channel 07/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96893:97384 [5] NCCL INFO Channel 07/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364903:1365409 [1] NCCL INFO Channel 01/0 : 9[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96889:97388 [1] NCCL INFO Channel 05/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364908:1365406 [6] NCCL INFO Channel 15/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364905:1365404 [3] NCCL INFO Channel 14/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364909:1365405 [7] NCCL INFO Channel 11/0 : 15[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124163:124661 [3] NCCL INFO Channel 09/0 : 11[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364904:1365408 [2] NCCL INFO Channel 15/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96891:97385 [3] NCCL INFO Channel 04/0 : 27[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557268:1557764 [2] NCCL INFO Connected all trees
+ip-26-0-161-103:96893:97384 [5] NCCL INFO Channel 08/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96892:97383 [4] NCCL INFO Channel 06/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96889:97388 [1] NCCL INFO Channel 06/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96891:97385 [3] NCCL INFO Channel 06/0 : 27[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364903:1365409 [1] NCCL INFO Channel 03/0 : 9[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364906:1365407 [4] NCCL INFO Channel 09/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:124163:124661 [3] NCCL INFO Channel 13/0 : 11[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364908:1365406 [6] NCCL INFO Channel 01/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364905:1365404 [3] NCCL INFO Channel 15/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364909:1365405 [7] NCCL INFO Channel 13/0 : 15[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364903:1365409 [1] NCCL INFO Channel 05/0 : 9[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364906:1365407 [4] NCCL INFO Channel 11/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722049:2722547 [4] NCCL INFO Connected all trees
+ip-26-0-161-103:96893:97384 [5] NCCL INFO Channel 09/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96892:97383 [4] NCCL INFO Channel 07/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:124163:124661 [3] NCCL INFO Channel 15/0 : 11[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96889:97388 [1] NCCL INFO Channel 07/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96891:97385 [3] NCCL INFO Channel 08/0 : 27[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364908:1365406 [6] NCCL INFO Channel 03/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364905:1365404 [3] NCCL INFO Channel 01/0 : 11[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364903:1365409 [1] NCCL INFO Channel 07/0 : 9[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364906:1365407 [4] NCCL INFO Channel 13/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364908:1365406 [6] NCCL INFO Channel 05/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96893:97384 [5] NCCL INFO Channel 10/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96892:97383 [4] NCCL INFO Channel 08/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96889:97388 [1] NCCL INFO Channel 08/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96891:97385 [3] NCCL INFO Channel 10/0 : 27[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96893:97384 [5] NCCL INFO Channel 11/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364905:1365404 [3] NCCL INFO Channel 03/0 : 11[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96892:97383 [4] NCCL INFO Channel 09/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364903:1365409 [1] NCCL INFO Channel 11/0 : 9[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364906:1365407 [4] NCCL INFO Channel 15/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364908:1365406 [6] NCCL INFO Channel 07/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96891:97385 [3] NCCL INFO Channel 12/0 : 27[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96889:97388 [1] NCCL INFO Channel 10/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96893:97384 [5] NCCL INFO Channel 12/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96892:97383 [4] NCCL INFO Channel 10/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364905:1365404 [3] NCCL INFO Channel 05/0 : 11[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364903:1365409 [1] NCCL INFO Channel 13/0 : 9[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364908:1365406 [6] NCCL INFO Channel 09/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96891:97385 [3] NCCL INFO Channel 14/0 : 27[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96893:97384 [5] NCCL INFO Channel 14/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96892:97383 [4] NCCL INFO Channel 11/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96891:97385 [3] NCCL INFO Channel 00/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364905:1365404 [3] NCCL INFO Channel 07/0 : 11[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364903:1365409 [1] NCCL INFO Channel 15/0 : 9[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364908:1365406 [6] NCCL INFO Channel 11/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96889:97388 [1] NCCL INFO Channel 11/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96893:97384 [5] NCCL INFO Channel 15/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364905:1365404 [3] NCCL INFO Channel 09/0 : 11[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96892:97383 [4] NCCL INFO Channel 13/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364908:1365406 [6] NCCL INFO Channel 13/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557268:1557764 [2] NCCL INFO NVLS comm 0xab81880 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-103:96891:97385 [3] NCCL INFO Channel 01/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96889:97388 [1] NCCL INFO Channel 12/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364905:1365404 [3] NCCL INFO Channel 13/0 : 11[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364908:1365406 [6] NCCL INFO Channel 15/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96892:97383 [4] NCCL INFO Channel 14/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722049:2722547 [4] NCCL INFO NVLS comm 0xa856380 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-103:96892:97383 [4] NCCL INFO Channel 15/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364905:1365404 [3] NCCL INFO Channel 15/0 : 11[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114514:115011 [2] NCCL INFO Channel 07/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96889:97388 [1] NCCL INFO Channel 13/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96891:97385 [3] NCCL INFO Channel 02/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114514:115011 [2] NCCL INFO Channel 09/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96889:97388 [1] NCCL INFO Channel 14/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96891:97385 [3] NCCL INFO Channel 04/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96891:97385 [3] NCCL INFO Channel 05/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96889:97388 [1] NCCL INFO Channel 15/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114514:115011 [2] NCCL INFO Channel 11/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96891:97385 [3] NCCL INFO Channel 06/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114514:115011 [2] NCCL INFO Channel 13/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96891:97385 [3] NCCL INFO Channel 07/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96891:97385 [3] NCCL INFO Channel 08/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114514:115011 [2] NCCL INFO Channel 15/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96891:97385 [3] NCCL INFO Channel 09/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96891:97385 [3] NCCL INFO Channel 10/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96891:97385 [3] NCCL INFO Channel 12/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96891:97385 [3] NCCL INFO Channel 13/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96891:97385 [3] NCCL INFO Channel 14/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96891:97385 [3] NCCL INFO Channel 15/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815966:2816474 [3] NCCL INFO Connected all trees
+ip-26-0-161-178:2815966:2816474 [3] NCCL INFO NVLS comm 0xac53fa0 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-59:1557266:1557759 [0] NCCL INFO Connected all trees
+ip-26-0-165-59:1557266:1557759 [0] NCCL INFO NVLS comm 0xc14cf00 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-168-52:2722045:2722544 [0] NCCL INFO Connected all trees
+ip-26-0-168-52:2722045:2722544 [0] NCCL INFO NVLS comm 0x9b08d30 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-59:1557273:1557760 [7] NCCL INFO Connected all trees
+ip-26-0-165-59:1557273:1557760 [7] NCCL INFO NVLS comm 0xb5cc7e0 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-168-52:2722046:2722541 [1] NCCL INFO Connected all trees
+ip-26-0-168-52:2722046:2722541 [1] NCCL INFO NVLS comm 0xadda970 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-168-52:2722052:2722546 [7] NCCL INFO Connected all trees
+ip-26-0-168-52:2722052:2722546 [7] NCCL INFO NVLS comm 0xb6b79f0 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-59:1557272:1557766 [6] NCCL INFO Connected all trees
+ip-26-0-168-52:2722051:2722543 [6] NCCL INFO Connected all trees
+ip-26-0-165-59:1557272:1557766 [6] NCCL INFO NVLS comm 0xa9e2270 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-168-52:2722051:2722543 [6] NCCL INFO NVLS comm 0xb298c00 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-59:1557269:1557765 [3] NCCL INFO Connected all trees
+ip-26-0-165-59:1557269:1557765 [3] NCCL INFO NVLS comm 0xb4cbdc0 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-178:2815967:2816473 [4] NCCL INFO Connected all trees
+ip-26-0-165-59:1557267:1557761 [1] NCCL INFO Connected all trees
+ip-26-0-161-178:2815967:2816473 [4] NCCL INFO NVLS comm 0xa848d80 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-59:1557267:1557761 [1] NCCL INFO NVLS comm 0x9e76b70 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-178:2815964:2816476 [1] NCCL INFO Connected all trees
+ip-26-0-161-178:2815964:2816476 [1] NCCL INFO NVLS comm 0x9f47930 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-138:96563:97086 [0] NCCL INFO Connected all trees
+ip-26-0-161-138:96563:97086 [0] NCCL INFO NVLS comm 0xb166280 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-138:96570:97093 [7] NCCL INFO Connected all trees
+ip-26-0-161-178:2815963:2816477 [0] NCCL INFO Connected all trees
+ip-26-0-161-178:2815968:2816472 [5] NCCL INFO Connected all trees
+ip-26-0-161-138:96570:97093 [7] NCCL INFO NVLS comm 0xb873da0 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-178:2815963:2816477 [0] NCCL INFO NVLS comm 0xab2e7e0 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-178:2815968:2816472 [5] NCCL INFO NVLS comm 0xb83d840 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-178:2815970:2816470 [7] NCCL INFO Connected all trees
+ip-26-0-161-178:2815970:2816470 [7] NCCL INFO NVLS comm 0xa873aa0 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-178:2815965:2816475 [2] NCCL INFO Connected all trees
+ip-26-0-161-178:2815965:2816475 [2] NCCL INFO NVLS comm 0xa383240 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-138:96565:97089 [2] NCCL INFO Connected all trees
+ip-26-0-161-138:96565:97089 [2] NCCL INFO NVLS comm 0xb9b8c10 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-138:96566:97087 [3] NCCL INFO Connected all trees
+ip-26-0-161-138:96566:97087 [3] NCCL INFO NVLS comm 0xb531f70 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-168-52:2722048:2722545 [3] NCCL INFO Connected all trees
+ip-26-0-168-52:2722048:2722545 [3] NCCL INFO NVLS comm 0xa691040 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-168-52:2722047:2722542 [2] NCCL INFO Connected all trees
+ip-26-0-168-52:2722047:2722542 [2] NCCL INFO NVLS comm 0xa9fbb80 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-138:96567:97090 [4] NCCL INFO Connected all trees
+ip-26-0-161-138:96567:97090 [4] NCCL INFO NVLS comm 0xa01fba0 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-178:2815969:2816471 [6] NCCL INFO Connected all trees
+ip-26-0-161-178:2815969:2816471 [6] NCCL INFO NVLS comm 0xb206db0 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-138:96569:97092 [6] NCCL INFO Connected all trees
+ip-26-0-161-138:96569:97092 [6] NCCL INFO NVLS comm 0xa638af0 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-168-52:2722050:2722548 [5] NCCL INFO Connected all trees
+ip-26-0-168-52:2722050:2722548 [5] NCCL INFO NVLS comm 0xa6c0c80 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-168-30:235651:236141 [3] NCCL INFO Connected all trees
+ip-26-0-168-30:235651:236141 [3] NCCL INFO NVLS comm 0xa4ad700 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-168-30:235653:236140 [5] NCCL INFO Connected all trees
+ip-26-0-168-30:235653:236140 [5] NCCL INFO NVLS comm 0xa6da390 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-142:62886:63378 [5] NCCL INFO Connected all trees
+ip-26-0-161-142:62886:63378 [5] NCCL INFO NVLS comm 0xb59e210 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-153:2636089:2636600 [5] NCCL INFO Connected all trees
+ip-26-0-161-153:2636089:2636600 [5] NCCL INFO NVLS comm 0x9edfc90 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-168-30:235649:236143 [1] NCCL INFO Connected all trees
+ip-26-0-168-34:272151:272646 [1] NCCL INFO Connected all trees
+ip-26-0-168-34:272151:272646 [1] NCCL INFO NVLS comm 0xb4d5a80 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-168-30:235649:236143 [1] NCCL INFO NVLS comm 0xb1b3870 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-59:1557271:1557762 [5] NCCL INFO Channel 01/0 : 5[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557271:1557762 [5] NCCL INFO Channel 03/0 : 5[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557267:1557761 [1] NCCL INFO Channel 01/0 : 1[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557271:1557762 [5] NCCL INFO Channel 05/0 : 5[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557269:1557765 [3] NCCL INFO Channel 01/0 : 3[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557267:1557761 [1] NCCL INFO Channel 03/0 : 1[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557271:1557762 [5] NCCL INFO Channel 07/0 : 5[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557272:1557766 [6] NCCL INFO Channel 01/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557273:1557760 [7] NCCL INFO Channel 01/0 : 7[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557267:1557761 [1] NCCL INFO Channel 05/0 : 1[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557269:1557765 [3] NCCL INFO Channel 03/0 : 3[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557266:1557759 [0] NCCL INFO Channel 01/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557271:1557762 [5] NCCL INFO Channel 09/0 : 5[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557268:1557764 [2] NCCL INFO Channel 01/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557273:1557760 [7] NCCL INFO Channel 03/0 : 7[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557267:1557761 [1] NCCL INFO Channel 07/0 : 1[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557272:1557766 [6] NCCL INFO Channel 03/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557270:1557763 [4] NCCL INFO Channel 01/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557266:1557759 [0] NCCL INFO Channel 03/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557269:1557765 [3] NCCL INFO Channel 05/0 : 3[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557271:1557762 [5] NCCL INFO Channel 11/0 : 5[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557268:1557764 [2] NCCL INFO Channel 03/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557273:1557760 [7] NCCL INFO Channel 05/0 : 7[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557267:1557761 [1] NCCL INFO Channel 11/0 : 1[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557266:1557759 [0] NCCL INFO Channel 05/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557270:1557763 [4] NCCL INFO Channel 03/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557272:1557766 [6] NCCL INFO Channel 05/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557269:1557765 [3] NCCL INFO Channel 07/0 : 3[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557271:1557762 [5] NCCL INFO Channel 15/0 : 5[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557268:1557764 [2] NCCL INFO Channel 05/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557273:1557760 [7] NCCL INFO Channel 07/0 : 7[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557267:1557761 [1] NCCL INFO Channel 13/0 : 1[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557270:1557763 [4] NCCL INFO Channel 05/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557272:1557766 [6] NCCL INFO Channel 07/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557269:1557765 [3] NCCL INFO Channel 09/0 : 3[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557266:1557759 [0] NCCL INFO Channel 07/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557268:1557764 [2] NCCL INFO Channel 07/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557267:1557761 [1] NCCL INFO Channel 15/0 : 1[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557273:1557760 [7] NCCL INFO Channel 09/0 : 7[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557270:1557763 [4] NCCL INFO Channel 07/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557272:1557766 [6] NCCL INFO Channel 09/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557269:1557765 [3] NCCL INFO Channel 13/0 : 3[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557268:1557764 [2] NCCL INFO Channel 09/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557266:1557759 [0] NCCL INFO Channel 09/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557273:1557760 [7] NCCL INFO Channel 11/0 : 7[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557270:1557763 [4] NCCL INFO Channel 09/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557272:1557766 [6] NCCL INFO Channel 11/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557269:1557765 [3] NCCL INFO Channel 15/0 : 3[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557268:1557764 [2] NCCL INFO Channel 11/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557266:1557759 [0] NCCL INFO Channel 11/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557273:1557760 [7] NCCL INFO Channel 13/0 : 7[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557270:1557763 [4] NCCL INFO Channel 11/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557272:1557766 [6] NCCL INFO Channel 13/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235655:236142 [7] NCCL INFO Connected all trees
+ip-26-0-165-59:1557268:1557764 [2] NCCL INFO Channel 13/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557266:1557759 [0] NCCL INFO Channel 13/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557270:1557763 [4] NCCL INFO Channel 13/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557272:1557766 [6] NCCL INFO Channel 15/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557268:1557764 [2] NCCL INFO Channel 15/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557266:1557759 [0] NCCL INFO Channel 15/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557270:1557763 [4] NCCL INFO Channel 15/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235655:236142 [7] NCCL INFO NVLS comm 0xb48d9f0 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-168-34:272157:272649 [7] NCCL INFO Connected all trees
+ip-26-0-161-153:2636088:2636601 [4] NCCL INFO Connected all trees
+ip-26-0-168-34:272153:272651 [3] NCCL INFO Connected all trees
+ip-26-0-168-34:272155:272652 [5] NCCL INFO Connected all trees
+ip-26-0-168-34:272157:272649 [7] NCCL INFO NVLS comm 0xb441140 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-168-34:272155:272652 [5] NCCL INFO NVLS comm 0xa77e610 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-168-34:272153:272651 [3] NCCL INFO NVLS comm 0x9b8ea10 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-153:2636088:2636601 [4] NCCL INFO NVLS comm 0xaa62b50 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-168-30:235648:236145 [0] NCCL INFO Connected all trees
+ip-26-0-161-153:2636087:2636603 [3] NCCL INFO Connected all trees
+ip-26-0-168-30:235648:236145 [0] NCCL INFO NVLS comm 0xa954a00 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-153:2636087:2636603 [3] NCCL INFO NVLS comm 0xa065880 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-168-34:272154:272650 [4] NCCL INFO Connected all trees
+ip-26-0-168-34:272150:272645 [0] NCCL INFO Connected all trees
+ip-26-0-168-34:272154:272650 [4] NCCL INFO NVLS comm 0xa853b80 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-168-34:272150:272645 [0] NCCL INFO NVLS comm 0xb33a540 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-168-34:272156:272648 [6] NCCL INFO Connected all trees
+ip-26-0-161-142:62885:63377 [4] NCCL INFO Connected all trees
+ip-26-0-168-34:272156:272648 [6] NCCL INFO NVLS comm 0xb76eb00 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-142:62885:63377 [4] NCCL INFO NVLS comm 0xb5cbf30 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-168-30:235652:236147 [4] NCCL INFO Connected all trees
+ip-26-0-168-30:235652:236147 [4] NCCL INFO NVLS comm 0xb25b2a0 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-168-30:235654:236146 [6] NCCL INFO Connected all trees
+ip-26-0-168-30:235654:236146 [6] NCCL INFO NVLS comm 0xb97f190 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-142:62884:63376 [3] NCCL INFO Connected all trees
+ip-26-0-161-142:62884:63376 [3] NCCL INFO NVLS comm 0xad6a3d0 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-153:2636091:2636599 [7] NCCL INFO Connected all trees
+ip-26-0-161-153:2636086:2636602 [2] NCCL INFO Connected all trees
+ip-26-0-161-153:2636091:2636599 [7] NCCL INFO NVLS comm 0xa7c32e0 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-153:2636084:2636605 [0] NCCL INFO Connected all trees
+ip-26-0-161-153:2636086:2636602 [2] NCCL INFO NVLS comm 0xb110fc0 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-153:2636090:2636598 [6] NCCL INFO Connected all trees
+ip-26-0-161-153:2636090:2636598 [6] NCCL INFO NVLS comm 0xb1c8e30 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-153:2636084:2636605 [0] NCCL INFO NVLS comm 0xb1686c0 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-168-34:272152:272647 [2] NCCL INFO Connected all trees
+ip-26-0-161-153:2636085:2636604 [1] NCCL INFO Connected all trees
+ip-26-0-168-34:272152:272647 [2] NCCL INFO NVLS comm 0xba40780 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-153:2636085:2636604 [1] NCCL INFO NVLS comm 0xb985620 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-168-30:235650:236144 [2] NCCL INFO Connected all trees
+ip-26-0-168-30:235650:236144 [2] NCCL INFO NVLS comm 0xa4cc2d0 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-142:62888:63380 [7] NCCL INFO Connected all trees
+ip-26-0-161-142:62888:63380 [7] NCCL INFO NVLS comm 0xa913bd0 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-142:62887:63379 [6] NCCL INFO Connected all trees
+ip-26-0-161-142:62887:63379 [6] NCCL INFO NVLS comm 0x9e2e830 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-142:62883:63373 [2] NCCL INFO Connected all trees
+ip-26-0-161-142:62883:63373 [2] NCCL INFO NVLS comm 0xa2edd90 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-142:62882:63374 [1] NCCL INFO Connected all trees
+ip-26-0-161-142:62881:63375 [0] NCCL INFO Connected all trees
+ip-26-0-161-142:62882:63374 [1] NCCL INFO NVLS comm 0xa6869b0 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-142:62881:63375 [0] NCCL INFO NVLS comm 0xb8c2040 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-161-138:96567:97090 [4] NCCL INFO Channel 01/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96569:97092 [6] NCCL INFO Channel 01/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96567:97090 [4] NCCL INFO Channel 03/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96569:97092 [6] NCCL INFO Channel 03/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96567:97090 [4] NCCL INFO Channel 05/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96569:97092 [6] NCCL INFO Channel 05/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96565:97089 [2] NCCL INFO Channel 01/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96567:97090 [4] NCCL INFO Channel 07/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96569:97092 [6] NCCL INFO Channel 07/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96567:97090 [4] NCCL INFO Channel 09/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96569:97092 [6] NCCL INFO Channel 09/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96565:97089 [2] NCCL INFO Channel 03/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96563:97086 [0] NCCL INFO Channel 01/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96566:97087 [3] NCCL INFO Channel 01/0 : 3[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96567:97090 [4] NCCL INFO Channel 11/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96569:97092 [6] NCCL INFO Channel 11/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96568:97091 [5] NCCL INFO Channel 01/0 : 5[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96570:97093 [7] NCCL INFO Channel 01/0 : 7[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96565:97089 [2] NCCL INFO Channel 05/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96564:97088 [1] NCCL INFO Channel 01/0 : 1[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96563:97086 [0] NCCL INFO Channel 03/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96567:97090 [4] NCCL INFO Channel 13/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96569:97092 [6] NCCL INFO Channel 13/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96566:97087 [3] NCCL INFO Channel 03/0 : 3[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96568:97091 [5] NCCL INFO Channel 03/0 : 5[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96570:97093 [7] NCCL INFO Channel 03/0 : 7[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96565:97089 [2] NCCL INFO Channel 07/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96564:97088 [1] NCCL INFO Channel 03/0 : 1[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96567:97090 [4] NCCL INFO Channel 15/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96563:97086 [0] NCCL INFO Channel 05/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96569:97092 [6] NCCL INFO Channel 15/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96566:97087 [3] NCCL INFO Channel 05/0 : 3[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96568:97091 [5] NCCL INFO Channel 05/0 : 5[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96570:97093 [7] NCCL INFO Channel 05/0 : 7[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96564:97088 [1] NCCL INFO Channel 05/0 : 1[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96565:97089 [2] NCCL INFO Channel 09/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96563:97086 [0] NCCL INFO Channel 07/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96566:97087 [3] NCCL INFO Channel 07/0 : 3[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96568:97091 [5] NCCL INFO Channel 07/0 : 5[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96570:97093 [7] NCCL INFO Channel 07/0 : 7[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96564:97088 [1] NCCL INFO Channel 07/0 : 1[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96563:97086 [0] NCCL INFO Channel 09/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96566:97087 [3] NCCL INFO Channel 09/0 : 3[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96565:97089 [2] NCCL INFO Channel 11/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96568:97091 [5] NCCL INFO Channel 09/0 : 5[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96564:97088 [1] NCCL INFO Channel 11/0 : 1[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96570:97093 [7] NCCL INFO Channel 09/0 : 7[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96563:97086 [0] NCCL INFO Channel 11/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96566:97087 [3] NCCL INFO Channel 13/0 : 3[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96565:97089 [2] NCCL INFO Channel 13/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96568:97091 [5] NCCL INFO Channel 11/0 : 5[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96564:97088 [1] NCCL INFO Channel 13/0 : 1[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96570:97093 [7] NCCL INFO Channel 11/0 : 7[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96563:97086 [0] NCCL INFO Channel 13/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96566:97087 [3] NCCL INFO Channel 15/0 : 3[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96568:97091 [5] NCCL INFO Channel 15/0 : 5[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96564:97088 [1] NCCL INFO Channel 15/0 : 1[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96565:97089 [2] NCCL INFO Channel 15/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96570:97093 [7] NCCL INFO Channel 13/0 : 7[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96563:97086 [0] NCCL INFO Channel 15/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815967:2816473 [4] NCCL INFO Channel 00/0 : 20[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815967:2816473 [4] NCCL INFO Channel 02/0 : 20[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815967:2816473 [4] NCCL INFO Channel 06/0 : 20[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815965:2816475 [2] NCCL INFO Channel 00/0 : 18[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815963:2816477 [0] NCCL INFO Channel 02/0 : 16[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815967:2816473 [4] NCCL INFO Channel 08/0 : 20[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815970:2816470 [7] NCCL INFO Channel 00/0 : 23[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815965:2816475 [2] NCCL INFO Channel 04/0 : 18[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815967:2816473 [4] NCCL INFO Channel 10/0 : 20[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815963:2816477 [0] NCCL INFO Channel 04/0 : 16[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815966:2816474 [3] NCCL INFO Channel 00/0 : 19[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815970:2816470 [7] NCCL INFO Channel 02/0 : 23[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815965:2816475 [2] NCCL INFO Channel 06/0 : 18[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815967:2816473 [4] NCCL INFO Channel 12/0 : 20[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815963:2816477 [0] NCCL INFO Channel 06/0 : 16[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815968:2816472 [5] NCCL INFO Channel 00/0 : 21[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815964:2816476 [1] NCCL INFO Channel 00/0 : 17[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815966:2816474 [3] NCCL INFO Channel 02/0 : 19[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815965:2816475 [2] NCCL INFO Channel 08/0 : 18[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815970:2816470 [7] NCCL INFO Channel 04/0 : 23[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815967:2816473 [4] NCCL INFO Channel 14/0 : 20[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815968:2816472 [5] NCCL INFO Channel 02/0 : 21[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815963:2816477 [0] NCCL INFO Channel 08/0 : 16[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815966:2816474 [3] NCCL INFO Channel 04/0 : 19[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815964:2816476 [1] NCCL INFO Channel 02/0 : 17[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815965:2816475 [2] NCCL INFO Channel 10/0 : 18[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815970:2816470 [7] NCCL INFO Channel 06/0 : 23[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815968:2816472 [5] NCCL INFO Channel 04/0 : 21[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815963:2816477 [0] NCCL INFO Channel 10/0 : 16[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815966:2816474 [3] NCCL INFO Channel 06/0 : 19[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815964:2816476 [1] NCCL INFO Channel 04/0 : 17[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815965:2816475 [2] NCCL INFO Channel 12/0 : 18[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815970:2816470 [7] NCCL INFO Channel 08/0 : 23[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815968:2816472 [5] NCCL INFO Channel 06/0 : 21[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815966:2816474 [3] NCCL INFO Channel 08/0 : 19[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815963:2816477 [0] NCCL INFO Channel 12/0 : 16[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815964:2816476 [1] NCCL INFO Channel 06/0 : 17[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815965:2816475 [2] NCCL INFO Channel 14/0 : 18[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815970:2816470 [7] NCCL INFO Channel 10/0 : 23[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815968:2816472 [5] NCCL INFO Channel 08/0 : 21[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815966:2816474 [3] NCCL INFO Channel 10/0 : 19[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815963:2816477 [0] NCCL INFO Channel 14/0 : 16[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815964:2816476 [1] NCCL INFO Channel 08/0 : 17[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815969:2816471 [6] NCCL INFO Channel 00/0 : 22[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815970:2816470 [7] NCCL INFO Channel 12/0 : 23[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815966:2816474 [3] NCCL INFO Channel 12/0 : 19[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815968:2816472 [5] NCCL INFO Channel 10/0 : 21[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815964:2816476 [1] NCCL INFO Channel 10/0 : 17[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815969:2816471 [6] NCCL INFO Channel 02/0 : 22[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815970:2816470 [7] NCCL INFO Channel 14/0 : 23[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815966:2816474 [3] NCCL INFO Channel 14/0 : 19[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815964:2816476 [1] NCCL INFO Channel 12/0 : 17[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815968:2816472 [5] NCCL INFO Channel 12/0 : 21[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815969:2816471 [6] NCCL INFO Channel 04/0 : 22[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815964:2816476 [1] NCCL INFO Channel 14/0 : 17[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815969:2816471 [6] NCCL INFO Channel 08/0 : 22[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815968:2816472 [5] NCCL INFO Channel 14/0 : 21[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815969:2816471 [6] NCCL INFO Channel 10/0 : 22[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815969:2816471 [6] NCCL INFO Channel 12/0 : 22[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815969:2816471 [6] NCCL INFO Channel 14/0 : 22[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-52:2722049:2722547 [4] NCCL INFO Channel 00/0 : 20[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722051:2722543 [6] NCCL INFO Channel 00/0 : 22[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-52:2722049:2722547 [4] NCCL INFO Channel 02/0 : 20[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722047:2722542 [2] NCCL INFO Channel 00/0 : 18[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-52:2722051:2722543 [6] NCCL INFO Channel 02/0 : 22[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-52:2722049:2722547 [4] NCCL INFO Channel 06/0 : 20[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722047:2722542 [2] NCCL INFO Channel 04/0 : 18[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-52:2722051:2722543 [6] NCCL INFO Channel 04/0 : 22[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-52:2722049:2722547 [4] NCCL INFO Channel 08/0 : 20[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722046:2722541 [1] NCCL INFO Channel 00/0 : 17[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-52:2722047:2722542 [2] NCCL INFO Channel 06/0 : 18[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-52:2722050:2722548 [5] NCCL INFO Channel 00/0 : 21[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-52:2722051:2722543 [6] NCCL INFO Channel 08/0 : 22[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-52:2722052:2722546 [7] NCCL INFO Channel 00/0 : 23[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722049:2722547 [4] NCCL INFO Channel 10/0 : 20[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722046:2722541 [1] NCCL INFO Channel 02/0 : 17[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-52:2722047:2722542 [2] NCCL INFO Channel 08/0 : 18[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-52:2722048:2722545 [3] NCCL INFO Channel 00/0 : 19[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-52:2722051:2722543 [6] NCCL INFO Channel 10/0 : 22[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-52:2722052:2722546 [7] NCCL INFO Channel 02/0 : 23[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722050:2722548 [5] NCCL INFO Channel 02/0 : 21[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-52:2722046:2722541 [1] NCCL INFO Channel 04/0 : 17[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-52:2722049:2722547 [4] NCCL INFO Channel 12/0 : 20[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722047:2722542 [2] NCCL INFO Channel 10/0 : 18[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-52:2722045:2722544 [0] NCCL INFO Channel 02/0 : 16[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-52:2722051:2722543 [6] NCCL INFO Channel 12/0 : 22[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-52:2722052:2722546 [7] NCCL INFO Channel 04/0 : 23[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722048:2722545 [3] NCCL INFO Channel 02/0 : 19[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-52:2722050:2722548 [5] NCCL INFO Channel 04/0 : 21[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-52:2722046:2722541 [1] NCCL INFO Channel 06/0 : 17[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-52:2722047:2722542 [2] NCCL INFO Channel 12/0 : 18[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-52:2722049:2722547 [4] NCCL INFO Channel 14/0 : 20[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722045:2722544 [0] NCCL INFO Channel 04/0 : 16[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-52:2722051:2722543 [6] NCCL INFO Channel 14/0 : 22[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-52:2722048:2722545 [3] NCCL INFO Channel 04/0 : 19[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-52:2722052:2722546 [7] NCCL INFO Channel 06/0 : 23[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722050:2722548 [5] NCCL INFO Channel 06/0 : 21[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-52:2722046:2722541 [1] NCCL INFO Channel 08/0 : 17[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-52:2722047:2722542 [2] NCCL INFO Channel 14/0 : 18[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-52:2722045:2722544 [0] NCCL INFO Channel 06/0 : 16[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-52:2722048:2722545 [3] NCCL INFO Channel 06/0 : 19[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-52:2722052:2722546 [7] NCCL INFO Channel 08/0 : 23[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722050:2722548 [5] NCCL INFO Channel 08/0 : 21[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-52:2722046:2722541 [1] NCCL INFO Channel 10/0 : 17[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-52:2722048:2722545 [3] NCCL INFO Channel 08/0 : 19[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-52:2722045:2722544 [0] NCCL INFO Channel 08/0 : 16[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-52:2722052:2722546 [7] NCCL INFO Channel 10/0 : 23[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722046:2722541 [1] NCCL INFO Channel 12/0 : 17[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-52:2722050:2722548 [5] NCCL INFO Channel 10/0 : 21[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-52:2722048:2722545 [3] NCCL INFO Channel 10/0 : 19[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-52:2722045:2722544 [0] NCCL INFO Channel 10/0 : 16[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-52:2722052:2722546 [7] NCCL INFO Channel 12/0 : 23[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722046:2722541 [1] NCCL INFO Channel 14/0 : 17[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-52:2722050:2722548 [5] NCCL INFO Channel 12/0 : 21[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-52:2722048:2722545 [3] NCCL INFO Channel 12/0 : 19[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-52:2722045:2722544 [0] NCCL INFO Channel 12/0 : 16[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-52:2722052:2722546 [7] NCCL INFO Channel 14/0 : 23[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722045:2722544 [0] NCCL INFO Channel 14/0 : 16[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-52:2722050:2722548 [5] NCCL INFO Channel 14/0 : 21[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-52:2722048:2722545 [3] NCCL INFO Channel 14/0 : 19[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272151:272646 [1] NCCL INFO Channel 00/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272153:272651 [3] NCCL INFO Channel 00/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272151:272646 [1] NCCL INFO Channel 02/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272156:272648 [6] NCCL INFO Channel 00/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272155:272652 [5] NCCL INFO Channel 00/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272157:272649 [7] NCCL INFO Channel 00/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272154:272650 [4] NCCL INFO Channel 00/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272151:272646 [1] NCCL INFO Channel 03/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272153:272651 [3] NCCL INFO Channel 01/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272156:272648 [6] NCCL INFO Channel 01/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272152:272647 [2] NCCL INFO Channel 00/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272155:272652 [5] NCCL INFO Channel 01/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272150:272645 [0] NCCL INFO Channel 01/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272157:272649 [7] NCCL INFO Channel 01/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272153:272651 [3] NCCL INFO Channel 02/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272154:272650 [4] NCCL INFO Channel 01/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272151:272646 [1] NCCL INFO Channel 04/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272156:272648 [6] NCCL INFO Channel 02/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272152:272647 [2] NCCL INFO Channel 01/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272155:272652 [5] NCCL INFO Channel 02/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272150:272645 [0] NCCL INFO Channel 02/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272157:272649 [7] NCCL INFO Channel 02/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272153:272651 [3] NCCL INFO Channel 04/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272151:272646 [1] NCCL INFO Channel 05/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272156:272648 [6] NCCL INFO Channel 03/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272155:272652 [5] NCCL INFO Channel 03/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272152:272647 [2] NCCL INFO Channel 03/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272154:272650 [4] NCCL INFO Channel 02/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272157:272649 [7] NCCL INFO Channel 03/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272153:272651 [3] NCCL INFO Channel 05/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272150:272645 [0] NCCL INFO Channel 03/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272151:272646 [1] NCCL INFO Channel 06/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272155:272652 [5] NCCL INFO Channel 04/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272152:272647 [2] NCCL INFO Channel 04/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272156:272648 [6] NCCL INFO Channel 04/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272154:272650 [4] NCCL INFO Channel 03/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272153:272651 [3] NCCL INFO Channel 06/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272157:272649 [7] NCCL INFO Channel 04/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272151:272646 [1] NCCL INFO Channel 07/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272150:272645 [0] NCCL INFO Channel 04/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272155:272652 [5] NCCL INFO Channel 06/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272156:272648 [6] NCCL INFO Channel 05/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272152:272647 [2] NCCL INFO Channel 05/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272157:272649 [7] NCCL INFO Channel 05/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272154:272650 [4] NCCL INFO Channel 05/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272153:272651 [3] NCCL INFO Channel 07/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272151:272646 [1] NCCL INFO Channel 08/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272150:272645 [0] NCCL INFO Channel 05/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272155:272652 [5] NCCL INFO Channel 07/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272156:272648 [6] NCCL INFO Channel 07/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272152:272647 [2] NCCL INFO Channel 06/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272157:272649 [7] NCCL INFO Channel 06/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272154:272650 [4] NCCL INFO Channel 06/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272153:272651 [3] NCCL INFO Channel 08/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272150:272645 [0] NCCL INFO Channel 06/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272151:272646 [1] NCCL INFO Channel 10/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272155:272652 [5] NCCL INFO Channel 08/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272156:272648 [6] NCCL INFO Channel 08/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272152:272647 [2] NCCL INFO Channel 07/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272151:272646 [1] NCCL INFO Channel 11/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272157:272649 [7] NCCL INFO Channel 08/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272154:272650 [4] NCCL INFO Channel 07/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272150:272645 [0] NCCL INFO Channel 07/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272155:272652 [5] NCCL INFO Channel 09/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272153:272651 [3] NCCL INFO Channel 09/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272156:272648 [6] NCCL INFO Channel 09/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272152:272647 [2] NCCL INFO Channel 08/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272151:272646 [1] NCCL INFO Channel 12/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272157:272649 [7] NCCL INFO Channel 09/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272154:272650 [4] NCCL INFO Channel 08/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272150:272645 [0] NCCL INFO Channel 09/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272155:272652 [5] NCCL INFO Channel 10/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272153:272651 [3] NCCL INFO Channel 10/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272152:272647 [2] NCCL INFO Channel 09/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272156:272648 [6] NCCL INFO Channel 10/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272151:272646 [1] NCCL INFO Channel 13/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272157:272649 [7] NCCL INFO Channel 10/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272154:272650 [4] NCCL INFO Channel 09/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272150:272645 [0] NCCL INFO Channel 10/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272156:272648 [6] NCCL INFO Channel 11/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272155:272652 [5] NCCL INFO Channel 11/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272153:272651 [3] NCCL INFO Channel 12/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272152:272647 [2] NCCL INFO Channel 11/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272157:272649 [7] NCCL INFO Channel 11/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272151:272646 [1] NCCL INFO Channel 14/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272154:272650 [4] NCCL INFO Channel 10/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272150:272645 [0] NCCL INFO Channel 11/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272156:272648 [6] NCCL INFO Channel 12/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272153:272651 [3] NCCL INFO Channel 13/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272157:272649 [7] NCCL INFO Channel 12/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272151:272646 [1] NCCL INFO Channel 15/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272155:272652 [5] NCCL INFO Channel 12/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272152:272647 [2] NCCL INFO Channel 12/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272154:272650 [4] NCCL INFO Channel 11/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272150:272645 [0] NCCL INFO Channel 12/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272156:272648 [6] NCCL INFO Channel 13/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272153:272651 [3] NCCL INFO Channel 14/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272151:272646 [1] NCCL INFO Channel 00/0 : 17[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272155:272652 [5] NCCL INFO Channel 14/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272157:272649 [7] NCCL INFO Channel 13/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272152:272647 [2] NCCL INFO Channel 13/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272154:272650 [4] NCCL INFO Channel 13/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272150:272645 [0] NCCL INFO Channel 13/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272156:272648 [6] NCCL INFO Channel 15/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272153:272651 [3] NCCL INFO Channel 15/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272151:272646 [1] NCCL INFO Channel 02/0 : 17[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272155:272652 [5] NCCL INFO Channel 15/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272157:272649 [7] NCCL INFO Channel 14/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272152:272647 [2] NCCL INFO Channel 14/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272154:272650 [4] NCCL INFO Channel 14/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272156:272648 [6] NCCL INFO Channel 00/0 : 22[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272153:272651 [3] NCCL INFO Channel 00/0 : 19[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272150:272645 [0] NCCL INFO Channel 14/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272151:272646 [1] NCCL INFO Channel 04/0 : 17[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272157:272649 [7] NCCL INFO Channel 00/0 : 23[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272155:272652 [5] NCCL INFO Channel 00/0 : 21[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272152:272647 [2] NCCL INFO Channel 15/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272154:272650 [4] NCCL INFO Channel 15/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272156:272648 [6] NCCL INFO Channel 02/0 : 22[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272150:272645 [0] NCCL INFO Channel 15/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272153:272651 [3] NCCL INFO Channel 02/0 : 19[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272151:272646 [1] NCCL INFO Channel 06/0 : 17[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272157:272649 [7] NCCL INFO Channel 02/0 : 23[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272155:272652 [5] NCCL INFO Channel 02/0 : 21[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272152:272647 [2] NCCL INFO Channel 00/0 : 18[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272154:272650 [4] NCCL INFO Channel 00/0 : 20[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272156:272648 [6] NCCL INFO Channel 04/0 : 22[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272150:272645 [0] NCCL INFO Channel 02/0 : 16[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272153:272651 [3] NCCL INFO Channel 04/0 : 19[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272151:272646 [1] NCCL INFO Channel 08/0 : 17[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272155:272652 [5] NCCL INFO Channel 04/0 : 21[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272157:272649 [7] NCCL INFO Channel 04/0 : 23[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272154:272650 [4] NCCL INFO Channel 02/0 : 20[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272152:272647 [2] NCCL INFO Channel 04/0 : 18[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272156:272648 [6] NCCL INFO Channel 08/0 : 22[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272150:272645 [0] NCCL INFO Channel 04/0 : 16[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272153:272651 [3] NCCL INFO Channel 06/0 : 19[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272151:272646 [1] NCCL INFO Channel 10/0 : 17[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272155:272652 [5] NCCL INFO Channel 06/0 : 21[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272157:272649 [7] NCCL INFO Channel 06/0 : 23[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272154:272650 [4] NCCL INFO Channel 06/0 : 20[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272156:272648 [6] NCCL INFO Channel 10/0 : 22[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272152:272647 [2] NCCL INFO Channel 06/0 : 18[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272150:272645 [0] NCCL INFO Channel 06/0 : 16[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272153:272651 [3] NCCL INFO Channel 08/0 : 19[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272151:272646 [1] NCCL INFO Channel 12/0 : 17[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272155:272652 [5] NCCL INFO Channel 08/0 : 21[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272157:272649 [7] NCCL INFO Channel 08/0 : 23[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272154:272650 [4] NCCL INFO Channel 08/0 : 20[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272156:272648 [6] NCCL INFO Channel 12/0 : 22[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272150:272645 [0] NCCL INFO Channel 08/0 : 16[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272152:272647 [2] NCCL INFO Channel 08/0 : 18[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272151:272646 [1] NCCL INFO Channel 14/0 : 17[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272153:272651 [3] NCCL INFO Channel 10/0 : 19[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272155:272652 [5] NCCL INFO Channel 10/0 : 21[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272157:272649 [7] NCCL INFO Channel 10/0 : 23[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272154:272650 [4] NCCL INFO Channel 10/0 : 20[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272156:272648 [6] NCCL INFO Channel 14/0 : 22[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272150:272645 [0] NCCL INFO Channel 10/0 : 16[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272152:272647 [2] NCCL INFO Channel 10/0 : 18[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272153:272651 [3] NCCL INFO Channel 12/0 : 19[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272154:272650 [4] NCCL INFO Channel 12/0 : 20[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272157:272649 [7] NCCL INFO Channel 12/0 : 23[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272155:272652 [5] NCCL INFO Channel 12/0 : 21[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272150:272645 [0] NCCL INFO Channel 12/0 : 16[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272152:272647 [2] NCCL INFO Channel 12/0 : 18[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272153:272651 [3] NCCL INFO Channel 14/0 : 19[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272157:272649 [7] NCCL INFO Channel 14/0 : 23[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272154:272650 [4] NCCL INFO Channel 14/0 : 20[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272155:272652 [5] NCCL INFO Channel 14/0 : 21[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272150:272645 [0] NCCL INFO Channel 14/0 : 16[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272152:272647 [2] NCCL INFO Channel 14/0 : 18[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-52:2722046:2722541 [1] NCCL INFO Channel 01/0 : 9[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-52:2722051:2722543 [6] NCCL INFO Channel 01/0 : 14[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-52:2722046:2722541 [1] NCCL INFO Channel 03/0 : 9[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-52:2722048:2722545 [3] NCCL INFO Channel 01/0 : 11[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235655:236142 [7] NCCL INFO Channel 01/0 : 7[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722051:2722543 [6] NCCL INFO Channel 03/0 : 14[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-52:2722046:2722541 [1] NCCL INFO Channel 05/0 : 9[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235648:236145 [0] NCCL INFO Channel 01/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-52:2722048:2722545 [3] NCCL INFO Channel 03/0 : 11[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235655:236142 [7] NCCL INFO Channel 03/0 : 7[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722049:2722547 [4] NCCL INFO Channel 01/0 : 12[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235648:236145 [0] NCCL INFO Channel 03/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-52:2722045:2722544 [0] NCCL INFO Channel 01/0 : 8[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-52:2722046:2722541 [1] NCCL INFO Channel 07/0 : 9[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-52:2722052:2722546 [7] NCCL INFO Channel 01/0 : 15[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722050:2722548 [5] NCCL INFO Channel 01/0 : 13[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-52:2722051:2722543 [6] NCCL INFO Channel 05/0 : 14[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-52:2722048:2722545 [3] NCCL INFO Channel 05/0 : 11[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-52:2722047:2722542 [2] NCCL INFO Channel 01/0 : 10[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235655:236142 [7] NCCL INFO Channel 05/0 : 7[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235652:236147 [4] NCCL INFO Channel 01/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235648:236145 [0] NCCL INFO Channel 05/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-52:2722049:2722547 [4] NCCL INFO Channel 03/0 : 12[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235655:236142 [7] NCCL INFO Channel 07/0 : 7[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235648:236145 [0] NCCL INFO Channel 07/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235653:236140 [5] NCCL INFO Channel 01/0 : 5[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-52:2722045:2722544 [0] NCCL INFO Channel 03/0 : 8[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-52:2722046:2722541 [1] NCCL INFO Channel 11/0 : 9[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235652:236147 [4] NCCL INFO Channel 03/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235655:236142 [7] NCCL INFO Channel 09/0 : 7[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722052:2722546 [7] NCCL INFO Channel 03/0 : 15[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722050:2722548 [5] NCCL INFO Channel 03/0 : 13[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-52:2722048:2722545 [3] NCCL INFO Channel 07/0 : 11[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-52:2722047:2722542 [2] NCCL INFO Channel 03/0 : 10[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-52:2722051:2722543 [6] NCCL INFO Channel 07/0 : 14[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-52:2722049:2722547 [4] NCCL INFO Channel 05/0 : 12[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235648:236145 [0] NCCL INFO Channel 09/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235650:236144 [2] NCCL INFO Channel 01/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235653:236140 [5] NCCL INFO Channel 03/0 : 5[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235649:236143 [1] NCCL INFO Channel 01/0 : 1[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-52:2722045:2722544 [0] NCCL INFO Channel 05/0 : 8[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-52:2722046:2722541 [1] NCCL INFO Channel 13/0 : 9[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235652:236147 [4] NCCL INFO Channel 05/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235655:236142 [7] NCCL INFO Channel 11/0 : 7[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235648:236145 [0] NCCL INFO Channel 11/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235654:236146 [6] NCCL INFO Channel 01/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-52:2722047:2722542 [2] NCCL INFO Channel 05/0 : 10[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-52:2722048:2722545 [3] NCCL INFO Channel 09/0 : 11[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-52:2722052:2722546 [7] NCCL INFO Channel 05/0 : 15[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722050:2722548 [5] NCCL INFO Channel 05/0 : 13[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235651:236141 [3] NCCL INFO Channel 01/0 : 3[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235650:236144 [2] NCCL INFO Channel 03/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-52:2722049:2722547 [4] NCCL INFO Channel 07/0 : 12[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722051:2722543 [6] NCCL INFO Channel 09/0 : 14[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-52:2722046:2722541 [1] NCCL INFO Channel 15/0 : 9[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-52:2722045:2722544 [0] NCCL INFO Channel 07/0 : 8[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235653:236140 [5] NCCL INFO Channel 05/0 : 5[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235649:236143 [1] NCCL INFO Channel 03/0 : 1[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235652:236147 [4] NCCL INFO Channel 07/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235655:236142 [7] NCCL INFO Channel 13/0 : 7[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235648:236145 [0] NCCL INFO Channel 13/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-52:2722048:2722545 [3] NCCL INFO Channel 13/0 : 11[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-52:2722047:2722542 [2] NCCL INFO Channel 07/0 : 10[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235654:236146 [6] NCCL INFO Channel 03/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235651:236141 [3] NCCL INFO Channel 03/0 : 3[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235650:236144 [2] NCCL INFO Channel 05/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235653:236140 [5] NCCL INFO Channel 07/0 : 5[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235649:236143 [1] NCCL INFO Channel 05/0 : 1[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235655:236142 [7] NCCL INFO Channel 00/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235652:236147 [4] NCCL INFO Channel 09/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235648:236145 [0] NCCL INFO Channel 15/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-52:2722050:2722548 [5] NCCL INFO Channel 07/0 : 13[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-52:2722046:2722541 [1] NCCL INFO Channel 01/0 : 25[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-52:2722045:2722544 [0] NCCL INFO Channel 09/0 : 8[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-52:2722052:2722546 [7] NCCL INFO Channel 07/0 : 15[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722049:2722547 [4] NCCL INFO Channel 09/0 : 12[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722051:2722543 [6] NCCL INFO Channel 11/0 : 14[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-52:2722048:2722545 [3] NCCL INFO Channel 15/0 : 11[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235651:236141 [3] NCCL INFO Channel 05/0 : 3[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-52:2722047:2722542 [2] NCCL INFO Channel 09/0 : 10[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235654:236146 [6] NCCL INFO Channel 05/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-52:2722050:2722548 [5] NCCL INFO Channel 09/0 : 13[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235653:236140 [5] NCCL INFO Channel 09/0 : 5[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235650:236144 [2] NCCL INFO Channel 07/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235649:236143 [1] NCCL INFO Channel 07/0 : 1[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-52:2722046:2722541 [1] NCCL INFO Channel 03/0 : 25[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-52:2722045:2722544 [0] NCCL INFO Channel 11/0 : 8[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-52:2722052:2722546 [7] NCCL INFO Channel 09/0 : 15[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235655:236142 [7] NCCL INFO Channel 01/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235652:236147 [4] NCCL INFO Channel 11/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235648:236145 [0] NCCL INFO Channel 01/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-52:2722048:2722545 [3] NCCL INFO Channel 01/0 : 27[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-52:2722049:2722547 [4] NCCL INFO Channel 11/0 : 12[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722047:2722542 [2] NCCL INFO Channel 11/0 : 10[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235654:236146 [6] NCCL INFO Channel 07/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-52:2722051:2722543 [6] NCCL INFO Channel 13/0 : 14[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-52:2722050:2722548 [5] NCCL INFO Channel 11/0 : 13[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235653:236140 [5] NCCL INFO Channel 11/0 : 5[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235651:236141 [3] NCCL INFO Channel 07/0 : 3[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-52:2722046:2722541 [1] NCCL INFO Channel 05/0 : 25[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235649:236143 [1] NCCL INFO Channel 11/0 : 1[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-52:2722045:2722544 [0] NCCL INFO Channel 13/0 : 8[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235650:236144 [2] NCCL INFO Channel 09/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-52:2722052:2722546 [7] NCCL INFO Channel 11/0 : 15[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235652:236147 [4] NCCL INFO Channel 13/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235648:236145 [0] NCCL INFO Channel 02/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235655:236142 [7] NCCL INFO Channel 02/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722048:2722545 [3] NCCL INFO Channel 03/0 : 27[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-52:2722049:2722547 [4] NCCL INFO Channel 13/0 : 12[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722047:2722542 [2] NCCL INFO Channel 13/0 : 10[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-52:2722046:2722541 [1] NCCL INFO Channel 07/0 : 25[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235654:236146 [6] NCCL INFO Channel 09/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-52:2722050:2722548 [5] NCCL INFO Channel 15/0 : 13[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-52:2722045:2722544 [0] NCCL INFO Channel 15/0 : 8[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-52:2722051:2722543 [6] NCCL INFO Channel 15/0 : 14[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235653:236140 [5] NCCL INFO Channel 15/0 : 5[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-52:2722052:2722546 [7] NCCL INFO Channel 13/0 : 15[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235651:236141 [3] NCCL INFO Channel 09/0 : 3[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235649:236143 [1] NCCL INFO Channel 13/0 : 1[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-52:2722048:2722545 [3] NCCL INFO Channel 05/0 : 27[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235648:236145 [0] NCCL INFO Channel 03/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235652:236147 [4] NCCL INFO Channel 15/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235655:236142 [7] NCCL INFO Channel 03/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235650:236144 [2] NCCL INFO Channel 11/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235653:236140 [5] NCCL INFO Channel 00/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235654:236146 [6] NCCL INFO Channel 11/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235649:236143 [1] NCCL INFO Channel 15/0 : 1[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-52:2722047:2722542 [2] NCCL INFO Channel 15/0 : 10[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235651:236141 [3] NCCL INFO Channel 13/0 : 3[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-52:2722046:2722541 [1] NCCL INFO Channel 11/0 : 25[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-52:2722049:2722547 [4] NCCL INFO Channel 15/0 : 12[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235652:236147 [4] NCCL INFO Channel 00/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235655:236142 [7] NCCL INFO Channel 04/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722045:2722544 [0] NCCL INFO Channel 01/0 : 24[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-52:2722050:2722548 [5] NCCL INFO Channel 01/0 : 29[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-52:2722051:2722543 [6] NCCL INFO Channel 01/0 : 30[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-52:2722048:2722545 [3] NCCL INFO Channel 07/0 : 27[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-52:2722052:2722546 [7] NCCL INFO Channel 01/0 : 31[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722047:2722542 [2] NCCL INFO Channel 01/0 : 26[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-52:2722046:2722541 [1] NCCL INFO Channel 13/0 : 25[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-52:2722049:2722547 [4] NCCL INFO Channel 01/0 : 28[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235648:236145 [0] NCCL INFO Channel 04/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235654:236146 [6] NCCL INFO Channel 13/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235649:236143 [1] NCCL INFO Channel 00/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235650:236144 [2] NCCL INFO Channel 13/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235653:236140 [5] NCCL INFO Channel 01/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235651:236141 [3] NCCL INFO Channel 15/0 : 3[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-52:2722050:2722548 [5] NCCL INFO Channel 03/0 : 29[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235652:236147 [4] NCCL INFO Channel 01/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235655:236142 [7] NCCL INFO Channel 05/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235648:236145 [0] NCCL INFO Channel 05/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235654:236146 [6] NCCL INFO Channel 15/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235649:236143 [1] NCCL INFO Channel 02/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-52:2722048:2722545 [3] NCCL INFO Channel 09/0 : 27[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-52:2722052:2722546 [7] NCCL INFO Channel 03/0 : 31[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235650:236144 [2] NCCL INFO Channel 15/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-52:2722045:2722544 [0] NCCL INFO Channel 03/0 : 24[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-52:2722047:2722542 [2] NCCL INFO Channel 03/0 : 26[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-52:2722051:2722543 [6] NCCL INFO Channel 03/0 : 30[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-52:2722046:2722541 [1] NCCL INFO Channel 15/0 : 25[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-52:2722049:2722547 [4] NCCL INFO Channel 03/0 : 28[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722050:2722548 [5] NCCL INFO Channel 05/0 : 29[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235651:236141 [3] NCCL INFO Channel 00/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235652:236147 [4] NCCL INFO Channel 02/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235655:236142 [7] NCCL INFO Channel 06/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235653:236140 [5] NCCL INFO Channel 02/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235648:236145 [0] NCCL INFO Channel 06/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-52:2722048:2722545 [3] NCCL INFO Channel 13/0 : 27[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-52:2722052:2722546 [7] NCCL INFO Channel 05/0 : 31[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722047:2722542 [2] NCCL INFO Channel 05/0 : 26[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-52:2722045:2722544 [0] NCCL INFO Channel 05/0 : 24[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235649:236143 [1] NCCL INFO Channel 03/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235654:236146 [6] NCCL INFO Channel 00/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235650:236144 [2] NCCL INFO Channel 00/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235651:236141 [3] NCCL INFO Channel 01/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235652:236147 [4] NCCL INFO Channel 03/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235655:236142 [7] NCCL INFO Channel 08/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722051:2722543 [6] NCCL INFO Channel 05/0 : 30[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-52:2722049:2722547 [4] NCCL INFO Channel 05/0 : 28[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722050:2722548 [5] NCCL INFO Channel 07/0 : 29[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-52:2722048:2722545 [3] NCCL INFO Channel 15/0 : 27[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235653:236140 [5] NCCL INFO Channel 03/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235648:236145 [0] NCCL INFO Channel 07/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235649:236143 [1] NCCL INFO Channel 04/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235654:236146 [6] NCCL INFO Channel 01/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235650:236144 [2] NCCL INFO Channel 01/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-52:2722047:2722542 [2] NCCL INFO Channel 07/0 : 26[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-52:2722052:2722546 [7] NCCL INFO Channel 07/0 : 31[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722045:2722544 [0] NCCL INFO Channel 07/0 : 24[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-52:2722049:2722547 [4] NCCL INFO Channel 07/0 : 28[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722050:2722548 [5] NCCL INFO Channel 09/0 : 29[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-52:2722051:2722543 [6] NCCL INFO Channel 07/0 : 30[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235651:236141 [3] NCCL INFO Channel 02/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235652:236147 [4] NCCL INFO Channel 05/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235655:236142 [7] NCCL INFO Channel 09/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235653:236140 [5] NCCL INFO Channel 04/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235648:236145 [0] NCCL INFO Channel 09/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-52:2722052:2722546 [7] NCCL INFO Channel 09/0 : 31[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722047:2722542 [2] NCCL INFO Channel 09/0 : 26[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-52:2722045:2722544 [0] NCCL INFO Channel 09/0 : 24[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235649:236143 [1] NCCL INFO Channel 05/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235654:236146 [6] NCCL INFO Channel 02/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235650:236144 [2] NCCL INFO Channel 03/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235652:236147 [4] NCCL INFO Channel 06/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235651:236141 [3] NCCL INFO Channel 04/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235655:236142 [7] NCCL INFO Channel 10/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722050:2722548 [5] NCCL INFO Channel 11/0 : 29[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-52:2722051:2722543 [6] NCCL INFO Channel 09/0 : 30[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-52:2722049:2722547 [4] NCCL INFO Channel 09/0 : 28[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722052:2722546 [7] NCCL INFO Channel 11/0 : 31[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722047:2722542 [2] NCCL INFO Channel 11/0 : 26[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-52:2722045:2722544 [0] NCCL INFO Channel 11/0 : 24[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235653:236140 [5] NCCL INFO Channel 06/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235648:236145 [0] NCCL INFO Channel 10/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235649:236143 [1] NCCL INFO Channel 06/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235654:236146 [6] NCCL INFO Channel 03/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235650:236144 [2] NCCL INFO Channel 04/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-52:2722050:2722548 [5] NCCL INFO Channel 15/0 : 29[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-52:2722051:2722543 [6] NCCL INFO Channel 11/0 : 30[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-52:2722049:2722547 [4] NCCL INFO Channel 11/0 : 28[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722052:2722546 [7] NCCL INFO Channel 13/0 : 31[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722047:2722542 [2] NCCL INFO Channel 13/0 : 26[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-52:2722045:2722544 [0] NCCL INFO Channel 13/0 : 24[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235652:236147 [4] NCCL INFO Channel 07/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235655:236142 [7] NCCL INFO Channel 11/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235651:236141 [3] NCCL INFO Channel 05/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235653:236140 [5] NCCL INFO Channel 07/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-52:2722051:2722543 [6] NCCL INFO Channel 13/0 : 30[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-52:2722049:2722547 [4] NCCL INFO Channel 13/0 : 28[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722047:2722542 [2] NCCL INFO Channel 15/0 : 26[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-52:2722045:2722544 [0] NCCL INFO Channel 15/0 : 24[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235648:236145 [0] NCCL INFO Channel 11/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235649:236143 [1] NCCL INFO Channel 07/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235654:236146 [6] NCCL INFO Channel 04/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235650:236144 [2] NCCL INFO Channel 05/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235652:236147 [4] NCCL INFO Channel 08/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235655:236142 [7] NCCL INFO Channel 12/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235651:236141 [3] NCCL INFO Channel 06/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235653:236140 [5] NCCL INFO Channel 08/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-52:2722049:2722547 [4] NCCL INFO Channel 15/0 : 28[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722051:2722543 [6] NCCL INFO Channel 15/0 : 30[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235649:236143 [1] NCCL INFO Channel 08/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235648:236145 [0] NCCL INFO Channel 12/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235654:236146 [6] NCCL INFO Channel 05/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235650:236144 [2] NCCL INFO Channel 06/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235652:236147 [4] NCCL INFO Channel 09/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235655:236142 [7] NCCL INFO Channel 13/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235653:236140 [5] NCCL INFO Channel 09/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235651:236141 [3] NCCL INFO Channel 07/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235649:236143 [1] NCCL INFO Channel 10/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235648:236145 [0] NCCL INFO Channel 13/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235654:236146 [6] NCCL INFO Channel 07/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235652:236147 [4] NCCL INFO Channel 10/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235650:236144 [2] NCCL INFO Channel 07/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235655:236142 [7] NCCL INFO Channel 14/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235653:236140 [5] NCCL INFO Channel 10/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235651:236141 [3] NCCL INFO Channel 08/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235649:236143 [1] NCCL INFO Channel 11/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235654:236146 [6] NCCL INFO Channel 08/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235648:236145 [0] NCCL INFO Channel 14/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235652:236147 [4] NCCL INFO Channel 11/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235650:236144 [2] NCCL INFO Channel 08/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235653:236140 [5] NCCL INFO Channel 11/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235651:236141 [3] NCCL INFO Channel 09/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235649:236143 [1] NCCL INFO Channel 12/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235654:236146 [6] NCCL INFO Channel 09/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235648:236145 [0] NCCL INFO Channel 15/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235652:236147 [4] NCCL INFO Channel 13/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235650:236144 [2] NCCL INFO Channel 09/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235653:236140 [5] NCCL INFO Channel 12/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235651:236141 [3] NCCL INFO Channel 10/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235649:236143 [1] NCCL INFO Channel 13/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235654:236146 [6] NCCL INFO Channel 10/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235652:236147 [4] NCCL INFO Channel 14/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235650:236144 [2] NCCL INFO Channel 11/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235653:236140 [5] NCCL INFO Channel 14/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235651:236141 [3] NCCL INFO Channel 12/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235649:236143 [1] NCCL INFO Channel 14/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235654:236146 [6] NCCL INFO Channel 11/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235652:236147 [4] NCCL INFO Channel 15/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235650:236144 [2] NCCL INFO Channel 12/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235653:236140 [5] NCCL INFO Channel 15/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235651:236141 [3] NCCL INFO Channel 13/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235649:236143 [1] NCCL INFO Channel 15/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235654:236146 [6] NCCL INFO Channel 12/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235655:236142 [7] NCCL INFO Channel 01/0 : 31[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272157:272649 [7] NCCL INFO Channel 00/0 : 7[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235650:236144 [2] NCCL INFO Channel 13/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235651:236141 [3] NCCL INFO Channel 14/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235654:236146 [6] NCCL INFO Channel 13/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272157:272649 [7] NCCL INFO Channel 02/0 : 7[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235655:236142 [7] NCCL INFO Channel 03/0 : 31[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235650:236144 [2] NCCL INFO Channel 14/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235648:236145 [0] NCCL INFO Channel 01/0 : 24[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272157:272649 [7] NCCL INFO Channel 04/0 : 7[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235654:236146 [6] NCCL INFO Channel 15/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636090:2636598 [6] NCCL INFO Channel 00/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235651:236141 [3] NCCL INFO Channel 15/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235655:236142 [7] NCCL INFO Channel 05/0 : 31[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235650:236144 [2] NCCL INFO Channel 15/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557270:1557763 [4] NCCL INFO Channel 00/0 : 20[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272150:272645 [0] NCCL INFO Channel 02/0 : 0[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636090:2636598 [6] NCCL INFO Channel 01/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235648:236145 [0] NCCL INFO Channel 03/0 : 24[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235655:236142 [7] NCCL INFO Channel 07/0 : 31[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235648:236145 [0] NCCL INFO Channel 05/0 : 24[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272157:272649 [7] NCCL INFO Channel 06/0 : 7[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557267:1557761 [1] NCCL INFO Channel 00/0 : 17[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235652:236147 [4] NCCL INFO Channel 01/0 : 28[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235655:236142 [7] NCCL INFO Channel 09/0 : 31[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235648:236145 [0] NCCL INFO Channel 07/0 : 24[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272150:272645 [0] NCCL INFO Channel 04/0 : 0[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636090:2636598 [6] NCCL INFO Channel 02/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557270:1557763 [4] NCCL INFO Channel 02/0 : 20[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235649:236143 [1] NCCL INFO Channel 01/0 : 25[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272154:272650 [4] NCCL INFO Channel 00/0 : 4[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272157:272649 [7] NCCL INFO Channel 08/0 : 7[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235653:236140 [5] NCCL INFO Channel 01/0 : 29[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272151:272646 [1] NCCL INFO Channel 00/0 : 1[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557267:1557761 [1] NCCL INFO Channel 02/0 : 17[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235652:236147 [4] NCCL INFO Channel 03/0 : 28[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636084:2636605 [0] NCCL INFO Channel 01/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557270:1557763 [4] NCCL INFO Channel 06/0 : 20[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235655:236142 [7] NCCL INFO Channel 11/0 : 31[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235648:236145 [0] NCCL INFO Channel 09/0 : 24[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272155:272652 [5] NCCL INFO Channel 00/0 : 5[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272150:272645 [0] NCCL INFO Channel 06/0 : 0[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272157:272649 [7] NCCL INFO Channel 10/0 : 7[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272154:272650 [4] NCCL INFO Channel 02/0 : 4[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636090:2636598 [6] NCCL INFO Channel 03/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557267:1557761 [1] NCCL INFO Channel 04/0 : 17[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557270:1557763 [4] NCCL INFO Channel 08/0 : 20[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235649:236143 [1] NCCL INFO Channel 03/0 : 25[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272151:272646 [1] NCCL INFO Channel 02/0 : 1[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235653:236140 [5] NCCL INFO Channel 03/0 : 29[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272155:272652 [5] NCCL INFO Channel 02/0 : 5[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235652:236147 [4] NCCL INFO Channel 05/0 : 28[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235655:236142 [7] NCCL INFO Channel 13/0 : 31[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272157:272649 [7] NCCL INFO Channel 12/0 : 7[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636084:2636605 [0] NCCL INFO Channel 02/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235648:236145 [0] NCCL INFO Channel 11/0 : 24[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557270:1557763 [4] NCCL INFO Channel 10/0 : 20[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557267:1557761 [1] NCCL INFO Channel 06/0 : 17[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272154:272650 [4] NCCL INFO Channel 06/0 : 4[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272151:272646 [1] NCCL INFO Channel 04/0 : 1[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636090:2636598 [6] NCCL INFO Channel 04/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636086:2636602 [2] NCCL INFO Channel 00/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235649:236143 [1] NCCL INFO Channel 05/0 : 25[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636084:2636605 [0] NCCL INFO Channel 03/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235654:236146 [6] NCCL INFO Channel 01/0 : 30[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235653:236140 [5] NCCL INFO Channel 05/0 : 29[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272150:272645 [0] NCCL INFO Channel 08/0 : 0[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557270:1557763 [4] NCCL INFO Channel 12/0 : 20[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272156:272648 [6] NCCL INFO Channel 00/0 : 6[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272153:272651 [3] NCCL INFO Channel 00/0 : 3[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272155:272652 [5] NCCL INFO Channel 04/0 : 5[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272157:272649 [7] NCCL INFO Channel 14/0 : 7[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272152:272647 [2] NCCL INFO Channel 00/0 : 2[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636088:2636601 [4] NCCL INFO Channel 00/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235655:236142 [7] NCCL INFO Channel 01/0 : 15[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636086:2636602 [2] NCCL INFO Channel 01/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235650:236144 [2] NCCL INFO Channel 01/0 : 26[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636090:2636598 [6] NCCL INFO Channel 05/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235652:236147 [4] NCCL INFO Channel 07/0 : 28[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235648:236145 [0] NCCL INFO Channel 13/0 : 24[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557267:1557761 [1] NCCL INFO Channel 08/0 : 17[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235649:236143 [1] NCCL INFO Channel 07/0 : 25[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235651:236141 [3] NCCL INFO Channel 01/0 : 27[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235654:236146 [6] NCCL INFO Channel 03/0 : 30[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272154:272650 [4] NCCL INFO Channel 08/0 : 4[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235653:236140 [5] NCCL INFO Channel 07/0 : 29[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272151:272646 [1] NCCL INFO Channel 06/0 : 1[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272150:272645 [0] NCCL INFO Channel 10/0 : 0[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636084:2636605 [0] NCCL INFO Channel 04/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636088:2636601 [4] NCCL INFO Channel 01/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636089:2636600 [5] NCCL INFO Channel 00/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557270:1557763 [4] NCCL INFO Channel 14/0 : 20[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557267:1557761 [1] NCCL INFO Channel 10/0 : 17[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235655:236142 [7] NCCL INFO Channel 03/0 : 15[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235652:236147 [4] NCCL INFO Channel 09/0 : 28[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272155:272652 [5] NCCL INFO Channel 06/0 : 5[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272153:272651 [3] NCCL INFO Channel 02/0 : 3[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235650:236144 [2] NCCL INFO Channel 03/0 : 26[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272156:272648 [6] NCCL INFO Channel 02/0 : 6[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272157:272649 [7] NCCL INFO Channel 00/0 : 23[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235648:236145 [0] NCCL INFO Channel 15/0 : 24[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235654:236146 [6] NCCL INFO Channel 05/0 : 30[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235649:236143 [1] NCCL INFO Channel 11/0 : 25[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235651:236141 [3] NCCL INFO Channel 03/0 : 27[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636090:2636598 [6] NCCL INFO Channel 07/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636086:2636602 [2] NCCL INFO Channel 03/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557270:1557763 [4] NCCL INFO Channel 00/0 : 4[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235653:236140 [5] NCCL INFO Channel 09/0 : 29[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636084:2636605 [0] NCCL INFO Channel 05/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557267:1557761 [1] NCCL INFO Channel 12/0 : 17[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235655:236142 [7] NCCL INFO Channel 05/0 : 15[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272154:272650 [4] NCCL INFO Channel 10/0 : 4[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272151:272646 [1] NCCL INFO Channel 08/0 : 1[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272152:272647 [2] NCCL INFO Channel 04/0 : 2[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272150:272645 [0] NCCL INFO Channel 12/0 : 0[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636088:2636601 [4] NCCL INFO Channel 02/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557266:1557759 [0] NCCL INFO Channel 02/0 : 16[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557270:1557763 [4] NCCL INFO Channel 02/0 : 4[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636089:2636600 [5] NCCL INFO Channel 01/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235652:236147 [4] NCCL INFO Channel 11/0 : 28[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636091:2636599 [7] NCCL INFO Channel 00/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636090:2636598 [6] NCCL INFO Channel 08/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235650:236144 [2] NCCL INFO Channel 05/0 : 26[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636086:2636602 [2] NCCL INFO Channel 04/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235648:236145 [0] NCCL INFO Channel 01/0 : 8[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235654:236146 [6] NCCL INFO Channel 07/0 : 30[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272155:272652 [5] NCCL INFO Channel 08/0 : 5[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557267:1557761 [1] NCCL INFO Channel 14/0 : 17[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272157:272649 [7] NCCL INFO Channel 02/0 : 23[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272156:272648 [6] NCCL INFO Channel 04/0 : 6[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557273:1557760 [7] NCCL INFO Channel 00/0 : 23[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557270:1557763 [4] NCCL INFO Channel 06/0 : 4[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636084:2636605 [0] NCCL INFO Channel 06/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235653:236140 [5] NCCL INFO Channel 11/0 : 29[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636088:2636601 [4] NCCL INFO Channel 03/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235655:236142 [7] NCCL INFO Channel 07/0 : 15[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235651:236141 [3] NCCL INFO Channel 05/0 : 27[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235649:236143 [1] NCCL INFO Channel 13/0 : 25[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272153:272651 [3] NCCL INFO Channel 04/0 : 3[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636089:2636600 [5] NCCL INFO Channel 02/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235652:236147 [4] NCCL INFO Channel 13/0 : 28[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272151:272646 [1] NCCL INFO Channel 10/0 : 1[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272154:272650 [4] NCCL INFO Channel 12/0 : 4[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235650:236144 [2] NCCL INFO Channel 07/0 : 26[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272150:272645 [0] NCCL INFO Channel 14/0 : 0[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272152:272647 [2] NCCL INFO Channel 06/0 : 2[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272155:272652 [5] NCCL INFO Channel 10/0 : 5[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557266:1557759 [0] NCCL INFO Channel 04/0 : 16[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235648:236145 [0] NCCL INFO Channel 03/0 : 8[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272157:272649 [7] NCCL INFO Channel 04/0 : 23[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557267:1557761 [1] NCCL INFO Channel 00/0 : 1[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235654:236146 [6] NCCL INFO Channel 09/0 : 30[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557273:1557760 [7] NCCL INFO Channel 02/0 : 23[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557270:1557763 [4] NCCL INFO Channel 08/0 : 4[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636090:2636598 [6] NCCL INFO Channel 09/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636091:2636599 [7] NCCL INFO Channel 01/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272156:272648 [6] NCCL INFO Channel 08/0 : 6[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235653:236140 [5] NCCL INFO Channel 15/0 : 29[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235655:236142 [7] NCCL INFO Channel 09/0 : 15[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557267:1557761 [1] NCCL INFO Channel 02/0 : 1[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235649:236143 [1] NCCL INFO Channel 15/0 : 25[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235652:236147 [4] NCCL INFO Channel 15/0 : 28[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235651:236141 [3] NCCL INFO Channel 07/0 : 27[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235650:236144 [2] NCCL INFO Channel 09/0 : 26[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636086:2636602 [2] NCCL INFO Channel 05/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557266:1557759 [0] NCCL INFO Channel 06/0 : 16[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235648:236145 [0] NCCL INFO Channel 05/0 : 8[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235654:236146 [6] NCCL INFO Channel 11/0 : 30[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272153:272651 [3] NCCL INFO Channel 06/0 : 3[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636089:2636600 [5] NCCL INFO Channel 03/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636088:2636601 [4] NCCL INFO Channel 05/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636084:2636605 [0] NCCL INFO Channel 07/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557270:1557763 [4] NCCL INFO Channel 10/0 : 4[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557273:1557760 [7] NCCL INFO Channel 04/0 : 23[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235653:236140 [5] NCCL INFO Channel 01/0 : 13[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272154:272650 [4] NCCL INFO Channel 14/0 : 4[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272150:272645 [0] NCCL INFO Channel 02/0 : 16[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272151:272646 [1] NCCL INFO Channel 12/0 : 1[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272152:272647 [2] NCCL INFO Channel 08/0 : 2[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272155:272652 [5] NCCL INFO Channel 12/0 : 5[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272157:272649 [7] NCCL INFO Channel 06/0 : 23[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272156:272648 [6] NCCL INFO Channel 10/0 : 6[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636085:2636604 [1] NCCL INFO Channel 00/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636087:2636603 [3] NCCL INFO Channel 00/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636090:2636598 [6] NCCL INFO Channel 10/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636091:2636599 [7] NCCL INFO Channel 02/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557267:1557761 [1] NCCL INFO Channel 04/0 : 1[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557266:1557759 [0] NCCL INFO Channel 08/0 : 16[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235655:236142 [7] NCCL INFO Channel 11/0 : 15[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636086:2636602 [2] NCCL INFO Channel 06/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557270:1557763 [4] NCCL INFO Channel 12/0 : 4[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557273:1557760 [7] NCCL INFO Channel 06/0 : 23[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235649:236143 [1] NCCL INFO Channel 01/0 : 9[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272153:272651 [3] NCCL INFO Channel 08/0 : 3[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272154:272650 [4] NCCL INFO Channel 00/0 : 20[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636084:2636605 [0] NCCL INFO Channel 09/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235652:236147 [4] NCCL INFO Channel 01/0 : 12[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235651:236141 [3] NCCL INFO Channel 09/0 : 27[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557266:1557759 [0] NCCL INFO Channel 10/0 : 16[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235650:236144 [2] NCCL INFO Channel 11/0 : 26[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235653:236140 [5] NCCL INFO Channel 03/0 : 13[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235648:236145 [0] NCCL INFO Channel 07/0 : 8[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235654:236146 [6] NCCL INFO Channel 13/0 : 30[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235655:236142 [7] NCCL INFO Channel 13/0 : 15[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272150:272645 [0] NCCL INFO Channel 04/0 : 16[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272151:272646 [1] NCCL INFO Channel 14/0 : 1[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235649:236143 [1] NCCL INFO Channel 03/0 : 9[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272155:272652 [5] NCCL INFO Channel 14/0 : 5[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272157:272649 [7] NCCL INFO Channel 08/0 : 23[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272156:272648 [6] NCCL INFO Channel 12/0 : 6[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272152:272647 [2] NCCL INFO Channel 10/0 : 2[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636088:2636601 [4] NCCL INFO Channel 06/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636089:2636600 [5] NCCL INFO Channel 04/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636090:2636598 [6] NCCL INFO Channel 11/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557273:1557760 [7] NCCL INFO Channel 08/0 : 23[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557267:1557761 [1] NCCL INFO Channel 06/0 : 1[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557270:1557763 [4] NCCL INFO Channel 14/0 : 4[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235652:236147 [4] NCCL INFO Channel 03/0 : 12[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272153:272651 [3] NCCL INFO Channel 10/0 : 3[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235651:236141 [3] NCCL INFO Channel 13/0 : 27[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272154:272650 [4] NCCL INFO Channel 02/0 : 20[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235650:236144 [2] NCCL INFO Channel 13/0 : 26[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272150:272645 [0] NCCL INFO Channel 06/0 : 16[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272151:272646 [1] NCCL INFO Channel 00/0 : 17[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557266:1557759 [0] NCCL INFO Channel 12/0 : 16[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636091:2636599 [7] NCCL INFO Channel 03/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636086:2636602 [2] NCCL INFO Channel 07/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636085:2636604 [1] NCCL INFO Channel 02/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636084:2636605 [0] NCCL INFO Channel 10/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636087:2636603 [3] NCCL INFO Channel 01/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557267:1557761 [1] NCCL INFO Channel 08/0 : 1[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557273:1557760 [7] NCCL INFO Channel 10/0 : 23[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235648:236145 [0] NCCL INFO Channel 09/0 : 8[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235653:236140 [5] NCCL INFO Channel 05/0 : 13[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235654:236146 [6] NCCL INFO Channel 15/0 : 30[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235649:236143 [1] NCCL INFO Channel 05/0 : 9[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272157:272649 [7] NCCL INFO Channel 10/0 : 23[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235652:236147 [4] NCCL INFO Channel 05/0 : 12[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272156:272648 [6] NCCL INFO Channel 14/0 : 6[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636088:2636601 [4] NCCL INFO Channel 07/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235651:236141 [3] NCCL INFO Channel 15/0 : 27[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272155:272652 [5] NCCL INFO Channel 00/0 : 21[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636090:2636598 [6] NCCL INFO Channel 12/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636089:2636600 [5] NCCL INFO Channel 06/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557273:1557760 [7] NCCL INFO Channel 12/0 : 23[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557266:1557759 [0] NCCL INFO Channel 14/0 : 16[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235650:236144 [2] NCCL INFO Channel 15/0 : 26[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272152:272647 [2] NCCL INFO Channel 12/0 : 2[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272154:272650 [4] NCCL INFO Channel 06/0 : 20[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722052:2722546 [7] NCCL INFO Channel 00/0 : 31[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636086:2636602 [2] NCCL INFO Channel 08/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636091:2636599 [7] NCCL INFO Channel 04/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557267:1557761 [1] NCCL INFO Channel 10/0 : 1[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235648:236145 [0] NCCL INFO Channel 11/0 : 8[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557273:1557760 [7] NCCL INFO Channel 14/0 : 23[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235654:236146 [6] NCCL INFO Channel 01/0 : 14[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235653:236140 [5] NCCL INFO Channel 07/0 : 13[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272153:272651 [3] NCCL INFO Channel 12/0 : 3[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-52:2722052:2722546 [7] NCCL INFO Channel 02/0 : 31[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235652:236147 [4] NCCL INFO Channel 07/0 : 12[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272150:272645 [0] NCCL INFO Channel 08/0 : 16[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272151:272646 [1] NCCL INFO Channel 02/0 : 17[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272157:272649 [7] NCCL INFO Channel 12/0 : 23[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235649:236143 [1] NCCL INFO Channel 07/0 : 9[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272156:272648 [6] NCCL INFO Channel 00/0 : 22[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272155:272652 [5] NCCL INFO Channel 02/0 : 21[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636084:2636605 [0] NCCL INFO Channel 11/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557266:1557759 [0] NCCL INFO Channel 02/0 : 0[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-52:2722052:2722546 [7] NCCL INFO Channel 04/0 : 31[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557267:1557761 [1] NCCL INFO Channel 12/0 : 1[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557273:1557760 [7] NCCL INFO Channel 00/0 : 7[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235655:236142 [7] NCCL INFO Channel 00/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636087:2636603 [3] NCCL INFO Channel 02/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636090:2636598 [6] NCCL INFO Channel 13/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636088:2636601 [4] NCCL INFO Channel 08/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636089:2636600 [5] NCCL INFO Channel 07/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636085:2636604 [1] NCCL INFO Channel 03/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235651:236141 [3] NCCL INFO Channel 01/0 : 11[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235650:236144 [2] NCCL INFO Channel 01/0 : 10[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235648:236145 [0] NCCL INFO Channel 13/0 : 8[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272154:272650 [4] NCCL INFO Channel 08/0 : 20[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272152:272647 [2] NCCL INFO Channel 14/0 : 2[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636086:2636602 [2] NCCL INFO Channel 09/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636091:2636599 [7] NCCL INFO Channel 05/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235654:236146 [6] NCCL INFO Channel 03/0 : 14[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272153:272651 [3] NCCL INFO Channel 14/0 : 3[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272157:272649 [7] NCCL INFO Channel 14/0 : 23[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722052:2722546 [7] NCCL INFO Channel 06/0 : 31[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557266:1557759 [0] NCCL INFO Channel 04/0 : 0[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272150:272645 [0] NCCL INFO Channel 10/0 : 16[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272151:272646 [1] NCCL INFO Channel 04/0 : 17[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557267:1557761 [1] NCCL INFO Channel 14/0 : 1[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272156:272648 [6] NCCL INFO Channel 02/0 : 22[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272155:272652 [5] NCCL INFO Channel 04/0 : 21[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557273:1557760 [7] NCCL INFO Channel 02/0 : 7[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272154:272650 [4] NCCL INFO Channel 10/0 : 20[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636084:2636605 [0] NCCL INFO Channel 12/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235653:236140 [5] NCCL INFO Channel 09/0 : 13[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235652:236147 [4] NCCL INFO Channel 09/0 : 12[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272152:272647 [2] NCCL INFO Channel 00/0 : 18[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235649:236143 [1] NCCL INFO Channel 11/0 : 9[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235655:236142 [7] NCCL INFO Channel 01/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235650:236144 [2] NCCL INFO Channel 03/0 : 10[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235651:236141 [3] NCCL INFO Channel 03/0 : 11[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235648:236145 [0] NCCL INFO Channel 15/0 : 8[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-52:2722052:2722546 [7] NCCL INFO Channel 08/0 : 31[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235654:236146 [6] NCCL INFO Channel 05/0 : 14[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557266:1557759 [0] NCCL INFO Channel 06/0 : 0[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235653:236140 [5] NCCL INFO Channel 11/0 : 13[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557273:1557760 [7] NCCL INFO Channel 04/0 : 7[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235652:236147 [4] NCCL INFO Channel 11/0 : 12[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272153:272651 [3] NCCL INFO Channel 00/0 : 19[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636090:2636598 [6] NCCL INFO Channel 15/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636089:2636600 [5] NCCL INFO Channel 08/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272150:272645 [0] NCCL INFO Channel 12/0 : 16[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636088:2636601 [4] NCCL INFO Channel 09/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272156:272648 [6] NCCL INFO Channel 04/0 : 22[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-52:2722052:2722546 [7] NCCL INFO Channel 10/0 : 31[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557266:1557759 [0] NCCL INFO Channel 08/0 : 0[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272151:272646 [1] NCCL INFO Channel 06/0 : 17[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557273:1557760 [7] NCCL INFO Channel 06/0 : 7[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235649:236143 [1] NCCL INFO Channel 13/0 : 9[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272155:272652 [5] NCCL INFO Channel 06/0 : 21[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235655:236142 [7] NCCL INFO Channel 02/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235650:236144 [2] NCCL INFO Channel 05/0 : 10[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272154:272650 [4] NCCL INFO Channel 12/0 : 20[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636085:2636604 [1] NCCL INFO Channel 04/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636087:2636603 [3] NCCL INFO Channel 04/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235654:236146 [6] NCCL INFO Channel 07/0 : 14[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272152:272647 [2] NCCL INFO Channel 04/0 : 18[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636091:2636599 [7] NCCL INFO Channel 06/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636086:2636602 [2] NCCL INFO Channel 11/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636084:2636605 [0] NCCL INFO Channel 13/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235651:236141 [3] NCCL INFO Channel 05/0 : 11[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235653:236140 [5] NCCL INFO Channel 15/0 : 13[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235652:236147 [4] NCCL INFO Channel 13/0 : 12[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272153:272651 [3] NCCL INFO Channel 02/0 : 19[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636090:2636598 [6] NCCL INFO Channel 00/0 : 22[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-52:2722052:2722546 [7] NCCL INFO Channel 12/0 : 31[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557266:1557759 [0] NCCL INFO Channel 10/0 : 0[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235649:236143 [1] NCCL INFO Channel 15/0 : 9[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272150:272645 [0] NCCL INFO Channel 14/0 : 16[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636089:2636600 [5] NCCL INFO Channel 09/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636088:2636601 [4] NCCL INFO Channel 10/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557273:1557760 [7] NCCL INFO Channel 08/0 : 7[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272156:272648 [6] NCCL INFO Channel 08/0 : 22[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636087:2636603 [3] NCCL INFO Channel 05/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-52:2722045:2722544 [0] NCCL INFO Channel 02/0 : 24[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557266:1557759 [0] NCCL INFO Channel 12/0 : 0[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235655:236142 [7] NCCL INFO Channel 03/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272151:272646 [1] NCCL INFO Channel 08/0 : 17[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636091:2636599 [7] NCCL INFO Channel 08/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722052:2722546 [7] NCCL INFO Channel 14/0 : 31[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557273:1557760 [7] NCCL INFO Channel 10/0 : 7[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235654:236146 [6] NCCL INFO Channel 09/0 : 14[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235650:236144 [2] NCCL INFO Channel 07/0 : 10[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272155:272652 [5] NCCL INFO Channel 08/0 : 21[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235652:236147 [4] NCCL INFO Channel 15/0 : 12[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235648:236145 [0] NCCL INFO Channel 01/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235651:236141 [3] NCCL INFO Channel 07/0 : 11[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272154:272650 [4] NCCL INFO Channel 14/0 : 20[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272153:272651 [3] NCCL INFO Channel 04/0 : 19[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272152:272647 [2] NCCL INFO Channel 06/0 : 18[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272156:272648 [6] NCCL INFO Channel 10/0 : 22[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636085:2636604 [1] NCCL INFO Channel 05/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272151:272646 [1] NCCL INFO Channel 10/0 : 17[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636090:2636598 [6] NCCL INFO Channel 02/0 : 22[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-52:2722045:2722544 [0] NCCL INFO Channel 04/0 : 24[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557266:1557759 [0] NCCL INFO Channel 14/0 : 0[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272155:272652 [5] NCCL INFO Channel 10/0 : 21[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636086:2636602 [2] NCCL INFO Channel 12/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636089:2636600 [5] NCCL INFO Channel 10/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636084:2636605 [0] NCCL INFO Channel 14/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-52:2722045:2722544 [0] NCCL INFO Channel 06/0 : 24[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557273:1557760 [7] NCCL INFO Channel 12/0 : 7[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636088:2636601 [4] NCCL INFO Channel 11/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235655:236142 [7] NCCL INFO Channel 04/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235654:236146 [6] NCCL INFO Channel 11/0 : 14[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235650:236144 [2] NCCL INFO Channel 09/0 : 10[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235651:236141 [3] NCCL INFO Channel 09/0 : 11[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272152:272647 [2] NCCL INFO Channel 08/0 : 18[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557273:1557760 [7] NCCL INFO Channel 14/0 : 7[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272153:272651 [3] NCCL INFO Channel 06/0 : 19[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-52:2722046:2722541 [1] NCCL INFO Channel 00/0 : 25[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557270:1557763 [4] NCCL INFO Channel 01/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272156:272648 [6] NCCL INFO Channel 12/0 : 22[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636091:2636599 [7] NCCL INFO Channel 09/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722050:2722548 [5] NCCL INFO Channel 00/0 : 29[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272151:272646 [1] NCCL INFO Channel 12/0 : 17[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636087:2636603 [3] NCCL INFO Channel 06/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-52:2722045:2722544 [0] NCCL INFO Channel 08/0 : 24[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272155:272652 [5] NCCL INFO Channel 12/0 : 21[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235655:236142 [7] NCCL INFO Channel 05/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272152:272647 [2] NCCL INFO Channel 10/0 : 18[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235648:236145 [0] NCCL INFO Channel 02/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272153:272651 [3] NCCL INFO Channel 08/0 : 19[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557270:1557763 [4] NCCL INFO Channel 03/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235654:236146 [6] NCCL INFO Channel 13/0 : 14[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272156:272648 [6] NCCL INFO Channel 14/0 : 22[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235649:236143 [1] NCCL INFO Channel 00/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272155:272652 [5] NCCL INFO Channel 14/0 : 21[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272151:272646 [1] NCCL INFO Channel 14/0 : 17[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636090:2636598 [6] NCCL INFO Channel 04/0 : 22[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235650:236144 [2] NCCL INFO Channel 11/0 : 10[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235653:236140 [5] NCCL INFO Channel 00/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235651:236141 [3] NCCL INFO Channel 13/0 : 11[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272154:272650 [4] NCCL INFO Channel 00/0 : 28[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636085:2636604 [1] NCCL INFO Channel 06/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-52:2722046:2722541 [1] NCCL INFO Channel 02/0 : 25[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-52:2722050:2722548 [5] NCCL INFO Channel 02/0 : 29[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636088:2636601 [4] NCCL INFO Channel 13/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636089:2636600 [5] NCCL INFO Channel 11/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636084:2636605 [0] NCCL INFO Channel 15/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-52:2722045:2722544 [0] NCCL INFO Channel 10/0 : 24[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636086:2636602 [2] NCCL INFO Channel 13/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636091:2636599 [7] NCCL INFO Channel 10/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557266:1557759 [0] NCCL INFO Channel 01/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557270:1557763 [4] NCCL INFO Channel 05/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235652:236147 [4] NCCL INFO Channel 00/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235655:236142 [7] NCCL INFO Channel 06/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235648:236145 [0] NCCL INFO Channel 03/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235654:236146 [6] NCCL INFO Channel 15/0 : 14[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272152:272647 [2] NCCL INFO Channel 12/0 : 18[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272153:272651 [3] NCCL INFO Channel 10/0 : 19[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-52:2722049:2722547 [4] NCCL INFO Channel 00/0 : 28[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235651:236141 [3] NCCL INFO Channel 15/0 : 11[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272154:272650 [4] NCCL INFO Channel 02/0 : 28[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636087:2636603 [3] NCCL INFO Channel 07/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235653:236140 [5] NCCL INFO Channel 01/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235650:236144 [2] NCCL INFO Channel 13/0 : 10[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235649:236143 [1] NCCL INFO Channel 02/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272153:272651 [3] NCCL INFO Channel 12/0 : 19[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272152:272647 [2] NCCL INFO Channel 14/0 : 18[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636090:2636598 [6] NCCL INFO Channel 08/0 : 22[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272154:272650 [4] NCCL INFO Channel 06/0 : 28[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636085:2636604 [1] NCCL INFO Channel 07/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272153:272651 [3] NCCL INFO Channel 14/0 : 19[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-52:2722046:2722541 [1] NCCL INFO Channel 04/0 : 25[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-52:2722050:2722548 [5] NCCL INFO Channel 04/0 : 29[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-52:2722045:2722544 [0] NCCL INFO Channel 12/0 : 24[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-52:2722049:2722547 [4] NCCL INFO Channel 02/0 : 28[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557266:1557759 [0] NCCL INFO Channel 03/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235652:236147 [4] NCCL INFO Channel 01/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235655:236142 [7] NCCL INFO Channel 08/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235648:236145 [0] NCCL INFO Channel 04/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272150:272645 [0] NCCL INFO Channel 02/0 : 24[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636089:2636600 [5] NCCL INFO Channel 12/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636088:2636601 [4] NCCL INFO Channel 14/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62883:63373 [2] NCCL INFO Channel 01/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272151:272646 [1] NCCL INFO Channel 00/0 : 25[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272154:272650 [4] NCCL INFO Channel 08/0 : 28[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557270:1557763 [4] NCCL INFO Channel 07/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272150:272645 [0] NCCL INFO Channel 04/0 : 24[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557273:1557760 [7] NCCL INFO Channel 01/0 : 15[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557267:1557761 [1] NCCL INFO Channel 01/0 : 9[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235650:236144 [2] NCCL INFO Channel 15/0 : 10[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235653:236140 [5] NCCL INFO Channel 02/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636084:2636605 [0] NCCL INFO Channel 02/0 : 16[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636091:2636599 [7] NCCL INFO Channel 11/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636086:2636602 [2] NCCL INFO Channel 14/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636090:2636598 [6] NCCL INFO Channel 10/0 : 22[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235649:236143 [1] NCCL INFO Channel 03/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636087:2636603 [3] NCCL INFO Channel 08/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636085:2636604 [1] NCCL INFO Channel 08/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62888:63380 [7] NCCL INFO Channel 01/0 : 7[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235655:236142 [7] NCCL INFO Channel 09/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235652:236147 [4] NCCL INFO Channel 02/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235648:236145 [0] NCCL INFO Channel 05/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62883:63373 [2] NCCL INFO Channel 03/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272157:272649 [7] NCCL INFO Channel 00/0 : 31[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557266:1557759 [0] NCCL INFO Channel 05/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272151:272646 [1] NCCL INFO Channel 02/0 : 25[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557270:1557763 [4] NCCL INFO Channel 09/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272154:272650 [4] NCCL INFO Channel 10/0 : 28[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557267:1557761 [1] NCCL INFO Channel 03/0 : 9[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557273:1557760 [7] NCCL INFO Channel 03/0 : 15[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272150:272645 [0] NCCL INFO Channel 06/0 : 24[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636088:2636601 [4] NCCL INFO Channel 15/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636089:2636600 [5] NCCL INFO Channel 14/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557266:1557759 [0] NCCL INFO Channel 07/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235653:236140 [5] NCCL INFO Channel 03/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272157:272649 [7] NCCL INFO Channel 02/0 : 31[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62888:63380 [7] NCCL INFO Channel 03/0 : 7[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235649:236143 [1] NCCL INFO Channel 04/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62883:63373 [2] NCCL INFO Channel 05/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235651:236141 [3] NCCL INFO Channel 00/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636091:2636599 [7] NCCL INFO Channel 12/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636084:2636605 [0] NCCL INFO Channel 04/0 : 16[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636090:2636598 [6] NCCL INFO Channel 12/0 : 22[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557270:1557763 [4] NCCL INFO Channel 11/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272151:272646 [1] NCCL INFO Channel 04/0 : 25[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272154:272650 [4] NCCL INFO Channel 12/0 : 28[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62887:63379 [6] NCCL INFO Channel 01/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62888:63380 [7] NCCL INFO Channel 05/0 : 7[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557267:1557761 [1] NCCL INFO Channel 05/0 : 9[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557273:1557760 [7] NCCL INFO Channel 05/0 : 15[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235655:236142 [7] NCCL INFO Channel 10/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235652:236147 [4] NCCL INFO Channel 03/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235648:236145 [0] NCCL INFO Channel 06/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235654:236146 [6] NCCL INFO Channel 00/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272150:272645 [0] NCCL INFO Channel 08/0 : 24[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235653:236140 [5] NCCL INFO Channel 04/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272157:272649 [7] NCCL INFO Channel 04/0 : 31[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235649:236143 [1] NCCL INFO Channel 05/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272154:272650 [4] NCCL INFO Channel 14/0 : 28[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272151:272646 [1] NCCL INFO Channel 06/0 : 25[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636086:2636602 [2] NCCL INFO Channel 15/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235650:236144 [2] NCCL INFO Channel 00/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235651:236141 [3] NCCL INFO Channel 01/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636085:2636604 [1] NCCL INFO Channel 10/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-52:2722047:2722542 [2] NCCL INFO Channel 00/0 : 26[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636087:2636603 [3] NCCL INFO Channel 09/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636088:2636601 [4] NCCL INFO Channel 00/0 : 20[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62883:63373 [2] NCCL INFO Channel 07/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636089:2636600 [5] NCCL INFO Channel 15/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557266:1557759 [0] NCCL INFO Channel 09/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557270:1557763 [4] NCCL INFO Channel 13/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636091:2636599 [7] NCCL INFO Channel 13/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557267:1557761 [1] NCCL INFO Channel 07/0 : 9[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272150:272645 [0] NCCL INFO Channel 10/0 : 24[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557273:1557760 [7] NCCL INFO Channel 07/0 : 15[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235655:236142 [7] NCCL INFO Channel 11/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272157:272649 [7] NCCL INFO Channel 06/0 : 31[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557266:1557759 [0] NCCL INFO Channel 11/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235652:236147 [4] NCCL INFO Channel 05/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272154:272650 [4] NCCL INFO Channel 00/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272151:272646 [1] NCCL INFO Channel 08/0 : 25[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62888:63380 [7] NCCL INFO Channel 07/0 : 7[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272150:272645 [0] NCCL INFO Channel 12/0 : 24[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272157:272649 [7] NCCL INFO Channel 08/0 : 31[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636084:2636605 [0] NCCL INFO Channel 06/0 : 16[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636090:2636598 [6] NCCL INFO Channel 14/0 : 22[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-52:2722047:2722542 [2] NCCL INFO Channel 04/0 : 26[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62885:63377 [4] NCCL INFO Channel 01/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62887:63379 [6] NCCL INFO Channel 03/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62883:63373 [2] NCCL INFO Channel 09/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557270:1557763 [4] NCCL INFO Channel 15/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235648:236145 [0] NCCL INFO Channel 07/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235654:236146 [6] NCCL INFO Channel 01/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557267:1557761 [1] NCCL INFO Channel 11/0 : 9[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235649:236143 [1] NCCL INFO Channel 06/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235653:236140 [5] NCCL INFO Channel 06/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557273:1557760 [7] NCCL INFO Channel 09/0 : 15[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557266:1557759 [0] NCCL INFO Channel 13/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235650:236144 [2] NCCL INFO Channel 01/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235651:236141 [3] NCCL INFO Channel 02/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235655:236142 [7] NCCL INFO Channel 12/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272154:272650 [4] NCCL INFO Channel 01/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636086:2636602 [2] NCCL INFO Channel 00/0 : 18[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636085:2636604 [1] NCCL INFO Channel 11/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557267:1557761 [1] NCCL INFO Channel 13/0 : 9[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272151:272646 [1] NCCL INFO Channel 10/0 : 25[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636088:2636601 [4] NCCL INFO Channel 02/0 : 20[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722047:2722542 [2] NCCL INFO Channel 06/0 : 26[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62881:63375 [0] NCCL INFO Channel 01/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62888:63380 [7] NCCL INFO Channel 09/0 : 7[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272150:272645 [0] NCCL INFO Channel 14/0 : 24[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272157:272649 [7] NCCL INFO Channel 10/0 : 31[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636089:2636600 [5] NCCL INFO Channel 00/0 : 21[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636091:2636599 [7] NCCL INFO Channel 14/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636087:2636603 [3] NCCL INFO Channel 10/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62885:63377 [4] NCCL INFO Channel 03/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62887:63379 [6] NCCL INFO Channel 05/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62883:63373 [2] NCCL INFO Channel 11/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235652:236147 [4] NCCL INFO Channel 06/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722047:2722542 [2] NCCL INFO Channel 08/0 : 26[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272154:272650 [4] NCCL INFO Channel 02/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62888:63380 [7] NCCL INFO Channel 11/0 : 7[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272151:272646 [1] NCCL INFO Channel 12/0 : 25[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636084:2636605 [0] NCCL INFO Channel 08/0 : 16[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62881:63375 [0] NCCL INFO Channel 03/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235648:236145 [0] NCCL INFO Channel 09/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235654:236146 [6] NCCL INFO Channel 02/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636086:2636602 [2] NCCL INFO Channel 04/0 : 18[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62885:63377 [4] NCCL INFO Channel 05/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235653:236140 [5] NCCL INFO Channel 07/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-52:2722047:2722542 [2] NCCL INFO Channel 10/0 : 26[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235649:236143 [1] NCCL INFO Channel 07/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235655:236142 [7] NCCL INFO Channel 13/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235650:236144 [2] NCCL INFO Channel 03/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235651:236141 [3] NCCL INFO Channel 04/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272150:272645 [0] NCCL INFO Channel 01/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272157:272649 [7] NCCL INFO Channel 12/0 : 31[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272154:272650 [4] NCCL INFO Channel 03/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272151:272646 [1] NCCL INFO Channel 14/0 : 25[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62887:63379 [6] NCCL INFO Channel 07/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62886:63378 [5] NCCL INFO Channel 01/0 : 5[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272150:272645 [0] NCCL INFO Channel 02/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636085:2636604 [1] NCCL INFO Channel 12/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62883:63373 [2] NCCL INFO Channel 13/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62884:63376 [3] NCCL INFO Channel 01/0 : 3[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62888:63380 [7] NCCL INFO Channel 13/0 : 7[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636088:2636601 [4] NCCL INFO Channel 06/0 : 20[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636091:2636599 [7] NCCL INFO Channel 00/0 : 23[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62885:63377 [4] NCCL INFO Channel 07/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62881:63375 [0] NCCL INFO Channel 05/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235652:236147 [4] NCCL INFO Channel 07/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636089:2636600 [5] NCCL INFO Channel 02/0 : 21[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235648:236145 [0] NCCL INFO Channel 10/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636084:2636605 [0] NCCL INFO Channel 10/0 : 16[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636087:2636603 [3] NCCL INFO Channel 12/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636086:2636602 [2] NCCL INFO Channel 06/0 : 18[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-52:2722047:2722542 [2] NCCL INFO Channel 12/0 : 26[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235654:236146 [6] NCCL INFO Channel 03/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235653:236140 [5] NCCL INFO Channel 08/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272157:272649 [7] NCCL INFO Channel 14/0 : 31[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272154:272650 [4] NCCL INFO Channel 05/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272151:272646 [1] NCCL INFO Channel 00/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272150:272645 [0] NCCL INFO Channel 03/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62887:63379 [6] NCCL INFO Channel 09/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272157:272649 [7] NCCL INFO Channel 00/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62886:63378 [5] NCCL INFO Channel 03/0 : 5[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636085:2636604 [1] NCCL INFO Channel 13/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-52:2722049:2722547 [4] NCCL INFO Channel 06/0 : 28[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636088:2636601 [4] NCCL INFO Channel 08/0 : 20[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235655:236142 [7] NCCL INFO Channel 14/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235649:236143 [1] NCCL INFO Channel 08/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235650:236144 [2] NCCL INFO Channel 04/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235652:236147 [4] NCCL INFO Channel 08/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235651:236141 [3] NCCL INFO Channel 05/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272151:272646 [1] NCCL INFO Channel 02/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272154:272650 [4] NCCL INFO Channel 06/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62883:63373 [2] NCCL INFO Channel 15/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62888:63380 [7] NCCL INFO Channel 00/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235648:236145 [0] NCCL INFO Channel 11/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235654:236146 [6] NCCL INFO Channel 04/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272150:272645 [0] NCCL INFO Channel 04/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272157:272649 [7] NCCL INFO Channel 01/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722047:2722542 [2] NCCL INFO Channel 14/0 : 26[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62882:63374 [1] NCCL INFO Channel 01/0 : 1[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235653:236140 [5] NCCL INFO Channel 09/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636091:2636599 [7] NCCL INFO Channel 02/0 : 23[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636089:2636600 [5] NCCL INFO Channel 04/0 : 21[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62885:63377 [4] NCCL INFO Channel 09/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636084:2636605 [0] NCCL INFO Channel 12/0 : 16[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62884:63376 [3] NCCL INFO Channel 03/0 : 3[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636087:2636603 [3] NCCL INFO Channel 13/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62887:63379 [6] NCCL INFO Channel 11/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62881:63375 [0] NCCL INFO Channel 07/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62886:63378 [5] NCCL INFO Channel 05/0 : 5[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272154:272650 [4] NCCL INFO Channel 07/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272151:272646 [1] NCCL INFO Channel 03/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235655:236142 [7] NCCL INFO Channel 01/0 : 15[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235649:236143 [1] NCCL INFO Channel 10/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272150:272645 [0] NCCL INFO Channel 05/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272157:272649 [7] NCCL INFO Channel 02/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722049:2722547 [4] NCCL INFO Channel 08/0 : 28[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235650:236144 [2] NCCL INFO Channel 05/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235652:236147 [4] NCCL INFO Channel 09/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722050:2722548 [5] NCCL INFO Channel 06/0 : 29[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235651:236141 [3] NCCL INFO Channel 06/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235654:236146 [6] NCCL INFO Channel 05/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235648:236145 [0] NCCL INFO Channel 12/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636085:2636604 [1] NCCL INFO Channel 14/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62888:63380 [7] NCCL INFO Channel 01/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235653:236140 [5] NCCL INFO Channel 10/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636088:2636601 [4] NCCL INFO Channel 10/0 : 20[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636086:2636602 [2] NCCL INFO Channel 08/0 : 18[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62883:63373 [2] NCCL INFO Channel 00/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636089:2636600 [5] NCCL INFO Channel 06/0 : 21[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636091:2636599 [7] NCCL INFO Channel 04/0 : 23[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62882:63374 [1] NCCL INFO Channel 03/0 : 1[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636084:2636605 [0] NCCL INFO Channel 14/0 : 16[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-52:2722046:2722541 [1] NCCL INFO Channel 06/0 : 25[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-52:2722049:2722547 [4] NCCL INFO Channel 10/0 : 28[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62885:63377 [4] NCCL INFO Channel 11/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272154:272650 [4] NCCL INFO Channel 08/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272151:272646 [1] NCCL INFO Channel 04/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636087:2636603 [3] NCCL INFO Channel 14/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-52:2722050:2722548 [5] NCCL INFO Channel 08/0 : 29[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235655:236142 [7] NCCL INFO Channel 03/0 : 15[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235649:236143 [1] NCCL INFO Channel 11/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235652:236147 [4] NCCL INFO Channel 10/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62887:63379 [6] NCCL INFO Channel 13/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62884:63376 [3] NCCL INFO Channel 05/0 : 3[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62886:63378 [5] NCCL INFO Channel 07/0 : 5[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62881:63375 [0] NCCL INFO Channel 09/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62888:63380 [7] NCCL INFO Channel 02/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272157:272649 [7] NCCL INFO Channel 03/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272150:272645 [0] NCCL INFO Channel 06/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-52:2722046:2722541 [1] NCCL INFO Channel 08/0 : 25[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62883:63373 [2] NCCL INFO Channel 01/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636085:2636604 [1] NCCL INFO Channel 15/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-52:2722049:2722547 [4] NCCL INFO Channel 12/0 : 28[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722048:2722545 [3] NCCL INFO Channel 00/0 : 27[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-52:2722050:2722548 [5] NCCL INFO Channel 10/0 : 29[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62885:63377 [4] NCCL INFO Channel 13/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62882:63374 [1] NCCL INFO Channel 05/0 : 1[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235650:236144 [2] NCCL INFO Channel 06/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235654:236146 [6] NCCL INFO Channel 07/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235653:236140 [5] NCCL INFO Channel 11/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235651:236141 [3] NCCL INFO Channel 07/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636088:2636601 [4] NCCL INFO Channel 12/0 : 20[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722046:2722541 [1] NCCL INFO Channel 10/0 : 25[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235648:236145 [0] NCCL INFO Channel 13/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636089:2636600 [5] NCCL INFO Channel 08/0 : 21[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636091:2636599 [7] NCCL INFO Channel 06/0 : 23[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636086:2636602 [2] NCCL INFO Channel 10/0 : 18[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235655:236142 [7] NCCL INFO Channel 05/0 : 15[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235649:236143 [1] NCCL INFO Channel 12/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235652:236147 [4] NCCL INFO Channel 11/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636087:2636603 [3] NCCL INFO Channel 15/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636085:2636604 [1] NCCL INFO Channel 00/0 : 17[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272154:272650 [4] NCCL INFO Channel 09/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636088:2636601 [4] NCCL INFO Channel 14/0 : 20[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272151:272646 [1] NCCL INFO Channel 05/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62887:63379 [6] NCCL INFO Channel 15/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272157:272649 [7] NCCL INFO Channel 04/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272150:272645 [0] NCCL INFO Channel 07/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62886:63378 [5] NCCL INFO Channel 09/0 : 5[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-52:2722050:2722548 [5] NCCL INFO Channel 12/0 : 29[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-52:2722048:2722545 [3] NCCL INFO Channel 02/0 : 27[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-52:2722049:2722547 [4] NCCL INFO Channel 14/0 : 28[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722046:2722541 [1] NCCL INFO Channel 12/0 : 25[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62884:63376 [3] NCCL INFO Channel 07/0 : 3[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62888:63380 [7] NCCL INFO Channel 03/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636089:2636600 [5] NCCL INFO Channel 10/0 : 21[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636091:2636599 [7] NCCL INFO Channel 08/0 : 23[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636085:2636604 [1] NCCL INFO Channel 02/0 : 17[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636086:2636602 [2] NCCL INFO Channel 12/0 : 18[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636087:2636603 [3] NCCL INFO Channel 00/0 : 19[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235654:236146 [6] NCCL INFO Channel 08/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272154:272650 [4] NCCL INFO Channel 10/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272151:272646 [1] NCCL INFO Channel 06/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62881:63375 [0] NCCL INFO Channel 11/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62883:63373 [2] NCCL INFO Channel 03/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62882:63374 [1] NCCL INFO Channel 07/0 : 1[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62885:63377 [4] NCCL INFO Channel 15/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62887:63379 [6] NCCL INFO Channel 00/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636089:2636600 [5] NCCL INFO Channel 12/0 : 21[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636091:2636599 [7] NCCL INFO Channel 10/0 : 23[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235653:236140 [5] NCCL INFO Channel 12/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235650:236144 [2] NCCL INFO Channel 07/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235648:236145 [0] NCCL INFO Channel 14/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272150:272645 [0] NCCL INFO Channel 09/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272157:272649 [7] NCCL INFO Channel 05/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235655:236142 [7] NCCL INFO Channel 07/0 : 15[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272154:272650 [4] NCCL INFO Channel 11/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722048:2722545 [3] NCCL INFO Channel 04/0 : 27[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-52:2722050:2722548 [5] NCCL INFO Channel 14/0 : 29[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-52:2722046:2722541 [1] NCCL INFO Channel 14/0 : 25[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235651:236141 [3] NCCL INFO Channel 08/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235652:236147 [4] NCCL INFO Channel 13/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235649:236143 [1] NCCL INFO Channel 13/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272151:272646 [1] NCCL INFO Channel 07/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62886:63378 [5] NCCL INFO Channel 11/0 : 5[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636086:2636602 [2] NCCL INFO Channel 14/0 : 18[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62888:63380 [7] NCCL INFO Channel 04/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62884:63376 [3] NCCL INFO Channel 09/0 : 3[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636087:2636603 [3] NCCL INFO Channel 02/0 : 19[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62881:63375 [0] NCCL INFO Channel 13/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636089:2636600 [5] NCCL INFO Channel 14/0 : 21[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636091:2636599 [7] NCCL INFO Channel 12/0 : 23[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636085:2636604 [1] NCCL INFO Channel 04/0 : 17[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235654:236146 [6] NCCL INFO Channel 09/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272150:272645 [0] NCCL INFO Channel 10/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272157:272649 [7] NCCL INFO Channel 06/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722048:2722545 [3] NCCL INFO Channel 06/0 : 27[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272154:272650 [4] NCCL INFO Channel 13/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272151:272646 [1] NCCL INFO Channel 08/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62883:63373 [2] NCCL INFO Channel 04/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62887:63379 [6] NCCL INFO Channel 01/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62882:63374 [1] NCCL INFO Channel 11/0 : 1[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62885:63377 [4] NCCL INFO Channel 00/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636087:2636603 [3] NCCL INFO Channel 04/0 : 19[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62886:63378 [5] NCCL INFO Channel 15/0 : 5[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62888:63380 [7] NCCL INFO Channel 05/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235653:236140 [5] NCCL INFO Channel 14/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636091:2636599 [7] NCCL INFO Channel 14/0 : 23[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722045:2722544 [0] NCCL INFO Channel 14/0 : 24[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62884:63376 [3] NCCL INFO Channel 13/0 : 3[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235648:236145 [0] NCCL INFO Channel 15/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235655:236142 [7] NCCL INFO Channel 09/0 : 15[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235650:236144 [2] NCCL INFO Channel 08/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636085:2636604 [1] NCCL INFO Channel 06/0 : 17[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-52:2722048:2722545 [3] NCCL INFO Channel 08/0 : 27[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62881:63375 [0] NCCL INFO Channel 15/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235652:236147 [4] NCCL INFO Channel 14/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272157:272649 [7] NCCL INFO Channel 08/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272154:272650 [4] NCCL INFO Channel 14/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235651:236141 [3] NCCL INFO Channel 09/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235649:236143 [1] NCCL INFO Channel 14/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235654:236146 [6] NCCL INFO Channel 10/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272151:272646 [1] NCCL INFO Channel 10/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272150:272645 [0] NCCL INFO Channel 11/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636087:2636603 [3] NCCL INFO Channel 06/0 : 19[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-52:2722051:2722543 [6] NCCL INFO Channel 00/0 : 30[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62883:63373 [2] NCCL INFO Channel 05/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235653:236140 [5] NCCL INFO Channel 15/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636085:2636604 [1] NCCL INFO Channel 08/0 : 17[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62887:63379 [6] NCCL INFO Channel 02/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272154:272650 [4] NCCL INFO Channel 15/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272157:272649 [7] NCCL INFO Channel 09/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636087:2636603 [3] NCCL INFO Channel 08/0 : 19[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62885:63377 [4] NCCL INFO Channel 01/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62882:63374 [1] NCCL INFO Channel 13/0 : 1[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272151:272646 [1] NCCL INFO Channel 11/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62886:63378 [5] NCCL INFO Channel 00/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272150:272645 [0] NCCL INFO Channel 12/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62888:63380 [7] NCCL INFO Channel 06/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62884:63376 [3] NCCL INFO Channel 15/0 : 3[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815967:2816473 [4] NCCL INFO Channel 01/0 : 12[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722048:2722545 [3] NCCL INFO Channel 10/0 : 27[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235655:236142 [7] NCCL INFO Channel 11/0 : 15[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235648:236145 [0] NCCL INFO Channel 01/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235652:236147 [4] NCCL INFO Channel 15/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235650:236144 [2] NCCL INFO Channel 09/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636085:2636604 [1] NCCL INFO Channel 10/0 : 17[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235649:236143 [1] NCCL INFO Channel 15/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235651:236141 [3] NCCL INFO Channel 10/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235653:236140 [5] NCCL INFO Channel 01/0 : 13[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235654:236146 [6] NCCL INFO Channel 11/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636087:2636603 [3] NCCL INFO Channel 10/0 : 19[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272157:272649 [7] NCCL INFO Channel 10/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636085:2636604 [1] NCCL INFO Channel 12/0 : 17[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62881:63375 [0] NCCL INFO Channel 01/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272151:272646 [1] NCCL INFO Channel 12/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-52:2722048:2722545 [3] NCCL INFO Channel 12/0 : 27[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-52:2722051:2722543 [6] NCCL INFO Channel 02/0 : 30[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62883:63373 [2] NCCL INFO Channel 06/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272150:272645 [0] NCCL INFO Channel 13/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815967:2816473 [4] NCCL INFO Channel 03/0 : 12[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62887:63379 [6] NCCL INFO Channel 03/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235655:236142 [7] NCCL INFO Channel 13/0 : 15[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62885:63377 [4] NCCL INFO Channel 02/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636087:2636603 [3] NCCL INFO Channel 12/0 : 19[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62882:63374 [1] NCCL INFO Channel 15/0 : 1[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636085:2636604 [1] NCCL INFO Channel 14/0 : 17[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-52:2722048:2722545 [3] NCCL INFO Channel 14/0 : 27[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62886:63378 [5] NCCL INFO Channel 01/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272157:272649 [7] NCCL INFO Channel 11/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62888:63380 [7] NCCL INFO Channel 08/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272151:272646 [1] NCCL INFO Channel 13/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272150:272645 [0] NCCL INFO Channel 14/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62884:63376 [3] NCCL INFO Channel 00/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235648:236145 [0] NCCL INFO Channel 03/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235652:236147 [4] NCCL INFO Channel 01/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235650:236144 [2] NCCL INFO Channel 11/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235653:236140 [5] NCCL INFO Channel 03/0 : 13[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235654:236146 [6] NCCL INFO Channel 12/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235649:236143 [1] NCCL INFO Channel 01/0 : 9[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636087:2636603 [3] NCCL INFO Channel 14/0 : 19[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-52:2722051:2722543 [6] NCCL INFO Channel 04/0 : 30[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235651:236141 [3] NCCL INFO Channel 12/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815967:2816473 [4] NCCL INFO Channel 05/0 : 12[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272157:272649 [7] NCCL INFO Channel 12/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62881:63375 [0] NCCL INFO Channel 02/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272151:272646 [1] NCCL INFO Channel 14/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272150:272645 [0] NCCL INFO Channel 15/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62887:63379 [6] NCCL INFO Channel 04/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62883:63373 [2] NCCL INFO Channel 07/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-52:2722051:2722543 [6] NCCL INFO Channel 08/0 : 30[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62885:63377 [4] NCCL INFO Channel 03/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235648:236145 [0] NCCL INFO Channel 05/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62886:63378 [5] NCCL INFO Channel 02/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62882:63374 [1] NCCL INFO Channel 00/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815965:2816475 [2] NCCL INFO Channel 01/0 : 10[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62888:63380 [7] NCCL INFO Channel 09/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62884:63376 [3] NCCL INFO Channel 01/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815969:2816471 [6] NCCL INFO Channel 01/0 : 14[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62881:63375 [0] NCCL INFO Channel 03/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272157:272649 [7] NCCL INFO Channel 13/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722051:2722543 [6] NCCL INFO Channel 10/0 : 30[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62887:63379 [6] NCCL INFO Channel 05/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272151:272646 [1] NCCL INFO Channel 15/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235650:236144 [2] NCCL INFO Channel 12/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272157:272649 [7] NCCL INFO Channel 14/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235652:236147 [4] NCCL INFO Channel 03/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235654:236146 [6] NCCL INFO Channel 13/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235653:236140 [5] NCCL INFO Channel 05/0 : 13[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235649:236143 [1] NCCL INFO Channel 03/0 : 9[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235651:236141 [3] NCCL INFO Channel 13/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235648:236145 [0] NCCL INFO Channel 07/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815967:2816473 [4] NCCL INFO Channel 07/0 : 12[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815968:2816472 [5] NCCL INFO Channel 01/0 : 13[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815965:2816475 [2] NCCL INFO Channel 03/0 : 10[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815969:2816471 [6] NCCL INFO Channel 03/0 : 14[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62885:63377 [4] NCCL INFO Channel 05/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62883:63373 [2] NCCL INFO Channel 08/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815970:2816470 [7] NCCL INFO Channel 01/0 : 15[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722051:2722543 [6] NCCL INFO Channel 12/0 : 30[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62886:63378 [5] NCCL INFO Channel 03/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62882:63374 [1] NCCL INFO Channel 02/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235650:236144 [2] NCCL INFO Channel 13/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62888:63380 [7] NCCL INFO Channel 10/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62884:63376 [3] NCCL INFO Channel 02/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235652:236147 [4] NCCL INFO Channel 05/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62887:63379 [6] NCCL INFO Channel 07/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62881:63375 [0] NCCL INFO Channel 04/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62885:63377 [4] NCCL INFO Channel 06/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62883:63373 [2] NCCL INFO Channel 09/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815967:2816473 [4] NCCL INFO Channel 09/0 : 12[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815968:2816472 [5] NCCL INFO Channel 03/0 : 13[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-52:2722051:2722543 [6] NCCL INFO Channel 14/0 : 30[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235654:236146 [6] NCCL INFO Channel 15/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62886:63378 [5] NCCL INFO Channel 04/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235653:236140 [5] NCCL INFO Channel 07/0 : 13[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62882:63374 [1] NCCL INFO Channel 03/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62888:63380 [7] NCCL INFO Channel 11/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235648:236145 [0] NCCL INFO Channel 09/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235649:236143 [1] NCCL INFO Channel 05/0 : 9[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62884:63376 [3] NCCL INFO Channel 04/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62887:63379 [6] NCCL INFO Channel 08/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62881:63375 [0] NCCL INFO Channel 05/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235651:236141 [3] NCCL INFO Channel 14/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815969:2816471 [6] NCCL INFO Channel 05/0 : 14[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62885:63377 [4] NCCL INFO Channel 07/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235652:236147 [4] NCCL INFO Channel 07/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235650:236144 [2] NCCL INFO Channel 14/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815970:2816470 [7] NCCL INFO Channel 03/0 : 15[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815965:2816475 [2] NCCL INFO Channel 05/0 : 10[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62883:63373 [2] NCCL INFO Channel 11/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62886:63378 [5] NCCL INFO Channel 06/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815968:2816472 [5] NCCL INFO Channel 05/0 : 13[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815967:2816473 [4] NCCL INFO Channel 11/0 : 12[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235654:236146 [6] NCCL INFO Channel 01/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235653:236140 [5] NCCL INFO Channel 09/0 : 13[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235648:236145 [0] NCCL INFO Channel 11/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235649:236143 [1] NCCL INFO Channel 07/0 : 9[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62888:63380 [7] NCCL INFO Channel 12/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235651:236141 [3] NCCL INFO Channel 15/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272155:272652 [5] NCCL INFO Channel 00/0 : 29[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815969:2816471 [6] NCCL INFO Channel 07/0 : 14[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62882:63374 [1] NCCL INFO Channel 04/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235652:236147 [4] NCCL INFO Channel 09/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815970:2816470 [7] NCCL INFO Channel 05/0 : 15[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62887:63379 [6] NCCL INFO Channel 09/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62881:63375 [0] NCCL INFO Channel 06/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235650:236144 [2] NCCL INFO Channel 15/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815965:2816475 [2] NCCL INFO Channel 07/0 : 10[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62885:63377 [4] NCCL INFO Channel 08/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235654:236146 [6] NCCL INFO Channel 03/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815964:2816476 [1] NCCL INFO Channel 01/0 : 9[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815968:2816472 [5] NCCL INFO Channel 07/0 : 13[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815967:2816473 [4] NCCL INFO Channel 13/0 : 12[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815969:2816471 [6] NCCL INFO Channel 09/0 : 14[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62884:63376 [3] NCCL INFO Channel 05/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272155:272652 [5] NCCL INFO Channel 02/0 : 29[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815970:2816470 [7] NCCL INFO Channel 07/0 : 15[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62883:63373 [2] NCCL INFO Channel 12/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62886:63378 [5] NCCL INFO Channel 07/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815965:2816475 [2] NCCL INFO Channel 09/0 : 10[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62888:63380 [7] NCCL INFO Channel 13/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235653:236140 [5] NCCL INFO Channel 11/0 : 13[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815966:2816474 [3] NCCL INFO Channel 01/0 : 11[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62882:63374 [1] NCCL INFO Channel 05/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235649:236143 [1] NCCL INFO Channel 11/0 : 9[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235648:236145 [0] NCCL INFO Channel 13/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62887:63379 [6] NCCL INFO Channel 10/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235651:236141 [3] NCCL INFO Channel 01/0 : 11[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62881:63375 [0] NCCL INFO Channel 07/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62885:63377 [4] NCCL INFO Channel 09/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272156:272648 [6] NCCL INFO Channel 00/0 : 30[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62884:63376 [3] NCCL INFO Channel 06/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815969:2816471 [6] NCCL INFO Channel 11/0 : 14[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815968:2816472 [5] NCCL INFO Channel 09/0 : 13[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235652:236147 [4] NCCL INFO Channel 11/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235654:236146 [6] NCCL INFO Channel 05/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815964:2816476 [1] NCCL INFO Channel 03/0 : 9[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815970:2816470 [7] NCCL INFO Channel 09/0 : 15[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815967:2816473 [4] NCCL INFO Channel 15/0 : 12[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235650:236144 [2] NCCL INFO Channel 01/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235653:236140 [5] NCCL INFO Channel 15/0 : 13[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815965:2816475 [2] NCCL INFO Channel 11/0 : 10[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235649:236143 [1] NCCL INFO Channel 13/0 : 9[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235648:236145 [0] NCCL INFO Channel 15/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272155:272652 [5] NCCL INFO Channel 04/0 : 29[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815966:2816474 [3] NCCL INFO Channel 03/0 : 11[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62886:63378 [5] NCCL INFO Channel 08/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62883:63373 [2] NCCL INFO Channel 13/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62888:63380 [7] NCCL INFO Channel 14/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235651:236141 [3] NCCL INFO Channel 03/0 : 11[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272156:272648 [6] NCCL INFO Channel 02/0 : 30[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62882:63374 [1] NCCL INFO Channel 06/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62887:63379 [6] NCCL INFO Channel 11/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62885:63377 [4] NCCL INFO Channel 10/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62881:63375 [0] NCCL INFO Channel 09/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62884:63376 [3] NCCL INFO Channel 07/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815969:2816471 [6] NCCL INFO Channel 13/0 : 14[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235652:236147 [4] NCCL INFO Channel 13/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815968:2816472 [5] NCCL INFO Channel 11/0 : 13[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235654:236146 [6] NCCL INFO Channel 07/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272155:272652 [5] NCCL INFO Channel 06/0 : 29[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815970:2816470 [7] NCCL INFO Channel 11/0 : 15[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235649:236143 [1] NCCL INFO Channel 15/0 : 9[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62886:63378 [5] NCCL INFO Channel 09/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235650:236144 [2] NCCL INFO Channel 03/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62883:63373 [2] NCCL INFO Channel 14/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235651:236141 [3] NCCL INFO Channel 05/0 : 11[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62882:63374 [1] NCCL INFO Channel 07/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62887:63379 [6] NCCL INFO Channel 12/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235652:236147 [4] NCCL INFO Channel 15/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62885:63377 [4] NCCL INFO Channel 11/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235654:236146 [6] NCCL INFO Channel 09/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272156:272648 [6] NCCL INFO Channel 04/0 : 30[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62881:63375 [0] NCCL INFO Channel 10/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62884:63376 [3] NCCL INFO Channel 08/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272155:272652 [5] NCCL INFO Channel 08/0 : 29[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815967:2816473 [4] NCCL INFO Channel 01/0 : 28[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62886:63378 [5] NCCL INFO Channel 10/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272156:272648 [6] NCCL INFO Channel 08/0 : 30[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815965:2816475 [2] NCCL INFO Channel 13/0 : 10[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815964:2816476 [1] NCCL INFO Channel 05/0 : 9[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815966:2816474 [3] NCCL INFO Channel 05/0 : 11[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815969:2816471 [6] NCCL INFO Channel 15/0 : 14[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235651:236141 [3] NCCL INFO Channel 07/0 : 11[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815968:2816472 [5] NCCL INFO Channel 15/0 : 13[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62883:63373 [2] NCCL INFO Channel 15/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235650:236144 [2] NCCL INFO Channel 05/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815970:2816470 [7] NCCL INFO Channel 13/0 : 15[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62887:63379 [6] NCCL INFO Channel 13/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62882:63374 [1] NCCL INFO Channel 08/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235654:236146 [6] NCCL INFO Channel 11/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272155:272652 [5] NCCL INFO Channel 10/0 : 29[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96570:97093 [7] NCCL INFO Channel 00/0 : 23[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62885:63377 [4] NCCL INFO Channel 13/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235651:236141 [3] NCCL INFO Channel 09/0 : 11[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272156:272648 [6] NCCL INFO Channel 10/0 : 30[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62881:63375 [0] NCCL INFO Channel 11/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62884:63376 [3] NCCL INFO Channel 09/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62886:63378 [5] NCCL INFO Channel 11/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235650:236144 [2] NCCL INFO Channel 07/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235654:236146 [6] NCCL INFO Channel 13/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815965:2816475 [2] NCCL INFO Channel 15/0 : 10[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815967:2816473 [4] NCCL INFO Channel 03/0 : 28[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815964:2816476 [1] NCCL INFO Channel 07/0 : 9[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815966:2816474 [3] NCCL INFO Channel 07/0 : 11[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96570:97093 [7] NCCL INFO Channel 02/0 : 23[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62887:63379 [6] NCCL INFO Channel 15/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62882:63374 [1] NCCL INFO Channel 10/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272155:272652 [5] NCCL INFO Channel 12/0 : 29[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815969:2816471 [6] NCCL INFO Channel 01/0 : 30[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815968:2816472 [5] NCCL INFO Channel 01/0 : 29[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62885:63377 [4] NCCL INFO Channel 14/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235651:236141 [3] NCCL INFO Channel 13/0 : 11[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272153:272651 [3] NCCL INFO Channel 00/0 : 27[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815970:2816470 [7] NCCL INFO Channel 01/0 : 31[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62881:63375 [0] NCCL INFO Channel 12/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62886:63378 [5] NCCL INFO Channel 12/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235654:236146 [6] NCCL INFO Channel 15/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235650:236144 [2] NCCL INFO Channel 09/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272156:272648 [6] NCCL INFO Channel 12/0 : 30[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815965:2816475 [2] NCCL INFO Channel 01/0 : 26[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62882:63374 [1] NCCL INFO Channel 11/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62884:63376 [3] NCCL INFO Channel 10/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62885:63377 [4] NCCL INFO Channel 15/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235651:236141 [3] NCCL INFO Channel 15/0 : 11[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272155:272652 [5] NCCL INFO Channel 14/0 : 29[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96570:97093 [7] NCCL INFO Channel 04/0 : 23[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815967:2816473 [4] NCCL INFO Channel 05/0 : 28[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636091:2636599 [7] NCCL INFO Channel 00/0 : 7[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62888:63380 [7] NCCL INFO Channel 01/0 : 31[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235650:236144 [2] NCCL INFO Channel 11/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272153:272651 [3] NCCL INFO Channel 02/0 : 27[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815964:2816476 [1] NCCL INFO Channel 11/0 : 9[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62881:63375 [0] NCCL INFO Channel 13/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62886:63378 [5] NCCL INFO Channel 14/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272156:272648 [6] NCCL INFO Channel 14/0 : 30[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815966:2816474 [3] NCCL INFO Channel 09/0 : 11[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815969:2816471 [6] NCCL INFO Channel 03/0 : 30[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815968:2816472 [5] NCCL INFO Channel 03/0 : 29[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815970:2816470 [7] NCCL INFO Channel 03/0 : 31[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96565:97089 [2] NCCL INFO Channel 00/0 : 18[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62882:63374 [1] NCCL INFO Channel 12/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62884:63376 [3] NCCL INFO Channel 12/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272155:272652 [5] NCCL INFO Channel 00/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815965:2816475 [2] NCCL INFO Channel 03/0 : 26[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272153:272651 [3] NCCL INFO Channel 04/0 : 27[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815967:2816473 [4] NCCL INFO Channel 07/0 : 28[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636091:2636599 [7] NCCL INFO Channel 02/0 : 7[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235650:236144 [2] NCCL INFO Channel 13/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272156:272648 [6] NCCL INFO Channel 00/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815964:2816476 [1] NCCL INFO Channel 13/0 : 9[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96570:97093 [7] NCCL INFO Channel 06/0 : 23[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62888:63380 [7] NCCL INFO Channel 03/0 : 31[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62881:63375 [0] NCCL INFO Channel 14/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636086:2636602 [2] NCCL INFO Channel 00/0 : 2[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62886:63378 [5] NCCL INFO Channel 15/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235650:236144 [2] NCCL INFO Channel 15/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62882:63374 [1] NCCL INFO Channel 13/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272155:272652 [5] NCCL INFO Channel 01/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815970:2816470 [7] NCCL INFO Channel 05/0 : 31[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815966:2816474 [3] NCCL INFO Channel 13/0 : 11[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815965:2816475 [2] NCCL INFO Channel 05/0 : 26[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62883:63373 [2] NCCL INFO Channel 01/0 : 26[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272153:272651 [3] NCCL INFO Channel 06/0 : 27[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815969:2816471 [6] NCCL INFO Channel 05/0 : 30[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815968:2816472 [5] NCCL INFO Channel 05/0 : 29[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96565:97089 [2] NCCL INFO Channel 04/0 : 18[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62884:63376 [3] NCCL INFO Channel 13/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272156:272648 [6] NCCL INFO Channel 01/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815967:2816473 [4] NCCL INFO Channel 09/0 : 28[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815964:2816476 [1] NCCL INFO Channel 15/0 : 9[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636091:2636599 [7] NCCL INFO Channel 04/0 : 7[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96570:97093 [7] NCCL INFO Channel 08/0 : 23[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272155:272652 [5] NCCL INFO Channel 02/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96569:97092 [6] NCCL INFO Channel 00/0 : 22[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272153:272651 [3] NCCL INFO Channel 08/0 : 27[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272156:272648 [6] NCCL INFO Channel 02/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62888:63380 [7] NCCL INFO Channel 05/0 : 31[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62881:63375 [0] NCCL INFO Channel 15/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62882:63374 [1] NCCL INFO Channel 14/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815966:2816474 [3] NCCL INFO Channel 15/0 : 11[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815970:2816470 [7] NCCL INFO Channel 07/0 : 31[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636086:2636602 [2] NCCL INFO Channel 04/0 : 2[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62884:63376 [3] NCCL INFO Channel 14/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62883:63373 [2] NCCL INFO Channel 03/0 : 26[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636091:2636599 [7] NCCL INFO Channel 06/0 : 7[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62888:63380 [7] NCCL INFO Channel 07/0 : 31[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272155:272652 [5] NCCL INFO Channel 03/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96567:97090 [4] NCCL INFO Channel 00/0 : 20[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62887:63379 [6] NCCL INFO Channel 01/0 : 30[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272156:272648 [6] NCCL INFO Channel 03/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96570:97093 [7] NCCL INFO Channel 10/0 : 23[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96569:97092 [6] NCCL INFO Channel 02/0 : 22[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62882:63374 [1] NCCL INFO Channel 15/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272153:272651 [3] NCCL INFO Channel 10/0 : 27[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96565:97089 [2] NCCL INFO Channel 06/0 : 18[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272155:272652 [5] NCCL INFO Channel 04/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815965:2816475 [2] NCCL INFO Channel 07/0 : 26[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815969:2816471 [6] NCCL INFO Channel 07/0 : 30[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815968:2816472 [5] NCCL INFO Channel 07/0 : 29[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815964:2816476 [1] NCCL INFO Channel 01/0 : 25[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815967:2816473 [4] NCCL INFO Channel 11/0 : 28[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636086:2636602 [2] NCCL INFO Channel 06/0 : 2[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636090:2636598 [6] NCCL INFO Channel 00/0 : 6[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636091:2636599 [7] NCCL INFO Channel 08/0 : 7[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62883:63373 [2] NCCL INFO Channel 05/0 : 26[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96567:97090 [4] NCCL INFO Channel 02/0 : 20[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62884:63376 [3] NCCL INFO Channel 15/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272156:272648 [6] NCCL INFO Channel 04/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815966:2816474 [3] NCCL INFO Channel 01/0 : 27[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96569:97092 [6] NCCL INFO Channel 04/0 : 22[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62888:63380 [7] NCCL INFO Channel 09/0 : 31[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62887:63379 [6] NCCL INFO Channel 03/0 : 30[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62885:63377 [4] NCCL INFO Channel 01/0 : 28[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272153:272651 [3] NCCL INFO Channel 12/0 : 27[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636088:2636601 [4] NCCL INFO Channel 00/0 : 4[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62883:63373 [2] NCCL INFO Channel 07/0 : 26[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272155:272652 [5] NCCL INFO Channel 06/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636086:2636602 [2] NCCL INFO Channel 08/0 : 2[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636090:2636598 [6] NCCL INFO Channel 02/0 : 6[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815970:2816470 [7] NCCL INFO Channel 09/0 : 31[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636091:2636599 [7] NCCL INFO Channel 10/0 : 7[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815964:2816476 [1] NCCL INFO Channel 03/0 : 25[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96570:97093 [7] NCCL INFO Channel 12/0 : 23[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815965:2816475 [2] NCCL INFO Channel 09/0 : 26[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96568:97091 [5] NCCL INFO Channel 00/0 : 21[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62888:63380 [7] NCCL INFO Channel 11/0 : 31[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272156:272648 [6] NCCL INFO Channel 05/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815967:2816473 [4] NCCL INFO Channel 13/0 : 28[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815968:2816472 [5] NCCL INFO Channel 09/0 : 29[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815969:2816471 [6] NCCL INFO Channel 09/0 : 30[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96567:97090 [4] NCCL INFO Channel 06/0 : 20[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62887:63379 [6] NCCL INFO Channel 05/0 : 30[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272153:272651 [3] NCCL INFO Channel 14/0 : 27[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272155:272652 [5] NCCL INFO Channel 07/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96565:97089 [2] NCCL INFO Channel 08/0 : 18[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62885:63377 [4] NCCL INFO Channel 03/0 : 28[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272156:272648 [6] NCCL INFO Channel 07/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636088:2636601 [4] NCCL INFO Channel 02/0 : 4[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636086:2636602 [2] NCCL INFO Channel 10/0 : 2[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96569:97092 [6] NCCL INFO Channel 08/0 : 22[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62883:63373 [2] NCCL INFO Channel 09/0 : 26[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636091:2636599 [7] NCCL INFO Channel 12/0 : 7[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636090:2636598 [6] NCCL INFO Channel 04/0 : 6[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62886:63378 [5] NCCL INFO Channel 01/0 : 29[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636089:2636600 [5] NCCL INFO Channel 00/0 : 5[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815966:2816474 [3] NCCL INFO Channel 03/0 : 27[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272153:272651 [3] NCCL INFO Channel 00/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815963:2816477 [0] NCCL INFO Channel 01/0 : 8[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272155:272652 [5] NCCL INFO Channel 08/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96570:97093 [7] NCCL INFO Channel 14/0 : 23[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272156:272648 [6] NCCL INFO Channel 08/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96568:97091 [5] NCCL INFO Channel 02/0 : 21[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62888:63380 [7] NCCL INFO Channel 13/0 : 31[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272153:272651 [3] NCCL INFO Channel 01/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636088:2636601 [4] NCCL INFO Channel 06/0 : 4[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96567:97090 [4] NCCL INFO Channel 08/0 : 20[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62887:63379 [6] NCCL INFO Channel 07/0 : 30[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636086:2636602 [2] NCCL INFO Channel 12/0 : 2[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96569:97092 [6] NCCL INFO Channel 10/0 : 22[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62885:63377 [4] NCCL INFO Channel 05/0 : 28[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62883:63373 [2] NCCL INFO Channel 11/0 : 26[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815970:2816470 [7] NCCL INFO Channel 11/0 : 31[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96564:97088 [1] NCCL INFO Channel 00/0 : 17[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96565:97089 [2] NCCL INFO Channel 10/0 : 18[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62881:63375 [0] NCCL INFO Channel 01/0 : 24[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62886:63378 [5] NCCL INFO Channel 03/0 : 29[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815964:2816476 [1] NCCL INFO Channel 05/0 : 25[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96563:97086 [0] NCCL INFO Channel 02/0 : 16[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62888:63380 [7] NCCL INFO Channel 01/0 : 15[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272155:272652 [5] NCCL INFO Channel 09/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272156:272648 [6] NCCL INFO Channel 09/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815965:2816475 [2] NCCL INFO Channel 11/0 : 26[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62887:63379 [6] NCCL INFO Channel 09/0 : 30[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62882:63374 [1] NCCL INFO Channel 01/0 : 25[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272153:272651 [3] NCCL INFO Channel 02/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815969:2816471 [6] NCCL INFO Channel 11/0 : 30[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815967:2816473 [4] NCCL INFO Channel 15/0 : 28[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815968:2816472 [5] NCCL INFO Channel 11/0 : 29[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636091:2636599 [7] NCCL INFO Channel 14/0 : 7[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636090:2636598 [6] NCCL INFO Channel 08/0 : 6[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815966:2816474 [3] NCCL INFO Channel 05/0 : 27[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636089:2636600 [5] NCCL INFO Channel 02/0 : 5[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96570:97093 [7] NCCL INFO Channel 00/0 : 7[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96568:97091 [5] NCCL INFO Channel 04/0 : 21[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96567:97090 [4] NCCL INFO Channel 10/0 : 20[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62885:63377 [4] NCCL INFO Channel 07/0 : 28[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272155:272652 [5] NCCL INFO Channel 10/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272156:272648 [6] NCCL INFO Channel 10/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96566:97087 [3] NCCL INFO Channel 00/0 : 19[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62883:63373 [2] NCCL INFO Channel 13/0 : 26[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62886:63378 [5] NCCL INFO Channel 05/0 : 29[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62881:63375 [0] NCCL INFO Channel 03/0 : 24[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62888:63380 [7] NCCL INFO Channel 03/0 : 15[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272153:272651 [3] NCCL INFO Channel 04/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96569:97092 [6] NCCL INFO Channel 12/0 : 22[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62887:63379 [6] NCCL INFO Channel 11/0 : 30[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62882:63374 [1] NCCL INFO Channel 03/0 : 25[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815963:2816477 [0] NCCL INFO Channel 03/0 : 8[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636084:2636605 [0] NCCL INFO Channel 02/0 : 0[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636085:2636604 [1] NCCL INFO Channel 00/0 : 1[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96564:97088 [1] NCCL INFO Channel 02/0 : 17[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815970:2816470 [7] NCCL INFO Channel 13/0 : 31[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636088:2636601 [4] NCCL INFO Channel 08/0 : 4[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96563:97086 [0] NCCL INFO Channel 04/0 : 16[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815964:2816476 [1] NCCL INFO Channel 07/0 : 25[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636086:2636602 [2] NCCL INFO Channel 14/0 : 2[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96570:97093 [7] NCCL INFO Channel 02/0 : 7[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272156:272648 [6] NCCL INFO Channel 11/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272155:272652 [5] NCCL INFO Channel 11/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815965:2816475 [2] NCCL INFO Channel 13/0 : 26[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96568:97091 [5] NCCL INFO Channel 06/0 : 21[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62884:63376 [3] NCCL INFO Channel 01/0 : 27[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62885:63377 [4] NCCL INFO Channel 09/0 : 28[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272153:272651 [3] NCCL INFO Channel 05/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815966:2816474 [3] NCCL INFO Channel 07/0 : 27[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815969:2816471 [6] NCCL INFO Channel 13/0 : 30[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815968:2816472 [5] NCCL INFO Channel 15/0 : 29[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815963:2816477 [0] NCCL INFO Channel 05/0 : 8[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815964:2816476 [1] NCCL INFO Channel 11/0 : 25[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815965:2816475 [2] NCCL INFO Channel 15/0 : 26[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636091:2636599 [7] NCCL INFO Channel 00/0 : 23[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62883:63373 [2] NCCL INFO Channel 15/0 : 26[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272156:272648 [6] NCCL INFO Channel 12/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636087:2636603 [3] NCCL INFO Channel 00/0 : 3[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96567:97090 [4] NCCL INFO Channel 12/0 : 20[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96565:97089 [2] NCCL INFO Channel 12/0 : 18[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62886:63378 [5] NCCL INFO Channel 07/0 : 29[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272155:272652 [5] NCCL INFO Channel 12/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636089:2636600 [5] NCCL INFO Channel 04/0 : 5[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636090:2636598 [6] NCCL INFO Channel 10/0 : 6[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96569:97092 [6] NCCL INFO Channel 14/0 : 22[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96566:97087 [3] NCCL INFO Channel 02/0 : 19[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62888:63380 [7] NCCL INFO Channel 05/0 : 15[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272153:272651 [3] NCCL INFO Channel 06/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96564:97088 [1] NCCL INFO Channel 04/0 : 17[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96563:97086 [0] NCCL INFO Channel 06/0 : 16[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62881:63375 [0] NCCL INFO Channel 05/0 : 24[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815966:2816474 [3] NCCL INFO Channel 09/0 : 27[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96570:97093 [7] NCCL INFO Channel 04/0 : 7[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62887:63379 [6] NCCL INFO Channel 13/0 : 30[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815969:2816471 [6] NCCL INFO Channel 15/0 : 30[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96568:97091 [5] NCCL INFO Channel 08/0 : 21[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62882:63374 [1] NCCL INFO Channel 05/0 : 25[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62885:63377 [4] NCCL INFO Channel 11/0 : 28[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815963:2816477 [0] NCCL INFO Channel 07/0 : 8[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62884:63376 [3] NCCL INFO Channel 03/0 : 27[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815964:2816476 [1] NCCL INFO Channel 13/0 : 25[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636084:2636605 [0] NCCL INFO Channel 04/0 : 0[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636085:2636604 [1] NCCL INFO Channel 02/0 : 1[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636088:2636601 [4] NCCL INFO Channel 10/0 : 4[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272156:272648 [6] NCCL INFO Channel 13/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815966:2816474 [3] NCCL INFO Channel 13/0 : 27[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636086:2636602 [2] NCCL INFO Channel 00/0 : 18[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272155:272652 [5] NCCL INFO Channel 14/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815964:2816476 [1] NCCL INFO Channel 15/0 : 25[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636091:2636599 [7] NCCL INFO Channel 02/0 : 23[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96567:97090 [4] NCCL INFO Channel 14/0 : 20[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272153:272651 [3] NCCL INFO Channel 07/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815963:2816477 [0] NCCL INFO Channel 09/0 : 8[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96569:97092 [6] NCCL INFO Channel 00/0 : 6[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96566:97087 [3] NCCL INFO Channel 04/0 : 19[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96565:97089 [2] NCCL INFO Channel 14/0 : 18[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62883:63373 [2] NCCL INFO Channel 01/0 : 10[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96563:97086 [0] NCCL INFO Channel 08/0 : 16[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62886:63378 [5] NCCL INFO Channel 09/0 : 29[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62888:63380 [7] NCCL INFO Channel 07/0 : 15[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96564:97088 [1] NCCL INFO Channel 06/0 : 17[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62881:63375 [0] NCCL INFO Channel 07/0 : 24[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272156:272648 [6] NCCL INFO Channel 15/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636090:2636598 [6] NCCL INFO Channel 12/0 : 6[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636089:2636600 [5] NCCL INFO Channel 06/0 : 5[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96570:97093 [7] NCCL INFO Channel 06/0 : 7[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62887:63379 [6] NCCL INFO Channel 15/0 : 30[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272155:272652 [5] NCCL INFO Channel 15/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815966:2816474 [3] NCCL INFO Channel 15/0 : 27[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636087:2636603 [3] NCCL INFO Channel 02/0 : 3[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96568:97091 [5] NCCL INFO Channel 10/0 : 21[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62882:63374 [1] NCCL INFO Channel 07/0 : 25[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62885:63377 [4] NCCL INFO Channel 13/0 : 28[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272153:272651 [3] NCCL INFO Channel 08/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815963:2816477 [0] NCCL INFO Channel 11/0 : 8[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96567:97090 [4] NCCL INFO Channel 00/0 : 4[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815963:2816477 [0] NCCL INFO Channel 13/0 : 8[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96569:97092 [6] NCCL INFO Channel 02/0 : 6[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636088:2636601 [4] NCCL INFO Channel 12/0 : 4[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636084:2636605 [0] NCCL INFO Channel 06/0 : 0[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636091:2636599 [7] NCCL INFO Channel 04/0 : 23[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636085:2636604 [1] NCCL INFO Channel 04/0 : 1[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636086:2636602 [2] NCCL INFO Channel 04/0 : 18[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62884:63376 [3] NCCL INFO Channel 05/0 : 27[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62888:63380 [7] NCCL INFO Channel 09/0 : 15[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62883:63373 [2] NCCL INFO Channel 03/0 : 10[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272153:272651 [3] NCCL INFO Channel 09/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815963:2816477 [0] NCCL INFO Channel 15/0 : 8[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636090:2636598 [6] NCCL INFO Channel 14/0 : 6[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636089:2636600 [5] NCCL INFO Channel 08/0 : 5[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636087:2636603 [3] NCCL INFO Channel 04/0 : 3[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96566:97087 [3] NCCL INFO Channel 06/0 : 19[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62886:63378 [5] NCCL INFO Channel 11/0 : 29[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272153:272651 [3] NCCL INFO Channel 10/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96563:97086 [0] NCCL INFO Channel 10/0 : 16[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96564:97088 [1] NCCL INFO Channel 08/0 : 17[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62881:63375 [0] NCCL INFO Channel 09/0 : 24[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96570:97093 [7] NCCL INFO Channel 08/0 : 7[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96568:97091 [5] NCCL INFO Channel 12/0 : 21[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96565:97089 [2] NCCL INFO Channel 00/0 : 2[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62887:63379 [6] NCCL INFO Channel 01/0 : 14[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96567:97090 [4] NCCL INFO Channel 02/0 : 4[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62882:63374 [1] NCCL INFO Channel 11/0 : 25[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815963:2816477 [0] NCCL INFO Channel 01/0 : 24[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96569:97092 [6] NCCL INFO Channel 04/0 : 6[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62885:63377 [4] NCCL INFO Channel 15/0 : 28[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636088:2636601 [4] NCCL INFO Channel 14/0 : 4[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96566:97087 [3] NCCL INFO Channel 08/0 : 19[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272153:272651 [3] NCCL INFO Channel 12/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636091:2636599 [7] NCCL INFO Channel 06/0 : 23[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96563:97086 [0] NCCL INFO Channel 12/0 : 16[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96564:97088 [1] NCCL INFO Channel 10/0 : 17[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815963:2816477 [0] NCCL INFO Channel 03/0 : 24[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62888:63380 [7] NCCL INFO Channel 11/0 : 15[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62884:63376 [3] NCCL INFO Channel 07/0 : 27[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62886:63378 [5] NCCL INFO Channel 15/0 : 29[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62883:63373 [2] NCCL INFO Channel 05/0 : 10[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272153:272651 [3] NCCL INFO Channel 13/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62881:63375 [0] NCCL INFO Channel 11/0 : 24[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62887:63379 [6] NCCL INFO Channel 03/0 : 14[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636084:2636605 [0] NCCL INFO Channel 08/0 : 0[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62885:63377 [4] NCCL INFO Channel 01/0 : 12[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62882:63374 [1] NCCL INFO Channel 13/0 : 25[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815963:2816477 [0] NCCL INFO Channel 05/0 : 24[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636085:2636604 [1] NCCL INFO Channel 06/0 : 1[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636086:2636602 [2] NCCL INFO Channel 06/0 : 18[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96570:97093 [7] NCCL INFO Channel 10/0 : 7[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62888:63380 [7] NCCL INFO Channel 13/0 : 15[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636089:2636600 [5] NCCL INFO Channel 10/0 : 5[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636090:2636598 [6] NCCL INFO Channel 00/0 : 22[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96568:97091 [5] NCCL INFO Channel 14/0 : 21[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96567:97090 [4] NCCL INFO Channel 06/0 : 4[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62884:63376 [3] NCCL INFO Channel 09/0 : 27[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272153:272651 [3] NCCL INFO Channel 14/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96565:97089 [2] NCCL INFO Channel 04/0 : 2[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96569:97092 [6] NCCL INFO Channel 08/0 : 6[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96566:97087 [3] NCCL INFO Channel 10/0 : 19[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96564:97088 [1] NCCL INFO Channel 12/0 : 17[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96563:97086 [0] NCCL INFO Channel 14/0 : 16[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815963:2816477 [0] NCCL INFO Channel 07/0 : 24[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96570:97093 [7] NCCL INFO Channel 12/0 : 7[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636087:2636603 [3] NCCL INFO Channel 06/0 : 3[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636088:2636601 [4] NCCL INFO Channel 00/0 : 20[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636091:2636599 [7] NCCL INFO Channel 08/0 : 23[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96567:97090 [4] NCCL INFO Channel 08/0 : 4[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62886:63378 [5] NCCL INFO Channel 01/0 : 13[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62883:63373 [2] NCCL INFO Channel 07/0 : 10[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272153:272651 [3] NCCL INFO Channel 15/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636084:2636605 [0] NCCL INFO Channel 10/0 : 0[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96568:97091 [5] NCCL INFO Channel 00/0 : 5[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62887:63379 [6] NCCL INFO Channel 05/0 : 14[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62881:63375 [0] NCCL INFO Channel 13/0 : 24[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62885:63377 [4] NCCL INFO Channel 03/0 : 12[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62882:63374 [1] NCCL INFO Channel 15/0 : 25[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815963:2816477 [0] NCCL INFO Channel 09/0 : 24[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62884:63376 [3] NCCL INFO Channel 13/0 : 27[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62886:63378 [5] NCCL INFO Channel 03/0 : 13[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636085:2636604 [1] NCCL INFO Channel 08/0 : 1[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636090:2636598 [6] NCCL INFO Channel 02/0 : 22[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636089:2636600 [5] NCCL INFO Channel 12/0 : 5[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636086:2636602 [2] NCCL INFO Channel 08/0 : 18[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96569:97092 [6] NCCL INFO Channel 10/0 : 6[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815970:2816470 [7] NCCL INFO Channel 00/0 : 31[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96565:97089 [2] NCCL INFO Channel 06/0 : 2[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96564:97088 [1] NCCL INFO Channel 14/0 : 17[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96563:97086 [0] NCCL INFO Channel 02/0 : 0[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96566:97087 [3] NCCL INFO Channel 12/0 : 19[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96570:97093 [7] NCCL INFO Channel 14/0 : 7[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96567:97090 [4] NCCL INFO Channel 10/0 : 4[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96568:97091 [5] NCCL INFO Channel 02/0 : 5[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62887:63379 [6] NCCL INFO Channel 07/0 : 14[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96569:97092 [6] NCCL INFO Channel 12/0 : 6[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62883:63373 [2] NCCL INFO Channel 09/0 : 10[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636091:2636599 [7] NCCL INFO Channel 10/0 : 23[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636088:2636601 [4] NCCL INFO Channel 02/0 : 20[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636087:2636603 [3] NCCL INFO Channel 08/0 : 3[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96565:97089 [2] NCCL INFO Channel 08/0 : 2[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96566:97087 [3] NCCL INFO Channel 14/0 : 19[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96563:97086 [0] NCCL INFO Channel 04/0 : 0[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62885:63377 [4] NCCL INFO Channel 05/0 : 12[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62882:63374 [1] NCCL INFO Channel 01/0 : 9[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62881:63375 [0] NCCL INFO Channel 15/0 : 24[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636084:2636605 [0] NCCL INFO Channel 12/0 : 0[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62884:63376 [3] NCCL INFO Channel 15/0 : 27[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815963:2816477 [0] NCCL INFO Channel 11/0 : 24[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636085:2636604 [1] NCCL INFO Channel 10/0 : 1[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62886:63378 [5] NCCL INFO Channel 05/0 : 13[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62888:63380 [7] NCCL INFO Channel 00/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62887:63379 [6] NCCL INFO Channel 09/0 : 14[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62883:63373 [2] NCCL INFO Channel 11/0 : 10[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96564:97088 [1] NCCL INFO Channel 00/0 : 1[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96568:97091 [5] NCCL INFO Channel 04/0 : 5[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96567:97090 [4] NCCL INFO Channel 12/0 : 4[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96569:97092 [6] NCCL INFO Channel 14/0 : 6[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815970:2816470 [7] NCCL INFO Channel 02/0 : 31[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636090:2636598 [6] NCCL INFO Channel 04/0 : 22[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96563:97086 [0] NCCL INFO Channel 06/0 : 0[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96566:97087 [3] NCCL INFO Channel 00/0 : 3[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815963:2816477 [0] NCCL INFO Channel 13/0 : 24[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636089:2636600 [5] NCCL INFO Channel 14/0 : 5[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96565:97089 [2] NCCL INFO Channel 10/0 : 2[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62885:63377 [4] NCCL INFO Channel 07/0 : 12[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62882:63374 [1] NCCL INFO Channel 03/0 : 9[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62884:63376 [3] NCCL INFO Channel 01/0 : 11[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62881:63375 [0] NCCL INFO Channel 01/0 : 8[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815970:2816470 [7] NCCL INFO Channel 04/0 : 31[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636086:2636602 [2] NCCL INFO Channel 10/0 : 18[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96564:97088 [1] NCCL INFO Channel 02/0 : 1[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96568:97091 [5] NCCL INFO Channel 06/0 : 5[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96567:97090 [4] NCCL INFO Channel 14/0 : 4[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815963:2816477 [0] NCCL INFO Channel 15/0 : 24[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636091:2636599 [7] NCCL INFO Channel 12/0 : 23[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96563:97086 [0] NCCL INFO Channel 08/0 : 0[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96566:97087 [3] NCCL INFO Channel 02/0 : 3[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62886:63378 [5] NCCL INFO Channel 07/0 : 13[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62888:63380 [7] NCCL INFO Channel 01/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636088:2636601 [4] NCCL INFO Channel 06/0 : 20[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96565:97089 [2] NCCL INFO Channel 12/0 : 2[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62883:63373 [2] NCCL INFO Channel 13/0 : 10[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62887:63379 [6] NCCL INFO Channel 11/0 : 14[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96564:97088 [1] NCCL INFO Channel 04/0 : 1[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62885:63377 [4] NCCL INFO Channel 09/0 : 12[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96568:97091 [5] NCCL INFO Channel 08/0 : 5[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815970:2816470 [7] NCCL INFO Channel 06/0 : 31[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815970:2816470 [7] NCCL INFO Channel 08/0 : 31[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636087:2636603 [3] NCCL INFO Channel 10/0 : 3[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636085:2636604 [1] NCCL INFO Channel 12/0 : 1[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636084:2636605 [0] NCCL INFO Channel 14/0 : 0[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636090:2636598 [6] NCCL INFO Channel 08/0 : 22[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62882:63374 [1] NCCL INFO Channel 05/0 : 9[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636089:2636600 [5] NCCL INFO Channel 00/0 : 21[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62884:63376 [3] NCCL INFO Channel 03/0 : 11[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636086:2636602 [2] NCCL INFO Channel 12/0 : 18[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96566:97087 [3] NCCL INFO Channel 04/0 : 3[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96563:97086 [0] NCCL INFO Channel 10/0 : 0[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62881:63375 [0] NCCL INFO Channel 03/0 : 8[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636091:2636599 [7] NCCL INFO Channel 14/0 : 23[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96564:97088 [1] NCCL INFO Channel 06/0 : 1[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62886:63378 [5] NCCL INFO Channel 09/0 : 13[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815970:2816470 [7] NCCL INFO Channel 10/0 : 31[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636088:2636601 [4] NCCL INFO Channel 08/0 : 20[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96565:97089 [2] NCCL INFO Channel 14/0 : 2[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96568:97091 [5] NCCL INFO Channel 10/0 : 5[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96563:97086 [0] NCCL INFO Channel 12/0 : 0[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62888:63380 [7] NCCL INFO Channel 02/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62887:63379 [6] NCCL INFO Channel 13/0 : 14[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815970:2816470 [7] NCCL INFO Channel 12/0 : 31[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636090:2636598 [6] NCCL INFO Channel 10/0 : 22[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636087:2636603 [3] NCCL INFO Channel 12/0 : 3[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62883:63373 [2] NCCL INFO Channel 15/0 : 10[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62885:63377 [4] NCCL INFO Channel 11/0 : 12[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636084:2636605 [0] NCCL INFO Channel 02/0 : 16[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62882:63374 [1] NCCL INFO Channel 07/0 : 9[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636089:2636600 [5] NCCL INFO Channel 02/0 : 21[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96566:97087 [3] NCCL INFO Channel 06/0 : 3[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96564:97088 [1] NCCL INFO Channel 08/0 : 1[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62884:63376 [3] NCCL INFO Channel 05/0 : 11[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96568:97091 [5] NCCL INFO Channel 12/0 : 5[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62881:63375 [0] NCCL INFO Channel 05/0 : 8[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96564:97088 [1] NCCL INFO Channel 10/0 : 1[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96566:97087 [3] NCCL INFO Channel 08/0 : 3[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96563:97086 [0] NCCL INFO Channel 14/0 : 0[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62886:63378 [5] NCCL INFO Channel 11/0 : 13[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815970:2816470 [7] NCCL INFO Channel 14/0 : 31[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96568:97091 [5] NCCL INFO Channel 14/0 : 5[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636088:2636601 [4] NCCL INFO Channel 10/0 : 20[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636086:2636602 [2] NCCL INFO Channel 14/0 : 18[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636085:2636604 [1] NCCL INFO Channel 14/0 : 1[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62887:63379 [6] NCCL INFO Channel 15/0 : 14[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62888:63380 [7] NCCL INFO Channel 03/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62885:63377 [4] NCCL INFO Channel 13/0 : 12[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96566:97087 [3] NCCL INFO Channel 10/0 : 3[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62882:63374 [1] NCCL INFO Channel 11/0 : 9[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96564:97088 [1] NCCL INFO Channel 12/0 : 1[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62881:63375 [0] NCCL INFO Channel 07/0 : 8[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62884:63376 [3] NCCL INFO Channel 07/0 : 11[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62886:63378 [5] NCCL INFO Channel 15/0 : 13[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62885:63377 [4] NCCL INFO Channel 15/0 : 12[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62888:63380 [7] NCCL INFO Channel 04/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636090:2636598 [6] NCCL INFO Channel 12/0 : 22[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636089:2636600 [5] NCCL INFO Channel 04/0 : 21[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62882:63374 [1] NCCL INFO Channel 13/0 : 9[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815965:2816475 [2] NCCL INFO Channel 00/0 : 26[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96570:97093 [7] NCCL INFO Channel 01/0 : 15[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96566:97087 [3] NCCL INFO Channel 12/0 : 3[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96564:97088 [1] NCCL INFO Channel 14/0 : 1[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636087:2636603 [3] NCCL INFO Channel 14/0 : 3[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62881:63375 [0] NCCL INFO Channel 09/0 : 8[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636088:2636601 [4] NCCL INFO Channel 12/0 : 20[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62884:63376 [3] NCCL INFO Channel 09/0 : 11[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62883:63373 [2] NCCL INFO Channel 00/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636084:2636605 [0] NCCL INFO Channel 04/0 : 16[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636091:2636599 [7] NCCL INFO Channel 00/0 : 31[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62888:63380 [7] NCCL INFO Channel 05/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62882:63374 [1] NCCL INFO Channel 15/0 : 9[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815965:2816475 [2] NCCL INFO Channel 04/0 : 26[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62881:63375 [0] NCCL INFO Channel 11/0 : 8[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96570:97093 [7] NCCL INFO Channel 03/0 : 15[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96566:97087 [3] NCCL INFO Channel 14/0 : 3[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96565:97089 [2] NCCL INFO Channel 01/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636090:2636598 [6] NCCL INFO Channel 14/0 : 22[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636085:2636604 [1] NCCL INFO Channel 00/0 : 17[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636089:2636600 [5] NCCL INFO Channel 06/0 : 21[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815969:2816471 [6] NCCL INFO Channel 00/0 : 30[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636088:2636601 [4] NCCL INFO Channel 14/0 : 20[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62884:63376 [3] NCCL INFO Channel 13/0 : 11[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62883:63373 [2] NCCL INFO Channel 01/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62887:63379 [6] NCCL INFO Channel 00/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96570:97093 [7] NCCL INFO Channel 05/0 : 15[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62888:63380 [7] NCCL INFO Channel 06/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62881:63375 [0] NCCL INFO Channel 13/0 : 8[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62886:63378 [5] NCCL INFO Channel 00/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815965:2816475 [2] NCCL INFO Channel 06/0 : 26[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815968:2816472 [5] NCCL INFO Channel 00/0 : 29[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636087:2636603 [3] NCCL INFO Channel 00/0 : 19[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636091:2636599 [7] NCCL INFO Channel 02/0 : 31[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815969:2816471 [6] NCCL INFO Channel 02/0 : 30[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636086:2636602 [2] NCCL INFO Channel 00/0 : 26[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636084:2636605 [0] NCCL INFO Channel 06/0 : 16[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815967:2816473 [4] NCCL INFO Channel 00/0 : 28[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636085:2636604 [1] NCCL INFO Channel 02/0 : 17[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636089:2636600 [5] NCCL INFO Channel 08/0 : 21[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96570:97093 [7] NCCL INFO Channel 07/0 : 15[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96565:97089 [2] NCCL INFO Channel 03/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62884:63376 [3] NCCL INFO Channel 15/0 : 11[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62887:63379 [6] NCCL INFO Channel 01/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62885:63377 [4] NCCL INFO Channel 00/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62883:63373 [2] NCCL INFO Channel 03/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62888:63380 [7] NCCL INFO Channel 08/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62881:63375 [0] NCCL INFO Channel 15/0 : 8[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815965:2816475 [2] NCCL INFO Channel 08/0 : 26[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815968:2816472 [5] NCCL INFO Channel 02/0 : 29[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96570:97093 [7] NCCL INFO Channel 09/0 : 15[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815969:2816471 [6] NCCL INFO Channel 04/0 : 30[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96565:97089 [2] NCCL INFO Channel 05/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815967:2816473 [4] NCCL INFO Channel 02/0 : 28[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636091:2636599 [7] NCCL INFO Channel 04/0 : 31[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636087:2636603 [3] NCCL INFO Channel 02/0 : 19[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96569:97092 [6] NCCL INFO Channel 01/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815964:2816476 [1] NCCL INFO Channel 00/0 : 25[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815965:2816475 [2] NCCL INFO Channel 10/0 : 26[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96570:97093 [7] NCCL INFO Channel 11/0 : 15[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815968:2816472 [5] NCCL INFO Channel 04/0 : 29[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96565:97089 [2] NCCL INFO Channel 07/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815969:2816471 [6] NCCL INFO Channel 08/0 : 30[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815967:2816473 [4] NCCL INFO Channel 06/0 : 28[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557273:1557760 [7] NCCL INFO Channel 11/0 : 15[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557266:1557759 [0] NCCL INFO Channel 15/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815965:2816475 [2] NCCL INFO Channel 12/0 : 26[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557267:1557761 [1] NCCL INFO Channel 15/0 : 9[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636084:2636605 [0] NCCL INFO Channel 08/0 : 16[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636089:2636600 [5] NCCL INFO Channel 10/0 : 21[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636086:2636602 [2] NCCL INFO Channel 04/0 : 26[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557273:1557760 [7] NCCL INFO Channel 13/0 : 15[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636090:2636598 [6] NCCL INFO Channel 00/0 : 30[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557271:1557762 [5] NCCL INFO Channel 00/0 : 21[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636085:2636604 [1] NCCL INFO Channel 04/0 : 17[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96569:97092 [6] NCCL INFO Channel 03/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557271:1557762 [5] NCCL INFO Channel 02/0 : 21[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96567:97090 [4] NCCL INFO Channel 01/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557272:1557766 [6] NCCL INFO Channel 00/0 : 22[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815968:2816472 [5] NCCL INFO Channel 06/0 : 29[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815964:2816476 [1] NCCL INFO Channel 02/0 : 25[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96570:97093 [7] NCCL INFO Channel 13/0 : 15[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557271:1557762 [5] NCCL INFO Channel 04/0 : 21[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815969:2816471 [6] NCCL INFO Channel 10/0 : 30[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96565:97089 [2] NCCL INFO Channel 09/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557272:1557766 [6] NCCL INFO Channel 02/0 : 22[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815967:2816473 [4] NCCL INFO Channel 08/0 : 28[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557271:1557762 [5] NCCL INFO Channel 06/0 : 21[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815965:2816475 [2] NCCL INFO Channel 14/0 : 26[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557272:1557766 [6] NCCL INFO Channel 04/0 : 22[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815966:2816474 [3] NCCL INFO Channel 00/0 : 27[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636088:2636601 [4] NCCL INFO Channel 00/0 : 28[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557271:1557762 [5] NCCL INFO Channel 08/0 : 21[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636091:2636599 [7] NCCL INFO Channel 06/0 : 31[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557269:1557765 [3] NCCL INFO Channel 00/0 : 19[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557272:1557766 [6] NCCL INFO Channel 08/0 : 22[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636087:2636603 [3] NCCL INFO Channel 04/0 : 19[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557271:1557762 [5] NCCL INFO Channel 10/0 : 21[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96569:97092 [6] NCCL INFO Channel 05/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557271:1557762 [5] NCCL INFO Channel 12/0 : 21[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96567:97090 [4] NCCL INFO Channel 03/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557272:1557766 [6] NCCL INFO Channel 10/0 : 22[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96565:97089 [2] NCCL INFO Channel 11/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557269:1557765 [3] NCCL INFO Channel 02/0 : 19[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557271:1557762 [5] NCCL INFO Channel 14/0 : 21[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636089:2636600 [5] NCCL INFO Channel 12/0 : 21[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636084:2636605 [0] NCCL INFO Channel 10/0 : 16[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557269:1557765 [3] NCCL INFO Channel 04/0 : 19[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557272:1557766 [6] NCCL INFO Channel 12/0 : 22[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636090:2636598 [6] NCCL INFO Channel 02/0 : 30[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557271:1557762 [5] NCCL INFO Channel 00/0 : 5[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636088:2636601 [4] NCCL INFO Channel 02/0 : 28[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636086:2636602 [2] NCCL INFO Channel 06/0 : 26[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557269:1557765 [3] NCCL INFO Channel 06/0 : 19[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636085:2636604 [1] NCCL INFO Channel 06/0 : 17[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96569:97092 [6] NCCL INFO Channel 07/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557272:1557766 [6] NCCL INFO Channel 14/0 : 22[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96567:97090 [4] NCCL INFO Channel 05/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557271:1557762 [5] NCCL INFO Channel 02/0 : 5[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96565:97089 [2] NCCL INFO Channel 13/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557269:1557765 [3] NCCL INFO Channel 08/0 : 19[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557272:1557766 [6] NCCL INFO Channel 00/0 : 6[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557271:1557762 [5] NCCL INFO Channel 04/0 : 5[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557269:1557765 [3] NCCL INFO Channel 10/0 : 19[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636091:2636599 [7] NCCL INFO Channel 08/0 : 31[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557271:1557762 [5] NCCL INFO Channel 06/0 : 5[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636087:2636603 [3] NCCL INFO Channel 06/0 : 19[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557272:1557766 [6] NCCL INFO Channel 02/0 : 6[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636089:2636600 [5] NCCL INFO Channel 14/0 : 21[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557269:1557765 [3] NCCL INFO Channel 12/0 : 19[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636084:2636605 [0] NCCL INFO Channel 12/0 : 16[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557271:1557762 [5] NCCL INFO Channel 08/0 : 5[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96569:97092 [6] NCCL INFO Channel 09/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96567:97090 [4] NCCL INFO Channel 07/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557272:1557766 [6] NCCL INFO Channel 04/0 : 6[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96565:97089 [2] NCCL INFO Channel 15/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557269:1557765 [3] NCCL INFO Channel 14/0 : 19[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557271:1557762 [5] NCCL INFO Channel 10/0 : 5[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557269:1557765 [3] NCCL INFO Channel 00/0 : 3[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557272:1557766 [6] NCCL INFO Channel 08/0 : 6[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557271:1557762 [5] NCCL INFO Channel 12/0 : 5[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557269:1557765 [3] NCCL INFO Channel 02/0 : 3[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557272:1557766 [6] NCCL INFO Channel 10/0 : 6[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636090:2636598 [6] NCCL INFO Channel 04/0 : 30[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557271:1557762 [5] NCCL INFO Channel 14/0 : 5[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636088:2636601 [4] NCCL INFO Channel 06/0 : 28[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557272:1557766 [6] NCCL INFO Channel 12/0 : 6[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96569:97092 [6] NCCL INFO Channel 11/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557269:1557765 [3] NCCL INFO Channel 04/0 : 3[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96567:97090 [4] NCCL INFO Channel 09/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557272:1557766 [6] NCCL INFO Channel 14/0 : 6[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557269:1557765 [3] NCCL INFO Channel 06/0 : 3[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557271:1557762 [5] NCCL INFO Channel 01/0 : 13[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636086:2636602 [2] NCCL INFO Channel 08/0 : 26[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636091:2636599 [7] NCCL INFO Channel 10/0 : 31[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557269:1557765 [3] NCCL INFO Channel 08/0 : 3[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636085:2636604 [1] NCCL INFO Channel 08/0 : 17[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636087:2636603 [3] NCCL INFO Channel 08/0 : 19[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557271:1557762 [5] NCCL INFO Channel 03/0 : 13[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636084:2636605 [0] NCCL INFO Channel 14/0 : 16[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96569:97092 [6] NCCL INFO Channel 13/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557269:1557765 [3] NCCL INFO Channel 10/0 : 3[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636090:2636598 [6] NCCL INFO Channel 08/0 : 30[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557271:1557762 [5] NCCL INFO Channel 05/0 : 13[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636088:2636601 [4] NCCL INFO Channel 08/0 : 28[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557272:1557766 [6] NCCL INFO Channel 01/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557269:1557765 [3] NCCL INFO Channel 12/0 : 3[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557271:1557762 [5] NCCL INFO Channel 07/0 : 13[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557272:1557766 [6] NCCL INFO Channel 03/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557269:1557765 [3] NCCL INFO Channel 14/0 : 3[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96569:97092 [6] NCCL INFO Channel 15/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557271:1557762 [5] NCCL INFO Channel 09/0 : 13[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96568:97091 [5] NCCL INFO Channel 01/0 : 13[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557272:1557766 [6] NCCL INFO Channel 05/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636091:2636599 [7] NCCL INFO Channel 12/0 : 31[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557271:1557762 [5] NCCL INFO Channel 11/0 : 13[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636086:2636602 [2] NCCL INFO Channel 10/0 : 26[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557272:1557766 [6] NCCL INFO Channel 07/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557271:1557762 [5] NCCL INFO Channel 15/0 : 13[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557272:1557766 [6] NCCL INFO Channel 09/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557269:1557765 [3] NCCL INFO Channel 01/0 : 11[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96568:97091 [5] NCCL INFO Channel 03/0 : 13[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557272:1557766 [6] NCCL INFO Channel 11/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557269:1557765 [3] NCCL INFO Channel 03/0 : 11[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636090:2636598 [6] NCCL INFO Channel 10/0 : 30[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557272:1557766 [6] NCCL INFO Channel 13/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636088:2636601 [4] NCCL INFO Channel 10/0 : 28[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636087:2636603 [3] NCCL INFO Channel 10/0 : 19[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636089:2636600 [5] NCCL INFO Channel 00/0 : 29[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636085:2636604 [1] NCCL INFO Channel 10/0 : 17[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557269:1557765 [3] NCCL INFO Channel 05/0 : 11[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636091:2636599 [7] NCCL INFO Channel 14/0 : 31[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557272:1557766 [6] NCCL INFO Channel 15/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96563:97086 [0] NCCL INFO Channel 01/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96568:97091 [5] NCCL INFO Channel 05/0 : 13[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557269:1557765 [3] NCCL INFO Channel 07/0 : 11[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557269:1557765 [3] NCCL INFO Channel 09/0 : 11[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557269:1557765 [3] NCCL INFO Channel 13/0 : 11[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557269:1557765 [3] NCCL INFO Channel 15/0 : 11[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557268:1557764 [2] NCCL INFO Channel 00/0 : 18[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636086:2636602 [2] NCCL INFO Channel 12/0 : 26[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636084:2636605 [0] NCCL INFO Channel 02/0 : 24[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557268:1557764 [2] NCCL INFO Channel 04/0 : 18[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636090:2636598 [6] NCCL INFO Channel 12/0 : 30[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557268:1557764 [2] NCCL INFO Channel 06/0 : 18[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557268:1557764 [2] NCCL INFO Channel 08/0 : 18[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96568:97091 [5] NCCL INFO Channel 07/0 : 13[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96563:97086 [0] NCCL INFO Channel 03/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636088:2636601 [4] NCCL INFO Channel 12/0 : 28[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557268:1557764 [2] NCCL INFO Channel 10/0 : 18[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557268:1557764 [2] NCCL INFO Channel 12/0 : 18[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557268:1557764 [2] NCCL INFO Channel 14/0 : 18[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96568:97091 [5] NCCL INFO Channel 09/0 : 13[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96563:97086 [0] NCCL INFO Channel 05/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557268:1557764 [2] NCCL INFO Channel 00/0 : 2[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557268:1557764 [2] NCCL INFO Channel 04/0 : 2[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557268:1557764 [2] NCCL INFO Channel 06/0 : 2[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636091:2636599 [7] NCCL INFO Channel 00/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636089:2636600 [5] NCCL INFO Channel 02/0 : 29[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557268:1557764 [2] NCCL INFO Channel 08/0 : 2[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636087:2636603 [3] NCCL INFO Channel 12/0 : 19[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557268:1557764 [2] NCCL INFO Channel 10/0 : 2[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636086:2636602 [2] NCCL INFO Channel 14/0 : 26[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636084:2636605 [0] NCCL INFO Channel 04/0 : 24[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636090:2636598 [6] NCCL INFO Channel 14/0 : 30[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557268:1557764 [2] NCCL INFO Channel 12/0 : 2[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557268:1557764 [2] NCCL INFO Channel 14/0 : 2[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96568:97091 [5] NCCL INFO Channel 11/0 : 13[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96563:97086 [0] NCCL INFO Channel 07/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636088:2636601 [4] NCCL INFO Channel 14/0 : 28[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636091:2636599 [7] NCCL INFO Channel 01/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557268:1557764 [2] NCCL INFO Channel 01/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636089:2636600 [5] NCCL INFO Channel 04/0 : 29[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272152:272647 [2] NCCL INFO Channel 00/0 : 26[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96568:97091 [5] NCCL INFO Channel 15/0 : 13[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96563:97086 [0] NCCL INFO Channel 09/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557268:1557764 [2] NCCL INFO Channel 03/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636087:2636603 [3] NCCL INFO Channel 14/0 : 19[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272152:272647 [2] NCCL INFO Channel 04/0 : 26[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96563:97086 [0] NCCL INFO Channel 11/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557268:1557764 [2] NCCL INFO Channel 05/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636086:2636602 [2] NCCL INFO Channel 00/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636090:2636598 [6] NCCL INFO Channel 00/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636088:2636601 [4] NCCL INFO Channel 00/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636084:2636605 [0] NCCL INFO Channel 06/0 : 24[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636091:2636599 [7] NCCL INFO Channel 02/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636089:2636600 [5] NCCL INFO Channel 06/0 : 29[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272152:272647 [2] NCCL INFO Channel 06/0 : 26[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557268:1557764 [2] NCCL INFO Channel 07/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96563:97086 [0] NCCL INFO Channel 13/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636088:2636601 [4] NCCL INFO Channel 01/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636090:2636598 [6] NCCL INFO Channel 01/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62886:63378 [5] NCCL INFO Channel 01/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62882:63374 [1] NCCL INFO Channel 00/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272152:272647 [2] NCCL INFO Channel 08/0 : 26[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62887:63379 [6] NCCL INFO Channel 02/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557268:1557764 [2] NCCL INFO Channel 09/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96567:97090 [4] NCCL INFO Channel 11/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62885:63377 [4] NCCL INFO Channel 01/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62888:63380 [7] NCCL INFO Channel 09/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62883:63373 [2] NCCL INFO Channel 04/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636091:2636599 [7] NCCL INFO Channel 03/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62886:63378 [5] NCCL INFO Channel 02/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62887:63379 [6] NCCL INFO Channel 03/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636086:2636602 [2] NCCL INFO Channel 01/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636089:2636600 [5] NCCL INFO Channel 08/0 : 29[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636084:2636605 [0] NCCL INFO Channel 08/0 : 24[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62882:63374 [1] NCCL INFO Channel 02/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636087:2636603 [3] NCCL INFO Channel 00/0 : 27[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62885:63377 [4] NCCL INFO Channel 02/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557268:1557764 [2] NCCL INFO Channel 11/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272152:272647 [2] NCCL INFO Channel 10/0 : 26[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96563:97086 [0] NCCL INFO Channel 15/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62884:63376 [3] NCCL INFO Channel 00/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96567:97090 [4] NCCL INFO Channel 13/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62888:63380 [7] NCCL INFO Channel 10/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96566:97087 [3] NCCL INFO Channel 01/0 : 11[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62883:63373 [2] NCCL INFO Channel 05/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62886:63378 [5] NCCL INFO Channel 03/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636090:2636598 [6] NCCL INFO Channel 02/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636088:2636601 [4] NCCL INFO Channel 02/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62887:63379 [6] NCCL INFO Channel 04/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557268:1557764 [2] NCCL INFO Channel 13/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636091:2636599 [7] NCCL INFO Channel 04/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62882:63374 [1] NCCL INFO Channel 03/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62881:63375 [0] NCCL INFO Channel 01/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62885:63377 [4] NCCL INFO Channel 03/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62884:63376 [3] NCCL INFO Channel 01/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272152:272647 [2] NCCL INFO Channel 12/0 : 26[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62888:63380 [7] NCCL INFO Channel 11/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96567:97090 [4] NCCL INFO Channel 15/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96566:97087 [3] NCCL INFO Channel 03/0 : 11[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62883:63373 [2] NCCL INFO Channel 06/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62886:63378 [5] NCCL INFO Channel 04/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62887:63379 [6] NCCL INFO Channel 05/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636089:2636600 [5] NCCL INFO Channel 10/0 : 29[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62882:63374 [1] NCCL INFO Channel 04/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557268:1557764 [2] NCCL INFO Channel 15/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636087:2636603 [3] NCCL INFO Channel 02/0 : 27[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636084:2636605 [0] NCCL INFO Channel 10/0 : 24[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62881:63375 [0] NCCL INFO Channel 02/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272152:272647 [2] NCCL INFO Channel 14/0 : 26[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96566:97087 [3] NCCL INFO Channel 05/0 : 11[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62885:63377 [4] NCCL INFO Channel 05/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62888:63380 [7] NCCL INFO Channel 12/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62884:63376 [3] NCCL INFO Channel 02/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62887:63379 [6] NCCL INFO Channel 07/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62886:63378 [5] NCCL INFO Channel 06/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62882:63374 [1] NCCL INFO Channel 05/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636086:2636602 [2] NCCL INFO Channel 03/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62881:63375 [0] NCCL INFO Channel 03/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636090:2636598 [6] NCCL INFO Channel 03/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636088:2636601 [4] NCCL INFO Channel 03/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62883:63373 [2] NCCL INFO Channel 07/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636091:2636599 [7] NCCL INFO Channel 05/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62885:63377 [4] NCCL INFO Channel 06/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636089:2636600 [5] NCCL INFO Channel 12/0 : 29[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62888:63380 [7] NCCL INFO Channel 13/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636085:2636604 [1] NCCL INFO Channel 12/0 : 17[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62884:63376 [3] NCCL INFO Channel 04/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272152:272647 [2] NCCL INFO Channel 00/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62886:63378 [5] NCCL INFO Channel 07/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96566:97087 [3] NCCL INFO Channel 07/0 : 11[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62887:63379 [6] NCCL INFO Channel 08/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62882:63374 [1] NCCL INFO Channel 06/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62885:63377 [4] NCCL INFO Channel 07/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62881:63375 [0] NCCL INFO Channel 04/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62888:63380 [7] NCCL INFO Channel 14/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62883:63373 [2] NCCL INFO Channel 08/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62884:63376 [3] NCCL INFO Channel 05/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636087:2636603 [3] NCCL INFO Channel 04/0 : 27[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62886:63378 [5] NCCL INFO Channel 08/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636084:2636605 [0] NCCL INFO Channel 12/0 : 24[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62887:63379 [6] NCCL INFO Channel 09/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272152:272647 [2] NCCL INFO Channel 01/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636086:2636602 [2] NCCL INFO Channel 04/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636088:2636601 [4] NCCL INFO Channel 05/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636090:2636598 [6] NCCL INFO Channel 04/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62882:63374 [1] NCCL INFO Channel 07/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96566:97087 [3] NCCL INFO Channel 09/0 : 11[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62885:63377 [4] NCCL INFO Channel 08/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62888:63380 [7] NCCL INFO Channel 01/0 : 15[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62886:63378 [5] NCCL INFO Channel 09/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62881:63375 [0] NCCL INFO Channel 05/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62884:63376 [3] NCCL INFO Channel 06/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62883:63373 [2] NCCL INFO Channel 09/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62887:63379 [6] NCCL INFO Channel 10/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62882:63374 [1] NCCL INFO Channel 08/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636091:2636599 [7] NCCL INFO Channel 06/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62885:63377 [4] NCCL INFO Channel 09/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636089:2636600 [5] NCCL INFO Channel 14/0 : 29[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62888:63380 [7] NCCL INFO Channel 03/0 : 15[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272152:272647 [2] NCCL INFO Channel 03/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636085:2636604 [1] NCCL INFO Channel 14/0 : 17[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636087:2636603 [3] NCCL INFO Channel 06/0 : 27[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62886:63378 [5] NCCL INFO Channel 10/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636084:2636605 [0] NCCL INFO Channel 14/0 : 24[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96566:97087 [3] NCCL INFO Channel 13/0 : 11[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62887:63379 [6] NCCL INFO Channel 11/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62881:63375 [0] NCCL INFO Channel 06/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62884:63376 [3] NCCL INFO Channel 07/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62883:63373 [2] NCCL INFO Channel 11/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62882:63374 [1] NCCL INFO Channel 10/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62885:63377 [4] NCCL INFO Channel 10/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62888:63380 [7] NCCL INFO Channel 05/0 : 15[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62886:63378 [5] NCCL INFO Channel 11/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272152:272647 [2] NCCL INFO Channel 04/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62887:63379 [6] NCCL INFO Channel 12/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636086:2636602 [2] NCCL INFO Channel 05/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96566:97087 [3] NCCL INFO Channel 15/0 : 11[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62884:63376 [3] NCCL INFO Channel 08/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636090:2636598 [6] NCCL INFO Channel 05/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62883:63373 [2] NCCL INFO Channel 12/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62888:63380 [7] NCCL INFO Channel 07/0 : 15[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62881:63375 [0] NCCL INFO Channel 07/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62882:63374 [1] NCCL INFO Channel 11/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62885:63377 [4] NCCL INFO Channel 11/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636091:2636599 [7] NCCL INFO Channel 08/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636089:2636600 [5] NCCL INFO Channel 00/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636088:2636601 [4] NCCL INFO Channel 06/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62886:63378 [5] NCCL INFO Channel 12/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62887:63379 [6] NCCL INFO Channel 13/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62884:63376 [3] NCCL INFO Channel 09/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62883:63373 [2] NCCL INFO Channel 13/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272152:272647 [2] NCCL INFO Channel 05/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62888:63380 [7] NCCL INFO Channel 09/0 : 15[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62885:63377 [4] NCCL INFO Channel 13/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636087:2636603 [3] NCCL INFO Channel 08/0 : 27[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636084:2636605 [0] NCCL INFO Channel 01/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636086:2636602 [2] NCCL INFO Channel 06/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96564:97088 [1] NCCL INFO Channel 01/0 : 9[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62882:63374 [1] NCCL INFO Channel 12/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62886:63378 [5] NCCL INFO Channel 14/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62887:63379 [6] NCCL INFO Channel 15/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62881:63375 [0] NCCL INFO Channel 09/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62884:63376 [3] NCCL INFO Channel 10/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636090:2636598 [6] NCCL INFO Channel 07/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272152:272647 [2] NCCL INFO Channel 06/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62883:63373 [2] NCCL INFO Channel 14/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636089:2636600 [5] NCCL INFO Channel 01/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96564:97088 [1] NCCL INFO Channel 03/0 : 9[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62888:63380 [7] NCCL INFO Channel 11/0 : 15[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636088:2636601 [4] NCCL INFO Channel 07/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636091:2636599 [7] NCCL INFO Channel 09/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62885:63377 [4] NCCL INFO Channel 14/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62887:63379 [6] NCCL INFO Channel 01/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62881:63375 [0] NCCL INFO Channel 10/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62882:63374 [1] NCCL INFO Channel 13/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272152:272647 [2] NCCL INFO Channel 07/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62886:63378 [5] NCCL INFO Channel 15/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62884:63376 [3] NCCL INFO Channel 12/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636085:2636604 [1] NCCL INFO Channel 00/0 : 25[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636084:2636605 [0] NCCL INFO Channel 02/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636090:2636598 [6] NCCL INFO Channel 08/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62888:63380 [7] NCCL INFO Channel 13/0 : 15[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62883:63373 [2] NCCL INFO Channel 15/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62885:63377 [4] NCCL INFO Channel 15/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636086:2636602 [2] NCCL INFO Channel 07/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636089:2636600 [5] NCCL INFO Channel 02/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96564:97088 [1] NCCL INFO Channel 05/0 : 9[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62887:63379 [6] NCCL INFO Channel 03/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636087:2636603 [3] NCCL INFO Channel 10/0 : 27[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62881:63375 [0] NCCL INFO Channel 11/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62882:63374 [1] NCCL INFO Channel 14/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62886:63378 [5] NCCL INFO Channel 01/0 : 13[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62884:63376 [3] NCCL INFO Channel 13/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62883:63373 [2] NCCL INFO Channel 01/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62885:63377 [4] NCCL INFO Channel 01/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272152:272647 [2] NCCL INFO Channel 08/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62887:63379 [6] NCCL INFO Channel 05/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62881:63375 [0] NCCL INFO Channel 12/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96564:97088 [1] NCCL INFO Channel 07/0 : 9[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62882:63374 [1] NCCL INFO Channel 15/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62886:63378 [5] NCCL INFO Channel 03/0 : 13[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636091:2636599 [7] NCCL INFO Channel 10/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636088:2636601 [4] NCCL INFO Channel 08/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62884:63376 [3] NCCL INFO Channel 14/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62885:63377 [4] NCCL INFO Channel 03/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62887:63379 [6] NCCL INFO Channel 07/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62883:63373 [2] NCCL INFO Channel 03/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62881:63375 [0] NCCL INFO Channel 13/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272152:272647 [2] NCCL INFO Channel 09/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62882:63374 [1] NCCL INFO Channel 01/0 : 9[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62886:63378 [5] NCCL INFO Channel 05/0 : 13[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62885:63377 [4] NCCL INFO Channel 05/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62887:63379 [6] NCCL INFO Channel 09/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62884:63376 [3] NCCL INFO Channel 15/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636085:2636604 [1] NCCL INFO Channel 02/0 : 25[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96564:97088 [1] NCCL INFO Channel 11/0 : 9[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62881:63375 [0] NCCL INFO Channel 14/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62883:63373 [2] NCCL INFO Channel 05/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62882:63374 [1] NCCL INFO Channel 03/0 : 9[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62886:63378 [5] NCCL INFO Channel 07/0 : 13[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62885:63377 [4] NCCL INFO Channel 07/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272152:272647 [2] NCCL INFO Channel 11/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636090:2636598 [6] NCCL INFO Channel 09/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636084:2636605 [0] NCCL INFO Channel 03/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636089:2636600 [5] NCCL INFO Channel 03/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96564:97088 [1] NCCL INFO Channel 13/0 : 9[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62887:63379 [6] NCCL INFO Channel 11/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62884:63376 [3] NCCL INFO Channel 01/0 : 11[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62881:63375 [0] NCCL INFO Channel 15/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62883:63373 [2] NCCL INFO Channel 07/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272152:272647 [2] NCCL INFO Channel 12/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62882:63374 [1] NCCL INFO Channel 05/0 : 9[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62886:63378 [5] NCCL INFO Channel 09/0 : 13[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636091:2636599 [7] NCCL INFO Channel 11/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636086:2636602 [2] NCCL INFO Channel 08/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636088:2636601 [4] NCCL INFO Channel 09/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62885:63377 [4] NCCL INFO Channel 09/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636087:2636603 [3] NCCL INFO Channel 12/0 : 27[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62887:63379 [6] NCCL INFO Channel 13/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636085:2636604 [1] NCCL INFO Channel 04/0 : 25[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96564:97088 [1] NCCL INFO Channel 15/0 : 9[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62884:63376 [3] NCCL INFO Channel 03/0 : 11[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636090:2636598 [6] NCCL INFO Channel 10/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62881:63375 [0] NCCL INFO Channel 01/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272152:272647 [2] NCCL INFO Channel 13/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62882:63374 [1] NCCL INFO Channel 07/0 : 9[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62883:63373 [2] NCCL INFO Channel 09/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62885:63377 [4] NCCL INFO Channel 11/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62886:63378 [5] NCCL INFO Channel 11/0 : 13[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62887:63379 [6] NCCL INFO Channel 15/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62884:63376 [3] NCCL INFO Channel 05/0 : 11[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62881:63375 [0] NCCL INFO Channel 03/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636084:2636605 [0] NCCL INFO Channel 04/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272152:272647 [2] NCCL INFO Channel 14/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62883:63373 [2] NCCL INFO Channel 11/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62882:63374 [1] NCCL INFO Channel 11/0 : 9[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62886:63378 [5] NCCL INFO Channel 15/0 : 13[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62885:63377 [4] NCCL INFO Channel 13/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636089:2636600 [5] NCCL INFO Channel 04/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636091:2636599 [7] NCCL INFO Channel 12/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62881:63375 [0] NCCL INFO Channel 05/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62884:63376 [3] NCCL INFO Channel 07/0 : 11[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636088:2636601 [4] NCCL INFO Channel 10/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62883:63373 [2] NCCL INFO Channel 13/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62885:63377 [4] NCCL INFO Channel 15/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62882:63374 [1] NCCL INFO Channel 13/0 : 9[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62881:63375 [0] NCCL INFO Channel 07/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62884:63376 [3] NCCL INFO Channel 09/0 : 11[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62883:63373 [2] NCCL INFO Channel 15/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272152:272647 [2] NCCL INFO Channel 15/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636086:2636602 [2] NCCL INFO Channel 09/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636087:2636603 [3] NCCL INFO Channel 14/0 : 27[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636090:2636598 [6] NCCL INFO Channel 11/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636085:2636604 [1] NCCL INFO Channel 06/0 : 25[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636084:2636605 [0] NCCL INFO Channel 05/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636089:2636600 [5] NCCL INFO Channel 06/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636091:2636599 [7] NCCL INFO Channel 13/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62881:63375 [0] NCCL INFO Channel 09/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62882:63374 [1] NCCL INFO Channel 15/0 : 9[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62884:63376 [3] NCCL INFO Channel 13/0 : 11[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62881:63375 [0] NCCL INFO Channel 11/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636088:2636601 [4] NCCL INFO Channel 11/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62884:63376 [3] NCCL INFO Channel 15/0 : 11[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62881:63375 [0] NCCL INFO Channel 13/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636090:2636598 [6] NCCL INFO Channel 12/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62881:63375 [0] NCCL INFO Channel 15/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636087:2636603 [3] NCCL INFO Channel 00/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636089:2636600 [5] NCCL INFO Channel 07/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636086:2636602 [2] NCCL INFO Channel 11/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636091:2636599 [7] NCCL INFO Channel 14/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636088:2636601 [4] NCCL INFO Channel 13/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636085:2636604 [1] NCCL INFO Channel 08/0 : 25[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636084:2636605 [0] NCCL INFO Channel 06/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636090:2636598 [6] NCCL INFO Channel 13/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636087:2636603 [3] NCCL INFO Channel 01/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636089:2636600 [5] NCCL INFO Channel 08/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636086:2636602 [2] NCCL INFO Channel 12/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636088:2636601 [4] NCCL INFO Channel 14/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636090:2636598 [6] NCCL INFO Channel 15/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636084:2636605 [0] NCCL INFO Channel 07/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636085:2636604 [1] NCCL INFO Channel 10/0 : 25[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636087:2636603 [3] NCCL INFO Channel 02/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636089:2636600 [5] NCCL INFO Channel 09/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636086:2636602 [2] NCCL INFO Channel 13/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636088:2636601 [4] NCCL INFO Channel 15/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636084:2636605 [0] NCCL INFO Channel 09/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636085:2636604 [1] NCCL INFO Channel 12/0 : 25[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636087:2636603 [3] NCCL INFO Channel 04/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636089:2636600 [5] NCCL INFO Channel 10/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636086:2636602 [2] NCCL INFO Channel 14/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636085:2636604 [1] NCCL INFO Channel 14/0 : 25[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636089:2636600 [5] NCCL INFO Channel 11/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636086:2636602 [2] NCCL INFO Channel 15/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636087:2636603 [3] NCCL INFO Channel 05/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636084:2636605 [0] NCCL INFO Channel 10/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636085:2636604 [1] NCCL INFO Channel 00/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636089:2636600 [5] NCCL INFO Channel 12/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636084:2636605 [0] NCCL INFO Channel 11/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636085:2636604 [1] NCCL INFO Channel 02/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636089:2636600 [5] NCCL INFO Channel 14/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636087:2636603 [3] NCCL INFO Channel 06/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636084:2636605 [0] NCCL INFO Channel 12/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636085:2636604 [1] NCCL INFO Channel 03/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636089:2636600 [5] NCCL INFO Channel 15/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636087:2636603 [3] NCCL INFO Channel 07/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636084:2636605 [0] NCCL INFO Channel 13/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636087:2636603 [3] NCCL INFO Channel 08/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636085:2636604 [1] NCCL INFO Channel 04/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636084:2636605 [0] NCCL INFO Channel 14/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636087:2636603 [3] NCCL INFO Channel 09/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636084:2636605 [0] NCCL INFO Channel 15/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636085:2636604 [1] NCCL INFO Channel 05/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636087:2636603 [3] NCCL INFO Channel 10/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636085:2636604 [1] NCCL INFO Channel 06/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636087:2636603 [3] NCCL INFO Channel 12/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636087:2636603 [3] NCCL INFO Channel 13/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636085:2636604 [1] NCCL INFO Channel 07/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636087:2636603 [3] NCCL INFO Channel 14/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636085:2636604 [1] NCCL INFO Channel 08/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636087:2636603 [3] NCCL INFO Channel 15/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636085:2636604 [1] NCCL INFO Channel 10/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636085:2636604 [1] NCCL INFO Channel 11/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636085:2636604 [1] NCCL INFO Channel 12/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636085:2636604 [1] NCCL INFO Channel 13/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636085:2636604 [1] NCCL INFO Channel 14/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636085:2636604 [1] NCCL INFO Channel 15/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815968:2816472 [5] NCCL INFO Channel 08/0 : 29[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815964:2816476 [1] NCCL INFO Channel 04/0 : 25[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815963:2816477 [0] NCCL INFO Channel 02/0 : 24[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815969:2816471 [6] NCCL INFO Channel 12/0 : 30[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815967:2816473 [4] NCCL INFO Channel 10/0 : 28[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815966:2816474 [3] NCCL INFO Channel 02/0 : 27[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815968:2816472 [5] NCCL INFO Channel 10/0 : 29[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815964:2816476 [1] NCCL INFO Channel 06/0 : 25[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815963:2816477 [0] NCCL INFO Channel 04/0 : 24[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815969:2816471 [6] NCCL INFO Channel 14/0 : 30[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815967:2816473 [4] NCCL INFO Channel 12/0 : 28[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815966:2816474 [3] NCCL INFO Channel 04/0 : 27[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815968:2816472 [5] NCCL INFO Channel 12/0 : 29[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815964:2816476 [1] NCCL INFO Channel 08/0 : 25[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815963:2816477 [0] NCCL INFO Channel 06/0 : 24[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815967:2816473 [4] NCCL INFO Channel 14/0 : 28[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815966:2816474 [3] NCCL INFO Channel 06/0 : 27[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815964:2816476 [1] NCCL INFO Channel 10/0 : 25[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815968:2816472 [5] NCCL INFO Channel 14/0 : 29[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815963:2816477 [0] NCCL INFO Channel 08/0 : 24[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815964:2816476 [1] NCCL INFO Channel 12/0 : 25[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815966:2816474 [3] NCCL INFO Channel 08/0 : 27[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815963:2816477 [0] NCCL INFO Channel 10/0 : 24[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815964:2816476 [1] NCCL INFO Channel 14/0 : 25[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815966:2816474 [3] NCCL INFO Channel 10/0 : 27[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815963:2816477 [0] NCCL INFO Channel 12/0 : 24[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815963:2816477 [0] NCCL INFO Channel 14/0 : 24[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815966:2816474 [3] NCCL INFO Channel 12/0 : 27[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815966:2816474 [3] NCCL INFO Channel 14/0 : 27[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153564:154057 [2] NCCL INFO Connected NVLS tree
+ip-26-0-165-202:153564:154057 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-202:153564:154057 [2] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-165-202:153564:154057 [2] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-78:114513:115008 [1] NCCL INFO Connected NVLS tree
+ip-26-0-161-78:114513:115008 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-78:114513:115008 [1] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-161-78:114513:115008 [1] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-78:114517:115013 [5] NCCL INFO Connected NVLS tree
+ip-26-0-161-78:114517:115013 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-78:114517:115013 [5] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-161-78:114517:115013 [5] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-78:114515:115009 [3] NCCL INFO Connected NVLS tree
+ip-26-0-161-78:114515:115009 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-78:114515:115009 [3] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-161-78:114515:115009 [3] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-78:114519:115014 [7] NCCL INFO Connected NVLS tree
+ip-26-0-161-78:114519:115014 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-78:114519:115014 [7] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-161-78:114519:115014 [7] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-123:32708:33198 [4] NCCL INFO Connected NVLS tree
+ip-26-0-161-123:32708:33198 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-123:32708:33198 [4] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-161-123:32708:33198 [4] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-123:32704:33197 [0] NCCL INFO Connected NVLS tree
+ip-26-0-161-123:32704:33197 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-123:32704:33197 [0] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-161-123:32704:33197 [0] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-160-225:124164:124663 [4] NCCL INFO Connected NVLS tree
+ip-26-0-160-225:124164:124663 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-225:124164:124663 [4] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-160-225:124164:124663 [4] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-160-225:124162:124660 [2] NCCL INFO Connected NVLS tree
+ip-26-0-160-225:124162:124660 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-225:124162:124660 [2] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-160-225:124162:124660 [2] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-123:32706:33199 [2] NCCL INFO Connected NVLS tree
+ip-26-0-161-123:32706:33199 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-123:32706:33199 [2] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-161-123:32706:33199 [2] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-160-225:124166:124665 [6] NCCL INFO Connected NVLS tree
+ip-26-0-160-225:124166:124665 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-225:124166:124665 [6] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-160-225:124166:124665 [6] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-160-225:124163:124661 [3] NCCL INFO Connected NVLS tree
+ip-26-0-160-225:124163:124661 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-225:124163:124661 [3] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-160-225:124163:124661 [3] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-160-225:124167:124664 [7] NCCL INFO Connected NVLS tree
+ip-26-0-160-225:124167:124664 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-225:124167:124664 [7] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-160-225:124167:124664 [7] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-160-225:124161:124659 [1] NCCL INFO Connected NVLS tree
+ip-26-0-160-225:124161:124659 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-225:124161:124659 [1] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-160-225:124161:124659 [1] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-160-225:124165:124662 [5] NCCL INFO Connected NVLS tree
+ip-26-0-160-225:124165:124662 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-225:124165:124662 [5] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-160-225:124165:124662 [5] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-165-202:153566:154062 [4] NCCL INFO Connected NVLS tree
+ip-26-0-165-202:153566:154062 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-202:153566:154062 [4] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-165-202:153566:154062 [4] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-123:32710:33202 [6] NCCL INFO Connected NVLS tree
+ip-26-0-161-123:32710:33202 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-123:32710:33202 [6] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-161-123:32710:33202 [6] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-165-202:153562:154061 [0] NCCL INFO Connected NVLS tree
+ip-26-0-165-202:153562:154061 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-202:153562:154061 [0] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-165-202:153562:154061 [0] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-165-202:153568:154056 [6] NCCL INFO Connected NVLS tree
+ip-26-0-165-202:153568:154056 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-202:153568:154056 [6] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-165-202:153568:154056 [6] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-103:96892:97383 [4] NCCL INFO Connected NVLS tree
+ip-26-0-161-103:96892:97383 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-103:96892:97383 [4] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-161-103:96892:97383 [4] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-138:96569:97092 [6] NCCL INFO Connected NVLS tree
+ip-26-0-161-138:96569:97092 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-138:96569:97092 [6] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-161-138:96569:97092 [6] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-138:96565:97089 [2] NCCL INFO Connected NVLS tree
+ip-26-0-161-138:96565:97089 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-138:96565:97089 [2] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-161-138:96565:97089 [2] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-138:96570:97093 [7] NCCL INFO Connected NVLS tree
+ip-26-0-161-138:96570:97093 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-138:96570:97093 [7] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-161-138:96570:97093 [7] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-138:96567:97090 [4] NCCL INFO Connected NVLS tree
+ip-26-0-161-138:96567:97090 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-138:96567:97090 [4] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-161-138:96567:97090 [4] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-123:32707:33200 [3] NCCL INFO Connected NVLS tree
+ip-26-0-161-123:32707:33200 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-123:32707:33200 [3] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-161-123:32707:33200 [3] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-165-59:1557270:1557763 [4] NCCL INFO Connected NVLS tree
+ip-26-0-165-59:1557270:1557763 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-59:1557270:1557763 [4] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-165-59:1557270:1557763 [4] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-138:96564:97088 [1] NCCL INFO Connected NVLS tree
+ip-26-0-161-138:96564:97088 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-138:96564:97088 [1] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-161-138:96564:97088 [1] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-103:96890:97386 [2] NCCL INFO Connected NVLS tree
+ip-26-0-161-103:96890:97386 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-103:96890:97386 [2] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-161-103:96890:97386 [2] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-123:32709:33201 [5] NCCL INFO Connected NVLS tree
+ip-26-0-161-123:32709:33201 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-123:32709:33201 [5] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-161-123:32709:33201 [5] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-165-59:1557266:1557759 [0] NCCL INFO Connected NVLS tree
+ip-26-0-165-59:1557266:1557759 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-59:1557266:1557759 [0] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-165-59:1557266:1557759 [0] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-123:32711:33204 [7] NCCL INFO Connected NVLS tree
+ip-26-0-161-123:32711:33204 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-123:32711:33204 [7] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-161-123:32711:33204 [7] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-165-59:1557272:1557766 [6] NCCL INFO Connected NVLS tree
+ip-26-0-165-59:1557272:1557766 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-59:1557272:1557766 [6] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-165-59:1557272:1557766 [6] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-138:96566:97087 [3] NCCL INFO Connected NVLS tree
+ip-26-0-161-138:96566:97087 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-138:96566:97087 [3] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-161-138:96566:97087 [3] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-165-59:1557269:1557765 [3] NCCL INFO Connected NVLS tree
+ip-26-0-165-59:1557269:1557765 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-59:1557269:1557765 [3] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-165-59:1557269:1557765 [3] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-138:96568:97091 [5] NCCL INFO Connected NVLS tree
+ip-26-0-161-138:96568:97091 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-138:96568:97091 [5] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-161-138:96568:97091 [5] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-165-59:1557267:1557761 [1] NCCL INFO Connected NVLS tree
+ip-26-0-165-59:1557267:1557761 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-59:1557267:1557761 [1] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-165-59:1557267:1557761 [1] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-165-59:1557268:1557764 [2] NCCL INFO Connected NVLS tree
+ip-26-0-165-59:1557268:1557764 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-59:1557268:1557764 [2] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-165-59:1557268:1557764 [2] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-165-59:1557273:1557760 [7] NCCL INFO Connected NVLS tree
+ip-26-0-165-59:1557273:1557760 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-59:1557273:1557760 [7] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-165-59:1557273:1557760 [7] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-103:96894:97381 [6] NCCL INFO Connected NVLS tree
+ip-26-0-161-103:96894:97381 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-103:96894:97381 [6] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-161-103:96894:97381 [6] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-165-59:1557271:1557762 [5] NCCL INFO Connected NVLS tree
+ip-26-0-165-59:1557271:1557762 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-59:1557271:1557762 [5] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-165-59:1557271:1557762 [5] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-103:96895:97382 [7] NCCL INFO Connected NVLS tree
+ip-26-0-161-103:96895:97382 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-103:96895:97382 [7] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-161-103:96895:97382 [7] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-160-242:1364909:1365405 [7] NCCL INFO Connected NVLS tree
+ip-26-0-160-242:1364909:1365405 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-123:32705:33203 [1] NCCL INFO Connected NVLS tree
+ip-26-0-160-242:1364909:1365405 [7] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-160-242:1364909:1365405 [7] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-123:32705:33203 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-123:32705:33203 [1] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-161-123:32705:33203 [1] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-103:96891:97385 [3] NCCL INFO Connected NVLS tree
+ip-26-0-161-103:96891:97385 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-103:96891:97385 [3] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-161-103:96891:97385 [3] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-160-242:1364904:1365408 [2] NCCL INFO Connected NVLS tree
+ip-26-0-160-242:1364904:1365408 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-242:1364904:1365408 [2] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-160-242:1364904:1365408 [2] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-160-225:124160:124658 [0] NCCL INFO Connected NVLS tree
+ip-26-0-160-225:124160:124658 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-225:124160:124658 [0] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-160-225:124160:124658 [0] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-103:96888:97387 [0] NCCL INFO Connected NVLS tree
+ip-26-0-161-103:96888:97387 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-103:96888:97387 [0] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-161-103:96888:97387 [0] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-123:32710:33202 [6] NCCL INFO comm 0xb0d3b90 rank 30 nranks 32 cudaDev 6 nvmlDev 6 busId b9000 commId 0x6ad24daef1fe15ce - Init COMPLETE
+ip-26-0-161-123:32704:33197 [0] NCCL INFO comm 0xbc119e0 rank 24 nranks 32 cudaDev 0 nvmlDev 0 busId 53000 commId 0x6ad24daef1fe15ce - Init COMPLETE
+ip-26-0-161-123:32706:33199 [2] NCCL INFO comm 0xbc2d3c0 rank 26 nranks 32 cudaDev 2 nvmlDev 2 busId 75000 commId 0x6ad24daef1fe15ce - Init COMPLETE
+ip-26-0-161-123:32707:33200 [3] NCCL INFO comm 0xa02c190 rank 27 nranks 32 cudaDev 3 nvmlDev 3 busId 86000 commId 0x6ad24daef1fe15ce - Init COMPLETE
+ip-26-0-161-123:32708:33198 [4] NCCL INFO comm 0xba333f0 rank 28 nranks 32 cudaDev 4 nvmlDev 4 busId 97000 commId 0x6ad24daef1fe15ce - Init COMPLETE
+ip-26-0-161-123:32709:33201 [5] NCCL INFO comm 0xb9f5040 rank 29 nranks 32 cudaDev 5 nvmlDev 5 busId a8000 commId 0x6ad24daef1fe15ce - Init COMPLETE
+ip-26-0-161-123:32705:33203 [1] NCCL INFO comm 0xb19e430 rank 25 nranks 32 cudaDev 1 nvmlDev 1 busId 64000 commId 0x6ad24daef1fe15ce - Init COMPLETE
+ip-26-0-161-123:32711:33204 [7] NCCL INFO comm 0xc08b660 rank 31 nranks 32 cudaDev 7 nvmlDev 7 busId ca000 commId 0x6ad24daef1fe15ce - Init COMPLETE
+ip-26-0-160-242:1364908:1365406 [6] NCCL INFO Connected NVLS tree
+ip-26-0-160-242:1364908:1365406 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-242:1364908:1365406 [6] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-160-242:1364908:1365406 [6] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-160-242:1364906:1365407 [4] NCCL INFO Connected NVLS tree
+ip-26-0-160-242:1364906:1365407 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-242:1364906:1365407 [4] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-160-242:1364906:1365407 [4] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-165-202:153565:154058 [3] NCCL INFO Connected NVLS tree
+ip-26-0-165-202:153565:154058 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-202:153565:154058 [3] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-165-202:153565:154058 [3] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-138:96563:97086 [0] NCCL INFO Connected NVLS tree
+ip-26-0-161-138:96563:97086 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-138:96563:97086 [0] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-161-138:96563:97086 [0] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-160-225:124160:124658 [0] NCCL INFO comm 0xb447d40 rank 0 nranks 32 cudaDev 0 nvmlDev 0 busId 53000 commId 0x6ad24daef1fe15ce - Init COMPLETE
+ip-26-0-160-225:124161:124659 [1] NCCL INFO comm 0x9fafb60 rank 1 nranks 32 cudaDev 1 nvmlDev 1 busId 64000 commId 0x6ad24daef1fe15ce - Init COMPLETE
+ip-26-0-160-225:124163:124661 [3] NCCL INFO comm 0xb6dea80 rank 3 nranks 32 cudaDev 3 nvmlDev 3 busId 86000 commId 0x6ad24daef1fe15ce - Init COMPLETE
+ip-26-0-160-225:124165:124662 [5] NCCL INFO comm 0x9fea760 rank 5 nranks 32 cudaDev 5 nvmlDev 5 busId a8000 commId 0x6ad24daef1fe15ce - Init COMPLETE
+ip-26-0-160-225:124164:124663 [4] NCCL INFO comm 0xa721840 rank 4 nranks 32 cudaDev 4 nvmlDev 4 busId 97000 commId 0x6ad24daef1fe15ce - Init COMPLETE
+ip-26-0-160-225:124167:124664 [7] NCCL INFO comm 0xb3f9260 rank 7 nranks 32 cudaDev 7 nvmlDev 7 busId ca000 commId 0x6ad24daef1fe15ce - Init COMPLETE
+ip-26-0-160-225:124166:124665 [6] NCCL INFO comm 0xaf0f490 rank 6 nranks 32 cudaDev 6 nvmlDev 6 busId b9000 commId 0x6ad24daef1fe15ce - Init COMPLETE
+ip-26-0-160-225:124162:124660 [2] NCCL INFO comm 0xba5d2d0 rank 2 nranks 32 cudaDev 2 nvmlDev 2 busId 75000 commId 0x6ad24daef1fe15ce - Init COMPLETE
+ip-26-0-165-59:1557268:1557764 [2] NCCL INFO comm 0xab81880 rank 2 nranks 32 cudaDev 2 nvmlDev 2 busId 75000 commId 0xe6d0f49db1af694c - Init COMPLETE
+ip-26-0-165-59:1557272:1557766 [6] NCCL INFO comm 0xa9e2270 rank 6 nranks 32 cudaDev 6 nvmlDev 6 busId b9000 commId 0xe6d0f49db1af694c - Init COMPLETE
+ip-26-0-165-59:1557266:1557759 [0] NCCL INFO comm 0xc14cf00 rank 0 nranks 32 cudaDev 0 nvmlDev 0 busId 53000 commId 0xe6d0f49db1af694c - Init COMPLETE
+ip-26-0-165-59:1557271:1557762 [5] NCCL INFO comm 0xa4e7e00 rank 5 nranks 32 cudaDev 5 nvmlDev 5 busId a8000 commId 0xe6d0f49db1af694c - Init COMPLETE
+ip-26-0-165-59:1557269:1557765 [3] NCCL INFO comm 0xb4cbdc0 rank 3 nranks 32 cudaDev 3 nvmlDev 3 busId 86000 commId 0xe6d0f49db1af694c - Init COMPLETE
+ip-26-0-165-59:1557270:1557763 [4] NCCL INFO comm 0xa956c80 rank 4 nranks 32 cudaDev 4 nvmlDev 4 busId 97000 commId 0xe6d0f49db1af694c - Init COMPLETE
+ip-26-0-165-59:1557267:1557761 [1] NCCL INFO comm 0x9e76b70 rank 1 nranks 32 cudaDev 1 nvmlDev 1 busId 64000 commId 0xe6d0f49db1af694c - Init COMPLETE
+ip-26-0-165-59:1557273:1557760 [7] NCCL INFO comm 0xb5cc7e0 rank 7 nranks 32 cudaDev 7 nvmlDev 7 busId ca000 commId 0xe6d0f49db1af694c - Init COMPLETE
+ip-26-0-161-103:96889:97388 [1] NCCL INFO Connected NVLS tree
+ip-26-0-161-103:96889:97388 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-103:96889:97388 [1] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-161-103:96889:97388 [1] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-160-242:1364905:1365404 [3] NCCL INFO Connected NVLS tree
+ip-26-0-160-242:1364905:1365404 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-242:1364905:1365404 [3] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-160-242:1364905:1365404 [3] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-160-242:1364902:1365410 [0] NCCL INFO Connected NVLS tree
+ip-26-0-160-242:1364902:1365410 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-242:1364902:1365410 [0] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-160-242:1364902:1365410 [0] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-160-242:1364903:1365409 [1] NCCL INFO Connected NVLS tree
+ip-26-0-160-242:1364903:1365409 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-242:1364903:1365409 [1] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-160-242:1364903:1365409 [1] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-165-202:153563:154060 [1] NCCL INFO Connected NVLS tree
+ip-26-0-165-202:153563:154060 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-202:153563:154060 [1] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-165-202:153563:154060 [1] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-138:96570:97093 [7] NCCL INFO comm 0xb873da0 rank 7 nranks 32 cudaDev 7 nvmlDev 7 busId ca000 commId 0xc9fdc2638087cf9c - Init COMPLETE
+ip-26-0-161-138:96568:97091 [5] NCCL INFO comm 0xa08c730 rank 5 nranks 32 cudaDev 5 nvmlDev 5 busId a8000 commId 0xc9fdc2638087cf9c - Init COMPLETE
+ip-26-0-161-138:96564:97088 [1] NCCL INFO comm 0xadd6e40 rank 1 nranks 32 cudaDev 1 nvmlDev 1 busId 64000 commId 0xc9fdc2638087cf9c - Init COMPLETE
+ip-26-0-161-138:96566:97087 [3] NCCL INFO comm 0xb531f70 rank 3 nranks 32 cudaDev 3 nvmlDev 3 busId 86000 commId 0xc9fdc2638087cf9c - Init COMPLETE
+ip-26-0-161-138:96565:97089 [2] NCCL INFO comm 0xb9b8c10 rank 2 nranks 32 cudaDev 2 nvmlDev 2 busId 75000 commId 0xc9fdc2638087cf9c - Init COMPLETE
+ip-26-0-161-138:96569:97092 [6] NCCL INFO comm 0xa638af0 rank 6 nranks 32 cudaDev 6 nvmlDev 6 busId b9000 commId 0xc9fdc2638087cf9c - Init COMPLETE
+ip-26-0-161-138:96567:97090 [4] NCCL INFO comm 0xa01fba0 rank 4 nranks 32 cudaDev 4 nvmlDev 4 busId 97000 commId 0xc9fdc2638087cf9c - Init COMPLETE
+ip-26-0-161-138:96563:97086 [0] NCCL INFO comm 0xb166280 rank 0 nranks 32 cudaDev 0 nvmlDev 0 busId 53000 commId 0xc9fdc2638087cf9c - Init COMPLETE
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+ip-26-0-165-59:1557271:1557795 [5] NCCL INFO Using network Libfabric
+ip-26-0-161-138:96568:97143 [5] NCCL INFO Using network Libfabric
+ip-26-0-161-138:96565:97145 [2] NCCL INFO Using network Libfabric
+ip-26-0-161-138:96569:97147 [6] NCCL INFO Using network Libfabric
+ip-26-0-161-138:96570:97146 [7] NCCL INFO Using network Libfabric
+ip-26-0-161-138:96567:97144 [4] NCCL INFO Using network Libfabric
+ip-26-0-161-138:96563:97148 [0] NCCL INFO Using network Libfabric
+ip-26-0-165-59:1557268:1557796 [2] NCCL INFO Using network Libfabric
+ip-26-0-161-138:96564:97149 [1] NCCL INFO Using network Libfabric
+ip-26-0-165-59:1557267:1557799 [1] NCCL INFO Using network Libfabric
+ip-26-0-165-59:1557266:1557798 [0] NCCL INFO Using network Libfabric
+ip-26-0-165-59:1557270:1557797 [4] NCCL INFO Using network Libfabric
+ip-26-0-165-59:1557272:1557800 [6] NCCL INFO Using network Libfabric
+ip-26-0-165-59:1557273:1557801 [7] NCCL INFO Using network Libfabric
+ip-26-0-161-138:96566:97151 [3] NCCL INFO Using network Libfabric
+ip-26-0-165-59:1557269:1557802 [3] NCCL INFO Using network Libfabric
+ip-26-0-161-138:96563:97148 [0] NCCL INFO comm 0xbbb44c0 rank 0 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0xd6a2a23e4b7c8957 - Init START
+ip-26-0-165-59:1557266:1557798 [0] NCCL INFO comm 0xc163230 rank 1 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0xd6a2a23e4b7c8957 - Init START
+ip-26-0-165-59:1557270:1557797 [4] NCCL INFO comm 0x9eef930 rank 1 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0xb4e67d13f14ca523 - Init START
+ip-26-0-161-138:96567:97144 [4] NCCL INFO comm 0xaa6d9d0 rank 0 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0xb4e67d13f14ca523 - Init START
+ip-26-0-165-59:1557266:1557798 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-59:1557270:1557797 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-138:96563:97148 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-138:96567:97144 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-59:1557269:1557802 [3] NCCL INFO comm 0xbf320c0 rank 1 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0xf91b029843677563 - Init START
+ip-26-0-161-138:96566:97151 [3] NCCL INFO comm 0xaac4830 rank 0 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0xf91b029843677563 - Init START
+ip-26-0-165-59:1557269:1557802 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-103:96893:97384 [5] NCCL INFO Connected NVLS tree
+ip-26-0-161-103:96893:97384 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-103:96893:97384 [5] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-161-103:96893:97384 [5] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-138:96566:97151 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-59:1557271:1557795 [5] NCCL INFO comm 0x9a81660 rank 1 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0x901e509d3ac79cef - Init START
+ip-26-0-161-138:96568:97143 [5] NCCL INFO comm 0xaaf0f50 rank 0 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0x901e509d3ac79cef - Init START
+ip-26-0-165-59:1557271:1557795 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-59:1557268:1557796 [2] NCCL INFO comm 0xa11a8f0 rank 1 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0x9392ca3f181752f6 - Init START
+ip-26-0-165-59:1557273:1557801 [7] NCCL INFO comm 0xc02c5d0 rank 1 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0xc5a3cb775d798355 - Init START
+ip-26-0-165-59:1557272:1557800 [6] NCCL INFO comm 0xb431a90 rank 1 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0x4e9bc1b782930120 - Init START
+ip-26-0-161-138:96565:97145 [2] NCCL INFO comm 0xb9305e0 rank 0 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0x9392ca3f181752f6 - Init START
+ip-26-0-161-138:96569:97147 [6] NCCL INFO comm 0xb09d550 rank 0 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0x4e9bc1b782930120 - Init START
+ip-26-0-161-138:96568:97143 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-138:96570:97146 [7] NCCL INFO comm 0xc2c1fb0 rank 0 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0xc5a3cb775d798355 - Init START
+ip-26-0-161-138:96565:97145 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-59:1557268:1557796 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-59:1557273:1557801 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-59:1557272:1557800 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-138:96570:97146 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-138:96569:97147 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-138:96564:97149 [1] NCCL INFO comm 0xb824ca0 rank 0 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0x6aa89d25d313e4b5 - Init START
+ip-26-0-165-59:1557267:1557799 [1] NCCL INFO comm 0xa8dce00 rank 1 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0x6aa89d25d313e4b5 - Init START
+ip-26-0-161-138:96564:97149 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-59:1557267:1557799 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-202:153569:154055 [7] NCCL INFO Connected NVLS tree
+ip-26-0-165-202:153569:154055 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-202:153569:154055 [7] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-165-202:153569:154055 [7] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-165-202:153567:154059 [5] NCCL INFO Connected NVLS tree
+ip-26-0-165-202:153567:154059 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-202:153567:154059 [5] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-165-202:153567:154059 [5] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-165-164:1227466:1227961 [3] NCCL INFO Connected NVLS tree
+ip-26-0-165-164:1227466:1227961 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-164:1227466:1227961 [3] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-165-164:1227466:1227961 [3] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-160-242:1364907:1365403 [5] NCCL INFO Connected NVLS tree
+ip-26-0-160-242:1364907:1365403 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-242:1364907:1365403 [5] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-160-242:1364907:1365403 [5] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-78:114516:115010 [4] NCCL INFO Connected NVLS tree
+ip-26-0-161-78:114516:115010 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-78:114516:115010 [4] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-161-78:114516:115010 [4] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-165-202:153569:154055 [7] NCCL INFO comm 0x9d016a0 rank 31 nranks 32 cudaDev 7 nvmlDev 7 busId ca000 commId 0x968ef8c3ebb85844 - Init COMPLETE
+ip-26-0-165-202:153565:154058 [3] NCCL INFO comm 0xb9d80c0 rank 27 nranks 32 cudaDev 3 nvmlDev 3 busId 86000 commId 0x968ef8c3ebb85844 - Init COMPLETE
+ip-26-0-165-202:153567:154059 [5] NCCL INFO comm 0xc2061c0 rank 29 nranks 32 cudaDev 5 nvmlDev 5 busId a8000 commId 0x968ef8c3ebb85844 - Init COMPLETE
+ip-26-0-165-202:153563:154060 [1] NCCL INFO comm 0xb9a2ae0 rank 25 nranks 32 cudaDev 1 nvmlDev 1 busId 64000 commId 0x968ef8c3ebb85844 - Init COMPLETE
+ip-26-0-165-202:153566:154062 [4] NCCL INFO comm 0xaa4e780 rank 28 nranks 32 cudaDev 4 nvmlDev 4 busId 97000 commId 0x968ef8c3ebb85844 - Init COMPLETE
+ip-26-0-165-202:153568:154056 [6] NCCL INFO comm 0xa7ee300 rank 30 nranks 32 cudaDev 6 nvmlDev 6 busId b9000 commId 0x968ef8c3ebb85844 - Init COMPLETE
+ip-26-0-165-202:153562:154061 [0] NCCL INFO comm 0xaa68a60 rank 24 nranks 32 cudaDev 0 nvmlDev 0 busId 53000 commId 0x968ef8c3ebb85844 - Init COMPLETE
+ip-26-0-165-202:153564:154057 [2] NCCL INFO comm 0x9d98af0 rank 26 nranks 32 cudaDev 2 nvmlDev 2 busId 75000 commId 0x968ef8c3ebb85844 - Init COMPLETE
+ip-26-0-165-164:1227467:1227960 [4] NCCL INFO Connected NVLS tree
+ip-26-0-165-164:1227467:1227960 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-164:1227467:1227960 [4] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-165-164:1227467:1227960 [4] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+ip-26-0-165-202:153569:154091 [7] NCCL INFO Using network Libfabric
+ip-26-0-161-123:32711:33234 [7] NCCL INFO Using network Libfabric
+ip-26-0-161-123:32707:33244 [3] NCCL INFO Using network Libfabric
+ip-26-0-161-123:32709:33243 [5] NCCL INFO Using network Libfabric
+ip-26-0-161-123:32710:33246 [6] NCCL INFO Using network Libfabric
+ip-26-0-161-123:32705:33248 [1] NCCL INFO Using network Libfabric
+ip-26-0-165-202:153564:154092 [2] NCCL INFO Using network Libfabric
+ip-26-0-161-123:32704:33247 [0] NCCL INFO Using network Libfabric
+ip-26-0-165-202:153567:154093 [5] NCCL INFO Using network Libfabric
+ip-26-0-165-202:153568:154096 [6] NCCL INFO Using network Libfabric
+ip-26-0-165-202:153562:154095 [0] NCCL INFO Using network Libfabric
+ip-26-0-165-202:153565:154094 [3] NCCL INFO Using network Libfabric
+ip-26-0-161-123:32711:33234 [7] NCCL INFO comm 0xb634450 rank 0 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0x6c1953a2a95d57e1 - Init START
+ip-26-0-165-202:153569:154091 [7] NCCL INFO comm 0xa751160 rank 1 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0x6c1953a2a95d57e1 - Init START
+ip-26-0-165-202:153563:154098 [1] NCCL INFO Using network Libfabric
+ip-26-0-165-202:153566:154097 [4] NCCL INFO Using network Libfabric
+ip-26-0-161-123:32711:33234 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-202:153569:154091 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-123:32706:33245 [2] NCCL INFO Using network Libfabric
+ip-26-0-161-123:32708:33242 [4] NCCL INFO Using network Libfabric
+ip-26-0-161-103:96890:97386 [2] NCCL INFO comm 0xa5cc060 rank 18 nranks 32 cudaDev 2 nvmlDev 2 busId 75000 commId 0x6ad24daef1fe15ce - Init COMPLETE
+ip-26-0-161-103:96894:97381 [6] NCCL INFO comm 0xa2306a0 rank 22 nranks 32 cudaDev 6 nvmlDev 6 busId b9000 commId 0x6ad24daef1fe15ce - Init COMPLETE
+ip-26-0-161-103:96892:97383 [4] NCCL INFO comm 0xb351080 rank 20 nranks 32 cudaDev 4 nvmlDev 4 busId 97000 commId 0x6ad24daef1fe15ce - Init COMPLETE
+ip-26-0-165-202:153567:154093 [5] NCCL INFO comm 0xb79ef70 rank 1 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0xa57a3be018a8b943 - Init START
+ip-26-0-161-123:32709:33243 [5] NCCL INFO comm 0xc4431c0 rank 0 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0xa57a3be018a8b943 - Init START
+ip-26-0-161-103:96893:97384 [5] NCCL INFO comm 0x9e9dd50 rank 21 nranks 32 cudaDev 5 nvmlDev 5 busId a8000 commId 0x6ad24daef1fe15ce - Init COMPLETE
+ip-26-0-161-103:96888:97387 [0] NCCL INFO comm 0xa5f0fe0 rank 16 nranks 32 cudaDev 0 nvmlDev 0 busId 53000 commId 0x6ad24daef1fe15ce - Init COMPLETE
+ip-26-0-161-103:96895:97382 [7] NCCL INFO comm 0xbbd4280 rank 23 nranks 32 cudaDev 7 nvmlDev 7 busId ca000 commId 0x6ad24daef1fe15ce - Init COMPLETE
+ip-26-0-161-103:96891:97385 [3] NCCL INFO comm 0xae813d0 rank 19 nranks 32 cudaDev 3 nvmlDev 3 busId 86000 commId 0x6ad24daef1fe15ce - Init COMPLETE
+ip-26-0-165-202:153568:154096 [6] NCCL INFO comm 0x9d87780 rank 1 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0x97bd314d7da0b64d - Init START
+ip-26-0-165-202:153562:154095 [0] NCCL INFO comm 0xb4b83a0 rank 1 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0x53b316fd2156f020 - Init START
+ip-26-0-161-103:96889:97388 [1] NCCL INFO comm 0x9f87fe0 rank 17 nranks 32 cudaDev 1 nvmlDev 1 busId 64000 commId 0x6ad24daef1fe15ce - Init COMPLETE
+ip-26-0-161-123:32710:33246 [6] NCCL INFO comm 0xbb21a40 rank 0 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0x97bd314d7da0b64d - Init START
+ip-26-0-161-123:32707:33244 [3] NCCL INFO comm 0xaa8f9e0 rank 0 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0x65195ce4f2c54799 - Init START
+ip-26-0-161-123:32704:33247 [0] NCCL INFO comm 0xc65f8b0 rank 0 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0x53b316fd2156f020 - Init START
+ip-26-0-165-202:153565:154094 [3] NCCL INFO comm 0xaf709b0 rank 1 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0x65195ce4f2c54799 - Init START
+ip-26-0-161-123:32709:33243 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-202:153567:154093 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-123:32710:33246 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-202:153568:154096 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-202:153563:154098 [1] NCCL INFO comm 0xc3f2ac0 rank 1 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0xe9de758c5f73d128 - Init START
+ip-26-0-161-123:32704:33247 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-123:32707:33244 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-202:153562:154095 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-202:153564:154092 [2] NCCL INFO comm 0xa7fefd0 rank 1 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0x5e0b17ed00861d01 - Init START
+ip-26-0-161-123:32705:33248 [1] NCCL INFO comm 0xbbec3d0 rank 0 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0xe9de758c5f73d128 - Init START
+ip-26-0-161-123:32706:33245 [2] NCCL INFO comm 0xb1c7ca0 rank 0 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0x5e0b17ed00861d01 - Init START
+ip-26-0-165-202:153566:154097 [4] NCCL INFO comm 0x9fe3290 rank 1 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0x5579512d5d31a4dd - Init START
+ip-26-0-161-123:32708:33242 [4] NCCL INFO comm 0xc480d20 rank 0 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0x5579512d5d31a4dd - Init START
+ip-26-0-161-123:32706:33245 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-202:153566:154097 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-123:32705:33248 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-123:32708:33242 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-202:153564:154092 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-202:153563:154098 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-202:153565:154094 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-178:2815963:2816477 [0] NCCL INFO Connected NVLS tree
+ip-26-0-161-178:2815963:2816477 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-178:2815963:2816477 [0] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-161-178:2815963:2816477 [0] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-165-164:1227470:1227958 [7] NCCL INFO Connected NVLS tree
+ip-26-0-165-164:1227470:1227958 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-164:1227470:1227958 [7] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-165-164:1227470:1227958 [7] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-178:2815967:2816473 [4] NCCL INFO Connected NVLS tree
+ip-26-0-161-178:2815967:2816473 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-178:2815967:2816473 [4] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-161-178:2815967:2816473 [4] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-165-164:1227468:1227959 [5] NCCL INFO Connected NVLS tree
+ip-26-0-165-164:1227468:1227959 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-164:1227468:1227959 [5] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-165-164:1227468:1227959 [5] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-165-164:1227464:1227963 [1] NCCL INFO Connected NVLS tree
+ip-26-0-165-164:1227464:1227963 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-164:1227464:1227963 [1] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-165-164:1227464:1227963 [1] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-165-131:1630208:1630729 [3] NCCL INFO Connected NVLS tree
+ip-26-0-165-131:1630208:1630729 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-131:1630208:1630729 [3] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-165-131:1630208:1630729 [3] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-165-131:1630210:1630732 [5] NCCL INFO Connected NVLS tree
+ip-26-0-165-131:1630210:1630732 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-131:1630210:1630732 [5] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-165-131:1630210:1630732 [5] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-165-131:1630212:1630734 [7] NCCL INFO Connected NVLS tree
+ip-26-0-165-131:1630212:1630734 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-131:1630212:1630734 [7] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-165-131:1630212:1630734 [7] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-165-131:1630209:1630730 [4] NCCL INFO Connected NVLS tree
+ip-26-0-165-131:1630209:1630730 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-131:1630209:1630730 [4] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-165-131:1630209:1630730 [4] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-160-242:1364904:1365408 [2] NCCL INFO comm 0xb849c50 rank 10 nranks 32 cudaDev 2 nvmlDev 2 busId 75000 commId 0x6ad24daef1fe15ce - Init COMPLETE
+ip-26-0-160-242:1364908:1365406 [6] NCCL INFO comm 0xbb61720 rank 14 nranks 32 cudaDev 6 nvmlDev 6 busId b9000 commId 0x6ad24daef1fe15ce - Init COMPLETE
+ip-26-0-160-242:1364906:1365407 [4] NCCL INFO comm 0xb742ff0 rank 12 nranks 32 cudaDev 4 nvmlDev 4 busId 97000 commId 0x6ad24daef1fe15ce - Init COMPLETE
+ip-26-0-160-242:1364902:1365410 [0] NCCL INFO comm 0xaf66250 rank 8 nranks 32 cudaDev 0 nvmlDev 0 busId 53000 commId 0x6ad24daef1fe15ce - Init COMPLETE
+ip-26-0-160-242:1364907:1365403 [5] NCCL INFO comm 0x9e43b50 rank 13 nranks 32 cudaDev 5 nvmlDev 5 busId a8000 commId 0x6ad24daef1fe15ce - Init COMPLETE
+ip-26-0-160-242:1364909:1365405 [7] NCCL INFO comm 0xb61ae00 rank 15 nranks 32 cudaDev 7 nvmlDev 7 busId ca000 commId 0x6ad24daef1fe15ce - Init COMPLETE
+ip-26-0-160-242:1364903:1365409 [1] NCCL INFO comm 0xbcea600 rank 9 nranks 32 cudaDev 1 nvmlDev 1 busId 64000 commId 0x6ad24daef1fe15ce - Init COMPLETE
+ip-26-0-160-242:1364905:1365404 [3] NCCL INFO comm 0xafe0340 rank 11 nranks 32 cudaDev 3 nvmlDev 3 busId 86000 commId 0x6ad24daef1fe15ce - Init COMPLETE
+ip-26-0-161-78:114518:115012 [6] NCCL INFO Connected NVLS tree
+ip-26-0-161-78:114518:115012 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-78:114518:115012 [6] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-161-78:114518:115012 [6] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-165-131:1630206:1630728 [1] NCCL INFO Connected NVLS tree
+ip-26-0-165-131:1630206:1630728 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-131:1630206:1630728 [1] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-165-131:1630206:1630728 [1] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-178:2815965:2816475 [2] NCCL INFO Connected NVLS tree
+ip-26-0-161-178:2815965:2816475 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-178:2815965:2816475 [2] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-161-178:2815965:2816475 [2] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-165-164:1227469:1227957 [6] NCCL INFO Connected NVLS tree
+ip-26-0-165-164:1227469:1227957 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-164:1227469:1227957 [6] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-165-164:1227469:1227957 [6] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-78:114512:115007 [0] NCCL INFO Connected NVLS tree
+ip-26-0-161-78:114512:115007 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-78:114512:115007 [0] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-161-78:114512:115007 [0] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-165-164:1227465:1227962 [2] NCCL INFO Connected NVLS tree
+ip-26-0-165-164:1227465:1227962 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-164:1227465:1227962 [2] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-165-164:1227465:1227962 [2] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-78:114514:115011 [2] NCCL INFO Connected NVLS tree
+ip-26-0-161-78:114514:115011 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-78:114514:115011 [2] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-161-78:114514:115011 [2] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-178:2815970:2816470 [7] NCCL INFO Connected NVLS tree
+ip-26-0-161-178:2815970:2816470 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-178:2815970:2816470 [7] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-161-178:2815970:2816470 [7] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-165-131:1630207:1630731 [2] NCCL INFO Connected NVLS tree
+ip-26-0-165-131:1630207:1630731 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-131:1630207:1630731 [2] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-165-131:1630207:1630731 [2] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-78:114519:115014 [7] NCCL INFO comm 0xa5040c0 rank 7 nranks 32 cudaDev 7 nvmlDev 7 busId ca000 commId 0x968ef8c3ebb85844 - Init COMPLETE
+ip-26-0-161-78:114517:115013 [5] NCCL INFO comm 0xb85b760 rank 5 nranks 32 cudaDev 5 nvmlDev 5 busId a8000 commId 0x968ef8c3ebb85844 - Init COMPLETE
+ip-26-0-161-78:114513:115008 [1] NCCL INFO comm 0xb40d9a0 rank 1 nranks 32 cudaDev 1 nvmlDev 1 busId 64000 commId 0x968ef8c3ebb85844 - Init COMPLETE
+ip-26-0-161-78:114512:115007 [0] NCCL INFO comm 0xa85e740 rank 0 nranks 32 cudaDev 0 nvmlDev 0 busId 53000 commId 0x968ef8c3ebb85844 - Init COMPLETE
+ip-26-0-161-78:114516:115010 [4] NCCL INFO comm 0xb3ab6d0 rank 4 nranks 32 cudaDev 4 nvmlDev 4 busId 97000 commId 0x968ef8c3ebb85844 - Init COMPLETE
+ip-26-0-161-78:114518:115012 [6] NCCL INFO comm 0xb8bc080 rank 6 nranks 32 cudaDev 6 nvmlDev 6 busId b9000 commId 0x968ef8c3ebb85844 - Init COMPLETE
+ip-26-0-161-78:114515:115009 [3] NCCL INFO comm 0xa3a81c0 rank 3 nranks 32 cudaDev 3 nvmlDev 3 busId 86000 commId 0x968ef8c3ebb85844 - Init COMPLETE
+ip-26-0-161-78:114514:115011 [2] NCCL INFO comm 0xb2d0ac0 rank 2 nranks 32 cudaDev 2 nvmlDev 2 busId 75000 commId 0x968ef8c3ebb85844 - Init COMPLETE
+ip-26-0-161-178:2815969:2816471 [6] NCCL INFO Connected NVLS tree
+ip-26-0-161-178:2815969:2816471 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-178:2815969:2816471 [6] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-161-178:2815969:2816471 [6] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+ip-26-0-160-225:124160:124698 [0] NCCL INFO Using network Libfabric
+ip-26-0-160-225:124164:124702 [4] NCCL INFO Using network Libfabric
+ip-26-0-161-78:114512:115045 [0] NCCL INFO Using network Libfabric
+ip-26-0-160-225:124161:124703 [1] NCCL INFO Using network Libfabric
+ip-26-0-161-78:114519:115047 [7] NCCL INFO Using network Libfabric
+ip-26-0-160-225:124165:124699 [5] NCCL INFO Using network Libfabric
+ip-26-0-161-78:114516:115044 [4] NCCL INFO Using network Libfabric
+ip-26-0-161-78:114517:115046 [5] NCCL INFO Using network Libfabric
+ip-26-0-160-225:124162:124705 [2] NCCL INFO Using network Libfabric
+ip-26-0-160-225:124163:124706 [3] NCCL INFO Using network Libfabric
+ip-26-0-161-78:114513:115048 [1] NCCL INFO Using network Libfabric
+ip-26-0-160-225:124167:124700 [7] NCCL INFO Using network Libfabric
+ip-26-0-161-78:114514:115049 [2] NCCL INFO Using network Libfabric
+ip-26-0-165-131:1630211:1630733 [6] NCCL INFO Connected NVLS tree
+ip-26-0-165-131:1630211:1630733 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-131:1630211:1630733 [6] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-165-131:1630211:1630733 [6] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-160-225:124160:124698 [0] NCCL INFO comm 0xa9dbd90 rank 0 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0x145a7e479ca528e9 - Init START
+NCCL version 2.18.5+cuda12.2
+ip-26-0-161-78:114512:115045 [0] NCCL INFO comm 0xb2c2490 rank 1 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0x145a7e479ca528e9 - Init START
+ip-26-0-160-225:124160:124698 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/541
+ip-26-0-161-78:114515:115050 [3] NCCL INFO Using network Libfabric
+ip-26-0-160-225:124164:124702 [4] NCCL INFO comm 0xb185300 rank 0 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0x72d1a31dbaf1f836 - Init START
+ip-26-0-161-78:114516:115044 [4] NCCL INFO comm 0xa945970 rank 1 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0x72d1a31dbaf1f836 - Init START
+ip-26-0-160-225:124165:124699 [5] NCCL INFO comm 0xaa36cd0 rank 0 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0x70adc5cb7dc315cd - Init START
+ip-26-0-160-225:124164:124702 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-78:114517:115046 [5] NCCL INFO comm 0xc2aa9f0 rank 1 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0x70adc5cb7dc315cd - Init START
+ip-26-0-161-78:114512:115045 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-160-225:124161:124703 [1] NCCL INFO comm 0xaa12d50 rank 0 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0x55893f50a8f6dd29 - Init START
+ip-26-0-160-225:124165:124699 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-78:114513:115048 [1] NCCL INFO comm 0xbe5c7e0 rank 1 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0x55893f50a8f6dd29 - Init START
+ip-26-0-161-78:114516:115044 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-78:114517:115046 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-160-225:124161:124703 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-160-225:124167:124700 [7] NCCL INFO comm 0xbe456f0 rank 0 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0x13466d39857b650f - Init START
+ip-26-0-161-78:114519:115047 [7] NCCL INFO comm 0x9a995d0 rank 1 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0x13466d39857b650f - Init START
+ip-26-0-160-225:124162:124705 [2] NCCL INFO comm 0xc4b2dd0 rank 0 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0x6160e37a54249843 - Init START
+ip-26-0-161-78:114514:115049 [2] NCCL INFO comm 0xbd1fab0 rank 1 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0x6160e37a54249843 - Init START
+ip-26-0-160-225:124167:124700 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-160-225:124162:124705 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-78:114513:115048 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-160-225:124163:124706 [3] NCCL INFO comm 0xac7b900 rank 0 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0x4177b0a43e9f62e1 - Init START
+ip-26-0-161-78:114519:115047 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-78:114515:115050 [3] NCCL INFO comm 0xadf7330 rank 1 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0x4177b0a43e9f62e1 - Init START
+ip-26-0-161-78:114514:115049 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-160-225:124163:124706 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-160-225:124166:124708 [6] NCCL INFO Using network Libfabric
+ip-26-0-161-78:114515:115050 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-78:114518:115051 [6] NCCL INFO Using network Libfabric
+ip-26-0-161-78:114518:115051 [6] NCCL INFO comm 0xae517a0 rank 1 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0x85ddc6e85d763ff9 - Init START
+ip-26-0-160-225:124166:124708 [6] NCCL INFO comm 0xb95b600 rank 0 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0x85ddc6e85d763ff9 - Init START
+ip-26-0-160-225:124166:124708 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-78:114518:115051 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-164:1227463:1227964 [0] NCCL INFO Connected NVLS tree
+ip-26-0-165-164:1227463:1227964 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-164:1227463:1227964 [0] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-165-164:1227463:1227964 [0] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-165-131:1630205:1630727 [0] NCCL INFO Connected NVLS tree
+ip-26-0-165-131:1630205:1630727 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-131:1630205:1630727 [0] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-165-131:1630205:1630727 [0] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-153:2636088:2636601 [4] NCCL INFO Connected NVLS tree
+ip-26-0-161-153:2636088:2636601 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-153:2636088:2636601 [4] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-161-153:2636088:2636601 [4] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-165-164:1227464:1227963 [1] NCCL INFO comm 0xb0a70f0 rank 17 nranks 32 cudaDev 1 nvmlDev 1 busId 64000 commId 0x968ef8c3ebb85844 - Init COMPLETE
+ip-26-0-165-164:1227465:1227962 [2] NCCL INFO comm 0xac59150 rank 18 nranks 32 cudaDev 2 nvmlDev 2 busId 75000 commId 0x968ef8c3ebb85844 - Init COMPLETE
+ip-26-0-165-164:1227466:1227961 [3] NCCL INFO comm 0xaebd080 rank 19 nranks 32 cudaDev 3 nvmlDev 3 busId 86000 commId 0x968ef8c3ebb85844 - Init COMPLETE
+ip-26-0-165-164:1227469:1227957 [6] NCCL INFO comm 0x9a95350 rank 22 nranks 32 cudaDev 6 nvmlDev 6 busId b9000 commId 0x968ef8c3ebb85844 - Init COMPLETE
+ip-26-0-165-164:1227463:1227964 [0] NCCL INFO comm 0xa7d40d0 rank 16 nranks 32 cudaDev 0 nvmlDev 0 busId 53000 commId 0x968ef8c3ebb85844 - Init COMPLETE
+ip-26-0-165-164:1227467:1227960 [4] NCCL INFO comm 0xa7d56c0 rank 20 nranks 32 cudaDev 4 nvmlDev 4 busId 97000 commId 0x968ef8c3ebb85844 - Init COMPLETE
+ip-26-0-165-164:1227470:1227958 [7] NCCL INFO comm 0xb2dbd00 rank 23 nranks 32 cudaDev 7 nvmlDev 7 busId ca000 commId 0x968ef8c3ebb85844 - Init COMPLETE
+ip-26-0-165-164:1227468:1227959 [5] NCCL INFO comm 0xb1614c0 rank 21 nranks 32 cudaDev 5 nvmlDev 5 busId a8000 commId 0x968ef8c3ebb85844 - Init COMPLETE
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+ip-26-0-165-164:1227463:1227994 [0] NCCL INFO Using network Libfabric
+ip-26-0-161-103:96894:97424 [6] NCCL INFO Using network Libfabric
+ip-26-0-165-164:1227469:1227995 [6] NCCL INFO Using network Libfabric
+ip-26-0-161-103:96889:97426 [1] NCCL INFO Using network Libfabric
+ip-26-0-161-103:96888:97423 [0] NCCL INFO Using network Libfabric
+ip-26-0-161-103:96890:97425 [2] NCCL INFO Using network Libfabric
+ip-26-0-161-103:96895:97427 [7] NCCL INFO Using network Libfabric
+ip-26-0-165-164:1227465:1227996 [2] NCCL INFO Using network Libfabric
+NCCL version 2.18.5+cuda12.2
+ip-26-0-165-164:1227464:1227997 [1] NCCL INFO Using network Libfabric
+ip-26-0-161-103:96892:97429 [4] NCCL INFO Using network Libfabric
+ip-26-0-165-164:1227470:1227998 [7] NCCL INFO Using network Libfabric
+ip-26-0-165-164:1227467:1227999 [4] NCCL INFO Using network Libfabric
+ip-26-0-165-164:1227463:1227994 [0] NCCL INFO comm 0x9d739e0 rank 1 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0xa047ea5e0359805a - Init START
+ip-26-0-161-103:96888:97423 [0] NCCL INFO comm 0xb0551c0 rank 0 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0xa047ea5e0359805a - Init START
+ip-26-0-165-164:1227469:1227995 [6] NCCL INFO comm 0xa4e4dc0 rank 1 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0x1d99a63c76dbb2eb - Init START
+ip-26-0-161-103:96894:97424 [6] NCCL INFO comm 0xac7ec60 rank 0 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0x1d99a63c76dbb2eb - Init START
+ip-26-0-161-103:96888:97423 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-164:1227463:1227994 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-103:96894:97424 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-164:1227469:1227995 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-164:1227465:1227996 [2] NCCL INFO comm 0xa2001a0 rank 1 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0xbaa57eddcf04b52 - Init START
+ip-26-0-161-103:96890:97425 [2] NCCL INFO comm 0xb030390 rank 0 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0xbaa57eddcf04b52 - Init START
+ip-26-0-165-164:1227464:1227997 [1] NCCL INFO comm 0xa63c090 rank 1 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0xf09cef77bd870ed5 - Init START
+ip-26-0-161-103:96889:97426 [1] NCCL INFO comm 0xa9ec050 rank 0 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0xf09cef77bd870ed5 - Init START
+ip-26-0-161-103:96890:97425 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-164:1227465:1227996 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-103:96893:97431 [5] NCCL INFO Using network Libfabric
+ip-26-0-161-103:96889:97426 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-164:1227470:1227998 [7] NCCL INFO comm 0xa8703f0 rank 1 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0xfc4908b6009a7d47 - Init START
+ip-26-0-161-103:96895:97427 [7] NCCL INFO comm 0xc6396c0 rank 0 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0xfc4908b6009a7d47 - Init START
+ip-26-0-161-142:62885:63377 [4] NCCL INFO Connected NVLS tree
+ip-26-0-165-164:1227464:1227997 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-142:62885:63377 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-103:96895:97427 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-142:62885:63377 [4] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-161-142:62885:63377 [4] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-165-164:1227470:1227998 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-164:1227467:1227999 [4] NCCL INFO comm 0xb225480 rank 1 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0xc0254c28e4109f52 - Init START
+ip-26-0-161-103:96892:97429 [4] NCCL INFO comm 0xbdb5c00 rank 0 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0xc0254c28e4109f52 - Init START
+ip-26-0-165-164:1227468:1228000 [5] NCCL INFO Using network Libfabric
+ip-26-0-161-103:96892:97429 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-164:1227467:1227999 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+NCCL version 2.18.5+cuda12.2
+ip-26-0-165-164:1227468:1228000 [5] NCCL INFO comm 0xbbc7e00 rank 1 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0xf57804cee76ef536 - Init START
+ip-26-0-161-103:96893:97431 [5] NCCL INFO comm 0xa902d80 rank 0 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0xf57804cee76ef536 - Init START
+ip-26-0-161-103:96893:97431 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-164:1227468:1228000 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-153:2636086:2636602 [2] NCCL INFO Connected NVLS tree
+ip-26-0-161-153:2636086:2636602 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-153:2636086:2636602 [2] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-161-153:2636086:2636602 [2] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-103:96891:97433 [3] NCCL INFO Using network Libfabric
+ip-26-0-165-164:1227466:1228001 [3] NCCL INFO Using network Libfabric
+ip-26-0-165-164:1227466:1228001 [3] NCCL INFO comm 0xa455b10 rank 1 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0x190cbf84dbe1f009 - Init START
+ip-26-0-161-103:96891:97433 [3] NCCL INFO comm 0xa41d6d0 rank 0 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0x190cbf84dbe1f009 - Init START
+ip-26-0-161-103:96891:97433 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-164:1227466:1228001 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-142:62883:63373 [2] NCCL INFO Connected NVLS tree
+ip-26-0-161-142:62883:63373 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-142:62887:63379 [6] NCCL INFO Connected NVLS tree
+ip-26-0-161-142:62883:63373 [2] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-161-142:62883:63373 [2] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-142:62887:63379 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-142:62887:63379 [6] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-161-142:62887:63379 [6] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-168-52:2722047:2722542 [2] NCCL INFO Connected NVLS tree
+ip-26-0-168-52:2722047:2722542 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-52:2722047:2722542 [2] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-168-52:2722047:2722542 [2] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-168-52:2722051:2722543 [6] NCCL INFO Connected NVLS tree
+ip-26-0-168-52:2722051:2722543 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-52:2722051:2722543 [6] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-168-52:2722051:2722543 [6] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-178:2815964:2816476 [1] NCCL INFO Connected NVLS tree
+ip-26-0-161-178:2815964:2816476 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-178:2815964:2816476 [1] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-161-178:2815964:2816476 [1] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-165-131:1630205:1630727 [0] NCCL INFO comm 0xaeb3380 rank 8 nranks 32 cudaDev 0 nvmlDev 0 busId 53000 commId 0x968ef8c3ebb85844 - Init COMPLETE
+ip-26-0-165-131:1630207:1630731 [2] NCCL INFO comm 0xab227f0 rank 10 nranks 32 cudaDev 2 nvmlDev 2 busId 75000 commId 0x968ef8c3ebb85844 - Init COMPLETE
+ip-26-0-165-131:1630209:1630730 [4] NCCL INFO comm 0xb1f6980 rank 12 nranks 32 cudaDev 4 nvmlDev 4 busId 97000 commId 0x968ef8c3ebb85844 - Init COMPLETE
+ip-26-0-165-131:1630211:1630733 [6] NCCL INFO comm 0xba47990 rank 14 nranks 32 cudaDev 6 nvmlDev 6 busId b9000 commId 0x968ef8c3ebb85844 - Init COMPLETE
+ip-26-0-168-52:2722049:2722547 [4] NCCL INFO Connected NVLS tree
+ip-26-0-168-52:2722049:2722547 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-52:2722049:2722547 [4] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-168-52:2722049:2722547 [4] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-165-131:1630206:1630728 [1] NCCL INFO comm 0xbb76e40 rank 9 nranks 32 cudaDev 1 nvmlDev 1 busId 64000 commId 0x968ef8c3ebb85844 - Init COMPLETE
+ip-26-0-165-131:1630210:1630732 [5] NCCL INFO comm 0xb46d6d0 rank 13 nranks 32 cudaDev 5 nvmlDev 5 busId a8000 commId 0x968ef8c3ebb85844 - Init COMPLETE
+ip-26-0-165-131:1630208:1630729 [3] NCCL INFO comm 0x9aa21d0 rank 11 nranks 32 cudaDev 3 nvmlDev 3 busId 86000 commId 0x968ef8c3ebb85844 - Init COMPLETE
+ip-26-0-165-131:1630212:1630734 [7] NCCL INFO comm 0x9c95060 rank 15 nranks 32 cudaDev 7 nvmlDev 7 busId ca000 commId 0x968ef8c3ebb85844 - Init COMPLETE
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+ip-26-0-160-242:1364908:1365452 [6] NCCL INFO Using network Libfabric
+ip-26-0-160-242:1364902:1365449 [0] NCCL INFO Using network Libfabric
+ip-26-0-160-242:1364904:1365451 [2] NCCL INFO Using network Libfabric
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+ip-26-0-160-242:1364906:1365453 [4] NCCL INFO Using network Libfabric
+NCCL version 2.18.5+cuda12.2
+ip-26-0-165-131:1630205:1630778 [0] NCCL INFO Using network Libfabric
+ip-26-0-165-131:1630211:1630779 [6] NCCL INFO Using network Libfabric
+ip-26-0-165-131:1630207:1630780 [2] NCCL INFO Using network Libfabric
+ip-26-0-165-131:1630209:1630781 [4] NCCL INFO Using network Libfabric
+ip-26-0-165-131:1630205:1630778 [0] NCCL INFO comm 0xa447920 rank 1 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0x708e697530715efe - Init START
+ip-26-0-160-242:1364905:1365456 [3] NCCL INFO Using network Libfabric
+ip-26-0-165-131:1630211:1630779 [6] NCCL INFO comm 0xc497e20 rank 1 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0x69dd6b894eb2fbd2 - Init START
+ip-26-0-160-242:1364902:1365449 [0] NCCL INFO comm 0xb9b3db0 rank 0 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0x708e697530715efe - Init START
+ip-26-0-160-242:1364908:1365452 [6] NCCL INFO comm 0xc5af960 rank 0 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0x69dd6b894eb2fbd2 - Init START
+ip-26-0-160-242:1364902:1365449 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-160-242:1364908:1365452 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-160-242:1364907:1365459 [5] NCCL INFO Using network Libfabric
+ip-26-0-165-131:1630205:1630778 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/409
+ip-26-0-160-242:1364909:1365460 [7] NCCL INFO Using network Libfabric
+ip-26-0-160-242:1364903:1365461 [1] NCCL INFO Using network Libfabric
+ip-26-0-165-131:1630211:1630779 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/409
+ip-26-0-160-242:1364904:1365451 [2] NCCL INFO comm 0xadf27a0 rank 0 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0x4a887d18c7bdcfe2 - Init START
+ip-26-0-165-131:1630207:1630780 [2] NCCL INFO comm 0xb5730b0 rank 1 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0x4a887d18c7bdcfe2 - Init START
+ip-26-0-160-242:1364906:1365453 [4] NCCL INFO comm 0xc194b30 rank 0 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0xb20b05f5e3ea6185 - Init START
+ip-26-0-165-131:1630209:1630781 [4] NCCL INFO comm 0xa78aba0 rank 1 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0xb20b05f5e3ea6185 - Init START
+ip-26-0-165-131:1630210:1630783 [5] NCCL INFO Using network Libfabric
+ip-26-0-165-131:1630206:1630784 [1] NCCL INFO Using network Libfabric
+ip-26-0-160-242:1364904:1365451 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-131:1630212:1630785 [7] NCCL INFO Using network Libfabric
+ip-26-0-161-153:2636090:2636598 [6] NCCL INFO Connected NVLS tree
+ip-26-0-161-153:2636090:2636598 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-153:2636090:2636598 [6] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-161-142:62881:63375 [0] NCCL INFO Connected NVLS tree
+ip-26-0-160-242:1364906:1365453 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-131:1630207:1630780 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/409
+ip-26-0-161-153:2636090:2636598 [6] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-153:2636084:2636605 [0] NCCL INFO Connected NVLS tree
+ip-26-0-161-142:62881:63375 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-142:62881:63375 [0] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-161-142:62881:63375 [0] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-153:2636084:2636605 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-153:2636084:2636605 [0] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-161-153:2636084:2636605 [0] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-165-131:1630209:1630781 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/409
+ip-26-0-165-131:1630208:1630782 [3] NCCL INFO Using network Libfabric
+ip-26-0-165-131:1630210:1630783 [5] NCCL INFO comm 0xbebd4f0 rank 1 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0xd8b0494bfaa0f0bf - Init START
+ip-26-0-165-131:1630206:1630784 [1] NCCL INFO comm 0xb10bcb0 rank 1 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0x173c75bc0100d108 - Init START
+ip-26-0-160-242:1364907:1365459 [5] NCCL INFO comm 0xa8a8580 rank 0 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0xd8b0494bfaa0f0bf - Init START
+ip-26-0-165-131:1630212:1630785 [7] NCCL INFO comm 0xa6e5da0 rank 1 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0x370c64acc5849eec - Init START
+ip-26-0-160-242:1364903:1365461 [1] NCCL INFO comm 0xb285650 rank 0 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0x173c75bc0100d108 - Init START
+ip-26-0-160-242:1364909:1365460 [7] NCCL INFO comm 0xc07f290 rank 0 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0x370c64acc5849eec - Init START
+ip-26-0-160-242:1364903:1365461 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-131:1630210:1630783 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/409
+ip-26-0-165-131:1630206:1630784 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/409
+ip-26-0-160-242:1364907:1365459 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-160-242:1364909:1365460 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-131:1630212:1630785 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/409
+ip-26-0-165-131:1630208:1630782 [3] NCCL INFO comm 0xa509410 rank 1 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0xaf1c7fecc0b3b264 - Init START
+ip-26-0-160-242:1364905:1365456 [3] NCCL INFO comm 0xba2e260 rank 0 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0xaf1c7fecc0b3b264 - Init START
+ip-26-0-160-242:1364905:1365456 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-131:1630208:1630782 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/409
+ip-26-0-161-142:62888:63380 [7] NCCL INFO Connected NVLS tree
+ip-26-0-161-142:62888:63380 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-142:62888:63380 [7] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-161-142:62888:63380 [7] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-168-52:2722045:2722544 [0] NCCL INFO Connected NVLS tree
+ip-26-0-168-52:2722045:2722544 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-52:2722045:2722544 [0] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-168-52:2722045:2722544 [0] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-153:2636091:2636599 [7] NCCL INFO Connected NVLS tree
+ip-26-0-161-153:2636091:2636599 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-153:2636091:2636599 [7] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-161-153:2636091:2636599 [7] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-142:62882:63374 [1] NCCL INFO Connected NVLS tree
+ip-26-0-161-142:62882:63374 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-142:62882:63374 [1] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-161-142:62882:63374 [1] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-178:2815966:2816474 [3] NCCL INFO Connected NVLS tree
+ip-26-0-161-178:2815966:2816474 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-178:2815966:2816474 [3] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-161-178:2815966:2816474 [3] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-153:2636085:2636604 [1] NCCL INFO Connected NVLS tree
+ip-26-0-161-153:2636085:2636604 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-153:2636085:2636604 [1] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-161-153:2636085:2636604 [1] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-178:2815968:2816472 [5] NCCL INFO Connected NVLS tree
+ip-26-0-161-178:2815968:2816472 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-178:2815968:2816472 [5] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-161-178:2815968:2816472 [5] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-168-30:235652:236147 [4] NCCL INFO Connected NVLS tree
+ip-26-0-161-142:62884:63376 [3] NCCL INFO Connected NVLS tree
+ip-26-0-161-142:62884:63376 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-30:235652:236147 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-30:235652:236147 [4] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-168-30:235652:236147 [4] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-142:62884:63376 [3] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-161-142:62884:63376 [3] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-168-52:2722052:2722546 [7] NCCL INFO Connected NVLS tree
+ip-26-0-168-52:2722052:2722546 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-52:2722052:2722546 [7] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-168-52:2722052:2722546 [7] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-142:62886:63378 [5] NCCL INFO Connected NVLS tree
+ip-26-0-161-142:62886:63378 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-142:62886:63378 [5] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-161-142:62886:63378 [5] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-168-30:235648:236145 [0] NCCL INFO Connected NVLS tree
+ip-26-0-168-30:235648:236145 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-30:235648:236145 [0] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-168-30:235648:236145 [0] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-153:2636087:2636603 [3] NCCL INFO Connected NVLS tree
+ip-26-0-161-153:2636087:2636603 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-153:2636087:2636603 [3] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-161-153:2636087:2636603 [3] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-153:2636089:2636600 [5] NCCL INFO Connected NVLS tree
+ip-26-0-161-153:2636089:2636600 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-153:2636089:2636600 [5] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-161-153:2636089:2636600 [5] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-178:2815965:2816475 [2] NCCL INFO comm 0xa383240 rank 26 nranks 32 cudaDev 2 nvmlDev 2 busId 75000 commId 0xc9fdc2638087cf9c - Init COMPLETE
+ip-26-0-161-178:2815970:2816470 [7] NCCL INFO comm 0xa873aa0 rank 31 nranks 32 cudaDev 7 nvmlDev 7 busId ca000 commId 0xc9fdc2638087cf9c - Init COMPLETE
+ip-26-0-161-178:2815966:2816474 [3] NCCL INFO comm 0xac53fa0 rank 27 nranks 32 cudaDev 3 nvmlDev 3 busId 86000 commId 0xc9fdc2638087cf9c - Init COMPLETE
+ip-26-0-161-178:2815969:2816471 [6] NCCL INFO comm 0xb206db0 rank 30 nranks 32 cudaDev 6 nvmlDev 6 busId b9000 commId 0xc9fdc2638087cf9c - Init COMPLETE
+ip-26-0-161-178:2815967:2816473 [4] NCCL INFO comm 0xa848d80 rank 28 nranks 32 cudaDev 4 nvmlDev 4 busId 97000 commId 0xc9fdc2638087cf9c - Init COMPLETE
+ip-26-0-161-178:2815963:2816477 [0] NCCL INFO comm 0xab2e7e0 rank 24 nranks 32 cudaDev 0 nvmlDev 0 busId 53000 commId 0xc9fdc2638087cf9c - Init COMPLETE
+ip-26-0-161-178:2815968:2816472 [5] NCCL INFO comm 0xb83d840 rank 29 nranks 32 cudaDev 5 nvmlDev 5 busId a8000 commId 0xc9fdc2638087cf9c - Init COMPLETE
+ip-26-0-161-178:2815964:2816476 [1] NCCL INFO comm 0x9f47930 rank 25 nranks 32 cudaDev 1 nvmlDev 1 busId 64000 commId 0xc9fdc2638087cf9c - Init COMPLETE
+ip-26-0-161-142:62883:63373 [2] NCCL INFO comm 0xa2edd90 rank 10 nranks 32 cudaDev 2 nvmlDev 2 busId 75000 commId 0xc9fdc2638087cf9c - Init COMPLETE
+ip-26-0-161-142:62887:63379 [6] NCCL INFO comm 0x9e2e830 rank 14 nranks 32 cudaDev 6 nvmlDev 6 busId b9000 commId 0xc9fdc2638087cf9c - Init COMPLETE
+ip-26-0-161-142:62886:63378 [5] NCCL INFO comm 0xb59e210 rank 13 nranks 32 cudaDev 5 nvmlDev 5 busId a8000 commId 0xc9fdc2638087cf9c - Init COMPLETE
+ip-26-0-161-142:62881:63375 [0] NCCL INFO comm 0xb8c2040 rank 8 nranks 32 cudaDev 0 nvmlDev 0 busId 53000 commId 0xc9fdc2638087cf9c - Init COMPLETE
+ip-26-0-161-142:62888:63380 [7] NCCL INFO comm 0xa913bd0 rank 15 nranks 32 cudaDev 7 nvmlDev 7 busId ca000 commId 0xc9fdc2638087cf9c - Init COMPLETE
+ip-26-0-161-142:62882:63374 [1] NCCL INFO comm 0xa6869b0 rank 9 nranks 32 cudaDev 1 nvmlDev 1 busId 64000 commId 0xc9fdc2638087cf9c - Init COMPLETE
+ip-26-0-161-142:62884:63376 [3] NCCL INFO comm 0xad6a3d0 rank 11 nranks 32 cudaDev 3 nvmlDev 3 busId 86000 commId 0xc9fdc2638087cf9c - Init COMPLETE
+ip-26-0-161-142:62885:63377 [4] NCCL INFO comm 0xb5cbf30 rank 12 nranks 32 cudaDev 4 nvmlDev 4 busId 97000 commId 0xc9fdc2638087cf9c - Init COMPLETE
+ip-26-0-168-30:235655:236142 [7] NCCL INFO Connected NVLS tree
+ip-26-0-168-30:235655:236142 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-30:235655:236142 [7] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-168-30:235655:236142 [7] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-168-34:272154:272650 [4] NCCL INFO Connected NVLS tree
+ip-26-0-168-34:272154:272650 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-34:272154:272650 [4] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-168-34:272154:272650 [4] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-153:2636090:2636598 [6] NCCL INFO comm 0xb1c8e30 rank 22 nranks 32 cudaDev 6 nvmlDev 6 busId b9000 commId 0xc9fdc2638087cf9c - Init COMPLETE
+ip-26-0-161-153:2636086:2636602 [2] NCCL INFO comm 0xb110fc0 rank 18 nranks 32 cudaDev 2 nvmlDev 2 busId 75000 commId 0xc9fdc2638087cf9c - Init COMPLETE
+ip-26-0-161-153:2636089:2636600 [5] NCCL INFO comm 0x9edfc90 rank 21 nranks 32 cudaDev 5 nvmlDev 5 busId a8000 commId 0xc9fdc2638087cf9c - Init COMPLETE
+ip-26-0-161-153:2636085:2636604 [1] NCCL INFO comm 0xb985620 rank 17 nranks 32 cudaDev 1 nvmlDev 1 busId 64000 commId 0xc9fdc2638087cf9c - Init COMPLETE
+ip-26-0-161-153:2636084:2636605 [0] NCCL INFO comm 0xb1686c0 rank 16 nranks 32 cudaDev 0 nvmlDev 0 busId 53000 commId 0xc9fdc2638087cf9c - Init COMPLETE
+ip-26-0-161-153:2636088:2636601 [4] NCCL INFO comm 0xaa62b50 rank 20 nranks 32 cudaDev 4 nvmlDev 4 busId 97000 commId 0xc9fdc2638087cf9c - Init COMPLETE
+ip-26-0-161-153:2636091:2636599 [7] NCCL INFO comm 0xa7c32e0 rank 23 nranks 32 cudaDev 7 nvmlDev 7 busId ca000 commId 0xc9fdc2638087cf9c - Init COMPLETE
+ip-26-0-161-153:2636087:2636603 [3] NCCL INFO comm 0xa065880 rank 19 nranks 32 cudaDev 3 nvmlDev 3 busId 86000 commId 0xc9fdc2638087cf9c - Init COMPLETE
+ip-26-0-168-34:272150:272645 [0] NCCL INFO Connected NVLS tree
+ip-26-0-168-34:272150:272645 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-34:272150:272645 [0] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-168-34:272150:272645 [0] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-168-52:2722046:2722541 [1] NCCL INFO Connected NVLS tree
+ip-26-0-168-52:2722046:2722541 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-52:2722046:2722541 [1] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-168-52:2722046:2722541 [1] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-168-30:235654:236146 [6] NCCL INFO Connected NVLS tree
+ip-26-0-168-30:235654:236146 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-30:235654:236146 [6] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-168-30:235654:236146 [6] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-168-34:272156:272648 [6] NCCL INFO Connected NVLS tree
+ip-26-0-168-34:272156:272648 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-34:272156:272648 [6] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-168-34:272156:272648 [6] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-168-52:2722048:2722545 [3] NCCL INFO Connected NVLS tree
+ip-26-0-168-52:2722048:2722545 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-52:2722048:2722545 [3] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-168-52:2722048:2722545 [3] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-168-34:272152:272647 [2] NCCL INFO Connected NVLS tree
+ip-26-0-168-34:272152:272647 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-34:272152:272647 [2] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-168-34:272152:272647 [2] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-168-52:2722050:2722548 [5] NCCL INFO Connected NVLS tree
+ip-26-0-168-52:2722050:2722548 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-52:2722050:2722548 [5] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-168-52:2722050:2722548 [5] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-168-30:235650:236144 [2] NCCL INFO Connected NVLS tree
+ip-26-0-168-30:235650:236144 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-30:235650:236144 [2] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-168-30:235650:236144 [2] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-168-30:235649:236143 [1] NCCL INFO Connected NVLS tree
+ip-26-0-168-30:235649:236143 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-30:235649:236143 [1] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-168-30:235649:236143 [1] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-168-52:2722052:2722546 [7] NCCL INFO comm 0xb6b79f0 rank 31 nranks 32 cudaDev 7 nvmlDev 7 busId ca000 commId 0xe6d0f49db1af694c - Init COMPLETE
+ip-26-0-168-52:2722049:2722547 [4] NCCL INFO comm 0xa856380 rank 28 nranks 32 cudaDev 4 nvmlDev 4 busId 97000 commId 0xe6d0f49db1af694c - Init COMPLETE
+ip-26-0-168-52:2722051:2722543 [6] NCCL INFO comm 0xb298c00 rank 30 nranks 32 cudaDev 6 nvmlDev 6 busId b9000 commId 0xe6d0f49db1af694c - Init COMPLETE
+ip-26-0-168-52:2722047:2722542 [2] NCCL INFO comm 0xa9fbb80 rank 26 nranks 32 cudaDev 2 nvmlDev 2 busId 75000 commId 0xe6d0f49db1af694c - Init COMPLETE
+ip-26-0-168-52:2722046:2722541 [1] NCCL INFO comm 0xadda970 rank 25 nranks 32 cudaDev 1 nvmlDev 1 busId 64000 commId 0xe6d0f49db1af694c - Init COMPLETE
+ip-26-0-168-52:2722048:2722545 [3] NCCL INFO comm 0xa691040 rank 27 nranks 32 cudaDev 3 nvmlDev 3 busId 86000 commId 0xe6d0f49db1af694c - Init COMPLETE
+ip-26-0-168-52:2722050:2722548 [5] NCCL INFO comm 0xa6c0c80 rank 29 nranks 32 cudaDev 5 nvmlDev 5 busId a8000 commId 0xe6d0f49db1af694c - Init COMPLETE
+ip-26-0-168-52:2722045:2722544 [0] NCCL INFO comm 0x9b08d30 rank 24 nranks 32 cudaDev 0 nvmlDev 0 busId 53000 commId 0xe6d0f49db1af694c - Init COMPLETE
+ip-26-0-168-34:272151:272646 [1] NCCL INFO Connected NVLS tree
+ip-26-0-168-34:272151:272646 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-34:272151:272646 [1] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-168-34:272151:272646 [1] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-168-34:272157:272649 [7] NCCL INFO Connected NVLS tree
+ip-26-0-168-34:272157:272649 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-34:272157:272649 [7] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-168-34:272157:272649 [7] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+ip-26-0-161-178:2815967:2816521 [4] NCCL INFO Using network Libfabric
+ip-26-0-161-178:2815970:2816523 [7] NCCL INFO Using network Libfabric
+ip-26-0-161-178:2815964:2816524 [1] NCCL INFO Using network Libfabric
+ip-26-0-161-178:2815966:2816526 [3] NCCL INFO Using network Libfabric
+ip-26-0-168-52:2722052:2722578 [7] NCCL INFO Using network Libfabric
+ip-26-0-168-52:2722049:2722579 [4] NCCL INFO Using network Libfabric
+ip-26-0-168-52:2722046:2722580 [1] NCCL INFO Using network Libfabric
+ip-26-0-161-178:2815965:2816528 [2] NCCL INFO Using network Libfabric
+ip-26-0-168-30:235651:236141 [3] NCCL INFO Connected NVLS tree
+ip-26-0-168-30:235651:236141 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-30:235651:236141 [3] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-168-30:235651:236141 [3] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-178:2815963:2816530 [0] NCCL INFO Using network Libfabric
+ip-26-0-168-52:2722048:2722581 [3] NCCL INFO Using network Libfabric
+ip-26-0-168-52:2722051:2722583 [6] NCCL INFO Using network Libfabric
+ip-26-0-161-178:2815968:2816533 [5] NCCL INFO Using network Libfabric
+ip-26-0-161-178:2815969:2816532 [6] NCCL INFO Using network Libfabric
+ip-26-0-168-52:2722045:2722584 [0] NCCL INFO Using network Libfabric
+ip-26-0-168-52:2722047:2722582 [2] NCCL INFO Using network Libfabric
+ip-26-0-168-52:2722052:2722578 [7] NCCL INFO comm 0xac4c980 rank 1 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0xe7e10d0278883222 - Init START
+ip-26-0-161-178:2815970:2816523 [7] NCCL INFO comm 0xb2c20f0 rank 0 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0xe7e10d0278883222 - Init START
+ip-26-0-168-52:2722049:2722579 [4] NCCL INFO comm 0x9df02f0 rank 1 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0x564424f9e54a802c - Init START
+ip-26-0-161-178:2815967:2816521 [4] NCCL INFO comm 0x9ddfca0 rank 0 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0x564424f9e54a802c - Init START
+ip-26-0-168-34:272153:272651 [3] NCCL INFO Connected NVLS tree
+ip-26-0-161-178:2815964:2816524 [1] NCCL INFO comm 0xa996570 rank 0 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0x60ec45dbd7f4bf99 - Init START
+ip-26-0-168-52:2722046:2722580 [1] NCCL INFO comm 0xb829990 rank 1 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0x60ec45dbd7f4bf99 - Init START
+ip-26-0-168-34:272153:272651 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-34:272153:272651 [3] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-168-34:272153:272651 [3] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-161-178:2815970:2816523 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-52:2722050:2722585 [5] NCCL INFO Using network Libfabric
+ip-26-0-161-178:2815967:2816521 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-52:2722052:2722578 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-178:2815964:2816524 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-52:2722049:2722579 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-52:2722046:2722580 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-52:2722048:2722581 [3] NCCL INFO comm 0x9c25f50 rank 1 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0x30479eeb41b2f950 - Init START
+ip-26-0-161-178:2815966:2816526 [3] NCCL INFO comm 0xb6a2800 rank 0 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0x30479eeb41b2f950 - Init START
+ip-26-0-161-178:2815966:2816526 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-52:2722051:2722583 [6] NCCL INFO comm 0xa82da40 rank 1 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0xef09316ae5338288 - Init START
+ip-26-0-161-178:2815969:2816532 [6] NCCL INFO comm 0xbc559e0 rank 0 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0xef09316ae5338288 - Init START
+ip-26-0-168-52:2722048:2722581 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-52:2722045:2722584 [0] NCCL INFO comm 0xa567690 rank 1 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0xce4b7fc6d9e0d602 - Init START
+ip-26-0-161-178:2815963:2816530 [0] NCCL INFO comm 0xb57d690 rank 0 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0xce4b7fc6d9e0d602 - Init START
+ip-26-0-161-178:2815965:2816528 [2] NCCL INFO comm 0xadd1650 rank 0 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0x864e31bd77042a79 - Init START
+ip-26-0-161-178:2815969:2816532 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-52:2722047:2722582 [2] NCCL INFO comm 0xb461dc0 rank 1 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0x864e31bd77042a79 - Init START
+ip-26-0-161-178:2815963:2816530 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-52:2722051:2722583 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-52:2722047:2722582 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-178:2815965:2816528 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-52:2722045:2722584 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-178:2815968:2816533 [5] NCCL INFO comm 0xc28c080 rank 0 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0x5034f31407405e64 - Init START
+ip-26-0-168-52:2722050:2722585 [5] NCCL INFO comm 0x9c5a090 rank 1 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0x5034f31407405e64 - Init START
+ip-26-0-168-52:2722050:2722585 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-178:2815968:2816533 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-30:235653:236140 [5] NCCL INFO Connected NVLS tree
+ip-26-0-168-30:235653:236140 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-30:235653:236140 [5] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-168-30:235653:236140 [5] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-168-30:235654:236146 [6] NCCL INFO comm 0xb97f190 rank 14 nranks 32 cudaDev 6 nvmlDev 6 busId b9000 commId 0xe6d0f49db1af694c - Init COMPLETE
+ip-26-0-168-30:235651:236141 [3] NCCL INFO comm 0xa4ad700 rank 11 nranks 32 cudaDev 3 nvmlDev 3 busId 86000 commId 0xe6d0f49db1af694c - Init COMPLETE
+ip-26-0-168-30:235650:236144 [2] NCCL INFO comm 0xa4cc2d0 rank 10 nranks 32 cudaDev 2 nvmlDev 2 busId 75000 commId 0xe6d0f49db1af694c - Init COMPLETE
+ip-26-0-168-30:235648:236145 [0] NCCL INFO comm 0xa954a00 rank 8 nranks 32 cudaDev 0 nvmlDev 0 busId 53000 commId 0xe6d0f49db1af694c - Init COMPLETE
+ip-26-0-168-30:235652:236147 [4] NCCL INFO comm 0xb25b2a0 rank 12 nranks 32 cudaDev 4 nvmlDev 4 busId 97000 commId 0xe6d0f49db1af694c - Init COMPLETE
+ip-26-0-168-30:235655:236142 [7] NCCL INFO comm 0xb48d9f0 rank 15 nranks 32 cudaDev 7 nvmlDev 7 busId ca000 commId 0xe6d0f49db1af694c - Init COMPLETE
+ip-26-0-168-30:235653:236140 [5] NCCL INFO comm 0xa6da390 rank 13 nranks 32 cudaDev 5 nvmlDev 5 busId a8000 commId 0xe6d0f49db1af694c - Init COMPLETE
+ip-26-0-168-30:235649:236143 [1] NCCL INFO comm 0xb1b3870 rank 9 nranks 32 cudaDev 1 nvmlDev 1 busId 64000 commId 0xe6d0f49db1af694c - Init COMPLETE
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+ip-26-0-161-142:62886:63411 [5] NCCL INFO Using network Libfabric
+ip-26-0-161-142:62883:63421 [2] NCCL INFO Using network Libfabric
+ip-26-0-161-142:62884:63417 [3] NCCL INFO Using network Libfabric
+ip-26-0-168-30:235653:236177 [5] NCCL INFO Using network Libfabric
+ip-26-0-161-142:62888:63422 [7] NCCL INFO Using network Libfabric
+ip-26-0-168-30:235648:236179 [0] NCCL INFO Using network Libfabric
+ip-26-0-161-142:62881:63420 [0] NCCL INFO Using network Libfabric
+ip-26-0-161-142:62885:63425 [4] NCCL INFO Using network Libfabric
+ip-26-0-161-142:62887:63423 [6] NCCL INFO Using network Libfabric
+ip-26-0-168-30:235651:236178 [3] NCCL INFO Using network Libfabric
+ip-26-0-161-142:62882:63424 [1] NCCL INFO Using network Libfabric
+ip-26-0-168-30:235650:236180 [2] NCCL INFO Using network Libfabric
+ip-26-0-168-30:235652:236183 [4] NCCL INFO Using network Libfabric
+ip-26-0-168-30:235655:236181 [7] NCCL INFO Using network Libfabric
+ip-26-0-168-30:235654:236182 [6] NCCL INFO Using network Libfabric
+ip-26-0-168-30:235649:236184 [1] NCCL INFO Using network Libfabric
+ip-26-0-168-30:235653:236177 [5] NCCL INFO comm 0xb129990 rank 1 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0xac4fbfdf97289f71 - Init START
+ip-26-0-161-142:62881:63420 [0] NCCL INFO comm 0xc328080 rank 0 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0x20c1c1aff1850363 - Init START
+ip-26-0-168-30:235648:236179 [0] NCCL INFO comm 0x9ee9b90 rank 1 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0x20c1c1aff1850363 - Init START
+ip-26-0-161-142:62886:63411 [5] NCCL INFO comm 0xbfec390 rank 0 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0xac4fbfdf97289f71 - Init START
+ip-26-0-168-30:235651:236178 [3] NCCL INFO comm 0xaefccc0 rank 1 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0x59659ca4831879fd - Init START
+ip-26-0-161-142:62884:63417 [3] NCCL INFO comm 0xb7b8cb0 rank 0 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0x59659ca4831879fd - Init START
+ip-26-0-161-142:62881:63420 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-30:235650:236180 [2] NCCL INFO comm 0xaf1bad0 rank 1 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0xdd28a795e7538c4 - Init START
+ip-26-0-161-142:62886:63411 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-142:62883:63421 [2] NCCL INFO comm 0xad52dc0 rank 0 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0xdd28a795e7538c4 - Init START
+ip-26-0-168-30:235648:236179 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-30:235653:236177 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-142:62884:63417 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-142:62888:63422 [7] NCCL INFO comm 0xb362750 rank 0 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0xeada9362f427fa83 - Init START
+ip-26-0-168-30:235655:236181 [7] NCCL INFO comm 0xaa22540 rank 1 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0xeada9362f427fa83 - Init START
+ip-26-0-168-30:235652:236183 [4] NCCL INFO comm 0xbcaab00 rank 1 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0xfa0e0ba119a417d5 - Init START
+ip-26-0-168-30:235651:236178 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-142:62885:63425 [4] NCCL INFO comm 0xc01a5d0 rank 0 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0xfa0e0ba119a417d5 - Init START
+ip-26-0-161-142:62883:63421 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-30:235650:236180 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-142:62888:63422 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-30:235654:236182 [6] NCCL INFO comm 0xaf27040 rank 1 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0x2f2b62613eb35974 - Init START
+ip-26-0-168-30:235655:236181 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-142:62887:63423 [6] NCCL INFO comm 0xa87cdc0 rank 0 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0x2f2b62613eb35974 - Init START
+ip-26-0-168-30:235649:236184 [1] NCCL INFO comm 0xbc1a340 rank 1 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0xc5f8d796be5089cb - Init START
+ip-26-0-168-30:235652:236183 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-142:62885:63425 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-142:62882:63424 [1] NCCL INFO comm 0xb0d54c0 rank 0 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0xc5f8d796be5089cb - Init START
+ip-26-0-161-142:62887:63423 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-30:235654:236182 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-142:62882:63424 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-30:235649:236184 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-34:272155:272652 [5] NCCL INFO Connected NVLS tree
+ip-26-0-168-34:272155:272652 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-34:272155:272652 [5] NCCL INFO threadThresholds 8/8/64 | 256/8/64 | 512 | 512
+ip-26-0-168-34:272155:272652 [5] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-165-59:1557266:1557798 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-168-34:272151:272646 [1] NCCL INFO comm 0xb4d5a80 rank 17 nranks 32 cudaDev 1 nvmlDev 1 busId 64000 commId 0xe6d0f49db1af694c - Init COMPLETE
+ip-26-0-168-34:272152:272647 [2] NCCL INFO comm 0xba40780 rank 18 nranks 32 cudaDev 2 nvmlDev 2 busId 75000 commId 0xe6d0f49db1af694c - Init COMPLETE
+ip-26-0-168-34:272154:272650 [4] NCCL INFO comm 0xa853b80 rank 20 nranks 32 cudaDev 4 nvmlDev 4 busId 97000 commId 0xe6d0f49db1af694c - Init COMPLETE
+ip-26-0-168-34:272156:272648 [6] NCCL INFO comm 0xb76eb00 rank 22 nranks 32 cudaDev 6 nvmlDev 6 busId b9000 commId 0xe6d0f49db1af694c - Init COMPLETE
+ip-26-0-168-34:272150:272645 [0] NCCL INFO comm 0xb33a540 rank 16 nranks 32 cudaDev 0 nvmlDev 0 busId 53000 commId 0xe6d0f49db1af694c - Init COMPLETE
+ip-26-0-168-34:272153:272651 [3] NCCL INFO comm 0x9b8ea10 rank 19 nranks 32 cudaDev 3 nvmlDev 3 busId 86000 commId 0xe6d0f49db1af694c - Init COMPLETE
+ip-26-0-168-34:272157:272649 [7] NCCL INFO comm 0xb441140 rank 23 nranks 32 cudaDev 7 nvmlDev 7 busId ca000 commId 0xe6d0f49db1af694c - Init COMPLETE
+ip-26-0-168-34:272155:272652 [5] NCCL INFO comm 0xa77e610 rank 21 nranks 32 cudaDev 5 nvmlDev 5 busId a8000 commId 0xe6d0f49db1af694c - Init COMPLETE
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+ip-26-0-161-153:2636090:2636647 [6] NCCL INFO Using network Libfabric
+ip-26-0-168-34:272156:272681 [6] NCCL INFO Using network Libfabric
+ip-26-0-161-153:2636086:2636652 [2] NCCL INFO Using network Libfabric
+ip-26-0-161-153:2636089:2636654 [5] NCCL INFO Using network Libfabric
+ip-26-0-161-153:2636088:2636653 [4] NCCL INFO Using network Libfabric
+ip-26-0-161-153:2636087:2636655 [3] NCCL INFO Using network Libfabric
+ip-26-0-168-34:272155:272684 [5] NCCL INFO Using network Libfabric
+ip-26-0-161-153:2636091:2636656 [7] NCCL INFO Using network Libfabric
+ip-26-0-168-34:272152:272682 [2] NCCL INFO Using network Libfabric
+ip-26-0-161-153:2636085:2636657 [1] NCCL INFO Using network Libfabric
+ip-26-0-168-34:272154:272683 [4] NCCL INFO Using network Libfabric
+ip-26-0-168-34:272151:272686 [1] NCCL INFO Using network Libfabric
+ip-26-0-168-34:272153:272685 [3] NCCL INFO Using network Libfabric
+ip-26-0-168-34:272157:272687 [7] NCCL INFO Using network Libfabric
+ip-26-0-161-153:2636090:2636647 [6] NCCL INFO comm 0xbc16d40 rank 0 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0x336838438ba1cbae - Init START
+ip-26-0-168-34:272156:272681 [6] NCCL INFO comm 0xc1be320 rank 1 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0x336838438ba1cbae - Init START
+ip-26-0-161-153:2636090:2636647 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-34:272156:272681 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-34:272155:272684 [5] NCCL INFO comm 0xb1cdf00 rank 1 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0x5cd2fe565188f730 - Init START
+ip-26-0-161-153:2636089:2636654 [5] NCCL INFO comm 0xa92d820 rank 0 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0x5cd2fe565188f730 - Init START
+ip-26-0-168-34:272154:272683 [4] NCCL INFO comm 0xb2a2df0 rank 1 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0xbf59abaaafdb1e45 - Init START
+ip-26-0-168-34:272152:272682 [2] NCCL INFO comm 0xc48fee0 rank 1 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0x52002fcca2d62604 - Init START
+ip-26-0-161-153:2636088:2636653 [4] NCCL INFO comm 0xb4b0490 rank 0 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0xbf59abaaafdb1e45 - Init START
+ip-26-0-161-153:2636084:2636659 [0] NCCL INFO Using network Libfabric
+ip-26-0-168-34:272151:272686 [1] NCCL INFO comm 0xaa6e840 rank 1 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0xdbcc72fe4166d27e - Init START
+ip-26-0-161-153:2636086:2636652 [2] NCCL INFO comm 0xbb5f0d0 rank 0 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0x52002fcca2d62604 - Init START
+ip-26-0-161-153:2636089:2636654 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-153:2636085:2636657 [1] NCCL INFO comm 0xc3d3070 rank 0 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0xdbcc72fe4166d27e - Init START
+ip-26-0-168-34:272155:272684 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-34:272153:272685 [3] NCCL INFO comm 0xa5de1a0 rank 1 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0x567d4d5e4099f109 - Init START
+ip-26-0-161-153:2636087:2636655 [3] NCCL INFO comm 0xaab34a0 rank 0 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0x567d4d5e4099f109 - Init START
+ip-26-0-161-153:2636088:2636653 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-34:272152:272682 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-153:2636086:2636652 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-153:2636085:2636657 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-34:272154:272683 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-34:272151:272686 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-153:2636087:2636655 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-34:272153:272685 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-34:272157:272687 [7] NCCL INFO comm 0xa9d6240 rank 1 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0x56bd7b6de7070786 - Init START
+ip-26-0-161-153:2636091:2636656 [7] NCCL INFO comm 0xb2113a0 rank 0 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0x56bd7b6de7070786 - Init START
+ip-26-0-161-153:2636091:2636656 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-34:272157:272687 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-34:272150:272688 [0] NCCL INFO Using network Libfabric
+ip-26-0-168-34:272150:272688 [0] NCCL INFO comm 0xbd8a030 rank 1 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0x8be89873e2d856ad - Init START
+ip-26-0-161-153:2636084:2636659 [0] NCCL INFO comm 0xa708e10 rank 0 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0x8be89873e2d856ad - Init START
+ip-26-0-161-153:2636084:2636659 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-34:272150:272688 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-59:1557273:1557801 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-165-59:1557272:1557800 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-165-59:1557267:1557799 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-161-138:96565:97145 [2] NCCL INFO Setting affinity for GPU 2 to 0fff,ffffffff
+ip-26-0-161-123:32711:33234 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-161-138:96569:97147 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,fff00000,00000000
+ip-26-0-161-138:96569:97147 [6] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-161-138:96569:97147 [6] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-161-138:96569:97147 [6] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-161-138:96569:97147 [6] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-165-59:1557272:1557800 [6] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-165-59:1557272:1557800 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-138:96569:97147 [6] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-161-138:96569:97147 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-59:1557269:1557802 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-161-123:32706:33245 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-165-59:1557268:1557796 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-165-59:1557268:1557796 [2] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-165-59:1557268:1557796 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-138:96565:97145 [2] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-161-138:96565:97145 [2] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-161-138:96565:97145 [2] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-161-138:96565:97145 [2] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-161-138:96565:97145 [2] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-161-138:96565:97145 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-59:1557271:1557795 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-161-138:96564:97149 [1] NCCL INFO Setting affinity for GPU 1 to 0fff,ffffffff
+ip-26-0-165-59:1557270:1557797 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-161-138:96564:97149 [1] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-161-138:96564:97149 [1] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-161-138:96564:97149 [1] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-161-138:96564:97149 [1] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-165-59:1557267:1557799 [1] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-165-59:1557267:1557799 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-138:96564:97149 [1] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-161-138:96564:97149 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-123:32705:33248 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-161-123:32709:33243 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-161-123:32708:33242 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-161-138:96566:97151 [3] NCCL INFO Setting affinity for GPU 3 to 0fff,ffffffff
+ip-26-0-161-138:96566:97151 [3] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-161-138:96566:97151 [3] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-161-138:96566:97151 [3] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-161-138:96566:97151 [3] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-161-138:96566:97151 [3] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-161-138:96566:97151 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-59:1557269:1557802 [3] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-165-59:1557269:1557802 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-123:32704:33247 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-161-123:32707:33244 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-161-138:96567:97144 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,fff00000,00000000
+ip-26-0-161-123:32710:33246 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-161-138:96570:97146 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,fff00000,00000000
+ip-26-0-161-138:96567:97144 [4] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-161-138:96567:97144 [4] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-161-138:96567:97144 [4] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-161-138:96567:97144 [4] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-161-138:96567:97144 [4] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-161-138:96567:97144 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-59:1557270:1557797 [4] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-165-59:1557270:1557797 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-138:96568:97143 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,fff00000,00000000
+ip-26-0-161-138:96570:97146 [7] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-161-138:96570:97146 [7] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-161-138:96570:97146 [7] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-161-138:96570:97146 [7] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-161-138:96570:97146 [7] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-161-138:96570:97146 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-59:1557273:1557801 [7] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-165-59:1557273:1557801 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-138:96568:97143 [5] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-161-138:96568:97143 [5] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-161-138:96568:97143 [5] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-165-59:1557271:1557795 [5] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-165-59:1557271:1557795 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-138:96568:97143 [5] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-161-138:96568:97143 [5] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-161-138:96568:97143 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-138:96563:97148 [0] NCCL INFO Setting affinity for GPU 0 to 0fff,ffffffff
+ip-26-0-161-138:96563:97148 [0] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-161-138:96563:97148 [0] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-161-138:96563:97148 [0] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-161-138:96563:97148 [0] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-165-59:1557266:1557798 [0] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-165-59:1557266:1557798 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-138:96563:97148 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-161-138:96563:97148 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-78:114516:115044 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-165-202:153563:154098 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-165-202:153563:154098 [1] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-165-202:153563:154098 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-123:32705:33248 [1] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-161-123:32705:33248 [1] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-161-123:32705:33248 [1] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-161-123:32705:33248 [1] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-161-123:32705:33248 [1] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-161-123:32705:33248 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-78:114514:115049 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-165-202:153564:154092 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-165-202:153564:154092 [2] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-165-202:153564:154092 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-123:32706:33245 [2] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-161-123:32706:33245 [2] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-161-123:32706:33245 [2] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-161-123:32706:33245 [2] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-161-123:32706:33245 [2] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-161-123:32706:33245 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-78:114513:115048 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-161-138:96569:97147 [6] NCCL INFO Channel 00/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96569:97147 [6] NCCL INFO Channel 01/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96569:97147 [6] NCCL INFO Channel 02/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96569:97147 [6] NCCL INFO Channel 03/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96569:97147 [6] NCCL INFO Channel 00/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96569:97147 [6] NCCL INFO Channel 01/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96569:97147 [6] NCCL INFO Channel 02/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96569:97147 [6] NCCL INFO Channel 03/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153566:154097 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-165-59:1557272:1557800 [6] NCCL INFO Channel 00/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557272:1557800 [6] NCCL INFO Channel 01/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557272:1557800 [6] NCCL INFO Channel 02/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557272:1557800 [6] NCCL INFO Channel 03/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557272:1557800 [6] NCCL INFO Channel 00/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557272:1557800 [6] NCCL INFO Channel 01/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557272:1557800 [6] NCCL INFO Channel 02/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557272:1557800 [6] NCCL INFO Channel 03/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153566:154097 [4] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-165-202:153566:154097 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-123:32708:33242 [4] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-161-123:32708:33242 [4] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-161-123:32708:33242 [4] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-161-123:32708:33242 [4] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-161-123:32708:33242 [4] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-161-123:32708:33242 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-138:96565:97145 [2] NCCL INFO Channel 00/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96565:97145 [2] NCCL INFO Channel 01/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96565:97145 [2] NCCL INFO Channel 02/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96565:97145 [2] NCCL INFO Channel 03/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96565:97145 [2] NCCL INFO Channel 00/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96565:97145 [2] NCCL INFO Channel 01/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96565:97145 [2] NCCL INFO Channel 02/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96565:97145 [2] NCCL INFO Channel 03/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557268:1557796 [2] NCCL INFO Channel 00/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557268:1557796 [2] NCCL INFO Channel 01/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557268:1557796 [2] NCCL INFO Channel 02/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557268:1557796 [2] NCCL INFO Channel 03/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557268:1557796 [2] NCCL INFO Channel 00/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557268:1557796 [2] NCCL INFO Channel 01/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557268:1557796 [2] NCCL INFO Channel 02/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557268:1557796 [2] NCCL INFO Channel 03/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114512:115045 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-161-78:114519:115047 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-165-202:153565:154094 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-160-225:124160:124698 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-161-138:96564:97149 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153565:154094 [3] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-165-202:153565:154094 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-138:96564:97149 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32707:33244 [3] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-161-123:32707:33244 [3] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-161-123:32707:33244 [3] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-161-123:32707:33244 [3] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-161-123:32707:33244 [3] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-161-123:32707:33244 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-138:96564:97149 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96564:97149 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96564:97149 [1] NCCL INFO Channel 00/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96564:97149 [1] NCCL INFO Channel 01/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96564:97149 [1] NCCL INFO Channel 02/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96564:97149 [1] NCCL INFO Channel 03/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124160:124698 [0] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-160-225:124160:124698 [0] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-160-225:124160:124698 [0] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-160-225:124160:124698 [0] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-160-225:124160:124698 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-160-225:124160:124698 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-78:114512:115045 [0] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-161-78:114512:115045 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-59:1557267:1557799 [1] NCCL INFO Channel 00/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557267:1557799 [1] NCCL INFO Channel 01/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557267:1557799 [1] NCCL INFO Channel 02/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557267:1557799 [1] NCCL INFO Channel 03/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153568:154096 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-165-59:1557267:1557799 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557267:1557799 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557267:1557799 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557267:1557799 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153568:154096 [6] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-165-202:153568:154096 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-123:32710:33246 [6] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-161-123:32710:33246 [6] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-161-123:32710:33246 [6] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-161-123:32710:33246 [6] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-161-123:32710:33246 [6] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-161-123:32710:33246 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-202:153569:154091 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-165-202:153569:154091 [7] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-165-202:153569:154091 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-123:32711:33234 [7] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-161-123:32711:33234 [7] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-161-123:32711:33234 [7] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-161-123:32711:33234 [7] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-161-123:32711:33234 [7] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-161-123:32711:33234 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-78:114517:115046 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-161-78:114515:115050 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-165-164:1227470:1227998 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-165-164:1227465:1227996 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-165-202:153567:154093 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-165-202:153567:154093 [5] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-165-202:153567:154093 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-123:32709:33243 [5] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-161-123:32709:33243 [5] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-161-123:32709:33243 [5] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-161-123:32709:33243 [5] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-161-123:32709:33243 [5] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-161-123:32709:33243 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-78:114518:115051 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-165-164:1227464:1227997 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-160-225:124164:124702 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-165-202:153562:154095 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-161-138:96566:97151 [3] NCCL INFO Channel 00/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96566:97151 [3] NCCL INFO Channel 01/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96566:97151 [3] NCCL INFO Channel 02/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96566:97151 [3] NCCL INFO Channel 03/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96566:97151 [3] NCCL INFO Channel 00/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96566:97151 [3] NCCL INFO Channel 01/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:124164:124702 [4] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-160-225:124164:124702 [4] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-160-225:124164:124702 [4] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-160-225:124164:124702 [4] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-160-225:124164:124702 [4] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-160-225:124164:124702 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-138:96566:97151 [3] NCCL INFO Channel 02/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114516:115044 [4] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-161-78:114516:115044 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-138:96566:97151 [3] NCCL INFO Channel 03/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153562:154095 [0] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-165-202:153562:154095 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-123:32704:33247 [0] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-161-123:32704:33247 [0] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-161-123:32704:33247 [0] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-161-123:32704:33247 [0] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-161-123:32704:33247 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-161-123:32704:33247 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-59:1557269:1557802 [3] NCCL INFO Channel 00/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557269:1557802 [3] NCCL INFO Channel 01/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557269:1557802 [3] NCCL INFO Channel 02/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557269:1557802 [3] NCCL INFO Channel 03/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557269:1557802 [3] NCCL INFO Channel 00/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557269:1557802 [3] NCCL INFO Channel 01/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557269:1557802 [3] NCCL INFO Channel 02/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557269:1557802 [3] NCCL INFO Channel 03/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557270:1557797 [4] NCCL INFO Channel 00/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557270:1557797 [4] NCCL INFO Channel 01/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557270:1557797 [4] NCCL INFO Channel 02/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557270:1557797 [4] NCCL INFO Channel 03/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557270:1557797 [4] NCCL INFO Channel 00/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557270:1557797 [4] NCCL INFO Channel 01/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557270:1557797 [4] NCCL INFO Channel 02/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557270:1557797 [4] NCCL INFO Channel 03/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96894:97424 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-161-138:96568:97143 [5] NCCL INFO Channel 00/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96568:97143 [5] NCCL INFO Channel 01/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96570:97146 [7] NCCL INFO Channel 00/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96568:97143 [5] NCCL INFO Channel 02/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96568:97143 [5] NCCL INFO Channel 03/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96570:97146 [7] NCCL INFO Channel 01/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96568:97143 [5] NCCL INFO Channel 00/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96567:97144 [4] NCCL INFO Channel 00/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96570:97146 [7] NCCL INFO Channel 02/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96568:97143 [5] NCCL INFO Channel 01/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96568:97143 [5] NCCL INFO Channel 02/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96570:97146 [7] NCCL INFO Channel 03/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96567:97144 [4] NCCL INFO Channel 01/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96568:97143 [5] NCCL INFO Channel 03/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96567:97144 [4] NCCL INFO Channel 02/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96567:97144 [4] NCCL INFO Channel 03/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96567:97144 [4] NCCL INFO Channel 00/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557271:1557795 [5] NCCL INFO Channel 00/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96567:97144 [4] NCCL INFO Channel 01/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96567:97144 [4] NCCL INFO Channel 02/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557271:1557795 [5] NCCL INFO Channel 01/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96567:97144 [4] NCCL INFO Channel 03/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557271:1557795 [5] NCCL INFO Channel 02/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557271:1557795 [5] NCCL INFO Channel 03/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557271:1557795 [5] NCCL INFO Channel 00/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557271:1557795 [5] NCCL INFO Channel 01/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557273:1557801 [7] NCCL INFO Channel 00/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557271:1557795 [5] NCCL INFO Channel 02/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557273:1557801 [7] NCCL INFO Channel 01/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557271:1557795 [5] NCCL INFO Channel 03/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557273:1557801 [7] NCCL INFO Channel 02/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557273:1557801 [7] NCCL INFO Channel 03/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557273:1557801 [7] NCCL INFO Channel 00/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96570:97146 [7] NCCL INFO Channel 00/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557273:1557801 [7] NCCL INFO Channel 01/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96570:97146 [7] NCCL INFO Channel 01/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557273:1557801 [7] NCCL INFO Channel 02/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96570:97146 [7] NCCL INFO Channel 02/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557273:1557801 [7] NCCL INFO Channel 03/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96570:97146 [7] NCCL INFO Channel 03/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96563:97148 [0] NCCL INFO Channel 00/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96563:97148 [0] NCCL INFO Channel 01/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96563:97148 [0] NCCL INFO Channel 02/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96563:97148 [0] NCCL INFO Channel 03/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227463:1227994 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-161-138:96563:97148 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96563:97148 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96563:97148 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96563:97148 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227468:1228000 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-165-59:1557266:1557798 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557266:1557798 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557266:1557798 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557266:1557798 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557266:1557798 [0] NCCL INFO Channel 00/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557266:1557798 [0] NCCL INFO Channel 01/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557266:1557798 [0] NCCL INFO Channel 02/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557266:1557798 [0] NCCL INFO Channel 03/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:124167:124700 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-160-225:124167:124700 [7] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-160-225:124167:124700 [7] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-160-225:124167:124700 [7] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-160-225:124167:124700 [7] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-160-225:124167:124700 [7] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-160-225:124167:124700 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-78:114519:115047 [7] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-161-78:114519:115047 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-202:153563:154098 [1] NCCL INFO Channel 00/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153563:154098 [1] NCCL INFO Channel 01/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227469:1227995 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-165-202:153563:154098 [1] NCCL INFO Channel 02/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153563:154098 [1] NCCL INFO Channel 03/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153563:154098 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153563:154098 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153563:154098 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153563:154098 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227469:1227995 [6] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-165-164:1227469:1227995 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-103:96894:97424 [6] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-161-103:96894:97424 [6] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-161-103:96894:97424 [6] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-161-103:96894:97424 [6] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-161-103:96894:97424 [6] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-161-103:96894:97424 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-202:153564:154092 [2] NCCL INFO Channel 00/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153564:154092 [2] NCCL INFO Channel 01/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153564:154092 [2] NCCL INFO Channel 02/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153564:154092 [2] NCCL INFO Channel 03/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153564:154092 [2] NCCL INFO Channel 00/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153564:154092 [2] NCCL INFO Channel 01/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32705:33248 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153564:154092 [2] NCCL INFO Channel 02/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153564:154092 [2] NCCL INFO Channel 03/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32705:33248 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32705:33248 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32705:33248 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32705:33248 [1] NCCL INFO Channel 00/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32705:33248 [1] NCCL INFO Channel 01/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32705:33248 [1] NCCL INFO Channel 02/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32705:33248 [1] NCCL INFO Channel 03/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227467:1227999 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-161-123:32706:33245 [2] NCCL INFO Channel 00/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32706:33245 [2] NCCL INFO Channel 01/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32706:33245 [2] NCCL INFO Channel 02/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32706:33245 [2] NCCL INFO Channel 03/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32706:33245 [2] NCCL INFO Channel 00/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32706:33245 [2] NCCL INFO Channel 01/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32706:33245 [2] NCCL INFO Channel 02/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32706:33245 [2] NCCL INFO Channel 03/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227466:1228001 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-165-131:1630211:1630779 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-165-202:153566:154097 [4] NCCL INFO Channel 00/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153566:154097 [4] NCCL INFO Channel 01/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153566:154097 [4] NCCL INFO Channel 02/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153566:154097 [4] NCCL INFO Channel 03/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153566:154097 [4] NCCL INFO Channel 00/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153566:154097 [4] NCCL INFO Channel 01/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153566:154097 [4] NCCL INFO Channel 02/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:124162:124705 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-165-202:153566:154097 [4] NCCL INFO Channel 03/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32708:33242 [4] NCCL INFO Channel 00/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32708:33242 [4] NCCL INFO Channel 01/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32708:33242 [4] NCCL INFO Channel 02/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32708:33242 [4] NCCL INFO Channel 03/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32708:33242 [4] NCCL INFO Channel 00/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630208:1630782 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-160-225:124162:124705 [2] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-160-225:124162:124705 [2] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-160-225:124162:124705 [2] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-160-225:124162:124705 [2] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-160-225:124162:124705 [2] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-161-123:32708:33242 [4] NCCL INFO Channel 01/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:124162:124705 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-78:114514:115049 [2] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-161-78:114514:115049 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-123:32708:33242 [4] NCCL INFO Channel 02/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32708:33242 [4] NCCL INFO Channel 03/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:124165:124699 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-160-225:124165:124699 [5] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-160-225:124165:124699 [5] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-160-225:124165:124699 [5] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-160-225:124165:124699 [5] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-160-225:124165:124699 [5] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-160-225:124165:124699 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-78:114517:115046 [5] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-161-78:114517:115046 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-202:153565:154094 [3] NCCL INFO Channel 00/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153565:154094 [3] NCCL INFO Channel 01/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153565:154094 [3] NCCL INFO Channel 02/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153565:154094 [3] NCCL INFO Channel 03/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153565:154094 [3] NCCL INFO Channel 00/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153565:154094 [3] NCCL INFO Channel 01/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153565:154094 [3] NCCL INFO Channel 02/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153565:154094 [3] NCCL INFO Channel 03/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114512:115045 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114512:115045 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114512:115045 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114512:115045 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114512:115045 [0] NCCL INFO Channel 00/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114512:115045 [0] NCCL INFO Channel 01/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114512:115045 [0] NCCL INFO Channel 02/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114512:115045 [0] NCCL INFO Channel 03/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:124160:124698 [0] NCCL INFO Channel 00/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:124160:124698 [0] NCCL INFO Channel 01/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:124160:124698 [0] NCCL INFO Channel 02/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:124160:124698 [0] NCCL INFO Channel 03/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:124160:124698 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32707:33244 [3] NCCL INFO Channel 00/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:124160:124698 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32707:33244 [3] NCCL INFO Channel 01/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:124160:124698 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:124160:124698 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32707:33244 [3] NCCL INFO Channel 02/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32707:33244 [3] NCCL INFO Channel 03/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32707:33244 [3] NCCL INFO Channel 00/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32707:33244 [3] NCCL INFO Channel 01/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32707:33244 [3] NCCL INFO Channel 02/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32707:33244 [3] NCCL INFO Channel 03/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:124166:124708 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-160-242:1364908:1365452 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,fff00000,00000000
+ip-26-0-165-202:153568:154096 [6] NCCL INFO Channel 00/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153568:154096 [6] NCCL INFO Channel 01/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153568:154096 [6] NCCL INFO Channel 02/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153568:154096 [6] NCCL INFO Channel 03/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153568:154096 [6] NCCL INFO Channel 00/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153568:154096 [6] NCCL INFO Channel 01/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153568:154096 [6] NCCL INFO Channel 02/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153568:154096 [6] NCCL INFO Channel 03/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32710:33246 [6] NCCL INFO Channel 00/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32710:33246 [6] NCCL INFO Channel 01/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32710:33246 [6] NCCL INFO Channel 02/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32710:33246 [6] NCCL INFO Channel 03/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32710:33246 [6] NCCL INFO Channel 00/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153569:154091 [7] NCCL INFO Channel 00/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32710:33246 [6] NCCL INFO Channel 01/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:124166:124708 [6] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-160-225:124166:124708 [6] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-160-225:124166:124708 [6] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-160-225:124166:124708 [6] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-160-225:124166:124708 [6] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-160-225:124166:124708 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-123:32710:33246 [6] NCCL INFO Channel 02/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153569:154091 [7] NCCL INFO Channel 01/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114518:115051 [6] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-161-78:114518:115051 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-123:32710:33246 [6] NCCL INFO Channel 03/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153569:154091 [7] NCCL INFO Channel 02/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364908:1365452 [6] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-160-242:1364908:1365452 [6] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-160-242:1364908:1365452 [6] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-165-202:153569:154091 [7] NCCL INFO Channel 03/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630211:1630779 [6] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-165-131:1630211:1630779 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-242:1364908:1365452 [6] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-160-242:1364908:1365452 [6] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-160-242:1364908:1365452 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-202:153569:154091 [7] NCCL INFO Channel 00/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153569:154091 [7] NCCL INFO Channel 01/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153569:154091 [7] NCCL INFO Channel 02/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153569:154091 [7] NCCL INFO Channel 03/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96895:97427 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-161-103:96890:97425 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-161-123:32711:33234 [7] NCCL INFO Channel 00/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96889:97426 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-161-123:32711:33234 [7] NCCL INFO Channel 01/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32711:33234 [7] NCCL INFO Channel 02/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32711:33234 [7] NCCL INFO Channel 03/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32711:33234 [7] NCCL INFO Channel 00/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32711:33234 [7] NCCL INFO Channel 01/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124163:124706 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-161-123:32711:33234 [7] NCCL INFO Channel 02/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227465:1227996 [2] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-165-164:1227465:1227996 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-103:96890:97425 [2] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-161-103:96890:97425 [2] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-161-103:96890:97425 [2] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-161-103:96890:97425 [2] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-161-103:96890:97425 [2] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-161-103:96890:97425 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-103:96895:97427 [7] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-161-103:96895:97427 [7] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-161-103:96895:97427 [7] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-161-103:96895:97427 [7] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-161-123:32711:33234 [7] NCCL INFO Channel 03/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96895:97427 [7] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-161-103:96895:97427 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-164:1227470:1227998 [7] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-165-164:1227470:1227998 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-103:96889:97426 [1] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-161-103:96889:97426 [1] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-161-103:96889:97426 [1] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-161-103:96889:97426 [1] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-161-103:96889:97426 [1] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-161-103:96889:97426 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-164:1227464:1227997 [1] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-165-164:1227464:1227997 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-242:1364907:1365459 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,fff00000,00000000
+ip-26-0-160-225:124163:124706 [3] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-160-225:124163:124706 [3] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-160-225:124163:124706 [3] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-160-225:124163:124706 [3] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-160-225:124163:124706 [3] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-160-225:124163:124706 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-78:114515:115050 [3] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-161-78:114515:115050 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-103:96888:97423 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-165-131:1630207:1630780 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-161-103:96892:97429 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-165-164:1227463:1227994 [0] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-165-164:1227463:1227994 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-103:96888:97423 [0] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-161-103:96888:97423 [0] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-161-103:96888:97423 [0] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-161-103:96888:97423 [0] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-161-103:96888:97423 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-161-103:96888:97423 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-103:96891:97433 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-161-103:96892:97429 [4] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-161-103:96892:97429 [4] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-161-103:96892:97429 [4] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-161-103:96892:97429 [4] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-161-103:96892:97429 [4] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-161-103:96892:97429 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-164:1227467:1227999 [4] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-165-164:1227467:1227999 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-103:96891:97433 [3] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-161-103:96891:97433 [3] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-161-103:96891:97433 [3] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-161-103:96891:97433 [3] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-161-103:96891:97433 [3] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-165-164:1227466:1228001 [3] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-165-164:1227466:1228001 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-103:96891:97433 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-131:1630206:1630784 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-161-103:96893:97431 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-165-202:153567:154093 [5] NCCL INFO Channel 00/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124161:124703 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-165-202:153567:154093 [5] NCCL INFO Channel 01/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153567:154093 [5] NCCL INFO Channel 02/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153567:154093 [5] NCCL INFO Channel 03/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153567:154093 [5] NCCL INFO Channel 00/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153567:154093 [5] NCCL INFO Channel 01/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630212:1630785 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-165-202:153567:154093 [5] NCCL INFO Channel 02/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153567:154093 [5] NCCL INFO Channel 03/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96893:97431 [5] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-161-103:96893:97431 [5] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-161-103:96893:97431 [5] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-161-103:96893:97431 [5] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-161-103:96893:97431 [5] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-161-103:96893:97431 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-164:1227468:1228000 [5] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-165-164:1227468:1228000 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-225:124161:124703 [1] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-160-225:124161:124703 [1] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-160-225:124161:124703 [1] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-160-225:124161:124703 [1] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-160-225:124161:124703 [1] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-160-225:124161:124703 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-78:114513:115048 [1] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-161-78:114513:115048 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-242:1364905:1365456 [3] NCCL INFO Setting affinity for GPU 3 to 0fff,ffffffff
+ip-26-0-160-242:1364903:1365461 [1] NCCL INFO Setting affinity for GPU 1 to 0fff,ffffffff
+ip-26-0-160-242:1364905:1365456 [3] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-160-242:1364905:1365456 [3] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-165-131:1630208:1630782 [3] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-165-131:1630208:1630782 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-242:1364905:1365456 [3] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-160-242:1364905:1365456 [3] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-160-242:1364905:1365456 [3] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-160-242:1364905:1365456 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-123:32709:33243 [5] NCCL INFO Channel 00/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32709:33243 [5] NCCL INFO Channel 01/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32709:33243 [5] NCCL INFO Channel 02/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32709:33243 [5] NCCL INFO Channel 03/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32709:33243 [5] NCCL INFO Channel 00/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32709:33243 [5] NCCL INFO Channel 01/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32709:33243 [5] NCCL INFO Channel 02/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32709:33243 [5] NCCL INFO Channel 03/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630209:1630781 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-160-225:124164:124702 [4] NCCL INFO Channel 00/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:124164:124702 [4] NCCL INFO Channel 01/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:124164:124702 [4] NCCL INFO Channel 02/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:124164:124702 [4] NCCL INFO Channel 03/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:124164:124702 [4] NCCL INFO Channel 00/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:124164:124702 [4] NCCL INFO Channel 01/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:124164:124702 [4] NCCL INFO Channel 02/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:124164:124702 [4] NCCL INFO Channel 03/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364904:1365451 [2] NCCL INFO Setting affinity for GPU 2 to 0fff,ffffffff
+ip-26-0-160-242:1364903:1365461 [1] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-160-242:1364903:1365461 [1] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-160-242:1364903:1365461 [1] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-160-242:1364903:1365461 [1] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-165-131:1630206:1630784 [1] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-165-131:1630206:1630784 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-242:1364903:1365461 [1] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-160-242:1364903:1365461 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-202:153562:154095 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153562:154095 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153562:154095 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153562:154095 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153562:154095 [0] NCCL INFO Channel 00/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153562:154095 [0] NCCL INFO Channel 01/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153562:154095 [0] NCCL INFO Channel 02/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153562:154095 [0] NCCL INFO Channel 03/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364904:1365451 [2] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-160-242:1364904:1365451 [2] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-160-242:1364904:1365451 [2] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-160-242:1364904:1365451 [2] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-160-242:1364904:1365451 [2] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-160-242:1364904:1365451 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-131:1630207:1630780 [2] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-165-131:1630207:1630780 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-78:114516:115044 [4] NCCL INFO Channel 00/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114516:115044 [4] NCCL INFO Channel 01/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114516:115044 [4] NCCL INFO Channel 02/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114516:115044 [4] NCCL INFO Channel 03/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114516:115044 [4] NCCL INFO Channel 00/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114516:115044 [4] NCCL INFO Channel 01/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114516:115044 [4] NCCL INFO Channel 02/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114516:115044 [4] NCCL INFO Channel 03/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32704:33247 [0] NCCL INFO Channel 00/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32704:33247 [0] NCCL INFO Channel 01/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32704:33247 [0] NCCL INFO Channel 02/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32704:33247 [0] NCCL INFO Channel 03/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32704:33247 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364906:1365453 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,fff00000,00000000
+ip-26-0-161-123:32704:33247 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32704:33247 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32704:33247 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364906:1365453 [4] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-160-242:1364906:1365453 [4] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-160-242:1364906:1365453 [4] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-160-242:1364906:1365453 [4] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-165-131:1630205:1630778 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-160-242:1364906:1365453 [4] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-160-242:1364906:1365453 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-131:1630209:1630781 [4] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-165-131:1630209:1630781 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-131:1630210:1630783 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-165-131:1630210:1630783 [5] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-165-131:1630210:1630783 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-242:1364907:1365459 [5] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-160-242:1364907:1365459 [5] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-160-242:1364907:1365459 [5] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-160-242:1364907:1365459 [5] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-160-242:1364907:1365459 [5] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-160-242:1364907:1365459 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-242:1364902:1365449 [0] NCCL INFO Setting affinity for GPU 0 to 0fff,ffffffff
+ip-26-0-160-242:1364902:1365449 [0] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-160-242:1364902:1365449 [0] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-160-242:1364902:1365449 [0] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-160-242:1364902:1365449 [0] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-165-131:1630205:1630778 [0] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-165-131:1630205:1630778 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-242:1364902:1365449 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-160-242:1364902:1365449 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-242:1364909:1365460 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,fff00000,00000000
+ip-26-0-160-242:1364909:1365460 [7] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-160-242:1364909:1365460 [7] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-160-242:1364909:1365460 [7] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-160-242:1364909:1365460 [7] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-165-131:1630212:1630785 [7] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-165-131:1630212:1630785 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-242:1364909:1365460 [7] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-160-242:1364909:1365460 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-78:114519:115047 [7] NCCL INFO Channel 00/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124167:124700 [7] NCCL INFO Channel 00/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114519:115047 [7] NCCL INFO Channel 01/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124167:124700 [7] NCCL INFO Channel 01/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114519:115047 [7] NCCL INFO Channel 02/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124167:124700 [7] NCCL INFO Channel 02/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114519:115047 [7] NCCL INFO Channel 03/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124167:124700 [7] NCCL INFO Channel 03/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124167:124700 [7] NCCL INFO Channel 00/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114519:115047 [7] NCCL INFO Channel 00/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114519:115047 [7] NCCL INFO Channel 01/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124167:124700 [7] NCCL INFO Channel 01/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114519:115047 [7] NCCL INFO Channel 02/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124167:124700 [7] NCCL INFO Channel 02/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114519:115047 [7] NCCL INFO Channel 03/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124167:124700 [7] NCCL INFO Channel 03/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227469:1227995 [6] NCCL INFO Channel 00/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227469:1227995 [6] NCCL INFO Channel 01/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227469:1227995 [6] NCCL INFO Channel 02/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227469:1227995 [6] NCCL INFO Channel 03/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227469:1227995 [6] NCCL INFO Channel 00/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227469:1227995 [6] NCCL INFO Channel 01/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227469:1227995 [6] NCCL INFO Channel 02/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227469:1227995 [6] NCCL INFO Channel 03/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96894:97424 [6] NCCL INFO Channel 00/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96894:97424 [6] NCCL INFO Channel 01/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96894:97424 [6] NCCL INFO Channel 02/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96894:97424 [6] NCCL INFO Channel 03/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96894:97424 [6] NCCL INFO Channel 00/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96894:97424 [6] NCCL INFO Channel 01/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96894:97424 [6] NCCL INFO Channel 02/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96894:97424 [6] NCCL INFO Channel 03/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114514:115049 [2] NCCL INFO Channel 00/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114514:115049 [2] NCCL INFO Channel 01/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114514:115049 [2] NCCL INFO Channel 02/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114514:115049 [2] NCCL INFO Channel 03/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114514:115049 [2] NCCL INFO Channel 00/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114514:115049 [2] NCCL INFO Channel 01/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114514:115049 [2] NCCL INFO Channel 02/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114514:115049 [2] NCCL INFO Channel 03/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:124162:124705 [2] NCCL INFO Channel 00/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:124162:124705 [2] NCCL INFO Channel 01/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:124162:124705 [2] NCCL INFO Channel 02/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:124162:124705 [2] NCCL INFO Channel 03/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:124162:124705 [2] NCCL INFO Channel 00/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:124162:124705 [2] NCCL INFO Channel 01/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:124162:124705 [2] NCCL INFO Channel 02/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:124162:124705 [2] NCCL INFO Channel 03/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114517:115046 [5] NCCL INFO Channel 00/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114517:115046 [5] NCCL INFO Channel 01/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114517:115046 [5] NCCL INFO Channel 02/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114517:115046 [5] NCCL INFO Channel 03/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114517:115046 [5] NCCL INFO Channel 00/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114517:115046 [5] NCCL INFO Channel 01/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114517:115046 [5] NCCL INFO Channel 02/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114517:115046 [5] NCCL INFO Channel 03/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124165:124699 [5] NCCL INFO Channel 00/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124165:124699 [5] NCCL INFO Channel 01/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124165:124699 [5] NCCL INFO Channel 02/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124165:124699 [5] NCCL INFO Channel 03/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124165:124699 [5] NCCL INFO Channel 00/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124165:124699 [5] NCCL INFO Channel 01/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124165:124699 [5] NCCL INFO Channel 02/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124165:124699 [5] NCCL INFO Channel 03/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364908:1365452 [6] NCCL INFO Channel 00/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364908:1365452 [6] NCCL INFO Channel 01/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364908:1365452 [6] NCCL INFO Channel 02/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364908:1365452 [6] NCCL INFO Channel 03/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364908:1365452 [6] NCCL INFO Channel 00/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364908:1365452 [6] NCCL INFO Channel 01/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364908:1365452 [6] NCCL INFO Channel 02/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364908:1365452 [6] NCCL INFO Channel 03/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630211:1630779 [6] NCCL INFO Channel 00/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630211:1630779 [6] NCCL INFO Channel 01/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:124166:124708 [6] NCCL INFO Channel 00/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630211:1630779 [6] NCCL INFO Channel 02/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:124166:124708 [6] NCCL INFO Channel 01/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630211:1630779 [6] NCCL INFO Channel 03/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:124166:124708 [6] NCCL INFO Channel 02/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630211:1630779 [6] NCCL INFO Channel 00/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:124166:124708 [6] NCCL INFO Channel 03/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630211:1630779 [6] NCCL INFO Channel 01/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630211:1630779 [6] NCCL INFO Channel 02/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:124166:124708 [6] NCCL INFO Channel 00/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630211:1630779 [6] NCCL INFO Channel 03/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:124166:124708 [6] NCCL INFO Channel 01/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:124166:124708 [6] NCCL INFO Channel 02/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:124166:124708 [6] NCCL INFO Channel 03/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114518:115051 [6] NCCL INFO Channel 00/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114518:115051 [6] NCCL INFO Channel 01/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114518:115051 [6] NCCL INFO Channel 02/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114518:115051 [6] NCCL INFO Channel 03/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114518:115051 [6] NCCL INFO Channel 00/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114518:115051 [6] NCCL INFO Channel 01/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114518:115051 [6] NCCL INFO Channel 02/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114518:115051 [6] NCCL INFO Channel 03/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96890:97425 [2] NCCL INFO Channel 00/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96890:97425 [2] NCCL INFO Channel 01/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96895:97427 [7] NCCL INFO Channel 00/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96890:97425 [2] NCCL INFO Channel 02/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96890:97425 [2] NCCL INFO Channel 03/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96890:97425 [2] NCCL INFO Channel 00/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96895:97427 [7] NCCL INFO Channel 01/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96890:97425 [2] NCCL INFO Channel 01/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96895:97427 [7] NCCL INFO Channel 02/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227470:1227998 [7] NCCL INFO Channel 00/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96890:97425 [2] NCCL INFO Channel 02/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96895:97427 [7] NCCL INFO Channel 03/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227470:1227998 [7] NCCL INFO Channel 01/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96890:97425 [2] NCCL INFO Channel 03/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96895:97427 [7] NCCL INFO Channel 00/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227470:1227998 [7] NCCL INFO Channel 02/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96895:97427 [7] NCCL INFO Channel 01/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96895:97427 [7] NCCL INFO Channel 02/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227465:1227996 [2] NCCL INFO Channel 00/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227470:1227998 [7] NCCL INFO Channel 03/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96895:97427 [7] NCCL INFO Channel 03/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227464:1227997 [1] NCCL INFO Channel 00/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227470:1227998 [7] NCCL INFO Channel 00/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227465:1227996 [2] NCCL INFO Channel 01/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227464:1227997 [1] NCCL INFO Channel 01/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227470:1227998 [7] NCCL INFO Channel 01/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227465:1227996 [2] NCCL INFO Channel 02/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227470:1227998 [7] NCCL INFO Channel 02/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227464:1227997 [1] NCCL INFO Channel 02/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227465:1227996 [2] NCCL INFO Channel 03/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227470:1227998 [7] NCCL INFO Channel 03/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227464:1227997 [1] NCCL INFO Channel 03/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227465:1227996 [2] NCCL INFO Channel 00/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227464:1227997 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227465:1227996 [2] NCCL INFO Channel 01/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227465:1227996 [2] NCCL INFO Channel 02/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:124163:124706 [3] NCCL INFO Channel 00/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227464:1227997 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227465:1227996 [2] NCCL INFO Channel 03/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:124163:124706 [3] NCCL INFO Channel 01/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227464:1227997 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227464:1227997 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124163:124706 [3] NCCL INFO Channel 02/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96889:97426 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124163:124706 [3] NCCL INFO Channel 03/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:124163:124706 [3] NCCL INFO Channel 00/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96889:97426 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124163:124706 [3] NCCL INFO Channel 01/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96889:97426 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124163:124706 [3] NCCL INFO Channel 02/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96889:97426 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124163:124706 [3] NCCL INFO Channel 03/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96889:97426 [1] NCCL INFO Channel 00/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96889:97426 [1] NCCL INFO Channel 01/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96889:97426 [1] NCCL INFO Channel 02/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96889:97426 [1] NCCL INFO Channel 03/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96892:97429 [4] NCCL INFO Channel 00/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96892:97429 [4] NCCL INFO Channel 01/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96892:97429 [4] NCCL INFO Channel 02/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96892:97429 [4] NCCL INFO Channel 03/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96892:97429 [4] NCCL INFO Channel 00/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96892:97429 [4] NCCL INFO Channel 01/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96892:97429 [4] NCCL INFO Channel 02/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114515:115050 [3] NCCL INFO Channel 00/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96892:97429 [4] NCCL INFO Channel 03/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114515:115050 [3] NCCL INFO Channel 01/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114515:115050 [3] NCCL INFO Channel 02/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227466:1228001 [3] NCCL INFO Channel 00/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114515:115050 [3] NCCL INFO Channel 03/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227466:1228001 [3] NCCL INFO Channel 01/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114515:115050 [3] NCCL INFO Channel 00/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227466:1228001 [3] NCCL INFO Channel 02/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114515:115050 [3] NCCL INFO Channel 01/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96888:97423 [0] NCCL INFO Channel 00/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227466:1228001 [3] NCCL INFO Channel 03/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114515:115050 [3] NCCL INFO Channel 02/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96888:97423 [0] NCCL INFO Channel 01/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227466:1228001 [3] NCCL INFO Channel 00/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114515:115050 [3] NCCL INFO Channel 03/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96888:97423 [0] NCCL INFO Channel 02/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227466:1228001 [3] NCCL INFO Channel 01/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96888:97423 [0] NCCL INFO Channel 03/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227466:1228001 [3] NCCL INFO Channel 02/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227466:1228001 [3] NCCL INFO Channel 03/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96888:97423 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96888:97423 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96888:97423 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96888:97423 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96891:97433 [3] NCCL INFO Channel 00/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227463:1227994 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96891:97433 [3] NCCL INFO Channel 01/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96891:97433 [3] NCCL INFO Channel 02/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227463:1227994 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96891:97433 [3] NCCL INFO Channel 03/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227463:1227994 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96891:97433 [3] NCCL INFO Channel 00/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227463:1227994 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96891:97433 [3] NCCL INFO Channel 01/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227463:1227994 [0] NCCL INFO Channel 00/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96891:97433 [3] NCCL INFO Channel 02/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227463:1227994 [0] NCCL INFO Channel 01/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96891:97433 [3] NCCL INFO Channel 03/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227463:1227994 [0] NCCL INFO Channel 02/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227463:1227994 [0] NCCL INFO Channel 03/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227467:1227999 [4] NCCL INFO Channel 00/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227467:1227999 [4] NCCL INFO Channel 01/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227467:1227999 [4] NCCL INFO Channel 02/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227467:1227999 [4] NCCL INFO Channel 03/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227467:1227999 [4] NCCL INFO Channel 00/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227467:1227999 [4] NCCL INFO Channel 01/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227467:1227999 [4] NCCL INFO Channel 02/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227467:1227999 [4] NCCL INFO Channel 03/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227468:1228000 [5] NCCL INFO Channel 00/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96893:97431 [5] NCCL INFO Channel 00/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227468:1228000 [5] NCCL INFO Channel 01/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96893:97431 [5] NCCL INFO Channel 01/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96893:97431 [5] NCCL INFO Channel 02/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227468:1228000 [5] NCCL INFO Channel 02/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96893:97431 [5] NCCL INFO Channel 03/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227468:1228000 [5] NCCL INFO Channel 03/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96893:97431 [5] NCCL INFO Channel 00/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227468:1228000 [5] NCCL INFO Channel 00/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96893:97431 [5] NCCL INFO Channel 01/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227468:1228000 [5] NCCL INFO Channel 01/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96893:97431 [5] NCCL INFO Channel 02/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227468:1228000 [5] NCCL INFO Channel 02/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96893:97431 [5] NCCL INFO Channel 03/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227468:1228000 [5] NCCL INFO Channel 03/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630208:1630782 [3] NCCL INFO Channel 00/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364905:1365456 [3] NCCL INFO Channel 00/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630208:1630782 [3] NCCL INFO Channel 01/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364905:1365456 [3] NCCL INFO Channel 01/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630208:1630782 [3] NCCL INFO Channel 02/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114513:115048 [1] NCCL INFO Channel 00/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364905:1365456 [3] NCCL INFO Channel 02/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630208:1630782 [3] NCCL INFO Channel 03/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364905:1365456 [3] NCCL INFO Channel 03/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630208:1630782 [3] NCCL INFO Channel 00/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114513:115048 [1] NCCL INFO Channel 01/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364905:1365456 [3] NCCL INFO Channel 00/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630208:1630782 [3] NCCL INFO Channel 01/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114513:115048 [1] NCCL INFO Channel 02/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364905:1365456 [3] NCCL INFO Channel 01/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630208:1630782 [3] NCCL INFO Channel 02/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114513:115048 [1] NCCL INFO Channel 03/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364905:1365456 [3] NCCL INFO Channel 02/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630208:1630782 [3] NCCL INFO Channel 03/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114513:115048 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364905:1365456 [3] NCCL INFO Channel 03/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114513:115048 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114513:115048 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114513:115048 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124161:124703 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124161:124703 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124161:124703 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124161:124703 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124161:124703 [1] NCCL INFO Channel 00/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124161:124703 [1] NCCL INFO Channel 01/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124161:124703 [1] NCCL INFO Channel 02/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124161:124703 [1] NCCL INFO Channel 03/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630206:1630784 [1] NCCL INFO Channel 00/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630206:1630784 [1] NCCL INFO Channel 01/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630206:1630784 [1] NCCL INFO Channel 02/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630206:1630784 [1] NCCL INFO Channel 03/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630206:1630784 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630206:1630784 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630206:1630784 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630206:1630784 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364903:1365461 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364904:1365451 [2] NCCL INFO Channel 00/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364903:1365461 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364904:1365451 [2] NCCL INFO Channel 01/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364903:1365461 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630207:1630780 [2] NCCL INFO Channel 00/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364904:1365451 [2] NCCL INFO Channel 02/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364903:1365461 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364904:1365451 [2] NCCL INFO Channel 03/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364903:1365461 [1] NCCL INFO Channel 00/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630207:1630780 [2] NCCL INFO Channel 01/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364904:1365451 [2] NCCL INFO Channel 00/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364903:1365461 [1] NCCL INFO Channel 01/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630207:1630780 [2] NCCL INFO Channel 02/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364904:1365451 [2] NCCL INFO Channel 01/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364903:1365461 [1] NCCL INFO Channel 02/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364904:1365451 [2] NCCL INFO Channel 02/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630207:1630780 [2] NCCL INFO Channel 03/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364903:1365461 [1] NCCL INFO Channel 03/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364904:1365451 [2] NCCL INFO Channel 03/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630207:1630780 [2] NCCL INFO Channel 00/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630207:1630780 [2] NCCL INFO Channel 01/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630207:1630780 [2] NCCL INFO Channel 02/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630207:1630780 [2] NCCL INFO Channel 03/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630209:1630781 [4] NCCL INFO Channel 00/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630209:1630781 [4] NCCL INFO Channel 01/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630209:1630781 [4] NCCL INFO Channel 02/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630209:1630781 [4] NCCL INFO Channel 03/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630209:1630781 [4] NCCL INFO Channel 00/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630209:1630781 [4] NCCL INFO Channel 01/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630209:1630781 [4] NCCL INFO Channel 02/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630209:1630781 [4] NCCL INFO Channel 03/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364906:1365453 [4] NCCL INFO Channel 00/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364906:1365453 [4] NCCL INFO Channel 01/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364906:1365453 [4] NCCL INFO Channel 02/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364906:1365453 [4] NCCL INFO Channel 03/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364906:1365453 [4] NCCL INFO Channel 00/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364906:1365453 [4] NCCL INFO Channel 01/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364906:1365453 [4] NCCL INFO Channel 02/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364906:1365453 [4] NCCL INFO Channel 03/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364907:1365459 [5] NCCL INFO Channel 00/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364907:1365459 [5] NCCL INFO Channel 01/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364907:1365459 [5] NCCL INFO Channel 02/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630210:1630783 [5] NCCL INFO Channel 00/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364907:1365459 [5] NCCL INFO Channel 03/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630210:1630783 [5] NCCL INFO Channel 01/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364907:1365459 [5] NCCL INFO Channel 00/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630210:1630783 [5] NCCL INFO Channel 02/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364907:1365459 [5] NCCL INFO Channel 01/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630210:1630783 [5] NCCL INFO Channel 03/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364907:1365459 [5] NCCL INFO Channel 02/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630210:1630783 [5] NCCL INFO Channel 00/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364907:1365459 [5] NCCL INFO Channel 03/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630210:1630783 [5] NCCL INFO Channel 01/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630210:1630783 [5] NCCL INFO Channel 02/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630210:1630783 [5] NCCL INFO Channel 03/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630205:1630778 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630205:1630778 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630205:1630778 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630205:1630778 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630205:1630778 [0] NCCL INFO Channel 00/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630205:1630778 [0] NCCL INFO Channel 01/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630205:1630778 [0] NCCL INFO Channel 02/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630205:1630778 [0] NCCL INFO Channel 03/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364902:1365449 [0] NCCL INFO Channel 00/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364902:1365449 [0] NCCL INFO Channel 01/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364902:1365449 [0] NCCL INFO Channel 02/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364902:1365449 [0] NCCL INFO Channel 03/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364902:1365449 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630212:1630785 [7] NCCL INFO Channel 00/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364902:1365449 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364902:1365449 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630212:1630785 [7] NCCL INFO Channel 01/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364902:1365449 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630212:1630785 [7] NCCL INFO Channel 02/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630212:1630785 [7] NCCL INFO Channel 03/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630212:1630785 [7] NCCL INFO Channel 00/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630212:1630785 [7] NCCL INFO Channel 01/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630212:1630785 [7] NCCL INFO Channel 02/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630212:1630785 [7] NCCL INFO Channel 03/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364909:1365460 [7] NCCL INFO Channel 00/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364909:1365460 [7] NCCL INFO Channel 01/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364909:1365460 [7] NCCL INFO Channel 02/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364909:1365460 [7] NCCL INFO Channel 03/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364909:1365460 [7] NCCL INFO Channel 00/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364909:1365460 [7] NCCL INFO Channel 01/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364909:1365460 [7] NCCL INFO Channel 02/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364909:1365460 [7] NCCL INFO Channel 03/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815968:2816533 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,fff00000,00000000
+ip-26-0-161-178:2815967:2816521 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,fff00000,00000000
+ip-26-0-165-59:1557272:1557800 [6] NCCL INFO Connected all rings
+ip-26-0-165-59:1557272:1557800 [6] NCCL INFO Connected all trees
+ip-26-0-165-59:1557272:1557800 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-59:1557272:1557800 [6] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-165-59:1557272:1557800 [6] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-168-52:2722045:2722584 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-161-178:2815970:2816523 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,fff00000,00000000
+ip-26-0-168-52:2722048:2722581 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-165-59:1557272:1557800 [6] NCCL INFO comm 0xb431a90 rank 1 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0x4e9bc1b782930120 - Init COMPLETE
+ip-26-0-165-59:1557268:1557796 [2] NCCL INFO Connected all rings
+ip-26-0-165-59:1557268:1557796 [2] NCCL INFO Connected all trees
+ip-26-0-165-59:1557268:1557796 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-59:1557268:1557796 [2] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-165-59:1557268:1557796 [2] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-161-178:2815964:2816524 [1] NCCL INFO Setting affinity for GPU 1 to 0fff,ffffffff
+ip-26-0-161-138:96569:97147 [6] NCCL INFO Connected all rings
+ip-26-0-161-138:96569:97147 [6] NCCL INFO Connected all trees
+ip-26-0-161-138:96569:97147 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-138:96569:97147 [6] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-161-138:96569:97147 [6] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-161-138:96565:97145 [2] NCCL INFO Connected all rings
+ip-26-0-161-138:96565:97145 [2] NCCL INFO Connected all trees
+ip-26-0-161-138:96565:97145 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-138:96565:97145 [2] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-161-138:96565:97145 [2] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-168-52:2722051:2722583 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-165-59:1557268:1557796 [2] NCCL INFO comm 0xa11a8f0 rank 1 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0x9392ca3f181752f6 - Init COMPLETE
+ip-26-0-161-178:2815966:2816526 [3] NCCL INFO Setting affinity for GPU 3 to 0fff,ffffffff
+ip-26-0-168-52:2722046:2722580 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-161-138:96569:97147 [6] NCCL INFO comm 0xb09d550 rank 0 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0x4e9bc1b782930120 - Init COMPLETE
+ip-26-0-168-52:2722046:2722580 [1] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-168-52:2722046:2722580 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-178:2815964:2816524 [1] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-161-178:2815964:2816524 [1] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-161-178:2815964:2816524 [1] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-161-178:2815964:2816524 [1] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-161-178:2815964:2816524 [1] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-161-178:2815964:2816524 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-168-52:2722047:2722582 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-161-178:2815966:2816526 [3] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-161-178:2815966:2816526 [3] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-161-178:2815966:2816526 [3] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-168-52:2722048:2722581 [3] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-168-52:2722048:2722581 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-178:2815966:2816526 [3] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-161-178:2815966:2816526 [3] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-161-178:2815966:2816526 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-138:96565:97145 [2] NCCL INFO comm 0xb9305e0 rank 0 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0x9392ca3f181752f6 - Init COMPLETE
+ip-26-0-165-59:1557267:1557799 [1] NCCL INFO Connected all rings
+ip-26-0-165-59:1557267:1557799 [1] NCCL INFO Connected all trees
+ip-26-0-165-59:1557267:1557799 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-59:1557267:1557799 [1] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-165-59:1557267:1557799 [1] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-161-138:96564:97149 [1] NCCL INFO Connected all rings
+ip-26-0-161-138:96564:97149 [1] NCCL INFO Connected all trees
+ip-26-0-161-138:96564:97149 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-138:96564:97149 [1] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-161-138:96564:97149 [1] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-161-178:2815969:2816532 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,fff00000,00000000
+ip-26-0-161-178:2815969:2816532 [6] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-161-178:2815969:2816532 [6] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-161-178:2815969:2816532 [6] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-168-52:2722051:2722583 [6] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-168-52:2722051:2722583 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-178:2815969:2816532 [6] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-161-178:2815969:2816532 [6] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-161-178:2815969:2816532 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-168-52:2722050:2722585 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-168-52:2722052:2722578 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-168-52:2722050:2722585 [5] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-168-52:2722050:2722585 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-178:2815968:2816533 [5] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-161-178:2815968:2816533 [5] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-161-178:2815968:2816533 [5] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-161-178:2815968:2816533 [5] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-161-178:2815968:2816533 [5] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-161-178:2815968:2816533 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-168-52:2722049:2722579 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-168-52:2722052:2722578 [7] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-168-52:2722052:2722578 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-178:2815970:2816523 [7] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-161-178:2815970:2816523 [7] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-161-178:2815970:2816523 [7] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-161-178:2815970:2816523 [7] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-161-178:2815970:2816523 [7] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-161-178:2815970:2816523 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-178:2815963:2816530 [0] NCCL INFO Setting affinity for GPU 0 to 0fff,ffffffff
+ip-26-0-168-52:2722049:2722579 [4] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-168-52:2722049:2722579 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-178:2815967:2816521 [4] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-161-178:2815967:2816521 [4] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-161-178:2815967:2816521 [4] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-161-178:2815967:2816521 [4] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-161-178:2815967:2816521 [4] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-161-178:2815967:2816521 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-178:2815963:2816530 [0] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-161-178:2815963:2816530 [0] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-161-178:2815963:2816530 [0] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-161-178:2815963:2816530 [0] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-161-178:2815963:2816530 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-161-178:2815963:2816530 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-168-52:2722045:2722584 [0] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-168-52:2722045:2722584 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-59:1557267:1557799 [1] NCCL INFO comm 0xa8dce00 rank 1 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0x6aa89d25d313e4b5 - Init COMPLETE
+ip-26-0-161-142:62887:63423 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-161-178:2815965:2816528 [2] NCCL INFO Setting affinity for GPU 2 to 0fff,ffffffff
+ip-26-0-168-30:235654:236182 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-168-30:235654:236182 [6] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-168-30:235654:236182 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-142:62887:63423 [6] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-161-142:62887:63423 [6] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-161-142:62887:63423 [6] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-161-142:62887:63423 [6] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-161-142:62887:63423 [6] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-161-142:62887:63423 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-178:2815965:2816528 [2] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-161-178:2815965:2816528 [2] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-161-178:2815965:2816528 [2] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-161-178:2815965:2816528 [2] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-168-52:2722047:2722582 [2] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-168-52:2722047:2722582 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-178:2815965:2816528 [2] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-161-178:2815965:2816528 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-225:124160:124698 [0] NCCL INFO Connected all rings
+ip-26-0-160-225:124160:124698 [0] NCCL INFO Connected all trees
+ip-26-0-160-225:124160:124698 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-225:124160:124698 [0] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-160-225:124160:124698 [0] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-161-138:96564:97149 [1] NCCL INFO comm 0xb824ca0 rank 0 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0x6aa89d25d313e4b5 - Init COMPLETE
+ip-26-0-161-78:114512:115045 [0] NCCL INFO Connected all rings
+ip-26-0-161-78:114512:115045 [0] NCCL INFO Connected all trees
+ip-26-0-161-78:114512:115045 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-78:114512:115045 [0] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-161-78:114512:115045 [0] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-165-59:1557270:1557797 [4] NCCL INFO Connected all rings
+ip-26-0-165-59:1557270:1557797 [4] NCCL INFO Connected all trees
+ip-26-0-165-59:1557270:1557797 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-59:1557270:1557797 [4] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-165-59:1557270:1557797 [4] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-165-59:1557269:1557802 [3] NCCL INFO Connected all rings
+ip-26-0-165-59:1557269:1557802 [3] NCCL INFO Connected all trees
+ip-26-0-165-59:1557269:1557802 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-59:1557269:1557802 [3] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-165-59:1557269:1557802 [3] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-168-30:235651:236178 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-160-225:124160:124698 [0] NCCL INFO comm 0xa9dbd90 rank 0 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0x145a7e479ca528e9 - Init COMPLETE
+ip-26-0-165-59:1557271:1557795 [5] NCCL INFO Connected all rings
+ip-26-0-165-59:1557271:1557795 [5] NCCL INFO Connected all trees
+ip-26-0-165-59:1557271:1557795 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-59:1557271:1557795 [5] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-165-59:1557271:1557795 [5] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-161-78:114512:115045 [0] NCCL INFO comm 0xb2c2490 rank 1 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0x145a7e479ca528e9 - Init COMPLETE
+ip-26-0-165-59:1557273:1557801 [7] NCCL INFO Connected all rings
+ip-26-0-165-59:1557273:1557801 [7] NCCL INFO Connected all trees
+ip-26-0-165-59:1557273:1557801 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-59:1557273:1557801 [7] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-165-59:1557273:1557801 [7] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-165-59:1557266:1557798 [0] NCCL INFO Connected all rings
+ip-26-0-165-59:1557266:1557798 [0] NCCL INFO Connected all trees
+ip-26-0-165-59:1557266:1557798 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-59:1557266:1557798 [0] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-165-59:1557266:1557798 [0] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-161-138:96566:97151 [3] NCCL INFO Connected all rings
+ip-26-0-161-138:96566:97151 [3] NCCL INFO Connected all trees
+ip-26-0-161-138:96566:97151 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-138:96566:97151 [3] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-161-138:96566:97151 [3] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-168-30:235655:236181 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-165-59:1557270:1557797 [4] NCCL INFO comm 0x9eef930 rank 1 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0xb4e67d13f14ca523 - Init COMPLETE
+ip-26-0-165-59:1557269:1557802 [3] NCCL INFO comm 0xbf320c0 rank 1 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0xf91b029843677563 - Init COMPLETE
+ip-26-0-161-142:62883:63421 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-165-59:1557271:1557795 [5] NCCL INFO comm 0x9a81660 rank 1 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0x901e509d3ac79cef - Init COMPLETE
+ip-26-0-161-138:96570:97146 [7] NCCL INFO Connected all rings
+ip-26-0-161-138:96570:97146 [7] NCCL INFO Connected all trees
+ip-26-0-161-138:96570:97146 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-138:96570:97146 [7] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-161-138:96570:97146 [7] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-165-59:1557273:1557801 [7] NCCL INFO comm 0xc02c5d0 rank 1 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0xc5a3cb775d798355 - Init COMPLETE
+ip-26-0-161-138:96567:97144 [4] NCCL INFO Connected all rings
+ip-26-0-161-138:96567:97144 [4] NCCL INFO Connected all trees
+ip-26-0-161-138:96567:97144 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-138:96567:97144 [4] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-161-138:96567:97144 [4] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-168-30:235649:236184 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-161-138:96566:97151 [3] NCCL INFO comm 0xaac4830 rank 0 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0xf91b029843677563 - Init COMPLETE
+ip-26-0-165-59:1557266:1557798 [0] NCCL INFO comm 0xc163230 rank 1 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0xd6a2a23e4b7c8957 - Init COMPLETE
+ip-26-0-161-138:96568:97143 [5] NCCL INFO Connected all rings
+ip-26-0-161-138:96568:97143 [5] NCCL INFO Connected all trees
+ip-26-0-161-138:96568:97143 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-138:96568:97143 [5] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-161-138:96568:97143 [5] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-161-142:62888:63422 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-161-138:96563:97148 [0] NCCL INFO Connected all rings
+ip-26-0-161-138:96563:97148 [0] NCCL INFO Connected all trees
+ip-26-0-161-138:96563:97148 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-138:96563:97148 [0] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-161-138:96563:97148 [0] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-161-142:62888:63422 [7] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-161-142:62888:63422 [7] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-161-142:62888:63422 [7] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-161-142:62888:63422 [7] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-168-30:235655:236181 [7] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-168-30:235655:236181 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-142:62888:63422 [7] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-161-142:62888:63422 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-168-30:235650:236180 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-168-30:235650:236180 [2] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-168-30:235650:236180 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-142:62883:63421 [2] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-161-142:62883:63421 [2] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-161-142:62883:63421 [2] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-161-142:62883:63421 [2] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-161-142:62883:63421 [2] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-161-142:62883:63421 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-138:96570:97146 [7] NCCL INFO comm 0xc2c1fb0 rank 0 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0xc5a3cb775d798355 - Init COMPLETE
+ip-26-0-161-142:62882:63424 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-161-138:96567:97144 [4] NCCL INFO comm 0xaa6d9d0 rank 0 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0xb4e67d13f14ca523 - Init COMPLETE
+ip-26-0-168-30:235653:236177 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-161-142:62882:63424 [1] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-161-142:62882:63424 [1] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-161-142:62882:63424 [1] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-161-142:62882:63424 [1] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-161-142:62882:63424 [1] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-161-142:62882:63424 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-168-30:235649:236184 [1] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-168-30:235649:236184 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-138:96568:97143 [5] NCCL INFO comm 0xaaf0f50 rank 0 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0x901e509d3ac79cef - Init COMPLETE
+ip-26-0-161-123:32706:33245 [2] NCCL INFO Connected all rings
+ip-26-0-161-123:32706:33245 [2] NCCL INFO Connected all trees
+ip-26-0-161-123:32706:33245 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-123:32706:33245 [2] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-161-123:32706:33245 [2] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-161-138:96563:97148 [0] NCCL INFO comm 0xbbb44c0 rank 0 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0xd6a2a23e4b7c8957 - Init COMPLETE
+ip-26-0-161-78:114516:115044 [4] NCCL INFO Connected all rings
+ip-26-0-161-78:114516:115044 [4] NCCL INFO Connected all trees
+ip-26-0-161-78:114516:115044 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-78:114516:115044 [4] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-161-78:114516:115044 [4] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-168-30:235648:236179 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-161-142:62884:63417 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-160-225:124164:124702 [4] NCCL INFO Connected all rings
+ip-26-0-160-225:124164:124702 [4] NCCL INFO Connected all trees
+ip-26-0-160-225:124164:124702 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-225:124164:124702 [4] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-160-225:124164:124702 [4] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-161-142:62884:63417 [3] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-161-142:62884:63417 [3] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-161-142:62884:63417 [3] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-161-142:62884:63417 [3] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-161-142:62884:63417 [3] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-161-142:62884:63417 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-168-30:235651:236178 [3] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-168-30:235651:236178 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-168-30:235652:236183 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-161-123:32705:33248 [1] NCCL INFO Connected all rings
+ip-26-0-161-123:32705:33248 [1] NCCL INFO Connected all trees
+ip-26-0-161-123:32705:33248 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-123:32705:33248 [1] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-161-123:32705:33248 [1] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-168-52:2722046:2722580 [1] NCCL INFO Channel 00/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-52:2722046:2722580 [1] NCCL INFO Channel 01/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-52:2722046:2722580 [1] NCCL INFO Channel 02/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-52:2722046:2722580 [1] NCCL INFO Channel 03/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-52:2722046:2722580 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815966:2816526 [3] NCCL INFO Channel 00/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-52:2722046:2722580 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-52:2722046:2722580 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815966:2816526 [3] NCCL INFO Channel 01/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-52:2722046:2722580 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815966:2816526 [3] NCCL INFO Channel 02/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815966:2816526 [3] NCCL INFO Channel 03/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815966:2816526 [3] NCCL INFO Channel 00/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815966:2816526 [3] NCCL INFO Channel 01/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815966:2816526 [3] NCCL INFO Channel 02/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815966:2816526 [3] NCCL INFO Channel 03/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-52:2722048:2722581 [3] NCCL INFO Channel 00/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-52:2722048:2722581 [3] NCCL INFO Channel 01/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-52:2722048:2722581 [3] NCCL INFO Channel 02/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-52:2722048:2722581 [3] NCCL INFO Channel 03/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-52:2722048:2722581 [3] NCCL INFO Channel 00/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-52:2722048:2722581 [3] NCCL INFO Channel 01/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-52:2722048:2722581 [3] NCCL INFO Channel 02/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-52:2722048:2722581 [3] NCCL INFO Channel 03/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32706:33245 [2] NCCL INFO comm 0xb1c7ca0 rank 0 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0x5e0b17ed00861d01 - Init COMPLETE
+ip-26-0-161-178:2815964:2816524 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815964:2816524 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815964:2816524 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815964:2816524 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815964:2816524 [1] NCCL INFO Channel 00/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114516:115044 [4] NCCL INFO comm 0xa945970 rank 1 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0x72d1a31dbaf1f836 - Init COMPLETE
+ip-26-0-161-178:2815964:2816524 [1] NCCL INFO Channel 01/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815964:2816524 [1] NCCL INFO Channel 02/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815964:2816524 [1] NCCL INFO Channel 03/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62881:63420 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-161-142:62881:63420 [0] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-161-142:62881:63420 [0] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-161-142:62881:63420 [0] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-161-142:62881:63420 [0] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-161-142:62881:63420 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-161-142:62881:63420 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-168-30:235648:236179 [0] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-168-30:235648:236179 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-225:124164:124702 [4] NCCL INFO comm 0xb185300 rank 0 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0x72d1a31dbaf1f836 - Init COMPLETE
+ip-26-0-161-142:62886:63411 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-161-142:62886:63411 [5] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-161-142:62886:63411 [5] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-161-142:62886:63411 [5] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-161-142:62886:63411 [5] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-168-30:235653:236177 [5] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-168-30:235653:236177 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-142:62886:63411 [5] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-161-142:62886:63411 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-123:32705:33248 [1] NCCL INFO comm 0xbbec3d0 rank 0 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0xe9de758c5f73d128 - Init COMPLETE
+ip-26-0-161-142:62885:63425 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-161-178:2815968:2816533 [5] NCCL INFO Channel 00/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815968:2816533 [5] NCCL INFO Channel 01/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815968:2816533 [5] NCCL INFO Channel 02/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815968:2816533 [5] NCCL INFO Channel 03/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815969:2816532 [6] NCCL INFO Channel 00/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815968:2816533 [5] NCCL INFO Channel 00/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815969:2816532 [6] NCCL INFO Channel 01/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815968:2816533 [5] NCCL INFO Channel 01/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815969:2816532 [6] NCCL INFO Channel 02/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815968:2816533 [5] NCCL INFO Channel 02/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815969:2816532 [6] NCCL INFO Channel 03/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815968:2816533 [5] NCCL INFO Channel 03/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815969:2816532 [6] NCCL INFO Channel 00/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815969:2816532 [6] NCCL INFO Channel 01/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815969:2816532 [6] NCCL INFO Channel 02/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815969:2816532 [6] NCCL INFO Channel 03/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-52:2722050:2722585 [5] NCCL INFO Channel 00/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815970:2816523 [7] NCCL INFO Channel 00/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722050:2722585 [5] NCCL INFO Channel 01/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62885:63425 [4] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-161-142:62885:63425 [4] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-161-142:62885:63425 [4] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-161-142:62885:63425 [4] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-161-142:62885:63425 [4] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-161-142:62885:63425 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-168-30:235652:236183 [4] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-168-30:235652:236183 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-168-52:2722050:2722585 [5] NCCL INFO Channel 02/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815970:2816523 [7] NCCL INFO Channel 01/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722050:2722585 [5] NCCL INFO Channel 03/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815970:2816523 [7] NCCL INFO Channel 02/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722050:2722585 [5] NCCL INFO Channel 00/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815970:2816523 [7] NCCL INFO Channel 03/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722050:2722585 [5] NCCL INFO Channel 01/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815967:2816521 [4] NCCL INFO Channel 00/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722050:2722585 [5] NCCL INFO Channel 02/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815970:2816523 [7] NCCL INFO Channel 00/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815970:2816523 [7] NCCL INFO Channel 01/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722050:2722585 [5] NCCL INFO Channel 03/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815967:2816521 [4] NCCL INFO Channel 01/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815970:2816523 [7] NCCL INFO Channel 02/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815967:2816521 [4] NCCL INFO Channel 02/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815970:2816523 [7] NCCL INFO Channel 03/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815967:2816521 [4] NCCL INFO Channel 03/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815967:2816521 [4] NCCL INFO Channel 00/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815967:2816521 [4] NCCL INFO Channel 01/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815967:2816521 [4] NCCL INFO Channel 02/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815967:2816521 [4] NCCL INFO Channel 03/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32708:33242 [4] NCCL INFO Connected all rings
+ip-26-0-161-123:32708:33242 [4] NCCL INFO Connected all trees
+ip-26-0-161-123:32708:33242 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-123:32708:33242 [4] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-161-123:32708:33242 [4] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-168-52:2722051:2722583 [6] NCCL INFO Channel 00/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-52:2722051:2722583 [6] NCCL INFO Channel 01/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-52:2722051:2722583 [6] NCCL INFO Channel 02/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-52:2722051:2722583 [6] NCCL INFO Channel 03/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-52:2722051:2722583 [6] NCCL INFO Channel 00/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-52:2722051:2722583 [6] NCCL INFO Channel 01/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-52:2722051:2722583 [6] NCCL INFO Channel 02/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-52:2722051:2722583 [6] NCCL INFO Channel 03/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-52:2722049:2722579 [4] NCCL INFO Channel 00/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722052:2722578 [7] NCCL INFO Channel 00/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722049:2722579 [4] NCCL INFO Channel 01/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722049:2722579 [4] NCCL INFO Channel 02/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722052:2722578 [7] NCCL INFO Channel 01/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722049:2722579 [4] NCCL INFO Channel 03/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722052:2722578 [7] NCCL INFO Channel 02/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722049:2722579 [4] NCCL INFO Channel 00/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722052:2722578 [7] NCCL INFO Channel 03/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722049:2722579 [4] NCCL INFO Channel 01/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722052:2722578 [7] NCCL INFO Channel 00/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722049:2722579 [4] NCCL INFO Channel 02/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722052:2722578 [7] NCCL INFO Channel 01/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722049:2722579 [4] NCCL INFO Channel 03/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722052:2722578 [7] NCCL INFO Channel 02/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722052:2722578 [7] NCCL INFO Channel 03/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815963:2816530 [0] NCCL INFO Channel 00/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815963:2816530 [0] NCCL INFO Channel 01/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815963:2816530 [0] NCCL INFO Channel 02/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815963:2816530 [0] NCCL INFO Channel 03/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815963:2816530 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815963:2816530 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815963:2816530 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815963:2816530 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62887:63423 [6] NCCL INFO Channel 00/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62887:63423 [6] NCCL INFO Channel 01/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62887:63423 [6] NCCL INFO Channel 02/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62887:63423 [6] NCCL INFO Channel 03/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62887:63423 [6] NCCL INFO Channel 00/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62887:63423 [6] NCCL INFO Channel 01/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-52:2722045:2722584 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62887:63423 [6] NCCL INFO Channel 02/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62887:63423 [6] NCCL INFO Channel 03/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-52:2722045:2722584 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-52:2722045:2722584 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-52:2722045:2722584 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-52:2722045:2722584 [0] NCCL INFO Channel 00/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-52:2722045:2722584 [0] NCCL INFO Channel 01/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-52:2722045:2722584 [0] NCCL INFO Channel 02/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-52:2722045:2722584 [0] NCCL INFO Channel 03/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815965:2816528 [2] NCCL INFO Channel 00/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235654:236182 [6] NCCL INFO Channel 00/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815965:2816528 [2] NCCL INFO Channel 01/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235654:236182 [6] NCCL INFO Channel 01/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815965:2816528 [2] NCCL INFO Channel 02/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235654:236182 [6] NCCL INFO Channel 02/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815965:2816528 [2] NCCL INFO Channel 03/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235654:236182 [6] NCCL INFO Channel 03/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815965:2816528 [2] NCCL INFO Channel 00/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815965:2816528 [2] NCCL INFO Channel 01/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235654:236182 [6] NCCL INFO Channel 00/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815965:2816528 [2] NCCL INFO Channel 02/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235654:236182 [6] NCCL INFO Channel 01/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815965:2816528 [2] NCCL INFO Channel 03/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235654:236182 [6] NCCL INFO Channel 02/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235654:236182 [6] NCCL INFO Channel 03/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-52:2722047:2722582 [2] NCCL INFO Channel 00/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-52:2722047:2722582 [2] NCCL INFO Channel 01/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-52:2722047:2722582 [2] NCCL INFO Channel 02/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-52:2722047:2722582 [2] NCCL INFO Channel 03/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-52:2722047:2722582 [2] NCCL INFO Channel 00/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-52:2722047:2722582 [2] NCCL INFO Channel 01/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-52:2722047:2722582 [2] NCCL INFO Channel 02/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-52:2722047:2722582 [2] NCCL INFO Channel 03/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32708:33242 [4] NCCL INFO comm 0xc480d20 rank 0 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0x5579512d5d31a4dd - Init COMPLETE
+ip-26-0-161-78:114519:115047 [7] NCCL INFO Connected all rings
+ip-26-0-161-78:114519:115047 [7] NCCL INFO Connected all trees
+ip-26-0-161-123:32707:33244 [3] NCCL INFO Connected all rings
+ip-26-0-161-123:32707:33244 [3] NCCL INFO Connected all trees
+ip-26-0-161-123:32707:33244 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-78:114519:115047 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-78:114519:115047 [7] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-161-78:114519:115047 [7] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-161-123:32707:33244 [3] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-161-123:32707:33244 [3] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-161-123:32711:33234 [7] NCCL INFO Connected all rings
+ip-26-0-161-123:32711:33234 [7] NCCL INFO Connected all trees
+ip-26-0-161-123:32711:33234 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-123:32711:33234 [7] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-161-123:32711:33234 [7] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-161-123:32710:33246 [6] NCCL INFO Connected all rings
+ip-26-0-161-123:32710:33246 [6] NCCL INFO Connected all trees
+ip-26-0-161-123:32710:33246 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-123:32710:33246 [6] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-161-123:32710:33246 [6] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-161-78:114519:115047 [7] NCCL INFO comm 0x9a995d0 rank 1 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0x13466d39857b650f - Init COMPLETE
+ip-26-0-161-123:32707:33244 [3] NCCL INFO comm 0xaa8f9e0 rank 0 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0x65195ce4f2c54799 - Init COMPLETE
+ip-26-0-160-225:124167:124700 [7] NCCL INFO Connected all rings
+ip-26-0-160-225:124167:124700 [7] NCCL INFO Connected all trees
+ip-26-0-160-225:124167:124700 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-225:124167:124700 [7] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-160-225:124167:124700 [7] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-161-123:32711:33234 [7] NCCL INFO comm 0xb634450 rank 0 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0x6c1953a2a95d57e1 - Init COMPLETE
+ip-26-0-161-103:96894:97424 [6] NCCL INFO Connected all rings
+ip-26-0-161-103:96894:97424 [6] NCCL INFO Connected all trees
+ip-26-0-161-103:96894:97424 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-103:96894:97424 [6] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-161-103:96894:97424 [6] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-168-34:272152:272682 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-161-123:32709:33243 [5] NCCL INFO Connected all rings
+ip-26-0-161-123:32709:33243 [5] NCCL INFO Connected all trees
+ip-26-0-161-123:32709:33243 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-123:32709:33243 [5] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-161-123:32709:33243 [5] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-161-123:32704:33247 [0] NCCL INFO Connected all rings
+ip-26-0-161-123:32704:33247 [0] NCCL INFO Connected all trees
+ip-26-0-161-123:32704:33247 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-123:32704:33247 [0] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-161-123:32704:33247 [0] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-161-123:32710:33246 [6] NCCL INFO comm 0xbb21a40 rank 0 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0x97bd314d7da0b64d - Init COMPLETE
+ip-26-0-160-225:124167:124700 [7] NCCL INFO comm 0xbe456f0 rank 0 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0x13466d39857b650f - Init COMPLETE
+ip-26-0-165-164:1227469:1227995 [6] NCCL INFO Connected all rings
+ip-26-0-165-164:1227469:1227995 [6] NCCL INFO Connected all trees
+ip-26-0-165-164:1227469:1227995 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-164:1227469:1227995 [6] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-165-164:1227469:1227995 [6] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-161-103:96894:97424 [6] NCCL INFO comm 0xac7ec60 rank 0 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0x1d99a63c76dbb2eb - Init COMPLETE
+ip-26-0-161-123:32709:33243 [5] NCCL INFO comm 0xc4431c0 rank 0 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0xa57a3be018a8b943 - Init COMPLETE
+ip-26-0-161-142:62888:63422 [7] NCCL INFO Channel 00/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32704:33247 [0] NCCL INFO comm 0xc65f8b0 rank 0 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0x53b316fd2156f020 - Init COMPLETE
+ip-26-0-161-142:62888:63422 [7] NCCL INFO Channel 01/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62888:63422 [7] NCCL INFO Channel 02/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62888:63422 [7] NCCL INFO Channel 03/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62888:63422 [7] NCCL INFO Channel 00/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62888:63422 [7] NCCL INFO Channel 01/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62888:63422 [7] NCCL INFO Channel 02/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62888:63422 [7] NCCL INFO Channel 03/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62883:63421 [2] NCCL INFO Channel 00/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62883:63421 [2] NCCL INFO Channel 01/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62883:63421 [2] NCCL INFO Channel 02/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62883:63421 [2] NCCL INFO Channel 03/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62883:63421 [2] NCCL INFO Channel 00/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62883:63421 [2] NCCL INFO Channel 01/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62883:63421 [2] NCCL INFO Channel 02/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62883:63421 [2] NCCL INFO Channel 03/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272153:272685 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-168-30:235655:236181 [7] NCCL INFO Channel 00/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235655:236181 [7] NCCL INFO Channel 01/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235655:236181 [7] NCCL INFO Channel 02/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235655:236181 [7] NCCL INFO Channel 03/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235655:236181 [7] NCCL INFO Channel 00/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235655:236181 [7] NCCL INFO Channel 01/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235655:236181 [7] NCCL INFO Channel 02/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235655:236181 [7] NCCL INFO Channel 03/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235650:236180 [2] NCCL INFO Channel 00/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235650:236180 [2] NCCL INFO Channel 01/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235650:236180 [2] NCCL INFO Channel 02/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235650:236180 [2] NCCL INFO Channel 03/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235650:236180 [2] NCCL INFO Channel 00/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235650:236180 [2] NCCL INFO Channel 01/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235650:236180 [2] NCCL INFO Channel 02/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235650:236180 [2] NCCL INFO Channel 03/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62882:63424 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62882:63424 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62882:63424 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62882:63424 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62882:63424 [1] NCCL INFO Channel 00/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62882:63424 [1] NCCL INFO Channel 01/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227469:1227995 [6] NCCL INFO comm 0xa4e4dc0 rank 1 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0x1d99a63c76dbb2eb - Init COMPLETE
+ip-26-0-161-142:62882:63424 [1] NCCL INFO Channel 02/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62882:63424 [1] NCCL INFO Channel 03/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114514:115049 [2] NCCL INFO Connected all rings
+ip-26-0-161-78:114514:115049 [2] NCCL INFO Connected all trees
+ip-26-0-161-78:114514:115049 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-78:114514:115049 [2] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-161-78:114514:115049 [2] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-168-30:235649:236184 [1] NCCL INFO Channel 00/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235649:236184 [1] NCCL INFO Channel 01/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235649:236184 [1] NCCL INFO Channel 02/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235649:236184 [1] NCCL INFO Channel 03/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235649:236184 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235649:236184 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235649:236184 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235649:236184 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114518:115051 [6] NCCL INFO Connected all rings
+ip-26-0-161-78:114518:115051 [6] NCCL INFO Connected all trees
+ip-26-0-161-78:114518:115051 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-34:272157:272687 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-161-78:114518:115051 [6] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-161-78:114518:115051 [6] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-160-225:124162:124705 [2] NCCL INFO Connected all rings
+ip-26-0-160-225:124162:124705 [2] NCCL INFO Connected all trees
+ip-26-0-160-225:124162:124705 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-225:124162:124705 [2] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-160-225:124162:124705 [2] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-161-78:114517:115046 [5] NCCL INFO Connected all rings
+ip-26-0-161-78:114517:115046 [5] NCCL INFO Connected all trees
+ip-26-0-161-78:114517:115046 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-78:114517:115046 [5] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-161-78:114517:115046 [5] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-161-153:2636090:2636647 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-161-142:62884:63417 [3] NCCL INFO Channel 00/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62884:63417 [3] NCCL INFO Channel 01/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62884:63417 [3] NCCL INFO Channel 02/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62884:63417 [3] NCCL INFO Channel 03/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62884:63417 [3] NCCL INFO Channel 00/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62884:63417 [3] NCCL INFO Channel 01/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62884:63417 [3] NCCL INFO Channel 02/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62884:63417 [3] NCCL INFO Channel 03/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235651:236178 [3] NCCL INFO Channel 00/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235651:236178 [3] NCCL INFO Channel 01/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235651:236178 [3] NCCL INFO Channel 02/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235651:236178 [3] NCCL INFO Channel 03/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235651:236178 [3] NCCL INFO Channel 00/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:124165:124699 [5] NCCL INFO Connected all rings
+ip-26-0-160-225:124165:124699 [5] NCCL INFO Connected all trees
+ip-26-0-160-225:124165:124699 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-225:124165:124699 [5] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-160-225:124165:124699 [5] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-168-30:235651:236178 [3] NCCL INFO Channel 01/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235651:236178 [3] NCCL INFO Channel 02/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235651:236178 [3] NCCL INFO Channel 03/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153565:154094 [3] NCCL INFO Connected all rings
+ip-26-0-165-202:153565:154094 [3] NCCL INFO Connected all trees
+ip-26-0-165-202:153565:154094 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-202:153565:154094 [3] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-165-202:153565:154094 [3] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-160-225:124166:124708 [6] NCCL INFO Connected all rings
+ip-26-0-160-225:124166:124708 [6] NCCL INFO Connected all trees
+ip-26-0-160-225:124166:124708 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-225:124166:124708 [6] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-160-225:124166:124708 [6] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-161-78:114513:115048 [1] NCCL INFO Connected all rings
+ip-26-0-161-78:114513:115048 [1] NCCL INFO Connected all trees
+ip-26-0-161-78:114513:115048 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-78:114513:115048 [1] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-161-78:114513:115048 [1] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-161-78:114514:115049 [2] NCCL INFO comm 0xbd1fab0 rank 1 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0x6160e37a54249843 - Init COMPLETE
+ip-26-0-161-78:114515:115050 [3] NCCL INFO Connected all rings
+ip-26-0-161-78:114515:115050 [3] NCCL INFO Connected all trees
+ip-26-0-161-78:114515:115050 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-202:153564:154092 [2] NCCL INFO Connected all rings
+ip-26-0-165-202:153564:154092 [2] NCCL INFO Connected all trees
+ip-26-0-165-202:153564:154092 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-78:114515:115050 [3] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-161-78:114515:115050 [3] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-165-202:153564:154092 [2] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-165-202:153564:154092 [2] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-161-78:114518:115051 [6] NCCL INFO comm 0xae517a0 rank 1 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0x85ddc6e85d763ff9 - Init COMPLETE
+ip-26-0-160-225:124163:124706 [3] NCCL INFO Connected all rings
+ip-26-0-160-225:124163:124706 [3] NCCL INFO Connected all trees
+ip-26-0-160-225:124163:124706 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-225:124163:124706 [3] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-160-225:124163:124706 [3] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-168-34:272154:272683 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-160-225:124162:124705 [2] NCCL INFO comm 0xc4b2dd0 rank 0 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0x6160e37a54249843 - Init COMPLETE
+ip-26-0-160-225:124161:124703 [1] NCCL INFO Connected all rings
+ip-26-0-160-225:124161:124703 [1] NCCL INFO Connected all trees
+ip-26-0-160-225:124161:124703 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-225:124161:124703 [1] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-160-225:124161:124703 [1] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-161-78:114517:115046 [5] NCCL INFO comm 0xc2aa9f0 rank 1 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0x70adc5cb7dc315cd - Init COMPLETE
+ip-26-0-168-34:272156:272681 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-161-142:62881:63420 [0] NCCL INFO Channel 00/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153563:154098 [1] NCCL INFO Connected all rings
+ip-26-0-165-202:153563:154098 [1] NCCL INFO Connected all trees
+ip-26-0-165-202:153563:154098 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-202:153563:154098 [1] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-165-202:153563:154098 [1] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-161-142:62881:63420 [0] NCCL INFO Channel 01/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62881:63420 [0] NCCL INFO Channel 02/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62881:63420 [0] NCCL INFO Channel 03/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62881:63420 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62881:63420 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62881:63420 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-142:62881:63420 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272156:272681 [6] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-168-34:272156:272681 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-153:2636090:2636647 [6] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-161-153:2636090:2636647 [6] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-161-153:2636090:2636647 [6] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-161-153:2636090:2636647 [6] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-161-153:2636090:2636647 [6] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-161-153:2636090:2636647 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-142:62886:63411 [5] NCCL INFO Channel 00/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62886:63411 [5] NCCL INFO Channel 01/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62886:63411 [5] NCCL INFO Channel 02/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62886:63411 [5] NCCL INFO Channel 03/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62886:63411 [5] NCCL INFO Channel 00/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62886:63411 [5] NCCL INFO Channel 01/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62886:63411 [5] NCCL INFO Channel 02/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62886:63411 [5] NCCL INFO Channel 03/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235648:236179 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235648:236179 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235653:236177 [5] NCCL INFO Channel 00/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235648:236179 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235648:236179 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235648:236179 [0] NCCL INFO Channel 00/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235653:236177 [5] NCCL INFO Channel 01/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235648:236179 [0] NCCL INFO Channel 01/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235653:236177 [5] NCCL INFO Channel 02/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235648:236179 [0] NCCL INFO Channel 02/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235653:236177 [5] NCCL INFO Channel 03/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235648:236179 [0] NCCL INFO Channel 03/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153568:154096 [6] NCCL INFO Connected all rings
+ip-26-0-165-202:153568:154096 [6] NCCL INFO Connected all trees
+ip-26-0-165-202:153568:154096 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-202:153568:154096 [6] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-165-202:153568:154096 [6] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-168-30:235653:236177 [5] NCCL INFO Channel 00/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235653:236177 [5] NCCL INFO Channel 01/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235653:236177 [5] NCCL INFO Channel 02/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272151:272686 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-168-30:235653:236177 [5] NCCL INFO Channel 03/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272150:272688 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-160-225:124165:124699 [5] NCCL INFO comm 0xaa36cd0 rank 0 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0x70adc5cb7dc315cd - Init COMPLETE
+ip-26-0-165-202:153562:154095 [0] NCCL INFO Connected all rings
+ip-26-0-165-202:153562:154095 [0] NCCL INFO Connected all trees
+ip-26-0-165-202:153562:154095 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-202:153562:154095 [0] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-165-202:153562:154095 [0] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-161-142:62885:63425 [4] NCCL INFO Channel 00/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:124166:124708 [6] NCCL INFO comm 0xb95b600 rank 0 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0x85ddc6e85d763ff9 - Init COMPLETE
+ip-26-0-161-142:62885:63425 [4] NCCL INFO Channel 01/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62885:63425 [4] NCCL INFO Channel 02/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62885:63425 [4] NCCL INFO Channel 03/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62885:63425 [4] NCCL INFO Channel 00/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62885:63425 [4] NCCL INFO Channel 01/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62885:63425 [4] NCCL INFO Channel 02/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153565:154094 [3] NCCL INFO comm 0xaf709b0 rank 1 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0x65195ce4f2c54799 - Init COMPLETE
+ip-26-0-161-142:62885:63425 [4] NCCL INFO Channel 03/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153569:154091 [7] NCCL INFO Connected all rings
+ip-26-0-165-202:153569:154091 [7] NCCL INFO Connected all trees
+ip-26-0-165-202:153569:154091 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-202:153569:154091 [7] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-165-202:153569:154091 [7] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-165-202:153567:154093 [5] NCCL INFO Connected all rings
+ip-26-0-165-202:153567:154093 [5] NCCL INFO Connected all trees
+ip-26-0-165-202:153567:154093 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-202:153567:154093 [5] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-165-202:153567:154093 [5] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-160-242:1364908:1365452 [6] NCCL INFO Connected all rings
+ip-26-0-160-242:1364908:1365452 [6] NCCL INFO Connected all trees
+ip-26-0-160-242:1364908:1365452 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-242:1364908:1365452 [6] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-160-242:1364908:1365452 [6] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-161-78:114513:115048 [1] NCCL INFO comm 0xbe5c7e0 rank 1 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0x55893f50a8f6dd29 - Init COMPLETE
+ip-26-0-168-30:235652:236183 [4] NCCL INFO Channel 00/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235652:236183 [4] NCCL INFO Channel 01/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235652:236183 [4] NCCL INFO Channel 02/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235652:236183 [4] NCCL INFO Channel 03/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114515:115050 [3] NCCL INFO comm 0xadf7330 rank 1 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0x4177b0a43e9f62e1 - Init COMPLETE
+ip-26-0-168-30:235652:236183 [4] NCCL INFO Channel 00/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235652:236183 [4] NCCL INFO Channel 01/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235652:236183 [4] NCCL INFO Channel 02/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235652:236183 [4] NCCL INFO Channel 03/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272155:272684 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-160-225:124163:124706 [3] NCCL INFO comm 0xac7b900 rank 0 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0x4177b0a43e9f62e1 - Init COMPLETE
+ip-26-0-165-202:153564:154092 [2] NCCL INFO comm 0xa7fefd0 rank 1 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0x5e0b17ed00861d01 - Init COMPLETE
+ip-26-0-160-225:124161:124703 [1] NCCL INFO comm 0xaa12d50 rank 0 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0x55893f50a8f6dd29 - Init COMPLETE
+ip-26-0-161-153:2636084:2636659 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-165-202:153563:154098 [1] NCCL INFO comm 0xc3f2ac0 rank 1 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0xe9de758c5f73d128 - Init COMPLETE
+ip-26-0-168-34:272150:272688 [0] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-168-34:272150:272688 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-153:2636084:2636659 [0] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-161-153:2636084:2636659 [0] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-161-153:2636084:2636659 [0] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-161-153:2636084:2636659 [0] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-161-153:2636084:2636659 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-161-153:2636084:2636659 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-202:153566:154097 [4] NCCL INFO Connected all rings
+ip-26-0-165-202:153566:154097 [4] NCCL INFO Connected all trees
+ip-26-0-165-202:153566:154097 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-202:153566:154097 [4] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-165-202:153566:154097 [4] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-165-202:153568:154096 [6] NCCL INFO comm 0x9d87780 rank 1 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0x97bd314d7da0b64d - Init COMPLETE
+ip-26-0-161-153:2636087:2636655 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-165-202:153562:154095 [0] NCCL INFO comm 0xb4b83a0 rank 1 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0x53b316fd2156f020 - Init COMPLETE
+ip-26-0-161-153:2636087:2636655 [3] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-161-153:2636087:2636655 [3] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-161-153:2636087:2636655 [3] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-161-153:2636087:2636655 [3] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-161-153:2636087:2636655 [3] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-161-153:2636087:2636655 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-168-34:272153:272685 [3] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-168-34:272153:272685 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-202:153569:154091 [7] NCCL INFO comm 0xa751160 rank 1 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0x6c1953a2a95d57e1 - Init COMPLETE
+ip-26-0-165-131:1630211:1630779 [6] NCCL INFO Connected all rings
+ip-26-0-165-131:1630211:1630779 [6] NCCL INFO Connected all trees
+ip-26-0-165-131:1630211:1630779 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-131:1630211:1630779 [6] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-165-202:153567:154093 [5] NCCL INFO comm 0xb79ef70 rank 1 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0xa57a3be018a8b943 - Init COMPLETE
+ip-26-0-165-131:1630211:1630779 [6] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-160-242:1364908:1365452 [6] NCCL INFO comm 0xc5af960 rank 0 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0x69dd6b894eb2fbd2 - Init COMPLETE
+ip-26-0-161-153:2636091:2636656 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-161-153:2636091:2636656 [7] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-161-153:2636091:2636656 [7] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-161-153:2636091:2636656 [7] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-161-153:2636091:2636656 [7] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-161-153:2636091:2636656 [7] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-161-153:2636091:2636656 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-168-34:272157:272687 [7] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-168-34:272157:272687 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-103:96892:97429 [4] NCCL INFO Connected all rings
+ip-26-0-161-103:96892:97429 [4] NCCL INFO Connected all trees
+ip-26-0-161-103:96892:97429 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-103:96892:97429 [4] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-161-103:96892:97429 [4] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-161-103:96888:97423 [0] NCCL INFO Connected all rings
+ip-26-0-161-103:96888:97423 [0] NCCL INFO Connected all trees
+ip-26-0-161-103:96888:97423 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-103:96888:97423 [0] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-161-103:96888:97423 [0] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-161-103:96895:97427 [7] NCCL INFO Connected all rings
+ip-26-0-161-103:96895:97427 [7] NCCL INFO Connected all trees
+ip-26-0-161-103:96895:97427 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-103:96895:97427 [7] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-161-103:96895:97427 [7] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-165-202:153566:154097 [4] NCCL INFO comm 0x9fe3290 rank 1 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0x5579512d5d31a4dd - Init COMPLETE
+ip-26-0-161-153:2636088:2636653 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-165-164:1227463:1227994 [0] NCCL INFO Connected all rings
+ip-26-0-165-164:1227463:1227994 [0] NCCL INFO Connected all trees
+ip-26-0-165-164:1227463:1227994 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-164:1227463:1227994 [0] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-165-164:1227463:1227994 [0] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-161-153:2636089:2636654 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-168-34:272154:272683 [4] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-168-34:272154:272683 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-153:2636088:2636653 [4] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-161-153:2636088:2636653 [4] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-161-153:2636088:2636653 [4] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-161-153:2636088:2636653 [4] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-161-153:2636088:2636653 [4] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-161-153:2636088:2636653 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-103:96893:97431 [5] NCCL INFO Connected all rings
+ip-26-0-161-103:96893:97431 [5] NCCL INFO Connected all trees
+ip-26-0-161-103:96893:97431 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-103:96893:97431 [5] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-161-103:96893:97431 [5] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-161-153:2636089:2636654 [5] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-161-153:2636089:2636654 [5] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-161-153:2636089:2636654 [5] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-161-153:2636089:2636654 [5] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-161-153:2636089:2636654 [5] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-161-153:2636089:2636654 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-168-34:272155:272684 [5] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-168-34:272155:272684 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-131:1630211:1630779 [6] NCCL INFO comm 0xc497e20 rank 1 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0x69dd6b894eb2fbd2 - Init COMPLETE
+ip-26-0-161-103:96891:97433 [3] NCCL INFO Connected all rings
+ip-26-0-161-103:96891:97433 [3] NCCL INFO Connected all trees
+ip-26-0-161-103:96891:97433 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-103:96891:97433 [3] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-161-103:96891:97433 [3] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-161-103:96890:97425 [2] NCCL INFO Connected all rings
+ip-26-0-161-103:96890:97425 [2] NCCL INFO Connected all trees
+ip-26-0-161-103:96890:97425 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-103:96890:97425 [2] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-161-103:96890:97425 [2] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-161-103:96889:97426 [1] NCCL INFO Connected all rings
+ip-26-0-161-103:96889:97426 [1] NCCL INFO Connected all trees
+ip-26-0-161-103:96889:97426 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-103:96889:97426 [1] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-161-103:96889:97426 [1] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-165-164:1227470:1227998 [7] NCCL INFO Connected all rings
+ip-26-0-165-164:1227470:1227998 [7] NCCL INFO Connected all trees
+ip-26-0-165-164:1227470:1227998 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-164:1227470:1227998 [7] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-165-164:1227470:1227998 [7] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-165-164:1227467:1227999 [4] NCCL INFO Connected all rings
+ip-26-0-165-164:1227467:1227999 [4] NCCL INFO Connected all trees
+ip-26-0-165-164:1227467:1227999 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-164:1227467:1227999 [4] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-165-164:1227467:1227999 [4] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-161-153:2636086:2636652 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-161-103:96892:97429 [4] NCCL INFO comm 0xbdb5c00 rank 0 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0xc0254c28e4109f52 - Init COMPLETE
+ip-26-0-161-103:96888:97423 [0] NCCL INFO comm 0xb0551c0 rank 0 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0xa047ea5e0359805a - Init COMPLETE
+ip-26-0-161-153:2636086:2636652 [2] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-161-153:2636086:2636652 [2] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-161-153:2636086:2636652 [2] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-161-153:2636086:2636652 [2] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-161-153:2636086:2636652 [2] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-161-153:2636086:2636652 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-168-34:272152:272682 [2] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-168-34:272152:272682 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-103:96895:97427 [7] NCCL INFO comm 0xc6396c0 rank 0 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0xfc4908b6009a7d47 - Init COMPLETE
+ip-26-0-165-164:1227468:1228000 [5] NCCL INFO Connected all rings
+ip-26-0-165-164:1227468:1228000 [5] NCCL INFO Connected all trees
+ip-26-0-165-164:1227468:1228000 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-164:1227468:1228000 [5] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-165-164:1227468:1228000 [5] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-165-164:1227463:1227994 [0] NCCL INFO comm 0x9d739e0 rank 1 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0xa047ea5e0359805a - Init COMPLETE
+ip-26-0-160-242:1364904:1365451 [2] NCCL INFO Connected all rings
+ip-26-0-160-242:1364904:1365451 [2] NCCL INFO Connected all trees
+ip-26-0-160-242:1364904:1365451 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-242:1364904:1365451 [2] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-160-242:1364904:1365451 [2] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-160-242:1364905:1365456 [3] NCCL INFO Connected all rings
+ip-26-0-160-242:1364905:1365456 [3] NCCL INFO Connected all trees
+ip-26-0-160-242:1364905:1365456 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-242:1364905:1365456 [3] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-160-242:1364905:1365456 [3] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-165-164:1227466:1228001 [3] NCCL INFO Connected all rings
+ip-26-0-165-164:1227466:1228001 [3] NCCL INFO Connected all trees
+ip-26-0-165-164:1227466:1228001 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-164:1227466:1228001 [3] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-165-164:1227466:1228001 [3] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-161-103:96893:97431 [5] NCCL INFO comm 0xa902d80 rank 0 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0xf57804cee76ef536 - Init COMPLETE
+ip-26-0-165-164:1227464:1227997 [1] NCCL INFO Connected all rings
+ip-26-0-165-164:1227464:1227997 [1] NCCL INFO Connected all trees
+ip-26-0-165-164:1227464:1227997 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-164:1227464:1227997 [1] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-165-164:1227464:1227997 [1] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-161-153:2636085:2636657 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-161-153:2636085:2636657 [1] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-161-153:2636085:2636657 [1] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-161-153:2636085:2636657 [1] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-161-153:2636085:2636657 [1] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-161-153:2636085:2636657 [1] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-161-153:2636085:2636657 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-168-34:272151:272686 [1] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-168-34:272151:272686 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-164:1227465:1227996 [2] NCCL INFO Connected all rings
+ip-26-0-165-164:1227465:1227996 [2] NCCL INFO Connected all trees
+ip-26-0-165-164:1227465:1227996 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-164:1227465:1227996 [2] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-165-164:1227465:1227996 [2] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-165-164:1227470:1227998 [7] NCCL INFO comm 0xa8703f0 rank 1 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0xfc4908b6009a7d47 - Init COMPLETE
+ip-26-0-165-131:1630208:1630782 [3] NCCL INFO Connected all rings
+ip-26-0-165-131:1630208:1630782 [3] NCCL INFO Connected all trees
+ip-26-0-165-131:1630208:1630782 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-131:1630208:1630782 [3] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-165-131:1630208:1630782 [3] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-161-103:96890:97425 [2] NCCL INFO comm 0xb030390 rank 0 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0xbaa57eddcf04b52 - Init COMPLETE
+ip-26-0-161-103:96891:97433 [3] NCCL INFO comm 0xa41d6d0 rank 0 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0x190cbf84dbe1f009 - Init COMPLETE
+ip-26-0-165-164:1227467:1227999 [4] NCCL INFO comm 0xb225480 rank 1 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0xc0254c28e4109f52 - Init COMPLETE
+ip-26-0-161-103:96889:97426 [1] NCCL INFO comm 0xa9ec050 rank 0 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0xf09cef77bd870ed5 - Init COMPLETE
+ip-26-0-165-164:1227468:1228000 [5] NCCL INFO comm 0xbbc7e00 rank 1 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0xf57804cee76ef536 - Init COMPLETE
+ip-26-0-160-242:1364903:1365461 [1] NCCL INFO Connected all rings
+ip-26-0-160-242:1364903:1365461 [1] NCCL INFO Connected all trees
+ip-26-0-165-131:1630210:1630783 [5] NCCL INFO Connected all rings
+ip-26-0-165-131:1630210:1630783 [5] NCCL INFO Connected all trees
+ip-26-0-160-242:1364903:1365461 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-131:1630210:1630783 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-131:1630210:1630783 [5] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-160-242:1364903:1365461 [1] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-160-242:1364903:1365461 [1] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-165-131:1630210:1630783 [5] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-160-242:1364906:1365453 [4] NCCL INFO Connected all rings
+ip-26-0-160-242:1364906:1365453 [4] NCCL INFO Connected all trees
+ip-26-0-160-242:1364906:1365453 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-242:1364906:1365453 [4] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-160-242:1364906:1365453 [4] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-165-164:1227466:1228001 [3] NCCL INFO comm 0xa455b10 rank 1 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0x190cbf84dbe1f009 - Init COMPLETE
+ip-26-0-160-242:1364904:1365451 [2] NCCL INFO comm 0xadf27a0 rank 0 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0x4a887d18c7bdcfe2 - Init COMPLETE
+ip-26-0-160-242:1364905:1365456 [3] NCCL INFO comm 0xba2e260 rank 0 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0xaf1c7fecc0b3b264 - Init COMPLETE
+ip-26-0-165-164:1227464:1227997 [1] NCCL INFO comm 0xa63c090 rank 1 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0xf09cef77bd870ed5 - Init COMPLETE
+ip-26-0-165-131:1630206:1630784 [1] NCCL INFO Connected all rings
+ip-26-0-165-131:1630206:1630784 [1] NCCL INFO Connected all trees
+ip-26-0-165-131:1630206:1630784 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-131:1630206:1630784 [1] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-165-131:1630206:1630784 [1] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-160-242:1364902:1365449 [0] NCCL INFO Connected all rings
+ip-26-0-160-242:1364902:1365449 [0] NCCL INFO Connected all trees
+ip-26-0-160-242:1364902:1365449 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-242:1364902:1365449 [0] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-160-242:1364902:1365449 [0] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-165-164:1227465:1227996 [2] NCCL INFO comm 0xa2001a0 rank 1 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0xbaa57eddcf04b52 - Init COMPLETE
+ip-26-0-165-131:1630209:1630781 [4] NCCL INFO Connected all rings
+ip-26-0-165-131:1630209:1630781 [4] NCCL INFO Connected all trees
+ip-26-0-165-131:1630209:1630781 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-131:1630209:1630781 [4] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-165-131:1630209:1630781 [4] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-165-131:1630208:1630782 [3] NCCL INFO comm 0xa509410 rank 1 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0xaf1c7fecc0b3b264 - Init COMPLETE
+ip-26-0-165-131:1630212:1630785 [7] NCCL INFO Connected all rings
+ip-26-0-165-131:1630212:1630785 [7] NCCL INFO Connected all trees
+ip-26-0-165-131:1630212:1630785 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-131:1630212:1630785 [7] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-165-131:1630212:1630785 [7] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-165-131:1630207:1630780 [2] NCCL INFO Connected all rings
+ip-26-0-165-131:1630207:1630780 [2] NCCL INFO Connected all trees
+ip-26-0-165-131:1630207:1630780 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-131:1630207:1630780 [2] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-165-131:1630207:1630780 [2] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-160-242:1364909:1365460 [7] NCCL INFO Connected all rings
+ip-26-0-160-242:1364909:1365460 [7] NCCL INFO Connected all trees
+ip-26-0-160-242:1364909:1365460 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-242:1364909:1365460 [7] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-160-242:1364909:1365460 [7] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-160-242:1364903:1365461 [1] NCCL INFO comm 0xb285650 rank 0 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0x173c75bc0100d108 - Init COMPLETE
+ip-26-0-160-242:1364906:1365453 [4] NCCL INFO comm 0xc194b30 rank 0 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0xb20b05f5e3ea6185 - Init COMPLETE
+ip-26-0-165-131:1630210:1630783 [5] NCCL INFO comm 0xbebd4f0 rank 1 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0xd8b0494bfaa0f0bf - Init COMPLETE
+ip-26-0-160-242:1364907:1365459 [5] NCCL INFO Connected all rings
+ip-26-0-160-242:1364907:1365459 [5] NCCL INFO Connected all trees
+ip-26-0-160-242:1364907:1365459 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-242:1364907:1365459 [5] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-160-242:1364907:1365459 [5] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-165-131:1630205:1630778 [0] NCCL INFO Connected all rings
+ip-26-0-165-131:1630205:1630778 [0] NCCL INFO Connected all trees
+ip-26-0-165-131:1630205:1630778 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-131:1630205:1630778 [0] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-165-131:1630205:1630778 [0] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-160-242:1364902:1365449 [0] NCCL INFO comm 0xb9b3db0 rank 0 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0x708e697530715efe - Init COMPLETE
+ip-26-0-165-131:1630206:1630784 [1] NCCL INFO comm 0xb10bcb0 rank 1 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0x173c75bc0100d108 - Init COMPLETE
+ip-26-0-165-131:1630209:1630781 [4] NCCL INFO comm 0xa78aba0 rank 1 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0xb20b05f5e3ea6185 - Init COMPLETE
+ip-26-0-165-131:1630212:1630785 [7] NCCL INFO comm 0xa6e5da0 rank 1 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0x370c64acc5849eec - Init COMPLETE
+ip-26-0-160-242:1364909:1365460 [7] NCCL INFO comm 0xc07f290 rank 0 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0x370c64acc5849eec - Init COMPLETE
+ip-26-0-165-131:1630207:1630780 [2] NCCL INFO comm 0xb5730b0 rank 1 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0x4a887d18c7bdcfe2 - Init COMPLETE
+ip-26-0-168-34:272156:272681 [6] NCCL INFO Channel 00/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364907:1365459 [5] NCCL INFO comm 0xa8a8580 rank 0 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0xd8b0494bfaa0f0bf - Init COMPLETE
+ip-26-0-168-34:272156:272681 [6] NCCL INFO Channel 01/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272156:272681 [6] NCCL INFO Channel 02/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272156:272681 [6] NCCL INFO Channel 03/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272156:272681 [6] NCCL INFO Channel 00/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272156:272681 [6] NCCL INFO Channel 01/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272156:272681 [6] NCCL INFO Channel 02/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272156:272681 [6] NCCL INFO Channel 03/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636090:2636647 [6] NCCL INFO Channel 00/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636090:2636647 [6] NCCL INFO Channel 01/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636090:2636647 [6] NCCL INFO Channel 02/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636090:2636647 [6] NCCL INFO Channel 03/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636090:2636647 [6] NCCL INFO Channel 00/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636090:2636647 [6] NCCL INFO Channel 01/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636090:2636647 [6] NCCL INFO Channel 02/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636090:2636647 [6] NCCL INFO Channel 03/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630205:1630778 [0] NCCL INFO comm 0xa447920 rank 1 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0x708e697530715efe - Init COMPLETE
+ip-26-0-168-34:272150:272688 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272150:272688 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272150:272688 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272150:272688 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272150:272688 [0] NCCL INFO Channel 00/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272150:272688 [0] NCCL INFO Channel 01/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272150:272688 [0] NCCL INFO Channel 02/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272150:272688 [0] NCCL INFO Channel 03/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636084:2636659 [0] NCCL INFO Channel 00/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636084:2636659 [0] NCCL INFO Channel 01/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636084:2636659 [0] NCCL INFO Channel 02/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636084:2636659 [0] NCCL INFO Channel 03/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636084:2636659 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636084:2636659 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272153:272685 [3] NCCL INFO Channel 00/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636084:2636659 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272153:272685 [3] NCCL INFO Channel 01/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636084:2636659 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272153:272685 [3] NCCL INFO Channel 02/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272153:272685 [3] NCCL INFO Channel 03/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272153:272685 [3] NCCL INFO Channel 00/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272153:272685 [3] NCCL INFO Channel 01/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272153:272685 [3] NCCL INFO Channel 02/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272153:272685 [3] NCCL INFO Channel 03/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636087:2636655 [3] NCCL INFO Channel 00/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636087:2636655 [3] NCCL INFO Channel 01/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636087:2636655 [3] NCCL INFO Channel 02/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636087:2636655 [3] NCCL INFO Channel 03/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636087:2636655 [3] NCCL INFO Channel 00/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636087:2636655 [3] NCCL INFO Channel 01/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636087:2636655 [3] NCCL INFO Channel 02/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636087:2636655 [3] NCCL INFO Channel 03/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272157:272687 [7] NCCL INFO Channel 00/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272157:272687 [7] NCCL INFO Channel 01/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272157:272687 [7] NCCL INFO Channel 02/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272157:272687 [7] NCCL INFO Channel 03/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272157:272687 [7] NCCL INFO Channel 00/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272157:272687 [7] NCCL INFO Channel 01/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272157:272687 [7] NCCL INFO Channel 02/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272157:272687 [7] NCCL INFO Channel 03/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636091:2636656 [7] NCCL INFO Channel 00/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636091:2636656 [7] NCCL INFO Channel 01/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636091:2636656 [7] NCCL INFO Channel 02/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636091:2636656 [7] NCCL INFO Channel 03/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636091:2636656 [7] NCCL INFO Channel 00/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636091:2636656 [7] NCCL INFO Channel 01/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636091:2636656 [7] NCCL INFO Channel 02/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636091:2636656 [7] NCCL INFO Channel 03/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272154:272683 [4] NCCL INFO Channel 00/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272154:272683 [4] NCCL INFO Channel 01/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272154:272683 [4] NCCL INFO Channel 02/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272154:272683 [4] NCCL INFO Channel 03/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272154:272683 [4] NCCL INFO Channel 00/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+12/28/2024 11:56:04 [INFO|DP=0|PP=0|TP=7|ip-26-0-160-225]: Local number of parameters: 7.45G (14212.00MiB)
+12/28/2024 11:56:04 [INFO|DP=0|PP=1|TP=0|ip-26-0-161-78]: Local number of parameters: 7.22G (13771.91MiB)
+ip-26-0-168-34:272154:272683 [4] NCCL INFO Channel 01/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+12/28/2024 11:56:04 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: Total number of parameters: 469G (895485.00MiB)
+12/28/2024 11:56:04 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: Local number of parameters: 7.45G (14212.00MiB)
+ip-26-0-168-34:272154:272683 [4] NCCL INFO Channel 02/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272154:272683 [4] NCCL INFO Channel 03/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+12/28/2024 11:56:04 [INFO|DP=0|PP=0|TP=7|ip-26-0-160-225]: [After model building] Memory usage: 14212.07MiB. Peak allocated: 14212.25MiB Peak reserved: 14314.00MiB
+12/28/2024 11:56:04 [INFO|DP=0|PP=1|TP=0|ip-26-0-161-78]: [After model building] Memory usage: 13771.98MiB. Peak allocated: 13772.15MiB Peak reserved: 13886.00MiB
+12/28/2024 11:56:04 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [After model building] Memory usage: 14212.07MiB. Peak allocated: 14212.25MiB Peak reserved: 14314.00MiB
+12/28/2024 11:56:04 [INFO|DP=0|PP=0|TP=31|ip-26-0-161-123]: Local number of parameters: 7.45G (14212.00MiB)
+12/28/2024 11:56:04 [INFO|DP=0|PP=0|TP=31|ip-26-0-161-123]: [After model building] Memory usage: 14212.07MiB. Peak allocated: 14212.25MiB Peak reserved: 14314.00MiB
+ip-26-0-161-153:2636088:2636653 [4] NCCL INFO Channel 00/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636089:2636654 [5] NCCL INFO Channel 00/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636088:2636653 [4] NCCL INFO Channel 01/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636089:2636654 [5] NCCL INFO Channel 01/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636088:2636653 [4] NCCL INFO Channel 02/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636089:2636654 [5] NCCL INFO Channel 02/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272155:272684 [5] NCCL INFO Channel 00/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636088:2636653 [4] NCCL INFO Channel 03/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636089:2636654 [5] NCCL INFO Channel 03/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272155:272684 [5] NCCL INFO Channel 01/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+12/28/2024 11:56:04 [INFO|DP=0|PP=0|TP=17|ip-26-0-161-103]: Local number of parameters: 7.45G (14212.00MiB)
+ip-26-0-161-153:2636088:2636653 [4] NCCL INFO Channel 00/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636089:2636654 [5] NCCL INFO Channel 00/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+12/28/2024 11:56:04 [INFO|DP=0|PP=0|TP=8|ip-26-0-160-242]: Local number of parameters: 7.45G (14212.00MiB)
+ip-26-0-168-34:272155:272684 [5] NCCL INFO Channel 02/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636088:2636653 [4] NCCL INFO Channel 01/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636089:2636654 [5] NCCL INFO Channel 01/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636088:2636653 [4] NCCL INFO Channel 02/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272155:272684 [5] NCCL INFO Channel 03/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636089:2636654 [5] NCCL INFO Channel 02/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636088:2636653 [4] NCCL INFO Channel 03/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636089:2636654 [5] NCCL INFO Channel 03/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272155:272684 [5] NCCL INFO Channel 00/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272155:272684 [5] NCCL INFO Channel 01/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272155:272684 [5] NCCL INFO Channel 02/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+12/28/2024 11:56:04 [INFO|DP=0|PP=0|TP=17|ip-26-0-161-103]: [After model building] Memory usage: 14212.07MiB. Peak allocated: 14212.25MiB Peak reserved: 14314.00MiB
+ip-26-0-168-34:272155:272684 [5] NCCL INFO Channel 03/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+12/28/2024 11:56:04 [INFO|DP=0|PP=0|TP=8|ip-26-0-160-242]: [After model building] Memory usage: 14212.07MiB. Peak allocated: 14212.25MiB Peak reserved: 14314.00MiB
+12/28/2024 11:56:04 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: No checkpoint path provided.
+12/28/2024 11:56:04 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: Parametrizing model parameters using StandardParametrizator
+12/28/2024 11:56:04 [INFO|DP=0|PP=1|TP=8|ip-26-0-165-131]: Local number of parameters: 7.22G (13771.91MiB)
+12/28/2024 11:56:04 [INFO|DP=0|PP=1|TP=8|ip-26-0-165-131]: [After model building] Memory usage: 13771.98MiB. Peak allocated: 13772.15MiB Peak reserved: 13886.00MiB
+ip-26-0-168-34:272152:272682 [2] NCCL INFO Channel 00/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272152:272682 [2] NCCL INFO Channel 01/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272152:272682 [2] NCCL INFO Channel 02/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272152:272682 [2] NCCL INFO Channel 03/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272152:272682 [2] NCCL INFO Channel 00/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272152:272682 [2] NCCL INFO Channel 01/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+12/28/2024 11:56:04 [INFO|DP=0|PP=1|TP=7|ip-26-0-161-78]: Local number of parameters: 7.22G (13771.91MiB)
+ip-26-0-168-34:272152:272682 [2] NCCL INFO Channel 02/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272152:272682 [2] NCCL INFO Channel 03/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+12/28/2024 11:56:04 [INFO|DP=0|PP=0|TP=3|ip-26-0-160-225]: Local number of parameters: 7.45G (14212.00MiB)
+12/28/2024 11:56:04 [INFO|DP=0|PP=1|TP=3|ip-26-0-161-78]: Local number of parameters: 7.22G (13771.91MiB)
+12/28/2024 11:56:04 [INFO|DP=0|PP=1|TP=7|ip-26-0-161-78]: [After model building] Memory usage: 13771.98MiB. Peak allocated: 13772.15MiB Peak reserved: 13886.00MiB
+12/28/2024 11:56:04 [INFO|DP=0|PP=0|TP=6|ip-26-0-160-225]: Local number of parameters: 7.45G (14212.00MiB)
+12/28/2024 11:56:04 [INFO|DP=0|PP=0|TP=1|ip-26-0-160-225]: Local number of parameters: 7.45G (14212.00MiB)
+12/28/2024 11:56:04 [INFO|DP=0|PP=0|TP=2|ip-26-0-160-225]: Local number of parameters: 7.45G (14212.00MiB)
+12/28/2024 11:56:04 [INFO|DP=0|PP=0|TP=3|ip-26-0-160-225]: [After model building] Memory usage: 14212.07MiB. Peak allocated: 14212.25MiB Peak reserved: 14314.00MiB
+12/28/2024 11:56:04 [INFO|DP=0|PP=1|TP=5|ip-26-0-161-78]: Local number of parameters: 7.22G (13771.91MiB)
+12/28/2024 11:56:04 [INFO|DP=0|PP=1|TP=3|ip-26-0-161-78]: [After model building] Memory usage: 13771.98MiB. Peak allocated: 13772.15MiB Peak reserved: 13886.00MiB
+12/28/2024 11:56:04 [INFO|DP=0|PP=0|TP=5|ip-26-0-160-225]: Local number of parameters: 7.45G (14212.00MiB)
+12/28/2024 11:56:04 [INFO|DP=0|PP=1|TP=2|ip-26-0-161-78]: Local number of parameters: 7.22G (13771.91MiB)
+12/28/2024 11:56:04 [INFO|DP=0|PP=0|TP=4|ip-26-0-160-225]: Local number of parameters: 7.45G (14212.00MiB)
+12/28/2024 11:56:04 [INFO|DP=0|PP=0|TP=6|ip-26-0-160-225]: [After model building] Memory usage: 14212.07MiB. Peak allocated: 14212.25MiB Peak reserved: 14314.00MiB
+12/28/2024 11:56:04 [INFO|DP=0|PP=0|TP=1|ip-26-0-160-225]: [After model building] Memory usage: 14212.07MiB. Peak allocated: 14212.25MiB Peak reserved: 14314.00MiB
+12/28/2024 11:56:04 [INFO|DP=0|PP=0|TP=2|ip-26-0-160-225]: [After model building] Memory usage: 14212.07MiB. Peak allocated: 14212.25MiB Peak reserved: 14314.00MiB
+12/28/2024 11:56:04 [INFO|DP=0|PP=1|TP=5|ip-26-0-161-78]: [After model building] Memory usage: 13771.98MiB. Peak allocated: 13772.15MiB Peak reserved: 13886.00MiB
+12/28/2024 11:56:04 [INFO|DP=0|PP=0|TP=5|ip-26-0-160-225]: [After model building] Memory usage: 14212.07MiB. Peak allocated: 14212.25MiB Peak reserved: 14314.00MiB
+12/28/2024 11:56:04 [INFO|DP=0|PP=1|TP=2|ip-26-0-161-78]: [After model building] Memory usage: 13771.98MiB. Peak allocated: 13772.15MiB Peak reserved: 13886.00MiB
+12/28/2024 11:56:04 [INFO|DP=0|PP=0|TP=4|ip-26-0-160-225]: [After model building] Memory usage: 14212.07MiB. Peak allocated: 14212.25MiB Peak reserved: 14314.00MiB
+ip-26-0-161-153:2636086:2636652 [2] NCCL INFO Channel 00/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+12/28/2024 11:56:04 [INFO|DP=0|PP=1|TP=17|ip-26-0-165-164]: Local number of parameters: 7.22G (13771.91MiB)
+ip-26-0-161-153:2636086:2636652 [2] NCCL INFO Channel 01/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636086:2636652 [2] NCCL INFO Channel 02/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636086:2636652 [2] NCCL INFO Channel 03/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636086:2636652 [2] NCCL INFO Channel 00/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636086:2636652 [2] NCCL INFO Channel 01/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636086:2636652 [2] NCCL INFO Channel 02/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+12/28/2024 11:56:04 [INFO|DP=0|PP=1|TP=17|ip-26-0-165-164]: [After model building] Memory usage: 13771.98MiB. Peak allocated: 13772.15MiB Peak reserved: 13886.00MiB
+ip-26-0-161-153:2636086:2636652 [2] NCCL INFO Channel 03/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+12/28/2024 11:56:04 [INFO|DP=0|PP=1|TP=31|ip-26-0-165-202]: Local number of parameters: 7.22G (13771.91MiB)
+12/28/2024 11:56:04 [INFO|DP=0|PP=0|TP=22|ip-26-0-161-103]: Local number of parameters: 7.45G (14212.00MiB)
+12/28/2024 11:56:04 [INFO|DP=0|PP=1|TP=31|ip-26-0-165-202]: [After model building] Memory usage: 13771.98MiB. Peak allocated: 13772.15MiB Peak reserved: 13886.00MiB
+12/28/2024 11:56:04 [INFO|DP=0|PP=0|TP=16|ip-26-0-161-103]: Local number of parameters: 7.45G (14212.00MiB)
+12/28/2024 11:56:04 [INFO|DP=0|PP=0|TP=22|ip-26-0-161-103]: [After model building] Memory usage: 14212.07MiB. Peak allocated: 14212.25MiB Peak reserved: 14314.00MiB
+12/28/2024 11:56:04 [INFO|DP=0|PP=0|TP=19|ip-26-0-161-103]: Local number of parameters: 7.45G (14212.00MiB)
+12/28/2024 11:56:04 [INFO|DP=0|PP=0|TP=10|ip-26-0-160-242]: Local number of parameters: 7.45G (14212.00MiB)
+12/28/2024 11:56:04 [INFO|DP=0|PP=0|TP=15|ip-26-0-160-242]: Local number of parameters: 7.45G (14212.00MiB)
+12/28/2024 11:56:04 [INFO|DP=0|PP=0|TP=20|ip-26-0-161-103]: Local number of parameters: 7.45G (14212.00MiB)
+12/28/2024 11:56:04 [INFO|DP=0|PP=0|TP=9|ip-26-0-160-242]: Local number of parameters: 7.45G (14212.00MiB)
+12/28/2024 11:56:04 [INFO|DP=0|PP=0|TP=28|ip-26-0-161-123]: Local number of parameters: 7.45G (14212.00MiB)
+12/28/2024 11:56:04 [INFO|DP=0|PP=0|TP=13|ip-26-0-160-242]: Local number of parameters: 7.45G (14212.00MiB)
+12/28/2024 11:56:04 [INFO|DP=0|PP=0|TP=21|ip-26-0-161-103]: Local number of parameters: 7.45G (14212.00MiB)
+12/28/2024 11:56:04 [INFO|DP=0|PP=0|TP=16|ip-26-0-161-103]: [After model building] Memory usage: 14212.07MiB. Peak allocated: 14212.25MiB Peak reserved: 14314.00MiB
+12/28/2024 11:56:04 [INFO|DP=0|PP=1|TP=10|ip-26-0-165-131]: Local number of parameters: 7.22G (13771.91MiB)
+12/28/2024 11:56:04 [INFO|DP=0|PP=0|TP=23|ip-26-0-161-103]: Local number of parameters: 7.45G (14212.00MiB)
+12/28/2024 11:56:04 [INFO|DP=0|PP=0|TP=18|ip-26-0-161-103]: Local number of parameters: 7.45G (14212.00MiB)
+12/28/2024 11:56:04 [INFO|DP=0|PP=1|TP=16|ip-26-0-165-164]: Local number of parameters: 7.22G (13771.91MiB)
+12/28/2024 11:56:04 [INFO|DP=0|PP=0|TP=14|ip-26-0-160-242]: Local number of parameters: 7.45G (14212.00MiB)
+12/28/2024 11:56:04 [INFO|DP=0|PP=0|TP=19|ip-26-0-161-103]: [After model building] Memory usage: 14212.07MiB. Peak allocated: 14212.25MiB Peak reserved: 14314.00MiB
+12/28/2024 11:56:04 [INFO|DP=0|PP=0|TP=12|ip-26-0-160-242]: Local number of parameters: 7.45G (14212.00MiB)
+12/28/2024 11:56:04 [INFO|DP=0|PP=0|TP=10|ip-26-0-160-242]: [After model building] Memory usage: 14212.07MiB. Peak allocated: 14212.25MiB Peak reserved: 14314.00MiB
+12/28/2024 11:56:04 [INFO|DP=0|PP=0|TP=27|ip-26-0-161-123]: Local number of parameters: 7.45G (14212.00MiB)
+12/28/2024 11:56:04 [INFO|DP=0|PP=0|TP=20|ip-26-0-161-103]: [After model building] Memory usage: 14212.07MiB. Peak allocated: 14212.25MiB Peak reserved: 14314.00MiB
+12/28/2024 11:56:04 [INFO|DP=0|PP=0|TP=15|ip-26-0-160-242]: [After model building] Memory usage: 14212.07MiB. Peak allocated: 14212.25MiB Peak reserved: 14314.00MiB
+12/28/2024 11:56:04 [INFO|DP=0|PP=0|TP=9|ip-26-0-160-242]: [After model building] Memory usage: 14212.07MiB. Peak allocated: 14212.25MiB Peak reserved: 14314.00MiB
+12/28/2024 11:56:04 [INFO|DP=0|PP=0|TP=28|ip-26-0-161-123]: [After model building] Memory usage: 14212.07MiB. Peak allocated: 14212.25MiB Peak reserved: 14314.00MiB
+12/28/2024 11:56:04 [INFO|DP=0|PP=0|TP=21|ip-26-0-161-103]: [After model building] Memory usage: 14212.07MiB. Peak allocated: 14212.25MiB Peak reserved: 14314.00MiB
+12/28/2024 11:56:04 [INFO|DP=0|PP=1|TP=16|ip-26-0-165-164]: [After model building] Memory usage: 13771.98MiB. Peak allocated: 13772.15MiB Peak reserved: 13886.00MiB
+12/28/2024 11:56:04 [INFO|DP=0|PP=1|TP=10|ip-26-0-165-131]: [After model building] Memory usage: 13771.98MiB. Peak allocated: 13772.15MiB Peak reserved: 13886.00MiB
+12/28/2024 11:56:04 [INFO|DP=0|PP=0|TP=13|ip-26-0-160-242]: [After model building] Memory usage: 14212.07MiB. Peak allocated: 14212.25MiB Peak reserved: 14314.00MiB
+12/28/2024 11:56:04 [INFO|DP=0|PP=0|TP=29|ip-26-0-161-123]: Local number of parameters: 7.45G (14212.00MiB)
+12/28/2024 11:56:04 [INFO|DP=0|PP=0|TP=23|ip-26-0-161-103]: [After model building] Memory usage: 14212.07MiB. Peak allocated: 14212.25MiB Peak reserved: 14314.00MiB
+12/28/2024 11:56:04 [INFO|DP=0|PP=0|TP=18|ip-26-0-161-103]: [After model building] Memory usage: 14212.07MiB. Peak allocated: 14212.25MiB Peak reserved: 14314.00MiB
+12/28/2024 11:56:04 [INFO|DP=0|PP=0|TP=14|ip-26-0-160-242]: [After model building] Memory usage: 14212.07MiB. Peak allocated: 14212.25MiB Peak reserved: 14314.00MiB
+ip-26-0-168-34:272151:272686 [1] NCCL INFO Channel 00/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+12/28/2024 11:56:04 [INFO|DP=0|PP=0|TP=27|ip-26-0-161-123]: [After model building] Memory usage: 14212.07MiB. Peak allocated: 14212.25MiB Peak reserved: 14314.00MiB
+ip-26-0-168-34:272151:272686 [1] NCCL INFO Channel 01/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+12/28/2024 11:56:04 [INFO|DP=0|PP=1|TP=15|ip-26-0-165-131]: Local number of parameters: 7.22G (13771.91MiB)
+12/28/2024 11:56:04 [INFO|DP=0|PP=1|TP=11|ip-26-0-165-131]: Local number of parameters: 7.22G (13771.91MiB)
+12/28/2024 11:56:04 [INFO|DP=0|PP=0|TP=12|ip-26-0-160-242]: [After model building] Memory usage: 14212.07MiB. Peak allocated: 14212.25MiB Peak reserved: 14314.00MiB
+12/28/2024 11:56:04 [INFO|DP=0|PP=1|TP=22|ip-26-0-165-164]: Local number of parameters: 7.22G (13771.91MiB)
+ip-26-0-168-34:272151:272686 [1] NCCL INFO Channel 02/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+12/28/2024 11:56:04 [INFO|DP=0|PP=1|TP=9|ip-26-0-165-131]: Local number of parameters: 7.22G (13771.91MiB)
+ip-26-0-168-34:272151:272686 [1] NCCL INFO Channel 03/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+12/28/2024 11:56:04 [INFO|DP=0|PP=0|TP=29|ip-26-0-161-123]: [After model building] Memory usage: 14212.07MiB. Peak allocated: 14212.25MiB Peak reserved: 14314.00MiB
+12/28/2024 11:56:04 [INFO|DP=0|PP=1|TP=18|ip-26-0-165-164]: Local number of parameters: 7.22G (13771.91MiB)
+12/28/2024 11:56:04 [INFO|DP=0|PP=1|TP=21|ip-26-0-165-164]: Local number of parameters: 7.22G (13771.91MiB)
+ip-26-0-168-34:272151:272686 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+12/28/2024 11:56:04 [INFO|DP=0|PP=0|TP=11|ip-26-0-160-242]: Local number of parameters: 7.45G (14212.00MiB)
+12/28/2024 11:56:04 [INFO|DP=0|PP=1|TP=12|ip-26-0-165-131]: Local number of parameters: 7.22G (13771.91MiB)
+12/28/2024 11:56:04 [INFO|DP=0|PP=1|TP=30|ip-26-0-165-202]: Local number of parameters: 7.22G (13771.91MiB)
+ip-26-0-168-34:272151:272686 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+12/28/2024 11:56:04 [INFO|DP=0|PP=1|TP=13|ip-26-0-165-131]: Local number of parameters: 7.22G (13771.91MiB)
+ip-26-0-168-34:272151:272686 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+12/28/2024 11:56:04 [INFO|DP=0|PP=0|TP=30|ip-26-0-161-123]: Local number of parameters: 7.45G (14212.00MiB)
+12/28/2024 11:56:04 [INFO|DP=0|PP=1|TP=11|ip-26-0-165-131]: [After model building] Memory usage: 13771.98MiB. Peak allocated: 13772.15MiB Peak reserved: 13886.00MiB
+ip-26-0-168-34:272151:272686 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+12/28/2024 11:56:04 [INFO|DP=0|PP=1|TP=15|ip-26-0-165-131]: [After model building] Memory usage: 13771.98MiB. Peak allocated: 13772.15MiB Peak reserved: 13886.00MiB
+12/28/2024 11:56:04 [INFO|DP=0|PP=1|TP=19|ip-26-0-165-164]: Local number of parameters: 7.22G (13771.91MiB)
+12/28/2024 11:56:04 [INFO|DP=0|PP=1|TP=22|ip-26-0-165-164]: [After model building] Memory usage: 13771.98MiB. Peak allocated: 13772.15MiB Peak reserved: 13886.00MiB
+12/28/2024 11:56:04 [INFO|DP=0|PP=1|TP=14|ip-26-0-165-131]: Local number of parameters: 7.22G (13771.91MiB)
+12/28/2024 11:56:04 [INFO|DP=0|PP=0|TP=24|ip-26-0-161-123]: Local number of parameters: 7.45G (14212.00MiB)
+12/28/2024 11:56:04 [INFO|DP=0|PP=1|TP=9|ip-26-0-165-131]: [After model building] Memory usage: 13771.98MiB. Peak allocated: 13772.15MiB Peak reserved: 13886.00MiB
+12/28/2024 11:56:04 [INFO|DP=0|PP=0|TP=25|ip-26-0-161-123]: Local number of parameters: 7.45G (14212.00MiB)
+12/28/2024 11:56:04 [INFO|DP=0|PP=1|TP=12|ip-26-0-165-131]: [After model building] Memory usage: 13771.98MiB. Peak allocated: 13772.15MiB Peak reserved: 13886.00MiB
+12/28/2024 11:56:04 [INFO|DP=0|PP=0|TP=26|ip-26-0-161-123]: Local number of parameters: 7.45G (14212.00MiB)
+12/28/2024 11:56:04 [INFO|DP=0|PP=1|TP=20|ip-26-0-165-164]: Local number of parameters: 7.22G (13771.91MiB)
+12/28/2024 11:56:04 [INFO|DP=0|PP=1|TP=18|ip-26-0-165-164]: [After model building] Memory usage: 13771.98MiB. Peak allocated: 13772.15MiB Peak reserved: 13886.00MiB
+12/28/2024 11:56:04 [INFO|DP=0|PP=1|TP=13|ip-26-0-165-131]: [After model building] Memory usage: 13771.98MiB. Peak allocated: 13772.15MiB Peak reserved: 13886.00MiB
+12/28/2024 11:56:04 [INFO|DP=0|PP=1|TP=30|ip-26-0-165-202]: [After model building] Memory usage: 13771.98MiB. Peak allocated: 13772.15MiB Peak reserved: 13886.00MiB
+12/28/2024 11:56:04 [INFO|DP=0|PP=1|TP=21|ip-26-0-165-164]: [After model building] Memory usage: 13771.98MiB. Peak allocated: 13772.15MiB Peak reserved: 13886.00MiB
+12/28/2024 11:56:04 [INFO|DP=0|PP=0|TP=30|ip-26-0-161-123]: [After model building] Memory usage: 14212.07MiB. Peak allocated: 14212.25MiB Peak reserved: 14314.00MiB
+12/28/2024 11:56:04 [INFO|DP=0|PP=1|TP=29|ip-26-0-165-202]: Local number of parameters: 7.22G (13771.91MiB)
+12/28/2024 11:56:04 [INFO|DP=0|PP=1|TP=27|ip-26-0-165-202]: Local number of parameters: 7.22G (13771.91MiB)
+12/28/2024 11:56:04 [INFO|DP=0|PP=1|TP=14|ip-26-0-165-131]: [After model building] Memory usage: 13771.98MiB. Peak allocated: 13772.15MiB Peak reserved: 13886.00MiB
+12/28/2024 11:56:04 [INFO|DP=0|PP=1|TP=19|ip-26-0-165-164]: [After model building] Memory usage: 13771.98MiB. Peak allocated: 13772.15MiB Peak reserved: 13886.00MiB
+12/28/2024 11:56:04 [INFO|DP=0|PP=0|TP=25|ip-26-0-161-123]: [After model building] Memory usage: 14212.07MiB. Peak allocated: 14212.25MiB Peak reserved: 14314.00MiB
+12/28/2024 11:56:04 [INFO|DP=0|PP=0|TP=24|ip-26-0-161-123]: [After model building] Memory usage: 14212.07MiB. Peak allocated: 14212.25MiB Peak reserved: 14314.00MiB
+12/28/2024 11:56:04 [INFO|DP=0|PP=0|TP=26|ip-26-0-161-123]: [After model building] Memory usage: 14212.07MiB. Peak allocated: 14212.25MiB Peak reserved: 14314.00MiB
+12/28/2024 11:56:04 [INFO|DP=0|PP=1|TP=20|ip-26-0-165-164]: [After model building] Memory usage: 13771.98MiB. Peak allocated: 13772.15MiB Peak reserved: 13886.00MiB
+12/28/2024 11:56:04 [INFO|DP=0|PP=1|TP=28|ip-26-0-165-202]: Local number of parameters: 7.22G (13771.91MiB)
+12/28/2024 11:56:04 [INFO|DP=0|PP=0|TP=11|ip-26-0-160-242]: [After model building] Memory usage: 14212.07MiB. Peak allocated: 14212.25MiB Peak reserved: 14314.00MiB
+12/28/2024 11:56:04 [INFO|DP=0|PP=1|TP=29|ip-26-0-165-202]: [After model building] Memory usage: 13771.98MiB. Peak allocated: 13772.15MiB Peak reserved: 13886.00MiB
+12/28/2024 11:56:04 [INFO|DP=0|PP=1|TP=27|ip-26-0-165-202]: [After model building] Memory usage: 13771.98MiB. Peak allocated: 13772.15MiB Peak reserved: 13886.00MiB
+12/28/2024 11:56:04 [INFO|DP=0|PP=1|TP=26|ip-26-0-165-202]: Local number of parameters: 7.22G (13771.91MiB)
+12/28/2024 11:56:04 [INFO|DP=0|PP=1|TP=28|ip-26-0-165-202]: [After model building] Memory usage: 13771.98MiB. Peak allocated: 13772.15MiB Peak reserved: 13886.00MiB
+12/28/2024 11:56:04 [INFO|DP=0|PP=1|TP=26|ip-26-0-165-202]: [After model building] Memory usage: 13771.98MiB. Peak allocated: 13772.15MiB Peak reserved: 13886.00MiB
+12/28/2024 11:56:04 [INFO|DP=0|PP=1|TP=1|ip-26-0-161-78]: Local number of parameters: 7.22G (13771.91MiB)
+ip-26-0-161-153:2636085:2636657 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636085:2636657 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636085:2636657 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636085:2636657 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636085:2636657 [1] NCCL INFO Channel 00/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+12/28/2024 11:56:04 [INFO|DP=0|PP=1|TP=1|ip-26-0-161-78]: [After model building] Memory usage: 13771.98MiB. Peak allocated: 13772.15MiB Peak reserved: 13886.00MiB
+ip-26-0-161-153:2636085:2636657 [1] NCCL INFO Channel 01/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+12/28/2024 11:56:04 [INFO|DP=0|PP=1|TP=6|ip-26-0-161-78]: Local number of parameters: 7.22G (13771.91MiB)
+ip-26-0-161-153:2636085:2636657 [1] NCCL INFO Channel 02/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636085:2636657 [1] NCCL INFO Channel 03/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+12/28/2024 11:56:04 [INFO|DP=0|PP=1|TP=6|ip-26-0-161-78]: [After model building] Memory usage: 13771.98MiB. Peak allocated: 13772.15MiB Peak reserved: 13886.00MiB
+12/28/2024 11:56:04 [INFO|DP=0|PP=1|TP=4|ip-26-0-161-78]: Local number of parameters: 7.22G (13771.91MiB)
+12/28/2024 11:56:04 [INFO|DP=0|PP=1|TP=4|ip-26-0-161-78]: [After model building] Memory usage: 13771.98MiB. Peak allocated: 13772.15MiB Peak reserved: 13886.00MiB
+12/28/2024 11:56:04 [INFO|DP=0|PP=1|TP=25|ip-26-0-165-202]: Local number of parameters: 7.22G (13771.91MiB)
+12/28/2024 11:56:04 [INFO|DP=0|PP=1|TP=23|ip-26-0-165-164]: Local number of parameters: 7.22G (13771.91MiB)
+12/28/2024 11:56:04 [INFO|DP=0|PP=1|TP=25|ip-26-0-165-202]: [After model building] Memory usage: 13771.98MiB. Peak allocated: 13772.15MiB Peak reserved: 13886.00MiB
+12/28/2024 11:56:04 [INFO|DP=0|PP=1|TP=23|ip-26-0-165-164]: [After model building] Memory usage: 13771.98MiB. Peak allocated: 13772.15MiB Peak reserved: 13886.00MiB
+12/28/2024 11:56:04 [INFO|DP=0|PP=1|TP=24|ip-26-0-165-202]: Local number of parameters: 7.22G (13771.91MiB)
+12/28/2024 11:56:04 [INFO|DP=0|PP=1|TP=24|ip-26-0-165-202]: [After model building] Memory usage: 13771.98MiB. Peak allocated: 13772.15MiB Peak reserved: 13886.00MiB
+ip-26-0-161-142:62887:63423 [6] NCCL INFO Connected all rings
+ip-26-0-161-142:62887:63423 [6] NCCL INFO Connected all trees
+ip-26-0-161-142:62887:63423 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-142:62887:63423 [6] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-161-142:62887:63423 [6] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-161-142:62887:63423 [6] NCCL INFO comm 0xa87cdc0 rank 0 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0x2f2b62613eb35974 - Init COMPLETE
+ip-26-0-161-178:2815964:2816524 [1] NCCL INFO Connected all rings
+ip-26-0-161-178:2815964:2816524 [1] NCCL INFO Connected all trees
+ip-26-0-161-178:2815964:2816524 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-178:2815964:2816524 [1] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-161-178:2815964:2816524 [1] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-161-178:2815964:2816524 [1] NCCL INFO comm 0xa996570 rank 0 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0x60ec45dbd7f4bf99 - Init COMPLETE
+ip-26-0-168-52:2722046:2722580 [1] NCCL INFO Connected all rings
+ip-26-0-168-52:2722046:2722580 [1] NCCL INFO Connected all trees
+ip-26-0-168-52:2722046:2722580 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-52:2722046:2722580 [1] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-168-52:2722046:2722580 [1] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-168-30:235654:236182 [6] NCCL INFO Connected all rings
+ip-26-0-168-30:235654:236182 [6] NCCL INFO Connected all trees
+ip-26-0-168-30:235654:236182 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-30:235654:236182 [6] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-168-30:235654:236182 [6] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-161-178:2815967:2816521 [4] NCCL INFO Connected all rings
+ip-26-0-161-178:2815967:2816521 [4] NCCL INFO Connected all trees
+ip-26-0-161-178:2815967:2816521 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-178:2815967:2816521 [4] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-161-178:2815967:2816521 [4] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-161-178:2815966:2816526 [3] NCCL INFO Connected all rings
+ip-26-0-161-178:2815966:2816526 [3] NCCL INFO Connected all trees
+ip-26-0-161-178:2815966:2816526 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-178:2815966:2816526 [3] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-161-178:2815966:2816526 [3] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-168-52:2722046:2722580 [1] NCCL INFO comm 0xb829990 rank 1 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0x60ec45dbd7f4bf99 - Init COMPLETE
+ip-26-0-168-30:235654:236182 [6] NCCL INFO comm 0xaf27040 rank 1 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0x2f2b62613eb35974 - Init COMPLETE
+ip-26-0-168-52:2722048:2722581 [3] NCCL INFO Connected all rings
+ip-26-0-168-52:2722048:2722581 [3] NCCL INFO Connected all trees
+ip-26-0-168-52:2722048:2722581 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-52:2722048:2722581 [3] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-168-52:2722048:2722581 [3] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-161-178:2815968:2816533 [5] NCCL INFO Connected all rings
+ip-26-0-161-178:2815968:2816533 [5] NCCL INFO Connected all trees
+ip-26-0-161-178:2815968:2816533 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-178:2815968:2816533 [5] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-161-178:2815968:2816533 [5] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-161-178:2815967:2816521 [4] NCCL INFO comm 0x9ddfca0 rank 0 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0x564424f9e54a802c - Init COMPLETE
+ip-26-0-168-52:2722050:2722585 [5] NCCL INFO Connected all rings
+ip-26-0-168-52:2722050:2722585 [5] NCCL INFO Connected all trees
+ip-26-0-168-52:2722050:2722585 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-52:2722050:2722585 [5] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-168-52:2722050:2722585 [5] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-161-178:2815966:2816526 [3] NCCL INFO comm 0xb6a2800 rank 0 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0x30479eeb41b2f950 - Init COMPLETE
+ip-26-0-168-52:2722049:2722579 [4] NCCL INFO Connected all rings
+ip-26-0-168-52:2722049:2722579 [4] NCCL INFO Connected all trees
+ip-26-0-168-52:2722049:2722579 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-52:2722049:2722579 [4] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-168-52:2722049:2722579 [4] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-161-178:2815963:2816530 [0] NCCL INFO Connected all rings
+ip-26-0-161-178:2815963:2816530 [0] NCCL INFO Connected all trees
+ip-26-0-161-178:2815963:2816530 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-178:2815963:2816530 [0] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-161-178:2815963:2816530 [0] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-168-52:2722045:2722584 [0] NCCL INFO Connected all rings
+ip-26-0-168-52:2722045:2722584 [0] NCCL INFO Connected all trees
+ip-26-0-168-52:2722045:2722584 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-52:2722045:2722584 [0] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-168-52:2722045:2722584 [0] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-161-178:2815969:2816532 [6] NCCL INFO Connected all rings
+ip-26-0-161-178:2815969:2816532 [6] NCCL INFO Connected all trees
+ip-26-0-161-178:2815969:2816532 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-178:2815969:2816532 [6] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-161-178:2815969:2816532 [6] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-161-178:2815970:2816523 [7] NCCL INFO Connected all rings
+ip-26-0-161-178:2815970:2816523 [7] NCCL INFO Connected all trees
+ip-26-0-161-178:2815970:2816523 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-178:2815970:2816523 [7] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-161-178:2815970:2816523 [7] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-168-52:2722048:2722581 [3] NCCL INFO comm 0x9c25f50 rank 1 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0x30479eeb41b2f950 - Init COMPLETE
+ip-26-0-168-52:2722052:2722578 [7] NCCL INFO Connected all rings
+ip-26-0-168-52:2722052:2722578 [7] NCCL INFO Connected all trees
+ip-26-0-168-52:2722052:2722578 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-52:2722052:2722578 [7] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-168-52:2722052:2722578 [7] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-161-178:2815968:2816533 [5] NCCL INFO comm 0xc28c080 rank 0 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0x5034f31407405e64 - Init COMPLETE
+ip-26-0-161-178:2815965:2816528 [2] NCCL INFO Connected all rings
+ip-26-0-161-178:2815965:2816528 [2] NCCL INFO Connected all trees
+ip-26-0-161-178:2815965:2816528 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-178:2815965:2816528 [2] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-161-178:2815965:2816528 [2] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-168-52:2722050:2722585 [5] NCCL INFO comm 0x9c5a090 rank 1 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0x5034f31407405e64 - Init COMPLETE
+ip-26-0-168-52:2722049:2722579 [4] NCCL INFO comm 0x9df02f0 rank 1 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0x564424f9e54a802c - Init COMPLETE
+ip-26-0-168-52:2722051:2722583 [6] NCCL INFO Connected all rings
+ip-26-0-168-52:2722051:2722583 [6] NCCL INFO Connected all trees
+ip-26-0-168-52:2722051:2722583 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-52:2722051:2722583 [6] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-168-52:2722051:2722583 [6] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-161-178:2815963:2816530 [0] NCCL INFO comm 0xb57d690 rank 0 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0xce4b7fc6d9e0d602 - Init COMPLETE
+ip-26-0-161-178:2815969:2816532 [6] NCCL INFO comm 0xbc559e0 rank 0 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0xef09316ae5338288 - Init COMPLETE
+ip-26-0-168-52:2722045:2722584 [0] NCCL INFO comm 0xa567690 rank 1 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0xce4b7fc6d9e0d602 - Init COMPLETE
+ip-26-0-168-52:2722047:2722582 [2] NCCL INFO Connected all rings
+ip-26-0-168-52:2722047:2722582 [2] NCCL INFO Connected all trees
+ip-26-0-168-52:2722047:2722582 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-52:2722047:2722582 [2] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-168-52:2722047:2722582 [2] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-161-178:2815970:2816523 [7] NCCL INFO comm 0xb2c20f0 rank 0 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0xe7e10d0278883222 - Init COMPLETE
+ip-26-0-161-178:2815965:2816528 [2] NCCL INFO comm 0xadd1650 rank 0 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0x864e31bd77042a79 - Init COMPLETE
+ip-26-0-168-52:2722052:2722578 [7] NCCL INFO comm 0xac4c980 rank 1 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0xe7e10d0278883222 - Init COMPLETE
+ip-26-0-161-142:62883:63421 [2] NCCL INFO Connected all rings
+ip-26-0-161-142:62883:63421 [2] NCCL INFO Connected all trees
+ip-26-0-161-142:62883:63421 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-142:62883:63421 [2] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-161-142:62883:63421 [2] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-168-52:2722051:2722583 [6] NCCL INFO comm 0xa82da40 rank 1 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0xef09316ae5338288 - Init COMPLETE
+ip-26-0-168-52:2722047:2722582 [2] NCCL INFO comm 0xb461dc0 rank 1 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0x864e31bd77042a79 - Init COMPLETE
+ip-26-0-161-142:62888:63422 [7] NCCL INFO Connected all rings
+ip-26-0-161-142:62888:63422 [7] NCCL INFO Connected all trees
+ip-26-0-161-142:62888:63422 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-142:62888:63422 [7] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-161-142:62888:63422 [7] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-161-142:62883:63421 [2] NCCL INFO comm 0xad52dc0 rank 0 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0xdd28a795e7538c4 - Init COMPLETE
+ip-26-0-161-142:62882:63424 [1] NCCL INFO Connected all rings
+ip-26-0-161-142:62882:63424 [1] NCCL INFO Connected all trees
+ip-26-0-161-142:62882:63424 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-142:62882:63424 [1] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-161-142:62882:63424 [1] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-161-142:62888:63422 [7] NCCL INFO comm 0xb362750 rank 0 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0xeada9362f427fa83 - Init COMPLETE
+ip-26-0-161-142:62882:63424 [1] NCCL INFO comm 0xb0d54c0 rank 0 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0xc5f8d796be5089cb - Init COMPLETE
+ip-26-0-168-30:235650:236180 [2] NCCL INFO Connected all rings
+ip-26-0-168-30:235650:236180 [2] NCCL INFO Connected all trees
+ip-26-0-168-30:235650:236180 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-30:235650:236180 [2] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-168-30:235650:236180 [2] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-168-30:235649:236184 [1] NCCL INFO Connected all rings
+ip-26-0-168-30:235649:236184 [1] NCCL INFO Connected all trees
+ip-26-0-168-30:235649:236184 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-30:235649:236184 [1] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-168-30:235649:236184 [1] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-168-30:235651:236178 [3] NCCL INFO Connected all rings
+ip-26-0-168-30:235651:236178 [3] NCCL INFO Connected all trees
+ip-26-0-168-30:235651:236178 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-142:62884:63417 [3] NCCL INFO Connected all rings
+ip-26-0-161-142:62884:63417 [3] NCCL INFO Connected all trees
+ip-26-0-161-142:62884:63417 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-30:235651:236178 [3] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-168-30:235651:236178 [3] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-161-142:62884:63417 [3] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-161-142:62884:63417 [3] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-161-142:62886:63411 [5] NCCL INFO Connected all rings
+ip-26-0-161-142:62886:63411 [5] NCCL INFO Connected all trees
+ip-26-0-161-142:62886:63411 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-142:62886:63411 [5] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-161-142:62886:63411 [5] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-161-142:62881:63420 [0] NCCL INFO Connected all rings
+ip-26-0-161-142:62881:63420 [0] NCCL INFO Connected all trees
+ip-26-0-161-142:62881:63420 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-142:62881:63420 [0] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-161-142:62881:63420 [0] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-161-142:62885:63425 [4] NCCL INFO Connected all rings
+ip-26-0-161-142:62885:63425 [4] NCCL INFO Connected all trees
+ip-26-0-161-142:62885:63425 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-142:62885:63425 [4] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-161-142:62885:63425 [4] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-168-30:235650:236180 [2] NCCL INFO comm 0xaf1bad0 rank 1 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0xdd28a795e7538c4 - Init COMPLETE
+ip-26-0-168-30:235649:236184 [1] NCCL INFO comm 0xbc1a340 rank 1 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0xc5f8d796be5089cb - Init COMPLETE
+ip-26-0-161-142:62884:63417 [3] NCCL INFO comm 0xb7b8cb0 rank 0 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0x59659ca4831879fd - Init COMPLETE
+ip-26-0-168-30:235648:236179 [0] NCCL INFO Connected all rings
+ip-26-0-168-30:235648:236179 [0] NCCL INFO Connected all trees
+ip-26-0-168-30:235648:236179 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-30:235648:236179 [0] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-168-30:235648:236179 [0] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-168-30:235651:236178 [3] NCCL INFO comm 0xaefccc0 rank 1 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0x59659ca4831879fd - Init COMPLETE
+ip-26-0-161-142:62886:63411 [5] NCCL INFO comm 0xbfec390 rank 0 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0xac4fbfdf97289f71 - Init COMPLETE
+ip-26-0-168-30:235653:236177 [5] NCCL INFO Connected all rings
+ip-26-0-168-30:235653:236177 [5] NCCL INFO Connected all trees
+ip-26-0-168-30:235653:236177 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-30:235653:236177 [5] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-168-30:235653:236177 [5] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-168-30:235655:236181 [7] NCCL INFO Connected all rings
+ip-26-0-168-30:235655:236181 [7] NCCL INFO Connected all trees
+ip-26-0-168-30:235655:236181 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-30:235655:236181 [7] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-168-30:235655:236181 [7] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-161-142:62881:63420 [0] NCCL INFO comm 0xc328080 rank 0 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0x20c1c1aff1850363 - Init COMPLETE
+ip-26-0-161-142:62885:63425 [4] NCCL INFO comm 0xc01a5d0 rank 0 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0xfa0e0ba119a417d5 - Init COMPLETE
+ip-26-0-168-30:235652:236183 [4] NCCL INFO Connected all rings
+ip-26-0-168-30:235652:236183 [4] NCCL INFO Connected all trees
+ip-26-0-168-30:235652:236183 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-30:235652:236183 [4] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-168-30:235652:236183 [4] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-168-30:235648:236179 [0] NCCL INFO comm 0x9ee9b90 rank 1 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0x20c1c1aff1850363 - Init COMPLETE
+ip-26-0-168-30:235655:236181 [7] NCCL INFO comm 0xaa22540 rank 1 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0xeada9362f427fa83 - Init COMPLETE
+ip-26-0-168-30:235653:236177 [5] NCCL INFO comm 0xb129990 rank 1 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0xac4fbfdf97289f71 - Init COMPLETE
+ip-26-0-168-30:235652:236183 [4] NCCL INFO comm 0xbcaab00 rank 1 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0xfa0e0ba119a417d5 - Init COMPLETE
+ip-26-0-161-153:2636090:2636647 [6] NCCL INFO Connected all rings
+ip-26-0-161-153:2636090:2636647 [6] NCCL INFO Connected all trees
+ip-26-0-161-153:2636090:2636647 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-153:2636090:2636647 [6] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-161-153:2636090:2636647 [6] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-168-34:272156:272681 [6] NCCL INFO Connected all rings
+ip-26-0-168-34:272156:272681 [6] NCCL INFO Connected all trees
+ip-26-0-168-34:272156:272681 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-34:272156:272681 [6] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-168-34:272156:272681 [6] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-161-153:2636090:2636647 [6] NCCL INFO comm 0xbc16d40 rank 0 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0x336838438ba1cbae - Init COMPLETE
+ip-26-0-168-34:272156:272681 [6] NCCL INFO comm 0xc1be320 rank 1 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0x336838438ba1cbae - Init COMPLETE
+ip-26-0-161-153:2636088:2636653 [4] NCCL INFO Connected all rings
+ip-26-0-161-153:2636088:2636653 [4] NCCL INFO Connected all trees
+ip-26-0-161-153:2636088:2636653 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-153:2636088:2636653 [4] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-161-153:2636088:2636653 [4] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-168-34:272150:272688 [0] NCCL INFO Connected all rings
+ip-26-0-168-34:272150:272688 [0] NCCL INFO Connected all trees
+ip-26-0-168-34:272150:272688 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-34:272150:272688 [0] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-168-34:272150:272688 [0] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-168-34:272157:272687 [7] NCCL INFO Connected all rings
+ip-26-0-168-34:272157:272687 [7] NCCL INFO Connected all trees
+ip-26-0-168-34:272157:272687 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-34:272157:272687 [7] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-168-34:272157:272687 [7] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-161-153:2636091:2636656 [7] NCCL INFO Connected all rings
+ip-26-0-161-153:2636091:2636656 [7] NCCL INFO Connected all trees
+ip-26-0-161-153:2636091:2636656 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-153:2636091:2636656 [7] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-161-153:2636091:2636656 [7] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-161-153:2636084:2636659 [0] NCCL INFO Connected all rings
+ip-26-0-161-153:2636084:2636659 [0] NCCL INFO Connected all trees
+ip-26-0-161-153:2636084:2636659 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-153:2636084:2636659 [0] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-161-153:2636084:2636659 [0] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-161-153:2636087:2636655 [3] NCCL INFO Connected all rings
+ip-26-0-161-153:2636087:2636655 [3] NCCL INFO Connected all trees
+ip-26-0-161-153:2636087:2636655 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-153:2636087:2636655 [3] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-161-153:2636087:2636655 [3] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-161-153:2636089:2636654 [5] NCCL INFO Connected all rings
+ip-26-0-161-153:2636089:2636654 [5] NCCL INFO Connected all trees
+ip-26-0-161-153:2636089:2636654 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-153:2636089:2636654 [5] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-161-153:2636089:2636654 [5] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-168-34:272153:272685 [3] NCCL INFO Connected all rings
+ip-26-0-168-34:272153:272685 [3] NCCL INFO Connected all trees
+ip-26-0-168-34:272153:272685 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-34:272153:272685 [3] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-168-34:272153:272685 [3] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-161-153:2636088:2636653 [4] NCCL INFO comm 0xb4b0490 rank 0 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0xbf59abaaafdb1e45 - Init COMPLETE
+ip-26-0-168-34:272150:272688 [0] NCCL INFO comm 0xbd8a030 rank 1 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0x8be89873e2d856ad - Init COMPLETE
+ip-26-0-168-34:272154:272683 [4] NCCL INFO Connected all rings
+ip-26-0-168-34:272154:272683 [4] NCCL INFO Connected all trees
+ip-26-0-168-34:272154:272683 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-34:272154:272683 [4] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-168-34:272154:272683 [4] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-168-34:272157:272687 [7] NCCL INFO comm 0xa9d6240 rank 1 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0x56bd7b6de7070786 - Init COMPLETE
+ip-26-0-161-153:2636085:2636657 [1] NCCL INFO Connected all rings
+ip-26-0-161-153:2636085:2636657 [1] NCCL INFO Connected all trees
+ip-26-0-161-153:2636085:2636657 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-153:2636085:2636657 [1] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-161-153:2636085:2636657 [1] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-161-153:2636091:2636656 [7] NCCL INFO comm 0xb2113a0 rank 0 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0x56bd7b6de7070786 - Init COMPLETE
+ip-26-0-168-34:272155:272684 [5] NCCL INFO Connected all rings
+ip-26-0-168-34:272155:272684 [5] NCCL INFO Connected all trees
+ip-26-0-168-34:272155:272684 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-34:272155:272684 [5] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-168-34:272155:272684 [5] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-161-153:2636086:2636652 [2] NCCL INFO Connected all rings
+ip-26-0-161-153:2636086:2636652 [2] NCCL INFO Connected all trees
+ip-26-0-161-153:2636086:2636652 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-153:2636086:2636652 [2] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-161-153:2636086:2636652 [2] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-168-34:272152:272682 [2] NCCL INFO Connected all rings
+ip-26-0-168-34:272152:272682 [2] NCCL INFO Connected all trees
+ip-26-0-168-34:272152:272682 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-34:272152:272682 [2] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-168-34:272152:272682 [2] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-168-34:272153:272685 [3] NCCL INFO comm 0xa5de1a0 rank 1 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0x567d4d5e4099f109 - Init COMPLETE
+ip-26-0-161-153:2636087:2636655 [3] NCCL INFO comm 0xaab34a0 rank 0 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0x567d4d5e4099f109 - Init COMPLETE
+ip-26-0-161-153:2636084:2636659 [0] NCCL INFO comm 0xa708e10 rank 0 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0x8be89873e2d856ad - Init COMPLETE
+ip-26-0-161-153:2636089:2636654 [5] NCCL INFO comm 0xa92d820 rank 0 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0x5cd2fe565188f730 - Init COMPLETE
+ip-26-0-168-34:272151:272686 [1] NCCL INFO Connected all rings
+ip-26-0-168-34:272151:272686 [1] NCCL INFO Connected all trees
+ip-26-0-168-34:272151:272686 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-34:272151:272686 [1] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-168-34:272151:272686 [1] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-168-34:272154:272683 [4] NCCL INFO comm 0xb2a2df0 rank 1 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0xbf59abaaafdb1e45 - Init COMPLETE
+ip-26-0-161-153:2636085:2636657 [1] NCCL INFO comm 0xc3d3070 rank 0 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0xdbcc72fe4166d27e - Init COMPLETE
+ip-26-0-168-34:272155:272684 [5] NCCL INFO comm 0xb1cdf00 rank 1 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0x5cd2fe565188f730 - Init COMPLETE
+ip-26-0-161-153:2636086:2636652 [2] NCCL INFO comm 0xbb5f0d0 rank 0 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0x52002fcca2d62604 - Init COMPLETE
+ip-26-0-168-34:272152:272682 [2] NCCL INFO comm 0xc48fee0 rank 1 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0x52002fcca2d62604 - Init COMPLETE
+ip-26-0-168-34:272151:272686 [1] NCCL INFO comm 0xaa6e840 rank 1 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0xdbcc72fe4166d27e - Init COMPLETE
+ip-26-0-161-123:32704:33268 [0] NCCL INFO Using network Libfabric
+ip-26-0-160-225:124160:124730 [0] NCCL INFO Using network Libfabric
+ip-26-0-161-178:2815963:2816553 [0] NCCL INFO Using network Libfabric
+ip-26-0-161-138:96563:97171 [0] NCCL INFO Using network Libfabric
+ip-26-0-160-225:124161:124732 [1] NCCL INFO Using network Libfabric
+ip-26-0-161-138:96564:97172 [1] NCCL INFO Using network Libfabric
+ip-26-0-161-178:2815963:2816553 [0] NCCL INFO comm 0xb593b90 rank 1 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0x36840ec6b4075829 - Init START
+ip-26-0-161-138:96563:97171 [0] NCCL INFO comm 0xbbcef70 rank 1 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0xfa4852f4aecd80ec - Init START
+ip-26-0-161-123:32704:33268 [0] NCCL INFO comm 0xc675d50 rank 0 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0x36840ec6b4075829 - Init START
+ip-26-0-160-225:124160:124730 [0] NCCL INFO comm 0xb44d4a0 rank 0 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0xfa4852f4aecd80ec - Init START
+ip-26-0-161-123:32704:33268 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-160-225:124160:124730 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/541
+ip-26-0-161-178:2815963:2816553 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-138:96563:97171 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-138:96564:97172 [1] NCCL INFO comm 0xb83a1e0 rank 1 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0x833fbeee0b434137 - Init START
+ip-26-0-160-225:124161:124732 [1] NCCL INFO comm 0xaa0bbf0 rank 0 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0x833fbeee0b434137 - Init START
+ip-26-0-160-225:124161:124732 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-138:96564:97172 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-103:96889:97453 [1] NCCL INFO Using network Libfabric
+ip-26-0-161-153:2636085:2636678 [1] NCCL INFO Using network Libfabric
+ip-26-0-161-123:32707:33271 [3] NCCL INFO Using network Libfabric
+ip-26-0-161-123:32706:33272 [2] NCCL INFO Using network Libfabric
+ip-26-0-161-178:2815965:2816555 [2] NCCL INFO Using network Libfabric
+ip-26-0-161-178:2815966:2816554 [3] NCCL INFO Using network Libfabric
+ip-26-0-161-153:2636085:2636678 [1] NCCL INFO comm 0xc3dc510 rank 1 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0xb2131d0b36358051 - Init START
+ip-26-0-161-103:96889:97453 [1] NCCL INFO comm 0xa9d5170 rank 0 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0xb2131d0b36358051 - Init START
+ip-26-0-161-123:32711:33274 [7] NCCL INFO Using network Libfabric
+ip-26-0-161-178:2815970:2816556 [7] NCCL INFO Using network Libfabric
+ip-26-0-161-153:2636085:2636678 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-103:96889:97453 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-178:2815965:2816555 [2] NCCL INFO comm 0xade8a60 rank 1 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0x167967f1c8f1f887 - Init START
+ip-26-0-161-123:32706:33272 [2] NCCL INFO comm 0xbc15fa0 rank 0 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0x167967f1c8f1f887 - Init START
+ip-26-0-161-178:2815966:2816554 [3] NCCL INFO comm 0xb6b8f90 rank 1 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0xc2f845d8f7e6ca39 - Init START
+ip-26-0-161-123:32707:33271 [3] NCCL INFO comm 0xaa77e30 rank 0 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0xc2f845d8f7e6ca39 - Init START
+ip-26-0-161-178:2815965:2816555 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-123:32706:33272 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-178:2815967:2816557 [4] NCCL INFO Using network Libfabric
+ip-26-0-161-123:32707:33271 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-178:2815966:2816554 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-123:32708:33276 [4] NCCL INFO Using network Libfabric
+ip-26-0-161-178:2815970:2816556 [7] NCCL INFO comm 0xb2d90c0 rank 1 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0x6be549eda214aab5 - Init START
+ip-26-0-161-123:32711:33274 [7] NCCL INFO comm 0xc082f40 rank 0 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0x6be549eda214aab5 - Init START
+ip-26-0-161-178:2815970:2816556 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-123:32705:33278 [1] NCCL INFO Using network Libfabric
+ip-26-0-161-123:32711:33274 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-178:2815964:2816558 [1] NCCL INFO Using network Libfabric
+ip-26-0-160-242:1364902:1365483 [0] NCCL INFO Using network Libfabric
+ip-26-0-161-142:62881:63445 [0] NCCL INFO Using network Libfabric
+ip-26-0-161-178:2815967:2816557 [4] NCCL INFO comm 0xa843590 rank 1 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0xe123c9967146b25f - Init START
+ip-26-0-161-123:32708:33276 [4] NCCL INFO comm 0xc48b100 rank 0 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0xe123c9967146b25f - Init START
+ip-26-0-161-178:2815968:2816559 [5] NCCL INFO Using network Libfabric
+ip-26-0-161-123:32708:33276 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-123:32709:33280 [5] NCCL INFO Using network Libfabric
+ip-26-0-161-178:2815967:2816557 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-178:2815964:2816558 [1] NCCL INFO comm 0xa9a5400 rank 1 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0xfd0f632a334dff00 - Init START
+ip-26-0-161-123:32705:33278 [1] NCCL INFO comm 0xbc02d30 rank 0 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0xfd0f632a334dff00 - Init START
+ip-26-0-161-123:32705:33278 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-123:32710:33282 [6] NCCL INFO Using network Libfabric
+ip-26-0-161-178:2815964:2816558 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-178:2815969:2816560 [6] NCCL INFO Using network Libfabric
+ip-26-0-161-142:62881:63445 [0] NCCL INFO comm 0xc311190 rank 1 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0xdc2325cededd95f8 - Init START
+ip-26-0-160-242:1364902:1365483 [0] NCCL INFO comm 0xb9c9000 rank 0 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0xdc2325cededd95f8 - Init START
+ip-26-0-161-178:2815968:2816559 [5] NCCL INFO comm 0xc2a34e0 rank 1 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0xba53db25a1463133 - Init START
+ip-26-0-161-123:32709:33280 [5] NCCL INFO comm 0xc45a330 rank 0 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0xba53db25a1463133 - Init START
+ip-26-0-161-142:62881:63445 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-160-242:1364902:1365483 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-123:32709:33280 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-178:2815968:2816559 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-103:96894:97456 [6] NCCL INFO Using network Libfabric
+ip-26-0-160-225:124164:124734 [4] NCCL INFO Using network Libfabric
+ip-26-0-161-138:96567:97173 [4] NCCL INFO Using network Libfabric
+ip-26-0-161-153:2636090:2636679 [6] NCCL INFO Using network Libfabric
+ip-26-0-161-103:96890:97457 [2] NCCL INFO Using network Libfabric
+ip-26-0-161-178:2815969:2816560 [6] NCCL INFO comm 0xbc6d8e0 rank 1 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0x457d552d1a4664ec - Init START
+ip-26-0-161-103:96888:97459 [0] NCCL INFO Using network Libfabric
+ip-26-0-161-123:32710:33282 [6] NCCL INFO comm 0xbb37cb0 rank 0 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0x457d552d1a4664ec - Init START
+ip-26-0-161-138:96568:97174 [5] NCCL INFO Using network Libfabric
+ip-26-0-160-225:124165:124736 [5] NCCL INFO Using network Libfabric
+ip-26-0-161-123:32710:33282 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+NCCL version 2.18.5+cuda12.2
+ip-26-0-161-178:2815969:2816560 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-153:2636086:2636680 [2] NCCL INFO Using network Libfabric
+ip-26-0-160-242:1364909:1365485 [7] NCCL INFO Using network Libfabric
+ip-26-0-161-153:2636084:2636681 [0] NCCL INFO Using network Libfabric
+NCCL version 2.18.5+cuda12.2
+ip-26-0-160-225:124162:124740 [2] NCCL INFO Using network Libfabric
+NCCL version 2.18.5+cuda12.2
+ip-26-0-161-103:96893:97462 [5] NCCL INFO Using network Libfabric
+ip-26-0-160-225:124166:124742 [6] NCCL INFO Using network Libfabric
+ip-26-0-160-225:124163:124743 [3] NCCL INFO Using network Libfabric
+ip-26-0-160-242:1364906:1365487 [4] NCCL INFO Using network Libfabric
+ip-26-0-161-138:96565:97175 [2] NCCL INFO Using network Libfabric
+ip-26-0-160-225:124167:124744 [7] NCCL INFO Using network Libfabric
+ip-26-0-161-103:96891:97463 [3] NCCL INFO Using network Libfabric
+ip-26-0-160-225:124164:124734 [4] NCCL INFO comm 0xb16e730 rank 0 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0x174515b09e2bfef0 - Init START
+ip-26-0-161-138:96567:97173 [4] NCCL INFO comm 0xaa843d0 rank 1 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0x174515b09e2bfef0 - Init START
+ip-26-0-161-142:62888:63446 [7] NCCL INFO Using network Libfabric
+NCCL version 2.18.5+cuda12.2
+ip-26-0-161-138:96570:97178 [7] NCCL INFO Using network Libfabric
+ip-26-0-161-153:2636089:2636682 [5] NCCL INFO Using network Libfabric
+ip-26-0-161-138:96566:97176 [3] NCCL INFO Using network Libfabric
+ip-26-0-161-138:96569:97177 [6] NCCL INFO Using network Libfabric
+NCCL version 2.18.5+cuda12.2
+ip-26-0-161-103:96894:97456 [6] NCCL INFO comm 0xac95050 rank 0 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0x8582362b5b377cb8 - Init START
+ip-26-0-160-225:124164:124734 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-142:62885:63447 [4] NCCL INFO Using network Libfabric
+ip-26-0-161-103:96895:97465 [7] NCCL INFO Using network Libfabric
+ip-26-0-160-242:1364907:1365491 [5] NCCL INFO Using network Libfabric
+ip-26-0-161-138:96567:97173 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-153:2636090:2636679 [6] NCCL INFO comm 0xbc324a0 rank 1 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0x8582362b5b377cb8 - Init START
+NCCL version 2.18.5+cuda12.2
+ip-26-0-160-242:1364904:1365489 [2] NCCL INFO Using network Libfabric
+ip-26-0-161-142:62883:63448 [2] NCCL INFO Using network Libfabric
+ip-26-0-161-138:96568:97174 [5] NCCL INFO comm 0xaadad80 rank 1 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0x3b05320dbc041b06 - Init START
+NCCL version 2.18.5+cuda12.2
+ip-26-0-160-225:124165:124736 [5] NCCL INFO comm 0xaa4d3d0 rank 0 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0x3b05320dbc041b06 - Init START
+ip-26-0-161-153:2636087:2636683 [3] NCCL INFO Using network Libfabric
+ip-26-0-161-142:62886:63449 [5] NCCL INFO Using network Libfabric
+ip-26-0-161-103:96894:97456 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-153:2636090:2636679 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-160-225:124165:124736 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-103:96890:97457 [2] NCCL INFO comm 0xb01a390 rank 0 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0x15a67f5c54f9ae39 - Init START
+ip-26-0-161-153:2636086:2636680 [2] NCCL INFO comm 0xbb75650 rank 1 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0x15a67f5c54f9ae39 - Init START
+ip-26-0-161-138:96568:97174 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-52:2722052:2722603 [7] NCCL INFO Using network Libfabric
+ip-26-0-165-202:153564:154120 [2] NCCL INFO Using network Libfabric
+ip-26-0-161-153:2636091:2636684 [7] NCCL INFO Using network Libfabric
+ip-26-0-161-103:96888:97459 [0] NCCL INFO comm 0xb03f0a0 rank 0 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0x67b3bc8359a8fd69 - Init START
+ip-26-0-165-202:153569:154118 [7] NCCL INFO Using network Libfabric
+ip-26-0-161-153:2636084:2636681 [0] NCCL INFO comm 0xb1631c0 rank 1 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0x67b3bc8359a8fd69 - Init START
+ip-26-0-161-103:96890:97457 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-153:2636086:2636680 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-138:96565:97175 [2] NCCL INFO comm 0xc420790 rank 1 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0x1a0854df454e0f6f - Init START
+NCCL version 2.18.5+cuda12.2
+ip-26-0-160-225:124162:124740 [2] NCCL INFO comm 0xc4a9dc0 rank 0 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0x1a0854df454e0f6f - Init START
+ip-26-0-168-52:2722047:2722604 [2] NCCL INFO Using network Libfabric
+ip-26-0-165-202:153565:154121 [3] NCCL INFO Using network Libfabric
+ip-26-0-161-153:2636084:2636681 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-103:96888:97459 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-160-225:124163:124743 [3] NCCL INFO comm 0xb6e2ee0 rank 0 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0xf118ca765e2897d0 - Init START
+ip-26-0-168-52:2722048:2722605 [3] NCCL INFO Using network Libfabric
+ip-26-0-161-138:96566:97176 [3] NCCL INFO comm 0xb52cf50 rank 1 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0xf118ca765e2897d0 - Init START
+ip-26-0-160-225:124162:124740 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-160-225:124167:124744 [7] NCCL INFO comm 0xbe5bb90 rank 0 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0x6dc7a8370918fa1 - Init START
+ip-26-0-161-142:62888:63446 [7] NCCL INFO comm 0xb379ed0 rank 1 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0xfdfbbee8bd17a511 - Init START
+NCCL version 2.18.5+cuda12.2
+ip-26-0-161-138:96570:97178 [7] NCCL INFO comm 0xc2d74c0 rank 1 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0x6dc7a8370918fa1 - Init START
+ip-26-0-160-242:1364909:1365485 [7] NCCL INFO comm 0xc067b00 rank 0 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0xfdfbbee8bd17a511 - Init START
+ip-26-0-160-225:124163:124743 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-160-225:124166:124742 [6] NCCL INFO comm 0xb971350 rank 0 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0x73aa7cd217337991 - Init START
+ip-26-0-161-138:96565:97175 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-202:153566:154124 [4] NCCL INFO Using network Libfabric
+ip-26-0-161-153:2636089:2636682 [5] NCCL INFO comm 0xa9432a0 rank 1 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0x20c1a1bfbac058 - Init START
+ip-26-0-161-103:96893:97462 [5] NCCL INFO comm 0xa8ed100 rank 0 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0x20c1a1bfbac058 - Init START
+ip-26-0-161-138:96569:97177 [6] NCCL INFO comm 0xb086aa0 rank 1 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0x73aa7cd217337991 - Init START
+ip-26-0-160-225:124167:124744 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-138:96566:97176 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-78:114513:115070 [1] NCCL INFO Using network Libfabric
+ip-26-0-160-242:1364909:1365485 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-142:62885:63447 [4] NCCL INFO comm 0xc030ad0 rank 1 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0x1cec219b593b5b6e - Init START
+ip-26-0-160-225:124166:124742 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-59:1557267:1557821 [1] NCCL INFO Using network Libfabric
+ip-26-0-161-138:96570:97178 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-138:96569:97177 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-160-242:1364906:1365487 [4] NCCL INFO comm 0xc1abec0 rank 0 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0x1cec219b593b5b6e - Init START
+ip-26-0-168-52:2722049:2722606 [4] NCCL INFO Using network Libfabric
+ip-26-0-161-142:62888:63446 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-142:62883:63448 [2] NCCL INFO comm 0xad3c290 rank 1 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0x3b173d6ca378a933 - Init START
+ip-26-0-161-103:96893:97462 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-160-242:1364904:1365489 [2] NCCL INFO comm 0xb840bf0 rank 0 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0x3b173d6ca378a933 - Init START
+ip-26-0-161-153:2636089:2636682 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-202:153567:154126 [5] NCCL INFO Using network Libfabric
+ip-26-0-165-202:153568:154127 [6] NCCL INFO Using network Libfabric
+ip-26-0-160-242:1364906:1365487 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-142:62882:63450 [1] NCCL INFO Using network Libfabric
+ip-26-0-161-142:62885:63447 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-160-242:1364904:1365489 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-153:2636087:2636683 [3] NCCL INFO comm 0xaac9fb0 rank 1 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0xa1e643239b1aeb8 - Init START
+ip-26-0-161-103:96891:97463 [3] NCCL INFO comm 0xae75180 rank 0 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0xa1e643239b1aeb8 - Init START
+ip-26-0-161-142:62886:63449 [5] NCCL INFO comm 0xc002a90 rank 1 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0xb5b1e5768650a111 - Init START
+ip-26-0-160-242:1364907:1365491 [5] NCCL INFO comm 0xa8af0a0 rank 0 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0xb5b1e5768650a111 - Init START
+ip-26-0-161-142:62883:63448 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-160-242:1364903:1365493 [1] NCCL INFO Using network Libfabric
+ip-26-0-168-52:2722051:2722608 [6] NCCL INFO Using network Libfabric
+NCCL version 2.18.5+cuda12.2
+ip-26-0-161-153:2636091:2636684 [7] NCCL INFO comm 0xb227b50 rank 1 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0x2eb8a3b534e743f8 - Init START
+ip-26-0-161-103:96895:97465 [7] NCCL INFO comm 0xc63f250 rank 0 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0x2eb8a3b534e743f8 - Init START
+ip-26-0-168-52:2722050:2722607 [5] NCCL INFO Using network Libfabric
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+ip-26-0-160-242:1364907:1365491 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-153:2636087:2636683 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-142:62886:63449 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-103:96891:97463 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-52:2722052:2722603 [7] NCCL INFO comm 0xb691400 rank 1 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0x8ccc8b29b5112548 - Init START
+ip-26-0-165-202:153569:154118 [7] NCCL INFO comm 0xa7463a0 rank 0 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0x8ccc8b29b5112548 - Init START
+NCCL version 2.18.5+cuda12.2
+ip-26-0-161-153:2636091:2636684 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-103:96895:97465 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-52:2722047:2722604 [2] NCCL INFO comm 0xb440480 rank 1 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0x7afd9f7327436e47 - Init START
+ip-26-0-165-202:153564:154120 [2] NCCL INFO comm 0xa7dd790 rank 0 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0x7afd9f7327436e47 - Init START
+ip-26-0-161-78:114512:115074 [0] NCCL INFO Using network Libfabric
+ip-26-0-168-52:2722052:2722603 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-52:2722048:2722605 [3] NCCL INFO comm 0xa66a8f0 rank 1 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0x6290c10a08bf7fdd - Init START
+ip-26-0-165-202:153569:154118 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-59:1557268:1557822 [2] NCCL INFO Using network Libfabric
+ip-26-0-161-78:114514:115073 [2] NCCL INFO Using network Libfabric
+ip-26-0-168-52:2722047:2722604 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-59:1557266:1557823 [0] NCCL INFO Using network Libfabric
+ip-26-0-165-202:153565:154121 [3] NCCL INFO comm 0xb9b5da0 rank 0 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0x6290c10a08bf7fdd - Init START
+ip-26-0-165-202:153564:154120 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-78:114517:115076 [5] NCCL INFO Using network Libfabric
+ip-26-0-165-202:153565:154121 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-59:1557271:1557824 [5] NCCL INFO Using network Libfabric
+ip-26-0-168-52:2722048:2722605 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-160-242:1364908:1365495 [6] NCCL INFO Using network Libfabric
+ip-26-0-165-202:153562:154129 [0] NCCL INFO Using network Libfabric
+ip-26-0-168-52:2722049:2722606 [4] NCCL INFO comm 0xa834490 rank 1 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0x9924b60493ed5710 - Init START
+ip-26-0-161-103:96892:97467 [4] NCCL INFO Using network Libfabric
+ip-26-0-165-202:153566:154124 [4] NCCL INFO comm 0xaa282c0 rank 0 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0x9924b60493ed5710 - Init START
+ip-26-0-161-142:62887:63451 [6] NCCL INFO Using network Libfabric
+ip-26-0-165-59:1557267:1557821 [1] NCCL INFO comm 0xa8bb440 rank 1 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0x317366f90460137 - Init START
+ip-26-0-160-242:1364903:1365493 [1] NCCL INFO comm 0xbcd3920 rank 0 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0x5a8efde32afc8c2a - Init START
+ip-26-0-165-131:1630205:1630805 [0] NCCL INFO Using network Libfabric
+ip-26-0-161-142:62882:63450 [1] NCCL INFO comm 0xb0ebc10 rank 1 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0x5a8efde32afc8c2a - Init START
+ip-26-0-165-202:153566:154124 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-160-242:1364905:1365497 [3] NCCL INFO Using network Libfabric
+ip-26-0-161-153:2636088:2636685 [4] NCCL INFO Using network Libfabric
+ip-26-0-161-78:114513:115070 [1] NCCL INFO comm 0xbe51a40 rank 0 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0x317366f90460137 - Init START
+ip-26-0-168-52:2722049:2722606 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-52:2722051:2722608 [6] NCCL INFO comm 0xb2724b0 rank 1 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0xd2565f661f0f5428 - Init START
+ip-26-0-161-142:62882:63450 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-160-242:1364903:1365493 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-202:153568:154127 [6] NCCL INFO comm 0xa7cc410 rank 0 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0xd2565f661f0f5428 - Init START
+ip-26-0-165-202:153567:154126 [5] NCCL INFO comm 0xc1e41e0 rank 0 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0xa28c5a970ab72e03 - Init START
+ip-26-0-168-30:235648:236202 [0] NCCL INFO Using network Libfabric
+ip-26-0-168-52:2722050:2722607 [5] NCCL INFO comm 0xa69e960 rank 1 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0xa28c5a970ab72e03 - Init START
+ip-26-0-165-131:1630209:1630807 [4] NCCL INFO Using network Libfabric
+ip-26-0-165-59:1557267:1557821 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-78:114513:115070 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-202:153568:154127 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-52:2722045:2722609 [0] NCCL INFO Using network Libfabric
+ip-26-0-165-202:153567:154126 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-78:114515:115081 [3] NCCL INFO Using network Libfabric
+ip-26-0-165-59:1557270:1557825 [4] NCCL INFO Using network Libfabric
+ip-26-0-161-78:114518:115082 [6] NCCL INFO Using network Libfabric
+ip-26-0-168-52:2722050:2722607 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-52:2722051:2722608 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-78:114516:115080 [4] NCCL INFO Using network Libfabric
+ip-26-0-161-142:62884:63452 [3] NCCL INFO Using network Libfabric
+ip-26-0-165-59:1557269:1557826 [3] NCCL INFO Using network Libfabric
+ip-26-0-165-59:1557266:1557823 [0] NCCL INFO comm 0xc141870 rank 1 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0xd3ea008df278b7a5 - Init START
+ip-26-0-165-59:1557268:1557822 [2] NCCL INFO comm 0xab5f560 rank 1 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0x3a70f37d11c42078 - Init START
+ip-26-0-165-59:1557272:1557827 [6] NCCL INFO Using network Libfabric
+ip-26-0-161-78:114512:115074 [0] NCCL INFO comm 0xb2a1520 rank 0 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0xd3ea008df278b7a5 - Init START
+ip-26-0-168-30:235652:236203 [4] NCCL INFO Using network Libfabric
+ip-26-0-161-78:114514:115073 [2] NCCL INFO comm 0xbd14d10 rank 0 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0x3a70f37d11c42078 - Init START
+ip-26-0-165-59:1557271:1557824 [5] NCCL INFO comm 0xa4c5ba0 rank 1 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0xf54fd7bc2ff2310e - Init START
+ip-26-0-161-78:114514:115073 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-59:1557266:1557823 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-59:1557268:1557822 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+NCCL version 2.18.5+cuda12.2
+ip-26-0-161-78:114517:115076 [5] NCCL INFO comm 0xc29fc50 rank 0 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0xf54fd7bc2ff2310e - Init START
+ip-26-0-161-78:114512:115074 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-142:62887:63451 [6] NCCL INFO comm 0xa894680 rank 1 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0x7ba2dc144f01ded2 - Init START
+ip-26-0-160-242:1364908:1365495 [6] NCCL INFO comm 0xc5c5120 rank 0 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0x7ba2dc144f01ded2 - Init START
+ip-26-0-161-78:114517:115076 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-59:1557271:1557824 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-153:2636088:2636685 [4] NCCL INFO comm 0xb4c73e0 rank 1 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0x142d040e61415d48 - Init START
+ip-26-0-161-103:96892:97467 [4] NCCL INFO comm 0xbdbb170 rank 0 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0x142d040e61415d48 - Init START
+NCCL version 2.18.5+cuda12.2
+ip-26-0-160-242:1364908:1365495 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-142:62887:63451 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-52:2722045:2722609 [0] NCCL INFO comm 0xa54d320 rank 1 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0x7a9a6457da866605 - Init START
+ip-26-0-168-30:235648:236202 [0] NCCL INFO comm 0xa92e1d0 rank 1 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0xa38d5ad285b7dfb1 - Init START
+ip-26-0-161-153:2636088:2636685 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-103:96892:97467 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-202:153562:154129 [0] NCCL INFO comm 0xb4ad5a0 rank 0 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0x7a9a6457da866605 - Init START
+ip-26-0-165-131:1630205:1630805 [0] NCCL INFO comm 0xae8cb50 rank 0 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0xa38d5ad285b7dfb1 - Init START
+ip-26-0-161-142:62884:63452 [3] NCCL INFO comm 0xb7c26d0 rank 1 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0x2007c24124eb1e16 - Init START
+ip-26-0-160-242:1364905:1365497 [3] NCCL INFO comm 0xba43580 rank 0 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0x2007c24124eb1e16 - Init START
+ip-26-0-168-52:2722045:2722609 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-202:153562:154129 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-78:114516:115080 [4] NCCL INFO comm 0xb3894c0 rank 0 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0x13fc5839256a0f0a - Init START
+ip-26-0-165-131:1630205:1630805 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/409
+ip-26-0-165-59:1557270:1557825 [4] NCCL INFO comm 0xa934a50 rank 1 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0x13fc5839256a0f0a - Init START
+ip-26-0-168-30:235648:236202 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-59:1557269:1557826 [3] NCCL INFO comm 0xbf10700 rank 1 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0xf02067e7b46b58ac - Init START
+ip-26-0-160-242:1364905:1365497 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-78:114515:115081 [3] NCCL INFO comm 0xadec590 rank 0 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0xf02067e7b46b58ac - Init START
+ip-26-0-161-142:62884:63452 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-59:1557272:1557827 [6] NCCL INFO comm 0xb426cd0 rank 1 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0x401d2ca93e58b7b3 - Init START
+ip-26-0-165-59:1557270:1557825 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-30:235652:236203 [4] NCCL INFO comm 0xbc9fd60 rank 1 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0x2d3503a923c539a9 - Init START
+ip-26-0-161-78:114516:115080 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-131:1630209:1630807 [4] NCCL INFO comm 0xb1d0150 rank 0 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0x2d3503a923c539a9 - Init START
+ip-26-0-161-78:114518:115082 [6] NCCL INFO comm 0xb895930 rank 0 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0x401d2ca93e58b7b3 - Init START
+ip-26-0-165-59:1557269:1557826 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-78:114515:115081 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+NCCL version 2.18.5+cuda12.2
+ip-26-0-165-131:1630209:1630807 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/409
+ip-26-0-165-131:1630210:1630809 [5] NCCL INFO Using network Libfabric
+NCCL version 2.18.5+cuda12.2
+ip-26-0-165-59:1557272:1557827 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-161-78:114518:115082 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-30:235652:236203 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-30:235653:236204 [5] NCCL INFO Using network Libfabric
+ip-26-0-165-131:1630208:1630811 [3] NCCL INFO Using network Libfabric
+ip-26-0-168-30:235651:236205 [3] NCCL INFO Using network Libfabric
+NCCL version 2.18.5+cuda12.2
+ip-26-0-165-131:1630211:1630813 [6] NCCL INFO Using network Libfabric
+NCCL version 2.18.5+cuda12.2
+ip-26-0-165-164:1227467:1228021 [4] NCCL INFO Using network Libfabric
+NCCL version 2.18.5+cuda12.2
+ip-26-0-168-30:235654:236206 [6] NCCL INFO Using network Libfabric
+ip-26-0-168-30:235653:236204 [5] NCCL INFO comm 0xb11ebf0 rank 1 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0x9fbcd95ac374d17c - Init START
+ip-26-0-165-131:1630210:1630809 [5] NCCL INFO comm 0xbeb2750 rank 0 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0x9fbcd95ac374d17c - Init START
+ip-26-0-168-30:235651:236205 [3] NCCL INFO comm 0xa4a2b40 rank 1 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0xcda45e7cac3649b2 - Init START
+ip-26-0-165-131:1630208:1630811 [3] NCCL INFO comm 0xa4e7ad0 rank 0 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0xcda45e7cac3649b2 - Init START
+ip-26-0-165-131:1630210:1630809 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/409
+ip-26-0-168-34:272154:272706 [4] NCCL INFO Using network Libfabric
+ip-26-0-168-30:235653:236204 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-131:1630208:1630811 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/409
+ip-26-0-168-30:235651:236205 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+NCCL version 2.18.5+cuda12.2
+ip-26-0-165-131:1630207:1630815 [2] NCCL INFO Using network Libfabric
+NCCL version 2.18.5+cuda12.2
+ip-26-0-168-30:235650:236207 [2] NCCL INFO Using network Libfabric
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+ip-26-0-168-30:235654:236206 [6] NCCL INFO comm 0xaf1c480 rank 1 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0xab67ef3e7d9c3e72 - Init START
+ip-26-0-165-131:1630211:1630813 [6] NCCL INFO comm 0xc48d080 rank 0 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0xab67ef3e7d9c3e72 - Init START
+ip-26-0-165-202:153563:154131 [1] NCCL INFO Using network Libfabric
+ip-26-0-165-164:1227468:1228023 [5] NCCL INFO Using network Libfabric
+NCCL version 2.18.5+cuda12.2
+ip-26-0-168-34:272155:272707 [5] NCCL INFO Using network Libfabric
+ip-26-0-165-131:1630211:1630813 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/409
+ip-26-0-168-52:2722046:2722610 [1] NCCL INFO Using network Libfabric
+ip-26-0-168-30:235654:236206 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-164:1227469:1228026 [6] NCCL INFO Using network Libfabric
+ip-26-0-165-164:1227467:1228021 [4] NCCL INFO comm 0xb21a6e0 rank 0 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0x162146216445f9b - Init START
+ip-26-0-168-34:272154:272706 [4] NCCL INFO comm 0xb298050 rank 1 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0x162146216445f9b - Init START
+ip-26-0-168-30:235650:236207 [2] NCCL INFO comm 0xaf10d30 rank 1 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0xafc8bbf8e2a35859 - Init START
+ip-26-0-165-131:1630207:1630815 [2] NCCL INFO comm 0xb568310 rank 0 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0xafc8bbf8e2a35859 - Init START
+ip-26-0-168-34:272156:272708 [6] NCCL INFO Using network Libfabric
+ip-26-0-165-164:1227465:1228028 [2] NCCL INFO Using network Libfabric
+ip-26-0-165-164:1227464:1228029 [1] NCCL INFO Using network Libfabric
+ip-26-0-168-34:272154:272706 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-164:1227467:1228021 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-131:1630207:1630815 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/409
+ip-26-0-168-30:235650:236207 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-34:272152:272709 [2] NCCL INFO Using network Libfabric
+ip-26-0-168-34:272151:272710 [1] NCCL INFO Using network Libfabric
+ip-26-0-165-164:1227466:1228032 [3] NCCL INFO Using network Libfabric
+ip-26-0-168-34:272155:272707 [5] NCCL INFO comm 0xb1c3160 rank 1 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0x218724ca16147a07 - Init START
+ip-26-0-168-52:2722046:2722610 [1] NCCL INFO comm 0xb81eb90 rank 1 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0xd27583e35160699c - Init START
+ip-26-0-165-164:1227470:1228033 [7] NCCL INFO Using network Libfabric
+ip-26-0-165-202:153563:154131 [1] NCCL INFO comm 0xc3e7cc0 rank 0 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0xd27583e35160699c - Init START
+ip-26-0-168-34:272153:272711 [3] NCCL INFO Using network Libfabric
+ip-26-0-165-164:1227468:1228023 [5] NCCL INFO comm 0xbba64c0 rank 0 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0x218724ca16147a07 - Init START
+ip-26-0-165-202:153563:154131 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-164:1227468:1228023 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-52:2722046:2722610 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-34:272157:272712 [7] NCCL INFO Using network Libfabric
+ip-26-0-168-34:272155:272707 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-34:272156:272708 [6] NCCL INFO comm 0xc1b3520 rank 1 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0xff9654fa091a52a3 - Init START
+ip-26-0-165-164:1227469:1228026 [6] NCCL INFO comm 0xa4d9fc0 rank 0 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0xff9654fa091a52a3 - Init START
+ip-26-0-168-34:272152:272709 [2] NCCL INFO comm 0xc485140 rank 1 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0x6f5a7a271888c3dd - Init START
+ip-26-0-165-164:1227469:1228026 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-164:1227465:1228028 [2] NCCL INFO comm 0xac45420 rank 0 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0x6f5a7a271888c3dd - Init START
+ip-26-0-168-34:272156:272708 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-164:1227463:1228035 [0] NCCL INFO Using network Libfabric
+ip-26-0-168-34:272150:272713 [0] NCCL INFO Using network Libfabric
+ip-26-0-165-164:1227464:1228029 [1] NCCL INFO comm 0xb080790 rank 0 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0xbd511d17f9a0d302 - Init START
+ip-26-0-168-34:272151:272710 [1] NCCL INFO comm 0xb4b3760 rank 1 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0xbd511d17f9a0d302 - Init START
+ip-26-0-168-34:272152:272709 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-34:272153:272711 [3] NCCL INFO comm 0xa5d3400 rank 1 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0x33412db31e485107 - Init START
+ip-26-0-165-164:1227465:1228028 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-164:1227466:1228032 [3] NCCL INFO comm 0xae9ae50 rank 0 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0x33412db31e485107 - Init START
+ip-26-0-168-34:272151:272710 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-164:1227464:1228029 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-164:1227466:1228032 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-34:272157:272712 [7] NCCL INFO comm 0xb41a9a0 rank 1 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0x3b2a736c8d71389 - Init START
+ip-26-0-168-34:272153:272711 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-164:1227470:1228033 [7] NCCL INFO comm 0xb2b5560 rank 0 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0x3b2a736c8d71389 - Init START
+ip-26-0-165-164:1227470:1228033 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-34:272157:272712 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-34:272150:272713 [0] NCCL INFO comm 0xbd7f290 rank 1 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0x390ca62f48410b65 - Init START
+ip-26-0-165-164:1227463:1228035 [0] NCCL INFO comm 0xa7b8b10 rank 0 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0x390ca62f48410b65 - Init START
+ip-26-0-165-164:1227463:1228035 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-34:272150:272713 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+ip-26-0-165-131:1630206:1630817 [1] NCCL INFO Using network Libfabric
+NCCL version 2.18.5+cuda12.2
+ip-26-0-168-30:235649:236208 [1] NCCL INFO Using network Libfabric
+ip-26-0-165-131:1630212:1630819 [7] NCCL INFO Using network Libfabric
+ip-26-0-168-30:235655:236209 [7] NCCL INFO Using network Libfabric
+ip-26-0-168-30:235649:236208 [1] NCCL INFO comm 0xbbf8980 rank 1 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0xfd6fa36a1b1a13e3 - Init START
+ip-26-0-161-78:114519:115084 [7] NCCL INFO Using network Libfabric
+ip-26-0-165-131:1630206:1630817 [1] NCCL INFO comm 0xbb509d0 rank 0 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0xfd6fa36a1b1a13e3 - Init START
+ip-26-0-165-131:1630206:1630817 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/409
+ip-26-0-165-59:1557273:1557828 [7] NCCL INFO Using network Libfabric
+ip-26-0-168-30:235649:236208 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-30:235655:236209 [7] NCCL INFO comm 0xb467400 rank 1 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0xda3cb15f18bb32c8 - Init START
+ip-26-0-165-131:1630212:1630819 [7] NCCL INFO comm 0xa6db000 rank 0 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0xda3cb15f18bb32c8 - Init START
+ip-26-0-165-131:1630212:1630819 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/409
+ip-26-0-168-30:235655:236209 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-59:1557273:1557828 [7] NCCL INFO comm 0xc011560 rank 1 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0x1bdd1fa5bfd14ebb - Init START
+ip-26-0-161-78:114519:115084 [7] NCCL INFO comm 0xa4dd970 rank 0 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0x1bdd1fa5bfd14ebb - Init START
+ip-26-0-161-78:114519:115084 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-165-59:1557273:1557828 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/408
+ip-26-0-168-30:235654:236206 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-161-103:96891:97463 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-161-142:62884:63452 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-168-52:2722047:2722604 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-168-30:235651:236205 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-161-78:114512:115074 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-160-242:1364905:1365497 [3] NCCL INFO Setting affinity for GPU 3 to 0fff,ffffffff
+ip-26-0-160-242:1364905:1365497 [3] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-160-242:1364905:1365497 [3] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-160-242:1364905:1365497 [3] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-161-142:62884:63452 [3] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-161-142:62884:63452 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-242:1364905:1365497 [3] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-160-242:1364905:1365497 [3] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-160-242:1364905:1365497 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-153:2636087:2636683 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-161-153:2636087:2636683 [3] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-161-153:2636087:2636683 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-103:96891:97463 [3] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-161-103:96891:97463 [3] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-161-103:96891:97463 [3] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-161-103:96891:97463 [3] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-161-103:96891:97463 [3] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-161-103:96891:97463 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-59:1557272:1557827 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-165-164:1227469:1228026 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-161-123:32707:33271 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-160-225:124161:124732 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-165-131:1630205:1630805 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-168-34:272151:272710 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-161-123:32711:33274 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-160-242:1364906:1365487 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,fff00000,00000000
+ip-26-0-168-52:2722051:2722608 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-165-202:153566:154124 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-161-78:114517:115076 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-165-164:1227468:1228023 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-161-138:96565:97175 [2] NCCL INFO Setting affinity for GPU 2 to 0fff,ffffffff
+ip-26-0-168-34:272153:272711 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-165-131:1630206:1630817 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-161-142:62885:63447 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-165-164:1227466:1228032 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-160-242:1364909:1365485 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,fff00000,00000000
+ip-26-0-161-142:62885:63447 [4] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-161-142:62885:63447 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-242:1364906:1365487 [4] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-160-242:1364906:1365487 [4] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-160-242:1364906:1365487 [4] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-165-164:1227466:1228032 [3] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-165-164:1227466:1228032 [3] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-160-242:1364906:1365487 [4] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-160-242:1364906:1365487 [4] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-160-242:1364906:1365487 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-164:1227466:1228032 [3] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-165-164:1227466:1228032 [3] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-165-164:1227466:1228032 [3] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-165-164:1227466:1228032 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-168-34:272153:272711 [3] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-168-34:272153:272711 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-131:1630208:1630811 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-161-138:96564:97172 [1] NCCL INFO Setting affinity for GPU 1 to 0fff,ffffffff
+ip-26-0-165-131:1630208:1630811 [3] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-165-131:1630208:1630811 [3] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-165-131:1630208:1630811 [3] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-165-131:1630208:1630811 [3] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-165-131:1630208:1630811 [3] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-165-131:1630208:1630811 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-168-30:235651:236205 [3] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-168-30:235651:236205 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-123:32705:33278 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-160-225:124161:124732 [1] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-160-225:124161:124732 [1] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-160-225:124161:124732 [1] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-160-225:124161:124732 [1] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-160-225:124161:124732 [1] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-161-138:96564:97172 [1] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-161-138:96564:97172 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-225:124161:124732 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-178:2815963:2816553 [0] NCCL INFO Setting affinity for GPU 0 to 0fff,ffffffff
+ip-26-0-161-123:32704:33268 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-161-123:32704:33268 [0] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-161-123:32704:33268 [0] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-161-123:32704:33268 [0] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-161-123:32704:33268 [0] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-161-123:32704:33268 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-161-123:32704:33268 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-178:2815963:2816553 [0] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-161-178:2815963:2816553 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-131:1630207:1630815 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-160-225:124162:124740 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-160-225:124162:124740 [2] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-160-225:124162:124740 [2] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-160-225:124162:124740 [2] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-160-225:124162:124740 [2] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-160-225:124162:124740 [2] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-160-225:124162:124740 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-138:96565:97175 [2] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-161-138:96565:97175 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-123:32706:33272 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-161-142:62886:63449 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-161-138:96568:97174 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,fff00000,00000000
+ip-26-0-165-202:153563:154131 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-161-123:32709:33280 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-168-30:235650:236207 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-165-202:153568:154127 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-160-225:124163:124743 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-161-142:62888:63446 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-165-131:1630209:1630807 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-168-30:235650:236207 [2] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-168-30:235650:236207 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-131:1630207:1630815 [2] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-165-131:1630207:1630815 [2] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-165-131:1630207:1630815 [2] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-165-131:1630207:1630815 [2] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-165-131:1630207:1630815 [2] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-165-131:1630207:1630815 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-123:32710:33282 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-165-202:153568:154127 [6] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-165-202:153568:154127 [6] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-165-202:153568:154127 [6] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-165-202:153568:154127 [6] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-168-52:2722051:2722608 [6] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-168-52:2722051:2722608 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-202:153568:154127 [6] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-165-202:153568:154127 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-142:62888:63446 [7] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-161-142:62888:63446 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-242:1364909:1365485 [7] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-160-242:1364909:1365485 [7] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-160-242:1364909:1365485 [7] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-160-242:1364909:1365485 [7] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-160-242:1364909:1365485 [7] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-160-242:1364909:1365485 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-242:1364902:1365483 [0] NCCL INFO Setting affinity for GPU 0 to 0fff,ffffffff
+ip-26-0-165-202:153565:154121 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-168-52:2722049:2722606 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-161-138:96563:97171 [0] NCCL INFO Setting affinity for GPU 0 to 0fff,ffffffff
+ip-26-0-161-78:114514:115073 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-168-52:2722049:2722606 [4] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-168-52:2722049:2722606 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-202:153566:154124 [4] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-165-202:153566:154124 [4] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-165-202:153566:154124 [4] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-165-202:153566:154124 [4] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-165-202:153566:154124 [4] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-165-202:153566:154124 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-168-30:235648:236202 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-160-225:124160:124730 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-161-123:32708:33276 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-160-225:124166:124742 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-168-30:235648:236202 [0] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-168-30:235648:236202 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-131:1630205:1630805 [0] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-165-131:1630205:1630805 [0] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-165-131:1630205:1630805 [0] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-165-131:1630205:1630805 [0] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-165-131:1630205:1630805 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-165-131:1630205:1630805 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-225:124160:124730 [0] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-160-225:124160:124730 [0] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-160-225:124160:124730 [0] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-160-225:124160:124730 [0] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-161-138:96563:97171 [0] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-161-138:96563:97171 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-225:124160:124730 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-160-225:124160:124730 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-142:62883:63448 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-168-30:235652:236203 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-165-59:1557273:1557828 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-165-131:1630210:1630809 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-160-225:124164:124734 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-161-138:96566:97176 [3] NCCL INFO Setting affinity for GPU 3 to 0fff,ffffffff
+ip-26-0-168-30:235652:236203 [4] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-168-30:235652:236203 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-131:1630209:1630807 [4] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-165-131:1630209:1630807 [4] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-165-131:1630209:1630807 [4] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-165-131:1630209:1630807 [4] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-165-131:1630209:1630807 [4] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-165-131:1630209:1630807 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-168-30:235649:236208 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-161-103:96895:97465 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-160-225:124167:124744 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-161-138:96570:97178 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,fff00000,00000000
+ip-26-0-161-138:96566:97176 [3] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-161-138:96566:97176 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-225:124163:124743 [3] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-160-225:124163:124743 [3] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-160-225:124163:124743 [3] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-160-225:124163:124743 [3] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-160-225:124163:124743 [3] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-160-225:124163:124743 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-138:96569:97177 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,fff00000,00000000
+ip-26-0-168-30:235649:236208 [1] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-168-30:235649:236208 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-131:1630206:1630817 [1] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-165-131:1630206:1630817 [1] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-165-131:1630206:1630817 [1] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-165-131:1630206:1630817 [1] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-165-131:1630206:1630817 [1] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-165-131:1630206:1630817 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-131:1630211:1630813 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-161-138:96570:97178 [7] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-161-138:96570:97178 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-225:124167:124744 [7] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-160-225:124167:124744 [7] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-160-225:124167:124744 [7] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-160-225:124167:124744 [7] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-160-225:124167:124744 [7] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-160-225:124167:124744 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-78:114519:115084 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-168-30:235653:236204 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-161-138:96569:97177 [6] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-161-138:96569:97177 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-225:124166:124742 [6] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-160-225:124166:124742 [6] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-160-225:124166:124742 [6] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-160-225:124166:124742 [6] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-160-225:124166:124742 [6] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-160-225:124166:124742 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-78:114518:115082 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-165-131:1630211:1630813 [6] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-165-131:1630211:1630813 [6] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-168-30:235654:236206 [6] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-168-30:235654:236206 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-131:1630211:1630813 [6] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-165-131:1630211:1630813 [6] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-165-131:1630211:1630813 [6] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-165-131:1630211:1630813 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-78:114519:115084 [7] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-161-78:114519:115084 [7] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-161-78:114519:115084 [7] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-161-78:114519:115084 [7] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-165-59:1557273:1557828 [7] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-165-59:1557273:1557828 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-168-30:235653:236204 [5] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-168-30:235653:236204 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-78:114519:115084 [7] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-161-78:114519:115084 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-131:1630210:1630809 [5] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-165-131:1630210:1630809 [5] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-165-131:1630210:1630809 [5] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-165-131:1630210:1630809 [5] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-160-225:124165:124736 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-165-131:1630210:1630809 [5] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-165-131:1630210:1630809 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-131:1630212:1630819 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-161-78:114518:115082 [6] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-161-78:114518:115082 [6] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-161-78:114518:115082 [6] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-161-78:114518:115082 [6] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-161-78:114518:115082 [6] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-165-59:1557272:1557827 [6] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-165-59:1557272:1557827 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-78:114518:115082 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-225:124165:124736 [5] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-160-225:124165:124736 [5] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-160-225:124165:124736 [5] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-160-225:124165:124736 [5] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-160-225:124165:124736 [5] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-160-225:124165:124736 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-138:96568:97174 [5] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-161-138:96568:97174 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-168-34:272155:272707 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-161-103:96894:97456 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-168-30:235655:236209 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-161-138:96567:97173 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,fff00000,00000000
+ip-26-0-168-34:272155:272707 [5] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-168-34:272155:272707 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-164:1227468:1228023 [5] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-165-164:1227468:1228023 [5] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-165-164:1227468:1228023 [5] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-165-164:1227468:1228023 [5] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-165-164:1227468:1228023 [5] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-165-164:1227468:1228023 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-168-30:235655:236209 [7] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-168-30:235655:236209 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-131:1630212:1630819 [7] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-165-131:1630212:1630819 [7] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-165-131:1630212:1630819 [7] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-165-131:1630212:1630819 [7] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-165-131:1630212:1630819 [7] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-165-131:1630212:1630819 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-138:96567:97173 [4] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-161-138:96567:97173 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-225:124164:124734 [4] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-160-225:124164:124734 [4] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-160-225:124164:124734 [4] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-160-225:124164:124734 [4] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-160-225:124164:124734 [4] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-160-225:124164:124734 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-78:114516:115080 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-168-34:272156:272708 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-168-34:272156:272708 [6] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-168-34:272156:272708 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-142:62881:63445 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-165-164:1227469:1228026 [6] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-165-164:1227469:1228026 [6] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-165-164:1227469:1228026 [6] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-165-164:1227469:1228026 [6] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-165-164:1227469:1228026 [6] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-165-164:1227469:1228026 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-142:62882:63450 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-161-78:114513:115070 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-161-142:62881:63445 [0] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-161-142:62881:63445 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-242:1364902:1365483 [0] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-160-242:1364902:1365483 [0] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-160-242:1364902:1365483 [0] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-160-242:1364902:1365483 [0] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-160-242:1364902:1365483 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-160-242:1364902:1365483 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-142:62887:63451 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-161-78:114515:115081 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-165-164:1227467:1228021 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-161-103:96893:97462 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-161-178:2815969:2816560 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,fff00000,00000000
+ip-26-0-165-202:153562:154129 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-168-52:2722045:2722609 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-161-178:2815969:2816560 [6] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-161-178:2815969:2816560 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-103:96889:97453 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-161-123:32710:33282 [6] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-161-123:32710:33282 [6] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-161-123:32710:33282 [6] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-161-123:32710:33282 [6] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-161-123:32710:33282 [6] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-161-123:32710:33282 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-168-52:2722045:2722609 [0] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-168-52:2722045:2722609 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-202:153562:154129 [0] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-165-202:153562:154129 [0] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-165-202:153562:154129 [0] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-165-202:153562:154129 [0] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-165-202:153562:154129 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-165-202:153562:154129 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-242:1364907:1365491 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,fff00000,00000000
+ip-26-0-168-34:272157:272712 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-160-242:1364907:1365491 [5] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-160-242:1364907:1365491 [5] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-160-242:1364907:1365491 [5] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-161-142:62886:63449 [5] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-161-142:62886:63449 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-242:1364907:1365491 [5] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-160-242:1364907:1365491 [5] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-160-242:1364907:1365491 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-168-34:272150:272713 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-165-202:153567:154126 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-160-242:1364905:1365497 [3] NCCL INFO Channel 00/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364905:1365497 [3] NCCL INFO Channel 01/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364905:1365497 [3] NCCL INFO Channel 02/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364905:1365497 [3] NCCL INFO Channel 03/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62884:63452 [3] NCCL INFO Channel 00/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364905:1365497 [3] NCCL INFO Channel 00/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62884:63452 [3] NCCL INFO Channel 01/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364905:1365497 [3] NCCL INFO Channel 01/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62884:63452 [3] NCCL INFO Channel 02/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364905:1365497 [3] NCCL INFO Channel 02/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153569:154118 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-161-142:62884:63452 [3] NCCL INFO Channel 03/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364905:1365497 [3] NCCL INFO Channel 03/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62884:63452 [3] NCCL INFO Channel 00/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96888:97459 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-161-142:62884:63452 [3] NCCL INFO Channel 01/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62884:63452 [3] NCCL INFO Channel 02/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62884:63452 [3] NCCL INFO Channel 03/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153564:154120 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-165-202:153564:154120 [2] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-165-202:153564:154120 [2] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-165-202:153564:154120 [2] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-165-202:153564:154120 [2] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-168-52:2722047:2722604 [2] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-168-52:2722047:2722604 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-202:153564:154120 [2] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-165-202:153564:154120 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-103:96891:97463 [3] NCCL INFO Channel 00/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96891:97463 [3] NCCL INFO Channel 01/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96891:97463 [3] NCCL INFO Channel 02/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96891:97463 [3] NCCL INFO Channel 03/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96891:97463 [3] NCCL INFO Channel 00/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96891:97463 [3] NCCL INFO Channel 01/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-103:96891:97463 [3] NCCL INFO Channel 02/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272154:272706 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-161-103:96891:97463 [3] NCCL INFO Channel 03/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815965:2816555 [2] NCCL INFO Setting affinity for GPU 2 to 0fff,ffffffff
+ip-26-0-165-164:1227463:1228035 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-160-242:1364904:1365489 [2] NCCL INFO Setting affinity for GPU 2 to 0fff,ffffffff
+ip-26-0-168-34:272152:272709 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-168-34:272154:272706 [4] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-168-34:272154:272706 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-164:1227467:1228021 [4] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-165-164:1227467:1228021 [4] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-165-164:1227467:1228021 [4] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-165-164:1227467:1228021 [4] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-165-164:1227467:1228021 [4] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-165-164:1227467:1228021 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-178:2815965:2816555 [2] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-161-178:2815965:2816555 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-164:1227463:1228035 [0] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-165-164:1227463:1228035 [0] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-165-164:1227463:1228035 [0] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-165-164:1227463:1228035 [0] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-165-164:1227463:1228035 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-165-164:1227463:1228035 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-168-34:272150:272713 [0] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-168-34:272150:272713 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-123:32706:33272 [2] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-161-123:32706:33272 [2] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-161-123:32706:33272 [2] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-161-123:32706:33272 [2] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-161-123:32706:33272 [2] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-161-123:32706:33272 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-242:1364904:1365489 [2] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-160-242:1364904:1365489 [2] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-160-242:1364904:1365489 [2] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-160-242:1364904:1365489 [2] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-161-142:62883:63448 [2] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-161-142:62883:63448 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-242:1364904:1365489 [2] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-160-242:1364904:1365489 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-59:1557271:1557824 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-165-164:1227464:1228029 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-161-153:2636087:2636683 [3] NCCL INFO Channel 00/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636087:2636683 [3] NCCL INFO Channel 01/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636087:2636683 [3] NCCL INFO Channel 02/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636087:2636683 [3] NCCL INFO Channel 03/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636087:2636683 [3] NCCL INFO Channel 00/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636087:2636683 [3] NCCL INFO Channel 01/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636087:2636683 [3] NCCL INFO Channel 02/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636087:2636683 [3] NCCL INFO Channel 03/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557271:1557824 [5] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-165-59:1557271:1557824 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-78:114517:115076 [5] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-161-78:114517:115076 [5] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-161-78:114517:115076 [5] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-161-78:114517:115076 [5] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-161-78:114517:115076 [5] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-161-78:114517:115076 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-164:1227464:1228029 [1] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-165-164:1227464:1228029 [1] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-165-164:1227464:1228029 [1] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-165-164:1227464:1228029 [1] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-165-164:1227464:1228029 [1] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-165-164:1227464:1228029 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-168-34:272151:272710 [1] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-168-34:272151:272710 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-59:1557266:1557823 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-165-164:1227470:1228033 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-161-103:96890:97457 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-161-178:2815967:2816557 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,fff00000,00000000
+ip-26-0-165-59:1557266:1557823 [0] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-165-59:1557266:1557823 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-164:1227465:1228028 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-161-78:114512:115074 [0] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-161-78:114512:115074 [0] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-161-78:114512:115074 [0] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-161-78:114512:115074 [0] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-161-78:114512:115074 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-161-78:114512:115074 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-103:96892:97467 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-161-178:2815967:2816557 [4] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-161-178:2815967:2816557 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-123:32708:33276 [4] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-161-123:32708:33276 [4] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-161-123:32708:33276 [4] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-161-123:32708:33276 [4] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-161-123:32708:33276 [4] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-161-123:32708:33276 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-164:1227470:1228033 [7] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-165-164:1227470:1228033 [7] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-165-164:1227470:1228033 [7] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-165-164:1227470:1228033 [7] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-165-164:1227470:1228033 [7] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-168-34:272157:272712 [7] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-168-34:272157:272712 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-164:1227470:1228033 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-164:1227465:1228028 [2] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-165-164:1227465:1228028 [2] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-165-164:1227465:1228028 [2] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-165-164:1227465:1228028 [2] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-165-164:1227465:1228028 [2] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-165-164:1227465:1228028 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-168-34:272152:272709 [2] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-168-34:272152:272709 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-242:1364908:1365495 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,fff00000,00000000
+ip-26-0-160-242:1364908:1365495 [6] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-160-242:1364908:1365495 [6] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-160-242:1364908:1365495 [6] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-160-242:1364908:1365495 [6] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-161-142:62887:63451 [6] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-161-142:62887:63451 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-242:1364908:1365495 [6] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-160-242:1364908:1365495 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-242:1364903:1365493 [1] NCCL INFO Setting affinity for GPU 1 to 0fff,ffffffff
+ip-26-0-165-164:1227466:1228032 [3] NCCL INFO Channel 00/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227466:1228032 [3] NCCL INFO Channel 01/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227466:1228032 [3] NCCL INFO Channel 02/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227466:1228032 [3] NCCL INFO Channel 03/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227466:1228032 [3] NCCL INFO Channel 00/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227466:1228032 [3] NCCL INFO Channel 01/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227466:1228032 [3] NCCL INFO Channel 02/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-164:1227466:1228032 [3] NCCL INFO Channel 03/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-142:62885:63447 [4] NCCL INFO Channel 00/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62885:63447 [4] NCCL INFO Channel 01/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62885:63447 [4] NCCL INFO Channel 02/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364903:1365493 [1] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-160-242:1364903:1365493 [1] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-160-242:1364903:1365493 [1] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-160-242:1364903:1365493 [1] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-161-142:62882:63450 [1] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-161-142:62885:63447 [4] NCCL INFO Channel 03/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62882:63450 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-242:1364903:1365493 [1] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-160-242:1364903:1365493 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-142:62885:63447 [4] NCCL INFO Channel 00/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62885:63447 [4] NCCL INFO Channel 01/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62885:63447 [4] NCCL INFO Channel 02/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-142:62885:63447 [4] NCCL INFO Channel 03/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364906:1365487 [4] NCCL INFO Channel 00/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364906:1365487 [4] NCCL INFO Channel 01/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364906:1365487 [4] NCCL INFO Channel 02/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364906:1365487 [4] NCCL INFO Channel 03/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364906:1365487 [4] NCCL INFO Channel 00/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364906:1365487 [4] NCCL INFO Channel 01/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364906:1365487 [4] NCCL INFO Channel 02/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272153:272711 [3] NCCL INFO Channel 00/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-242:1364906:1365487 [4] NCCL INFO Channel 03/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272153:272711 [3] NCCL INFO Channel 01/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272153:272711 [3] NCCL INFO Channel 02/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272153:272711 [3] NCCL INFO Channel 03/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272153:272711 [3] NCCL INFO Channel 00/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272153:272711 [3] NCCL INFO Channel 01/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272153:272711 [3] NCCL INFO Channel 02/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-34:272153:272711 [3] NCCL INFO Channel 03/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636089:2636682 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-161-153:2636089:2636682 [5] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-161-153:2636089:2636682 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-103:96893:97462 [5] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-161-103:96893:97462 [5] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-161-103:96893:97462 [5] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-161-103:96893:97462 [5] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-161-103:96893:97462 [5] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-161-103:96893:97462 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-178:2815964:2816558 [1] NCCL INFO Setting affinity for GPU 1 to 0fff,ffffffff
+ip-26-0-161-178:2815964:2816558 [1] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-161-178:2815964:2816558 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-123:32705:33278 [1] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-161-123:32705:33278 [1] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-161-123:32705:33278 [1] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-161-123:32705:33278 [1] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-161-123:32705:33278 [1] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-161-123:32705:33278 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-131:1630208:1630811 [3] NCCL INFO Channel 00/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630208:1630811 [3] NCCL INFO Channel 01/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630208:1630811 [3] NCCL INFO Channel 02/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630208:1630811 [3] NCCL INFO Channel 03/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235651:236205 [3] NCCL INFO Channel 00/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630208:1630811 [3] NCCL INFO Channel 00/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815966:2816554 [3] NCCL INFO Setting affinity for GPU 3 to 0fff,ffffffff
+ip-26-0-165-131:1630208:1630811 [3] NCCL INFO Channel 01/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235651:236205 [3] NCCL INFO Channel 01/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96564:97172 [1] NCCL INFO Channel 00/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630208:1630811 [3] NCCL INFO Channel 02/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235651:236205 [3] NCCL INFO Channel 02/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630208:1630811 [3] NCCL INFO Channel 03/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96564:97172 [1] NCCL INFO Channel 01/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235651:236205 [3] NCCL INFO Channel 03/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96564:97172 [1] NCCL INFO Channel 02/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235651:236205 [3] NCCL INFO Channel 00/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96564:97172 [1] NCCL INFO Channel 03/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235651:236205 [3] NCCL INFO Channel 01/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96564:97172 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235651:236205 [3] NCCL INFO Channel 02/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:124161:124732 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96564:97172 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235651:236205 [3] NCCL INFO Channel 03/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:124161:124732 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96564:97172 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124161:124732 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96564:97172 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124161:124732 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124161:124732 [1] NCCL INFO Channel 00/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815970:2816556 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,fff00000,00000000
+ip-26-0-160-225:124161:124732 [1] NCCL INFO Channel 01/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124161:124732 [1] NCCL INFO Channel 02/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124161:124732 [1] NCCL INFO Channel 03/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815966:2816554 [3] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-161-178:2815966:2816554 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-123:32707:33271 [3] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-161-123:32707:33271 [3] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-161-123:32707:33271 [3] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-161-123:32707:33271 [3] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-161-123:32707:33271 [3] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-161-123:32707:33271 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-178:2815970:2816556 [7] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-161-178:2815970:2816556 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-123:32711:33274 [7] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-161-123:32711:33274 [7] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-161-123:32711:33274 [7] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-161-123:32711:33274 [7] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-161-123:32711:33274 [7] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-161-123:32711:33274 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-178:2815968:2816559 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,fff00000,00000000
+ip-26-0-161-153:2636085:2636678 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-161-178:2815968:2816559 [5] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-161-178:2815968:2816559 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-123:32709:33280 [5] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-161-123:32709:33280 [5] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-161-123:32709:33280 [5] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-161-123:32709:33280 [5] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-161-123:32709:33280 [5] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-161-123:32709:33280 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-103:96889:97453 [1] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-161-103:96889:97453 [1] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-161-103:96889:97453 [1] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-161-103:96889:97453 [1] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-161-103:96889:97453 [1] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-161-153:2636085:2636678 [1] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-161-153:2636085:2636678 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-103:96889:97453 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-123:32704:33268 [0] NCCL INFO Channel 00/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32704:33268 [0] NCCL INFO Channel 01/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96565:97175 [2] NCCL INFO Channel 00/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32704:33268 [0] NCCL INFO Channel 02/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96565:97175 [2] NCCL INFO Channel 01/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32704:33268 [0] NCCL INFO Channel 03/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96565:97175 [2] NCCL INFO Channel 02/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32704:33268 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96565:97175 [2] NCCL INFO Channel 03/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32704:33268 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96565:97175 [2] NCCL INFO Channel 00/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32704:33268 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96565:97175 [2] NCCL INFO Channel 01/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32704:33268 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96565:97175 [2] NCCL INFO Channel 02/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-138:96565:97175 [2] NCCL INFO Channel 03/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557268:1557822 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-161-178:2815963:2816553 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815963:2816553 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815963:2816553 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815963:2816553 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815963:2816553 [0] NCCL INFO Channel 00/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815963:2816553 [0] NCCL INFO Channel 01/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815963:2816553 [0] NCCL INFO Channel 02/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815963:2816553 [0] NCCL INFO Channel 03/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557268:1557822 [2] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-165-59:1557268:1557822 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-59:1557270:1557825 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-161-78:114514:115073 [2] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-161-78:114514:115073 [2] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-161-78:114514:115073 [2] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-161-78:114514:115073 [2] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-161-78:114514:115073 [2] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-161-78:114514:115073 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-225:124162:124740 [2] NCCL INFO Channel 00/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:124162:124740 [2] NCCL INFO Channel 01/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557270:1557825 [4] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-165-59:1557270:1557825 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-78:114516:115080 [4] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-161-78:114516:115080 [4] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-161-78:114516:115080 [4] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-161-78:114516:115080 [4] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-161-78:114516:115080 [4] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-161-78:114516:115080 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-225:124162:124740 [2] NCCL INFO Channel 02/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:124162:124740 [2] NCCL INFO Channel 03/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:124162:124740 [2] NCCL INFO Channel 00/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:124162:124740 [2] NCCL INFO Channel 01/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:124162:124740 [2] NCCL INFO Channel 02/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:124162:124740 [2] NCCL INFO Channel 03/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-52:2722046:2722610 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-165-59:1557269:1557826 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-168-52:2722046:2722610 [1] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-168-52:2722046:2722610 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-202:153563:154131 [1] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-165-202:153563:154131 [1] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-165-202:153563:154131 [1] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-165-202:153563:154131 [1] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-165-202:153563:154131 [1] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-165-202:153563:154131 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-59:1557269:1557826 [3] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-165-59:1557269:1557826 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-78:114515:115081 [3] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-161-78:114515:115081 [3] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-161-78:114515:115081 [3] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-161-78:114515:115081 [3] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-161-78:114515:115081 [3] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-161-78:114515:115081 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-142:62888:63446 [7] NCCL INFO Channel 00/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62888:63446 [7] NCCL INFO Channel 01/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62888:63446 [7] NCCL INFO Channel 02/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557267:1557821 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-161-142:62888:63446 [7] NCCL INFO Channel 03/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62888:63446 [7] NCCL INFO Channel 00/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62888:63446 [7] NCCL INFO Channel 01/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62888:63446 [7] NCCL INFO Channel 02/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-142:62888:63446 [7] NCCL INFO Channel 03/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153568:154127 [6] NCCL INFO Channel 00/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153568:154127 [6] NCCL INFO Channel 01/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153568:154127 [6] NCCL INFO Channel 02/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153568:154127 [6] NCCL INFO Channel 03/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153568:154127 [6] NCCL INFO Channel 00/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-202:153568:154127 [6] NCCL INFO Channel 01/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630207:1630815 [2] NCCL INFO Channel 00/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153568:154127 [6] NCCL INFO Channel 02/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630207:1630815 [2] NCCL INFO Channel 01/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153568:154127 [6] NCCL INFO Channel 03/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630207:1630815 [2] NCCL INFO Channel 02/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630207:1630815 [2] NCCL INFO Channel 03/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630207:1630815 [2] NCCL INFO Channel 00/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630207:1630815 [2] NCCL INFO Channel 01/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630207:1630815 [2] NCCL INFO Channel 02/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-131:1630207:1630815 [2] NCCL INFO Channel 03/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364909:1365485 [7] NCCL INFO Channel 00/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557267:1557821 [1] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-165-59:1557267:1557821 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-242:1364909:1365485 [7] NCCL INFO Channel 01/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114513:115070 [1] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-161-78:114513:115070 [1] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-161-78:114513:115070 [1] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-161-78:114513:115070 [1] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-161-78:114513:115070 [1] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-161-78:114513:115070 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-242:1364909:1365485 [7] NCCL INFO Channel 02/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364909:1365485 [7] NCCL INFO Channel 03/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364909:1365485 [7] NCCL INFO Channel 00/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364909:1365485 [7] NCCL INFO Channel 01/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364909:1365485 [7] NCCL INFO Channel 02/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-242:1364909:1365485 [7] NCCL INFO Channel 03/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722051:2722608 [6] NCCL INFO Channel 00/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-52:2722051:2722608 [6] NCCL INFO Channel 01/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-52:2722051:2722608 [6] NCCL INFO Channel 02/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-52:2722051:2722608 [6] NCCL INFO Channel 03/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-52:2722051:2722608 [6] NCCL INFO Channel 00/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-52:2722051:2722608 [6] NCCL INFO Channel 01/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235650:236207 [2] NCCL INFO Channel 00/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-52:2722051:2722608 [6] NCCL INFO Channel 02/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235650:236207 [2] NCCL INFO Channel 01/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-52:2722051:2722608 [6] NCCL INFO Channel 03/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235650:236207 [2] NCCL INFO Channel 02/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235650:236207 [2] NCCL INFO Channel 03/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235650:236207 [2] NCCL INFO Channel 00/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235650:236207 [2] NCCL INFO Channel 01/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235650:236207 [2] NCCL INFO Channel 02/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-30:235650:236207 [2] NCCL INFO Channel 03/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153566:154124 [4] NCCL INFO Channel 00/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153566:154124 [4] NCCL INFO Channel 01/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722049:2722606 [4] NCCL INFO Channel 00/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153566:154124 [4] NCCL INFO Channel 02/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722049:2722606 [4] NCCL INFO Channel 01/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153566:154124 [4] NCCL INFO Channel 03/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722049:2722606 [4] NCCL INFO Channel 02/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153566:154124 [4] NCCL INFO Channel 00/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722049:2722606 [4] NCCL INFO Channel 03/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153566:154124 [4] NCCL INFO Channel 01/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722049:2722606 [4] NCCL INFO Channel 00/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96563:97171 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153566:154124 [4] NCCL INFO Channel 02/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722049:2722606 [4] NCCL INFO Channel 01/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-52:2722049:2722606 [4] NCCL INFO Channel 02/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153566:154124 [4] NCCL INFO Channel 03/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96563:97171 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-52:2722049:2722606 [4] NCCL INFO Channel 03/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96563:97171 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96563:97171 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96563:97171 [0] NCCL INFO Channel 00/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96563:97171 [0] NCCL INFO Channel 01/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96563:97171 [0] NCCL INFO Channel 02/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-138:96563:97171 [0] NCCL INFO Channel 03/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:124160:124730 [0] NCCL INFO Channel 00/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:124160:124730 [0] NCCL INFO Channel 01/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:124160:124730 [0] NCCL INFO Channel 02/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:124160:124730 [0] NCCL INFO Channel 03/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:124160:124730 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:124160:124730 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:124160:124730 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:124160:124730 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235648:236202 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235648:236202 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235648:236202 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235648:236202 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235648:236202 [0] NCCL INFO Channel 00/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630205:1630805 [0] NCCL INFO Channel 00/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235648:236202 [0] NCCL INFO Channel 01/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630205:1630805 [0] NCCL INFO Channel 01/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-30:235648:236202 [0] NCCL INFO Channel 02/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630205:1630805 [0] NCCL INFO Channel 02/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:124163:124743 [3] NCCL INFO Channel 00/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235648:236202 [0] NCCL INFO Channel 03/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630205:1630805 [0] NCCL INFO Channel 03/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-131:1630205:1630805 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:124163:124743 [3] NCCL INFO Channel 01/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630205:1630805 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:124163:124743 [3] NCCL INFO Channel 02/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630205:1630805 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:124163:124743 [3] NCCL INFO Channel 03/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630205:1630805 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:124163:124743 [3] NCCL INFO Channel 00/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:124163:124743 [3] NCCL INFO Channel 01/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:124163:124743 [3] NCCL INFO Channel 02/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:124163:124743 [3] NCCL INFO Channel 03/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630209:1630807 [4] NCCL INFO Channel 00/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630209:1630807 [4] NCCL INFO Channel 01/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630209:1630807 [4] NCCL INFO Channel 02/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630209:1630807 [4] NCCL INFO Channel 03/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630209:1630807 [4] NCCL INFO Channel 00/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630209:1630807 [4] NCCL INFO Channel 01/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630209:1630807 [4] NCCL INFO Channel 02/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630209:1630807 [4] NCCL INFO Channel 03/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96570:97178 [7] NCCL INFO Channel 00/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96570:97178 [7] NCCL INFO Channel 01/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96570:97178 [7] NCCL INFO Channel 02/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235649:236208 [1] NCCL INFO Channel 00/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96570:97178 [7] NCCL INFO Channel 03/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235649:236208 [1] NCCL INFO Channel 01/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96570:97178 [7] NCCL INFO Channel 00/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235649:236208 [1] NCCL INFO Channel 02/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96570:97178 [7] NCCL INFO Channel 01/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96570:97178 [7] NCCL INFO Channel 02/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235649:236208 [1] NCCL INFO Channel 03/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124167:124744 [7] NCCL INFO Channel 00/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96570:97178 [7] NCCL INFO Channel 03/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235649:236208 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124167:124744 [7] NCCL INFO Channel 01/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96566:97176 [3] NCCL INFO Channel 00/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235649:236208 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-30:235649:236208 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124166:124742 [6] NCCL INFO Channel 00/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:124167:124744 [7] NCCL INFO Channel 02/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96566:97176 [3] NCCL INFO Channel 01/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-30:235649:236208 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96569:97177 [6] NCCL INFO Channel 00/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630206:1630817 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124167:124744 [7] NCCL INFO Channel 03/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124166:124742 [6] NCCL INFO Channel 01/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96566:97176 [3] NCCL INFO Channel 02/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:124167:124744 [7] NCCL INFO Channel 00/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630206:1630817 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124166:124742 [6] NCCL INFO Channel 02/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96566:97176 [3] NCCL INFO Channel 03/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:124167:124744 [7] NCCL INFO Channel 01/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96569:97177 [6] NCCL INFO Channel 01/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114519:115084 [7] NCCL INFO Channel 00/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630211:1630813 [6] NCCL INFO Channel 00/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630206:1630817 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124166:124742 [6] NCCL INFO Channel 03/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235652:236203 [4] NCCL INFO Channel 00/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630206:1630817 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:124167:124744 [7] NCCL INFO Channel 02/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124166:124742 [6] NCCL INFO Channel 00/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96566:97176 [3] NCCL INFO Channel 00/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114519:115084 [7] NCCL INFO Channel 01/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124167:124744 [7] NCCL INFO Channel 03/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-138:96569:97177 [6] NCCL INFO Channel 02/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630211:1630813 [6] NCCL INFO Channel 01/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:124166:124742 [6] NCCL INFO Channel 01/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96566:97176 [3] NCCL INFO Channel 01/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114519:115084 [7] NCCL INFO Channel 02/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235652:236203 [4] NCCL INFO Channel 01/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630206:1630817 [1] NCCL INFO Channel 00/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96569:97177 [6] NCCL INFO Channel 03/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:124166:124742 [6] NCCL INFO Channel 02/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96566:97176 [3] NCCL INFO Channel 02/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-138:96569:97177 [6] NCCL INFO Channel 00/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114519:115084 [7] NCCL INFO Channel 03/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235652:236203 [4] NCCL INFO Channel 02/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630211:1630813 [6] NCCL INFO Channel 02/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:124166:124742 [6] NCCL INFO Channel 03/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630206:1630817 [1] NCCL INFO Channel 01/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96566:97176 [3] NCCL INFO Channel 03/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-131:1630211:1630813 [6] NCCL INFO Channel 03/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96569:97177 [6] NCCL INFO Channel 01/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114519:115084 [7] NCCL INFO Channel 00/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630206:1630817 [1] NCCL INFO Channel 02/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-138:96569:97177 [6] NCCL INFO Channel 02/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235652:236203 [4] NCCL INFO Channel 03/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114519:115084 [7] NCCL INFO Channel 01/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630211:1630813 [6] NCCL INFO Channel 00/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96569:97177 [6] NCCL INFO Channel 03/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630206:1630817 [1] NCCL INFO Channel 03/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-131:1630211:1630813 [6] NCCL INFO Channel 01/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114519:115084 [7] NCCL INFO Channel 02/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235652:236203 [4] NCCL INFO Channel 00/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630211:1630813 [6] NCCL INFO Channel 02/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114519:115084 [7] NCCL INFO Channel 03/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235652:236203 [4] NCCL INFO Channel 01/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630211:1630813 [6] NCCL INFO Channel 03/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235652:236203 [4] NCCL INFO Channel 02/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-30:235652:236203 [4] NCCL INFO Channel 03/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-138:96568:97174 [5] NCCL INFO Channel 00/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-52:2722050:2722607 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-161-138:96568:97174 [5] NCCL INFO Channel 01/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96568:97174 [5] NCCL INFO Channel 02/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96568:97174 [5] NCCL INFO Channel 03/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235654:236206 [6] NCCL INFO Channel 00/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96568:97174 [5] NCCL INFO Channel 00/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96568:97174 [5] NCCL INFO Channel 01/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235654:236206 [6] NCCL INFO Channel 01/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96568:97174 [5] NCCL INFO Channel 02/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235653:236204 [5] NCCL INFO Channel 00/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235654:236206 [6] NCCL INFO Channel 02/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-138:96568:97174 [5] NCCL INFO Channel 03/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630210:1630809 [5] NCCL INFO Channel 00/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235654:236206 [6] NCCL INFO Channel 03/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235653:236204 [5] NCCL INFO Channel 01/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630210:1630809 [5] NCCL INFO Channel 01/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235654:236206 [6] NCCL INFO Channel 00/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630210:1630809 [5] NCCL INFO Channel 02/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235653:236204 [5] NCCL INFO Channel 02/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235654:236206 [6] NCCL INFO Channel 01/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630210:1630809 [5] NCCL INFO Channel 03/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235654:236206 [6] NCCL INFO Channel 02/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557272:1557827 [6] NCCL INFO Channel 00/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235653:236204 [5] NCCL INFO Channel 03/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630210:1630809 [5] NCCL INFO Channel 00/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235654:236206 [6] NCCL INFO Channel 03/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-131:1630210:1630809 [5] NCCL INFO Channel 01/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-30:235653:236204 [5] NCCL INFO Channel 00/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557272:1557827 [6] NCCL INFO Channel 01/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235653:236204 [5] NCCL INFO Channel 01/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630210:1630809 [5] NCCL INFO Channel 02/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114518:115082 [6] NCCL INFO Channel 00/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557272:1557827 [6] NCCL INFO Channel 02/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-52:2722052:2722603 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-168-30:235653:236204 [5] NCCL INFO Channel 02/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-131:1630210:1630809 [5] NCCL INFO Channel 03/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557272:1557827 [6] NCCL INFO Channel 03/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114518:115082 [6] NCCL INFO Channel 01/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-30:235653:236204 [5] NCCL INFO Channel 03/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557272:1557827 [6] NCCL INFO Channel 00/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114518:115082 [6] NCCL INFO Channel 02/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557272:1557827 [6] NCCL INFO Channel 01/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114518:115082 [6] NCCL INFO Channel 03/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:124165:124736 [5] NCCL INFO Channel 00/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-52:2722050:2722607 [5] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-168-52:2722050:2722607 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-59:1557272:1557827 [6] NCCL INFO Channel 02/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114518:115082 [6] NCCL INFO Channel 00/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:124165:124736 [5] NCCL INFO Channel 01/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153567:154126 [5] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-165-202:153567:154126 [5] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-165-202:153567:154126 [5] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-165-202:153567:154126 [5] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-165-202:153567:154126 [5] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-165-202:153567:154126 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-59:1557272:1557827 [6] NCCL INFO Channel 03/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-78:114518:115082 [6] NCCL INFO Channel 01/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:124165:124736 [5] NCCL INFO Channel 02/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557273:1557828 [7] NCCL INFO Channel 00/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636088:2636685 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-161-78:114518:115082 [6] NCCL INFO Channel 02/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557273:1557828 [7] NCCL INFO Channel 01/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124165:124736 [5] NCCL INFO Channel 03/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114518:115082 [6] NCCL INFO Channel 03/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-59:1557273:1557828 [7] NCCL INFO Channel 02/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124165:124736 [5] NCCL INFO Channel 00/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227468:1228023 [5] NCCL INFO Channel 00/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557273:1557828 [7] NCCL INFO Channel 03/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124164:124734 [4] NCCL INFO Channel 00/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:124165:124736 [5] NCCL INFO Channel 01/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557273:1557828 [7] NCCL INFO Channel 00/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227468:1228023 [5] NCCL INFO Channel 01/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124165:124736 [5] NCCL INFO Channel 02/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124164:124734 [4] NCCL INFO Channel 01/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557273:1557828 [7] NCCL INFO Channel 01/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227468:1228023 [5] NCCL INFO Channel 02/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124165:124736 [5] NCCL INFO Channel 03/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557273:1557828 [7] NCCL INFO Channel 02/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227468:1228023 [5] NCCL INFO Channel 03/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124164:124734 [4] NCCL INFO Channel 02/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557273:1557828 [7] NCCL INFO Channel 03/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:124164:124734 [4] NCCL INFO Channel 03/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227468:1228023 [5] NCCL INFO Channel 00/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227468:1228023 [5] NCCL INFO Channel 01/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124164:124734 [4] NCCL INFO Channel 00/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227468:1228023 [5] NCCL INFO Channel 02/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-52:2722052:2722603 [7] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-168-52:2722052:2722603 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-225:124164:124734 [4] NCCL INFO Channel 01/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153569:154118 [7] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-165-202:153569:154118 [7] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-165-202:153569:154118 [7] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-165-202:153569:154118 [7] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-165-202:153569:154118 [7] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-165-202:153569:154118 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-164:1227468:1228023 [5] NCCL INFO Channel 03/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:124164:124734 [4] NCCL INFO Channel 02/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:124164:124734 [4] NCCL INFO Channel 03/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-131:1630212:1630819 [7] NCCL INFO Channel 00/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630212:1630819 [7] NCCL INFO Channel 01/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630212:1630819 [7] NCCL INFO Channel 02/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630212:1630819 [7] NCCL INFO Channel 03/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630212:1630819 [7] NCCL INFO Channel 00/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-131:1630212:1630819 [7] NCCL INFO Channel 01/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636088:2636685 [4] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-161-153:2636088:2636685 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-103:96892:97467 [4] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-161-103:96892:97467 [4] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-161-103:96892:97467 [4] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-161-103:96892:97467 [4] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-165-131:1630212:1630819 [7] NCCL INFO Channel 02/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96892:97467 [4] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-161-103:96892:97467 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-131:1630212:1630819 [7] NCCL INFO Channel 03/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272155:272707 [5] NCCL INFO Channel 00/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96567:97173 [4] NCCL INFO Channel 00/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272155:272707 [5] NCCL INFO Channel 01/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96567:97173 [4] NCCL INFO Channel 01/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272155:272707 [5] NCCL INFO Channel 02/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96567:97173 [4] NCCL INFO Channel 02/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272155:272707 [5] NCCL INFO Channel 03/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96567:97173 [4] NCCL INFO Channel 03/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272155:272707 [5] NCCL INFO Channel 00/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96567:97173 [4] NCCL INFO Channel 00/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272155:272707 [5] NCCL INFO Channel 01/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96567:97173 [4] NCCL INFO Channel 01/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272155:272707 [5] NCCL INFO Channel 02/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96567:97173 [4] NCCL INFO Channel 02/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272155:272707 [5] NCCL INFO Channel 03/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-138:96567:97173 [4] NCCL INFO Channel 03/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636084:2636681 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-168-30:235655:236209 [7] NCCL INFO Channel 00/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235655:236209 [7] NCCL INFO Channel 01/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722048:2722605 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-168-30:235655:236209 [7] NCCL INFO Channel 02/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235655:236209 [7] NCCL INFO Channel 03/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235655:236209 [7] NCCL INFO Channel 00/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235655:236209 [7] NCCL INFO Channel 01/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235655:236209 [7] NCCL INFO Channel 02/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-30:235655:236209 [7] NCCL INFO Channel 03/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636084:2636681 [0] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-161-153:2636084:2636681 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-103:96888:97459 [0] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-161-103:96888:97459 [0] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-161-103:96888:97459 [0] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-161-103:96888:97459 [0] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-161-103:96888:97459 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-161-103:96888:97459 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-168-52:2722048:2722605 [3] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-168-52:2722048:2722605 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-202:153565:154121 [3] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-165-202:153565:154121 [3] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-165-202:153565:154121 [3] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-165-202:153565:154121 [3] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-165-202:153565:154121 [3] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-165-202:153565:154121 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-164:1227469:1228026 [6] NCCL INFO Channel 00/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227469:1228026 [6] NCCL INFO Channel 01/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272156:272708 [6] NCCL INFO Channel 00/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227469:1228026 [6] NCCL INFO Channel 02/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62881:63445 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272156:272708 [6] NCCL INFO Channel 01/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227469:1228026 [6] NCCL INFO Channel 03/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62881:63445 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272156:272708 [6] NCCL INFO Channel 02/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227469:1228026 [6] NCCL INFO Channel 00/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62881:63445 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227469:1228026 [6] NCCL INFO Channel 01/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272156:272708 [6] NCCL INFO Channel 03/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-164:1227469:1228026 [6] NCCL INFO Channel 02/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-34:272156:272708 [6] NCCL INFO Channel 00/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62881:63445 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227469:1228026 [6] NCCL INFO Channel 03/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62881:63445 [0] NCCL INFO Channel 00/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272156:272708 [6] NCCL INFO Channel 01/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62881:63445 [0] NCCL INFO Channel 01/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272156:272708 [6] NCCL INFO Channel 02/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62881:63445 [0] NCCL INFO Channel 02/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272156:272708 [6] NCCL INFO Channel 03/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62881:63445 [0] NCCL INFO Channel 03/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153562:154129 [0] NCCL INFO Channel 00/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153562:154129 [0] NCCL INFO Channel 01/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153562:154129 [0] NCCL INFO Channel 02/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153562:154129 [0] NCCL INFO Channel 03/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153562:154129 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153562:154129 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153562:154129 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153562:154129 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815969:2816560 [6] NCCL INFO Channel 00/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815969:2816560 [6] NCCL INFO Channel 01/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815969:2816560 [6] NCCL INFO Channel 02/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815969:2816560 [6] NCCL INFO Channel 03/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815969:2816560 [6] NCCL INFO Channel 00/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815969:2816560 [6] NCCL INFO Channel 01/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815969:2816560 [6] NCCL INFO Channel 02/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-178:2815969:2816560 [6] NCCL INFO Channel 03/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62886:63449 [5] NCCL INFO Channel 00/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62886:63449 [5] NCCL INFO Channel 01/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62886:63449 [5] NCCL INFO Channel 02/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62886:63449 [5] NCCL INFO Channel 03/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62886:63449 [5] NCCL INFO Channel 00/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62886:63449 [5] NCCL INFO Channel 01/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62886:63449 [5] NCCL INFO Channel 02/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-142:62886:63449 [5] NCCL INFO Channel 03/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364902:1365483 [0] NCCL INFO Channel 00/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364902:1365483 [0] NCCL INFO Channel 01/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364902:1365483 [0] NCCL INFO Channel 02/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364902:1365483 [0] NCCL INFO Channel 03/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364902:1365483 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364902:1365483 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364902:1365483 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364902:1365483 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32710:33282 [6] NCCL INFO Channel 00/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32710:33282 [6] NCCL INFO Channel 01/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32710:33282 [6] NCCL INFO Channel 02/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32710:33282 [6] NCCL INFO Channel 03/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32710:33282 [6] NCCL INFO Channel 00/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32710:33282 [6] NCCL INFO Channel 01/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32710:33282 [6] NCCL INFO Channel 02/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-123:32710:33282 [6] NCCL INFO Channel 03/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-168-52:2722045:2722609 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-52:2722045:2722609 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-52:2722045:2722609 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-52:2722045:2722609 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-52:2722045:2722609 [0] NCCL INFO Channel 00/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-52:2722045:2722609 [0] NCCL INFO Channel 01/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-52:2722045:2722609 [0] NCCL INFO Channel 02/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-52:2722045:2722609 [0] NCCL INFO Channel 03/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364907:1365491 [5] NCCL INFO Channel 00/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364907:1365491 [5] NCCL INFO Channel 01/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364907:1365491 [5] NCCL INFO Channel 02/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364907:1365491 [5] NCCL INFO Channel 03/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364907:1365491 [5] NCCL INFO Channel 00/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364907:1365491 [5] NCCL INFO Channel 01/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364907:1365491 [5] NCCL INFO Channel 02/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-242:1364907:1365491 [5] NCCL INFO Channel 03/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153564:154120 [2] NCCL INFO Channel 00/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153564:154120 [2] NCCL INFO Channel 01/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153564:154120 [2] NCCL INFO Channel 02/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153564:154120 [2] NCCL INFO Channel 03/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153564:154120 [2] NCCL INFO Channel 00/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153564:154120 [2] NCCL INFO Channel 01/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153564:154120 [2] NCCL INFO Channel 02/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-202:153564:154120 [2] NCCL INFO Channel 03/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-52:2722047:2722604 [2] NCCL INFO Channel 00/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-52:2722047:2722604 [2] NCCL INFO Channel 01/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-52:2722047:2722604 [2] NCCL INFO Channel 02/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-52:2722047:2722604 [2] NCCL INFO Channel 03/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636086:2636680 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-168-52:2722047:2722604 [2] NCCL INFO Channel 00/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-52:2722047:2722604 [2] NCCL INFO Channel 01/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-52:2722047:2722604 [2] NCCL INFO Channel 02/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-52:2722047:2722604 [2] NCCL INFO Channel 03/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62883:63448 [2] NCCL INFO Channel 00/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62883:63448 [2] NCCL INFO Channel 01/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62883:63448 [2] NCCL INFO Channel 02/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62883:63448 [2] NCCL INFO Channel 03/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62883:63448 [2] NCCL INFO Channel 00/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62883:63448 [2] NCCL INFO Channel 01/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62883:63448 [2] NCCL INFO Channel 02/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-142:62883:63448 [2] NCCL INFO Channel 03/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636086:2636680 [2] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-161-153:2636086:2636680 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-103:96890:97457 [2] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-161-103:96890:97457 [2] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-161-103:96890:97457 [2] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-161-103:96890:97457 [2] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-161-103:96890:97457 [2] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-161-103:96890:97457 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-164:1227467:1228021 [4] NCCL INFO Channel 00/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227467:1228021 [4] NCCL INFO Channel 01/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227467:1228021 [4] NCCL INFO Channel 02/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227467:1228021 [4] NCCL INFO Channel 03/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227467:1228021 [4] NCCL INFO Channel 00/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227467:1228021 [4] NCCL INFO Channel 01/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227467:1228021 [4] NCCL INFO Channel 02/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364904:1365489 [2] NCCL INFO Channel 00/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227467:1228021 [4] NCCL INFO Channel 03/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-242:1364904:1365489 [2] NCCL INFO Channel 01/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364904:1365489 [2] NCCL INFO Channel 02/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364904:1365489 [2] NCCL INFO Channel 03/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815965:2816555 [2] NCCL INFO Channel 00/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364904:1365489 [2] NCCL INFO Channel 00/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364904:1365489 [2] NCCL INFO Channel 01/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815965:2816555 [2] NCCL INFO Channel 01/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-242:1364904:1365489 [2] NCCL INFO Channel 02/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815965:2816555 [2] NCCL INFO Channel 02/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227463:1228035 [0] NCCL INFO Channel 00/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-242:1364904:1365489 [2] NCCL INFO Channel 03/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815965:2816555 [2] NCCL INFO Channel 03/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227463:1228035 [0] NCCL INFO Channel 01/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815965:2816555 [2] NCCL INFO Channel 00/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227463:1228035 [0] NCCL INFO Channel 02/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815965:2816555 [2] NCCL INFO Channel 01/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227463:1228035 [0] NCCL INFO Channel 03/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815965:2816555 [2] NCCL INFO Channel 02/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227463:1228035 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-178:2815965:2816555 [2] NCCL INFO Channel 03/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-164:1227463:1228035 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227463:1228035 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227463:1228035 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227464:1228029 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114517:115076 [5] NCCL INFO Channel 00/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227464:1228029 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272150:272713 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114517:115076 [5] NCCL INFO Channel 01/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227464:1228029 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114517:115076 [5] NCCL INFO Channel 02/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-34:272150:272713 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114517:115076 [5] NCCL INFO Channel 03/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227464:1228029 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272150:272713 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114517:115076 [5] NCCL INFO Channel 00/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227464:1228029 [1] NCCL INFO Channel 00/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272150:272713 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114517:115076 [5] NCCL INFO Channel 01/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227464:1228029 [1] NCCL INFO Channel 01/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272150:272713 [0] NCCL INFO Channel 00/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114517:115076 [5] NCCL INFO Channel 02/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227464:1228029 [1] NCCL INFO Channel 02/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272150:272713 [0] NCCL INFO Channel 01/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114517:115076 [5] NCCL INFO Channel 03/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-164:1227464:1228029 [1] NCCL INFO Channel 03/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-34:272150:272713 [0] NCCL INFO Channel 02/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272150:272713 [0] NCCL INFO Channel 03/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272154:272706 [4] NCCL INFO Channel 00/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272154:272706 [4] NCCL INFO Channel 01/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272154:272706 [4] NCCL INFO Channel 02/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272154:272706 [4] NCCL INFO Channel 03/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272154:272706 [4] NCCL INFO Channel 00/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32706:33272 [2] NCCL INFO Channel 00/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272154:272706 [4] NCCL INFO Channel 01/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32706:33272 [2] NCCL INFO Channel 01/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272154:272706 [4] NCCL INFO Channel 02/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32706:33272 [2] NCCL INFO Channel 02/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272154:272706 [4] NCCL INFO Channel 03/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32706:33272 [2] NCCL INFO Channel 03/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557271:1557824 [5] NCCL INFO Channel 00/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32706:33272 [2] NCCL INFO Channel 00/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557271:1557824 [5] NCCL INFO Channel 01/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32706:33272 [2] NCCL INFO Channel 01/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557271:1557824 [5] NCCL INFO Channel 02/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32706:33272 [2] NCCL INFO Channel 02/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557271:1557824 [5] NCCL INFO Channel 03/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32706:33272 [2] NCCL INFO Channel 03/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557271:1557824 [5] NCCL INFO Channel 00/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557271:1557824 [5] NCCL INFO Channel 01/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557271:1557824 [5] NCCL INFO Channel 02/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-59:1557271:1557824 [5] NCCL INFO Channel 03/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636091:2636684 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-161-178:2815967:2816557 [4] NCCL INFO Channel 00/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815967:2816557 [4] NCCL INFO Channel 01/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272151:272710 [1] NCCL INFO Channel 00/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815967:2816557 [4] NCCL INFO Channel 02/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-178:2815967:2816557 [4] NCCL INFO Channel 03/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-164:1227465:1228028 [2] NCCL INFO Channel 00/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272151:272710 [1] NCCL INFO Channel 01/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815967:2816557 [4] NCCL INFO Channel 00/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272151:272710 [1] NCCL INFO Channel 02/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227465:1228028 [2] NCCL INFO Channel 01/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815967:2816557 [4] NCCL INFO Channel 01/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272151:272710 [1] NCCL INFO Channel 03/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227465:1228028 [2] NCCL INFO Channel 02/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815967:2816557 [4] NCCL INFO Channel 02/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272151:272710 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227465:1228028 [2] NCCL INFO Channel 03/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-178:2815967:2816557 [4] NCCL INFO Channel 03/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114512:115074 [0] NCCL INFO Channel 00/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272151:272710 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227465:1228028 [2] NCCL INFO Channel 00/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272151:272710 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227465:1228028 [2] NCCL INFO Channel 01/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114512:115074 [0] NCCL INFO Channel 01/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-168-34:272151:272710 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-164:1227465:1228028 [2] NCCL INFO Channel 02/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114512:115074 [0] NCCL INFO Channel 02/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227465:1228028 [2] NCCL INFO Channel 03/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114512:115074 [0] NCCL INFO Channel 03/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114512:115074 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114512:115074 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-78:114512:115074 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227470:1228033 [7] NCCL INFO Channel 00/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-78:114512:115074 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-164:1227470:1228033 [7] NCCL INFO Channel 01/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227470:1228033 [7] NCCL INFO Channel 02/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227470:1228033 [7] NCCL INFO Channel 03/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636091:2636684 [7] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-161-153:2636091:2636684 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-103:96895:97465 [7] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-161-103:96895:97465 [7] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-161-103:96895:97465 [7] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-161-103:96895:97465 [7] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-161-103:96895:97465 [7] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-161-103:96895:97465 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-164:1227470:1228033 [7] NCCL INFO Channel 00/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227470:1228033 [7] NCCL INFO Channel 01/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272157:272712 [7] NCCL INFO Channel 00/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227470:1228033 [7] NCCL INFO Channel 02/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-164:1227470:1228033 [7] NCCL INFO Channel 03/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272157:272712 [7] NCCL INFO Channel 01/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272157:272712 [7] NCCL INFO Channel 02/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272157:272712 [7] NCCL INFO Channel 03/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272157:272712 [7] NCCL INFO Channel 00/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272157:272712 [7] NCCL INFO Channel 01/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272157:272712 [7] NCCL INFO Channel 02/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-34:272157:272712 [7] NCCL INFO Channel 03/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557266:1557823 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32708:33276 [4] NCCL INFO Channel 00/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557266:1557823 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32708:33276 [4] NCCL INFO Channel 01/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557266:1557823 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32708:33276 [4] NCCL INFO Channel 02/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557266:1557823 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32708:33276 [4] NCCL INFO Channel 03/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557266:1557823 [0] NCCL INFO Channel 00/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32708:33276 [4] NCCL INFO Channel 00/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557266:1557823 [0] NCCL INFO Channel 01/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32708:33276 [4] NCCL INFO Channel 01/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636090:2636679 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-165-59:1557266:1557823 [0] NCCL INFO Channel 02/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-59:1557266:1557823 [0] NCCL INFO Channel 03/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-123:32708:33276 [4] NCCL INFO Channel 02/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-123:32708:33276 [4] NCCL INFO Channel 03/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-168-34:272152:272709 [2] NCCL INFO Channel 00/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272152:272709 [2] NCCL INFO Channel 01/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272152:272709 [2] NCCL INFO Channel 02/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272152:272709 [2] NCCL INFO Channel 03/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272152:272709 [2] NCCL INFO Channel 00/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272152:272709 [2] NCCL INFO Channel 01/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272152:272709 [2] NCCL INFO Channel 02/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-168-34:272152:272709 [2] NCCL INFO Channel 03/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96894:97456 [6] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-161-103:96894:97456 [6] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-161-103:96894:97456 [6] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-161-103:96894:97456 [6] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-161-153:2636090:2636679 [6] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-161-153:2636090:2636679 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-103:96894:97456 [6] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-161-103:96894:97456 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-161-142:62887:63451 [6] NCCL INFO Channel 00/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62887:63451 [6] NCCL INFO Channel 01/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62887:63451 [6] NCCL INFO Channel 02/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62887:63451 [6] NCCL INFO Channel 03/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62887:63451 [6] NCCL INFO Channel 00/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62887:63451 [6] NCCL INFO Channel 01/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62887:63451 [6] NCCL INFO Channel 02/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62887:63451 [6] NCCL INFO Channel 03/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364908:1365495 [6] NCCL INFO Channel 00/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364908:1365495 [6] NCCL INFO Channel 01/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364908:1365495 [6] NCCL INFO Channel 02/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364908:1365495 [6] NCCL INFO Channel 03/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364908:1365495 [6] NCCL INFO Channel 00/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364908:1365495 [6] NCCL INFO Channel 01/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364908:1365495 [6] NCCL INFO Channel 02/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-242:1364908:1365495 [6] NCCL INFO Channel 03/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-142:62882:63450 [1] NCCL INFO Channel 00/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62882:63450 [1] NCCL INFO Channel 01/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62882:63450 [1] NCCL INFO Channel 02/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62882:63450 [1] NCCL INFO Channel 03/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62882:63450 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62882:63450 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62882:63450 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-142:62882:63450 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364903:1365493 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364903:1365493 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364903:1365493 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364903:1365493 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364903:1365493 [1] NCCL INFO Channel 00/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364903:1365493 [1] NCCL INFO Channel 01/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364903:1365493 [1] NCCL INFO Channel 02/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-242:1364903:1365493 [1] NCCL INFO Channel 03/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96893:97462 [5] NCCL INFO Channel 00/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96893:97462 [5] NCCL INFO Channel 01/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96893:97462 [5] NCCL INFO Channel 02/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96893:97462 [5] NCCL INFO Channel 03/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96893:97462 [5] NCCL INFO Channel 00/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96893:97462 [5] NCCL INFO Channel 01/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96893:97462 [5] NCCL INFO Channel 02/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-103:96893:97462 [5] NCCL INFO Channel 03/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636089:2636682 [5] NCCL INFO Channel 00/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636089:2636682 [5] NCCL INFO Channel 01/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636089:2636682 [5] NCCL INFO Channel 02/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636089:2636682 [5] NCCL INFO Channel 03/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636089:2636682 [5] NCCL INFO Channel 00/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636089:2636682 [5] NCCL INFO Channel 01/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636089:2636682 [5] NCCL INFO Channel 02/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-153:2636089:2636682 [5] NCCL INFO Channel 03/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815964:2816558 [1] NCCL INFO Channel 00/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815964:2816558 [1] NCCL INFO Channel 01/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815964:2816558 [1] NCCL INFO Channel 02/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815964:2816558 [1] NCCL INFO Channel 03/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815964:2816558 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815964:2816558 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815964:2816558 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815964:2816558 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815970:2816556 [7] NCCL INFO Channel 00/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815970:2816556 [7] NCCL INFO Channel 01/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815970:2816556 [7] NCCL INFO Channel 02/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815970:2816556 [7] NCCL INFO Channel 03/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815970:2816556 [7] NCCL INFO Channel 00/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815970:2816556 [7] NCCL INFO Channel 01/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815970:2816556 [7] NCCL INFO Channel 02/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815970:2816556 [7] NCCL INFO Channel 03/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-178:2815968:2816559 [5] NCCL INFO Channel 00/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815968:2816559 [5] NCCL INFO Channel 01/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815968:2816559 [5] NCCL INFO Channel 02/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815968:2816559 [5] NCCL INFO Channel 03/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815968:2816559 [5] NCCL INFO Channel 00/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815968:2816559 [5] NCCL INFO Channel 01/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815966:2816554 [3] NCCL INFO Channel 00/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815968:2816559 [5] NCCL INFO Channel 02/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815968:2816559 [5] NCCL INFO Channel 03/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-178:2815966:2816554 [3] NCCL INFO Channel 01/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815966:2816554 [3] NCCL INFO Channel 02/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-178:2815966:2816554 [3] NCCL INFO Channel 03/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32705:33278 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815966:2816554 [3] NCCL INFO Channel 00/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32705:33278 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96889:97453 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815966:2816554 [3] NCCL INFO Channel 01/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32705:33278 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96889:97453 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815966:2816554 [3] NCCL INFO Channel 02/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32705:33278 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96889:97453 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-178:2815966:2816554 [3] NCCL INFO Channel 03/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32705:33278 [1] NCCL INFO Channel 00/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96889:97453 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32705:33278 [1] NCCL INFO Channel 01/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96889:97453 [1] NCCL INFO Channel 00/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96889:97453 [1] NCCL INFO Channel 01/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32705:33278 [1] NCCL INFO Channel 02/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96889:97453 [1] NCCL INFO Channel 02/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32705:33278 [1] NCCL INFO Channel 03/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96889:97453 [1] NCCL INFO Channel 03/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32707:33271 [3] NCCL INFO Channel 00/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32707:33271 [3] NCCL INFO Channel 01/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32707:33271 [3] NCCL INFO Channel 02/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32707:33271 [3] NCCL INFO Channel 03/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32707:33271 [3] NCCL INFO Channel 00/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32707:33271 [3] NCCL INFO Channel 01/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32707:33271 [3] NCCL INFO Channel 02/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32707:33271 [3] NCCL INFO Channel 03/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-123:32711:33274 [7] NCCL INFO Channel 00/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32711:33274 [7] NCCL INFO Channel 01/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32711:33274 [7] NCCL INFO Channel 02/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32711:33274 [7] NCCL INFO Channel 03/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32711:33274 [7] NCCL INFO Channel 00/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32711:33274 [7] NCCL INFO Channel 01/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32711:33274 [7] NCCL INFO Channel 02/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-123:32711:33274 [7] NCCL INFO Channel 03/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-59:1557268:1557822 [2] NCCL INFO Channel 00/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557268:1557822 [2] NCCL INFO Channel 01/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557268:1557822 [2] NCCL INFO Channel 02/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636085:2636678 [1] NCCL INFO Channel 00/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557268:1557822 [2] NCCL INFO Channel 03/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-59:1557268:1557822 [2] NCCL INFO Channel 00/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636085:2636678 [1] NCCL INFO Channel 01/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636085:2636678 [1] NCCL INFO Channel 02/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557268:1557822 [2] NCCL INFO Channel 01/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636085:2636678 [1] NCCL INFO Channel 03/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557268:1557822 [2] NCCL INFO Channel 02/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636085:2636678 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557268:1557822 [2] NCCL INFO Channel 03/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636085:2636678 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636085:2636678 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-153:2636085:2636678 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-123:32709:33280 [5] NCCL INFO Channel 00/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32709:33280 [5] NCCL INFO Channel 01/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32709:33280 [5] NCCL INFO Channel 02/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114514:115073 [2] NCCL INFO Channel 00/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32709:33280 [5] NCCL INFO Channel 03/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32709:33280 [5] NCCL INFO Channel 00/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114514:115073 [2] NCCL INFO Channel 01/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32709:33280 [5] NCCL INFO Channel 01/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-123:32709:33280 [5] NCCL INFO Channel 02/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114514:115073 [2] NCCL INFO Channel 02/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-123:32709:33280 [5] NCCL INFO Channel 03/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-161-78:114514:115073 [2] NCCL INFO Channel 03/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114514:115073 [2] NCCL INFO Channel 00/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114514:115073 [2] NCCL INFO Channel 01/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114514:115073 [2] NCCL INFO Channel 02/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114514:115073 [2] NCCL INFO Channel 03/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-78:114516:115080 [4] NCCL INFO Channel 00/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114516:115080 [4] NCCL INFO Channel 01/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114516:115080 [4] NCCL INFO Channel 02/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114516:115080 [4] NCCL INFO Channel 03/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114516:115080 [4] NCCL INFO Channel 00/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114516:115080 [4] NCCL INFO Channel 01/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114516:115080 [4] NCCL INFO Channel 02/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-78:114516:115080 [4] NCCL INFO Channel 03/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557270:1557825 [4] NCCL INFO Channel 00/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557270:1557825 [4] NCCL INFO Channel 01/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557270:1557825 [4] NCCL INFO Channel 02/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557270:1557825 [4] NCCL INFO Channel 03/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557270:1557825 [4] NCCL INFO Channel 00/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557270:1557825 [4] NCCL INFO Channel 01/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557270:1557825 [4] NCCL INFO Channel 02/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-59:1557270:1557825 [4] NCCL INFO Channel 03/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153563:154131 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-52:2722046:2722610 [1] NCCL INFO Channel 00/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153563:154131 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-52:2722046:2722610 [1] NCCL INFO Channel 01/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153563:154131 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-52:2722046:2722610 [1] NCCL INFO Channel 02/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153563:154131 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-52:2722046:2722610 [1] NCCL INFO Channel 03/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-52:2722046:2722610 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153563:154131 [1] NCCL INFO Channel 00/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-52:2722046:2722610 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153563:154131 [1] NCCL INFO Channel 01/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-52:2722046:2722610 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153563:154131 [1] NCCL INFO Channel 02/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-168-52:2722046:2722610 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-202:153563:154131 [1] NCCL INFO Channel 03/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114515:115081 [3] NCCL INFO Channel 00/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114515:115081 [3] NCCL INFO Channel 01/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114515:115081 [3] NCCL INFO Channel 02/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114515:115081 [3] NCCL INFO Channel 03/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114515:115081 [3] NCCL INFO Channel 00/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114515:115081 [3] NCCL INFO Channel 01/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114515:115081 [3] NCCL INFO Channel 02/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114515:115081 [3] NCCL INFO Channel 03/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557269:1557826 [3] NCCL INFO Channel 00/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557269:1557826 [3] NCCL INFO Channel 01/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557269:1557826 [3] NCCL INFO Channel 02/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557269:1557826 [3] NCCL INFO Channel 03/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557269:1557826 [3] NCCL INFO Channel 00/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557269:1557826 [3] NCCL INFO Channel 01/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557269:1557826 [3] NCCL INFO Channel 02/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-59:1557269:1557826 [3] NCCL INFO Channel 03/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-78:114513:115070 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114513:115070 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114513:115070 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114513:115070 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114513:115070 [1] NCCL INFO Channel 00/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114513:115070 [1] NCCL INFO Channel 01/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114513:115070 [1] NCCL INFO Channel 02/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557267:1557821 [1] NCCL INFO Channel 00/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-78:114513:115070 [1] NCCL INFO Channel 03/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557267:1557821 [1] NCCL INFO Channel 01/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557267:1557821 [1] NCCL INFO Channel 02/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557267:1557821 [1] NCCL INFO Channel 03/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557267:1557821 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557267:1557821 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557267:1557821 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-59:1557267:1557821 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-161-103:96892:97467 [4] NCCL INFO Channel 00/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96892:97467 [4] NCCL INFO Channel 01/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96892:97467 [4] NCCL INFO Channel 02/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96892:97467 [4] NCCL INFO Channel 03/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96892:97467 [4] NCCL INFO Channel 00/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96892:97467 [4] NCCL INFO Channel 01/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96892:97467 [4] NCCL INFO Channel 02/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96892:97467 [4] NCCL INFO Channel 03/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-202:153567:154126 [5] NCCL INFO Channel 00/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153567:154126 [5] NCCL INFO Channel 01/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153567:154126 [5] NCCL INFO Channel 02/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-52:2722050:2722607 [5] NCCL INFO Channel 00/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-52:2722050:2722607 [5] NCCL INFO Channel 01/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153567:154126 [5] NCCL INFO Channel 03/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-52:2722050:2722607 [5] NCCL INFO Channel 02/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153567:154126 [5] NCCL INFO Channel 00/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-52:2722050:2722607 [5] NCCL INFO Channel 03/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153567:154126 [5] NCCL INFO Channel 01/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-52:2722050:2722607 [5] NCCL INFO Channel 00/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153567:154126 [5] NCCL INFO Channel 02/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-52:2722050:2722607 [5] NCCL INFO Channel 01/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153567:154126 [5] NCCL INFO Channel 03/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-52:2722050:2722607 [5] NCCL INFO Channel 02/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-168-52:2722050:2722607 [5] NCCL INFO Channel 03/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-202:153569:154118 [7] NCCL INFO Channel 00/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153569:154118 [7] NCCL INFO Channel 01/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153569:154118 [7] NCCL INFO Channel 02/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153569:154118 [7] NCCL INFO Channel 03/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153569:154118 [7] NCCL INFO Channel 00/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153569:154118 [7] NCCL INFO Channel 01/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153569:154118 [7] NCCL INFO Channel 02/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-202:153569:154118 [7] NCCL INFO Channel 03/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722052:2722603 [7] NCCL INFO Channel 00/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722052:2722603 [7] NCCL INFO Channel 01/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722052:2722603 [7] NCCL INFO Channel 02/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722052:2722603 [7] NCCL INFO Channel 03/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722052:2722603 [7] NCCL INFO Channel 00/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722052:2722603 [7] NCCL INFO Channel 01/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722052:2722603 [7] NCCL INFO Channel 02/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-168-52:2722052:2722603 [7] NCCL INFO Channel 03/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636088:2636685 [4] NCCL INFO Channel 00/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636088:2636685 [4] NCCL INFO Channel 01/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636088:2636685 [4] NCCL INFO Channel 02/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96888:97459 [0] NCCL INFO Channel 00/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636088:2636685 [4] NCCL INFO Channel 03/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-153:2636088:2636685 [4] NCCL INFO Channel 00/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96888:97459 [0] NCCL INFO Channel 01/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636088:2636685 [4] NCCL INFO Channel 01/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96888:97459 [0] NCCL INFO Channel 02/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96888:97459 [0] NCCL INFO Channel 03/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636088:2636685 [4] NCCL INFO Channel 02/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96888:97459 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636088:2636685 [4] NCCL INFO Channel 03/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-161-103:96888:97459 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96888:97459 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96888:97459 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-202:153565:154121 [3] NCCL INFO Channel 00/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153565:154121 [3] NCCL INFO Channel 01/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153565:154121 [3] NCCL INFO Channel 02/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153565:154121 [3] NCCL INFO Channel 03/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153565:154121 [3] NCCL INFO Channel 00/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153565:154121 [3] NCCL INFO Channel 01/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153565:154121 [3] NCCL INFO Channel 02/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-202:153565:154121 [3] NCCL INFO Channel 03/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-52:2722048:2722605 [3] NCCL INFO Channel 00/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-52:2722048:2722605 [3] NCCL INFO Channel 01/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-52:2722048:2722605 [3] NCCL INFO Channel 02/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-52:2722048:2722605 [3] NCCL INFO Channel 03/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-52:2722048:2722605 [3] NCCL INFO Channel 00/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-52:2722048:2722605 [3] NCCL INFO Channel 01/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-52:2722048:2722605 [3] NCCL INFO Channel 02/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-168-52:2722048:2722605 [3] NCCL INFO Channel 03/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-161-153:2636084:2636681 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636084:2636681 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636084:2636681 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636084:2636681 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636084:2636681 [0] NCCL INFO Channel 00/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636084:2636681 [0] NCCL INFO Channel 01/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636084:2636681 [0] NCCL INFO Channel 02/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-153:2636084:2636681 [0] NCCL INFO Channel 03/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-161-103:96890:97457 [2] NCCL INFO Channel 00/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96890:97457 [2] NCCL INFO Channel 01/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96890:97457 [2] NCCL INFO Channel 02/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96890:97457 [2] NCCL INFO Channel 03/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96890:97457 [2] NCCL INFO Channel 00/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96890:97457 [2] NCCL INFO Channel 01/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96890:97457 [2] NCCL INFO Channel 02/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96890:97457 [2] NCCL INFO Channel 03/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636086:2636680 [2] NCCL INFO Channel 00/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636086:2636680 [2] NCCL INFO Channel 01/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636086:2636680 [2] NCCL INFO Channel 02/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636086:2636680 [2] NCCL INFO Channel 03/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636086:2636680 [2] NCCL INFO Channel 00/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636086:2636680 [2] NCCL INFO Channel 01/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636086:2636680 [2] NCCL INFO Channel 02/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-153:2636086:2636680 [2] NCCL INFO Channel 03/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-161-103:96895:97465 [7] NCCL INFO Channel 00/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96895:97465 [7] NCCL INFO Channel 01/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96895:97465 [7] NCCL INFO Channel 02/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96895:97465 [7] NCCL INFO Channel 03/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96895:97465 [7] NCCL INFO Channel 00/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96895:97465 [7] NCCL INFO Channel 01/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96895:97465 [7] NCCL INFO Channel 02/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96895:97465 [7] NCCL INFO Channel 03/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636091:2636684 [7] NCCL INFO Channel 00/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-153:2636091:2636684 [7] NCCL INFO Channel 01/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96894:97456 [6] NCCL INFO Channel 00/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636091:2636684 [7] NCCL INFO Channel 02/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96894:97456 [6] NCCL INFO Channel 01/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636091:2636684 [7] NCCL INFO Channel 03/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96894:97456 [6] NCCL INFO Channel 02/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636091:2636684 [7] NCCL INFO Channel 00/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96894:97456 [6] NCCL INFO Channel 03/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636091:2636684 [7] NCCL INFO Channel 01/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96894:97456 [6] NCCL INFO Channel 00/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636091:2636684 [7] NCCL INFO Channel 02/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96894:97456 [6] NCCL INFO Channel 01/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636091:2636684 [7] NCCL INFO Channel 03/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-161-103:96894:97456 [6] NCCL INFO Channel 02/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96894:97456 [6] NCCL INFO Channel 03/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636090:2636679 [6] NCCL INFO Channel 00/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636090:2636679 [6] NCCL INFO Channel 01/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636090:2636679 [6] NCCL INFO Channel 02/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636090:2636679 [6] NCCL INFO Channel 03/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636090:2636679 [6] NCCL INFO Channel 00/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636090:2636679 [6] NCCL INFO Channel 01/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636090:2636679 [6] NCCL INFO Channel 02/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-153:2636090:2636679 [6] NCCL INFO Channel 03/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-161-103:96891:97463 [3] NCCL INFO Connected all rings
+ip-26-0-161-103:96891:97463 [3] NCCL INFO Connected all trees
+ip-26-0-161-103:96891:97463 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-103:96891:97463 [3] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-161-103:96891:97463 [3] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-161-153:2636087:2636683 [3] NCCL INFO Connected all rings
+ip-26-0-161-153:2636087:2636683 [3] NCCL INFO Connected all trees
+ip-26-0-161-153:2636087:2636683 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-153:2636087:2636683 [3] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-161-153:2636087:2636683 [3] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-161-103:96891:97463 [3] NCCL INFO comm 0xae75180 rank 0 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0xa1e643239b1aeb8 - Init COMPLETE
+ip-26-0-161-153:2636087:2636683 [3] NCCL INFO comm 0xaac9fb0 rank 1 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0xa1e643239b1aeb8 - Init COMPLETE
+ip-26-0-161-142:62884:63452 [3] NCCL INFO Connected all rings
+ip-26-0-161-142:62884:63452 [3] NCCL INFO Connected all trees
+ip-26-0-161-142:62884:63452 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-142:62884:63452 [3] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-161-142:62884:63452 [3] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-161-142:62884:63452 [3] NCCL INFO comm 0xb7c26d0 rank 1 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0x2007c24124eb1e16 - Init COMPLETE
+ip-26-0-165-164:1227466:1228032 [3] NCCL INFO Connected all rings
+ip-26-0-165-164:1227466:1228032 [3] NCCL INFO Connected all trees
+ip-26-0-165-164:1227466:1228032 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-164:1227466:1228032 [3] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-165-164:1227466:1228032 [3] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-160-242:1364905:1365497 [3] NCCL INFO Connected all rings
+ip-26-0-160-242:1364905:1365497 [3] NCCL INFO Connected all trees
+ip-26-0-160-242:1364905:1365497 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-242:1364905:1365497 [3] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-160-242:1364905:1365497 [3] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-165-164:1227466:1228032 [3] NCCL INFO comm 0xae9ae50 rank 0 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0x33412db31e485107 - Init COMPLETE
+ip-26-0-168-34:272153:272711 [3] NCCL INFO Connected all rings
+ip-26-0-168-34:272153:272711 [3] NCCL INFO Connected all trees
+ip-26-0-168-34:272153:272711 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-34:272153:272711 [3] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-168-34:272153:272711 [3] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-160-242:1364906:1365487 [4] NCCL INFO Connected all rings
+ip-26-0-160-242:1364906:1365487 [4] NCCL INFO Connected all trees
+ip-26-0-160-242:1364906:1365487 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-242:1364906:1365487 [4] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-160-242:1364906:1365487 [4] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-160-242:1364905:1365497 [3] NCCL INFO comm 0xba43580 rank 0 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0x2007c24124eb1e16 - Init COMPLETE
+ip-26-0-161-123:32704:33268 [0] NCCL INFO Connected all rings
+ip-26-0-161-123:32704:33268 [0] NCCL INFO Connected all trees
+ip-26-0-161-123:32704:33268 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-123:32704:33268 [0] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-161-123:32704:33268 [0] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-160-242:1364906:1365487 [4] NCCL INFO comm 0xc1abec0 rank 0 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0x1cec219b593b5b6e - Init COMPLETE
+ip-26-0-168-34:272153:272711 [3] NCCL INFO comm 0xa5d3400 rank 1 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0x33412db31e485107 - Init COMPLETE
+ip-26-0-161-178:2815963:2816553 [0] NCCL INFO Connected all rings
+ip-26-0-161-178:2815963:2816553 [0] NCCL INFO Connected all trees
+ip-26-0-161-178:2815963:2816553 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-178:2815963:2816553 [0] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-161-178:2815963:2816553 [0] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-161-142:62885:63447 [4] NCCL INFO Connected all rings
+ip-26-0-161-142:62885:63447 [4] NCCL INFO Connected all trees
+ip-26-0-161-142:62885:63447 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-142:62885:63447 [4] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-161-142:62885:63447 [4] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-161-123:32704:33268 [0] NCCL INFO comm 0xc675d50 rank 0 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0x36840ec6b4075829 - Init COMPLETE
+ip-26-0-161-178:2815963:2816553 [0] NCCL INFO comm 0xb593b90 rank 1 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0x36840ec6b4075829 - Init COMPLETE
+ip-26-0-168-30:235651:236205 [3] NCCL INFO Connected all rings
+ip-26-0-168-30:235651:236205 [3] NCCL INFO Connected all trees
+ip-26-0-168-30:235651:236205 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-30:235651:236205 [3] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-168-30:235651:236205 [3] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-160-225:124161:124732 [1] NCCL INFO Connected all rings
+ip-26-0-160-225:124161:124732 [1] NCCL INFO Connected all trees
+ip-26-0-160-225:124161:124732 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-225:124161:124732 [1] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-160-225:124161:124732 [1] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-161-142:62885:63447 [4] NCCL INFO comm 0xc030ad0 rank 1 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0x1cec219b593b5b6e - Init COMPLETE
+ip-26-0-165-59:1557272:1557827 [6] NCCL INFO Connected all rings
+ip-26-0-165-59:1557272:1557827 [6] NCCL INFO Connected all trees
+ip-26-0-165-59:1557272:1557827 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-59:1557272:1557827 [6] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-165-59:1557272:1557827 [6] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-165-59:1557273:1557828 [7] NCCL INFO Connected all rings
+ip-26-0-165-59:1557273:1557828 [7] NCCL INFO Connected all trees
+ip-26-0-165-59:1557273:1557828 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-59:1557273:1557828 [7] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-165-59:1557273:1557828 [7] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-168-30:235651:236205 [3] NCCL INFO comm 0xa4a2b40 rank 1 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0xcda45e7cac3649b2 - Init COMPLETE
+ip-26-0-161-78:114518:115082 [6] NCCL INFO Connected all rings
+ip-26-0-161-78:114518:115082 [6] NCCL INFO Connected all trees
+ip-26-0-161-78:114518:115082 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-78:114518:115082 [6] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-161-78:114518:115082 [6] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-160-242:1364909:1365485 [7] NCCL INFO Connected all rings
+ip-26-0-160-242:1364909:1365485 [7] NCCL INFO Connected all trees
+ip-26-0-160-242:1364909:1365485 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-242:1364909:1365485 [7] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-160-242:1364909:1365485 [7] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-168-52:2722051:2722608 [6] NCCL INFO Connected all rings
+ip-26-0-168-52:2722051:2722608 [6] NCCL INFO Connected all trees
+ip-26-0-168-52:2722051:2722608 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-52:2722051:2722608 [6] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-160-225:124161:124732 [1] NCCL INFO comm 0xaa0bbf0 rank 0 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0x833fbeee0b434137 - Init COMPLETE
+ip-26-0-168-52:2722051:2722608 [6] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-161-138:96564:97172 [1] NCCL INFO Connected all rings
+ip-26-0-161-138:96564:97172 [1] NCCL INFO Connected all trees
+ip-26-0-161-138:96564:97172 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-138:96564:97172 [1] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-161-138:96564:97172 [1] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-161-78:114519:115084 [7] NCCL INFO Connected all rings
+ip-26-0-161-78:114519:115084 [7] NCCL INFO Connected all trees
+ip-26-0-161-78:114519:115084 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-78:114519:115084 [7] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-161-78:114519:115084 [7] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-165-59:1557272:1557827 [6] NCCL INFO comm 0xb426cd0 rank 1 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0x401d2ca93e58b7b3 - Init COMPLETE
+ip-26-0-161-142:62888:63446 [7] NCCL INFO Connected all rings
+ip-26-0-161-142:62888:63446 [7] NCCL INFO Connected all trees
+ip-26-0-161-142:62888:63446 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-142:62888:63446 [7] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-161-142:62888:63446 [7] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-165-131:1630208:1630811 [3] NCCL INFO Connected all rings
+ip-26-0-165-131:1630208:1630811 [3] NCCL INFO Connected all trees
+ip-26-0-165-131:1630208:1630811 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-131:1630208:1630811 [3] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-165-131:1630208:1630811 [3] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-161-78:114518:115082 [6] NCCL INFO comm 0xb895930 rank 0 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0x401d2ca93e58b7b3 - Init COMPLETE
+ip-26-0-165-59:1557273:1557828 [7] NCCL INFO comm 0xc011560 rank 1 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0x1bdd1fa5bfd14ebb - Init COMPLETE
+ip-26-0-168-52:2722049:2722606 [4] NCCL INFO Connected all rings
+ip-26-0-168-52:2722049:2722606 [4] NCCL INFO Connected all trees
+ip-26-0-168-52:2722049:2722606 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-52:2722049:2722606 [4] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-168-52:2722049:2722606 [4] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-160-242:1364909:1365485 [7] NCCL INFO comm 0xc067b00 rank 0 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0xfdfbbee8bd17a511 - Init COMPLETE
+ip-26-0-168-52:2722051:2722608 [6] NCCL INFO comm 0xb2724b0 rank 1 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0xd2565f661f0f5428 - Init COMPLETE
+ip-26-0-161-78:114519:115084 [7] NCCL INFO comm 0xa4dd970 rank 0 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0x1bdd1fa5bfd14ebb - Init COMPLETE
+ip-26-0-161-138:96564:97172 [1] NCCL INFO comm 0xb83a1e0 rank 1 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0x833fbeee0b434137 - Init COMPLETE
+ip-26-0-160-225:124162:124740 [2] NCCL INFO Connected all rings
+ip-26-0-160-225:124162:124740 [2] NCCL INFO Connected all trees
+ip-26-0-160-225:124162:124740 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-225:124162:124740 [2] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-160-225:124162:124740 [2] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-161-142:62888:63446 [7] NCCL INFO comm 0xb379ed0 rank 1 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0xfdfbbee8bd17a511 - Init COMPLETE
+ip-26-0-165-131:1630208:1630811 [3] NCCL INFO comm 0xa4e7ad0 rank 0 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0xcda45e7cac3649b2 - Init COMPLETE
+ip-26-0-161-138:96565:97175 [2] NCCL INFO Connected all rings
+ip-26-0-161-138:96565:97175 [2] NCCL INFO Connected all trees
+ip-26-0-161-138:96565:97175 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-138:96565:97175 [2] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-161-138:96565:97175 [2] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-168-52:2722049:2722606 [4] NCCL INFO comm 0xa834490 rank 1 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0x9924b60493ed5710 - Init COMPLETE
+ip-26-0-165-59:1557271:1557824 [5] NCCL INFO Connected all rings
+ip-26-0-165-59:1557271:1557824 [5] NCCL INFO Connected all trees
+ip-26-0-165-59:1557271:1557824 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-59:1557271:1557824 [5] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-165-59:1557271:1557824 [5] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-161-78:114517:115076 [5] NCCL INFO Connected all rings
+ip-26-0-161-78:114517:115076 [5] NCCL INFO Connected all trees
+ip-26-0-161-78:114517:115076 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-78:114517:115076 [5] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-161-78:114517:115076 [5] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-160-225:124162:124740 [2] NCCL INFO comm 0xc4a9dc0 rank 0 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0x1a0854df454e0f6f - Init COMPLETE
+ip-26-0-161-138:96565:97175 [2] NCCL INFO comm 0xc420790 rank 1 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0x1a0854df454e0f6f - Init COMPLETE
+ip-26-0-161-178:2815969:2816560 [6] NCCL INFO Connected all rings
+ip-26-0-161-178:2815969:2816560 [6] NCCL INFO Connected all trees
+ip-26-0-161-123:32710:33282 [6] NCCL INFO Connected all rings
+ip-26-0-161-123:32710:33282 [6] NCCL INFO Connected all trees
+ip-26-0-161-123:32710:33282 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-178:2815969:2816560 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-178:2815969:2816560 [6] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-161-178:2815969:2816560 [6] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-161-123:32710:33282 [6] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-161-123:32710:33282 [6] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-161-103:96893:97462 [5] NCCL INFO Connected all rings
+ip-26-0-161-103:96893:97462 [5] NCCL INFO Connected all trees
+ip-26-0-161-103:96893:97462 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-103:96893:97462 [5] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-161-103:96893:97462 [5] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-160-225:124160:124730 [0] NCCL INFO Connected all rings
+ip-26-0-160-225:124160:124730 [0] NCCL INFO Connected all trees
+ip-26-0-160-225:124160:124730 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-225:124160:124730 [0] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-160-225:124160:124730 [0] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-168-30:235648:236202 [0] NCCL INFO Connected all rings
+ip-26-0-168-30:235648:236202 [0] NCCL INFO Connected all trees
+ip-26-0-168-30:235648:236202 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-30:235648:236202 [0] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-168-30:235648:236202 [0] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-160-225:124166:124742 [6] NCCL INFO Connected all rings
+ip-26-0-160-225:124166:124742 [6] NCCL INFO Connected all trees
+ip-26-0-160-225:124166:124742 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-225:124166:124742 [6] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-160-225:124166:124742 [6] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-165-59:1557271:1557824 [5] NCCL INFO comm 0xa4c5ba0 rank 1 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0xf54fd7bc2ff2310e - Init COMPLETE
+ip-26-0-165-164:1227468:1228023 [5] NCCL INFO Connected all rings
+ip-26-0-165-164:1227468:1228023 [5] NCCL INFO Connected all trees
+ip-26-0-165-164:1227468:1228023 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-164:1227468:1228023 [5] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-165-164:1227468:1228023 [5] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-160-225:124163:124743 [3] NCCL INFO Connected all rings
+ip-26-0-160-225:124163:124743 [3] NCCL INFO Connected all trees
+ip-26-0-160-225:124163:124743 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-225:124163:124743 [3] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-160-225:124163:124743 [3] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-161-123:32706:33272 [2] NCCL INFO Connected all rings
+ip-26-0-161-123:32706:33272 [2] NCCL INFO Connected all trees
+ip-26-0-161-123:32706:33272 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-123:32706:33272 [2] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-161-123:32706:33272 [2] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-161-178:2815965:2816555 [2] NCCL INFO Connected all rings
+ip-26-0-161-178:2815965:2816555 [2] NCCL INFO Connected all trees
+ip-26-0-161-178:2815965:2816555 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-178:2815965:2816555 [2] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-161-178:2815965:2816555 [2] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-160-225:124164:124734 [4] NCCL INFO Connected all rings
+ip-26-0-160-225:124164:124734 [4] NCCL INFO Connected all trees
+ip-26-0-160-225:124164:124734 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-138:96566:97176 [3] NCCL INFO Connected all rings
+ip-26-0-161-138:96566:97176 [3] NCCL INFO Connected all trees
+ip-26-0-160-225:124164:124734 [4] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-160-225:124164:124734 [4] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-161-138:96566:97176 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-138:96566:97176 [3] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-161-138:96566:97176 [3] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-165-131:1630207:1630815 [2] NCCL INFO Connected all rings
+ip-26-0-165-131:1630207:1630815 [2] NCCL INFO Connected all trees
+ip-26-0-165-131:1630207:1630815 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-131:1630207:1630815 [2] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-165-131:1630207:1630815 [2] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-168-34:272156:272708 [6] NCCL INFO Connected all rings
+ip-26-0-168-34:272156:272708 [6] NCCL INFO Connected all trees
+ip-26-0-168-34:272156:272708 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-34:272156:272708 [6] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-168-34:272156:272708 [6] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-168-30:235655:236209 [7] NCCL INFO Connected all rings
+ip-26-0-168-30:235655:236209 [7] NCCL INFO Connected all trees
+ip-26-0-168-30:235655:236209 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-30:235655:236209 [7] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-168-30:235655:236209 [7] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-160-225:124165:124736 [5] NCCL INFO Connected all rings
+ip-26-0-160-225:124165:124736 [5] NCCL INFO Connected all trees
+ip-26-0-160-225:124165:124736 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-225:124165:124736 [5] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-160-225:124165:124736 [5] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-161-78:114517:115076 [5] NCCL INFO comm 0xc29fc50 rank 0 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0xf54fd7bc2ff2310e - Init COMPLETE
+ip-26-0-168-30:235649:236208 [1] NCCL INFO Connected all rings
+ip-26-0-168-30:235649:236208 [1] NCCL INFO Connected all trees
+ip-26-0-168-30:235649:236208 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-30:235649:236208 [1] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-168-30:235649:236208 [1] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-161-153:2636089:2636682 [5] NCCL INFO Connected all rings
+ip-26-0-161-153:2636089:2636682 [5] NCCL INFO Connected all trees
+ip-26-0-161-153:2636089:2636682 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-153:2636089:2636682 [5] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-161-153:2636089:2636682 [5] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-161-142:62883:63448 [2] NCCL INFO Connected all rings
+ip-26-0-161-142:62883:63448 [2] NCCL INFO Connected all trees
+ip-26-0-161-142:62883:63448 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-142:62883:63448 [2] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-161-142:62883:63448 [2] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-168-30:235654:236206 [6] NCCL INFO Connected all rings
+ip-26-0-168-30:235654:236206 [6] NCCL INFO Connected all trees
+ip-26-0-168-30:235654:236206 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-30:235654:236206 [6] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-168-30:235654:236206 [6] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-161-142:62881:63445 [0] NCCL INFO Connected all rings
+ip-26-0-161-142:62881:63445 [0] NCCL INFO Connected all trees
+ip-26-0-161-142:62881:63445 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-142:62881:63445 [0] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-161-142:62881:63445 [0] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-161-138:96563:97171 [0] NCCL INFO Connected all rings
+ip-26-0-161-138:96563:97171 [0] NCCL INFO Connected all trees
+ip-26-0-161-138:96563:97171 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-138:96563:97171 [0] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-161-138:96563:97171 [0] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-168-30:235652:236203 [4] NCCL INFO Connected all rings
+ip-26-0-168-30:235652:236203 [4] NCCL INFO Connected all trees
+ip-26-0-168-30:235652:236203 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-225:124167:124744 [7] NCCL INFO Connected all rings
+ip-26-0-160-225:124167:124744 [7] NCCL INFO Connected all trees
+ip-26-0-160-225:124167:124744 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-225:124167:124744 [7] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-160-225:124167:124744 [7] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-168-30:235652:236203 [4] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-168-30:235652:236203 [4] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-161-123:32710:33282 [6] NCCL INFO comm 0xbb37cb0 rank 0 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0x457d552d1a4664ec - Init COMPLETE
+ip-26-0-161-178:2815969:2816560 [6] NCCL INFO comm 0xbc6d8e0 rank 1 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0x457d552d1a4664ec - Init COMPLETE
+ip-26-0-165-164:1227469:1228026 [6] NCCL INFO Connected all rings
+ip-26-0-165-164:1227469:1228026 [6] NCCL INFO Connected all trees
+ip-26-0-165-164:1227469:1228026 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-164:1227469:1228026 [6] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-165-164:1227469:1228026 [6] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-161-142:62886:63449 [5] NCCL INFO Connected all rings
+ip-26-0-161-142:62886:63449 [5] NCCL INFO Connected all trees
+ip-26-0-161-142:62886:63449 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-142:62886:63449 [5] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-161-142:62886:63449 [5] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-168-30:235653:236204 [5] NCCL INFO Connected all rings
+ip-26-0-168-30:235653:236204 [5] NCCL INFO Connected all trees
+ip-26-0-168-30:235653:236204 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-30:235653:236204 [5] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-168-30:235653:236204 [5] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-161-103:96893:97462 [5] NCCL INFO comm 0xa8ed100 rank 0 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0x20c1a1bfbac058 - Init COMPLETE
+ip-26-0-161-153:2636085:2636678 [1] NCCL INFO Connected all rings
+ip-26-0-161-153:2636085:2636678 [1] NCCL INFO Connected all trees
+ip-26-0-161-153:2636085:2636678 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-225:124160:124730 [0] NCCL INFO comm 0xb44d4a0 rank 0 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0xfa4852f4aecd80ec - Init COMPLETE
+ip-26-0-161-153:2636085:2636678 [1] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-161-153:2636085:2636678 [1] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-165-59:1557266:1557823 [0] NCCL INFO Connected all rings
+ip-26-0-165-59:1557266:1557823 [0] NCCL INFO Connected all trees
+ip-26-0-165-59:1557266:1557823 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-59:1557266:1557823 [0] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-165-59:1557266:1557823 [0] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-165-131:1630210:1630809 [5] NCCL INFO Connected all rings
+ip-26-0-165-131:1630210:1630809 [5] NCCL INFO Connected all trees
+ip-26-0-165-131:1630210:1630809 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-131:1630210:1630809 [5] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-165-131:1630210:1630809 [5] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-168-52:2722045:2722609 [0] NCCL INFO Connected all rings
+ip-26-0-168-52:2722045:2722609 [0] NCCL INFO Connected all trees
+ip-26-0-168-52:2722045:2722609 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-52:2722045:2722609 [0] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-168-52:2722045:2722609 [0] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-160-225:124166:124742 [6] NCCL INFO comm 0xb971350 rank 0 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0x73aa7cd217337991 - Init COMPLETE
+ip-26-0-161-138:96568:97174 [5] NCCL INFO Connected all rings
+ip-26-0-161-138:96568:97174 [5] NCCL INFO Connected all trees
+ip-26-0-161-138:96568:97174 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-138:96568:97174 [5] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-161-138:96568:97174 [5] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-168-30:235648:236202 [0] NCCL INFO comm 0xa92e1d0 rank 1 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0xa38d5ad285b7dfb1 - Init COMPLETE
+ip-26-0-161-138:96570:97178 [7] NCCL INFO Connected all rings
+ip-26-0-161-178:2815967:2816557 [4] NCCL INFO Connected all rings
+ip-26-0-161-178:2815967:2816557 [4] NCCL INFO Connected all trees
+ip-26-0-161-138:96570:97178 [7] NCCL INFO Connected all trees
+ip-26-0-165-164:1227468:1228023 [5] NCCL INFO comm 0xbba64c0 rank 0 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0x218724ca16147a07 - Init COMPLETE
+ip-26-0-161-138:96570:97178 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-138:96570:97178 [7] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-161-138:96570:97178 [7] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-161-178:2815967:2816557 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-178:2815967:2816557 [4] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-161-178:2815967:2816557 [4] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-165-131:1630209:1630807 [4] NCCL INFO Connected all rings
+ip-26-0-165-131:1630209:1630807 [4] NCCL INFO Connected all trees
+ip-26-0-165-131:1630209:1630807 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-78:114512:115074 [0] NCCL INFO Connected all rings
+ip-26-0-161-78:114512:115074 [0] NCCL INFO Connected all trees
+ip-26-0-161-78:114512:115074 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-131:1630209:1630807 [4] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-165-131:1630209:1630807 [4] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-161-78:114512:115074 [0] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-161-78:114512:115074 [0] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-160-242:1364907:1365491 [5] NCCL INFO Connected all rings
+ip-26-0-160-242:1364907:1365491 [5] NCCL INFO Connected all trees
+ip-26-0-161-138:96569:97177 [6] NCCL INFO Connected all rings
+ip-26-0-160-242:1364907:1365491 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-242:1364907:1365491 [5] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-160-242:1364907:1365491 [5] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-161-138:96569:97177 [6] NCCL INFO Connected all trees
+ip-26-0-161-138:96569:97177 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-138:96569:97177 [6] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-161-138:96569:97177 [6] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-165-131:1630206:1630817 [1] NCCL INFO Connected all rings
+ip-26-0-165-131:1630206:1630817 [1] NCCL INFO Connected all trees
+ip-26-0-165-131:1630206:1630817 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-131:1630206:1630817 [1] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-165-131:1630206:1630817 [1] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-161-123:32708:33276 [4] NCCL INFO Connected all rings
+ip-26-0-161-123:32708:33276 [4] NCCL INFO Connected all trees
+ip-26-0-161-123:32708:33276 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-123:32708:33276 [4] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-161-123:32708:33276 [4] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-165-131:1630212:1630819 [7] NCCL INFO Connected all rings
+ip-26-0-165-131:1630212:1630819 [7] NCCL INFO Connected all trees
+ip-26-0-165-131:1630212:1630819 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-131:1630212:1630819 [7] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-165-131:1630212:1630819 [7] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-165-131:1630205:1630805 [0] NCCL INFO Connected all rings
+ip-26-0-165-131:1630205:1630805 [0] NCCL INFO Connected all trees
+ip-26-0-165-131:1630205:1630805 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-30:235650:236207 [2] NCCL INFO Connected all rings
+ip-26-0-168-30:235650:236207 [2] NCCL INFO Connected all trees
+ip-26-0-168-30:235650:236207 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-131:1630205:1630805 [0] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-165-131:1630205:1630805 [0] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-168-30:235650:236207 [2] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-168-30:235650:236207 [2] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-161-123:32706:33272 [2] NCCL INFO comm 0xbc15fa0 rank 0 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0x167967f1c8f1f887 - Init COMPLETE
+ip-26-0-160-242:1364904:1365489 [2] NCCL INFO Connected all rings
+ip-26-0-160-242:1364904:1365489 [2] NCCL INFO Connected all trees
+ip-26-0-160-242:1364904:1365489 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-242:1364904:1365489 [2] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-160-242:1364904:1365489 [2] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-161-142:62887:63451 [6] NCCL INFO Connected all rings
+ip-26-0-161-142:62887:63451 [6] NCCL INFO Connected all trees
+ip-26-0-161-142:62887:63451 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-142:62887:63451 [6] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-161-142:62887:63451 [6] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-168-34:272155:272707 [5] NCCL INFO Connected all rings
+ip-26-0-168-34:272155:272707 [5] NCCL INFO Connected all trees
+ip-26-0-168-34:272155:272707 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-34:272155:272707 [5] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-168-34:272155:272707 [5] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-160-242:1364902:1365483 [0] NCCL INFO Connected all rings
+ip-26-0-160-242:1364902:1365483 [0] NCCL INFO Connected all trees
+ip-26-0-160-242:1364902:1365483 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-242:1364902:1365483 [0] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-160-242:1364902:1365483 [0] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-161-178:2815965:2816555 [2] NCCL INFO comm 0xade8a60 rank 1 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0x167967f1c8f1f887 - Init COMPLETE
+ip-26-0-161-138:96566:97176 [3] NCCL INFO comm 0xb52cf50 rank 1 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0xf118ca765e2897d0 - Init COMPLETE
+ip-26-0-168-34:272156:272708 [6] NCCL INFO comm 0xc1b3520 rank 1 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0xff9654fa091a52a3 - Init COMPLETE
+ip-26-0-168-30:235655:236209 [7] NCCL INFO comm 0xb467400 rank 1 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0xda3cb15f18bb32c8 - Init COMPLETE
+ip-26-0-165-131:1630207:1630815 [2] NCCL INFO comm 0xb568310 rank 0 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0xafc8bbf8e2a35859 - Init COMPLETE
+ip-26-0-160-225:124164:124734 [4] NCCL INFO comm 0xb16e730 rank 0 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0x174515b09e2bfef0 - Init COMPLETE
+ip-26-0-160-225:124165:124736 [5] NCCL INFO comm 0xaa4d3d0 rank 0 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0x3b05320dbc041b06 - Init COMPLETE
+ip-26-0-161-138:96567:97173 [4] NCCL INFO Connected all rings
+ip-26-0-161-138:96567:97173 [4] NCCL INFO Connected all trees
+ip-26-0-161-138:96567:97173 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-153:2636089:2636682 [5] NCCL INFO comm 0xa9432a0 rank 1 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0x20c1a1bfbac058 - Init COMPLETE
+ip-26-0-161-138:96567:97173 [4] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-161-138:96567:97173 [4] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-160-225:124163:124743 [3] NCCL INFO comm 0xb6e2ee0 rank 0 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0xf118ca765e2897d0 - Init COMPLETE
+ip-26-0-165-59:1557268:1557822 [2] NCCL INFO Connected all rings
+ip-26-0-165-59:1557268:1557822 [2] NCCL INFO Connected all trees
+ip-26-0-165-59:1557268:1557822 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-59:1557268:1557822 [2] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-165-59:1557268:1557822 [2] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-161-138:96563:97171 [0] NCCL INFO comm 0xbbcef70 rank 1 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0xfa4852f4aecd80ec - Init COMPLETE
+ip-26-0-160-225:124167:124744 [7] NCCL INFO comm 0xbe5bb90 rank 0 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0x6dc7a8370918fa1 - Init COMPLETE
+ip-26-0-165-164:1227469:1228026 [6] NCCL INFO comm 0xa4d9fc0 rank 0 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0xff9654fa091a52a3 - Init COMPLETE
+ip-26-0-160-242:1364908:1365495 [6] NCCL INFO Connected all rings
+ip-26-0-161-142:62883:63448 [2] NCCL INFO comm 0xad3c290 rank 1 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0x3b173d6ca378a933 - Init COMPLETE
+ip-26-0-160-242:1364908:1365495 [6] NCCL INFO Connected all trees
+ip-26-0-160-242:1364908:1365495 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-242:1364908:1365495 [6] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-160-242:1364908:1365495 [6] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-168-30:235649:236208 [1] NCCL INFO comm 0xbbf8980 rank 1 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0xfd6fa36a1b1a13e3 - Init COMPLETE
+ip-26-0-161-142:62881:63445 [0] NCCL INFO comm 0xc311190 rank 1 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0xdc2325cededd95f8 - Init COMPLETE
+ip-26-0-168-30:235654:236206 [6] NCCL INFO comm 0xaf1c480 rank 1 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0xab67ef3e7d9c3e72 - Init COMPLETE
+ip-26-0-168-30:235652:236203 [4] NCCL INFO comm 0xbc9fd60 rank 1 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0x2d3503a923c539a9 - Init COMPLETE
+ip-26-0-161-103:96889:97453 [1] NCCL INFO Connected all rings
+ip-26-0-161-103:96889:97453 [1] NCCL INFO Connected all trees
+ip-26-0-161-103:96889:97453 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-131:1630211:1630813 [6] NCCL INFO Connected all rings
+ip-26-0-165-131:1630211:1630813 [6] NCCL INFO Connected all trees
+ip-26-0-165-131:1630211:1630813 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-103:96889:97453 [1] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-161-103:96889:97453 [1] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-165-131:1630211:1630813 [6] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-165-131:1630211:1630813 [6] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-161-142:62886:63449 [5] NCCL INFO comm 0xc002a90 rank 1 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0xb5b1e5768650a111 - Init COMPLETE
+ip-26-0-165-59:1557266:1557823 [0] NCCL INFO comm 0xc141870 rank 1 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0xd3ea008df278b7a5 - Init COMPLETE
+ip-26-0-168-30:235653:236204 [5] NCCL INFO comm 0xb11ebf0 rank 1 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0x9fbcd95ac374d17c - Init COMPLETE
+ip-26-0-168-52:2722045:2722609 [0] NCCL INFO comm 0xa54d320 rank 1 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0x7a9a6457da866605 - Init COMPLETE
+ip-26-0-161-153:2636085:2636678 [1] NCCL INFO comm 0xc3dc510 rank 1 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0xb2131d0b36358051 - Init COMPLETE
+ip-26-0-161-142:62882:63450 [1] NCCL INFO Connected all rings
+ip-26-0-161-142:62882:63450 [1] NCCL INFO Connected all trees
+ip-26-0-161-142:62882:63450 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-142:62882:63450 [1] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-161-142:62882:63450 [1] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-168-34:272150:272713 [0] NCCL INFO Connected all rings
+ip-26-0-168-34:272150:272713 [0] NCCL INFO Connected all trees
+ip-26-0-168-34:272150:272713 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-34:272150:272713 [0] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-168-34:272150:272713 [0] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-160-242:1364903:1365493 [1] NCCL INFO Connected all rings
+ip-26-0-160-242:1364903:1365493 [1] NCCL INFO Connected all trees
+ip-26-0-160-242:1364903:1365493 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-242:1364903:1365493 [1] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-160-242:1364903:1365493 [1] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-168-52:2722047:2722604 [2] NCCL INFO Connected all rings
+ip-26-0-168-52:2722047:2722604 [2] NCCL INFO Connected all trees
+ip-26-0-168-52:2722047:2722604 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-52:2722047:2722604 [2] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-168-52:2722047:2722604 [2] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-161-138:96568:97174 [5] NCCL INFO comm 0xaadad80 rank 1 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0x3b05320dbc041b06 - Init COMPLETE
+ip-26-0-165-131:1630210:1630809 [5] NCCL INFO comm 0xbeb2750 rank 0 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0x9fbcd95ac374d17c - Init COMPLETE
+ip-26-0-161-138:96570:97178 [7] NCCL INFO comm 0xc2d74c0 rank 1 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0x6dc7a8370918fa1 - Init COMPLETE
+ip-26-0-161-78:114512:115074 [0] NCCL INFO comm 0xb2a1520 rank 0 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0xd3ea008df278b7a5 - Init COMPLETE
+ip-26-0-165-164:1227467:1228021 [4] NCCL INFO Connected all rings
+ip-26-0-165-164:1227467:1228021 [4] NCCL INFO Connected all trees
+ip-26-0-165-164:1227467:1228021 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-164:1227467:1228021 [4] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-165-164:1227467:1228021 [4] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-161-178:2815967:2816557 [4] NCCL INFO comm 0xa843590 rank 1 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0xe123c9967146b25f - Init COMPLETE
+ip-26-0-161-123:32708:33276 [4] NCCL INFO comm 0xc48b100 rank 0 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0xe123c9967146b25f - Init COMPLETE
+ip-26-0-160-242:1364907:1365491 [5] NCCL INFO comm 0xa8af0a0 rank 0 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0xb5b1e5768650a111 - Init COMPLETE
+ip-26-0-168-30:235650:236207 [2] NCCL INFO comm 0xaf10d30 rank 1 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0xafc8bbf8e2a35859 - Init COMPLETE
+ip-26-0-165-131:1630209:1630807 [4] NCCL INFO comm 0xb1d0150 rank 0 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0x2d3503a923c539a9 - Init COMPLETE
+ip-26-0-161-138:96569:97177 [6] NCCL INFO comm 0xb086aa0 rank 1 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0x73aa7cd217337991 - Init COMPLETE
+ip-26-0-165-131:1630206:1630817 [1] NCCL INFO comm 0xbb509d0 rank 0 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0xfd6fa36a1b1a13e3 - Init COMPLETE
+ip-26-0-161-142:62887:63451 [6] NCCL INFO comm 0xa894680 rank 1 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0x7ba2dc144f01ded2 - Init COMPLETE
+ip-26-0-165-131:1630212:1630819 [7] NCCL INFO comm 0xa6db000 rank 0 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0xda3cb15f18bb32c8 - Init COMPLETE
+ip-26-0-160-242:1364904:1365489 [2] NCCL INFO comm 0xb840bf0 rank 0 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0x3b173d6ca378a933 - Init COMPLETE
+ip-26-0-168-34:272157:272712 [7] NCCL INFO Connected all rings
+ip-26-0-168-34:272157:272712 [7] NCCL INFO Connected all trees
+ip-26-0-168-34:272157:272712 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-34:272157:272712 [7] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-168-34:272157:272712 [7] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-168-34:272155:272707 [5] NCCL INFO comm 0xb1c3160 rank 1 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0x218724ca16147a07 - Init COMPLETE
+ip-26-0-165-164:1227463:1228035 [0] NCCL INFO Connected all rings
+ip-26-0-165-164:1227463:1228035 [0] NCCL INFO Connected all trees
+ip-26-0-165-164:1227463:1228035 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-164:1227463:1228035 [0] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-165-164:1227463:1228035 [0] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-168-34:272154:272706 [4] NCCL INFO Connected all rings
+ip-26-0-168-34:272154:272706 [4] NCCL INFO Connected all trees
+ip-26-0-168-34:272154:272706 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-34:272154:272706 [4] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-168-34:272154:272706 [4] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-168-34:272151:272710 [1] NCCL INFO Connected all rings
+ip-26-0-168-34:272151:272710 [1] NCCL INFO Connected all trees
+ip-26-0-168-34:272151:272710 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-34:272151:272710 [1] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-168-34:272151:272710 [1] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-165-164:1227464:1228029 [1] NCCL INFO Connected all rings
+ip-26-0-165-164:1227464:1228029 [1] NCCL INFO Connected all trees
+ip-26-0-165-164:1227464:1228029 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-164:1227470:1228033 [7] NCCL INFO Connected all rings
+ip-26-0-165-164:1227470:1228033 [7] NCCL INFO Connected all trees
+ip-26-0-165-164:1227470:1228033 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-59:1557270:1557825 [4] NCCL INFO Connected all rings
+ip-26-0-165-59:1557270:1557825 [4] NCCL INFO Connected all trees
+ip-26-0-165-164:1227464:1228029 [1] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-165-164:1227464:1228029 [1] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-165-59:1557270:1557825 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-164:1227470:1228033 [7] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-165-164:1227470:1228033 [7] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-165-59:1557270:1557825 [4] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-165-59:1557270:1557825 [4] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-161-123:32711:33274 [7] NCCL INFO Connected all rings
+ip-26-0-161-123:32711:33274 [7] NCCL INFO Connected all trees
+ip-26-0-161-123:32711:33274 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-123:32711:33274 [7] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-161-123:32711:33274 [7] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-161-138:96567:97173 [4] NCCL INFO comm 0xaa843d0 rank 1 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0x174515b09e2bfef0 - Init COMPLETE
+ip-26-0-165-59:1557268:1557822 [2] NCCL INFO comm 0xab5f560 rank 1 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0x3a70f37d11c42078 - Init COMPLETE
+ip-26-0-165-164:1227465:1228028 [2] NCCL INFO Connected all rings
+ip-26-0-165-164:1227465:1228028 [2] NCCL INFO Connected all trees
+ip-26-0-165-164:1227465:1228028 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-164:1227465:1228028 [2] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-165-164:1227465:1228028 [2] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-165-131:1630205:1630805 [0] NCCL INFO comm 0xae8cb50 rank 0 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0xa38d5ad285b7dfb1 - Init COMPLETE
+ip-26-0-168-34:272152:272709 [2] NCCL INFO Connected all rings
+ip-26-0-168-34:272152:272709 [2] NCCL INFO Connected all trees
+ip-26-0-168-34:272152:272709 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-34:272152:272709 [2] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-168-34:272152:272709 [2] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-160-242:1364908:1365495 [6] NCCL INFO comm 0xc5c5120 rank 0 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0x7ba2dc144f01ded2 - Init COMPLETE
+ip-26-0-161-178:2815970:2816556 [7] NCCL INFO Connected all rings
+ip-26-0-161-178:2815970:2816556 [7] NCCL INFO Connected all trees
+ip-26-0-161-178:2815970:2816556 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-178:2815970:2816556 [7] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-161-178:2815970:2816556 [7] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-161-103:96889:97453 [1] NCCL INFO comm 0xa9d5170 rank 0 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0xb2131d0b36358051 - Init COMPLETE
+ip-26-0-165-202:153566:154124 [4] NCCL INFO Connected all rings
+ip-26-0-165-202:153566:154124 [4] NCCL INFO Connected all trees
+ip-26-0-165-202:153566:154124 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-202:153566:154124 [4] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-165-202:153566:154124 [4] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-161-142:62882:63450 [1] NCCL INFO comm 0xb0ebc10 rank 1 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0x5a8efde32afc8c2a - Init COMPLETE
+ip-26-0-160-242:1364902:1365483 [0] NCCL INFO comm 0xb9c9000 rank 0 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0xdc2325cededd95f8 - Init COMPLETE
+ip-26-0-165-131:1630211:1630813 [6] NCCL INFO comm 0xc48d080 rank 0 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0xab67ef3e7d9c3e72 - Init COMPLETE
+ip-26-0-168-34:272150:272713 [0] NCCL INFO comm 0xbd7f290 rank 1 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0x390ca62f48410b65 - Init COMPLETE
+ip-26-0-161-123:32707:33271 [3] NCCL INFO Connected all rings
+ip-26-0-161-123:32707:33271 [3] NCCL INFO Connected all trees
+ip-26-0-161-123:32707:33271 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-123:32707:33271 [3] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-161-123:32707:33271 [3] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-168-52:2722047:2722604 [2] NCCL INFO comm 0xb440480 rank 1 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0x7afd9f7327436e47 - Init COMPLETE
+ip-26-0-160-242:1364903:1365493 [1] NCCL INFO comm 0xbcd3920 rank 0 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0x5a8efde32afc8c2a - Init COMPLETE
+ip-26-0-165-59:1557267:1557821 [1] NCCL INFO Connected all rings
+ip-26-0-161-123:32705:33278 [1] NCCL INFO Connected all rings
+ip-26-0-161-123:32705:33278 [1] NCCL INFO Connected all trees
+ip-26-0-165-59:1557267:1557821 [1] NCCL INFO Connected all trees
+ip-26-0-165-59:1557267:1557821 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-59:1557267:1557821 [1] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-165-59:1557267:1557821 [1] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-161-123:32705:33278 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-123:32705:33278 [1] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-161-123:32705:33278 [1] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-161-78:114516:115080 [4] NCCL INFO Connected all rings
+ip-26-0-161-78:114516:115080 [4] NCCL INFO Connected all trees
+ip-26-0-161-78:114516:115080 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-78:114516:115080 [4] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-161-78:114516:115080 [4] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-161-123:32709:33280 [5] NCCL INFO Connected all rings
+ip-26-0-161-123:32709:33280 [5] NCCL INFO Connected all trees
+ip-26-0-161-123:32709:33280 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-123:32709:33280 [5] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-161-123:32709:33280 [5] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-161-178:2815966:2816554 [3] NCCL INFO Connected all rings
+ip-26-0-161-178:2815966:2816554 [3] NCCL INFO Connected all trees
+ip-26-0-161-178:2815966:2816554 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-178:2815966:2816554 [3] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-161-178:2815966:2816554 [3] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-161-178:2815968:2816559 [5] NCCL INFO Connected all rings
+ip-26-0-161-178:2815968:2816559 [5] NCCL INFO Connected all trees
+ip-26-0-161-178:2815968:2816559 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-178:2815968:2816559 [5] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-161-178:2815968:2816559 [5] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-165-164:1227467:1228021 [4] NCCL INFO comm 0xb21a6e0 rank 0 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0x162146216445f9b - Init COMPLETE
+ip-26-0-165-59:1557269:1557826 [3] NCCL INFO Connected all rings
+ip-26-0-165-59:1557269:1557826 [3] NCCL INFO Connected all trees
+ip-26-0-165-59:1557269:1557826 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-59:1557269:1557826 [3] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-165-59:1557269:1557826 [3] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-161-178:2815964:2816558 [1] NCCL INFO Connected all rings
+ip-26-0-161-178:2815964:2816558 [1] NCCL INFO Connected all trees
+ip-26-0-161-178:2815964:2816558 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-178:2815964:2816558 [1] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-161-178:2815964:2816558 [1] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-165-164:1227463:1228035 [0] NCCL INFO comm 0xa7b8b10 rank 0 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0x390ca62f48410b65 - Init COMPLETE
+ip-26-0-168-34:272157:272712 [7] NCCL INFO comm 0xb41a9a0 rank 1 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0x3b2a736c8d71389 - Init COMPLETE
+ip-26-0-165-59:1557270:1557825 [4] NCCL INFO comm 0xa934a50 rank 1 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0x13fc5839256a0f0a - Init COMPLETE
+ip-26-0-168-34:272154:272706 [4] NCCL INFO comm 0xb298050 rank 1 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0x162146216445f9b - Init COMPLETE
+ip-26-0-168-34:272151:272710 [1] NCCL INFO comm 0xb4b3760 rank 1 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0xbd511d17f9a0d302 - Init COMPLETE
+ip-26-0-165-164:1227470:1228033 [7] NCCL INFO comm 0xb2b5560 rank 0 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0x3b2a736c8d71389 - Init COMPLETE
+ip-26-0-165-164:1227464:1228029 [1] NCCL INFO comm 0xb080790 rank 0 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0xbd511d17f9a0d302 - Init COMPLETE
+ip-26-0-161-123:32711:33274 [7] NCCL INFO comm 0xc082f40 rank 0 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0x6be549eda214aab5 - Init COMPLETE
+ip-26-0-161-78:114515:115081 [3] NCCL INFO Connected all rings
+ip-26-0-161-78:114515:115081 [3] NCCL INFO Connected all trees
+ip-26-0-161-78:114515:115081 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-78:114515:115081 [3] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-161-78:114515:115081 [3] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-161-78:114513:115070 [1] NCCL INFO Connected all rings
+ip-26-0-161-78:114513:115070 [1] NCCL INFO Connected all trees
+ip-26-0-161-78:114513:115070 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-34:272152:272709 [2] NCCL INFO comm 0xc485140 rank 1 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0x6f5a7a271888c3dd - Init COMPLETE
+ip-26-0-161-78:114513:115070 [1] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-161-78:114513:115070 [1] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-165-164:1227465:1228028 [2] NCCL INFO comm 0xac45420 rank 0 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0x6f5a7a271888c3dd - Init COMPLETE
+ip-26-0-161-78:114514:115073 [2] NCCL INFO Connected all rings
+ip-26-0-161-78:114514:115073 [2] NCCL INFO Connected all trees
+ip-26-0-161-78:114514:115073 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-78:114514:115073 [2] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-161-78:114514:115073 [2] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-161-178:2815970:2816556 [7] NCCL INFO comm 0xb2d90c0 rank 1 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0x6be549eda214aab5 - Init COMPLETE
+ip-26-0-161-103:96892:97467 [4] NCCL INFO Connected all rings
+ip-26-0-161-103:96892:97467 [4] NCCL INFO Connected all trees
+ip-26-0-161-103:96892:97467 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-103:96892:97467 [4] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-161-103:96892:97467 [4] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-161-123:32707:33271 [3] NCCL INFO comm 0xaa77e30 rank 0 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0xc2f845d8f7e6ca39 - Init COMPLETE
+ip-26-0-161-78:114516:115080 [4] NCCL INFO comm 0xb3894c0 rank 0 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0x13fc5839256a0f0a - Init COMPLETE
+ip-26-0-161-123:32705:33278 [1] NCCL INFO comm 0xbc02d30 rank 0 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0xfd0f632a334dff00 - Init COMPLETE
+ip-26-0-161-123:32709:33280 [5] NCCL INFO comm 0xc45a330 rank 0 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0xba53db25a1463133 - Init COMPLETE
+ip-26-0-161-178:2815966:2816554 [3] NCCL INFO comm 0xb6b8f90 rank 1 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0xc2f845d8f7e6ca39 - Init COMPLETE
+ip-26-0-161-178:2815968:2816559 [5] NCCL INFO comm 0xc2a34e0 rank 1 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0xba53db25a1463133 - Init COMPLETE
+ip-26-0-165-59:1557269:1557826 [3] NCCL INFO comm 0xbf10700 rank 1 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0xf02067e7b46b58ac - Init COMPLETE
+ip-26-0-165-59:1557267:1557821 [1] NCCL INFO comm 0xa8bb440 rank 1 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0x317366f90460137 - Init COMPLETE
+ip-26-0-161-178:2815964:2816558 [1] NCCL INFO comm 0xa9a5400 rank 1 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0xfd0f632a334dff00 - Init COMPLETE
+ip-26-0-161-153:2636088:2636685 [4] NCCL INFO Connected all rings
+ip-26-0-161-153:2636088:2636685 [4] NCCL INFO Connected all trees
+ip-26-0-161-153:2636088:2636685 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-153:2636088:2636685 [4] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-161-153:2636088:2636685 [4] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-161-78:114515:115081 [3] NCCL INFO comm 0xadec590 rank 0 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0xf02067e7b46b58ac - Init COMPLETE
+ip-26-0-161-78:114513:115070 [1] NCCL INFO comm 0xbe51a40 rank 0 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0x317366f90460137 - Init COMPLETE
+ip-26-0-161-103:96892:97467 [4] NCCL INFO comm 0xbdbb170 rank 0 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0x142d040e61415d48 - Init COMPLETE
+ip-26-0-161-78:114514:115073 [2] NCCL INFO comm 0xbd14d10 rank 0 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0x3a70f37d11c42078 - Init COMPLETE
+ip-26-0-161-103:96888:97459 [0] NCCL INFO Connected all rings
+ip-26-0-161-103:96888:97459 [0] NCCL INFO Connected all trees
+ip-26-0-161-103:96888:97459 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-103:96888:97459 [0] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-161-103:96888:97459 [0] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-168-52:2722046:2722610 [1] NCCL INFO Connected all rings
+ip-26-0-168-52:2722046:2722610 [1] NCCL INFO Connected all trees
+ip-26-0-168-52:2722046:2722610 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-52:2722046:2722610 [1] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-168-52:2722046:2722610 [1] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-161-103:96895:97465 [7] NCCL INFO Connected all rings
+ip-26-0-161-103:96895:97465 [7] NCCL INFO Connected all trees
+ip-26-0-161-103:96895:97465 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-103:96895:97465 [7] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-161-103:96895:97465 [7] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-161-103:96890:97457 [2] NCCL INFO Connected all rings
+ip-26-0-161-103:96890:97457 [2] NCCL INFO Connected all trees
+ip-26-0-161-103:96890:97457 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-103:96890:97457 [2] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-161-103:96890:97457 [2] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-161-153:2636088:2636685 [4] NCCL INFO comm 0xb4c73e0 rank 1 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0x142d040e61415d48 - Init COMPLETE
+ip-26-0-168-52:2722046:2722610 [1] NCCL INFO comm 0xb81eb90 rank 1 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0xd27583e35160699c - Init COMPLETE
+ip-26-0-165-202:153568:154127 [6] NCCL INFO Connected all rings
+ip-26-0-165-202:153568:154127 [6] NCCL INFO Connected all trees
+ip-26-0-165-202:153568:154127 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-202:153568:154127 [6] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-165-202:153568:154127 [6] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-161-153:2636084:2636681 [0] NCCL INFO Connected all rings
+ip-26-0-161-153:2636084:2636681 [0] NCCL INFO Connected all trees
+ip-26-0-161-153:2636084:2636681 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-153:2636084:2636681 [0] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-161-153:2636084:2636681 [0] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-161-103:96888:97459 [0] NCCL INFO comm 0xb03f0a0 rank 0 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0x67b3bc8359a8fd69 - Init COMPLETE
+ip-26-0-161-103:96894:97456 [6] NCCL INFO Connected all rings
+ip-26-0-161-103:96894:97456 [6] NCCL INFO Connected all trees
+ip-26-0-161-103:96894:97456 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-103:96894:97456 [6] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-161-103:96894:97456 [6] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-165-202:153566:154124 [4] NCCL INFO comm 0xaa282c0 rank 0 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0x9924b60493ed5710 - Init COMPLETE
+ip-26-0-161-153:2636086:2636680 [2] NCCL INFO Connected all rings
+ip-26-0-161-153:2636086:2636680 [2] NCCL INFO Connected all trees
+ip-26-0-161-153:2636086:2636680 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-153:2636086:2636680 [2] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-161-153:2636086:2636680 [2] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-168-52:2722052:2722603 [7] NCCL INFO Connected all rings
+ip-26-0-168-52:2722052:2722603 [7] NCCL INFO Connected all trees
+ip-26-0-168-52:2722052:2722603 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-52:2722052:2722603 [7] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-168-52:2722052:2722603 [7] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-168-52:2722048:2722605 [3] NCCL INFO Connected all rings
+ip-26-0-168-52:2722048:2722605 [3] NCCL INFO Connected all trees
+ip-26-0-168-52:2722048:2722605 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-52:2722048:2722605 [3] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-168-52:2722048:2722605 [3] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-161-103:96895:97465 [7] NCCL INFO comm 0xc63f250 rank 0 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0x2eb8a3b534e743f8 - Init COMPLETE
+ip-26-0-168-52:2722050:2722607 [5] NCCL INFO Connected all rings
+ip-26-0-168-52:2722050:2722607 [5] NCCL INFO Connected all trees
+ip-26-0-168-52:2722050:2722607 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-168-52:2722050:2722607 [5] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-168-52:2722050:2722607 [5] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-161-103:96890:97457 [2] NCCL INFO comm 0xb01a390 rank 0 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0x15a67f5c54f9ae39 - Init COMPLETE
+ip-26-0-165-202:153568:154127 [6] NCCL INFO comm 0xa7cc410 rank 0 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0xd2565f661f0f5428 - Init COMPLETE
+ip-26-0-161-153:2636084:2636681 [0] NCCL INFO comm 0xb1631c0 rank 1 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0x67b3bc8359a8fd69 - Init COMPLETE
+ip-26-0-161-153:2636091:2636684 [7] NCCL INFO Connected all rings
+ip-26-0-161-153:2636091:2636684 [7] NCCL INFO Connected all trees
+ip-26-0-161-153:2636091:2636684 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-153:2636091:2636684 [7] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-161-153:2636091:2636684 [7] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-161-103:96894:97456 [6] NCCL INFO comm 0xac95050 rank 0 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0x8582362b5b377cb8 - Init COMPLETE
+ip-26-0-165-202:153562:154129 [0] NCCL INFO Connected all rings
+ip-26-0-165-202:153562:154129 [0] NCCL INFO Connected all trees
+ip-26-0-165-202:153562:154129 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-202:153562:154129 [0] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-165-202:153562:154129 [0] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-168-52:2722052:2722603 [7] NCCL INFO comm 0xb691400 rank 1 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0x8ccc8b29b5112548 - Init COMPLETE
+ip-26-0-168-52:2722048:2722605 [3] NCCL INFO comm 0xa66a8f0 rank 1 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0x6290c10a08bf7fdd - Init COMPLETE
+ip-26-0-161-153:2636086:2636680 [2] NCCL INFO comm 0xbb75650 rank 1 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0x15a67f5c54f9ae39 - Init COMPLETE
+ip-26-0-168-52:2722050:2722607 [5] NCCL INFO comm 0xa69e960 rank 1 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0xa28c5a970ab72e03 - Init COMPLETE
+ip-26-0-161-153:2636090:2636679 [6] NCCL INFO Connected all rings
+ip-26-0-161-153:2636090:2636679 [6] NCCL INFO Connected all trees
+ip-26-0-161-153:2636090:2636679 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-161-153:2636090:2636679 [6] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-161-153:2636090:2636679 [6] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-161-153:2636091:2636684 [7] NCCL INFO comm 0xb227b50 rank 1 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0x2eb8a3b534e743f8 - Init COMPLETE
+ip-26-0-165-202:153562:154129 [0] NCCL INFO comm 0xb4ad5a0 rank 0 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0x7a9a6457da866605 - Init COMPLETE
+ip-26-0-161-153:2636090:2636679 [6] NCCL INFO comm 0xbc324a0 rank 1 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0x8582362b5b377cb8 - Init COMPLETE
+ip-26-0-165-202:153564:154120 [2] NCCL INFO Connected all rings
+ip-26-0-165-202:153564:154120 [2] NCCL INFO Connected all trees
+ip-26-0-165-202:153564:154120 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-202:153564:154120 [2] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-165-202:153564:154120 [2] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-165-202:153564:154120 [2] NCCL INFO comm 0xa7dd790 rank 0 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0x7afd9f7327436e47 - Init COMPLETE
+ip-26-0-165-202:153563:154131 [1] NCCL INFO Connected all rings
+ip-26-0-165-202:153563:154131 [1] NCCL INFO Connected all trees
+ip-26-0-165-202:153563:154131 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-202:153563:154131 [1] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-165-202:153563:154131 [1] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-165-202:153565:154121 [3] NCCL INFO Connected all rings
+ip-26-0-165-202:153565:154121 [3] NCCL INFO Connected all trees
+ip-26-0-165-202:153565:154121 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-202:153565:154121 [3] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-165-202:153565:154121 [3] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-165-202:153567:154126 [5] NCCL INFO Connected all rings
+ip-26-0-165-202:153567:154126 [5] NCCL INFO Connected all trees
+ip-26-0-165-202:153567:154126 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-202:153567:154126 [5] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-165-202:153567:154126 [5] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-165-202:153569:154118 [7] NCCL INFO Connected all rings
+ip-26-0-165-202:153569:154118 [7] NCCL INFO Connected all trees
+ip-26-0-165-202:153569:154118 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-202:153569:154118 [7] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-165-202:153569:154118 [7] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-165-202:153563:154131 [1] NCCL INFO comm 0xc3e7cc0 rank 0 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0xd27583e35160699c - Init COMPLETE
+ip-26-0-165-202:153565:154121 [3] NCCL INFO comm 0xb9b5da0 rank 0 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0x6290c10a08bf7fdd - Init COMPLETE
+ip-26-0-165-202:153567:154126 [5] NCCL INFO comm 0xc1e41e0 rank 0 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0xa28c5a970ab72e03 - Init COMPLETE
+ip-26-0-165-202:153569:154118 [7] NCCL INFO comm 0xa7463a0 rank 0 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0x8ccc8b29b5112548 - Init COMPLETE
+12/28/2024 11:56:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [Optimizer Building] Using LearningRateForSP as learning rate
+12/28/2024 11:56:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [ZeRO sharding] Size of optimizer params per rank:
+12/28/2024 11:56:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [ZeRO sharding] DP Rank 0 has 3.73G out of 7.45G (50.00%) params' optimizer states
+12/28/2024 11:56:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [ZeRO sharding] DP Rank 1 has 3.73G out of 7.45G (50.00%) params' optimizer states
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.token_position_embeddings | PP: 0/2 | Block rank: 0
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.0 | PP: 0/2 | Block rank: 1
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.1 | PP: 0/2 | Block rank: 2
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.2 | PP: 0/2 | Block rank: 3
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.3 | PP: 0/2 | Block rank: 4
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.4 | PP: 0/2 | Block rank: 5
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.5 | PP: 0/2 | Block rank: 6
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.6 | PP: 0/2 | Block rank: 7
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.7 | PP: 0/2 | Block rank: 8
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.8 | PP: 0/2 | Block rank: 9
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.9 | PP: 0/2 | Block rank: 10
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.10 | PP: 0/2 | Block rank: 11
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.11 | PP: 0/2 | Block rank: 12
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.12 | PP: 0/2 | Block rank: 13
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.13 | PP: 0/2 | Block rank: 14
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.14 | PP: 0/2 | Block rank: 15
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.15 | PP: 0/2 | Block rank: 16
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.16 | PP: 0/2 | Block rank: 17
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.17 | PP: 0/2 | Block rank: 18
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.18 | PP: 0/2 | Block rank: 19
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.19 | PP: 0/2 | Block rank: 20
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.20 | PP: 0/2 | Block rank: 21
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.21 | PP: 0/2 | Block rank: 22
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.22 | PP: 0/2 | Block rank: 23
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.23 | PP: 0/2 | Block rank: 24
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.24 | PP: 0/2 | Block rank: 25
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.25 | PP: 0/2 | Block rank: 26
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.26 | PP: 0/2 | Block rank: 27
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.27 | PP: 0/2 | Block rank: 28
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.28 | PP: 0/2 | Block rank: 29
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.29 | PP: 0/2 | Block rank: 30
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.30 | PP: 0/2 | Block rank: 31
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.31 | PP: 0/2 | Block rank: 32
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.32 | PP: 0/2 | Block rank: 33
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.33 | PP: 0/2 | Block rank: 34
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.34 | PP: 0/2 | Block rank: 35
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.35 | PP: 0/2 | Block rank: 36
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.36 | PP: 0/2 | Block rank: 37
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.37 | PP: 0/2 | Block rank: 38
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.38 | PP: 0/2 | Block rank: 39
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.39 | PP: 0/2 | Block rank: 40
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.40 | PP: 0/2 | Block rank: 41
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.41 | PP: 0/2 | Block rank: 42
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.42 | PP: 0/2 | Block rank: 43
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.43 | PP: 0/2 | Block rank: 44
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.44 | PP: 0/2 | Block rank: 45
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.45 | PP: 0/2 | Block rank: 46
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.46 | PP: 0/2 | Block rank: 47
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.47 | PP: 0/2 | Block rank: 48
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.48 | PP: 0/2 | Block rank: 49
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.49 | PP: 0/2 | Block rank: 50
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.50 | PP: 0/2 | Block rank: 51
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.51 | PP: 0/2 | Block rank: 52
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.52 | PP: 0/2 | Block rank: 53
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.53 | PP: 0/2 | Block rank: 54
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.54 | PP: 0/2 | Block rank: 55
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.55 | PP: 0/2 | Block rank: 56
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.56 | PP: 0/2 | Block rank: 57
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.57 | PP: 0/2 | Block rank: 58
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.58 | PP: 0/2 | Block rank: 59
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.59 | PP: 0/2 | Block rank: 60
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.60 | PP: 0/2 | Block rank: 61
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.61 | PP: 0/2 | Block rank: 62
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.62 | PP: 0/2 | Block rank: 63
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.63 | PP: 0/2 | Block rank: 64
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.64 | PP: 1/2 | Block rank: 0
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.65 | PP: 1/2 | Block rank: 1
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.66 | PP: 1/2 | Block rank: 2
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.67 | PP: 1/2 | Block rank: 3
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.68 | PP: 1/2 | Block rank: 4
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.69 | PP: 1/2 | Block rank: 5
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.70 | PP: 1/2 | Block rank: 6
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.71 | PP: 1/2 | Block rank: 7
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.72 | PP: 1/2 | Block rank: 8
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.73 | PP: 1/2 | Block rank: 9
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.74 | PP: 1/2 | Block rank: 10
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.75 | PP: 1/2 | Block rank: 11
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.76 | PP: 1/2 | Block rank: 12
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.77 | PP: 1/2 | Block rank: 13
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.78 | PP: 1/2 | Block rank: 14
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.79 | PP: 1/2 | Block rank: 15
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.80 | PP: 1/2 | Block rank: 16
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.81 | PP: 1/2 | Block rank: 17
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.82 | PP: 1/2 | Block rank: 18
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.83 | PP: 1/2 | Block rank: 19
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.84 | PP: 1/2 | Block rank: 20
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.85 | PP: 1/2 | Block rank: 21
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.86 | PP: 1/2 | Block rank: 22
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.87 | PP: 1/2 | Block rank: 23
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.88 | PP: 1/2 | Block rank: 24
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.89 | PP: 1/2 | Block rank: 25
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.90 | PP: 1/2 | Block rank: 26
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.91 | PP: 1/2 | Block rank: 27
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.92 | PP: 1/2 | Block rank: 28
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.93 | PP: 1/2 | Block rank: 29
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.94 | PP: 1/2 | Block rank: 30
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.95 | PP: 1/2 | Block rank: 31
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.96 | PP: 1/2 | Block rank: 32
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.97 | PP: 1/2 | Block rank: 33
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.98 | PP: 1/2 | Block rank: 34
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.99 | PP: 1/2 | Block rank: 35
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.100 | PP: 1/2 | Block rank: 36
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.101 | PP: 1/2 | Block rank: 37
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.102 | PP: 1/2 | Block rank: 38
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.103 | PP: 1/2 | Block rank: 39
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.104 | PP: 1/2 | Block rank: 40
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.105 | PP: 1/2 | Block rank: 41
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.106 | PP: 1/2 | Block rank: 42
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.107 | PP: 1/2 | Block rank: 43
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.108 | PP: 1/2 | Block rank: 44
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.109 | PP: 1/2 | Block rank: 45
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.110 | PP: 1/2 | Block rank: 46
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.111 | PP: 1/2 | Block rank: 47
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.112 | PP: 1/2 | Block rank: 48
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.113 | PP: 1/2 | Block rank: 49
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.114 | PP: 1/2 | Block rank: 50
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.115 | PP: 1/2 | Block rank: 51
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.116 | PP: 1/2 | Block rank: 52
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.117 | PP: 1/2 | Block rank: 53
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.118 | PP: 1/2 | Block rank: 54
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.119 | PP: 1/2 | Block rank: 55
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.120 | PP: 1/2 | Block rank: 56
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.121 | PP: 1/2 | Block rank: 57
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.122 | PP: 1/2 | Block rank: 58
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.123 | PP: 1/2 | Block rank: 59
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.124 | PP: 1/2 | Block rank: 60
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.125 | PP: 1/2 | Block rank: 61
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.final_layer_norm | PP: 1/2 | Block rank: 62
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.lm_head | PP: 1/2 | Block rank: 63
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.cast_to_fp32 | PP: 1/2 | Block rank: 64
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: loss | PP: 1/2 | Block rank: 65
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [Training Plan] Stage Stable Training Stage has 99 remaining training steps and has consumed 0 samples
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: Using dummy data generator
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [Training Plan] There are 1 training stages 
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [Stage Stable Training Stage] start from step 1 
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: 
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [Start training] datetime: 2024-12-28 11:56:08.333487 | mbs: 64 | grad_accum: 2 | global_batch_size: 256 | sequence_length: 4096 | train_steps: 100 | start_iteration_step: 0 | consumed_train_samples: 0
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: Resuming training from stage Stable Training Stage, it has trained for 0 samples and has 99 remaining train steps
+12/28/2024 11:56:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:  Memory usage: 56848.07MiB. Peak allocated 56848.07MiB. Peak reserved: 56950.00MiB
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+Traceback (most recent call last):
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+        trainer.train(dataloader)trainer.train(dataloader)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    Traceback (most recent call last):
+trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    trainer.train(dataloader)    
+return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+        return f(*args, **kwargs)outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)    
+outputs = self.pipeline_engine.train_batch_iter(  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)    
+output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+        return forward_call(*args, **kwargs)outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+        return forward_call(*args, **kwargs)return forward_call(*args, **kwargs)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    sharded_logits = self.model(
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 896, in forward_with_hidden_states
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 896, in forward_with_hidden_states
+return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]    
+return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 896, in forward_with_hidden_states
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 896, in forward_with_hidden_states
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]Traceback (most recent call last):
+    
+sharded_logits = self.model(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 896, in forward_with_hidden_states
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 896, in forward_with_hidden_states
+    output = self.token_position_embeddings(input_ids=input_ids, input_mask=input_mask)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.token_position_embeddings(input_ids=input_ids, input_mask=input_mask)    
+    output = self.token_position_embeddings(input_ids=input_ids, input_mask=input_mask)    output = self.token_position_embeddings(input_ids=input_ids, input_mask=input_mask)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+
+
+output = self.token_position_embeddings(input_ids=input_ids, input_mask=input_mask)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.token_position_embeddings(input_ids=input_ids, input_mask=input_mask)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 160, in forward
+            return self._call_impl(*args, **kwargs)return forward_call(*args, **kwargs)return forward_call(*args, **kwargs)
+
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 160, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 160, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 160, in forward
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 160, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 160, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 896, in forward_with_hidden_states
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+    output = self.pp_block(**new_kwargs)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+        output = self.token_position_embeddings(input_ids=input_ids, input_mask=input_mask)output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+        output = self.pp_block(**new_kwargs)return self._call_impl(*args, **kwargs)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+return forward_call(*args, **kwargs)    
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 786, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 786, in forward
+        return forward_call(*args, **kwargs)return forward_call(*args, **kwargs)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 786, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 786, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 786, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 786, in forward
+    input_embeds = self.token_embedding(input_ids)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 160, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    input_embeds = self.token_embedding(input_ids)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    input_embeds = self.token_embedding(input_ids)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    input_embeds = self.token_embedding(input_ids)    
+input_embeds = self.token_embedding(input_ids)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    input_embeds = self.token_embedding(input_ids)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+                return self._call_impl(*args, **kwargs)return self._call_impl(*args, **kwargs)
+return self._call_impl(*args, **kwargs)return forward_call(*args, **kwargs)
+
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 290, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 786, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+        return forward_call(*args, **kwargs)return forward_call(*args, **kwargs)    
+
+return forward_call(*args, **kwargs)  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 290, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 290, in forward
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 290, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 290, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 290, in forward
+    out = out * (~input_mask[..., None])
+torch.cuda    .input_embeds = self.token_embedding(input_ids)OutOfMemoryError
+: CUDA out of memory. Tried to allocate 8.00 GiB. GPU 3 has a total capacty of 79.33 GiB of which 3.81 GiB is free. Including non-PyTorch memory, this process has 75.51 GiB memory in use. Of the allocated memory 63.52 GiB is allocated by PyTorch, and 89.18 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    out = out * (~input_mask[..., None])
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+torch.cuda  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 896, in forward_with_hidden_states
+.OutOfMemoryError:     CUDA out of memory. Tried to allocate 8.00 GiB. GPU 5 has a total capacty of 79.33 GiB of which 3.99 GiB is free. Including non-PyTorch memory, this process has 75.33 GiB memory in use. Of the allocated memory 63.52 GiB is allocated by PyTorch, and 89.18 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONFout = out * (~input_mask[..., None])
+
+    out = out * (~input_mask[..., None])
+torch.cuda.OutOfMemoryError: torch.cudaCUDA out of memory. Tried to allocate 8.00 GiB. GPU 6 has a total capacty of 79.33 GiB of which 4.11 GiB is free. Including non-PyTorch memory, this process has 75.21 GiB memory in use. Of the allocated memory 63.52 GiB is allocated by PyTorch, and 89.18 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF.OutOfMemoryError
+: CUDA out of memory. Tried to allocate 8.00 GiB. GPU 2 has a total capacty of 79.33 GiB of which 4.11 GiB is free. Including non-PyTorch memory, this process has 75.21 GiB memory in use. Of the allocated memory 63.52 GiB is allocated by PyTorch, and 89.18 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    out = out * (~input_mask[..., None])
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 8.00 GiB. GPU 4 has a total capacty of 79.33 GiB of which 4.08 GiB is free. Including non-PyTorch memory, this process has 75.24 GiB memory in use. Of the allocated memory 63.52 GiB is allocated by PyTorch, and 89.18 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 290, in forward
+    out = out * (~input_mask[..., None])
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 8.00 GiB. GPU 7 has a total capacty of 79.33 GiB of which 4.16 GiB is free. Including non-PyTorch memory, this process has 75.15 GiB memory in use. Of the allocated memory 63.52 GiB is allocated by PyTorch, and 89.18 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    output = self.token_position_embeddings(input_ids=input_ids, input_mask=input_mask)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    out = out * (~input_mask[..., None])
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 8.00 GiB. GPU 0 has a total capacty of 79.33 GiB of which 4.13 GiB is free. Including non-PyTorch memory, this process has 75.19 GiB memory in use. Of the allocated memory 63.52 GiB is allocated by PyTorch, and 89.18 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 160, in forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 786, in forward
+    input_embeds = self.token_embedding(input_ids)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 290, in forward
+    out = out * (~input_mask[..., None])
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 8.00 GiB. GPU 1 has a total capacty of 79.33 GiB of which 3.99 GiB is free. Including non-PyTorch memory, this process has 75.33 GiB memory in use. Of the allocated memory 63.52 GiB is allocated by PyTorch, and 89.18 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 896, in forward_with_hidden_states
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 896, in forward_with_hidden_states
+    output = self.token_position_embeddings(input_ids=input_ids, input_mask=input_mask)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.token_position_embeddings(input_ids=input_ids, input_mask=input_mask)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 160, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 160, in forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 786, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 786, in forward
+    input_embeds = self.token_embedding(input_ids)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    input_embeds = self.token_embedding(input_ids)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 290, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 290, in forward
+    out = out * (~input_mask[..., None])
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 8.00 GiB. GPU 7 has a total capacty of 79.33 GiB of which 5.15 GiB is free. Including non-PyTorch memory, this process has 74.17 GiB memory in use. Of the allocated memory 63.52 GiB is allocated by PyTorch, and 97.18 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    out = out * (~input_mask[..., None])
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 8.00 GiB. GPU 5 has a total capacty of 79.33 GiB of which 4.98 GiB is free. Including non-PyTorch memory, this process has 74.34 GiB memory in use. Of the allocated memory 63.52 GiB is allocated by PyTorch, and 97.18 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 896, in forward_with_hidden_states
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 896, in forward_with_hidden_states
+output = self.token_position_embeddings(input_ids=input_ids, input_mask=input_mask)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 160, in forward
+    output = self.token_position_embeddings(input_ids=input_ids, input_mask=input_mask)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 160, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 896, in forward_with_hidden_states
+    return self._call_impl(*args, **kwargs)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    output = self.token_position_embeddings(input_ids=input_ids, input_mask=input_mask)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 786, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 786, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 160, in forward
+    input_embeds = self.token_embedding(input_ids)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    input_embeds = self.token_embedding(input_ids)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 290, in forward
+        output = self.pp_block(**new_kwargs)return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]    
+
+output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 896, in forward_with_hidden_states
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 290, in forward
+        trainer.train(dataloader)return self._call_impl(*args, **kwargs)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    out = out * (~input_mask[..., None])
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 8.00 GiB. GPU 4 has a total capacty of 79.33 GiB of which 4.95 GiB is free. Including non-PyTorch memory, this process has 74.37 GiB memory in use. Of the allocated memory 63.52 GiB is allocated by PyTorch, and 97.18 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.token_position_embeddings(input_ids=input_ids, input_mask=input_mask)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    out = out * (~input_mask[..., None])
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 8.00 GiB. GPU 1 has a total capacty of 79.33 GiB of which 4.98 GiB is free. Including non-PyTorch memory, this process has 74.34 GiB memory in use. Of the allocated memory 63.52 GiB is allocated by PyTorch, and 97.18 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 786, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+        return forward_call(*args, **kwargs)input_embeds = self.token_embedding(input_ids)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 160, in forward
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 290, in forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    out = out * (~input_mask[..., None])
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 8.00 GiB. GPU 2 has a total capacty of 79.33 GiB of which 4.98 GiB is free. Including non-PyTorch memory, this process has 74.34 GiB memory in use. Of the allocated memory 63.52 GiB is allocated by PyTorch, and 97.18 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 786, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 896, in forward_with_hidden_states
+    input_embeds = self.token_embedding(input_ids)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = self.token_position_embeddings(input_ids=input_ids, input_mask=input_mask)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 290, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    out = out * (~input_mask[..., None])
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 8.00 GiB. GPU 3 has a total capacty of 79.33 GiB of which 4.80 GiB is free. Including non-PyTorch memory, this process has 74.52 GiB memory in use. Of the allocated memory 63.52 GiB is allocated by PyTorch, and 97.18 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF    
+sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 160, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 896, in forward_with_hidden_states
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 786, in forward
+    output = self.token_position_embeddings(input_ids=input_ids, input_mask=input_mask)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    input_embeds = self.token_embedding(input_ids)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 160, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 290, in forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    out = out * (~input_mask[..., None])
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 8.00 GiB. GPU 6 has a total capacty of 79.33 GiB of which 4.98 GiB is free. Including non-PyTorch memory, this process has 74.34 GiB memory in use. Of the allocated memory 63.52 GiB is allocated by PyTorch, and 97.18 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 786, in forward
+    input_embeds = self.token_embedding(input_ids)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 290, in forward
+    out = out * (~input_mask[..., None])
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 8.00 GiB. GPU 0 has a total capacty of 79.33 GiB of which 5.00 GiB is free. Including non-PyTorch memory, this process has 74.32 GiB memory in use. Of the allocated memory 63.52 GiB is allocated by PyTorch, and 97.18 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    trainer.train(dataloader)
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    trainer.train(dataloader)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return f(*args, **kwargs)
+    return f(*args, **kwargs)  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    outputs = self.pipeline_engine.train_batch_iter(  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs = self.pipeline_engine.train_batch_iter(    
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)    
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+          File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+return forward_call(*args, **kwargs)return self._call_impl(*args, **kwargs)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 896, in forward_with_hidden_states
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 896, in forward_with_hidden_states
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+            sharded_logits = self.model(sharded_logits = self.model(sharded_logits = self.model(
+
+
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+output = self.token_position_embeddings(input_ids=input_ids, input_mask=input_mask)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.token_position_embeddings(input_ids=input_ids, input_mask=input_mask)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    return self._call_impl(*args, **kwargs)    return self._call_impl(*args, **kwargs)
+return self._call_impl(*args, **kwargs)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 160, in forward
+        return forward_call(*args, **kwargs)return forward_call(*args, **kwargs)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 160, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 896, in forward_with_hidden_states
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+        return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 896, in forward_with_hidden_states
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 896, in forward_with_hidden_states
+    output = self.token_position_embeddings(input_ids=input_ids, input_mask=input_mask)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = self.token_position_embeddings(input_ids=input_ids, input_mask=input_mask)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+        return self._call_impl(*args, **kwargs)return self._call_impl(*args, **kwargs)
+
+    return self._call_impl(*args, **kwargs)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 160, in forward
+    return forward_call(*args, **kwargs)
+    return forward_call(*args, **kwargs)  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 786, in forward
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 786, in forward
+    output = self.token_position_embeddings(input_ids=input_ids, input_mask=input_mask)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 160, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 896, in forward_with_hidden_states
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    input_embeds = self.token_embedding(input_ids)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    input_embeds = self.token_embedding(input_ids)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.token_position_embeddings(input_ids=input_ids, input_mask=input_mask)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+        return self._call_impl(*args, **kwargs)return self._call_impl(*args, **kwargs)
+
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 160, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)return forward_call(*args, **kwargs)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 786, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 290, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 290, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 786, in forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 160, in forward
+    out = out * (~input_mask[..., None])
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 8.00 GiB. GPU 5 has a total capacty of 79.33 GiB of which 4.10 GiB is free. Including non-PyTorch memory, this process has 75.22 GiB memory in use. Of the allocated memory 63.52 GiB is allocated by PyTorch, and 97.18 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    input_embeds = self.token_embedding(input_ids)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    out = out * (~input_mask[..., None])
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 8.00 GiB. GPU 6 has a total capacty of 79.33 GiB of which 3.98 GiB is free. Including non-PyTorch memory, this process has 75.34 GiB memory in use. Of the allocated memory 63.52 GiB is allocated by PyTorch, and 97.18 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    input_embeds = self.token_embedding(input_ids)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+        return self._call_impl(*args, **kwargs)return self._call_impl(*args, **kwargs)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 786, in forward
+        return forward_call(*args, **kwargs)return forward_call(*args, **kwargs)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 290, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 290, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 786, in forward
+    input_embeds = self.token_embedding(input_ids)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    out = out * (~input_mask[..., None])
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 8.00 GiB. GPU 0 has a total capacty of 79.33 GiB of which 4.00 GiB is free. Including non-PyTorch memory, this process has 75.31 GiB memory in use. Of the allocated memory 63.52 GiB is allocated by PyTorch, and 97.18 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    input_embeds = self.token_embedding(input_ids)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    out = out * (~input_mask[..., None])
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 8.00 GiB. GPU 1 has a total capacty of 79.33 GiB of which 4.10 GiB is free. Including non-PyTorch memory, this process has 75.22 GiB memory in use. Of the allocated memory 63.52 GiB is allocated by PyTorch, and 97.18 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 290, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 290, in forward
+    out = out * (~input_mask[..., None])
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 8.00 GiB. GPU 7 has a total capacty of 79.33 GiB of which 4.27 GiB is free. Including non-PyTorch memory, this process has 75.04 GiB memory in use. Of the allocated memory 63.52 GiB is allocated by PyTorch, and 97.18 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    out = out * (~input_mask[..., None])
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 8.00 GiB. GPU 2 has a total capacty of 79.33 GiB of which 3.98 GiB is free. Including non-PyTorch memory, this process has 75.34 GiB memory in use. Of the allocated memory 63.52 GiB is allocated by PyTorch, and 97.18 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    trainer.train(dataloader)    
+trainer.train(dataloader)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+trainer.train(dataloader)
+    trainer.train(dataloader)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+
+    trainer.train(dataloader)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return self._call_impl(*args, **kwargs)
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return f(*args, **kwargs)
+    return f(*args, **kwargs)  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+return f(*args, **kwargs)
+    outputs = self.pipeline_engine.train_batch_iter(  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 896, in forward_with_hidden_states
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    output = self.token_position_embeddings(input_ids=input_ids, input_mask=input_mask)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+        outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)    
+outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+    outputs = self.pipeline_engine.train_batch_iter(    
+outputs = self.pipeline_engine.train_batch_iter(
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 160, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+        output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)output = model(**micro_batch)    
+
+output = model(**micro_batch)  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 786, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+        return forward_call(*args, **kwargs)return forward_call(*args, **kwargs)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    input_embeds = self.token_embedding(input_ids)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    sharded_logits = self.model(
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 896, in forward_with_hidden_states
+sharded_logits = self.model(
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 290, in forward
+    output = self.token_position_embeddings(input_ids=input_ids, input_mask=input_mask)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    out = out * (~input_mask[..., None])
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 8.00 GiB. GPU 3 has a total capacty of 79.33 GiB of which 3.92 GiB is free. Including non-PyTorch memory, this process has 75.40 GiB memory in use. Of the allocated memory 63.52 GiB is allocated by PyTorch, and 97.18 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+        return self._call_impl(*args, **kwargs)return self._call_impl(*args, **kwargs)
+    
+return self._call_impl(*args, **kwargs)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 160, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 896, in forward_with_hidden_states
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 896, in forward_with_hidden_states
+return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 896, in forward_with_hidden_states
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 896, in forward_with_hidden_states
+return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 896, in forward_with_hidden_states
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 896, in forward_with_hidden_states
+    output = self.token_position_embeddings(input_ids=input_ids, input_mask=input_mask)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.token_position_embeddings(input_ids=input_ids, input_mask=input_mask)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.token_position_embeddings(input_ids=input_ids, input_mask=input_mask)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.token_position_embeddings(input_ids=input_ids, input_mask=input_mask)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.token_position_embeddings(input_ids=input_ids, input_mask=input_mask)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 786, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+        return self._call_impl(*args, **kwargs)return self._call_impl(*args, **kwargs)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = self.token_position_embeddings(input_ids=input_ids, input_mask=input_mask)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 160, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 160, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 160, in forward
+        return forward_call(*args, **kwargs)return forward_call(*args, **kwargs)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 160, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 160, in forward
+    input_embeds = self.token_embedding(input_ids)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 160, in forward
+    output = self.pp_block(**new_kwargs)
+    output = self.pp_block(**new_kwargs)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 290, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+    output = self.pp_block(**new_kwargs)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 786, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 786, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 786, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 786, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    out = out * (~input_mask[..., None])
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 8.00 GiB. GPU 2 has a total capacty of 79.33 GiB of which 4.10 GiB is free. Including non-PyTorch memory, this process has 75.22 GiB memory in use. Of the allocated memory 63.52 GiB is allocated by PyTorch, and 97.18 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 786, in forward
+    input_embeds = self.token_embedding(input_ids)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 786, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    input_embeds = self.token_embedding(input_ids)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    input_embeds = self.token_embedding(input_ids)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+        input_embeds = self.token_embedding(input_ids)input_embeds = self.token_embedding(input_ids)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    input_embeds = self.token_embedding(input_ids)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 290, in forward
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+        return self._call_impl(*args, **kwargs)return self._call_impl(*args, **kwargs)
+    
+return self._call_impl(*args, **kwargs)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    out = out * (~input_mask[..., None])
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 8.00 GiB. GPU 1 has a total capacty of 79.33 GiB of which 3.98 GiB is free. Including non-PyTorch memory, this process has 75.34 GiB memory in use. Of the allocated memory 63.52 GiB is allocated by PyTorch, and 97.18 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return self._call_impl(*args, **kwargs)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 290, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 290, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 290, in forward
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 290, in forward
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 290, in forward
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    out = out * (~input_mask[..., None])
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 8.00 GiB. GPU 4 has a total capacty of 79.33 GiB of which 4.07 GiB is free. Including non-PyTorch memory, this process has 75.24 GiB memory in use. Of the allocated memory 63.52 GiB is allocated by PyTorch, and 97.18 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    out = out * (~input_mask[..., None])
+    out = out * (~input_mask[..., None])torch.cuda
+.OutOfMemoryError: CUDA out of memory. Tried to allocate 8.00 GiB. GPU 7 has a total capacty of 79.33 GiB of which 4.16 GiB is free. Including non-PyTorch memory, this process has 75.16 GiB memory in use. Of the allocated memory 63.52 GiB is allocated by PyTorch, and 97.18 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONFtorch.cuda
+.OutOfMemoryError: CUDA out of memory. Tried to allocate 8.00 GiB. GPU 0 has a total capacty of 79.33 GiB of which 4.12 GiB is free. Including non-PyTorch memory, this process has 75.20 GiB memory in use. Of the allocated memory 63.52 GiB is allocated by PyTorch, and 97.18 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    out = out * (~input_mask[..., None])
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 8.00 GiB. GPU 6 has a total capacty of 79.33 GiB of which 4.10 GiB is free. Including non-PyTorch memory, this process has 75.22 GiB memory in use. Of the allocated memory 63.52 GiB is allocated by PyTorch, and 97.18 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    out = out * (~input_mask[..., None])
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 8.00 GiB. GPU 5 has a total capacty of 79.33 GiB of which 3.98 GiB is free. Including non-PyTorch memory, this process has 75.34 GiB memory in use. Of the allocated memory 63.52 GiB is allocated by PyTorch, and 97.18 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs = self.pipeline_engine.train_batch_iter(
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+    trainer.train(dataloader)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+trainer.train(dataloader)    
+trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+        trainer.train(dataloader)outputs = self.pipeline_engine.train_batch_iter(
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+        trainer.train(dataloader)return f(*args, **kwargs)
+
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    trainer.train(dataloader)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return f(*args, **kwargs)
+    return f(*args, **kwargs)  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+    output = model(**micro_batch)
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return f(*args, **kwargs)
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+        outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+        return forward_call(*args, **kwargs)return forward_call(*args, **kwargs)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+        outputs = self.pipeline_engine.train_batch_iter(outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    return forward_call(*args, **kwargs)
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+        sharded_logits = self.model(output = model(**micro_batch)
+    
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    sharded_logits = self.model(
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+        return forward_call(*args, **kwargs)    return self._call_impl(*args, **kwargs)
+return forward_call(*args, **kwargs)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+return forward_call(*args, **kwargs)  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    return self._call_impl(*args, **kwargs)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 896, in forward_with_hidden_states
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 896, in forward_with_hidden_states
+    output = self.token_position_embeddings(input_ids=input_ids, input_mask=input_mask)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 896, in forward_with_hidden_states
+sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 896, in forward_with_hidden_states
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 896, in forward_with_hidden_states
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.token_position_embeddings(input_ids=input_ids, input_mask=input_mask)    
+output = self.token_position_embeddings(input_ids=input_ids, input_mask=input_mask)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.token_position_embeddings(input_ids=input_ids, input_mask=input_mask)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.token_position_embeddings(input_ids=input_ids, input_mask=input_mask)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 160, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+        return self._call_impl(*args, **kwargs)return self._call_impl(*args, **kwargs)
+    
+return self._call_impl(*args, **kwargs)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+        return forward_call(*args, **kwargs)return forward_call(*args, **kwargs)
+    
+return forward_call(*args, **kwargs)  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 160, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 160, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 160, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 160, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 896, in forward_with_hidden_states
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 896, in forward_with_hidden_states
+    output = self.pp_block(**new_kwargs)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+output = self.pp_block(**new_kwargs)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 896, in forward_with_hidden_states
+return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 896, in forward_with_hidden_states
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 786, in forward
+        return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]output = self.token_position_embeddings(input_ids=input_ids, input_mask=input_mask)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 896, in forward_with_hidden_states
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.token_position_embeddings(input_ids=input_ids, input_mask=input_mask)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.token_position_embeddings(input_ids=input_ids, input_mask=input_mask)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.token_position_embeddings(input_ids=input_ids, input_mask=input_mask)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.token_position_embeddings(input_ids=input_ids, input_mask=input_mask)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    input_embeds = self.token_embedding(input_ids)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+          File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+return self._call_impl(*args, **kwargs)    return self._call_impl(*args, **kwargs)
+    
+return self._call_impl(*args, **kwargs)return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 786, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 160, in forward
+    return forward_call(*args, **kwargs)    
+return forward_call(*args, **kwargs)      File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 160, in forward
+
+return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 786, in forward
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 786, in forward
+return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 786, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 160, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = self.pp_block(**new_kwargs)
+    input_embeds = self.token_embedding(input_ids)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 160, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 160, in forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 290, in forward
+        input_embeds = self.token_embedding(input_ids)    input_embeds = self.token_embedding(input_ids)
+output = self.pp_block(**new_kwargs)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = self.pp_block(**new_kwargs)
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    input_embeds = self.token_embedding(input_ids)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+return self._call_impl(*args, **kwargs)    return self._call_impl(*args, **kwargs)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+return self._call_impl(*args, **kwargs)    
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 786, in forward
+    return self._call_impl(*args, **kwargs)
+    out = out * (~input_mask[..., None])
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 8.00 GiB. GPU 4 has a total capacty of 79.33 GiB of which 3.96 GiB is free. Including non-PyTorch memory, this process has 75.36 GiB memory in use. Of the allocated memory 63.52 GiB is allocated by PyTorch, and 97.18 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 786, in forward
+    return forward_call(*args, **kwargs)
+    return forward_call(*args, **kwargs)  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 290, in forward
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 290, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 786, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 290, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 786, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 290, in forward
+    input_embeds = self.token_embedding(input_ids)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    input_embeds = self.token_embedding(input_ids)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    out = out * (~input_mask[..., None])
+torch.cuda    .out = out * (~input_mask[..., None])OutOfMemoryError
+: CUDA out of memory. Tried to allocate 8.00 GiB. GPU 4 has a total capacty of 79.33 GiB of which 4.25 GiB is free. Including non-PyTorch memory, this process has 75.07 GiB memory in use. Of the allocated memory 63.52 GiB is allocated by PyTorch, and 97.18 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 8.00 GiB. GPU 6 has a total capacty of 79.33 GiB of which 4.27 GiB is free. Including non-PyTorch memory, this process has 75.04 GiB memory in use. Of the allocated memory 63.52 GiB is allocated by PyTorch, and 97.18 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    input_embeds = self.token_embedding(input_ids)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    out = out * (~input_mask[..., None])
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 8.00 GiB. GPU 2 has a total capacty of 79.33 GiB of which 4.27 GiB is free. Including non-PyTorch memory, this process has 75.04 GiB memory in use. Of the allocated memory 63.52 GiB is allocated by PyTorch, and 97.18 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 786, in forward
+        return self._call_impl(*args, **kwargs)return self._call_impl(*args, **kwargs)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    out = out * (~input_mask[..., None])
+    input_embeds = self.token_embedding(input_ids)torch.cuda
+.OutOfMemoryError  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+: CUDA out of memory. Tried to allocate 8.00 GiB. GPU 1 has a total capacty of 79.33 GiB of which 4.35 GiB is free. Including non-PyTorch memory, this process has 74.97 GiB memory in use. Of the allocated memory 63.52 GiB is allocated by PyTorch, and 97.18 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 290, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 290, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 290, in forward
+    input_embeds = self.token_embedding(input_ids)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 290, in forward
+    out = out * (~input_mask[..., None])
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 8.00 GiB. GPU 7 has a total capacty of 79.33 GiB of which 4.53 GiB is free. Including non-PyTorch memory, this process has 74.79 GiB memory in use. Of the allocated memory 63.52 GiB is allocated by PyTorch, and 97.18 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    out = out * (~input_mask[..., None])
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 8.00 GiB. GPU 5 has a total capacty of 79.33 GiB of which 4.35 GiB is free. Including non-PyTorch memory, this process has 74.97 GiB memory in use. Of the allocated memory 63.52 GiB is allocated by PyTorch, and 97.18 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    out = out * (~input_mask[..., None])
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 8.00 GiB. GPU 3 has a total capacty of 79.33 GiB of which 4.18 GiB is free. Including non-PyTorch memory, this process has 75.14 GiB memory in use. Of the allocated memory 63.52 GiB is allocated by PyTorch, and 97.18 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    out = out * (~input_mask[..., None])
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 8.00 GiB. GPU 0 has a total capacty of 79.33 GiB of which 4.30 GiB is free. Including non-PyTorch memory, this process has 75.02 GiB memory in use. Of the allocated memory 63.52 GiB is allocated by PyTorch, and 97.18 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 290, in forward
+    out = out * (~input_mask[..., None])
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 8.00 GiB. GPU 3 has a total capacty of 79.33 GiB of which 3.80 GiB is free. Including non-PyTorch memory, this process has 75.51 GiB memory in use. Of the allocated memory 63.52 GiB is allocated by PyTorch, and 97.18 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    trainer.train(dataloader)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+trainer.train(dataloader)
+    trainer.train(dataloader)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return f(*args, **kwargs)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+return f(*args, **kwargs)    
+return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+return f(*args, **kwargs)
+    return f(*args, **kwargs)
+    return f(*args, **kwargs)  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+        outputs = self.pipeline_engine.train_batch_iter(outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+        
+outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+        outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+        outputs = self.pipeline_engine.train_batch_iter(return f(*args, **kwargs)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs = self.pipeline_engine.train_batch_iter(    
+outputs = self.pipeline_engine.train_batch_iter(  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+    
+outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+        output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+Traceback (most recent call last):
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+        output = model(**micro_batch)output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    
+output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    output = model(**micro_batch)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return f(*args, **kwargs)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+Traceback (most recent call last):
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+        trainer.train(dataloader)trainer.train(dataloader)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    trainer.train(dataloader)
+return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return self._call_impl(*args, **kwargs)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)    
+outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)    
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs = self.pipeline_engine.train_batch_iter(
+        outputs = self.pipeline_engine.train_batch_iter(  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+outputs = self.pipeline_engine.train_batch_iter(
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+    return f(*args, **kwargs)
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+        return f(*args, **kwargs)return f(*args, **kwargs)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    return forward_call(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return f(*args, **kwargs)
+    return forward_call(*args, **kwargs)
+          File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+return forward_call(*args, **kwargs)return forward_call(*args, **kwargs)
+
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return forward_call(*args, **kwargs)  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+Traceback (most recent call last):
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+        trainer.train(dataloader)outputs = self.pipeline_engine.train_batch_iter(
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)    
+outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    outputs = self.pipeline_engine.train_batch_iter(
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+        output = model(**micro_batch)
+            outputs = self.pipeline_engine.train_batch_iter(outputs = self.pipeline_engine.train_batch_iter(trainer.train(dataloader)
+
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    output = model(**micro_batch)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return f(*args, **kwargs)
+    sharded_logits = self.model(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+    return self._call_impl(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    sharded_logits = self.model(    
+sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    sharded_logits = self.model(  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    sharded_logits = self.model(
+    sharded_logits = self.model(  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return forward_call(*args, **kwargs)
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    sharded_logits = self.model(
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    output = model(**micro_batch)
+            return self._call_impl(*args, **kwargs)return self._call_impl(*args, **kwargs)return self._call_impl(*args, **kwargs)
+
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 896, in forward_with_hidden_states
+    output = model(**micro_batch)    
+output = model(**micro_batch)    
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 896, in forward_with_hidden_states
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = self.token_position_embeddings(input_ids=input_ids, input_mask=input_mask)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)    
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+        return forward_call(*args, **kwargs)    return forward_call(*args, **kwargs)
+return forward_call(*args, **kwargs)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+        return forward_call(*args, **kwargs)return forward_call(*args, **kwargs)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    return forward_call(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    output = self.token_position_embeddings(input_ids=input_ids, input_mask=input_mask)
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+        return forward_call(*args, **kwargs)return forward_call(*args, **kwargs)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 896, in forward_with_hidden_states
+    output = model(**micro_batch)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 160, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+    return forward_call(*args, **kwargs)
+        return self._call_impl(*args, **kwargs)  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+return self._call_impl(*args, **kwargs)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 160, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 896, in forward_with_hidden_states
+return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 896, in forward_with_hidden_states
+return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 896, in forward_with_hidden_states
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 896, in forward_with_hidden_states
+    output = self.token_position_embeddings(input_ids=input_ids, input_mask=input_mask)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 896, in forward_with_hidden_states
+    output = self.token_position_embeddings(input_ids=input_ids, input_mask=input_mask)        
+output = self.token_position_embeddings(input_ids=input_ids, input_mask=input_mask)
+output = self.token_position_embeddings(input_ids=input_ids, input_mask=input_mask)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.pp_block(**new_kwargs)
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.token_position_embeddings(input_ids=input_ids, input_mask=input_mask)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 786, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 160, in forward
+    return forward_call(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)        
+    return forward_call(*args, **kwargs)return forward_call(*args, **kwargs)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    sharded_logits = self.model(
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)    
+return forward_call(*args, **kwargs)
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    return forward_call(*args, **kwargs)
+
+output = self.token_position_embeddings(input_ids=input_ids, input_mask=input_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 160, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 160, in forward
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 160, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 786, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 160, in forward
+    output = self.pp_block(**new_kwargs)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return self._call_impl(*args, **kwargs)
+    sharded_logits = self.model(
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 160, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    sharded_logits = self.model(
+    input_embeds = self.token_embedding(input_ids)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+        return self._call_impl(*args, **kwargs)return self._call_impl(*args, **kwargs)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    trainer.train(dataloader)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 896, in forward_with_hidden_states
+    sharded_logits = self.model(
+    return forward_call(*args, **kwargs)
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 786, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.pp_block(**new_kwargs)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    input_embeds = self.token_embedding(input_ids)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+        return forward_call(*args, **kwargs)
+return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self._call_impl(*args, **kwargs)
+        return self._call_impl(*args, **kwargs)
+    output = self.token_position_embeddings(input_ids=input_ids, input_mask=input_mask)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 290, in forward
+    return self._call_impl(*args, **kwargs)
+return forward_call(*args, **kwargs)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+    return forward_call(*args, **kwargs)  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self._call_impl(*args, **kwargs)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = self.pp_block(**new_kwargs)
+    output = self.pp_block(**new_kwargs)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 290, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    input_embeds = self.token_embedding(input_ids)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 786, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 786, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return forward_call(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+    return forward_call(*args, **kwargs)  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 786, in forward
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 290, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 786, in forward
+    out = out * (~input_mask[..., None])
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 896, in forward_with_hidden_states
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 8.00 GiB. GPU 6 has a total capacty of 79.33 GiB of which 4.35 GiB is free. Including non-PyTorch memory, this process has 74.97 GiB memory in use. Of the allocated memory 63.52 GiB is allocated by PyTorch, and 97.18 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    out = out * (~input_mask[..., None])
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 8.00 GiB. GPU 4 has a total capacty of 79.33 GiB of which 4.33 GiB is free. Including non-PyTorch memory, this process has 74.99 GiB memory in use. Of the allocated memory 63.52 GiB is allocated by PyTorch, and 97.18 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    input_embeds = self.token_embedding(input_ids)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 896, in forward_with_hidden_states
+        input_embeds = self.token_embedding(input_ids)input_embeds = self.token_embedding(input_ids)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    out = out * (~input_mask[..., None])
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 8.00 GiB. GPU 1 has a total capacty of 79.33 GiB of which 4.27 GiB is free. Including non-PyTorch memory, this process has 75.04 GiB memory in use. Of the allocated memory 63.52 GiB is allocated by PyTorch, and 97.18 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    input_embeds = self.token_embedding(input_ids)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 896, in forward_with_hidden_states
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 896, in forward_with_hidden_states
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 896, in forward_with_hidden_states
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 896, in forward_with_hidden_states
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 896, in forward_with_hidden_states
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 896, in forward_with_hidden_states
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 896, in forward_with_hidden_states
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 160, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 896, in forward_with_hidden_states
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 896, in forward_with_hidden_states
+    output = self.token_position_embeddings(input_ids=input_ids, input_mask=input_mask)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.token_position_embeddings(input_ids=input_ids, input_mask=input_mask)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.token_position_embeddings(input_ids=input_ids, input_mask=input_mask)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+output = self.token_position_embeddings(input_ids=input_ids, input_mask=input_mask)    
+output = self.token_position_embeddings(input_ids=input_ids, input_mask=input_mask)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.token_position_embeddings(input_ids=input_ids, input_mask=input_mask)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.token_position_embeddings(input_ids=input_ids, input_mask=input_mask)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.token_position_embeddings(input_ids=input_ids, input_mask=input_mask)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.token_position_embeddings(input_ids=input_ids, input_mask=input_mask)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+    output = self.token_position_embeddings(input_ids=input_ids, input_mask=input_mask)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+        return self._call_impl(*args, **kwargs)return self._call_impl(*args, **kwargs)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = self.token_position_embeddings(input_ids=input_ids, input_mask=input_mask)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 160, in forward
+    output = self.pp_block(**new_kwargs)
+        return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]return forward_call(*args, **kwargs)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+        return forward_call(*args, **kwargs)  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 896, in forward_with_hidden_states
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 160, in forward
+return forward_call(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 160, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 160, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 160, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 786, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 290, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    return self._call_impl(*args, **kwargs)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 290, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 160, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 290, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    input_embeds = self.token_embedding(input_ids)
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+        return self._call_impl(*args, **kwargs)return self._call_impl(*args, **kwargs)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 160, in forward
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 160, in forward
+return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 160, in forward
+    return forward_call(*args, **kwargs)
+    output = self.token_position_embeddings(input_ids=input_ids, input_mask=input_mask)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)    
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 290, in forward
+return forward_call(*args, **kwargs)  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 160, in forward
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 160, in forward
+    output = self.pp_block(**new_kwargs)
+    out = out * (~input_mask[..., None])
+    output = self.pp_block(**new_kwargs)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+output = self.pp_block(**new_kwargs)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+output = self.pp_block(**new_kwargs)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    out = out * (~input_mask[..., None])
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 8.00 GiB. GPU 3 has a total capacty of 79.33 GiB of which 4.10 GiB is free. Including non-PyTorch memory, this process has 75.22 GiB memory in use. Of the allocated memory 63.52 GiB is allocated by PyTorch, and 97.18 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 8.00 GiB. GPU 0 has a total capacty of 79.33 GiB of which 4.37 GiB is free. Including non-PyTorch memory, this process has 74.94 GiB memory in use. Of the allocated memory 63.52 GiB is allocated by PyTorch, and 97.18 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 786, in forward
+    out = out * (~input_mask[..., None])
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 8.00 GiB. GPU 2 has a total capacty of 79.33 GiB of which 4.35 GiB is free. Including non-PyTorch memory, this process has 74.97 GiB memory in use. Of the allocated memory 63.52 GiB is allocated by PyTorch, and 97.18 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+        output = self.pp_block(**new_kwargs)return self._call_impl(*args, **kwargs)
+
+    out = out * (~input_mask[..., None])
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 8.00 GiB. GPU 5 has a total capacty of 79.33 GiB of which 4.27 GiB is free. Including non-PyTorch memory, this process has 75.04 GiB memory in use. Of the allocated memory 63.52 GiB is allocated by PyTorch, and 97.18 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    output = self.pp_block(**new_kwargs)
+    return self._call_impl(*args, **kwargs)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    input_embeds = self.token_embedding(input_ids)
+            return self._call_impl(*args, **kwargs)return self._call_impl(*args, **kwargs)return self._call_impl(*args, **kwargs)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = self.pp_block(**new_kwargs)
+        return self._call_impl(*args, **kwargs)return self._call_impl(*args, **kwargs)
+    
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+return self._call_impl(*args, **kwargs)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+return forward_call(*args, **kwargs)
+return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 786, in forward
+    return self._call_impl(*args, **kwargs)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+    return forward_call(*args, **kwargs)  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 786, in forward
+    return self._call_impl(*args, **kwargs)return self._call_impl(*args, **kwargs)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 786, in forward
+    return self._call_impl(*args, **kwargs)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 786, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 160, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 786, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 786, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 786, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 786, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 786, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 290, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 290, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 786, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 786, in forward
+    input_embeds = self.token_embedding(input_ids)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    out = out * (~input_mask[..., None])
+    sharded_logits = self.model(
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 8.00 GiB. GPU 5 has a total capacty of 79.33 GiB of which 4.35 GiB is free. Including non-PyTorch memory, this process has 74.97 GiB memory in use. Of the allocated memory 63.52 GiB is allocated by PyTorch, and 97.18 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    out = out * (~input_mask[..., None])
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 8.00 GiB. GPU 7 has a total capacty of 79.33 GiB of which 4.45 GiB is free. Including non-PyTorch memory, this process has 74.87 GiB memory in use. Of the allocated memory 63.52 GiB is allocated by PyTorch, and 97.18 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    input_embeds = self.token_embedding(input_ids)
+    input_embeds = self.token_embedding(input_ids)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+
+    input_embeds = self.token_embedding(input_ids)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    input_embeds = self.token_embedding(input_ids)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    input_embeds = self.token_embedding(input_ids)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    input_embeds = self.token_embedding(input_ids)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    input_embeds = self.token_embedding(input_ids)
+    input_embeds = self.token_embedding(input_ids)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    input_embeds = self.token_embedding(input_ids)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 786, in forward
+    input_embeds = self.token_embedding(input_ids)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+return self._call_impl(*args, **kwargs)    
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    input_embeds = self.token_embedding(input_ids)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 290, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 290, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 290, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 290, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 290, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 896, in forward_with_hidden_states
+    return self._call_impl(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 290, in forward
+        return forward_call(*args, **kwargs)return forward_call(*args, **kwargs)
+
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 290, in forward
+return forward_call(*args, **kwargs)  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 290, in forward
+
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 290, in forward
+return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 290, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 290, in forward
+    out = out * (~input_mask[..., None])
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 8.00 GiB. GPU 6 has a total capacty of 79.33 GiB of which 4.27 GiB is free. Including non-PyTorch memory, this process has 75.04 GiB memory in use. Of the allocated memory 63.52 GiB is allocated by PyTorch, and 97.18 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 290, in forward
+    output = self.token_position_embeddings(input_ids=input_ids, input_mask=input_mask)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    out = out * (~input_mask[..., None])
+    torch.cudaout = out * (~input_mask[..., None]).
+    OutOfMemoryErrorout = out * (~input_mask[..., None]): 
+CUDA out of memory. Tried to allocate 8.00 GiB. GPU 2 has a total capacty of 79.33 GiB of which 4.27 GiB is free. Including non-PyTorch memory, this process has 75.04 GiB memory in use. Of the allocated memory 63.52 GiB is allocated by PyTorch, and 97.18 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONFtorch.cuda
+.torch.cudaOutOfMemoryError.: OutOfMemoryErrorCUDA out of memory. Tried to allocate 8.00 GiB. GPU 4 has a total capacty of 79.33 GiB of which 4.25 GiB is free. Including non-PyTorch memory, this process has 75.07 GiB memory in use. Of the allocated memory 63.52 GiB is allocated by PyTorch, and 97.18 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF: CUDA out of memory. Tried to allocate 8.00 GiB. GPU 7 has a total capacty of 79.33 GiB of which 4.53 GiB is free. Including non-PyTorch memory, this process has 74.79 GiB memory in use. Of the allocated memory 63.52 GiB is allocated by PyTorch, and 97.18 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+
+    out = out * (~input_mask[..., None])
+    out = out * (~input_mask[..., None])
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 8.00 GiB. GPU 7 has a total capacty of 79.33 GiB of which 4.27 GiB is free. Including non-PyTorch memory, this process has 75.04 GiB memory in use. Of the allocated memory 63.52 GiB is allocated by PyTorch, and 97.18 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 8.00 GiB. GPU 1 has a total capacty of 79.33 GiB of which 4.35 GiB is free. Including non-PyTorch memory, this process has 74.97 GiB memory in use. Of the allocated memory 63.52 GiB is allocated by PyTorch, and 97.18 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    out = out * (~input_mask[..., None])
+    out = out * (~input_mask[..., None])
+torch.cuda.    OutOfMemoryErrortorch.cudaout = out * (~input_mask[..., None]): .    
+CUDA out of memory. Tried to allocate 8.00 GiB. GPU 0 has a total capacty of 79.33 GiB of which 4.00 GiB is free. Including non-PyTorch memory, this process has 75.31 GiB memory in use. Of the allocated memory 63.52 GiB is allocated by PyTorch, and 97.18 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONFOutOfMemoryError    out = out * (~input_mask[..., None])
+: torch.cudaout = out * (~input_mask[..., None]).
+CUDA out of memory. Tried to allocate 8.00 GiB. GPU 5 has a total capacty of 79.33 GiB of which 4.10 GiB is free. Including non-PyTorch memory, this process has 75.22 GiB memory in use. Of the allocated memory 63.52 GiB is allocated by PyTorch, and 97.18 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+OutOfMemoryError
+torch.cuda: .torch.cudaCUDA out of memory. Tried to allocate 8.00 GiB. GPU 4 has a total capacty of 79.33 GiB of which 3.96 GiB is free. Including non-PyTorch memory, this process has 75.36 GiB memory in use. Of the allocated memory 63.52 GiB is allocated by PyTorch, and 97.18 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONFOutOfMemoryError.
+: OutOfMemoryErrorCUDA out of memory. Tried to allocate 8.00 GiB. GPU 3 has a total capacty of 79.33 GiB of which 3.92 GiB is free. Including non-PyTorch memory, this process has 75.40 GiB memory in use. Of the allocated memory 63.52 GiB is allocated by PyTorch, and 97.18 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF: 
+CUDA out of memory. Tried to allocate 8.00 GiB. GPU 6 has a total capacty of 79.33 GiB of which 3.98 GiB is free. Including non-PyTorch memory, this process has 75.34 GiB memory in use. Of the allocated memory 63.52 GiB is allocated by PyTorch, and 97.18 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    out = out * (~input_mask[..., None])
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 8.00 GiB. GPU 1 has a total capacty of 79.33 GiB of which 4.10 GiB is free. Including non-PyTorch memory, this process has 75.22 GiB memory in use. Of the allocated memory 63.52 GiB is allocated by PyTorch, and 97.18 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 160, in forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 786, in forward
+    input_embeds = self.token_embedding(input_ids)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 290, in forward
+    out = out * (~input_mask[..., None])
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 8.00 GiB. GPU 2 has a total capacty of 79.33 GiB of which 3.98 GiB is free. Including non-PyTorch memory, this process has 75.34 GiB memory in use. Of the allocated memory 63.52 GiB is allocated by PyTorch, and 97.18 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 896, in forward_with_hidden_states
+    output = self.token_position_embeddings(input_ids=input_ids, input_mask=input_mask)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 160, in forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 786, in forward
+    input_embeds = self.token_embedding(input_ids)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 290, in forward
+    out = out * (~input_mask[..., None])
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 8.00 GiB. GPU 0 has a total capacty of 79.33 GiB of which 4.30 GiB is free. Including non-PyTorch memory, this process has 75.02 GiB memory in use. Of the allocated memory 63.52 GiB is allocated by PyTorch, and 97.18 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 896, in forward_with_hidden_states
+    output = self.token_position_embeddings(input_ids=input_ids, input_mask=input_mask)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 160, in forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 786, in forward
+    input_embeds = self.token_embedding(input_ids)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 290, in forward
+    out = out * (~input_mask[..., None])
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 8.00 GiB. GPU 3 has a total capacty of 79.33 GiB of which 4.18 GiB is free. Including non-PyTorch memory, this process has 75.14 GiB memory in use. Of the allocated memory 63.52 GiB is allocated by PyTorch, and 97.18 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+ip-26-0-160-225:124164:124674 [0] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-160-225:124165:124678 [0] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-160-225:124163:124677 [0] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-160-225:124162:124679 [0] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-160-225:124161:124675 [0] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-160-225:124166:124676 [0] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-160-225:124160:124680 [0] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-160-225:124160:124641 [0] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-160-225:124164:124603 [0] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-160-225:124165:124604 [0] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-160-225:124163:124605 [0] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-160-225:124162:124601 [0] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-160-225:124161:124599 [0] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-160-225:124160:124598 [0] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-160-225:124166:124602 [0] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-160-225:124165:124358 [0] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-160-225:124164:124359 [0] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-160-225:124163:124362 [0] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-160-225:124162:124363 [0] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-160-225:124161:124361 [0] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-160-225:124166:124357 [0] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-160-225:124160:124360 [0] NCCL INFO [Service thread] Connection closed by localRank 7
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+12/28/2024 11:56:10 [ERROR|DP=0|PP=1|TP=23|ip-26-0-165-164]: [RECV META] Rank 1: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 11:56:10 [ERROR|DP=0|PP=1|TP=4|ip-26-0-161-78]: [RECV META] Rank 1: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 11:56:10 [ERROR|DP=0|PP=1|TP=7|ip-26-0-161-78]: [RECV META] Rank 1: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 11:56:10 [ERROR|DP=0|PP=1|TP=21|ip-26-0-165-164]: [RECV META] Rank 1: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 11:56:10 [ERROR|DP=0|PP=1|TP=22|ip-26-0-165-164]: [RECV META] Rank 1: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 11:56:10 [ERROR|DP=0|PP=1|TP=6|ip-26-0-161-78]: [RECV META] Rank 1: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 11:56:10 [ERROR|DP=0|PP=1|TP=0|ip-26-0-161-78]: [RECV META] Rank 1: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 11:56:10 [ERROR|DP=0|PP=1|TP=19|ip-26-0-165-164]: [RECV META] Rank 1: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 11:56:10 [ERROR|DP=0|PP=1|TP=5|ip-26-0-161-78]: [RECV META] Rank 1: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 11:56:10 [WARNING|DP=0|PP=1|TP=23|ip-26-0-165-164]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 11:56:10 [ERROR|DP=0|PP=1|TP=20|ip-26-0-165-164]: [RECV META] Rank 1: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 11:56:10 [ERROR|DP=0|PP=1|TP=3|ip-26-0-161-78]: [RECV META] Rank 1: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 11:56:10 [ERROR|DP=0|PP=1|TP=1|ip-26-0-161-78]: [RECV META] Rank 1: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 11:56:10 [ERROR|DP=0|PP=1|TP=18|ip-26-0-165-164]: [RECV META] Rank 1: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 11:56:10 [ERROR|DP=0|PP=1|TP=17|ip-26-0-165-164]: [RECV META] Rank 1: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 11:56:10 [WARNING|DP=0|PP=1|TP=23|ip-26-0-165-164]: No progress made in communication iteration, attempt 1/1
+12/28/2024 11:56:10 [WARNING|DP=0|PP=1|TP=6|ip-26-0-161-78]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 11:56:10 [WARNING|DP=0|PP=1|TP=4|ip-26-0-161-78]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 11:56:10 [WARNING|DP=0|PP=1|TP=21|ip-26-0-165-164]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 11:56:10 [WARNING|DP=0|PP=1|TP=19|ip-26-0-165-164]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 11:56:10 [ERROR|DP=0|PP=1|TP=23|ip-26-0-165-164]: Max retries reached, giving up on communication
+12/28/2024 11:56:10 [WARNING|DP=0|PP=1|TP=0|ip-26-0-161-78]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 11:56:10 [WARNING|DP=0|PP=1|TP=7|ip-26-0-161-78]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 11:56:10 [WARNING|DP=0|PP=1|TP=22|ip-26-0-165-164]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 11:56:10 [WARNING|DP=0|PP=1|TP=5|ip-26-0-161-78]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 11:56:10 [WARNING|DP=0|PP=1|TP=6|ip-26-0-161-78]: No progress made in communication iteration, attempt 1/1
+12/28/2024 11:56:10 [WARNING|DP=0|PP=1|TP=0|ip-26-0-161-78]: No progress made in communication iteration, attempt 1/1
+12/28/2024 11:56:10 [WARNING|DP=0|PP=1|TP=20|ip-26-0-165-164]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 11:56:10 [WARNING|DP=0|PP=1|TP=21|ip-26-0-165-164]: No progress made in communication iteration, attempt 1/1
+12/28/2024 11:56:10 [WARNING|DP=0|PP=1|TP=18|ip-26-0-165-164]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 11:56:10 [WARNING|DP=0|PP=1|TP=3|ip-26-0-161-78]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 11:56:10 [WARNING|DP=0|PP=1|TP=4|ip-26-0-161-78]: No progress made in communication iteration, attempt 1/1
+12/28/2024 11:56:10 [WARNING|DP=0|PP=1|TP=1|ip-26-0-161-78]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 11:56:10 [WARNING|DP=0|PP=1|TP=17|ip-26-0-165-164]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 11:56:10 [WARNING|DP=0|PP=1|TP=19|ip-26-0-165-164]: No progress made in communication iteration, attempt 1/1
+12/28/2024 11:56:10 [WARNING|DP=0|PP=1|TP=7|ip-26-0-161-78]: No progress made in communication iteration, attempt 1/1
+12/28/2024 11:56:10 [WARNING|DP=0|PP=1|TP=5|ip-26-0-161-78]: No progress made in communication iteration, attempt 1/1
+12/28/2024 11:56:10 [WARNING|DP=0|PP=1|TP=22|ip-26-0-165-164]: No progress made in communication iteration, attempt 1/1
+12/28/2024 11:56:10 [WARNING|DP=0|PP=1|TP=20|ip-26-0-165-164]: No progress made in communication iteration, attempt 1/1
+12/28/2024 11:56:10 [ERROR|DP=0|PP=1|TP=21|ip-26-0-165-164]: Max retries reached, giving up on communication
+12/28/2024 11:56:10 [WARNING|DP=0|PP=1|TP=18|ip-26-0-165-164]: No progress made in communication iteration, attempt 1/1
+12/28/2024 11:56:10 [WARNING|DP=0|PP=1|TP=17|ip-26-0-165-164]: No progress made in communication iteration, attempt 1/1
+12/28/2024 11:56:10 [ERROR|DP=0|PP=1|TP=19|ip-26-0-165-164]: Max retries reached, giving up on communication
+12/28/2024 11:56:10 [ERROR|DP=0|PP=1|TP=6|ip-26-0-161-78]: Max retries reached, giving up on communication
+12/28/2024 11:56:10 [ERROR|DP=0|PP=1|TP=0|ip-26-0-161-78]: Max retries reached, giving up on communication
+12/28/2024 11:56:10 [ERROR|DP=0|PP=1|TP=4|ip-26-0-161-78]: Max retries reached, giving up on communication
+12/28/2024 11:56:10 [WARNING|DP=0|PP=1|TP=3|ip-26-0-161-78]: No progress made in communication iteration, attempt 1/1
+12/28/2024 11:56:10 [WARNING|DP=0|PP=1|TP=1|ip-26-0-161-78]: No progress made in communication iteration, attempt 1/1
+12/28/2024 11:56:10 [ERROR|DP=0|PP=1|TP=7|ip-26-0-161-78]: Max retries reached, giving up on communication
+12/28/2024 11:56:10 [ERROR|DP=0|PP=1|TP=22|ip-26-0-165-164]: Max retries reached, giving up on communication
+12/28/2024 11:56:10 [ERROR|DP=0|PP=1|TP=20|ip-26-0-165-164]: Max retries reached, giving up on communication
+12/28/2024 11:56:10 [ERROR|DP=0|PP=1|TP=5|ip-26-0-161-78]: Max retries reached, giving up on communication
+12/28/2024 11:56:10 [ERROR|DP=0|PP=1|TP=18|ip-26-0-165-164]: Max retries reached, giving up on communication
+12/28/2024 11:56:10 [ERROR|DP=0|PP=1|TP=17|ip-26-0-165-164]: Max retries reached, giving up on communication
+12/28/2024 11:56:10 [ERROR|DP=0|PP=1|TP=3|ip-26-0-161-78]: Max retries reached, giving up on communication
+12/28/2024 11:56:10 [ERROR|DP=0|PP=1|TP=16|ip-26-0-165-164]: [RECV META] Rank 1: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 11:56:10 [ERROR|DP=0|PP=1|TP=1|ip-26-0-161-78]: Max retries reached, giving up on communication
+12/28/2024 11:56:10 [ERROR|DP=0|PP=1|TP=2|ip-26-0-161-78]: [RECV META] Rank 1: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 11:56:10 [ERROR|DP=1|PP=1|TP=22|ip-26-0-168-34]: [RECV META] Rank 1: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 11:56:10 [ERROR|DP=1|PP=1|TP=19|ip-26-0-168-34]: [RECV META] Rank 1: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 11:56:10 [WARNING|DP=0|PP=1|TP=2|ip-26-0-161-78]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 11:56:10 [WARNING|DP=0|PP=1|TP=16|ip-26-0-165-164]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 11:56:10 [WARNING|DP=0|PP=1|TP=2|ip-26-0-161-78]: No progress made in communication iteration, attempt 1/1
+12/28/2024 11:56:10 [ERROR|DP=1|PP=1|TP=20|ip-26-0-168-34]: [RECV META] Rank 1: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 11:56:10 [WARNING|DP=0|PP=1|TP=16|ip-26-0-165-164]: No progress made in communication iteration, attempt 1/1
+12/28/2024 11:56:10 [ERROR|DP=0|PP=1|TP=2|ip-26-0-161-78]: Max retries reached, giving up on communication
+12/28/2024 11:56:10 [ERROR|DP=1|PP=1|TP=21|ip-26-0-168-34]: [RECV META] Rank 1: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 11:56:10 [ERROR|DP=1|PP=1|TP=17|ip-26-0-168-34]: [RECV META] Rank 1: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 11:56:10 [ERROR|DP=0|PP=1|TP=16|ip-26-0-165-164]: Max retries reached, giving up on communication
+12/28/2024 11:56:10 [ERROR|DP=1|PP=1|TP=23|ip-26-0-168-34]: [RECV META] Rank 1: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 11:56:10 [WARNING|DP=1|PP=1|TP=22|ip-26-0-168-34]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 11:56:10 [WARNING|DP=1|PP=1|TP=19|ip-26-0-168-34]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 11:56:10 [WARNING|DP=1|PP=1|TP=20|ip-26-0-168-34]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 11:56:10 [WARNING|DP=1|PP=1|TP=21|ip-26-0-168-34]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 11:56:10 [ERROR|DP=1|PP=1|TP=16|ip-26-0-168-34]: [RECV META] Rank 1: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 11:56:10 [WARNING|DP=1|PP=1|TP=17|ip-26-0-168-34]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 11:56:10 [WARNING|DP=1|PP=1|TP=22|ip-26-0-168-34]: No progress made in communication iteration, attempt 1/1
+12/28/2024 11:56:10 [WARNING|DP=1|PP=1|TP=23|ip-26-0-168-34]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 11:56:10 [WARNING|DP=1|PP=1|TP=19|ip-26-0-168-34]: No progress made in communication iteration, attempt 1/1
+12/28/2024 11:56:10 [WARNING|DP=1|PP=1|TP=20|ip-26-0-168-34]: No progress made in communication iteration, attempt 1/1
+12/28/2024 11:56:10 [WARNING|DP=1|PP=1|TP=21|ip-26-0-168-34]: No progress made in communication iteration, attempt 1/1
+12/28/2024 11:56:10 [ERROR|DP=1|PP=1|TP=22|ip-26-0-168-34]: Max retries reached, giving up on communication
+12/28/2024 11:56:10 [WARNING|DP=1|PP=1|TP=17|ip-26-0-168-34]: No progress made in communication iteration, attempt 1/1
+12/28/2024 11:56:10 [ERROR|DP=1|PP=1|TP=19|ip-26-0-168-34]: Max retries reached, giving up on communication
+12/28/2024 11:56:10 [WARNING|DP=1|PP=1|TP=23|ip-26-0-168-34]: No progress made in communication iteration, attempt 1/1
+12/28/2024 11:56:10 [ERROR|DP=1|PP=1|TP=20|ip-26-0-168-34]: Max retries reached, giving up on communication
+12/28/2024 11:56:10 [ERROR|DP=1|PP=1|TP=21|ip-26-0-168-34]: Max retries reached, giving up on communication
+12/28/2024 11:56:10 [WARNING|DP=1|PP=1|TP=16|ip-26-0-168-34]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 11:56:10 [ERROR|DP=1|PP=1|TP=17|ip-26-0-168-34]: Max retries reached, giving up on communication
+12/28/2024 11:56:10 [WARNING|DP=1|PP=1|TP=16|ip-26-0-168-34]: No progress made in communication iteration, attempt 1/1
+12/28/2024 11:56:10 [ERROR|DP=1|PP=1|TP=18|ip-26-0-168-34]: [RECV META] Rank 1: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 11:56:10 [ERROR|DP=1|PP=1|TP=23|ip-26-0-168-34]: Max retries reached, giving up on communication
+12/28/2024 11:56:10 [ERROR|DP=1|PP=1|TP=16|ip-26-0-168-34]: Max retries reached, giving up on communication
+12/28/2024 11:56:10 [WARNING|DP=1|PP=1|TP=18|ip-26-0-168-34]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 11:56:10 [WARNING|DP=1|PP=1|TP=18|ip-26-0-168-34]: No progress made in communication iteration, attempt 1/1
+12/28/2024 11:56:10 [ERROR|DP=1|PP=1|TP=1|ip-26-0-165-59]: [RECV META] Rank 1: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 11:56:10 [ERROR|DP=1|PP=1|TP=4|ip-26-0-165-59]: [RECV META] Rank 1: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 11:56:10 [ERROR|DP=1|PP=1|TP=18|ip-26-0-168-34]: Max retries reached, giving up on communication
+12/28/2024 11:56:10 [ERROR|DP=1|PP=1|TP=6|ip-26-0-165-59]: [RECV META] Rank 1: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 11:56:10 [ERROR|DP=1|PP=1|TP=2|ip-26-0-165-59]: [RECV META] Rank 1: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 11:56:10 [ERROR|DP=1|PP=1|TP=3|ip-26-0-165-59]: [RECV META] Rank 1: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 11:56:10 [ERROR|DP=1|PP=1|TP=7|ip-26-0-165-59]: [RECV META] Rank 1: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 11:56:10 [ERROR|DP=0|PP=1|TP=9|ip-26-0-165-131]: [RECV META] Rank 1: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 11:56:10 [ERROR|DP=0|PP=1|TP=14|ip-26-0-165-131]: [RECV META] Rank 1: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 11:56:10 [ERROR|DP=1|PP=1|TP=5|ip-26-0-165-59]: [RECV META] Rank 1: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 11:56:10 [ERROR|DP=1|PP=1|TP=0|ip-26-0-165-59]: [RECV META] Rank 1: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 11:56:10 [ERROR|DP=0|PP=1|TP=11|ip-26-0-165-131]: [RECV META] Rank 1: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 11:56:10 [ERROR|DP=0|PP=1|TP=12|ip-26-0-165-131]: [RECV META] Rank 1: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 11:56:10 [WARNING|DP=1|PP=1|TP=1|ip-26-0-165-59]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 11:56:10 [WARNING|DP=1|PP=1|TP=4|ip-26-0-165-59]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 11:56:10 [ERROR|DP=0|PP=1|TP=8|ip-26-0-165-131]: [RECV META] Rank 1: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 11:56:10 [WARNING|DP=1|PP=1|TP=6|ip-26-0-165-59]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 11:56:10 [WARNING|DP=1|PP=1|TP=2|ip-26-0-165-59]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 11:56:10 [ERROR|DP=0|PP=1|TP=13|ip-26-0-165-131]: [RECV META] Rank 1: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 11:56:10 [WARNING|DP=0|PP=1|TP=9|ip-26-0-165-131]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 11:56:10 [WARNING|DP=1|PP=1|TP=7|ip-26-0-165-59]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 11:56:10 [WARNING|DP=1|PP=1|TP=3|ip-26-0-165-59]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 11:56:10 [WARNING|DP=0|PP=1|TP=14|ip-26-0-165-131]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 11:56:10 [WARNING|DP=0|PP=1|TP=12|ip-26-0-165-131]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 11:56:10 [WARNING|DP=1|PP=1|TP=5|ip-26-0-165-59]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 11:56:10 [WARNING|DP=1|PP=1|TP=1|ip-26-0-165-59]: No progress made in communication iteration, attempt 1/1
+12/28/2024 11:56:10 [ERROR|DP=0|PP=1|TP=10|ip-26-0-165-131]: [RECV META] Rank 1: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 11:56:10 [WARNING|DP=1|PP=1|TP=4|ip-26-0-165-59]: No progress made in communication iteration, attempt 1/1
+12/28/2024 11:56:10 [WARNING|DP=1|PP=1|TP=6|ip-26-0-165-59]: No progress made in communication iteration, attempt 1/1
+12/28/2024 11:56:10 [WARNING|DP=1|PP=1|TP=0|ip-26-0-165-59]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 11:56:10 [WARNING|DP=1|PP=1|TP=2|ip-26-0-165-59]: No progress made in communication iteration, attempt 1/1
+12/28/2024 11:56:10 [WARNING|DP=1|PP=1|TP=7|ip-26-0-165-59]: No progress made in communication iteration, attempt 1/1
+12/28/2024 11:56:10 [ERROR|DP=1|PP=1|TP=1|ip-26-0-165-59]: Max retries reached, giving up on communication
+12/28/2024 11:56:10 [WARNING|DP=0|PP=1|TP=11|ip-26-0-165-131]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 11:56:10 [WARNING|DP=0|PP=1|TP=8|ip-26-0-165-131]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 11:56:10 [WARNING|DP=1|PP=1|TP=3|ip-26-0-165-59]: No progress made in communication iteration, attempt 1/1
+12/28/2024 11:56:10 [WARNING|DP=1|PP=1|TP=5|ip-26-0-165-59]: No progress made in communication iteration, attempt 1/1
+12/28/2024 11:56:10 [ERROR|DP=1|PP=1|TP=6|ip-26-0-165-59]: Max retries reached, giving up on communication
+12/28/2024 11:56:10 [ERROR|DP=1|PP=1|TP=4|ip-26-0-165-59]: Max retries reached, giving up on communication
+12/28/2024 11:56:10 [ERROR|DP=1|PP=1|TP=2|ip-26-0-165-59]: Max retries reached, giving up on communication
+12/28/2024 11:56:10 [WARNING|DP=1|PP=1|TP=0|ip-26-0-165-59]: No progress made in communication iteration, attempt 1/1
+12/28/2024 11:56:10 [ERROR|DP=0|PP=1|TP=15|ip-26-0-165-131]: [RECV META] Rank 1: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 11:56:10 [WARNING|DP=0|PP=1|TP=13|ip-26-0-165-131]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 11:56:10 [WARNING|DP=0|PP=1|TP=9|ip-26-0-165-131]: No progress made in communication iteration, attempt 1/1
+12/28/2024 11:56:10 [WARNING|DP=0|PP=1|TP=14|ip-26-0-165-131]: No progress made in communication iteration, attempt 1/1
+12/28/2024 11:56:10 [ERROR|DP=1|PP=1|TP=7|ip-26-0-165-59]: Max retries reached, giving up on communication
+12/28/2024 11:56:10 [ERROR|DP=1|PP=1|TP=3|ip-26-0-165-59]: Max retries reached, giving up on communication
+12/28/2024 11:56:10 [ERROR|DP=1|PP=1|TP=5|ip-26-0-165-59]: Max retries reached, giving up on communication
+12/28/2024 11:56:10 [WARNING|DP=0|PP=1|TP=12|ip-26-0-165-131]: No progress made in communication iteration, attempt 1/1
+12/28/2024 11:56:10 [WARNING|DP=0|PP=1|TP=11|ip-26-0-165-131]: No progress made in communication iteration, attempt 1/1
+12/28/2024 11:56:10 [WARNING|DP=0|PP=1|TP=8|ip-26-0-165-131]: No progress made in communication iteration, attempt 1/1
+12/28/2024 11:56:10 [ERROR|DP=1|PP=1|TP=0|ip-26-0-165-59]: Max retries reached, giving up on communication
+12/28/2024 11:56:10 [ERROR|DP=0|PP=1|TP=14|ip-26-0-165-131]: Max retries reached, giving up on communication
+12/28/2024 11:56:10 [ERROR|DP=0|PP=1|TP=9|ip-26-0-165-131]: Max retries reached, giving up on communication
+12/28/2024 11:56:10 [WARNING|DP=0|PP=1|TP=13|ip-26-0-165-131]: No progress made in communication iteration, attempt 1/1
+12/28/2024 11:56:10 [WARNING|DP=0|PP=1|TP=10|ip-26-0-165-131]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 11:56:10 [ERROR|DP=0|PP=1|TP=12|ip-26-0-165-131]: Max retries reached, giving up on communication
+12/28/2024 11:56:10 [WARNING|DP=0|PP=1|TP=15|ip-26-0-165-131]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 11:56:10 [ERROR|DP=0|PP=1|TP=11|ip-26-0-165-131]: Max retries reached, giving up on communication
+12/28/2024 11:56:10 [ERROR|DP=0|PP=1|TP=8|ip-26-0-165-131]: Max retries reached, giving up on communication
+12/28/2024 11:56:10 [ERROR|DP=0|PP=1|TP=13|ip-26-0-165-131]: Max retries reached, giving up on communication
+12/28/2024 11:56:10 [WARNING|DP=0|PP=1|TP=10|ip-26-0-165-131]: No progress made in communication iteration, attempt 1/1
+12/28/2024 11:56:10 [WARNING|DP=0|PP=1|TP=15|ip-26-0-165-131]: No progress made in communication iteration, attempt 1/1
+12/28/2024 11:56:10 [ERROR|DP=0|PP=1|TP=10|ip-26-0-165-131]: Max retries reached, giving up on communication
+12/28/2024 11:56:10 [ERROR|DP=0|PP=1|TP=15|ip-26-0-165-131]: Max retries reached, giving up on communication
+12/28/2024 11:56:10 [ERROR|DP=1|PP=1|TP=25|ip-26-0-168-52]: [RECV META] Rank 1: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 11:56:10 [ERROR|DP=1|PP=1|TP=31|ip-26-0-168-52]: [RECV META] Rank 1: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 11:56:10 [ERROR|DP=1|PP=1|TP=28|ip-26-0-168-52]: [RECV META] Rank 1: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 11:56:10 [ERROR|DP=1|PP=1|TP=30|ip-26-0-168-52]: [RECV META] Rank 1: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 11:56:10 [ERROR|DP=1|PP=1|TP=29|ip-26-0-168-52]: [RECV META] Rank 1: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 11:56:10 [ERROR|DP=1|PP=1|TP=26|ip-26-0-168-52]: [RECV META] Rank 1: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 11:56:10 [ERROR|DP=1|PP=1|TP=27|ip-26-0-168-52]: [RECV META] Rank 1: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 11:56:10 [WARNING|DP=1|PP=1|TP=25|ip-26-0-168-52]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 11:56:10 [ERROR|DP=1|PP=1|TP=24|ip-26-0-168-52]: [RECV META] Rank 1: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 11:56:10 [WARNING|DP=1|PP=1|TP=28|ip-26-0-168-52]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 11:56:10 [WARNING|DP=1|PP=1|TP=31|ip-26-0-168-52]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 11:56:10 [WARNING|DP=1|PP=1|TP=30|ip-26-0-168-52]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 11:56:10 [WARNING|DP=1|PP=1|TP=26|ip-26-0-168-52]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 11:56:10 [WARNING|DP=1|PP=1|TP=29|ip-26-0-168-52]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 11:56:10 [WARNING|DP=1|PP=1|TP=25|ip-26-0-168-52]: No progress made in communication iteration, attempt 1/1
+12/28/2024 11:56:10 [WARNING|DP=1|PP=1|TP=27|ip-26-0-168-52]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 11:56:10 [WARNING|DP=1|PP=1|TP=31|ip-26-0-168-52]: No progress made in communication iteration, attempt 1/1
+12/28/2024 11:56:10 [WARNING|DP=1|PP=1|TP=28|ip-26-0-168-52]: No progress made in communication iteration, attempt 1/1
+12/28/2024 11:56:10 [WARNING|DP=1|PP=1|TP=30|ip-26-0-168-52]: No progress made in communication iteration, attempt 1/1
+12/28/2024 11:56:10 [WARNING|DP=1|PP=1|TP=26|ip-26-0-168-52]: No progress made in communication iteration, attempt 1/1
+12/28/2024 11:56:10 [WARNING|DP=1|PP=1|TP=24|ip-26-0-168-52]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 11:56:10 [ERROR|DP=1|PP=1|TP=25|ip-26-0-168-52]: Max retries reached, giving up on communication
+12/28/2024 11:56:10 [WARNING|DP=1|PP=1|TP=29|ip-26-0-168-52]: No progress made in communication iteration, attempt 1/1
+12/28/2024 11:56:10 [WARNING|DP=1|PP=1|TP=27|ip-26-0-168-52]: No progress made in communication iteration, attempt 1/1
+12/28/2024 11:56:10 [ERROR|DP=1|PP=1|TP=31|ip-26-0-168-52]: Max retries reached, giving up on communication
+12/28/2024 11:56:10 [ERROR|DP=1|PP=1|TP=28|ip-26-0-168-52]: Max retries reached, giving up on communication
+12/28/2024 11:56:10 [ERROR|DP=1|PP=1|TP=30|ip-26-0-168-52]: Max retries reached, giving up on communication
+12/28/2024 11:56:10 [ERROR|DP=1|PP=1|TP=26|ip-26-0-168-52]: Max retries reached, giving up on communication
+12/28/2024 11:56:10 [ERROR|DP=1|PP=1|TP=27|ip-26-0-168-52]: Max retries reached, giving up on communication
+12/28/2024 11:56:10 [ERROR|DP=1|PP=1|TP=29|ip-26-0-168-52]: Max retries reached, giving up on communication
+12/28/2024 11:56:10 [WARNING|DP=1|PP=1|TP=24|ip-26-0-168-52]: No progress made in communication iteration, attempt 1/1
+12/28/2024 11:56:10 [ERROR|DP=1|PP=1|TP=24|ip-26-0-168-52]: Max retries reached, giving up on communication
+12/28/2024 11:56:10 [ERROR|DP=0|PP=1|TP=25|ip-26-0-165-202]: [RECV META] Rank 1: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 11:56:10 [ERROR|DP=0|PP=1|TP=30|ip-26-0-165-202]: [RECV META] Rank 1: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 11:56:10 [ERROR|DP=0|PP=1|TP=31|ip-26-0-165-202]: [RECV META] Rank 1: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 11:56:10 [ERROR|DP=0|PP=1|TP=29|ip-26-0-165-202]: [RECV META] Rank 1: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 11:56:10 [ERROR|DP=0|PP=1|TP=28|ip-26-0-165-202]: [RECV META] Rank 1: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 11:56:10 [ERROR|DP=0|PP=1|TP=24|ip-26-0-165-202]: [RECV META] Rank 1: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 11:56:10 [WARNING|DP=0|PP=1|TP=25|ip-26-0-165-202]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 11:56:10 [ERROR|DP=0|PP=1|TP=27|ip-26-0-165-202]: [RECV META] Rank 1: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 11:56:10 [ERROR|DP=0|PP=1|TP=26|ip-26-0-165-202]: [RECV META] Rank 1: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 11:56:10 [WARNING|DP=0|PP=1|TP=30|ip-26-0-165-202]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 11:56:10 [WARNING|DP=0|PP=1|TP=25|ip-26-0-165-202]: No progress made in communication iteration, attempt 1/1
+12/28/2024 11:56:10 [WARNING|DP=0|PP=1|TP=31|ip-26-0-165-202]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 11:56:10 [WARNING|DP=0|PP=1|TP=28|ip-26-0-165-202]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 11:56:10 [WARNING|DP=0|PP=1|TP=29|ip-26-0-165-202]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 11:56:10 [WARNING|DP=0|PP=1|TP=24|ip-26-0-165-202]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 11:56:10 [ERROR|DP=0|PP=1|TP=25|ip-26-0-165-202]: Max retries reached, giving up on communication
+12/28/2024 11:56:10 [WARNING|DP=0|PP=1|TP=27|ip-26-0-165-202]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 11:56:10 [WARNING|DP=0|PP=1|TP=30|ip-26-0-165-202]: No progress made in communication iteration, attempt 1/1
+12/28/2024 11:56:10 [WARNING|DP=0|PP=1|TP=31|ip-26-0-165-202]: No progress made in communication iteration, attempt 1/1
+12/28/2024 11:56:10 [WARNING|DP=0|PP=1|TP=29|ip-26-0-165-202]: No progress made in communication iteration, attempt 1/1
+12/28/2024 11:56:10 [WARNING|DP=0|PP=1|TP=28|ip-26-0-165-202]: No progress made in communication iteration, attempt 1/1
+12/28/2024 11:56:10 [WARNING|DP=0|PP=1|TP=26|ip-26-0-165-202]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 11:56:10 [WARNING|DP=0|PP=1|TP=24|ip-26-0-165-202]: No progress made in communication iteration, attempt 1/1
+12/28/2024 11:56:10 [ERROR|DP=0|PP=1|TP=30|ip-26-0-165-202]: Max retries reached, giving up on communication
+12/28/2024 11:56:10 [WARNING|DP=0|PP=1|TP=27|ip-26-0-165-202]: No progress made in communication iteration, attempt 1/1
+12/28/2024 11:56:10 [ERROR|DP=0|PP=1|TP=31|ip-26-0-165-202]: Max retries reached, giving up on communication
+12/28/2024 11:56:10 [ERROR|DP=0|PP=1|TP=29|ip-26-0-165-202]: Max retries reached, giving up on communication
+12/28/2024 11:56:10 [ERROR|DP=0|PP=1|TP=28|ip-26-0-165-202]: Max retries reached, giving up on communication
+12/28/2024 11:56:10 [WARNING|DP=0|PP=1|TP=26|ip-26-0-165-202]: No progress made in communication iteration, attempt 1/1
+12/28/2024 11:56:10 [ERROR|DP=0|PP=1|TP=27|ip-26-0-165-202]: Max retries reached, giving up on communication
+12/28/2024 11:56:10 [ERROR|DP=0|PP=1|TP=24|ip-26-0-165-202]: Max retries reached, giving up on communication
+12/28/2024 11:56:10 [ERROR|DP=0|PP=1|TP=26|ip-26-0-165-202]: Max retries reached, giving up on communication
+12/28/2024 11:56:10 [ERROR|DP=1|PP=1|TP=8|ip-26-0-168-30]: [RECV META] Rank 1: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 11:56:10 [ERROR|DP=1|PP=1|TP=12|ip-26-0-168-30]: [RECV META] Rank 1: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 11:56:10 [ERROR|DP=1|PP=1|TP=10|ip-26-0-168-30]: [RECV META] Rank 1: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 11:56:10 [ERROR|DP=1|PP=1|TP=13|ip-26-0-168-30]: [RECV META] Rank 1: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 11:56:10 [ERROR|DP=1|PP=1|TP=14|ip-26-0-168-30]: [RECV META] Rank 1: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 11:56:10 [ERROR|DP=1|PP=1|TP=11|ip-26-0-168-30]: [RECV META] Rank 1: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 11:56:10 [ERROR|DP=1|PP=1|TP=15|ip-26-0-168-30]: [RECV META] Rank 1: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 11:56:10 [WARNING|DP=1|PP=1|TP=8|ip-26-0-168-30]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 11:56:10 [WARNING|DP=1|PP=1|TP=12|ip-26-0-168-30]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 11:56:10 [WARNING|DP=1|PP=1|TP=10|ip-26-0-168-30]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 11:56:10 [ERROR|DP=1|PP=1|TP=9|ip-26-0-168-30]: [RECV META] Rank 1: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 11:56:10 [WARNING|DP=1|PP=1|TP=13|ip-26-0-168-30]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 11:56:10 [WARNING|DP=1|PP=1|TP=8|ip-26-0-168-30]: No progress made in communication iteration, attempt 1/1
+12/28/2024 11:56:10 [WARNING|DP=1|PP=1|TP=12|ip-26-0-168-30]: No progress made in communication iteration, attempt 1/1
+12/28/2024 11:56:10 [WARNING|DP=1|PP=1|TP=10|ip-26-0-168-30]: No progress made in communication iteration, attempt 1/1
+12/28/2024 11:56:10 [WARNING|DP=1|PP=1|TP=14|ip-26-0-168-30]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 11:56:10 [WARNING|DP=1|PP=1|TP=11|ip-26-0-168-30]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 11:56:10 [WARNING|DP=1|PP=1|TP=13|ip-26-0-168-30]: No progress made in communication iteration, attempt 1/1
+12/28/2024 11:56:10 [ERROR|DP=1|PP=1|TP=8|ip-26-0-168-30]: Max retries reached, giving up on communication
+12/28/2024 11:56:10 [WARNING|DP=1|PP=1|TP=15|ip-26-0-168-30]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 11:56:10 [ERROR|DP=1|PP=1|TP=12|ip-26-0-168-30]: Max retries reached, giving up on communication
+12/28/2024 11:56:10 [ERROR|DP=1|PP=1|TP=10|ip-26-0-168-30]: Max retries reached, giving up on communication
+12/28/2024 11:56:10 [WARNING|DP=1|PP=1|TP=9|ip-26-0-168-30]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 0: [1] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '0:1', but store->get('0:1') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+12/28/2024 11:56:10 [WARNING|DP=1|PP=1|TP=14|ip-26-0-168-30]: No progress made in communication iteration, attempt 1/1
+12/28/2024 11:56:10 [ERROR|DP=1|PP=1|TP=13|ip-26-0-168-30]: Max retries reached, giving up on communication
+12/28/2024 11:56:10 [WARNING|DP=1|PP=1|TP=11|ip-26-0-168-30]: No progress made in communication iteration, attempt 1/1
+12/28/2024 11:56:10 [WARNING|DP=1|PP=1|TP=15|ip-26-0-168-30]: No progress made in communication iteration, attempt 1/1
+12/28/2024 11:56:10 [WARNING|DP=1|PP=1|TP=9|ip-26-0-168-30]: No progress made in communication iteration, attempt 1/1
+12/28/2024 11:56:10 [ERROR|DP=1|PP=1|TP=14|ip-26-0-168-30]: Max retries reached, giving up on communication
+12/28/2024 11:56:10 [ERROR|DP=1|PP=1|TP=11|ip-26-0-168-30]: Max retries reached, giving up on communication
+12/28/2024 11:56:10 [ERROR|DP=1|PP=1|TP=15|ip-26-0-168-30]: Max retries reached, giving up on communication
+12/28/2024 11:56:10 [ERROR|DP=1|PP=1|TP=9|ip-26-0-168-30]: Max retries reached, giving up on communication
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+Traceback (most recent call last):
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+Traceback (most recent call last):
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    trainer.train(dataloader)
+    trainer.train(dataloader)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+trainer.train(dataloader)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+trainer.train(dataloader)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+trainer.train(dataloader)
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+Traceback (most recent call last):
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    trainer.train(dataloader)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+trainer.train(dataloader)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+    trainer.train(dataloader)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+trainer.train(dataloader)    
+trainer.train(dataloader)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+trainer.train(dataloader)
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+trainer.train(dataloader)
+    trainer.train(dataloader)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+trainer.train(dataloader)    
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    sharded_logits = self.model(
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+trainer.train(dataloader)
+trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+    activation = pipeline_state.activations_buffer.popleft()
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+IndexError: pop from an empty deque
+    trainer.train(dataloader)    
+return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    trainer.train(dataloader)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+trainer.train(dataloader)
+    trainer.train(dataloader)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+
+        trainer.train(dataloader)return f(*args, **kwargs)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+trainer.train(dataloader)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    trainer.train(dataloader)
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    trainer.train(dataloader)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+
+    trainer.train(dataloader)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+
+    trainer.train(dataloader)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+trainer.train(dataloader)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+        trainer.train(dataloader)return f(*args, **kwargs)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+Traceback (most recent call last):
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return f(*args, **kwargs)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+return f(*args, **kwargs)
+    trainer.train(dataloader)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return f(*args, **kwargs)  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+return f(*args, **kwargs)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+trainer.train(dataloader)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+trainer.train(dataloader)
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    trainer.train(dataloader)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+return f(*args, **kwargs)
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+trainer.train(dataloader)    
+trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return f(*args, **kwargs)  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+
+    return f(*args, **kwargs)  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return f(*args, **kwargs)  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+
+    return f(*args, **kwargs)  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs = self.pipeline_engine.train_batch_iter(
+    return f(*args, **kwargs)    
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+        return f(*args, **kwargs)return f(*args, **kwargs)
+
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    return f(*args, **kwargs)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return f(*args, **kwargs)
+    return f(*args, **kwargs)  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+        outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs = self.pipeline_engine.train_batch_iter(    
+outputs = self.pipeline_engine.train_batch_iter(  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    outputs = self.pipeline_engine.train_batch_iter(
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    
+output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+            outputs = self.pipeline_engine.train_batch_iter(outputs = self.pipeline_engine.train_batch_iter(outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+        output = model(**micro_batch)output = model(**micro_batch)
+
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+output = model(**micro_batch)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    output = model(**micro_batch)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+        trainer.train(dataloader)trainer.train(dataloader)
+
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+trainer.train(dataloader)    trainer.train(dataloader)
+trainer.train(dataloader)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs = self.pipeline_engine.train_batch_iter(
+          File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+outputs = self.pipeline_engine.train_batch_iter(outputs = self.pipeline_engine.train_batch_iter(
+
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    return f(*args, **kwargs)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+        output = model(**micro_batch)
+output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+return f(*args, **kwargs)    
+return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)    
+outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    output = model(**micro_batch)
+    output = model(**micro_batch)
+outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    return self._call_impl(*args, **kwargs)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+return self._call_impl(*args, **kwargs)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+        outputs = self.pipeline_engine.train_batch_iter(
+outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    outputs = self.pipeline_engine.train_batch_iter(
+    return self._call_impl(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)    
+outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return self._call_impl(*args, **kwargs)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    return forward_call(*args, **kwargs)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+        return forward_call(*args, **kwargs)return forward_call(*args, **kwargs)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+        return self._call_impl(*args, **kwargs)
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = model(**micro_batch)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+        return self._call_impl(*args, **kwargs)return forward_call(*args, **kwargs)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+        return forward_call(*args, **kwargs)return forward_call(*args, **kwargs)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    return f(*args, **kwargs)
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return f(*args, **kwargs)  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+return f(*args, **kwargs)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+return f(*args, **kwargs)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+Traceback (most recent call last):
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return self._call_impl(*args, **kwargs)
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    output = model(**micro_batch)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+Traceback (most recent call last):
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+    return forward_call(*args, **kwargs)
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs = self.pipeline_engine.train_batch_iter(
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    output = model(**micro_batch)
+                outputs = self.pipeline_engine.train_batch_iter(outputs = self.pipeline_engine.train_batch_iter(outputs = self.pipeline_engine.train_batch_iter(outputs = self.pipeline_engine.train_batch_iter(
+
+
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    sharded_logits = self.model(
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)    
+output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    sharded_logits = self.model(
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+sharded_logits = self.model(
+        output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)output = model(**micro_batch)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    output = model(**micro_batch)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+sharded_logits = self.model(    
+sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)    
+output = model(**micro_batch)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = model(**micro_batch)
+    return self._call_impl(*args, **kwargs)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    return self._call_impl(*args, **kwargs)    
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+            sharded_logits = self.model(sharded_logits = self.model(sharded_logits = self.model(
+
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+        return self._call_impl(*args, **kwargs)return self._call_impl(*args, **kwargs)    
+
+    return self._call_impl(*args, **kwargs)
+return self._call_impl(*args, **kwargs)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    sharded_logits = self.model(
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    sharded_logits = self.model(  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return f(*args, **kwargs)
+        return forward_call(*args, **kwargs)return forward_call(*args, **kwargs)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+        return forward_call(*args, **kwargs)        return self._call_impl(*args, **kwargs)
+return forward_call(*args, **kwargs)return forward_call(*args, **kwargs)
+
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+return forward_call(*args, **kwargs)          File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+return forward_call(*args, **kwargs)  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+    return self._call_impl(*args, **kwargs)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    sharded_logits = self.model(
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    trainer.train(dataloader)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self._call_impl(*args, **kwargs)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+trainer.train(dataloader)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+
+
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+trainer.train(dataloader)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+trainer.train(dataloader)    
+trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+return self._call_impl(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+        output = model(**micro_batch)    output = model(**micro_batch)
+output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)    
+
+output = model(**micro_batch)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = model(**micro_batch)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+        outputs = self.pipeline_engine.train_batch_iter(outputs = self.pipeline_engine.train_batch_iter(
+
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+    return self._call_impl(*args, **kwargs)    
+    return self._call_impl(*args, **kwargs)return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return forward_call(*args, **kwargs)
+          File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+return forward_call(*args, **kwargs)return forward_call(*args, **kwargs)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+        return forward_call(*args, **kwargs)
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    activation = pipeline_state.activations_buffer.popleft()
+IndexError: pop from an empty deque
+    return self._call_impl(*args, **kwargs)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+return forward_call(*args, **kwargs)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]    
+return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return self._call_impl(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    activation = pipeline_state.activations_buffer.popleft()
+IndexError: pop from an empty deque
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+    return self._call_impl(*args, **kwargs)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+return self._call_impl(*args, **kwargs)    
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+        return forward_call(*args, **kwargs)    return forward_call(*args, **kwargs)
+
+return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+    return forward_call(*args, **kwargs)
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+        hidden_encoder_states = encoder_block(**hidden_encoder_states)hidden_encoder_states = encoder_block(**hidden_encoder_states)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    sharded_logits = self.model(
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+        return forward_call(*args, **kwargs)return self._call_impl(*args, **kwargs)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    activation = pipeline_state.activations_buffer.popleft()
+    sharded_logits = self.model(
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+sharded_logits = self.model(
+    sharded_logits = self.model(
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+IndexError: pop from an empty deque
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)        
+sharded_logits = self.model(return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+sharded_logits = self.model(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return forward_call(*args, **kwargs)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+return forward_call(*args, **kwargs)
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    return forward_call(*args, **kwargs)
+    activation = pipeline_state.activations_buffer.popleft()
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)    
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    sharded_logits = self.model(  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+
+IndexError: pop from an empty deque
+    return f(*args, **kwargs)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+return f(*args, **kwargs)
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+
+        return forward_call(*args, **kwargs)return self._call_impl(*args, **kwargs)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return f(*args, **kwargs)  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+
+    return f(*args, **kwargs)  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+
+return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)    
+output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+        output = model(**micro_batch)output = model(**micro_batch)
+
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+        sharded_logits = self.model(sharded_logits = self.model(    
+
+        output = model(**micro_batch)output = model(**micro_batch)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+sharded_logits = self.model(    
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+        return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+        return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+        hidden_encoder_states = encoder_block(**hidden_encoder_states)hidden_encoder_states = encoder_block(**hidden_encoder_states)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = model(**micro_batch)    
+return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    activation = pipeline_state.activations_buffer.popleft()
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+IndexError: pop from an empty deque
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+        return self._call_impl(*args, **kwargs)return self._call_impl(*args, **kwargs)    
+
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    activation = pipeline_state.activations_buffer.popleft()
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+IndexError: pop from an empty deque
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    sharded_logits = self.model(
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+        return self._call_impl(*args, **kwargs)return forward_call(*args, **kwargs)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+        return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+        return self._call_impl(*args, **kwargs)return forward_call(*args, **kwargs)
+
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+    return forward_call(*args, **kwargs)    
+return forward_call(*args, **kwargs)return forward_call(*args, **kwargs)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+    return forward_call(*args, **kwargs)
+        return self._call_impl(*args, **kwargs)return self._call_impl(*args, **kwargs)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+        return self._call_impl(*args, **kwargs)hidden_encoder_states = encoder_block(**hidden_encoder_states)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)      File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+
+outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+        return forward_call(*args, **kwargs)return forward_call(*args, **kwargs)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    sharded_logits = self.model(
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+sharded_logits = self.model(
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+    outputs = self.pipeline_engine.train_batch_iter(
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+          File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    sharded_logits = self.model(
+outputs = self.pipeline_engine.train_batch_iter(outputs = self.pipeline_engine.train_batch_iter(
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+        return forward_call(*args, **kwargs)return self._call_impl(*args, **kwargs)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)    
+return forward_call(*args, **kwargs)  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    return forward_call(*args, **kwargs)
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+        return self._call_impl(*args, **kwargs)return self._call_impl(*args, **kwargs)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+return self._call_impl(*args, **kwargs)    
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    sharded_logits = self.model(
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+        return forward_call(*args, **kwargs)return self._call_impl(*args, **kwargs)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    activation = pipeline_state.activations_buffer.popleft()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)    
+return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+        return forward_call(*args, **kwargs)return forward_call(*args, **kwargs)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+IndexError: pop from an empty deque
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+        return self._call_impl(*args, **kwargs)return forward_call(*args, **kwargs)
+
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]    
+return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+        return self._call_impl(*args, **kwargs)
+return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+    return forward_call(*args, **kwargs)
+        return forward_call(*args, **kwargs)return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return forward_call(*args, **kwargs)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+    return forward_call(*args, **kwargs)
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]    
+return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+new_kwargs[name] = recv_from_pipeline_state_buffer(
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+new_kwargs[name] = recv_from_pipeline_state_buffer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+        hidden_encoder_states = encoder_block(**hidden_encoder_states)
+return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    activation = pipeline_state.activations_buffer.popleft()
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+    return forward_call(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)
+        hidden_encoder_states = encoder_block(**hidden_encoder_states)hidden_encoder_states = encoder_block(**hidden_encoder_states)
+    
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+hidden_encoder_states = encoder_block(**hidden_encoder_states)
+IndexError: pop from an empty deque
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    activation = pipeline_state.activations_buffer.popleft()
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)    
+return self._call_impl(*args, **kwargs)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+return self._call_impl(*args, **kwargs)
+IndexError: pop from an empty deque
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+    new_kwargs[name] = recv_from_pipeline_state_buffer(  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+    return self._call_impl(*args, **kwargs)
+        output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    
+output = model(**micro_batch)  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+output = model(**micro_batch)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+return self._call_impl(*args, **kwargs)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    activation = pipeline_state.activations_buffer.popleft()
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+IndexError: pop from an empty deque
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+        return forward_call(*args, **kwargs)return forward_call(*args, **kwargs)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+        new_kwargs[name] = recv_from_pipeline_state_buffer(new_kwargs[name] = recv_from_pipeline_state_buffer(
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+return forward_call(*args, **kwargs)    
+    return forward_call(*args, **kwargs)  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+return forward_call(*args, **kwargs)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    activation = pipeline_state.activations_buffer.popleft()
+IndexError: pop from an empty deque
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    activation = pipeline_state.activations_buffer.popleft()
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)    
+return forward_call(*args, **kwargs)  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+        return self._call_impl(*args, **kwargs)return forward_call(*args, **kwargs)
+
+    return forward_call(*args, **kwargs)  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+IndexError: pop from an empty deque
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    activation = pipeline_state.activations_buffer.popleft()
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+        return forward_call(*args, **kwargs)return forward_call(*args, **kwargs)
+
+    activation = pipeline_state.activations_buffer.popleft()
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+IndexError: pop from an empty deque
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+IndexError: pop from an empty deque
+    activation = pipeline_state.activations_buffer.popleft()
+IndexError: pop from an empty deque
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+        return self._call_impl(*args, **kwargs)return self._call_impl(*args, **kwargs)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    activation = pipeline_state.activations_buffer.popleft()
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+IndexError: pop from an empty deque
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+    return forward_call(*args, **kwargs)  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    activation = pipeline_state.activations_buffer.popleft()
+        return forward_call(*args, **kwargs)new_kwargs[name] = recv_from_pipeline_state_buffer(
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+IndexError: pop from an empty deque
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+    activation = pipeline_state.activations_buffer.popleft()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+    IndexErroractivation = pipeline_state.activations_buffer.popleft()    : 
+activation = pipeline_state.activations_buffer.popleft()pop from an empty deque
+    
+IndexError    activation = pipeline_state.activations_buffer.popleft(): activation = pipeline_state.activations_buffer.popleft()pop from an empty deque
+    activation = pipeline_state.activations_buffer.popleft()
+IndexError
+
+: IndexErrorpop from an empty deque: IndexError
+: pop from an empty dequepop from an empty deque
+
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+IndexError: pop from an empty deque
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+    activation = pipeline_state.activations_buffer.popleft()
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+IndexError: pop from an empty deque
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+        activation = pipeline_state.activations_buffer.popleft()activation = pipeline_state.activations_buffer.popleft()
+
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+return forward_call(*args, **kwargs)
+    return forward_call(*args, **kwargs)  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+    activation = pipeline_state.activations_buffer.popleft()
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+    activation = pipeline_state.activations_buffer.popleft()
+IndexError: pop from an empty deque
+IndexErrorIndexError: : pop from an empty dequepop from an empty deque
+
+IndexError: pop from an empty deque
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+    activation = pipeline_state.activations_buffer.popleft()
+IndexError: pop from an empty deque
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    activation = pipeline_state.activations_buffer.popleft()
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    new_kwargs[name] = recv_from_pipeline_state_buffer(    
+new_kwargs[name] = recv_from_pipeline_state_buffer(    
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+    activation = pipeline_state.activations_buffer.popleft()
+IndexError: pop from an empty deque
+IndexError: pop from an empty deque
+        activation = pipeline_state.activations_buffer.popleft()
+hidden_encoder_states = encoder_block(**hidden_encoder_states)
+        hidden_encoder_states = encoder_block(**hidden_encoder_states)return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+new_kwargs[name] = recv_from_pipeline_state_buffer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+IndexError: pop from an empty deque
+        hidden_encoder_states = encoder_block(**hidden_encoder_states)    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+
+hidden_encoder_states = encoder_block(**hidden_encoder_states)      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+    sharded_logits = self.model(
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+sharded_logits = self.model(
+    activation = pipeline_state.activations_buffer.popleft()
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+sharded_logits = self.model(
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+hidden_encoder_states = encoder_block(**hidden_encoder_states)    
+return self._call_impl(*args, **kwargs)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+sharded_logits = self.model(  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+IndexError: pop from an empty deque
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+    activation = pipeline_state.activations_buffer.popleft()
+    IndexErroractivation = pipeline_state.activations_buffer.popleft(): 
+pop from an empty deque
+IndexError: pop from an empty deque
+    activation = pipeline_state.activations_buffer.popleft()
+    activation = pipeline_state.activations_buffer.popleft()
+IndexError: pop from an empty deque
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+IndexError: pop from an empty deque
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+    activation = pipeline_state.activations_buffer.popleft()
+IndexError: pop from an empty deque
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+new_kwargs[name] = recv_from_pipeline_state_buffer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    activation = pipeline_state.activations_buffer.popleft()
+    activation = pipeline_state.activations_buffer.popleft()IndexError
+: pop from an empty deque
+IndexError: pop from an empty deque
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+    return self._call_impl(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+
+    activation = pipeline_state.activations_buffer.popleft()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    activation = pipeline_state.activations_buffer.popleft()
+IndexError: pop from an empty deque
+IndexError: pop from an empty deque
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+        return forward_call(*args, **kwargs)return forward_call(*args, **kwargs)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+        return forward_call(*args, **kwargs)return forward_call(*args, **kwargs)
+
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+return forward_call(*args, **kwargs)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)    
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    activation = pipeline_state.activations_buffer.popleft()
+    activation = pipeline_state.activations_buffer.popleft()
+IndexError: pop from an empty deque
+    return self._call_impl(*args, **kwargs)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+IndexError: pop from an empty deque
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return forward_call(*args, **kwargs)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    activation = pipeline_state.activations_buffer.popleft()
+    activation = pipeline_state.activations_buffer.popleft()IndexError
+    : pop from an empty dequeactivation = pipeline_state.activations_buffer.popleft()
+IndexError
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+: pop from an empty dequeIndexError
+: pop from an empty deque
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+    activation = pipeline_state.activations_buffer.popleft()
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+IndexError: pop from an empty deque
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    activation = pipeline_state.activations_buffer.popleft()
+IndexError: pop from an empty deque
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    new_kwargs[name] = recv_from_pipeline_state_buffer(    
+new_kwargs[name] = recv_from_pipeline_state_buffer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+    new_kwargs[name] = recv_from_pipeline_state_buffer(  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+    activation = pipeline_state.activations_buffer.popleft()
+IndexError: pop from an empty deque
+        return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]    hidden_encoder_states = encoder_block(**hidden_encoder_states)return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+    activation = pipeline_state.activations_buffer.popleft()
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+IndexError: pop from an empty deque
+    activation = pipeline_state.activations_buffer.popleft()
+        return self._call_impl(*args, **kwargs)activation = pipeline_state.activations_buffer.popleft()
+
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    IndexErroractivation = pipeline_state.activations_buffer.popleft(): 
+pop from an empty deque    
+activation = pipeline_state.activations_buffer.popleft()
+IndexErrorreturn self._call_impl(*args, **kwargs): 
+pop from an empty deque
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+IndexError    activation = pipeline_state.activations_buffer.popleft(): IndexError
+pop from an empty deque: pop from an empty deque
+
+IndexError: pop from an empty deque
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    activation = pipeline_state.activations_buffer.popleft()
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+IndexError: pop from an empty deque
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+    new_kwargs[name] = recv_from_pipeline_state_buffer(    
+new_kwargs[name] = recv_from_pipeline_state_buffer(
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+new_kwargs[name] = recv_from_pipeline_state_buffer(  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+    activation = pipeline_state.activations_buffer.popleft()
+IndexError: pop from an empty deque
+    activation = pipeline_state.activations_buffer.popleft()
+    activation = pipeline_state.activations_buffer.popleft()
+IndexError: pop from an empty deque
+IndexError: pop from an empty deque
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+    activation = pipeline_state.activations_buffer.popleft()
+IndexError: pop from an empty deque
+    activation = pipeline_state.activations_buffer.popleft()
+IndexError: pop from an empty deque
+    activation = pipeline_state.activations_buffer.popleft()
+IndexError: pop from an empty deque
+ip-26-0-161-123:32709:33215 [0] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-161-123:32706:33213 [0] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-161-123:32708:33214 [0] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-161-123:32705:33217 [0] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-161-123:32710:33218 [0] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-161-123:32704:33220 [0] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-161-123:32704:33176 [0] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-161-123:32709:33139 [0] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-161-123:32708:33142 [0] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-161-123:32706:33141 [0] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-161-123:32705:33145 [0] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-161-123:32710:33138 [0] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-161-123:32704:33144 [0] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-161-123:32708:32897 [0] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-161-123:32709:32902 [0] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-161-123:32706:32900 [0] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-161-123:32705:32904 [0] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-161-123:32710:32898 [0] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-161-123:32704:32903 [0] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-161-123:32710:33218 [0] NCCL INFO [Service thread] Connection closed by localRank 3
+ip-26-0-161-123:32709:33215 [0] NCCL INFO [Service thread] Connection closed by localRank 3
+ip-26-0-161-123:32708:33214 [0] NCCL INFO [Service thread] Connection closed by localRank 3
+ip-26-0-161-123:32705:33217 [0] NCCL INFO [Service thread] Connection closed by localRank 3
+ip-26-0-161-123:32704:33220 [0] NCCL INFO [Service thread] Connection closed by localRank 3
+ip-26-0-161-123:32706:33213 [0] NCCL INFO [Service thread] Connection closed by localRank 3
+ip-26-0-161-123:32704:33176 [0] NCCL INFO [Service thread] Connection closed by localRank 3
+ip-26-0-161-123:32710:33138 [0] NCCL INFO [Service thread] Connection closed by localRank 3
+ip-26-0-161-123:32709:33139 [0] NCCL INFO [Service thread] Connection closed by localRank 3
+ip-26-0-161-123:32708:33142 [0] NCCL INFO [Service thread] Connection closed by localRank 3
+ip-26-0-161-123:32705:33145 [0] NCCL INFO [Service thread] Connection closed by localRank 3
+ip-26-0-161-123:32704:33144 [0] NCCL INFO [Service thread] Connection closed by localRank 3
+ip-26-0-161-123:32706:33141 [0] NCCL INFO [Service thread] Connection closed by localRank 3
+ip-26-0-161-123:32710:32898 [0] NCCL INFO [Service thread] Connection closed by localRank 3
+ip-26-0-161-123:32709:32902 [0] NCCL INFO [Service thread] Connection closed by localRank 3
+ip-26-0-161-123:32708:32897 [0] NCCL INFO [Service thread] Connection closed by localRank 3
+ip-26-0-161-123:32705:32904 [0] NCCL INFO [Service thread] Connection closed by localRank 3
+ip-26-0-161-123:32704:32903 [0] NCCL INFO [Service thread] Connection closed by localRank 3
+ip-26-0-161-123:32706:32900 [0] NCCL INFO [Service thread] Connection closed by localRank 3
+ip-26-0-161-178:2815970:2816485 [0] NCCL INFO [Service thread] Connection closed by localRank 5
+ip-26-0-161-178:2815969:2816487 [0] NCCL INFO [Service thread] Connection closed by localRank 5
+ip-26-0-161-178:2815966:2816490 [0] NCCL INFO [Service thread] Connection closed by localRank 5
+ip-26-0-161-178:2815967:2816488 [0] NCCL INFO [Service thread] Connection closed by localRank 5
+ip-26-0-161-178:2815963:2816492 [0] NCCL INFO [Service thread] Connection closed by localRank 5
+ip-26-0-161-178:2815963:2816453 [0] NCCL INFO [Service thread] Connection closed by localRank 5
+ip-26-0-161-178:2815970:2816413 [0] NCCL INFO [Service thread] Connection closed by localRank 5
+ip-26-0-161-178:2815969:2816417 [0] NCCL INFO [Service thread] Connection closed by localRank 5
+ip-26-0-161-178:2815966:2816415 [0] NCCL INFO [Service thread] Connection closed by localRank 5
+ip-26-0-161-178:2815963:2816411 [0] NCCL INFO [Service thread] Connection closed by localRank 5
+ip-26-0-161-178:2815967:2816416 [0] NCCL INFO [Service thread] Connection closed by localRank 5
+ip-26-0-161-178:2815970:2816173 [0] NCCL INFO [Service thread] Connection closed by localRank 5
+ip-26-0-161-178:2815969:2816174 [0] NCCL INFO [Service thread] Connection closed by localRank 5
+ip-26-0-161-178:2815966:2816177 [0] NCCL INFO [Service thread] Connection closed by localRank 5
+ip-26-0-161-178:2815963:2816176 [0] NCCL INFO [Service thread] Connection closed by localRank 5
+ip-26-0-161-178:2815967:2816170 [0] NCCL INFO [Service thread] Connection closed by localRank 5
+ip-26-0-161-138:96568:97104 [0] NCCL INFO [Service thread] Connection closed by localRank 2
+ip-26-0-161-138:96566:97106 [0] NCCL INFO [Service thread] Connection closed by localRank 2
+ip-26-0-161-138:96564:97111 [0] NCCL INFO [Service thread] Connection closed by localRank 2
+ip-26-0-161-138:96568:97021 [0] NCCL INFO [Service thread] Connection closed by localRank 2
+ip-26-0-161-138:96566:97027 [0] NCCL INFO [Service thread] Connection closed by localRank 2
+ip-26-0-161-138:96564:97028 [0] NCCL INFO [Service thread] Connection closed by localRank 2
+ip-26-0-161-138:96568:96777 [0] NCCL INFO [Service thread] Connection closed by localRank 2
+ip-26-0-161-138:96566:96778 [0] NCCL INFO [Service thread] Connection closed by localRank 2
+ip-26-0-161-138:96564:96782 [0] NCCL INFO [Service thread] Connection closed by localRank 2
+ip-26-0-161-178:2815963:2816492 [0] NCCL INFO [Service thread] Connection closed by localRank 1
+ip-26-0-161-178:2815963:2816453 [0] NCCL INFO [Service thread] Connection closed by localRank 1
+ip-26-0-161-138:96566:97106 [0] NCCL INFO [Service thread] Connection closed by localRank 6
+ip-26-0-161-138:96566:97027 [0] NCCL INFO [Service thread] Connection closed by localRank 6
+ip-26-0-161-138:96566:96778 [0] NCCL INFO [Service thread] Connection closed by localRank 6
+ip-26-0-160-242:1364906:1365421 [0] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-160-242:1364905:1365420 [0] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-160-242:1364903:1365425 [0] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-160-242:1364908:1365422 [0] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-160-242:1364902:1365423 [0] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-160-242:1364902:1365388 [0] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-160-242:1364906:1365348 [0] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-160-242:1364905:1365346 [0] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-160-242:1364903:1365351 [0] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-160-242:1364908:1365345 [0] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-160-242:1364902:1365347 [0] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-160-242:1364905:1365111 [0] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-160-242:1364906:1365110 [0] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-160-242:1364903:1365105 [0] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-160-242:1364908:1365106 [0] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-160-242:1364902:1365104 [0] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-161-142:62886:63395 [0] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-161-142:62885:63394 [0] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-161-142:62883:63391 [0] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-161-142:62882:63396 [0] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-161-142:62881:63397 [0] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-161-142:62881:63357 [0] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-161-142:62886:63316 [0] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-161-142:62885:63320 [0] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-161-142:62883:63318 [0] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-161-142:62882:63322 [0] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-161-142:62881:63321 [0] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-161-142:62886:63077 [0] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-161-142:62883:63081 [0] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-161-142:62885:63082 [0] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-161-142:62882:63080 [0] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-161-142:62881:63078 [0] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-160-242:1364905:1365420 [0] NCCL INFO [Service thread] Connection closed by localRank 5
+ip-26-0-160-242:1364903:1365425 [0] NCCL INFO [Service thread] Connection closed by localRank 5
+ip-26-0-160-242:1364902:1365423 [0] NCCL INFO [Service thread] Connection closed by localRank 5
+ip-26-0-160-242:1364906:1365421 [0] NCCL INFO [Service thread] Connection closed by localRank 5
+ip-26-0-160-242:1364902:1365388 [0] NCCL INFO [Service thread] Connection closed by localRank 5
+ip-26-0-160-242:1364905:1365346 [0] NCCL INFO [Service thread] Connection closed by localRank 5
+ip-26-0-160-242:1364903:1365351 [0] NCCL INFO [Service thread] Connection closed by localRank 5
+ip-26-0-160-242:1364902:1365347 [0] NCCL INFO [Service thread] Connection closed by localRank 5
+ip-26-0-160-242:1364906:1365348 [0] NCCL INFO [Service thread] Connection closed by localRank 5
+ip-26-0-160-242:1364905:1365111 [0] NCCL INFO [Service thread] Connection closed by localRank 5
+ip-26-0-160-242:1364903:1365105 [0] NCCL INFO [Service thread] Connection closed by localRank 5
+ip-26-0-160-242:1364902:1365104 [0] NCCL INFO [Service thread] Connection closed by localRank 5
+ip-26-0-160-242:1364906:1365110 [0] NCCL INFO [Service thread] Connection closed by localRank 5
+ip-26-0-160-242:1364905:1365420 [0] NCCL INFO [Service thread] Connection closed by localRank 6
+ip-26-0-160-242:1364906:1365421 [0] NCCL INFO [Service thread] Connection closed by localRank 6
+ip-26-0-160-242:1364906:1365348 [0] NCCL INFO [Service thread] Connection closed by localRank 6
+ip-26-0-160-242:1364905:1365346 [0] NCCL INFO [Service thread] Connection closed by localRank 6
+ip-26-0-160-242:1364906:1365110 [0] NCCL INFO [Service thread] Connection closed by localRank 6
+ip-26-0-160-242:1364905:1365111 [0] NCCL INFO [Service thread] Connection closed by localRank 6
+ip-26-0-160-242:1364906:1365421 [0] NCCL INFO [Service thread] Connection closed by localRank 2
+ip-26-0-160-242:1364906:1365348 [0] NCCL INFO [Service thread] Connection closed by localRank 2
+ip-26-0-160-242:1364906:1365110 [0] NCCL INFO [Service thread] Connection closed by localRank 2
+ip-26-0-161-142:62881:63397 [0] NCCL INFO [Service thread] Connection closed by localRank 6
+ip-26-0-161-142:62881:63357 [0] NCCL INFO [Service thread] Connection closed by localRank 6
+ip-26-0-161-103:96893:97397 [0] NCCL INFO [Service thread] Connection closed by localRank 4
+ip-26-0-161-103:96890:97402 [0] NCCL INFO [Service thread] Connection closed by localRank 4
+ip-26-0-161-103:96889:97401 [0] NCCL INFO [Service thread] Connection closed by localRank 4
+ip-26-0-161-103:96888:97403 [0] NCCL INFO [Service thread] Connection closed by localRank 4
+ip-26-0-161-103:96895:97398 [0] NCCL INFO [Service thread] Connection closed by localRank 4
+ip-26-0-161-103:96891:97404 [0] NCCL INFO [Service thread] Connection closed by localRank 4
+ip-26-0-161-103:96888:97363 [0] NCCL INFO [Service thread] Connection closed by localRank 4
+ip-26-0-161-103:96893:97323 [0] NCCL INFO [Service thread] Connection closed by localRank 4
+ip-26-0-161-103:96889:97328 [0] NCCL INFO [Service thread] Connection closed by localRank 4
+ip-26-0-161-103:96890:97329 [0] NCCL INFO [Service thread] Connection closed by localRank 4
+ip-26-0-161-103:96888:97327 [0] NCCL INFO [Service thread] Connection closed by localRank 4
+ip-26-0-161-103:96895:97326 [0] NCCL INFO [Service thread] Connection closed by localRank 4
+ip-26-0-161-103:96891:97330 [0] NCCL INFO [Service thread] Connection closed by localRank 4
+ip-26-0-161-103:96893:97083 [0] NCCL INFO [Service thread] Connection closed by localRank 4
+ip-26-0-161-103:96890:97090 [0] NCCL INFO [Service thread] Connection closed by localRank 4
+ip-26-0-161-103:96889:97087 [0] NCCL INFO [Service thread] Connection closed by localRank 4
+ip-26-0-161-103:96888:97084 [0] NCCL INFO [Service thread] Connection closed by localRank 4
+ip-26-0-161-103:96895:97086 [0] NCCL INFO [Service thread] Connection closed by localRank 4
+ip-26-0-161-103:96891:97089 [0] NCCL INFO [Service thread] Connection closed by localRank 4
+ip-26-0-161-103:96891:97404 [0] NCCL INFO [Service thread] Connection closed by localRank 6
+ip-26-0-161-103:96893:97397 [0] NCCL INFO [Service thread] Connection closed by localRank 6
+ip-26-0-161-103:96891:97330 [0] NCCL INFO [Service thread] Connection closed by localRank 6
+ip-26-0-161-103:96893:97323 [0] NCCL INFO [Service thread] Connection closed by localRank 6
+ip-26-0-161-103:96891:97089 [0] NCCL INFO [Service thread] Connection closed by localRank 6
+ip-26-0-161-103:96893:97083 [0] NCCL INFO [Service thread] Connection closed by localRank 6
+ip-26-0-161-103:96893:97397 [0] NCCL INFO [Service thread] Connection closed by localRank 2
+ip-26-0-161-103:96893:97323 [0] NCCL INFO [Service thread] Connection closed by localRank 2
+ip-26-0-161-103:96893:97083 [0] NCCL INFO [Service thread] Connection closed by localRank 2
+ip-26-0-161-103:96893:97397 [0] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-161-103:96893:97323 [0] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-161-103:96893:97083 [0] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-161-78:114517:115023 [0] NCCL INFO [Service thread] Connection closed by localRank 4
+ip-26-0-161-78:114514:115026 [0] NCCL INFO [Service thread] Connection closed by localRank 4
+ip-26-0-161-78:114513:115029 [0] NCCL INFO [Service thread] Connection closed by localRank 4
+ip-26-0-161-78:114512:115027 [0] NCCL INFO [Service thread] Connection closed by localRank 4
+ip-26-0-161-78:114519:115022 [0] NCCL INFO [Service thread] Connection closed by localRank 4
+ip-26-0-161-78:114512:114986 [0] NCCL INFO [Service thread] Connection closed by localRank 4
+ip-26-0-161-78:114517:114950 [0] NCCL INFO [Service thread] Connection closed by localRank 4
+ip-26-0-161-78:114514:114952 [0] NCCL INFO [Service thread] Connection closed by localRank 4
+ip-26-0-161-78:114513:114953 [0] NCCL INFO [Service thread] Connection closed by localRank 4
+ip-26-0-161-78:114512:114951 [0] NCCL INFO [Service thread] Connection closed by localRank 4
+ip-26-0-161-78:114519:114948 [0] NCCL INFO [Service thread] Connection closed by localRank 4
+ip-26-0-161-78:114517:114709 [0] NCCL INFO [Service thread] Connection closed by localRank 4
+ip-26-0-161-78:114514:114711 [0] NCCL INFO [Service thread] Connection closed by localRank 4
+ip-26-0-161-78:114513:114714 [0] NCCL INFO [Service thread] Connection closed by localRank 4
+ip-26-0-161-78:114519:114707 [0] NCCL INFO [Service thread] Connection closed by localRank 4
+ip-26-0-161-78:114512:114712 [0] NCCL INFO [Service thread] Connection closed by localRank 4
+ip-26-0-161-78:114517:115023 [0] NCCL INFO [Service thread] Connection closed by localRank 3
+ip-26-0-161-78:114513:115029 [0] NCCL INFO [Service thread] Connection closed by localRank 3
+ip-26-0-161-78:114512:115027 [0] NCCL INFO [Service thread] Connection closed by localRank 3
+ip-26-0-161-78:114512:114986 [0] NCCL INFO [Service thread] Connection closed by localRank 3
+ip-26-0-161-78:114517:114950 [0] NCCL INFO [Service thread] Connection closed by localRank 3
+ip-26-0-161-78:114513:114953 [0] NCCL INFO [Service thread] Connection closed by localRank 3
+ip-26-0-161-78:114512:114951 [0] NCCL INFO [Service thread] Connection closed by localRank 3
+ip-26-0-161-78:114517:114709 [0] NCCL INFO [Service thread] Connection closed by localRank 3
+ip-26-0-161-78:114513:114714 [0] NCCL INFO [Service thread] Connection closed by localRank 3
+ip-26-0-161-78:114512:114712 [0] NCCL INFO [Service thread] Connection closed by localRank 3
+ip-26-0-161-78:114513:115029 [0] NCCL INFO [Service thread] Connection closed by localRank 6
+ip-26-0-161-78:114512:115027 [0] NCCL INFO [Service thread] Connection closed by localRank 6
+ip-26-0-161-78:114517:115023 [0] NCCL INFO [Service thread] Connection closed by localRank 6
+ip-26-0-161-78:114512:114986 [0] NCCL INFO [Service thread] Connection closed by localRank 6
+ip-26-0-161-78:114513:114953 [0] NCCL INFO [Service thread] Connection closed by localRank 6
+ip-26-0-161-78:114512:114951 [0] NCCL INFO [Service thread] Connection closed by localRank 6
+ip-26-0-161-78:114517:114950 [0] NCCL INFO [Service thread] Connection closed by localRank 6
+ip-26-0-161-78:114513:114714 [0] NCCL INFO [Service thread] Connection closed by localRank 6
+ip-26-0-161-78:114512:114712 [0] NCCL INFO [Service thread] Connection closed by localRank 6
+ip-26-0-161-78:114517:114709 [0] NCCL INFO [Service thread] Connection closed by localRank 6
+ip-26-0-161-78:114517:115023 [0] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-161-78:114517:114950 [0] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-161-78:114517:114709 [0] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-168-30:235652:236162 [0] NCCL INFO [Service thread] Connection closed by localRank 6
+ip-26-0-168-30:235648:236159 [0] NCCL INFO [Service thread] Connection closed by localRank 6
+ip-26-0-168-30:235653:236163 [0] NCCL INFO [Service thread] Connection closed by localRank 6
+ip-26-0-168-30:235648:236122 [0] NCCL INFO [Service thread] Connection closed by localRank 6
+ip-26-0-168-30:235652:236082 [0] NCCL INFO [Service thread] Connection closed by localRank 6
+ip-26-0-168-30:235653:236081 [0] NCCL INFO [Service thread] Connection closed by localRank 6
+ip-26-0-168-30:235648:236084 [0] NCCL INFO [Service thread] Connection closed by localRank 6
+ip-26-0-168-30:235652:235843 [0] NCCL INFO [Service thread] Connection closed by localRank 6
+ip-26-0-168-30:235648:235842 [0] NCCL INFO [Service thread] Connection closed by localRank 6
+ip-26-0-168-30:235653:235847 [0] NCCL INFO [Service thread] Connection closed by localRank 6
+ip-26-0-168-30:235652:236162 [0] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-168-30:235652:236082 [0] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-168-34:272157:272666 [0] NCCL INFO [Service thread] Connection closed by localRank 6
+ip-26-0-168-34:272154:272661 [0] NCCL INFO [Service thread] Connection closed by localRank 6
+ip-26-0-168-34:272151:272667 [0] NCCL INFO [Service thread] Connection closed by localRank 6
+ip-26-0-168-34:272155:272663 [0] NCCL INFO [Service thread] Connection closed by localRank 6
+ip-26-0-168-34:272154:272587 [0] NCCL INFO [Service thread] Connection closed by localRank 6
+ip-26-0-168-34:272157:272592 [0] NCCL INFO [Service thread] Connection closed by localRank 6
+ip-26-0-168-34:272151:272585 [0] NCCL INFO [Service thread] Connection closed by localRank 6
+ip-26-0-168-34:272155:272591 [0] NCCL INFO [Service thread] Connection closed by localRank 6
+ip-26-0-168-34:272154:272347 [0] NCCL INFO [Service thread] Connection closed by localRank 6
+ip-26-0-168-34:272157:272346 [0] NCCL INFO [Service thread] Connection closed by localRank 6
+ip-26-0-168-34:272151:272350 [0] NCCL INFO [Service thread] Connection closed by localRank 6
+ip-26-0-168-34:272155:272349 [0] NCCL INFO [Service thread] Connection closed by localRank 6
+[2024-12-28 11:56:32,422] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2815963 closing signal SIGTERM
+[2024-12-28 11:56:32,423] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2815964 closing signal SIGTERM
+[2024-12-28 11:56:32,423] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2815965 closing signal SIGTERM
+[2024-12-28 11:56:32,423] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2815966 closing signal SIGTERM
+[2024-12-28 11:56:32,423] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2815967 closing signal SIGTERM
+[2024-12-28 11:56:32,423] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2815969 closing signal SIGTERM
+[2024-12-28 11:56:32,423] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2815970 closing signal SIGTERM
+[2024-12-28 11:56:32,425] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2722046 closing signal SIGTERM
+[2024-12-28 11:56:32,426] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2722047 closing signal SIGTERM
+[2024-12-28 11:56:32,426] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2722048 closing signal SIGTERM
+[2024-12-28 11:56:32,426] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2722049 closing signal SIGTERM
+[2024-12-28 11:56:32,426] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2722050 closing signal SIGTERM
+[2024-12-28 11:56:32,426] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2722051 closing signal SIGTERM
+[2024-12-28 11:56:32,426] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2722052 closing signal SIGTERM
+[2024-12-28 11:56:32,427] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 32704 closing signal SIGTERM
+[2024-12-28 11:56:32,428] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 32705 closing signal SIGTERM
+[2024-12-28 11:56:32,428] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 32706 closing signal SIGTERM
+[2024-12-28 11:56:32,428] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 32707 closing signal SIGTERM
+[2024-12-28 11:56:32,428] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 32708 closing signal SIGTERM
+[2024-12-28 11:56:32,428] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 32709 closing signal SIGTERM
+[2024-12-28 11:56:32,428] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 32710 closing signal SIGTERM
+[2024-12-28 11:56:32,430] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 124161 closing signal SIGTERM
+[2024-12-28 11:56:32,430] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 124162 closing signal SIGTERM
+[2024-12-28 11:56:32,430] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 124163 closing signal SIGTERM
+[2024-12-28 11:56:32,430] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 124164 closing signal SIGTERM
+[2024-12-28 11:56:32,430] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 124166 closing signal SIGTERM
+[2024-12-28 11:56:33,559] torch.distributed.elastic.multiprocessing.api: [ERROR] failed (exitcode: 1) local_rank: 0 (pid: 124160) of binary: /fsx/nouamane/miniconda/envs/2-1-cu121/bin/python
+[2024-12-28 11:56:33,601] torch.distributed.elastic.multiprocessing.errors.error_handler: [ERROR] no error file defined for parent, to copy child error file (/tmp/torchelastic_3p9kp7cj/13849095_eogth9d3/attempt_0/0/error.json)
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 264, in launch_agent
+    raise ChildFailedError(
+torch.distributed.elastic.multiprocessing.errors.ChildFailedError: 
+============================================================
+/fsx/nouamane/projects/nanotron/run_train.py FAILED
+------------------------------------------------------------
+Failures:
+[1]:
+  time      : 2024-12-28_11:56:08
+  host      : ip-26-0-160-225.ec2.internal
+  rank      : 5 (local_rank: 5)
+  exitcode  : 1 (pid: 124165)
+  error_file: /tmp/torchelastic_3p9kp7cj/13849095_eogth9d3/attempt_0/5/error.json
+  traceback : Traceback (most recent call last):
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+      return f(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+      outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+      outputs = self.pipeline_engine.train_batch_iter(
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+      output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+      output = model(**micro_batch)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+      sharded_logits = self.model(
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+      return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 896, in forward_with_hidden_states
+      output = self.token_position_embeddings(input_ids=input_ids, input_mask=input_mask)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 160, in forward
+      output = self.pp_block(**new_kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 786, in forward
+      input_embeds = self.token_embedding(input_ids)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 290, in forward
+      out = out * (~input_mask[..., None])
+  torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 8.00 GiB. GPU 5 has a total capacty of 79.33 GiB of which 4.98 GiB is free. Including non-PyTorch memory, this process has 74.34 GiB memory in use. Of the allocated memory 63.52 GiB is allocated by PyTorch, and 97.18 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  
+[2]:
+  time      : 2024-12-28_11:56:08
+  host      : ip-26-0-160-225.ec2.internal
+  rank      : 7 (local_rank: 7)
+  exitcode  : 1 (pid: 124167)
+  error_file: /tmp/torchelastic_3p9kp7cj/13849095_eogth9d3/attempt_0/7/error.json
+  traceback : Traceback (most recent call last):
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+      return f(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+      outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+      outputs = self.pipeline_engine.train_batch_iter(
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+      output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+      output = model(**micro_batch)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+      sharded_logits = self.model(
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+      return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 896, in forward_with_hidden_states
+      output = self.token_position_embeddings(input_ids=input_ids, input_mask=input_mask)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 160, in forward
+      output = self.pp_block(**new_kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 786, in forward
+      input_embeds = self.token_embedding(input_ids)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 290, in forward
+      out = out * (~input_mask[..., None])
+  torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 8.00 GiB. GPU 7 has a total capacty of 79.33 GiB of which 5.15 GiB is free. Including non-PyTorch memory, this process has 74.17 GiB memory in use. Of the allocated memory 63.52 GiB is allocated by PyTorch, and 97.18 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  
+------------------------------------------------------------
+Root Cause (first observed failure):
+[0]:
+  time      : 2024-12-28_11:56:08
+  host      : ip-26-0-160-225.ec2.internal
+  rank      : 0 (local_rank: 0)
+  exitcode  : 1 (pid: 124160)
+  error_file: /tmp/torchelastic_3p9kp7cj/13849095_eogth9d3/attempt_0/0/error.json
+  traceback : Traceback (most recent call last):
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+      return f(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+      outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+      outputs = self.pipeline_engine.train_batch_iter(
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+      output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+      output = model(**micro_batch)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+      sharded_logits = self.model(
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+      return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 896, in forward_with_hidden_states
+      output = self.token_position_embeddings(input_ids=input_ids, input_mask=input_mask)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 160, in forward
+      output = self.pp_block(**new_kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 786, in forward
+      input_embeds = self.token_embedding(input_ids)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 290, in forward
+      out = out * (~input_mask[..., None])
+  torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 8.00 GiB. GPU 0 has a total capacty of 79.33 GiB of which 5.00 GiB is free. Including non-PyTorch memory, this process has 74.32 GiB memory in use. Of the allocated memory 63.52 GiB is allocated by PyTorch, and 97.18 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  
+============================================================
+srun: error: ip-26-0-160-225: task 0: Exited with exit code 1
+srun: launch/slurm: _step_signal: Terminating StepId=13849095.0
+[2024-12-28 11:56:34,461] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-28 11:56:34,461] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 96564 closing signal SIGTERM
+[2024-12-28 11:56:34,462] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 96566 closing signal SIGTERM
+[2024-12-28 11:56:34,462] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 96567 closing signal SIGTERM
+[2024-12-28 11:56:34,462] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 96568 closing signal SIGTERM
+[2024-12-28 11:56:34,462] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 96569 closing signal SIGTERM
+[2024-12-28 11:56:34,462] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 96570 closing signal SIGTERM
+[2024-12-28 11:56:34,462] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-28 11:56:34,462] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-28 11:56:34,461] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-28 11:56:34,463] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1630205 closing signal SIGTERM
+[2024-12-28 11:56:34,463] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2815965 closing signal SIGTERM
+[2024-12-28 11:56:34,463] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1630206 closing signal SIGTERM
+[2024-12-28 11:56:34,463] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1630207 closing signal SIGTERM
+[2024-12-28 11:56:34,463] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-28 11:56:34,461] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 62881 closing signal SIGTERM
+[2024-12-28 11:56:34,463] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2815966 closing signal SIGTERM
+[2024-12-28 11:56:34,463] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1630208 closing signal SIGTERM
+[2024-12-28 11:56:34,463] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1630209 closing signal SIGTERM
+[2024-12-28 11:56:34,461] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 62882 closing signal SIGTERM
+[2024-12-28 11:56:34,463] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2815967 closing signal SIGTERM
+[2024-12-28 11:56:34,463] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 114512 closing signal SIGTERM
+[2024-12-28 11:56:34,463] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 114513 closing signal SIGTERM
+[2024-12-28 11:56:34,463] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1630210 closing signal SIGTERM
+[2024-12-28 11:56:34,463] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1630211 closing signal SIGTERM
+[2024-12-28 11:56:34,461] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 62883 closing signal SIGTERM
+[2024-12-28 11:56:34,463] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2815969 closing signal SIGTERM
+[2024-12-28 11:56:34,461] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 62884 closing signal SIGTERM
+[2024-12-28 11:56:34,463] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 114514 closing signal SIGTERM
+[2024-12-28 11:56:34,463] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 114515 closing signal SIGTERM
+[2024-12-28 11:56:34,464] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-28 11:56:34,463] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2815970 closing signal SIGTERM
+[2024-12-28 11:56:34,463] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1630212 closing signal SIGTERM
+[2024-12-28 11:56:34,461] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 62885 closing signal SIGTERM
+[2024-12-28 11:56:34,463] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 114517 closing signal SIGTERM
+[2024-12-28 11:56:34,461] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 62886 closing signal SIGTERM
+[2024-12-28 11:56:34,462] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-28 11:56:34,464] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1227463 closing signal SIGTERM
+[2024-12-28 11:56:34,463] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-28 11:56:34,461] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 62887 closing signal SIGTERM
+[2024-12-28 11:56:34,462] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-28 11:56:34,464] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1227464 closing signal SIGTERM
+[2024-12-28 11:56:34,463] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 114518 closing signal SIGTERM
+[2024-12-28 11:56:34,464] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1227465 closing signal SIGTERM
+[2024-12-28 11:56:34,463] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 153562 closing signal SIGTERM
+[2024-12-28 11:56:34,463] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 114519 closing signal SIGTERM
+[2024-12-28 11:56:34,464] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1227466 closing signal SIGTERM
+[2024-12-28 11:56:34,462] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 62888 closing signal SIGTERM
+[2024-12-28 11:56:34,463] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-28 11:56:34,463] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1557267 closing signal SIGTERM
+[2024-12-28 11:56:34,463] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 32705 closing signal SIGTERM
+[2024-12-28 11:56:34,464] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1227467 closing signal SIGTERM
+[2024-12-28 11:56:34,463] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 153563 closing signal SIGTERM
+[2024-12-28 11:56:34,463] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 32706 closing signal SIGTERM
+[2024-12-28 11:56:34,464] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1227468 closing signal SIGTERM
+[2024-12-28 11:56:34,463] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 153564 closing signal SIGTERM
+[2024-12-28 11:56:34,462] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-28 11:56:34,464] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 153565 closing signal SIGTERM
+[2024-12-28 11:56:34,461] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-28 11:56:34,463] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 32708 closing signal SIGTERM
+[2024-12-28 11:56:34,463] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2636084 closing signal SIGTERM
+[2024-12-28 11:56:34,464] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1227469 closing signal SIGTERM
+[2024-12-28 11:56:34,463] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2636085 closing signal SIGTERM
+[2024-12-28 11:56:34,463] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1557268 closing signal SIGTERM
+[2024-12-28 11:56:34,463] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 32709 closing signal SIGTERM
+[2024-12-28 11:56:34,464] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1227470 closing signal SIGTERM
+[2024-12-28 11:56:34,462] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 235648 closing signal SIGTERM
+[2024-12-28 11:56:34,463] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2636086 closing signal SIGTERM
+[2024-12-28 11:56:34,464] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 153566 closing signal SIGTERM
+[2024-12-28 11:56:34,462] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 272150 closing signal SIGTERM
+[2024-12-28 11:56:34,463] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2636087 closing signal SIGTERM
+[2024-12-28 11:56:34,464] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 153567 closing signal SIGTERM
+[2024-12-28 11:56:34,463] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-28 11:56:34,464] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 153568 closing signal SIGTERM
+[2024-12-28 11:56:34,463] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 32710 closing signal SIGTERM
+[2024-12-28 11:56:34,463] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2636088 closing signal SIGTERM
+[2024-12-28 11:56:34,463] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1557269 closing signal SIGTERM
+[2024-12-28 11:56:34,462] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 235649 closing signal SIGTERM
+[2024-12-28 11:56:34,462] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 272151 closing signal SIGTERM
+[2024-12-28 11:56:34,463] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2636089 closing signal SIGTERM
+[2024-12-28 11:56:34,462] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 235650 closing signal SIGTERM
+[2024-12-28 11:56:34,462] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 272152 closing signal SIGTERM
+[2024-12-28 11:56:34,462] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 235651 closing signal SIGTERM
+[2024-12-28 11:56:34,464] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 153569 closing signal SIGTERM
+[2024-12-28 11:56:34,462] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 272153 closing signal SIGTERM
+[2024-12-28 11:56:34,463] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2636090 closing signal SIGTERM
+[2024-12-28 11:56:34,463] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 96888 closing signal SIGTERM
+[2024-12-28 11:56:34,463] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2636091 closing signal SIGTERM
+[2024-12-28 11:56:34,463] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1557271 closing signal SIGTERM
+[2024-12-28 11:56:34,463] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 96889 closing signal SIGTERM
+[2024-12-28 11:56:34,463] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 235652 closing signal SIGTERM
+[2024-12-28 11:56:34,463] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 235653 closing signal SIGTERM
+[2024-12-28 11:56:34,462] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 272154 closing signal SIGTERM
+[2024-12-28 11:56:34,463] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 235654 closing signal SIGTERM
+[2024-12-28 11:56:34,462] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 272155 closing signal SIGTERM
+[2024-12-28 11:56:34,463] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 96890 closing signal SIGTERM
+[2024-12-28 11:56:34,462] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 272156 closing signal SIGTERM
+[2024-12-28 11:56:34,463] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 96891 closing signal SIGTERM
+[2024-12-28 11:56:34,463] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1557272 closing signal SIGTERM
+[2024-12-28 11:56:34,463] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 96892 closing signal SIGTERM
+[2024-12-28 11:56:34,463] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1557273 closing signal SIGTERM
+[2024-12-28 11:56:34,463] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 235655 closing signal SIGTERM
+[2024-12-28 11:56:34,463] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 96893 closing signal SIGTERM
+[2024-12-28 11:56:34,462] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 272157 closing signal SIGTERM
+[2024-12-28 11:56:34,463] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 96894 closing signal SIGTERM
+[2024-12-28 11:56:34,463] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2024-12-28 11:56:34,463] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 96895 closing signal SIGTERM
+[2024-12-28 11:56:34,463] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2722046 closing signal SIGTERM
+[2024-12-28 11:56:34,463] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2722047 closing signal SIGTERM
+[2024-12-28 11:56:34,463] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2722048 closing signal SIGTERM
+[2024-12-28 11:56:34,463] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2722051 closing signal SIGTERM
+[2024-12-28 11:56:34,463] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2722052 closing signal SIGTERM
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 878, in _invoke_run
+    run_result = self._monitor_workers(self._worker_group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/local_elastic_agent.py", line 307, in _monitor_workers
+    result = self._pcontext.wait(0)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 288, in wait
+    return self._poll()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 668, in _poll
+    self.close()  # terminate all running procs
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 331, in close
+    self._close(death_sig=death_sig, timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 713, in _close
+    handler.proc.wait(time_to_wait)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1209, in wait
+    return self._wait(timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1953, in _wait
+    time.sleep(delay)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 2721970 got signal: 15
+srun: error: ip-26-0-168-52: task 15: Exited with exit code 1
+[2024-12-28 11:56:36,441] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-161-142.ec2.internal_62807_0' has failed to send a keep-alive heartbeat to the rendezvous '13849095' due to an error of type RendezvousConnectionError.
+[2024-12-28 11:56:36,469] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-161-153.ec2.internal_2636009_0' has failed to send a keep-alive heartbeat to the rendezvous '13849095' due to an error of type RendezvousConnectionError.
+[2024-12-28 11:56:37,203] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-161-78.ec2.internal_114438_0' has failed to send a keep-alive heartbeat to the rendezvous '13849095' due to an error of type RendezvousConnectionError.
+[2024-12-28 11:56:37,209] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-161-123.ec2.internal_32628_0' has failed to send a keep-alive heartbeat to the rendezvous '13849095' due to an error of type RendezvousConnectionError.
+[2024-12-28 11:56:37,215] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-165-131.ec2.internal_1630127_0' has failed to send a keep-alive heartbeat to the rendezvous '13849095' due to an error of type RendezvousConnectionError.
+[2024-12-28 11:56:37,233] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-165-59.ec2.internal_1557190_0' has failed to send a keep-alive heartbeat to the rendezvous '13849095' due to an error of type RendezvousConnectionError.
+[2024-12-28 11:56:37,247] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-168-30.ec2.internal_235573_0' has failed to send a keep-alive heartbeat to the rendezvous '13849095' due to an error of type RendezvousConnectionError.
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 878, in _invoke_run
+    run_result = self._monitor_workers(self._worker_group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/local_elastic_agent.py", line 307, in _monitor_workers
+    result = self._pcontext.wait(0)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 288, in wait
+    return self._poll()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 668, in _poll
+    self.close()  # terminate all running procs
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 331, in close
+    self._close(death_sig=death_sig, timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 713, in _close
+    handler.proc.wait(time_to_wait)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1209, in wait
+    return self._wait(timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1953, in _wait
+    time.sleep(delay)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 32628 got signal: 15
+[2024-12-28 11:56:37,285] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-161-178.ec2.internal_2815889_0' has failed to send a keep-alive heartbeat to the rendezvous '13849095' due to an error of type RendezvousConnectionError.
+[2024-12-28 11:56:37,311] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-168-34.ec2.internal_272077_0' has failed to send a keep-alive heartbeat to the rendezvous '13849095' due to an error of type RendezvousConnectionError.
+[2024-12-28 11:56:37,349] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-161-138.ec2.internal_96487_0' has failed to send a keep-alive heartbeat to the rendezvous '13849095' due to an error of type RendezvousConnectionError.
+[2024-12-28 11:56:37,352] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-160-242.ec2.internal_1364827_0' has failed to send a keep-alive heartbeat to the rendezvous '13849095' due to an error of type RendezvousConnectionError.
+[2024-12-28 11:56:37,379] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-161-103.ec2.internal_96814_0' has failed to send a keep-alive heartbeat to the rendezvous '13849095' due to an error of type RendezvousConnectionError.
+[2024-12-28 11:56:37,381] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-165-202.ec2.internal_153487_0' has failed to send a keep-alive heartbeat to the rendezvous '13849095' due to an error of type RendezvousConnectionError.
+[2024-12-28 11:56:37,387] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-165-164.ec2.internal_1227389_0' has failed to send a keep-alive heartbeat to the rendezvous '13849095' due to an error of type RendezvousConnectionError.
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 878, in _invoke_run
+    run_result = self._monitor_workers(self._worker_group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/local_elastic_agent.py", line 307, in _monitor_workers
+    result = self._pcontext.wait(0)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 288, in wait
+    return self._poll()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 668, in _poll
+    self.close()  # terminate all running procs
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 331, in close
+    self._close(death_sig=death_sig, timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 713, in _close
+    handler.proc.wait(time_to_wait)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1209, in wait
+    return self._wait(timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1953, in _wait
+    time.sleep(delay)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 2815889 got signal: 15
+[2024-12-28 11:56:37,420] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1364903 closing signal SIGTERM
+[2024-12-28 11:56:37,420] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1364904 closing signal SIGTERM
+[2024-12-28 11:56:37,420] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1364905 closing signal SIGTERM
+[2024-12-28 11:56:37,420] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1364906 closing signal SIGTERM
+[2024-12-28 11:56:37,420] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1364908 closing signal SIGTERM
+srun: error: ip-26-0-161-123: task 4: Exited with exit code 1
+srun: error: ip-26-0-161-178: task 8: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 96487 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 1557190 got signal: 15
+srun: error: ip-26-0-161-138: task 5: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 114438 got signal: 15
+srun: error: ip-26-0-161-78: task 2: Exited with exit code 1
+srun: error: ip-26-0-165-59: task 9: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 62807 got signal: 15
+srun: error: ip-26-0-161-142: task 6: Exited with exit code 1
+[2024-12-28 11:56:39,901] torch.distributed.elastic.multiprocessing.api: [ERROR] failed (exitcode: 1) local_rank: 0 (pid: 1364902) of binary: /fsx/nouamane/miniconda/envs/2-1-cu121/bin/python
+[2024-12-28 11:56:39,984] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-160-242.ec2.internal_1364827_0' has failed to shutdown the rendezvous '13849095' due to an error of type RendezvousConnectionError.
+[2024-12-28 11:56:39,984] torch.distributed.elastic.multiprocessing.errors.error_handler: [ERROR] no error file defined for parent, to copy child error file (/tmp/torchelastic_iw__ftcn/13849095_k3rbzj4l/attempt_0/0/error.json)
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 264, in launch_agent
+    raise ChildFailedError(
+torch.distributed.elastic.multiprocessing.errors.ChildFailedError: 
+============================================================
+/fsx/nouamane/projects/nanotron/run_train.py FAILED
+------------------------------------------------------------
+Failures:
+[1]:
+  time      : 2024-12-28_11:56:08
+  host      : ip-26-0-160-242.ec2.internal
+  rank      : 13 (local_rank: 5)
+  exitcode  : 1 (pid: 1364907)
+  error_file: /tmp/torchelastic_iw__ftcn/13849095_k3rbzj4l/attempt_0/5/error.json
+  traceback : Traceback (most recent call last):
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+      return f(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+      outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+      outputs = self.pipeline_engine.train_batch_iter(
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+      output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+      output = model(**micro_batch)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+      sharded_logits = self.model(
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+      return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 896, in forward_with_hidden_states
+      output = self.token_position_embeddings(input_ids=input_ids, input_mask=input_mask)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 160, in forward
+      output = self.pp_block(**new_kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 786, in forward
+      input_embeds = self.token_embedding(input_ids)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 290, in forward
+      out = out * (~input_mask[..., None])
+  torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 8.00 GiB. GPU 5 has a total capacty of 79.33 GiB of which 4.10 GiB is free. Including non-PyTorch memory, this process has 75.22 GiB memory in use. Of the allocated memory 63.52 GiB is allocated by PyTorch, and 97.18 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  
+[2]:
+  time      : 2024-12-28_11:56:08
+  host      : ip-26-0-160-242.ec2.internal
+  rank      : 15 (local_rank: 7)
+  exitcode  : 1 (pid: 1364909)
+  error_file: /tmp/torchelastic_iw__ftcn/13849095_k3rbzj4l/attempt_0/7/error.json
+  traceback : Traceback (most recent call last):
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+      return f(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+      outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+      outputs = self.pipeline_engine.train_batch_iter(
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+      output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+      output = model(**micro_batch)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+      sharded_logits = self.model(
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+      return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 896, in forward_with_hidden_states
+      output = self.token_position_embeddings(input_ids=input_ids, input_mask=input_mask)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 160, in forward
+      output = self.pp_block(**new_kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 786, in forward
+      input_embeds = self.token_embedding(input_ids)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 290, in forward
+      out = out * (~input_mask[..., None])
+  torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 8.00 GiB. GPU 7 has a total capacty of 79.33 GiB of which 4.27 GiB is free. Including non-PyTorch memory, this process has 75.04 GiB memory in use. Of the allocated memory 63.52 GiB is allocated by PyTorch, and 97.18 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  
+------------------------------------------------------------
+Root Cause (first observed failure):
+[0]:
+  time      : 2024-12-28_11:56:08
+  host      : ip-26-0-160-242.ec2.internal
+  rank      : 8 (local_rank: 0)
+  exitcode  : 1 (pid: 1364902)
+  error_file: /tmp/torchelastic_iw__ftcn/13849095_k3rbzj4l/attempt_0/0/error.json
+  traceback : Traceback (most recent call last):
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+      return f(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+      outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+      outputs = self.pipeline_engine.train_batch_iter(
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+      output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+      output = model(**micro_batch)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+      sharded_logits = self.model(
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+      return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 896, in forward_with_hidden_states
+      output = self.token_position_embeddings(input_ids=input_ids, input_mask=input_mask)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 160, in forward
+      output = self.pp_block(**new_kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 786, in forward
+      input_embeds = self.token_embedding(input_ids)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 290, in forward
+      out = out * (~input_mask[..., None])
+  torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 8.00 GiB. GPU 0 has a total capacty of 79.33 GiB of which 4.00 GiB is free. Including non-PyTorch memory, this process has 75.31 GiB memory in use. Of the allocated memory 63.52 GiB is allocated by PyTorch, and 97.18 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  
+============================================================
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 2636009 got signal: 15
+srun: error: ip-26-0-161-153: task 7: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 1227389 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 96814 got signal: 15
+srun: error: ip-26-0-165-164: task 11: Exited with exit code 1
+srun: error: ip-26-0-161-103: task 3: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 153487 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 235573 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 1630127 got signal: 15
+srun: error: ip-26-0-165-202: task 12: Exited with exit code 1
+srun: error: ip-26-0-168-30: task 13: Exited with exit code 1
+srun: error: ip-26-0-165-131: task 10: Exited with exit code 1
+[2024-12-28 11:56:42,313] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-168-34.ec2.internal_272077_0' has failed to send a keep-alive heartbeat to the rendezvous '13849095' due to an error of type RendezvousConnectionError.
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 272077 got signal: 15
+srun: error: ip-26-0-168-34: task 14: Exited with exit code 1
+srun: error: Node failure on ip-26-0-160-242
+slurmstepd: error: *** JOB 13849095 ON ip-26-0-160-225 CANCELLED AT 2024-12-28T12:10:12 DUE TO NODE FAILURE, SEE SLURMCTLD LOG FOR DETAILS ***
+srun: Job step aborted: Waiting up to 32 seconds for job step to finish.
diff --git a/logs/13849414-bench_3.57G_dp2_tp16_pp8_acc32_mbs4_seq4096_zero1_tpmodeRED_vocab131k.out b/logs/13849414-bench_3.57G_dp2_tp16_pp8_acc32_mbs4_seq4096_zero1_tpmodeRED_vocab131k.out
new file mode 100644
index 0000000000000000000000000000000000000000..fb84207fc68c2f473a67f3c275926e5ffb8ab2ce
--- /dev/null
+++ b/logs/13849414-bench_3.57G_dp2_tp16_pp8_acc32_mbs4_seq4096_zero1_tpmodeRED_vocab131k.out
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:35580475883496d64443dcf240a2a7f9ca2871be63339964448811da31de893b
+size 17075014
diff --git a/logs/13849632-bench_8.86G_dp8_tp16_pp2_acc4_mbs8_seq4096_zero1_tpmodeRED_vocab131k.out b/logs/13849632-bench_8.86G_dp8_tp16_pp2_acc4_mbs8_seq4096_zero1_tpmodeRED_vocab131k.out
new file mode 100644
index 0000000000000000000000000000000000000000..5b0b3d5ea5619bc0bd1d70ba60c37e76e0de58ce
--- /dev/null
+++ b/logs/13849632-bench_8.86G_dp8_tp16_pp2_acc4_mbs8_seq4096_zero1_tpmodeRED_vocab131k.out
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f9c7b0f02ca03489ae9c68d7edd13d177a9c4f5ad2715b72d16e67b910692c26
+size 16079014
diff --git a/logs/14018389-bench_1.34G_dp1_tp8_pp2_acc8_mbs32_seq4096_zero0_tpmodeRED_vocab131k.out b/logs/14018389-bench_1.34G_dp1_tp8_pp2_acc8_mbs32_seq4096_zero0_tpmodeRED_vocab131k.out
new file mode 100644
index 0000000000000000000000000000000000000000..c838ba299589e37b367eba11b21923435357fe6a
--- /dev/null
+++ b/logs/14018389-bench_1.34G_dp1_tp8_pp2_acc8_mbs32_seq4096_zero0_tpmodeRED_vocab131k.out
@@ -0,0 +1,814 @@
++ '[' -z 14018389 ']'
++ source /etc/profile.d/modules.sh
+++ . /usr/share/modules/init/bash
++++ unset _mlshdbg
++++ '[' 0 = 1 ']'
++++ unset _mlre _mlIFS
++++ '[' -n x ']'
++++ _mlIFS=' 	
+'
++++ IFS=' '
++++ '[' -n '' ']'
+++++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash autoinit
++++ _mlcode='module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+MODULEPATH_modshare=/etc/environment-modules/modules:1:/usr/share/modules/\$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1; export MODULEPATH_modshare;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+LOADEDMODULES=; export LOADEDMODULES;
+MODULEPATH=/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/\$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles; export MODULEPATH;
+test 0;'
++++ _mlret=0
++++ '[' -n x ']'
++++ IFS=' 	
+'
++++ unset _mlIFS
++++ unset _mlre _mlv _mlrv
++++ '[' 0 -eq 0 ']'
++++ eval 'module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+MODULEPATH_modshare=/etc/environment-modules/modules:1:/usr/share/modules/\$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1; export MODULEPATH_modshare;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+LOADEDMODULES=; export LOADEDMODULES;
+MODULEPATH=/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/\$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles; export MODULEPATH;
+test 0;'
+++++ MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl
+++++ export MODULES_CMD
+++++ ENV=/usr/share/modules/init/profile.sh
+++++ export ENV
+++++ MODULEPATH_modshare='/etc/environment-modules/modules:1:/usr/share/modules/$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1'
+++++ export MODULEPATH_modshare
+++++ BASH_ENV=/usr/share/modules/init/bash
+++++ export BASH_ENV
+++++ MODULESHOME=/usr/share/modules
+++++ export MODULESHOME
+++++ LOADEDMODULES=
+++++ export LOADEDMODULES
+++++ MODULEPATH='/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles'
+++++ export MODULEPATH
+++++ test 0
++++ '[' 0 = 1 ']'
++++ '[' -t 2 ']'
++++ export -f module
++++ export -f switchml
++++ '[' 5 -ge 3 ']'
++++ [[ ehxB =~ i ]]
++++ [[ ! :/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin: =~ :/usr/bin: ]]
++++ '[' '!' -n '' ']'
++++ MANPATH=:
++++ export MANPATH
+++++ manpath
++++ [[ ! :/fsx/nouamane/miniconda/envs/2-1-cu121/man:/fsx/nouamane/miniconda/envs/2-1-cu121/share/man:/fsx/nouamane/miniconda/man:/fsx/nouamane/miniconda/share/man:/usr/local/man:/usr/local/share/man:/usr/share/man:/opt/slurm/share/man:/admin/home/nouamane/.local/share/man:/admin/home/nouamane/.fzf/man:: =~ :/usr/share/man: ]]
++++ unset _mlcode _mlret
++++ '[' -n '' ']'
++ module load cuda/12.1
++ unset _mlshdbg
++ '[' 0 = 1 ']'
++ unset _mlre _mlIFS
++ '[' -n x ']'
++ _mlIFS=' 	
+'
++ IFS=' '
++ '[' -n '' ']'
+++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash load cuda/12.1
++ eval 'CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include;' export 'CPATH;
+LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:;' export 'LD_LIBRARY_PATH;
+FI_EFA_FORK_SAFE=1;' export 'FI_EFA_FORK_SAFE;
+MANPATH=/usr/local/cuda-12.1/share/man::;' export 'MANPATH;
+LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64;' export 'LIBRARY_PATH;
+_LMFILES_=/usr/share/modules/modulefiles/cuda/12.1;' export '_LMFILES_;
+LOADEDMODULES=cuda/12.1;' export 'LOADEDMODULES;
+MPI_PATH=/opt/amazon/openmpi;' export 'MPI_PATH;
+NCCL_HOME_modshare=/opt/nccl/build:1;' export 'NCCL_HOME_modshare;
+NCCL_PROTO=simple;' export 'NCCL_PROTO;
+MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1;' export 'MANPATH_modshare;
+LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1;' export 'LIBRARY_PATH_modshare;
+NCCL_SOCKET_IFNAME=enp;' export 'NCCL_SOCKET_IFNAME;
+AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl;' export 'AWS_OFI_NCCL_HOME;
+NCCL_HOME=/opt/nccl/build;' export 'NCCL_HOME;
+FI_PROVIDER=efa;' export 'FI_PROVIDER;
+AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1;' export 'AWS_OFI_NCCL_HOME_modshare;
+CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1;' export 'CPATH_modshare;
+LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2::1:/opt/nccl/build/lib:1:/opt/aws-ofi-nccl/lib:2:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/efa/lib:1;' export 'LD_LIBRARY_PATH_modshare;
+FI_EFA_ENABLE_SHM_TRANSFER=1;' export 'FI_EFA_ENABLE_SHM_TRANSFER;
+_LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1;' export '_LMFILES__modshare;
+LOADEDMODULES_modshare=cuda/12.1:1;' export 'LOADEDMODULES_modshare;
+MPI_PATH_modshare=/opt/amazon/openmpi:1;' export 'MPI_PATH_modshare;
+PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin;' export 'PATH;
+CUDA_HOME=/usr/local/cuda-12.1;' export 'CUDA_HOME;
+PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/opt/slurm/bin:1:/usr/sbin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1;' export 'PATH_modshare;
+test' '0;'
+++ CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include
+++ export CPATH
+++ LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:
+++ export LD_LIBRARY_PATH
+++ FI_EFA_FORK_SAFE=1
+++ export FI_EFA_FORK_SAFE
+++ MANPATH=/usr/local/cuda-12.1/share/man::
+++ export MANPATH
+++ LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64
+++ export LIBRARY_PATH
+++ _LMFILES_=/usr/share/modules/modulefiles/cuda/12.1
+++ export _LMFILES_
+++ LOADEDMODULES=cuda/12.1
+++ export LOADEDMODULES
+++ MPI_PATH=/opt/amazon/openmpi
+++ export MPI_PATH
+++ NCCL_HOME_modshare=/opt/nccl/build:1
+++ export NCCL_HOME_modshare
+++ NCCL_PROTO=simple
+++ export NCCL_PROTO
+++ MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1
+++ export MANPATH_modshare
+++ LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1
+++ export LIBRARY_PATH_modshare
+++ NCCL_SOCKET_IFNAME=enp
+++ export NCCL_SOCKET_IFNAME
+++ AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl
+++ export AWS_OFI_NCCL_HOME
+++ NCCL_HOME=/opt/nccl/build
+++ export NCCL_HOME
+++ FI_PROVIDER=efa
+++ export FI_PROVIDER
+++ AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1
+++ export AWS_OFI_NCCL_HOME_modshare
+++ CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1
+++ export CPATH_modshare
+++ LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2::1:/opt/nccl/build/lib:1:/opt/aws-ofi-nccl/lib:2:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/efa/lib:1
+++ export LD_LIBRARY_PATH_modshare
+++ FI_EFA_ENABLE_SHM_TRANSFER=1
+++ export FI_EFA_ENABLE_SHM_TRANSFER
+++ _LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1
+++ export _LMFILES__modshare
+++ LOADEDMODULES_modshare=cuda/12.1:1
+++ export LOADEDMODULES_modshare
+++ MPI_PATH_modshare=/opt/amazon/openmpi:1
+++ export MPI_PATH_modshare
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
+++ export PATH
+++ CUDA_HOME=/usr/local/cuda-12.1
+++ export CUDA_HOME
+++ PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/opt/slurm/bin:1:/usr/sbin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1
+++ export PATH_modshare
+++ test 0
++ _mlstatus=0
++ '[' -n x ']'
++ IFS=' 	
+'
++ unset _mlre _mlv _mlrv _mlIFS
++ '[' -n '' ']'
++ unset _mlshdbg
++ return 0
++ source /fsx/nouamane/miniconda/bin/activate
+++ _CONDA_ROOT=/fsx/nouamane/miniconda
+++ . /fsx/nouamane/miniconda/etc/profile.d/conda.sh
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ '[' -z x ']'
+++ conda activate
+++ local cmd=activate
+++ case "$cmd" in
+++ __conda_activate activate
+++ '[' -n '' ']'
+++ local ask_conda
++++ PS1=
++++ __conda_exe shell.posix activate
++++ /fsx/nouamane/miniconda/bin/conda shell.posix activate
+Error while loading conda entry point: conda-libmamba-solver (module 'libmambapy' has no attribute 'QueryFormat')
+++ ask_conda='. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''2'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_1='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/etc/conda/activate.d/libxml2_activate.sh"'
+++ eval '. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''2'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_1='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/etc/conda/activate.d/libxml2_activate.sh"'
++++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh
+++++ test -n ''
+++++ unset XML_CATALOG_FILES
+++++ unset xml_catalog_files_libxml2
++++ PS1='(base) '
++++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
++++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
++++ export CONDA_PREFIX=/fsx/nouamane/miniconda
++++ CONDA_PREFIX=/fsx/nouamane/miniconda
++++ export CONDA_SHLVL=2
++++ CONDA_SHLVL=2
++++ export CONDA_DEFAULT_ENV=base
++++ CONDA_DEFAULT_ENV=base
++++ export 'CONDA_PROMPT_MODIFIER=(base) '
++++ CONDA_PROMPT_MODIFIER='(base) '
++++ export CONDA_PREFIX_1=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ CONDA_PREFIX_1=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ . /fsx/nouamane/miniconda/etc/conda/activate.d/libxml2_activate.sh
+++++ test -n ''
+++++ xml_catalog_files_libxml2=
+++++ XML_CATALOG_FILES=
+++++ conda_catalog_files=
+++++ ifs_libxml2=' 	
+'
+++++ IFS=' '
+++++ rem=/fsx/nouamane/miniconda
+++++ for pre in ${rem}
+++++ test '' = /fsx/nouamane/miniconda
+++++ conda_catalog_files=/fsx/nouamane/miniconda
+++++ rem=
+++++ IFS=' 	
+'
+++++ conda_catalog_files='file:///fsx/nouamane/miniconda/etc/xml/catalog file:///etc/xml/catalog'
+++++ export 'XML_CATALOG_FILES=file:///fsx/nouamane/miniconda/etc/xml/catalog file:///etc/xml/catalog'
+++++ XML_CATALOG_FILES='file:///fsx/nouamane/miniconda/etc/xml/catalog file:///etc/xml/catalog'
+++++ unset conda_catalog_files ifs_libxml2 rem
+++ __conda_hashr
+++ '[' -n '' ']'
+++ '[' -n '' ']'
+++ hash -r
++ conda activate 2-1-cu121
++ local cmd=activate
++ case "$cmd" in
++ __conda_activate activate 2-1-cu121
++ '[' -n '' ']'
++ local ask_conda
+++ PS1='(base) '
+++ __conda_exe shell.posix activate 2-1-cu121
+++ /fsx/nouamane/miniconda/bin/conda shell.posix activate 2-1-cu121
+Error while loading conda entry point: conda-libmamba-solver (module 'libmambapy' has no attribute 'QueryFormat')
++ ask_conda='. "/fsx/nouamane/miniconda/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_2='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
++ eval '. "/fsx/nouamane/miniconda/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_2='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
+++ . /fsx/nouamane/miniconda/etc/conda/deactivate.d/libxml2_deactivate.sh
++++ test -n ''
++++ unset XML_CATALOG_FILES
++++ unset xml_catalog_files_libxml2
+++ PS1='(2-1-cu121) '
+++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
+++ export CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ export CONDA_SHLVL=3
+++ CONDA_SHLVL=3
+++ export CONDA_DEFAULT_ENV=2-1-cu121
+++ CONDA_DEFAULT_ENV=2-1-cu121
+++ export 'CONDA_PROMPT_MODIFIER=(2-1-cu121) '
+++ CONDA_PROMPT_MODIFIER='(2-1-cu121) '
+++ export CONDA_PREFIX_2=/fsx/nouamane/miniconda
+++ CONDA_PREFIX_2=/fsx/nouamane/miniconda
+++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ export _CE_M=
+++ _CE_M=
+++ export _CE_CONDA=
+++ _CE_CONDA=
+++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh
++++ test -n ''
++++ xml_catalog_files_libxml2=
++++ XML_CATALOG_FILES=
++++ conda_catalog_files=
++++ ifs_libxml2=' 	
+'
++++ IFS=' '
++++ rem=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ for pre in ${rem}
++++ test '' = /fsx/nouamane/miniconda/envs/2-1-cu121
++++ conda_catalog_files=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ rem=
++++ IFS=' 	
+'
++++ conda_catalog_files='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ export 'XML_CATALOG_FILES=file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ XML_CATALOG_FILES='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ unset conda_catalog_files ifs_libxml2 rem
++ __conda_hashr
++ '[' -n '' ']'
++ '[' -n '' ']'
++ hash -r
++ export PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
++ PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
++ '[' -z '' ']'
+++ scontrol show hostnames 'ip-26-0-163-[43,58]'
++ export 'NODELIST=ip-26-0-163-43
+ip-26-0-163-58'
++ NODELIST='ip-26-0-163-43
+ip-26-0-163-58'
+++ head -n1
+++ scontrol show hostnames 'ip-26-0-163-[43,58]'
++ export MASTER_NODE=ip-26-0-163-43
++ MASTER_NODE=ip-26-0-163-43
++ export MASTER_PORT=12356
++ MASTER_PORT=12356
++ export NNODES=2
++ NNODES=2
++ export GPUS_PER_NODE=8
++ GPUS_PER_NODE=8
++ export WORLD_SIZE=16
++ WORLD_SIZE=16
++ export CUDA_DEVICE_MAX_CONNECTIONS=1
++ CUDA_DEVICE_MAX_CONNECTIONS=1
++ export NCCL_DEBUG=WARN
++ NCCL_DEBUG=WARN
++ export NANOTRON_BENCHMARK=1
++ NANOTRON_BENCHMARK=1
++ export WANDB_MODE=disabled
++ WANDB_MODE=disabled
++ export TORCH_NCCL_ASYNC_ERROR_HANDLING=1
++ TORCH_NCCL_ASYNC_ERROR_HANDLING=1
++ export TORCH_DISTRIBUTED_DEBUG=DETAIL
++ TORCH_DISTRIBUTED_DEBUG=DETAIL
++ '[' -z '' ']'
++ echo '=== GPU Topology ==='
+=== GPU Topology ===
++ nvidia-smi topo -m
+	GPU0	GPU1	GPU2	GPU3	GPU4	GPU5	GPU6	GPU7	CPU Affinity	NUMA Affinity	GPU NUMA ID
+GPU0	 X 	NV18	NV18	NV18	NV18	NV18	NV18	NV18	0-43	0		N/A
+GPU1	NV18	 X 	NV18	NV18	NV18	NV18	NV18	NV18	0-43	0		N/A
+GPU2	NV18	NV18	 X 	NV18	NV18	NV18	NV18	NV18	0-43	0		N/A
+GPU3	NV18	NV18	NV18	 X 	NV18	NV18	NV18	NV18	0-43	0		N/A
+GPU4	NV18	NV18	NV18	NV18	 X 	NV18	NV18	NV18	52-95	1		N/A
+GPU5	NV18	NV18	NV18	NV18	NV18	 X 	NV18	NV18	52-95	1		N/A
+GPU6	NV18	NV18	NV18	NV18	NV18	NV18	 X 	NV18	52-95	1		N/A
+GPU7	NV18	NV18	NV18	NV18	NV18	NV18	NV18	 X 	52-95	1		N/A
+
+Legend:
+
+  X    = Self
+  SYS  = Connection traversing PCIe as well as the SMP interconnect between NUMA nodes (e.g., QPI/UPI)
+  NODE = Connection traversing PCIe as well as the interconnect between PCIe Host Bridges within a NUMA node
+  PHB  = Connection traversing PCIe as well as a PCIe Host Bridge (typically the CPU)
+  PXB  = Connection traversing multiple PCIe bridges (without traversing the PCIe Host Bridge)
+  PIX  = Connection traversing at most a single PCIe bridge
+  NV#  = Connection traversing a bonded set of # NVLinks
++ echo ==================
+==================
++ export SRUN_ALLOC_ARGS=
++ SRUN_ALLOC_ARGS=
++ echo 'Master node: ip-26-0-163-43'
+Master node: ip-26-0-163-43
++ echo 'All nodes: ip-26-0-163-43
+ip-26-0-163-58'
+All nodes: ip-26-0-163-43
+ip-26-0-163-58
++ echo 'World size: 16'
+World size: 16
++ '[' -n '' ']'
++ srun --wait=0 --kill-on-bad-exit=1 torchrun --nnodes=2 --nproc_per_node=8 --rdzv_id=14018389 --rdzv_backend=c10d --rdzv_endpoint=ip-26-0-163-43:12356 --max_restarts 0 --rdzv_conf timeout=60 /fsx/nouamane/projects/nanotron/run_train.py --config-file benchmark/configs/config_1.34G_dp1_tp8_pp2_acc8_mbs32_seq4096_zero0_tpmodeRED_vocab131k.yaml
+[2025-01-01 23:02:01,030] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-01 23:02:01,030] torch.distributed.run: [WARNING] 
+[2025-01-01 23:02:01,030] torch.distributed.run: [WARNING] *****************************************
+[2025-01-01 23:02:01,030] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-01 23:02:01,030] torch.distributed.run: [WARNING] *****************************************
+[2025-01-01 23:02:01,313] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-01 23:02:01,313] torch.distributed.run: [WARNING] 
+[2025-01-01 23:02:01,313] torch.distributed.run: [WARNING] *****************************************
+[2025-01-01 23:02:01,313] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-01 23:02:01,313] torch.distributed.run: [WARNING] *****************************************
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+01/01/2025 23:02:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]: Config:
+01/01/2025 23:02:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]: Config(general=GeneralArgs(project='debug',
+01/01/2025 23:02:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]:                            run='1.34G_dp1_tp8_pp2_acc8_mbs32_seq4096_zero0_tpmodeRED_vocab131k',
+01/01/2025 23:02:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]:                            seed=42,
+01/01/2025 23:02:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]:                            step=None,
+01/01/2025 23:02:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]:                            consumed_train_samples=None,
+01/01/2025 23:02:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]:                            benchmark_csv_path=PosixPath('benchmark/results/bench_final2.csv'),
+01/01/2025 23:02:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]:                            ignore_sanity_checks=True),
+01/01/2025 23:02:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]:        parallelism=ParallelismArgs(dp=1,
+01/01/2025 23:02:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]:                                    pp=2,
+01/01/2025 23:02:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]:                                    tp=8,
+01/01/2025 23:02:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]:                                    pp_engine=<nanotron.parallel.pipeline_parallel.engine.OneForwardOneBackwardPipelineEngine object at 0x7f108aa84a30>,
+01/01/2025 23:02:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]:                                    tp_mode=<TensorParallelLinearMode.REDUCE_SCATTER: 2>,
+01/01/2025 23:02:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]:                                    tp_linear_async_communication=True,
+01/01/2025 23:02:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]:                                    recompute_layer=False,
+01/01/2025 23:02:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]:                                    tp_recompute_allgather=True,
+01/01/2025 23:02:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]:                                    expert_parallel_size=1),
+01/01/2025 23:02:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]:        model=ModelArgs(model_config=LlamaConfig(bos_token_id=0,
+01/01/2025 23:02:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]:                                                 eos_token_id=0,
+01/01/2025 23:02:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]:                                                 hidden_act='silu',
+01/01/2025 23:02:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]:                                                 hidden_size=2048,
+01/01/2025 23:02:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]:                                                 initializer_range=0.02,
+01/01/2025 23:02:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]:                                                 intermediate_size=8192,
+01/01/2025 23:02:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]:                                                 is_llama_config=True,
+01/01/2025 23:02:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]:                                                 max_position_embeddings=4096,
+01/01/2025 23:02:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]:                                                 num_attention_heads=32,
+01/01/2025 23:02:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]:                                                 num_hidden_layers=16,
+01/01/2025 23:02:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]:                                                 num_key_value_heads=32,
+01/01/2025 23:02:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]:                                                 pad_token_id=None,
+01/01/2025 23:02:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]:                                                 pretraining_tp=1,
+01/01/2025 23:02:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]:                                                 rms_norm_eps=1e-05,
+01/01/2025 23:02:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]:                                                 rope_scaling=None,
+01/01/2025 23:02:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]:                                                 rope_theta=10000.0,
+01/01/2025 23:02:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]:                                                 rope_interleaved=False,
+01/01/2025 23:02:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]:                                                 tie_word_embeddings=True,
+01/01/2025 23:02:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]:                                                 use_cache=True,
+01/01/2025 23:02:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]:                                                 vocab_size=131072),
+01/01/2025 23:02:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]:                        init_method=RandomInit(std=0.02),
+01/01/2025 23:02:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]:                        dtype=torch.bfloat16,
+01/01/2025 23:02:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]:                        make_vocab_size_divisible_by=1,
+01/01/2025 23:02:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]:                        ddp_bucket_cap_mb=25),
+01/01/2025 23:02:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]:        tokenizer=TokenizerArgs(tokenizer_name_or_path='robot-test/dummy-tokenizer-wordlevel',
+01/01/2025 23:02:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]:                                tokenizer_revision=None,
+01/01/2025 23:02:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]:                                tokenizer_max_length=None),
+01/01/2025 23:02:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]:        checkpoints=CheckpointsArgs(checkpoints_path=PosixPath('checkpoints'),
+01/01/2025 23:02:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]:                                    checkpoint_interval=10000,
+01/01/2025 23:02:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]:                                    save_initial_state=False,
+01/01/2025 23:02:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]:                                    save_final_state=False,
+01/01/2025 23:02:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]:                                    resume_checkpoint_path=None,
+01/01/2025 23:02:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]:                                    checkpoints_path_is_shared_file_system=False),
+01/01/2025 23:02:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]:        logging=LoggingArgs(log_level='info',
+01/01/2025 23:02:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]:                            log_level_replica='info',
+01/01/2025 23:02:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]:                            iteration_step_info_interval=1),
+01/01/2025 23:02:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]:        tokens=TokensArgs(sequence_length=4096,
+01/01/2025 23:02:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]:                          train_steps=100,
+01/01/2025 23:02:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]:                          micro_batch_size=32,
+01/01/2025 23:02:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]:                          batch_accumulation_per_replica=8,
+01/01/2025 23:02:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]:                          val_check_interval=100,
+01/01/2025 23:02:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]:                          limit_val_batches=0,
+01/01/2025 23:02:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]:                          limit_test_batches=0),
+01/01/2025 23:02:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]:        optimizer=OptimizerArgs(optimizer_factory=AdamWOptimizerArgs(adam_eps=1e-08,
+01/01/2025 23:02:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]:                                                                     adam_beta1=0.9,
+01/01/2025 23:02:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]:                                                                     adam_beta2=0.95,
+01/01/2025 23:02:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]:                                                                     torch_adam_is_fused=True,
+01/01/2025 23:02:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]:                                                                     name='adamW'),
+01/01/2025 23:02:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]:                                zero_stage=0,
+01/01/2025 23:02:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]:                                weight_decay=0.01,
+01/01/2025 23:02:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]:                                clip_grad=1.0,
+01/01/2025 23:02:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]:                                accumulate_grad_in_fp32=True,
+01/01/2025 23:02:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]:                                learning_rate_scheduler=LRSchedulerArgs(learning_rate=0.0003,
+01/01/2025 23:02:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]:                                                                        lr_warmup_steps=2,
+01/01/2025 23:02:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]:                                                                        lr_warmup_style='linear',
+01/01/2025 23:02:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]:                                                                        lr_decay_style='cosine',
+01/01/2025 23:02:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]:                                                                        lr_decay_steps=13,
+01/01/2025 23:02:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]:                                                                        lr_decay_starting_step=None,
+01/01/2025 23:02:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]:                                                                        min_decay_lr=1e-05)),
+01/01/2025 23:02:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]:        data_stages=[DatasetStageArgs(name='Stable Training Stage',
+01/01/2025 23:02:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]:                                      start_training_step=1,
+01/01/2025 23:02:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]:                                      data=DataArgs(dataset=None,
+01/01/2025 23:02:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]:                                                    seed=42,
+01/01/2025 23:02:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]:                                                    num_loading_workers=1))],
+01/01/2025 23:02:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]:        profiler=None,
+01/01/2025 23:02:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]:        lighteval=None,
+01/01/2025 23:02:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]:        s3_upload=None)
+01/01/2025 23:02:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]: Model Config:
+01/01/2025 23:02:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]: LlamaConfig(bos_token_id=0,
+01/01/2025 23:02:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]:             eos_token_id=0,
+01/01/2025 23:02:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]:             hidden_act='silu',
+01/01/2025 23:02:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]:             hidden_size=2048,
+01/01/2025 23:02:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]:             initializer_range=0.02,
+01/01/2025 23:02:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]:             intermediate_size=8192,
+01/01/2025 23:02:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]:             is_llama_config=True,
+01/01/2025 23:02:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]:             max_position_embeddings=4096,
+01/01/2025 23:02:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]:             num_attention_heads=32,
+01/01/2025 23:02:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]:             num_hidden_layers=16,
+01/01/2025 23:02:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]:             num_key_value_heads=32,
+01/01/2025 23:02:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]:             pad_token_id=None,
+01/01/2025 23:02:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]:             pretraining_tp=1,
+01/01/2025 23:02:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]:             rms_norm_eps=1e-05,
+01/01/2025 23:02:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]:             rope_scaling=None,
+01/01/2025 23:02:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]:             rope_theta=10000.0,
+01/01/2025 23:02:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]:             rope_interleaved=False,
+01/01/2025 23:02:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]:             tie_word_embeddings=True,
+01/01/2025 23:02:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]:             use_cache=True,
+01/01/2025 23:02:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]:             vocab_size=131072)
+01/01/2025 23:02:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]: Building model..
+01/01/2025 23:02:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]: Initialize RoPE Theta = 10000.0
+01/01/2025 23:02:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]: Setting PP block ranks...
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+01/01/2025 23:02:48 [INFO|DP=0|PP=0|TP=2|ip-26-0-163-43]: Local number of parameters: 126M (240.09MiB)
+01/01/2025 23:02:48 [INFO|DP=0|PP=1|TP=2|ip-26-0-163-58]: Local number of parameters: 75.5M (144.04MiB)
+01/01/2025 23:02:48 [INFO|DP=0|PP=0|TP=1|ip-26-0-163-43]: Local number of parameters: 126M (240.09MiB)
+01/01/2025 23:02:48 [INFO|DP=0|PP=1|TP=7|ip-26-0-163-58]: Local number of parameters: 75.5M (144.04MiB)
+01/01/2025 23:02:48 [INFO|DP=0|PP=0|TP=2|ip-26-0-163-43]: [After model building] Memory usage: 241.11MiB. Peak allocated: 5504.00MiB Peak reserved: 12684.00MiB
+01/01/2025 23:02:48 [INFO|DP=0|PP=0|TP=7|ip-26-0-163-43]: Local number of parameters: 126M (240.09MiB)
+01/01/2025 23:02:48 [INFO|DP=0|PP=1|TP=2|ip-26-0-163-58]: [After model building] Memory usage: 145.06MiB. Peak allocated: 5504.00MiB Peak reserved: 12678.00MiB
+01/01/2025 23:02:48 [INFO|DP=0|PP=0|TP=1|ip-26-0-163-43]: [After model building] Memory usage: 241.11MiB. Peak allocated: 5504.00MiB Peak reserved: 12684.00MiB
+01/01/2025 23:02:48 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]: Total number of parameters: 1.61G (3073.03MiB)
+01/01/2025 23:02:48 [INFO|DP=0|PP=1|TP=7|ip-26-0-163-58]: [After model building] Memory usage: 145.06MiB. Peak allocated: 5504.00MiB Peak reserved: 12678.00MiB
+01/01/2025 23:02:48 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]: Local number of parameters: 126M (240.09MiB)
+01/01/2025 23:02:48 [INFO|DP=0|PP=0|TP=7|ip-26-0-163-43]: [After model building] Memory usage: 241.11MiB. Peak allocated: 5504.00MiB Peak reserved: 12684.00MiB
+01/01/2025 23:02:48 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]: [After model building] Memory usage: 241.11MiB. Peak allocated: 5504.00MiB Peak reserved: 12684.00MiB
+01/01/2025 23:02:48 [INFO|DP=0|PP=0|TP=5|ip-26-0-163-43]: Local number of parameters: 126M (240.09MiB)
+01/01/2025 23:02:48 [INFO|DP=0|PP=1|TP=1|ip-26-0-163-58]: Local number of parameters: 75.5M (144.04MiB)
+01/01/2025 23:02:48 [INFO|DP=0|PP=1|TP=5|ip-26-0-163-58]: Local number of parameters: 75.5M (144.04MiB)
+01/01/2025 23:02:48 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]: No checkpoint path provided.
+01/01/2025 23:02:48 [INFO|DP=0|PP=0|TP=6|ip-26-0-163-43]: Local number of parameters: 126M (240.09MiB)
+01/01/2025 23:02:48 [INFO|DP=0|PP=0|TP=4|ip-26-0-163-43]: Local number of parameters: 126M (240.09MiB)
+01/01/2025 23:02:48 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]: Parametrizing model parameters using StandardParametrizator
+01/01/2025 23:02:48 [INFO|DP=0|PP=0|TP=5|ip-26-0-163-43]: [After model building] Memory usage: 241.11MiB. Peak allocated: 5504.00MiB Peak reserved: 12684.00MiB
+01/01/2025 23:02:48 [INFO|DP=0|PP=1|TP=6|ip-26-0-163-58]: Local number of parameters: 75.5M (144.04MiB)
+01/01/2025 23:02:48 [INFO|DP=0|PP=1|TP=1|ip-26-0-163-58]: [After model building] Memory usage: 145.06MiB. Peak allocated: 5504.00MiB Peak reserved: 12678.00MiB
+01/01/2025 23:02:48 [INFO|DP=0|PP=1|TP=5|ip-26-0-163-58]: [After model building] Memory usage: 145.06MiB. Peak allocated: 5504.00MiB Peak reserved: 12678.00MiB
+01/01/2025 23:02:48 [INFO|DP=0|PP=0|TP=3|ip-26-0-163-43]: Local number of parameters: 126M (240.09MiB)
+01/01/2025 23:02:48 [INFO|DP=0|PP=1|TP=3|ip-26-0-163-58]: Local number of parameters: 75.5M (144.04MiB)
+01/01/2025 23:02:48 [INFO|DP=0|PP=0|TP=6|ip-26-0-163-43]: [After model building] Memory usage: 241.11MiB. Peak allocated: 5504.00MiB Peak reserved: 12684.00MiB
+01/01/2025 23:02:48 [INFO|DP=0|PP=0|TP=4|ip-26-0-163-43]: [After model building] Memory usage: 241.11MiB. Peak allocated: 5504.00MiB Peak reserved: 12684.00MiB
+01/01/2025 23:02:48 [INFO|DP=0|PP=1|TP=6|ip-26-0-163-58]: [After model building] Memory usage: 145.06MiB. Peak allocated: 5504.00MiB Peak reserved: 12678.00MiB
+01/01/2025 23:02:48 [INFO|DP=0|PP=1|TP=3|ip-26-0-163-58]: [After model building] Memory usage: 145.06MiB. Peak allocated: 5504.00MiB Peak reserved: 12678.00MiB
+01/01/2025 23:02:48 [INFO|DP=0|PP=0|TP=3|ip-26-0-163-43]: [After model building] Memory usage: 241.11MiB. Peak allocated: 5504.00MiB Peak reserved: 12684.00MiB
+01/01/2025 23:02:48 [INFO|DP=0|PP=1|TP=4|ip-26-0-163-58]: Local number of parameters: 75.5M (144.04MiB)
+01/01/2025 23:02:48 [INFO|DP=0|PP=1|TP=4|ip-26-0-163-58]: [After model building] Memory usage: 145.06MiB. Peak allocated: 5504.00MiB Peak reserved: 12678.00MiB
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+01/01/2025 23:02:48 [INFO|DP=0|PP=1|TP=0|ip-26-0-163-58]: Local number of parameters: 75.5M (144.04MiB)
+01/01/2025 23:02:48 [INFO|DP=0|PP=1|TP=0|ip-26-0-163-58]: [After model building] Memory usage: 145.06MiB. Peak allocated: 5504.00MiB Peak reserved: 12678.00MiB
+01/01/2025 23:02:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]: [Optimizer Building] Using LearningRateForSP as learning rate
+01/01/2025 23:02:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]: module_name: model.token_position_embeddings | PP: 0/2 | Block rank: 0
+01/01/2025 23:02:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]: module_name: model.decoder.0 | PP: 0/2 | Block rank: 1
+01/01/2025 23:02:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]: module_name: model.decoder.1 | PP: 0/2 | Block rank: 2
+01/01/2025 23:02:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]: module_name: model.decoder.2 | PP: 0/2 | Block rank: 3
+01/01/2025 23:02:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]: module_name: model.decoder.3 | PP: 0/2 | Block rank: 4
+01/01/2025 23:02:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]: module_name: model.decoder.4 | PP: 0/2 | Block rank: 5
+01/01/2025 23:02:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]: module_name: model.decoder.5 | PP: 0/2 | Block rank: 6
+01/01/2025 23:02:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]: module_name: model.decoder.6 | PP: 0/2 | Block rank: 7
+01/01/2025 23:02:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]: module_name: model.decoder.7 | PP: 0/2 | Block rank: 8
+01/01/2025 23:02:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]: module_name: model.decoder.8 | PP: 0/2 | Block rank: 9
+01/01/2025 23:02:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]: module_name: model.decoder.9 | PP: 0/2 | Block rank: 10
+01/01/2025 23:02:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]: module_name: model.decoder.10 | PP: 0/2 | Block rank: 11
+01/01/2025 23:02:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]: module_name: model.decoder.11 | PP: 1/2 | Block rank: 0
+01/01/2025 23:02:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]: module_name: model.decoder.12 | PP: 1/2 | Block rank: 1
+01/01/2025 23:02:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]: module_name: model.decoder.13 | PP: 1/2 | Block rank: 2
+01/01/2025 23:02:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]: module_name: model.decoder.14 | PP: 1/2 | Block rank: 3
+01/01/2025 23:02:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]: module_name: model.decoder.15 | PP: 1/2 | Block rank: 4
+01/01/2025 23:02:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]: module_name: model.final_layer_norm | PP: 1/2 | Block rank: 5
+01/01/2025 23:02:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]: module_name: model.lm_head | PP: 1/2 | Block rank: 6
+01/01/2025 23:02:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]: module_name: model.cast_to_fp32 | PP: 1/2 | Block rank: 7
+01/01/2025 23:02:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]: module_name: loss | PP: 1/2 | Block rank: 8
+01/01/2025 23:02:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]: [Training Plan] Stage Stable Training Stage has 99 remaining training steps and has consumed 0 samples
+01/01/2025 23:02:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]: Using dummy data generator
+01/01/2025 23:02:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]: [Training Plan] There are 1 training stages 
+01/01/2025 23:02:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]: [Stage Stable Training Stage] start from step 1 
+01/01/2025 23:02:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]: 
+01/01/2025 23:02:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]: [Start training] datetime: 2025-01-01 23:02:50.074719 | mbs: 32 | grad_accum: 8 | global_batch_size: 256 | sequence_length: 4096 | train_steps: 100 | start_iteration_step: 0 | consumed_train_samples: 0
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+01/01/2025 23:02:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]: Resuming training from stage Stable Training Stage, it has trained for 0 samples and has 99 remaining train steps
+01/01/2025 23:02:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]:  Memory usage: 1201.45MiB. Peak allocated 5504.00MiB. Peak reserved: 12684.00MiB
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+01/01/2025 23:03:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]:  Memory usage: 2296.07MiB. Peak allocated 39593.23MiB. Peak reserved: 51944.00MiB
+01/01/2025 23:03:10 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]:  Memory usage: 3256.45MiB. Peak allocated 3256.45MiB. Peak reserved: 51944.00MiB
+01/01/2025 23:03:10 [INFO|DP=0|PP=1|TP=0|ip-26-0-163-58]: iteration: 1 / 100 | consumed_tokens: 1.05M | elapsed_time_per_iteration_ms: 18.9K | tokens_per_sec: 55.5K | tokens_per_sec_per_gpu: 3.47K | global_batch_size: 256 | lm_loss: 12 | lr: 0.00015 | model_tflops_per_gpu: 33.5 | hardware_tflops_per_gpu: 33.5 | grad_norm: 0.389 | cuda_memory_allocated: 2.51G | cuda_max_memory_reserved: 41G | hd_total_memory_tb: 312G | hd_used_memory_tb: 71.9G | hd_free_memory_tb: 240G
+01/01/2025 23:03:17 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]:  Memory usage: 3256.45MiB. Peak allocated 40553.61MiB. Peak reserved: 53480.00MiB
+01/01/2025 23:03:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]:  Memory usage: 3256.45MiB. Peak allocated 3256.49MiB. Peak reserved: 53480.00MiB
+01/01/2025 23:03:18 [INFO|DP=0|PP=1|TP=0|ip-26-0-163-58]: iteration: 2 / 100 | consumed_tokens: 2.1M | elapsed_time_per_iteration_ms: 7.14K | tokens_per_sec: 147K | tokens_per_sec_per_gpu: 9.18K | global_batch_size: 256 | lm_loss: 12 | lr: 0.0003 | model_tflops_per_gpu: 88.7 | hardware_tflops_per_gpu: 88.7 | grad_norm: 0.388 | cuda_memory_allocated: 2.51G | cuda_max_memory_reserved: 41G | hd_total_memory_tb: 312G | hd_used_memory_tb: 71.9G | hd_free_memory_tb: 240G
+01/01/2025 23:03:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]:  Memory usage: 3256.45MiB. Peak allocated 40553.61MiB. Peak reserved: 53480.00MiB
+01/01/2025 23:03:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]: | job_id   | name                                                           | nodes | seq_len | mbs | batch_accum | gbs | mTFLOPs | hTFLOPs | tok/s/gpu | AllReduce (GB/s) | AllGather (GB/s) | ReduceScatter (GB/s) | AR Intra-node (GB/s) | AG Intra-node (GB/s) | RS Intra-node (GB/s) | Mem Alloc (GB) | Mem Res (GB) | dp | pp | tp | pp_engine                           | tp_mode        | tp_async_comm | hidden_size | hidden_act | num_layers | num_heads | num_kv_heads | max_pos | vocab_size | tie_word_embeddings | dtype          | zero_stage | ddp_bucket_cap_mb | accumulate_grad_in_fp32 | Total Params | Local Params |
+01/01/2025 23:03:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]: | -------- | -------------------------------------------------------------- | ----- | ------- | --- | ----------- | --- | ------- | ------- | --------- | ---------------- | ---------------- | -------------------- | -------------------- | -------------------- | -------------------- | -------------- | ------------ | -- | -- | -- | ----------------------------------- | -------------- | ------------- | ----------- | ---------- | ---------- | --------- | ------------ | ------- | ---------- | ------------------- | -------------- | ---------- | ----------------- | ----------------------- | ------------ | ------------ |
+01/01/2025 23:03:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]: | 14018389 | 1.34G_dp1_tp8_pp2_acc8_mbs32_seq4096_zero0_tpmodeRED_vocab131k | 2     | 4096    | 32  | 8           | 256 | 87.80   | 87.80   | 9085.45   | 342.76           | 171.26           | 164.61               | 428.68               | 244.81               | 244.41               | 3.18           | 52.23        | 1  | 2  | 8  | OneForwardOneBackwardPipelineEngine | REDUCE_SCATTER | True          | 2048        | silu       | 16         | 32        | 32           | 4096    | 131072     | True                | torch.bfloat16 | 0          | 25                | True                    | 1.61G        | 126M         |
+01/01/2025 23:03:25 [INFO|DP=0|PP=1|TP=0|ip-26-0-163-58]: iteration: 3 / 100 | consumed_tokens: 3.15M | elapsed_time_per_iteration_ms: 7.21K | tokens_per_sec: 145K | tokens_per_sec_per_gpu: 9.09K | global_batch_size: 256 | lm_loss: 12 | lr: 0.000296 | model_tflops_per_gpu: 87.8 | hardware_tflops_per_gpu: 87.8 | grad_norm: 0.383 | cuda_memory_allocated: 2.51G | cuda_max_memory_reserved: 41G | hd_total_memory_tb: 312G | hd_used_memory_tb: 71.9G | hd_free_memory_tb: 240G
+01/01/2025 23:03:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]: Attempting to write benchmark results to CSV file: benchmark/results/bench_final2.csv
+01/01/2025 23:03:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]: Acquired lock for CSV file: benchmark/results/bench_final2.csv
+01/01/2025 23:03:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]: Successfully wrote to CSV file: benchmark/results/bench_final2.csv. Releasing lock...
+01/01/2025 23:03:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-43]: Throughput logging complete
+[2025-01-01 23:03:47,635] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 53809 closing signal SIGTERM
+[2025-01-01 23:03:47,635] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 53810 closing signal SIGTERM
+[2025-01-01 23:03:47,635] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 53811 closing signal SIGTERM
+[2025-01-01 23:03:47,635] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 53812 closing signal SIGTERM
+[2025-01-01 23:03:47,635] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 53813 closing signal SIGTERM
+[2025-01-01 23:03:47,635] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 53814 closing signal SIGTERM
+[2025-01-01 23:03:47,635] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 53815 closing signal SIGTERM
+[2025-01-01 23:03:51,305] torch.distributed.elastic.multiprocessing.api: [ERROR] failed (exitcode: 1) local_rank: 0 (pid: 53808) of binary: /fsx/nouamane/miniconda/envs/2-1-cu121/bin/python
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 264, in launch_agent
+    raise ChildFailedError(
+torch.distributed.elastic.multiprocessing.errors.ChildFailedError: 
+============================================================
+/fsx/nouamane/projects/nanotron/run_train.py FAILED
+------------------------------------------------------------
+Failures:
+  <NO_OTHER_FAILURES>
+------------------------------------------------------------
+Root Cause (first observed failure):
+[0]:
+  time      : 2025-01-01_23:03:47
+  host      : ip-26-0-163-58.ec2.internal
+  rank      : 8 (local_rank: 0)
+  exitcode  : 1 (pid: 53808)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+============================================================
+srun: error: ip-26-0-163-58: task 1: Exited with exit code 1
+srun: launch/slurm: _step_signal: Terminating StepId=14018389.0
+slurmstepd: error: *** STEP 14018389.0 ON ip-26-0-163-43 CANCELLED AT 2025-01-01T23:03:51 ***
+[2025-01-01 23:03:51,664] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-01 23:03:51,664] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 58484 closing signal SIGTERM
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 58317 got signal: 15
+srun: error: ip-26-0-163-43: task 0: Exited with exit code 1
diff --git a/logs/14018953-bench_8.86G_dp1_tp16_pp8_acc16_mbs16_seq4096_zero0_tpmodeRED_vocab131k.out b/logs/14018953-bench_8.86G_dp1_tp16_pp8_acc16_mbs16_seq4096_zero0_tpmodeRED_vocab131k.out
new file mode 100644
index 0000000000000000000000000000000000000000..596c485abbef4dda85b2208ca89daa5850f33b4c
--- /dev/null
+++ b/logs/14018953-bench_8.86G_dp1_tp16_pp8_acc16_mbs16_seq4096_zero0_tpmodeRED_vocab131k.out
@@ -0,0 +1,2030 @@
++ '[' -z 14018953 ']'
++ source /etc/profile.d/modules.sh
+++ . /usr/share/modules/init/bash
++++ unset _mlshdbg
++++ '[' 0 = 1 ']'
++++ unset _mlre _mlIFS
++++ '[' -n x ']'
++++ _mlIFS=' 	
+'
++++ IFS=' '
++++ '[' -n '' ']'
+++++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash autoinit
++++ _mlcode='module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+MODULEPATH_modshare=/etc/environment-modules/modules:1:/usr/share/modules/\$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1; export MODULEPATH_modshare;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+LOADEDMODULES=; export LOADEDMODULES;
+MODULEPATH=/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/\$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles; export MODULEPATH;
+test 0;'
++++ _mlret=0
++++ '[' -n x ']'
++++ IFS=' 	
+'
++++ unset _mlIFS
++++ unset _mlre _mlv _mlrv
++++ '[' 0 -eq 0 ']'
++++ eval 'module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+MODULEPATH_modshare=/etc/environment-modules/modules:1:/usr/share/modules/\$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1; export MODULEPATH_modshare;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+LOADEDMODULES=; export LOADEDMODULES;
+MODULEPATH=/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/\$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles; export MODULEPATH;
+test 0;'
+++++ MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl
+++++ export MODULES_CMD
+++++ ENV=/usr/share/modules/init/profile.sh
+++++ export ENV
+++++ MODULEPATH_modshare='/etc/environment-modules/modules:1:/usr/share/modules/$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1'
+++++ export MODULEPATH_modshare
+++++ BASH_ENV=/usr/share/modules/init/bash
+++++ export BASH_ENV
+++++ MODULESHOME=/usr/share/modules
+++++ export MODULESHOME
+++++ LOADEDMODULES=
+++++ export LOADEDMODULES
+++++ MODULEPATH='/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles'
+++++ export MODULEPATH
+++++ test 0
++++ '[' 0 = 1 ']'
++++ '[' -t 2 ']'
++++ export -f module
++++ export -f switchml
++++ '[' 5 -ge 3 ']'
++++ [[ ehxB =~ i ]]
++++ [[ ! :/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin: =~ :/usr/bin: ]]
++++ '[' '!' -n '' ']'
++++ MANPATH=:
++++ export MANPATH
+++++ manpath
++++ [[ ! :/fsx/nouamane/miniconda/envs/2-1-cu121/man:/fsx/nouamane/miniconda/envs/2-1-cu121/share/man:/fsx/nouamane/miniconda/man:/fsx/nouamane/miniconda/share/man:/usr/local/man:/usr/local/share/man:/usr/share/man:/opt/slurm/share/man:/admin/home/nouamane/.local/share/man:/admin/home/nouamane/.fzf/man:: =~ :/usr/share/man: ]]
++++ unset _mlcode _mlret
++++ '[' -n '' ']'
++ module load cuda/12.1
++ unset _mlshdbg
++ '[' 0 = 1 ']'
++ unset _mlre _mlIFS
++ '[' -n x ']'
++ _mlIFS=' 	
+'
++ IFS=' '
++ '[' -n '' ']'
+++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash load cuda/12.1
++ eval 'CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include;' export 'CPATH;
+LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:;' export 'LD_LIBRARY_PATH;
+FI_EFA_FORK_SAFE=1;' export 'FI_EFA_FORK_SAFE;
+MANPATH=/usr/local/cuda-12.1/share/man::;' export 'MANPATH;
+LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64;' export 'LIBRARY_PATH;
+_LMFILES_=/usr/share/modules/modulefiles/cuda/12.1;' export '_LMFILES_;
+LOADEDMODULES=cuda/12.1;' export 'LOADEDMODULES;
+MPI_PATH=/opt/amazon/openmpi;' export 'MPI_PATH;
+NCCL_HOME_modshare=/opt/nccl/build:1;' export 'NCCL_HOME_modshare;
+NCCL_PROTO=simple;' export 'NCCL_PROTO;
+MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1;' export 'MANPATH_modshare;
+LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1;' export 'LIBRARY_PATH_modshare;
+NCCL_SOCKET_IFNAME=enp;' export 'NCCL_SOCKET_IFNAME;
+AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl;' export 'AWS_OFI_NCCL_HOME;
+NCCL_HOME=/opt/nccl/build;' export 'NCCL_HOME;
+FI_PROVIDER=efa;' export 'FI_PROVIDER;
+AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1;' export 'AWS_OFI_NCCL_HOME_modshare;
+CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1;' export 'CPATH_modshare;
+LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2::1:/opt/nccl/build/lib:1:/opt/aws-ofi-nccl/lib:2:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/efa/lib:1;' export 'LD_LIBRARY_PATH_modshare;
+FI_EFA_ENABLE_SHM_TRANSFER=1;' export 'FI_EFA_ENABLE_SHM_TRANSFER;
+_LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1;' export '_LMFILES__modshare;
+LOADEDMODULES_modshare=cuda/12.1:1;' export 'LOADEDMODULES_modshare;
+MPI_PATH_modshare=/opt/amazon/openmpi:1;' export 'MPI_PATH_modshare;
+PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin;' export 'PATH;
+CUDA_HOME=/usr/local/cuda-12.1;' export 'CUDA_HOME;
+PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/opt/slurm/bin:1:/usr/sbin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1;' export 'PATH_modshare;
+test' '0;'
+++ CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include
+++ export CPATH
+++ LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:
+++ export LD_LIBRARY_PATH
+++ FI_EFA_FORK_SAFE=1
+++ export FI_EFA_FORK_SAFE
+++ MANPATH=/usr/local/cuda-12.1/share/man::
+++ export MANPATH
+++ LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64
+++ export LIBRARY_PATH
+++ _LMFILES_=/usr/share/modules/modulefiles/cuda/12.1
+++ export _LMFILES_
+++ LOADEDMODULES=cuda/12.1
+++ export LOADEDMODULES
+++ MPI_PATH=/opt/amazon/openmpi
+++ export MPI_PATH
+++ NCCL_HOME_modshare=/opt/nccl/build:1
+++ export NCCL_HOME_modshare
+++ NCCL_PROTO=simple
+++ export NCCL_PROTO
+++ MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1
+++ export MANPATH_modshare
+++ LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1
+++ export LIBRARY_PATH_modshare
+++ NCCL_SOCKET_IFNAME=enp
+++ export NCCL_SOCKET_IFNAME
+++ AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl
+++ export AWS_OFI_NCCL_HOME
+++ NCCL_HOME=/opt/nccl/build
+++ export NCCL_HOME
+++ FI_PROVIDER=efa
+++ export FI_PROVIDER
+++ AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1
+++ export AWS_OFI_NCCL_HOME_modshare
+++ CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1
+++ export CPATH_modshare
+++ LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2::1:/opt/nccl/build/lib:1:/opt/aws-ofi-nccl/lib:2:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/efa/lib:1
+++ export LD_LIBRARY_PATH_modshare
+++ FI_EFA_ENABLE_SHM_TRANSFER=1
+++ export FI_EFA_ENABLE_SHM_TRANSFER
+++ _LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1
+++ export _LMFILES__modshare
+++ LOADEDMODULES_modshare=cuda/12.1:1
+++ export LOADEDMODULES_modshare
+++ MPI_PATH_modshare=/opt/amazon/openmpi:1
+++ export MPI_PATH_modshare
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
+++ export PATH
+++ CUDA_HOME=/usr/local/cuda-12.1
+++ export CUDA_HOME
+++ PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/opt/slurm/bin:1:/usr/sbin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1
+++ export PATH_modshare
+++ test 0
++ _mlstatus=0
++ '[' -n x ']'
++ IFS=' 	
+'
++ unset _mlre _mlv _mlrv _mlIFS
++ '[' -n '' ']'
++ unset _mlshdbg
++ return 0
++ source /fsx/nouamane/miniconda/bin/activate
+++ _CONDA_ROOT=/fsx/nouamane/miniconda
+++ . /fsx/nouamane/miniconda/etc/profile.d/conda.sh
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ '[' -z x ']'
+++ conda activate
+++ local cmd=activate
+++ case "$cmd" in
+++ __conda_activate activate
+++ '[' -n '' ']'
+++ local ask_conda
++++ PS1=
++++ __conda_exe shell.posix activate
++++ /fsx/nouamane/miniconda/bin/conda shell.posix activate
+Error while loading conda entry point: conda-libmamba-solver (module 'libmambapy' has no attribute 'QueryFormat')
+++ ask_conda='. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''2'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_1='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/etc/conda/activate.d/libxml2_activate.sh"'
+++ eval '. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''2'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_1='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/etc/conda/activate.d/libxml2_activate.sh"'
++++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh
+++++ test -n ''
+++++ unset XML_CATALOG_FILES
+++++ unset xml_catalog_files_libxml2
++++ PS1='(base) '
++++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
++++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
++++ export CONDA_PREFIX=/fsx/nouamane/miniconda
++++ CONDA_PREFIX=/fsx/nouamane/miniconda
++++ export CONDA_SHLVL=2
++++ CONDA_SHLVL=2
++++ export CONDA_DEFAULT_ENV=base
++++ CONDA_DEFAULT_ENV=base
++++ export 'CONDA_PROMPT_MODIFIER=(base) '
++++ CONDA_PROMPT_MODIFIER='(base) '
++++ export CONDA_PREFIX_1=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ CONDA_PREFIX_1=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ . /fsx/nouamane/miniconda/etc/conda/activate.d/libxml2_activate.sh
+++++ test -n ''
+++++ xml_catalog_files_libxml2=
+++++ XML_CATALOG_FILES=
+++++ conda_catalog_files=
+++++ ifs_libxml2=' 	
+'
+++++ IFS=' '
+++++ rem=/fsx/nouamane/miniconda
+++++ for pre in ${rem}
+++++ test '' = /fsx/nouamane/miniconda
+++++ conda_catalog_files=/fsx/nouamane/miniconda
+++++ rem=
+++++ IFS=' 	
+'
+++++ conda_catalog_files='file:///fsx/nouamane/miniconda/etc/xml/catalog file:///etc/xml/catalog'
+++++ export 'XML_CATALOG_FILES=file:///fsx/nouamane/miniconda/etc/xml/catalog file:///etc/xml/catalog'
+++++ XML_CATALOG_FILES='file:///fsx/nouamane/miniconda/etc/xml/catalog file:///etc/xml/catalog'
+++++ unset conda_catalog_files ifs_libxml2 rem
+++ __conda_hashr
+++ '[' -n '' ']'
+++ '[' -n '' ']'
+++ hash -r
++ conda activate 2-1-cu121
++ local cmd=activate
++ case "$cmd" in
++ __conda_activate activate 2-1-cu121
++ '[' -n '' ']'
++ local ask_conda
+++ PS1='(base) '
+++ __conda_exe shell.posix activate 2-1-cu121
+++ /fsx/nouamane/miniconda/bin/conda shell.posix activate 2-1-cu121
+Error while loading conda entry point: conda-libmamba-solver (module 'libmambapy' has no attribute 'QueryFormat')
++ ask_conda='. "/fsx/nouamane/miniconda/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_2='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
++ eval '. "/fsx/nouamane/miniconda/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_2='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
+++ . /fsx/nouamane/miniconda/etc/conda/deactivate.d/libxml2_deactivate.sh
++++ test -n ''
++++ unset XML_CATALOG_FILES
++++ unset xml_catalog_files_libxml2
+++ PS1='(2-1-cu121) '
+++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
+++ export CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ export CONDA_SHLVL=3
+++ CONDA_SHLVL=3
+++ export CONDA_DEFAULT_ENV=2-1-cu121
+++ CONDA_DEFAULT_ENV=2-1-cu121
+++ export 'CONDA_PROMPT_MODIFIER=(2-1-cu121) '
+++ CONDA_PROMPT_MODIFIER='(2-1-cu121) '
+++ export CONDA_PREFIX_2=/fsx/nouamane/miniconda
+++ CONDA_PREFIX_2=/fsx/nouamane/miniconda
+++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ export _CE_M=
+++ _CE_M=
+++ export _CE_CONDA=
+++ _CE_CONDA=
+++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh
++++ test -n ''
++++ xml_catalog_files_libxml2=
++++ XML_CATALOG_FILES=
++++ conda_catalog_files=
++++ ifs_libxml2=' 	
+'
++++ IFS=' '
++++ rem=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ for pre in ${rem}
++++ test '' = /fsx/nouamane/miniconda/envs/2-1-cu121
++++ conda_catalog_files=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ rem=
++++ IFS=' 	
+'
++++ conda_catalog_files='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ export 'XML_CATALOG_FILES=file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ XML_CATALOG_FILES='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ unset conda_catalog_files ifs_libxml2 rem
++ __conda_hashr
++ '[' -n '' ']'
++ '[' -n '' ']'
++ hash -r
++ export PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
++ PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
++ '[' -z '' ']'
+++ scontrol show hostnames 'ip-26-0-160-225,ip-26-0-167-[217,245],ip-26-0-168-[30,34,52,95,120,238],ip-26-0-169-[86,132,139,207,239,247],ip-26-0-170-31'
++ export 'NODELIST=ip-26-0-160-225
+ip-26-0-167-217
+ip-26-0-167-245
+ip-26-0-168-30
+ip-26-0-168-34
+ip-26-0-168-52
+ip-26-0-168-95
+ip-26-0-168-120
+ip-26-0-168-238
+ip-26-0-169-86
+ip-26-0-169-132
+ip-26-0-169-139
+ip-26-0-169-207
+ip-26-0-169-239
+ip-26-0-169-247
+ip-26-0-170-31'
++ NODELIST='ip-26-0-160-225
+ip-26-0-167-217
+ip-26-0-167-245
+ip-26-0-168-30
+ip-26-0-168-34
+ip-26-0-168-52
+ip-26-0-168-95
+ip-26-0-168-120
+ip-26-0-168-238
+ip-26-0-169-86
+ip-26-0-169-132
+ip-26-0-169-139
+ip-26-0-169-207
+ip-26-0-169-239
+ip-26-0-169-247
+ip-26-0-170-31'
+++ head -n1
+++ scontrol show hostnames 'ip-26-0-160-225,ip-26-0-167-[217,245],ip-26-0-168-[30,34,52,95,120,238],ip-26-0-169-[86,132,139,207,239,247],ip-26-0-170-31'
++ export MASTER_NODE=ip-26-0-160-225
++ MASTER_NODE=ip-26-0-160-225
++ export MASTER_PORT=12356
++ MASTER_PORT=12356
++ export NNODES=16
++ NNODES=16
++ export GPUS_PER_NODE=8
++ GPUS_PER_NODE=8
++ export WORLD_SIZE=128
++ WORLD_SIZE=128
++ export CUDA_DEVICE_MAX_CONNECTIONS=1
++ CUDA_DEVICE_MAX_CONNECTIONS=1
++ export NCCL_DEBUG=WARN
++ NCCL_DEBUG=WARN
++ export NANOTRON_BENCHMARK=1
++ NANOTRON_BENCHMARK=1
++ export WANDB_MODE=disabled
++ WANDB_MODE=disabled
++ export TORCH_NCCL_ASYNC_ERROR_HANDLING=1
++ TORCH_NCCL_ASYNC_ERROR_HANDLING=1
++ export TORCH_DISTRIBUTED_DEBUG=DETAIL
++ TORCH_DISTRIBUTED_DEBUG=DETAIL
++ '[' -z '' ']'
++ echo '=== GPU Topology ==='
+=== GPU Topology ===
++ nvidia-smi topo -m
+	GPU0	GPU1	GPU2	GPU3	GPU4	GPU5	GPU6	GPU7	CPU Affinity	NUMA Affinity	GPU NUMA ID
+GPU0	 X 	NV18	NV18	NV18	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU1	NV18	 X 	NV18	NV18	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU2	NV18	NV18	 X 	NV18	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU3	NV18	NV18	NV18	 X 	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU4	NV18	NV18	NV18	NV18	 X 	NV18	NV18	NV18	48-95	1		N/A
+GPU5	NV18	NV18	NV18	NV18	NV18	 X 	NV18	NV18	48-95	1		N/A
+GPU6	NV18	NV18	NV18	NV18	NV18	NV18	 X 	NV18	48-95	1		N/A
+GPU7	NV18	NV18	NV18	NV18	NV18	NV18	NV18	 X 	48-95	1		N/A
+
+Legend:
+
+  X    = Self
+  SYS  = Connection traversing PCIe as well as the SMP interconnect between NUMA nodes (e.g., QPI/UPI)
+  NODE = Connection traversing PCIe as well as the interconnect between PCIe Host Bridges within a NUMA node
+  PHB  = Connection traversing PCIe as well as a PCIe Host Bridge (typically the CPU)
+  PXB  = Connection traversing multiple PCIe bridges (without traversing the PCIe Host Bridge)
+  PIX  = Connection traversing at most a single PCIe bridge
+  NV#  = Connection traversing a bonded set of # NVLinks
++ echo ==================
+==================
++ export SRUN_ALLOC_ARGS=
++ SRUN_ALLOC_ARGS=
++ echo 'Master node: ip-26-0-160-225'
+Master node: ip-26-0-160-225
++ echo 'All nodes: ip-26-0-160-225
+ip-26-0-167-217
+ip-26-0-167-245
+ip-26-0-168-30
+ip-26-0-168-34
+ip-26-0-168-52
+ip-26-0-168-95
+ip-26-0-168-120
+ip-26-0-168-238
+ip-26-0-169-86
+ip-26-0-169-132
+ip-26-0-169-139
+ip-26-0-169-207
+ip-26-0-169-239
+ip-26-0-169-247
+ip-26-0-170-31'
+All nodes: ip-26-0-160-225
+ip-26-0-167-217
+ip-26-0-167-245
+ip-26-0-168-30
+ip-26-0-168-34
+ip-26-0-168-52
+ip-26-0-168-95
+ip-26-0-168-120
+ip-26-0-168-238
+ip-26-0-169-86
+ip-26-0-169-132
+ip-26-0-169-139
+ip-26-0-169-207
+ip-26-0-169-239
+ip-26-0-169-247
+ip-26-0-170-31
++ echo 'World size: 128'
+World size: 128
++ '[' -n '' ']'
++ srun --wait=0 --kill-on-bad-exit=1 torchrun --nnodes=16 --nproc_per_node=8 --rdzv_id=14018953 --rdzv_backend=c10d --rdzv_endpoint=ip-26-0-160-225:12356 --max_restarts 0 --rdzv_conf timeout=60 /fsx/nouamane/projects/nanotron/run_train.py --config-file benchmark/configs/config_8.86G_dp1_tp16_pp8_acc16_mbs16_seq4096_zero0_tpmodeRED_vocab131k.yaml
+[2025-01-03 02:29:40,744] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 02:29:40,744] torch.distributed.run: [WARNING] 
+[2025-01-03 02:29:40,744] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 02:29:40,744] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 02:29:40,744] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 02:29:40,744] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 02:29:40,744] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 02:29:40,744] torch.distributed.run: [WARNING] 
+[2025-01-03 02:29:40,744] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 02:29:40,744] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 02:29:40,744] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 02:29:40,744] torch.distributed.run: [WARNING] 
+[2025-01-03 02:29:40,744] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 02:29:40,744] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 02:29:40,744] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 02:29:40,744] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 02:29:40,744] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 02:29:40,744] torch.distributed.run: [WARNING] 
+[2025-01-03 02:29:40,744] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 02:29:40,744] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 02:29:40,744] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 02:29:40,744] torch.distributed.run: [WARNING] 
+[2025-01-03 02:29:40,744] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 02:29:40,744] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 02:29:40,744] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 02:29:40,744] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 02:29:40,744] torch.distributed.run: [WARNING] 
+[2025-01-03 02:29:40,744] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 02:29:40,744] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 02:29:40,744] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 02:29:40,746] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 02:29:40,746] torch.distributed.run: [WARNING] 
+[2025-01-03 02:29:40,746] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 02:29:40,746] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 02:29:40,746] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 02:29:40,746] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 02:29:40,747] torch.distributed.run: [WARNING] 
+[2025-01-03 02:29:40,747] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 02:29:40,747] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 02:29:40,747] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 02:29:40,746] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 02:29:40,746] torch.distributed.run: [WARNING] 
+[2025-01-03 02:29:40,746] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 02:29:40,746] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 02:29:40,746] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 02:29:40,749] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 02:29:40,749] torch.distributed.run: [WARNING] 
+[2025-01-03 02:29:40,749] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 02:29:40,749] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 02:29:40,749] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 02:29:40,751] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 02:29:40,751] torch.distributed.run: [WARNING] 
+[2025-01-03 02:29:40,751] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 02:29:40,751] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 02:29:40,751] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 02:29:40,751] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 02:29:40,752] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 02:29:40,752] torch.distributed.run: [WARNING] 
+[2025-01-03 02:29:40,752] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 02:29:40,752] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 02:29:40,752] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 02:29:40,752] torch.distributed.run: [WARNING] 
+[2025-01-03 02:29:40,752] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 02:29:40,752] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 02:29:40,752] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 02:29:40,758] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 02:29:40,758] torch.distributed.run: [WARNING] 
+[2025-01-03 02:29:40,758] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 02:29:40,758] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 02:29:40,758] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 02:29:40,899] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 02:29:40,899] torch.distributed.run: [WARNING] 
+[2025-01-03 02:29:40,899] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 02:29:40,899] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 02:29:40,899] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 02:29:40,942] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 02:29:40,943] torch.distributed.run: [WARNING] 
+[2025-01-03 02:29:40,943] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 02:29:40,943] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 02:29:40,943] torch.distributed.run: [WARNING] *****************************************
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+01/03/2025 02:30:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: Config:
+01/03/2025 02:30:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: Config(general=GeneralArgs(project='debug',
+01/03/2025 02:30:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                            run='8.86G_dp1_tp16_pp8_acc16_mbs16_seq4096_zero0_tpmodeRED_vocab131k',
+01/03/2025 02:30:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                            seed=42,
+01/03/2025 02:30:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                            step=None,
+01/03/2025 02:30:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                            consumed_train_samples=None,
+01/03/2025 02:30:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                            benchmark_csv_path=PosixPath('benchmark/results/bench_final2.csv'),
+01/03/2025 02:30:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                            ignore_sanity_checks=True),
+01/03/2025 02:30:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:        parallelism=ParallelismArgs(dp=1,
+01/03/2025 02:30:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                    pp=8,
+01/03/2025 02:30:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                    tp=16,
+01/03/2025 02:30:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                    pp_engine=<nanotron.parallel.pipeline_parallel.engine.OneForwardOneBackwardPipelineEngine object at 0x7f92d9434c10>,
+01/03/2025 02:30:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                    tp_mode=<TensorParallelLinearMode.REDUCE_SCATTER: 2>,
+01/03/2025 02:30:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                    tp_linear_async_communication=True,
+01/03/2025 02:30:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                    recompute_layer=False,
+01/03/2025 02:30:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                    tp_recompute_allgather=True,
+01/03/2025 02:30:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                    expert_parallel_size=1),
+01/03/2025 02:30:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:        model=ModelArgs(model_config=LlamaConfig(bos_token_id=0,
+01/03/2025 02:30:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 eos_token_id=0,
+01/03/2025 02:30:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 hidden_act='silu',
+01/03/2025 02:30:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 hidden_size=4096,
+01/03/2025 02:30:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 initializer_range=0.02,
+01/03/2025 02:30:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 intermediate_size=14336,
+01/03/2025 02:30:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 is_llama_config=True,
+01/03/2025 02:30:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 max_position_embeddings=4096,
+01/03/2025 02:30:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 num_attention_heads=32,
+01/03/2025 02:30:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 num_hidden_layers=32,
+01/03/2025 02:30:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 num_key_value_heads=32,
+01/03/2025 02:30:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 pad_token_id=None,
+01/03/2025 02:30:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 pretraining_tp=1,
+01/03/2025 02:30:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 rms_norm_eps=1e-05,
+01/03/2025 02:30:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 rope_scaling=None,
+01/03/2025 02:30:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 rope_theta=10000.0,
+01/03/2025 02:30:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 rope_interleaved=False,
+01/03/2025 02:30:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 tie_word_embeddings=False,
+01/03/2025 02:30:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 use_cache=True,
+01/03/2025 02:30:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 vocab_size=131072),
+01/03/2025 02:30:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                        init_method=RandomInit(std=0.02),
+01/03/2025 02:30:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                        dtype=torch.bfloat16,
+01/03/2025 02:30:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                        make_vocab_size_divisible_by=1,
+01/03/2025 02:30:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                        ddp_bucket_cap_mb=25),
+01/03/2025 02:30:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:        tokenizer=TokenizerArgs(tokenizer_name_or_path='robot-test/dummy-tokenizer-wordlevel',
+01/03/2025 02:30:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                tokenizer_revision=None,
+01/03/2025 02:30:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                tokenizer_max_length=None),
+01/03/2025 02:30:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:        checkpoints=CheckpointsArgs(checkpoints_path=PosixPath('checkpoints'),
+01/03/2025 02:30:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                    checkpoint_interval=10000,
+01/03/2025 02:30:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                    save_initial_state=False,
+01/03/2025 02:30:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                    save_final_state=False,
+01/03/2025 02:30:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                    resume_checkpoint_path=None,
+01/03/2025 02:30:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                    checkpoints_path_is_shared_file_system=False),
+01/03/2025 02:30:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:        logging=LoggingArgs(log_level='info',
+01/03/2025 02:30:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                            log_level_replica='info',
+01/03/2025 02:30:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                            iteration_step_info_interval=1),
+01/03/2025 02:30:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:        tokens=TokensArgs(sequence_length=4096,
+01/03/2025 02:30:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                          train_steps=100,
+01/03/2025 02:30:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                          micro_batch_size=16,
+01/03/2025 02:30:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                          batch_accumulation_per_replica=16,
+01/03/2025 02:30:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                          val_check_interval=100,
+01/03/2025 02:30:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                          limit_val_batches=0,
+01/03/2025 02:30:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                          limit_test_batches=0),
+01/03/2025 02:30:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:        optimizer=OptimizerArgs(optimizer_factory=AdamWOptimizerArgs(adam_eps=1e-08,
+01/03/2025 02:30:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                                     adam_beta1=0.9,
+01/03/2025 02:30:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                                     adam_beta2=0.95,
+01/03/2025 02:30:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                                     torch_adam_is_fused=True,
+01/03/2025 02:30:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                                     name='adamW'),
+01/03/2025 02:30:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                zero_stage=0,
+01/03/2025 02:30:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                weight_decay=0.01,
+01/03/2025 02:30:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                clip_grad=1.0,
+01/03/2025 02:30:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                accumulate_grad_in_fp32=True,
+01/03/2025 02:30:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                learning_rate_scheduler=LRSchedulerArgs(learning_rate=0.0003,
+01/03/2025 02:30:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                                        lr_warmup_steps=2,
+01/03/2025 02:30:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                                        lr_warmup_style='linear',
+01/03/2025 02:30:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                                        lr_decay_style='cosine',
+01/03/2025 02:30:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                                        lr_decay_steps=13,
+01/03/2025 02:30:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                                        lr_decay_starting_step=None,
+01/03/2025 02:30:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                                        min_decay_lr=1e-05)),
+01/03/2025 02:30:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:        data_stages=[DatasetStageArgs(name='Stable Training Stage',
+01/03/2025 02:30:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                      start_training_step=1,
+01/03/2025 02:30:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                      data=DataArgs(dataset=None,
+01/03/2025 02:30:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                    seed=42,
+01/03/2025 02:30:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                    num_loading_workers=1))],
+01/03/2025 02:30:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:        profiler=None,
+01/03/2025 02:30:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:        lighteval=None,
+01/03/2025 02:30:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:        s3_upload=None)
+01/03/2025 02:30:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: Model Config:
+01/03/2025 02:30:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: LlamaConfig(bos_token_id=0,
+01/03/2025 02:30:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             eos_token_id=0,
+01/03/2025 02:30:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             hidden_act='silu',
+01/03/2025 02:30:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             hidden_size=4096,
+01/03/2025 02:30:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             initializer_range=0.02,
+01/03/2025 02:30:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             intermediate_size=14336,
+01/03/2025 02:30:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             is_llama_config=True,
+01/03/2025 02:30:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             max_position_embeddings=4096,
+01/03/2025 02:30:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             num_attention_heads=32,
+01/03/2025 02:30:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             num_hidden_layers=32,
+01/03/2025 02:30:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             num_key_value_heads=32,
+01/03/2025 02:30:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             pad_token_id=None,
+01/03/2025 02:30:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             pretraining_tp=1,
+01/03/2025 02:30:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             rms_norm_eps=1e-05,
+01/03/2025 02:30:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             rope_scaling=None,
+01/03/2025 02:30:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             rope_theta=10000.0,
+01/03/2025 02:30:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             rope_interleaved=False,
+01/03/2025 02:30:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             tie_word_embeddings=False,
+01/03/2025 02:30:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             use_cache=True,
+01/03/2025 02:30:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             vocab_size=131072)
+01/03/2025 02:30:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: Building model..
+01/03/2025 02:30:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: Initialize RoPE Theta = 10000.0
+01/03/2025 02:30:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: Setting PP block ranks...
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+01/03/2025 02:30:39 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: Total number of parameters: 8.86G (16904.12MiB)
+01/03/2025 02:30:39 [INFO|DP=0|PP=4|TP=0|ip-26-0-168-95]: Local number of parameters: 60.9M (116.06MiB)
+01/03/2025 02:30:39 [INFO|DP=0|PP=2|TP=0|ip-26-0-168-238]: Local number of parameters: 60.9M (116.06MiB)
+01/03/2025 02:30:39 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: Local number of parameters: 110M (209.08MiB)
+01/03/2025 02:30:39 [INFO|DP=0|PP=6|TP=0|ip-26-0-169-239]: Local number of parameters: 60.9M (116.06MiB)
+01/03/2025 02:30:39 [INFO|DP=0|PP=2|TP=0|ip-26-0-168-238]: [After model building] Memory usage: 117.08MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:30:39 [INFO|DP=0|PP=4|TP=0|ip-26-0-168-95]: [After model building] Memory usage: 117.08MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:30:39 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [After model building] Memory usage: 211.09MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:30:39 [INFO|DP=0|PP=6|TP=0|ip-26-0-169-239]: [After model building] Memory usage: 117.08MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:30:39 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: No checkpoint path provided.
+01/03/2025 02:30:39 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: Parametrizing model parameters using StandardParametrizator
+01/03/2025 02:30:39 [INFO|DP=0|PP=7|TP=0|ip-26-0-169-86]: Local number of parameters: 64M (122.04MiB)
+01/03/2025 02:30:39 [INFO|DP=0|PP=7|TP=0|ip-26-0-169-86]: [After model building] Memory usage: 122.05MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:30:39 [INFO|DP=0|PP=0|TP=6|ip-26-0-160-225]: Local number of parameters: 110M (209.08MiB)
+01/03/2025 02:30:39 [INFO|DP=0|PP=5|TP=0|ip-26-0-169-139]: Local number of parameters: 60.9M (116.06MiB)
+01/03/2025 02:30:39 [INFO|DP=0|PP=0|TP=7|ip-26-0-160-225]: Local number of parameters: 110M (209.08MiB)
+01/03/2025 02:30:39 [INFO|DP=0|PP=0|TP=3|ip-26-0-160-225]: Local number of parameters: 110M (209.08MiB)
+01/03/2025 02:30:39 [INFO|DP=0|PP=0|TP=6|ip-26-0-160-225]: [After model building] Memory usage: 211.09MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:30:39 [INFO|DP=0|PP=0|TP=2|ip-26-0-160-225]: Local number of parameters: 110M (209.08MiB)
+01/03/2025 02:30:39 [INFO|DP=0|PP=0|TP=7|ip-26-0-160-225]: [After model building] Memory usage: 211.09MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:30:39 [INFO|DP=0|PP=5|TP=0|ip-26-0-169-139]: [After model building] Memory usage: 117.08MiB. Peak allocated: 5392.00MiB Peak reserved: 9494.00MiB
+01/03/2025 02:30:39 [INFO|DP=0|PP=0|TP=3|ip-26-0-160-225]: [After model building] Memory usage: 211.09MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:30:39 [INFO|DP=0|PP=0|TP=2|ip-26-0-160-225]: [After model building] Memory usage: 211.09MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:30:39 [INFO|DP=0|PP=0|TP=4|ip-26-0-160-225]: Local number of parameters: 110M (209.08MiB)
+01/03/2025 02:30:39 [INFO|DP=0|PP=0|TP=4|ip-26-0-160-225]: [After model building] Memory usage: 211.09MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:30:39 [INFO|DP=0|PP=0|TP=8|ip-26-0-167-217]: Local number of parameters: 110M (209.08MiB)
+01/03/2025 02:30:39 [INFO|DP=0|PP=0|TP=8|ip-26-0-167-217]: [After model building] Memory usage: 211.09MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:30:39 [INFO|DP=0|PP=3|TP=0|ip-26-0-168-34]: Local number of parameters: 76.1M (145.08MiB)
+01/03/2025 02:30:39 [INFO|DP=0|PP=6|TP=8|ip-26-0-169-247]: Local number of parameters: 60.9M (116.06MiB)
+01/03/2025 02:30:39 [INFO|DP=0|PP=4|TP=8|ip-26-0-169-132]: Local number of parameters: 60.9M (116.06MiB)
+01/03/2025 02:30:39 [INFO|DP=0|PP=1|TP=0|ip-26-0-167-245]: Local number of parameters: 60.9M (116.06MiB)
+01/03/2025 02:30:39 [INFO|DP=0|PP=3|TP=0|ip-26-0-168-34]: [After model building] Memory usage: 146.09MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:30:39 [INFO|DP=0|PP=6|TP=8|ip-26-0-169-247]: [After model building] Memory usage: 117.08MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:30:39 [INFO|DP=0|PP=4|TP=8|ip-26-0-169-132]: [After model building] Memory usage: 117.08MiB. Peak allocated: 5392.00MiB Peak reserved: 9494.00MiB
+01/03/2025 02:30:39 [INFO|DP=0|PP=1|TP=0|ip-26-0-167-245]: [After model building] Memory usage: 117.08MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:30:39 [INFO|DP=0|PP=4|TP=6|ip-26-0-168-95]: Local number of parameters: 60.9M (116.06MiB)
+01/03/2025 02:30:39 [INFO|DP=0|PP=1|TP=6|ip-26-0-167-245]: Local number of parameters: 60.9M (116.06MiB)
+01/03/2025 02:30:39 [INFO|DP=0|PP=2|TP=6|ip-26-0-168-238]: Local number of parameters: 60.9M (116.06MiB)
+01/03/2025 02:30:39 [INFO|DP=0|PP=4|TP=6|ip-26-0-168-95]: [After model building] Memory usage: 117.08MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:30:39 [INFO|DP=0|PP=1|TP=6|ip-26-0-167-245]: [After model building] Memory usage: 117.08MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:30:39 [INFO|DP=0|PP=4|TP=3|ip-26-0-168-95]: Local number of parameters: 60.9M (116.06MiB)
+01/03/2025 02:30:39 [INFO|DP=0|PP=2|TP=3|ip-26-0-168-238]: Local number of parameters: 60.9M (116.06MiB)
+01/03/2025 02:30:39 [INFO|DP=0|PP=1|TP=7|ip-26-0-167-245]: Local number of parameters: 60.9M (116.06MiB)
+01/03/2025 02:30:39 [INFO|DP=0|PP=1|TP=3|ip-26-0-167-245]: Local number of parameters: 60.9M (116.06MiB)
+01/03/2025 02:30:39 [INFO|DP=0|PP=2|TP=7|ip-26-0-168-238]: Local number of parameters: 60.9M (116.06MiB)
+01/03/2025 02:30:39 [INFO|DP=0|PP=6|TP=3|ip-26-0-169-239]: Local number of parameters: 60.9M (116.06MiB)
+01/03/2025 02:30:39 [INFO|DP=0|PP=2|TP=6|ip-26-0-168-238]: [After model building] Memory usage: 117.08MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:30:39 [INFO|DP=0|PP=7|TP=7|ip-26-0-169-86]: Local number of parameters: 64M (122.04MiB)
+01/03/2025 02:30:39 [INFO|DP=0|PP=7|TP=3|ip-26-0-169-86]: Local number of parameters: 64M (122.04MiB)
+01/03/2025 02:30:39 [INFO|DP=0|PP=4|TP=2|ip-26-0-168-95]: Local number of parameters: 60.9M (116.06MiB)
+01/03/2025 02:30:39 [INFO|DP=0|PP=6|TP=7|ip-26-0-169-239]: Local number of parameters: 60.9M (116.06MiB)
+01/03/2025 02:30:39 [INFO|DP=0|PP=1|TP=2|ip-26-0-167-245]: Local number of parameters: 60.9M (116.06MiB)
+01/03/2025 02:30:39 [INFO|DP=0|PP=4|TP=7|ip-26-0-168-95]: Local number of parameters: 60.9M (116.06MiB)
+01/03/2025 02:30:39 [INFO|DP=0|PP=6|TP=2|ip-26-0-169-239]: Local number of parameters: 60.9M (116.06MiB)
+01/03/2025 02:30:39 [INFO|DP=0|PP=7|TP=2|ip-26-0-169-86]: Local number of parameters: 64M (122.04MiB)
+01/03/2025 02:30:39 [INFO|DP=0|PP=4|TP=3|ip-26-0-168-95]: [After model building] Memory usage: 117.08MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:30:39 [INFO|DP=0|PP=1|TP=7|ip-26-0-167-245]: [After model building] Memory usage: 117.08MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:30:39 [INFO|DP=0|PP=2|TP=3|ip-26-0-168-238]: [After model building] Memory usage: 117.08MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:30:39 [INFO|DP=0|PP=2|TP=7|ip-26-0-168-238]: [After model building] Memory usage: 117.08MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:30:39 [INFO|DP=0|PP=1|TP=3|ip-26-0-167-245]: [After model building] Memory usage: 117.08MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:30:39 [INFO|DP=0|PP=2|TP=2|ip-26-0-168-238]: Local number of parameters: 60.9M (116.06MiB)
+01/03/2025 02:30:39 [INFO|DP=0|PP=6|TP=3|ip-26-0-169-239]: [After model building] Memory usage: 117.08MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:30:39 [INFO|DP=0|PP=4|TP=2|ip-26-0-168-95]: [After model building] Memory usage: 117.08MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:30:39 [INFO|DP=0|PP=7|TP=3|ip-26-0-169-86]: [After model building] Memory usage: 122.05MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:30:39 [INFO|DP=0|PP=7|TP=7|ip-26-0-169-86]: [After model building] Memory usage: 122.05MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:30:39 [INFO|DP=0|PP=6|TP=7|ip-26-0-169-239]: [After model building] Memory usage: 117.08MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:30:39 [INFO|DP=0|PP=2|TP=8|ip-26-0-168-30]: Local number of parameters: 60.9M (116.06MiB)
+01/03/2025 02:30:39 [INFO|DP=0|PP=1|TP=2|ip-26-0-167-245]: [After model building] Memory usage: 117.08MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:30:39 [INFO|DP=0|PP=7|TP=2|ip-26-0-169-86]: [After model building] Memory usage: 122.05MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:30:39 [INFO|DP=0|PP=6|TP=6|ip-26-0-169-239]: Local number of parameters: 60.9M (116.06MiB)
+01/03/2025 02:30:39 [INFO|DP=0|PP=6|TP=2|ip-26-0-169-239]: [After model building] Memory usage: 117.08MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:30:39 [INFO|DP=0|PP=4|TP=7|ip-26-0-168-95]: [After model building] Memory usage: 117.08MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:30:39 [INFO|DP=0|PP=7|TP=6|ip-26-0-169-86]: Local number of parameters: 64M (122.04MiB)
+01/03/2025 02:30:39 [INFO|DP=0|PP=2|TP=4|ip-26-0-168-238]: Local number of parameters: 60.9M (116.06MiB)
+01/03/2025 02:30:39 [INFO|DP=0|PP=0|TP=1|ip-26-0-160-225]: Local number of parameters: 110M (209.08MiB)
+01/03/2025 02:30:39 [INFO|DP=0|PP=2|TP=2|ip-26-0-168-238]: [After model building] Memory usage: 117.08MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:30:39 [INFO|DP=0|PP=6|TP=6|ip-26-0-169-239]: [After model building] Memory usage: 117.08MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:30:39 [INFO|DP=0|PP=1|TP=4|ip-26-0-167-245]: Local number of parameters: 60.9M (116.06MiB)
+01/03/2025 02:30:39 [INFO|DP=0|PP=7|TP=4|ip-26-0-169-86]: Local number of parameters: 64M (122.04MiB)
+01/03/2025 02:30:39 [INFO|DP=0|PP=6|TP=4|ip-26-0-169-239]: Local number of parameters: 60.9M (116.06MiB)
+01/03/2025 02:30:39 [INFO|DP=0|PP=7|TP=6|ip-26-0-169-86]: [After model building] Memory usage: 122.05MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:30:39 [INFO|DP=0|PP=2|TP=4|ip-26-0-168-238]: [After model building] Memory usage: 117.08MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:30:39 [INFO|DP=0|PP=2|TP=8|ip-26-0-168-30]: [After model building] Memory usage: 117.08MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:30:39 [INFO|DP=0|PP=0|TP=5|ip-26-0-160-225]: Local number of parameters: 110M (209.08MiB)
+01/03/2025 02:30:39 [INFO|DP=0|PP=0|TP=1|ip-26-0-160-225]: [After model building] Memory usage: 211.09MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:30:39 [INFO|DP=0|PP=7|TP=4|ip-26-0-169-86]: [After model building] Memory usage: 122.05MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:30:39 [INFO|DP=0|PP=6|TP=4|ip-26-0-169-239]: [After model building] Memory usage: 117.08MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:30:39 [INFO|DP=0|PP=1|TP=4|ip-26-0-167-245]: [After model building] Memory usage: 117.08MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:30:39 [INFO|DP=0|PP=0|TP=5|ip-26-0-160-225]: [After model building] Memory usage: 211.09MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:30:39 [INFO|DP=0|PP=4|TP=4|ip-26-0-168-95]: Local number of parameters: 60.9M (116.06MiB)
+01/03/2025 02:30:39 [INFO|DP=0|PP=5|TP=8|ip-26-0-169-207]: Local number of parameters: 60.9M (116.06MiB)
+NCCL version 2.18.5+cuda12.2
+01/03/2025 02:30:39 [INFO|DP=0|PP=4|TP=4|ip-26-0-168-95]: [After model building] Memory usage: 117.08MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+NCCL version 2.18.5+cuda12.2
+01/03/2025 02:30:39 [INFO|DP=0|PP=5|TP=8|ip-26-0-169-207]: [After model building] Memory usage: 117.08MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:30:39 [INFO|DP=0|PP=0|TP=13|ip-26-0-167-217]: Local number of parameters: 110M (209.08MiB)
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+01/03/2025 02:30:39 [INFO|DP=0|PP=3|TP=6|ip-26-0-168-34]: Local number of parameters: 76.1M (145.08MiB)
+01/03/2025 02:30:39 [INFO|DP=0|PP=0|TP=13|ip-26-0-167-217]: [After model building] Memory usage: 211.09MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+01/03/2025 02:30:39 [INFO|DP=0|PP=3|TP=6|ip-26-0-168-34]: [After model building] Memory usage: 146.09MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:30:39 [INFO|DP=0|PP=5|TP=3|ip-26-0-169-139]: Local number of parameters: 60.9M (116.06MiB)
+01/03/2025 02:30:39 [INFO|DP=0|PP=7|TP=8|ip-26-0-170-31]: Local number of parameters: 64M (122.04MiB)
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+01/03/2025 02:30:39 [INFO|DP=0|PP=3|TP=7|ip-26-0-168-34]: Local number of parameters: 76.1M (145.08MiB)
+01/03/2025 02:30:39 [INFO|DP=0|PP=0|TP=12|ip-26-0-167-217]: Local number of parameters: 110M (209.08MiB)
+01/03/2025 02:30:39 [INFO|DP=0|PP=5|TP=7|ip-26-0-169-139]: Local number of parameters: 60.9M (116.06MiB)
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+01/03/2025 02:30:39 [INFO|DP=0|PP=5|TP=2|ip-26-0-169-139]: Local number of parameters: 60.9M (116.06MiB)
+NCCL version 2.18.5+cuda12.2
+01/03/2025 02:30:39 [INFO|DP=0|PP=5|TP=3|ip-26-0-169-139]: [After model building] Memory usage: 117.08MiB. Peak allocated: 5392.00MiB Peak reserved: 9494.00MiB
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+01/03/2025 02:30:39 [INFO|DP=0|PP=4|TP=1|ip-26-0-168-95]: Local number of parameters: 60.9M (116.06MiB)
+01/03/2025 02:30:39 [INFO|DP=0|PP=3|TP=2|ip-26-0-168-34]: Local number of parameters: 76.1M (145.08MiB)
+01/03/2025 02:30:39 [INFO|DP=0|PP=1|TP=1|ip-26-0-167-245]: Local number of parameters: 60.9M (116.06MiB)
+01/03/2025 02:30:39 [INFO|DP=0|PP=3|TP=7|ip-26-0-168-34]: [After model building] Memory usage: 146.09MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:30:39 [INFO|DP=0|PP=7|TP=8|ip-26-0-170-31]: [After model building] Memory usage: 122.05MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:30:39 [INFO|DP=0|PP=7|TP=1|ip-26-0-169-86]: Local number of parameters: 64M (122.04MiB)
+01/03/2025 02:30:39 [INFO|DP=0|PP=2|TP=1|ip-26-0-168-238]: Local number of parameters: 60.9M (116.06MiB)
+01/03/2025 02:30:39 [INFO|DP=0|PP=5|TP=1|ip-26-0-169-139]: Local number of parameters: 60.9M (116.06MiB)
+NCCL version 2.18.5+cuda12.2
+01/03/2025 02:30:39 [INFO|DP=0|PP=5|TP=6|ip-26-0-169-139]: Local number of parameters: 60.9M (116.06MiB)
+01/03/2025 02:30:39 [INFO|DP=0|PP=3|TP=8|ip-26-0-168-52]: Local number of parameters: 76.1M (145.08MiB)
+01/03/2025 02:30:39 [INFO|DP=0|PP=5|TP=7|ip-26-0-169-139]: [After model building] Memory usage: 117.08MiB. Peak allocated: 5392.00MiB Peak reserved: 9494.00MiB
+01/03/2025 02:30:39 [INFO|DP=0|PP=0|TP=15|ip-26-0-167-217]: Local number of parameters: 110M (209.08MiB)
+01/03/2025 02:30:39 [INFO|DP=0|PP=5|TP=2|ip-26-0-169-139]: [After model building] Memory usage: 117.08MiB. Peak allocated: 5392.00MiB Peak reserved: 9494.00MiB
+01/03/2025 02:30:39 [INFO|DP=0|PP=0|TP=12|ip-26-0-167-217]: [After model building] Memory usage: 211.09MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+NCCL version 2.18.5+cuda12.2
+01/03/2025 02:30:39 [INFO|DP=0|PP=4|TP=1|ip-26-0-168-95]: [After model building] Memory usage: 117.08MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:30:39 [INFO|DP=0|PP=6|TP=1|ip-26-0-169-239]: Local number of parameters: 60.9M (116.06MiB)
+NCCL version 2.18.5+cuda12.2
+01/03/2025 02:30:39 [INFO|DP=0|PP=3|TP=2|ip-26-0-168-34]: [After model building] Memory usage: 146.09MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:30:39 [INFO|DP=0|PP=7|TP=1|ip-26-0-169-86]: [After model building] Memory usage: 122.05MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:30:39 [INFO|DP=0|PP=2|TP=1|ip-26-0-168-238]: [After model building] Memory usage: 117.08MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:30:39 [INFO|DP=0|PP=1|TP=1|ip-26-0-167-245]: [After model building] Memory usage: 117.08MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:30:39 [INFO|DP=0|PP=5|TP=1|ip-26-0-169-139]: [After model building] Memory usage: 117.08MiB. Peak allocated: 5392.00MiB Peak reserved: 9494.00MiB
+01/03/2025 02:30:39 [INFO|DP=0|PP=3|TP=8|ip-26-0-168-52]: [After model building] Memory usage: 146.09MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:30:39 [INFO|DP=0|PP=3|TP=4|ip-26-0-168-34]: Local number of parameters: 76.1M (145.08MiB)
+01/03/2025 02:30:39 [INFO|DP=0|PP=0|TP=9|ip-26-0-167-217]: Local number of parameters: 110M (209.08MiB)
+01/03/2025 02:30:39 [INFO|DP=0|PP=0|TP=15|ip-26-0-167-217]: [After model building] Memory usage: 211.09MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:30:39 [INFO|DP=0|PP=3|TP=3|ip-26-0-168-34]: Local number of parameters: 76.1M (145.08MiB)
+01/03/2025 02:30:39 [INFO|DP=0|PP=5|TP=6|ip-26-0-169-139]: [After model building] Memory usage: 117.08MiB. Peak allocated: 5392.00MiB Peak reserved: 9494.00MiB
+01/03/2025 02:30:39 [INFO|DP=0|PP=1|TP=5|ip-26-0-167-245]: Local number of parameters: 60.9M (116.06MiB)
+01/03/2025 02:30:39 [INFO|DP=0|PP=7|TP=5|ip-26-0-169-86]: Local number of parameters: 64M (122.04MiB)
+01/03/2025 02:30:39 [INFO|DP=0|PP=5|TP=5|ip-26-0-169-139]: Local number of parameters: 60.9M (116.06MiB)
+01/03/2025 02:30:39 [INFO|DP=0|PP=2|TP=5|ip-26-0-168-238]: Local number of parameters: 60.9M (116.06MiB)
+01/03/2025 02:30:39 [INFO|DP=0|PP=6|TP=5|ip-26-0-169-239]: Local number of parameters: 60.9M (116.06MiB)
+01/03/2025 02:30:39 [INFO|DP=0|PP=6|TP=1|ip-26-0-169-239]: [After model building] Memory usage: 117.08MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:30:39 [INFO|DP=0|PP=3|TP=3|ip-26-0-168-34]: [After model building] Memory usage: 146.09MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:30:39 [INFO|DP=0|PP=0|TP=9|ip-26-0-167-217]: [After model building] Memory usage: 211.09MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:30:39 [INFO|DP=0|PP=3|TP=4|ip-26-0-168-34]: [After model building] Memory usage: 146.09MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:30:39 [INFO|DP=0|PP=4|TP=5|ip-26-0-168-95]: Local number of parameters: 60.9M (116.06MiB)
+01/03/2025 02:30:39 [INFO|DP=0|PP=1|TP=5|ip-26-0-167-245]: [After model building] Memory usage: 117.08MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:30:39 [INFO|DP=0|PP=7|TP=5|ip-26-0-169-86]: [After model building] Memory usage: 122.05MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:30:39 [INFO|DP=0|PP=2|TP=5|ip-26-0-168-238]: [After model building] Memory usage: 117.08MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:30:39 [INFO|DP=0|PP=5|TP=5|ip-26-0-169-139]: [After model building] Memory usage: 117.08MiB. Peak allocated: 5392.00MiB Peak reserved: 9494.00MiB
+01/03/2025 02:30:39 [INFO|DP=0|PP=6|TP=5|ip-26-0-169-239]: [After model building] Memory usage: 117.08MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+NCCL version 2.18.5+cuda12.2
+01/03/2025 02:30:39 [INFO|DP=0|PP=4|TP=5|ip-26-0-168-95]: [After model building] Memory usage: 117.08MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:30:39 [INFO|DP=0|PP=5|TP=4|ip-26-0-169-139]: Local number of parameters: 60.9M (116.06MiB)
+01/03/2025 02:30:39 [INFO|DP=0|PP=0|TP=10|ip-26-0-167-217]: Local number of parameters: 110M (209.08MiB)
+01/03/2025 02:30:39 [INFO|DP=0|PP=5|TP=4|ip-26-0-169-139]: [After model building] Memory usage: 117.08MiB. Peak allocated: 5392.00MiB Peak reserved: 9494.00MiB
+01/03/2025 02:30:39 [INFO|DP=0|PP=1|TP=8|ip-26-0-168-120]: Local number of parameters: 60.9M (116.06MiB)
+01/03/2025 02:30:39 [INFO|DP=0|PP=0|TP=14|ip-26-0-167-217]: Local number of parameters: 110M (209.08MiB)
+01/03/2025 02:30:39 [INFO|DP=0|PP=0|TP=10|ip-26-0-167-217]: [After model building] Memory usage: 211.09MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+NCCL version 2.18.5+cuda12.2
+01/03/2025 02:30:39 [INFO|DP=0|PP=1|TP=8|ip-26-0-168-120]: [After model building] Memory usage: 117.08MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+01/03/2025 02:30:39 [INFO|DP=0|PP=0|TP=11|ip-26-0-167-217]: Local number of parameters: 110M (209.08MiB)
+01/03/2025 02:30:39 [INFO|DP=0|PP=0|TP=14|ip-26-0-167-217]: [After model building] Memory usage: 211.09MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+01/03/2025 02:30:39 [INFO|DP=0|PP=0|TP=11|ip-26-0-167-217]: [After model building] Memory usage: 211.09MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+01/03/2025 02:30:39 [INFO|DP=0|PP=3|TP=5|ip-26-0-168-34]: Local number of parameters: 76.1M (145.08MiB)
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+01/03/2025 02:30:39 [INFO|DP=0|PP=3|TP=1|ip-26-0-168-34]: Local number of parameters: 76.1M (145.08MiB)
+NCCL version 2.18.5+cuda12.2
+01/03/2025 02:30:39 [INFO|DP=0|PP=3|TP=5|ip-26-0-168-34]: [After model building] Memory usage: 146.09MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:30:39 [INFO|DP=0|PP=3|TP=1|ip-26-0-168-34]: [After model building] Memory usage: 146.09MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:30:39 [INFO|DP=0|PP=2|TP=13|ip-26-0-168-30]: Local number of parameters: 60.9M (116.06MiB)
+01/03/2025 02:30:39 [INFO|DP=0|PP=4|TP=13|ip-26-0-169-132]: Local number of parameters: 60.9M (116.06MiB)
+NCCL version 2.18.5+cuda12.2
+01/03/2025 02:30:39 [INFO|DP=0|PP=2|TP=13|ip-26-0-168-30]: [After model building] Memory usage: 117.08MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+01/03/2025 02:30:39 [INFO|DP=0|PP=4|TP=13|ip-26-0-169-132]: [After model building] Memory usage: 117.08MiB. Peak allocated: 5392.00MiB Peak reserved: 9494.00MiB
+01/03/2025 02:30:39 [INFO|DP=0|PP=3|TP=13|ip-26-0-168-52]: Local number of parameters: 76.1M (145.08MiB)
+01/03/2025 02:30:39 [INFO|DP=0|PP=2|TP=12|ip-26-0-168-30]: Local number of parameters: 60.9M (116.06MiB)
+01/03/2025 02:30:39 [INFO|DP=0|PP=4|TP=12|ip-26-0-169-132]: Local number of parameters: 60.9M (116.06MiB)
+01/03/2025 02:30:39 [INFO|DP=0|PP=3|TP=13|ip-26-0-168-52]: [After model building] Memory usage: 146.09MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:30:39 [INFO|DP=0|PP=2|TP=12|ip-26-0-168-30]: [After model building] Memory usage: 117.08MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:30:39 [INFO|DP=0|PP=4|TP=15|ip-26-0-169-132]: Local number of parameters: 60.9M (116.06MiB)
+01/03/2025 02:30:39 [INFO|DP=0|PP=4|TP=12|ip-26-0-169-132]: [After model building] Memory usage: 117.08MiB. Peak allocated: 5392.00MiB Peak reserved: 9494.00MiB
+01/03/2025 02:30:39 [INFO|DP=0|PP=4|TP=9|ip-26-0-169-132]: Local number of parameters: 60.9M (116.06MiB)
+01/03/2025 02:30:39 [INFO|DP=0|PP=4|TP=15|ip-26-0-169-132]: [After model building] Memory usage: 117.08MiB. Peak allocated: 5392.00MiB Peak reserved: 9494.00MiB
+01/03/2025 02:30:39 [INFO|DP=0|PP=3|TP=12|ip-26-0-168-52]: Local number of parameters: 76.1M (145.08MiB)
+01/03/2025 02:30:39 [INFO|DP=0|PP=4|TP=9|ip-26-0-169-132]: [After model building] Memory usage: 117.08MiB. Peak allocated: 5392.00MiB Peak reserved: 9494.00MiB
+NCCL version 2.18.5+cuda12.2
+01/03/2025 02:30:39 [INFO|DP=0|PP=4|TP=10|ip-26-0-169-132]: Local number of parameters: 60.9M (116.06MiB)
+01/03/2025 02:30:39 [INFO|DP=0|PP=3|TP=12|ip-26-0-168-52]: [After model building] Memory usage: 146.09MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:30:39 [INFO|DP=0|PP=2|TP=15|ip-26-0-168-30]: Local number of parameters: 60.9M (116.06MiB)
+01/03/2025 02:30:39 [INFO|DP=0|PP=3|TP=9|ip-26-0-168-52]: Local number of parameters: 76.1M (145.08MiB)
+01/03/2025 02:30:39 [INFO|DP=0|PP=6|TP=9|ip-26-0-169-247]: Local number of parameters: 60.9M (116.06MiB)
+01/03/2025 02:30:39 [INFO|DP=0|PP=1|TP=9|ip-26-0-168-120]: Local number of parameters: 60.9M (116.06MiB)
+01/03/2025 02:30:39 [INFO|DP=0|PP=6|TP=13|ip-26-0-169-247]: Local number of parameters: 60.9M (116.06MiB)
+NCCL version 2.18.5+cuda12.2
+01/03/2025 02:30:39 [INFO|DP=0|PP=2|TP=9|ip-26-0-168-30]: Local number of parameters: 60.9M (116.06MiB)
+01/03/2025 02:30:39 [INFO|DP=0|PP=2|TP=10|ip-26-0-168-30]: Local number of parameters: 60.9M (116.06MiB)
+01/03/2025 02:30:39 [INFO|DP=0|PP=5|TP=9|ip-26-0-169-207]: Local number of parameters: 60.9M (116.06MiB)
+01/03/2025 02:30:39 [INFO|DP=0|PP=5|TP=13|ip-26-0-169-207]: Local number of parameters: 60.9M (116.06MiB)
+01/03/2025 02:30:39 [INFO|DP=0|PP=2|TP=15|ip-26-0-168-30]: [After model building] Memory usage: 117.08MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:30:39 [INFO|DP=0|PP=4|TP=10|ip-26-0-169-132]: [After model building] Memory usage: 117.08MiB. Peak allocated: 5392.00MiB Peak reserved: 9494.00MiB
+01/03/2025 02:30:39 [INFO|DP=0|PP=7|TP=9|ip-26-0-170-31]: Local number of parameters: 64M (122.04MiB)
+01/03/2025 02:30:39 [INFO|DP=0|PP=3|TP=9|ip-26-0-168-52]: [After model building] Memory usage: 146.09MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:30:39 [INFO|DP=0|PP=6|TP=9|ip-26-0-169-247]: [After model building] Memory usage: 117.08MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:30:39 [INFO|DP=0|PP=1|TP=9|ip-26-0-168-120]: [After model building] Memory usage: 117.08MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:30:39 [INFO|DP=0|PP=6|TP=13|ip-26-0-169-247]: [After model building] Memory usage: 117.08MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:30:39 [INFO|DP=0|PP=5|TP=13|ip-26-0-169-207]: [After model building] Memory usage: 117.08MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:30:39 [INFO|DP=0|PP=5|TP=9|ip-26-0-169-207]: [After model building] Memory usage: 117.08MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:30:39 [INFO|DP=0|PP=2|TP=10|ip-26-0-168-30]: [After model building] Memory usage: 117.08MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+NCCL version 2.18.5+cuda12.2
+01/03/2025 02:30:39 [INFO|DP=0|PP=2|TP=9|ip-26-0-168-30]: [After model building] Memory usage: 117.08MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:30:39 [INFO|DP=0|PP=2|TP=11|ip-26-0-168-30]: Local number of parameters: 60.9M (116.06MiB)
+01/03/2025 02:30:39 [INFO|DP=0|PP=1|TP=13|ip-26-0-168-120]: Local number of parameters: 60.9M (116.06MiB)
+01/03/2025 02:30:39 [INFO|DP=0|PP=1|TP=11|ip-26-0-168-120]: Local number of parameters: 60.9M (116.06MiB)
+01/03/2025 02:30:39 [INFO|DP=0|PP=6|TP=10|ip-26-0-169-247]: Local number of parameters: 60.9M (116.06MiB)
+01/03/2025 02:30:39 [INFO|DP=0|PP=3|TP=10|ip-26-0-168-52]: Local number of parameters: 76.1M (145.08MiB)
+01/03/2025 02:30:39 [INFO|DP=0|PP=4|TP=11|ip-26-0-169-132]: Local number of parameters: 60.9M (116.06MiB)
+NCCL version 2.18.5+cuda12.2
+01/03/2025 02:30:39 [INFO|DP=0|PP=5|TP=10|ip-26-0-169-207]: Local number of parameters: 60.9M (116.06MiB)
+01/03/2025 02:30:39 [INFO|DP=0|PP=7|TP=9|ip-26-0-170-31]: [After model building] Memory usage: 122.05MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:30:39 [INFO|DP=0|PP=4|TP=14|ip-26-0-169-132]: Local number of parameters: 60.9M (116.06MiB)
+01/03/2025 02:30:39 [INFO|DP=0|PP=3|TP=15|ip-26-0-168-52]: Local number of parameters: 76.1M (145.08MiB)
+01/03/2025 02:30:39 [INFO|DP=0|PP=7|TP=13|ip-26-0-170-31]: Local number of parameters: 64M (122.04MiB)
+01/03/2025 02:30:39 [INFO|DP=0|PP=2|TP=11|ip-26-0-168-30]: [After model building] Memory usage: 117.08MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:30:39 [INFO|DP=0|PP=1|TP=11|ip-26-0-168-120]: [After model building] Memory usage: 117.08MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:30:39 [INFO|DP=0|PP=6|TP=10|ip-26-0-169-247]: [After model building] Memory usage: 117.08MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:30:39 [INFO|DP=0|PP=3|TP=10|ip-26-0-168-52]: [After model building] Memory usage: 146.09MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:30:39 [INFO|DP=0|PP=5|TP=12|ip-26-0-169-207]: Local number of parameters: 60.9M (116.06MiB)
+01/03/2025 02:30:39 [INFO|DP=0|PP=1|TP=13|ip-26-0-168-120]: [After model building] Memory usage: 117.08MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:30:39 [INFO|DP=0|PP=4|TP=11|ip-26-0-169-132]: [After model building] Memory usage: 117.08MiB. Peak allocated: 5392.00MiB Peak reserved: 9494.00MiB
+01/03/2025 02:30:39 [INFO|DP=0|PP=6|TP=12|ip-26-0-169-247]: Local number of parameters: 60.9M (116.06MiB)
+01/03/2025 02:30:39 [INFO|DP=0|PP=5|TP=10|ip-26-0-169-207]: [After model building] Memory usage: 117.08MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:30:39 [INFO|DP=0|PP=3|TP=15|ip-26-0-168-52]: [After model building] Memory usage: 146.09MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:30:39 [INFO|DP=0|PP=4|TP=14|ip-26-0-169-132]: [After model building] Memory usage: 117.08MiB. Peak allocated: 5392.00MiB Peak reserved: 9494.00MiB
+01/03/2025 02:30:39 [INFO|DP=0|PP=7|TP=13|ip-26-0-170-31]: [After model building] Memory usage: 122.05MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:30:39 [INFO|DP=0|PP=5|TP=12|ip-26-0-169-207]: [After model building] Memory usage: 117.08MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:30:39 [INFO|DP=0|PP=6|TP=12|ip-26-0-169-247]: [After model building] Memory usage: 117.08MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:30:39 [INFO|DP=0|PP=2|TP=14|ip-26-0-168-30]: Local number of parameters: 60.9M (116.06MiB)
+01/03/2025 02:30:39 [INFO|DP=0|PP=1|TP=12|ip-26-0-168-120]: Local number of parameters: 60.9M (116.06MiB)
+01/03/2025 02:30:39 [INFO|DP=0|PP=5|TP=15|ip-26-0-169-207]: Local number of parameters: 60.9M (116.06MiB)
+01/03/2025 02:30:39 [INFO|DP=0|PP=3|TP=11|ip-26-0-168-52]: Local number of parameters: 76.1M (145.08MiB)
+01/03/2025 02:30:39 [INFO|DP=0|PP=7|TP=10|ip-26-0-170-31]: Local number of parameters: 64M (122.04MiB)
+01/03/2025 02:30:39 [INFO|DP=0|PP=1|TP=12|ip-26-0-168-120]: [After model building] Memory usage: 117.08MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+NCCL version 2.18.5+cuda12.2
+01/03/2025 02:30:39 [INFO|DP=0|PP=2|TP=14|ip-26-0-168-30]: [After model building] Memory usage: 117.08MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:30:39 [INFO|DP=0|PP=7|TP=12|ip-26-0-170-31]: Local number of parameters: 64M (122.04MiB)
+NCCL version 2.18.5+cuda12.2
+01/03/2025 02:30:39 [INFO|DP=0|PP=6|TP=15|ip-26-0-169-247]: Local number of parameters: 60.9M (116.06MiB)
+01/03/2025 02:30:39 [INFO|DP=0|PP=3|TP=11|ip-26-0-168-52]: [After model building] Memory usage: 146.09MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:30:39 [INFO|DP=0|PP=5|TP=15|ip-26-0-169-207]: [After model building] Memory usage: 117.08MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:30:39 [INFO|DP=0|PP=7|TP=10|ip-26-0-170-31]: [After model building] Memory usage: 122.05MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+01/03/2025 02:30:39 [INFO|DP=0|PP=7|TP=12|ip-26-0-170-31]: [After model building] Memory usage: 122.05MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:30:39 [INFO|DP=0|PP=3|TP=14|ip-26-0-168-52]: Local number of parameters: 76.1M (145.08MiB)
+01/03/2025 02:30:39 [INFO|DP=0|PP=7|TP=15|ip-26-0-170-31]: Local number of parameters: 64M (122.04MiB)
+01/03/2025 02:30:39 [INFO|DP=0|PP=6|TP=15|ip-26-0-169-247]: [After model building] Memory usage: 117.08MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+01/03/2025 02:30:39 [INFO|DP=0|PP=7|TP=15|ip-26-0-170-31]: [After model building] Memory usage: 122.05MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:30:39 [INFO|DP=0|PP=3|TP=14|ip-26-0-168-52]: [After model building] Memory usage: 146.09MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:30:39 [INFO|DP=0|PP=6|TP=11|ip-26-0-169-247]: Local number of parameters: 60.9M (116.06MiB)
+NCCL version 2.18.5+cuda12.2
+01/03/2025 02:30:39 [INFO|DP=0|PP=1|TP=10|ip-26-0-168-120]: Local number of parameters: 60.9M (116.06MiB)
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+01/03/2025 02:30:39 [INFO|DP=0|PP=1|TP=15|ip-26-0-168-120]: Local number of parameters: 60.9M (116.06MiB)
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+01/03/2025 02:30:39 [INFO|DP=0|PP=6|TP=11|ip-26-0-169-247]: [After model building] Memory usage: 117.08MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:30:39 [INFO|DP=0|PP=5|TP=11|ip-26-0-169-207]: Local number of parameters: 60.9M (116.06MiB)
+01/03/2025 02:30:39 [INFO|DP=0|PP=5|TP=14|ip-26-0-169-207]: Local number of parameters: 60.9M (116.06MiB)
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+01/03/2025 02:30:39 [INFO|DP=0|PP=1|TP=10|ip-26-0-168-120]: [After model building] Memory usage: 117.08MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:30:39 [INFO|DP=0|PP=6|TP=14|ip-26-0-169-247]: Local number of parameters: 60.9M (116.06MiB)
+01/03/2025 02:30:39 [INFO|DP=0|PP=1|TP=15|ip-26-0-168-120]: [After model building] Memory usage: 117.08MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+01/03/2025 02:30:39 [INFO|DP=0|PP=5|TP=14|ip-26-0-169-207]: [After model building] Memory usage: 117.08MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:30:39 [INFO|DP=0|PP=5|TP=11|ip-26-0-169-207]: [After model building] Memory usage: 117.08MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+01/03/2025 02:30:39 [INFO|DP=0|PP=6|TP=14|ip-26-0-169-247]: [After model building] Memory usage: 117.08MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+01/03/2025 02:30:39 [INFO|DP=0|PP=7|TP=11|ip-26-0-170-31]: Local number of parameters: 64M (122.04MiB)
+01/03/2025 02:30:39 [INFO|DP=0|PP=7|TP=14|ip-26-0-170-31]: Local number of parameters: 64M (122.04MiB)
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+01/03/2025 02:30:39 [INFO|DP=0|PP=1|TP=14|ip-26-0-168-120]: Local number of parameters: 60.9M (116.06MiB)
+01/03/2025 02:30:39 [INFO|DP=0|PP=7|TP=11|ip-26-0-170-31]: [After model building] Memory usage: 122.05MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:30:39 [INFO|DP=0|PP=7|TP=14|ip-26-0-170-31]: [After model building] Memory usage: 122.05MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+01/03/2025 02:30:39 [INFO|DP=0|PP=1|TP=14|ip-26-0-168-120]: [After model building] Memory usage: 117.08MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+01/03/2025 02:30:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [Optimizer Building] Using LearningRateForSP as learning rate
+01/03/2025 02:30:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.token_position_embeddings | PP: 0/8 | Block rank: 0
+01/03/2025 02:30:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.0 | PP: 0/8 | Block rank: 1
+01/03/2025 02:30:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.1 | PP: 0/8 | Block rank: 2
+01/03/2025 02:30:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.2 | PP: 0/8 | Block rank: 3
+01/03/2025 02:30:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.3 | PP: 0/8 | Block rank: 4
+01/03/2025 02:30:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.4 | PP: 0/8 | Block rank: 5
+01/03/2025 02:30:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.5 | PP: 1/8 | Block rank: 0
+01/03/2025 02:30:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.6 | PP: 1/8 | Block rank: 1
+01/03/2025 02:30:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.7 | PP: 1/8 | Block rank: 2
+01/03/2025 02:30:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.8 | PP: 1/8 | Block rank: 3
+01/03/2025 02:30:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.9 | PP: 2/8 | Block rank: 0
+01/03/2025 02:30:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.10 | PP: 2/8 | Block rank: 1
+01/03/2025 02:30:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.11 | PP: 2/8 | Block rank: 2
+01/03/2025 02:30:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.12 | PP: 2/8 | Block rank: 3
+01/03/2025 02:30:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.13 | PP: 3/8 | Block rank: 0
+01/03/2025 02:30:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.14 | PP: 3/8 | Block rank: 1
+01/03/2025 02:30:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.15 | PP: 3/8 | Block rank: 2
+01/03/2025 02:30:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.16 | PP: 3/8 | Block rank: 3
+01/03/2025 02:30:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.17 | PP: 3/8 | Block rank: 4
+01/03/2025 02:30:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.18 | PP: 4/8 | Block rank: 0
+01/03/2025 02:30:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.19 | PP: 4/8 | Block rank: 1
+01/03/2025 02:30:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.20 | PP: 4/8 | Block rank: 2
+01/03/2025 02:30:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.21 | PP: 4/8 | Block rank: 3
+01/03/2025 02:30:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.22 | PP: 5/8 | Block rank: 0
+01/03/2025 02:30:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.23 | PP: 5/8 | Block rank: 1
+01/03/2025 02:30:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.24 | PP: 5/8 | Block rank: 2
+01/03/2025 02:30:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.25 | PP: 5/8 | Block rank: 3
+01/03/2025 02:30:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.26 | PP: 6/8 | Block rank: 0
+01/03/2025 02:30:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.27 | PP: 6/8 | Block rank: 1
+01/03/2025 02:30:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.28 | PP: 6/8 | Block rank: 2
+01/03/2025 02:30:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.29 | PP: 6/8 | Block rank: 3
+01/03/2025 02:30:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.30 | PP: 7/8 | Block rank: 0
+01/03/2025 02:30:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.31 | PP: 7/8 | Block rank: 1
+01/03/2025 02:30:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.final_layer_norm | PP: 7/8 | Block rank: 2
+01/03/2025 02:30:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.lm_head | PP: 7/8 | Block rank: 3
+01/03/2025 02:30:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.cast_to_fp32 | PP: 7/8 | Block rank: 4
+01/03/2025 02:30:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: loss | PP: 7/8 | Block rank: 5
+01/03/2025 02:30:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [Training Plan] Stage Stable Training Stage has 99 remaining training steps and has consumed 0 samples
+01/03/2025 02:30:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: Using dummy data generator
+01/03/2025 02:30:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [Training Plan] There are 1 training stages 
+01/03/2025 02:30:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [Stage Stable Training Stage] start from step 1 
+01/03/2025 02:30:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: 
+01/03/2025 02:30:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [Start training] datetime: 2025-01-03 02:30:40.896806 | mbs: 16 | grad_accum: 16 | global_batch_size: 256 | sequence_length: 4096 | train_steps: 100 | start_iteration_step: 0 | consumed_train_samples: 0
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+01/03/2025 02:30:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: Resuming training from stage Stable Training Stage, it has trained for 0 samples and has 99 remaining train steps
+01/03/2025 02:30:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:  Memory usage: 1047.40MiB. Peak allocated 5392.00MiB. Peak reserved: 10518.00MiB
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+01/03/2025 02:31:30 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:  Memory usage: 2140.97MiB. Peak allocated 33463.84MiB. Peak reserved: 34566.00MiB
+01/03/2025 02:31:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:  Memory usage: 2977.30MiB. Peak allocated 2977.30MiB. Peak reserved: 34566.00MiB
+01/03/2025 02:31:40 [INFO|DP=0|PP=7|TP=0|ip-26-0-169-86]: iteration: 1 / 100 | consumed_tokens: 1.05M | elapsed_time_per_iteration_ms: 57.2K | tokens_per_sec: 18.3K | tokens_per_sec_per_gpu: 143 | global_batch_size: 256 | lm_loss: 12.6 | lr: 0.00015 | model_tflops_per_gpu: 8.07 | hardware_tflops_per_gpu: 8.07 | grad_norm: 1.45 | cuda_memory_allocated: 2.3G | cuda_max_memory_reserved: 9.29G | hd_total_memory_tb: 312G | hd_used_memory_tb: 69.9G | hd_free_memory_tb: 242G
+01/03/2025 02:32:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:  Memory usage: 2977.30MiB. Peak allocated 34300.17MiB. Peak reserved: 36102.00MiB
+01/03/2025 02:32:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:  Memory usage: 2977.30MiB. Peak allocated 2977.31MiB. Peak reserved: 36102.00MiB
+01/03/2025 02:32:06 [INFO|DP=0|PP=7|TP=0|ip-26-0-169-86]: iteration: 2 / 100 | consumed_tokens: 2.1M | elapsed_time_per_iteration_ms: 26.7K | tokens_per_sec: 39.3K | tokens_per_sec_per_gpu: 307 | global_batch_size: 256 | lm_loss: 12.6 | lr: 0.0003 | model_tflops_per_gpu: 17.3 | hardware_tflops_per_gpu: 17.3 | grad_norm: 1.45 | cuda_memory_allocated: 2.3G | cuda_max_memory_reserved: 11.4G | hd_total_memory_tb: 312G | hd_used_memory_tb: 69.9G | hd_free_memory_tb: 242G
+01/03/2025 02:32:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:  Memory usage: 2977.30MiB. Peak allocated 34300.17MiB. Peak reserved: 36102.00MiB
+01/03/2025 02:32:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: | job_id   | name                                                             | nodes | seq_len | mbs | batch_accum | gbs | mTFLOPs | hTFLOPs | tok/s/gpu | AllReduce (GB/s) | AllGather (GB/s) | ReduceScatter (GB/s) | AR Intra-node (GB/s) | AG Intra-node (GB/s) | RS Intra-node (GB/s) | Mem Alloc (GB) | Mem Res (GB) | dp | pp | tp | pp_engine                           | tp_mode        | tp_async_comm | hidden_size | hidden_act | num_layers | num_heads | num_kv_heads | max_pos | vocab_size | tie_word_embeddings | dtype          | zero_stage | ddp_bucket_cap_mb | accumulate_grad_in_fp32 | Total Params | Local Params |
+01/03/2025 02:32:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: | -------- | ---------------------------------------------------------------- | ----- | ------- | --- | ----------- | --- | ------- | ------- | --------- | ---------------- | ---------------- | -------------------- | -------------------- | -------------------- | -------------------- | -------------- | ------------ | -- | -- | -- | ----------------------------------- | -------------- | ------------- | ----------- | ---------- | ---------- | --------- | ------------ | ------- | ---------- | ------------------- | -------------- | ---------- | ----------------- | ----------------------- | ------------ | ------------ |
+01/03/2025 02:32:33 [INFO|DP=0|PP=7|TP=0|ip-26-0-169-86]: iteration: 3 / 100 | consumed_tokens: 3.15M | elapsed_time_per_iteration_ms: 26.6K | tokens_per_sec: 39.5K | tokens_per_sec_per_gpu: 308 | global_batch_size: 256 | lm_loss: 12.6 | lr: 0.000296 | model_tflops_per_gpu: 17.4 | hardware_tflops_per_gpu: 17.4 | grad_norm: 1.41 | cuda_memory_allocated: 2.3G | cuda_max_memory_reserved: 11.4G | hd_total_memory_tb: 312G | hd_used_memory_tb: 69.9G | hd_free_memory_tb: 242G
+01/03/2025 02:32:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: | 14018953 | 8.86G_dp1_tp16_pp8_acc16_mbs16_seq4096_zero0_tpmodeRED_vocab131k | 16    | 4096    | 16  | 16          | 256 | 17.38   | 17.38   | 308.30    | 50.43            | 26.52            | 26.16                | 424.10               | 242.04               | 241.73               | 2.91           | 35.26        | 1  | 8  | 16 | OneForwardOneBackwardPipelineEngine | REDUCE_SCATTER | True          | 4096        | silu       | 32         | 32        | 32           | 4096    | 131072     | False               | torch.bfloat16 | 0          | 25                | True                    | 8.86G        | 110M         |
+01/03/2025 02:32:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: Attempting to write benchmark results to CSV file: benchmark/results/bench_final2.csv
+01/03/2025 02:33:30 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: Acquired lock for CSV file: benchmark/results/bench_final2.csv
+01/03/2025 02:33:30 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: Successfully wrote to CSV file: benchmark/results/bench_final2.csv. Releasing lock...
+01/03/2025 02:33:31 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: Throughput logging complete
+[2025-01-03 02:33:58,309] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 377285 closing signal SIGTERM
+[2025-01-03 02:33:58,310] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 377286 closing signal SIGTERM
+[2025-01-03 02:33:58,310] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 377288 closing signal SIGTERM
+[2025-01-03 02:33:58,310] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 377289 closing signal SIGTERM
+[2025-01-03 02:33:58,311] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 662957 closing signal SIGTERM
+[2025-01-03 02:33:58,311] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 662958 closing signal SIGTERM
+[2025-01-03 02:33:58,311] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 662960 closing signal SIGTERM
+[2025-01-03 02:33:58,311] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 662961 closing signal SIGTERM
+[2025-01-03 02:33:59,327] torch.distributed.elastic.multiprocessing.api: [ERROR] failed (exitcode: 1) local_rank: 0 (pid: 662954) of binary: /fsx/nouamane/miniconda/envs/2-1-cu121/bin/python
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 264, in launch_agent
+    raise ChildFailedError(
+torch.distributed.elastic.multiprocessing.errors.ChildFailedError: 
+============================================================
+/fsx/nouamane/projects/nanotron/run_train.py FAILED
+------------------------------------------------------------
+Failures:
+[1]:
+  time      : 2025-01-03_02:33:58
+  host      : ip-26-0-170-31.ec2.internal
+  rank      : 121 (local_rank: 1)
+  exitcode  : 1 (pid: 662955)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[2]:
+  time      : 2025-01-03_02:33:58
+  host      : ip-26-0-170-31.ec2.internal
+  rank      : 122 (local_rank: 2)
+  exitcode  : 1 (pid: 662956)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[3]:
+  time      : 2025-01-03_02:33:58
+  host      : ip-26-0-170-31.ec2.internal
+  rank      : 125 (local_rank: 5)
+  exitcode  : 1 (pid: 662959)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+------------------------------------------------------------
+Root Cause (first observed failure):
+[0]:
+  time      : 2025-01-03_02:33:58
+  host      : ip-26-0-170-31.ec2.internal
+  rank      : 120 (local_rank: 0)
+  exitcode  : 1 (pid: 662954)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+============================================================
+srun: error: ip-26-0-170-31: task 15: Exited with exit code 1
+srun: launch/slurm: _step_signal: Terminating StepId=14018953.0
+slurmstepd: error: *** STEP 14018953.0 ON ip-26-0-160-225 CANCELLED AT 2025-01-03T02:33:59 ***
+[2025-01-03 02:33:59,721] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 02:33:59,721] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 377286 closing signal SIGTERM
+[2025-01-03 02:33:59,720] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 02:33:59,720] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 4096477 closing signal SIGTERM
+[2025-01-03 02:33:59,720] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 4096478 closing signal SIGTERM
+[2025-01-03 02:33:59,720] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 4096479 closing signal SIGTERM
+[2025-01-03 02:33:59,720] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 4096480 closing signal SIGTERM
+[2025-01-03 02:33:59,720] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 4096481 closing signal SIGTERM
+[2025-01-03 02:33:59,721] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 4096482 closing signal SIGTERM
+[2025-01-03 02:33:59,721] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 4096483 closing signal SIGTERM
+[2025-01-03 02:33:59,721] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 4096484 closing signal SIGTERM
+[2025-01-03 02:33:59,719] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 02:33:59,720] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 02:33:59,720] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 02:33:59,721] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 51461 closing signal SIGTERM
+[2025-01-03 02:33:59,720] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 30237 closing signal SIGTERM
+[2025-01-03 02:33:59,720] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 50294 closing signal SIGTERM
+[2025-01-03 02:33:59,720] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 50295 closing signal SIGTERM
+[2025-01-03 02:33:59,721] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 51462 closing signal SIGTERM
+[2025-01-03 02:33:59,720] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 30238 closing signal SIGTERM
+[2025-01-03 02:33:59,720] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 02:33:59,721] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 50296 closing signal SIGTERM
+[2025-01-03 02:33:59,721] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 51463 closing signal SIGTERM
+[2025-01-03 02:33:59,720] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 30239 closing signal SIGTERM
+[2025-01-03 02:33:59,721] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 50297 closing signal SIGTERM
+[2025-01-03 02:33:59,720] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 02:33:59,720] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 02:33:59,721] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 51464 closing signal SIGTERM
+[2025-01-03 02:33:59,720] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 30240 closing signal SIGTERM
+[2025-01-03 02:33:59,721] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 227036 closing signal SIGTERM
+[2025-01-03 02:33:59,721] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 50298 closing signal SIGTERM
+[2025-01-03 02:33:59,721] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 51465 closing signal SIGTERM
+[2025-01-03 02:33:59,721] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 227037 closing signal SIGTERM
+[2025-01-03 02:33:59,721] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 50299 closing signal SIGTERM
+[2025-01-03 02:33:59,720] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 30241 closing signal SIGTERM
+[2025-01-03 02:33:59,721] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 227038 closing signal SIGTERM
+[2025-01-03 02:33:59,720] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 165976 closing signal SIGTERM
+[2025-01-03 02:33:59,720] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 575560 closing signal SIGTERM
+[2025-01-03 02:33:59,721] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 51466 closing signal SIGTERM
+[2025-01-03 02:33:59,720] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 30242 closing signal SIGTERM
+[2025-01-03 02:33:59,721] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 227039 closing signal SIGTERM
+[2025-01-03 02:33:59,721] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 50300 closing signal SIGTERM
+[2025-01-03 02:33:59,721] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 51467 closing signal SIGTERM
+[2025-01-03 02:33:59,720] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 02:33:59,721] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 227040 closing signal SIGTERM
+[2025-01-03 02:33:59,721] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 50301 closing signal SIGTERM
+[2025-01-03 02:33:59,720] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 30243 closing signal SIGTERM
+[2025-01-03 02:33:59,721] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 227041 closing signal SIGTERM
+[2025-01-03 02:33:59,720] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 165977 closing signal SIGTERM
+[2025-01-03 02:33:59,721] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 51468 closing signal SIGTERM
+[2025-01-03 02:33:59,721] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 227042 closing signal SIGTERM
+[2025-01-03 02:33:59,720] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 165978 closing signal SIGTERM
+[2025-01-03 02:33:59,720] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 575561 closing signal SIGTERM
+[2025-01-03 02:33:59,720] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 30244 closing signal SIGTERM
+[2025-01-03 02:33:59,720] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 413404 closing signal SIGTERM
+[2025-01-03 02:33:59,720] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 165979 closing signal SIGTERM
+[2025-01-03 02:33:59,721] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 227043 closing signal SIGTERM
+[2025-01-03 02:33:59,720] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 575562 closing signal SIGTERM
+[2025-01-03 02:33:59,720] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 02:33:59,720] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 413405 closing signal SIGTERM
+[2025-01-03 02:33:59,720] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 165980 closing signal SIGTERM
+[2025-01-03 02:33:59,720] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 165981 closing signal SIGTERM
+[2025-01-03 02:33:59,720] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 02:33:59,721] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 575563 closing signal SIGTERM
+[2025-01-03 02:33:59,721] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 264548 closing signal SIGTERM
+[2025-01-03 02:33:59,720] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 413406 closing signal SIGTERM
+[2025-01-03 02:33:59,720] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 165982 closing signal SIGTERM
+[2025-01-03 02:33:59,720] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 165983 closing signal SIGTERM
+[2025-01-03 02:33:59,721] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 172109 closing signal SIGTERM
+[2025-01-03 02:33:59,721] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 575564 closing signal SIGTERM
+[2025-01-03 02:33:59,720] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 413407 closing signal SIGTERM
+[2025-01-03 02:33:59,721] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 264549 closing signal SIGTERM
+[2025-01-03 02:33:59,721] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 264550 closing signal SIGTERM
+[2025-01-03 02:33:59,721] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 172110 closing signal SIGTERM
+[2025-01-03 02:33:59,720] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 413408 closing signal SIGTERM
+[2025-01-03 02:33:59,721] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 575565 closing signal SIGTERM
+[2025-01-03 02:33:59,720] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 413409 closing signal SIGTERM
+[2025-01-03 02:33:59,721] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 264551 closing signal SIGTERM
+[2025-01-03 02:33:59,721] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 264552 closing signal SIGTERM
+[2025-01-03 02:33:59,719] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 02:33:59,721] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 172111 closing signal SIGTERM
+[2025-01-03 02:33:59,720] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 413410 closing signal SIGTERM
+[2025-01-03 02:33:59,721] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 264553 closing signal SIGTERM
+[2025-01-03 02:33:59,721] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 264554 closing signal SIGTERM
+[2025-01-03 02:33:59,721] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 413411 closing signal SIGTERM
+[2025-01-03 02:33:59,721] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 172112 closing signal SIGTERM
+[2025-01-03 02:33:59,721] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 172113 closing signal SIGTERM
+[2025-01-03 02:33:59,720] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 46690 closing signal SIGTERM
+[2025-01-03 02:33:59,721] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 575566 closing signal SIGTERM
+[2025-01-03 02:33:59,720] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 46691 closing signal SIGTERM
+[2025-01-03 02:33:59,721] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 264555 closing signal SIGTERM
+[2025-01-03 02:33:59,721] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 172114 closing signal SIGTERM
+[2025-01-03 02:33:59,721] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 172115 closing signal SIGTERM
+[2025-01-03 02:33:59,721] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 02:33:59,721] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 575567 closing signal SIGTERM
+[2025-01-03 02:33:59,720] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 46692 closing signal SIGTERM
+[2025-01-03 02:33:59,720] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 46693 closing signal SIGTERM
+[2025-01-03 02:33:59,721] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 172116 closing signal SIGTERM
+[2025-01-03 02:33:59,721] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 51500 closing signal SIGTERM
+[2025-01-03 02:33:59,720] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 46694 closing signal SIGTERM
+[2025-01-03 02:33:59,721] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 51501 closing signal SIGTERM
+[2025-01-03 02:33:59,720] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 46695 closing signal SIGTERM
+[2025-01-03 02:33:59,720] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 02:33:59,720] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 46696 closing signal SIGTERM
+[2025-01-03 02:33:59,721] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 51502 closing signal SIGTERM
+[2025-01-03 02:33:59,721] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 51503 closing signal SIGTERM
+[2025-01-03 02:33:59,721] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 51504 closing signal SIGTERM
+[2025-01-03 02:33:59,721] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 24441 closing signal SIGTERM
+[2025-01-03 02:33:59,721] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 24442 closing signal SIGTERM
+[2025-01-03 02:33:59,720] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 46697 closing signal SIGTERM
+[2025-01-03 02:33:59,721] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 51505 closing signal SIGTERM
+[2025-01-03 02:33:59,721] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 51506 closing signal SIGTERM
+[2025-01-03 02:33:59,721] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 24443 closing signal SIGTERM
+[2025-01-03 02:33:59,721] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 24444 closing signal SIGTERM
+[2025-01-03 02:33:59,721] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 51507 closing signal SIGTERM
+[2025-01-03 02:33:59,721] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 24445 closing signal SIGTERM
+[2025-01-03 02:33:59,721] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 24446 closing signal SIGTERM
+[2025-01-03 02:33:59,721] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 24447 closing signal SIGTERM
+[2025-01-03 02:33:59,721] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 02:33:59,721] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 24448 closing signal SIGTERM
+[2025-01-03 02:33:59,721] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 186487 closing signal SIGTERM
+[2025-01-03 02:33:59,721] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 186488 closing signal SIGTERM
+[2025-01-03 02:33:59,721] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 186489 closing signal SIGTERM
+[2025-01-03 02:33:59,721] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 186490 closing signal SIGTERM
+[2025-01-03 02:33:59,722] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 186491 closing signal SIGTERM
+[2025-01-03 02:33:59,722] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 186492 closing signal SIGTERM
+[2025-01-03 02:33:59,722] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 186493 closing signal SIGTERM
+[2025-01-03 02:33:59,722] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 186494 closing signal SIGTERM
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 878, in _invoke_run
+    run_result = self._monitor_workers(self._worker_group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/local_elastic_agent.py", line 307, in _monitor_workers
+    result = self._pcontext.wait(0)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 288, in wait
+    return self._poll()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 668, in _poll
+    self.close()  # terminate all running procs
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 331, in close
+    self._close(death_sig=death_sig, timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 713, in _close
+    handler.proc.wait(time_to_wait)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1209, in wait
+    return self._wait(timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1953, in _wait
+    time.sleep(delay)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 377210 got signal: 15
+srun: error: ip-26-0-160-225: task 0: Exited with exit code 1
+[2025-01-03 02:34:02,237] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-168-34.ec2.internal_30164_0' has failed to send a keep-alive heartbeat to the rendezvous '14018953' due to an error of type RendezvousConnectionError.
+[2025-01-03 02:34:02,968] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-168-238.ec2.internal_51389_0' has failed to send a keep-alive heartbeat to the rendezvous '14018953' due to an error of type RendezvousConnectionError.
+[2025-01-03 02:34:02,985] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-168-30.ec2.internal_575486_0' has failed to send a keep-alive heartbeat to the rendezvous '14018953' due to an error of type RendezvousConnectionError.
+[2025-01-03 02:34:02,992] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-169-86.ec2.internal_50222_0' has failed to send a keep-alive heartbeat to the rendezvous '14018953' due to an error of type RendezvousConnectionError.
+[2025-01-03 02:34:03,043] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-169-207.ec2.internal_186415_0' has failed to send a keep-alive heartbeat to the rendezvous '14018953' due to an error of type RendezvousConnectionError.
+[2025-01-03 02:34:03,084] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-169-239.ec2.internal_172038_0' has failed to send a keep-alive heartbeat to the rendezvous '14018953' due to an error of type RendezvousConnectionError.
+[2025-01-03 02:34:03,109] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-167-245.ec2.internal_4096405_0' has failed to send a keep-alive heartbeat to the rendezvous '14018953' due to an error of type RendezvousConnectionError.
+[2025-01-03 02:34:03,123] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-169-247.ec2.internal_51428_0' has failed to send a keep-alive heartbeat to the rendezvous '14018953' due to an error of type RendezvousConnectionError.
+[2025-01-03 02:34:03,128] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-168-120.ec2.internal_165904_0' has failed to send a keep-alive heartbeat to the rendezvous '14018953' due to an error of type RendezvousConnectionError.
+[2025-01-03 02:34:03,155] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-168-95.ec2.internal_264476_0' has failed to send a keep-alive heartbeat to the rendezvous '14018953' due to an error of type RendezvousConnectionError.
+[2025-01-03 02:34:03,186] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-169-132.ec2.internal_46619_0' has failed to send a keep-alive heartbeat to the rendezvous '14018953' due to an error of type RendezvousConnectionError.
+[2025-01-03 02:34:03,189] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-168-52.ec2.internal_413332_0' has failed to send a keep-alive heartbeat to the rendezvous '14018953' due to an error of type RendezvousConnectionError.
+[2025-01-03 02:34:03,200] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-169-139.ec2.internal_24369_0' has failed to send a keep-alive heartbeat to the rendezvous '14018953' due to an error of type RendezvousConnectionError.
+[2025-01-03 02:34:03,207] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-167-217.ec2.internal_226965_0' has failed to send a keep-alive heartbeat to the rendezvous '14018953' due to an error of type RendezvousConnectionError.
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 50222 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 226965 got signal: 15
+srun: error: ip-26-0-169-86: task 9: Exited with exit code 1
+srun: error: ip-26-0-167-217: task 1: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 51428 got signal: 15
+srun: error: ip-26-0-169-247: task 14: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 24369 got signal: 15
+[2025-01-03 02:34:07,239] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-168-34.ec2.internal_30164_0' has failed to send a keep-alive heartbeat to the rendezvous '14018953' due to an error of type RendezvousConnectionError.
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 51389 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 46619 got signal: 15
+srun: error: ip-26-0-169-139: task 11: Exited with exit code 1
+srun: error: ip-26-0-168-238: task 8: Exited with exit code 1
+srun: error: ip-26-0-169-132: task 10: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 165904 got signal: 15
+[2025-01-03 02:34:07,986] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-168-30.ec2.internal_575486_0' has failed to send a keep-alive heartbeat to the rendezvous '14018953' due to an error of type RendezvousConnectionError.
+[2025-01-03 02:34:08,045] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-169-207.ec2.internal_186415_0' has failed to send a keep-alive heartbeat to the rendezvous '14018953' due to an error of type RendezvousConnectionError.
+[2025-01-03 02:34:08,085] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-169-239.ec2.internal_172038_0' has failed to send a keep-alive heartbeat to the rendezvous '14018953' due to an error of type RendezvousConnectionError.
+[2025-01-03 02:34:08,110] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-167-245.ec2.internal_4096405_0' has failed to send a keep-alive heartbeat to the rendezvous '14018953' due to an error of type RendezvousConnectionError.
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 186415 got signal: 15
+[2025-01-03 02:34:08,157] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-168-95.ec2.internal_264476_0' has failed to send a keep-alive heartbeat to the rendezvous '14018953' due to an error of type RendezvousConnectionError.
+[2025-01-03 02:34:08,191] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-168-52.ec2.internal_413332_0' has failed to send a keep-alive heartbeat to the rendezvous '14018953' due to an error of type RendezvousConnectionError.
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 4096405 got signal: 15
+srun: error: ip-26-0-168-120: task 7: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 30164 got signal: 15
+srun: error: ip-26-0-169-207: task 12: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 264476 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 575486 got signal: 15
+srun: error: ip-26-0-167-245: task 2: Exited with exit code 1
+srun: error: ip-26-0-168-34: task 4: Exited with exit code 1
+srun: error: ip-26-0-168-30: task 3: Exited with exit code 1
+srun: error: ip-26-0-168-95: task 6: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 172038 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 413332 got signal: 15
+srun: error: ip-26-0-169-239: task 13: Exited with exit code 1
+srun: error: ip-26-0-168-52: task 5: Exited with exit code 1
diff --git a/logs/14018985-bench_3.57G_dp4_tp1_pp4_acc64_mbs1_seq4096_zero0_tpmodeRED_vocab131k.out b/logs/14018985-bench_3.57G_dp4_tp1_pp4_acc64_mbs1_seq4096_zero0_tpmodeRED_vocab131k.out
new file mode 100644
index 0000000000000000000000000000000000000000..07da74699fd6cbc99bf12fb7c47b9bbba9c815c2
--- /dev/null
+++ b/logs/14018985-bench_3.57G_dp4_tp1_pp4_acc64_mbs1_seq4096_zero0_tpmodeRED_vocab131k.out
@@ -0,0 +1,1081 @@
++ '[' -z 14018985 ']'
++ source /etc/profile.d/modules.sh
+++ . /usr/share/modules/init/bash
++++ unset _mlshdbg
++++ '[' 0 = 1 ']'
++++ unset _mlre _mlIFS
++++ '[' -n x ']'
++++ _mlIFS=' 	
+'
++++ IFS=' '
++++ '[' -n '' ']'
+++++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash autoinit
++++ _mlcode='module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+MODULEPATH_modshare=/etc/environment-modules/modules:1:/usr/share/modules/\$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1; export MODULEPATH_modshare;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+LOADEDMODULES=; export LOADEDMODULES;
+MODULEPATH=/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/\$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles; export MODULEPATH;
+test 0;'
++++ _mlret=0
++++ '[' -n x ']'
++++ IFS=' 	
+'
++++ unset _mlIFS
++++ unset _mlre _mlv _mlrv
++++ '[' 0 -eq 0 ']'
++++ eval 'module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+MODULEPATH_modshare=/etc/environment-modules/modules:1:/usr/share/modules/\$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1; export MODULEPATH_modshare;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+LOADEDMODULES=; export LOADEDMODULES;
+MODULEPATH=/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/\$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles; export MODULEPATH;
+test 0;'
+++++ MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl
+++++ export MODULES_CMD
+++++ ENV=/usr/share/modules/init/profile.sh
+++++ export ENV
+++++ MODULEPATH_modshare='/etc/environment-modules/modules:1:/usr/share/modules/$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1'
+++++ export MODULEPATH_modshare
+++++ BASH_ENV=/usr/share/modules/init/bash
+++++ export BASH_ENV
+++++ MODULESHOME=/usr/share/modules
+++++ export MODULESHOME
+++++ LOADEDMODULES=
+++++ export LOADEDMODULES
+++++ MODULEPATH='/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles'
+++++ export MODULEPATH
+++++ test 0
++++ '[' 0 = 1 ']'
++++ '[' -t 2 ']'
++++ export -f module
++++ export -f switchml
++++ '[' 5 -ge 3 ']'
++++ [[ ehxB =~ i ]]
++++ [[ ! :/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin: =~ :/usr/bin: ]]
++++ '[' '!' -n '' ']'
++++ MANPATH=:
++++ export MANPATH
+++++ manpath
++++ [[ ! :/fsx/nouamane/miniconda/envs/2-1-cu121/man:/fsx/nouamane/miniconda/envs/2-1-cu121/share/man:/fsx/nouamane/miniconda/man:/fsx/nouamane/miniconda/share/man:/usr/local/man:/usr/local/share/man:/usr/share/man:/opt/slurm/share/man:/admin/home/nouamane/.local/share/man:/admin/home/nouamane/.fzf/man:: =~ :/usr/share/man: ]]
++++ unset _mlcode _mlret
++++ '[' -n '' ']'
++ module load cuda/12.1
++ unset _mlshdbg
++ '[' 0 = 1 ']'
++ unset _mlre _mlIFS
++ '[' -n x ']'
++ _mlIFS=' 	
+'
++ IFS=' '
++ '[' -n '' ']'
+++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash load cuda/12.1
++ eval 'CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include;' export 'CPATH;
+LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:;' export 'LD_LIBRARY_PATH;
+FI_EFA_FORK_SAFE=1;' export 'FI_EFA_FORK_SAFE;
+MANPATH=/usr/local/cuda-12.1/share/man::;' export 'MANPATH;
+LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64;' export 'LIBRARY_PATH;
+_LMFILES_=/usr/share/modules/modulefiles/cuda/12.1;' export '_LMFILES_;
+LOADEDMODULES=cuda/12.1;' export 'LOADEDMODULES;
+MPI_PATH=/opt/amazon/openmpi;' export 'MPI_PATH;
+NCCL_HOME_modshare=/opt/nccl/build:1;' export 'NCCL_HOME_modshare;
+NCCL_PROTO=simple;' export 'NCCL_PROTO;
+MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1;' export 'MANPATH_modshare;
+LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1;' export 'LIBRARY_PATH_modshare;
+NCCL_SOCKET_IFNAME=enp;' export 'NCCL_SOCKET_IFNAME;
+AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl;' export 'AWS_OFI_NCCL_HOME;
+NCCL_HOME=/opt/nccl/build;' export 'NCCL_HOME;
+FI_PROVIDER=efa;' export 'FI_PROVIDER;
+AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1;' export 'AWS_OFI_NCCL_HOME_modshare;
+CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1;' export 'CPATH_modshare;
+LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2::1:/opt/nccl/build/lib:1:/opt/aws-ofi-nccl/lib:2:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/efa/lib:1;' export 'LD_LIBRARY_PATH_modshare;
+FI_EFA_ENABLE_SHM_TRANSFER=1;' export 'FI_EFA_ENABLE_SHM_TRANSFER;
+_LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1;' export '_LMFILES__modshare;
+LOADEDMODULES_modshare=cuda/12.1:1;' export 'LOADEDMODULES_modshare;
+MPI_PATH_modshare=/opt/amazon/openmpi:1;' export 'MPI_PATH_modshare;
+PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin;' export 'PATH;
+CUDA_HOME=/usr/local/cuda-12.1;' export 'CUDA_HOME;
+PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/opt/slurm/bin:1:/usr/sbin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1;' export 'PATH_modshare;
+test' '0;'
+++ CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include
+++ export CPATH
+++ LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:
+++ export LD_LIBRARY_PATH
+++ FI_EFA_FORK_SAFE=1
+++ export FI_EFA_FORK_SAFE
+++ MANPATH=/usr/local/cuda-12.1/share/man::
+++ export MANPATH
+++ LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64
+++ export LIBRARY_PATH
+++ _LMFILES_=/usr/share/modules/modulefiles/cuda/12.1
+++ export _LMFILES_
+++ LOADEDMODULES=cuda/12.1
+++ export LOADEDMODULES
+++ MPI_PATH=/opt/amazon/openmpi
+++ export MPI_PATH
+++ NCCL_HOME_modshare=/opt/nccl/build:1
+++ export NCCL_HOME_modshare
+++ NCCL_PROTO=simple
+++ export NCCL_PROTO
+++ MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1
+++ export MANPATH_modshare
+++ LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1
+++ export LIBRARY_PATH_modshare
+++ NCCL_SOCKET_IFNAME=enp
+++ export NCCL_SOCKET_IFNAME
+++ AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl
+++ export AWS_OFI_NCCL_HOME
+++ NCCL_HOME=/opt/nccl/build
+++ export NCCL_HOME
+++ FI_PROVIDER=efa
+++ export FI_PROVIDER
+++ AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1
+++ export AWS_OFI_NCCL_HOME_modshare
+++ CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1
+++ export CPATH_modshare
+++ LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2::1:/opt/nccl/build/lib:1:/opt/aws-ofi-nccl/lib:2:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/efa/lib:1
+++ export LD_LIBRARY_PATH_modshare
+++ FI_EFA_ENABLE_SHM_TRANSFER=1
+++ export FI_EFA_ENABLE_SHM_TRANSFER
+++ _LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1
+++ export _LMFILES__modshare
+++ LOADEDMODULES_modshare=cuda/12.1:1
+++ export LOADEDMODULES_modshare
+++ MPI_PATH_modshare=/opt/amazon/openmpi:1
+++ export MPI_PATH_modshare
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
+++ export PATH
+++ CUDA_HOME=/usr/local/cuda-12.1
+++ export CUDA_HOME
+++ PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/opt/slurm/bin:1:/usr/sbin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1
+++ export PATH_modshare
+++ test 0
++ _mlstatus=0
++ '[' -n x ']'
++ IFS=' 	
+'
++ unset _mlre _mlv _mlrv _mlIFS
++ '[' -n '' ']'
++ unset _mlshdbg
++ return 0
++ source /fsx/nouamane/miniconda/bin/activate
+++ _CONDA_ROOT=/fsx/nouamane/miniconda
+++ . /fsx/nouamane/miniconda/etc/profile.d/conda.sh
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ '[' -z x ']'
+++ conda activate
+++ local cmd=activate
+++ case "$cmd" in
+++ __conda_activate activate
+++ '[' -n '' ']'
+++ local ask_conda
++++ PS1=
++++ __conda_exe shell.posix activate
++++ /fsx/nouamane/miniconda/bin/conda shell.posix activate
+Error while loading conda entry point: conda-libmamba-solver (module 'libmambapy' has no attribute 'QueryFormat')
+++ ask_conda='. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''2'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_1='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/etc/conda/activate.d/libxml2_activate.sh"'
+++ eval '. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''2'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_1='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/etc/conda/activate.d/libxml2_activate.sh"'
++++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh
+++++ test -n ''
+++++ unset XML_CATALOG_FILES
+++++ unset xml_catalog_files_libxml2
++++ PS1='(base) '
++++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
++++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
++++ export CONDA_PREFIX=/fsx/nouamane/miniconda
++++ CONDA_PREFIX=/fsx/nouamane/miniconda
++++ export CONDA_SHLVL=2
++++ CONDA_SHLVL=2
++++ export CONDA_DEFAULT_ENV=base
++++ CONDA_DEFAULT_ENV=base
++++ export 'CONDA_PROMPT_MODIFIER=(base) '
++++ CONDA_PROMPT_MODIFIER='(base) '
++++ export CONDA_PREFIX_1=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ CONDA_PREFIX_1=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ . /fsx/nouamane/miniconda/etc/conda/activate.d/libxml2_activate.sh
+++++ test -n ''
+++++ xml_catalog_files_libxml2=
+++++ XML_CATALOG_FILES=
+++++ conda_catalog_files=
+++++ ifs_libxml2=' 	
+'
+++++ IFS=' '
+++++ rem=/fsx/nouamane/miniconda
+++++ for pre in ${rem}
+++++ test '' = /fsx/nouamane/miniconda
+++++ conda_catalog_files=/fsx/nouamane/miniconda
+++++ rem=
+++++ IFS=' 	
+'
+++++ conda_catalog_files='file:///fsx/nouamane/miniconda/etc/xml/catalog file:///etc/xml/catalog'
+++++ export 'XML_CATALOG_FILES=file:///fsx/nouamane/miniconda/etc/xml/catalog file:///etc/xml/catalog'
+++++ XML_CATALOG_FILES='file:///fsx/nouamane/miniconda/etc/xml/catalog file:///etc/xml/catalog'
+++++ unset conda_catalog_files ifs_libxml2 rem
+++ __conda_hashr
+++ '[' -n '' ']'
+++ '[' -n '' ']'
+++ hash -r
++ conda activate 2-1-cu121
++ local cmd=activate
++ case "$cmd" in
++ __conda_activate activate 2-1-cu121
++ '[' -n '' ']'
++ local ask_conda
+++ PS1='(base) '
+++ __conda_exe shell.posix activate 2-1-cu121
+++ /fsx/nouamane/miniconda/bin/conda shell.posix activate 2-1-cu121
+Error while loading conda entry point: conda-libmamba-solver (module 'libmambapy' has no attribute 'QueryFormat')
++ ask_conda='. "/fsx/nouamane/miniconda/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_2='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
++ eval '. "/fsx/nouamane/miniconda/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_2='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
+++ . /fsx/nouamane/miniconda/etc/conda/deactivate.d/libxml2_deactivate.sh
++++ test -n ''
++++ unset XML_CATALOG_FILES
++++ unset xml_catalog_files_libxml2
+++ PS1='(2-1-cu121) '
+++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
+++ export CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ export CONDA_SHLVL=3
+++ CONDA_SHLVL=3
+++ export CONDA_DEFAULT_ENV=2-1-cu121
+++ CONDA_DEFAULT_ENV=2-1-cu121
+++ export 'CONDA_PROMPT_MODIFIER=(2-1-cu121) '
+++ CONDA_PROMPT_MODIFIER='(2-1-cu121) '
+++ export CONDA_PREFIX_2=/fsx/nouamane/miniconda
+++ CONDA_PREFIX_2=/fsx/nouamane/miniconda
+++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ export _CE_M=
+++ _CE_M=
+++ export _CE_CONDA=
+++ _CE_CONDA=
+++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh
++++ test -n ''
++++ xml_catalog_files_libxml2=
++++ XML_CATALOG_FILES=
++++ conda_catalog_files=
++++ ifs_libxml2=' 	
+'
++++ IFS=' '
++++ rem=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ for pre in ${rem}
++++ test '' = /fsx/nouamane/miniconda/envs/2-1-cu121
++++ conda_catalog_files=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ rem=
++++ IFS=' 	
+'
++++ conda_catalog_files='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ export 'XML_CATALOG_FILES=file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ XML_CATALOG_FILES='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ unset conda_catalog_files ifs_libxml2 rem
++ __conda_hashr
++ '[' -n '' ']'
++ '[' -n '' ']'
++ hash -r
++ export PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
++ PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
++ '[' -z '' ']'
+++ scontrol show hostnames 'ip-26-0-175-[170,241]'
++ export 'NODELIST=ip-26-0-175-170
+ip-26-0-175-241'
++ NODELIST='ip-26-0-175-170
+ip-26-0-175-241'
+++ head -n1
+++ scontrol show hostnames 'ip-26-0-175-[170,241]'
++ export MASTER_NODE=ip-26-0-175-170
++ MASTER_NODE=ip-26-0-175-170
++ export MASTER_PORT=12356
++ MASTER_PORT=12356
++ export NNODES=2
++ NNODES=2
++ export GPUS_PER_NODE=8
++ GPUS_PER_NODE=8
++ export WORLD_SIZE=16
++ WORLD_SIZE=16
++ export CUDA_DEVICE_MAX_CONNECTIONS=1
++ CUDA_DEVICE_MAX_CONNECTIONS=1
++ export NCCL_DEBUG=WARN
++ NCCL_DEBUG=WARN
++ export NANOTRON_BENCHMARK=1
++ NANOTRON_BENCHMARK=1
++ export WANDB_MODE=disabled
++ WANDB_MODE=disabled
++ export TORCH_NCCL_ASYNC_ERROR_HANDLING=1
++ TORCH_NCCL_ASYNC_ERROR_HANDLING=1
++ export TORCH_DISTRIBUTED_DEBUG=DETAIL
++ TORCH_DISTRIBUTED_DEBUG=DETAIL
++ '[' -z '' ']'
++ echo '=== GPU Topology ==='
+=== GPU Topology ===
++ nvidia-smi topo -m
+	GPU0	GPU1	GPU2	GPU3	GPU4	GPU5	GPU6	GPU7	CPU Affinity	NUMA Affinity	GPU NUMA ID
+GPU0	 X 	NV18	NV18	NV18	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU1	NV18	 X 	NV18	NV18	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU2	NV18	NV18	 X 	NV18	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU3	NV18	NV18	NV18	 X 	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU4	NV18	NV18	NV18	NV18	 X 	NV18	NV18	NV18	48-95	1		N/A
+GPU5	NV18	NV18	NV18	NV18	NV18	 X 	NV18	NV18	48-95	1		N/A
+GPU6	NV18	NV18	NV18	NV18	NV18	NV18	 X 	NV18	48-95	1		N/A
+GPU7	NV18	NV18	NV18	NV18	NV18	NV18	NV18	 X 	48-95	1		N/A
+
+Legend:
+
+  X    = Self
+  SYS  = Connection traversing PCIe as well as the SMP interconnect between NUMA nodes (e.g., QPI/UPI)
+  NODE = Connection traversing PCIe as well as the interconnect between PCIe Host Bridges within a NUMA node
+  PHB  = Connection traversing PCIe as well as a PCIe Host Bridge (typically the CPU)
+  PXB  = Connection traversing multiple PCIe bridges (without traversing the PCIe Host Bridge)
+  PIX  = Connection traversing at most a single PCIe bridge
+  NV#  = Connection traversing a bonded set of # NVLinks
++ echo ==================
+==================
++ export SRUN_ALLOC_ARGS=
++ SRUN_ALLOC_ARGS=
++ echo 'Master node: ip-26-0-175-170'
+Master node: ip-26-0-175-170
++ echo 'All nodes: ip-26-0-175-170
+ip-26-0-175-241'
+All nodes: ip-26-0-175-170
+ip-26-0-175-241
++ echo 'World size: 16'
+World size: 16
++ '[' -n '' ']'
++ srun --wait=0 --kill-on-bad-exit=1 torchrun --nnodes=2 --nproc_per_node=8 --rdzv_id=14018985 --rdzv_backend=c10d --rdzv_endpoint=ip-26-0-175-170:12356 --max_restarts 0 --rdzv_conf timeout=60 /fsx/nouamane/projects/nanotron/run_train.py --config-file benchmark/configs/config_3.57G_dp4_tp1_pp4_acc64_mbs1_seq4096_zero0_tpmodeRED_vocab131k.yaml
+[2025-01-02 13:45:15,123] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-02 13:45:15,152] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-02 13:45:15,124] torch.distributed.run: [WARNING] 
+[2025-01-02 13:45:15,124] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 13:45:15,124] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-02 13:45:15,124] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 13:45:15,153] torch.distributed.run: [WARNING] 
+[2025-01-02 13:45:15,153] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 13:45:15,153] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-02 13:45:15,153] torch.distributed.run: [WARNING] *****************************************
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+01/02/2025 13:45:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]: Config:
+01/02/2025 13:45:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]: Config(general=GeneralArgs(project='debug',
+01/02/2025 13:45:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                            run='3.57G_dp4_tp1_pp4_acc64_mbs1_seq4096_zero0_tpmodeRED_vocab131k',
+01/02/2025 13:45:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                            seed=42,
+01/02/2025 13:45:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                            step=None,
+01/02/2025 13:45:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                            consumed_train_samples=None,
+01/02/2025 13:45:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                            benchmark_csv_path=PosixPath('benchmark/results/bench_final2.csv'),
+01/02/2025 13:45:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                            ignore_sanity_checks=True),
+01/02/2025 13:45:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:        parallelism=ParallelismArgs(dp=4,
+01/02/2025 13:45:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                                    pp=4,
+01/02/2025 13:45:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                                    tp=1,
+01/02/2025 13:45:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                                    pp_engine=<nanotron.parallel.pipeline_parallel.engine.OneForwardOneBackwardPipelineEngine object at 0x7fcd9f620dc0>,
+01/02/2025 13:45:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                                    tp_mode=<TensorParallelLinearMode.REDUCE_SCATTER: 2>,
+01/02/2025 13:45:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                                    tp_linear_async_communication=True,
+01/02/2025 13:45:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                                    recompute_layer=False,
+01/02/2025 13:45:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                                    tp_recompute_allgather=True,
+01/02/2025 13:45:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                                    expert_parallel_size=1),
+01/02/2025 13:45:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:        model=ModelArgs(model_config=LlamaConfig(bos_token_id=0,
+01/02/2025 13:45:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                                                 eos_token_id=0,
+01/02/2025 13:45:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                                                 hidden_act='silu',
+01/02/2025 13:45:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                                                 hidden_size=3072,
+01/02/2025 13:45:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                                                 initializer_range=0.02,
+01/02/2025 13:45:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                                                 intermediate_size=8192,
+01/02/2025 13:45:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                                                 is_llama_config=True,
+01/02/2025 13:45:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                                                 max_position_embeddings=4096,
+01/02/2025 13:45:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                                                 num_attention_heads=32,
+01/02/2025 13:45:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                                                 num_hidden_layers=28,
+01/02/2025 13:45:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                                                 num_key_value_heads=32,
+01/02/2025 13:45:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                                                 pad_token_id=None,
+01/02/2025 13:45:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                                                 pretraining_tp=1,
+01/02/2025 13:45:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                                                 rms_norm_eps=1e-05,
+01/02/2025 13:45:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                                                 rope_scaling=None,
+01/02/2025 13:45:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                                                 rope_theta=10000.0,
+01/02/2025 13:45:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                                                 rope_interleaved=False,
+01/02/2025 13:45:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                                                 tie_word_embeddings=True,
+01/02/2025 13:45:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                                                 use_cache=True,
+01/02/2025 13:45:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                                                 vocab_size=131072),
+01/02/2025 13:45:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                        init_method=RandomInit(std=0.02),
+01/02/2025 13:45:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                        dtype=torch.bfloat16,
+01/02/2025 13:45:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                        make_vocab_size_divisible_by=1,
+01/02/2025 13:45:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                        ddp_bucket_cap_mb=25),
+01/02/2025 13:45:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:        tokenizer=TokenizerArgs(tokenizer_name_or_path='robot-test/dummy-tokenizer-wordlevel',
+01/02/2025 13:45:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                                tokenizer_revision=None,
+01/02/2025 13:45:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                                tokenizer_max_length=None),
+01/02/2025 13:45:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:        checkpoints=CheckpointsArgs(checkpoints_path=PosixPath('checkpoints'),
+01/02/2025 13:45:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                                    checkpoint_interval=10000,
+01/02/2025 13:45:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                                    save_initial_state=False,
+01/02/2025 13:45:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                                    save_final_state=False,
+01/02/2025 13:45:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                                    resume_checkpoint_path=None,
+01/02/2025 13:45:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                                    checkpoints_path_is_shared_file_system=False),
+01/02/2025 13:45:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:        logging=LoggingArgs(log_level='info',
+01/02/2025 13:45:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                            log_level_replica='info',
+01/02/2025 13:45:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                            iteration_step_info_interval=1),
+01/02/2025 13:45:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:        tokens=TokensArgs(sequence_length=4096,
+01/02/2025 13:45:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                          train_steps=100,
+01/02/2025 13:45:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                          micro_batch_size=1,
+01/02/2025 13:45:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                          batch_accumulation_per_replica=64,
+01/02/2025 13:45:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                          val_check_interval=100,
+01/02/2025 13:45:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                          limit_val_batches=0,
+01/02/2025 13:45:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                          limit_test_batches=0),
+01/02/2025 13:45:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:        optimizer=OptimizerArgs(optimizer_factory=AdamWOptimizerArgs(adam_eps=1e-08,
+01/02/2025 13:45:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                                                                     adam_beta1=0.9,
+01/02/2025 13:45:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                                                                     adam_beta2=0.95,
+01/02/2025 13:45:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                                                                     torch_adam_is_fused=True,
+01/02/2025 13:45:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                                                                     name='adamW'),
+01/02/2025 13:45:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                                zero_stage=0,
+01/02/2025 13:45:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                                weight_decay=0.01,
+01/02/2025 13:45:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                                clip_grad=1.0,
+01/02/2025 13:45:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                                accumulate_grad_in_fp32=True,
+01/02/2025 13:45:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                                learning_rate_scheduler=LRSchedulerArgs(learning_rate=0.0003,
+01/02/2025 13:45:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                                                                        lr_warmup_steps=2,
+01/02/2025 13:45:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                                                                        lr_warmup_style='linear',
+01/02/2025 13:45:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                                                                        lr_decay_style='cosine',
+01/02/2025 13:45:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                                                                        lr_decay_steps=13,
+01/02/2025 13:45:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                                                                        lr_decay_starting_step=None,
+01/02/2025 13:45:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                                                                        min_decay_lr=1e-05)),
+01/02/2025 13:45:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:        data_stages=[DatasetStageArgs(name='Stable Training Stage',
+01/02/2025 13:45:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                                      start_training_step=1,
+01/02/2025 13:45:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                                      data=DataArgs(dataset=None,
+01/02/2025 13:45:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                                                    seed=42,
+01/02/2025 13:45:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:                                                    num_loading_workers=1))],
+01/02/2025 13:45:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:        profiler=None,
+01/02/2025 13:45:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:        lighteval=None,
+01/02/2025 13:45:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:        s3_upload=None)
+01/02/2025 13:45:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]: Model Config:
+01/02/2025 13:45:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]: LlamaConfig(bos_token_id=0,
+01/02/2025 13:45:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:             eos_token_id=0,
+01/02/2025 13:45:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:             hidden_act='silu',
+01/02/2025 13:45:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:             hidden_size=3072,
+01/02/2025 13:45:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:             initializer_range=0.02,
+01/02/2025 13:45:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:             intermediate_size=8192,
+01/02/2025 13:45:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:             is_llama_config=True,
+01/02/2025 13:45:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:             max_position_embeddings=4096,
+01/02/2025 13:45:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:             num_attention_heads=32,
+01/02/2025 13:45:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:             num_hidden_layers=28,
+01/02/2025 13:45:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:             num_key_value_heads=32,
+01/02/2025 13:45:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:             pad_token_id=None,
+01/02/2025 13:45:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:             pretraining_tp=1,
+01/02/2025 13:45:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:             rms_norm_eps=1e-05,
+01/02/2025 13:45:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:             rope_scaling=None,
+01/02/2025 13:45:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:             rope_theta=10000.0,
+01/02/2025 13:45:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:             rope_interleaved=False,
+01/02/2025 13:45:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:             tie_word_embeddings=True,
+01/02/2025 13:45:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:             use_cache=True,
+01/02/2025 13:45:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:             vocab_size=131072)
+01/02/2025 13:45:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]: Building model..
+01/02/2025 13:45:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]: Initialize RoPE Theta = 10000.0
+01/02/2025 13:45:53 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]: Setting PP block ranks...
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+01/02/2025 13:45:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]: Total number of parameters: 3.98G (7584.33MiB)
+01/02/2025 13:45:56 [INFO|DP=0|PP=1|TP=0|ip-26-0-175-170]: Local number of parameters: 906M (1728.09MiB)
+01/02/2025 13:45:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]: Local number of parameters: 1.31G (2496.09MiB)
+01/02/2025 13:45:56 [INFO|DP=0|PP=3|TP=0|ip-26-0-175-241]: Local number of parameters: 856M (1632.05MiB)
+01/02/2025 13:45:56 [INFO|DP=0|PP=2|TP=0|ip-26-0-175-241]: Local number of parameters: 906M (1728.09MiB)
+01/02/2025 13:45:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]: [After model building] Memory usage: 2496.11MiB. Peak allocated: 5504.00MiB Peak reserved: 12674.00MiB
+01/02/2025 13:45:56 [INFO|DP=0|PP=1|TP=0|ip-26-0-175-170]: [After model building] Memory usage: 1728.11MiB. Peak allocated: 5504.00MiB Peak reserved: 12674.00MiB
+01/02/2025 13:45:56 [INFO|DP=0|PP=3|TP=0|ip-26-0-175-241]: [After model building] Memory usage: 1632.07MiB. Peak allocated: 5504.00MiB Peak reserved: 11650.00MiB
+01/02/2025 13:45:56 [INFO|DP=0|PP=2|TP=0|ip-26-0-175-241]: [After model building] Memory usage: 1728.11MiB. Peak allocated: 5504.00MiB Peak reserved: 11650.00MiB
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+        return f(*args, **kwargs)raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+
+        outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)model.p2p.clear_history()
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+01/02/2025 13:46:01 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]: No checkpoint path provided.
+01/02/2025 13:46:01 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]: Parametrizing model parameters using StandardParametrizator
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+01/02/2025 13:46:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]: [Optimizer Building] Using LearningRateForSP as learning rate
+
+ip-26-0-175-241:109572:110101 [7] register_rail_mr_buffer:539 NCCL WARN NET/OFI Unable to register memory (type = 2) for device 7. RC: -14, Error: Bad address
+
+ip-26-0-175-241:109572:110101 [7] proxy.cc:1485 NCCL WARN [Service thread] Error encountered progressing operation=Connect, res=3, closing connection
+
+ip-26-0-175-241:109572:110101 [7] proxy.cc:1519 NCCL WARN [Proxy Service 1] Failed to execute operation Connect from rank 1, retcode 3
+
+ip-26-0-175-241:109572:110090 [7] misc/socket.cc:49 NCCL WARN socketProgress: Connection closed by remote peer ip-26-0-175-241.ec2.internal<44949>
+
+ip-26-0-175-241:109572:110090 [7] proxy.cc:1143 NCCL WARN Socket recv failed while polling for opId=0x7f2998459e90
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 171, in __init__
+    self.model = self.init_model()  # Defines self.model
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 721, in init_model
+    model = self._load_model_checkpoint(model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 784, in _load_model_checkpoint
+    dist.all_reduce(param, op=dist.ReduceOp.AVG, group=group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/c10d_logger.py", line 47, in wrapper
+    return func(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/distributed_c10d.py", line 2050, in all_reduce
+    work = group.allreduce([tensor], opts)
+torch.distributed.DistBackendError: NCCL error in: ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1333, internal error - please report this issue to the NCCL developers, NCCL version 2.18.5
+ncclInternalError: Internal check failed.
+Last error:
+Socket recv failed while polling for opId=0x7f2998459e90
+01/02/2025 13:46:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]: module_name: model.token_position_embeddings | PP: 0/4 | Block rank: 0
+01/02/2025 13:46:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]: module_name: model.decoder.0 | PP: 0/4 | Block rank: 1
+01/02/2025 13:46:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]: module_name: model.decoder.1 | PP: 0/4 | Block rank: 2
+01/02/2025 13:46:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]: module_name: model.decoder.2 | PP: 0/4 | Block rank: 3
+01/02/2025 13:46:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]: module_name: model.decoder.3 | PP: 0/4 | Block rank: 4
+01/02/2025 13:46:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]: module_name: model.decoder.4 | PP: 0/4 | Block rank: 5
+01/02/2025 13:46:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]: module_name: model.decoder.5 | PP: 0/4 | Block rank: 6
+01/02/2025 13:46:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]: module_name: model.decoder.6 | PP: 0/4 | Block rank: 7
+01/02/2025 13:46:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]: module_name: model.decoder.7 | PP: 0/4 | Block rank: 8
+01/02/2025 13:46:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]: module_name: model.decoder.8 | PP: 1/4 | Block rank: 0
+01/02/2025 13:46:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]: module_name: model.decoder.9 | PP: 1/4 | Block rank: 1
+01/02/2025 13:46:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]: module_name: model.decoder.10 | PP: 1/4 | Block rank: 2
+01/02/2025 13:46:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]: module_name: model.decoder.11 | PP: 1/4 | Block rank: 3
+01/02/2025 13:46:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]: module_name: model.decoder.12 | PP: 1/4 | Block rank: 4
+01/02/2025 13:46:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]: module_name: model.decoder.13 | PP: 1/4 | Block rank: 5
+01/02/2025 13:46:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]: module_name: model.decoder.14 | PP: 1/4 | Block rank: 6
+01/02/2025 13:46:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]: module_name: model.decoder.15 | PP: 1/4 | Block rank: 7
+01/02/2025 13:46:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]: module_name: model.decoder.16 | PP: 2/4 | Block rank: 0
+01/02/2025 13:46:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]: module_name: model.decoder.17 | PP: 2/4 | Block rank: 1
+01/02/2025 13:46:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]: module_name: model.decoder.18 | PP: 2/4 | Block rank: 2
+01/02/2025 13:46:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]: module_name: model.decoder.19 | PP: 2/4 | Block rank: 3
+01/02/2025 13:46:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]: module_name: model.decoder.20 | PP: 2/4 | Block rank: 4
+01/02/2025 13:46:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]: module_name: model.decoder.21 | PP: 2/4 | Block rank: 5
+01/02/2025 13:46:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]: module_name: model.decoder.22 | PP: 2/4 | Block rank: 6
+01/02/2025 13:46:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]: module_name: model.decoder.23 | PP: 2/4 | Block rank: 7
+01/02/2025 13:46:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]: module_name: model.decoder.24 | PP: 3/4 | Block rank: 0
+01/02/2025 13:46:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]: module_name: model.decoder.25 | PP: 3/4 | Block rank: 1
+01/02/2025 13:46:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]: module_name: model.decoder.26 | PP: 3/4 | Block rank: 2
+01/02/2025 13:46:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]: module_name: model.decoder.27 | PP: 3/4 | Block rank: 3
+01/02/2025 13:46:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]: module_name: model.final_layer_norm | PP: 3/4 | Block rank: 4
+01/02/2025 13:46:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]: module_name: model.lm_head | PP: 3/4 | Block rank: 5
+01/02/2025 13:46:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]: module_name: model.cast_to_fp32 | PP: 3/4 | Block rank: 6
+01/02/2025 13:46:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]: module_name: loss | PP: 3/4 | Block rank: 7
+01/02/2025 13:46:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]: [Training Plan] Stage Stable Training Stage has 99 remaining training steps and has consumed 0 samples
+01/02/2025 13:46:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]: Using dummy data generator
+01/02/2025 13:46:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]: [Training Plan] There are 1 training stages 
+01/02/2025 13:46:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]: [Stage Stable Training Stage] start from step 1 
+01/02/2025 13:46:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]: 
+01/02/2025 13:46:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]: [Start training] datetime: 2025-01-02 13:46:02.357640 | mbs: 1 | grad_accum: 64 | global_batch_size: 256 | sequence_length: 4096 | train_steps: 100 | start_iteration_step: 0 | consumed_train_samples: 0
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+01/02/2025 13:46:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]: Resuming training from stage Stable Training Stage, it has trained for 0 samples and has 99 remaining train steps
+01/02/2025 13:46:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-175-170]:  Memory usage: 14976.58MiB. Peak allocated 14976.58MiB. Peak reserved: 25160.00MiB
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+[2025-01-02 13:46:16,459] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 109569 closing signal SIGTERM
+[2025-01-02 13:46:16,460] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 109570 closing signal SIGTERM
+[2025-01-02 13:46:16,460] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 109571 closing signal SIGTERM
+[2025-01-02 13:46:16,460] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 109572 closing signal SIGTERM
+[2025-01-02 13:46:21,456] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 85391 closing signal SIGTERM
+[2025-01-02 13:46:33,239] torch.distributed.elastic.multiprocessing.api: [ERROR] failed (exitcode: 1) local_rank: 4 (pid: 85392) of binary: /fsx/nouamane/miniconda/envs/2-1-cu121/bin/python
+[2025-01-02 13:46:33,290] torch.distributed.elastic.multiprocessing.errors.error_handler: [ERROR] no error file defined for parent, to copy child error file (/tmp/torchelastic_1o9jbak1/14018985_h56rd8l3/attempt_0/4/error.json)
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 264, in launch_agent
+    raise ChildFailedError(
+torch.distributed.elastic.multiprocessing.errors.ChildFailedError: 
+============================================================
+/fsx/nouamane/projects/nanotron/run_train.py FAILED
+------------------------------------------------------------
+Failures:
+[0]:
+  time      : 2025-01-02_13:46:02
+  host      : ip-26-0-175-170.ec2.internal
+  rank      : 0 (local_rank: 0)
+  exitcode  : 1 (pid: 85388)
+  error_file: /tmp/torchelastic_1o9jbak1/14018985_h56rd8l3/attempt_0/0/error.json
+  traceback : Traceback (most recent call last):
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+      return f(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+      outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+      outputs = self.pipeline_engine.train_batch_iter(
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+      model.p2p.clear_history()
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+      raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+  AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+  
+[1]:
+  time      : 2025-01-02_13:46:02
+  host      : ip-26-0-175-170.ec2.internal
+  rank      : 1 (local_rank: 1)
+  exitcode  : 1 (pid: 85389)
+  error_file: /tmp/torchelastic_1o9jbak1/14018985_h56rd8l3/attempt_0/1/error.json
+  traceback : Traceback (most recent call last):
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+      return f(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+      outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+      outputs = self.pipeline_engine.train_batch_iter(
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+      model.p2p.clear_history()
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+      raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+  AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+  
+[2]:
+  time      : 2025-01-02_13:46:02
+  host      : ip-26-0-175-170.ec2.internal
+  rank      : 2 (local_rank: 2)
+  exitcode  : 1 (pid: 85390)
+  error_file: /tmp/torchelastic_1o9jbak1/14018985_h56rd8l3/attempt_0/2/error.json
+  traceback : Traceback (most recent call last):
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+      return f(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+      outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+      outputs = self.pipeline_engine.train_batch_iter(
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+      model.p2p.clear_history()
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+      raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+  AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+  
+[4]:
+  time      : 2025-01-02_13:46:01
+  host      : ip-26-0-175-170.ec2.internal
+  rank      : 5 (local_rank: 5)
+  exitcode  : 1 (pid: 85393)
+  error_file: /tmp/torchelastic_1o9jbak1/14018985_h56rd8l3/attempt_0/5/error.json
+  traceback : Traceback (most recent call last):
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+      return f(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+      outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+      outputs = self.pipeline_engine.train_batch_iter(
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+      model.p2p.clear_history()
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+      raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+  AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+  
+[5]:
+  time      : 2025-01-02_13:46:01
+  host      : ip-26-0-175-170.ec2.internal
+  rank      : 6 (local_rank: 6)
+  exitcode  : 1 (pid: 85394)
+  error_file: /tmp/torchelastic_1o9jbak1/14018985_h56rd8l3/attempt_0/6/error.json
+  traceback : Traceback (most recent call last):
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+      return f(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+      outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+      outputs = self.pipeline_engine.train_batch_iter(
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+      model.p2p.clear_history()
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+      raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+  AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+  
+[6]:
+  time      : 2025-01-02_13:46:01
+  host      : ip-26-0-175-170.ec2.internal
+  rank      : 7 (local_rank: 7)
+  exitcode  : 1 (pid: 85395)
+  error_file: /tmp/torchelastic_1o9jbak1/14018985_h56rd8l3/attempt_0/7/error.json
+  traceback : Traceback (most recent call last):
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+      return f(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+      outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+      outputs = self.pipeline_engine.train_batch_iter(
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+      model.p2p.clear_history()
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+      raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+  AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+  
+------------------------------------------------------------
+Root Cause (first observed failure):
+[3]:
+  time      : 2025-01-02_13:46:01
+  host      : ip-26-0-175-170.ec2.internal
+  rank      : 4 (local_rank: 4)
+  exitcode  : 1 (pid: 85392)
+  error_file: /tmp/torchelastic_1o9jbak1/14018985_h56rd8l3/attempt_0/4/error.json
+  traceback : Traceback (most recent call last):
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+      return f(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+      outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+      outputs = self.pipeline_engine.train_batch_iter(
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+      model.p2p.clear_history()
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+      raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+  AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+  
+============================================================
+srun: error: ip-26-0-175-170: task 0: Exited with exit code 1
+srun: launch/slurm: _step_signal: Terminating StepId=14018985.0
+[2025-01-02 13:46:33,553] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-02 13:46:33,553] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 109570 closing signal SIGTERM
+[2025-01-02 13:46:33,553] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 109571 closing signal SIGTERM
+[2025-01-02 13:46:33,553] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 109572 closing signal SIGTERM
+[2025-01-02 13:46:36,243] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-175-241.ec2.internal_109494_0' has failed to send a keep-alive heartbeat to the rendezvous '14018985' due to an error of type RendezvousConnectionError.
+[2025-01-02 13:46:41,245] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-175-241.ec2.internal_109494_0' has failed to send a keep-alive heartbeat to the rendezvous '14018985' due to an error of type RendezvousConnectionError.
+[2025-01-02 13:46:46,247] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-175-241.ec2.internal_109494_0' has failed to send a keep-alive heartbeat to the rendezvous '14018985' due to an error of type RendezvousConnectionError.
+[2025-01-02 13:46:51,249] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-175-241.ec2.internal_109494_0' has failed to send a keep-alive heartbeat to the rendezvous '14018985' due to an error of type RendezvousConnectionError.
+[2025-01-02 13:46:56,250] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-175-241.ec2.internal_109494_0' has failed to send a keep-alive heartbeat to the rendezvous '14018985' due to an error of type RendezvousConnectionError.
+[2025-01-02 13:47:01,252] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-175-241.ec2.internal_109494_0' has failed to send a keep-alive heartbeat to the rendezvous '14018985' due to an error of type RendezvousConnectionError.
+slurmstepd: error: *** STEP 14018985.0 STEPD TERMINATED ON ip-26-0-175-241 AT 2025-01-02T13:50:03 DUE TO JOB NOT ENDING WITH SIGNALS ***
+srun: error: ip-26-0-175-241: task 1: Killed
+srun: Force Terminated StepId=14018985.0
diff --git a/logs/14019012-bench_3.57G_dp4_tp2_pp4_acc64_mbs1_seq4096_zero0_tpmodeRED_vocab131k.out b/logs/14019012-bench_3.57G_dp4_tp2_pp4_acc64_mbs1_seq4096_zero0_tpmodeRED_vocab131k.out
new file mode 100644
index 0000000000000000000000000000000000000000..cadc7c56e159ac53aeb6e115fee6c8c6bc7adf68
--- /dev/null
+++ b/logs/14019012-bench_3.57G_dp4_tp2_pp4_acc64_mbs1_seq4096_zero0_tpmodeRED_vocab131k.out
@@ -0,0 +1,1340 @@
++ '[' -z 14019012 ']'
++ source /etc/profile.d/modules.sh
+++ . /usr/share/modules/init/bash
++++ unset _mlshdbg
++++ '[' 0 = 1 ']'
++++ unset _mlre _mlIFS
++++ '[' -n x ']'
++++ _mlIFS=' 	
+'
++++ IFS=' '
++++ '[' -n '' ']'
+++++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash autoinit
++++ _mlcode='module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+MODULEPATH_modshare=/etc/environment-modules/modules:1:/usr/share/modules/\$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1; export MODULEPATH_modshare;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+LOADEDMODULES=; export LOADEDMODULES;
+MODULEPATH=/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/\$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles; export MODULEPATH;
+test 0;'
++++ _mlret=0
++++ '[' -n x ']'
++++ IFS=' 	
+'
++++ unset _mlIFS
++++ unset _mlre _mlv _mlrv
++++ '[' 0 -eq 0 ']'
++++ eval 'module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+MODULEPATH_modshare=/etc/environment-modules/modules:1:/usr/share/modules/\$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1; export MODULEPATH_modshare;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+LOADEDMODULES=; export LOADEDMODULES;
+MODULEPATH=/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/\$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles; export MODULEPATH;
+test 0;'
+++++ MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl
+++++ export MODULES_CMD
+++++ ENV=/usr/share/modules/init/profile.sh
+++++ export ENV
+++++ MODULEPATH_modshare='/etc/environment-modules/modules:1:/usr/share/modules/$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1'
+++++ export MODULEPATH_modshare
+++++ BASH_ENV=/usr/share/modules/init/bash
+++++ export BASH_ENV
+++++ MODULESHOME=/usr/share/modules
+++++ export MODULESHOME
+++++ LOADEDMODULES=
+++++ export LOADEDMODULES
+++++ MODULEPATH='/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles'
+++++ export MODULEPATH
+++++ test 0
++++ '[' 0 = 1 ']'
++++ '[' -t 2 ']'
++++ export -f module
++++ export -f switchml
++++ '[' 5 -ge 3 ']'
++++ [[ ehxB =~ i ]]
++++ [[ ! :/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin: =~ :/usr/bin: ]]
++++ '[' '!' -n '' ']'
++++ MANPATH=:
++++ export MANPATH
+++++ manpath
++++ [[ ! :/fsx/nouamane/miniconda/envs/2-1-cu121/man:/fsx/nouamane/miniconda/envs/2-1-cu121/share/man:/fsx/nouamane/miniconda/man:/fsx/nouamane/miniconda/share/man:/usr/local/man:/usr/local/share/man:/usr/share/man:/opt/slurm/share/man:/admin/home/nouamane/.local/share/man:/admin/home/nouamane/.fzf/man:: =~ :/usr/share/man: ]]
++++ unset _mlcode _mlret
++++ '[' -n '' ']'
++ module load cuda/12.1
++ unset _mlshdbg
++ '[' 0 = 1 ']'
++ unset _mlre _mlIFS
++ '[' -n x ']'
++ _mlIFS=' 	
+'
++ IFS=' '
++ '[' -n '' ']'
+++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash load cuda/12.1
++ eval 'CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include;' export 'CPATH;
+LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:;' export 'LD_LIBRARY_PATH;
+FI_EFA_FORK_SAFE=1;' export 'FI_EFA_FORK_SAFE;
+MANPATH=/usr/local/cuda-12.1/share/man::;' export 'MANPATH;
+LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64;' export 'LIBRARY_PATH;
+_LMFILES_=/usr/share/modules/modulefiles/cuda/12.1;' export '_LMFILES_;
+LOADEDMODULES=cuda/12.1;' export 'LOADEDMODULES;
+MPI_PATH=/opt/amazon/openmpi;' export 'MPI_PATH;
+NCCL_HOME_modshare=/opt/nccl/build:1;' export 'NCCL_HOME_modshare;
+NCCL_PROTO=simple;' export 'NCCL_PROTO;
+MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1;' export 'MANPATH_modshare;
+LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1;' export 'LIBRARY_PATH_modshare;
+NCCL_SOCKET_IFNAME=enp;' export 'NCCL_SOCKET_IFNAME;
+AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl;' export 'AWS_OFI_NCCL_HOME;
+NCCL_HOME=/opt/nccl/build;' export 'NCCL_HOME;
+FI_PROVIDER=efa;' export 'FI_PROVIDER;
+AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1;' export 'AWS_OFI_NCCL_HOME_modshare;
+CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1;' export 'CPATH_modshare;
+LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2::1:/opt/nccl/build/lib:1:/opt/aws-ofi-nccl/lib:2:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/efa/lib:1;' export 'LD_LIBRARY_PATH_modshare;
+FI_EFA_ENABLE_SHM_TRANSFER=1;' export 'FI_EFA_ENABLE_SHM_TRANSFER;
+_LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1;' export '_LMFILES__modshare;
+LOADEDMODULES_modshare=cuda/12.1:1;' export 'LOADEDMODULES_modshare;
+MPI_PATH_modshare=/opt/amazon/openmpi:1;' export 'MPI_PATH_modshare;
+PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin;' export 'PATH;
+CUDA_HOME=/usr/local/cuda-12.1;' export 'CUDA_HOME;
+PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/opt/slurm/bin:1:/usr/sbin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1;' export 'PATH_modshare;
+test' '0;'
+++ CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include
+++ export CPATH
+++ LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:
+++ export LD_LIBRARY_PATH
+++ FI_EFA_FORK_SAFE=1
+++ export FI_EFA_FORK_SAFE
+++ MANPATH=/usr/local/cuda-12.1/share/man::
+++ export MANPATH
+++ LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64
+++ export LIBRARY_PATH
+++ _LMFILES_=/usr/share/modules/modulefiles/cuda/12.1
+++ export _LMFILES_
+++ LOADEDMODULES=cuda/12.1
+++ export LOADEDMODULES
+++ MPI_PATH=/opt/amazon/openmpi
+++ export MPI_PATH
+++ NCCL_HOME_modshare=/opt/nccl/build:1
+++ export NCCL_HOME_modshare
+++ NCCL_PROTO=simple
+++ export NCCL_PROTO
+++ MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1
+++ export MANPATH_modshare
+++ LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1
+++ export LIBRARY_PATH_modshare
+++ NCCL_SOCKET_IFNAME=enp
+++ export NCCL_SOCKET_IFNAME
+++ AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl
+++ export AWS_OFI_NCCL_HOME
+++ NCCL_HOME=/opt/nccl/build
+++ export NCCL_HOME
+++ FI_PROVIDER=efa
+++ export FI_PROVIDER
+++ AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1
+++ export AWS_OFI_NCCL_HOME_modshare
+++ CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1
+++ export CPATH_modshare
+++ LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2::1:/opt/nccl/build/lib:1:/opt/aws-ofi-nccl/lib:2:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/efa/lib:1
+++ export LD_LIBRARY_PATH_modshare
+++ FI_EFA_ENABLE_SHM_TRANSFER=1
+++ export FI_EFA_ENABLE_SHM_TRANSFER
+++ _LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1
+++ export _LMFILES__modshare
+++ LOADEDMODULES_modshare=cuda/12.1:1
+++ export LOADEDMODULES_modshare
+++ MPI_PATH_modshare=/opt/amazon/openmpi:1
+++ export MPI_PATH_modshare
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
+++ export PATH
+++ CUDA_HOME=/usr/local/cuda-12.1
+++ export CUDA_HOME
+++ PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/opt/slurm/bin:1:/usr/sbin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1
+++ export PATH_modshare
+++ test 0
++ _mlstatus=0
++ '[' -n x ']'
++ IFS=' 	
+'
++ unset _mlre _mlv _mlrv _mlIFS
++ '[' -n '' ']'
++ unset _mlshdbg
++ return 0
++ source /fsx/nouamane/miniconda/bin/activate
+++ _CONDA_ROOT=/fsx/nouamane/miniconda
+++ . /fsx/nouamane/miniconda/etc/profile.d/conda.sh
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ '[' -z x ']'
+++ conda activate
+++ local cmd=activate
+++ case "$cmd" in
+++ __conda_activate activate
+++ '[' -n '' ']'
+++ local ask_conda
++++ PS1=
++++ __conda_exe shell.posix activate
++++ /fsx/nouamane/miniconda/bin/conda shell.posix activate
+Error while loading conda entry point: conda-libmamba-solver (module 'libmambapy' has no attribute 'QueryFormat')
+++ ask_conda='. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''2'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_1='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/etc/conda/activate.d/libxml2_activate.sh"'
+++ eval '. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''2'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_1='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/etc/conda/activate.d/libxml2_activate.sh"'
++++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh
+++++ test -n ''
+++++ unset XML_CATALOG_FILES
+++++ unset xml_catalog_files_libxml2
++++ PS1='(base) '
++++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
++++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
++++ export CONDA_PREFIX=/fsx/nouamane/miniconda
++++ CONDA_PREFIX=/fsx/nouamane/miniconda
++++ export CONDA_SHLVL=2
++++ CONDA_SHLVL=2
++++ export CONDA_DEFAULT_ENV=base
++++ CONDA_DEFAULT_ENV=base
++++ export 'CONDA_PROMPT_MODIFIER=(base) '
++++ CONDA_PROMPT_MODIFIER='(base) '
++++ export CONDA_PREFIX_1=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ CONDA_PREFIX_1=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ . /fsx/nouamane/miniconda/etc/conda/activate.d/libxml2_activate.sh
+++++ test -n ''
+++++ xml_catalog_files_libxml2=
+++++ XML_CATALOG_FILES=
+++++ conda_catalog_files=
+++++ ifs_libxml2=' 	
+'
+++++ IFS=' '
+++++ rem=/fsx/nouamane/miniconda
+++++ for pre in ${rem}
+++++ test '' = /fsx/nouamane/miniconda
+++++ conda_catalog_files=/fsx/nouamane/miniconda
+++++ rem=
+++++ IFS=' 	
+'
+++++ conda_catalog_files='file:///fsx/nouamane/miniconda/etc/xml/catalog file:///etc/xml/catalog'
+++++ export 'XML_CATALOG_FILES=file:///fsx/nouamane/miniconda/etc/xml/catalog file:///etc/xml/catalog'
+++++ XML_CATALOG_FILES='file:///fsx/nouamane/miniconda/etc/xml/catalog file:///etc/xml/catalog'
+++++ unset conda_catalog_files ifs_libxml2 rem
+++ __conda_hashr
+++ '[' -n '' ']'
+++ '[' -n '' ']'
+++ hash -r
++ conda activate 2-1-cu121
++ local cmd=activate
++ case "$cmd" in
++ __conda_activate activate 2-1-cu121
++ '[' -n '' ']'
++ local ask_conda
+++ PS1='(base) '
+++ __conda_exe shell.posix activate 2-1-cu121
+++ /fsx/nouamane/miniconda/bin/conda shell.posix activate 2-1-cu121
+Error while loading conda entry point: conda-libmamba-solver (module 'libmambapy' has no attribute 'QueryFormat')
++ ask_conda='. "/fsx/nouamane/miniconda/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_2='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
++ eval '. "/fsx/nouamane/miniconda/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_2='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
+++ . /fsx/nouamane/miniconda/etc/conda/deactivate.d/libxml2_deactivate.sh
++++ test -n ''
++++ unset XML_CATALOG_FILES
++++ unset xml_catalog_files_libxml2
+++ PS1='(2-1-cu121) '
+++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
+++ export CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ export CONDA_SHLVL=3
+++ CONDA_SHLVL=3
+++ export CONDA_DEFAULT_ENV=2-1-cu121
+++ CONDA_DEFAULT_ENV=2-1-cu121
+++ export 'CONDA_PROMPT_MODIFIER=(2-1-cu121) '
+++ CONDA_PROMPT_MODIFIER='(2-1-cu121) '
+++ export CONDA_PREFIX_2=/fsx/nouamane/miniconda
+++ CONDA_PREFIX_2=/fsx/nouamane/miniconda
+++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ export _CE_M=
+++ _CE_M=
+++ export _CE_CONDA=
+++ _CE_CONDA=
+++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh
++++ test -n ''
++++ xml_catalog_files_libxml2=
++++ XML_CATALOG_FILES=
++++ conda_catalog_files=
++++ ifs_libxml2=' 	
+'
++++ IFS=' '
++++ rem=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ for pre in ${rem}
++++ test '' = /fsx/nouamane/miniconda/envs/2-1-cu121
++++ conda_catalog_files=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ rem=
++++ IFS=' 	
+'
++++ conda_catalog_files='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ export 'XML_CATALOG_FILES=file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ XML_CATALOG_FILES='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ unset conda_catalog_files ifs_libxml2 rem
++ __conda_hashr
++ '[' -n '' ']'
++ '[' -n '' ']'
++ hash -r
++ export PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
++ PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
++ '[' -z '' ']'
+++ scontrol show hostnames 'ip-26-0-160-225,ip-26-0-161-138,ip-26-0-175-[170,241]'
++ export 'NODELIST=ip-26-0-160-225
+ip-26-0-161-138
+ip-26-0-175-170
+ip-26-0-175-241'
++ NODELIST='ip-26-0-160-225
+ip-26-0-161-138
+ip-26-0-175-170
+ip-26-0-175-241'
+++ head -n1
+++ scontrol show hostnames 'ip-26-0-160-225,ip-26-0-161-138,ip-26-0-175-[170,241]'
++ export MASTER_NODE=ip-26-0-160-225
++ MASTER_NODE=ip-26-0-160-225
++ export MASTER_PORT=12356
++ MASTER_PORT=12356
++ export NNODES=4
++ NNODES=4
++ export GPUS_PER_NODE=8
++ GPUS_PER_NODE=8
++ export WORLD_SIZE=32
++ WORLD_SIZE=32
++ export CUDA_DEVICE_MAX_CONNECTIONS=1
++ CUDA_DEVICE_MAX_CONNECTIONS=1
++ export NCCL_DEBUG=WARN
++ NCCL_DEBUG=WARN
++ export NANOTRON_BENCHMARK=1
++ NANOTRON_BENCHMARK=1
++ export WANDB_MODE=disabled
++ WANDB_MODE=disabled
++ export TORCH_NCCL_ASYNC_ERROR_HANDLING=1
++ TORCH_NCCL_ASYNC_ERROR_HANDLING=1
++ export TORCH_DISTRIBUTED_DEBUG=DETAIL
++ TORCH_DISTRIBUTED_DEBUG=DETAIL
++ '[' -z '' ']'
++ echo '=== GPU Topology ==='
+=== GPU Topology ===
++ nvidia-smi topo -m
+	GPU0	GPU1	GPU2	GPU3	GPU4	GPU5	GPU6	GPU7	CPU Affinity	NUMA Affinity	GPU NUMA ID
+GPU0	 X 	NV18	NV18	NV18	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU1	NV18	 X 	NV18	NV18	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU2	NV18	NV18	 X 	NV18	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU3	NV18	NV18	NV18	 X 	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU4	NV18	NV18	NV18	NV18	 X 	NV18	NV18	NV18	48-95	1		N/A
+GPU5	NV18	NV18	NV18	NV18	NV18	 X 	NV18	NV18	48-95	1		N/A
+GPU6	NV18	NV18	NV18	NV18	NV18	NV18	 X 	NV18	48-95	1		N/A
+GPU7	NV18	NV18	NV18	NV18	NV18	NV18	NV18	 X 	48-95	1		N/A
+
+Legend:
+
+  X    = Self
+  SYS  = Connection traversing PCIe as well as the SMP interconnect between NUMA nodes (e.g., QPI/UPI)
+  NODE = Connection traversing PCIe as well as the interconnect between PCIe Host Bridges within a NUMA node
+  PHB  = Connection traversing PCIe as well as a PCIe Host Bridge (typically the CPU)
+  PXB  = Connection traversing multiple PCIe bridges (without traversing the PCIe Host Bridge)
+  PIX  = Connection traversing at most a single PCIe bridge
+  NV#  = Connection traversing a bonded set of # NVLinks
++ echo ==================
+==================
++ export SRUN_ALLOC_ARGS=
++ SRUN_ALLOC_ARGS=
++ echo 'Master node: ip-26-0-160-225'
+Master node: ip-26-0-160-225
++ echo 'All nodes: ip-26-0-160-225
+ip-26-0-161-138
+ip-26-0-175-170
+ip-26-0-175-241'
+All nodes: ip-26-0-160-225
+ip-26-0-161-138
+ip-26-0-175-170
+ip-26-0-175-241
++ echo 'World size: 32'
+World size: 32
++ '[' -n '' ']'
++ srun --wait=0 --kill-on-bad-exit=1 torchrun --nnodes=4 --nproc_per_node=8 --rdzv_id=14019012 --rdzv_backend=c10d --rdzv_endpoint=ip-26-0-160-225:12356 --max_restarts 0 --rdzv_conf timeout=60 /fsx/nouamane/projects/nanotron/run_train.py --config-file benchmark/configs/config_3.57G_dp4_tp2_pp4_acc64_mbs1_seq4096_zero0_tpmodeRED_vocab131k.yaml
+[2025-01-02 22:09:12,887] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-02 22:09:12,886] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-02 22:09:12,892] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-02 22:09:12,887] torch.distributed.run: [WARNING] 
+[2025-01-02 22:09:12,887] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 22:09:12,887] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-02 22:09:12,887] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 22:09:12,887] torch.distributed.run: [WARNING] 
+[2025-01-02 22:09:12,887] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 22:09:12,887] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-02 22:09:12,887] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 22:09:12,892] torch.distributed.run: [WARNING] 
+[2025-01-02 22:09:12,892] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 22:09:12,892] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-02 22:09:12,892] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 22:09:13,110] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-02 22:09:13,110] torch.distributed.run: [WARNING] 
+[2025-01-02 22:09:13,110] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 22:09:13,110] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-02 22:09:13,110] torch.distributed.run: [WARNING] *****************************************
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+01/02/2025 22:09:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: Config:
+01/02/2025 22:09:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: Config(general=GeneralArgs(project='debug',
+01/02/2025 22:09:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                            run='3.57G_dp4_tp2_pp4_acc64_mbs1_seq4096_zero0_tpmodeRED_vocab131k',
+01/02/2025 22:09:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                            seed=42,
+01/02/2025 22:09:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                            step=None,
+01/02/2025 22:09:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                            consumed_train_samples=None,
+01/02/2025 22:09:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                            benchmark_csv_path=PosixPath('benchmark/results/bench_final2.csv'),
+01/02/2025 22:09:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                            ignore_sanity_checks=True),
+01/02/2025 22:09:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:        parallelism=ParallelismArgs(dp=4,
+01/02/2025 22:09:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                    pp=4,
+01/02/2025 22:09:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                    tp=2,
+01/02/2025 22:09:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                    pp_engine=<nanotron.parallel.pipeline_parallel.engine.OneForwardOneBackwardPipelineEngine object at 0x7f3e41bd0dc0>,
+01/02/2025 22:09:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                    tp_mode=<TensorParallelLinearMode.REDUCE_SCATTER: 2>,
+01/02/2025 22:09:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                    tp_linear_async_communication=True,
+01/02/2025 22:09:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                    recompute_layer=False,
+01/02/2025 22:09:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                    tp_recompute_allgather=True,
+01/02/2025 22:09:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                    expert_parallel_size=1),
+01/02/2025 22:09:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:        model=ModelArgs(model_config=LlamaConfig(bos_token_id=0,
+01/02/2025 22:09:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 eos_token_id=0,
+01/02/2025 22:09:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 hidden_act='silu',
+01/02/2025 22:09:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 hidden_size=3072,
+01/02/2025 22:09:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 initializer_range=0.02,
+01/02/2025 22:09:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 intermediate_size=8192,
+01/02/2025 22:09:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 is_llama_config=True,
+01/02/2025 22:09:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 max_position_embeddings=4096,
+01/02/2025 22:09:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 num_attention_heads=32,
+01/02/2025 22:09:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 num_hidden_layers=28,
+01/02/2025 22:09:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 num_key_value_heads=32,
+01/02/2025 22:09:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 pad_token_id=None,
+01/02/2025 22:09:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 pretraining_tp=1,
+01/02/2025 22:09:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 rms_norm_eps=1e-05,
+01/02/2025 22:09:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 rope_scaling=None,
+01/02/2025 22:09:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 rope_theta=10000.0,
+01/02/2025 22:09:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 rope_interleaved=False,
+01/02/2025 22:09:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 tie_word_embeddings=True,
+01/02/2025 22:09:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 use_cache=True,
+01/02/2025 22:09:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 vocab_size=131072),
+01/02/2025 22:09:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                        init_method=RandomInit(std=0.02),
+01/02/2025 22:09:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                        dtype=torch.bfloat16,
+01/02/2025 22:09:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                        make_vocab_size_divisible_by=1,
+01/02/2025 22:09:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                        ddp_bucket_cap_mb=25),
+01/02/2025 22:09:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:        tokenizer=TokenizerArgs(tokenizer_name_or_path='robot-test/dummy-tokenizer-wordlevel',
+01/02/2025 22:09:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                tokenizer_revision=None,
+01/02/2025 22:09:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                tokenizer_max_length=None),
+01/02/2025 22:09:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:        checkpoints=CheckpointsArgs(checkpoints_path=PosixPath('checkpoints'),
+01/02/2025 22:09:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                    checkpoint_interval=10000,
+01/02/2025 22:09:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                    save_initial_state=False,
+01/02/2025 22:09:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                    save_final_state=False,
+01/02/2025 22:09:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                    resume_checkpoint_path=None,
+01/02/2025 22:09:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                    checkpoints_path_is_shared_file_system=False),
+01/02/2025 22:09:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:        logging=LoggingArgs(log_level='info',
+01/02/2025 22:09:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                            log_level_replica='info',
+01/02/2025 22:09:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                            iteration_step_info_interval=1),
+01/02/2025 22:09:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:        tokens=TokensArgs(sequence_length=4096,
+01/02/2025 22:09:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                          train_steps=100,
+01/02/2025 22:09:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                          micro_batch_size=1,
+01/02/2025 22:09:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                          batch_accumulation_per_replica=64,
+01/02/2025 22:09:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                          val_check_interval=100,
+01/02/2025 22:09:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                          limit_val_batches=0,
+01/02/2025 22:09:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                          limit_test_batches=0),
+01/02/2025 22:09:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:        optimizer=OptimizerArgs(optimizer_factory=AdamWOptimizerArgs(adam_eps=1e-08,
+01/02/2025 22:09:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                                     adam_beta1=0.9,
+01/02/2025 22:09:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                                     adam_beta2=0.95,
+01/02/2025 22:09:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                                     torch_adam_is_fused=True,
+01/02/2025 22:09:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                                     name='adamW'),
+01/02/2025 22:09:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                zero_stage=0,
+01/02/2025 22:09:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                weight_decay=0.01,
+01/02/2025 22:09:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                clip_grad=1.0,
+01/02/2025 22:09:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                accumulate_grad_in_fp32=True,
+01/02/2025 22:09:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                learning_rate_scheduler=LRSchedulerArgs(learning_rate=0.0003,
+01/02/2025 22:09:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                                        lr_warmup_steps=2,
+01/02/2025 22:09:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                                        lr_warmup_style='linear',
+01/02/2025 22:09:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                                        lr_decay_style='cosine',
+01/02/2025 22:09:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                                        lr_decay_steps=13,
+01/02/2025 22:09:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                                        lr_decay_starting_step=None,
+01/02/2025 22:09:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                                        min_decay_lr=1e-05)),
+01/02/2025 22:09:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:        data_stages=[DatasetStageArgs(name='Stable Training Stage',
+01/02/2025 22:09:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                      start_training_step=1,
+01/02/2025 22:09:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                      data=DataArgs(dataset=None,
+01/02/2025 22:09:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                    seed=42,
+01/02/2025 22:09:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                    num_loading_workers=1))],
+01/02/2025 22:09:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:        profiler=None,
+01/02/2025 22:09:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:        lighteval=None,
+01/02/2025 22:09:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:        s3_upload=None)
+01/02/2025 22:09:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: Model Config:
+01/02/2025 22:09:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: LlamaConfig(bos_token_id=0,
+01/02/2025 22:09:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             eos_token_id=0,
+01/02/2025 22:09:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             hidden_act='silu',
+01/02/2025 22:09:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             hidden_size=3072,
+01/02/2025 22:09:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             initializer_range=0.02,
+01/02/2025 22:09:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             intermediate_size=8192,
+01/02/2025 22:09:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             is_llama_config=True,
+01/02/2025 22:09:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             max_position_embeddings=4096,
+01/02/2025 22:09:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             num_attention_heads=32,
+01/02/2025 22:09:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             num_hidden_layers=28,
+01/02/2025 22:09:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             num_key_value_heads=32,
+01/02/2025 22:09:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             pad_token_id=None,
+01/02/2025 22:09:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             pretraining_tp=1,
+01/02/2025 22:09:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             rms_norm_eps=1e-05,
+01/02/2025 22:09:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             rope_scaling=None,
+01/02/2025 22:09:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             rope_theta=10000.0,
+01/02/2025 22:09:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             rope_interleaved=False,
+01/02/2025 22:09:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             tie_word_embeddings=True,
+01/02/2025 22:09:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             use_cache=True,
+01/02/2025 22:09:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             vocab_size=131072)
+01/02/2025 22:09:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: Building model..
+01/02/2025 22:09:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: Initialize RoPE Theta = 10000.0
+01/02/2025 22:09:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: Setting PP block ranks...
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+01/02/2025 22:10:00 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: Total number of parameters: 3.98G (7584.67MiB)
+01/02/2025 22:10:00 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: Local number of parameters: 654M (1248.09MiB)
+01/02/2025 22:10:00 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [After model building] Memory usage: 1248.11MiB. Peak allocated: 5440.00MiB Peak reserved: 14658.00MiB
+01/02/2025 22:10:00 [INFO|DP=0|PP=2|TP=0|ip-26-0-175-170]: Local number of parameters: 453M (864.09MiB)
+01/02/2025 22:10:00 [INFO|DP=0|PP=0|TP=1|ip-26-0-160-225]: Local number of parameters: 654M (1248.09MiB)
+01/02/2025 22:10:00 [INFO|DP=0|PP=2|TP=0|ip-26-0-175-170]: [After model building] Memory usage: 864.11MiB. Peak allocated: 5440.00MiB Peak reserved: 14658.00MiB
+01/02/2025 22:10:00 [INFO|DP=0|PP=0|TP=1|ip-26-0-160-225]: [After model building] Memory usage: 1248.11MiB. Peak allocated: 5440.00MiB Peak reserved: 14658.00MiB
+01/02/2025 22:10:00 [INFO|DP=0|PP=3|TP=0|ip-26-0-175-241]: Local number of parameters: 428M (816.05MiB)
+01/02/2025 22:10:00 [INFO|DP=0|PP=1|TP=0|ip-26-0-161-138]: Local number of parameters: 453M (864.09MiB)
+01/02/2025 22:10:00 [INFO|DP=0|PP=3|TP=1|ip-26-0-175-241]: Local number of parameters: 428M (816.05MiB)
+01/02/2025 22:10:00 [INFO|DP=0|PP=2|TP=1|ip-26-0-175-170]: Local number of parameters: 453M (864.09MiB)
+01/02/2025 22:10:00 [INFO|DP=0|PP=1|TP=1|ip-26-0-161-138]: Local number of parameters: 453M (864.09MiB)
+01/02/2025 22:10:00 [INFO|DP=0|PP=1|TP=0|ip-26-0-161-138]: [After model building] Memory usage: 864.11MiB. Peak allocated: 5440.00MiB Peak reserved: 14658.00MiB
+01/02/2025 22:10:00 [INFO|DP=0|PP=3|TP=0|ip-26-0-175-241]: [After model building] Memory usage: 816.07MiB. Peak allocated: 5440.00MiB Peak reserved: 14658.00MiB
+01/02/2025 22:10:00 [INFO|DP=0|PP=3|TP=1|ip-26-0-175-241]: [After model building] Memory usage: 816.07MiB. Peak allocated: 5440.00MiB Peak reserved: 14658.00MiB
+01/02/2025 22:10:00 [INFO|DP=0|PP=2|TP=1|ip-26-0-175-170]: [After model building] Memory usage: 864.11MiB. Peak allocated: 5440.00MiB Peak reserved: 14658.00MiB
+01/02/2025 22:10:00 [INFO|DP=0|PP=1|TP=1|ip-26-0-161-138]: [After model building] Memory usage: 864.11MiB. Peak allocated: 5440.00MiB Peak reserved: 14658.00MiB
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+01/02/2025 22:10:04 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: No checkpoint path provided.
+01/02/2025 22:10:04 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: Parametrizing model parameters using StandardParametrizator
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+01/02/2025 22:10:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [Optimizer Building] Using LearningRateForSP as learning rate
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+01/02/2025 22:10:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.token_position_embeddings | PP: 0/4 | Block rank: 0
+01/02/2025 22:10:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.0 | PP: 0/4 | Block rank: 1
+01/02/2025 22:10:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.1 | PP: 0/4 | Block rank: 2
+01/02/2025 22:10:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.2 | PP: 0/4 | Block rank: 3
+01/02/2025 22:10:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.3 | PP: 0/4 | Block rank: 4
+01/02/2025 22:10:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.4 | PP: 0/4 | Block rank: 5
+01/02/2025 22:10:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.5 | PP: 0/4 | Block rank: 6
+01/02/2025 22:10:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.6 | PP: 0/4 | Block rank: 7
+01/02/2025 22:10:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.7 | PP: 0/4 | Block rank: 8
+01/02/2025 22:10:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.8 | PP: 1/4 | Block rank: 0
+01/02/2025 22:10:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.9 | PP: 1/4 | Block rank: 1
+01/02/2025 22:10:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.10 | PP: 1/4 | Block rank: 2
+01/02/2025 22:10:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.11 | PP: 1/4 | Block rank: 3
+01/02/2025 22:10:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.12 | PP: 1/4 | Block rank: 4
+01/02/2025 22:10:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.13 | PP: 1/4 | Block rank: 5
+01/02/2025 22:10:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.14 | PP: 1/4 | Block rank: 6
+01/02/2025 22:10:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.15 | PP: 1/4 | Block rank: 7
+01/02/2025 22:10:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.16 | PP: 2/4 | Block rank: 0
+01/02/2025 22:10:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.17 | PP: 2/4 | Block rank: 1
+01/02/2025 22:10:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.18 | PP: 2/4 | Block rank: 2
+01/02/2025 22:10:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.19 | PP: 2/4 | Block rank: 3
+01/02/2025 22:10:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.20 | PP: 2/4 | Block rank: 4
+01/02/2025 22:10:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.21 | PP: 2/4 | Block rank: 5
+01/02/2025 22:10:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.22 | PP: 2/4 | Block rank: 6
+01/02/2025 22:10:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.23 | PP: 2/4 | Block rank: 7
+01/02/2025 22:10:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.24 | PP: 3/4 | Block rank: 0
+01/02/2025 22:10:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.25 | PP: 3/4 | Block rank: 1
+01/02/2025 22:10:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.26 | PP: 3/4 | Block rank: 2
+01/02/2025 22:10:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.27 | PP: 3/4 | Block rank: 3
+01/02/2025 22:10:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.final_layer_norm | PP: 3/4 | Block rank: 4
+01/02/2025 22:10:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.lm_head | PP: 3/4 | Block rank: 5
+01/02/2025 22:10:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.cast_to_fp32 | PP: 3/4 | Block rank: 6
+01/02/2025 22:10:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: loss | PP: 3/4 | Block rank: 7
+01/02/2025 22:10:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [Training Plan] Stage Stable Training Stage has 99 remaining training steps and has consumed 0 samples
+01/02/2025 22:10:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: Using dummy data generator
+01/02/2025 22:10:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [Training Plan] There are 1 training stages 
+01/02/2025 22:10:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [Stage Stable Training Stage] start from step 1 
+01/02/2025 22:10:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: 
+01/02/2025 22:10:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [Start training] datetime: 2025-01-02 22:10:06.266109 | mbs: 1 | grad_accum: 64 | global_batch_size: 256 | sequence_length: 4096 | train_steps: 100 | start_iteration_step: 0 | consumed_train_samples: 0
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+01/02/2025 22:10:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: Resuming training from stage Stable Training Stage, it has trained for 0 samples and has 99 remaining train steps
+01/02/2025 22:10:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:  Memory usage: 7488.58MiB. Peak allocated 7488.58MiB. Peak reserved: 20904.00MiB
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+Traceback (most recent call last):
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    return f(*args, **kwargs)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+        trainer.train(dataloader)outputs = self.pipeline_engine.train_batch_iter(
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    model.p2p.clear_history()
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)AttributeError
+: 'DistributedDataParallel' object has no attribute 'p2p'  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+Traceback (most recent call last):
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+[2025-01-02 22:10:25,284] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1868065 closing signal SIGTERM
+[2025-01-02 22:10:25,284] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1868066 closing signal SIGTERM
+[2025-01-02 22:10:25,284] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1868067 closing signal SIGTERM
+[2025-01-02 22:10:25,284] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1868068 closing signal SIGTERM
+[2025-01-02 22:10:25,284] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1868070 closing signal SIGTERM
+[2025-01-02 22:10:25,284] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1868071 closing signal SIGTERM
+[2025-01-02 22:10:25,284] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1868072 closing signal SIGTERM
+[2025-01-02 22:10:25,294] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 30992 closing signal SIGTERM
+[2025-01-02 22:10:25,294] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 30993 closing signal SIGTERM
+[2025-01-02 22:10:25,294] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 30994 closing signal SIGTERM
+[2025-01-02 22:10:25,294] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 30995 closing signal SIGTERM
+[2025-01-02 22:10:25,294] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 30997 closing signal SIGTERM
+[2025-01-02 22:10:25,294] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 30998 closing signal SIGTERM
+[2025-01-02 22:10:25,295] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 30999 closing signal SIGTERM
+[2025-01-02 22:10:25,297] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 40939 closing signal SIGTERM
+[2025-01-02 22:10:25,297] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 40940 closing signal SIGTERM
+[2025-01-02 22:10:25,297] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 40941 closing signal SIGTERM
+[2025-01-02 22:10:25,297] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 40942 closing signal SIGTERM
+[2025-01-02 22:10:25,297] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 40943 closing signal SIGTERM
+[2025-01-02 22:10:25,297] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 40944 closing signal SIGTERM
+[2025-01-02 22:10:25,297] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 40945 closing signal SIGTERM
+[2025-01-02 22:10:28,519] torch.distributed.elastic.multiprocessing.api: [ERROR] failed (exitcode: 1) local_rank: 7 (pid: 40946) of binary: /fsx/nouamane/miniconda/envs/2-1-cu121/bin/python
+[2025-01-02 22:10:28,553] torch.distributed.elastic.multiprocessing.errors.error_handler: [ERROR] no error file defined for parent, to copy child error file (/tmp/torchelastic_6kbdver4/14019012_a4dszyax/attempt_0/7/error.json)
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 264, in launch_agent
+    raise ChildFailedError(
+torch.distributed.elastic.multiprocessing.errors.ChildFailedError: 
+============================================================
+/fsx/nouamane/projects/nanotron/run_train.py FAILED
+------------------------------------------------------------
+Failures:
+  <NO_OTHER_FAILURES>
+------------------------------------------------------------
+Root Cause (first observed failure):
+[0]:
+  time      : 2025-01-02_22:10:06
+  host      : ip-26-0-175-241.ec2.internal
+  rank      : 31 (local_rank: 7)
+  exitcode  : 1 (pid: 40946)
+  error_file: /tmp/torchelastic_6kbdver4/14019012_a4dszyax/attempt_0/7/error.json
+  traceback : Traceback (most recent call last):
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+      return f(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+      outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+      outputs = self.pipeline_engine.train_batch_iter(
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+      model.p2p.clear_history()
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+      raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+  AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+  
+============================================================
+srun: error: ip-26-0-175-241: task 3: Exited with exit code 1
+srun: launch/slurm: _step_signal: Terminating StepId=14019012.0
+slurmstepd: error: *** STEP 14019012.0 ON ip-26-0-160-225 CANCELLED AT 2025-01-02T22:10:28 ***
+[2025-01-02 22:10:28,906] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-02 22:10:28,906] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1868066 closing signal SIGTERM
+[2025-01-02 22:10:28,906] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1868067 closing signal SIGTERM
+[2025-01-02 22:10:28,906] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1868070 closing signal SIGTERM
+[2025-01-02 22:10:28,906] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1868072 closing signal SIGTERM
+[2025-01-02 22:10:28,908] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-02 22:10:28,908] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 284925 closing signal SIGTERM
+[2025-01-02 22:10:28,908] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-02 22:10:28,908] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 30999 closing signal SIGTERM
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 284846 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 878, in _invoke_run
+    run_result = self._monitor_workers(self._worker_group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/local_elastic_agent.py", line 307, in _monitor_workers
+    result = self._pcontext.wait(0)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 288, in wait
+    return self._poll()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 668, in _poll
+    self.close()  # terminate all running procs
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 331, in close
+    self._close(death_sig=death_sig, timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 713, in _close
+    handler.proc.wait(time_to_wait)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1209, in wait
+    return self._wait(timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1953, in _wait
+    time.sleep(delay)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 30920 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 878, in _invoke_run
+    run_result = self._monitor_workers(self._worker_group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/local_elastic_agent.py", line 307, in _monitor_workers
+    result = self._pcontext.wait(0)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 288, in wait
+    return self._poll()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 668, in _poll
+    self.close()  # terminate all running procs
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 331, in close
+    self._close(death_sig=death_sig, timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 713, in _close
+    handler.proc.wait(time_to_wait)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1209, in wait
+    return self._wait(timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1953, in _wait
+    time.sleep(delay)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 1867992 got signal: 15
+srun: error: ip-26-0-160-225: task 0: Exited with exit code 1
+srun: error: ip-26-0-175-170: task 2: Exited with exit code 1
+srun: error: ip-26-0-161-138: task 1: Exited with exit code 1
diff --git a/logs/14019076-bench_8.86G_dp1_tp8_pp16_acc256_mbs1_seq4096_zero0_tpmodeRED_vocab131k.out b/logs/14019076-bench_8.86G_dp1_tp8_pp16_acc256_mbs1_seq4096_zero0_tpmodeRED_vocab131k.out
new file mode 100644
index 0000000000000000000000000000000000000000..e78ec469c56c582de0849c56e084cf69d64b120f
--- /dev/null
+++ b/logs/14019076-bench_8.86G_dp1_tp8_pp16_acc256_mbs1_seq4096_zero0_tpmodeRED_vocab131k.out
@@ -0,0 +1,2005 @@
++ '[' -z 14019076 ']'
++ source /etc/profile.d/modules.sh
+++ . /usr/share/modules/init/bash
++++ unset _mlshdbg
++++ '[' 0 = 1 ']'
++++ unset _mlre _mlIFS
++++ '[' -n x ']'
++++ _mlIFS=' 	
+'
++++ IFS=' '
++++ '[' -n '' ']'
+++++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash autoinit
++++ _mlcode='module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+MODULEPATH_modshare=/etc/environment-modules/modules:1:/usr/share/modules/\$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1; export MODULEPATH_modshare;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+LOADEDMODULES=; export LOADEDMODULES;
+MODULEPATH=/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/\$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles; export MODULEPATH;
+test 0;'
++++ _mlret=0
++++ '[' -n x ']'
++++ IFS=' 	
+'
++++ unset _mlIFS
++++ unset _mlre _mlv _mlrv
++++ '[' 0 -eq 0 ']'
++++ eval 'module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+MODULEPATH_modshare=/etc/environment-modules/modules:1:/usr/share/modules/\$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1; export MODULEPATH_modshare;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+LOADEDMODULES=; export LOADEDMODULES;
+MODULEPATH=/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/\$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles; export MODULEPATH;
+test 0;'
+++++ MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl
+++++ export MODULES_CMD
+++++ ENV=/usr/share/modules/init/profile.sh
+++++ export ENV
+++++ MODULEPATH_modshare='/etc/environment-modules/modules:1:/usr/share/modules/$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1'
+++++ export MODULEPATH_modshare
+++++ BASH_ENV=/usr/share/modules/init/bash
+++++ export BASH_ENV
+++++ MODULESHOME=/usr/share/modules
+++++ export MODULESHOME
+++++ LOADEDMODULES=
+++++ export LOADEDMODULES
+++++ MODULEPATH='/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles'
+++++ export MODULEPATH
+++++ test 0
++++ '[' 0 = 1 ']'
++++ '[' -t 2 ']'
++++ export -f module
++++ export -f switchml
++++ '[' 5 -ge 3 ']'
++++ [[ ehxB =~ i ]]
++++ [[ ! :/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin: =~ :/usr/bin: ]]
++++ '[' '!' -n '' ']'
++++ MANPATH=:
++++ export MANPATH
+++++ manpath
++++ [[ ! :/fsx/nouamane/miniconda/envs/2-1-cu121/man:/fsx/nouamane/miniconda/envs/2-1-cu121/share/man:/fsx/nouamane/miniconda/man:/fsx/nouamane/miniconda/share/man:/usr/local/man:/usr/local/share/man:/usr/share/man:/opt/slurm/share/man:/admin/home/nouamane/.local/share/man:/admin/home/nouamane/.fzf/man:: =~ :/usr/share/man: ]]
++++ unset _mlcode _mlret
++++ '[' -n '' ']'
++ module load cuda/12.1
++ unset _mlshdbg
++ '[' 0 = 1 ']'
++ unset _mlre _mlIFS
++ '[' -n x ']'
++ _mlIFS=' 	
+'
++ IFS=' '
++ '[' -n '' ']'
+++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash load cuda/12.1
++ eval 'CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include;' export 'CPATH;
+LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:;' export 'LD_LIBRARY_PATH;
+FI_EFA_FORK_SAFE=1;' export 'FI_EFA_FORK_SAFE;
+MANPATH=/usr/local/cuda-12.1/share/man::;' export 'MANPATH;
+LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64;' export 'LIBRARY_PATH;
+_LMFILES_=/usr/share/modules/modulefiles/cuda/12.1;' export '_LMFILES_;
+LOADEDMODULES=cuda/12.1;' export 'LOADEDMODULES;
+MPI_PATH=/opt/amazon/openmpi;' export 'MPI_PATH;
+NCCL_HOME_modshare=/opt/nccl/build:1;' export 'NCCL_HOME_modshare;
+NCCL_PROTO=simple;' export 'NCCL_PROTO;
+MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1;' export 'MANPATH_modshare;
+LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1;' export 'LIBRARY_PATH_modshare;
+NCCL_SOCKET_IFNAME=enp;' export 'NCCL_SOCKET_IFNAME;
+AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl;' export 'AWS_OFI_NCCL_HOME;
+NCCL_HOME=/opt/nccl/build;' export 'NCCL_HOME;
+FI_PROVIDER=efa;' export 'FI_PROVIDER;
+AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1;' export 'AWS_OFI_NCCL_HOME_modshare;
+CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1;' export 'CPATH_modshare;
+LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2::1:/opt/nccl/build/lib:1:/opt/aws-ofi-nccl/lib:2:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/efa/lib:1;' export 'LD_LIBRARY_PATH_modshare;
+FI_EFA_ENABLE_SHM_TRANSFER=1;' export 'FI_EFA_ENABLE_SHM_TRANSFER;
+_LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1;' export '_LMFILES__modshare;
+LOADEDMODULES_modshare=cuda/12.1:1;' export 'LOADEDMODULES_modshare;
+MPI_PATH_modshare=/opt/amazon/openmpi:1;' export 'MPI_PATH_modshare;
+PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin;' export 'PATH;
+CUDA_HOME=/usr/local/cuda-12.1;' export 'CUDA_HOME;
+PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/opt/slurm/bin:1:/usr/sbin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1;' export 'PATH_modshare;
+test' '0;'
+++ CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include
+++ export CPATH
+++ LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:
+++ export LD_LIBRARY_PATH
+++ FI_EFA_FORK_SAFE=1
+++ export FI_EFA_FORK_SAFE
+++ MANPATH=/usr/local/cuda-12.1/share/man::
+++ export MANPATH
+++ LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64
+++ export LIBRARY_PATH
+++ _LMFILES_=/usr/share/modules/modulefiles/cuda/12.1
+++ export _LMFILES_
+++ LOADEDMODULES=cuda/12.1
+++ export LOADEDMODULES
+++ MPI_PATH=/opt/amazon/openmpi
+++ export MPI_PATH
+++ NCCL_HOME_modshare=/opt/nccl/build:1
+++ export NCCL_HOME_modshare
+++ NCCL_PROTO=simple
+++ export NCCL_PROTO
+++ MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1
+++ export MANPATH_modshare
+++ LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1
+++ export LIBRARY_PATH_modshare
+++ NCCL_SOCKET_IFNAME=enp
+++ export NCCL_SOCKET_IFNAME
+++ AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl
+++ export AWS_OFI_NCCL_HOME
+++ NCCL_HOME=/opt/nccl/build
+++ export NCCL_HOME
+++ FI_PROVIDER=efa
+++ export FI_PROVIDER
+++ AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1
+++ export AWS_OFI_NCCL_HOME_modshare
+++ CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1
+++ export CPATH_modshare
+++ LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2::1:/opt/nccl/build/lib:1:/opt/aws-ofi-nccl/lib:2:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/efa/lib:1
+++ export LD_LIBRARY_PATH_modshare
+++ FI_EFA_ENABLE_SHM_TRANSFER=1
+++ export FI_EFA_ENABLE_SHM_TRANSFER
+++ _LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1
+++ export _LMFILES__modshare
+++ LOADEDMODULES_modshare=cuda/12.1:1
+++ export LOADEDMODULES_modshare
+++ MPI_PATH_modshare=/opt/amazon/openmpi:1
+++ export MPI_PATH_modshare
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
+++ export PATH
+++ CUDA_HOME=/usr/local/cuda-12.1
+++ export CUDA_HOME
+++ PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/opt/slurm/bin:1:/usr/sbin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1
+++ export PATH_modshare
+++ test 0
++ _mlstatus=0
++ '[' -n x ']'
++ IFS=' 	
+'
++ unset _mlre _mlv _mlrv _mlIFS
++ '[' -n '' ']'
++ unset _mlshdbg
++ return 0
++ source /fsx/nouamane/miniconda/bin/activate
+++ _CONDA_ROOT=/fsx/nouamane/miniconda
+++ . /fsx/nouamane/miniconda/etc/profile.d/conda.sh
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ '[' -z x ']'
+++ conda activate
+++ local cmd=activate
+++ case "$cmd" in
+++ __conda_activate activate
+++ '[' -n '' ']'
+++ local ask_conda
++++ PS1=
++++ __conda_exe shell.posix activate
++++ /fsx/nouamane/miniconda/bin/conda shell.posix activate
+Error while loading conda entry point: conda-libmamba-solver (module 'libmambapy' has no attribute 'QueryFormat')
+++ ask_conda='. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''2'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_1='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/etc/conda/activate.d/libxml2_activate.sh"'
+++ eval '. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''2'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_1='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/etc/conda/activate.d/libxml2_activate.sh"'
++++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh
+++++ test -n ''
+++++ unset XML_CATALOG_FILES
+++++ unset xml_catalog_files_libxml2
++++ PS1='(base) '
++++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
++++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
++++ export CONDA_PREFIX=/fsx/nouamane/miniconda
++++ CONDA_PREFIX=/fsx/nouamane/miniconda
++++ export CONDA_SHLVL=2
++++ CONDA_SHLVL=2
++++ export CONDA_DEFAULT_ENV=base
++++ CONDA_DEFAULT_ENV=base
++++ export 'CONDA_PROMPT_MODIFIER=(base) '
++++ CONDA_PROMPT_MODIFIER='(base) '
++++ export CONDA_PREFIX_1=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ CONDA_PREFIX_1=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ . /fsx/nouamane/miniconda/etc/conda/activate.d/libxml2_activate.sh
+++++ test -n ''
+++++ xml_catalog_files_libxml2=
+++++ XML_CATALOG_FILES=
+++++ conda_catalog_files=
+++++ ifs_libxml2=' 	
+'
+++++ IFS=' '
+++++ rem=/fsx/nouamane/miniconda
+++++ for pre in ${rem}
+++++ test '' = /fsx/nouamane/miniconda
+++++ conda_catalog_files=/fsx/nouamane/miniconda
+++++ rem=
+++++ IFS=' 	
+'
+++++ conda_catalog_files='file:///fsx/nouamane/miniconda/etc/xml/catalog file:///etc/xml/catalog'
+++++ export 'XML_CATALOG_FILES=file:///fsx/nouamane/miniconda/etc/xml/catalog file:///etc/xml/catalog'
+++++ XML_CATALOG_FILES='file:///fsx/nouamane/miniconda/etc/xml/catalog file:///etc/xml/catalog'
+++++ unset conda_catalog_files ifs_libxml2 rem
+++ __conda_hashr
+++ '[' -n '' ']'
+++ '[' -n '' ']'
+++ hash -r
++ conda activate 2-1-cu121
++ local cmd=activate
++ case "$cmd" in
++ __conda_activate activate 2-1-cu121
++ '[' -n '' ']'
++ local ask_conda
+++ PS1='(base) '
+++ __conda_exe shell.posix activate 2-1-cu121
+++ /fsx/nouamane/miniconda/bin/conda shell.posix activate 2-1-cu121
+Error while loading conda entry point: conda-libmamba-solver (module 'libmambapy' has no attribute 'QueryFormat')
++ ask_conda='. "/fsx/nouamane/miniconda/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_2='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
++ eval '. "/fsx/nouamane/miniconda/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_2='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
+++ . /fsx/nouamane/miniconda/etc/conda/deactivate.d/libxml2_deactivate.sh
++++ test -n ''
++++ unset XML_CATALOG_FILES
++++ unset xml_catalog_files_libxml2
+++ PS1='(2-1-cu121) '
+++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
+++ export CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ export CONDA_SHLVL=3
+++ CONDA_SHLVL=3
+++ export CONDA_DEFAULT_ENV=2-1-cu121
+++ CONDA_DEFAULT_ENV=2-1-cu121
+++ export 'CONDA_PROMPT_MODIFIER=(2-1-cu121) '
+++ CONDA_PROMPT_MODIFIER='(2-1-cu121) '
+++ export CONDA_PREFIX_2=/fsx/nouamane/miniconda
+++ CONDA_PREFIX_2=/fsx/nouamane/miniconda
+++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ export _CE_M=
+++ _CE_M=
+++ export _CE_CONDA=
+++ _CE_CONDA=
+++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh
++++ test -n ''
++++ xml_catalog_files_libxml2=
++++ XML_CATALOG_FILES=
++++ conda_catalog_files=
++++ ifs_libxml2=' 	
+'
++++ IFS=' '
++++ rem=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ for pre in ${rem}
++++ test '' = /fsx/nouamane/miniconda/envs/2-1-cu121
++++ conda_catalog_files=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ rem=
++++ IFS=' 	
+'
++++ conda_catalog_files='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ export 'XML_CATALOG_FILES=file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ XML_CATALOG_FILES='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ unset conda_catalog_files ifs_libxml2 rem
++ __conda_hashr
++ '[' -n '' ']'
++ '[' -n '' ']'
++ hash -r
++ export PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
++ PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
++ '[' -z '' ']'
+++ scontrol show hostnames 'ip-26-0-165-213,ip-26-0-166-[15,36,125,214,244],ip-26-0-167-[9,51,175,177,217,245],ip-26-0-168-[30,34,52,95]'
++ export 'NODELIST=ip-26-0-165-213
+ip-26-0-166-15
+ip-26-0-166-36
+ip-26-0-166-125
+ip-26-0-166-214
+ip-26-0-166-244
+ip-26-0-167-9
+ip-26-0-167-51
+ip-26-0-167-175
+ip-26-0-167-177
+ip-26-0-167-217
+ip-26-0-167-245
+ip-26-0-168-30
+ip-26-0-168-34
+ip-26-0-168-52
+ip-26-0-168-95'
++ NODELIST='ip-26-0-165-213
+ip-26-0-166-15
+ip-26-0-166-36
+ip-26-0-166-125
+ip-26-0-166-214
+ip-26-0-166-244
+ip-26-0-167-9
+ip-26-0-167-51
+ip-26-0-167-175
+ip-26-0-167-177
+ip-26-0-167-217
+ip-26-0-167-245
+ip-26-0-168-30
+ip-26-0-168-34
+ip-26-0-168-52
+ip-26-0-168-95'
+++ head -n1
+++ scontrol show hostnames 'ip-26-0-165-213,ip-26-0-166-[15,36,125,214,244],ip-26-0-167-[9,51,175,177,217,245],ip-26-0-168-[30,34,52,95]'
++ export MASTER_NODE=ip-26-0-165-213
++ MASTER_NODE=ip-26-0-165-213
++ export MASTER_PORT=12356
++ MASTER_PORT=12356
++ export NNODES=16
++ NNODES=16
++ export GPUS_PER_NODE=8
++ GPUS_PER_NODE=8
++ export WORLD_SIZE=128
++ WORLD_SIZE=128
++ export CUDA_DEVICE_MAX_CONNECTIONS=1
++ CUDA_DEVICE_MAX_CONNECTIONS=1
++ export NCCL_DEBUG=WARN
++ NCCL_DEBUG=WARN
++ export NANOTRON_BENCHMARK=1
++ NANOTRON_BENCHMARK=1
++ export WANDB_MODE=disabled
++ WANDB_MODE=disabled
++ export TORCH_NCCL_ASYNC_ERROR_HANDLING=1
++ TORCH_NCCL_ASYNC_ERROR_HANDLING=1
++ export TORCH_DISTRIBUTED_DEBUG=DETAIL
++ TORCH_DISTRIBUTED_DEBUG=DETAIL
++ '[' -z '' ']'
++ echo '=== GPU Topology ==='
+=== GPU Topology ===
++ nvidia-smi topo -m
+	GPU0	GPU1	GPU2	GPU3	GPU4	GPU5	GPU6	GPU7	CPU Affinity	NUMA Affinity	GPU NUMA ID
+GPU0	 X 	NV18	NV18	NV18	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU1	NV18	 X 	NV18	NV18	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU2	NV18	NV18	 X 	NV18	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU3	NV18	NV18	NV18	 X 	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU4	NV18	NV18	NV18	NV18	 X 	NV18	NV18	NV18	48-95	1		N/A
+GPU5	NV18	NV18	NV18	NV18	NV18	 X 	NV18	NV18	48-95	1		N/A
+GPU6	NV18	NV18	NV18	NV18	NV18	NV18	 X 	NV18	48-95	1		N/A
+GPU7	NV18	NV18	NV18	NV18	NV18	NV18	NV18	 X 	48-95	1		N/A
+
+Legend:
+
+  X    = Self
+  SYS  = Connection traversing PCIe as well as the SMP interconnect between NUMA nodes (e.g., QPI/UPI)
+  NODE = Connection traversing PCIe as well as the interconnect between PCIe Host Bridges within a NUMA node
+  PHB  = Connection traversing PCIe as well as a PCIe Host Bridge (typically the CPU)
+  PXB  = Connection traversing multiple PCIe bridges (without traversing the PCIe Host Bridge)
+  PIX  = Connection traversing at most a single PCIe bridge
+  NV#  = Connection traversing a bonded set of # NVLinks
++ echo ==================
+==================
++ export SRUN_ALLOC_ARGS=
++ SRUN_ALLOC_ARGS=
++ echo 'Master node: ip-26-0-165-213'
+Master node: ip-26-0-165-213
++ echo 'All nodes: ip-26-0-165-213
+ip-26-0-166-15
+ip-26-0-166-36
+ip-26-0-166-125
+ip-26-0-166-214
+ip-26-0-166-244
+ip-26-0-167-9
+ip-26-0-167-51
+ip-26-0-167-175
+ip-26-0-167-177
+ip-26-0-167-217
+ip-26-0-167-245
+ip-26-0-168-30
+ip-26-0-168-34
+ip-26-0-168-52
+ip-26-0-168-95'
+All nodes: ip-26-0-165-213
+ip-26-0-166-15
+ip-26-0-166-36
+ip-26-0-166-125
+ip-26-0-166-214
+ip-26-0-166-244
+ip-26-0-167-9
+ip-26-0-167-51
+ip-26-0-167-175
+ip-26-0-167-177
+ip-26-0-167-217
+ip-26-0-167-245
+ip-26-0-168-30
+ip-26-0-168-34
+ip-26-0-168-52
+ip-26-0-168-95
++ echo 'World size: 128'
+World size: 128
++ '[' -n '' ']'
++ srun --wait=0 --kill-on-bad-exit=1 torchrun --nnodes=16 --nproc_per_node=8 --rdzv_id=14019076 --rdzv_backend=c10d --rdzv_endpoint=ip-26-0-165-213:12356 --max_restarts 0 --rdzv_conf timeout=60 /fsx/nouamane/projects/nanotron/run_train.py --config-file benchmark/configs/config_8.86G_dp1_tp8_pp16_acc256_mbs1_seq4096_zero0_tpmodeRED_vocab131k.yaml
+[2025-01-03 02:37:30,778] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 02:37:30,777] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 02:37:30,778] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 02:37:30,777] torch.distributed.run: [WARNING] 
+[2025-01-03 02:37:30,777] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 02:37:30,777] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 02:37:30,777] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 02:37:30,778] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 02:37:30,778] torch.distributed.run: [WARNING] 
+[2025-01-03 02:37:30,778] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 02:37:30,778] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 02:37:30,778] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 02:37:30,778] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 02:37:30,778] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 02:37:30,778] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 02:37:30,782] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 02:37:30,784] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 02:37:30,785] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 02:37:30,786] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 02:37:30,786] torch.distributed.run: [WARNING] 
+[2025-01-03 02:37:30,786] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 02:37:30,786] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 02:37:30,786] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 02:37:30,792] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 02:37:30,796] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 02:37:30,806] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 02:37:30,778] torch.distributed.run: [WARNING] 
+[2025-01-03 02:37:30,778] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 02:37:30,778] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 02:37:30,778] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 02:37:30,778] torch.distributed.run: [WARNING] 
+[2025-01-03 02:37:30,778] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 02:37:30,778] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 02:37:30,778] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 02:37:30,778] torch.distributed.run: [WARNING] 
+[2025-01-03 02:37:30,778] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 02:37:30,778] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 02:37:30,778] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 02:37:30,778] torch.distributed.run: [WARNING] 
+[2025-01-03 02:37:30,778] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 02:37:30,778] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 02:37:30,778] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 02:37:30,778] torch.distributed.run: [WARNING] 
+[2025-01-03 02:37:30,778] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 02:37:30,778] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 02:37:30,778] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 02:37:30,782] torch.distributed.run: [WARNING] 
+[2025-01-03 02:37:30,782] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 02:37:30,782] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 02:37:30,782] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 02:37:30,785] torch.distributed.run: [WARNING] 
+[2025-01-03 02:37:30,785] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 02:37:30,785] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 02:37:30,785] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 02:37:30,785] torch.distributed.run: [WARNING] 
+[2025-01-03 02:37:30,785] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 02:37:30,785] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 02:37:30,785] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 02:37:30,792] torch.distributed.run: [WARNING] 
+[2025-01-03 02:37:30,792] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 02:37:30,792] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 02:37:30,792] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 02:37:30,796] torch.distributed.run: [WARNING] 
+[2025-01-03 02:37:30,796] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 02:37:30,796] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 02:37:30,796] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 02:37:30,845] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 02:37:30,806] torch.distributed.run: [WARNING] 
+[2025-01-03 02:37:30,806] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 02:37:30,806] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 02:37:30,806] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 02:37:30,845] torch.distributed.run: [WARNING] 
+[2025-01-03 02:37:30,845] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 02:37:30,845] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 02:37:30,845] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 02:37:31,059] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 02:37:31,059] torch.distributed.run: [WARNING] 
+[2025-01-03 02:37:31,059] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 02:37:31,059] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 02:37:31,059] torch.distributed.run: [WARNING] *****************************************
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+01/03/2025 02:38:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]: Config:
+01/03/2025 02:38:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]: Config(general=GeneralArgs(project='debug',
+01/03/2025 02:38:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]:                            run='8.86G_dp1_tp8_pp16_acc256_mbs1_seq4096_zero0_tpmodeRED_vocab131k',
+01/03/2025 02:38:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]:                            seed=42,
+01/03/2025 02:38:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]:                            step=None,
+01/03/2025 02:38:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]:                            consumed_train_samples=None,
+01/03/2025 02:38:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]:                            benchmark_csv_path=PosixPath('benchmark/results/bench_final2.csv'),
+01/03/2025 02:38:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]:                            ignore_sanity_checks=True),
+01/03/2025 02:38:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]:        parallelism=ParallelismArgs(dp=1,
+01/03/2025 02:38:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]:                                    pp=16,
+01/03/2025 02:38:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]:                                    tp=8,
+01/03/2025 02:38:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]:                                    pp_engine=<nanotron.parallel.pipeline_parallel.engine.OneForwardOneBackwardPipelineEngine object at 0x7f3f24574dc0>,
+01/03/2025 02:38:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]:                                    tp_mode=<TensorParallelLinearMode.REDUCE_SCATTER: 2>,
+01/03/2025 02:38:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]:                                    tp_linear_async_communication=True,
+01/03/2025 02:38:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]:                                    recompute_layer=False,
+01/03/2025 02:38:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]:                                    tp_recompute_allgather=True,
+01/03/2025 02:38:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]:                                    expert_parallel_size=1),
+01/03/2025 02:38:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]:        model=ModelArgs(model_config=LlamaConfig(bos_token_id=0,
+01/03/2025 02:38:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]:                                                 eos_token_id=0,
+01/03/2025 02:38:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]:                                                 hidden_act='silu',
+01/03/2025 02:38:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]:                                                 hidden_size=4096,
+01/03/2025 02:38:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]:                                                 initializer_range=0.02,
+01/03/2025 02:38:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]:                                                 intermediate_size=14336,
+01/03/2025 02:38:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]:                                                 is_llama_config=True,
+01/03/2025 02:38:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]:                                                 max_position_embeddings=4096,
+01/03/2025 02:38:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]:                                                 num_attention_heads=32,
+01/03/2025 02:38:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]:                                                 num_hidden_layers=32,
+01/03/2025 02:38:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]:                                                 num_key_value_heads=32,
+01/03/2025 02:38:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]:                                                 pad_token_id=None,
+01/03/2025 02:38:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]:                                                 pretraining_tp=1,
+01/03/2025 02:38:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]:                                                 rms_norm_eps=1e-05,
+01/03/2025 02:38:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]:                                                 rope_scaling=None,
+01/03/2025 02:38:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]:                                                 rope_theta=10000.0,
+01/03/2025 02:38:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]:                                                 rope_interleaved=False,
+01/03/2025 02:38:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]:                                                 tie_word_embeddings=False,
+01/03/2025 02:38:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]:                                                 use_cache=True,
+01/03/2025 02:38:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]:                                                 vocab_size=131072),
+01/03/2025 02:38:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]:                        init_method=RandomInit(std=0.02),
+01/03/2025 02:38:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]:                        dtype=torch.bfloat16,
+01/03/2025 02:38:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]:                        make_vocab_size_divisible_by=1,
+01/03/2025 02:38:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]:                        ddp_bucket_cap_mb=25),
+01/03/2025 02:38:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]:        tokenizer=TokenizerArgs(tokenizer_name_or_path='robot-test/dummy-tokenizer-wordlevel',
+01/03/2025 02:38:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]:                                tokenizer_revision=None,
+01/03/2025 02:38:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]:                                tokenizer_max_length=None),
+01/03/2025 02:38:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]:        checkpoints=CheckpointsArgs(checkpoints_path=PosixPath('checkpoints'),
+01/03/2025 02:38:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]:                                    checkpoint_interval=10000,
+01/03/2025 02:38:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]:                                    save_initial_state=False,
+01/03/2025 02:38:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]:                                    save_final_state=False,
+01/03/2025 02:38:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]:                                    resume_checkpoint_path=None,
+01/03/2025 02:38:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]:                                    checkpoints_path_is_shared_file_system=False),
+01/03/2025 02:38:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]:        logging=LoggingArgs(log_level='info',
+01/03/2025 02:38:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]:                            log_level_replica='info',
+01/03/2025 02:38:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]:                            iteration_step_info_interval=1),
+01/03/2025 02:38:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]:        tokens=TokensArgs(sequence_length=4096,
+01/03/2025 02:38:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]:                          train_steps=100,
+01/03/2025 02:38:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]:                          micro_batch_size=1,
+01/03/2025 02:38:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]:                          batch_accumulation_per_replica=256,
+01/03/2025 02:38:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]:                          val_check_interval=100,
+01/03/2025 02:38:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]:                          limit_val_batches=0,
+01/03/2025 02:38:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]:                          limit_test_batches=0),
+01/03/2025 02:38:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]:        optimizer=OptimizerArgs(optimizer_factory=AdamWOptimizerArgs(adam_eps=1e-08,
+01/03/2025 02:38:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]:                                                                     adam_beta1=0.9,
+01/03/2025 02:38:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]:                                                                     adam_beta2=0.95,
+01/03/2025 02:38:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]:                                                                     torch_adam_is_fused=True,
+01/03/2025 02:38:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]:                                                                     name='adamW'),
+01/03/2025 02:38:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]:                                zero_stage=0,
+01/03/2025 02:38:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]:                                weight_decay=0.01,
+01/03/2025 02:38:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]:                                clip_grad=1.0,
+01/03/2025 02:38:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]:                                accumulate_grad_in_fp32=True,
+01/03/2025 02:38:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]:                                learning_rate_scheduler=LRSchedulerArgs(learning_rate=0.0003,
+01/03/2025 02:38:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]:                                                                        lr_warmup_steps=2,
+01/03/2025 02:38:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]:                                                                        lr_warmup_style='linear',
+01/03/2025 02:38:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]:                                                                        lr_decay_style='cosine',
+01/03/2025 02:38:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]:                                                                        lr_decay_steps=13,
+01/03/2025 02:38:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]:                                                                        lr_decay_starting_step=None,
+01/03/2025 02:38:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]:                                                                        min_decay_lr=1e-05)),
+01/03/2025 02:38:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]:        data_stages=[DatasetStageArgs(name='Stable Training Stage',
+01/03/2025 02:38:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]:                                      start_training_step=1,
+01/03/2025 02:38:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]:                                      data=DataArgs(dataset=None,
+01/03/2025 02:38:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]:                                                    seed=42,
+01/03/2025 02:38:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]:                                                    num_loading_workers=1))],
+01/03/2025 02:38:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]:        profiler=None,
+01/03/2025 02:38:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]:        lighteval=None,
+01/03/2025 02:38:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]:        s3_upload=None)
+01/03/2025 02:38:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]: Model Config:
+01/03/2025 02:38:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]: LlamaConfig(bos_token_id=0,
+01/03/2025 02:38:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]:             eos_token_id=0,
+01/03/2025 02:38:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]:             hidden_act='silu',
+01/03/2025 02:38:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]:             hidden_size=4096,
+01/03/2025 02:38:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]:             initializer_range=0.02,
+01/03/2025 02:38:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]:             intermediate_size=14336,
+01/03/2025 02:38:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]:             is_llama_config=True,
+01/03/2025 02:38:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]:             max_position_embeddings=4096,
+01/03/2025 02:38:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]:             num_attention_heads=32,
+01/03/2025 02:38:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]:             num_hidden_layers=32,
+01/03/2025 02:38:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]:             num_key_value_heads=32,
+01/03/2025 02:38:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]:             pad_token_id=None,
+01/03/2025 02:38:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]:             pretraining_tp=1,
+01/03/2025 02:38:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]:             rms_norm_eps=1e-05,
+01/03/2025 02:38:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]:             rope_scaling=None,
+01/03/2025 02:38:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]:             rope_theta=10000.0,
+01/03/2025 02:38:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]:             rope_interleaved=False,
+01/03/2025 02:38:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]:             tie_word_embeddings=False,
+01/03/2025 02:38:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]:             use_cache=True,
+01/03/2025 02:38:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]:             vocab_size=131072)
+01/03/2025 02:38:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]: Building model..
+01/03/2025 02:38:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]: Initialize RoPE Theta = 10000.0
+01/03/2025 02:38:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]: Setting PP block ranks...
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+01/03/2025 02:38:21 [INFO|DP=0|PP=0|TP=6|ip-26-0-165-213]: Local number of parameters: 158M (302.05MiB)
+01/03/2025 02:38:21 [INFO|DP=0|PP=0|TP=6|ip-26-0-165-213]: [After model building] Memory usage: 302.06MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:38:21 [INFO|DP=0|PP=0|TP=1|ip-26-0-165-213]: Local number of parameters: 158M (302.05MiB)
+01/03/2025 02:38:21 [INFO|DP=0|PP=0|TP=3|ip-26-0-165-213]: Local number of parameters: 158M (302.05MiB)
+01/03/2025 02:38:21 [INFO|DP=0|PP=0|TP=1|ip-26-0-165-213]: [After model building] Memory usage: 302.06MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:38:21 [INFO|DP=0|PP=0|TP=3|ip-26-0-165-213]: [After model building] Memory usage: 302.06MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:38:21 [INFO|DP=0|PP=0|TP=7|ip-26-0-165-213]: Local number of parameters: 158M (302.05MiB)
+01/03/2025 02:38:21 [INFO|DP=0|PP=0|TP=4|ip-26-0-165-213]: Local number of parameters: 158M (302.05MiB)
+01/03/2025 02:38:21 [INFO|DP=0|PP=0|TP=2|ip-26-0-165-213]: Local number of parameters: 158M (302.05MiB)
+01/03/2025 02:38:21 [INFO|DP=0|PP=0|TP=5|ip-26-0-165-213]: Local number of parameters: 158M (302.05MiB)
+01/03/2025 02:38:21 [INFO|DP=0|PP=10|TP=6|ip-26-0-167-51]: Local number of parameters: 60.8M (116.03MiB)
+01/03/2025 02:38:21 [INFO|DP=0|PP=12|TP=6|ip-26-0-168-30]: Local number of parameters: 60.8M (116.03MiB)
+01/03/2025 02:38:21 [INFO|DP=0|PP=4|TP=6|ip-26-0-166-244]: Local number of parameters: 60.8M (116.03MiB)
+01/03/2025 02:38:21 [INFO|DP=0|PP=3|TP=6|ip-26-0-166-214]: Local number of parameters: 60.8M (116.03MiB)
+01/03/2025 02:38:21 [INFO|DP=0|PP=9|TP=6|ip-26-0-167-245]: Local number of parameters: 60.8M (116.03MiB)
+01/03/2025 02:38:21 [INFO|DP=0|PP=14|TP=6|ip-26-0-168-52]: Local number of parameters: 128M (244.04MiB)
+01/03/2025 02:38:21 [INFO|DP=0|PP=1|TP=6|ip-26-0-166-125]: Local number of parameters: 60.8M (116.03MiB)
+01/03/2025 02:38:21 [INFO|DP=0|PP=0|TP=7|ip-26-0-165-213]: [After model building] Memory usage: 302.06MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:38:21 [INFO|DP=0|PP=2|TP=6|ip-26-0-166-15]: Local number of parameters: 60.8M (116.03MiB)
+01/03/2025 02:38:21 [INFO|DP=0|PP=15|TP=6|ip-26-0-168-95]: Local number of parameters: 0 (0.00MiB)
+01/03/2025 02:38:21 [INFO|DP=0|PP=13|TP=6|ip-26-0-168-34]: Local number of parameters: 60.8M (116.03MiB)
+01/03/2025 02:38:21 [INFO|DP=0|PP=8|TP=6|ip-26-0-167-217]: Local number of parameters: 60.8M (116.03MiB)
+01/03/2025 02:38:21 [INFO|DP=0|PP=11|TP=6|ip-26-0-167-9]: Local number of parameters: 60.8M (116.03MiB)
+01/03/2025 02:38:21 [INFO|DP=0|PP=10|TP=6|ip-26-0-167-51]: [After model building] Memory usage: 116.04MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:38:21 [INFO|DP=0|PP=0|TP=4|ip-26-0-165-213]: [After model building] Memory usage: 302.06MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:38:21 [INFO|DP=0|PP=9|TP=6|ip-26-0-167-245]: [After model building] Memory usage: 116.04MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:38:21 [INFO|DP=0|PP=3|TP=6|ip-26-0-166-214]: [After model building] Memory usage: 116.04MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:38:21 [INFO|DP=0|PP=14|TP=6|ip-26-0-168-52]: [After model building] Memory usage: 244.05MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:38:21 [INFO|DP=0|PP=12|TP=6|ip-26-0-168-30]: [After model building] Memory usage: 116.04MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:38:21 [INFO|DP=0|PP=2|TP=6|ip-26-0-166-15]: [After model building] Memory usage: 116.04MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:38:21 [INFO|DP=0|PP=4|TP=6|ip-26-0-166-244]: [After model building] Memory usage: 116.04MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:38:21 [INFO|DP=0|PP=1|TP=6|ip-26-0-166-125]: [After model building] Memory usage: 116.04MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:38:21 [INFO|DP=0|PP=15|TP=6|ip-26-0-168-95]: [After model building] Memory usage: 0.01MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:38:21 [INFO|DP=0|PP=13|TP=6|ip-26-0-168-34]: [After model building] Memory usage: 116.04MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:38:21 [INFO|DP=0|PP=0|TP=2|ip-26-0-165-213]: [After model building] Memory usage: 302.06MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:38:21 [INFO|DP=0|PP=0|TP=5|ip-26-0-165-213]: [After model building] Memory usage: 302.06MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:38:21 [INFO|DP=0|PP=11|TP=6|ip-26-0-167-9]: [After model building] Memory usage: 116.04MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:38:21 [INFO|DP=0|PP=8|TP=6|ip-26-0-167-217]: [After model building] Memory usage: 116.04MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:38:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]: Total number of parameters: 8.86G (16900.06MiB)
+01/03/2025 02:38:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]: Local number of parameters: 158M (302.05MiB)
+01/03/2025 02:38:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]: [After model building] Memory usage: 302.06MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:38:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]: No checkpoint path provided.
+01/03/2025 02:38:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]: Parametrizing model parameters using StandardParametrizator
+01/03/2025 02:38:21 [INFO|DP=0|PP=6|TP=6|ip-26-0-167-175]: Local number of parameters: 60.8M (116.03MiB)
+01/03/2025 02:38:21 [INFO|DP=0|PP=5|TP=6|ip-26-0-166-36]: Local number of parameters: 60.8M (116.03MiB)
+01/03/2025 02:38:21 [INFO|DP=0|PP=7|TP=6|ip-26-0-167-177]: Local number of parameters: 91.3M (174.05MiB)
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+01/03/2025 02:38:21 [INFO|DP=0|PP=6|TP=6|ip-26-0-167-175]: [After model building] Memory usage: 116.04MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:38:21 [INFO|DP=0|PP=5|TP=6|ip-26-0-166-36]: [After model building] Memory usage: 116.04MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+01/03/2025 02:38:21 [INFO|DP=0|PP=7|TP=6|ip-26-0-167-177]: [After model building] Memory usage: 174.06MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+01/03/2025 02:38:21 [INFO|DP=0|PP=2|TP=1|ip-26-0-166-15]: Local number of parameters: 60.8M (116.03MiB)
+01/03/2025 02:38:21 [INFO|DP=0|PP=4|TP=1|ip-26-0-166-244]: Local number of parameters: 60.8M (116.03MiB)
+01/03/2025 02:38:21 [INFO|DP=0|PP=5|TP=1|ip-26-0-166-36]: Local number of parameters: 60.8M (116.03MiB)
+NCCL version 2.18.5+cuda12.2
+01/03/2025 02:38:21 [INFO|DP=0|PP=9|TP=1|ip-26-0-167-245]: Local number of parameters: 60.8M (116.03MiB)
+01/03/2025 02:38:21 [INFO|DP=0|PP=3|TP=1|ip-26-0-166-214]: Local number of parameters: 60.8M (116.03MiB)
+01/03/2025 02:38:21 [INFO|DP=0|PP=6|TP=1|ip-26-0-167-175]: Local number of parameters: 60.8M (116.03MiB)
+01/03/2025 02:38:21 [INFO|DP=0|PP=2|TP=3|ip-26-0-166-15]: Local number of parameters: 60.8M (116.03MiB)
+01/03/2025 02:38:21 [INFO|DP=0|PP=4|TP=3|ip-26-0-166-244]: Local number of parameters: 60.8M (116.03MiB)
+01/03/2025 02:38:21 [INFO|DP=0|PP=13|TP=1|ip-26-0-168-34]: Local number of parameters: 60.8M (116.03MiB)
+01/03/2025 02:38:21 [INFO|DP=0|PP=12|TP=3|ip-26-0-168-30]: Local number of parameters: 60.8M (116.03MiB)
+01/03/2025 02:38:21 [INFO|DP=0|PP=8|TP=1|ip-26-0-167-217]: Local number of parameters: 60.8M (116.03MiB)
+01/03/2025 02:38:21 [INFO|DP=0|PP=1|TP=1|ip-26-0-166-125]: Local number of parameters: 60.8M (116.03MiB)
+01/03/2025 02:38:21 [INFO|DP=0|PP=1|TP=3|ip-26-0-166-125]: Local number of parameters: 60.8M (116.03MiB)
+01/03/2025 02:38:21 [INFO|DP=0|PP=15|TP=1|ip-26-0-168-95]: Local number of parameters: 0 (0.00MiB)
+01/03/2025 02:38:21 [INFO|DP=0|PP=11|TP=1|ip-26-0-167-9]: Local number of parameters: 60.8M (116.03MiB)
+01/03/2025 02:38:21 [INFO|DP=0|PP=8|TP=3|ip-26-0-167-217]: Local number of parameters: 60.8M (116.03MiB)
+01/03/2025 02:38:21 [INFO|DP=0|PP=10|TP=1|ip-26-0-167-51]: Local number of parameters: 60.8M (116.03MiB)
+01/03/2025 02:38:21 [INFO|DP=0|PP=10|TP=3|ip-26-0-167-51]: Local number of parameters: 60.8M (116.03MiB)
+01/03/2025 02:38:21 [INFO|DP=0|PP=9|TP=3|ip-26-0-167-245]: Local number of parameters: 60.8M (116.03MiB)
+01/03/2025 02:38:21 [INFO|DP=0|PP=14|TP=3|ip-26-0-168-52]: Local number of parameters: 128M (244.04MiB)
+01/03/2025 02:38:21 [INFO|DP=0|PP=3|TP=3|ip-26-0-166-214]: Local number of parameters: 60.8M (116.03MiB)
+01/03/2025 02:38:21 [INFO|DP=0|PP=12|TP=1|ip-26-0-168-30]: Local number of parameters: 60.8M (116.03MiB)
+01/03/2025 02:38:21 [INFO|DP=0|PP=4|TP=7|ip-26-0-166-244]: Local number of parameters: 60.8M (116.03MiB)
+01/03/2025 02:38:21 [INFO|DP=0|PP=14|TP=1|ip-26-0-168-52]: Local number of parameters: 128M (244.04MiB)
+NCCL version 2.18.5+cuda12.2
+01/03/2025 02:38:21 [INFO|DP=0|PP=5|TP=1|ip-26-0-166-36]: [After model building] Memory usage: 116.04MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:38:21 [INFO|DP=0|PP=2|TP=1|ip-26-0-166-15]: [After model building] Memory usage: 116.04MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:38:21 [INFO|DP=0|PP=8|TP=7|ip-26-0-167-217]: Local number of parameters: 60.8M (116.03MiB)
+01/03/2025 02:38:21 [INFO|DP=0|PP=4|TP=1|ip-26-0-166-244]: [After model building] Memory usage: 116.04MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:38:21 [INFO|DP=0|PP=9|TP=1|ip-26-0-167-245]: [After model building] Memory usage: 116.04MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:38:21 [INFO|DP=0|PP=3|TP=1|ip-26-0-166-214]: [After model building] Memory usage: 116.04MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:38:21 [INFO|DP=0|PP=1|TP=7|ip-26-0-166-125]: Local number of parameters: 60.8M (116.03MiB)
+01/03/2025 02:38:21 [INFO|DP=0|PP=11|TP=3|ip-26-0-167-9]: Local number of parameters: 60.8M (116.03MiB)
+01/03/2025 02:38:21 [INFO|DP=0|PP=6|TP=1|ip-26-0-167-175]: [After model building] Memory usage: 116.04MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:38:21 [INFO|DP=0|PP=4|TP=3|ip-26-0-166-244]: [After model building] Memory usage: 116.04MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:38:21 [INFO|DP=0|PP=15|TP=3|ip-26-0-168-95]: Local number of parameters: 0 (0.00MiB)
+01/03/2025 02:38:21 [INFO|DP=0|PP=13|TP=3|ip-26-0-168-34]: Local number of parameters: 60.8M (116.03MiB)
+01/03/2025 02:38:21 [INFO|DP=0|PP=2|TP=3|ip-26-0-166-15]: [After model building] Memory usage: 116.04MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:38:21 [INFO|DP=0|PP=3|TP=7|ip-26-0-166-214]: Local number of parameters: 60.8M (116.03MiB)
+01/03/2025 02:38:21 [INFO|DP=0|PP=10|TP=7|ip-26-0-167-51]: Local number of parameters: 60.8M (116.03MiB)
+01/03/2025 02:38:21 [INFO|DP=0|PP=7|TP=1|ip-26-0-167-177]: Local number of parameters: 91.3M (174.05MiB)
+01/03/2025 02:38:21 [INFO|DP=0|PP=5|TP=7|ip-26-0-166-36]: Local number of parameters: 60.8M (116.03MiB)
+01/03/2025 02:38:21 [INFO|DP=0|PP=13|TP=1|ip-26-0-168-34]: [After model building] Memory usage: 116.04MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:38:21 [INFO|DP=0|PP=9|TP=7|ip-26-0-167-245]: Local number of parameters: 60.8M (116.03MiB)
+01/03/2025 02:38:21 [INFO|DP=0|PP=7|TP=7|ip-26-0-167-177]: Local number of parameters: 91.3M (174.05MiB)
+01/03/2025 02:38:21 [INFO|DP=0|PP=2|TP=7|ip-26-0-166-15]: Local number of parameters: 60.8M (116.03MiB)
+01/03/2025 02:38:21 [INFO|DP=0|PP=11|TP=7|ip-26-0-167-9]: Local number of parameters: 60.8M (116.03MiB)
+01/03/2025 02:38:21 [INFO|DP=0|PP=12|TP=7|ip-26-0-168-30]: Local number of parameters: 60.8M (116.03MiB)
+01/03/2025 02:38:21 [INFO|DP=0|PP=12|TP=3|ip-26-0-168-30]: [After model building] Memory usage: 116.04MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:38:21 [INFO|DP=0|PP=14|TP=7|ip-26-0-168-52]: Local number of parameters: 128M (244.04MiB)
+01/03/2025 02:38:21 [INFO|DP=0|PP=8|TP=1|ip-26-0-167-217]: [After model building] Memory usage: 116.04MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:38:21 [INFO|DP=0|PP=1|TP=1|ip-26-0-166-125]: [After model building] Memory usage: 116.04MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:38:21 [INFO|DP=0|PP=11|TP=1|ip-26-0-167-9]: [After model building] Memory usage: 116.04MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:38:21 [INFO|DP=0|PP=9|TP=3|ip-26-0-167-245]: [After model building] Memory usage: 116.04MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:38:21 [INFO|DP=0|PP=6|TP=7|ip-26-0-167-175]: Local number of parameters: 60.8M (116.03MiB)
+01/03/2025 02:38:21 [INFO|DP=0|PP=13|TP=7|ip-26-0-168-34]: Local number of parameters: 60.8M (116.03MiB)
+01/03/2025 02:38:21 [INFO|DP=0|PP=8|TP=3|ip-26-0-167-217]: [After model building] Memory usage: 116.04MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:38:21 [INFO|DP=0|PP=3|TP=3|ip-26-0-166-214]: [After model building] Memory usage: 116.04MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:38:21 [INFO|DP=0|PP=1|TP=3|ip-26-0-166-125]: [After model building] Memory usage: 116.04MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:38:21 [INFO|DP=0|PP=10|TP=3|ip-26-0-167-51]: [After model building] Memory usage: 116.04MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:38:21 [INFO|DP=0|PP=10|TP=1|ip-26-0-167-51]: [After model building] Memory usage: 116.04MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:38:21 [INFO|DP=0|PP=4|TP=7|ip-26-0-166-244]: [After model building] Memory usage: 116.04MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:38:21 [INFO|DP=0|PP=14|TP=3|ip-26-0-168-52]: [After model building] Memory usage: 244.05MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:38:21 [INFO|DP=0|PP=8|TP=7|ip-26-0-167-217]: [After model building] Memory usage: 116.04MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:38:21 [INFO|DP=0|PP=15|TP=7|ip-26-0-168-95]: Local number of parameters: 0 (0.00MiB)
+01/03/2025 02:38:21 [INFO|DP=0|PP=15|TP=1|ip-26-0-168-95]: [After model building] Memory usage: 0.01MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:38:21 [INFO|DP=0|PP=12|TP=1|ip-26-0-168-30]: [After model building] Memory usage: 116.04MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:38:21 [INFO|DP=0|PP=11|TP=3|ip-26-0-167-9]: [After model building] Memory usage: 116.04MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:38:21 [INFO|DP=0|PP=14|TP=1|ip-26-0-168-52]: [After model building] Memory usage: 244.05MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:38:21 [INFO|DP=0|PP=13|TP=3|ip-26-0-168-34]: [After model building] Memory usage: 116.04MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:38:21 [INFO|DP=0|PP=1|TP=7|ip-26-0-166-125]: [After model building] Memory usage: 116.04MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:38:21 [INFO|DP=0|PP=10|TP=7|ip-26-0-167-51]: [After model building] Memory usage: 116.04MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:38:21 [INFO|DP=0|PP=3|TP=7|ip-26-0-166-214]: [After model building] Memory usage: 116.04MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:38:21 [INFO|DP=0|PP=5|TP=7|ip-26-0-166-36]: [After model building] Memory usage: 116.04MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:38:21 [INFO|DP=0|PP=7|TP=7|ip-26-0-167-177]: [After model building] Memory usage: 174.06MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:38:21 [INFO|DP=0|PP=11|TP=7|ip-26-0-167-9]: [After model building] Memory usage: 116.04MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:38:21 [INFO|DP=0|PP=2|TP=7|ip-26-0-166-15]: [After model building] Memory usage: 116.04MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:38:21 [INFO|DP=0|PP=14|TP=7|ip-26-0-168-52]: [After model building] Memory usage: 244.05MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:38:21 [INFO|DP=0|PP=15|TP=3|ip-26-0-168-95]: [After model building] Memory usage: 0.01MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:38:21 [INFO|DP=0|PP=9|TP=7|ip-26-0-167-245]: [After model building] Memory usage: 116.04MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:38:21 [INFO|DP=0|PP=7|TP=1|ip-26-0-167-177]: [After model building] Memory usage: 174.06MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:38:21 [INFO|DP=0|PP=12|TP=5|ip-26-0-168-30]: Local number of parameters: 60.8M (116.03MiB)
+01/03/2025 02:38:21 [INFO|DP=0|PP=12|TP=7|ip-26-0-168-30]: [After model building] Memory usage: 116.04MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:38:21 [INFO|DP=0|PP=8|TP=5|ip-26-0-167-217]: Local number of parameters: 60.8M (116.03MiB)
+01/03/2025 02:38:21 [INFO|DP=0|PP=6|TP=7|ip-26-0-167-175]: [After model building] Memory usage: 116.04MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:38:21 [INFO|DP=0|PP=13|TP=7|ip-26-0-168-34]: [After model building] Memory usage: 116.04MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:38:21 [INFO|DP=0|PP=1|TP=5|ip-26-0-166-125]: Local number of parameters: 60.8M (116.03MiB)
+01/03/2025 02:38:21 [INFO|DP=0|PP=6|TP=5|ip-26-0-167-175]: Local number of parameters: 60.8M (116.03MiB)
+01/03/2025 02:38:21 [INFO|DP=0|PP=4|TP=5|ip-26-0-166-244]: Local number of parameters: 60.8M (116.03MiB)
+01/03/2025 02:38:21 [INFO|DP=0|PP=13|TP=5|ip-26-0-168-34]: Local number of parameters: 60.8M (116.03MiB)
+01/03/2025 02:38:21 [INFO|DP=0|PP=5|TP=5|ip-26-0-166-36]: Local number of parameters: 60.8M (116.03MiB)
+01/03/2025 02:38:21 [INFO|DP=0|PP=7|TP=5|ip-26-0-167-177]: Local number of parameters: 91.3M (174.05MiB)
+01/03/2025 02:38:21 [INFO|DP=0|PP=8|TP=4|ip-26-0-167-217]: Local number of parameters: 60.8M (116.03MiB)
+01/03/2025 02:38:21 [INFO|DP=0|PP=10|TP=5|ip-26-0-167-51]: Local number of parameters: 60.8M (116.03MiB)
+01/03/2025 02:38:21 [INFO|DP=0|PP=15|TP=7|ip-26-0-168-95]: [After model building] Memory usage: 0.01MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:38:21 [INFO|DP=0|PP=4|TP=4|ip-26-0-166-244]: Local number of parameters: 60.8M (116.03MiB)
+01/03/2025 02:38:21 [INFO|DP=0|PP=11|TP=5|ip-26-0-167-9]: Local number of parameters: 60.8M (116.03MiB)
+01/03/2025 02:38:21 [INFO|DP=0|PP=2|TP=5|ip-26-0-166-15]: Local number of parameters: 60.8M (116.03MiB)
+01/03/2025 02:38:21 [INFO|DP=0|PP=9|TP=5|ip-26-0-167-245]: Local number of parameters: 60.8M (116.03MiB)
+01/03/2025 02:38:21 [INFO|DP=0|PP=15|TP=5|ip-26-0-168-95]: Local number of parameters: 0 (0.00MiB)
+01/03/2025 02:38:21 [INFO|DP=0|PP=3|TP=5|ip-26-0-166-214]: Local number of parameters: 60.8M (116.03MiB)
+01/03/2025 02:38:21 [INFO|DP=0|PP=2|TP=4|ip-26-0-166-15]: Local number of parameters: 60.8M (116.03MiB)
+01/03/2025 02:38:21 [INFO|DP=0|PP=1|TP=4|ip-26-0-166-125]: Local number of parameters: 60.8M (116.03MiB)
+01/03/2025 02:38:21 [INFO|DP=0|PP=11|TP=4|ip-26-0-167-9]: Local number of parameters: 60.8M (116.03MiB)
+01/03/2025 02:38:21 [INFO|DP=0|PP=5|TP=3|ip-26-0-166-36]: Local number of parameters: 60.8M (116.03MiB)
+01/03/2025 02:38:21 [INFO|DP=0|PP=14|TP=5|ip-26-0-168-52]: Local number of parameters: 128M (244.04MiB)
+01/03/2025 02:38:21 [INFO|DP=0|PP=3|TP=4|ip-26-0-166-214]: Local number of parameters: 60.8M (116.03MiB)
+01/03/2025 02:38:21 [INFO|DP=0|PP=9|TP=4|ip-26-0-167-245]: Local number of parameters: 60.8M (116.03MiB)
+01/03/2025 02:38:21 [INFO|DP=0|PP=6|TP=4|ip-26-0-167-175]: Local number of parameters: 60.8M (116.03MiB)
+01/03/2025 02:38:21 [INFO|DP=0|PP=10|TP=4|ip-26-0-167-51]: Local number of parameters: 60.8M (116.03MiB)
+01/03/2025 02:38:21 [INFO|DP=0|PP=4|TP=2|ip-26-0-166-244]: Local number of parameters: 60.8M (116.03MiB)
+01/03/2025 02:38:21 [INFO|DP=0|PP=14|TP=4|ip-26-0-168-52]: Local number of parameters: 128M (244.04MiB)
+01/03/2025 02:38:21 [INFO|DP=0|PP=8|TP=2|ip-26-0-167-217]: Local number of parameters: 60.8M (116.03MiB)
+01/03/2025 02:38:21 [INFO|DP=0|PP=6|TP=3|ip-26-0-167-175]: Local number of parameters: 60.8M (116.03MiB)
+01/03/2025 02:38:21 [INFO|DP=0|PP=12|TP=4|ip-26-0-168-30]: Local number of parameters: 60.8M (116.03MiB)
+01/03/2025 02:38:21 [INFO|DP=0|PP=12|TP=5|ip-26-0-168-30]: [After model building] Memory usage: 116.04MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:38:21 [INFO|DP=0|PP=5|TP=4|ip-26-0-166-36]: Local number of parameters: 60.8M (116.03MiB)
+01/03/2025 02:38:21 [INFO|DP=0|PP=13|TP=4|ip-26-0-168-34]: Local number of parameters: 60.8M (116.03MiB)
+01/03/2025 02:38:21 [INFO|DP=0|PP=2|TP=2|ip-26-0-166-15]: Local number of parameters: 60.8M (116.03MiB)
+01/03/2025 02:38:21 [INFO|DP=0|PP=8|TP=5|ip-26-0-167-217]: [After model building] Memory usage: 116.04MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:38:21 [INFO|DP=0|PP=1|TP=5|ip-26-0-166-125]: [After model building] Memory usage: 116.04MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:38:21 [INFO|DP=0|PP=6|TP=5|ip-26-0-167-175]: [After model building] Memory usage: 116.04MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:38:21 [INFO|DP=0|PP=4|TP=5|ip-26-0-166-244]: [After model building] Memory usage: 116.04MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:38:21 [INFO|DP=0|PP=7|TP=4|ip-26-0-167-177]: Local number of parameters: 91.3M (174.05MiB)
+01/03/2025 02:38:21 [INFO|DP=0|PP=13|TP=5|ip-26-0-168-34]: [After model building] Memory usage: 116.04MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:38:21 [INFO|DP=0|PP=1|TP=2|ip-26-0-166-125]: Local number of parameters: 60.8M (116.03MiB)
+01/03/2025 02:38:21 [INFO|DP=0|PP=6|TP=2|ip-26-0-167-175]: Local number of parameters: 60.8M (116.03MiB)
+01/03/2025 02:38:21 [INFO|DP=0|PP=8|TP=4|ip-26-0-167-217]: [After model building] Memory usage: 116.04MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:38:21 [INFO|DP=0|PP=4|TP=4|ip-26-0-166-244]: [After model building] Memory usage: 116.04MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:38:21 [INFO|DP=0|PP=7|TP=3|ip-26-0-167-177]: Local number of parameters: 91.3M (174.05MiB)
+01/03/2025 02:38:21 [INFO|DP=0|PP=10|TP=5|ip-26-0-167-51]: [After model building] Memory usage: 116.04MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:38:21 [INFO|DP=0|PP=11|TP=5|ip-26-0-167-9]: [After model building] Memory usage: 116.04MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:38:21 [INFO|DP=0|PP=3|TP=2|ip-26-0-166-214]: Local number of parameters: 60.8M (116.03MiB)
+01/03/2025 02:38:21 [INFO|DP=0|PP=15|TP=4|ip-26-0-168-95]: Local number of parameters: 0 (0.00MiB)
+01/03/2025 02:38:21 [INFO|DP=0|PP=5|TP=2|ip-26-0-166-36]: Local number of parameters: 60.8M (116.03MiB)
+01/03/2025 02:38:21 [INFO|DP=0|PP=5|TP=5|ip-26-0-166-36]: [After model building] Memory usage: 116.04MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:38:21 [INFO|DP=0|PP=13|TP=2|ip-26-0-168-34]: Local number of parameters: 60.8M (116.03MiB)
+01/03/2025 02:38:21 [INFO|DP=0|PP=2|TP=5|ip-26-0-166-15]: [After model building] Memory usage: 116.04MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:38:21 [INFO|DP=0|PP=12|TP=2|ip-26-0-168-30]: Local number of parameters: 60.8M (116.03MiB)
+01/03/2025 02:38:21 [INFO|DP=0|PP=7|TP=5|ip-26-0-167-177]: [After model building] Memory usage: 174.06MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:38:21 [INFO|DP=0|PP=7|TP=2|ip-26-0-167-177]: Local number of parameters: 91.3M (174.05MiB)
+01/03/2025 02:38:21 [INFO|DP=0|PP=9|TP=2|ip-26-0-167-245]: Local number of parameters: 60.8M (116.03MiB)
+01/03/2025 02:38:21 [INFO|DP=0|PP=15|TP=5|ip-26-0-168-95]: [After model building] Memory usage: 0.01MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:38:21 [INFO|DP=0|PP=3|TP=4|ip-26-0-166-214]: [After model building] Memory usage: 116.04MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:38:21 [INFO|DP=0|PP=3|TP=5|ip-26-0-166-214]: [After model building] Memory usage: 116.04MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:38:21 [INFO|DP=0|PP=10|TP=2|ip-26-0-167-51]: Local number of parameters: 60.8M (116.03MiB)
+01/03/2025 02:38:21 [INFO|DP=0|PP=1|TP=4|ip-26-0-166-125]: [After model building] Memory usage: 116.04MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+NCCL version 2.18.5+cuda12.2
+01/03/2025 02:38:21 [INFO|DP=0|PP=14|TP=2|ip-26-0-168-52]: Local number of parameters: 128M (244.04MiB)
+01/03/2025 02:38:21 [INFO|DP=0|PP=2|TP=4|ip-26-0-166-15]: [After model building] Memory usage: 116.04MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:38:21 [INFO|DP=0|PP=6|TP=4|ip-26-0-167-175]: [After model building] Memory usage: 116.04MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+NCCL version 2.18.5+cuda12.2
+01/03/2025 02:38:21 [INFO|DP=0|PP=11|TP=2|ip-26-0-167-9]: Local number of parameters: 60.8M (116.03MiB)
+01/03/2025 02:38:21 [INFO|DP=0|PP=9|TP=5|ip-26-0-167-245]: [After model building] Memory usage: 116.04MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:38:21 [INFO|DP=0|PP=8|TP=2|ip-26-0-167-217]: [After model building] Memory usage: 116.04MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:38:21 [INFO|DP=0|PP=4|TP=2|ip-26-0-166-244]: [After model building] Memory usage: 116.04MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:38:21 [INFO|DP=0|PP=5|TP=3|ip-26-0-166-36]: [After model building] Memory usage: 116.04MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:38:21 [INFO|DP=0|PP=14|TP=5|ip-26-0-168-52]: [After model building] Memory usage: 244.05MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:38:21 [INFO|DP=0|PP=10|TP=4|ip-26-0-167-51]: [After model building] Memory usage: 116.04MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:38:21 [INFO|DP=0|PP=14|TP=4|ip-26-0-168-52]: [After model building] Memory usage: 244.05MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:38:21 [INFO|DP=0|PP=11|TP=4|ip-26-0-167-9]: [After model building] Memory usage: 116.04MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:38:21 [INFO|DP=0|PP=9|TP=4|ip-26-0-167-245]: [After model building] Memory usage: 116.04MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:38:21 [INFO|DP=0|PP=6|TP=3|ip-26-0-167-175]: [After model building] Memory usage: 116.04MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:38:21 [INFO|DP=0|PP=15|TP=2|ip-26-0-168-95]: Local number of parameters: 0 (0.00MiB)
+01/03/2025 02:38:21 [INFO|DP=0|PP=13|TP=4|ip-26-0-168-34]: [After model building] Memory usage: 116.04MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:38:21 [INFO|DP=0|PP=12|TP=4|ip-26-0-168-30]: [After model building] Memory usage: 116.04MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:38:21 [INFO|DP=0|PP=5|TP=4|ip-26-0-166-36]: [After model building] Memory usage: 116.04MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:38:21 [INFO|DP=0|PP=2|TP=2|ip-26-0-166-15]: [After model building] Memory usage: 116.04MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:38:21 [INFO|DP=0|PP=1|TP=2|ip-26-0-166-125]: [After model building] Memory usage: 116.04MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:38:21 [INFO|DP=0|PP=4|TP=0|ip-26-0-166-244]: Local number of parameters: 60.8M (116.03MiB)
+01/03/2025 02:38:21 [INFO|DP=0|PP=7|TP=3|ip-26-0-167-177]: [After model building] Memory usage: 174.06MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:38:21 [INFO|DP=0|PP=6|TP=2|ip-26-0-167-175]: [After model building] Memory usage: 116.04MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:38:21 [INFO|DP=0|PP=5|TP=2|ip-26-0-166-36]: [After model building] Memory usage: 116.04MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:38:21 [INFO|DP=0|PP=13|TP=2|ip-26-0-168-34]: [After model building] Memory usage: 116.04MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:38:21 [INFO|DP=0|PP=3|TP=2|ip-26-0-166-214]: [After model building] Memory usage: 116.04MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:38:21 [INFO|DP=0|PP=2|TP=0|ip-26-0-166-15]: Local number of parameters: 60.8M (116.03MiB)
+01/03/2025 02:38:21 [INFO|DP=0|PP=12|TP=2|ip-26-0-168-30]: [After model building] Memory usage: 116.04MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:38:21 [INFO|DP=0|PP=12|TP=0|ip-26-0-168-30]: Local number of parameters: 60.8M (116.03MiB)
+01/03/2025 02:38:21 [INFO|DP=0|PP=15|TP=4|ip-26-0-168-95]: [After model building] Memory usage: 0.01MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:38:21 [INFO|DP=0|PP=11|TP=2|ip-26-0-167-9]: [After model building] Memory usage: 116.04MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:38:21 [INFO|DP=0|PP=7|TP=4|ip-26-0-167-177]: [After model building] Memory usage: 174.06MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:38:21 [INFO|DP=0|PP=10|TP=0|ip-26-0-167-51]: Local number of parameters: 60.8M (116.03MiB)
+01/03/2025 02:38:21 [INFO|DP=0|PP=6|TP=0|ip-26-0-167-175]: Local number of parameters: 60.8M (116.03MiB)
+01/03/2025 02:38:21 [INFO|DP=0|PP=8|TP=0|ip-26-0-167-217]: Local number of parameters: 60.8M (116.03MiB)
+01/03/2025 02:38:21 [INFO|DP=0|PP=7|TP=2|ip-26-0-167-177]: [After model building] Memory usage: 174.06MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:38:21 [INFO|DP=0|PP=10|TP=2|ip-26-0-167-51]: [After model building] Memory usage: 116.04MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:38:21 [INFO|DP=0|PP=15|TP=0|ip-26-0-168-95]: Local number of parameters: 0 (0.00MiB)
+01/03/2025 02:38:21 [INFO|DP=0|PP=9|TP=2|ip-26-0-167-245]: [After model building] Memory usage: 116.04MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:38:21 [INFO|DP=0|PP=14|TP=2|ip-26-0-168-52]: [After model building] Memory usage: 244.05MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:38:21 [INFO|DP=0|PP=14|TP=0|ip-26-0-168-52]: Local number of parameters: 128M (244.04MiB)
+01/03/2025 02:38:21 [INFO|DP=0|PP=11|TP=0|ip-26-0-167-9]: Local number of parameters: 60.8M (116.03MiB)
+01/03/2025 02:38:21 [INFO|DP=0|PP=15|TP=2|ip-26-0-168-95]: [After model building] Memory usage: 0.01MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+NCCL version 2.18.5+cuda12.2
+01/03/2025 02:38:21 [INFO|DP=0|PP=13|TP=0|ip-26-0-168-34]: Local number of parameters: 60.8M (116.03MiB)
+01/03/2025 02:38:21 [INFO|DP=0|PP=4|TP=0|ip-26-0-166-244]: [After model building] Memory usage: 116.04MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:38:21 [INFO|DP=0|PP=2|TP=0|ip-26-0-166-15]: [After model building] Memory usage: 116.04MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:38:21 [INFO|DP=0|PP=10|TP=0|ip-26-0-167-51]: [After model building] Memory usage: 116.04MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:38:21 [INFO|DP=0|PP=12|TP=0|ip-26-0-168-30]: [After model building] Memory usage: 116.04MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:38:21 [INFO|DP=0|PP=6|TP=0|ip-26-0-167-175]: [After model building] Memory usage: 116.04MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:38:21 [INFO|DP=0|PP=8|TP=0|ip-26-0-167-217]: [After model building] Memory usage: 116.04MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:38:21 [INFO|DP=0|PP=15|TP=0|ip-26-0-168-95]: [After model building] Memory usage: 0.01MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+01/03/2025 02:38:21 [INFO|DP=0|PP=14|TP=0|ip-26-0-168-52]: [After model building] Memory usage: 244.05MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:38:21 [INFO|DP=0|PP=11|TP=0|ip-26-0-167-9]: [After model building] Memory usage: 116.04MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:38:21 [INFO|DP=0|PP=13|TP=0|ip-26-0-168-34]: [After model building] Memory usage: 116.04MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+01/03/2025 02:38:21 [INFO|DP=0|PP=1|TP=0|ip-26-0-166-125]: Local number of parameters: 60.8M (116.03MiB)
+01/03/2025 02:38:21 [INFO|DP=0|PP=7|TP=0|ip-26-0-167-177]: Local number of parameters: 91.3M (174.05MiB)
+01/03/2025 02:38:21 [INFO|DP=0|PP=5|TP=0|ip-26-0-166-36]: Local number of parameters: 60.8M (116.03MiB)
+01/03/2025 02:38:21 [INFO|DP=0|PP=3|TP=0|ip-26-0-166-214]: Local number of parameters: 60.8M (116.03MiB)
+01/03/2025 02:38:21 [INFO|DP=0|PP=9|TP=0|ip-26-0-167-245]: Local number of parameters: 60.8M (116.03MiB)
+01/03/2025 02:38:21 [INFO|DP=0|PP=1|TP=0|ip-26-0-166-125]: [After model building] Memory usage: 116.04MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:38:21 [INFO|DP=0|PP=7|TP=0|ip-26-0-167-177]: [After model building] Memory usage: 174.06MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:38:21 [INFO|DP=0|PP=3|TP=0|ip-26-0-166-214]: [After model building] Memory usage: 116.04MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:38:21 [INFO|DP=0|PP=5|TP=0|ip-26-0-166-36]: [After model building] Memory usage: 116.04MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 02:38:21 [INFO|DP=0|PP=9|TP=0|ip-26-0-167-245]: [After model building] Memory usage: 116.04MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+01/03/2025 02:38:22 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]: [Optimizer Building] Using LearningRateForSP as learning rate
+01/03/2025 02:38:22 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]: module_name: model.token_position_embeddings | PP: 0/16 | Block rank: 0
+01/03/2025 02:38:22 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]: module_name: model.decoder.0 | PP: 0/16 | Block rank: 1
+01/03/2025 02:38:22 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]: module_name: model.decoder.1 | PP: 0/16 | Block rank: 2
+01/03/2025 02:38:22 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]: module_name: model.decoder.2 | PP: 0/16 | Block rank: 3
+01/03/2025 02:38:22 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]: module_name: model.decoder.3 | PP: 1/16 | Block rank: 0
+01/03/2025 02:38:22 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]: module_name: model.decoder.4 | PP: 1/16 | Block rank: 1
+01/03/2025 02:38:22 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]: module_name: model.decoder.5 | PP: 2/16 | Block rank: 0
+01/03/2025 02:38:22 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]: module_name: model.decoder.6 | PP: 2/16 | Block rank: 1
+01/03/2025 02:38:22 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]: module_name: model.decoder.7 | PP: 3/16 | Block rank: 0
+01/03/2025 02:38:22 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]: module_name: model.decoder.8 | PP: 3/16 | Block rank: 1
+01/03/2025 02:38:22 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]: module_name: model.decoder.9 | PP: 4/16 | Block rank: 0
+01/03/2025 02:38:22 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]: module_name: model.decoder.10 | PP: 4/16 | Block rank: 1
+01/03/2025 02:38:22 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]: module_name: model.decoder.11 | PP: 5/16 | Block rank: 0
+01/03/2025 02:38:22 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]: module_name: model.decoder.12 | PP: 5/16 | Block rank: 1
+01/03/2025 02:38:22 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]: module_name: model.decoder.13 | PP: 6/16 | Block rank: 0
+01/03/2025 02:38:22 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]: module_name: model.decoder.14 | PP: 6/16 | Block rank: 1
+01/03/2025 02:38:22 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]: module_name: model.decoder.15 | PP: 7/16 | Block rank: 0
+01/03/2025 02:38:22 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]: module_name: model.decoder.16 | PP: 7/16 | Block rank: 1
+01/03/2025 02:38:22 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]: module_name: model.decoder.17 | PP: 7/16 | Block rank: 2
+01/03/2025 02:38:22 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]: module_name: model.decoder.18 | PP: 8/16 | Block rank: 0
+01/03/2025 02:38:22 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]: module_name: model.decoder.19 | PP: 8/16 | Block rank: 1
+01/03/2025 02:38:22 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]: module_name: model.decoder.20 | PP: 9/16 | Block rank: 0
+01/03/2025 02:38:22 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]: module_name: model.decoder.21 | PP: 9/16 | Block rank: 1
+01/03/2025 02:38:22 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]: module_name: model.decoder.22 | PP: 10/16 | Block rank: 0
+01/03/2025 02:38:22 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]: module_name: model.decoder.23 | PP: 10/16 | Block rank: 1
+01/03/2025 02:38:22 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]: module_name: model.decoder.24 | PP: 11/16 | Block rank: 0
+01/03/2025 02:38:22 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]: module_name: model.decoder.25 | PP: 11/16 | Block rank: 1
+01/03/2025 02:38:22 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]: module_name: model.decoder.26 | PP: 12/16 | Block rank: 0
+01/03/2025 02:38:22 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]: module_name: model.decoder.27 | PP: 12/16 | Block rank: 1
+01/03/2025 02:38:22 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]: module_name: model.decoder.28 | PP: 13/16 | Block rank: 0
+01/03/2025 02:38:22 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]: module_name: model.decoder.29 | PP: 13/16 | Block rank: 1
+01/03/2025 02:38:22 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]: module_name: model.decoder.30 | PP: 14/16 | Block rank: 0
+01/03/2025 02:38:22 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]: module_name: model.decoder.31 | PP: 14/16 | Block rank: 1
+01/03/2025 02:38:22 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]: module_name: model.final_layer_norm | PP: 14/16 | Block rank: 2
+01/03/2025 02:38:22 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]: module_name: model.lm_head | PP: 14/16 | Block rank: 3
+01/03/2025 02:38:22 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]: module_name: model.cast_to_fp32 | PP: 15/16 | Block rank: 0
+01/03/2025 02:38:22 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]: module_name: loss | PP: 15/16 | Block rank: 1
+01/03/2025 02:38:22 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]: [Training Plan] Stage Stable Training Stage has 99 remaining training steps and has consumed 0 samples
+01/03/2025 02:38:22 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]: Using dummy data generator
+01/03/2025 02:38:22 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]: [Training Plan] There are 1 training stages 
+01/03/2025 02:38:22 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]: [Stage Stable Training Stage] start from step 1 
+01/03/2025 02:38:22 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]: 
+01/03/2025 02:38:22 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]: [Start training] datetime: 2025-01-03 02:38:22.926808 | mbs: 1 | grad_accum: 256 | global_batch_size: 256 | sequence_length: 4096 | train_steps: 100 | start_iteration_step: 0 | consumed_train_samples: 0
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+01/03/2025 02:38:23 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]: Resuming training from stage Stable Training Stage, it has trained for 0 samples and has 99 remaining train steps
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+01/03/2025 02:38:23 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]:  Memory usage: 1510.25MiB. Peak allocated 5392.00MiB. Peak reserved: 10518.00MiB
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+01/03/2025 02:44:22 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]:  Memory usage: 1641.28MiB. Peak allocated 6493.31MiB. Peak reserved: 7030.00MiB
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+01/03/2025 02:44:33 [INFO|DP=0|PP=15|TP=0|ip-26-0-168-95]: iteration: 1 / 100 | consumed_tokens: 1.05M | elapsed_time_per_iteration_ms: 369K | tokens_per_sec: 2.84K | tokens_per_sec_per_gpu: 22.2 | global_batch_size: 256 | lm_loss: 12.6 | lr: 0.00015 | model_tflops_per_gpu: 1.25 | hardware_tflops_per_gpu: 1.25 | grad_norm: 1.81 | cuda_memory_allocated: 179K | cuda_max_memory_reserved: 942M | hd_total_memory_tb: 312G | hd_used_memory_tb: 69.4G | hd_free_memory_tb: 243G
+01/03/2025 02:44:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]:  Memory usage: 2849.48MiB. Peak allocated 2849.48MiB. Peak reserved: 7030.00MiB
+01/03/2025 02:49:44 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]:  Memory usage: 2849.48MiB. Peak allocated 7577.73MiB. Peak reserved: 8290.00MiB
+01/03/2025 02:49:44 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]:  Memory usage: 2849.48MiB. Peak allocated 2849.49MiB. Peak reserved: 8290.00MiB
+01/03/2025 02:49:44 [INFO|DP=0|PP=15|TP=0|ip-26-0-168-95]: iteration: 2 / 100 | consumed_tokens: 2.1M | elapsed_time_per_iteration_ms: 312K | tokens_per_sec: 3.37K | tokens_per_sec_per_gpu: 26.3 | global_batch_size: 256 | lm_loss: 12.6 | lr: 0.0003 | model_tflops_per_gpu: 1.48 | hardware_tflops_per_gpu: 1.48 | grad_norm: 1.81 | cuda_memory_allocated: 179K | cuda_max_memory_reserved: 942M | hd_total_memory_tb: 312G | hd_used_memory_tb: 69.4G | hd_free_memory_tb: 243G
+01/03/2025 02:54:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]:  Memory usage: 2849.48MiB. Peak allocated 7577.73MiB. Peak reserved: 8362.00MiB
+01/03/2025 02:54:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]: | job_id   | name                                                             | nodes | seq_len | mbs | batch_accum | gbs | mTFLOPs | hTFLOPs | tok/s/gpu | AllReduce (GB/s) | AllGather (GB/s) | ReduceScatter (GB/s) | AR Intra-node (GB/s) | AG Intra-node (GB/s) | RS Intra-node (GB/s) | Mem Alloc (GB) | Mem Res (GB) | dp | pp | tp | pp_engine                           | tp_mode        | tp_async_comm | hidden_size | hidden_act | num_layers | num_heads | num_kv_heads | max_pos | vocab_size | tie_word_embeddings | dtype          | zero_stage | ddp_bucket_cap_mb | accumulate_grad_in_fp32 | Total Params | Local Params |
+01/03/2025 02:54:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]: | -------- | ---------------------------------------------------------------- | ----- | ------- | --- | ----------- | --- | ------- | ------- | --------- | ---------------- | ---------------- | -------------------- | -------------------- | -------------------- | -------------------- | -------------- | ------------ | -- | -- | -- | ----------------------------------- | -------------- | ------------- | ----------- | ---------- | ---------- | --------- | ------------ | ------- | ---------- | ------------------- | -------------- | ---------- | ----------------- | ----------------------- | ------------ | ------------ |
+01/03/2025 02:54:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]: | 14019076 | 8.86G_dp1_tp8_pp16_acc256_mbs1_seq4096_zero0_tpmodeRED_vocab131k | 16    | 4096    | 1   | 256         | 256 | 1.48    | 1.48    | 26.33     | 53.02            | 28.26            | 27.78                | 427.82               | 242.82               | 242.70               | 2.78           | 8.17         | 1  | 16 | 8  | OneForwardOneBackwardPipelineEngine | REDUCE_SCATTER | True          | 4096        | silu       | 32         | 32        | 32           | 4096    | 131072     | False               | torch.bfloat16 | 0          | 25                | True                    | 8.86G        | 158M         |
+01/03/2025 02:54:55 [INFO|DP=0|PP=15|TP=0|ip-26-0-168-95]: iteration: 3 / 100 | consumed_tokens: 3.15M | elapsed_time_per_iteration_ms: 311K | tokens_per_sec: 3.37K | tokens_per_sec_per_gpu: 26.3 | global_batch_size: 256 | lm_loss: 12.6 | lr: 0.000296 | model_tflops_per_gpu: 1.48 | hardware_tflops_per_gpu: 1.48 | grad_norm: 1.79 | cuda_memory_allocated: 179K | cuda_max_memory_reserved: 942M | hd_total_memory_tb: 312G | hd_used_memory_tb: 69.4G | hd_free_memory_tb: 243G
+01/03/2025 02:54:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]: Attempting to write benchmark results to CSV file: benchmark/results/bench_final2.csv
+01/03/2025 02:54:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]: Acquired lock for CSV file: benchmark/results/bench_final2.csv
+01/03/2025 02:54:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]: Successfully wrote to CSV file: benchmark/results/bench_final2.csv. Releasing lock...
+01/03/2025 02:54:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-165-213]: Throughput logging complete
+terminate called after throwing an instance of 'c10::Error'
+  what():  CUDA driver error: unknown error
+Exception raised from _hasPrimaryContext at ../aten/src/ATen/cuda/detail/CUDAHooks.cpp:67 (most recent call first):
+frame #0: c10::Error::Error(c10::SourceLocation, std::string) + 0x57 (0x7f2d8c707617 in /fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/lib/libc10.so)
+frame #1: c10::detail::torchCheckFail(char const*, char const*, unsigned int, char const*) + 0x68 (0x7f2d8c6c2a56 in /fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/lib/libc10.so)
+frame #2: <unknown function> + 0x11a00bf (0x7f2d8da230bf in /fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/lib/libtorch_cuda.so)
+frame #3: c10::cuda::MaybeSetDevice(int) + 0xc (0x7f2d8c7b8bdc in /fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/lib/libc10_cuda.so)
+frame #4: std::_Sp_counted_ptr_inplace<std::vector<at::cuda::CUDAEvent, std::allocator<at::cuda::CUDAEvent> >, std::allocator<std::vector<at::cuda::CUDAEvent, std::allocator<at::cuda::CUDAEvent> > >, (__gnu_cxx::_Lock_policy)2>::_M_dispose() + 0x98 (0x7f2d8db32ae8 in /fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/lib/libtorch_cuda.so)
+frame #5: std::_Sp_counted_base<(__gnu_cxx::_Lock_policy)2>::_M_release() + 0x48 (0x7f2deba85048 in /fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/lib/libtorch_python.so)
+frame #6: c10d::ProcessGroupNCCL::WorkNCCL::~WorkNCCL() + 0x135 (0x7f2d8dafdc55 in /fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/lib/libtorch_cuda.so)
+frame #7: c10d::ProcessGroupNCCL::workCleanupLoop() + 0x3c5 (0x7f2d8db13585 in /fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/lib/libtorch_cuda.so)
+frame #8: c10d::ProcessGroupNCCL::ncclCommWatchdog() + 0x78 (0x7f2d8db13718 in /fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/lib/libtorch_cuda.so)
+frame #9: <unknown function> + 0xd3e95 (0x7f2e1b80be95 in /fsx/nouamane/miniconda/envs/2-1-cu121/bin/../lib/libstdc++.so.6)
+frame #10: <unknown function> + 0x8609 (0x7f2e1bc7b609 in /lib/x86_64-linux-gnu/libpthread.so.0)
+frame #11: clone + 0x43 (0x7f2e1ba44353 in /lib/x86_64-linux-gnu/libc.so.6)
+
+[2025-01-03 02:55:18,959] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 140752 closing signal SIGTERM
+[2025-01-03 02:55:18,959] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 140753 closing signal SIGTERM
+[2025-01-03 02:55:18,959] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 140754 closing signal SIGTERM
+[2025-01-03 02:55:18,959] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 140755 closing signal SIGTERM
+[2025-01-03 02:55:18,959] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 140756 closing signal SIGTERM
+[2025-01-03 02:55:18,959] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 140757 closing signal SIGTERM
+[2025-01-03 02:55:18,959] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 140758 closing signal SIGTERM
+[2025-01-03 02:55:19,110] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 267814 closing signal SIGTERM
+[2025-01-03 02:55:19,110] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 267815 closing signal SIGTERM
+[2025-01-03 02:55:19,110] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 267816 closing signal SIGTERM
+[2025-01-03 02:55:19,110] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 267817 closing signal SIGTERM
+[2025-01-03 02:55:19,110] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 267818 closing signal SIGTERM
+[2025-01-03 02:55:19,110] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 267820 closing signal SIGTERM
+[2025-01-03 02:55:21,541] torch.distributed.elastic.multiprocessing.api: [ERROR] failed (exitcode: 1) local_rank: 0 (pid: 267813) of binary: /fsx/nouamane/miniconda/envs/2-1-cu121/bin/python
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 264, in launch_agent
+    raise ChildFailedError(
+torch.distributed.elastic.multiprocessing.errors.ChildFailedError: 
+============================================================
+/fsx/nouamane/projects/nanotron/run_train.py FAILED
+------------------------------------------------------------
+Failures:
+[1]:
+  time      : 2025-01-03_02:55:19
+  host      : ip-26-0-168-95.ec2.internal
+  rank      : 126 (local_rank: 6)
+  exitcode  : 1 (pid: 267819)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+------------------------------------------------------------
+Root Cause (first observed failure):
+[0]:
+  time      : 2025-01-03_02:55:19
+  host      : ip-26-0-168-95.ec2.internal
+  rank      : 120 (local_rank: 0)
+  exitcode  : 1 (pid: 267813)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+============================================================
+srun: error: ip-26-0-168-95: task 15: Exited with exit code 1
+srun: launch/slurm: _step_signal: Terminating StepId=14019076.0
+slurmstepd: error: *** STEP 14019076.0 ON ip-26-0-165-213 CANCELLED AT 2025-01-03T02:55:21 ***
+[2025-01-03 02:55:21,864] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 02:55:21,864] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 140754 closing signal SIGTERM
+[2025-01-03 02:55:21,864] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 140756 closing signal SIGTERM
+[2025-01-03 02:55:21,864] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 140758 closing signal SIGTERM
+[2025-01-03 02:55:21,864] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 02:55:21,864] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 641272 closing signal SIGTERM
+[2025-01-03 02:55:21,864] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 641273 closing signal SIGTERM
+[2025-01-03 02:55:21,864] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 641274 closing signal SIGTERM
+[2025-01-03 02:55:21,864] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 641275 closing signal SIGTERM
+[2025-01-03 02:55:21,864] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 641276 closing signal SIGTERM
+[2025-01-03 02:55:21,864] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 02:55:21,864] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 02:55:21,864] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 641277 closing signal SIGTERM
+[2025-01-03 02:55:21,864] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 641278 closing signal SIGTERM
+[2025-01-03 02:55:21,865] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 130731 closing signal SIGTERM
+[2025-01-03 02:55:21,864] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 4100622 closing signal SIGTERM
+[2025-01-03 02:55:21,864] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 641279 closing signal SIGTERM
+[2025-01-03 02:55:21,865] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 130732 closing signal SIGTERM
+[2025-01-03 02:55:21,864] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 4100623 closing signal SIGTERM
+[2025-01-03 02:55:21,863] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 02:55:21,865] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 130733 closing signal SIGTERM
+[2025-01-03 02:55:21,864] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 4100624 closing signal SIGTERM
+[2025-01-03 02:55:21,865] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 130734 closing signal SIGTERM
+[2025-01-03 02:55:21,865] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 4100625 closing signal SIGTERM
+[2025-01-03 02:55:21,863] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 339052 closing signal SIGTERM
+[2025-01-03 02:55:21,865] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 130735 closing signal SIGTERM
+[2025-01-03 02:55:21,865] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 4100626 closing signal SIGTERM
+[2025-01-03 02:55:21,865] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 130736 closing signal SIGTERM
+[2025-01-03 02:55:21,865] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 4100627 closing signal SIGTERM
+[2025-01-03 02:55:21,865] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 130737 closing signal SIGTERM
+[2025-01-03 02:55:21,865] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 4100628 closing signal SIGTERM
+[2025-01-03 02:55:21,863] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 339053 closing signal SIGTERM
+[2025-01-03 02:55:21,864] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 02:55:21,865] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 130738 closing signal SIGTERM
+[2025-01-03 02:55:21,865] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 4100629 closing signal SIGTERM
+[2025-01-03 02:55:21,863] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 339054 closing signal SIGTERM
+[2025-01-03 02:55:21,864] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 02:55:21,865] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 195139 closing signal SIGTERM
+[2025-01-03 02:55:21,863] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 339055 closing signal SIGTERM
+[2025-01-03 02:55:21,865] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 195140 closing signal SIGTERM
+[2025-01-03 02:55:21,863] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 339056 closing signal SIGTERM
+[2025-01-03 02:55:21,864] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 578825 closing signal SIGTERM
+[2025-01-03 02:55:21,863] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 339057 closing signal SIGTERM
+[2025-01-03 02:55:21,865] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 195141 closing signal SIGTERM
+[2025-01-03 02:55:21,864] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 02:55:21,865] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 195142 closing signal SIGTERM
+[2025-01-03 02:55:21,864] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 578826 closing signal SIGTERM
+[2025-01-03 02:55:21,864] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 02:55:21,864] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 578827 closing signal SIGTERM
+[2025-01-03 02:55:21,863] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 339058 closing signal SIGTERM
+[2025-01-03 02:55:21,865] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 195143 closing signal SIGTERM
+[2025-01-03 02:55:21,863] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 339059 closing signal SIGTERM
+[2025-01-03 02:55:21,865] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 440133 closing signal SIGTERM
+[2025-01-03 02:55:21,865] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 195144 closing signal SIGTERM
+[2025-01-03 02:55:21,864] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 578828 closing signal SIGTERM
+[2025-01-03 02:55:21,864] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 578829 closing signal SIGTERM
+[2025-01-03 02:55:21,865] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 195145 closing signal SIGTERM
+[2025-01-03 02:55:21,865] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 195146 closing signal SIGTERM
+[2025-01-03 02:55:21,864] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 02:55:21,865] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 230436 closing signal SIGTERM
+[2025-01-03 02:55:21,864] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 578830 closing signal SIGTERM
+[2025-01-03 02:55:21,864] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 578831 closing signal SIGTERM
+[2025-01-03 02:55:21,864] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 416702 closing signal SIGTERM
+[2025-01-03 02:55:21,864] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 578832 closing signal SIGTERM
+[2025-01-03 02:55:21,865] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 230437 closing signal SIGTERM
+[2025-01-03 02:55:21,864] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 416703 closing signal SIGTERM
+[2025-01-03 02:55:21,864] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 416704 closing signal SIGTERM
+[2025-01-03 02:55:21,865] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 230438 closing signal SIGTERM
+[2025-01-03 02:55:21,865] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 440134 closing signal SIGTERM
+[2025-01-03 02:55:21,864] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 416705 closing signal SIGTERM
+[2025-01-03 02:55:21,864] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 416706 closing signal SIGTERM
+[2025-01-03 02:55:21,865] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 230439 closing signal SIGTERM
+[2025-01-03 02:55:21,865] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 440135 closing signal SIGTERM
+[2025-01-03 02:55:21,864] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 416707 closing signal SIGTERM
+[2025-01-03 02:55:21,864] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 416708 closing signal SIGTERM
+[2025-01-03 02:55:21,865] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 230440 closing signal SIGTERM
+[2025-01-03 02:55:21,865] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 440136 closing signal SIGTERM
+[2025-01-03 02:55:21,865] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 440137 closing signal SIGTERM
+[2025-01-03 02:55:21,864] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 416709 closing signal SIGTERM
+[2025-01-03 02:55:21,865] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 230441 closing signal SIGTERM
+[2025-01-03 02:55:21,865] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 440138 closing signal SIGTERM
+[2025-01-03 02:55:21,865] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 440139 closing signal SIGTERM
+[2025-01-03 02:55:21,864] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 02:55:21,865] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 440140 closing signal SIGTERM
+[2025-01-03 02:55:21,865] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 230442 closing signal SIGTERM
+[2025-01-03 02:55:21,863] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 02:55:21,865] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 230443 closing signal SIGTERM
+[2025-01-03 02:55:21,865] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 381863 closing signal SIGTERM
+[2025-01-03 02:55:21,864] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 491417 closing signal SIGTERM
+[2025-01-03 02:55:21,864] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 02:55:21,864] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 33499 closing signal SIGTERM
+[2025-01-03 02:55:21,865] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 381864 closing signal SIGTERM
+[2025-01-03 02:55:21,864] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 491418 closing signal SIGTERM
+[2025-01-03 02:55:21,865] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 33500 closing signal SIGTERM
+[2025-01-03 02:55:21,865] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 381865 closing signal SIGTERM
+[2025-01-03 02:55:21,865] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 02:55:21,864] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 491419 closing signal SIGTERM
+[2025-01-03 02:55:21,865] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 381866 closing signal SIGTERM
+[2025-01-03 02:55:21,865] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 33501 closing signal SIGTERM
+[2025-01-03 02:55:21,866] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 381867 closing signal SIGTERM
+[2025-01-03 02:55:21,864] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 491420 closing signal SIGTERM
+[2025-01-03 02:55:21,866] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 136336 closing signal SIGTERM
+[2025-01-03 02:55:21,866] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 381868 closing signal SIGTERM
+[2025-01-03 02:55:21,865] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 33502 closing signal SIGTERM
+[2025-01-03 02:55:21,865] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 33503 closing signal SIGTERM
+[2025-01-03 02:55:21,864] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 491421 closing signal SIGTERM
+[2025-01-03 02:55:21,864] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 491422 closing signal SIGTERM
+[2025-01-03 02:55:21,866] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 381869 closing signal SIGTERM
+[2025-01-03 02:55:21,866] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 381870 closing signal SIGTERM
+[2025-01-03 02:55:21,865] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 33504 closing signal SIGTERM
+[2025-01-03 02:55:21,865] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 33505 closing signal SIGTERM
+[2025-01-03 02:55:21,864] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 491423 closing signal SIGTERM
+[2025-01-03 02:55:21,864] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 491424 closing signal SIGTERM
+[2025-01-03 02:55:21,865] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 33506 closing signal SIGTERM
+[2025-01-03 02:55:21,866] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 136337 closing signal SIGTERM
+[2025-01-03 02:55:21,866] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 136338 closing signal SIGTERM
+[2025-01-03 02:55:21,866] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 136339 closing signal SIGTERM
+[2025-01-03 02:55:21,866] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 136340 closing signal SIGTERM
+[2025-01-03 02:55:21,866] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 136341 closing signal SIGTERM
+[2025-01-03 02:55:21,866] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 136342 closing signal SIGTERM
+[2025-01-03 02:55:21,866] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 136343 closing signal SIGTERM
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 878, in _invoke_run
+    run_result = self._monitor_workers(self._worker_group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/local_elastic_agent.py", line 307, in _monitor_workers
+    result = self._pcontext.wait(0)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 288, in wait
+    return self._poll()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 668, in _poll
+    self.close()  # terminate all running procs
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 331, in close
+    self._close(death_sig=death_sig, timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 713, in _close
+    handler.proc.wait(time_to_wait)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1209, in wait
+    return self._wait(timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1953, in _wait
+    time.sleep(delay)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 140677 got signal: 15
+srun: error: ip-26-0-165-213: task 0: Exited with exit code 1
+[2025-01-03 02:55:22,835] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-166-244.ec2.internal_195067_0' has failed to send a keep-alive heartbeat to the rendezvous '14019076' due to an error of type RendezvousConnectionError.
+[2025-01-03 02:55:22,907] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-166-15.ec2.internal_641199_0' has failed to send a keep-alive heartbeat to the rendezvous '14019076' due to an error of type RendezvousConnectionError.
+[2025-01-03 02:55:23,564] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-167-217.ec2.internal_230364_0' has failed to send a keep-alive heartbeat to the rendezvous '14019076' due to an error of type RendezvousConnectionError.
+[2025-01-03 02:55:23,583] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-166-125.ec2.internal_440059_0' has failed to send a keep-alive heartbeat to the rendezvous '14019076' due to an error of type RendezvousConnectionError.
+[2025-01-03 02:55:23,618] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-168-52.ec2.internal_416631_0' has failed to send a keep-alive heartbeat to the rendezvous '14019076' due to an error of type RendezvousConnectionError.
+[2025-01-03 02:55:23,619] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-168-34.ec2.internal_33426_0' has failed to send a keep-alive heartbeat to the rendezvous '14019076' due to an error of type RendezvousConnectionError.
+[2025-01-03 02:55:23,672] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-167-245.ec2.internal_4100551_0' has failed to send a keep-alive heartbeat to the rendezvous '14019076' due to an error of type RendezvousConnectionError.
+[2025-01-03 02:55:23,697] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-168-30.ec2.internal_578753_0' has failed to send a keep-alive heartbeat to the rendezvous '14019076' due to an error of type RendezvousConnectionError.
+[2025-01-03 02:55:23,720] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-167-177.ec2.internal_338981_0' has failed to send a keep-alive heartbeat to the rendezvous '14019076' due to an error of type RendezvousConnectionError.
+[2025-01-03 02:55:23,725] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-167-9.ec2.internal_130658_0' has failed to send a keep-alive heartbeat to the rendezvous '14019076' due to an error of type RendezvousConnectionError.
+[2025-01-03 02:55:23,753] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-167-51.ec2.internal_136264_0' has failed to send a keep-alive heartbeat to the rendezvous '14019076' due to an error of type RendezvousConnectionError.
+[2025-01-03 02:55:23,756] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-166-214.ec2.internal_381790_0' has failed to send a keep-alive heartbeat to the rendezvous '14019076' due to an error of type RendezvousConnectionError.
+[2025-01-03 02:55:23,759] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-166-36.ec2.internal_491344_0' has failed to send a keep-alive heartbeat to the rendezvous '14019076' due to an error of type RendezvousConnectionError.
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 136264 got signal: 15
+[2025-01-03 02:55:27,836] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-166-244.ec2.internal_195067_0' has failed to send a keep-alive heartbeat to the rendezvous '14019076' due to an error of type RendezvousConnectionError.
+[2025-01-03 02:55:27,908] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-166-15.ec2.internal_641199_0' has failed to send a keep-alive heartbeat to the rendezvous '14019076' due to an error of type RendezvousConnectionError.
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 33426 got signal: 15
+srun: error: ip-26-0-167-51: task 7: Exited with exit code 1
+srun: error: ip-26-0-168-34: task 13: Exited with exit code 1
+[2025-01-03 02:55:28,565] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-167-217.ec2.internal_230364_0' has failed to send a keep-alive heartbeat to the rendezvous '14019076' due to an error of type RendezvousConnectionError.
+[2025-01-03 02:55:28,584] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-166-125.ec2.internal_440059_0' has failed to send a keep-alive heartbeat to the rendezvous '14019076' due to an error of type RendezvousConnectionError.
+[2025-01-03 02:55:28,620] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-168-52.ec2.internal_416631_0' has failed to send a keep-alive heartbeat to the rendezvous '14019076' due to an error of type RendezvousConnectionError.
+[2025-01-03 02:55:28,674] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-167-245.ec2.internal_4100551_0' has failed to send a keep-alive heartbeat to the rendezvous '14019076' due to an error of type RendezvousConnectionError.
+[2025-01-03 02:55:28,698] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-168-30.ec2.internal_578753_0' has failed to send a keep-alive heartbeat to the rendezvous '14019076' due to an error of type RendezvousConnectionError.
+[2025-01-03 02:55:28,722] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-167-177.ec2.internal_338981_0' has failed to send a keep-alive heartbeat to the rendezvous '14019076' due to an error of type RendezvousConnectionError.
+[2025-01-03 02:55:28,727] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-167-9.ec2.internal_130658_0' has failed to send a keep-alive heartbeat to the rendezvous '14019076' due to an error of type RendezvousConnectionError.
+[2025-01-03 02:55:28,758] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-166-214.ec2.internal_381790_0' has failed to send a keep-alive heartbeat to the rendezvous '14019076' due to an error of type RendezvousConnectionError.
+[2025-01-03 02:55:28,761] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-166-36.ec2.internal_491344_0' has failed to send a keep-alive heartbeat to the rendezvous '14019076' due to an error of type RendezvousConnectionError.
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 440059 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 381790 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+srun: error: ip-26-0-166-125: task 3: Exited with exit code 1
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 230364 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 491344 got signal: 15
+srun: error: ip-26-0-166-214: task 4: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 578753 got signal: 15
+srun: error: ip-26-0-167-217: task 10: Exited with exit code 1
+srun: error: ip-26-0-168-30: task 12: Exited with exit code 1
+srun: error: ip-26-0-166-36: task 2: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 130658 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 195067 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 338981 got signal: 15
+srun: error: ip-26-0-166-244: task 5: Exited with exit code 1
+srun: error: ip-26-0-167-9: task 6: Exited with exit code 1
+srun: error: ip-26-0-167-177: task 9: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 4100551 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 416631 got signal: 15
+srun: error: ip-26-0-167-245: task 11: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 641199 got signal: 15
+srun: error: ip-26-0-168-52: task 14: Exited with exit code 1
+srun: error: ip-26-0-166-15: task 1: Exited with exit code 1
+srun: error: Node failure on ip-26-0-167-175
+slurmstepd: error: *** JOB 14019076 ON ip-26-0-165-213 CANCELLED AT 2025-01-03T02:56:43 DUE TO NODE FAILURE, SEE SLURMCTLD LOG FOR DETAILS ***
+srun: Job step aborted: Waiting up to 32 seconds for job step to finish.
diff --git a/logs/14019149-bench_80G_dp16_tp4_pp2_acc16_mbs1_seq4096_zero0_tpmodeRED_vocab131k.out b/logs/14019149-bench_80G_dp16_tp4_pp2_acc16_mbs1_seq4096_zero0_tpmodeRED_vocab131k.out
new file mode 100644
index 0000000000000000000000000000000000000000..d8772f618618352b29b76f10b2bccb41d674447f
--- /dev/null
+++ b/logs/14019149-bench_80G_dp16_tp4_pp2_acc16_mbs1_seq4096_zero0_tpmodeRED_vocab131k.out
@@ -0,0 +1,3928 @@
++ '[' -z 14019149 ']'
++ source /etc/profile.d/modules.sh
+++ . /usr/share/modules/init/bash
++++ unset _mlshdbg
++++ '[' 0 = 1 ']'
++++ unset _mlre _mlIFS
++++ '[' -n x ']'
++++ _mlIFS=' 	
+'
++++ IFS=' '
++++ '[' -n '' ']'
+++++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash autoinit
++++ _mlcode='module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+MODULEPATH_modshare=/etc/environment-modules/modules:1:/usr/share/modules/\$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1; export MODULEPATH_modshare;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+LOADEDMODULES=; export LOADEDMODULES;
+MODULEPATH=/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/\$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles; export MODULEPATH;
+test 0;'
++++ _mlret=0
++++ '[' -n x ']'
++++ IFS=' 	
+'
++++ unset _mlIFS
++++ unset _mlre _mlv _mlrv
++++ '[' 0 -eq 0 ']'
++++ eval 'module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+MODULEPATH_modshare=/etc/environment-modules/modules:1:/usr/share/modules/\$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1; export MODULEPATH_modshare;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+LOADEDMODULES=; export LOADEDMODULES;
+MODULEPATH=/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/\$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles; export MODULEPATH;
+test 0;'
+++++ MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl
+++++ export MODULES_CMD
+++++ ENV=/usr/share/modules/init/profile.sh
+++++ export ENV
+++++ MODULEPATH_modshare='/etc/environment-modules/modules:1:/usr/share/modules/$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1'
+++++ export MODULEPATH_modshare
+++++ BASH_ENV=/usr/share/modules/init/bash
+++++ export BASH_ENV
+++++ MODULESHOME=/usr/share/modules
+++++ export MODULESHOME
+++++ LOADEDMODULES=
+++++ export LOADEDMODULES
+++++ MODULEPATH='/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles'
+++++ export MODULEPATH
+++++ test 0
++++ '[' 0 = 1 ']'
++++ '[' -t 2 ']'
++++ export -f module
++++ export -f switchml
++++ '[' 5 -ge 3 ']'
++++ [[ ehxB =~ i ]]
++++ [[ ! :/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin: =~ :/usr/bin: ]]
++++ '[' '!' -n '' ']'
++++ MANPATH=:
++++ export MANPATH
+++++ manpath
++++ [[ ! :/fsx/nouamane/miniconda/envs/2-1-cu121/man:/fsx/nouamane/miniconda/envs/2-1-cu121/share/man:/fsx/nouamane/miniconda/man:/fsx/nouamane/miniconda/share/man:/usr/local/man:/usr/local/share/man:/usr/share/man:/opt/slurm/share/man:/admin/home/nouamane/.local/share/man:/admin/home/nouamane/.fzf/man:: =~ :/usr/share/man: ]]
++++ unset _mlcode _mlret
++++ '[' -n '' ']'
++ module load cuda/12.1
++ unset _mlshdbg
++ '[' 0 = 1 ']'
++ unset _mlre _mlIFS
++ '[' -n x ']'
++ _mlIFS=' 	
+'
++ IFS=' '
++ '[' -n '' ']'
+++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash load cuda/12.1
++ eval 'CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include;' export 'CPATH;
+LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:;' export 'LD_LIBRARY_PATH;
+FI_EFA_FORK_SAFE=1;' export 'FI_EFA_FORK_SAFE;
+MANPATH=/usr/local/cuda-12.1/share/man::;' export 'MANPATH;
+LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64;' export 'LIBRARY_PATH;
+_LMFILES_=/usr/share/modules/modulefiles/cuda/12.1;' export '_LMFILES_;
+LOADEDMODULES=cuda/12.1;' export 'LOADEDMODULES;
+MPI_PATH=/opt/amazon/openmpi;' export 'MPI_PATH;
+NCCL_HOME_modshare=/opt/nccl/build:1;' export 'NCCL_HOME_modshare;
+NCCL_PROTO=simple;' export 'NCCL_PROTO;
+MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1;' export 'MANPATH_modshare;
+LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1;' export 'LIBRARY_PATH_modshare;
+NCCL_SOCKET_IFNAME=enp;' export 'NCCL_SOCKET_IFNAME;
+AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl;' export 'AWS_OFI_NCCL_HOME;
+NCCL_HOME=/opt/nccl/build;' export 'NCCL_HOME;
+FI_PROVIDER=efa;' export 'FI_PROVIDER;
+AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1;' export 'AWS_OFI_NCCL_HOME_modshare;
+CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1;' export 'CPATH_modshare;
+LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2::1:/opt/nccl/build/lib:1:/opt/aws-ofi-nccl/lib:2:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/efa/lib:1;' export 'LD_LIBRARY_PATH_modshare;
+FI_EFA_ENABLE_SHM_TRANSFER=1;' export 'FI_EFA_ENABLE_SHM_TRANSFER;
+_LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1;' export '_LMFILES__modshare;
+LOADEDMODULES_modshare=cuda/12.1:1;' export 'LOADEDMODULES_modshare;
+MPI_PATH_modshare=/opt/amazon/openmpi:1;' export 'MPI_PATH_modshare;
+PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin;' export 'PATH;
+CUDA_HOME=/usr/local/cuda-12.1;' export 'CUDA_HOME;
+PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/opt/slurm/bin:1:/usr/sbin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1;' export 'PATH_modshare;
+test' '0;'
+++ CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include
+++ export CPATH
+++ LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:
+++ export LD_LIBRARY_PATH
+++ FI_EFA_FORK_SAFE=1
+++ export FI_EFA_FORK_SAFE
+++ MANPATH=/usr/local/cuda-12.1/share/man::
+++ export MANPATH
+++ LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64
+++ export LIBRARY_PATH
+++ _LMFILES_=/usr/share/modules/modulefiles/cuda/12.1
+++ export _LMFILES_
+++ LOADEDMODULES=cuda/12.1
+++ export LOADEDMODULES
+++ MPI_PATH=/opt/amazon/openmpi
+++ export MPI_PATH
+++ NCCL_HOME_modshare=/opt/nccl/build:1
+++ export NCCL_HOME_modshare
+++ NCCL_PROTO=simple
+++ export NCCL_PROTO
+++ MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1
+++ export MANPATH_modshare
+++ LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1
+++ export LIBRARY_PATH_modshare
+++ NCCL_SOCKET_IFNAME=enp
+++ export NCCL_SOCKET_IFNAME
+++ AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl
+++ export AWS_OFI_NCCL_HOME
+++ NCCL_HOME=/opt/nccl/build
+++ export NCCL_HOME
+++ FI_PROVIDER=efa
+++ export FI_PROVIDER
+++ AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1
+++ export AWS_OFI_NCCL_HOME_modshare
+++ CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1
+++ export CPATH_modshare
+++ LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2::1:/opt/nccl/build/lib:1:/opt/aws-ofi-nccl/lib:2:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/efa/lib:1
+++ export LD_LIBRARY_PATH_modshare
+++ FI_EFA_ENABLE_SHM_TRANSFER=1
+++ export FI_EFA_ENABLE_SHM_TRANSFER
+++ _LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1
+++ export _LMFILES__modshare
+++ LOADEDMODULES_modshare=cuda/12.1:1
+++ export LOADEDMODULES_modshare
+++ MPI_PATH_modshare=/opt/amazon/openmpi:1
+++ export MPI_PATH_modshare
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
+++ export PATH
+++ CUDA_HOME=/usr/local/cuda-12.1
+++ export CUDA_HOME
+++ PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/opt/slurm/bin:1:/usr/sbin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1
+++ export PATH_modshare
+++ test 0
++ _mlstatus=0
++ '[' -n x ']'
++ IFS=' 	
+'
++ unset _mlre _mlv _mlrv _mlIFS
++ '[' -n '' ']'
++ unset _mlshdbg
++ return 0
++ source /fsx/nouamane/miniconda/bin/activate
+++ _CONDA_ROOT=/fsx/nouamane/miniconda
+++ . /fsx/nouamane/miniconda/etc/profile.d/conda.sh
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ '[' -z x ']'
+++ conda activate
+++ local cmd=activate
+++ case "$cmd" in
+++ __conda_activate activate
+++ '[' -n '' ']'
+++ local ask_conda
++++ PS1=
++++ __conda_exe shell.posix activate
++++ /fsx/nouamane/miniconda/bin/conda shell.posix activate
+Error while loading conda entry point: conda-libmamba-solver (module 'libmambapy' has no attribute 'QueryFormat')
+++ ask_conda='. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''2'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_1='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/etc/conda/activate.d/libxml2_activate.sh"'
+++ eval '. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''2'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_1='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/etc/conda/activate.d/libxml2_activate.sh"'
++++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh
+++++ test -n ''
+++++ unset XML_CATALOG_FILES
+++++ unset xml_catalog_files_libxml2
++++ PS1='(base) '
++++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
++++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
++++ export CONDA_PREFIX=/fsx/nouamane/miniconda
++++ CONDA_PREFIX=/fsx/nouamane/miniconda
++++ export CONDA_SHLVL=2
++++ CONDA_SHLVL=2
++++ export CONDA_DEFAULT_ENV=base
++++ CONDA_DEFAULT_ENV=base
++++ export 'CONDA_PROMPT_MODIFIER=(base) '
++++ CONDA_PROMPT_MODIFIER='(base) '
++++ export CONDA_PREFIX_1=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ CONDA_PREFIX_1=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ . /fsx/nouamane/miniconda/etc/conda/activate.d/libxml2_activate.sh
+++++ test -n ''
+++++ xml_catalog_files_libxml2=
+++++ XML_CATALOG_FILES=
+++++ conda_catalog_files=
+++++ ifs_libxml2=' 	
+'
+++++ IFS=' '
+++++ rem=/fsx/nouamane/miniconda
+++++ for pre in ${rem}
+++++ test '' = /fsx/nouamane/miniconda
+++++ conda_catalog_files=/fsx/nouamane/miniconda
+++++ rem=
+++++ IFS=' 	
+'
+++++ conda_catalog_files='file:///fsx/nouamane/miniconda/etc/xml/catalog file:///etc/xml/catalog'
+++++ export 'XML_CATALOG_FILES=file:///fsx/nouamane/miniconda/etc/xml/catalog file:///etc/xml/catalog'
+++++ XML_CATALOG_FILES='file:///fsx/nouamane/miniconda/etc/xml/catalog file:///etc/xml/catalog'
+++++ unset conda_catalog_files ifs_libxml2 rem
+++ __conda_hashr
+++ '[' -n '' ']'
+++ '[' -n '' ']'
+++ hash -r
++ conda activate 2-1-cu121
++ local cmd=activate
++ case "$cmd" in
++ __conda_activate activate 2-1-cu121
++ '[' -n '' ']'
++ local ask_conda
+++ PS1='(base) '
+++ __conda_exe shell.posix activate 2-1-cu121
+++ /fsx/nouamane/miniconda/bin/conda shell.posix activate 2-1-cu121
+Error while loading conda entry point: conda-libmamba-solver (module 'libmambapy' has no attribute 'QueryFormat')
++ ask_conda='. "/fsx/nouamane/miniconda/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_2='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
++ eval '. "/fsx/nouamane/miniconda/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_2='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
+++ . /fsx/nouamane/miniconda/etc/conda/deactivate.d/libxml2_deactivate.sh
++++ test -n ''
++++ unset XML_CATALOG_FILES
++++ unset xml_catalog_files_libxml2
+++ PS1='(2-1-cu121) '
+++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
+++ export CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ export CONDA_SHLVL=3
+++ CONDA_SHLVL=3
+++ export CONDA_DEFAULT_ENV=2-1-cu121
+++ CONDA_DEFAULT_ENV=2-1-cu121
+++ export 'CONDA_PROMPT_MODIFIER=(2-1-cu121) '
+++ CONDA_PROMPT_MODIFIER='(2-1-cu121) '
+++ export CONDA_PREFIX_2=/fsx/nouamane/miniconda
+++ CONDA_PREFIX_2=/fsx/nouamane/miniconda
+++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ export _CE_M=
+++ _CE_M=
+++ export _CE_CONDA=
+++ _CE_CONDA=
+++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh
++++ test -n ''
++++ xml_catalog_files_libxml2=
++++ XML_CATALOG_FILES=
++++ conda_catalog_files=
++++ ifs_libxml2=' 	
+'
++++ IFS=' '
++++ rem=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ for pre in ${rem}
++++ test '' = /fsx/nouamane/miniconda/envs/2-1-cu121
++++ conda_catalog_files=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ rem=
++++ IFS=' 	
+'
++++ conda_catalog_files='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ export 'XML_CATALOG_FILES=file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ XML_CATALOG_FILES='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ unset conda_catalog_files ifs_libxml2 rem
++ __conda_hashr
++ '[' -n '' ']'
++ '[' -n '' ']'
++ hash -r
++ export PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
++ PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
++ '[' -z '' ']'
+++ scontrol show hostnames 'ip-26-0-171-[56,62,88,102,168,230,249],ip-26-0-172-[57,73,116,142,147,252],ip-26-0-173-[7,121,202]'
++ export 'NODELIST=ip-26-0-171-56
+ip-26-0-171-62
+ip-26-0-171-88
+ip-26-0-171-102
+ip-26-0-171-168
+ip-26-0-171-230
+ip-26-0-171-249
+ip-26-0-172-57
+ip-26-0-172-73
+ip-26-0-172-116
+ip-26-0-172-142
+ip-26-0-172-147
+ip-26-0-172-252
+ip-26-0-173-7
+ip-26-0-173-121
+ip-26-0-173-202'
++ NODELIST='ip-26-0-171-56
+ip-26-0-171-62
+ip-26-0-171-88
+ip-26-0-171-102
+ip-26-0-171-168
+ip-26-0-171-230
+ip-26-0-171-249
+ip-26-0-172-57
+ip-26-0-172-73
+ip-26-0-172-116
+ip-26-0-172-142
+ip-26-0-172-147
+ip-26-0-172-252
+ip-26-0-173-7
+ip-26-0-173-121
+ip-26-0-173-202'
+++ head -n1
+++ scontrol show hostnames 'ip-26-0-171-[56,62,88,102,168,230,249],ip-26-0-172-[57,73,116,142,147,252],ip-26-0-173-[7,121,202]'
++ export MASTER_NODE=ip-26-0-171-56
++ MASTER_NODE=ip-26-0-171-56
++ export MASTER_PORT=12356
++ MASTER_PORT=12356
++ export NNODES=16
++ NNODES=16
++ export GPUS_PER_NODE=8
++ GPUS_PER_NODE=8
++ export WORLD_SIZE=128
++ WORLD_SIZE=128
++ export CUDA_DEVICE_MAX_CONNECTIONS=1
++ CUDA_DEVICE_MAX_CONNECTIONS=1
++ export NCCL_DEBUG=WARN
++ NCCL_DEBUG=WARN
++ export NANOTRON_BENCHMARK=1
++ NANOTRON_BENCHMARK=1
++ export WANDB_MODE=disabled
++ WANDB_MODE=disabled
++ export TORCH_NCCL_ASYNC_ERROR_HANDLING=1
++ TORCH_NCCL_ASYNC_ERROR_HANDLING=1
++ export TORCH_DISTRIBUTED_DEBUG=DETAIL
++ TORCH_DISTRIBUTED_DEBUG=DETAIL
++ '[' -z '' ']'
++ echo '=== GPU Topology ==='
+=== GPU Topology ===
++ nvidia-smi topo -m
+	GPU0	GPU1	GPU2	GPU3	GPU4	GPU5	GPU6	GPU7	CPU Affinity	NUMA Affinity	GPU NUMA ID
+GPU0	 X 	NV18	NV18	NV18	NV18	NV18	NV18	NV18	0-43	0		N/A
+GPU1	NV18	 X 	NV18	NV18	NV18	NV18	NV18	NV18	0-43	0		N/A
+GPU2	NV18	NV18	 X 	NV18	NV18	NV18	NV18	NV18	0-43	0		N/A
+GPU3	NV18	NV18	NV18	 X 	NV18	NV18	NV18	NV18	0-43	0		N/A
+GPU4	NV18	NV18	NV18	NV18	 X 	NV18	NV18	NV18	52-95	1		N/A
+GPU5	NV18	NV18	NV18	NV18	NV18	 X 	NV18	NV18	52-95	1		N/A
+GPU6	NV18	NV18	NV18	NV18	NV18	NV18	 X 	NV18	52-95	1		N/A
+GPU7	NV18	NV18	NV18	NV18	NV18	NV18	NV18	 X 	52-95	1		N/A
+
+Legend:
+
+  X    = Self
+  SYS  = Connection traversing PCIe as well as the SMP interconnect between NUMA nodes (e.g., QPI/UPI)
+  NODE = Connection traversing PCIe as well as the interconnect between PCIe Host Bridges within a NUMA node
+  PHB  = Connection traversing PCIe as well as a PCIe Host Bridge (typically the CPU)
+  PXB  = Connection traversing multiple PCIe bridges (without traversing the PCIe Host Bridge)
+  PIX  = Connection traversing at most a single PCIe bridge
+  NV#  = Connection traversing a bonded set of # NVLinks
++ echo ==================
+==================
++ export SRUN_ALLOC_ARGS=
++ SRUN_ALLOC_ARGS=
++ echo 'Master node: ip-26-0-171-56'
+Master node: ip-26-0-171-56
++ echo 'All nodes: ip-26-0-171-56
+ip-26-0-171-62
+ip-26-0-171-88
+ip-26-0-171-102
+ip-26-0-171-168
+ip-26-0-171-230
+ip-26-0-171-249
+ip-26-0-172-57
+ip-26-0-172-73
+ip-26-0-172-116
+ip-26-0-172-142
+ip-26-0-172-147
+ip-26-0-172-252
+ip-26-0-173-7
+ip-26-0-173-121
+ip-26-0-173-202'
+All nodes: ip-26-0-171-56
+ip-26-0-171-62
+ip-26-0-171-88
+ip-26-0-171-102
+ip-26-0-171-168
+ip-26-0-171-230
+ip-26-0-171-249
+ip-26-0-172-57
+ip-26-0-172-73
+ip-26-0-172-116
+ip-26-0-172-142
+ip-26-0-172-147
+ip-26-0-172-252
+ip-26-0-173-7
+ip-26-0-173-121
+ip-26-0-173-202
++ echo 'World size: 128'
+World size: 128
++ '[' -n '' ']'
++ srun --wait=0 --kill-on-bad-exit=1 torchrun --nnodes=16 --nproc_per_node=8 --rdzv_id=14019149 --rdzv_backend=c10d --rdzv_endpoint=ip-26-0-171-56:12356 --max_restarts 0 --rdzv_conf timeout=60 /fsx/nouamane/projects/nanotron/run_train.py --config-file benchmark/configs/config_80G_dp16_tp4_pp2_acc16_mbs1_seq4096_zero0_tpmodeRED_vocab131k.yaml
+[2025-01-03 03:59:38,856] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 03:59:38,856] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 03:59:38,857] torch.distributed.run: [WARNING] 
+[2025-01-03 03:59:38,857] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 03:59:38,857] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 03:59:38,857] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 03:59:38,857] torch.distributed.run: [WARNING] 
+[2025-01-03 03:59:38,857] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 03:59:38,857] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 03:59:38,857] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 03:59:38,857] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 03:59:38,857] torch.distributed.run: [WARNING] 
+[2025-01-03 03:59:38,857] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 03:59:38,857] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 03:59:38,857] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 03:59:38,858] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 03:59:38,858] torch.distributed.run: [WARNING] 
+[2025-01-03 03:59:38,858] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 03:59:38,858] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 03:59:38,858] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 03:59:38,875] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 03:59:38,880] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 03:59:38,880] torch.distributed.run: [WARNING] 
+[2025-01-03 03:59:38,880] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 03:59:38,880] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 03:59:38,880] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 03:59:38,875] torch.distributed.run: [WARNING] 
+[2025-01-03 03:59:38,875] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 03:59:38,875] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 03:59:38,875] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 03:59:38,941] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 03:59:38,941] torch.distributed.run: [WARNING] 
+[2025-01-03 03:59:38,941] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 03:59:38,941] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 03:59:38,941] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 03:59:38,970] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 03:59:38,970] torch.distributed.run: [WARNING] 
+[2025-01-03 03:59:38,970] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 03:59:38,970] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 03:59:38,970] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 03:59:38,978] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 03:59:38,978] torch.distributed.run: [WARNING] 
+[2025-01-03 03:59:38,978] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 03:59:38,978] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 03:59:38,978] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 03:59:38,982] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 03:59:38,983] torch.distributed.run: [WARNING] 
+[2025-01-03 03:59:38,983] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 03:59:38,983] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 03:59:38,983] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 03:59:38,989] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 03:59:38,989] torch.distributed.run: [WARNING] 
+[2025-01-03 03:59:38,989] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 03:59:38,989] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 03:59:38,989] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 03:59:39,011] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 03:59:39,011] torch.distributed.run: [WARNING] 
+[2025-01-03 03:59:39,011] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 03:59:39,011] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 03:59:39,011] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 03:59:39,064] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 03:59:39,064] torch.distributed.run: [WARNING] 
+[2025-01-03 03:59:39,064] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 03:59:39,064] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 03:59:39,064] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 03:59:39,104] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 03:59:39,104] torch.distributed.run: [WARNING] 
+[2025-01-03 03:59:39,104] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 03:59:39,104] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 03:59:39,104] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 03:59:39,205] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 03:59:39,205] torch.distributed.run: [WARNING] 
+[2025-01-03 03:59:39,205] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 03:59:39,205] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 03:59:39,205] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 03:59:39,227] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 03:59:39,227] torch.distributed.run: [WARNING] 
+[2025-01-03 03:59:39,227] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 03:59:39,227] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 03:59:39,227] torch.distributed.run: [WARNING] *****************************************
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+01/03/2025 04:00:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]: Config:
+01/03/2025 04:00:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]: Config(general=GeneralArgs(project='debug',
+01/03/2025 04:00:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                            run='80G_dp16_tp4_pp2_acc16_mbs1_seq4096_zero0_tpmodeRED_vocab131k',
+01/03/2025 04:00:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                            seed=42,
+01/03/2025 04:00:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                            step=None,
+01/03/2025 04:00:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                            consumed_train_samples=None,
+01/03/2025 04:00:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                            benchmark_csv_path=PosixPath('benchmark/results/bench_final2.csv'),
+01/03/2025 04:00:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                            ignore_sanity_checks=True),
+01/03/2025 04:00:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:        parallelism=ParallelismArgs(dp=16,
+01/03/2025 04:00:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                                    pp=2,
+01/03/2025 04:00:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                                    tp=4,
+01/03/2025 04:00:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                                    pp_engine=<nanotron.parallel.pipeline_parallel.engine.OneForwardOneBackwardPipelineEngine object at 0x7f9664e84dc0>,
+01/03/2025 04:00:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                                    tp_mode=<TensorParallelLinearMode.REDUCE_SCATTER: 2>,
+01/03/2025 04:00:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                                    tp_linear_async_communication=True,
+01/03/2025 04:00:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                                    recompute_layer=False,
+01/03/2025 04:00:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                                    tp_recompute_allgather=True,
+01/03/2025 04:00:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                                    expert_parallel_size=1),
+01/03/2025 04:00:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:        model=ModelArgs(model_config=LlamaConfig(bos_token_id=0,
+01/03/2025 04:00:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                                                 eos_token_id=0,
+01/03/2025 04:00:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                                                 hidden_act='silu',
+01/03/2025 04:00:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                                                 hidden_size=8192,
+01/03/2025 04:00:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                                                 initializer_range=0.02,
+01/03/2025 04:00:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                                                 intermediate_size=28672,
+01/03/2025 04:00:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                                                 is_llama_config=True,
+01/03/2025 04:00:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                                                 max_position_embeddings=4096,
+01/03/2025 04:00:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                                                 num_attention_heads=64,
+01/03/2025 04:00:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                                                 num_hidden_layers=80,
+01/03/2025 04:00:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                                                 num_key_value_heads=64,
+01/03/2025 04:00:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                                                 pad_token_id=None,
+01/03/2025 04:00:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                                                 pretraining_tp=1,
+01/03/2025 04:00:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                                                 rms_norm_eps=1e-05,
+01/03/2025 04:00:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                                                 rope_scaling=None,
+01/03/2025 04:00:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                                                 rope_theta=10000.0,
+01/03/2025 04:00:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                                                 rope_interleaved=False,
+01/03/2025 04:00:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                                                 tie_word_embeddings=False,
+01/03/2025 04:00:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                                                 use_cache=True,
+01/03/2025 04:00:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                                                 vocab_size=131072),
+01/03/2025 04:00:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                        init_method=RandomInit(std=0.02),
+01/03/2025 04:00:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                        dtype=torch.bfloat16,
+01/03/2025 04:00:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                        make_vocab_size_divisible_by=1,
+01/03/2025 04:00:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                        ddp_bucket_cap_mb=25),
+01/03/2025 04:00:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:        tokenizer=TokenizerArgs(tokenizer_name_or_path='robot-test/dummy-tokenizer-wordlevel',
+01/03/2025 04:00:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                                tokenizer_revision=None,
+01/03/2025 04:00:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                                tokenizer_max_length=None),
+01/03/2025 04:00:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:        checkpoints=CheckpointsArgs(checkpoints_path=PosixPath('checkpoints'),
+01/03/2025 04:00:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                                    checkpoint_interval=10000,
+01/03/2025 04:00:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                                    save_initial_state=False,
+01/03/2025 04:00:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                                    save_final_state=False,
+01/03/2025 04:00:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                                    resume_checkpoint_path=None,
+01/03/2025 04:00:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                                    checkpoints_path_is_shared_file_system=False),
+01/03/2025 04:00:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:        logging=LoggingArgs(log_level='info',
+01/03/2025 04:00:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                            log_level_replica='info',
+01/03/2025 04:00:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                            iteration_step_info_interval=1),
+01/03/2025 04:00:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:        tokens=TokensArgs(sequence_length=4096,
+01/03/2025 04:00:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                          train_steps=100,
+01/03/2025 04:00:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                          micro_batch_size=1,
+01/03/2025 04:00:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                          batch_accumulation_per_replica=16,
+01/03/2025 04:00:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                          val_check_interval=100,
+01/03/2025 04:00:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                          limit_val_batches=0,
+01/03/2025 04:00:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                          limit_test_batches=0),
+01/03/2025 04:00:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:        optimizer=OptimizerArgs(optimizer_factory=AdamWOptimizerArgs(adam_eps=1e-08,
+01/03/2025 04:00:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                                                                     adam_beta1=0.9,
+01/03/2025 04:00:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                                                                     adam_beta2=0.95,
+01/03/2025 04:00:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                                                                     torch_adam_is_fused=True,
+01/03/2025 04:00:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                                                                     name='adamW'),
+01/03/2025 04:00:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                                zero_stage=0,
+01/03/2025 04:00:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                                weight_decay=0.01,
+01/03/2025 04:00:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                                clip_grad=1.0,
+01/03/2025 04:00:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                                accumulate_grad_in_fp32=True,
+01/03/2025 04:00:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                                learning_rate_scheduler=LRSchedulerArgs(learning_rate=0.0003,
+01/03/2025 04:00:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                                                                        lr_warmup_steps=2,
+01/03/2025 04:00:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                                                                        lr_warmup_style='linear',
+01/03/2025 04:00:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                                                                        lr_decay_style='cosine',
+01/03/2025 04:00:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                                                                        lr_decay_steps=13,
+01/03/2025 04:00:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                                                                        lr_decay_starting_step=None,
+01/03/2025 04:00:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                                                                        min_decay_lr=1e-05)),
+01/03/2025 04:00:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:        data_stages=[DatasetStageArgs(name='Stable Training Stage',
+01/03/2025 04:00:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                                      start_training_step=1,
+01/03/2025 04:00:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                                      data=DataArgs(dataset=None,
+01/03/2025 04:00:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                                                    seed=42,
+01/03/2025 04:00:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                                                    num_loading_workers=1))],
+01/03/2025 04:00:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:        profiler=None,
+01/03/2025 04:00:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:        lighteval=None,
+01/03/2025 04:00:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:        s3_upload=None)
+01/03/2025 04:00:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]: Model Config:
+01/03/2025 04:00:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]: LlamaConfig(bos_token_id=0,
+01/03/2025 04:00:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:             eos_token_id=0,
+01/03/2025 04:00:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:             hidden_act='silu',
+01/03/2025 04:00:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:             hidden_size=8192,
+01/03/2025 04:00:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:             initializer_range=0.02,
+01/03/2025 04:00:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:             intermediate_size=28672,
+01/03/2025 04:00:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:             is_llama_config=True,
+01/03/2025 04:00:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:             max_position_embeddings=4096,
+01/03/2025 04:00:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:             num_attention_heads=64,
+01/03/2025 04:00:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:             num_hidden_layers=80,
+01/03/2025 04:00:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:             num_key_value_heads=64,
+01/03/2025 04:00:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:             pad_token_id=None,
+01/03/2025 04:00:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:             pretraining_tp=1,
+01/03/2025 04:00:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:             rms_norm_eps=1e-05,
+01/03/2025 04:00:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:             rope_scaling=None,
+01/03/2025 04:00:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:             rope_theta=10000.0,
+01/03/2025 04:00:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:             rope_interleaved=False,
+01/03/2025 04:00:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:             tie_word_embeddings=False,
+01/03/2025 04:00:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:             use_cache=True,
+01/03/2025 04:00:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:             vocab_size=131072)
+01/03/2025 04:00:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]: Building model..
+01/03/2025 04:00:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]: Initialize RoPE Theta = 10000.0
+01/03/2025 04:00:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]: Setting PP block ranks...
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+01/03/2025 04:00:31 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]: Total number of parameters: 80G (152586.06MiB)
+01/03/2025 04:00:31 [INFO|DP=0|PP=1|TP=0|ip-26-0-172-142]: Local number of parameters: 9.76G (18609.23MiB)
+01/03/2025 04:00:31 [INFO|DP=0|PP=0|TP=1|ip-26-0-171-102]: Local number of parameters: 10.2G (19537.28MiB)
+01/03/2025 04:00:31 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]: Local number of parameters: 10.2G (19537.28MiB)
+01/03/2025 04:00:31 [INFO|DP=0|PP=1|TP=1|ip-26-0-172-142]: Local number of parameters: 9.76G (18609.23MiB)
+01/03/2025 04:00:31 [INFO|DP=0|PP=0|TP=3|ip-26-0-171-102]: Local number of parameters: 10.2G (19537.28MiB)
+01/03/2025 04:00:31 [INFO|DP=0|PP=0|TP=1|ip-26-0-171-102]: [After model building] Memory usage: 19537.33MiB. Peak allocated: 19537.38MiB Peak reserved: 19670.00MiB
+01/03/2025 04:00:31 [INFO|DP=0|PP=1|TP=3|ip-26-0-172-142]: Local number of parameters: 9.76G (18609.23MiB)
+01/03/2025 04:00:31 [INFO|DP=0|PP=1|TP=0|ip-26-0-172-142]: [After model building] Memory usage: 18609.28MiB. Peak allocated: 18609.33MiB Peak reserved: 18726.00MiB
+01/03/2025 04:00:31 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]: [After model building] Memory usage: 19537.33MiB. Peak allocated: 19537.38MiB Peak reserved: 19670.00MiB
+01/03/2025 04:00:31 [INFO|DP=0|PP=1|TP=1|ip-26-0-172-142]: [After model building] Memory usage: 18609.28MiB. Peak allocated: 18609.33MiB Peak reserved: 18726.00MiB
+01/03/2025 04:00:31 [INFO|DP=0|PP=0|TP=3|ip-26-0-171-102]: [After model building] Memory usage: 19537.33MiB. Peak allocated: 19537.38MiB Peak reserved: 19670.00MiB
+01/03/2025 04:00:31 [INFO|DP=0|PP=1|TP=3|ip-26-0-172-142]: [After model building] Memory usage: 18609.28MiB. Peak allocated: 18609.33MiB Peak reserved: 18726.00MiB
+01/03/2025 04:00:31 [INFO|DP=0|PP=0|TP=2|ip-26-0-171-102]: Local number of parameters: 10.2G (19537.28MiB)
+01/03/2025 04:00:31 [INFO|DP=0|PP=1|TP=2|ip-26-0-172-142]: Local number of parameters: 9.76G (18609.23MiB)
+01/03/2025 04:00:31 [INFO|DP=0|PP=0|TP=2|ip-26-0-171-102]: [After model building] Memory usage: 19537.33MiB. Peak allocated: 19537.38MiB Peak reserved: 19670.00MiB
+01/03/2025 04:00:31 [INFO|DP=0|PP=1|TP=2|ip-26-0-172-142]: [After model building] Memory usage: 18609.28MiB. Peak allocated: 18609.33MiB Peak reserved: 18726.00MiB
+[E ProcessGroupGloo.cpp:138] Rank 1 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+[E ProcessGroupGloo.cpp:138] Rank 1 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+[E ProcessGroupGloo.cpp:138] Rank 1 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+[E ProcessGroupGloo.cpp:138] Rank 1 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+[E ProcessGroupGloo.cpp:138] Rank 1 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+[E ProcessGroupGloo.cpp:138] Rank 1 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+[E ProcessGroupGloo.cpp:138] Rank 1 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+[E ProcessGroupGloo.cpp:138] Rank 1 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+Traceback (most recent call last):
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 171, in __init__
+    trainer = DistributedTrainer(config_file)
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 171, in __init__
+    trainer = DistributedTrainer(config_file)  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 171, in __init__
+
+        trainer = DistributedTrainer(config_file)    trainer = DistributedTrainer(config_file)  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 171, in __init__
+
+trainer = DistributedTrainer(config_file)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 171, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 171, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 171, in __init__
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 171, in __init__
+    self.model = self.init_model()  # Defines self.model    
+self.model = self.init_model()  # Defines self.model
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 720, in init_model
+self.model = self.init_model()  # Defines self.model
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 720, in init_model
+    self.model = self.init_model()  # Defines self.model    
+self.model = self.init_model()  # Defines self.model  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 720, in init_model
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 720, in init_model
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 720, in init_model
+    self.model = self.init_model()  # Defines self.model    
+self.model = self.init_model()  # Defines self.model
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 720, in init_model
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 720, in init_model
+    self.model = self.init_model()  # Defines self.model
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 720, in init_model
+    model = self._init_model_instance()
+                    model = self._init_model_instance()  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 730, in _init_model_instance
+
+    model = self._init_model_instance()model = self._init_model_instance()model = self._init_model_instance()model = self._init_model_instance()model = self._init_model_instance()
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 730, in _init_model_instance
+
+
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 730, in _init_model_instance
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 730, in _init_model_instance
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 730, in _init_model_instance
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 730, in _init_model_instance
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 730, in _init_model_instance
+    model = self._init_model_instance()
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 730, in _init_model_instance
+    model = self._init_model(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 826, in _init_model
+    model = self._init_model(
+    model = self._init_model(  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 826, in _init_model
+
+    model = self._init_model(  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 826, in _init_model
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 826, in _init_model
+    model = self._init_model(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 826, in _init_model
+    model = self._init_model(
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 826, in _init_model
+model = self._init_model(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 826, in _init_model
+    model = self._init_model(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 826, in _init_model
+    dist.all_reduce(total_params, group=parallel_context.pp_pg, async_op=False, op=dist.ReduceOp.SUM)  # PP
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/c10d_logger.py", line 47, in wrapper
+    dist.all_reduce(total_params, group=parallel_context.pp_pg, async_op=False, op=dist.ReduceOp.SUM)  # PP
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/c10d_logger.py", line 47, in wrapper
+dist.all_reduce(total_params, group=parallel_context.pp_pg, async_op=False, op=dist.ReduceOp.SUM)  # PP    
+dist.all_reduce(total_params, group=parallel_context.pp_pg, async_op=False, op=dist.ReduceOp.SUM)  # PP
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/c10d_logger.py", line 47, in wrapper
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/c10d_logger.py", line 47, in wrapper
+    dist.all_reduce(total_params, group=parallel_context.pp_pg, async_op=False, op=dist.ReduceOp.SUM)  # PP
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/c10d_logger.py", line 47, in wrapper
+    dist.all_reduce(total_params, group=parallel_context.pp_pg, async_op=False, op=dist.ReduceOp.SUM)  # PP    
+dist.all_reduce(total_params, group=parallel_context.pp_pg, async_op=False, op=dist.ReduceOp.SUM)  # PP
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/c10d_logger.py", line 47, in wrapper
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/c10d_logger.py", line 47, in wrapper
+    dist.all_reduce(total_params, group=parallel_context.pp_pg, async_op=False, op=dist.ReduceOp.SUM)  # PP
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/c10d_logger.py", line 47, in wrapper
+    return func(*args, **kwargs)
+    return func(*args, **kwargs)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/distributed_c10d.py", line 2050, in all_reduce
+return func(*args, **kwargs)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/distributed_c10d.py", line 2050, in all_reduce
+return func(*args, **kwargs)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/distributed_c10d.py", line 2050, in all_reduce
+
+    return func(*args, **kwargs)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/distributed_c10d.py", line 2050, in all_reduce
+
+        return func(*args, **kwargs)return func(*args, **kwargs)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/distributed_c10d.py", line 2050, in all_reduce
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/distributed_c10d.py", line 2050, in all_reduce
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/distributed_c10d.py", line 2050, in all_reduce
+    return func(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/distributed_c10d.py", line 2050, in all_reduce
+        work = group.allreduce([tensor], opts)    work = group.allreduce([tensor], opts)    
+work = group.allreduce([tensor], opts)
+work = group.allreduce([tensor], opts)            
+work = group.allreduce([tensor], opts)
+work = group.allreduce([tensor], opts)work = group.allreduce([tensor], opts)
+
+
+RuntimeErrorRuntimeError: RuntimeError: : RuntimeErrorRank 1 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+ Original exception: 
+[../third_party/gloo/gloo/transport/tcp/unbound_buffer.cc:133] Timed out waiting 1200000ms for send operation to complete: Rank 1 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+ Original exception: 
+[../third_party/gloo/gloo/transport/tcp/unbound_buffer.cc:133] Timed out waiting 1200000ms for send operation to complete
+RuntimeErrorRuntimeErrorRuntimeErrorRank 1 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+ Original exception: 
+[../third_party/gloo/gloo/transport/tcp/unbound_buffer.cc:133] Timed out waiting 1200000ms for send operation to complete: 
+Rank 1 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+ Original exception: 
+[../third_party/gloo/gloo/transport/tcp/unbound_buffer.cc:133] Timed out waiting 1200000ms for send operation to complete: : 
+Rank 1 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+ Original exception: 
+[../third_party/gloo/gloo/transport/tcp/unbound_buffer.cc:133] Timed out waiting 1200000ms for send operation to complete
+Rank 1 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+ Original exception: 
+[../third_party/gloo/gloo/transport/tcp/unbound_buffer.cc:133] Timed out waiting 1200000ms for send operation to completeRank 1 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+ Original exception: 
+[../third_party/gloo/gloo/transport/tcp/unbound_buffer.cc:133] Timed out waiting 1200000ms for send operation to complete
+
+
+    work = group.allreduce([tensor], opts)
+RuntimeError: Rank 1 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+ Original exception: 
+[../third_party/gloo/gloo/transport/tcp/unbound_buffer.cc:133] Timed out waiting 1200000ms for send operation to complete
+[E ProcessGroupGloo.cpp:138] Rank 1 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+[E ProcessGroupGloo.cpp:138] Rank 1 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+[E ProcessGroupGloo.cpp:138] Rank 1 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+[E ProcessGroupGloo.cpp:138] Rank 1 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+[E ProcessGroupGloo.cpp:138] Rank 1 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+[E ProcessGroupGloo.cpp:138] Rank 1 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+[E ProcessGroupGloo.cpp:138] Rank 1 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+[E ProcessGroupGloo.cpp:138] Rank 1 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+[E ProcessGroupGloo.cpp:138] Rank 5 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+[E ProcessGroupGloo.cpp:138] Rank 5 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+[E ProcessGroupGloo.cpp:138] Rank 5 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+[E ProcessGroupGloo.cpp:138] Rank 5 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+[E ProcessGroupGloo.cpp:138] Rank 5 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+[E ProcessGroupGloo.cpp:138] Rank 5 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+[E ProcessGroupGloo.cpp:138] Rank 5 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+[E ProcessGroupGloo.cpp:138] Rank 5 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+[E ProcessGroupGloo.cpp:138] Rank 6 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+[E ProcessGroupGloo.cpp:138] Rank 7 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+[E ProcessGroupGloo.cpp:138] Rank 7 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+[E ProcessGroupGloo.cpp:138] Rank 7 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+[E ProcessGroupGloo.cpp:138] Rank 7 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+[E ProcessGroupGloo.cpp:138] Rank 6 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+[E ProcessGroupGloo.cpp:138] Rank 6 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+[E ProcessGroupGloo.cpp:138] Rank 6 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+[E ProcessGroupGloo.cpp:138] Rank 7 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+[E ProcessGroupGloo.cpp:138] Rank 7 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+[E ProcessGroupGloo.cpp:138] Rank 6 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+[E ProcessGroupGloo.cpp:138] Rank 7 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+[E ProcessGroupGloo.cpp:138] Rank 6 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+[E ProcessGroupGloo.cpp:138] Rank 6 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+[E ProcessGroupGloo.cpp:138] Rank 6 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+[E ProcessGroupGloo.cpp:138] Rank 7 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+[E ProcessGroupGloo.cpp:138] Rank 11 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+[E ProcessGroupGloo.cpp:138] Rank 11 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+[E ProcessGroupGloo.cpp:138] Rank 11 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+[E ProcessGroupGloo.cpp:138] Rank 11 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+[E ProcessGroupGloo.cpp:138] Rank 11 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+[E ProcessGroupGloo.cpp:138] Rank 11 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+[E ProcessGroupGloo.cpp:138] Rank 11 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+[E ProcessGroupGloo.cpp:138] Rank 11 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+[E ProcessGroupGloo.cpp:138] Rank 10 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+[E ProcessGroupGloo.cpp:138] Rank 4 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+[E ProcessGroupGloo.cpp:138] Rank 10 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+[E ProcessGroupGloo.cpp:138] Rank 4 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+[E ProcessGroupGloo.cpp:138] Rank 4 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+[E ProcessGroupGloo.cpp:138] Rank 10 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+[E ProcessGroupGloo.cpp:138] Rank 4 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+[E ProcessGroupGloo.cpp:138] Rank 10 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+[E ProcessGroupGloo.cpp:138] Rank 10 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+[E ProcessGroupGloo.cpp:138] Rank 10 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+[E ProcessGroupGloo.cpp:138] Rank 10 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+[E ProcessGroupGloo.cpp:138] Rank 10 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+[E ProcessGroupGloo.cpp:138] Rank 15 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+[E ProcessGroupGloo.cpp:138] Rank 14 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+[E ProcessGroupGloo.cpp:138] Rank 15 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+[E ProcessGroupGloo.cpp:138] Rank 14 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+[E ProcessGroupGloo.cpp:138] Rank 14 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+[E ProcessGroupGloo.cpp:138] Rank 15 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+[E ProcessGroupGloo.cpp:138] Rank 15 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+[E ProcessGroupGloo.cpp:138] Rank 14 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+[E ProcessGroupGloo.cpp:138] Rank 15 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+[E ProcessGroupGloo.cpp:138] Rank 15 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+[E ProcessGroupGloo.cpp:138] Rank 15 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+[E ProcessGroupGloo.cpp:138] Rank 15 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+[E ProcessGroupGloo.cpp:138] Rank 14 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+[E ProcessGroupGloo.cpp:138] Rank 14 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+[E ProcessGroupGloo.cpp:138] Rank 14 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+[E ProcessGroupGloo.cpp:138] Rank 14 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+[E ProcessGroupGloo.cpp:138] Rank 4 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+Traceback (most recent call last):
+Traceback (most recent call last):
+Traceback (most recent call last):
+[E ProcessGroupGloo.cpp:138] Rank 4 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+[E ProcessGroupGloo.cpp:138] Rank 4 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+[E ProcessGroupGloo.cpp:138] Rank 4 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 171, in __init__
+    trainer = DistributedTrainer(config_file)    trainer = DistributedTrainer(config_file)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 171, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 171, in __init__
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 171, in __init__
+    self.model = self.init_model()  # Defines self.model
+        self.model = self.init_model()  # Defines self.modelself.model = self.init_model()  # Defines self.model
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 720, in init_model
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 720, in init_model
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 720, in init_model
+    self.model = self.init_model()  # Defines self.model
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 720, in init_model
+    model = self._init_model_instance()
+        model = self._init_model_instance()  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 730, in _init_model_instance
+model = self._init_model_instance()
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 730, in _init_model_instance
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 730, in _init_model_instance
+    model = self._init_model_instance()
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 730, in _init_model_instance
+    model = self._init_model(
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 862, in _init_model
+model = self._init_model(    
+model = self._init_model(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 862, in _init_model
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 862, in _init_model
+    model = self._init_model(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 862, in _init_model
+    model = DistributedDataParallel(
+    model = DistributedDataParallel(  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 795, in __init__
+
+    model = DistributedDataParallel(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 795, in __init__
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 795, in __init__
+    model = DistributedDataParallel(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 795, in __init__
+    _verify_param_shape_across_processes(self.process_group, parameters)
+          File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/utils.py", line 265, in _verify_param_shape_across_processes
+_verify_param_shape_across_processes(self.process_group, parameters)_verify_param_shape_across_processes(self.process_group, parameters)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/utils.py", line 265, in _verify_param_shape_across_processes
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/utils.py", line 265, in _verify_param_shape_across_processes
+    _verify_param_shape_across_processes(self.process_group, parameters)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/utils.py", line 265, in _verify_param_shape_across_processes
+    return dist._verify_params_across_processes(process_group, tensors, logger)
+        return dist._verify_params_across_processes(process_group, tensors, logger)return dist._verify_params_across_processes(process_group, tensors, logger)
+
+RuntimeErrorRuntimeError: : Rank 1 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+ Original exception: 
+[../third_party/gloo/gloo/transport/tcp/unbound_buffer.cc:81] Timed out waiting 1200000ms for recv operation to completeRank 1 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+ Original exception: 
+[../third_party/gloo/gloo/transport/tcp/unbound_buffer.cc:81] Timed out waiting 1200000ms for recv operation to complete
+
+RuntimeError: Rank 1 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+ Original exception: 
+[../third_party/gloo/gloo/transport/tcp/unbound_buffer.cc:81] Timed out waiting 1200000ms for recv operation to complete
+    return dist._verify_params_across_processes(process_group, tensors, logger)
+RuntimeError: Rank 1 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+ Original exception: 
+[../third_party/gloo/gloo/transport/tcp/unbound_buffer.cc:81] Timed out waiting 1200000ms for recv operation to complete
+Traceback (most recent call last):
+Traceback (most recent call last):
+Traceback (most recent call last):
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+            trainer = DistributedTrainer(config_file)trainer = DistributedTrainer(config_file)trainer = DistributedTrainer(config_file)
+
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 171, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 171, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 171, in __init__
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 171, in __init__
+    self.model = self.init_model()  # Defines self.model
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 720, in init_model
+        self.model = self.init_model()  # Defines self.modelself.model = self.init_model()  # Defines self.model
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 720, in init_model
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 720, in init_model
+    self.model = self.init_model()  # Defines self.model
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 720, in init_model
+            model = self._init_model_instance()model = self._init_model_instance()model = self._init_model_instance()
+
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 730, in _init_model_instance
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 730, in _init_model_instance
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 730, in _init_model_instance
+    model = self._init_model_instance()
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 730, in _init_model_instance
+    model = self._init_model(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 862, in _init_model
+    model = self._init_model(
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 862, in _init_model
+model = self._init_model(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 862, in _init_model
+    model = self._init_model(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 862, in _init_model
+    model = DistributedDataParallel(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 795, in __init__
+    model = DistributedDataParallel(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 795, in __init__
+    model = DistributedDataParallel(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 795, in __init__
+    model = DistributedDataParallel(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 795, in __init__
+    _verify_param_shape_across_processes(self.process_group, parameters)
+        _verify_param_shape_across_processes(self.process_group, parameters)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/utils.py", line 265, in _verify_param_shape_across_processes
+_verify_param_shape_across_processes(self.process_group, parameters)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/utils.py", line 265, in _verify_param_shape_across_processes
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/utils.py", line 265, in _verify_param_shape_across_processes
+    _verify_param_shape_across_processes(self.process_group, parameters)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/utils.py", line 265, in _verify_param_shape_across_processes
+            return dist._verify_params_across_processes(process_group, tensors, logger)return dist._verify_params_across_processes(process_group, tensors, logger)return dist._verify_params_across_processes(process_group, tensors, logger)
+
+
+RuntimeError: RuntimeErrorRank 1 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+ Original exception: 
+[../third_party/gloo/gloo/transport/tcp/unbound_buffer.cc:81] Timed out waiting 1200000ms for recv operation to completeRuntimeError
+: : Rank 1 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+ Original exception: 
+[../third_party/gloo/gloo/transport/tcp/unbound_buffer.cc:81] Timed out waiting 1200000ms for recv operation to completeRank 1 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+ Original exception: 
+[../third_party/gloo/gloo/transport/tcp/unbound_buffer.cc:81] Timed out waiting 1200000ms for recv operation to complete
+
+    return dist._verify_params_across_processes(process_group, tensors, logger)
+RuntimeError: Rank 1 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+ Original exception: 
+[../third_party/gloo/gloo/transport/tcp/unbound_buffer.cc:81] Timed out waiting 1200000ms for recv operation to complete
+[E ProcessGroupGloo.cpp:138] Rank 9 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+[E ProcessGroupGloo.cpp:138] Rank 8 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+[E ProcessGroupGloo.cpp:138] Rank 9 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+[E ProcessGroupGloo.cpp:138] Rank 8 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+[E ProcessGroupGloo.cpp:138] Rank 8 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+[E ProcessGroupGloo.cpp:138] Rank 9 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+[E ProcessGroupGloo.cpp:138] Rank 8 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+[E ProcessGroupGloo.cpp:138] Rank 9 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+[E ProcessGroupGloo.cpp:138] Rank 2 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+[E ProcessGroupGloo.cpp:138] Rank 2 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+[E ProcessGroupGloo.cpp:138] Rank 2 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+[E ProcessGroupGloo.cpp:138] Rank 2 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+[E ProcessGroupGloo.cpp:138] Rank 9 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+[E ProcessGroupGloo.cpp:138] Rank 9 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+[E ProcessGroupGloo.cpp:138] Rank 8 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+[E ProcessGroupGloo.cpp:138] Rank 9 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+[E ProcessGroupGloo.cpp:138] Rank 8 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+[E ProcessGroupGloo.cpp:138] Rank 3 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+[E ProcessGroupGloo.cpp:138] Rank 8 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+[E ProcessGroupGloo.cpp:138] Rank 9 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+[E ProcessGroupGloo.cpp:138] Rank 8 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+[E ProcessGroupGloo.cpp:2810] [Rank 0]: Rank 12 failed to pass monitoredBarrier in 1200000 ms
+[E ProcessGroupGloo.cpp:138] Rank 3 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+[E ProcessGroupGloo.cpp:2810] [Rank 0]: Rank 12 failed to pass monitoredBarrier in 1200000 ms
+[E ProcessGroupGloo.cpp:2810] [Rank 0]: Rank 13 failed to pass monitoredBarrier in 1200000 ms
+[E ProcessGroupGloo.cpp:2810] [Rank 0]: Rank 13 failed to pass monitoredBarrier in 1200000 ms
+[E ProcessGroupGloo.cpp:138] [Rank 0]: Ranks 12, 13 failed to pass monitoredBarrier in 1200000 ms
+[E ProcessGroupGloo.cpp:138] [Rank 0]: Ranks 12, 13 failed to pass monitoredBarrier in 1200000 ms
+[E ProcessGroupGloo.cpp:2810] [Rank 0]: Rank 12 failed to pass monitoredBarrier in 1200000 ms
+[E ProcessGroupGloo.cpp:138] Rank 3 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+[E ProcessGroupGloo.cpp:2810] [Rank 0]: Rank 12 failed to pass monitoredBarrier in 1200000 ms
+[E ProcessGroupGloo.cpp:2810] [Rank 0]: Rank 13 failed to pass monitoredBarrier in 1200000 ms
+[E ProcessGroupGloo.cpp:138] [Rank 0]: Ranks 12, 13 failed to pass monitoredBarrier in 1200000 ms
+[E ProcessGroupGloo.cpp:2810] [Rank 0]: Rank 13 failed to pass monitoredBarrier in 1200000 ms
+[E ProcessGroupGloo.cpp:2810] [Rank 0]: Rank 12 failed to pass monitoredBarrier in 1200000 ms
+[E ProcessGroupGloo.cpp:2810] [Rank 0]: Rank 13 failed to pass monitoredBarrier in 1200000 ms
+[E ProcessGroupGloo.cpp:138] [Rank 0]: Ranks 12, 13 failed to pass monitoredBarrier in 1200000 ms
+[E ProcessGroupGloo.cpp:138] [Rank 0]: Ranks 12, 13 failed to pass monitoredBarrier in 1200000 ms
+[E ProcessGroupGloo.cpp:2810] [Rank 0]: Rank 12 failed to pass monitoredBarrier in 1200000 ms
+[E ProcessGroupGloo.cpp:2810] [Rank 0]: Rank 13 failed to pass monitoredBarrier in 1200000 ms
+[E ProcessGroupGloo.cpp:138] [Rank 0]: Ranks 12, 13 failed to pass monitoredBarrier in 1200000 ms
+[E ProcessGroupGloo.cpp:138] Rank 2 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+[E ProcessGroupGloo.cpp:138] Rank 3 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+[E ProcessGroupGloo.cpp:138] Rank 3 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+[E ProcessGroupGloo.cpp:138] Rank 2 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+[E ProcessGroupGloo.cpp:138] Rank 2 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+[E ProcessGroupGloo.cpp:138] Rank 2 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+[E ProcessGroupGloo.cpp:138] Rank 3 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+[E ProcessGroupGloo.cpp:138] Rank 3 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+[E ProcessGroupGloo.cpp:2810] [Rank 0]: Rank 12 failed to pass monitoredBarrier in 1200000 ms
+[E ProcessGroupGloo.cpp:2810] [Rank 0]: Rank 13 failed to pass monitoredBarrier in 1200000 ms
+[E ProcessGroupGloo.cpp:138] [Rank 0]: Ranks 12, 13 failed to pass monitoredBarrier in 1200000 ms
+[E ProcessGroupGloo.cpp:2810] [Rank 0]: Rank 12 failed to pass monitoredBarrier in 1200000 ms
+[E ProcessGroupGloo.cpp:2810] [Rank 0]: Rank 13 failed to pass monitoredBarrier in 1200000 ms
+[E ProcessGroupGloo.cpp:138] Rank 3 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+[E ProcessGroupGloo.cpp:138] [Rank 0]: Ranks 12, 13 failed to pass monitoredBarrier in 1200000 ms
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 171, in __init__
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 171, in __init__
+    self.model = self.init_model()  # Defines self.model
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 720, in init_model
+    self.model = self.init_model()  # Defines self.model
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 720, in init_model
+    model = self._init_model_instance()
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 730, in _init_model_instance
+    model = self._init_model_instance()
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 730, in _init_model_instance
+    model = self._init_model(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 862, in _init_model
+    model = self._init_model(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 862, in _init_model
+    model = DistributedDataParallel(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 795, in __init__
+    model = DistributedDataParallel(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 795, in __init__
+    _verify_param_shape_across_processes(self.process_group, parameters)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/utils.py", line 265, in _verify_param_shape_across_processes
+    _verify_param_shape_across_processes(self.process_group, parameters)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/utils.py", line 265, in _verify_param_shape_across_processes
+    return dist._verify_params_across_processes(process_group, tensors, logger)
+RuntimeError: [Rank 0]: Ranks 12, 13 failed to pass monitoredBarrier in 1200000 ms
+    return dist._verify_params_across_processes(process_group, tensors, logger)
+RuntimeError: [Rank 0]: Ranks 12, 13 failed to pass monitoredBarrier in 1200000 ms
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 171, in __init__
+    self.model = self.init_model()  # Defines self.model
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 720, in init_model
+    model = self._init_model_instance()
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 730, in _init_model_instance
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    model = self._init_model(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 862, in _init_model
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    trainer = DistributedTrainer(config_file)
+    model = DistributedDataParallel(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 171, in __init__
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 795, in __init__
+    self.model = self.init_model()  # Defines self.model
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 720, in init_model
+    _verify_param_shape_across_processes(self.process_group, parameters)
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/utils.py", line 265, in _verify_param_shape_across_processes
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 171, in __init__
+    model = self._init_model_instance()
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 730, in _init_model_instance
+    self.model = self.init_model()  # Defines self.model
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 720, in init_model
+    model = self._init_model(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 862, in _init_model
+    return dist._verify_params_across_processes(process_group, tensors, logger)
+    model = self._init_model_instance()
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 730, in _init_model_instance
+RuntimeError: [Rank 0]: Ranks 12, 13 failed to pass monitoredBarrier in 1200000 ms
+Traceback (most recent call last):
+    model = DistributedDataParallel(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 795, in __init__
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    model = self._init_model(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 862, in _init_model
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 171, in __init__
+    _verify_param_shape_across_processes(self.process_group, parameters)
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/utils.py", line 265, in _verify_param_shape_across_processes
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    model = DistributedDataParallel(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 795, in __init__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+Traceback (most recent call last):
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 171, in __init__
+Traceback (most recent call last):
+    self.model = self.init_model()  # Defines self.model
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 720, in init_model
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    return dist._verify_params_across_processes(process_group, tensors, logger)
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+RuntimeError: [Rank 0]: Ranks 12, 13 failed to pass monitoredBarrier in 1200000 ms
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    _verify_param_shape_across_processes(self.process_group, parameters)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/utils.py", line 265, in _verify_param_shape_across_processes
+    self.model = self.init_model()  # Defines self.model
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 720, in init_model
+    model = self._init_model_instance()
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 730, in _init_model_instance
+    return dist._verify_params_across_processes(process_group, tensors, logger)
+RuntimeError: [Rank 0]: Ranks 12, 13 failed to pass monitoredBarrier in 1200000 ms
+    model = self._init_model_instance()
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 730, in _init_model_instance
+    trainer = DistributedTrainer(config_file)
+    model = self._init_model(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 862, in _init_model
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 171, in __init__
+    model = self._init_model(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 862, in _init_model
+    self.model = self.init_model()  # Defines self.model
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 720, in init_model
+    model = DistributedDataParallel(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 795, in __init__
+                trainer = DistributedTrainer(config_file)trainer = DistributedTrainer(config_file)trainer = DistributedTrainer(config_file)
+
+    trainer = DistributedTrainer(config_file)          File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 171, in __init__
+trainer = DistributedTrainer(config_file)  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 171, in __init__
+
+trainer = DistributedTrainer(config_file)
+
+trainer = DistributedTrainer(config_file)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 171, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 171, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 171, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 171, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 171, in __init__
+    model = DistributedDataParallel(
+    model = self._init_model_instance()
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 730, in _init_model_instance
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 795, in __init__
+    _verify_param_shape_across_processes(self.process_group, parameters)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/utils.py", line 265, in _verify_param_shape_across_processes
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 171, in __init__
+    _verify_param_shape_across_processes(self.process_group, parameters)
+    model = self._init_model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/utils.py", line 265, in _verify_param_shape_across_processes
+    return dist._verify_params_across_processes(process_group, tensors, logger)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 862, in _init_model
+RuntimeError: [Rank 0]: Ranks 12, 13 failed to pass monitoredBarrier in 1200000 ms
+    return dist._verify_params_across_processes(process_group, tensors, logger)
+RuntimeError: [Rank 0]: Ranks 12, 13 failed to pass monitoredBarrier in 1200000 ms
+    model = DistributedDataParallel(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 795, in __init__
+    _verify_param_shape_across_processes(self.process_group, parameters)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/utils.py", line 265, in _verify_param_shape_across_processes
+            self.model = self.init_model()  # Defines self.modelself.model = self.init_model()  # Defines self.model    self.model = self.init_model()  # Defines self.model
+    
+    return dist._verify_params_across_processes(process_group, tensors, logger)
+        
+self.model = self.init_model()  # Defines self.modelself.model = self.init_model()  # Defines self.modelself.model = self.init_model()  # Defines self.modelself.model = self.init_model()  # Defines self.model  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 720, in init_model
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 720, in init_model
+
+
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 720, in init_model
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 720, in init_model
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 720, in init_model
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 720, in init_model
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 720, in init_model
+RuntimeError: [Rank 0]: Ranks 12, 13 failed to pass monitoredBarrier in 1200000 ms
+    self.model = self.init_model()  # Defines self.model
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 720, in init_model
+    model = self._init_model_instance()    
+model = self._init_model_instance()
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 730, in _init_model_instance
+model = self._init_model_instance()  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 730, in _init_model_instance
+    
+model = self._init_model_instance()
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 730, in _init_model_instance
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 730, in _init_model_instance
+    model = self._init_model_instance()
+        model = self._init_model_instance()  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 730, in _init_model_instance
+model = self._init_model_instance()
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 730, in _init_model_instance
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 730, in _init_model_instance
+    model = self._init_model_instance()
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 730, in _init_model_instance
+        model = self._init_model(model = self._init_model(        model = self._init_model(
+
+    model = self._init_model(
+    
+model = self._init_model(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 862, in _init_model
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 862, in _init_model
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 862, in _init_model
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 862, in _init_model
+model = self._init_model(  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 862, in _init_model
+
+    model = self._init_model(  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 862, in _init_model
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 862, in _init_model
+    model = self._init_model(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 862, in _init_model
+        model = DistributedDataParallel(    model = DistributedDataParallel(
+model = DistributedDataParallel(
+    
+model = DistributedDataParallel(            
+model = DistributedDataParallel(  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 795, in __init__
+model = DistributedDataParallel(  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 795, in __init__
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 795, in __init__
+model = DistributedDataParallel(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 795, in __init__
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 795, in __init__
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 795, in __init__
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 795, in __init__
+    model = DistributedDataParallel(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 795, in __init__
+                _verify_param_shape_across_processes(self.process_group, parameters)    _verify_param_shape_across_processes(self.process_group, parameters)_verify_param_shape_across_processes(self.process_group, parameters)_verify_param_shape_across_processes(self.process_group, parameters)
+    _verify_param_shape_across_processes(self.process_group, parameters)
+
+
+_verify_param_shape_across_processes(self.process_group, parameters)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/utils.py", line 265, in _verify_param_shape_across_processes
+    _verify_param_shape_across_processes(self.process_group, parameters)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/utils.py", line 265, in _verify_param_shape_across_processes
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/utils.py", line 265, in _verify_param_shape_across_processes
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/utils.py", line 265, in _verify_param_shape_across_processes
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/utils.py", line 265, in _verify_param_shape_across_processes
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/utils.py", line 265, in _verify_param_shape_across_processes
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/utils.py", line 265, in _verify_param_shape_across_processes
+    _verify_param_shape_across_processes(self.process_group, parameters)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/utils.py", line 265, in _verify_param_shape_across_processes
+    return dist._verify_params_across_processes(process_group, tensors, logger)
+    return dist._verify_params_across_processes(process_group, tensors, logger)
+    return dist._verify_params_across_processes(process_group, tensors, logger)
+    return dist._verify_params_across_processes(process_group, tensors, logger)RuntimeErrorRuntimeError
+:     Rank 4 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+ Original exception: 
+[../third_party/gloo/gloo/transport/tcp/unbound_buffer.cc:81] Timed out waiting 1200000ms for recv operation to completereturn dist._verify_params_across_processes(process_group, tensors, logger): 
+
+Rank 5 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+ Original exception: 
+[../third_party/gloo/gloo/transport/tcp/unbound_buffer.cc:81] Timed out waiting 1200000ms for recv operation to complete
+RuntimeError    : return dist._verify_params_across_processes(process_group, tensors, logger)
+    Rank 5 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+ Original exception: 
+[../third_party/gloo/gloo/transport/tcp/unbound_buffer.cc:81] Timed out waiting 1200000ms for recv operation to completereturn dist._verify_params_across_processes(process_group, tensors, logger)
+RuntimeError
+: RuntimeErrorRank 4 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+ Original exception: 
+[../third_party/gloo/gloo/transport/tcp/unbound_buffer.cc:81] Timed out waiting 1200000ms for recv operation to complete
+: RuntimeErrorRank 5 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+ Original exception: 
+[../third_party/gloo/gloo/transport/tcp/unbound_buffer.cc:81] Timed out waiting 1200000ms for recv operation to complete
+: Rank 5 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+ Original exception: 
+[../third_party/gloo/gloo/transport/tcp/unbound_buffer.cc:81] Timed out waiting 1200000ms for recv operation to complete
+RuntimeError: Rank 4 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+ Original exception: 
+[../third_party/gloo/gloo/transport/tcp/unbound_buffer.cc:81] Timed out waiting 1200000ms for recv operation to complete
+    return dist._verify_params_across_processes(process_group, tensors, logger)
+RuntimeError: Rank 4 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+ Original exception: 
+[../third_party/gloo/gloo/transport/tcp/unbound_buffer.cc:81] Timed out waiting 1200000ms for recv operation to complete
+Traceback (most recent call last):
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+Traceback (most recent call last):
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+Traceback (most recent call last):
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+        trainer = DistributedTrainer(config_file)trainer = DistributedTrainer(config_file)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 171, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 171, in __init__
+    trainer = DistributedTrainer(config_file)
+        trainer = DistributedTrainer(config_file)trainer = DistributedTrainer(config_file)  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 171, in __init__
+    
+
+trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 171, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 171, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 171, in __init__
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 171, in __init__
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 171, in __init__
+    self.model = self.init_model()  # Defines self.model
+        self.model = self.init_model()  # Defines self.modelself.model = self.init_model()  # Defines self.model    
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 720, in init_model
+self.model = self.init_model()  # Defines self.model
+
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 720, in init_model
+self.model = self.init_model()  # Defines self.model  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 720, in init_model
+
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 720, in init_model
+self.model = self.init_model()  # Defines self.model
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 720, in init_model
+    self.model = self.init_model()  # Defines self.model
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 720, in init_model
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 720, in init_model
+    self.model = self.init_model()  # Defines self.model
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 720, in init_model
+        model = self._init_model_instance()model = self._init_model_instance()    
+        
+    model = self._init_model_instance()      File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 730, in _init_model_instance
+model = self._init_model_instance()model = self._init_model_instance()model = self._init_model_instance()  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 730, in _init_model_instance
+
+model = self._init_model_instance()
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 730, in _init_model_instance
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 730, in _init_model_instance
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 730, in _init_model_instance
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 730, in _init_model_instance
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 730, in _init_model_instance
+    model = self._init_model_instance()
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 730, in _init_model_instance
+    model = self._init_model(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 862, in _init_model
+    model = self._init_model(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 862, in _init_model
+    model = self._init_model(    
+model = self._init_model(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 862, in _init_model
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 862, in _init_model
+    model = self._init_model(
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 862, in _init_model
+model = self._init_model(    
+model = self._init_model(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 862, in _init_model
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 862, in _init_model
+    model = self._init_model(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 862, in _init_model
+    model = DistributedDataParallel(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 795, in __init__
+    model = DistributedDataParallel(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 795, in __init__
+    model = DistributedDataParallel(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 795, in __init__
+    model = DistributedDataParallel(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 795, in __init__
+    model = DistributedDataParallel(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 795, in __init__
+        model = DistributedDataParallel(
+model = DistributedDataParallel(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 795, in __init__
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 795, in __init__
+    model = DistributedDataParallel(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 795, in __init__
+Traceback (most recent call last):
+Traceback (most recent call last):
+Traceback (most recent call last):
+Traceback (most recent call last):
+Traceback (most recent call last):
+Traceback (most recent call last):
+Traceback (most recent call last):
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+            _verify_param_shape_across_processes(self.process_group, parameters)_verify_param_shape_across_processes(self.process_group, parameters)        
+        _verify_param_shape_across_processes(self.process_group, parameters)
+_verify_param_shape_across_processes(self.process_group, parameters)_verify_param_shape_across_processes(self.process_group, parameters)_verify_param_shape_across_processes(self.process_group, parameters)
+
+_verify_param_shape_across_processes(self.process_group, parameters)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/utils.py", line 265, in _verify_param_shape_across_processes
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/utils.py", line 265, in _verify_param_shape_across_processes
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/utils.py", line 265, in _verify_param_shape_across_processes
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/utils.py", line 265, in _verify_param_shape_across_processes
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/utils.py", line 265, in _verify_param_shape_across_processes
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/utils.py", line 265, in _verify_param_shape_across_processes
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/utils.py", line 265, in _verify_param_shape_across_processes
+    _verify_param_shape_across_processes(self.process_group, parameters)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/utils.py", line 265, in _verify_param_shape_across_processes
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+Traceback (most recent call last):
+Traceback (most recent call last):
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    return dist._verify_params_across_processes(process_group, tensors, logger)
+        return dist._verify_params_across_processes(process_group, tensors, logger)return dist._verify_params_across_processes(process_group, tensors, logger)
+
+    return dist._verify_params_across_processes(process_group, tensors, logger)
+    return dist._verify_params_across_processes(process_group, tensors, logger)
+    RuntimeErrorreturn dist._verify_params_across_processes(process_group, tensors, logger)
+: RuntimeErrorRuntimeError    : : Rank 6 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+ Original exception: 
+[../third_party/gloo/gloo/transport/tcp/unbound_buffer.cc:81] Timed out waiting 1200000ms for recv operation to completereturn dist._verify_params_across_processes(process_group, tensors, logger)Rank 7 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+ Original exception: 
+[../third_party/gloo/gloo/transport/tcp/unbound_buffer.cc:81] Timed out waiting 1200000ms for recv operation to completeRuntimeErrorRank 7 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+ Original exception: 
+[../third_party/gloo/gloo/transport/tcp/unbound_buffer.cc:81] Timed out waiting 1200000ms for recv operation to complete
+RuntimeError
+: 
+
+RuntimeErrorRank 7 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+ Original exception: 
+[../third_party/gloo/gloo/transport/tcp/unbound_buffer.cc:81] Timed out waiting 1200000ms for recv operation to complete: 
+: Rank 6 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+ Original exception: 
+[../third_party/gloo/gloo/transport/tcp/unbound_buffer.cc:81] Timed out waiting 1200000ms for recv operation to completeRank 7 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+ Original exception: 
+[../third_party/gloo/gloo/transport/tcp/unbound_buffer.cc:81] Timed out waiting 1200000ms for recv operation to complete
+
+RuntimeError: Rank 6 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+ Original exception: 
+[../third_party/gloo/gloo/transport/tcp/unbound_buffer.cc:81] Timed out waiting 1200000ms for recv operation to complete
+    return dist._verify_params_across_processes(process_group, tensors, logger)
+RuntimeError: Rank 6 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+ Original exception: 
+[../third_party/gloo/gloo/transport/tcp/unbound_buffer.cc:81] Timed out waiting 1200000ms for recv operation to complete
+    trainer = DistributedTrainer(config_file)
+    trainer = DistributedTrainer(config_file)  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 171, in __init__
+
+    trainer = DistributedTrainer(config_file)      File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 171, in __init__
+trainer = DistributedTrainer(config_file)
+
+        trainer = DistributedTrainer(config_file)  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 171, in __init__
+trainer = DistributedTrainer(config_file)  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 171, in __init__
+
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 171, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 171, in __init__
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 171, in __init__
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 171, in __init__
+    self.model = self.init_model()  # Defines self.model
+                self.model = self.init_model()  # Defines self.model  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 720, in init_model
+
+self.model = self.init_model()  # Defines self.modelself.model = self.init_model()  # Defines self.model        
+self.model = self.init_model()  # Defines self.model
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 720, in init_model
+self.model = self.init_model()  # Defines self.modelself.model = self.init_model()  # Defines self.model
+
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 720, in init_model
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 720, in init_model
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 720, in init_model
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 720, in init_model
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 720, in init_model
+    self.model = self.init_model()  # Defines self.model
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 720, in init_model
+    model = self._init_model_instance()
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 730, in _init_model_instance
+    model = self._init_model_instance()    
+model = self._init_model_instance()
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 730, in _init_model_instance
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 730, in _init_model_instance
+model = self._init_model_instance()    
+model = self._init_model_instance()    
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 730, in _init_model_instance
+model = self._init_model_instance()
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 730, in _init_model_instance
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 730, in _init_model_instance
+model = self._init_model_instance()
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 730, in _init_model_instance
+    model = self._init_model_instance()    
+model = self._init_model(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 730, in _init_model_instance
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 862, in _init_model
+    model = self._init_model(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 862, in _init_model
+    model = self._init_model(
+    model = self._init_model(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 862, in _init_model
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 862, in _init_model
+            model = self._init_model(model = self._init_model(model = self._init_model(
+
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 862, in _init_model
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 862, in _init_model
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 862, in _init_model
+        model = DistributedDataParallel(model = self._init_model(
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 862, in _init_model
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 795, in __init__
+    model = DistributedDataParallel(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 795, in __init__
+        model = DistributedDataParallel(model = DistributedDataParallel(
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 795, in __init__
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 795, in __init__
+            model = DistributedDataParallel(model = DistributedDataParallel(model = DistributedDataParallel(
+
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 795, in __init__
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 795, in __init__
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 795, in __init__
+        trainer = DistributedTrainer(config_file)trainer = DistributedTrainer(config_file)
+
+        trainer = DistributedTrainer(config_file)      File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 171, in __init__
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 171, in __init__
+    trainer = DistributedTrainer(config_file)
+trainer = DistributedTrainer(config_file)trainer = DistributedTrainer(config_file)trainer = DistributedTrainer(config_file)
+
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 171, in __init__
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 171, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 171, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 171, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 171, in __init__
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 171, in __init__
+    model = DistributedDataParallel(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 795, in __init__
+    self.model = self.init_model()  # Defines self.model
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 720, in init_model
+        self.model = self.init_model()  # Defines self.model
+self.model = self.init_model()  # Defines self.model        
+self.model = self.init_model()  # Defines self.model      File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 720, in init_model
+
+self.model = self.init_model()  # Defines self.model      File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 720, in init_model
+self.model = self.init_model()  # Defines self.model
+self.model = self.init_model()  # Defines self.model
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 720, in init_model
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 720, in init_model
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 720, in init_model
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 720, in init_model
+    self.model = self.init_model()  # Defines self.model
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 720, in init_model
+    _verify_param_shape_across_processes(self.process_group, parameters)            
+_verify_param_shape_across_processes(self.process_group, parameters)_verify_param_shape_across_processes(self.process_group, parameters)    _verify_param_shape_across_processes(self.process_group, parameters)    
+
+
+    _verify_param_shape_across_processes(self.process_group, parameters)_verify_param_shape_across_processes(self.process_group, parameters)_verify_param_shape_across_processes(self.process_group, parameters)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/utils.py", line 265, in _verify_param_shape_across_processes
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/utils.py", line 265, in _verify_param_shape_across_processes
+
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/utils.py", line 265, in _verify_param_shape_across_processes
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/utils.py", line 265, in _verify_param_shape_across_processes
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/utils.py", line 265, in _verify_param_shape_across_processes
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/utils.py", line 265, in _verify_param_shape_across_processes
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/utils.py", line 265, in _verify_param_shape_across_processes
+    _verify_param_shape_across_processes(self.process_group, parameters)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/utils.py", line 265, in _verify_param_shape_across_processes
+        model = self._init_model_instance()model = self._init_model_instance()
+    
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 730, in _init_model_instance
+model = self._init_model_instance()
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 730, in _init_model_instance
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 730, in _init_model_instance
+        model = self._init_model_instance()model = self._init_model_instance()    
+
+model = self._init_model_instance()    
+model = self._init_model_instance()  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 730, in _init_model_instance
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 730, in _init_model_instance
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 730, in _init_model_instance
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 730, in _init_model_instance
+    model = self._init_model_instance()
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 730, in _init_model_instance
+    model = self._init_model(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 862, in _init_model
+        model = self._init_model(    model = self._init_model(
+model = self._init_model(
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 862, in _init_model
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 862, in _init_model
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 862, in _init_model
+    model = self._init_model(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 862, in _init_model
+    model = self._init_model(
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 862, in _init_model
+model = self._init_model(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 862, in _init_model
+    model = self._init_model(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 862, in _init_model
+    model = DistributedDataParallel(    
+model = DistributedDataParallel(    
+model = DistributedDataParallel(  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 795, in __init__
+    return dist._verify_params_across_processes(process_group, tensors, logger)    
+    
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 795, in __init__
+model = DistributedDataParallel(            
+model = DistributedDataParallel(  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 795, in __init__
+model = DistributedDataParallel(model = DistributedDataParallel(
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 795, in __init__
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 795, in __init__
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 795, in __init__
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 795, in __init__
+return dist._verify_params_across_processes(process_group, tensors, logger)
+    return dist._verify_params_across_processes(process_group, tensors, logger)
+    return dist._verify_params_across_processes(process_group, tensors, logger)
+RuntimeError    : return dist._verify_params_across_processes(process_group, tensors, logger)RuntimeError
+Rank 7 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+ Original exception: 
+[../third_party/gloo/gloo/transport/tcp/unbound_buffer.cc:81] Timed out waiting 1200000ms for recv operation to complete    
+return dist._verify_params_across_processes(process_group, tensors, logger): RuntimeError
+    : RuntimeErrorRank 6 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+ Original exception: 
+[../third_party/gloo/gloo/transport/tcp/unbound_buffer.cc:81] Timed out waiting 1200000ms for recv operation to completereturn dist._verify_params_across_processes(process_group, tensors, logger): 
+Rank 6 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+ Original exception: 
+[../third_party/gloo/gloo/transport/tcp/unbound_buffer.cc:81] Timed out waiting 1200000ms for recv operation to complete
+Rank 7 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+ Original exception: 
+[../third_party/gloo/gloo/transport/tcp/unbound_buffer.cc:81] Timed out waiting 1200000ms for recv operation to complete
+
+    model = DistributedDataParallel(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 795, in __init__
+RuntimeErrorRuntimeError: Rank 7 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+ Original exception: 
+[../third_party/gloo/gloo/transport/tcp/unbound_buffer.cc:81] Timed out waiting 1200000ms for recv operation to complete
+: RuntimeErrorRank 6 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+ Original exception: 
+[../third_party/gloo/gloo/transport/tcp/unbound_buffer.cc:81] Timed out waiting 1200000ms for recv operation to complete: 
+Rank 7 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+ Original exception: 
+[../third_party/gloo/gloo/transport/tcp/unbound_buffer.cc:81] Timed out waiting 1200000ms for recv operation to complete
+    return dist._verify_params_across_processes(process_group, tensors, logger)
+RuntimeError: Rank 6 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+ Original exception: 
+[../third_party/gloo/gloo/transport/tcp/unbound_buffer.cc:81] Timed out waiting 1200000ms for recv operation to complete
+    _verify_param_shape_across_processes(self.process_group, parameters)
+    _verify_param_shape_across_processes(self.process_group, parameters)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/utils.py", line 265, in _verify_param_shape_across_processes
+    _verify_param_shape_across_processes(self.process_group, parameters)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/utils.py", line 265, in _verify_param_shape_across_processes
+    _verify_param_shape_across_processes(self.process_group, parameters)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/utils.py", line 265, in _verify_param_shape_across_processes
+
+    _verify_param_shape_across_processes(self.process_group, parameters)
+    _verify_param_shape_across_processes(self.process_group, parameters)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/utils.py", line 265, in _verify_param_shape_across_processes
+    
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/utils.py", line 265, in _verify_param_shape_across_processes
+_verify_param_shape_across_processes(self.process_group, parameters)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/utils.py", line 265, in _verify_param_shape_across_processes
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/utils.py", line 265, in _verify_param_shape_across_processes
+    _verify_param_shape_across_processes(self.process_group, parameters)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/utils.py", line 265, in _verify_param_shape_across_processes
+    return dist._verify_params_across_processes(process_group, tensors, logger)
+    return dist._verify_params_across_processes(process_group, tensors, logger)
+    return dist._verify_params_across_processes(process_group, tensors, logger)
+    RuntimeErrorreturn dist._verify_params_across_processes(process_group, tensors, logger)
+: Rank 5 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+ Original exception: 
+[../third_party/gloo/gloo/transport/tcp/unbound_buffer.cc:81] Timed out waiting 1200000ms for recv operation to complete
+RuntimeError    return dist._verify_params_across_processes(process_group, tensors, logger)    RuntimeError: return dist._verify_params_across_processes(process_group, tensors, logger)
+Rank 5 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+ Original exception: 
+[../third_party/gloo/gloo/transport/tcp/unbound_buffer.cc:81] Timed out waiting 1200000ms for recv operation to complete: 
+
+RuntimeErrorRank 5 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+ Original exception: 
+[../third_party/gloo/gloo/transport/tcp/unbound_buffer.cc:81] Timed out waiting 1200000ms for recv operation to complete
+: RuntimeError: Rank 5 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+ Original exception: 
+[../third_party/gloo/gloo/transport/tcp/unbound_buffer.cc:81] Timed out waiting 1200000ms for recv operation to completeRuntimeErrorRank 4 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+ Original exception: 
+[../third_party/gloo/gloo/transport/tcp/unbound_buffer.cc:81] Timed out waiting 1200000ms for recv operation to complete
+
+: Rank 4 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+ Original exception: 
+[../third_party/gloo/gloo/transport/tcp/unbound_buffer.cc:81] Timed out waiting 1200000ms for recv operation to complete
+    return dist._verify_params_across_processes(process_group, tensors, logger)
+RuntimeError: Rank 4 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+ Original exception: 
+[../third_party/gloo/gloo/transport/tcp/unbound_buffer.cc:81] Timed out waiting 1200000ms for recv operation to complete
+    return dist._verify_params_across_processes(process_group, tensors, logger)
+RuntimeError: Rank 4 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+ Original exception: 
+[../third_party/gloo/gloo/transport/tcp/unbound_buffer.cc:81] Timed out waiting 1200000ms for recv operation to complete
+Traceback (most recent call last):
+Traceback (most recent call last):
+Traceback (most recent call last):
+Traceback (most recent call last):
+Traceback (most recent call last):
+Traceback (most recent call last):
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    trainer = DistributedTrainer(config_file)    
+trainer = DistributedTrainer(config_file)    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 171, in __init__
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 171, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 171, in __init__
+    trainer = DistributedTrainer(config_file)
+            trainer = DistributedTrainer(config_file)trainer = DistributedTrainer(config_file)trainer = DistributedTrainer(config_file)  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 171, in __init__
+
+
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 171, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 171, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 171, in __init__
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 171, in __init__
+            self.model = self.init_model()  # Defines self.modelself.model = self.init_model()  # Defines self.model        self.model = self.init_model()  # Defines self.model
+
+self.model = self.init_model()  # Defines self.modelself.model = self.init_model()  # Defines self.model
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 720, in init_model
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 720, in init_model
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 720, in init_model
+
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 720, in init_model
+self.model = self.init_model()  # Defines self.modelself.model = self.init_model()  # Defines self.model  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 720, in init_model
+
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 720, in init_model
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 720, in init_model
+    self.model = self.init_model()  # Defines self.model
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 720, in init_model
+    model = self._init_model_instance()    
+model = self._init_model_instance()
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 730, in _init_model_instance
+model = self._init_model_instance()
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 730, in _init_model_instance
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 730, in _init_model_instance
+model = self._init_model_instance()
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 730, in _init_model_instance
+model = self._init_model_instance()
+    model = self._init_model_instance()  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 730, in _init_model_instance
+
+    model = self._init_model_instance()  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 730, in _init_model_instance
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 730, in _init_model_instance
+    model = self._init_model_instance()
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 730, in _init_model_instance
+    model = self._init_model(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 862, in _init_model
+    model = self._init_model(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 862, in _init_model
+    model = self._init_model(
+          File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 862, in _init_model
+model = self._init_model(    model = self._init_model(
+model = self._init_model(
+    
+model = self._init_model(  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 862, in _init_model
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 862, in _init_model
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 862, in _init_model
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 862, in _init_model
+    model = self._init_model(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 862, in _init_model
+        model = DistributedDataParallel(model = DistributedDataParallel(
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 795, in __init__
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 795, in __init__
+    model = DistributedDataParallel(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 795, in __init__
+    model = DistributedDataParallel(
+        model = DistributedDataParallel(model = DistributedDataParallel(
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 795, in __init__
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 795, in __init__
+model = DistributedDataParallel(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 795, in __init__
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 795, in __init__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+Traceback (most recent call last):
+Traceback (most recent call last):
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+Traceback (most recent call last):
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    model = DistributedDataParallel(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 795, in __init__
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 171, in __init__
+    trainer = DistributedTrainer(config_file)
+        trainer = DistributedTrainer(config_file)          File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 171, in __init__
+trainer = DistributedTrainer(config_file)
+trainer = DistributedTrainer(config_file)trainer = DistributedTrainer(config_file)    
+trainer = DistributedTrainer(config_file)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 171, in __init__
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 171, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 171, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 171, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 171, in __init__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+Traceback (most recent call last):
+Traceback (most recent call last):
+Traceback (most recent call last):
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+Traceback (most recent call last):
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 171, in __init__
+    _verify_param_shape_across_processes(self.process_group, parameters)
+    _verify_param_shape_across_processes(self.process_group, parameters)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/utils.py", line 265, in _verify_param_shape_across_processes
+    _verify_param_shape_across_processes(self.process_group, parameters)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/utils.py", line 265, in _verify_param_shape_across_processes
+
+    _verify_param_shape_across_processes(self.process_group, parameters)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/utils.py", line 265, in _verify_param_shape_across_processes
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/utils.py", line 265, in _verify_param_shape_across_processes
+_verify_param_shape_across_processes(self.process_group, parameters)    
+_verify_param_shape_across_processes(self.process_group, parameters)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/utils.py", line 265, in _verify_param_shape_across_processes
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/utils.py", line 265, in _verify_param_shape_across_processes
+    _verify_param_shape_across_processes(self.process_group, parameters)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/utils.py", line 265, in _verify_param_shape_across_processes
+    _verify_param_shape_across_processes(self.process_group, parameters)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/utils.py", line 265, in _verify_param_shape_across_processes
+    self.model = self.init_model()  # Defines self.model
+            self.model = self.init_model()  # Defines self.modelself.model = self.init_model()  # Defines self.model  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 720, in init_model
+    
+self.model = self.init_model()  # Defines self.model    self.model = self.init_model()  # Defines self.model    self.model = self.init_model()  # Defines self.model
+
+
+self.model = self.init_model()  # Defines self.model  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 720, in init_model
+
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 720, in init_model
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 720, in init_model
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 720, in init_model
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 720, in init_model
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 720, in init_model
+    return dist._verify_params_across_processes(process_group, tensors, logger)
+    self.model = self.init_model()  # Defines self.model
+    return dist._verify_params_across_processes(process_group, tensors, logger)
+    return dist._verify_params_across_processes(process_group, tensors, logger)
+    return dist._verify_params_across_processes(process_group, tensors, logger)RuntimeError
+    : return dist._verify_params_across_processes(process_group, tensors, logger)Rank 11 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+ Original exception: 
+[../third_party/gloo/gloo/transport/tcp/unbound_buffer.cc:81] Timed out waiting 1200000ms for recv operation to completeRuntimeError
+    
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 720, in init_model
+RuntimeErrorreturn dist._verify_params_across_processes(process_group, tensors, logger): 
+: Rank 10 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+ Original exception: 
+[../third_party/gloo/gloo/transport/tcp/unbound_buffer.cc:81] Timed out waiting 1200000ms for recv operation to completeRank 10 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+ Original exception: 
+[../third_party/gloo/gloo/transport/tcp/unbound_buffer.cc:81] Timed out waiting 1200000ms for recv operation to complete
+RuntimeError
+: Rank 10 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+ Original exception: 
+[../third_party/gloo/gloo/transport/tcp/unbound_buffer.cc:81] Timed out waiting 1200000ms for recv operation to complete
+RuntimeErrorRuntimeError: : Rank 11 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+ Original exception: 
+[../third_party/gloo/gloo/transport/tcp/unbound_buffer.cc:81] Timed out waiting 1200000ms for recv operation to completeRank 10 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+ Original exception: 
+[../third_party/gloo/gloo/transport/tcp/unbound_buffer.cc:81] Timed out waiting 1200000ms for recv operation to complete
+
+    return dist._verify_params_across_processes(process_group, tensors, logger)
+RuntimeError: Rank 11 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+ Original exception: 
+[../third_party/gloo/gloo/transport/tcp/unbound_buffer.cc:81] Timed out waiting 1200000ms for recv operation to complete
+    return dist._verify_params_across_processes(process_group, tensors, logger)
+RuntimeError: Rank 11 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+ Original exception: 
+[../third_party/gloo/gloo/transport/tcp/unbound_buffer.cc:81] Timed out waiting 1200000ms for recv operation to complete
+    trainer = DistributedTrainer(config_file)        
+trainer = DistributedTrainer(config_file)trainer = DistributedTrainer(config_file)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 171, in __init__
+    trainer = DistributedTrainer(config_file)  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 171, in __init__
+    trainer = DistributedTrainer(config_file)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 171, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 171, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 171, in __init__
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 171, in __init__
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 171, in __init__
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 171, in __init__
+    model = self._init_model_instance()
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 730, in _init_model_instance
+model = self._init_model_instance()
+    model = self._init_model_instance()  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 730, in _init_model_instance
+
+    model = self._init_model_instance()  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 730, in _init_model_instance
+
+    model = self._init_model_instance()  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 730, in _init_model_instance
+
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 730, in _init_model_instance
+model = self._init_model_instance()    
+model = self._init_model_instance()
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 730, in _init_model_instance
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 730, in _init_model_instance
+    model = self._init_model_instance()
+          File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 730, in _init_model_instance
+model = self._init_model(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 862, in _init_model
+        model = self._init_model(model = self._init_model(    
+model = self._init_model(
+model = self._init_model(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 862, in _init_model
+    
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 862, in _init_model
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 862, in _init_model
+model = self._init_model(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 862, in _init_model
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 862, in _init_model
+    model = self._init_model(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 862, in _init_model
+    model = DistributedDataParallel(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 795, in __init__
+    model = DistributedDataParallel(
+    model = self._init_model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 795, in __init__
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 862, in _init_model
+    model = DistributedDataParallel(model = DistributedDataParallel(    
+
+model = DistributedDataParallel(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 795, in __init__
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 795, in __init__
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 795, in __init__
+model = DistributedDataParallel(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 795, in __init__
+    model = DistributedDataParallel(
+            self.model = self.init_model()  # Defines self.modelself.model = self.init_model()  # Defines self.modelself.model = self.init_model()  # Defines self.model
+
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 795, in __init__
+    self.model = self.init_model()  # Defines self.model  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 720, in init_model
+          File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 720, in init_model
+self.model = self.init_model()  # Defines self.model  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 720, in init_model
+    
+self.model = self.init_model()  # Defines self.model
+self.model = self.init_model()  # Defines self.model
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 720, in init_model
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 720, in init_model
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 720, in init_model
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 720, in init_model
+    self.model = self.init_model()  # Defines self.model
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 720, in init_model
+    model = DistributedDataParallel(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 795, in __init__
+    model = self._init_model_instance()
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 730, in _init_model_instance
+    model = self._init_model_instance()
+    model = self._init_model_instance()  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 730, in _init_model_instance
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 730, in _init_model_instance
+    model = self._init_model_instance()
+    model = self._init_model_instance()  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 730, in _init_model_instance
+
+            _verify_param_shape_across_processes(self.process_group, parameters)            _verify_param_shape_across_processes(self.process_group, parameters)_verify_param_shape_across_processes(self.process_group, parameters)    _verify_param_shape_across_processes(self.process_group, parameters)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 730, in _init_model_instance
+model = self._init_model_instance()
+    model = self._init_model_instance()
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 730, in _init_model_instance
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 730, in _init_model_instance
+_verify_param_shape_across_processes(self.process_group, parameters)
+_verify_param_shape_across_processes(self.process_group, parameters)
+_verify_param_shape_across_processes(self.process_group, parameters)
+
+
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/utils.py", line 265, in _verify_param_shape_across_processes
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/utils.py", line 265, in _verify_param_shape_across_processes
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/utils.py", line 265, in _verify_param_shape_across_processes
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/utils.py", line 265, in _verify_param_shape_across_processes
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/utils.py", line 265, in _verify_param_shape_across_processes
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/utils.py", line 265, in _verify_param_shape_across_processes
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/utils.py", line 265, in _verify_param_shape_across_processes
+    model = self._init_model(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 862, in _init_model
+        model = self._init_model(model = self._init_model_instance()
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 862, in _init_model
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 730, in _init_model_instance
+    model = self._init_model(
+    _verify_param_shape_across_processes(self.process_group, parameters)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 862, in _init_model
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/utils.py", line 265, in _verify_param_shape_across_processes
+    model = self._init_model(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 862, in _init_model
+    model = self._init_model(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 862, in _init_model
+    model = self._init_model(    
+model = self._init_model(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 862, in _init_model
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 862, in _init_model
+    model = DistributedDataParallel(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 795, in __init__
+    model = self._init_model(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 862, in _init_model
+    model = DistributedDataParallel(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 795, in __init__
+    model = DistributedDataParallel(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 795, in __init__
+    model = DistributedDataParallel(
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 795, in __init__
+model = DistributedDataParallel(
+    model = DistributedDataParallel(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 795, in __init__
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 795, in __init__
+    model = DistributedDataParallel(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 795, in __init__
+    model = DistributedDataParallel(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 795, in __init__
+    return dist._verify_params_across_processes(process_group, tensors, logger)    
+return dist._verify_params_across_processes(process_group, tensors, logger)
+    return dist._verify_params_across_processes(process_group, tensors, logger)
+    return dist._verify_params_across_processes(process_group, tensors, logger)
+    RuntimeErrorreturn dist._verify_params_across_processes(process_group, tensors, logger)
+RuntimeError: Rank 11 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+ Original exception: 
+[../third_party/gloo/gloo/transport/tcp/unbound_buffer.cc:81] Timed out waiting 1200000ms for recv operation to complete: 
+    Rank 10 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+ Original exception: 
+[../third_party/gloo/gloo/transport/tcp/unbound_buffer.cc:81] Timed out waiting 1200000ms for recv operation to completereturn dist._verify_params_across_processes(process_group, tensors, logger)
+
+    return dist._verify_params_across_processes(process_group, tensors, logger)RuntimeErrorRuntimeError
+RuntimeError: Rank 11 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+ Original exception: 
+[../third_party/gloo/gloo/transport/tcp/unbound_buffer.cc:81] Timed out waiting 1200000ms for recv operation to complete: 
+Rank 11 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+ Original exception: 
+[../third_party/gloo/gloo/transport/tcp/unbound_buffer.cc:81] Timed out waiting 1200000ms for recv operation to complete
+RuntimeError: : RuntimeErrorRank 10 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+ Original exception: 
+[../third_party/gloo/gloo/transport/tcp/unbound_buffer.cc:81] Timed out waiting 1200000ms for recv operation to completeRank 10 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+ Original exception: 
+[../third_party/gloo/gloo/transport/tcp/unbound_buffer.cc:81] Timed out waiting 1200000ms for recv operation to complete
+
+: Rank 10 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+ Original exception: 
+[../third_party/gloo/gloo/transport/tcp/unbound_buffer.cc:81] Timed out waiting 1200000ms for recv operation to complete
+    return dist._verify_params_across_processes(process_group, tensors, logger)
+RuntimeError: Rank 11 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+ Original exception: 
+[../third_party/gloo/gloo/transport/tcp/unbound_buffer.cc:81] Timed out waiting 1200000ms for recv operation to complete
+        _verify_param_shape_across_processes(self.process_group, parameters)_verify_param_shape_across_processes(self.process_group, parameters)
+
+        _verify_param_shape_across_processes(self.process_group, parameters)          File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/utils.py", line 265, in _verify_param_shape_across_processes
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/utils.py", line 265, in _verify_param_shape_across_processes
+_verify_param_shape_across_processes(self.process_group, parameters)_verify_param_shape_across_processes(self.process_group, parameters)
+
+    _verify_param_shape_across_processes(self.process_group, parameters)
+_verify_param_shape_across_processes(self.process_group, parameters)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/utils.py", line 265, in _verify_param_shape_across_processes
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/utils.py", line 265, in _verify_param_shape_across_processes
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/utils.py", line 265, in _verify_param_shape_across_processes
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/utils.py", line 265, in _verify_param_shape_across_processes
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/utils.py", line 265, in _verify_param_shape_across_processes
+    _verify_param_shape_across_processes(self.process_group, parameters)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/utils.py", line 265, in _verify_param_shape_across_processes
+    return dist._verify_params_across_processes(process_group, tensors, logger)    
+    return dist._verify_params_across_processes(process_group, tensors, logger)    
+return dist._verify_params_across_processes(process_group, tensors, logger)return dist._verify_params_across_processes(process_group, tensors, logger)
+
+    return dist._verify_params_across_processes(process_group, tensors, logger)RuntimeError
+: Rank 2 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+ Original exception: 
+[../third_party/gloo/gloo/transport/tcp/unbound_buffer.cc:81] Timed out waiting 1200000ms for recv operation to complete    
+    RuntimeErrorreturn dist._verify_params_across_processes(process_group, tensors, logger)RuntimeErrorreturn dist._verify_params_across_processes(process_group, tensors, logger)
+
+: : RuntimeErrorRuntimeErrorRank 2 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+ Original exception: 
+[../third_party/gloo/gloo/transport/tcp/unbound_buffer.cc:81] Timed out waiting 1200000ms for recv operation to completeRank 3 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+ Original exception: 
+[../third_party/gloo/gloo/transport/tcp/unbound_buffer.cc:81] Timed out waiting 1200000ms for recv operation to complete: 
+
+: Rank 3 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+ Original exception: 
+[../third_party/gloo/gloo/transport/tcp/unbound_buffer.cc:81] Timed out waiting 1200000ms for recv operation to complete
+Rank 3 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+ Original exception: 
+[../third_party/gloo/gloo/transport/tcp/unbound_buffer.cc:81] Timed out waiting 1200000ms for recv operation to complete
+RuntimeError: RuntimeErrorRank 2 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+ Original exception: 
+[../third_party/gloo/gloo/transport/tcp/unbound_buffer.cc:81] Timed out waiting 1200000ms for recv operation to complete
+: Rank 2 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+ Original exception: 
+[../third_party/gloo/gloo/transport/tcp/unbound_buffer.cc:81] Timed out waiting 1200000ms for recv operation to complete
+    return dist._verify_params_across_processes(process_group, tensors, logger)
+RuntimeError: Rank 3 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+ Original exception: 
+[../third_party/gloo/gloo/transport/tcp/unbound_buffer.cc:81] Timed out waiting 1200000ms for recv operation to complete
+Traceback (most recent call last):
+Traceback (most recent call last):
+Traceback (most recent call last):
+Traceback (most recent call last):
+Traceback (most recent call last):
+Traceback (most recent call last):
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+        trainer = DistributedTrainer(config_file)        trainer = DistributedTrainer(config_file)    
+trainer = DistributedTrainer(config_file)
+trainer = DistributedTrainer(config_file)trainer = DistributedTrainer(config_file)    
+
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 171, in __init__
+trainer = DistributedTrainer(config_file)      File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 171, in __init__
+trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 171, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 171, in __init__
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 171, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 171, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 171, in __init__
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 171, in __init__
+        self.model = self.init_model()  # Defines self.modelself.model = self.init_model()  # Defines self.model
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 720, in init_model
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 720, in init_model
+    self.model = self.init_model()  # Defines self.model
+        self.model = self.init_model()  # Defines self.model  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 720, in init_model
+
+self.model = self.init_model()  # Defines self.model    
+self.model = self.init_model()  # Defines self.model
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 720, in init_model
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 720, in init_model
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 720, in init_model
+    self.model = self.init_model()  # Defines self.model
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 720, in init_model
+    self.model = self.init_model()  # Defines self.model
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 720, in init_model
+        model = self._init_model_instance()    model = self._init_model_instance()
+model = self._init_model_instance()    
+
+          File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 730, in _init_model_instance
+model = self._init_model_instance()  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 730, in _init_model_instance
+    model = self._init_model_instance()model = self._init_model_instance()  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 730, in _init_model_instance
+
+
+model = self._init_model_instance()
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 730, in _init_model_instance
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 730, in _init_model_instance
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 730, in _init_model_instance
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 730, in _init_model_instance
+    model = self._init_model_instance()
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 730, in _init_model_instance
+    model = self._init_model(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 862, in _init_model
+    model = self._init_model(    
+model = self._init_model(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 862, in _init_model
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 862, in _init_model
+    model = self._init_model(
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 862, in _init_model
+model = self._init_model(
+    model = self._init_model(  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 862, in _init_model
+    
+model = self._init_model(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 862, in _init_model
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 862, in _init_model
+    model = self._init_model(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 862, in _init_model
+    model = DistributedDataParallel(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 795, in __init__
+    model = DistributedDataParallel(    
+model = DistributedDataParallel(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 795, in __init__
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 795, in __init__
+        model = DistributedDataParallel(model = DistributedDataParallel(
+    
+model = DistributedDataParallel(
+    model = DistributedDataParallel(  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 795, in __init__
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 795, in __init__
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 795, in __init__
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 795, in __init__
+    model = DistributedDataParallel(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 795, in __init__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+Traceback (most recent call last):
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+            _verify_param_shape_across_processes(self.process_group, parameters)_verify_param_shape_across_processes(self.process_group, parameters)    _verify_param_shape_across_processes(self.process_group, parameters)
+
+        _verify_param_shape_across_processes(self.process_group, parameters)_verify_param_shape_across_processes(self.process_group, parameters)
+    _verify_param_shape_across_processes(self.process_group, parameters)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/utils.py", line 265, in _verify_param_shape_across_processes
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/utils.py", line 265, in _verify_param_shape_across_processes
+_verify_param_shape_across_processes(self.process_group, parameters)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/utils.py", line 265, in _verify_param_shape_across_processes
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/utils.py", line 265, in _verify_param_shape_across_processes
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/utils.py", line 265, in _verify_param_shape_across_processes
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/utils.py", line 265, in _verify_param_shape_across_processes
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/utils.py", line 265, in _verify_param_shape_across_processes
+    _verify_param_shape_across_processes(self.process_group, parameters)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/utils.py", line 265, in _verify_param_shape_across_processes
+        trainer = DistributedTrainer(config_file)trainer = DistributedTrainer(config_file)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 171, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 171, in __init__
+        trainer = DistributedTrainer(config_file)trainer = DistributedTrainer(config_file)    
+trainer = DistributedTrainer(config_file)
+    
+trainer = DistributedTrainer(config_file)  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 171, in __init__
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 171, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 171, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 171, in __init__
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 171, in __init__
+        return dist._verify_params_across_processes(process_group, tensors, logger)return dist._verify_params_across_processes(process_group, tensors, logger)
+
+    return dist._verify_params_across_processes(process_group, tensors, logger)
+    return dist._verify_params_across_processes(process_group, tensors, logger)
+    return dist._verify_params_across_processes(process_group, tensors, logger)
+    return dist._verify_params_across_processes(process_group, tensors, logger)RuntimeErrorRuntimeError    
+return dist._verify_params_across_processes(process_group, tensors, logger)
+: RuntimeErrorRank 14 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+ Original exception: 
+[../third_party/gloo/gloo/transport/tcp/unbound_buffer.cc:81] Timed out waiting 1200000ms for recv operation to complete: 
+: RuntimeErrorRank 15 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+ Original exception: 
+[../third_party/gloo/gloo/transport/tcp/unbound_buffer.cc:81] Timed out waiting 1200000ms for recv operation to completeRuntimeError
+Rank 14 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+ Original exception: 
+[../third_party/gloo/gloo/transport/tcp/unbound_buffer.cc:81] Timed out waiting 1200000ms for recv operation to complete: RuntimeError: 
+Rank 14 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+ Original exception: 
+[../third_party/gloo/gloo/transport/tcp/unbound_buffer.cc:81] Timed out waiting 1200000ms for recv operation to complete
+Rank 15 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+ Original exception: 
+[../third_party/gloo/gloo/transport/tcp/unbound_buffer.cc:81] Timed out waiting 1200000ms for recv operation to completeRuntimeError
+    trainer = DistributedTrainer(config_file)
+: : Rank 15 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+ Original exception: 
+[../third_party/gloo/gloo/transport/tcp/unbound_buffer.cc:81] Timed out waiting 1200000ms for recv operation to completeRank 15 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+ Original exception: 
+[../third_party/gloo/gloo/transport/tcp/unbound_buffer.cc:81] Timed out waiting 1200000ms for recv operation to complete
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 171, in __init__
+    return dist._verify_params_across_processes(process_group, tensors, logger)
+RuntimeError: Rank 14 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+ Original exception: 
+[../third_party/gloo/gloo/transport/tcp/unbound_buffer.cc:81] Timed out waiting 1200000ms for recv operation to complete
+        self.model = self.init_model()  # Defines self.model
+self.model = self.init_model()  # Defines self.model
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 720, in init_model
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 720, in init_model
+self.model = self.init_model()  # Defines self.model    
+self.model = self.init_model()  # Defines self.model
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 720, in init_model
+self.model = self.init_model()  # Defines self.model  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 720, in init_model
+
+    self.model = self.init_model()  # Defines self.model    
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 720, in init_model
+self.model = self.init_model()  # Defines self.model
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 720, in init_model
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 720, in init_model
+    self.model = self.init_model()  # Defines self.model
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 720, in init_model
+        model = self._init_model_instance()    model = self._init_model_instance()
+    model = self._init_model_instance()        
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 730, in _init_model_instance
+model = self._init_model_instance()
+model = self._init_model_instance()model = self._init_model_instance()model = self._init_model_instance()  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 730, in _init_model_instance
+
+
+
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 730, in _init_model_instance
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 730, in _init_model_instance
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 730, in _init_model_instance
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 730, in _init_model_instance
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 730, in _init_model_instance
+    model = self._init_model_instance()
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 730, in _init_model_instance
+    model = self._init_model(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 862, in _init_model
+    model = self._init_model(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 862, in _init_model
+    model = self._init_model(
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 862, in _init_model
+model = self._init_model(
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 862, in _init_model
+model = self._init_model(
+    model = self._init_model(      File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 862, in _init_model
+
+model = self._init_model(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 862, in _init_model
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 862, in _init_model
+    model = self._init_model(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 862, in _init_model
+    model = DistributedDataParallel(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 795, in __init__
+    model = DistributedDataParallel(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 795, in __init__
+    model = DistributedDataParallel(
+    model = DistributedDataParallel(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 795, in __init__
+              File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 795, in __init__
+model = DistributedDataParallel(model = DistributedDataParallel(model = DistributedDataParallel(
+
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 795, in __init__
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 795, in __init__
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 795, in __init__
+    model = DistributedDataParallel(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 795, in __init__
+        _verify_param_shape_across_processes(self.process_group, parameters)        _verify_param_shape_across_processes(self.process_group, parameters)
+_verify_param_shape_across_processes(self.process_group, parameters)_verify_param_shape_across_processes(self.process_group, parameters)
+
+
+              File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/utils.py", line 265, in _verify_param_shape_across_processes
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/utils.py", line 265, in _verify_param_shape_across_processes
+_verify_param_shape_across_processes(self.process_group, parameters)_verify_param_shape_across_processes(self.process_group, parameters)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/utils.py", line 265, in _verify_param_shape_across_processes
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/utils.py", line 265, in _verify_param_shape_across_processes
+_verify_param_shape_across_processes(self.process_group, parameters)
+
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/utils.py", line 265, in _verify_param_shape_across_processes
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/utils.py", line 265, in _verify_param_shape_across_processes
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/utils.py", line 265, in _verify_param_shape_across_processes
+    _verify_param_shape_across_processes(self.process_group, parameters)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/utils.py", line 265, in _verify_param_shape_across_processes
+    return dist._verify_params_across_processes(process_group, tensors, logger)
+RuntimeError: Rank 14 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+ Original exception: 
+[../third_party/gloo/gloo/transport/tcp/unbound_buffer.cc:81] Timed out waiting 1200000ms for recv operation to complete
+    return dist._verify_params_across_processes(process_group, tensors, logger)    
+return dist._verify_params_across_processes(process_group, tensors, logger)
+    return dist._verify_params_across_processes(process_group, tensors, logger)    
+return dist._verify_params_across_processes(process_group, tensors, logger)
+    return dist._verify_params_across_processes(process_group, tensors, logger)
+    RuntimeErrorRuntimeErrorreturn dist._verify_params_across_processes(process_group, tensors, logger)RuntimeError
+: : : Rank 14 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+ Original exception: 
+[../third_party/gloo/gloo/transport/tcp/unbound_buffer.cc:81] Timed out waiting 1200000ms for recv operation to completeRank 14 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+ Original exception: 
+[../third_party/gloo/gloo/transport/tcp/unbound_buffer.cc:81] Timed out waiting 1200000ms for recv operation to completeRank 15 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+ Original exception: 
+[../third_party/gloo/gloo/transport/tcp/unbound_buffer.cc:81] Timed out waiting 1200000ms for recv operation to complete
+RuntimeErrorRuntimeError
+
+: : Rank 14 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+ Original exception: 
+[../third_party/gloo/gloo/transport/tcp/unbound_buffer.cc:81] Timed out waiting 1200000ms for recv operation to completeRank 15 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+ Original exception: 
+[../third_party/gloo/gloo/transport/tcp/unbound_buffer.cc:81] Timed out waiting 1200000ms for recv operation to complete
+
+RuntimeError: Rank 15 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+ Original exception: 
+[../third_party/gloo/gloo/transport/tcp/unbound_buffer.cc:81] Timed out waiting 1200000ms for recv operation to complete
+    return dist._verify_params_across_processes(process_group, tensors, logger)
+RuntimeError: Rank 15 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+ Original exception: 
+[../third_party/gloo/gloo/transport/tcp/unbound_buffer.cc:81] Timed out waiting 1200000ms for recv operation to complete
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+Traceback (most recent call last):
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+        trainer = DistributedTrainer(config_file)trainer = DistributedTrainer(config_file)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 171, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 171, in __init__
+        trainer = DistributedTrainer(config_file)trainer = DistributedTrainer(config_file)        
+trainer = DistributedTrainer(config_file)
+trainer = DistributedTrainer(config_file)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 171, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 171, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 171, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 171, in __init__
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 171, in __init__
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 171, in __init__
+    self.model = self.init_model()  # Defines self.model
+    self.model = self.init_model()  # Defines self.model    
+self.model = self.init_model()  # Defines self.model  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 720, in init_model
+
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 720, in init_model
+self.model = self.init_model()  # Defines self.model
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 720, in init_model
+    self.model = self.init_model()  # Defines self.model  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 720, in init_model
+
+    self.model = self.init_model()  # Defines self.model
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 720, in init_model
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 720, in init_model
+self.model = self.init_model()  # Defines self.model
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 720, in init_model
+    self.model = self.init_model()  # Defines self.model
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 720, in init_model
+            model = self._init_model_instance()    model = self._init_model_instance()    model = self._init_model_instance()
+model = self._init_model_instance()
+    
+    
+model = self._init_model_instance()  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 730, in _init_model_instance
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 730, in _init_model_instance
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 730, in _init_model_instance
+model = self._init_model_instance()model = self._init_model_instance()
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 730, in _init_model_instance
+
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 730, in _init_model_instance
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 730, in _init_model_instance
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 730, in _init_model_instance
+    model = self._init_model_instance()
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 730, in _init_model_instance
+    model = self._init_model(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 862, in _init_model
+    model = self._init_model(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 862, in _init_model
+    model = self._init_model(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 862, in _init_model
+    model = self._init_model(
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 862, in _init_model
+model = self._init_model(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 862, in _init_model
+    model = self._init_model(
+    model = self._init_model(  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 862, in _init_model
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 862, in _init_model
+    model = self._init_model(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 862, in _init_model
+    model = DistributedDataParallel(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 795, in __init__
+    model = DistributedDataParallel(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 795, in __init__
+        model = DistributedDataParallel(model = DistributedDataParallel(
+
+    model = DistributedDataParallel(  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 795, in __init__
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 795, in __init__
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 795, in __init__
+    model = DistributedDataParallel(
+    model = DistributedDataParallel(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 795, in __init__
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 795, in __init__
+    model = DistributedDataParallel(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 795, in __init__
+    _verify_param_shape_across_processes(self.process_group, parameters)
+    _verify_param_shape_across_processes(self.process_group, parameters)      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/utils.py", line 265, in _verify_param_shape_across_processes
+
+_verify_param_shape_across_processes(self.process_group, parameters)
+    _verify_param_shape_across_processes(self.process_group, parameters)    
+_verify_param_shape_across_processes(self.process_group, parameters)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/utils.py", line 265, in _verify_param_shape_across_processes
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/utils.py", line 265, in _verify_param_shape_across_processes
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/utils.py", line 265, in _verify_param_shape_across_processes
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/utils.py", line 265, in _verify_param_shape_across_processes
+_verify_param_shape_across_processes(self.process_group, parameters)    
+_verify_param_shape_across_processes(self.process_group, parameters)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/utils.py", line 265, in _verify_param_shape_across_processes
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/utils.py", line 265, in _verify_param_shape_across_processes
+    _verify_param_shape_across_processes(self.process_group, parameters)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/utils.py", line 265, in _verify_param_shape_across_processes
+        return dist._verify_params_across_processes(process_group, tensors, logger)
+return dist._verify_params_across_processes(process_group, tensors, logger)
+    return dist._verify_params_across_processes(process_group, tensors, logger)
+    return dist._verify_params_across_processes(process_group, tensors, logger)
+    RuntimeErrorreturn dist._verify_params_across_processes(process_group, tensors, logger)
+: Rank 9 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+ Original exception: 
+[../third_party/gloo/gloo/transport/tcp/unbound_buffer.cc:81] Timed out waiting 1200000ms for recv operation to complete    
+return dist._verify_params_across_processes(process_group, tensors, logger)RuntimeErrorRuntimeError
+: : RuntimeErrorRank 8 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+ Original exception: 
+[../third_party/gloo/gloo/transport/tcp/unbound_buffer.cc:81] Timed out waiting 1200000ms for recv operation to completeRank 9 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+ Original exception: 
+[../third_party/gloo/gloo/transport/tcp/unbound_buffer.cc:81] Timed out waiting 1200000ms for recv operation to complete: 
+
+Rank 8 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+ Original exception: 
+[../third_party/gloo/gloo/transport/tcp/unbound_buffer.cc:81] Timed out waiting 1200000ms for recv operation to completeRuntimeError
+: RuntimeErrorRank 8 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+ Original exception: 
+[../third_party/gloo/gloo/transport/tcp/unbound_buffer.cc:81] Timed out waiting 1200000ms for recv operation to complete
+: Rank 9 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+ Original exception: 
+[../third_party/gloo/gloo/transport/tcp/unbound_buffer.cc:81] Timed out waiting 1200000ms for recv operation to complete
+    return dist._verify_params_across_processes(process_group, tensors, logger)
+RuntimeError: Rank 9 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+ Original exception: 
+[../third_party/gloo/gloo/transport/tcp/unbound_buffer.cc:81] Timed out waiting 1200000ms for recv operation to complete
+    return dist._verify_params_across_processes(process_group, tensors, logger)
+RuntimeError: Rank 8 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+ Original exception: 
+[../third_party/gloo/gloo/transport/tcp/unbound_buffer.cc:81] Timed out waiting 1200000ms for recv operation to complete
+Traceback (most recent call last):
+Traceback (most recent call last):
+Traceback (most recent call last):
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+        trainer = DistributedTrainer(config_file)trainer = DistributedTrainer(config_file)    
+trainer = DistributedTrainer(config_file)
+    trainer = DistributedTrainer(config_file)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 171, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 171, in __init__
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 171, in __init__
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 171, in __init__
+trainer = DistributedTrainer(config_file)trainer = DistributedTrainer(config_file)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 171, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 171, in __init__
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 171, in __init__
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 171, in __init__
+    self.model = self.init_model()  # Defines self.model    
+self.model = self.init_model()  # Defines self.model
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 720, in init_model
+self.model = self.init_model()  # Defines self.model  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 720, in init_model
+    
+self.model = self.init_model()  # Defines self.model
+    self.model = self.init_model()  # Defines self.model  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 720, in init_model
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 720, in init_model
+
+    self.model = self.init_model()  # Defines self.model  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 720, in init_model
+
+    self.model = self.init_model()  # Defines self.model
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 720, in init_model
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 720, in init_model
+    self.model = self.init_model()  # Defines self.model
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 720, in init_model
+                model = self._init_model_instance()model = self._init_model_instance()model = self._init_model_instance()model = self._init_model_instance()
+
+
+            model = self._init_model_instance()  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 730, in _init_model_instance
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 730, in _init_model_instance
+model = self._init_model_instance()  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 730, in _init_model_instance
+model = self._init_model_instance()
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 730, in _init_model_instance
+
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 730, in _init_model_instance
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 730, in _init_model_instance
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 730, in _init_model_instance
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    model = self._init_model_instance()
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 730, in _init_model_instance
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+Traceback (most recent call last):
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+Traceback (most recent call last):
+Traceback (most recent call last):
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    model = self._init_model(
+    model = self._init_model(  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 862, in _init_model
+
+    model = self._init_model(  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 862, in _init_model
+
+    model = self._init_model(  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 862, in _init_model
+
+    model = self._init_model(  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 862, in _init_model
+
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 862, in _init_model
+model = self._init_model(
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 862, in _init_model
+model = self._init_model(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 862, in _init_model
+    model = self._init_model(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 862, in _init_model
+    model = DistributedDataParallel(
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 795, in __init__
+    model = DistributedDataParallel(    model = DistributedDataParallel(
+model = DistributedDataParallel(
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 795, in __init__
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 795, in __init__
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 795, in __init__
+            model = DistributedDataParallel(model = DistributedDataParallel(model = DistributedDataParallel(
+
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 795, in __init__
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 795, in __init__
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 795, in __init__
+    model = DistributedDataParallel(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 795, in __init__
+    trainer = DistributedTrainer(config_file)
+    trainer = DistributedTrainer(config_file)    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 171, in __init__
+    
+trainer = DistributedTrainer(config_file)  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 171, in __init__
+
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 171, in __init__
+    trainer = DistributedTrainer(config_file)trainer = DistributedTrainer(config_file)  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 171, in __init__
+
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 171, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 171, in __init__
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 171, in __init__
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 171, in __init__
+        _verify_param_shape_across_processes(self.process_group, parameters)_verify_param_shape_across_processes(self.process_group, parameters)
+        
+            _verify_param_shape_across_processes(self.process_group, parameters)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/utils.py", line 265, in _verify_param_shape_across_processes
+_verify_param_shape_across_processes(self.process_group, parameters)_verify_param_shape_across_processes(self.process_group, parameters)_verify_param_shape_across_processes(self.process_group, parameters)_verify_param_shape_across_processes(self.process_group, parameters)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/utils.py", line 265, in _verify_param_shape_across_processes
+
+
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/utils.py", line 265, in _verify_param_shape_across_processes
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/utils.py", line 265, in _verify_param_shape_across_processes
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/utils.py", line 265, in _verify_param_shape_across_processes
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/utils.py", line 265, in _verify_param_shape_across_processes
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/utils.py", line 265, in _verify_param_shape_across_processes
+    _verify_param_shape_across_processes(self.process_group, parameters)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/utils.py", line 265, in _verify_param_shape_across_processes
+        self.model = self.init_model()  # Defines self.model    self.model = self.init_model()  # Defines self.model
+self.model = self.init_model()  # Defines self.model
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 720, in init_model
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 720, in init_model
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 720, in init_model
+    self.model = self.init_model()  # Defines self.model
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 720, in init_model
+    self.model = self.init_model()  # Defines self.model
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 720, in init_model
+    self.model = self.init_model()  # Defines self.model
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 720, in init_model
+self.model = self.init_model()  # Defines self.model
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 720, in init_model
+    self.model = self.init_model()  # Defines self.model
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 720, in init_model
+        return dist._verify_params_across_processes(process_group, tensors, logger)                return dist._verify_params_across_processes(process_group, tensors, logger)
+return dist._verify_params_across_processes(process_group, tensors, logger)return dist._verify_params_across_processes(process_group, tensors, logger)return dist._verify_params_across_processes(process_group, tensors, logger)return dist._verify_params_across_processes(process_group, tensors, logger)
+
+
+
+
+RuntimeError: RuntimeErrorRank 3 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+ Original exception: 
+[../third_party/gloo/gloo/transport/tcp/unbound_buffer.cc:81] Timed out waiting 1200000ms for recv operation to completeRuntimeError
+: : Rank 2 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+ Original exception: 
+[../third_party/gloo/gloo/transport/tcp/unbound_buffer.cc:81] Timed out waiting 1200000ms for recv operation to complete
+Rank 2 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+ Original exception: 
+[../third_party/gloo/gloo/transport/tcp/unbound_buffer.cc:81] Timed out waiting 1200000ms for recv operation to complete
+RuntimeErrorRuntimeError: Rank 3 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+ Original exception: 
+[../third_party/gloo/gloo/transport/tcp/unbound_buffer.cc:81] Timed out waiting 1200000ms for recv operation to completeRuntimeError
+: Rank 2 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+ Original exception: 
+[../third_party/gloo/gloo/transport/tcp/unbound_buffer.cc:81] Timed out waiting 1200000ms for recv operation to complete: 
+Rank 2 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+ Original exception: 
+[../third_party/gloo/gloo/transport/tcp/unbound_buffer.cc:81] Timed out waiting 1200000ms for recv operation to complete
+    return dist._verify_params_across_processes(process_group, tensors, logger)
+RuntimeError: Rank 3 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+ Original exception: 
+[../third_party/gloo/gloo/transport/tcp/unbound_buffer.cc:81] Timed out waiting 1200000ms for recv operation to complete
+    model = self._init_model_instance()        
+    model = self._init_model_instance()    model = self._init_model_instance()
+    model = self._init_model_instance()  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 730, in _init_model_instance
+
+model = self._init_model_instance()    
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 730, in _init_model_instance
+model = self._init_model_instance()
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 730, in _init_model_instance
+model = self._init_model_instance()  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 730, in _init_model_instance
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 730, in _init_model_instance
+
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 730, in _init_model_instance
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 730, in _init_model_instance
+    return dist._verify_params_across_processes(process_group, tensors, logger)
+RuntimeError: Rank 3 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+ Original exception: 
+[../third_party/gloo/gloo/transport/tcp/unbound_buffer.cc:81] Timed out waiting 1200000ms for recv operation to complete
+    model = self._init_model_instance()
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 730, in _init_model_instance
+    model = self._init_model(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 862, in _init_model
+    model = self._init_model(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 862, in _init_model
+    model = self._init_model(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 862, in _init_model
+    model = self._init_model(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 862, in _init_model
+    model = self._init_model(    
+model = self._init_model(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 862, in _init_model
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 862, in _init_model
+    model = self._init_model(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 862, in _init_model
+    model = self._init_model(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 862, in _init_model
+    model = DistributedDataParallel(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 795, in __init__
+    model = DistributedDataParallel(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 795, in __init__
+    model = DistributedDataParallel(    
+model = DistributedDataParallel(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 795, in __init__
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 795, in __init__
+        model = DistributedDataParallel(model = DistributedDataParallel(
+
+    model = DistributedDataParallel(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 795, in __init__
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 795, in __init__
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 795, in __init__
+    model = DistributedDataParallel(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 795, in __init__
+    _verify_param_shape_across_processes(self.process_group, parameters)
+    _verify_param_shape_across_processes(self.process_group, parameters)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/utils.py", line 265, in _verify_param_shape_across_processes
+    _verify_param_shape_across_processes(self.process_group, parameters)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/utils.py", line 265, in _verify_param_shape_across_processes
+
+        _verify_param_shape_across_processes(self.process_group, parameters)
+_verify_param_shape_across_processes(self.process_group, parameters)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/utils.py", line 265, in _verify_param_shape_across_processes
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/utils.py", line 265, in _verify_param_shape_across_processes
+_verify_param_shape_across_processes(self.process_group, parameters)      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/utils.py", line 265, in _verify_param_shape_across_processes
+
+_verify_param_shape_across_processes(self.process_group, parameters)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/utils.py", line 265, in _verify_param_shape_across_processes
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/utils.py", line 265, in _verify_param_shape_across_processes
+    _verify_param_shape_across_processes(self.process_group, parameters)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/utils.py", line 265, in _verify_param_shape_across_processes
+    return dist._verify_params_across_processes(process_group, tensors, logger)
+    return dist._verify_params_across_processes(process_group, tensors, logger)
+    return dist._verify_params_across_processes(process_group, tensors, logger)
+    return dist._verify_params_across_processes(process_group, tensors, logger)
+RuntimeError    return dist._verify_params_across_processes(process_group, tensors, logger)    
+: return dist._verify_params_across_processes(process_group, tensors, logger)
+Rank 9 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+ Original exception: 
+[../third_party/gloo/gloo/transport/tcp/unbound_buffer.cc:81] Timed out waiting 1200000ms for recv operation to complete
+RuntimeErrorRuntimeError    return dist._verify_params_across_processes(process_group, tensors, logger): : 
+Rank 8 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+ Original exception: 
+[../third_party/gloo/gloo/transport/tcp/unbound_buffer.cc:81] Timed out waiting 1200000ms for recv operation to completeRank 8 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+ Original exception: 
+[../third_party/gloo/gloo/transport/tcp/unbound_buffer.cc:81] Timed out waiting 1200000ms for recv operation to complete
+RuntimeError
+RuntimeError: : Rank 9 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+ Original exception: 
+[../third_party/gloo/gloo/transport/tcp/unbound_buffer.cc:81] Timed out waiting 1200000ms for recv operation to completeRank 9 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+ Original exception: 
+[../third_party/gloo/gloo/transport/tcp/unbound_buffer.cc:81] Timed out waiting 1200000ms for recv operation to complete
+
+RuntimeError: RuntimeErrorRank 9 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+ Original exception: 
+[../third_party/gloo/gloo/transport/tcp/unbound_buffer.cc:81] Timed out waiting 1200000ms for recv operation to complete
+: Rank 8 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+ Original exception: 
+[../third_party/gloo/gloo/transport/tcp/unbound_buffer.cc:81] Timed out waiting 1200000ms for recv operation to complete
+    return dist._verify_params_across_processes(process_group, tensors, logger)
+RuntimeError: Rank 8 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+ Original exception: 
+[../third_party/gloo/gloo/transport/tcp/unbound_buffer.cc:81] Timed out waiting 1200000ms for recv operation to complete
+[2025-01-03 04:20:52,281] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 161759 closing signal SIGTERM
+[2025-01-03 04:20:52,281] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 161760 closing signal SIGTERM
+[2025-01-03 04:20:52,281] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 161761 closing signal SIGTERM
+[2025-01-03 04:20:52,281] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 161762 closing signal SIGTERM
+[2025-01-03 04:20:52,281] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 161763 closing signal SIGTERM
+[2025-01-03 04:20:52,281] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 161764 closing signal SIGTERM
+[2025-01-03 04:20:52,335] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 327907 closing signal SIGTERM
+[2025-01-03 04:20:52,336] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 327908 closing signal SIGTERM
+[2025-01-03 04:20:52,336] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 327909 closing signal SIGTERM
+[2025-01-03 04:20:52,336] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 327910 closing signal SIGTERM
+[2025-01-03 04:20:52,336] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 327911 closing signal SIGTERM
+[2025-01-03 04:20:52,336] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 327912 closing signal SIGTERM
+[2025-01-03 04:20:52,336] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 327913 closing signal SIGTERM
+[2025-01-03 04:20:52,346] torch.distributed.elastic.multiprocessing.api: [ERROR] failed (exitcode: 1) local_rank: 0 (pid: 24653) of binary: /fsx/nouamane/miniconda/envs/2-1-cu121/bin/python
+[2025-01-03 04:20:52,374] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 320012 closing signal SIGTERM
+[2025-01-03 04:20:52,375] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 320013 closing signal SIGTERM
+[2025-01-03 04:20:52,375] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 320015 closing signal SIGTERM
+[2025-01-03 04:20:52,375] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 320017 closing signal SIGTERM
+[2025-01-03 04:20:52,375] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 320018 closing signal SIGTERM
+[2025-01-03 04:20:52,378] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 888642 closing signal SIGTERM
+[2025-01-03 04:20:52,379] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 888643 closing signal SIGTERM
+[2025-01-03 04:20:52,379] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 888645 closing signal SIGTERM
+[2025-01-03 04:20:52,379] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 888646 closing signal SIGTERM
+[2025-01-03 04:20:52,379] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 888647 closing signal SIGTERM
+[2025-01-03 04:20:52,405] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 261432 closing signal SIGTERM
+[2025-01-03 04:20:52,405] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 261433 closing signal SIGTERM
+[2025-01-03 04:20:52,405] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 261434 closing signal SIGTERM
+[2025-01-03 04:20:52,405] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 261435 closing signal SIGTERM
+[2025-01-03 04:20:52,405] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 261436 closing signal SIGTERM
+[2025-01-03 04:20:52,405] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 261437 closing signal SIGTERM
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 264, in launch_agent
+    raise ChildFailedError(
+torch.distributed.elastic.multiprocessing.errors.ChildFailedError: 
+============================================================
+/fsx/nouamane/projects/nanotron/run_train.py FAILED
+------------------------------------------------------------
+Failures:
+[1]:
+  time      : 2025-01-03_04:20:52
+  host      : ip-26-0-173-7.ec2.internal
+  rank      : 121 (local_rank: 1)
+  exitcode  : 1 (pid: 24654)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[2]:
+  time      : 2025-01-03_04:20:52
+  host      : ip-26-0-173-7.ec2.internal
+  rank      : 122 (local_rank: 2)
+  exitcode  : 1 (pid: 24655)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[3]:
+  time      : 2025-01-03_04:20:52
+  host      : ip-26-0-173-7.ec2.internal
+  rank      : 123 (local_rank: 3)
+  exitcode  : 1 (pid: 24656)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[4]:
+  time      : 2025-01-03_04:20:52
+  host      : ip-26-0-173-7.ec2.internal
+  rank      : 124 (local_rank: 4)
+  exitcode  : 1 (pid: 24657)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[5]:
+  time      : 2025-01-03_04:20:52
+  host      : ip-26-0-173-7.ec2.internal
+  rank      : 125 (local_rank: 5)
+  exitcode  : 1 (pid: 24658)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[6]:
+  time      : 2025-01-03_04:20:52
+  host      : ip-26-0-173-7.ec2.internal
+  rank      : 126 (local_rank: 6)
+  exitcode  : 1 (pid: 24659)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[7]:
+  time      : 2025-01-03_04:20:52
+  host      : ip-26-0-173-7.ec2.internal
+  rank      : 127 (local_rank: 7)
+  exitcode  : 1 (pid: 24660)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+------------------------------------------------------------
+Root Cause (first observed failure):
+[0]:
+  time      : 2025-01-03_04:20:52
+  host      : ip-26-0-173-7.ec2.internal
+  rank      : 120 (local_rank: 0)
+  exitcode  : 1 (pid: 24653)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+============================================================
+[2025-01-03 04:20:52,442] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1163536 closing signal SIGTERM
+[2025-01-03 04:20:52,442] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1163537 closing signal SIGTERM
+[2025-01-03 04:20:52,442] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1163538 closing signal SIGTERM
+[2025-01-03 04:20:52,442] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1163539 closing signal SIGTERM
+[2025-01-03 04:20:52,442] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1163540 closing signal SIGTERM
+[2025-01-03 04:20:52,442] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1163541 closing signal SIGTERM
+[2025-01-03 04:20:52,444] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 137589 closing signal SIGTERM
+[2025-01-03 04:20:52,444] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 137590 closing signal SIGTERM
+[2025-01-03 04:20:52,444] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 137591 closing signal SIGTERM
+[2025-01-03 04:20:52,444] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 137592 closing signal SIGTERM
+[2025-01-03 04:20:52,444] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 137593 closing signal SIGTERM
+[2025-01-03 04:20:52,444] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 137595 closing signal SIGTERM
+[2025-01-03 04:20:52,469] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 143073 closing signal SIGTERM
+[2025-01-03 04:20:52,469] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 143074 closing signal SIGTERM
+[2025-01-03 04:20:52,469] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 143075 closing signal SIGTERM
+[2025-01-03 04:20:52,470] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 143076 closing signal SIGTERM
+[2025-01-03 04:20:52,470] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 143077 closing signal SIGTERM
+[2025-01-03 04:20:52,470] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 143078 closing signal SIGTERM
+[2025-01-03 04:20:52,499] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 237304 closing signal SIGTERM
+[2025-01-03 04:20:52,499] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 237305 closing signal SIGTERM
+[2025-01-03 04:20:52,499] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 237306 closing signal SIGTERM
+[2025-01-03 04:20:52,499] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 237308 closing signal SIGTERM
+[2025-01-03 04:20:52,499] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 237309 closing signal SIGTERM
+[2025-01-03 04:20:52,499] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 237310 closing signal SIGTERM
+[2025-01-03 04:20:52,506] torch.distributed.elastic.multiprocessing.api: [ERROR] failed (exitcode: 1) local_rank: 0 (pid: 317938) of binary: /fsx/nouamane/miniconda/envs/2-1-cu121/bin/python
+[2025-01-03 04:20:52,554] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 22442 closing signal SIGTERM
+[2025-01-03 04:20:52,557] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 22443 closing signal SIGTERM
+[2025-01-03 04:20:52,557] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 22444 closing signal SIGTERM
+[2025-01-03 04:20:52,557] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 22445 closing signal SIGTERM
+[2025-01-03 04:20:52,557] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 22446 closing signal SIGTERM
+[2025-01-03 04:20:52,557] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 22448 closing signal SIGTERM
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 264, in launch_agent
+    raise ChildFailedError(
+torch.distributed.elastic.multiprocessing.errors.ChildFailedError: 
+============================================================
+/fsx/nouamane/projects/nanotron/run_train.py FAILED
+------------------------------------------------------------
+Failures:
+[1]:
+  time      : 2025-01-03_04:20:52
+  host      : ip-26-0-171-102.ec2.internal
+  rank      : 1 (local_rank: 1)
+  exitcode  : 1 (pid: 317939)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[2]:
+  time      : 2025-01-03_04:20:52
+  host      : ip-26-0-171-102.ec2.internal
+  rank      : 2 (local_rank: 2)
+  exitcode  : 1 (pid: 317940)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[3]:
+  time      : 2025-01-03_04:20:52
+  host      : ip-26-0-171-102.ec2.internal
+  rank      : 3 (local_rank: 3)
+  exitcode  : 1 (pid: 317941)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[4]:
+  time      : 2025-01-03_04:20:52
+  host      : ip-26-0-171-102.ec2.internal
+  rank      : 4 (local_rank: 4)
+  exitcode  : 1 (pid: 317942)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[5]:
+  time      : 2025-01-03_04:20:52
+  host      : ip-26-0-171-102.ec2.internal
+  rank      : 5 (local_rank: 5)
+  exitcode  : 1 (pid: 317943)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[6]:
+  time      : 2025-01-03_04:20:52
+  host      : ip-26-0-171-102.ec2.internal
+  rank      : 6 (local_rank: 6)
+  exitcode  : 1 (pid: 317944)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[7]:
+  time      : 2025-01-03_04:20:52
+  host      : ip-26-0-171-102.ec2.internal
+  rank      : 7 (local_rank: 7)
+  exitcode  : 1 (pid: 317945)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+------------------------------------------------------------
+Root Cause (first observed failure):
+[0]:
+  time      : 2025-01-03_04:20:52
+  host      : ip-26-0-171-102.ec2.internal
+  rank      : 0 (local_rank: 0)
+  exitcode  : 1 (pid: 317938)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+============================================================
+[2025-01-03 04:20:52,590] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3352598 closing signal SIGTERM
+[2025-01-03 04:20:52,590] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3352599 closing signal SIGTERM
+[2025-01-03 04:20:52,590] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3352600 closing signal SIGTERM
+[2025-01-03 04:20:52,591] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3352602 closing signal SIGTERM
+[2025-01-03 04:20:52,591] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3352603 closing signal SIGTERM
+[2025-01-03 04:20:52,591] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 37626 closing signal SIGTERM
+[2025-01-03 04:20:52,591] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 37627 closing signal SIGTERM
+[2025-01-03 04:20:52,591] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 37628 closing signal SIGTERM
+srun: error: ip-26-0-173-7: task 13: Exited with exit code 1
+srun: launch/slurm: _step_signal: Terminating StepId=14019149.0
+slurmstepd: error: *** STEP 14019149.0 ON ip-26-0-171-56 CANCELLED AT 2025-01-03T04:20:52 ***
+[2025-01-03 04:20:52,713] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 04:20:52,713] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 04:20:52,713] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 327908 closing signal SIGTERM
+[2025-01-03 04:20:52,713] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 344415 closing signal SIGTERM
+[2025-01-03 04:20:52,713] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 327911 closing signal SIGTERM
+[2025-01-03 04:20:52,713] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 344416 closing signal SIGTERM
+[2025-01-03 04:20:52,713] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 344417 closing signal SIGTERM
+[2025-01-03 04:20:52,713] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 327912 closing signal SIGTERM
+[2025-01-03 04:20:52,714] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 344418 closing signal SIGTERM
+[2025-01-03 04:20:52,713] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 327913 closing signal SIGTERM
+[2025-01-03 04:20:52,714] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 344419 closing signal SIGTERM
+[2025-01-03 04:20:52,714] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 344420 closing signal SIGTERM
+[2025-01-03 04:20:52,714] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 344421 closing signal SIGTERM
+[2025-01-03 04:20:52,714] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 04:20:52,714] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 04:20:52,714] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 888642 closing signal SIGTERM
+[2025-01-03 04:20:52,713] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 04:20:52,714] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 04:20:52,714] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 237304 closing signal SIGTERM
+[2025-01-03 04:20:52,714] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 237305 closing signal SIGTERM
+[2025-01-03 04:20:52,714] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 888643 closing signal SIGTERM
+[2025-01-03 04:20:52,714] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 261432 closing signal SIGTERM
+[2025-01-03 04:20:52,714] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1163536 closing signal SIGTERM
+[2025-01-03 04:20:52,714] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 237306 closing signal SIGTERM
+[2025-01-03 04:20:52,714] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 237308 closing signal SIGTERM
+[2025-01-03 04:20:52,714] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 888645 closing signal SIGTERM
+[2025-01-03 04:20:52,714] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 888646 closing signal SIGTERM
+[2025-01-03 04:20:52,714] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 261433 closing signal SIGTERM
+[2025-01-03 04:20:52,714] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1163537 closing signal SIGTERM
+[2025-01-03 04:20:52,714] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 04:20:52,714] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 04:20:52,714] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 261434 closing signal SIGTERM
+[2025-01-03 04:20:52,714] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 888647 closing signal SIGTERM
+[2025-01-03 04:20:52,714] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 237309 closing signal SIGTERM
+[2025-01-03 04:20:52,714] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 04:20:52,714] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 237310 closing signal SIGTERM
+[2025-01-03 04:20:52,714] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 261435 closing signal SIGTERM
+[2025-01-03 04:20:52,714] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1163538 closing signal SIGTERM
+[2025-01-03 04:20:52,714] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 143073 closing signal SIGTERM
+[2025-01-03 04:20:52,714] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 137589 closing signal SIGTERM
+[2025-01-03 04:20:52,714] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 161760 closing signal SIGTERM
+[2025-01-03 04:20:52,714] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 261436 closing signal SIGTERM
+[2025-01-03 04:20:52,714] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1163539 closing signal SIGTERM
+[2025-01-03 04:20:52,714] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 261437 closing signal SIGTERM
+[2025-01-03 04:20:52,714] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 143074 closing signal SIGTERM
+[2025-01-03 04:20:52,714] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 161761 closing signal SIGTERM
+[2025-01-03 04:20:52,714] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 137590 closing signal SIGTERM
+[2025-01-03 04:20:52,714] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 137591 closing signal SIGTERM
+[2025-01-03 04:20:52,714] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1163540 closing signal SIGTERM
+[2025-01-03 04:20:52,714] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 143075 closing signal SIGTERM
+[2025-01-03 04:20:52,714] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 161762 closing signal SIGTERM
+[2025-01-03 04:20:52,714] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 161763 closing signal SIGTERM
+[2025-01-03 04:20:52,714] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 137592 closing signal SIGTERM
+[2025-01-03 04:20:52,715] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1163541 closing signal SIGTERM
+[2025-01-03 04:20:52,713] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 04:20:52,714] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 137593 closing signal SIGTERM
+[2025-01-03 04:20:52,714] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 143076 closing signal SIGTERM
+[2025-01-03 04:20:52,714] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 137595 closing signal SIGTERM
+[2025-01-03 04:20:52,714] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 161764 closing signal SIGTERM
+[2025-01-03 04:20:52,714] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 143077 closing signal SIGTERM
+[2025-01-03 04:20:52,713] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 320012 closing signal SIGTERM
+[2025-01-03 04:20:52,713] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 320013 closing signal SIGTERM
+[2025-01-03 04:20:52,714] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 143078 closing signal SIGTERM
+[2025-01-03 04:20:52,713] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 320015 closing signal SIGTERM
+[2025-01-03 04:20:52,713] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 320017 closing signal SIGTERM
+[2025-01-03 04:20:52,713] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 320018 closing signal SIGTERM
+[2025-01-03 04:20:52,714] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 04:20:52,713] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 04:20:52,714] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 22442 closing signal SIGTERM
+[2025-01-03 04:20:52,713] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 37626 closing signal SIGTERM
+[2025-01-03 04:20:52,714] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 22443 closing signal SIGTERM
+[2025-01-03 04:20:52,714] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 22444 closing signal SIGTERM
+[2025-01-03 04:20:52,714] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 04:20:52,713] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 37628 closing signal SIGTERM
+[2025-01-03 04:20:52,714] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 22445 closing signal SIGTERM
+[2025-01-03 04:20:52,714] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 22446 closing signal SIGTERM
+[2025-01-03 04:20:52,714] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 22448 closing signal SIGTERM
+[2025-01-03 04:20:52,714] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3352598 closing signal SIGTERM
+[2025-01-03 04:20:52,714] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3352599 closing signal SIGTERM
+[2025-01-03 04:20:52,714] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3352600 closing signal SIGTERM
+[2025-01-03 04:20:52,714] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3352602 closing signal SIGTERM
+[2025-01-03 04:20:52,714] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3352603 closing signal SIGTERM
+[2025-01-03 04:20:52,718] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 04:20:52,724] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 155808 closing signal SIGTERM
+[2025-01-03 04:20:52,724] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 155809 closing signal SIGTERM
+[2025-01-03 04:20:52,724] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 155810 closing signal SIGTERM
+[2025-01-03 04:20:52,724] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 155811 closing signal SIGTERM
+[2025-01-03 04:20:52,724] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 155812 closing signal SIGTERM
+[2025-01-03 04:20:52,724] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 155813 closing signal SIGTERM
+[2025-01-03 04:20:52,725] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 155814 closing signal SIGTERM
+[2025-01-03 04:20:52,728] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 155815 closing signal SIGTERM
+srun: error: ip-26-0-171-102: task 3: Terminated
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 878, in _invoke_run
+    run_result = self._monitor_workers(self._worker_group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/local_elastic_agent.py", line 307, in _monitor_workers
+    result = self._pcontext.wait(0)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 288, in wait
+    return self._poll()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 668, in _poll
+    self.close()  # terminate all running procs
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 331, in close
+    self._close(death_sig=death_sig, timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 713, in _close
+    handler.proc.wait(time_to_wait)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1209, in wait
+    return self._wait(timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1953, in _wait
+    time.sleep(delay)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 37551 got signal: 15
+srun: error: ip-26-0-173-202: task 15: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 878, in _invoke_run
+    run_result = self._monitor_workers(self._worker_group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/local_elastic_agent.py", line 307, in _monitor_workers
+    result = self._pcontext.wait(0)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 288, in wait
+    return self._poll()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 668, in _poll
+    self.close()  # terminate all running procs
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 331, in close
+    self._close(death_sig=death_sig, timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 713, in _close
+    handler.proc.wait(time_to_wait)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1209, in wait
+    return self._wait(timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1953, in _wait
+    time.sleep(delay)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 888567 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 878, in _invoke_run
+    run_result = self._monitor_workers(self._worker_group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/local_elastic_agent.py", line 307, in _monitor_workers
+    result = self._pcontext.wait(0)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 288, in wait
+    return self._poll()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 668, in _poll
+    self.close()  # terminate all running procs
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 331, in close
+    self._close(death_sig=death_sig, timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 713, in _close
+    handler.proc.wait(time_to_wait)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1209, in wait
+    return self._wait(timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1953, in _wait
+    time.sleep(delay)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 161685 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 878, in _invoke_run
+    run_result = self._monitor_workers(self._worker_group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/local_elastic_agent.py", line 307, in _monitor_workers
+    result = self._pcontext.wait(0)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 288, in wait
+    return self._poll()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 668, in _poll
+    self.close()  # terminate all running procs
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 331, in close
+    self._close(death_sig=death_sig, timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 713, in _close
+    handler.proc.wait(time_to_wait)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1209, in wait
+    return self._wait(timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1953, in _wait
+    time.sleep(delay)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 327836 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 878, in _invoke_run
+    run_result = self._monitor_workers(self._worker_group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/local_elastic_agent.py", line 307, in _monitor_workers
+    result = self._pcontext.wait(0)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 288, in wait
+    return self._poll()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 668, in _poll
+    self.close()  # terminate all running procs
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 331, in close
+    self._close(death_sig=death_sig, timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 713, in _close
+    handler.proc.wait(time_to_wait)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1209, in wait
+    return self._wait(timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1953, in _wait
+    time.sleep(delay)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 319938 got signal: 15
+srun: error: ip-26-0-171-230: task 5: Exited with exit code 1
+srun: error: ip-26-0-172-147: task 11: Exited with exit code 1
+srun: error: ip-26-0-171-62: task 1: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+srun: error: ip-26-0-172-116: task 9: Exited with exit code 1
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 878, in _invoke_run
+    run_result = self._monitor_workers(self._worker_group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/local_elastic_agent.py", line 307, in _monitor_workers
+    result = self._pcontext.wait(0)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 288, in wait
+    return self._poll()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 668, in _poll
+    self.close()  # terminate all running procs
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 331, in close
+    self._close(death_sig=death_sig, timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 713, in _close
+    handler.proc.wait(time_to_wait)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1209, in wait
+    return self._wait(timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1953, in _wait
+    time.sleep(delay)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 3352525 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 878, in _invoke_run
+    run_result = self._monitor_workers(self._worker_group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/local_elastic_agent.py", line 307, in _monitor_workers
+    result = self._pcontext.wait(0)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 288, in wait
+    return self._poll()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 668, in _poll
+    self.close()  # terminate all running procs
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 331, in close
+    self._close(death_sig=death_sig, timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 713, in _close
+    handler.proc.wait(time_to_wait)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1209, in wait
+    return self._wait(timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1953, in _wait
+    time.sleep(delay)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 137516 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 878, in _invoke_run
+    run_result = self._monitor_workers(self._worker_group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/local_elastic_agent.py", line 307, in _monitor_workers
+    result = self._pcontext.wait(0)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 288, in wait
+    return self._poll()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 668, in _poll
+    self.close()  # terminate all running procs
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 331, in close
+    self._close(death_sig=death_sig, timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 713, in _close
+    handler.proc.wait(time_to_wait)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1209, in wait
+    return self._wait(timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1953, in _wait
+    time.sleep(delay)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 237232 got signal: 15
+srun: error: ip-26-0-173-121: task 14: Exited with exit code 1
+srun: error: ip-26-0-172-142: task 10: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 878, in _invoke_run
+    run_result = self._monitor_workers(self._worker_group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/local_elastic_agent.py", line 307, in _monitor_workers
+    result = self._pcontext.wait(0)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 288, in wait
+    return self._poll()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 668, in _poll
+    self.close()  # terminate all running procs
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 331, in close
+    self._close(death_sig=death_sig, timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 713, in _close
+    handler.proc.wait(time_to_wait)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1209, in wait
+    return self._wait(timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1953, in _wait
+    time.sleep(delay)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 261358 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 878, in _invoke_run
+    run_result = self._monitor_workers(self._worker_group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/local_elastic_agent.py", line 307, in _monitor_workers
+    result = self._pcontext.wait(0)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 288, in wait
+    return self._poll()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 668, in _poll
+    self.close()  # terminate all running procs
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 331, in close
+    self._close(death_sig=death_sig, timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 713, in _close
+    handler.proc.wait(time_to_wait)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1209, in wait
+    return self._wait(timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1953, in _wait
+    time.sleep(delay)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 143001 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 344341 got signal: 15
+srun: error: ip-26-0-172-73: task 8: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 878, in _invoke_run
+    run_result = self._monitor_workers(self._worker_group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/local_elastic_agent.py", line 307, in _monitor_workers
+    result = self._pcontext.wait(0)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 288, in wait
+    return self._poll()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 668, in _poll
+    self.close()  # terminate all running procs
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 331, in close
+    self._close(death_sig=death_sig, timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 713, in _close
+    handler.proc.wait(time_to_wait)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1209, in wait
+    return self._wait(timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1953, in _wait
+    time.sleep(delay)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 22365 got signal: 15
+srun: error: ip-26-0-171-249: task 6: Exited with exit code 1
+srun: error: ip-26-0-172-57: task 7: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 878, in _invoke_run
+    run_result = self._monitor_workers(self._worker_group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/local_elastic_agent.py", line 307, in _monitor_workers
+    result = self._pcontext.wait(0)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 288, in wait
+    return self._poll()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 668, in _poll
+    self.close()  # terminate all running procs
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 331, in close
+    self._close(death_sig=death_sig, timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 713, in _close
+    handler.proc.wait(time_to_wait)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1209, in wait
+    return self._wait(timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1953, in _wait
+    time.sleep(delay)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 1163462 got signal: 15
+srun: error: ip-26-0-171-56: task 0: Exited with exit code 1
+srun: error: ip-26-0-172-252: task 12: Exited with exit code 1
+srun: error: ip-26-0-171-168: task 4: Exited with exit code 1
+[2025-01-03 04:20:57,008] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-171-88.ec2.internal_155734_0' has failed to send a keep-alive heartbeat to the rendezvous '14019149' due to an error of type RendezvousConnectionError.
+[2025-01-03 04:21:02,011] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-171-88.ec2.internal_155734_0' has failed to send a keep-alive heartbeat to the rendezvous '14019149' due to an error of type RendezvousConnectionError.
+[2025-01-03 04:21:07,012] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-171-88.ec2.internal_155734_0' has failed to send a keep-alive heartbeat to the rendezvous '14019149' due to an error of type RendezvousConnectionError.
+[2025-01-03 04:21:12,014] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-171-88.ec2.internal_155734_0' has failed to send a keep-alive heartbeat to the rendezvous '14019149' due to an error of type RendezvousConnectionError.
+[2025-01-03 04:21:17,016] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-171-88.ec2.internal_155734_0' has failed to send a keep-alive heartbeat to the rendezvous '14019149' due to an error of type RendezvousConnectionError.
+[2025-01-03 04:21:22,017] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-171-88.ec2.internal_155734_0' has failed to send a keep-alive heartbeat to the rendezvous '14019149' due to an error of type RendezvousConnectionError.
+slurmstepd: error: *** STEP 14019149.0 STEPD TERMINATED ON ip-26-0-171-88 AT 2025-01-03T04:24:22 DUE TO JOB NOT ENDING WITH SIGNALS ***
+srun: error: ip-26-0-171-88: task 2: Killed
+srun: Force Terminated StepId=14019149.0
diff --git a/logs/14019182-bench_469G_dp16_tp32_pp1_acc1_mbs16_seq4096_zero0_tpmodeRED_vocab131k.out b/logs/14019182-bench_469G_dp16_tp32_pp1_acc1_mbs16_seq4096_zero0_tpmodeRED_vocab131k.out
new file mode 100644
index 0000000000000000000000000000000000000000..381c990645f85e143dabbcbd402df8be35c49b56
--- /dev/null
+++ b/logs/14019182-bench_469G_dp16_tp32_pp1_acc1_mbs16_seq4096_zero0_tpmodeRED_vocab131k.out
@@ -0,0 +1,13065 @@
++ '[' -z 14019182 ']'
++ source /etc/profile.d/modules.sh
+++ . /usr/share/modules/init/bash
++++ unset _mlshdbg
++++ '[' 0 = 1 ']'
++++ unset _mlre _mlIFS
++++ '[' -n x ']'
++++ _mlIFS=' 	
+'
++++ IFS=' '
++++ '[' -n '' ']'
+++++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash autoinit
++++ _mlcode='module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+MODULEPATH_modshare=/etc/environment-modules/modules:1:/usr/share/modules/\$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1; export MODULEPATH_modshare;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+LOADEDMODULES=; export LOADEDMODULES;
+MODULEPATH=/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/\$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles; export MODULEPATH;
+test 0;'
++++ _mlret=0
++++ '[' -n x ']'
++++ IFS=' 	
+'
++++ unset _mlIFS
++++ unset _mlre _mlv _mlrv
++++ '[' 0 -eq 0 ']'
++++ eval 'module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+MODULEPATH_modshare=/etc/environment-modules/modules:1:/usr/share/modules/\$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1; export MODULEPATH_modshare;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+LOADEDMODULES=; export LOADEDMODULES;
+MODULEPATH=/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/\$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles; export MODULEPATH;
+test 0;'
+++++ MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl
+++++ export MODULES_CMD
+++++ ENV=/usr/share/modules/init/profile.sh
+++++ export ENV
+++++ MODULEPATH_modshare='/etc/environment-modules/modules:1:/usr/share/modules/$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1'
+++++ export MODULEPATH_modshare
+++++ BASH_ENV=/usr/share/modules/init/bash
+++++ export BASH_ENV
+++++ MODULESHOME=/usr/share/modules
+++++ export MODULESHOME
+++++ LOADEDMODULES=
+++++ export LOADEDMODULES
+++++ MODULEPATH='/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles'
+++++ export MODULEPATH
+++++ test 0
++++ '[' 0 = 1 ']'
++++ '[' -t 2 ']'
++++ export -f module
++++ export -f switchml
++++ '[' 5 -ge 3 ']'
++++ [[ ehxB =~ i ]]
++++ [[ ! :/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin: =~ :/usr/bin: ]]
++++ '[' '!' -n '' ']'
++++ MANPATH=:
++++ export MANPATH
+++++ manpath
++++ [[ ! :/fsx/nouamane/miniconda/envs/2-1-cu121/man:/fsx/nouamane/miniconda/envs/2-1-cu121/share/man:/fsx/nouamane/miniconda/man:/fsx/nouamane/miniconda/share/man:/usr/local/man:/usr/local/share/man:/usr/share/man:/opt/slurm/share/man:/admin/home/nouamane/.local/share/man:/admin/home/nouamane/.fzf/man:: =~ :/usr/share/man: ]]
++++ unset _mlcode _mlret
++++ '[' -n '' ']'
++ module load cuda/12.1
++ unset _mlshdbg
++ '[' 0 = 1 ']'
++ unset _mlre _mlIFS
++ '[' -n x ']'
++ _mlIFS=' 	
+'
++ IFS=' '
++ '[' -n '' ']'
+++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash load cuda/12.1
++ eval 'CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include;' export 'CPATH;
+LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:;' export 'LD_LIBRARY_PATH;
+FI_EFA_FORK_SAFE=1;' export 'FI_EFA_FORK_SAFE;
+MANPATH=/usr/local/cuda-12.1/share/man::;' export 'MANPATH;
+LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64;' export 'LIBRARY_PATH;
+_LMFILES_=/usr/share/modules/modulefiles/cuda/12.1;' export '_LMFILES_;
+LOADEDMODULES=cuda/12.1;' export 'LOADEDMODULES;
+MPI_PATH=/opt/amazon/openmpi;' export 'MPI_PATH;
+NCCL_HOME_modshare=/opt/nccl/build:1;' export 'NCCL_HOME_modshare;
+NCCL_PROTO=simple;' export 'NCCL_PROTO;
+MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1;' export 'MANPATH_modshare;
+LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1;' export 'LIBRARY_PATH_modshare;
+NCCL_SOCKET_IFNAME=enp;' export 'NCCL_SOCKET_IFNAME;
+AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl;' export 'AWS_OFI_NCCL_HOME;
+NCCL_HOME=/opt/nccl/build;' export 'NCCL_HOME;
+FI_PROVIDER=efa;' export 'FI_PROVIDER;
+AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1;' export 'AWS_OFI_NCCL_HOME_modshare;
+CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1;' export 'CPATH_modshare;
+LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2::1:/opt/nccl/build/lib:1:/opt/aws-ofi-nccl/lib:2:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/efa/lib:1;' export 'LD_LIBRARY_PATH_modshare;
+FI_EFA_ENABLE_SHM_TRANSFER=1;' export 'FI_EFA_ENABLE_SHM_TRANSFER;
+_LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1;' export '_LMFILES__modshare;
+LOADEDMODULES_modshare=cuda/12.1:1;' export 'LOADEDMODULES_modshare;
+MPI_PATH_modshare=/opt/amazon/openmpi:1;' export 'MPI_PATH_modshare;
+PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin;' export 'PATH;
+CUDA_HOME=/usr/local/cuda-12.1;' export 'CUDA_HOME;
+PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/opt/slurm/bin:1:/usr/sbin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1;' export 'PATH_modshare;
+test' '0;'
+++ CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include
+++ export CPATH
+++ LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:
+++ export LD_LIBRARY_PATH
+++ FI_EFA_FORK_SAFE=1
+++ export FI_EFA_FORK_SAFE
+++ MANPATH=/usr/local/cuda-12.1/share/man::
+++ export MANPATH
+++ LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64
+++ export LIBRARY_PATH
+++ _LMFILES_=/usr/share/modules/modulefiles/cuda/12.1
+++ export _LMFILES_
+++ LOADEDMODULES=cuda/12.1
+++ export LOADEDMODULES
+++ MPI_PATH=/opt/amazon/openmpi
+++ export MPI_PATH
+++ NCCL_HOME_modshare=/opt/nccl/build:1
+++ export NCCL_HOME_modshare
+++ NCCL_PROTO=simple
+++ export NCCL_PROTO
+++ MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1
+++ export MANPATH_modshare
+++ LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1
+++ export LIBRARY_PATH_modshare
+++ NCCL_SOCKET_IFNAME=enp
+++ export NCCL_SOCKET_IFNAME
+++ AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl
+++ export AWS_OFI_NCCL_HOME
+++ NCCL_HOME=/opt/nccl/build
+++ export NCCL_HOME
+++ FI_PROVIDER=efa
+++ export FI_PROVIDER
+++ AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1
+++ export AWS_OFI_NCCL_HOME_modshare
+++ CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1
+++ export CPATH_modshare
+++ LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2::1:/opt/nccl/build/lib:1:/opt/aws-ofi-nccl/lib:2:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/efa/lib:1
+++ export LD_LIBRARY_PATH_modshare
+++ FI_EFA_ENABLE_SHM_TRANSFER=1
+++ export FI_EFA_ENABLE_SHM_TRANSFER
+++ _LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1
+++ export _LMFILES__modshare
+++ LOADEDMODULES_modshare=cuda/12.1:1
+++ export LOADEDMODULES_modshare
+++ MPI_PATH_modshare=/opt/amazon/openmpi:1
+++ export MPI_PATH_modshare
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
+++ export PATH
+++ CUDA_HOME=/usr/local/cuda-12.1
+++ export CUDA_HOME
+++ PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/opt/slurm/bin:1:/usr/sbin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1
+++ export PATH_modshare
+++ test 0
++ _mlstatus=0
++ '[' -n x ']'
++ IFS=' 	
+'
++ unset _mlre _mlv _mlrv _mlIFS
++ '[' -n '' ']'
++ unset _mlshdbg
++ return 0
++ source /fsx/nouamane/miniconda/bin/activate
+++ _CONDA_ROOT=/fsx/nouamane/miniconda
+++ . /fsx/nouamane/miniconda/etc/profile.d/conda.sh
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ '[' -z x ']'
+++ conda activate
+++ local cmd=activate
+++ case "$cmd" in
+++ __conda_activate activate
+++ '[' -n '' ']'
+++ local ask_conda
++++ PS1=
++++ __conda_exe shell.posix activate
++++ /fsx/nouamane/miniconda/bin/conda shell.posix activate
+Error while loading conda entry point: conda-libmamba-solver (module 'libmambapy' has no attribute 'QueryFormat')
+++ ask_conda='. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''2'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_1='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/etc/conda/activate.d/libxml2_activate.sh"'
+++ eval '. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''2'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_1='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/etc/conda/activate.d/libxml2_activate.sh"'
++++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh
+++++ test -n ''
+++++ unset XML_CATALOG_FILES
+++++ unset xml_catalog_files_libxml2
++++ PS1='(base) '
++++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
++++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
++++ export CONDA_PREFIX=/fsx/nouamane/miniconda
++++ CONDA_PREFIX=/fsx/nouamane/miniconda
++++ export CONDA_SHLVL=2
++++ CONDA_SHLVL=2
++++ export CONDA_DEFAULT_ENV=base
++++ CONDA_DEFAULT_ENV=base
++++ export 'CONDA_PROMPT_MODIFIER=(base) '
++++ CONDA_PROMPT_MODIFIER='(base) '
++++ export CONDA_PREFIX_1=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ CONDA_PREFIX_1=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ . /fsx/nouamane/miniconda/etc/conda/activate.d/libxml2_activate.sh
+++++ test -n ''
+++++ xml_catalog_files_libxml2=
+++++ XML_CATALOG_FILES=
+++++ conda_catalog_files=
+++++ ifs_libxml2=' 	
+'
+++++ IFS=' '
+++++ rem=/fsx/nouamane/miniconda
+++++ for pre in ${rem}
+++++ test '' = /fsx/nouamane/miniconda
+++++ conda_catalog_files=/fsx/nouamane/miniconda
+++++ rem=
+++++ IFS=' 	
+'
+++++ conda_catalog_files='file:///fsx/nouamane/miniconda/etc/xml/catalog file:///etc/xml/catalog'
+++++ export 'XML_CATALOG_FILES=file:///fsx/nouamane/miniconda/etc/xml/catalog file:///etc/xml/catalog'
+++++ XML_CATALOG_FILES='file:///fsx/nouamane/miniconda/etc/xml/catalog file:///etc/xml/catalog'
+++++ unset conda_catalog_files ifs_libxml2 rem
+++ __conda_hashr
+++ '[' -n '' ']'
+++ '[' -n '' ']'
+++ hash -r
++ conda activate 2-1-cu121
++ local cmd=activate
++ case "$cmd" in
++ __conda_activate activate 2-1-cu121
++ '[' -n '' ']'
++ local ask_conda
+++ PS1='(base) '
+++ __conda_exe shell.posix activate 2-1-cu121
+++ /fsx/nouamane/miniconda/bin/conda shell.posix activate 2-1-cu121
+Error while loading conda entry point: conda-libmamba-solver (module 'libmambapy' has no attribute 'QueryFormat')
++ ask_conda='. "/fsx/nouamane/miniconda/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_2='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
++ eval '. "/fsx/nouamane/miniconda/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_2='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
+++ . /fsx/nouamane/miniconda/etc/conda/deactivate.d/libxml2_deactivate.sh
++++ test -n ''
++++ unset XML_CATALOG_FILES
++++ unset xml_catalog_files_libxml2
+++ PS1='(2-1-cu121) '
+++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
+++ export CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ export CONDA_SHLVL=3
+++ CONDA_SHLVL=3
+++ export CONDA_DEFAULT_ENV=2-1-cu121
+++ CONDA_DEFAULT_ENV=2-1-cu121
+++ export 'CONDA_PROMPT_MODIFIER=(2-1-cu121) '
+++ CONDA_PROMPT_MODIFIER='(2-1-cu121) '
+++ export CONDA_PREFIX_2=/fsx/nouamane/miniconda
+++ CONDA_PREFIX_2=/fsx/nouamane/miniconda
+++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ export _CE_M=
+++ _CE_M=
+++ export _CE_CONDA=
+++ _CE_CONDA=
+++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh
++++ test -n ''
++++ xml_catalog_files_libxml2=
++++ XML_CATALOG_FILES=
++++ conda_catalog_files=
++++ ifs_libxml2=' 	
+'
++++ IFS=' '
++++ rem=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ for pre in ${rem}
++++ test '' = /fsx/nouamane/miniconda/envs/2-1-cu121
++++ conda_catalog_files=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ rem=
++++ IFS=' 	
+'
++++ conda_catalog_files='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ export 'XML_CATALOG_FILES=file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ XML_CATALOG_FILES='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ unset conda_catalog_files ifs_libxml2 rem
++ __conda_hashr
++ '[' -n '' ']'
++ '[' -n '' ']'
++ hash -r
++ export PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
++ PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
++ '[' -z '' ']'
+++ scontrol show hostnames 'ip-26-0-160-[100,103,192],ip-26-0-163-236,ip-26-0-164-[0,18,45,75,187,207,236],ip-26-0-165-[24,38,59,131,164,202,213],ip-26-0-166-[15,36,125,214,244],ip-26-0-167-[9,51,175,177,217,245],ip-26-0-168-[30,34,52,95,120,238],ip-26-0-169-[86,132,139,207,239,247],ip-26-0-170-[31,132,143,160],ip-26-0-171-[21,56,62,88,102,168,230,249],ip-26-0-172-[57,73,116,142,147,252],ip-26-0-173-[7,121,202,246],ip-26-0-174-36'
++ export 'NODELIST=ip-26-0-160-100
+ip-26-0-160-103
+ip-26-0-160-192
+ip-26-0-163-236
+ip-26-0-164-0
+ip-26-0-164-18
+ip-26-0-164-45
+ip-26-0-164-75
+ip-26-0-164-187
+ip-26-0-164-207
+ip-26-0-164-236
+ip-26-0-165-24
+ip-26-0-165-38
+ip-26-0-165-59
+ip-26-0-165-131
+ip-26-0-165-164
+ip-26-0-165-202
+ip-26-0-165-213
+ip-26-0-166-15
+ip-26-0-166-36
+ip-26-0-166-125
+ip-26-0-166-214
+ip-26-0-166-244
+ip-26-0-167-9
+ip-26-0-167-51
+ip-26-0-167-175
+ip-26-0-167-177
+ip-26-0-167-217
+ip-26-0-167-245
+ip-26-0-168-30
+ip-26-0-168-34
+ip-26-0-168-52
+ip-26-0-168-95
+ip-26-0-168-120
+ip-26-0-168-238
+ip-26-0-169-86
+ip-26-0-169-132
+ip-26-0-169-139
+ip-26-0-169-207
+ip-26-0-169-239
+ip-26-0-169-247
+ip-26-0-170-31
+ip-26-0-170-132
+ip-26-0-170-143
+ip-26-0-170-160
+ip-26-0-171-21
+ip-26-0-171-56
+ip-26-0-171-62
+ip-26-0-171-88
+ip-26-0-171-102
+ip-26-0-171-168
+ip-26-0-171-230
+ip-26-0-171-249
+ip-26-0-172-57
+ip-26-0-172-73
+ip-26-0-172-116
+ip-26-0-172-142
+ip-26-0-172-147
+ip-26-0-172-252
+ip-26-0-173-7
+ip-26-0-173-121
+ip-26-0-173-202
+ip-26-0-173-246
+ip-26-0-174-36'
++ NODELIST='ip-26-0-160-100
+ip-26-0-160-103
+ip-26-0-160-192
+ip-26-0-163-236
+ip-26-0-164-0
+ip-26-0-164-18
+ip-26-0-164-45
+ip-26-0-164-75
+ip-26-0-164-187
+ip-26-0-164-207
+ip-26-0-164-236
+ip-26-0-165-24
+ip-26-0-165-38
+ip-26-0-165-59
+ip-26-0-165-131
+ip-26-0-165-164
+ip-26-0-165-202
+ip-26-0-165-213
+ip-26-0-166-15
+ip-26-0-166-36
+ip-26-0-166-125
+ip-26-0-166-214
+ip-26-0-166-244
+ip-26-0-167-9
+ip-26-0-167-51
+ip-26-0-167-175
+ip-26-0-167-177
+ip-26-0-167-217
+ip-26-0-167-245
+ip-26-0-168-30
+ip-26-0-168-34
+ip-26-0-168-52
+ip-26-0-168-95
+ip-26-0-168-120
+ip-26-0-168-238
+ip-26-0-169-86
+ip-26-0-169-132
+ip-26-0-169-139
+ip-26-0-169-207
+ip-26-0-169-239
+ip-26-0-169-247
+ip-26-0-170-31
+ip-26-0-170-132
+ip-26-0-170-143
+ip-26-0-170-160
+ip-26-0-171-21
+ip-26-0-171-56
+ip-26-0-171-62
+ip-26-0-171-88
+ip-26-0-171-102
+ip-26-0-171-168
+ip-26-0-171-230
+ip-26-0-171-249
+ip-26-0-172-57
+ip-26-0-172-73
+ip-26-0-172-116
+ip-26-0-172-142
+ip-26-0-172-147
+ip-26-0-172-252
+ip-26-0-173-7
+ip-26-0-173-121
+ip-26-0-173-202
+ip-26-0-173-246
+ip-26-0-174-36'
+++ head -n1
+++ scontrol show hostnames 'ip-26-0-160-[100,103,192],ip-26-0-163-236,ip-26-0-164-[0,18,45,75,187,207,236],ip-26-0-165-[24,38,59,131,164,202,213],ip-26-0-166-[15,36,125,214,244],ip-26-0-167-[9,51,175,177,217,245],ip-26-0-168-[30,34,52,95,120,238],ip-26-0-169-[86,132,139,207,239,247],ip-26-0-170-[31,132,143,160],ip-26-0-171-[21,56,62,88,102,168,230,249],ip-26-0-172-[57,73,116,142,147,252],ip-26-0-173-[7,121,202,246],ip-26-0-174-36'
++ export MASTER_NODE=ip-26-0-160-100
++ MASTER_NODE=ip-26-0-160-100
++ export MASTER_PORT=12356
++ MASTER_PORT=12356
++ export NNODES=64
++ NNODES=64
++ export GPUS_PER_NODE=8
++ GPUS_PER_NODE=8
++ export WORLD_SIZE=512
++ WORLD_SIZE=512
++ export CUDA_DEVICE_MAX_CONNECTIONS=1
++ CUDA_DEVICE_MAX_CONNECTIONS=1
++ export NCCL_DEBUG=WARN
++ NCCL_DEBUG=WARN
++ export NANOTRON_BENCHMARK=1
++ NANOTRON_BENCHMARK=1
++ export WANDB_MODE=disabled
++ WANDB_MODE=disabled
++ export TORCH_NCCL_ASYNC_ERROR_HANDLING=1
++ TORCH_NCCL_ASYNC_ERROR_HANDLING=1
++ export TORCH_DISTRIBUTED_DEBUG=DETAIL
++ TORCH_DISTRIBUTED_DEBUG=DETAIL
++ '[' -z '' ']'
++ echo '=== GPU Topology ==='
+=== GPU Topology ===
++ nvidia-smi topo -m
+	GPU0	GPU1	GPU2	GPU3	GPU4	GPU5	GPU6	GPU7	CPU Affinity	NUMA Affinity	GPU NUMA ID
+GPU0	 X 	NV18	NV18	NV18	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU1	NV18	 X 	NV18	NV18	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU2	NV18	NV18	 X 	NV18	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU3	NV18	NV18	NV18	 X 	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU4	NV18	NV18	NV18	NV18	 X 	NV18	NV18	NV18	48-95	1		N/A
+GPU5	NV18	NV18	NV18	NV18	NV18	 X 	NV18	NV18	48-95	1		N/A
+GPU6	NV18	NV18	NV18	NV18	NV18	NV18	 X 	NV18	48-95	1		N/A
+GPU7	NV18	NV18	NV18	NV18	NV18	NV18	NV18	 X 	48-95	1		N/A
+
+Legend:
+
+  X    = Self
+  SYS  = Connection traversing PCIe as well as the SMP interconnect between NUMA nodes (e.g., QPI/UPI)
+  NODE = Connection traversing PCIe as well as the interconnect between PCIe Host Bridges within a NUMA node
+  PHB  = Connection traversing PCIe as well as a PCIe Host Bridge (typically the CPU)
+  PXB  = Connection traversing multiple PCIe bridges (without traversing the PCIe Host Bridge)
+  PIX  = Connection traversing at most a single PCIe bridge
+  NV#  = Connection traversing a bonded set of # NVLinks
++ echo ==================
+==================
++ export SRUN_ALLOC_ARGS=
++ SRUN_ALLOC_ARGS=
++ echo 'Master node: ip-26-0-160-100'
+Master node: ip-26-0-160-100
++ echo 'All nodes: ip-26-0-160-100
+ip-26-0-160-103
+ip-26-0-160-192
+ip-26-0-163-236
+ip-26-0-164-0
+ip-26-0-164-18
+ip-26-0-164-45
+ip-26-0-164-75
+ip-26-0-164-187
+ip-26-0-164-207
+ip-26-0-164-236
+ip-26-0-165-24
+ip-26-0-165-38
+ip-26-0-165-59
+ip-26-0-165-131
+ip-26-0-165-164
+ip-26-0-165-202
+ip-26-0-165-213
+ip-26-0-166-15
+ip-26-0-166-36
+ip-26-0-166-125
+ip-26-0-166-214
+ip-26-0-166-244
+ip-26-0-167-9
+ip-26-0-167-51
+ip-26-0-167-175
+ip-26-0-167-177
+ip-26-0-167-217
+ip-26-0-167-245
+ip-26-0-168-30
+ip-26-0-168-34
+ip-26-0-168-52
+ip-26-0-168-95
+ip-26-0-168-120
+ip-26-0-168-238
+ip-26-0-169-86
+ip-26-0-169-132
+ip-26-0-169-139
+ip-26-0-169-207
+ip-26-0-169-239
+ip-26-0-169-247
+ip-26-0-170-31
+ip-26-0-170-132
+ip-26-0-170-143
+ip-26-0-170-160
+ip-26-0-171-21
+ip-26-0-171-56
+ip-26-0-171-62
+ip-26-0-171-88
+ip-26-0-171-102
+ip-26-0-171-168
+ip-26-0-171-230
+ip-26-0-171-249
+ip-26-0-172-57
+ip-26-0-172-73
+ip-26-0-172-116
+ip-26-0-172-142
+ip-26-0-172-147
+ip-26-0-172-252
+ip-26-0-173-7
+ip-26-0-173-121
+ip-26-0-173-202
+ip-26-0-173-246
+ip-26-0-174-36'
+All nodes: ip-26-0-160-100
+ip-26-0-160-103
+ip-26-0-160-192
+ip-26-0-163-236
+ip-26-0-164-0
+ip-26-0-164-18
+ip-26-0-164-45
+ip-26-0-164-75
+ip-26-0-164-187
+ip-26-0-164-207
+ip-26-0-164-236
+ip-26-0-165-24
+ip-26-0-165-38
+ip-26-0-165-59
+ip-26-0-165-131
+ip-26-0-165-164
+ip-26-0-165-202
+ip-26-0-165-213
+ip-26-0-166-15
+ip-26-0-166-36
+ip-26-0-166-125
+ip-26-0-166-214
+ip-26-0-166-244
+ip-26-0-167-9
+ip-26-0-167-51
+ip-26-0-167-175
+ip-26-0-167-177
+ip-26-0-167-217
+ip-26-0-167-245
+ip-26-0-168-30
+ip-26-0-168-34
+ip-26-0-168-52
+ip-26-0-168-95
+ip-26-0-168-120
+ip-26-0-168-238
+ip-26-0-169-86
+ip-26-0-169-132
+ip-26-0-169-139
+ip-26-0-169-207
+ip-26-0-169-239
+ip-26-0-169-247
+ip-26-0-170-31
+ip-26-0-170-132
+ip-26-0-170-143
+ip-26-0-170-160
+ip-26-0-171-21
+ip-26-0-171-56
+ip-26-0-171-62
+ip-26-0-171-88
+ip-26-0-171-102
+ip-26-0-171-168
+ip-26-0-171-230
+ip-26-0-171-249
+ip-26-0-172-57
+ip-26-0-172-73
+ip-26-0-172-116
+ip-26-0-172-142
+ip-26-0-172-147
+ip-26-0-172-252
+ip-26-0-173-7
+ip-26-0-173-121
+ip-26-0-173-202
+ip-26-0-173-246
+ip-26-0-174-36
++ echo 'World size: 512'
+World size: 512
++ '[' -n '' ']'
++ srun --wait=0 --kill-on-bad-exit=1 torchrun --nnodes=64 --nproc_per_node=8 --rdzv_id=14019182 --rdzv_backend=c10d --rdzv_endpoint=ip-26-0-160-100:12356 --max_restarts 0 --rdzv_conf timeout=60 /fsx/nouamane/projects/nanotron/run_train.py --config-file benchmark/configs/config_469G_dp16_tp32_pp1_acc1_mbs16_seq4096_zero0_tpmodeRED_vocab131k.yaml
+[2025-01-02 19:08:23,976] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-02 19:08:23,976] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-02 19:08:23,977] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-02 19:08:23,977] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-02 19:08:23,977] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-02 19:08:23,977] torch.distributed.run: [WARNING] 
+[2025-01-02 19:08:23,977] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 19:08:23,977] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-02 19:08:23,977] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 19:08:23,976] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-02 19:08:23,978] torch.distributed.run: [WARNING] 
+[2025-01-02 19:08:23,978] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 19:08:23,978] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-02 19:08:23,978] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 19:08:23,977] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-02 19:08:23,976] torch.distributed.run: [WARNING] 
+[2025-01-02 19:08:23,976] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 19:08:23,976] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-02 19:08:23,976] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 19:08:23,977] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-02 19:08:23,977] torch.distributed.run: [WARNING] 
+[2025-01-02 19:08:23,977] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 19:08:23,977] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-02 19:08:23,977] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 19:08:23,976] torch.distributed.run: [WARNING] 
+[2025-01-02 19:08:23,976] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 19:08:23,976] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-02 19:08:23,976] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 19:08:23,977] torch.distributed.run: [WARNING] 
+[2025-01-02 19:08:23,977] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 19:08:23,977] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-02 19:08:23,977] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 19:08:23,977] torch.distributed.run: [WARNING] 
+[2025-01-02 19:08:23,977] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 19:08:23,977] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-02 19:08:23,977] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 19:08:23,977] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-02 19:08:23,977] torch.distributed.run: [WARNING] 
+[2025-01-02 19:08:23,977] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 19:08:23,977] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-02 19:08:23,977] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 19:08:23,977] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-02 19:08:23,975] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-02 19:08:23,975] torch.distributed.run: [WARNING] 
+[2025-01-02 19:08:23,975] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 19:08:23,975] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-02 19:08:23,975] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 19:08:23,977] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-02 19:08:23,975] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-02 19:08:23,977] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-02 19:08:23,977] torch.distributed.run: [WARNING] 
+[2025-01-02 19:08:23,977] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 19:08:23,977] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-02 19:08:23,977] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 19:08:23,975] torch.distributed.run: [WARNING] 
+[2025-01-02 19:08:23,975] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 19:08:23,975] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-02 19:08:23,975] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 19:08:23,978] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-02 19:08:23,978] torch.distributed.run: [WARNING] 
+[2025-01-02 19:08:23,978] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 19:08:23,978] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-02 19:08:23,978] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 19:08:23,977] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-02 19:08:23,977] torch.distributed.run: [WARNING] 
+[2025-01-02 19:08:23,977] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 19:08:23,977] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-02 19:08:23,977] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 19:08:23,977] torch.distributed.run: [WARNING] 
+[2025-01-02 19:08:23,977] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 19:08:23,977] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-02 19:08:23,977] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 19:08:23,979] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-02 19:08:23,979] torch.distributed.run: [WARNING] 
+[2025-01-02 19:08:23,979] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 19:08:23,979] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-02 19:08:23,979] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 19:08:23,980] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-02 19:08:23,980] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-02 19:08:23,980] torch.distributed.run: [WARNING] 
+[2025-01-02 19:08:23,980] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 19:08:23,980] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-02 19:08:23,980] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 19:08:23,980] torch.distributed.run: [WARNING] 
+[2025-01-02 19:08:23,980] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 19:08:23,980] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-02 19:08:23,980] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 19:08:23,981] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-02 19:08:23,981] torch.distributed.run: [WARNING] 
+[2025-01-02 19:08:23,981] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 19:08:23,981] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-02 19:08:23,981] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 19:08:23,982] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-02 19:08:23,982] torch.distributed.run: [WARNING] 
+[2025-01-02 19:08:23,982] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 19:08:23,982] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-02 19:08:23,982] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 19:08:23,983] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-02 19:08:23,983] torch.distributed.run: [WARNING] 
+[2025-01-02 19:08:23,983] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 19:08:23,983] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-02 19:08:23,983] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 19:08:23,984] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-02 19:08:23,984] torch.distributed.run: [WARNING] 
+[2025-01-02 19:08:23,984] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 19:08:23,984] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-02 19:08:23,984] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 19:08:23,983] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-02 19:08:23,983] torch.distributed.run: [WARNING] 
+[2025-01-02 19:08:23,983] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 19:08:23,983] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-02 19:08:23,983] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 19:08:23,983] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-02 19:08:23,983] torch.distributed.run: [WARNING] 
+[2025-01-02 19:08:23,983] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 19:08:23,983] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-02 19:08:23,983] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 19:08:23,984] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-02 19:08:23,984] torch.distributed.run: [WARNING] 
+[2025-01-02 19:08:23,984] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 19:08:23,984] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-02 19:08:23,984] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 19:08:23,984] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-02 19:08:23,984] torch.distributed.run: [WARNING] 
+[2025-01-02 19:08:23,984] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 19:08:23,984] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-02 19:08:23,984] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 19:08:23,986] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-02 19:08:23,986] torch.distributed.run: [WARNING] 
+[2025-01-02 19:08:23,986] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 19:08:23,986] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-02 19:08:23,986] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 19:08:23,985] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-02 19:08:23,985] torch.distributed.run: [WARNING] 
+[2025-01-02 19:08:23,985] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 19:08:23,985] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-02 19:08:23,985] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 19:08:23,985] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-02 19:08:23,986] torch.distributed.run: [WARNING] 
+[2025-01-02 19:08:23,986] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 19:08:23,986] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-02 19:08:23,986] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 19:08:23,986] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-02 19:08:23,986] torch.distributed.run: [WARNING] 
+[2025-01-02 19:08:23,986] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 19:08:23,986] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-02 19:08:23,986] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 19:08:23,986] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-02 19:08:23,987] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-02 19:08:23,987] torch.distributed.run: [WARNING] 
+[2025-01-02 19:08:23,987] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 19:08:23,987] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-02 19:08:23,987] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 19:08:23,987] torch.distributed.run: [WARNING] 
+[2025-01-02 19:08:23,987] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 19:08:23,987] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-02 19:08:23,987] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 19:08:23,987] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-02 19:08:23,987] torch.distributed.run: [WARNING] 
+[2025-01-02 19:08:23,987] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 19:08:23,987] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-02 19:08:23,987] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 19:08:23,988] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-02 19:08:23,988] torch.distributed.run: [WARNING] 
+[2025-01-02 19:08:23,988] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 19:08:23,988] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-02 19:08:23,988] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 19:08:23,987] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-02 19:08:23,987] torch.distributed.run: [WARNING] 
+[2025-01-02 19:08:23,987] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 19:08:23,987] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-02 19:08:23,987] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 19:08:23,989] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-02 19:08:23,989] torch.distributed.run: [WARNING] 
+[2025-01-02 19:08:23,989] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 19:08:23,989] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-02 19:08:23,989] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 19:08:23,989] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-02 19:08:23,989] torch.distributed.run: [WARNING] 
+[2025-01-02 19:08:23,989] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 19:08:23,989] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-02 19:08:23,989] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 19:08:23,991] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-02 19:08:23,991] torch.distributed.run: [WARNING] 
+[2025-01-02 19:08:23,991] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 19:08:23,991] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-02 19:08:23,991] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 19:08:23,990] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-02 19:08:23,990] torch.distributed.run: [WARNING] 
+[2025-01-02 19:08:23,990] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 19:08:23,990] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-02 19:08:23,990] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 19:08:23,994] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-02 19:08:23,994] torch.distributed.run: [WARNING] 
+[2025-01-02 19:08:23,994] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 19:08:23,994] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-02 19:08:23,994] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 19:08:23,995] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-02 19:08:23,995] torch.distributed.run: [WARNING] 
+[2025-01-02 19:08:23,995] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 19:08:23,995] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-02 19:08:23,995] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 19:08:23,995] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-02 19:08:23,995] torch.distributed.run: [WARNING] 
+[2025-01-02 19:08:23,995] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 19:08:23,995] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-02 19:08:23,995] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 19:08:23,996] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-02 19:08:23,996] torch.distributed.run: [WARNING] 
+[2025-01-02 19:08:23,996] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 19:08:23,996] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-02 19:08:23,996] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 19:08:24,000] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-02 19:08:24,000] torch.distributed.run: [WARNING] 
+[2025-01-02 19:08:24,000] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 19:08:24,000] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-02 19:08:24,000] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 19:08:23,999] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-02 19:08:23,999] torch.distributed.run: [WARNING] 
+[2025-01-02 19:08:23,999] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 19:08:23,999] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-02 19:08:23,999] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 19:08:24,001] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-02 19:08:24,001] torch.distributed.run: [WARNING] 
+[2025-01-02 19:08:24,001] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 19:08:24,001] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-02 19:08:24,001] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 19:08:24,005] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-02 19:08:24,005] torch.distributed.run: [WARNING] 
+[2025-01-02 19:08:24,005] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 19:08:24,005] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-02 19:08:24,005] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 19:08:24,012] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-02 19:08:24,012] torch.distributed.run: [WARNING] 
+[2025-01-02 19:08:24,012] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 19:08:24,012] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-02 19:08:24,012] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 19:08:24,015] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-02 19:08:24,016] torch.distributed.run: [WARNING] 
+[2025-01-02 19:08:24,016] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 19:08:24,016] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-02 19:08:24,016] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 19:08:24,014] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-02 19:08:24,021] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-02 19:08:24,021] torch.distributed.run: [WARNING] 
+[2025-01-02 19:08:24,021] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 19:08:24,021] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-02 19:08:24,021] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 19:08:24,022] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-02 19:08:24,023] torch.distributed.run: [WARNING] 
+[2025-01-02 19:08:24,023] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 19:08:24,023] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-02 19:08:24,023] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 19:08:23,977] torch.distributed.run: [WARNING] 
+[2025-01-02 19:08:23,977] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 19:08:23,977] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-02 19:08:23,977] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 19:08:23,977] torch.distributed.run: [WARNING] 
+[2025-01-02 19:08:23,977] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 19:08:23,977] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-02 19:08:23,977] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 19:08:24,025] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-02 19:08:24,025] torch.distributed.run: [WARNING] 
+[2025-01-02 19:08:24,025] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 19:08:24,025] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-02 19:08:24,025] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 19:08:24,027] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-02 19:08:24,027] torch.distributed.run: [WARNING] 
+[2025-01-02 19:08:24,027] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 19:08:24,027] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-02 19:08:24,027] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 19:08:24,045] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-02 19:08:24,045] torch.distributed.run: [WARNING] 
+[2025-01-02 19:08:24,045] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 19:08:24,045] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-02 19:08:24,045] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 19:08:24,059] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-02 19:08:24,059] torch.distributed.run: [WARNING] 
+[2025-01-02 19:08:24,059] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 19:08:24,059] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-02 19:08:24,059] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 19:08:24,014] torch.distributed.run: [WARNING] 
+[2025-01-02 19:08:24,014] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 19:08:24,014] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-02 19:08:24,014] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 19:08:24,064] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-02 19:08:24,064] torch.distributed.run: [WARNING] 
+[2025-01-02 19:08:24,064] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 19:08:24,064] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-02 19:08:24,064] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 19:08:24,075] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-02 19:08:24,075] torch.distributed.run: [WARNING] 
+[2025-01-02 19:08:24,075] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 19:08:24,075] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-02 19:08:24,075] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 19:08:24,078] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-02 19:08:24,078] torch.distributed.run: [WARNING] 
+[2025-01-02 19:08:24,078] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 19:08:24,078] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-02 19:08:24,078] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 19:08:24,081] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-02 19:08:24,081] torch.distributed.run: [WARNING] 
+[2025-01-02 19:08:24,081] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 19:08:24,081] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-02 19:08:24,081] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 19:08:24,098] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-02 19:08:24,098] torch.distributed.run: [WARNING] 
+[2025-01-02 19:08:24,098] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 19:08:24,098] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-02 19:08:24,098] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 19:08:24,168] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-02 19:08:24,168] torch.distributed.run: [WARNING] 
+[2025-01-02 19:08:24,168] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 19:08:24,168] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-02 19:08:24,168] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 19:08:27,588] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-02 19:08:27,588] torch.distributed.run: [WARNING] 
+[2025-01-02 19:08:27,588] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 19:08:27,588] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-02 19:08:27,588] torch.distributed.run: [WARNING] *****************************************
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+01/02/2025 19:10:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]: Config:
+01/02/2025 19:10:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]: Config(general=GeneralArgs(project='debug',
+01/02/2025 19:10:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                            run='469G_dp16_tp32_pp1_acc1_mbs16_seq4096_zero0_tpmodeRED_vocab131k',
+01/02/2025 19:10:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                            seed=42,
+01/02/2025 19:10:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                            step=None,
+01/02/2025 19:10:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                            consumed_train_samples=None,
+01/02/2025 19:10:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                            benchmark_csv_path=PosixPath('benchmark/results/bench_final2.csv'),
+01/02/2025 19:10:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                            ignore_sanity_checks=True),
+01/02/2025 19:10:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:        parallelism=ParallelismArgs(dp=16,
+01/02/2025 19:10:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                    pp=1,
+01/02/2025 19:10:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                    tp=32,
+01/02/2025 19:10:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                    pp_engine=<nanotron.parallel.pipeline_parallel.engine.OneForwardOneBackwardPipelineEngine object at 0x7f5a95524c40>,
+01/02/2025 19:10:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                    tp_mode=<TensorParallelLinearMode.REDUCE_SCATTER: 2>,
+01/02/2025 19:10:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                    tp_linear_async_communication=True,
+01/02/2025 19:10:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                    recompute_layer=False,
+01/02/2025 19:10:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                    tp_recompute_allgather=True,
+01/02/2025 19:10:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                    expert_parallel_size=1),
+01/02/2025 19:10:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:        model=ModelArgs(model_config=LlamaConfig(bos_token_id=0,
+01/02/2025 19:10:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                                 eos_token_id=0,
+01/02/2025 19:10:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                                 hidden_act='silu',
+01/02/2025 19:10:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                                 hidden_size=16384,
+01/02/2025 19:10:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                                 initializer_range=0.02,
+01/02/2025 19:10:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                                 intermediate_size=53248,
+01/02/2025 19:10:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                                 is_llama_config=True,
+01/02/2025 19:10:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                                 max_position_embeddings=4096,
+01/02/2025 19:10:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                                 num_attention_heads=128,
+01/02/2025 19:10:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                                 num_hidden_layers=126,
+01/02/2025 19:10:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                                 num_key_value_heads=128,
+01/02/2025 19:10:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                                 pad_token_id=None,
+01/02/2025 19:10:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                                 pretraining_tp=1,
+01/02/2025 19:10:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                                 rms_norm_eps=1e-05,
+01/02/2025 19:10:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                                 rope_scaling=None,
+01/02/2025 19:10:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                                 rope_theta=10000.0,
+01/02/2025 19:10:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                                 rope_interleaved=False,
+01/02/2025 19:10:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                                 tie_word_embeddings=False,
+01/02/2025 19:10:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                                 use_cache=True,
+01/02/2025 19:10:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                                 vocab_size=131072),
+01/02/2025 19:10:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                        init_method=RandomInit(std=0.02),
+01/02/2025 19:10:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                        dtype=torch.bfloat16,
+01/02/2025 19:10:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                        make_vocab_size_divisible_by=1,
+01/02/2025 19:10:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                        ddp_bucket_cap_mb=25),
+01/02/2025 19:10:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:        tokenizer=TokenizerArgs(tokenizer_name_or_path='robot-test/dummy-tokenizer-wordlevel',
+01/02/2025 19:10:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                tokenizer_revision=None,
+01/02/2025 19:10:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                tokenizer_max_length=None),
+01/02/2025 19:10:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:        checkpoints=CheckpointsArgs(checkpoints_path=PosixPath('checkpoints'),
+01/02/2025 19:10:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                    checkpoint_interval=10000,
+01/02/2025 19:10:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                    save_initial_state=False,
+01/02/2025 19:10:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                    save_final_state=False,
+01/02/2025 19:10:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                    resume_checkpoint_path=None,
+01/02/2025 19:10:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                    checkpoints_path_is_shared_file_system=False),
+01/02/2025 19:10:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:        logging=LoggingArgs(log_level='info',
+01/02/2025 19:10:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                            log_level_replica='info',
+01/02/2025 19:10:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                            iteration_step_info_interval=1),
+01/02/2025 19:10:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:        tokens=TokensArgs(sequence_length=4096,
+01/02/2025 19:10:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                          train_steps=100,
+01/02/2025 19:10:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                          micro_batch_size=16,
+01/02/2025 19:10:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                          batch_accumulation_per_replica=1,
+01/02/2025 19:10:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                          val_check_interval=100,
+01/02/2025 19:10:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                          limit_val_batches=0,
+01/02/2025 19:10:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                          limit_test_batches=0),
+01/02/2025 19:10:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:        optimizer=OptimizerArgs(optimizer_factory=AdamWOptimizerArgs(adam_eps=1e-08,
+01/02/2025 19:10:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                                                     adam_beta1=0.9,
+01/02/2025 19:10:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                                                     adam_beta2=0.95,
+01/02/2025 19:10:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                                                     torch_adam_is_fused=True,
+01/02/2025 19:10:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                                                     name='adamW'),
+01/02/2025 19:10:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                zero_stage=0,
+01/02/2025 19:10:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                weight_decay=0.01,
+01/02/2025 19:10:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                clip_grad=1.0,
+01/02/2025 19:10:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                accumulate_grad_in_fp32=True,
+01/02/2025 19:10:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                learning_rate_scheduler=LRSchedulerArgs(learning_rate=0.0003,
+01/02/2025 19:10:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                                                        lr_warmup_steps=2,
+01/02/2025 19:10:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                                                        lr_warmup_style='linear',
+01/02/2025 19:10:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                                                        lr_decay_style='cosine',
+01/02/2025 19:10:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                                                        lr_decay_steps=13,
+01/02/2025 19:10:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                                                        lr_decay_starting_step=None,
+01/02/2025 19:10:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                                                        min_decay_lr=1e-05)),
+01/02/2025 19:10:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:        data_stages=[DatasetStageArgs(name='Stable Training Stage',
+01/02/2025 19:10:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                      start_training_step=1,
+01/02/2025 19:10:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                      data=DataArgs(dataset=None,
+01/02/2025 19:10:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                                    seed=42,
+01/02/2025 19:10:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                                    num_loading_workers=1))],
+01/02/2025 19:10:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:        profiler=None,
+01/02/2025 19:10:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:        lighteval=None,
+01/02/2025 19:10:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:        s3_upload=None)
+01/02/2025 19:10:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]: Model Config:
+01/02/2025 19:10:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]: LlamaConfig(bos_token_id=0,
+01/02/2025 19:10:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:             eos_token_id=0,
+01/02/2025 19:10:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:             hidden_act='silu',
+01/02/2025 19:10:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:             hidden_size=16384,
+01/02/2025 19:10:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:             initializer_range=0.02,
+01/02/2025 19:10:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:             intermediate_size=53248,
+01/02/2025 19:10:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:             is_llama_config=True,
+01/02/2025 19:10:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:             max_position_embeddings=4096,
+01/02/2025 19:10:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:             num_attention_heads=128,
+01/02/2025 19:10:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:             num_hidden_layers=126,
+01/02/2025 19:10:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:             num_key_value_heads=128,
+01/02/2025 19:10:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:             pad_token_id=None,
+01/02/2025 19:10:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:             pretraining_tp=1,
+01/02/2025 19:10:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:             rms_norm_eps=1e-05,
+01/02/2025 19:10:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:             rope_scaling=None,
+01/02/2025 19:10:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:             rope_theta=10000.0,
+01/02/2025 19:10:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:             rope_interleaved=False,
+01/02/2025 19:10:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:             tie_word_embeddings=False,
+01/02/2025 19:10:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:             use_cache=True,
+01/02/2025 19:10:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:             vocab_size=131072)
+01/02/2025 19:10:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]: Building model..
+01/02/2025 19:10:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]: Initialize RoPE Theta = 10000.0
+01/02/2025 19:10:03 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]: Setting PP block ranks...
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+01/02/2025 19:10:15 [INFO|DP=0|PP=0|TP=6|ip-26-0-160-100]: Local number of parameters: 14.7G (27983.91MiB)
+01/02/2025 19:10:15 [INFO|DP=0|PP=0|TP=7|ip-26-0-160-100]: Local number of parameters: 14.7G (27983.91MiB)
+01/02/2025 19:10:15 [INFO|DP=0|PP=0|TP=4|ip-26-0-160-100]: Local number of parameters: 14.7G (27983.91MiB)
+01/02/2025 19:10:15 [INFO|DP=0|PP=0|TP=5|ip-26-0-160-100]: Local number of parameters: 14.7G (27983.91MiB)
+01/02/2025 19:10:15 [INFO|DP=0|PP=0|TP=2|ip-26-0-160-100]: Local number of parameters: 14.7G (27983.91MiB)
+01/02/2025 19:10:15 [INFO|DP=0|PP=0|TP=1|ip-26-0-160-100]: Local number of parameters: 14.7G (27983.91MiB)
+01/02/2025 19:10:15 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]: Total number of parameters: 469G (895485.00MiB)
+01/02/2025 19:10:15 [INFO|DP=0|PP=0|TP=3|ip-26-0-160-100]: Local number of parameters: 14.7G (27983.91MiB)
+01/02/2025 19:10:15 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]: Local number of parameters: 14.7G (27983.91MiB)
+01/02/2025 19:10:15 [INFO|DP=0|PP=0|TP=6|ip-26-0-160-100]: [After model building] Memory usage: 27984.04MiB. Peak allocated: 27984.74MiB Peak reserved: 28086.00MiB
+01/02/2025 19:10:15 [INFO|DP=0|PP=0|TP=4|ip-26-0-160-100]: [After model building] Memory usage: 27984.04MiB. Peak allocated: 27984.74MiB Peak reserved: 28086.00MiB
+01/02/2025 19:10:15 [INFO|DP=0|PP=0|TP=5|ip-26-0-160-100]: [After model building] Memory usage: 27984.04MiB. Peak allocated: 27984.74MiB Peak reserved: 28086.00MiB
+01/02/2025 19:10:15 [INFO|DP=0|PP=0|TP=7|ip-26-0-160-100]: [After model building] Memory usage: 27984.04MiB. Peak allocated: 27984.74MiB Peak reserved: 28086.00MiB
+01/02/2025 19:10:15 [INFO|DP=0|PP=0|TP=2|ip-26-0-160-100]: [After model building] Memory usage: 27984.04MiB. Peak allocated: 27984.74MiB Peak reserved: 28086.00MiB
+01/02/2025 19:10:15 [INFO|DP=0|PP=0|TP=1|ip-26-0-160-100]: [After model building] Memory usage: 27984.04MiB. Peak allocated: 27984.74MiB Peak reserved: 28086.00MiB
+01/02/2025 19:10:15 [INFO|DP=0|PP=0|TP=3|ip-26-0-160-100]: [After model building] Memory usage: 27984.04MiB. Peak allocated: 27984.74MiB Peak reserved: 28086.00MiB
+01/02/2025 19:10:15 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]: [After model building] Memory usage: 27984.04MiB. Peak allocated: 27984.74MiB Peak reserved: 28086.00MiB
+01/02/2025 19:10:15 [INFO|DP=0|PP=0|TP=21|ip-26-0-160-192]: Local number of parameters: 14.7G (27983.91MiB)
+01/02/2025 19:10:15 [INFO|DP=0|PP=0|TP=17|ip-26-0-160-192]: Local number of parameters: 14.7G (27983.91MiB)
+01/02/2025 19:10:15 [INFO|DP=0|PP=0|TP=27|ip-26-0-163-236]: Local number of parameters: 14.7G (27983.91MiB)
+01/02/2025 19:10:15 [INFO|DP=0|PP=0|TP=26|ip-26-0-163-236]: Local number of parameters: 14.7G (27983.91MiB)
+01/02/2025 19:10:15 [INFO|DP=0|PP=0|TP=28|ip-26-0-163-236]: Local number of parameters: 14.7G (27983.91MiB)
+01/02/2025 19:10:15 [INFO|DP=0|PP=0|TP=23|ip-26-0-160-192]: Local number of parameters: 14.7G (27983.91MiB)
+01/02/2025 19:10:15 [INFO|DP=0|PP=0|TP=22|ip-26-0-160-192]: Local number of parameters: 14.7G (27983.91MiB)
+01/02/2025 19:10:15 [INFO|DP=0|PP=0|TP=19|ip-26-0-160-192]: Local number of parameters: 14.7G (27983.91MiB)
+01/02/2025 19:10:15 [INFO|DP=0|PP=0|TP=20|ip-26-0-160-192]: Local number of parameters: 14.7G (27983.91MiB)
+01/02/2025 19:10:15 [INFO|DP=0|PP=0|TP=30|ip-26-0-163-236]: Local number of parameters: 14.7G (27983.91MiB)
+01/02/2025 19:10:15 [INFO|DP=0|PP=0|TP=29|ip-26-0-163-236]: Local number of parameters: 14.7G (27983.91MiB)
+01/02/2025 19:10:15 [INFO|DP=0|PP=0|TP=31|ip-26-0-163-236]: Local number of parameters: 14.7G (27983.91MiB)
+01/02/2025 19:10:15 [INFO|DP=0|PP=0|TP=16|ip-26-0-160-192]: Local number of parameters: 14.7G (27983.91MiB)
+01/02/2025 19:10:15 [INFO|DP=0|PP=0|TP=18|ip-26-0-160-192]: Local number of parameters: 14.7G (27983.91MiB)
+01/02/2025 19:10:15 [INFO|DP=0|PP=0|TP=24|ip-26-0-163-236]: Local number of parameters: 14.7G (27983.91MiB)
+01/02/2025 19:10:15 [INFO|DP=0|PP=0|TP=17|ip-26-0-160-192]: [After model building] Memory usage: 27984.04MiB. Peak allocated: 27984.74MiB Peak reserved: 28098.00MiB
+01/02/2025 19:10:15 [INFO|DP=0|PP=0|TP=25|ip-26-0-163-236]: Local number of parameters: 14.7G (27983.91MiB)
+01/02/2025 19:10:15 [INFO|DP=0|PP=0|TP=27|ip-26-0-163-236]: [After model building] Memory usage: 27984.04MiB. Peak allocated: 27984.74MiB Peak reserved: 28086.00MiB
+01/02/2025 19:10:15 [INFO|DP=0|PP=0|TP=21|ip-26-0-160-192]: [After model building] Memory usage: 27984.04MiB. Peak allocated: 27984.74MiB Peak reserved: 28098.00MiB
+01/02/2025 19:10:15 [INFO|DP=0|PP=0|TP=13|ip-26-0-160-103]: Local number of parameters: 14.7G (27983.91MiB)
+01/02/2025 19:10:15 [INFO|DP=0|PP=0|TP=23|ip-26-0-160-192]: [After model building] Memory usage: 27984.04MiB. Peak allocated: 27984.74MiB Peak reserved: 28098.00MiB
+01/02/2025 19:10:15 [INFO|DP=0|PP=0|TP=15|ip-26-0-160-103]: Local number of parameters: 14.7G (27983.91MiB)
+01/02/2025 19:10:15 [INFO|DP=0|PP=0|TP=26|ip-26-0-163-236]: [After model building] Memory usage: 27984.04MiB. Peak allocated: 27984.74MiB Peak reserved: 28086.00MiB
+01/02/2025 19:10:15 [INFO|DP=0|PP=0|TP=28|ip-26-0-163-236]: [After model building] Memory usage: 27984.04MiB. Peak allocated: 27984.74MiB Peak reserved: 28086.00MiB
+01/02/2025 19:10:15 [INFO|DP=0|PP=0|TP=31|ip-26-0-163-236]: [After model building] Memory usage: 27984.04MiB. Peak allocated: 27984.74MiB Peak reserved: 28086.00MiB
+01/02/2025 19:10:15 [INFO|DP=0|PP=0|TP=30|ip-26-0-163-236]: [After model building] Memory usage: 27984.04MiB. Peak allocated: 27984.74MiB Peak reserved: 28086.00MiB
+01/02/2025 19:10:15 [INFO|DP=0|PP=0|TP=22|ip-26-0-160-192]: [After model building] Memory usage: 27984.04MiB. Peak allocated: 27984.74MiB Peak reserved: 28098.00MiB
+01/02/2025 19:10:15 [INFO|DP=0|PP=0|TP=19|ip-26-0-160-192]: [After model building] Memory usage: 27984.04MiB. Peak allocated: 27984.74MiB Peak reserved: 28098.00MiB
+01/02/2025 19:10:15 [INFO|DP=0|PP=0|TP=20|ip-26-0-160-192]: [After model building] Memory usage: 27984.04MiB. Peak allocated: 27984.74MiB Peak reserved: 28098.00MiB
+01/02/2025 19:10:15 [INFO|DP=0|PP=0|TP=16|ip-26-0-160-192]: [After model building] Memory usage: 27984.04MiB. Peak allocated: 27984.74MiB Peak reserved: 28098.00MiB
+01/02/2025 19:10:15 [INFO|DP=0|PP=0|TP=14|ip-26-0-160-103]: Local number of parameters: 14.7G (27983.91MiB)
+01/02/2025 19:10:15 [INFO|DP=0|PP=0|TP=9|ip-26-0-160-103]: Local number of parameters: 14.7G (27983.91MiB)
+01/02/2025 19:10:15 [INFO|DP=0|PP=0|TP=10|ip-26-0-160-103]: Local number of parameters: 14.7G (27983.91MiB)
+01/02/2025 19:10:15 [INFO|DP=0|PP=0|TP=8|ip-26-0-160-103]: Local number of parameters: 14.7G (27983.91MiB)
+01/02/2025 19:10:15 [INFO|DP=0|PP=0|TP=29|ip-26-0-163-236]: [After model building] Memory usage: 27984.04MiB. Peak allocated: 27984.74MiB Peak reserved: 28086.00MiB
+01/02/2025 19:10:15 [INFO|DP=0|PP=0|TP=18|ip-26-0-160-192]: [After model building] Memory usage: 27984.04MiB. Peak allocated: 27984.74MiB Peak reserved: 28098.00MiB
+01/02/2025 19:10:15 [INFO|DP=0|PP=0|TP=11|ip-26-0-160-103]: Local number of parameters: 14.7G (27983.91MiB)
+01/02/2025 19:10:15 [INFO|DP=0|PP=0|TP=24|ip-26-0-163-236]: [After model building] Memory usage: 27984.04MiB. Peak allocated: 27984.74MiB Peak reserved: 28086.00MiB
+01/02/2025 19:10:15 [INFO|DP=0|PP=0|TP=12|ip-26-0-160-103]: Local number of parameters: 14.7G (27983.91MiB)
+01/02/2025 19:10:15 [INFO|DP=0|PP=0|TP=13|ip-26-0-160-103]: [After model building] Memory usage: 27984.04MiB. Peak allocated: 27984.74MiB Peak reserved: 28086.00MiB
+01/02/2025 19:10:15 [INFO|DP=0|PP=0|TP=15|ip-26-0-160-103]: [After model building] Memory usage: 27984.04MiB. Peak allocated: 27984.74MiB Peak reserved: 28086.00MiB
+01/02/2025 19:10:15 [INFO|DP=0|PP=0|TP=25|ip-26-0-163-236]: [After model building] Memory usage: 27984.04MiB. Peak allocated: 27984.74MiB Peak reserved: 28086.00MiB
+01/02/2025 19:10:15 [INFO|DP=0|PP=0|TP=14|ip-26-0-160-103]: [After model building] Memory usage: 27984.04MiB. Peak allocated: 27984.74MiB Peak reserved: 28086.00MiB
+01/02/2025 19:10:15 [INFO|DP=0|PP=0|TP=9|ip-26-0-160-103]: [After model building] Memory usage: 27984.04MiB. Peak allocated: 27984.74MiB Peak reserved: 28086.00MiB
+01/02/2025 19:10:15 [INFO|DP=0|PP=0|TP=10|ip-26-0-160-103]: [After model building] Memory usage: 27984.04MiB. Peak allocated: 27984.74MiB Peak reserved: 28086.00MiB
+01/02/2025 19:10:15 [INFO|DP=0|PP=0|TP=8|ip-26-0-160-103]: [After model building] Memory usage: 27984.04MiB. Peak allocated: 27984.74MiB Peak reserved: 28086.00MiB
+01/02/2025 19:10:15 [INFO|DP=0|PP=0|TP=11|ip-26-0-160-103]: [After model building] Memory usage: 27984.04MiB. Peak allocated: 27984.74MiB Peak reserved: 28086.00MiB
+01/02/2025 19:10:15 [INFO|DP=0|PP=0|TP=12|ip-26-0-160-103]: [After model building] Memory usage: 27984.04MiB. Peak allocated: 27984.74MiB Peak reserved: 28086.00MiB
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+01/02/2025 19:10:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]: No checkpoint path provided.
+01/02/2025 19:10:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]: Parametrizing model parameters using StandardParametrizator
+01/02/2025 19:10:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]: [Optimizer Building] Using LearningRateForSP as learning rate
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 1 has a total capacty of 79.33 GiB of which 12.76 GiB is free. Including non-PyTorch memory, this process has 66.56 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    trainer = DistributedTrainer(config_file)
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    optimizer = optimizer_builder(named_param_groups)
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 5 has a total capacty of 79.33 GiB of which 13.13 GiB is free. Including non-PyTorch memory, this process has 66.19 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    result = OptimizerFromGradientAccumulator(
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 5 has a total capacty of 79.33 GiB of which 12.76 GiB is free. Including non-PyTorch memory, this process has 66.56 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    optimizer = optimizer_builder(named_param_groups)
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    result = OptimizerFromGradientAccumulator(
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 3 has a total capacty of 79.33 GiB of which 12.59 GiB is free. Including non-PyTorch memory, this process has 66.73 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 2 has a total capacty of 79.33 GiB of which 12.88 GiB is free. Including non-PyTorch memory, this process has 66.44 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    trainer = DistributedTrainer(config_file)
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 6 has a total capacty of 79.33 GiB of which 12.76 GiB is free. Including non-PyTorch memory, this process has 66.56 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 4 has a total capacty of 79.33 GiB of which 13.03 GiB is free. Including non-PyTorch memory, this process has 66.29 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+Traceback (most recent call last):
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    result = OptimizerFromGradientAccumulator(
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    trainer = DistributedTrainer(config_file)
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 7 has a total capacty of 79.33 GiB of which 12.94 GiB is free. Including non-PyTorch memory, this process has 66.38 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    optimizer = optimizer_builder(named_param_groups)
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    result = OptimizerFromGradientAccumulator(
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    result = OptimizerFromGradientAccumulator(
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 2 has a total capacty of 79.33 GiB of which 12.76 GiB is free. Including non-PyTorch memory, this process has 66.56 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    optimizer = optimizer_builder(named_param_groups)
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 7 has a total capacty of 79.33 GiB of which 13.05 GiB is free. Including non-PyTorch memory, this process has 66.26 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 6 has a total capacty of 79.33 GiB of which 12.88 GiB is free. Including non-PyTorch memory, this process has 66.44 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 7 has a total capacty of 79.33 GiB of which 12.94 GiB is free. Including non-PyTorch memory, this process has 66.38 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    trainer = DistributedTrainer(config_file)
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+Traceback (most recent call last):
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+Traceback (most recent call last):
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 7 has a total capacty of 79.33 GiB of which 13.23 GiB is free. Including non-PyTorch memory, this process has 66.09 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    trainer = DistributedTrainer(config_file)
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    trainer = DistributedTrainer(config_file)
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    result = OptimizerFromGradientAccumulator(
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+Traceback (most recent call last):
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    result = OptimizerFromGradientAccumulator(
+    trainer = DistributedTrainer(config_file)
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    result = OptimizerFromGradientAccumulator(
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    result = OptimizerFromGradientAccumulator(
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    result = OptimizerFromGradientAccumulator(
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    result = OptimizerFromGradientAccumulator(
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 4 has a total capacty of 79.33 GiB of which 13.11 GiB is free. Including non-PyTorch memory, this process has 66.21 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    result = OptimizerFromGradientAccumulator(
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 1 has a total capacty of 79.33 GiB of which 12.92 GiB is free. Including non-PyTorch memory, this process has 66.40 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+Traceback (most recent call last):
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 1 has a total capacty of 79.33 GiB of which 12.88 GiB is free. Including non-PyTorch memory, this process has 66.44 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 0 has a total capacty of 79.33 GiB of which 12.90 GiB is free. Including non-PyTorch memory, this process has 66.42 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 3 has a total capacty of 79.33 GiB of which 12.70 GiB is free. Including non-PyTorch memory, this process has 66.62 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 4 has a total capacty of 79.33 GiB of which 12.86 GiB is free. Including non-PyTorch memory, this process has 66.46 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 5 has a total capacty of 79.33 GiB of which 13.05 GiB is free. Including non-PyTorch memory, this process has 66.26 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 4 has a total capacty of 79.33 GiB of which 13.02 GiB is free. Including non-PyTorch memory, this process has 66.30 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 113.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 1 has a total capacty of 79.33 GiB of which 13.05 GiB is free. Including non-PyTorch memory, this process has 66.26 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+Traceback (most recent call last):
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    trainer = DistributedTrainer(config_file)
+Traceback (most recent call last):
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    optimizer = optimizer_builder(named_param_groups)
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 5 has a total capacty of 79.33 GiB of which 12.88 GiB is free. Including non-PyTorch memory, this process has 66.44 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+Traceback (most recent call last):
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    trainer = DistributedTrainer(config_file)
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    result = OptimizerFromGradientAccumulator(
+    optimizer = optimizer_builder(named_param_groups)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 4 has a total capacty of 79.33 GiB of which 12.86 GiB is free. Including non-PyTorch memory, this process has 66.46 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    optimizer = optimizer_builder(named_param_groups)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 3 has a total capacty of 79.33 GiB of which 12.59 GiB is free. Including non-PyTorch memory, this process has 66.73 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    result = OptimizerFromGradientAccumulator(
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 1 has a total capacty of 79.33 GiB of which 13.05 GiB is free. Including non-PyTorch memory, this process has 66.26 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    trainer = DistributedTrainer(config_file)
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    optimizer = optimizer_builder(named_param_groups)
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+Traceback (most recent call last):
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    result = OptimizerFromGradientAccumulator(
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    result = OptimizerFromGradientAccumulator(
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())    
+    optimizer = optimizer_builder(named_param_groups)
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 5 has a total capacty of 79.33 GiB of which 12.76 GiB is free. Including non-PyTorch memory, this process has 66.56 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+trainer = DistributedTrainer(config_file)  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 5 has a total capacty of 79.33 GiB of which 13.05 GiB is free. Including non-PyTorch memory, this process has 66.26 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 6 has a total capacty of 79.33 GiB of which 12.88 GiB is free. Including non-PyTorch memory, this process has 66.44 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    trainer = DistributedTrainer(config_file)
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 5 has a total capacty of 79.33 GiB of which 12.88 GiB is free. Including non-PyTorch memory, this process has 66.44 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    result = OptimizerFromGradientAccumulator(
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    optimizer = optimizer_builder(named_param_groups)
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 0 has a total capacty of 79.33 GiB of which 12.79 GiB is free. Including non-PyTorch memory, this process has 66.53 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    trainer = DistributedTrainer(config_file)
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    optimizer = optimizer_builder(named_param_groups)
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+    result = OptimizerFromGradientAccumulator(
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 5 has a total capacty of 79.33 GiB of which 12.92 GiB is free. Including non-PyTorch memory, this process has 66.40 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    result = OptimizerFromGradientAccumulator(
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    result = OptimizerFromGradientAccumulator(
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 7 has a total capacty of 79.33 GiB of which 13.05 GiB is free. Including non-PyTorch memory, this process has 66.26 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    optimizer = optimizer_builder(named_param_groups)
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+Traceback (most recent call last):
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 6 has a total capacty of 79.33 GiB of which 12.91 GiB is free. Including non-PyTorch memory, this process has 66.41 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 113.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    result = OptimizerFromGradientAccumulator(
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 1 has a total capacty of 79.33 GiB of which 12.88 GiB is free. Including non-PyTorch memory, this process has 66.44 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    trainer = DistributedTrainer(config_file)
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 2 has a total capacty of 79.33 GiB of which 13.13 GiB is free. Including non-PyTorch memory, this process has 66.19 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 7 has a total capacty of 79.33 GiB of which 13.09 GiB is free. Including non-PyTorch memory, this process has 66.22 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 3 has a total capacty of 79.33 GiB of which 12.96 GiB is free. Including non-PyTorch memory, this process has 66.36 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 3 has a total capacty of 79.33 GiB of which 12.88 GiB is free. Including non-PyTorch memory, this process has 66.44 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 0 has a total capacty of 79.33 GiB of which 13.16 GiB is free. Including non-PyTorch memory, this process has 66.16 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    trainer = DistributedTrainer(config_file)
+    optimizer = optimizer_builder(named_param_groups)
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    optimizer = optimizer_builder(named_param_groups)
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    result = OptimizerFromGradientAccumulator(
+    trainer = DistributedTrainer(config_file)
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 2 has a total capacty of 79.33 GiB of which 12.76 GiB is free. Including non-PyTorch memory, this process has 66.56 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    trainer = DistributedTrainer(config_file)
+    trainer = DistributedTrainer(config_file)
+Traceback (most recent call last):
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    optimizer = optimizer_builder(named_param_groups)
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    trainer = DistributedTrainer(config_file)
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    result = OptimizerFromGradientAccumulator(
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+Traceback (most recent call last):
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+Traceback (most recent call last):
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    result = OptimizerFromGradientAccumulator(
+    result = OptimizerFromGradientAccumulator(
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    trainer = DistributedTrainer(config_file)
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 6 has a total capacty of 79.33 GiB of which 12.76 GiB is free. Including non-PyTorch memory, this process has 66.56 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 3 has a total capacty of 79.33 GiB of which 13.00 GiB is free. Including non-PyTorch memory, this process has 66.32 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    trainer = DistributedTrainer(config_file)    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+
+    result = OptimizerFromGradientAccumulator(
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+Traceback (most recent call last):
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 0 has a total capacty of 79.33 GiB of which 12.90 GiB is free. Including non-PyTorch memory, this process has 66.42 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 1 has a total capacty of 79.33 GiB of which 12.76 GiB is free. Including non-PyTorch memory, this process has 66.56 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 6 has a total capacty of 79.33 GiB of which 13.13 GiB is free. Including non-PyTorch memory, this process has 66.19 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    optimizer = optimizer_builder(named_param_groups)
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+    optimizer = optimizer_builder(named_param_groups)
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+    optimizer = optimizer_builder(named_param_groups)
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    optimizer = optimizer_builder(named_param_groups)
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    result = OptimizerFromGradientAccumulator(
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 2 has a total capacty of 79.33 GiB of which 12.80 GiB is free. Including non-PyTorch memory, this process has 66.52 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 5 has a total capacty of 79.33 GiB of which 12.79 GiB is free. Including non-PyTorch memory, this process has 66.53 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 113.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 6 has a total capacty of 79.33 GiB of which 13.05 GiB is free. Including non-PyTorch memory, this process has 66.26 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+trainer = DistributedTrainer(config_file)
+        gradient_accumulator = gradient_accumulator_builder(name_to_param.items())result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    result = OptimizerFromGradientAccumulator(
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    trainer = DistributedTrainer(config_file)
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 4 has a total capacty of 79.33 GiB of which 12.74 GiB is free. Including non-PyTorch memory, this process has 66.58 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(    
+optimizer = optimizer_builder(named_param_groups)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 4 has a total capacty of 79.33 GiB of which 12.86 GiB is free. Including non-PyTorch memory, this process has 66.46 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+        gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    trainer = DistributedTrainer(config_file)
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    optimizer = optimizer_builder(named_param_groups)
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+Traceback (most recent call last):
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    trainer = DistributedTrainer(config_file)
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 6 has a total capacty of 79.33 GiB of which 13.13 GiB is free. Including non-PyTorch memory, this process has 66.19 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    trainer = DistributedTrainer(config_file)
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    optimizer = optimizer_builder(named_param_groups)
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 0 has a total capacty of 79.33 GiB of which 13.12 GiB is free. Including non-PyTorch memory, this process has 66.20 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    trainer = DistributedTrainer(config_file)
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+Traceback (most recent call last):
+    trainer = DistributedTrainer(config_file)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 1 has a total capacty of 79.33 GiB of which 13.13 GiB is free. Including non-PyTorch memory, this process has 66.19 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 2 has a total capacty of 79.33 GiB of which 13.13 GiB is free. Including non-PyTorch memory, this process has 66.19 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+Traceback (most recent call last):
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 7 has a total capacty of 79.33 GiB of which 12.96 GiB is free. Including non-PyTorch memory, this process has 66.35 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 113.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    trainer = DistributedTrainer(config_file)
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    trainer = DistributedTrainer(config_file)
+    result = OptimizerFromGradientAccumulator(
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    trainer = DistributedTrainer(config_file)
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    result = OptimizerFromGradientAccumulator(
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+Traceback (most recent call last):
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 3 has a total capacty of 79.33 GiB of which 12.61 GiB is free. Including non-PyTorch memory, this process has 66.71 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 113.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    result = OptimizerFromGradientAccumulator(
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 0 has a total capacty of 79.33 GiB of which 13.08 GiB is free. Including non-PyTorch memory, this process has 66.24 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    trainer = DistributedTrainer(config_file)
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    optimizer = optimizer_builder(named_param_groups)
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    trainer = DistributedTrainer(config_file)    optimizer = optimizer_builder(named_param_groups)
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 5 has a total capacty of 79.33 GiB of which 12.76 GiB is free. Including non-PyTorch memory, this process has 66.56 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    trainer = DistributedTrainer(config_file)
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    result = OptimizerFromGradientAccumulator(
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    result = OptimizerFromGradientAccumulator(
+    result = OptimizerFromGradientAccumulator(
+    result = OptimizerFromGradientAccumulator(
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    result = OptimizerFromGradientAccumulator(
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    trainer = DistributedTrainer(config_file)
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    optimizer = optimizer_builder(named_param_groups)    
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 5 has a total capacty of 79.33 GiB of which 13.17 GiB is free. Including non-PyTorch memory, this process has 66.15 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+Traceback (most recent call last):
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 7 has a total capacty of 79.33 GiB of which 13.35 GiB is free. Including non-PyTorch memory, this process has 65.97 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    result = OptimizerFromGradientAccumulator(
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 4 has a total capacty of 79.33 GiB of which 13.11 GiB is free. Including non-PyTorch memory, this process has 66.21 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    result = OptimizerFromGradientAccumulator(
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 0 has a total capacty of 79.33 GiB of which 12.90 GiB is free. Including non-PyTorch memory, this process has 66.42 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 2 has a total capacty of 79.33 GiB of which 13.04 GiB is free. Including non-PyTorch memory, this process has 66.28 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 113.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 5 has a total capacty of 79.33 GiB of which 13.12 GiB is free. Including non-PyTorch memory, this process has 66.20 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 113.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    trainer = DistributedTrainer(config_file)
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    trainer = DistributedTrainer(config_file)
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 0 has a total capacty of 79.33 GiB of which 12.79 GiB is free. Including non-PyTorch memory, this process has 66.53 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 7 has a total capacty of 79.33 GiB of which 13.22 GiB is free. Including non-PyTorch memory, this process has 66.10 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 113.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    trainer = DistributedTrainer(config_file)
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 3 has a total capacty of 79.33 GiB of which 12.59 GiB is free. Including non-PyTorch memory, this process has 66.73 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 0 has a total capacty of 79.33 GiB of which 13.14 GiB is free. Including non-PyTorch memory, this process has 66.17 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 113.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    result = OptimizerFromGradientAccumulator(
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    trainer = DistributedTrainer(config_file)
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    trainer = DistributedTrainer(config_file)
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+Traceback (most recent call last):
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    optimizer = optimizer_builder(named_param_groups)
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 3 has a total capacty of 79.33 GiB of which 12.70 GiB is free. Including non-PyTorch memory, this process has 66.62 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+        self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(result = OptimizerFromGradientAccumulator(
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 2 has a total capacty of 79.33 GiB of which 12.88 GiB is free. Including non-PyTorch memory, this process has 66.44 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 3 has a total capacty of 79.33 GiB of which 12.88 GiB is free. Including non-PyTorch memory, this process has 66.44 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    result = OptimizerFromGradientAccumulator(
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")    
+optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 4 has a total capacty of 79.33 GiB of which 13.11 GiB is free. Including non-PyTorch memory, this process has 66.21 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    trainer = DistributedTrainer(config_file)
+    trainer = DistributedTrainer(config_file)
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    result = OptimizerFromGradientAccumulator(
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    trainer = DistributedTrainer(config_file)
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 1 has a total capacty of 79.33 GiB of which 13.13 GiB is free. Including non-PyTorch memory, this process has 66.19 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    optimizer = optimizer_builder(named_param_groups)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 1 has a total capacty of 79.33 GiB of which 12.79 GiB is free. Including non-PyTorch memory, this process has 66.53 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 113.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+Traceback (most recent call last):
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 4 has a total capacty of 79.33 GiB of which 12.74 GiB is free. Including non-PyTorch memory, this process has 66.58 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    optimizer = optimizer_builder(named_param_groups)
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    result = OptimizerFromGradientAccumulator(
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 2 has a total capacty of 79.33 GiB of which 12.91 GiB is free. Including non-PyTorch memory, this process has 66.41 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 113.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 7 has a total capacty of 79.33 GiB of which 12.94 GiB is free. Including non-PyTorch memory, this process has 66.38 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    trainer = DistributedTrainer(config_file)
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    result = OptimizerFromGradientAccumulator(
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 1 has a total capacty of 79.33 GiB of which 13.80 GiB is free. Including non-PyTorch memory, this process has 65.52 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 0 has a total capacty of 79.33 GiB of which 12.93 GiB is free. Including non-PyTorch memory, this process has 66.39 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 113.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 6 has a total capacty of 79.33 GiB of which 13.13 GiB is free. Including non-PyTorch memory, this process has 66.19 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    optimizer = optimizer_builder(named_param_groups)
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 7 has a total capacty of 79.33 GiB of which 13.30 GiB is free. Including non-PyTorch memory, this process has 66.02 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 113.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    result = OptimizerFromGradientAccumulator(
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 7 has a total capacty of 79.33 GiB of which 13.31 GiB is free. Including non-PyTorch memory, this process has 66.01 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    trainer = DistributedTrainer(config_file)
+    optimizer = optimizer_builder(named_param_groups)
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    result = OptimizerFromGradientAccumulator(
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 1 has a total capacty of 79.33 GiB of which 13.17 GiB is free. Including non-PyTorch memory, this process has 66.15 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 5 has a total capacty of 79.33 GiB of which 13.13 GiB is free. Including non-PyTorch memory, this process has 66.19 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 4 has a total capacty of 79.33 GiB of which 13.07 GiB is free. Including non-PyTorch memory, this process has 66.25 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    trainer = DistributedTrainer(config_file)
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    trainer = DistributedTrainer(config_file)
+Traceback (most recent call last):
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 1 has a total capacty of 79.33 GiB of which 13.12 GiB is free. Including non-PyTorch memory, this process has 66.20 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 113.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 1 has a total capacty of 79.33 GiB of which 12.76 GiB is free. Including non-PyTorch memory, this process has 66.56 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    optimizer = optimizer_builder(named_param_groups)
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    trainer = DistributedTrainer(config_file)
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+Traceback (most recent call last):
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 2 has a total capacty of 79.33 GiB of which 13.05 GiB is free. Including non-PyTorch memory, this process has 66.26 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    result = OptimizerFromGradientAccumulator(
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 7 has a total capacty of 79.33 GiB of which 13.02 GiB is free. Including non-PyTorch memory, this process has 66.30 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    trainer = DistributedTrainer(config_file)
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 4 has a total capacty of 79.33 GiB of which 12.88 GiB is free. Including non-PyTorch memory, this process has 66.44 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 113.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    optimizer = optimizer_builder(named_param_groups)
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    trainer = DistributedTrainer(config_file)
+    result = OptimizerFromGradientAccumulator(
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    trainer = DistributedTrainer(config_file)
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 0 has a total capacty of 79.33 GiB of which 13.08 GiB is free. Including non-PyTorch memory, this process has 66.24 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+    optimizer = optimizer_builder(named_param_groups)
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    trainer = DistributedTrainer(config_file)
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    optimizer = optimizer_builder(named_param_groups)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 7 has a total capacty of 79.33 GiB of which 13.97 GiB is free. Including non-PyTorch memory, this process has 65.35 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    result = OptimizerFromGradientAccumulator(
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    trainer = DistributedTrainer(config_file)
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    result = OptimizerFromGradientAccumulator(
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 6 has a total capacty of 79.33 GiB of which 12.88 GiB is free. Including non-PyTorch memory, this process has 66.44 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    optimizer = optimizer_builder(named_param_groups)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 0 has a total capacty of 79.33 GiB of which 12.86 GiB is free. Including non-PyTorch memory, this process has 66.46 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    result = OptimizerFromGradientAccumulator(
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 5 has a total capacty of 79.33 GiB of which 13.13 GiB is free. Including non-PyTorch memory, this process has 66.19 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(    
+optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+Traceback (most recent call last):
+Traceback (most recent call last):
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+Traceback (most recent call last):
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    optimizer = optimizer_builder(named_param_groups)
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+Traceback (most recent call last):
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    trainer = DistributedTrainer(config_file)
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 0 has a total capacty of 79.33 GiB of which 13.07 GiB is free. Including non-PyTorch memory, this process has 66.25 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 113.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    trainer = DistributedTrainer(config_file)
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    trainer = DistributedTrainer(config_file)
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+Traceback (most recent call last):
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    trainer = DistributedTrainer(config_file)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 3 has a total capacty of 79.33 GiB of which 12.66 GiB is free. Including non-PyTorch memory, this process has 66.65 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 0 has a total capacty of 79.33 GiB of which 12.82 GiB is free. Including non-PyTorch memory, this process has 66.49 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    optimizer = optimizer_builder(named_param_groups)
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    optimizer = optimizer_builder(named_param_groups)
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    optimizer = optimizer_builder(named_param_groups)
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    optimizer = optimizer_builder(named_param_groups)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 6 has a total capacty of 79.33 GiB of which 13.08 GiB is free. Including non-PyTorch memory, this process has 66.24 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 113.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    optimizer = optimizer_builder(named_param_groups)
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 3 has a total capacty of 79.33 GiB of which 12.74 GiB is free. Including non-PyTorch memory, this process has 66.58 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    result = OptimizerFromGradientAccumulator(
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    trainer = DistributedTrainer(config_file)
+    result = OptimizerFromGradientAccumulator(
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 6 has a total capacty of 79.33 GiB of which 13.09 GiB is free. Including non-PyTorch memory, this process has 66.22 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    result = OptimizerFromGradientAccumulator(
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+optimizer = optimizer_builder(named_param_groups)
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 2 has a total capacty of 79.33 GiB of which 12.88 GiB is free. Including non-PyTorch memory, this process has 66.44 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 2 has a total capacty of 79.33 GiB of which 13.80 GiB is free. Including non-PyTorch memory, this process has 65.52 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 2 has a total capacty of 79.33 GiB of which 13.09 GiB is free. Including non-PyTorch memory, this process has 66.22 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 2 has a total capacty of 79.33 GiB of which 12.84 GiB is free. Including non-PyTorch memory, this process has 66.48 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    result = OptimizerFromGradientAccumulator(
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    result = OptimizerFromGradientAccumulator(
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 6 has a total capacty of 79.33 GiB of which 12.84 GiB is free. Including non-PyTorch memory, this process has 66.48 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    result = OptimizerFromGradientAccumulator(
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 2 has a total capacty of 79.33 GiB of which 13.05 GiB is free. Including non-PyTorch memory, this process has 66.26 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 6 has a total capacty of 79.33 GiB of which 13.04 GiB is free. Including non-PyTorch memory, this process has 66.28 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 113.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 7 has a total capacty of 79.33 GiB of which 13.31 GiB is free. Including non-PyTorch memory, this process has 66.01 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 4 has a total capacty of 79.33 GiB of which 13.03 GiB is free. Including non-PyTorch memory, this process has 66.29 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 7 has a total capacty of 79.33 GiB of which 13.23 GiB is free. Including non-PyTorch memory, this process has 66.09 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 6 has a total capacty of 79.33 GiB of which 13.80 GiB is free. Including non-PyTorch memory, this process has 65.52 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 1 has a total capacty of 79.33 GiB of which 13.05 GiB is free. Including non-PyTorch memory, this process has 66.26 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    trainer = DistributedTrainer(config_file)
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 3 has a total capacty of 79.33 GiB of which 12.96 GiB is free. Including non-PyTorch memory, this process has 66.36 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    trainer = DistributedTrainer(config_file)
+    trainer = DistributedTrainer(config_file)
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 6 has a total capacty of 79.33 GiB of which 12.80 GiB is free. Including non-PyTorch memory, this process has 66.52 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+Traceback (most recent call last):
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 3 has a total capacty of 79.33 GiB of which 12.88 GiB is free. Including non-PyTorch memory, this process has 66.44 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    result = OptimizerFromGradientAccumulator(
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    optimizer = optimizer_builder(named_param_groups)
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    trainer = DistributedTrainer(config_file)
+    result = OptimizerFromGradientAccumulator(
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    result = OptimizerFromGradientAccumulator(
+    result = OptimizerFromGradientAccumulator(
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    optimizer = optimizer_builder(named_param_groups)
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    result = OptimizerFromGradientAccumulator(
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    optimizer = optimizer_builder(named_param_groups)
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+Traceback (most recent call last):
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    optimizer = optimizer_builder(named_param_groups)
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    optimizer = optimizer_builder(named_param_groups)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 3 has a total capacty of 79.33 GiB of which 12.83 GiB is free. Including non-PyTorch memory, this process has 66.49 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 113.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    result = OptimizerFromGradientAccumulator(
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    trainer = DistributedTrainer(config_file)
+Traceback (most recent call last):
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    result = OptimizerFromGradientAccumulator(
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 6 has a total capacty of 79.33 GiB of which 13.05 GiB is free. Including non-PyTorch memory, this process has 66.26 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    trainer = DistributedTrainer(config_file)
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 1 has a total capacty of 79.33 GiB of which 13.00 GiB is free. Including non-PyTorch memory, this process has 66.31 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 113.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+Traceback (most recent call last):
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 7 has a total capacty of 79.33 GiB of which 13.05 GiB is free. Including non-PyTorch memory, this process has 66.26 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 0 has a total capacty of 79.33 GiB of which 13.08 GiB is free. Including non-PyTorch memory, this process has 66.24 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    trainer = DistributedTrainer(config_file)
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 3 has a total capacty of 79.33 GiB of which 13.62 GiB is free. Including non-PyTorch memory, this process has 65.70 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 4 has a total capacty of 79.33 GiB of which 13.03 GiB is free. Including non-PyTorch memory, this process has 66.29 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+    optimizer = optimizer_builder(named_param_groups)
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 5 has a total capacty of 79.33 GiB of which 13.80 GiB is free. Including non-PyTorch memory, this process has 65.52 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    optimizer = optimizer_builder(named_param_groups)
+    optimizer = optimizer_builder(named_param_groups)
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 4 has a total capacty of 79.33 GiB of which 12.78 GiB is free. Including non-PyTorch memory, this process has 66.54 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 3 has a total capacty of 79.33 GiB of which 12.95 GiB is free. Including non-PyTorch memory, this process has 66.37 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 113.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    result = OptimizerFromGradientAccumulator(
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+    result = OptimizerFromGradientAccumulator(
+    result = OptimizerFromGradientAccumulator(    
+result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    trainer = DistributedTrainer(config_file)
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 6 has a total capacty of 79.33 GiB of which 12.72 GiB is free. Including non-PyTorch memory, this process has 66.60 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 3 has a total capacty of 79.33 GiB of which 12.96 GiB is free. Including non-PyTorch memory, this process has 66.36 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+Traceback (most recent call last):
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+Traceback (most recent call last):
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    trainer = DistributedTrainer(config_file)
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+        gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    trainer = DistributedTrainer(config_file)    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+    optimizer = optimizer_builder(named_param_groups)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 5 has a total capacty of 79.33 GiB of which 13.05 GiB is free. Including non-PyTorch memory, this process has 66.26 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    trainer = DistributedTrainer(config_file)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 0 has a total capacty of 79.33 GiB of which 13.16 GiB is free. Including non-PyTorch memory, this process has 66.16 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 5 has a total capacty of 79.33 GiB of which 12.84 GiB is free. Including non-PyTorch memory, this process has 66.48 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+        contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")result = OptimizerFromGradientAccumulator(
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 4 has a total capacty of 79.33 GiB of which 13.77 GiB is free. Including non-PyTorch memory, this process has 65.54 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 2 has a total capacty of 79.33 GiB of which 12.72 GiB is free. Including non-PyTorch memory, this process has 66.60 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    optimizer = optimizer_builder(named_param_groups)
+    optimizer = optimizer_builder(named_param_groups)
+    optimizer = optimizer_builder(named_param_groups)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 1 has a total capacty of 79.33 GiB of which 12.84 GiB is free. Including non-PyTorch memory, this process has 66.48 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    optimizer = optimizer_builder(named_param_groups)
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    result = OptimizerFromGradientAccumulator(
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    trainer = DistributedTrainer(config_file)
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    result = OptimizerFromGradientAccumulator(
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 0 has a total capacty of 79.33 GiB of which 13.82 GiB is free. Including non-PyTorch memory, this process has 65.50 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    result = OptimizerFromGradientAccumulator(
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 7 has a total capacty of 79.33 GiB of which 12.90 GiB is free. Including non-PyTorch memory, this process has 66.42 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 2 has a total capacty of 79.33 GiB of which 13.13 GiB is free. Including non-PyTorch memory, this process has 66.19 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 0 has a total capacty of 79.33 GiB of which 13.11 GiB is free. Including non-PyTorch memory, this process has 66.21 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 113.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    trainer = DistributedTrainer(config_file)
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 5 has a total capacty of 79.33 GiB of which 13.00 GiB is free. Including non-PyTorch memory, this process has 66.31 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 113.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 2 has a total capacty of 79.33 GiB of which 12.76 GiB is free. Including non-PyTorch memory, this process has 66.56 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 5 has a total capacty of 79.33 GiB of which 12.72 GiB is free. Including non-PyTorch memory, this process has 66.60 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 1 has a total capacty of 79.33 GiB of which 12.88 GiB is free. Including non-PyTorch memory, this process has 66.44 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+Traceback (most recent call last):
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 0 has a total capacty of 79.33 GiB of which 12.73 GiB is free. Including non-PyTorch memory, this process has 66.58 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 113.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+    trainer = DistributedTrainer(config_file)
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+        gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 5 has a total capacty of 79.33 GiB of which 13.09 GiB is free. Including non-PyTorch memory, this process has 66.22 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    trainer = DistributedTrainer(config_file)
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+    optimizer = optimizer_builder(named_param_groups)
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    result = OptimizerFromGradientAccumulator(
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 3 has a total capacty of 79.33 GiB of which 12.92 GiB is free. Including non-PyTorch memory, this process has 66.40 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    trainer = DistributedTrainer(config_file)
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+        self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(optimizer = optimizer_builder(named_param_groups)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 3 has a total capacty of 79.33 GiB of which 12.70 GiB is free. Including non-PyTorch memory, this process has 66.62 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    trainer = DistributedTrainer(config_file)
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 4 has a total capacty of 79.33 GiB of which 12.70 GiB is free. Including non-PyTorch memory, this process has 66.62 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    optimizer = optimizer_builder(named_param_groups)
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 4 has a total capacty of 79.33 GiB of which 13.06 GiB is free. Including non-PyTorch memory, this process has 66.26 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 113.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 6 has a total capacty of 79.33 GiB of which 12.76 GiB is free. Including non-PyTorch memory, this process has 66.56 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 6 has a total capacty of 79.33 GiB of which 13.05 GiB is free. Including non-PyTorch memory, this process has 66.26 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    trainer = DistributedTrainer(config_file)
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    trainer = DistributedTrainer(config_file)
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 7 has a total capacty of 79.33 GiB of which 13.31 GiB is free. Including non-PyTorch memory, this process has 66.01 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 1 has a total capacty of 79.33 GiB of which 13.13 GiB is free. Including non-PyTorch memory, this process has 66.19 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+    result = OptimizerFromGradientAccumulator(
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 4 has a total capacty of 79.33 GiB of which 12.82 GiB is free. Including non-PyTorch memory, this process has 66.50 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    result = OptimizerFromGradientAccumulator(
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 2 has a total capacty of 79.33 GiB of which 13.05 GiB is free. Including non-PyTorch memory, this process has 66.26 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    result = OptimizerFromGradientAccumulator(
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    optimizer = optimizer_builder(named_param_groups)
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 5 has a total capacty of 79.33 GiB of which 12.88 GiB is free. Including non-PyTorch memory, this process has 66.44 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    result = OptimizerFromGradientAccumulator(
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 6 has a total capacty of 79.33 GiB of which 13.02 GiB is free. Including non-PyTorch memory, this process has 66.30 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    trainer = DistributedTrainer(config_file)
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 3 has a total capacty of 79.33 GiB of which 12.55 GiB is free. Including non-PyTorch memory, this process has 66.77 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    optimizer = optimizer_builder(named_param_groups)
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    trainer = DistributedTrainer(config_file)
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    result = OptimizerFromGradientAccumulator(
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    trainer = DistributedTrainer(config_file)
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 1 has a total capacty of 79.33 GiB of which 12.72 GiB is free. Including non-PyTorch memory, this process has 66.60 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 7 has a total capacty of 79.33 GiB of which 13.18 GiB is free. Including non-PyTorch memory, this process has 66.14 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 113.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    trainer = DistributedTrainer(config_file)
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 2 has a total capacty of 79.33 GiB of which 13.08 GiB is free. Including non-PyTorch memory, this process has 66.24 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 113.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    optimizer = optimizer_builder(named_param_groups)
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    trainer = DistributedTrainer(config_file)
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    result = OptimizerFromGradientAccumulator(
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 0 has a total capacty of 79.33 GiB of which 12.79 GiB is free. Including non-PyTorch memory, this process has 66.53 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    result = OptimizerFromGradientAccumulator(
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    trainer = DistributedTrainer(config_file)
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    optimizer = optimizer_builder(named_param_groups)
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+Traceback (most recent call last):
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 0 has a total capacty of 79.33 GiB of which 13.20 GiB is free. Including non-PyTorch memory, this process has 66.12 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    result = OptimizerFromGradientAccumulator(
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    trainer = DistributedTrainer(config_file)
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 2 has a total capacty of 79.33 GiB of which 13.02 GiB is free. Including non-PyTorch memory, this process has 66.30 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 5 has a total capacty of 79.33 GiB of which 13.02 GiB is free. Including non-PyTorch memory, this process has 66.30 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 6 has a total capacty of 79.33 GiB of which 12.88 GiB is free. Including non-PyTorch memory, this process has 66.44 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 1 has a total capacty of 79.33 GiB of which 13.09 GiB is free. Including non-PyTorch memory, this process has 66.22 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 4 has a total capacty of 79.33 GiB of which 12.74 GiB is free. Including non-PyTorch memory, this process has 66.58 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 0 has a total capacty of 79.33 GiB of which 13.04 GiB is free. Including non-PyTorch memory, this process has 66.28 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+Traceback (most recent call last):
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 6 has a total capacty of 79.33 GiB of which 13.17 GiB is free. Including non-PyTorch memory, this process has 66.15 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 5 has a total capacty of 79.33 GiB of which 12.93 GiB is free. Including non-PyTorch memory, this process has 66.39 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 93.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    optimizer = optimizer_builder(named_param_groups)
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    result = OptimizerFromGradientAccumulator(
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 6 has a total capacty of 79.33 GiB of which 12.81 GiB is free. Including non-PyTorch memory, this process has 66.51 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 93.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 4 has a total capacty of 79.33 GiB of which 12.99 GiB is free. Including non-PyTorch memory, this process has 66.33 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 4 has a total capacty of 79.33 GiB of which 12.99 GiB is free. Including non-PyTorch memory, this process has 66.33 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 7 has a total capacty of 79.33 GiB of which 13.27 GiB is free. Including non-PyTorch memory, this process has 66.05 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    trainer = DistributedTrainer(config_file)
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    result = OptimizerFromGradientAccumulator(
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    optimizer = optimizer_builder(named_param_groups)
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 5 has a total capacty of 79.33 GiB of which 12.76 GiB is free. Including non-PyTorch memory, this process has 66.56 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 0 has a total capacty of 79.33 GiB of which 12.90 GiB is free. Including non-PyTorch memory, this process has 66.42 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    optimizer = optimizer_builder(named_param_groups)
+    result = OptimizerFromGradientAccumulator(
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    result = OptimizerFromGradientAccumulator(
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+    result = OptimizerFromGradientAccumulator(
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 3 has a total capacty of 79.33 GiB of which 12.92 GiB is free. Including non-PyTorch memory, this process has 66.40 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+Traceback (most recent call last):
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 1 has a total capacty of 79.33 GiB of which 12.88 GiB is free. Including non-PyTorch memory, this process has 66.44 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    trainer = DistributedTrainer(config_file)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 0 has a total capacty of 79.33 GiB of which 13.12 GiB is free. Including non-PyTorch memory, this process has 66.20 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    optimizer = optimizer_builder(named_param_groups)
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    trainer = DistributedTrainer(config_file)
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+Traceback (most recent call last):
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 4 has a total capacty of 79.33 GiB of which 12.86 GiB is free. Including non-PyTorch memory, this process has 66.46 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    optimizer = optimizer_builder(named_param_groups)
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+Traceback (most recent call last):
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    result = OptimizerFromGradientAccumulator(
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    optimizer = optimizer_builder(named_param_groups)
+    optimizer = optimizer_builder(named_param_groups)
+Traceback (most recent call last):
+Traceback (most recent call last):
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 1 has a total capacty of 79.33 GiB of which 12.76 GiB is free. Including non-PyTorch memory, this process has 66.56 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    trainer = DistributedTrainer(config_file)
+    result = OptimizerFromGradientAccumulator(
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 1 has a total capacty of 79.33 GiB of which 12.88 GiB is free. Including non-PyTorch memory, this process has 66.44 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    optimizer = optimizer_builder(named_param_groups)
+    trainer = DistributedTrainer(config_file)
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    result = OptimizerFromGradientAccumulator(
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    trainer = DistributedTrainer(config_file)
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    optimizer = optimizer_builder(named_param_groups)
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    optimizer = optimizer_builder(named_param_groups)
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 3 has a total capacty of 79.33 GiB of which 12.88 GiB is free. Including non-PyTorch memory, this process has 66.44 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    result = OptimizerFromGradientAccumulator(
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 3 has a total capacty of 79.33 GiB of which 12.59 GiB is free. Including non-PyTorch memory, this process has 66.73 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 2 has a total capacty of 79.33 GiB of which 13.17 GiB is free. Including non-PyTorch memory, this process has 66.15 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    result = OptimizerFromGradientAccumulator(
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 2 has a total capacty of 79.33 GiB of which 13.02 GiB is free. Including non-PyTorch memory, this process has 66.30 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 1 has a total capacty of 79.33 GiB of which 13.09 GiB is free. Including non-PyTorch memory, this process has 66.22 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 7 has a total capacty of 79.33 GiB of which 13.22 GiB is free. Including non-PyTorch memory, this process has 66.10 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 113.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 7 has a total capacty of 79.33 GiB of which 13.05 GiB is free. Including non-PyTorch memory, this process has 66.26 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+    trainer = DistributedTrainer(config_file)
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+    trainer = DistributedTrainer(config_file)
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 4 has a total capacty of 79.33 GiB of which 13.15 GiB is free. Including non-PyTorch memory, this process has 66.17 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+Traceback (most recent call last):
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 7 has a total capacty of 79.33 GiB of which 13.23 GiB is free. Including non-PyTorch memory, this process has 66.09 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 7 has a total capacty of 79.33 GiB of which 12.94 GiB is free. Including non-PyTorch memory, this process has 66.38 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 2 has a total capacty of 79.33 GiB of which 12.76 GiB is free. Including non-PyTorch memory, this process has 66.56 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+Traceback (most recent call last):
+Traceback (most recent call last):
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    optimizer = optimizer_builder(named_param_groups)
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 1 has a total capacty of 79.33 GiB of which 12.93 GiB is free. Including non-PyTorch memory, this process has 66.39 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 93.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 1 has a total capacty of 79.33 GiB of which 13.80 GiB is free. Including non-PyTorch memory, this process has 65.52 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    result = OptimizerFromGradientAccumulator(
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+Traceback (most recent call last):
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    trainer = DistributedTrainer(config_file)
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    optimizer = optimizer_builder(named_param_groups)
+    result = OptimizerFromGradientAccumulator(
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    trainer = DistributedTrainer(config_file)
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    result = OptimizerFromGradientAccumulator(
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    result = OptimizerFromGradientAccumulator(
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    optimizer = optimizer_builder(named_param_groups)
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    optimizer = optimizer_builder(named_param_groups)
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    result = OptimizerFromGradientAccumulator(
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    result = OptimizerFromGradientAccumulator(
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 0 has a total capacty of 79.33 GiB of which 13.14 GiB is free. Including non-PyTorch memory, this process has 66.17 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 113.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    trainer = DistributedTrainer(config_file)
+    result = OptimizerFromGradientAccumulator(
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 0 has a total capacty of 79.33 GiB of which 13.12 GiB is free. Including non-PyTorch memory, this process has 66.20 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 7 has a total capacty of 79.33 GiB of which 13.27 GiB is free. Including non-PyTorch memory, this process has 66.05 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 7 has a total capacty of 79.33 GiB of which 13.05 GiB is free. Including non-PyTorch memory, this process has 66.26 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 7 has a total capacty of 79.33 GiB of which 13.27 GiB is free. Including non-PyTorch memory, this process has 66.05 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 2 has a total capacty of 79.33 GiB of which 12.88 GiB is free. Including non-PyTorch memory, this process has 66.44 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    optimizer = optimizer_builder(named_param_groups)
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 6 has a total capacty of 79.33 GiB of which 12.76 GiB is free. Including non-PyTorch memory, this process has 66.56 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    result = OptimizerFromGradientAccumulator(
+    torch.cudatrainer = DistributedTrainer(config_file).OutOfMemoryError: 
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+CUDA out of memory. Tried to allocate 54.66 GiB. GPU 7 has a total capacty of 79.33 GiB of which 13.97 GiB is free. Including non-PyTorch memory, this process has 65.35 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    result = OptimizerFromGradientAccumulator(
+Traceback (most recent call last):
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    result = OptimizerFromGradientAccumulator(
+trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+Traceback (most recent call last):
+    trainer = DistributedTrainer(config_file)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 5 has a total capacty of 79.33 GiB of which 12.76 GiB is free. Including non-PyTorch memory, this process has 66.56 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 0 has a total capacty of 79.33 GiB of which 13.16 GiB is free. Including non-PyTorch memory, this process has 66.16 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    trainer = DistributedTrainer(config_file)
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    trainer = DistributedTrainer(config_file)
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 5 has a total capacty of 79.33 GiB of which 12.88 GiB is free. Including non-PyTorch memory, this process has 66.44 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    trainer = DistributedTrainer(config_file)
+Traceback (most recent call last):
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    trainer = DistributedTrainer(config_file)
+torch.cuda.OutOfMemoryError  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 1 has a total capacty of 79.33 GiB of which 12.76 GiB is free. Including non-PyTorch memory, this process has 66.56 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    optimizer = optimizer_builder(named_param_groups)
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    optimizer = optimizer_builder(named_param_groups)
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+Traceback (most recent call last):
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    optimizer = optimizer_builder(named_param_groups)
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    result = OptimizerFromGradientAccumulator(
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    result = OptimizerFromGradientAccumulator(
+    trainer = DistributedTrainer(config_file)
+    optimizer = optimizer_builder(named_param_groups)
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    result = OptimizerFromGradientAccumulator(
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    result = OptimizerFromGradientAccumulator(
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 7 has a total capacty of 79.33 GiB of which 13.10 GiB is free. Including non-PyTorch memory, this process has 66.22 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 93.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 0 has a total capacty of 79.33 GiB of which 13.04 GiB is free. Including non-PyTorch memory, this process has 66.28 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+Traceback (most recent call last):
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 2 has a total capacty of 79.33 GiB of which 12.81 GiB is free. Including non-PyTorch memory, this process has 66.51 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 93.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    optimizer = optimizer_builder(named_param_groups)
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+Traceback (most recent call last):
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    result = OptimizerFromGradientAccumulator(
+        optimizer = optimizer_builder(named_param_groups)trainer = DistributedTrainer(config_file)
+    trainer = DistributedTrainer(config_file)
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+    optimizer = optimizer_builder(named_param_groups)
+Traceback (most recent call last):
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    optimizer = optimizer_builder(named_param_groups)
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 3 has a total capacty of 79.33 GiB of which 12.66 GiB is free. Including non-PyTorch memory, this process has 66.65 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 2 has a total capacty of 79.33 GiB of which 12.88 GiB is free. Including non-PyTorch memory, this process has 66.44 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    result = OptimizerFromGradientAccumulator(
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+Traceback (most recent call last):
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+Traceback (most recent call last):
+    result = OptimizerFromGradientAccumulator(
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+    result = OptimizerFromGradientAccumulator(
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 6 has a total capacty of 79.33 GiB of which 13.02 GiB is free. Including non-PyTorch memory, this process has 66.30 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 0 has a total capacty of 79.33 GiB of which 12.90 GiB is free. Including non-PyTorch memory, this process has 66.42 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 2 has a total capacty of 79.33 GiB of which 13.80 GiB is free. Including non-PyTorch memory, this process has 65.52 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    result = OptimizerFromGradientAccumulator(
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 5 has a total capacty of 79.33 GiB of which 13.09 GiB is free. Including non-PyTorch memory, this process has 66.22 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 3 has a total capacty of 79.33 GiB of which 12.88 GiB is free. Including non-PyTorch memory, this process has 66.44 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 4 has a total capacty of 79.33 GiB of which 13.10 GiB is free. Including non-PyTorch memory, this process has 66.22 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 113.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    trainer = DistributedTrainer(config_file)
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+Traceback (most recent call last):
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+Traceback (most recent call last):
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    trainer = DistributedTrainer(config_file)
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    optimizer = optimizer_builder(named_param_groups)
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 4 has a total capacty of 79.33 GiB of which 12.86 GiB is free. Including non-PyTorch memory, this process has 66.46 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    trainer = DistributedTrainer(config_file)
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 6 has a total capacty of 79.33 GiB of which 13.09 GiB is free. Including non-PyTorch memory, this process has 66.22 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 0 has a total capacty of 79.33 GiB of which 13.04 GiB is free. Including non-PyTorch memory, this process has 66.28 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    optimizer = optimizer_builder(named_param_groups)
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    trainer = DistributedTrainer(config_file)
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    result = OptimizerFromGradientAccumulator(
+    trainer = DistributedTrainer(config_file)
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    result = OptimizerFromGradientAccumulator(
+    trainer = DistributedTrainer(config_file)
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+Traceback (most recent call last):
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    optimizer = optimizer_builder(named_param_groups)
+    trainer = DistributedTrainer(config_file)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 5 has a total capacty of 79.33 GiB of which 13.09 GiB is free. Including non-PyTorch memory, this process has 66.22 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    result = OptimizerFromGradientAccumulator(
+        self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(optimizer = optimizer_builder(named_param_groups)
+
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+        self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 1 has a total capacty of 79.33 GiB of which 13.13 GiB is free. Including non-PyTorch memory, this process has 66.19 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 3 has a total capacty of 79.33 GiB of which 12.92 GiB is free. Including non-PyTorch memory, this process has 66.40 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+Traceback (most recent call last):
+    trainer = DistributedTrainer(config_file)
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+Traceback (most recent call last):
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+        optimizer = optimizer_builder(named_param_groups)self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    result = OptimizerFromGradientAccumulator(
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 3 has a total capacty of 79.33 GiB of which 12.84 GiB is free. Including non-PyTorch memory, this process has 66.48 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    result = OptimizerFromGradientAccumulator(
+    optimizer = optimizer_builder(named_param_groups)
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 1 has a total capacty of 79.33 GiB of which 12.84 GiB is free. Including non-PyTorch memory, this process has 66.48 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    optimizer = optimizer_builder(named_param_groups)
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    result = OptimizerFromGradientAccumulator(
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    optimizer = optimizer_builder(named_param_groups)
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+Traceback (most recent call last):
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 6 has a total capacty of 79.33 GiB of which 12.88 GiB is free. Including non-PyTorch memory, this process has 66.44 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    result = OptimizerFromGradientAccumulator(
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    trainer = DistributedTrainer(config_file)
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    result = OptimizerFromGradientAccumulator(
+    result = OptimizerFromGradientAccumulator(
+Traceback (most recent call last):
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    result = OptimizerFromGradientAccumulator(
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    trainer = DistributedTrainer(config_file)
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    result = OptimizerFromGradientAccumulator(
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 7 has a total capacty of 79.33 GiB of which 13.27 GiB is free. Including non-PyTorch memory, this process has 66.05 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 3 has a total capacty of 79.33 GiB of which 12.59 GiB is free. Including non-PyTorch memory, this process has 66.73 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    result = OptimizerFromGradientAccumulator(
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 2 has a total capacty of 79.33 GiB of which 13.13 GiB is free. Including non-PyTorch memory, this process has 66.19 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+Traceback (most recent call last):
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    optimizer = optimizer_builder(named_param_groups)
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 4 has a total capacty of 79.33 GiB of which 13.11 GiB is free. Including non-PyTorch memory, this process has 66.21 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    result = OptimizerFromGradientAccumulator(
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    optimizer = optimizer_builder(named_param_groups)
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    optimizer = optimizer_builder(named_param_groups)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 5 has a total capacty of 79.33 GiB of which 13.80 GiB is free. Including non-PyTorch memory, this process has 65.52 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 2 has a total capacty of 79.33 GiB of which 13.12 GiB is free. Including non-PyTorch memory, this process has 66.20 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 113.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 6 has a total capacty of 79.33 GiB of which 13.80 GiB is free. Including non-PyTorch memory, this process has 65.52 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    result = OptimizerFromGradientAccumulator(
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 2 has a total capacty of 79.33 GiB of which 13.13 GiB is free. Including non-PyTorch memory, this process has 66.19 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 7 has a total capacty of 79.33 GiB of which 13.19 GiB is free. Including non-PyTorch memory, this process has 66.13 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    optimizer = optimizer_builder(named_param_groups)
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    result = OptimizerFromGradientAccumulator(
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(    
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+Traceback (most recent call last):
+    optimizer = optimizer_builder(named_param_groups)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 5 has a total capacty of 79.33 GiB of which 12.88 GiB is free. Including non-PyTorch memory, this process has 66.44 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 2 has a total capacty of 79.33 GiB of which 13.05 GiB is free. Including non-PyTorch memory, this process has 66.26 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 2 has a total capacty of 79.33 GiB of which 13.05 GiB is free. Including non-PyTorch memory, this process has 66.26 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    result = OptimizerFromGradientAccumulator(
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 1 has a total capacty of 79.33 GiB of which 13.04 GiB is free. Including non-PyTorch memory, this process has 66.28 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 113.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    result = OptimizerFromGradientAccumulator(
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 7 has a total capacty of 79.33 GiB of which 12.94 GiB is free. Including non-PyTorch memory, this process has 66.38 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 4 has a total capacty of 79.33 GiB of which 13.07 GiB is free. Including non-PyTorch memory, this process has 66.25 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    result = OptimizerFromGradientAccumulator(
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    result = OptimizerFromGradientAccumulator(
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 6 has a total capacty of 79.33 GiB of which 12.76 GiB is free. Including non-PyTorch memory, this process has 66.56 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    optimizer = optimizer_builder(named_param_groups)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 1 has a total capacty of 79.33 GiB of which 13.09 GiB is free. Including non-PyTorch memory, this process has 66.22 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 6 has a total capacty of 79.33 GiB of which 13.09 GiB is free. Including non-PyTorch memory, this process has 66.22 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    trainer = DistributedTrainer(config_file)
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    result = OptimizerFromGradientAccumulator(
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    trainer = DistributedTrainer(config_file)
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    optimizer = optimizer_builder(named_param_groups)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 1 has a total capacty of 79.33 GiB of which 13.02 GiB is free. Including non-PyTorch memory, this process has 66.30 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 7 has a total capacty of 79.33 GiB of which 13.02 GiB is free. Including non-PyTorch memory, this process has 66.30 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+Traceback (most recent call last):
+    result = OptimizerFromGradientAccumulator(
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 0 has a total capacty of 79.33 GiB of which 12.79 GiB is free. Including non-PyTorch memory, this process has 66.53 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 0 has a total capacty of 79.33 GiB of which 13.16 GiB is free. Including non-PyTorch memory, this process has 66.16 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    result = OptimizerFromGradientAccumulator(
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+Traceback (most recent call last):
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    optimizer = optimizer_builder(named_param_groups)
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 3 has a total capacty of 79.33 GiB of which 12.92 GiB is free. Including non-PyTorch memory, this process has 66.40 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    trainer = DistributedTrainer(config_file)
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    optimizer = optimizer_builder(named_param_groups)
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    trainer = DistributedTrainer(config_file)
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+Traceback (most recent call last):
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    result = OptimizerFromGradientAccumulator(
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 5 has a total capacty of 79.33 GiB of which 13.09 GiB is free. Including non-PyTorch memory, this process has 66.22 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    trainer = DistributedTrainer(config_file)
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 3 has a total capacty of 79.33 GiB of which 12.75 GiB is free. Including non-PyTorch memory, this process has 66.57 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 93.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    trainer = DistributedTrainer(config_file)
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 0 has a total capacty of 79.33 GiB of which 12.83 GiB is free. Including non-PyTorch memory, this process has 66.49 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 93.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    optimizer = optimizer_builder(named_param_groups)
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 2 has a total capacty of 79.33 GiB of which 13.09 GiB is free. Including non-PyTorch memory, this process has 66.22 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    result = OptimizerFromGradientAccumulator(
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    result = OptimizerFromGradientAccumulator(
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 4 has a total capacty of 79.33 GiB of which 13.77 GiB is free. Including non-PyTorch memory, this process has 65.54 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    result = OptimizerFromGradientAccumulator(
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    optimizer = optimizer_builder(named_param_groups)
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 6 has a total capacty of 79.33 GiB of which 13.05 GiB is free. Including non-PyTorch memory, this process has 66.26 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    optimizer = optimizer_builder(named_param_groups)
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    trainer = DistributedTrainer(config_file)
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    result = OptimizerFromGradientAccumulator(
+Traceback (most recent call last):
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    trainer = DistributedTrainer(config_file)
+    trainer = DistributedTrainer(config_file)
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 3 has a total capacty of 79.33 GiB of which 12.70 GiB is free. Including non-PyTorch memory, this process has 66.62 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    result = OptimizerFromGradientAccumulator(
+    trainer = DistributedTrainer(config_file)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 1 has a total capacty of 79.33 GiB of which 12.84 GiB is free. Including non-PyTorch memory, this process has 66.48 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+Traceback (most recent call last):
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    trainer = DistributedTrainer(config_file)
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+    trainer = DistributedTrainer(config_file)
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    trainer = DistributedTrainer(config_file)
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+    optimizer = optimizer_builder(named_param_groups)  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    trainer = DistributedTrainer(config_file)
+    trainer = DistributedTrainer(config_file)
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    optimizer = optimizer_builder(named_param_groups)
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 3 has a total capacty of 79.33 GiB of which 12.70 GiB is free. Including non-PyTorch memory, this process has 66.62 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    optimizer = optimizer_builder(named_param_groups)
+    optimizer = optimizer_builder(named_param_groups)
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+    result = OptimizerFromGradientAccumulator(
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 6 has a total capacty of 79.33 GiB of which 13.13 GiB is free. Including non-PyTorch memory, this process has 66.19 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    trainer = DistributedTrainer(config_file)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 4 has a total capacty of 79.33 GiB of which 13.03 GiB is free. Including non-PyTorch memory, this process has 66.29 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 3 has a total capacty of 79.33 GiB of which 12.55 GiB is free. Including non-PyTorch memory, this process has 66.77 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    optimizer = optimizer_builder(named_param_groups)
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+    result = OptimizerFromGradientAccumulator(
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    result = OptimizerFromGradientAccumulator(
+    result = OptimizerFromGradientAccumulator(
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 4 has a total capacty of 79.33 GiB of which 12.99 GiB is free. Including non-PyTorch memory, this process has 66.33 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    optimizer = optimizer_builder(named_param_groups)
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    result = OptimizerFromGradientAccumulator(
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 3 has a total capacty of 79.33 GiB of which 12.96 GiB is free. Including non-PyTorch memory, this process has 66.36 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 5 has a total capacty of 79.33 GiB of which 13.05 GiB is free. Including non-PyTorch memory, this process has 66.26 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    result = OptimizerFromGradientAccumulator(
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    result = OptimizerFromGradientAccumulator(
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    result = OptimizerFromGradientAccumulator(
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 1 has a total capacty of 79.33 GiB of which 13.05 GiB is free. Including non-PyTorch memory, this process has 66.26 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    optimizer = optimizer_builder(named_param_groups)
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    result = OptimizerFromGradientAccumulator(
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 5 has a total capacty of 79.33 GiB of which 13.04 GiB is free. Including non-PyTorch memory, this process has 66.28 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 113.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    trainer = DistributedTrainer(config_file)
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 3 has a total capacty of 79.33 GiB of which 12.87 GiB is free. Including non-PyTorch memory, this process has 66.45 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 113.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    result = OptimizerFromGradientAccumulator(
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 7 has a total capacty of 79.33 GiB of which 13.19 GiB is free. Including non-PyTorch memory, this process has 66.13 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+Traceback (most recent call last):
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 2 has a total capacty of 79.33 GiB of which 13.09 GiB is free. Including non-PyTorch memory, this process has 66.22 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+Traceback (most recent call last):
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 4 has a total capacty of 79.33 GiB of which 13.03 GiB is free. Including non-PyTorch memory, this process has 66.29 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    result = OptimizerFromGradientAccumulator(
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 1 has a total capacty of 79.33 GiB of which 13.05 GiB is free. Including non-PyTorch memory, this process has 66.26 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 3 has a total capacty of 79.33 GiB of which 12.66 GiB is free. Including non-PyTorch memory, this process has 66.65 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 3 has a total capacty of 79.33 GiB of which 12.84 GiB is free. Including non-PyTorch memory, this process has 66.48 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    optimizer = optimizer_builder(named_param_groups)
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 4 has a total capacty of 79.33 GiB of which 12.79 GiB is free. Including non-PyTorch memory, this process has 66.53 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 93.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 2 has a total capacty of 79.33 GiB of which 12.72 GiB is free. Including non-PyTorch memory, this process has 66.60 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 0 has a total capacty of 79.33 GiB of which 12.79 GiB is free. Including non-PyTorch memory, this process has 66.53 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 6 has a total capacty of 79.33 GiB of which 13.12 GiB is free. Including non-PyTorch memory, this process has 66.20 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 113.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 5 has a total capacty of 79.33 GiB of which 13.02 GiB is free. Including non-PyTorch memory, this process has 66.30 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+    trainer = DistributedTrainer(config_file)
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    result = OptimizerFromGradientAccumulator(
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 0 has a total capacty of 79.33 GiB of which 13.82 GiB is free. Including non-PyTorch memory, this process has 65.50 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    optimizer = optimizer_builder(named_param_groups)
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 4 has a total capacty of 79.33 GiB of which 12.74 GiB is free. Including non-PyTorch memory, this process has 66.58 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    trainer = DistributedTrainer(config_file)
+    result = OptimizerFromGradientAccumulator(
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    optimizer = optimizer_builder(named_param_groups)
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    optimizer = optimizer_builder(named_param_groups)
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 3 has a total capacty of 79.33 GiB of which 13.62 GiB is free. Including non-PyTorch memory, this process has 65.70 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    optimizer = optimizer_builder(named_param_groups)
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    trainer = DistributedTrainer(config_file)
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    optimizer = optimizer_builder(named_param_groups)
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+Traceback (most recent call last):
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 7 has a total capacty of 79.33 GiB of which 13.02 GiB is free. Including non-PyTorch memory, this process has 66.30 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 6 has a total capacty of 79.33 GiB of which 13.13 GiB is free. Including non-PyTorch memory, this process has 66.19 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 6 has a total capacty of 79.33 GiB of which 13.05 GiB is free. Including non-PyTorch memory, this process has 66.26 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 5 has a total capacty of 79.33 GiB of which 12.84 GiB is free. Including non-PyTorch memory, this process has 66.48 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 1 has a total capacty of 79.33 GiB of which 13.13 GiB is free. Including non-PyTorch memory, this process has 66.19 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    optimizer = optimizer_builder(named_param_groups)
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    optimizer = optimizer_builder(named_param_groups)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 2 has a total capacty of 79.33 GiB of which 12.76 GiB is free. Including non-PyTorch memory, this process has 66.56 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    result = OptimizerFromGradientAccumulator(
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 2 has a total capacty of 79.33 GiB of which 13.02 GiB is free. Including non-PyTorch memory, this process has 66.30 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+    result = OptimizerFromGradientAccumulator(
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 6 has a total capacty of 79.33 GiB of which 13.02 GiB is free. Including non-PyTorch memory, this process has 66.30 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+Traceback (most recent call last):
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 6 has a total capacty of 79.33 GiB of which 12.84 GiB is free. Including non-PyTorch memory, this process has 66.48 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+Traceback (most recent call last):
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    trainer = DistributedTrainer(config_file)
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 6 has a total capacty of 79.33 GiB of which 12.84 GiB is free. Including non-PyTorch memory, this process has 66.48 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    trainer = DistributedTrainer(config_file)
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+Traceback (most recent call last):
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    trainer = DistributedTrainer(config_file)
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+Traceback (most recent call last):
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 0 has a total capacty of 79.33 GiB of which 12.86 GiB is free. Including non-PyTorch memory, this process has 66.46 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    optimizer = optimizer_builder(named_param_groups)
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    optimizer = optimizer_builder(named_param_groups)
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+Traceback (most recent call last):
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 4 has a total capacty of 79.33 GiB of which 13.07 GiB is free. Including non-PyTorch memory, this process has 66.25 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    trainer = DistributedTrainer(config_file)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 1 has a total capacty of 79.33 GiB of which 13.02 GiB is free. Including non-PyTorch memory, this process has 66.30 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    result = OptimizerFromGradientAccumulator(
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    result = OptimizerFromGradientAccumulator(
+    result = OptimizerFromGradientAccumulator(
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    optimizer = optimizer_builder(named_param_groups)
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    trainer = DistributedTrainer(config_file)
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    result = OptimizerFromGradientAccumulator(
+    result = OptimizerFromGradientAccumulator(
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    optimizer = optimizer_builder(named_param_groups)
+    optimizer = optimizer_builder(named_param_groups)
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    trainer = DistributedTrainer(config_file)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 4 has a total capacty of 79.33 GiB of which 13.11 GiB is free. Including non-PyTorch memory, this process has 66.21 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    result = OptimizerFromGradientAccumulator(
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    optimizer = optimizer_builder(named_param_groups)
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 5 has a total capacty of 79.33 GiB of which 13.05 GiB is free. Including non-PyTorch memory, this process has 66.26 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    result = OptimizerFromGradientAccumulator(
+Traceback (most recent call last):
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    result = OptimizerFromGradientAccumulator(
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 5 has a total capacty of 79.33 GiB of which 13.13 GiB is free. Including non-PyTorch memory, this process has 66.19 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+Traceback (most recent call last):
+Traceback (most recent call last):
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    trainer = DistributedTrainer(config_file)
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 4 has a total capacty of 79.33 GiB of which 12.74 GiB is free. Including non-PyTorch memory, this process has 66.58 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    optimizer = optimizer_builder(named_param_groups)
+    result = OptimizerFromGradientAccumulator(
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 6 has a total capacty of 79.33 GiB of which 12.72 GiB is free. Including non-PyTorch memory, this process has 66.60 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 0 has a total capacty of 79.33 GiB of which 13.08 GiB is free. Including non-PyTorch memory, this process has 66.24 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    trainer = DistributedTrainer(config_file)
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    result = OptimizerFromGradientAccumulator(
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    trainer = DistributedTrainer(config_file)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 7 has a total capacty of 79.33 GiB of which 13.31 GiB is free. Including non-PyTorch memory, this process has 66.01 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 6 has a total capacty of 79.33 GiB of which 12.76 GiB is free. Including non-PyTorch memory, this process has 66.56 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+Traceback (most recent call last):
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    trainer = DistributedTrainer(config_file)torch.cuda
+Traceback (most recent call last):
+.OutOfMemoryError  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 0 has a total capacty of 79.33 GiB of which 13.08 GiB is free. Including non-PyTorch memory, this process has 66.24 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 0 has a total capacty of 79.33 GiB of which 12.75 GiB is free. Including non-PyTorch memory, this process has 66.57 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 0 has a total capacty of 79.33 GiB of which 13.12 GiB is free. Including non-PyTorch memory, this process has 66.20 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 1 has a total capacty of 79.33 GiB of which 13.09 GiB is free. Including non-PyTorch memory, this process has 66.22 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    optimizer = optimizer_builder(named_param_groups)
+    trainer = DistributedTrainer(config_file)
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    trainer = DistributedTrainer(config_file)
+        gradient_accumulator = gradient_accumulator_builder(name_to_param.items())self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 2 has a total capacty of 79.33 GiB of which 12.76 GiB is free. Including non-PyTorch memory, this process has 66.56 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 1 has a total capacty of 79.33 GiB of which 12.72 GiB is free. Including non-PyTorch memory, this process has 66.60 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    result = OptimizerFromGradientAccumulator(
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 7 has a total capacty of 79.33 GiB of which 12.90 GiB is free. Including non-PyTorch memory, this process has 66.42 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 6 has a total capacty of 79.33 GiB of which 12.72 GiB is free. Including non-PyTorch memory, this process has 66.60 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    result = OptimizerFromGradientAccumulator(
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    optimizer = optimizer_builder(named_param_groups)
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 2 has a total capacty of 79.33 GiB of which 12.72 GiB is free. Including non-PyTorch memory, this process has 66.60 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    result = OptimizerFromGradientAccumulator(
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    optimizer = optimizer_builder(named_param_groups)
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    trainer = DistributedTrainer(config_file)
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 0 has a total capacty of 79.33 GiB of which 12.75 GiB is free. Including non-PyTorch memory, this process has 66.57 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    optimizer = optimizer_builder(named_param_groups)
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 7 has a total capacty of 79.33 GiB of which 13.23 GiB is free. Including non-PyTorch memory, this process has 66.09 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 7 has a total capacty of 79.33 GiB of which 12.94 GiB is free. Including non-PyTorch memory, this process has 66.38 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    result = OptimizerFromGradientAccumulator(
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 5 has a total capacty of 79.33 GiB of which 12.84 GiB is free. Including non-PyTorch memory, this process has 66.48 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 7 has a total capacty of 79.33 GiB of which 13.31 GiB is free. Including non-PyTorch memory, this process has 66.01 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    optimizer = optimizer_builder(named_param_groups)
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    result = OptimizerFromGradientAccumulator(
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 6 has a total capacty of 79.33 GiB of which 12.72 GiB is free. Including non-PyTorch memory, this process has 66.60 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 1 has a total capacty of 79.33 GiB of which 12.88 GiB is free. Including non-PyTorch memory, this process has 66.44 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 5 has a total capacty of 79.33 GiB of which 12.72 GiB is free. Including non-PyTorch memory, this process has 66.60 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    result = OptimizerFromGradientAccumulator(
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    trainer = DistributedTrainer(config_file)    trainer = DistributedTrainer(config_file)
+
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 5 has a total capacty of 79.33 GiB of which 13.13 GiB is free. Including non-PyTorch memory, this process has 66.19 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 5 has a total capacty of 79.33 GiB of which 12.72 GiB is free. Including non-PyTorch memory, this process has 66.60 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    optimizer = optimizer_builder(named_param_groups)
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+        optimizer = optimizer_builder(named_param_groups)result = OptimizerFromGradientAccumulator(
+
+    optimizer = optimizer_builder(named_param_groups)
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    result = OptimizerFromGradientAccumulator(
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 4 has a total capacty of 79.33 GiB of which 12.70 GiB is free. Including non-PyTorch memory, this process has 66.62 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    result = OptimizerFromGradientAccumulator(
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 4 has a total capacty of 79.33 GiB of which 12.70 GiB is free. Including non-PyTorch memory, this process has 66.62 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    trainer = DistributedTrainer(config_file)
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    trainer = DistributedTrainer(config_file)
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 3 has a total capacty of 79.33 GiB of which 12.96 GiB is free. Including non-PyTorch memory, this process has 66.36 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 1 has a total capacty of 79.33 GiB of which 12.80 GiB is free. Including non-PyTorch memory, this process has 66.52 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 7 has a total capacty of 79.33 GiB of which 12.98 GiB is free. Including non-PyTorch memory, this process has 66.34 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 4 has a total capacty of 79.33 GiB of which 13.07 GiB is free. Including non-PyTorch memory, this process has 66.25 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 6 has a total capacty of 79.33 GiB of which 13.09 GiB is free. Including non-PyTorch memory, this process has 66.22 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    optimizer = optimizer_builder(named_param_groups)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 7 has a total capacty of 79.33 GiB of which 13.19 GiB is free. Including non-PyTorch memory, this process has 66.13 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 0 has a total capacty of 79.33 GiB of which 12.75 GiB is free. Including non-PyTorch memory, this process has 66.57 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    optimizer = optimizer_builder(named_param_groups)
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    result = OptimizerFromGradientAccumulator(
+    result = OptimizerFromGradientAccumulator(
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 4 has a total capacty of 79.33 GiB of which 12.86 GiB is free. Including non-PyTorch memory, this process has 66.46 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+        gradient_accumulator = gradient_accumulator_builder(name_to_param.items())contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 5 has a total capacty of 79.33 GiB of which 12.76 GiB is free. Including non-PyTorch memory, this process has 66.56 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+Traceback (most recent call last):
+torch.cuda.  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 0 has a total capacty of 79.33 GiB of which 12.90 GiB is free. Including non-PyTorch memory, this process has 66.42 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    result = OptimizerFromGradientAccumulator(
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 5 has a total capacty of 79.33 GiB of which 12.88 GiB is free. Including non-PyTorch memory, this process has 66.44 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 2 has a total capacty of 79.33 GiB of which 12.84 GiB is free. Including non-PyTorch memory, this process has 66.48 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    trainer = DistributedTrainer(config_file)
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    trainer = DistributedTrainer(config_file)
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    trainer = DistributedTrainer(config_file)
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    optimizer = optimizer_builder(named_param_groups)
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+Traceback (most recent call last):
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+      File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    result = OptimizerFromGradientAccumulator(
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    optimizer = optimizer_builder(named_param_groups)
+    trainer = DistributedTrainer(config_file)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 7 has a total capacty of 79.33 GiB of which 12.90 GiB is free. Including non-PyTorch memory, this process has 66.42 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    result = OptimizerFromGradientAccumulator(
+    trainer = DistributedTrainer(config_file)
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    result = OptimizerFromGradientAccumulator(
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    optimizer = optimizer_builder(named_param_groups)
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 5 has a total capacty of 79.33 GiB of which 13.02 GiB is free. Including non-PyTorch memory, this process has 66.30 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 3 has a total capacty of 79.33 GiB of which 12.84 GiB is free. Including non-PyTorch memory, this process has 66.48 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    optimizer = optimizer_builder(named_param_groups)
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    result = OptimizerFromGradientAccumulator(
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    result = OptimizerFromGradientAccumulator(
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 1 has a total capacty of 79.33 GiB of which 12.72 GiB is free. Including non-PyTorch memory, this process has 66.60 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 6 has a total capacty of 79.33 GiB of which 12.88 GiB is free. Including non-PyTorch memory, this process has 66.44 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 7 has a total capacty of 79.33 GiB of which 13.05 GiB is free. Including non-PyTorch memory, this process has 66.26 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 4 has a total capacty of 79.33 GiB of which 12.82 GiB is free. Including non-PyTorch memory, this process has 66.50 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 6 has a total capacty of 79.33 GiB of which 12.92 GiB is free. Including non-PyTorch memory, this process has 66.40 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    result = OptimizerFromGradientAccumulator(
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 5 has a total capacty of 79.33 GiB of which 12.80 GiB is free. Including non-PyTorch memory, this process has 66.52 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 2 has a total capacty of 79.33 GiB of which 12.84 GiB is free. Including non-PyTorch memory, this process has 66.48 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+Traceback (most recent call last):
+Traceback (most recent call last):
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+Traceback (most recent call last):
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+Traceback (most recent call last):
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 1 has a total capacty of 79.33 GiB of which 13.02 GiB is free. Including non-PyTorch memory, this process has 66.30 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    trainer = DistributedTrainer(config_file)
+    trainer = DistributedTrainer(config_file)
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    trainer = DistributedTrainer(config_file)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+        trainer = DistributedTrainer(config_file)trainer = DistributedTrainer(config_file)
+
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+trainer = DistributedTrainer(config_file)    
+trainer = DistributedTrainer(config_file)  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+        self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    optimizer = optimizer_builder(named_param_groups)
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+            self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    optimizer = optimizer_builder(named_param_groups)
+self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    result = OptimizerFromGradientAccumulator(
+    trainer = DistributedTrainer(config_file)
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    trainer = DistributedTrainer(config_file)
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    optimizer = optimizer_builder(named_param_groups)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    optimizer = optimizer_builder(named_param_groups)
+    optimizer = optimizer_builder(named_param_groups)    
+optimizer = optimizer_builder(named_param_groups)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+optimizer = optimizer_builder(named_param_groups)  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    
+optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+        result = OptimizerFromGradientAccumulator(result = OptimizerFromGradientAccumulator(    
+
+    result = OptimizerFromGradientAccumulator(
+result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+          File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+optimizer = optimizer_builder(named_param_groups)result = OptimizerFromGradientAccumulator(
+
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    result = OptimizerFromGradientAccumulator(
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 2 has a total capacty of 79.33 GiB of which 13.09 GiB is free. Including non-PyTorch memory, this process has 66.22 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 1 has a total capacty of 79.33 GiB of which 12.84 GiB is free. Including non-PyTorch memory, this process has 66.48 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 1 has a total capacty of 79.33 GiB of which 12.76 GiB is free. Including non-PyTorch memory, this process has 66.56 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+gradient_accumulator = gradient_accumulator_builder(name_to_param.items())    
+gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    result = OptimizerFromGradientAccumulator(
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 4 has a total capacty of 79.33 GiB of which 12.82 GiB is free. Including non-PyTorch memory, this process has 66.50 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 3 has a total capacty of 79.33 GiB of which 12.55 GiB is free. Including non-PyTorch memory, this process has 66.77 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    trainer = DistributedTrainer(config_file)
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 3 has a total capacty of 79.33 GiB of which 12.59 GiB is free. Including non-PyTorch memory, this process has 66.73 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    optimizer = optimizer_builder(named_param_groups)
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    torch.cudaself.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(.
+OutOfMemoryError:   File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+CUDA out of memory. Tried to allocate 54.66 GiB. GPU 6 has a total capacty of 79.33 GiB of which 12.84 GiB is free. Including non-PyTorch memory, this process has 66.48 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    result = OptimizerFromGradientAccumulator(
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 2 has a total capacty of 79.33 GiB of which 12.88 GiB is free. Including non-PyTorch memory, this process has 66.44 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+
+          File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+torch.cuda    .self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(OutOfMemoryError
+: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 4 has a total capacty of 79.33 GiB of which 12.82 GiB is free. Including non-PyTorch memory, this process has 66.50 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+            contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")torch.cudacontiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+.
+
+OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 5 has a total capacty of 79.33 GiB of which 12.72 GiB is free. Including non-PyTorch memory, this process has 66.60 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    torch.cudacontiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")torch.cuda.
+torch.cuda.OutOfMemoryError.OutOfMemoryError: : CUDA out of memory. Tried to allocate 54.66 GiB. GPU 2 has a total capacty of 79.33 GiB of which 12.84 GiB is free. Including non-PyTorch memory, this process has 66.48 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONFCUDA out of memory. Tried to allocate 54.66 GiB. GPU 3 has a total capacty of 79.33 GiB of which 12.55 GiB is free. Including non-PyTorch memory, this process has 66.77 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONFOutOfMemoryError
+
+: torch.cudaCUDA out of memory. Tried to allocate 54.66 GiB. GPU 0 has a total capacty of 79.33 GiB of which 12.86 GiB is free. Including non-PyTorch memory, this process has 66.46 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF.
+OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 7 has a total capacty of 79.33 GiB of which 12.90 GiB is free. Including non-PyTorch memory, this process has 66.42 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 1 has a total capacty of 79.33 GiB of which 12.72 GiB is free. Including non-PyTorch memory, this process has 66.60 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+Traceback (most recent call last):
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 0 has a total capacty of 79.33 GiB of which 12.86 GiB is free. Including non-PyTorch memory, this process has 66.46 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+        self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(optimizer = optimizer_builder(named_param_groups)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    result = OptimizerFromGradientAccumulator(
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    trainer = DistributedTrainer(config_file)
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 4 has a total capacty of 79.33 GiB of which 12.89 GiB is free. Including non-PyTorch memory, this process has 66.42 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 2 has a total capacty of 79.33 GiB of which 12.92 GiB is free. Including non-PyTorch memory, this process has 66.40 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 3 has a total capacty of 79.33 GiB of which 12.62 GiB is free. Including non-PyTorch memory, this process has 66.69 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 5 has a total capacty of 79.33 GiB of which 12.84 GiB is free. Including non-PyTorch memory, this process has 66.48 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 0 has a total capacty of 79.33 GiB of which 12.94 GiB is free. Including non-PyTorch memory, this process has 66.38 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 2 has a total capacty of 79.33 GiB of which 12.72 GiB is free. Including non-PyTorch memory, this process has 66.60 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())    
+trainer = DistributedTrainer(config_file)  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+        self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 3 has a total capacty of 79.33 GiB of which 12.66 GiB is free. Including non-PyTorch memory, this process has 66.65 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 7 has a total capacty of 79.33 GiB of which 13.02 GiB is free. Including non-PyTorch memory, this process has 66.30 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 4 has a total capacty of 79.33 GiB of which 12.70 GiB is free. Including non-PyTorch memory, this process has 66.62 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 1 has a total capacty of 79.33 GiB of which 13.09 GiB is free. Including non-PyTorch memory, this process has 66.22 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 5 has a total capacty of 79.33 GiB of which 13.09 GiB is free. Including non-PyTorch memory, this process has 66.22 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 0 has a total capacty of 79.33 GiB of which 12.79 GiB is free. Including non-PyTorch memory, this process has 66.53 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    result = OptimizerFromGradientAccumulator(
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    result = OptimizerFromGradientAccumulator(
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    optimizer = optimizer_builder(named_param_groups)
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 4 has a total capacty of 79.33 GiB of which 12.74 GiB is free. Including non-PyTorch memory, this process has 66.58 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 0 has a total capacty of 79.33 GiB of which 13.04 GiB is free. Including non-PyTorch memory, this process has 66.28 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 7 has a total capacty of 79.33 GiB of which 13.27 GiB is free. Including non-PyTorch memory, this process has 66.05 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 2 has a total capacty of 79.33 GiB of which 13.02 GiB is free. Including non-PyTorch memory, this process has 66.30 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 3 has a total capacty of 79.33 GiB of which 12.70 GiB is free. Including non-PyTorch memory, this process has 66.62 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 4 has a total capacty of 79.33 GiB of which 12.99 GiB is free. Including non-PyTorch memory, this process has 66.33 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 6 has a total capacty of 79.33 GiB of which 13.02 GiB is free. Including non-PyTorch memory, this process has 66.30 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 3 has a total capacty of 79.33 GiB of which 12.92 GiB is free. Including non-PyTorch memory, this process has 66.40 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 5 has a total capacty of 79.33 GiB of which 13.09 GiB is free. Including non-PyTorch memory, this process has 66.22 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+        self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(optimizer = optimizer_builder(named_param_groups)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 5 has a total capacty of 79.33 GiB of which 13.02 GiB is free. Including non-PyTorch memory, this process has 66.30 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 7 has a total capacty of 79.33 GiB of which 13.19 GiB is free. Including non-PyTorch memory, this process has 66.13 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 1 has a total capacty of 79.33 GiB of which 13.09 GiB is free. Including non-PyTorch memory, this process has 66.22 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+        trainer = DistributedTrainer(config_file)optimizer = optimizer_builder(named_param_groups)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    result = OptimizerFromGradientAccumulator(
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 6 has a total capacty of 79.33 GiB of which 13.09 GiB is free. Including non-PyTorch memory, this process has 66.22 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+        gradient_accumulator = gradient_accumulator_builder(name_to_param.items())gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+        self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 4 has a total capacty of 79.33 GiB of which 13.07 GiB is free. Including non-PyTorch memory, this process has 66.25 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 7 has a total capacty of 79.33 GiB of which 13.27 GiB is free. Including non-PyTorch memory, this process has 66.05 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 3 has a total capacty of 79.33 GiB of which 12.84 GiB is free. Including non-PyTorch memory, this process has 66.48 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 0 has a total capacty of 79.33 GiB of which 13.12 GiB is free. Including non-PyTorch memory, this process has 66.20 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 5 has a total capacty of 79.33 GiB of which 12.72 GiB is free. Including non-PyTorch memory, this process has 66.60 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    result = OptimizerFromGradientAccumulator(
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    result = OptimizerFromGradientAccumulator(
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 0 has a total capacty of 79.33 GiB of which 13.04 GiB is free. Including non-PyTorch memory, this process has 66.28 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+Traceback (most recent call last):
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 3 has a total capacty of 79.33 GiB of which 12.92 GiB is free. Including non-PyTorch memory, this process has 66.40 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 3 has a total capacty of 79.33 GiB of which 12.55 GiB is free. Including non-PyTorch memory, this process has 66.77 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    trainer = DistributedTrainer(config_file)
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 6 has a total capacty of 79.33 GiB of which 12.84 GiB is free. Including non-PyTorch memory, this process has 66.48 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    optimizer = optimizer_builder(named_param_groups)
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 7 has a total capacty of 79.33 GiB of which 12.90 GiB is free. Including non-PyTorch memory, this process has 66.42 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 2 has a total capacty of 79.33 GiB of which 13.09 GiB is free. Including non-PyTorch memory, this process has 66.22 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    trainer = DistributedTrainer(config_file)
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    result = OptimizerFromGradientAccumulator(
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    trainer = DistributedTrainer(config_file)
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    optimizer = optimizer_builder(named_param_groups)
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 4 has a total capacty of 79.33 GiB of which 12.82 GiB is free. Including non-PyTorch memory, this process has 66.50 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    result = OptimizerFromGradientAccumulator(
+    result = OptimizerFromGradientAccumulator(
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    optimizer = optimizer_builder(named_param_groups)
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 1 has a total capacty of 79.33 GiB of which 13.02 GiB is free. Including non-PyTorch memory, this process has 66.30 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    result = OptimizerFromGradientAccumulator(
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 5 has a total capacty of 79.33 GiB of which 13.02 GiB is free. Including non-PyTorch memory, this process has 66.30 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 6 has a total capacty of 79.33 GiB of which 13.02 GiB is free. Including non-PyTorch memory, this process has 66.30 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 2 has a total capacty of 79.33 GiB of which 13.02 GiB is free. Including non-PyTorch memory, this process has 66.30 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 2 has a total capacty of 79.33 GiB of which 12.84 GiB is free. Including non-PyTorch memory, this process has 66.48 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    result = OptimizerFromGradientAccumulator(
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+Traceback (most recent call last):
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 1 has a total capacty of 79.33 GiB of which 12.84 GiB is free. Including non-PyTorch memory, this process has 66.48 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 7 has a total capacty of 79.33 GiB of which 13.27 GiB is free. Including non-PyTorch memory, this process has 66.05 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 4 has a total capacty of 79.33 GiB of which 12.99 GiB is free. Including non-PyTorch memory, this process has 66.33 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    result = OptimizerFromGradientAccumulator(
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 6 has a total capacty of 79.33 GiB of which 13.09 GiB is free. Including non-PyTorch memory, this process has 66.22 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 0 has a total capacty of 79.33 GiB of which 13.12 GiB is free. Including non-PyTorch memory, this process has 66.20 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 1 has a total capacty of 79.33 GiB of which 12.72 GiB is free. Including non-PyTorch memory, this process has 66.60 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 0 has a total capacty of 79.33 GiB of which 12.86 GiB is free. Including non-PyTorch memory, this process has 66.46 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 2 has a total capacty of 79.33 GiB of which 12.84 GiB is free. Including non-PyTorch memory, this process has 66.48 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+        result = OptimizerFromGradientAccumulator(
+trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+        gradient_accumulator = gradient_accumulator_builder(name_to_param.items())self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    trainer = DistributedTrainer(config_file)
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 7 has a total capacty of 79.33 GiB of which 13.19 GiB is free. Including non-PyTorch memory, this process has 66.13 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    trainer = DistributedTrainer(config_file)
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 1 has a total capacty of 79.33 GiB of which 13.02 GiB is free. Including non-PyTorch memory, this process has 66.30 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+Traceback (most recent call last):
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 2 has a total capacty of 79.33 GiB of which 12.72 GiB is free. Including non-PyTorch memory, this process has 66.60 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+Traceback (most recent call last):
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 0 has a total capacty of 79.33 GiB of which 12.75 GiB is free. Including non-PyTorch memory, this process has 66.57 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    result = OptimizerFromGradientAccumulator(
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+        gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    result = OptimizerFromGradientAccumulator(
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 0 has a total capacty of 79.33 GiB of which 12.75 GiB is free. Including non-PyTorch memory, this process has 66.57 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 7 has a total capacty of 79.33 GiB of which 13.02 GiB is free. Including non-PyTorch memory, this process has 66.30 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    optimizer = optimizer_builder(named_param_groups)
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 7 has a total capacty of 79.33 GiB of which 12.90 GiB is free. Including non-PyTorch memory, this process has 66.42 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    trainer = DistributedTrainer(config_file)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 4 has a total capacty of 79.33 GiB of which 12.82 GiB is free. Including non-PyTorch memory, this process has 66.50 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    trainer = DistributedTrainer(config_file)
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 6 has a total capacty of 79.33 GiB of which 12.84 GiB is free. Including non-PyTorch memory, this process has 66.48 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 2 has a total capacty of 79.33 GiB of which 13.09 GiB is free. Including non-PyTorch memory, this process has 66.22 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+Traceback (most recent call last):
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 1 has a total capacty of 79.33 GiB of which 12.84 GiB is free. Including non-PyTorch memory, this process has 66.48 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+Traceback (most recent call last):
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 6 has a total capacty of 79.33 GiB of which 12.72 GiB is free. Including non-PyTorch memory, this process has 66.60 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    trainer = DistributedTrainer(config_file)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 5 has a total capacty of 79.33 GiB of which 12.84 GiB is free. Including non-PyTorch memory, this process has 66.48 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    trainer = DistributedTrainer(config_file)
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 7 has a total capacty of 79.33 GiB of which 13.02 GiB is free. Including non-PyTorch memory, this process has 66.30 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 3 has a total capacty of 79.33 GiB of which 12.84 GiB is free. Including non-PyTorch memory, this process has 66.48 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 0 has a total capacty of 79.33 GiB of which 12.86 GiB is free. Including non-PyTorch memory, this process has 66.46 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 5 has a total capacty of 79.33 GiB of which 12.72 GiB is free. Including non-PyTorch memory, this process has 66.60 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    trainer = DistributedTrainer(config_file)
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 4 has a total capacty of 79.33 GiB of which 12.70 GiB is free. Including non-PyTorch memory, this process has 66.62 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    result = OptimizerFromGradientAccumulator(
+    trainer = DistributedTrainer(config_file)
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    trainer = DistributedTrainer(config_file)
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    optimizer = optimizer_builder(named_param_groups)
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    result = OptimizerFromGradientAccumulator(
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 4 has a total capacty of 79.33 GiB of which 13.07 GiB is free. Including non-PyTorch memory, this process has 66.25 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 1 has a total capacty of 79.33 GiB of which 12.72 GiB is free. Including non-PyTorch memory, this process has 66.60 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    trainer = DistributedTrainer(config_file)
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 3 has a total capacty of 79.33 GiB of which 12.66 GiB is free. Including non-PyTorch memory, this process has 66.65 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    optimizer = optimizer_builder(named_param_groups)
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    optimizer = optimizer_builder(named_param_groups)
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    result = OptimizerFromGradientAccumulator(
+    result = OptimizerFromGradientAccumulator(
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    trainer = DistributedTrainer(config_file)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 1 has a total capacty of 79.33 GiB of which 13.13 GiB is free. Including non-PyTorch memory, this process has 66.19 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+Traceback (most recent call last):
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 3 has a total capacty of 79.33 GiB of which 12.69 GiB is free. Including non-PyTorch memory, this process has 66.63 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 113.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    result = OptimizerFromGradientAccumulator(
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    result = OptimizerFromGradientAccumulator(
+    trainer = DistributedTrainer(config_file)    optimizer = optimizer_builder(named_param_groups)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    result = OptimizerFromGradientAccumulator(
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 3 has a total capacty of 79.33 GiB of which 12.55 GiB is free. Including non-PyTorch memory, this process has 66.77 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 5 has a total capacty of 79.33 GiB of which 13.09 GiB is free. Including non-PyTorch memory, this process has 66.22 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+        gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    result = OptimizerFromGradientAccumulator(
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 4 has a total capacty of 79.33 GiB of which 13.03 GiB is free. Including non-PyTorch memory, this process has 66.29 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    trainer = DistributedTrainer(config_file)
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")    
+contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+Traceback (most recent call last):
+torch.cuda  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 7 has a total capacty of 79.33 GiB of which 13.31 GiB is free. Including non-PyTorch memory, this process has 66.01 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+torch.cuda.OutOfMemoryErrortorch.cuda: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 3 has a total capacty of 79.33 GiB of which 12.66 GiB is free. Including non-PyTorch memory, this process has 66.65 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF.
+OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 6 has a total capacty of 79.33 GiB of which 12.72 GiB is free. Including non-PyTorch memory, this process has 66.60 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+        gradient_accumulator = gradient_accumulator_builder(name_to_param.items())contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 5 has a total capacty of 79.33 GiB of which 12.84 GiB is free. Including non-PyTorch memory, this process has 66.48 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    trainer = DistributedTrainer(config_file)
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    result = OptimizerFromGradientAccumulator(
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    result = OptimizerFromGradientAccumulator(
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 2 has a total capacty of 79.33 GiB of which 12.72 GiB is free. Including non-PyTorch memory, this process has 66.60 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 3 has a total capacty of 79.33 GiB of which 12.96 GiB is free. Including non-PyTorch memory, this process has 66.36 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 2 has a total capacty of 79.33 GiB of which 13.05 GiB is free. Including non-PyTorch memory, this process has 66.26 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 1 has a total capacty of 79.33 GiB of which 13.09 GiB is free. Including non-PyTorch memory, this process has 66.22 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 4 has a total capacty of 79.33 GiB of which 12.70 GiB is free. Including non-PyTorch memory, this process has 66.62 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    trainer = DistributedTrainer(config_file)
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    result = OptimizerFromGradientAccumulator(
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 7 has a total capacty of 79.33 GiB of which 13.24 GiB is free. Including non-PyTorch memory, this process has 66.08 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 93.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    optimizer = optimizer_builder(named_param_groups)
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    optimizer = optimizer_builder(named_param_groups)
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 6 has a total capacty of 79.33 GiB of which 13.05 GiB is free. Including non-PyTorch memory, this process has 66.26 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 0 has a total capacty of 79.33 GiB of which 13.16 GiB is free. Including non-PyTorch memory, this process has 66.15 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 93.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    result = OptimizerFromGradientAccumulator(
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 4 has a total capacty of 79.33 GiB of which 13.12 GiB is free. Including non-PyTorch memory, this process has 66.20 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 93.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 0 has a total capacty of 79.33 GiB of which 13.08 GiB is free. Including non-PyTorch memory, this process has 66.24 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 6 has a total capacty of 79.33 GiB of which 13.14 GiB is free. Including non-PyTorch memory, this process has 66.18 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 93.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 6 has a total capacty of 79.33 GiB of which 12.87 GiB is free. Including non-PyTorch memory, this process has 66.45 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 113.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    result = OptimizerFromGradientAccumulator(
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 6 has a total capacty of 79.33 GiB of which 12.75 GiB is free. Including non-PyTorch memory, this process has 66.57 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 113.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    trainer = DistributedTrainer(config_file)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 5 has a total capacty of 79.33 GiB of which 13.13 GiB is free. Including non-PyTorch memory, this process has 66.19 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 4 has a total capacty of 79.33 GiB of which 12.99 GiB is free. Including non-PyTorch memory, this process has 66.33 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    result = OptimizerFromGradientAccumulator(
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 3 has a total capacty of 79.33 GiB of which 12.89 GiB is free. Including non-PyTorch memory, this process has 66.43 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 93.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+Traceback (most recent call last):
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 7 has a total capacty of 79.33 GiB of which 13.04 GiB is free. Including non-PyTorch memory, this process has 66.28 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 113.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 1 has a total capacty of 79.33 GiB of which 12.87 GiB is free. Including non-PyTorch memory, this process has 66.45 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 113.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 2 has a total capacty of 79.33 GiB of which 13.14 GiB is free. Including non-PyTorch memory, this process has 66.18 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 93.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+        gradient_accumulator = gradient_accumulator_builder(name_to_param.items())self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 2 has a total capacty of 79.33 GiB of which 12.75 GiB is free. Including non-PyTorch memory, this process has 66.57 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 113.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 3 has a total capacty of 79.33 GiB of which 12.92 GiB is free. Including non-PyTorch memory, this process has 66.40 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 0 has a total capacty of 79.33 GiB of which 13.04 GiB is free. Including non-PyTorch memory, this process has 66.28 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    optimizer = optimizer_builder(named_param_groups)
+Traceback (most recent call last):
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 1 has a total capacty of 79.33 GiB of which 13.06 GiB is free. Including non-PyTorch memory, this process has 66.26 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 93.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    trainer = DistributedTrainer(config_file)
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 5 has a total capacty of 79.33 GiB of which 12.75 GiB is free. Including non-PyTorch memory, this process has 66.57 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 113.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 5 has a total capacty of 79.33 GiB of which 12.87 GiB is free. Including non-PyTorch memory, this process has 66.45 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 113.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    optimizer = optimizer_builder(named_param_groups)
+        result = OptimizerFromGradientAccumulator(self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 0 has a total capacty of 79.33 GiB of which 12.77 GiB is free. Including non-PyTorch memory, this process has 66.54 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 113.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 4 has a total capacty of 79.33 GiB of which 12.73 GiB is free. Including non-PyTorch memory, this process has 66.59 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 113.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 1 has a total capacty of 79.33 GiB of which 12.75 GiB is free. Including non-PyTorch memory, this process has 66.57 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 113.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 5 has a total capacty of 79.33 GiB of which 13.06 GiB is free. Including non-PyTorch memory, this process has 66.26 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 93.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+Traceback (most recent call last):
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    trainer = DistributedTrainer(config_file)
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    optimizer = optimizer_builder(named_param_groups)
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    result = OptimizerFromGradientAccumulator(
+torch.cuda  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 4 has a total capacty of 79.33 GiB of which 12.84 GiB is free. Including non-PyTorch memory, this process has 66.47 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 113.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 3 has a total capacty of 79.33 GiB of which 12.57 GiB is free. Including non-PyTorch memory, this process has 66.74 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 113.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 2 has a total capacty of 79.33 GiB of which 13.02 GiB is free. Including non-PyTorch memory, this process has 66.30 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 6 has a total capacty of 79.33 GiB of which 13.02 GiB is free. Including non-PyTorch memory, this process has 66.30 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 0 has a total capacty of 79.33 GiB of which 12.89 GiB is free. Including non-PyTorch memory, this process has 66.43 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 113.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 7 has a total capacty of 79.33 GiB of which 12.93 GiB is free. Including non-PyTorch memory, this process has 66.39 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 113.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 2 has a total capacty of 79.33 GiB of which 12.87 GiB is free. Including non-PyTorch memory, this process has 66.45 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 113.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 2 has a total capacty of 79.33 GiB of which 12.88 GiB is free. Including non-PyTorch memory, this process has 66.44 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    result = OptimizerFromGradientAccumulator(
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 7 has a total capacty of 79.33 GiB of which 13.31 GiB is free. Including non-PyTorch memory, this process has 66.01 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    trainer = DistributedTrainer(config_file)
+    result = OptimizerFromGradientAccumulator(
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 4 has a total capacty of 79.33 GiB of which 13.12 GiB is free. Including non-PyTorch memory, this process has 66.20 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 93.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+        gradient_accumulator = gradient_accumulator_builder(name_to_param.items())self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+        contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 1 has a total capacty of 79.33 GiB of which 12.76 GiB is free. Including non-PyTorch memory, this process has 66.56 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    optimizer = optimizer_builder(named_param_groups)
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 6 has a total capacty of 79.33 GiB of which 12.88 GiB is free. Including non-PyTorch memory, this process has 66.44 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 7 has a total capacty of 79.33 GiB of which 12.94 GiB is free. Including non-PyTorch memory, this process has 66.38 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 6 has a total capacty of 79.33 GiB of which 13.05 GiB is free. Including non-PyTorch memory, this process has 66.26 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    trainer = DistributedTrainer(config_file)
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 2 has a total capacty of 79.33 GiB of which 13.14 GiB is free. Including non-PyTorch memory, this process has 66.18 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 93.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 3 has a total capacty of 79.33 GiB of which 12.59 GiB is free. Including non-PyTorch memory, this process has 66.73 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    trainer = DistributedTrainer(config_file)
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+Traceback (most recent call last):
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 4 has a total capacty of 79.33 GiB of which 12.86 GiB is free. Including non-PyTorch memory, this process has 66.46 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    trainer = DistributedTrainer(config_file)
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    result = OptimizerFromGradientAccumulator(
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    optimizer = optimizer_builder(named_param_groups)
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 5 has a total capacty of 79.33 GiB of which 12.76 GiB is free. Including non-PyTorch memory, this process has 66.56 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 1 has a total capacty of 79.33 GiB of which 13.06 GiB is free. Including non-PyTorch memory, this process has 66.26 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 93.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 3 has a total capacty of 79.33 GiB of which 12.89 GiB is free. Including non-PyTorch memory, this process has 66.43 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 93.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 3 has a total capacty of 79.33 GiB of which 12.96 GiB is free. Including non-PyTorch memory, this process has 66.36 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 5 has a total capacty of 79.33 GiB of which 13.13 GiB is free. Including non-PyTorch memory, this process has 66.19 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 7 has a total capacty of 79.33 GiB of which 13.24 GiB is free. Including non-PyTorch memory, this process has 66.08 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 93.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 5 has a total capacty of 79.33 GiB of which 13.06 GiB is free. Including non-PyTorch memory, this process has 66.26 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 93.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    trainer = DistributedTrainer(config_file)
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 2 has a total capacty of 79.33 GiB of which 12.76 GiB is free. Including non-PyTorch memory, this process has 66.56 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 1 has a total capacty of 79.33 GiB of which 12.88 GiB is free. Including non-PyTorch memory, this process has 66.44 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    result = OptimizerFromGradientAccumulator(
+Traceback (most recent call last):
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 0 has a total capacty of 79.33 GiB of which 13.08 GiB is free. Including non-PyTorch memory, this process has 66.24 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 6 has a total capacty of 79.33 GiB of which 12.76 GiB is free. Including non-PyTorch memory, this process has 66.56 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    result = OptimizerFromGradientAccumulator(
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 0 has a total capacty of 79.33 GiB of which 13.16 GiB is free. Including non-PyTorch memory, this process has 66.15 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 93.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 2 has a total capacty of 79.33 GiB of which 13.05 GiB is free. Including non-PyTorch memory, this process has 66.26 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 1 has a total capacty of 79.33 GiB of which 13.13 GiB is free. Including non-PyTorch memory, this process has 66.19 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 5 has a total capacty of 79.33 GiB of which 12.88 GiB is free. Including non-PyTorch memory, this process has 66.44 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 6 has a total capacty of 79.33 GiB of which 13.14 GiB is free. Including non-PyTorch memory, this process has 66.18 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 93.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 4 has a total capacty of 79.33 GiB of which 13.03 GiB is free. Including non-PyTorch memory, this process has 66.29 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 7 has a total capacty of 79.33 GiB of which 13.05 GiB is free. Including non-PyTorch memory, this process has 66.26 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 0 has a total capacty of 79.33 GiB of which 12.90 GiB is free. Including non-PyTorch memory, this process has 66.42 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    result = OptimizerFromGradientAccumulator(
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 3 has a total capacty of 79.33 GiB of which 12.70 GiB is free. Including non-PyTorch memory, this process has 66.62 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 0 has a total capacty of 79.33 GiB of which 12.79 GiB is free. Including non-PyTorch memory, this process has 66.53 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 78, in __init__
+    self.fp32_grad_buffers, self._contiguous_fp32_grad_buffer = self.build_grad_buffers(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 173, in build_grad_buffers
+    contiguous_buffer_f32_gradients = torch.zeros(needed_buffer_size, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 54.66 GiB. GPU 4 has a total capacty of 79.33 GiB of which 12.74 GiB is free. Including non-PyTorch memory, this process has 66.58 GiB memory in use. Of the allocated memory 54.66 GiB is allocated by PyTorch, and 101.96 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+[2025-01-02 19:10:43,563] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 73437 closing signal SIGTERM
+[2025-01-02 19:10:43,563] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 73439 closing signal SIGTERM
+[2025-01-02 19:10:43,563] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 73440 closing signal SIGTERM
+[2025-01-02 19:10:43,563] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 73441 closing signal SIGTERM
+[2025-01-02 19:10:43,564] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 73442 closing signal SIGTERM
+[2025-01-02 19:10:43,564] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 73443 closing signal SIGTERM
+[2025-01-02 19:10:43,564] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 73444 closing signal SIGTERM
+[2025-01-02 19:10:47,461] torch.distributed.elastic.multiprocessing.api: [ERROR] failed (exitcode: 1) local_rank: 1 (pid: 73438) of binary: /fsx/nouamane/miniconda/envs/2-1-cu121/bin/python
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 264, in launch_agent
+    raise ChildFailedError(
+torch.distributed.elastic.multiprocessing.errors.ChildFailedError: 
+============================================================
+/fsx/nouamane/projects/nanotron/run_train.py FAILED
+------------------------------------------------------------
+Failures:
+  <NO_OTHER_FAILURES>
+------------------------------------------------------------
+Root Cause (first observed failure):
+[0]:
+  time      : 2025-01-02_19:10:43
+  host      : ip-26-0-160-100.ec2.internal
+  rank      : 1 (local_rank: 1)
+  exitcode  : 1 (pid: 73438)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+============================================================
+srun: error: ip-26-0-160-100: task 0: Exited with exit code 1
+srun: launch/slurm: _step_signal: Terminating StepId=14019182.0
+[2025-01-02 19:10:47,764] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-02 19:10:47,765] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 53795 closing signal SIGTERM
+[2025-01-02 19:10:47,765] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 53796 closing signal SIGTERM
+[2025-01-02 19:10:47,765] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 53797 closing signal SIGTERM
+[2025-01-02 19:10:47,765] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 53798 closing signal SIGTERM
+[2025-01-02 19:10:47,766] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-02 19:10:47,765] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 53799 closing signal SIGTERM
+[2025-01-02 19:10:47,763] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-02 19:10:47,765] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 53800 closing signal SIGTERM
+[2025-01-02 19:10:47,765] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 53801 closing signal SIGTERM
+[2025-01-02 19:10:47,766] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 410580 closing signal SIGTERM
+[2025-01-02 19:10:47,766] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-02 19:10:47,765] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 53802 closing signal SIGTERM
+[2025-01-02 19:10:47,766] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-02 19:10:47,763] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 112579 closing signal SIGTERM
+[2025-01-02 19:10:47,766] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 410581 closing signal SIGTERM
+[2025-01-02 19:10:47,764] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 112580 closing signal SIGTERM
+[2025-01-02 19:10:47,766] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-02 19:10:47,766] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-02 19:10:47,766] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 410582 closing signal SIGTERM
+[2025-01-02 19:10:47,766] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 281142 closing signal SIGTERM
+[2025-01-02 19:10:47,764] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 112581 closing signal SIGTERM
+[2025-01-02 19:10:47,766] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 410583 closing signal SIGTERM
+[2025-01-02 19:10:47,767] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 273114 closing signal SIGTERM
+[2025-01-02 19:10:47,766] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-02 19:10:47,764] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 112582 closing signal SIGTERM
+[2025-01-02 19:10:47,766] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 92004 closing signal SIGTERM
+[2025-01-02 19:10:47,766] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 69405 closing signal SIGTERM
+[2025-01-02 19:10:47,766] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 410584 closing signal SIGTERM
+[2025-01-02 19:10:47,767] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 273115 closing signal SIGTERM
+[2025-01-02 19:10:47,766] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 281143 closing signal SIGTERM
+[2025-01-02 19:10:47,765] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-02 19:10:47,766] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 92005 closing signal SIGTERM
+[2025-01-02 19:10:47,767] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 273116 closing signal SIGTERM
+[2025-01-02 19:10:47,766] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 153492 closing signal SIGTERM
+[2025-01-02 19:10:47,766] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 281144 closing signal SIGTERM
+[2025-01-02 19:10:47,765] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 330200 closing signal SIGTERM
+[2025-01-02 19:10:47,764] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 112583 closing signal SIGTERM
+[2025-01-02 19:10:47,766] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 92006 closing signal SIGTERM
+[2025-01-02 19:10:47,766] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 69406 closing signal SIGTERM
+[2025-01-02 19:10:47,766] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 410585 closing signal SIGTERM
+[2025-01-02 19:10:47,767] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 273117 closing signal SIGTERM
+[2025-01-02 19:10:47,766] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 153493 closing signal SIGTERM
+[2025-01-02 19:10:47,765] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 330201 closing signal SIGTERM
+[2025-01-02 19:10:47,764] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 112584 closing signal SIGTERM
+[2025-01-02 19:10:47,766] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 69407 closing signal SIGTERM
+[2025-01-02 19:10:47,766] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 410586 closing signal SIGTERM
+[2025-01-02 19:10:47,766] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 153494 closing signal SIGTERM
+[2025-01-02 19:10:47,767] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 281145 closing signal SIGTERM
+[2025-01-02 19:10:47,765] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 330202 closing signal SIGTERM
+[2025-01-02 19:10:47,766] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-02 19:10:47,764] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 112585 closing signal SIGTERM
+[2025-01-02 19:10:47,766] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 92008 closing signal SIGTERM
+[2025-01-02 19:10:47,766] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 410587 closing signal SIGTERM
+[2025-01-02 19:10:47,767] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 273118 closing signal SIGTERM
+[2025-01-02 19:10:47,766] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 153495 closing signal SIGTERM
+[2025-01-02 19:10:47,767] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 281146 closing signal SIGTERM
+[2025-01-02 19:10:47,765] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 330203 closing signal SIGTERM
+[2025-01-02 19:10:47,766] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 149150 closing signal SIGTERM
+[2025-01-02 19:10:47,764] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 112586 closing signal SIGTERM
+[2025-01-02 19:10:47,766] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 92009 closing signal SIGTERM
+[2025-01-02 19:10:47,766] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 69408 closing signal SIGTERM
+[2025-01-02 19:10:47,767] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 273119 closing signal SIGTERM
+[2025-01-02 19:10:47,766] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 153497 closing signal SIGTERM
+[2025-01-02 19:10:47,765] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 330204 closing signal SIGTERM
+[2025-01-02 19:10:47,766] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-02 19:10:47,766] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 92010 closing signal SIGTERM
+[2025-01-02 19:10:47,767] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 69410 closing signal SIGTERM
+[2025-01-02 19:10:47,767] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 273120 closing signal SIGTERM
+[2025-01-02 19:10:47,766] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-02 19:10:47,767] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 281147 closing signal SIGTERM
+[2025-01-02 19:10:47,766] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 330205 closing signal SIGTERM
+[2025-01-02 19:10:47,767] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 149151 closing signal SIGTERM
+[2025-01-02 19:10:47,767] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-02 19:10:47,766] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 92011 closing signal SIGTERM
+[2025-01-02 19:10:47,766] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-02 19:10:47,766] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 150328 closing signal SIGTERM
+[2025-01-02 19:10:47,766] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 153498 closing signal SIGTERM
+[2025-01-02 19:10:47,767] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 281148 closing signal SIGTERM
+[2025-01-02 19:10:47,767] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 149152 closing signal SIGTERM
+[2025-01-02 19:10:47,767] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 141711 closing signal SIGTERM
+[2025-01-02 19:10:47,767] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2778789 closing signal SIGTERM
+[2025-01-02 19:10:47,766] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-02 19:10:47,767] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 69411 closing signal SIGTERM
+[2025-01-02 19:10:47,766] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 37999 closing signal SIGTERM
+[2025-01-02 19:10:47,767] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 150329 closing signal SIGTERM
+[2025-01-02 19:10:47,766] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 153499 closing signal SIGTERM
+[2025-01-02 19:10:47,767] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 281149 closing signal SIGTERM
+[2025-01-02 19:10:47,766] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 330206 closing signal SIGTERM
+[2025-01-02 19:10:47,767] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 141712 closing signal SIGTERM
+[2025-01-02 19:10:47,767] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2778790 closing signal SIGTERM
+[2025-01-02 19:10:47,767] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-02 19:10:47,767] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-02 19:10:47,766] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 596519 closing signal SIGTERM
+[2025-01-02 19:10:47,766] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-02 19:10:47,767] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 38000 closing signal SIGTERM
+[2025-01-02 19:10:47,767] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 150330 closing signal SIGTERM
+[2025-01-02 19:10:47,766] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 330207 closing signal SIGTERM
+[2025-01-02 19:10:47,767] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 149153 closing signal SIGTERM
+[2025-01-02 19:10:47,767] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 141714 closing signal SIGTERM
+[2025-01-02 19:10:47,767] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2778791 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 537954 closing signal SIGTERM
+[2025-01-02 19:10:47,767] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 323254 closing signal SIGTERM
+[2025-01-02 19:10:47,766] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 596520 closing signal SIGTERM
+[2025-01-02 19:10:47,767] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-02 19:10:47,766] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 387824 closing signal SIGTERM
+[2025-01-02 19:10:47,767] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 38001 closing signal SIGTERM
+[2025-01-02 19:10:47,767] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 150331 closing signal SIGTERM
+[2025-01-02 19:10:47,767] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 149154 closing signal SIGTERM
+[2025-01-02 19:10:47,767] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2778792 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 537955 closing signal SIGTERM
+[2025-01-02 19:10:47,767] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 323255 closing signal SIGTERM
+[2025-01-02 19:10:47,766] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 596521 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 27510 closing signal SIGTERM
+[2025-01-02 19:10:47,766] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 387825 closing signal SIGTERM
+[2025-01-02 19:10:47,767] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-02 19:10:47,767] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 149155 closing signal SIGTERM
+[2025-01-02 19:10:47,767] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 141715 closing signal SIGTERM
+[2025-01-02 19:10:47,767] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 537956 closing signal SIGTERM
+[2025-01-02 19:10:47,767] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 323256 closing signal SIGTERM
+[2025-01-02 19:10:47,766] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-02 19:10:47,766] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 596522 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 27511 closing signal SIGTERM
+[2025-01-02 19:10:47,766] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 387826 closing signal SIGTERM
+[2025-01-02 19:10:47,767] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 38002 closing signal SIGTERM
+[2025-01-02 19:10:47,767] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 150332 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-02 19:10:47,767] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 149156 closing signal SIGTERM
+[2025-01-02 19:10:47,767] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 141716 closing signal SIGTERM
+[2025-01-02 19:10:47,767] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2778793 closing signal SIGTERM
+[2025-01-02 19:10:47,767] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 34237 closing signal SIGTERM
+[2025-01-02 19:10:47,767] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 134811 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 27512 closing signal SIGTERM
+[2025-01-02 19:10:47,766] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 387827 closing signal SIGTERM
+[2025-01-02 19:10:47,767] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 38003 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3958422 closing signal SIGTERM
+[2025-01-02 19:10:47,767] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 150333 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 91834 closing signal SIGTERM
+[2025-01-02 19:10:47,767] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 141717 closing signal SIGTERM
+[2025-01-02 19:10:47,767] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2778794 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 537957 closing signal SIGTERM
+[2025-01-02 19:10:47,767] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 323257 closing signal SIGTERM
+[2025-01-02 19:10:47,767] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 134812 closing signal SIGTERM
+[2025-01-02 19:10:47,767] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 387828 closing signal SIGTERM
+[2025-01-02 19:10:47,767] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 38004 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3958423 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 91835 closing signal SIGTERM
+[2025-01-02 19:10:47,767] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 149157 closing signal SIGTERM
+[2025-01-02 19:10:47,767] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 141718 closing signal SIGTERM
+[2025-01-02 19:10:47,767] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2778795 closing signal SIGTERM
+[2025-01-02 19:10:47,767] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 34238 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 537958 closing signal SIGTERM
+[2025-01-02 19:10:47,767] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 323258 closing signal SIGTERM
+[2025-01-02 19:10:47,767] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 134813 closing signal SIGTERM
+[2025-01-02 19:10:47,766] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 596524 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 27514 closing signal SIGTERM
+[2025-01-02 19:10:47,767] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 38005 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3958424 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 91836 closing signal SIGTERM
+[2025-01-02 19:10:47,767] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-02 19:10:47,767] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2778796 closing signal SIGTERM
+[2025-01-02 19:10:47,767] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 34239 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 537959 closing signal SIGTERM
+[2025-01-02 19:10:47,767] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 323259 closing signal SIGTERM
+[2025-01-02 19:10:47,767] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 134814 closing signal SIGTERM
+[2025-01-02 19:10:47,766] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 596525 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 27515 closing signal SIGTERM
+[2025-01-02 19:10:47,767] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 387829 closing signal SIGTERM
+[2025-01-02 19:10:47,767] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 38006 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3958425 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 279713 closing signal SIGTERM
+[2025-01-02 19:10:47,767] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 537960 closing signal SIGTERM
+[2025-01-02 19:10:47,767] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 134815 closing signal SIGTERM
+[2025-01-02 19:10:47,765] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-02 19:10:47,766] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 596526 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 27516 closing signal SIGTERM
+[2025-01-02 19:10:47,767] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 387830 closing signal SIGTERM
+[2025-01-02 19:10:47,767] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3958426 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 91837 closing signal SIGTERM
+[2025-01-02 19:10:47,767] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 34241 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 537961 closing signal SIGTERM
+[2025-01-02 19:10:47,767] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 323260 closing signal SIGTERM
+[2025-01-02 19:10:47,767] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 387831 closing signal SIGTERM
+[2025-01-02 19:10:47,767] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 34242 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 287318 closing signal SIGTERM
+[2025-01-02 19:10:47,767] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 323261 closing signal SIGTERM
+[2025-01-02 19:10:47,767] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 134816 closing signal SIGTERM
+[2025-01-02 19:10:47,766] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 150510 closing signal SIGTERM
+[2025-01-02 19:10:47,769] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 160172 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 19327 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3958427 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 91838 closing signal SIGTERM
+[2025-01-02 19:10:47,767] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 279714 closing signal SIGTERM
+[2025-01-02 19:10:47,767] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 134817 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3958428 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 91839 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 287319 closing signal SIGTERM
+[2025-01-02 19:10:47,767] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 134818 closing signal SIGTERM
+[2025-01-02 19:10:47,769] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 160173 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 19328 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3958429 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 91840 closing signal SIGTERM
+[2025-01-02 19:10:47,767] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 481873 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 279715 closing signal SIGTERM
+[2025-01-02 19:10:47,767] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 34243 closing signal SIGTERM
+[2025-01-02 19:10:47,766] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 19329 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 91841 closing signal SIGTERM
+[2025-01-02 19:10:47,767] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 481874 closing signal SIGTERM
+[2025-01-02 19:10:47,767] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-02 19:10:47,766] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 287320 closing signal SIGTERM
+[2025-01-02 19:10:47,766] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 150513 closing signal SIGTERM
+[2025-01-02 19:10:47,769] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 160174 closing signal SIGTERM
+[2025-01-02 19:10:47,767] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 481875 closing signal SIGTERM
+[2025-01-02 19:10:47,767] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 133301 closing signal SIGTERM
+[2025-01-02 19:10:47,767] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 120970 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 279716 closing signal SIGTERM
+[2025-01-02 19:10:47,766] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 150514 closing signal SIGTERM
+[2025-01-02 19:10:47,767] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 236046 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 19330 closing signal SIGTERM
+[2025-01-02 19:10:47,767] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 481876 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 287321 closing signal SIGTERM
+[2025-01-02 19:10:47,767] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 236047 closing signal SIGTERM
+[2025-01-02 19:10:47,767] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 481877 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 133302 closing signal SIGTERM
+[2025-01-02 19:10:47,767] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 120971 closing signal SIGTERM
+[2025-01-02 19:10:47,769] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 160175 closing signal SIGTERM
+[2025-01-02 19:10:47,767] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 236048 closing signal SIGTERM
+[2025-01-02 19:10:47,767] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 133303 closing signal SIGTERM
+[2025-01-02 19:10:47,767] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 120972 closing signal SIGTERM
+[2025-01-02 19:10:47,766] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 150515 closing signal SIGTERM
+[2025-01-02 19:10:47,769] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 160177 closing signal SIGTERM
+[2025-01-02 19:10:47,767] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 236050 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 19331 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 133304 closing signal SIGTERM
+[2025-01-02 19:10:47,767] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 120973 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 287322 closing signal SIGTERM
+[2025-01-02 19:10:47,766] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 150516 closing signal SIGTERM
+[2025-01-02 19:10:47,767] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 236051 closing signal SIGTERM
+[2025-01-02 19:10:47,767] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 317496 closing signal SIGTERM
+[2025-01-02 19:10:47,767] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 19332 closing signal SIGTERM
+[2025-01-02 19:10:47,767] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 481878 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 279718 closing signal SIGTERM
+[2025-01-02 19:10:47,767] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 317497 closing signal SIGTERM
+[2025-01-02 19:10:47,767] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 216030 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 19333 closing signal SIGTERM
+[2025-01-02 19:10:47,767] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 481879 closing signal SIGTERM
+[2025-01-02 19:10:47,767] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 133305 closing signal SIGTERM
+[2025-01-02 19:10:47,767] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 120974 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 72038 closing signal SIGTERM
+[2025-01-02 19:10:47,767] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 317498 closing signal SIGTERM
+[2025-01-02 19:10:47,767] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 481880 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 133306 closing signal SIGTERM
+[2025-01-02 19:10:47,767] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 120975 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 287324 closing signal SIGTERM
+[2025-01-02 19:10:47,767] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 236052 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 72039 closing signal SIGTERM
+[2025-01-02 19:10:47,767] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 317499 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 216031 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 77717 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 133307 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 279719 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-02 19:10:47,767] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 236053 closing signal SIGTERM
+[2025-01-02 19:10:47,767] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 317500 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 216032 closing signal SIGTERM
+[2025-01-02 19:10:47,767] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 77718 closing signal SIGTERM
+[2025-01-02 19:10:47,767] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 120977 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 279720 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 287325 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 72040 closing signal SIGTERM
+[2025-01-02 19:10:47,765] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-02 19:10:47,768] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 77719 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 133308 closing signal SIGTERM
+[2025-01-02 19:10:47,767] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 171566 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 317502 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 318476 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 216033 closing signal SIGTERM
+[2025-01-02 19:10:47,766] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 92235 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 501629 closing signal SIGTERM
+[2025-01-02 19:10:47,767] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 199265 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 77720 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 171567 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 72041 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 317503 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 318477 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 216034 closing signal SIGTERM
+[2025-01-02 19:10:47,766] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 92236 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 77721 closing signal SIGTERM
+[2025-01-02 19:10:47,767] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 568895 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 171568 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 318478 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 216035 closing signal SIGTERM
+[2025-01-02 19:10:47,766] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 92237 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 501630 closing signal SIGTERM
+[2025-01-02 19:10:47,767] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 199266 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 568896 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 171569 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 72042 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 318479 closing signal SIGTERM
+[2025-01-02 19:10:47,766] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 92238 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 501631 closing signal SIGTERM
+[2025-01-02 19:10:47,769] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 77722 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 171570 closing signal SIGTERM
+[2025-01-02 19:10:47,767] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 72043 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 318480 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 216036 closing signal SIGTERM
+[2025-01-02 19:10:47,766] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 92239 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 501632 closing signal SIGTERM
+[2025-01-02 19:10:47,767] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 199267 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 77723 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 568897 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 146935 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 318481 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 216037 closing signal SIGTERM
+[2025-01-02 19:10:47,766] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 92240 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 567627 closing signal SIGTERM
+[2025-01-02 19:10:47,769] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 523949 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 77724 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 171571 closing signal SIGTERM
+[2025-01-02 19:10:47,767] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 54849 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 72044 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 146936 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 28562 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 567628 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 501633 closing signal SIGTERM
+[2025-01-02 19:10:47,767] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 199268 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 568898 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 171572 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-02 19:10:47,769] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 72045 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 28563 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 318482 closing signal SIGTERM
+[2025-01-02 19:10:47,766] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 92241 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 567629 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 501634 closing signal SIGTERM
+[2025-01-02 19:10:47,767] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 199269 closing signal SIGTERM
+[2025-01-02 19:10:47,769] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 523950 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 568899 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 171573 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 31125 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 54850 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 146937 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 28564 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 318483 closing signal SIGTERM
+[2025-01-02 19:10:47,766] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 92242 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 567630 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 501636 closing signal SIGTERM
+[2025-01-02 19:10:47,769] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 523952 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 568900 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 54851 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 146938 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 28565 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 568901 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 31126 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 567631 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 567632 closing signal SIGTERM
+[2025-01-02 19:10:47,769] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 523953 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 31127 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 54852 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 146939 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 28566 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 567633 closing signal SIGTERM
+[2025-01-02 19:10:47,769] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 523954 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 568902 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 54853 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 146940 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 28567 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 24757 closing signal SIGTERM
+[2025-01-02 19:10:47,769] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 523955 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 54854 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 146941 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 28568 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 24758 closing signal SIGTERM
+[2025-01-02 19:10:47,769] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 89880 closing signal SIGTERM
+[2025-01-02 19:10:47,769] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 31128 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 146942 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 28569 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 24759 closing signal SIGTERM
+[2025-01-02 19:10:47,769] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 149776 closing signal SIGTERM
+[2025-01-02 19:10:47,769] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 24760 closing signal SIGTERM
+[2025-01-02 19:10:47,769] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 523956 closing signal SIGTERM
+[2025-01-02 19:10:47,769] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 149777 closing signal SIGTERM
+[2025-01-02 19:10:47,769] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 89881 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 54855 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 24761 closing signal SIGTERM
+[2025-01-02 19:10:47,769] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 149778 closing signal SIGTERM
+[2025-01-02 19:10:47,769] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 557157 closing signal SIGTERM
+[2025-01-02 19:10:47,769] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 31129 closing signal SIGTERM
+[2025-01-02 19:10:47,769] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 149779 closing signal SIGTERM
+[2025-01-02 19:10:47,769] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 557158 closing signal SIGTERM
+[2025-01-02 19:10:47,769] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-02 19:10:47,769] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 31130 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-02 19:10:47,769] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 557159 closing signal SIGTERM
+[2025-01-02 19:10:47,769] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 89883 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 124814 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 24762 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 124815 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 24763 closing signal SIGTERM
+[2025-01-02 19:10:47,769] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 149780 closing signal SIGTERM
+[2025-01-02 19:10:47,769] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 557160 closing signal SIGTERM
+[2025-01-02 19:10:47,769] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 31131 closing signal SIGTERM
+[2025-01-02 19:10:47,769] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 592825 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 24764 closing signal SIGTERM
+[2025-01-02 19:10:47,769] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 149781 closing signal SIGTERM
+[2025-01-02 19:10:47,769] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 557161 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 124816 closing signal SIGTERM
+[2025-01-02 19:10:47,769] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 31132 closing signal SIGTERM
+[2025-01-02 19:10:47,769] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 149782 closing signal SIGTERM
+[2025-01-02 19:10:47,769] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 459050 closing signal SIGTERM
+[2025-01-02 19:10:47,769] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-02 19:10:47,769] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 89884 closing signal SIGTERM
+[2025-01-02 19:10:47,767] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-02 19:10:47,769] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 48936 closing signal SIGTERM
+[2025-01-02 19:10:47,769] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 592826 closing signal SIGTERM
+[2025-01-02 19:10:47,769] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-02 19:10:47,769] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 557162 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 124817 closing signal SIGTERM
+[2025-01-02 19:10:47,769] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 592827 closing signal SIGTERM
+[2025-01-02 19:10:47,769] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 149783 closing signal SIGTERM
+[2025-01-02 19:10:47,769] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 557163 closing signal SIGTERM
+[2025-01-02 19:10:47,769] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1043834 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 124818 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 207968 closing signal SIGTERM
+[2025-01-02 19:10:47,769] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 48937 closing signal SIGTERM
+[2025-01-02 19:10:47,769] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 592828 closing signal SIGTERM
+[2025-01-02 19:10:47,769] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 773714 closing signal SIGTERM
+[2025-01-02 19:10:47,769] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 557164 closing signal SIGTERM
+[2025-01-02 19:10:47,769] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 89885 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 124819 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 207969 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-02 19:10:47,769] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 773715 closing signal SIGTERM
+[2025-01-02 19:10:47,770] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 459051 closing signal SIGTERM
+[2025-01-02 19:10:47,769] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1043835 closing signal SIGTERM
+[2025-01-02 19:10:47,769] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 89887 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 207970 closing signal SIGTERM
+[2025-01-02 19:10:47,769] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 48938 closing signal SIGTERM
+[2025-01-02 19:10:47,769] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 592829 closing signal SIGTERM
+[2025-01-02 19:10:47,769] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 773716 closing signal SIGTERM
+[2025-01-02 19:10:47,770] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 459052 closing signal SIGTERM
+[2025-01-02 19:10:47,769] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1043836 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 207971 closing signal SIGTERM
+[2025-01-02 19:10:47,769] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 592830 closing signal SIGTERM
+[2025-01-02 19:10:47,769] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 773717 closing signal SIGTERM
+[2025-01-02 19:10:47,769] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1043837 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 124820 closing signal SIGTERM
+[2025-01-02 19:10:47,769] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 592831 closing signal SIGTERM
+[2025-01-02 19:10:47,769] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 773718 closing signal SIGTERM
+[2025-01-02 19:10:47,769] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1043838 closing signal SIGTERM
+[2025-01-02 19:10:47,769] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 124821 closing signal SIGTERM
+[2025-01-02 19:10:47,769] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 279767 closing signal SIGTERM
+[2025-01-02 19:10:47,769] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 48939 closing signal SIGTERM
+[2025-01-02 19:10:47,769] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 773720 closing signal SIGTERM
+[2025-01-02 19:10:47,770] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 459053 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 207972 closing signal SIGTERM
+[2025-01-02 19:10:47,769] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 48940 closing signal SIGTERM
+[2025-01-02 19:10:47,769] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 592832 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 207973 closing signal SIGTERM
+[2025-01-02 19:10:47,769] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1043839 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 207974 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-02 19:10:47,769] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 773721 closing signal SIGTERM
+[2025-01-02 19:10:47,770] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-02 19:10:47,770] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 459054 closing signal SIGTERM
+[2025-01-02 19:10:47,769] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1043840 closing signal SIGTERM
+[2025-01-02 19:10:47,769] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 279768 closing signal SIGTERM
+[2025-01-02 19:10:47,769] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1043841 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 207975 closing signal SIGTERM
+[2025-01-02 19:10:47,769] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 279769 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3239640 closing signal SIGTERM
+[2025-01-02 19:10:47,770] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 459055 closing signal SIGTERM
+[2025-01-02 19:10:47,770] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 459056 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3239641 closing signal SIGTERM
+[2025-01-02 19:10:47,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3239642 closing signal SIGTERM
+[2025-01-02 19:10:47,769] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 279770 closing signal SIGTERM
+[2025-01-02 19:10:47,769] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3239643 closing signal SIGTERM
+[2025-01-02 19:10:47,770] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 591675 closing signal SIGTERM
+[2025-01-02 19:10:47,769] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3239644 closing signal SIGTERM
+[2025-01-02 19:10:47,769] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3239645 closing signal SIGTERM
+[2025-01-02 19:10:47,769] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3239646 closing signal SIGTERM
+[2025-01-02 19:10:47,771] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 591676 closing signal SIGTERM
+[2025-01-02 19:10:47,769] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3239647 closing signal SIGTERM
+[2025-01-02 19:10:47,771] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 591677 closing signal SIGTERM
+[2025-01-02 19:10:47,771] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 591678 closing signal SIGTERM
+[2025-01-02 19:10:47,770] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-02 19:10:47,771] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 591679 closing signal SIGTERM
+[2025-01-02 19:10:47,771] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 591681 closing signal SIGTERM
+[2025-01-02 19:10:47,770] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 88578 closing signal SIGTERM
+[2025-01-02 19:10:47,770] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 88579 closing signal SIGTERM
+[2025-01-02 19:10:47,770] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 88580 closing signal SIGTERM
+[2025-01-02 19:10:47,771] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 88581 closing signal SIGTERM
+[2025-01-02 19:10:47,771] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 88582 closing signal SIGTERM
+[2025-01-02 19:10:47,994] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-168-238.ec2.internal_19250_0' has failed to send a keep-alive heartbeat to the rendezvous '14019182' due to an error of type RendezvousConnectionError.
+[2025-01-02 19:10:48,139] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-164-18.ec2.internal_410504_0' has failed to send a keep-alive heartbeat to the rendezvous '14019182' due to an error of type RendezvousConnectionError.
+[2025-01-02 19:10:48,139] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-164-0.ec2.internal_112504_0' has failed to send a keep-alive heartbeat to the rendezvous '14019182' due to an error of type RendezvousConnectionError.
+[2025-01-02 19:10:48,145] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-170-31.ec2.internal_523874_0' has failed to send a keep-alive heartbeat to the rendezvous '14019182' due to an error of type RendezvousConnectionError.
+[2025-01-02 19:10:48,148] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-169-239.ec2.internal_77642_0' has failed to send a keep-alive heartbeat to the rendezvous '14019182' due to an error of type RendezvousConnectionError.
+[2025-01-02 19:10:48,156] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-171-230.ec2.internal_773639_0' has failed to send a keep-alive heartbeat to the rendezvous '14019182' due to an error of type RendezvousConnectionError.
+[2025-01-02 19:10:48,159] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-165-164.ec2.internal_141637_0' has failed to send a keep-alive heartbeat to the rendezvous '14019182' due to an error of type RendezvousConnectionError.
+[2025-01-02 19:10:48,165] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-165-213.ec2.internal_37924_0' has failed to send a keep-alive heartbeat to the rendezvous '14019182' due to an error of type RendezvousConnectionError.
+[2025-01-02 19:10:48,168] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-167-9.ec2.internal_27433_0' has failed to send a keep-alive heartbeat to the rendezvous '14019182' due to an error of type RendezvousConnectionError.
+[2025-01-02 19:10:48,169] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-167-177.ec2.internal_235972_0' has failed to send a keep-alive heartbeat to the rendezvous '14019182' due to an error of type RendezvousConnectionError.
+[2025-01-02 19:10:48,171] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-170-160.ec2.internal_89805_0' has failed to send a keep-alive heartbeat to the rendezvous '14019182' due to an error of type RendezvousConnectionError.
+[2025-01-02 19:10:48,184] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-168-30.ec2.internal_481798_0' has failed to send a keep-alive heartbeat to the rendezvous '14019182' due to an error of type RendezvousConnectionError.
+[2025-01-02 19:10:48,190] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-171-249.ec2.internal_31050_0' has failed to send a keep-alive heartbeat to the rendezvous '14019182' due to an error of type RendezvousConnectionError.
+[2025-01-02 19:10:48,194] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-168-52.ec2.internal_318401_0' has failed to send a keep-alive heartbeat to the rendezvous '14019182' due to an error of type RendezvousConnectionError.
+[2025-01-02 19:10:48,199] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-167-217.ec2.internal_133226_0' has failed to send a keep-alive heartbeat to the rendezvous '14019182' due to an error of type RendezvousConnectionError.
+[2025-01-02 19:10:48,202] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-172-116.ec2.internal_207894_0' has failed to send a keep-alive heartbeat to the rendezvous '14019182' due to an error of type RendezvousConnectionError.
+[2025-01-02 19:10:48,211] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-171-168.ec2.internal_1043759_0' has failed to send a keep-alive heartbeat to the rendezvous '14019182' due to an error of type RendezvousConnectionError.
+[2025-01-02 19:10:48,233] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-168-95.ec2.internal_171491_0' has failed to send a keep-alive heartbeat to the rendezvous '14019182' due to an error of type RendezvousConnectionError.
+[2025-01-02 19:10:48,237] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-166-15.ec2.internal_537880_0' has failed to send a keep-alive heartbeat to the rendezvous '14019182' due to an error of type RendezvousConnectionError.
+[2025-01-02 19:10:48,244] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-165-202.ec2.internal_134736_0' has failed to send a keep-alive heartbeat to the rendezvous '14019182' due to an error of type RendezvousConnectionError.
+[2025-01-02 19:10:48,254] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-167-245.ec2.internal_3958245_0' has failed to send a keep-alive heartbeat to the rendezvous '14019182' due to an error of type RendezvousConnectionError.
+[2025-01-02 19:10:48,264] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-171-102.ec2.internal_146861_0' has failed to send a keep-alive heartbeat to the rendezvous '14019182' due to an error of type RendezvousConnectionError.
+[2025-01-02 19:10:48,277] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-174-36.ec2.internal_279689_0' has failed to send a keep-alive heartbeat to the rendezvous '14019182' due to an error of type RendezvousConnectionError.
+[2025-01-02 19:10:48,279] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-165-131.ec2.internal_596442_0' has failed to send a keep-alive heartbeat to the rendezvous '14019182' due to an error of type RendezvousConnectionError.
+[2025-01-02 19:10:48,282] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-164-187.ec2.internal_153417_0' has failed to send a keep-alive heartbeat to the rendezvous '14019182' due to an error of type RendezvousConnectionError.
+[2025-01-02 19:10:48,285] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-173-202.ec2.internal_591598_0' has failed to send a keep-alive heartbeat to the rendezvous '14019182' due to an error of type RendezvousConnectionError.
+[2025-01-02 19:10:48,288] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-169-86.ec2.internal_120894_0' has failed to send a keep-alive heartbeat to the rendezvous '14019182' due to an error of type RendezvousConnectionError.
+[2025-01-02 19:10:48,292] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-165-59.ec2.internal_149076_0' has failed to send a keep-alive heartbeat to the rendezvous '14019182' due to an error of type RendezvousConnectionError.
+[2025-01-02 19:10:48,294] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-171-56.ec2.internal_215955_0' has failed to send a keep-alive heartbeat to the rendezvous '14019182' due to an error of type RendezvousConnectionError.
+[2025-01-02 19:10:48,304] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-167-175.ec2.internal_160096_0' has failed to send a keep-alive heartbeat to the rendezvous '14019182' due to an error of type RendezvousConnectionError.
+[2025-01-02 19:10:48,306] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-165-24.ec2.internal_273037_0' has failed to send a keep-alive heartbeat to the rendezvous '14019182' due to an error of type RendezvousConnectionError.
+[2025-01-02 19:10:48,308] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-163-236.ec2.internal_91930_0' has failed to send a keep-alive heartbeat to the rendezvous '14019182' due to an error of type RendezvousConnectionError.
+[2025-01-02 19:10:48,311] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-164-207.ec2.internal_69330_0' has failed to send a keep-alive heartbeat to the rendezvous '14019182' due to an error of type RendezvousConnectionError.
+[2025-01-02 19:10:48,322] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-169-132.ec2.internal_88502_0' has failed to send a keep-alive heartbeat to the rendezvous '14019182' due to an error of type RendezvousConnectionError.
+[2025-01-02 19:10:48,324] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-168-120.ec2.internal_71963_0' has failed to send a keep-alive heartbeat to the rendezvous '14019182' due to an error of type RendezvousConnectionError.
+[2025-01-02 19:10:48,326] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-172-142.ec2.internal_24682_0' has failed to send a keep-alive heartbeat to the rendezvous '14019182' due to an error of type RendezvousConnectionError.
+[2025-01-02 19:10:48,330] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-166-214.ec2.internal_279639_0' has failed to send a keep-alive heartbeat to the rendezvous '14019182' due to an error of type RendezvousConnectionError.
+[2025-01-02 19:10:48,331] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-173-7.ec2.internal_557082_0' has failed to send a keep-alive heartbeat to the rendezvous '14019182' due to an error of type RendezvousConnectionError.
+[2025-01-02 19:10:48,334] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-168-34.ec2.internal_287243_0' has failed to send a keep-alive heartbeat to the rendezvous '14019182' due to an error of type RendezvousConnectionError.
+[2025-01-02 19:10:48,339] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-166-244.ec2.internal_91759_0' has failed to send a keep-alive heartbeat to the rendezvous '14019182' due to an error of type RendezvousConnectionError.
+[2025-01-02 19:10:48,344] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-164-75.ec2.internal_330124_0' has failed to send a keep-alive heartbeat to the rendezvous '14019182' due to an error of type RendezvousConnectionError.
+[2025-01-02 19:10:48,351] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-172-57.ec2.internal_149702_0' has failed to send a keep-alive heartbeat to the rendezvous '14019182' due to an error of type RendezvousConnectionError.
+[2025-01-02 19:10:48,353] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-171-62.ec2.internal_199189_0' has failed to send a keep-alive heartbeat to the rendezvous '14019182' due to an error of type RendezvousConnectionError.
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+[2025-01-02 19:10:48,355] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-166-36.ec2.internal_387749_0' has failed to send a keep-alive heartbeat to the rendezvous '14019182' due to an error of type RendezvousConnectionError.
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 279689 got signal: 15
+[2025-01-02 19:10:48,365] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-165-38.ec2.internal_150433_0' has failed to send a keep-alive heartbeat to the rendezvous '14019182' due to an error of type RendezvousConnectionError.
+[2025-01-02 19:10:48,367] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-170-132.ec2.internal_568819_0' has failed to send a keep-alive heartbeat to the rendezvous '14019182' due to an error of type RendezvousConnectionError.
+[2025-01-02 19:10:48,370] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-172-73.ec2.internal_124739_0' has failed to send a keep-alive heartbeat to the rendezvous '14019182' due to an error of type RendezvousConnectionError.
+[2025-01-02 19:10:48,372] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-164-236.ec2.internal_150252_0' has failed to send a keep-alive heartbeat to the rendezvous '14019182' due to an error of type RendezvousConnectionError.
+[2025-01-02 19:10:48,379] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-170-143.ec2.internal_28487_0' has failed to send a keep-alive heartbeat to the rendezvous '14019182' due to an error of type RendezvousConnectionError.
+[2025-01-02 19:10:48,383] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-169-207.ec2.internal_92160_0' has failed to send a keep-alive heartbeat to the rendezvous '14019182' due to an error of type RendezvousConnectionError.
+[2025-01-02 19:10:48,390] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-172-147.ec2.internal_48859_0' has failed to send a keep-alive heartbeat to the rendezvous '14019182' due to an error of type RendezvousConnectionError.
+[2025-01-02 19:10:48,394] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-164-45.ec2.internal_281066_0' has failed to send a keep-alive heartbeat to the rendezvous '14019182' due to an error of type RendezvousConnectionError.
+[2025-01-02 19:10:48,395] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-171-21.ec2.internal_317421_0' has failed to send a keep-alive heartbeat to the rendezvous '14019182' due to an error of type RendezvousConnectionError.
+[2025-01-02 19:10:48,402] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-167-51.ec2.internal_34161_0' has failed to send a keep-alive heartbeat to the rendezvous '14019182' due to an error of type RendezvousConnectionError.
+[2025-01-02 19:10:48,405] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-160-192.ec2.internal_53721_0' has failed to send a keep-alive heartbeat to the rendezvous '14019182' due to an error of type RendezvousConnectionError.
+[2025-01-02 19:10:48,419] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-171-88.ec2.internal_567550_0' has failed to send a keep-alive heartbeat to the rendezvous '14019182' due to an error of type RendezvousConnectionError.
+[2025-01-02 19:10:48,428] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-169-139.ec2.internal_54773_0' has failed to send a keep-alive heartbeat to the rendezvous '14019182' due to an error of type RendezvousConnectionError.
+[2025-01-02 19:10:48,431] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-169-247.ec2.internal_501555_0' has failed to send a keep-alive heartbeat to the rendezvous '14019182' due to an error of type RendezvousConnectionError.
+[2025-01-02 19:10:48,444] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-173-246.ec2.internal_592749_0' has failed to send a keep-alive heartbeat to the rendezvous '14019182' due to an error of type RendezvousConnectionError.
+[2025-01-02 19:10:48,451] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-160-103.ec2.internal_2778715_0' has failed to send a keep-alive heartbeat to the rendezvous '14019182' due to an error of type RendezvousConnectionError.
+[2025-01-02 19:10:48,471] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-166-125.ec2.internal_323177_0' has failed to send a keep-alive heartbeat to the rendezvous '14019182' due to an error of type RendezvousConnectionError.
+[2025-01-02 19:10:48,481] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-173-121.ec2.internal_3239565_0' has failed to send a keep-alive heartbeat to the rendezvous '14019182' due to an error of type RendezvousConnectionError.
+[2025-01-02 19:10:48,496] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-172-252.ec2.internal_458973_0' has failed to send a keep-alive heartbeat to the rendezvous '14019182' due to an error of type RendezvousConnectionError.
+srun: error: ip-26-0-174-36: task 63: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 160096 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 596442 got signal: 15
+srun: error: ip-26-0-167-175: task 25: Exited with exit code 1
+srun: error: ip-26-0-165-131: task 14: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 88502 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 150433 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 27433 got signal: 15
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 199189 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 89805 got signal: 15
+srun: error: ip-26-0-169-132: task 36: Exited with exit code 1
+srun: error: ip-26-0-165-38: task 12: Exited with exit code 1
+srun: error: ip-26-0-171-62: task 47: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 69330 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 48859 got signal: 15
+srun: error: ip-26-0-167-9: task 23: Exited with exit code 1
+srun: error: ip-26-0-170-160: task 44: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 273037 got signal: 15
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 773639 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 150252 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 591598 got signal: 15
+srun: error: ip-26-0-164-207: task 9: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 523874 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 34161 got signal: 15
+srun: error: ip-26-0-165-24: task 11: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 120894 got signal: 15
+srun: error: ip-26-0-171-230: task 51: Exited with exit code 1
+srun: error: ip-26-0-172-147: task 57: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+srun: error: ip-26-0-173-202: task 61: Exited with exit code 1
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 91930 got signal: 15
+srun: error: ip-26-0-164-236: task 10: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+srun: error: ip-26-0-169-86: task 35: Exited with exit code 1
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 287243 got signal: 15
+srun: error: ip-26-0-170-31: task 41: Exited with exit code 1
+srun: error: ip-26-0-167-51: task 24: Exited with exit code 1
+srun: error: ip-26-0-163-236: task 3: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 279639 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 317421 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 537880 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 112504 got signal: 15
+srun: error: ip-26-0-168-34: task 30: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 153417 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 3239565 got signal: 15
+srun: error: ip-26-0-166-214: task 21: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 141637 got signal: 15
+srun: error: ip-26-0-171-21: task 45: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 501555 got signal: 15
+srun: error: ip-26-0-164-0: task 4: Exited with exit code 1
+srun: error: ip-26-0-164-187: task 8: Exited with exit code 1
+srun: error: ip-26-0-166-15: task 18: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 235972 got signal: 15
+[2025-01-02 19:10:52,995] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-168-238.ec2.internal_19250_0' has failed to send a keep-alive heartbeat to the rendezvous '14019182' due to an error of type RendezvousConnectionError.
+srun: error: ip-26-0-173-121: task 60: Exited with exit code 1
+[2025-01-02 19:10:53,141] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-164-18.ec2.internal_410504_0' has failed to send a keep-alive heartbeat to the rendezvous '14019182' due to an error of type RendezvousConnectionError.
+[2025-01-02 19:10:53,150] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-169-239.ec2.internal_77642_0' has failed to send a keep-alive heartbeat to the rendezvous '14019182' due to an error of type RendezvousConnectionError.
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 207894 got signal: 15
+[2025-01-02 19:10:53,166] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-165-213.ec2.internal_37924_0' has failed to send a keep-alive heartbeat to the rendezvous '14019182' due to an error of type RendezvousConnectionError.
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+srun: error: ip-26-0-165-164: task 15: Exited with exit code 1
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    elastic_launch(
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = f(*args, **kwargs)
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 2778715 got signal: 15
+    result = agent.run()
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 481798 got signal: 15
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 318401 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 592749 got signal: 15
+[2025-01-02 19:10:53,192] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-171-249.ec2.internal_31050_0' has failed to send a keep-alive heartbeat to the rendezvous '14019182' due to an error of type RendezvousConnectionError.
+[2025-01-02 19:10:53,200] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-167-217.ec2.internal_133226_0' has failed to send a keep-alive heartbeat to the rendezvous '14019182' due to an error of type RendezvousConnectionError.
+srun: error: ip-26-0-169-247: task 40: Exited with exit code 1
+[2025-01-02 19:10:53,213] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-171-168.ec2.internal_1043759_0' has failed to send a keep-alive heartbeat to the rendezvous '14019182' due to an error of type RendezvousConnectionError.
+[2025-01-02 19:10:53,234] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-168-95.ec2.internal_171491_0' has failed to send a keep-alive heartbeat to the rendezvous '14019182' due to an error of type RendezvousConnectionError.
+[2025-01-02 19:10:53,245] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-165-202.ec2.internal_134736_0' has failed to send a keep-alive heartbeat to the rendezvous '14019182' due to an error of type RendezvousConnectionError.
+[2025-01-02 19:10:53,256] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-167-245.ec2.internal_3958245_0' has failed to send a keep-alive heartbeat to the rendezvous '14019182' due to an error of type RendezvousConnectionError.
+srun: error: ip-26-0-167-177: task 26: Exited with exit code 1
+[2025-01-02 19:10:53,265] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-171-102.ec2.internal_146861_0' has failed to send a keep-alive heartbeat to the rendezvous '14019182' due to an error of type RendezvousConnectionError.
+[2025-01-02 19:10:53,294] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-165-59.ec2.internal_149076_0' has failed to send a keep-alive heartbeat to the rendezvous '14019182' due to an error of type RendezvousConnectionError.
+[2025-01-02 19:10:53,296] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-171-56.ec2.internal_215955_0' has failed to send a keep-alive heartbeat to the rendezvous '14019182' due to an error of type RendezvousConnectionError.
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    run(args)
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 31050 got signal: 15
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 458973 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+[2025-01-02 19:10:53,326] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-168-120.ec2.internal_71963_0' has failed to send a keep-alive heartbeat to the rendezvous '14019182' due to an error of type RendezvousConnectionError.
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+[2025-01-02 19:10:53,328] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-172-142.ec2.internal_24682_0' has failed to send a keep-alive heartbeat to the rendezvous '14019182' due to an error of type RendezvousConnectionError.
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 149702 got signal: 15
+[2025-01-02 19:10:53,333] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-173-7.ec2.internal_557082_0' has failed to send a keep-alive heartbeat to the rendezvous '14019182' due to an error of type RendezvousConnectionError.
+[2025-01-02 19:10:53,341] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-166-244.ec2.internal_91759_0' has failed to send a keep-alive heartbeat to the rendezvous '14019182' due to an error of type RendezvousConnectionError.
+[2025-01-02 19:10:53,346] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-164-75.ec2.internal_330124_0' has failed to send a keep-alive heartbeat to the rendezvous '14019182' due to an error of type RendezvousConnectionError.
+[2025-01-02 19:10:53,356] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-166-36.ec2.internal_387749_0' has failed to send a keep-alive heartbeat to the rendezvous '14019182' due to an error of type RendezvousConnectionError.
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 19250 got signal: 15
+[2025-01-02 19:10:53,369] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-170-132.ec2.internal_568819_0' has failed to send a keep-alive heartbeat to the rendezvous '14019182' due to an error of type RendezvousConnectionError.
+[2025-01-02 19:10:53,372] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-172-73.ec2.internal_124739_0' has failed to send a keep-alive heartbeat to the rendezvous '14019182' due to an error of type RendezvousConnectionError.
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 133226 got signal: 15
+[2025-01-02 19:10:53,380] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-170-143.ec2.internal_28487_0' has failed to send a keep-alive heartbeat to the rendezvous '14019182' due to an error of type RendezvousConnectionError.
+[2025-01-02 19:10:53,385] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-169-207.ec2.internal_92160_0' has failed to send a keep-alive heartbeat to the rendezvous '14019182' due to an error of type RendezvousConnectionError.
+[2025-01-02 19:10:53,395] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-164-45.ec2.internal_281066_0' has failed to send a keep-alive heartbeat to the rendezvous '14019182' due to an error of type RendezvousConnectionError.
+[2025-01-02 19:10:53,406] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-160-192.ec2.internal_53721_0' has failed to send a keep-alive heartbeat to the rendezvous '14019182' due to an error of type RendezvousConnectionError.
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+srun: error: ip-26-0-172-116: task 55: Exited with exit code 1
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 387749 got signal: 15
+[2025-01-02 19:10:53,421] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-171-88.ec2.internal_567550_0' has failed to send a keep-alive heartbeat to the rendezvous '14019182' due to an error of type RendezvousConnectionError.
+[2025-01-02 19:10:53,430] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-169-139.ec2.internal_54773_0' has failed to send a keep-alive heartbeat to the rendezvous '14019182' due to an error of type RendezvousConnectionError.
+srun: error: ip-26-0-160-103: task 1: Exited with exit code 1
+srun: error: ip-26-0-173-246: task 62: Exited with exit code 1
+srun: error: ip-26-0-168-52: task 31: Exited with exit code 1
+[2025-01-02 19:10:53,473] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-166-125.ec2.internal_323177_0' has failed to send a keep-alive heartbeat to the rendezvous '14019182' due to an error of type RendezvousConnectionError.
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 28487 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 323177 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 92160 got signal: 15
+srun: error: ip-26-0-168-30: task 29: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 568819 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+srun: error: ip-26-0-172-252: task 58: Exited with exit code 1
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 281066 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 149076 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+srun: error: ip-26-0-168-238: task 34: Exited with exit code 1
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 53721 got signal: 15
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 171491 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 410504 got signal: 15
+srun: error: ip-26-0-167-217: task 27: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 54773 got signal: 15
+srun: error: ip-26-0-172-57: task 53: Exited with exit code 1
+srun: error: ip-26-0-171-249: task 52: Exited with exit code 1
+srun: error: ip-26-0-166-36: task 19: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 124739 got signal: 15
+srun: error: ip-26-0-166-125: task 20: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+srun: error: ip-26-0-170-143: task 43: Exited with exit code 1
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 71963 got signal: 15
+srun: error: ip-26-0-169-207: task 38: Exited with exit code 1
+srun: error: ip-26-0-170-132: task 42: Exited with exit code 1
+srun: error: ip-26-0-164-45: task 6: Exited with exit code 1
+srun: error: ip-26-0-160-192: task 2: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 146861 got signal: 15
+srun: error: ip-26-0-168-95: task 32: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 567550 got signal: 15
+srun: error: ip-26-0-164-18: task 5: Exited with exit code 1
+srun: error: ip-26-0-165-59: task 13: Exited with exit code 1
+srun: error: ip-26-0-172-73: task 54: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 37924 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 134736 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 215955 got signal: 15
+srun: error: ip-26-0-168-120: task 33: Exited with exit code 1
+srun: error: ip-26-0-171-102: task 49: Exited with exit code 1
+srun: error: ip-26-0-169-139: task 37: Exited with exit code 1
+srun: error: ip-26-0-171-88: task 48: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 24682 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 330124 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 91759 got signal: 15
+srun: error: ip-26-0-165-213: task 17: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+srun: error: ip-26-0-165-202: task 16: Exited with exit code 1
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 77642 got signal: 15
+srun: error: ip-26-0-171-56: task 46: Exited with exit code 1
+srun: error: ip-26-0-172-142: task 56: Exited with exit code 1
+srun: error: ip-26-0-166-244: task 22: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 1043759 got signal: 15
+srun: error: ip-26-0-164-75: task 7: Exited with exit code 1
+srun: error: ip-26-0-169-239: task 39: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+srun: error: ip-26-0-171-168: task 50: Exited with exit code 1
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 3958245 got signal: 15
+srun: error: ip-26-0-167-245: task 28: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 557082 got signal: 15
+srun: error: ip-26-0-173-7: task 59: Exited with exit code 1
diff --git a/logs/14019215-bench_8.86G_dp64_tp8_pp1_acc2_mbs2_seq4096_zero0_tpmodeRED_vocab131k.out b/logs/14019215-bench_8.86G_dp64_tp8_pp1_acc2_mbs2_seq4096_zero0_tpmodeRED_vocab131k.out
new file mode 100644
index 0000000000000000000000000000000000000000..16460801f341772812235e4c65fae5977cf3315f
--- /dev/null
+++ b/logs/14019215-bench_8.86G_dp64_tp8_pp1_acc2_mbs2_seq4096_zero0_tpmodeRED_vocab131k.out
@@ -0,0 +1,11094 @@
++ '[' -z 14019215 ']'
++ source /etc/profile.d/modules.sh
+++ . /usr/share/modules/init/bash
++++ unset _mlshdbg
++++ '[' 0 = 1 ']'
++++ unset _mlre _mlIFS
++++ '[' -n x ']'
++++ _mlIFS=' 	
+'
++++ IFS=' '
++++ '[' -n '' ']'
+++++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash autoinit
++++ _mlcode='module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+MODULEPATH_modshare=/etc/environment-modules/modules:1:/usr/share/modules/\$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1; export MODULEPATH_modshare;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+LOADEDMODULES=; export LOADEDMODULES;
+MODULEPATH=/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/\$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles; export MODULEPATH;
+test 0;'
++++ _mlret=0
++++ '[' -n x ']'
++++ IFS=' 	
+'
++++ unset _mlIFS
++++ unset _mlre _mlv _mlrv
++++ '[' 0 -eq 0 ']'
++++ eval 'module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+MODULEPATH_modshare=/etc/environment-modules/modules:1:/usr/share/modules/\$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1; export MODULEPATH_modshare;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+LOADEDMODULES=; export LOADEDMODULES;
+MODULEPATH=/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/\$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles; export MODULEPATH;
+test 0;'
+++++ MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl
+++++ export MODULES_CMD
+++++ ENV=/usr/share/modules/init/profile.sh
+++++ export ENV
+++++ MODULEPATH_modshare='/etc/environment-modules/modules:1:/usr/share/modules/$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1'
+++++ export MODULEPATH_modshare
+++++ BASH_ENV=/usr/share/modules/init/bash
+++++ export BASH_ENV
+++++ MODULESHOME=/usr/share/modules
+++++ export MODULESHOME
+++++ LOADEDMODULES=
+++++ export LOADEDMODULES
+++++ MODULEPATH='/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles'
+++++ export MODULEPATH
+++++ test 0
++++ '[' 0 = 1 ']'
++++ '[' -t 2 ']'
++++ export -f module
++++ export -f switchml
++++ '[' 5 -ge 3 ']'
++++ [[ ehxB =~ i ]]
++++ [[ ! :/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin: =~ :/usr/bin: ]]
++++ '[' '!' -n '' ']'
++++ MANPATH=:
++++ export MANPATH
+++++ manpath
++++ [[ ! :/fsx/nouamane/miniconda/envs/2-1-cu121/man:/fsx/nouamane/miniconda/envs/2-1-cu121/share/man:/fsx/nouamane/miniconda/man:/fsx/nouamane/miniconda/share/man:/usr/local/man:/usr/local/share/man:/usr/share/man:/opt/slurm/share/man:/admin/home/nouamane/.local/share/man:/admin/home/nouamane/.fzf/man:: =~ :/usr/share/man: ]]
++++ unset _mlcode _mlret
++++ '[' -n '' ']'
++ module load cuda/12.1
++ unset _mlshdbg
++ '[' 0 = 1 ']'
++ unset _mlre _mlIFS
++ '[' -n x ']'
++ _mlIFS=' 	
+'
++ IFS=' '
++ '[' -n '' ']'
+++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash load cuda/12.1
++ eval 'CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include;' export 'CPATH;
+LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:;' export 'LD_LIBRARY_PATH;
+FI_EFA_FORK_SAFE=1;' export 'FI_EFA_FORK_SAFE;
+MANPATH=/usr/local/cuda-12.1/share/man::;' export 'MANPATH;
+LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64;' export 'LIBRARY_PATH;
+_LMFILES_=/usr/share/modules/modulefiles/cuda/12.1;' export '_LMFILES_;
+LOADEDMODULES=cuda/12.1;' export 'LOADEDMODULES;
+MPI_PATH=/opt/amazon/openmpi;' export 'MPI_PATH;
+NCCL_HOME_modshare=/opt/nccl/build:1;' export 'NCCL_HOME_modshare;
+NCCL_PROTO=simple;' export 'NCCL_PROTO;
+MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1;' export 'MANPATH_modshare;
+LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1;' export 'LIBRARY_PATH_modshare;
+NCCL_SOCKET_IFNAME=enp;' export 'NCCL_SOCKET_IFNAME;
+AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl;' export 'AWS_OFI_NCCL_HOME;
+NCCL_HOME=/opt/nccl/build;' export 'NCCL_HOME;
+FI_PROVIDER=efa;' export 'FI_PROVIDER;
+AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1;' export 'AWS_OFI_NCCL_HOME_modshare;
+CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1;' export 'CPATH_modshare;
+LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2::1:/opt/nccl/build/lib:1:/opt/aws-ofi-nccl/lib:2:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/efa/lib:1;' export 'LD_LIBRARY_PATH_modshare;
+FI_EFA_ENABLE_SHM_TRANSFER=1;' export 'FI_EFA_ENABLE_SHM_TRANSFER;
+_LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1;' export '_LMFILES__modshare;
+LOADEDMODULES_modshare=cuda/12.1:1;' export 'LOADEDMODULES_modshare;
+MPI_PATH_modshare=/opt/amazon/openmpi:1;' export 'MPI_PATH_modshare;
+PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin;' export 'PATH;
+CUDA_HOME=/usr/local/cuda-12.1;' export 'CUDA_HOME;
+PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/opt/slurm/bin:1:/usr/sbin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1;' export 'PATH_modshare;
+test' '0;'
+++ CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include
+++ export CPATH
+++ LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:
+++ export LD_LIBRARY_PATH
+++ FI_EFA_FORK_SAFE=1
+++ export FI_EFA_FORK_SAFE
+++ MANPATH=/usr/local/cuda-12.1/share/man::
+++ export MANPATH
+++ LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64
+++ export LIBRARY_PATH
+++ _LMFILES_=/usr/share/modules/modulefiles/cuda/12.1
+++ export _LMFILES_
+++ LOADEDMODULES=cuda/12.1
+++ export LOADEDMODULES
+++ MPI_PATH=/opt/amazon/openmpi
+++ export MPI_PATH
+++ NCCL_HOME_modshare=/opt/nccl/build:1
+++ export NCCL_HOME_modshare
+++ NCCL_PROTO=simple
+++ export NCCL_PROTO
+++ MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1
+++ export MANPATH_modshare
+++ LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1
+++ export LIBRARY_PATH_modshare
+++ NCCL_SOCKET_IFNAME=enp
+++ export NCCL_SOCKET_IFNAME
+++ AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl
+++ export AWS_OFI_NCCL_HOME
+++ NCCL_HOME=/opt/nccl/build
+++ export NCCL_HOME
+++ FI_PROVIDER=efa
+++ export FI_PROVIDER
+++ AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1
+++ export AWS_OFI_NCCL_HOME_modshare
+++ CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1
+++ export CPATH_modshare
+++ LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2::1:/opt/nccl/build/lib:1:/opt/aws-ofi-nccl/lib:2:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/efa/lib:1
+++ export LD_LIBRARY_PATH_modshare
+++ FI_EFA_ENABLE_SHM_TRANSFER=1
+++ export FI_EFA_ENABLE_SHM_TRANSFER
+++ _LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1
+++ export _LMFILES__modshare
+++ LOADEDMODULES_modshare=cuda/12.1:1
+++ export LOADEDMODULES_modshare
+++ MPI_PATH_modshare=/opt/amazon/openmpi:1
+++ export MPI_PATH_modshare
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
+++ export PATH
+++ CUDA_HOME=/usr/local/cuda-12.1
+++ export CUDA_HOME
+++ PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/opt/slurm/bin:1:/usr/sbin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1
+++ export PATH_modshare
+++ test 0
++ _mlstatus=0
++ '[' -n x ']'
++ IFS=' 	
+'
++ unset _mlre _mlv _mlrv _mlIFS
++ '[' -n '' ']'
++ unset _mlshdbg
++ return 0
++ source /fsx/nouamane/miniconda/bin/activate
+++ _CONDA_ROOT=/fsx/nouamane/miniconda
+++ . /fsx/nouamane/miniconda/etc/profile.d/conda.sh
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ '[' -z x ']'
+++ conda activate
+++ local cmd=activate
+++ case "$cmd" in
+++ __conda_activate activate
+++ '[' -n '' ']'
+++ local ask_conda
++++ PS1=
++++ __conda_exe shell.posix activate
++++ /fsx/nouamane/miniconda/bin/conda shell.posix activate
+Error while loading conda entry point: conda-libmamba-solver (module 'libmambapy' has no attribute 'QueryFormat')
+++ ask_conda='. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''2'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_1='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/etc/conda/activate.d/libxml2_activate.sh"'
+++ eval '. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''2'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_1='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/etc/conda/activate.d/libxml2_activate.sh"'
++++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh
+++++ test -n ''
+++++ unset XML_CATALOG_FILES
+++++ unset xml_catalog_files_libxml2
++++ PS1='(base) '
++++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
++++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
++++ export CONDA_PREFIX=/fsx/nouamane/miniconda
++++ CONDA_PREFIX=/fsx/nouamane/miniconda
++++ export CONDA_SHLVL=2
++++ CONDA_SHLVL=2
++++ export CONDA_DEFAULT_ENV=base
++++ CONDA_DEFAULT_ENV=base
++++ export 'CONDA_PROMPT_MODIFIER=(base) '
++++ CONDA_PROMPT_MODIFIER='(base) '
++++ export CONDA_PREFIX_1=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ CONDA_PREFIX_1=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ . /fsx/nouamane/miniconda/etc/conda/activate.d/libxml2_activate.sh
+++++ test -n ''
+++++ xml_catalog_files_libxml2=
+++++ XML_CATALOG_FILES=
+++++ conda_catalog_files=
+++++ ifs_libxml2=' 	
+'
+++++ IFS=' '
+++++ rem=/fsx/nouamane/miniconda
+++++ for pre in ${rem}
+++++ test '' = /fsx/nouamane/miniconda
+++++ conda_catalog_files=/fsx/nouamane/miniconda
+++++ rem=
+++++ IFS=' 	
+'
+++++ conda_catalog_files='file:///fsx/nouamane/miniconda/etc/xml/catalog file:///etc/xml/catalog'
+++++ export 'XML_CATALOG_FILES=file:///fsx/nouamane/miniconda/etc/xml/catalog file:///etc/xml/catalog'
+++++ XML_CATALOG_FILES='file:///fsx/nouamane/miniconda/etc/xml/catalog file:///etc/xml/catalog'
+++++ unset conda_catalog_files ifs_libxml2 rem
+++ __conda_hashr
+++ '[' -n '' ']'
+++ '[' -n '' ']'
+++ hash -r
++ conda activate 2-1-cu121
++ local cmd=activate
++ case "$cmd" in
++ __conda_activate activate 2-1-cu121
++ '[' -n '' ']'
++ local ask_conda
+++ PS1='(base) '
+++ __conda_exe shell.posix activate 2-1-cu121
+++ /fsx/nouamane/miniconda/bin/conda shell.posix activate 2-1-cu121
+Error while loading conda entry point: conda-libmamba-solver (module 'libmambapy' has no attribute 'QueryFormat')
++ ask_conda='. "/fsx/nouamane/miniconda/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_2='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
++ eval '. "/fsx/nouamane/miniconda/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_2='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
+++ . /fsx/nouamane/miniconda/etc/conda/deactivate.d/libxml2_deactivate.sh
++++ test -n ''
++++ unset XML_CATALOG_FILES
++++ unset xml_catalog_files_libxml2
+++ PS1='(2-1-cu121) '
+++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
+++ export CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ export CONDA_SHLVL=3
+++ CONDA_SHLVL=3
+++ export CONDA_DEFAULT_ENV=2-1-cu121
+++ CONDA_DEFAULT_ENV=2-1-cu121
+++ export 'CONDA_PROMPT_MODIFIER=(2-1-cu121) '
+++ CONDA_PROMPT_MODIFIER='(2-1-cu121) '
+++ export CONDA_PREFIX_2=/fsx/nouamane/miniconda
+++ CONDA_PREFIX_2=/fsx/nouamane/miniconda
+++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ export _CE_M=
+++ _CE_M=
+++ export _CE_CONDA=
+++ _CE_CONDA=
+++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh
++++ test -n ''
++++ xml_catalog_files_libxml2=
++++ XML_CATALOG_FILES=
++++ conda_catalog_files=
++++ ifs_libxml2=' 	
+'
++++ IFS=' '
++++ rem=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ for pre in ${rem}
++++ test '' = /fsx/nouamane/miniconda/envs/2-1-cu121
++++ conda_catalog_files=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ rem=
++++ IFS=' 	
+'
++++ conda_catalog_files='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ export 'XML_CATALOG_FILES=file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ XML_CATALOG_FILES='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ unset conda_catalog_files ifs_libxml2 rem
++ __conda_hashr
++ '[' -n '' ']'
++ '[' -n '' ']'
++ hash -r
++ export PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
++ PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
++ '[' -z '' ']'
+++ scontrol show hostnames 'ip-26-0-161-[78,103,123],ip-26-0-163-236,ip-26-0-164-[0,18,45,75,187,207,236],ip-26-0-165-[24,38,59,131,164,202,213],ip-26-0-166-[15,36,125,214,244],ip-26-0-167-[9,51,175,177,217,245],ip-26-0-168-[95,120,238],ip-26-0-169-[86,132,139,207,239,247],ip-26-0-170-[31,132,143,160],ip-26-0-171-[21,56,62,88,102,168,230,249],ip-26-0-172-[57,73,116,142],ip-26-0-174-[100,186,196,240],ip-26-0-175-[19,34,132,165,170,241]'
++ export 'NODELIST=ip-26-0-161-78
+ip-26-0-161-103
+ip-26-0-161-123
+ip-26-0-163-236
+ip-26-0-164-0
+ip-26-0-164-18
+ip-26-0-164-45
+ip-26-0-164-75
+ip-26-0-164-187
+ip-26-0-164-207
+ip-26-0-164-236
+ip-26-0-165-24
+ip-26-0-165-38
+ip-26-0-165-59
+ip-26-0-165-131
+ip-26-0-165-164
+ip-26-0-165-202
+ip-26-0-165-213
+ip-26-0-166-15
+ip-26-0-166-36
+ip-26-0-166-125
+ip-26-0-166-214
+ip-26-0-166-244
+ip-26-0-167-9
+ip-26-0-167-51
+ip-26-0-167-175
+ip-26-0-167-177
+ip-26-0-167-217
+ip-26-0-167-245
+ip-26-0-168-95
+ip-26-0-168-120
+ip-26-0-168-238
+ip-26-0-169-86
+ip-26-0-169-132
+ip-26-0-169-139
+ip-26-0-169-207
+ip-26-0-169-239
+ip-26-0-169-247
+ip-26-0-170-31
+ip-26-0-170-132
+ip-26-0-170-143
+ip-26-0-170-160
+ip-26-0-171-21
+ip-26-0-171-56
+ip-26-0-171-62
+ip-26-0-171-88
+ip-26-0-171-102
+ip-26-0-171-168
+ip-26-0-171-230
+ip-26-0-171-249
+ip-26-0-172-57
+ip-26-0-172-73
+ip-26-0-172-116
+ip-26-0-172-142
+ip-26-0-174-100
+ip-26-0-174-186
+ip-26-0-174-196
+ip-26-0-174-240
+ip-26-0-175-19
+ip-26-0-175-34
+ip-26-0-175-132
+ip-26-0-175-165
+ip-26-0-175-170
+ip-26-0-175-241'
++ NODELIST='ip-26-0-161-78
+ip-26-0-161-103
+ip-26-0-161-123
+ip-26-0-163-236
+ip-26-0-164-0
+ip-26-0-164-18
+ip-26-0-164-45
+ip-26-0-164-75
+ip-26-0-164-187
+ip-26-0-164-207
+ip-26-0-164-236
+ip-26-0-165-24
+ip-26-0-165-38
+ip-26-0-165-59
+ip-26-0-165-131
+ip-26-0-165-164
+ip-26-0-165-202
+ip-26-0-165-213
+ip-26-0-166-15
+ip-26-0-166-36
+ip-26-0-166-125
+ip-26-0-166-214
+ip-26-0-166-244
+ip-26-0-167-9
+ip-26-0-167-51
+ip-26-0-167-175
+ip-26-0-167-177
+ip-26-0-167-217
+ip-26-0-167-245
+ip-26-0-168-95
+ip-26-0-168-120
+ip-26-0-168-238
+ip-26-0-169-86
+ip-26-0-169-132
+ip-26-0-169-139
+ip-26-0-169-207
+ip-26-0-169-239
+ip-26-0-169-247
+ip-26-0-170-31
+ip-26-0-170-132
+ip-26-0-170-143
+ip-26-0-170-160
+ip-26-0-171-21
+ip-26-0-171-56
+ip-26-0-171-62
+ip-26-0-171-88
+ip-26-0-171-102
+ip-26-0-171-168
+ip-26-0-171-230
+ip-26-0-171-249
+ip-26-0-172-57
+ip-26-0-172-73
+ip-26-0-172-116
+ip-26-0-172-142
+ip-26-0-174-100
+ip-26-0-174-186
+ip-26-0-174-196
+ip-26-0-174-240
+ip-26-0-175-19
+ip-26-0-175-34
+ip-26-0-175-132
+ip-26-0-175-165
+ip-26-0-175-170
+ip-26-0-175-241'
+++ head -n1
+++ scontrol show hostnames 'ip-26-0-161-[78,103,123],ip-26-0-163-236,ip-26-0-164-[0,18,45,75,187,207,236],ip-26-0-165-[24,38,59,131,164,202,213],ip-26-0-166-[15,36,125,214,244],ip-26-0-167-[9,51,175,177,217,245],ip-26-0-168-[95,120,238],ip-26-0-169-[86,132,139,207,239,247],ip-26-0-170-[31,132,143,160],ip-26-0-171-[21,56,62,88,102,168,230,249],ip-26-0-172-[57,73,116,142],ip-26-0-174-[100,186,196,240],ip-26-0-175-[19,34,132,165,170,241]'
++ export MASTER_NODE=ip-26-0-161-78
++ MASTER_NODE=ip-26-0-161-78
++ export MASTER_PORT=12356
++ MASTER_PORT=12356
++ export NNODES=64
++ NNODES=64
++ export GPUS_PER_NODE=8
++ GPUS_PER_NODE=8
++ export WORLD_SIZE=512
++ WORLD_SIZE=512
++ export CUDA_DEVICE_MAX_CONNECTIONS=1
++ CUDA_DEVICE_MAX_CONNECTIONS=1
++ export NCCL_DEBUG=WARN
++ NCCL_DEBUG=WARN
++ export NANOTRON_BENCHMARK=1
++ NANOTRON_BENCHMARK=1
++ export WANDB_MODE=disabled
++ WANDB_MODE=disabled
++ export TORCH_NCCL_ASYNC_ERROR_HANDLING=1
++ TORCH_NCCL_ASYNC_ERROR_HANDLING=1
++ export TORCH_DISTRIBUTED_DEBUG=DETAIL
++ TORCH_DISTRIBUTED_DEBUG=DETAIL
++ '[' -z '' ']'
++ echo '=== GPU Topology ==='
+=== GPU Topology ===
++ nvidia-smi topo -m
+	GPU0	GPU1	GPU2	GPU3	GPU4	GPU5	GPU6	GPU7	CPU Affinity	NUMA Affinity	GPU NUMA ID
+GPU0	 X 	NV18	NV18	NV18	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU1	NV18	 X 	NV18	NV18	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU2	NV18	NV18	 X 	NV18	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU3	NV18	NV18	NV18	 X 	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU4	NV18	NV18	NV18	NV18	 X 	NV18	NV18	NV18	48-95	1		N/A
+GPU5	NV18	NV18	NV18	NV18	NV18	 X 	NV18	NV18	48-95	1		N/A
+GPU6	NV18	NV18	NV18	NV18	NV18	NV18	 X 	NV18	48-95	1		N/A
+GPU7	NV18	NV18	NV18	NV18	NV18	NV18	NV18	 X 	48-95	1		N/A
+
+Legend:
+
+  X    = Self
+  SYS  = Connection traversing PCIe as well as the SMP interconnect between NUMA nodes (e.g., QPI/UPI)
+  NODE = Connection traversing PCIe as well as the interconnect between PCIe Host Bridges within a NUMA node
+  PHB  = Connection traversing PCIe as well as a PCIe Host Bridge (typically the CPU)
+  PXB  = Connection traversing multiple PCIe bridges (without traversing the PCIe Host Bridge)
+  PIX  = Connection traversing at most a single PCIe bridge
+  NV#  = Connection traversing a bonded set of # NVLinks
++ echo ==================
+==================
++ export SRUN_ALLOC_ARGS=
++ SRUN_ALLOC_ARGS=
++ echo 'Master node: ip-26-0-161-78'
+Master node: ip-26-0-161-78
++ echo 'All nodes: ip-26-0-161-78
+ip-26-0-161-103
+ip-26-0-161-123
+ip-26-0-163-236
+ip-26-0-164-0
+ip-26-0-164-18
+ip-26-0-164-45
+ip-26-0-164-75
+ip-26-0-164-187
+ip-26-0-164-207
+ip-26-0-164-236
+ip-26-0-165-24
+ip-26-0-165-38
+ip-26-0-165-59
+ip-26-0-165-131
+ip-26-0-165-164
+ip-26-0-165-202
+ip-26-0-165-213
+ip-26-0-166-15
+ip-26-0-166-36
+ip-26-0-166-125
+ip-26-0-166-214
+ip-26-0-166-244
+ip-26-0-167-9
+ip-26-0-167-51
+ip-26-0-167-175
+ip-26-0-167-177
+ip-26-0-167-217
+ip-26-0-167-245
+ip-26-0-168-95
+ip-26-0-168-120
+ip-26-0-168-238
+ip-26-0-169-86
+ip-26-0-169-132
+ip-26-0-169-139
+ip-26-0-169-207
+ip-26-0-169-239
+ip-26-0-169-247
+ip-26-0-170-31
+ip-26-0-170-132
+ip-26-0-170-143
+ip-26-0-170-160
+ip-26-0-171-21
+ip-26-0-171-56
+ip-26-0-171-62
+ip-26-0-171-88
+ip-26-0-171-102
+ip-26-0-171-168
+ip-26-0-171-230
+ip-26-0-171-249
+ip-26-0-172-57
+ip-26-0-172-73
+ip-26-0-172-116
+ip-26-0-172-142
+ip-26-0-174-100
+ip-26-0-174-186
+ip-26-0-174-196
+ip-26-0-174-240
+ip-26-0-175-19
+ip-26-0-175-34
+ip-26-0-175-132
+ip-26-0-175-165
+ip-26-0-175-170
+ip-26-0-175-241'
+All nodes: ip-26-0-161-78
+ip-26-0-161-103
+ip-26-0-161-123
+ip-26-0-163-236
+ip-26-0-164-0
+ip-26-0-164-18
+ip-26-0-164-45
+ip-26-0-164-75
+ip-26-0-164-187
+ip-26-0-164-207
+ip-26-0-164-236
+ip-26-0-165-24
+ip-26-0-165-38
+ip-26-0-165-59
+ip-26-0-165-131
+ip-26-0-165-164
+ip-26-0-165-202
+ip-26-0-165-213
+ip-26-0-166-15
+ip-26-0-166-36
+ip-26-0-166-125
+ip-26-0-166-214
+ip-26-0-166-244
+ip-26-0-167-9
+ip-26-0-167-51
+ip-26-0-167-175
+ip-26-0-167-177
+ip-26-0-167-217
+ip-26-0-167-245
+ip-26-0-168-95
+ip-26-0-168-120
+ip-26-0-168-238
+ip-26-0-169-86
+ip-26-0-169-132
+ip-26-0-169-139
+ip-26-0-169-207
+ip-26-0-169-239
+ip-26-0-169-247
+ip-26-0-170-31
+ip-26-0-170-132
+ip-26-0-170-143
+ip-26-0-170-160
+ip-26-0-171-21
+ip-26-0-171-56
+ip-26-0-171-62
+ip-26-0-171-88
+ip-26-0-171-102
+ip-26-0-171-168
+ip-26-0-171-230
+ip-26-0-171-249
+ip-26-0-172-57
+ip-26-0-172-73
+ip-26-0-172-116
+ip-26-0-172-142
+ip-26-0-174-100
+ip-26-0-174-186
+ip-26-0-174-196
+ip-26-0-174-240
+ip-26-0-175-19
+ip-26-0-175-34
+ip-26-0-175-132
+ip-26-0-175-165
+ip-26-0-175-170
+ip-26-0-175-241
++ echo 'World size: 512'
+World size: 512
++ '[' -n '' ']'
++ srun --wait=0 --kill-on-bad-exit=1 torchrun --nnodes=64 --nproc_per_node=8 --rdzv_id=14019215 --rdzv_backend=c10d --rdzv_endpoint=ip-26-0-161-78:12356 --max_restarts 0 --rdzv_conf timeout=60 /fsx/nouamane/projects/nanotron/run_train.py --config-file benchmark/configs/config_8.86G_dp64_tp8_pp1_acc2_mbs2_seq4096_zero0_tpmodeRED_vocab131k.yaml
+[2025-01-02 08:31:52,814] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-02 08:31:52,814] torch.distributed.run: [WARNING] 
+[2025-01-02 08:31:52,814] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 08:31:52,814] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-02 08:31:52,814] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 08:31:52,812] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-02 08:31:52,812] torch.distributed.run: [WARNING] 
+[2025-01-02 08:31:52,812] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 08:31:52,812] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-02 08:31:52,812] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 08:31:52,814] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-02 08:31:52,814] torch.distributed.run: [WARNING] 
+[2025-01-02 08:31:52,814] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 08:31:52,814] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-02 08:31:52,814] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 08:31:52,815] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-02 08:31:52,815] torch.distributed.run: [WARNING] 
+[2025-01-02 08:31:52,815] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 08:31:52,815] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-02 08:31:52,815] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 08:31:52,815] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-02 08:31:52,815] torch.distributed.run: [WARNING] 
+[2025-01-02 08:31:52,815] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 08:31:52,815] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-02 08:31:52,815] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 08:31:52,815] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-02 08:31:52,816] torch.distributed.run: [WARNING] 
+[2025-01-02 08:31:52,816] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 08:31:52,816] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-02 08:31:52,816] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 08:31:52,816] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-02 08:31:52,816] torch.distributed.run: [WARNING] 
+[2025-01-02 08:31:52,816] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 08:31:52,816] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-02 08:31:52,816] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 08:31:52,816] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-02 08:31:52,816] torch.distributed.run: [WARNING] 
+[2025-01-02 08:31:52,816] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 08:31:52,816] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-02 08:31:52,816] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 08:31:52,819] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-02 08:31:52,819] torch.distributed.run: [WARNING] 
+[2025-01-02 08:31:52,819] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 08:31:52,819] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-02 08:31:52,819] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 08:31:52,820] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-02 08:31:52,820] torch.distributed.run: [WARNING] 
+[2025-01-02 08:31:52,820] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 08:31:52,820] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-02 08:31:52,820] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 08:31:52,822] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-02 08:31:52,822] torch.distributed.run: [WARNING] 
+[2025-01-02 08:31:52,822] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 08:31:52,822] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-02 08:31:52,822] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 08:31:52,822] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-02 08:31:52,823] torch.distributed.run: [WARNING] 
+[2025-01-02 08:31:52,823] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 08:31:52,823] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-02 08:31:52,823] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 08:31:52,823] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-02 08:31:52,823] torch.distributed.run: [WARNING] 
+[2025-01-02 08:31:52,823] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 08:31:52,823] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-02 08:31:52,823] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 08:31:52,826] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-02 08:31:52,826] torch.distributed.run: [WARNING] 
+[2025-01-02 08:31:52,826] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 08:31:52,826] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-02 08:31:52,826] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 08:31:52,826] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-02 08:31:52,826] torch.distributed.run: [WARNING] 
+[2025-01-02 08:31:52,826] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 08:31:52,826] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-02 08:31:52,826] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 08:31:52,828] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-02 08:31:52,829] torch.distributed.run: [WARNING] 
+[2025-01-02 08:31:52,829] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 08:31:52,829] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-02 08:31:52,829] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 08:31:52,833] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-02 08:31:52,833] torch.distributed.run: [WARNING] 
+[2025-01-02 08:31:52,833] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 08:31:52,833] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-02 08:31:52,833] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 08:31:52,835] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-02 08:31:52,835] torch.distributed.run: [WARNING] 
+[2025-01-02 08:31:52,835] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 08:31:52,835] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-02 08:31:52,835] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 08:31:52,840] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-02 08:31:52,841] torch.distributed.run: [WARNING] 
+[2025-01-02 08:31:52,841] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 08:31:52,841] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-02 08:31:52,841] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 08:31:52,845] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-02 08:31:52,845] torch.distributed.run: [WARNING] 
+[2025-01-02 08:31:52,845] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 08:31:52,845] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-02 08:31:52,845] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 08:31:52,850] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-02 08:31:52,850] torch.distributed.run: [WARNING] 
+[2025-01-02 08:31:52,850] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 08:31:52,850] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-02 08:31:52,850] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 08:31:52,853] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-02 08:31:52,854] torch.distributed.run: [WARNING] 
+[2025-01-02 08:31:52,854] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 08:31:52,854] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-02 08:31:52,854] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 08:31:52,854] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-02 08:31:52,854] torch.distributed.run: [WARNING] 
+[2025-01-02 08:31:52,854] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 08:31:52,854] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-02 08:31:52,854] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 08:31:52,855] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-02 08:31:52,855] torch.distributed.run: [WARNING] 
+[2025-01-02 08:31:52,855] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 08:31:52,855] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-02 08:31:52,855] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 08:31:52,857] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-02 08:31:52,857] torch.distributed.run: [WARNING] 
+[2025-01-02 08:31:52,857] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 08:31:52,857] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-02 08:31:52,857] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 08:31:52,869] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-02 08:31:52,869] torch.distributed.run: [WARNING] 
+[2025-01-02 08:31:52,869] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 08:31:52,869] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-02 08:31:52,869] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 08:31:52,870] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-02 08:31:52,870] torch.distributed.run: [WARNING] 
+[2025-01-02 08:31:52,870] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 08:31:52,870] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-02 08:31:52,870] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 08:31:52,870] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-02 08:31:52,870] torch.distributed.run: [WARNING] 
+[2025-01-02 08:31:52,870] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 08:31:52,870] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-02 08:31:52,870] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 08:31:52,872] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-02 08:31:52,872] torch.distributed.run: [WARNING] 
+[2025-01-02 08:31:52,872] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 08:31:52,872] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-02 08:31:52,872] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 08:31:52,875] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-02 08:31:52,875] torch.distributed.run: [WARNING] 
+[2025-01-02 08:31:52,875] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 08:31:52,875] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-02 08:31:52,875] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 08:31:52,880] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-02 08:31:52,880] torch.distributed.run: [WARNING] 
+[2025-01-02 08:31:52,880] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 08:31:52,880] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-02 08:31:52,880] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 08:31:52,886] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-02 08:31:52,886] torch.distributed.run: [WARNING] 
+[2025-01-02 08:31:52,886] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 08:31:52,886] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-02 08:31:52,886] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 08:31:52,887] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-02 08:31:52,888] torch.distributed.run: [WARNING] 
+[2025-01-02 08:31:52,888] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 08:31:52,888] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-02 08:31:52,888] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 08:31:52,900] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-02 08:31:52,900] torch.distributed.run: [WARNING] 
+[2025-01-02 08:31:52,900] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 08:31:52,900] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-02 08:31:52,900] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 08:31:52,901] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-02 08:31:52,901] torch.distributed.run: [WARNING] 
+[2025-01-02 08:31:52,901] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 08:31:52,901] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-02 08:31:52,901] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 08:31:52,911] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-02 08:31:52,911] torch.distributed.run: [WARNING] 
+[2025-01-02 08:31:52,911] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 08:31:52,911] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-02 08:31:52,911] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 08:31:52,910] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-02 08:31:52,910] torch.distributed.run: [WARNING] 
+[2025-01-02 08:31:52,910] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 08:31:52,910] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-02 08:31:52,910] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 08:31:52,916] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-02 08:31:52,917] torch.distributed.run: [WARNING] 
+[2025-01-02 08:31:52,917] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 08:31:52,917] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-02 08:31:52,917] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 08:31:52,919] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-02 08:31:52,919] torch.distributed.run: [WARNING] 
+[2025-01-02 08:31:52,919] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 08:31:52,919] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-02 08:31:52,919] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 08:31:52,929] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-02 08:31:52,929] torch.distributed.run: [WARNING] 
+[2025-01-02 08:31:52,929] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 08:31:52,929] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-02 08:31:52,929] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 08:31:52,935] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-02 08:31:52,935] torch.distributed.run: [WARNING] 
+[2025-01-02 08:31:52,935] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 08:31:52,935] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-02 08:31:52,935] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 08:31:52,943] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-02 08:31:52,943] torch.distributed.run: [WARNING] 
+[2025-01-02 08:31:52,943] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 08:31:52,943] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-02 08:31:52,943] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 08:31:52,944] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-02 08:31:52,944] torch.distributed.run: [WARNING] 
+[2025-01-02 08:31:52,944] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 08:31:52,944] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-02 08:31:52,944] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 08:31:52,945] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-02 08:31:52,946] torch.distributed.run: [WARNING] 
+[2025-01-02 08:31:52,946] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 08:31:52,946] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-02 08:31:52,946] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 08:31:52,946] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-02 08:31:52,946] torch.distributed.run: [WARNING] 
+[2025-01-02 08:31:52,946] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 08:31:52,946] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-02 08:31:52,946] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 08:31:52,949] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-02 08:31:52,949] torch.distributed.run: [WARNING] 
+[2025-01-02 08:31:52,949] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 08:31:52,949] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-02 08:31:52,949] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 08:31:52,950] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-02 08:31:52,950] torch.distributed.run: [WARNING] 
+[2025-01-02 08:31:52,950] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 08:31:52,950] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-02 08:31:52,950] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 08:31:52,981] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-02 08:31:52,981] torch.distributed.run: [WARNING] 
+[2025-01-02 08:31:52,981] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 08:31:52,981] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-02 08:31:52,981] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 08:31:52,985] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-02 08:31:52,985] torch.distributed.run: [WARNING] 
+[2025-01-02 08:31:52,985] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 08:31:52,985] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-02 08:31:52,985] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 08:31:52,996] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-02 08:31:52,997] torch.distributed.run: [WARNING] 
+[2025-01-02 08:31:52,997] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 08:31:52,997] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-02 08:31:52,997] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 08:31:53,013] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-02 08:31:53,013] torch.distributed.run: [WARNING] 
+[2025-01-02 08:31:53,013] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 08:31:53,013] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-02 08:31:53,013] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 08:31:53,028] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-02 08:31:53,029] torch.distributed.run: [WARNING] 
+[2025-01-02 08:31:53,029] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 08:31:53,029] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-02 08:31:53,029] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 08:31:53,048] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-02 08:31:53,049] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-02 08:31:53,048] torch.distributed.run: [WARNING] 
+[2025-01-02 08:31:53,048] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 08:31:53,048] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-02 08:31:53,048] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 08:31:53,049] torch.distributed.run: [WARNING] 
+[2025-01-02 08:31:53,049] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 08:31:53,049] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-02 08:31:53,049] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 08:31:53,052] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-02 08:31:53,053] torch.distributed.run: [WARNING] 
+[2025-01-02 08:31:53,053] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 08:31:53,053] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-02 08:31:53,053] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 08:31:53,056] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-02 08:31:53,056] torch.distributed.run: [WARNING] 
+[2025-01-02 08:31:53,056] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 08:31:53,056] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-02 08:31:53,056] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 08:31:53,068] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-02 08:31:53,068] torch.distributed.run: [WARNING] 
+[2025-01-02 08:31:53,068] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 08:31:53,068] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-02 08:31:53,068] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 08:31:53,093] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-02 08:31:53,093] torch.distributed.run: [WARNING] 
+[2025-01-02 08:31:53,093] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 08:31:53,093] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-02 08:31:53,093] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 08:31:53,096] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-02 08:31:53,096] torch.distributed.run: [WARNING] 
+[2025-01-02 08:31:53,096] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 08:31:53,096] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-02 08:31:53,096] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 08:31:53,102] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-02 08:31:53,102] torch.distributed.run: [WARNING] 
+[2025-01-02 08:31:53,102] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 08:31:53,102] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-02 08:31:53,102] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 08:31:53,105] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-02 08:31:53,105] torch.distributed.run: [WARNING] 
+[2025-01-02 08:31:53,105] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 08:31:53,105] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-02 08:31:53,105] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 08:31:53,132] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-02 08:31:53,132] torch.distributed.run: [WARNING] 
+[2025-01-02 08:31:53,132] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 08:31:53,132] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-02 08:31:53,132] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 08:31:53,163] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-02 08:31:53,163] torch.distributed.run: [WARNING] 
+[2025-01-02 08:31:53,163] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 08:31:53,163] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-02 08:31:53,163] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 08:31:53,333] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-02 08:31:53,333] torch.distributed.run: [WARNING] 
+[2025-01-02 08:31:53,333] torch.distributed.run: [WARNING] *****************************************
+[2025-01-02 08:31:53,333] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-02 08:31:53,333] torch.distributed.run: [WARNING] *****************************************
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+01/02/2025 08:33:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]: Config:
+01/02/2025 08:33:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]: Config(general=GeneralArgs(project='debug',
+01/02/2025 08:33:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                            run='8.86G_dp64_tp8_pp1_acc2_mbs2_seq4096_zero0_tpmodeRED_vocab131k',
+01/02/2025 08:33:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                            seed=42,
+01/02/2025 08:33:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                            step=None,
+01/02/2025 08:33:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                            consumed_train_samples=None,
+01/02/2025 08:33:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                            benchmark_csv_path=PosixPath('benchmark/results/bench_final2.csv'),
+01/02/2025 08:33:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                            ignore_sanity_checks=True),
+01/02/2025 08:33:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:        parallelism=ParallelismArgs(dp=64,
+01/02/2025 08:33:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                                    pp=1,
+01/02/2025 08:33:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                                    tp=8,
+01/02/2025 08:33:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                                    pp_engine=<nanotron.parallel.pipeline_parallel.engine.OneForwardOneBackwardPipelineEngine object at 0x7fb556d4cdc0>,
+01/02/2025 08:33:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                                    tp_mode=<TensorParallelLinearMode.REDUCE_SCATTER: 2>,
+01/02/2025 08:33:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                                    tp_linear_async_communication=True,
+01/02/2025 08:33:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                                    recompute_layer=False,
+01/02/2025 08:33:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                                    tp_recompute_allgather=True,
+01/02/2025 08:33:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                                    expert_parallel_size=1),
+01/02/2025 08:33:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:        model=ModelArgs(model_config=LlamaConfig(bos_token_id=0,
+01/02/2025 08:33:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                                                 eos_token_id=0,
+01/02/2025 08:33:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                                                 hidden_act='silu',
+01/02/2025 08:33:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                                                 hidden_size=4096,
+01/02/2025 08:33:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                                                 initializer_range=0.02,
+01/02/2025 08:33:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                                                 intermediate_size=14336,
+01/02/2025 08:33:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                                                 is_llama_config=True,
+01/02/2025 08:33:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                                                 max_position_embeddings=4096,
+01/02/2025 08:33:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                                                 num_attention_heads=32,
+01/02/2025 08:33:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                                                 num_hidden_layers=32,
+01/02/2025 08:33:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                                                 num_key_value_heads=32,
+01/02/2025 08:33:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                                                 pad_token_id=None,
+01/02/2025 08:33:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                                                 pretraining_tp=1,
+01/02/2025 08:33:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                                                 rms_norm_eps=1e-05,
+01/02/2025 08:33:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                                                 rope_scaling=None,
+01/02/2025 08:33:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                                                 rope_theta=10000.0,
+01/02/2025 08:33:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                                                 rope_interleaved=False,
+01/02/2025 08:33:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                                                 tie_word_embeddings=False,
+01/02/2025 08:33:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                                                 use_cache=True,
+01/02/2025 08:33:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                                                 vocab_size=131072),
+01/02/2025 08:33:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                        init_method=RandomInit(std=0.02),
+01/02/2025 08:33:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                        dtype=torch.bfloat16,
+01/02/2025 08:33:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                        make_vocab_size_divisible_by=1,
+01/02/2025 08:33:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                        ddp_bucket_cap_mb=25),
+01/02/2025 08:33:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:        tokenizer=TokenizerArgs(tokenizer_name_or_path='robot-test/dummy-tokenizer-wordlevel',
+01/02/2025 08:33:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                                tokenizer_revision=None,
+01/02/2025 08:33:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                                tokenizer_max_length=None),
+01/02/2025 08:33:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:        checkpoints=CheckpointsArgs(checkpoints_path=PosixPath('checkpoints'),
+01/02/2025 08:33:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                                    checkpoint_interval=10000,
+01/02/2025 08:33:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                                    save_initial_state=False,
+01/02/2025 08:33:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                                    save_final_state=False,
+01/02/2025 08:33:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                                    resume_checkpoint_path=None,
+01/02/2025 08:33:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                                    checkpoints_path_is_shared_file_system=False),
+01/02/2025 08:33:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:        logging=LoggingArgs(log_level='info',
+01/02/2025 08:33:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                            log_level_replica='info',
+01/02/2025 08:33:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                            iteration_step_info_interval=1),
+01/02/2025 08:33:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:        tokens=TokensArgs(sequence_length=4096,
+01/02/2025 08:33:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                          train_steps=100,
+01/02/2025 08:33:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                          micro_batch_size=2,
+01/02/2025 08:33:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                          batch_accumulation_per_replica=2,
+01/02/2025 08:33:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                          val_check_interval=100,
+01/02/2025 08:33:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                          limit_val_batches=0,
+01/02/2025 08:33:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                          limit_test_batches=0),
+01/02/2025 08:33:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:        optimizer=OptimizerArgs(optimizer_factory=AdamWOptimizerArgs(adam_eps=1e-08,
+01/02/2025 08:33:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                                                                     adam_beta1=0.9,
+01/02/2025 08:33:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                                                                     adam_beta2=0.95,
+01/02/2025 08:33:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                                                                     torch_adam_is_fused=True,
+01/02/2025 08:33:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                                                                     name='adamW'),
+01/02/2025 08:33:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                                zero_stage=0,
+01/02/2025 08:33:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                                weight_decay=0.01,
+01/02/2025 08:33:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                                clip_grad=1.0,
+01/02/2025 08:33:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                                accumulate_grad_in_fp32=True,
+01/02/2025 08:33:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                                learning_rate_scheduler=LRSchedulerArgs(learning_rate=0.0003,
+01/02/2025 08:33:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                                                                        lr_warmup_steps=2,
+01/02/2025 08:33:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                                                                        lr_warmup_style='linear',
+01/02/2025 08:33:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                                                                        lr_decay_style='cosine',
+01/02/2025 08:33:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                                                                        lr_decay_steps=13,
+01/02/2025 08:33:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                                                                        lr_decay_starting_step=None,
+01/02/2025 08:33:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                                                                        min_decay_lr=1e-05)),
+01/02/2025 08:33:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:        data_stages=[DatasetStageArgs(name='Stable Training Stage',
+01/02/2025 08:33:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                                      start_training_step=1,
+01/02/2025 08:33:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                                      data=DataArgs(dataset=None,
+01/02/2025 08:33:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                                                    seed=42,
+01/02/2025 08:33:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:                                                    num_loading_workers=1))],
+01/02/2025 08:33:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:        profiler=None,
+01/02/2025 08:33:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:        lighteval=None,
+01/02/2025 08:33:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:        s3_upload=None)
+01/02/2025 08:33:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]: Model Config:
+01/02/2025 08:33:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]: LlamaConfig(bos_token_id=0,
+01/02/2025 08:33:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:             eos_token_id=0,
+01/02/2025 08:33:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:             hidden_act='silu',
+01/02/2025 08:33:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:             hidden_size=4096,
+01/02/2025 08:33:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:             initializer_range=0.02,
+01/02/2025 08:33:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:             intermediate_size=14336,
+01/02/2025 08:33:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:             is_llama_config=True,
+01/02/2025 08:33:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:             max_position_embeddings=4096,
+01/02/2025 08:33:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:             num_attention_heads=32,
+01/02/2025 08:33:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:             num_hidden_layers=32,
+01/02/2025 08:33:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:             num_key_value_heads=32,
+01/02/2025 08:33:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:             pad_token_id=None,
+01/02/2025 08:33:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:             pretraining_tp=1,
+01/02/2025 08:33:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:             rms_norm_eps=1e-05,
+01/02/2025 08:33:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:             rope_scaling=None,
+01/02/2025 08:33:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:             rope_theta=10000.0,
+01/02/2025 08:33:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:             rope_interleaved=False,
+01/02/2025 08:33:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:             tie_word_embeddings=False,
+01/02/2025 08:33:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:             use_cache=True,
+01/02/2025 08:33:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:             vocab_size=131072)
+01/02/2025 08:33:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]: Building model..
+01/02/2025 08:33:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]: Initialize RoPE Theta = 10000.0
+01/02/2025 08:33:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]: Setting PP block ranks...
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+01/02/2025 08:33:35 [INFO|DP=0|PP=0|TP=7|ip-26-0-161-103]: Local number of parameters: 1.11G (2112.51MiB)
+01/02/2025 08:33:35 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]: Total number of parameters: 8.86G (16900.06MiB)
+01/02/2025 08:33:35 [INFO|DP=0|PP=0|TP=5|ip-26-0-161-103]: Local number of parameters: 1.11G (2112.51MiB)
+01/02/2025 08:33:35 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]: Local number of parameters: 1.11G (2112.51MiB)
+01/02/2025 08:33:35 [INFO|DP=0|PP=0|TP=6|ip-26-0-161-103]: Local number of parameters: 1.11G (2112.51MiB)
+01/02/2025 08:33:35 [INFO|DP=0|PP=0|TP=4|ip-26-0-161-103]: Local number of parameters: 1.11G (2112.51MiB)
+01/02/2025 08:33:35 [INFO|DP=0|PP=0|TP=1|ip-26-0-161-103]: Local number of parameters: 1.11G (2112.51MiB)
+01/02/2025 08:33:35 [INFO|DP=0|PP=0|TP=7|ip-26-0-161-103]: [After model building] Memory usage: 2112.55MiB. Peak allocated: 5380.00MiB Peak reserved: 10518.00MiB
+01/02/2025 08:33:35 [INFO|DP=0|PP=0|TP=5|ip-26-0-161-103]: [After model building] Memory usage: 2112.55MiB. Peak allocated: 5380.00MiB Peak reserved: 10518.00MiB
+01/02/2025 08:33:35 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]: [After model building] Memory usage: 2112.55MiB. Peak allocated: 5380.00MiB Peak reserved: 10518.00MiB
+01/02/2025 08:33:35 [INFO|DP=0|PP=0|TP=4|ip-26-0-161-103]: [After model building] Memory usage: 2112.55MiB. Peak allocated: 5380.00MiB Peak reserved: 10518.00MiB
+01/02/2025 08:33:35 [INFO|DP=0|PP=0|TP=2|ip-26-0-161-103]: Local number of parameters: 1.11G (2112.51MiB)
+01/02/2025 08:33:35 [INFO|DP=0|PP=0|TP=3|ip-26-0-161-103]: Local number of parameters: 1.11G (2112.51MiB)
+01/02/2025 08:33:35 [INFO|DP=0|PP=0|TP=1|ip-26-0-161-103]: [After model building] Memory usage: 2112.55MiB. Peak allocated: 5380.00MiB Peak reserved: 10518.00MiB
+01/02/2025 08:33:35 [INFO|DP=0|PP=0|TP=6|ip-26-0-161-103]: [After model building] Memory usage: 2112.55MiB. Peak allocated: 5380.00MiB Peak reserved: 10518.00MiB
+01/02/2025 08:33:35 [INFO|DP=0|PP=0|TP=2|ip-26-0-161-103]: [After model building] Memory usage: 2112.55MiB. Peak allocated: 5380.00MiB Peak reserved: 10518.00MiB
+01/02/2025 08:33:35 [INFO|DP=0|PP=0|TP=3|ip-26-0-161-103]: [After model building] Memory usage: 2112.55MiB. Peak allocated: 5380.00MiB Peak reserved: 10518.00MiB
+01/02/2025 08:33:37 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]: No checkpoint path provided.
+01/02/2025 08:33:37 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]: Parametrizing model parameters using StandardParametrizator
+01/02/2025 08:33:42 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]: [Optimizer Building] Using LearningRateForSP as learning rate
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    trainer.train(dataloader)
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return f(*args, **kwargs)
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    trainer.train(dataloader)
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    model.p2p.clear_history()
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return f(*args, **kwargs)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    trainer.train(dataloader)
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    return f(*args, **kwargs)
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs = self.pipeline_engine.train_batch_iter(
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    outputs = self.pipeline_engine.train_batch_iter(
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    model.p2p.clear_history()
+    model.p2p.clear_history()  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    model.p2p.clear_history()
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    outputs = self.pipeline_engine.train_batch_iter(
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    model.p2p.clear_history()
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    model.p2p.clear_history()
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    trainer.train(dataloader)
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    trainer.train(dataloader)
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    model.p2p.clear_history()
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+Traceback (most recent call last):
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    trainer.train(dataloader)
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    trainer.train(dataloader)
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return f(*args, **kwargs)
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    outputs = self.pipeline_engine.train_batch_iter(
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    trainer.train(dataloader)
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+Traceback (most recent call last):
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    model.p2p.clear_history()
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    return f(*args, **kwargs)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    outputs = self.pipeline_engine.train_batch_iter(
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    model.p2p.clear_history()
+        return f(*args, **kwargs)outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    model.p2p.clear_history()
+    outputs = self.pipeline_engine.train_batch_iter(
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    model.p2p.clear_history()
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    model.p2p.clear_history()
+    trainer.train(dataloader)
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+Traceback (most recent call last):
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+
+    return f(*args, **kwargs)
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)    
+    outputs = self.pipeline_engine.train_batch_iter(
+trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    model.p2p.clear_history()
+    outputs = self.pipeline_engine.train_batch_iter(
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    trainer.train(dataloader)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    outputs = self.pipeline_engine.train_batch_iter(
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    return f(*args, **kwargs)
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    model.p2p.clear_history()
+Traceback (most recent call last):
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+Traceback (most recent call last):
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return f(*args, **kwargs)
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs = self.pipeline_engine.train_batch_iter(
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    model.p2p.clear_history()
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    return f(*args, **kwargs)
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    model.p2p.clear_history()
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return f(*args, **kwargs)
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    outputs = self.pipeline_engine.train_batch_iter(
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+Traceback (most recent call last):
+    model.p2p.clear_history()
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    return f(*args, **kwargs)
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+Traceback (most recent call last):
+    trainer.train(dataloader)
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return f(*args, **kwargs)
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+Traceback (most recent call last):
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    trainer.train(dataloader)
+    return f(*args, **kwargs)
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    trainer.train(dataloader)
+Traceback (most recent call last):
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+Traceback (most recent call last):
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return f(*args, **kwargs)
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'    model.p2p.clear_history()
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+Traceback (most recent call last):
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+        return f(*args, **kwargs)outputs = self.pipeline_engine.train_batch_iter(
+
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    trainer.train(dataloader)
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    return f(*args, **kwargs)AttributeError
+: 'DistributedDataParallel' object has no attribute 'p2p'  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs = self.pipeline_engine.train_batch_iter(
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+Traceback (most recent call last):
+    trainer.train(dataloader)
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    trainer.train(dataloader)
+Traceback (most recent call last):
+    model.p2p.clear_history()
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return f(*args, **kwargs)
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs = self.pipeline_engine.train_batch_iter(
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    return f(*args, **kwargs)
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    model.p2p.clear_history()
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'    
+model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    model.p2p.clear_history()
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+Traceback (most recent call last):
+    outputs = self.pipeline_engine.train_batch_iter(
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    outputs = self.pipeline_engine.train_batch_iter(
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+Traceback (most recent call last):
+    model.p2p.clear_history()
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    outputs = self.pipeline_engine.train_batch_iter(
+    model.p2p.clear_history()
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+Traceback (most recent call last):
+    trainer.train(dataloader)
+    return f(*args, **kwargs)
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    model.p2p.clear_history()
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+Traceback (most recent call last):
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+Traceback (most recent call last):
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    outputs = self.pipeline_engine.train_batch_iter(
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    trainer.train(dataloader)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    return f(*args, **kwargs)
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    model.p2p.clear_history()
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    outputs = self.pipeline_engine.train_batch_iter(
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    model.p2p.clear_history()
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    return f(*args, **kwargs)
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+Traceback (most recent call last):
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    outputs = self.pipeline_engine.train_batch_iter(
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    model.p2p.clear_history()
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    trainer.train(dataloader)
+    trainer.train(dataloader)
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+Traceback (most recent call last):
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return f(*args, **kwargs)
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    trainer.train(dataloader)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+Traceback (most recent call last):
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return f(*args, **kwargs)
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    trainer.train(dataloader)
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+Traceback (most recent call last):
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    outputs = self.pipeline_engine.train_batch_iter(
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    model.p2p.clear_history()
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    model.p2p.clear_history()
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    model.p2p.clear_history()
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    outputs = self.pipeline_engine.train_batch_iter(
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    return f(*args, **kwargs)
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+Traceback (most recent call last):
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    model.p2p.clear_history()
+
+    trainer.train(dataloader)
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+Traceback (most recent call last):
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    outputs = self.pipeline_engine.train_batch_iter(
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    return f(*args, **kwargs)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    trainer.train(dataloader)
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    model.p2p.clear_history()
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    return f(*args, **kwargs)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+Traceback (most recent call last):
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+Traceback (most recent call last):
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    trainer.train(dataloader)
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+        raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")return f(*args, **kwargs)
+
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    trainer.train(dataloader)
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+        outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)trainer.train(dataloader)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    return f(*args, **kwargs)
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    model.p2p.clear_history()
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    return f(*args, **kwargs)
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    trainer.train(dataloader)
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    model.p2p.clear_history()
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return f(*args, **kwargs)
+    model.p2p.clear_history()
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+Traceback (most recent call last):
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    return f(*args, **kwargs)
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    trainer.train(dataloader)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    model.p2p.clear_history()
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    model.p2p.clear_history()
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+Traceback (most recent call last):
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return f(*args, **kwargs)
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    trainer.train(dataloader)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    model.p2p.clear_history()
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs = self.pipeline_engine.train_batch_iter(
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+    model.p2p.clear_history()
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    trainer.train(dataloader)
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    return f(*args, **kwargs)
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+Traceback (most recent call last):
+    model.p2p.clear_history()
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    model.p2p.clear_history()
+    return f(*args, **kwargs)
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+Traceback (most recent call last):
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    trainer.train(dataloader)
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+Traceback (most recent call last):
+AttributeError:     'DistributedDataParallel' object has no attribute 'p2p'trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+
+    return f(*args, **kwargs)
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return f(*args, **kwargs)
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    trainer.train(dataloader)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    model.p2p.clear_history()
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return f(*args, **kwargs)
+    outputs = self.pipeline_engine.train_batch_iter(
+Traceback (most recent call last):
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    outputs = self.pipeline_engine.train_batch_iter(
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+Traceback (most recent call last):
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    model.p2p.clear_history()
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    model.p2p.clear_history()
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    trainer.train(dataloader)
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    trainer.train(dataloader)
+    outputs = self.pipeline_engine.train_batch_iter(
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    return f(*args, **kwargs)
+    trainer.train(dataloader)
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    return f(*args, **kwargs)
+    model.p2p.clear_history()
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+Traceback (most recent call last):
+Traceback (most recent call last):
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    trainer.train(dataloader)
+    model.p2p.clear_history()
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    outputs = self.pipeline_engine.train_batch_iter(
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    model.p2p.clear_history()
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return f(*args, **kwargs)
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    model.p2p.clear_history()
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    trainer.train(dataloader)
+    trainer.train(dataloader)
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    return f(*args, **kwargs)
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return f(*args, **kwargs)
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    outputs = self.pipeline_engine.train_batch_iter(
+    outputs = self.pipeline_engine.train_batch_iter(
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    model.p2p.clear_history()
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    model.p2p.clear_history()
+    model.p2p.clear_history()
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+Traceback (most recent call last):
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    trainer.train(dataloader)
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    return f(*args, **kwargs)
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs = self.pipeline_engine.train_batch_iter(
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+Traceback (most recent call last):
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs = self.pipeline_engine.train_batch_iter(
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    outputs = self.pipeline_engine.train_batch_iter(
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    model.p2p.clear_history()
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+Traceback (most recent call last):
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    model.p2p.clear_history()
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+Traceback (most recent call last):
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    trainer.train(dataloader)
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    trainer.train(dataloader)
+    trainer.train(dataloader)    
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    model.p2p.clear_history()
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return f(*args, **kwargs)
+    model.p2p.clear_history()
+Traceback (most recent call last):
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return f(*args, **kwargs)
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    trainer.train(dataloader)
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    trainer.train(dataloader)
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return f(*args, **kwargs)
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    outputs = self.pipeline_engine.train_batch_iter(
+    model.p2p.clear_history()
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return f(*args, **kwargs)
+    outputs = self.pipeline_engine.train_batch_iter(
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    model.p2p.clear_history()
+    model.p2p.clear_history()
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+        return f(*args, **kwargs)raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    outputs = self.pipeline_engine.train_batch_iter(
+    model.p2p.clear_history()
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    outputs = self.pipeline_engine.train_batch_iter(
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    model.p2p.clear_history()
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+Traceback (most recent call last):
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    trainer.train(dataloader)
+    trainer.train(dataloader)
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs = self.pipeline_engine.train_batch_iter(
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    model.p2p.clear_history()
+    return f(*args, **kwargs)
+    return f(*args, **kwargs)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    model.p2p.clear_history()
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    return f(*args, **kwargs)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    outputs = self.pipeline_engine.train_batch_iter(
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs = self.pipeline_engine.train_batch_iter(
+    trainer.train(dataloader)
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    trainer.train(dataloader)
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    model.p2p.clear_history()
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return f(*args, **kwargs)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    trainer.train(dataloader)
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+Traceback (most recent call last):
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+Traceback (most recent call last):
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    outputs = self.pipeline_engine.train_batch_iter(
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    model.p2p.clear_history()
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+Traceback (most recent call last):
+    trainer.train(dataloader)
+    outputs = self.pipeline_engine.train_batch_iter(
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    model.p2p.clear_history()
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    return f(*args, **kwargs)
+    model.p2p.clear_history()
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+Traceback (most recent call last):
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'Traceback (most recent call last):
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+
+
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    return f(*args, **kwargs)
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    trainer.train(dataloader)
+    trainer.train(dataloader)
+    outputs = self.pipeline_engine.train_batch_iter(
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+Traceback (most recent call last):
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    model.p2p.clear_history()
+Traceback (most recent call last):
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return f(*args, **kwargs)
+    model.p2p.clear_history()
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    trainer.train(dataloader)
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return f(*args, **kwargs)
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    trainer.train(dataloader)
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    model.p2p.clear_history()
+Traceback (most recent call last):
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+Traceback (most recent call last):
+    return f(*args, **kwargs)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+01/02/2025 08:33:43 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]: module_name: model.token_position_embeddings | PP: 0/1 | Block rank: 0
+01/02/2025 08:33:43 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]: module_name: model.decoder.0 | PP: 0/1 | Block rank: 1
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+01/02/2025 08:33:43 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]: module_name: model.decoder.1 | PP: 0/1 | Block rank: 2
+    outputs = self.pipeline_engine.train_batch_iter(
+01/02/2025 08:33:43 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]: module_name: model.decoder.2 | PP: 0/1 | Block rank: 3
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+01/02/2025 08:33:43 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]: module_name: model.decoder.3 | PP: 0/1 | Block rank: 4
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+01/02/2025 08:33:43 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]: module_name: model.decoder.4 | PP: 0/1 | Block rank: 5
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+01/02/2025 08:33:43 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]: module_name: model.decoder.5 | PP: 0/1 | Block rank: 6
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+01/02/2025 08:33:43 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]: module_name: model.decoder.6 | PP: 0/1 | Block rank: 7
+        model.p2p.clear_history()outputs = self.pipeline_engine.train_batch_iter(
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+01/02/2025 08:33:43 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]: module_name: model.decoder.7 | PP: 0/1 | Block rank: 8
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+01/02/2025 08:33:43 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]: module_name: model.decoder.8 | PP: 0/1 | Block rank: 9
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+01/02/2025 08:33:43 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]: module_name: model.decoder.9 | PP: 0/1 | Block rank: 10
+
+Traceback (most recent call last):
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+01/02/2025 08:33:43 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]: module_name: model.decoder.10 | PP: 0/1 | Block rank: 11
+01/02/2025 08:33:43 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]: module_name: model.decoder.11 | PP: 0/1 | Block rank: 12
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+01/02/2025 08:33:43 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]: module_name: model.decoder.12 | PP: 0/1 | Block rank: 13
+01/02/2025 08:33:43 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]: module_name: model.decoder.13 | PP: 0/1 | Block rank: 14
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return f(*args, **kwargs)
+01/02/2025 08:33:43 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]: module_name: model.decoder.14 | PP: 0/1 | Block rank: 15
+01/02/2025 08:33:43 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]: module_name: model.decoder.15 | PP: 0/1 | Block rank: 16
+01/02/2025 08:33:43 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]: module_name: model.decoder.16 | PP: 0/1 | Block rank: 17
+01/02/2025 08:33:43 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]: module_name: model.decoder.17 | PP: 0/1 | Block rank: 18
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    model.p2p.clear_history()
+01/02/2025 08:33:43 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]: module_name: model.decoder.18 | PP: 0/1 | Block rank: 19
+01/02/2025 08:33:43 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]: module_name: model.decoder.19 | PP: 0/1 | Block rank: 20
+01/02/2025 08:33:43 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]: module_name: model.decoder.20 | PP: 0/1 | Block rank: 21
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    outputs = self.pipeline_engine.train_batch_iter(
+01/02/2025 08:33:43 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]: module_name: model.decoder.21 | PP: 0/1 | Block rank: 22
+01/02/2025 08:33:43 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]: module_name: model.decoder.22 | PP: 0/1 | Block rank: 23
+01/02/2025 08:33:43 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]: module_name: model.decoder.23 | PP: 0/1 | Block rank: 24
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+Traceback (most recent call last):
+01/02/2025 08:33:43 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]: module_name: model.decoder.24 | PP: 0/1 | Block rank: 25
+01/02/2025 08:33:43 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]: module_name: model.decoder.25 | PP: 0/1 | Block rank: 26
+01/02/2025 08:33:43 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]: module_name: model.decoder.26 | PP: 0/1 | Block rank: 27
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+01/02/2025 08:33:43 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]: module_name: model.decoder.27 | PP: 0/1 | Block rank: 28
+01/02/2025 08:33:43 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]: module_name: model.decoder.28 | PP: 0/1 | Block rank: 29
+01/02/2025 08:33:43 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]: module_name: model.decoder.29 | PP: 0/1 | Block rank: 30
+    model.p2p.clear_history()
+Traceback (most recent call last):
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+01/02/2025 08:33:43 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]: module_name: model.decoder.30 | PP: 0/1 | Block rank: 31
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+01/02/2025 08:33:43 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]: module_name: model.decoder.31 | PP: 0/1 | Block rank: 32
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+01/02/2025 08:33:43 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]: module_name: model.final_layer_norm | PP: 0/1 | Block rank: 33
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+01/02/2025 08:33:43 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]: module_name: model.lm_head | PP: 0/1 | Block rank: 34
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+01/02/2025 08:33:43 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]: module_name: model.cast_to_fp32 | PP: 0/1 | Block rank: 35
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+01/02/2025 08:33:43 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]: module_name: loss | PP: 0/1 | Block rank: 36
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+01/02/2025 08:33:43 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]: [Training Plan] Stage Stable Training Stage has 99 remaining training steps and has consumed 0 samples
+    model.p2p.clear_history()
+01/02/2025 08:33:43 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]: Using dummy data generator
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+01/02/2025 08:33:43 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]: [Training Plan] There are 1 training stages 
+01/02/2025 08:33:43 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]: [Stage Stable Training Stage] start from step 1 
+01/02/2025 08:33:43 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]: 
+    trainer.train(dataloader)
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+01/02/2025 08:33:43 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]: [Start training] datetime: 2025-01-02 08:33:43.120292 | mbs: 2 | grad_accum: 2 | global_batch_size: 256 | sequence_length: 4096 | train_steps: 100 | start_iteration_step: 0 | consumed_train_samples: 0
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs = self.pipeline_engine.train_batch_iter(
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    Traceback (most recent call last):
+outputs = self.pipeline_engine.train_batch_iter(
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    return f(*args, **kwargs)
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    model.p2p.clear_history()
+Traceback (most recent call last):
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return f(*args, **kwargs)
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    trainer.train(dataloader)
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    trainer.train(dataloader)
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    model.p2p.clear_history()
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    return f(*args, **kwargs)
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+
+    model.p2p.clear_history()
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+Traceback (most recent call last):
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    model.p2p.clear_history()
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    model.p2p.clear_history()
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+Traceback (most recent call last):
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+Traceback (most recent call last):
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+Traceback (most recent call last):
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+Traceback (most recent call last):
+    trainer.train(dataloader)
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    model.p2p.clear_history()
+    trainer.train(dataloader)
+    trainer.train(dataloader)
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+Traceback (most recent call last):
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    trainer.train(dataloader)
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    return f(*args, **kwargs)
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    model.p2p.clear_history()
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    return f(*args, **kwargs)
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+
+    trainer.train(dataloader)
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    return f(*args, **kwargs)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+Traceback (most recent call last):
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return f(*args, **kwargs)
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+Traceback (most recent call last):
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs = self.pipeline_engine.train_batch_iter(
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    trainer.train(dataloader)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+Traceback (most recent call last):
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    model.p2p.clear_history()
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    model.p2p.clear_history()
+    outputs = self.pipeline_engine.train_batch_iter(
+    return f(*args, **kwargs)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    outputs = self.pipeline_engine.train_batch_iter(
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+Traceback (most recent call last):
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    outputs = self.pipeline_engine.train_batch_iter(
+    trainer.train(dataloader)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    model.p2p.clear_history()
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    model.p2p.clear_history()
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+
+    return f(*args, **kwargs)
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'    trainer.train(dataloader)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'    model.p2p.clear_history()
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+
+    return f(*args, **kwargs)
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs = self.pipeline_engine.train_batch_iter(
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+Traceback (most recent call last):
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+Traceback (most recent call last):
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return f(*args, **kwargs)
+
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+Traceback (most recent call last):
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    outputs = self.pipeline_engine.train_batch_iter(
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    trainer.train(dataloader)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+    model.p2p.clear_history()
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    trainer.train(dataloader)
+    return f(*args, **kwargs)
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return f(*args, **kwargs)
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    model.p2p.clear_history()
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    outputs = self.pipeline_engine.train_batch_iter(
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    trainer.train(dataloader)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    model.p2p.clear_history()
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    outputs = self.pipeline_engine.train_batch_iter(
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs = self.pipeline_engine.train_batch_iter(
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    return f(*args, **kwargs)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    model.p2p.clear_history()
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    model.p2p.clear_history()
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    model.p2p.clear_history()
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    model.p2p.clear_history()
+    outputs = self.pipeline_engine.train_batch_iter(
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    outputs = self.pipeline_engine.train_batch_iter(
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    model.p2p.clear_history()
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    model.p2p.clear_history()
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    trainer.train(dataloader)
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    trainer.train(dataloader)
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs = self.pipeline_engine.train_batch_iter(
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    trainer.train(dataloader)
+Traceback (most recent call last):
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+Traceback (most recent call last):
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    return f(*args, **kwargs)
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    return f(*args, **kwargs)
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    trainer.train(dataloader)
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    model.p2p.clear_history()
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    return f(*args, **kwargs)
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    return f(*args, **kwargs)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+Traceback (most recent call last):
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    trainer.train(dataloader)
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    return f(*args, **kwargs)
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return f(*args, **kwargs)
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    trainer.train(dataloader)
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    trainer.train(dataloader)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return f(*args, **kwargs)
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+Traceback (most recent call last):
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+Traceback (most recent call last):
+    model.p2p.clear_history()
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+    outputs = self.pipeline_engine.train_batch_iter(
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    trainer.train(dataloader)
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+Traceback (most recent call last):
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    model.p2p.clear_history()
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+Traceback (most recent call last):
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+Traceback (most recent call last):
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    trainer.train(dataloader)
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+        trainer.train(dataloader)raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    trainer.train(dataloader)
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return f(*args, **kwargs)
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+Traceback (most recent call last):
+    model.p2p.clear_history()
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    outputs = self.pipeline_engine.train_batch_iter(
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    model.p2p.clear_history()
+    outputs = self.pipeline_engine.train_batch_iter(
+    trainer.train(dataloader)
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    model.p2p.clear_history()
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    outputs = self.pipeline_engine.train_batch_iter(
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    trainer.train(dataloader)
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    model.p2p.clear_history()
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    return f(*args, **kwargs)
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    outputs = self.pipeline_engine.train_batch_iter(
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    outputs = self.pipeline_engine.train_batch_iter(
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    model.p2p.clear_history()
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+Traceback (most recent call last):
+    model.p2p.clear_history()
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+Traceback (most recent call last):
+    model.p2p.clear_history()
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    outputs = self.pipeline_engine.train_batch_iter(
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    model.p2p.clear_history()
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return f(*args, **kwargs)
+Traceback (most recent call last):
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    trainer.train(dataloader)
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    trainer.train(dataloader)
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return f(*args, **kwargs)
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    trainer.train(dataloader)
+    trainer.train(dataloader)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+Traceback (most recent call last):
+    return f(*args, **kwargs)
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+Traceback (most recent call last):
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    model.p2p.clear_history()
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+        model.p2p.clear_history()return f(*args, **kwargs)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs = self.pipeline_engine.train_batch_iter(
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    return f(*args, **kwargs)
+    trainer.train(dataloader)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    trainer.train(dataloader)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    model.p2p.clear_history()
+Traceback (most recent call last):
+    model.p2p.clear_history()
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+Traceback (most recent call last):
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return f(*args, **kwargs)
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    model.p2p.clear_history()
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    trainer.train(dataloader)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    trainer.train(dataloader)
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    model.p2p.clear_history()
+    outputs = self.pipeline_engine.train_batch_iter(
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+
+    model.p2p.clear_history()
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    model.p2p.clear_history()
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    outputs = self.pipeline_engine.train_batch_iter(
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+Traceback (most recent call last):
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    trainer.train(dataloader)
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return f(*args, **kwargs)
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    outputs = self.pipeline_engine.train_batch_iter(
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    trainer.train(dataloader)
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+Traceback (most recent call last):
+    model.p2p.clear_history()
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    return f(*args, **kwargs)
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return f(*args, **kwargs)
+    model.p2p.clear_history()
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+Traceback (most recent call last):
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    trainer.train(dataloader)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    trainer.train(dataloader)
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    outputs = self.pipeline_engine.train_batch_iter(
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    model.p2p.clear_history()
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    model.p2p.clear_history()
+    trainer.train(dataloader)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    model.p2p.clear_history()
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+Traceback (most recent call last):
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    trainer.train(dataloader)
+    outputs = self.pipeline_engine.train_batch_iter(
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    trainer.train(dataloader)
+    model.p2p.clear_history()
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    model.p2p.clear_history()
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return f(*args, **kwargs)
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    outputs = self.pipeline_engine.train_batch_iter(
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    trainer.train(dataloader)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    model.p2p.clear_history()
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    outputs = self.pipeline_engine.train_batch_iter(
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    model.p2p.clear_history()
+Traceback (most recent call last):
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+Traceback (most recent call last):
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+Traceback (most recent call last):
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    trainer.train(dataloader)
+    trainer.train(dataloader)
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    model.p2p.clear_history()
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    model.p2p.clear_history()
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return f(*args, **kwargs)
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    model.p2p.clear_history()
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    trainer.train(dataloader)
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return f(*args, **kwargs)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return f(*args, **kwargs)
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    model.p2p.clear_history()
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+Traceback (most recent call last):
+    model.p2p.clear_history()
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    trainer.train(dataloader)
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    model.p2p.clear_history()
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    trainer.train(dataloader)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+Traceback (most recent call last):
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+Traceback (most recent call last):
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    trainer.train(dataloader)
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    trainer.train(dataloader)
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+Traceback (most recent call last):
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    trainer.train(dataloader)
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+Traceback (most recent call last):
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    trainer.train(dataloader)
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+        trainer.train(dataloader)model.p2p.clear_history()
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")    
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return f(*args, **kwargs)
+raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+AttributeErrorAttributeError: : 'DistributedDataParallel' object has no attribute 'p2p''DistributedDataParallel' object has no attribute 'p2p'
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+
+    return f(*args, **kwargs)
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+Traceback (most recent call last):
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    model.p2p.clear_history()
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs = self.pipeline_engine.train_batch_iter(
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    model.p2p.clear_history()
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    model.p2p.clear_history()
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    outputs = self.pipeline_engine.train_batch_iter(
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'    trainer.train(dataloader)
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+Traceback (most recent call last):
+
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+        trainer.train(dataloader)outputs = self.pipeline_engine.train_batch_iter(
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+Traceback (most recent call last):
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+        model.p2p.clear_history()outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+Traceback (most recent call last):
+Traceback (most recent call last):
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+
+    trainer.train(dataloader)
+    model.p2p.clear_history()
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    trainer.train(dataloader)
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    return f(*args, **kwargs)
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    outputs = self.pipeline_engine.train_batch_iter(
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    model.p2p.clear_history()
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    model.p2p.clear_history()
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+Traceback (most recent call last):
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    outputs = self.pipeline_engine.train_batch_iter(
+Traceback (most recent call last):
+    return f(*args, **kwargs)
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return f(*args, **kwargs)
+    model.p2p.clear_history()
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    model.p2p.clear_history()
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return f(*args, **kwargs)
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    outputs = self.pipeline_engine.train_batch_iter(
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    trainer.train(dataloader)
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    AttributeErroroutputs = self.pipeline_engine.train_batch_iter(: 
+'DistributedDataParallel' object has no attribute 'p2p'  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    return f(*args, **kwargs)
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs = self.pipeline_engine.train_batch_iter(
+    model.p2p.clear_history()
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    model.p2p.clear_history()
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    model.p2p.clear_history()
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'Traceback (most recent call last):
+
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    outputs = self.pipeline_engine.train_batch_iter(
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    trainer.train(dataloader)
+Traceback (most recent call last):
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    outputs = self.pipeline_engine.train_batch_iter(
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    trainer.train(dataloader)
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    trainer.train(dataloader)
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+
+Traceback (most recent call last):
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    model.p2p.clear_history()
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    outputs = self.pipeline_engine.train_batch_iter(
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    return f(*args, **kwargs)
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    trainer.train(dataloader)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs = self.pipeline_engine.train_batch_iter(
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    return f(*args, **kwargs)
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+Traceback (most recent call last):
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    trainer.train(dataloader)
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs = self.pipeline_engine.train_batch_iter(
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    model.p2p.clear_history()
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    model.p2p.clear_history()
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    model.p2p.clear_history()
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    outputs = self.pipeline_engine.train_batch_iter(
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return f(*args, **kwargs)
+    model.p2p.clear_history()
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    trainer.train(dataloader)
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    return f(*args, **kwargs)
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    model.p2p.clear_history()
+Traceback (most recent call last):
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    trainer.train(dataloader)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    outputs = self.pipeline_engine.train_batch_iter(
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    model.p2p.clear_history()
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+Traceback (most recent call last):
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'    trainer.train(dataloader)
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    trainer.train(dataloader)
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    return f(*args, **kwargs)
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    trainer.train(dataloader)
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    outputs = self.pipeline_engine.train_batch_iter(
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+Traceback (most recent call last):
+    return f(*args, **kwargs)
+    outputs = self.pipeline_engine.train_batch_iter(
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+Traceback (most recent call last):
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    model.p2p.clear_history()
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    trainer.train(dataloader)
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    model.p2p.clear_history()
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return f(*args, **kwargs)
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+Traceback (most recent call last):
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    model.p2p.clear_history()
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+        return f(*args, **kwargs)outputs = self.pipeline_engine.train_batch_iter(
+
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    outputs = self.pipeline_engine.train_batch_iter(
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    outputs = self.pipeline_engine.train_batch_iter(
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    return f(*args, **kwargs)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    model.p2p.clear_history()
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    trainer.train(dataloader)
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    model.p2p.clear_history()
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+Traceback (most recent call last):
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    model.p2p.clear_history()
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    trainer.train(dataloader)
+Traceback (most recent call last):
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    outputs = self.pipeline_engine.train_batch_iter(
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    model.p2p.clear_history()
+    model.p2p.clear_history()
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    outputs = self.pipeline_engine.train_batch_iter(
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    model.p2p.clear_history()
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs = self.pipeline_engine.train_batch_iter(
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    model.p2p.clear_history()
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+Traceback (most recent call last):
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+Traceback (most recent call last):
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return f(*args, **kwargs)
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    outputs = self.pipeline_engine.train_batch_iter(
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    model.p2p.clear_history()
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    return f(*args, **kwargs)
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    model.p2p.clear_history()
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    return f(*args, **kwargs)
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+Traceback (most recent call last):
+    trainer.train(dataloader)
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    outputs = self.pipeline_engine.train_batch_iter(
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    trainer.train(dataloader)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+Traceback (most recent call last):
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    trainer.train(dataloader)
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    return f(*args, **kwargs)
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+Traceback (most recent call last):
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return f(*args, **kwargs)
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    model.p2p.clear_history()
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+Traceback (most recent call last):
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    model.p2p.clear_history()
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    trainer.train(dataloader)
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    trainer.train(dataloader)
+    model.p2p.clear_history()
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+Traceback (most recent call last):
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    return f(*args, **kwargs)
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return f(*args, **kwargs)
+    return f(*args, **kwargs)
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    outputs = self.pipeline_engine.train_batch_iter(
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    trainer.train(dataloader)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+Traceback (most recent call last):
+    model.p2p.clear_history()
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    outputs = self.pipeline_engine.train_batch_iter(
+    model.p2p.clear_history()
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    trainer.train(dataloader)
+    model.p2p.clear_history()
+    model.p2p.clear_history()
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    return f(*args, **kwargs)
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    model.p2p.clear_history()
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+Traceback (most recent call last):
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    model.p2p.clear_history()
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    trainer.train(dataloader)
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return f(*args, **kwargs)
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    return f(*args, **kwargs)
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    trainer.train(dataloader)
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    model.p2p.clear_history()
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    model.p2p.clear_history()
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    outputs = self.pipeline_engine.train_batch_iter(
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return f(*args, **kwargs)
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    trainer.train(dataloader)
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    model.p2p.clear_history()
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+Traceback (most recent call last):
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+      File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    outputs = self.pipeline_engine.train_batch_iter(
+    trainer.train(dataloader)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    return f(*args, **kwargs)
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    model.p2p.clear_history()
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    model.p2p.clear_history()
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    trainer.train(dataloader)
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+Traceback (most recent call last):
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    model.p2p.clear_history()
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+        return f(*args, **kwargs)model.p2p.clear_history()
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    model.p2p.clear_history()
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+Traceback (most recent call last):
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    outputs = self.pipeline_engine.train_batch_iter(
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs = self.pipeline_engine.train_batch_iter(
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    trainer.train(dataloader)
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    return f(*args, **kwargs)
+    model.p2p.clear_history()
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    return f(*args, **kwargs)
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    model.p2p.clear_history()
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    trainer.train(dataloader)
+    outputs = self.pipeline_engine.train_batch_iter(
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    model.p2p.clear_history()
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+Traceback (most recent call last):
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+Traceback (most recent call last):
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+Traceback (most recent call last):
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    trainer.train(dataloader)
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    trainer.train(dataloader)
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+Traceback (most recent call last):
+    model.p2p.clear_history()
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    return f(*args, **kwargs)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    model.p2p.clear_history()
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return f(*args, **kwargs)
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    model.p2p.clear_history()
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    model.p2p.clear_history()
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs = self.pipeline_engine.train_batch_iter(
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    model.p2p.clear_history()
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+Traceback (most recent call last):
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    model.p2p.clear_history()
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    model.p2p.clear_history()
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    model.p2p.clear_history()
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    outputs = self.pipeline_engine.train_batch_iter(
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    model.p2p.clear_history()
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    return f(*args, **kwargs)
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+Traceback (most recent call last):
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    outputs = self.pipeline_engine.train_batch_iter(
+    trainer.train(dataloader)  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    model.p2p.clear_history()
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return f(*args, **kwargs)
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    return f(*args, **kwargs)
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    trainer.train(dataloader)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    outputs = self.pipeline_engine.train_batch_iter(
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+Traceback (most recent call last):
+    model.p2p.clear_history()
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    trainer.train(dataloader)
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    model.p2p.clear_history()
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")    
+outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+AttributeError  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+: 'DistributedDataParallel' object has no attribute 'p2p'
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    model.p2p.clear_history()
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+Traceback (most recent call last):
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+        trainer.train(dataloader)trainer.train(dataloader)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return f(*args, **kwargs)
+    return f(*args, **kwargs)  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)    
+outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)    
+outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    model.p2p.clear_history()    
+model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError    : raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")    'DistributedDataParallel' object has no attribute 'p2p'
+raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: AttributeError'DistributedDataParallel' object has no attribute 'p2p': 'DistributedDataParallel' object has no attribute 'p2p'
+
+
+Traceback (most recent call last):
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    model.p2p.clear_history()
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+Traceback (most recent call last):
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+01/02/2025 08:33:44 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]: Resuming training from stage Stable Training Stage, it has trained for 0 samples and has 99 remaining train steps
+01/02/2025 08:33:44 [INFO|DP=0|PP=0|TP=0|ip-26-0-161-103]:  Memory usage: 12677.06MiB. Peak allocated 12677.06MiB. Peak reserved: 21084.00MiB
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+    model.p2p.clear_history()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+    raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+[2025-01-02 08:34:00,585] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 88451 closing signal SIGTERM
+[2025-01-02 08:34:00,585] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 88452 closing signal SIGTERM
+[2025-01-02 08:34:00,585] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 88453 closing signal SIGTERM
+[2025-01-02 08:34:00,585] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 88454 closing signal SIGTERM
+[2025-01-02 08:34:00,586] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 88455 closing signal SIGTERM
+[2025-01-02 08:34:00,586] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 88456 closing signal SIGTERM
+[2025-01-02 08:34:00,586] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 88458 closing signal SIGTERM
+[2025-01-02 08:34:00,589] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 105245 closing signal SIGTERM
+[2025-01-02 08:34:00,590] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 105246 closing signal SIGTERM
+[2025-01-02 08:34:00,590] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 105247 closing signal SIGTERM
+[2025-01-02 08:34:00,590] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 105248 closing signal SIGTERM
+[2025-01-02 08:34:00,590] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 105249 closing signal SIGTERM
+[2025-01-02 08:34:00,590] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 105251 closing signal SIGTERM
+[2025-01-02 08:34:00,590] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 105252 closing signal SIGTERM
+[2025-01-02 08:34:00,591] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 441951 closing signal SIGTERM
+[2025-01-02 08:34:00,591] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 441952 closing signal SIGTERM
+[2025-01-02 08:34:00,591] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 441954 closing signal SIGTERM
+[2025-01-02 08:34:00,591] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 24354 closing signal SIGTERM
+[2025-01-02 08:34:00,591] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 441955 closing signal SIGTERM
+[2025-01-02 08:34:00,591] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 441956 closing signal SIGTERM
+[2025-01-02 08:34:00,593] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 351865 closing signal SIGTERM
+[2025-01-02 08:34:00,591] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 441957 closing signal SIGTERM
+[2025-01-02 08:34:00,591] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 24355 closing signal SIGTERM
+[2025-01-02 08:34:00,591] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 441958 closing signal SIGTERM
+[2025-01-02 08:34:00,591] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 24357 closing signal SIGTERM
+[2025-01-02 08:34:00,591] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 24358 closing signal SIGTERM
+[2025-01-02 08:34:00,593] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 351866 closing signal SIGTERM
+[2025-01-02 08:34:00,593] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 351867 closing signal SIGTERM
+[2025-01-02 08:34:00,591] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 24359 closing signal SIGTERM
+[2025-01-02 08:34:00,593] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 351869 closing signal SIGTERM
+[2025-01-02 08:34:00,591] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 24360 closing signal SIGTERM
+[2025-01-02 08:34:00,593] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 351870 closing signal SIGTERM
+[2025-01-02 08:34:00,591] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 24361 closing signal SIGTERM
+[2025-01-02 08:34:00,593] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 351871 closing signal SIGTERM
+[2025-01-02 08:34:00,593] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 351872 closing signal SIGTERM
+[2025-01-02 08:34:00,593] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 229843 closing signal SIGTERM
+[2025-01-02 08:34:00,593] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 229844 closing signal SIGTERM
+[2025-01-02 08:34:00,593] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 229845 closing signal SIGTERM
+[2025-01-02 08:34:00,593] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 229847 closing signal SIGTERM
+[2025-01-02 08:34:00,593] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 229848 closing signal SIGTERM
+[2025-01-02 08:34:00,593] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 229849 closing signal SIGTERM
+[2025-01-02 08:34:00,593] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 229850 closing signal SIGTERM
+[2025-01-02 08:34:00,598] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 148541 closing signal SIGTERM
+[2025-01-02 08:34:00,598] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 148543 closing signal SIGTERM
+[2025-01-02 08:34:00,598] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 148544 closing signal SIGTERM
+[2025-01-02 08:34:00,598] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 148545 closing signal SIGTERM
+[2025-01-02 08:34:00,598] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 148546 closing signal SIGTERM
+[2025-01-02 08:34:00,598] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 148547 closing signal SIGTERM
+[2025-01-02 08:34:00,599] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 24185 closing signal SIGTERM
+[2025-01-02 08:34:00,599] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 24187 closing signal SIGTERM
+[2025-01-02 08:34:00,599] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 24188 closing signal SIGTERM
+[2025-01-02 08:34:00,599] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1024207 closing signal SIGTERM
+[2025-01-02 08:34:00,599] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 24189 closing signal SIGTERM
+[2025-01-02 08:34:00,599] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 24190 closing signal SIGTERM
+[2025-01-02 08:34:00,599] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 24191 closing signal SIGTERM
+[2025-01-02 08:34:00,599] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1024208 closing signal SIGTERM
+[2025-01-02 08:34:00,599] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 24192 closing signal SIGTERM
+[2025-01-02 08:34:00,599] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1024209 closing signal SIGTERM
+[2025-01-02 08:34:00,599] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1024210 closing signal SIGTERM
+[2025-01-02 08:34:00,599] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1024211 closing signal SIGTERM
+[2025-01-02 08:34:00,599] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1024212 closing signal SIGTERM
+[2025-01-02 08:34:00,603] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 172482 closing signal SIGTERM
+[2025-01-02 08:34:00,603] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 172483 closing signal SIGTERM
+[2025-01-02 08:34:00,603] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 172484 closing signal SIGTERM
+[2025-01-02 08:34:00,603] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 172485 closing signal SIGTERM
+[2025-01-02 08:34:00,603] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 172486 closing signal SIGTERM
+[2025-01-02 08:34:00,603] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 172488 closing signal SIGTERM
+[2025-01-02 08:34:00,603] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 172489 closing signal SIGTERM
+[2025-01-02 08:34:00,602] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 73349 closing signal SIGTERM
+[2025-01-02 08:34:00,603] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 73350 closing signal SIGTERM
+[2025-01-02 08:34:00,603] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 73351 closing signal SIGTERM
+[2025-01-02 08:34:00,603] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 73352 closing signal SIGTERM
+[2025-01-02 08:34:00,603] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 73354 closing signal SIGTERM
+[2025-01-02 08:34:00,603] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 73355 closing signal SIGTERM
+[2025-01-02 08:34:00,603] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 73356 closing signal SIGTERM
+[2025-01-02 08:34:02,255] torch.distributed.elastic.multiprocessing.api: [ERROR] failed (exitcode: 1) local_rank: 0 (pid: 148540) of binary: /fsx/nouamane/miniconda/envs/2-1-cu121/bin/python
+[2025-01-02 08:34:02,291] torch.distributed.elastic.multiprocessing.errors.error_handler: [ERROR] no error file defined for parent, to copy child error file (/tmp/torchelastic_ep36_na8/14019215_u9437snt/attempt_0/0/error.json)
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 264, in launch_agent
+    raise ChildFailedError(
+torch.distributed.elastic.multiprocessing.errors.ChildFailedError: 
+============================================================
+/fsx/nouamane/projects/nanotron/run_train.py FAILED
+------------------------------------------------------------
+Failures:
+[1]:
+  time      : 2025-01-02_08:33:43
+  host      : ip-26-0-175-34.ec2.internal
+  rank      : 506 (local_rank: 2)
+  exitcode  : 1 (pid: 148542)
+  error_file: /tmp/torchelastic_ep36_na8/14019215_u9437snt/attempt_0/2/error.json
+  traceback : Traceback (most recent call last):
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+      return f(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+      outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+      outputs = self.pipeline_engine.train_batch_iter(
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+      model.p2p.clear_history()
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+      raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+  AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+  
+------------------------------------------------------------
+Root Cause (first observed failure):
+[0]:
+  time      : 2025-01-02_08:33:43
+  host      : ip-26-0-175-34.ec2.internal
+  rank      : 504 (local_rank: 0)
+  exitcode  : 1 (pid: 148540)
+  error_file: /tmp/torchelastic_ep36_na8/14019215_u9437snt/attempt_0/0/error.json
+  traceback : Traceback (most recent call last):
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+      return f(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+      outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+      outputs = self.pipeline_engine.train_batch_iter(
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+      model.p2p.clear_history()
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+      raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+  AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+  
+============================================================
+[2025-01-02 08:34:02,366] torch.distributed.elastic.multiprocessing.api: [ERROR] failed (exitcode: 1) local_rank: 7 (pid: 1024213) of binary: /fsx/nouamane/miniconda/envs/2-1-cu121/bin/python
+[2025-01-02 08:34:02,409] torch.distributed.elastic.multiprocessing.errors.error_handler: [ERROR] no error file defined for parent, to copy child error file (/tmp/torchelastic_tf7cy0yv/14019215_83t4dahr/attempt_0/7/error.json)
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 264, in launch_agent
+    raise ChildFailedError(
+torch.distributed.elastic.multiprocessing.errors.ChildFailedError: 
+============================================================
+/fsx/nouamane/projects/nanotron/run_train.py FAILED
+------------------------------------------------------------
+Failures:
+[0]:
+  time      : 2025-01-02_08:33:44
+  host      : ip-26-0-161-103.ec2.internal
+  rank      : 0 (local_rank: 0)
+  exitcode  : 1 (pid: 1024206)
+  error_file: /tmp/torchelastic_tf7cy0yv/14019215_83t4dahr/attempt_0/0/error.json
+  traceback : Traceback (most recent call last):
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+      return f(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+      outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+      outputs = self.pipeline_engine.train_batch_iter(
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+      model.p2p.clear_history()
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+      raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+  AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+  
+------------------------------------------------------------
+Root Cause (first observed failure):
+[1]:
+  time      : 2025-01-02_08:33:43
+  host      : ip-26-0-161-103.ec2.internal
+  rank      : 7 (local_rank: 7)
+  exitcode  : 1 (pid: 1024213)
+  error_file: /tmp/torchelastic_tf7cy0yv/14019215_83t4dahr/attempt_0/7/error.json
+  traceback : Traceback (most recent call last):
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+      return f(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+      outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+      outputs = self.pipeline_engine.train_batch_iter(
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 245, in train_batch_iter
+      model.p2p.clear_history()
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1695, in __getattr__
+      raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
+  AttributeError: 'DistributedDataParallel' object has no attribute 'p2p'
+  
+============================================================
+srun: error: ip-26-0-175-34: task 59: Exited with exit code 1
+srun: launch/slurm: _step_signal: Terminating StepId=14019215.0
+slurmstepd: error: *** STEP 14019215.0 ON ip-26-0-161-78 CANCELLED AT 2025-01-02T08:34:02 ***
+[2025-01-02 08:34:02,647] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-02 08:34:02,648] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 963312 closing signal SIGTERM
+[2025-01-02 08:34:02,648] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 963313 closing signal SIGTERM
+[2025-01-02 08:34:02,648] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 963314 closing signal SIGTERM
+[2025-01-02 08:34:02,648] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 963315 closing signal SIGTERM
+[2025-01-02 08:34:02,648] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 963316 closing signal SIGTERM
+[2025-01-02 08:34:02,648] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-02 08:34:02,648] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 963317 closing signal SIGTERM
+[2025-01-02 08:34:02,648] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-02 08:34:02,648] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 963318 closing signal SIGTERM
+[2025-01-02 08:34:02,647] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-02 08:34:02,648] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 172483 closing signal SIGTERM
+[2025-01-02 08:34:02,648] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 172484 closing signal SIGTERM
+[2025-01-02 08:34:02,649] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 511244 closing signal SIGTERM
+[2025-01-02 08:34:02,648] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 172485 closing signal SIGTERM
+[2025-01-02 08:34:02,647] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-02 08:34:02,648] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 90542 closing signal SIGTERM
+[2025-01-02 08:34:02,647] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-02 08:34:02,648] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 172486 closing signal SIGTERM
+[2025-01-02 08:34:02,649] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 511245 closing signal SIGTERM
+[2025-01-02 08:34:02,648] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 172488 closing signal SIGTERM
+[2025-01-02 08:34:02,648] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 90543 closing signal SIGTERM
+[2025-01-02 08:34:02,649] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 511246 closing signal SIGTERM
+[2025-01-02 08:34:02,648] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 90544 closing signal SIGTERM
+[2025-01-02 08:34:02,648] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-02 08:34:02,647] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 41139 closing signal SIGTERM
+[2025-01-02 08:34:02,648] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 172489 closing signal SIGTERM
+[2025-01-02 08:34:02,647] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 187271 closing signal SIGTERM
+[2025-01-02 08:34:02,648] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 90545 closing signal SIGTERM
+[2025-01-02 08:34:02,647] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-02 08:34:02,649] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 511247 closing signal SIGTERM
+[2025-01-02 08:34:02,648] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-02 08:34:02,647] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 187272 closing signal SIGTERM
+[2025-01-02 08:34:02,647] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 41140 closing signal SIGTERM
+[2025-01-02 08:34:02,649] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 511248 closing signal SIGTERM
+[2025-01-02 08:34:02,649] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 146306 closing signal SIGTERM
+[2025-01-02 08:34:02,647] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 41141 closing signal SIGTERM
+[2025-01-02 08:34:02,649] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 511249 closing signal SIGTERM
+[2025-01-02 08:34:02,647] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 187273 closing signal SIGTERM
+[2025-01-02 08:34:02,648] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 90546 closing signal SIGTERM
+[2025-01-02 08:34:02,647] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 303187 closing signal SIGTERM
+[2025-01-02 08:34:02,649] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 146308 closing signal SIGTERM
+[2025-01-02 08:34:02,648] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 195241 closing signal SIGTERM
+[2025-01-02 08:34:02,647] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 187274 closing signal SIGTERM
+[2025-01-02 08:34:02,648] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 90548 closing signal SIGTERM
+[2025-01-02 08:34:02,647] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 41142 closing signal SIGTERM
+[2025-01-02 08:34:02,649] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 511250 closing signal SIGTERM
+[2025-01-02 08:34:02,648] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 195242 closing signal SIGTERM
+[2025-01-02 08:34:02,646] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-02 08:34:02,648] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 303188 closing signal SIGTERM
+[2025-01-02 08:34:02,647] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 41143 closing signal SIGTERM
+[2025-01-02 08:34:02,647] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-02 08:34:02,649] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 146309 closing signal SIGTERM
+[2025-01-02 08:34:02,648] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 195243 closing signal SIGTERM
+[2025-01-02 08:34:02,647] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-02 08:34:02,648] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 187276 closing signal SIGTERM
+[2025-01-02 08:34:02,647] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 222246 closing signal SIGTERM
+[2025-01-02 08:34:02,648] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 195244 closing signal SIGTERM
+[2025-01-02 08:34:02,648] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 187277 closing signal SIGTERM
+[2025-01-02 08:34:02,648] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 303189 closing signal SIGTERM
+[2025-01-02 08:34:02,647] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 222247 closing signal SIGTERM
+[2025-01-02 08:34:02,647] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 41145 closing signal SIGTERM
+[2025-01-02 08:34:02,649] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 146310 closing signal SIGTERM
+[2025-01-02 08:34:02,648] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 195245 closing signal SIGTERM
+[2025-01-02 08:34:02,648] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 44423 closing signal SIGTERM
+[2025-01-02 08:34:02,647] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 222248 closing signal SIGTERM
+[2025-01-02 08:34:02,648] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-02 08:34:02,649] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 146311 closing signal SIGTERM
+[2025-01-02 08:34:02,649] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 195247 closing signal SIGTERM
+[2025-01-02 08:34:02,647] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-02 08:34:02,647] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 222249 closing signal SIGTERM
+[2025-01-02 08:34:02,648] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 303190 closing signal SIGTERM
+[2025-01-02 08:34:02,649] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 146312 closing signal SIGTERM
+[2025-01-02 08:34:02,648] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 469889 closing signal SIGTERM
+[2025-01-02 08:34:02,648] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 41522 closing signal SIGTERM
+[2025-01-02 08:34:02,648] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 44424 closing signal SIGTERM
+[2025-01-02 08:34:02,647] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 222250 closing signal SIGTERM
+[2025-01-02 08:34:02,648] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 303191 closing signal SIGTERM
+[2025-01-02 08:34:02,648] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 469890 closing signal SIGTERM
+[2025-01-02 08:34:02,647] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 222252 closing signal SIGTERM
+[2025-01-02 08:34:02,648] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 41523 closing signal SIGTERM
+[2025-01-02 08:34:02,648] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 44425 closing signal SIGTERM
+[2025-01-02 08:34:02,648] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 469891 closing signal SIGTERM
+[2025-01-02 08:34:02,648] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 303193 closing signal SIGTERM
+[2025-01-02 08:34:02,648] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 41524 closing signal SIGTERM
+[2025-01-02 08:34:02,648] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-02 08:34:02,648] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-02 08:34:02,649] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-02 08:34:02,648] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 44426 closing signal SIGTERM
+[2025-01-02 08:34:02,648] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 275003 closing signal SIGTERM
+[2025-01-02 08:34:02,648] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 469892 closing signal SIGTERM
+[2025-01-02 08:34:02,648] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-02 08:34:02,648] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 469893 closing signal SIGTERM
+[2025-01-02 08:34:02,648] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 41525 closing signal SIGTERM
+[2025-01-02 08:34:02,648] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 469894 closing signal SIGTERM
+[2025-01-02 08:34:02,648] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 26883 closing signal SIGTERM
+[2025-01-02 08:34:02,648] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-02 08:34:02,648] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 44427 closing signal SIGTERM
+[2025-01-02 08:34:02,649] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 429783 closing signal SIGTERM
+[2025-01-02 08:34:02,648] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 44428 closing signal SIGTERM
+[2025-01-02 08:34:02,648] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 26884 closing signal SIGTERM
+[2025-01-02 08:34:02,648] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 41526 closing signal SIGTERM
+[2025-01-02 08:34:02,648] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 275004 closing signal SIGTERM
+[2025-01-02 08:34:02,648] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 229844 closing signal SIGTERM
+[2025-01-02 08:34:02,648] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 279031 closing signal SIGTERM
+[2025-01-02 08:34:02,648] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 32702 closing signal SIGTERM
+[2025-01-02 08:34:02,648] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 229845 closing signal SIGTERM
+[2025-01-02 08:34:02,648] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-02 08:34:02,650] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 429784 closing signal SIGTERM
+[2025-01-02 08:34:02,648] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 26885 closing signal SIGTERM
+[2025-01-02 08:34:02,648] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 229847 closing signal SIGTERM
+[2025-01-02 08:34:02,648] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 44429 closing signal SIGTERM
+[2025-01-02 08:34:02,648] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 41527 closing signal SIGTERM
+[2025-01-02 08:34:02,648] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 229848 closing signal SIGTERM
+[2025-01-02 08:34:02,648] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 275005 closing signal SIGTERM
+[2025-01-02 08:34:02,650] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 429785 closing signal SIGTERM
+[2025-01-02 08:34:02,648] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 229849 closing signal SIGTERM
+[2025-01-02 08:34:02,650] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 429786 closing signal SIGTERM
+[2025-01-02 08:34:02,648] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 26886 closing signal SIGTERM
+[2025-01-02 08:34:02,648] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 24355 closing signal SIGTERM
+[2025-01-02 08:34:02,648] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 26889 closing signal SIGTERM
+[2025-01-02 08:34:02,648] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 24357 closing signal SIGTERM
+[2025-01-02 08:34:02,648] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 279032 closing signal SIGTERM
+[2025-01-02 08:34:02,649] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 32703 closing signal SIGTERM
+[2025-01-02 08:34:02,648] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 275006 closing signal SIGTERM
+[2025-01-02 08:34:02,650] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 429788 closing signal SIGTERM
+[2025-01-02 08:34:02,648] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-02 08:34:02,648] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 24358 closing signal SIGTERM
+[2025-01-02 08:34:02,649] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 32704 closing signal SIGTERM
+[2025-01-02 08:34:02,648] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 24359 closing signal SIGTERM
+[2025-01-02 08:34:02,650] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-02 08:34:02,648] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 279033 closing signal SIGTERM
+[2025-01-02 08:34:02,648] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 24360 closing signal SIGTERM
+[2025-01-02 08:34:02,648] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 275007 closing signal SIGTERM
+[2025-01-02 08:34:02,650] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 429789 closing signal SIGTERM
+[2025-01-02 08:34:02,649] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 26890 closing signal SIGTERM
+[2025-01-02 08:34:02,648] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 24361 closing signal SIGTERM
+[2025-01-02 08:34:02,648] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 275008 closing signal SIGTERM
+[2025-01-02 08:34:02,649] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 32705 closing signal SIGTERM
+[2025-01-02 08:34:02,648] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 279034 closing signal SIGTERM
+[2025-01-02 08:34:02,649] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 88856 closing signal SIGTERM
+[2025-01-02 08:34:02,650] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3786916 closing signal SIGTERM
+[2025-01-02 08:34:02,648] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 279035 closing signal SIGTERM
+[2025-01-02 08:34:02,649] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 32706 closing signal SIGTERM
+[2025-01-02 08:34:02,648] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 279036 closing signal SIGTERM
+[2025-01-02 08:34:02,649] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 88857 closing signal SIGTERM
+[2025-01-02 08:34:02,650] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3786917 closing signal SIGTERM
+[2025-01-02 08:34:02,649] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 32707 closing signal SIGTERM
+[2025-01-02 08:34:02,649] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-02 08:34:02,647] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-02 08:34:02,649] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 88858 closing signal SIGTERM
+[2025-01-02 08:34:02,650] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3786918 closing signal SIGTERM
+[2025-01-02 08:34:02,649] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-02 08:34:02,648] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-02 08:34:02,649] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 32708 closing signal SIGTERM
+[2025-01-02 08:34:02,649] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-02 08:34:02,648] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-02 08:34:02,649] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 88860 closing signal SIGTERM
+[2025-01-02 08:34:02,650] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3786919 closing signal SIGTERM
+[2025-01-02 08:34:02,649] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 62523 closing signal SIGTERM
+[2025-01-02 08:34:02,649] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 88861 closing signal SIGTERM
+[2025-01-02 08:34:02,649] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-02 08:34:02,649] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 61744 closing signal SIGTERM
+[2025-01-02 08:34:02,650] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3786920 closing signal SIGTERM
+[2025-01-02 08:34:02,649] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 170886 closing signal SIGTERM
+[2025-01-02 08:34:02,648] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 41575 closing signal SIGTERM
+[2025-01-02 08:34:02,649] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 62524 closing signal SIGTERM
+[2025-01-02 08:34:02,648] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 73350 closing signal SIGTERM
+[2025-01-02 08:34:02,648] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 41576 closing signal SIGTERM
+[2025-01-02 08:34:02,649] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-02 08:34:02,649] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 61745 closing signal SIGTERM
+[2025-01-02 08:34:02,649] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 206253 closing signal SIGTERM
+[2025-01-02 08:34:02,649] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 170887 closing signal SIGTERM
+[2025-01-02 08:34:02,648] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 41577 closing signal SIGTERM
+[2025-01-02 08:34:02,649] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 127642 closing signal SIGTERM
+[2025-01-02 08:34:02,650] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3786921 closing signal SIGTERM
+[2025-01-02 08:34:02,648] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 73352 closing signal SIGTERM
+[2025-01-02 08:34:02,649] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 206254 closing signal SIGTERM
+[2025-01-02 08:34:02,650] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3786922 closing signal SIGTERM
+[2025-01-02 08:34:02,649] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 62525 closing signal SIGTERM
+[2025-01-02 08:34:02,648] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 41578 closing signal SIGTERM
+[2025-01-02 08:34:02,649] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 127643 closing signal SIGTERM
+[2025-01-02 08:34:02,649] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 61746 closing signal SIGTERM
+[2025-01-02 08:34:02,649] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 170888 closing signal SIGTERM
+[2025-01-02 08:34:02,649] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 167119 closing signal SIGTERM
+[2025-01-02 08:34:02,649] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-02 08:34:02,648] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-02 08:34:02,649] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-02 08:34:02,649] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 73354 closing signal SIGTERM
+[2025-01-02 08:34:02,649] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 127645 closing signal SIGTERM
+[2025-01-02 08:34:02,649] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 206256 closing signal SIGTERM
+[2025-01-02 08:34:02,649] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 62526 closing signal SIGTERM
+[2025-01-02 08:34:02,649] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 73355 closing signal SIGTERM
+[2025-01-02 08:34:02,648] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 41580 closing signal SIGTERM
+[2025-01-02 08:34:02,648] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-02 08:34:02,649] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 61747 closing signal SIGTERM
+[2025-01-02 08:34:02,649] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 72027 closing signal SIGTERM
+[2025-01-02 08:34:02,649] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 170889 closing signal SIGTERM
+[2025-01-02 08:34:02,649] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 62527 closing signal SIGTERM
+[2025-01-02 08:34:02,648] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 41581 closing signal SIGTERM
+[2025-01-02 08:34:02,649] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 88453 closing signal SIGTERM
+[2025-01-02 08:34:02,649] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 61748 closing signal SIGTERM
+[2025-01-02 08:34:02,649] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 170890 closing signal SIGTERM
+[2025-01-02 08:34:02,649] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 88454 closing signal SIGTERM
+[2025-01-02 08:34:02,649] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 127646 closing signal SIGTERM
+[2025-01-02 08:34:02,649] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 472415 closing signal SIGTERM
+[2025-01-02 08:34:02,649] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 206257 closing signal SIGTERM
+[2025-01-02 08:34:02,649] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 170891 closing signal SIGTERM
+[2025-01-02 08:34:02,649] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 167120 closing signal SIGTERM
+[2025-01-02 08:34:02,649] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 92911 closing signal SIGTERM
+[2025-01-02 08:34:02,649] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 88455 closing signal SIGTERM
+[2025-01-02 08:34:02,649] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 127647 closing signal SIGTERM
+[2025-01-02 08:34:02,649] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 72028 closing signal SIGTERM
+[2025-01-02 08:34:02,649] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-02 08:34:02,649] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 62529 closing signal SIGTERM
+[2025-01-02 08:34:02,649] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 73356 closing signal SIGTERM
+[2025-01-02 08:34:02,649] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 472416 closing signal SIGTERM
+[2025-01-02 08:34:02,649] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 61749 closing signal SIGTERM
+[2025-01-02 08:34:02,649] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 206258 closing signal SIGTERM
+[2025-01-02 08:34:02,649] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 167121 closing signal SIGTERM
+[2025-01-02 08:34:02,649] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 92913 closing signal SIGTERM
+[2025-01-02 08:34:02,649] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 88456 closing signal SIGTERM
+[2025-01-02 08:34:02,650] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-02 08:34:02,649] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-02 08:34:02,649] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 105246 closing signal SIGTERM
+[2025-01-02 08:34:02,649] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 170892 closing signal SIGTERM
+[2025-01-02 08:34:02,648] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-02 08:34:02,649] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 92914 closing signal SIGTERM
+[2025-01-02 08:34:02,649] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 127648 closing signal SIGTERM
+[2025-01-02 08:34:02,649] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 472417 closing signal SIGTERM
+[2025-01-02 08:34:02,649] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 206259 closing signal SIGTERM
+[2025-01-02 08:34:02,650] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 72030 closing signal SIGTERM
+[2025-01-02 08:34:02,650] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-02 08:34:02,649] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 397178 closing signal SIGTERM
+[2025-01-02 08:34:02,649] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 88458 closing signal SIGTERM
+[2025-01-02 08:34:02,649] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 395030 closing signal SIGTERM
+[2025-01-02 08:34:02,649] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 105247 closing signal SIGTERM
+[2025-01-02 08:34:02,651] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 612900 closing signal SIGTERM
+[2025-01-02 08:34:02,649] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 167123 closing signal SIGTERM
+[2025-01-02 08:34:02,649] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 92915 closing signal SIGTERM
+[2025-01-02 08:34:02,649] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-02 08:34:02,649] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 472418 closing signal SIGTERM
+[2025-01-02 08:34:02,650] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 70194 closing signal SIGTERM
+[2025-01-02 08:34:02,650] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 72031 closing signal SIGTERM
+[2025-01-02 08:34:02,649] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 105248 closing signal SIGTERM
+[2025-01-02 08:34:02,650] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-02 08:34:02,649] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 397179 closing signal SIGTERM
+[2025-01-02 08:34:02,649] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 92916 closing signal SIGTERM
+[2025-01-02 08:34:02,649] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 472419 closing signal SIGTERM
+[2025-01-02 08:34:02,651] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-02 08:34:02,650] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 70195 closing signal SIGTERM
+[2025-01-02 08:34:02,649] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 395031 closing signal SIGTERM
+[2025-01-02 08:34:02,651] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 612901 closing signal SIGTERM
+[2025-01-02 08:34:02,650] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-02 08:34:02,649] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 397180 closing signal SIGTERM
+[2025-01-02 08:34:02,649] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 92917 closing signal SIGTERM
+[2025-01-02 08:34:02,649] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 472420 closing signal SIGTERM
+[2025-01-02 08:34:02,650] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 70196 closing signal SIGTERM
+[2025-01-02 08:34:02,650] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 72033 closing signal SIGTERM
+[2025-01-02 08:34:02,649] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-02 08:34:02,651] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 612902 closing signal SIGTERM
+[2025-01-02 08:34:02,649] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 167124 closing signal SIGTERM
+[2025-01-02 08:34:02,649] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 397181 closing signal SIGTERM
+[2025-01-02 08:34:02,649] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 441954 closing signal SIGTERM
+[2025-01-02 08:34:02,649] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-02 08:34:02,651] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 351866 closing signal SIGTERM
+[2025-01-02 08:34:02,650] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 70197 closing signal SIGTERM
+[2025-01-02 08:34:02,649] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-02 08:34:02,649] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 395032 closing signal SIGTERM
+[2025-01-02 08:34:02,649] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 105249 closing signal SIGTERM
+[2025-01-02 08:34:02,651] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 50845 closing signal SIGTERM
+[2025-01-02 08:34:02,650] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 441955 closing signal SIGTERM
+[2025-01-02 08:34:02,649] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 139334 closing signal SIGTERM
+[2025-01-02 08:34:02,651] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 351867 closing signal SIGTERM
+[2025-01-02 08:34:02,650] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 70198 closing signal SIGTERM
+[2025-01-02 08:34:02,650] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 204751 closing signal SIGTERM
+[2025-01-02 08:34:02,650] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 94168 closing signal SIGTERM
+[2025-01-02 08:34:02,651] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 612903 closing signal SIGTERM
+[2025-01-02 08:34:02,651] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 50846 closing signal SIGTERM
+[2025-01-02 08:34:02,649] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 167125 closing signal SIGTERM
+[2025-01-02 08:34:02,649] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 397182 closing signal SIGTERM
+[2025-01-02 08:34:02,650] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 441956 closing signal SIGTERM
+[2025-01-02 08:34:02,649] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 139335 closing signal SIGTERM
+[2025-01-02 08:34:02,651] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 351869 closing signal SIGTERM
+[2025-01-02 08:34:02,650] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 58030 closing signal SIGTERM
+[2025-01-02 08:34:02,649] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 395033 closing signal SIGTERM
+[2025-01-02 08:34:02,650] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 204752 closing signal SIGTERM
+[2025-01-02 08:34:02,651] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 612904 closing signal SIGTERM
+[2025-01-02 08:34:02,651] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 50847 closing signal SIGTERM
+[2025-01-02 08:34:02,649] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 397184 closing signal SIGTERM
+[2025-01-02 08:34:02,650] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 441957 closing signal SIGTERM
+[2025-01-02 08:34:02,648] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-02 08:34:02,651] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 351870 closing signal SIGTERM
+[2025-01-02 08:34:02,650] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 70199 closing signal SIGTERM
+[2025-01-02 08:34:02,649] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 395034 closing signal SIGTERM
+[2025-01-02 08:34:02,650] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 441958 closing signal SIGTERM
+[2025-01-02 08:34:02,651] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 351871 closing signal SIGTERM
+[2025-01-02 08:34:02,650] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 70200 closing signal SIGTERM
+[2025-01-02 08:34:02,650] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 58031 closing signal SIGTERM
+[2025-01-02 08:34:02,649] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 395035 closing signal SIGTERM
+[2025-01-02 08:34:02,650] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 204753 closing signal SIGTERM
+[2025-01-02 08:34:02,649] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 139336 closing signal SIGTERM
+[2025-01-02 08:34:02,651] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 351872 closing signal SIGTERM
+[2025-01-02 08:34:02,649] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 395036 closing signal SIGTERM
+[2025-01-02 08:34:02,650] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 204754 closing signal SIGTERM
+[2025-01-02 08:34:02,650] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 94170 closing signal SIGTERM
+[2025-01-02 08:34:02,651] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 612905 closing signal SIGTERM
+[2025-01-02 08:34:02,651] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 50848 closing signal SIGTERM
+[2025-01-02 08:34:02,650] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 58032 closing signal SIGTERM
+[2025-01-02 08:34:02,651] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 612906 closing signal SIGTERM
+[2025-01-02 08:34:02,651] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 50849 closing signal SIGTERM
+[2025-01-02 08:34:02,649] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 24398 closing signal SIGTERM
+[2025-01-02 08:34:02,650] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 94171 closing signal SIGTERM
+[2025-01-02 08:34:02,651] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 50850 closing signal SIGTERM
+[2025-01-02 08:34:02,649] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 139337 closing signal SIGTERM
+[2025-01-02 08:34:02,650] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 204755 closing signal SIGTERM
+[2025-01-02 08:34:02,649] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 139338 closing signal SIGTERM
+[2025-01-02 08:34:02,650] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 58033 closing signal SIGTERM
+[2025-01-02 08:34:02,650] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 204756 closing signal SIGTERM
+[2025-01-02 08:34:02,649] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 139339 closing signal SIGTERM
+[2025-01-02 08:34:02,650] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 58034 closing signal SIGTERM
+[2025-01-02 08:34:02,650] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 204757 closing signal SIGTERM
+[2025-01-02 08:34:02,651] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 94172 closing signal SIGTERM
+[2025-01-02 08:34:02,650] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 58036 closing signal SIGTERM
+[2025-01-02 08:34:02,651] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 94173 closing signal SIGTERM
+[2025-01-02 08:34:02,649] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 24399 closing signal SIGTERM
+[2025-01-02 08:34:02,652] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-02 08:34:02,650] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-02 08:34:02,649] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-02 08:34:02,650] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-02 08:34:02,650] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-02 08:34:02,650] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-02 08:34:02,653] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 102707 closing signal SIGTERM
+[2025-01-02 08:34:02,649] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 24401 closing signal SIGTERM
+[2025-01-02 08:34:02,649] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-02 08:34:02,650] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 829130 closing signal SIGTERM
+[2025-01-02 08:34:02,650] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 667059 closing signal SIGTERM
+[2025-01-02 08:34:02,650] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 95606 closing signal SIGTERM
+[2025-01-02 08:34:02,653] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 102708 closing signal SIGTERM
+[2025-01-02 08:34:02,650] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 42983 closing signal SIGTERM
+[2025-01-02 08:34:02,650] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 198907 closing signal SIGTERM
+[2025-01-02 08:34:02,650] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 829131 closing signal SIGTERM
+[2025-01-02 08:34:02,649] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 24402 closing signal SIGTERM
+[2025-01-02 08:34:02,649] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 101626 closing signal SIGTERM
+[2025-01-02 08:34:02,650] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 667060 closing signal SIGTERM
+[2025-01-02 08:34:02,650] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 95607 closing signal SIGTERM
+[2025-01-02 08:34:02,650] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-02 08:34:02,653] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 102709 closing signal SIGTERM
+[2025-01-02 08:34:02,650] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 198909 closing signal SIGTERM
+[2025-01-02 08:34:02,650] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 829132 closing signal SIGTERM
+[2025-01-02 08:34:02,649] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 24403 closing signal SIGTERM
+[2025-01-02 08:34:02,649] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 101627 closing signal SIGTERM
+[2025-01-02 08:34:02,650] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 667061 closing signal SIGTERM
+[2025-01-02 08:34:02,650] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 95608 closing signal SIGTERM
+[2025-01-02 08:34:02,649] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 101628 closing signal SIGTERM
+[2025-01-02 08:34:02,653] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 102710 closing signal SIGTERM
+[2025-01-02 08:34:02,650] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 42985 closing signal SIGTERM
+[2025-01-02 08:34:02,650] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 198910 closing signal SIGTERM
+[2025-01-02 08:34:02,650] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 95609 closing signal SIGTERM
+[2025-01-02 08:34:02,653] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 102711 closing signal SIGTERM
+[2025-01-02 08:34:02,650] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 829134 closing signal SIGTERM
+[2025-01-02 08:34:02,650] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 402623 closing signal SIGTERM
+[2025-01-02 08:34:02,653] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 102712 closing signal SIGTERM
+[2025-01-02 08:34:02,650] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 42986 closing signal SIGTERM
+[2025-01-02 08:34:02,650] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 667062 closing signal SIGTERM
+[2025-01-02 08:34:02,649] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 101630 closing signal SIGTERM
+[2025-01-02 08:34:02,650] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-02 08:34:02,650] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 667063 closing signal SIGTERM
+[2025-01-02 08:34:02,650] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 198911 closing signal SIGTERM
+[2025-01-02 08:34:02,650] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 95610 closing signal SIGTERM
+[2025-01-02 08:34:02,650] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 402624 closing signal SIGTERM
+[2025-01-02 08:34:02,650] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 198912 closing signal SIGTERM
+[2025-01-02 08:34:02,650] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 95611 closing signal SIGTERM
+[2025-01-02 08:34:02,650] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 829135 closing signal SIGTERM
+[2025-01-02 08:34:02,650] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 42987 closing signal SIGTERM
+[2025-01-02 08:34:02,650] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 24187 closing signal SIGTERM
+[2025-01-02 08:34:02,650] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 198913 closing signal SIGTERM
+[2025-01-02 08:34:02,650] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 95612 closing signal SIGTERM
+[2025-01-02 08:34:02,650] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-02 08:34:02,650] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 829136 closing signal SIGTERM
+[2025-01-02 08:34:02,650] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 402625 closing signal SIGTERM
+[2025-01-02 08:34:02,649] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 101631 closing signal SIGTERM
+[2025-01-02 08:34:02,650] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 667065 closing signal SIGTERM
+[2025-01-02 08:34:02,649] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 101632 closing signal SIGTERM
+[2025-01-02 08:34:02,650] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 24188 closing signal SIGTERM
+[2025-01-02 08:34:02,650] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-02 08:34:02,650] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 42989 closing signal SIGTERM
+[2025-01-02 08:34:02,650] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 24189 closing signal SIGTERM
+[2025-01-02 08:34:02,650] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 24190 closing signal SIGTERM
+[2025-01-02 08:34:02,650] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 402626 closing signal SIGTERM
+[2025-01-02 08:34:02,650] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 587973 closing signal SIGTERM
+[2025-01-02 08:34:02,650] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 383213 closing signal SIGTERM
+[2025-01-02 08:34:02,650] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 24191 closing signal SIGTERM
+[2025-01-02 08:34:02,650] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 402627 closing signal SIGTERM
+[2025-01-02 08:34:02,650] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 587974 closing signal SIGTERM
+[2025-01-02 08:34:02,650] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 587975 closing signal SIGTERM
+[2025-01-02 08:34:02,650] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-02 08:34:02,651] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 587976 closing signal SIGTERM
+[2025-01-02 08:34:02,650] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 383214 closing signal SIGTERM
+[2025-01-02 08:34:02,651] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 402628 closing signal SIGTERM
+[2025-01-02 08:34:02,651] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 402629 closing signal SIGTERM
+[2025-01-02 08:34:02,651] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 587977 closing signal SIGTERM
+[2025-01-02 08:34:02,650] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-02 08:34:02,650] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 383215 closing signal SIGTERM
+[2025-01-02 08:34:02,651] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 587979 closing signal SIGTERM
+[2025-01-02 08:34:02,651] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 93269 closing signal SIGTERM
+[2025-01-02 08:34:02,651] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 383216 closing signal SIGTERM
+[2025-01-02 08:34:02,651] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 935187 closing signal SIGTERM
+[2025-01-02 08:34:02,651] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 93270 closing signal SIGTERM
+[2025-01-02 08:34:02,652] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-02 08:34:02,651] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 383217 closing signal SIGTERM
+[2025-01-02 08:34:02,651] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 93271 closing signal SIGTERM
+[2025-01-02 08:34:02,652] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 51047 closing signal SIGTERM
+[2025-01-02 08:34:02,651] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 383218 closing signal SIGTERM
+[2025-01-02 08:34:02,651] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 93273 closing signal SIGTERM
+[2025-01-02 08:34:02,651] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 935188 closing signal SIGTERM
+[2025-01-02 08:34:02,651] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 93274 closing signal SIGTERM
+[2025-01-02 08:34:02,651] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 935189 closing signal SIGTERM
+[2025-01-02 08:34:02,651] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 93275 closing signal SIGTERM
+[2025-01-02 08:34:02,652] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 51049 closing signal SIGTERM
+[2025-01-02 08:34:02,651] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 935190 closing signal SIGTERM
+[2025-01-02 08:34:02,653] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 51050 closing signal SIGTERM
+[2025-01-02 08:34:02,653] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 51052 closing signal SIGTERM
+[2025-01-02 08:34:02,651] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 935191 closing signal SIGTERM
+[2025-01-02 08:34:02,653] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 51053 closing signal SIGTERM
+[2025-01-02 08:34:02,651] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 935192 closing signal SIGTERM
+srun: error: ip-26-0-161-103: task 1: Terminated
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 24323 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 50973 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 878, in _invoke_run
+    run_result = self._monitor_workers(self._worker_group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/local_elastic_agent.py", line 307, in _monitor_workers
+    result = self._pcontext.wait(0)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 288, in wait
+    return self._poll()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 668, in _poll
+    self.close()  # terminate all running procs
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 331, in close
+    self._close(death_sig=death_sig, timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 713, in _close
+    handler.proc.wait(time_to_wait)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1209, in wait
+    return self._wait(timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1953, in _wait
+    time.sleep(delay)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 24112 got signal: 15
+srun: error: ip-26-0-169-139: task 34: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 878, in _invoke_run
+    run_result = self._monitor_workers(self._worker_group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/local_elastic_agent.py", line 307, in _monitor_workers
+    result = self._pcontext.wait(0)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 288, in wait
+    return self._poll()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 668, in _poll
+    self.close()  # terminate all running procs
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 331, in close
+    self._close(death_sig=death_sig, timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 713, in _close
+    handler.proc.wait(time_to_wait)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1209, in wait
+    return self._wait(timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1953, in _wait
+    time.sleep(delay)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 73276 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 878, in _invoke_run
+    run_result = self._monitor_workers(self._worker_group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/local_elastic_agent.py", line 307, in _monitor_workers
+    result = self._pcontext.wait(0)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 288, in wait
+    return self._poll()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 668, in _poll
+    self.close()  # terminate all running procs
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 331, in close
+    self._close(death_sig=death_sig, timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 713, in _close
+    handler.proc.wait(time_to_wait)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1209, in wait
+    return self._wait(timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1953, in _wait
+    time.sleep(delay)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 229771 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 878, in _invoke_run
+    run_result = self._monitor_workers(self._worker_group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/local_elastic_agent.py", line 307, in _monitor_workers
+    result = self._pcontext.wait(0)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 288, in wait
+    return self._poll()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 668, in _poll
+    self.close()  # terminate all running procs
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 331, in close
+    self._close(death_sig=death_sig, timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 713, in _close
+    handler.proc.wait(time_to_wait)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1209, in wait
+    return self._wait(timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1953, in _wait
+    time.sleep(delay)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 441877 got signal: 15
+srun: error: ip-26-0-175-241: task 63: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 878, in _invoke_run
+    run_result = self._monitor_workers(self._worker_group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/local_elastic_agent.py", line 307, in _monitor_workers
+    result = self._pcontext.wait(0)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 288, in wait
+    return self._poll()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 668, in _poll
+    self.close()  # terminate all running procs
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 331, in close
+    self._close(death_sig=death_sig, timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 713, in _close
+    handler.proc.wait(time_to_wait)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1209, in wait
+    return self._wait(timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1953, in _wait
+    time.sleep(delay)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 105172 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 878, in _invoke_run
+    run_result = self._monitor_workers(self._worker_group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/local_elastic_agent.py", line 307, in _monitor_workers
+    result = self._pcontext.wait(0)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 288, in wait
+    return self._poll()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 668, in _poll
+    self.close()  # terminate all running procs
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 331, in close
+    self._close(death_sig=death_sig, timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 713, in _close
+    handler.proc.wait(time_to_wait)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1209, in wait
+    return self._wait(timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1953, in _wait
+    time.sleep(delay)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 351791 got signal: 15
+srun: error: ip-26-0-175-170: task 62: Exited with exit code 1
+srun: error: ip-26-0-169-86: task 32: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 71954 got signal: 15
+srun: error: ip-26-0-171-56: task 43: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 878, in _invoke_run
+    run_result = self._monitor_workers(self._worker_group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/local_elastic_agent.py", line 307, in _monitor_workers
+    result = self._pcontext.wait(0)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 288, in wait
+    return self._poll()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 668, in _poll
+    self.close()  # terminate all running procs
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 331, in close
+    self._close(death_sig=death_sig, timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 713, in _close
+    handler.proc.wait(time_to_wait)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1209, in wait
+    return self._wait(timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1953, in _wait
+    time.sleep(delay)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+srun: error: ip-26-0-165-213: task 17: Exited with exit code 1
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 88378 got signal: 15
+srun: error: ip-26-0-170-31: task 38: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 303114 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 878, in _invoke_run
+    run_result = self._monitor_workers(self._worker_group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/local_elastic_agent.py", line 307, in _monitor_workers
+    result = self._pcontext.wait(0)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 288, in wait
+    return self._poll()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 668, in _poll
+    self.close()  # terminate all running procs
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 331, in close
+    self._close(death_sig=death_sig, timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 713, in _close
+    handler.proc.wait(time_to_wait)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1209, in wait
+    return self._wait(timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1953, in _wait
+    time.sleep(delay)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 24282 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 50770 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 88782 got signal: 15
+srun: error: ip-26-0-171-88: task 45: Exited with exit code 1
+srun: error: ip-26-0-171-62: task 44: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 94095 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+srun: error: ip-26-0-169-132: task 33: Exited with exit code 1
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 397103 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 170813 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 93195 got signal: 15
+srun: error: ip-26-0-164-18: task 5: Exited with exit code 1
+srun: error: ip-26-0-167-217: task 27: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 146231 got signal: 15
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 429710 got signal: 15
+srun: error: ip-26-0-167-175: task 25: Exited with exit code 1
+srun: error: ip-26-0-167-9: task 23: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 41448 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 878, in _invoke_run
+    run_result = self._monitor_workers(self._worker_group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/local_elastic_agent.py", line 307, in _monitor_workers
+    result = self._pcontext.wait(0)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 288, in wait
+    return self._poll()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 668, in _poll
+    self.close()  # terminate all running procs
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 331, in close
+    self._close(death_sig=death_sig, timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 713, in _close
+    handler.proc.wait(time_to_wait)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1209, in wait
+    return self._wait(timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1953, in _wait
+    time.sleep(delay)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 172410 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 102617 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 383140 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 587900 got signal: 15
+srun: error: ip-26-0-170-143: task 40: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    run(args)
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 187197 got signal: 15
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 139261 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 666985 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 41501 got signal: 15
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 42909 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 195167 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 90468 got signal: 15
+srun: error: ip-26-0-170-132: task 39: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 92837 got signal: 15
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 61670 got signal: 15
+srun: error: ip-26-0-166-214: task 21: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 41066 got signal: 15
+srun: error: ip-26-0-166-15: task 18: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    result = self._invoke_run(role)
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    time.sleep(monitor_interval)
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+    result = f(*args, **kwargs)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 62450 got signal: 15
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 935114 got signal: 15
+srun: error: ip-26-0-175-19: task 58: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 26811 got signal: 15
+srun: error: ip-26-0-174-240: task 57: Exited with exit code 1
+srun: error: ip-26-0-164-45: task 6: Exited with exit code 1
+srun: error: ip-26-0-165-24: task 11: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    sys.exit(main())
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 3786842 got signal: 15
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 57956 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+srun: error: ip-26-0-165-59: task 13: Exited with exit code 1
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 222173 got signal: 15
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 469813 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+srun: error: ip-26-0-175-165: task 61: Exited with exit code 1
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 127569 got signal: 15
+srun: error: ip-26-0-166-125: task 20: Exited with exit code 1
+srun: error: ip-26-0-164-207: task 9: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = agent.run()
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = self._invoke_run(role)
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 167045 got signal: 15
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 206180 got signal: 15
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 274929 got signal: 15
+srun: error: ip-26-0-161-123: task 2: Exited with exit code 1
+srun: error: ip-26-0-169-239: task 36: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 101542 got signal: 15
+srun: error: ip-26-0-171-230: task 48: Exited with exit code 1
+srun: error: ip-26-0-172-57: task 50: Exited with exit code 1
+srun: error: ip-26-0-164-236: task 10: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 198833 got signal: 15
+srun: error: ip-26-0-166-244: task 22: Exited with exit code 1
+srun: error: ip-26-0-169-207: task 35: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 829057 got signal: 15
+srun: error: ip-26-0-174-186: task 55: Exited with exit code 1
+srun: error: ip-26-0-168-95: task 29: Exited with exit code 1
+srun: error: ip-26-0-171-168: task 47: Exited with exit code 1
+srun: error: ip-26-0-170-160: task 41: Exited with exit code 1
+srun: error: ip-26-0-165-202: task 16: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 278956 got signal: 15
+srun: error: ip-26-0-165-38: task 12: Exited with exit code 1
+srun: error: ip-26-0-167-245: task 28: Exited with exit code 1
+srun: error: ip-26-0-165-131: task 14: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    sys.exit(main())
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    return f(*args, **kwargs)
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+srun: error: ip-26-0-171-21: task 42: Exited with exit code 1
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+srun: error: ip-26-0-172-116: task 52: Exited with exit code 1
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 472341 got signal: 15
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 70120 got signal: 15
+srun: error: ip-26-0-164-75: task 7: Exited with exit code 1
+srun: error: ip-26-0-167-177: task 26: Exited with exit code 1
+srun: error: ip-26-0-168-238: task 31: Exited with exit code 1
+srun: error: ip-26-0-175-132: task 60: Exited with exit code 1
+srun: error: ip-26-0-171-249: task 49: Exited with exit code 1
+srun: error: ip-26-0-163-236: task 3: Exited with exit code 1
+srun: error: ip-26-0-166-36: task 19: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 44349 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 963238 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 402550 got signal: 15
+srun: error: ip-26-0-167-51: task 24: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 394956 got signal: 15
+srun: error: ip-26-0-168-120: task 30: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 204677 got signal: 15
+srun: error: ip-26-0-164-187: task 8: Exited with exit code 1
+srun: error: ip-26-0-169-247: task 37: Exited with exit code 1
+srun: error: ip-26-0-161-78: task 0: Exited with exit code 1
+srun: error: ip-26-0-174-196: task 56: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 32627 got signal: 15
+srun: error: ip-26-0-174-100: task 54: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 95533 got signal: 15
+srun: error: ip-26-0-165-164: task 15: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 511171 got signal: 15
+srun: error: ip-26-0-172-142: task 53: Exited with exit code 1
+srun: error: ip-26-0-164-0: task 4: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 612828 got signal: 15
+srun: error: ip-26-0-171-102: task 46: Exited with exit code 1
+srun: error: _server_read: fd 41 error reading header: Connection reset by peer
+srun: error: Aborting, io error and missing step on node 51
+srun: Job step aborted: Waiting up to 32 seconds for job step to finish.
+srun: error: Timed out waiting for job step to complete
diff --git a/logs/14019308-bench_469G_dp4_tp8_pp8_acc64_mbs1_seq4096_zero1_tpmodeRED_vocab131k.out b/logs/14019308-bench_469G_dp4_tp8_pp8_acc64_mbs1_seq4096_zero1_tpmodeRED_vocab131k.out
new file mode 100644
index 0000000000000000000000000000000000000000..e6b224c24218d2ff6fe2469c7cd7187ab8a4c229
--- /dev/null
+++ b/logs/14019308-bench_469G_dp4_tp8_pp8_acc64_mbs1_seq4096_zero1_tpmodeRED_vocab131k.out
@@ -0,0 +1,10341 @@
++ '[' -z 14019308 ']'
++ source /etc/profile.d/modules.sh
+++ . /usr/share/modules/init/bash
++++ unset _mlshdbg
++++ '[' 0 = 1 ']'
++++ unset _mlre _mlIFS
++++ '[' -n x ']'
++++ _mlIFS=' 	
+'
++++ IFS=' '
++++ '[' -n '' ']'
+++++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash autoinit
++++ _mlcode='module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+MODULEPATH_modshare=/etc/environment-modules/modules:1:/usr/share/modules/\$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1; export MODULEPATH_modshare;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+LOADEDMODULES=; export LOADEDMODULES;
+MODULEPATH=/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/\$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles; export MODULEPATH;
+test 0;'
++++ _mlret=0
++++ '[' -n x ']'
++++ IFS=' 	
+'
++++ unset _mlIFS
++++ unset _mlre _mlv _mlrv
++++ '[' 0 -eq 0 ']'
++++ eval 'module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+MODULEPATH_modshare=/etc/environment-modules/modules:1:/usr/share/modules/\$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1; export MODULEPATH_modshare;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+LOADEDMODULES=; export LOADEDMODULES;
+MODULEPATH=/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/\$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles; export MODULEPATH;
+test 0;'
+++++ MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl
+++++ export MODULES_CMD
+++++ ENV=/usr/share/modules/init/profile.sh
+++++ export ENV
+++++ MODULEPATH_modshare='/etc/environment-modules/modules:1:/usr/share/modules/$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1'
+++++ export MODULEPATH_modshare
+++++ BASH_ENV=/usr/share/modules/init/bash
+++++ export BASH_ENV
+++++ MODULESHOME=/usr/share/modules
+++++ export MODULESHOME
+++++ LOADEDMODULES=
+++++ export LOADEDMODULES
+++++ MODULEPATH='/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles'
+++++ export MODULEPATH
+++++ test 0
++++ '[' 0 = 1 ']'
++++ '[' -t 2 ']'
++++ export -f module
++++ export -f switchml
++++ '[' 5 -ge 3 ']'
++++ [[ ehxB =~ i ]]
++++ [[ ! :/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin: =~ :/usr/bin: ]]
++++ '[' '!' -n '' ']'
++++ MANPATH=:
++++ export MANPATH
+++++ manpath
++++ [[ ! :/fsx/nouamane/miniconda/envs/2-1-cu121/man:/fsx/nouamane/miniconda/envs/2-1-cu121/share/man:/fsx/nouamane/miniconda/man:/fsx/nouamane/miniconda/share/man:/usr/local/man:/usr/local/share/man:/usr/share/man:/opt/slurm/share/man:/admin/home/nouamane/.local/share/man:/admin/home/nouamane/.fzf/man:: =~ :/usr/share/man: ]]
++++ unset _mlcode _mlret
++++ '[' -n '' ']'
++ module load cuda/12.1
++ unset _mlshdbg
++ '[' 0 = 1 ']'
++ unset _mlre _mlIFS
++ '[' -n x ']'
++ _mlIFS=' 	
+'
++ IFS=' '
++ '[' -n '' ']'
+++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash load cuda/12.1
++ eval 'CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include;' export 'CPATH;
+LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:;' export 'LD_LIBRARY_PATH;
+FI_EFA_FORK_SAFE=1;' export 'FI_EFA_FORK_SAFE;
+MANPATH=/usr/local/cuda-12.1/share/man::;' export 'MANPATH;
+LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64;' export 'LIBRARY_PATH;
+_LMFILES_=/usr/share/modules/modulefiles/cuda/12.1;' export '_LMFILES_;
+LOADEDMODULES=cuda/12.1;' export 'LOADEDMODULES;
+MPI_PATH=/opt/amazon/openmpi;' export 'MPI_PATH;
+NCCL_HOME_modshare=/opt/nccl/build:1;' export 'NCCL_HOME_modshare;
+NCCL_PROTO=simple;' export 'NCCL_PROTO;
+MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1;' export 'MANPATH_modshare;
+LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1;' export 'LIBRARY_PATH_modshare;
+NCCL_SOCKET_IFNAME=enp;' export 'NCCL_SOCKET_IFNAME;
+AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl;' export 'AWS_OFI_NCCL_HOME;
+NCCL_HOME=/opt/nccl/build;' export 'NCCL_HOME;
+FI_PROVIDER=efa;' export 'FI_PROVIDER;
+AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1;' export 'AWS_OFI_NCCL_HOME_modshare;
+CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1;' export 'CPATH_modshare;
+LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2::1:/opt/nccl/build/lib:1:/opt/aws-ofi-nccl/lib:2:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/efa/lib:1;' export 'LD_LIBRARY_PATH_modshare;
+FI_EFA_ENABLE_SHM_TRANSFER=1;' export 'FI_EFA_ENABLE_SHM_TRANSFER;
+_LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1;' export '_LMFILES__modshare;
+LOADEDMODULES_modshare=cuda/12.1:1;' export 'LOADEDMODULES_modshare;
+MPI_PATH_modshare=/opt/amazon/openmpi:1;' export 'MPI_PATH_modshare;
+PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin;' export 'PATH;
+CUDA_HOME=/usr/local/cuda-12.1;' export 'CUDA_HOME;
+PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/opt/slurm/bin:1:/usr/sbin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1;' export 'PATH_modshare;
+test' '0;'
+++ CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include
+++ export CPATH
+++ LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:
+++ export LD_LIBRARY_PATH
+++ FI_EFA_FORK_SAFE=1
+++ export FI_EFA_FORK_SAFE
+++ MANPATH=/usr/local/cuda-12.1/share/man::
+++ export MANPATH
+++ LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64
+++ export LIBRARY_PATH
+++ _LMFILES_=/usr/share/modules/modulefiles/cuda/12.1
+++ export _LMFILES_
+++ LOADEDMODULES=cuda/12.1
+++ export LOADEDMODULES
+++ MPI_PATH=/opt/amazon/openmpi
+++ export MPI_PATH
+++ NCCL_HOME_modshare=/opt/nccl/build:1
+++ export NCCL_HOME_modshare
+++ NCCL_PROTO=simple
+++ export NCCL_PROTO
+++ MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1
+++ export MANPATH_modshare
+++ LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1
+++ export LIBRARY_PATH_modshare
+++ NCCL_SOCKET_IFNAME=enp
+++ export NCCL_SOCKET_IFNAME
+++ AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl
+++ export AWS_OFI_NCCL_HOME
+++ NCCL_HOME=/opt/nccl/build
+++ export NCCL_HOME
+++ FI_PROVIDER=efa
+++ export FI_PROVIDER
+++ AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1
+++ export AWS_OFI_NCCL_HOME_modshare
+++ CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1
+++ export CPATH_modshare
+++ LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2::1:/opt/nccl/build/lib:1:/opt/aws-ofi-nccl/lib:2:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/efa/lib:1
+++ export LD_LIBRARY_PATH_modshare
+++ FI_EFA_ENABLE_SHM_TRANSFER=1
+++ export FI_EFA_ENABLE_SHM_TRANSFER
+++ _LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1
+++ export _LMFILES__modshare
+++ LOADEDMODULES_modshare=cuda/12.1:1
+++ export LOADEDMODULES_modshare
+++ MPI_PATH_modshare=/opt/amazon/openmpi:1
+++ export MPI_PATH_modshare
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
+++ export PATH
+++ CUDA_HOME=/usr/local/cuda-12.1
+++ export CUDA_HOME
+++ PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/opt/slurm/bin:1:/usr/sbin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1
+++ export PATH_modshare
+++ test 0
++ _mlstatus=0
++ '[' -n x ']'
++ IFS=' 	
+'
++ unset _mlre _mlv _mlrv _mlIFS
++ '[' -n '' ']'
++ unset _mlshdbg
++ return 0
++ source /fsx/nouamane/miniconda/bin/activate
+++ _CONDA_ROOT=/fsx/nouamane/miniconda
+++ . /fsx/nouamane/miniconda/etc/profile.d/conda.sh
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ '[' -z x ']'
+++ conda activate
+++ local cmd=activate
+++ case "$cmd" in
+++ __conda_activate activate
+++ '[' -n '' ']'
+++ local ask_conda
++++ PS1=
++++ __conda_exe shell.posix activate
++++ /fsx/nouamane/miniconda/bin/conda shell.posix activate
+Error while loading conda entry point: conda-libmamba-solver (module 'libmambapy' has no attribute 'QueryFormat')
+++ ask_conda='. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''2'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_1='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/etc/conda/activate.d/libxml2_activate.sh"'
+++ eval '. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''2'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_1='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/etc/conda/activate.d/libxml2_activate.sh"'
++++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh
+++++ test -n ''
+++++ unset XML_CATALOG_FILES
+++++ unset xml_catalog_files_libxml2
++++ PS1='(base) '
++++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
++++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
++++ export CONDA_PREFIX=/fsx/nouamane/miniconda
++++ CONDA_PREFIX=/fsx/nouamane/miniconda
++++ export CONDA_SHLVL=2
++++ CONDA_SHLVL=2
++++ export CONDA_DEFAULT_ENV=base
++++ CONDA_DEFAULT_ENV=base
++++ export 'CONDA_PROMPT_MODIFIER=(base) '
++++ CONDA_PROMPT_MODIFIER='(base) '
++++ export CONDA_PREFIX_1=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ CONDA_PREFIX_1=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ . /fsx/nouamane/miniconda/etc/conda/activate.d/libxml2_activate.sh
+++++ test -n ''
+++++ xml_catalog_files_libxml2=
+++++ XML_CATALOG_FILES=
+++++ conda_catalog_files=
+++++ ifs_libxml2=' 	
+'
+++++ IFS=' '
+++++ rem=/fsx/nouamane/miniconda
+++++ for pre in ${rem}
+++++ test '' = /fsx/nouamane/miniconda
+++++ conda_catalog_files=/fsx/nouamane/miniconda
+++++ rem=
+++++ IFS=' 	
+'
+++++ conda_catalog_files='file:///fsx/nouamane/miniconda/etc/xml/catalog file:///etc/xml/catalog'
+++++ export 'XML_CATALOG_FILES=file:///fsx/nouamane/miniconda/etc/xml/catalog file:///etc/xml/catalog'
+++++ XML_CATALOG_FILES='file:///fsx/nouamane/miniconda/etc/xml/catalog file:///etc/xml/catalog'
+++++ unset conda_catalog_files ifs_libxml2 rem
+++ __conda_hashr
+++ '[' -n '' ']'
+++ '[' -n '' ']'
+++ hash -r
++ conda activate 2-1-cu121
++ local cmd=activate
++ case "$cmd" in
++ __conda_activate activate 2-1-cu121
++ '[' -n '' ']'
++ local ask_conda
+++ PS1='(base) '
+++ __conda_exe shell.posix activate 2-1-cu121
+++ /fsx/nouamane/miniconda/bin/conda shell.posix activate 2-1-cu121
+Error while loading conda entry point: conda-libmamba-solver (module 'libmambapy' has no attribute 'QueryFormat')
++ ask_conda='. "/fsx/nouamane/miniconda/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_2='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
++ eval '. "/fsx/nouamane/miniconda/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_2='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
+++ . /fsx/nouamane/miniconda/etc/conda/deactivate.d/libxml2_deactivate.sh
++++ test -n ''
++++ unset XML_CATALOG_FILES
++++ unset xml_catalog_files_libxml2
+++ PS1='(2-1-cu121) '
+++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
+++ export CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ export CONDA_SHLVL=3
+++ CONDA_SHLVL=3
+++ export CONDA_DEFAULT_ENV=2-1-cu121
+++ CONDA_DEFAULT_ENV=2-1-cu121
+++ export 'CONDA_PROMPT_MODIFIER=(2-1-cu121) '
+++ CONDA_PROMPT_MODIFIER='(2-1-cu121) '
+++ export CONDA_PREFIX_2=/fsx/nouamane/miniconda
+++ CONDA_PREFIX_2=/fsx/nouamane/miniconda
+++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ export _CE_M=
+++ _CE_M=
+++ export _CE_CONDA=
+++ _CE_CONDA=
+++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh
++++ test -n ''
++++ xml_catalog_files_libxml2=
++++ XML_CATALOG_FILES=
++++ conda_catalog_files=
++++ ifs_libxml2=' 	
+'
++++ IFS=' '
++++ rem=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ for pre in ${rem}
++++ test '' = /fsx/nouamane/miniconda/envs/2-1-cu121
++++ conda_catalog_files=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ rem=
++++ IFS=' 	
+'
++++ conda_catalog_files='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ export 'XML_CATALOG_FILES=file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ XML_CATALOG_FILES='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ unset conda_catalog_files ifs_libxml2 rem
++ __conda_hashr
++ '[' -n '' ']'
++ '[' -n '' ']'
++ hash -r
++ export PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
++ PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
++ '[' -z '' ']'
+++ scontrol show hostnames 'ip-26-0-163-236,ip-26-0-164-[0,18,45,75,187,207,236],ip-26-0-165-[24,38,59,131,164,202,213],ip-26-0-166-[15,36,125,214,244],ip-26-0-167-[9,51,175,177],ip-26-0-170-[143,160],ip-26-0-171-[21,56,62,88,102,168]'
++ export 'NODELIST=ip-26-0-163-236
+ip-26-0-164-0
+ip-26-0-164-18
+ip-26-0-164-45
+ip-26-0-164-75
+ip-26-0-164-187
+ip-26-0-164-207
+ip-26-0-164-236
+ip-26-0-165-24
+ip-26-0-165-38
+ip-26-0-165-59
+ip-26-0-165-131
+ip-26-0-165-164
+ip-26-0-165-202
+ip-26-0-165-213
+ip-26-0-166-15
+ip-26-0-166-36
+ip-26-0-166-125
+ip-26-0-166-214
+ip-26-0-166-244
+ip-26-0-167-9
+ip-26-0-167-51
+ip-26-0-167-175
+ip-26-0-167-177
+ip-26-0-170-143
+ip-26-0-170-160
+ip-26-0-171-21
+ip-26-0-171-56
+ip-26-0-171-62
+ip-26-0-171-88
+ip-26-0-171-102
+ip-26-0-171-168'
++ NODELIST='ip-26-0-163-236
+ip-26-0-164-0
+ip-26-0-164-18
+ip-26-0-164-45
+ip-26-0-164-75
+ip-26-0-164-187
+ip-26-0-164-207
+ip-26-0-164-236
+ip-26-0-165-24
+ip-26-0-165-38
+ip-26-0-165-59
+ip-26-0-165-131
+ip-26-0-165-164
+ip-26-0-165-202
+ip-26-0-165-213
+ip-26-0-166-15
+ip-26-0-166-36
+ip-26-0-166-125
+ip-26-0-166-214
+ip-26-0-166-244
+ip-26-0-167-9
+ip-26-0-167-51
+ip-26-0-167-175
+ip-26-0-167-177
+ip-26-0-170-143
+ip-26-0-170-160
+ip-26-0-171-21
+ip-26-0-171-56
+ip-26-0-171-62
+ip-26-0-171-88
+ip-26-0-171-102
+ip-26-0-171-168'
+++ head -n1
+++ scontrol show hostnames 'ip-26-0-163-236,ip-26-0-164-[0,18,45,75,187,207,236],ip-26-0-165-[24,38,59,131,164,202,213],ip-26-0-166-[15,36,125,214,244],ip-26-0-167-[9,51,175,177],ip-26-0-170-[143,160],ip-26-0-171-[21,56,62,88,102,168]'
++ export MASTER_NODE=ip-26-0-163-236
++ MASTER_NODE=ip-26-0-163-236
++ export MASTER_PORT=12356
++ MASTER_PORT=12356
++ export NNODES=32
++ NNODES=32
++ export GPUS_PER_NODE=8
++ GPUS_PER_NODE=8
++ export WORLD_SIZE=256
++ WORLD_SIZE=256
++ export CUDA_DEVICE_MAX_CONNECTIONS=1
++ CUDA_DEVICE_MAX_CONNECTIONS=1
++ export NCCL_DEBUG=WARN
++ NCCL_DEBUG=WARN
++ export NANOTRON_BENCHMARK=1
++ NANOTRON_BENCHMARK=1
++ export WANDB_MODE=disabled
++ WANDB_MODE=disabled
++ export TORCH_NCCL_ASYNC_ERROR_HANDLING=1
++ TORCH_NCCL_ASYNC_ERROR_HANDLING=1
++ export TORCH_DISTRIBUTED_DEBUG=DETAIL
++ TORCH_DISTRIBUTED_DEBUG=DETAIL
++ '[' -z '' ']'
++ echo '=== GPU Topology ==='
+=== GPU Topology ===
++ nvidia-smi topo -m
+	GPU0	GPU1	GPU2	GPU3	GPU4	GPU5	GPU6	GPU7	CPU Affinity	NUMA Affinity	GPU NUMA ID
+GPU0	 X 	NV18	NV18	NV18	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU1	NV18	 X 	NV18	NV18	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU2	NV18	NV18	 X 	NV18	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU3	NV18	NV18	NV18	 X 	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU4	NV18	NV18	NV18	NV18	 X 	NV18	NV18	NV18	48-95	1		N/A
+GPU5	NV18	NV18	NV18	NV18	NV18	 X 	NV18	NV18	48-95	1		N/A
+GPU6	NV18	NV18	NV18	NV18	NV18	NV18	 X 	NV18	48-95	1		N/A
+GPU7	NV18	NV18	NV18	NV18	NV18	NV18	NV18	 X 	48-95	1		N/A
+
+Legend:
+
+  X    = Self
+  SYS  = Connection traversing PCIe as well as the SMP interconnect between NUMA nodes (e.g., QPI/UPI)
+  NODE = Connection traversing PCIe as well as the interconnect between PCIe Host Bridges within a NUMA node
+  PHB  = Connection traversing PCIe as well as a PCIe Host Bridge (typically the CPU)
+  PXB  = Connection traversing multiple PCIe bridges (without traversing the PCIe Host Bridge)
+  PIX  = Connection traversing at most a single PCIe bridge
+  NV#  = Connection traversing a bonded set of # NVLinks
++ echo ==================
+==================
++ export SRUN_ALLOC_ARGS=
++ SRUN_ALLOC_ARGS=
++ echo 'Master node: ip-26-0-163-236'
+Master node: ip-26-0-163-236
++ echo 'All nodes: ip-26-0-163-236
+ip-26-0-164-0
+ip-26-0-164-18
+ip-26-0-164-45
+ip-26-0-164-75
+ip-26-0-164-187
+ip-26-0-164-207
+ip-26-0-164-236
+ip-26-0-165-24
+ip-26-0-165-38
+ip-26-0-165-59
+ip-26-0-165-131
+ip-26-0-165-164
+ip-26-0-165-202
+ip-26-0-165-213
+ip-26-0-166-15
+ip-26-0-166-36
+ip-26-0-166-125
+ip-26-0-166-214
+ip-26-0-166-244
+ip-26-0-167-9
+ip-26-0-167-51
+ip-26-0-167-175
+ip-26-0-167-177
+ip-26-0-170-143
+ip-26-0-170-160
+ip-26-0-171-21
+ip-26-0-171-56
+ip-26-0-171-62
+ip-26-0-171-88
+ip-26-0-171-102
+ip-26-0-171-168'
+All nodes: ip-26-0-163-236
+ip-26-0-164-0
+ip-26-0-164-18
+ip-26-0-164-45
+ip-26-0-164-75
+ip-26-0-164-187
+ip-26-0-164-207
+ip-26-0-164-236
+ip-26-0-165-24
+ip-26-0-165-38
+ip-26-0-165-59
+ip-26-0-165-131
+ip-26-0-165-164
+ip-26-0-165-202
+ip-26-0-165-213
+ip-26-0-166-15
+ip-26-0-166-36
+ip-26-0-166-125
+ip-26-0-166-214
+ip-26-0-166-244
+ip-26-0-167-9
+ip-26-0-167-51
+ip-26-0-167-175
+ip-26-0-167-177
+ip-26-0-170-143
+ip-26-0-170-160
+ip-26-0-171-21
+ip-26-0-171-56
+ip-26-0-171-62
+ip-26-0-171-88
+ip-26-0-171-102
+ip-26-0-171-168
++ echo 'World size: 256'
+World size: 256
++ '[' -n '' ']'
++ srun --wait=0 --kill-on-bad-exit=1 torchrun --nnodes=32 --nproc_per_node=8 --rdzv_id=14019308 --rdzv_backend=c10d --rdzv_endpoint=ip-26-0-163-236:12356 --max_restarts 0 --rdzv_conf timeout=60 /fsx/nouamane/projects/nanotron/run_train.py --config-file benchmark/configs/config_469G_dp4_tp8_pp8_acc64_mbs1_seq4096_zero1_tpmodeRED_vocab131k.yaml
+[2025-01-03 01:30:01,272] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 01:30:01,272] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 01:30:01,273] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 01:30:01,273] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 01:30:01,272] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 01:30:01,273] torch.distributed.run: [WARNING] 
+[2025-01-03 01:30:01,273] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 01:30:01,273] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 01:30:01,273] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 01:30:01,273] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 01:30:01,272] torch.distributed.run: [WARNING] 
+[2025-01-03 01:30:01,272] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 01:30:01,272] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 01:30:01,272] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 01:30:01,274] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 01:30:01,274] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 01:30:01,275] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 01:30:01,275] torch.distributed.run: [WARNING] 
+[2025-01-03 01:30:01,275] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 01:30:01,275] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 01:30:01,275] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 01:30:01,273] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 01:30:01,273] torch.distributed.run: [WARNING] 
+[2025-01-03 01:30:01,273] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 01:30:01,273] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 01:30:01,273] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 01:30:01,274] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 01:30:01,275] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 01:30:01,272] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 01:30:01,278] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 01:30:01,277] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 01:30:01,279] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 01:30:01,280] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 01:30:01,281] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 01:30:01,283] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 01:30:01,283] torch.distributed.run: [WARNING] 
+[2025-01-03 01:30:01,283] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 01:30:01,283] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 01:30:01,283] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 01:30:01,282] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 01:30:01,282] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 01:30:01,284] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 01:30:01,301] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 01:30:01,301] torch.distributed.run: [WARNING] 
+[2025-01-03 01:30:01,301] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 01:30:01,301] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 01:30:01,301] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 01:30:01,274] torch.distributed.run: [WARNING] 
+[2025-01-03 01:30:01,274] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 01:30:01,274] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 01:30:01,274] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 01:30:01,273] torch.distributed.run: [WARNING] 
+[2025-01-03 01:30:01,273] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 01:30:01,273] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 01:30:01,273] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 01:30:01,273] torch.distributed.run: [WARNING] 
+[2025-01-03 01:30:01,273] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 01:30:01,273] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 01:30:01,273] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 01:30:01,273] torch.distributed.run: [WARNING] 
+[2025-01-03 01:30:01,273] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 01:30:01,273] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 01:30:01,273] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 01:30:01,272] torch.distributed.run: [WARNING] 
+[2025-01-03 01:30:01,272] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 01:30:01,272] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 01:30:01,272] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 01:30:01,274] torch.distributed.run: [WARNING] 
+[2025-01-03 01:30:01,274] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 01:30:01,274] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 01:30:01,274] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 01:30:01,316] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 01:30:01,275] torch.distributed.run: [WARNING] 
+[2025-01-03 01:30:01,275] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 01:30:01,275] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 01:30:01,275] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 01:30:01,274] torch.distributed.run: [WARNING] 
+[2025-01-03 01:30:01,274] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 01:30:01,274] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 01:30:01,274] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 01:30:01,272] torch.distributed.run: [WARNING] 
+[2025-01-03 01:30:01,272] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 01:30:01,272] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 01:30:01,272] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 01:30:01,277] torch.distributed.run: [WARNING] 
+[2025-01-03 01:30:01,277] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 01:30:01,277] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 01:30:01,277] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 01:30:01,281] torch.distributed.run: [WARNING] 
+[2025-01-03 01:30:01,281] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 01:30:01,281] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 01:30:01,281] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 01:30:01,280] torch.distributed.run: [WARNING] 
+[2025-01-03 01:30:01,280] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 01:30:01,280] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 01:30:01,280] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 01:30:01,278] torch.distributed.run: [WARNING] 
+[2025-01-03 01:30:01,278] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 01:30:01,278] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 01:30:01,278] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 01:30:01,279] torch.distributed.run: [WARNING] 
+[2025-01-03 01:30:01,279] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 01:30:01,279] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 01:30:01,279] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 01:30:01,326] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 01:30:01,326] torch.distributed.run: [WARNING] 
+[2025-01-03 01:30:01,326] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 01:30:01,326] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 01:30:01,326] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 01:30:01,282] torch.distributed.run: [WARNING] 
+[2025-01-03 01:30:01,282] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 01:30:01,282] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 01:30:01,282] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 01:30:01,284] torch.distributed.run: [WARNING] 
+[2025-01-03 01:30:01,284] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 01:30:01,284] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 01:30:01,284] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 01:30:01,283] torch.distributed.run: [WARNING] 
+[2025-01-03 01:30:01,283] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 01:30:01,283] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 01:30:01,283] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 01:30:01,335] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 01:30:01,316] torch.distributed.run: [WARNING] 
+[2025-01-03 01:30:01,316] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 01:30:01,316] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 01:30:01,316] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 01:30:01,362] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 01:30:01,368] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 01:30:01,336] torch.distributed.run: [WARNING] 
+[2025-01-03 01:30:01,336] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 01:30:01,336] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 01:30:01,336] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 01:30:01,362] torch.distributed.run: [WARNING] 
+[2025-01-03 01:30:01,362] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 01:30:01,362] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 01:30:01,362] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 01:30:01,368] torch.distributed.run: [WARNING] 
+[2025-01-03 01:30:01,368] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 01:30:01,368] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 01:30:01,368] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 01:30:01,410] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 01:30:01,410] torch.distributed.run: [WARNING] 
+[2025-01-03 01:30:01,410] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 01:30:01,410] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 01:30:01,410] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 01:30:01,459] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 01:30:01,484] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 01:30:01,484] torch.distributed.run: [WARNING] 
+[2025-01-03 01:30:01,484] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 01:30:01,484] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 01:30:01,484] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 01:30:01,459] torch.distributed.run: [WARNING] 
+[2025-01-03 01:30:01,459] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 01:30:01,459] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 01:30:01,459] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 01:30:01,515] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 01:30:01,516] torch.distributed.run: [WARNING] 
+[2025-01-03 01:30:01,516] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 01:30:01,516] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 01:30:01,516] torch.distributed.run: [WARNING] *****************************************
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+01/03/2025 01:30:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: Config:
+01/03/2025 01:30:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: Config(general=GeneralArgs(project='debug',
+01/03/2025 01:30:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                            run='469G_dp4_tp8_pp8_acc64_mbs1_seq4096_zero1_tpmodeRED_vocab131k',
+01/03/2025 01:30:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                            seed=42,
+01/03/2025 01:30:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                            step=None,
+01/03/2025 01:30:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                            consumed_train_samples=None,
+01/03/2025 01:30:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                            benchmark_csv_path=PosixPath('benchmark/results/bench_final2.csv'),
+01/03/2025 01:30:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                            ignore_sanity_checks=True),
+01/03/2025 01:30:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:        parallelism=ParallelismArgs(dp=4,
+01/03/2025 01:30:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                    pp=8,
+01/03/2025 01:30:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                    tp=8,
+01/03/2025 01:30:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                    pp_engine=<nanotron.parallel.pipeline_parallel.engine.OneForwardOneBackwardPipelineEngine object at 0x7fc30a71cdc0>,
+01/03/2025 01:30:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                    tp_mode=<TensorParallelLinearMode.REDUCE_SCATTER: 2>,
+01/03/2025 01:30:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                    tp_linear_async_communication=True,
+01/03/2025 01:30:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                    recompute_layer=False,
+01/03/2025 01:30:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                    tp_recompute_allgather=True,
+01/03/2025 01:30:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                    expert_parallel_size=1),
+01/03/2025 01:30:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:        model=ModelArgs(model_config=LlamaConfig(bos_token_id=0,
+01/03/2025 01:30:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                                 eos_token_id=0,
+01/03/2025 01:30:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                                 hidden_act='silu',
+01/03/2025 01:30:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                                 hidden_size=16384,
+01/03/2025 01:30:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                                 initializer_range=0.02,
+01/03/2025 01:30:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                                 intermediate_size=53248,
+01/03/2025 01:30:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                                 is_llama_config=True,
+01/03/2025 01:30:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                                 max_position_embeddings=4096,
+01/03/2025 01:30:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                                 num_attention_heads=128,
+01/03/2025 01:30:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                                 num_hidden_layers=126,
+01/03/2025 01:30:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                                 num_key_value_heads=128,
+01/03/2025 01:30:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                                 pad_token_id=None,
+01/03/2025 01:30:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                                 pretraining_tp=1,
+01/03/2025 01:30:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                                 rms_norm_eps=1e-05,
+01/03/2025 01:30:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                                 rope_scaling=None,
+01/03/2025 01:30:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                                 rope_theta=10000.0,
+01/03/2025 01:30:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                                 rope_interleaved=False,
+01/03/2025 01:30:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                                 tie_word_embeddings=False,
+01/03/2025 01:30:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                                 use_cache=True,
+01/03/2025 01:30:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                                 vocab_size=131072),
+01/03/2025 01:30:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                        init_method=RandomInit(std=0.02),
+01/03/2025 01:30:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                        dtype=torch.bfloat16,
+01/03/2025 01:30:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                        make_vocab_size_divisible_by=1,
+01/03/2025 01:30:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                        ddp_bucket_cap_mb=25),
+01/03/2025 01:30:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:        tokenizer=TokenizerArgs(tokenizer_name_or_path='robot-test/dummy-tokenizer-wordlevel',
+01/03/2025 01:30:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                tokenizer_revision=None,
+01/03/2025 01:30:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                tokenizer_max_length=None),
+01/03/2025 01:30:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:        checkpoints=CheckpointsArgs(checkpoints_path=PosixPath('checkpoints'),
+01/03/2025 01:30:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                    checkpoint_interval=10000,
+01/03/2025 01:30:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                    save_initial_state=False,
+01/03/2025 01:30:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                    save_final_state=False,
+01/03/2025 01:30:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                    resume_checkpoint_path=None,
+01/03/2025 01:30:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                    checkpoints_path_is_shared_file_system=False),
+01/03/2025 01:30:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:        logging=LoggingArgs(log_level='info',
+01/03/2025 01:30:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                            log_level_replica='info',
+01/03/2025 01:30:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                            iteration_step_info_interval=1),
+01/03/2025 01:30:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:        tokens=TokensArgs(sequence_length=4096,
+01/03/2025 01:30:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                          train_steps=100,
+01/03/2025 01:30:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                          micro_batch_size=1,
+01/03/2025 01:30:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                          batch_accumulation_per_replica=64,
+01/03/2025 01:30:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                          val_check_interval=100,
+01/03/2025 01:30:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                          limit_val_batches=0,
+01/03/2025 01:30:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                          limit_test_batches=0),
+01/03/2025 01:30:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:        optimizer=OptimizerArgs(optimizer_factory=AdamWOptimizerArgs(adam_eps=1e-08,
+01/03/2025 01:30:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                                                     adam_beta1=0.9,
+01/03/2025 01:30:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                                                     adam_beta2=0.95,
+01/03/2025 01:30:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                                                     torch_adam_is_fused=True,
+01/03/2025 01:30:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                                                     name='adamW'),
+01/03/2025 01:30:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                zero_stage=1,
+01/03/2025 01:30:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                weight_decay=0.01,
+01/03/2025 01:30:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                clip_grad=1.0,
+01/03/2025 01:30:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                accumulate_grad_in_fp32=True,
+01/03/2025 01:30:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                learning_rate_scheduler=LRSchedulerArgs(learning_rate=0.0003,
+01/03/2025 01:30:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                                                        lr_warmup_steps=2,
+01/03/2025 01:30:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                                                        lr_warmup_style='linear',
+01/03/2025 01:30:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                                                        lr_decay_style='cosine',
+01/03/2025 01:30:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                                                        lr_decay_steps=13,
+01/03/2025 01:30:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                                                        lr_decay_starting_step=None,
+01/03/2025 01:30:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                                                        min_decay_lr=1e-05)),
+01/03/2025 01:30:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:        data_stages=[DatasetStageArgs(name='Stable Training Stage',
+01/03/2025 01:30:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                      start_training_step=1,
+01/03/2025 01:30:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                      data=DataArgs(dataset=None,
+01/03/2025 01:30:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                                    seed=42,
+01/03/2025 01:30:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                                    num_loading_workers=1))],
+01/03/2025 01:30:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:        profiler=None,
+01/03/2025 01:30:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:        lighteval=None,
+01/03/2025 01:30:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:        s3_upload=None)
+01/03/2025 01:30:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: Model Config:
+01/03/2025 01:30:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: LlamaConfig(bos_token_id=0,
+01/03/2025 01:30:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:             eos_token_id=0,
+01/03/2025 01:30:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:             hidden_act='silu',
+01/03/2025 01:30:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:             hidden_size=16384,
+01/03/2025 01:30:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:             initializer_range=0.02,
+01/03/2025 01:30:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:             intermediate_size=53248,
+01/03/2025 01:30:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:             is_llama_config=True,
+01/03/2025 01:30:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:             max_position_embeddings=4096,
+01/03/2025 01:30:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:             num_attention_heads=128,
+01/03/2025 01:30:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:             num_hidden_layers=126,
+01/03/2025 01:30:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:             num_key_value_heads=128,
+01/03/2025 01:30:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:             pad_token_id=None,
+01/03/2025 01:30:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:             pretraining_tp=1,
+01/03/2025 01:30:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:             rms_norm_eps=1e-05,
+01/03/2025 01:30:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:             rope_scaling=None,
+01/03/2025 01:30:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:             rope_theta=10000.0,
+01/03/2025 01:30:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:             rope_interleaved=False,
+01/03/2025 01:30:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:             tie_word_embeddings=False,
+01/03/2025 01:30:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:             use_cache=True,
+01/03/2025 01:30:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:             vocab_size=131072)
+01/03/2025 01:30:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: Building model..
+01/03/2025 01:30:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: Initialize RoPE Theta = 10000.0
+01/03/2025 01:30:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: Setting PP block ranks...
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+01/03/2025 01:31:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: Total number of parameters: 469G (895295.25MiB)
+01/03/2025 01:31:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: Local number of parameters: 7.65G (14593.00MiB)
+01/03/2025 01:31:02 [INFO|DP=0|PP=7|TP=0|ip-26-0-171-21]: Local number of parameters: 7.19G (13712.97MiB)
+01/03/2025 01:31:02 [INFO|DP=0|PP=4|TP=0|ip-26-0-166-15]: Local number of parameters: 7.38G (14081.00MiB)
+01/03/2025 01:31:02 [INFO|DP=0|PP=5|TP=0|ip-26-0-167-175]: Local number of parameters: 6.92G (13200.94MiB)
+01/03/2025 01:31:02 [INFO|DP=0|PP=3|TP=0|ip-26-0-165-24]: Local number of parameters: 7.38G (14081.00MiB)
+01/03/2025 01:31:02 [INFO|DP=0|PP=6|TP=0|ip-26-0-170-143]: Local number of parameters: 7.38G (14081.00MiB)
+01/03/2025 01:31:02 [INFO|DP=0|PP=2|TP=0|ip-26-0-165-131]: Local number of parameters: 7.38G (14081.00MiB)
+01/03/2025 01:31:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: [After model building] Memory usage: 14593.03MiB. Peak allocated: 14593.15MiB Peak reserved: 14726.00MiB
+01/03/2025 01:31:02 [INFO|DP=0|PP=7|TP=0|ip-26-0-171-21]: [After model building] Memory usage: 13712.99MiB. Peak allocated: 13713.12MiB Peak reserved: 13814.00MiB
+01/03/2025 01:31:02 [INFO|DP=0|PP=3|TP=0|ip-26-0-165-24]: [After model building] Memory usage: 14081.03MiB. Peak allocated: 14081.15MiB Peak reserved: 14182.00MiB
+01/03/2025 01:31:02 [INFO|DP=0|PP=4|TP=0|ip-26-0-166-15]: [After model building] Memory usage: 14081.03MiB. Peak allocated: 14081.15MiB Peak reserved: 14182.00MiB
+01/03/2025 01:31:02 [INFO|DP=0|PP=6|TP=0|ip-26-0-170-143]: [After model building] Memory usage: 14081.03MiB. Peak allocated: 14081.15MiB Peak reserved: 14182.00MiB
+01/03/2025 01:31:02 [INFO|DP=0|PP=5|TP=0|ip-26-0-167-175]: [After model building] Memory usage: 13200.96MiB. Peak allocated: 13201.09MiB Peak reserved: 13302.00MiB
+01/03/2025 01:31:02 [INFO|DP=0|PP=2|TP=0|ip-26-0-165-131]: [After model building] Memory usage: 14081.03MiB. Peak allocated: 14081.15MiB Peak reserved: 14182.00MiB
+01/03/2025 01:31:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: No checkpoint path provided.
+01/03/2025 01:31:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: Parametrizing model parameters using StandardParametrizator
+01/03/2025 01:31:02 [INFO|DP=0|PP=1|TP=0|ip-26-0-164-207]: Local number of parameters: 7.38G (14081.00MiB)
+01/03/2025 01:31:02 [INFO|DP=0|PP=1|TP=0|ip-26-0-164-207]: [After model building] Memory usage: 14081.03MiB. Peak allocated: 14081.15MiB Peak reserved: 14182.00MiB
+01/03/2025 01:31:02 [INFO|DP=0|PP=0|TP=5|ip-26-0-163-236]: Local number of parameters: 7.65G (14593.00MiB)
+01/03/2025 01:31:02 [INFO|DP=0|PP=0|TP=5|ip-26-0-163-236]: [After model building] Memory usage: 14593.03MiB. Peak allocated: 14593.15MiB Peak reserved: 14726.00MiB
+01/03/2025 01:31:02 [INFO|DP=0|PP=0|TP=7|ip-26-0-163-236]: Local number of parameters: 7.65G (14593.00MiB)
+01/03/2025 01:31:02 [INFO|DP=0|PP=0|TP=1|ip-26-0-163-236]: Local number of parameters: 7.65G (14593.00MiB)
+01/03/2025 01:31:02 [INFO|DP=0|PP=0|TP=4|ip-26-0-163-236]: Local number of parameters: 7.65G (14593.00MiB)
+01/03/2025 01:31:02 [INFO|DP=0|PP=0|TP=7|ip-26-0-163-236]: [After model building] Memory usage: 14593.03MiB. Peak allocated: 14593.15MiB Peak reserved: 14726.00MiB
+01/03/2025 01:31:02 [INFO|DP=0|PP=0|TP=1|ip-26-0-163-236]: [After model building] Memory usage: 14593.03MiB. Peak allocated: 14593.15MiB Peak reserved: 14726.00MiB
+01/03/2025 01:31:02 [INFO|DP=0|PP=0|TP=4|ip-26-0-163-236]: [After model building] Memory usage: 14593.03MiB. Peak allocated: 14593.15MiB Peak reserved: 14726.00MiB
+01/03/2025 01:31:02 [INFO|DP=0|PP=0|TP=2|ip-26-0-163-236]: Local number of parameters: 7.65G (14593.00MiB)
+01/03/2025 01:31:02 [INFO|DP=0|PP=0|TP=2|ip-26-0-163-236]: [After model building] Memory usage: 14593.03MiB. Peak allocated: 14593.15MiB Peak reserved: 14726.00MiB
+01/03/2025 01:31:02 [INFO|DP=0|PP=0|TP=3|ip-26-0-163-236]: Local number of parameters: 7.65G (14593.00MiB)
+01/03/2025 01:31:02 [INFO|DP=0|PP=0|TP=3|ip-26-0-163-236]: [After model building] Memory usage: 14593.03MiB. Peak allocated: 14593.15MiB Peak reserved: 14726.00MiB
+01/03/2025 01:31:02 [INFO|DP=0|PP=0|TP=6|ip-26-0-163-236]: Local number of parameters: 7.65G (14593.00MiB)
+01/03/2025 01:31:02 [INFO|DP=0|PP=0|TP=6|ip-26-0-163-236]: [After model building] Memory usage: 14593.03MiB. Peak allocated: 14593.15MiB Peak reserved: 14726.00MiB
+01/03/2025 01:31:02 [INFO|DP=0|PP=6|TP=1|ip-26-0-170-143]: Local number of parameters: 7.38G (14081.00MiB)
+01/03/2025 01:31:02 [INFO|DP=0|PP=7|TP=1|ip-26-0-171-21]: Local number of parameters: 7.19G (13712.97MiB)
+01/03/2025 01:31:02 [INFO|DP=0|PP=4|TP=1|ip-26-0-166-15]: Local number of parameters: 7.38G (14081.00MiB)
+01/03/2025 01:31:02 [INFO|DP=0|PP=4|TP=4|ip-26-0-166-15]: Local number of parameters: 7.38G (14081.00MiB)
+01/03/2025 01:31:02 [INFO|DP=0|PP=3|TP=4|ip-26-0-165-24]: Local number of parameters: 7.38G (14081.00MiB)
+01/03/2025 01:31:02 [INFO|DP=0|PP=2|TP=4|ip-26-0-165-131]: Local number of parameters: 7.38G (14081.00MiB)
+01/03/2025 01:31:02 [INFO|DP=0|PP=1|TP=4|ip-26-0-164-207]: Local number of parameters: 7.38G (14081.00MiB)
+01/03/2025 01:31:02 [INFO|DP=0|PP=2|TP=1|ip-26-0-165-131]: Local number of parameters: 7.38G (14081.00MiB)
+01/03/2025 01:31:02 [INFO|DP=0|PP=6|TP=7|ip-26-0-170-143]: Local number of parameters: 7.38G (14081.00MiB)
+01/03/2025 01:31:02 [INFO|DP=0|PP=3|TP=7|ip-26-0-165-24]: Local number of parameters: 7.38G (14081.00MiB)
+01/03/2025 01:31:02 [INFO|DP=0|PP=1|TP=1|ip-26-0-164-207]: Local number of parameters: 7.38G (14081.00MiB)
+01/03/2025 01:31:02 [INFO|DP=0|PP=4|TP=7|ip-26-0-166-15]: Local number of parameters: 7.38G (14081.00MiB)
+01/03/2025 01:31:02 [INFO|DP=0|PP=7|TP=7|ip-26-0-171-21]: Local number of parameters: 7.19G (13712.97MiB)
+01/03/2025 01:31:02 [INFO|DP=0|PP=2|TP=7|ip-26-0-165-131]: Local number of parameters: 7.38G (14081.00MiB)
+01/03/2025 01:31:02 [INFO|DP=0|PP=6|TP=1|ip-26-0-170-143]: [After model building] Memory usage: 14081.03MiB. Peak allocated: 14081.15MiB Peak reserved: 14182.00MiB
+01/03/2025 01:31:02 [INFO|DP=0|PP=4|TP=1|ip-26-0-166-15]: [After model building] Memory usage: 14081.03MiB. Peak allocated: 14081.15MiB Peak reserved: 14182.00MiB
+01/03/2025 01:31:02 [INFO|DP=0|PP=4|TP=4|ip-26-0-166-15]: [After model building] Memory usage: 14081.03MiB. Peak allocated: 14081.15MiB Peak reserved: 14182.00MiB
+01/03/2025 01:31:02 [INFO|DP=0|PP=1|TP=7|ip-26-0-164-207]: Local number of parameters: 7.38G (14081.00MiB)
+01/03/2025 01:31:02 [INFO|DP=0|PP=7|TP=1|ip-26-0-171-21]: [After model building] Memory usage: 13712.99MiB. Peak allocated: 13713.12MiB Peak reserved: 13814.00MiB
+01/03/2025 01:31:02 [INFO|DP=0|PP=4|TP=5|ip-26-0-166-15]: Local number of parameters: 7.38G (14081.00MiB)
+01/03/2025 01:31:02 [INFO|DP=0|PP=7|TP=5|ip-26-0-171-21]: Local number of parameters: 7.19G (13712.97MiB)
+01/03/2025 01:31:02 [INFO|DP=0|PP=3|TP=4|ip-26-0-165-24]: [After model building] Memory usage: 14081.03MiB. Peak allocated: 14081.15MiB Peak reserved: 14182.00MiB
+01/03/2025 01:31:02 [INFO|DP=0|PP=5|TP=7|ip-26-0-167-175]: Local number of parameters: 6.92G (13200.94MiB)
+01/03/2025 01:31:02 [INFO|DP=0|PP=2|TP=4|ip-26-0-165-131]: [After model building] Memory usage: 14081.03MiB. Peak allocated: 14081.15MiB Peak reserved: 14182.00MiB
+01/03/2025 01:31:02 [INFO|DP=0|PP=2|TP=5|ip-26-0-165-131]: Local number of parameters: 7.38G (14081.00MiB)
+01/03/2025 01:31:02 [INFO|DP=0|PP=1|TP=4|ip-26-0-164-207]: [After model building] Memory usage: 14081.03MiB. Peak allocated: 14081.15MiB Peak reserved: 14182.00MiB
+01/03/2025 01:31:02 [INFO|DP=0|PP=3|TP=5|ip-26-0-165-24]: Local number of parameters: 7.38G (14081.00MiB)
+01/03/2025 01:31:02 [INFO|DP=0|PP=6|TP=5|ip-26-0-170-143]: Local number of parameters: 7.38G (14081.00MiB)
+01/03/2025 01:31:02 [INFO|DP=0|PP=2|TP=1|ip-26-0-165-131]: [After model building] Memory usage: 14081.03MiB. Peak allocated: 14081.15MiB Peak reserved: 14182.00MiB
+01/03/2025 01:31:02 [INFO|DP=0|PP=5|TP=5|ip-26-0-167-175]: Local number of parameters: 6.92G (13200.94MiB)
+01/03/2025 01:31:02 [INFO|DP=0|PP=6|TP=7|ip-26-0-170-143]: [After model building] Memory usage: 14081.03MiB. Peak allocated: 14081.15MiB Peak reserved: 14182.00MiB
+01/03/2025 01:31:02 [INFO|DP=0|PP=1|TP=5|ip-26-0-164-207]: Local number of parameters: 7.38G (14081.00MiB)
+01/03/2025 01:31:02 [INFO|DP=0|PP=1|TP=1|ip-26-0-164-207]: [After model building] Memory usage: 14081.03MiB. Peak allocated: 14081.15MiB Peak reserved: 14182.00MiB
+01/03/2025 01:31:02 [INFO|DP=0|PP=3|TP=7|ip-26-0-165-24]: [After model building] Memory usage: 14081.03MiB. Peak allocated: 14081.15MiB Peak reserved: 14182.00MiB
+01/03/2025 01:31:02 [INFO|DP=0|PP=4|TP=7|ip-26-0-166-15]: [After model building] Memory usage: 14081.03MiB. Peak allocated: 14081.15MiB Peak reserved: 14182.00MiB
+01/03/2025 01:31:02 [INFO|DP=0|PP=2|TP=7|ip-26-0-165-131]: [After model building] Memory usage: 14081.03MiB. Peak allocated: 14081.15MiB Peak reserved: 14182.00MiB
+01/03/2025 01:31:02 [INFO|DP=0|PP=7|TP=7|ip-26-0-171-21]: [After model building] Memory usage: 13712.99MiB. Peak allocated: 13713.12MiB Peak reserved: 13814.00MiB
+01/03/2025 01:31:02 [INFO|DP=0|PP=1|TP=7|ip-26-0-164-207]: [After model building] Memory usage: 14081.03MiB. Peak allocated: 14081.15MiB Peak reserved: 14182.00MiB
+01/03/2025 01:31:02 [INFO|DP=0|PP=4|TP=5|ip-26-0-166-15]: [After model building] Memory usage: 14081.03MiB. Peak allocated: 14081.15MiB Peak reserved: 14182.00MiB
+01/03/2025 01:31:02 [INFO|DP=0|PP=7|TP=5|ip-26-0-171-21]: [After model building] Memory usage: 13712.99MiB. Peak allocated: 13713.12MiB Peak reserved: 13814.00MiB
+01/03/2025 01:31:02 [INFO|DP=0|PP=5|TP=7|ip-26-0-167-175]: [After model building] Memory usage: 13200.96MiB. Peak allocated: 13201.09MiB Peak reserved: 13302.00MiB
+01/03/2025 01:31:02 [INFO|DP=0|PP=6|TP=5|ip-26-0-170-143]: [After model building] Memory usage: 14081.03MiB. Peak allocated: 14081.15MiB Peak reserved: 14182.00MiB
+01/03/2025 01:31:02 [INFO|DP=0|PP=2|TP=5|ip-26-0-165-131]: [After model building] Memory usage: 14081.03MiB. Peak allocated: 14081.15MiB Peak reserved: 14182.00MiB
+01/03/2025 01:31:02 [INFO|DP=0|PP=3|TP=5|ip-26-0-165-24]: [After model building] Memory usage: 14081.03MiB. Peak allocated: 14081.15MiB Peak reserved: 14182.00MiB
+01/03/2025 01:31:02 [INFO|DP=0|PP=7|TP=4|ip-26-0-171-21]: Local number of parameters: 7.19G (13712.97MiB)
+01/03/2025 01:31:02 [INFO|DP=0|PP=5|TP=5|ip-26-0-167-175]: [After model building] Memory usage: 13200.96MiB. Peak allocated: 13201.09MiB Peak reserved: 13302.00MiB
+01/03/2025 01:31:02 [INFO|DP=0|PP=1|TP=5|ip-26-0-164-207]: [After model building] Memory usage: 14081.03MiB. Peak allocated: 14081.15MiB Peak reserved: 14182.00MiB
+01/03/2025 01:31:02 [INFO|DP=0|PP=5|TP=4|ip-26-0-167-175]: Local number of parameters: 6.92G (13200.94MiB)
+01/03/2025 01:31:02 [INFO|DP=0|PP=6|TP=4|ip-26-0-170-143]: Local number of parameters: 7.38G (14081.00MiB)
+01/03/2025 01:31:02 [INFO|DP=0|PP=7|TP=4|ip-26-0-171-21]: [After model building] Memory usage: 13712.99MiB. Peak allocated: 13713.12MiB Peak reserved: 13814.00MiB
+01/03/2025 01:31:02 [INFO|DP=0|PP=5|TP=4|ip-26-0-167-175]: [After model building] Memory usage: 13200.96MiB. Peak allocated: 13201.09MiB Peak reserved: 13302.00MiB
+01/03/2025 01:31:02 [INFO|DP=0|PP=6|TP=4|ip-26-0-170-143]: [After model building] Memory usage: 14081.03MiB. Peak allocated: 14081.15MiB Peak reserved: 14182.00MiB
+01/03/2025 01:31:02 [INFO|DP=0|PP=1|TP=3|ip-26-0-164-207]: Local number of parameters: 7.38G (14081.00MiB)
+01/03/2025 01:31:02 [INFO|DP=0|PP=6|TP=3|ip-26-0-170-143]: Local number of parameters: 7.38G (14081.00MiB)
+01/03/2025 01:31:02 [INFO|DP=0|PP=4|TP=2|ip-26-0-166-15]: Local number of parameters: 7.38G (14081.00MiB)
+01/03/2025 01:31:02 [INFO|DP=0|PP=3|TP=3|ip-26-0-165-24]: Local number of parameters: 7.38G (14081.00MiB)
+01/03/2025 01:31:02 [INFO|DP=0|PP=7|TP=3|ip-26-0-171-21]: Local number of parameters: 7.19G (13712.97MiB)
+01/03/2025 01:31:02 [INFO|DP=0|PP=1|TP=2|ip-26-0-164-207]: Local number of parameters: 7.38G (14081.00MiB)
+01/03/2025 01:31:02 [INFO|DP=0|PP=2|TP=3|ip-26-0-165-131]: Local number of parameters: 7.38G (14081.00MiB)
+01/03/2025 01:31:02 [INFO|DP=0|PP=2|TP=2|ip-26-0-165-131]: Local number of parameters: 7.38G (14081.00MiB)
+01/03/2025 01:31:02 [INFO|DP=0|PP=4|TP=3|ip-26-0-166-15]: Local number of parameters: 7.38G (14081.00MiB)
+01/03/2025 01:31:02 [INFO|DP=0|PP=7|TP=2|ip-26-0-171-21]: Local number of parameters: 7.19G (13712.97MiB)
+01/03/2025 01:31:02 [INFO|DP=0|PP=5|TP=3|ip-26-0-167-175]: Local number of parameters: 6.92G (13200.94MiB)
+01/03/2025 01:31:02 [INFO|DP=0|PP=6|TP=2|ip-26-0-170-143]: Local number of parameters: 7.38G (14081.00MiB)
+01/03/2025 01:31:02 [INFO|DP=0|PP=3|TP=2|ip-26-0-165-24]: Local number of parameters: 7.38G (14081.00MiB)
+01/03/2025 01:31:02 [INFO|DP=0|PP=1|TP=3|ip-26-0-164-207]: [After model building] Memory usage: 14081.03MiB. Peak allocated: 14081.15MiB Peak reserved: 14182.00MiB
+01/03/2025 01:31:02 [INFO|DP=0|PP=5|TP=2|ip-26-0-167-175]: Local number of parameters: 6.92G (13200.94MiB)
+01/03/2025 01:31:02 [INFO|DP=0|PP=4|TP=2|ip-26-0-166-15]: [After model building] Memory usage: 14081.03MiB. Peak allocated: 14081.15MiB Peak reserved: 14182.00MiB
+01/03/2025 01:31:02 [INFO|DP=0|PP=6|TP=3|ip-26-0-170-143]: [After model building] Memory usage: 14081.03MiB. Peak allocated: 14081.15MiB Peak reserved: 14182.00MiB
+01/03/2025 01:31:02 [INFO|DP=0|PP=3|TP=3|ip-26-0-165-24]: [After model building] Memory usage: 14081.03MiB. Peak allocated: 14081.15MiB Peak reserved: 14182.00MiB
+01/03/2025 01:31:02 [INFO|DP=0|PP=7|TP=3|ip-26-0-171-21]: [After model building] Memory usage: 13712.99MiB. Peak allocated: 13713.12MiB Peak reserved: 13814.00MiB
+01/03/2025 01:31:02 [INFO|DP=0|PP=1|TP=2|ip-26-0-164-207]: [After model building] Memory usage: 14081.03MiB. Peak allocated: 14081.15MiB Peak reserved: 14182.00MiB
+01/03/2025 01:31:02 [INFO|DP=0|PP=2|TP=2|ip-26-0-165-131]: [After model building] Memory usage: 14081.03MiB. Peak allocated: 14081.15MiB Peak reserved: 14182.00MiB
+01/03/2025 01:31:02 [INFO|DP=0|PP=2|TP=3|ip-26-0-165-131]: [After model building] Memory usage: 14081.03MiB. Peak allocated: 14081.15MiB Peak reserved: 14182.00MiB
+01/03/2025 01:31:02 [INFO|DP=0|PP=4|TP=3|ip-26-0-166-15]: [After model building] Memory usage: 14081.03MiB. Peak allocated: 14081.15MiB Peak reserved: 14182.00MiB
+01/03/2025 01:31:02 [INFO|DP=0|PP=6|TP=6|ip-26-0-170-143]: Local number of parameters: 7.38G (14081.00MiB)
+01/03/2025 01:31:02 [INFO|DP=0|PP=7|TP=6|ip-26-0-171-21]: Local number of parameters: 7.19G (13712.97MiB)
+01/03/2025 01:31:02 [INFO|DP=0|PP=3|TP=6|ip-26-0-165-24]: Local number of parameters: 7.38G (14081.00MiB)
+01/03/2025 01:31:02 [INFO|DP=0|PP=2|TP=6|ip-26-0-165-131]: Local number of parameters: 7.38G (14081.00MiB)
+01/03/2025 01:31:02 [INFO|DP=0|PP=5|TP=3|ip-26-0-167-175]: [After model building] Memory usage: 13200.96MiB. Peak allocated: 13201.09MiB Peak reserved: 13302.00MiB
+01/03/2025 01:31:02 [INFO|DP=0|PP=7|TP=2|ip-26-0-171-21]: [After model building] Memory usage: 13712.99MiB. Peak allocated: 13713.12MiB Peak reserved: 13814.00MiB
+01/03/2025 01:31:02 [INFO|DP=0|PP=6|TP=2|ip-26-0-170-143]: [After model building] Memory usage: 14081.03MiB. Peak allocated: 14081.15MiB Peak reserved: 14182.00MiB
+01/03/2025 01:31:02 [INFO|DP=0|PP=4|TP=6|ip-26-0-166-15]: Local number of parameters: 7.38G (14081.00MiB)
+01/03/2025 01:31:02 [INFO|DP=0|PP=1|TP=6|ip-26-0-164-207]: Local number of parameters: 7.38G (14081.00MiB)
+01/03/2025 01:31:02 [INFO|DP=0|PP=5|TP=6|ip-26-0-167-175]: Local number of parameters: 6.92G (13200.94MiB)
+01/03/2025 01:31:02 [INFO|DP=0|PP=3|TP=2|ip-26-0-165-24]: [After model building] Memory usage: 14081.03MiB. Peak allocated: 14081.15MiB Peak reserved: 14182.00MiB
+01/03/2025 01:31:02 [INFO|DP=0|PP=5|TP=2|ip-26-0-167-175]: [After model building] Memory usage: 13200.96MiB. Peak allocated: 13201.09MiB Peak reserved: 13302.00MiB
+01/03/2025 01:31:02 [INFO|DP=0|PP=7|TP=6|ip-26-0-171-21]: [After model building] Memory usage: 13712.99MiB. Peak allocated: 13713.12MiB Peak reserved: 13814.00MiB
+01/03/2025 01:31:02 [INFO|DP=0|PP=3|TP=6|ip-26-0-165-24]: [After model building] Memory usage: 14081.03MiB. Peak allocated: 14081.15MiB Peak reserved: 14182.00MiB
+01/03/2025 01:31:02 [INFO|DP=0|PP=6|TP=6|ip-26-0-170-143]: [After model building] Memory usage: 14081.03MiB. Peak allocated: 14081.15MiB Peak reserved: 14182.00MiB
+01/03/2025 01:31:02 [INFO|DP=0|PP=2|TP=6|ip-26-0-165-131]: [After model building] Memory usage: 14081.03MiB. Peak allocated: 14081.15MiB Peak reserved: 14182.00MiB
+01/03/2025 01:31:02 [INFO|DP=0|PP=4|TP=6|ip-26-0-166-15]: [After model building] Memory usage: 14081.03MiB. Peak allocated: 14081.15MiB Peak reserved: 14182.00MiB
+01/03/2025 01:31:02 [INFO|DP=0|PP=1|TP=6|ip-26-0-164-207]: [After model building] Memory usage: 14081.03MiB. Peak allocated: 14081.15MiB Peak reserved: 14182.00MiB
+01/03/2025 01:31:02 [INFO|DP=0|PP=5|TP=6|ip-26-0-167-175]: [After model building] Memory usage: 13200.96MiB. Peak allocated: 13201.09MiB Peak reserved: 13302.00MiB
+01/03/2025 01:31:02 [INFO|DP=0|PP=3|TP=1|ip-26-0-165-24]: Local number of parameters: 7.38G (14081.00MiB)
+01/03/2025 01:31:02 [INFO|DP=0|PP=5|TP=1|ip-26-0-167-175]: Local number of parameters: 6.92G (13200.94MiB)
+01/03/2025 01:31:02 [INFO|DP=0|PP=5|TP=1|ip-26-0-167-175]: [After model building] Memory usage: 13200.96MiB. Peak allocated: 13201.09MiB Peak reserved: 13302.00MiB
+01/03/2025 01:31:02 [INFO|DP=0|PP=3|TP=1|ip-26-0-165-24]: [After model building] Memory usage: 14081.03MiB. Peak allocated: 14081.15MiB Peak reserved: 14182.00MiB
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+01/03/2025 01:31:04 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: [Optimizer Building] Using LearningRateForSP as learning rate
+01/03/2025 01:31:04 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: [ZeRO sharding] Size of optimizer params per rank:
+01/03/2025 01:31:04 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: [ZeRO sharding] DP Rank 0 has 1.91G out of 7.65G (25.00%) params' optimizer states
+01/03/2025 01:31:04 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: [ZeRO sharding] DP Rank 1 has 1.91G out of 7.65G (25.00%) params' optimizer states
+01/03/2025 01:31:04 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: [ZeRO sharding] DP Rank 2 has 1.91G out of 7.65G (25.00%) params' optimizer states
+01/03/2025 01:31:04 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: [ZeRO sharding] DP Rank 3 has 1.91G out of 7.65G (25.00%) params' optimizer states
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.token_position_embeddings | PP: 0/8 | Block rank: 0
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.0 | PP: 0/8 | Block rank: 1
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.1 | PP: 0/8 | Block rank: 2
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.2 | PP: 0/8 | Block rank: 3
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.3 | PP: 0/8 | Block rank: 4
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.4 | PP: 0/8 | Block rank: 5
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.5 | PP: 0/8 | Block rank: 6
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.6 | PP: 0/8 | Block rank: 7
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.7 | PP: 0/8 | Block rank: 8
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.8 | PP: 0/8 | Block rank: 9
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.9 | PP: 0/8 | Block rank: 10
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.10 | PP: 0/8 | Block rank: 11
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.11 | PP: 0/8 | Block rank: 12
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.12 | PP: 0/8 | Block rank: 13
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.13 | PP: 0/8 | Block rank: 14
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.14 | PP: 0/8 | Block rank: 15
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.15 | PP: 0/8 | Block rank: 16
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.16 | PP: 1/8 | Block rank: 0
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.17 | PP: 1/8 | Block rank: 1
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.18 | PP: 1/8 | Block rank: 2
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.19 | PP: 1/8 | Block rank: 3
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.20 | PP: 1/8 | Block rank: 4
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.21 | PP: 1/8 | Block rank: 5
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.22 | PP: 1/8 | Block rank: 6
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.23 | PP: 1/8 | Block rank: 7
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.24 | PP: 1/8 | Block rank: 8
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.25 | PP: 1/8 | Block rank: 9
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.26 | PP: 1/8 | Block rank: 10
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.27 | PP: 1/8 | Block rank: 11
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.28 | PP: 1/8 | Block rank: 12
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.29 | PP: 1/8 | Block rank: 13
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.30 | PP: 1/8 | Block rank: 14
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.31 | PP: 1/8 | Block rank: 15
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.32 | PP: 2/8 | Block rank: 0
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.33 | PP: 2/8 | Block rank: 1
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.34 | PP: 2/8 | Block rank: 2
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.35 | PP: 2/8 | Block rank: 3
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.36 | PP: 2/8 | Block rank: 4
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.37 | PP: 2/8 | Block rank: 5
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.38 | PP: 2/8 | Block rank: 6
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.39 | PP: 2/8 | Block rank: 7
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.40 | PP: 2/8 | Block rank: 8
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.41 | PP: 2/8 | Block rank: 9
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.42 | PP: 2/8 | Block rank: 10
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.43 | PP: 2/8 | Block rank: 11
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.44 | PP: 2/8 | Block rank: 12
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.45 | PP: 2/8 | Block rank: 13
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.46 | PP: 2/8 | Block rank: 14
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.47 | PP: 2/8 | Block rank: 15
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.48 | PP: 3/8 | Block rank: 0
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.49 | PP: 3/8 | Block rank: 1
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.50 | PP: 3/8 | Block rank: 2
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.51 | PP: 3/8 | Block rank: 3
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.52 | PP: 3/8 | Block rank: 4
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.53 | PP: 3/8 | Block rank: 5
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.54 | PP: 3/8 | Block rank: 6
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.55 | PP: 3/8 | Block rank: 7
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.56 | PP: 3/8 | Block rank: 8
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.57 | PP: 3/8 | Block rank: 9
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.58 | PP: 3/8 | Block rank: 10
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.59 | PP: 3/8 | Block rank: 11
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.60 | PP: 3/8 | Block rank: 12
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.61 | PP: 3/8 | Block rank: 13
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.62 | PP: 3/8 | Block rank: 14
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.63 | PP: 3/8 | Block rank: 15
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.64 | PP: 4/8 | Block rank: 0
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.65 | PP: 4/8 | Block rank: 1
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.66 | PP: 4/8 | Block rank: 2
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.67 | PP: 4/8 | Block rank: 3
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.68 | PP: 4/8 | Block rank: 4
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.69 | PP: 4/8 | Block rank: 5
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.70 | PP: 4/8 | Block rank: 6
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.71 | PP: 4/8 | Block rank: 7
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.72 | PP: 4/8 | Block rank: 8
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.73 | PP: 4/8 | Block rank: 9
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.74 | PP: 4/8 | Block rank: 10
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.75 | PP: 4/8 | Block rank: 11
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.76 | PP: 4/8 | Block rank: 12
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.77 | PP: 4/8 | Block rank: 13
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.78 | PP: 4/8 | Block rank: 14
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.79 | PP: 4/8 | Block rank: 15
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.80 | PP: 5/8 | Block rank: 0
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.81 | PP: 5/8 | Block rank: 1
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.82 | PP: 5/8 | Block rank: 2
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.83 | PP: 5/8 | Block rank: 3
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.84 | PP: 5/8 | Block rank: 4
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.85 | PP: 5/8 | Block rank: 5
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.86 | PP: 5/8 | Block rank: 6
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.87 | PP: 5/8 | Block rank: 7
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.88 | PP: 5/8 | Block rank: 8
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.89 | PP: 5/8 | Block rank: 9
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.90 | PP: 5/8 | Block rank: 10
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.91 | PP: 5/8 | Block rank: 11
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.92 | PP: 5/8 | Block rank: 12
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.93 | PP: 5/8 | Block rank: 13
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.94 | PP: 5/8 | Block rank: 14
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.95 | PP: 6/8 | Block rank: 0
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.96 | PP: 6/8 | Block rank: 1
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.97 | PP: 6/8 | Block rank: 2
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.98 | PP: 6/8 | Block rank: 3
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.99 | PP: 6/8 | Block rank: 4
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.100 | PP: 6/8 | Block rank: 5
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.101 | PP: 6/8 | Block rank: 6
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.102 | PP: 6/8 | Block rank: 7
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.103 | PP: 6/8 | Block rank: 8
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.104 | PP: 6/8 | Block rank: 9
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.105 | PP: 6/8 | Block rank: 10
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.106 | PP: 6/8 | Block rank: 11
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.107 | PP: 6/8 | Block rank: 12
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.108 | PP: 6/8 | Block rank: 13
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.109 | PP: 6/8 | Block rank: 14
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.110 | PP: 6/8 | Block rank: 15
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.111 | PP: 7/8 | Block rank: 0
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.112 | PP: 7/8 | Block rank: 1
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.113 | PP: 7/8 | Block rank: 2
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.114 | PP: 7/8 | Block rank: 3
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.115 | PP: 7/8 | Block rank: 4
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.116 | PP: 7/8 | Block rank: 5
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.117 | PP: 7/8 | Block rank: 6
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.118 | PP: 7/8 | Block rank: 7
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.119 | PP: 7/8 | Block rank: 8
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.120 | PP: 7/8 | Block rank: 9
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.121 | PP: 7/8 | Block rank: 10
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.122 | PP: 7/8 | Block rank: 11
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.123 | PP: 7/8 | Block rank: 12
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.124 | PP: 7/8 | Block rank: 13
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.125 | PP: 7/8 | Block rank: 14
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.final_layer_norm | PP: 7/8 | Block rank: 15
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.lm_head | PP: 7/8 | Block rank: 16
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.cast_to_fp32 | PP: 7/8 | Block rank: 17
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: loss | PP: 7/8 | Block rank: 18
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: [Training Plan] Stage Stable Training Stage has 99 remaining training steps and has consumed 0 samples
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: Using dummy data generator
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: [Training Plan] There are 1 training stages 
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: [Stage Stable Training Stage] start from step 1 
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: 
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: [Start training] datetime: 2025-01-03 01:31:05.097587 | mbs: 1 | grad_accum: 64 | global_batch_size: 256 | sequence_length: 4096 | train_steps: 100 | start_iteration_step: 0 | consumed_train_samples: 0
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: Resuming training from stage Stable Training Stage, it has trained for 0 samples and has 99 remaining train steps
+01/03/2025 01:31:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:  Memory usage: 51075.52MiB. Peak allocated 51075.52MiB. Peak reserved: 51210.00MiB
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+NCCL version 2.18.5+cuda12.2
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+NCCL version 2.18.5+cuda12.2
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+NCCL version 2.18.5+cuda12.2
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+NCCL version 2.18.5+cuda12.2
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+NCCL version 2.18.5+cuda12.2
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+NCCL version 2.18.5+cuda12.2
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+NCCL version 2.18.5+cuda12.2
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+NCCL version 2.18.5+cuda12.2
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+NCCL version 2.18.5+cuda12.2
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 160, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 160, in forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+        return self._call_impl(*args, **kwargs)return self._call_impl(*args, **kwargs)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 201, in forward
+    Traceback (most recent call last):
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return self.act(gate_states) * up_states
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 52.00 MiB. GPU 0 has a total capacty of 79.33 GiB of which 33.94 MiB is free. Including non-PyTorch memory, this process has 79.29 GiB memory in use. Of the allocated memory 69.08 GiB is allocated by PyTorch, and 568.76 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 201, in forward
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+        return self.act(gate_states) * up_statesreturn f(*args, **kwargs)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/nn/activations.py", line 149, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return nn.functional.silu(input)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/functional.py", line 2072, in silu
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return torch._C._nn.silu(input)
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 52.00 MiB. GPU 3 has a total capacty of 79.33 GiB of which 13.94 MiB is free. Including non-PyTorch memory, this process has 79.30 GiB memory in use. Of the allocated memory 69.02 GiB is allocated by PyTorch, and 568.76 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self._call_impl(*args, **kwargs)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 160, in forward
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 160, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+Traceback (most recent call last):
+    trainer.train(dataloader)  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+        return self._call_impl(*args, **kwargs)hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return forward_call(*args, **kwargs)    
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 201, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+    return self.act(gate_states) * up_states
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 52.00 MiB. GPU 6 has a total capacty of 79.33 GiB of which 1.94 MiB is free. Including non-PyTorch memory, this process has 79.32 GiB memory in use. Of the allocated memory 69.08 GiB is allocated by PyTorch, and 568.76 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 201, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return self.act(gate_states) * up_states
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/nn/activations.py", line 149, in forward
+return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 160, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 160, in forward
+        return nn.functional.silu(input)output = self.pp_block(**new_kwargs)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/functional.py", line 2072, in silu
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+        return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]output = self.pp_block(**new_kwargs)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    return torch._C._nn.silu(input)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 52.00 MiB. GPU 4 has a total capacty of 79.33 GiB of which 37.94 MiB is free. Including non-PyTorch memory, this process has 79.28 GiB memory in use. Of the allocated memory 69.02 GiB is allocated by PyTorch, and 568.76 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 160, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 201, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    return self.act(gate_states) * up_states
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 52.00 MiB. GPU 2 has a total capacty of 79.33 GiB of which 5.94 MiB is free. Including non-PyTorch memory, this process has 79.31 GiB memory in use. Of the allocated memory 69.08 GiB is allocated by PyTorch, and 564.76 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    out = F.linear(tensor, weight, bias)
+    return self._call_impl(*args, **kwargs)
+torch.cuda.OutOfMemoryError  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+: CUDA out of memory. Tried to allocate 128.00 MiB. GPU 5 has a total capacty of 79.33 GiB of which 29.94 MiB is free. Including non-PyTorch memory, this process has 79.29 GiB memory in use. Of the allocated memory 69.13 GiB is allocated by PyTorch, and 568.76 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    out = F.linear(tensor, weight, bias)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 128.00 MiB. GPU 1 has a total capacty of 79.33 GiB of which 29.94 MiB is free. Including non-PyTorch memory, this process has 79.29 GiB memory in use. Of the allocated memory 69.13 GiB is allocated by PyTorch, and 568.76 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    trainer.train(dataloader)    
+trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    return self._call_impl(*args, **kwargs)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+return self._call_impl(*args, **kwargs)    
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = model(**micro_batch)    
+return forward_call(*args, **kwargs)    
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+return forward_call(*args, **kwargs)    
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+return forward_call(*args, **kwargs)  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+        hidden_encoder_states = encoder_block(**hidden_encoder_states)return forward_call(*args, **kwargs)
+
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+        return self._call_impl(*args, **kwargs)return self._call_impl(*args, **kwargs)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 160, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 160, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 160, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 160, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.pp_block(**new_kwargs)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    output = self.pp_block(**new_kwargs)
+          File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+return self._call_impl(*args, **kwargs)return self._call_impl(*args, **kwargs)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+Traceback (most recent call last):
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 160, in forward
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+        output = self.pp_block(**new_kwargs)return forward_call(*args, **kwargs)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+        hidden_states = self.down_proj(self.split_silu_mul(merged_states))return self._call_impl(*args, **kwargs)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 201, in forward
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 201, in forward
+    return self.act(gate_states) * up_states
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 52.00 MiB. GPU 2 has a total capacty of 79.33 GiB of which 13.94 MiB is free. Including non-PyTorch memory, this process has 79.30 GiB memory in use. Of the allocated memory 69.08 GiB is allocated by PyTorch, and 556.76 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self.act(gate_states) * up_states
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 52.00 MiB. GPU 6 has a total capacty of 79.33 GiB of which 45.94 MiB is free. Including non-PyTorch memory, this process has 79.27 GiB memory in use. Of the allocated memory 69.08 GiB is allocated by PyTorch, and 524.76 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+return self._call_impl(*args, **kwargs)    
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    return forward_call(*args, **kwargs)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 201, in forward
+return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 201, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+    return self.act(gate_states) * up_states
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 52.00 MiB. GPU 0 has a total capacty of 79.33 GiB of which 45.94 MiB is free. Including non-PyTorch memory, this process has 79.27 GiB memory in use. Of the allocated memory 69.08 GiB is allocated by PyTorch, and 556.76 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return self.act(gate_states) * up_states
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 52.00 MiB. GPU 4 has a total capacty of 79.33 GiB of which 29.94 MiB is free. Including non-PyTorch memory, this process has 79.29 GiB memory in use. Of the allocated memory 69.08 GiB is allocated by PyTorch, and 524.76 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 201, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self.act(gate_states) * up_states
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/nn/activations.py", line 149, in forward
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+        return self._call_impl(*args, **kwargs)return nn.functional.silu(input)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/functional.py", line 2072, in silu
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 160, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 160, in forward
+    return torch._C._nn.silu(input)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 52.00 MiB. GPU 3 has a total capacty of 79.33 GiB of which 25.94 MiB is free. Including non-PyTorch memory, this process has 79.29 GiB memory in use. Of the allocated memory 69.02 GiB is allocated by PyTorch, and 556.76 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    out = F.linear(tensor, weight, bias)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 128.00 MiB. GPU 1 has a total capacty of 79.33 GiB of which 41.94 MiB is free. Including non-PyTorch memory, this process has 79.28 GiB memory in use. Of the allocated memory 69.13 GiB is allocated by PyTorch, and 556.76 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    out = F.linear(tensor, weight, bias)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 128.00 MiB. GPU 5 has a total capacty of 79.33 GiB of which 41.94 MiB is free. Including non-PyTorch memory, this process has 79.28 GiB memory in use. Of the allocated memory 69.13 GiB is allocated by PyTorch, and 556.76 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 160, in forward
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 725, in _core_forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    output = self.attn(hidden_states=hidden_states, sequence_mask=sequence_mask)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 690, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    output = self.o_proj(attention_output)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    out = F.linear(tensor, weight, bias)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 128.00 MiB. GPU 3 has a total capacty of 79.33 GiB of which 33.94 MiB is free. Including non-PyTorch memory, this process has 79.29 GiB memory in use. Of the allocated memory 68.92 GiB is allocated by PyTorch, and 572.76 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 160, in forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 725, in _core_forward
+    output = self.attn(hidden_states=hidden_states, sequence_mask=sequence_mask)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 690, in forward
+    output = self.o_proj(attention_output)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+    out = F.linear(tensor, weight, bias)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 128.00 MiB. GPU 3 has a total capacty of 79.33 GiB of which 109.94 MiB is free. Including non-PyTorch memory, this process has 79.21 GiB memory in use. Of the allocated memory 69.64 GiB is allocated by PyTorch, and 480.25 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+NCCL version 2.18.5+cuda12.2
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+NCCL version 2.18.5+cuda12.2
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+NCCL version 2.18.5+cuda12.2
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+NCCL version 2.18.5+cuda12.2
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+NCCL version 2.18.5+cuda12.2
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+NCCL version 2.18.5+cuda12.2
+[E ProcessGroupGloo.cpp:2810] [Rank 0]: Rank 3 failed to pass monitoredBarrier in 1200000 ms
+[E ProcessGroupGloo.cpp:138] [Rank 0]: Ranks 3 failed to pass monitoredBarrier in 1200000 ms
+[E ProcessGroupGloo.cpp:2810] [Rank 0]: Rank 3 failed to pass monitoredBarrier in 1200000 ms
+[E ProcessGroupGloo.cpp:138] [Rank 0]: Ranks 3 failed to pass monitoredBarrier in 1200000 ms
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/c10d_logger.py", line 47, in wrapper
+    return func(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/distributed_c10d.py", line 3375, in reduce_scatter_tensor
+    work = group._reduce_scatter_base(output, input, opts)
+RuntimeError: [Rank 0]: Ranks 3 failed to pass monitoredBarrier in 1200000 ms
+
+During handling of the above exception, another exception occurred:
+
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 160, in forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 725, in _core_forward
+    output = self.attn(hidden_states=hidden_states, sequence_mask=sequence_mask)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 690, in forward
+    output = self.o_proj(attention_output)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 469, in forward
+    out = differentiable_reduce_scatter_sum(out, group=group)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/distributed_differentiable_primitives.py", line 146, in differentiable_reduce_scatter_sum
+    return DifferentiableReduceScatterSum.apply(tensor, group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/distributed_differentiable_primitives.py", line 119, in forward
+    dist.reduce_scatter_tensor(sharded_tensor, tensor, group=group, op=dist.ReduceOp.SUM)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/distributed.py", line 43, in reduce_scatter_tensor
+    return dist.reduce_scatter_tensor(output=output, input=input, group=group, op=op, async_op=async_op)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/c10d_logger.py", line 52, in wrapper
+    "args": f"{args}, {kwargs}",
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/_tensor.py", line 431, in __repr__
+    return torch._tensor_str._str(self, tensor_contents=tensor_contents)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/_tensor_str.py", line 664, in _str
+    return _str_intern(self, tensor_contents=tensor_contents)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/_tensor_str.py", line 595, in _str_intern
+    tensor_str = _tensor_str(self, indent)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/_tensor_str.py", line 329, in _tensor_str
+    self = self.float()
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 256.00 MiB. GPU 0 has a total capacty of 79.33 GiB of which 133.94 MiB is free. Including non-PyTorch memory, this process has 79.19 GiB memory in use. Of the allocated memory 69.78 GiB is allocated by PyTorch, and 464.25 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/c10d_logger.py", line 47, in wrapper
+    return func(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/distributed_c10d.py", line 3375, in reduce_scatter_tensor
+    work = group._reduce_scatter_base(output, input, opts)
+RuntimeError: [Rank 0]: Ranks 3 failed to pass monitoredBarrier in 1200000 ms
+
+During handling of the above exception, another exception occurred:
+
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 160, in forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 725, in _core_forward
+    output = self.attn(hidden_states=hidden_states, sequence_mask=sequence_mask)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 690, in forward
+    output = self.o_proj(attention_output)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 469, in forward
+    out = differentiable_reduce_scatter_sum(out, group=group)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/distributed_differentiable_primitives.py", line 146, in differentiable_reduce_scatter_sum
+    return DifferentiableReduceScatterSum.apply(tensor, group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/distributed_differentiable_primitives.py", line 119, in forward
+    dist.reduce_scatter_tensor(sharded_tensor, tensor, group=group, op=dist.ReduceOp.SUM)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/distributed.py", line 43, in reduce_scatter_tensor
+    return dist.reduce_scatter_tensor(output=output, input=input, group=group, op=op, async_op=async_op)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/c10d_logger.py", line 52, in wrapper
+    "args": f"{args}, {kwargs}",
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/_tensor.py", line 431, in __repr__
+    return torch._tensor_str._str(self, tensor_contents=tensor_contents)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/_tensor_str.py", line 664, in _str
+    return _str_intern(self, tensor_contents=tensor_contents)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/_tensor_str.py", line 595, in _str_intern
+    tensor_str = _tensor_str(self, indent)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/_tensor_str.py", line 329, in _tensor_str
+    self = self.float()
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 256.00 MiB. GPU 0 has a total capacty of 79.33 GiB of which 137.94 MiB is free. Including non-PyTorch memory, this process has 79.18 GiB memory in use. Of the allocated memory 69.06 GiB is allocated by PyTorch, and 556.76 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+[E ProcessGroupGloo.cpp:138] Rank 7 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/c10d_logger.py", line 47, in wrapper
+    return func(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/distributed_c10d.py", line 3375, in reduce_scatter_tensor
+    work = group._reduce_scatter_base(output, input, opts)
+RuntimeError: Rank 7 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+ Original exception: 
+[../third_party/gloo/gloo/transport/tcp/pair.cc:534] Connection closed by peer [26.0.164.0]:4319
+
+During handling of the above exception, another exception occurred:
+
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 160, in forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 469, in forward
+    out = differentiable_reduce_scatter_sum(out, group=group)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/distributed_differentiable_primitives.py", line 146, in differentiable_reduce_scatter_sum
+    return DifferentiableReduceScatterSum.apply(tensor, group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/distributed_differentiable_primitives.py", line 119, in forward
+    dist.reduce_scatter_tensor(sharded_tensor, tensor, group=group, op=dist.ReduceOp.SUM)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/distributed.py", line 43, in reduce_scatter_tensor
+    return dist.reduce_scatter_tensor(output=output, input=input, group=group, op=op, async_op=async_op)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/c10d_logger.py", line 52, in wrapper
+    "args": f"{args}, {kwargs}",
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/_tensor.py", line 431, in __repr__
+    return torch._tensor_str._str(self, tensor_contents=tensor_contents)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/_tensor_str.py", line 664, in _str
+    return _str_intern(self, tensor_contents=tensor_contents)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/_tensor_str.py", line 595, in _str_intern
+    tensor_str = _tensor_str(self, indent)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/_tensor_str.py", line 329, in _tensor_str
+    self = self.float()
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 256.00 MiB. GPU 7 has a total capacty of 79.33 GiB of which 101.94 MiB is free. Including non-PyTorch memory, this process has 79.22 GiB memory in use. Of the allocated memory 69.27 GiB is allocated by PyTorch, and 552.76 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+[E ProcessGroupGloo.cpp:138] Rank 7 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+[E ProcessGroupGloo.cpp:138] Rank 1 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+[E ProcessGroupGloo.cpp:138] Rank 5 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+[E ProcessGroupGloo.cpp:138] Rank 2 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+[E ProcessGroupGloo.cpp:138] Rank 4 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+[E ProcessGroupGloo.cpp:138] Rank 6 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+[E ProcessGroupGloo.cpp:138] Rank 7 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+[E TraceUtils.h:35] Store is down while updating #60 with key NCCL_0_trace_start
+[E TraceUtils.h:35] Store is down while updating #60 with key NCCL_0_trace_start
+[E TraceUtils.h:35] Store is down while updating #90 with key NCCL_0_trace_start
+[E TraceUtils.h:35] Store is down while updating #60 with key NCCL_0_trace_end
+[E TraceUtils.h:35] Store is down while updating #60 with key NCCL_0_trace_start
+[E TraceUtils.h:35] Store is down while updating #63 with key NCCL_0_trace_start
+[E TraceUtils.h:35] Store is down while updating #60 with key NCCL_0_trace_start
+[E TraceUtils.h:35] Store is down while updating #63 with key NCCL_0_trace_end
+[E TraceUtils.h:35] Store is down while updating #69 with key NCCL_1_trace_start
+[E TraceUtils.h:35] Store is down while updating #77 with key NCCL_3_trace_start
+[E TraceUtils.h:35] Store is down while updating #74 with key NCCL_3_trace_start
+[E TraceUtils.h:35] Store is down while updating #72 with key NCCL_0_trace_start
+[E TraceUtils.h:35] Store is down while updating #93 with key NCCL_3_trace_start
+[E TraceUtils.h:35] Store is down while updating #72 with key NCCL_3_trace_start
+[E TraceUtils.h:35] Store is down while updating #60 with key NCCL_2_trace_start
+[E TraceUtils.h:35] Store is down while updating #66 with key NCCL_0_trace_start
+[E TraceUtils.h:35] Store is down while updating #60 with key NCCL_3_trace_start
+[E TraceUtils.h:35] Store is down while updating #66 with key NCCL_0_trace_start
+[E TraceUtils.h:35] Store is down while updating #63 with key NCCL_3_trace_start
+[E TraceUtils.h:35] Store is down while updating #63 with key NCCL_3_trace_start
+[E TraceUtils.h:35] Store is down while updating #62 with key NCCL_3_trace_start
+[E TraceUtils.h:35] Store is down while updating #69 with key NCCL_0_trace_start
+[E TraceUtils.h:35] Store is down while updating #68 with key NCCL_0_trace_start
+[E TraceUtils.h:35] Store is down while updating #74 with key NCCL_3_trace_start
+[E TraceUtils.h:35] Store is down while updating #68 with key NCCL_1_trace_start
+[E TraceUtils.h:35] Store is down while updating #74 with key NCCL_1_trace_start
+[E TraceUtils.h:35] Store is down while updating #66 with key NCCL_0_trace_start
+[E TraceUtils.h:35] Store is down while updating #60 with key NCCL_3_trace_start
+[E TraceUtils.h:35] Store is down while updating #78 with key NCCL_3_trace_end
+[E TraceUtils.h:35] Store is down while updating #66 with key NCCL_1_trace_start
+[E TraceUtils.h:35] Store is down while updating #66 with key NCCL_3_trace_start
+[E TraceUtils.h:35] Store is down while updating #60 with key NCCL_3_trace_start
+[E TraceUtils.h:35] Store is down while updating #69 with key NCCL_2_trace_start
+[E TraceUtils.h:35] Store is down while updating #72 with key NCCL_2_trace_start
+[E TraceUtils.h:35] Store is down while updating #68 with key NCCL_2_trace_start
+[E TraceUtils.h:35] Store is down while updating #66 with key NCCL_1_trace_start
+[E TraceUtils.h:35] Store is down while updating #71 with key NCCL_3_trace_end
+[E TraceUtils.h:35] Store is down while updating #66 with key NCCL_3_trace_start
+[E TraceUtils.h:35] Store is down while updating #66 with key NCCL_3_trace_start
+[E TraceUtils.h:35] Store is down while updating #68 with key NCCL_0_trace_start
+[E TraceUtils.h:35] Store is down while updating #69 with key NCCL_1_trace_start
+[E TraceUtils.h:35] Store is down while updating #72 with key NCCL_2_trace_start
+[E TraceUtils.h:35] Store is down while updating #66 with key NCCL_3_trace_start
+[E TraceUtils.h:35] Store is down while updating #69 with key NCCL_3_trace_start
+[E TraceUtils.h:35] Store is down while updating #66 with key NCCL_3_trace_start
+[E TraceUtils.h:35] Store is down while updating #68 with key NCCL_2_trace_end
+[E TraceUtils.h:35] Store is down while updating #75 with key NCCL_2_trace_start
+[E TraceUtils.h:35] Store is down while updating #66 with key NCCL_2_trace_start
+[E TraceUtils.h:35] Store is down while updating #69 with key NCCL_1_trace_start
+[E TraceUtils.h:35] Store is down while updating #65 with key NCCL_1_trace_start
+[E TraceUtils.h:35] Store is down while updating #60 with key NCCL_1_trace_start
+[E TraceUtils.h:35] Store is down while updating #72 with key NCCL_3_trace_end
+[E TraceUtils.h:35] Store is down while updating #60 with key NCCL_1_trace_start
+[E TraceUtils.h:35] Store is down while updating #59 with key NCCL_1_trace_start
+[E TraceUtils.h:35] Store is down while updating #66 with key NCCL_1_trace_start
+[E TraceUtils.h:35] Store is down while updating #60 with key NCCL_1_trace_start
+[E TraceUtils.h:35] Store is down while updating #63 with key NCCL_1_trace_start
+[E TraceUtils.h:35] Store is down while updating #66 with key NCCL_1_trace_start
+[E TraceUtils.h:35] Store is down while updating #72 with key NCCL_1_trace_start
+[E TraceUtils.h:35] Store is down while updating #71 with key NCCL_2_trace_start
+[E TraceUtils.h:35] Store is down while updating #75 with key NCCL_3_trace_start
+[E TraceUtils.h:35] Store is down while updating #90 with key NCCL_1_trace_start
+[E TraceUtils.h:35] Store is down while updating #71 with key NCCL_1_trace_start
+[E TraceUtils.h:35] Store is down while updating #71 with key NCCL_2_trace_end
+[E TraceUtils.h:35] Store is down while updating #66 with key NCCL_1_trace_start
+[E TraceUtils.h:35] Store is down while updating #69 with key NCCL_1_trace_start
+[E TraceUtils.h:35] Store is down while updating #66 with key NCCL_2_trace_start
+[E TraceUtils.h:35] Store is down while updating #63 with key NCCL_1_trace_start
+[E TraceUtils.h:35] Store is down while updating #72 with key NCCL_2_trace_start
+[E TraceUtils.h:35] Store is down while updating #69 with key NCCL_1_trace_start
+[E TraceUtils.h:35] Store is down while updating #66 with key NCCL_2_trace_start
+[E TraceUtils.h:35] Store is down while updating #69 with key NCCL_2_trace_start
+[E TraceUtils.h:35] Store is down while updating #69 with key NCCL_2_trace_start
+[E TraceUtils.h:35] Store is down while updating #60 with key NCCL_2_trace_start
+[E TraceUtils.h:35] Store is down while updating #66 with key NCCL_0_trace_start
+[E TraceUtils.h:35] Store is down while updating #66 with key NCCL_2_trace_start
+[E TraceUtils.h:35] Store is down while updating #69 with key NCCL_2_trace_start
+[E TraceUtils.h:35] Store is down while updating #69 with key NCCL_0_trace_start
+[E TraceUtils.h:35] Store is down while updating #90 with key NCCL_0_trace_start
+[E TraceUtils.h:35] Store is down while updating #63 with key NCCL_2_trace_start
+[E TraceUtils.h:35] Store is down while updating #60 with key NCCL_2_trace_start
+[E TraceUtils.h:35] Store is down while updating #90 with key NCCL_2_trace_start
+[E TraceUtils.h:35] Store is down while updating #66 with key NCCL_0_trace_start
+[E TraceUtils.h:35] Store is down while updating #71 with key NCCL_0_trace_start
+[E TraceUtils.h:35] Store is down while updating #66 with key NCCL_0_trace_start
+[E TraceUtils.h:35] Store is down while updating #60 with key NCCL_2_trace_start
+[E TraceUtils.h:35] Store is down while updating #69 with key NCCL_0_trace_start
+01/03/2025 01:31:16 [ERROR|DP=0|PP=6|TP=4|ip-26-0-170-143]: [RECV META] Rank 6: Failed to receive metadata from rank 5: [6] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '5:6', but store->get('5:6') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [WARNING|DP=0|PP=6|TP=4|ip-26-0-170-143]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 5: [6] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '5:6', but store->get('5:6') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [ERROR|DP=0|PP=6|TP=5|ip-26-0-170-143]: [RECV META] Rank 6: Failed to receive metadata from rank 5: [6] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '5:6', but store->get('5:6') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [WARNING|DP=0|PP=6|TP=4|ip-26-0-170-143]: No progress made in communication iteration, attempt 1/1
+01/03/2025 01:31:16 [ERROR|DP=0|PP=6|TP=4|ip-26-0-170-143]: Max retries reached, giving up on communication
+01/03/2025 01:31:16 [WARNING|DP=0|PP=6|TP=5|ip-26-0-170-143]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 5: [6] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '5:6', but store->get('5:6') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [ERROR|DP=0|PP=6|TP=3|ip-26-0-170-143]: [RECV META] Rank 6: Failed to receive metadata from rank 5: [6] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '5:6', but store->get('5:6') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [ERROR|DP=0|PP=6|TP=1|ip-26-0-170-143]: [RECV META] Rank 6: Failed to receive metadata from rank 5: [6] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '5:6', but store->get('5:6') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [WARNING|DP=0|PP=6|TP=5|ip-26-0-170-143]: No progress made in communication iteration, attempt 1/1
+01/03/2025 01:31:16 [ERROR|DP=0|PP=6|TP=6|ip-26-0-170-143]: [RECV META] Rank 6: Failed to receive metadata from rank 5: [6] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '5:6', but store->get('5:6') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [ERROR|DP=0|PP=6|TP=5|ip-26-0-170-143]: Max retries reached, giving up on communication
+01/03/2025 01:31:16 [ERROR|DP=0|PP=6|TP=0|ip-26-0-170-143]: [RECV META] Rank 6: Failed to receive metadata from rank 5: [6] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '5:6', but store->get('5:6') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [WARNING|DP=0|PP=6|TP=3|ip-26-0-170-143]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 5: [6] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '5:6', but store->get('5:6') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [WARNING|DP=0|PP=6|TP=1|ip-26-0-170-143]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 5: [6] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '5:6', but store->get('5:6') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [WARNING|DP=0|PP=6|TP=6|ip-26-0-170-143]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 5: [6] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '5:6', but store->get('5:6') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [WARNING|DP=0|PP=6|TP=3|ip-26-0-170-143]: No progress made in communication iteration, attempt 1/1
+01/03/2025 01:31:16 [WARNING|DP=0|PP=6|TP=1|ip-26-0-170-143]: No progress made in communication iteration, attempt 1/1
+01/03/2025 01:31:16 [ERROR|DP=0|PP=6|TP=2|ip-26-0-170-143]: [RECV META] Rank 6: Failed to receive metadata from rank 5: [6] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '5:6', but store->get('5:6') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [WARNING|DP=0|PP=6|TP=0|ip-26-0-170-143]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 5: [6] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '5:6', but store->get('5:6') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [ERROR|DP=0|PP=6|TP=3|ip-26-0-170-143]: Max retries reached, giving up on communication
+01/03/2025 01:31:16 [WARNING|DP=0|PP=6|TP=6|ip-26-0-170-143]: No progress made in communication iteration, attempt 1/1
+01/03/2025 01:31:16 [ERROR|DP=0|PP=6|TP=1|ip-26-0-170-143]: Max retries reached, giving up on communication
+01/03/2025 01:31:16 [WARNING|DP=0|PP=6|TP=0|ip-26-0-170-143]: No progress made in communication iteration, attempt 1/1
+01/03/2025 01:31:16 [ERROR|DP=0|PP=6|TP=6|ip-26-0-170-143]: Max retries reached, giving up on communication
+01/03/2025 01:31:16 [WARNING|DP=0|PP=6|TP=2|ip-26-0-170-143]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 5: [6] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '5:6', but store->get('5:6') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [ERROR|DP=0|PP=6|TP=0|ip-26-0-170-143]: Max retries reached, giving up on communication
+01/03/2025 01:31:16 [ERROR|DP=1|PP=7|TP=4|ip-26-0-171-56]: [RECV META] Rank 7: Failed to receive metadata from rank 6: [7] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '6:7', but store->get('6:7') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [WARNING|DP=0|PP=6|TP=2|ip-26-0-170-143]: No progress made in communication iteration, attempt 1/1
+01/03/2025 01:31:16 [WARNING|DP=1|PP=7|TP=4|ip-26-0-171-56]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 6: [7] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '6:7', but store->get('6:7') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [WARNING|DP=1|PP=7|TP=4|ip-26-0-171-56]: No progress made in communication iteration, attempt 1/1
+01/03/2025 01:31:16 [ERROR|DP=0|PP=6|TP=2|ip-26-0-170-143]: Max retries reached, giving up on communication
+01/03/2025 01:31:16 [ERROR|DP=0|PP=6|TP=7|ip-26-0-170-143]: [RECV META] Rank 6: Failed to receive metadata from rank 5: [6] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '5:6', but store->get('5:6') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [ERROR|DP=1|PP=7|TP=4|ip-26-0-171-56]: Max retries reached, giving up on communication
+01/03/2025 01:31:16 [ERROR|DP=0|PP=7|TP=6|ip-26-0-171-21]: [RECV META] Rank 7: Failed to receive metadata from rank 6: [7] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '6:7', but store->get('6:7') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [WARNING|DP=0|PP=6|TP=7|ip-26-0-170-143]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 5: [6] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '5:6', but store->get('5:6') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [WARNING|DP=0|PP=6|TP=7|ip-26-0-170-143]: No progress made in communication iteration, attempt 1/1
+01/03/2025 01:31:16 [ERROR|DP=0|PP=6|TP=7|ip-26-0-170-143]: Max retries reached, giving up on communication
+01/03/2025 01:31:16 [WARNING|DP=0|PP=7|TP=6|ip-26-0-171-21]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 6: [7] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '6:7', but store->get('6:7') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [ERROR|DP=2|PP=6|TP=6|ip-26-0-171-102]: [RECV META] Rank 6: Failed to receive metadata from rank 5: [6] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '5:6', but store->get('5:6') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [ERROR|DP=0|PP=7|TP=7|ip-26-0-171-21]: [RECV META] Rank 7: Failed to receive metadata from rank 6: [7] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '6:7', but store->get('6:7') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [ERROR|DP=0|PP=7|TP=3|ip-26-0-171-21]: [RECV META] Rank 7: Failed to receive metadata from rank 6: [7] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '6:7', but store->get('6:7') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [ERROR|DP=3|PP=6|TP=6|ip-26-0-171-168]: [RECV META] Rank 6: Failed to receive metadata from rank 5: [6] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '5:6', but store->get('5:6') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [ERROR|DP=1|PP=7|TP=1|ip-26-0-171-56]: [RECV META] Rank 7: Failed to receive metadata from rank 6: [7] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '6:7', but store->get('6:7') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [WARNING|DP=0|PP=7|TP=6|ip-26-0-171-21]: No progress made in communication iteration, attempt 1/1
+01/03/2025 01:31:16 [ERROR|DP=0|PP=7|TP=6|ip-26-0-171-21]: Max retries reached, giving up on communication
+01/03/2025 01:31:16 [WARNING|DP=0|PP=7|TP=7|ip-26-0-171-21]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 6: [7] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '6:7', but store->get('6:7') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [WARNING|DP=0|PP=7|TP=3|ip-26-0-171-21]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 6: [7] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '6:7', but store->get('6:7') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [WARNING|DP=2|PP=6|TP=6|ip-26-0-171-102]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 5: [6] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '5:6', but store->get('5:6') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [ERROR|DP=3|PP=6|TP=2|ip-26-0-171-168]: [RECV META] Rank 6: Failed to receive metadata from rank 5: [6] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '5:6', but store->get('5:6') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [ERROR|DP=3|PP=6|TP=1|ip-26-0-171-168]: [RECV META] Rank 6: Failed to receive metadata from rank 5: [6] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '5:6', but store->get('5:6') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [WARNING|DP=2|PP=6|TP=6|ip-26-0-171-102]: No progress made in communication iteration, attempt 1/1
+01/03/2025 01:31:16 [WARNING|DP=1|PP=7|TP=1|ip-26-0-171-56]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 6: [7] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '6:7', but store->get('6:7') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [WARNING|DP=3|PP=6|TP=6|ip-26-0-171-168]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 5: [6] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '5:6', but store->get('5:6') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [WARNING|DP=3|PP=6|TP=2|ip-26-0-171-168]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 5: [6] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '5:6', but store->get('5:6') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [WARNING|DP=1|PP=7|TP=1|ip-26-0-171-56]: No progress made in communication iteration, attempt 1/1
+01/03/2025 01:31:16 [WARNING|DP=0|PP=7|TP=7|ip-26-0-171-21]: No progress made in communication iteration, attempt 1/1
+01/03/2025 01:31:16 [WARNING|DP=0|PP=7|TP=3|ip-26-0-171-21]: No progress made in communication iteration, attempt 1/1
+01/03/2025 01:31:16 [ERROR|DP=3|PP=6|TP=5|ip-26-0-171-168]: [RECV META] Rank 6: Failed to receive metadata from rank 5: [6] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '5:6', but store->get('5:6') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [WARNING|DP=3|PP=6|TP=1|ip-26-0-171-168]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 5: [6] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '5:6', but store->get('5:6') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [ERROR|DP=0|PP=7|TP=7|ip-26-0-171-21]: Max retries reached, giving up on communication
+01/03/2025 01:31:16 [ERROR|DP=0|PP=7|TP=3|ip-26-0-171-21]: Max retries reached, giving up on communication
+01/03/2025 01:31:16 [WARNING|DP=3|PP=6|TP=6|ip-26-0-171-168]: No progress made in communication iteration, attempt 1/1
+01/03/2025 01:31:16 [WARNING|DP=3|PP=6|TP=2|ip-26-0-171-168]: No progress made in communication iteration, attempt 1/1
+01/03/2025 01:31:16 [ERROR|DP=0|PP=7|TP=4|ip-26-0-171-21]: [RECV META] Rank 7: Failed to receive metadata from rank 6: [7] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '6:7', but store->get('6:7') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [ERROR|DP=2|PP=7|TP=7|ip-26-0-171-62]: [RECV META] Rank 7: Failed to receive metadata from rank 6: [7] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '6:7', but store->get('6:7') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [ERROR|DP=1|PP=7|TP=1|ip-26-0-171-56]: Max retries reached, giving up on communication
+01/03/2025 01:31:16 [ERROR|DP=2|PP=6|TP=6|ip-26-0-171-102]: Max retries reached, giving up on communication
+01/03/2025 01:31:16 [ERROR|DP=2|PP=7|TP=2|ip-26-0-171-62]: [RECV META] Rank 7: Failed to receive metadata from rank 6: [7] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '6:7', but store->get('6:7') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [WARNING|DP=2|PP=7|TP=7|ip-26-0-171-62]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 6: [7] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '6:7', but store->get('6:7') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [WARNING|DP=3|PP=6|TP=1|ip-26-0-171-168]: No progress made in communication iteration, attempt 1/1
+01/03/2025 01:31:16 [ERROR|DP=3|PP=6|TP=2|ip-26-0-171-168]: Max retries reached, giving up on communication
+01/03/2025 01:31:16 [ERROR|DP=3|PP=6|TP=6|ip-26-0-171-168]: Max retries reached, giving up on communication
+01/03/2025 01:31:16 [WARNING|DP=3|PP=6|TP=5|ip-26-0-171-168]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 5: [6] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '5:6', but store->get('5:6') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [ERROR|DP=0|PP=7|TP=1|ip-26-0-171-21]: [RECV META] Rank 7: Failed to receive metadata from rank 6: [7] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '6:7', but store->get('6:7') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [ERROR|DP=3|PP=6|TP=4|ip-26-0-171-168]: [RECV META] Rank 6: Failed to receive metadata from rank 5: [6] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '5:6', but store->get('5:6') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [ERROR|DP=3|PP=6|TP=1|ip-26-0-171-168]: Max retries reached, giving up on communication
+01/03/2025 01:31:16 [WARNING|DP=0|PP=7|TP=4|ip-26-0-171-21]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 6: [7] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '6:7', but store->get('6:7') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [WARNING|DP=3|PP=6|TP=5|ip-26-0-171-168]: No progress made in communication iteration, attempt 1/1
+01/03/2025 01:31:16 [ERROR|DP=3|PP=6|TP=5|ip-26-0-171-168]: Max retries reached, giving up on communication
+01/03/2025 01:31:16 [ERROR|DP=1|PP=7|TP=2|ip-26-0-171-56]: [RECV META] Rank 7: Failed to receive metadata from rank 6: [7] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '6:7', but store->get('6:7') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [ERROR|DP=1|PP=7|TP=7|ip-26-0-171-56]: [RECV META] Rank 7: Failed to receive metadata from rank 6: [7] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '6:7', but store->get('6:7') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [WARNING|DP=0|PP=7|TP=1|ip-26-0-171-21]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 6: [7] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '6:7', but store->get('6:7') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [WARNING|DP=0|PP=7|TP=4|ip-26-0-171-21]: No progress made in communication iteration, attempt 1/1
+01/03/2025 01:31:16 [ERROR|DP=3|PP=5|TP=6|ip-26-0-167-9]: [RECV META] Rank 5: Failed to receive metadata from rank 4: [5] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '4:5', but store->get('4:5') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [ERROR|DP=3|PP=5|TP=3|ip-26-0-167-9]: [RECV META] Rank 5: Failed to receive metadata from rank 4: [5] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '4:5', but store->get('4:5') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [WARNING|DP=3|PP=6|TP=4|ip-26-0-171-168]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 5: [6] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '5:6', but store->get('5:6') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [ERROR|DP=3|PP=6|TP=7|ip-26-0-171-168]: [RECV META] Rank 6: Failed to receive metadata from rank 5: [6] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '5:6', but store->get('5:6') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [WARNING|DP=0|PP=7|TP=1|ip-26-0-171-21]: No progress made in communication iteration, attempt 1/1
+01/03/2025 01:31:16 [ERROR|DP=0|PP=7|TP=4|ip-26-0-171-21]: Max retries reached, giving up on communication
+01/03/2025 01:31:16 [ERROR|DP=2|PP=7|TP=5|ip-26-0-171-62]: [RECV META] Rank 7: Failed to receive metadata from rank 6: [7] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '6:7', but store->get('6:7') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [WARNING|DP=2|PP=7|TP=7|ip-26-0-171-62]: No progress made in communication iteration, attempt 1/1
+01/03/2025 01:31:16 [WARNING|DP=3|PP=6|TP=4|ip-26-0-171-168]: No progress made in communication iteration, attempt 1/1
+01/03/2025 01:31:16 [ERROR|DP=0|PP=7|TP=1|ip-26-0-171-21]: Max retries reached, giving up on communication
+01/03/2025 01:31:16 [ERROR|DP=2|PP=7|TP=3|ip-26-0-171-62]: [RECV META] Rank 7: Failed to receive metadata from rank 6: [7] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '6:7', but store->get('6:7') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [WARNING|DP=3|PP=5|TP=6|ip-26-0-167-9]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 4: [5] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '4:5', but store->get('4:5') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [WARNING|DP=3|PP=5|TP=3|ip-26-0-167-9]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 4: [5] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '4:5', but store->get('4:5') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [ERROR|DP=2|PP=7|TP=7|ip-26-0-171-62]: Max retries reached, giving up on communication
+01/03/2025 01:31:16 [WARNING|DP=2|PP=7|TP=2|ip-26-0-171-62]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 6: [7] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '6:7', but store->get('6:7') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [ERROR|DP=1|PP=7|TP=3|ip-26-0-171-56]: [RECV META] Rank 7: Failed to receive metadata from rank 6: [7] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '6:7', but store->get('6:7') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [WARNING|DP=1|PP=7|TP=2|ip-26-0-171-56]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 6: [7] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '6:7', but store->get('6:7') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [WARNING|DP=3|PP=5|TP=6|ip-26-0-167-9]: No progress made in communication iteration, attempt 1/1
+01/03/2025 01:31:16 [ERROR|DP=2|PP=7|TP=6|ip-26-0-171-62]: [RECV META] Rank 7: Failed to receive metadata from rank 6: [7] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '6:7', but store->get('6:7') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [WARNING|DP=1|PP=7|TP=7|ip-26-0-171-56]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 6: [7] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '6:7', but store->get('6:7') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [ERROR|DP=1|PP=7|TP=5|ip-26-0-171-56]: [RECV META] Rank 7: Failed to receive metadata from rank 6: [7] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '6:7', but store->get('6:7') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [WARNING|DP=3|PP=5|TP=3|ip-26-0-167-9]: No progress made in communication iteration, attempt 1/1
+01/03/2025 01:31:16 [WARNING|DP=2|PP=7|TP=5|ip-26-0-171-62]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 6: [7] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '6:7', but store->get('6:7') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [WARNING|DP=2|PP=7|TP=2|ip-26-0-171-62]: No progress made in communication iteration, attempt 1/1
+01/03/2025 01:31:16 [ERROR|DP=3|PP=6|TP=4|ip-26-0-171-168]: Max retries reached, giving up on communication
+01/03/2025 01:31:16 [WARNING|DP=3|PP=6|TP=7|ip-26-0-171-168]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 5: [6] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '5:6', but store->get('5:6') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [ERROR|DP=0|PP=7|TP=2|ip-26-0-171-21]: [RECV META] Rank 7: Failed to receive metadata from rank 6: [7] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '6:7', but store->get('6:7') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [WARNING|DP=3|PP=6|TP=7|ip-26-0-171-168]: No progress made in communication iteration, attempt 1/1
+01/03/2025 01:31:16 [ERROR|DP=2|PP=6|TP=0|ip-26-0-171-102]: [RECV META] Rank 6: Failed to receive metadata from rank 5: [6] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '5:6', but store->get('5:6') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [ERROR|DP=1|PP=5|TP=1|ip-26-0-167-177]: [RECV META] Rank 5: Failed to receive metadata from rank 4: [5] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '4:5', but store->get('4:5') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [WARNING|DP=1|PP=7|TP=2|ip-26-0-171-56]: No progress made in communication iteration, attempt 1/1
+01/03/2025 01:31:16 [WARNING|DP=1|PP=7|TP=7|ip-26-0-171-56]: No progress made in communication iteration, attempt 1/1
+01/03/2025 01:31:16 [ERROR|DP=3|PP=6|TP=7|ip-26-0-171-168]: Max retries reached, giving up on communication
+01/03/2025 01:31:16 [ERROR|DP=1|PP=7|TP=2|ip-26-0-171-56]: Max retries reached, giving up on communication
+01/03/2025 01:31:16 [WARNING|DP=1|PP=7|TP=3|ip-26-0-171-56]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 6: [7] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '6:7', but store->get('6:7') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [ERROR|DP=1|PP=7|TP=7|ip-26-0-171-56]: Max retries reached, giving up on communication
+01/03/2025 01:31:16 [ERROR|DP=3|PP=5|TP=5|ip-26-0-167-9]: [RECV META] Rank 5: Failed to receive metadata from rank 4: [5] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '4:5', but store->get('4:5') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [ERROR|DP=3|PP=5|TP=2|ip-26-0-167-9]: [RECV META] Rank 5: Failed to receive metadata from rank 4: [5] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '4:5', but store->get('4:5') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [ERROR|DP=3|PP=5|TP=6|ip-26-0-167-9]: Max retries reached, giving up on communication
+01/03/2025 01:31:16 [ERROR|DP=3|PP=5|TP=3|ip-26-0-167-9]: Max retries reached, giving up on communication
+01/03/2025 01:31:16 [WARNING|DP=2|PP=7|TP=3|ip-26-0-171-62]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 6: [7] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '6:7', but store->get('6:7') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [ERROR|DP=2|PP=7|TP=2|ip-26-0-171-62]: Max retries reached, giving up on communication
+01/03/2025 01:31:16 [WARNING|DP=2|PP=7|TP=5|ip-26-0-171-62]: No progress made in communication iteration, attempt 1/1
+01/03/2025 01:31:16 [ERROR|DP=3|PP=6|TP=0|ip-26-0-171-168]: [RECV META] Rank 6: Failed to receive metadata from rank 5: [6] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '5:6', but store->get('5:6') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [WARNING|DP=1|PP=7|TP=5|ip-26-0-171-56]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 6: [7] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '6:7', but store->get('6:7') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [ERROR|DP=0|PP=7|TP=5|ip-26-0-171-21]: [RECV META] Rank 7: Failed to receive metadata from rank 6: [7] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '6:7', but store->get('6:7') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [ERROR|DP=3|PP=5|TP=4|ip-26-0-167-9]: [RECV META] Rank 5: Failed to receive metadata from rank 4: [5] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '4:5', but store->get('4:5') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [WARNING|DP=3|PP=5|TP=5|ip-26-0-167-9]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 4: [5] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '4:5', but store->get('4:5') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [WARNING|DP=2|PP=7|TP=6|ip-26-0-171-62]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 6: [7] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '6:7', but store->get('6:7') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [ERROR|DP=3|PP=6|TP=3|ip-26-0-171-168]: [RECV META] Rank 6: Failed to receive metadata from rank 5: [6] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '5:6', but store->get('5:6') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [ERROR|DP=1|PP=6|TP=5|ip-26-0-170-160]: [RECV META] Rank 6: Failed to receive metadata from rank 5: [6] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '5:6', but store->get('5:6') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [ERROR|DP=1|PP=7|TP=6|ip-26-0-171-56]: [RECV META] Rank 7: Failed to receive metadata from rank 6: [7] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '6:7', but store->get('6:7') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [WARNING|DP=0|PP=7|TP=2|ip-26-0-171-21]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 6: [7] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '6:7', but store->get('6:7') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [WARNING|DP=3|PP=5|TP=2|ip-26-0-167-9]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 4: [5] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '4:5', but store->get('4:5') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [ERROR|DP=3|PP=5|TP=7|ip-26-0-167-9]: [RECV META] Rank 5: Failed to receive metadata from rank 4: [5] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '4:5', but store->get('4:5') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [WARNING|DP=3|PP=5|TP=2|ip-26-0-167-9]: No progress made in communication iteration, attempt 1/1
+01/03/2025 01:31:16 [WARNING|DP=3|PP=5|TP=5|ip-26-0-167-9]: No progress made in communication iteration, attempt 1/1
+01/03/2025 01:31:16 [WARNING|DP=2|PP=7|TP=3|ip-26-0-171-62]: No progress made in communication iteration, attempt 1/1
+01/03/2025 01:31:16 [ERROR|DP=2|PP=7|TP=5|ip-26-0-171-62]: Max retries reached, giving up on communication
+01/03/2025 01:31:16 [WARNING|DP=3|PP=6|TP=0|ip-26-0-171-168]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 5: [6] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '5:6', but store->get('5:6') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [WARNING|DP=2|PP=6|TP=0|ip-26-0-171-102]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 5: [6] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '5:6', but store->get('5:6') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [ERROR|DP=1|PP=5|TP=4|ip-26-0-167-177]: [RECV META] Rank 5: Failed to receive metadata from rank 4: [5] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '4:5', but store->get('4:5') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [WARNING|DP=1|PP=7|TP=3|ip-26-0-171-56]: No progress made in communication iteration, attempt 1/1
+01/03/2025 01:31:16 [WARNING|DP=1|PP=7|TP=5|ip-26-0-171-56]: No progress made in communication iteration, attempt 1/1
+01/03/2025 01:31:16 [WARNING|DP=0|PP=7|TP=5|ip-26-0-171-21]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 6: [7] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '6:7', but store->get('6:7') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [WARNING|DP=0|PP=7|TP=2|ip-26-0-171-21]: No progress made in communication iteration, attempt 1/1
+01/03/2025 01:31:16 [ERROR|DP=3|PP=5|TP=1|ip-26-0-167-9]: [RECV META] Rank 5: Failed to receive metadata from rank 4: [5] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '4:5', but store->get('4:5') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [ERROR|DP=3|PP=5|TP=5|ip-26-0-167-9]: Max retries reached, giving up on communication
+01/03/2025 01:31:16 [ERROR|DP=3|PP=5|TP=2|ip-26-0-167-9]: Max retries reached, giving up on communication
+01/03/2025 01:31:16 [WARNING|DP=3|PP=5|TP=4|ip-26-0-167-9]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 4: [5] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '4:5', but store->get('4:5') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [WARNING|DP=2|PP=7|TP=6|ip-26-0-171-62]: No progress made in communication iteration, attempt 1/1
+01/03/2025 01:31:16 [WARNING|DP=3|PP=6|TP=3|ip-26-0-171-168]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 5: [6] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '5:6', but store->get('5:6') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [WARNING|DP=2|PP=6|TP=0|ip-26-0-171-102]: No progress made in communication iteration, attempt 1/1
+01/03/2025 01:31:16 [ERROR|DP=0|PP=5|TP=0|ip-26-0-167-175]: [RECV META] Rank 5: Failed to receive metadata from rank 4: [5] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '4:5', but store->get('4:5') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [WARNING|DP=1|PP=5|TP=1|ip-26-0-167-177]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 4: [5] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '4:5', but store->get('4:5') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [ERROR|DP=1|PP=7|TP=3|ip-26-0-171-56]: Max retries reached, giving up on communication
+01/03/2025 01:31:16 [ERROR|DP=1|PP=7|TP=5|ip-26-0-171-56]: Max retries reached, giving up on communication
+01/03/2025 01:31:16 [ERROR|DP=0|PP=7|TP=2|ip-26-0-171-21]: Max retries reached, giving up on communication
+01/03/2025 01:31:16 [WARNING|DP=0|PP=7|TP=5|ip-26-0-171-21]: No progress made in communication iteration, attempt 1/1
+01/03/2025 01:31:16 [ERROR|DP=2|PP=7|TP=3|ip-26-0-171-62]: Max retries reached, giving up on communication
+01/03/2025 01:31:16 [WARNING|DP=3|PP=6|TP=0|ip-26-0-171-168]: No progress made in communication iteration, attempt 1/1
+01/03/2025 01:31:16 [WARNING|DP=1|PP=6|TP=5|ip-26-0-170-160]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 5: [6] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '5:6', but store->get('5:6') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [WARNING|DP=1|PP=5|TP=1|ip-26-0-167-177]: No progress made in communication iteration, attempt 1/1
+01/03/2025 01:31:16 [WARNING|DP=1|PP=7|TP=6|ip-26-0-171-56]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 6: [7] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '6:7', but store->get('6:7') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [WARNING|DP=3|PP=5|TP=7|ip-26-0-167-9]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 4: [5] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '4:5', but store->get('4:5') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [WARNING|DP=3|PP=5|TP=4|ip-26-0-167-9]: No progress made in communication iteration, attempt 1/1
+01/03/2025 01:31:16 [ERROR|DP=2|PP=7|TP=1|ip-26-0-171-62]: [RECV META] Rank 7: Failed to receive metadata from rank 6: [7] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '6:7', but store->get('6:7') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [ERROR|DP=2|PP=7|TP=6|ip-26-0-171-62]: Max retries reached, giving up on communication
+01/03/2025 01:31:16 [WARNING|DP=3|PP=6|TP=3|ip-26-0-171-168]: No progress made in communication iteration, attempt 1/1
+01/03/2025 01:31:16 [ERROR|DP=2|PP=6|TP=0|ip-26-0-171-102]: Max retries reached, giving up on communication
+01/03/2025 01:31:16 [WARNING|DP=1|PP=6|TP=5|ip-26-0-170-160]: No progress made in communication iteration, attempt 1/1
+01/03/2025 01:31:16 [WARNING|DP=1|PP=5|TP=4|ip-26-0-167-177]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 4: [5] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '4:5', but store->get('4:5') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [WARNING|DP=3|PP=5|TP=1|ip-26-0-167-9]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 4: [5] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '4:5', but store->get('4:5') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [WARNING|DP=2|PP=7|TP=1|ip-26-0-171-62]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 6: [7] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '6:7', but store->get('6:7') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [ERROR|DP=3|PP=6|TP=0|ip-26-0-171-168]: Max retries reached, giving up on communication
+01/03/2025 01:31:16 [ERROR|DP=3|PP=6|TP=3|ip-26-0-171-168]: Max retries reached, giving up on communication
+01/03/2025 01:31:16 [ERROR|DP=1|PP=6|TP=5|ip-26-0-170-160]: Max retries reached, giving up on communication
+01/03/2025 01:31:16 [WARNING|DP=0|PP=5|TP=0|ip-26-0-167-175]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 4: [5] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '4:5', but store->get('4:5') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [ERROR|DP=1|PP=5|TP=1|ip-26-0-167-177]: Max retries reached, giving up on communication
+01/03/2025 01:31:16 [WARNING|DP=1|PP=7|TP=6|ip-26-0-171-56]: No progress made in communication iteration, attempt 1/1
+01/03/2025 01:31:16 [ERROR|DP=0|PP=7|TP=5|ip-26-0-171-21]: Max retries reached, giving up on communication
+01/03/2025 01:31:16 [ERROR|DP=0|PP=7|TP=0|ip-26-0-171-21]: [RECV META] Rank 7: Failed to receive metadata from rank 6: [7] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '6:7', but store->get('6:7') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [ERROR|DP=3|PP=5|TP=4|ip-26-0-167-9]: Max retries reached, giving up on communication
+01/03/2025 01:31:16 [WARNING|DP=3|PP=5|TP=7|ip-26-0-167-9]: No progress made in communication iteration, attempt 1/1
+01/03/2025 01:31:16 [ERROR|DP=2|PP=5|TP=7|ip-26-0-167-51]: [RECV META] Rank 5: Failed to receive metadata from rank 4: [5] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '4:5', but store->get('4:5') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [WARNING|DP=2|PP=7|TP=1|ip-26-0-171-62]: No progress made in communication iteration, attempt 1/1
+01/03/2025 01:31:16 [WARNING|DP=0|PP=5|TP=0|ip-26-0-167-175]: No progress made in communication iteration, attempt 1/1
+01/03/2025 01:31:16 [WARNING|DP=1|PP=5|TP=4|ip-26-0-167-177]: No progress made in communication iteration, attempt 1/1
+01/03/2025 01:31:16 [ERROR|DP=1|PP=5|TP=6|ip-26-0-167-177]: [RECV META] Rank 5: Failed to receive metadata from rank 4: [5] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '4:5', but store->get('4:5') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [ERROR|DP=1|PP=7|TP=6|ip-26-0-171-56]: Max retries reached, giving up on communication
+01/03/2025 01:31:16 [WARNING|DP=0|PP=7|TP=0|ip-26-0-171-21]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 6: [7] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '6:7', but store->get('6:7') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [WARNING|DP=3|PP=5|TP=1|ip-26-0-167-9]: No progress made in communication iteration, attempt 1/1
+01/03/2025 01:31:16 [ERROR|DP=2|PP=7|TP=4|ip-26-0-171-62]: [RECV META] Rank 7: Failed to receive metadata from rank 6: [7] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '6:7', but store->get('6:7') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [ERROR|DP=3|PP=7|TP=6|ip-26-0-171-88]: [RECV META] Rank 7: Failed to receive metadata from rank 6: [7] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '6:7', but store->get('6:7') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [ERROR|DP=3|PP=7|TP=2|ip-26-0-171-88]: [RECV META] Rank 7: Failed to receive metadata from rank 6: [7] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '6:7', but store->get('6:7') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [ERROR|DP=0|PP=5|TP=0|ip-26-0-167-175]: Max retries reached, giving up on communication
+01/03/2025 01:31:16 [ERROR|DP=1|PP=5|TP=4|ip-26-0-167-177]: Max retries reached, giving up on communication
+01/03/2025 01:31:16 [WARNING|DP=0|PP=7|TP=0|ip-26-0-171-21]: No progress made in communication iteration, attempt 1/1
+01/03/2025 01:31:16 [ERROR|DP=3|PP=5|TP=7|ip-26-0-167-9]: Max retries reached, giving up on communication
+01/03/2025 01:31:16 [ERROR|DP=2|PP=7|TP=1|ip-26-0-171-62]: Max retries reached, giving up on communication
+01/03/2025 01:31:16 [ERROR|DP=1|PP=6|TP=3|ip-26-0-170-160]: [RECV META] Rank 6: Failed to receive metadata from rank 5: [6] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '5:6', but store->get('5:6') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [ERROR|DP=1|PP=7|TP=0|ip-26-0-171-56]: [RECV META] Rank 7: Failed to receive metadata from rank 6: [7] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '6:7', but store->get('6:7') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [ERROR|DP=0|PP=7|TP=0|ip-26-0-171-21]: Max retries reached, giving up on communication
+01/03/2025 01:31:16 [ERROR|DP=3|PP=5|TP=1|ip-26-0-167-9]: Max retries reached, giving up on communication
+01/03/2025 01:31:16 [WARNING|DP=2|PP=5|TP=7|ip-26-0-167-51]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 4: [5] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '4:5', but store->get('4:5') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [WARNING|DP=1|PP=6|TP=3|ip-26-0-170-160]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 5: [6] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '5:6', but store->get('5:6') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [WARNING|DP=1|PP=7|TP=0|ip-26-0-171-56]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 6: [7] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '6:7', but store->get('6:7') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [ERROR|DP=3|PP=5|TP=0|ip-26-0-167-9]: [RECV META] Rank 5: Failed to receive metadata from rank 4: [5] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '4:5', but store->get('4:5') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [WARNING|DP=2|PP=5|TP=7|ip-26-0-167-51]: No progress made in communication iteration, attempt 1/1
+01/03/2025 01:31:16 [ERROR|DP=3|PP=7|TP=5|ip-26-0-171-88]: [RECV META] Rank 7: Failed to receive metadata from rank 6: [7] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '6:7', but store->get('6:7') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [WARNING|DP=1|PP=6|TP=3|ip-26-0-170-160]: No progress made in communication iteration, attempt 1/1
+01/03/2025 01:31:16 [WARNING|DP=1|PP=5|TP=6|ip-26-0-167-177]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 4: [5] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '4:5', but store->get('4:5') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [WARNING|DP=2|PP=7|TP=4|ip-26-0-171-62]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 6: [7] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '6:7', but store->get('6:7') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [WARNING|DP=3|PP=7|TP=6|ip-26-0-171-88]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 6: [7] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '6:7', but store->get('6:7') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [WARNING|DP=3|PP=7|TP=2|ip-26-0-171-88]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 6: [7] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '6:7', but store->get('6:7') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [ERROR|DP=1|PP=6|TP=3|ip-26-0-170-160]: Max retries reached, giving up on communication
+01/03/2025 01:31:16 [WARNING|DP=1|PP=5|TP=6|ip-26-0-167-177]: No progress made in communication iteration, attempt 1/1
+01/03/2025 01:31:16 [WARNING|DP=2|PP=7|TP=4|ip-26-0-171-62]: No progress made in communication iteration, attempt 1/1
+01/03/2025 01:31:16 [WARNING|DP=3|PP=7|TP=5|ip-26-0-171-88]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 6: [7] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '6:7', but store->get('6:7') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [WARNING|DP=3|PP=7|TP=2|ip-26-0-171-88]: No progress made in communication iteration, attempt 1/1
+01/03/2025 01:31:16 [WARNING|DP=3|PP=7|TP=6|ip-26-0-171-88]: No progress made in communication iteration, attempt 1/1
+01/03/2025 01:31:16 [ERROR|DP=1|PP=5|TP=6|ip-26-0-167-177]: Max retries reached, giving up on communication
+01/03/2025 01:31:16 [WARNING|DP=1|PP=7|TP=0|ip-26-0-171-56]: No progress made in communication iteration, attempt 1/1
+01/03/2025 01:31:16 [WARNING|DP=3|PP=5|TP=0|ip-26-0-167-9]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 4: [5] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '4:5', but store->get('4:5') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [ERROR|DP=2|PP=5|TP=3|ip-26-0-167-51]: [RECV META] Rank 5: Failed to receive metadata from rank 4: [5] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '4:5', but store->get('4:5') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [ERROR|DP=2|PP=5|TP=1|ip-26-0-167-51]: [RECV META] Rank 5: Failed to receive metadata from rank 4: [5] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '4:5', but store->get('4:5') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [ERROR|DP=2|PP=5|TP=7|ip-26-0-167-51]: Max retries reached, giving up on communication
+01/03/2025 01:31:16 [ERROR|DP=2|PP=7|TP=4|ip-26-0-171-62]: Max retries reached, giving up on communication
+01/03/2025 01:31:16 [WARNING|DP=3|PP=7|TP=5|ip-26-0-171-88]: No progress made in communication iteration, attempt 1/1
+01/03/2025 01:31:16 [ERROR|DP=3|PP=7|TP=2|ip-26-0-171-88]: Max retries reached, giving up on communication
+01/03/2025 01:31:16 [ERROR|DP=3|PP=7|TP=6|ip-26-0-171-88]: Max retries reached, giving up on communication
+01/03/2025 01:31:16 [ERROR|DP=1|PP=7|TP=0|ip-26-0-171-56]: Max retries reached, giving up on communication
+01/03/2025 01:31:16 [WARNING|DP=3|PP=5|TP=0|ip-26-0-167-9]: No progress made in communication iteration, attempt 1/1
+01/03/2025 01:31:16 [WARNING|DP=2|PP=5|TP=3|ip-26-0-167-51]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 4: [5] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '4:5', but store->get('4:5') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [WARNING|DP=2|PP=5|TP=1|ip-26-0-167-51]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 4: [5] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '4:5', but store->get('4:5') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [ERROR|DP=2|PP=7|TP=0|ip-26-0-171-62]: [RECV META] Rank 7: Failed to receive metadata from rank 6: [7] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '6:7', but store->get('6:7') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [ERROR|DP=1|PP=6|TP=4|ip-26-0-170-160]: [RECV META] Rank 6: Failed to receive metadata from rank 5: [6] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '5:6', but store->get('5:6') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [ERROR|DP=3|PP=5|TP=0|ip-26-0-167-9]: Max retries reached, giving up on communication
+01/03/2025 01:31:16 [WARNING|DP=2|PP=5|TP=3|ip-26-0-167-51]: No progress made in communication iteration, attempt 1/1
+01/03/2025 01:31:16 [WARNING|DP=2|PP=5|TP=1|ip-26-0-167-51]: No progress made in communication iteration, attempt 1/1
+01/03/2025 01:31:16 [WARNING|DP=2|PP=7|TP=0|ip-26-0-171-62]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 6: [7] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '6:7', but store->get('6:7') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [WARNING|DP=1|PP=6|TP=4|ip-26-0-170-160]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 5: [6] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '5:6', but store->get('5:6') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [ERROR|DP=1|PP=6|TP=2|ip-26-0-170-160]: [RECV META] Rank 6: Failed to receive metadata from rank 5: [6] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '5:6', but store->get('5:6') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [ERROR|DP=1|PP=5|TP=7|ip-26-0-167-177]: [RECV META] Rank 5: Failed to receive metadata from rank 4: [5] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '4:5', but store->get('4:5') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [ERROR|DP=2|PP=5|TP=4|ip-26-0-167-51]: [RECV META] Rank 5: Failed to receive metadata from rank 4: [5] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '4:5', but store->get('4:5') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [ERROR|DP=2|PP=5|TP=3|ip-26-0-167-51]: Max retries reached, giving up on communication
+01/03/2025 01:31:16 [ERROR|DP=2|PP=5|TP=1|ip-26-0-167-51]: Max retries reached, giving up on communication
+01/03/2025 01:31:16 [WARNING|DP=2|PP=7|TP=0|ip-26-0-171-62]: No progress made in communication iteration, attempt 1/1
+01/03/2025 01:31:16 [ERROR|DP=3|PP=7|TP=5|ip-26-0-171-88]: Max retries reached, giving up on communication
+01/03/2025 01:31:16 [WARNING|DP=1|PP=6|TP=4|ip-26-0-170-160]: No progress made in communication iteration, attempt 1/1
+01/03/2025 01:31:16 [WARNING|DP=2|PP=5|TP=4|ip-26-0-167-51]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 4: [5] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '4:5', but store->get('4:5') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [ERROR|DP=2|PP=7|TP=0|ip-26-0-171-62]: Max retries reached, giving up on communication
+01/03/2025 01:31:16 [ERROR|DP=3|PP=7|TP=4|ip-26-0-171-88]: [RECV META] Rank 7: Failed to receive metadata from rank 6: [7] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '6:7', but store->get('6:7') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [ERROR|DP=1|PP=6|TP=7|ip-26-0-170-160]: [RECV META] Rank 6: Failed to receive metadata from rank 5: [6] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '5:6', but store->get('5:6') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [ERROR|DP=1|PP=6|TP=0|ip-26-0-170-160]: [RECV META] Rank 6: Failed to receive metadata from rank 5: [6] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '5:6', but store->get('5:6') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [ERROR|DP=1|PP=6|TP=4|ip-26-0-170-160]: Max retries reached, giving up on communication
+01/03/2025 01:31:16 [WARNING|DP=2|PP=5|TP=4|ip-26-0-167-51]: No progress made in communication iteration, attempt 1/1
+01/03/2025 01:31:16 [WARNING|DP=3|PP=7|TP=4|ip-26-0-171-88]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 6: [7] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '6:7', but store->get('6:7') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [ERROR|DP=2|PP=6|TP=2|ip-26-0-171-102]: [RECV META] Rank 6: Failed to receive metadata from rank 5: [6] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '5:6', but store->get('5:6') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [ERROR|DP=2|PP=6|TP=3|ip-26-0-171-102]: [RECV META] Rank 6: Failed to receive metadata from rank 5: [6] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '5:6', but store->get('5:6') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [WARNING|DP=1|PP=6|TP=2|ip-26-0-170-160]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 5: [6] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '5:6', but store->get('5:6') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [ERROR|DP=3|PP=7|TP=7|ip-26-0-171-88]: [RECV META] Rank 7: Failed to receive metadata from rank 6: [7] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '6:7', but store->get('6:7') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [ERROR|DP=1|PP=6|TP=6|ip-26-0-170-160]: [RECV META] Rank 6: Failed to receive metadata from rank 5: [6] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '5:6', but store->get('5:6') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [WARNING|DP=1|PP=6|TP=2|ip-26-0-170-160]: No progress made in communication iteration, attempt 1/1
+01/03/2025 01:31:16 [ERROR|DP=0|PP=5|TP=4|ip-26-0-167-175]: [RECV META] Rank 5: Failed to receive metadata from rank 4: [5] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '4:5', but store->get('4:5') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [ERROR|DP=0|PP=5|TP=1|ip-26-0-167-175]: [RECV META] Rank 5: Failed to receive metadata from rank 4: [5] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '4:5', but store->get('4:5') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [WARNING|DP=1|PP=5|TP=7|ip-26-0-167-177]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 4: [5] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '4:5', but store->get('4:5') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [WARNING|DP=3|PP=7|TP=4|ip-26-0-171-88]: No progress made in communication iteration, attempt 1/1
+01/03/2025 01:31:16 [WARNING|DP=1|PP=6|TP=7|ip-26-0-170-160]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 5: [6] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '5:6', but store->get('5:6') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [ERROR|DP=1|PP=6|TP=1|ip-26-0-170-160]: [RECV META] Rank 6: Failed to receive metadata from rank 5: [6] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '5:6', but store->get('5:6') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [WARNING|DP=1|PP=5|TP=7|ip-26-0-167-177]: No progress made in communication iteration, attempt 1/1
+01/03/2025 01:31:16 [ERROR|DP=2|PP=5|TP=4|ip-26-0-167-51]: Max retries reached, giving up on communication
+01/03/2025 01:31:16 [ERROR|DP=3|PP=7|TP=4|ip-26-0-171-88]: Max retries reached, giving up on communication
+01/03/2025 01:31:16 [WARNING|DP=2|PP=6|TP=3|ip-26-0-171-102]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 5: [6] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '5:6', but store->get('5:6') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [WARNING|DP=2|PP=6|TP=2|ip-26-0-171-102]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 5: [6] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '5:6', but store->get('5:6') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [WARNING|DP=1|PP=6|TP=0|ip-26-0-170-160]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 5: [6] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '5:6', but store->get('5:6') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [ERROR|DP=1|PP=6|TP=2|ip-26-0-170-160]: Max retries reached, giving up on communication
+01/03/2025 01:31:16 [ERROR|DP=2|PP=5|TP=2|ip-26-0-167-51]: [RECV META] Rank 5: Failed to receive metadata from rank 4: [5] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '4:5', but store->get('4:5') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [WARNING|DP=2|PP=6|TP=2|ip-26-0-171-102]: No progress made in communication iteration, attempt 1/1
+01/03/2025 01:31:16 [WARNING|DP=2|PP=6|TP=3|ip-26-0-171-102]: No progress made in communication iteration, attempt 1/1
+01/03/2025 01:31:16 [WARNING|DP=0|PP=5|TP=1|ip-26-0-167-175]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 4: [5] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '4:5', but store->get('4:5') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [WARNING|DP=0|PP=5|TP=4|ip-26-0-167-175]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 4: [5] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '4:5', but store->get('4:5') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [WARNING|DP=2|PP=5|TP=2|ip-26-0-167-51]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 4: [5] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '4:5', but store->get('4:5') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [ERROR|DP=2|PP=6|TP=2|ip-26-0-171-102]: Max retries reached, giving up on communication
+01/03/2025 01:31:16 [ERROR|DP=2|PP=6|TP=3|ip-26-0-171-102]: Max retries reached, giving up on communication
+01/03/2025 01:31:16 [WARNING|DP=0|PP=5|TP=4|ip-26-0-167-175]: No progress made in communication iteration, attempt 1/1
+01/03/2025 01:31:16 [WARNING|DP=0|PP=5|TP=1|ip-26-0-167-175]: No progress made in communication iteration, attempt 1/1
+01/03/2025 01:31:16 [ERROR|DP=1|PP=5|TP=5|ip-26-0-167-177]: [RECV META] Rank 5: Failed to receive metadata from rank 4: [5] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '4:5', but store->get('4:5') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [ERROR|DP=1|PP=5|TP=7|ip-26-0-167-177]: Max retries reached, giving up on communication
+01/03/2025 01:31:16 [ERROR|DP=1|PP=5|TP=2|ip-26-0-167-177]: [RECV META] Rank 5: Failed to receive metadata from rank 4: [5] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '4:5', but store->get('4:5') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [WARNING|DP=2|PP=5|TP=2|ip-26-0-167-51]: No progress made in communication iteration, attempt 1/1
+01/03/2025 01:31:16 [ERROR|DP=0|PP=5|TP=4|ip-26-0-167-175]: Max retries reached, giving up on communication
+01/03/2025 01:31:16 [WARNING|DP=1|PP=5|TP=2|ip-26-0-167-177]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 4: [5] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '4:5', but store->get('4:5') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [WARNING|DP=1|PP=5|TP=5|ip-26-0-167-177]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 4: [5] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '4:5', but store->get('4:5') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [ERROR|DP=2|PP=5|TP=5|ip-26-0-167-51]: [RECV META] Rank 5: Failed to receive metadata from rank 4: [5] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '4:5', but store->get('4:5') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [ERROR|DP=2|PP=5|TP=2|ip-26-0-167-51]: Max retries reached, giving up on communication
+01/03/2025 01:31:16 [WARNING|DP=3|PP=7|TP=7|ip-26-0-171-88]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 6: [7] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '6:7', but store->get('6:7') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [ERROR|DP=3|PP=7|TP=3|ip-26-0-171-88]: [RECV META] Rank 7: Failed to receive metadata from rank 6: [7] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '6:7', but store->get('6:7') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [WARNING|DP=1|PP=6|TP=7|ip-26-0-170-160]: No progress made in communication iteration, attempt 1/1
+01/03/2025 01:31:16 [WARNING|DP=1|PP=6|TP=6|ip-26-0-170-160]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 5: [6] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '5:6', but store->get('5:6') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [WARNING|DP=1|PP=6|TP=0|ip-26-0-170-160]: No progress made in communication iteration, attempt 1/1
+01/03/2025 01:31:16 [ERROR|DP=0|PP=5|TP=1|ip-26-0-167-175]: Max retries reached, giving up on communication
+01/03/2025 01:31:16 [WARNING|DP=1|PP=5|TP=2|ip-26-0-167-177]: No progress made in communication iteration, attempt 1/1
+01/03/2025 01:31:16 [WARNING|DP=1|PP=5|TP=5|ip-26-0-167-177]: No progress made in communication iteration, attempt 1/1
+01/03/2025 01:31:16 [WARNING|DP=2|PP=5|TP=5|ip-26-0-167-51]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 4: [5] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '4:5', but store->get('4:5') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [WARNING|DP=3|PP=7|TP=7|ip-26-0-171-88]: No progress made in communication iteration, attempt 1/1
+01/03/2025 01:31:16 [ERROR|DP=2|PP=6|TP=5|ip-26-0-171-102]: [RECV META] Rank 6: Failed to receive metadata from rank 5: [6] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '5:6', but store->get('5:6') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [ERROR|DP=1|PP=6|TP=7|ip-26-0-170-160]: Max retries reached, giving up on communication
+01/03/2025 01:31:16 [WARNING|DP=1|PP=6|TP=1|ip-26-0-170-160]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 5: [6] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '5:6', but store->get('5:6') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [ERROR|DP=1|PP=6|TP=0|ip-26-0-170-160]: Max retries reached, giving up on communication
+01/03/2025 01:31:16 [WARNING|DP=1|PP=6|TP=6|ip-26-0-170-160]: No progress made in communication iteration, attempt 1/1
+01/03/2025 01:31:16 [ERROR|DP=1|PP=5|TP=2|ip-26-0-167-177]: Max retries reached, giving up on communication
+01/03/2025 01:31:16 [ERROR|DP=1|PP=5|TP=5|ip-26-0-167-177]: Max retries reached, giving up on communication
+01/03/2025 01:31:16 [ERROR|DP=3|PP=7|TP=7|ip-26-0-171-88]: Max retries reached, giving up on communication
+01/03/2025 01:31:16 [WARNING|DP=2|PP=6|TP=5|ip-26-0-171-102]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 5: [6] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '5:6', but store->get('5:6') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [WARNING|DP=1|PP=6|TP=1|ip-26-0-170-160]: No progress made in communication iteration, attempt 1/1
+01/03/2025 01:31:16 [ERROR|DP=1|PP=6|TP=6|ip-26-0-170-160]: Max retries reached, giving up on communication
+01/03/2025 01:31:16 [ERROR|DP=1|PP=5|TP=0|ip-26-0-167-177]: [RECV META] Rank 5: Failed to receive metadata from rank 4: [5] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '4:5', but store->get('4:5') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [WARNING|DP=3|PP=7|TP=3|ip-26-0-171-88]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 6: [7] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '6:7', but store->get('6:7') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [ERROR|DP=2|PP=6|TP=4|ip-26-0-171-102]: [RECV META] Rank 6: Failed to receive metadata from rank 5: [6] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '5:6', but store->get('5:6') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [WARNING|DP=2|PP=6|TP=5|ip-26-0-171-102]: No progress made in communication iteration, attempt 1/1
+01/03/2025 01:31:16 [ERROR|DP=2|PP=6|TP=1|ip-26-0-171-102]: [RECV META] Rank 6: Failed to receive metadata from rank 5: [6] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '5:6', but store->get('5:6') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [ERROR|DP=1|PP=6|TP=1|ip-26-0-170-160]: Max retries reached, giving up on communication
+01/03/2025 01:31:16 [WARNING|DP=3|PP=7|TP=3|ip-26-0-171-88]: No progress made in communication iteration, attempt 1/1
+01/03/2025 01:31:16 [ERROR|DP=2|PP=6|TP=5|ip-26-0-171-102]: Max retries reached, giving up on communication
+01/03/2025 01:31:16 [ERROR|DP=0|PP=5|TP=6|ip-26-0-167-175]: [RECV META] Rank 5: Failed to receive metadata from rank 4: [5] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '4:5', but store->get('4:5') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [ERROR|DP=0|PP=5|TP=7|ip-26-0-167-175]: [RECV META] Rank 5: Failed to receive metadata from rank 4: [5] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '4:5', but store->get('4:5') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [WARNING|DP=2|PP=5|TP=5|ip-26-0-167-51]: No progress made in communication iteration, attempt 1/1
+01/03/2025 01:31:16 [ERROR|DP=3|PP=7|TP=3|ip-26-0-171-88]: Max retries reached, giving up on communication
+01/03/2025 01:31:16 [WARNING|DP=0|PP=5|TP=6|ip-26-0-167-175]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 4: [5] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '4:5', but store->get('4:5') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [WARNING|DP=0|PP=5|TP=7|ip-26-0-167-175]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 4: [5] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '4:5', but store->get('4:5') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [ERROR|DP=2|PP=5|TP=5|ip-26-0-167-51]: Max retries reached, giving up on communication
+01/03/2025 01:31:16 [ERROR|DP=0|PP=5|TP=5|ip-26-0-167-175]: [RECV META] Rank 5: Failed to receive metadata from rank 4: [5] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '4:5', but store->get('4:5') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [WARNING|DP=0|PP=5|TP=6|ip-26-0-167-175]: No progress made in communication iteration, attempt 1/1
+01/03/2025 01:31:16 [WARNING|DP=0|PP=5|TP=7|ip-26-0-167-175]: No progress made in communication iteration, attempt 1/1
+01/03/2025 01:31:16 [WARNING|DP=1|PP=5|TP=0|ip-26-0-167-177]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 4: [5] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '4:5', but store->get('4:5') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [ERROR|DP=2|PP=5|TP=0|ip-26-0-167-51]: [RECV META] Rank 5: Failed to receive metadata from rank 4: [5] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '4:5', but store->get('4:5') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [ERROR|DP=0|PP=5|TP=6|ip-26-0-167-175]: Max retries reached, giving up on communication
+01/03/2025 01:31:16 [ERROR|DP=0|PP=5|TP=7|ip-26-0-167-175]: Max retries reached, giving up on communication
+01/03/2025 01:31:16 [WARNING|DP=1|PP=5|TP=0|ip-26-0-167-177]: No progress made in communication iteration, attempt 1/1
+01/03/2025 01:31:16 [WARNING|DP=2|PP=5|TP=0|ip-26-0-167-51]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 4: [5] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '4:5', but store->get('4:5') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [ERROR|DP=2|PP=6|TP=7|ip-26-0-171-102]: [RECV META] Rank 6: Failed to receive metadata from rank 5: [6] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '5:6', but store->get('5:6') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [WARNING|DP=0|PP=5|TP=5|ip-26-0-167-175]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 4: [5] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '4:5', but store->get('4:5') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [ERROR|DP=1|PP=5|TP=0|ip-26-0-167-177]: Max retries reached, giving up on communication
+01/03/2025 01:31:16 [WARNING|DP=2|PP=5|TP=0|ip-26-0-167-51]: No progress made in communication iteration, attempt 1/1
+01/03/2025 01:31:16 [ERROR|DP=3|PP=7|TP=0|ip-26-0-171-88]: [RECV META] Rank 7: Failed to receive metadata from rank 6: [7] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '6:7', but store->get('6:7') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [WARNING|DP=2|PP=6|TP=4|ip-26-0-171-102]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 5: [6] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '5:6', but store->get('5:6') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [ERROR|DP=0|PP=5|TP=3|ip-26-0-167-175]: [RECV META] Rank 5: Failed to receive metadata from rank 4: [5] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '4:5', but store->get('4:5') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [ERROR|DP=2|PP=5|TP=0|ip-26-0-167-51]: Max retries reached, giving up on communication
+01/03/2025 01:31:16 [WARNING|DP=3|PP=7|TP=0|ip-26-0-171-88]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 6: [7] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '6:7', but store->get('6:7') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [WARNING|DP=2|PP=6|TP=1|ip-26-0-171-102]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 5: [6] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '5:6', but store->get('5:6') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [WARNING|DP=0|PP=5|TP=5|ip-26-0-167-175]: No progress made in communication iteration, attempt 1/1
+01/03/2025 01:31:16 [WARNING|DP=3|PP=7|TP=0|ip-26-0-171-88]: No progress made in communication iteration, attempt 1/1
+01/03/2025 01:31:16 [WARNING|DP=2|PP=6|TP=4|ip-26-0-171-102]: No progress made in communication iteration, attempt 1/1
+01/03/2025 01:31:16 [WARNING|DP=2|PP=6|TP=1|ip-26-0-171-102]: No progress made in communication iteration, attempt 1/1
+01/03/2025 01:31:16 [ERROR|DP=3|PP=7|TP=0|ip-26-0-171-88]: Max retries reached, giving up on communication
+01/03/2025 01:31:16 [WARNING|DP=2|PP=6|TP=7|ip-26-0-171-102]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 5: [6] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '5:6', but store->get('5:6') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [ERROR|DP=2|PP=6|TP=4|ip-26-0-171-102]: Max retries reached, giving up on communication
+01/03/2025 01:31:16 [ERROR|DP=2|PP=6|TP=1|ip-26-0-171-102]: Max retries reached, giving up on communication
+01/03/2025 01:31:16 [ERROR|DP=2|PP=5|TP=6|ip-26-0-167-51]: [RECV META] Rank 5: Failed to receive metadata from rank 4: [5] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '4:5', but store->get('4:5') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [WARNING|DP=2|PP=6|TP=7|ip-26-0-171-102]: No progress made in communication iteration, attempt 1/1
+01/03/2025 01:31:16 [ERROR|DP=2|PP=6|TP=7|ip-26-0-171-102]: Max retries reached, giving up on communication
+01/03/2025 01:31:16 [ERROR|DP=3|PP=7|TP=1|ip-26-0-171-88]: [RECV META] Rank 7: Failed to receive metadata from rank 6: [7] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '6:7', but store->get('6:7') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [WARNING|DP=2|PP=5|TP=6|ip-26-0-167-51]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 4: [5] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '4:5', but store->get('4:5') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [ERROR|DP=0|PP=5|TP=5|ip-26-0-167-175]: Max retries reached, giving up on communication
+01/03/2025 01:31:16 [ERROR|DP=0|PP=5|TP=2|ip-26-0-167-175]: [RECV META] Rank 5: Failed to receive metadata from rank 4: [5] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '4:5', but store->get('4:5') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [WARNING|DP=2|PP=5|TP=6|ip-26-0-167-51]: No progress made in communication iteration, attempt 1/1
+01/03/2025 01:31:16 [WARNING|DP=0|PP=5|TP=3|ip-26-0-167-175]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 4: [5] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '4:5', but store->get('4:5') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [ERROR|DP=2|PP=5|TP=6|ip-26-0-167-51]: Max retries reached, giving up on communication
+01/03/2025 01:31:16 [WARNING|DP=0|PP=5|TP=3|ip-26-0-167-175]: No progress made in communication iteration, attempt 1/1
+01/03/2025 01:31:16 [WARNING|DP=3|PP=7|TP=1|ip-26-0-171-88]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 6: [7] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '6:7', but store->get('6:7') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [WARNING|DP=0|PP=5|TP=2|ip-26-0-167-175]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 4: [5] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '4:5', but store->get('4:5') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [ERROR|DP=0|PP=5|TP=3|ip-26-0-167-175]: Max retries reached, giving up on communication
+01/03/2025 01:31:16 [WARNING|DP=0|PP=5|TP=2|ip-26-0-167-175]: No progress made in communication iteration, attempt 1/1
+01/03/2025 01:31:16 [ERROR|DP=0|PP=5|TP=2|ip-26-0-167-175]: Max retries reached, giving up on communication
+01/03/2025 01:31:16 [WARNING|DP=3|PP=7|TP=1|ip-26-0-171-88]: No progress made in communication iteration, attempt 1/1
+01/03/2025 01:31:16 [ERROR|DP=3|PP=7|TP=1|ip-26-0-171-88]: Max retries reached, giving up on communication
+01/03/2025 01:31:16 [ERROR|DP=1|PP=5|TP=3|ip-26-0-167-177]: [RECV META] Rank 5: Failed to receive metadata from rank 4: [5] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '4:5', but store->get('4:5') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [WARNING|DP=1|PP=5|TP=3|ip-26-0-167-177]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 4: [5] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '4:5', but store->get('4:5') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/03/2025 01:31:16 [WARNING|DP=1|PP=5|TP=3|ip-26-0-167-177]: No progress made in communication iteration, attempt 1/1
+01/03/2025 01:31:16 [ERROR|DP=1|PP=5|TP=3|ip-26-0-167-177]: Max retries reached, giving up on communication
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+Traceback (most recent call last):
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)    
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+Traceback (most recent call last):
+    trainer.train(dataloader)
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    output = model(**micro_batch)
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+    trainer.train(dataloader)
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+    activation = pipeline_state.activations_buffer.popleft()
+IndexError: pop from an empty deque
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+        trainer.train(dataloader)output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+Traceback (most recent call last):
+    return f(*args, **kwargs)
+Traceback (most recent call last):
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    trainer.train(dataloader)
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+Traceback (most recent call last):
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    trainer.train(dataloader)
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+Traceback (most recent call last):
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+Traceback (most recent call last):
+    trainer.train(dataloader)
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    trainer.train(dataloader)
+    trainer.train(dataloader)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+Traceback (most recent call last):
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    trainer.train(dataloader)
+    trainer.train(dataloader)
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+Traceback (most recent call last):
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return f(*args, **kwargs)
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+    trainer.train(dataloader)
+        trainer.train(dataloader)outputs = self.pipeline_engine.train_batch_iter(
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    return f(*args, **kwargs)
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    outputs = self.pipeline_engine.train_batch_iter(
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return f(*args, **kwargs)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return f(*args, **kwargs)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+trainer.train(dataloader)
+    return f(*args, **kwargs)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return f(*args, **kwargs)
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    return f(*args, **kwargs)
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+Traceback (most recent call last):
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return f(*args, **kwargs)
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    trainer.train(dataloader)
+    return f(*args, **kwargs)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+Traceback (most recent call last):
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return f(*args, **kwargs)  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs = self.pipeline_engine.train_batch_iter(
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs = self.pipeline_engine.train_batch_iter(
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    trainer.train(dataloader)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)    
+trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs = self.pipeline_engine.train_batch_iter(
+    output = model(**micro_batch)  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+trainer.train(dataloader)
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    outputs = self.pipeline_engine.train_batch_iter(
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return f(*args, **kwargs)
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    return f(*args, **kwargs)
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    output = model(**micro_batch)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+Traceback (most recent call last):
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+Traceback (most recent call last):
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+    Traceback (most recent call last):
+return f(*args, **kwargs)
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return self._call_impl(*args, **kwargs)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+        return f(*args, **kwargs)return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+Traceback (most recent call last):
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+    return f(*args, **kwargs)
+        output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)return self._call_impl(*args, **kwargs)
+
+    trainer.train(dataloader)
+    outputs = self.pipeline_engine.train_batch_iter(
+    output = model(**micro_batch)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)    
+outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = model(**micro_batch)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+Traceback (most recent call last):
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+        outputs = self.pipeline_engine.train_batch_iter(    outputs = self.pipeline_engine.train_batch_iter(
+outputs = self.pipeline_engine.train_batch_iter(    
+        outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)outputs = self.pipeline_engine.train_batch_iter(
+
+    trainer.train(dataloader)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+output = model(**micro_batch)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    trainer.train(dataloader)
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+Traceback (most recent call last):
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    trainer.train(dataloader)
+    trainer.train(dataloader)
+    output = model(**micro_batch)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+Traceback (most recent call last):
+    output = model(**micro_batch)
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+        output = model(**micro_batch)outputs = self.pipeline_engine.train_batch_iter(
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+trainer.train(dataloader)    
+trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return f(*args, **kwargs)
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+Traceback (most recent call last):
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+trainer.train(dataloader)    
+trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return f(*args, **kwargs)
+        output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    trainer.train(dataloader)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    output = model(**micro_batch)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    output = model(**micro_batch)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+Traceback (most recent call last):
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)Traceback (most recent call last):
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+Traceback (most recent call last):
+        return self._call_impl(*args, **kwargs)return forward_call(*args, **kwargs)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    output = model(**micro_batch)
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    return f(*args, **kwargs)
+    return f(*args, **kwargs)
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs = self.pipeline_engine.train_batch_iter(
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    output = model(**micro_batch)
+    output = model(**micro_batch)
+    return self._call_impl(*args, **kwargs)
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+    return self._call_impl(*args, **kwargs)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = model(**micro_batch)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+        return forward_call(*args, **kwargs)return forward_call(*args, **kwargs)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+    outputs = self.pipeline_engine.train_batch_iter(
+    return f(*args, **kwargs)
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+Traceback (most recent call last):
+Traceback (most recent call last):
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    outputs = self.pipeline_engine.train_batch_iter(
+    return forward_call(*args, **kwargs)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return f(*args, **kwargs)
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+sharded_logits = self.model(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    return self._call_impl(*args, **kwargs)
+        outputs = self.pipeline_engine.train_batch_iter(trainer.train(dataloader)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs = self.pipeline_engine.train_batch_iter(
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    output = model(**micro_batch)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+Traceback (most recent call last):
+    return f(*args, **kwargs)
+    return forward_call(*args, **kwargs)
+    outputs = self.pipeline_engine.train_batch_iter(
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return f(*args, **kwargs)
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+Traceback (most recent call last):
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return f(*args, **kwargs)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return self._call_impl(*args, **kwargs)
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+Traceback (most recent call last):
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    return self._call_impl(*args, **kwargs)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return f(*args, **kwargs)
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    trainer.train(dataloader)
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    sharded_logits = self.model(    
+    trainer.train(dataloader)
+    sharded_logits = self.model(
+        return self._call_impl(*args, **kwargs)outputs = self.pipeline_engine.train_batch_iter(
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    return self._call_impl(*args, **kwargs)
+sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+Traceback (most recent call last):
+    return self._call_impl(*args, **kwargs)  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+
+    return self._call_impl(*args, **kwargs)
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+    outputs = self.pipeline_engine.train_batch_iter(
+    return forward_call(*args, **kwargs)
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    return self._call_impl(*args, **kwargs)    
+return forward_call(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)    
+trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    output = model(**micro_batch)
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    trainer.train(dataloader)    
+    output = model(**micro_batch)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+        trainer.train(dataloader)output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    return forward_call(*args, **kwargs)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+        output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)output = model(**micro_batch)
+
+return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    return self._call_impl(*args, **kwargs)    
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+        return forward_call(*args, **kwargs)return forward_call(*args, **kwargs)
+
+    return f(*args, **kwargs)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    output = model(**micro_batch)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    outputs = self.pipeline_engine.train_batch_iter(
+    output = model(**micro_batch)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    sharded_logits = self.model(
+    trainer.train(dataloader)
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    return f(*args, **kwargs)
+        output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)        output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+return self._call_impl(*args, **kwargs)trainer.train(dataloader)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    outputs = self.pipeline_engine.train_batch_iter(
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    trainer.train(dataloader)    
+    return self._call_impl(*args, **kwargs)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    outputs = self.pipeline_engine.train_batch_iter(
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    output = model(**micro_batch)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    activation = pipeline_state.activations_buffer.popleft()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs = self.pipeline_engine.train_batch_iter(
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+IndexError: pop from an empty deque
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+        output = model(**micro_batch)return self._call_impl(*args, **kwargs)    
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+return forward_call(*args, **kwargs)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    trainer.train(dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return forward_call(*args, **kwargs)
+    return f(*args, **kwargs)
+    trainer.train(dataloader)    
+return self._call_impl(*args, **kwargs)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    activation = pipeline_state.activations_buffer.popleft()
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    outputs = self.pipeline_engine.train_batch_iter(
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+trainer.train(dataloader)    
+trainer.train(dataloader)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+IndexError: pop from an empty deque
+        return f(*args, **kwargs)return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    output = model(**micro_batch)
+    output = model(**micro_batch)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+        outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)    outputs = self.pipeline_engine.train_batch_iter(
+output = model(**micro_batch)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    output = model(**micro_batch)
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+    trainer.train(dataloader)
+    output = model(**micro_batch)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+        return f(*args, **kwargs)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return self._call_impl(*args, **kwargs)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    output = model(**micro_batch)
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    sharded_logits = self.model(
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+return f(*args, **kwargs)    
+return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return self._call_impl(*args, **kwargs)
+    return f(*args, **kwargs)
+        output = model(**micro_batch)return self._call_impl(*args, **kwargs)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return f(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return f(*args, **kwargs)
+    output = model(**micro_batch)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+        return self._call_impl(*args, **kwargs)return self._call_impl(*args, **kwargs)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    sharded_logits = self.model(
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    output = model(**micro_batch)
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+        output = model(**micro_batch)return forward_call(*args, **kwargs)
+
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+        return self._call_impl(*args, **kwargs)return forward_call(*args, **kwargs)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+return self._call_impl(*args, **kwargs)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+return f(*args, **kwargs)
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+    return self._call_impl(*args, **kwargs)
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)    
+return forward_call(*args, **kwargs)  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+
+    return forward_call(*args, **kwargs)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+        return forward_call(*args, **kwargs)outputs = self.pipeline_engine.train_batch_iter(
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    sharded_logits = self.model(
+    return f(*args, **kwargs)
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    sharded_logits = self.model(
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return self._call_impl(*args, **kwargs)
+    sharded_logits = self.model(  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    sharded_logits = self.model(  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    sharded_logits = self.model(
+    return forward_call(*args, **kwargs)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return f(*args, **kwargs)
+hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    sharded_logits = self.model(
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    output = model(**micro_batch)
+    sharded_logits = self.model(
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    output = model(**micro_batch)
+output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    sharded_logits = self.model(
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+        hidden_encoder_states = encoder_block(**hidden_encoder_states)output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)    
+
+sharded_logits = self.model(  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+            return self._call_impl(*args, **kwargs)return self._call_impl(*args, **kwargs)return self._call_impl(*args, **kwargs)
+
+
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+return self._call_impl(*args, **kwargs)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+
+    output = model(**micro_batch)
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    sharded_logits = self.model(
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    output = model(**micro_batch)
+    return forward_call(*args, **kwargs)
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return self._call_impl(*args, **kwargs)  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    trainer.train(dataloader)
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+    return forward_call(*args, **kwargs)
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)    
+output = model(**micro_batch)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+        return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+return self._call_impl(*args, **kwargs)  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    output = model(**micro_batch)
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return self._call_impl(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+    return forward_call(*args, **kwargs)
+    return forward_call(*args, **kwargs)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+        return forward_call(*args, **kwargs)return forward_call(*args, **kwargs)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+        outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)sharded_logits = self.model(
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return self._call_impl(*args, **kwargs)
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+sharded_logits = self.model(
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+        return self._call_impl(*args, **kwargs)return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+        outputs = self.pipeline_engine.train_batch_iter(outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    return self._call_impl(*args, **kwargs)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+        return forward_call(*args, **kwargs)return forward_call(*args, **kwargs)
+
+    return forward_call(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)    
+outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return self._call_impl(*args, **kwargs)
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return forward_call(*args, **kwargs)
+    return forward_call(*args, **kwargs)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+        outputs = self.pipeline_engine.train_batch_iter(outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    sharded_logits = self.model(
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    outputs = self.pipeline_engine.train_batch_iter(    
+outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+hidden_encoder_states = encoder_block(**hidden_encoder_states)
+    sharded_logits = self.model(
+    outputs = self.pipeline_engine.train_batch_iter(
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+        outputs = self.pipeline_engine.train_batch_iter(outputs = self.pipeline_engine.train_batch_iter(
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    outputs = self.pipeline_engine.train_batch_iter(
+    return self._call_impl(*args, **kwargs)
+        return forward_call(*args, **kwargs)return forward_call(*args, **kwargs)
+
+        output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)return self._call_impl(*args, **kwargs)
+
+        
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+return self._call_impl(*args, **kwargs)output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    outputs = self.pipeline_engine.train_batch_iter(
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+        return self._call_impl(*args, **kwargs)
+    outputs = self.pipeline_engine.train_batch_iter(
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return forward_call(*args, **kwargs)
+return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+    outputs = self.pipeline_engine.train_batch_iter(
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+    sharded_logits = self.model(
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+    sharded_logits = self.model(
+outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]    
+return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    return self._call_impl(*args, **kwargs)    
+return self._call_impl(*args, **kwargs)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    sharded_logits = self.model(
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+    return forward_call(*args, **kwargs)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return forward_call(*args, **kwargs)
+    sharded_logits = self.model(
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+    sharded_logits = self.model(  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return self._call_impl(*args, **kwargs)    
+return self._call_impl(*args, **kwargs)    
+        hidden_encoder_states = encoder_block(**hidden_encoder_states)hidden_encoder_states = encoder_block(**hidden_encoder_states)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)    
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+return self._call_impl(*args, **kwargs)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    return self._call_impl(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+    return forward_call(*args, **kwargs)
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+        outputs = self.pipeline_engine.train_batch_iter(outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+    sharded_logits = self.model(
+    sharded_logits = self.model(
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+    return self._call_impl(*args, **kwargs)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+sharded_logits = self.model(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+        return forward_call(*args, **kwargs)output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+
+    return forward_call(*args, **kwargs)
+    sharded_logits = self.model(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    outputs = self.pipeline_engine.train_batch_iter(
+    output = model(**micro_batch)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+    return self._call_impl(*args, **kwargs)
+    sharded_logits = self.model(
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]    
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)    
+output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]    
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    sharded_logits = self.model(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+        sharded_logits = self.model(activation = pipeline_state.activations_buffer.popleft()
+
+    return self._call_impl(*args, **kwargs)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+    activation = pipeline_state.activations_buffer.popleft()
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    return forward_call(*args, **kwargs)    
+return forward_call(*args, **kwargs)  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+IndexError: pop from an empty deque
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+        hidden_encoder_states = encoder_block(**hidden_encoder_states)hidden_encoder_states = encoder_block(**hidden_encoder_states)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+        outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)hidden_encoder_states = encoder_block(**hidden_encoder_states)
+
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+IndexError: pop from an empty deque
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    output = model(**micro_batch)
+    output = model(**micro_batch)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    activation = pipeline_state.activations_buffer.popleft()
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+IndexError: pop from an empty deque
+    output = model(**micro_batch)
+    output = model(**micro_batch)
+        return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+sharded_logits = self.model(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    sharded_logits = self.model(
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    outputs = self.pipeline_engine.train_batch_iter(
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+hidden_encoder_states = encoder_block(**hidden_encoder_states)
+    activation = pipeline_state.activations_buffer.popleft()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+IndexError: pop from an empty deque
+    output = model(**micro_batch)
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = model(**micro_batch)
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+        return self._call_impl(*args, **kwargs)return self._call_impl(*args, **kwargs)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = model(**micro_batch)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+        return self._call_impl(*args, **kwargs)return self._call_impl(*args, **kwargs)
+
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+output = model(**micro_batch)
+        return forward_call(*args, **kwargs)return self._call_impl(*args, **kwargs)
+
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+        return forward_call(*args, **kwargs)return forward_call(*args, **kwargs)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    output = model(**micro_batch)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    return self._call_impl(*args, **kwargs)
+    return forward_call(*args, **kwargs)
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+    return forward_call(*args, **kwargs)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+    return forward_call(*args, **kwargs)
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    output = model(**micro_batch)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+    return forward_call(*args, **kwargs)
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)    
+sharded_logits = self.model(      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+
+sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+    activation = pipeline_state.activations_buffer.popleft()
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+IndexError: pop from an empty deque
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+        activation = pipeline_state.activations_buffer.popleft()hidden_encoder_states = encoder_block(**hidden_encoder_states)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+IndexError: pop from an empty deque
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+    activation = pipeline_state.activations_buffer.popleft()
+IndexError: pop from an empty deque
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    activation = pipeline_state.activations_buffer.popleft()
+IndexError:     pop from an empty deque
+return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    activation = pipeline_state.activations_buffer.popleft()
+IndexError: pop from an empty deque
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+        hidden_encoder_states = encoder_block(**hidden_encoder_states)hidden_encoder_states = encoder_block(**hidden_encoder_states)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+    return self._call_impl(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+        return self._call_impl(*args, **kwargs)return self._call_impl(*args, **kwargs)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+return self._call_impl(*args, **kwargs)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)
+return forward_call(*args, **kwargs)
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+        return forward_call(*args, **kwargs)return self._call_impl(*args, **kwargs)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+return self._call_impl(*args, **kwargs)    
+return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+return self._call_impl(*args, **kwargs)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+    return forward_call(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self._call_impl(*args, **kwargs)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+        return self._call_impl(*args, **kwargs)return self._call_impl(*args, **kwargs)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+return self._call_impl(*args, **kwargs)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+return self._call_impl(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+            return forward_call(*args, **kwargs)    return forward_call(*args, **kwargs)return forward_call(*args, **kwargs)return forward_call(*args, **kwargs)
+    
+
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self._call_impl(*args, **kwargs)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+return self._call_impl(*args, **kwargs)
+
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+return forward_call(*args, **kwargs)  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+return self._call_impl(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+return self._call_impl(*args, **kwargs)
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+return self._call_impl(*args, **kwargs)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+        return self._call_impl(*args, **kwargs)
+hidden_encoder_states = encoder_block(**hidden_encoder_states)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+    return forward_call(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+        return forward_call(*args, **kwargs)new_kwargs[name] = recv_from_pipeline_state_buffer(
+
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+        return self._call_impl(*args, **kwargs)return self._call_impl(*args, **kwargs)    
+
+            return forward_call(*args, **kwargs)return forward_call(*args, **kwargs)    return forward_call(*args, **kwargs)
+    
+return forward_call(*args, **kwargs)  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+
+return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+        return forward_call(*args, **kwargs)    return forward_call(*args, **kwargs)
+return self._call_impl(*args, **kwargs)
+    
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+    return self._call_impl(*args, **kwargs)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+    return self._call_impl(*args, **kwargs)
+    return forward_call(*args, **kwargs)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+        return forward_call(*args, **kwargs)return forward_call(*args, **kwargs)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    return forward_call(*args, **kwargs)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    return forward_call(*args, **kwargs)
+    return forward_call(*args, **kwargs)
+    return forward_call(*args, **kwargs)  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+
+    activation = pipeline_state.activations_buffer.popleft()
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+IndexError: pop from an empty deque
+    return forward_call(*args, **kwargs)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    return forward_call(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+    return forward_call(*args, **kwargs)  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+    return forward_call(*args, **kwargs)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+    sharded_logits = self.model(
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+    return forward_call(*args, **kwargs)  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+
+    activation = pipeline_state.activations_buffer.popleft()
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    IndexErrorhidden_encoder_states = encoder_block(**hidden_encoder_states)    
+    return forward_call(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)return forward_call(*args, **kwargs)
+
+: hidden_encoder_states = encoder_block(**hidden_encoder_states)pop from an empty deque
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    activation = pipeline_state.activations_buffer.popleft()
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    activation = pipeline_state.activations_buffer.popleft()
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+    return forward_call(*args, **kwargs)
+        new_kwargs[name] = recv_from_pipeline_state_buffer(new_kwargs[name] = recv_from_pipeline_state_buffer(
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+IndexError: pop from an empty deque
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+IndexError: pop from an empty deque
+    return self._call_impl(*args, **kwargs)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+    sharded_logits = self.model(
+    activation = pipeline_state.activations_buffer.popleft()
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+IndexError: pop from an empty deque
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+new_kwargs[name] = recv_from_pipeline_state_buffer(
+    sharded_logits = self.model(
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+        new_kwargs[name] = recv_from_pipeline_state_buffer(return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+    return self._call_impl(*args, **kwargs)
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+activation = pipeline_state.activations_buffer.popleft()
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+    activation = pipeline_state.activations_buffer.popleft()
+    return self._call_impl(*args, **kwargs)
+    activation = pipeline_state.activations_buffer.popleft()
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+IndexError: pop from an empty deque
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+IndexError: pop from an empty deque
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+IndexError: pop from an empty deque
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    activation = pipeline_state.activations_buffer.popleft()
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    activation = pipeline_state.activations_buffer.popleft()
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+    activation = pipeline_state.activations_buffer.popleft()
+    sharded_logits = self.model(
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    activation = pipeline_state.activations_buffer.popleft()
+IndexError: pop from an empty deque
+IndexError: pop from an empty deque
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+IndexError: pop from an empty deque
+sharded_logits = self.model(    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+IndexError: pop from an empty deque
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+    activation = pipeline_state.activations_buffer.popleft()
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+IndexError: pop from an empty deque
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    activation = pipeline_state.activations_buffer.popleft()
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+        activation = pipeline_state.activations_buffer.popleft()hidden_encoder_states = encoder_block(**hidden_encoder_states)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    activation = pipeline_state.activations_buffer.popleft()
+IndexError: pop from an empty deque
+    activation = pipeline_state.activations_buffer.popleft()        
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+IndexError: pop from an empty deque
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+return self._call_impl(*args, **kwargs)hidden_encoder_states = encoder_block(**hidden_encoder_states)
+
+IndexError  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+: pop from an empty deque  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+    activation = pipeline_state.activations_buffer.popleft()
+        activation = pipeline_state.activations_buffer.popleft()hidden_encoder_states = encoder_block(**hidden_encoder_states)
+
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+IndexError: pop from an empty deque
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    activation = pipeline_state.activations_buffer.popleft()
+
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+IndexError: pop from an empty deque
+IndexError: pop from an empty deque
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    IndexError  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+new_kwargs[name] = recv_from_pipeline_state_buffer(    
+    return forward_call(*args, **kwargs)
+    sharded_logits = self.model(
+        hidden_encoder_states = encoder_block(**hidden_encoder_states)hidden_encoder_states = encoder_block(**hidden_encoder_states)
+
+return forward_call(*args, **kwargs): 
+pop from an empty deque
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+new_kwargs[name] = recv_from_pipeline_state_buffer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+sharded_logits = self.model(
+    activation = pipeline_state.activations_buffer.popleft()    
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+    sharded_logits = self.model(  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+activation = pipeline_state.activations_buffer.popleft()
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    sharded_logits = self.model(sharded_logits = self.model(
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+IndexErrorIndexError: : pop from an empty dequepop from an empty deque
+
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)return self._call_impl(*args, **kwargs)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    activation = pipeline_state.activations_buffer.popleft()
+    activation = pipeline_state.activations_buffer.popleft()
+    return forward_call(*args, **kwargs)
+    activation = pipeline_state.activations_buffer.popleft()
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+        return self._call_impl(*args, **kwargs)return self._call_impl(*args, **kwargs)
+
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+    return self._call_impl(*args, **kwargs)    
+IndexError: pop from an empty deque
+IndexError: pop from an empty deque
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+IndexError: pop from an empty deque
+    activation = pipeline_state.activations_buffer.popleft()
+return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+IndexError: pop from an empty deque
+    return self._call_impl(*args, **kwargs)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+
+    return self._call_impl(*args, **kwargs)  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    activation = pipeline_state.activations_buffer.popleft()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+IndexError: pop from an empty deque
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    activation = pipeline_state.activations_buffer.popleft()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+    activation = pipeline_state.activations_buffer.popleft()
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+IndexError: pop from an empty deque
+IndexError: pop from an empty deque
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+    activation = pipeline_state.activations_buffer.popleft()
+    return forward_call(*args, **kwargs)
+    return forward_call(*args, **kwargs)  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+    activation = pipeline_state.activations_buffer.popleft()
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+IndexError: pop from an empty deque
+IndexError: pop from an empty deque
+    sharded_logits = self.model(
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+sharded_logits = self.model(
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+        sharded_logits = self.model(return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    sharded_logits = self.model(    
+sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+    sharded_logits = self.model(
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    sharded_logits = self.model(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    sharded_logits = self.model(  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    activation = pipeline_state.activations_buffer.popleft()
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    activation = pipeline_state.activations_buffer.popleft()
+IndexError: pop from an empty deque
+    return self._call_impl(*args, **kwargs)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+    activation = pipeline_state.activations_buffer.popleft()
+IndexError    : activation = pipeline_state.activations_buffer.popleft()pop from an empty deque
+IndexError
+    return forward_call(*args, **kwargs)    
+: pop from an empty deque
+IndexError: pop from an empty deque
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+    activation = pipeline_state.activations_buffer.popleft()
+IndexError: pop from an empty deque
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+        hidden_encoder_states = encoder_block(**hidden_encoder_states)return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+
+    return forward_call(*args, **kwargs)
+          File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    activation = pipeline_state.activations_buffer.popleft()
+IndexError: pop from an empty deque
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+return forward_call(*args, **kwargs)return self._call_impl(*args, **kwargs)
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    activation = pipeline_state.activations_buffer.popleft()
+    return self._call_impl(*args, **kwargs)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+IndexError: pop from an empty deque
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+    activation = pipeline_state.activations_buffer.popleft()
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+IndexError: pop from an empty deque
+    activation = pipeline_state.activations_buffer.popleft()
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+IndexError: pop from an empty deque
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+    return forward_call(*args, **kwargs)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+    return self._call_impl(*args, **kwargs)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+    activation = pipeline_state.activations_buffer.popleft()
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+IndexError: pop from an empty deque
+    activation = pipeline_state.activations_buffer.popleft()
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+IndexError: pop from an empty deque
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+        return self._call_impl(*args, **kwargs)return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    activation = pipeline_state.activations_buffer.popleft()
+IndexError: pop from an empty deque
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+            return forward_call(*args, **kwargs)return forward_call(*args, **kwargs)return forward_call(*args, **kwargs)
+
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+    activation = pipeline_state.activations_buffer.popleft()
+    return self._call_impl(*args, **kwargs)
+    activation = pipeline_state.activations_buffer.popleft()
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)    
+return self._call_impl(*args, **kwargs)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+
+IndexError: pop from an empty deque
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+IndexError: pop from an empty deque
+    return self._call_impl(*args, **kwargs)
+    return forward_call(*args, **kwargs)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)    
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    activation = pipeline_state.activations_buffer.popleft()
+    return forward_call(*args, **kwargs)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+IndexError: pop from an empty deque
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+    return self._call_impl(*args, **kwargs)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+
+    return self._call_impl(*args, **kwargs)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    activation = pipeline_state.activations_buffer.popleft()
+    return forward_call(*args, **kwargs)
+    return forward_call(*args, **kwargs)
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+        return forward_call(*args, **kwargs)return forward_call(*args, **kwargs)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+IndexError  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+: pop from an empty deque
+new_kwargs[name] = recv_from_pipeline_state_buffer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+        return forward_call(*args, **kwargs)return forward_call(*args, **kwargs)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+    activation = pipeline_state.activations_buffer.popleft()
+    return self._call_impl(*args, **kwargs)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+return self._call_impl(*args, **kwargs)
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+IndexError: pop from an empty deque
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+            return forward_call(*args, **kwargs)return forward_call(*args, **kwargs)    return forward_call(*args, **kwargs)
+return self._call_impl(*args, **kwargs)
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+    
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+return forward_call(*args, **kwargs)  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    activation = pipeline_state.activations_buffer.popleft()
+    activation = pipeline_state.activations_buffer.popleft()
+IndexError    : activation = pipeline_state.activations_buffer.popleft()
+pop from an empty deque
+IndexError: pop from an empty dequeIndexError
+: pop from an empty deque
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+    activation = pipeline_state.activations_buffer.popleft()
+IndexError: pop from an empty deque
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+    activation = pipeline_state.activations_buffer.popleft()
+    activation = pipeline_state.activations_buffer.popleft()
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+IndexError: pop from an empty deque
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+IndexError: pop from an empty deque
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self._call_impl(*args, **kwargs)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+    activation = pipeline_state.activations_buffer.popleft()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    activation = pipeline_state.activations_buffer.popleft()    
+activation = pipeline_state.activations_buffer.popleft()IndexError
+: IndexErrorpop from an empty deque: 
+pop from an empty dequeIndexError
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+: pop from an empty deque
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    activation = pipeline_state.activations_buffer.popleft()
+    activation = pipeline_state.activations_buffer.popleft()
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]    
+return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+IndexError: pop from an empty deque
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    activation = pipeline_state.activations_buffer.popleft()
+IndexError: pop from an empty deque
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+    return forward_call(*args, **kwargs)
+IndexError: pop from an empty deque
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+        hidden_encoder_states = encoder_block(**hidden_encoder_states)hidden_encoder_states = encoder_block(**hidden_encoder_states)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+    activation = pipeline_state.activations_buffer.popleft()
+IndexError: pop from an empty deque
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+    activation = pipeline_state.activations_buffer.popleft()
+IndexError: pop from an empty deque
+    return self._call_impl(*args, **kwargs)
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+    activation = pipeline_state.activations_buffer.popleft()
+    new_kwargs[name] = recv_from_pipeline_state_buffer(    
+return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+IndexError: pop from an empty deque
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    activation = pipeline_state.activations_buffer.popleft()
+IndexError: pop from an empty deque
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]    
+return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]    
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return self._call_impl(*args, **kwargs)
+return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    activation = pipeline_state.activations_buffer.popleft()
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+IndexError: pop from an empty deque
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+        hidden_encoder_states = encoder_block(**hidden_encoder_states)hidden_encoder_states = encoder_block(**hidden_encoder_states)    
+
+hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+
+              File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+return forward_call(*args, **kwargs)return forward_call(*args, **kwargs)return forward_call(*args, **kwargs)
+
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    activation = pipeline_state.activations_buffer.popleft()
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+    return forward_call(*args, **kwargs)
+IndexError: pop from an empty deque
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+        return forward_call(*args, **kwargs)return self._call_impl(*args, **kwargs)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+    activation = pipeline_state.activations_buffer.popleft()
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    activation = pipeline_state.activations_buffer.popleft()
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+IndexError: pop from an empty deque
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+return self._call_impl(*args, **kwargs)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+IndexError: pop from an empty deque
+return self._call_impl(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+new_kwargs[name] = recv_from_pipeline_state_buffer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+    new_kwargs[name] = recv_from_pipeline_state_buffer(  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+    activation = pipeline_state.activations_buffer.popleft()
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+IndexError: pop from an empty deque
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+    activation = pipeline_state.activations_buffer.popleft()
+new_kwargs[name] = recv_from_pipeline_state_buffer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+IndexError: pop from an empty deque
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+            return self._call_impl(*args, **kwargs)return forward_call(*args, **kwargs)return forward_call(*args, **kwargs)
+    
+
+return forward_call(*args, **kwargs)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+    activation = pipeline_state.activations_buffer.popleft()
+IndexError: pop from an empty deque
+    activation = pipeline_state.activations_buffer.popleft()
+IndexError: pop from an empty deque
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+    activation = pipeline_state.activations_buffer.popleft()
+IndexError:     pop from an empty dequeactivation = pipeline_state.activations_buffer.popleft()
+
+    activation = pipeline_state.activations_buffer.popleft()
+    IndexErroractivation = pipeline_state.activations_buffer.popleft(): pop from an empty deque
+IndexError
+: IndexErrorpop from an empty deque: 
+pop from an empty deque
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+    activation = pipeline_state.activations_buffer.popleft()
+IndexError: pop from an empty deque
+new_kwargs[name] = recv_from_pipeline_state_buffer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+    activation = pipeline_state.activations_buffer.popleft()
+IndexError: pop from an empty deque
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+    activation = pipeline_state.activations_buffer.popleft()
+IndexError: pop from an empty deque
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+    activation = pipeline_state.activations_buffer.popleft()
+IndexError: pop from an empty deque
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+    activation = pipeline_state.activations_buffer.popleft()
+IndexError: pop from an empty deque
+    activation = pipeline_state.activations_buffer.popleft()
+    IndexErroractivation = pipeline_state.activations_buffer.popleft(): 
+pop from an empty deque
+IndexError: pop from an empty deque
+    activation = pipeline_state.activations_buffer.popleft()
+IndexError: pop from an empty deque
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 160, in forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 725, in _core_forward
+    output = self.attn(hidden_states=hidden_states, sequence_mask=sequence_mask)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 690, in forward
+    output = self.o_proj(attention_output)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 469, in forward
+    out = differentiable_reduce_scatter_sum(out, group=group)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/distributed_differentiable_primitives.py", line 146, in differentiable_reduce_scatter_sum
+    return DifferentiableReduceScatterSum.apply(tensor, group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/distributed_differentiable_primitives.py", line 119, in forward
+    dist.reduce_scatter_tensor(sharded_tensor, tensor, group=group, op=dist.ReduceOp.SUM)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/distributed.py", line 43, in reduce_scatter_tensor
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/c10d_logger.py", line 47, in wrapper
+    return dist.reduce_scatter_tensor(output=output, input=input, group=group, op=op, async_op=async_op)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/c10d_logger.py", line 47, in wrapper
+    return func(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/distributed_c10d.py", line 3375, in reduce_scatter_tensor
+    return func(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/distributed_c10d.py", line 3375, in reduce_scatter_tensor
+    work = group._reduce_scatter_base(output, input, opts)
+RuntimeError: Rank 2 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+ Original exception: 
+[../third_party/gloo/gloo/transport/tcp/pair.cc:534] Connection closed by peer [26.0.163.236]:27035
+
+During handling of the above exception, another exception occurred:
+
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    work = group._reduce_scatter_base(output, input, opts)
+RuntimeError: Rank 7 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+ Original exception: 
+[../third_party/gloo/gloo/transport/tcp/pair.cc:534] Connection closed by peer [26.0.163.236]:27035
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/c10d_logger.py", line 47, in wrapper
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return func(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/distributed_c10d.py", line 3375, in reduce_scatter_tensor
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    work = group._reduce_scatter_base(output, input, opts)
+RuntimeError: Rank 5 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+ Original exception: 
+[../third_party/gloo/gloo/transport/tcp/pair.cc:534] Connection closed by peer [26.0.163.236]:27035
+
+During handling of the above exception, another exception occurred:
+
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+        return forward_call(*args, **kwargs)outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/c10d_logger.py", line 47, in wrapper
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return func(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/distributed_c10d.py", line 3375, in reduce_scatter_tensor
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    work = group._reduce_scatter_base(output, input, opts)
+RuntimeError: Rank 1 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+ Original exception: 
+[../third_party/gloo/gloo/transport/tcp/pair.cc:534] Connection closed by peer [26.0.163.236]:27035
+
+During handling of the above exception, another exception occurred:
+
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/c10d_logger.py", line 47, in wrapper
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/c10d_logger.py", line 47, in wrapper
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+    return f(*args, **kwargs)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return func(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/distributed_c10d.py", line 3375, in reduce_scatter_tensor
+    return func(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/distributed_c10d.py", line 3375, in reduce_scatter_tensor
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+        return forward_call(*args, **kwargs)work = group._reduce_scatter_base(output, input, opts)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 160, in forward
+RuntimeError: Rank 4 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+ Original exception: 
+[../third_party/gloo/gloo/transport/tcp/pair.cc:534] Connection closed by peer [26.0.163.236]:27035
+
+During handling of the above exception, another exception occurred:
+
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    work = group._reduce_scatter_base(output, input, opts)
+RuntimeError: Rank 6 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+ Original exception: 
+[../third_party/gloo/gloo/transport/tcp/pair.cc:534] Connection closed by peer [26.0.163.236]:27035
+
+During handling of the above exception, another exception occurred:
+
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+        output = self.pp_block(**new_kwargs)
+trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return forward_call(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 160, in forward
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 725, in _core_forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+    output = self.pp_block(**new_kwargs)
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    output = self.attn(hidden_states=hidden_states, sequence_mask=sequence_mask)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 690, in forward
+return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 725, in _core_forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.attn(hidden_states=hidden_states, sequence_mask=sequence_mask)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+output = self.o_proj(attention_output)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)    
+return self._call_impl(*args, **kwargs)    
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+return self._call_impl(*args, **kwargs)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 690, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+        sharded_logits = self.model(return row_linear(
+
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.o_proj(attention_output)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return self._call_impl(*args, **kwargs)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 160, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 469, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+        return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]return row_linear(
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    out = differentiable_reduce_scatter_sum(out, group=group)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/distributed_differentiable_primitives.py", line 146, in differentiable_reduce_scatter_sum
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 160, in forward
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return DifferentiableReduceScatterSum.apply(tensor, group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 469, in forward
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/distributed_differentiable_primitives.py", line 119, in forward
+    return forward_call(*args, **kwargs)    
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    out = differentiable_reduce_scatter_sum(out, group=group)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/distributed_differentiable_primitives.py", line 146, in differentiable_reduce_scatter_sum
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 160, in forward
+    dist.reduce_scatter_tensor(sharded_tensor, tensor, group=group, op=dist.ReduceOp.SUM)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/distributed.py", line 43, in reduce_scatter_tensor
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 725, in _core_forward
+    return DifferentiableReduceScatterSum.apply(tensor, group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return dist.reduce_scatter_tensor(output=output, input=input, group=group, op=op, async_op=async_op)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/c10d_logger.py", line 52, in wrapper
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 725, in _core_forward
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/distributed_differentiable_primitives.py", line 119, in forward
+    output = self.attn(hidden_states=hidden_states, sequence_mask=sequence_mask)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.attn(hidden_states=hidden_states, sequence_mask=sequence_mask)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    "args": f"{args}, {kwargs}",
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/_tensor.py", line 431, in __repr__
+    dist.reduce_scatter_tensor(sharded_tensor, tensor, group=group, op=dist.ReduceOp.SUM)
+    return self._call_impl(*args, **kwargs)  File "/fsx/nouamane/projects/nanotron/src/nanotron/distributed.py", line 43, in reduce_scatter_tensor
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return torch._tensor_str._str(self, tensor_contents=tensor_contents)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/_tensor_str.py", line 664, in _str
+    return forward_call(*args, **kwargs)    
+return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 690, in forward
+    return dist.reduce_scatter_tensor(output=output, input=input, group=group, op=op, async_op=async_op)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/c10d_logger.py", line 52, in wrapper
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 690, in forward
+    output = self.o_proj(attention_output)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return _str_intern(self, tensor_contents=tensor_contents)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/_tensor_str.py", line 595, in _str_intern
+    "args": f"{args}, {kwargs}",
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/_tensor.py", line 431, in __repr__
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 725, in _core_forward
+    output = self.o_proj(attention_output)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = self.attn(hidden_states=hidden_states, sequence_mask=sequence_mask)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    tensor_str = _tensor_str(self, indent)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/_tensor_str.py", line 329, in _tensor_str
+    return torch._tensor_str._str(self, tensor_contents=tensor_contents)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/_tensor_str.py", line 664, in _str
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    self = self.float()
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 256.00 MiB. GPU 2 has a total capacty of 79.33 GiB of which 101.94 MiB is free. Including non-PyTorch memory, this process has 79.22 GiB memory in use. Of the allocated memory 69.78 GiB is allocated by PyTorch, and 464.25 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    return _str_intern(self, tensor_contents=tensor_contents)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/_tensor_str.py", line 595, in _str_intern
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 690, in forward
+        return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)tensor_str = _tensor_str(self, indent)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/_tensor_str.py", line 329, in _tensor_str
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 469, in forward
+    output = self.o_proj(attention_output)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    self = self.float()
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 256.00 MiB. GPU 5 has a total capacty of 79.33 GiB of which 101.94 MiB is free. Including non-PyTorch memory, this process has 79.22 GiB memory in use. Of the allocated memory 69.78 GiB is allocated by PyTorch, and 464.25 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 469, in forward
+    out = differentiable_reduce_scatter_sum(out, group=group)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/distributed_differentiable_primitives.py", line 146, in differentiable_reduce_scatter_sum
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    out = differentiable_reduce_scatter_sum(out, group=group)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/distributed_differentiable_primitives.py", line 146, in differentiable_reduce_scatter_sum
+    return DifferentiableReduceScatterSum.apply(tensor, group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    return DifferentiableReduceScatterSum.apply(tensor, group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/distributed_differentiable_primitives.py", line 119, in forward
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/distributed_differentiable_primitives.py", line 119, in forward
+    dist.reduce_scatter_tensor(sharded_tensor, tensor, group=group, op=dist.ReduceOp.SUM)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/distributed.py", line 43, in reduce_scatter_tensor
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    dist.reduce_scatter_tensor(sharded_tensor, tensor, group=group, op=dist.ReduceOp.SUM)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/distributed.py", line 43, in reduce_scatter_tensor
+    return dist.reduce_scatter_tensor(output=output, input=input, group=group, op=op, async_op=async_op)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/c10d_logger.py", line 52, in wrapper
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 469, in forward
+    return dist.reduce_scatter_tensor(output=output, input=input, group=group, op=op, async_op=async_op)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/c10d_logger.py", line 52, in wrapper
+    "args": f"{args}, {kwargs}",
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/_tensor.py", line 431, in __repr__
+    out = differentiable_reduce_scatter_sum(out, group=group)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/distributed_differentiable_primitives.py", line 146, in differentiable_reduce_scatter_sum
+    "args": f"{args}, {kwargs}",
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/_tensor.py", line 431, in __repr__
+    return torch._tensor_str._str(self, tensor_contents=tensor_contents)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/_tensor_str.py", line 664, in _str
+    return DifferentiableReduceScatterSum.apply(tensor, group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return torch._tensor_str._str(self, tensor_contents=tensor_contents)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/_tensor_str.py", line 664, in _str
+    return _str_intern(self, tensor_contents=tensor_contents)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/_tensor_str.py", line 595, in _str_intern
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/distributed_differentiable_primitives.py", line 119, in forward
+    return _str_intern(self, tensor_contents=tensor_contents)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/_tensor_str.py", line 595, in _str_intern
+    tensor_str = _tensor_str(self, indent)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/_tensor_str.py", line 329, in _tensor_str
+    tensor_str = _tensor_str(self, indent)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/_tensor_str.py", line 329, in _tensor_str
+    self = self.float()
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 256.00 MiB. GPU 1 has a total capacty of 79.33 GiB of which 101.94 MiB is free. Including non-PyTorch memory, this process has 79.22 GiB memory in use. Of the allocated memory 69.78 GiB is allocated by PyTorch, and 464.25 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    dist.reduce_scatter_tensor(sharded_tensor, tensor, group=group, op=dist.ReduceOp.SUM)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/distributed.py", line 43, in reduce_scatter_tensor
+    self = self.float()
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 256.00 MiB. GPU 4 has a total capacty of 79.33 GiB of which 85.94 MiB is free. Including non-PyTorch memory, this process has 79.23 GiB memory in use. Of the allocated memory 69.78 GiB is allocated by PyTorch, and 464.25 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return dist.reduce_scatter_tensor(output=output, input=input, group=group, op=op, async_op=async_op)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/c10d_logger.py", line 52, in wrapper
+    "args": f"{args}, {kwargs}",
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/_tensor.py", line 431, in __repr__
+    return torch._tensor_str._str(self, tensor_contents=tensor_contents)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/_tensor_str.py", line 664, in _str
+    return _str_intern(self, tensor_contents=tensor_contents)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/_tensor_str.py", line 595, in _str_intern
+    tensor_str = _tensor_str(self, indent)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/_tensor_str.py", line 329, in _tensor_str
+    self = self.float()
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 256.00 MiB. GPU 6 has a total capacty of 79.33 GiB of which 101.94 MiB is free. Including non-PyTorch memory, this process has 79.22 GiB memory in use. Of the allocated memory 69.78 GiB is allocated by PyTorch, and 464.25 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/c10d_logger.py", line 47, in wrapper
+    return func(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/distributed_c10d.py", line 3375, in reduce_scatter_tensor
+    work = group._reduce_scatter_base(output, input, opts)
+RuntimeError: Rank 7 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+ Original exception: 
+[../third_party/gloo/gloo/transport/tcp/pair.cc:534] Connection closed by peer [26.0.164.187]:11425
+
+During handling of the above exception, another exception occurred:
+
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 160, in forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 731, in _core_forward
+    hidden_states = self.mlp(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 245, in forward
+    hidden_states = self.down_proj(self.split_silu_mul(merged_states))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 469, in forward
+    out = differentiable_reduce_scatter_sum(out, group=group)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/distributed_differentiable_primitives.py", line 146, in differentiable_reduce_scatter_sum
+    return DifferentiableReduceScatterSum.apply(tensor, group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/distributed_differentiable_primitives.py", line 119, in forward
+    dist.reduce_scatter_tensor(sharded_tensor, tensor, group=group, op=dist.ReduceOp.SUM)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/distributed.py", line 43, in reduce_scatter_tensor
+    return dist.reduce_scatter_tensor(output=output, input=input, group=group, op=op, async_op=async_op)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/c10d_logger.py", line 52, in wrapper
+    "args": f"{args}, {kwargs}",
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/_tensor.py", line 431, in __repr__
+    return torch._tensor_str._str(self, tensor_contents=tensor_contents)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/_tensor_str.py", line 664, in _str
+    return _str_intern(self, tensor_contents=tensor_contents)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/_tensor_str.py", line 595, in _str_intern
+    tensor_str = _tensor_str(self, indent)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/_tensor_str.py", line 329, in _tensor_str
+    self = self.float()
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 256.00 MiB. GPU 7 has a total capacty of 79.33 GiB of which 145.94 MiB is free. Including non-PyTorch memory, this process has 79.18 GiB memory in use. Of the allocated memory 69.27 GiB is allocated by PyTorch, and 508.76 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+[E ProcessGroupGloo.cpp:138] Rank 4 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+[E ProcessGroupGloo.cpp:138] Rank 7 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+[E ProcessGroupGloo.cpp:138] Rank 5 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+[E ProcessGroupGloo.cpp:138] Rank 6 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+[E ProcessGroupGloo.cpp:138] Rank 1 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+[E ProcessGroupGloo.cpp:138] Rank 2 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/c10d_logger.py", line 47, in wrapper
+    return func(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/distributed_c10d.py", line 3375, in reduce_scatter_tensor
+    work = group._reduce_scatter_base(output, input, opts)
+RuntimeError: Rank 7 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+ Original exception: 
+[../third_party/gloo/gloo/transport/tcp/pair.cc:534] Connection closed by peer [26.0.164.18]:15041
+
+During handling of the above exception, another exception occurred:
+
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/c10d_logger.py", line 47, in wrapper
+    return func(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/distributed_c10d.py", line 3375, in reduce_scatter_tensor
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    work = group._reduce_scatter_base(output, input, opts)
+RuntimeError: Rank 6 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+ Original exception: 
+[../third_party/gloo/gloo/transport/tcp/pair.cc:534] Connection closed by peer [26.0.164.18]:53658
+
+During handling of the above exception, another exception occurred:
+
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/c10d_logger.py", line 47, in wrapper
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+    return func(*args, **kwargs)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/distributed_c10d.py", line 3375, in reduce_scatter_tensor
+return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 160, in forward
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/c10d_logger.py", line 47, in wrapper
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    return func(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/distributed_c10d.py", line 3375, in reduce_scatter_tensor
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    work = group._reduce_scatter_base(output, input, opts)
+RuntimeError: Rank 4 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+ Original exception: 
+[../third_party/gloo/gloo/transport/tcp/pair.cc:534] Connection closed by peer [26.0.164.18]:5793
+
+During handling of the above exception, another exception occurred:
+
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    work = group._reduce_scatter_base(output, input, opts)
+    return self._call_impl(*args, **kwargs)
+RuntimeError  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+: Rank 2 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+ Original exception: 
+[../third_party/gloo/gloo/transport/tcp/pair.cc:534] Connection closed by peer [26.0.164.18]:50910
+
+During handling of the above exception, another exception occurred:
+
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 725, in _core_forward
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    output = self.attn(hidden_states=hidden_states, sequence_mask=sequence_mask)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/c10d_logger.py", line 47, in wrapper
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+    return forward_call(*args, **kwargs)  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 690, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.o_proj(attention_output)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    return func(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/distributed_c10d.py", line 3375, in reduce_scatter_tensor
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    work = group._reduce_scatter_base(output, input, opts)
+RuntimeError: Rank 5 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+ Original exception: 
+[../third_party/gloo/gloo/transport/tcp/pair.cc:534] Connection closed by peer [26.0.164.18]:40617
+
+During handling of the above exception, another exception occurred:
+
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 160, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return self._call_impl(*args, **kwargs)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 160, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 469, in forward
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 725, in _core_forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 160, in forward
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/c10d_logger.py", line 47, in wrapper
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+    out = differentiable_reduce_scatter_sum(out, group=group)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/distributed_differentiable_primitives.py", line 146, in differentiable_reduce_scatter_sum
+        return self._call_impl(*args, **kwargs)
+output = self.attn(hidden_states=hidden_states, sequence_mask=sequence_mask)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return func(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/distributed_c10d.py", line 3375, in reduce_scatter_tensor
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    return DifferentiableReduceScatterSum.apply(tensor, group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return forward_call(*args, **kwargs)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+        return forward_call(*args, **kwargs)return super().apply(*args, **kwargs)  # type: ignore[misc]
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/distributed_differentiable_primitives.py", line 119, in forward
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 725, in _core_forward
+    work = group._reduce_scatter_base(output, input, opts)
+RuntimeError: Rank 1 successfully reached monitoredBarrier, but received errors while waiting for send/recv from rank 0. Please check rank 0 logs for faulty rank.
+ Original exception: 
+[../third_party/gloo/gloo/transport/tcp/pair.cc:534] Connection closed by peer [26.0.164.18]:19646
+
+During handling of the above exception, another exception occurred:
+
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 690, in forward
+    output = self.o_proj(attention_output)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 725, in _core_forward
+    dist.reduce_scatter_tensor(sharded_tensor, tensor, group=group, op=dist.ReduceOp.SUM)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/distributed.py", line 43, in reduce_scatter_tensor
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    trainer.train(dataloader)    
+output = self.attn(hidden_states=hidden_states, sequence_mask=sequence_mask)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = self.attn(hidden_states=hidden_states, sequence_mask=sequence_mask)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return dist.reduce_scatter_tensor(output=output, input=input, group=group, op=op, async_op=async_op)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/c10d_logger.py", line 52, in wrapper
+        return self._call_impl(*args, **kwargs)
+return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+        return self._call_impl(*args, **kwargs)return forward_call(*args, **kwargs)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    "args": f"{args}, {kwargs}",
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/_tensor.py", line 431, in __repr__
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 690, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    output = self.o_proj(attention_output)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+        return forward_call(*args, **kwargs)return torch._tensor_str._str(self, tensor_contents=tensor_contents)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 690, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/_tensor_str.py", line 664, in _str
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    output = self.o_proj(attention_output)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return _str_intern(self, tensor_contents=tensor_contents)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/_tensor_str.py", line 595, in _str_intern
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+        return forward_call(*args, **kwargs)return forward_call(*args, **kwargs)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 469, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    tensor_str = _tensor_str(self, indent)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/_tensor_str.py", line 329, in _tensor_str
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+    return row_linear(  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    out = differentiable_reduce_scatter_sum(out, group=group)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/distributed_differentiable_primitives.py", line 146, in differentiable_reduce_scatter_sum
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    self = self.float()
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 256.00 MiB. GPU 7 has a total capacty of 79.33 GiB of which 225.94 MiB is free. Including non-PyTorch memory, this process has 79.10 GiB memory in use. Of the allocated memory 69.06 GiB is allocated by PyTorch, and 556.76 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return DifferentiableReduceScatterSum.apply(tensor, group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+        hidden_encoder_states = encoder_block(**hidden_encoder_states)return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)    
+
+return row_linear(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/distributed_differentiable_primitives.py", line 119, in forward
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 469, in forward
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 160, in forward
+    dist.reduce_scatter_tensor(sharded_tensor, tensor, group=group, op=dist.ReduceOp.SUM)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/distributed.py", line 43, in reduce_scatter_tensor
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    out = differentiable_reduce_scatter_sum(out, group=group)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/distributed_differentiable_primitives.py", line 146, in differentiable_reduce_scatter_sum
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 469, in forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return dist.reduce_scatter_tensor(output=output, input=input, group=group, op=op, async_op=async_op)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/c10d_logger.py", line 52, in wrapper
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)    
+return DifferentiableReduceScatterSum.apply(tensor, group)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    out = differentiable_reduce_scatter_sum(out, group=group)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/distributed_differentiable_primitives.py", line 146, in differentiable_reduce_scatter_sum
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+    "args": f"{args}, {kwargs}",
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/_tensor.py", line 431, in __repr__
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/distributed_differentiable_primitives.py", line 119, in forward
+    return DifferentiableReduceScatterSum.apply(tensor, group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 725, in _core_forward
+    hidden_encoder_states = encoder_block(**hidden_encoder_states)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+        output = self.attn(hidden_states=hidden_states, sequence_mask=sequence_mask)return torch._tensor_str._str(self, tensor_contents=tensor_contents)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/_tensor_str.py", line 664, in _str
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    dist.reduce_scatter_tensor(sharded_tensor, tensor, group=group, op=dist.ReduceOp.SUM)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/distributed.py", line 43, in reduce_scatter_tensor
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/distributed_differentiable_primitives.py", line 119, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 690, in forward
+        return dist.reduce_scatter_tensor(output=output, input=input, group=group, op=op, async_op=async_op)return _str_intern(self, tensor_contents=tensor_contents)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/c10d_logger.py", line 52, in wrapper
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/_tensor_str.py", line 595, in _str_intern
+    dist.reduce_scatter_tensor(sharded_tensor, tensor, group=group, op=dist.ReduceOp.SUM)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/distributed.py", line 43, in reduce_scatter_tensor
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 160, in forward
+    output = self.o_proj(attention_output)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    "args": f"{args}, {kwargs}",
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/_tensor.py", line 431, in __repr__
+return dist.reduce_scatter_tensor(output=output, input=input, group=group, op=op, async_op=async_op)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/c10d_logger.py", line 52, in wrapper
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    tensor_str = _tensor_str(self, indent)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/_tensor_str.py", line 329, in _tensor_str
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    "args": f"{args}, {kwargs}",
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/_tensor.py", line 431, in __repr__
+    return torch._tensor_str._str(self, tensor_contents=tensor_contents)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/_tensor_str.py", line 664, in _str
+self = self.float()
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 256.00 MiB. GPU 6 has a total capacty of 79.33 GiB of which 105.94 MiB is free. Including non-PyTorch memory, this process has 79.21 GiB memory in use. Of the allocated memory 69.06 GiB is allocated by PyTorch, and 556.76 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    return torch._tensor_str._str(self, tensor_contents=tensor_contents)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/_tensor_str.py", line 664, in _str
+    return _str_intern(self, tensor_contents=tensor_contents)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/_tensor_str.py", line 595, in _str_intern
+    hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 725, in _core_forward
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return _str_intern(self, tensor_contents=tensor_contents)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/_tensor_str.py", line 595, in _str_intern
+    tensor_str = _tensor_str(self, indent)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/_tensor_str.py", line 329, in _tensor_str
+    output = self.attn(hidden_states=hidden_states, sequence_mask=sequence_mask)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 469, in forward
+    tensor_str = _tensor_str(self, indent)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/_tensor_str.py", line 329, in _tensor_str
+    self = self.float()
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 256.00 MiB. GPU 2 has a total capacty of 79.33 GiB of which 105.94 MiB is free. Including non-PyTorch memory, this process has 79.21 GiB memory in use. Of the allocated memory 69.06 GiB is allocated by PyTorch, and 556.76 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    self = self.float()
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 256.00 MiB. GPU 4 has a total capacty of 79.33 GiB of which 105.94 MiB is free. Including non-PyTorch memory, this process has 79.21 GiB memory in use. Of the allocated memory 69.06 GiB is allocated by PyTorch, and 540.76 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    out = differentiable_reduce_scatter_sum(out, group=group)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/distributed_differentiable_primitives.py", line 146, in differentiable_reduce_scatter_sum
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return DifferentiableReduceScatterSum.apply(tensor, group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 690, in forward
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/distributed_differentiable_primitives.py", line 119, in forward
+    output = self.o_proj(attention_output)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    dist.reduce_scatter_tensor(sharded_tensor, tensor, group=group, op=dist.ReduceOp.SUM)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/distributed.py", line 43, in reduce_scatter_tensor
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return dist.reduce_scatter_tensor(output=output, input=input, group=group, op=op, async_op=async_op)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/c10d_logger.py", line 52, in wrapper
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+    "args": f"{args}, {kwargs}",
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/_tensor.py", line 431, in __repr__
+    return row_linear(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+    return torch._tensor_str._str(self, tensor_contents=tensor_contents)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/_tensor_str.py", line 664, in _str
+    return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return _str_intern(self, tensor_contents=tensor_contents)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/_tensor_str.py", line 595, in _str_intern
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 469, in forward
+    tensor_str = _tensor_str(self, indent)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/_tensor_str.py", line 329, in _tensor_str
+    out = differentiable_reduce_scatter_sum(out, group=group)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/distributed_differentiable_primitives.py", line 146, in differentiable_reduce_scatter_sum
+    self = self.float()
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 256.00 MiB. GPU 5 has a total capacty of 79.33 GiB of which 41.94 MiB is free. Including non-PyTorch memory, this process has 79.28 GiB memory in use. Of the allocated memory 69.06 GiB is allocated by PyTorch, and 540.76 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    return DifferentiableReduceScatterSum.apply(tensor, group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/distributed_differentiable_primitives.py", line 119, in forward
+    dist.reduce_scatter_tensor(sharded_tensor, tensor, group=group, op=dist.ReduceOp.SUM)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/distributed.py", line 43, in reduce_scatter_tensor
+    return dist.reduce_scatter_tensor(output=output, input=input, group=group, op=op, async_op=async_op)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/c10d_logger.py", line 52, in wrapper
+    "args": f"{args}, {kwargs}",
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/_tensor.py", line 431, in __repr__
+    return torch._tensor_str._str(self, tensor_contents=tensor_contents)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/_tensor_str.py", line 664, in _str
+    return _str_intern(self, tensor_contents=tensor_contents)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/_tensor_str.py", line 595, in _str_intern
+    tensor_str = _tensor_str(self, indent)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/_tensor_str.py", line 329, in _tensor_str
+    self = self.float()
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 256.00 MiB. GPU 1 has a total capacty of 79.33 GiB of which 25.94 MiB is free. Including non-PyTorch memory, this process has 79.29 GiB memory in use. Of the allocated memory 69.06 GiB is allocated by PyTorch, and 556.76 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+NCCL version 2.18.5+cuda12.2
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+NCCL version 2.18.5+cuda12.2
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+[E TraceUtils.h:35] Store is down while updating #88 with key NCCL_4_trace_start
+[E TraceUtils.h:35] Store is down while updating #88 with key NCCL_1_trace_start
+[E TraceUtils.h:35] Store is down while updating #153 with key NCCL_4_trace_start
+[E TraceUtils.h:35] Store is down while updating #217 with key NCCL_2_trace_end
+[E TraceUtils.h:35] Store is down while updating #153 with key NCCL_7_trace_start
+[E TraceUtils.h:35] Store is down while updating #154 with key NCCL_5_trace_start
+[E TraceUtils.h:35] Store is down while updating #90 with key NCCL_7_trace_start
+[E TraceUtils.h:35] Store is down while updating #88 with key NCCL_0_trace_end
+[E TraceUtils.h:35] Store is down while updating #91 with key NCCL_6_trace_end
+[E TraceUtils.h:35] Store is down while updating #152 with key NCCL_1_trace_end
+[E TraceUtils.h:35] Store is down while updating #152 with key NCCL_6_trace_end
+[E TraceUtils.h:35] Store is down while updating #88 with key NCCL_5_trace_end
+[E TraceUtils.h:35] Store is down while updating #216 with key NCCL_1_trace_end
+[E TraceUtils.h:35] Store is down while updating #88 with key NCCL_2_trace_start
+[E TraceUtils.h:35] Store is down while updating #217 with key NCCL_6_trace_end
+[E TraceUtils.h:35] Store is down while updating #217 with key NCCL_2_trace_start
+[E TraceUtils.h:35] Store is down while updating #153 with key NCCL_3_trace_start
+[E TraceUtils.h:35] Store is down while updating #217 with key NCCL_6_trace_end
+[E TraceUtils.h:35] Store is down while updating #216 with key NCCL_7_trace_end
+[E TraceUtils.h:35] Store is down while updating #217 with key NCCL_0_trace_start
+[E TraceUtils.h:35] Store is down while updating #88 with key NCCL_3_trace_end
+[E TraceUtils.h:35] Store is down while updating #217 with key NCCL_0_trace_start
+[E TraceUtils.h:35] Store is down while updating #218 with key NCCL_7_trace_end
+[E TraceUtils.h:35] Store is down while updating #216 with key NCCL_5_trace_end
+[E TraceUtils.h:35] Store is down while updating #217 with key NCCL_6_trace_end
+[E TraceUtils.h:35] Store is down while updating #217 with key NCCL_5_trace_start
+[E TraceUtils.h:35] Store is down while updating #217 with key NCCL_4_trace_start
+[E TraceUtils.h:35] Store is down while updating #217 with key NCCL_6_trace_start
+[E TraceUtils.h:35] Store is down while updating #216 with key NCCL_4_trace_end
+[E TraceUtils.h:35] Store is down while updating #153 with key NCCL_1_trace_start
+[E TraceUtils.h:35] Store is down while updating #153 with key NCCL_4_trace_start
+[E TraceUtils.h:35] Store is down while updating #153 with key NCCL_2_trace_start
+[E TraceUtils.h:35] Store is down while updating #217 with key NCCL_2_trace_start
+[E TraceUtils.h:35] Store is down while updating #153 with key NCCL_5_trace_end
+[E TraceUtils.h:35] Store is down while updating #153 with key NCCL_1_trace_end
+[E TraceUtils.h:35] Store is down while updating #152 with key NCCL_2_trace_end
+[E TraceUtils.h:35] Store is down while updating #153 with key NCCL_3_trace_start
+[E TraceUtils.h:35] Store is down while updating #152 with key NCCL_0_trace_end
+[E TraceUtils.h:35] Store is down while updating #217 with key NCCL_0_trace_end
+[E TraceUtils.h:35] Store is down while updating #218 with key NCCL_3_trace_start
+[E TraceUtils.h:35] Store is down while updating #217 with key NCCL_4_trace_end
+[E TraceUtils.h:35] Store is down while updating #216 with key NCCL_5_trace_start
+[E TraceUtils.h:35] Store is down while updating #217 with key NCCL_7_trace_start
+[E TraceUtils.h:35] Store is down while updating #217 with key NCCL_3_trace_end
+[E TraceUtils.h:35] Store is down while updating #217 with key NCCL_1_trace_end
+[E TraceUtils.h:35] Store is down while updating #153 with key NCCL_6_trace_start
+[E TraceUtils.h:35] Store is down while updating #153 with key NCCL_7_trace_end
+[E TraceUtils.h:35] Store is down while updating #217 with key NCCL_1_trace_start
+[E TraceUtils.h:35] Store is down while updating #216 with key NCCL_3_trace_start
+[E TraceUtils.h:35] Store is down while updating #217 with key NCCL_2_trace_start
+[E TraceUtils.h:35] Store is down while updating #152 with key NCCL_2_trace_end
+[E TraceUtils.h:35] Store is down while updating #153 with key NCCL_0_trace_start
+[E TraceUtils.h:35] Store is down while updating #217 with key NCCL_5_trace_end
+[E TraceUtils.h:35] Store is down while updating #217 with key NCCL_0_trace_end
+[E TraceUtils.h:35] Store is down while updating #217 with key NCCL_4_trace_end
+[E TraceUtils.h:35] Store is down while updating #153 with key NCCL_7_trace_end
+[E TraceUtils.h:35] Store is down while updating #216 with key NCCL_7_trace_end
+[E TraceUtils.h:35] Store is down while updating #153 with key NCCL_5_trace_end
+[E TraceUtils.h:35] Store is down while updating #216 with key NCCL_1_trace_end
+[E TraceUtils.h:35] Store is down while updating #153 with key NCCL_4_trace_start
+[E TraceUtils.h:35] Store is down while updating #217 with key NCCL_3_trace_start
+[E TraceUtils.h:35] Store is down while updating #87 with key NCCL_1_trace_start
+[E TraceUtils.h:35] Store is down while updating #153 with key NCCL_3_trace_end
+[E TraceUtils.h:35] Store is down while updating #87 with key NCCL_0_trace_start
+[E TraceUtils.h:35] Store is down while updating #153 with key NCCL_6_trace_start
+[E TraceUtils.h:35] Store is down while updating #88 with key NCCL_2_trace_end
+[E TraceUtils.h:35] Store is down while updating #87 with key NCCL_5_trace_start
+[E TraceUtils.h:35] Store is down while updating #87 with key NCCL_4_trace_start
+[E TraceUtils.h:35] Store is down while updating #87 with key NCCL_3_trace_start
+[E TraceUtils.h:35] Store is down while updating #87 with key NCCL_7_trace_start
+[E TraceUtils.h:35] Store is down while updating #88 with key NCCL_5_trace_end
+[E TraceUtils.h:35] Store is down while updating #89 with key NCCL_6_trace_start
+[E TraceUtils.h:35] Store is down while updating #87 with key NCCL_6_trace_start
+[E TraceUtils.h:35] Store is down while updating #88 with key NCCL_4_trace_end
+[E TraceUtils.h:35] Store is down while updating #89 with key NCCL_0_trace_start
+[E TraceUtils.h:35] Store is down while updating #87 with key NCCL_2_trace_start
+[E TraceUtils.h:35] Store is down while updating #153 with key NCCL_0_trace_start
+[E TraceUtils.h:35] Store is down while updating #87 with key NCCL_2_trace_start
+[E TraceUtils.h:35] Store is down while updating #89 with key NCCL_7_trace_start
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[E TraceUtils.h:35] Store is down while updating #87 with key NCCL_4_trace_start
+01/03/2025 01:31:17 [ERROR|DP=0|PP=4|TP=7|ip-26-0-166-15]: [SEND META] Rank 4: Failed to send metadata to rank 5: Broken pipe
+01/03/2025 01:31:17 [ERROR|DP=0|PP=4|TP=6|ip-26-0-166-15]: [SEND META] Rank 4: Failed to send metadata to rank 5: Broken pipe
+01/03/2025 01:31:17 [ERROR|DP=0|PP=4|TP=3|ip-26-0-166-15]: [SEND META] Rank 4: Failed to send metadata to rank 5: Broken pipe
+[E TraceUtils.h:35] Store is down while updating #87 with key NCCL_3_trace_start
+01/03/2025 01:31:17 [ERROR|DP=0|PP=4|TP=0|ip-26-0-166-15]: [SEND META] Rank 4: Failed to send metadata to rank 5: Broken pipe
+01/03/2025 01:31:17 [ERROR|DP=0|PP=4|TP=1|ip-26-0-166-15]: [SEND META] Rank 4: Failed to send metadata to rank 5: Broken pipe
+01/03/2025 01:31:17 [ERROR|DP=0|PP=4|TP=4|ip-26-0-166-15]: [SEND META] Rank 4: Failed to send metadata to rank 5: Broken pipe
+01/03/2025 01:31:17 [ERROR|DP=0|PP=4|TP=2|ip-26-0-166-15]: [SEND META] Rank 4: Failed to send metadata to rank 5: Broken pipe
+01/03/2025 01:31:17 [ERROR|DP=0|PP=4|TP=5|ip-26-0-166-15]: [SEND META] Rank 4: Failed to send metadata to rank 5: Broken pipe
+[E TraceUtils.h:35] Store is down while updating #87 with key NCCL_0_trace_start
+[E TraceUtils.h:35] Store is down while updating #88 with key NCCL_1_trace_start
+[E TraceUtils.h:35] Store is down while updating #87 with key NCCL_6_trace_start
+[E TraceUtils.h:35] Store is down while updating #87 with key NCCL_7_trace_start
+[E TraceUtils.h:35] Store is down while updating #87 with key NCCL_1_trace_start
+[E TraceUtils.h:35] Store is down while updating #152 with key NCCL_1_trace_end
+[E TraceUtils.h:35] Store is down while updating #153 with key NCCL_0_trace_end
+[E TraceUtils.h:35] Store is down while updating #153 with key NCCL_6_trace_start
+[E TraceUtils.h:35] Store is down while updating #88 with key NCCL_3_trace_start
+[E TraceUtils.h:35] Store is down while updating #152 with key NCCL_5_trace_end
+[E TraceUtils.h:35] Store is down while updating #152 with key NCCL_7_trace_end
+[E TraceUtils.h:35] Store is down while updating #152 with key NCCL_4_trace_start
+[E TraceUtils.h:35] Store is down while updating #153 with key NCCL_3_trace_start
+[E TraceUtils.h:35] Store is down while updating #153 with key NCCL_2_trace_start
+[E TraceUtils.h:35] Store is down while updating #87 with key NCCL_5_trace_start
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/p2p.py", line 279, in _send_meta
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/p2p.py", line 279, in _send_meta
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/p2p.py", line 279, in _send_meta
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/p2p.py", line 279, in _send_meta
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/p2p.py", line 279, in _send_meta
+Traceback (most recent call last):
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/p2p.py", line 279, in _send_meta
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/p2p.py", line 279, in _send_meta
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/p2p.py", line 279, in _send_meta
+    work = dist.isend(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/distributed_c10d.py", line 1528, in isend
+    work = dist.isend(
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/distributed_c10d.py", line 1528, in isend
+work = dist.isend(    
+work = dist.isend(    
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/distributed_c10d.py", line 1528, in isend
+work = dist.isend(    
+work = dist.isend(  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/distributed_c10d.py", line 1528, in isend
+    
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/distributed_c10d.py", line 1528, in isend
+work = dist.isend(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/distributed_c10d.py", line 1528, in isend
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/distributed_c10d.py", line 1528, in isend
+    return group.send([tensor], group_dst_rank, tag)
+RuntimeError: Broken pipe
+
+During handling of the above exception, another exception occurred:
+
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    work = dist.isend(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/distributed_c10d.py", line 1528, in isend
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return group.send([tensor], group_dst_rank, tag)
+RuntimeError: Broken pipe
+
+During handling of the above exception, another exception occurred:
+
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return group.send([tensor], group_dst_rank, tag)
+    return group.send([tensor], group_dst_rank, tag)
+RuntimeError: Broken pipe
+    RuntimeError
+During handling of the above exception, another exception occurred:
+
+return group.send([tensor], group_dst_rank, tag): Traceback (most recent call last):
+    Broken pipereturn group.send([tensor], group_dst_rank, tag)
+    
+return group.send([tensor], group_dst_rank, tag)
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+
+RuntimeError
+During handling of the above exception, another exception occurred:
+
+: RuntimeErrorRuntimeErrorTraceback (most recent call last):
+Broken pipe: :   File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    
+Broken pipeBroken pipereturn f(*args, **kwargs)
+During handling of the above exception, another exception occurred:
+
+
+
+
+Traceback (most recent call last):
+
+During handling of the above exception, another exception occurred:
+
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+
+During handling of the above exception, another exception occurred:
+
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+      File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return group.send([tensor], group_dst_rank, tag)
+RuntimeError: Broken pipe
+
+During handling of the above exception, another exception occurred:
+
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 263, in train_batch_iter
+    return f(*args, **kwargs)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+return f(*args, **kwargs)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+        outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)return f(*args, **kwargs)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    send_activation()
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/state.py", line 23, in __call__
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    self.p2p.send_tensors([self.activation], to_rank=self.to_rank)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/p2p.py", line 498, in send_tensors
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)        
+futures = self.isend_tensors(tensors=tensors, to_rank=to_rank, tag=tag)outputs = self.pipeline_engine.train_batch_iter(
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/p2p.py", line 441, in isend_tensors
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 263, in train_batch_iter
+    outputs = self.pipeline_engine.train_batch_iter(    
+outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 263, in train_batch_iter
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 263, in train_batch_iter
+        outputs = self.pipeline_engine.train_batch_iter(outputs = self.pipeline_engine.train_batch_iter(
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 263, in train_batch_iter
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 263, in train_batch_iter
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 263, in train_batch_iter
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 263, in train_batch_iter
+    self._send_meta(tensor, to_rank=to_rank, tag=tag)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/p2p.py", line 339, in _send_meta
+    send_activation()
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/state.py", line 23, in __call__
+    raise P2PCommunicationError(f"Failed to send metadata to rank {to_rank}: {str(e)}")
+nanotron.parallel.pipeline_parallel.p2p.P2PCommunicationError: Failed to send metadata to rank 5: Broken pipe
+    self.p2p.send_tensors([self.activation], to_rank=self.to_rank)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/p2p.py", line 498, in send_tensors
+    send_activation()    
+send_activation()
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/state.py", line 23, in __call__
+    send_activation()  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/state.py", line 23, in __call__
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/state.py", line 23, in __call__
+    send_activation()
+    send_activation()  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/state.py", line 23, in __call__
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/state.py", line 23, in __call__
+        send_activation()futures = self.isend_tensors(tensors=tensors, to_rank=to_rank, tag=tag)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/p2p.py", line 441, in isend_tensors
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/state.py", line 23, in __call__
+    self.p2p.send_tensors([self.activation], to_rank=self.to_rank)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/p2p.py", line 498, in send_tensors
+    self._send_meta(tensor, to_rank=to_rank, tag=tag)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/p2p.py", line 339, in _send_meta
+    self.p2p.send_tensors([self.activation], to_rank=self.to_rank)    
+self.p2p.send_tensors([self.activation], to_rank=self.to_rank)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/p2p.py", line 498, in send_tensors
+    self.p2p.send_tensors([self.activation], to_rank=self.to_rank)  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/p2p.py", line 498, in send_tensors
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/p2p.py", line 498, in send_tensors
+    self.p2p.send_tensors([self.activation], to_rank=self.to_rank)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/p2p.py", line 498, in send_tensors
+    raise P2PCommunicationError(f"Failed to send metadata to rank {to_rank}: {str(e)}")
+nanotron.parallel.pipeline_parallel.p2p.P2PCommunicationError: Failed to send metadata to rank 5: Broken pipe
+    futures = self.isend_tensors(tensors=tensors, to_rank=to_rank, tag=tag)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/p2p.py", line 441, in isend_tensors
+        futures = self.isend_tensors(tensors=tensors, to_rank=to_rank, tag=tag)futures = self.isend_tensors(tensors=tensors, to_rank=to_rank, tag=tag)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/p2p.py", line 441, in isend_tensors
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/p2p.py", line 441, in isend_tensors
+    futures = self.isend_tensors(tensors=tensors, to_rank=to_rank, tag=tag)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/p2p.py", line 441, in isend_tensors
+    futures = self.isend_tensors(tensors=tensors, to_rank=to_rank, tag=tag)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/p2p.py", line 441, in isend_tensors
+    self._send_meta(tensor, to_rank=to_rank, tag=tag)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/p2p.py", line 339, in _send_meta
+    self.p2p.send_tensors([self.activation], to_rank=self.to_rank)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/p2p.py", line 498, in send_tensors
+self._send_meta(tensor, to_rank=to_rank, tag=tag)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/p2p.py", line 339, in _send_meta
+    self._send_meta(tensor, to_rank=to_rank, tag=tag)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/p2p.py", line 339, in _send_meta
+        self._send_meta(tensor, to_rank=to_rank, tag=tag)self._send_meta(tensor, to_rank=to_rank, tag=tag)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/p2p.py", line 339, in _send_meta
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/p2p.py", line 339, in _send_meta
+    raise P2PCommunicationError(f"Failed to send metadata to rank {to_rank}: {str(e)}")
+nanotron.parallel.pipeline_parallel.p2p.P2PCommunicationError: Failed to send metadata to rank 5: Broken pipe
+    raise P2PCommunicationError(f"Failed to send metadata to rank {to_rank}: {str(e)}")
+nanotron.parallel.pipeline_parallel.p2p.P2PCommunicationError: Failed to send metadata to rank 5: Broken pipe
+    raise P2PCommunicationError(f"Failed to send metadata to rank {to_rank}: {str(e)}")
+nanotron.parallel.pipeline_parallel.p2p.P2PCommunicationError: Failed to send metadata to rank 5: Broken pipe    
+futures = self.isend_tensors(tensors=tensors, to_rank=to_rank, tag=tag)    
+raise P2PCommunicationError(f"Failed to send metadata to rank {to_rank}: {str(e)}")  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/p2p.py", line 441, in isend_tensors
+
+nanotron.parallel.pipeline_parallel.p2p.P2PCommunicationError: Failed to send metadata to rank 5: Broken pipe
+    raise P2PCommunicationError(f"Failed to send metadata to rank {to_rank}: {str(e)}")
+nanotron.parallel.pipeline_parallel.p2p.P2PCommunicationError: Failed to send metadata to rank 5: Broken pipe
+    self._send_meta(tensor, to_rank=to_rank, tag=tag)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/p2p.py", line 339, in _send_meta
+    raise P2PCommunicationError(f"Failed to send metadata to rank {to_rank}: {str(e)}")
+nanotron.parallel.pipeline_parallel.p2p.P2PCommunicationError: Failed to send metadata to rank 5: Broken pipe
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+01/03/2025 01:31:18 [ERROR|DP=2|PP=4|TP=6|ip-26-0-166-244]: [SEND META] Rank 4: Failed to send metadata to rank 5: Broken pipe
+01/03/2025 01:31:18 [ERROR|DP=2|PP=4|TP=7|ip-26-0-166-244]: [SEND META] Rank 4: Failed to send metadata to rank 5: Broken pipe
+01/03/2025 01:31:18 [ERROR|DP=2|PP=4|TP=4|ip-26-0-166-244]: [SEND META] Rank 4: Failed to send metadata to rank 5: Broken pipe
+01/03/2025 01:31:18 [ERROR|DP=2|PP=4|TP=0|ip-26-0-166-244]: [SEND META] Rank 4: Failed to send metadata to rank 5: Broken pipe
+01/03/2025 01:31:18 [ERROR|DP=2|PP=4|TP=1|ip-26-0-166-244]: [SEND META] Rank 4: Failed to send metadata to rank 5: Broken pipe
+01/03/2025 01:31:18 [ERROR|DP=2|PP=4|TP=2|ip-26-0-166-244]: [SEND META] Rank 4: Failed to send metadata to rank 5: Broken pipe
+01/03/2025 01:31:18 [ERROR|DP=2|PP=4|TP=5|ip-26-0-166-244]: [SEND META] Rank 4: Failed to send metadata to rank 5: Broken pipe
+01/03/2025 01:31:18 [ERROR|DP=2|PP=4|TP=3|ip-26-0-166-244]: [SEND META] Rank 4: Failed to send metadata to rank 5: Broken pipe
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/p2p.py", line 279, in _send_meta
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/p2p.py", line 279, in _send_meta
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/p2p.py", line 279, in _send_meta
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/p2p.py", line 279, in _send_meta
+Traceback (most recent call last):
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/p2p.py", line 279, in _send_meta
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/p2p.py", line 279, in _send_meta
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/p2p.py", line 279, in _send_meta
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/p2p.py", line 279, in _send_meta
+    work = dist.isend(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/distributed_c10d.py", line 1528, in isend
+    work = dist.isend(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/distributed_c10d.py", line 1528, in isend
+    return group.send([tensor], group_dst_rank, tag)
+RuntimeError: Broken pipe
+
+During handling of the above exception, another exception occurred:
+
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return group.send([tensor], group_dst_rank, tag)
+RuntimeError: Broken pipe
+
+During handling of the above exception, another exception occurred:
+
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    work = dist.isend(
+    work = dist.isend(  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/distributed_c10d.py", line 1528, in isend
+
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/distributed_c10d.py", line 1528, in isend
+work = dist.isend(
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/distributed_c10d.py", line 1528, in isend
+work = dist.isend(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/distributed_c10d.py", line 1528, in isend
+    work = dist.isend(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/distributed_c10d.py", line 1528, in isend
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    work = dist.isend(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/distributed_c10d.py", line 1528, in isend
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return group.send([tensor], group_dst_rank, tag)
+RuntimeError: Broken pipe
+
+During handling of the above exception, another exception occurred:
+
+    Traceback (most recent call last):
+return group.send([tensor], group_dst_rank, tag)    
+return group.send([tensor], group_dst_rank, tag)
+      File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+return group.send([tensor], group_dst_rank, tag)RuntimeErrorRuntimeError
+: : Broken pipeBroken pipe
+RuntimeError
+: 
+During handling of the above exception, another exception occurred:
+
+
+During handling of the above exception, another exception occurred:
+
+Broken pipeTraceback (most recent call last):
+Traceback (most recent call last):
+
+
+During handling of the above exception, another exception occurred:
+
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+        outputs = self.pipeline_engine.train_batch_iter(return group.send([tensor], group_dst_rank, tag)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 263, in train_batch_iter
+RuntimeError: Broken pipe
+
+During handling of the above exception, another exception occurred:
+
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 263, in train_batch_iter
+    return group.send([tensor], group_dst_rank, tag)
+RuntimeError: Broken pipe
+
+During handling of the above exception, another exception occurred:
+
+Traceback (most recent call last):
+      File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+send_activation()
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/state.py", line 23, in __call__
+    send_activation()
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/state.py", line 23, in __call__
+    self.p2p.send_tensors([self.activation], to_rank=self.to_rank)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/p2p.py", line 498, in send_tensors
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    self.p2p.send_tensors([self.activation], to_rank=self.to_rank)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/p2p.py", line 498, in send_tensors
+    futures = self.isend_tensors(tensors=tensors, to_rank=to_rank, tag=tag)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/p2p.py", line 441, in isend_tensors
+    trainer.train(dataloader)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+trainer.train(dataloader)
+    trainer.train(dataloader)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    futures = self.isend_tensors(tensors=tensors, to_rank=to_rank, tag=tag)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/p2p.py", line 441, in isend_tensors
+return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+        return f(*args, **kwargs)return f(*args, **kwargs)
+
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+return f(*args, **kwargs)  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+    self._send_meta(tensor, to_rank=to_rank, tag=tag)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/p2p.py", line 339, in _send_meta
+    self._send_meta(tensor, to_rank=to_rank, tag=tag)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/p2p.py", line 339, in _send_meta
+outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+        raise P2PCommunicationError(f"Failed to send metadata to rank {to_rank}: {str(e)}")outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+
+nanotron.parallel.pipeline_parallel.p2p      File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+.outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)P2PCommunicationError
+: Failed to send metadata to rank 5: Broken pipe  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 263, in train_batch_iter
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 263, in train_batch_iter
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 263, in train_batch_iter
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 263, in train_batch_iter
+    raise P2PCommunicationError(f"Failed to send metadata to rank {to_rank}: {str(e)}")
+nanotron.parallel.pipeline_parallel.p2p.P2PCommunicationError: Failed to send metadata to rank 5: Broken pipe
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 263, in train_batch_iter
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 263, in train_batch_iter
+    send_activation()
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/state.py", line 23, in __call__
+    self.p2p.send_tensors([self.activation], to_rank=self.to_rank)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/p2p.py", line 498, in send_tensors
+    send_activation()
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/state.py", line 23, in __call__
+    send_activation()
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/state.py", line 23, in __call__
+    send_activation()    
+send_activation()  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/state.py", line 23, in __call__
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/state.py", line 23, in __call__
+    send_activation()
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/state.py", line 23, in __call__
+    futures = self.isend_tensors(tensors=tensors, to_rank=to_rank, tag=tag)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/p2p.py", line 441, in isend_tensors
+    self.p2p.send_tensors([self.activation], to_rank=self.to_rank)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/p2p.py", line 498, in send_tensors
+    self.p2p.send_tensors([self.activation], to_rank=self.to_rank)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/p2p.py", line 498, in send_tensors
+    self.p2p.send_tensors([self.activation], to_rank=self.to_rank)
+    self.p2p.send_tensors([self.activation], to_rank=self.to_rank)  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/p2p.py", line 498, in send_tensors
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/p2p.py", line 498, in send_tensors
+    self._send_meta(tensor, to_rank=to_rank, tag=tag)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/p2p.py", line 339, in _send_meta
+    futures = self.isend_tensors(tensors=tensors, to_rank=to_rank, tag=tag)    
+raise P2PCommunicationError(f"Failed to send metadata to rank {to_rank}: {str(e)}")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/p2p.py", line 441, in isend_tensors
+nanotron.parallel.pipeline_parallel.p2p.P2PCommunicationError: Failed to send metadata to rank 5: Broken pipe
+        futures = self.isend_tensors(tensors=tensors, to_rank=to_rank, tag=tag)futures = self.isend_tensors(tensors=tensors, to_rank=to_rank, tag=tag)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/p2p.py", line 441, in isend_tensors
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/p2p.py", line 441, in isend_tensors
+    self.p2p.send_tensors([self.activation], to_rank=self.to_rank)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/p2p.py", line 498, in send_tensors
+    futures = self.isend_tensors(tensors=tensors, to_rank=to_rank, tag=tag)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/p2p.py", line 441, in isend_tensors
+    self._send_meta(tensor, to_rank=to_rank, tag=tag)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/p2p.py", line 339, in _send_meta
+    self._send_meta(tensor, to_rank=to_rank, tag=tag)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/p2p.py", line 339, in _send_meta
+    self._send_meta(tensor, to_rank=to_rank, tag=tag)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/p2p.py", line 339, in _send_meta
+    futures = self.isend_tensors(tensors=tensors, to_rank=to_rank, tag=tag)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/p2p.py", line 441, in isend_tensors
+    self._send_meta(tensor, to_rank=to_rank, tag=tag)
+    raise P2PCommunicationError(f"Failed to send metadata to rank {to_rank}: {str(e)}")  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/p2p.py", line 339, in _send_meta
+
+nanotron.parallel.pipeline_parallel.p2p.P2PCommunicationError:     Failed to send metadata to rank 5: Broken pipe
+raise P2PCommunicationError(f"Failed to send metadata to rank {to_rank}: {str(e)}")
+nanotron.parallel.pipeline_parallel.p2p    .P2PCommunicationErrorraise P2PCommunicationError(f"Failed to send metadata to rank {to_rank}: {str(e)}"): 
+Failed to send metadata to rank 5: Broken pipe
+nanotron.parallel.pipeline_parallel.p2p.P2PCommunicationError:     Failed to send metadata to rank 5: Broken pipeself._send_meta(tensor, to_rank=to_rank, tag=tag)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/p2p.py", line 339, in _send_meta
+    raise P2PCommunicationError(f"Failed to send metadata to rank {to_rank}: {str(e)}")
+nanotron.parallel.pipeline_parallel.p2p.P2PCommunicationError: Failed to send metadata to rank 5: Broken pipe
+    raise P2PCommunicationError(f"Failed to send metadata to rank {to_rank}: {str(e)}")
+nanotron.parallel.pipeline_parallel.p2p.P2PCommunicationError: Failed to send metadata to rank 5: Broken pipe
+[2025-01-03 01:31:33,765] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 168462 closing signal SIGTERM
+[2025-01-03 01:31:33,766] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 168463 closing signal SIGTERM
+[2025-01-03 01:31:33,766] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 168464 closing signal SIGTERM
+[2025-01-03 01:31:33,766] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 168466 closing signal SIGTERM
+[2025-01-03 01:31:33,766] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 168467 closing signal SIGTERM
+[2025-01-03 01:31:33,766] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 168468 closing signal SIGTERM
+[2025-01-03 01:31:33,766] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 168469 closing signal SIGTERM
+[2025-01-03 01:31:33,767] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 41865 closing signal SIGTERM
+[2025-01-03 01:31:33,767] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 41866 closing signal SIGTERM
+[2025-01-03 01:31:33,767] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 41867 closing signal SIGTERM
+[2025-01-03 01:31:33,767] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 41869 closing signal SIGTERM
+[2025-01-03 01:31:33,767] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 41870 closing signal SIGTERM
+[2025-01-03 01:31:33,767] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 41871 closing signal SIGTERM
+[2025-01-03 01:31:33,767] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 41872 closing signal SIGTERM
+[2025-01-03 01:31:38,763] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 165979 closing signal SIGTERM
+[2025-01-03 01:31:38,764] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 165980 closing signal SIGTERM
+[2025-01-03 01:31:38,764] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 165981 closing signal SIGTERM
+[2025-01-03 01:31:38,764] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 165982 closing signal SIGTERM
+[2025-01-03 01:31:38,764] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 165983 closing signal SIGTERM
+[2025-01-03 01:31:38,764] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 165984 closing signal SIGTERM
+[2025-01-03 01:31:38,759] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 188366 closing signal SIGTERM
+[2025-01-03 01:31:38,764] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 165985 closing signal SIGTERM
+[2025-01-03 01:31:38,760] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 188367 closing signal SIGTERM
+[2025-01-03 01:31:38,760] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 188368 closing signal SIGTERM
+[2025-01-03 01:31:38,760] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 188369 closing signal SIGTERM
+[2025-01-03 01:31:38,760] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 188370 closing signal SIGTERM
+[2025-01-03 01:31:38,760] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 188371 closing signal SIGTERM
+[2025-01-03 01:31:38,760] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 188372 closing signal SIGTERM
+[2025-01-03 01:31:38,763] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 277602 closing signal SIGTERM
+[2025-01-03 01:31:38,763] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 277603 closing signal SIGTERM
+[2025-01-03 01:31:38,763] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 277604 closing signal SIGTERM
+[2025-01-03 01:31:38,763] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 277605 closing signal SIGTERM
+[2025-01-03 01:31:38,763] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 277606 closing signal SIGTERM
+[2025-01-03 01:31:38,763] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 277607 closing signal SIGTERM
+[2025-01-03 01:31:38,763] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 277608 closing signal SIGTERM
+[2025-01-03 01:31:38,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1120675 closing signal SIGTERM
+[2025-01-03 01:31:38,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1120677 closing signal SIGTERM
+[2025-01-03 01:31:38,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1120678 closing signal SIGTERM
+[2025-01-03 01:31:38,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1120679 closing signal SIGTERM
+[2025-01-03 01:31:38,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1120680 closing signal SIGTERM
+[2025-01-03 01:31:38,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 261313 closing signal SIGTERM
+[2025-01-03 01:31:38,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1120681 closing signal SIGTERM
+[2025-01-03 01:31:38,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 261314 closing signal SIGTERM
+[2025-01-03 01:31:38,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 261315 closing signal SIGTERM
+[2025-01-03 01:31:38,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 261316 closing signal SIGTERM
+[2025-01-03 01:31:38,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 261318 closing signal SIGTERM
+[2025-01-03 01:31:38,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 261319 closing signal SIGTERM
+[2025-01-03 01:31:38,771] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 294657 closing signal SIGTERM
+[2025-01-03 01:31:38,771] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 109529 closing signal SIGTERM
+[2025-01-03 01:31:38,771] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 109530 closing signal SIGTERM
+[2025-01-03 01:31:38,771] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 109531 closing signal SIGTERM
+[2025-01-03 01:31:38,771] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 294658 closing signal SIGTERM
+[2025-01-03 01:31:38,771] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 109532 closing signal SIGTERM
+[2025-01-03 01:31:38,771] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 294659 closing signal SIGTERM
+[2025-01-03 01:31:38,771] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 109533 closing signal SIGTERM
+[2025-01-03 01:31:38,771] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 294661 closing signal SIGTERM
+[2025-01-03 01:31:38,770] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 396154 closing signal SIGTERM
+[2025-01-03 01:31:38,771] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 396155 closing signal SIGTERM
+[2025-01-03 01:31:38,771] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 396156 closing signal SIGTERM
+[2025-01-03 01:31:38,771] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 396157 closing signal SIGTERM
+[2025-01-03 01:31:38,771] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 396158 closing signal SIGTERM
+[2025-01-03 01:31:38,771] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 396160 closing signal SIGTERM
+[2025-01-03 01:31:38,771] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 229223 closing signal SIGTERM
+[2025-01-03 01:31:38,771] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 229224 closing signal SIGTERM
+[2025-01-03 01:31:38,771] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 229226 closing signal SIGTERM
+[2025-01-03 01:31:38,771] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 229227 closing signal SIGTERM
+[2025-01-03 01:31:38,771] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 229228 closing signal SIGTERM
+[2025-01-03 01:31:38,771] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 229229 closing signal SIGTERM
+[2025-01-03 01:31:38,772] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 235567 closing signal SIGTERM
+[2025-01-03 01:31:38,772] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 235568 closing signal SIGTERM
+[2025-01-03 01:31:38,772] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 235569 closing signal SIGTERM
+[2025-01-03 01:31:38,772] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 235570 closing signal SIGTERM
+[2025-01-03 01:31:38,772] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 235572 closing signal SIGTERM
+[2025-01-03 01:31:38,772] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 235573 closing signal SIGTERM
+[2025-01-03 01:31:38,773] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 26749 closing signal SIGTERM
+[2025-01-03 01:31:38,773] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 26750 closing signal SIGTERM
+[2025-01-03 01:31:38,773] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 26751 closing signal SIGTERM
+[2025-01-03 01:31:38,773] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 26753 closing signal SIGTERM
+[2025-01-03 01:31:38,773] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 26754 closing signal SIGTERM
+[2025-01-03 01:31:38,775] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 102803 closing signal SIGTERM
+[2025-01-03 01:31:38,773] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 26755 closing signal SIGTERM
+[2025-01-03 01:31:38,775] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 102804 closing signal SIGTERM
+[2025-01-03 01:31:38,775] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 102805 closing signal SIGTERM
+[2025-01-03 01:31:38,775] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 102806 closing signal SIGTERM
+[2025-01-03 01:31:38,775] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 102807 closing signal SIGTERM
+[2025-01-03 01:31:38,775] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 102808 closing signal SIGTERM
+[2025-01-03 01:31:38,777] torch.distributed.elastic.multiprocessing.api: [ERROR] failed (exitcode: 1) local_rank: 0 (pid: 102583) of binary: /fsx/nouamane/miniconda/envs/2-1-cu121/bin/python
+[2025-01-03 01:31:38,776] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 312289 closing signal SIGTERM
+[2025-01-03 01:31:38,777] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 312290 closing signal SIGTERM
+[2025-01-03 01:31:38,777] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 312291 closing signal SIGTERM
+[2025-01-03 01:31:38,777] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 312292 closing signal SIGTERM
+[2025-01-03 01:31:38,777] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 312293 closing signal SIGTERM
+[2025-01-03 01:31:38,777] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 312294 closing signal SIGTERM
+[2025-01-03 01:31:38,833] torch.distributed.elastic.multiprocessing.errors.error_handler: [ERROR] no error file defined for parent, to copy child error file (/tmp/torchelastic_a2xcs83l/14019308_5ar9l1w_/attempt_0/0/error.json)
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 264, in launch_agent
+    raise ChildFailedError(
+torch.distributed.elastic.multiprocessing.errors.ChildFailedError: 
+============================================================
+/fsx/nouamane/projects/nanotron/run_train.py FAILED
+------------------------------------------------------------
+Failures:
+[1]:
+  time      : 2025-01-03_01:31:16
+  host      : ip-26-0-171-88.ec2.internal
+  rank      : 249 (local_rank: 1)
+  exitcode  : 1 (pid: 102584)
+  error_file: /tmp/torchelastic_a2xcs83l/14019308_5ar9l1w_/attempt_0/1/error.json
+  traceback : Traceback (most recent call last):
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+      return f(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+      outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+      outputs = self.pipeline_engine.train_batch_iter(
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+      output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+      output = model(**micro_batch)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+      sharded_logits = self.model(
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+      return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+      hidden_encoder_states = encoder_block(**hidden_encoder_states)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+      new_kwargs[name] = recv_from_pipeline_state_buffer(
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+      activation = pipeline_state.activations_buffer.popleft()
+  IndexError: pop from an empty deque
+  
+[2]:
+  time      : 2025-01-03_01:31:16
+  host      : ip-26-0-171-88.ec2.internal
+  rank      : 250 (local_rank: 2)
+  exitcode  : 1 (pid: 102585)
+  error_file: /tmp/torchelastic_a2xcs83l/14019308_5ar9l1w_/attempt_0/2/error.json
+  traceback : Traceback (most recent call last):
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+      return f(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+      outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+      outputs = self.pipeline_engine.train_batch_iter(
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+      output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+      output = model(**micro_batch)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+      sharded_logits = self.model(
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+      return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+      hidden_encoder_states = encoder_block(**hidden_encoder_states)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+      new_kwargs[name] = recv_from_pipeline_state_buffer(
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+      activation = pipeline_state.activations_buffer.popleft()
+  IndexError: pop from an empty deque
+  
+[3]:
+  time      : 2025-01-03_01:31:16
+  host      : ip-26-0-171-88.ec2.internal
+  rank      : 251 (local_rank: 3)
+  exitcode  : 1 (pid: 102586)
+  error_file: /tmp/torchelastic_a2xcs83l/14019308_5ar9l1w_/attempt_0/3/error.json
+  traceback : Traceback (most recent call last):
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+      return f(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+      outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+      outputs = self.pipeline_engine.train_batch_iter(
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+      output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+      output = model(**micro_batch)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+      sharded_logits = self.model(
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+      return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+      hidden_encoder_states = encoder_block(**hidden_encoder_states)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+      new_kwargs[name] = recv_from_pipeline_state_buffer(
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+      activation = pipeline_state.activations_buffer.popleft()
+  IndexError: pop from an empty deque
+  
+[4]:
+  time      : 2025-01-03_01:31:16
+  host      : ip-26-0-171-88.ec2.internal
+  rank      : 252 (local_rank: 4)
+  exitcode  : 1 (pid: 102587)
+  error_file: /tmp/torchelastic_a2xcs83l/14019308_5ar9l1w_/attempt_0/4/error.json
+  traceback : Traceback (most recent call last):
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+      return f(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+      outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+      outputs = self.pipeline_engine.train_batch_iter(
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+      output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+      output = model(**micro_batch)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+      sharded_logits = self.model(
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+      return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+      hidden_encoder_states = encoder_block(**hidden_encoder_states)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+      new_kwargs[name] = recv_from_pipeline_state_buffer(
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+      activation = pipeline_state.activations_buffer.popleft()
+  IndexError: pop from an empty deque
+  
+[5]:
+  time      : 2025-01-03_01:31:16
+  host      : ip-26-0-171-88.ec2.internal
+  rank      : 253 (local_rank: 5)
+  exitcode  : 1 (pid: 102588)
+  error_file: /tmp/torchelastic_a2xcs83l/14019308_5ar9l1w_/attempt_0/5/error.json
+  traceback : Traceback (most recent call last):
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+      return f(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+      outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+      outputs = self.pipeline_engine.train_batch_iter(
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+      output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+      output = model(**micro_batch)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+      sharded_logits = self.model(
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+      return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+      hidden_encoder_states = encoder_block(**hidden_encoder_states)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+      new_kwargs[name] = recv_from_pipeline_state_buffer(
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+      activation = pipeline_state.activations_buffer.popleft()
+  IndexError: pop from an empty deque
+  
+[6]:
+  time      : 2025-01-03_01:31:16
+  host      : ip-26-0-171-88.ec2.internal
+  rank      : 254 (local_rank: 6)
+  exitcode  : 1 (pid: 102589)
+  error_file: /tmp/torchelastic_a2xcs83l/14019308_5ar9l1w_/attempt_0/6/error.json
+  traceback : Traceback (most recent call last):
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+      return f(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+      outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+      outputs = self.pipeline_engine.train_batch_iter(
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+      output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+      output = model(**micro_batch)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+      sharded_logits = self.model(
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+      return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+      hidden_encoder_states = encoder_block(**hidden_encoder_states)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+      new_kwargs[name] = recv_from_pipeline_state_buffer(
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+      activation = pipeline_state.activations_buffer.popleft()
+  IndexError: pop from an empty deque
+  
+[7]:
+  time      : 2025-01-03_01:31:16
+  host      : ip-26-0-171-88.ec2.internal
+  rank      : 255 (local_rank: 7)
+  exitcode  : 1 (pid: 102590)
+  error_file: /tmp/torchelastic_a2xcs83l/14019308_5ar9l1w_/attempt_0/7/error.json
+  traceback : Traceback (most recent call last):
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+      return f(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+      outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+      outputs = self.pipeline_engine.train_batch_iter(
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+      output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+      output = model(**micro_batch)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+      sharded_logits = self.model(
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+      return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+      hidden_encoder_states = encoder_block(**hidden_encoder_states)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+      new_kwargs[name] = recv_from_pipeline_state_buffer(
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+      activation = pipeline_state.activations_buffer.popleft()
+  IndexError: pop from an empty deque
+  
+------------------------------------------------------------
+Root Cause (first observed failure):
+[0]:
+  time      : 2025-01-03_01:31:16
+  host      : ip-26-0-171-88.ec2.internal
+  rank      : 248 (local_rank: 0)
+  exitcode  : 1 (pid: 102583)
+  error_file: /tmp/torchelastic_a2xcs83l/14019308_5ar9l1w_/attempt_0/0/error.json
+  traceback : Traceback (most recent call last):
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+      return f(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+      outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+      outputs = self.pipeline_engine.train_batch_iter(
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 284, in train_batch_iter
+      output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+      output = model(**micro_batch)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+      sharded_logits = self.model(
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+      return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+      hidden_encoder_states = encoder_block(**hidden_encoder_states)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+      new_kwargs[name] = recv_from_pipeline_state_buffer(
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+      activation = pipeline_state.activations_buffer.popleft()
+  IndexError: pop from an empty deque
+  
+============================================================
+[2025-01-03 01:31:38,851] torch.distributed.elastic.multiprocessing.api: [ERROR] failed (exitcode: 1) local_rank: 3 (pid: 168465) of binary: /fsx/nouamane/miniconda/envs/2-1-cu121/bin/python
+[2025-01-03 01:31:38,892] torch.distributed.elastic.multiprocessing.errors.error_handler: [ERROR] no error file defined for parent, to copy child error file (/tmp/torchelastic_xvb4rx7p/14019308_wscx6fdh/attempt_0/3/error.json)
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 264, in launch_agent
+    raise ChildFailedError(
+torch.distributed.elastic.multiprocessing.errors.ChildFailedError: 
+============================================================
+/fsx/nouamane/projects/nanotron/run_train.py FAILED
+------------------------------------------------------------
+Failures:
+  <NO_OTHER_FAILURES>
+------------------------------------------------------------
+Root Cause (first observed failure):
+[0]:
+  time      : 2025-01-03_01:31:14
+  host      : ip-26-0-163-236.ec2.internal
+  rank      : 3 (local_rank: 3)
+  exitcode  : 1 (pid: 168465)
+  error_file: /tmp/torchelastic_xvb4rx7p/14019308_wscx6fdh/attempt_0/3/error.json
+  traceback : Traceback (most recent call last):
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+      return f(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 463, in train
+      outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 500, in training_step
+      outputs = self.pipeline_engine.train_batch_iter(
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 257, in train_batch_iter
+      output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+      output = model(**micro_batch)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1014, in forward
+      sharded_logits = self.model(
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 887, in forward
+      return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+      hidden_encoder_states = encoder_block(**hidden_encoder_states)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 160, in forward
+      output = self.pp_block(**new_kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 752, in forward
+      hidden_states, sequence_mask = self._core_forward(hidden_states, sequence_mask)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 725, in _core_forward
+      output = self.attn(hidden_states=hidden_states, sequence_mask=sequence_mask)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 690, in forward
+      output = self.o_proj(attention_output)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/nn.py", line 162, in forward
+      return row_linear(
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 593, in row_linear
+      return _RowLinearAsyncCommunication.apply(input, weight, bias, group, tp_mode)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+      return super().apply(*args, **kwargs)  # type: ignore[misc]
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 466, in forward
+      out = F.linear(tensor, weight, bias)
+  torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 128.00 MiB. GPU 3 has a total capacty of 79.33 GiB of which 109.94 MiB is free. Including non-PyTorch memory, this process has 79.21 GiB memory in use. Of the allocated memory 69.64 GiB is allocated by PyTorch, and 480.25 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  
+============================================================
+srun: error: ip-26-0-171-88: task 29: Exited with exit code 1
+srun: launch/slurm: _step_signal: Terminating StepId=14019308.0
+slurmstepd: error: *** STEP 14019308.0 ON ip-26-0-163-236 CANCELLED AT 2025-01-03T01:31:39 ***
+[2025-01-03 01:31:39,082] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 01:31:39,082] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 188366 closing signal SIGTERM
+[2025-01-03 01:31:39,082] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 188367 closing signal SIGTERM
+[2025-01-03 01:31:39,082] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 188368 closing signal SIGTERM
+[2025-01-03 01:31:39,085] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 01:31:39,082] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 188369 closing signal SIGTERM
+[2025-01-03 01:31:39,082] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 188370 closing signal SIGTERM
+[2025-01-03 01:31:39,086] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 41866 closing signal SIGTERM
+[2025-01-03 01:31:39,082] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 188371 closing signal SIGTERM
+[2025-01-03 01:31:39,086] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 41867 closing signal SIGTERM
+[2025-01-03 01:31:39,086] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 41869 closing signal SIGTERM
+[2025-01-03 01:31:39,086] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 41870 closing signal SIGTERM
+[2025-01-03 01:31:39,086] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 01:31:39,082] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 188372 closing signal SIGTERM
+[2025-01-03 01:31:39,085] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 01:31:39,086] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 41872 closing signal SIGTERM
+[2025-01-03 01:31:39,086] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 356679 closing signal SIGTERM
+[2025-01-03 01:31:39,086] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 01:31:39,086] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 356680 closing signal SIGTERM
+[2025-01-03 01:31:39,086] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 226831 closing signal SIGTERM
+[2025-01-03 01:31:39,086] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 226832 closing signal SIGTERM
+[2025-01-03 01:31:39,086] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 614255 closing signal SIGTERM
+[2025-01-03 01:31:39,087] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 614256 closing signal SIGTERM
+[2025-01-03 01:31:39,086] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 01:31:39,086] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 226833 closing signal SIGTERM
+[2025-01-03 01:31:39,085] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 01:31:39,085] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 01:31:39,087] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 614257 closing signal SIGTERM
+[2025-01-03 01:31:39,086] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 01:31:39,087] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 614258 closing signal SIGTERM
+[2025-01-03 01:31:39,085] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 01:31:39,086] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 217250 closing signal SIGTERM
+[2025-01-03 01:31:39,086] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 226834 closing signal SIGTERM
+[2025-01-03 01:31:39,087] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 614259 closing signal SIGTERM
+[2025-01-03 01:31:39,086] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 01:31:39,085] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 312289 closing signal SIGTERM
+[2025-01-03 01:31:39,087] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 614260 closing signal SIGTERM
+[2025-01-03 01:31:39,086] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 224461 closing signal SIGTERM
+[2025-01-03 01:31:39,086] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 75945 closing signal SIGTERM
+[2025-01-03 01:31:39,086] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 229223 closing signal SIGTERM
+[2025-01-03 01:31:39,086] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 217251 closing signal SIGTERM
+[2025-01-03 01:31:39,085] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 312290 closing signal SIGTERM
+[2025-01-03 01:31:39,087] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 614261 closing signal SIGTERM
+[2025-01-03 01:31:39,086] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 224462 closing signal SIGTERM
+[2025-01-03 01:31:39,087] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 01:31:39,086] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 261313 closing signal SIGTERM
+[2025-01-03 01:31:39,086] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 01:31:39,085] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 312291 closing signal SIGTERM
+[2025-01-03 01:31:39,087] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 614262 closing signal SIGTERM
+[2025-01-03 01:31:39,086] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 01:31:39,086] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 75946 closing signal SIGTERM
+[2025-01-03 01:31:39,086] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 229224 closing signal SIGTERM
+[2025-01-03 01:31:39,086] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 224463 closing signal SIGTERM
+[2025-01-03 01:31:39,086] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 217252 closing signal SIGTERM
+[2025-01-03 01:31:39,087] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 01:31:39,087] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 109529 closing signal SIGTERM
+[2025-01-03 01:31:39,086] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 01:31:39,086] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 261314 closing signal SIGTERM
+[2025-01-03 01:31:39,086] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 235567 closing signal SIGTERM
+[2025-01-03 01:31:39,085] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 312292 closing signal SIGTERM
+[2025-01-03 01:31:39,086] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 01:31:39,086] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 145102 closing signal SIGTERM
+[2025-01-03 01:31:39,086] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 75947 closing signal SIGTERM
+[2025-01-03 01:31:39,086] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 229226 closing signal SIGTERM
+[2025-01-03 01:31:39,086] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 224464 closing signal SIGTERM
+[2025-01-03 01:31:39,087] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 102803 closing signal SIGTERM
+[2025-01-03 01:31:39,087] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 109530 closing signal SIGTERM
+[2025-01-03 01:31:39,086] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 01:31:39,087] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 167142 closing signal SIGTERM
+[2025-01-03 01:31:39,086] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 261315 closing signal SIGTERM
+[2025-01-03 01:31:39,087] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 235568 closing signal SIGTERM
+[2025-01-03 01:31:39,085] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 312293 closing signal SIGTERM
+[2025-01-03 01:31:39,086] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 01:31:39,086] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 75948 closing signal SIGTERM
+[2025-01-03 01:31:39,086] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 229227 closing signal SIGTERM
+[2025-01-03 01:31:39,086] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 224465 closing signal SIGTERM
+[2025-01-03 01:31:39,086] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 217253 closing signal SIGTERM
+[2025-01-03 01:31:39,087] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 102804 closing signal SIGTERM
+[2025-01-03 01:31:39,087] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 109531 closing signal SIGTERM
+[2025-01-03 01:31:39,086] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 277602 closing signal SIGTERM
+[2025-01-03 01:31:39,086] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 261316 closing signal SIGTERM
+[2025-01-03 01:31:39,087] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 01:31:39,087] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 235569 closing signal SIGTERM
+[2025-01-03 01:31:39,085] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 312294 closing signal SIGTERM
+[2025-01-03 01:31:39,087] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 113630 closing signal SIGTERM
+[2025-01-03 01:31:39,086] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 145103 closing signal SIGTERM
+[2025-01-03 01:31:39,086] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 229228 closing signal SIGTERM
+[2025-01-03 01:31:39,086] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 224466 closing signal SIGTERM
+[2025-01-03 01:31:39,086] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 217254 closing signal SIGTERM
+[2025-01-03 01:31:39,087] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 102805 closing signal SIGTERM
+[2025-01-03 01:31:39,087] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 109532 closing signal SIGTERM
+[2025-01-03 01:31:39,086] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 277603 closing signal SIGTERM
+[2025-01-03 01:31:39,087] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 167143 closing signal SIGTERM
+[2025-01-03 01:31:39,086] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 261318 closing signal SIGTERM
+[2025-01-03 01:31:39,087] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 235570 closing signal SIGTERM
+[2025-01-03 01:31:39,087] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 113631 closing signal SIGTERM
+[2025-01-03 01:31:39,086] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 145104 closing signal SIGTERM
+[2025-01-03 01:31:39,087] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 410498 closing signal SIGTERM
+[2025-01-03 01:31:39,086] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 229229 closing signal SIGTERM
+[2025-01-03 01:31:39,087] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 01:31:39,086] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 01:31:39,086] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 217255 closing signal SIGTERM
+[2025-01-03 01:31:39,087] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 102806 closing signal SIGTERM
+[2025-01-03 01:31:39,087] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 109533 closing signal SIGTERM
+[2025-01-03 01:31:39,085] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 01:31:39,086] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 277604 closing signal SIGTERM
+[2025-01-03 01:31:39,087] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 167144 closing signal SIGTERM
+[2025-01-03 01:31:39,086] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 01:31:39,086] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 261319 closing signal SIGTERM
+[2025-01-03 01:31:39,087] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 165979 closing signal SIGTERM
+[2025-01-03 01:31:39,087] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 235572 closing signal SIGTERM
+[2025-01-03 01:31:39,087] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 113632 closing signal SIGTERM
+[2025-01-03 01:31:39,086] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 145105 closing signal SIGTERM
+[2025-01-03 01:31:39,087] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 410499 closing signal SIGTERM
+[2025-01-03 01:31:39,084] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 01:31:39,087] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 294657 closing signal SIGTERM
+[2025-01-03 01:31:39,086] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 396154 closing signal SIGTERM
+[2025-01-03 01:31:39,086] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 01:31:39,087] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 102807 closing signal SIGTERM
+[2025-01-03 01:31:39,086] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 26749 closing signal SIGTERM
+[2025-01-03 01:31:39,086] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 277605 closing signal SIGTERM
+[2025-01-03 01:31:39,087] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 167145 closing signal SIGTERM
+[2025-01-03 01:31:39,087] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1120675 closing signal SIGTERM
+[2025-01-03 01:31:39,087] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 165980 closing signal SIGTERM
+[2025-01-03 01:31:39,087] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 235573 closing signal SIGTERM
+[2025-01-03 01:31:39,087] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 113633 closing signal SIGTERM
+[2025-01-03 01:31:39,087] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 410500 closing signal SIGTERM
+[2025-01-03 01:31:39,087] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 294661 closing signal SIGTERM
+[2025-01-03 01:31:39,086] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 396155 closing signal SIGTERM
+[2025-01-03 01:31:39,087] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 102808 closing signal SIGTERM
+[2025-01-03 01:31:39,086] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 26750 closing signal SIGTERM
+[2025-01-03 01:31:39,087] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 167146 closing signal SIGTERM
+[2025-01-03 01:31:39,087] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 165981 closing signal SIGTERM
+[2025-01-03 01:31:39,087] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 410501 closing signal SIGTERM
+[2025-01-03 01:31:39,086] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 396156 closing signal SIGTERM
+[2025-01-03 01:31:39,087] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 355099 closing signal SIGTERM
+[2025-01-03 01:31:39,086] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 26751 closing signal SIGTERM
+[2025-01-03 01:31:39,086] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 277606 closing signal SIGTERM
+[2025-01-03 01:31:39,087] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1120677 closing signal SIGTERM
+[2025-01-03 01:31:39,087] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 165982 closing signal SIGTERM
+[2025-01-03 01:31:39,086] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 145106 closing signal SIGTERM
+[2025-01-03 01:31:39,085] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 405879 closing signal SIGTERM
+[2025-01-03 01:31:39,086] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 396157 closing signal SIGTERM
+[2025-01-03 01:31:39,087] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 355100 closing signal SIGTERM
+[2025-01-03 01:31:39,086] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 26753 closing signal SIGTERM
+[2025-01-03 01:31:39,087] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1120678 closing signal SIGTERM
+[2025-01-03 01:31:39,087] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 165983 closing signal SIGTERM
+[2025-01-03 01:31:39,085] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 01:31:39,086] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 396158 closing signal SIGTERM
+[2025-01-03 01:31:39,087] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 355101 closing signal SIGTERM
+[2025-01-03 01:31:39,086] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 26754 closing signal SIGTERM
+[2025-01-03 01:31:39,087] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 167147 closing signal SIGTERM
+[2025-01-03 01:31:39,087] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 165984 closing signal SIGTERM
+[2025-01-03 01:31:39,086] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 396160 closing signal SIGTERM
+[2025-01-03 01:31:39,087] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 355102 closing signal SIGTERM
+[2025-01-03 01:31:39,086] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 26755 closing signal SIGTERM
+[2025-01-03 01:31:39,086] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 277607 closing signal SIGTERM
+[2025-01-03 01:31:39,087] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 167148 closing signal SIGTERM
+[2025-01-03 01:31:39,087] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 165985 closing signal SIGTERM
+[2025-01-03 01:31:39,087] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 167149 closing signal SIGTERM
+[2025-01-03 01:31:39,087] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1120679 closing signal SIGTERM
+[2025-01-03 01:31:39,086] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 463099 closing signal SIGTERM
+[2025-01-03 01:31:39,086] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 277608 closing signal SIGTERM
+[2025-01-03 01:31:39,085] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 405880 closing signal SIGTERM
+[2025-01-03 01:31:39,087] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1120680 closing signal SIGTERM
+[2025-01-03 01:31:39,086] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 01:31:39,086] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 75949 closing signal SIGTERM
+[2025-01-03 01:31:39,087] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 75950 closing signal SIGTERM
+[2025-01-03 01:31:39,085] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 405881 closing signal SIGTERM
+[2025-01-03 01:31:39,085] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 405882 closing signal SIGTERM
+[2025-01-03 01:31:39,085] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 405883 closing signal SIGTERM
+[2025-01-03 01:31:39,086] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 463100 closing signal SIGTERM
+[2025-01-03 01:31:39,087] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1120681 closing signal SIGTERM
+[2025-01-03 01:31:39,086] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 211608 closing signal SIGTERM
+[2025-01-03 01:31:39,087] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 01:31:39,086] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 463101 closing signal SIGTERM
+[2025-01-03 01:31:39,086] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 211609 closing signal SIGTERM
+[2025-01-03 01:31:39,087] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 217256 closing signal SIGTERM
+[2025-01-03 01:31:39,087] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 360155 closing signal SIGTERM
+[2025-01-03 01:31:39,087] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 360156 closing signal SIGTERM
+[2025-01-03 01:31:39,087] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 211610 closing signal SIGTERM
+[2025-01-03 01:31:39,087] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 211611 closing signal SIGTERM
+[2025-01-03 01:31:39,087] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 211612 closing signal SIGTERM
+[2025-01-03 01:31:39,087] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 360157 closing signal SIGTERM
+[2025-01-03 01:31:39,087] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 360158 closing signal SIGTERM
+[2025-01-03 01:31:39,087] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 113634 closing signal SIGTERM
+[2025-01-03 01:31:39,087] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 113635 closing signal SIGTERM
+[2025-01-03 01:31:39,087] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 113636 closing signal SIGTERM
+[2025-01-03 01:31:39,087] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 360159 closing signal SIGTERM
+[2025-01-03 01:31:39,086] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 226835 closing signal SIGTERM
+[2025-01-03 01:31:39,086] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 226836 closing signal SIGTERM
+[2025-01-03 01:31:39,087] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 355103 closing signal SIGTERM
+[2025-01-03 01:31:39,087] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 355104 closing signal SIGTERM
+[2025-01-03 01:31:39,087] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 75951 closing signal SIGTERM
+[2025-01-03 01:31:39,086] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 01:31:39,088] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 360160 closing signal SIGTERM
+[2025-01-03 01:31:39,088] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 113637 closing signal SIGTERM
+[2025-01-03 01:31:39,086] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 463102 closing signal SIGTERM
+[2025-01-03 01:31:39,086] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 463103 closing signal SIGTERM
+[2025-01-03 01:31:39,086] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 405884 closing signal SIGTERM
+[2025-01-03 01:31:39,087] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 226246 closing signal SIGTERM
+[2025-01-03 01:31:39,088] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 355105 closing signal SIGTERM
+[2025-01-03 01:31:39,087] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 226247 closing signal SIGTERM
+[2025-01-03 01:31:39,088] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 217257 closing signal SIGTERM
+[2025-01-03 01:31:39,087] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 226248 closing signal SIGTERM
+[2025-01-03 01:31:39,088] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 360161 closing signal SIGTERM
+[2025-01-03 01:31:39,087] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 463104 closing signal SIGTERM
+[2025-01-03 01:31:39,087] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 226249 closing signal SIGTERM
+[2025-01-03 01:31:39,087] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 226250 closing signal SIGTERM
+[2025-01-03 01:31:39,086] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 356681 closing signal SIGTERM
+[2025-01-03 01:31:39,086] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 356682 closing signal SIGTERM
+[2025-01-03 01:31:39,087] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 356683 closing signal SIGTERM
+[2025-01-03 01:31:39,087] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 356684 closing signal SIGTERM
+[2025-01-03 01:31:39,086] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 405885 closing signal SIGTERM
+[2025-01-03 01:31:39,087] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 211613 closing signal SIGTERM
+[2025-01-03 01:31:39,088] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 145107 closing signal SIGTERM
+[2025-01-03 01:31:39,089] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 360162 closing signal SIGTERM
+[2025-01-03 01:31:39,088] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 463105 closing signal SIGTERM
+[2025-01-03 01:31:39,089] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 356685 closing signal SIGTERM
+[2025-01-03 01:31:39,089] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 410502 closing signal SIGTERM
+[2025-01-03 01:31:39,089] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 224467 closing signal SIGTERM
+[2025-01-03 01:31:39,088] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 463106 closing signal SIGTERM
+[2025-01-03 01:31:39,088] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 355106 closing signal SIGTERM
+[2025-01-03 01:31:39,090] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 145108 closing signal SIGTERM
+[2025-01-03 01:31:39,088] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 405886 closing signal SIGTERM
+[2025-01-03 01:31:39,089] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 224468 closing signal SIGTERM
+[2025-01-03 01:31:39,090] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 410503 closing signal SIGTERM
+[2025-01-03 01:31:39,090] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 410504 closing signal SIGTERM
+[2025-01-03 01:31:39,090] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 410505 closing signal SIGTERM
+[2025-01-03 01:31:39,089] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 226837 closing signal SIGTERM
+[2025-01-03 01:31:39,087] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 226251 closing signal SIGTERM
+[2025-01-03 01:31:39,087] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 226252 closing signal SIGTERM
+[2025-01-03 01:31:39,087] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 226253 closing signal SIGTERM
+[2025-01-03 01:31:39,089] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 211614 closing signal SIGTERM
+[2025-01-03 01:31:39,089] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 211615 closing signal SIGTERM
+[2025-01-03 01:31:39,091] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 356686 closing signal SIGTERM
+[2025-01-03 01:31:39,088] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 75952 closing signal SIGTERM
+[2025-01-03 01:31:39,091] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 145109 closing signal SIGTERM
+[2025-01-03 01:31:39,090] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 226838 closing signal SIGTERM
+srun: error: ip-26-0-163-236: task 0: Terminated
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 878, in _invoke_run
+    run_result = self._monitor_workers(self._worker_group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/local_elastic_agent.py", line 307, in _monitor_workers
+    result = self._pcontext.wait(0)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 288, in wait
+    return self._poll()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 668, in _poll
+    self.close()  # terminate all running procs
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 331, in close
+    self._close(death_sig=death_sig, timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 713, in _close
+    handler.proc.wait(time_to_wait)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1209, in wait
+    return self._wait(timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1953, in _wait
+    time.sleep(delay)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 294540 got signal: 15
+srun: error: ip-26-0-171-56: task 27: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 878, in _invoke_run
+    run_result = self._monitor_workers(self._worker_group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/local_elastic_agent.py", line 307, in _monitor_workers
+    result = self._pcontext.wait(0)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 288, in wait
+    return self._poll()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 668, in _poll
+    self.close()  # terminate all running procs
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 331, in close
+    self._close(death_sig=death_sig, timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 713, in _close
+    handler.proc.wait(time_to_wait)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1209, in wait
+    return self._wait(timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1953, in _wait
+    time.sleep(delay)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 109361 got signal: 15
+srun: error: ip-26-0-167-51: task 21: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 878, in _invoke_run
+    run_result = self._monitor_workers(self._worker_group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/local_elastic_agent.py", line 307, in _monitor_workers
+    result = self._pcontext.wait(0)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 288, in wait
+    return self._poll()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 668, in _poll
+    self.close()  # terminate all running procs
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 331, in close
+    self._close(death_sig=death_sig, timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 713, in _close
+    handler.proc.wait(time_to_wait)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1209, in wait
+    return self._wait(timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1953, in _wait
+    time.sleep(delay)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 261192 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 878, in _invoke_run
+    run_result = self._monitor_workers(self._worker_group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/local_elastic_agent.py", line 307, in _monitor_workers
+    result = self._pcontext.wait(0)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 288, in wait
+    return self._poll()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 668, in _poll
+    self.close()  # terminate all running procs
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 331, in close
+    self._close(death_sig=death_sig, timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 713, in _close
+    handler.proc.wait(time_to_wait)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1209, in wait
+    return self._wait(timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1953, in _wait
+    time.sleep(delay)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 26640 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 878, in _invoke_run
+    run_result = self._monitor_workers(self._worker_group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/local_elastic_agent.py", line 307, in _monitor_workers
+    result = self._pcontext.wait(0)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 288, in wait
+    return self._poll()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 668, in _poll
+    self.close()  # terminate all running procs
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 331, in close
+    self._close(death_sig=death_sig, timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 713, in _close
+    handler.proc.wait(time_to_wait)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1209, in wait
+    return self._wait(timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1953, in _wait
+    time.sleep(delay)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 312120 got signal: 15
+srun: error: ip-26-0-171-102: task 30: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 878, in _invoke_run
+    run_result = self._monitor_workers(self._worker_group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/local_elastic_agent.py", line 307, in _monitor_workers
+    result = self._pcontext.wait(0)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 288, in wait
+    return self._poll()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 668, in _poll
+    self.close()  # terminate all running procs
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 331, in close
+    self._close(death_sig=death_sig, timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 713, in _close
+    handler.proc.wait(time_to_wait)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1209, in wait
+    return self._wait(timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1953, in _wait
+    time.sleep(delay)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 396038 got signal: 15
+srun: error: ip-26-0-170-143: task 24: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 878, in _invoke_run
+    run_result = self._monitor_workers(self._worker_group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/local_elastic_agent.py", line 307, in _monitor_workers
+    result = self._pcontext.wait(0)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 288, in wait
+    return self._poll()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 668, in _poll
+    self.close()  # terminate all running procs
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 331, in close
+srun: error: ip-26-0-167-177: task 23: Exited with exit code 1
+    self._close(death_sig=death_sig, timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 713, in _close
+    handler.proc.wait(time_to_wait)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1209, in wait
+    return self._wait(timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1953, in _wait
+    time.sleep(delay)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 1120567 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 878, in _invoke_run
+    run_result = self._monitor_workers(self._worker_group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/local_elastic_agent.py", line 307, in _monitor_workers
+    result = self._pcontext.wait(0)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 288, in wait
+    return self._poll()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 668, in _poll
+    self.close()  # terminate all running procs
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 331, in close
+    self._close(death_sig=death_sig, timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 713, in _close
+    handler.proc.wait(time_to_wait)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1209, in wait
+    return self._wait(timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1953, in _wait
+    time.sleep(delay)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 235398 got signal: 15
+srun: error: ip-26-0-171-21: task 26: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 878, in _invoke_run
+    run_result = self._monitor_workers(self._worker_group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/local_elastic_agent.py", line 307, in _monitor_workers
+    result = self._pcontext.wait(0)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 288, in wait
+    return self._poll()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 668, in _poll
+    self.close()  # terminate all running procs
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 331, in close
+    self._close(death_sig=death_sig, timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 713, in _close
+    handler.proc.wait(time_to_wait)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1209, in wait
+    return self._wait(timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1953, in _wait
+    time.sleep(delay)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 277486 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 878, in _invoke_run
+    run_result = self._monitor_workers(self._worker_group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/local_elastic_agent.py", line 307, in _monitor_workers
+    result = self._pcontext.wait(0)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 288, in wait
+    return self._poll()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 668, in _poll
+    self.close()  # terminate all running procs
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 331, in close
+    self._close(death_sig=death_sig, timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 713, in _close
+    handler.proc.wait(time_to_wait)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1209, in wait
+    return self._wait(timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1953, in _wait
+    time.sleep(delay)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 41698 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 878, in _invoke_run
+    run_result = self._monitor_workers(self._worker_group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/local_elastic_agent.py", line 307, in _monitor_workers
+    result = self._pcontext.wait(0)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 288, in wait
+    return self._poll()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 668, in _poll
+    self.close()  # terminate all running procs
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 331, in close
+    self._close(death_sig=death_sig, timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 713, in _close
+    handler.proc.wait(time_to_wait)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1209, in wait
+    return self._wait(timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1953, in _wait
+    time.sleep(delay)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 165870 got signal: 15
+srun: error: ip-26-0-171-168: task 31: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 878, in _invoke_run
+    run_result = self._monitor_workers(self._worker_group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/local_elastic_agent.py", line 307, in _monitor_workers
+    result = self._pcontext.wait(0)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 288, in wait
+    return self._poll()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 668, in _poll
+    self.close()  # terminate all running procs
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 331, in close
+    self._close(death_sig=death_sig, timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 713, in _close
+    handler.proc.wait(time_to_wait)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1209, in wait
+    return self._wait(timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1953, in _wait
+    time.sleep(delay)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 229054 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 878, in _invoke_run
+    run_result = self._monitor_workers(self._worker_group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/local_elastic_agent.py", line 307, in _monitor_workers
+    result = self._pcontext.wait(0)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 288, in wait
+    return self._poll()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 668, in _poll
+    self.close()  # terminate all running procs
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 331, in close
+    self._close(death_sig=death_sig, timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 713, in _close
+    handler.proc.wait(time_to_wait)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1209, in wait
+    return self._wait(timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1953, in _wait
+    time.sleep(delay)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 188197 got signal: 15
+srun: error: ip-26-0-167-175: task 22: Exited with exit code 1
+srun: error: ip-26-0-171-62: task 28: Exited with exit code 1
+srun: error: ip-26-0-164-18: task 2: Exited with exit code 1
+srun: error: ip-26-0-170-160: task 25: Exited with exit code 1
+srun: error: ip-26-0-164-187: task 5: Exited with exit code 1
+srun: error: ip-26-0-164-0: task 1: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 878, in _invoke_run
+    run_result = self._monitor_workers(self._worker_group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/local_elastic_agent.py", line 307, in _monitor_workers
+    result = self._pcontext.wait(0)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 288, in wait
+    return self._poll()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 668, in _poll
+    self.close()  # terminate all running procs
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 331, in close
+    self._close(death_sig=death_sig, timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 713, in _close
+    handler.proc.wait(time_to_wait)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1209, in wait
+    return self._wait(timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1953, in _wait
+    time.sleep(delay)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 102633 got signal: 15
+srun: error: ip-26-0-167-9: task 20: Exited with exit code 1
+[2025-01-03 01:31:43,418] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-165-213.ec2.internal_113440_0' has failed to send a keep-alive heartbeat to the rendezvous '14019308' due to an error of type RendezvousConnectionError.
+[2025-01-03 01:31:43,430] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-166-244.ec2.internal_166951_0' has failed to send a keep-alive heartbeat to the rendezvous '14019308' due to an error of type RendezvousConnectionError.
+[2025-01-03 01:31:43,446] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-165-131.ec2.internal_75765_0' has failed to send a keep-alive heartbeat to the rendezvous '14019308' due to an error of type RendezvousConnectionError.
+[2025-01-03 01:31:43,452] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-164-236.ec2.internal_226664_0' has failed to send a keep-alive heartbeat to the rendezvous '14019308' due to an error of type RendezvousConnectionError.
+[2025-01-03 01:31:43,487] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-164-45.ec2.internal_356510_0' has failed to send a keep-alive heartbeat to the rendezvous '14019308' due to an error of type RendezvousConnectionError.
+[2025-01-03 01:31:43,517] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-166-214.ec2.internal_354932_0' has failed to send a keep-alive heartbeat to the rendezvous '14019308' due to an error of type RendezvousConnectionError.
+[2025-01-03 01:31:43,544] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-165-202.ec2.internal_211419_0' has failed to send a keep-alive heartbeat to the rendezvous '14019308' due to an error of type RendezvousConnectionError.
+[2025-01-03 01:31:43,557] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-164-207.ec2.internal_144935_0' has failed to send a keep-alive heartbeat to the rendezvous '14019308' due to an error of type RendezvousConnectionError.
+[2025-01-03 01:31:43,567] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-164-75.ec2.internal_405711_0' has failed to send a keep-alive heartbeat to the rendezvous '14019308' due to an error of type RendezvousConnectionError.
+[2025-01-03 01:31:43,573] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-166-125.ec2.internal_410317_0' has failed to send a keep-alive heartbeat to the rendezvous '14019308' due to an error of type RendezvousConnectionError.
+[2025-01-03 01:31:43,577] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-165-24.ec2.internal_359952_0' has failed to send a keep-alive heartbeat to the rendezvous '14019308' due to an error of type RendezvousConnectionError.
+[2025-01-03 01:31:43,601] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-165-38.ec2.internal_226056_0' has failed to send a keep-alive heartbeat to the rendezvous '14019308' due to an error of type RendezvousConnectionError.
+[2025-01-03 01:31:43,604] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-165-164.ec2.internal_217060_0' has failed to send a keep-alive heartbeat to the rendezvous '14019308' due to an error of type RendezvousConnectionError.
+[2025-01-03 01:31:43,608] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-166-15.ec2.internal_614065_0' has failed to send a keep-alive heartbeat to the rendezvous '14019308' due to an error of type RendezvousConnectionError.
+[2025-01-03 01:31:43,663] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-165-59.ec2.internal_224271_0' has failed to send a keep-alive heartbeat to the rendezvous '14019308' due to an error of type RendezvousConnectionError.
+[2025-01-03 01:31:43,677] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-166-36.ec2.internal_462909_0' has failed to send a keep-alive heartbeat to the rendezvous '14019308' due to an error of type RendezvousConnectionError.
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 166951 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 614065 got signal: 15
+srun: error: ip-26-0-166-244: task 19: Exited with exit code 1
+srun: error: ip-26-0-166-15: task 15: Exited with exit code 1
+[2025-01-03 01:31:48,419] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-165-213.ec2.internal_113440_0' has failed to send a keep-alive heartbeat to the rendezvous '14019308' due to an error of type RendezvousConnectionError.
+[2025-01-03 01:31:48,447] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-165-131.ec2.internal_75765_0' has failed to send a keep-alive heartbeat to the rendezvous '14019308' due to an error of type RendezvousConnectionError.
+[2025-01-03 01:31:48,453] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-164-236.ec2.internal_226664_0' has failed to send a keep-alive heartbeat to the rendezvous '14019308' due to an error of type RendezvousConnectionError.
+[2025-01-03 01:31:48,488] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-164-45.ec2.internal_356510_0' has failed to send a keep-alive heartbeat to the rendezvous '14019308' due to an error of type RendezvousConnectionError.
+[2025-01-03 01:31:48,519] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-166-214.ec2.internal_354932_0' has failed to send a keep-alive heartbeat to the rendezvous '14019308' due to an error of type RendezvousConnectionError.
+[2025-01-03 01:31:48,545] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-165-202.ec2.internal_211419_0' has failed to send a keep-alive heartbeat to the rendezvous '14019308' due to an error of type RendezvousConnectionError.
+[2025-01-03 01:31:48,558] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-164-207.ec2.internal_144935_0' has failed to send a keep-alive heartbeat to the rendezvous '14019308' due to an error of type RendezvousConnectionError.
+[2025-01-03 01:31:48,568] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-164-75.ec2.internal_405711_0' has failed to send a keep-alive heartbeat to the rendezvous '14019308' due to an error of type RendezvousConnectionError.
+[2025-01-03 01:31:48,574] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-166-125.ec2.internal_410317_0' has failed to send a keep-alive heartbeat to the rendezvous '14019308' due to an error of type RendezvousConnectionError.
+[2025-01-03 01:31:48,578] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-165-24.ec2.internal_359952_0' has failed to send a keep-alive heartbeat to the rendezvous '14019308' due to an error of type RendezvousConnectionError.
+[2025-01-03 01:31:48,602] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-165-38.ec2.internal_226056_0' has failed to send a keep-alive heartbeat to the rendezvous '14019308' due to an error of type RendezvousConnectionError.
+[2025-01-03 01:31:48,605] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-165-164.ec2.internal_217060_0' has failed to send a keep-alive heartbeat to the rendezvous '14019308' due to an error of type RendezvousConnectionError.
+[2025-01-03 01:31:48,664] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-165-59.ec2.internal_224271_0' has failed to send a keep-alive heartbeat to the rendezvous '14019308' due to an error of type RendezvousConnectionError.
+[2025-01-03 01:31:48,678] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-166-36.ec2.internal_462909_0' has failed to send a keep-alive heartbeat to the rendezvous '14019308' due to an error of type RendezvousConnectionError.
+[2025-01-03 01:31:53,421] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-165-213.ec2.internal_113440_0' has failed to send a keep-alive heartbeat to the rendezvous '14019308' due to an error of type RendezvousConnectionError.
+[2025-01-03 01:31:53,448] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-165-131.ec2.internal_75765_0' has failed to send a keep-alive heartbeat to the rendezvous '14019308' due to an error of type RendezvousConnectionError.
+[2025-01-03 01:31:53,454] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-164-236.ec2.internal_226664_0' has failed to send a keep-alive heartbeat to the rendezvous '14019308' due to an error of type RendezvousConnectionError.
+[2025-01-03 01:31:53,489] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-164-45.ec2.internal_356510_0' has failed to send a keep-alive heartbeat to the rendezvous '14019308' due to an error of type RendezvousConnectionError.
+[2025-01-03 01:31:53,520] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-166-214.ec2.internal_354932_0' has failed to send a keep-alive heartbeat to the rendezvous '14019308' due to an error of type RendezvousConnectionError.
+[2025-01-03 01:31:53,546] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-165-202.ec2.internal_211419_0' has failed to send a keep-alive heartbeat to the rendezvous '14019308' due to an error of type RendezvousConnectionError.
+[2025-01-03 01:31:53,560] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-164-207.ec2.internal_144935_0' has failed to send a keep-alive heartbeat to the rendezvous '14019308' due to an error of type RendezvousConnectionError.
+[2025-01-03 01:31:53,570] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-164-75.ec2.internal_405711_0' has failed to send a keep-alive heartbeat to the rendezvous '14019308' due to an error of type RendezvousConnectionError.
+[2025-01-03 01:31:53,576] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-166-125.ec2.internal_410317_0' has failed to send a keep-alive heartbeat to the rendezvous '14019308' due to an error of type RendezvousConnectionError.
+[2025-01-03 01:31:53,579] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-165-24.ec2.internal_359952_0' has failed to send a keep-alive heartbeat to the rendezvous '14019308' due to an error of type RendezvousConnectionError.
+[2025-01-03 01:31:53,604] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-165-38.ec2.internal_226056_0' has failed to send a keep-alive heartbeat to the rendezvous '14019308' due to an error of type RendezvousConnectionError.
+[2025-01-03 01:31:53,607] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-165-164.ec2.internal_217060_0' has failed to send a keep-alive heartbeat to the rendezvous '14019308' due to an error of type RendezvousConnectionError.
+[2025-01-03 01:31:53,665] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-165-59.ec2.internal_224271_0' has failed to send a keep-alive heartbeat to the rendezvous '14019308' due to an error of type RendezvousConnectionError.
+[2025-01-03 01:31:53,680] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-166-36.ec2.internal_462909_0' has failed to send a keep-alive heartbeat to the rendezvous '14019308' due to an error of type RendezvousConnectionError.
+[2025-01-03 01:31:58,422] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-165-213.ec2.internal_113440_0' has failed to send a keep-alive heartbeat to the rendezvous '14019308' due to an error of type RendezvousConnectionError.
+[2025-01-03 01:31:58,450] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-165-131.ec2.internal_75765_0' has failed to send a keep-alive heartbeat to the rendezvous '14019308' due to an error of type RendezvousConnectionError.
+[2025-01-03 01:31:58,456] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-164-236.ec2.internal_226664_0' has failed to send a keep-alive heartbeat to the rendezvous '14019308' due to an error of type RendezvousConnectionError.
+[2025-01-03 01:31:58,491] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-164-45.ec2.internal_356510_0' has failed to send a keep-alive heartbeat to the rendezvous '14019308' due to an error of type RendezvousConnectionError.
+[2025-01-03 01:31:58,522] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-166-214.ec2.internal_354932_0' has failed to send a keep-alive heartbeat to the rendezvous '14019308' due to an error of type RendezvousConnectionError.
+[2025-01-03 01:31:58,548] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-165-202.ec2.internal_211419_0' has failed to send a keep-alive heartbeat to the rendezvous '14019308' due to an error of type RendezvousConnectionError.
+[2025-01-03 01:31:58,561] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-164-207.ec2.internal_144935_0' has failed to send a keep-alive heartbeat to the rendezvous '14019308' due to an error of type RendezvousConnectionError.
+[2025-01-03 01:31:58,571] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-164-75.ec2.internal_405711_0' has failed to send a keep-alive heartbeat to the rendezvous '14019308' due to an error of type RendezvousConnectionError.
+[2025-01-03 01:31:58,577] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-166-125.ec2.internal_410317_0' has failed to send a keep-alive heartbeat to the rendezvous '14019308' due to an error of type RendezvousConnectionError.
+[2025-01-03 01:31:58,580] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-165-24.ec2.internal_359952_0' has failed to send a keep-alive heartbeat to the rendezvous '14019308' due to an error of type RendezvousConnectionError.
+[2025-01-03 01:31:58,606] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-165-38.ec2.internal_226056_0' has failed to send a keep-alive heartbeat to the rendezvous '14019308' due to an error of type RendezvousConnectionError.
+[2025-01-03 01:31:58,609] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-165-164.ec2.internal_217060_0' has failed to send a keep-alive heartbeat to the rendezvous '14019308' due to an error of type RendezvousConnectionError.
+[2025-01-03 01:31:58,667] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-165-59.ec2.internal_224271_0' has failed to send a keep-alive heartbeat to the rendezvous '14019308' due to an error of type RendezvousConnectionError.
+[2025-01-03 01:31:58,681] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-166-36.ec2.internal_462909_0' has failed to send a keep-alive heartbeat to the rendezvous '14019308' due to an error of type RendezvousConnectionError.
+[2025-01-03 01:32:03,424] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-165-213.ec2.internal_113440_0' has failed to send a keep-alive heartbeat to the rendezvous '14019308' due to an error of type RendezvousConnectionError.
+[2025-01-03 01:32:03,451] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-165-131.ec2.internal_75765_0' has failed to send a keep-alive heartbeat to the rendezvous '14019308' due to an error of type RendezvousConnectionError.
+[2025-01-03 01:32:03,457] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-164-236.ec2.internal_226664_0' has failed to send a keep-alive heartbeat to the rendezvous '14019308' due to an error of type RendezvousConnectionError.
+[2025-01-03 01:32:03,493] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-164-45.ec2.internal_356510_0' has failed to send a keep-alive heartbeat to the rendezvous '14019308' due to an error of type RendezvousConnectionError.
+[2025-01-03 01:32:03,523] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-166-214.ec2.internal_354932_0' has failed to send a keep-alive heartbeat to the rendezvous '14019308' due to an error of type RendezvousConnectionError.
+[2025-01-03 01:32:03,550] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-165-202.ec2.internal_211419_0' has failed to send a keep-alive heartbeat to the rendezvous '14019308' due to an error of type RendezvousConnectionError.
+[2025-01-03 01:32:03,563] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-164-207.ec2.internal_144935_0' has failed to send a keep-alive heartbeat to the rendezvous '14019308' due to an error of type RendezvousConnectionError.
+[2025-01-03 01:32:03,573] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-164-75.ec2.internal_405711_0' has failed to send a keep-alive heartbeat to the rendezvous '14019308' due to an error of type RendezvousConnectionError.
+[2025-01-03 01:32:03,578] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-166-125.ec2.internal_410317_0' has failed to send a keep-alive heartbeat to the rendezvous '14019308' due to an error of type RendezvousConnectionError.
+[2025-01-03 01:32:03,582] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-165-24.ec2.internal_359952_0' has failed to send a keep-alive heartbeat to the rendezvous '14019308' due to an error of type RendezvousConnectionError.
+[2025-01-03 01:32:03,608] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-165-38.ec2.internal_226056_0' has failed to send a keep-alive heartbeat to the rendezvous '14019308' due to an error of type RendezvousConnectionError.
+[2025-01-03 01:32:03,610] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-165-164.ec2.internal_217060_0' has failed to send a keep-alive heartbeat to the rendezvous '14019308' due to an error of type RendezvousConnectionError.
+[2025-01-03 01:32:03,669] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-165-59.ec2.internal_224271_0' has failed to send a keep-alive heartbeat to the rendezvous '14019308' due to an error of type RendezvousConnectionError.
+[2025-01-03 01:32:03,683] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-166-36.ec2.internal_462909_0' has failed to send a keep-alive heartbeat to the rendezvous '14019308' due to an error of type RendezvousConnectionError.
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 354932 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 462909 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 75765 got signal: 15
+srun: error: ip-26-0-166-214: task 18: Exited with exit code 1
+srun: error: ip-26-0-166-36: task 16: Exited with exit code 1
+srun: error: ip-26-0-165-131: task 11: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 113440 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 224271 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 405711 got signal: 15
+srun: error: ip-26-0-165-213: task 14: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 226056 got signal: 15
+srun: error: ip-26-0-165-59: task 10: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 217060 got signal: 15
+srun: error: ip-26-0-164-75: task 4: Exited with exit code 1
+srun: error: ip-26-0-165-38: task 9: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 359952 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 211419 got signal: 15
+srun: error: ip-26-0-165-164: task 12: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 410317 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 226664 got signal: 15
+srun: error: ip-26-0-165-24: task 8: Exited with exit code 1
+srun: error: ip-26-0-165-202: task 13: Exited with exit code 1
+srun: error: ip-26-0-166-125: task 17: Exited with exit code 1
+srun: error: ip-26-0-164-236: task 7: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 144935 got signal: 15
+srun: error: ip-26-0-164-207: task 6: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 356510 got signal: 15
+srun: error: ip-26-0-164-45: task 3: Exited with exit code 1
+srun: Force Terminated StepId=14019308.0
diff --git a/logs/14019312-bench_3.57G_dp4_tp32_pp2_acc16_mbs4_seq4096_zero1_tpmodeRED_vocab131k.out b/logs/14019312-bench_3.57G_dp4_tp32_pp2_acc16_mbs4_seq4096_zero1_tpmodeRED_vocab131k.out
new file mode 100644
index 0000000000000000000000000000000000000000..77e413e4e31e459bde93b80d0b149877c3874185
--- /dev/null
+++ b/logs/14019312-bench_3.57G_dp4_tp32_pp2_acc16_mbs4_seq4096_zero1_tpmodeRED_vocab131k.out
@@ -0,0 +1,3275 @@
++ '[' -z 14019312 ']'
++ source /etc/profile.d/modules.sh
+++ . /usr/share/modules/init/bash
++++ unset _mlshdbg
++++ '[' 0 = 1 ']'
++++ unset _mlre _mlIFS
++++ '[' -n x ']'
++++ _mlIFS=' 	
+'
++++ IFS=' '
++++ '[' -n '' ']'
+++++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash autoinit
++++ _mlcode='module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+MODULEPATH_modshare=/etc/environment-modules/modules:1:/usr/share/modules/\$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1; export MODULEPATH_modshare;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+LOADEDMODULES=; export LOADEDMODULES;
+MODULEPATH=/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/\$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles; export MODULEPATH;
+test 0;'
++++ _mlret=0
++++ '[' -n x ']'
++++ IFS=' 	
+'
++++ unset _mlIFS
++++ unset _mlre _mlv _mlrv
++++ '[' 0 -eq 0 ']'
++++ eval 'module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+MODULEPATH_modshare=/etc/environment-modules/modules:1:/usr/share/modules/\$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1; export MODULEPATH_modshare;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+LOADEDMODULES=; export LOADEDMODULES;
+MODULEPATH=/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/\$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles; export MODULEPATH;
+test 0;'
+++++ MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl
+++++ export MODULES_CMD
+++++ ENV=/usr/share/modules/init/profile.sh
+++++ export ENV
+++++ MODULEPATH_modshare='/etc/environment-modules/modules:1:/usr/share/modules/$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1'
+++++ export MODULEPATH_modshare
+++++ BASH_ENV=/usr/share/modules/init/bash
+++++ export BASH_ENV
+++++ MODULESHOME=/usr/share/modules
+++++ export MODULESHOME
+++++ LOADEDMODULES=
+++++ export LOADEDMODULES
+++++ MODULEPATH='/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles'
+++++ export MODULEPATH
+++++ test 0
++++ '[' 0 = 1 ']'
++++ '[' -t 2 ']'
++++ export -f module
++++ export -f switchml
++++ '[' 5 -ge 3 ']'
++++ [[ ehxB =~ i ]]
++++ [[ ! :/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin: =~ :/usr/bin: ]]
++++ '[' '!' -n '' ']'
++++ MANPATH=:
++++ export MANPATH
+++++ manpath
++++ [[ ! :/fsx/nouamane/miniconda/envs/2-1-cu121/man:/fsx/nouamane/miniconda/envs/2-1-cu121/share/man:/fsx/nouamane/miniconda/man:/fsx/nouamane/miniconda/share/man:/usr/local/man:/usr/local/share/man:/usr/share/man:/opt/slurm/share/man:/admin/home/nouamane/.local/share/man:/admin/home/nouamane/.fzf/man:: =~ :/usr/share/man: ]]
++++ unset _mlcode _mlret
++++ '[' -n '' ']'
++ module load cuda/12.1
++ unset _mlshdbg
++ '[' 0 = 1 ']'
++ unset _mlre _mlIFS
++ '[' -n x ']'
++ _mlIFS=' 	
+'
++ IFS=' '
++ '[' -n '' ']'
+++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash load cuda/12.1
++ eval 'CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include;' export 'CPATH;
+LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:;' export 'LD_LIBRARY_PATH;
+FI_EFA_FORK_SAFE=1;' export 'FI_EFA_FORK_SAFE;
+MANPATH=/usr/local/cuda-12.1/share/man::;' export 'MANPATH;
+LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64;' export 'LIBRARY_PATH;
+_LMFILES_=/usr/share/modules/modulefiles/cuda/12.1;' export '_LMFILES_;
+LOADEDMODULES=cuda/12.1;' export 'LOADEDMODULES;
+MPI_PATH=/opt/amazon/openmpi;' export 'MPI_PATH;
+NCCL_HOME_modshare=/opt/nccl/build:1;' export 'NCCL_HOME_modshare;
+NCCL_PROTO=simple;' export 'NCCL_PROTO;
+MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1;' export 'MANPATH_modshare;
+LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1;' export 'LIBRARY_PATH_modshare;
+NCCL_SOCKET_IFNAME=enp;' export 'NCCL_SOCKET_IFNAME;
+AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl;' export 'AWS_OFI_NCCL_HOME;
+NCCL_HOME=/opt/nccl/build;' export 'NCCL_HOME;
+FI_PROVIDER=efa;' export 'FI_PROVIDER;
+AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1;' export 'AWS_OFI_NCCL_HOME_modshare;
+CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1;' export 'CPATH_modshare;
+LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2::1:/opt/nccl/build/lib:1:/opt/aws-ofi-nccl/lib:2:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/efa/lib:1;' export 'LD_LIBRARY_PATH_modshare;
+FI_EFA_ENABLE_SHM_TRANSFER=1;' export 'FI_EFA_ENABLE_SHM_TRANSFER;
+_LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1;' export '_LMFILES__modshare;
+LOADEDMODULES_modshare=cuda/12.1:1;' export 'LOADEDMODULES_modshare;
+MPI_PATH_modshare=/opt/amazon/openmpi:1;' export 'MPI_PATH_modshare;
+PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin;' export 'PATH;
+CUDA_HOME=/usr/local/cuda-12.1;' export 'CUDA_HOME;
+PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/opt/slurm/bin:1:/usr/sbin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1;' export 'PATH_modshare;
+test' '0;'
+++ CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include
+++ export CPATH
+++ LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:
+++ export LD_LIBRARY_PATH
+++ FI_EFA_FORK_SAFE=1
+++ export FI_EFA_FORK_SAFE
+++ MANPATH=/usr/local/cuda-12.1/share/man::
+++ export MANPATH
+++ LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64
+++ export LIBRARY_PATH
+++ _LMFILES_=/usr/share/modules/modulefiles/cuda/12.1
+++ export _LMFILES_
+++ LOADEDMODULES=cuda/12.1
+++ export LOADEDMODULES
+++ MPI_PATH=/opt/amazon/openmpi
+++ export MPI_PATH
+++ NCCL_HOME_modshare=/opt/nccl/build:1
+++ export NCCL_HOME_modshare
+++ NCCL_PROTO=simple
+++ export NCCL_PROTO
+++ MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1
+++ export MANPATH_modshare
+++ LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1
+++ export LIBRARY_PATH_modshare
+++ NCCL_SOCKET_IFNAME=enp
+++ export NCCL_SOCKET_IFNAME
+++ AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl
+++ export AWS_OFI_NCCL_HOME
+++ NCCL_HOME=/opt/nccl/build
+++ export NCCL_HOME
+++ FI_PROVIDER=efa
+++ export FI_PROVIDER
+++ AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1
+++ export AWS_OFI_NCCL_HOME_modshare
+++ CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1
+++ export CPATH_modshare
+++ LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2::1:/opt/nccl/build/lib:1:/opt/aws-ofi-nccl/lib:2:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/efa/lib:1
+++ export LD_LIBRARY_PATH_modshare
+++ FI_EFA_ENABLE_SHM_TRANSFER=1
+++ export FI_EFA_ENABLE_SHM_TRANSFER
+++ _LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1
+++ export _LMFILES__modshare
+++ LOADEDMODULES_modshare=cuda/12.1:1
+++ export LOADEDMODULES_modshare
+++ MPI_PATH_modshare=/opt/amazon/openmpi:1
+++ export MPI_PATH_modshare
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
+++ export PATH
+++ CUDA_HOME=/usr/local/cuda-12.1
+++ export CUDA_HOME
+++ PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/opt/slurm/bin:1:/usr/sbin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1
+++ export PATH_modshare
+++ test 0
++ _mlstatus=0
++ '[' -n x ']'
++ IFS=' 	
+'
++ unset _mlre _mlv _mlrv _mlIFS
++ '[' -n '' ']'
++ unset _mlshdbg
++ return 0
++ source /fsx/nouamane/miniconda/bin/activate
+++ _CONDA_ROOT=/fsx/nouamane/miniconda
+++ . /fsx/nouamane/miniconda/etc/profile.d/conda.sh
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ '[' -z x ']'
+++ conda activate
+++ local cmd=activate
+++ case "$cmd" in
+++ __conda_activate activate
+++ '[' -n '' ']'
+++ local ask_conda
++++ PS1=
++++ __conda_exe shell.posix activate
++++ /fsx/nouamane/miniconda/bin/conda shell.posix activate
+Error while loading conda entry point: conda-libmamba-solver (module 'libmambapy' has no attribute 'QueryFormat')
+++ ask_conda='. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''2'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_1='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/etc/conda/activate.d/libxml2_activate.sh"'
+++ eval '. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''2'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_1='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/etc/conda/activate.d/libxml2_activate.sh"'
++++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh
+++++ test -n ''
+++++ unset XML_CATALOG_FILES
+++++ unset xml_catalog_files_libxml2
++++ PS1='(base) '
++++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
++++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
++++ export CONDA_PREFIX=/fsx/nouamane/miniconda
++++ CONDA_PREFIX=/fsx/nouamane/miniconda
++++ export CONDA_SHLVL=2
++++ CONDA_SHLVL=2
++++ export CONDA_DEFAULT_ENV=base
++++ CONDA_DEFAULT_ENV=base
++++ export 'CONDA_PROMPT_MODIFIER=(base) '
++++ CONDA_PROMPT_MODIFIER='(base) '
++++ export CONDA_PREFIX_1=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ CONDA_PREFIX_1=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ . /fsx/nouamane/miniconda/etc/conda/activate.d/libxml2_activate.sh
+++++ test -n ''
+++++ xml_catalog_files_libxml2=
+++++ XML_CATALOG_FILES=
+++++ conda_catalog_files=
+++++ ifs_libxml2=' 	
+'
+++++ IFS=' '
+++++ rem=/fsx/nouamane/miniconda
+++++ for pre in ${rem}
+++++ test '' = /fsx/nouamane/miniconda
+++++ conda_catalog_files=/fsx/nouamane/miniconda
+++++ rem=
+++++ IFS=' 	
+'
+++++ conda_catalog_files='file:///fsx/nouamane/miniconda/etc/xml/catalog file:///etc/xml/catalog'
+++++ export 'XML_CATALOG_FILES=file:///fsx/nouamane/miniconda/etc/xml/catalog file:///etc/xml/catalog'
+++++ XML_CATALOG_FILES='file:///fsx/nouamane/miniconda/etc/xml/catalog file:///etc/xml/catalog'
+++++ unset conda_catalog_files ifs_libxml2 rem
+++ __conda_hashr
+++ '[' -n '' ']'
+++ '[' -n '' ']'
+++ hash -r
++ conda activate 2-1-cu121
++ local cmd=activate
++ case "$cmd" in
++ __conda_activate activate 2-1-cu121
++ '[' -n '' ']'
++ local ask_conda
+++ PS1='(base) '
+++ __conda_exe shell.posix activate 2-1-cu121
+++ /fsx/nouamane/miniconda/bin/conda shell.posix activate 2-1-cu121
+Error while loading conda entry point: conda-libmamba-solver (module 'libmambapy' has no attribute 'QueryFormat')
++ ask_conda='. "/fsx/nouamane/miniconda/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_2='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
++ eval '. "/fsx/nouamane/miniconda/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_2='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
+++ . /fsx/nouamane/miniconda/etc/conda/deactivate.d/libxml2_deactivate.sh
++++ test -n ''
++++ unset XML_CATALOG_FILES
++++ unset xml_catalog_files_libxml2
+++ PS1='(2-1-cu121) '
+++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
+++ export CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ export CONDA_SHLVL=3
+++ CONDA_SHLVL=3
+++ export CONDA_DEFAULT_ENV=2-1-cu121
+++ CONDA_DEFAULT_ENV=2-1-cu121
+++ export 'CONDA_PROMPT_MODIFIER=(2-1-cu121) '
+++ CONDA_PROMPT_MODIFIER='(2-1-cu121) '
+++ export CONDA_PREFIX_2=/fsx/nouamane/miniconda
+++ CONDA_PREFIX_2=/fsx/nouamane/miniconda
+++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ export _CE_M=
+++ _CE_M=
+++ export _CE_CONDA=
+++ _CE_CONDA=
+++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh
++++ test -n ''
++++ xml_catalog_files_libxml2=
++++ XML_CATALOG_FILES=
++++ conda_catalog_files=
++++ ifs_libxml2=' 	
+'
++++ IFS=' '
++++ rem=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ for pre in ${rem}
++++ test '' = /fsx/nouamane/miniconda/envs/2-1-cu121
++++ conda_catalog_files=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ rem=
++++ IFS=' 	
+'
++++ conda_catalog_files='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ export 'XML_CATALOG_FILES=file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ XML_CATALOG_FILES='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ unset conda_catalog_files ifs_libxml2 rem
++ __conda_hashr
++ '[' -n '' ']'
++ '[' -n '' ']'
++ hash -r
++ export PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
++ PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
++ '[' -z '' ']'
+++ scontrol show hostnames 'ip-26-0-163-236,ip-26-0-164-[0,18,45,75,187,207,236],ip-26-0-165-[24,38,59,131,164,202,213],ip-26-0-166-[15,36,125],ip-26-0-169-[139,207,239,247],ip-26-0-170-[31,132,143,160],ip-26-0-171-21,ip-26-0-172-252,ip-26-0-173-[7,121,202,246]'
++ export 'NODELIST=ip-26-0-163-236
+ip-26-0-164-0
+ip-26-0-164-18
+ip-26-0-164-45
+ip-26-0-164-75
+ip-26-0-164-187
+ip-26-0-164-207
+ip-26-0-164-236
+ip-26-0-165-24
+ip-26-0-165-38
+ip-26-0-165-59
+ip-26-0-165-131
+ip-26-0-165-164
+ip-26-0-165-202
+ip-26-0-165-213
+ip-26-0-166-15
+ip-26-0-166-36
+ip-26-0-166-125
+ip-26-0-169-139
+ip-26-0-169-207
+ip-26-0-169-239
+ip-26-0-169-247
+ip-26-0-170-31
+ip-26-0-170-132
+ip-26-0-170-143
+ip-26-0-170-160
+ip-26-0-171-21
+ip-26-0-172-252
+ip-26-0-173-7
+ip-26-0-173-121
+ip-26-0-173-202
+ip-26-0-173-246'
++ NODELIST='ip-26-0-163-236
+ip-26-0-164-0
+ip-26-0-164-18
+ip-26-0-164-45
+ip-26-0-164-75
+ip-26-0-164-187
+ip-26-0-164-207
+ip-26-0-164-236
+ip-26-0-165-24
+ip-26-0-165-38
+ip-26-0-165-59
+ip-26-0-165-131
+ip-26-0-165-164
+ip-26-0-165-202
+ip-26-0-165-213
+ip-26-0-166-15
+ip-26-0-166-36
+ip-26-0-166-125
+ip-26-0-169-139
+ip-26-0-169-207
+ip-26-0-169-239
+ip-26-0-169-247
+ip-26-0-170-31
+ip-26-0-170-132
+ip-26-0-170-143
+ip-26-0-170-160
+ip-26-0-171-21
+ip-26-0-172-252
+ip-26-0-173-7
+ip-26-0-173-121
+ip-26-0-173-202
+ip-26-0-173-246'
+++ head -n1
+++ scontrol show hostnames 'ip-26-0-163-236,ip-26-0-164-[0,18,45,75,187,207,236],ip-26-0-165-[24,38,59,131,164,202,213],ip-26-0-166-[15,36,125],ip-26-0-169-[139,207,239,247],ip-26-0-170-[31,132,143,160],ip-26-0-171-21,ip-26-0-172-252,ip-26-0-173-[7,121,202,246]'
++ export MASTER_NODE=ip-26-0-163-236
++ MASTER_NODE=ip-26-0-163-236
++ export MASTER_PORT=12356
++ MASTER_PORT=12356
++ export NNODES=32
++ NNODES=32
++ export GPUS_PER_NODE=8
++ GPUS_PER_NODE=8
++ export WORLD_SIZE=256
++ WORLD_SIZE=256
++ export CUDA_DEVICE_MAX_CONNECTIONS=1
++ CUDA_DEVICE_MAX_CONNECTIONS=1
++ export NCCL_DEBUG=WARN
++ NCCL_DEBUG=WARN
++ export NANOTRON_BENCHMARK=1
++ NANOTRON_BENCHMARK=1
++ export WANDB_MODE=disabled
++ WANDB_MODE=disabled
++ export TORCH_NCCL_ASYNC_ERROR_HANDLING=1
++ TORCH_NCCL_ASYNC_ERROR_HANDLING=1
++ export TORCH_DISTRIBUTED_DEBUG=DETAIL
++ TORCH_DISTRIBUTED_DEBUG=DETAIL
++ '[' -z '' ']'
++ echo '=== GPU Topology ==='
+=== GPU Topology ===
++ nvidia-smi topo -m
+	GPU0	GPU1	GPU2	GPU3	GPU4	GPU5	GPU6	GPU7	CPU Affinity	NUMA Affinity	GPU NUMA ID
+GPU0	 X 	NV18	NV18	NV18	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU1	NV18	 X 	NV18	NV18	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU2	NV18	NV18	 X 	NV18	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU3	NV18	NV18	NV18	 X 	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU4	NV18	NV18	NV18	NV18	 X 	NV18	NV18	NV18	48-95	1		N/A
+GPU5	NV18	NV18	NV18	NV18	NV18	 X 	NV18	NV18	48-95	1		N/A
+GPU6	NV18	NV18	NV18	NV18	NV18	NV18	 X 	NV18	48-95	1		N/A
+GPU7	NV18	NV18	NV18	NV18	NV18	NV18	NV18	 X 	48-95	1		N/A
+
+Legend:
+
+  X    = Self
+  SYS  = Connection traversing PCIe as well as the SMP interconnect between NUMA nodes (e.g., QPI/UPI)
+  NODE = Connection traversing PCIe as well as the interconnect between PCIe Host Bridges within a NUMA node
+  PHB  = Connection traversing PCIe as well as a PCIe Host Bridge (typically the CPU)
+  PXB  = Connection traversing multiple PCIe bridges (without traversing the PCIe Host Bridge)
+  PIX  = Connection traversing at most a single PCIe bridge
+  NV#  = Connection traversing a bonded set of # NVLinks
++ echo ==================
+==================
++ export SRUN_ALLOC_ARGS=
++ SRUN_ALLOC_ARGS=
++ echo 'Master node: ip-26-0-163-236'
+Master node: ip-26-0-163-236
++ echo 'All nodes: ip-26-0-163-236
+ip-26-0-164-0
+ip-26-0-164-18
+ip-26-0-164-45
+ip-26-0-164-75
+ip-26-0-164-187
+ip-26-0-164-207
+ip-26-0-164-236
+ip-26-0-165-24
+ip-26-0-165-38
+ip-26-0-165-59
+ip-26-0-165-131
+ip-26-0-165-164
+ip-26-0-165-202
+ip-26-0-165-213
+ip-26-0-166-15
+ip-26-0-166-36
+ip-26-0-166-125
+ip-26-0-169-139
+ip-26-0-169-207
+ip-26-0-169-239
+ip-26-0-169-247
+ip-26-0-170-31
+ip-26-0-170-132
+ip-26-0-170-143
+ip-26-0-170-160
+ip-26-0-171-21
+ip-26-0-172-252
+ip-26-0-173-7
+ip-26-0-173-121
+ip-26-0-173-202
+ip-26-0-173-246'
+All nodes: ip-26-0-163-236
+ip-26-0-164-0
+ip-26-0-164-18
+ip-26-0-164-45
+ip-26-0-164-75
+ip-26-0-164-187
+ip-26-0-164-207
+ip-26-0-164-236
+ip-26-0-165-24
+ip-26-0-165-38
+ip-26-0-165-59
+ip-26-0-165-131
+ip-26-0-165-164
+ip-26-0-165-202
+ip-26-0-165-213
+ip-26-0-166-15
+ip-26-0-166-36
+ip-26-0-166-125
+ip-26-0-169-139
+ip-26-0-169-207
+ip-26-0-169-239
+ip-26-0-169-247
+ip-26-0-170-31
+ip-26-0-170-132
+ip-26-0-170-143
+ip-26-0-170-160
+ip-26-0-171-21
+ip-26-0-172-252
+ip-26-0-173-7
+ip-26-0-173-121
+ip-26-0-173-202
+ip-26-0-173-246
++ echo 'World size: 256'
+World size: 256
++ '[' -n '' ']'
++ srun --wait=0 --kill-on-bad-exit=1 torchrun --nnodes=32 --nproc_per_node=8 --rdzv_id=14019312 --rdzv_backend=c10d --rdzv_endpoint=ip-26-0-163-236:12356 --max_restarts 0 --rdzv_conf timeout=60 /fsx/nouamane/projects/nanotron/run_train.py --config-file benchmark/configs/config_3.57G_dp4_tp32_pp2_acc16_mbs4_seq4096_zero1_tpmodeRED_vocab131k.yaml
+[2025-01-03 08:08:09,632] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 08:08:09,632] torch.distributed.run: [WARNING] 
+[2025-01-03 08:08:09,632] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 08:08:09,632] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 08:08:09,632] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 08:08:09,632] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 08:08:09,632] torch.distributed.run: [WARNING] 
+[2025-01-03 08:08:09,632] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 08:08:09,632] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 08:08:09,632] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 08:08:09,632] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 08:08:09,632] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 08:08:09,633] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 08:08:09,632] torch.distributed.run: [WARNING] 
+[2025-01-03 08:08:09,632] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 08:08:09,632] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 08:08:09,632] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 08:08:09,632] torch.distributed.run: [WARNING] 
+[2025-01-03 08:08:09,632] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 08:08:09,632] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 08:08:09,632] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 08:08:09,633] torch.distributed.run: [WARNING] 
+[2025-01-03 08:08:09,633] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 08:08:09,633] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 08:08:09,633] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 08:08:09,633] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 08:08:09,633] torch.distributed.run: [WARNING] 
+[2025-01-03 08:08:09,633] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 08:08:09,633] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 08:08:09,633] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 08:08:09,635] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 08:08:09,635] torch.distributed.run: [WARNING] 
+[2025-01-03 08:08:09,635] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 08:08:09,635] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 08:08:09,635] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 08:08:09,636] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 08:08:09,636] torch.distributed.run: [WARNING] 
+[2025-01-03 08:08:09,636] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 08:08:09,636] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 08:08:09,636] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 08:08:09,635] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 08:08:09,636] torch.distributed.run: [WARNING] 
+[2025-01-03 08:08:09,636] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 08:08:09,636] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 08:08:09,636] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 08:08:09,637] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 08:08:09,639] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 08:08:09,639] torch.distributed.run: [WARNING] 
+[2025-01-03 08:08:09,639] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 08:08:09,639] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 08:08:09,639] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 08:08:09,640] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 08:08:09,640] torch.distributed.run: [WARNING] 
+[2025-01-03 08:08:09,640] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 08:08:09,640] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 08:08:09,640] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 08:08:09,638] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 08:08:09,638] torch.distributed.run: [WARNING] 
+[2025-01-03 08:08:09,638] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 08:08:09,638] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 08:08:09,638] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 08:08:09,638] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 08:08:09,639] torch.distributed.run: [WARNING] 
+[2025-01-03 08:08:09,639] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 08:08:09,639] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 08:08:09,639] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 08:08:09,639] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 08:08:09,638] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 08:08:09,638] torch.distributed.run: [WARNING] 
+[2025-01-03 08:08:09,638] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 08:08:09,638] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 08:08:09,638] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 08:08:09,640] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 08:08:09,640] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 08:08:09,640] torch.distributed.run: [WARNING] 
+[2025-01-03 08:08:09,640] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 08:08:09,640] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 08:08:09,640] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 08:08:09,640] torch.distributed.run: [WARNING] 
+[2025-01-03 08:08:09,640] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 08:08:09,640] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 08:08:09,640] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 08:08:09,639] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 08:08:09,639] torch.distributed.run: [WARNING] 
+[2025-01-03 08:08:09,639] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 08:08:09,639] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 08:08:09,639] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 08:08:09,640] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 08:08:09,640] torch.distributed.run: [WARNING] 
+[2025-01-03 08:08:09,640] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 08:08:09,640] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 08:08:09,640] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 08:08:09,641] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 08:08:09,641] torch.distributed.run: [WARNING] 
+[2025-01-03 08:08:09,641] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 08:08:09,641] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 08:08:09,641] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 08:08:09,642] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 08:08:09,642] torch.distributed.run: [WARNING] 
+[2025-01-03 08:08:09,642] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 08:08:09,642] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 08:08:09,642] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 08:08:09,643] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 08:08:09,643] torch.distributed.run: [WARNING] 
+[2025-01-03 08:08:09,643] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 08:08:09,643] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 08:08:09,643] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 08:08:09,642] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 08:08:09,642] torch.distributed.run: [WARNING] 
+[2025-01-03 08:08:09,642] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 08:08:09,642] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 08:08:09,642] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 08:08:09,644] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 08:08:09,644] torch.distributed.run: [WARNING] 
+[2025-01-03 08:08:09,644] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 08:08:09,644] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 08:08:09,644] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 08:08:09,643] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 08:08:09,643] torch.distributed.run: [WARNING] 
+[2025-01-03 08:08:09,643] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 08:08:09,643] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 08:08:09,643] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 08:08:09,645] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 08:08:09,645] torch.distributed.run: [WARNING] 
+[2025-01-03 08:08:09,645] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 08:08:09,645] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 08:08:09,645] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 08:08:09,645] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 08:08:09,645] torch.distributed.run: [WARNING] 
+[2025-01-03 08:08:09,645] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 08:08:09,645] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 08:08:09,645] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 08:08:09,639] torch.distributed.run: [WARNING] 
+[2025-01-03 08:08:09,639] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 08:08:09,639] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 08:08:09,639] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 08:08:09,637] torch.distributed.run: [WARNING] 
+[2025-01-03 08:08:09,637] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 08:08:09,637] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 08:08:09,637] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 08:08:09,719] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 08:08:09,719] torch.distributed.run: [WARNING] 
+[2025-01-03 08:08:09,719] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 08:08:09,719] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 08:08:09,719] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 08:08:09,731] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 08:08:09,731] torch.distributed.run: [WARNING] 
+[2025-01-03 08:08:09,731] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 08:08:09,731] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 08:08:09,731] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 08:08:09,910] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 08:08:09,952] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 08:08:09,952] torch.distributed.run: [WARNING] 
+[2025-01-03 08:08:09,952] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 08:08:09,952] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 08:08:09,952] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 08:08:09,910] torch.distributed.run: [WARNING] 
+[2025-01-03 08:08:09,910] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 08:08:09,910] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 08:08:09,910] torch.distributed.run: [WARNING] *****************************************
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+01/03/2025 08:09:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: Config:
+01/03/2025 08:09:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: Config(general=GeneralArgs(project='debug',
+01/03/2025 08:09:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                            run='3.57G_dp4_tp32_pp2_acc16_mbs4_seq4096_zero1_tpmodeRED_vocab131k',
+01/03/2025 08:09:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                            seed=42,
+01/03/2025 08:09:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                            step=None,
+01/03/2025 08:09:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                            consumed_train_samples=None,
+01/03/2025 08:09:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                            benchmark_csv_path=PosixPath('benchmark/results/bench_final2.csv'),
+01/03/2025 08:09:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                            ignore_sanity_checks=True),
+01/03/2025 08:09:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:        parallelism=ParallelismArgs(dp=4,
+01/03/2025 08:09:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                    pp=2,
+01/03/2025 08:09:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                    tp=32,
+01/03/2025 08:09:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                    pp_engine=<nanotron.parallel.pipeline_parallel.engine.OneForwardOneBackwardPipelineEngine object at 0x7f6b8e664dc0>,
+01/03/2025 08:09:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                    tp_mode=<TensorParallelLinearMode.REDUCE_SCATTER: 2>,
+01/03/2025 08:09:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                    tp_linear_async_communication=True,
+01/03/2025 08:09:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                    recompute_layer=False,
+01/03/2025 08:09:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                    tp_recompute_allgather=True,
+01/03/2025 08:09:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                    expert_parallel_size=1),
+01/03/2025 08:09:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:        model=ModelArgs(model_config=LlamaConfig(bos_token_id=0,
+01/03/2025 08:09:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                                 eos_token_id=0,
+01/03/2025 08:09:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                                 hidden_act='silu',
+01/03/2025 08:09:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                                 hidden_size=3072,
+01/03/2025 08:09:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                                 initializer_range=0.02,
+01/03/2025 08:09:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                                 intermediate_size=8192,
+01/03/2025 08:09:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                                 is_llama_config=True,
+01/03/2025 08:09:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                                 max_position_embeddings=4096,
+01/03/2025 08:09:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                                 num_attention_heads=32,
+01/03/2025 08:09:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                                 num_hidden_layers=28,
+01/03/2025 08:09:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                                 num_key_value_heads=32,
+01/03/2025 08:09:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                                 pad_token_id=None,
+01/03/2025 08:09:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                                 pretraining_tp=1,
+01/03/2025 08:09:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                                 rms_norm_eps=1e-05,
+01/03/2025 08:09:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                                 rope_scaling=None,
+01/03/2025 08:09:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                                 rope_theta=10000.0,
+01/03/2025 08:09:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                                 rope_interleaved=False,
+01/03/2025 08:09:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                                 tie_word_embeddings=True,
+01/03/2025 08:09:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                                 use_cache=True,
+01/03/2025 08:09:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                                 vocab_size=131072),
+01/03/2025 08:09:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                        init_method=RandomInit(std=0.02),
+01/03/2025 08:09:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                        dtype=torch.bfloat16,
+01/03/2025 08:09:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                        make_vocab_size_divisible_by=1,
+01/03/2025 08:09:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                        ddp_bucket_cap_mb=25),
+01/03/2025 08:09:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:        tokenizer=TokenizerArgs(tokenizer_name_or_path='robot-test/dummy-tokenizer-wordlevel',
+01/03/2025 08:09:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                tokenizer_revision=None,
+01/03/2025 08:09:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                tokenizer_max_length=None),
+01/03/2025 08:09:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:        checkpoints=CheckpointsArgs(checkpoints_path=PosixPath('checkpoints'),
+01/03/2025 08:09:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                    checkpoint_interval=10000,
+01/03/2025 08:09:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                    save_initial_state=False,
+01/03/2025 08:09:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                    save_final_state=False,
+01/03/2025 08:09:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                    resume_checkpoint_path=None,
+01/03/2025 08:09:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                    checkpoints_path_is_shared_file_system=False),
+01/03/2025 08:09:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:        logging=LoggingArgs(log_level='info',
+01/03/2025 08:09:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                            log_level_replica='info',
+01/03/2025 08:09:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                            iteration_step_info_interval=1),
+01/03/2025 08:09:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:        tokens=TokensArgs(sequence_length=4096,
+01/03/2025 08:09:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                          train_steps=100,
+01/03/2025 08:09:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                          micro_batch_size=4,
+01/03/2025 08:09:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                          batch_accumulation_per_replica=16,
+01/03/2025 08:09:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                          val_check_interval=100,
+01/03/2025 08:09:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                          limit_val_batches=0,
+01/03/2025 08:09:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                          limit_test_batches=0),
+01/03/2025 08:09:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:        optimizer=OptimizerArgs(optimizer_factory=AdamWOptimizerArgs(adam_eps=1e-08,
+01/03/2025 08:09:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                                                     adam_beta1=0.9,
+01/03/2025 08:09:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                                                     adam_beta2=0.95,
+01/03/2025 08:09:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                                                     torch_adam_is_fused=True,
+01/03/2025 08:09:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                                                     name='adamW'),
+01/03/2025 08:09:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                zero_stage=1,
+01/03/2025 08:09:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                weight_decay=0.01,
+01/03/2025 08:09:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                clip_grad=1.0,
+01/03/2025 08:09:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                accumulate_grad_in_fp32=True,
+01/03/2025 08:09:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                learning_rate_scheduler=LRSchedulerArgs(learning_rate=0.0003,
+01/03/2025 08:09:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                                                        lr_warmup_steps=2,
+01/03/2025 08:09:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                                                        lr_warmup_style='linear',
+01/03/2025 08:09:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                                                        lr_decay_style='cosine',
+01/03/2025 08:09:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                                                        lr_decay_steps=13,
+01/03/2025 08:09:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                                                        lr_decay_starting_step=None,
+01/03/2025 08:09:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                                                        min_decay_lr=1e-05)),
+01/03/2025 08:09:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:        data_stages=[DatasetStageArgs(name='Stable Training Stage',
+01/03/2025 08:09:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                      start_training_step=1,
+01/03/2025 08:09:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                      data=DataArgs(dataset=None,
+01/03/2025 08:09:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                                    seed=42,
+01/03/2025 08:09:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                                    num_loading_workers=1))],
+01/03/2025 08:09:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:        profiler=None,
+01/03/2025 08:09:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:        lighteval=None,
+01/03/2025 08:09:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:        s3_upload=None)
+01/03/2025 08:09:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: Model Config:
+01/03/2025 08:09:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: LlamaConfig(bos_token_id=0,
+01/03/2025 08:09:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:             eos_token_id=0,
+01/03/2025 08:09:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:             hidden_act='silu',
+01/03/2025 08:09:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:             hidden_size=3072,
+01/03/2025 08:09:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:             initializer_range=0.02,
+01/03/2025 08:09:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:             intermediate_size=8192,
+01/03/2025 08:09:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:             is_llama_config=True,
+01/03/2025 08:09:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:             max_position_embeddings=4096,
+01/03/2025 08:09:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:             num_attention_heads=32,
+01/03/2025 08:09:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:             num_hidden_layers=28,
+01/03/2025 08:09:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:             num_key_value_heads=32,
+01/03/2025 08:09:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:             pad_token_id=None,
+01/03/2025 08:09:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:             pretraining_tp=1,
+01/03/2025 08:09:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:             rms_norm_eps=1e-05,
+01/03/2025 08:09:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:             rope_scaling=None,
+01/03/2025 08:09:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:             rope_theta=10000.0,
+01/03/2025 08:09:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:             rope_interleaved=False,
+01/03/2025 08:09:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:             tie_word_embeddings=True,
+01/03/2025 08:09:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:             use_cache=True,
+01/03/2025 08:09:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:             vocab_size=131072)
+01/03/2025 08:09:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: Building model..
+01/03/2025 08:09:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: Initialize RoPE Theta = 10000.0
+01/03/2025 08:09:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: Setting PP block ranks...
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+01/03/2025 08:09:19 [INFO|DP=0|PP=0|TP=2|ip-26-0-163-236]: Local number of parameters: 69.3M (132.19MiB)
+01/03/2025 08:09:19 [INFO|DP=0|PP=1|TP=0|ip-26-0-166-15]: Local number of parameters: 55.1M (105.15MiB)
+01/03/2025 08:09:19 [INFO|DP=0|PP=0|TP=2|ip-26-0-163-236]: [After model building] Memory usage: 132.21MiB. Peak allocated: 5384.00MiB Peak reserved: 10528.00MiB
+01/03/2025 08:09:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: Total number of parameters: 3.98G (7594.69MiB)
+01/03/2025 08:09:19 [INFO|DP=0|PP=1|TP=0|ip-26-0-166-15]: [After model building] Memory usage: 105.17MiB. Peak allocated: 5384.00MiB Peak reserved: 10524.00MiB
+01/03/2025 08:09:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: Local number of parameters: 69.3M (132.19MiB)
+01/03/2025 08:09:19 [INFO|DP=0|PP=0|TP=24|ip-26-0-164-187]: Local number of parameters: 69.3M (132.19MiB)
+01/03/2025 08:09:19 [INFO|DP=0|PP=0|TP=24|ip-26-0-164-187]: [After model building] Memory usage: 132.21MiB. Peak allocated: 5384.00MiB Peak reserved: 9504.00MiB
+01/03/2025 08:09:19 [INFO|DP=0|PP=1|TP=24|ip-26-0-169-207]: Local number of parameters: 55.1M (105.15MiB)
+01/03/2025 08:09:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: [After model building] Memory usage: 132.21MiB. Peak allocated: 5384.00MiB Peak reserved: 10528.00MiB
+01/03/2025 08:09:19 [INFO|DP=0|PP=0|TP=8|ip-26-0-164-0]: Local number of parameters: 69.3M (132.19MiB)
+01/03/2025 08:09:19 [INFO|DP=0|PP=0|TP=10|ip-26-0-164-0]: Local number of parameters: 69.3M (132.19MiB)
+01/03/2025 08:09:19 [INFO|DP=0|PP=0|TP=11|ip-26-0-164-0]: Local number of parameters: 69.3M (132.19MiB)
+01/03/2025 08:09:19 [INFO|DP=0|PP=1|TP=24|ip-26-0-169-207]: [After model building] Memory usage: 105.17MiB. Peak allocated: 5384.00MiB Peak reserved: 10524.00MiB
+01/03/2025 08:09:19 [INFO|DP=0|PP=0|TP=11|ip-26-0-164-0]: [After model building] Memory usage: 132.21MiB. Peak allocated: 5384.00MiB Peak reserved: 10528.00MiB
+01/03/2025 08:09:19 [INFO|DP=0|PP=0|TP=10|ip-26-0-164-0]: [After model building] Memory usage: 132.21MiB. Peak allocated: 5384.00MiB Peak reserved: 10528.00MiB
+01/03/2025 08:09:19 [INFO|DP=0|PP=0|TP=8|ip-26-0-164-0]: [After model building] Memory usage: 132.21MiB. Peak allocated: 5384.00MiB Peak reserved: 10528.00MiB
+01/03/2025 08:09:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: No checkpoint path provided.
+01/03/2025 08:09:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: Parametrizing model parameters using StandardParametrizator
+01/03/2025 08:09:19 [INFO|DP=0|PP=1|TP=1|ip-26-0-166-15]: Local number of parameters: 55.1M (105.15MiB)
+01/03/2025 08:09:19 [INFO|DP=0|PP=0|TP=3|ip-26-0-163-236]: Local number of parameters: 69.3M (132.19MiB)
+01/03/2025 08:09:19 [INFO|DP=0|PP=1|TP=2|ip-26-0-166-15]: Local number of parameters: 55.1M (105.15MiB)
+01/03/2025 08:09:19 [INFO|DP=0|PP=1|TP=3|ip-26-0-166-15]: Local number of parameters: 55.1M (105.15MiB)
+01/03/2025 08:09:19 [INFO|DP=0|PP=0|TP=6|ip-26-0-163-236]: Local number of parameters: 69.3M (132.19MiB)
+01/03/2025 08:09:19 [INFO|DP=0|PP=1|TP=6|ip-26-0-166-15]: Local number of parameters: 55.1M (105.15MiB)
+01/03/2025 08:09:19 [INFO|DP=0|PP=0|TP=1|ip-26-0-163-236]: Local number of parameters: 69.3M (132.19MiB)
+01/03/2025 08:09:19 [INFO|DP=0|PP=1|TP=1|ip-26-0-166-15]: [After model building] Memory usage: 105.17MiB. Peak allocated: 5384.00MiB Peak reserved: 10524.00MiB
+01/03/2025 08:09:19 [INFO|DP=0|PP=1|TP=3|ip-26-0-166-15]: [After model building] Memory usage: 105.17MiB. Peak allocated: 5384.00MiB Peak reserved: 10524.00MiB
+01/03/2025 08:09:19 [INFO|DP=0|PP=1|TP=2|ip-26-0-166-15]: [After model building] Memory usage: 105.17MiB. Peak allocated: 5384.00MiB Peak reserved: 10524.00MiB
+01/03/2025 08:09:19 [INFO|DP=0|PP=1|TP=5|ip-26-0-166-15]: Local number of parameters: 55.1M (105.15MiB)
+01/03/2025 08:09:19 [INFO|DP=0|PP=0|TP=7|ip-26-0-163-236]: Local number of parameters: 69.3M (132.19MiB)
+01/03/2025 08:09:19 [INFO|DP=0|PP=0|TP=3|ip-26-0-163-236]: [After model building] Memory usage: 132.21MiB. Peak allocated: 5384.00MiB Peak reserved: 10528.00MiB
+01/03/2025 08:09:19 [INFO|DP=0|PP=0|TP=6|ip-26-0-163-236]: [After model building] Memory usage: 132.21MiB. Peak allocated: 5384.00MiB Peak reserved: 10528.00MiB
+01/03/2025 08:09:19 [INFO|DP=0|PP=0|TP=5|ip-26-0-163-236]: Local number of parameters: 69.3M (132.19MiB)
+01/03/2025 08:09:19 [INFO|DP=0|PP=1|TP=6|ip-26-0-166-15]: [After model building] Memory usage: 105.17MiB. Peak allocated: 5384.00MiB Peak reserved: 10524.00MiB
+01/03/2025 08:09:19 [INFO|DP=0|PP=0|TP=1|ip-26-0-163-236]: [After model building] Memory usage: 132.21MiB. Peak allocated: 5384.00MiB Peak reserved: 10528.00MiB
+01/03/2025 08:09:19 [INFO|DP=0|PP=1|TP=5|ip-26-0-166-15]: [After model building] Memory usage: 105.17MiB. Peak allocated: 5384.00MiB Peak reserved: 10524.00MiB
+01/03/2025 08:09:19 [INFO|DP=0|PP=0|TP=7|ip-26-0-163-236]: [After model building] Memory usage: 132.21MiB. Peak allocated: 5384.00MiB Peak reserved: 10528.00MiB
+01/03/2025 08:09:19 [INFO|DP=0|PP=0|TP=5|ip-26-0-163-236]: [After model building] Memory usage: 132.21MiB. Peak allocated: 5384.00MiB Peak reserved: 10528.00MiB
+01/03/2025 08:09:19 [INFO|DP=0|PP=0|TP=19|ip-26-0-164-18]: Local number of parameters: 69.3M (132.19MiB)
+01/03/2025 08:09:19 [INFO|DP=0|PP=1|TP=7|ip-26-0-166-15]: Local number of parameters: 55.1M (105.15MiB)
+01/03/2025 08:09:19 [INFO|DP=0|PP=0|TP=9|ip-26-0-164-0]: Local number of parameters: 69.3M (132.19MiB)
+01/03/2025 08:09:19 [INFO|DP=0|PP=1|TP=9|ip-26-0-166-36]: Local number of parameters: 55.1M (105.15MiB)
+01/03/2025 08:09:19 [INFO|DP=0|PP=0|TP=19|ip-26-0-164-18]: [After model building] Memory usage: 132.21MiB. Peak allocated: 5384.00MiB Peak reserved: 10528.00MiB
+01/03/2025 08:09:19 [INFO|DP=0|PP=1|TP=9|ip-26-0-166-36]: [After model building] Memory usage: 105.17MiB. Peak allocated: 5384.00MiB Peak reserved: 10524.00MiB
+01/03/2025 08:09:19 [INFO|DP=0|PP=1|TP=7|ip-26-0-166-15]: [After model building] Memory usage: 105.17MiB. Peak allocated: 5384.00MiB Peak reserved: 10524.00MiB
+01/03/2025 08:09:19 [INFO|DP=0|PP=0|TP=9|ip-26-0-164-0]: [After model building] Memory usage: 132.21MiB. Peak allocated: 5384.00MiB Peak reserved: 10528.00MiB
+01/03/2025 08:09:19 [INFO|DP=0|PP=0|TP=4|ip-26-0-163-236]: Local number of parameters: 69.3M (132.19MiB)
+01/03/2025 08:09:19 [INFO|DP=0|PP=0|TP=16|ip-26-0-164-18]: Local number of parameters: 69.3M (132.19MiB)
+01/03/2025 08:09:19 [INFO|DP=0|PP=1|TP=19|ip-26-0-169-139]: Local number of parameters: 55.1M (105.15MiB)
+01/03/2025 08:09:19 [INFO|DP=0|PP=1|TP=16|ip-26-0-169-139]: Local number of parameters: 55.1M (105.15MiB)
+01/03/2025 08:09:19 [INFO|DP=0|PP=0|TP=4|ip-26-0-163-236]: [After model building] Memory usage: 132.21MiB. Peak allocated: 5384.00MiB Peak reserved: 10528.00MiB
+01/03/2025 08:09:19 [INFO|DP=0|PP=0|TP=16|ip-26-0-164-18]: [After model building] Memory usage: 132.21MiB. Peak allocated: 5384.00MiB Peak reserved: 10528.00MiB
+01/03/2025 08:09:19 [INFO|DP=0|PP=1|TP=19|ip-26-0-169-139]: [After model building] Memory usage: 105.17MiB. Peak allocated: 5384.00MiB Peak reserved: 10524.00MiB
+01/03/2025 08:09:19 [INFO|DP=0|PP=1|TP=11|ip-26-0-166-36]: Local number of parameters: 55.1M (105.15MiB)
+01/03/2025 08:09:19 [INFO|DP=0|PP=1|TP=16|ip-26-0-169-139]: [After model building] Memory usage: 105.17MiB. Peak allocated: 5384.00MiB Peak reserved: 10524.00MiB
+01/03/2025 08:09:19 [INFO|DP=0|PP=1|TP=11|ip-26-0-166-36]: [After model building] Memory usage: 105.17MiB. Peak allocated: 5384.00MiB Peak reserved: 10524.00MiB
+01/03/2025 08:09:19 [INFO|DP=0|PP=0|TP=27|ip-26-0-164-187]: Local number of parameters: 69.3M (132.19MiB)
+01/03/2025 08:09:19 [INFO|DP=0|PP=0|TP=27|ip-26-0-164-187]: [After model building] Memory usage: 132.21MiB. Peak allocated: 5384.00MiB Peak reserved: 9504.00MiB
+01/03/2025 08:09:19 [INFO|DP=0|PP=0|TP=26|ip-26-0-164-187]: Local number of parameters: 69.3M (132.19MiB)
+01/03/2025 08:09:19 [INFO|DP=0|PP=0|TP=25|ip-26-0-164-187]: Local number of parameters: 69.3M (132.19MiB)
+01/03/2025 08:09:19 [INFO|DP=0|PP=0|TP=31|ip-26-0-164-187]: Local number of parameters: 69.3M (132.19MiB)
+01/03/2025 08:09:19 [INFO|DP=0|PP=0|TP=17|ip-26-0-164-18]: Local number of parameters: 69.3M (132.19MiB)
+01/03/2025 08:09:19 [INFO|DP=0|PP=0|TP=13|ip-26-0-164-0]: Local number of parameters: 69.3M (132.19MiB)
+01/03/2025 08:09:19 [INFO|DP=0|PP=0|TP=26|ip-26-0-164-187]: [After model building] Memory usage: 132.21MiB. Peak allocated: 5384.00MiB Peak reserved: 9504.00MiB
+01/03/2025 08:09:19 [INFO|DP=0|PP=1|TP=22|ip-26-0-169-139]: Local number of parameters: 55.1M (105.15MiB)
+01/03/2025 08:09:19 [INFO|DP=0|PP=0|TP=22|ip-26-0-164-18]: Local number of parameters: 69.3M (132.19MiB)
+01/03/2025 08:09:19 [INFO|DP=0|PP=1|TP=14|ip-26-0-166-36]: Local number of parameters: 55.1M (105.15MiB)
+01/03/2025 08:09:19 [INFO|DP=0|PP=0|TP=14|ip-26-0-164-0]: Local number of parameters: 69.3M (132.19MiB)
+01/03/2025 08:09:19 [INFO|DP=0|PP=0|TP=15|ip-26-0-164-0]: Local number of parameters: 69.3M (132.19MiB)
+01/03/2025 08:09:19 [INFO|DP=0|PP=0|TP=12|ip-26-0-164-0]: Local number of parameters: 69.3M (132.19MiB)
+01/03/2025 08:09:19 [INFO|DP=0|PP=0|TP=28|ip-26-0-164-187]: Local number of parameters: 69.3M (132.19MiB)
+01/03/2025 08:09:19 [INFO|DP=0|PP=0|TP=31|ip-26-0-164-187]: [After model building] Memory usage: 132.21MiB. Peak allocated: 5384.00MiB Peak reserved: 9504.00MiB
+01/03/2025 08:09:19 [INFO|DP=0|PP=0|TP=25|ip-26-0-164-187]: [After model building] Memory usage: 132.21MiB. Peak allocated: 5384.00MiB Peak reserved: 9504.00MiB
+01/03/2025 08:09:19 [INFO|DP=0|PP=0|TP=18|ip-26-0-164-18]: Local number of parameters: 69.3M (132.19MiB)
+01/03/2025 08:09:19 [INFO|DP=0|PP=0|TP=21|ip-26-0-164-18]: Local number of parameters: 69.3M (132.19MiB)
+01/03/2025 08:09:19 [INFO|DP=0|PP=0|TP=23|ip-26-0-164-18]: Local number of parameters: 69.3M (132.19MiB)
+01/03/2025 08:09:19 [INFO|DP=0|PP=0|TP=20|ip-26-0-164-18]: Local number of parameters: 69.3M (132.19MiB)
+01/03/2025 08:09:19 [INFO|DP=0|PP=0|TP=13|ip-26-0-164-0]: [After model building] Memory usage: 132.21MiB. Peak allocated: 5384.00MiB Peak reserved: 10528.00MiB
+01/03/2025 08:09:19 [INFO|DP=0|PP=1|TP=25|ip-26-0-169-207]: Local number of parameters: 55.1M (105.15MiB)
+01/03/2025 08:09:19 [INFO|DP=0|PP=1|TP=22|ip-26-0-169-139]: [After model building] Memory usage: 105.17MiB. Peak allocated: 5384.00MiB Peak reserved: 10524.00MiB
+01/03/2025 08:09:19 [INFO|DP=0|PP=0|TP=14|ip-26-0-164-0]: [After model building] Memory usage: 132.21MiB. Peak allocated: 5384.00MiB Peak reserved: 10528.00MiB
+01/03/2025 08:09:19 [INFO|DP=0|PP=0|TP=22|ip-26-0-164-18]: [After model building] Memory usage: 132.21MiB. Peak allocated: 5384.00MiB Peak reserved: 10528.00MiB
+01/03/2025 08:09:19 [INFO|DP=0|PP=1|TP=14|ip-26-0-166-36]: [After model building] Memory usage: 105.17MiB. Peak allocated: 5384.00MiB Peak reserved: 10524.00MiB
+01/03/2025 08:09:19 [INFO|DP=0|PP=0|TP=15|ip-26-0-164-0]: [After model building] Memory usage: 132.21MiB. Peak allocated: 5384.00MiB Peak reserved: 10528.00MiB
+01/03/2025 08:09:19 [INFO|DP=0|PP=0|TP=28|ip-26-0-164-187]: [After model building] Memory usage: 132.21MiB. Peak allocated: 5384.00MiB Peak reserved: 9504.00MiB
+01/03/2025 08:09:19 [INFO|DP=0|PP=0|TP=12|ip-26-0-164-0]: [After model building] Memory usage: 132.21MiB. Peak allocated: 5384.00MiB Peak reserved: 10528.00MiB
+01/03/2025 08:09:19 [INFO|DP=0|PP=1|TP=26|ip-26-0-169-207]: Local number of parameters: 55.1M (105.15MiB)
+01/03/2025 08:09:19 [INFO|DP=0|PP=1|TP=31|ip-26-0-169-207]: Local number of parameters: 55.1M (105.15MiB)
+01/03/2025 08:09:19 [INFO|DP=0|PP=0|TP=17|ip-26-0-164-18]: [After model building] Memory usage: 132.21MiB. Peak allocated: 5384.00MiB Peak reserved: 10528.00MiB
+01/03/2025 08:09:19 [INFO|DP=0|PP=0|TP=21|ip-26-0-164-18]: [After model building] Memory usage: 132.21MiB. Peak allocated: 5384.00MiB Peak reserved: 10528.00MiB
+01/03/2025 08:09:19 [INFO|DP=0|PP=0|TP=18|ip-26-0-164-18]: [After model building] Memory usage: 132.21MiB. Peak allocated: 5384.00MiB Peak reserved: 10528.00MiB
+01/03/2025 08:09:19 [INFO|DP=0|PP=0|TP=23|ip-26-0-164-18]: [After model building] Memory usage: 132.21MiB. Peak allocated: 5384.00MiB Peak reserved: 10528.00MiB
+01/03/2025 08:09:19 [INFO|DP=0|PP=0|TP=20|ip-26-0-164-18]: [After model building] Memory usage: 132.21MiB. Peak allocated: 5384.00MiB Peak reserved: 10528.00MiB
+01/03/2025 08:09:19 [INFO|DP=0|PP=1|TP=25|ip-26-0-169-207]: [After model building] Memory usage: 105.17MiB. Peak allocated: 5384.00MiB Peak reserved: 10524.00MiB
+01/03/2025 08:09:19 [INFO|DP=0|PP=1|TP=13|ip-26-0-166-36]: Local number of parameters: 55.1M (105.15MiB)
+01/03/2025 08:09:19 [INFO|DP=0|PP=1|TP=10|ip-26-0-166-36]: Local number of parameters: 55.1M (105.15MiB)
+01/03/2025 08:09:19 [INFO|DP=0|PP=1|TP=18|ip-26-0-169-139]: Local number of parameters: 55.1M (105.15MiB)
+01/03/2025 08:09:19 [INFO|DP=0|PP=1|TP=31|ip-26-0-169-207]: [After model building] Memory usage: 105.17MiB. Peak allocated: 5384.00MiB Peak reserved: 10524.00MiB
+01/03/2025 08:09:19 [INFO|DP=0|PP=0|TP=30|ip-26-0-164-187]: Local number of parameters: 69.3M (132.19MiB)
+01/03/2025 08:09:19 [INFO|DP=0|PP=1|TP=8|ip-26-0-166-36]: Local number of parameters: 55.1M (105.15MiB)
+01/03/2025 08:09:19 [INFO|DP=0|PP=1|TP=21|ip-26-0-169-139]: Local number of parameters: 55.1M (105.15MiB)
+01/03/2025 08:09:19 [INFO|DP=0|PP=1|TP=26|ip-26-0-169-207]: [After model building] Memory usage: 105.17MiB. Peak allocated: 5384.00MiB Peak reserved: 10524.00MiB
+01/03/2025 08:09:19 [INFO|DP=0|PP=0|TP=29|ip-26-0-164-187]: Local number of parameters: 69.3M (132.19MiB)
+01/03/2025 08:09:19 [INFO|DP=0|PP=1|TP=15|ip-26-0-166-36]: Local number of parameters: 55.1M (105.15MiB)
+01/03/2025 08:09:19 [INFO|DP=0|PP=1|TP=23|ip-26-0-169-139]: Local number of parameters: 55.1M (105.15MiB)
+01/03/2025 08:09:19 [INFO|DP=0|PP=1|TP=10|ip-26-0-166-36]: [After model building] Memory usage: 105.17MiB. Peak allocated: 5384.00MiB Peak reserved: 10524.00MiB
+01/03/2025 08:09:19 [INFO|DP=0|PP=1|TP=13|ip-26-0-166-36]: [After model building] Memory usage: 105.17MiB. Peak allocated: 5384.00MiB Peak reserved: 10524.00MiB
+01/03/2025 08:09:19 [INFO|DP=0|PP=1|TP=20|ip-26-0-169-139]: Local number of parameters: 55.1M (105.15MiB)
+01/03/2025 08:09:19 [INFO|DP=0|PP=1|TP=18|ip-26-0-169-139]: [After model building] Memory usage: 105.17MiB. Peak allocated: 5384.00MiB Peak reserved: 10524.00MiB
+01/03/2025 08:09:19 [INFO|DP=0|PP=1|TP=21|ip-26-0-169-139]: [After model building] Memory usage: 105.17MiB. Peak allocated: 5384.00MiB Peak reserved: 10524.00MiB
+01/03/2025 08:09:19 [INFO|DP=0|PP=0|TP=29|ip-26-0-164-187]: [After model building] Memory usage: 132.21MiB. Peak allocated: 5384.00MiB Peak reserved: 9504.00MiB
+01/03/2025 08:09:19 [INFO|DP=0|PP=0|TP=30|ip-26-0-164-187]: [After model building] Memory usage: 132.21MiB. Peak allocated: 5384.00MiB Peak reserved: 9504.00MiB
+01/03/2025 08:09:19 [INFO|DP=0|PP=1|TP=8|ip-26-0-166-36]: [After model building] Memory usage: 105.17MiB. Peak allocated: 5384.00MiB Peak reserved: 10524.00MiB
+NCCL version 2.18.5+cuda12.2
+01/03/2025 08:09:19 [INFO|DP=0|PP=1|TP=20|ip-26-0-169-139]: [After model building] Memory usage: 105.17MiB. Peak allocated: 5384.00MiB Peak reserved: 10524.00MiB
+01/03/2025 08:09:19 [INFO|DP=0|PP=1|TP=23|ip-26-0-169-139]: [After model building] Memory usage: 105.17MiB. Peak allocated: 5384.00MiB Peak reserved: 10524.00MiB
+01/03/2025 08:09:19 [INFO|DP=0|PP=1|TP=30|ip-26-0-169-207]: Local number of parameters: 55.1M (105.15MiB)
+01/03/2025 08:09:19 [INFO|DP=0|PP=1|TP=15|ip-26-0-166-36]: [After model building] Memory usage: 105.17MiB. Peak allocated: 5384.00MiB Peak reserved: 10524.00MiB
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+01/03/2025 08:09:19 [INFO|DP=0|PP=1|TP=29|ip-26-0-169-207]: Local number of parameters: 55.1M (105.15MiB)
+01/03/2025 08:09:19 [INFO|DP=0|PP=1|TP=30|ip-26-0-169-207]: [After model building] Memory usage: 105.17MiB. Peak allocated: 5384.00MiB Peak reserved: 10524.00MiB
+01/03/2025 08:09:19 [INFO|DP=0|PP=1|TP=29|ip-26-0-169-207]: [After model building] Memory usage: 105.17MiB. Peak allocated: 5384.00MiB Peak reserved: 10524.00MiB
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+01/03/2025 08:09:19 [INFO|DP=0|PP=1|TP=4|ip-26-0-166-15]: Local number of parameters: 55.1M (105.15MiB)
+01/03/2025 08:09:19 [INFO|DP=0|PP=1|TP=4|ip-26-0-166-15]: [After model building] Memory usage: 105.17MiB. Peak allocated: 5384.00MiB Peak reserved: 10524.00MiB
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+01/03/2025 08:09:19 [INFO|DP=0|PP=1|TP=12|ip-26-0-166-36]: Local number of parameters: 55.1M (105.15MiB)
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+01/03/2025 08:09:19 [INFO|DP=0|PP=1|TP=27|ip-26-0-169-207]: Local number of parameters: 55.1M (105.15MiB)
+NCCL version 2.18.5+cuda12.2
+01/03/2025 08:09:19 [INFO|DP=0|PP=1|TP=12|ip-26-0-166-36]: [After model building] Memory usage: 105.17MiB. Peak allocated: 5384.00MiB Peak reserved: 10524.00MiB
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+01/03/2025 08:09:19 [INFO|DP=0|PP=1|TP=27|ip-26-0-169-207]: [After model building] Memory usage: 105.17MiB. Peak allocated: 5384.00MiB Peak reserved: 10524.00MiB
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+01/03/2025 08:09:19 [INFO|DP=0|PP=1|TP=17|ip-26-0-169-139]: Local number of parameters: 55.1M (105.15MiB)
+01/03/2025 08:09:19 [INFO|DP=0|PP=1|TP=17|ip-26-0-169-139]: [After model building] Memory usage: 105.17MiB. Peak allocated: 5384.00MiB Peak reserved: 10524.00MiB
+01/03/2025 08:09:19 [INFO|DP=0|PP=1|TP=28|ip-26-0-169-207]: Local number of parameters: 55.1M (105.15MiB)
+NCCL version 2.18.5+cuda12.2
+01/03/2025 08:09:19 [INFO|DP=0|PP=1|TP=28|ip-26-0-169-207]: [After model building] Memory usage: 105.17MiB. Peak allocated: 5384.00MiB Peak reserved: 10524.00MiB
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+01/03/2025 08:09:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: [Optimizer Building] Using LearningRateForSP as learning rate
+01/03/2025 08:09:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: [ZeRO sharding] Size of optimizer params per rank:
+01/03/2025 08:09:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: [ZeRO sharding] DP Rank 0 has 17.3M out of 69.3M (25.00%) params' optimizer states
+01/03/2025 08:09:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: [ZeRO sharding] DP Rank 1 has 17.3M out of 69.3M (25.00%) params' optimizer states
+01/03/2025 08:09:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: [ZeRO sharding] DP Rank 2 has 17.3M out of 69.3M (25.00%) params' optimizer states
+01/03/2025 08:09:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: [ZeRO sharding] DP Rank 3 has 17.3M out of 69.3M (25.00%) params' optimizer states
+01/03/2025 08:09:22 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.token_position_embeddings | PP: 0/2 | Block rank: 0
+01/03/2025 08:09:22 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.0 | PP: 0/2 | Block rank: 1
+01/03/2025 08:09:22 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.1 | PP: 0/2 | Block rank: 2
+01/03/2025 08:09:22 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.2 | PP: 0/2 | Block rank: 3
+01/03/2025 08:09:22 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.3 | PP: 0/2 | Block rank: 4
+01/03/2025 08:09:22 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.4 | PP: 0/2 | Block rank: 5
+01/03/2025 08:09:22 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.5 | PP: 0/2 | Block rank: 6
+01/03/2025 08:09:22 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.6 | PP: 0/2 | Block rank: 7
+01/03/2025 08:09:22 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.7 | PP: 0/2 | Block rank: 8
+01/03/2025 08:09:22 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.8 | PP: 0/2 | Block rank: 9
+01/03/2025 08:09:22 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.9 | PP: 0/2 | Block rank: 10
+01/03/2025 08:09:22 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.10 | PP: 0/2 | Block rank: 11
+01/03/2025 08:09:22 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.11 | PP: 0/2 | Block rank: 12
+01/03/2025 08:09:22 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.12 | PP: 0/2 | Block rank: 13
+01/03/2025 08:09:22 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.13 | PP: 0/2 | Block rank: 14
+01/03/2025 08:09:22 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.14 | PP: 0/2 | Block rank: 15
+01/03/2025 08:09:22 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.15 | PP: 0/2 | Block rank: 16
+01/03/2025 08:09:22 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.16 | PP: 1/2 | Block rank: 0
+01/03/2025 08:09:22 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.17 | PP: 1/2 | Block rank: 1
+01/03/2025 08:09:22 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.18 | PP: 1/2 | Block rank: 2
+01/03/2025 08:09:22 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.19 | PP: 1/2 | Block rank: 3
+01/03/2025 08:09:22 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.20 | PP: 1/2 | Block rank: 4
+01/03/2025 08:09:22 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.21 | PP: 1/2 | Block rank: 5
+01/03/2025 08:09:22 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.22 | PP: 1/2 | Block rank: 6
+01/03/2025 08:09:22 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.23 | PP: 1/2 | Block rank: 7
+01/03/2025 08:09:22 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.24 | PP: 1/2 | Block rank: 8
+01/03/2025 08:09:22 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.25 | PP: 1/2 | Block rank: 9
+01/03/2025 08:09:22 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.26 | PP: 1/2 | Block rank: 10
+01/03/2025 08:09:22 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.27 | PP: 1/2 | Block rank: 11
+01/03/2025 08:09:22 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.final_layer_norm | PP: 1/2 | Block rank: 12
+01/03/2025 08:09:22 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.lm_head | PP: 1/2 | Block rank: 13
+01/03/2025 08:09:22 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.cast_to_fp32 | PP: 1/2 | Block rank: 14
+01/03/2025 08:09:22 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: loss | PP: 1/2 | Block rank: 15
+01/03/2025 08:09:22 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: [Training Plan] Stage Stable Training Stage has 99 remaining training steps and has consumed 0 samples
+01/03/2025 08:09:22 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: Using dummy data generator
+01/03/2025 08:09:22 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: [Training Plan] There are 1 training stages 
+01/03/2025 08:09:22 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: [Stage Stable Training Stage] start from step 1 
+01/03/2025 08:09:22 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: 
+01/03/2025 08:09:22 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: [Start training] datetime: 2025-01-03 08:09:22.026723 | mbs: 4 | grad_accum: 16 | global_batch_size: 256 | sequence_length: 4096 | train_steps: 100 | start_iteration_step: 0 | consumed_train_samples: 0
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+01/03/2025 08:09:22 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: Resuming training from stage Stable Training Stage, it has trained for 0 samples and has 99 remaining train steps
+01/03/2025 08:09:22 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:  Memory usage: 462.68MiB. Peak allocated 5384.00MiB. Peak reserved: 10528.00MiB
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+01/03/2025 08:10:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:  Memory usage: 730.82MiB. Peak allocated 2969.96MiB. Peak reserved: 3150.00MiB
+01/03/2025 08:10:12 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:  Memory usage: 863.06MiB. Peak allocated 1160.48MiB. Peak reserved: 3202.00MiB
+01/03/2025 08:10:12 [INFO|DP=0|PP=1|TP=0|ip-26-0-166-15]: iteration: 1 / 100 | consumed_tokens: 1.05M | elapsed_time_per_iteration_ms: 49K | tokens_per_sec: 21.4K | tokens_per_sec_per_gpu: 83.6 | global_batch_size: 256 | lm_loss: 12.1 | lr: 0.00015 | model_tflops_per_gpu: 2.15 | hardware_tflops_per_gpu: 2.15 | grad_norm: 0.622 | cuda_memory_allocated: 774M | cuda_max_memory_reserved: 2.47G | hd_total_memory_tb: 312G | hd_used_memory_tb: 70.3G | hd_free_memory_tb: 242G
+01/03/2025 08:10:47 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:  Memory usage: 863.06MiB. Peak allocated 2991.53MiB. Peak reserved: 3212.00MiB
+01/03/2025 08:10:47 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:  Memory usage: 863.06MiB. Peak allocated 1160.48MiB. Peak reserved: 3212.00MiB
+01/03/2025 08:10:47 [INFO|DP=0|PP=1|TP=0|ip-26-0-166-15]: iteration: 2 / 100 | consumed_tokens: 2.1M | elapsed_time_per_iteration_ms: 34.8K | tokens_per_sec: 30.1K | tokens_per_sec_per_gpu: 118 | global_batch_size: 256 | lm_loss: 12.1 | lr: 0.0003 | model_tflops_per_gpu: 3.02 | hardware_tflops_per_gpu: 3.02 | grad_norm: 0.624 | cuda_memory_allocated: 774M | cuda_max_memory_reserved: 2.48G | hd_total_memory_tb: 312G | hd_used_memory_tb: 70.3G | hd_free_memory_tb: 242G
+01/03/2025 08:11:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:  Memory usage: 863.06MiB. Peak allocated 2991.53MiB. Peak reserved: 3212.00MiB
+01/03/2025 08:11:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: | job_id   | name                                                            | nodes | seq_len | mbs | batch_accum | gbs | mTFLOPs | hTFLOPs | tok/s/gpu | AllReduce (GB/s) | AllGather (GB/s) | ReduceScatter (GB/s) | AR Intra-node (GB/s) | AG Intra-node (GB/s) | RS Intra-node (GB/s) | Mem Alloc (GB) | Mem Res (GB) | dp | pp | tp | pp_engine                           | tp_mode        | tp_async_comm | hidden_size | hidden_act | num_layers | num_heads | num_kv_heads | max_pos | vocab_size | tie_word_embeddings | dtype          | zero_stage | ddp_bucket_cap_mb | accumulate_grad_in_fp32 | Total Params | Local Params |
+01/03/2025 08:11:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: | -------- | --------------------------------------------------------------- | ----- | ------- | --- | ----------- | --- | ------- | ------- | --------- | ---------------- | ---------------- | -------------------- | -------------------- | -------------------- | -------------------- | -------------- | ------------ | -- | -- | -- | ----------------------------------- | -------------- | ------------- | ----------- | ---------- | ---------- | --------- | ------------ | ------- | ---------- | ------------------- | -------------- | ---------- | ----------------- | ----------------------- | ------------ | ------------ |
+01/03/2025 08:11:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: | 14019312 | 3.57G_dp4_tp32_pp2_acc16_mbs4_seq4096_zero1_tpmodeRED_vocab131k | 32    | 4096    | 4   | 16          | 256 | 3.11    | 3.11    | 121.17    | 27.77            | 13.42            | 13.45                | 426.00               | 241.83               | 241.67               | 1.13           | 3.14         | 4  | 2  | 32 | OneForwardOneBackwardPipelineEngine | REDUCE_SCATTER | True          | 3072        | silu       | 28         | 32        | 32           | 4096    | 131072     | True                | torch.bfloat16 | 1          | 25                | True                    | 3.98G        | 69.3M        |
+01/03/2025 08:11:21 [INFO|DP=0|PP=1|TP=0|ip-26-0-166-15]: iteration: 3 / 100 | consumed_tokens: 3.15M | elapsed_time_per_iteration_ms: 33.8K | tokens_per_sec: 31K | tokens_per_sec_per_gpu: 121 | global_batch_size: 256 | lm_loss: 12.1 | lr: 0.000296 | model_tflops_per_gpu: 3.11 | hardware_tflops_per_gpu: 3.11 | grad_norm: 0.592 | cuda_memory_allocated: 774M | cuda_max_memory_reserved: 2.48G | hd_total_memory_tb: 312G | hd_used_memory_tb: 70.3G | hd_free_memory_tb: 242G
+01/03/2025 08:11:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: Attempting to write benchmark results to CSV file: benchmark/results/bench_final2.csv
+01/03/2025 08:11:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: Acquired lock for CSV file: benchmark/results/bench_final2.csv
+01/03/2025 08:11:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: Successfully wrote to CSV file: benchmark/results/bench_final2.csv. Releasing lock...
+01/03/2025 08:11:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: Throughput logging complete
+terminate called after throwing an instance of 'c10::Error'
+  what():  CUDA driver error: unknown error
+Exception raised from _hasPrimaryContext at ../aten/src/ATen/cuda/detail/CUDAHooks.cpp:67 (most recent call first):
+frame #0: c10::Error::Error(c10::SourceLocation, std::string) + 0x57 (0x7fc338970617 in /fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/lib/libc10.so)
+frame #1: c10::detail::torchCheckFail(char const*, char const*, unsigned int, char const*) + 0x68 (0x7fc33892ba56 in /fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/lib/libc10.so)
+frame #2: <unknown function> + 0x11a00bf (0x7fc339c8c0bf in /fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/lib/libtorch_cuda.so)
+frame #3: c10::cuda::MaybeSetDevice(int) + 0xc (0x7fc338a21bdc in /fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/lib/libc10_cuda.so)
+frame #4: std::_Sp_counted_ptr_inplace<std::vector<at::cuda::CUDAEvent, std::allocator<at::cuda::CUDAEvent> >, std::allocator<std::vector<at::cuda::CUDAEvent, std::allocator<at::cuda::CUDAEvent> > >, (__gnu_cxx::_Lock_policy)2>::_M_dispose() + 0x98 (0x7fc339d9bae8 in /fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/lib/libtorch_cuda.so)
+frame #5: std::_Sp_counted_base<(__gnu_cxx::_Lock_policy)2>::_M_release() + 0x48 (0x7fc397cee048 in /fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/lib/libtorch_python.so)
+frame #6: c10d::ProcessGroupNCCL::WorkNCCL::~WorkNCCL() + 0x135 (0x7fc339d66c55 in /fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/lib/libtorch_cuda.so)
+frame #7: c10d::ProcessGroupNCCL::workCleanupLoop() + 0x3c5 (0x7fc339d7c585 in /fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/lib/libtorch_cuda.so)
+frame #8: c10d::ProcessGroupNCCL::ncclCommWatchdog() + 0x78 (0x7fc339d7c718 in /fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/lib/libtorch_cuda.so)
+frame #9: <unknown function> + 0xd3e95 (0x7fc3c7a74e95 in /fsx/nouamane/miniconda/envs/2-1-cu121/bin/../lib/libstdc++.so.6)
+frame #10: <unknown function> + 0x8609 (0x7fc3c7ee4609 in /lib/x86_64-linux-gnu/libpthread.so.0)
+frame #11: clone + 0x43 (0x7fc3c7cad353 in /lib/x86_64-linux-gnu/libc.so.6)
+
+double free or corruption (out)
+[2025-01-03 08:11:52,159] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 136897 closing signal SIGTERM
+[2025-01-03 08:11:52,159] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 136898 closing signal SIGTERM
+[2025-01-03 08:11:52,159] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 136899 closing signal SIGTERM
+[2025-01-03 08:11:52,159] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 136900 closing signal SIGTERM
+[2025-01-03 08:11:52,159] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 136902 closing signal SIGTERM
+[2025-01-03 08:11:52,168] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 520387 closing signal SIGTERM
+[2025-01-03 08:11:52,168] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 520388 closing signal SIGTERM
+[2025-01-03 08:11:52,168] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 520389 closing signal SIGTERM
+[2025-01-03 08:11:52,168] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 520390 closing signal SIGTERM
+[2025-01-03 08:11:52,168] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 520391 closing signal SIGTERM
+[2025-01-03 08:11:52,168] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 520393 closing signal SIGTERM
+[2025-01-03 08:11:52,168] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 520394 closing signal SIGTERM
+[2025-01-03 08:11:52,170] torch.distributed.elastic.multiprocessing.api: [ERROR] failed (exitcode: 1) local_rank: 0 (pid: 266547) of binary: /fsx/nouamane/miniconda/envs/2-1-cu121/bin/python
+[2025-01-03 08:11:52,172] torch.distributed.elastic.multiprocessing.api: [ERROR] failed (exitcode: 1) local_rank: 0 (pid: 71950) of binary: /fsx/nouamane/miniconda/envs/2-1-cu121/bin/python
+[2025-01-03 08:11:52,179] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 37735 closing signal SIGTERM
+[2025-01-03 08:11:52,179] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 37737 closing signal SIGTERM
+[2025-01-03 08:11:52,179] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 37739 closing signal SIGTERM
+[2025-01-03 08:11:52,179] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 37740 closing signal SIGTERM
+[2025-01-03 08:11:52,178] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 27296 closing signal SIGTERM
+[2025-01-03 08:11:52,178] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 27297 closing signal SIGTERM
+[2025-01-03 08:11:52,178] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 27298 closing signal SIGTERM
+[2025-01-03 08:11:52,178] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 27299 closing signal SIGTERM
+[2025-01-03 08:11:52,178] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 27300 closing signal SIGTERM
+[2025-01-03 08:11:52,178] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 27301 closing signal SIGTERM
+[2025-01-03 08:11:52,181] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 108381 closing signal SIGTERM
+[2025-01-03 08:11:52,182] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 108383 closing signal SIGTERM
+[2025-01-03 08:11:52,182] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 108384 closing signal SIGTERM
+[2025-01-03 08:11:52,182] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 108385 closing signal SIGTERM
+[2025-01-03 08:11:52,182] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 108386 closing signal SIGTERM
+[2025-01-03 08:11:52,182] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 108387 closing signal SIGTERM
+[2025-01-03 08:11:52,183] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 56953 closing signal SIGTERM
+[2025-01-03 08:11:52,183] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 56954 closing signal SIGTERM
+[2025-01-03 08:11:52,183] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 56955 closing signal SIGTERM
+[2025-01-03 08:11:52,183] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 56956 closing signal SIGTERM
+[2025-01-03 08:11:52,183] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 56958 closing signal SIGTERM
+[2025-01-03 08:11:52,183] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 56959 closing signal SIGTERM
+[2025-01-03 08:11:52,183] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 56960 closing signal SIGTERM
+[2025-01-03 08:11:52,187] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 67370 closing signal SIGTERM
+[2025-01-03 08:11:52,187] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 67371 closing signal SIGTERM
+[2025-01-03 08:11:52,187] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 67372 closing signal SIGTERM
+[2025-01-03 08:11:52,187] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 67373 closing signal SIGTERM
+[2025-01-03 08:11:52,187] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 67374 closing signal SIGTERM
+[2025-01-03 08:11:52,187] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 67376 closing signal SIGTERM
+[2025-01-03 08:11:52,188] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 67377 closing signal SIGTERM
+[2025-01-03 08:11:52,189] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 26597 closing signal SIGTERM
+[2025-01-03 08:11:52,195] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 263666 closing signal SIGTERM
+[2025-01-03 08:11:52,195] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 263667 closing signal SIGTERM
+[2025-01-03 08:11:52,195] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 263668 closing signal SIGTERM
+[2025-01-03 08:11:52,195] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 263669 closing signal SIGTERM
+[2025-01-03 08:11:52,195] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 263670 closing signal SIGTERM
+[2025-01-03 08:11:52,195] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 263671 closing signal SIGTERM
+[2025-01-03 08:11:52,202] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 65775 closing signal SIGTERM
+[2025-01-03 08:11:52,202] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 65776 closing signal SIGTERM
+[2025-01-03 08:11:52,203] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 65777 closing signal SIGTERM
+[2025-01-03 08:11:52,203] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 65778 closing signal SIGTERM
+[2025-01-03 08:11:52,203] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 65779 closing signal SIGTERM
+[2025-01-03 08:11:52,203] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 65780 closing signal SIGTERM
+[2025-01-03 08:11:52,203] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 65782 closing signal SIGTERM
+[2025-01-03 08:11:52,208] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 33579 closing signal SIGTERM
+[2025-01-03 08:11:52,208] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 33580 closing signal SIGTERM
+[2025-01-03 08:11:52,208] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 33581 closing signal SIGTERM
+[2025-01-03 08:11:52,208] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 33582 closing signal SIGTERM
+[2025-01-03 08:11:52,208] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 33583 closing signal SIGTERM
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 264, in launch_agent
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 264, in launch_agent
+    raise ChildFailedError(
+torch.distributed.elastic.multiprocessing.errors.ChildFailedError: 
+============================================================
+/fsx/nouamane/projects/nanotron/run_train.py FAILED
+------------------------------------------------------------
+Failures:
+[1]:
+  time      : 2025-01-03_08:11:52
+  host      : ip-26-0-163-236.ec2.internal
+  rank      : 1 (local_rank: 1)
+  exitcode  : 1 (pid: 266548)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[2]:
+  time      : 2025-01-03_08:11:52
+  host      : ip-26-0-163-236.ec2.internal
+  rank      : 2 (local_rank: 2)
+  exitcode  : 1 (pid: 266549)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[3]:
+  time      : 2025-01-03_08:11:52
+  host      : ip-26-0-163-236.ec2.internal
+  rank      : 3 (local_rank: 3)
+  exitcode  : 1 (pid: 266550)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[4]:
+  time      : 2025-01-03_08:11:52
+  host      : ip-26-0-163-236.ec2.internal
+  rank      : 4 (local_rank: 4)
+  exitcode  : 1 (pid: 266551)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[5]:
+  time      : 2025-01-03_08:11:52
+  host      : ip-26-0-163-236.ec2.internal
+  rank      : 5 (local_rank: 5)
+  exitcode  : 1 (pid: 266552)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[6]:
+  time      : 2025-01-03_08:11:52
+  host      : ip-26-0-163-236.ec2.internal
+  rank      : 6 (local_rank: 6)
+  exitcode  : 1 (pid: 266553)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[7]:
+  time      : 2025-01-03_08:11:52
+  host      : ip-26-0-163-236.ec2.internal
+  rank      : 7 (local_rank: 7)
+  exitcode  : 1 (pid: 266554)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+    raise ChildFailedError(
+------------------------------------------------------------
+Root Cause (first observed failure):
+[0]:
+  time      : 2025-01-03_08:11:52
+  host      : ip-26-0-163-236.ec2.internal
+  rank      : 0 (local_rank: 0)
+  exitcode  : 1 (pid: 266547)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+============================================================
+torch.distributed.elastic.multiprocessing.errors.ChildFailedError: 
+============================================================
+/fsx/nouamane/projects/nanotron/run_train.py FAILED
+------------------------------------------------------------
+Failures:
+[1]:
+  time      : 2025-01-03_08:11:52
+  host      : ip-26-0-173-7.ec2.internal
+  rank      : 249 (local_rank: 1)
+  exitcode  : 1 (pid: 71951)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[2]:
+  time      : 2025-01-03_08:11:52
+  host      : ip-26-0-173-7.ec2.internal
+  rank      : 250 (local_rank: 2)
+  exitcode  : 1 (pid: 71952)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[3]:
+  time      : 2025-01-03_08:11:52
+  host      : ip-26-0-173-7.ec2.internal
+  rank      : 251 (local_rank: 3)
+  exitcode  : 1 (pid: 71953)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[4]:
+  time      : 2025-01-03_08:11:52
+  host      : ip-26-0-173-7.ec2.internal
+  rank      : 252 (local_rank: 4)
+  exitcode  : 1 (pid: 71954)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[5]:
+  time      : 2025-01-03_08:11:52
+  host      : ip-26-0-173-7.ec2.internal
+  rank      : 253 (local_rank: 5)
+  exitcode  : 1 (pid: 71955)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[6]:
+  time      : 2025-01-03_08:11:52
+  host      : ip-26-0-173-7.ec2.internal
+  rank      : 254 (local_rank: 6)
+  exitcode  : 1 (pid: 71956)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[7]:
+  time      : 2025-01-03_08:11:52
+  host      : ip-26-0-173-7.ec2.internal
+  rank      : 255 (local_rank: 7)
+  exitcode  : 1 (pid: 71957)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+------------------------------------------------------------
+Root Cause (first observed failure):
+[0]:
+  time      : 2025-01-03_08:11:52
+  host      : ip-26-0-173-7.ec2.internal
+  rank      : 248 (local_rank: 0)
+  exitcode  : 1 (pid: 71950)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+============================================================
+[2025-01-03 08:11:52,252] torch.distributed.elastic.multiprocessing.api: [ERROR] failed (exitcode: 1) local_rank: 0 (pid: 26596) of binary: /fsx/nouamane/miniconda/envs/2-1-cu121/bin/python
+[2025-01-03 08:11:52,305] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-164-207.ec2.internal_26524_0' has failed to shutdown the rendezvous '14019312' due to an error of type RendezvousConnectionError.
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 264, in launch_agent
+    raise ChildFailedError(
+torch.distributed.elastic.multiprocessing.errors.ChildFailedError: 
+============================================================
+/fsx/nouamane/projects/nanotron/run_train.py FAILED
+------------------------------------------------------------
+Failures:
+[1]:
+  time      : 2025-01-03_08:11:52
+  host      : ip-26-0-164-207.ec2.internal
+  rank      : 34 (local_rank: 2)
+  exitcode  : 1 (pid: 26598)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[2]:
+  time      : 2025-01-03_08:11:52
+  host      : ip-26-0-164-207.ec2.internal
+  rank      : 35 (local_rank: 3)
+  exitcode  : 1 (pid: 26599)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[3]:
+  time      : 2025-01-03_08:11:52
+  host      : ip-26-0-164-207.ec2.internal
+  rank      : 36 (local_rank: 4)
+  exitcode  : 1 (pid: 26600)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[4]:
+  time      : 2025-01-03_08:11:52
+  host      : ip-26-0-164-207.ec2.internal
+  rank      : 37 (local_rank: 5)
+  exitcode  : 1 (pid: 26601)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[5]:
+  time      : 2025-01-03_08:11:52
+  host      : ip-26-0-164-207.ec2.internal
+  rank      : 38 (local_rank: 6)
+  exitcode  : 1 (pid: 26602)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[6]:
+  time      : 2025-01-03_08:11:52
+  host      : ip-26-0-164-207.ec2.internal
+  rank      : 39 (local_rank: 7)
+  exitcode  : 1 (pid: 26603)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+------------------------------------------------------------
+Root Cause (first observed failure):
+[0]:
+  time      : 2025-01-03_08:11:52
+  host      : ip-26-0-164-207.ec2.internal
+  rank      : 32 (local_rank: 0)
+  exitcode  : 1 (pid: 26596)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+============================================================
+srun: error: ip-26-0-163-236: task 0: Exited with exit code 1
+srun: launch/slurm: _step_signal: Terminating StepId=14019312.0
+[2025-01-03 08:11:52,520] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 08:11:52,520] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 67370 closing signal SIGTERM
+[2025-01-03 08:11:52,520] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 67371 closing signal SIGTERM
+[2025-01-03 08:11:52,520] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 67372 closing signal SIGTERM
+[2025-01-03 08:11:52,521] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 67374 closing signal SIGTERM
+[2025-01-03 08:11:52,521] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 67376 closing signal SIGTERM
+[2025-01-03 08:11:52,521] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 08:11:52,521] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 08:11:52,521] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 67377 closing signal SIGTERM
+[2025-01-03 08:11:52,523] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 08:11:52,521] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 27297 closing signal SIGTERM
+[2025-01-03 08:11:52,521] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 455466 closing signal SIGTERM
+[2025-01-03 08:11:52,521] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 455467 closing signal SIGTERM
+[2025-01-03 08:11:52,522] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 455468 closing signal SIGTERM
+[2025-01-03 08:11:52,522] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 455469 closing signal SIGTERM
+[2025-01-03 08:11:52,523] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 33579 closing signal SIGTERM
+[2025-01-03 08:11:52,523] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 33580 closing signal SIGTERM
+[2025-01-03 08:11:52,521] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 27298 closing signal SIGTERM
+[2025-01-03 08:11:52,522] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 455470 closing signal SIGTERM
+[2025-01-03 08:11:52,521] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 27299 closing signal SIGTERM
+[2025-01-03 08:11:52,522] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 455471 closing signal SIGTERM
+[2025-01-03 08:11:52,522] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 08:11:52,522] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 455472 closing signal SIGTERM
+[2025-01-03 08:11:52,521] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 27300 closing signal SIGTERM
+[2025-01-03 08:11:52,522] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 455473 closing signal SIGTERM
+[2025-01-03 08:11:52,521] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 27301 closing signal SIGTERM
+[2025-01-03 08:11:52,521] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 08:11:52,523] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 33582 closing signal SIGTERM
+[2025-01-03 08:11:52,522] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 520387 closing signal SIGTERM
+[2025-01-03 08:11:52,522] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 08:11:52,522] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 520388 closing signal SIGTERM
+[2025-01-03 08:11:52,523] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 33583 closing signal SIGTERM
+[2025-01-03 08:11:52,522] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 520389 closing signal SIGTERM
+[2025-01-03 08:11:52,521] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 61024 closing signal SIGTERM
+[2025-01-03 08:11:52,522] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 520390 closing signal SIGTERM
+[2025-01-03 08:11:52,522] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 65776 closing signal SIGTERM
+[2025-01-03 08:11:52,522] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 65777 closing signal SIGTERM
+[2025-01-03 08:11:52,522] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 520391 closing signal SIGTERM
+[2025-01-03 08:11:52,521] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 61025 closing signal SIGTERM
+[2025-01-03 08:11:52,522] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 520393 closing signal SIGTERM
+[2025-01-03 08:11:52,521] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 61026 closing signal SIGTERM
+[2025-01-03 08:11:52,521] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 08:11:52,522] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 65778 closing signal SIGTERM
+[2025-01-03 08:11:52,521] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 61027 closing signal SIGTERM
+[2025-01-03 08:11:52,522] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 65779 closing signal SIGTERM
+[2025-01-03 08:11:52,521] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 61028 closing signal SIGTERM
+[2025-01-03 08:11:52,522] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 520394 closing signal SIGTERM
+[2025-01-03 08:11:52,522] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 140988 closing signal SIGTERM
+[2025-01-03 08:11:52,522] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 65780 closing signal SIGTERM
+[2025-01-03 08:11:52,522] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 65782 closing signal SIGTERM
+[2025-01-03 08:11:52,522] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 61029 closing signal SIGTERM
+[2025-01-03 08:11:52,522] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 140989 closing signal SIGTERM
+[2025-01-03 08:11:52,521] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 08:11:52,522] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 61030 closing signal SIGTERM
+[2025-01-03 08:11:52,522] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 140990 closing signal SIGTERM
+[2025-01-03 08:11:52,521] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 08:11:52,522] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 61031 closing signal SIGTERM
+[2025-01-03 08:11:52,522] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 140991 closing signal SIGTERM
+[2025-01-03 08:11:52,521] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 56954 closing signal SIGTERM
+[2025-01-03 08:11:52,521] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 56955 closing signal SIGTERM
+[2025-01-03 08:11:52,521] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 26350 closing signal SIGTERM
+[2025-01-03 08:11:52,522] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 140993 closing signal SIGTERM
+[2025-01-03 08:11:52,521] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 56956 closing signal SIGTERM
+[2025-01-03 08:11:52,521] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 56958 closing signal SIGTERM
+[2025-01-03 08:11:52,521] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 26351 closing signal SIGTERM
+[2025-01-03 08:11:52,521] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 08:11:52,522] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 140994 closing signal SIGTERM
+[2025-01-03 08:11:52,522] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 140995 closing signal SIGTERM
+[2025-01-03 08:11:52,522] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 56959 closing signal SIGTERM
+[2025-01-03 08:11:52,521] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 26352 closing signal SIGTERM
+[2025-01-03 08:11:52,521] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 08:11:52,522] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 56960 closing signal SIGTERM
+[2025-01-03 08:11:52,522] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 26353 closing signal SIGTERM
+[2025-01-03 08:11:52,521] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 45898 closing signal SIGTERM
+[2025-01-03 08:11:52,521] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 45899 closing signal SIGTERM
+[2025-01-03 08:11:52,522] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 26354 closing signal SIGTERM
+[2025-01-03 08:11:52,522] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 26355 closing signal SIGTERM
+[2025-01-03 08:11:52,521] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 263668 closing signal SIGTERM
+[2025-01-03 08:11:52,521] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 45900 closing signal SIGTERM
+[2025-01-03 08:11:52,521] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 08:11:52,521] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 45901 closing signal SIGTERM
+[2025-01-03 08:11:52,522] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 26356 closing signal SIGTERM
+[2025-01-03 08:11:52,522] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 08:11:52,521] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 08:11:52,521] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 263669 closing signal SIGTERM
+[2025-01-03 08:11:52,521] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 263670 closing signal SIGTERM
+[2025-01-03 08:11:52,521] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 45902 closing signal SIGTERM
+[2025-01-03 08:11:52,521] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 45903 closing signal SIGTERM
+[2025-01-03 08:11:52,521] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 27771 closing signal SIGTERM
+[2025-01-03 08:11:52,522] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 117518 closing signal SIGTERM
+[2025-01-03 08:11:52,521] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 108381 closing signal SIGTERM
+[2025-01-03 08:11:52,521] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 263671 closing signal SIGTERM
+[2025-01-03 08:11:52,521] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 08:11:52,521] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 45904 closing signal SIGTERM
+[2025-01-03 08:11:52,521] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 45905 closing signal SIGTERM
+[2025-01-03 08:11:52,522] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 117519 closing signal SIGTERM
+[2025-01-03 08:11:52,522] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 27772 closing signal SIGTERM
+[2025-01-03 08:11:52,522] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 117520 closing signal SIGTERM
+[2025-01-03 08:11:52,521] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 08:11:52,522] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 27773 closing signal SIGTERM
+[2025-01-03 08:11:52,521] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 108384 closing signal SIGTERM
+[2025-01-03 08:11:52,521] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 08:11:52,521] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 47538 closing signal SIGTERM
+[2025-01-03 08:11:52,522] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 117521 closing signal SIGTERM
+[2025-01-03 08:11:52,521] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 47539 closing signal SIGTERM
+[2025-01-03 08:11:52,521] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 08:11:52,522] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 117522 closing signal SIGTERM
+[2025-01-03 08:11:52,522] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 27774 closing signal SIGTERM
+[2025-01-03 08:11:52,521] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 108385 closing signal SIGTERM
+[2025-01-03 08:11:52,522] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 45074 closing signal SIGTERM
+[2025-01-03 08:11:52,522] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 27775 closing signal SIGTERM
+[2025-01-03 08:11:52,522] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 117523 closing signal SIGTERM
+[2025-01-03 08:11:52,522] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 254466 closing signal SIGTERM
+[2025-01-03 08:11:52,521] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 47540 closing signal SIGTERM
+[2025-01-03 08:11:52,522] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 117524 closing signal SIGTERM
+[2025-01-03 08:11:52,522] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 27776 closing signal SIGTERM
+[2025-01-03 08:11:52,521] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 47541 closing signal SIGTERM
+[2025-01-03 08:11:52,521] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 108386 closing signal SIGTERM
+[2025-01-03 08:11:52,522] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 45075 closing signal SIGTERM
+[2025-01-03 08:11:52,522] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 33210 closing signal SIGTERM
+[2025-01-03 08:11:52,521] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 108387 closing signal SIGTERM
+[2025-01-03 08:11:52,522] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 117525 closing signal SIGTERM
+[2025-01-03 08:11:52,522] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 254467 closing signal SIGTERM
+[2025-01-03 08:11:52,522] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 47542 closing signal SIGTERM
+[2025-01-03 08:11:52,522] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 08:11:52,521] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 08:11:52,522] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 27777 closing signal SIGTERM
+[2025-01-03 08:11:52,522] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 47543 closing signal SIGTERM
+[2025-01-03 08:11:52,522] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 33211 closing signal SIGTERM
+[2025-01-03 08:11:52,522] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 45076 closing signal SIGTERM
+[2025-01-03 08:11:52,522] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 27778 closing signal SIGTERM
+[2025-01-03 08:11:52,522] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 254468 closing signal SIGTERM
+[2025-01-03 08:11:52,522] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 47544 closing signal SIGTERM
+[2025-01-03 08:11:52,522] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 33212 closing signal SIGTERM
+[2025-01-03 08:11:52,522] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 45077 closing signal SIGTERM
+[2025-01-03 08:11:52,521] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 136897 closing signal SIGTERM
+[2025-01-03 08:11:52,522] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 47545 closing signal SIGTERM
+[2025-01-03 08:11:52,522] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 31730 closing signal SIGTERM
+[2025-01-03 08:11:52,522] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 45078 closing signal SIGTERM
+[2025-01-03 08:11:52,521] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 136898 closing signal SIGTERM
+[2025-01-03 08:11:52,522] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 33213 closing signal SIGTERM
+[2025-01-03 08:11:52,522] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 45079 closing signal SIGTERM
+[2025-01-03 08:11:52,522] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 08:11:52,522] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 254469 closing signal SIGTERM
+[2025-01-03 08:11:52,522] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 33214 closing signal SIGTERM
+[2025-01-03 08:11:52,521] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 136899 closing signal SIGTERM
+[2025-01-03 08:11:52,522] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 254470 closing signal SIGTERM
+[2025-01-03 08:11:52,522] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 31731 closing signal SIGTERM
+[2025-01-03 08:11:52,522] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 45080 closing signal SIGTERM
+[2025-01-03 08:11:52,521] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 136900 closing signal SIGTERM
+[2025-01-03 08:11:52,522] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 254471 closing signal SIGTERM
+[2025-01-03 08:11:52,522] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 33215 closing signal SIGTERM
+[2025-01-03 08:11:52,522] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 45081 closing signal SIGTERM
+[2025-01-03 08:11:52,522] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 33216 closing signal SIGTERM
+[2025-01-03 08:11:52,522] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 31732 closing signal SIGTERM
+[2025-01-03 08:11:52,522] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 76745 closing signal SIGTERM
+[2025-01-03 08:11:52,522] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 33217 closing signal SIGTERM
+[2025-01-03 08:11:52,522] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 254472 closing signal SIGTERM
+[2025-01-03 08:11:52,522] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 76746 closing signal SIGTERM
+[2025-01-03 08:11:52,522] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 254473 closing signal SIGTERM
+[2025-01-03 08:11:52,522] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 31733 closing signal SIGTERM
+[2025-01-03 08:11:52,522] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 76747 closing signal SIGTERM
+[2025-01-03 08:11:52,522] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 31734 closing signal SIGTERM
+[2025-01-03 08:11:52,523] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 08:11:52,523] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 08:11:52,522] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 76748 closing signal SIGTERM
+[2025-01-03 08:11:52,522] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 76749 closing signal SIGTERM
+[2025-01-03 08:11:52,522] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 31735 closing signal SIGTERM
+[2025-01-03 08:11:52,522] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 76750 closing signal SIGTERM
+[2025-01-03 08:11:52,522] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 08:11:52,524] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 37735 closing signal SIGTERM
+[2025-01-03 08:11:52,522] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 76751 closing signal SIGTERM
+[2025-01-03 08:11:52,524] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 778959 closing signal SIGTERM
+[2025-01-03 08:11:52,524] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 778960 closing signal SIGTERM
+[2025-01-03 08:11:52,522] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 31736 closing signal SIGTERM
+[2025-01-03 08:11:52,524] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 778961 closing signal SIGTERM
+[2025-01-03 08:11:52,522] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 276449 closing signal SIGTERM
+[2025-01-03 08:11:52,521] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 08:11:52,522] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 31737 closing signal SIGTERM
+[2025-01-03 08:11:52,524] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 37737 closing signal SIGTERM
+[2025-01-03 08:11:52,524] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 778962 closing signal SIGTERM
+[2025-01-03 08:11:52,522] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 76752 closing signal SIGTERM
+[2025-01-03 08:11:52,524] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 778963 closing signal SIGTERM
+[2025-01-03 08:11:52,521] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3398281 closing signal SIGTERM
+[2025-01-03 08:11:52,524] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 37739 closing signal SIGTERM
+[2025-01-03 08:11:52,522] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 276450 closing signal SIGTERM
+[2025-01-03 08:11:52,524] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 778964 closing signal SIGTERM
+[2025-01-03 08:11:52,524] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 778965 closing signal SIGTERM
+[2025-01-03 08:11:52,523] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 08:11:52,522] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 08:11:52,524] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 37740 closing signal SIGTERM
+[2025-01-03 08:11:52,522] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 276451 closing signal SIGTERM
+[2025-01-03 08:11:52,524] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 778966 closing signal SIGTERM
+[2025-01-03 08:11:52,521] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3398282 closing signal SIGTERM
+[2025-01-03 08:11:52,523] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 66554 closing signal SIGTERM
+[2025-01-03 08:11:52,522] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 276452 closing signal SIGTERM
+[2025-01-03 08:11:52,522] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 83177 closing signal SIGTERM
+[2025-01-03 08:11:52,522] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 276453 closing signal SIGTERM
+[2025-01-03 08:11:52,522] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 83178 closing signal SIGTERM
+[2025-01-03 08:11:52,522] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 08:11:52,522] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3398283 closing signal SIGTERM
+[2025-01-03 08:11:52,522] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 276454 closing signal SIGTERM
+[2025-01-03 08:11:52,522] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 83179 closing signal SIGTERM
+[2025-01-03 08:11:52,523] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 66555 closing signal SIGTERM
+[2025-01-03 08:11:52,522] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3398284 closing signal SIGTERM
+[2025-01-03 08:11:52,522] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 83180 closing signal SIGTERM
+[2025-01-03 08:11:52,523] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 136914 closing signal SIGTERM
+[2025-01-03 08:11:52,522] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 276455 closing signal SIGTERM
+[2025-01-03 08:11:52,522] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 83181 closing signal SIGTERM
+[2025-01-03 08:11:52,523] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 136915 closing signal SIGTERM
+[2025-01-03 08:11:52,523] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 66556 closing signal SIGTERM
+[2025-01-03 08:11:52,522] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 276456 closing signal SIGTERM
+[2025-01-03 08:11:52,522] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3398285 closing signal SIGTERM
+[2025-01-03 08:11:52,522] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 83182 closing signal SIGTERM
+[2025-01-03 08:11:52,523] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 136916 closing signal SIGTERM
+[2025-01-03 08:11:52,523] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 66557 closing signal SIGTERM
+[2025-01-03 08:11:52,523] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 136917 closing signal SIGTERM
+[2025-01-03 08:11:52,523] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 66558 closing signal SIGTERM
+[2025-01-03 08:11:52,522] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 83183 closing signal SIGTERM
+[2025-01-03 08:11:52,522] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 83184 closing signal SIGTERM
+[2025-01-03 08:11:52,523] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 136918 closing signal SIGTERM
+[2025-01-03 08:11:52,523] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 66559 closing signal SIGTERM
+[2025-01-03 08:11:52,522] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3398286 closing signal SIGTERM
+[2025-01-03 08:11:52,523] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 136919 closing signal SIGTERM
+[2025-01-03 08:11:52,523] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 66560 closing signal SIGTERM
+[2025-01-03 08:11:52,523] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 136920 closing signal SIGTERM
+[2025-01-03 08:11:52,523] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 136921 closing signal SIGTERM
+[2025-01-03 08:11:52,522] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3398287 closing signal SIGTERM
+[2025-01-03 08:11:52,523] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 66561 closing signal SIGTERM
+[2025-01-03 08:11:52,522] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3398288 closing signal SIGTERM
+srun: error: ip-26-0-164-207: task 6: Exited with exit code 1
+srun: error: ip-26-0-173-7: task 28: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 878, in _invoke_run
+    run_result = self._monitor_workers(self._worker_group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/local_elastic_agent.py", line 307, in _monitor_workers
+    result = self._pcontext.wait(0)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 288, in wait
+    return self._poll()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 668, in _poll
+    self.close()  # terminate all running procs
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 331, in close
+    self._close(death_sig=death_sig, timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 713, in _close
+    handler.proc.wait(time_to_wait)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1209, in wait
+    return self._wait(timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1953, in _wait
+    time.sleep(delay)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 37661 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 878, in _invoke_run
+    run_result = self._monitor_workers(self._worker_group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/local_elastic_agent.py", line 307, in _monitor_workers
+    result = self._pcontext.wait(0)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 288, in wait
+    return self._poll()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 668, in _poll
+    self.close()  # terminate all running procs
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 331, in close
+    self._close(death_sig=death_sig, timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 713, in _close
+    handler.proc.wait(time_to_wait)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1209, in wait
+    return self._wait(timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1953, in _wait
+    time.sleep(delay)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 33507 got signal: 15
+srun: error: ip-26-0-172-252: task 27: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 878, in _invoke_run
+    run_result = self._monitor_workers(self._worker_group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/local_elastic_agent.py", line 307, in _monitor_workers
+    result = self._pcontext.wait(0)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 288, in wait
+    return self._poll()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 668, in _poll
+    self.close()  # terminate all running procs
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 331, in close
+    self._close(death_sig=death_sig, timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 713, in _close
+    handler.proc.wait(time_to_wait)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1209, in wait
+    return self._wait(timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1953, in _wait
+    time.sleep(delay)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 27220 got signal: 15
+srun: error: ip-26-0-164-75: task 4: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 878, in _invoke_run
+    run_result = self._monitor_workers(self._worker_group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/local_elastic_agent.py", line 307, in _monitor_workers
+    result = self._pcontext.wait(0)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 288, in wait
+    return self._poll()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 668, in _poll
+    self.close()  # terminate all running procs
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 331, in close
+    self._close(death_sig=death_sig, timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 713, in _close
+    handler.proc.wait(time_to_wait)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1209, in wait
+    return self._wait(timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1953, in _wait
+    time.sleep(delay)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 263592 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 878, in _invoke_run
+    run_result = self._monitor_workers(self._worker_group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/local_elastic_agent.py", line 307, in _monitor_workers
+    result = self._pcontext.wait(0)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 288, in wait
+    return self._poll()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 668, in _poll
+    self.close()  # terminate all running procs
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 331, in close
+    self._close(death_sig=death_sig, timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 713, in _close
+    handler.proc.wait(time_to_wait)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1209, in wait
+    return self._wait(timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1953, in _wait
+    time.sleep(delay)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 136825 got signal: 15
+srun: error: ip-26-0-165-24: task 8: Exited with exit code 1
+srun: error: ip-26-0-169-207: task 19: Exited with exit code 1
+srun: error: ip-26-0-170-143: task 24: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 878, in _invoke_run
+    run_result = self._monitor_workers(self._worker_group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/local_elastic_agent.py", line 307, in _monitor_workers
+    result = self._pcontext.wait(0)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 288, in wait
+    return self._poll()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 668, in _poll
+    self.close()  # terminate all running procs
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 331, in close
+    self._close(death_sig=death_sig, timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 713, in _close
+    handler.proc.wait(time_to_wait)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1209, in wait
+    return self._wait(timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1953, in _wait
+    time.sleep(delay)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 108307 got signal: 15
+srun: error: ip-26-0-165-131: task 11: Exited with exit code 1
+[2025-01-03 08:11:56,125] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-166-15.ec2.internal_65704_0' has failed to send a keep-alive heartbeat to the rendezvous '14019312' due to an error of type RendezvousConnectionError.
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 878, in _invoke_run
+    run_result = self._monitor_workers(self._worker_group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/local_elastic_agent.py", line 307, in _monitor_workers
+    result = self._pcontext.wait(0)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 288, in wait
+    return self._poll()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 668, in _poll
+    self.close()  # terminate all running procs
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 331, in close
+    self._close(death_sig=death_sig, timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 713, in _close
+    handler.proc.wait(time_to_wait)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1209, in wait
+    return self._wait(timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1953, in _wait
+    time.sleep(delay)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 56881 got signal: 15
+[2025-01-03 08:11:56,292] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-173-121.ec2.internal_3398209_0' has failed to send a keep-alive heartbeat to the rendezvous '14019312' due to an error of type RendezvousConnectionError.
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 26277 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 878, in _invoke_run
+    run_result = self._monitor_workers(self._worker_group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/local_elastic_agent.py", line 307, in _monitor_workers
+    result = self._pcontext.wait(0)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 288, in wait
+    return self._poll()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 668, in _poll
+    self.close()  # terminate all running procs
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 331, in close
+    self._close(death_sig=death_sig, timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 713, in _close
+    handler.proc.wait(time_to_wait)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1209, in wait
+    return self._wait(timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1953, in _wait
+    time.sleep(delay)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 67298 got signal: 15
+srun: error: ip-26-0-164-187: task 5: Exited with exit code 1
+[2025-01-03 08:11:56,824] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-169-239.ec2.internal_254394_0' has failed to send a keep-alive heartbeat to the rendezvous '14019312' due to an error of type RendezvousConnectionError.
+[2025-01-03 08:11:56,835] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-170-132.ec2.internal_136843_0' has failed to send a keep-alive heartbeat to the rendezvous '14019312' due to an error of type RendezvousConnectionError.
+[2025-01-03 08:11:56,844] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-170-31.ec2.internal_778888_0' has failed to send a keep-alive heartbeat to the rendezvous '14019312' due to an error of type RendezvousConnectionError.
+[2025-01-03 08:11:56,882] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-166-36.ec2.internal_45002_0' has failed to send a keep-alive heartbeat to the rendezvous '14019312' due to an error of type RendezvousConnectionError.
+[2025-01-03 08:11:56,887] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-164-45.ec2.internal_455395_0' has failed to send a keep-alive heartbeat to the rendezvous '14019312' due to an error of type RendezvousConnectionError.
+[2025-01-03 08:11:56,907] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-165-202.ec2.internal_45826_0' has failed to send a keep-alive heartbeat to the rendezvous '14019312' due to an error of type RendezvousConnectionError.
+[2025-01-03 08:11:56,922] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-165-164.ec2.internal_117447_0' has failed to send a keep-alive heartbeat to the rendezvous '14019312' due to an error of type RendezvousConnectionError.
+srun: error: ip-26-0-165-59: task 10: Exited with exit code 1
+[2025-01-03 08:11:56,938] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-170-160.ec2.internal_276378_0' has failed to send a keep-alive heartbeat to the rendezvous '14019312' due to an error of type RendezvousConnectionError.
+[2025-01-03 08:11:56,953] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-164-236.ec2.internal_31657_0' has failed to send a keep-alive heartbeat to the rendezvous '14019312' due to an error of type RendezvousConnectionError.
+[2025-01-03 08:11:56,977] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-171-21.ec2.internal_76673_0' has failed to send a keep-alive heartbeat to the rendezvous '14019312' due to an error of type RendezvousConnectionError.
+[2025-01-03 08:11:56,991] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-164-18.ec2.internal_140917_0' has failed to send a keep-alive heartbeat to the rendezvous '14019312' due to an error of type RendezvousConnectionError.
+[2025-01-03 08:11:56,999] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-166-125.ec2.internal_520314_0' has failed to send a keep-alive heartbeat to the rendezvous '14019312' due to an error of type RendezvousConnectionError.
+[2025-01-03 08:11:57,009] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-165-213.ec2.internal_47466_0' has failed to send a keep-alive heartbeat to the rendezvous '14019312' due to an error of type RendezvousConnectionError.
+[2025-01-03 08:11:57,018] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-169-139.ec2.internal_27701_0' has failed to send a keep-alive heartbeat to the rendezvous '14019312' due to an error of type RendezvousConnectionError.
+[2025-01-03 08:11:57,036] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-169-247.ec2.internal_33138_0' has failed to send a keep-alive heartbeat to the rendezvous '14019312' due to an error of type RendezvousConnectionError.
+srun: error: ip-26-0-164-0: task 1: Exited with exit code 1
+[2025-01-03 08:11:57,048] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-173-246.ec2.internal_66481_0' has failed to send a keep-alive heartbeat to the rendezvous '14019312' due to an error of type RendezvousConnectionError.
+[2025-01-03 08:11:57,053] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-173-202.ec2.internal_83105_0' has failed to send a keep-alive heartbeat to the rendezvous '14019312' due to an error of type RendezvousConnectionError.
+[2025-01-03 08:11:57,084] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-165-38.ec2.internal_60952_0' has failed to send a keep-alive heartbeat to the rendezvous '14019312' due to an error of type RendezvousConnectionError.
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 878, in _invoke_run
+    run_result = self._monitor_workers(self._worker_group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/local_elastic_agent.py", line 307, in _monitor_workers
+    result = self._pcontext.wait(0)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 288, in wait
+    return self._poll()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 668, in _poll
+    self.close()  # terminate all running procs
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 331, in close
+    self._close(death_sig=death_sig, timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 713, in _close
+    handler.proc.wait(time_to_wait)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1209, in wait
+    return self._wait(timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1953, in _wait
+    time.sleep(delay)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 65704 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 45002 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 60952 got signal: 15
+srun: error: ip-26-0-166-15: task 15: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 47466 got signal: 15
+srun: error: ip-26-0-166-36: task 16: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 878, in _invoke_run
+    run_result = self._monitor_workers(self._worker_group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/local_elastic_agent.py", line 307, in _monitor_workers
+    result = self._pcontext.wait(0)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 288, in wait
+    return self._poll()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 668, in _poll
+    self.close()  # terminate all running procs
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 331, in close
+    self._close(death_sig=death_sig, timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 713, in _close
+    handler.proc.wait(time_to_wait)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1209, in wait
+    return self._wait(timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1953, in _wait
+    time.sleep(delay)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 520314 got signal: 15
+srun: error: ip-26-0-165-38: task 9: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 254394 got signal: 15
+srun: error: ip-26-0-165-213: task 14: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 83105 got signal: 15
+srun: error: ip-26-0-166-125: task 17: Exited with exit code 1
+srun: error: ip-26-0-169-239: task 20: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 276378 got signal: 15
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 3398209 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 45826 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 140917 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 31657 got signal: 15
+srun: error: ip-26-0-173-202: task 30: Exited with exit code 1
+srun: error: ip-26-0-170-160: task 25: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 117447 got signal: 15
+srun: error: ip-26-0-173-121: task 29: Exited with exit code 1
+srun: error: ip-26-0-164-18: task 2: Exited with exit code 1
+srun: error: ip-26-0-165-202: task 13: Exited with exit code 1
+srun: error: ip-26-0-164-236: task 7: Exited with exit code 1
+srun: error: ip-26-0-165-164: task 12: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 33138 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 66481 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 778888 got signal: 15
+srun: error: ip-26-0-169-247: task 21: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 27701 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 76673 got signal: 15
+srun: error: ip-26-0-173-246: task 31: Exited with exit code 1
+srun: error: ip-26-0-170-31: task 22: Exited with exit code 1
+srun: error: ip-26-0-171-21: task 26: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 136843 got signal: 15
+srun: error: ip-26-0-169-139: task 18: Exited with exit code 1
+srun: error: ip-26-0-170-132: task 23: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 455395 got signal: 15
+srun: error: ip-26-0-164-45: task 3: Exited with exit code 1
diff --git a/logs/14019313-bench_80G_dp4_tp32_pp2_acc32_mbs2_seq4096_zero1_tpmodeRED_vocab131k.out b/logs/14019313-bench_80G_dp4_tp32_pp2_acc32_mbs2_seq4096_zero1_tpmodeRED_vocab131k.out
new file mode 100644
index 0000000000000000000000000000000000000000..75bf8d51453846d5fd4bb8ebb3b51937aca5d1a8
--- /dev/null
+++ b/logs/14019313-bench_80G_dp4_tp32_pp2_acc32_mbs2_seq4096_zero1_tpmodeRED_vocab131k.out
@@ -0,0 +1,3168 @@
++ '[' -z 14019313 ']'
++ source /etc/profile.d/modules.sh
+++ . /usr/share/modules/init/bash
++++ unset _mlshdbg
++++ '[' 0 = 1 ']'
++++ unset _mlre _mlIFS
++++ '[' -n x ']'
++++ _mlIFS=' 	
+'
++++ IFS=' '
++++ '[' -n '' ']'
+++++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash autoinit
++++ _mlcode='module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+MODULEPATH_modshare=/etc/environment-modules/modules:1:/usr/share/modules/\$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1; export MODULEPATH_modshare;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+LOADEDMODULES=; export LOADEDMODULES;
+MODULEPATH=/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/\$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles; export MODULEPATH;
+test 0;'
++++ _mlret=0
++++ '[' -n x ']'
++++ IFS=' 	
+'
++++ unset _mlIFS
++++ unset _mlre _mlv _mlrv
++++ '[' 0 -eq 0 ']'
++++ eval 'module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+MODULEPATH_modshare=/etc/environment-modules/modules:1:/usr/share/modules/\$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1; export MODULEPATH_modshare;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+LOADEDMODULES=; export LOADEDMODULES;
+MODULEPATH=/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/\$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles; export MODULEPATH;
+test 0;'
+++++ MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl
+++++ export MODULES_CMD
+++++ ENV=/usr/share/modules/init/profile.sh
+++++ export ENV
+++++ MODULEPATH_modshare='/etc/environment-modules/modules:1:/usr/share/modules/$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1'
+++++ export MODULEPATH_modshare
+++++ BASH_ENV=/usr/share/modules/init/bash
+++++ export BASH_ENV
+++++ MODULESHOME=/usr/share/modules
+++++ export MODULESHOME
+++++ LOADEDMODULES=
+++++ export LOADEDMODULES
+++++ MODULEPATH='/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles'
+++++ export MODULEPATH
+++++ test 0
++++ '[' 0 = 1 ']'
++++ '[' -t 2 ']'
++++ export -f module
++++ export -f switchml
++++ '[' 5 -ge 3 ']'
++++ [[ ehxB =~ i ]]
++++ [[ ! :/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin: =~ :/usr/bin: ]]
++++ '[' '!' -n '' ']'
++++ MANPATH=:
++++ export MANPATH
+++++ manpath
++++ [[ ! :/fsx/nouamane/miniconda/envs/2-1-cu121/man:/fsx/nouamane/miniconda/envs/2-1-cu121/share/man:/fsx/nouamane/miniconda/man:/fsx/nouamane/miniconda/share/man:/usr/local/man:/usr/local/share/man:/usr/share/man:/opt/slurm/share/man:/admin/home/nouamane/.local/share/man:/admin/home/nouamane/.fzf/man:: =~ :/usr/share/man: ]]
++++ unset _mlcode _mlret
++++ '[' -n '' ']'
++ module load cuda/12.1
++ unset _mlshdbg
++ '[' 0 = 1 ']'
++ unset _mlre _mlIFS
++ '[' -n x ']'
++ _mlIFS=' 	
+'
++ IFS=' '
++ '[' -n '' ']'
+++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash load cuda/12.1
++ eval 'CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include;' export 'CPATH;
+LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:;' export 'LD_LIBRARY_PATH;
+FI_EFA_FORK_SAFE=1;' export 'FI_EFA_FORK_SAFE;
+MANPATH=/usr/local/cuda-12.1/share/man::;' export 'MANPATH;
+LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64;' export 'LIBRARY_PATH;
+_LMFILES_=/usr/share/modules/modulefiles/cuda/12.1;' export '_LMFILES_;
+LOADEDMODULES=cuda/12.1;' export 'LOADEDMODULES;
+MPI_PATH=/opt/amazon/openmpi;' export 'MPI_PATH;
+NCCL_HOME_modshare=/opt/nccl/build:1;' export 'NCCL_HOME_modshare;
+NCCL_PROTO=simple;' export 'NCCL_PROTO;
+MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1;' export 'MANPATH_modshare;
+LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1;' export 'LIBRARY_PATH_modshare;
+NCCL_SOCKET_IFNAME=enp;' export 'NCCL_SOCKET_IFNAME;
+AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl;' export 'AWS_OFI_NCCL_HOME;
+NCCL_HOME=/opt/nccl/build;' export 'NCCL_HOME;
+FI_PROVIDER=efa;' export 'FI_PROVIDER;
+AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1;' export 'AWS_OFI_NCCL_HOME_modshare;
+CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1;' export 'CPATH_modshare;
+LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2::1:/opt/nccl/build/lib:1:/opt/aws-ofi-nccl/lib:2:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/efa/lib:1;' export 'LD_LIBRARY_PATH_modshare;
+FI_EFA_ENABLE_SHM_TRANSFER=1;' export 'FI_EFA_ENABLE_SHM_TRANSFER;
+_LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1;' export '_LMFILES__modshare;
+LOADEDMODULES_modshare=cuda/12.1:1;' export 'LOADEDMODULES_modshare;
+MPI_PATH_modshare=/opt/amazon/openmpi:1;' export 'MPI_PATH_modshare;
+PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin;' export 'PATH;
+CUDA_HOME=/usr/local/cuda-12.1;' export 'CUDA_HOME;
+PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/opt/slurm/bin:1:/usr/sbin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1;' export 'PATH_modshare;
+test' '0;'
+++ CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include
+++ export CPATH
+++ LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:
+++ export LD_LIBRARY_PATH
+++ FI_EFA_FORK_SAFE=1
+++ export FI_EFA_FORK_SAFE
+++ MANPATH=/usr/local/cuda-12.1/share/man::
+++ export MANPATH
+++ LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64
+++ export LIBRARY_PATH
+++ _LMFILES_=/usr/share/modules/modulefiles/cuda/12.1
+++ export _LMFILES_
+++ LOADEDMODULES=cuda/12.1
+++ export LOADEDMODULES
+++ MPI_PATH=/opt/amazon/openmpi
+++ export MPI_PATH
+++ NCCL_HOME_modshare=/opt/nccl/build:1
+++ export NCCL_HOME_modshare
+++ NCCL_PROTO=simple
+++ export NCCL_PROTO
+++ MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1
+++ export MANPATH_modshare
+++ LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1
+++ export LIBRARY_PATH_modshare
+++ NCCL_SOCKET_IFNAME=enp
+++ export NCCL_SOCKET_IFNAME
+++ AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl
+++ export AWS_OFI_NCCL_HOME
+++ NCCL_HOME=/opt/nccl/build
+++ export NCCL_HOME
+++ FI_PROVIDER=efa
+++ export FI_PROVIDER
+++ AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1
+++ export AWS_OFI_NCCL_HOME_modshare
+++ CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1
+++ export CPATH_modshare
+++ LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2::1:/opt/nccl/build/lib:1:/opt/aws-ofi-nccl/lib:2:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/efa/lib:1
+++ export LD_LIBRARY_PATH_modshare
+++ FI_EFA_ENABLE_SHM_TRANSFER=1
+++ export FI_EFA_ENABLE_SHM_TRANSFER
+++ _LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1
+++ export _LMFILES__modshare
+++ LOADEDMODULES_modshare=cuda/12.1:1
+++ export LOADEDMODULES_modshare
+++ MPI_PATH_modshare=/opt/amazon/openmpi:1
+++ export MPI_PATH_modshare
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
+++ export PATH
+++ CUDA_HOME=/usr/local/cuda-12.1
+++ export CUDA_HOME
+++ PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/opt/slurm/bin:1:/usr/sbin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1
+++ export PATH_modshare
+++ test 0
++ _mlstatus=0
++ '[' -n x ']'
++ IFS=' 	
+'
++ unset _mlre _mlv _mlrv _mlIFS
++ '[' -n '' ']'
++ unset _mlshdbg
++ return 0
++ source /fsx/nouamane/miniconda/bin/activate
+++ _CONDA_ROOT=/fsx/nouamane/miniconda
+++ . /fsx/nouamane/miniconda/etc/profile.d/conda.sh
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ '[' -z x ']'
+++ conda activate
+++ local cmd=activate
+++ case "$cmd" in
+++ __conda_activate activate
+++ '[' -n '' ']'
+++ local ask_conda
++++ PS1=
++++ __conda_exe shell.posix activate
++++ /fsx/nouamane/miniconda/bin/conda shell.posix activate
+Error while loading conda entry point: conda-libmamba-solver (module 'libmambapy' has no attribute 'QueryFormat')
+++ ask_conda='. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''2'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_1='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/etc/conda/activate.d/libxml2_activate.sh"'
+++ eval '. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''2'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_1='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/etc/conda/activate.d/libxml2_activate.sh"'
++++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh
+++++ test -n ''
+++++ unset XML_CATALOG_FILES
+++++ unset xml_catalog_files_libxml2
++++ PS1='(base) '
++++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
++++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
++++ export CONDA_PREFIX=/fsx/nouamane/miniconda
++++ CONDA_PREFIX=/fsx/nouamane/miniconda
++++ export CONDA_SHLVL=2
++++ CONDA_SHLVL=2
++++ export CONDA_DEFAULT_ENV=base
++++ CONDA_DEFAULT_ENV=base
++++ export 'CONDA_PROMPT_MODIFIER=(base) '
++++ CONDA_PROMPT_MODIFIER='(base) '
++++ export CONDA_PREFIX_1=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ CONDA_PREFIX_1=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ . /fsx/nouamane/miniconda/etc/conda/activate.d/libxml2_activate.sh
+++++ test -n ''
+++++ xml_catalog_files_libxml2=
+++++ XML_CATALOG_FILES=
+++++ conda_catalog_files=
+++++ ifs_libxml2=' 	
+'
+++++ IFS=' '
+++++ rem=/fsx/nouamane/miniconda
+++++ for pre in ${rem}
+++++ test '' = /fsx/nouamane/miniconda
+++++ conda_catalog_files=/fsx/nouamane/miniconda
+++++ rem=
+++++ IFS=' 	
+'
+++++ conda_catalog_files='file:///fsx/nouamane/miniconda/etc/xml/catalog file:///etc/xml/catalog'
+++++ export 'XML_CATALOG_FILES=file:///fsx/nouamane/miniconda/etc/xml/catalog file:///etc/xml/catalog'
+++++ XML_CATALOG_FILES='file:///fsx/nouamane/miniconda/etc/xml/catalog file:///etc/xml/catalog'
+++++ unset conda_catalog_files ifs_libxml2 rem
+++ __conda_hashr
+++ '[' -n '' ']'
+++ '[' -n '' ']'
+++ hash -r
++ conda activate 2-1-cu121
++ local cmd=activate
++ case "$cmd" in
++ __conda_activate activate 2-1-cu121
++ '[' -n '' ']'
++ local ask_conda
+++ PS1='(base) '
+++ __conda_exe shell.posix activate 2-1-cu121
+++ /fsx/nouamane/miniconda/bin/conda shell.posix activate 2-1-cu121
+Error while loading conda entry point: conda-libmamba-solver (module 'libmambapy' has no attribute 'QueryFormat')
++ ask_conda='. "/fsx/nouamane/miniconda/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_2='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
++ eval '. "/fsx/nouamane/miniconda/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_2='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
+++ . /fsx/nouamane/miniconda/etc/conda/deactivate.d/libxml2_deactivate.sh
++++ test -n ''
++++ unset XML_CATALOG_FILES
++++ unset xml_catalog_files_libxml2
+++ PS1='(2-1-cu121) '
+++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
+++ export CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ export CONDA_SHLVL=3
+++ CONDA_SHLVL=3
+++ export CONDA_DEFAULT_ENV=2-1-cu121
+++ CONDA_DEFAULT_ENV=2-1-cu121
+++ export 'CONDA_PROMPT_MODIFIER=(2-1-cu121) '
+++ CONDA_PROMPT_MODIFIER='(2-1-cu121) '
+++ export CONDA_PREFIX_2=/fsx/nouamane/miniconda
+++ CONDA_PREFIX_2=/fsx/nouamane/miniconda
+++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ export _CE_M=
+++ _CE_M=
+++ export _CE_CONDA=
+++ _CE_CONDA=
+++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh
++++ test -n ''
++++ xml_catalog_files_libxml2=
++++ XML_CATALOG_FILES=
++++ conda_catalog_files=
++++ ifs_libxml2=' 	
+'
++++ IFS=' '
++++ rem=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ for pre in ${rem}
++++ test '' = /fsx/nouamane/miniconda/envs/2-1-cu121
++++ conda_catalog_files=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ rem=
++++ IFS=' 	
+'
++++ conda_catalog_files='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ export 'XML_CATALOG_FILES=file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ XML_CATALOG_FILES='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ unset conda_catalog_files ifs_libxml2 rem
++ __conda_hashr
++ '[' -n '' ']'
++ '[' -n '' ']'
++ hash -r
++ export PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
++ PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
++ '[' -z '' ']'
+++ scontrol show hostnames 'ip-26-0-163-236,ip-26-0-164-[0,18,45,75,187,207,236],ip-26-0-165-[24,38,59,131,164,202,213],ip-26-0-166-[15,36,125],ip-26-0-169-[139,207,239,247],ip-26-0-170-[31,132,143,160],ip-26-0-171-21,ip-26-0-172-252,ip-26-0-173-[7,121,202,246]'
++ export 'NODELIST=ip-26-0-163-236
+ip-26-0-164-0
+ip-26-0-164-18
+ip-26-0-164-45
+ip-26-0-164-75
+ip-26-0-164-187
+ip-26-0-164-207
+ip-26-0-164-236
+ip-26-0-165-24
+ip-26-0-165-38
+ip-26-0-165-59
+ip-26-0-165-131
+ip-26-0-165-164
+ip-26-0-165-202
+ip-26-0-165-213
+ip-26-0-166-15
+ip-26-0-166-36
+ip-26-0-166-125
+ip-26-0-169-139
+ip-26-0-169-207
+ip-26-0-169-239
+ip-26-0-169-247
+ip-26-0-170-31
+ip-26-0-170-132
+ip-26-0-170-143
+ip-26-0-170-160
+ip-26-0-171-21
+ip-26-0-172-252
+ip-26-0-173-7
+ip-26-0-173-121
+ip-26-0-173-202
+ip-26-0-173-246'
++ NODELIST='ip-26-0-163-236
+ip-26-0-164-0
+ip-26-0-164-18
+ip-26-0-164-45
+ip-26-0-164-75
+ip-26-0-164-187
+ip-26-0-164-207
+ip-26-0-164-236
+ip-26-0-165-24
+ip-26-0-165-38
+ip-26-0-165-59
+ip-26-0-165-131
+ip-26-0-165-164
+ip-26-0-165-202
+ip-26-0-165-213
+ip-26-0-166-15
+ip-26-0-166-36
+ip-26-0-166-125
+ip-26-0-169-139
+ip-26-0-169-207
+ip-26-0-169-239
+ip-26-0-169-247
+ip-26-0-170-31
+ip-26-0-170-132
+ip-26-0-170-143
+ip-26-0-170-160
+ip-26-0-171-21
+ip-26-0-172-252
+ip-26-0-173-7
+ip-26-0-173-121
+ip-26-0-173-202
+ip-26-0-173-246'
+++ head -n1
+++ scontrol show hostnames 'ip-26-0-163-236,ip-26-0-164-[0,18,45,75,187,207,236],ip-26-0-165-[24,38,59,131,164,202,213],ip-26-0-166-[15,36,125],ip-26-0-169-[139,207,239,247],ip-26-0-170-[31,132,143,160],ip-26-0-171-21,ip-26-0-172-252,ip-26-0-173-[7,121,202,246]'
++ export MASTER_NODE=ip-26-0-163-236
++ MASTER_NODE=ip-26-0-163-236
++ export MASTER_PORT=12356
++ MASTER_PORT=12356
++ export NNODES=32
++ NNODES=32
++ export GPUS_PER_NODE=8
++ GPUS_PER_NODE=8
++ export WORLD_SIZE=256
++ WORLD_SIZE=256
++ export CUDA_DEVICE_MAX_CONNECTIONS=1
++ CUDA_DEVICE_MAX_CONNECTIONS=1
++ export NCCL_DEBUG=WARN
++ NCCL_DEBUG=WARN
++ export NANOTRON_BENCHMARK=1
++ NANOTRON_BENCHMARK=1
++ export WANDB_MODE=disabled
++ WANDB_MODE=disabled
++ export TORCH_NCCL_ASYNC_ERROR_HANDLING=1
++ TORCH_NCCL_ASYNC_ERROR_HANDLING=1
++ export TORCH_DISTRIBUTED_DEBUG=DETAIL
++ TORCH_DISTRIBUTED_DEBUG=DETAIL
++ '[' -z '' ']'
++ echo '=== GPU Topology ==='
+=== GPU Topology ===
++ nvidia-smi topo -m
+	GPU0	GPU1	GPU2	GPU3	GPU4	GPU5	GPU6	GPU7	CPU Affinity	NUMA Affinity	GPU NUMA ID
+GPU0	 X 	NV18	NV18	NV18	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU1	NV18	 X 	NV18	NV18	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU2	NV18	NV18	 X 	NV18	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU3	NV18	NV18	NV18	 X 	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU4	NV18	NV18	NV18	NV18	 X 	NV18	NV18	NV18	48-95	1		N/A
+GPU5	NV18	NV18	NV18	NV18	NV18	 X 	NV18	NV18	48-95	1		N/A
+GPU6	NV18	NV18	NV18	NV18	NV18	NV18	 X 	NV18	48-95	1		N/A
+GPU7	NV18	NV18	NV18	NV18	NV18	NV18	NV18	 X 	48-95	1		N/A
+
+Legend:
+
+  X    = Self
+  SYS  = Connection traversing PCIe as well as the SMP interconnect between NUMA nodes (e.g., QPI/UPI)
+  NODE = Connection traversing PCIe as well as the interconnect between PCIe Host Bridges within a NUMA node
+  PHB  = Connection traversing PCIe as well as a PCIe Host Bridge (typically the CPU)
+  PXB  = Connection traversing multiple PCIe bridges (without traversing the PCIe Host Bridge)
+  PIX  = Connection traversing at most a single PCIe bridge
+  NV#  = Connection traversing a bonded set of # NVLinks
++ echo ==================
+==================
++ export SRUN_ALLOC_ARGS=
++ SRUN_ALLOC_ARGS=
++ echo 'Master node: ip-26-0-163-236'
+Master node: ip-26-0-163-236
++ echo 'All nodes: ip-26-0-163-236
+ip-26-0-164-0
+ip-26-0-164-18
+ip-26-0-164-45
+ip-26-0-164-75
+ip-26-0-164-187
+ip-26-0-164-207
+ip-26-0-164-236
+ip-26-0-165-24
+ip-26-0-165-38
+ip-26-0-165-59
+ip-26-0-165-131
+ip-26-0-165-164
+ip-26-0-165-202
+ip-26-0-165-213
+ip-26-0-166-15
+ip-26-0-166-36
+ip-26-0-166-125
+ip-26-0-169-139
+ip-26-0-169-207
+ip-26-0-169-239
+ip-26-0-169-247
+ip-26-0-170-31
+ip-26-0-170-132
+ip-26-0-170-143
+ip-26-0-170-160
+ip-26-0-171-21
+ip-26-0-172-252
+ip-26-0-173-7
+ip-26-0-173-121
+ip-26-0-173-202
+ip-26-0-173-246'
+All nodes: ip-26-0-163-236
+ip-26-0-164-0
+ip-26-0-164-18
+ip-26-0-164-45
+ip-26-0-164-75
+ip-26-0-164-187
+ip-26-0-164-207
+ip-26-0-164-236
+ip-26-0-165-24
+ip-26-0-165-38
+ip-26-0-165-59
+ip-26-0-165-131
+ip-26-0-165-164
+ip-26-0-165-202
+ip-26-0-165-213
+ip-26-0-166-15
+ip-26-0-166-36
+ip-26-0-166-125
+ip-26-0-169-139
+ip-26-0-169-207
+ip-26-0-169-239
+ip-26-0-169-247
+ip-26-0-170-31
+ip-26-0-170-132
+ip-26-0-170-143
+ip-26-0-170-160
+ip-26-0-171-21
+ip-26-0-172-252
+ip-26-0-173-7
+ip-26-0-173-121
+ip-26-0-173-202
+ip-26-0-173-246
++ echo 'World size: 256'
+World size: 256
++ '[' -n '' ']'
++ srun --wait=0 --kill-on-bad-exit=1 torchrun --nnodes=32 --nproc_per_node=8 --rdzv_id=14019313 --rdzv_backend=c10d --rdzv_endpoint=ip-26-0-163-236:12356 --max_restarts 0 --rdzv_conf timeout=60 /fsx/nouamane/projects/nanotron/run_train.py --config-file benchmark/configs/config_80G_dp4_tp32_pp2_acc32_mbs2_seq4096_zero1_tpmodeRED_vocab131k.yaml
+[2025-01-03 08:12:10,262] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 08:12:10,261] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 08:12:10,261] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 08:12:10,262] torch.distributed.run: [WARNING] 
+[2025-01-03 08:12:10,262] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 08:12:10,262] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 08:12:10,262] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 08:12:10,262] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 08:12:10,261] torch.distributed.run: [WARNING] 
+[2025-01-03 08:12:10,261] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 08:12:10,261] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 08:12:10,261] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 08:12:10,261] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 08:12:10,262] torch.distributed.run: [WARNING] 
+[2025-01-03 08:12:10,262] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 08:12:10,262] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 08:12:10,262] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 08:12:10,262] torch.distributed.run: [WARNING] 
+[2025-01-03 08:12:10,262] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 08:12:10,262] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 08:12:10,262] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 08:12:10,261] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 08:12:10,261] torch.distributed.run: [WARNING] 
+[2025-01-03 08:12:10,261] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 08:12:10,261] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 08:12:10,261] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 08:12:10,261] torch.distributed.run: [WARNING] 
+[2025-01-03 08:12:10,261] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 08:12:10,261] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 08:12:10,261] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 08:12:10,262] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 08:12:10,262] torch.distributed.run: [WARNING] 
+[2025-01-03 08:12:10,262] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 08:12:10,262] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 08:12:10,262] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 08:12:10,263] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 08:12:10,263] torch.distributed.run: [WARNING] 
+[2025-01-03 08:12:10,263] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 08:12:10,263] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 08:12:10,263] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 08:12:10,263] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 08:12:10,264] torch.distributed.run: [WARNING] 
+[2025-01-03 08:12:10,264] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 08:12:10,264] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 08:12:10,264] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 08:12:10,263] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 08:12:10,264] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 08:12:10,263] torch.distributed.run: [WARNING] 
+[2025-01-03 08:12:10,263] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 08:12:10,263] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 08:12:10,263] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 08:12:10,264] torch.distributed.run: [WARNING] 
+[2025-01-03 08:12:10,264] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 08:12:10,264] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 08:12:10,264] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 08:12:10,264] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 08:12:10,264] torch.distributed.run: [WARNING] 
+[2025-01-03 08:12:10,264] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 08:12:10,264] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 08:12:10,264] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 08:12:10,264] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 08:12:10,264] torch.distributed.run: [WARNING] 
+[2025-01-03 08:12:10,264] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 08:12:10,264] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 08:12:10,264] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 08:12:10,265] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 08:12:10,265] torch.distributed.run: [WARNING] 
+[2025-01-03 08:12:10,265] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 08:12:10,265] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 08:12:10,265] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 08:12:10,268] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 08:12:10,268] torch.distributed.run: [WARNING] 
+[2025-01-03 08:12:10,268] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 08:12:10,268] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 08:12:10,268] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 08:12:10,267] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 08:12:10,267] torch.distributed.run: [WARNING] 
+[2025-01-03 08:12:10,267] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 08:12:10,267] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 08:12:10,267] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 08:12:10,270] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 08:12:10,270] torch.distributed.run: [WARNING] 
+[2025-01-03 08:12:10,270] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 08:12:10,270] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 08:12:10,270] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 08:12:10,268] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 08:12:10,268] torch.distributed.run: [WARNING] 
+[2025-01-03 08:12:10,268] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 08:12:10,268] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 08:12:10,268] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 08:12:10,269] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 08:12:10,269] torch.distributed.run: [WARNING] 
+[2025-01-03 08:12:10,269] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 08:12:10,269] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 08:12:10,269] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 08:12:10,270] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 08:12:10,270] torch.distributed.run: [WARNING] 
+[2025-01-03 08:12:10,270] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 08:12:10,270] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 08:12:10,270] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 08:12:10,270] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 08:12:10,270] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 08:12:10,270] torch.distributed.run: [WARNING] 
+[2025-01-03 08:12:10,270] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 08:12:10,270] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 08:12:10,270] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 08:12:10,271] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 08:12:10,271] torch.distributed.run: [WARNING] 
+[2025-01-03 08:12:10,271] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 08:12:10,271] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 08:12:10,271] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 08:12:10,271] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 08:12:10,271] torch.distributed.run: [WARNING] 
+[2025-01-03 08:12:10,271] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 08:12:10,271] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 08:12:10,271] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 08:12:10,273] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 08:12:10,273] torch.distributed.run: [WARNING] 
+[2025-01-03 08:12:10,273] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 08:12:10,273] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 08:12:10,273] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 08:12:10,283] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 08:12:10,283] torch.distributed.run: [WARNING] 
+[2025-01-03 08:12:10,283] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 08:12:10,283] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 08:12:10,283] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 08:12:10,294] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 08:12:10,294] torch.distributed.run: [WARNING] 
+[2025-01-03 08:12:10,294] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 08:12:10,294] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 08:12:10,294] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 08:12:10,306] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 08:12:10,307] torch.distributed.run: [WARNING] 
+[2025-01-03 08:12:10,307] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 08:12:10,307] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 08:12:10,307] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 08:12:10,311] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 08:12:10,312] torch.distributed.run: [WARNING] 
+[2025-01-03 08:12:10,312] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 08:12:10,312] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 08:12:10,312] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 08:12:10,270] torch.distributed.run: [WARNING] 
+[2025-01-03 08:12:10,270] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 08:12:10,270] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 08:12:10,270] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 08:12:10,364] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 08:12:10,364] torch.distributed.run: [WARNING] 
+[2025-01-03 08:12:10,364] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 08:12:10,364] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 08:12:10,364] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 08:12:10,551] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 08:12:10,551] torch.distributed.run: [WARNING] 
+[2025-01-03 08:12:10,551] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 08:12:10,551] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 08:12:10,551] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 08:12:10,601] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 08:12:10,601] torch.distributed.run: [WARNING] 
+[2025-01-03 08:12:10,601] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 08:12:10,601] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 08:12:10,601] torch.distributed.run: [WARNING] *****************************************
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+01/03/2025 08:13:09 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: Config:
+01/03/2025 08:13:09 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: Config(general=GeneralArgs(project='debug',
+01/03/2025 08:13:09 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                            run='80G_dp4_tp32_pp2_acc32_mbs2_seq4096_zero1_tpmodeRED_vocab131k',
+01/03/2025 08:13:09 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                            seed=42,
+01/03/2025 08:13:09 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                            step=None,
+01/03/2025 08:13:09 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                            consumed_train_samples=None,
+01/03/2025 08:13:09 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                            benchmark_csv_path=PosixPath('benchmark/results/bench_final2.csv'),
+01/03/2025 08:13:09 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                            ignore_sanity_checks=True),
+01/03/2025 08:13:09 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:        parallelism=ParallelismArgs(dp=4,
+01/03/2025 08:13:09 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                    pp=2,
+01/03/2025 08:13:09 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                    tp=32,
+01/03/2025 08:13:09 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                    pp_engine=<nanotron.parallel.pipeline_parallel.engine.OneForwardOneBackwardPipelineEngine object at 0x7f7670f88c10>,
+01/03/2025 08:13:09 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                    tp_mode=<TensorParallelLinearMode.REDUCE_SCATTER: 2>,
+01/03/2025 08:13:09 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                    tp_linear_async_communication=True,
+01/03/2025 08:13:09 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                    recompute_layer=False,
+01/03/2025 08:13:09 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                    tp_recompute_allgather=True,
+01/03/2025 08:13:09 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                    expert_parallel_size=1),
+01/03/2025 08:13:09 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:        model=ModelArgs(model_config=LlamaConfig(bos_token_id=0,
+01/03/2025 08:13:09 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                                 eos_token_id=0,
+01/03/2025 08:13:09 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                                 hidden_act='silu',
+01/03/2025 08:13:09 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                                 hidden_size=8192,
+01/03/2025 08:13:09 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                                 initializer_range=0.02,
+01/03/2025 08:13:09 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                                 intermediate_size=28672,
+01/03/2025 08:13:09 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                                 is_llama_config=True,
+01/03/2025 08:13:09 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                                 max_position_embeddings=4096,
+01/03/2025 08:13:09 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                                 num_attention_heads=64,
+01/03/2025 08:13:09 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                                 num_hidden_layers=80,
+01/03/2025 08:13:09 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                                 num_key_value_heads=64,
+01/03/2025 08:13:09 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                                 pad_token_id=None,
+01/03/2025 08:13:09 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                                 pretraining_tp=1,
+01/03/2025 08:13:09 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                                 rms_norm_eps=1e-05,
+01/03/2025 08:13:09 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                                 rope_scaling=None,
+01/03/2025 08:13:09 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                                 rope_theta=10000.0,
+01/03/2025 08:13:09 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                                 rope_interleaved=False,
+01/03/2025 08:13:09 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                                 tie_word_embeddings=False,
+01/03/2025 08:13:09 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                                 use_cache=True,
+01/03/2025 08:13:09 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                                 vocab_size=131072),
+01/03/2025 08:13:09 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                        init_method=RandomInit(std=0.02),
+01/03/2025 08:13:09 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                        dtype=torch.bfloat16,
+01/03/2025 08:13:09 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                        make_vocab_size_divisible_by=1,
+01/03/2025 08:13:09 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                        ddp_bucket_cap_mb=25),
+01/03/2025 08:13:09 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:        tokenizer=TokenizerArgs(tokenizer_name_or_path='robot-test/dummy-tokenizer-wordlevel',
+01/03/2025 08:13:09 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                tokenizer_revision=None,
+01/03/2025 08:13:09 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                tokenizer_max_length=None),
+01/03/2025 08:13:09 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:        checkpoints=CheckpointsArgs(checkpoints_path=PosixPath('checkpoints'),
+01/03/2025 08:13:09 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                    checkpoint_interval=10000,
+01/03/2025 08:13:09 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                    save_initial_state=False,
+01/03/2025 08:13:09 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                    save_final_state=False,
+01/03/2025 08:13:09 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                    resume_checkpoint_path=None,
+01/03/2025 08:13:09 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                    checkpoints_path_is_shared_file_system=False),
+01/03/2025 08:13:09 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:        logging=LoggingArgs(log_level='info',
+01/03/2025 08:13:09 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                            log_level_replica='info',
+01/03/2025 08:13:09 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                            iteration_step_info_interval=1),
+01/03/2025 08:13:09 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:        tokens=TokensArgs(sequence_length=4096,
+01/03/2025 08:13:09 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                          train_steps=100,
+01/03/2025 08:13:09 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                          micro_batch_size=2,
+01/03/2025 08:13:09 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                          batch_accumulation_per_replica=32,
+01/03/2025 08:13:09 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                          val_check_interval=100,
+01/03/2025 08:13:09 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                          limit_val_batches=0,
+01/03/2025 08:13:09 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                          limit_test_batches=0),
+01/03/2025 08:13:09 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:        optimizer=OptimizerArgs(optimizer_factory=AdamWOptimizerArgs(adam_eps=1e-08,
+01/03/2025 08:13:09 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                                                     adam_beta1=0.9,
+01/03/2025 08:13:09 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                                                     adam_beta2=0.95,
+01/03/2025 08:13:09 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                                                     torch_adam_is_fused=True,
+01/03/2025 08:13:09 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                                                     name='adamW'),
+01/03/2025 08:13:09 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                zero_stage=1,
+01/03/2025 08:13:09 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                weight_decay=0.01,
+01/03/2025 08:13:09 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                clip_grad=1.0,
+01/03/2025 08:13:09 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                accumulate_grad_in_fp32=True,
+01/03/2025 08:13:09 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                learning_rate_scheduler=LRSchedulerArgs(learning_rate=0.0003,
+01/03/2025 08:13:09 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                                                        lr_warmup_steps=2,
+01/03/2025 08:13:09 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                                                        lr_warmup_style='linear',
+01/03/2025 08:13:09 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                                                        lr_decay_style='cosine',
+01/03/2025 08:13:09 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                                                        lr_decay_steps=13,
+01/03/2025 08:13:09 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                                                        lr_decay_starting_step=None,
+01/03/2025 08:13:09 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                                                        min_decay_lr=1e-05)),
+01/03/2025 08:13:09 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:        data_stages=[DatasetStageArgs(name='Stable Training Stage',
+01/03/2025 08:13:09 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                      start_training_step=1,
+01/03/2025 08:13:09 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                      data=DataArgs(dataset=None,
+01/03/2025 08:13:09 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                                    seed=42,
+01/03/2025 08:13:09 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:                                                    num_loading_workers=1))],
+01/03/2025 08:13:09 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:        profiler=None,
+01/03/2025 08:13:09 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:        lighteval=None,
+01/03/2025 08:13:09 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:        s3_upload=None)
+01/03/2025 08:13:09 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: Model Config:
+01/03/2025 08:13:09 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: LlamaConfig(bos_token_id=0,
+01/03/2025 08:13:09 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:             eos_token_id=0,
+01/03/2025 08:13:09 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:             hidden_act='silu',
+01/03/2025 08:13:09 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:             hidden_size=8192,
+01/03/2025 08:13:09 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:             initializer_range=0.02,
+01/03/2025 08:13:09 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:             intermediate_size=28672,
+01/03/2025 08:13:09 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:             is_llama_config=True,
+01/03/2025 08:13:09 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:             max_position_embeddings=4096,
+01/03/2025 08:13:09 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:             num_attention_heads=64,
+01/03/2025 08:13:09 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:             num_hidden_layers=80,
+01/03/2025 08:13:09 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:             num_key_value_heads=64,
+01/03/2025 08:13:09 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:             pad_token_id=None,
+01/03/2025 08:13:09 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:             pretraining_tp=1,
+01/03/2025 08:13:09 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:             rms_norm_eps=1e-05,
+01/03/2025 08:13:09 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:             rope_scaling=None,
+01/03/2025 08:13:09 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:             rope_theta=10000.0,
+01/03/2025 08:13:09 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:             rope_interleaved=False,
+01/03/2025 08:13:09 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:             tie_word_embeddings=False,
+01/03/2025 08:13:09 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:             use_cache=True,
+01/03/2025 08:13:09 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:             vocab_size=131072)
+01/03/2025 08:13:09 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: Building model..
+01/03/2025 08:13:09 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: Initialize RoPE Theta = 10000.0
+01/03/2025 08:13:09 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: Setting PP block ranks...
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+01/03/2025 08:13:20 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: Total number of parameters: 80G (152656.50MiB)
+01/03/2025 08:13:20 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: Local number of parameters: 1.28G (2443.28MiB)
+01/03/2025 08:13:20 [INFO|DP=0|PP=0|TP=3|ip-26-0-163-236]: Local number of parameters: 1.28G (2443.28MiB)
+01/03/2025 08:13:20 [INFO|DP=0|PP=1|TP=0|ip-26-0-166-15]: Local number of parameters: 1.22G (2327.23MiB)
+01/03/2025 08:13:20 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: [After model building] Memory usage: 2443.33MiB. Peak allocated: 5384.00MiB Peak reserved: 10518.00MiB
+01/03/2025 08:13:20 [INFO|DP=0|PP=0|TP=3|ip-26-0-163-236]: [After model building] Memory usage: 2443.33MiB. Peak allocated: 5384.00MiB Peak reserved: 10518.00MiB
+01/03/2025 08:13:20 [INFO|DP=0|PP=1|TP=0|ip-26-0-166-15]: [After model building] Memory usage: 2327.28MiB. Peak allocated: 5384.00MiB Peak reserved: 10518.00MiB
+01/03/2025 08:13:20 [INFO|DP=0|PP=0|TP=30|ip-26-0-164-187]: Local number of parameters: 1.28G (2443.28MiB)
+01/03/2025 08:13:20 [INFO|DP=0|PP=0|TP=28|ip-26-0-164-187]: Local number of parameters: 1.28G (2443.28MiB)
+01/03/2025 08:13:20 [INFO|DP=0|PP=0|TP=30|ip-26-0-164-187]: [After model building] Memory usage: 2443.33MiB. Peak allocated: 5384.00MiB Peak reserved: 9494.00MiB
+01/03/2025 08:13:20 [INFO|DP=0|PP=0|TP=28|ip-26-0-164-187]: [After model building] Memory usage: 2443.33MiB. Peak allocated: 5384.00MiB Peak reserved: 9494.00MiB
+01/03/2025 08:13:20 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: No checkpoint path provided.
+01/03/2025 08:13:20 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: Parametrizing model parameters using StandardParametrizator
+01/03/2025 08:13:20 [INFO|DP=0|PP=0|TP=9|ip-26-0-164-0]: Local number of parameters: 1.28G (2443.28MiB)
+01/03/2025 08:13:20 [INFO|DP=0|PP=0|TP=12|ip-26-0-164-0]: Local number of parameters: 1.28G (2443.28MiB)
+01/03/2025 08:13:20 [INFO|DP=0|PP=0|TP=11|ip-26-0-164-0]: Local number of parameters: 1.28G (2443.28MiB)
+01/03/2025 08:13:20 [INFO|DP=0|PP=0|TP=9|ip-26-0-164-0]: [After model building] Memory usage: 2443.33MiB. Peak allocated: 5384.00MiB Peak reserved: 10518.00MiB
+01/03/2025 08:13:20 [INFO|DP=0|PP=0|TP=12|ip-26-0-164-0]: [After model building] Memory usage: 2443.33MiB. Peak allocated: 5384.00MiB Peak reserved: 10518.00MiB
+01/03/2025 08:13:20 [INFO|DP=0|PP=0|TP=11|ip-26-0-164-0]: [After model building] Memory usage: 2443.33MiB. Peak allocated: 5384.00MiB Peak reserved: 10518.00MiB
+01/03/2025 08:13:20 [INFO|DP=0|PP=0|TP=2|ip-26-0-163-236]: Local number of parameters: 1.28G (2443.28MiB)
+01/03/2025 08:13:20 [INFO|DP=0|PP=0|TP=2|ip-26-0-163-236]: [After model building] Memory usage: 2443.33MiB. Peak allocated: 5384.00MiB Peak reserved: 10518.00MiB
+01/03/2025 08:13:20 [INFO|DP=0|PP=1|TP=3|ip-26-0-166-15]: Local number of parameters: 1.22G (2327.23MiB)
+01/03/2025 08:13:20 [INFO|DP=0|PP=0|TP=7|ip-26-0-163-236]: Local number of parameters: 1.28G (2443.28MiB)
+01/03/2025 08:13:20 [INFO|DP=0|PP=1|TP=7|ip-26-0-166-15]: Local number of parameters: 1.22G (2327.23MiB)
+01/03/2025 08:13:20 [INFO|DP=0|PP=1|TP=3|ip-26-0-166-15]: [After model building] Memory usage: 2327.28MiB. Peak allocated: 5384.00MiB Peak reserved: 10518.00MiB
+01/03/2025 08:13:20 [INFO|DP=0|PP=0|TP=5|ip-26-0-163-236]: Local number of parameters: 1.28G (2443.28MiB)
+01/03/2025 08:13:20 [INFO|DP=0|PP=0|TP=6|ip-26-0-163-236]: Local number of parameters: 1.28G (2443.28MiB)
+01/03/2025 08:13:20 [INFO|DP=0|PP=0|TP=1|ip-26-0-163-236]: Local number of parameters: 1.28G (2443.28MiB)
+01/03/2025 08:13:20 [INFO|DP=0|PP=1|TP=1|ip-26-0-166-15]: Local number of parameters: 1.22G (2327.23MiB)
+01/03/2025 08:13:20 [INFO|DP=0|PP=0|TP=7|ip-26-0-163-236]: [After model building] Memory usage: 2443.33MiB. Peak allocated: 5384.00MiB Peak reserved: 10518.00MiB
+01/03/2025 08:13:20 [INFO|DP=0|PP=1|TP=7|ip-26-0-166-15]: [After model building] Memory usage: 2327.28MiB. Peak allocated: 5384.00MiB Peak reserved: 10518.00MiB
+01/03/2025 08:13:20 [INFO|DP=0|PP=0|TP=4|ip-26-0-163-236]: Local number of parameters: 1.28G (2443.28MiB)
+01/03/2025 08:13:20 [INFO|DP=0|PP=0|TP=5|ip-26-0-163-236]: [After model building] Memory usage: 2443.33MiB. Peak allocated: 5384.00MiB Peak reserved: 10518.00MiB
+01/03/2025 08:13:20 [INFO|DP=0|PP=1|TP=1|ip-26-0-166-15]: [After model building] Memory usage: 2327.28MiB. Peak allocated: 5384.00MiB Peak reserved: 10518.00MiB
+01/03/2025 08:13:20 [INFO|DP=0|PP=0|TP=1|ip-26-0-163-236]: [After model building] Memory usage: 2443.33MiB. Peak allocated: 5384.00MiB Peak reserved: 10518.00MiB
+01/03/2025 08:13:20 [INFO|DP=0|PP=0|TP=6|ip-26-0-163-236]: [After model building] Memory usage: 2443.33MiB. Peak allocated: 5384.00MiB Peak reserved: 10518.00MiB
+01/03/2025 08:13:20 [INFO|DP=0|PP=0|TP=4|ip-26-0-163-236]: [After model building] Memory usage: 2443.33MiB. Peak allocated: 5384.00MiB Peak reserved: 10518.00MiB
+01/03/2025 08:13:20 [INFO|DP=0|PP=0|TP=8|ip-26-0-164-0]: Local number of parameters: 1.28G (2443.28MiB)
+01/03/2025 08:13:20 [INFO|DP=0|PP=1|TP=30|ip-26-0-169-207]: Local number of parameters: 1.22G (2327.23MiB)
+01/03/2025 08:13:20 [INFO|DP=0|PP=1|TP=11|ip-26-0-166-36]: Local number of parameters: 1.22G (2327.23MiB)
+01/03/2025 08:13:20 [INFO|DP=0|PP=0|TP=8|ip-26-0-164-0]: [After model building] Memory usage: 2443.33MiB. Peak allocated: 5384.00MiB Peak reserved: 10518.00MiB
+01/03/2025 08:13:20 [INFO|DP=0|PP=0|TP=21|ip-26-0-164-18]: Local number of parameters: 1.28G (2443.28MiB)
+01/03/2025 08:13:20 [INFO|DP=0|PP=1|TP=30|ip-26-0-169-207]: [After model building] Memory usage: 2327.28MiB. Peak allocated: 5384.00MiB Peak reserved: 10518.00MiB
+01/03/2025 08:13:20 [INFO|DP=0|PP=1|TP=11|ip-26-0-166-36]: [After model building] Memory usage: 2327.28MiB. Peak allocated: 5384.00MiB Peak reserved: 10518.00MiB
+01/03/2025 08:13:20 [INFO|DP=0|PP=1|TP=28|ip-26-0-169-207]: Local number of parameters: 1.22G (2327.23MiB)
+01/03/2025 08:13:20 [INFO|DP=0|PP=0|TP=21|ip-26-0-164-18]: [After model building] Memory usage: 2443.33MiB. Peak allocated: 5384.00MiB Peak reserved: 10518.00MiB
+01/03/2025 08:13:20 [INFO|DP=0|PP=1|TP=19|ip-26-0-169-139]: Local number of parameters: 1.22G (2327.23MiB)
+01/03/2025 08:13:20 [INFO|DP=0|PP=0|TP=19|ip-26-0-164-18]: Local number of parameters: 1.28G (2443.28MiB)
+01/03/2025 08:13:20 [INFO|DP=0|PP=0|TP=22|ip-26-0-164-18]: Local number of parameters: 1.28G (2443.28MiB)
+01/03/2025 08:13:20 [INFO|DP=0|PP=0|TP=25|ip-26-0-164-187]: Local number of parameters: 1.28G (2443.28MiB)
+01/03/2025 08:13:20 [INFO|DP=0|PP=1|TP=28|ip-26-0-169-207]: [After model building] Memory usage: 2327.28MiB. Peak allocated: 5384.00MiB Peak reserved: 10518.00MiB
+01/03/2025 08:13:20 [INFO|DP=0|PP=0|TP=29|ip-26-0-164-187]: Local number of parameters: 1.28G (2443.28MiB)
+01/03/2025 08:13:20 [INFO|DP=0|PP=0|TP=16|ip-26-0-164-18]: Local number of parameters: 1.28G (2443.28MiB)
+01/03/2025 08:13:20 [INFO|DP=0|PP=0|TP=17|ip-26-0-164-18]: Local number of parameters: 1.28G (2443.28MiB)
+01/03/2025 08:13:20 [INFO|DP=0|PP=1|TP=19|ip-26-0-169-139]: [After model building] Memory usage: 2327.28MiB. Peak allocated: 5384.00MiB Peak reserved: 10518.00MiB
+01/03/2025 08:13:20 [INFO|DP=0|PP=0|TP=24|ip-26-0-164-187]: Local number of parameters: 1.28G (2443.28MiB)
+01/03/2025 08:13:20 [INFO|DP=0|PP=0|TP=20|ip-26-0-164-18]: Local number of parameters: 1.28G (2443.28MiB)
+01/03/2025 08:13:20 [INFO|DP=0|PP=0|TP=19|ip-26-0-164-18]: [After model building] Memory usage: 2443.33MiB. Peak allocated: 5384.00MiB Peak reserved: 10518.00MiB
+01/03/2025 08:13:20 [INFO|DP=0|PP=0|TP=23|ip-26-0-164-18]: Local number of parameters: 1.28G (2443.28MiB)
+01/03/2025 08:13:20 [INFO|DP=0|PP=0|TP=25|ip-26-0-164-187]: [After model building] Memory usage: 2443.33MiB. Peak allocated: 5384.00MiB Peak reserved: 9494.00MiB
+01/03/2025 08:13:20 [INFO|DP=0|PP=0|TP=22|ip-26-0-164-18]: [After model building] Memory usage: 2443.33MiB. Peak allocated: 5384.00MiB Peak reserved: 10518.00MiB
+01/03/2025 08:13:20 [INFO|DP=0|PP=0|TP=29|ip-26-0-164-187]: [After model building] Memory usage: 2443.33MiB. Peak allocated: 5384.00MiB Peak reserved: 9494.00MiB
+01/03/2025 08:13:20 [INFO|DP=0|PP=0|TP=16|ip-26-0-164-18]: [After model building] Memory usage: 2443.33MiB. Peak allocated: 5384.00MiB Peak reserved: 10518.00MiB
+01/03/2025 08:13:20 [INFO|DP=0|PP=0|TP=26|ip-26-0-164-187]: Local number of parameters: 1.28G (2443.28MiB)
+01/03/2025 08:13:20 [INFO|DP=0|PP=0|TP=31|ip-26-0-164-187]: Local number of parameters: 1.28G (2443.28MiB)
+01/03/2025 08:13:20 [INFO|DP=0|PP=0|TP=24|ip-26-0-164-187]: [After model building] Memory usage: 2443.33MiB. Peak allocated: 5384.00MiB Peak reserved: 9494.00MiB
+01/03/2025 08:13:20 [INFO|DP=0|PP=0|TP=18|ip-26-0-164-18]: Local number of parameters: 1.28G (2443.28MiB)
+01/03/2025 08:13:20 [INFO|DP=0|PP=0|TP=27|ip-26-0-164-187]: Local number of parameters: 1.28G (2443.28MiB)
+01/03/2025 08:13:20 [INFO|DP=0|PP=0|TP=17|ip-26-0-164-18]: [After model building] Memory usage: 2443.33MiB. Peak allocated: 5384.00MiB Peak reserved: 10518.00MiB
+01/03/2025 08:13:20 [INFO|DP=0|PP=1|TP=22|ip-26-0-169-139]: Local number of parameters: 1.22G (2327.23MiB)
+01/03/2025 08:13:20 [INFO|DP=0|PP=0|TP=20|ip-26-0-164-18]: [After model building] Memory usage: 2443.33MiB. Peak allocated: 5384.00MiB Peak reserved: 10518.00MiB
+01/03/2025 08:13:20 [INFO|DP=0|PP=0|TP=10|ip-26-0-164-0]: Local number of parameters: 1.28G (2443.28MiB)
+01/03/2025 08:13:20 [INFO|DP=0|PP=0|TP=23|ip-26-0-164-18]: [After model building] Memory usage: 2443.33MiB. Peak allocated: 5384.00MiB Peak reserved: 10518.00MiB
+01/03/2025 08:13:20 [INFO|DP=0|PP=1|TP=29|ip-26-0-169-207]: Local number of parameters: 1.22G (2327.23MiB)
+01/03/2025 08:13:20 [INFO|DP=0|PP=1|TP=24|ip-26-0-169-207]: Local number of parameters: 1.22G (2327.23MiB)
+01/03/2025 08:13:20 [INFO|DP=0|PP=0|TP=18|ip-26-0-164-18]: [After model building] Memory usage: 2443.33MiB. Peak allocated: 5384.00MiB Peak reserved: 10518.00MiB
+01/03/2025 08:13:20 [INFO|DP=0|PP=0|TP=15|ip-26-0-164-0]: Local number of parameters: 1.28G (2443.28MiB)
+01/03/2025 08:13:20 [INFO|DP=0|PP=0|TP=13|ip-26-0-164-0]: Local number of parameters: 1.28G (2443.28MiB)
+01/03/2025 08:13:20 [INFO|DP=0|PP=1|TP=22|ip-26-0-169-139]: [After model building] Memory usage: 2327.28MiB. Peak allocated: 5384.00MiB Peak reserved: 10518.00MiB
+01/03/2025 08:13:20 [INFO|DP=0|PP=0|TP=26|ip-26-0-164-187]: [After model building] Memory usage: 2443.33MiB. Peak allocated: 5384.00MiB Peak reserved: 9494.00MiB
+01/03/2025 08:13:20 [INFO|DP=0|PP=1|TP=21|ip-26-0-169-139]: Local number of parameters: 1.22G (2327.23MiB)
+01/03/2025 08:13:20 [INFO|DP=0|PP=0|TP=14|ip-26-0-164-0]: Local number of parameters: 1.28G (2443.28MiB)
+01/03/2025 08:13:20 [INFO|DP=0|PP=1|TP=20|ip-26-0-169-139]: Local number of parameters: 1.22G (2327.23MiB)
+01/03/2025 08:13:20 [INFO|DP=0|PP=1|TP=26|ip-26-0-169-207]: Local number of parameters: 1.22G (2327.23MiB)
+01/03/2025 08:13:20 [INFO|DP=0|PP=0|TP=31|ip-26-0-164-187]: [After model building] Memory usage: 2443.33MiB. Peak allocated: 5384.00MiB Peak reserved: 9494.00MiB
+01/03/2025 08:13:20 [INFO|DP=0|PP=0|TP=27|ip-26-0-164-187]: [After model building] Memory usage: 2443.33MiB. Peak allocated: 5384.00MiB Peak reserved: 9494.00MiB
+01/03/2025 08:13:20 [INFO|DP=0|PP=0|TP=10|ip-26-0-164-0]: [After model building] Memory usage: 2443.33MiB. Peak allocated: 5384.00MiB Peak reserved: 10518.00MiB
+01/03/2025 08:13:20 [INFO|DP=0|PP=1|TP=9|ip-26-0-166-36]: Local number of parameters: 1.22G (2327.23MiB)
+01/03/2025 08:13:20 [INFO|DP=0|PP=1|TP=12|ip-26-0-166-36]: Local number of parameters: 1.22G (2327.23MiB)
+01/03/2025 08:13:20 [INFO|DP=0|PP=1|TP=29|ip-26-0-169-207]: [After model building] Memory usage: 2327.28MiB. Peak allocated: 5384.00MiB Peak reserved: 10518.00MiB
+01/03/2025 08:13:20 [INFO|DP=0|PP=1|TP=27|ip-26-0-169-207]: Local number of parameters: 1.22G (2327.23MiB)
+01/03/2025 08:13:20 [INFO|DP=0|PP=0|TP=13|ip-26-0-164-0]: [After model building] Memory usage: 2443.33MiB. Peak allocated: 5384.00MiB Peak reserved: 10518.00MiB
+01/03/2025 08:13:20 [INFO|DP=0|PP=1|TP=23|ip-26-0-169-139]: Local number of parameters: 1.22G (2327.23MiB)
+01/03/2025 08:13:20 [INFO|DP=0|PP=1|TP=24|ip-26-0-169-207]: [After model building] Memory usage: 2327.28MiB. Peak allocated: 5384.00MiB Peak reserved: 10518.00MiB
+01/03/2025 08:13:20 [INFO|DP=0|PP=0|TP=15|ip-26-0-164-0]: [After model building] Memory usage: 2443.33MiB. Peak allocated: 5384.00MiB Peak reserved: 10518.00MiB
+01/03/2025 08:13:20 [INFO|DP=0|PP=1|TP=10|ip-26-0-166-36]: Local number of parameters: 1.22G (2327.23MiB)
+01/03/2025 08:13:20 [INFO|DP=0|PP=1|TP=21|ip-26-0-169-139]: [After model building] Memory usage: 2327.28MiB. Peak allocated: 5384.00MiB Peak reserved: 10518.00MiB
+01/03/2025 08:13:20 [INFO|DP=0|PP=0|TP=14|ip-26-0-164-0]: [After model building] Memory usage: 2443.33MiB. Peak allocated: 5384.00MiB Peak reserved: 10518.00MiB
+01/03/2025 08:13:20 [INFO|DP=0|PP=1|TP=18|ip-26-0-169-139]: Local number of parameters: 1.22G (2327.23MiB)
+01/03/2025 08:13:20 [INFO|DP=0|PP=1|TP=20|ip-26-0-169-139]: [After model building] Memory usage: 2327.28MiB. Peak allocated: 5384.00MiB Peak reserved: 10518.00MiB
+01/03/2025 08:13:20 [INFO|DP=0|PP=1|TP=26|ip-26-0-169-207]: [After model building] Memory usage: 2327.28MiB. Peak allocated: 5384.00MiB Peak reserved: 10518.00MiB
+01/03/2025 08:13:20 [INFO|DP=0|PP=1|TP=12|ip-26-0-166-36]: [After model building] Memory usage: 2327.28MiB. Peak allocated: 5384.00MiB Peak reserved: 10518.00MiB
+01/03/2025 08:13:20 [INFO|DP=0|PP=1|TP=9|ip-26-0-166-36]: [After model building] Memory usage: 2327.28MiB. Peak allocated: 5384.00MiB Peak reserved: 10518.00MiB
+01/03/2025 08:13:20 [INFO|DP=0|PP=1|TP=27|ip-26-0-169-207]: [After model building] Memory usage: 2327.28MiB. Peak allocated: 5384.00MiB Peak reserved: 10518.00MiB
+01/03/2025 08:13:20 [INFO|DP=0|PP=1|TP=23|ip-26-0-169-139]: [After model building] Memory usage: 2327.28MiB. Peak allocated: 5384.00MiB Peak reserved: 10518.00MiB
+01/03/2025 08:13:20 [INFO|DP=0|PP=1|TP=15|ip-26-0-166-36]: Local number of parameters: 1.22G (2327.23MiB)
+01/03/2025 08:13:20 [INFO|DP=0|PP=1|TP=10|ip-26-0-166-36]: [After model building] Memory usage: 2327.28MiB. Peak allocated: 5384.00MiB Peak reserved: 10518.00MiB
+01/03/2025 08:13:20 [INFO|DP=0|PP=1|TP=13|ip-26-0-166-36]: Local number of parameters: 1.22G (2327.23MiB)
+01/03/2025 08:13:20 [INFO|DP=0|PP=1|TP=18|ip-26-0-169-139]: [After model building] Memory usage: 2327.28MiB. Peak allocated: 5384.00MiB Peak reserved: 10518.00MiB
+01/03/2025 08:13:20 [INFO|DP=0|PP=1|TP=14|ip-26-0-166-36]: Local number of parameters: 1.22G (2327.23MiB)
+01/03/2025 08:13:20 [INFO|DP=0|PP=1|TP=15|ip-26-0-166-36]: [After model building] Memory usage: 2327.28MiB. Peak allocated: 5384.00MiB Peak reserved: 10518.00MiB
+01/03/2025 08:13:20 [INFO|DP=0|PP=1|TP=13|ip-26-0-166-36]: [After model building] Memory usage: 2327.28MiB. Peak allocated: 5384.00MiB Peak reserved: 10518.00MiB
+01/03/2025 08:13:20 [INFO|DP=0|PP=1|TP=14|ip-26-0-166-36]: [After model building] Memory usage: 2327.28MiB. Peak allocated: 5384.00MiB Peak reserved: 10518.00MiB
+01/03/2025 08:13:20 [INFO|DP=0|PP=1|TP=2|ip-26-0-166-15]: Local number of parameters: 1.22G (2327.23MiB)
+01/03/2025 08:13:20 [INFO|DP=0|PP=1|TP=2|ip-26-0-166-15]: [After model building] Memory usage: 2327.28MiB. Peak allocated: 5384.00MiB Peak reserved: 10518.00MiB
+01/03/2025 08:13:20 [INFO|DP=0|PP=1|TP=5|ip-26-0-166-15]: Local number of parameters: 1.22G (2327.23MiB)
+01/03/2025 08:13:20 [INFO|DP=0|PP=1|TP=6|ip-26-0-166-15]: Local number of parameters: 1.22G (2327.23MiB)
+01/03/2025 08:13:20 [INFO|DP=0|PP=1|TP=5|ip-26-0-166-15]: [After model building] Memory usage: 2327.28MiB. Peak allocated: 5384.00MiB Peak reserved: 10518.00MiB
+01/03/2025 08:13:20 [INFO|DP=0|PP=1|TP=6|ip-26-0-166-15]: [After model building] Memory usage: 2327.28MiB. Peak allocated: 5384.00MiB Peak reserved: 10518.00MiB
+01/03/2025 08:13:20 [INFO|DP=0|PP=1|TP=4|ip-26-0-166-15]: Local number of parameters: 1.22G (2327.23MiB)
+01/03/2025 08:13:20 [INFO|DP=0|PP=1|TP=4|ip-26-0-166-15]: [After model building] Memory usage: 2327.28MiB. Peak allocated: 5384.00MiB Peak reserved: 10518.00MiB
+01/03/2025 08:13:20 [INFO|DP=0|PP=1|TP=8|ip-26-0-166-36]: Local number of parameters: 1.22G (2327.23MiB)
+01/03/2025 08:13:20 [INFO|DP=0|PP=1|TP=8|ip-26-0-166-36]: [After model building] Memory usage: 2327.28MiB. Peak allocated: 5384.00MiB Peak reserved: 10518.00MiB
+01/03/2025 08:13:20 [INFO|DP=0|PP=1|TP=31|ip-26-0-169-207]: Local number of parameters: 1.22G (2327.23MiB)
+01/03/2025 08:13:20 [INFO|DP=0|PP=1|TP=25|ip-26-0-169-207]: Local number of parameters: 1.22G (2327.23MiB)
+01/03/2025 08:13:20 [INFO|DP=0|PP=1|TP=31|ip-26-0-169-207]: [After model building] Memory usage: 2327.28MiB. Peak allocated: 5384.00MiB Peak reserved: 10518.00MiB
+01/03/2025 08:13:20 [INFO|DP=0|PP=1|TP=17|ip-26-0-169-139]: Local number of parameters: 1.22G (2327.23MiB)
+01/03/2025 08:13:20 [INFO|DP=0|PP=1|TP=25|ip-26-0-169-207]: [After model building] Memory usage: 2327.28MiB. Peak allocated: 5384.00MiB Peak reserved: 10518.00MiB
+01/03/2025 08:13:20 [INFO|DP=0|PP=1|TP=17|ip-26-0-169-139]: [After model building] Memory usage: 2327.28MiB. Peak allocated: 5384.00MiB Peak reserved: 10518.00MiB
+01/03/2025 08:13:20 [INFO|DP=0|PP=1|TP=16|ip-26-0-169-139]: Local number of parameters: 1.22G (2327.23MiB)
+01/03/2025 08:13:20 [INFO|DP=0|PP=1|TP=16|ip-26-0-169-139]: [After model building] Memory usage: 2327.28MiB. Peak allocated: 5384.00MiB Peak reserved: 10518.00MiB
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+01/03/2025 08:13:23 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: [Optimizer Building] Using LearningRateForSP as learning rate
+01/03/2025 08:13:23 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: [ZeRO sharding] Size of optimizer params per rank:
+01/03/2025 08:13:23 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: [ZeRO sharding] DP Rank 0 has 320M out of 1.28G (25.00%) params' optimizer states
+01/03/2025 08:13:23 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: [ZeRO sharding] DP Rank 1 has 320M out of 1.28G (25.00%) params' optimizer states
+01/03/2025 08:13:23 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: [ZeRO sharding] DP Rank 2 has 320M out of 1.28G (25.00%) params' optimizer states
+01/03/2025 08:13:23 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: [ZeRO sharding] DP Rank 3 has 320M out of 1.28G (25.00%) params' optimizer states
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+01/03/2025 08:13:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.token_position_embeddings | PP: 0/2 | Block rank: 0
+01/03/2025 08:13:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.0 | PP: 0/2 | Block rank: 1
+01/03/2025 08:13:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.1 | PP: 0/2 | Block rank: 2
+01/03/2025 08:13:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.2 | PP: 0/2 | Block rank: 3
+01/03/2025 08:13:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.3 | PP: 0/2 | Block rank: 4
+01/03/2025 08:13:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.4 | PP: 0/2 | Block rank: 5
+01/03/2025 08:13:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.5 | PP: 0/2 | Block rank: 6
+01/03/2025 08:13:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.6 | PP: 0/2 | Block rank: 7
+01/03/2025 08:13:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.7 | PP: 0/2 | Block rank: 8
+01/03/2025 08:13:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.8 | PP: 0/2 | Block rank: 9
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+01/03/2025 08:13:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.9 | PP: 0/2 | Block rank: 10
+01/03/2025 08:13:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.10 | PP: 0/2 | Block rank: 11
+01/03/2025 08:13:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.11 | PP: 0/2 | Block rank: 12
+01/03/2025 08:13:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.12 | PP: 0/2 | Block rank: 13
+01/03/2025 08:13:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.13 | PP: 0/2 | Block rank: 14
+01/03/2025 08:13:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.14 | PP: 0/2 | Block rank: 15
+01/03/2025 08:13:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.15 | PP: 0/2 | Block rank: 16
+01/03/2025 08:13:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.16 | PP: 0/2 | Block rank: 17
+01/03/2025 08:13:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.17 | PP: 0/2 | Block rank: 18
+01/03/2025 08:13:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.18 | PP: 0/2 | Block rank: 19
+01/03/2025 08:13:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.19 | PP: 0/2 | Block rank: 20
+01/03/2025 08:13:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.20 | PP: 0/2 | Block rank: 21
+01/03/2025 08:13:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.21 | PP: 0/2 | Block rank: 22
+01/03/2025 08:13:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.22 | PP: 0/2 | Block rank: 23
+01/03/2025 08:13:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.23 | PP: 0/2 | Block rank: 24
+01/03/2025 08:13:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.24 | PP: 0/2 | Block rank: 25
+01/03/2025 08:13:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.25 | PP: 0/2 | Block rank: 26
+01/03/2025 08:13:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.26 | PP: 0/2 | Block rank: 27
+01/03/2025 08:13:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.27 | PP: 0/2 | Block rank: 28
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+01/03/2025 08:13:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.28 | PP: 0/2 | Block rank: 29
+01/03/2025 08:13:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.29 | PP: 0/2 | Block rank: 30
+01/03/2025 08:13:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.30 | PP: 0/2 | Block rank: 31
+01/03/2025 08:13:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.31 | PP: 0/2 | Block rank: 32
+01/03/2025 08:13:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.32 | PP: 0/2 | Block rank: 33
+01/03/2025 08:13:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.33 | PP: 0/2 | Block rank: 34
+01/03/2025 08:13:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.34 | PP: 0/2 | Block rank: 35
+01/03/2025 08:13:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.35 | PP: 0/2 | Block rank: 36
+01/03/2025 08:13:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.36 | PP: 0/2 | Block rank: 37
+01/03/2025 08:13:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.37 | PP: 0/2 | Block rank: 38
+01/03/2025 08:13:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.38 | PP: 0/2 | Block rank: 39
+01/03/2025 08:13:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.39 | PP: 0/2 | Block rank: 40
+01/03/2025 08:13:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.40 | PP: 0/2 | Block rank: 41
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+01/03/2025 08:13:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.41 | PP: 1/2 | Block rank: 0
+01/03/2025 08:13:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.42 | PP: 1/2 | Block rank: 1
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+01/03/2025 08:13:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.43 | PP: 1/2 | Block rank: 2
+01/03/2025 08:13:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.44 | PP: 1/2 | Block rank: 3
+01/03/2025 08:13:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.45 | PP: 1/2 | Block rank: 4
+01/03/2025 08:13:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.46 | PP: 1/2 | Block rank: 5
+01/03/2025 08:13:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.47 | PP: 1/2 | Block rank: 6
+01/03/2025 08:13:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.48 | PP: 1/2 | Block rank: 7
+01/03/2025 08:13:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.49 | PP: 1/2 | Block rank: 8
+01/03/2025 08:13:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.50 | PP: 1/2 | Block rank: 9
+01/03/2025 08:13:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.51 | PP: 1/2 | Block rank: 10
+01/03/2025 08:13:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.52 | PP: 1/2 | Block rank: 11
+01/03/2025 08:13:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.53 | PP: 1/2 | Block rank: 12
+01/03/2025 08:13:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.54 | PP: 1/2 | Block rank: 13
+01/03/2025 08:13:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.55 | PP: 1/2 | Block rank: 14
+01/03/2025 08:13:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.56 | PP: 1/2 | Block rank: 15
+01/03/2025 08:13:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.57 | PP: 1/2 | Block rank: 16
+01/03/2025 08:13:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.58 | PP: 1/2 | Block rank: 17
+01/03/2025 08:13:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.59 | PP: 1/2 | Block rank: 18
+01/03/2025 08:13:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.60 | PP: 1/2 | Block rank: 19
+01/03/2025 08:13:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.61 | PP: 1/2 | Block rank: 20
+01/03/2025 08:13:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.62 | PP: 1/2 | Block rank: 21
+01/03/2025 08:13:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.63 | PP: 1/2 | Block rank: 22
+01/03/2025 08:13:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.64 | PP: 1/2 | Block rank: 23
+01/03/2025 08:13:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.65 | PP: 1/2 | Block rank: 24
+01/03/2025 08:13:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.66 | PP: 1/2 | Block rank: 25
+01/03/2025 08:13:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.67 | PP: 1/2 | Block rank: 26
+01/03/2025 08:13:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.68 | PP: 1/2 | Block rank: 27
+01/03/2025 08:13:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.69 | PP: 1/2 | Block rank: 28
+01/03/2025 08:13:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.70 | PP: 1/2 | Block rank: 29
+01/03/2025 08:13:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.71 | PP: 1/2 | Block rank: 30
+01/03/2025 08:13:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.72 | PP: 1/2 | Block rank: 31
+01/03/2025 08:13:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.73 | PP: 1/2 | Block rank: 32
+01/03/2025 08:13:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.74 | PP: 1/2 | Block rank: 33
+01/03/2025 08:13:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.75 | PP: 1/2 | Block rank: 34
+01/03/2025 08:13:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.76 | PP: 1/2 | Block rank: 35
+01/03/2025 08:13:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.77 | PP: 1/2 | Block rank: 36
+01/03/2025 08:13:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.78 | PP: 1/2 | Block rank: 37
+01/03/2025 08:13:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.decoder.79 | PP: 1/2 | Block rank: 38
+01/03/2025 08:13:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.final_layer_norm | PP: 1/2 | Block rank: 39
+01/03/2025 08:13:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.lm_head | PP: 1/2 | Block rank: 40
+01/03/2025 08:13:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: model.cast_to_fp32 | PP: 1/2 | Block rank: 41
+01/03/2025 08:13:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: module_name: loss | PP: 1/2 | Block rank: 42
+01/03/2025 08:13:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: [Training Plan] Stage Stable Training Stage has 99 remaining training steps and has consumed 0 samples
+01/03/2025 08:13:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: Using dummy data generator
+01/03/2025 08:13:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: [Training Plan] There are 1 training stages 
+01/03/2025 08:13:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: [Stage Stable Training Stage] start from step 1 
+01/03/2025 08:13:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: 
+01/03/2025 08:13:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: [Start training] datetime: 2025-01-03 08:13:24.075419 | mbs: 2 | grad_accum: 32 | global_batch_size: 256 | sequence_length: 4096 | train_steps: 100 | start_iteration_step: 0 | consumed_train_samples: 0
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+01/03/2025 08:13:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: Resuming training from stage Stable Training Stage, it has trained for 0 samples and has 99 remaining train steps
+01/03/2025 08:13:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:  Memory usage: 8551.89MiB. Peak allocated 8551.89MiB. Peak reserved: 16628.00MiB
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+01/03/2025 08:15:58 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:  Memory usage: 8912.96MiB. Peak allocated 17032.78MiB. Peak reserved: 17406.00MiB
+01/03/2025 08:16:08 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:  Memory usage: 11363.37MiB. Peak allocated 16861.47MiB. Peak reserved: 22298.00MiB
+01/03/2025 08:16:08 [INFO|DP=0|PP=1|TP=0|ip-26-0-166-15]: iteration: 1 / 100 | consumed_tokens: 1.05M | elapsed_time_per_iteration_ms: 162K | tokens_per_sec: 6.46K | tokens_per_sec_per_gpu: 25.2 | global_batch_size: 256 | lm_loss: 13.4 | lr: 0.00015 | model_tflops_per_gpu: 12.8 | hardware_tflops_per_gpu: 12.8 | grad_norm: 7.27 | cuda_memory_allocated: 11.4G | cuda_max_memory_reserved: 18.2G | hd_total_memory_tb: 312G | hd_used_memory_tb: 70.3G | hd_free_memory_tb: 242G
+01/03/2025 08:18:27 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:  Memory usage: 11363.37MiB. Peak allocated 19391.92MiB. Peak reserved: 22298.00MiB
+01/03/2025 08:18:27 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:  Memory usage: 11363.37MiB. Peak allocated 16861.47MiB. Peak reserved: 22298.00MiB
+01/03/2025 08:18:27 [INFO|DP=0|PP=1|TP=0|ip-26-0-166-15]: iteration: 2 / 100 | consumed_tokens: 2.1M | elapsed_time_per_iteration_ms: 139K | tokens_per_sec: 7.53K | tokens_per_sec_per_gpu: 29.4 | global_batch_size: 256 | lm_loss: 13.4 | lr: 0.0003 | model_tflops_per_gpu: 14.9 | hardware_tflops_per_gpu: 14.9 | grad_norm: 7.27 | cuda_memory_allocated: 11.4G | cuda_max_memory_reserved: 18.2G | hd_total_memory_tb: 312G | hd_used_memory_tb: 70.3G | hd_free_memory_tb: 242G
+01/03/2025 08:20:47 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]:  Memory usage: 11363.37MiB. Peak allocated 19391.92MiB. Peak reserved: 22298.00MiB
+01/03/2025 08:20:47 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: | job_id   | name                                                          | nodes | seq_len | mbs | batch_accum | gbs | mTFLOPs | hTFLOPs | tok/s/gpu | AllReduce (GB/s) | AllGather (GB/s) | ReduceScatter (GB/s) | AR Intra-node (GB/s) | AG Intra-node (GB/s) | RS Intra-node (GB/s) | Mem Alloc (GB) | Mem Res (GB) | dp | pp | tp | pp_engine                           | tp_mode        | tp_async_comm | hidden_size | hidden_act | num_layers | num_heads | num_kv_heads | max_pos | vocab_size | tie_word_embeddings | dtype          | zero_stage | ddp_bucket_cap_mb | accumulate_grad_in_fp32 | Total Params | Local Params |
+01/03/2025 08:20:47 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: | -------- | ------------------------------------------------------------- | ----- | ------- | --- | ----------- | --- | ------- | ------- | --------- | ---------------- | ---------------- | -------------------- | -------------------- | -------------------- | -------------------- | -------------- | ------------ | -- | -- | -- | ----------------------------------- | -------------- | ------------- | ----------- | ---------- | ---------- | --------- | ------------ | ------- | ---------- | ------------------- | -------------- | ---------- | ----------------- | ----------------------- | ------------ | ------------ |
+01/03/2025 08:20:47 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: | 14019313 | 80G_dp4_tp32_pp2_acc32_mbs2_seq4096_zero1_tpmodeRED_vocab131k | 32    | 4096    | 2   | 32          | 256 | 14.75   | 14.75   | 29.16     | 27.34            | 13.46            | 13.85                | 427.07               | 243.93               | 243.61               | 16.47          | 21.78        | 4  | 2  | 32 | OneForwardOneBackwardPipelineEngine | REDUCE_SCATTER | True          | 8192        | silu       | 80         | 64        | 64           | 4096    | 131072     | False               | torch.bfloat16 | 1          | 25                | True                    | 80G          | 1.28G        |
+01/03/2025 08:20:47 [INFO|DP=0|PP=1|TP=0|ip-26-0-166-15]: iteration: 3 / 100 | consumed_tokens: 3.15M | elapsed_time_per_iteration_ms: 140K | tokens_per_sec: 7.46K | tokens_per_sec_per_gpu: 29.2 | global_batch_size: 256 | lm_loss: 13.4 | lr: 0.000296 | model_tflops_per_gpu: 14.7 | hardware_tflops_per_gpu: 14.7 | grad_norm: 6.92 | cuda_memory_allocated: 11.4G | cuda_max_memory_reserved: 18.2G | hd_total_memory_tb: 312G | hd_used_memory_tb: 70.3G | hd_free_memory_tb: 242G
+01/03/2025 08:20:47 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: Attempting to write benchmark results to CSV file: benchmark/results/bench_final2.csv
+01/03/2025 08:20:47 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: Acquired lock for CSV file: benchmark/results/bench_final2.csv
+01/03/2025 08:20:47 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: Successfully wrote to CSV file: benchmark/results/bench_final2.csv. Releasing lock...
+01/03/2025 08:20:47 [INFO|DP=0|PP=0|TP=0|ip-26-0-163-236]: Throughput logging complete
+[2025-01-03 08:21:18,075] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 267939 closing signal SIGTERM
+[2025-01-03 08:21:18,075] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 267940 closing signal SIGTERM
+[2025-01-03 08:21:18,075] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 267941 closing signal SIGTERM
+[2025-01-03 08:21:18,075] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 267942 closing signal SIGTERM
+[2025-01-03 08:21:18,075] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 264954 closing signal SIGTERM
+[2025-01-03 08:21:18,075] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 264955 closing signal SIGTERM
+[2025-01-03 08:21:18,075] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 264958 closing signal SIGTERM
+[2025-01-03 08:21:18,076] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 264959 closing signal SIGTERM
+[2025-01-03 08:21:18,076] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 264960 closing signal SIGTERM
+[2025-01-03 08:21:18,080] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 27896 closing signal SIGTERM
+[2025-01-03 08:21:18,080] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 27899 closing signal SIGTERM
+[2025-01-03 08:21:18,080] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 27900 closing signal SIGTERM
+[2025-01-03 08:21:18,080] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 27901 closing signal SIGTERM
+[2025-01-03 08:21:18,080] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 27902 closing signal SIGTERM
+[2025-01-03 08:21:18,089] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 109802 closing signal SIGTERM
+[2025-01-03 08:21:18,089] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 109803 closing signal SIGTERM
+[2025-01-03 08:21:18,089] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 109804 closing signal SIGTERM
+[2025-01-03 08:21:18,089] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 109805 closing signal SIGTERM
+[2025-01-03 08:21:18,089] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 109807 closing signal SIGTERM
+[2025-01-03 08:21:18,089] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 109808 closing signal SIGTERM
+[2025-01-03 08:21:18,090] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 138179 closing signal SIGTERM
+[2025-01-03 08:21:18,091] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 138180 closing signal SIGTERM
+[2025-01-03 08:21:18,091] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 138181 closing signal SIGTERM
+[2025-01-03 08:21:18,091] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 138182 closing signal SIGTERM
+[2025-01-03 08:21:18,091] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 138184 closing signal SIGTERM
+[2025-01-03 08:21:18,091] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 138185 closing signal SIGTERM
+[2025-01-03 08:21:18,100] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 28731 closing signal SIGTERM
+[2025-01-03 08:21:18,101] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 28732 closing signal SIGTERM
+[2025-01-03 08:21:18,101] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 28733 closing signal SIGTERM
+[2025-01-03 08:21:18,101] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 28734 closing signal SIGTERM
+[2025-01-03 08:21:18,101] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 28735 closing signal SIGTERM
+[2025-01-03 08:21:18,101] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 28736 closing signal SIGTERM
+[2025-01-03 08:21:18,117] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 277753 closing signal SIGTERM
+[2025-01-03 08:21:18,117] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 277755 closing signal SIGTERM
+[2025-01-03 08:21:18,117] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 277756 closing signal SIGTERM
+[2025-01-03 08:21:18,117] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 277757 closing signal SIGTERM
+[2025-01-03 08:21:18,117] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 277758 closing signal SIGTERM
+[2025-01-03 08:21:18,117] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 277759 closing signal SIGTERM
+[2025-01-03 08:21:18,117] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 277760 closing signal SIGTERM
+[2025-01-03 08:21:18,120] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 39015 closing signal SIGTERM
+[2025-01-03 08:21:18,120] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 39017 closing signal SIGTERM
+[2025-01-03 08:21:18,120] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 39019 closing signal SIGTERM
+[2025-01-03 08:21:18,120] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 39020 closing signal SIGTERM
+[2025-01-03 08:21:18,120] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 39021 closing signal SIGTERM
+[2025-01-03 08:21:18,120] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 39022 closing signal SIGTERM
+[2025-01-03 08:21:18,134] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 73264 closing signal SIGTERM
+[2025-01-03 08:21:18,135] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 73265 closing signal SIGTERM
+[2025-01-03 08:21:18,549] torch.distributed.elastic.multiprocessing.api: [ERROR] failed (exitcode: 1) local_rank: 0 (pid: 73259) of binary: /fsx/nouamane/miniconda/envs/2-1-cu121/bin/python
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 264, in launch_agent
+    raise ChildFailedError(
+torch.distributed.elastic.multiprocessing.errors.ChildFailedError: 
+============================================================
+/fsx/nouamane/projects/nanotron/run_train.py FAILED
+------------------------------------------------------------
+Failures:
+[1]:
+  time      : 2025-01-03_08:21:18
+  host      : ip-26-0-173-7.ec2.internal
+  rank      : 249 (local_rank: 1)
+  exitcode  : 1 (pid: 73260)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[2]:
+  time      : 2025-01-03_08:21:18
+  host      : ip-26-0-173-7.ec2.internal
+  rank      : 250 (local_rank: 2)
+  exitcode  : 1 (pid: 73261)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[3]:
+  time      : 2025-01-03_08:21:18
+  host      : ip-26-0-173-7.ec2.internal
+  rank      : 251 (local_rank: 3)
+  exitcode  : 1 (pid: 73262)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[4]:
+  time      : 2025-01-03_08:21:18
+  host      : ip-26-0-173-7.ec2.internal
+  rank      : 252 (local_rank: 4)
+  exitcode  : 1 (pid: 73263)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[5]:
+  time      : 2025-01-03_08:21:18
+  host      : ip-26-0-173-7.ec2.internal
+  rank      : 255 (local_rank: 7)
+  exitcode  : 1 (pid: 73266)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+------------------------------------------------------------
+Root Cause (first observed failure):
+[0]:
+  time      : 2025-01-03_08:21:18
+  host      : ip-26-0-173-7.ec2.internal
+  rank      : 248 (local_rank: 0)
+  exitcode  : 1 (pid: 73259)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+============================================================
+srun: error: ip-26-0-173-7: task 28: Exited with exit code 1
+srun: launch/slurm: _step_signal: Terminating StepId=14019313.0
+slurmstepd: error: *** STEP 14019313.0 ON ip-26-0-163-236 CANCELLED AT 2025-01-03T08:21:18 ***
+[2025-01-03 08:21:18,889] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 08:21:18,889] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 08:21:18,890] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 267942 closing signal SIGTERM
+[2025-01-03 08:21:18,890] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 08:21:18,890] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 68701 closing signal SIGTERM
+[2025-01-03 08:21:18,890] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 28731 closing signal SIGTERM
+[2025-01-03 08:21:18,890] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 68702 closing signal SIGTERM
+[2025-01-03 08:21:18,890] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 28732 closing signal SIGTERM
+[2025-01-03 08:21:18,890] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 28733 closing signal SIGTERM
+[2025-01-03 08:21:18,890] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 68703 closing signal SIGTERM
+[2025-01-03 08:21:18,890] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 68704 closing signal SIGTERM
+[2025-01-03 08:21:18,890] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 28734 closing signal SIGTERM
+[2025-01-03 08:21:18,890] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 68705 closing signal SIGTERM
+[2025-01-03 08:21:18,890] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 28735 closing signal SIGTERM
+[2025-01-03 08:21:18,890] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 28736 closing signal SIGTERM
+[2025-01-03 08:21:18,890] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 68706 closing signal SIGTERM
+[2025-01-03 08:21:18,890] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 68707 closing signal SIGTERM
+[2025-01-03 08:21:18,890] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 08:21:18,890] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 68708 closing signal SIGTERM
+[2025-01-03 08:21:18,891] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 456812 closing signal SIGTERM
+[2025-01-03 08:21:18,890] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 08:21:18,890] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 08:21:18,890] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 08:21:18,891] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 62350 closing signal SIGTERM
+[2025-01-03 08:21:18,891] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 456813 closing signal SIGTERM
+[2025-01-03 08:21:18,890] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 27896 closing signal SIGTERM
+[2025-01-03 08:21:18,891] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 62351 closing signal SIGTERM
+[2025-01-03 08:21:18,891] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 142291 closing signal SIGTERM
+[2025-01-03 08:21:18,891] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 456814 closing signal SIGTERM
+[2025-01-03 08:21:18,890] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 08:21:18,890] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 08:21:18,890] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 27899 closing signal SIGTERM
+[2025-01-03 08:21:18,891] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 62352 closing signal SIGTERM
+[2025-01-03 08:21:18,891] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 142292 closing signal SIGTERM
+[2025-01-03 08:21:18,891] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 62353 closing signal SIGTERM
+[2025-01-03 08:21:18,890] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 109802 closing signal SIGTERM
+[2025-01-03 08:21:18,890] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 27900 closing signal SIGTERM
+[2025-01-03 08:21:18,891] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 456815 closing signal SIGTERM
+[2025-01-03 08:21:18,890] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 33027 closing signal SIGTERM
+[2025-01-03 08:21:18,890] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 08:21:18,891] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 62354 closing signal SIGTERM
+[2025-01-03 08:21:18,891] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 142293 closing signal SIGTERM
+[2025-01-03 08:21:18,891] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 27901 closing signal SIGTERM
+[2025-01-03 08:21:18,891] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 456816 closing signal SIGTERM
+[2025-01-03 08:21:18,890] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 33028 closing signal SIGTERM
+[2025-01-03 08:21:18,891] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 62355 closing signal SIGTERM
+[2025-01-03 08:21:18,890] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 109803 closing signal SIGTERM
+[2025-01-03 08:21:18,891] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 142294 closing signal SIGTERM
+[2025-01-03 08:21:18,891] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 27902 closing signal SIGTERM
+[2025-01-03 08:21:18,890] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 33029 closing signal SIGTERM
+[2025-01-03 08:21:18,890] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 264954 closing signal SIGTERM
+[2025-01-03 08:21:18,891] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 62356 closing signal SIGTERM
+[2025-01-03 08:21:18,890] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 109804 closing signal SIGTERM
+[2025-01-03 08:21:18,891] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 08:21:18,891] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 142295 closing signal SIGTERM
+[2025-01-03 08:21:18,891] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 08:21:18,891] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 456817 closing signal SIGTERM
+[2025-01-03 08:21:18,890] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 33030 closing signal SIGTERM
+[2025-01-03 08:21:18,890] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 264955 closing signal SIGTERM
+[2025-01-03 08:21:18,890] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 08:21:18,890] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 109805 closing signal SIGTERM
+[2025-01-03 08:21:18,891] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 521833 closing signal SIGTERM
+[2025-01-03 08:21:18,891] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 456818 closing signal SIGTERM
+[2025-01-03 08:21:18,890] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 33031 closing signal SIGTERM
+[2025-01-03 08:21:18,890] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 08:21:18,890] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 109807 closing signal SIGTERM
+[2025-01-03 08:21:18,891] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 118813 closing signal SIGTERM
+[2025-01-03 08:21:18,891] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 456819 closing signal SIGTERM
+[2025-01-03 08:21:18,890] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 33032 closing signal SIGTERM
+[2025-01-03 08:21:18,890] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 264959 closing signal SIGTERM
+[2025-01-03 08:21:18,891] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 62357 closing signal SIGTERM
+[2025-01-03 08:21:18,890] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 109808 closing signal SIGTERM
+[2025-01-03 08:21:18,891] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 521834 closing signal SIGTERM
+[2025-01-03 08:21:18,891] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 142296 closing signal SIGTERM
+[2025-01-03 08:21:18,891] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 33033 closing signal SIGTERM
+[2025-01-03 08:21:18,890] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 264960 closing signal SIGTERM
+[2025-01-03 08:21:18,891] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 255750 closing signal SIGTERM
+[2025-01-03 08:21:18,890] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 48836 closing signal SIGTERM
+[2025-01-03 08:21:18,891] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 521835 closing signal SIGTERM
+[2025-01-03 08:21:18,891] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 33034 closing signal SIGTERM
+[2025-01-03 08:21:18,891] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 118814 closing signal SIGTERM
+[2025-01-03 08:21:18,890] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 08:21:18,890] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 48837 closing signal SIGTERM
+[2025-01-03 08:21:18,891] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 521836 closing signal SIGTERM
+[2025-01-03 08:21:18,891] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 118815 closing signal SIGTERM
+[2025-01-03 08:21:18,891] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 255751 closing signal SIGTERM
+[2025-01-03 08:21:18,891] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 48838 closing signal SIGTERM
+[2025-01-03 08:21:18,891] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 142297 closing signal SIGTERM
+[2025-01-03 08:21:18,891] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 08:21:18,891] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 118816 closing signal SIGTERM
+[2025-01-03 08:21:18,891] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 118817 closing signal SIGTERM
+[2025-01-03 08:21:18,891] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 48839 closing signal SIGTERM
+[2025-01-03 08:21:18,891] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 118818 closing signal SIGTERM
+[2025-01-03 08:21:18,891] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 255752 closing signal SIGTERM
+[2025-01-03 08:21:18,891] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 48840 closing signal SIGTERM
+[2025-01-03 08:21:18,892] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 521837 closing signal SIGTERM
+[2025-01-03 08:21:18,891] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 08:21:18,891] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 08:21:18,890] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 08:21:18,891] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 58258 closing signal SIGTERM
+[2025-01-03 08:21:18,891] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 48841 closing signal SIGTERM
+[2025-01-03 08:21:18,891] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 46362 closing signal SIGTERM
+[2025-01-03 08:21:18,891] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 118819 closing signal SIGTERM
+[2025-01-03 08:21:18,891] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 58260 closing signal SIGTERM
+[2025-01-03 08:21:18,891] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 08:21:18,892] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 521838 closing signal SIGTERM
+[2025-01-03 08:21:18,891] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 118820 closing signal SIGTERM
+[2025-01-03 08:21:18,891] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 255754 closing signal SIGTERM
+[2025-01-03 08:21:18,891] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 58261 closing signal SIGTERM
+[2025-01-03 08:21:18,891] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 27644 closing signal SIGTERM
+[2025-01-03 08:21:18,891] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 142298 closing signal SIGTERM
+[2025-01-03 08:21:18,891] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 48842 closing signal SIGTERM
+[2025-01-03 08:21:18,891] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 27645 closing signal SIGTERM
+[2025-01-03 08:21:18,891] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 46363 closing signal SIGTERM
+[2025-01-03 08:21:18,892] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 67076 closing signal SIGTERM
+[2025-01-03 08:21:18,891] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 29063 closing signal SIGTERM
+[2025-01-03 08:21:18,891] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 255755 closing signal SIGTERM
+[2025-01-03 08:21:18,891] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 58262 closing signal SIGTERM
+[2025-01-03 08:21:18,891] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 48843 closing signal SIGTERM
+[2025-01-03 08:21:18,891] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 277755 closing signal SIGTERM
+[2025-01-03 08:21:18,891] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 27647 closing signal SIGTERM
+[2025-01-03 08:21:18,892] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 521839 closing signal SIGTERM
+[2025-01-03 08:21:18,892] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 67077 closing signal SIGTERM
+[2025-01-03 08:21:18,891] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 58263 closing signal SIGTERM
+[2025-01-03 08:21:18,891] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 27648 closing signal SIGTERM
+[2025-01-03 08:21:18,892] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 521840 closing signal SIGTERM
+[2025-01-03 08:21:18,891] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 46364 closing signal SIGTERM
+[2025-01-03 08:21:18,891] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 08:21:18,892] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 67078 closing signal SIGTERM
+[2025-01-03 08:21:18,890] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 08:21:18,891] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 58264 closing signal SIGTERM
+[2025-01-03 08:21:18,891] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 277756 closing signal SIGTERM
+[2025-01-03 08:21:18,891] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 27649 closing signal SIGTERM
+[2025-01-03 08:21:18,890] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 08:21:18,891] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 46365 closing signal SIGTERM
+[2025-01-03 08:21:18,892] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 08:21:18,892] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 67079 closing signal SIGTERM
+[2025-01-03 08:21:18,891] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 29064 closing signal SIGTERM
+[2025-01-03 08:21:18,891] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 255756 closing signal SIGTERM
+[2025-01-03 08:21:18,891] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 58265 closing signal SIGTERM
+[2025-01-03 08:21:18,891] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 277757 closing signal SIGTERM
+[2025-01-03 08:21:18,891] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 27650 closing signal SIGTERM
+[2025-01-03 08:21:18,891] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 46366 closing signal SIGTERM
+[2025-01-03 08:21:18,891] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 78055 closing signal SIGTERM
+[2025-01-03 08:21:18,892] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 67080 closing signal SIGTERM
+[2025-01-03 08:21:18,891] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 47196 closing signal SIGTERM
+[2025-01-03 08:21:18,891] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 08:21:18,891] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 277758 closing signal SIGTERM
+[2025-01-03 08:21:18,891] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 27651 closing signal SIGTERM
+[2025-01-03 08:21:18,891] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 46367 closing signal SIGTERM
+[2025-01-03 08:21:18,891] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 78056 closing signal SIGTERM
+[2025-01-03 08:21:18,892] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 67081 closing signal SIGTERM
+[2025-01-03 08:21:18,891] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 29065 closing signal SIGTERM
+[2025-01-03 08:21:18,891] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 277759 closing signal SIGTERM
+[2025-01-03 08:21:18,891] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 27652 closing signal SIGTERM
+[2025-01-03 08:21:18,891] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 84458 closing signal SIGTERM
+[2025-01-03 08:21:18,891] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 46368 closing signal SIGTERM
+[2025-01-03 08:21:18,893] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 780599 closing signal SIGTERM
+[2025-01-03 08:21:18,891] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 08:21:18,891] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 47197 closing signal SIGTERM
+[2025-01-03 08:21:18,891] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 277760 closing signal SIGTERM
+[2025-01-03 08:21:18,891] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 84459 closing signal SIGTERM
+[2025-01-03 08:21:18,891] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 46369 closing signal SIGTERM
+[2025-01-03 08:21:18,891] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 78057 closing signal SIGTERM
+[2025-01-03 08:21:18,890] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 08:21:18,892] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 67082 closing signal SIGTERM
+[2025-01-03 08:21:18,891] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 34493 closing signal SIGTERM
+[2025-01-03 08:21:18,891] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 84460 closing signal SIGTERM
+[2025-01-03 08:21:18,892] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 67083 closing signal SIGTERM
+[2025-01-03 08:21:18,891] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 47198 closing signal SIGTERM
+[2025-01-03 08:21:18,891] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 29066 closing signal SIGTERM
+[2025-01-03 08:21:18,891] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 84461 closing signal SIGTERM
+[2025-01-03 08:21:18,891] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 78058 closing signal SIGTERM
+[2025-01-03 08:21:18,893] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 780600 closing signal SIGTERM
+[2025-01-03 08:21:18,892] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 138554 closing signal SIGTERM
+[2025-01-03 08:21:18,891] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 47199 closing signal SIGTERM
+[2025-01-03 08:21:18,891] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 78059 closing signal SIGTERM
+[2025-01-03 08:21:18,890] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 138179 closing signal SIGTERM
+[2025-01-03 08:21:18,893] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 780601 closing signal SIGTERM
+[2025-01-03 08:21:18,891] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 47200 closing signal SIGTERM
+[2025-01-03 08:21:18,891] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 29067 closing signal SIGTERM
+[2025-01-03 08:21:18,891] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 84462 closing signal SIGTERM
+[2025-01-03 08:21:18,891] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 78060 closing signal SIGTERM
+[2025-01-03 08:21:18,890] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 138180 closing signal SIGTERM
+[2025-01-03 08:21:18,893] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 780602 closing signal SIGTERM
+[2025-01-03 08:21:18,892] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 138555 closing signal SIGTERM
+[2025-01-03 08:21:18,892] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 08:21:18,891] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 47201 closing signal SIGTERM
+[2025-01-03 08:21:18,891] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 34494 closing signal SIGTERM
+[2025-01-03 08:21:18,891] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 84463 closing signal SIGTERM
+[2025-01-03 08:21:18,891] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 78061 closing signal SIGTERM
+[2025-01-03 08:21:18,893] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 780603 closing signal SIGTERM
+[2025-01-03 08:21:18,891] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 47202 closing signal SIGTERM
+[2025-01-03 08:21:18,893] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 08:21:18,891] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 84464 closing signal SIGTERM
+[2025-01-03 08:21:18,890] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 138181 closing signal SIGTERM
+[2025-01-03 08:21:18,893] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 780604 closing signal SIGTERM
+[2025-01-03 08:21:18,892] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 138556 closing signal SIGTERM
+[2025-01-03 08:21:18,891] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 47203 closing signal SIGTERM
+[2025-01-03 08:21:18,891] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 29068 closing signal SIGTERM
+[2025-01-03 08:21:18,891] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 34495 closing signal SIGTERM
+[2025-01-03 08:21:18,891] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 84465 closing signal SIGTERM
+[2025-01-03 08:21:18,893] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 780605 closing signal SIGTERM
+[2025-01-03 08:21:18,892] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 138557 closing signal SIGTERM
+[2025-01-03 08:21:18,892] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 67958 closing signal SIGTERM
+[2025-01-03 08:21:18,891] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 29069 closing signal SIGTERM
+[2025-01-03 08:21:18,890] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 08:21:18,891] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 78062 closing signal SIGTERM
+[2025-01-03 08:21:18,892] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 138558 closing signal SIGTERM
+[2025-01-03 08:21:18,893] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 39017 closing signal SIGTERM
+[2025-01-03 08:21:18,893] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 780606 closing signal SIGTERM
+[2025-01-03 08:21:18,892] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 138559 closing signal SIGTERM
+[2025-01-03 08:21:18,892] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 67959 closing signal SIGTERM
+[2025-01-03 08:21:18,891] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 34496 closing signal SIGTERM
+[2025-01-03 08:21:18,890] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 138182 closing signal SIGTERM
+[2025-01-03 08:21:18,892] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 138560 closing signal SIGTERM
+[2025-01-03 08:21:18,893] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 08:21:18,892] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 67960 closing signal SIGTERM
+[2025-01-03 08:21:18,893] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 39019 closing signal SIGTERM
+[2025-01-03 08:21:18,891] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3399563 closing signal SIGTERM
+[2025-01-03 08:21:18,892] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 138561 closing signal SIGTERM
+[2025-01-03 08:21:18,891] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 29070 closing signal SIGTERM
+[2025-01-03 08:21:18,891] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3399564 closing signal SIGTERM
+[2025-01-03 08:21:18,890] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 138184 closing signal SIGTERM
+[2025-01-03 08:21:18,892] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 67961 closing signal SIGTERM
+[2025-01-03 08:21:18,891] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3399565 closing signal SIGTERM
+[2025-01-03 08:21:18,891] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 34497 closing signal SIGTERM
+[2025-01-03 08:21:18,892] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 67962 closing signal SIGTERM
+[2025-01-03 08:21:18,893] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 39020 closing signal SIGTERM
+[2025-01-03 08:21:18,894] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 34874 closing signal SIGTERM
+[2025-01-03 08:21:18,892] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 67963 closing signal SIGTERM
+[2025-01-03 08:21:18,893] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 39021 closing signal SIGTERM
+[2025-01-03 08:21:18,892] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 67964 closing signal SIGTERM
+[2025-01-03 08:21:18,891] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3399566 closing signal SIGTERM
+[2025-01-03 08:21:18,890] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 138185 closing signal SIGTERM
+[2025-01-03 08:21:18,893] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 39022 closing signal SIGTERM
+[2025-01-03 08:21:18,891] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 34498 closing signal SIGTERM
+[2025-01-03 08:21:18,891] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 34499 closing signal SIGTERM
+[2025-01-03 08:21:18,894] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 34876 closing signal SIGTERM
+[2025-01-03 08:21:18,892] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 67965 closing signal SIGTERM
+[2025-01-03 08:21:18,891] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3399568 closing signal SIGTERM
+[2025-01-03 08:21:18,892] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 34500 closing signal SIGTERM
+[2025-01-03 08:21:18,894] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 34877 closing signal SIGTERM
+[2025-01-03 08:21:18,891] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3399569 closing signal SIGTERM
+[2025-01-03 08:21:18,894] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 34878 closing signal SIGTERM
+[2025-01-03 08:21:18,894] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 34879 closing signal SIGTERM
+[2025-01-03 08:21:18,891] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3399570 closing signal SIGTERM
+[2025-01-03 08:21:18,894] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 34880 closing signal SIGTERM
+[2025-01-03 08:21:18,894] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 34881 closing signal SIGTERM
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 878, in _invoke_run
+    run_result = self._monitor_workers(self._worker_group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/local_elastic_agent.py", line 307, in _monitor_workers
+    result = self._pcontext.wait(0)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 288, in wait
+    return self._poll()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 668, in _poll
+    self.close()  # terminate all running procs
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 331, in close
+    self._close(death_sig=death_sig, timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 713, in _close
+    handler.proc.wait(time_to_wait)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1209, in wait
+    return self._wait(timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1953, in _wait
+    time.sleep(delay)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 267863 got signal: 15
+srun: error: ip-26-0-163-236: task 0: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 878, in _invoke_run
+    run_result = self._monitor_workers(self._worker_group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/local_elastic_agent.py", line 307, in _monitor_workers
+    result = self._pcontext.wait(0)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 288, in wait
+    return self._poll()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 668, in _poll
+    self.close()  # terminate all running procs
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 331, in close
+    self._close(death_sig=death_sig, timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 713, in _close
+    handler.proc.wait(time_to_wait)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1209, in wait
+    return self._wait(timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1953, in _wait
+    time.sleep(delay)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 28657 got signal: 15
+srun: error: ip-26-0-165-24: task 8: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 878, in _invoke_run
+    run_result = self._monitor_workers(self._worker_group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/local_elastic_agent.py", line 307, in _monitor_workers
+    result = self._pcontext.wait(0)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 288, in wait
+    return self._poll()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 668, in _poll
+    self.close()  # terminate all running procs
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 331, in close
+    self._close(death_sig=death_sig, timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 713, in _close
+    handler.proc.wait(time_to_wait)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1209, in wait
+    return self._wait(timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1953, in _wait
+    time.sleep(delay)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 264882 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 878, in _invoke_run
+    run_result = self._monitor_workers(self._worker_group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/local_elastic_agent.py", line 307, in _monitor_workers
+    result = self._pcontext.wait(0)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 288, in wait
+    return self._poll()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 668, in _poll
+    self.close()  # terminate all running procs
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 331, in close
+    self._close(death_sig=death_sig, timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 713, in _close
+    handler.proc.wait(time_to_wait)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1209, in wait
+    return self._wait(timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1953, in _wait
+    time.sleep(delay)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 27823 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 878, in _invoke_run
+    run_result = self._monitor_workers(self._worker_group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/local_elastic_agent.py", line 307, in _monitor_workers
+    result = self._pcontext.wait(0)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 288, in wait
+    return self._poll()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 668, in _poll
+    self.close()  # terminate all running procs
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 331, in close
+    self._close(death_sig=death_sig, timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 713, in _close
+    handler.proc.wait(time_to_wait)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1209, in wait
+    return self._wait(timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1953, in _wait
+    time.sleep(delay)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 109728 got signal: 15
+srun: error: ip-26-0-164-207: task 6: Exited with exit code 1
+srun: error: ip-26-0-169-207: task 19: Exited with exit code 1
+[2025-01-03 08:21:22,010] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-166-125.ec2.internal_521759_0' has failed to send a keep-alive heartbeat to the rendezvous '14019313' due to an error of type RendezvousConnectionError.
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 878, in _invoke_run
+    run_result = self._monitor_workers(self._worker_group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/local_elastic_agent.py", line 307, in _monitor_workers
+    result = self._pcontext.wait(0)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 288, in wait
+    return self._poll()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 668, in _poll
+    self.close()  # terminate all running procs
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 331, in close
+    self._close(death_sig=death_sig, timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 713, in _close
+    handler.proc.wait(time_to_wait)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1209, in wait
+    return self._wait(timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1953, in _wait
+    time.sleep(delay)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 38944 got signal: 15
+srun: error: ip-26-0-165-131: task 11: Exited with exit code 1
+[2025-01-03 08:21:22,212] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-173-121.ec2.internal_3399490_0' has failed to send a keep-alive heartbeat to the rendezvous '14019313' due to an error of type RendezvousConnectionError.
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 878, in _invoke_run
+    run_result = self._monitor_workers(self._worker_group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/local_elastic_agent.py", line 307, in _monitor_workers
+    result = self._pcontext.wait(0)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 288, in wait
+    return self._poll()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 668, in _poll
+    self.close()  # terminate all running procs
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 331, in close
+    self._close(death_sig=death_sig, timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 713, in _close
+    handler.proc.wait(time_to_wait)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1209, in wait
+    return self._wait(timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1953, in _wait
+    time.sleep(delay)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 138106 got signal: 15
+srun: error: ip-26-0-172-252: task 27: Exited with exit code 1
+srun: error: ip-26-0-170-143: task 24: Exited with exit code 1
+[2025-01-03 08:21:22,682] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-165-164.ec2.internal_118741_0' has failed to send a keep-alive heartbeat to the rendezvous '14019313' due to an error of type RendezvousConnectionError.
+[2025-01-03 08:21:22,686] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-164-236.ec2.internal_32955_0' has failed to send a keep-alive heartbeat to the rendezvous '14019313' due to an error of type RendezvousConnectionError.
+[2025-01-03 08:21:22,691] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-170-132.ec2.internal_138482_0' has failed to send a keep-alive heartbeat to the rendezvous '14019313' due to an error of type RendezvousConnectionError.
+[2025-01-03 08:21:22,704] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-170-31.ec2.internal_780527_0' has failed to send a keep-alive heartbeat to the rendezvous '14019313' due to an error of type RendezvousConnectionError.
+[2025-01-03 08:21:22,707] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-164-75.ec2.internal_34803_0' has failed to send a keep-alive heartbeat to the rendezvous '14019313' due to an error of type RendezvousConnectionError.
+[2025-01-03 08:21:22,706] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-165-213.ec2.internal_48763_0' has failed to send a keep-alive heartbeat to the rendezvous '14019313' due to an error of type RendezvousConnectionError.
+[2025-01-03 08:21:22,710] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-169-239.ec2.internal_255677_0' has failed to send a keep-alive heartbeat to the rendezvous '14019313' due to an error of type RendezvousConnectionError.
+[2025-01-03 08:21:22,718] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-165-202.ec2.internal_47124_0' has failed to send a keep-alive heartbeat to the rendezvous '14019313' due to an error of type RendezvousConnectionError.
+[2025-01-03 08:21:22,766] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-169-247.ec2.internal_34421_0' has failed to send a keep-alive heartbeat to the rendezvous '14019313' due to an error of type RendezvousConnectionError.
+[2025-01-03 08:21:22,768] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-169-139.ec2.internal_28992_0' has failed to send a keep-alive heartbeat to the rendezvous '14019313' due to an error of type RendezvousConnectionError.
+[2025-01-03 08:21:22,771] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-164-45.ec2.internal_456740_0' has failed to send a keep-alive heartbeat to the rendezvous '14019313' due to an error of type RendezvousConnectionError.
+[2025-01-03 08:21:22,773] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-173-202.ec2.internal_84385_0' has failed to send a keep-alive heartbeat to the rendezvous '14019313' due to an error of type RendezvousConnectionError.
+[2025-01-03 08:21:22,785] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-171-21.ec2.internal_77983_0' has failed to send a keep-alive heartbeat to the rendezvous '14019313' due to an error of type RendezvousConnectionError.
+[2025-01-03 08:21:22,800] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-165-59.ec2.internal_27573_0' has failed to send a keep-alive heartbeat to the rendezvous '14019313' due to an error of type RendezvousConnectionError.
+[2025-01-03 08:21:22,835] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-166-15.ec2.internal_67004_0' has failed to send a keep-alive heartbeat to the rendezvous '14019313' due to an error of type RendezvousConnectionError.
+[2025-01-03 08:21:22,837] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-164-0.ec2.internal_68630_0' has failed to send a keep-alive heartbeat to the rendezvous '14019313' due to an error of type RendezvousConnectionError.
+[2025-01-03 08:21:22,839] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-164-187.ec2.internal_58187_0' has failed to send a keep-alive heartbeat to the rendezvous '14019313' due to an error of type RendezvousConnectionError.
+[2025-01-03 08:21:22,847] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-165-38.ec2.internal_62278_0' has failed to send a keep-alive heartbeat to the rendezvous '14019313' due to an error of type RendezvousConnectionError.
+[2025-01-03 08:21:22,928] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-164-18.ec2.internal_142219_0' has failed to send a keep-alive heartbeat to the rendezvous '14019313' due to an error of type RendezvousConnectionError.
+[2025-01-03 08:21:22,991] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-173-246.ec2.internal_67885_0' has failed to send a keep-alive heartbeat to the rendezvous '14019313' due to an error of type RendezvousConnectionError.
+[2025-01-03 08:21:22,993] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-166-36.ec2.internal_46290_0' has failed to send a keep-alive heartbeat to the rendezvous '14019313' due to an error of type RendezvousConnectionError.
+[2025-01-03 08:21:22,996] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-170-160.ec2.internal_277680_0' has failed to send a keep-alive heartbeat to the rendezvous '14019313' due to an error of type RendezvousConnectionError.
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 67004 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 48763 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 58187 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 34803 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 255677 got signal: 15
+srun: error: ip-26-0-166-15: task 15: Exited with exit code 1
+srun: error: ip-26-0-165-213: task 14: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 878, in _invoke_run
+    run_result = self._monitor_workers(self._worker_group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/local_elastic_agent.py", line 307, in _monitor_workers
+    result = self._pcontext.wait(0)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 288, in wait
+    return self._poll()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 668, in _poll
+    self.close()  # terminate all running procs
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 331, in close
+    self._close(death_sig=death_sig, timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 713, in _close
+    handler.proc.wait(time_to_wait)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1209, in wait
+    return self._wait(timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1953, in _wait
+    time.sleep(delay)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 277680 got signal: 15
+srun: error: ip-26-0-164-75: task 4: Exited with exit code 1
+srun: error: ip-26-0-164-187: task 5: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 32955 got signal: 15
+srun: error: ip-26-0-169-239: task 20: Exited with exit code 1
+srun: error: ip-26-0-170-160: task 25: Exited with exit code 1
+srun: error: ip-26-0-164-236: task 7: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 67885 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 521759 got signal: 15
+srun: error: ip-26-0-173-246: task 31: Exited with exit code 1
+srun: error: ip-26-0-166-125: task 17: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 142219 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 34421 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 3399490 got signal: 15
+srun: error: ip-26-0-164-18: task 2: Exited with exit code 1
+srun: error: ip-26-0-169-247: task 21: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 46290 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 68630 got signal: 15
+srun: error: ip-26-0-173-121: task 29: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 28992 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 27573 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 138482 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 47124 got signal: 15
+srun: error: ip-26-0-166-36: task 16: Exited with exit code 1
+srun: error: ip-26-0-164-0: task 1: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 84385 got signal: 15
+srun: error: ip-26-0-169-139: task 18: Exited with exit code 1
+srun: error: ip-26-0-165-59: task 10: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 456740 got signal: 15
+srun: error: ip-26-0-165-202: task 13: Exited with exit code 1
+srun: error: ip-26-0-170-132: task 23: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 118741 got signal: 15
+srun: error: ip-26-0-173-202: task 30: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 780527 got signal: 15
+srun: error: ip-26-0-164-45: task 3: Exited with exit code 1
+srun: error: ip-26-0-165-164: task 12: Exited with exit code 1
+srun: error: ip-26-0-170-31: task 22: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 77983 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 62278 got signal: 15
+srun: error: ip-26-0-171-21: task 26: Exited with exit code 1
+srun: error: ip-26-0-165-38: task 9: Exited with exit code 1
diff --git a/logs/14023604-bench_3.57G_dp2_tp8_pp8_acc128_mbs1_seq4096_zero0_tpmodeRED_vocab131k.out b/logs/14023604-bench_3.57G_dp2_tp8_pp8_acc128_mbs1_seq4096_zero0_tpmodeRED_vocab131k.out
new file mode 100644
index 0000000000000000000000000000000000000000..fb6006387fd92ded69e1db42eff984aa1c2fb125
--- /dev/null
+++ b/logs/14023604-bench_3.57G_dp2_tp8_pp8_acc128_mbs1_seq4096_zero0_tpmodeRED_vocab131k.out
@@ -0,0 +1,2057 @@
++ '[' -z 14023604 ']'
++ source /etc/profile.d/modules.sh
+++ . /usr/share/modules/init/bash
++++ unset _mlshdbg
++++ '[' 0 = 1 ']'
++++ unset _mlre _mlIFS
++++ '[' -n x ']'
++++ _mlIFS=' 	
+'
++++ IFS=' '
++++ '[' -n '' ']'
+++++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash autoinit
++++ _mlcode='module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+test 0;'
++++ _mlret=0
++++ '[' -n x ']'
++++ IFS=' 	
+'
++++ unset _mlIFS
++++ unset _mlre _mlv _mlrv
++++ '[' 0 -eq 0 ']'
++++ eval 'module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+test 0;'
+++++ MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl
+++++ export MODULES_CMD
+++++ ENV=/usr/share/modules/init/profile.sh
+++++ export ENV
+++++ BASH_ENV=/usr/share/modules/init/bash
+++++ export BASH_ENV
+++++ MODULESHOME=/usr/share/modules
+++++ export MODULESHOME
+++++ test 0
++++ '[' 0 = 1 ']'
++++ '[' -t 2 ']'
++++ export -f module
++++ export -f switchml
++++ '[' 5 -ge 3 ']'
++++ [[ ehxB =~ i ]]
++++ [[ ! :/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin: =~ :/usr/bin: ]]
++++ '[' '!' -n x ']'
+++++ manpath
++++ [[ ! :/fsx/nouamane/miniconda/envs/2-1-cu121/man:/fsx/nouamane/miniconda/envs/2-1-cu121/share/man:/fsx/nouamane/miniconda/man:/fsx/nouamane/miniconda/share/man:/opt/amazon/openmpi/share/man:/opt/amazon/efa/share/man:/opt/slurm/share/man:/usr/local/man:/usr/local/share/man:/usr/share/man:/admin/home/nouamane/.local/share/man:/admin/home/nouamane/.fzf/man:: =~ :/usr/share/man: ]]
++++ unset _mlcode _mlret
++++ '[' -n '' ']'
++ module load cuda/12.1
++ unset _mlshdbg
++ '[' 0 = 1 ']'
++ unset _mlre _mlIFS
++ '[' -n x ']'
++ _mlIFS=' 	
+'
++ IFS=' '
++ '[' -n '' ']'
+++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash load cuda/12.1
++ eval 'CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include;' export 'CPATH;
+LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/opt/aws-ofi-nccl/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/targets/x86_64-linux/lib/:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/lib:/usr/lib::/opt/amazon/openmpi/lib:/opt/amazon/efa/lib;' export 'LD_LIBRARY_PATH;
+FI_EFA_FORK_SAFE=1;' export 'FI_EFA_FORK_SAFE;
+MANPATH=/usr/local/cuda-12.1/share/man::;' export 'MANPATH;
+LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64;' export 'LIBRARY_PATH;
+_LMFILES_=/usr/share/modules/modulefiles/cuda/12.1;' export '_LMFILES_;
+LOADEDMODULES=cuda/12.1;' export 'LOADEDMODULES;
+MPI_PATH=/opt/amazon/openmpi;' export 'MPI_PATH;
+NCCL_HOME_modshare=/opt/nccl/build:1;' export 'NCCL_HOME_modshare;
+NCCL_PROTO=simple;' export 'NCCL_PROTO;
+MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1;' export 'MANPATH_modshare;
+LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1;' export 'LIBRARY_PATH_modshare;
+NCCL_SOCKET_IFNAME=enp;' export 'NCCL_SOCKET_IFNAME;
+AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl;' export 'AWS_OFI_NCCL_HOME;
+NCCL_HOME=/opt/nccl/build;' export 'NCCL_HOME;
+FI_PROVIDER=efa;' export 'FI_PROVIDER;
+AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1;' export 'AWS_OFI_NCCL_HOME_modshare;
+CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1;' export 'CPATH_modshare;
+LD_LIBRARY_PATH_modshare=:1:/opt/amazon/efa/lib:3:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib/:1:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2:/opt/aws-ofi-nccl/lib:2:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/efa/lib:1:/usr/lib:1;' export 'LD_LIBRARY_PATH_modshare;
+FI_EFA_ENABLE_SHM_TRANSFER=1;' export 'FI_EFA_ENABLE_SHM_TRANSFER;
+_LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1;' export '_LMFILES__modshare;
+LOADEDMODULES_modshare=cuda/12.1:1;' export 'LOADEDMODULES_modshare;
+MPI_PATH_modshare=/opt/amazon/openmpi:1;' export 'MPI_PATH_modshare;
+PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin;' export 'PATH;
+CUDA_HOME=/usr/local/cuda-12.1;' export 'CUDA_HOME;
+PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/local/cuda-12.1/include:1:/usr/bin:1:/opt/amazon/efa/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/usr/sbin:1:/opt/slurm/bin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:2:/opt/amazon/openmpi/bin:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1;' export 'PATH_modshare;
+test' '0;'
+++ CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include
+++ export CPATH
+++ LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/opt/aws-ofi-nccl/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/targets/x86_64-linux/lib/:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/lib:/usr/lib::/opt/amazon/openmpi/lib:/opt/amazon/efa/lib
+++ export LD_LIBRARY_PATH
+++ FI_EFA_FORK_SAFE=1
+++ export FI_EFA_FORK_SAFE
+++ MANPATH=/usr/local/cuda-12.1/share/man::
+++ export MANPATH
+++ LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64
+++ export LIBRARY_PATH
+++ _LMFILES_=/usr/share/modules/modulefiles/cuda/12.1
+++ export _LMFILES_
+++ LOADEDMODULES=cuda/12.1
+++ export LOADEDMODULES
+++ MPI_PATH=/opt/amazon/openmpi
+++ export MPI_PATH
+++ NCCL_HOME_modshare=/opt/nccl/build:1
+++ export NCCL_HOME_modshare
+++ NCCL_PROTO=simple
+++ export NCCL_PROTO
+++ MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1
+++ export MANPATH_modshare
+++ LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1
+++ export LIBRARY_PATH_modshare
+++ NCCL_SOCKET_IFNAME=enp
+++ export NCCL_SOCKET_IFNAME
+++ AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl
+++ export AWS_OFI_NCCL_HOME
+++ NCCL_HOME=/opt/nccl/build
+++ export NCCL_HOME
+++ FI_PROVIDER=efa
+++ export FI_PROVIDER
+++ AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1
+++ export AWS_OFI_NCCL_HOME_modshare
+++ CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1
+++ export CPATH_modshare
+++ LD_LIBRARY_PATH_modshare=:1:/opt/amazon/efa/lib:3:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib/:1:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2:/opt/aws-ofi-nccl/lib:2:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/efa/lib:1:/usr/lib:1
+++ export LD_LIBRARY_PATH_modshare
+++ FI_EFA_ENABLE_SHM_TRANSFER=1
+++ export FI_EFA_ENABLE_SHM_TRANSFER
+++ _LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1
+++ export _LMFILES__modshare
+++ LOADEDMODULES_modshare=cuda/12.1:1
+++ export LOADEDMODULES_modshare
+++ MPI_PATH_modshare=/opt/amazon/openmpi:1
+++ export MPI_PATH_modshare
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin
+++ export PATH
+++ CUDA_HOME=/usr/local/cuda-12.1
+++ export CUDA_HOME
+++ PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/local/cuda-12.1/include:1:/usr/bin:1:/opt/amazon/efa/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/usr/sbin:1:/opt/slurm/bin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:2:/opt/amazon/openmpi/bin:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1
+++ export PATH_modshare
+++ test 0
++ _mlstatus=0
++ '[' -n x ']'
++ IFS=' 	
+'
++ unset _mlre _mlv _mlrv _mlIFS
++ '[' -n '' ']'
++ unset _mlshdbg
++ return 0
++ source /fsx/nouamane/miniconda/bin/activate
+++ _CONDA_ROOT=/fsx/nouamane/miniconda
+++ . /fsx/nouamane/miniconda/etc/profile.d/conda.sh
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ '[' -z x ']'
+++ conda activate
+++ local cmd=activate
+++ case "$cmd" in
+++ __conda_activate activate
+++ '[' -n '' ']'
+++ local ask_conda
++++ PS1=
++++ __conda_exe shell.posix activate
++++ /fsx/nouamane/miniconda/bin/conda shell.posix activate
+Error while loading conda entry point: conda-libmamba-solver (module 'libmambapy' has no attribute 'QueryFormat')
+++ ask_conda='. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''2'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_1='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/etc/conda/activate.d/libxml2_activate.sh"'
+++ eval '. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''2'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_1='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/etc/conda/activate.d/libxml2_activate.sh"'
++++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh
+++++ test -n ''
+++++ unset XML_CATALOG_FILES
+++++ unset xml_catalog_files_libxml2
++++ PS1='(base) '
++++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin
++++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin
++++ export CONDA_PREFIX=/fsx/nouamane/miniconda
++++ CONDA_PREFIX=/fsx/nouamane/miniconda
++++ export CONDA_SHLVL=2
++++ CONDA_SHLVL=2
++++ export CONDA_DEFAULT_ENV=base
++++ CONDA_DEFAULT_ENV=base
++++ export 'CONDA_PROMPT_MODIFIER=(base) '
++++ CONDA_PROMPT_MODIFIER='(base) '
++++ export CONDA_PREFIX_1=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ CONDA_PREFIX_1=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ . /fsx/nouamane/miniconda/etc/conda/activate.d/libxml2_activate.sh
+++++ test -n ''
+++++ xml_catalog_files_libxml2=
+++++ XML_CATALOG_FILES=
+++++ conda_catalog_files=
+++++ ifs_libxml2=' 	
+'
+++++ IFS=' '
+++++ rem=/fsx/nouamane/miniconda
+++++ for pre in ${rem}
+++++ test '' = /fsx/nouamane/miniconda
+++++ conda_catalog_files=/fsx/nouamane/miniconda
+++++ rem=
+++++ IFS=' 	
+'
+++++ conda_catalog_files='file:///fsx/nouamane/miniconda/etc/xml/catalog file:///etc/xml/catalog'
+++++ export 'XML_CATALOG_FILES=file:///fsx/nouamane/miniconda/etc/xml/catalog file:///etc/xml/catalog'
+++++ XML_CATALOG_FILES='file:///fsx/nouamane/miniconda/etc/xml/catalog file:///etc/xml/catalog'
+++++ unset conda_catalog_files ifs_libxml2 rem
+++ __conda_hashr
+++ '[' -n '' ']'
+++ '[' -n '' ']'
+++ hash -r
++ conda activate 2-1-cu121
++ local cmd=activate
++ case "$cmd" in
++ __conda_activate activate 2-1-cu121
++ '[' -n '' ']'
++ local ask_conda
+++ PS1='(base) '
+++ __conda_exe shell.posix activate 2-1-cu121
+++ /fsx/nouamane/miniconda/bin/conda shell.posix activate 2-1-cu121
+Error while loading conda entry point: conda-libmamba-solver (module 'libmambapy' has no attribute 'QueryFormat')
++ ask_conda='. "/fsx/nouamane/miniconda/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_2='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
++ eval '. "/fsx/nouamane/miniconda/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_2='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
+++ . /fsx/nouamane/miniconda/etc/conda/deactivate.d/libxml2_deactivate.sh
++++ test -n ''
++++ unset XML_CATALOG_FILES
++++ unset xml_catalog_files_libxml2
+++ PS1='(2-1-cu121) '
+++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin
+++ export CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ export CONDA_SHLVL=3
+++ CONDA_SHLVL=3
+++ export CONDA_DEFAULT_ENV=2-1-cu121
+++ CONDA_DEFAULT_ENV=2-1-cu121
+++ export 'CONDA_PROMPT_MODIFIER=(2-1-cu121) '
+++ CONDA_PROMPT_MODIFIER='(2-1-cu121) '
+++ export CONDA_PREFIX_2=/fsx/nouamane/miniconda
+++ CONDA_PREFIX_2=/fsx/nouamane/miniconda
+++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ export _CE_M=
+++ _CE_M=
+++ export _CE_CONDA=
+++ _CE_CONDA=
+++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh
++++ test -n ''
++++ xml_catalog_files_libxml2=
++++ XML_CATALOG_FILES=
++++ conda_catalog_files=
++++ ifs_libxml2=' 	
+'
++++ IFS=' '
++++ rem=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ for pre in ${rem}
++++ test '' = /fsx/nouamane/miniconda/envs/2-1-cu121
++++ conda_catalog_files=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ rem=
++++ IFS=' 	
+'
++++ conda_catalog_files='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ export 'XML_CATALOG_FILES=file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ XML_CATALOG_FILES='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ unset conda_catalog_files ifs_libxml2 rem
++ __conda_hashr
++ '[' -n '' ']'
++ '[' -n '' ']'
++ hash -r
++ export PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin
++ PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/opt/slurm/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/amazon/efa/bin:/opt/amazon/openmpi/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin
++ '[' -z '' ']'
+++ scontrol show hostnames 'ip-26-0-160-[100,103],ip-26-0-168-[34,52,95,120,238],ip-26-0-169-[86,132,139,207,239,247],ip-26-0-170-[31,132,143]'
++ export 'NODELIST=ip-26-0-160-100
+ip-26-0-160-103
+ip-26-0-168-34
+ip-26-0-168-52
+ip-26-0-168-95
+ip-26-0-168-120
+ip-26-0-168-238
+ip-26-0-169-86
+ip-26-0-169-132
+ip-26-0-169-139
+ip-26-0-169-207
+ip-26-0-169-239
+ip-26-0-169-247
+ip-26-0-170-31
+ip-26-0-170-132
+ip-26-0-170-143'
++ NODELIST='ip-26-0-160-100
+ip-26-0-160-103
+ip-26-0-168-34
+ip-26-0-168-52
+ip-26-0-168-95
+ip-26-0-168-120
+ip-26-0-168-238
+ip-26-0-169-86
+ip-26-0-169-132
+ip-26-0-169-139
+ip-26-0-169-207
+ip-26-0-169-239
+ip-26-0-169-247
+ip-26-0-170-31
+ip-26-0-170-132
+ip-26-0-170-143'
+++ head -n1
+++ scontrol show hostnames 'ip-26-0-160-[100,103],ip-26-0-168-[34,52,95,120,238],ip-26-0-169-[86,132,139,207,239,247],ip-26-0-170-[31,132,143]'
++ export MASTER_NODE=ip-26-0-160-100
++ MASTER_NODE=ip-26-0-160-100
++ export MASTER_PORT=12356
++ MASTER_PORT=12356
++ export NNODES=16
++ NNODES=16
++ export GPUS_PER_NODE=8
++ GPUS_PER_NODE=8
++ export WORLD_SIZE=128
++ WORLD_SIZE=128
++ export CUDA_DEVICE_MAX_CONNECTIONS=1
++ CUDA_DEVICE_MAX_CONNECTIONS=1
++ export NCCL_DEBUG=WARN
++ NCCL_DEBUG=WARN
++ export NANOTRON_BENCHMARK=1
++ NANOTRON_BENCHMARK=1
++ export WANDB_MODE=disabled
++ WANDB_MODE=disabled
++ export TORCH_NCCL_ASYNC_ERROR_HANDLING=1
++ TORCH_NCCL_ASYNC_ERROR_HANDLING=1
++ export TORCH_DISTRIBUTED_DEBUG=DETAIL
++ TORCH_DISTRIBUTED_DEBUG=DETAIL
++ '[' -z '' ']'
++ echo '=== GPU Topology ==='
+=== GPU Topology ===
++ nvidia-smi topo -m
+	GPU0	GPU1	GPU2	GPU3	GPU4	GPU5	GPU6	GPU7	CPU Affinity	NUMA Affinity	GPU NUMA ID
+GPU0	 X 	NV18	NV18	NV18	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU1	NV18	 X 	NV18	NV18	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU2	NV18	NV18	 X 	NV18	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU3	NV18	NV18	NV18	 X 	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU4	NV18	NV18	NV18	NV18	 X 	NV18	NV18	NV18	48-95	1		N/A
+GPU5	NV18	NV18	NV18	NV18	NV18	 X 	NV18	NV18	48-95	1		N/A
+GPU6	NV18	NV18	NV18	NV18	NV18	NV18	 X 	NV18	48-95	1		N/A
+GPU7	NV18	NV18	NV18	NV18	NV18	NV18	NV18	 X 	48-95	1		N/A
+
+Legend:
+
+  X    = Self
+  SYS  = Connection traversing PCIe as well as the SMP interconnect between NUMA nodes (e.g., QPI/UPI)
+  NODE = Connection traversing PCIe as well as the interconnect between PCIe Host Bridges within a NUMA node
+  PHB  = Connection traversing PCIe as well as a PCIe Host Bridge (typically the CPU)
+  PXB  = Connection traversing multiple PCIe bridges (without traversing the PCIe Host Bridge)
+  PIX  = Connection traversing at most a single PCIe bridge
+  NV#  = Connection traversing a bonded set of # NVLinks
++ echo ==================
+==================
++ export SRUN_ALLOC_ARGS=
++ SRUN_ALLOC_ARGS=
++ echo 'Master node: ip-26-0-160-100'
+Master node: ip-26-0-160-100
++ echo 'All nodes: ip-26-0-160-100
+ip-26-0-160-103
+ip-26-0-168-34
+ip-26-0-168-52
+ip-26-0-168-95
+ip-26-0-168-120
+ip-26-0-168-238
+ip-26-0-169-86
+ip-26-0-169-132
+ip-26-0-169-139
+ip-26-0-169-207
+ip-26-0-169-239
+ip-26-0-169-247
+ip-26-0-170-31
+ip-26-0-170-132
+ip-26-0-170-143'
+All nodes: ip-26-0-160-100
+ip-26-0-160-103
+ip-26-0-168-34
+ip-26-0-168-52
+ip-26-0-168-95
+ip-26-0-168-120
+ip-26-0-168-238
+ip-26-0-169-86
+ip-26-0-169-132
+ip-26-0-169-139
+ip-26-0-169-207
+ip-26-0-169-239
+ip-26-0-169-247
+ip-26-0-170-31
+ip-26-0-170-132
+ip-26-0-170-143
++ echo 'World size: 128'
+World size: 128
++ '[' -n '' ']'
++ srun --wait=0 --kill-on-bad-exit=1 torchrun --nnodes=16 --nproc_per_node=8 --rdzv_id=14023604 --rdzv_backend=c10d --rdzv_endpoint=ip-26-0-160-100:12356 --max_restarts 0 --rdzv_conf timeout=60 /fsx/nouamane/projects/nanotron/run_train.py --config-file benchmark/configs/config_3.57G_dp2_tp8_pp8_acc128_mbs1_seq4096_zero0_tpmodeRED_vocab131k.yaml
+[2025-01-03 11:11:00,796] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 11:11:00,795] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 11:11:00,795] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 11:11:00,795] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 11:11:00,796] torch.distributed.run: [WARNING] 
+[2025-01-03 11:11:00,796] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 11:11:00,796] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 11:11:00,796] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 11:11:00,796] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 11:11:00,796] torch.distributed.run: [WARNING] 
+[2025-01-03 11:11:00,796] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 11:11:00,796] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 11:11:00,796] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 11:11:00,795] torch.distributed.run: [WARNING] 
+[2025-01-03 11:11:00,795] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 11:11:00,795] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 11:11:00,795] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 11:11:00,796] torch.distributed.run: [WARNING] 
+[2025-01-03 11:11:00,796] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 11:11:00,796] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 11:11:00,796] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 11:11:00,796] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 11:11:00,796] torch.distributed.run: [WARNING] 
+[2025-01-03 11:11:00,796] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 11:11:00,796] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 11:11:00,796] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 11:11:00,796] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 11:11:00,796] torch.distributed.run: [WARNING] 
+[2025-01-03 11:11:00,796] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 11:11:00,796] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 11:11:00,796] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 11:11:00,802] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 11:11:00,802] torch.distributed.run: [WARNING] 
+[2025-01-03 11:11:00,802] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 11:11:00,802] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 11:11:00,802] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 11:11:00,804] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 11:11:00,804] torch.distributed.run: [WARNING] 
+[2025-01-03 11:11:00,804] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 11:11:00,804] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 11:11:00,804] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 11:11:00,806] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 11:11:00,806] torch.distributed.run: [WARNING] 
+[2025-01-03 11:11:00,806] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 11:11:00,806] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 11:11:00,806] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 11:11:00,822] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 11:11:00,822] torch.distributed.run: [WARNING] 
+[2025-01-03 11:11:00,822] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 11:11:00,822] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 11:11:00,822] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 11:11:00,825] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 11:11:00,832] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 11:11:00,832] torch.distributed.run: [WARNING] 
+[2025-01-03 11:11:00,832] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 11:11:00,832] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 11:11:00,832] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 11:11:00,837] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 11:11:00,838] torch.distributed.run: [WARNING] 
+[2025-01-03 11:11:00,838] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 11:11:00,838] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 11:11:00,838] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 11:11:00,795] torch.distributed.run: [WARNING] 
+[2025-01-03 11:11:00,795] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 11:11:00,795] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 11:11:00,795] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 11:11:00,843] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 11:11:00,843] torch.distributed.run: [WARNING] 
+[2025-01-03 11:11:00,843] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 11:11:00,843] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 11:11:00,843] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 11:11:00,825] torch.distributed.run: [WARNING] 
+[2025-01-03 11:11:00,825] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 11:11:00,825] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 11:11:00,825] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 11:11:04,472] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-03 11:11:04,472] torch.distributed.run: [WARNING] 
+[2025-01-03 11:11:04,472] torch.distributed.run: [WARNING] *****************************************
+[2025-01-03 11:11:04,472] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-03 11:11:04,472] torch.distributed.run: [WARNING] *****************************************
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+01/03/2025 11:11:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]: Config:
+01/03/2025 11:11:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]: Config(general=GeneralArgs(project='debug',
+01/03/2025 11:11:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                            run='3.57G_dp2_tp8_pp8_acc128_mbs1_seq4096_zero0_tpmodeRED_vocab131k',
+01/03/2025 11:11:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                            seed=42,
+01/03/2025 11:11:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                            step=None,
+01/03/2025 11:11:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                            consumed_train_samples=None,
+01/03/2025 11:11:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                            benchmark_csv_path=PosixPath('benchmark/results/bench_final2.csv'),
+01/03/2025 11:11:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                            ignore_sanity_checks=True),
+01/03/2025 11:11:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:        parallelism=ParallelismArgs(dp=2,
+01/03/2025 11:11:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                    pp=8,
+01/03/2025 11:11:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                    tp=8,
+01/03/2025 11:11:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                    pp_engine=<nanotron.parallel.pipeline_parallel.engine.OneForwardOneBackwardPipelineEngine object at 0x7f04b9518dc0>,
+01/03/2025 11:11:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                    tp_mode=<TensorParallelLinearMode.REDUCE_SCATTER: 2>,
+01/03/2025 11:11:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                    tp_linear_async_communication=True,
+01/03/2025 11:11:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                    recompute_layer=False,
+01/03/2025 11:11:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                    tp_recompute_allgather=True,
+01/03/2025 11:11:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                    expert_parallel_size=1),
+01/03/2025 11:11:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:        model=ModelArgs(model_config=LlamaConfig(bos_token_id=0,
+01/03/2025 11:11:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                                 eos_token_id=0,
+01/03/2025 11:11:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                                 hidden_act='silu',
+01/03/2025 11:11:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                                 hidden_size=3072,
+01/03/2025 11:11:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                                 initializer_range=0.02,
+01/03/2025 11:11:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                                 intermediate_size=8192,
+01/03/2025 11:11:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                                 is_llama_config=True,
+01/03/2025 11:11:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                                 max_position_embeddings=4096,
+01/03/2025 11:11:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                                 num_attention_heads=32,
+01/03/2025 11:11:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                                 num_hidden_layers=28,
+01/03/2025 11:11:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                                 num_key_value_heads=32,
+01/03/2025 11:11:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                                 pad_token_id=None,
+01/03/2025 11:11:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                                 pretraining_tp=1,
+01/03/2025 11:11:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                                 rms_norm_eps=1e-05,
+01/03/2025 11:11:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                                 rope_scaling=None,
+01/03/2025 11:11:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                                 rope_theta=10000.0,
+01/03/2025 11:11:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                                 rope_interleaved=False,
+01/03/2025 11:11:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                                 tie_word_embeddings=True,
+01/03/2025 11:11:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                                 use_cache=True,
+01/03/2025 11:11:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                                 vocab_size=131072),
+01/03/2025 11:11:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                        init_method=RandomInit(std=0.02),
+01/03/2025 11:11:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                        dtype=torch.bfloat16,
+01/03/2025 11:11:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                        make_vocab_size_divisible_by=1,
+01/03/2025 11:11:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                        ddp_bucket_cap_mb=25),
+01/03/2025 11:11:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:        tokenizer=TokenizerArgs(tokenizer_name_or_path='robot-test/dummy-tokenizer-wordlevel',
+01/03/2025 11:11:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                tokenizer_revision=None,
+01/03/2025 11:11:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                tokenizer_max_length=None),
+01/03/2025 11:11:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:        checkpoints=CheckpointsArgs(checkpoints_path=PosixPath('checkpoints'),
+01/03/2025 11:11:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                    checkpoint_interval=10000,
+01/03/2025 11:11:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                    save_initial_state=False,
+01/03/2025 11:11:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                    save_final_state=False,
+01/03/2025 11:11:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                    resume_checkpoint_path=None,
+01/03/2025 11:11:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                    checkpoints_path_is_shared_file_system=False),
+01/03/2025 11:11:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:        logging=LoggingArgs(log_level='info',
+01/03/2025 11:11:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                            log_level_replica='info',
+01/03/2025 11:11:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                            iteration_step_info_interval=1),
+01/03/2025 11:11:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:        tokens=TokensArgs(sequence_length=4096,
+01/03/2025 11:11:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                          train_steps=100,
+01/03/2025 11:11:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                          micro_batch_size=1,
+01/03/2025 11:11:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                          batch_accumulation_per_replica=128,
+01/03/2025 11:11:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                          val_check_interval=100,
+01/03/2025 11:11:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                          limit_val_batches=0,
+01/03/2025 11:11:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                          limit_test_batches=0),
+01/03/2025 11:11:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:        optimizer=OptimizerArgs(optimizer_factory=AdamWOptimizerArgs(adam_eps=1e-08,
+01/03/2025 11:11:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                                                     adam_beta1=0.9,
+01/03/2025 11:11:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                                                     adam_beta2=0.95,
+01/03/2025 11:11:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                                                     torch_adam_is_fused=True,
+01/03/2025 11:11:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                                                     name='adamW'),
+01/03/2025 11:11:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                zero_stage=0,
+01/03/2025 11:11:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                weight_decay=0.01,
+01/03/2025 11:11:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                clip_grad=1.0,
+01/03/2025 11:11:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                accumulate_grad_in_fp32=True,
+01/03/2025 11:11:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                learning_rate_scheduler=LRSchedulerArgs(learning_rate=0.0003,
+01/03/2025 11:11:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                                                        lr_warmup_steps=2,
+01/03/2025 11:11:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                                                        lr_warmup_style='linear',
+01/03/2025 11:11:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                                                        lr_decay_style='cosine',
+01/03/2025 11:11:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                                                        lr_decay_steps=13,
+01/03/2025 11:11:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                                                        lr_decay_starting_step=None,
+01/03/2025 11:11:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                                                        min_decay_lr=1e-05)),
+01/03/2025 11:11:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:        data_stages=[DatasetStageArgs(name='Stable Training Stage',
+01/03/2025 11:11:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                      start_training_step=1,
+01/03/2025 11:11:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                      data=DataArgs(dataset=None,
+01/03/2025 11:11:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                                    seed=42,
+01/03/2025 11:11:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:                                                    num_loading_workers=1))],
+01/03/2025 11:11:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:        profiler=None,
+01/03/2025 11:11:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:        lighteval=None,
+01/03/2025 11:11:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:        s3_upload=None)
+01/03/2025 11:11:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]: Model Config:
+01/03/2025 11:11:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]: LlamaConfig(bos_token_id=0,
+01/03/2025 11:11:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:             eos_token_id=0,
+01/03/2025 11:11:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:             hidden_act='silu',
+01/03/2025 11:11:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:             hidden_size=3072,
+01/03/2025 11:11:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:             initializer_range=0.02,
+01/03/2025 11:11:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:             intermediate_size=8192,
+01/03/2025 11:11:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:             is_llama_config=True,
+01/03/2025 11:11:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:             max_position_embeddings=4096,
+01/03/2025 11:11:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:             num_attention_heads=32,
+01/03/2025 11:11:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:             num_hidden_layers=28,
+01/03/2025 11:11:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:             num_key_value_heads=32,
+01/03/2025 11:11:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:             pad_token_id=None,
+01/03/2025 11:11:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:             pretraining_tp=1,
+01/03/2025 11:11:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:             rms_norm_eps=1e-05,
+01/03/2025 11:11:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:             rope_scaling=None,
+01/03/2025 11:11:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:             rope_theta=10000.0,
+01/03/2025 11:11:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:             rope_interleaved=False,
+01/03/2025 11:11:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:             tie_word_embeddings=True,
+01/03/2025 11:11:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:             use_cache=True,
+01/03/2025 11:11:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:             vocab_size=131072)
+01/03/2025 11:11:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]: Building model..
+01/03/2025 11:11:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]: Initialize RoPE Theta = 10000.0
+01/03/2025 11:11:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]: Setting PP block ranks...
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+01/03/2025 11:11:58 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]: Total number of parameters: 3.98G (7586.67MiB)
+01/03/2025 11:11:58 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]: Local number of parameters: 107M (204.05MiB)
+01/03/2025 11:11:58 [INFO|DP=0|PP=0|TP=4|ip-26-0-160-100]: Local number of parameters: 107M (204.05MiB)
+01/03/2025 11:11:58 [INFO|DP=0|PP=0|TP=1|ip-26-0-160-100]: Local number of parameters: 107M (204.05MiB)
+01/03/2025 11:11:58 [INFO|DP=0|PP=0|TP=2|ip-26-0-160-100]: Local number of parameters: 107M (204.05MiB)
+01/03/2025 11:11:58 [INFO|DP=0|PP=0|TP=3|ip-26-0-160-100]: Local number of parameters: 107M (204.05MiB)
+01/03/2025 11:11:58 [INFO|DP=0|PP=0|TP=6|ip-26-0-160-100]: Local number of parameters: 107M (204.05MiB)
+01/03/2025 11:11:58 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]: [After model building] Memory usage: 204.56MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 11:11:58 [INFO|DP=0|PP=0|TP=5|ip-26-0-160-100]: Local number of parameters: 107M (204.05MiB)
+01/03/2025 11:11:58 [INFO|DP=0|PP=0|TP=4|ip-26-0-160-100]: [After model building] Memory usage: 204.56MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 11:11:58 [INFO|DP=0|PP=0|TP=1|ip-26-0-160-100]: [After model building] Memory usage: 204.56MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 11:11:58 [INFO|DP=0|PP=0|TP=2|ip-26-0-160-100]: [After model building] Memory usage: 204.56MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 11:11:58 [INFO|DP=0|PP=0|TP=6|ip-26-0-160-100]: [After model building] Memory usage: 204.56MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 11:11:58 [INFO|DP=0|PP=0|TP=3|ip-26-0-160-100]: [After model building] Memory usage: 204.56MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 11:11:58 [INFO|DP=0|PP=0|TP=5|ip-26-0-160-100]: [After model building] Memory usage: 204.56MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 11:11:58 [INFO|DP=0|PP=0|TP=7|ip-26-0-160-100]: Local number of parameters: 107M (204.05MiB)
+01/03/2025 11:11:58 [INFO|DP=0|PP=0|TP=7|ip-26-0-160-100]: [After model building] Memory usage: 204.56MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 11:11:58 [INFO|DP=0|PP=1|TP=3|ip-26-0-168-120]: Local number of parameters: 56.6M (108.05MiB)
+01/03/2025 11:11:58 [INFO|DP=0|PP=6|TP=3|ip-26-0-169-86]: Local number of parameters: 56.6M (108.05MiB)
+01/03/2025 11:11:58 [INFO|DP=0|PP=5|TP=3|ip-26-0-169-239]: Local number of parameters: 56.6M (108.05MiB)
+01/03/2025 11:11:58 [INFO|DP=0|PP=2|TP=3|ip-26-0-168-34]: Local number of parameters: 56.6M (108.05MiB)
+01/03/2025 11:11:58 [INFO|DP=0|PP=3|TP=3|ip-26-0-168-95]: Local number of parameters: 56.6M (108.05MiB)
+01/03/2025 11:11:58 [INFO|DP=0|PP=4|TP=3|ip-26-0-169-139]: Local number of parameters: 56.6M (108.05MiB)
+01/03/2025 11:11:58 [INFO|DP=0|PP=7|TP=3|ip-26-0-170-143]: Local number of parameters: 50.3M (96.01MiB)
+01/03/2025 11:11:58 [INFO|DP=0|PP=4|TP=1|ip-26-0-169-139]: Local number of parameters: 56.6M (108.05MiB)
+01/03/2025 11:11:58 [INFO|DP=0|PP=1|TP=1|ip-26-0-168-120]: Local number of parameters: 56.6M (108.05MiB)
+01/03/2025 11:11:58 [INFO|DP=0|PP=2|TP=4|ip-26-0-168-34]: Local number of parameters: 56.6M (108.05MiB)
+01/03/2025 11:11:58 [INFO|DP=0|PP=1|TP=3|ip-26-0-168-120]: [After model building] Memory usage: 108.56MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 11:11:58 [INFO|DP=0|PP=6|TP=4|ip-26-0-169-86]: Local number of parameters: 56.6M (108.05MiB)
+01/03/2025 11:11:58 [INFO|DP=0|PP=2|TP=6|ip-26-0-168-34]: Local number of parameters: 56.6M (108.05MiB)
+01/03/2025 11:11:58 [INFO|DP=0|PP=3|TP=3|ip-26-0-168-95]: [After model building] Memory usage: 108.56MiB. Peak allocated: 5392.00MiB Peak reserved: 9494.00MiB
+01/03/2025 11:11:58 [INFO|DP=0|PP=3|TP=1|ip-26-0-168-95]: Local number of parameters: 56.6M (108.05MiB)
+01/03/2025 11:11:58 [INFO|DP=0|PP=4|TP=4|ip-26-0-169-139]: Local number of parameters: 56.6M (108.05MiB)
+01/03/2025 11:11:58 [INFO|DP=0|PP=4|TP=6|ip-26-0-169-139]: Local number of parameters: 56.6M (108.05MiB)
+01/03/2025 11:11:58 [INFO|DP=0|PP=5|TP=1|ip-26-0-169-239]: Local number of parameters: 56.6M (108.05MiB)
+01/03/2025 11:11:58 [INFO|DP=0|PP=5|TP=4|ip-26-0-169-239]: Local number of parameters: 56.6M (108.05MiB)
+01/03/2025 11:11:58 [INFO|DP=0|PP=1|TP=2|ip-26-0-168-120]: Local number of parameters: 56.6M (108.05MiB)
+01/03/2025 11:11:58 [INFO|DP=0|PP=6|TP=3|ip-26-0-169-86]: [After model building] Memory usage: 108.56MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 11:11:58 [INFO|DP=0|PP=6|TP=1|ip-26-0-169-86]: Local number of parameters: 56.6M (108.05MiB)
+01/03/2025 11:11:58 [INFO|DP=0|PP=2|TP=5|ip-26-0-168-34]: Local number of parameters: 56.6M (108.05MiB)
+01/03/2025 11:11:58 [INFO|DP=0|PP=2|TP=1|ip-26-0-168-34]: Local number of parameters: 56.6M (108.05MiB)
+01/03/2025 11:11:58 [INFO|DP=0|PP=5|TP=3|ip-26-0-169-239]: [After model building] Memory usage: 108.56MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 11:11:58 [INFO|DP=0|PP=4|TP=2|ip-26-0-169-139]: Local number of parameters: 56.6M (108.05MiB)
+01/03/2025 11:11:58 [INFO|DP=0|PP=4|TP=3|ip-26-0-169-139]: [After model building] Memory usage: 108.56MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 11:11:58 [INFO|DP=0|PP=4|TP=5|ip-26-0-169-139]: Local number of parameters: 56.6M (108.05MiB)
+01/03/2025 11:11:58 [INFO|DP=0|PP=6|TP=5|ip-26-0-169-86]: Local number of parameters: 56.6M (108.05MiB)
+01/03/2025 11:11:58 [INFO|DP=0|PP=1|TP=4|ip-26-0-168-120]: Local number of parameters: 56.6M (108.05MiB)
+01/03/2025 11:11:58 [INFO|DP=0|PP=2|TP=3|ip-26-0-168-34]: [After model building] Memory usage: 108.56MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 11:11:58 [INFO|DP=0|PP=3|TP=2|ip-26-0-168-95]: Local number of parameters: 56.6M (108.05MiB)
+01/03/2025 11:11:58 [INFO|DP=0|PP=3|TP=4|ip-26-0-168-95]: Local number of parameters: 56.6M (108.05MiB)
+01/03/2025 11:11:58 [INFO|DP=0|PP=7|TP=4|ip-26-0-170-143]: Local number of parameters: 50.3M (96.01MiB)
+01/03/2025 11:11:58 [INFO|DP=0|PP=7|TP=6|ip-26-0-170-143]: Local number of parameters: 50.3M (96.01MiB)
+01/03/2025 11:11:58 [INFO|DP=0|PP=7|TP=2|ip-26-0-170-143]: Local number of parameters: 50.3M (96.01MiB)
+01/03/2025 11:11:58 [INFO|DP=0|PP=1|TP=6|ip-26-0-168-120]: Local number of parameters: 56.6M (108.05MiB)
+01/03/2025 11:11:58 [INFO|DP=0|PP=5|TP=6|ip-26-0-169-239]: Local number of parameters: 56.6M (108.05MiB)
+01/03/2025 11:11:58 [INFO|DP=0|PP=7|TP=5|ip-26-0-170-143]: Local number of parameters: 50.3M (96.01MiB)
+01/03/2025 11:11:58 [INFO|DP=0|PP=7|TP=1|ip-26-0-170-143]: Local number of parameters: 50.3M (96.01MiB)
+01/03/2025 11:11:58 [INFO|DP=0|PP=6|TP=6|ip-26-0-169-86]: Local number of parameters: 56.6M (108.05MiB)
+01/03/2025 11:11:58 [INFO|DP=0|PP=6|TP=2|ip-26-0-169-86]: Local number of parameters: 56.6M (108.05MiB)
+01/03/2025 11:11:58 [INFO|DP=0|PP=7|TP=3|ip-26-0-170-143]: [After model building] Memory usage: 96.02MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 11:11:58 [INFO|DP=0|PP=5|TP=5|ip-26-0-169-239]: Local number of parameters: 56.6M (108.05MiB)
+01/03/2025 11:11:58 [INFO|DP=0|PP=2|TP=2|ip-26-0-168-34]: Local number of parameters: 56.6M (108.05MiB)
+01/03/2025 11:11:58 [INFO|DP=0|PP=5|TP=2|ip-26-0-169-239]: Local number of parameters: 56.6M (108.05MiB)
+01/03/2025 11:11:58 [INFO|DP=0|PP=4|TP=0|ip-26-0-169-139]: Local number of parameters: 56.6M (108.05MiB)
+01/03/2025 11:11:58 [INFO|DP=0|PP=1|TP=1|ip-26-0-168-120]: [After model building] Memory usage: 108.56MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 11:11:58 [INFO|DP=0|PP=2|TP=4|ip-26-0-168-34]: [After model building] Memory usage: 108.56MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 11:11:58 [INFO|DP=0|PP=3|TP=5|ip-26-0-168-95]: Local number of parameters: 56.6M (108.05MiB)
+01/03/2025 11:11:58 [INFO|DP=0|PP=4|TP=1|ip-26-0-169-139]: [After model building] Memory usage: 108.56MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 11:11:58 [INFO|DP=0|PP=7|TP=0|ip-26-0-170-143]: Local number of parameters: 50.3M (96.01MiB)
+01/03/2025 11:11:58 [INFO|DP=0|PP=2|TP=6|ip-26-0-168-34]: [After model building] Memory usage: 108.56MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 11:11:58 [INFO|DP=0|PP=6|TP=4|ip-26-0-169-86]: [After model building] Memory usage: 108.56MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 11:11:58 [INFO|DP=0|PP=1|TP=5|ip-26-0-168-120]: Local number of parameters: 56.6M (108.05MiB)
+01/03/2025 11:11:58 [INFO|DP=0|PP=5|TP=1|ip-26-0-169-239]: [After model building] Memory usage: 108.56MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 11:11:58 [INFO|DP=0|PP=4|TP=6|ip-26-0-169-139]: [After model building] Memory usage: 108.56MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 11:11:58 [INFO|DP=0|PP=2|TP=5|ip-26-0-168-34]: [After model building] Memory usage: 108.56MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 11:11:58 [INFO|DP=0|PP=1|TP=2|ip-26-0-168-120]: [After model building] Memory usage: 108.56MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 11:11:58 [INFO|DP=0|PP=5|TP=4|ip-26-0-169-239]: [After model building] Memory usage: 108.56MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 11:11:58 [INFO|DP=0|PP=3|TP=6|ip-26-0-168-95]: Local number of parameters: 56.6M (108.05MiB)
+01/03/2025 11:11:58 [INFO|DP=0|PP=4|TP=4|ip-26-0-169-139]: [After model building] Memory usage: 108.56MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 11:11:58 [INFO|DP=0|PP=6|TP=0|ip-26-0-169-86]: Local number of parameters: 56.6M (108.05MiB)
+01/03/2025 11:11:58 [INFO|DP=0|PP=6|TP=1|ip-26-0-169-86]: [After model building] Memory usage: 108.56MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 11:11:58 [INFO|DP=0|PP=7|TP=6|ip-26-0-170-143]: [After model building] Memory usage: 96.02MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 11:11:58 [INFO|DP=0|PP=3|TP=1|ip-26-0-168-95]: [After model building] Memory usage: 108.56MiB. Peak allocated: 5392.00MiB Peak reserved: 9494.00MiB
+01/03/2025 11:11:58 [INFO|DP=0|PP=4|TP=5|ip-26-0-169-139]: [After model building] Memory usage: 108.56MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 11:11:58 [INFO|DP=0|PP=6|TP=5|ip-26-0-169-86]: [After model building] Memory usage: 108.56MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 11:11:58 [INFO|DP=0|PP=7|TP=5|ip-26-0-170-143]: [After model building] Memory usage: 96.02MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 11:11:58 [INFO|DP=0|PP=7|TP=4|ip-26-0-170-143]: [After model building] Memory usage: 96.02MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 11:11:58 [INFO|DP=0|PP=4|TP=2|ip-26-0-169-139]: [After model building] Memory usage: 108.56MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 11:11:58 [INFO|DP=0|PP=5|TP=6|ip-26-0-169-239]: [After model building] Memory usage: 108.56MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 11:11:58 [INFO|DP=0|PP=7|TP=2|ip-26-0-170-143]: [After model building] Memory usage: 96.02MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 11:11:58 [INFO|DP=0|PP=2|TP=1|ip-26-0-168-34]: [After model building] Memory usage: 108.56MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 11:11:58 [INFO|DP=0|PP=2|TP=0|ip-26-0-168-34]: Local number of parameters: 56.6M (108.05MiB)
+01/03/2025 11:11:58 [INFO|DP=0|PP=1|TP=6|ip-26-0-168-120]: [After model building] Memory usage: 108.56MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 11:11:58 [INFO|DP=0|PP=6|TP=2|ip-26-0-169-86]: [After model building] Memory usage: 108.56MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 11:11:58 [INFO|DP=0|PP=3|TP=4|ip-26-0-168-95]: [After model building] Memory usage: 108.56MiB. Peak allocated: 5392.00MiB Peak reserved: 9494.00MiB
+01/03/2025 11:11:58 [INFO|DP=0|PP=3|TP=2|ip-26-0-168-95]: [After model building] Memory usage: 108.56MiB. Peak allocated: 5392.00MiB Peak reserved: 9494.00MiB
+01/03/2025 11:11:58 [INFO|DP=0|PP=7|TP=1|ip-26-0-170-143]: [After model building] Memory usage: 96.02MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 11:11:58 [INFO|DP=0|PP=1|TP=4|ip-26-0-168-120]: [After model building] Memory usage: 108.56MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 11:11:58 [INFO|DP=0|PP=6|TP=6|ip-26-0-169-86]: [After model building] Memory usage: 108.56MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 11:11:58 [INFO|DP=0|PP=2|TP=2|ip-26-0-168-34]: [After model building] Memory usage: 108.56MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 11:11:58 [INFO|DP=0|PP=5|TP=5|ip-26-0-169-239]: [After model building] Memory usage: 108.56MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 11:11:58 [INFO|DP=0|PP=4|TP=0|ip-26-0-169-139]: [After model building] Memory usage: 108.56MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 11:11:58 [INFO|DP=0|PP=7|TP=0|ip-26-0-170-143]: [After model building] Memory usage: 96.02MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 11:11:58 [INFO|DP=0|PP=5|TP=2|ip-26-0-169-239]: [After model building] Memory usage: 108.56MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 11:11:58 [INFO|DP=0|PP=1|TP=5|ip-26-0-168-120]: [After model building] Memory usage: 108.56MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 11:11:58 [INFO|DP=0|PP=3|TP=5|ip-26-0-168-95]: [After model building] Memory usage: 108.56MiB. Peak allocated: 5392.00MiB Peak reserved: 9494.00MiB
+01/03/2025 11:11:58 [INFO|DP=0|PP=3|TP=6|ip-26-0-168-95]: [After model building] Memory usage: 108.56MiB. Peak allocated: 5392.00MiB Peak reserved: 9494.00MiB
+01/03/2025 11:11:58 [INFO|DP=0|PP=6|TP=0|ip-26-0-169-86]: [After model building] Memory usage: 108.56MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 11:11:58 [INFO|DP=0|PP=2|TP=0|ip-26-0-168-34]: [After model building] Memory usage: 108.56MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 11:11:58 [INFO|DP=0|PP=4|TP=7|ip-26-0-169-139]: Local number of parameters: 56.6M (108.05MiB)
+01/03/2025 11:11:58 [INFO|DP=0|PP=3|TP=7|ip-26-0-168-95]: Local number of parameters: 56.6M (108.05MiB)
+01/03/2025 11:11:58 [INFO|DP=0|PP=1|TP=7|ip-26-0-168-120]: Local number of parameters: 56.6M (108.05MiB)
+01/03/2025 11:11:58 [INFO|DP=0|PP=2|TP=7|ip-26-0-168-34]: Local number of parameters: 56.6M (108.05MiB)
+NCCL version 2.18.5+cuda12.2
+01/03/2025 11:11:58 [INFO|DP=0|PP=7|TP=7|ip-26-0-170-143]: Local number of parameters: 50.3M (96.01MiB)
+NCCL version 2.18.5+cuda12.2
+01/03/2025 11:11:58 [INFO|DP=0|PP=6|TP=7|ip-26-0-169-86]: Local number of parameters: 56.6M (108.05MiB)
+01/03/2025 11:11:58 [INFO|DP=0|PP=5|TP=7|ip-26-0-169-239]: Local number of parameters: 56.6M (108.05MiB)
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+01/03/2025 11:11:58 [INFO|DP=0|PP=3|TP=7|ip-26-0-168-95]: [After model building] Memory usage: 108.56MiB. Peak allocated: 5392.00MiB Peak reserved: 9494.00MiB
+NCCL version 2.18.5+cuda12.2
+01/03/2025 11:11:58 [INFO|DP=0|PP=4|TP=7|ip-26-0-169-139]: [After model building] Memory usage: 108.56MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+01/03/2025 11:11:58 [INFO|DP=0|PP=1|TP=7|ip-26-0-168-120]: [After model building] Memory usage: 108.56MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+NCCL version 2.18.5+cuda12.2
+01/03/2025 11:11:58 [INFO|DP=0|PP=2|TP=7|ip-26-0-168-34]: [After model building] Memory usage: 108.56MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+01/03/2025 11:11:58 [INFO|DP=0|PP=7|TP=7|ip-26-0-170-143]: [After model building] Memory usage: 96.02MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+NCCL version 2.18.5+cuda12.2
+01/03/2025 11:11:58 [INFO|DP=0|PP=6|TP=7|ip-26-0-169-86]: [After model building] Memory usage: 108.56MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+NCCL version 2.18.5+cuda12.2
+01/03/2025 11:11:58 [INFO|DP=0|PP=5|TP=7|ip-26-0-169-239]: [After model building] Memory usage: 108.56MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+01/03/2025 11:11:58 [INFO|DP=0|PP=5|TP=0|ip-26-0-169-239]: Local number of parameters: 56.6M (108.05MiB)
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+01/03/2025 11:11:58 [INFO|DP=0|PP=3|TP=0|ip-26-0-168-95]: Local number of parameters: 56.6M (108.05MiB)
+01/03/2025 11:11:58 [INFO|DP=0|PP=1|TP=0|ip-26-0-168-120]: Local number of parameters: 56.6M (108.05MiB)
+01/03/2025 11:11:58 [INFO|DP=0|PP=5|TP=0|ip-26-0-169-239]: [After model building] Memory usage: 108.56MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 11:11:58 [INFO|DP=0|PP=1|TP=0|ip-26-0-168-120]: [After model building] Memory usage: 108.56MiB. Peak allocated: 5392.00MiB Peak reserved: 10518.00MiB
+01/03/2025 11:11:58 [INFO|DP=0|PP=3|TP=0|ip-26-0-168-95]: [After model building] Memory usage: 108.56MiB. Peak allocated: 5392.00MiB Peak reserved: 9494.00MiB
+01/03/2025 11:11:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]: No checkpoint path provided.
+01/03/2025 11:11:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]: Parametrizing model parameters using StandardParametrizator
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+01/03/2025 11:12:00 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]: [Optimizer Building] Using LearningRateForSP as learning rate
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+01/03/2025 11:12:00 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]: module_name: model.token_position_embeddings | PP: 0/8 | Block rank: 0
+01/03/2025 11:12:00 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]: module_name: model.decoder.0 | PP: 0/8 | Block rank: 1
+01/03/2025 11:12:00 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]: module_name: model.decoder.1 | PP: 0/8 | Block rank: 2
+01/03/2025 11:12:00 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]: module_name: model.decoder.2 | PP: 0/8 | Block rank: 3
+01/03/2025 11:12:00 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]: module_name: model.decoder.3 | PP: 0/8 | Block rank: 4
+01/03/2025 11:12:00 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]: module_name: model.decoder.4 | PP: 1/8 | Block rank: 0
+01/03/2025 11:12:00 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]: module_name: model.decoder.5 | PP: 1/8 | Block rank: 1
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+01/03/2025 11:12:00 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]: module_name: model.decoder.6 | PP: 1/8 | Block rank: 2
+01/03/2025 11:12:00 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]: module_name: model.decoder.7 | PP: 1/8 | Block rank: 3
+01/03/2025 11:12:00 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]: module_name: model.decoder.8 | PP: 2/8 | Block rank: 0
+01/03/2025 11:12:00 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]: module_name: model.decoder.9 | PP: 2/8 | Block rank: 1
+01/03/2025 11:12:00 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]: module_name: model.decoder.10 | PP: 2/8 | Block rank: 2
+01/03/2025 11:12:00 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]: module_name: model.decoder.11 | PP: 2/8 | Block rank: 3
+01/03/2025 11:12:00 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]: module_name: model.decoder.12 | PP: 3/8 | Block rank: 0
+01/03/2025 11:12:00 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]: module_name: model.decoder.13 | PP: 3/8 | Block rank: 1
+01/03/2025 11:12:00 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]: module_name: model.decoder.14 | PP: 3/8 | Block rank: 2
+01/03/2025 11:12:00 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]: module_name: model.decoder.15 | PP: 3/8 | Block rank: 3
+01/03/2025 11:12:00 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]: module_name: model.decoder.16 | PP: 4/8 | Block rank: 0
+01/03/2025 11:12:00 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]: module_name: model.decoder.17 | PP: 4/8 | Block rank: 1
+01/03/2025 11:12:00 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]: module_name: model.decoder.18 | PP: 4/8 | Block rank: 2
+01/03/2025 11:12:00 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]: module_name: model.decoder.19 | PP: 4/8 | Block rank: 3
+01/03/2025 11:12:00 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]: module_name: model.decoder.20 | PP: 5/8 | Block rank: 0
+01/03/2025 11:12:00 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]: module_name: model.decoder.21 | PP: 5/8 | Block rank: 1
+01/03/2025 11:12:00 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]: module_name: model.decoder.22 | PP: 5/8 | Block rank: 2
+01/03/2025 11:12:00 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]: module_name: model.decoder.23 | PP: 5/8 | Block rank: 3
+01/03/2025 11:12:00 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]: module_name: model.decoder.24 | PP: 6/8 | Block rank: 0
+01/03/2025 11:12:00 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]: module_name: model.decoder.25 | PP: 6/8 | Block rank: 1
+01/03/2025 11:12:00 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]: module_name: model.decoder.26 | PP: 6/8 | Block rank: 2
+01/03/2025 11:12:00 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]: module_name: model.decoder.27 | PP: 6/8 | Block rank: 3
+01/03/2025 11:12:00 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]: module_name: model.final_layer_norm | PP: 7/8 | Block rank: 0
+01/03/2025 11:12:00 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]: module_name: model.lm_head | PP: 7/8 | Block rank: 1
+01/03/2025 11:12:00 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]: module_name: model.cast_to_fp32 | PP: 7/8 | Block rank: 2
+01/03/2025 11:12:00 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]: module_name: loss | PP: 7/8 | Block rank: 3
+01/03/2025 11:12:00 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]: [Training Plan] Stage Stable Training Stage has 99 remaining training steps and has consumed 0 samples
+01/03/2025 11:12:00 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]: Using dummy data generator
+01/03/2025 11:12:00 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]: [Training Plan] There are 1 training stages 
+01/03/2025 11:12:00 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]: [Stage Stable Training Stage] start from step 1 
+01/03/2025 11:12:00 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]: 
+01/03/2025 11:12:00 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]: [Start training] datetime: 2025-01-03 11:12:00.756855 | mbs: 1 | grad_accum: 128 | global_batch_size: 256 | sequence_length: 4096 | train_steps: 100 | start_iteration_step: 0 | consumed_train_samples: 0
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+01/03/2025 11:12:00 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]: Resuming training from stage Stable Training Stage, it has trained for 0 samples and has 99 remaining train steps
+01/03/2025 11:12:00 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:  Memory usage: 1224.79MiB. Peak allocated 5392.00MiB. Peak reserved: 10518.00MiB
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+NCCL version 2.18.5+cuda12.2
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+NCCL version 2.18.5+cuda12.2
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+NCCL version 2.18.5+cuda12.2
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+NCCL version 2.18.5+cuda12.2
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+NCCL version 2.18.5+cuda12.2
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+NCCL version 2.18.5+cuda12.2
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+NCCL version 2.18.5+cuda12.2
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+NCCL version 2.18.5+cuda12.2
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+NCCL version 2.18.5+cuda12.2
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+NCCL version 2.18.5+cuda12.2
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+01/03/2025 11:15:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:  Memory usage: 1339.83MiB. Peak allocated 3578.19MiB. Peak reserved: 3892.00MiB
+01/03/2025 11:15:16 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:  Memory usage: 2156.03MiB. Peak allocated 2156.03MiB. Peak reserved: 3892.00MiB
+01/03/2025 11:15:16 [INFO|DP=0|PP=7|TP=0|ip-26-0-170-143]: iteration: 1 / 100 | consumed_tokens: 1.05M | elapsed_time_per_iteration_ms: 194K | tokens_per_sec: 5.41K | tokens_per_sec_per_gpu: 42.3 | global_batch_size: 256 | lm_loss: 12.1 | lr: 0.00015 | model_tflops_per_gpu: 1.09 | hardware_tflops_per_gpu: 1.09 | grad_norm: 0.901 | cuda_memory_allocated: 1.1G | cuda_max_memory_reserved: 1.75G | hd_total_memory_tb: 312G | hd_used_memory_tb: 71.3G | hd_free_memory_tb: 241G
+01/03/2025 11:17:58 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:  Memory usage: 2156.03MiB. Peak allocated 4225.62MiB. Peak reserved: 4804.00MiB
+01/03/2025 11:17:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:  Memory usage: 2156.03MiB. Peak allocated 2156.04MiB. Peak reserved: 4804.00MiB
+01/03/2025 11:17:59 [INFO|DP=0|PP=7|TP=0|ip-26-0-170-143]: iteration: 2 / 100 | consumed_tokens: 2.1M | elapsed_time_per_iteration_ms: 163K | tokens_per_sec: 6.43K | tokens_per_sec_per_gpu: 50.3 | global_batch_size: 256 | lm_loss: 12.1 | lr: 0.0003 | model_tflops_per_gpu: 1.29 | hardware_tflops_per_gpu: 1.29 | grad_norm: 0.901 | cuda_memory_allocated: 1.1G | cuda_max_memory_reserved: 2.02G | hd_total_memory_tb: 312G | hd_used_memory_tb: 71.3G | hd_free_memory_tb: 241G
+01/03/2025 11:20:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]:  Memory usage: 2156.03MiB. Peak allocated 4242.48MiB. Peak reserved: 4804.00MiB
+01/03/2025 11:20:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]: | job_id   | name                                                            | nodes | seq_len | mbs | batch_accum | gbs | mTFLOPs | hTFLOPs | tok/s/gpu | AllReduce (GB/s) | AllGather (GB/s) | ReduceScatter (GB/s) | AR Intra-node (GB/s) | AG Intra-node (GB/s) | RS Intra-node (GB/s) | Mem Alloc (GB) | Mem Res (GB) | dp | pp | tp | pp_engine                           | tp_mode        | tp_async_comm | hidden_size | hidden_act | num_layers | num_heads | num_kv_heads | max_pos | vocab_size | tie_word_embeddings | dtype          | zero_stage | ddp_bucket_cap_mb | accumulate_grad_in_fp32 | Total Params | Local Params |
+01/03/2025 11:20:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]: | -------- | --------------------------------------------------------------- | ----- | ------- | --- | ----------- | --- | ------- | ------- | --------- | ---------------- | ---------------- | -------------------- | -------------------- | -------------------- | -------------------- | -------------- | ------------ | -- | -- | -- | ----------------------------------- | -------------- | ------------- | ----------- | ---------- | ---------- | --------- | ------------ | ------- | ---------- | ------------------- | -------------- | ---------- | ----------------- | ----------------------- | ------------ | ------------ |
+01/03/2025 11:20:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]: | 14023604 | 3.57G_dp2_tp8_pp8_acc128_mbs1_seq4096_zero0_tpmodeRED_vocab131k | 16    | 4096    | 1   | 128         | 256 | 1.30    | 1.30    | 50.48     | 47.27            | 25.54            | 25.74                | 424.52               | 241.66               | 240.51               | 2.11           | 4.69         | 2  | 8  | 8  | OneForwardOneBackwardPipelineEngine | REDUCE_SCATTER | True          | 3072        | silu       | 28         | 32        | 32           | 4096    | 131072     | True                | torch.bfloat16 | 0          | 25                | True                    | 3.98G        | 107M         |
+01/03/2025 11:20:41 [INFO|DP=0|PP=7|TP=0|ip-26-0-170-143]: iteration: 3 / 100 | consumed_tokens: 3.15M | elapsed_time_per_iteration_ms: 162K | tokens_per_sec: 6.46K | tokens_per_sec_per_gpu: 50.5 | global_batch_size: 256 | lm_loss: 12.1 | lr: 0.000296 | model_tflops_per_gpu: 1.3 | hardware_tflops_per_gpu: 1.3 | grad_norm: 0.871 | cuda_memory_allocated: 1.1G | cuda_max_memory_reserved: 2.02G | hd_total_memory_tb: 312G | hd_used_memory_tb: 71.3G | hd_free_memory_tb: 241G
+01/03/2025 11:20:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]: Attempting to write benchmark results to CSV file: benchmark/results/bench_final2.csv
+01/03/2025 11:20:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]: Acquired lock for CSV file: benchmark/results/bench_final2.csv
+01/03/2025 11:20:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]: Successfully wrote to CSV file: benchmark/results/bench_final2.csv. Releasing lock...
+01/03/2025 11:20:41 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-100]: Throughput logging complete
+[2025-01-03 11:21:11,645] torch.distributed.elastic.multiprocessing.api: [ERROR] failed (exitcode: 1) local_rank: 0 (pid: 62915) of binary: /fsx/nouamane/miniconda/envs/2-1-cu121/bin/python
+[2025-01-03 11:21:11,666] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 300994 closing signal SIGTERM
+[2025-01-03 11:21:11,666] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 300995 closing signal SIGTERM
+[2025-01-03 11:21:11,666] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 300996 closing signal SIGTERM
+[2025-01-03 11:21:11,666] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 300997 closing signal SIGTERM
+[2025-01-03 11:21:11,666] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 300998 closing signal SIGTERM
+[2025-01-03 11:21:11,666] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 300999 closing signal SIGTERM
+[2025-01-03 11:21:11,666] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 301001 closing signal SIGTERM
+[2025-01-03 11:21:11,673] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 87660 closing signal SIGTERM
+[2025-01-03 11:21:11,673] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 61917 closing signal SIGTERM
+[2025-01-03 11:21:11,673] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 87661 closing signal SIGTERM
+[2025-01-03 11:21:11,673] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 87662 closing signal SIGTERM
+[2025-01-03 11:21:11,673] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 61918 closing signal SIGTERM
+[2025-01-03 11:21:11,673] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 87663 closing signal SIGTERM
+[2025-01-03 11:21:11,673] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 61919 closing signal SIGTERM
+[2025-01-03 11:21:11,673] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 87664 closing signal SIGTERM
+[2025-01-03 11:21:11,673] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 87665 closing signal SIGTERM
+[2025-01-03 11:21:11,673] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 61920 closing signal SIGTERM
+[2025-01-03 11:21:11,673] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 87666 closing signal SIGTERM
+[2025-01-03 11:21:11,673] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 61921 closing signal SIGTERM
+[2025-01-03 11:21:11,673] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 61922 closing signal SIGTERM
+[2025-01-03 11:21:11,673] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 61923 closing signal SIGTERM
+[2025-01-03 11:21:11,697] torch.distributed.elastic.multiprocessing.api: [ERROR] failed (exitcode: 1) local_rank: 0 (pid: 288815) of binary: /fsx/nouamane/miniconda/envs/2-1-cu121/bin/python
+[2025-01-03 11:21:11,708] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 303651 closing signal SIGTERM
+[2025-01-03 11:21:11,709] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 303652 closing signal SIGTERM
+[2025-01-03 11:21:11,709] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 303653 closing signal SIGTERM
+[2025-01-03 11:21:11,710] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 197675 closing signal SIGTERM
+[2025-01-03 11:21:11,709] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 303655 closing signal SIGTERM
+[2025-01-03 11:21:11,709] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 303656 closing signal SIGTERM
+[2025-01-03 11:21:11,709] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 303657 closing signal SIGTERM
+[2025-01-03 11:21:11,710] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 197676 closing signal SIGTERM
+[2025-01-03 11:21:11,709] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 303658 closing signal SIGTERM
+[2025-01-03 11:21:11,710] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 197677 closing signal SIGTERM
+[2025-01-03 11:21:11,711] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 197678 closing signal SIGTERM
+[2025-01-03 11:21:11,711] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 197679 closing signal SIGTERM
+[2025-01-03 11:21:11,711] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 197680 closing signal SIGTERM
+[2025-01-03 11:21:11,711] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 197682 closing signal SIGTERM
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 264, in launch_agent
+    raise ChildFailedError(
+torch.distributed.elastic.multiprocessing.errors.ChildFailedError: 
+============================================================
+/fsx/nouamane/projects/nanotron/run_train.py FAILED
+------------------------------------------------------------
+Failures:
+[1]:
+  time      : 2025-01-03_11:21:11
+  host      : ip-26-0-170-31.ec2.internal
+  rank      : 121 (local_rank: 1)
+  exitcode  : 1 (pid: 62916)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[2]:
+  time      : 2025-01-03_11:21:11
+  host      : ip-26-0-170-31.ec2.internal
+  rank      : 122 (local_rank: 2)
+  exitcode  : 1 (pid: 62917)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[3]:
+  time      : 2025-01-03_11:21:11
+  host      : ip-26-0-170-31.ec2.internal
+  rank      : 123 (local_rank: 3)
+  exitcode  : 1 (pid: 62918)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[4]:
+  time      : 2025-01-03_11:21:11
+  host      : ip-26-0-170-31.ec2.internal
+  rank      : 124 (local_rank: 4)
+  exitcode  : 1 (pid: 62919)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[5]:
+  time      : 2025-01-03_11:21:11
+  host      : ip-26-0-170-31.ec2.internal
+  rank      : 125 (local_rank: 5)
+  exitcode  : 1 (pid: 62920)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[6]:
+  time      : 2025-01-03_11:21:11
+  host      : ip-26-0-170-31.ec2.internal
+  rank      : 126 (local_rank: 6)
+  exitcode  : 1 (pid: 62921)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[7]:
+  time      : 2025-01-03_11:21:11
+  host      : ip-26-0-170-31.ec2.internal
+  rank      : 127 (local_rank: 7)
+  exitcode  : 1 (pid: 62922)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+------------------------------------------------------------
+Root Cause (first observed failure):
+[0]:
+  time      : 2025-01-03_11:21:11
+  host      : ip-26-0-170-31.ec2.internal
+  rank      : 120 (local_rank: 0)
+  exitcode  : 1 (pid: 62915)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+============================================================
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+[2025-01-03 11:21:11,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 177775 closing signal SIGTERM
+[2025-01-03 11:21:11,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 177776 closing signal SIGTERM
+[2025-01-03 11:21:11,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 177777 closing signal SIGTERM
+[2025-01-03 11:21:11,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 177778 closing signal SIGTERM
+[2025-01-03 11:21:11,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 177779 closing signal SIGTERM
+[2025-01-03 11:21:11,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 177780 closing signal SIGTERM
+[2025-01-03 11:21:11,768] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 177781 closing signal SIGTERM
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 264, in launch_agent
+    raise ChildFailedError(
+torch.distributed.elastic.multiprocessing.errors.ChildFailedError: 
+============================================================
+/fsx/nouamane/projects/nanotron/run_train.py FAILED
+------------------------------------------------------------
+Failures:
+[1]:
+  time      : 2025-01-03_11:21:11
+  host      : ip-26-0-160-100.ec2.internal
+  rank      : 1 (local_rank: 1)
+  exitcode  : 1 (pid: 288816)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[2]:
+  time      : 2025-01-03_11:21:11
+  host      : ip-26-0-160-100.ec2.internal
+  rank      : 2 (local_rank: 2)
+  exitcode  : 1 (pid: 288817)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[3]:
+  time      : 2025-01-03_11:21:11
+  host      : ip-26-0-160-100.ec2.internal
+  rank      : 3 (local_rank: 3)
+  exitcode  : 1 (pid: 288818)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[4]:
+  time      : 2025-01-03_11:21:11
+  host      : ip-26-0-160-100.ec2.internal
+  rank      : 4 (local_rank: 4)
+  exitcode  : 1 (pid: 288819)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[5]:
+  time      : 2025-01-03_11:21:11
+  host      : ip-26-0-160-100.ec2.internal
+  rank      : 5 (local_rank: 5)
+  exitcode  : 1 (pid: 288820)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[6]:
+  time      : 2025-01-03_11:21:11
+  host      : ip-26-0-160-100.ec2.internal
+  rank      : 6 (local_rank: 6)
+  exitcode  : 1 (pid: 288821)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[7]:
+  time      : 2025-01-03_11:21:11
+  host      : ip-26-0-160-100.ec2.internal
+  rank      : 7 (local_rank: 7)
+  exitcode  : 1 (pid: 288822)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+------------------------------------------------------------
+Root Cause (first observed failure):
+[0]:
+  time      : 2025-01-03_11:21:11
+  host      : ip-26-0-160-100.ec2.internal
+  rank      : 0 (local_rank: 0)
+  exitcode  : 1 (pid: 288815)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+============================================================
+srun: error: ip-26-0-160-100: task 0: Exited with exit code 1
+srun: launch/slurm: _step_signal: Terminating StepId=14023604.0
+[2025-01-03 11:21:12,056] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 11:21:12,056] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 61918 closing signal SIGTERM
+[2025-01-03 11:21:12,056] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 61919 closing signal SIGTERM
+[2025-01-03 11:21:12,057] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 61920 closing signal SIGTERM
+[2025-01-03 11:21:12,057] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 61921 closing signal SIGTERM
+[2025-01-03 11:21:12,057] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 61922 closing signal SIGTERM
+[2025-01-03 11:21:12,057] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 61923 closing signal SIGTERM
+[2025-01-03 11:21:12,056] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 11:21:12,057] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 11:21:12,057] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 166390 closing signal SIGTERM
+[2025-01-03 11:21:12,057] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 166391 closing signal SIGTERM
+[2025-01-03 11:21:12,057] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 300994 closing signal SIGTERM
+[2025-01-03 11:21:12,057] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 166392 closing signal SIGTERM
+[2025-01-03 11:21:12,057] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 166393 closing signal SIGTERM
+[2025-01-03 11:21:12,057] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 300995 closing signal SIGTERM
+[2025-01-03 11:21:12,058] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 300996 closing signal SIGTERM
+[2025-01-03 11:21:12,058] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 300997 closing signal SIGTERM
+[2025-01-03 11:21:12,057] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 11:21:12,057] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 166394 closing signal SIGTERM
+[2025-01-03 11:21:12,058] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 300998 closing signal SIGTERM
+[2025-01-03 11:21:12,057] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 166395 closing signal SIGTERM
+[2025-01-03 11:21:12,056] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 11:21:12,058] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 300999 closing signal SIGTERM
+[2025-01-03 11:21:12,057] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 11:21:12,057] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 87660 closing signal SIGTERM
+[2025-01-03 11:21:12,057] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 166396 closing signal SIGTERM
+[2025-01-03 11:21:12,057] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 11:21:12,058] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 301001 closing signal SIGTERM
+[2025-01-03 11:21:12,056] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 11:21:12,057] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 166397 closing signal SIGTERM
+[2025-01-03 11:21:12,056] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 303651 closing signal SIGTERM
+[2025-01-03 11:21:12,058] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 11:21:12,057] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 87661 closing signal SIGTERM
+[2025-01-03 11:21:12,056] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 303652 closing signal SIGTERM
+[2025-01-03 11:21:12,057] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 295709 closing signal SIGTERM
+[2025-01-03 11:21:12,056] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 96287 closing signal SIGTERM
+[2025-01-03 11:21:12,057] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 69207 closing signal SIGTERM
+[2025-01-03 11:21:12,057] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 87662 closing signal SIGTERM
+[2025-01-03 11:21:12,056] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 303653 closing signal SIGTERM
+[2025-01-03 11:21:12,058] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 197675 closing signal SIGTERM
+[2025-01-03 11:21:12,057] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 295710 closing signal SIGTERM
+[2025-01-03 11:21:12,056] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 303655 closing signal SIGTERM
+[2025-01-03 11:21:12,057] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 96288 closing signal SIGTERM
+[2025-01-03 11:21:12,058] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 197676 closing signal SIGTERM
+[2025-01-03 11:21:12,057] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 69208 closing signal SIGTERM
+[2025-01-03 11:21:12,057] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 295711 closing signal SIGTERM
+[2025-01-03 11:21:12,057] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 87663 closing signal SIGTERM
+[2025-01-03 11:21:12,057] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 11:21:12,056] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 303657 closing signal SIGTERM
+[2025-01-03 11:21:12,058] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 197677 closing signal SIGTERM
+[2025-01-03 11:21:12,057] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 69209 closing signal SIGTERM
+[2025-01-03 11:21:12,057] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 69210 closing signal SIGTERM
+[2025-01-03 11:21:12,058] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 197678 closing signal SIGTERM
+[2025-01-03 11:21:12,058] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 295712 closing signal SIGTERM
+[2025-01-03 11:21:12,057] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 96289 closing signal SIGTERM
+[2025-01-03 11:21:12,058] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 98145 closing signal SIGTERM
+[2025-01-03 11:21:12,056] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 303658 closing signal SIGTERM
+[2025-01-03 11:21:12,057] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 69211 closing signal SIGTERM
+[2025-01-03 11:21:12,057] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 69212 closing signal SIGTERM
+[2025-01-03 11:21:12,058] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 295714 closing signal SIGTERM
+[2025-01-03 11:21:12,058] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 197679 closing signal SIGTERM
+[2025-01-03 11:21:12,058] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 197680 closing signal SIGTERM
+[2025-01-03 11:21:12,057] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 87665 closing signal SIGTERM
+[2025-01-03 11:21:12,057] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 69213 closing signal SIGTERM
+[2025-01-03 11:21:12,057] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 96290 closing signal SIGTERM
+[2025-01-03 11:21:12,057] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 69214 closing signal SIGTERM
+[2025-01-03 11:21:12,058] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 295715 closing signal SIGTERM
+[2025-01-03 11:21:12,058] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 98146 closing signal SIGTERM
+[2025-01-03 11:21:12,058] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 197682 closing signal SIGTERM
+[2025-01-03 11:21:12,058] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 87666 closing signal SIGTERM
+[2025-01-03 11:21:12,057] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 11:21:12,057] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 96291 closing signal SIGTERM
+[2025-01-03 11:21:12,058] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 295716 closing signal SIGTERM
+[2025-01-03 11:21:12,057] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 11:21:12,057] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 11:21:12,058] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 98147 closing signal SIGTERM
+[2025-01-03 11:21:12,058] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 98148 closing signal SIGTERM
+[2025-01-03 11:21:12,058] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 98149 closing signal SIGTERM
+[2025-01-03 11:21:12,057] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 96292 closing signal SIGTERM
+[2025-01-03 11:21:12,058] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 73127 closing signal SIGTERM
+[2025-01-03 11:21:12,057] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 96293 closing signal SIGTERM
+[2025-01-03 11:21:12,057] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 177775 closing signal SIGTERM
+[2025-01-03 11:21:12,058] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 98150 closing signal SIGTERM
+[2025-01-03 11:21:12,057] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 166706 closing signal SIGTERM
+[2025-01-03 11:21:12,058] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 98151 closing signal SIGTERM
+[2025-01-03 11:21:12,058] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 73128 closing signal SIGTERM
+[2025-01-03 11:21:12,057] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 96294 closing signal SIGTERM
+[2025-01-03 11:21:12,058] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 73129 closing signal SIGTERM
+[2025-01-03 11:21:12,057] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 166707 closing signal SIGTERM
+[2025-01-03 11:21:12,057] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 177776 closing signal SIGTERM
+[2025-01-03 11:21:12,058] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 73130 closing signal SIGTERM
+[2025-01-03 11:21:12,058] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 73131 closing signal SIGTERM
+[2025-01-03 11:21:12,057] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 177777 closing signal SIGTERM
+[2025-01-03 11:21:12,057] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 166708 closing signal SIGTERM
+[2025-01-03 11:21:12,058] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 73132 closing signal SIGTERM
+[2025-01-03 11:21:12,058] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 73133 closing signal SIGTERM
+[2025-01-03 11:21:12,057] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 177778 closing signal SIGTERM
+[2025-01-03 11:21:12,057] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 166709 closing signal SIGTERM
+[2025-01-03 11:21:12,057] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 166710 closing signal SIGTERM
+[2025-01-03 11:21:12,058] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 73134 closing signal SIGTERM
+[2025-01-03 11:21:12,057] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 177779 closing signal SIGTERM
+[2025-01-03 11:21:12,057] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 166711 closing signal SIGTERM
+[2025-01-03 11:21:12,057] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 166712 closing signal SIGTERM
+[2025-01-03 11:21:12,057] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 177780 closing signal SIGTERM
+[2025-01-03 11:21:12,057] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 166713 closing signal SIGTERM
+[2025-01-03 11:21:12,057] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 177781 closing signal SIGTERM
+[2025-01-03 11:21:12,059] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-03 11:21:12,059] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 175322 closing signal SIGTERM
+[2025-01-03 11:21:12,059] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 175323 closing signal SIGTERM
+[2025-01-03 11:21:12,059] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 175324 closing signal SIGTERM
+[2025-01-03 11:21:12,059] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 175325 closing signal SIGTERM
+[2025-01-03 11:21:12,059] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 175326 closing signal SIGTERM
+[2025-01-03 11:21:12,059] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 175327 closing signal SIGTERM
+[2025-01-03 11:21:12,059] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 175328 closing signal SIGTERM
+[2025-01-03 11:21:12,059] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 175329 closing signal SIGTERM
+srun: error: ip-26-0-170-31: task 13: Terminated
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 878, in _invoke_run
+    run_result = self._monitor_workers(self._worker_group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/local_elastic_agent.py", line 307, in _monitor_workers
+    result = self._pcontext.wait(0)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 288, in wait
+    return self._poll()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 668, in _poll
+    self.close()  # terminate all running procs
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 331, in close
+    self._close(death_sig=death_sig, timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 713, in _close
+    handler.proc.wait(time_to_wait)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1209, in wait
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    return self._wait(timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1953, in _wait
+    time.sleep(delay)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 177699 got signal: 15
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 878, in _invoke_run
+    run_result = self._monitor_workers(self._worker_group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/local_elastic_agent.py", line 307, in _monitor_workers
+    result = self._pcontext.wait(0)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 288, in wait
+    return self._poll()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 668, in _poll
+    self.close()  # terminate all running procs
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 331, in close
+    self._close(death_sig=death_sig, timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 713, in _close
+    handler.proc.wait(time_to_wait)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1209, in wait
+    return self._wait(timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1953, in _wait
+    time.sleep(delay)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 61841 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 878, in _invoke_run
+    run_result = self._monitor_workers(self._worker_group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/local_elastic_agent.py", line 307, in _monitor_workers
+    result = self._pcontext.wait(0)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 288, in wait
+    return self._poll()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 668, in _poll
+    self.close()  # terminate all running procs
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 331, in close
+    self._close(death_sig=death_sig, timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 713, in _close
+    handler.proc.wait(time_to_wait)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1209, in wait
+    return self._wait(timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1953, in _wait
+    time.sleep(delay)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 87585 got signal: 15
+srun: error: ip-26-0-170-143: task 15: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 878, in _invoke_run
+    run_result = self._monitor_workers(self._worker_group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/local_elastic_agent.py", line 307, in _monitor_workers
+    result = self._pcontext.wait(0)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 288, in wait
+    return self._poll()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 668, in _poll
+    self.close()  # terminate all running procs
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 331, in close
+    self._close(death_sig=death_sig, timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 713, in _close
+    handler.proc.wait(time_to_wait)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1209, in wait
+    return self._wait(timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1953, in _wait
+    time.sleep(delay)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 303574 got signal: 15
+srun: error: ip-26-0-168-52: task 3: Exited with exit code 1
+srun: error: ip-26-0-160-103: task 1: Exited with exit code 1
+[2025-01-03 11:21:15,566] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-169-86.ec2.internal_175246_0' has failed to send a keep-alive heartbeat to the rendezvous '14023604' due to an error of type RendezvousConnectionError.
+srun: error: ip-26-0-169-207: task 10: Exited with exit code 1
+[2025-01-03 11:21:16,267] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-170-132.ec2.internal_197587_0' has failed to send a keep-alive heartbeat to the rendezvous '14023604' due to an error of type RendezvousConnectionError.
+[2025-01-03 11:21:16,358] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-168-95.ec2.internal_98070_0' has failed to send a keep-alive heartbeat to the rendezvous '14023604' due to an error of type RendezvousConnectionError.
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 878, in _invoke_run
+    run_result = self._monitor_workers(self._worker_group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/local_elastic_agent.py", line 307, in _monitor_workers
+    result = self._pcontext.wait(0)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 288, in wait
+    return self._poll()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 668, in _poll
+    self.close()  # terminate all running procs
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 331, in close
+    self._close(death_sig=death_sig, timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 713, in _close
+    handler.proc.wait(time_to_wait)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1209, in wait
+    return self._wait(timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1953, in _wait
+    time.sleep(delay)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 300917 got signal: 15
+terminate called without an active exception
+Fatal Python error: Aborted
+
+Thread 0x00007f1e71e36000 (most recent call first):
+  <no Python frame>
+
+Extension modules: numpy.core._multiarray_umath, numpy.core._multiarray_tests, numpy.linalg._umath_linalg, numpy.fft._pocketfft_internal, numpy.random._common, numpy.random.bit_generator, numpy.random._bounded_integers, numpy.random._mt19937, numpy.random.mtrand, numpy.random._philox, numpy.random._pcg64, numpy.random._sfc64, numpy.random._generator, torch._C, torch._C._fft, torch._C._linalg, torch._C._nested, torch._C._nn, torch._C._sparse, torch._C._special, gmpy2.gmpy2 (total: 21)
+[2025-01-03 11:21:16,448] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-169-139.ec2.internal_69130_0' has failed to send a keep-alive heartbeat to the rendezvous '14023604' due to an error of type RendezvousConnectionError.
+[2025-01-03 11:21:16,450] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-168-238.ec2.internal_96210_0' has failed to send a keep-alive heartbeat to the rendezvous '14023604' due to an error of type RendezvousConnectionError.
+[2025-01-03 11:21:16,477] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-169-247.ec2.internal_73051_0' has failed to send a keep-alive heartbeat to the rendezvous '14023604' due to an error of type RendezvousConnectionError.
+[2025-01-03 11:21:16,490] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-168-34.ec2.internal_166314_0' has failed to send a keep-alive heartbeat to the rendezvous '14023604' due to an error of type RendezvousConnectionError.
+[2025-01-03 11:21:16,493] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-169-239.ec2.internal_295634_0' has failed to send a keep-alive heartbeat to the rendezvous '14023604' due to an error of type RendezvousConnectionError.
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 295634 got signal: 15
+srun: error: ip-26-0-168-120: task 5: Aborted (core dumped)
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 878, in _invoke_run
+    run_result = self._monitor_workers(self._worker_group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/local_elastic_agent.py", line 307, in _monitor_workers
+    result = self._pcontext.wait(0)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 288, in wait
+    return self._poll()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 668, in _poll
+    self.close()  # terminate all running procs
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 331, in close
+    self._close(death_sig=death_sig, timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 713, in _close
+    handler.proc.wait(time_to_wait)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1209, in wait
+    return self._wait(timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1953, in _wait
+    time.sleep(delay)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 197587 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 98070 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 175246 got signal: 15
+[2025-01-03 11:21:16,853] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-169-132.ec2.internal_166623_0' has failed to send a keep-alive heartbeat to the rendezvous '14023604' due to an error of type RendezvousConnectionError.
+srun: error: ip-26-0-169-239: task 11: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 166623 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 73051 got signal: 15
+srun: error: ip-26-0-170-132: task 14: Exited with exit code 1
+srun: error: ip-26-0-168-95: task 4: Exited with exit code 1
+srun: error: ip-26-0-169-86: task 7: Exited with exit code 1
+srun: error: ip-26-0-169-132: task 8: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 69130 got signal: 15
+srun: error: ip-26-0-169-247: task 12: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 166314 got signal: 15
+srun: error: ip-26-0-169-139: task 9: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 96210 got signal: 15
+srun: error: ip-26-0-168-34: task 2: Exited with exit code 1
+srun: error: ip-26-0-168-238: task 6: Exited with exit code 1
+srun: Force Terminated StepId=14023604.0
diff --git a/logs/14098592-bench_1.34G_dp2_tp1_pp8_acc64_mbs2_seq4096_zero1_tpmodeRED_vocab131k.out b/logs/14098592-bench_1.34G_dp2_tp1_pp8_acc64_mbs2_seq4096_zero1_tpmodeRED_vocab131k.out
new file mode 100644
index 0000000000000000000000000000000000000000..a78b4f6e1563ae406f36ee37d3c7ce3c78a1d545
--- /dev/null
+++ b/logs/14098592-bench_1.34G_dp2_tp1_pp8_acc64_mbs2_seq4096_zero1_tpmodeRED_vocab131k.out
@@ -0,0 +1,10472 @@
++ '[' -z 14098592 ']'
++ source /etc/profile.d/modules.sh
+++ . /usr/share/modules/init/bash
++++ unset _mlshdbg
++++ '[' 0 = 1 ']'
++++ unset _mlre _mlIFS
++++ '[' -n x ']'
++++ _mlIFS=' 	
+'
++++ IFS=' '
++++ '[' -n '' ']'
+++++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash autoinit
++++ _mlcode='module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+test 0;'
++++ _mlret=0
++++ '[' -n x ']'
++++ IFS=' 	
+'
++++ unset _mlIFS
++++ unset _mlre _mlv _mlrv
++++ '[' 0 -eq 0 ']'
++++ eval 'module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+test 0;'
+++++ MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl
+++++ export MODULES_CMD
+++++ ENV=/usr/share/modules/init/profile.sh
+++++ export ENV
+++++ BASH_ENV=/usr/share/modules/init/bash
+++++ export BASH_ENV
+++++ MODULESHOME=/usr/share/modules
+++++ export MODULESHOME
+++++ test 0
++++ '[' 0 = 1 ']'
++++ '[' -t 2 ']'
++++ export -f module
++++ export -f switchml
++++ '[' 5 -ge 3 ']'
++++ [[ ehxB =~ i ]]
++++ [[ ! :/admin/home/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/admin/home/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin: =~ :/usr/bin: ]]
++++ '[' '!' -n x ']'
+++++ manpath
++++ [[ ! :/admin/home/nouamane/miniconda/envs/2-1-cu121/man:/admin/home/nouamane/miniconda/envs/2-1-cu121/share/man:/fsx/nouamane/miniconda/man:/fsx/nouamane/miniconda/share/man:/opt/slurm/share/man:/opt/amazon/openmpi/share/man:/opt/amazon/efa/share/man:/usr/local/man:/usr/local/share/man:/usr/share/man:/admin/home/nouamane/.local/share/man:/admin/home/nouamane/.fzf/man:: =~ :/usr/share/man: ]]
++++ unset _mlcode _mlret
++++ '[' -n '' ']'
++ module load cuda/12.1
++ unset _mlshdbg
++ '[' 0 = 1 ']'
++ unset _mlre _mlIFS
++ '[' -n x ']'
++ _mlIFS=' 	
+'
++ IFS=' '
++ '[' -n '' ']'
+++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash load cuda/12.1
++ eval 'CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include;' export 'CPATH;
+LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/opt/aws-ofi-nccl/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/targets/x86_64-linux/lib/:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/lib:/usr/lib;' export 'LD_LIBRARY_PATH;
+FI_EFA_FORK_SAFE=1;' export 'FI_EFA_FORK_SAFE;
+MANPATH=/usr/local/cuda-12.1/share/man::;' export 'MANPATH;
+LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64;' export 'LIBRARY_PATH;
+_LMFILES_=/usr/share/modules/modulefiles/cuda/12.1;' export '_LMFILES_;
+LOADEDMODULES=cuda/12.1;' export 'LOADEDMODULES;
+MPI_PATH=/opt/amazon/openmpi;' export 'MPI_PATH;
+NCCL_HOME_modshare=/opt/nccl/build:1;' export 'NCCL_HOME_modshare;
+NCCL_PROTO=simple;' export 'NCCL_PROTO;
+MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1;' export 'MANPATH_modshare;
+LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1;' export 'LIBRARY_PATH_modshare;
+NCCL_SOCKET_IFNAME=enp;' export 'NCCL_SOCKET_IFNAME;
+AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl;' export 'AWS_OFI_NCCL_HOME;
+NCCL_HOME=/opt/nccl/build;' export 'NCCL_HOME;
+FI_PROVIDER=efa;' export 'FI_PROVIDER;
+AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1;' export 'AWS_OFI_NCCL_HOME_modshare;
+CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1;' export 'CPATH_modshare;
+LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib/:1:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2:/opt/aws-ofi-nccl/lib:2:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/efa/lib:1:/usr/lib:1;' export 'LD_LIBRARY_PATH_modshare;
+FI_EFA_ENABLE_SHM_TRANSFER=1;' export 'FI_EFA_ENABLE_SHM_TRANSFER;
+_LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1;' export '_LMFILES__modshare;
+LOADEDMODULES_modshare=cuda/12.1:1;' export 'LOADEDMODULES_modshare;
+MPI_PATH_modshare=/opt/amazon/openmpi:1;' export 'MPI_PATH_modshare;
+PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/admin/home/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin;' export 'PATH;
+CUDA_HOME=/usr/local/cuda-12.1;' export 'CUDA_HOME;
+PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/local/cuda-12.1/include:1:/usr/bin:1:/opt/amazon/efa/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/opt/slurm/bin:1:/usr/sbin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:2:/opt/amazon/openmpi/bin:1:/admin/home/nouamane/.fzf/bin:1:/admin/home/nouamane/miniconda/envs/2-1-cu121/bin:1:/usr/local/games:1;' export 'PATH_modshare;
+test' '0;'
+++ CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include
+++ export CPATH
+++ LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/opt/aws-ofi-nccl/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/targets/x86_64-linux/lib/:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/lib:/usr/lib
+++ export LD_LIBRARY_PATH
+++ FI_EFA_FORK_SAFE=1
+++ export FI_EFA_FORK_SAFE
+++ MANPATH=/usr/local/cuda-12.1/share/man::
+++ export MANPATH
+++ LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64
+++ export LIBRARY_PATH
+++ _LMFILES_=/usr/share/modules/modulefiles/cuda/12.1
+++ export _LMFILES_
+++ LOADEDMODULES=cuda/12.1
+++ export LOADEDMODULES
+++ MPI_PATH=/opt/amazon/openmpi
+++ export MPI_PATH
+++ NCCL_HOME_modshare=/opt/nccl/build:1
+++ export NCCL_HOME_modshare
+++ NCCL_PROTO=simple
+++ export NCCL_PROTO
+++ MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1
+++ export MANPATH_modshare
+++ LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1
+++ export LIBRARY_PATH_modshare
+++ NCCL_SOCKET_IFNAME=enp
+++ export NCCL_SOCKET_IFNAME
+++ AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl
+++ export AWS_OFI_NCCL_HOME
+++ NCCL_HOME=/opt/nccl/build
+++ export NCCL_HOME
+++ FI_PROVIDER=efa
+++ export FI_PROVIDER
+++ AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1
+++ export AWS_OFI_NCCL_HOME_modshare
+++ CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1
+++ export CPATH_modshare
+++ LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib/:1:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2:/opt/aws-ofi-nccl/lib:2:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/efa/lib:1:/usr/lib:1
+++ export LD_LIBRARY_PATH_modshare
+++ FI_EFA_ENABLE_SHM_TRANSFER=1
+++ export FI_EFA_ENABLE_SHM_TRANSFER
+++ _LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1
+++ export _LMFILES__modshare
+++ LOADEDMODULES_modshare=cuda/12.1:1
+++ export LOADEDMODULES_modshare
+++ MPI_PATH_modshare=/opt/amazon/openmpi:1
+++ export MPI_PATH_modshare
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/admin/home/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
+++ export PATH
+++ CUDA_HOME=/usr/local/cuda-12.1
+++ export CUDA_HOME
+++ PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/local/cuda-12.1/include:1:/usr/bin:1:/opt/amazon/efa/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/opt/slurm/bin:1:/usr/sbin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:2:/opt/amazon/openmpi/bin:1:/admin/home/nouamane/.fzf/bin:1:/admin/home/nouamane/miniconda/envs/2-1-cu121/bin:1:/usr/local/games:1
+++ export PATH_modshare
+++ test 0
++ _mlstatus=0
++ '[' -n x ']'
++ IFS=' 	
+'
++ unset _mlre _mlv _mlrv _mlIFS
++ '[' -n '' ']'
++ unset _mlshdbg
++ return 0
++ source /fsx/nouamane/miniconda/bin/activate
+++ _CONDA_ROOT=/fsx/nouamane/miniconda
+++ . /fsx/nouamane/miniconda/etc/profile.d/conda.sh
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ '[' -z x ']'
+++ conda activate
+++ local cmd=activate
+++ case "$cmd" in
+++ __conda_activate activate
+++ '[' -n '' ']'
+++ local ask_conda
++++ PS1=
++++ __conda_exe shell.posix activate
++++ /fsx/nouamane/miniconda/bin/conda shell.posix activate
+Error while loading conda entry point: conda-libmamba-solver (module 'libmambapy' has no attribute 'QueryFormat')
+++ ask_conda='. "/admin/home/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/fsx/nouamane/miniconda/bin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/admin/home/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_2='\''/admin/home/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/etc/conda/activate.d/libxml2_activate.sh"'
+++ eval '. "/admin/home/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/fsx/nouamane/miniconda/bin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/admin/home/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_2='\''/admin/home/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/etc/conda/activate.d/libxml2_activate.sh"'
++++ . /admin/home/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh
+++++ test -n ''
+++++ unset XML_CATALOG_FILES
+++++ unset xml_catalog_files_libxml2
++++ PS1='(base) '
++++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/fsx/nouamane/miniconda/bin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/admin/home/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
++++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/fsx/nouamane/miniconda/bin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/admin/home/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
++++ export CONDA_PREFIX=/fsx/nouamane/miniconda
++++ CONDA_PREFIX=/fsx/nouamane/miniconda
++++ export CONDA_SHLVL=3
++++ CONDA_SHLVL=3
++++ export CONDA_DEFAULT_ENV=base
++++ CONDA_DEFAULT_ENV=base
++++ export 'CONDA_PROMPT_MODIFIER=(base) '
++++ CONDA_PROMPT_MODIFIER='(base) '
++++ export CONDA_PREFIX_2=/admin/home/nouamane/miniconda/envs/2-1-cu121
++++ CONDA_PREFIX_2=/admin/home/nouamane/miniconda/envs/2-1-cu121
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ . /fsx/nouamane/miniconda/etc/conda/activate.d/libxml2_activate.sh
+++++ test -n ''
+++++ xml_catalog_files_libxml2=
+++++ XML_CATALOG_FILES=
+++++ conda_catalog_files=
+++++ ifs_libxml2=' 	
+'
+++++ IFS=' '
+++++ rem=/fsx/nouamane/miniconda
+++++ for pre in ${rem}
+++++ test '' = /fsx/nouamane/miniconda
+++++ conda_catalog_files=/fsx/nouamane/miniconda
+++++ rem=
+++++ IFS=' 	
+'
+++++ conda_catalog_files='file:///fsx/nouamane/miniconda/etc/xml/catalog file:///etc/xml/catalog'
+++++ export 'XML_CATALOG_FILES=file:///fsx/nouamane/miniconda/etc/xml/catalog file:///etc/xml/catalog'
+++++ XML_CATALOG_FILES='file:///fsx/nouamane/miniconda/etc/xml/catalog file:///etc/xml/catalog'
+++++ unset conda_catalog_files ifs_libxml2 rem
+++ __conda_hashr
+++ '[' -n '' ']'
+++ '[' -n '' ']'
+++ hash -r
++ conda activate 2-1-cu121
++ local cmd=activate
++ case "$cmd" in
++ __conda_activate activate 2-1-cu121
++ '[' -n '' ']'
++ local ask_conda
+++ PS1='(base) '
+++ __conda_exe shell.posix activate 2-1-cu121
+++ /fsx/nouamane/miniconda/bin/conda shell.posix activate 2-1-cu121
+Error while loading conda entry point: conda-libmamba-solver (module 'libmambapy' has no attribute 'QueryFormat')
++ ask_conda='. "/fsx/nouamane/miniconda/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/admin/home/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''4'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_3='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
++ eval '. "/fsx/nouamane/miniconda/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/admin/home/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''4'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_3='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
+++ . /fsx/nouamane/miniconda/etc/conda/deactivate.d/libxml2_deactivate.sh
++++ test -n ''
++++ unset XML_CATALOG_FILES
++++ unset xml_catalog_files_libxml2
+++ PS1='(2-1-cu121) '
+++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/admin/home/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/admin/home/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
+++ export CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ export CONDA_SHLVL=4
+++ CONDA_SHLVL=4
+++ export CONDA_DEFAULT_ENV=2-1-cu121
+++ CONDA_DEFAULT_ENV=2-1-cu121
+++ export 'CONDA_PROMPT_MODIFIER=(2-1-cu121) '
+++ CONDA_PROMPT_MODIFIER='(2-1-cu121) '
+++ export CONDA_PREFIX_3=/fsx/nouamane/miniconda
+++ CONDA_PREFIX_3=/fsx/nouamane/miniconda
+++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ export _CE_M=
+++ _CE_M=
+++ export _CE_CONDA=
+++ _CE_CONDA=
+++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh
++++ test -n ''
++++ xml_catalog_files_libxml2=
++++ XML_CATALOG_FILES=
++++ conda_catalog_files=
++++ ifs_libxml2=' 	
+'
++++ IFS=' '
++++ rem=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ for pre in ${rem}
++++ test '' = /fsx/nouamane/miniconda/envs/2-1-cu121
++++ conda_catalog_files=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ rem=
++++ IFS=' 	
+'
++++ conda_catalog_files='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ export 'XML_CATALOG_FILES=file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ XML_CATALOG_FILES='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ unset conda_catalog_files ifs_libxml2 rem
++ __conda_hashr
++ '[' -n '' ']'
++ '[' -n '' ']'
++ hash -r
++ export PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/admin/home/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
++ PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/admin/home/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
++ '[' -z '' ']'
+++ scontrol show hostnames ip-26-0-172-142,ip-26-0-173-121
++ export 'NODELIST=ip-26-0-172-142
+ip-26-0-173-121'
++ NODELIST='ip-26-0-172-142
+ip-26-0-173-121'
+++ scontrol show hostnames ip-26-0-172-142,ip-26-0-173-121
+++ head -n1
++ export MASTER_NODE=ip-26-0-172-142
++ MASTER_NODE=ip-26-0-172-142
++ export MASTER_PORT=12356
++ MASTER_PORT=12356
++ export NNODES=2
++ NNODES=2
++ export GPUS_PER_NODE=8
++ GPUS_PER_NODE=8
++ export WORLD_SIZE=16
++ WORLD_SIZE=16
++ export CUDA_DEVICE_MAX_CONNECTIONS=1
++ CUDA_DEVICE_MAX_CONNECTIONS=1
++ export NCCL_DEBUG=INFO
++ NCCL_DEBUG=INFO
++ export NANOTRON_BENCHMARK=1
++ NANOTRON_BENCHMARK=1
++ export WANDB_MODE=disabled
++ WANDB_MODE=disabled
++ export TORCH_NCCL_ASYNC_ERROR_HANDLING=1
++ TORCH_NCCL_ASYNC_ERROR_HANDLING=1
++ export TORCH_DISTRIBUTED_DEBUG=DETAIL
++ TORCH_DISTRIBUTED_DEBUG=DETAIL
++ '[' -z '' ']'
++ echo '=== GPU Topology ==='
+=== GPU Topology ===
++ nvidia-smi topo -m
+	GPU0	GPU1	GPU2	GPU3	GPU4	GPU5	GPU6	GPU7	CPU Affinity	NUMA Affinity	GPU NUMA ID
+GPU0	 X 	NV18	NV18	NV18	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU1	NV18	 X 	NV18	NV18	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU2	NV18	NV18	 X 	NV18	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU3	NV18	NV18	NV18	 X 	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU4	NV18	NV18	NV18	NV18	 X 	NV18	NV18	NV18	48-95	1		N/A
+GPU5	NV18	NV18	NV18	NV18	NV18	 X 	NV18	NV18	48-95	1		N/A
+GPU6	NV18	NV18	NV18	NV18	NV18	NV18	 X 	NV18	48-95	1		N/A
+GPU7	NV18	NV18	NV18	NV18	NV18	NV18	NV18	 X 	48-95	1		N/A
+
+Legend:
+
+  X    = Self
+  SYS  = Connection traversing PCIe as well as the SMP interconnect between NUMA nodes (e.g., QPI/UPI)
+  NODE = Connection traversing PCIe as well as the interconnect between PCIe Host Bridges within a NUMA node
+  PHB  = Connection traversing PCIe as well as a PCIe Host Bridge (typically the CPU)
+  PXB  = Connection traversing multiple PCIe bridges (without traversing the PCIe Host Bridge)
+  PIX  = Connection traversing at most a single PCIe bridge
+  NV#  = Connection traversing a bonded set of # NVLinks
++ echo ==================
+==================
++ export SRUN_ALLOC_ARGS=
++ SRUN_ALLOC_ARGS=
++ echo 'Master node: ip-26-0-172-142'
+Master node: ip-26-0-172-142
++ echo 'All nodes: ip-26-0-172-142
+ip-26-0-173-121'
+All nodes: ip-26-0-172-142
+ip-26-0-173-121
++ echo 'World size: 16'
+World size: 16
++ '[' -n '' ']'
++ srun --wait=0 --kill-on-bad-exit=1 torchrun --nnodes=2 --nproc_per_node=8 --rdzv_id=14098592 --rdzv_backend=c10d --rdzv_endpoint=ip-26-0-172-142:12356 --max_restarts 0 --rdzv_conf timeout=60 /fsx/nouamane/projects/nanotron/run_train.py --config-file benchmark/configs/config_1.34G_dp2_tp1_pp8_acc64_mbs2_seq4096_zero1_tpmodeRED_vocab131k.yaml
+[2025-01-07 00:36:23,601] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-07 00:36:23,601] torch.distributed.run: [WARNING] 
+[2025-01-07 00:36:23,601] torch.distributed.run: [WARNING] *****************************************
+[2025-01-07 00:36:23,601] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-07 00:36:23,601] torch.distributed.run: [WARNING] *****************************************
+[2025-01-07 00:36:24,050] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-07 00:36:24,050] torch.distributed.run: [WARNING] 
+[2025-01-07 00:36:24,050] torch.distributed.run: [WARNING] *****************************************
+[2025-01-07 00:36:24,050] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-07 00:36:24,050] torch.distributed.run: [WARNING] *****************************************
+ip-26-0-172-142:789255:789255 [0] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-172-142:789255:789255 [0] NCCL INFO Bootstrap : Using enp71s0:26.0.172.142<0>
+ip-26-0-172-142:789255:789255 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-172-142:789255:789255 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-172-142:789255:789255 [0] NCCL INFO cudaDriverVersion 12020
+ip-26-0-172-142:789262:789262 [7] NCCL INFO cudaDriverVersion 12020
+NCCL version 2.18.6+cuda12.1
+ip-26-0-172-142:789256:789256 [1] NCCL INFO cudaDriverVersion 12020
+ip-26-0-172-142:789260:789260 [5] NCCL INFO cudaDriverVersion 12020
+ip-26-0-172-142:789257:789257 [2] NCCL INFO cudaDriverVersion 12020
+ip-26-0-172-142:789261:789261 [6] NCCL INFO cudaDriverVersion 12020
+ip-26-0-172-142:789259:789259 [4] NCCL INFO cudaDriverVersion 12020
+ip-26-0-172-142:789258:789258 [3] NCCL INFO cudaDriverVersion 12020
+ip-26-0-172-142:789262:789262 [7] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-172-142:789256:789256 [1] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-172-142:789260:789260 [5] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-172-142:789257:789257 [2] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-173-121:3987798:3987798 [4] NCCL INFO cudaDriverVersion 12020
+ip-26-0-173-121:3987800:3987800 [6] NCCL INFO cudaDriverVersion 12020
+ip-26-0-173-121:3987797:3987797 [3] NCCL INFO cudaDriverVersion 12020
+ip-26-0-173-121:3987794:3987794 [0] NCCL INFO cudaDriverVersion 12020
+ip-26-0-173-121:3987796:3987796 [2] NCCL INFO cudaDriverVersion 12020
+ip-26-0-173-121:3987795:3987795 [1] NCCL INFO cudaDriverVersion 12020
+ip-26-0-172-142:789261:789261 [6] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-172-142:789259:789259 [4] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-173-121:3987799:3987799 [5] NCCL INFO cudaDriverVersion 12020
+ip-26-0-173-121:3987801:3987801 [7] NCCL INFO cudaDriverVersion 12020
+ip-26-0-172-142:789258:789258 [3] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-173-121:3987795:3987795 [1] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-173-121:3987799:3987799 [5] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-173-121:3987794:3987794 [0] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-173-121:3987798:3987798 [4] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-173-121:3987797:3987797 [3] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-173-121:3987796:3987796 [2] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-173-121:3987801:3987801 [7] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-173-121:3987800:3987800 [6] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-172-142:789262:789262 [7] NCCL INFO Bootstrap : Using enp71s0:26.0.172.142<0>
+ip-26-0-172-142:789257:789257 [2] NCCL INFO Bootstrap : Using enp71s0:26.0.172.142<0>
+ip-26-0-172-142:789260:789260 [5] NCCL INFO Bootstrap : Using enp71s0:26.0.172.142<0>
+ip-26-0-172-142:789256:789256 [1] NCCL INFO Bootstrap : Using enp71s0:26.0.172.142<0>
+ip-26-0-172-142:789259:789259 [4] NCCL INFO Bootstrap : Using enp71s0:26.0.172.142<0>
+ip-26-0-172-142:789261:789261 [6] NCCL INFO Bootstrap : Using enp71s0:26.0.172.142<0>
+ip-26-0-172-142:789258:789258 [3] NCCL INFO Bootstrap : Using enp71s0:26.0.172.142<0>
+ip-26-0-172-142:789261:789261 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-172-142:789262:789262 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-172-142:789260:789260 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-172-142:789256:789256 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-172-142:789257:789257 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-172-142:789261:789261 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-172-142:789262:789262 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-172-142:789260:789260 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-172-142:789256:789256 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-172-142:789257:789257 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-172-142:789259:789259 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-172-142:789259:789259 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-172-142:789258:789258 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-172-142:789258:789258 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-173-121:3987795:3987795 [1] NCCL INFO Bootstrap : Using enp71s0:26.0.173.121<0>
+ip-26-0-173-121:3987794:3987794 [0] NCCL INFO Bootstrap : Using enp71s0:26.0.173.121<0>
+ip-26-0-173-121:3987799:3987799 [5] NCCL INFO Bootstrap : Using enp71s0:26.0.173.121<0>
+ip-26-0-173-121:3987796:3987796 [2] NCCL INFO Bootstrap : Using enp71s0:26.0.173.121<0>
+ip-26-0-173-121:3987800:3987800 [6] NCCL INFO Bootstrap : Using enp71s0:26.0.173.121<0>
+ip-26-0-173-121:3987797:3987797 [3] NCCL INFO Bootstrap : Using enp71s0:26.0.173.121<0>
+ip-26-0-173-121:3987801:3987801 [7] NCCL INFO Bootstrap : Using enp71s0:26.0.173.121<0>
+ip-26-0-173-121:3987798:3987798 [4] NCCL INFO Bootstrap : Using enp71s0:26.0.173.121<0>
+ip-26-0-173-121:3987795:3987795 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-173-121:3987795:3987795 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-173-121:3987796:3987796 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-173-121:3987797:3987797 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-173-121:3987800:3987800 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-173-121:3987796:3987796 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-173-121:3987800:3987800 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-173-121:3987797:3987797 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-173-121:3987798:3987798 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-173-121:3987799:3987799 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-173-121:3987798:3987798 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-173-121:3987799:3987799 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-173-121:3987801:3987801 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-173-121:3987801:3987801 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-173-121:3987794:3987794 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-173-121:3987794:3987794 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-172-142:789255:789416 [0] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-172-142:789255:789416 [0] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-172-142:789255:789416 [0] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-172-142:789255:789416 [0] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-172-142:789255:789416 [0] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-172-142:789255:789416 [0] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-172-142:789255:789416 [0] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-172-142:789255:789416 [0] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-172-142:789255:789416 [0] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-172-142:789259:789417 [4] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-172-142:789259:789417 [4] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-172-142:789262:789418 [7] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-172-142:789262:789418 [7] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-172-142:789255:789416 [0] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-172-142:789260:789419 [5] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-172-142:789260:789419 [5] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-172-142:789261:789423 [6] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-172-142:789261:789423 [6] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-172-142:789257:789422 [2] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-172-142:789257:789422 [2] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-173-121:3987799:3987958 [5] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-173-121:3987799:3987958 [5] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-172-142:789259:789417 [4] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-172-142:789259:789417 [4] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-172-142:789259:789417 [4] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-172-142:789259:789417 [4] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-172-142:789259:789417 [4] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-172-142:789259:789417 [4] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-172-142:789259:789417 [4] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-172-142:789262:789418 [7] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-172-142:789262:789418 [7] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-172-142:789262:789418 [7] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-172-142:789262:789418 [7] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-172-142:789262:789418 [7] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-172-142:789262:789418 [7] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-172-142:789262:789418 [7] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-172-142:789258:789420 [3] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-172-142:789258:789420 [3] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-173-121:3987799:3987958 [5] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-173-121:3987799:3987958 [5] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-173-121:3987799:3987958 [5] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-173-121:3987799:3987958 [5] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-173-121:3987799:3987958 [5] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-173-121:3987799:3987958 [5] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-173-121:3987799:3987958 [5] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-172-142:789261:789423 [6] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-172-142:789261:789423 [6] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-172-142:789261:789423 [6] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-172-142:789261:789423 [6] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-172-142:789261:789423 [6] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-172-142:789261:789423 [6] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-172-142:789261:789423 [6] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-172-142:789260:789419 [5] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-172-142:789260:789419 [5] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-172-142:789260:789419 [5] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-172-142:789260:789419 [5] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-172-142:789260:789419 [5] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-172-142:789260:789419 [5] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-172-142:789260:789419 [5] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-172-142:789256:789421 [1] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-172-142:789256:789421 [1] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-172-142:789257:789422 [2] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-172-142:789257:789422 [2] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-172-142:789257:789422 [2] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-172-142:789257:789422 [2] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-172-142:789257:789422 [2] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-172-142:789257:789422 [2] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-172-142:789257:789422 [2] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-172-142:789258:789420 [3] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-172-142:789258:789420 [3] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-172-142:789258:789420 [3] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-172-142:789258:789420 [3] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-172-142:789258:789420 [3] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-172-142:789258:789420 [3] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-172-142:789258:789420 [3] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-173-121:3987797:3987954 [3] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-173-121:3987797:3987954 [3] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-173-121:3987796:3987957 [2] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-173-121:3987796:3987957 [2] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-172-142:789256:789421 [1] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-172-142:789256:789421 [1] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-172-142:789256:789421 [1] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-172-142:789256:789421 [1] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-172-142:789256:789421 [1] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-172-142:789256:789421 [1] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-172-142:789256:789421 [1] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-173-121:3987795:3987961 [1] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-173-121:3987795:3987961 [1] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-173-121:3987798:3987955 [4] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-173-121:3987798:3987955 [4] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-173-121:3987794:3987956 [0] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-173-121:3987794:3987956 [0] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-173-121:3987797:3987954 [3] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-173-121:3987797:3987954 [3] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-173-121:3987797:3987954 [3] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-173-121:3987797:3987954 [3] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-173-121:3987797:3987954 [3] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-173-121:3987797:3987954 [3] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-173-121:3987797:3987954 [3] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-173-121:3987796:3987957 [2] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-173-121:3987796:3987957 [2] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-173-121:3987796:3987957 [2] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-173-121:3987796:3987957 [2] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-173-121:3987796:3987957 [2] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-173-121:3987796:3987957 [2] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-173-121:3987796:3987957 [2] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-173-121:3987801:3987960 [7] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-173-121:3987801:3987960 [7] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-173-121:3987795:3987961 [1] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-173-121:3987795:3987961 [1] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-173-121:3987795:3987961 [1] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-173-121:3987795:3987961 [1] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-173-121:3987795:3987961 [1] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-173-121:3987795:3987961 [1] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-173-121:3987795:3987961 [1] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-173-121:3987798:3987955 [4] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-173-121:3987798:3987955 [4] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-173-121:3987798:3987955 [4] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-173-121:3987798:3987955 [4] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-173-121:3987798:3987955 [4] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-173-121:3987798:3987955 [4] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-173-121:3987798:3987955 [4] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-173-121:3987794:3987956 [0] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-173-121:3987794:3987956 [0] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-173-121:3987794:3987956 [0] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-173-121:3987794:3987956 [0] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-173-121:3987794:3987956 [0] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-173-121:3987794:3987956 [0] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-173-121:3987794:3987956 [0] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-173-121:3987801:3987960 [7] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-173-121:3987801:3987960 [7] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-173-121:3987801:3987960 [7] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-173-121:3987801:3987960 [7] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-173-121:3987801:3987960 [7] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-173-121:3987801:3987960 [7] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-173-121:3987801:3987960 [7] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-173-121:3987800:3987959 [6] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-173-121:3987800:3987959 [6] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-173-121:3987800:3987959 [6] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-173-121:3987800:3987959 [6] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-173-121:3987800:3987959 [6] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-173-121:3987800:3987959 [6] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-173-121:3987800:3987959 [6] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-173-121:3987800:3987959 [6] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-173-121:3987800:3987959 [6] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-172-142:789259:789417 [4] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-173-121:3987799:3987958 [5] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-173-121:3987797:3987954 [3] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-172-142:789255:789416 [0] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-172-142:789255:789416 [0] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-172-142:789255:789416 [0] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-172-142:789255:789416 [0] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-172-142:789255:789416 [0] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-172-142:789255:789416 [0] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-172-142:789255:789416 [0] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-172-142:789255:789416 [0] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-172-142:789255:789416 [0] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-172-142:789255:789416 [0] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-172-142:789255:789416 [0] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-172-142:789255:789416 [0] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-173-121:3987796:3987957 [2] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-173-121:3987795:3987961 [1] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-173-121:3987794:3987956 [0] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-173-121:3987798:3987955 [4] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-173-121:3987801:3987960 [7] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-173-121:3987800:3987959 [6] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-172-142:789255:789416 [0] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-172-142:789255:789416 [0] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-172-142:789255:789416 [0] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-172-142:789255:789416 [0] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-172-142:789255:789416 [0] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-172-142:789255:789416 [0] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-172-142:789255:789416 [0] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-172-142:789255:789416 [0] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-172-142:789255:789416 [0] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-172-142:789255:789416 [0] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-172-142:789255:789416 [0] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-172-142:789255:789416 [0] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-172-142:789255:789416 [0] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-172-142:789255:789416 [0] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-172-142:789255:789416 [0] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-172-142:789255:789416 [0] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-172-142:789255:789416 [0] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-172-142:789255:789416 [0] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-172-142:789255:789416 [0] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-172-142:789255:789416 [0] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-172-142:789255:789416 [0] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/205
+ip-26-0-172-142:789255:789416 [0] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-172-142:789262:789418 [7] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-172-142:789258:789420 [3] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-172-142:789257:789422 [2] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-172-142:789261:789423 [6] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-172-142:789256:789421 [1] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-172-142:789260:789419 [5] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-172-142:789259:789417 [4] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-172-142:789259:789417 [4] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-172-142:789259:789417 [4] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-172-142:789259:789417 [4] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-172-142:789259:789417 [4] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-172-142:789259:789417 [4] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-172-142:789259:789417 [4] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-172-142:789259:789417 [4] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-172-142:789259:789417 [4] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-172-142:789259:789417 [4] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-172-142:789259:789417 [4] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-172-142:789259:789417 [4] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-172-142:789259:789417 [4] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-172-142:789259:789417 [4] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-172-142:789259:789417 [4] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-172-142:789259:789417 [4] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-172-142:789259:789417 [4] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-172-142:789259:789417 [4] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-172-142:789259:789417 [4] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-172-142:789259:789417 [4] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-172-142:789259:789417 [4] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-172-142:789259:789417 [4] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-172-142:789259:789417 [4] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-172-142:789259:789417 [4] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-172-142:789259:789417 [4] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-172-142:789259:789417 [4] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-172-142:789259:789417 [4] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-172-142:789259:789417 [4] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-172-142:789259:789417 [4] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-172-142:789259:789417 [4] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-172-142:789259:789417 [4] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-172-142:789259:789417 [4] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-172-142:789259:789417 [4] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/184
+ip-26-0-172-142:789259:789417 [4] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-173-121:3987799:3987958 [5] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-173-121:3987799:3987958 [5] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-173-121:3987799:3987958 [5] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-173-121:3987799:3987958 [5] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-173-121:3987799:3987958 [5] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-173-121:3987799:3987958 [5] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-173-121:3987799:3987958 [5] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-173-121:3987799:3987958 [5] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-173-121:3987799:3987958 [5] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-173-121:3987799:3987958 [5] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-173-121:3987799:3987958 [5] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-173-121:3987799:3987958 [5] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-173-121:3987799:3987958 [5] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-173-121:3987799:3987958 [5] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-173-121:3987799:3987958 [5] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-173-121:3987799:3987958 [5] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-173-121:3987799:3987958 [5] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-173-121:3987799:3987958 [5] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-173-121:3987799:3987958 [5] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-173-121:3987799:3987958 [5] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-173-121:3987799:3987958 [5] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-173-121:3987799:3987958 [5] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-173-121:3987799:3987958 [5] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-173-121:3987799:3987958 [5] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-173-121:3987799:3987958 [5] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-173-121:3987799:3987958 [5] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-173-121:3987799:3987958 [5] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-173-121:3987799:3987958 [5] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-173-121:3987799:3987958 [5] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-173-121:3987799:3987958 [5] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-173-121:3987799:3987958 [5] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-173-121:3987799:3987958 [5] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-172-142:789255:789416 [0] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-172-142:789255:789416 [0] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-172-142:789255:789416 [0] NCCL INFO Using network Libfabric
+ip-26-0-173-121:3987799:3987958 [5] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/184
+ip-26-0-173-121:3987799:3987958 [5] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-173-121:3987799:3987958 [5] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-173-121:3987799:3987958 [5] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-173-121:3987799:3987958 [5] NCCL INFO Using network Libfabric
+ip-26-0-173-121:3987797:3987954 [3] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-173-121:3987797:3987954 [3] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-173-121:3987797:3987954 [3] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-173-121:3987797:3987954 [3] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-173-121:3987797:3987954 [3] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-173-121:3987797:3987954 [3] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-173-121:3987797:3987954 [3] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-173-121:3987797:3987954 [3] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-173-121:3987797:3987954 [3] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-173-121:3987797:3987954 [3] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-173-121:3987797:3987954 [3] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-173-121:3987797:3987954 [3] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-173-121:3987797:3987954 [3] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-173-121:3987797:3987954 [3] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-173-121:3987797:3987954 [3] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-173-121:3987797:3987954 [3] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-173-121:3987797:3987954 [3] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-173-121:3987797:3987954 [3] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-173-121:3987797:3987954 [3] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-173-121:3987797:3987954 [3] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-173-121:3987797:3987954 [3] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-173-121:3987797:3987954 [3] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-173-121:3987797:3987954 [3] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-173-121:3987797:3987954 [3] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-173-121:3987797:3987954 [3] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-173-121:3987797:3987954 [3] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-173-121:3987797:3987954 [3] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-173-121:3987797:3987954 [3] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-173-121:3987797:3987954 [3] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-173-121:3987797:3987954 [3] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-173-121:3987797:3987954 [3] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-173-121:3987797:3987954 [3] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-172-142:789257:789422 [2] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-172-142:789257:789422 [2] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-172-142:789257:789422 [2] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-172-142:789257:789422 [2] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-172-142:789257:789422 [2] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-172-142:789257:789422 [2] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-172-142:789257:789422 [2] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-172-142:789257:789422 [2] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-172-142:789257:789422 [2] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-172-142:789257:789422 [2] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-172-142:789257:789422 [2] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-172-142:789257:789422 [2] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-172-142:789257:789422 [2] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-172-142:789257:789422 [2] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-172-142:789257:789422 [2] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-172-142:789257:789422 [2] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-172-142:789257:789422 [2] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-172-142:789257:789422 [2] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-172-142:789257:789422 [2] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-172-142:789257:789422 [2] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-172-142:789257:789422 [2] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-172-142:789257:789422 [2] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-172-142:789257:789422 [2] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-172-142:789257:789422 [2] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-172-142:789257:789422 [2] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-172-142:789257:789422 [2] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-172-142:789257:789422 [2] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-172-142:789257:789422 [2] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-172-142:789257:789422 [2] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-172-142:789257:789422 [2] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-172-142:789257:789422 [2] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-172-142:789257:789422 [2] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-172-142:789257:789422 [2] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/184
+ip-26-0-172-142:789257:789422 [2] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-172-142:789261:789423 [6] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-172-142:789261:789423 [6] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-172-142:789261:789423 [6] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-172-142:789261:789423 [6] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-172-142:789261:789423 [6] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-172-142:789261:789423 [6] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-172-142:789261:789423 [6] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-172-142:789261:789423 [6] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-172-142:789261:789423 [6] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-172-142:789261:789423 [6] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-172-142:789261:789423 [6] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-172-142:789261:789423 [6] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-172-142:789261:789423 [6] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-172-142:789261:789423 [6] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-172-142:789261:789423 [6] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-172-142:789261:789423 [6] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-172-142:789261:789423 [6] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-172-142:789261:789423 [6] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-172-142:789261:789423 [6] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-172-142:789261:789423 [6] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-172-142:789261:789423 [6] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-172-142:789261:789423 [6] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-172-142:789261:789423 [6] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-172-142:789261:789423 [6] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-172-142:789261:789423 [6] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-172-142:789261:789423 [6] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-172-142:789261:789423 [6] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-172-142:789261:789423 [6] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-172-142:789261:789423 [6] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-172-142:789261:789423 [6] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-172-142:789261:789423 [6] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-172-142:789261:789423 [6] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-172-142:789261:789423 [6] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/184
+ip-26-0-172-142:789261:789423 [6] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-173-121:3987797:3987954 [3] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/184
+ip-26-0-173-121:3987797:3987954 [3] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-172-142:789259:789417 [4] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-172-142:789259:789417 [4] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-172-142:789259:789417 [4] NCCL INFO Using network Libfabric
+ip-26-0-172-142:789258:789420 [3] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-172-142:789258:789420 [3] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-172-142:789258:789420 [3] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-172-142:789258:789420 [3] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-172-142:789258:789420 [3] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-172-142:789258:789420 [3] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-172-142:789258:789420 [3] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-172-142:789258:789420 [3] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-172-142:789258:789420 [3] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-172-142:789258:789420 [3] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-172-142:789258:789420 [3] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-172-142:789258:789420 [3] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-172-142:789258:789420 [3] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-172-142:789258:789420 [3] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-172-142:789258:789420 [3] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-172-142:789258:789420 [3] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-172-142:789258:789420 [3] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-172-142:789258:789420 [3] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-172-142:789258:789420 [3] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-172-142:789258:789420 [3] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-172-142:789258:789420 [3] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-172-142:789258:789420 [3] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-172-142:789258:789420 [3] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-172-142:789258:789420 [3] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-172-142:789258:789420 [3] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-172-142:789258:789420 [3] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-172-142:789258:789420 [3] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-172-142:789258:789420 [3] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-172-142:789258:789420 [3] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-172-142:789258:789420 [3] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-172-142:789258:789420 [3] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-172-142:789258:789420 [3] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-172-142:789258:789420 [3] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/184
+ip-26-0-172-142:789258:789420 [3] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-173-121:3987795:3987961 [1] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-173-121:3987795:3987961 [1] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-173-121:3987795:3987961 [1] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-173-121:3987795:3987961 [1] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-173-121:3987795:3987961 [1] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-173-121:3987795:3987961 [1] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-173-121:3987795:3987961 [1] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-173-121:3987795:3987961 [1] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-173-121:3987795:3987961 [1] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-173-121:3987795:3987961 [1] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-173-121:3987795:3987961 [1] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-173-121:3987795:3987961 [1] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-173-121:3987795:3987961 [1] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-173-121:3987795:3987961 [1] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-173-121:3987795:3987961 [1] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-173-121:3987795:3987961 [1] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-173-121:3987795:3987961 [1] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-173-121:3987795:3987961 [1] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-173-121:3987795:3987961 [1] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-173-121:3987795:3987961 [1] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-173-121:3987795:3987961 [1] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-173-121:3987795:3987961 [1] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-173-121:3987795:3987961 [1] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-173-121:3987795:3987961 [1] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-173-121:3987795:3987961 [1] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-173-121:3987795:3987961 [1] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-173-121:3987795:3987961 [1] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-173-121:3987795:3987961 [1] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-173-121:3987795:3987961 [1] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-173-121:3987795:3987961 [1] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-173-121:3987795:3987961 [1] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-173-121:3987795:3987961 [1] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-172-142:789262:789418 [7] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-172-142:789262:789418 [7] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-172-142:789262:789418 [7] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-172-142:789262:789418 [7] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-172-142:789262:789418 [7] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-172-142:789262:789418 [7] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-172-142:789262:789418 [7] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-172-142:789262:789418 [7] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-172-142:789262:789418 [7] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-172-142:789262:789418 [7] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-172-142:789262:789418 [7] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-172-142:789262:789418 [7] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-172-142:789262:789418 [7] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-172-142:789262:789418 [7] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-172-142:789262:789418 [7] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-172-142:789262:789418 [7] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-172-142:789262:789418 [7] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-172-142:789262:789418 [7] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-172-142:789262:789418 [7] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-172-142:789262:789418 [7] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-172-142:789262:789418 [7] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-172-142:789262:789418 [7] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-172-142:789262:789418 [7] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-172-142:789262:789418 [7] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-172-142:789262:789418 [7] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-172-142:789262:789418 [7] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-172-142:789262:789418 [7] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-172-142:789262:789418 [7] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-172-142:789262:789418 [7] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-172-142:789262:789418 [7] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-172-142:789262:789418 [7] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-172-142:789262:789418 [7] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-172-142:789260:789419 [5] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-172-142:789260:789419 [5] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-172-142:789260:789419 [5] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-172-142:789260:789419 [5] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-172-142:789260:789419 [5] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-172-142:789260:789419 [5] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-172-142:789260:789419 [5] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-172-142:789260:789419 [5] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-172-142:789260:789419 [5] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-172-142:789260:789419 [5] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-172-142:789260:789419 [5] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-172-142:789260:789419 [5] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-172-142:789260:789419 [5] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-172-142:789260:789419 [5] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-172-142:789260:789419 [5] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-172-142:789260:789419 [5] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-172-142:789260:789419 [5] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-172-142:789260:789419 [5] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-172-142:789260:789419 [5] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-172-142:789260:789419 [5] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-172-142:789260:789419 [5] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-172-142:789260:789419 [5] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-172-142:789260:789419 [5] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-172-142:789260:789419 [5] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-172-142:789260:789419 [5] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-172-142:789260:789419 [5] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-172-142:789260:789419 [5] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-172-142:789260:789419 [5] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-172-142:789260:789419 [5] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-172-142:789260:789419 [5] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-172-142:789260:789419 [5] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-172-142:789260:789419 [5] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-172-142:789256:789421 [1] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-172-142:789256:789421 [1] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-172-142:789256:789421 [1] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-172-142:789256:789421 [1] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-172-142:789256:789421 [1] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-172-142:789256:789421 [1] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-172-142:789256:789421 [1] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-172-142:789256:789421 [1] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-172-142:789256:789421 [1] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-172-142:789256:789421 [1] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-172-142:789256:789421 [1] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-172-142:789256:789421 [1] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-172-142:789256:789421 [1] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-172-142:789256:789421 [1] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-172-142:789256:789421 [1] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-172-142:789256:789421 [1] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-172-142:789256:789421 [1] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-172-142:789256:789421 [1] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-172-142:789256:789421 [1] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-172-142:789256:789421 [1] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-172-142:789256:789421 [1] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-172-142:789256:789421 [1] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-172-142:789256:789421 [1] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-172-142:789256:789421 [1] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-172-142:789256:789421 [1] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-172-142:789256:789421 [1] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-172-142:789256:789421 [1] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-172-142:789256:789421 [1] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-172-142:789256:789421 [1] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-172-142:789256:789421 [1] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-172-142:789256:789421 [1] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-172-142:789256:789421 [1] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-172-142:789262:789418 [7] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/184
+ip-26-0-172-142:789262:789418 [7] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-172-142:789260:789419 [5] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/184
+ip-26-0-172-142:789260:789419 [5] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-172-142:789256:789421 [1] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/184
+ip-26-0-172-142:789256:789421 [1] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-173-121:3987795:3987961 [1] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/184
+ip-26-0-173-121:3987795:3987961 [1] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-173-121:3987796:3987957 [2] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-173-121:3987796:3987957 [2] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-173-121:3987796:3987957 [2] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-173-121:3987796:3987957 [2] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-173-121:3987796:3987957 [2] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-173-121:3987796:3987957 [2] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-173-121:3987796:3987957 [2] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-173-121:3987796:3987957 [2] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-173-121:3987796:3987957 [2] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-173-121:3987796:3987957 [2] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-173-121:3987796:3987957 [2] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-173-121:3987796:3987957 [2] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-173-121:3987796:3987957 [2] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-173-121:3987796:3987957 [2] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-173-121:3987796:3987957 [2] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-173-121:3987796:3987957 [2] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-173-121:3987796:3987957 [2] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-173-121:3987796:3987957 [2] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-173-121:3987796:3987957 [2] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-173-121:3987796:3987957 [2] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-173-121:3987796:3987957 [2] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-173-121:3987796:3987957 [2] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-173-121:3987796:3987957 [2] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-173-121:3987796:3987957 [2] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-173-121:3987796:3987957 [2] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-173-121:3987796:3987957 [2] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-173-121:3987796:3987957 [2] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-173-121:3987796:3987957 [2] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-173-121:3987796:3987957 [2] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-173-121:3987796:3987957 [2] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-173-121:3987796:3987957 [2] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-173-121:3987796:3987957 [2] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-173-121:3987796:3987957 [2] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/184
+ip-26-0-173-121:3987796:3987957 [2] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-173-121:3987794:3987956 [0] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-173-121:3987794:3987956 [0] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-173-121:3987794:3987956 [0] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-173-121:3987794:3987956 [0] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-173-121:3987794:3987956 [0] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-173-121:3987794:3987956 [0] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-173-121:3987794:3987956 [0] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-173-121:3987794:3987956 [0] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-173-121:3987794:3987956 [0] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-173-121:3987794:3987956 [0] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-173-121:3987794:3987956 [0] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-173-121:3987794:3987956 [0] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-173-121:3987794:3987956 [0] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-173-121:3987794:3987956 [0] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-173-121:3987794:3987956 [0] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-173-121:3987794:3987956 [0] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-173-121:3987794:3987956 [0] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-173-121:3987794:3987956 [0] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-173-121:3987794:3987956 [0] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-173-121:3987794:3987956 [0] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-173-121:3987794:3987956 [0] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-173-121:3987794:3987956 [0] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-173-121:3987794:3987956 [0] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-173-121:3987794:3987956 [0] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-173-121:3987794:3987956 [0] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-173-121:3987794:3987956 [0] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-173-121:3987794:3987956 [0] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-173-121:3987794:3987956 [0] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-173-121:3987794:3987956 [0] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-173-121:3987794:3987956 [0] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-173-121:3987794:3987956 [0] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-173-121:3987794:3987956 [0] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-173-121:3987800:3987959 [6] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-173-121:3987800:3987959 [6] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-173-121:3987800:3987959 [6] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-173-121:3987800:3987959 [6] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-173-121:3987800:3987959 [6] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-173-121:3987800:3987959 [6] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-173-121:3987800:3987959 [6] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-173-121:3987800:3987959 [6] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-173-121:3987800:3987959 [6] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-173-121:3987800:3987959 [6] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-173-121:3987800:3987959 [6] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-173-121:3987800:3987959 [6] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-173-121:3987800:3987959 [6] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-173-121:3987800:3987959 [6] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-173-121:3987800:3987959 [6] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-173-121:3987800:3987959 [6] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-173-121:3987800:3987959 [6] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-173-121:3987800:3987959 [6] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-173-121:3987800:3987959 [6] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-173-121:3987800:3987959 [6] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-173-121:3987800:3987959 [6] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-173-121:3987800:3987959 [6] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-173-121:3987800:3987959 [6] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-173-121:3987800:3987959 [6] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-173-121:3987800:3987959 [6] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-173-121:3987800:3987959 [6] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-173-121:3987800:3987959 [6] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-173-121:3987800:3987959 [6] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-173-121:3987800:3987959 [6] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-173-121:3987800:3987959 [6] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-173-121:3987800:3987959 [6] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-173-121:3987800:3987959 [6] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-173-121:3987801:3987960 [7] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-173-121:3987801:3987960 [7] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-173-121:3987801:3987960 [7] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-173-121:3987801:3987960 [7] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-173-121:3987801:3987960 [7] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-173-121:3987801:3987960 [7] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-173-121:3987801:3987960 [7] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-173-121:3987801:3987960 [7] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-173-121:3987801:3987960 [7] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-173-121:3987801:3987960 [7] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-173-121:3987801:3987960 [7] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-173-121:3987801:3987960 [7] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-173-121:3987801:3987960 [7] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-173-121:3987801:3987960 [7] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-173-121:3987801:3987960 [7] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-173-121:3987801:3987960 [7] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-173-121:3987801:3987960 [7] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-173-121:3987801:3987960 [7] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-173-121:3987801:3987960 [7] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-173-121:3987801:3987960 [7] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-173-121:3987801:3987960 [7] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-173-121:3987801:3987960 [7] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-173-121:3987801:3987960 [7] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-173-121:3987801:3987960 [7] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-173-121:3987801:3987960 [7] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-173-121:3987801:3987960 [7] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-173-121:3987801:3987960 [7] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-173-121:3987801:3987960 [7] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-173-121:3987801:3987960 [7] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-173-121:3987801:3987960 [7] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-173-121:3987801:3987960 [7] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-173-121:3987801:3987960 [7] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-173-121:3987794:3987956 [0] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/184
+ip-26-0-173-121:3987794:3987956 [0] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-173-121:3987798:3987955 [4] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-173-121:3987798:3987955 [4] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-173-121:3987798:3987955 [4] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-173-121:3987798:3987955 [4] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-173-121:3987798:3987955 [4] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-173-121:3987798:3987955 [4] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-173-121:3987798:3987955 [4] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-173-121:3987798:3987955 [4] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-173-121:3987798:3987955 [4] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-173-121:3987798:3987955 [4] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-173-121:3987798:3987955 [4] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-173-121:3987798:3987955 [4] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-173-121:3987798:3987955 [4] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-173-121:3987798:3987955 [4] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-173-121:3987798:3987955 [4] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-173-121:3987798:3987955 [4] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-173-121:3987798:3987955 [4] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-173-121:3987798:3987955 [4] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-173-121:3987798:3987955 [4] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-173-121:3987798:3987955 [4] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-173-121:3987798:3987955 [4] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-173-121:3987798:3987955 [4] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-173-121:3987798:3987955 [4] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-173-121:3987798:3987955 [4] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-173-121:3987798:3987955 [4] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-173-121:3987798:3987955 [4] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-173-121:3987798:3987955 [4] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-173-121:3987798:3987955 [4] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-173-121:3987798:3987955 [4] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-173-121:3987798:3987955 [4] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-173-121:3987798:3987955 [4] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-173-121:3987798:3987955 [4] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-173-121:3987800:3987959 [6] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/184
+ip-26-0-173-121:3987800:3987959 [6] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-173-121:3987801:3987960 [7] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/184
+ip-26-0-173-121:3987801:3987960 [7] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-173-121:3987798:3987955 [4] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/184
+ip-26-0-173-121:3987798:3987955 [4] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-173-121:3987797:3987954 [3] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-173-121:3987797:3987954 [3] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-173-121:3987797:3987954 [3] NCCL INFO Using network Libfabric
+ip-26-0-173-121:3987795:3987961 [1] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-173-121:3987795:3987961 [1] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-173-121:3987795:3987961 [1] NCCL INFO Using network Libfabric
+ip-26-0-172-142:789257:789422 [2] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-172-142:789257:789422 [2] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-172-142:789257:789422 [2] NCCL INFO Using network Libfabric
+ip-26-0-173-121:3987796:3987957 [2] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-173-121:3987796:3987957 [2] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-173-121:3987796:3987957 [2] NCCL INFO Using network Libfabric
+ip-26-0-173-121:3987801:3987960 [7] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-173-121:3987801:3987960 [7] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-173-121:3987800:3987959 [6] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-173-121:3987800:3987959 [6] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-173-121:3987800:3987959 [6] NCCL INFO Using network Libfabric
+ip-26-0-173-121:3987801:3987960 [7] NCCL INFO Using network Libfabric
+ip-26-0-173-121:3987794:3987956 [0] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-173-121:3987794:3987956 [0] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-173-121:3987794:3987956 [0] NCCL INFO Using network Libfabric
+ip-26-0-173-121:3987798:3987955 [4] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-173-121:3987798:3987955 [4] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-173-121:3987798:3987955 [4] NCCL INFO Using network Libfabric
+ip-26-0-172-142:789261:789423 [6] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-172-142:789261:789423 [6] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-172-142:789261:789423 [6] NCCL INFO Using network Libfabric
+ip-26-0-172-142:789258:789420 [3] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-172-142:789258:789420 [3] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-172-142:789258:789420 [3] NCCL INFO Using network Libfabric
+ip-26-0-172-142:789260:789419 [5] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-172-142:789260:789419 [5] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-172-142:789260:789419 [5] NCCL INFO Using network Libfabric
+ip-26-0-172-142:789262:789418 [7] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-172-142:789262:789418 [7] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-172-142:789262:789418 [7] NCCL INFO Using network Libfabric
+ip-26-0-172-142:789256:789421 [1] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-172-142:789256:789421 [1] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-172-142:789256:789421 [1] NCCL INFO Using network Libfabric
+ip-26-0-172-142:789262:789418 [7] NCCL INFO comm 0x87c3e80 rank 7 nranks 16 cudaDev 7 nvmlDev 7 busId ca000 commId 0xf40a4bd2dbd7f17c - Init START
+ip-26-0-172-142:789259:789417 [4] NCCL INFO comm 0x8e10500 rank 4 nranks 16 cudaDev 4 nvmlDev 4 busId 97000 commId 0xf40a4bd2dbd7f17c - Init START
+ip-26-0-172-142:789260:789419 [5] NCCL INFO comm 0x9e37670 rank 5 nranks 16 cudaDev 5 nvmlDev 5 busId a8000 commId 0xf40a4bd2dbd7f17c - Init START
+ip-26-0-172-142:789258:789420 [3] NCCL INFO comm 0x88eba40 rank 3 nranks 16 cudaDev 3 nvmlDev 3 busId 86000 commId 0xf40a4bd2dbd7f17c - Init START
+ip-26-0-172-142:789257:789422 [2] NCCL INFO comm 0x916e000 rank 2 nranks 16 cudaDev 2 nvmlDev 2 busId 75000 commId 0xf40a4bd2dbd7f17c - Init START
+ip-26-0-172-142:789256:789421 [1] NCCL INFO comm 0xa0887a0 rank 1 nranks 16 cudaDev 1 nvmlDev 1 busId 64000 commId 0xf40a4bd2dbd7f17c - Init START
+ip-26-0-173-121:3987796:3987957 [2] NCCL INFO comm 0x8e72160 rank 10 nranks 16 cudaDev 2 nvmlDev 2 busId 75000 commId 0xf40a4bd2dbd7f17c - Init START
+ip-26-0-173-121:3987795:3987961 [1] NCCL INFO comm 0x95d08c0 rank 9 nranks 16 cudaDev 1 nvmlDev 1 busId 64000 commId 0xf40a4bd2dbd7f17c - Init START
+ip-26-0-173-121:3987798:3987955 [4] NCCL INFO comm 0x9b44720 rank 12 nranks 16 cudaDev 4 nvmlDev 4 busId 97000 commId 0xf40a4bd2dbd7f17c - Init START
+ip-26-0-172-142:789261:789423 [6] NCCL INFO comm 0x9bb0c50 rank 6 nranks 16 cudaDev 6 nvmlDev 6 busId b9000 commId 0xf40a4bd2dbd7f17c - Init START
+ip-26-0-173-121:3987801:3987960 [7] NCCL INFO comm 0x9388d10 rank 15 nranks 16 cudaDev 7 nvmlDev 7 busId ca000 commId 0xf40a4bd2dbd7f17c - Init START
+ip-26-0-173-121:3987794:3987956 [0] NCCL INFO comm 0x8a09bb0 rank 8 nranks 16 cudaDev 0 nvmlDev 0 busId 53000 commId 0xf40a4bd2dbd7f17c - Init START
+ip-26-0-173-121:3987799:3987958 [5] NCCL INFO comm 0x8f9e050 rank 13 nranks 16 cudaDev 5 nvmlDev 5 busId a8000 commId 0xf40a4bd2dbd7f17c - Init START
+ip-26-0-173-121:3987797:3987954 [3] NCCL INFO comm 0x88c5420 rank 11 nranks 16 cudaDev 3 nvmlDev 3 busId 86000 commId 0xf40a4bd2dbd7f17c - Init START
+ip-26-0-172-142:789255:789416 [0] NCCL INFO comm 0x9d36820 rank 0 nranks 16 cudaDev 0 nvmlDev 0 busId 53000 commId 0xf40a4bd2dbd7f17c - Init START
+ip-26-0-173-121:3987800:3987959 [6] NCCL INFO comm 0x8feba80 rank 14 nranks 16 cudaDev 6 nvmlDev 6 busId b9000 commId 0xf40a4bd2dbd7f17c - Init START
+ip-26-0-172-142:789262:789418 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-172-142:789259:789417 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-172-142:789260:789419 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-172-142:789261:789423 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-172-142:789258:789420 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-172-142:789256:789421 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-172-142:789257:789422 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-173-121:3987797:3987954 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-173-121:3987801:3987960 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-173-121:3987798:3987955 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-173-121:3987794:3987956 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-173-121:3987795:3987961 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-172-142:789255:789416 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/205
+ip-26-0-173-121:3987800:3987959 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-173-121:3987799:3987958 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-173-121:3987796:3987957 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-173-121:3987799:3987958 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-173-121:3987799:3987958 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-173-121:3987796:3987957 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-173-121:3987796:3987957 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-173-121:3987800:3987959 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-173-121:3987800:3987959 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-173-121:3987798:3987955 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-173-121:3987798:3987955 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-173-121:3987797:3987954 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-173-121:3987797:3987954 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-173-121:3987794:3987956 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-173-121:3987794:3987956 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-173-121:3987801:3987960 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-173-121:3987801:3987960 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-173-121:3987795:3987961 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-173-121:3987795:3987961 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-172-142:789261:789423 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-172-142:789261:789423 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-172-142:789256:789421 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-172-142:789256:789421 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-172-142:789262:789418 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-172-142:789262:789418 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-172-142:789255:789416 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-172-142:789255:789416 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-172-142:789259:789417 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-172-142:789259:789417 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-172-142:789258:789420 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-172-142:789260:789419 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-172-142:789260:789419 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-172-142:789258:789420 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-172-142:789257:789422 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-172-142:789257:789422 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-173-121:3987801:3987960 [7] NCCL INFO Trees [0] -1/-1/-1->15->14 [1] 8/-1/-1->15->14 [2] 8/-1/-1->15->14 [3] 8/-1/-1->15->14 [4] 8/-1/-1->15->14 [5] 8/-1/-1->15->14 [6] 8/-1/-1->15->14 [7] 8/-1/-1->15->7 [8] -1/-1/-1->15->14 [9] 8/-1/-1->15->14 [10] 8/-1/-1->15->14 [11] 8/-1/-1->15->14 [12] 8/-1/-1->15->14 [13] 8/-1/-1->15->14 [14] 8/-1/-1->15->14 [15] 8/7/-1->15->-1
+ip-26-0-173-121:3987801:3987960 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-142:789256:789421 [1] NCCL INFO NVLS Head  0:  0  8
+ip-26-0-172-142:789260:789419 [5] NCCL INFO NVLS Head  0:  0  8
+ip-26-0-172-142:789260:789419 [5] NCCL INFO NVLS Head  1:  1  9
+ip-26-0-172-142:789256:789421 [1] NCCL INFO NVLS Head  1:  1  9
+ip-26-0-172-142:789261:789423 [6] NCCL INFO NVLS Head  0:  0  8
+ip-26-0-172-142:789260:789419 [5] NCCL INFO NVLS Head  2:  2 10
+ip-26-0-172-142:789256:789421 [1] NCCL INFO NVLS Head  2:  2 10
+ip-26-0-172-142:789255:789416 [0] NCCL INFO NVLS Head  0:  0  8
+ip-26-0-172-142:789260:789419 [5] NCCL INFO NVLS Head  3:  3 11
+ip-26-0-172-142:789261:789423 [6] NCCL INFO NVLS Head  1:  1  9
+ip-26-0-172-142:789256:789421 [1] NCCL INFO NVLS Head  3:  3 11
+ip-26-0-172-142:789255:789416 [0] NCCL INFO NVLS Head  1:  1  9
+ip-26-0-172-142:789262:789418 [7] NCCL INFO NVLS Head  0:  0  8
+ip-26-0-172-142:789260:789419 [5] NCCL INFO NVLS Head  4:  4 12
+ip-26-0-172-142:789256:789421 [1] NCCL INFO NVLS Head  4:  4 12
+ip-26-0-172-142:789259:789417 [4] NCCL INFO NVLS Head  0:  0  8
+ip-26-0-172-142:789261:789423 [6] NCCL INFO NVLS Head  2:  2 10
+ip-26-0-172-142:789255:789416 [0] NCCL INFO NVLS Head  2:  2 10
+ip-26-0-172-142:789262:789418 [7] NCCL INFO NVLS Head  1:  1  9
+ip-26-0-172-142:789259:789417 [4] NCCL INFO NVLS Head  1:  1  9
+ip-26-0-172-142:789255:789416 [0] NCCL INFO NVLS Head  3:  3 11
+ip-26-0-172-142:789260:789419 [5] NCCL INFO NVLS Head  5:  5 13
+ip-26-0-172-142:789256:789421 [1] NCCL INFO NVLS Head  5:  5 13
+ip-26-0-172-142:789261:789423 [6] NCCL INFO NVLS Head  3:  3 11
+ip-26-0-172-142:789258:789420 [3] NCCL INFO NVLS Head  0:  0  8
+ip-26-0-172-142:789262:789418 [7] NCCL INFO NVLS Head  2:  2 10
+ip-26-0-172-142:789259:789417 [4] NCCL INFO NVLS Head  2:  2 10
+ip-26-0-172-142:789255:789416 [0] NCCL INFO NVLS Head  4:  4 12
+ip-26-0-172-142:789259:789417 [4] NCCL INFO NVLS Head  3:  3 11
+ip-26-0-173-121:3987796:3987957 [2] NCCL INFO Trees [0] 11/-1/-1->10->9 [1] 11/-1/-1->10->9 [2] 11/-1/-1->10->2 [3] -1/-1/-1->10->9 [4] 11/-1/-1->10->9 [5] 11/-1/-1->10->9 [6] 11/-1/-1->10->9 [7] 11/-1/-1->10->9 [8] 11/-1/-1->10->9 [9] 11/-1/-1->10->9 [10] 11/2/-1->10->-1 [11] -1/-1/-1->10->9 [12] 11/-1/-1->10->9 [13] 11/-1/-1->10->9 [14] 11/-1/-1->10->9 [15] 11/-1/-1->10->9
+ip-26-0-172-142:789260:789419 [5] NCCL INFO NVLS Head  6:  6 14
+ip-26-0-172-142:789256:789421 [1] NCCL INFO NVLS Head  6:  6 14
+ip-26-0-172-142:789259:789417 [4] NCCL INFO NVLS Head  4:  4 12
+ip-26-0-172-142:789261:789423 [6] NCCL INFO NVLS Head  4:  4 12
+ip-26-0-172-142:789257:789422 [2] NCCL INFO NVLS Head  0:  0  8
+ip-26-0-172-142:789256:789421 [1] NCCL INFO NVLS Head  7:  7 15
+ip-26-0-172-142:789261:789423 [6] NCCL INFO NVLS Head  5:  5 13
+ip-26-0-172-142:789262:789418 [7] NCCL INFO NVLS Head  3:  3 11
+ip-26-0-172-142:789257:789422 [2] NCCL INFO NVLS Head  1:  1  9
+ip-26-0-172-142:789258:789420 [3] NCCL INFO NVLS Head  1:  1  9
+ip-26-0-172-142:789255:789416 [0] NCCL INFO NVLS Head  5:  5 13
+ip-26-0-173-121:3987796:3987957 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-173-121:3987799:3987958 [5] NCCL INFO Trees [0] 14/-1/-1->13->12 [1] 14/-1/-1->13->12 [2] 14/-1/-1->13->12 [3] 14/-1/-1->13->12 [4] 14/-1/-1->13->12 [5] 14/-1/-1->13->5 [6] -1/-1/-1->13->12 [7] 14/-1/-1->13->12 [8] 14/-1/-1->13->12 [9] 14/-1/-1->13->12 [10] 14/-1/-1->13->12 [11] 14/-1/-1->13->12 [12] 14/-1/-1->13->12 [13] 14/5/-1->13->-1 [14] -1/-1/-1->13->12 [15] 14/-1/-1->13->12
+ip-26-0-172-142:789260:789419 [5] NCCL INFO NVLS Head  7:  7 15
+ip-26-0-172-142:789259:789417 [4] NCCL INFO NVLS Head  5:  5 13
+ip-26-0-172-142:789261:789423 [6] NCCL INFO NVLS Head  6:  6 14
+ip-26-0-172-142:789262:789418 [7] NCCL INFO NVLS Head  4:  4 12
+ip-26-0-172-142:789257:789422 [2] NCCL INFO NVLS Head  2:  2 10
+ip-26-0-172-142:789262:789418 [7] NCCL INFO NVLS Head  5:  5 13
+ip-26-0-172-142:789258:789420 [3] NCCL INFO NVLS Head  2:  2 10
+ip-26-0-172-142:789255:789416 [0] NCCL INFO NVLS Head  6:  6 14
+ip-26-0-172-142:789259:789417 [4] NCCL INFO NVLS Head  6:  6 14
+ip-26-0-172-142:789261:789423 [6] NCCL INFO NVLS Head  7:  7 15
+ip-26-0-172-142:789255:789416 [0] NCCL INFO NVLS Head  7:  7 15
+ip-26-0-173-121:3987799:3987958 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-142:789256:789421 [1] NCCL INFO Trees [0] 2/-1/-1->1->0 [1] 2/9/-1->1->-1 [2] -1/-1/-1->1->0 [3] 2/-1/-1->1->0 [4] 2/-1/-1->1->0 [5] 2/-1/-1->1->0 [6] 2/-1/-1->1->0 [7] 2/-1/-1->1->0 [8] 2/-1/-1->1->0 [9] 2/-1/-1->1->9 [10] -1/-1/-1->1->0 [11] 2/-1/-1->1->0 [12] 2/-1/-1->1->0 [13] 2/-1/-1->1->0 [14] 2/-1/-1->1->0 [15] 2/-1/-1->1->0
+ip-26-0-172-142:789257:789422 [2] NCCL INFO NVLS Head  3:  3 11
+ip-26-0-172-142:789262:789418 [7] NCCL INFO NVLS Head  6:  6 14
+ip-26-0-172-142:789258:789420 [3] NCCL INFO NVLS Head  3:  3 11
+ip-26-0-172-142:789262:789418 [7] NCCL INFO NVLS Head  7:  7 15
+ip-26-0-172-142:789259:789417 [4] NCCL INFO NVLS Head  7:  7 15
+ip-26-0-172-142:789256:789421 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-142:789257:789422 [2] NCCL INFO NVLS Head  4:  4 12
+ip-26-0-172-142:789258:789420 [3] NCCL INFO NVLS Head  4:  4 12
+ip-26-0-173-121:3987800:3987959 [6] NCCL INFO Trees [0] 15/-1/-1->14->13 [1] 15/-1/-1->14->13 [2] 15/-1/-1->14->13 [3] 15/-1/-1->14->13 [4] 15/-1/-1->14->13 [5] 15/-1/-1->14->13 [6] 15/-1/-1->14->6 [7] -1/-1/-1->14->13 [8] 15/-1/-1->14->13 [9] 15/-1/-1->14->13 [10] 15/-1/-1->14->13 [11] 15/-1/-1->14->13 [12] 15/-1/-1->14->13 [13] 15/-1/-1->14->13 [14] 15/6/-1->14->-1 [15] -1/-1/-1->14->13
+ip-26-0-172-142:789260:789419 [5] NCCL INFO Trees [0] 6/-1/-1->5->4 [1] 6/-1/-1->5->4 [2] 6/-1/-1->5->4 [3] 6/-1/-1->5->4 [4] 6/-1/-1->5->4 [5] 6/13/-1->5->-1 [6] -1/-1/-1->5->4 [7] 6/-1/-1->5->4 [8] 6/-1/-1->5->4 [9] 6/-1/-1->5->4 [10] 6/-1/-1->5->4 [11] 6/-1/-1->5->4 [12] 6/-1/-1->5->4 [13] 6/-1/-1->5->13 [14] -1/-1/-1->5->4 [15] 6/-1/-1->5->4
+ip-26-0-172-142:789258:789420 [3] NCCL INFO NVLS Head  5:  5 13
+ip-26-0-172-142:789255:789416 [0] NCCL INFO Channel 00/16 :    0   1   2   3   4   5   6   7   8   9  10  11  12  13  14  15
+ip-26-0-172-142:789258:789420 [3] NCCL INFO NVLS Head  6:  6 14
+ip-26-0-172-142:789257:789422 [2] NCCL INFO NVLS Head  5:  5 13
+ip-26-0-172-142:789260:789419 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-142:789258:789420 [3] NCCL INFO NVLS Head  7:  7 15
+ip-26-0-173-121:3987800:3987959 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-142:789261:789423 [6] NCCL INFO Trees [0] 7/-1/-1->6->5 [1] 7/-1/-1->6->5 [2] 7/-1/-1->6->5 [3] 7/-1/-1->6->5 [4] 7/-1/-1->6->5 [5] 7/-1/-1->6->5 [6] 7/14/-1->6->-1 [7] -1/-1/-1->6->5 [8] 7/-1/-1->6->5 [9] 7/-1/-1->6->5 [10] 7/-1/-1->6->5 [11] 7/-1/-1->6->5 [12] 7/-1/-1->6->5 [13] 7/-1/-1->6->5 [14] 7/-1/-1->6->14 [15] -1/-1/-1->6->5
+ip-26-0-172-142:789255:789416 [0] NCCL INFO Channel 01/16 :    0   7   6   5   4   3   2   9   8  15  14  13  12  11  10   1
+ip-26-0-173-121:3987794:3987956 [0] NCCL INFO Trees [0] 9/-1/-1->8->0 [1] -1/-1/-1->8->15 [2] 9/-1/-1->8->15 [3] 9/-1/-1->8->15 [4] 9/-1/-1->8->15 [5] 9/-1/-1->8->15 [6] 9/-1/-1->8->15 [7] 9/-1/-1->8->15 [8] 9/0/-1->8->-1 [9] -1/-1/-1->8->15 [10] 9/-1/-1->8->15 [11] 9/-1/-1->8->15 [12] 9/-1/-1->8->15 [13] 9/-1/-1->8->15 [14] 9/-1/-1->8->15 [15] 9/-1/-1->8->15
+ip-26-0-173-121:3987794:3987956 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-142:789259:789417 [4] NCCL INFO Trees [0] 5/-1/-1->4->3 [1] 5/-1/-1->4->3 [2] 5/-1/-1->4->3 [3] 5/-1/-1->4->3 [4] 5/12/-1->4->-1 [5] -1/-1/-1->4->3 [6] 5/-1/-1->4->3 [7] 5/-1/-1->4->3 [8] 5/-1/-1->4->3 [9] 5/-1/-1->4->3 [10] 5/-1/-1->4->3 [11] 5/-1/-1->4->3 [12] 5/-1/-1->4->12 [13] -1/-1/-1->4->3 [14] 5/-1/-1->4->3 [15] 5/-1/-1->4->3
+ip-26-0-172-142:789257:789422 [2] NCCL INFO NVLS Head  6:  6 14
+ip-26-0-172-142:789261:789423 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-142:789255:789416 [0] NCCL INFO Channel 02/16 :    0   7   6   5   4   3  10   9   8  15  14  13  12  11   2   1
+ip-26-0-172-142:789259:789417 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-173-121:3987797:3987954 [3] NCCL INFO Trees [0] 12/-1/-1->11->10 [1] 12/-1/-1->11->10 [2] 12/-1/-1->11->10 [3] 12/-1/-1->11->3 [4] -1/-1/-1->11->10 [5] 12/-1/-1->11->10 [6] 12/-1/-1->11->10 [7] 12/-1/-1->11->10 [8] 12/-1/-1->11->10 [9] 12/-1/-1->11->10 [10] 12/-1/-1->11->10 [11] 12/3/-1->11->-1 [12] -1/-1/-1->11->10 [13] 12/-1/-1->11->10 [14] 12/-1/-1->11->10 [15] 12/-1/-1->11->10
+ip-26-0-173-121:3987795:3987961 [1] NCCL INFO Trees [0] 10/-1/-1->9->8 [1] 10/-1/-1->9->1 [2] -1/-1/-1->9->8 [3] 10/-1/-1->9->8 [4] 10/-1/-1->9->8 [5] 10/-1/-1->9->8 [6] 10/-1/-1->9->8 [7] 10/-1/-1->9->8 [8] 10/-1/-1->9->8 [9] 10/1/-1->9->-1 [10] -1/-1/-1->9->8 [11] 10/-1/-1->9->8 [12] 10/-1/-1->9->8 [13] 10/-1/-1->9->8 [14] 10/-1/-1->9->8 [15] 10/-1/-1->9->8
+ip-26-0-173-121:3987795:3987961 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-142:789262:789418 [7] NCCL INFO Trees [0] -1/-1/-1->7->6 [1] 0/-1/-1->7->6 [2] 0/-1/-1->7->6 [3] 0/-1/-1->7->6 [4] 0/-1/-1->7->6 [5] 0/-1/-1->7->6 [6] 0/-1/-1->7->6 [7] 0/15/-1->7->-1 [8] -1/-1/-1->7->6 [9] 0/-1/-1->7->6 [10] 0/-1/-1->7->6 [11] 0/-1/-1->7->6 [12] 0/-1/-1->7->6 [13] 0/-1/-1->7->6 [14] 0/-1/-1->7->6 [15] 0/-1/-1->7->15
+ip-26-0-172-142:789257:789422 [2] NCCL INFO NVLS Head  7:  7 15
+ip-26-0-172-142:789255:789416 [0] NCCL INFO Channel 03/16 :    0   7   6   5   4  11  10   9   8  15  14  13  12   3   2   1
+ip-26-0-172-142:789262:789418 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-142:789255:789416 [0] NCCL INFO Channel 04/16 :    0   7   6   5  12  11  10   9   8  15  14  13   4   3   2   1
+ip-26-0-172-142:789255:789416 [0] NCCL INFO Channel 05/16 :    0   4   7   6  13  11  10   9   8  12  15  14   5   3   2   1
+ip-26-0-172-142:789255:789416 [0] NCCL INFO Channel 06/16 :    0   5   4   7  14  11  10   9   8  13  12  15   6   3   2   1
+ip-26-0-173-121:3987797:3987954 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-173-121:3987798:3987955 [4] NCCL INFO Trees [0] 13/-1/-1->12->11 [1] 13/-1/-1->12->11 [2] 13/-1/-1->12->11 [3] 13/-1/-1->12->11 [4] 13/-1/-1->12->4 [5] -1/-1/-1->12->11 [6] 13/-1/-1->12->11 [7] 13/-1/-1->12->11 [8] 13/-1/-1->12->11 [9] 13/-1/-1->12->11 [10] 13/-1/-1->12->11 [11] 13/-1/-1->12->11 [12] 13/4/-1->12->-1 [13] -1/-1/-1->12->11 [14] 13/-1/-1->12->11 [15] 13/-1/-1->12->11
+ip-26-0-172-142:789255:789416 [0] NCCL INFO Channel 07/16 :    0   6   5   4  15  11  10   9   8  14  13  12   7   3   2   1
+ip-26-0-172-142:789258:789420 [3] NCCL INFO Trees [0] 4/-1/-1->3->2 [1] 4/-1/-1->3->2 [2] 4/-1/-1->3->2 [3] 4/11/-1->3->-1 [4] -1/-1/-1->3->2 [5] 4/-1/-1->3->2 [6] 4/-1/-1->3->2 [7] 4/-1/-1->3->2 [8] 4/-1/-1->3->2 [9] 4/-1/-1->3->2 [10] 4/-1/-1->3->2 [11] 4/-1/-1->3->11 [12] -1/-1/-1->3->2 [13] 4/-1/-1->3->2 [14] 4/-1/-1->3->2 [15] 4/-1/-1->3->2
+ip-26-0-172-142:789255:789416 [0] NCCL INFO Channel 08/16 :    0   1   2   3   4   5   6   7   8   9  10  11  12  13  14  15
+ip-26-0-172-142:789257:789422 [2] NCCL INFO Trees [0] 3/-1/-1->2->1 [1] 3/-1/-1->2->1 [2] 3/10/-1->2->-1 [3] -1/-1/-1->2->1 [4] 3/-1/-1->2->1 [5] 3/-1/-1->2->1 [6] 3/-1/-1->2->1 [7] 3/-1/-1->2->1 [8] 3/-1/-1->2->1 [9] 3/-1/-1->2->1 [10] 3/-1/-1->2->10 [11] -1/-1/-1->2->1 [12] 3/-1/-1->2->1 [13] 3/-1/-1->2->1 [14] 3/-1/-1->2->1 [15] 3/-1/-1->2->1
+ip-26-0-172-142:789258:789420 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-173-121:3987798:3987955 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-142:789255:789416 [0] NCCL INFO Channel 09/16 :    0   7   6   5   4   3   2   9   8  15  14  13  12  11  10   1
+ip-26-0-172-142:789257:789422 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-142:789255:789416 [0] NCCL INFO Channel 10/16 :    0   7   6   5   4   3  10   9   8  15  14  13  12  11   2   1
+ip-26-0-172-142:789255:789416 [0] NCCL INFO Channel 11/16 :    0   7   6   5   4  11  10   9   8  15  14  13  12   3   2   1
+ip-26-0-172-142:789255:789416 [0] NCCL INFO Channel 12/16 :    0   7   6   5  12  11  10   9   8  15  14  13   4   3   2   1
+ip-26-0-172-142:789255:789416 [0] NCCL INFO Channel 13/16 :    0   4   7   6  13  11  10   9   8  12  15  14   5   3   2   1
+ip-26-0-172-142:789255:789416 [0] NCCL INFO Channel 14/16 :    0   5   4   7  14  11  10   9   8  13  12  15   6   3   2   1
+ip-26-0-172-142:789255:789416 [0] NCCL INFO Channel 15/16 :    0   6   5   4  15  11  10   9   8  14  13  12   7   3   2   1
+ip-26-0-172-142:789255:789416 [0] NCCL INFO Trees [0] 1/8/-1->0->-1 [1] -1/-1/-1->0->7 [2] 1/-1/-1->0->7 [3] 1/-1/-1->0->7 [4] 1/-1/-1->0->7 [5] 1/-1/-1->0->7 [6] 1/-1/-1->0->7 [7] 1/-1/-1->0->7 [8] 1/-1/-1->0->8 [9] -1/-1/-1->0->7 [10] 1/-1/-1->0->7 [11] 1/-1/-1->0->7 [12] 1/-1/-1->0->7 [13] 1/-1/-1->0->7 [14] 1/-1/-1->0->7 [15] 1/-1/-1->0->7
+ip-26-0-172-142:789255:789416 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-173-121:3987794:3987956 [0] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-173-121:3987796:3987957 [2] NCCL INFO Channel 00/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-173-121:3987801:3987960 [7] NCCL INFO Channel 00/0 : 15[7] -> 0[0] [send] via NET/Libfabric/0(8)/GDRDMA
+ip-26-0-173-121:3987799:3987958 [5] NCCL INFO Channel 00/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-173-121:3987800:3987959 [6] NCCL INFO Channel 00/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-173-121:3987795:3987961 [1] NCCL INFO Channel 00/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-173-121:3987796:3987957 [2] NCCL INFO Channel 08/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-173-121:3987799:3987958 [5] NCCL INFO Channel 08/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-173-121:3987800:3987959 [6] NCCL INFO Channel 08/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-173-121:3987795:3987961 [1] NCCL INFO Channel 08/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-173-121:3987798:3987955 [4] NCCL INFO Channel 00/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-173-121:3987797:3987954 [3] NCCL INFO Channel 00/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-173-121:3987798:3987955 [4] NCCL INFO Channel 08/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-173-121:3987797:3987954 [3] NCCL INFO Channel 08/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-173-121:3987796:3987957 [2] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-173-121:3987798:3987955 [4] NCCL INFO Channel 05/0 : 12[4] -> 15[7] via P2P/IPC
+ip-26-0-173-121:3987797:3987954 [3] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-173-121:3987798:3987955 [4] NCCL INFO Channel 06/0 : 12[4] -> 15[7] via P2P/IPC
+ip-26-0-173-121:3987798:3987955 [4] NCCL INFO Channel 13/0 : 12[4] -> 15[7] via P2P/IPC
+ip-26-0-173-121:3987798:3987955 [4] NCCL INFO Channel 14/0 : 12[4] -> 15[7] via P2P/IPC
+ip-26-0-172-142:789255:789416 [0] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-172-142:789260:789419 [5] NCCL INFO Channel 00/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-142:789261:789423 [6] NCCL INFO Channel 00/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-142:789260:789419 [5] NCCL INFO Channel 08/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-142:789261:789423 [6] NCCL INFO Channel 08/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-142:789256:789421 [1] NCCL INFO Channel 00/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-142:789259:789417 [4] NCCL INFO Channel 00/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-142:789258:789420 [3] NCCL INFO Channel 00/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-142:789256:789421 [1] NCCL INFO Channel 08/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-142:789257:789422 [2] NCCL INFO Channel 00/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-142:789259:789417 [4] NCCL INFO Channel 08/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-142:789258:789420 [3] NCCL INFO Channel 08/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-142:789257:789422 [2] NCCL INFO Channel 08/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-142:789259:789417 [4] NCCL INFO Channel 05/0 : 4[4] -> 7[7] via P2P/IPC
+ip-26-0-172-142:789258:789420 [3] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-172-142:789257:789422 [2] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-172-142:789259:789417 [4] NCCL INFO Channel 06/0 : 4[4] -> 7[7] via P2P/IPC
+ip-26-0-172-142:789259:789417 [4] NCCL INFO Channel 13/0 : 4[4] -> 7[7] via P2P/IPC
+ip-26-0-172-142:789259:789417 [4] NCCL INFO Channel 14/0 : 4[4] -> 7[7] via P2P/IPC
+ip-26-0-172-142:789255:789416 [0] NCCL INFO Channel 00/0 : 15[7] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:789255:789416 [0] NCCL INFO Channel 08/0 : 15[7] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:789255:789416 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-142:789262:789418 [7] NCCL INFO Channel 00/0 : 7[7] -> 8[0] [send] via NET/Libfabric/0(0)/GDRDMA
+ip-26-0-172-142:789255:789416 [0] NCCL INFO Channel 08/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-142:789262:789418 [7] NCCL INFO Channel 08/0 : 7[7] -> 8[0] [send] via NET/Libfabric/0(0)/GDRDMA
+ip-26-0-172-142:789256:789421 [1] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-173-121:3987794:3987956 [0] NCCL INFO Channel 00/0 : 7[7] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-121:3987801:3987960 [7] NCCL INFO Channel 08/0 : 15[7] -> 0[0] [send] via NET/Libfabric/0(8)/GDRDMA
+ip-26-0-173-121:3987794:3987956 [0] NCCL INFO Channel 08/0 : 7[7] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-121:3987794:3987956 [0] NCCL INFO Channel 00/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-172-142:789255:789416 [0] NCCL INFO Channel 05/0 : 0[0] -> 4[4] via P2P/IPC
+ip-26-0-172-142:789255:789416 [0] NCCL INFO Channel 13/0 : 0[0] -> 4[4] via P2P/IPC
+ip-26-0-173-121:3987794:3987956 [0] NCCL INFO Channel 08/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-173-121:3987795:3987961 [1] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-173-121:3987794:3987956 [0] NCCL INFO Channel 05/0 : 8[0] -> 12[4] via P2P/IPC
+ip-26-0-173-121:3987794:3987956 [0] NCCL INFO Channel 13/0 : 8[0] -> 12[4] via P2P/IPC
+ip-26-0-173-121:3987794:3987956 [0] NCCL INFO Channel 06/0 : 8[0] -> 13[5] via P2P/IPC
+ip-26-0-173-121:3987798:3987955 [4] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-172-142:789255:789416 [0] NCCL INFO Channel 06/0 : 0[0] -> 5[5] via P2P/IPC
+ip-26-0-173-121:3987794:3987956 [0] NCCL INFO Channel 14/0 : 8[0] -> 13[5] via P2P/IPC
+ip-26-0-172-142:789259:789417 [4] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-172-142:789255:789416 [0] NCCL INFO Channel 14/0 : 0[0] -> 5[5] via P2P/IPC
+ip-26-0-173-121:3987794:3987956 [0] NCCL INFO Channel 07/0 : 8[0] -> 14[6] via P2P/IPC
+ip-26-0-172-142:789255:789416 [0] NCCL INFO Channel 07/0 : 0[0] -> 6[6] via P2P/IPC
+ip-26-0-173-121:3987794:3987956 [0] NCCL INFO Channel 15/0 : 8[0] -> 14[6] via P2P/IPC
+ip-26-0-172-142:789260:789419 [5] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-172-142:789257:789422 [2] NCCL INFO Channel 02/0 : 11[3] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:789255:789416 [0] NCCL INFO Channel 15/0 : 0[0] -> 6[6] via P2P/IPC
+ip-26-0-172-142:789257:789422 [2] NCCL INFO Channel 10/0 : 11[3] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-121:3987794:3987956 [0] NCCL INFO Channel 01/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-173-121:3987794:3987956 [0] NCCL INFO Channel 02/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-172-142:789261:789423 [6] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-172-142:789255:789416 [0] NCCL INFO Channel 01/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-172-142:789258:789420 [3] NCCL INFO Channel 03/0 : 12[4] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:789258:789420 [3] NCCL INFO Channel 11/0 : 12[4] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-121:3987794:3987956 [0] NCCL INFO Channel 03/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-172-142:789255:789416 [0] NCCL INFO Channel 02/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-172-142:789258:789420 [3] NCCL INFO Channel 02/0 : 3[3] -> 10[2] [send] via NET/Libfabric/2(2)/GDRDMA
+ip-26-0-173-121:3987797:3987954 [3] NCCL INFO Channel 03/0 : 4[4] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:789255:789416 [0] NCCL INFO Channel 03/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-172-142:789258:789420 [3] NCCL INFO Channel 10/0 : 3[3] -> 10[2] [send] via NET/Libfabric/2(2)/GDRDMA
+ip-26-0-173-121:3987794:3987956 [0] NCCL INFO Channel 04/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-173-121:3987797:3987954 [3] NCCL INFO Channel 11/0 : 4[4] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:789255:789416 [0] NCCL INFO Channel 04/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-173-121:3987794:3987956 [0] NCCL INFO Channel 09/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-172-142:789255:789416 [0] NCCL INFO Channel 09/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-173-121:3987796:3987957 [2] NCCL INFO Channel 02/0 : 3[3] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-121:3987796:3987957 [2] NCCL INFO Channel 10/0 : 3[3] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-121:3987797:3987954 [3] NCCL INFO Channel 02/0 : 11[3] -> 2[2] [send] via NET/Libfabric/2(10)/GDRDMA
+ip-26-0-173-121:3987797:3987954 [3] NCCL INFO Channel 10/0 : 11[3] -> 2[2] [send] via NET/Libfabric/2(10)/GDRDMA
+ip-26-0-172-142:789255:789416 [0] NCCL INFO Channel 10/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-173-121:3987794:3987956 [0] NCCL INFO Channel 10/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-173-121:3987796:3987957 [2] NCCL INFO Channel 01/0 : 10[2] -> 1[1] [send] via NET/Libfabric/1(9)/GDRDMA
+ip-26-0-173-121:3987796:3987957 [2] NCCL INFO Channel 09/0 : 10[2] -> 1[1] [send] via NET/Libfabric/1(9)/GDRDMA
+ip-26-0-172-142:789262:789418 [7] NCCL INFO Channel 06/0 : 7[7] -> 14[6] [send] via NET/Libfabric/6(6)/GDRDMA
+ip-26-0-173-121:3987794:3987956 [0] NCCL INFO Channel 11/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-172-142:789255:789416 [0] NCCL INFO Channel 11/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-173-121:3987801:3987960 [7] NCCL INFO Channel 06/0 : 15[7] -> 6[6] [send] via NET/Libfabric/6(14)/GDRDMA
+ip-26-0-172-142:789255:789416 [0] NCCL INFO Channel 12/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-173-121:3987801:3987960 [7] NCCL INFO Channel 14/0 : 15[7] -> 6[6] [send] via NET/Libfabric/6(14)/GDRDMA
+ip-26-0-173-121:3987799:3987958 [5] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-173-121:3987794:3987956 [0] NCCL INFO Channel 12/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-173-121:3987800:3987959 [6] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-172-142:789256:789421 [1] NCCL INFO Channel 01/0 : 10[2] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:789256:789421 [1] NCCL INFO Channel 09/0 : 10[2] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:789257:789422 [2] NCCL INFO Channel 01/0 : 2[2] -> 9[1] [send] via NET/Libfabric/1(1)/GDRDMA
+ip-26-0-172-142:789257:789422 [2] NCCL INFO Channel 09/0 : 2[2] -> 9[1] [send] via NET/Libfabric/1(1)/GDRDMA
+ip-26-0-172-142:789256:789421 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789256:789421 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789256:789421 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987796:3987957 [2] NCCL INFO Channel 02/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-172-142:789256:789421 [1] NCCL INFO Channel 04/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987796:3987957 [2] NCCL INFO Channel 03/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-172-142:789256:789421 [1] NCCL INFO Channel 05/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987796:3987957 [2] NCCL INFO Channel 04/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-172-142:789256:789421 [1] NCCL INFO Channel 06/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987796:3987957 [2] NCCL INFO Channel 05/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-172-142:789256:789421 [1] NCCL INFO Channel 07/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789256:789421 [1] NCCL INFO Channel 09/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987796:3987957 [2] NCCL INFO Channel 06/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-172-142:789256:789421 [1] NCCL INFO Channel 10/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987796:3987957 [2] NCCL INFO Channel 07/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-172-142:789256:789421 [1] NCCL INFO Channel 11/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987796:3987957 [2] NCCL INFO Channel 10/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-172-142:789256:789421 [1] NCCL INFO Channel 12/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987796:3987957 [2] NCCL INFO Channel 11/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-172-142:789256:789421 [1] NCCL INFO Channel 13/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987796:3987957 [2] NCCL INFO Channel 12/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-172-142:789256:789421 [1] NCCL INFO Channel 14/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987796:3987957 [2] NCCL INFO Channel 13/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-172-142:789256:789421 [1] NCCL INFO Channel 15/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789259:789417 [4] NCCL INFO Channel 04/0 : 13[5] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-121:3987796:3987957 [2] NCCL INFO Channel 14/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-172-142:789259:789417 [4] NCCL INFO Channel 12/0 : 13[5] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:789259:789417 [4] NCCL INFO Channel 03/0 : 4[4] -> 11[3] [send] via NET/Libfabric/3(3)/GDRDMA
+ip-26-0-173-121:3987796:3987957 [2] NCCL INFO Channel 15/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-172-142:789259:789417 [4] NCCL INFO Channel 11/0 : 4[4] -> 11[3] [send] via NET/Libfabric/3(3)/GDRDMA
+ip-26-0-172-142:789260:789419 [5] NCCL INFO Channel 05/0 : 14[6] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:789260:789419 [5] NCCL INFO Channel 13/0 : 14[6] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:789261:789423 [6] NCCL INFO Channel 06/0 : 15[7] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:789262:789418 [7] NCCL INFO Channel 14/0 : 7[7] -> 14[6] [send] via NET/Libfabric/6(6)/GDRDMA
+ip-26-0-172-142:789260:789419 [5] NCCL INFO Channel 04/0 : 5[5] -> 12[4] [send] via NET/Libfabric/4(4)/GDRDMA
+ip-26-0-172-142:789261:789423 [6] NCCL INFO Channel 14/0 : 15[7] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:789260:789419 [5] NCCL INFO Channel 12/0 : 5[5] -> 12[4] [send] via NET/Libfabric/4(4)/GDRDMA
+ip-26-0-172-142:789261:789423 [6] NCCL INFO Channel 05/0 : 6[6] -> 13[5] [send] via NET/Libfabric/5(5)/GDRDMA
+ip-26-0-172-142:789261:789423 [6] NCCL INFO Channel 13/0 : 6[6] -> 13[5] [send] via NET/Libfabric/5(5)/GDRDMA
+ip-26-0-173-121:3987801:3987960 [7] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-173-121:3987799:3987958 [5] NCCL INFO Channel 05/0 : 6[6] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-121:3987799:3987958 [5] NCCL INFO Channel 13/0 : 6[6] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-121:3987795:3987961 [1] NCCL INFO Channel 01/0 : 2[2] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-121:3987795:3987961 [1] NCCL INFO Channel 09/0 : 2[2] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-121:3987800:3987959 [6] NCCL INFO Channel 06/0 : 7[7] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-121:3987800:3987959 [6] NCCL INFO Channel 14/0 : 7[7] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-121:3987800:3987959 [6] NCCL INFO Channel 05/0 : 14[6] -> 5[5] [send] via NET/Libfabric/5(13)/GDRDMA
+ip-26-0-173-121:3987800:3987959 [6] NCCL INFO Channel 13/0 : 14[6] -> 5[5] [send] via NET/Libfabric/5(13)/GDRDMA
+ip-26-0-172-142:789262:789418 [7] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-173-121:3987800:3987959 [6] NCCL INFO Channel 06/0 : 14[6] -> 11[3] via P2P/IPC
+ip-26-0-173-121:3987800:3987959 [6] NCCL INFO Channel 14/0 : 14[6] -> 11[3] via P2P/IPC
+ip-26-0-173-121:3987798:3987955 [4] NCCL INFO Channel 04/0 : 5[5] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-121:3987799:3987958 [5] NCCL INFO Channel 04/0 : 13[5] -> 4[4] [send] via NET/Libfabric/4(12)/GDRDMA
+ip-26-0-173-121:3987798:3987955 [4] NCCL INFO Channel 12/0 : 5[5] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-121:3987799:3987958 [5] NCCL INFO Channel 12/0 : 13[5] -> 4[4] [send] via NET/Libfabric/4(12)/GDRDMA
+ip-26-0-173-121:3987798:3987955 [4] NCCL INFO Channel 03/0 : 12[4] -> 3[3] [send] via NET/Libfabric/3(11)/GDRDMA
+ip-26-0-172-142:789261:789423 [6] NCCL INFO Channel 06/0 : 6[6] -> 3[3] via P2P/IPC
+ip-26-0-173-121:3987798:3987955 [4] NCCL INFO Channel 11/0 : 12[4] -> 3[3] [send] via NET/Libfabric/3(11)/GDRDMA
+ip-26-0-173-121:3987799:3987958 [5] NCCL INFO Channel 05/0 : 13[5] -> 11[3] via P2P/IPC
+ip-26-0-172-142:789260:789419 [5] NCCL INFO Channel 05/0 : 5[5] -> 3[3] via P2P/IPC
+ip-26-0-173-121:3987799:3987958 [5] NCCL INFO Channel 13/0 : 13[5] -> 11[3] via P2P/IPC
+ip-26-0-173-121:3987795:3987961 [1] NCCL INFO Channel 01/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-173-121:3987795:3987961 [1] NCCL INFO Channel 02/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-173-121:3987795:3987961 [1] NCCL INFO Channel 03/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-172-142:789261:789423 [6] NCCL INFO Channel 14/0 : 6[6] -> 3[3] via P2P/IPC
+ip-26-0-173-121:3987795:3987961 [1] NCCL INFO Channel 04/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-172-142:789260:789419 [5] NCCL INFO Channel 13/0 : 5[5] -> 3[3] via P2P/IPC
+ip-26-0-173-121:3987795:3987961 [1] NCCL INFO Channel 05/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-173-121:3987795:3987961 [1] NCCL INFO Channel 06/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-173-121:3987795:3987961 [1] NCCL INFO Channel 07/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-173-121:3987795:3987961 [1] NCCL INFO Channel 09/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-173-121:3987795:3987961 [1] NCCL INFO Channel 10/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-173-121:3987795:3987961 [1] NCCL INFO Channel 11/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-173-121:3987795:3987961 [1] NCCL INFO Channel 12/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-173-121:3987795:3987961 [1] NCCL INFO Channel 13/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-173-121:3987795:3987961 [1] NCCL INFO Channel 14/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-173-121:3987795:3987961 [1] NCCL INFO Channel 15/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-172-142:789257:789422 [2] NCCL INFO Channel 02/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-142:789257:789422 [2] NCCL INFO Channel 03/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-142:789257:789422 [2] NCCL INFO Channel 04/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-142:789257:789422 [2] NCCL INFO Channel 05/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-142:789257:789422 [2] NCCL INFO Channel 06/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-142:789257:789422 [2] NCCL INFO Channel 07/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-142:789257:789422 [2] NCCL INFO Channel 10/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-142:789257:789422 [2] NCCL INFO Channel 11/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-142:789257:789422 [2] NCCL INFO Channel 12/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-142:789257:789422 [2] NCCL INFO Channel 13/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-142:789257:789422 [2] NCCL INFO Channel 14/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-142:789257:789422 [2] NCCL INFO Channel 15/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-173-121:3987801:3987960 [7] NCCL INFO Channel 07/0 : 4[4] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-121:3987801:3987960 [7] NCCL INFO Channel 15/0 : 4[4] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-121:3987798:3987955 [4] NCCL INFO Channel 07/0 : 12[4] -> 7[7] [send] via NET/Libfabric/7(15)/GDRDMA
+ip-26-0-173-121:3987798:3987955 [4] NCCL INFO Channel 15/0 : 12[4] -> 7[7] [send] via NET/Libfabric/7(15)/GDRDMA
+ip-26-0-172-142:789262:789418 [7] NCCL INFO Channel 07/0 : 12[4] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:789262:789418 [7] NCCL INFO Channel 15/0 : 12[4] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:789259:789417 [4] NCCL INFO Channel 07/0 : 4[4] -> 15[7] [send] via NET/Libfabric/7(7)/GDRDMA
+ip-26-0-172-142:789259:789417 [4] NCCL INFO Channel 15/0 : 4[4] -> 15[7] [send] via NET/Libfabric/7(7)/GDRDMA
+ip-26-0-172-142:789262:789418 [7] NCCL INFO Channel 07/0 : 7[7] -> 3[3] via P2P/IPC
+ip-26-0-173-121:3987801:3987960 [7] NCCL INFO Channel 07/0 : 15[7] -> 11[3] via P2P/IPC
+ip-26-0-172-142:789262:789418 [7] NCCL INFO Channel 15/0 : 7[7] -> 3[3] via P2P/IPC
+ip-26-0-173-121:3987801:3987960 [7] NCCL INFO Channel 15/0 : 15[7] -> 11[3] via P2P/IPC
+ip-26-0-172-142:789262:789418 [7] NCCL INFO Channel 01/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-142:789262:789418 [7] NCCL INFO Channel 02/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-173-121:3987801:3987960 [7] NCCL INFO Channel 01/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-172-142:789262:789418 [7] NCCL INFO Channel 03/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-142:789262:789418 [7] NCCL INFO Channel 04/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-173-121:3987801:3987960 [7] NCCL INFO Channel 02/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-172-142:789262:789418 [7] NCCL INFO Channel 05/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-142:789262:789418 [7] NCCL INFO Channel 09/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-173-121:3987801:3987960 [7] NCCL INFO Channel 03/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-172-142:789262:789418 [7] NCCL INFO Channel 10/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-142:789262:789418 [7] NCCL INFO Channel 11/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-142:789259:789417 [4] NCCL INFO Channel 01/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-173-121:3987801:3987960 [7] NCCL INFO Channel 04/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-172-142:789262:789418 [7] NCCL INFO Channel 12/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-142:789259:789417 [4] NCCL INFO Channel 02/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-142:789262:789418 [7] NCCL INFO Channel 13/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-142:789259:789417 [4] NCCL INFO Channel 04/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-142:789259:789417 [4] NCCL INFO Channel 09/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-142:789259:789417 [4] NCCL INFO Channel 10/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-142:789259:789417 [4] NCCL INFO Channel 12/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-142:789258:789420 [3] NCCL INFO Channel 01/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-142:789258:789420 [3] NCCL INFO Channel 03/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-173-121:3987801:3987960 [7] NCCL INFO Channel 05/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-172-142:789258:789420 [3] NCCL INFO Channel 04/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-142:789258:789420 [3] NCCL INFO Channel 05/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-142:789258:789420 [3] NCCL INFO Channel 06/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-142:789261:789423 [6] NCCL INFO Channel 01/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-142:789258:789420 [3] NCCL INFO Channel 07/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-142:789261:789423 [6] NCCL INFO Channel 02/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-142:789260:789419 [5] NCCL INFO Channel 01/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-173-121:3987801:3987960 [7] NCCL INFO Channel 09/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-172-142:789258:789420 [3] NCCL INFO Channel 09/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-142:789261:789423 [6] NCCL INFO Channel 03/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-142:789260:789419 [5] NCCL INFO Channel 02/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-173-121:3987801:3987960 [7] NCCL INFO Channel 10/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-172-142:789258:789420 [3] NCCL INFO Channel 11/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-142:789260:789419 [5] NCCL INFO Channel 03/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-142:789261:789423 [6] NCCL INFO Channel 04/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-173-121:3987801:3987960 [7] NCCL INFO Channel 11/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-172-142:789258:789420 [3] NCCL INFO Channel 12/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-173-121:3987801:3987960 [7] NCCL INFO Channel 12/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-172-142:789260:789419 [5] NCCL INFO Channel 06/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-142:789261:789423 [6] NCCL INFO Channel 07/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-173-121:3987798:3987955 [4] NCCL INFO Channel 01/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-172-142:789258:789420 [3] NCCL INFO Channel 13/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-173-121:3987801:3987960 [7] NCCL INFO Channel 13/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-172-142:789261:789423 [6] NCCL INFO Channel 09/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-142:789260:789419 [5] NCCL INFO Channel 07/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-142:789258:789420 [3] NCCL INFO Channel 14/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-173-121:3987798:3987955 [4] NCCL INFO Channel 02/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-172-142:789261:789423 [6] NCCL INFO Channel 10/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-142:789260:789419 [5] NCCL INFO Channel 09/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-142:789258:789420 [3] NCCL INFO Channel 15/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-142:789261:789423 [6] NCCL INFO Channel 11/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-142:789260:789419 [5] NCCL INFO Channel 10/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-173-121:3987798:3987955 [4] NCCL INFO Channel 04/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-172-142:789260:789419 [5] NCCL INFO Channel 11/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-142:789261:789423 [6] NCCL INFO Channel 12/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-173-121:3987798:3987955 [4] NCCL INFO Channel 09/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-173-121:3987798:3987955 [4] NCCL INFO Channel 10/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-173-121:3987798:3987955 [4] NCCL INFO Channel 12/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-172-142:789261:789423 [6] NCCL INFO Channel 15/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-173-121:3987797:3987954 [3] NCCL INFO Channel 01/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-173-121:3987797:3987954 [3] NCCL INFO Channel 03/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-173-121:3987800:3987959 [6] NCCL INFO Channel 01/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-173-121:3987797:3987954 [3] NCCL INFO Channel 04/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-173-121:3987797:3987954 [3] NCCL INFO Channel 05/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-173-121:3987800:3987959 [6] NCCL INFO Channel 02/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-173-121:3987797:3987954 [3] NCCL INFO Channel 06/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-173-121:3987800:3987959 [6] NCCL INFO Channel 03/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-173-121:3987797:3987954 [3] NCCL INFO Channel 07/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-173-121:3987800:3987959 [6] NCCL INFO Channel 04/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-173-121:3987797:3987954 [3] NCCL INFO Channel 09/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-173-121:3987800:3987959 [6] NCCL INFO Channel 07/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-173-121:3987797:3987954 [3] NCCL INFO Channel 11/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-173-121:3987800:3987959 [6] NCCL INFO Channel 09/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-173-121:3987799:3987958 [5] NCCL INFO Channel 01/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-173-121:3987797:3987954 [3] NCCL INFO Channel 12/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-173-121:3987800:3987959 [6] NCCL INFO Channel 10/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-173-121:3987799:3987958 [5] NCCL INFO Channel 02/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-173-121:3987797:3987954 [3] NCCL INFO Channel 13/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-173-121:3987800:3987959 [6] NCCL INFO Channel 11/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-173-121:3987799:3987958 [5] NCCL INFO Channel 03/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-173-121:3987797:3987954 [3] NCCL INFO Channel 14/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-173-121:3987799:3987958 [5] NCCL INFO Channel 06/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-173-121:3987800:3987959 [6] NCCL INFO Channel 12/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-173-121:3987797:3987954 [3] NCCL INFO Channel 15/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-173-121:3987799:3987958 [5] NCCL INFO Channel 07/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-173-121:3987800:3987959 [6] NCCL INFO Channel 15/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-173-121:3987799:3987958 [5] NCCL INFO Channel 09/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-172-142:789260:789419 [5] NCCL INFO Channel 14/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-173-121:3987799:3987958 [5] NCCL INFO Channel 10/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-172-142:789260:789419 [5] NCCL INFO Channel 15/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-173-121:3987799:3987958 [5] NCCL INFO Channel 11/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-173-121:3987799:3987958 [5] NCCL INFO Channel 14/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-173-121:3987799:3987958 [5] NCCL INFO Channel 15/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-172-142:789258:789420 [3] NCCL INFO Connected all rings
+ip-26-0-172-142:789255:789416 [0] NCCL INFO Connected all rings
+ip-26-0-172-142:789255:789416 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-142:789255:789416 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-142:789255:789416 [0] NCCL INFO Channel 04/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-142:789255:789416 [0] NCCL INFO Channel 05/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-142:789255:789416 [0] NCCL INFO Channel 06/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-142:789255:789416 [0] NCCL INFO Channel 07/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-142:789255:789416 [0] NCCL INFO Channel 10/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-142:789258:789420 [3] NCCL INFO Channel 01/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-142:789255:789416 [0] NCCL INFO Channel 11/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-142:789258:789420 [3] NCCL INFO Channel 02/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-142:789255:789416 [0] NCCL INFO Channel 12/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-142:789258:789420 [3] NCCL INFO Channel 03/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-142:789255:789416 [0] NCCL INFO Channel 13/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-142:789258:789420 [3] NCCL INFO Channel 05/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-142:789255:789416 [0] NCCL INFO Channel 14/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-142:789255:789416 [0] NCCL INFO Channel 15/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-142:789257:789422 [2] NCCL INFO Connected all rings
+ip-26-0-172-142:789261:789423 [6] NCCL INFO Connected all rings
+ip-26-0-173-121:3987795:3987961 [1] NCCL INFO Connected all rings
+ip-26-0-172-142:789257:789422 [2] NCCL INFO Channel 01/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-142:789257:789422 [2] NCCL INFO Channel 02/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-142:789262:789418 [7] NCCL INFO Connected all rings
+ip-26-0-172-142:789257:789422 [2] NCCL INFO Channel 04/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-142:789257:789422 [2] NCCL INFO Channel 05/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-142:789257:789422 [2] NCCL INFO Channel 06/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-142:789257:789422 [2] NCCL INFO Channel 07/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-142:789257:789422 [2] NCCL INFO Channel 09/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-142:789257:789422 [2] NCCL INFO Channel 10/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-142:789257:789422 [2] NCCL INFO Channel 12/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-142:789257:789422 [2] NCCL INFO Channel 13/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-142:789261:789423 [6] NCCL INFO Channel 01/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-173-121:3987794:3987956 [0] NCCL INFO Connected all rings
+ip-26-0-173-121:3987794:3987956 [0] NCCL INFO Channel 02/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-172-142:789257:789422 [2] NCCL INFO Channel 14/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-142:789261:789423 [6] NCCL INFO Channel 02/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-142:789257:789422 [2] NCCL INFO Channel 15/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-142:789261:789423 [6] NCCL INFO Channel 03/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-142:789259:789417 [4] NCCL INFO Connected all rings
+ip-26-0-172-142:789261:789423 [6] NCCL INFO Channel 04/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-142:789258:789420 [3] NCCL INFO Channel 06/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-142:789261:789423 [6] NCCL INFO Channel 05/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-142:789258:789420 [3] NCCL INFO Channel 07/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-142:789261:789423 [6] NCCL INFO Channel 06/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-142:789258:789420 [3] NCCL INFO Channel 09/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-142:789261:789423 [6] NCCL INFO Channel 09/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-142:789258:789420 [3] NCCL INFO Channel 10/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-142:789261:789423 [6] NCCL INFO Channel 10/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-142:789258:789420 [3] NCCL INFO Channel 11/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-142:789261:789423 [6] NCCL INFO Channel 11/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-142:789260:789419 [5] NCCL INFO Connected all rings
+ip-26-0-172-142:789258:789420 [3] NCCL INFO Channel 13/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-142:789261:789423 [6] NCCL INFO Channel 12/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-142:789258:789420 [3] NCCL INFO Channel 14/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-142:789261:789423 [6] NCCL INFO Channel 13/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-142:789258:789420 [3] NCCL INFO Channel 15/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-142:789261:789423 [6] NCCL INFO Channel 14/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-142:789259:789417 [4] NCCL INFO Channel 01/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-142:789259:789417 [4] NCCL INFO Channel 02/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-142:789260:789419 [5] NCCL INFO Channel 01/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-142:789259:789417 [4] NCCL INFO Channel 03/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-142:789260:789419 [5] NCCL INFO Channel 02/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-142:789259:789417 [4] NCCL INFO Channel 04/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-142:789260:789419 [5] NCCL INFO Channel 03/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-142:789259:789417 [4] NCCL INFO Channel 06/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-142:789260:789419 [5] NCCL INFO Channel 04/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-142:789259:789417 [4] NCCL INFO Channel 07/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-142:789260:789419 [5] NCCL INFO Channel 05/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-142:789259:789417 [4] NCCL INFO Channel 09/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-142:789260:789419 [5] NCCL INFO Channel 07/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-142:789259:789417 [4] NCCL INFO Channel 10/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-142:789260:789419 [5] NCCL INFO Channel 09/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-142:789259:789417 [4] NCCL INFO Channel 11/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-142:789260:789419 [5] NCCL INFO Channel 10/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-142:789259:789417 [4] NCCL INFO Channel 12/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-142:789260:789419 [5] NCCL INFO Channel 11/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-142:789259:789417 [4] NCCL INFO Channel 14/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-142:789260:789419 [5] NCCL INFO Channel 12/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-142:789259:789417 [4] NCCL INFO Channel 15/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-142:789260:789419 [5] NCCL INFO Channel 13/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-142:789260:789419 [5] NCCL INFO Channel 15/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-142:789258:789420 [3] NCCL INFO Channel 03/0 : 11[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:789258:789420 [3] NCCL INFO Channel 11/0 : 11[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:789258:789420 [3] NCCL INFO Channel 03/0 : 3[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:789258:789420 [3] NCCL INFO Channel 11/0 : 3[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:789259:789417 [4] NCCL INFO Channel 04/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:789261:789423 [6] NCCL INFO Channel 06/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:789259:789417 [4] NCCL INFO Channel 12/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:789260:789419 [5] NCCL INFO Channel 05/0 : 13[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:789261:789423 [6] NCCL INFO Channel 14/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:789259:789417 [4] NCCL INFO Channel 04/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:789260:789419 [5] NCCL INFO Channel 13/0 : 13[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:789259:789417 [4] NCCL INFO Channel 12/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:789261:789423 [6] NCCL INFO Channel 06/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:789260:789419 [5] NCCL INFO Channel 05/0 : 5[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:789261:789423 [6] NCCL INFO Channel 14/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:789260:789419 [5] NCCL INFO Channel 13/0 : 5[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-121:3987794:3987956 [0] NCCL INFO Channel 03/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-173-121:3987794:3987956 [0] NCCL INFO Channel 04/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-173-121:3987794:3987956 [0] NCCL INFO Channel 05/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-172-142:789256:789421 [1] NCCL INFO Connected all rings
+ip-26-0-173-121:3987794:3987956 [0] NCCL INFO Channel 06/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-173-121:3987794:3987956 [0] NCCL INFO Channel 07/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-172-142:789256:789421 [1] NCCL INFO Channel 01/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-173-121:3987794:3987956 [0] NCCL INFO Channel 10/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-172-142:789256:789421 [1] NCCL INFO Channel 03/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-142:789256:789421 [1] NCCL INFO Channel 04/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-142:789256:789421 [1] NCCL INFO Channel 05/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-142:789256:789421 [1] NCCL INFO Channel 06/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-173-121:3987794:3987956 [0] NCCL INFO Channel 11/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-172-142:789256:789421 [1] NCCL INFO Channel 07/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-142:789256:789421 [1] NCCL INFO Channel 09/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-142:789256:789421 [1] NCCL INFO Channel 11/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-142:789256:789421 [1] NCCL INFO Channel 12/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-142:789256:789421 [1] NCCL INFO Channel 13/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-142:789256:789421 [1] NCCL INFO Channel 14/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-173-121:3987794:3987956 [0] NCCL INFO Channel 12/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-172-142:789256:789421 [1] NCCL INFO Channel 15/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-142:789255:789416 [0] NCCL INFO Channel 05/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-172-142:789256:789421 [1] NCCL INFO Channel 01/0 : 9[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:789256:789421 [1] NCCL INFO Channel 09/0 : 9[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:789255:789416 [0] NCCL INFO Channel 06/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-172-142:789256:789421 [1] NCCL INFO Channel 01/0 : 1[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:789256:789421 [1] NCCL INFO Channel 09/0 : 1[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-121:3987794:3987956 [0] NCCL INFO Channel 13/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-172-142:789255:789416 [0] NCCL INFO Channel 07/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-172-142:789255:789416 [0] NCCL INFO Channel 13/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-172-142:789255:789416 [0] NCCL INFO Channel 14/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-172-142:789255:789416 [0] NCCL INFO Channel 15/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-172-142:789257:789422 [2] NCCL INFO Channel 02/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:789257:789422 [2] NCCL INFO Channel 10/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-121:3987794:3987956 [0] NCCL INFO Channel 14/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-172-142:789257:789422 [2] NCCL INFO Channel 02/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:789257:789422 [2] NCCL INFO Channel 10/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:789262:789418 [7] NCCL INFO Channel 07/0 : 15[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:789255:789416 [0] NCCL INFO Channel 00/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:789255:789416 [0] NCCL INFO Channel 08/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:789262:789418 [7] NCCL INFO Channel 15/0 : 15[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:789255:789416 [0] NCCL INFO Channel 00/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:789262:789418 [7] NCCL INFO Channel 07/0 : 7[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:789255:789416 [0] NCCL INFO Channel 08/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:789262:789418 [7] NCCL INFO Channel 15/0 : 7[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-121:3987794:3987956 [0] NCCL INFO Channel 15/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-173-121:3987798:3987955 [4] NCCL INFO Connected all rings
+ip-26-0-173-121:3987799:3987958 [5] NCCL INFO Connected all rings
+ip-26-0-173-121:3987796:3987957 [2] NCCL INFO Connected all rings
+ip-26-0-173-121:3987801:3987960 [7] NCCL INFO Connected all rings
+ip-26-0-173-121:3987797:3987954 [3] NCCL INFO Connected all rings
+ip-26-0-173-121:3987800:3987959 [6] NCCL INFO Connected all rings
+ip-26-0-173-121:3987795:3987961 [1] NCCL INFO Channel 01/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-173-121:3987795:3987961 [1] NCCL INFO Channel 03/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-173-121:3987795:3987961 [1] NCCL INFO Channel 04/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-173-121:3987795:3987961 [1] NCCL INFO Channel 05/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-173-121:3987795:3987961 [1] NCCL INFO Channel 06/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-173-121:3987795:3987961 [1] NCCL INFO Channel 07/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-173-121:3987795:3987961 [1] NCCL INFO Channel 09/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-173-121:3987796:3987957 [2] NCCL INFO Channel 01/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-173-121:3987795:3987961 [1] NCCL INFO Channel 11/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-173-121:3987796:3987957 [2] NCCL INFO Channel 02/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-173-121:3987795:3987961 [1] NCCL INFO Channel 12/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-173-121:3987798:3987955 [4] NCCL INFO Channel 01/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-173-121:3987796:3987957 [2] NCCL INFO Channel 04/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-173-121:3987795:3987961 [1] NCCL INFO Channel 13/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-173-121:3987798:3987955 [4] NCCL INFO Channel 02/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-173-121:3987796:3987957 [2] NCCL INFO Channel 05/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-173-121:3987795:3987961 [1] NCCL INFO Channel 14/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-173-121:3987799:3987958 [5] NCCL INFO Channel 01/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-173-121:3987798:3987955 [4] NCCL INFO Channel 03/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-173-121:3987796:3987957 [2] NCCL INFO Channel 06/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-173-121:3987795:3987961 [1] NCCL INFO Channel 15/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-173-121:3987799:3987958 [5] NCCL INFO Channel 02/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-173-121:3987798:3987955 [4] NCCL INFO Channel 04/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-173-121:3987796:3987957 [2] NCCL INFO Channel 07/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-173-121:3987800:3987959 [6] NCCL INFO Channel 01/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-173-121:3987794:3987956 [0] NCCL INFO Channel 05/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-173-121:3987799:3987958 [5] NCCL INFO Channel 03/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-173-121:3987798:3987955 [4] NCCL INFO Channel 06/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-173-121:3987796:3987957 [2] NCCL INFO Channel 09/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-173-121:3987800:3987959 [6] NCCL INFO Channel 02/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-173-121:3987794:3987956 [0] NCCL INFO Channel 06/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-173-121:3987799:3987958 [5] NCCL INFO Channel 04/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-173-121:3987798:3987955 [4] NCCL INFO Channel 07/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-173-121:3987796:3987957 [2] NCCL INFO Channel 10/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-173-121:3987800:3987959 [6] NCCL INFO Channel 03/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-173-121:3987794:3987956 [0] NCCL INFO Channel 07/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-173-121:3987799:3987958 [5] NCCL INFO Channel 05/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-173-121:3987798:3987955 [4] NCCL INFO Channel 09/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-173-121:3987796:3987957 [2] NCCL INFO Channel 12/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-173-121:3987800:3987959 [6] NCCL INFO Channel 04/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-173-121:3987794:3987956 [0] NCCL INFO Channel 13/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-173-121:3987799:3987958 [5] NCCL INFO Channel 07/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-173-121:3987798:3987955 [4] NCCL INFO Channel 10/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-173-121:3987796:3987957 [2] NCCL INFO Channel 13/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-173-121:3987800:3987959 [6] NCCL INFO Channel 05/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-173-121:3987794:3987956 [0] NCCL INFO Channel 14/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-173-121:3987799:3987958 [5] NCCL INFO Channel 09/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-173-121:3987798:3987955 [4] NCCL INFO Channel 11/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-173-121:3987796:3987957 [2] NCCL INFO Channel 14/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-173-121:3987800:3987959 [6] NCCL INFO Channel 06/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-173-121:3987794:3987956 [0] NCCL INFO Channel 15/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-173-121:3987799:3987958 [5] NCCL INFO Channel 10/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-173-121:3987798:3987955 [4] NCCL INFO Channel 12/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-173-121:3987796:3987957 [2] NCCL INFO Channel 15/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-173-121:3987800:3987959 [6] NCCL INFO Channel 09/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-173-121:3987799:3987958 [5] NCCL INFO Channel 11/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-173-121:3987798:3987955 [4] NCCL INFO Channel 14/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-173-121:3987797:3987954 [3] NCCL INFO Channel 01/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-173-121:3987800:3987959 [6] NCCL INFO Channel 10/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-173-121:3987795:3987961 [1] NCCL INFO Channel 01/0 : 1[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-121:3987795:3987961 [1] NCCL INFO Channel 09/0 : 1[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-121:3987799:3987958 [5] NCCL INFO Channel 12/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-173-121:3987797:3987954 [3] NCCL INFO Channel 02/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-173-121:3987795:3987961 [1] NCCL INFO Channel 01/0 : 9[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-121:3987798:3987955 [4] NCCL INFO Channel 15/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-173-121:3987795:3987961 [1] NCCL INFO Channel 09/0 : 9[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-121:3987800:3987959 [6] NCCL INFO Channel 11/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-173-121:3987799:3987958 [5] NCCL INFO Channel 13/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-173-121:3987797:3987954 [3] NCCL INFO Channel 03/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-173-121:3987800:3987959 [6] NCCL INFO Channel 12/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-172-142:789256:789421 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987799:3987958 [5] NCCL INFO Channel 15/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-173-121:3987797:3987954 [3] NCCL INFO Channel 05/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-172-142:789256:789421 [1] NCCL INFO Channel 08/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987800:3987959 [6] NCCL INFO Channel 13/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-173-121:3987797:3987954 [3] NCCL INFO Channel 06/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-173-121:3987800:3987959 [6] NCCL INFO Channel 14/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-173-121:3987797:3987954 [3] NCCL INFO Channel 07/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-173-121:3987795:3987961 [1] NCCL INFO Channel 00/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-173-121:3987797:3987954 [3] NCCL INFO Channel 09/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-173-121:3987799:3987958 [5] NCCL INFO Channel 05/0 : 5[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-121:3987800:3987959 [6] NCCL INFO Channel 06/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-121:3987799:3987958 [5] NCCL INFO Channel 13/0 : 5[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-121:3987800:3987959 [6] NCCL INFO Channel 14/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-121:3987799:3987958 [5] NCCL INFO Channel 05/0 : 13[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-121:3987800:3987959 [6] NCCL INFO Channel 06/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-121:3987799:3987958 [5] NCCL INFO Channel 13/0 : 13[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-121:3987795:3987961 [1] NCCL INFO Channel 08/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-173-121:3987800:3987959 [6] NCCL INFO Channel 14/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-121:3987797:3987954 [3] NCCL INFO Channel 10/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-173-121:3987797:3987954 [3] NCCL INFO Channel 11/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-172-142:789260:789419 [5] NCCL INFO Channel 00/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-142:789260:789419 [5] NCCL INFO Channel 04/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-142:789261:789423 [6] NCCL INFO Channel 00/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-142:789260:789419 [5] NCCL INFO Channel 08/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-142:789261:789423 [6] NCCL INFO Channel 05/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-142:789260:789419 [5] NCCL INFO Channel 12/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-142:789261:789423 [6] NCCL INFO Channel 08/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-142:789261:789423 [6] NCCL INFO Channel 13/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-173-121:3987797:3987954 [3] NCCL INFO Channel 13/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-173-121:3987797:3987954 [3] NCCL INFO Channel 14/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-173-121:3987799:3987958 [5] NCCL INFO Channel 00/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-173-121:3987800:3987959 [6] NCCL INFO Channel 00/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-173-121:3987797:3987954 [3] NCCL INFO Channel 15/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-173-121:3987799:3987958 [5] NCCL INFO Channel 04/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-173-121:3987799:3987958 [5] NCCL INFO Channel 08/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-173-121:3987800:3987959 [6] NCCL INFO Channel 05/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-173-121:3987801:3987960 [7] NCCL INFO Channel 07/0 : 7[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-121:3987794:3987956 [0] NCCL INFO Channel 00/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-121:3987801:3987960 [7] NCCL INFO Channel 15/0 : 7[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-121:3987794:3987956 [0] NCCL INFO Channel 08/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-121:3987801:3987960 [7] NCCL INFO Channel 07/0 : 15[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-121:3987794:3987956 [0] NCCL INFO Channel 00/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-121:3987801:3987960 [7] NCCL INFO Channel 15/0 : 15[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-121:3987794:3987956 [0] NCCL INFO Channel 08/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-121:3987796:3987957 [2] NCCL INFO Channel 02/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-121:3987799:3987958 [5] NCCL INFO Channel 12/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-173-121:3987798:3987955 [4] NCCL INFO Channel 04/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-121:3987800:3987959 [6] NCCL INFO Channel 08/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-173-121:3987796:3987957 [2] NCCL INFO Channel 10/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-121:3987797:3987954 [3] NCCL INFO Channel 03/0 : 3[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-121:3987801:3987960 [7] NCCL INFO Channel 01/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-172-142:789262:789418 [7] NCCL INFO Channel 01/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987798:3987955 [4] NCCL INFO Channel 12/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-121:3987796:3987957 [2] NCCL INFO Channel 02/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-121:3987797:3987954 [3] NCCL INFO Channel 11/0 : 3[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-121:3987798:3987955 [4] NCCL INFO Channel 04/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-121:3987796:3987957 [2] NCCL INFO Channel 10/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-121:3987797:3987954 [3] NCCL INFO Channel 03/0 : 11[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-121:3987798:3987955 [4] NCCL INFO Channel 12/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-121:3987797:3987954 [3] NCCL INFO Channel 11/0 : 11[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:789262:789418 [7] NCCL INFO Channel 02/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789262:789418 [7] NCCL INFO Channel 03/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789262:789418 [7] NCCL INFO Channel 04/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789259:789417 [4] NCCL INFO Channel 00/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-142:789262:789418 [7] NCCL INFO Channel 05/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789259:789417 [4] NCCL INFO Channel 03/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-142:789259:789417 [4] NCCL INFO Channel 05/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-173-121:3987801:3987960 [7] NCCL INFO Channel 02/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-172-142:789262:789418 [7] NCCL INFO Channel 06/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987800:3987959 [6] NCCL INFO Channel 13/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-172-142:789257:789422 [2] NCCL INFO Channel 00/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-173-121:3987801:3987960 [7] NCCL INFO Channel 03/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-172-142:789259:789417 [4] NCCL INFO Channel 06/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-142:789262:789418 [7] NCCL INFO Channel 07/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789257:789422 [2] NCCL INFO Channel 01/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-142:789259:789417 [4] NCCL INFO Channel 07/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-173-121:3987801:3987960 [7] NCCL INFO Channel 04/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-172-142:789262:789418 [7] NCCL INFO Channel 09/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789257:789422 [2] NCCL INFO Channel 08/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-142:789259:789417 [4] NCCL INFO Channel 08/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-142:789262:789418 [7] NCCL INFO Channel 10/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789257:789422 [2] NCCL INFO Channel 09/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-142:789259:789417 [4] NCCL INFO Channel 11/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-142:789262:789418 [7] NCCL INFO Channel 11/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789259:789417 [4] NCCL INFO Channel 13/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-142:789258:789420 [3] NCCL INFO Channel 00/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-142:789262:789418 [7] NCCL INFO Channel 12/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789259:789417 [4] NCCL INFO Channel 14/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-173-121:3987801:3987960 [7] NCCL INFO Channel 05/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-172-142:789258:789420 [3] NCCL INFO Channel 02/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-142:789262:789418 [7] NCCL INFO Channel 13/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789259:789417 [4] NCCL INFO Channel 15/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-142:789258:789420 [3] NCCL INFO Channel 08/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-173-121:3987796:3987957 [2] NCCL INFO Channel 00/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-173-121:3987801:3987960 [7] NCCL INFO Channel 06/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-172-142:789262:789418 [7] NCCL INFO Channel 14/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987798:3987955 [4] NCCL INFO Channel 00/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-172-142:789258:789420 [3] NCCL INFO Channel 10/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-173-121:3987796:3987957 [2] NCCL INFO Channel 01/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-173-121:3987801:3987960 [7] NCCL INFO Channel 07/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-172-142:789262:789418 [7] NCCL INFO Channel 15/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987798:3987955 [4] NCCL INFO Channel 03/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-173-121:3987796:3987957 [2] NCCL INFO Channel 08/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-173-121:3987801:3987960 [7] NCCL INFO Channel 09/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-172-142:789262:789418 [7] NCCL INFO Channel 00/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-173-121:3987801:3987960 [7] NCCL INFO Channel 10/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-173-121:3987796:3987957 [2] NCCL INFO Channel 09/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-173-121:3987798:3987955 [4] NCCL INFO Channel 05/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-173-121:3987801:3987960 [7] NCCL INFO Channel 11/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-173-121:3987798:3987955 [4] NCCL INFO Channel 06/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-173-121:3987801:3987960 [7] NCCL INFO Channel 12/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-173-121:3987798:3987955 [4] NCCL INFO Channel 07/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-173-121:3987801:3987960 [7] NCCL INFO Channel 13/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-173-121:3987798:3987955 [4] NCCL INFO Channel 08/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-173-121:3987801:3987960 [7] NCCL INFO Channel 14/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-173-121:3987798:3987955 [4] NCCL INFO Channel 11/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-173-121:3987801:3987960 [7] NCCL INFO Channel 15/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-173-121:3987798:3987955 [4] NCCL INFO Channel 13/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-173-121:3987801:3987960 [7] NCCL INFO Channel 00/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-173-121:3987798:3987955 [4] NCCL INFO Channel 14/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-173-121:3987798:3987955 [4] NCCL INFO Channel 15/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-173-121:3987801:3987960 [7] NCCL INFO Channel 06/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-173-121:3987797:3987954 [3] NCCL INFO Channel 00/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-173-121:3987801:3987960 [7] NCCL INFO Channel 08/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-173-121:3987797:3987954 [3] NCCL INFO Channel 02/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-172-142:789262:789418 [7] NCCL INFO Channel 06/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-173-121:3987801:3987960 [7] NCCL INFO Channel 14/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-173-121:3987797:3987954 [3] NCCL INFO Channel 08/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-172-142:789262:789418 [7] NCCL INFO Channel 08/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-173-121:3987797:3987954 [3] NCCL INFO Channel 10/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-172-142:789262:789418 [7] NCCL INFO Channel 14/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-142:789255:789416 [0] NCCL INFO Connected all trees
+ip-26-0-172-142:789255:789416 [0] NCCL INFO NVLS comm 0x9d36820 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-142:789262:789418 [7] NCCL INFO Connected all trees
+ip-26-0-172-142:789260:789419 [5] NCCL INFO Connected all trees
+ip-26-0-172-142:789261:789423 [6] NCCL INFO Connected all trees
+ip-26-0-172-142:789262:789418 [7] NCCL INFO NVLS comm 0x87c3e80 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-142:789260:789419 [5] NCCL INFO NVLS comm 0x9e37670 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-142:789261:789423 [6] NCCL INFO NVLS comm 0x9bb0c50 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-142:789256:789421 [1] NCCL INFO Connected all trees
+ip-26-0-172-142:789256:789421 [1] NCCL INFO NVLS comm 0xa0887a0 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-142:789259:789417 [4] NCCL INFO Connected all trees
+ip-26-0-172-142:789259:789417 [4] NCCL INFO NVLS comm 0x8e10500 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-142:789258:789420 [3] NCCL INFO Connected all trees
+ip-26-0-172-142:789258:789420 [3] NCCL INFO NVLS comm 0x88eba40 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-142:789257:789422 [2] NCCL INFO Connected all trees
+ip-26-0-172-142:789257:789422 [2] NCCL INFO NVLS comm 0x916e000 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-173-121:3987799:3987958 [5] NCCL INFO Connected all trees
+ip-26-0-173-121:3987794:3987956 [0] NCCL INFO Connected all trees
+ip-26-0-173-121:3987799:3987958 [5] NCCL INFO NVLS comm 0x8f9e050 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-173-121:3987794:3987956 [0] NCCL INFO NVLS comm 0x8a09bb0 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-173-121:3987797:3987954 [3] NCCL INFO Connected all trees
+ip-26-0-173-121:3987801:3987960 [7] NCCL INFO Connected all trees
+ip-26-0-173-121:3987798:3987955 [4] NCCL INFO Connected all trees
+ip-26-0-173-121:3987800:3987959 [6] NCCL INFO Connected all trees
+ip-26-0-173-121:3987797:3987954 [3] NCCL INFO NVLS comm 0x88c5420 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-173-121:3987801:3987960 [7] NCCL INFO NVLS comm 0x9388d10 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-173-121:3987795:3987961 [1] NCCL INFO Connected all trees
+ip-26-0-173-121:3987798:3987955 [4] NCCL INFO NVLS comm 0x9b44720 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-173-121:3987800:3987959 [6] NCCL INFO NVLS comm 0x8feba80 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-173-121:3987795:3987961 [1] NCCL INFO NVLS comm 0x95d08c0 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-173-121:3987796:3987957 [2] NCCL INFO Connected all trees
+ip-26-0-173-121:3987796:3987957 [2] NCCL INFO NVLS comm 0x8e72160 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-142:789259:789417 [4] NCCL INFO Channel 00/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:789259:789417 [4] NCCL INFO Channel 01/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:789259:789417 [4] NCCL INFO Channel 02/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:789259:789417 [4] NCCL INFO Channel 03/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:789259:789417 [4] NCCL INFO Channel 05/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:789259:789417 [4] NCCL INFO Channel 06/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:789259:789417 [4] NCCL INFO Channel 07/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:789259:789417 [4] NCCL INFO Channel 08/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:789259:789417 [4] NCCL INFO Channel 09/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:789259:789417 [4] NCCL INFO Channel 10/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:789259:789417 [4] NCCL INFO Channel 11/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:789259:789417 [4] NCCL INFO Channel 13/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:789259:789417 [4] NCCL INFO Channel 14/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:789259:789417 [4] NCCL INFO Channel 15/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:789259:789417 [4] NCCL INFO Channel 00/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:789259:789417 [4] NCCL INFO Channel 01/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:789259:789417 [4] NCCL INFO Channel 02/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:789259:789417 [4] NCCL INFO Channel 03/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:789259:789417 [4] NCCL INFO Channel 05/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:789255:789416 [0] NCCL INFO Channel 01/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:789259:789417 [4] NCCL INFO Channel 06/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:789262:789418 [7] NCCL INFO Channel 00/0 : 15[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:789259:789417 [4] NCCL INFO Channel 07/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:789255:789416 [0] NCCL INFO Channel 02/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:789262:789418 [7] NCCL INFO Channel 01/0 : 15[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:789259:789417 [4] NCCL INFO Channel 08/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:789256:789421 [1] NCCL INFO Channel 00/0 : 9[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:789261:789423 [6] NCCL INFO Channel 00/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:789260:789419 [5] NCCL INFO Channel 00/0 : 13[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:789255:789416 [0] NCCL INFO Channel 03/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:789262:789418 [7] NCCL INFO Channel 02/0 : 15[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:789259:789417 [4] NCCL INFO Channel 09/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:789256:789421 [1] NCCL INFO Channel 02/0 : 9[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:789257:789422 [2] NCCL INFO Channel 00/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:789261:789423 [6] NCCL INFO Channel 01/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:789255:789416 [0] NCCL INFO Channel 04/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:789262:789418 [7] NCCL INFO Channel 03/0 : 15[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:789260:789419 [5] NCCL INFO Channel 01/0 : 13[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:789259:789417 [4] NCCL INFO Channel 10/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:789256:789421 [1] NCCL INFO Channel 03/0 : 9[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:789257:789422 [2] NCCL INFO Channel 01/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:789261:789423 [6] NCCL INFO Channel 02/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:789258:789420 [3] NCCL INFO Channel 00/0 : 11[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:789262:789418 [7] NCCL INFO Channel 04/0 : 15[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:789255:789416 [0] NCCL INFO Channel 05/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:789259:789417 [4] NCCL INFO Channel 11/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:789260:789419 [5] NCCL INFO Channel 02/0 : 13[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:789256:789421 [1] NCCL INFO Channel 04/0 : 9[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:789257:789422 [2] NCCL INFO Channel 03/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:789261:789423 [6] NCCL INFO Channel 03/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:789258:789420 [3] NCCL INFO Channel 01/0 : 11[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:789262:789418 [7] NCCL INFO Channel 05/0 : 15[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:789259:789417 [4] NCCL INFO Channel 13/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:789255:789416 [0] NCCL INFO Channel 06/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:789260:789419 [5] NCCL INFO Channel 03/0 : 13[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:789256:789421 [1] NCCL INFO Channel 05/0 : 9[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:789257:789422 [2] NCCL INFO Channel 04/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:789261:789423 [6] NCCL INFO Channel 04/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:789258:789420 [3] NCCL INFO Channel 02/0 : 11[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:789262:789418 [7] NCCL INFO Channel 06/0 : 15[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:789259:789417 [4] NCCL INFO Channel 14/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:789255:789416 [0] NCCL INFO Channel 07/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:789260:789419 [5] NCCL INFO Channel 04/0 : 13[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:789256:789421 [1] NCCL INFO Channel 06/0 : 9[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:789257:789422 [2] NCCL INFO Channel 05/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:789261:789423 [6] NCCL INFO Channel 05/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:789262:789418 [7] NCCL INFO Channel 08/0 : 15[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:789259:789417 [4] NCCL INFO Channel 15/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:789258:789420 [3] NCCL INFO Channel 04/0 : 11[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:789255:789416 [0] NCCL INFO Channel 09/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:789260:789419 [5] NCCL INFO Channel 06/0 : 13[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:789256:789421 [1] NCCL INFO Channel 07/0 : 9[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:789257:789422 [2] NCCL INFO Channel 06/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:789262:789418 [7] NCCL INFO Channel 09/0 : 15[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:789261:789423 [6] NCCL INFO Channel 07/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:789258:789420 [3] NCCL INFO Channel 05/0 : 11[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:789255:789416 [0] NCCL INFO Channel 10/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:789260:789419 [5] NCCL INFO Channel 07/0 : 13[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:789257:789422 [2] NCCL INFO Channel 07/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:789256:789421 [1] NCCL INFO Channel 08/0 : 9[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:789262:789418 [7] NCCL INFO Channel 10/0 : 15[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:789261:789423 [6] NCCL INFO Channel 08/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:789258:789420 [3] NCCL INFO Channel 06/0 : 11[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:789255:789416 [0] NCCL INFO Channel 11/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:789260:789419 [5] NCCL INFO Channel 08/0 : 13[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:789256:789421 [1] NCCL INFO Channel 10/0 : 9[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:789257:789422 [2] NCCL INFO Channel 08/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:789262:789418 [7] NCCL INFO Channel 11/0 : 15[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:789261:789423 [6] NCCL INFO Channel 09/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:789258:789420 [3] NCCL INFO Channel 07/0 : 11[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:789255:789416 [0] NCCL INFO Channel 12/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:789260:789419 [5] NCCL INFO Channel 09/0 : 13[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:789257:789422 [2] NCCL INFO Channel 09/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:789256:789421 [1] NCCL INFO Channel 11/0 : 9[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:789262:789418 [7] NCCL INFO Channel 12/0 : 15[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:789255:789416 [0] NCCL INFO Channel 13/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:789261:789423 [6] NCCL INFO Channel 10/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:789258:789420 [3] NCCL INFO Channel 08/0 : 11[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:789260:789419 [5] NCCL INFO Channel 10/0 : 13[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:789256:789421 [1] NCCL INFO Channel 12/0 : 9[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:789262:789418 [7] NCCL INFO Channel 13/0 : 15[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:789257:789422 [2] NCCL INFO Channel 11/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:789258:789420 [3] NCCL INFO Channel 09/0 : 11[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:789255:789416 [0] NCCL INFO Channel 14/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:789260:789419 [5] NCCL INFO Channel 11/0 : 13[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:789261:789423 [6] NCCL INFO Channel 11/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:789256:789421 [1] NCCL INFO Channel 13/0 : 9[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:789262:789418 [7] NCCL INFO Channel 14/0 : 15[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:789257:789422 [2] NCCL INFO Channel 12/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:789258:789420 [3] NCCL INFO Channel 10/0 : 11[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:789255:789416 [0] NCCL INFO Channel 15/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:789256:789421 [1] NCCL INFO Channel 14/0 : 9[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:789260:789419 [5] NCCL INFO Channel 12/0 : 13[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:789261:789423 [6] NCCL INFO Channel 12/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:789262:789418 [7] NCCL INFO Channel 00/0 : 7[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:789257:789422 [2] NCCL INFO Channel 13/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:789258:789420 [3] NCCL INFO Channel 12/0 : 11[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:789255:789416 [0] NCCL INFO Channel 01/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:789256:789421 [1] NCCL INFO Channel 15/0 : 9[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:789260:789419 [5] NCCL INFO Channel 14/0 : 13[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:789262:789418 [7] NCCL INFO Channel 01/0 : 7[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:789261:789423 [6] NCCL INFO Channel 13/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:789257:789422 [2] NCCL INFO Channel 14/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:789255:789416 [0] NCCL INFO Channel 02/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:789258:789420 [3] NCCL INFO Channel 13/0 : 11[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:789256:789421 [1] NCCL INFO Channel 00/0 : 1[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:789262:789418 [7] NCCL INFO Channel 02/0 : 7[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:789260:789419 [5] NCCL INFO Channel 15/0 : 13[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:789261:789423 [6] NCCL INFO Channel 15/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:789257:789422 [2] NCCL INFO Channel 15/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:789258:789420 [3] NCCL INFO Channel 14/0 : 11[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:789255:789416 [0] NCCL INFO Channel 03/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:789256:789421 [1] NCCL INFO Channel 02/0 : 1[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:789262:789418 [7] NCCL INFO Channel 03/0 : 7[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:789260:789419 [5] NCCL INFO Channel 00/0 : 5[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:789261:789423 [6] NCCL INFO Channel 00/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:789258:789420 [3] NCCL INFO Channel 15/0 : 11[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:789256:789421 [1] NCCL INFO Channel 03/0 : 1[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:789257:789422 [2] NCCL INFO Channel 00/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:789255:789416 [0] NCCL INFO Channel 04/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:789262:789418 [7] NCCL INFO Channel 04/0 : 7[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:789260:789419 [5] NCCL INFO Channel 01/0 : 5[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:789261:789423 [6] NCCL INFO Channel 01/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:789258:789420 [3] NCCL INFO Channel 00/0 : 3[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:789256:789421 [1] NCCL INFO Channel 04/0 : 1[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:789255:789416 [0] NCCL INFO Channel 05/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:789257:789422 [2] NCCL INFO Channel 01/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:789262:789418 [7] NCCL INFO Channel 05/0 : 7[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:789260:789419 [5] NCCL INFO Channel 02/0 : 5[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:789261:789423 [6] NCCL INFO Channel 02/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:789258:789420 [3] NCCL INFO Channel 01/0 : 3[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:789256:789421 [1] NCCL INFO Channel 05/0 : 1[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:789255:789416 [0] NCCL INFO Channel 06/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:789257:789422 [2] NCCL INFO Channel 03/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:789262:789418 [7] NCCL INFO Channel 06/0 : 7[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:789260:789419 [5] NCCL INFO Channel 03/0 : 5[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:789261:789423 [6] NCCL INFO Channel 03/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:789258:789420 [3] NCCL INFO Channel 02/0 : 3[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:789256:789421 [1] NCCL INFO Channel 06/0 : 1[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:789255:789416 [0] NCCL INFO Channel 07/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:789257:789422 [2] NCCL INFO Channel 04/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:789262:789418 [7] NCCL INFO Channel 08/0 : 7[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:789260:789419 [5] NCCL INFO Channel 04/0 : 5[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:789261:789423 [6] NCCL INFO Channel 04/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:789258:789420 [3] NCCL INFO Channel 04/0 : 3[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:789256:789421 [1] NCCL INFO Channel 07/0 : 1[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:789255:789416 [0] NCCL INFO Channel 09/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:789257:789422 [2] NCCL INFO Channel 05/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:789260:789419 [5] NCCL INFO Channel 06/0 : 5[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:789261:789423 [6] NCCL INFO Channel 05/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:789256:789421 [1] NCCL INFO Channel 08/0 : 1[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:789262:789418 [7] NCCL INFO Channel 09/0 : 7[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:789255:789416 [0] NCCL INFO Channel 10/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:789257:789422 [2] NCCL INFO Channel 06/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:789258:789420 [3] NCCL INFO Channel 05/0 : 3[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:789260:789419 [5] NCCL INFO Channel 07/0 : 5[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:789261:789423 [6] NCCL INFO Channel 07/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:789255:789416 [0] NCCL INFO Channel 11/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:789257:789422 [2] NCCL INFO Channel 07/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:789258:789420 [3] NCCL INFO Channel 06/0 : 3[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:789260:789419 [5] NCCL INFO Channel 08/0 : 5[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:789261:789423 [6] NCCL INFO Channel 08/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:789256:789421 [1] NCCL INFO Channel 10/0 : 1[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:789262:789418 [7] NCCL INFO Channel 10/0 : 7[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:789255:789416 [0] NCCL INFO Channel 12/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:789257:789422 [2] NCCL INFO Channel 08/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:789258:789420 [3] NCCL INFO Channel 07/0 : 3[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:789261:789423 [6] NCCL INFO Channel 09/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:789260:789419 [5] NCCL INFO Channel 09/0 : 5[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:789256:789421 [1] NCCL INFO Channel 11/0 : 1[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:789255:789416 [0] NCCL INFO Channel 13/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:789257:789422 [2] NCCL INFO Channel 09/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:789258:789420 [3] NCCL INFO Channel 08/0 : 3[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:789262:789418 [7] NCCL INFO Channel 11/0 : 7[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:789261:789423 [6] NCCL INFO Channel 10/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:789260:789419 [5] NCCL INFO Channel 10/0 : 5[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:789256:789421 [1] NCCL INFO Channel 12/0 : 1[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:789255:789416 [0] NCCL INFO Channel 14/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:789257:789422 [2] NCCL INFO Channel 11/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:789258:789420 [3] NCCL INFO Channel 09/0 : 3[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:789262:789418 [7] NCCL INFO Channel 12/0 : 7[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:789261:789423 [6] NCCL INFO Channel 11/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:789256:789421 [1] NCCL INFO Channel 13/0 : 1[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:789260:789419 [5] NCCL INFO Channel 11/0 : 5[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:789255:789416 [0] NCCL INFO Channel 15/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:789257:789422 [2] NCCL INFO Channel 12/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:789258:789420 [3] NCCL INFO Channel 10/0 : 3[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:789262:789418 [7] NCCL INFO Channel 13/0 : 7[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:789261:789423 [6] NCCL INFO Channel 12/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:789256:789421 [1] NCCL INFO Channel 14/0 : 1[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:789260:789419 [5] NCCL INFO Channel 12/0 : 5[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:789257:789422 [2] NCCL INFO Channel 13/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:789258:789420 [3] NCCL INFO Channel 12/0 : 3[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:789261:789423 [6] NCCL INFO Channel 13/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:789262:789418 [7] NCCL INFO Channel 14/0 : 7[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:789256:789421 [1] NCCL INFO Channel 15/0 : 1[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:789257:789422 [2] NCCL INFO Channel 14/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:789258:789420 [3] NCCL INFO Channel 13/0 : 3[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:789261:789423 [6] NCCL INFO Channel 15/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:789257:789422 [2] NCCL INFO Channel 15/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:789258:789420 [3] NCCL INFO Channel 14/0 : 3[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:789260:789419 [5] NCCL INFO Channel 14/0 : 5[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:789258:789420 [3] NCCL INFO Channel 15/0 : 3[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:789260:789419 [5] NCCL INFO Channel 15/0 : 5[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-121:3987794:3987956 [0] NCCL INFO Channel 01/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-121:3987801:3987960 [7] NCCL INFO Channel 00/0 : 7[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-121:3987798:3987955 [4] NCCL INFO Channel 00/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-121:3987794:3987956 [0] NCCL INFO Channel 02/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-121:3987801:3987960 [7] NCCL INFO Channel 01/0 : 7[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-121:3987798:3987955 [4] NCCL INFO Channel 01/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-121:3987794:3987956 [0] NCCL INFO Channel 03/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-121:3987801:3987960 [7] NCCL INFO Channel 02/0 : 7[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-121:3987798:3987955 [4] NCCL INFO Channel 02/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-121:3987794:3987956 [0] NCCL INFO Channel 04/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-121:3987801:3987960 [7] NCCL INFO Channel 03/0 : 7[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-121:3987798:3987955 [4] NCCL INFO Channel 03/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-121:3987799:3987958 [5] NCCL INFO Channel 00/0 : 5[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-121:3987796:3987957 [2] NCCL INFO Channel 00/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-121:3987794:3987956 [0] NCCL INFO Channel 05/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-121:3987801:3987960 [7] NCCL INFO Channel 04/0 : 7[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-121:3987800:3987959 [6] NCCL INFO Channel 00/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-121:3987798:3987955 [4] NCCL INFO Channel 05/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-121:3987799:3987958 [5] NCCL INFO Channel 01/0 : 5[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-121:3987795:3987961 [1] NCCL INFO Channel 00/0 : 1[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-121:3987796:3987957 [2] NCCL INFO Channel 01/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-121:3987797:3987954 [3] NCCL INFO Channel 00/0 : 3[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-121:3987794:3987956 [0] NCCL INFO Channel 06/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-121:3987801:3987960 [7] NCCL INFO Channel 05/0 : 7[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-121:3987800:3987959 [6] NCCL INFO Channel 01/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-121:3987798:3987955 [4] NCCL INFO Channel 06/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-121:3987799:3987958 [5] NCCL INFO Channel 02/0 : 5[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-121:3987795:3987961 [1] NCCL INFO Channel 02/0 : 1[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-121:3987796:3987957 [2] NCCL INFO Channel 03/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-121:3987797:3987954 [3] NCCL INFO Channel 01/0 : 3[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-121:3987794:3987956 [0] NCCL INFO Channel 07/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-121:3987801:3987960 [7] NCCL INFO Channel 06/0 : 7[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-121:3987798:3987955 [4] NCCL INFO Channel 07/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-121:3987800:3987959 [6] NCCL INFO Channel 02/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-121:3987796:3987957 [2] NCCL INFO Channel 04/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-121:3987799:3987958 [5] NCCL INFO Channel 03/0 : 5[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-121:3987795:3987961 [1] NCCL INFO Channel 03/0 : 1[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-121:3987797:3987954 [3] NCCL INFO Channel 02/0 : 3[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-121:3987794:3987956 [0] NCCL INFO Channel 09/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-121:3987801:3987960 [7] NCCL INFO Channel 08/0 : 7[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-121:3987798:3987955 [4] NCCL INFO Channel 08/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-121:3987800:3987959 [6] NCCL INFO Channel 03/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-121:3987796:3987957 [2] NCCL INFO Channel 05/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-121:3987799:3987958 [5] NCCL INFO Channel 04/0 : 5[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-121:3987797:3987954 [3] NCCL INFO Channel 04/0 : 3[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-121:3987795:3987961 [1] NCCL INFO Channel 04/0 : 1[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-121:3987794:3987956 [0] NCCL INFO Channel 10/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-121:3987798:3987955 [4] NCCL INFO Channel 09/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-121:3987801:3987960 [7] NCCL INFO Channel 09/0 : 7[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-121:3987800:3987959 [6] NCCL INFO Channel 04/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-121:3987796:3987957 [2] NCCL INFO Channel 06/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-121:3987799:3987958 [5] NCCL INFO Channel 06/0 : 5[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-121:3987797:3987954 [3] NCCL INFO Channel 05/0 : 3[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-121:3987795:3987961 [1] NCCL INFO Channel 05/0 : 1[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-121:3987794:3987956 [0] NCCL INFO Channel 11/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-121:3987801:3987960 [7] NCCL INFO Channel 10/0 : 7[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-121:3987796:3987957 [2] NCCL INFO Channel 07/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-121:3987798:3987955 [4] NCCL INFO Channel 10/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-121:3987800:3987959 [6] NCCL INFO Channel 05/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-121:3987797:3987954 [3] NCCL INFO Channel 06/0 : 3[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-121:3987799:3987958 [5] NCCL INFO Channel 07/0 : 5[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-121:3987794:3987956 [0] NCCL INFO Channel 12/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-121:3987795:3987961 [1] NCCL INFO Channel 06/0 : 1[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-121:3987801:3987960 [7] NCCL INFO Channel 11/0 : 7[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-121:3987796:3987957 [2] NCCL INFO Channel 08/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-121:3987798:3987955 [4] NCCL INFO Channel 11/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-121:3987797:3987954 [3] NCCL INFO Channel 07/0 : 3[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-121:3987799:3987958 [5] NCCL INFO Channel 08/0 : 5[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-121:3987794:3987956 [0] NCCL INFO Channel 13/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-121:3987800:3987959 [6] NCCL INFO Channel 07/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-121:3987795:3987961 [1] NCCL INFO Channel 07/0 : 1[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-121:3987801:3987960 [7] NCCL INFO Channel 12/0 : 7[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-121:3987796:3987957 [2] NCCL INFO Channel 09/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-121:3987797:3987954 [3] NCCL INFO Channel 08/0 : 3[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-121:3987798:3987955 [4] NCCL INFO Channel 13/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-121:3987799:3987958 [5] NCCL INFO Channel 09/0 : 5[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-121:3987794:3987956 [0] NCCL INFO Channel 14/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-121:3987800:3987959 [6] NCCL INFO Channel 08/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-121:3987795:3987961 [1] NCCL INFO Channel 08/0 : 1[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-121:3987801:3987960 [7] NCCL INFO Channel 13/0 : 7[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-121:3987797:3987954 [3] NCCL INFO Channel 09/0 : 3[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-121:3987796:3987957 [2] NCCL INFO Channel 11/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-121:3987798:3987955 [4] NCCL INFO Channel 14/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-121:3987799:3987958 [5] NCCL INFO Channel 10/0 : 5[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-121:3987794:3987956 [0] NCCL INFO Channel 15/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-121:3987800:3987959 [6] NCCL INFO Channel 09/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-121:3987795:3987961 [1] NCCL INFO Channel 10/0 : 1[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-121:3987801:3987960 [7] NCCL INFO Channel 14/0 : 7[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-121:3987797:3987954 [3] NCCL INFO Channel 10/0 : 3[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-121:3987798:3987955 [4] NCCL INFO Channel 15/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-121:3987796:3987957 [2] NCCL INFO Channel 12/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-121:3987794:3987956 [0] NCCL INFO Channel 01/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-121:3987799:3987958 [5] NCCL INFO Channel 11/0 : 5[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-121:3987800:3987959 [6] NCCL INFO Channel 10/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-121:3987795:3987961 [1] NCCL INFO Channel 11/0 : 1[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-121:3987797:3987954 [3] NCCL INFO Channel 12/0 : 3[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-121:3987801:3987960 [7] NCCL INFO Channel 00/0 : 15[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-121:3987796:3987957 [2] NCCL INFO Channel 13/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-121:3987798:3987955 [4] NCCL INFO Channel 00/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-121:3987794:3987956 [0] NCCL INFO Channel 02/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-121:3987799:3987958 [5] NCCL INFO Channel 12/0 : 5[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-121:3987795:3987961 [1] NCCL INFO Channel 12/0 : 1[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-121:3987800:3987959 [6] NCCL INFO Channel 11/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-121:3987797:3987954 [3] NCCL INFO Channel 13/0 : 3[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-121:3987801:3987960 [7] NCCL INFO Channel 01/0 : 15[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-121:3987796:3987957 [2] NCCL INFO Channel 14/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-121:3987794:3987956 [0] NCCL INFO Channel 03/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-121:3987798:3987955 [4] NCCL INFO Channel 01/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-121:3987795:3987961 [1] NCCL INFO Channel 13/0 : 1[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-121:3987799:3987958 [5] NCCL INFO Channel 14/0 : 5[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-121:3987801:3987960 [7] NCCL INFO Channel 02/0 : 15[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-121:3987797:3987954 [3] NCCL INFO Channel 14/0 : 3[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-121:3987800:3987959 [6] NCCL INFO Channel 12/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-121:3987796:3987957 [2] NCCL INFO Channel 15/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-121:3987794:3987956 [0] NCCL INFO Channel 04/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-121:3987798:3987955 [4] NCCL INFO Channel 02/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-121:3987799:3987958 [5] NCCL INFO Channel 15/0 : 5[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-121:3987795:3987961 [1] NCCL INFO Channel 14/0 : 1[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-121:3987797:3987954 [3] NCCL INFO Channel 15/0 : 3[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-121:3987801:3987960 [7] NCCL INFO Channel 03/0 : 15[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-121:3987800:3987959 [6] NCCL INFO Channel 13/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-121:3987796:3987957 [2] NCCL INFO Channel 00/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-121:3987794:3987956 [0] NCCL INFO Channel 05/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-121:3987798:3987955 [4] NCCL INFO Channel 03/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-121:3987799:3987958 [5] NCCL INFO Channel 00/0 : 13[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-121:3987795:3987961 [1] NCCL INFO Channel 15/0 : 1[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-121:3987797:3987954 [3] NCCL INFO Channel 00/0 : 11[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-121:3987801:3987960 [7] NCCL INFO Channel 04/0 : 15[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-121:3987800:3987959 [6] NCCL INFO Channel 15/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-121:3987796:3987957 [2] NCCL INFO Channel 01/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-121:3987794:3987956 [0] NCCL INFO Channel 06/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-121:3987798:3987955 [4] NCCL INFO Channel 05/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-121:3987795:3987961 [1] NCCL INFO Channel 00/0 : 9[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-121:3987799:3987958 [5] NCCL INFO Channel 01/0 : 13[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-121:3987797:3987954 [3] NCCL INFO Channel 01/0 : 11[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-121:3987801:3987960 [7] NCCL INFO Channel 05/0 : 15[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-121:3987800:3987959 [6] NCCL INFO Channel 00/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-121:3987794:3987956 [0] NCCL INFO Channel 07/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-121:3987796:3987957 [2] NCCL INFO Channel 03/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-121:3987798:3987955 [4] NCCL INFO Channel 06/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-121:3987799:3987958 [5] NCCL INFO Channel 02/0 : 13[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-121:3987797:3987954 [3] NCCL INFO Channel 02/0 : 11[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-121:3987795:3987961 [1] NCCL INFO Channel 02/0 : 9[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-121:3987801:3987960 [7] NCCL INFO Channel 06/0 : 15[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-121:3987800:3987959 [6] NCCL INFO Channel 01/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-121:3987794:3987956 [0] NCCL INFO Channel 09/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-121:3987796:3987957 [2] NCCL INFO Channel 04/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-121:3987798:3987955 [4] NCCL INFO Channel 07/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-121:3987799:3987958 [5] NCCL INFO Channel 03/0 : 13[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-121:3987797:3987954 [3] NCCL INFO Channel 04/0 : 11[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-121:3987795:3987961 [1] NCCL INFO Channel 03/0 : 9[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-121:3987801:3987960 [7] NCCL INFO Channel 08/0 : 15[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-121:3987800:3987959 [6] NCCL INFO Channel 02/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-121:3987794:3987956 [0] NCCL INFO Channel 10/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-121:3987796:3987957 [2] NCCL INFO Channel 05/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-121:3987799:3987958 [5] NCCL INFO Channel 04/0 : 13[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-121:3987798:3987955 [4] NCCL INFO Channel 08/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-121:3987797:3987954 [3] NCCL INFO Channel 05/0 : 11[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-121:3987795:3987961 [1] NCCL INFO Channel 04/0 : 9[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-121:3987801:3987960 [7] NCCL INFO Channel 09/0 : 15[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-121:3987800:3987959 [6] NCCL INFO Channel 03/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-121:3987794:3987956 [0] NCCL INFO Channel 11/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-121:3987796:3987957 [2] NCCL INFO Channel 06/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-121:3987799:3987958 [5] NCCL INFO Channel 06/0 : 13[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-121:3987797:3987954 [3] NCCL INFO Channel 06/0 : 11[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-121:3987798:3987955 [4] NCCL INFO Channel 09/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-121:3987795:3987961 [1] NCCL INFO Channel 05/0 : 9[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-121:3987794:3987956 [0] NCCL INFO Channel 12/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-121:3987801:3987960 [7] NCCL INFO Channel 10/0 : 15[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-121:3987800:3987959 [6] NCCL INFO Channel 04/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-121:3987796:3987957 [2] NCCL INFO Channel 07/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-121:3987799:3987958 [5] NCCL INFO Channel 07/0 : 13[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-121:3987797:3987954 [3] NCCL INFO Channel 07/0 : 11[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-121:3987795:3987961 [1] NCCL INFO Channel 06/0 : 9[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-121:3987798:3987955 [4] NCCL INFO Channel 10/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-121:3987794:3987956 [0] NCCL INFO Channel 13/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-121:3987800:3987959 [6] NCCL INFO Channel 05/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-121:3987801:3987960 [7] NCCL INFO Channel 11/0 : 15[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-121:3987796:3987957 [2] NCCL INFO Channel 08/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-121:3987799:3987958 [5] NCCL INFO Channel 08/0 : 13[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-121:3987797:3987954 [3] NCCL INFO Channel 08/0 : 11[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-121:3987795:3987961 [1] NCCL INFO Channel 07/0 : 9[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-121:3987798:3987955 [4] NCCL INFO Channel 11/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-121:3987794:3987956 [0] NCCL INFO Channel 14/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-121:3987800:3987959 [6] NCCL INFO Channel 07/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-121:3987796:3987957 [2] NCCL INFO Channel 09/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-121:3987801:3987960 [7] NCCL INFO Channel 12/0 : 15[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-121:3987799:3987958 [5] NCCL INFO Channel 09/0 : 13[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-121:3987797:3987954 [3] NCCL INFO Channel 09/0 : 11[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-121:3987795:3987961 [1] NCCL INFO Channel 08/0 : 9[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-121:3987798:3987955 [4] NCCL INFO Channel 13/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-121:3987794:3987956 [0] NCCL INFO Channel 15/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-121:3987800:3987959 [6] NCCL INFO Channel 08/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-121:3987796:3987957 [2] NCCL INFO Channel 11/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-121:3987801:3987960 [7] NCCL INFO Channel 13/0 : 15[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-121:3987799:3987958 [5] NCCL INFO Channel 10/0 : 13[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-121:3987795:3987961 [1] NCCL INFO Channel 10/0 : 9[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-121:3987797:3987954 [3] NCCL INFO Channel 10/0 : 11[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-121:3987798:3987955 [4] NCCL INFO Channel 14/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-121:3987800:3987959 [6] NCCL INFO Channel 09/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-121:3987796:3987957 [2] NCCL INFO Channel 12/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-121:3987801:3987960 [7] NCCL INFO Channel 14/0 : 15[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-121:3987799:3987958 [5] NCCL INFO Channel 11/0 : 13[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-121:3987797:3987954 [3] NCCL INFO Channel 12/0 : 11[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-121:3987795:3987961 [1] NCCL INFO Channel 11/0 : 9[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-121:3987798:3987955 [4] NCCL INFO Channel 15/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-121:3987800:3987959 [6] NCCL INFO Channel 10/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-121:3987796:3987957 [2] NCCL INFO Channel 13/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-121:3987799:3987958 [5] NCCL INFO Channel 12/0 : 13[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-121:3987795:3987961 [1] NCCL INFO Channel 12/0 : 9[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-121:3987797:3987954 [3] NCCL INFO Channel 13/0 : 11[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-121:3987800:3987959 [6] NCCL INFO Channel 11/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-121:3987796:3987957 [2] NCCL INFO Channel 14/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-121:3987799:3987958 [5] NCCL INFO Channel 14/0 : 13[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-121:3987795:3987961 [1] NCCL INFO Channel 13/0 : 9[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-121:3987797:3987954 [3] NCCL INFO Channel 14/0 : 11[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-121:3987800:3987959 [6] NCCL INFO Channel 12/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-121:3987799:3987958 [5] NCCL INFO Channel 15/0 : 13[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-121:3987796:3987957 [2] NCCL INFO Channel 15/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-121:3987797:3987954 [3] NCCL INFO Channel 15/0 : 11[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-121:3987795:3987961 [1] NCCL INFO Channel 14/0 : 9[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-121:3987800:3987959 [6] NCCL INFO Channel 13/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-121:3987800:3987959 [6] NCCL INFO Channel 15/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-121:3987795:3987961 [1] NCCL INFO Channel 15/0 : 9[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:789255:789416 [0] NCCL INFO Connected NVLS tree
+ip-26-0-172-142:789255:789416 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-142:789255:789416 [0] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+ip-26-0-172-142:789255:789416 [0] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-173-121:3987796:3987957 [2] NCCL INFO Connected NVLS tree
+ip-26-0-173-121:3987796:3987957 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-173-121:3987796:3987957 [2] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+ip-26-0-173-121:3987796:3987957 [2] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-173-121:3987798:3987955 [4] NCCL INFO Connected NVLS tree
+ip-26-0-173-121:3987798:3987955 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-173-121:3987798:3987955 [4] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+ip-26-0-173-121:3987798:3987955 [4] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-173-121:3987799:3987958 [5] NCCL INFO Connected NVLS tree
+ip-26-0-173-121:3987799:3987958 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-173-121:3987799:3987958 [5] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+ip-26-0-173-121:3987799:3987958 [5] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-172-142:789259:789417 [4] NCCL INFO Connected NVLS tree
+ip-26-0-172-142:789259:789417 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-142:789259:789417 [4] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+ip-26-0-172-142:789259:789417 [4] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-173-121:3987794:3987956 [0] NCCL INFO Connected NVLS tree
+ip-26-0-173-121:3987794:3987956 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-173-121:3987794:3987956 [0] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+ip-26-0-173-121:3987794:3987956 [0] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-172-142:789257:789422 [2] NCCL INFO Connected NVLS tree
+ip-26-0-172-142:789257:789422 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-142:789257:789422 [2] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+ip-26-0-172-142:789257:789422 [2] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-172-142:789260:789419 [5] NCCL INFO Connected NVLS tree
+ip-26-0-172-142:789260:789419 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-142:789260:789419 [5] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+ip-26-0-172-142:789260:789419 [5] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-173-121:3987801:3987960 [7] NCCL INFO Connected NVLS tree
+ip-26-0-173-121:3987801:3987960 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-173-121:3987801:3987960 [7] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+ip-26-0-173-121:3987801:3987960 [7] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-173-121:3987797:3987954 [3] NCCL INFO Connected NVLS tree
+ip-26-0-173-121:3987797:3987954 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-173-121:3987797:3987954 [3] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+ip-26-0-173-121:3987797:3987954 [3] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-172-142:789258:789420 [3] NCCL INFO Connected NVLS tree
+ip-26-0-172-142:789258:789420 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-142:789258:789420 [3] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+ip-26-0-172-142:789258:789420 [3] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-173-121:3987800:3987959 [6] NCCL INFO Connected NVLS tree
+ip-26-0-173-121:3987800:3987959 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-173-121:3987800:3987959 [6] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+ip-26-0-173-121:3987800:3987959 [6] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-172-142:789262:789418 [7] NCCL INFO Connected NVLS tree
+ip-26-0-172-142:789262:789418 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-142:789262:789418 [7] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+ip-26-0-172-142:789262:789418 [7] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-172-142:789261:789423 [6] NCCL INFO Connected NVLS tree
+ip-26-0-172-142:789261:789423 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-142:789261:789423 [6] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+ip-26-0-172-142:789261:789423 [6] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-173-121:3987795:3987961 [1] NCCL INFO Connected NVLS tree
+ip-26-0-173-121:3987795:3987961 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-142:789256:789421 [1] NCCL INFO Connected NVLS tree
+ip-26-0-172-142:789256:789421 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-142:789256:789421 [1] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+ip-26-0-172-142:789256:789421 [1] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-173-121:3987795:3987961 [1] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+ip-26-0-173-121:3987795:3987961 [1] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-172-142:789262:789418 [7] NCCL INFO comm 0x87c3e80 rank 7 nranks 16 cudaDev 7 nvmlDev 7 busId ca000 commId 0xf40a4bd2dbd7f17c - Init COMPLETE
+ip-26-0-172-142:789259:789417 [4] NCCL INFO comm 0x8e10500 rank 4 nranks 16 cudaDev 4 nvmlDev 4 busId 97000 commId 0xf40a4bd2dbd7f17c - Init COMPLETE
+ip-26-0-172-142:789261:789423 [6] NCCL INFO comm 0x9bb0c50 rank 6 nranks 16 cudaDev 6 nvmlDev 6 busId b9000 commId 0xf40a4bd2dbd7f17c - Init COMPLETE
+ip-26-0-172-142:789258:789420 [3] NCCL INFO comm 0x88eba40 rank 3 nranks 16 cudaDev 3 nvmlDev 3 busId 86000 commId 0xf40a4bd2dbd7f17c - Init COMPLETE
+ip-26-0-172-142:789260:789419 [5] NCCL INFO comm 0x9e37670 rank 5 nranks 16 cudaDev 5 nvmlDev 5 busId a8000 commId 0xf40a4bd2dbd7f17c - Init COMPLETE
+ip-26-0-172-142:789256:789421 [1] NCCL INFO comm 0xa0887a0 rank 1 nranks 16 cudaDev 1 nvmlDev 1 busId 64000 commId 0xf40a4bd2dbd7f17c - Init COMPLETE
+ip-26-0-172-142:789255:789416 [0] NCCL INFO comm 0x9d36820 rank 0 nranks 16 cudaDev 0 nvmlDev 0 busId 53000 commId 0xf40a4bd2dbd7f17c - Init COMPLETE
+ip-26-0-172-142:789257:789422 [2] NCCL INFO comm 0x916e000 rank 2 nranks 16 cudaDev 2 nvmlDev 2 busId 75000 commId 0xf40a4bd2dbd7f17c - Init COMPLETE
+ip-26-0-173-121:3987801:3987960 [7] NCCL INFO comm 0x9388d10 rank 15 nranks 16 cudaDev 7 nvmlDev 7 busId ca000 commId 0xf40a4bd2dbd7f17c - Init COMPLETE
+ip-26-0-173-121:3987797:3987954 [3] NCCL INFO comm 0x88c5420 rank 11 nranks 16 cudaDev 3 nvmlDev 3 busId 86000 commId 0xf40a4bd2dbd7f17c - Init COMPLETE
+ip-26-0-173-121:3987799:3987958 [5] NCCL INFO comm 0x8f9e050 rank 13 nranks 16 cudaDev 5 nvmlDev 5 busId a8000 commId 0xf40a4bd2dbd7f17c - Init COMPLETE
+ip-26-0-173-121:3987795:3987961 [1] NCCL INFO comm 0x95d08c0 rank 9 nranks 16 cudaDev 1 nvmlDev 1 busId 64000 commId 0xf40a4bd2dbd7f17c - Init COMPLETE
+ip-26-0-173-121:3987796:3987957 [2] NCCL INFO comm 0x8e72160 rank 10 nranks 16 cudaDev 2 nvmlDev 2 busId 75000 commId 0xf40a4bd2dbd7f17c - Init COMPLETE
+ip-26-0-173-121:3987794:3987956 [0] NCCL INFO comm 0x8a09bb0 rank 8 nranks 16 cudaDev 0 nvmlDev 0 busId 53000 commId 0xf40a4bd2dbd7f17c - Init COMPLETE
+ip-26-0-173-121:3987800:3987959 [6] NCCL INFO comm 0x8feba80 rank 14 nranks 16 cudaDev 6 nvmlDev 6 busId b9000 commId 0xf40a4bd2dbd7f17c - Init COMPLETE
+ip-26-0-173-121:3987798:3987955 [4] NCCL INFO comm 0x9b44720 rank 12 nranks 16 cudaDev 4 nvmlDev 4 busId 97000 commId 0xf40a4bd2dbd7f17c - Init COMPLETE
+ip-26-0-172-142:789255:789586 [0] NCCL INFO Using network Libfabric
+ip-26-0-172-142:789257:789588 [2] NCCL INFO Using network Libfabric
+ip-26-0-172-142:789258:789592 [3] NCCL INFO Using network Libfabric
+ip-26-0-172-142:789260:789591 [5] NCCL INFO Using network Libfabric
+ip-26-0-172-142:789261:789587 [6] NCCL INFO Using network Libfabric
+ip-26-0-173-121:3987794:3988123 [0] NCCL INFO Using network Libfabric
+ip-26-0-173-121:3987796:3988122 [2] NCCL INFO Using network Libfabric
+ip-26-0-173-121:3987797:3988125 [3] NCCL INFO Using network Libfabric
+ip-26-0-172-142:789256:789593 [1] NCCL INFO Using network Libfabric
+ip-26-0-173-121:3987801:3988128 [7] NCCL INFO Using network Libfabric
+ip-26-0-172-142:789262:789590 [7] NCCL INFO Using network Libfabric
+ip-26-0-173-121:3987800:3988126 [6] NCCL INFO Using network Libfabric
+ip-26-0-172-142:789259:789589 [4] NCCL INFO Using network Libfabric
+ip-26-0-173-121:3987798:3988129 [4] NCCL INFO Using network Libfabric
+ip-26-0-173-121:3987795:3988124 [1] NCCL INFO Using network Libfabric
+ip-26-0-173-121:3987799:3988127 [5] NCCL INFO Using network Libfabric
+ip-26-0-173-121:3987801:3988128 [7] NCCL INFO comm 0x9851480 rank 15 nranks 16 cudaDev 7 nvmlDev 7 busId ca000 commId 0xa2326dfe0277fd9 - Init START
+ip-26-0-172-142:789255:789586 [0] NCCL INFO comm 0xa1ffd50 rank 0 nranks 16 cudaDev 0 nvmlDev 0 busId 53000 commId 0xa2326dfe0277fd9 - Init START
+ip-26-0-173-121:3987800:3988126 [6] NCCL INFO comm 0x943f720 rank 14 nranks 16 cudaDev 6 nvmlDev 6 busId b9000 commId 0xa2326dfe0277fd9 - Init START
+ip-26-0-172-142:789262:789590 [7] NCCL INFO comm 0x8c22120 rank 7 nranks 16 cudaDev 7 nvmlDev 7 busId ca000 commId 0xa2326dfe0277fd9 - Init START
+ip-26-0-172-142:789261:789587 [6] NCCL INFO comm 0xa007010 rank 6 nranks 16 cudaDev 6 nvmlDev 6 busId b9000 commId 0xa2326dfe0277fd9 - Init START
+ip-26-0-172-142:789256:789593 [1] NCCL INFO comm 0xa553960 rank 1 nranks 16 cudaDev 1 nvmlDev 1 busId 64000 commId 0xa2326dfe0277fd9 - Init START
+ip-26-0-172-142:789260:789591 [5] NCCL INFO comm 0xa28a6a0 rank 5 nranks 16 cudaDev 5 nvmlDev 5 busId a8000 commId 0xa2326dfe0277fd9 - Init START
+ip-26-0-172-142:789258:789592 [3] NCCL INFO comm 0x8d44de0 rank 3 nranks 16 cudaDev 3 nvmlDev 3 busId 86000 commId 0xa2326dfe0277fd9 - Init START
+ip-26-0-172-142:789259:789589 [4] NCCL INFO comm 0x92665b0 rank 4 nranks 16 cudaDev 4 nvmlDev 4 busId 97000 commId 0xa2326dfe0277fd9 - Init START
+ip-26-0-172-142:789257:789588 [2] NCCL INFO comm 0x95c0680 rank 2 nranks 16 cudaDev 2 nvmlDev 2 busId 75000 commId 0xa2326dfe0277fd9 - Init START
+ip-26-0-173-121:3987799:3988127 [5] NCCL INFO comm 0x93f88d0 rank 13 nranks 16 cudaDev 5 nvmlDev 5 busId a8000 commId 0xa2326dfe0277fd9 - Init START
+ip-26-0-173-121:3987798:3988129 [4] NCCL INFO comm 0x9df7d00 rank 12 nranks 16 cudaDev 4 nvmlDev 4 busId 97000 commId 0xa2326dfe0277fd9 - Init START
+ip-26-0-173-121:3987797:3988125 [3] NCCL INFO comm 0x8d9bd80 rank 11 nranks 16 cudaDev 3 nvmlDev 3 busId 86000 commId 0xa2326dfe0277fd9 - Init START
+ip-26-0-173-121:3987796:3988122 [2] NCCL INFO comm 0x92c5260 rank 10 nranks 16 cudaDev 2 nvmlDev 2 busId 75000 commId 0xa2326dfe0277fd9 - Init START
+ip-26-0-173-121:3987795:3988124 [1] NCCL INFO comm 0x9a28d90 rank 9 nranks 16 cudaDev 1 nvmlDev 1 busId 64000 commId 0xa2326dfe0277fd9 - Init START
+ip-26-0-173-121:3987794:3988123 [0] NCCL INFO comm 0x8e641b0 rank 8 nranks 16 cudaDev 0 nvmlDev 0 busId 53000 commId 0xa2326dfe0277fd9 - Init START
+ip-26-0-172-142:789260:789591 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-172-142:789259:789589 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-172-142:789261:789587 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-172-142:789262:789590 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-173-121:3987797:3988125 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-173-121:3987794:3988123 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-172-142:789255:789586 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/205
+ip-26-0-172-142:789256:789593 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-172-142:789257:789588 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-172-142:789258:789592 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-173-121:3987795:3988124 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-173-121:3987796:3988122 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-173-121:3987800:3988126 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-173-121:3987798:3988129 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-173-121:3987799:3988127 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-173-121:3987801:3988128 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-173-121:3987795:3988124 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-173-121:3987795:3988124 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-172-142:789255:789586 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-172-142:789255:789586 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-172-142:789257:789588 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-172-142:789257:789588 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-172-142:789258:789592 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-172-142:789258:789592 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-172-142:789259:789589 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-172-142:789259:789589 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-172-142:789262:789590 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-172-142:789262:789590 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-172-142:789261:789587 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-172-142:789261:789587 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-173-121:3987798:3988129 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-173-121:3987798:3988129 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-172-142:789256:789593 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-172-142:789256:789593 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-172-142:789260:789591 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-172-142:789260:789591 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-173-121:3987800:3988126 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-173-121:3987800:3988126 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-173-121:3987801:3988128 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-173-121:3987801:3988128 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-173-121:3987796:3988122 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-173-121:3987796:3988122 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-173-121:3987799:3988127 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-173-121:3987799:3988127 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-173-121:3987794:3988123 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-173-121:3987794:3988123 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-173-121:3987797:3988125 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-173-121:3987797:3988125 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-173-121:3987797:3988125 [3] NCCL INFO Trees [0] 12/-1/-1->11->10 [1] 12/-1/-1->11->10 [2] 12/-1/-1->11->10 [3] 12/-1/-1->11->3 [4] -1/-1/-1->11->10 [5] 12/-1/-1->11->10 [6] 12/-1/-1->11->10 [7] 12/-1/-1->11->10 [8] 12/-1/-1->11->10 [9] 12/-1/-1->11->10 [10] 12/-1/-1->11->10 [11] 12/3/-1->11->-1 [12] -1/-1/-1->11->10 [13] 12/-1/-1->11->10 [14] 12/-1/-1->11->10 [15] 12/-1/-1->11->10
+ip-26-0-173-121:3987797:3988125 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-142:789261:789587 [6] NCCL INFO NVLS Head  0:  0  8
+ip-26-0-172-142:789262:789590 [7] NCCL INFO NVLS Head  0:  0  8
+ip-26-0-172-142:789261:789587 [6] NCCL INFO NVLS Head  1:  1  9
+ip-26-0-172-142:789262:789590 [7] NCCL INFO NVLS Head  1:  1  9
+ip-26-0-172-142:789261:789587 [6] NCCL INFO NVLS Head  2:  2 10
+ip-26-0-172-142:789262:789590 [7] NCCL INFO NVLS Head  2:  2 10
+ip-26-0-172-142:789261:789587 [6] NCCL INFO NVLS Head  3:  3 11
+ip-26-0-172-142:789262:789590 [7] NCCL INFO NVLS Head  3:  3 11
+ip-26-0-172-142:789261:789587 [6] NCCL INFO NVLS Head  4:  4 12
+ip-26-0-172-142:789262:789590 [7] NCCL INFO NVLS Head  4:  4 12
+ip-26-0-172-142:789256:789593 [1] NCCL INFO NVLS Head  0:  0  8
+ip-26-0-172-142:789261:789587 [6] NCCL INFO NVLS Head  5:  5 13
+ip-26-0-173-121:3987801:3988128 [7] NCCL INFO Trees [0] -1/-1/-1->15->14 [1] 8/-1/-1->15->14 [2] 8/-1/-1->15->14 [3] 8/-1/-1->15->14 [4] 8/-1/-1->15->14 [5] 8/-1/-1->15->14 [6] 8/-1/-1->15->14 [7] 8/-1/-1->15->7 [8] -1/-1/-1->15->14 [9] 8/-1/-1->15->14 [10] 8/-1/-1->15->14 [11] 8/-1/-1->15->14 [12] 8/-1/-1->15->14 [13] 8/-1/-1->15->14 [14] 8/-1/-1->15->14 [15] 8/7/-1->15->-1
+ip-26-0-173-121:3987801:3988128 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-173-121:3987800:3988126 [6] NCCL INFO Trees [0] 15/-1/-1->14->13 [1] 15/-1/-1->14->13 [2] 15/-1/-1->14->13 [3] 15/-1/-1->14->13 [4] 15/-1/-1->14->13 [5] 15/-1/-1->14->13 [6] 15/-1/-1->14->6 [7] -1/-1/-1->14->13 [8] 15/-1/-1->14->13 [9] 15/-1/-1->14->13 [10] 15/-1/-1->14->13 [11] 15/-1/-1->14->13 [12] 15/-1/-1->14->13 [13] 15/-1/-1->14->13 [14] 15/6/-1->14->-1 [15] -1/-1/-1->14->13
+ip-26-0-173-121:3987800:3988126 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-142:789262:789590 [7] NCCL INFO NVLS Head  5:  5 13
+ip-26-0-172-142:789261:789587 [6] NCCL INFO NVLS Head  6:  6 14
+ip-26-0-172-142:789256:789593 [1] NCCL INFO NVLS Head  1:  1  9
+ip-26-0-172-142:789262:789590 [7] NCCL INFO NVLS Head  6:  6 14
+ip-26-0-172-142:789259:789589 [4] NCCL INFO NVLS Head  0:  0  8
+ip-26-0-172-142:789261:789587 [6] NCCL INFO NVLS Head  7:  7 15
+ip-26-0-172-142:789255:789586 [0] NCCL INFO NVLS Head  0:  0  8
+ip-26-0-172-142:789260:789591 [5] NCCL INFO NVLS Head  0:  0  8
+ip-26-0-172-142:789256:789593 [1] NCCL INFO NVLS Head  2:  2 10
+ip-26-0-172-142:789262:789590 [7] NCCL INFO NVLS Head  7:  7 15
+ip-26-0-172-142:789259:789589 [4] NCCL INFO NVLS Head  1:  1  9
+ip-26-0-172-142:789260:789591 [5] NCCL INFO NVLS Head  1:  1  9
+ip-26-0-172-142:789256:789593 [1] NCCL INFO NVLS Head  3:  3 11
+ip-26-0-172-142:789255:789586 [0] NCCL INFO NVLS Head  1:  1  9
+ip-26-0-172-142:789259:789589 [4] NCCL INFO NVLS Head  2:  2 10
+ip-26-0-172-142:789256:789593 [1] NCCL INFO NVLS Head  4:  4 12
+ip-26-0-172-142:789260:789591 [5] NCCL INFO NVLS Head  2:  2 10
+ip-26-0-172-142:789257:789588 [2] NCCL INFO NVLS Head  0:  0  8
+ip-26-0-172-142:789255:789586 [0] NCCL INFO NVLS Head  2:  2 10
+ip-26-0-172-142:789259:789589 [4] NCCL INFO NVLS Head  3:  3 11
+ip-26-0-172-142:789261:789587 [6] NCCL INFO Trees [0] 7/-1/-1->6->5 [1] 7/-1/-1->6->5 [2] 7/-1/-1->6->5 [3] 7/-1/-1->6->5 [4] 7/-1/-1->6->5 [5] 7/-1/-1->6->5 [6] 7/14/-1->6->-1 [7] -1/-1/-1->6->5 [8] 7/-1/-1->6->5 [9] 7/-1/-1->6->5 [10] 7/-1/-1->6->5 [11] 7/-1/-1->6->5 [12] 7/-1/-1->6->5 [13] 7/-1/-1->6->5 [14] 7/-1/-1->6->14 [15] -1/-1/-1->6->5
+ip-26-0-172-142:789262:789590 [7] NCCL INFO Trees [0] -1/-1/-1->7->6 [1] 0/-1/-1->7->6 [2] 0/-1/-1->7->6 [3] 0/-1/-1->7->6 [4] 0/-1/-1->7->6 [5] 0/-1/-1->7->6 [6] 0/-1/-1->7->6 [7] 0/15/-1->7->-1 [8] -1/-1/-1->7->6 [9] 0/-1/-1->7->6 [10] 0/-1/-1->7->6 [11] 0/-1/-1->7->6 [12] 0/-1/-1->7->6 [13] 0/-1/-1->7->6 [14] 0/-1/-1->7->6 [15] 0/-1/-1->7->15
+ip-26-0-172-142:789256:789593 [1] NCCL INFO NVLS Head  5:  5 13
+ip-26-0-172-142:789255:789586 [0] NCCL INFO NVLS Head  3:  3 11
+ip-26-0-172-142:789260:789591 [5] NCCL INFO NVLS Head  3:  3 11
+ip-26-0-172-142:789259:789589 [4] NCCL INFO NVLS Head  4:  4 12
+ip-26-0-172-142:789261:789587 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-142:789260:789591 [5] NCCL INFO NVLS Head  4:  4 12
+ip-26-0-172-142:789257:789588 [2] NCCL INFO NVLS Head  1:  1  9
+ip-26-0-172-142:789256:789593 [1] NCCL INFO NVLS Head  6:  6 14
+ip-26-0-172-142:789262:789590 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-142:789255:789586 [0] NCCL INFO NVLS Head  4:  4 12
+ip-26-0-173-121:3987799:3988127 [5] NCCL INFO Trees [0] 14/-1/-1->13->12 [1] 14/-1/-1->13->12 [2] 14/-1/-1->13->12 [3] 14/-1/-1->13->12 [4] 14/-1/-1->13->12 [5] 14/-1/-1->13->5 [6] -1/-1/-1->13->12 [7] 14/-1/-1->13->12 [8] 14/-1/-1->13->12 [9] 14/-1/-1->13->12 [10] 14/-1/-1->13->12 [11] 14/-1/-1->13->12 [12] 14/-1/-1->13->12 [13] 14/5/-1->13->-1 [14] -1/-1/-1->13->12 [15] 14/-1/-1->13->12
+ip-26-0-173-121:3987799:3988127 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-142:789259:789589 [4] NCCL INFO NVLS Head  5:  5 13
+ip-26-0-172-142:789260:789591 [5] NCCL INFO NVLS Head  5:  5 13
+ip-26-0-172-142:789257:789588 [2] NCCL INFO NVLS Head  2:  2 10
+ip-26-0-172-142:789256:789593 [1] NCCL INFO NVLS Head  7:  7 15
+ip-26-0-172-142:789257:789588 [2] NCCL INFO NVLS Head  3:  3 11
+ip-26-0-172-142:789255:789586 [0] NCCL INFO NVLS Head  5:  5 13
+ip-26-0-172-142:789259:789589 [4] NCCL INFO NVLS Head  6:  6 14
+ip-26-0-172-142:789255:789586 [0] NCCL INFO NVLS Head  6:  6 14
+ip-26-0-172-142:789260:789591 [5] NCCL INFO NVLS Head  6:  6 14
+ip-26-0-172-142:789257:789588 [2] NCCL INFO NVLS Head  4:  4 12
+ip-26-0-172-142:789259:789589 [4] NCCL INFO NVLS Head  7:  7 15
+ip-26-0-172-142:789255:789586 [0] NCCL INFO NVLS Head  7:  7 15
+ip-26-0-172-142:789260:789591 [5] NCCL INFO NVLS Head  7:  7 15
+ip-26-0-172-142:789257:789588 [2] NCCL INFO NVLS Head  5:  5 13
+ip-26-0-172-142:789257:789588 [2] NCCL INFO NVLS Head  6:  6 14
+ip-26-0-173-121:3987798:3988129 [4] NCCL INFO Trees [0] 13/-1/-1->12->11 [1] 13/-1/-1->12->11 [2] 13/-1/-1->12->11 [3] 13/-1/-1->12->11 [4] 13/-1/-1->12->4 [5] -1/-1/-1->12->11 [6] 13/-1/-1->12->11 [7] 13/-1/-1->12->11 [8] 13/-1/-1->12->11 [9] 13/-1/-1->12->11 [10] 13/-1/-1->12->11 [11] 13/-1/-1->12->11 [12] 13/4/-1->12->-1 [13] -1/-1/-1->12->11 [14] 13/-1/-1->12->11 [15] 13/-1/-1->12->11
+ip-26-0-173-121:3987798:3988129 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-173-121:3987796:3988122 [2] NCCL INFO Trees [0] 11/-1/-1->10->9 [1] 11/-1/-1->10->9 [2] 11/-1/-1->10->2 [3] -1/-1/-1->10->9 [4] 11/-1/-1->10->9 [5] 11/-1/-1->10->9 [6] 11/-1/-1->10->9 [7] 11/-1/-1->10->9 [8] 11/-1/-1->10->9 [9] 11/-1/-1->10->9 [10] 11/2/-1->10->-1 [11] -1/-1/-1->10->9 [12] 11/-1/-1->10->9 [13] 11/-1/-1->10->9 [14] 11/-1/-1->10->9 [15] 11/-1/-1->10->9
+ip-26-0-172-142:789257:789588 [2] NCCL INFO NVLS Head  7:  7 15
+ip-26-0-172-142:789256:789593 [1] NCCL INFO Trees [0] 2/-1/-1->1->0 [1] 2/9/-1->1->-1 [2] -1/-1/-1->1->0 [3] 2/-1/-1->1->0 [4] 2/-1/-1->1->0 [5] 2/-1/-1->1->0 [6] 2/-1/-1->1->0 [7] 2/-1/-1->1->0 [8] 2/-1/-1->1->0 [9] 2/-1/-1->1->9 [10] -1/-1/-1->1->0 [11] 2/-1/-1->1->0 [12] 2/-1/-1->1->0 [13] 2/-1/-1->1->0 [14] 2/-1/-1->1->0 [15] 2/-1/-1->1->0
+ip-26-0-172-142:789255:789586 [0] NCCL INFO Channel 00/16 :    0   1   2   3   4   5   6   7   8   9  10  11  12  13  14  15
+ip-26-0-172-142:789256:789593 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-173-121:3987796:3988122 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-173-121:3987794:3988123 [0] NCCL INFO Trees [0] 9/-1/-1->8->0 [1] -1/-1/-1->8->15 [2] 9/-1/-1->8->15 [3] 9/-1/-1->8->15 [4] 9/-1/-1->8->15 [5] 9/-1/-1->8->15 [6] 9/-1/-1->8->15 [7] 9/-1/-1->8->15 [8] 9/0/-1->8->-1 [9] -1/-1/-1->8->15 [10] 9/-1/-1->8->15 [11] 9/-1/-1->8->15 [12] 9/-1/-1->8->15 [13] 9/-1/-1->8->15 [14] 9/-1/-1->8->15 [15] 9/-1/-1->8->15
+ip-26-0-173-121:3987794:3988123 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-173-121:3987795:3988124 [1] NCCL INFO Trees [0] 10/-1/-1->9->8 [1] 10/-1/-1->9->1 [2] -1/-1/-1->9->8 [3] 10/-1/-1->9->8 [4] 10/-1/-1->9->8 [5] 10/-1/-1->9->8 [6] 10/-1/-1->9->8 [7] 10/-1/-1->9->8 [8] 10/-1/-1->9->8 [9] 10/1/-1->9->-1 [10] -1/-1/-1->9->8 [11] 10/-1/-1->9->8 [12] 10/-1/-1->9->8 [13] 10/-1/-1->9->8 [14] 10/-1/-1->9->8 [15] 10/-1/-1->9->8
+ip-26-0-172-142:789255:789586 [0] NCCL INFO Channel 01/16 :    0   7   6   5   4   3   2   9   8  15  14  13  12  11  10   1
+ip-26-0-172-142:789259:789589 [4] NCCL INFO Trees [0] 5/-1/-1->4->3 [1] 5/-1/-1->4->3 [2] 5/-1/-1->4->3 [3] 5/-1/-1->4->3 [4] 5/12/-1->4->-1 [5] -1/-1/-1->4->3 [6] 5/-1/-1->4->3 [7] 5/-1/-1->4->3 [8] 5/-1/-1->4->3 [9] 5/-1/-1->4->3 [10] 5/-1/-1->4->3 [11] 5/-1/-1->4->3 [12] 5/-1/-1->4->12 [13] -1/-1/-1->4->3 [14] 5/-1/-1->4->3 [15] 5/-1/-1->4->3
+ip-26-0-172-142:789255:789586 [0] NCCL INFO Channel 02/16 :    0   7   6   5   4   3  10   9   8  15  14  13  12  11   2   1
+ip-26-0-172-142:789260:789591 [5] NCCL INFO Trees [0] 6/-1/-1->5->4 [1] 6/-1/-1->5->4 [2] 6/-1/-1->5->4 [3] 6/-1/-1->5->4 [4] 6/-1/-1->5->4 [5] 6/13/-1->5->-1 [6] -1/-1/-1->5->4 [7] 6/-1/-1->5->4 [8] 6/-1/-1->5->4 [9] 6/-1/-1->5->4 [10] 6/-1/-1->5->4 [11] 6/-1/-1->5->4 [12] 6/-1/-1->5->4 [13] 6/-1/-1->5->13 [14] -1/-1/-1->5->4 [15] 6/-1/-1->5->4
+ip-26-0-172-142:789258:789592 [3] NCCL INFO NVLS Head  0:  0  8
+ip-26-0-173-121:3987795:3988124 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-142:789259:789589 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-142:789255:789586 [0] NCCL INFO Channel 03/16 :    0   7   6   5   4  11  10   9   8  15  14  13  12   3   2   1
+ip-26-0-172-142:789260:789591 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-142:789257:789588 [2] NCCL INFO Trees [0] 3/-1/-1->2->1 [1] 3/-1/-1->2->1 [2] 3/10/-1->2->-1 [3] -1/-1/-1->2->1 [4] 3/-1/-1->2->1 [5] 3/-1/-1->2->1 [6] 3/-1/-1->2->1 [7] 3/-1/-1->2->1 [8] 3/-1/-1->2->1 [9] 3/-1/-1->2->1 [10] 3/-1/-1->2->10 [11] -1/-1/-1->2->1 [12] 3/-1/-1->2->1 [13] 3/-1/-1->2->1 [14] 3/-1/-1->2->1 [15] 3/-1/-1->2->1
+ip-26-0-172-142:789258:789592 [3] NCCL INFO NVLS Head  1:  1  9
+ip-26-0-172-142:789255:789586 [0] NCCL INFO Channel 04/16 :    0   7   6   5  12  11  10   9   8  15  14  13   4   3   2   1
+ip-26-0-172-142:789258:789592 [3] NCCL INFO NVLS Head  2:  2 10
+ip-26-0-172-142:789257:789588 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-142:789258:789592 [3] NCCL INFO NVLS Head  3:  3 11
+ip-26-0-172-142:789255:789586 [0] NCCL INFO Channel 05/16 :    0   4   7   6  13  11  10   9   8  12  15  14   5   3   2   1
+ip-26-0-172-142:789258:789592 [3] NCCL INFO NVLS Head  4:  4 12
+ip-26-0-172-142:789258:789592 [3] NCCL INFO NVLS Head  5:  5 13
+ip-26-0-172-142:789255:789586 [0] NCCL INFO Channel 06/16 :    0   5   4   7  14  11  10   9   8  13  12  15   6   3   2   1
+ip-26-0-172-142:789258:789592 [3] NCCL INFO NVLS Head  6:  6 14
+ip-26-0-172-142:789258:789592 [3] NCCL INFO NVLS Head  7:  7 15
+ip-26-0-172-142:789255:789586 [0] NCCL INFO Channel 07/16 :    0   6   5   4  15  11  10   9   8  14  13  12   7   3   2   1
+ip-26-0-172-142:789255:789586 [0] NCCL INFO Channel 08/16 :    0   1   2   3   4   5   6   7   8   9  10  11  12  13  14  15
+ip-26-0-172-142:789255:789586 [0] NCCL INFO Channel 09/16 :    0   7   6   5   4   3   2   9   8  15  14  13  12  11  10   1
+ip-26-0-172-142:789255:789586 [0] NCCL INFO Channel 10/16 :    0   7   6   5   4   3  10   9   8  15  14  13  12  11   2   1
+ip-26-0-172-142:789255:789586 [0] NCCL INFO Channel 11/16 :    0   7   6   5   4  11  10   9   8  15  14  13  12   3   2   1
+ip-26-0-172-142:789255:789586 [0] NCCL INFO Channel 12/16 :    0   7   6   5  12  11  10   9   8  15  14  13   4   3   2   1
+ip-26-0-172-142:789255:789586 [0] NCCL INFO Channel 13/16 :    0   4   7   6  13  11  10   9   8  12  15  14   5   3   2   1
+ip-26-0-172-142:789258:789592 [3] NCCL INFO Trees [0] 4/-1/-1->3->2 [1] 4/-1/-1->3->2 [2] 4/-1/-1->3->2 [3] 4/11/-1->3->-1 [4] -1/-1/-1->3->2 [5] 4/-1/-1->3->2 [6] 4/-1/-1->3->2 [7] 4/-1/-1->3->2 [8] 4/-1/-1->3->2 [9] 4/-1/-1->3->2 [10] 4/-1/-1->3->2 [11] 4/-1/-1->3->11 [12] -1/-1/-1->3->2 [13] 4/-1/-1->3->2 [14] 4/-1/-1->3->2 [15] 4/-1/-1->3->2
+ip-26-0-172-142:789255:789586 [0] NCCL INFO Channel 14/16 :    0   5   4   7  14  11  10   9   8  13  12  15   6   3   2   1
+ip-26-0-172-142:789255:789586 [0] NCCL INFO Channel 15/16 :    0   6   5   4  15  11  10   9   8  14  13  12   7   3   2   1
+ip-26-0-172-142:789258:789592 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-142:789255:789586 [0] NCCL INFO Trees [0] 1/8/-1->0->-1 [1] -1/-1/-1->0->7 [2] 1/-1/-1->0->7 [3] 1/-1/-1->0->7 [4] 1/-1/-1->0->7 [5] 1/-1/-1->0->7 [6] 1/-1/-1->0->7 [7] 1/-1/-1->0->7 [8] 1/-1/-1->0->8 [9] -1/-1/-1->0->7 [10] 1/-1/-1->0->7 [11] 1/-1/-1->0->7 [12] 1/-1/-1->0->7 [13] 1/-1/-1->0->7 [14] 1/-1/-1->0->7 [15] 1/-1/-1->0->7
+ip-26-0-172-142:789255:789586 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-142:789256:789593 [1] NCCL INFO Channel 00/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-142:789261:789587 [6] NCCL INFO Channel 00/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-173-121:3987797:3988125 [3] NCCL INFO Channel 00/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-172-142:789259:789589 [4] NCCL INFO Channel 00/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-142:789260:789591 [5] NCCL INFO Channel 00/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-142:789258:789592 [3] NCCL INFO Channel 00/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-142:789257:789588 [2] NCCL INFO Channel 00/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-142:789261:789587 [6] NCCL INFO Channel 08/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-173-121:3987800:3988126 [6] NCCL INFO Channel 00/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-172-142:789256:789593 [1] NCCL INFO Channel 08/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-173-121:3987796:3988122 [2] NCCL INFO Channel 00/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-173-121:3987798:3988129 [4] NCCL INFO Channel 00/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-173-121:3987799:3988127 [5] NCCL INFO Channel 00/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-173-121:3987797:3988125 [3] NCCL INFO Channel 08/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-172-142:789259:789589 [4] NCCL INFO Channel 08/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-142:789260:789591 [5] NCCL INFO Channel 08/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-142:789258:789592 [3] NCCL INFO Channel 08/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-173-121:3987800:3988126 [6] NCCL INFO Channel 08/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-172-142:789257:789588 [2] NCCL INFO Channel 08/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-173-121:3987798:3988129 [4] NCCL INFO Channel 08/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-173-121:3987796:3988122 [2] NCCL INFO Channel 08/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-173-121:3987799:3988127 [5] NCCL INFO Channel 08/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-173-121:3987795:3988124 [1] NCCL INFO Channel 00/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-173-121:3987795:3988124 [1] NCCL INFO Channel 08/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-172-142:789259:789589 [4] NCCL INFO Channel 05/0 : 4[4] -> 7[7] via P2P/IPC
+ip-26-0-172-142:789259:789589 [4] NCCL INFO Channel 06/0 : 4[4] -> 7[7] via P2P/IPC
+ip-26-0-172-142:789259:789589 [4] NCCL INFO Channel 13/0 : 4[4] -> 7[7] via P2P/IPC
+ip-26-0-172-142:789259:789589 [4] NCCL INFO Channel 14/0 : 4[4] -> 7[7] via P2P/IPC
+ip-26-0-173-121:3987798:3988129 [4] NCCL INFO Channel 05/0 : 12[4] -> 15[7] via P2P/IPC
+ip-26-0-173-121:3987798:3988129 [4] NCCL INFO Channel 06/0 : 12[4] -> 15[7] via P2P/IPC
+ip-26-0-173-121:3987798:3988129 [4] NCCL INFO Channel 13/0 : 12[4] -> 15[7] via P2P/IPC
+ip-26-0-173-121:3987798:3988129 [4] NCCL INFO Channel 14/0 : 12[4] -> 15[7] via P2P/IPC
+ip-26-0-172-142:789255:789586 [0] NCCL INFO Channel 00/0 : 15[7] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:789262:789590 [7] NCCL INFO Channel 00/0 : 7[7] -> 8[0] [send] via NET/Libfabric/0(0)/GDRDMA
+ip-26-0-172-142:789255:789586 [0] NCCL INFO Channel 08/0 : 15[7] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:789255:789586 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-142:789262:789590 [7] NCCL INFO Channel 08/0 : 7[7] -> 8[0] [send] via NET/Libfabric/0(0)/GDRDMA
+ip-26-0-172-142:789255:789586 [0] NCCL INFO Channel 08/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-142:789257:789588 [2] NCCL INFO Channel 02/0 : 11[3] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:789257:789588 [2] NCCL INFO Channel 10/0 : 11[3] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-121:3987794:3988123 [0] NCCL INFO Channel 00/0 : 7[7] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-121:3987801:3988128 [7] NCCL INFO Channel 00/0 : 15[7] -> 0[0] [send] via NET/Libfabric/0(8)/GDRDMA
+ip-26-0-173-121:3987794:3988123 [0] NCCL INFO Channel 08/0 : 7[7] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-121:3987794:3988123 [0] NCCL INFO Channel 00/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-173-121:3987801:3988128 [7] NCCL INFO Channel 08/0 : 15[7] -> 0[0] [send] via NET/Libfabric/0(8)/GDRDMA
+ip-26-0-172-142:789255:789586 [0] NCCL INFO Channel 05/0 : 0[0] -> 4[4] via P2P/IPC
+ip-26-0-172-142:789255:789586 [0] NCCL INFO Channel 13/0 : 0[0] -> 4[4] via P2P/IPC
+ip-26-0-173-121:3987794:3988123 [0] NCCL INFO Channel 08/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-172-142:789258:789592 [3] NCCL INFO Channel 03/0 : 12[4] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:789258:789592 [3] NCCL INFO Channel 11/0 : 12[4] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:789258:789592 [3] NCCL INFO Channel 02/0 : 3[3] -> 10[2] [send] via NET/Libfabric/2(2)/GDRDMA
+ip-26-0-172-142:789258:789592 [3] NCCL INFO Channel 10/0 : 3[3] -> 10[2] [send] via NET/Libfabric/2(2)/GDRDMA
+ip-26-0-173-121:3987794:3988123 [0] NCCL INFO Channel 05/0 : 8[0] -> 12[4] via P2P/IPC
+ip-26-0-173-121:3987794:3988123 [0] NCCL INFO Channel 13/0 : 8[0] -> 12[4] via P2P/IPC
+ip-26-0-173-121:3987797:3988125 [3] NCCL INFO Channel 03/0 : 4[4] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-121:3987797:3988125 [3] NCCL INFO Channel 11/0 : 4[4] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:789255:789586 [0] NCCL INFO Channel 06/0 : 0[0] -> 5[5] via P2P/IPC
+ip-26-0-172-142:789255:789586 [0] NCCL INFO Channel 14/0 : 0[0] -> 5[5] via P2P/IPC
+ip-26-0-173-121:3987794:3988123 [0] NCCL INFO Channel 06/0 : 8[0] -> 13[5] via P2P/IPC
+ip-26-0-172-142:789255:789586 [0] NCCL INFO Channel 07/0 : 0[0] -> 6[6] via P2P/IPC
+ip-26-0-173-121:3987794:3988123 [0] NCCL INFO Channel 14/0 : 8[0] -> 13[5] via P2P/IPC
+ip-26-0-172-142:789255:789586 [0] NCCL INFO Channel 15/0 : 0[0] -> 6[6] via P2P/IPC
+ip-26-0-173-121:3987796:3988122 [2] NCCL INFO Channel 02/0 : 3[3] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-121:3987794:3988123 [0] NCCL INFO Channel 07/0 : 8[0] -> 14[6] via P2P/IPC
+ip-26-0-173-121:3987796:3988122 [2] NCCL INFO Channel 10/0 : 3[3] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-121:3987797:3988125 [3] NCCL INFO Channel 02/0 : 11[3] -> 2[2] [send] via NET/Libfabric/2(10)/GDRDMA
+ip-26-0-173-121:3987794:3988123 [0] NCCL INFO Channel 15/0 : 8[0] -> 14[6] via P2P/IPC
+ip-26-0-172-142:789255:789586 [0] NCCL INFO Channel 01/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-173-121:3987797:3988125 [3] NCCL INFO Channel 10/0 : 11[3] -> 2[2] [send] via NET/Libfabric/2(10)/GDRDMA
+ip-26-0-173-121:3987796:3988122 [2] NCCL INFO Channel 01/0 : 10[2] -> 1[1] [send] via NET/Libfabric/1(9)/GDRDMA
+ip-26-0-173-121:3987796:3988122 [2] NCCL INFO Channel 09/0 : 10[2] -> 1[1] [send] via NET/Libfabric/1(9)/GDRDMA
+ip-26-0-172-142:789255:789586 [0] NCCL INFO Channel 02/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-173-121:3987794:3988123 [0] NCCL INFO Channel 01/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-172-142:789255:789586 [0] NCCL INFO Channel 03/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-173-121:3987794:3988123 [0] NCCL INFO Channel 02/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-172-142:789255:789586 [0] NCCL INFO Channel 04/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-173-121:3987794:3988123 [0] NCCL INFO Channel 03/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-172-142:789255:789586 [0] NCCL INFO Channel 09/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-173-121:3987794:3988123 [0] NCCL INFO Channel 04/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-172-142:789255:789586 [0] NCCL INFO Channel 10/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-173-121:3987794:3988123 [0] NCCL INFO Channel 09/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-173-121:3987801:3988128 [7] NCCL INFO Channel 06/0 : 15[7] -> 6[6] [send] via NET/Libfabric/6(14)/GDRDMA
+ip-26-0-172-142:789255:789586 [0] NCCL INFO Channel 11/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-173-121:3987794:3988123 [0] NCCL INFO Channel 10/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-172-142:789255:789586 [0] NCCL INFO Channel 12/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-173-121:3987794:3988123 [0] NCCL INFO Channel 11/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-173-121:3987794:3988123 [0] NCCL INFO Channel 12/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-172-142:789256:789593 [1] NCCL INFO Channel 01/0 : 10[2] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:789256:789593 [1] NCCL INFO Channel 09/0 : 10[2] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:789257:789588 [2] NCCL INFO Channel 01/0 : 2[2] -> 9[1] [send] via NET/Libfabric/1(1)/GDRDMA
+ip-26-0-172-142:789257:789588 [2] NCCL INFO Channel 09/0 : 2[2] -> 9[1] [send] via NET/Libfabric/1(1)/GDRDMA
+ip-26-0-172-142:789256:789593 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789256:789593 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789256:789593 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789256:789593 [1] NCCL INFO Channel 04/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987796:3988122 [2] NCCL INFO Channel 02/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-173-121:3987796:3988122 [2] NCCL INFO Channel 03/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-172-142:789256:789593 [1] NCCL INFO Channel 05/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987796:3988122 [2] NCCL INFO Channel 04/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-172-142:789256:789593 [1] NCCL INFO Channel 06/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987796:3988122 [2] NCCL INFO Channel 05/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-172-142:789256:789593 [1] NCCL INFO Channel 07/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987796:3988122 [2] NCCL INFO Channel 06/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-172-142:789256:789593 [1] NCCL INFO Channel 09/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987796:3988122 [2] NCCL INFO Channel 07/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-173-121:3987796:3988122 [2] NCCL INFO Channel 10/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-173-121:3987796:3988122 [2] NCCL INFO Channel 11/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-173-121:3987796:3988122 [2] NCCL INFO Channel 12/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-173-121:3987796:3988122 [2] NCCL INFO Channel 13/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-172-142:789256:789593 [1] NCCL INFO Channel 10/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987796:3988122 [2] NCCL INFO Channel 14/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-172-142:789256:789593 [1] NCCL INFO Channel 11/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987796:3988122 [2] NCCL INFO Channel 15/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-172-142:789256:789593 [1] NCCL INFO Channel 12/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789256:789593 [1] NCCL INFO Channel 13/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789256:789593 [1] NCCL INFO Channel 14/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789256:789593 [1] NCCL INFO Channel 15/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789259:789589 [4] NCCL INFO Channel 04/0 : 13[5] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:789259:789589 [4] NCCL INFO Channel 12/0 : 13[5] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:789259:789589 [4] NCCL INFO Channel 03/0 : 4[4] -> 11[3] [send] via NET/Libfabric/3(3)/GDRDMA
+ip-26-0-172-142:789259:789589 [4] NCCL INFO Channel 11/0 : 4[4] -> 11[3] [send] via NET/Libfabric/3(3)/GDRDMA
+ip-26-0-172-142:789261:789587 [6] NCCL INFO Channel 06/0 : 15[7] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:789262:789590 [7] NCCL INFO Channel 06/0 : 7[7] -> 14[6] [send] via NET/Libfabric/6(6)/GDRDMA
+ip-26-0-172-142:789261:789587 [6] NCCL INFO Channel 14/0 : 15[7] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:789262:789590 [7] NCCL INFO Channel 14/0 : 7[7] -> 14[6] [send] via NET/Libfabric/6(6)/GDRDMA
+ip-26-0-172-142:789260:789591 [5] NCCL INFO Channel 05/0 : 14[6] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:789261:789587 [6] NCCL INFO Channel 05/0 : 6[6] -> 13[5] [send] via NET/Libfabric/5(5)/GDRDMA
+ip-26-0-172-142:789260:789591 [5] NCCL INFO Channel 13/0 : 14[6] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:789261:789587 [6] NCCL INFO Channel 13/0 : 6[6] -> 13[5] [send] via NET/Libfabric/5(5)/GDRDMA
+ip-26-0-172-142:789260:789591 [5] NCCL INFO Channel 04/0 : 5[5] -> 12[4] [send] via NET/Libfabric/4(4)/GDRDMA
+ip-26-0-172-142:789260:789591 [5] NCCL INFO Channel 12/0 : 5[5] -> 12[4] [send] via NET/Libfabric/4(4)/GDRDMA
+ip-26-0-173-121:3987799:3988127 [5] NCCL INFO Channel 05/0 : 6[6] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-121:3987799:3988127 [5] NCCL INFO Channel 13/0 : 6[6] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-121:3987798:3988129 [4] NCCL INFO Channel 04/0 : 5[5] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-121:3987800:3988126 [6] NCCL INFO Channel 06/0 : 7[7] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-121:3987799:3988127 [5] NCCL INFO Channel 04/0 : 13[5] -> 4[4] [send] via NET/Libfabric/4(12)/GDRDMA
+ip-26-0-173-121:3987798:3988129 [4] NCCL INFO Channel 12/0 : 5[5] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-121:3987801:3988128 [7] NCCL INFO Channel 14/0 : 15[7] -> 6[6] [send] via NET/Libfabric/6(14)/GDRDMA
+ip-26-0-173-121:3987800:3988126 [6] NCCL INFO Channel 14/0 : 7[7] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-121:3987799:3988127 [5] NCCL INFO Channel 12/0 : 13[5] -> 4[4] [send] via NET/Libfabric/4(12)/GDRDMA
+ip-26-0-173-121:3987798:3988129 [4] NCCL INFO Channel 03/0 : 12[4] -> 3[3] [send] via NET/Libfabric/3(11)/GDRDMA
+ip-26-0-173-121:3987800:3988126 [6] NCCL INFO Channel 05/0 : 14[6] -> 5[5] [send] via NET/Libfabric/5(13)/GDRDMA
+ip-26-0-173-121:3987800:3988126 [6] NCCL INFO Channel 13/0 : 14[6] -> 5[5] [send] via NET/Libfabric/5(13)/GDRDMA
+ip-26-0-172-142:789261:789587 [6] NCCL INFO Channel 06/0 : 6[6] -> 3[3] via P2P/IPC
+ip-26-0-173-121:3987798:3988129 [4] NCCL INFO Channel 11/0 : 12[4] -> 3[3] [send] via NET/Libfabric/3(11)/GDRDMA
+ip-26-0-173-121:3987799:3988127 [5] NCCL INFO Channel 05/0 : 13[5] -> 11[3] via P2P/IPC
+ip-26-0-173-121:3987800:3988126 [6] NCCL INFO Channel 06/0 : 14[6] -> 11[3] via P2P/IPC
+ip-26-0-172-142:789260:789591 [5] NCCL INFO Channel 05/0 : 5[5] -> 3[3] via P2P/IPC
+ip-26-0-172-142:789261:789587 [6] NCCL INFO Channel 14/0 : 6[6] -> 3[3] via P2P/IPC
+ip-26-0-172-142:789260:789591 [5] NCCL INFO Channel 13/0 : 5[5] -> 3[3] via P2P/IPC
+ip-26-0-172-142:789259:789589 [4] NCCL INFO Channel 07/0 : 4[4] -> 15[7] [send] via NET/Libfabric/7(7)/GDRDMA
+ip-26-0-173-121:3987799:3988127 [5] NCCL INFO Channel 13/0 : 13[5] -> 11[3] via P2P/IPC
+ip-26-0-173-121:3987800:3988126 [6] NCCL INFO Channel 14/0 : 14[6] -> 11[3] via P2P/IPC
+ip-26-0-173-121:3987795:3988124 [1] NCCL INFO Channel 01/0 : 2[2] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-121:3987795:3988124 [1] NCCL INFO Channel 09/0 : 2[2] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-121:3987795:3988124 [1] NCCL INFO Channel 01/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-173-121:3987795:3988124 [1] NCCL INFO Channel 02/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-173-121:3987795:3988124 [1] NCCL INFO Channel 03/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-173-121:3987795:3988124 [1] NCCL INFO Channel 04/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-173-121:3987795:3988124 [1] NCCL INFO Channel 05/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-173-121:3987795:3988124 [1] NCCL INFO Channel 06/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-173-121:3987795:3988124 [1] NCCL INFO Channel 07/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-173-121:3987795:3988124 [1] NCCL INFO Channel 09/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-173-121:3987795:3988124 [1] NCCL INFO Channel 10/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-173-121:3987795:3988124 [1] NCCL INFO Channel 11/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-173-121:3987795:3988124 [1] NCCL INFO Channel 12/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-173-121:3987795:3988124 [1] NCCL INFO Channel 13/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-173-121:3987795:3988124 [1] NCCL INFO Channel 14/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-173-121:3987795:3988124 [1] NCCL INFO Channel 15/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-172-142:789257:789588 [2] NCCL INFO Channel 02/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-142:789257:789588 [2] NCCL INFO Channel 03/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-142:789257:789588 [2] NCCL INFO Channel 04/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-142:789257:789588 [2] NCCL INFO Channel 05/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-142:789257:789588 [2] NCCL INFO Channel 06/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-142:789257:789588 [2] NCCL INFO Channel 07/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-142:789257:789588 [2] NCCL INFO Channel 10/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-142:789257:789588 [2] NCCL INFO Channel 11/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-142:789257:789588 [2] NCCL INFO Channel 12/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-142:789257:789588 [2] NCCL INFO Channel 13/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-142:789257:789588 [2] NCCL INFO Channel 14/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-142:789257:789588 [2] NCCL INFO Channel 15/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-142:789262:789590 [7] NCCL INFO Channel 07/0 : 12[4] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:789259:789589 [4] NCCL INFO Channel 15/0 : 4[4] -> 15[7] [send] via NET/Libfabric/7(7)/GDRDMA
+ip-26-0-172-142:789262:789590 [7] NCCL INFO Channel 15/0 : 12[4] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-121:3987801:3988128 [7] NCCL INFO Channel 07/0 : 4[4] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-121:3987798:3988129 [4] NCCL INFO Channel 07/0 : 12[4] -> 7[7] [send] via NET/Libfabric/7(15)/GDRDMA
+ip-26-0-173-121:3987801:3988128 [7] NCCL INFO Channel 15/0 : 4[4] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-121:3987798:3988129 [4] NCCL INFO Channel 15/0 : 12[4] -> 7[7] [send] via NET/Libfabric/7(15)/GDRDMA
+ip-26-0-173-121:3987801:3988128 [7] NCCL INFO Channel 07/0 : 15[7] -> 11[3] via P2P/IPC
+ip-26-0-172-142:789262:789590 [7] NCCL INFO Channel 07/0 : 7[7] -> 3[3] via P2P/IPC
+ip-26-0-172-142:789262:789590 [7] NCCL INFO Channel 15/0 : 7[7] -> 3[3] via P2P/IPC
+ip-26-0-173-121:3987801:3988128 [7] NCCL INFO Channel 15/0 : 15[7] -> 11[3] via P2P/IPC
+ip-26-0-172-142:789262:789590 [7] NCCL INFO Channel 01/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-142:789262:789590 [7] NCCL INFO Channel 02/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-173-121:3987801:3988128 [7] NCCL INFO Channel 01/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-172-142:789262:789590 [7] NCCL INFO Channel 03/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-142:789262:789590 [7] NCCL INFO Channel 04/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-173-121:3987801:3988128 [7] NCCL INFO Channel 02/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-172-142:789262:789590 [7] NCCL INFO Channel 05/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-142:789262:789590 [7] NCCL INFO Channel 09/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-142:789262:789590 [7] NCCL INFO Channel 10/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-142:789262:789590 [7] NCCL INFO Channel 11/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-173-121:3987801:3988128 [7] NCCL INFO Channel 03/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-172-142:789262:789590 [7] NCCL INFO Channel 12/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-142:789262:789590 [7] NCCL INFO Channel 13/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-142:789259:789589 [4] NCCL INFO Channel 01/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-142:789258:789592 [3] NCCL INFO Channel 01/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-142:789259:789589 [4] NCCL INFO Channel 02/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-142:789258:789592 [3] NCCL INFO Channel 03/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-142:789259:789589 [4] NCCL INFO Channel 04/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-142:789261:789587 [6] NCCL INFO Channel 01/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-142:789258:789592 [3] NCCL INFO Channel 04/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-142:789259:789589 [4] NCCL INFO Channel 09/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-142:789261:789587 [6] NCCL INFO Channel 02/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-142:789259:789589 [4] NCCL INFO Channel 10/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-142:789258:789592 [3] NCCL INFO Channel 05/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-142:789261:789587 [6] NCCL INFO Channel 03/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-142:789259:789589 [4] NCCL INFO Channel 12/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-142:789258:789592 [3] NCCL INFO Channel 06/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-142:789261:789587 [6] NCCL INFO Channel 04/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-142:789258:789592 [3] NCCL INFO Channel 07/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-142:789261:789587 [6] NCCL INFO Channel 07/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-142:789258:789592 [3] NCCL INFO Channel 09/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-142:789261:789587 [6] NCCL INFO Channel 09/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-142:789260:789591 [5] NCCL INFO Channel 01/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-142:789258:789592 [3] NCCL INFO Channel 11/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-142:789261:789587 [6] NCCL INFO Channel 10/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-142:789260:789591 [5] NCCL INFO Channel 02/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-142:789258:789592 [3] NCCL INFO Channel 12/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-142:789261:789587 [6] NCCL INFO Channel 11/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-142:789260:789591 [5] NCCL INFO Channel 03/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-173-121:3987801:3988128 [7] NCCL INFO Channel 04/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-172-142:789261:789587 [6] NCCL INFO Channel 12/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-142:789258:789592 [3] NCCL INFO Channel 13/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-142:789260:789591 [5] NCCL INFO Channel 06/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-173-121:3987801:3988128 [7] NCCL INFO Channel 05/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-173-121:3987801:3988128 [7] NCCL INFO Channel 09/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-173-121:3987801:3988128 [7] NCCL INFO Channel 10/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-173-121:3987798:3988129 [4] NCCL INFO Channel 01/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-173-121:3987801:3988128 [7] NCCL INFO Channel 11/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-173-121:3987798:3988129 [4] NCCL INFO Channel 02/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-173-121:3987801:3988128 [7] NCCL INFO Channel 12/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-173-121:3987798:3988129 [4] NCCL INFO Channel 04/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-173-121:3987797:3988125 [3] NCCL INFO Channel 01/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-173-121:3987801:3988128 [7] NCCL INFO Channel 13/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-173-121:3987798:3988129 [4] NCCL INFO Channel 09/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-173-121:3987797:3988125 [3] NCCL INFO Channel 03/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-173-121:3987798:3988129 [4] NCCL INFO Channel 10/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-173-121:3987797:3988125 [3] NCCL INFO Channel 04/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-173-121:3987798:3988129 [4] NCCL INFO Channel 12/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-173-121:3987797:3988125 [3] NCCL INFO Channel 05/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-173-121:3987797:3988125 [3] NCCL INFO Channel 06/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-173-121:3987797:3988125 [3] NCCL INFO Channel 07/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-173-121:3987800:3988126 [6] NCCL INFO Channel 01/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-173-121:3987797:3988125 [3] NCCL INFO Channel 09/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-173-121:3987799:3988127 [5] NCCL INFO Channel 01/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-173-121:3987800:3988126 [6] NCCL INFO Channel 02/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-173-121:3987797:3988125 [3] NCCL INFO Channel 11/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-173-121:3987799:3988127 [5] NCCL INFO Channel 02/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-173-121:3987800:3988126 [6] NCCL INFO Channel 03/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-173-121:3987799:3988127 [5] NCCL INFO Channel 03/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-173-121:3987797:3988125 [3] NCCL INFO Channel 12/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-173-121:3987800:3988126 [6] NCCL INFO Channel 04/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-173-121:3987797:3988125 [3] NCCL INFO Channel 13/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-173-121:3987799:3988127 [5] NCCL INFO Channel 06/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-173-121:3987800:3988126 [6] NCCL INFO Channel 07/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-173-121:3987797:3988125 [3] NCCL INFO Channel 14/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-173-121:3987799:3988127 [5] NCCL INFO Channel 07/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-173-121:3987800:3988126 [6] NCCL INFO Channel 09/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-173-121:3987799:3988127 [5] NCCL INFO Channel 09/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-173-121:3987797:3988125 [3] NCCL INFO Channel 15/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-173-121:3987800:3988126 [6] NCCL INFO Channel 10/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-173-121:3987799:3988127 [5] NCCL INFO Channel 10/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-173-121:3987800:3988126 [6] NCCL INFO Channel 11/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-173-121:3987799:3988127 [5] NCCL INFO Channel 11/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-173-121:3987800:3988126 [6] NCCL INFO Channel 12/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-173-121:3987799:3988127 [5] NCCL INFO Channel 14/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-173-121:3987800:3988126 [6] NCCL INFO Channel 15/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-173-121:3987799:3988127 [5] NCCL INFO Channel 15/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-172-142:789261:789587 [6] NCCL INFO Channel 15/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-142:789258:789592 [3] NCCL INFO Channel 14/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-142:789260:789591 [5] NCCL INFO Channel 07/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-142:789258:789592 [3] NCCL INFO Channel 15/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-142:789260:789591 [5] NCCL INFO Channel 09/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-142:789260:789591 [5] NCCL INFO Channel 10/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-142:789260:789591 [5] NCCL INFO Channel 11/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-142:789260:789591 [5] NCCL INFO Channel 14/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-142:789260:789591 [5] NCCL INFO Channel 15/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-142:789256:789593 [1] NCCL INFO Connected all rings
+ip-26-0-173-121:3987794:3988123 [0] NCCL INFO Connected all rings
+ip-26-0-173-121:3987794:3988123 [0] NCCL INFO Channel 02/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-173-121:3987794:3988123 [0] NCCL INFO Channel 03/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-173-121:3987794:3988123 [0] NCCL INFO Channel 04/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-173-121:3987794:3988123 [0] NCCL INFO Channel 05/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-173-121:3987794:3988123 [0] NCCL INFO Channel 06/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-173-121:3987794:3988123 [0] NCCL INFO Channel 07/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-173-121:3987794:3988123 [0] NCCL INFO Channel 10/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-173-121:3987794:3988123 [0] NCCL INFO Channel 11/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-173-121:3987794:3988123 [0] NCCL INFO Channel 12/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-173-121:3987794:3988123 [0] NCCL INFO Channel 13/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-173-121:3987794:3988123 [0] NCCL INFO Channel 14/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-173-121:3987794:3988123 [0] NCCL INFO Channel 15/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-172-142:789255:789586 [0] NCCL INFO Connected all rings
+ip-26-0-172-142:789255:789586 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-142:789255:789586 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-142:789255:789586 [0] NCCL INFO Channel 04/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-142:789255:789586 [0] NCCL INFO Channel 05/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-142:789255:789586 [0] NCCL INFO Channel 06/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-142:789255:789586 [0] NCCL INFO Channel 07/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-173-121:3987796:3988122 [2] NCCL INFO Connected all rings
+ip-26-0-173-121:3987798:3988129 [4] NCCL INFO Connected all rings
+ip-26-0-173-121:3987800:3988126 [6] NCCL INFO Connected all rings
+ip-26-0-172-142:789257:789588 [2] NCCL INFO Connected all rings
+ip-26-0-173-121:3987795:3988124 [1] NCCL INFO Connected all rings
+ip-26-0-172-142:789255:789586 [0] NCCL INFO Channel 10/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-173-121:3987799:3988127 [5] NCCL INFO Connected all rings
+ip-26-0-173-121:3987801:3988128 [7] NCCL INFO Connected all rings
+ip-26-0-172-142:789255:789586 [0] NCCL INFO Channel 11/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-142:789256:789593 [1] NCCL INFO Channel 01/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-142:789255:789586 [0] NCCL INFO Channel 12/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-142:789256:789593 [1] NCCL INFO Channel 03/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-142:789256:789593 [1] NCCL INFO Channel 04/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-142:789255:789586 [0] NCCL INFO Channel 13/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-142:789256:789593 [1] NCCL INFO Channel 05/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-142:789255:789586 [0] NCCL INFO Channel 14/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-142:789256:789593 [1] NCCL INFO Channel 06/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-142:789255:789586 [0] NCCL INFO Channel 15/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-142:789256:789593 [1] NCCL INFO Channel 07/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-142:789256:789593 [1] NCCL INFO Channel 09/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-142:789256:789593 [1] NCCL INFO Channel 11/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-142:789256:789593 [1] NCCL INFO Channel 12/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-173-121:3987796:3988122 [2] NCCL INFO Channel 01/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-172-142:789256:789593 [1] NCCL INFO Channel 13/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-173-121:3987796:3988122 [2] NCCL INFO Channel 02/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-172-142:789256:789593 [1] NCCL INFO Channel 14/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-173-121:3987800:3988126 [6] NCCL INFO Channel 01/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-173-121:3987798:3988129 [4] NCCL INFO Channel 01/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-173-121:3987796:3988122 [2] NCCL INFO Channel 04/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-172-142:789256:789593 [1] NCCL INFO Channel 15/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-142:789260:789591 [5] NCCL INFO Connected all rings
+ip-26-0-173-121:3987795:3988124 [1] NCCL INFO Channel 01/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-172-142:789255:789586 [0] NCCL INFO Channel 05/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-173-121:3987800:3988126 [6] NCCL INFO Channel 02/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-172-142:789255:789586 [0] NCCL INFO Channel 06/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-173-121:3987800:3988126 [6] NCCL INFO Channel 03/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-173-121:3987796:3988122 [2] NCCL INFO Channel 05/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-173-121:3987800:3988126 [6] NCCL INFO Channel 04/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-172-142:789255:789586 [0] NCCL INFO Channel 07/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-172-142:789255:789586 [0] NCCL INFO Channel 13/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-172-142:789255:789586 [0] NCCL INFO Channel 14/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-173-121:3987795:3988124 [1] NCCL INFO Channel 03/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-173-121:3987800:3988126 [6] NCCL INFO Channel 05/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-173-121:3987799:3988127 [5] NCCL INFO Channel 01/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-173-121:3987798:3988129 [4] NCCL INFO Channel 02/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-172-142:789258:789592 [3] NCCL INFO Connected all rings
+ip-26-0-173-121:3987795:3988124 [1] NCCL INFO Channel 04/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-172-142:789255:789586 [0] NCCL INFO Channel 15/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-173-121:3987795:3988124 [1] NCCL INFO Channel 05/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-173-121:3987796:3988122 [2] NCCL INFO Channel 06/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-173-121:3987799:3988127 [5] NCCL INFO Channel 02/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-173-121:3987798:3988129 [4] NCCL INFO Channel 03/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-173-121:3987800:3988126 [6] NCCL INFO Channel 06/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-173-121:3987796:3988122 [2] NCCL INFO Channel 07/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-173-121:3987795:3988124 [1] NCCL INFO Channel 06/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-173-121:3987799:3988127 [5] NCCL INFO Channel 03/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-173-121:3987798:3988129 [4] NCCL INFO Channel 04/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-173-121:3987800:3988126 [6] NCCL INFO Channel 09/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-172-142:789257:789588 [2] NCCL INFO Channel 01/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-173-121:3987799:3988127 [5] NCCL INFO Channel 04/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-172-142:789257:789588 [2] NCCL INFO Channel 02/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-173-121:3987795:3988124 [1] NCCL INFO Channel 07/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-173-121:3987800:3988126 [6] NCCL INFO Channel 10/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-172-142:789257:789588 [2] NCCL INFO Channel 04/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-173-121:3987799:3988127 [5] NCCL INFO Channel 05/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-173-121:3987796:3988122 [2] NCCL INFO Channel 09/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-172-142:789257:789588 [2] NCCL INFO Channel 05/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-173-121:3987795:3988124 [1] NCCL INFO Channel 09/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-173-121:3987800:3988126 [6] NCCL INFO Channel 11/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-173-121:3987796:3988122 [2] NCCL INFO Channel 10/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-173-121:3987799:3988127 [5] NCCL INFO Channel 07/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-172-142:789257:789588 [2] NCCL INFO Channel 06/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-173-121:3987795:3988124 [1] NCCL INFO Channel 11/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-172-142:789257:789588 [2] NCCL INFO Channel 07/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-173-121:3987800:3988126 [6] NCCL INFO Channel 12/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-173-121:3987798:3988129 [4] NCCL INFO Channel 06/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-173-121:3987796:3988122 [2] NCCL INFO Channel 12/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-173-121:3987799:3988127 [5] NCCL INFO Channel 09/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-172-142:789257:789588 [2] NCCL INFO Channel 09/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-142:789257:789588 [2] NCCL INFO Channel 10/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-173-121:3987795:3988124 [1] NCCL INFO Channel 12/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-173-121:3987800:3988126 [6] NCCL INFO Channel 13/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-173-121:3987798:3988129 [4] NCCL INFO Channel 07/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-172-142:789257:789588 [2] NCCL INFO Channel 12/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-142:789262:789590 [7] NCCL INFO Connected all rings
+ip-26-0-173-121:3987796:3988122 [2] NCCL INFO Channel 13/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-173-121:3987799:3988127 [5] NCCL INFO Channel 10/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-173-121:3987795:3988124 [1] NCCL INFO Channel 13/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-172-142:789257:789588 [2] NCCL INFO Channel 13/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-173-121:3987800:3988126 [6] NCCL INFO Channel 14/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-173-121:3987798:3988129 [4] NCCL INFO Channel 09/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-173-121:3987796:3988122 [2] NCCL INFO Channel 14/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-173-121:3987799:3988127 [5] NCCL INFO Channel 11/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-173-121:3987795:3988124 [1] NCCL INFO Channel 14/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-173-121:3987798:3988129 [4] NCCL INFO Channel 10/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-173-121:3987796:3988122 [2] NCCL INFO Channel 15/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-173-121:3987799:3988127 [5] NCCL INFO Channel 12/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-172-142:789259:789589 [4] NCCL INFO Connected all rings
+ip-26-0-173-121:3987795:3988124 [1] NCCL INFO Channel 15/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-173-121:3987798:3988129 [4] NCCL INFO Channel 11/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-172-142:789257:789588 [2] NCCL INFO Channel 14/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-173-121:3987799:3988127 [5] NCCL INFO Channel 13/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-173-121:3987798:3988129 [4] NCCL INFO Channel 12/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-173-121:3987794:3988123 [0] NCCL INFO Channel 05/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-172-142:789257:789588 [2] NCCL INFO Channel 15/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-173-121:3987795:3988124 [1] NCCL INFO Channel 01/0 : 1[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-121:3987795:3988124 [1] NCCL INFO Channel 09/0 : 1[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-121:3987799:3988127 [5] NCCL INFO Channel 15/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-173-121:3987795:3988124 [1] NCCL INFO Channel 01/0 : 9[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-121:3987795:3988124 [1] NCCL INFO Channel 09/0 : 9[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-121:3987798:3988129 [4] NCCL INFO Channel 14/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-173-121:3987794:3988123 [0] NCCL INFO Channel 06/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-172-142:789256:789593 [1] NCCL INFO Channel 01/0 : 9[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:789256:789593 [1] NCCL INFO Channel 09/0 : 9[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:789256:789593 [1] NCCL INFO Channel 01/0 : 1[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-121:3987798:3988129 [4] NCCL INFO Channel 15/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-172-142:789256:789593 [1] NCCL INFO Channel 09/0 : 1[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-121:3987800:3988126 [6] NCCL INFO Channel 06/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-121:3987800:3988126 [6] NCCL INFO Channel 14/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-121:3987800:3988126 [6] NCCL INFO Channel 06/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-121:3987794:3988123 [0] NCCL INFO Channel 07/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-173-121:3987800:3988126 [6] NCCL INFO Channel 14/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-121:3987799:3988127 [5] NCCL INFO Channel 05/0 : 5[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-121:3987799:3988127 [5] NCCL INFO Channel 13/0 : 5[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-121:3987794:3988123 [0] NCCL INFO Channel 13/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-173-121:3987799:3988127 [5] NCCL INFO Channel 05/0 : 13[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-121:3987799:3988127 [5] NCCL INFO Channel 13/0 : 13[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-121:3987794:3988123 [0] NCCL INFO Channel 14/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-173-121:3987794:3988123 [0] NCCL INFO Channel 15/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-172-142:789261:789587 [6] NCCL INFO Connected all rings
+ip-26-0-173-121:3987795:3988124 [1] NCCL INFO Channel 00/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-173-121:3987801:3988128 [7] NCCL INFO Channel 07/0 : 7[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-121:3987794:3988123 [0] NCCL INFO Channel 00/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-121:3987795:3988124 [1] NCCL INFO Channel 08/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-173-121:3987794:3988123 [0] NCCL INFO Channel 08/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-121:3987801:3988128 [7] NCCL INFO Channel 15/0 : 7[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-121:3987794:3988123 [0] NCCL INFO Channel 00/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-121:3987801:3988128 [7] NCCL INFO Channel 07/0 : 15[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-121:3987794:3988123 [0] NCCL INFO Channel 08/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-121:3987801:3988128 [7] NCCL INFO Channel 15/0 : 15[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:789256:789593 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789256:789593 [1] NCCL INFO Channel 08/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987797:3988125 [3] NCCL INFO Connected all rings
+ip-26-0-172-142:789258:789592 [3] NCCL INFO Channel 01/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-142:789259:789589 [4] NCCL INFO Channel 01/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-142:789258:789592 [3] NCCL INFO Channel 02/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-142:789259:789589 [4] NCCL INFO Channel 02/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-142:789258:789592 [3] NCCL INFO Channel 03/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-142:789259:789589 [4] NCCL INFO Channel 03/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-142:789258:789592 [3] NCCL INFO Channel 05/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-142:789259:789589 [4] NCCL INFO Channel 04/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-142:789260:789591 [5] NCCL INFO Channel 01/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-142:789258:789592 [3] NCCL INFO Channel 06/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-173-121:3987797:3988125 [3] NCCL INFO Channel 01/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-172-142:789259:789589 [4] NCCL INFO Channel 06/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-173-121:3987797:3988125 [3] NCCL INFO Channel 02/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-173-121:3987797:3988125 [3] NCCL INFO Channel 03/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-172-142:789260:789591 [5] NCCL INFO Channel 02/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-142:789258:789592 [3] NCCL INFO Channel 07/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-142:789259:789589 [4] NCCL INFO Channel 07/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-173-121:3987797:3988125 [3] NCCL INFO Channel 05/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-173-121:3987797:3988125 [3] NCCL INFO Channel 06/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-172-142:789260:789591 [5] NCCL INFO Channel 03/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-142:789261:789587 [6] NCCL INFO Channel 01/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-142:789258:789592 [3] NCCL INFO Channel 09/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-173-121:3987797:3988125 [3] NCCL INFO Channel 07/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-172-142:789259:789589 [4] NCCL INFO Channel 09/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-173-121:3987797:3988125 [3] NCCL INFO Channel 09/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-173-121:3987797:3988125 [3] NCCL INFO Channel 10/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-172-142:789260:789591 [5] NCCL INFO Channel 04/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-142:789261:789587 [6] NCCL INFO Channel 02/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-142:789258:789592 [3] NCCL INFO Channel 10/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-142:789259:789589 [4] NCCL INFO Channel 10/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-173-121:3987797:3988125 [3] NCCL INFO Channel 11/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-173-121:3987797:3988125 [3] NCCL INFO Channel 13/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-173-121:3987797:3988125 [3] NCCL INFO Channel 14/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-172-142:789260:789591 [5] NCCL INFO Channel 05/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-142:789261:789587 [6] NCCL INFO Channel 03/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-142:789258:789592 [3] NCCL INFO Channel 11/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-142:789259:789589 [4] NCCL INFO Channel 11/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-173-121:3987797:3988125 [3] NCCL INFO Channel 15/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-173-121:3987796:3988122 [2] NCCL INFO Channel 02/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:789261:789587 [6] NCCL INFO Channel 04/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-142:789260:789591 [5] NCCL INFO Channel 07/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-142:789258:789592 [3] NCCL INFO Channel 13/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-142:789259:789589 [4] NCCL INFO Channel 12/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-173-121:3987798:3988129 [4] NCCL INFO Channel 04/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-121:3987796:3988122 [2] NCCL INFO Channel 10/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-121:3987797:3988125 [3] NCCL INFO Channel 03/0 : 3[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-121:3987798:3988129 [4] NCCL INFO Channel 12/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-121:3987796:3988122 [2] NCCL INFO Channel 02/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-121:3987797:3988125 [3] NCCL INFO Channel 11/0 : 3[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-121:3987798:3988129 [4] NCCL INFO Channel 04/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-121:3987796:3988122 [2] NCCL INFO Channel 10/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-121:3987797:3988125 [3] NCCL INFO Channel 03/0 : 11[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-121:3987798:3988129 [4] NCCL INFO Channel 12/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-121:3987797:3988125 [3] NCCL INFO Channel 11/0 : 11[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:789261:789587 [6] NCCL INFO Channel 05/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-142:789260:789591 [5] NCCL INFO Channel 09/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-142:789258:789592 [3] NCCL INFO Channel 14/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-142:789259:789589 [4] NCCL INFO Channel 14/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-142:789260:789591 [5] NCCL INFO Channel 10/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-142:789258:789592 [3] NCCL INFO Channel 15/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-142:789261:789587 [6] NCCL INFO Channel 06/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-142:789259:789589 [4] NCCL INFO Channel 15/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-142:789260:789591 [5] NCCL INFO Channel 11/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-142:789261:789587 [6] NCCL INFO Channel 09/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-142:789257:789588 [2] NCCL INFO Channel 02/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:789260:789591 [5] NCCL INFO Channel 12/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-142:789257:789588 [2] NCCL INFO Channel 10/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:789258:789592 [3] NCCL INFO Channel 03/0 : 11[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:789261:789587 [6] NCCL INFO Channel 10/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-142:789257:789588 [2] NCCL INFO Channel 02/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:789258:789592 [3] NCCL INFO Channel 11/0 : 11[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:789257:789588 [2] NCCL INFO Channel 10/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:789258:789592 [3] NCCL INFO Channel 03/0 : 3[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:789258:789592 [3] NCCL INFO Channel 11/0 : 3[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:789260:789591 [5] NCCL INFO Channel 13/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-142:789261:789587 [6] NCCL INFO Channel 11/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-142:789260:789591 [5] NCCL INFO Channel 15/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-142:789261:789587 [6] NCCL INFO Channel 12/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-142:789261:789587 [6] NCCL INFO Channel 13/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-142:789259:789589 [4] NCCL INFO Channel 04/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:789259:789589 [4] NCCL INFO Channel 12/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:789259:789589 [4] NCCL INFO Channel 04/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:789259:789589 [4] NCCL INFO Channel 12/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:789261:789587 [6] NCCL INFO Channel 14/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-173-121:3987796:3988122 [2] NCCL INFO Channel 00/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-172-142:789260:789591 [5] NCCL INFO Channel 05/0 : 13[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:789260:789591 [5] NCCL INFO Channel 13/0 : 13[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:789261:789587 [6] NCCL INFO Channel 06/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:789261:789587 [6] NCCL INFO Channel 14/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:789260:789591 [5] NCCL INFO Channel 05/0 : 5[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:789260:789591 [5] NCCL INFO Channel 13/0 : 5[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:789261:789587 [6] NCCL INFO Channel 06/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:789261:789587 [6] NCCL INFO Channel 14/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:789257:789588 [2] NCCL INFO Channel 00/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-173-121:3987796:3988122 [2] NCCL INFO Channel 01/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-172-142:789257:789588 [2] NCCL INFO Channel 01/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-173-121:3987796:3988122 [2] NCCL INFO Channel 08/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-173-121:3987798:3988129 [4] NCCL INFO Channel 00/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-173-121:3987796:3988122 [2] NCCL INFO Channel 09/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-172-142:789257:789588 [2] NCCL INFO Channel 08/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-142:789259:789589 [4] NCCL INFO Channel 00/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-173-121:3987797:3988125 [3] NCCL INFO Channel 00/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-173-121:3987797:3988125 [3] NCCL INFO Channel 02/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-172-142:789257:789588 [2] NCCL INFO Channel 09/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-173-121:3987799:3988127 [5] NCCL INFO Channel 00/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-172-142:789259:789589 [4] NCCL INFO Channel 03/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-173-121:3987797:3988125 [3] NCCL INFO Channel 08/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-173-121:3987799:3988127 [5] NCCL INFO Channel 04/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-173-121:3987797:3988125 [3] NCCL INFO Channel 10/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-172-142:789260:789591 [5] NCCL INFO Channel 00/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-142:789259:789589 [4] NCCL INFO Channel 05/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-142:789261:789587 [6] NCCL INFO Channel 00/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-173-121:3987799:3988127 [5] NCCL INFO Channel 08/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-173-121:3987798:3988129 [4] NCCL INFO Channel 03/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-173-121:3987799:3988127 [5] NCCL INFO Channel 12/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-172-142:789258:789592 [3] NCCL INFO Channel 00/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-173-121:3987798:3988129 [4] NCCL INFO Channel 05/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-172-142:789260:789591 [5] NCCL INFO Channel 04/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-142:789259:789589 [4] NCCL INFO Channel 06/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-142:789261:789587 [6] NCCL INFO Channel 05/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-173-121:3987798:3988129 [4] NCCL INFO Channel 06/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-173-121:3987798:3988129 [4] NCCL INFO Channel 07/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-172-142:789258:789592 [3] NCCL INFO Channel 02/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-173-121:3987800:3988126 [6] NCCL INFO Channel 00/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-172-142:789262:789590 [7] NCCL INFO Channel 07/0 : 15[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:789255:789586 [0] NCCL INFO Channel 00/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-121:3987798:3988129 [4] NCCL INFO Channel 08/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-172-142:789260:789591 [5] NCCL INFO Channel 08/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-142:789259:789589 [4] NCCL INFO Channel 07/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-142:789261:789587 [6] NCCL INFO Channel 08/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-142:789262:789590 [7] NCCL INFO Channel 15/0 : 15[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:789255:789586 [0] NCCL INFO Channel 08/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:789262:789590 [7] NCCL INFO Channel 07/0 : 7[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-121:3987800:3988126 [6] NCCL INFO Channel 05/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-172-142:789255:789586 [0] NCCL INFO Channel 00/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:789262:789590 [7] NCCL INFO Channel 15/0 : 7[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:789255:789586 [0] NCCL INFO Channel 08/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-121:3987798:3988129 [4] NCCL INFO Channel 11/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-172-142:789262:789590 [7] NCCL INFO Channel 01/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987801:3988128 [7] NCCL INFO Channel 01/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-173-121:3987800:3988126 [6] NCCL INFO Channel 08/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-173-121:3987798:3988129 [4] NCCL INFO Channel 13/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-172-142:789258:789592 [3] NCCL INFO Channel 08/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-142:789261:789587 [6] NCCL INFO Channel 13/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-142:789260:789591 [5] NCCL INFO Channel 12/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-142:789258:789592 [3] NCCL INFO Channel 10/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-142:789259:789589 [4] NCCL INFO Channel 08/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-142:789259:789589 [4] NCCL INFO Channel 11/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-173-121:3987801:3988128 [7] NCCL INFO Channel 02/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-172-142:789262:789590 [7] NCCL INFO Channel 02/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987800:3988126 [6] NCCL INFO Channel 13/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-173-121:3987798:3988129 [4] NCCL INFO Channel 14/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-172-142:789259:789589 [4] NCCL INFO Channel 13/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-142:789262:789590 [7] NCCL INFO Channel 03/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987801:3988128 [7] NCCL INFO Channel 03/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-173-121:3987798:3988129 [4] NCCL INFO Channel 15/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-173-121:3987801:3988128 [7] NCCL INFO Channel 04/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-172-142:789259:789589 [4] NCCL INFO Channel 14/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-142:789262:789590 [7] NCCL INFO Channel 04/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987801:3988128 [7] NCCL INFO Channel 05/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-172-142:789259:789589 [4] NCCL INFO Channel 15/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-142:789262:789590 [7] NCCL INFO Channel 05/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789262:789590 [7] NCCL INFO Channel 06/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987801:3988128 [7] NCCL INFO Channel 06/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-172-142:789262:789590 [7] NCCL INFO Channel 07/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987801:3988128 [7] NCCL INFO Channel 07/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-172-142:789262:789590 [7] NCCL INFO Channel 09/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987801:3988128 [7] NCCL INFO Channel 09/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-173-121:3987801:3988128 [7] NCCL INFO Channel 10/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-172-142:789262:789590 [7] NCCL INFO Channel 10/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987801:3988128 [7] NCCL INFO Channel 11/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-173-121:3987801:3988128 [7] NCCL INFO Channel 12/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-172-142:789262:789590 [7] NCCL INFO Channel 11/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789262:789590 [7] NCCL INFO Channel 12/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987801:3988128 [7] NCCL INFO Channel 13/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-172-142:789262:789590 [7] NCCL INFO Channel 13/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987801:3988128 [7] NCCL INFO Channel 14/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-172-142:789262:789590 [7] NCCL INFO Channel 14/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789262:789590 [7] NCCL INFO Channel 15/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987801:3988128 [7] NCCL INFO Channel 15/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-173-121:3987801:3988128 [7] NCCL INFO Channel 00/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-173-121:3987801:3988128 [7] NCCL INFO Channel 06/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-173-121:3987801:3988128 [7] NCCL INFO Channel 08/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-173-121:3987801:3988128 [7] NCCL INFO Channel 14/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-172-142:789262:789590 [7] NCCL INFO Channel 00/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-142:789262:789590 [7] NCCL INFO Channel 06/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-142:789262:789590 [7] NCCL INFO Channel 08/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-142:789262:789590 [7] NCCL INFO Channel 14/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-142:789258:789592 [3] NCCL INFO Connected all trees
+ip-26-0-172-142:789258:789592 [3] NCCL INFO NVLS comm 0x8d44de0 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-142:789257:789588 [2] NCCL INFO Connected all trees
+ip-26-0-172-142:789257:789588 [2] NCCL INFO NVLS comm 0x95c0680 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-142:789259:789589 [4] NCCL INFO Connected all trees
+ip-26-0-172-142:789260:789591 [5] NCCL INFO Connected all trees
+ip-26-0-172-142:789259:789589 [4] NCCL INFO NVLS comm 0x92665b0 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-173-121:3987796:3988122 [2] NCCL INFO Connected all trees
+ip-26-0-172-142:789260:789591 [5] NCCL INFO NVLS comm 0xa28a6a0 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-173-121:3987796:3988122 [2] NCCL INFO NVLS comm 0x92c5260 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-173-121:3987799:3988127 [5] NCCL INFO Connected all trees
+ip-26-0-173-121:3987799:3988127 [5] NCCL INFO NVLS comm 0x93f88d0 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-142:789256:789593 [1] NCCL INFO Connected all trees
+ip-26-0-172-142:789256:789593 [1] NCCL INFO NVLS comm 0xa553960 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-173-121:3987797:3988125 [3] NCCL INFO Connected all trees
+ip-26-0-173-121:3987798:3988129 [4] NCCL INFO Connected all trees
+ip-26-0-173-121:3987797:3988125 [3] NCCL INFO NVLS comm 0x8d9bd80 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-173-121:3987798:3988129 [4] NCCL INFO NVLS comm 0x9df7d00 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-142:789255:789586 [0] NCCL INFO Connected all trees
+ip-26-0-172-142:789262:789590 [7] NCCL INFO Connected all trees
+ip-26-0-172-142:789255:789586 [0] NCCL INFO NVLS comm 0xa1ffd50 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-142:789262:789590 [7] NCCL INFO NVLS comm 0x8c22120 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-173-121:3987794:3988123 [0] NCCL INFO Connected all trees
+ip-26-0-173-121:3987794:3988123 [0] NCCL INFO NVLS comm 0x8e641b0 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-173-121:3987801:3988128 [7] NCCL INFO Connected all trees
+ip-26-0-173-121:3987801:3988128 [7] NCCL INFO NVLS comm 0x9851480 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-142:789261:789587 [6] NCCL INFO Connected all trees
+ip-26-0-172-142:789261:789587 [6] NCCL INFO NVLS comm 0xa007010 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-173-121:3987800:3988126 [6] NCCL INFO Connected all trees
+ip-26-0-173-121:3987800:3988126 [6] NCCL INFO NVLS comm 0x943f720 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-173-121:3987795:3988124 [1] NCCL INFO Connected all trees
+ip-26-0-173-121:3987795:3988124 [1] NCCL INFO NVLS comm 0x9a28d90 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-142:789260:789591 [5] NCCL INFO Channel 00/0 : 13[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:789260:789591 [5] NCCL INFO Channel 01/0 : 13[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:789260:789591 [5] NCCL INFO Channel 02/0 : 13[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:789262:789590 [7] NCCL INFO Channel 00/0 : 15[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:789260:789591 [5] NCCL INFO Channel 03/0 : 13[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:789262:789590 [7] NCCL INFO Channel 01/0 : 15[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:789256:789593 [1] NCCL INFO Channel 00/0 : 9[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:789260:789591 [5] NCCL INFO Channel 04/0 : 13[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:789258:789592 [3] NCCL INFO Channel 00/0 : 11[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:789259:789589 [4] NCCL INFO Channel 00/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:789262:789590 [7] NCCL INFO Channel 02/0 : 15[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:789256:789593 [1] NCCL INFO Channel 02/0 : 9[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:789260:789591 [5] NCCL INFO Channel 06/0 : 13[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:789258:789592 [3] NCCL INFO Channel 01/0 : 11[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:789259:789589 [4] NCCL INFO Channel 01/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:789262:789590 [7] NCCL INFO Channel 03/0 : 15[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:789258:789592 [3] NCCL INFO Channel 02/0 : 11[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:789256:789593 [1] NCCL INFO Channel 03/0 : 9[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:789260:789591 [5] NCCL INFO Channel 07/0 : 13[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:789261:789587 [6] NCCL INFO Channel 00/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:789259:789589 [4] NCCL INFO Channel 02/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:789255:789586 [0] NCCL INFO Channel 01/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:789262:789590 [7] NCCL INFO Channel 04/0 : 15[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:789258:789592 [3] NCCL INFO Channel 04/0 : 11[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:789257:789588 [2] NCCL INFO Channel 00/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:789260:789591 [5] NCCL INFO Channel 08/0 : 13[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:789256:789593 [1] NCCL INFO Channel 04/0 : 9[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:789259:789589 [4] NCCL INFO Channel 03/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:789262:789590 [7] NCCL INFO Channel 05/0 : 15[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:789255:789586 [0] NCCL INFO Channel 02/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:789261:789587 [6] NCCL INFO Channel 01/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:789258:789592 [3] NCCL INFO Channel 05/0 : 11[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:789257:789588 [2] NCCL INFO Channel 01/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:789256:789593 [1] NCCL INFO Channel 05/0 : 9[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:789260:789591 [5] NCCL INFO Channel 09/0 : 13[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:789262:789590 [7] NCCL INFO Channel 06/0 : 15[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:789259:789589 [4] NCCL INFO Channel 05/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:789255:789586 [0] NCCL INFO Channel 03/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:789258:789592 [3] NCCL INFO Channel 06/0 : 11[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:789261:789587 [6] NCCL INFO Channel 02/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:789257:789588 [2] NCCL INFO Channel 03/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:789260:789591 [5] NCCL INFO Channel 10/0 : 13[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:789256:789593 [1] NCCL INFO Channel 06/0 : 9[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:789262:789590 [7] NCCL INFO Channel 08/0 : 15[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:789259:789589 [4] NCCL INFO Channel 06/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:789255:789586 [0] NCCL INFO Channel 04/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:789258:789592 [3] NCCL INFO Channel 07/0 : 11[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:789261:789587 [6] NCCL INFO Channel 03/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:789256:789593 [1] NCCL INFO Channel 07/0 : 9[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:789257:789588 [2] NCCL INFO Channel 04/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:789260:789591 [5] NCCL INFO Channel 11/0 : 13[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:789262:789590 [7] NCCL INFO Channel 09/0 : 15[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:789255:789586 [0] NCCL INFO Channel 05/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:789259:789589 [4] NCCL INFO Channel 07/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:789258:789592 [3] NCCL INFO Channel 08/0 : 11[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:789261:789587 [6] NCCL INFO Channel 04/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:789256:789593 [1] NCCL INFO Channel 08/0 : 9[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:789257:789588 [2] NCCL INFO Channel 05/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:789260:789591 [5] NCCL INFO Channel 12/0 : 13[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:789262:789590 [7] NCCL INFO Channel 10/0 : 15[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:789258:789592 [3] NCCL INFO Channel 09/0 : 11[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:789259:789589 [4] NCCL INFO Channel 08/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:789255:789586 [0] NCCL INFO Channel 06/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:789256:789593 [1] NCCL INFO Channel 10/0 : 9[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:789261:789587 [6] NCCL INFO Channel 05/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:789257:789588 [2] NCCL INFO Channel 06/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:789260:789591 [5] NCCL INFO Channel 14/0 : 13[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:789262:789590 [7] NCCL INFO Channel 11/0 : 15[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:789258:789592 [3] NCCL INFO Channel 10/0 : 11[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:789255:789586 [0] NCCL INFO Channel 07/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:789259:789589 [4] NCCL INFO Channel 09/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:789256:789593 [1] NCCL INFO Channel 11/0 : 9[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:789257:789588 [2] NCCL INFO Channel 07/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:789261:789587 [6] NCCL INFO Channel 07/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:789260:789591 [5] NCCL INFO Channel 15/0 : 13[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:789262:789590 [7] NCCL INFO Channel 12/0 : 15[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:789258:789592 [3] NCCL INFO Channel 12/0 : 11[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:789255:789586 [0] NCCL INFO Channel 09/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:789259:789589 [4] NCCL INFO Channel 10/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:789256:789593 [1] NCCL INFO Channel 12/0 : 9[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:789257:789588 [2] NCCL INFO Channel 08/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:789261:789587 [6] NCCL INFO Channel 08/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:789260:789591 [5] NCCL INFO Channel 00/0 : 5[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:789258:789592 [3] NCCL INFO Channel 13/0 : 11[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:789262:789590 [7] NCCL INFO Channel 13/0 : 15[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:789255:789586 [0] NCCL INFO Channel 10/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:789259:789589 [4] NCCL INFO Channel 11/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:789256:789593 [1] NCCL INFO Channel 13/0 : 9[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:789257:789588 [2] NCCL INFO Channel 09/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:789261:789587 [6] NCCL INFO Channel 09/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:789258:789592 [3] NCCL INFO Channel 14/0 : 11[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:789260:789591 [5] NCCL INFO Channel 01/0 : 5[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:789255:789586 [0] NCCL INFO Channel 11/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:789262:789590 [7] NCCL INFO Channel 14/0 : 15[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:789259:789589 [4] NCCL INFO Channel 13/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:789256:789593 [1] NCCL INFO Channel 14/0 : 9[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:789257:789588 [2] NCCL INFO Channel 11/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:789261:789587 [6] NCCL INFO Channel 10/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:789258:789592 [3] NCCL INFO Channel 15/0 : 11[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:789255:789586 [0] NCCL INFO Channel 12/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:789260:789591 [5] NCCL INFO Channel 02/0 : 5[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:789262:789590 [7] NCCL INFO Channel 00/0 : 7[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:789259:789589 [4] NCCL INFO Channel 14/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:789256:789593 [1] NCCL INFO Channel 15/0 : 9[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:789257:789588 [2] NCCL INFO Channel 12/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:789261:789587 [6] NCCL INFO Channel 11/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:789258:789592 [3] NCCL INFO Channel 00/0 : 3[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:789255:789586 [0] NCCL INFO Channel 13/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:789260:789591 [5] NCCL INFO Channel 03/0 : 5[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:789262:789590 [7] NCCL INFO Channel 01/0 : 7[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:789259:789589 [4] NCCL INFO Channel 15/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:789256:789593 [1] NCCL INFO Channel 00/0 : 1[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:789257:789588 [2] NCCL INFO Channel 13/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:789261:789587 [6] NCCL INFO Channel 12/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:789258:789592 [3] NCCL INFO Channel 01/0 : 3[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:789255:789586 [0] NCCL INFO Channel 14/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:789260:789591 [5] NCCL INFO Channel 04/0 : 5[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:789262:789590 [7] NCCL INFO Channel 02/0 : 7[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:789259:789589 [4] NCCL INFO Channel 00/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:789256:789593 [1] NCCL INFO Channel 02/0 : 1[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:789257:789588 [2] NCCL INFO Channel 14/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:789258:789592 [3] NCCL INFO Channel 02/0 : 3[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:789261:789587 [6] NCCL INFO Channel 13/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:789255:789586 [0] NCCL INFO Channel 15/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:789262:789590 [7] NCCL INFO Channel 03/0 : 7[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:789260:789591 [5] NCCL INFO Channel 06/0 : 5[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:789256:789593 [1] NCCL INFO Channel 03/0 : 1[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:789259:789589 [4] NCCL INFO Channel 01/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:789257:789588 [2] NCCL INFO Channel 15/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:789258:789592 [3] NCCL INFO Channel 04/0 : 3[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:789261:789587 [6] NCCL INFO Channel 15/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:789255:789586 [0] NCCL INFO Channel 01/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:789262:789590 [7] NCCL INFO Channel 04/0 : 7[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:789260:789591 [5] NCCL INFO Channel 07/0 : 5[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:789256:789593 [1] NCCL INFO Channel 04/0 : 1[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:789259:789589 [4] NCCL INFO Channel 02/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:789257:789588 [2] NCCL INFO Channel 00/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:789258:789592 [3] NCCL INFO Channel 05/0 : 3[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:789261:789587 [6] NCCL INFO Channel 00/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:789255:789586 [0] NCCL INFO Channel 02/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:789260:789591 [5] NCCL INFO Channel 08/0 : 5[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:789262:789590 [7] NCCL INFO Channel 05/0 : 7[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:789256:789593 [1] NCCL INFO Channel 05/0 : 1[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:789257:789588 [2] NCCL INFO Channel 01/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:789259:789589 [4] NCCL INFO Channel 03/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:789258:789592 [3] NCCL INFO Channel 06/0 : 3[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:789261:789587 [6] NCCL INFO Channel 01/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:789255:789586 [0] NCCL INFO Channel 03/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:789260:789591 [5] NCCL INFO Channel 09/0 : 5[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:789262:789590 [7] NCCL INFO Channel 06/0 : 7[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:789256:789593 [1] NCCL INFO Channel 06/0 : 1[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:789257:789588 [2] NCCL INFO Channel 03/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:789258:789592 [3] NCCL INFO Channel 07/0 : 3[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:789259:789589 [4] NCCL INFO Channel 05/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:789261:789587 [6] NCCL INFO Channel 02/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:789255:789586 [0] NCCL INFO Channel 04/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:789260:789591 [5] NCCL INFO Channel 10/0 : 5[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:789262:789590 [7] NCCL INFO Channel 08/0 : 7[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:789256:789593 [1] NCCL INFO Channel 07/0 : 1[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:789257:789588 [2] NCCL INFO Channel 04/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:789258:789592 [3] NCCL INFO Channel 08/0 : 3[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:789261:789587 [6] NCCL INFO Channel 03/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:789259:789589 [4] NCCL INFO Channel 06/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:789255:789586 [0] NCCL INFO Channel 05/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:789260:789591 [5] NCCL INFO Channel 11/0 : 5[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:789262:789590 [7] NCCL INFO Channel 09/0 : 7[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:789256:789593 [1] NCCL INFO Channel 08/0 : 1[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:789257:789588 [2] NCCL INFO Channel 05/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:789258:789592 [3] NCCL INFO Channel 09/0 : 3[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:789261:789587 [6] NCCL INFO Channel 04/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:789259:789589 [4] NCCL INFO Channel 07/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:789255:789586 [0] NCCL INFO Channel 06/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:789260:789591 [5] NCCL INFO Channel 12/0 : 5[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:789256:789593 [1] NCCL INFO Channel 10/0 : 1[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:789262:789590 [7] NCCL INFO Channel 10/0 : 7[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:789257:789588 [2] NCCL INFO Channel 06/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:789258:789592 [3] NCCL INFO Channel 10/0 : 3[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:789261:789587 [6] NCCL INFO Channel 05/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:789259:789589 [4] NCCL INFO Channel 08/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:789255:789586 [0] NCCL INFO Channel 07/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:789260:789591 [5] NCCL INFO Channel 14/0 : 5[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:789256:789593 [1] NCCL INFO Channel 11/0 : 1[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:789257:789588 [2] NCCL INFO Channel 07/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:789258:789592 [3] NCCL INFO Channel 12/0 : 3[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:789262:789590 [7] NCCL INFO Channel 11/0 : 7[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:789261:789587 [6] NCCL INFO Channel 07/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:789259:789589 [4] NCCL INFO Channel 09/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:789255:789586 [0] NCCL INFO Channel 09/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:789260:789591 [5] NCCL INFO Channel 15/0 : 5[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:789256:789593 [1] NCCL INFO Channel 12/0 : 1[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:789257:789588 [2] NCCL INFO Channel 08/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:789258:789592 [3] NCCL INFO Channel 13/0 : 3[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:789262:789590 [7] NCCL INFO Channel 12/0 : 7[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:789261:789587 [6] NCCL INFO Channel 08/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:789255:789586 [0] NCCL INFO Channel 10/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:789259:789589 [4] NCCL INFO Channel 10/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:789256:789593 [1] NCCL INFO Channel 13/0 : 1[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:789258:789592 [3] NCCL INFO Channel 14/0 : 3[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:789257:789588 [2] NCCL INFO Channel 09/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:789262:789590 [7] NCCL INFO Channel 13/0 : 7[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:789261:789587 [6] NCCL INFO Channel 09/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:789255:789586 [0] NCCL INFO Channel 11/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:789256:789593 [1] NCCL INFO Channel 14/0 : 1[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:789259:789589 [4] NCCL INFO Channel 11/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:789258:789592 [3] NCCL INFO Channel 15/0 : 3[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:789257:789588 [2] NCCL INFO Channel 11/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:789262:789590 [7] NCCL INFO Channel 14/0 : 7[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:789261:789587 [6] NCCL INFO Channel 10/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:789256:789593 [1] NCCL INFO Channel 15/0 : 1[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:789255:789586 [0] NCCL INFO Channel 12/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:789259:789589 [4] NCCL INFO Channel 13/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:789257:789588 [2] NCCL INFO Channel 12/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:789261:789587 [6] NCCL INFO Channel 11/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:789255:789586 [0] NCCL INFO Channel 13/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:789257:789588 [2] NCCL INFO Channel 13/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:789259:789589 [4] NCCL INFO Channel 14/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:789261:789587 [6] NCCL INFO Channel 12/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:789255:789586 [0] NCCL INFO Channel 14/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:789257:789588 [2] NCCL INFO Channel 14/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:789259:789589 [4] NCCL INFO Channel 15/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:789261:789587 [6] NCCL INFO Channel 13/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:789255:789586 [0] NCCL INFO Channel 15/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:789257:789588 [2] NCCL INFO Channel 15/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:789261:789587 [6] NCCL INFO Channel 15/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-121:3987794:3988123 [0] NCCL INFO Channel 01/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-121:3987796:3988122 [2] NCCL INFO Channel 00/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-121:3987794:3988123 [0] NCCL INFO Channel 02/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-121:3987798:3988129 [4] NCCL INFO Channel 00/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-121:3987796:3988122 [2] NCCL INFO Channel 01/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-121:3987800:3988126 [6] NCCL INFO Channel 00/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-121:3987794:3988123 [0] NCCL INFO Channel 03/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-121:3987798:3988129 [4] NCCL INFO Channel 01/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-121:3987796:3988122 [2] NCCL INFO Channel 03/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-121:3987800:3988126 [6] NCCL INFO Channel 01/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-121:3987794:3988123 [0] NCCL INFO Channel 04/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-121:3987798:3988129 [4] NCCL INFO Channel 02/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-121:3987799:3988127 [5] NCCL INFO Channel 00/0 : 5[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-121:3987796:3988122 [2] NCCL INFO Channel 04/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-121:3987795:3988124 [1] NCCL INFO Channel 00/0 : 1[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-121:3987794:3988123 [0] NCCL INFO Channel 05/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-121:3987800:3988126 [6] NCCL INFO Channel 02/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-121:3987798:3988129 [4] NCCL INFO Channel 03/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-121:3987799:3988127 [5] NCCL INFO Channel 01/0 : 5[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-121:3987801:3988128 [7] NCCL INFO Channel 00/0 : 7[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-121:3987797:3988125 [3] NCCL INFO Channel 00/0 : 3[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-121:3987796:3988122 [2] NCCL INFO Channel 05/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-121:3987800:3988126 [6] NCCL INFO Channel 03/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-121:3987795:3988124 [1] NCCL INFO Channel 02/0 : 1[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-121:3987798:3988129 [4] NCCL INFO Channel 05/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-121:3987794:3988123 [0] NCCL INFO Channel 06/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-121:3987799:3988127 [5] NCCL INFO Channel 02/0 : 5[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-121:3987801:3988128 [7] NCCL INFO Channel 01/0 : 7[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-121:3987796:3988122 [2] NCCL INFO Channel 06/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-121:3987797:3988125 [3] NCCL INFO Channel 01/0 : 3[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-121:3987800:3988126 [6] NCCL INFO Channel 04/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-121:3987795:3988124 [1] NCCL INFO Channel 03/0 : 1[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-121:3987798:3988129 [4] NCCL INFO Channel 06/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-121:3987794:3988123 [0] NCCL INFO Channel 07/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-121:3987799:3988127 [5] NCCL INFO Channel 03/0 : 5[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-121:3987796:3988122 [2] NCCL INFO Channel 07/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-121:3987797:3988125 [3] NCCL INFO Channel 02/0 : 3[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-121:3987801:3988128 [7] NCCL INFO Channel 02/0 : 7[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-121:3987800:3988126 [6] NCCL INFO Channel 05/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-121:3987795:3988124 [1] NCCL INFO Channel 04/0 : 1[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-121:3987798:3988129 [4] NCCL INFO Channel 07/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-121:3987794:3988123 [0] NCCL INFO Channel 09/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-121:3987796:3988122 [2] NCCL INFO Channel 08/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-121:3987799:3988127 [5] NCCL INFO Channel 04/0 : 5[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-121:3987797:3988125 [3] NCCL INFO Channel 04/0 : 3[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-121:3987801:3988128 [7] NCCL INFO Channel 03/0 : 7[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-121:3987795:3988124 [1] NCCL INFO Channel 05/0 : 1[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-121:3987800:3988126 [6] NCCL INFO Channel 07/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-121:3987798:3988129 [4] NCCL INFO Channel 08/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-121:3987794:3988123 [0] NCCL INFO Channel 10/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-121:3987796:3988122 [2] NCCL INFO Channel 09/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-121:3987799:3988127 [5] NCCL INFO Channel 06/0 : 5[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-121:3987797:3988125 [3] NCCL INFO Channel 05/0 : 3[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-121:3987795:3988124 [1] NCCL INFO Channel 06/0 : 1[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-121:3987801:3988128 [7] NCCL INFO Channel 04/0 : 7[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-121:3987800:3988126 [6] NCCL INFO Channel 08/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-121:3987794:3988123 [0] NCCL INFO Channel 11/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-121:3987796:3988122 [2] NCCL INFO Channel 11/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-121:3987798:3988129 [4] NCCL INFO Channel 09/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-121:3987799:3988127 [5] NCCL INFO Channel 07/0 : 5[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-121:3987797:3988125 [3] NCCL INFO Channel 06/0 : 3[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-121:3987795:3988124 [1] NCCL INFO Channel 07/0 : 1[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-121:3987801:3988128 [7] NCCL INFO Channel 05/0 : 7[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-121:3987800:3988126 [6] NCCL INFO Channel 09/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-121:3987794:3988123 [0] NCCL INFO Channel 12/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-121:3987796:3988122 [2] NCCL INFO Channel 12/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-121:3987798:3988129 [4] NCCL INFO Channel 10/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-121:3987799:3988127 [5] NCCL INFO Channel 08/0 : 5[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-121:3987797:3988125 [3] NCCL INFO Channel 07/0 : 3[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-121:3987795:3988124 [1] NCCL INFO Channel 08/0 : 1[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-121:3987801:3988128 [7] NCCL INFO Channel 06/0 : 7[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-121:3987800:3988126 [6] NCCL INFO Channel 10/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-121:3987794:3988123 [0] NCCL INFO Channel 13/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-121:3987796:3988122 [2] NCCL INFO Channel 13/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-121:3987798:3988129 [4] NCCL INFO Channel 11/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-121:3987799:3988127 [5] NCCL INFO Channel 09/0 : 5[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-121:3987795:3988124 [1] NCCL INFO Channel 10/0 : 1[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-121:3987797:3988125 [3] NCCL INFO Channel 08/0 : 3[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-121:3987801:3988128 [7] NCCL INFO Channel 08/0 : 7[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-121:3987800:3988126 [6] NCCL INFO Channel 11/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-121:3987794:3988123 [0] NCCL INFO Channel 14/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-121:3987796:3988122 [2] NCCL INFO Channel 14/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-121:3987798:3988129 [4] NCCL INFO Channel 13/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-121:3987799:3988127 [5] NCCL INFO Channel 10/0 : 5[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-121:3987795:3988124 [1] NCCL INFO Channel 11/0 : 1[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-121:3987797:3988125 [3] NCCL INFO Channel 09/0 : 3[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-121:3987801:3988128 [7] NCCL INFO Channel 09/0 : 7[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-121:3987794:3988123 [0] NCCL INFO Channel 15/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-121:3987800:3988126 [6] NCCL INFO Channel 12/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-121:3987796:3988122 [2] NCCL INFO Channel 15/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-121:3987798:3988129 [4] NCCL INFO Channel 14/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-121:3987799:3988127 [5] NCCL INFO Channel 11/0 : 5[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-121:3987795:3988124 [1] NCCL INFO Channel 12/0 : 1[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-121:3987797:3988125 [3] NCCL INFO Channel 10/0 : 3[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-121:3987801:3988128 [7] NCCL INFO Channel 10/0 : 7[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-121:3987794:3988123 [0] NCCL INFO Channel 01/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-121:3987796:3988122 [2] NCCL INFO Channel 00/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-121:3987800:3988126 [6] NCCL INFO Channel 13/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-121:3987799:3988127 [5] NCCL INFO Channel 12/0 : 5[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-121:3987798:3988129 [4] NCCL INFO Channel 15/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-121:3987795:3988124 [1] NCCL INFO Channel 13/0 : 1[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-121:3987797:3988125 [3] NCCL INFO Channel 12/0 : 3[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-121:3987794:3988123 [0] NCCL INFO Channel 02/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-121:3987801:3988128 [7] NCCL INFO Channel 11/0 : 7[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-121:3987796:3988122 [2] NCCL INFO Channel 01/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-121:3987799:3988127 [5] NCCL INFO Channel 14/0 : 5[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-121:3987800:3988126 [6] NCCL INFO Channel 15/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-121:3987798:3988129 [4] NCCL INFO Channel 00/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-121:3987795:3988124 [1] NCCL INFO Channel 14/0 : 1[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-121:3987797:3988125 [3] NCCL INFO Channel 13/0 : 3[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-121:3987794:3988123 [0] NCCL INFO Channel 03/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-121:3987801:3988128 [7] NCCL INFO Channel 12/0 : 7[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-121:3987796:3988122 [2] NCCL INFO Channel 03/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-121:3987800:3988126 [6] NCCL INFO Channel 00/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-121:3987799:3988127 [5] NCCL INFO Channel 15/0 : 5[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-121:3987798:3988129 [4] NCCL INFO Channel 01/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-121:3987795:3988124 [1] NCCL INFO Channel 15/0 : 1[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-121:3987797:3988125 [3] NCCL INFO Channel 14/0 : 3[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-121:3987794:3988123 [0] NCCL INFO Channel 04/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-121:3987796:3988122 [2] NCCL INFO Channel 04/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-121:3987801:3988128 [7] NCCL INFO Channel 13/0 : 7[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-121:3987799:3988127 [5] NCCL INFO Channel 00/0 : 13[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-121:3987800:3988126 [6] NCCL INFO Channel 01/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-121:3987798:3988129 [4] NCCL INFO Channel 02/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-121:3987795:3988124 [1] NCCL INFO Channel 00/0 : 9[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-121:3987797:3988125 [3] NCCL INFO Channel 15/0 : 3[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-121:3987794:3988123 [0] NCCL INFO Channel 05/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-121:3987796:3988122 [2] NCCL INFO Channel 05/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-121:3987801:3988128 [7] NCCL INFO Channel 14/0 : 7[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-121:3987795:3988124 [1] NCCL INFO Channel 02/0 : 9[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-121:3987800:3988126 [6] NCCL INFO Channel 02/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-121:3987798:3988129 [4] NCCL INFO Channel 03/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-121:3987797:3988125 [3] NCCL INFO Channel 00/0 : 11[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-121:3987794:3988123 [0] NCCL INFO Channel 06/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-121:3987799:3988127 [5] NCCL INFO Channel 01/0 : 13[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-121:3987796:3988122 [2] NCCL INFO Channel 06/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-121:3987801:3988128 [7] NCCL INFO Channel 00/0 : 15[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-121:3987795:3988124 [1] NCCL INFO Channel 03/0 : 9[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-121:3987800:3988126 [6] NCCL INFO Channel 03/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-121:3987797:3988125 [3] NCCL INFO Channel 01/0 : 11[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-121:3987794:3988123 [0] NCCL INFO Channel 07/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-121:3987798:3988129 [4] NCCL INFO Channel 05/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-121:3987796:3988122 [2] NCCL INFO Channel 07/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-121:3987799:3988127 [5] NCCL INFO Channel 02/0 : 13[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-121:3987801:3988128 [7] NCCL INFO Channel 01/0 : 15[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-121:3987795:3988124 [1] NCCL INFO Channel 04/0 : 9[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-121:3987800:3988126 [6] NCCL INFO Channel 04/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-121:3987797:3988125 [3] NCCL INFO Channel 02/0 : 11[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-121:3987796:3988122 [2] NCCL INFO Channel 08/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-121:3987794:3988123 [0] NCCL INFO Channel 09/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-121:3987798:3988129 [4] NCCL INFO Channel 06/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-121:3987799:3988127 [5] NCCL INFO Channel 03/0 : 13[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-121:3987795:3988124 [1] NCCL INFO Channel 05/0 : 9[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-121:3987801:3988128 [7] NCCL INFO Channel 02/0 : 15[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-121:3987800:3988126 [6] NCCL INFO Channel 05/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-121:3987797:3988125 [3] NCCL INFO Channel 04/0 : 11[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-121:3987796:3988122 [2] NCCL INFO Channel 09/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-121:3987794:3988123 [0] NCCL INFO Channel 10/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-121:3987798:3988129 [4] NCCL INFO Channel 07/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-121:3987799:3988127 [5] NCCL INFO Channel 04/0 : 13[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-121:3987797:3988125 [3] NCCL INFO Channel 05/0 : 11[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-121:3987801:3988128 [7] NCCL INFO Channel 03/0 : 15[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-121:3987800:3988126 [6] NCCL INFO Channel 07/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-121:3987795:3988124 [1] NCCL INFO Channel 06/0 : 9[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-121:3987796:3988122 [2] NCCL INFO Channel 11/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-121:3987794:3988123 [0] NCCL INFO Channel 11/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-121:3987798:3988129 [4] NCCL INFO Channel 08/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-121:3987797:3988125 [3] NCCL INFO Channel 06/0 : 11[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-121:3987795:3988124 [1] NCCL INFO Channel 07/0 : 9[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-121:3987796:3988122 [2] NCCL INFO Channel 12/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-121:3987801:3988128 [7] NCCL INFO Channel 04/0 : 15[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-121:3987800:3988126 [6] NCCL INFO Channel 08/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-121:3987799:3988127 [5] NCCL INFO Channel 06/0 : 13[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-121:3987794:3988123 [0] NCCL INFO Channel 12/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-121:3987797:3988125 [3] NCCL INFO Channel 07/0 : 11[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-121:3987798:3988129 [4] NCCL INFO Channel 09/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-121:3987796:3988122 [2] NCCL INFO Channel 13/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-121:3987795:3988124 [1] NCCL INFO Channel 08/0 : 9[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-121:3987794:3988123 [0] NCCL INFO Channel 13/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-121:3987800:3988126 [6] NCCL INFO Channel 09/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-121:3987799:3988127 [5] NCCL INFO Channel 07/0 : 13[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-121:3987801:3988128 [7] NCCL INFO Channel 05/0 : 15[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-121:3987797:3988125 [3] NCCL INFO Channel 08/0 : 11[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-121:3987798:3988129 [4] NCCL INFO Channel 10/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-121:3987796:3988122 [2] NCCL INFO Channel 14/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-121:3987795:3988124 [1] NCCL INFO Channel 10/0 : 9[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-121:3987794:3988123 [0] NCCL INFO Channel 14/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-121:3987797:3988125 [3] NCCL INFO Channel 09/0 : 11[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-121:3987799:3988127 [5] NCCL INFO Channel 08/0 : 13[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-121:3987800:3988126 [6] NCCL INFO Channel 10/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-121:3987801:3988128 [7] NCCL INFO Channel 06/0 : 15[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-121:3987798:3988129 [4] NCCL INFO Channel 11/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-121:3987796:3988122 [2] NCCL INFO Channel 15/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-121:3987795:3988124 [1] NCCL INFO Channel 11/0 : 9[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-121:3987794:3988123 [0] NCCL INFO Channel 15/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-121:3987797:3988125 [3] NCCL INFO Channel 10/0 : 11[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-121:3987800:3988126 [6] NCCL INFO Channel 11/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-121:3987799:3988127 [5] NCCL INFO Channel 09/0 : 13[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-121:3987801:3988128 [7] NCCL INFO Channel 08/0 : 15[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-121:3987798:3988129 [4] NCCL INFO Channel 13/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-121:3987795:3988124 [1] NCCL INFO Channel 12/0 : 9[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-121:3987797:3988125 [3] NCCL INFO Channel 12/0 : 11[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-121:3987801:3988128 [7] NCCL INFO Channel 09/0 : 15[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-121:3987800:3988126 [6] NCCL INFO Channel 12/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-121:3987798:3988129 [4] NCCL INFO Channel 14/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-121:3987799:3988127 [5] NCCL INFO Channel 10/0 : 13[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-121:3987797:3988125 [3] NCCL INFO Channel 13/0 : 11[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-121:3987795:3988124 [1] NCCL INFO Channel 13/0 : 9[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-121:3987798:3988129 [4] NCCL INFO Channel 15/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-121:3987800:3988126 [6] NCCL INFO Channel 13/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-121:3987799:3988127 [5] NCCL INFO Channel 11/0 : 13[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-121:3987801:3988128 [7] NCCL INFO Channel 10/0 : 15[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-121:3987797:3988125 [3] NCCL INFO Channel 14/0 : 11[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-121:3987795:3988124 [1] NCCL INFO Channel 14/0 : 9[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-121:3987800:3988126 [6] NCCL INFO Channel 15/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-121:3987799:3988127 [5] NCCL INFO Channel 12/0 : 13[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-121:3987797:3988125 [3] NCCL INFO Channel 15/0 : 11[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-121:3987795:3988124 [1] NCCL INFO Channel 15/0 : 9[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-121:3987801:3988128 [7] NCCL INFO Channel 11/0 : 15[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-121:3987799:3988127 [5] NCCL INFO Channel 14/0 : 13[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-121:3987799:3988127 [5] NCCL INFO Channel 15/0 : 13[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-121:3987801:3988128 [7] NCCL INFO Channel 12/0 : 15[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-121:3987801:3988128 [7] NCCL INFO Channel 13/0 : 15[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-121:3987801:3988128 [7] NCCL INFO Channel 14/0 : 15[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:789256:789593 [1] NCCL INFO Connected NVLS tree
+ip-26-0-172-142:789256:789593 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-142:789256:789593 [1] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+ip-26-0-172-142:789256:789593 [1] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-173-121:3987795:3988124 [1] NCCL INFO Connected NVLS tree
+ip-26-0-173-121:3987795:3988124 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-173-121:3987795:3988124 [1] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+ip-26-0-173-121:3987795:3988124 [1] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-173-121:3987801:3988128 [7] NCCL INFO Connected NVLS tree
+ip-26-0-173-121:3987801:3988128 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-173-121:3987801:3988128 [7] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+ip-26-0-173-121:3987801:3988128 [7] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-173-121:3987796:3988122 [2] NCCL INFO Connected NVLS tree
+ip-26-0-173-121:3987796:3988122 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-173-121:3987796:3988122 [2] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+ip-26-0-173-121:3987796:3988122 [2] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-173-121:3987798:3988129 [4] NCCL INFO Connected NVLS tree
+ip-26-0-173-121:3987798:3988129 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-173-121:3987798:3988129 [4] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+ip-26-0-173-121:3987798:3988129 [4] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-173-121:3987799:3988127 [5] NCCL INFO Connected NVLS tree
+ip-26-0-173-121:3987799:3988127 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-173-121:3987799:3988127 [5] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+ip-26-0-173-121:3987799:3988127 [5] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-172-142:789262:789590 [7] NCCL INFO Connected NVLS tree
+ip-26-0-172-142:789262:789590 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-142:789262:789590 [7] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+ip-26-0-172-142:789262:789590 [7] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-173-121:3987794:3988123 [0] NCCL INFO Connected NVLS tree
+ip-26-0-173-121:3987794:3988123 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-173-121:3987794:3988123 [0] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+ip-26-0-173-121:3987794:3988123 [0] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-172-142:789257:789588 [2] NCCL INFO Connected NVLS tree
+ip-26-0-172-142:789257:789588 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-142:789257:789588 [2] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+ip-26-0-172-142:789257:789588 [2] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-172-142:789259:789589 [4] NCCL INFO Connected NVLS tree
+ip-26-0-172-142:789259:789589 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-142:789259:789589 [4] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+ip-26-0-172-142:789259:789589 [4] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-173-121:3987797:3988125 [3] NCCL INFO Connected NVLS tree
+ip-26-0-173-121:3987797:3988125 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-173-121:3987797:3988125 [3] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+ip-26-0-173-121:3987797:3988125 [3] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-173-121:3987800:3988126 [6] NCCL INFO Connected NVLS tree
+ip-26-0-173-121:3987800:3988126 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-173-121:3987800:3988126 [6] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+ip-26-0-173-121:3987800:3988126 [6] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-172-142:789255:789586 [0] NCCL INFO Connected NVLS tree
+ip-26-0-172-142:789255:789586 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-142:789255:789586 [0] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+ip-26-0-172-142:789255:789586 [0] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-172-142:789260:789591 [5] NCCL INFO Connected NVLS tree
+ip-26-0-172-142:789260:789591 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-142:789260:789591 [5] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+ip-26-0-172-142:789260:789591 [5] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-172-142:789258:789592 [3] NCCL INFO Connected NVLS tree
+ip-26-0-172-142:789258:789592 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-142:789258:789592 [3] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+ip-26-0-172-142:789258:789592 [3] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-172-142:789261:789587 [6] NCCL INFO Connected NVLS tree
+ip-26-0-172-142:789261:789587 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-142:789261:789587 [6] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+ip-26-0-172-142:789261:789587 [6] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-173-121:3987798:3988129 [4] NCCL INFO comm 0x9df7d00 rank 12 nranks 16 cudaDev 4 nvmlDev 4 busId 97000 commId 0xa2326dfe0277fd9 - Init COMPLETE
+ip-26-0-173-121:3987794:3988123 [0] NCCL INFO comm 0x8e641b0 rank 8 nranks 16 cudaDev 0 nvmlDev 0 busId 53000 commId 0xa2326dfe0277fd9 - Init COMPLETE
+ip-26-0-173-121:3987799:3988127 [5] NCCL INFO comm 0x93f88d0 rank 13 nranks 16 cudaDev 5 nvmlDev 5 busId a8000 commId 0xa2326dfe0277fd9 - Init COMPLETE
+ip-26-0-173-121:3987795:3988124 [1] NCCL INFO comm 0x9a28d90 rank 9 nranks 16 cudaDev 1 nvmlDev 1 busId 64000 commId 0xa2326dfe0277fd9 - Init COMPLETE
+ip-26-0-173-121:3987801:3988128 [7] NCCL INFO comm 0x9851480 rank 15 nranks 16 cudaDev 7 nvmlDev 7 busId ca000 commId 0xa2326dfe0277fd9 - Init COMPLETE
+ip-26-0-173-121:3987797:3988125 [3] NCCL INFO comm 0x8d9bd80 rank 11 nranks 16 cudaDev 3 nvmlDev 3 busId 86000 commId 0xa2326dfe0277fd9 - Init COMPLETE
+ip-26-0-173-121:3987800:3988126 [6] NCCL INFO comm 0x943f720 rank 14 nranks 16 cudaDev 6 nvmlDev 6 busId b9000 commId 0xa2326dfe0277fd9 - Init COMPLETE
+ip-26-0-173-121:3987796:3988122 [2] NCCL INFO comm 0x92c5260 rank 10 nranks 16 cudaDev 2 nvmlDev 2 busId 75000 commId 0xa2326dfe0277fd9 - Init COMPLETE
+ip-26-0-172-142:789261:789587 [6] NCCL INFO comm 0xa007010 rank 6 nranks 16 cudaDev 6 nvmlDev 6 busId b9000 commId 0xa2326dfe0277fd9 - Init COMPLETE
+ip-26-0-172-142:789259:789589 [4] NCCL INFO comm 0x92665b0 rank 4 nranks 16 cudaDev 4 nvmlDev 4 busId 97000 commId 0xa2326dfe0277fd9 - Init COMPLETE
+ip-26-0-172-142:789262:789590 [7] NCCL INFO comm 0x8c22120 rank 7 nranks 16 cudaDev 7 nvmlDev 7 busId ca000 commId 0xa2326dfe0277fd9 - Init COMPLETE
+ip-26-0-172-142:789256:789593 [1] NCCL INFO comm 0xa553960 rank 1 nranks 16 cudaDev 1 nvmlDev 1 busId 64000 commId 0xa2326dfe0277fd9 - Init COMPLETE
+ip-26-0-172-142:789255:789586 [0] NCCL INFO comm 0xa1ffd50 rank 0 nranks 16 cudaDev 0 nvmlDev 0 busId 53000 commId 0xa2326dfe0277fd9 - Init COMPLETE
+ip-26-0-172-142:789257:789588 [2] NCCL INFO comm 0x95c0680 rank 2 nranks 16 cudaDev 2 nvmlDev 2 busId 75000 commId 0xa2326dfe0277fd9 - Init COMPLETE
+ip-26-0-172-142:789258:789592 [3] NCCL INFO comm 0x8d44de0 rank 3 nranks 16 cudaDev 3 nvmlDev 3 busId 86000 commId 0xa2326dfe0277fd9 - Init COMPLETE
+ip-26-0-172-142:789260:789591 [5] NCCL INFO comm 0xa28a6a0 rank 5 nranks 16 cudaDev 5 nvmlDev 5 busId a8000 commId 0xa2326dfe0277fd9 - Init COMPLETE
+NCCL version 2.18.6+cuda12.1
+ip-26-0-172-142:789261:789623 [6] NCCL INFO Using network Libfabric
+ip-26-0-172-142:789255:789622 [0] NCCL INFO Using network Libfabric
+ip-26-0-172-142:789260:789624 [5] NCCL INFO Using network Libfabric
+ip-26-0-172-142:789257:789625 [2] NCCL INFO Using network Libfabric
+ip-26-0-172-142:789259:789626 [4] NCCL INFO Using network Libfabric
+ip-26-0-172-142:789258:789629 [3] NCCL INFO Using network Libfabric
+ip-26-0-172-142:789256:789628 [1] NCCL INFO Using network Libfabric
+ip-26-0-172-142:789262:789627 [7] NCCL INFO Using network Libfabric
+ip-26-0-172-142:789255:789622 [0] NCCL INFO comm 0xa2175e0 rank 0 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0x98e054a6265db20 - Init START
+ip-26-0-172-142:789260:789624 [5] NCCL INFO comm 0xa2a0b50 rank 5 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0x98e054a6265db20 - Init START
+ip-26-0-172-142:789261:789623 [6] NCCL INFO comm 0xa01c8d0 rank 6 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0x98e054a6265db20 - Init START
+ip-26-0-172-142:789262:789627 [7] NCCL INFO comm 0x8c38eb0 rank 7 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0x98e054a6265db20 - Init START
+ip-26-0-172-142:789259:789626 [4] NCCL INFO comm 0x927d120 rank 4 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0x98e054a6265db20 - Init START
+ip-26-0-172-142:789258:789629 [3] NCCL INFO comm 0x8d5bae0 rank 3 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0x98e054a6265db20 - Init START
+ip-26-0-172-142:789256:789628 [1] NCCL INFO comm 0xa56a1d0 rank 1 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0x98e054a6265db20 - Init START
+ip-26-0-172-142:789257:789625 [2] NCCL INFO comm 0x95d7160 rank 2 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0x98e054a6265db20 - Init START
+ip-26-0-173-121:3987797:3988159 [3] NCCL INFO Using network Libfabric
+ip-26-0-173-121:3987795:3988160 [1] NCCL INFO Using network Libfabric
+ip-26-0-172-142:789255:789622 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/205
+ip-26-0-172-142:789260:789624 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-172-142:789262:789627 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-172-142:789258:789629 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-172-142:789257:789625 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-172-142:789261:789623 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-172-142:789259:789626 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-172-142:789256:789628 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-173-121:3987794:3988157 [0] NCCL INFO Using network Libfabric
+ip-26-0-173-121:3987798:3988162 [4] NCCL INFO Using network Libfabric
+ip-26-0-173-121:3987796:3988163 [2] NCCL INFO Using network Libfabric
+ip-26-0-173-121:3987800:3988161 [6] NCCL INFO Using network Libfabric
+ip-26-0-173-121:3987799:3988158 [5] NCCL INFO Using network Libfabric
+ip-26-0-173-121:3987801:3988164 [7] NCCL INFO Using network Libfabric
+ip-26-0-173-121:3987797:3988159 [3] NCCL INFO comm 0x8db2350 rank 3 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0xcef8427ac1151735 - Init START
+ip-26-0-173-121:3987796:3988163 [2] NCCL INFO comm 0x92db530 rank 2 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0xcef8427ac1151735 - Init START
+ip-26-0-173-121:3987795:3988160 [1] NCCL INFO comm 0x9a3fc10 rank 1 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0xcef8427ac1151735 - Init START
+ip-26-0-173-121:3987794:3988157 [0] NCCL INFO comm 0x8e7b3f0 rank 0 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0xcef8427ac1151735 - Init START
+ip-26-0-173-121:3987800:3988161 [6] NCCL INFO comm 0x9456580 rank 6 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0xcef8427ac1151735 - Init START
+ip-26-0-173-121:3987801:3988164 [7] NCCL INFO comm 0x98674d0 rank 7 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0xcef8427ac1151735 - Init START
+ip-26-0-173-121:3987799:3988158 [5] NCCL INFO comm 0x940dfd0 rank 5 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0xcef8427ac1151735 - Init START
+ip-26-0-173-121:3987798:3988162 [4] NCCL INFO comm 0x9e0dcf0 rank 4 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0xcef8427ac1151735 - Init START
+ip-26-0-173-121:3987797:3988159 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-173-121:3987796:3988163 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-173-121:3987798:3988162 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-173-121:3987799:3988158 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-173-121:3987794:3988157 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-173-121:3987795:3988160 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-173-121:3987801:3988164 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-173-121:3987800:3988161 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-173-121:3987794:3988157 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-173-121:3987794:3988157 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-173-121:3987801:3988164 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-173-121:3987801:3988164 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-173-121:3987799:3988158 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-173-121:3987799:3988158 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-173-121:3987796:3988163 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-173-121:3987796:3988163 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-173-121:3987795:3988160 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-173-121:3987795:3988160 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-173-121:3987797:3988159 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-173-121:3987797:3988159 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-173-121:3987800:3988161 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-173-121:3987800:3988161 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-173-121:3987798:3988162 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-173-121:3987798:3988162 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-173-121:3987798:3988162 [4] NCCL INFO Trees [0] 5/-1/-1->4->3 [1] 5/-1/-1->4->3 [2] 5/-1/-1->4->3 [3] 5/-1/-1->4->3 [4] 5/-1/-1->4->3 [5] 5/-1/-1->4->3 [6] 5/-1/-1->4->3 [7] 5/-1/-1->4->3 [8] 5/-1/-1->4->3 [9] 5/-1/-1->4->3 [10] 5/-1/-1->4->3 [11] 5/-1/-1->4->3 [12] 5/-1/-1->4->3 [13] 5/-1/-1->4->3 [14] 5/-1/-1->4->3 [15] 5/-1/-1->4->3 [16] 5/-1/-1->4->3 [17] 5/-1/-1->4->3 [18] 5/-1/-1->4->3 [19] 5/-1/-1->4->3 [20] 5/-1/-1->4->3 [21] 5/-1/-1->4->3 [22] 5/-1/-1->4->3 [23] 5/-1/-1->4->3
+ip-26-0-173-121:3987798:3988162 [4] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-173-121:3987799:3988158 [5] NCCL INFO Trees [0] 6/-1/-1->5->4 [1] 6/-1/-1->5->4 [2] 6/-1/-1->5->4 [3] 6/-1/-1->5->4 [4] 6/-1/-1->5->4 [5] 6/-1/-1->5->4 [6] 6/-1/-1->5->4 [7] 6/-1/-1->5->4 [8] 6/-1/-1->5->4 [9] 6/-1/-1->5->4 [10] 6/-1/-1->5->4 [11] 6/-1/-1->5->4 [12] 6/-1/-1->5->4 [13] 6/-1/-1->5->4 [14] 6/-1/-1->5->4 [15] 6/-1/-1->5->4 [16] 6/-1/-1->5->4 [17] 6/-1/-1->5->4 [18] 6/-1/-1->5->4 [19] 6/-1/-1->5->4 [20] 6/-1/-1->5->4 [21] 6/-1/-1->5->4 [22] 6/-1/-1->5->4 [23] 6/-1/-1->5->4
+ip-26-0-173-121:3987797:3988159 [3] NCCL INFO Trees [0] 4/-1/-1->3->2 [1] 4/-1/-1->3->2 [2] 4/-1/-1->3->2 [3] 4/-1/-1->3->2 [4] 4/-1/-1->3->2 [5] 4/-1/-1->3->2 [6] 4/-1/-1->3->2 [7] 4/-1/-1->3->2 [8] 4/-1/-1->3->2 [9] 4/-1/-1->3->2 [10] 4/-1/-1->3->2 [11] 4/-1/-1->3->2 [12] 4/-1/-1->3->2 [13] 4/-1/-1->3->2 [14] 4/-1/-1->3->2 [15] 4/-1/-1->3->2 [16] 4/-1/-1->3->2 [17] 4/-1/-1->3->2 [18] 4/-1/-1->3->2 [19] 4/-1/-1->3->2 [20] 4/-1/-1->3->2 [21] 4/-1/-1->3->2 [22] 4/-1/-1->3->2 [23] 4/-1/-1->3->2
+ip-26-0-173-121:3987800:3988161 [6] NCCL INFO Trees [0] 7/-1/-1->6->5 [1] 7/-1/-1->6->5 [2] 7/-1/-1->6->5 [3] 7/-1/-1->6->5 [4] 7/-1/-1->6->5 [5] 7/-1/-1->6->5 [6] 7/-1/-1->6->5 [7] 7/-1/-1->6->5 [8] 7/-1/-1->6->5 [9] 7/-1/-1->6->5 [10] 7/-1/-1->6->5 [11] 7/-1/-1->6->5 [12] 7/-1/-1->6->5 [13] 7/-1/-1->6->5 [14] 7/-1/-1->6->5 [15] 7/-1/-1->6->5 [16] 7/-1/-1->6->5 [17] 7/-1/-1->6->5 [18] 7/-1/-1->6->5 [19] 7/-1/-1->6->5 [20] 7/-1/-1->6->5 [21] 7/-1/-1->6->5 [22] 7/-1/-1->6->5 [23] 7/-1/-1->6->5
+ip-26-0-173-121:3987796:3988163 [2] NCCL INFO Trees [0] 3/-1/-1->2->1 [1] 3/-1/-1->2->1 [2] 3/-1/-1->2->1 [3] 3/-1/-1->2->1 [4] 3/-1/-1->2->1 [5] 3/-1/-1->2->1 [6] 3/-1/-1->2->1 [7] 3/-1/-1->2->1 [8] 3/-1/-1->2->1 [9] 3/-1/-1->2->1 [10] 3/-1/-1->2->1 [11] 3/-1/-1->2->1 [12] 3/-1/-1->2->1 [13] 3/-1/-1->2->1 [14] 3/-1/-1->2->1 [15] 3/-1/-1->2->1 [16] 3/-1/-1->2->1 [17] 3/-1/-1->2->1 [18] 3/-1/-1->2->1 [19] 3/-1/-1->2->1 [20] 3/-1/-1->2->1 [21] 3/-1/-1->2->1 [22] 3/-1/-1->2->1 [23] 3/-1/-1->2->1
+ip-26-0-173-121:3987801:3988164 [7] NCCL INFO Trees [0] -1/-1/-1->7->6 [1] -1/-1/-1->7->6 [2] -1/-1/-1->7->6 [3] -1/-1/-1->7->6 [4] -1/-1/-1->7->6 [5] -1/-1/-1->7->6 [6] -1/-1/-1->7->6 [7] -1/-1/-1->7->6 [8] -1/-1/-1->7->6 [9] -1/-1/-1->7->6 [10] -1/-1/-1->7->6 [11] -1/-1/-1->7->6 [12] -1/-1/-1->7->6 [13] -1/-1/-1->7->6 [14] -1/-1/-1->7->6 [15] -1/-1/-1->7->6 [16] -1/-1/-1->7->6 [17] -1/-1/-1->7->6 [18] -1/-1/-1->7->6 [19] -1/-1/-1->7->6 [20] -1/-1/-1->7->6 [21] -1/-1/-1->7->6 [22] -1/-1/-1->7->6 [23] -1/-1/-1->7->6
+ip-26-0-173-121:3987797:3988159 [3] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-173-121:3987799:3988158 [5] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-173-121:3987800:3988161 [6] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-173-121:3987796:3988163 [2] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-173-121:3987801:3988164 [7] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-173-121:3987794:3988157 [0] NCCL INFO Channel 00/24 :    0   1   2   3   4   5   6   7
+ip-26-0-173-121:3987794:3988157 [0] NCCL INFO Channel 01/24 :    0   1   2   3   4   5   6   7
+ip-26-0-173-121:3987794:3988157 [0] NCCL INFO Channel 02/24 :    0   1   2   3   4   5   6   7
+ip-26-0-173-121:3987794:3988157 [0] NCCL INFO Channel 03/24 :    0   1   2   3   4   5   6   7
+ip-26-0-173-121:3987794:3988157 [0] NCCL INFO Channel 04/24 :    0   1   2   3   4   5   6   7
+ip-26-0-173-121:3987795:3988160 [1] NCCL INFO Trees [0] 2/-1/-1->1->0 [1] 2/-1/-1->1->0 [2] 2/-1/-1->1->0 [3] 2/-1/-1->1->0 [4] 2/-1/-1->1->0 [5] 2/-1/-1->1->0 [6] 2/-1/-1->1->0 [7] 2/-1/-1->1->0 [8] 2/-1/-1->1->0 [9] 2/-1/-1->1->0 [10] 2/-1/-1->1->0 [11] 2/-1/-1->1->0 [12] 2/-1/-1->1->0 [13] 2/-1/-1->1->0 [14] 2/-1/-1->1->0 [15] 2/-1/-1->1->0 [16] 2/-1/-1->1->0 [17] 2/-1/-1->1->0 [18] 2/-1/-1->1->0 [19] 2/-1/-1->1->0 [20] 2/-1/-1->1->0 [21] 2/-1/-1->1->0 [22] 2/-1/-1->1->0 [23] 2/-1/-1->1->0
+ip-26-0-173-121:3987794:3988157 [0] NCCL INFO Channel 05/24 :    0   1   2   3   4   5   6   7
+ip-26-0-173-121:3987794:3988157 [0] NCCL INFO Channel 06/24 :    0   1   2   3   4   5   6   7
+ip-26-0-173-121:3987794:3988157 [0] NCCL INFO Channel 07/24 :    0   1   2   3   4   5   6   7
+ip-26-0-173-121:3987794:3988157 [0] NCCL INFO Channel 08/24 :    0   1   2   3   4   5   6   7
+ip-26-0-173-121:3987794:3988157 [0] NCCL INFO Channel 09/24 :    0   1   2   3   4   5   6   7
+ip-26-0-173-121:3987794:3988157 [0] NCCL INFO Channel 10/24 :    0   1   2   3   4   5   6   7
+ip-26-0-173-121:3987794:3988157 [0] NCCL INFO Channel 11/24 :    0   1   2   3   4   5   6   7
+ip-26-0-173-121:3987794:3988157 [0] NCCL INFO Channel 12/24 :    0   1   2   3   4   5   6   7
+ip-26-0-173-121:3987794:3988157 [0] NCCL INFO Channel 13/24 :    0   1   2   3   4   5   6   7
+ip-26-0-173-121:3987794:3988157 [0] NCCL INFO Channel 14/24 :    0   1   2   3   4   5   6   7
+ip-26-0-173-121:3987794:3988157 [0] NCCL INFO Channel 15/24 :    0   1   2   3   4   5   6   7
+ip-26-0-173-121:3987794:3988157 [0] NCCL INFO Channel 16/24 :    0   1   2   3   4   5   6   7
+ip-26-0-173-121:3987794:3988157 [0] NCCL INFO Channel 17/24 :    0   1   2   3   4   5   6   7
+ip-26-0-173-121:3987794:3988157 [0] NCCL INFO Channel 18/24 :    0   1   2   3   4   5   6   7
+ip-26-0-173-121:3987794:3988157 [0] NCCL INFO Channel 19/24 :    0   1   2   3   4   5   6   7
+ip-26-0-173-121:3987794:3988157 [0] NCCL INFO Channel 20/24 :    0   1   2   3   4   5   6   7
+ip-26-0-173-121:3987794:3988157 [0] NCCL INFO Channel 21/24 :    0   1   2   3   4   5   6   7
+ip-26-0-173-121:3987794:3988157 [0] NCCL INFO Channel 22/24 :    0   1   2   3   4   5   6   7
+ip-26-0-173-121:3987794:3988157 [0] NCCL INFO Channel 23/24 :    0   1   2   3   4   5   6   7
+ip-26-0-173-121:3987794:3988157 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] 1/-1/-1->0->-1 [7] 1/-1/-1->0->-1 [8] 1/-1/-1->0->-1 [9] 1/-1/-1->0->-1 [10] 1/-1/-1->0->-1 [11] 1/-1/-1->0->-1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] 1/-1/-1->0->-1 [19] 1/-1/-1->0->-1 [20] 1/-1/-1->0->-1 [21] 1/-1/-1->0->-1 [22] 1/-1/-1->0->-1 [23] 1/-1/-1->0->-1
+ip-26-0-173-121:3987795:3988160 [1] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-173-121:3987794:3988157 [0] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-173-121:3987800:3988161 [6] NCCL INFO Channel 00/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-173-121:3987796:3988163 [2] NCCL INFO Channel 00/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-173-121:3987800:3988161 [6] NCCL INFO Channel 01/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-173-121:3987796:3988163 [2] NCCL INFO Channel 01/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-173-121:3987796:3988163 [2] NCCL INFO Channel 02/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-173-121:3987800:3988161 [6] NCCL INFO Channel 02/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-173-121:3987796:3988163 [2] NCCL INFO Channel 03/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-173-121:3987797:3988159 [3] NCCL INFO Channel 00/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-173-121:3987801:3988164 [7] NCCL INFO Channel 00/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987800:3988161 [6] NCCL INFO Channel 03/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-173-121:3987796:3988163 [2] NCCL INFO Channel 04/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-173-121:3987797:3988159 [3] NCCL INFO Channel 01/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-173-121:3987801:3988164 [7] NCCL INFO Channel 01/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987798:3988162 [4] NCCL INFO Channel 00/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-173-121:3987800:3988161 [6] NCCL INFO Channel 04/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-173-121:3987796:3988163 [2] NCCL INFO Channel 05/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-173-121:3987797:3988159 [3] NCCL INFO Channel 02/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-173-121:3987801:3988164 [7] NCCL INFO Channel 02/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987798:3988162 [4] NCCL INFO Channel 01/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-173-121:3987800:3988161 [6] NCCL INFO Channel 05/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-173-121:3987796:3988163 [2] NCCL INFO Channel 06/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-142:789261:789623 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-172-142:789261:789623 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-173-121:3987797:3988159 [3] NCCL INFO Channel 03/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-173-121:3987801:3988164 [7] NCCL INFO Channel 03/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987798:3988162 [4] NCCL INFO Channel 02/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-173-121:3987800:3988161 [6] NCCL INFO Channel 06/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-173-121:3987796:3988163 [2] NCCL INFO Channel 07/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-173-121:3987797:3988159 [3] NCCL INFO Channel 04/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-173-121:3987801:3988164 [7] NCCL INFO Channel 04/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987798:3988162 [4] NCCL INFO Channel 03/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-173-121:3987796:3988163 [2] NCCL INFO Channel 08/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-173-121:3987798:3988162 [4] NCCL INFO Channel 04/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-173-121:3987796:3988163 [2] NCCL INFO Channel 09/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-173-121:3987798:3988162 [4] NCCL INFO Channel 05/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-173-121:3987797:3988159 [3] NCCL INFO Channel 05/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-173-121:3987796:3988163 [2] NCCL INFO Channel 10/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-173-121:3987798:3988162 [4] NCCL INFO Channel 06/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-173-121:3987797:3988159 [3] NCCL INFO Channel 06/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-173-121:3987796:3988163 [2] NCCL INFO Channel 11/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-173-121:3987794:3988157 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-173-121:3987798:3988162 [4] NCCL INFO Channel 07/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-173-121:3987796:3988163 [2] NCCL INFO Channel 12/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-173-121:3987801:3988164 [7] NCCL INFO Channel 05/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987799:3988158 [5] NCCL INFO Channel 00/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-173-121:3987797:3988159 [3] NCCL INFO Channel 07/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-173-121:3987800:3988161 [6] NCCL INFO Channel 07/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-173-121:3987800:3988161 [6] NCCL INFO Channel 08/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-173-121:3987795:3988160 [1] NCCL INFO Channel 00/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-173-121:3987794:3988157 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-173-121:3987800:3988161 [6] NCCL INFO Channel 09/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-173-121:3987795:3988160 [1] NCCL INFO Channel 01/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-173-121:3987794:3988157 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-173-121:3987798:3988162 [4] NCCL INFO Channel 08/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-173-121:3987800:3988161 [6] NCCL INFO Channel 10/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-173-121:3987795:3988160 [1] NCCL INFO Channel 02/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-173-121:3987794:3988157 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-173-121:3987798:3988162 [4] NCCL INFO Channel 09/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-173-121:3987800:3988161 [6] NCCL INFO Channel 11/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-173-121:3987795:3988160 [1] NCCL INFO Channel 03/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-173-121:3987796:3988163 [2] NCCL INFO Channel 13/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-173-121:3987794:3988157 [0] NCCL INFO Channel 04/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-173-121:3987799:3988158 [5] NCCL INFO Channel 01/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-173-121:3987801:3988164 [7] NCCL INFO Channel 06/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987796:3988163 [2] NCCL INFO Channel 14/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-173-121:3987797:3988159 [3] NCCL INFO Channel 08/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-173-121:3987798:3988162 [4] NCCL INFO Channel 10/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-173-121:3987795:3988160 [1] NCCL INFO Channel 04/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-173-121:3987800:3988161 [6] NCCL INFO Channel 12/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-173-121:3987801:3988164 [7] NCCL INFO Channel 07/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987798:3988162 [4] NCCL INFO Channel 11/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-173-121:3987795:3988160 [1] NCCL INFO Channel 05/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-173-121:3987800:3988161 [6] NCCL INFO Channel 13/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-173-121:3987801:3988164 [7] NCCL INFO Channel 08/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987798:3988162 [4] NCCL INFO Channel 12/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-173-121:3987795:3988160 [1] NCCL INFO Channel 06/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-173-121:3987799:3988158 [5] NCCL INFO Channel 02/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-173-121:3987800:3988161 [6] NCCL INFO Channel 14/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-173-121:3987801:3988164 [7] NCCL INFO Channel 09/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987798:3988162 [4] NCCL INFO Channel 13/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-173-121:3987795:3988160 [1] NCCL INFO Channel 07/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-173-121:3987799:3988158 [5] NCCL INFO Channel 03/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-173-121:3987800:3988161 [6] NCCL INFO Channel 15/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-173-121:3987801:3988164 [7] NCCL INFO Channel 10/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987796:3988163 [2] NCCL INFO Channel 15/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-173-121:3987798:3988162 [4] NCCL INFO Channel 14/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-173-121:3987795:3988160 [1] NCCL INFO Channel 08/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-173-121:3987799:3988158 [5] NCCL INFO Channel 04/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-173-121:3987800:3988161 [6] NCCL INFO Channel 16/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-173-121:3987801:3988164 [7] NCCL INFO Channel 11/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987797:3988159 [3] NCCL INFO Channel 09/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-173-121:3987796:3988163 [2] NCCL INFO Channel 16/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-173-121:3987795:3988160 [1] NCCL INFO Channel 09/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-173-121:3987798:3988162 [4] NCCL INFO Channel 15/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-173-121:3987799:3988158 [5] NCCL INFO Channel 05/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-173-121:3987800:3988161 [6] NCCL INFO Channel 17/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-173-121:3987801:3988164 [7] NCCL INFO Channel 12/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987794:3988157 [0] NCCL INFO Channel 05/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-173-121:3987797:3988159 [3] NCCL INFO Channel 10/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-173-121:3987795:3988160 [1] NCCL INFO Channel 10/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-173-121:3987796:3988163 [2] NCCL INFO Channel 17/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-173-121:3987798:3988162 [4] NCCL INFO Channel 16/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-173-121:3987799:3988158 [5] NCCL INFO Channel 06/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-173-121:3987800:3988161 [6] NCCL INFO Channel 18/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-173-121:3987801:3988164 [7] NCCL INFO Channel 13/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987794:3988157 [0] NCCL INFO Channel 06/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-173-121:3987795:3988160 [1] NCCL INFO Channel 11/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-173-121:3987797:3988159 [3] NCCL INFO Channel 11/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-173-121:3987796:3988163 [2] NCCL INFO Channel 18/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-173-121:3987798:3988162 [4] NCCL INFO Channel 17/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-173-121:3987800:3988161 [6] NCCL INFO Channel 19/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-173-121:3987799:3988158 [5] NCCL INFO Channel 07/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-173-121:3987801:3988164 [7] NCCL INFO Channel 14/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987794:3988157 [0] NCCL INFO Channel 07/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-173-121:3987796:3988163 [2] NCCL INFO Channel 19/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-173-121:3987795:3988160 [1] NCCL INFO Channel 12/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-173-121:3987797:3988159 [3] NCCL INFO Channel 12/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-173-121:3987800:3988161 [6] NCCL INFO Channel 20/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-173-121:3987798:3988162 [4] NCCL INFO Channel 18/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-173-121:3987799:3988158 [5] NCCL INFO Channel 08/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-173-121:3987801:3988164 [7] NCCL INFO Channel 15/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987794:3988157 [0] NCCL INFO Channel 08/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-173-121:3987795:3988160 [1] NCCL INFO Channel 13/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-173-121:3987799:3988158 [5] NCCL INFO Channel 09/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-173-121:3987801:3988164 [7] NCCL INFO Channel 16/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987794:3988157 [0] NCCL INFO Channel 09/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-173-121:3987795:3988160 [1] NCCL INFO Channel 14/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-173-121:3987801:3988164 [7] NCCL INFO Channel 17/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987799:3988158 [5] NCCL INFO Channel 10/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-173-121:3987794:3988157 [0] NCCL INFO Channel 10/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-173-121:3987795:3988160 [1] NCCL INFO Channel 15/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-173-121:3987801:3988164 [7] NCCL INFO Channel 18/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987799:3988158 [5] NCCL INFO Channel 11/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-173-121:3987794:3988157 [0] NCCL INFO Channel 11/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-173-121:3987796:3988163 [2] NCCL INFO Channel 20/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-173-121:3987801:3988164 [7] NCCL INFO Channel 19/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987797:3988159 [3] NCCL INFO Channel 13/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-173-121:3987798:3988162 [4] NCCL INFO Channel 19/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-173-121:3987796:3988163 [2] NCCL INFO Channel 21/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-173-121:3987801:3988164 [7] NCCL INFO Channel 20/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987797:3988159 [3] NCCL INFO Channel 14/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-173-121:3987798:3988162 [4] NCCL INFO Channel 20/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-173-121:3987796:3988163 [2] NCCL INFO Channel 22/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-173-121:3987801:3988164 [7] NCCL INFO Channel 21/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987798:3988162 [4] NCCL INFO Channel 21/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-173-121:3987797:3988159 [3] NCCL INFO Channel 15/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-173-121:3987800:3988161 [6] NCCL INFO Channel 21/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-173-121:3987796:3988163 [2] NCCL INFO Channel 23/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-173-121:3987798:3988162 [4] NCCL INFO Channel 22/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-173-121:3987801:3988164 [7] NCCL INFO Channel 22/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987797:3988159 [3] NCCL INFO Channel 16/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-173-121:3987800:3988161 [6] NCCL INFO Channel 22/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-173-121:3987798:3988162 [4] NCCL INFO Channel 23/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-173-121:3987801:3988164 [7] NCCL INFO Channel 23/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987797:3988159 [3] NCCL INFO Channel 17/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-173-121:3987795:3988160 [1] NCCL INFO Channel 16/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-173-121:3987800:3988161 [6] NCCL INFO Channel 23/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-173-121:3987799:3988158 [5] NCCL INFO Channel 12/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-173-121:3987794:3988157 [0] NCCL INFO Channel 12/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-173-121:3987797:3988159 [3] NCCL INFO Channel 18/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-173-121:3987795:3988160 [1] NCCL INFO Channel 17/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-173-121:3987794:3988157 [0] NCCL INFO Channel 13/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-173-121:3987799:3988158 [5] NCCL INFO Channel 13/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-173-121:3987797:3988159 [3] NCCL INFO Channel 19/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-173-121:3987795:3988160 [1] NCCL INFO Channel 18/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-173-121:3987794:3988157 [0] NCCL INFO Channel 14/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-173-121:3987799:3988158 [5] NCCL INFO Channel 14/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-173-121:3987797:3988159 [3] NCCL INFO Channel 20/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-173-121:3987795:3988160 [1] NCCL INFO Channel 19/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-173-121:3987794:3988157 [0] NCCL INFO Channel 15/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-173-121:3987799:3988158 [5] NCCL INFO Channel 15/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-173-121:3987797:3988159 [3] NCCL INFO Channel 21/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-173-121:3987795:3988160 [1] NCCL INFO Channel 20/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-173-121:3987794:3988157 [0] NCCL INFO Channel 16/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-173-121:3987799:3988158 [5] NCCL INFO Channel 16/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-173-121:3987797:3988159 [3] NCCL INFO Channel 22/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-173-121:3987795:3988160 [1] NCCL INFO Channel 21/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-173-121:3987794:3988157 [0] NCCL INFO Channel 17/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-173-121:3987799:3988158 [5] NCCL INFO Channel 17/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-173-121:3987795:3988160 [1] NCCL INFO Channel 22/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-173-121:3987797:3988159 [3] NCCL INFO Channel 23/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-173-121:3987794:3988157 [0] NCCL INFO Channel 18/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-173-121:3987799:3988158 [5] NCCL INFO Channel 18/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-173-121:3987795:3988160 [1] NCCL INFO Channel 23/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-173-121:3987794:3988157 [0] NCCL INFO Channel 19/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-173-121:3987799:3988158 [5] NCCL INFO Channel 19/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-173-121:3987794:3988157 [0] NCCL INFO Channel 20/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-173-121:3987799:3988158 [5] NCCL INFO Channel 20/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-142:789256:789628 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-172-142:789256:789628 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-173-121:3987794:3988157 [0] NCCL INFO Channel 21/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-173-121:3987799:3988158 [5] NCCL INFO Channel 21/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-173-121:3987794:3988157 [0] NCCL INFO Channel 22/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-173-121:3987799:3988158 [5] NCCL INFO Channel 22/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-173-121:3987794:3988157 [0] NCCL INFO Channel 23/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-173-121:3987799:3988158 [5] NCCL INFO Channel 23/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-142:789255:789622 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-172-142:789255:789622 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-172-142:789257:789625 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-172-142:789257:789625 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-172-142:789262:789627 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-172-142:789262:789627 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-172-142:789258:789629 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-172-142:789258:789629 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-172-142:789259:789626 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-172-142:789259:789626 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-172-142:789260:789624 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-172-142:789260:789624 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-172-142:789260:789624 [5] NCCL INFO Trees [0] 6/-1/-1->5->4 [1] 6/-1/-1->5->4 [2] 6/-1/-1->5->4 [3] 6/-1/-1->5->4 [4] 6/-1/-1->5->4 [5] 6/-1/-1->5->4 [6] 6/-1/-1->5->4 [7] 6/-1/-1->5->4 [8] 6/-1/-1->5->4 [9] 6/-1/-1->5->4 [10] 6/-1/-1->5->4 [11] 6/-1/-1->5->4 [12] 6/-1/-1->5->4 [13] 6/-1/-1->5->4 [14] 6/-1/-1->5->4 [15] 6/-1/-1->5->4 [16] 6/-1/-1->5->4 [17] 6/-1/-1->5->4 [18] 6/-1/-1->5->4 [19] 6/-1/-1->5->4 [20] 6/-1/-1->5->4 [21] 6/-1/-1->5->4 [22] 6/-1/-1->5->4 [23] 6/-1/-1->5->4
+ip-26-0-172-142:789260:789624 [5] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-172-142:789261:789623 [6] NCCL INFO Trees [0] 7/-1/-1->6->5 [1] 7/-1/-1->6->5 [2] 7/-1/-1->6->5 [3] 7/-1/-1->6->5 [4] 7/-1/-1->6->5 [5] 7/-1/-1->6->5 [6] 7/-1/-1->6->5 [7] 7/-1/-1->6->5 [8] 7/-1/-1->6->5 [9] 7/-1/-1->6->5 [10] 7/-1/-1->6->5 [11] 7/-1/-1->6->5 [12] 7/-1/-1->6->5 [13] 7/-1/-1->6->5 [14] 7/-1/-1->6->5 [15] 7/-1/-1->6->5 [16] 7/-1/-1->6->5 [17] 7/-1/-1->6->5 [18] 7/-1/-1->6->5 [19] 7/-1/-1->6->5 [20] 7/-1/-1->6->5 [21] 7/-1/-1->6->5 [22] 7/-1/-1->6->5 [23] 7/-1/-1->6->5
+ip-26-0-172-142:789261:789623 [6] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-172-142:789259:789626 [4] NCCL INFO Trees [0] 5/-1/-1->4->3 [1] 5/-1/-1->4->3 [2] 5/-1/-1->4->3 [3] 5/-1/-1->4->3 [4] 5/-1/-1->4->3 [5] 5/-1/-1->4->3 [6] 5/-1/-1->4->3 [7] 5/-1/-1->4->3 [8] 5/-1/-1->4->3 [9] 5/-1/-1->4->3 [10] 5/-1/-1->4->3 [11] 5/-1/-1->4->3 [12] 5/-1/-1->4->3 [13] 5/-1/-1->4->3 [14] 5/-1/-1->4->3 [15] 5/-1/-1->4->3 [16] 5/-1/-1->4->3 [17] 5/-1/-1->4->3 [18] 5/-1/-1->4->3 [19] 5/-1/-1->4->3 [20] 5/-1/-1->4->3 [21] 5/-1/-1->4->3 [22] 5/-1/-1->4->3 [23] 5/-1/-1->4->3
+ip-26-0-172-142:789259:789626 [4] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-172-142:789255:789622 [0] NCCL INFO Channel 00/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-142:789258:789629 [3] NCCL INFO Trees [0] 4/-1/-1->3->2 [1] 4/-1/-1->3->2 [2] 4/-1/-1->3->2 [3] 4/-1/-1->3->2 [4] 4/-1/-1->3->2 [5] 4/-1/-1->3->2 [6] 4/-1/-1->3->2 [7] 4/-1/-1->3->2 [8] 4/-1/-1->3->2 [9] 4/-1/-1->3->2 [10] 4/-1/-1->3->2 [11] 4/-1/-1->3->2 [12] 4/-1/-1->3->2 [13] 4/-1/-1->3->2 [14] 4/-1/-1->3->2 [15] 4/-1/-1->3->2 [16] 4/-1/-1->3->2 [17] 4/-1/-1->3->2 [18] 4/-1/-1->3->2 [19] 4/-1/-1->3->2 [20] 4/-1/-1->3->2 [21] 4/-1/-1->3->2 [22] 4/-1/-1->3->2 [23] 4/-1/-1->3->2
+ip-26-0-172-142:789255:789622 [0] NCCL INFO Channel 01/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-142:789258:789629 [3] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-172-142:789257:789625 [2] NCCL INFO Trees [0] 3/-1/-1->2->1 [1] 3/-1/-1->2->1 [2] 3/-1/-1->2->1 [3] 3/-1/-1->2->1 [4] 3/-1/-1->2->1 [5] 3/-1/-1->2->1 [6] 3/-1/-1->2->1 [7] 3/-1/-1->2->1 [8] 3/-1/-1->2->1 [9] 3/-1/-1->2->1 [10] 3/-1/-1->2->1 [11] 3/-1/-1->2->1 [12] 3/-1/-1->2->1 [13] 3/-1/-1->2->1 [14] 3/-1/-1->2->1 [15] 3/-1/-1->2->1 [16] 3/-1/-1->2->1 [17] 3/-1/-1->2->1 [18] 3/-1/-1->2->1 [19] 3/-1/-1->2->1 [20] 3/-1/-1->2->1 [21] 3/-1/-1->2->1 [22] 3/-1/-1->2->1 [23] 3/-1/-1->2->1
+ip-26-0-172-142:789255:789622 [0] NCCL INFO Channel 02/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-142:789262:789627 [7] NCCL INFO Trees [0] -1/-1/-1->7->6 [1] -1/-1/-1->7->6 [2] -1/-1/-1->7->6 [3] -1/-1/-1->7->6 [4] -1/-1/-1->7->6 [5] -1/-1/-1->7->6 [6] -1/-1/-1->7->6 [7] -1/-1/-1->7->6 [8] -1/-1/-1->7->6 [9] -1/-1/-1->7->6 [10] -1/-1/-1->7->6 [11] -1/-1/-1->7->6 [12] -1/-1/-1->7->6 [13] -1/-1/-1->7->6 [14] -1/-1/-1->7->6 [15] -1/-1/-1->7->6 [16] -1/-1/-1->7->6 [17] -1/-1/-1->7->6 [18] -1/-1/-1->7->6 [19] -1/-1/-1->7->6 [20] -1/-1/-1->7->6 [21] -1/-1/-1->7->6 [22] -1/-1/-1->7->6 [23] -1/-1/-1->7->6
+ip-26-0-172-142:789256:789628 [1] NCCL INFO Trees [0] 2/-1/-1->1->0 [1] 2/-1/-1->1->0 [2] 2/-1/-1->1->0 [3] 2/-1/-1->1->0 [4] 2/-1/-1->1->0 [5] 2/-1/-1->1->0 [6] 2/-1/-1->1->0 [7] 2/-1/-1->1->0 [8] 2/-1/-1->1->0 [9] 2/-1/-1->1->0 [10] 2/-1/-1->1->0 [11] 2/-1/-1->1->0 [12] 2/-1/-1->1->0 [13] 2/-1/-1->1->0 [14] 2/-1/-1->1->0 [15] 2/-1/-1->1->0 [16] 2/-1/-1->1->0 [17] 2/-1/-1->1->0 [18] 2/-1/-1->1->0 [19] 2/-1/-1->1->0 [20] 2/-1/-1->1->0 [21] 2/-1/-1->1->0 [22] 2/-1/-1->1->0 [23] 2/-1/-1->1->0
+ip-26-0-172-142:789257:789625 [2] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-172-142:789255:789622 [0] NCCL INFO Channel 03/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-142:789262:789627 [7] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-172-142:789256:789628 [1] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-172-142:789255:789622 [0] NCCL INFO Channel 04/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-142:789255:789622 [0] NCCL INFO Channel 05/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-142:789255:789622 [0] NCCL INFO Channel 06/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-142:789255:789622 [0] NCCL INFO Channel 07/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-142:789255:789622 [0] NCCL INFO Channel 08/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-142:789255:789622 [0] NCCL INFO Channel 09/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-142:789255:789622 [0] NCCL INFO Channel 10/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-142:789255:789622 [0] NCCL INFO Channel 11/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-142:789255:789622 [0] NCCL INFO Channel 12/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-142:789255:789622 [0] NCCL INFO Channel 13/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-142:789255:789622 [0] NCCL INFO Channel 14/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-142:789255:789622 [0] NCCL INFO Channel 15/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-142:789255:789622 [0] NCCL INFO Channel 16/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-142:789255:789622 [0] NCCL INFO Channel 17/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-142:789255:789622 [0] NCCL INFO Channel 18/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-142:789255:789622 [0] NCCL INFO Channel 19/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-142:789255:789622 [0] NCCL INFO Channel 20/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-142:789255:789622 [0] NCCL INFO Channel 21/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-142:789255:789622 [0] NCCL INFO Channel 22/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-142:789255:789622 [0] NCCL INFO Channel 23/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-142:789255:789622 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] 1/-1/-1->0->-1 [7] 1/-1/-1->0->-1 [8] 1/-1/-1->0->-1 [9] 1/-1/-1->0->-1 [10] 1/-1/-1->0->-1 [11] 1/-1/-1->0->-1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] 1/-1/-1->0->-1 [19] 1/-1/-1->0->-1 [20] 1/-1/-1->0->-1 [21] 1/-1/-1->0->-1 [22] 1/-1/-1->0->-1 [23] 1/-1/-1->0->-1
+ip-26-0-172-142:789255:789622 [0] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-172-142:789256:789628 [1] NCCL INFO Channel 00/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-142:789261:789623 [6] NCCL INFO Channel 00/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-142:789262:789627 [7] NCCL INFO Channel 00/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789256:789628 [1] NCCL INFO Channel 01/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-142:789261:789623 [6] NCCL INFO Channel 01/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-142:789260:789624 [5] NCCL INFO Channel 00/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-142:789262:789627 [7] NCCL INFO Channel 01/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789256:789628 [1] NCCL INFO Channel 02/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-142:789260:789624 [5] NCCL INFO Channel 01/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-142:789261:789623 [6] NCCL INFO Channel 02/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-142:789255:789622 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-142:789262:789627 [7] NCCL INFO Channel 02/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789256:789628 [1] NCCL INFO Channel 03/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-142:789260:789624 [5] NCCL INFO Channel 02/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-142:789261:789623 [6] NCCL INFO Channel 03/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-142:789255:789622 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-142:789262:789627 [7] NCCL INFO Channel 03/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789260:789624 [5] NCCL INFO Channel 03/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-142:789256:789628 [1] NCCL INFO Channel 04/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-142:789261:789623 [6] NCCL INFO Channel 04/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-142:789255:789622 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-142:789262:789627 [7] NCCL INFO Channel 04/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789260:789624 [5] NCCL INFO Channel 04/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-142:789256:789628 [1] NCCL INFO Channel 05/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-142:789261:789623 [6] NCCL INFO Channel 05/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-142:789255:789622 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-142:789256:789628 [1] NCCL INFO Channel 06/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-142:789260:789624 [5] NCCL INFO Channel 05/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-142:789258:789629 [3] NCCL INFO Channel 00/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-142:789255:789622 [0] NCCL INFO Channel 04/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-142:789256:789628 [1] NCCL INFO Channel 07/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-142:789260:789624 [5] NCCL INFO Channel 06/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-142:789258:789629 [3] NCCL INFO Channel 01/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-142:789257:789625 [2] NCCL INFO Channel 00/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-142:789256:789628 [1] NCCL INFO Channel 08/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-142:789255:789622 [0] NCCL INFO Channel 05/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-142:789260:789624 [5] NCCL INFO Channel 07/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-142:789258:789629 [3] NCCL INFO Channel 02/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-142:789257:789625 [2] NCCL INFO Channel 01/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-142:789256:789628 [1] NCCL INFO Channel 09/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-142:789255:789622 [0] NCCL INFO Channel 06/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-142:789260:789624 [5] NCCL INFO Channel 08/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-142:789258:789629 [3] NCCL INFO Channel 03/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-142:789257:789625 [2] NCCL INFO Channel 02/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-142:789256:789628 [1] NCCL INFO Channel 10/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-142:789255:789622 [0] NCCL INFO Channel 07/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-142:789260:789624 [5] NCCL INFO Channel 09/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-142:789257:789625 [2] NCCL INFO Channel 03/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-142:789258:789629 [3] NCCL INFO Channel 04/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-142:789256:789628 [1] NCCL INFO Channel 11/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-142:789255:789622 [0] NCCL INFO Channel 08/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-142:789260:789624 [5] NCCL INFO Channel 10/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-142:789257:789625 [2] NCCL INFO Channel 04/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-142:789258:789629 [3] NCCL INFO Channel 05/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-142:789261:789623 [6] NCCL INFO Channel 06/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-142:789255:789622 [0] NCCL INFO Channel 09/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-142:789256:789628 [1] NCCL INFO Channel 12/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-142:789260:789624 [5] NCCL INFO Channel 11/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-142:789257:789625 [2] NCCL INFO Channel 05/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-142:789258:789629 [3] NCCL INFO Channel 06/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-142:789262:789627 [7] NCCL INFO Channel 05/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789261:789623 [6] NCCL INFO Channel 07/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-142:789255:789622 [0] NCCL INFO Channel 10/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-142:789256:789628 [1] NCCL INFO Channel 13/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-142:789260:789624 [5] NCCL INFO Channel 12/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-142:789257:789625 [2] NCCL INFO Channel 06/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-142:789258:789629 [3] NCCL INFO Channel 07/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-142:789262:789627 [7] NCCL INFO Channel 06/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789255:789622 [0] NCCL INFO Channel 11/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-142:789261:789623 [6] NCCL INFO Channel 08/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-142:789256:789628 [1] NCCL INFO Channel 14/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-142:789260:789624 [5] NCCL INFO Channel 13/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-142:789259:789626 [4] NCCL INFO Channel 00/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-142:789257:789625 [2] NCCL INFO Channel 07/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-142:789258:789629 [3] NCCL INFO Channel 08/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-142:789255:789622 [0] NCCL INFO Channel 12/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-142:789261:789623 [6] NCCL INFO Channel 09/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-142:789262:789627 [7] NCCL INFO Channel 07/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789260:789624 [5] NCCL INFO Channel 14/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-142:789257:789625 [2] NCCL INFO Channel 08/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-142:789256:789628 [1] NCCL INFO Channel 15/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-142:789259:789626 [4] NCCL INFO Channel 01/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-142:789258:789629 [3] NCCL INFO Channel 09/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-142:789255:789622 [0] NCCL INFO Channel 13/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-142:789262:789627 [7] NCCL INFO Channel 08/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789261:789623 [6] NCCL INFO Channel 10/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-142:789260:789624 [5] NCCL INFO Channel 15/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-142:789257:789625 [2] NCCL INFO Channel 09/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-142:789256:789628 [1] NCCL INFO Channel 16/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-142:789259:789626 [4] NCCL INFO Channel 02/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-142:789258:789629 [3] NCCL INFO Channel 10/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-142:789255:789622 [0] NCCL INFO Channel 14/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-142:789262:789627 [7] NCCL INFO Channel 09/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789261:789623 [6] NCCL INFO Channel 11/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-142:789260:789624 [5] NCCL INFO Channel 16/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-142:789257:789625 [2] NCCL INFO Channel 10/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-142:789259:789626 [4] NCCL INFO Channel 03/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-142:789256:789628 [1] NCCL INFO Channel 17/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-142:789258:789629 [3] NCCL INFO Channel 11/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-142:789255:789622 [0] NCCL INFO Channel 15/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-142:789262:789627 [7] NCCL INFO Channel 10/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789261:789623 [6] NCCL INFO Channel 12/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-142:789260:789624 [5] NCCL INFO Channel 17/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-142:789257:789625 [2] NCCL INFO Channel 11/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-142:789256:789628 [1] NCCL INFO Channel 18/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-142:789259:789626 [4] NCCL INFO Channel 04/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-142:789258:789629 [3] NCCL INFO Channel 12/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-142:789255:789622 [0] NCCL INFO Channel 16/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-142:789262:789627 [7] NCCL INFO Channel 11/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789261:789623 [6] NCCL INFO Channel 13/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-142:789260:789624 [5] NCCL INFO Channel 18/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-142:789257:789625 [2] NCCL INFO Channel 12/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-142:789256:789628 [1] NCCL INFO Channel 19/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-142:789259:789626 [4] NCCL INFO Channel 05/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-142:789258:789629 [3] NCCL INFO Channel 13/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-142:789255:789622 [0] NCCL INFO Channel 17/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-142:789260:789624 [5] NCCL INFO Channel 19/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-142:789256:789628 [1] NCCL INFO Channel 20/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-142:789257:789625 [2] NCCL INFO Channel 13/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-142:789259:789626 [4] NCCL INFO Channel 06/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-142:789258:789629 [3] NCCL INFO Channel 14/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-142:789261:789623 [6] NCCL INFO Channel 14/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-142:789255:789622 [0] NCCL INFO Channel 18/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-142:789260:789624 [5] NCCL INFO Channel 20/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-142:789257:789625 [2] NCCL INFO Channel 14/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-142:789256:789628 [1] NCCL INFO Channel 21/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-142:789259:789626 [4] NCCL INFO Channel 07/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-142:789258:789629 [3] NCCL INFO Channel 15/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-142:789262:789627 [7] NCCL INFO Channel 12/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789261:789623 [6] NCCL INFO Channel 15/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-142:789255:789622 [0] NCCL INFO Channel 19/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-142:789260:789624 [5] NCCL INFO Channel 21/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-142:789257:789625 [2] NCCL INFO Channel 15/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-142:789256:789628 [1] NCCL INFO Channel 22/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-142:789259:789626 [4] NCCL INFO Channel 08/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-142:789258:789629 [3] NCCL INFO Channel 16/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-142:789262:789627 [7] NCCL INFO Channel 13/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789261:789623 [6] NCCL INFO Channel 16/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-142:789255:789622 [0] NCCL INFO Channel 20/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-142:789260:789624 [5] NCCL INFO Channel 22/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-142:789257:789625 [2] NCCL INFO Channel 16/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-142:789259:789626 [4] NCCL INFO Channel 09/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-142:789256:789628 [1] NCCL INFO Channel 23/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-142:789258:789629 [3] NCCL INFO Channel 17/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-142:789262:789627 [7] NCCL INFO Channel 14/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789255:789622 [0] NCCL INFO Channel 21/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-142:789261:789623 [6] NCCL INFO Channel 17/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-142:789260:789624 [5] NCCL INFO Channel 23/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-142:789257:789625 [2] NCCL INFO Channel 17/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-142:789259:789626 [4] NCCL INFO Channel 10/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-142:789258:789629 [3] NCCL INFO Channel 18/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-142:789262:789627 [7] NCCL INFO Channel 15/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789255:789622 [0] NCCL INFO Channel 22/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-142:789261:789623 [6] NCCL INFO Channel 18/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-142:789259:789626 [4] NCCL INFO Channel 11/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-142:789257:789625 [2] NCCL INFO Channel 18/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-142:789258:789629 [3] NCCL INFO Channel 19/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-142:789262:789627 [7] NCCL INFO Channel 16/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789255:789622 [0] NCCL INFO Channel 23/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-142:789261:789623 [6] NCCL INFO Channel 19/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-142:789259:789626 [4] NCCL INFO Channel 12/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-142:789257:789625 [2] NCCL INFO Channel 19/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-142:789258:789629 [3] NCCL INFO Channel 20/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-142:789262:789627 [7] NCCL INFO Channel 17/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789261:789623 [6] NCCL INFO Channel 20/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-142:789257:789625 [2] NCCL INFO Channel 20/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-142:789259:789626 [4] NCCL INFO Channel 13/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-142:789258:789629 [3] NCCL INFO Channel 21/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-142:789262:789627 [7] NCCL INFO Channel 18/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789259:789626 [4] NCCL INFO Channel 14/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-142:789261:789623 [6] NCCL INFO Channel 21/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-142:789257:789625 [2] NCCL INFO Channel 21/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-142:789258:789629 [3] NCCL INFO Channel 22/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-142:789262:789627 [7] NCCL INFO Channel 19/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789259:789626 [4] NCCL INFO Channel 15/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-142:789257:789625 [2] NCCL INFO Channel 22/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-142:789261:789623 [6] NCCL INFO Channel 22/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-142:789258:789629 [3] NCCL INFO Channel 23/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-142:789262:789627 [7] NCCL INFO Channel 20/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789259:789626 [4] NCCL INFO Channel 16/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-142:789257:789625 [2] NCCL INFO Channel 23/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-142:789261:789623 [6] NCCL INFO Channel 23/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-142:789259:789626 [4] NCCL INFO Channel 17/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-142:789262:789627 [7] NCCL INFO Channel 21/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789259:789626 [4] NCCL INFO Channel 18/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-142:789262:789627 [7] NCCL INFO Channel 22/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789262:789627 [7] NCCL INFO Channel 23/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789259:789626 [4] NCCL INFO Channel 19/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-142:789259:789626 [4] NCCL INFO Channel 20/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-142:789259:789626 [4] NCCL INFO Channel 21/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-142:789259:789626 [4] NCCL INFO Channel 22/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-142:789259:789626 [4] NCCL INFO Channel 23/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-173-121:3987796:3988163 [2] NCCL INFO Connected all rings
+ip-26-0-173-121:3987797:3988159 [3] NCCL INFO Connected all rings
+ip-26-0-173-121:3987801:3988164 [7] NCCL INFO Connected all rings
+ip-26-0-173-121:3987801:3988164 [7] NCCL INFO Channel 00/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-173-121:3987801:3988164 [7] NCCL INFO Channel 01/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-173-121:3987798:3988162 [4] NCCL INFO Connected all rings
+ip-26-0-173-121:3987801:3988164 [7] NCCL INFO Channel 02/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-173-121:3987801:3988164 [7] NCCL INFO Channel 03/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-173-121:3987799:3988158 [5] NCCL INFO Connected all rings
+ip-26-0-173-121:3987795:3988160 [1] NCCL INFO Connected all rings
+ip-26-0-173-121:3987801:3988164 [7] NCCL INFO Channel 04/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-173-121:3987794:3988157 [0] NCCL INFO Connected all rings
+ip-26-0-173-121:3987801:3988164 [7] NCCL INFO Channel 05/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-173-121:3987800:3988161 [6] NCCL INFO Connected all rings
+ip-26-0-173-121:3987801:3988164 [7] NCCL INFO Channel 06/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-173-121:3987801:3988164 [7] NCCL INFO Channel 07/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-173-121:3987801:3988164 [7] NCCL INFO Channel 08/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-173-121:3987801:3988164 [7] NCCL INFO Channel 09/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-173-121:3987801:3988164 [7] NCCL INFO Channel 10/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-173-121:3987801:3988164 [7] NCCL INFO Channel 11/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-173-121:3987801:3988164 [7] NCCL INFO Channel 12/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-173-121:3987801:3988164 [7] NCCL INFO Channel 13/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-173-121:3987801:3988164 [7] NCCL INFO Channel 14/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-173-121:3987801:3988164 [7] NCCL INFO Channel 15/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-173-121:3987801:3988164 [7] NCCL INFO Channel 16/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-173-121:3987801:3988164 [7] NCCL INFO Channel 17/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-173-121:3987801:3988164 [7] NCCL INFO Channel 18/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-173-121:3987801:3988164 [7] NCCL INFO Channel 19/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-173-121:3987801:3988164 [7] NCCL INFO Channel 20/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-173-121:3987801:3988164 [7] NCCL INFO Channel 21/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-142:789257:789625 [2] NCCL INFO Connected all rings
+ip-26-0-173-121:3987801:3988164 [7] NCCL INFO Channel 22/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-173-121:3987801:3988164 [7] NCCL INFO Channel 23/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-173-121:3987797:3988159 [3] NCCL INFO Channel 00/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-173-121:3987796:3988163 [2] NCCL INFO Channel 00/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-173-121:3987796:3988163 [2] NCCL INFO Channel 01/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-173-121:3987797:3988159 [3] NCCL INFO Channel 01/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-173-121:3987796:3988163 [2] NCCL INFO Channel 02/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-173-121:3987797:3988159 [3] NCCL INFO Channel 02/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-173-121:3987796:3988163 [2] NCCL INFO Channel 03/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-173-121:3987797:3988159 [3] NCCL INFO Channel 03/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-173-121:3987796:3988163 [2] NCCL INFO Channel 04/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-173-121:3987797:3988159 [3] NCCL INFO Channel 04/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-173-121:3987796:3988163 [2] NCCL INFO Channel 05/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-173-121:3987797:3988159 [3] NCCL INFO Channel 05/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-173-121:3987796:3988163 [2] NCCL INFO Channel 06/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-173-121:3987797:3988159 [3] NCCL INFO Channel 06/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-173-121:3987796:3988163 [2] NCCL INFO Channel 07/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-173-121:3987797:3988159 [3] NCCL INFO Channel 07/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-173-121:3987796:3988163 [2] NCCL INFO Channel 08/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-173-121:3987797:3988159 [3] NCCL INFO Channel 08/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-173-121:3987796:3988163 [2] NCCL INFO Channel 09/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-173-121:3987797:3988159 [3] NCCL INFO Channel 09/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-173-121:3987799:3988158 [5] NCCL INFO Channel 00/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-173-121:3987795:3988160 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987799:3988158 [5] NCCL INFO Channel 01/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-173-121:3987796:3988163 [2] NCCL INFO Channel 10/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-173-121:3987797:3988159 [3] NCCL INFO Channel 10/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-173-121:3987799:3988158 [5] NCCL INFO Channel 02/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-173-121:3987796:3988163 [2] NCCL INFO Channel 11/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-173-121:3987797:3988159 [3] NCCL INFO Channel 11/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-173-121:3987795:3988160 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987799:3988158 [5] NCCL INFO Channel 03/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-173-121:3987796:3988163 [2] NCCL INFO Channel 12/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-173-121:3987797:3988159 [3] NCCL INFO Channel 12/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-173-121:3987797:3988159 [3] NCCL INFO Channel 13/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-173-121:3987799:3988158 [5] NCCL INFO Channel 04/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-173-121:3987796:3988163 [2] NCCL INFO Channel 13/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-173-121:3987799:3988158 [5] NCCL INFO Channel 05/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-173-121:3987797:3988159 [3] NCCL INFO Channel 14/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-173-121:3987795:3988160 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987797:3988159 [3] NCCL INFO Channel 15/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-173-121:3987798:3988162 [4] NCCL INFO Channel 00/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-173-121:3987795:3988160 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987796:3988163 [2] NCCL INFO Channel 14/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-173-121:3987797:3988159 [3] NCCL INFO Channel 16/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-173-121:3987798:3988162 [4] NCCL INFO Channel 01/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-173-121:3987795:3988160 [1] NCCL INFO Channel 04/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987796:3988163 [2] NCCL INFO Channel 15/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-173-121:3987799:3988158 [5] NCCL INFO Channel 06/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-173-121:3987795:3988160 [1] NCCL INFO Channel 05/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987798:3988162 [4] NCCL INFO Channel 02/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-173-121:3987797:3988159 [3] NCCL INFO Channel 17/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-173-121:3987796:3988163 [2] NCCL INFO Channel 16/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-173-121:3987799:3988158 [5] NCCL INFO Channel 07/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-173-121:3987795:3988160 [1] NCCL INFO Channel 06/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987798:3988162 [4] NCCL INFO Channel 03/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-173-121:3987796:3988163 [2] NCCL INFO Channel 17/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-173-121:3987797:3988159 [3] NCCL INFO Channel 18/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-173-121:3987799:3988158 [5] NCCL INFO Channel 08/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-173-121:3987795:3988160 [1] NCCL INFO Channel 07/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987798:3988162 [4] NCCL INFO Channel 04/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-173-121:3987796:3988163 [2] NCCL INFO Channel 18/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-173-121:3987797:3988159 [3] NCCL INFO Channel 19/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-173-121:3987799:3988158 [5] NCCL INFO Channel 09/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-173-121:3987795:3988160 [1] NCCL INFO Channel 08/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987798:3988162 [4] NCCL INFO Channel 05/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-173-121:3987796:3988163 [2] NCCL INFO Channel 19/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-173-121:3987797:3988159 [3] NCCL INFO Channel 20/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-173-121:3987795:3988160 [1] NCCL INFO Channel 09/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987800:3988161 [6] NCCL INFO Channel 00/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-173-121:3987799:3988158 [5] NCCL INFO Channel 10/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-173-121:3987797:3988159 [3] NCCL INFO Channel 21/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-173-121:3987796:3988163 [2] NCCL INFO Channel 20/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-173-121:3987795:3988160 [1] NCCL INFO Channel 10/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987800:3988161 [6] NCCL INFO Channel 01/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-173-121:3987797:3988159 [3] NCCL INFO Channel 22/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-173-121:3987796:3988163 [2] NCCL INFO Channel 21/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-173-121:3987798:3988162 [4] NCCL INFO Channel 06/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-173-121:3987795:3988160 [1] NCCL INFO Channel 11/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987797:3988159 [3] NCCL INFO Channel 23/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-173-121:3987800:3988161 [6] NCCL INFO Channel 02/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-173-121:3987796:3988163 [2] NCCL INFO Channel 22/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-173-121:3987798:3988162 [4] NCCL INFO Channel 07/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-173-121:3987795:3988160 [1] NCCL INFO Channel 12/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987796:3988163 [2] NCCL INFO Channel 23/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-173-121:3987800:3988161 [6] NCCL INFO Channel 03/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-173-121:3987798:3988162 [4] NCCL INFO Channel 08/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-173-121:3987795:3988160 [1] NCCL INFO Channel 13/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987799:3988158 [5] NCCL INFO Channel 11/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-173-121:3987800:3988161 [6] NCCL INFO Channel 04/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-173-121:3987798:3988162 [4] NCCL INFO Channel 09/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-173-121:3987795:3988160 [1] NCCL INFO Channel 14/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987799:3988158 [5] NCCL INFO Channel 12/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-173-121:3987798:3988162 [4] NCCL INFO Channel 10/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-173-121:3987800:3988161 [6] NCCL INFO Channel 05/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-173-121:3987795:3988160 [1] NCCL INFO Channel 15/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987799:3988158 [5] NCCL INFO Channel 13/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-173-121:3987798:3988162 [4] NCCL INFO Channel 11/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-173-121:3987800:3988161 [6] NCCL INFO Channel 06/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-173-121:3987795:3988160 [1] NCCL INFO Channel 16/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987800:3988161 [6] NCCL INFO Channel 07/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-173-121:3987798:3988162 [4] NCCL INFO Channel 12/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-173-121:3987799:3988158 [5] NCCL INFO Channel 14/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-173-121:3987795:3988160 [1] NCCL INFO Channel 17/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987800:3988161 [6] NCCL INFO Channel 08/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-173-121:3987798:3988162 [4] NCCL INFO Channel 13/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-173-121:3987799:3988158 [5] NCCL INFO Channel 15/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-173-121:3987795:3988160 [1] NCCL INFO Channel 18/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987799:3988158 [5] NCCL INFO Channel 16/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-173-121:3987798:3988162 [4] NCCL INFO Channel 14/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-173-121:3987800:3988161 [6] NCCL INFO Channel 09/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-173-121:3987795:3988160 [1] NCCL INFO Channel 19/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987800:3988161 [6] NCCL INFO Channel 10/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-173-121:3987799:3988158 [5] NCCL INFO Channel 17/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-173-121:3987798:3988162 [4] NCCL INFO Channel 15/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-173-121:3987795:3988160 [1] NCCL INFO Channel 20/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987800:3988161 [6] NCCL INFO Channel 11/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-173-121:3987799:3988158 [5] NCCL INFO Channel 18/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-173-121:3987798:3988162 [4] NCCL INFO Channel 16/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-173-121:3987795:3988160 [1] NCCL INFO Channel 21/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987800:3988161 [6] NCCL INFO Channel 12/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-173-121:3987799:3988158 [5] NCCL INFO Channel 19/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-173-121:3987798:3988162 [4] NCCL INFO Channel 17/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-173-121:3987795:3988160 [1] NCCL INFO Channel 22/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987799:3988158 [5] NCCL INFO Channel 20/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-173-121:3987800:3988161 [6] NCCL INFO Channel 13/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-173-121:3987798:3988162 [4] NCCL INFO Channel 18/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-173-121:3987795:3988160 [1] NCCL INFO Channel 23/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987798:3988162 [4] NCCL INFO Channel 19/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-173-121:3987800:3988161 [6] NCCL INFO Channel 14/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-173-121:3987799:3988158 [5] NCCL INFO Channel 21/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-173-121:3987800:3988161 [6] NCCL INFO Channel 15/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-173-121:3987798:3988162 [4] NCCL INFO Channel 20/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-173-121:3987799:3988158 [5] NCCL INFO Channel 22/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-173-121:3987798:3988162 [4] NCCL INFO Channel 21/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-173-121:3987798:3988162 [4] NCCL INFO Channel 22/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-173-121:3987799:3988158 [5] NCCL INFO Channel 23/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-173-121:3987800:3988161 [6] NCCL INFO Channel 16/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-142:789256:789628 [1] NCCL INFO Connected all rings
+ip-26-0-173-121:3987798:3988162 [4] NCCL INFO Channel 23/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-173-121:3987800:3988161 [6] NCCL INFO Channel 17/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-173-121:3987800:3988161 [6] NCCL INFO Channel 18/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-173-121:3987800:3988161 [6] NCCL INFO Channel 19/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-173-121:3987800:3988161 [6] NCCL INFO Channel 20/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-173-121:3987800:3988161 [6] NCCL INFO Channel 21/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-173-121:3987800:3988161 [6] NCCL INFO Channel 22/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-142:789261:789623 [6] NCCL INFO Connected all rings
+ip-26-0-173-121:3987800:3988161 [6] NCCL INFO Channel 23/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-142:789262:789627 [7] NCCL INFO Connected all rings
+ip-26-0-172-142:789262:789627 [7] NCCL INFO Channel 00/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-142:789259:789626 [4] NCCL INFO Connected all rings
+ip-26-0-172-142:789260:789624 [5] NCCL INFO Connected all rings
+ip-26-0-172-142:789262:789627 [7] NCCL INFO Channel 01/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-142:789255:789622 [0] NCCL INFO Connected all rings
+ip-26-0-172-142:789262:789627 [7] NCCL INFO Channel 02/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-142:789258:789629 [3] NCCL INFO Connected all rings
+ip-26-0-172-142:789262:789627 [7] NCCL INFO Channel 03/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-142:789262:789627 [7] NCCL INFO Channel 04/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-142:789262:789627 [7] NCCL INFO Channel 05/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-142:789262:789627 [7] NCCL INFO Channel 06/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-142:789262:789627 [7] NCCL INFO Channel 07/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-142:789262:789627 [7] NCCL INFO Channel 08/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-142:789262:789627 [7] NCCL INFO Channel 09/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-142:789262:789627 [7] NCCL INFO Channel 10/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-142:789262:789627 [7] NCCL INFO Channel 11/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-142:789262:789627 [7] NCCL INFO Channel 12/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-142:789262:789627 [7] NCCL INFO Channel 13/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-142:789262:789627 [7] NCCL INFO Channel 14/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-142:789262:789627 [7] NCCL INFO Channel 15/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-142:789257:789625 [2] NCCL INFO Channel 00/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-142:789262:789627 [7] NCCL INFO Channel 16/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-142:789257:789625 [2] NCCL INFO Channel 01/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-142:789262:789627 [7] NCCL INFO Channel 17/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-142:789257:789625 [2] NCCL INFO Channel 02/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-142:789262:789627 [7] NCCL INFO Channel 18/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-142:789257:789625 [2] NCCL INFO Channel 03/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-142:789262:789627 [7] NCCL INFO Channel 19/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-142:789257:789625 [2] NCCL INFO Channel 04/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-142:789259:789626 [4] NCCL INFO Channel 00/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-142:789256:789628 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789262:789627 [7] NCCL INFO Channel 20/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-142:789262:789627 [7] NCCL INFO Channel 21/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-142:789257:789625 [2] NCCL INFO Channel 05/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-142:789256:789628 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789261:789623 [6] NCCL INFO Channel 00/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-142:789257:789625 [2] NCCL INFO Channel 06/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-142:789256:789628 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789261:789623 [6] NCCL INFO Channel 01/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-142:789257:789625 [2] NCCL INFO Channel 07/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-142:789256:789628 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789261:789623 [6] NCCL INFO Channel 02/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-142:789257:789625 [2] NCCL INFO Channel 08/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-142:789256:789628 [1] NCCL INFO Channel 04/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789261:789623 [6] NCCL INFO Channel 03/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-142:789257:789625 [2] NCCL INFO Channel 09/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-142:789256:789628 [1] NCCL INFO Channel 05/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789261:789623 [6] NCCL INFO Channel 04/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-142:789257:789625 [2] NCCL INFO Channel 10/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-142:789256:789628 [1] NCCL INFO Channel 06/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789261:789623 [6] NCCL INFO Channel 05/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-142:789257:789625 [2] NCCL INFO Channel 11/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-142:789256:789628 [1] NCCL INFO Channel 07/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789261:789623 [6] NCCL INFO Channel 06/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-142:789256:789628 [1] NCCL INFO Channel 08/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789257:789625 [2] NCCL INFO Channel 12/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-142:789259:789626 [4] NCCL INFO Channel 01/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-142:789261:789623 [6] NCCL INFO Channel 07/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-142:789262:789627 [7] NCCL INFO Channel 22/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-142:789256:789628 [1] NCCL INFO Channel 09/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789257:789625 [2] NCCL INFO Channel 13/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-142:789259:789626 [4] NCCL INFO Channel 02/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-142:789261:789623 [6] NCCL INFO Channel 08/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-142:789260:789624 [5] NCCL INFO Channel 00/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-142:789262:789627 [7] NCCL INFO Channel 23/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-142:789257:789625 [2] NCCL INFO Channel 14/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-142:789256:789628 [1] NCCL INFO Channel 10/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789259:789626 [4] NCCL INFO Channel 03/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-142:789261:789623 [6] NCCL INFO Channel 09/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-142:789260:789624 [5] NCCL INFO Channel 01/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-142:789257:789625 [2] NCCL INFO Channel 15/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-142:789256:789628 [1] NCCL INFO Channel 11/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789261:789623 [6] NCCL INFO Channel 10/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-142:789259:789626 [4] NCCL INFO Channel 04/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-142:789260:789624 [5] NCCL INFO Channel 02/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-142:789257:789625 [2] NCCL INFO Channel 16/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-142:789256:789628 [1] NCCL INFO Channel 12/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789261:789623 [6] NCCL INFO Channel 11/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-142:789259:789626 [4] NCCL INFO Channel 05/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-142:789258:789629 [3] NCCL INFO Channel 00/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-142:789260:789624 [5] NCCL INFO Channel 03/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-142:789257:789625 [2] NCCL INFO Channel 17/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-142:789256:789628 [1] NCCL INFO Channel 13/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789259:789626 [4] NCCL INFO Channel 06/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-142:789261:789623 [6] NCCL INFO Channel 12/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-142:789258:789629 [3] NCCL INFO Channel 01/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-173-121:3987794:3988157 [0] NCCL INFO Connected all trees
+ip-26-0-172-142:789257:789625 [2] NCCL INFO Channel 18/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-142:789260:789624 [5] NCCL INFO Channel 04/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-142:789256:789628 [1] NCCL INFO Channel 14/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987794:3988157 [0] NCCL INFO NVLS comm 0x8e7b3f0 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-142:789259:789626 [4] NCCL INFO Channel 07/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-142:789261:789623 [6] NCCL INFO Channel 13/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-142:789258:789629 [3] NCCL INFO Channel 02/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-142:789260:789624 [5] NCCL INFO Channel 05/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-142:789259:789626 [4] NCCL INFO Channel 08/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-142:789258:789629 [3] NCCL INFO Channel 03/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-142:789260:789624 [5] NCCL INFO Channel 06/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-142:789256:789628 [1] NCCL INFO Channel 15/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789259:789626 [4] NCCL INFO Channel 09/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-142:789258:789629 [3] NCCL INFO Channel 04/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-142:789256:789628 [1] NCCL INFO Channel 16/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789260:789624 [5] NCCL INFO Channel 07/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-142:789258:789629 [3] NCCL INFO Channel 05/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-142:789259:789626 [4] NCCL INFO Channel 10/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-142:789256:789628 [1] NCCL INFO Channel 17/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789260:789624 [5] NCCL INFO Channel 08/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-142:789258:789629 [3] NCCL INFO Channel 06/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-142:789259:789626 [4] NCCL INFO Channel 11/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-142:789256:789628 [1] NCCL INFO Channel 18/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789260:789624 [5] NCCL INFO Channel 09/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-142:789258:789629 [3] NCCL INFO Channel 07/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-142:789259:789626 [4] NCCL INFO Channel 12/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-142:789260:789624 [5] NCCL INFO Channel 10/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-142:789256:789628 [1] NCCL INFO Channel 19/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789257:789625 [2] NCCL INFO Channel 19/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-142:789258:789629 [3] NCCL INFO Channel 08/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-142:789259:789626 [4] NCCL INFO Channel 13/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-142:789260:789624 [5] NCCL INFO Channel 11/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-142:789256:789628 [1] NCCL INFO Channel 20/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789261:789623 [6] NCCL INFO Channel 14/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-142:789257:789625 [2] NCCL INFO Channel 20/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-142:789258:789629 [3] NCCL INFO Channel 09/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-142:789259:789626 [4] NCCL INFO Channel 14/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-142:789260:789624 [5] NCCL INFO Channel 12/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-142:789256:789628 [1] NCCL INFO Channel 21/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789261:789623 [6] NCCL INFO Channel 15/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-142:789257:789625 [2] NCCL INFO Channel 21/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-142:789258:789629 [3] NCCL INFO Channel 10/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-142:789259:789626 [4] NCCL INFO Channel 15/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-142:789260:789624 [5] NCCL INFO Channel 13/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-142:789256:789628 [1] NCCL INFO Channel 22/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789261:789623 [6] NCCL INFO Channel 16/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-142:789258:789629 [3] NCCL INFO Channel 11/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-142:789257:789625 [2] NCCL INFO Channel 22/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-142:789259:789626 [4] NCCL INFO Channel 16/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-142:789260:789624 [5] NCCL INFO Channel 14/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-142:789256:789628 [1] NCCL INFO Channel 23/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789261:789623 [6] NCCL INFO Channel 17/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-142:789257:789625 [2] NCCL INFO Channel 23/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-142:789258:789629 [3] NCCL INFO Channel 12/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-142:789259:789626 [4] NCCL INFO Channel 17/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-142:789260:789624 [5] NCCL INFO Channel 15/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-142:789261:789623 [6] NCCL INFO Channel 18/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-142:789258:789629 [3] NCCL INFO Channel 13/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-142:789259:789626 [4] NCCL INFO Channel 18/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-142:789260:789624 [5] NCCL INFO Channel 16/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-142:789261:789623 [6] NCCL INFO Channel 19/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-142:789258:789629 [3] NCCL INFO Channel 14/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-142:789259:789626 [4] NCCL INFO Channel 19/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-142:789260:789624 [5] NCCL INFO Channel 17/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-142:789261:789623 [6] NCCL INFO Channel 20/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-142:789258:789629 [3] NCCL INFO Channel 15/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-142:789259:789626 [4] NCCL INFO Channel 20/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-142:789260:789624 [5] NCCL INFO Channel 18/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-142:789261:789623 [6] NCCL INFO Channel 21/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-142:789258:789629 [3] NCCL INFO Channel 16/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-142:789259:789626 [4] NCCL INFO Channel 21/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-142:789260:789624 [5] NCCL INFO Channel 19/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-142:789261:789623 [6] NCCL INFO Channel 22/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-142:789259:789626 [4] NCCL INFO Channel 22/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-142:789260:789624 [5] NCCL INFO Channel 20/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-142:789261:789623 [6] NCCL INFO Channel 23/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-142:789260:789624 [5] NCCL INFO Channel 21/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-142:789260:789624 [5] NCCL INFO Channel 22/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-142:789260:789624 [5] NCCL INFO Channel 23/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-142:789258:789629 [3] NCCL INFO Channel 17/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-142:789259:789626 [4] NCCL INFO Channel 23/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-142:789258:789629 [3] NCCL INFO Channel 18/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-142:789258:789629 [3] NCCL INFO Channel 19/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-142:789258:789629 [3] NCCL INFO Channel 20/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-142:789258:789629 [3] NCCL INFO Channel 21/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-142:789258:789629 [3] NCCL INFO Channel 22/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-142:789258:789629 [3] NCCL INFO Channel 23/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-173-121:3987801:3988164 [7] NCCL INFO Connected all trees
+ip-26-0-173-121:3987801:3988164 [7] NCCL INFO NVLS comm 0x98674d0 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-142:789255:789622 [0] NCCL INFO Connected all trees
+ip-26-0-172-142:789255:789622 [0] NCCL INFO NVLS comm 0xa2175e0 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-173-121:3987795:3988160 [1] NCCL INFO Connected all trees
+ip-26-0-173-121:3987795:3988160 [1] NCCL INFO NVLS comm 0x9a3fc10 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-173-121:3987796:3988163 [2] NCCL INFO Connected all trees
+ip-26-0-173-121:3987796:3988163 [2] NCCL INFO NVLS comm 0x92db530 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-173-121:3987797:3988159 [3] NCCL INFO Connected all trees
+ip-26-0-173-121:3987797:3988159 [3] NCCL INFO NVLS comm 0x8db2350 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-173-121:3987798:3988162 [4] NCCL INFO Connected all trees
+ip-26-0-173-121:3987798:3988162 [4] NCCL INFO NVLS comm 0x9e0dcf0 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-173-121:3987800:3988161 [6] NCCL INFO Connected all trees
+ip-26-0-173-121:3987800:3988161 [6] NCCL INFO NVLS comm 0x9456580 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-142:789262:789627 [7] NCCL INFO Connected all trees
+ip-26-0-172-142:789262:789627 [7] NCCL INFO NVLS comm 0x8c38eb0 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-173-121:3987799:3988158 [5] NCCL INFO Connected all trees
+ip-26-0-173-121:3987799:3988158 [5] NCCL INFO NVLS comm 0x940dfd0 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-173-121:3987797:3988159 [3] NCCL INFO Connected NVLS tree
+ip-26-0-173-121:3987797:3988159 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-173-121:3987797:3988159 [3] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-173-121:3987797:3988159 [3] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-173-121:3987799:3988158 [5] NCCL INFO Connected NVLS tree
+ip-26-0-173-121:3987799:3988158 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-173-121:3987799:3988158 [5] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-173-121:3987799:3988158 [5] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-173-121:3987795:3988160 [1] NCCL INFO Connected NVLS tree
+ip-26-0-173-121:3987795:3988160 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-173-121:3987795:3988160 [1] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-173-121:3987795:3988160 [1] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-173-121:3987801:3988164 [7] NCCL INFO Connected NVLS tree
+ip-26-0-173-121:3987801:3988164 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-173-121:3987801:3988164 [7] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-173-121:3987801:3988164 [7] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-173-121:3987800:3988161 [6] NCCL INFO Connected NVLS tree
+ip-26-0-173-121:3987800:3988161 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-173-121:3987800:3988161 [6] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-173-121:3987800:3988161 [6] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-173-121:3987794:3988157 [0] NCCL INFO Connected NVLS tree
+ip-26-0-173-121:3987794:3988157 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-173-121:3987796:3988163 [2] NCCL INFO Connected NVLS tree
+ip-26-0-173-121:3987796:3988163 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-173-121:3987796:3988163 [2] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-173-121:3987796:3988163 [2] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-173-121:3987794:3988157 [0] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-173-121:3987794:3988157 [0] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-173-121:3987798:3988162 [4] NCCL INFO Connected NVLS tree
+ip-26-0-173-121:3987798:3988162 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-173-121:3987798:3988162 [4] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-173-121:3987798:3988162 [4] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-142:789256:789628 [1] NCCL INFO Connected all trees
+ip-26-0-172-142:789256:789628 [1] NCCL INFO NVLS comm 0xa56a1d0 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-173-121:3987800:3988161 [6] NCCL INFO comm 0x9456580 rank 6 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0xcef8427ac1151735 - Init COMPLETE
+ip-26-0-173-121:3987796:3988163 [2] NCCL INFO comm 0x92db530 rank 2 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0xcef8427ac1151735 - Init COMPLETE
+ip-26-0-173-121:3987794:3988157 [0] NCCL INFO comm 0x8e7b3f0 rank 0 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0xcef8427ac1151735 - Init COMPLETE
+ip-26-0-173-121:3987798:3988162 [4] NCCL INFO comm 0x9e0dcf0 rank 4 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0xcef8427ac1151735 - Init COMPLETE
+ip-26-0-173-121:3987799:3988158 [5] NCCL INFO comm 0x940dfd0 rank 5 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0xcef8427ac1151735 - Init COMPLETE
+ip-26-0-173-121:3987797:3988159 [3] NCCL INFO comm 0x8db2350 rank 3 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0xcef8427ac1151735 - Init COMPLETE
+ip-26-0-173-121:3987801:3988164 [7] NCCL INFO comm 0x98674d0 rank 7 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0xcef8427ac1151735 - Init COMPLETE
+ip-26-0-173-121:3987795:3988160 [1] NCCL INFO comm 0x9a3fc10 rank 1 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0xcef8427ac1151735 - Init COMPLETE
+ip-26-0-172-142:789260:789624 [5] NCCL INFO Connected all trees
+ip-26-0-172-142:789260:789624 [5] NCCL INFO NVLS comm 0xa2a0b50 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-142:789261:789623 [6] NCCL INFO Connected all trees
+ip-26-0-172-142:789261:789623 [6] NCCL INFO NVLS comm 0xa01c8d0 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-142:789259:789626 [4] NCCL INFO Connected all trees
+ip-26-0-172-142:789259:789626 [4] NCCL INFO NVLS comm 0x927d120 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-142:789258:789629 [3] NCCL INFO Connected all trees
+ip-26-0-172-142:789258:789629 [3] NCCL INFO NVLS comm 0x8d5bae0 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-142:789257:789625 [2] NCCL INFO Connected all trees
+ip-26-0-172-142:789257:789625 [2] NCCL INFO NVLS comm 0x95d7160 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-142:789262:789627 [7] NCCL INFO Connected NVLS tree
+ip-26-0-172-142:789262:789627 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-142:789262:789627 [7] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-172-142:789262:789627 [7] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-142:789260:789624 [5] NCCL INFO Connected NVLS tree
+ip-26-0-172-142:789260:789624 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-142:789260:789624 [5] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-172-142:789256:789628 [1] NCCL INFO Connected NVLS tree
+ip-26-0-172-142:789260:789624 [5] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-142:789256:789628 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-142:789256:789628 [1] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-172-142:789256:789628 [1] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-142:789258:789629 [3] NCCL INFO Connected NVLS tree
+ip-26-0-172-142:789258:789629 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-142:789258:789629 [3] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-172-142:789258:789629 [3] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-142:789261:789623 [6] NCCL INFO Connected NVLS tree
+ip-26-0-172-142:789261:789623 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-142:789261:789623 [6] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-172-142:789261:789623 [6] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-142:789259:789626 [4] NCCL INFO Connected NVLS tree
+ip-26-0-172-142:789259:789626 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-142:789259:789626 [4] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-172-142:789259:789626 [4] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-142:789257:789625 [2] NCCL INFO Connected NVLS tree
+ip-26-0-172-142:789257:789625 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-142:789257:789625 [2] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-172-142:789257:789625 [2] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-142:789255:789622 [0] NCCL INFO Connected NVLS tree
+ip-26-0-172-142:789255:789622 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-142:789255:789622 [0] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-172-142:789255:789622 [0] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-142:789259:789626 [4] NCCL INFO comm 0x927d120 rank 4 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0x98e054a6265db20 - Init COMPLETE
+ip-26-0-172-142:789257:789625 [2] NCCL INFO comm 0x95d7160 rank 2 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0x98e054a6265db20 - Init COMPLETE
+ip-26-0-172-142:789261:789623 [6] NCCL INFO comm 0xa01c8d0 rank 6 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0x98e054a6265db20 - Init COMPLETE
+ip-26-0-172-142:789256:789628 [1] NCCL INFO comm 0xa56a1d0 rank 1 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0x98e054a6265db20 - Init COMPLETE
+ip-26-0-172-142:789255:789622 [0] NCCL INFO comm 0xa2175e0 rank 0 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0x98e054a6265db20 - Init COMPLETE
+ip-26-0-172-142:789260:789624 [5] NCCL INFO comm 0xa2a0b50 rank 5 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0x98e054a6265db20 - Init COMPLETE
+ip-26-0-172-142:789258:789629 [3] NCCL INFO comm 0x8d5bae0 rank 3 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0x98e054a6265db20 - Init COMPLETE
+ip-26-0-172-142:789262:789627 [7] NCCL INFO comm 0x8c38eb0 rank 7 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0x98e054a6265db20 - Init COMPLETE
+01/07/2025 00:37:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]: Config:
+01/07/2025 00:37:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]: Config(general=GeneralArgs(project='debug',
+01/07/2025 00:37:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]:                            run='1.34G_dp2_tp1_pp8_acc64_mbs2_seq4096_zero1_tpmodeRED_vocab131k',
+01/07/2025 00:37:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]:                            seed=42,
+01/07/2025 00:37:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]:                            step=None,
+01/07/2025 00:37:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]:                            consumed_train_samples=None,
+01/07/2025 00:37:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]:                            benchmark_csv_path=PosixPath('benchmark/results/bench_final2.csv'),
+01/07/2025 00:37:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]:                            ignore_sanity_checks=True),
+01/07/2025 00:37:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]:        parallelism=ParallelismArgs(dp=2,
+01/07/2025 00:37:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]:                                    pp=8,
+01/07/2025 00:37:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]:                                    tp=1,
+01/07/2025 00:37:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]:                                    pp_engine=<nanotron.parallel.pipeline_parallel.engine.OneForwardOneBackwardPipelineEngine object at 0x7f4af7ed4df0>,
+01/07/2025 00:37:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]:                                    tp_mode=<TensorParallelLinearMode.REDUCE_SCATTER: 2>,
+01/07/2025 00:37:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]:                                    tp_linear_async_communication=True,
+01/07/2025 00:37:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]:                                    recompute_layer=False,
+01/07/2025 00:37:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]:                                    tp_recompute_allgather=True,
+01/07/2025 00:37:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]:                                    expert_parallel_size=1),
+01/07/2025 00:37:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]:        model=ModelArgs(model_config=LlamaConfig(bos_token_id=0,
+01/07/2025 00:37:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]:                                                 eos_token_id=0,
+01/07/2025 00:37:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]:                                                 hidden_act='silu',
+01/07/2025 00:37:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]:                                                 hidden_size=2048,
+01/07/2025 00:37:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]:                                                 initializer_range=0.02,
+01/07/2025 00:37:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]:                                                 intermediate_size=8192,
+01/07/2025 00:37:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]:                                                 is_llama_config=True,
+01/07/2025 00:37:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]:                                                 max_position_embeddings=4096,
+01/07/2025 00:37:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]:                                                 num_attention_heads=32,
+01/07/2025 00:37:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]:                                                 num_hidden_layers=16,
+01/07/2025 00:37:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]:                                                 num_key_value_heads=32,
+01/07/2025 00:37:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]:                                                 pad_token_id=None,
+01/07/2025 00:37:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]:                                                 pretraining_tp=1,
+01/07/2025 00:37:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]:                                                 rms_norm_eps=1e-05,
+01/07/2025 00:37:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]:                                                 rope_scaling=None,
+01/07/2025 00:37:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]:                                                 rope_theta=10000.0,
+01/07/2025 00:37:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]:                                                 rope_interleaved=False,
+01/07/2025 00:37:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]:                                                 tie_word_embeddings=True,
+01/07/2025 00:37:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]:                                                 use_cache=True,
+01/07/2025 00:37:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]:                                                 vocab_size=131072),
+01/07/2025 00:37:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]:                        init_method=RandomInit(std=0.02),
+01/07/2025 00:37:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]:                        dtype=torch.bfloat16,
+01/07/2025 00:37:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]:                        make_vocab_size_divisible_by=1,
+01/07/2025 00:37:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]:                        ddp_bucket_cap_mb=25),
+01/07/2025 00:37:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]:        tokenizer=TokenizerArgs(tokenizer_name_or_path='robot-test/dummy-tokenizer-wordlevel',
+01/07/2025 00:37:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]:                                tokenizer_revision=None,
+01/07/2025 00:37:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]:                                tokenizer_max_length=None),
+01/07/2025 00:37:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]:        checkpoints=CheckpointsArgs(checkpoints_path=PosixPath('checkpoints'),
+01/07/2025 00:37:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]:                                    checkpoint_interval=10000,
+01/07/2025 00:37:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]:                                    save_initial_state=False,
+01/07/2025 00:37:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]:                                    save_final_state=False,
+01/07/2025 00:37:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]:                                    resume_checkpoint_path=None,
+01/07/2025 00:37:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]:                                    checkpoints_path_is_shared_file_system=False),
+01/07/2025 00:37:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]:        logging=LoggingArgs(log_level='info',
+01/07/2025 00:37:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]:                            log_level_replica='info',
+01/07/2025 00:37:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]:                            iteration_step_info_interval=1),
+01/07/2025 00:37:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]:        tokens=TokensArgs(sequence_length=4096,
+01/07/2025 00:37:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]:                          train_steps=100,
+01/07/2025 00:37:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]:                          micro_batch_size=2,
+01/07/2025 00:37:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]:                          batch_accumulation_per_replica=64,
+01/07/2025 00:37:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]:                          val_check_interval=100,
+01/07/2025 00:37:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]:                          limit_val_batches=0,
+01/07/2025 00:37:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]:                          limit_test_batches=0),
+01/07/2025 00:37:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]:        optimizer=OptimizerArgs(optimizer_factory=AdamWOptimizerArgs(adam_eps=1e-08,
+01/07/2025 00:37:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]:                                                                     adam_beta1=0.9,
+01/07/2025 00:37:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]:                                                                     adam_beta2=0.95,
+01/07/2025 00:37:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]:                                                                     torch_adam_is_fused=True,
+01/07/2025 00:37:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]:                                                                     name='adamW'),
+01/07/2025 00:37:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]:                                zero_stage=1,
+01/07/2025 00:37:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]:                                weight_decay=0.01,
+01/07/2025 00:37:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]:                                clip_grad=1.0,
+01/07/2025 00:37:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]:                                accumulate_grad_in_fp32=True,
+01/07/2025 00:37:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]:                                learning_rate_scheduler=LRSchedulerArgs(learning_rate=0.0003,
+01/07/2025 00:37:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]:                                                                        lr_warmup_steps=2,
+01/07/2025 00:37:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]:                                                                        lr_warmup_style='linear',
+01/07/2025 00:37:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]:                                                                        lr_decay_style='cosine',
+01/07/2025 00:37:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]:                                                                        lr_decay_steps=13,
+01/07/2025 00:37:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]:                                                                        lr_decay_starting_step=None,
+01/07/2025 00:37:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]:                                                                        min_decay_lr=1e-05)),
+01/07/2025 00:37:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]:        data_stages=[DatasetStageArgs(name='Stable Training Stage',
+01/07/2025 00:37:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]:                                      start_training_step=1,
+01/07/2025 00:37:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]:                                      data=DataArgs(dataset=None,
+01/07/2025 00:37:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]:                                                    seed=42,
+01/07/2025 00:37:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]:                                                    num_loading_workers=1))],
+01/07/2025 00:37:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]:        profiler=None,
+01/07/2025 00:37:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]:        lighteval=None,
+01/07/2025 00:37:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]:        s3_upload=None)
+01/07/2025 00:37:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]: Model Config:
+01/07/2025 00:37:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]: LlamaConfig(bos_token_id=0,
+01/07/2025 00:37:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]:             eos_token_id=0,
+01/07/2025 00:37:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]:             hidden_act='silu',
+01/07/2025 00:37:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]:             hidden_size=2048,
+01/07/2025 00:37:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]:             initializer_range=0.02,
+01/07/2025 00:37:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]:             intermediate_size=8192,
+01/07/2025 00:37:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]:             is_llama_config=True,
+01/07/2025 00:37:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]:             max_position_embeddings=4096,
+01/07/2025 00:37:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]:             num_attention_heads=32,
+01/07/2025 00:37:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]:             num_hidden_layers=16,
+01/07/2025 00:37:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]:             num_key_value_heads=32,
+01/07/2025 00:37:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]:             pad_token_id=None,
+01/07/2025 00:37:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]:             pretraining_tp=1,
+01/07/2025 00:37:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]:             rms_norm_eps=1e-05,
+01/07/2025 00:37:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]:             rope_scaling=None,
+01/07/2025 00:37:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]:             rope_theta=10000.0,
+01/07/2025 00:37:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]:             rope_interleaved=False,
+01/07/2025 00:37:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]:             tie_word_embeddings=True,
+01/07/2025 00:37:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]:             use_cache=True,
+01/07/2025 00:37:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]:             vocab_size=131072)
+01/07/2025 00:37:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]: Building model..
+01/07/2025 00:37:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]: Initialize RoPE Theta = 10000.0
+01/07/2025 00:37:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]: Setting PP block ranks...
+01/07/2025 00:37:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]: module_name: model.token_position_embeddings | PP: 0/8 | Block rank: 0
+01/07/2025 00:37:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]: module_name: model.decoder.0 | PP: 0/8 | Block rank: 1
+01/07/2025 00:37:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]: module_name: model.decoder.1 | PP: 0/8 | Block rank: 2
+01/07/2025 00:37:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]: module_name: model.decoder.2 | PP: 0/8 | Block rank: 3
+01/07/2025 00:37:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]: module_name: model.decoder.3 | PP: 1/8 | Block rank: 0
+01/07/2025 00:37:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]: module_name: model.decoder.4 | PP: 1/8 | Block rank: 1
+01/07/2025 00:37:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]: module_name: model.decoder.5 | PP: 1/8 | Block rank: 2
+01/07/2025 00:37:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]: module_name: model.decoder.6 | PP: 2/8 | Block rank: 0
+01/07/2025 00:37:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]: module_name: model.decoder.7 | PP: 2/8 | Block rank: 1
+01/07/2025 00:37:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]: module_name: model.decoder.8 | PP: 2/8 | Block rank: 2
+01/07/2025 00:37:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]: module_name: model.decoder.9 | PP: 3/8 | Block rank: 0
+01/07/2025 00:37:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]: module_name: model.decoder.10 | PP: 3/8 | Block rank: 1
+01/07/2025 00:37:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]: module_name: model.decoder.11 | PP: 3/8 | Block rank: 2
+01/07/2025 00:37:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]: module_name: model.decoder.12 | PP: 4/8 | Block rank: 0
+01/07/2025 00:37:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]: module_name: model.decoder.13 | PP: 4/8 | Block rank: 1
+01/07/2025 00:37:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]: module_name: model.decoder.14 | PP: 5/8 | Block rank: 0
+01/07/2025 00:37:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]: module_name: model.decoder.15 | PP: 6/8 | Block rank: 0
+01/07/2025 00:37:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]: module_name: model.final_layer_norm | PP: 7/8 | Block rank: 0
+01/07/2025 00:37:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]: module_name: model.lm_head | PP: 7/8 | Block rank: 1
+01/07/2025 00:37:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]: module_name: model.cast_to_fp32 | PP: 7/8 | Block rank: 2
+01/07/2025 00:37:02 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]: module_name: loss | PP: 7/8 | Block rank: 3
+NCCL version 2.18.6+cuda12.1
+NCCL version 2.18.6+cuda12.1
+NCCL version 2.18.6+cuda12.1
+NCCL version 2.18.6+cuda12.1
+NCCL version 2.18.6+cuda12.1
+NCCL version 2.18.6+cuda12.1
+NCCL version 2.18.6+cuda12.1
+NCCL version 2.18.6+cuda12.1
+NCCL version 2.18.6+cuda12.1
+NCCL version 2.18.6+cuda12.1
+NCCL version 2.18.6+cuda12.1
+ip-26-0-172-142:789261:789678 [6] NCCL INFO Using network Libfabric
+ip-26-0-172-142:789260:789677 [5] NCCL INFO Using network Libfabric
+ip-26-0-172-142:789259:789683 [4] NCCL INFO Using network Libfabric
+ip-26-0-172-142:789261:789678 [6] NCCL INFO comm 0xbd1d5f0 rank 0 nranks 1 cudaDev 6 nvmlDev 6 busId b9000 commId 0x65112659cd2179b2 - Init START
+ip-26-0-172-142:789262:789680 [7] NCCL INFO Using network Libfabric
+ip-26-0-172-142:789260:789677 [5] NCCL INFO comm 0xbfa1a80 rank 0 nranks 1 cudaDev 5 nvmlDev 5 busId a8000 commId 0xfc05f86a91bb9bd - Init START
+ip-26-0-172-142:789258:789684 [3] NCCL INFO Using network Libfabric
+ip-26-0-172-142:789260:789677 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-172-142:789259:789683 [4] NCCL INFO comm 0xaf7be30 rank 0 nranks 1 cudaDev 4 nvmlDev 4 busId 97000 commId 0x299f92c3f3ff3cc8 - Init START
+ip-26-0-172-142:789261:789678 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-172-142:789257:789685 [2] NCCL INFO Using network Libfabric
+ip-26-0-172-142:789259:789683 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-173-121:3987797:3988211 [3] NCCL INFO Using network Libfabric
+ip-26-0-173-121:3987794:3988214 [0] NCCL INFO Using network Libfabric
+ip-26-0-172-142:789262:789680 [7] NCCL INFO comm 0xa93a770 rank 0 nranks 1 cudaDev 7 nvmlDev 7 busId ca000 commId 0x7f37ce22192b3cb9 - Init START
+ip-26-0-172-142:789258:789684 [3] NCCL INFO comm 0xaa5bbb0 rank 0 nranks 1 cudaDev 3 nvmlDev 3 busId 86000 commId 0x3d2b6343e711457 - Init START
+ip-26-0-172-142:789262:789680 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-173-121:3987798:3988213 [4] NCCL INFO Using network Libfabric
+ip-26-0-173-121:3987796:3988210 [2] NCCL INFO Using network Libfabric
+ip-26-0-172-142:789257:789685 [2] NCCL INFO comm 0xb2d77f0 rank 0 nranks 1 cudaDev 2 nvmlDev 2 busId 75000 commId 0x31ea480d69bce7af - Init START
+ip-26-0-172-142:789258:789684 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-172-142:789257:789685 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-173-121:3987795:3988215 [1] NCCL INFO Using network Libfabric
+ip-26-0-173-121:3987797:3988211 [3] NCCL INFO comm 0xaaad830 rank 0 nranks 1 cudaDev 3 nvmlDev 3 busId 86000 commId 0xa04a18d93d84e9ae - Init START
+ip-26-0-173-121:3987794:3988214 [0] NCCL INFO comm 0xab7c0d0 rank 0 nranks 1 cudaDev 0 nvmlDev 0 busId 53000 commId 0x111566d954b745ba - Init START
+ip-26-0-173-121:3987797:3988211 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-173-121:3987794:3988214 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-173-121:3987798:3988213 [4] NCCL INFO comm 0xbb0a880 rank 0 nranks 1 cudaDev 4 nvmlDev 4 busId 97000 commId 0x953f1b590b4d8b44 - Init START
+ip-26-0-173-121:3987796:3988210 [2] NCCL INFO comm 0xafd75a0 rank 0 nranks 1 cudaDev 2 nvmlDev 2 busId 75000 commId 0xf536fd0616c61a14 - Init START
+ip-26-0-173-121:3987798:3988213 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-173-121:3987796:3988210 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-173-121:3987795:3988215 [1] NCCL INFO comm 0xb73fa40 rank 0 nranks 1 cudaDev 1 nvmlDev 1 busId 64000 commId 0xa5fcc36c4b51c4c - Init START
+ip-26-0-173-121:3987799:3988217 [5] NCCL INFO Using network Libfabric
+ip-26-0-173-121:3987795:3988215 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-173-121:3987799:3988217 [5] NCCL INFO comm 0xb109ef0 rank 0 nranks 1 cudaDev 5 nvmlDev 5 busId a8000 commId 0x5cf0af34367bb8c4 - Init START
+ip-26-0-173-121:3987799:3988217 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+NCCL version 2.18.6+cuda12.1
+NCCL version 2.18.6+cuda12.1
+NCCL version 2.18.6+cuda12.1
+ip-26-0-172-142:789255:789693 [0] NCCL INFO Using network Libfabric
+ip-26-0-172-142:789255:789693 [0] NCCL INFO comm 0xc0dfdc0 rank 0 nranks 1 cudaDev 0 nvmlDev 0 busId 53000 commId 0xa15da757bb15a5fd - Init START
+ip-26-0-172-142:789255:789693 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/205
+ip-26-0-173-121:3987800:3988226 [6] NCCL INFO Using network Libfabric
+ip-26-0-173-121:3987800:3988226 [6] NCCL INFO comm 0xa72f590 rank 0 nranks 1 cudaDev 6 nvmlDev 6 busId b9000 commId 0xb636512b59c0f485 - Init START
+ip-26-0-173-121:3987800:3988226 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-173-121:3987801:3988227 [7] NCCL INFO Using network Libfabric
+ip-26-0-172-142:789256:789695 [1] NCCL INFO Using network Libfabric
+ip-26-0-172-142:789256:789695 [1] NCCL INFO comm 0xc42ddb0 rank 0 nranks 1 cudaDev 1 nvmlDev 1 busId 64000 commId 0x2486e57acb78929c - Init START
+ip-26-0-173-121:3987801:3988227 [7] NCCL INFO comm 0xab41930 rank 0 nranks 1 cudaDev 7 nvmlDev 7 busId ca000 commId 0x89a5658a982d7fda - Init START
+ip-26-0-172-142:789256:789695 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-173-121:3987801:3988227 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-173-121:3987800:3988226 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-173-121:3987800:3988226 [6] NCCL INFO Channel 00/32 :    0
+ip-26-0-173-121:3987800:3988226 [6] NCCL INFO Channel 01/32 :    0
+ip-26-0-173-121:3987800:3988226 [6] NCCL INFO Channel 02/32 :    0
+ip-26-0-173-121:3987800:3988226 [6] NCCL INFO Channel 03/32 :    0
+ip-26-0-173-121:3987800:3988226 [6] NCCL INFO Channel 04/32 :    0
+ip-26-0-173-121:3987800:3988226 [6] NCCL INFO Channel 05/32 :    0
+ip-26-0-173-121:3987800:3988226 [6] NCCL INFO Channel 06/32 :    0
+ip-26-0-173-121:3987800:3988226 [6] NCCL INFO Channel 07/32 :    0
+ip-26-0-173-121:3987800:3988226 [6] NCCL INFO Channel 08/32 :    0
+ip-26-0-173-121:3987800:3988226 [6] NCCL INFO Channel 09/32 :    0
+ip-26-0-173-121:3987800:3988226 [6] NCCL INFO Channel 10/32 :    0
+ip-26-0-173-121:3987800:3988226 [6] NCCL INFO Channel 11/32 :    0
+ip-26-0-173-121:3987800:3988226 [6] NCCL INFO Channel 12/32 :    0
+ip-26-0-173-121:3987800:3988226 [6] NCCL INFO Channel 13/32 :    0
+ip-26-0-173-121:3987800:3988226 [6] NCCL INFO Channel 14/32 :    0
+ip-26-0-173-121:3987800:3988226 [6] NCCL INFO Channel 15/32 :    0
+ip-26-0-173-121:3987800:3988226 [6] NCCL INFO Channel 16/32 :    0
+ip-26-0-173-121:3987800:3988226 [6] NCCL INFO Channel 17/32 :    0
+ip-26-0-173-121:3987800:3988226 [6] NCCL INFO Channel 18/32 :    0
+ip-26-0-173-121:3987800:3988226 [6] NCCL INFO Channel 19/32 :    0
+ip-26-0-173-121:3987800:3988226 [6] NCCL INFO Channel 20/32 :    0
+ip-26-0-173-121:3987800:3988226 [6] NCCL INFO Channel 21/32 :    0
+ip-26-0-173-121:3987800:3988226 [6] NCCL INFO Channel 22/32 :    0
+ip-26-0-173-121:3987800:3988226 [6] NCCL INFO Channel 23/32 :    0
+ip-26-0-173-121:3987800:3988226 [6] NCCL INFO Channel 24/32 :    0
+ip-26-0-173-121:3987800:3988226 [6] NCCL INFO Channel 25/32 :    0
+ip-26-0-173-121:3987800:3988226 [6] NCCL INFO Channel 26/32 :    0
+ip-26-0-173-121:3987800:3988226 [6] NCCL INFO Channel 27/32 :    0
+ip-26-0-173-121:3987800:3988226 [6] NCCL INFO Channel 28/32 :    0
+ip-26-0-173-121:3987800:3988226 [6] NCCL INFO Channel 29/32 :    0
+ip-26-0-173-121:3987800:3988226 [6] NCCL INFO Channel 30/32 :    0
+ip-26-0-173-121:3987800:3988226 [6] NCCL INFO Channel 31/32 :    0
+ip-26-0-173-121:3987800:3988226 [6] NCCL INFO Trees [0] -1/-1/-1->0->-1 [1] -1/-1/-1->0->-1 [2] -1/-1/-1->0->-1 [3] -1/-1/-1->0->-1 [4] -1/-1/-1->0->-1 [5] -1/-1/-1->0->-1 [6] -1/-1/-1->0->-1 [7] -1/-1/-1->0->-1 [8] -1/-1/-1->0->-1 [9] -1/-1/-1->0->-1 [10] -1/-1/-1->0->-1 [11] -1/-1/-1->0->-1 [12] -1/-1/-1->0->-1 [13] -1/-1/-1->0->-1 [14] -1/-1/-1->0->-1 [15] -1/-1/-1->0->-1 [16] -1/-1/-1->0->-1 [17] -1/-1/-1->0->-1 [18] -1/-1/-1->0->-1 [19] -1/-1/-1->0->-1 [20] -1/-1/-1->0->-1 [21] -1/-1/-1->0->-1 [22] -1/-1/-1->0->-1 [23] -1/-1/-1->0->-1 [24] -1/-1/-1->0->-1 [25] -1/-1/-1->0->-1 [26] -1/-1/-1->0->-1 [27] -1/-1/-1->0->-1 [28] -1/-1/-1->0->-1 [29] -1/-1/-1->0->-1 [30] -1/-1/-1->0->-1 [31] -1/-1/-1->0->-1
+ip-26-0-173-121:3987800:3988226 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-173-121:3987800:3988226 [6] NCCL INFO Connected all rings
+ip-26-0-173-121:3987800:3988226 [6] NCCL INFO Connected all trees
+ip-26-0-173-121:3987800:3988226 [6] NCCL INFO 32 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-142:789256:789695 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-172-142:789256:789695 [1] NCCL INFO Channel 00/32 :    0
+ip-26-0-172-142:789256:789695 [1] NCCL INFO Channel 01/32 :    0
+ip-26-0-172-142:789256:789695 [1] NCCL INFO Channel 02/32 :    0
+ip-26-0-172-142:789256:789695 [1] NCCL INFO Channel 03/32 :    0
+ip-26-0-172-142:789256:789695 [1] NCCL INFO Channel 04/32 :    0
+ip-26-0-172-142:789256:789695 [1] NCCL INFO Channel 05/32 :    0
+ip-26-0-172-142:789256:789695 [1] NCCL INFO Channel 06/32 :    0
+ip-26-0-172-142:789256:789695 [1] NCCL INFO Channel 07/32 :    0
+ip-26-0-172-142:789256:789695 [1] NCCL INFO Channel 08/32 :    0
+ip-26-0-172-142:789256:789695 [1] NCCL INFO Channel 09/32 :    0
+ip-26-0-172-142:789256:789695 [1] NCCL INFO Channel 10/32 :    0
+ip-26-0-172-142:789256:789695 [1] NCCL INFO Channel 11/32 :    0
+ip-26-0-172-142:789256:789695 [1] NCCL INFO Channel 12/32 :    0
+ip-26-0-172-142:789256:789695 [1] NCCL INFO Channel 13/32 :    0
+ip-26-0-172-142:789256:789695 [1] NCCL INFO Channel 14/32 :    0
+ip-26-0-172-142:789256:789695 [1] NCCL INFO Channel 15/32 :    0
+ip-26-0-172-142:789256:789695 [1] NCCL INFO Channel 16/32 :    0
+ip-26-0-172-142:789256:789695 [1] NCCL INFO Channel 17/32 :    0
+ip-26-0-172-142:789256:789695 [1] NCCL INFO Channel 18/32 :    0
+ip-26-0-172-142:789256:789695 [1] NCCL INFO Channel 19/32 :    0
+ip-26-0-172-142:789256:789695 [1] NCCL INFO Channel 20/32 :    0
+ip-26-0-172-142:789256:789695 [1] NCCL INFO Channel 21/32 :    0
+ip-26-0-172-142:789256:789695 [1] NCCL INFO Channel 22/32 :    0
+ip-26-0-172-142:789256:789695 [1] NCCL INFO Channel 23/32 :    0
+ip-26-0-172-142:789256:789695 [1] NCCL INFO Channel 24/32 :    0
+ip-26-0-172-142:789256:789695 [1] NCCL INFO Channel 25/32 :    0
+ip-26-0-172-142:789256:789695 [1] NCCL INFO Channel 26/32 :    0
+ip-26-0-172-142:789256:789695 [1] NCCL INFO Channel 27/32 :    0
+ip-26-0-172-142:789256:789695 [1] NCCL INFO Channel 28/32 :    0
+ip-26-0-172-142:789256:789695 [1] NCCL INFO Channel 29/32 :    0
+ip-26-0-172-142:789256:789695 [1] NCCL INFO Channel 30/32 :    0
+ip-26-0-172-142:789256:789695 [1] NCCL INFO Channel 31/32 :    0
+ip-26-0-172-142:789256:789695 [1] NCCL INFO Trees [0] -1/-1/-1->0->-1 [1] -1/-1/-1->0->-1 [2] -1/-1/-1->0->-1 [3] -1/-1/-1->0->-1 [4] -1/-1/-1->0->-1 [5] -1/-1/-1->0->-1 [6] -1/-1/-1->0->-1 [7] -1/-1/-1->0->-1 [8] -1/-1/-1->0->-1 [9] -1/-1/-1->0->-1 [10] -1/-1/-1->0->-1 [11] -1/-1/-1->0->-1 [12] -1/-1/-1->0->-1 [13] -1/-1/-1->0->-1 [14] -1/-1/-1->0->-1 [15] -1/-1/-1->0->-1 [16] -1/-1/-1->0->-1 [17] -1/-1/-1->0->-1 [18] -1/-1/-1->0->-1 [19] -1/-1/-1->0->-1 [20] -1/-1/-1->0->-1 [21] -1/-1/-1->0->-1 [22] -1/-1/-1->0->-1 [23] -1/-1/-1->0->-1 [24] -1/-1/-1->0->-1 [25] -1/-1/-1->0->-1 [26] -1/-1/-1->0->-1 [27] -1/-1/-1->0->-1 [28] -1/-1/-1->0->-1 [29] -1/-1/-1->0->-1 [30] -1/-1/-1->0->-1 [31] -1/-1/-1->0->-1
+ip-26-0-172-142:789256:789695 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-142:789256:789695 [1] NCCL INFO Connected all rings
+ip-26-0-172-142:789256:789695 [1] NCCL INFO Connected all trees
+ip-26-0-172-142:789256:789695 [1] NCCL INFO 32 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-173-121:3987800:3988226 [6] NCCL INFO comm 0xa72f590 rank 0 nranks 1 cudaDev 6 nvmlDev 6 busId b9000 commId 0xb636512b59c0f485 - Init COMPLETE
+ip-26-0-173-121:3987798:3988213 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-173-121:3987798:3988213 [4] NCCL INFO Channel 00/32 :    0
+ip-26-0-173-121:3987798:3988213 [4] NCCL INFO Channel 01/32 :    0
+ip-26-0-173-121:3987798:3988213 [4] NCCL INFO Channel 02/32 :    0
+ip-26-0-173-121:3987798:3988213 [4] NCCL INFO Channel 03/32 :    0
+ip-26-0-173-121:3987798:3988213 [4] NCCL INFO Channel 04/32 :    0
+ip-26-0-173-121:3987798:3988213 [4] NCCL INFO Channel 05/32 :    0
+ip-26-0-173-121:3987798:3988213 [4] NCCL INFO Channel 06/32 :    0
+ip-26-0-173-121:3987798:3988213 [4] NCCL INFO Channel 07/32 :    0
+ip-26-0-173-121:3987798:3988213 [4] NCCL INFO Channel 08/32 :    0
+ip-26-0-173-121:3987798:3988213 [4] NCCL INFO Channel 09/32 :    0
+ip-26-0-173-121:3987798:3988213 [4] NCCL INFO Channel 10/32 :    0
+ip-26-0-173-121:3987798:3988213 [4] NCCL INFO Channel 11/32 :    0
+ip-26-0-173-121:3987798:3988213 [4] NCCL INFO Channel 12/32 :    0
+ip-26-0-173-121:3987798:3988213 [4] NCCL INFO Channel 13/32 :    0
+ip-26-0-173-121:3987798:3988213 [4] NCCL INFO Channel 14/32 :    0
+ip-26-0-173-121:3987798:3988213 [4] NCCL INFO Channel 15/32 :    0
+ip-26-0-173-121:3987798:3988213 [4] NCCL INFO Channel 16/32 :    0
+ip-26-0-173-121:3987798:3988213 [4] NCCL INFO Channel 17/32 :    0
+ip-26-0-173-121:3987798:3988213 [4] NCCL INFO Channel 18/32 :    0
+ip-26-0-173-121:3987798:3988213 [4] NCCL INFO Channel 19/32 :    0
+ip-26-0-173-121:3987798:3988213 [4] NCCL INFO Channel 20/32 :    0
+ip-26-0-173-121:3987798:3988213 [4] NCCL INFO Channel 21/32 :    0
+ip-26-0-173-121:3987798:3988213 [4] NCCL INFO Channel 22/32 :    0
+ip-26-0-173-121:3987798:3988213 [4] NCCL INFO Channel 23/32 :    0
+ip-26-0-173-121:3987798:3988213 [4] NCCL INFO Channel 24/32 :    0
+ip-26-0-173-121:3987798:3988213 [4] NCCL INFO Channel 25/32 :    0
+ip-26-0-173-121:3987798:3988213 [4] NCCL INFO Channel 26/32 :    0
+ip-26-0-173-121:3987798:3988213 [4] NCCL INFO Channel 27/32 :    0
+ip-26-0-173-121:3987798:3988213 [4] NCCL INFO Channel 28/32 :    0
+ip-26-0-173-121:3987798:3988213 [4] NCCL INFO Channel 29/32 :    0
+ip-26-0-173-121:3987798:3988213 [4] NCCL INFO Channel 30/32 :    0
+ip-26-0-173-121:3987798:3988213 [4] NCCL INFO Channel 31/32 :    0
+ip-26-0-173-121:3987798:3988213 [4] NCCL INFO Trees [0] -1/-1/-1->0->-1 [1] -1/-1/-1->0->-1 [2] -1/-1/-1->0->-1 [3] -1/-1/-1->0->-1 [4] -1/-1/-1->0->-1 [5] -1/-1/-1->0->-1 [6] -1/-1/-1->0->-1 [7] -1/-1/-1->0->-1 [8] -1/-1/-1->0->-1 [9] -1/-1/-1->0->-1 [10] -1/-1/-1->0->-1 [11] -1/-1/-1->0->-1 [12] -1/-1/-1->0->-1 [13] -1/-1/-1->0->-1 [14] -1/-1/-1->0->-1 [15] -1/-1/-1->0->-1 [16] -1/-1/-1->0->-1 [17] -1/-1/-1->0->-1 [18] -1/-1/-1->0->-1 [19] -1/-1/-1->0->-1 [20] -1/-1/-1->0->-1 [21] -1/-1/-1->0->-1 [22] -1/-1/-1->0->-1 [23] -1/-1/-1->0->-1 [24] -1/-1/-1->0->-1 [25] -1/-1/-1->0->-1 [26] -1/-1/-1->0->-1 [27] -1/-1/-1->0->-1 [28] -1/-1/-1->0->-1 [29] -1/-1/-1->0->-1 [30] -1/-1/-1->0->-1 [31] -1/-1/-1->0->-1
+ip-26-0-173-121:3987798:3988213 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-173-121:3987798:3988213 [4] NCCL INFO Connected all rings
+ip-26-0-173-121:3987798:3988213 [4] NCCL INFO Connected all trees
+ip-26-0-173-121:3987798:3988213 [4] NCCL INFO 32 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-142:789257:789685 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-172-142:789257:789685 [2] NCCL INFO Channel 00/32 :    0
+ip-26-0-172-142:789257:789685 [2] NCCL INFO Channel 01/32 :    0
+ip-26-0-172-142:789257:789685 [2] NCCL INFO Channel 02/32 :    0
+ip-26-0-172-142:789257:789685 [2] NCCL INFO Channel 03/32 :    0
+ip-26-0-172-142:789257:789685 [2] NCCL INFO Channel 04/32 :    0
+ip-26-0-172-142:789257:789685 [2] NCCL INFO Channel 05/32 :    0
+ip-26-0-172-142:789257:789685 [2] NCCL INFO Channel 06/32 :    0
+ip-26-0-172-142:789257:789685 [2] NCCL INFO Channel 07/32 :    0
+ip-26-0-172-142:789257:789685 [2] NCCL INFO Channel 08/32 :    0
+ip-26-0-172-142:789257:789685 [2] NCCL INFO Channel 09/32 :    0
+ip-26-0-172-142:789257:789685 [2] NCCL INFO Channel 10/32 :    0
+ip-26-0-172-142:789257:789685 [2] NCCL INFO Channel 11/32 :    0
+ip-26-0-172-142:789257:789685 [2] NCCL INFO Channel 12/32 :    0
+ip-26-0-172-142:789257:789685 [2] NCCL INFO Channel 13/32 :    0
+ip-26-0-172-142:789257:789685 [2] NCCL INFO Channel 14/32 :    0
+ip-26-0-172-142:789257:789685 [2] NCCL INFO Channel 15/32 :    0
+ip-26-0-172-142:789257:789685 [2] NCCL INFO Channel 16/32 :    0
+ip-26-0-172-142:789257:789685 [2] NCCL INFO Channel 17/32 :    0
+ip-26-0-172-142:789257:789685 [2] NCCL INFO Channel 18/32 :    0
+ip-26-0-172-142:789257:789685 [2] NCCL INFO Channel 19/32 :    0
+ip-26-0-172-142:789257:789685 [2] NCCL INFO Channel 20/32 :    0
+ip-26-0-172-142:789257:789685 [2] NCCL INFO Channel 21/32 :    0
+ip-26-0-172-142:789257:789685 [2] NCCL INFO Channel 22/32 :    0
+ip-26-0-172-142:789257:789685 [2] NCCL INFO Channel 23/32 :    0
+ip-26-0-172-142:789257:789685 [2] NCCL INFO Channel 24/32 :    0
+ip-26-0-172-142:789257:789685 [2] NCCL INFO Channel 25/32 :    0
+ip-26-0-172-142:789257:789685 [2] NCCL INFO Channel 26/32 :    0
+ip-26-0-172-142:789257:789685 [2] NCCL INFO Channel 27/32 :    0
+ip-26-0-172-142:789257:789685 [2] NCCL INFO Channel 28/32 :    0
+ip-26-0-172-142:789257:789685 [2] NCCL INFO Channel 29/32 :    0
+ip-26-0-172-142:789257:789685 [2] NCCL INFO Channel 30/32 :    0
+ip-26-0-172-142:789257:789685 [2] NCCL INFO Channel 31/32 :    0
+ip-26-0-172-142:789257:789685 [2] NCCL INFO Trees [0] -1/-1/-1->0->-1 [1] -1/-1/-1->0->-1 [2] -1/-1/-1->0->-1 [3] -1/-1/-1->0->-1 [4] -1/-1/-1->0->-1 [5] -1/-1/-1->0->-1 [6] -1/-1/-1->0->-1 [7] -1/-1/-1->0->-1 [8] -1/-1/-1->0->-1 [9] -1/-1/-1->0->-1 [10] -1/-1/-1->0->-1 [11] -1/-1/-1->0->-1 [12] -1/-1/-1->0->-1 [13] -1/-1/-1->0->-1 [14] -1/-1/-1->0->-1 [15] -1/-1/-1->0->-1 [16] -1/-1/-1->0->-1 [17] -1/-1/-1->0->-1 [18] -1/-1/-1->0->-1 [19] -1/-1/-1->0->-1 [20] -1/-1/-1->0->-1 [21] -1/-1/-1->0->-1 [22] -1/-1/-1->0->-1 [23] -1/-1/-1->0->-1 [24] -1/-1/-1->0->-1 [25] -1/-1/-1->0->-1 [26] -1/-1/-1->0->-1 [27] -1/-1/-1->0->-1 [28] -1/-1/-1->0->-1 [29] -1/-1/-1->0->-1 [30] -1/-1/-1->0->-1 [31] -1/-1/-1->0->-1
+ip-26-0-172-142:789257:789685 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-142:789257:789685 [2] NCCL INFO Connected all rings
+ip-26-0-172-142:789257:789685 [2] NCCL INFO Connected all trees
+ip-26-0-172-142:789257:789685 [2] NCCL INFO 32 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-142:789261:789678 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-172-142:789261:789678 [6] NCCL INFO Channel 00/32 :    0
+ip-26-0-172-142:789261:789678 [6] NCCL INFO Channel 01/32 :    0
+ip-26-0-172-142:789261:789678 [6] NCCL INFO Channel 02/32 :    0
+ip-26-0-172-142:789261:789678 [6] NCCL INFO Channel 03/32 :    0
+ip-26-0-172-142:789261:789678 [6] NCCL INFO Channel 04/32 :    0
+ip-26-0-172-142:789261:789678 [6] NCCL INFO Channel 05/32 :    0
+ip-26-0-172-142:789261:789678 [6] NCCL INFO Channel 06/32 :    0
+ip-26-0-172-142:789261:789678 [6] NCCL INFO Channel 07/32 :    0
+ip-26-0-172-142:789261:789678 [6] NCCL INFO Channel 08/32 :    0
+ip-26-0-172-142:789261:789678 [6] NCCL INFO Channel 09/32 :    0
+ip-26-0-172-142:789261:789678 [6] NCCL INFO Channel 10/32 :    0
+ip-26-0-172-142:789261:789678 [6] NCCL INFO Channel 11/32 :    0
+ip-26-0-172-142:789261:789678 [6] NCCL INFO Channel 12/32 :    0
+ip-26-0-172-142:789261:789678 [6] NCCL INFO Channel 13/32 :    0
+ip-26-0-172-142:789261:789678 [6] NCCL INFO Channel 14/32 :    0
+ip-26-0-172-142:789261:789678 [6] NCCL INFO Channel 15/32 :    0
+ip-26-0-172-142:789261:789678 [6] NCCL INFO Channel 16/32 :    0
+ip-26-0-172-142:789261:789678 [6] NCCL INFO Channel 17/32 :    0
+ip-26-0-172-142:789261:789678 [6] NCCL INFO Channel 18/32 :    0
+ip-26-0-172-142:789261:789678 [6] NCCL INFO Channel 19/32 :    0
+ip-26-0-172-142:789261:789678 [6] NCCL INFO Channel 20/32 :    0
+ip-26-0-172-142:789261:789678 [6] NCCL INFO Channel 21/32 :    0
+ip-26-0-172-142:789261:789678 [6] NCCL INFO Channel 22/32 :    0
+ip-26-0-172-142:789261:789678 [6] NCCL INFO Channel 23/32 :    0
+ip-26-0-172-142:789261:789678 [6] NCCL INFO Channel 24/32 :    0
+ip-26-0-172-142:789261:789678 [6] NCCL INFO Channel 25/32 :    0
+ip-26-0-172-142:789261:789678 [6] NCCL INFO Channel 26/32 :    0
+ip-26-0-172-142:789261:789678 [6] NCCL INFO Channel 27/32 :    0
+ip-26-0-172-142:789261:789678 [6] NCCL INFO Channel 28/32 :    0
+ip-26-0-172-142:789261:789678 [6] NCCL INFO Channel 29/32 :    0
+ip-26-0-172-142:789261:789678 [6] NCCL INFO Channel 30/32 :    0
+ip-26-0-172-142:789261:789678 [6] NCCL INFO Channel 31/32 :    0
+ip-26-0-172-142:789261:789678 [6] NCCL INFO Trees [0] -1/-1/-1->0->-1 [1] -1/-1/-1->0->-1 [2] -1/-1/-1->0->-1 [3] -1/-1/-1->0->-1 [4] -1/-1/-1->0->-1 [5] -1/-1/-1->0->-1 [6] -1/-1/-1->0->-1 [7] -1/-1/-1->0->-1 [8] -1/-1/-1->0->-1 [9] -1/-1/-1->0->-1 [10] -1/-1/-1->0->-1 [11] -1/-1/-1->0->-1 [12] -1/-1/-1->0->-1 [13] -1/-1/-1->0->-1 [14] -1/-1/-1->0->-1 [15] -1/-1/-1->0->-1 [16] -1/-1/-1->0->-1 [17] -1/-1/-1->0->-1 [18] -1/-1/-1->0->-1 [19] -1/-1/-1->0->-1 [20] -1/-1/-1->0->-1 [21] -1/-1/-1->0->-1 [22] -1/-1/-1->0->-1 [23] -1/-1/-1->0->-1 [24] -1/-1/-1->0->-1 [25] -1/-1/-1->0->-1 [26] -1/-1/-1->0->-1 [27] -1/-1/-1->0->-1 [28] -1/-1/-1->0->-1 [29] -1/-1/-1->0->-1 [30] -1/-1/-1->0->-1 [31] -1/-1/-1->0->-1
+ip-26-0-172-142:789261:789678 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-142:789261:789678 [6] NCCL INFO Connected all rings
+ip-26-0-172-142:789261:789678 [6] NCCL INFO Connected all trees
+ip-26-0-172-142:789261:789678 [6] NCCL INFO 32 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-142:789255:789693 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-172-142:789255:789693 [0] NCCL INFO Channel 00/32 :    0
+ip-26-0-172-142:789255:789693 [0] NCCL INFO Channel 01/32 :    0
+ip-26-0-172-142:789255:789693 [0] NCCL INFO Channel 02/32 :    0
+ip-26-0-172-142:789255:789693 [0] NCCL INFO Channel 03/32 :    0
+ip-26-0-172-142:789255:789693 [0] NCCL INFO Channel 04/32 :    0
+ip-26-0-172-142:789255:789693 [0] NCCL INFO Channel 05/32 :    0
+ip-26-0-172-142:789255:789693 [0] NCCL INFO Channel 06/32 :    0
+ip-26-0-172-142:789255:789693 [0] NCCL INFO Channel 07/32 :    0
+ip-26-0-172-142:789255:789693 [0] NCCL INFO Channel 08/32 :    0
+ip-26-0-172-142:789255:789693 [0] NCCL INFO Channel 09/32 :    0
+ip-26-0-172-142:789255:789693 [0] NCCL INFO Channel 10/32 :    0
+ip-26-0-172-142:789255:789693 [0] NCCL INFO Channel 11/32 :    0
+ip-26-0-172-142:789255:789693 [0] NCCL INFO Channel 12/32 :    0
+ip-26-0-172-142:789255:789693 [0] NCCL INFO Channel 13/32 :    0
+ip-26-0-172-142:789255:789693 [0] NCCL INFO Channel 14/32 :    0
+ip-26-0-172-142:789255:789693 [0] NCCL INFO Channel 15/32 :    0
+ip-26-0-172-142:789255:789693 [0] NCCL INFO Channel 16/32 :    0
+ip-26-0-172-142:789255:789693 [0] NCCL INFO Channel 17/32 :    0
+ip-26-0-172-142:789255:789693 [0] NCCL INFO Channel 18/32 :    0
+ip-26-0-172-142:789255:789693 [0] NCCL INFO Channel 19/32 :    0
+ip-26-0-172-142:789255:789693 [0] NCCL INFO Channel 20/32 :    0
+ip-26-0-172-142:789255:789693 [0] NCCL INFO Channel 21/32 :    0
+ip-26-0-172-142:789255:789693 [0] NCCL INFO Channel 22/32 :    0
+ip-26-0-172-142:789255:789693 [0] NCCL INFO Channel 23/32 :    0
+ip-26-0-172-142:789255:789693 [0] NCCL INFO Channel 24/32 :    0
+ip-26-0-172-142:789255:789693 [0] NCCL INFO Channel 25/32 :    0
+ip-26-0-172-142:789255:789693 [0] NCCL INFO Channel 26/32 :    0
+ip-26-0-172-142:789255:789693 [0] NCCL INFO Channel 27/32 :    0
+ip-26-0-172-142:789255:789693 [0] NCCL INFO Channel 28/32 :    0
+ip-26-0-172-142:789255:789693 [0] NCCL INFO Channel 29/32 :    0
+ip-26-0-172-142:789255:789693 [0] NCCL INFO Channel 30/32 :    0
+ip-26-0-172-142:789255:789693 [0] NCCL INFO Channel 31/32 :    0
+ip-26-0-172-142:789255:789693 [0] NCCL INFO Trees [0] -1/-1/-1->0->-1 [1] -1/-1/-1->0->-1 [2] -1/-1/-1->0->-1 [3] -1/-1/-1->0->-1 [4] -1/-1/-1->0->-1 [5] -1/-1/-1->0->-1 [6] -1/-1/-1->0->-1 [7] -1/-1/-1->0->-1 [8] -1/-1/-1->0->-1 [9] -1/-1/-1->0->-1 [10] -1/-1/-1->0->-1 [11] -1/-1/-1->0->-1 [12] -1/-1/-1->0->-1 [13] -1/-1/-1->0->-1 [14] -1/-1/-1->0->-1 [15] -1/-1/-1->0->-1 [16] -1/-1/-1->0->-1 [17] -1/-1/-1->0->-1 [18] -1/-1/-1->0->-1 [19] -1/-1/-1->0->-1 [20] -1/-1/-1->0->-1 [21] -1/-1/-1->0->-1 [22] -1/-1/-1->0->-1 [23] -1/-1/-1->0->-1 [24] -1/-1/-1->0->-1 [25] -1/-1/-1->0->-1 [26] -1/-1/-1->0->-1 [27] -1/-1/-1->0->-1 [28] -1/-1/-1->0->-1 [29] -1/-1/-1->0->-1 [30] -1/-1/-1->0->-1 [31] -1/-1/-1->0->-1
+ip-26-0-172-142:789255:789693 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-142:789255:789693 [0] NCCL INFO Connected all rings
+ip-26-0-172-142:789255:789693 [0] NCCL INFO Connected all trees
+ip-26-0-172-142:789255:789693 [0] NCCL INFO 32 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-142:789256:789695 [1] NCCL INFO comm 0xc42ddb0 rank 0 nranks 1 cudaDev 1 nvmlDev 1 busId 64000 commId 0x2486e57acb78929c - Init COMPLETE
+ip-26-0-173-121:3987796:3988210 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-173-121:3987796:3988210 [2] NCCL INFO Channel 00/32 :    0
+ip-26-0-173-121:3987796:3988210 [2] NCCL INFO Channel 01/32 :    0
+ip-26-0-173-121:3987796:3988210 [2] NCCL INFO Channel 02/32 :    0
+ip-26-0-173-121:3987796:3988210 [2] NCCL INFO Channel 03/32 :    0
+ip-26-0-173-121:3987796:3988210 [2] NCCL INFO Channel 04/32 :    0
+ip-26-0-173-121:3987796:3988210 [2] NCCL INFO Channel 05/32 :    0
+ip-26-0-173-121:3987796:3988210 [2] NCCL INFO Channel 06/32 :    0
+ip-26-0-173-121:3987796:3988210 [2] NCCL INFO Channel 07/32 :    0
+ip-26-0-173-121:3987796:3988210 [2] NCCL INFO Channel 08/32 :    0
+ip-26-0-173-121:3987796:3988210 [2] NCCL INFO Channel 09/32 :    0
+ip-26-0-173-121:3987796:3988210 [2] NCCL INFO Channel 10/32 :    0
+ip-26-0-173-121:3987796:3988210 [2] NCCL INFO Channel 11/32 :    0
+ip-26-0-173-121:3987796:3988210 [2] NCCL INFO Channel 12/32 :    0
+ip-26-0-173-121:3987796:3988210 [2] NCCL INFO Channel 13/32 :    0
+ip-26-0-173-121:3987796:3988210 [2] NCCL INFO Channel 14/32 :    0
+ip-26-0-173-121:3987796:3988210 [2] NCCL INFO Channel 15/32 :    0
+ip-26-0-173-121:3987796:3988210 [2] NCCL INFO Channel 16/32 :    0
+ip-26-0-173-121:3987796:3988210 [2] NCCL INFO Channel 17/32 :    0
+ip-26-0-173-121:3987796:3988210 [2] NCCL INFO Channel 18/32 :    0
+ip-26-0-173-121:3987796:3988210 [2] NCCL INFO Channel 19/32 :    0
+ip-26-0-173-121:3987796:3988210 [2] NCCL INFO Channel 20/32 :    0
+ip-26-0-173-121:3987796:3988210 [2] NCCL INFO Channel 21/32 :    0
+ip-26-0-173-121:3987796:3988210 [2] NCCL INFO Channel 22/32 :    0
+ip-26-0-173-121:3987796:3988210 [2] NCCL INFO Channel 23/32 :    0
+ip-26-0-173-121:3987796:3988210 [2] NCCL INFO Channel 24/32 :    0
+ip-26-0-173-121:3987796:3988210 [2] NCCL INFO Channel 25/32 :    0
+ip-26-0-173-121:3987796:3988210 [2] NCCL INFO Channel 26/32 :    0
+ip-26-0-173-121:3987796:3988210 [2] NCCL INFO Channel 27/32 :    0
+ip-26-0-173-121:3987796:3988210 [2] NCCL INFO Channel 28/32 :    0
+ip-26-0-173-121:3987796:3988210 [2] NCCL INFO Channel 29/32 :    0
+ip-26-0-173-121:3987796:3988210 [2] NCCL INFO Channel 30/32 :    0
+ip-26-0-173-121:3987796:3988210 [2] NCCL INFO Channel 31/32 :    0
+ip-26-0-173-121:3987796:3988210 [2] NCCL INFO Trees [0] -1/-1/-1->0->-1 [1] -1/-1/-1->0->-1 [2] -1/-1/-1->0->-1 [3] -1/-1/-1->0->-1 [4] -1/-1/-1->0->-1 [5] -1/-1/-1->0->-1 [6] -1/-1/-1->0->-1 [7] -1/-1/-1->0->-1 [8] -1/-1/-1->0->-1 [9] -1/-1/-1->0->-1 [10] -1/-1/-1->0->-1 [11] -1/-1/-1->0->-1 [12] -1/-1/-1->0->-1 [13] -1/-1/-1->0->-1 [14] -1/-1/-1->0->-1 [15] -1/-1/-1->0->-1 [16] -1/-1/-1->0->-1 [17] -1/-1/-1->0->-1 [18] -1/-1/-1->0->-1 [19] -1/-1/-1->0->-1 [20] -1/-1/-1->0->-1 [21] -1/-1/-1->0->-1 [22] -1/-1/-1->0->-1 [23] -1/-1/-1->0->-1 [24] -1/-1/-1->0->-1 [25] -1/-1/-1->0->-1 [26] -1/-1/-1->0->-1 [27] -1/-1/-1->0->-1 [28] -1/-1/-1->0->-1 [29] -1/-1/-1->0->-1 [30] -1/-1/-1->0->-1 [31] -1/-1/-1->0->-1
+ip-26-0-173-121:3987796:3988210 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-173-121:3987796:3988210 [2] NCCL INFO Connected all rings
+ip-26-0-173-121:3987796:3988210 [2] NCCL INFO Connected all trees
+ip-26-0-173-121:3987796:3988210 [2] NCCL INFO 32 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-173-121:3987798:3988213 [4] NCCL INFO comm 0xbb0a880 rank 0 nranks 1 cudaDev 4 nvmlDev 4 busId 97000 commId 0x953f1b590b4d8b44 - Init COMPLETE
+ip-26-0-172-142:789262:789680 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-172-142:789262:789680 [7] NCCL INFO Channel 00/32 :    0
+ip-26-0-172-142:789262:789680 [7] NCCL INFO Channel 01/32 :    0
+ip-26-0-172-142:789262:789680 [7] NCCL INFO Channel 02/32 :    0
+ip-26-0-172-142:789262:789680 [7] NCCL INFO Channel 03/32 :    0
+ip-26-0-172-142:789262:789680 [7] NCCL INFO Channel 04/32 :    0
+ip-26-0-172-142:789262:789680 [7] NCCL INFO Channel 05/32 :    0
+ip-26-0-172-142:789262:789680 [7] NCCL INFO Channel 06/32 :    0
+ip-26-0-172-142:789262:789680 [7] NCCL INFO Channel 07/32 :    0
+ip-26-0-172-142:789262:789680 [7] NCCL INFO Channel 08/32 :    0
+ip-26-0-172-142:789262:789680 [7] NCCL INFO Channel 09/32 :    0
+ip-26-0-172-142:789262:789680 [7] NCCL INFO Channel 10/32 :    0
+ip-26-0-172-142:789262:789680 [7] NCCL INFO Channel 11/32 :    0
+ip-26-0-172-142:789262:789680 [7] NCCL INFO Channel 12/32 :    0
+ip-26-0-172-142:789262:789680 [7] NCCL INFO Channel 13/32 :    0
+ip-26-0-172-142:789262:789680 [7] NCCL INFO Channel 14/32 :    0
+ip-26-0-172-142:789262:789680 [7] NCCL INFO Channel 15/32 :    0
+ip-26-0-172-142:789262:789680 [7] NCCL INFO Channel 16/32 :    0
+ip-26-0-172-142:789262:789680 [7] NCCL INFO Channel 17/32 :    0
+ip-26-0-172-142:789262:789680 [7] NCCL INFO Channel 18/32 :    0
+ip-26-0-172-142:789262:789680 [7] NCCL INFO Channel 19/32 :    0
+ip-26-0-172-142:789262:789680 [7] NCCL INFO Channel 20/32 :    0
+ip-26-0-172-142:789262:789680 [7] NCCL INFO Channel 21/32 :    0
+ip-26-0-172-142:789262:789680 [7] NCCL INFO Channel 22/32 :    0
+ip-26-0-172-142:789262:789680 [7] NCCL INFO Channel 23/32 :    0
+ip-26-0-172-142:789262:789680 [7] NCCL INFO Channel 24/32 :    0
+ip-26-0-172-142:789262:789680 [7] NCCL INFO Channel 25/32 :    0
+ip-26-0-172-142:789262:789680 [7] NCCL INFO Channel 26/32 :    0
+ip-26-0-172-142:789262:789680 [7] NCCL INFO Channel 27/32 :    0
+ip-26-0-172-142:789262:789680 [7] NCCL INFO Channel 28/32 :    0
+ip-26-0-172-142:789262:789680 [7] NCCL INFO Channel 29/32 :    0
+ip-26-0-172-142:789262:789680 [7] NCCL INFO Channel 30/32 :    0
+ip-26-0-172-142:789262:789680 [7] NCCL INFO Channel 31/32 :    0
+ip-26-0-172-142:789262:789680 [7] NCCL INFO Trees [0] -1/-1/-1->0->-1 [1] -1/-1/-1->0->-1 [2] -1/-1/-1->0->-1 [3] -1/-1/-1->0->-1 [4] -1/-1/-1->0->-1 [5] -1/-1/-1->0->-1 [6] -1/-1/-1->0->-1 [7] -1/-1/-1->0->-1 [8] -1/-1/-1->0->-1 [9] -1/-1/-1->0->-1 [10] -1/-1/-1->0->-1 [11] -1/-1/-1->0->-1 [12] -1/-1/-1->0->-1 [13] -1/-1/-1->0->-1 [14] -1/-1/-1->0->-1 [15] -1/-1/-1->0->-1 [16] -1/-1/-1->0->-1 [17] -1/-1/-1->0->-1 [18] -1/-1/-1->0->-1 [19] -1/-1/-1->0->-1 [20] -1/-1/-1->0->-1 [21] -1/-1/-1->0->-1 [22] -1/-1/-1->0->-1 [23] -1/-1/-1->0->-1 [24] -1/-1/-1->0->-1 [25] -1/-1/-1->0->-1 [26] -1/-1/-1->0->-1 [27] -1/-1/-1->0->-1 [28] -1/-1/-1->0->-1 [29] -1/-1/-1->0->-1 [30] -1/-1/-1->0->-1 [31] -1/-1/-1->0->-1
+ip-26-0-172-142:789262:789680 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-142:789262:789680 [7] NCCL INFO Connected all rings
+ip-26-0-172-142:789262:789680 [7] NCCL INFO Connected all trees
+ip-26-0-172-142:789262:789680 [7] NCCL INFO 32 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-142:789261:789678 [6] NCCL INFO comm 0xbd1d5f0 rank 0 nranks 1 cudaDev 6 nvmlDev 6 busId b9000 commId 0x65112659cd2179b2 - Init COMPLETE
+ip-26-0-172-142:789257:789685 [2] NCCL INFO comm 0xb2d77f0 rank 0 nranks 1 cudaDev 2 nvmlDev 2 busId 75000 commId 0x31ea480d69bce7af - Init COMPLETE
+ip-26-0-172-142:789259:789683 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-172-142:789259:789683 [4] NCCL INFO Channel 00/32 :    0
+ip-26-0-172-142:789259:789683 [4] NCCL INFO Channel 01/32 :    0
+ip-26-0-172-142:789259:789683 [4] NCCL INFO Channel 02/32 :    0
+ip-26-0-172-142:789259:789683 [4] NCCL INFO Channel 03/32 :    0
+ip-26-0-172-142:789259:789683 [4] NCCL INFO Channel 04/32 :    0
+ip-26-0-172-142:789259:789683 [4] NCCL INFO Channel 05/32 :    0
+ip-26-0-172-142:789259:789683 [4] NCCL INFO Channel 06/32 :    0
+ip-26-0-172-142:789259:789683 [4] NCCL INFO Channel 07/32 :    0
+ip-26-0-172-142:789259:789683 [4] NCCL INFO Channel 08/32 :    0
+ip-26-0-172-142:789259:789683 [4] NCCL INFO Channel 09/32 :    0
+ip-26-0-172-142:789259:789683 [4] NCCL INFO Channel 10/32 :    0
+ip-26-0-172-142:789259:789683 [4] NCCL INFO Channel 11/32 :    0
+ip-26-0-172-142:789259:789683 [4] NCCL INFO Channel 12/32 :    0
+ip-26-0-172-142:789259:789683 [4] NCCL INFO Channel 13/32 :    0
+ip-26-0-172-142:789259:789683 [4] NCCL INFO Channel 14/32 :    0
+ip-26-0-172-142:789259:789683 [4] NCCL INFO Channel 15/32 :    0
+ip-26-0-172-142:789259:789683 [4] NCCL INFO Channel 16/32 :    0
+ip-26-0-172-142:789259:789683 [4] NCCL INFO Channel 17/32 :    0
+ip-26-0-172-142:789259:789683 [4] NCCL INFO Channel 18/32 :    0
+ip-26-0-172-142:789259:789683 [4] NCCL INFO Channel 19/32 :    0
+ip-26-0-172-142:789259:789683 [4] NCCL INFO Channel 20/32 :    0
+ip-26-0-172-142:789259:789683 [4] NCCL INFO Channel 21/32 :    0
+ip-26-0-172-142:789259:789683 [4] NCCL INFO Channel 22/32 :    0
+ip-26-0-172-142:789259:789683 [4] NCCL INFO Channel 23/32 :    0
+ip-26-0-172-142:789259:789683 [4] NCCL INFO Channel 24/32 :    0
+ip-26-0-172-142:789259:789683 [4] NCCL INFO Channel 25/32 :    0
+ip-26-0-172-142:789259:789683 [4] NCCL INFO Channel 26/32 :    0
+ip-26-0-172-142:789259:789683 [4] NCCL INFO Channel 27/32 :    0
+ip-26-0-172-142:789259:789683 [4] NCCL INFO Channel 28/32 :    0
+ip-26-0-172-142:789259:789683 [4] NCCL INFO Channel 29/32 :    0
+ip-26-0-172-142:789259:789683 [4] NCCL INFO Channel 30/32 :    0
+ip-26-0-172-142:789259:789683 [4] NCCL INFO Channel 31/32 :    0
+ip-26-0-172-142:789259:789683 [4] NCCL INFO Trees [0] -1/-1/-1->0->-1 [1] -1/-1/-1->0->-1 [2] -1/-1/-1->0->-1 [3] -1/-1/-1->0->-1 [4] -1/-1/-1->0->-1 [5] -1/-1/-1->0->-1 [6] -1/-1/-1->0->-1 [7] -1/-1/-1->0->-1 [8] -1/-1/-1->0->-1 [9] -1/-1/-1->0->-1 [10] -1/-1/-1->0->-1 [11] -1/-1/-1->0->-1 [12] -1/-1/-1->0->-1 [13] -1/-1/-1->0->-1 [14] -1/-1/-1->0->-1 [15] -1/-1/-1->0->-1 [16] -1/-1/-1->0->-1 [17] -1/-1/-1->0->-1 [18] -1/-1/-1->0->-1 [19] -1/-1/-1->0->-1 [20] -1/-1/-1->0->-1 [21] -1/-1/-1->0->-1 [22] -1/-1/-1->0->-1 [23] -1/-1/-1->0->-1 [24] -1/-1/-1->0->-1 [25] -1/-1/-1->0->-1 [26] -1/-1/-1->0->-1 [27] -1/-1/-1->0->-1 [28] -1/-1/-1->0->-1 [29] -1/-1/-1->0->-1 [30] -1/-1/-1->0->-1 [31] -1/-1/-1->0->-1
+ip-26-0-172-142:789259:789683 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-142:789259:789683 [4] NCCL INFO Connected all rings
+ip-26-0-172-142:789259:789683 [4] NCCL INFO Connected all trees
+ip-26-0-172-142:789259:789683 [4] NCCL INFO 32 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-142:789258:789684 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-172-142:789258:789684 [3] NCCL INFO Channel 00/32 :    0
+ip-26-0-172-142:789258:789684 [3] NCCL INFO Channel 01/32 :    0
+ip-26-0-172-142:789258:789684 [3] NCCL INFO Channel 02/32 :    0
+ip-26-0-172-142:789258:789684 [3] NCCL INFO Channel 03/32 :    0
+ip-26-0-172-142:789258:789684 [3] NCCL INFO Channel 04/32 :    0
+ip-26-0-172-142:789258:789684 [3] NCCL INFO Channel 05/32 :    0
+ip-26-0-172-142:789258:789684 [3] NCCL INFO Channel 06/32 :    0
+ip-26-0-172-142:789258:789684 [3] NCCL INFO Channel 07/32 :    0
+ip-26-0-172-142:789258:789684 [3] NCCL INFO Channel 08/32 :    0
+ip-26-0-172-142:789258:789684 [3] NCCL INFO Channel 09/32 :    0
+ip-26-0-172-142:789258:789684 [3] NCCL INFO Channel 10/32 :    0
+ip-26-0-172-142:789258:789684 [3] NCCL INFO Channel 11/32 :    0
+ip-26-0-172-142:789258:789684 [3] NCCL INFO Channel 12/32 :    0
+ip-26-0-172-142:789258:789684 [3] NCCL INFO Channel 13/32 :    0
+ip-26-0-172-142:789258:789684 [3] NCCL INFO Channel 14/32 :    0
+ip-26-0-172-142:789258:789684 [3] NCCL INFO Channel 15/32 :    0
+ip-26-0-172-142:789258:789684 [3] NCCL INFO Channel 16/32 :    0
+ip-26-0-172-142:789258:789684 [3] NCCL INFO Channel 17/32 :    0
+ip-26-0-172-142:789258:789684 [3] NCCL INFO Channel 18/32 :    0
+ip-26-0-172-142:789258:789684 [3] NCCL INFO Channel 19/32 :    0
+ip-26-0-172-142:789258:789684 [3] NCCL INFO Channel 20/32 :    0
+ip-26-0-172-142:789258:789684 [3] NCCL INFO Channel 21/32 :    0
+ip-26-0-172-142:789258:789684 [3] NCCL INFO Channel 22/32 :    0
+ip-26-0-172-142:789258:789684 [3] NCCL INFO Channel 23/32 :    0
+ip-26-0-172-142:789258:789684 [3] NCCL INFO Channel 24/32 :    0
+ip-26-0-172-142:789258:789684 [3] NCCL INFO Channel 25/32 :    0
+ip-26-0-172-142:789258:789684 [3] NCCL INFO Channel 26/32 :    0
+ip-26-0-172-142:789258:789684 [3] NCCL INFO Channel 27/32 :    0
+ip-26-0-172-142:789258:789684 [3] NCCL INFO Channel 28/32 :    0
+ip-26-0-172-142:789258:789684 [3] NCCL INFO Channel 29/32 :    0
+ip-26-0-172-142:789258:789684 [3] NCCL INFO Channel 30/32 :    0
+ip-26-0-172-142:789258:789684 [3] NCCL INFO Channel 31/32 :    0
+ip-26-0-172-142:789258:789684 [3] NCCL INFO Trees [0] -1/-1/-1->0->-1 [1] -1/-1/-1->0->-1 [2] -1/-1/-1->0->-1 [3] -1/-1/-1->0->-1 [4] -1/-1/-1->0->-1 [5] -1/-1/-1->0->-1 [6] -1/-1/-1->0->-1 [7] -1/-1/-1->0->-1 [8] -1/-1/-1->0->-1 [9] -1/-1/-1->0->-1 [10] -1/-1/-1->0->-1 [11] -1/-1/-1->0->-1 [12] -1/-1/-1->0->-1 [13] -1/-1/-1->0->-1 [14] -1/-1/-1->0->-1 [15] -1/-1/-1->0->-1 [16] -1/-1/-1->0->-1 [17] -1/-1/-1->0->-1 [18] -1/-1/-1->0->-1 [19] -1/-1/-1->0->-1 [20] -1/-1/-1->0->-1 [21] -1/-1/-1->0->-1 [22] -1/-1/-1->0->-1 [23] -1/-1/-1->0->-1 [24] -1/-1/-1->0->-1 [25] -1/-1/-1->0->-1 [26] -1/-1/-1->0->-1 [27] -1/-1/-1->0->-1 [28] -1/-1/-1->0->-1 [29] -1/-1/-1->0->-1 [30] -1/-1/-1->0->-1 [31] -1/-1/-1->0->-1
+ip-26-0-172-142:789258:789684 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-142:789258:789684 [3] NCCL INFO Connected all rings
+ip-26-0-172-142:789258:789684 [3] NCCL INFO Connected all trees
+ip-26-0-172-142:789258:789684 [3] NCCL INFO 32 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-173-121:3987797:3988211 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-173-121:3987797:3988211 [3] NCCL INFO Channel 00/32 :    0
+ip-26-0-173-121:3987797:3988211 [3] NCCL INFO Channel 01/32 :    0
+ip-26-0-173-121:3987797:3988211 [3] NCCL INFO Channel 02/32 :    0
+ip-26-0-173-121:3987797:3988211 [3] NCCL INFO Channel 03/32 :    0
+ip-26-0-173-121:3987797:3988211 [3] NCCL INFO Channel 04/32 :    0
+ip-26-0-173-121:3987797:3988211 [3] NCCL INFO Channel 05/32 :    0
+ip-26-0-173-121:3987797:3988211 [3] NCCL INFO Channel 06/32 :    0
+ip-26-0-173-121:3987797:3988211 [3] NCCL INFO Channel 07/32 :    0
+ip-26-0-173-121:3987797:3988211 [3] NCCL INFO Channel 08/32 :    0
+ip-26-0-173-121:3987797:3988211 [3] NCCL INFO Channel 09/32 :    0
+ip-26-0-173-121:3987797:3988211 [3] NCCL INFO Channel 10/32 :    0
+ip-26-0-173-121:3987797:3988211 [3] NCCL INFO Channel 11/32 :    0
+ip-26-0-173-121:3987797:3988211 [3] NCCL INFO Channel 12/32 :    0
+ip-26-0-173-121:3987797:3988211 [3] NCCL INFO Channel 13/32 :    0
+ip-26-0-173-121:3987797:3988211 [3] NCCL INFO Channel 14/32 :    0
+ip-26-0-173-121:3987797:3988211 [3] NCCL INFO Channel 15/32 :    0
+ip-26-0-173-121:3987797:3988211 [3] NCCL INFO Channel 16/32 :    0
+ip-26-0-173-121:3987797:3988211 [3] NCCL INFO Channel 17/32 :    0
+ip-26-0-173-121:3987797:3988211 [3] NCCL INFO Channel 18/32 :    0
+ip-26-0-173-121:3987797:3988211 [3] NCCL INFO Channel 19/32 :    0
+ip-26-0-173-121:3987797:3988211 [3] NCCL INFO Channel 20/32 :    0
+ip-26-0-173-121:3987797:3988211 [3] NCCL INFO Channel 21/32 :    0
+ip-26-0-173-121:3987797:3988211 [3] NCCL INFO Channel 22/32 :    0
+ip-26-0-173-121:3987797:3988211 [3] NCCL INFO Channel 23/32 :    0
+ip-26-0-173-121:3987797:3988211 [3] NCCL INFO Channel 24/32 :    0
+ip-26-0-173-121:3987797:3988211 [3] NCCL INFO Channel 25/32 :    0
+ip-26-0-173-121:3987797:3988211 [3] NCCL INFO Channel 26/32 :    0
+ip-26-0-173-121:3987797:3988211 [3] NCCL INFO Channel 27/32 :    0
+ip-26-0-173-121:3987797:3988211 [3] NCCL INFO Channel 28/32 :    0
+ip-26-0-173-121:3987797:3988211 [3] NCCL INFO Channel 29/32 :    0
+ip-26-0-173-121:3987797:3988211 [3] NCCL INFO Channel 30/32 :    0
+ip-26-0-173-121:3987797:3988211 [3] NCCL INFO Channel 31/32 :    0
+ip-26-0-173-121:3987797:3988211 [3] NCCL INFO Trees [0] -1/-1/-1->0->-1 [1] -1/-1/-1->0->-1 [2] -1/-1/-1->0->-1 [3] -1/-1/-1->0->-1 [4] -1/-1/-1->0->-1 [5] -1/-1/-1->0->-1 [6] -1/-1/-1->0->-1 [7] -1/-1/-1->0->-1 [8] -1/-1/-1->0->-1 [9] -1/-1/-1->0->-1 [10] -1/-1/-1->0->-1 [11] -1/-1/-1->0->-1 [12] -1/-1/-1->0->-1 [13] -1/-1/-1->0->-1 [14] -1/-1/-1->0->-1 [15] -1/-1/-1->0->-1 [16] -1/-1/-1->0->-1 [17] -1/-1/-1->0->-1 [18] -1/-1/-1->0->-1 [19] -1/-1/-1->0->-1 [20] -1/-1/-1->0->-1 [21] -1/-1/-1->0->-1 [22] -1/-1/-1->0->-1 [23] -1/-1/-1->0->-1 [24] -1/-1/-1->0->-1 [25] -1/-1/-1->0->-1 [26] -1/-1/-1->0->-1 [27] -1/-1/-1->0->-1 [28] -1/-1/-1->0->-1 [29] -1/-1/-1->0->-1 [30] -1/-1/-1->0->-1 [31] -1/-1/-1->0->-1
+ip-26-0-173-121:3987797:3988211 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-142:789255:789693 [0] NCCL INFO comm 0xc0dfdc0 rank 0 nranks 1 cudaDev 0 nvmlDev 0 busId 53000 commId 0xa15da757bb15a5fd - Init COMPLETE
+ip-26-0-173-121:3987797:3988211 [3] NCCL INFO Connected all rings
+ip-26-0-173-121:3987797:3988211 [3] NCCL INFO Connected all trees
+ip-26-0-173-121:3987797:3988211 [3] NCCL INFO 32 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-142:789262:789680 [7] NCCL INFO comm 0xa93a770 rank 0 nranks 1 cudaDev 7 nvmlDev 7 busId ca000 commId 0x7f37ce22192b3cb9 - Init COMPLETE
+ip-26-0-172-142:789260:789677 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-172-142:789260:789677 [5] NCCL INFO Channel 00/32 :    0
+ip-26-0-172-142:789260:789677 [5] NCCL INFO Channel 01/32 :    0
+ip-26-0-172-142:789260:789677 [5] NCCL INFO Channel 02/32 :    0
+ip-26-0-172-142:789260:789677 [5] NCCL INFO Channel 03/32 :    0
+ip-26-0-172-142:789260:789677 [5] NCCL INFO Channel 04/32 :    0
+ip-26-0-172-142:789260:789677 [5] NCCL INFO Channel 05/32 :    0
+ip-26-0-172-142:789260:789677 [5] NCCL INFO Channel 06/32 :    0
+ip-26-0-172-142:789260:789677 [5] NCCL INFO Channel 07/32 :    0
+ip-26-0-172-142:789260:789677 [5] NCCL INFO Channel 08/32 :    0
+ip-26-0-172-142:789260:789677 [5] NCCL INFO Channel 09/32 :    0
+ip-26-0-172-142:789260:789677 [5] NCCL INFO Channel 10/32 :    0
+ip-26-0-172-142:789260:789677 [5] NCCL INFO Channel 11/32 :    0
+ip-26-0-172-142:789260:789677 [5] NCCL INFO Channel 12/32 :    0
+ip-26-0-172-142:789260:789677 [5] NCCL INFO Channel 13/32 :    0
+ip-26-0-172-142:789260:789677 [5] NCCL INFO Channel 14/32 :    0
+ip-26-0-172-142:789260:789677 [5] NCCL INFO Channel 15/32 :    0
+ip-26-0-172-142:789260:789677 [5] NCCL INFO Channel 16/32 :    0
+ip-26-0-172-142:789260:789677 [5] NCCL INFO Channel 17/32 :    0
+ip-26-0-172-142:789260:789677 [5] NCCL INFO Channel 18/32 :    0
+ip-26-0-172-142:789260:789677 [5] NCCL INFO Channel 19/32 :    0
+ip-26-0-172-142:789260:789677 [5] NCCL INFO Channel 20/32 :    0
+ip-26-0-172-142:789260:789677 [5] NCCL INFO Channel 21/32 :    0
+ip-26-0-172-142:789260:789677 [5] NCCL INFO Channel 22/32 :    0
+ip-26-0-172-142:789260:789677 [5] NCCL INFO Channel 23/32 :    0
+ip-26-0-172-142:789260:789677 [5] NCCL INFO Channel 24/32 :    0
+ip-26-0-172-142:789260:789677 [5] NCCL INFO Channel 25/32 :    0
+ip-26-0-172-142:789260:789677 [5] NCCL INFO Channel 26/32 :    0
+ip-26-0-172-142:789260:789677 [5] NCCL INFO Channel 27/32 :    0
+ip-26-0-172-142:789260:789677 [5] NCCL INFO Channel 28/32 :    0
+ip-26-0-172-142:789260:789677 [5] NCCL INFO Channel 29/32 :    0
+ip-26-0-172-142:789260:789677 [5] NCCL INFO Channel 30/32 :    0
+ip-26-0-172-142:789260:789677 [5] NCCL INFO Channel 31/32 :    0
+ip-26-0-172-142:789260:789677 [5] NCCL INFO Trees [0] -1/-1/-1->0->-1 [1] -1/-1/-1->0->-1 [2] -1/-1/-1->0->-1 [3] -1/-1/-1->0->-1 [4] -1/-1/-1->0->-1 [5] -1/-1/-1->0->-1 [6] -1/-1/-1->0->-1 [7] -1/-1/-1->0->-1 [8] -1/-1/-1->0->-1 [9] -1/-1/-1->0->-1 [10] -1/-1/-1->0->-1 [11] -1/-1/-1->0->-1 [12] -1/-1/-1->0->-1 [13] -1/-1/-1->0->-1 [14] -1/-1/-1->0->-1 [15] -1/-1/-1->0->-1 [16] -1/-1/-1->0->-1 [17] -1/-1/-1->0->-1 [18] -1/-1/-1->0->-1 [19] -1/-1/-1->0->-1 [20] -1/-1/-1->0->-1 [21] -1/-1/-1->0->-1 [22] -1/-1/-1->0->-1 [23] -1/-1/-1->0->-1 [24] -1/-1/-1->0->-1 [25] -1/-1/-1->0->-1 [26] -1/-1/-1->0->-1 [27] -1/-1/-1->0->-1 [28] -1/-1/-1->0->-1 [29] -1/-1/-1->0->-1 [30] -1/-1/-1->0->-1 [31] -1/-1/-1->0->-1
+ip-26-0-172-142:789260:789677 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-142:789260:789677 [5] NCCL INFO Connected all rings
+ip-26-0-172-142:789260:789677 [5] NCCL INFO Connected all trees
+ip-26-0-172-142:789260:789677 [5] NCCL INFO 32 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-173-121:3987795:3988215 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-173-121:3987795:3988215 [1] NCCL INFO Channel 00/32 :    0
+ip-26-0-173-121:3987795:3988215 [1] NCCL INFO Channel 01/32 :    0
+ip-26-0-173-121:3987795:3988215 [1] NCCL INFO Channel 02/32 :    0
+ip-26-0-173-121:3987795:3988215 [1] NCCL INFO Channel 03/32 :    0
+ip-26-0-173-121:3987795:3988215 [1] NCCL INFO Channel 04/32 :    0
+ip-26-0-173-121:3987795:3988215 [1] NCCL INFO Channel 05/32 :    0
+ip-26-0-173-121:3987795:3988215 [1] NCCL INFO Channel 06/32 :    0
+ip-26-0-173-121:3987795:3988215 [1] NCCL INFO Channel 07/32 :    0
+ip-26-0-173-121:3987795:3988215 [1] NCCL INFO Channel 08/32 :    0
+ip-26-0-173-121:3987795:3988215 [1] NCCL INFO Channel 09/32 :    0
+ip-26-0-173-121:3987795:3988215 [1] NCCL INFO Channel 10/32 :    0
+ip-26-0-173-121:3987795:3988215 [1] NCCL INFO Channel 11/32 :    0
+ip-26-0-173-121:3987795:3988215 [1] NCCL INFO Channel 12/32 :    0
+ip-26-0-173-121:3987795:3988215 [1] NCCL INFO Channel 13/32 :    0
+ip-26-0-173-121:3987795:3988215 [1] NCCL INFO Channel 14/32 :    0
+ip-26-0-173-121:3987795:3988215 [1] NCCL INFO Channel 15/32 :    0
+ip-26-0-173-121:3987795:3988215 [1] NCCL INFO Channel 16/32 :    0
+ip-26-0-173-121:3987795:3988215 [1] NCCL INFO Channel 17/32 :    0
+ip-26-0-173-121:3987795:3988215 [1] NCCL INFO Channel 18/32 :    0
+ip-26-0-173-121:3987795:3988215 [1] NCCL INFO Channel 19/32 :    0
+ip-26-0-173-121:3987795:3988215 [1] NCCL INFO Channel 20/32 :    0
+ip-26-0-173-121:3987795:3988215 [1] NCCL INFO Channel 21/32 :    0
+ip-26-0-173-121:3987795:3988215 [1] NCCL INFO Channel 22/32 :    0
+ip-26-0-173-121:3987795:3988215 [1] NCCL INFO Channel 23/32 :    0
+ip-26-0-173-121:3987795:3988215 [1] NCCL INFO Channel 24/32 :    0
+ip-26-0-173-121:3987795:3988215 [1] NCCL INFO Channel 25/32 :    0
+ip-26-0-173-121:3987795:3988215 [1] NCCL INFO Channel 26/32 :    0
+ip-26-0-173-121:3987795:3988215 [1] NCCL INFO Channel 27/32 :    0
+ip-26-0-173-121:3987795:3988215 [1] NCCL INFO Channel 28/32 :    0
+ip-26-0-173-121:3987795:3988215 [1] NCCL INFO Channel 29/32 :    0
+ip-26-0-173-121:3987795:3988215 [1] NCCL INFO Channel 30/32 :    0
+ip-26-0-173-121:3987795:3988215 [1] NCCL INFO Channel 31/32 :    0
+ip-26-0-173-121:3987795:3988215 [1] NCCL INFO Trees [0] -1/-1/-1->0->-1 [1] -1/-1/-1->0->-1 [2] -1/-1/-1->0->-1 [3] -1/-1/-1->0->-1 [4] -1/-1/-1->0->-1 [5] -1/-1/-1->0->-1 [6] -1/-1/-1->0->-1 [7] -1/-1/-1->0->-1 [8] -1/-1/-1->0->-1 [9] -1/-1/-1->0->-1 [10] -1/-1/-1->0->-1 [11] -1/-1/-1->0->-1 [12] -1/-1/-1->0->-1 [13] -1/-1/-1->0->-1 [14] -1/-1/-1->0->-1 [15] -1/-1/-1->0->-1 [16] -1/-1/-1->0->-1 [17] -1/-1/-1->0->-1 [18] -1/-1/-1->0->-1 [19] -1/-1/-1->0->-1 [20] -1/-1/-1->0->-1 [21] -1/-1/-1->0->-1 [22] -1/-1/-1->0->-1 [23] -1/-1/-1->0->-1 [24] -1/-1/-1->0->-1 [25] -1/-1/-1->0->-1 [26] -1/-1/-1->0->-1 [27] -1/-1/-1->0->-1 [28] -1/-1/-1->0->-1 [29] -1/-1/-1->0->-1 [30] -1/-1/-1->0->-1 [31] -1/-1/-1->0->-1
+ip-26-0-173-121:3987795:3988215 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-173-121:3987795:3988215 [1] NCCL INFO Connected all rings
+ip-26-0-173-121:3987795:3988215 [1] NCCL INFO Connected all trees
+ip-26-0-173-121:3987795:3988215 [1] NCCL INFO 32 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-142:789259:789683 [4] NCCL INFO comm 0xaf7be30 rank 0 nranks 1 cudaDev 4 nvmlDev 4 busId 97000 commId 0x299f92c3f3ff3cc8 - Init COMPLETE
+ip-26-0-173-121:3987799:3988217 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-173-121:3987799:3988217 [5] NCCL INFO Channel 00/32 :    0
+ip-26-0-173-121:3987799:3988217 [5] NCCL INFO Channel 01/32 :    0
+ip-26-0-173-121:3987799:3988217 [5] NCCL INFO Channel 02/32 :    0
+ip-26-0-173-121:3987799:3988217 [5] NCCL INFO Channel 03/32 :    0
+ip-26-0-173-121:3987799:3988217 [5] NCCL INFO Channel 04/32 :    0
+ip-26-0-173-121:3987799:3988217 [5] NCCL INFO Channel 05/32 :    0
+ip-26-0-173-121:3987799:3988217 [5] NCCL INFO Channel 06/32 :    0
+ip-26-0-173-121:3987799:3988217 [5] NCCL INFO Channel 07/32 :    0
+ip-26-0-173-121:3987799:3988217 [5] NCCL INFO Channel 08/32 :    0
+ip-26-0-173-121:3987799:3988217 [5] NCCL INFO Channel 09/32 :    0
+ip-26-0-173-121:3987799:3988217 [5] NCCL INFO Channel 10/32 :    0
+ip-26-0-173-121:3987799:3988217 [5] NCCL INFO Channel 11/32 :    0
+ip-26-0-173-121:3987799:3988217 [5] NCCL INFO Channel 12/32 :    0
+ip-26-0-173-121:3987799:3988217 [5] NCCL INFO Channel 13/32 :    0
+ip-26-0-173-121:3987799:3988217 [5] NCCL INFO Channel 14/32 :    0
+ip-26-0-173-121:3987799:3988217 [5] NCCL INFO Channel 15/32 :    0
+ip-26-0-173-121:3987799:3988217 [5] NCCL INFO Channel 16/32 :    0
+ip-26-0-173-121:3987799:3988217 [5] NCCL INFO Channel 17/32 :    0
+ip-26-0-173-121:3987799:3988217 [5] NCCL INFO Channel 18/32 :    0
+ip-26-0-173-121:3987799:3988217 [5] NCCL INFO Channel 19/32 :    0
+ip-26-0-173-121:3987799:3988217 [5] NCCL INFO Channel 20/32 :    0
+ip-26-0-173-121:3987799:3988217 [5] NCCL INFO Channel 21/32 :    0
+ip-26-0-173-121:3987799:3988217 [5] NCCL INFO Channel 22/32 :    0
+ip-26-0-173-121:3987799:3988217 [5] NCCL INFO Channel 23/32 :    0
+ip-26-0-173-121:3987799:3988217 [5] NCCL INFO Channel 24/32 :    0
+ip-26-0-173-121:3987799:3988217 [5] NCCL INFO Channel 25/32 :    0
+ip-26-0-173-121:3987799:3988217 [5] NCCL INFO Channel 26/32 :    0
+ip-26-0-173-121:3987799:3988217 [5] NCCL INFO Channel 27/32 :    0
+ip-26-0-173-121:3987799:3988217 [5] NCCL INFO Channel 28/32 :    0
+ip-26-0-173-121:3987799:3988217 [5] NCCL INFO Channel 29/32 :    0
+ip-26-0-173-121:3987799:3988217 [5] NCCL INFO Channel 30/32 :    0
+ip-26-0-173-121:3987799:3988217 [5] NCCL INFO Channel 31/32 :    0
+ip-26-0-173-121:3987799:3988217 [5] NCCL INFO Trees [0] -1/-1/-1->0->-1 [1] -1/-1/-1->0->-1 [2] -1/-1/-1->0->-1 [3] -1/-1/-1->0->-1 [4] -1/-1/-1->0->-1 [5] -1/-1/-1->0->-1 [6] -1/-1/-1->0->-1 [7] -1/-1/-1->0->-1 [8] -1/-1/-1->0->-1 [9] -1/-1/-1->0->-1 [10] -1/-1/-1->0->-1 [11] -1/-1/-1->0->-1 [12] -1/-1/-1->0->-1 [13] -1/-1/-1->0->-1 [14] -1/-1/-1->0->-1 [15] -1/-1/-1->0->-1 [16] -1/-1/-1->0->-1 [17] -1/-1/-1->0->-1 [18] -1/-1/-1->0->-1 [19] -1/-1/-1->0->-1 [20] -1/-1/-1->0->-1 [21] -1/-1/-1->0->-1 [22] -1/-1/-1->0->-1 [23] -1/-1/-1->0->-1 [24] -1/-1/-1->0->-1 [25] -1/-1/-1->0->-1 [26] -1/-1/-1->0->-1 [27] -1/-1/-1->0->-1 [28] -1/-1/-1->0->-1 [29] -1/-1/-1->0->-1 [30] -1/-1/-1->0->-1 [31] -1/-1/-1->0->-1
+ip-26-0-173-121:3987799:3988217 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-142:789258:789684 [3] NCCL INFO comm 0xaa5bbb0 rank 0 nranks 1 cudaDev 3 nvmlDev 3 busId 86000 commId 0x3d2b6343e711457 - Init COMPLETE
+ip-26-0-173-121:3987799:3988217 [5] NCCL INFO Connected all rings
+ip-26-0-173-121:3987799:3988217 [5] NCCL INFO Connected all trees
+ip-26-0-173-121:3987799:3988217 [5] NCCL INFO 32 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-173-121:3987796:3988210 [2] NCCL INFO comm 0xafd75a0 rank 0 nranks 1 cudaDev 2 nvmlDev 2 busId 75000 commId 0xf536fd0616c61a14 - Init COMPLETE
+ip-26-0-172-142:789260:789677 [5] NCCL INFO comm 0xbfa1a80 rank 0 nranks 1 cudaDev 5 nvmlDev 5 busId a8000 commId 0xfc05f86a91bb9bd - Init COMPLETE
+ip-26-0-173-121:3987794:3988214 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-173-121:3987794:3988214 [0] NCCL INFO Channel 00/32 :    0
+ip-26-0-173-121:3987794:3988214 [0] NCCL INFO Channel 01/32 :    0
+ip-26-0-173-121:3987794:3988214 [0] NCCL INFO Channel 02/32 :    0
+ip-26-0-173-121:3987794:3988214 [0] NCCL INFO Channel 03/32 :    0
+ip-26-0-173-121:3987794:3988214 [0] NCCL INFO Channel 04/32 :    0
+ip-26-0-173-121:3987794:3988214 [0] NCCL INFO Channel 05/32 :    0
+ip-26-0-173-121:3987794:3988214 [0] NCCL INFO Channel 06/32 :    0
+ip-26-0-173-121:3987794:3988214 [0] NCCL INFO Channel 07/32 :    0
+ip-26-0-173-121:3987794:3988214 [0] NCCL INFO Channel 08/32 :    0
+ip-26-0-173-121:3987794:3988214 [0] NCCL INFO Channel 09/32 :    0
+ip-26-0-173-121:3987794:3988214 [0] NCCL INFO Channel 10/32 :    0
+ip-26-0-173-121:3987794:3988214 [0] NCCL INFO Channel 11/32 :    0
+ip-26-0-173-121:3987794:3988214 [0] NCCL INFO Channel 12/32 :    0
+ip-26-0-173-121:3987794:3988214 [0] NCCL INFO Channel 13/32 :    0
+ip-26-0-173-121:3987794:3988214 [0] NCCL INFO Channel 14/32 :    0
+ip-26-0-173-121:3987794:3988214 [0] NCCL INFO Channel 15/32 :    0
+ip-26-0-173-121:3987794:3988214 [0] NCCL INFO Channel 16/32 :    0
+ip-26-0-173-121:3987794:3988214 [0] NCCL INFO Channel 17/32 :    0
+ip-26-0-173-121:3987794:3988214 [0] NCCL INFO Channel 18/32 :    0
+ip-26-0-173-121:3987794:3988214 [0] NCCL INFO Channel 19/32 :    0
+ip-26-0-173-121:3987794:3988214 [0] NCCL INFO Channel 20/32 :    0
+ip-26-0-173-121:3987794:3988214 [0] NCCL INFO Channel 21/32 :    0
+ip-26-0-173-121:3987797:3988211 [3] NCCL INFO comm 0xaaad830 rank 0 nranks 1 cudaDev 3 nvmlDev 3 busId 86000 commId 0xa04a18d93d84e9ae - Init COMPLETE
+ip-26-0-173-121:3987794:3988214 [0] NCCL INFO Channel 22/32 :    0
+ip-26-0-173-121:3987794:3988214 [0] NCCL INFO Channel 23/32 :    0
+ip-26-0-173-121:3987794:3988214 [0] NCCL INFO Channel 24/32 :    0
+ip-26-0-173-121:3987794:3988214 [0] NCCL INFO Channel 25/32 :    0
+ip-26-0-173-121:3987794:3988214 [0] NCCL INFO Channel 26/32 :    0
+ip-26-0-173-121:3987794:3988214 [0] NCCL INFO Channel 27/32 :    0
+ip-26-0-173-121:3987794:3988214 [0] NCCL INFO Channel 28/32 :    0
+ip-26-0-173-121:3987794:3988214 [0] NCCL INFO Channel 29/32 :    0
+ip-26-0-173-121:3987794:3988214 [0] NCCL INFO Channel 30/32 :    0
+ip-26-0-173-121:3987794:3988214 [0] NCCL INFO Channel 31/32 :    0
+ip-26-0-173-121:3987794:3988214 [0] NCCL INFO Trees [0] -1/-1/-1->0->-1 [1] -1/-1/-1->0->-1 [2] -1/-1/-1->0->-1 [3] -1/-1/-1->0->-1 [4] -1/-1/-1->0->-1 [5] -1/-1/-1->0->-1 [6] -1/-1/-1->0->-1 [7] -1/-1/-1->0->-1 [8] -1/-1/-1->0->-1 [9] -1/-1/-1->0->-1 [10] -1/-1/-1->0->-1 [11] -1/-1/-1->0->-1 [12] -1/-1/-1->0->-1 [13] -1/-1/-1->0->-1 [14] -1/-1/-1->0->-1 [15] -1/-1/-1->0->-1 [16] -1/-1/-1->0->-1 [17] -1/-1/-1->0->-1 [18] -1/-1/-1->0->-1 [19] -1/-1/-1->0->-1 [20] -1/-1/-1->0->-1 [21] -1/-1/-1->0->-1 [22] -1/-1/-1->0->-1 [23] -1/-1/-1->0->-1 [24] -1/-1/-1->0->-1 [25] -1/-1/-1->0->-1 [26] -1/-1/-1->0->-1 [27] -1/-1/-1->0->-1 [28] -1/-1/-1->0->-1 [29] -1/-1/-1->0->-1 [30] -1/-1/-1->0->-1 [31] -1/-1/-1->0->-1
+ip-26-0-173-121:3987794:3988214 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-173-121:3987794:3988214 [0] NCCL INFO Connected all rings
+ip-26-0-173-121:3987794:3988214 [0] NCCL INFO Connected all trees
+ip-26-0-173-121:3987794:3988214 [0] NCCL INFO 32 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-173-121:3987801:3988227 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-173-121:3987801:3988227 [7] NCCL INFO Channel 00/32 :    0
+ip-26-0-173-121:3987801:3988227 [7] NCCL INFO Channel 01/32 :    0
+ip-26-0-173-121:3987801:3988227 [7] NCCL INFO Channel 02/32 :    0
+ip-26-0-173-121:3987801:3988227 [7] NCCL INFO Channel 03/32 :    0
+ip-26-0-173-121:3987801:3988227 [7] NCCL INFO Channel 04/32 :    0
+ip-26-0-173-121:3987801:3988227 [7] NCCL INFO Channel 05/32 :    0
+ip-26-0-173-121:3987801:3988227 [7] NCCL INFO Channel 06/32 :    0
+ip-26-0-173-121:3987801:3988227 [7] NCCL INFO Channel 07/32 :    0
+ip-26-0-173-121:3987801:3988227 [7] NCCL INFO Channel 08/32 :    0
+ip-26-0-173-121:3987801:3988227 [7] NCCL INFO Channel 09/32 :    0
+ip-26-0-173-121:3987801:3988227 [7] NCCL INFO Channel 10/32 :    0
+ip-26-0-173-121:3987801:3988227 [7] NCCL INFO Channel 11/32 :    0
+ip-26-0-173-121:3987801:3988227 [7] NCCL INFO Channel 12/32 :    0
+ip-26-0-173-121:3987801:3988227 [7] NCCL INFO Channel 13/32 :    0
+ip-26-0-173-121:3987801:3988227 [7] NCCL INFO Channel 14/32 :    0
+ip-26-0-173-121:3987801:3988227 [7] NCCL INFO Channel 15/32 :    0
+ip-26-0-173-121:3987801:3988227 [7] NCCL INFO Channel 16/32 :    0
+ip-26-0-173-121:3987801:3988227 [7] NCCL INFO Channel 17/32 :    0
+ip-26-0-173-121:3987801:3988227 [7] NCCL INFO Channel 18/32 :    0
+ip-26-0-173-121:3987801:3988227 [7] NCCL INFO Channel 19/32 :    0
+ip-26-0-173-121:3987801:3988227 [7] NCCL INFO Channel 20/32 :    0
+ip-26-0-173-121:3987801:3988227 [7] NCCL INFO Channel 21/32 :    0
+ip-26-0-173-121:3987801:3988227 [7] NCCL INFO Channel 22/32 :    0
+ip-26-0-173-121:3987801:3988227 [7] NCCL INFO Channel 23/32 :    0
+ip-26-0-173-121:3987801:3988227 [7] NCCL INFO Channel 24/32 :    0
+ip-26-0-173-121:3987801:3988227 [7] NCCL INFO Channel 25/32 :    0
+ip-26-0-173-121:3987801:3988227 [7] NCCL INFO Channel 26/32 :    0
+ip-26-0-173-121:3987801:3988227 [7] NCCL INFO Channel 27/32 :    0
+ip-26-0-173-121:3987801:3988227 [7] NCCL INFO Channel 28/32 :    0
+ip-26-0-173-121:3987801:3988227 [7] NCCL INFO Channel 29/32 :    0
+ip-26-0-173-121:3987801:3988227 [7] NCCL INFO Channel 30/32 :    0
+ip-26-0-173-121:3987801:3988227 [7] NCCL INFO Channel 31/32 :    0
+ip-26-0-173-121:3987801:3988227 [7] NCCL INFO Trees [0] -1/-1/-1->0->-1 [1] -1/-1/-1->0->-1 [2] -1/-1/-1->0->-1 [3] -1/-1/-1->0->-1 [4] -1/-1/-1->0->-1 [5] -1/-1/-1->0->-1 [6] -1/-1/-1->0->-1 [7] -1/-1/-1->0->-1 [8] -1/-1/-1->0->-1 [9] -1/-1/-1->0->-1 [10] -1/-1/-1->0->-1 [11] -1/-1/-1->0->-1 [12] -1/-1/-1->0->-1 [13] -1/-1/-1->0->-1 [14] -1/-1/-1->0->-1 [15] -1/-1/-1->0->-1 [16] -1/-1/-1->0->-1 [17] -1/-1/-1->0->-1 [18] -1/-1/-1->0->-1 [19] -1/-1/-1->0->-1 [20] -1/-1/-1->0->-1 [21] -1/-1/-1->0->-1 [22] -1/-1/-1->0->-1 [23] -1/-1/-1->0->-1 [24] -1/-1/-1->0->-1 [25] -1/-1/-1->0->-1 [26] -1/-1/-1->0->-1 [27] -1/-1/-1->0->-1 [28] -1/-1/-1->0->-1 [29] -1/-1/-1->0->-1 [30] -1/-1/-1->0->-1 [31] -1/-1/-1->0->-1
+ip-26-0-173-121:3987801:3988227 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-173-121:3987801:3988227 [7] NCCL INFO Connected all rings
+ip-26-0-173-121:3987801:3988227 [7] NCCL INFO Connected all trees
+ip-26-0-173-121:3987801:3988227 [7] NCCL INFO 32 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-173-121:3987795:3988215 [1] NCCL INFO comm 0xb73fa40 rank 0 nranks 1 cudaDev 1 nvmlDev 1 busId 64000 commId 0xa5fcc36c4b51c4c - Init COMPLETE
+ip-26-0-173-121:3987799:3988217 [5] NCCL INFO comm 0xb109ef0 rank 0 nranks 1 cudaDev 5 nvmlDev 5 busId a8000 commId 0x5cf0af34367bb8c4 - Init COMPLETE
+ip-26-0-173-121:3987794:3988214 [0] NCCL INFO comm 0xab7c0d0 rank 0 nranks 1 cudaDev 0 nvmlDev 0 busId 53000 commId 0x111566d954b745ba - Init COMPLETE
+ip-26-0-173-121:3987801:3988227 [7] NCCL INFO comm 0xab41930 rank 0 nranks 1 cudaDev 7 nvmlDev 7 busId ca000 commId 0x89a5658a982d7fda - Init COMPLETE
+ip-26-0-172-142:789255:789714 [0] NCCL INFO Using network Libfabric
+ip-26-0-172-142:789261:789716 [6] NCCL INFO Using network Libfabric
+ip-26-0-172-142:789257:789715 [2] NCCL INFO Using network Libfabric
+ip-26-0-173-121:3987798:3988247 [4] NCCL INFO Using network Libfabric
+ip-26-0-172-142:789259:789717 [4] NCCL INFO Using network Libfabric
+ip-26-0-173-121:3987794:3988245 [0] NCCL INFO Using network Libfabric
+ip-26-0-173-121:3987796:3988246 [2] NCCL INFO Using network Libfabric
+ip-26-0-173-121:3987800:3988248 [6] NCCL INFO Using network Libfabric
+ip-26-0-172-142:789256:789719 [1] NCCL INFO Using network Libfabric
+ip-26-0-172-142:789258:789720 [3] NCCL INFO Using network Libfabric
+ip-26-0-173-121:3987801:3988249 [7] NCCL INFO Using network Libfabric
+ip-26-0-172-142:789260:789721 [5] NCCL INFO Using network Libfabric
+ip-26-0-172-142:789262:789722 [7] NCCL INFO Using network Libfabric
+ip-26-0-173-121:3987799:3988250 [5] NCCL INFO Using network Libfabric
+ip-26-0-173-121:3987797:3988251 [3] NCCL INFO Using network Libfabric
+ip-26-0-173-121:3987795:3988252 [1] NCCL INFO Using network Libfabric
+ip-26-0-172-142:789261:789716 [6] NCCL INFO comm 0xbd202a0 rank 3 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0xd3c41587a72a5b5b - Init START
+ip-26-0-172-142:789259:789717 [4] NCCL INFO comm 0xaf7eae0 rank 2 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0xd3c41587a72a5b5b - Init START
+ip-26-0-172-142:789257:789715 [2] NCCL INFO comm 0xb2da4a0 rank 1 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0xd3c41587a72a5b5b - Init START
+ip-26-0-172-142:789255:789714 [0] NCCL INFO comm 0xc0e2a70 rank 0 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0xd3c41587a72a5b5b - Init START
+ip-26-0-173-121:3987800:3988248 [6] NCCL INFO comm 0xa732240 rank 7 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0xd3c41587a72a5b5b - Init START
+ip-26-0-173-121:3987798:3988247 [4] NCCL INFO comm 0xbb0d530 rank 6 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0xd3c41587a72a5b5b - Init START
+ip-26-0-173-121:3987796:3988246 [2] NCCL INFO comm 0xafda250 rank 5 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0xd3c41587a72a5b5b - Init START
+ip-26-0-173-121:3987794:3988245 [0] NCCL INFO comm 0xab7ed80 rank 4 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0xd3c41587a72a5b5b - Init START
+ip-26-0-172-142:789261:789716 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-172-142:789259:789717 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-172-142:789255:789714 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/205
+ip-26-0-172-142:789257:789715 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-173-121:3987794:3988245 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-173-121:3987800:3988248 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-173-121:3987796:3988246 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-173-121:3987798:3988247 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-172-142:789262:789722 [7] NCCL INFO comm 0xa93d420 rank 3 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0x700714da38272a85 - Init START
+ip-26-0-172-142:789260:789721 [5] NCCL INFO comm 0xbfa4730 rank 2 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0x700714da38272a85 - Init START
+ip-26-0-173-121:3987801:3988249 [7] NCCL INFO comm 0xab445e0 rank 7 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0x700714da38272a85 - Init START
+ip-26-0-172-142:789258:789720 [3] NCCL INFO comm 0xaa5e860 rank 1 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0x700714da38272a85 - Init START
+ip-26-0-172-142:789256:789719 [1] NCCL INFO comm 0xc430a60 rank 0 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0x700714da38272a85 - Init START
+ip-26-0-173-121:3987799:3988250 [5] NCCL INFO comm 0xb10cba0 rank 6 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0x700714da38272a85 - Init START
+ip-26-0-173-121:3987797:3988251 [3] NCCL INFO comm 0xaab04e0 rank 5 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0x700714da38272a85 - Init START
+ip-26-0-173-121:3987795:3988252 [1] NCCL INFO comm 0xb7426f0 rank 4 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0x700714da38272a85 - Init START
+ip-26-0-173-121:3987799:3988250 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-173-121:3987801:3988249 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-172-142:789262:789722 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-172-142:789260:789721 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-172-142:789256:789719 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-172-142:789258:789720 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-173-121:3987795:3988252 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-173-121:3987797:3988251 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-172-142:789258:789720 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-172-142:789258:789720 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-173-121:3987799:3988250 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-173-121:3987799:3988250 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-172-142:789256:789719 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-172-142:789256:789719 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-172-142:789262:789722 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-172-142:789262:789722 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-172-142:789259:789717 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-172-142:789259:789717 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-173-121:3987798:3988247 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-173-121:3987798:3988247 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-172-142:789255:789714 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-172-142:789255:789714 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-172-142:789260:789721 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-172-142:789260:789721 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-172-142:789261:789716 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-172-142:789261:789716 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-172-142:789257:789715 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-172-142:789257:789715 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-173-121:3987796:3988246 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-173-121:3987796:3988246 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-173-121:3987794:3988245 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-173-121:3987794:3988245 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-173-121:3987797:3988251 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-173-121:3987797:3988251 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-173-121:3987795:3988252 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-173-121:3987795:3988252 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-173-121:3987801:3988249 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-173-121:3987801:3988249 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-173-121:3987800:3988248 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-173-121:3987800:3988248 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-173-121:3987801:3988249 [7] NCCL INFO Trees [0] -1/-1/-1->7->6 [1] -1/-1/-1->7->6 [2] 4/-1/-1->7->6 [3] 4/-1/-1->7->6 [4] 4/-1/-1->7->6 [5] 4/-1/-1->7->6 [6] 4/-1/-1->7->3 [7] 4/-1/-1->7->3 [8] -1/-1/-1->7->6 [9] -1/-1/-1->7->6 [10] 4/-1/-1->7->6 [11] 4/-1/-1->7->6 [12] 4/-1/-1->7->6 [13] 4/-1/-1->7->6 [14] 4/3/-1->7->-1 [15] 4/3/-1->7->-1
+ip-26-0-173-121:3987801:3988249 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-142:789256:789719 [1] NCCL INFO NVLS Head  0:  0  4
+ip-26-0-172-142:789256:789719 [1] NCCL INFO NVLS Head  1:  1  5
+ip-26-0-172-142:789256:789719 [1] NCCL INFO NVLS Head  2:  2  6
+ip-26-0-172-142:789256:789719 [1] NCCL INFO NVLS Head  3:  3  7
+ip-26-0-172-142:789256:789719 [1] NCCL INFO Channel 00/16 :    0   1   2   3   4   5   6   7
+ip-26-0-172-142:789256:789719 [1] NCCL INFO Channel 01/16 :    0   3   2   5   4   7   6   1
+ip-26-0-172-142:789256:789719 [1] NCCL INFO Channel 02/16 :    0   3   6   5   4   7   2   1
+ip-26-0-172-142:789256:789719 [1] NCCL INFO Channel 03/16 :    0   2   7   5   4   6   3   1
+ip-26-0-172-142:789256:789719 [1] NCCL INFO Channel 04/16 :    0   1   2   3   4   5   6   7
+ip-26-0-172-142:789256:789719 [1] NCCL INFO Channel 05/16 :    0   3   2   5   4   7   6   1
+ip-26-0-172-142:789256:789719 [1] NCCL INFO Channel 06/16 :    0   3   6   5   4   7   2   1
+ip-26-0-172-142:789256:789719 [1] NCCL INFO Channel 07/16 :    0   2   7   5   4   6   3   1
+ip-26-0-172-142:789256:789719 [1] NCCL INFO Channel 08/16 :    0   1   2   3   4   5   6   7
+ip-26-0-172-142:789256:789719 [1] NCCL INFO Channel 09/16 :    0   3   2   5   4   7   6   1
+ip-26-0-172-142:789256:789719 [1] NCCL INFO Channel 10/16 :    0   3   6   5   4   7   2   1
+ip-26-0-172-142:789256:789719 [1] NCCL INFO Channel 11/16 :    0   2   7   5   4   6   3   1
+ip-26-0-172-142:789256:789719 [1] NCCL INFO Channel 12/16 :    0   1   2   3   4   5   6   7
+ip-26-0-172-142:789256:789719 [1] NCCL INFO Channel 13/16 :    0   3   2   5   4   7   6   1
+ip-26-0-172-142:789256:789719 [1] NCCL INFO Channel 14/16 :    0   3   6   5   4   7   2   1
+ip-26-0-172-142:789262:789722 [7] NCCL INFO NVLS Head  0:  0  4
+ip-26-0-172-142:789256:789719 [1] NCCL INFO Channel 15/16 :    0   2   7   5   4   6   3   1
+ip-26-0-172-142:789262:789722 [7] NCCL INFO NVLS Head  1:  1  5
+ip-26-0-172-142:789262:789722 [7] NCCL INFO NVLS Head  2:  2  6
+ip-26-0-173-121:3987799:3988250 [5] NCCL INFO Trees [0] 7/-1/-1->6->5 [1] 7/-1/-1->6->5 [2] 7/-1/-1->6->5 [3] 7/-1/-1->6->5 [4] 7/-1/-1->6->2 [5] 7/-1/-1->6->2 [6] -1/-1/-1->6->5 [7] -1/-1/-1->6->5 [8] 7/-1/-1->6->5 [9] 7/-1/-1->6->5 [10] 7/-1/-1->6->5 [11] 7/-1/-1->6->5 [12] 7/2/-1->6->-1 [13] 7/2/-1->6->-1 [14] -1/-1/-1->6->5 [15] -1/-1/-1->6->5
+ip-26-0-173-121:3987799:3988250 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-173-121:3987797:3988251 [3] NCCL INFO Trees [0] 6/-1/-1->5->4 [1] 6/-1/-1->5->4 [2] 6/-1/-1->5->1 [3] 6/-1/-1->5->1 [4] -1/-1/-1->5->4 [5] -1/-1/-1->5->4 [6] 6/-1/-1->5->4 [7] 6/-1/-1->5->4 [8] 6/-1/-1->5->4 [9] 6/-1/-1->5->4 [10] 6/1/-1->5->-1 [11] 6/1/-1->5->-1 [12] -1/-1/-1->5->4 [13] -1/-1/-1->5->4 [14] 6/-1/-1->5->4 [15] 6/-1/-1->5->4
+ip-26-0-172-142:789260:789721 [5] NCCL INFO NVLS Head  0:  0  4
+ip-26-0-172-142:789262:789722 [7] NCCL INFO NVLS Head  3:  3  7
+ip-26-0-172-142:789256:789719 [1] NCCL INFO Trees [0] 1/4/-1->0->-1 [1] 1/4/-1->0->-1 [2] -1/-1/-1->0->3 [3] -1/-1/-1->0->3 [4] 1/-1/-1->0->3 [5] 1/-1/-1->0->3 [6] 1/-1/-1->0->3 [7] 1/-1/-1->0->3 [8] 1/-1/-1->0->4 [9] 1/-1/-1->0->4 [10] -1/-1/-1->0->3 [11] -1/-1/-1->0->3 [12] 1/-1/-1->0->3 [13] 1/-1/-1->0->3 [14] 1/-1/-1->0->3 [15] 1/-1/-1->0->3
+ip-26-0-172-142:789260:789721 [5] NCCL INFO NVLS Head  1:  1  5
+ip-26-0-172-142:789256:789719 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-142:789260:789721 [5] NCCL INFO NVLS Head  2:  2  6
+ip-26-0-172-142:789260:789721 [5] NCCL INFO NVLS Head  3:  3  7
+ip-26-0-173-121:3987795:3988252 [1] NCCL INFO Trees [0] 5/-1/-1->4->0 [1] 5/-1/-1->4->0 [2] -1/-1/-1->4->7 [3] -1/-1/-1->4->7 [4] 5/-1/-1->4->7 [5] 5/-1/-1->4->7 [6] 5/-1/-1->4->7 [7] 5/-1/-1->4->7 [8] 5/0/-1->4->-1 [9] 5/0/-1->4->-1 [10] -1/-1/-1->4->7 [11] -1/-1/-1->4->7 [12] 5/-1/-1->4->7 [13] 5/-1/-1->4->7 [14] 5/-1/-1->4->7 [15] 5/-1/-1->4->7
+ip-26-0-173-121:3987797:3988251 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-173-121:3987795:3988252 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-142:789262:789722 [7] NCCL INFO Trees [0] -1/-1/-1->3->2 [1] -1/-1/-1->3->2 [2] 0/-1/-1->3->2 [3] 0/-1/-1->3->2 [4] 0/-1/-1->3->2 [5] 0/-1/-1->3->2 [6] 0/7/-1->3->-1 [7] 0/7/-1->3->-1 [8] -1/-1/-1->3->2 [9] -1/-1/-1->3->2 [10] 0/-1/-1->3->2 [11] 0/-1/-1->3->2 [12] 0/-1/-1->3->2 [13] 0/-1/-1->3->2 [14] 0/-1/-1->3->7 [15] 0/-1/-1->3->7
+ip-26-0-172-142:789262:789722 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-142:789260:789721 [5] NCCL INFO Trees [0] 3/-1/-1->2->1 [1] 3/-1/-1->2->1 [2] 3/-1/-1->2->1 [3] 3/-1/-1->2->1 [4] 3/6/-1->2->-1 [5] 3/6/-1->2->-1 [6] -1/-1/-1->2->1 [7] -1/-1/-1->2->1 [8] 3/-1/-1->2->1 [9] 3/-1/-1->2->1 [10] 3/-1/-1->2->1 [11] 3/-1/-1->2->1 [12] 3/-1/-1->2->6 [13] 3/-1/-1->2->6 [14] -1/-1/-1->2->1 [15] -1/-1/-1->2->1
+ip-26-0-172-142:789260:789721 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-142:789258:789720 [3] NCCL INFO NVLS Head  0:  0  4
+ip-26-0-172-142:789258:789720 [3] NCCL INFO NVLS Head  1:  1  5
+ip-26-0-172-142:789258:789720 [3] NCCL INFO NVLS Head  2:  2  6
+ip-26-0-172-142:789258:789720 [3] NCCL INFO NVLS Head  3:  3  7
+ip-26-0-172-142:789258:789720 [3] NCCL INFO Trees [0] 2/-1/-1->1->0 [1] 2/-1/-1->1->0 [2] 2/5/-1->1->-1 [3] 2/5/-1->1->-1 [4] -1/-1/-1->1->0 [5] -1/-1/-1->1->0 [6] 2/-1/-1->1->0 [7] 2/-1/-1->1->0 [8] 2/-1/-1->1->0 [9] 2/-1/-1->1->0 [10] 2/-1/-1->1->5 [11] 2/-1/-1->1->5 [12] -1/-1/-1->1->0 [13] -1/-1/-1->1->0 [14] 2/-1/-1->1->0 [15] 2/-1/-1->1->0
+ip-26-0-172-142:789258:789720 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-173-121:3987797:3988251 [3] NCCL INFO Channel 00/0 : 5[3] -> 6[5] via P2P/IPC
+ip-26-0-173-121:3987799:3988250 [5] NCCL INFO Channel 00/0 : 6[5] -> 7[7] via P2P/IPC
+ip-26-0-172-142:789260:789721 [5] NCCL INFO Channel 00/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-173-121:3987797:3988251 [3] NCCL INFO Channel 04/0 : 5[3] -> 6[5] via P2P/IPC
+ip-26-0-173-121:3987799:3988250 [5] NCCL INFO Channel 04/0 : 6[5] -> 7[7] via P2P/IPC
+ip-26-0-172-142:789260:789721 [5] NCCL INFO Channel 04/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-173-121:3987797:3988251 [3] NCCL INFO Channel 08/0 : 5[3] -> 6[5] via P2P/IPC
+ip-26-0-173-121:3987799:3988250 [5] NCCL INFO Channel 08/0 : 6[5] -> 7[7] via P2P/IPC
+ip-26-0-172-142:789258:789720 [3] NCCL INFO Channel 00/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-172-142:789260:789721 [5] NCCL INFO Channel 08/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-173-121:3987797:3988251 [3] NCCL INFO Channel 12/0 : 5[3] -> 6[5] via P2P/IPC
+ip-26-0-173-121:3987799:3988250 [5] NCCL INFO Channel 12/0 : 6[5] -> 7[7] via P2P/IPC
+ip-26-0-172-142:789258:789720 [3] NCCL INFO Channel 04/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-172-142:789260:789721 [5] NCCL INFO Channel 12/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-172-142:789258:789720 [3] NCCL INFO Channel 08/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-172-142:789258:789720 [3] NCCL INFO Channel 12/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-173-121:3987800:3988248 [6] NCCL INFO Trees [0] -1/-1/-1->7->6 [1] -1/-1/-1->7->6 [2] 4/-1/-1->7->6 [3] 4/-1/-1->7->6 [4] 4/-1/-1->7->6 [5] 4/-1/-1->7->6 [6] 4/-1/-1->7->3 [7] 4/-1/-1->7->3 [8] -1/-1/-1->7->6 [9] -1/-1/-1->7->6 [10] 4/-1/-1->7->6 [11] 4/-1/-1->7->6 [12] 4/-1/-1->7->6 [13] 4/-1/-1->7->6 [14] 4/3/-1->7->-1 [15] 4/3/-1->7->-1
+ip-26-0-173-121:3987800:3988248 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-142:789255:789714 [0] NCCL INFO NVLS Head  0:  0  4
+ip-26-0-172-142:789255:789714 [0] NCCL INFO NVLS Head  1:  1  5
+ip-26-0-172-142:789259:789717 [4] NCCL INFO NVLS Head  0:  0  4
+ip-26-0-172-142:789255:789714 [0] NCCL INFO NVLS Head  2:  2  6
+ip-26-0-172-142:789255:789714 [0] NCCL INFO NVLS Head  3:  3  7
+ip-26-0-172-142:789259:789717 [4] NCCL INFO NVLS Head  1:  1  5
+ip-26-0-172-142:789259:789717 [4] NCCL INFO NVLS Head  2:  2  6
+ip-26-0-172-142:789259:789717 [4] NCCL INFO NVLS Head  3:  3  7
+ip-26-0-172-142:789255:789714 [0] NCCL INFO Channel 00/16 :    0   1   2   3   4   5   6   7
+ip-26-0-172-142:789255:789714 [0] NCCL INFO Channel 01/16 :    0   3   2   5   4   7   6   1
+ip-26-0-172-142:789255:789714 [0] NCCL INFO Channel 02/16 :    0   3   6   5   4   7   2   1
+ip-26-0-172-142:789261:789716 [6] NCCL INFO NVLS Head  0:  0  4
+ip-26-0-172-142:789255:789714 [0] NCCL INFO Channel 03/16 :    0   2   7   5   4   6   3   1
+ip-26-0-172-142:789255:789714 [0] NCCL INFO Channel 04/16 :    0   1   2   3   4   5   6   7
+ip-26-0-172-142:789259:789717 [4] NCCL INFO Trees [0] 3/-1/-1->2->1 [1] 3/-1/-1->2->1 [2] 3/-1/-1->2->1 [3] 3/-1/-1->2->1 [4] 3/6/-1->2->-1 [5] 3/6/-1->2->-1 [6] -1/-1/-1->2->1 [7] -1/-1/-1->2->1 [8] 3/-1/-1->2->1 [9] 3/-1/-1->2->1 [10] 3/-1/-1->2->1 [11] 3/-1/-1->2->1 [12] 3/-1/-1->2->6 [13] 3/-1/-1->2->6 [14] -1/-1/-1->2->1 [15] -1/-1/-1->2->1
+ip-26-0-172-142:789261:789716 [6] NCCL INFO NVLS Head  1:  1  5
+ip-26-0-172-142:789255:789714 [0] NCCL INFO Channel 05/16 :    0   3   2   5   4   7   6   1
+ip-26-0-172-142:789259:789717 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-173-121:3987796:3988246 [2] NCCL INFO Trees [0] 6/-1/-1->5->4 [1] 6/-1/-1->5->4 [2] 6/-1/-1->5->1 [3] 6/-1/-1->5->1 [4] -1/-1/-1->5->4 [5] -1/-1/-1->5->4 [6] 6/-1/-1->5->4 [7] 6/-1/-1->5->4 [8] 6/-1/-1->5->4 [9] 6/-1/-1->5->4 [10] 6/1/-1->5->-1 [11] 6/1/-1->5->-1 [12] -1/-1/-1->5->4 [13] -1/-1/-1->5->4 [14] 6/-1/-1->5->4 [15] 6/-1/-1->5->4
+ip-26-0-173-121:3987796:3988246 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-173-121:3987798:3988247 [4] NCCL INFO Trees [0] 7/-1/-1->6->5 [1] 7/-1/-1->6->5 [2] 7/-1/-1->6->5 [3] 7/-1/-1->6->5 [4] 7/-1/-1->6->2 [5] 7/-1/-1->6->2 [6] -1/-1/-1->6->5 [7] -1/-1/-1->6->5 [8] 7/-1/-1->6->5 [9] 7/-1/-1->6->5 [10] 7/-1/-1->6->5 [11] 7/-1/-1->6->5 [12] 7/2/-1->6->-1 [13] 7/2/-1->6->-1 [14] -1/-1/-1->6->5 [15] -1/-1/-1->6->5
+ip-26-0-173-121:3987798:3988247 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-142:789261:789716 [6] NCCL INFO NVLS Head  2:  2  6
+ip-26-0-172-142:789255:789714 [0] NCCL INFO Channel 06/16 :    0   3   6   5   4   7   2   1
+ip-26-0-172-142:789261:789716 [6] NCCL INFO NVLS Head  3:  3  7
+ip-26-0-172-142:789255:789714 [0] NCCL INFO Channel 07/16 :    0   2   7   5   4   6   3   1
+ip-26-0-172-142:789255:789714 [0] NCCL INFO Channel 08/16 :    0   1   2   3   4   5   6   7
+ip-26-0-172-142:789255:789714 [0] NCCL INFO Channel 09/16 :    0   3   2   5   4   7   6   1
+ip-26-0-172-142:789255:789714 [0] NCCL INFO Channel 10/16 :    0   3   6   5   4   7   2   1
+ip-26-0-172-142:789257:789715 [2] NCCL INFO NVLS Head  0:  0  4
+ip-26-0-172-142:789255:789714 [0] NCCL INFO Channel 11/16 :    0   2   7   5   4   6   3   1
+ip-26-0-172-142:789255:789714 [0] NCCL INFO Channel 12/16 :    0   1   2   3   4   5   6   7
+ip-26-0-172-142:789257:789715 [2] NCCL INFO NVLS Head  1:  1  5
+ip-26-0-172-142:789261:789716 [6] NCCL INFO Trees [0] -1/-1/-1->3->2 [1] -1/-1/-1->3->2 [2] 0/-1/-1->3->2 [3] 0/-1/-1->3->2 [4] 0/-1/-1->3->2 [5] 0/-1/-1->3->2 [6] 0/7/-1->3->-1 [7] 0/7/-1->3->-1 [8] -1/-1/-1->3->2 [9] -1/-1/-1->3->2 [10] 0/-1/-1->3->2 [11] 0/-1/-1->3->2 [12] 0/-1/-1->3->2 [13] 0/-1/-1->3->2 [14] 0/-1/-1->3->7 [15] 0/-1/-1->3->7
+ip-26-0-172-142:789257:789715 [2] NCCL INFO NVLS Head  2:  2  6
+ip-26-0-172-142:789255:789714 [0] NCCL INFO Channel 13/16 :    0   3   2   5   4   7   6   1
+ip-26-0-172-142:789261:789716 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-142:789257:789715 [2] NCCL INFO NVLS Head  3:  3  7
+ip-26-0-172-142:789255:789714 [0] NCCL INFO Channel 14/16 :    0   3   6   5   4   7   2   1
+ip-26-0-172-142:789255:789714 [0] NCCL INFO Channel 15/16 :    0   2   7   5   4   6   3   1
+ip-26-0-172-142:789255:789714 [0] NCCL INFO Trees [0] 1/4/-1->0->-1 [1] 1/4/-1->0->-1 [2] -1/-1/-1->0->3 [3] -1/-1/-1->0->3 [4] 1/-1/-1->0->3 [5] 1/-1/-1->0->3 [6] 1/-1/-1->0->3 [7] 1/-1/-1->0->3 [8] 1/-1/-1->0->4 [9] 1/-1/-1->0->4 [10] -1/-1/-1->0->3 [11] -1/-1/-1->0->3 [12] 1/-1/-1->0->3 [13] 1/-1/-1->0->3 [14] 1/-1/-1->0->3 [15] 1/-1/-1->0->3
+ip-26-0-172-142:789257:789715 [2] NCCL INFO Trees [0] 2/-1/-1->1->0 [1] 2/-1/-1->1->0 [2] 2/5/-1->1->-1 [3] 2/5/-1->1->-1 [4] -1/-1/-1->1->0 [5] -1/-1/-1->1->0 [6] 2/-1/-1->1->0 [7] 2/-1/-1->1->0 [8] 2/-1/-1->1->0 [9] 2/-1/-1->1->0 [10] 2/-1/-1->1->5 [11] 2/-1/-1->1->5 [12] -1/-1/-1->1->0 [13] -1/-1/-1->1->0 [14] 2/-1/-1->1->0 [15] 2/-1/-1->1->0
+ip-26-0-173-121:3987794:3988245 [0] NCCL INFO Trees [0] 5/-1/-1->4->0 [1] 5/-1/-1->4->0 [2] -1/-1/-1->4->7 [3] -1/-1/-1->4->7 [4] 5/-1/-1->4->7 [5] 5/-1/-1->4->7 [6] 5/-1/-1->4->7 [7] 5/-1/-1->4->7 [8] 5/0/-1->4->-1 [9] 5/0/-1->4->-1 [10] -1/-1/-1->4->7 [11] -1/-1/-1->4->7 [12] 5/-1/-1->4->7 [13] 5/-1/-1->4->7 [14] 5/-1/-1->4->7 [15] 5/-1/-1->4->7
+ip-26-0-173-121:3987794:3988245 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-142:789255:789714 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-142:789257:789715 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-142:789259:789717 [4] NCCL INFO Channel 00/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-173-121:3987796:3988246 [2] NCCL INFO Channel 00/0 : 5[2] -> 6[4] via P2P/IPC
+ip-26-0-172-142:789259:789717 [4] NCCL INFO Channel 04/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-173-121:3987796:3988246 [2] NCCL INFO Channel 04/0 : 5[2] -> 6[4] via P2P/IPC
+ip-26-0-172-142:789257:789715 [2] NCCL INFO Channel 00/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-172-142:789259:789717 [4] NCCL INFO Channel 08/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-173-121:3987798:3988247 [4] NCCL INFO Channel 00/0 : 6[4] -> 7[6] via P2P/IPC
+ip-26-0-173-121:3987796:3988246 [2] NCCL INFO Channel 08/0 : 5[2] -> 6[4] via P2P/IPC
+ip-26-0-172-142:789259:789717 [4] NCCL INFO Channel 12/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-172-142:789257:789715 [2] NCCL INFO Channel 04/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-173-121:3987798:3988247 [4] NCCL INFO Channel 04/0 : 6[4] -> 7[6] via P2P/IPC
+ip-26-0-173-121:3987796:3988246 [2] NCCL INFO Channel 12/0 : 5[2] -> 6[4] via P2P/IPC
+ip-26-0-172-142:789257:789715 [2] NCCL INFO Channel 08/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-173-121:3987798:3988247 [4] NCCL INFO Channel 08/0 : 6[4] -> 7[6] via P2P/IPC
+ip-26-0-172-142:789257:789715 [2] NCCL INFO Channel 12/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-173-121:3987798:3988247 [4] NCCL INFO Channel 12/0 : 6[4] -> 7[6] via P2P/IPC
+ip-26-0-173-121:3987800:3988248 [6] NCCL INFO Channel 00/0 : 7[6] -> 0[0] [send] via NET/Libfabric/0(4)/GDRDMA
+ip-26-0-172-142:789256:789719 [1] NCCL INFO Channel 00/0 : 7[7] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:789256:789719 [1] NCCL INFO Channel 04/0 : 7[7] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:789262:789722 [7] NCCL INFO Channel 00/0 : 3[7] -> 4[1] [send] via NET/Libfabric/1(0)/GDRDMA
+ip-26-0-172-142:789256:789719 [1] NCCL INFO Channel 08/0 : 7[7] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:789262:789722 [7] NCCL INFO Channel 04/0 : 3[7] -> 4[1] [send] via NET/Libfabric/1(0)/GDRDMA
+ip-26-0-172-142:789256:789719 [1] NCCL INFO Channel 12/0 : 7[7] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:789256:789719 [1] NCCL INFO Channel 00/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-172-142:789262:789722 [7] NCCL INFO Channel 08/0 : 3[7] -> 4[1] [send] via NET/Libfabric/1(0)/GDRDMA
+ip-26-0-172-142:789256:789719 [1] NCCL INFO Channel 04/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-172-142:789262:789722 [7] NCCL INFO Channel 12/0 : 3[7] -> 4[1] [send] via NET/Libfabric/1(0)/GDRDMA
+ip-26-0-172-142:789256:789719 [1] NCCL INFO Channel 08/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-172-142:789256:789719 [1] NCCL INFO Channel 12/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-173-121:3987795:3988252 [1] NCCL INFO Channel 00/0 : 3[7] -> 4[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-121:3987795:3988252 [1] NCCL INFO Channel 04/0 : 3[7] -> 4[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-121:3987801:3988249 [7] NCCL INFO Channel 00/0 : 7[7] -> 0[1] [send] via NET/Libfabric/1(4)/GDRDMA
+ip-26-0-173-121:3987795:3988252 [1] NCCL INFO Channel 08/0 : 3[7] -> 4[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-121:3987801:3988249 [7] NCCL INFO Channel 04/0 : 7[7] -> 0[1] [send] via NET/Libfabric/1(4)/GDRDMA
+ip-26-0-173-121:3987795:3988252 [1] NCCL INFO Channel 12/0 : 3[7] -> 4[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-121:3987795:3988252 [1] NCCL INFO Channel 00/0 : 4[1] -> 5[3] via P2P/IPC
+ip-26-0-173-121:3987801:3988249 [7] NCCL INFO Channel 08/0 : 7[7] -> 0[1] [send] via NET/Libfabric/1(4)/GDRDMA
+ip-26-0-173-121:3987795:3988252 [1] NCCL INFO Channel 04/0 : 4[1] -> 5[3] via P2P/IPC
+ip-26-0-173-121:3987801:3988249 [7] NCCL INFO Channel 12/0 : 7[7] -> 0[1] [send] via NET/Libfabric/1(4)/GDRDMA
+ip-26-0-173-121:3987795:3988252 [1] NCCL INFO Channel 08/0 : 4[1] -> 5[3] via P2P/IPC
+ip-26-0-172-142:789256:789719 [1] NCCL INFO Channel 03/0 : 0[1] -> 2[5] via P2P/IPC
+ip-26-0-172-142:789256:789719 [1] NCCL INFO Channel 07/0 : 0[1] -> 2[5] via P2P/IPC
+ip-26-0-173-121:3987795:3988252 [1] NCCL INFO Channel 12/0 : 4[1] -> 5[3] via P2P/IPC
+ip-26-0-172-142:789256:789719 [1] NCCL INFO Channel 11/0 : 0[1] -> 2[5] via P2P/IPC
+ip-26-0-172-142:789256:789719 [1] NCCL INFO Channel 15/0 : 0[1] -> 2[5] via P2P/IPC
+ip-26-0-172-142:789256:789719 [1] NCCL INFO Channel 01/0 : 0[1] -> 3[7] via P2P/IPC
+ip-26-0-173-121:3987795:3988252 [1] NCCL INFO Channel 03/0 : 4[1] -> 6[5] via P2P/IPC
+ip-26-0-172-142:789256:789719 [1] NCCL INFO Channel 02/0 : 0[1] -> 3[7] via P2P/IPC
+ip-26-0-172-142:789256:789719 [1] NCCL INFO Channel 05/0 : 0[1] -> 3[7] via P2P/IPC
+ip-26-0-173-121:3987795:3988252 [1] NCCL INFO Channel 07/0 : 4[1] -> 6[5] via P2P/IPC
+ip-26-0-172-142:789256:789719 [1] NCCL INFO Channel 06/0 : 0[1] -> 3[7] via P2P/IPC
+ip-26-0-173-121:3987795:3988252 [1] NCCL INFO Channel 11/0 : 4[1] -> 6[5] via P2P/IPC
+ip-26-0-172-142:789256:789719 [1] NCCL INFO Channel 09/0 : 0[1] -> 3[7] via P2P/IPC
+ip-26-0-173-121:3987795:3988252 [1] NCCL INFO Channel 15/0 : 4[1] -> 6[5] via P2P/IPC
+ip-26-0-172-142:789256:789719 [1] NCCL INFO Channel 10/0 : 0[1] -> 3[7] via P2P/IPC
+ip-26-0-173-121:3987795:3988252 [1] NCCL INFO Channel 01/0 : 4[1] -> 7[7] via P2P/IPC
+ip-26-0-173-121:3987795:3988252 [1] NCCL INFO Channel 02/0 : 4[1] -> 7[7] via P2P/IPC
+ip-26-0-173-121:3987795:3988252 [1] NCCL INFO Channel 05/0 : 4[1] -> 7[7] via P2P/IPC
+ip-26-0-173-121:3987795:3988252 [1] NCCL INFO Channel 06/0 : 4[1] -> 7[7] via P2P/IPC
+ip-26-0-172-142:789255:789714 [0] NCCL INFO Channel 00/0 : 7[6] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-121:3987795:3988252 [1] NCCL INFO Channel 09/0 : 4[1] -> 7[7] via P2P/IPC
+ip-26-0-172-142:789255:789714 [0] NCCL INFO Channel 04/0 : 7[6] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:789261:789716 [6] NCCL INFO Channel 00/0 : 3[6] -> 4[0] [send] via NET/Libfabric/0(0)/GDRDMA
+ip-26-0-172-142:789255:789714 [0] NCCL INFO Channel 08/0 : 7[6] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:789261:789716 [6] NCCL INFO Channel 04/0 : 3[6] -> 4[0] [send] via NET/Libfabric/0(0)/GDRDMA
+ip-26-0-172-142:789255:789714 [0] NCCL INFO Channel 12/0 : 7[6] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-121:3987795:3988252 [1] NCCL INFO Channel 10/0 : 4[1] -> 7[7] via P2P/IPC
+ip-26-0-172-142:789255:789714 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-172-142:789261:789716 [6] NCCL INFO Channel 08/0 : 3[6] -> 4[0] [send] via NET/Libfabric/0(0)/GDRDMA
+ip-26-0-173-121:3987795:3988252 [1] NCCL INFO Channel 13/0 : 4[1] -> 7[7] via P2P/IPC
+ip-26-0-173-121:3987795:3988252 [1] NCCL INFO Channel 14/0 : 4[1] -> 7[7] via P2P/IPC
+ip-26-0-172-142:789256:789719 [1] NCCL INFO Channel 13/0 : 0[1] -> 3[7] via P2P/IPC
+ip-26-0-172-142:789256:789719 [1] NCCL INFO Channel 14/0 : 0[1] -> 3[7] via P2P/IPC
+ip-26-0-172-142:789255:789714 [0] NCCL INFO Channel 04/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-172-142:789261:789716 [6] NCCL INFO Channel 12/0 : 3[6] -> 4[0] [send] via NET/Libfabric/0(0)/GDRDMA
+ip-26-0-172-142:789255:789714 [0] NCCL INFO Channel 08/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-172-142:789255:789714 [0] NCCL INFO Channel 12/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-173-121:3987794:3988245 [0] NCCL INFO Channel 00/0 : 3[6] -> 4[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-121:3987800:3988248 [6] NCCL INFO Channel 04/0 : 7[6] -> 0[0] [send] via NET/Libfabric/0(4)/GDRDMA
+ip-26-0-173-121:3987794:3988245 [0] NCCL INFO Channel 04/0 : 3[6] -> 4[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-121:3987800:3988248 [6] NCCL INFO Channel 08/0 : 7[6] -> 0[0] [send] via NET/Libfabric/0(4)/GDRDMA
+ip-26-0-173-121:3987794:3988245 [0] NCCL INFO Channel 08/0 : 3[6] -> 4[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-121:3987800:3988248 [6] NCCL INFO Channel 12/0 : 7[6] -> 0[0] [send] via NET/Libfabric/0(4)/GDRDMA
+ip-26-0-173-121:3987794:3988245 [0] NCCL INFO Channel 12/0 : 3[6] -> 4[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-121:3987794:3988245 [0] NCCL INFO Channel 00/0 : 4[0] -> 5[2] via P2P/IPC
+ip-26-0-172-142:789255:789714 [0] NCCL INFO Channel 03/0 : 0[0] -> 2[4] via P2P/IPC
+ip-26-0-172-142:789255:789714 [0] NCCL INFO Channel 07/0 : 0[0] -> 2[4] via P2P/IPC
+ip-26-0-172-142:789255:789714 [0] NCCL INFO Channel 11/0 : 0[0] -> 2[4] via P2P/IPC
+ip-26-0-172-142:789255:789714 [0] NCCL INFO Channel 15/0 : 0[0] -> 2[4] via P2P/IPC
+ip-26-0-172-142:789255:789714 [0] NCCL INFO Channel 01/0 : 0[0] -> 3[6] via P2P/IPC
+ip-26-0-173-121:3987794:3988245 [0] NCCL INFO Channel 04/0 : 4[0] -> 5[2] via P2P/IPC
+ip-26-0-172-142:789255:789714 [0] NCCL INFO Channel 02/0 : 0[0] -> 3[6] via P2P/IPC
+ip-26-0-172-142:789255:789714 [0] NCCL INFO Channel 05/0 : 0[0] -> 3[6] via P2P/IPC
+ip-26-0-173-121:3987794:3988245 [0] NCCL INFO Channel 08/0 : 4[0] -> 5[2] via P2P/IPC
+ip-26-0-172-142:789255:789714 [0] NCCL INFO Channel 06/0 : 0[0] -> 3[6] via P2P/IPC
+ip-26-0-173-121:3987794:3988245 [0] NCCL INFO Channel 12/0 : 4[0] -> 5[2] via P2P/IPC
+ip-26-0-173-121:3987794:3988245 [0] NCCL INFO Channel 03/0 : 4[0] -> 6[4] via P2P/IPC
+ip-26-0-172-142:789255:789714 [0] NCCL INFO Channel 09/0 : 0[0] -> 3[6] via P2P/IPC
+ip-26-0-173-121:3987794:3988245 [0] NCCL INFO Channel 07/0 : 4[0] -> 6[4] via P2P/IPC
+ip-26-0-172-142:789255:789714 [0] NCCL INFO Channel 10/0 : 0[0] -> 3[6] via P2P/IPC
+ip-26-0-173-121:3987794:3988245 [0] NCCL INFO Channel 11/0 : 4[0] -> 6[4] via P2P/IPC
+ip-26-0-172-142:789255:789714 [0] NCCL INFO Channel 13/0 : 0[0] -> 3[6] via P2P/IPC
+ip-26-0-173-121:3987794:3988245 [0] NCCL INFO Channel 15/0 : 4[0] -> 6[4] via P2P/IPC
+ip-26-0-172-142:789255:789714 [0] NCCL INFO Channel 14/0 : 0[0] -> 3[6] via P2P/IPC
+ip-26-0-173-121:3987794:3988245 [0] NCCL INFO Channel 01/0 : 4[0] -> 7[6] via P2P/IPC
+ip-26-0-173-121:3987794:3988245 [0] NCCL INFO Channel 02/0 : 4[0] -> 7[6] via P2P/IPC
+ip-26-0-173-121:3987794:3988245 [0] NCCL INFO Channel 05/0 : 4[0] -> 7[6] via P2P/IPC
+ip-26-0-173-121:3987794:3988245 [0] NCCL INFO Channel 06/0 : 4[0] -> 7[6] via P2P/IPC
+ip-26-0-173-121:3987794:3988245 [0] NCCL INFO Channel 09/0 : 4[0] -> 7[6] via P2P/IPC
+ip-26-0-173-121:3987794:3988245 [0] NCCL INFO Channel 10/0 : 4[0] -> 7[6] via P2P/IPC
+ip-26-0-173-121:3987800:3988248 [6] NCCL INFO Channel 02/0 : 7[6] -> 2[4] [send] via NET/Libfabric/4(6)/GDRDMA
+ip-26-0-173-121:3987794:3988245 [0] NCCL INFO Channel 13/0 : 4[0] -> 7[6] via P2P/IPC
+ip-26-0-173-121:3987794:3988245 [0] NCCL INFO Channel 14/0 : 4[0] -> 7[6] via P2P/IPC
+ip-26-0-172-142:789258:789720 [3] NCCL INFO Channel 01/0 : 6[5] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:789258:789720 [3] NCCL INFO Channel 05/0 : 6[5] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:789258:789720 [3] NCCL INFO Channel 09/0 : 6[5] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:789258:789720 [3] NCCL INFO Channel 13/0 : 6[5] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:789260:789721 [5] NCCL INFO Channel 02/0 : 7[7] -> 2[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:789262:789722 [7] NCCL INFO Channel 02/0 : 3[7] -> 6[5] [send] via NET/Libfabric/5(2)/GDRDMA
+ip-26-0-172-142:789260:789721 [5] NCCL INFO Channel 06/0 : 7[7] -> 2[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:789262:789722 [7] NCCL INFO Channel 06/0 : 3[7] -> 6[5] [send] via NET/Libfabric/5(2)/GDRDMA
+ip-26-0-172-142:789260:789721 [5] NCCL INFO Channel 10/0 : 7[7] -> 2[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:789262:789722 [7] NCCL INFO Channel 10/0 : 3[7] -> 6[5] [send] via NET/Libfabric/5(2)/GDRDMA
+ip-26-0-172-142:789260:789721 [5] NCCL INFO Channel 14/0 : 7[7] -> 2[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:789262:789722 [7] NCCL INFO Channel 14/0 : 3[7] -> 6[5] [send] via NET/Libfabric/5(2)/GDRDMA
+ip-26-0-172-142:789260:789721 [5] NCCL INFO Channel 01/0 : 2[5] -> 5[3] [send] via NET/Libfabric/3(1)/GDRDMA
+ip-26-0-172-142:789260:789721 [5] NCCL INFO Channel 05/0 : 2[5] -> 5[3] [send] via NET/Libfabric/3(1)/GDRDMA
+ip-26-0-172-142:789260:789721 [5] NCCL INFO Channel 09/0 : 2[5] -> 5[3] [send] via NET/Libfabric/3(1)/GDRDMA
+ip-26-0-172-142:789260:789721 [5] NCCL INFO Channel 13/0 : 2[5] -> 5[3] [send] via NET/Libfabric/3(1)/GDRDMA
+ip-26-0-173-121:3987797:3988251 [3] NCCL INFO Channel 01/0 : 2[5] -> 5[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-121:3987797:3988251 [3] NCCL INFO Channel 05/0 : 2[5] -> 5[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-121:3987797:3988251 [3] NCCL INFO Channel 09/0 : 2[5] -> 5[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-121:3987797:3988251 [3] NCCL INFO Channel 13/0 : 2[5] -> 5[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-121:3987799:3988250 [5] NCCL INFO Channel 02/0 : 3[7] -> 6[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-121:3987801:3988249 [7] NCCL INFO Channel 02/0 : 7[7] -> 2[5] [send] via NET/Libfabric/5(6)/GDRDMA
+ip-26-0-173-121:3987799:3988250 [5] NCCL INFO Channel 06/0 : 3[7] -> 6[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-121:3987801:3988249 [7] NCCL INFO Channel 06/0 : 7[7] -> 2[5] [send] via NET/Libfabric/5(6)/GDRDMA
+ip-26-0-173-121:3987799:3988250 [5] NCCL INFO Channel 10/0 : 3[7] -> 6[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-121:3987801:3988249 [7] NCCL INFO Channel 10/0 : 7[7] -> 2[5] [send] via NET/Libfabric/5(6)/GDRDMA
+ip-26-0-173-121:3987799:3988250 [5] NCCL INFO Channel 14/0 : 3[7] -> 6[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-121:3987801:3988249 [7] NCCL INFO Channel 14/0 : 7[7] -> 2[5] [send] via NET/Libfabric/5(6)/GDRDMA
+ip-26-0-173-121:3987799:3988250 [5] NCCL INFO Channel 01/0 : 6[5] -> 1[3] [send] via NET/Libfabric/3(5)/GDRDMA
+ip-26-0-173-121:3987799:3988250 [5] NCCL INFO Channel 05/0 : 6[5] -> 1[3] [send] via NET/Libfabric/3(5)/GDRDMA
+ip-26-0-173-121:3987799:3988250 [5] NCCL INFO Channel 09/0 : 6[5] -> 1[3] [send] via NET/Libfabric/3(5)/GDRDMA
+ip-26-0-173-121:3987799:3988250 [5] NCCL INFO Channel 13/0 : 6[5] -> 1[3] [send] via NET/Libfabric/3(5)/GDRDMA
+ip-26-0-172-142:789257:789715 [2] NCCL INFO Channel 01/0 : 6[4] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:789257:789715 [2] NCCL INFO Channel 05/0 : 6[4] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:789257:789715 [2] NCCL INFO Channel 09/0 : 6[4] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:789257:789715 [2] NCCL INFO Channel 13/0 : 6[4] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:789260:789721 [5] NCCL INFO Channel 03/0 : 2[5] -> 7[7] [send] via NET/Libfabric/7(3)/GDRDMA
+ip-26-0-172-142:789259:789717 [4] NCCL INFO Channel 02/0 : 7[6] -> 2[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:789261:789716 [6] NCCL INFO Channel 02/0 : 3[6] -> 6[4] [send] via NET/Libfabric/4(2)/GDRDMA
+ip-26-0-172-142:789259:789717 [4] NCCL INFO Channel 06/0 : 7[6] -> 2[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:789261:789716 [6] NCCL INFO Channel 06/0 : 3[6] -> 6[4] [send] via NET/Libfabric/4(2)/GDRDMA
+ip-26-0-172-142:789259:789717 [4] NCCL INFO Channel 10/0 : 7[6] -> 2[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:789261:789716 [6] NCCL INFO Channel 10/0 : 3[6] -> 6[4] [send] via NET/Libfabric/4(2)/GDRDMA
+ip-26-0-172-142:789259:789717 [4] NCCL INFO Channel 14/0 : 7[6] -> 2[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:789261:789716 [6] NCCL INFO Channel 14/0 : 3[6] -> 6[4] [send] via NET/Libfabric/4(2)/GDRDMA
+ip-26-0-172-142:789259:789717 [4] NCCL INFO Channel 01/0 : 2[4] -> 5[2] [send] via NET/Libfabric/2(1)/GDRDMA
+ip-26-0-172-142:789259:789717 [4] NCCL INFO Channel 05/0 : 2[4] -> 5[2] [send] via NET/Libfabric/2(1)/GDRDMA
+ip-26-0-172-142:789259:789717 [4] NCCL INFO Channel 09/0 : 2[4] -> 5[2] [send] via NET/Libfabric/2(1)/GDRDMA
+ip-26-0-172-142:789259:789717 [4] NCCL INFO Channel 13/0 : 2[4] -> 5[2] [send] via NET/Libfabric/2(1)/GDRDMA
+ip-26-0-173-121:3987796:3988246 [2] NCCL INFO Channel 01/0 : 2[4] -> 5[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-121:3987796:3988246 [2] NCCL INFO Channel 05/0 : 2[4] -> 5[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-121:3987796:3988246 [2] NCCL INFO Channel 09/0 : 2[4] -> 5[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-121:3987796:3988246 [2] NCCL INFO Channel 13/0 : 2[4] -> 5[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-121:3987798:3988247 [4] NCCL INFO Channel 02/0 : 3[6] -> 6[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-121:3987800:3988248 [6] NCCL INFO Channel 06/0 : 7[6] -> 2[4] [send] via NET/Libfabric/4(6)/GDRDMA
+ip-26-0-173-121:3987798:3988247 [4] NCCL INFO Channel 06/0 : 3[6] -> 6[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-121:3987800:3988248 [6] NCCL INFO Channel 10/0 : 7[6] -> 2[4] [send] via NET/Libfabric/4(6)/GDRDMA
+ip-26-0-173-121:3987798:3988247 [4] NCCL INFO Channel 10/0 : 3[6] -> 6[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-121:3987800:3988248 [6] NCCL INFO Channel 14/0 : 7[6] -> 2[4] [send] via NET/Libfabric/4(6)/GDRDMA
+ip-26-0-173-121:3987798:3988247 [4] NCCL INFO Channel 14/0 : 3[6] -> 6[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-121:3987798:3988247 [4] NCCL INFO Channel 01/0 : 6[4] -> 1[2] [send] via NET/Libfabric/2(5)/GDRDMA
+ip-26-0-173-121:3987798:3988247 [4] NCCL INFO Channel 05/0 : 6[4] -> 1[2] [send] via NET/Libfabric/2(5)/GDRDMA
+ip-26-0-173-121:3987798:3988247 [4] NCCL INFO Channel 09/0 : 6[4] -> 1[2] [send] via NET/Libfabric/2(5)/GDRDMA
+ip-26-0-173-121:3987798:3988247 [4] NCCL INFO Channel 13/0 : 6[4] -> 1[2] [send] via NET/Libfabric/2(5)/GDRDMA
+ip-26-0-172-142:789259:789717 [4] NCCL INFO Channel 03/0 : 2[4] -> 7[6] [send] via NET/Libfabric/6(3)/GDRDMA
+ip-26-0-172-142:789262:789722 [7] NCCL INFO Channel 03/0 : 6[5] -> 3[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:789260:789721 [5] NCCL INFO Channel 07/0 : 2[5] -> 7[7] [send] via NET/Libfabric/7(3)/GDRDMA
+ip-26-0-172-142:789262:789722 [7] NCCL INFO Channel 07/0 : 6[5] -> 3[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:789260:789721 [5] NCCL INFO Channel 11/0 : 2[5] -> 7[7] [send] via NET/Libfabric/7(3)/GDRDMA
+ip-26-0-172-142:789262:789722 [7] NCCL INFO Channel 11/0 : 6[5] -> 3[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:789260:789721 [5] NCCL INFO Channel 15/0 : 2[5] -> 7[7] [send] via NET/Libfabric/7(3)/GDRDMA
+ip-26-0-172-142:789262:789722 [7] NCCL INFO Channel 15/0 : 6[5] -> 3[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-121:3987801:3988249 [7] NCCL INFO Channel 03/0 : 2[5] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-121:3987799:3988250 [5] NCCL INFO Channel 03/0 : 6[5] -> 3[7] [send] via NET/Libfabric/7(7)/GDRDMA
+ip-26-0-173-121:3987801:3988249 [7] NCCL INFO Channel 07/0 : 2[5] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-121:3987799:3988250 [5] NCCL INFO Channel 07/0 : 6[5] -> 3[7] [send] via NET/Libfabric/7(7)/GDRDMA
+ip-26-0-173-121:3987801:3988249 [7] NCCL INFO Channel 11/0 : 2[5] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-121:3987799:3988250 [5] NCCL INFO Channel 11/0 : 6[5] -> 3[7] [send] via NET/Libfabric/7(7)/GDRDMA
+ip-26-0-173-121:3987801:3988249 [7] NCCL INFO Channel 15/0 : 2[5] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-121:3987799:3988250 [5] NCCL INFO Channel 15/0 : 6[5] -> 3[7] [send] via NET/Libfabric/7(7)/GDRDMA
+ip-26-0-173-121:3987801:3988249 [7] NCCL INFO Channel 03/0 : 7[7] -> 5[3] via P2P/IPC
+ip-26-0-172-142:789262:789722 [7] NCCL INFO Channel 03/0 : 3[7] -> 1[3] via P2P/IPC
+ip-26-0-172-142:789262:789722 [7] NCCL INFO Channel 07/0 : 3[7] -> 1[3] via P2P/IPC
+ip-26-0-172-142:789262:789722 [7] NCCL INFO Channel 11/0 : 3[7] -> 1[3] via P2P/IPC
+ip-26-0-173-121:3987801:3988249 [7] NCCL INFO Channel 07/0 : 7[7] -> 5[3] via P2P/IPC
+ip-26-0-172-142:789262:789722 [7] NCCL INFO Channel 15/0 : 3[7] -> 1[3] via P2P/IPC
+ip-26-0-173-121:3987801:3988249 [7] NCCL INFO Channel 11/0 : 7[7] -> 5[3] via P2P/IPC
+ip-26-0-172-142:789260:789721 [5] NCCL INFO Channel 02/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-173-121:3987801:3988249 [7] NCCL INFO Channel 15/0 : 7[7] -> 5[3] via P2P/IPC
+ip-26-0-172-142:789262:789722 [7] NCCL INFO Channel 01/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-172-142:789260:789721 [5] NCCL INFO Channel 06/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-172-142:789262:789722 [7] NCCL INFO Channel 05/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-173-121:3987801:3988249 [7] NCCL INFO Channel 01/0 : 7[7] -> 6[5] via P2P/IPC
+ip-26-0-173-121:3987799:3988250 [5] NCCL INFO Channel 02/0 : 6[5] -> 5[3] via P2P/IPC
+ip-26-0-172-142:789260:789721 [5] NCCL INFO Channel 10/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-172-142:789262:789722 [7] NCCL INFO Channel 09/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-172-142:789260:789721 [5] NCCL INFO Channel 14/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-173-121:3987799:3988250 [5] NCCL INFO Channel 06/0 : 6[5] -> 5[3] via P2P/IPC
+ip-26-0-173-121:3987801:3988249 [7] NCCL INFO Channel 05/0 : 7[7] -> 6[5] via P2P/IPC
+ip-26-0-172-142:789262:789722 [7] NCCL INFO Channel 13/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-173-121:3987799:3988250 [5] NCCL INFO Channel 10/0 : 6[5] -> 5[3] via P2P/IPC
+ip-26-0-173-121:3987801:3988249 [7] NCCL INFO Channel 09/0 : 7[7] -> 6[5] via P2P/IPC
+ip-26-0-173-121:3987799:3988250 [5] NCCL INFO Channel 14/0 : 6[5] -> 5[3] via P2P/IPC
+ip-26-0-173-121:3987801:3988249 [7] NCCL INFO Channel 13/0 : 7[7] -> 6[5] via P2P/IPC
+ip-26-0-173-121:3987797:3988251 [3] NCCL INFO Channel 01/0 : 5[3] -> 4[1] via P2P/IPC
+ip-26-0-172-142:789258:789720 [3] NCCL INFO Channel 01/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-172-142:789258:789720 [3] NCCL INFO Channel 02/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-172-142:789258:789720 [3] NCCL INFO Channel 03/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-172-142:789261:789716 [6] NCCL INFO Channel 03/0 : 6[4] -> 3[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:789259:789717 [4] NCCL INFO Channel 07/0 : 2[4] -> 7[6] [send] via NET/Libfabric/6(3)/GDRDMA
+ip-26-0-172-142:789261:789716 [6] NCCL INFO Channel 07/0 : 6[4] -> 3[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:789259:789717 [4] NCCL INFO Channel 11/0 : 2[4] -> 7[6] [send] via NET/Libfabric/6(3)/GDRDMA
+ip-26-0-172-142:789261:789716 [6] NCCL INFO Channel 11/0 : 6[4] -> 3[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:789259:789717 [4] NCCL INFO Channel 15/0 : 2[4] -> 7[6] [send] via NET/Libfabric/6(3)/GDRDMA
+ip-26-0-172-142:789261:789716 [6] NCCL INFO Channel 15/0 : 6[4] -> 3[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-121:3987797:3988251 [3] NCCL INFO Channel 02/0 : 5[3] -> 4[1] via P2P/IPC
+ip-26-0-173-121:3987800:3988248 [6] NCCL INFO Channel 03/0 : 2[4] -> 7[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-121:3987798:3988247 [4] NCCL INFO Channel 03/0 : 6[4] -> 3[6] [send] via NET/Libfabric/6(7)/GDRDMA
+ip-26-0-173-121:3987800:3988248 [6] NCCL INFO Channel 07/0 : 2[4] -> 7[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-121:3987798:3988247 [4] NCCL INFO Channel 07/0 : 6[4] -> 3[6] [send] via NET/Libfabric/6(7)/GDRDMA
+ip-26-0-173-121:3987800:3988248 [6] NCCL INFO Channel 11/0 : 2[4] -> 7[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-121:3987798:3988247 [4] NCCL INFO Channel 11/0 : 6[4] -> 3[6] [send] via NET/Libfabric/6(7)/GDRDMA
+ip-26-0-172-142:789258:789720 [3] NCCL INFO Channel 05/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-173-121:3987800:3988248 [6] NCCL INFO Channel 15/0 : 2[4] -> 7[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-121:3987798:3988247 [4] NCCL INFO Channel 15/0 : 6[4] -> 3[6] [send] via NET/Libfabric/6(7)/GDRDMA
+ip-26-0-173-121:3987800:3988248 [6] NCCL INFO Channel 03/0 : 7[6] -> 5[2] via P2P/IPC
+ip-26-0-172-142:789261:789716 [6] NCCL INFO Channel 03/0 : 3[6] -> 1[2] via P2P/IPC
+ip-26-0-173-121:3987797:3988251 [3] NCCL INFO Channel 03/0 : 5[3] -> 4[1] via P2P/IPC
+ip-26-0-172-142:789261:789716 [6] NCCL INFO Channel 07/0 : 3[6] -> 1[2] via P2P/IPC
+ip-26-0-172-142:789258:789720 [3] NCCL INFO Channel 06/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-173-121:3987800:3988248 [6] NCCL INFO Channel 07/0 : 7[6] -> 5[2] via P2P/IPC
+ip-26-0-173-121:3987797:3988251 [3] NCCL INFO Channel 05/0 : 5[3] -> 4[1] via P2P/IPC
+ip-26-0-172-142:789261:789716 [6] NCCL INFO Channel 11/0 : 3[6] -> 1[2] via P2P/IPC
+ip-26-0-172-142:789258:789720 [3] NCCL INFO Channel 07/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-173-121:3987800:3988248 [6] NCCL INFO Channel 11/0 : 7[6] -> 5[2] via P2P/IPC
+ip-26-0-173-121:3987797:3988251 [3] NCCL INFO Channel 06/0 : 5[3] -> 4[1] via P2P/IPC
+ip-26-0-172-142:789261:789716 [6] NCCL INFO Channel 15/0 : 3[6] -> 1[2] via P2P/IPC
+ip-26-0-172-142:789258:789720 [3] NCCL INFO Channel 09/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-173-121:3987800:3988248 [6] NCCL INFO Channel 15/0 : 7[6] -> 5[2] via P2P/IPC
+ip-26-0-173-121:3987797:3988251 [3] NCCL INFO Channel 07/0 : 5[3] -> 4[1] via P2P/IPC
+ip-26-0-172-142:789258:789720 [3] NCCL INFO Channel 10/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-172-142:789261:789716 [6] NCCL INFO Channel 01/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-172-142:789259:789717 [4] NCCL INFO Channel 02/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-173-121:3987797:3988251 [3] NCCL INFO Channel 09/0 : 5[3] -> 4[1] via P2P/IPC
+ip-26-0-173-121:3987800:3988248 [6] NCCL INFO Channel 01/0 : 7[6] -> 6[4] via P2P/IPC
+ip-26-0-173-121:3987798:3988247 [4] NCCL INFO Channel 02/0 : 6[4] -> 5[2] via P2P/IPC
+ip-26-0-172-142:789258:789720 [3] NCCL INFO Channel 11/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-172-142:789259:789717 [4] NCCL INFO Channel 06/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-172-142:789261:789716 [6] NCCL INFO Channel 05/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-173-121:3987797:3988251 [3] NCCL INFO Channel 10/0 : 5[3] -> 4[1] via P2P/IPC
+ip-26-0-172-142:789258:789720 [3] NCCL INFO Channel 13/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-173-121:3987798:3988247 [4] NCCL INFO Channel 06/0 : 6[4] -> 5[2] via P2P/IPC
+ip-26-0-173-121:3987800:3988248 [6] NCCL INFO Channel 05/0 : 7[6] -> 6[4] via P2P/IPC
+ip-26-0-172-142:789259:789717 [4] NCCL INFO Channel 10/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-172-142:789261:789716 [6] NCCL INFO Channel 09/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-173-121:3987797:3988251 [3] NCCL INFO Channel 11/0 : 5[3] -> 4[1] via P2P/IPC
+ip-26-0-173-121:3987798:3988247 [4] NCCL INFO Channel 10/0 : 6[4] -> 5[2] via P2P/IPC
+ip-26-0-172-142:789258:789720 [3] NCCL INFO Channel 14/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-172-142:789259:789717 [4] NCCL INFO Channel 14/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-173-121:3987800:3988248 [6] NCCL INFO Channel 09/0 : 7[6] -> 6[4] via P2P/IPC
+ip-26-0-172-142:789261:789716 [6] NCCL INFO Channel 13/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-173-121:3987797:3988251 [3] NCCL INFO Channel 13/0 : 5[3] -> 4[1] via P2P/IPC
+ip-26-0-172-142:789258:789720 [3] NCCL INFO Channel 15/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-173-121:3987798:3988247 [4] NCCL INFO Channel 14/0 : 6[4] -> 5[2] via P2P/IPC
+ip-26-0-172-142:789257:789715 [2] NCCL INFO Channel 01/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987800:3988248 [6] NCCL INFO Channel 13/0 : 7[6] -> 6[4] via P2P/IPC
+ip-26-0-173-121:3987797:3988251 [3] NCCL INFO Channel 14/0 : 5[3] -> 4[1] via P2P/IPC
+ip-26-0-173-121:3987797:3988251 [3] NCCL INFO Channel 15/0 : 5[3] -> 4[1] via P2P/IPC
+ip-26-0-173-121:3987796:3988246 [2] NCCL INFO Channel 01/0 : 5[2] -> 4[0] via P2P/IPC
+ip-26-0-172-142:789257:789715 [2] NCCL INFO Channel 02/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987796:3988246 [2] NCCL INFO Channel 02/0 : 5[2] -> 4[0] via P2P/IPC
+ip-26-0-172-142:789257:789715 [2] NCCL INFO Channel 03/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789257:789715 [2] NCCL INFO Channel 05/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987796:3988246 [2] NCCL INFO Channel 03/0 : 5[2] -> 4[0] via P2P/IPC
+ip-26-0-173-121:3987796:3988246 [2] NCCL INFO Channel 05/0 : 5[2] -> 4[0] via P2P/IPC
+ip-26-0-173-121:3987796:3988246 [2] NCCL INFO Channel 06/0 : 5[2] -> 4[0] via P2P/IPC
+ip-26-0-173-121:3987796:3988246 [2] NCCL INFO Channel 07/0 : 5[2] -> 4[0] via P2P/IPC
+ip-26-0-173-121:3987796:3988246 [2] NCCL INFO Channel 09/0 : 5[2] -> 4[0] via P2P/IPC
+ip-26-0-173-121:3987796:3988246 [2] NCCL INFO Channel 10/0 : 5[2] -> 4[0] via P2P/IPC
+ip-26-0-173-121:3987796:3988246 [2] NCCL INFO Channel 11/0 : 5[2] -> 4[0] via P2P/IPC
+ip-26-0-172-142:789257:789715 [2] NCCL INFO Channel 06/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987796:3988246 [2] NCCL INFO Channel 13/0 : 5[2] -> 4[0] via P2P/IPC
+ip-26-0-172-142:789257:789715 [2] NCCL INFO Channel 07/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987796:3988246 [2] NCCL INFO Channel 14/0 : 5[2] -> 4[0] via P2P/IPC
+ip-26-0-173-121:3987796:3988246 [2] NCCL INFO Channel 15/0 : 5[2] -> 4[0] via P2P/IPC
+ip-26-0-172-142:789257:789715 [2] NCCL INFO Channel 09/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789257:789715 [2] NCCL INFO Channel 10/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789257:789715 [2] NCCL INFO Channel 11/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789257:789715 [2] NCCL INFO Channel 13/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789257:789715 [2] NCCL INFO Channel 14/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789257:789715 [2] NCCL INFO Channel 15/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987797:3988251 [3] NCCL INFO Connected all rings
+ip-26-0-172-142:789256:789719 [1] NCCL INFO Connected all rings
+ip-26-0-172-142:789256:789719 [1] NCCL INFO Channel 01/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-172-142:789256:789719 [1] NCCL INFO Channel 05/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-172-142:789256:789719 [1] NCCL INFO Channel 06/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-172-142:789256:789719 [1] NCCL INFO Channel 07/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-172-142:789255:789714 [0] NCCL INFO Connected all rings
+ip-26-0-172-142:789255:789714 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-172-142:789257:789715 [2] NCCL INFO Connected all rings
+ip-26-0-172-142:789255:789714 [0] NCCL INFO Channel 05/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-172-142:789256:789719 [1] NCCL INFO Channel 09/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-172-142:789256:789719 [1] NCCL INFO Channel 13/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-172-142:789256:789719 [1] NCCL INFO Channel 14/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-172-142:789256:789719 [1] NCCL INFO Channel 15/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-173-121:3987801:3988249 [7] NCCL INFO Connected all rings
+ip-26-0-172-142:789260:789721 [5] NCCL INFO Connected all rings
+ip-26-0-172-142:789255:789714 [0] NCCL INFO Channel 06/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-172-142:789255:789714 [0] NCCL INFO Channel 07/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-172-142:789255:789714 [0] NCCL INFO Channel 09/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-172-142:789255:789714 [0] NCCL INFO Channel 13/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-172-142:789255:789714 [0] NCCL INFO Channel 14/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-172-142:789255:789714 [0] NCCL INFO Channel 15/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-173-121:3987797:3988251 [3] NCCL INFO Channel 01/0 : 5[3] -> 6[5] via P2P/IPC
+ip-26-0-173-121:3987797:3988251 [3] NCCL INFO Channel 02/0 : 5[3] -> 6[5] via P2P/IPC
+ip-26-0-173-121:3987797:3988251 [3] NCCL INFO Channel 03/0 : 5[3] -> 6[5] via P2P/IPC
+ip-26-0-173-121:3987794:3988245 [0] NCCL INFO Connected all rings
+ip-26-0-173-121:3987794:3988245 [0] NCCL INFO Channel 01/0 : 4[0] -> 5[2] via P2P/IPC
+ip-26-0-173-121:3987797:3988251 [3] NCCL INFO Channel 06/0 : 5[3] -> 6[5] via P2P/IPC
+ip-26-0-173-121:3987797:3988251 [3] NCCL INFO Channel 07/0 : 5[3] -> 6[5] via P2P/IPC
+ip-26-0-173-121:3987797:3988251 [3] NCCL INFO Channel 09/0 : 5[3] -> 6[5] via P2P/IPC
+ip-26-0-173-121:3987797:3988251 [3] NCCL INFO Channel 10/0 : 5[3] -> 6[5] via P2P/IPC
+ip-26-0-173-121:3987797:3988251 [3] NCCL INFO Channel 11/0 : 5[3] -> 6[5] via P2P/IPC
+ip-26-0-172-142:789260:789721 [5] NCCL INFO Channel 01/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-173-121:3987794:3988245 [0] NCCL INFO Channel 05/0 : 4[0] -> 5[2] via P2P/IPC
+ip-26-0-173-121:3987797:3988251 [3] NCCL INFO Channel 14/0 : 5[3] -> 6[5] via P2P/IPC
+ip-26-0-173-121:3987794:3988245 [0] NCCL INFO Channel 06/0 : 4[0] -> 5[2] via P2P/IPC
+ip-26-0-172-142:789261:789716 [6] NCCL INFO Connected all rings
+ip-26-0-173-121:3987794:3988245 [0] NCCL INFO Channel 07/0 : 4[0] -> 5[2] via P2P/IPC
+ip-26-0-173-121:3987794:3988245 [0] NCCL INFO Channel 09/0 : 4[0] -> 5[2] via P2P/IPC
+ip-26-0-172-142:789259:789717 [4] NCCL INFO Connected all rings
+ip-26-0-172-142:789260:789721 [5] NCCL INFO Channel 02/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-173-121:3987794:3988245 [0] NCCL INFO Channel 13/0 : 4[0] -> 5[2] via P2P/IPC
+ip-26-0-173-121:3987795:3988252 [1] NCCL INFO Connected all rings
+ip-26-0-173-121:3987795:3988252 [1] NCCL INFO Channel 01/0 : 4[1] -> 5[3] via P2P/IPC
+ip-26-0-173-121:3987794:3988245 [0] NCCL INFO Channel 14/0 : 4[0] -> 5[2] via P2P/IPC
+ip-26-0-172-142:789260:789721 [5] NCCL INFO Channel 03/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-173-121:3987795:3988252 [1] NCCL INFO Channel 05/0 : 4[1] -> 5[3] via P2P/IPC
+ip-26-0-173-121:3987794:3988245 [0] NCCL INFO Channel 15/0 : 4[0] -> 5[2] via P2P/IPC
+ip-26-0-173-121:3987795:3988252 [1] NCCL INFO Channel 06/0 : 4[1] -> 5[3] via P2P/IPC
+ip-26-0-172-142:789260:789721 [5] NCCL INFO Channel 05/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-173-121:3987795:3988252 [1] NCCL INFO Channel 07/0 : 4[1] -> 5[3] via P2P/IPC
+ip-26-0-173-121:3987795:3988252 [1] NCCL INFO Channel 09/0 : 4[1] -> 5[3] via P2P/IPC
+ip-26-0-172-142:789260:789721 [5] NCCL INFO Channel 09/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-173-121:3987795:3988252 [1] NCCL INFO Channel 13/0 : 4[1] -> 5[3] via P2P/IPC
+ip-26-0-172-142:789260:789721 [5] NCCL INFO Channel 10/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-172-142:789257:789715 [2] NCCL INFO Channel 01/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-172-142:789260:789721 [5] NCCL INFO Channel 11/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-172-142:789257:789715 [2] NCCL INFO Channel 02/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-172-142:789257:789715 [2] NCCL INFO Channel 03/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-172-142:789260:789721 [5] NCCL INFO Channel 13/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-173-121:3987795:3988252 [1] NCCL INFO Channel 14/0 : 4[1] -> 5[3] via P2P/IPC
+ip-26-0-172-142:789257:789715 [2] NCCL INFO Channel 06/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-173-121:3987795:3988252 [1] NCCL INFO Channel 15/0 : 4[1] -> 5[3] via P2P/IPC
+ip-26-0-172-142:789257:789715 [2] NCCL INFO Channel 07/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-172-142:789257:789715 [2] NCCL INFO Channel 09/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-172-142:789257:789715 [2] NCCL INFO Channel 10/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-172-142:789259:789717 [4] NCCL INFO Channel 01/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-173-121:3987796:3988246 [2] NCCL INFO Connected all rings
+ip-26-0-172-142:789257:789715 [2] NCCL INFO Channel 11/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-172-142:789259:789717 [4] NCCL INFO Channel 02/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-172-142:789259:789717 [4] NCCL INFO Channel 03/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-172-142:789257:789715 [2] NCCL INFO Channel 14/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-172-142:789259:789717 [4] NCCL INFO Channel 05/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-172-142:789257:789715 [2] NCCL INFO Channel 15/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-172-142:789259:789717 [4] NCCL INFO Channel 09/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-172-142:789255:789714 [0] NCCL INFO Channel 03/0 : 0[0] -> 3[6] via P2P/IPC
+ip-26-0-172-142:789259:789717 [4] NCCL INFO Channel 10/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-172-142:789255:789714 [0] NCCL INFO Channel 04/0 : 0[0] -> 3[6] via P2P/IPC
+ip-26-0-172-142:789259:789717 [4] NCCL INFO Channel 11/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-172-142:789255:789714 [0] NCCL INFO Channel 07/0 : 0[0] -> 3[6] via P2P/IPC
+ip-26-0-173-121:3987798:3988247 [4] NCCL INFO Connected all rings
+ip-26-0-172-142:789259:789717 [4] NCCL INFO Channel 13/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-173-121:3987796:3988246 [2] NCCL INFO Channel 01/0 : 5[2] -> 6[4] via P2P/IPC
+ip-26-0-172-142:789255:789714 [0] NCCL INFO Channel 11/0 : 0[0] -> 3[6] via P2P/IPC
+ip-26-0-173-121:3987797:3988251 [3] NCCL INFO Channel 15/0 : 5[3] -> 6[5] via P2P/IPC
+ip-26-0-172-142:789255:789714 [0] NCCL INFO Channel 12/0 : 0[0] -> 3[6] via P2P/IPC
+ip-26-0-173-121:3987796:3988246 [2] NCCL INFO Channel 02/0 : 5[2] -> 6[4] via P2P/IPC
+ip-26-0-172-142:789257:789715 [2] NCCL INFO Channel 02/0 : 5[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:789257:789715 [2] NCCL INFO Channel 03/0 : 5[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:789259:789717 [4] NCCL INFO Channel 04/0 : 6[4] -> 2[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:789257:789715 [2] NCCL INFO Channel 10/0 : 5[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:789259:789717 [4] NCCL INFO Channel 05/0 : 6[4] -> 2[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:789257:789715 [2] NCCL INFO Channel 11/0 : 5[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:789259:789717 [4] NCCL INFO Channel 12/0 : 6[4] -> 2[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:789257:789715 [2] NCCL INFO Channel 02/0 : 1[2] -> 5[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:789259:789717 [4] NCCL INFO Channel 13/0 : 6[4] -> 2[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:789257:789715 [2] NCCL INFO Channel 03/0 : 1[2] -> 5[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:789259:789717 [4] NCCL INFO Channel 04/0 : 2[4] -> 6[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:789257:789715 [2] NCCL INFO Channel 10/0 : 1[2] -> 5[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:789259:789717 [4] NCCL INFO Channel 05/0 : 2[4] -> 6[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:789257:789715 [2] NCCL INFO Channel 11/0 : 1[2] -> 5[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:789259:789717 [4] NCCL INFO Channel 12/0 : 2[4] -> 6[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:789259:789717 [4] NCCL INFO Channel 13/0 : 2[4] -> 6[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-121:3987799:3988250 [5] NCCL INFO Connected all rings
+ip-26-0-173-121:3987795:3988252 [1] NCCL INFO Channel 03/0 : 4[1] -> 7[7] via P2P/IPC
+ip-26-0-173-121:3987796:3988246 [2] NCCL INFO Channel 03/0 : 5[2] -> 6[4] via P2P/IPC
+ip-26-0-173-121:3987795:3988252 [1] NCCL INFO Channel 04/0 : 4[1] -> 7[7] via P2P/IPC
+ip-26-0-173-121:3987796:3988246 [2] NCCL INFO Channel 06/0 : 5[2] -> 6[4] via P2P/IPC
+ip-26-0-172-142:789255:789714 [0] NCCL INFO Channel 15/0 : 0[0] -> 3[6] via P2P/IPC
+ip-26-0-173-121:3987795:3988252 [1] NCCL INFO Channel 07/0 : 4[1] -> 7[7] via P2P/IPC
+ip-26-0-173-121:3987796:3988246 [2] NCCL INFO Channel 07/0 : 5[2] -> 6[4] via P2P/IPC
+ip-26-0-172-142:789261:789716 [6] NCCL INFO Channel 06/0 : 7[6] -> 3[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:789255:789714 [0] NCCL INFO Channel 00/0 : 4[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:789261:789716 [6] NCCL INFO Channel 07/0 : 7[6] -> 3[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:789255:789714 [0] NCCL INFO Channel 01/0 : 4[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:789261:789716 [6] NCCL INFO Channel 14/0 : 7[6] -> 3[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:789255:789714 [0] NCCL INFO Channel 08/0 : 4[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:789261:789716 [6] NCCL INFO Channel 15/0 : 7[6] -> 3[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:789255:789714 [0] NCCL INFO Channel 09/0 : 4[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:789261:789716 [6] NCCL INFO Channel 06/0 : 3[6] -> 7[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:789255:789714 [0] NCCL INFO Channel 00/0 : 0[0] -> 4[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:789261:789716 [6] NCCL INFO Channel 07/0 : 3[6] -> 7[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:789255:789714 [0] NCCL INFO Channel 01/0 : 0[0] -> 4[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-121:3987795:3988252 [1] NCCL INFO Channel 11/0 : 4[1] -> 7[7] via P2P/IPC
+ip-26-0-172-142:789261:789716 [6] NCCL INFO Channel 14/0 : 3[6] -> 7[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:789255:789714 [0] NCCL INFO Channel 08/0 : 0[0] -> 4[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:789261:789716 [6] NCCL INFO Channel 15/0 : 3[6] -> 7[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:789255:789714 [0] NCCL INFO Channel 09/0 : 0[0] -> 4[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-121:3987796:3988246 [2] NCCL INFO Channel 09/0 : 5[2] -> 6[4] via P2P/IPC
+ip-26-0-172-142:789258:789720 [3] NCCL INFO Connected all rings
+ip-26-0-173-121:3987795:3988252 [1] NCCL INFO Channel 12/0 : 4[1] -> 7[7] via P2P/IPC
+ip-26-0-173-121:3987796:3988246 [2] NCCL INFO Channel 10/0 : 5[2] -> 6[4] via P2P/IPC
+ip-26-0-173-121:3987795:3988252 [1] NCCL INFO Channel 15/0 : 4[1] -> 7[7] via P2P/IPC
+ip-26-0-173-121:3987796:3988246 [2] NCCL INFO Channel 11/0 : 5[2] -> 6[4] via P2P/IPC
+ip-26-0-173-121:3987796:3988246 [2] NCCL INFO Channel 14/0 : 5[2] -> 6[4] via P2P/IPC
+ip-26-0-173-121:3987800:3988248 [6] NCCL INFO Connected all rings
+ip-26-0-172-142:789258:789720 [3] NCCL INFO Channel 01/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-173-121:3987798:3988247 [4] NCCL INFO Channel 01/0 : 6[4] -> 7[6] via P2P/IPC
+ip-26-0-173-121:3987796:3988246 [2] NCCL INFO Channel 15/0 : 5[2] -> 6[4] via P2P/IPC
+ip-26-0-172-142:789258:789720 [3] NCCL INFO Channel 02/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-172-142:789258:789720 [3] NCCL INFO Channel 03/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-173-121:3987798:3988247 [4] NCCL INFO Channel 02/0 : 6[4] -> 7[6] via P2P/IPC
+ip-26-0-172-142:789258:789720 [3] NCCL INFO Channel 06/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-173-121:3987794:3988245 [0] NCCL INFO Channel 03/0 : 4[0] -> 7[6] via P2P/IPC
+ip-26-0-172-142:789258:789720 [3] NCCL INFO Channel 07/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-173-121:3987798:3988247 [4] NCCL INFO Channel 03/0 : 6[4] -> 7[6] via P2P/IPC
+ip-26-0-172-142:789258:789720 [3] NCCL INFO Channel 09/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-172-142:789262:789722 [7] NCCL INFO Connected all rings
+ip-26-0-173-121:3987794:3988245 [0] NCCL INFO Channel 04/0 : 4[0] -> 7[6] via P2P/IPC
+ip-26-0-172-142:789258:789720 [3] NCCL INFO Channel 10/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-173-121:3987799:3988250 [5] NCCL INFO Channel 01/0 : 6[5] -> 7[7] via P2P/IPC
+ip-26-0-173-121:3987798:3988247 [4] NCCL INFO Channel 05/0 : 6[4] -> 7[6] via P2P/IPC
+ip-26-0-172-142:789258:789720 [3] NCCL INFO Channel 11/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-173-121:3987794:3988245 [0] NCCL INFO Channel 07/0 : 4[0] -> 7[6] via P2P/IPC
+ip-26-0-172-142:789258:789720 [3] NCCL INFO Channel 14/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-173-121:3987799:3988250 [5] NCCL INFO Channel 02/0 : 6[5] -> 7[7] via P2P/IPC
+ip-26-0-173-121:3987798:3988247 [4] NCCL INFO Channel 09/0 : 6[4] -> 7[6] via P2P/IPC
+ip-26-0-173-121:3987794:3988245 [0] NCCL INFO Channel 11/0 : 4[0] -> 7[6] via P2P/IPC
+ip-26-0-172-142:789258:789720 [3] NCCL INFO Channel 15/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-173-121:3987799:3988250 [5] NCCL INFO Channel 03/0 : 6[5] -> 7[7] via P2P/IPC
+ip-26-0-173-121:3987798:3988247 [4] NCCL INFO Channel 10/0 : 6[4] -> 7[6] via P2P/IPC
+ip-26-0-172-142:789256:789719 [1] NCCL INFO Channel 03/0 : 0[1] -> 3[7] via P2P/IPC
+ip-26-0-173-121:3987794:3988245 [0] NCCL INFO Channel 12/0 : 4[0] -> 7[6] via P2P/IPC
+ip-26-0-172-142:789258:789720 [3] NCCL INFO Channel 02/0 : 5[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:789258:789720 [3] NCCL INFO Channel 03/0 : 5[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-121:3987799:3988250 [5] NCCL INFO Channel 05/0 : 6[5] -> 7[7] via P2P/IPC
+ip-26-0-172-142:789258:789720 [3] NCCL INFO Channel 10/0 : 5[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:789258:789720 [3] NCCL INFO Channel 11/0 : 5[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:789256:789719 [1] NCCL INFO Channel 04/0 : 0[1] -> 3[7] via P2P/IPC
+ip-26-0-173-121:3987798:3988247 [4] NCCL INFO Channel 11/0 : 6[4] -> 7[6] via P2P/IPC
+ip-26-0-172-142:789258:789720 [3] NCCL INFO Channel 02/0 : 1[3] -> 5[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:789258:789720 [3] NCCL INFO Channel 03/0 : 1[3] -> 5[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-121:3987794:3988245 [0] NCCL INFO Channel 15/0 : 4[0] -> 7[6] via P2P/IPC
+ip-26-0-172-142:789258:789720 [3] NCCL INFO Channel 10/0 : 1[3] -> 5[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:789258:789720 [3] NCCL INFO Channel 11/0 : 1[3] -> 5[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:789256:789719 [1] NCCL INFO Channel 07/0 : 0[1] -> 3[7] via P2P/IPC
+ip-26-0-173-121:3987799:3988250 [5] NCCL INFO Channel 09/0 : 6[5] -> 7[7] via P2P/IPC
+ip-26-0-172-142:789256:789719 [1] NCCL INFO Channel 11/0 : 0[1] -> 3[7] via P2P/IPC
+ip-26-0-173-121:3987798:3988247 [4] NCCL INFO Channel 13/0 : 6[4] -> 7[6] via P2P/IPC
+ip-26-0-172-142:789260:789721 [5] NCCL INFO Channel 04/0 : 6[5] -> 2[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:789260:789721 [5] NCCL INFO Channel 05/0 : 6[5] -> 2[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:789256:789719 [1] NCCL INFO Channel 12/0 : 0[1] -> 3[7] via P2P/IPC
+ip-26-0-172-142:789260:789721 [5] NCCL INFO Channel 12/0 : 6[5] -> 2[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-121:3987799:3988250 [5] NCCL INFO Channel 10/0 : 6[5] -> 7[7] via P2P/IPC
+ip-26-0-172-142:789260:789721 [5] NCCL INFO Channel 13/0 : 6[5] -> 2[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:789260:789721 [5] NCCL INFO Channel 04/0 : 2[5] -> 6[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:789260:789721 [5] NCCL INFO Channel 05/0 : 2[5] -> 6[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:789256:789719 [1] NCCL INFO Channel 15/0 : 0[1] -> 3[7] via P2P/IPC
+ip-26-0-172-142:789260:789721 [5] NCCL INFO Channel 12/0 : 2[5] -> 6[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-121:3987799:3988250 [5] NCCL INFO Channel 11/0 : 6[5] -> 7[7] via P2P/IPC
+ip-26-0-172-142:789260:789721 [5] NCCL INFO Channel 13/0 : 2[5] -> 6[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-121:3987796:3988246 [2] NCCL INFO Channel 02/0 : 1[2] -> 5[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-121:3987796:3988246 [2] NCCL INFO Channel 03/0 : 1[2] -> 5[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-121:3987798:3988247 [4] NCCL INFO Channel 04/0 : 2[4] -> 6[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-121:3987796:3988246 [2] NCCL INFO Channel 10/0 : 1[2] -> 5[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-121:3987798:3988247 [4] NCCL INFO Channel 05/0 : 2[4] -> 6[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-121:3987796:3988246 [2] NCCL INFO Channel 11/0 : 1[2] -> 5[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-121:3987798:3988247 [4] NCCL INFO Channel 12/0 : 2[4] -> 6[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-121:3987799:3988250 [5] NCCL INFO Channel 13/0 : 6[5] -> 7[7] via P2P/IPC
+ip-26-0-173-121:3987796:3988246 [2] NCCL INFO Channel 02/0 : 5[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-121:3987798:3988247 [4] NCCL INFO Channel 13/0 : 2[4] -> 6[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-121:3987796:3988246 [2] NCCL INFO Channel 03/0 : 5[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-121:3987798:3988247 [4] NCCL INFO Channel 04/0 : 6[4] -> 2[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-121:3987796:3988246 [2] NCCL INFO Channel 10/0 : 5[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-121:3987798:3988247 [4] NCCL INFO Channel 05/0 : 6[4] -> 2[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-121:3987796:3988246 [2] NCCL INFO Channel 11/0 : 5[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-121:3987798:3988247 [4] NCCL INFO Channel 12/0 : 6[4] -> 2[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-121:3987798:3988247 [4] NCCL INFO Channel 13/0 : 6[4] -> 2[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-121:3987797:3988251 [3] NCCL INFO Channel 02/0 : 1[3] -> 5[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-121:3987799:3988250 [5] NCCL INFO Channel 04/0 : 2[5] -> 6[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-121:3987797:3988251 [3] NCCL INFO Channel 03/0 : 1[3] -> 5[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-121:3987799:3988250 [5] NCCL INFO Channel 05/0 : 2[5] -> 6[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-121:3987797:3988251 [3] NCCL INFO Channel 10/0 : 1[3] -> 5[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-121:3987799:3988250 [5] NCCL INFO Channel 12/0 : 2[5] -> 6[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:789256:789719 [1] NCCL INFO Channel 00/0 : 4[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:789262:789722 [7] NCCL INFO Channel 06/0 : 7[7] -> 3[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-121:3987797:3988251 [3] NCCL INFO Channel 11/0 : 1[3] -> 5[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-121:3987799:3988250 [5] NCCL INFO Channel 13/0 : 2[5] -> 6[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-121:3987797:3988251 [3] NCCL INFO Channel 02/0 : 5[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:789256:789719 [1] NCCL INFO Channel 01/0 : 4[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-121:3987799:3988250 [5] NCCL INFO Channel 04/0 : 6[5] -> 2[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:789262:789722 [7] NCCL INFO Channel 07/0 : 7[7] -> 3[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-121:3987797:3988251 [3] NCCL INFO Channel 03/0 : 5[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-121:3987799:3988250 [5] NCCL INFO Channel 05/0 : 6[5] -> 2[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:789256:789719 [1] NCCL INFO Channel 08/0 : 4[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:789262:789722 [7] NCCL INFO Channel 14/0 : 7[7] -> 3[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-121:3987797:3988251 [3] NCCL INFO Channel 10/0 : 5[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-121:3987799:3988250 [5] NCCL INFO Channel 12/0 : 6[5] -> 2[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:789256:789719 [1] NCCL INFO Channel 09/0 : 4[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:789262:789722 [7] NCCL INFO Channel 15/0 : 7[7] -> 3[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-121:3987797:3988251 [3] NCCL INFO Channel 11/0 : 5[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-121:3987799:3988250 [5] NCCL INFO Channel 13/0 : 6[5] -> 2[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:789256:789719 [1] NCCL INFO Channel 00/0 : 0[1] -> 4[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:789262:789722 [7] NCCL INFO Channel 06/0 : 3[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:789256:789719 [1] NCCL INFO Channel 01/0 : 0[1] -> 4[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:789262:789722 [7] NCCL INFO Channel 07/0 : 3[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:789256:789719 [1] NCCL INFO Channel 08/0 : 0[1] -> 4[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:789262:789722 [7] NCCL INFO Channel 14/0 : 3[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:789256:789719 [1] NCCL INFO Channel 09/0 : 0[1] -> 4[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:789262:789722 [7] NCCL INFO Channel 15/0 : 3[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-121:3987800:3988248 [6] NCCL INFO Channel 06/0 : 3[6] -> 7[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-121:3987794:3988245 [0] NCCL INFO Channel 00/0 : 0[0] -> 4[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-121:3987800:3988248 [6] NCCL INFO Channel 07/0 : 3[6] -> 7[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-121:3987794:3988245 [0] NCCL INFO Channel 01/0 : 0[0] -> 4[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-121:3987800:3988248 [6] NCCL INFO Channel 14/0 : 3[6] -> 7[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-121:3987794:3988245 [0] NCCL INFO Channel 08/0 : 0[0] -> 4[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-121:3987800:3988248 [6] NCCL INFO Channel 15/0 : 3[6] -> 7[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-121:3987794:3988245 [0] NCCL INFO Channel 09/0 : 0[0] -> 4[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-121:3987800:3988248 [6] NCCL INFO Channel 06/0 : 7[6] -> 3[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-121:3987794:3988245 [0] NCCL INFO Channel 00/0 : 4[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-121:3987800:3988248 [6] NCCL INFO Channel 07/0 : 7[6] -> 3[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-121:3987794:3988245 [0] NCCL INFO Channel 01/0 : 4[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-121:3987800:3988248 [6] NCCL INFO Channel 14/0 : 7[6] -> 3[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-121:3987794:3988245 [0] NCCL INFO Channel 08/0 : 4[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-121:3987800:3988248 [6] NCCL INFO Channel 15/0 : 7[6] -> 3[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-121:3987794:3988245 [0] NCCL INFO Channel 09/0 : 4[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-121:3987800:3988248 [6] NCCL INFO Channel 02/0 : 7[6] -> 4[0] via P2P/IPC
+ip-26-0-172-142:789261:789716 [6] NCCL INFO Channel 02/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789259:789717 [4] NCCL INFO Channel 00/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-172-142:789257:789715 [2] NCCL INFO Channel 00/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789261:789716 [6] NCCL INFO Channel 03/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789259:789717 [4] NCCL INFO Channel 01/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-172-142:789257:789715 [2] NCCL INFO Channel 04/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789261:789716 [6] NCCL INFO Channel 04/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789259:789717 [4] NCCL INFO Channel 03/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-172-142:789257:789715 [2] NCCL INFO Channel 08/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789258:789720 [3] NCCL INFO Channel 00/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-172-142:789261:789716 [6] NCCL INFO Channel 05/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789257:789715 [2] NCCL INFO Channel 12/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789258:789720 [3] NCCL INFO Channel 04/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-173-121:3987801:3988249 [7] NCCL INFO Channel 06/0 : 3[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-121:3987795:3988252 [1] NCCL INFO Channel 00/0 : 0[1] -> 4[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:789260:789721 [5] NCCL INFO Channel 00/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-173-121:3987800:3988248 [6] NCCL INFO Channel 03/0 : 7[6] -> 4[0] via P2P/IPC
+ip-26-0-173-121:3987795:3988252 [1] NCCL INFO Channel 01/0 : 0[1] -> 4[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-121:3987801:3988249 [7] NCCL INFO Channel 07/0 : 3[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-121:3987801:3988249 [7] NCCL INFO Channel 14/0 : 3[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-121:3987795:3988252 [1] NCCL INFO Channel 08/0 : 0[1] -> 4[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-121:3987801:3988249 [7] NCCL INFO Channel 15/0 : 3[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-121:3987795:3988252 [1] NCCL INFO Channel 09/0 : 0[1] -> 4[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-121:3987801:3988249 [7] NCCL INFO Channel 06/0 : 7[7] -> 3[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-121:3987795:3988252 [1] NCCL INFO Channel 00/0 : 4[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-121:3987801:3988249 [7] NCCL INFO Channel 07/0 : 7[7] -> 3[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-121:3987795:3988252 [1] NCCL INFO Channel 01/0 : 4[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-121:3987801:3988249 [7] NCCL INFO Channel 14/0 : 7[7] -> 3[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-121:3987795:3988252 [1] NCCL INFO Channel 08/0 : 4[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-121:3987801:3988249 [7] NCCL INFO Channel 15/0 : 7[7] -> 3[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-121:3987795:3988252 [1] NCCL INFO Channel 09/0 : 4[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-121:3987801:3988249 [7] NCCL INFO Channel 02/0 : 7[7] -> 4[1] via P2P/IPC
+ip-26-0-172-142:789262:789722 [7] NCCL INFO Channel 02/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-173-121:3987800:3988248 [6] NCCL INFO Channel 04/0 : 7[6] -> 4[0] via P2P/IPC
+ip-26-0-172-142:789259:789717 [4] NCCL INFO Channel 07/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-172-142:789261:789716 [6] NCCL INFO Channel 06/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789259:789717 [4] NCCL INFO Channel 08/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-172-142:789261:789716 [6] NCCL INFO Channel 07/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987801:3988249 [7] NCCL INFO Channel 03/0 : 7[7] -> 4[1] via P2P/IPC
+ip-26-0-173-121:3987798:3988247 [4] NCCL INFO Channel 00/0 : 6[4] -> 5[2] via P2P/IPC
+ip-26-0-172-142:789259:789717 [4] NCCL INFO Channel 09/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-172-142:789261:789716 [6] NCCL INFO Channel 10/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987801:3988249 [7] NCCL INFO Channel 04/0 : 7[7] -> 4[1] via P2P/IPC
+ip-26-0-172-142:789260:789721 [5] NCCL INFO Channel 01/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-173-121:3987799:3988250 [5] NCCL INFO Channel 00/0 : 6[5] -> 5[3] via P2P/IPC
+ip-26-0-173-121:3987798:3988247 [4] NCCL INFO Channel 01/0 : 6[4] -> 5[2] via P2P/IPC
+ip-26-0-172-142:789259:789717 [4] NCCL INFO Channel 11/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-172-142:789261:789716 [6] NCCL INFO Channel 11/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987801:3988249 [7] NCCL INFO Channel 05/0 : 7[7] -> 4[1] via P2P/IPC
+ip-26-0-173-121:3987800:3988248 [6] NCCL INFO Channel 05/0 : 7[6] -> 4[0] via P2P/IPC
+ip-26-0-173-121:3987799:3988250 [5] NCCL INFO Channel 01/0 : 6[5] -> 5[3] via P2P/IPC
+ip-26-0-173-121:3987798:3988247 [4] NCCL INFO Channel 03/0 : 6[4] -> 5[2] via P2P/IPC
+ip-26-0-172-142:789262:789722 [7] NCCL INFO Channel 03/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-173-121:3987801:3988249 [7] NCCL INFO Channel 06/0 : 7[7] -> 4[1] via P2P/IPC
+ip-26-0-172-142:789258:789720 [3] NCCL INFO Channel 08/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-173-121:3987800:3988248 [6] NCCL INFO Channel 06/0 : 7[6] -> 4[0] via P2P/IPC
+ip-26-0-172-142:789262:789722 [7] NCCL INFO Channel 04/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-173-121:3987799:3988250 [5] NCCL INFO Channel 03/0 : 6[5] -> 5[3] via P2P/IPC
+ip-26-0-173-121:3987798:3988247 [4] NCCL INFO Channel 07/0 : 6[4] -> 5[2] via P2P/IPC
+ip-26-0-172-142:789258:789720 [3] NCCL INFO Channel 12/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-172-142:789261:789716 [6] NCCL INFO Channel 12/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987801:3988249 [7] NCCL INFO Channel 07/0 : 7[7] -> 4[1] via P2P/IPC
+ip-26-0-172-142:789262:789722 [7] NCCL INFO Channel 05/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-172-142:789260:789721 [5] NCCL INFO Channel 03/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-173-121:3987800:3988248 [6] NCCL INFO Channel 07/0 : 7[6] -> 4[0] via P2P/IPC
+ip-26-0-173-121:3987799:3988250 [5] NCCL INFO Channel 07/0 : 6[5] -> 5[3] via P2P/IPC
+ip-26-0-173-121:3987797:3988251 [3] NCCL INFO Channel 00/0 : 5[3] -> 4[1] via P2P/IPC
+ip-26-0-173-121:3987798:3988247 [4] NCCL INFO Channel 08/0 : 6[4] -> 5[2] via P2P/IPC
+ip-26-0-172-142:789261:789716 [6] NCCL INFO Channel 13/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789262:789722 [7] NCCL INFO Channel 06/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-172-142:789260:789721 [5] NCCL INFO Channel 07/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-173-121:3987801:3988249 [7] NCCL INFO Channel 10/0 : 7[7] -> 4[1] via P2P/IPC
+ip-26-0-173-121:3987800:3988248 [6] NCCL INFO Channel 10/0 : 7[6] -> 4[0] via P2P/IPC
+ip-26-0-172-142:789261:789716 [6] NCCL INFO Channel 14/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987799:3988250 [5] NCCL INFO Channel 08/0 : 6[5] -> 5[3] via P2P/IPC
+ip-26-0-172-142:789262:789722 [7] NCCL INFO Channel 07/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-172-142:789260:789721 [5] NCCL INFO Channel 08/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-173-121:3987797:3988251 [3] NCCL INFO Channel 04/0 : 5[3] -> 4[1] via P2P/IPC
+ip-26-0-173-121:3987798:3988247 [4] NCCL INFO Channel 09/0 : 6[4] -> 5[2] via P2P/IPC
+ip-26-0-172-142:789261:789716 [6] NCCL INFO Channel 15/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987796:3988246 [2] NCCL INFO Channel 00/0 : 5[2] -> 4[0] via P2P/IPC
+ip-26-0-172-142:789262:789722 [7] NCCL INFO Channel 10/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-172-142:789260:789721 [5] NCCL INFO Channel 09/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-173-121:3987800:3988248 [6] NCCL INFO Channel 11/0 : 7[6] -> 4[0] via P2P/IPC
+ip-26-0-172-142:789259:789717 [4] NCCL INFO Channel 15/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-173-121:3987799:3988250 [5] NCCL INFO Channel 09/0 : 6[5] -> 5[3] via P2P/IPC
+ip-26-0-173-121:3987797:3988251 [3] NCCL INFO Channel 08/0 : 5[3] -> 4[1] via P2P/IPC
+ip-26-0-173-121:3987798:3988247 [4] NCCL INFO Channel 11/0 : 6[4] -> 5[2] via P2P/IPC
+ip-26-0-172-142:789262:789722 [7] NCCL INFO Channel 11/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-172-142:789260:789721 [5] NCCL INFO Channel 11/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-173-121:3987796:3988246 [2] NCCL INFO Channel 04/0 : 5[2] -> 4[0] via P2P/IPC
+ip-26-0-173-121:3987800:3988248 [6] NCCL INFO Channel 12/0 : 7[6] -> 4[0] via P2P/IPC
+ip-26-0-172-142:789262:789722 [7] NCCL INFO Channel 12/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-172-142:789260:789721 [5] NCCL INFO Channel 15/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-173-121:3987799:3988250 [5] NCCL INFO Channel 11/0 : 6[5] -> 5[3] via P2P/IPC
+ip-26-0-173-121:3987797:3988251 [3] NCCL INFO Channel 12/0 : 5[3] -> 4[1] via P2P/IPC
+ip-26-0-173-121:3987798:3988247 [4] NCCL INFO Channel 15/0 : 6[4] -> 5[2] via P2P/IPC
+ip-26-0-172-142:789262:789722 [7] NCCL INFO Channel 13/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-172-142:789262:789722 [7] NCCL INFO Channel 14/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-172-142:789261:789716 [6] NCCL INFO Channel 00/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-173-121:3987801:3988249 [7] NCCL INFO Channel 11/0 : 7[7] -> 4[1] via P2P/IPC
+ip-26-0-172-142:789262:789722 [7] NCCL INFO Channel 15/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-172-142:789261:789716 [6] NCCL INFO Channel 02/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-173-121:3987800:3988248 [6] NCCL INFO Channel 13/0 : 7[6] -> 4[0] via P2P/IPC
+ip-26-0-172-142:789261:789716 [6] NCCL INFO Channel 03/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-172-142:789262:789722 [7] NCCL INFO Channel 00/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-173-121:3987796:3988246 [2] NCCL INFO Channel 08/0 : 5[2] -> 4[0] via P2P/IPC
+ip-26-0-173-121:3987800:3988248 [6] NCCL INFO Channel 14/0 : 7[6] -> 4[0] via P2P/IPC
+ip-26-0-172-142:789261:789716 [6] NCCL INFO Channel 04/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-172-142:789262:789722 [7] NCCL INFO Channel 02/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-173-121:3987799:3988250 [5] NCCL INFO Channel 15/0 : 6[5] -> 5[3] via P2P/IPC
+ip-26-0-173-121:3987796:3988246 [2] NCCL INFO Channel 12/0 : 5[2] -> 4[0] via P2P/IPC
+ip-26-0-173-121:3987800:3988248 [6] NCCL INFO Channel 15/0 : 7[6] -> 4[0] via P2P/IPC
+ip-26-0-172-142:789261:789716 [6] NCCL INFO Channel 08/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-172-142:789262:789722 [7] NCCL INFO Channel 03/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-172-142:789261:789716 [6] NCCL INFO Channel 10/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-173-121:3987801:3988249 [7] NCCL INFO Channel 12/0 : 7[7] -> 4[1] via P2P/IPC
+ip-26-0-173-121:3987801:3988249 [7] NCCL INFO Channel 13/0 : 7[7] -> 4[1] via P2P/IPC
+ip-26-0-173-121:3987801:3988249 [7] NCCL INFO Channel 14/0 : 7[7] -> 4[1] via P2P/IPC
+ip-26-0-173-121:3987800:3988248 [6] NCCL INFO Channel 00/0 : 7[6] -> 6[4] via P2P/IPC
+ip-26-0-173-121:3987801:3988249 [7] NCCL INFO Channel 15/0 : 7[7] -> 4[1] via P2P/IPC
+ip-26-0-173-121:3987800:3988248 [6] NCCL INFO Channel 02/0 : 7[6] -> 6[4] via P2P/IPC
+ip-26-0-173-121:3987801:3988249 [7] NCCL INFO Channel 00/0 : 7[7] -> 6[5] via P2P/IPC
+ip-26-0-173-121:3987800:3988248 [6] NCCL INFO Channel 03/0 : 7[6] -> 6[4] via P2P/IPC
+ip-26-0-173-121:3987801:3988249 [7] NCCL INFO Channel 02/0 : 7[7] -> 6[5] via P2P/IPC
+ip-26-0-173-121:3987800:3988248 [6] NCCL INFO Channel 04/0 : 7[6] -> 6[4] via P2P/IPC
+ip-26-0-172-142:789262:789722 [7] NCCL INFO Channel 04/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-173-121:3987801:3988249 [7] NCCL INFO Channel 03/0 : 7[7] -> 6[5] via P2P/IPC
+ip-26-0-173-121:3987800:3988248 [6] NCCL INFO Channel 08/0 : 7[6] -> 6[4] via P2P/IPC
+ip-26-0-173-121:3987801:3988249 [7] NCCL INFO Channel 04/0 : 7[7] -> 6[5] via P2P/IPC
+ip-26-0-173-121:3987800:3988248 [6] NCCL INFO Channel 10/0 : 7[6] -> 6[4] via P2P/IPC
+ip-26-0-172-142:789261:789716 [6] NCCL INFO Channel 11/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-172-142:789262:789722 [7] NCCL INFO Channel 08/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-173-121:3987801:3988249 [7] NCCL INFO Channel 08/0 : 7[7] -> 6[5] via P2P/IPC
+ip-26-0-173-121:3987800:3988248 [6] NCCL INFO Channel 11/0 : 7[6] -> 6[4] via P2P/IPC
+ip-26-0-172-142:789261:789716 [6] NCCL INFO Channel 12/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-173-121:3987801:3988249 [7] NCCL INFO Channel 10/0 : 7[7] -> 6[5] via P2P/IPC
+ip-26-0-173-121:3987800:3988248 [6] NCCL INFO Channel 12/0 : 7[6] -> 6[4] via P2P/IPC
+ip-26-0-172-142:789262:789722 [7] NCCL INFO Channel 10/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-172-142:789262:789722 [7] NCCL INFO Channel 11/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-173-121:3987801:3988249 [7] NCCL INFO Channel 11/0 : 7[7] -> 6[5] via P2P/IPC
+ip-26-0-172-142:789262:789722 [7] NCCL INFO Channel 12/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-173-121:3987801:3988249 [7] NCCL INFO Channel 12/0 : 7[7] -> 6[5] via P2P/IPC
+ip-26-0-172-142:789255:789714 [0] NCCL INFO Connected all trees
+ip-26-0-172-142:789255:789714 [0] NCCL INFO NVLS comm 0xc0e2a70 headRank 0 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-172-142:789260:789721 [5] NCCL INFO Connected all trees
+ip-26-0-172-142:789260:789721 [5] NCCL INFO NVLS comm 0xbfa4730 headRank 2 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-172-142:789256:789719 [1] NCCL INFO Connected all trees
+ip-26-0-172-142:789256:789719 [1] NCCL INFO NVLS comm 0xc430a60 headRank 0 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-172-142:789262:789722 [7] NCCL INFO Connected all trees
+ip-26-0-172-142:789262:789722 [7] NCCL INFO NVLS comm 0xa93d420 headRank 3 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-172-142:789257:789715 [2] NCCL INFO Connected all trees
+ip-26-0-172-142:789257:789715 [2] NCCL INFO NVLS comm 0xb2da4a0 headRank 1 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-172-142:789258:789720 [3] NCCL INFO Connected all trees
+ip-26-0-172-142:789258:789720 [3] NCCL INFO NVLS comm 0xaa5e860 headRank 1 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-173-121:3987795:3988252 [1] NCCL INFO Connected all trees
+ip-26-0-173-121:3987794:3988245 [0] NCCL INFO Connected all trees
+ip-26-0-173-121:3987795:3988252 [1] NCCL INFO NVLS comm 0xb7426f0 headRank 0 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-173-121:3987794:3988245 [0] NCCL INFO NVLS comm 0xab7ed80 headRank 0 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-173-121:3987796:3988246 [2] NCCL INFO Connected all trees
+ip-26-0-173-121:3987801:3988249 [7] NCCL INFO Connected all trees
+ip-26-0-173-121:3987796:3988246 [2] NCCL INFO NVLS comm 0xafda250 headRank 1 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-173-121:3987801:3988249 [7] NCCL INFO NVLS comm 0xab445e0 headRank 3 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-173-121:3987799:3988250 [5] NCCL INFO Connected all trees
+ip-26-0-173-121:3987799:3988250 [5] NCCL INFO NVLS comm 0xb10cba0 headRank 2 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-173-121:3987797:3988251 [3] NCCL INFO Connected all trees
+ip-26-0-173-121:3987797:3988251 [3] NCCL INFO NVLS comm 0xaab04e0 headRank 1 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-172-142:789259:789717 [4] NCCL INFO Connected all trees
+ip-26-0-172-142:789261:789716 [6] NCCL INFO Connected all trees
+ip-26-0-172-142:789259:789717 [4] NCCL INFO NVLS comm 0xaf7eae0 headRank 2 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-172-142:789261:789716 [6] NCCL INFO NVLS comm 0xbd202a0 headRank 3 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-173-121:3987800:3988248 [6] NCCL INFO Connected all trees
+ip-26-0-173-121:3987800:3988248 [6] NCCL INFO NVLS comm 0xa732240 headRank 3 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-173-121:3987798:3988247 [4] NCCL INFO Connected all trees
+ip-26-0-173-121:3987798:3988247 [4] NCCL INFO NVLS comm 0xbb0d530 headRank 2 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-172-142:789256:789719 [1] NCCL INFO Channel 02/0 : 4[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:789260:789721 [5] NCCL INFO Channel 00/0 : 6[5] -> 2[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:789256:789719 [1] NCCL INFO Channel 03/0 : 4[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:789260:789721 [5] NCCL INFO Channel 01/0 : 6[5] -> 2[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:789256:789719 [1] NCCL INFO Channel 04/0 : 4[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:789262:789722 [7] NCCL INFO Channel 00/0 : 7[7] -> 3[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:789260:789721 [5] NCCL INFO Channel 02/0 : 6[5] -> 2[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:789256:789719 [1] NCCL INFO Channel 05/0 : 4[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:789262:789722 [7] NCCL INFO Channel 01/0 : 7[7] -> 3[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:789260:789721 [5] NCCL INFO Channel 03/0 : 6[5] -> 2[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:789256:789719 [1] NCCL INFO Channel 06/0 : 4[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:789262:789722 [7] NCCL INFO Channel 02/0 : 7[7] -> 3[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:789260:789721 [5] NCCL INFO Channel 06/0 : 6[5] -> 2[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:789258:789720 [3] NCCL INFO Channel 00/0 : 5[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:789256:789719 [1] NCCL INFO Channel 07/0 : 4[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:789262:789722 [7] NCCL INFO Channel 03/0 : 7[7] -> 3[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:789258:789720 [3] NCCL INFO Channel 01/0 : 5[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:789260:789721 [5] NCCL INFO Channel 07/0 : 6[5] -> 2[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:789256:789719 [1] NCCL INFO Channel 10/0 : 4[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:789262:789722 [7] NCCL INFO Channel 04/0 : 7[7] -> 3[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:789258:789720 [3] NCCL INFO Channel 04/0 : 5[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:789256:789719 [1] NCCL INFO Channel 11/0 : 4[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:789260:789721 [5] NCCL INFO Channel 08/0 : 6[5] -> 2[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:789262:789722 [7] NCCL INFO Channel 05/0 : 7[7] -> 3[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:789258:789720 [3] NCCL INFO Channel 05/0 : 5[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:789256:789719 [1] NCCL INFO Channel 12/0 : 4[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:789260:789721 [5] NCCL INFO Channel 09/0 : 6[5] -> 2[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:789258:789720 [3] NCCL INFO Channel 06/0 : 5[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:789262:789722 [7] NCCL INFO Channel 08/0 : 7[7] -> 3[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:789256:789719 [1] NCCL INFO Channel 13/0 : 4[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:789260:789721 [5] NCCL INFO Channel 10/0 : 6[5] -> 2[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:789258:789720 [3] NCCL INFO Channel 07/0 : 5[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:789256:789719 [1] NCCL INFO Channel 14/0 : 4[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:789262:789722 [7] NCCL INFO Channel 09/0 : 7[7] -> 3[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:789260:789721 [5] NCCL INFO Channel 11/0 : 6[5] -> 2[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:789258:789720 [3] NCCL INFO Channel 08/0 : 5[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:789256:789719 [1] NCCL INFO Channel 15/0 : 4[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:789262:789722 [7] NCCL INFO Channel 10/0 : 7[7] -> 3[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:789260:789721 [5] NCCL INFO Channel 14/0 : 6[5] -> 2[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:789258:789720 [3] NCCL INFO Channel 09/0 : 5[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:789256:789719 [1] NCCL INFO Channel 02/0 : 0[1] -> 4[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:789262:789722 [7] NCCL INFO Channel 11/0 : 7[7] -> 3[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:789260:789721 [5] NCCL INFO Channel 15/0 : 6[5] -> 2[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:789258:789720 [3] NCCL INFO Channel 12/0 : 5[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:789256:789719 [1] NCCL INFO Channel 03/0 : 0[1] -> 4[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:789262:789722 [7] NCCL INFO Channel 12/0 : 7[7] -> 3[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:789260:789721 [5] NCCL INFO Channel 00/0 : 2[5] -> 6[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:789258:789720 [3] NCCL INFO Channel 13/0 : 5[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:789256:789719 [1] NCCL INFO Channel 04/0 : 0[1] -> 4[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:789262:789722 [7] NCCL INFO Channel 13/0 : 7[7] -> 3[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:789260:789721 [5] NCCL INFO Channel 01/0 : 2[5] -> 6[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:789258:789720 [3] NCCL INFO Channel 14/0 : 5[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:789256:789719 [1] NCCL INFO Channel 05/0 : 0[1] -> 4[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:789262:789722 [7] NCCL INFO Channel 00/0 : 3[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:789260:789721 [5] NCCL INFO Channel 02/0 : 2[5] -> 6[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:789258:789720 [3] NCCL INFO Channel 15/0 : 5[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:789256:789719 [1] NCCL INFO Channel 06/0 : 0[1] -> 4[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:789262:789722 [7] NCCL INFO Channel 01/0 : 3[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:789260:789721 [5] NCCL INFO Channel 03/0 : 2[5] -> 6[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:789258:789720 [3] NCCL INFO Channel 00/0 : 1[3] -> 5[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:789256:789719 [1] NCCL INFO Channel 07/0 : 0[1] -> 4[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:789262:789722 [7] NCCL INFO Channel 02/0 : 3[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:789260:789721 [5] NCCL INFO Channel 06/0 : 2[5] -> 6[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:789258:789720 [3] NCCL INFO Channel 01/0 : 1[3] -> 5[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:789256:789719 [1] NCCL INFO Channel 10/0 : 0[1] -> 4[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:789260:789721 [5] NCCL INFO Channel 07/0 : 2[5] -> 6[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:789262:789722 [7] NCCL INFO Channel 03/0 : 3[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:789258:789720 [3] NCCL INFO Channel 04/0 : 1[3] -> 5[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:789256:789719 [1] NCCL INFO Channel 11/0 : 0[1] -> 4[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:789260:789721 [5] NCCL INFO Channel 08/0 : 2[5] -> 6[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:789262:789722 [7] NCCL INFO Channel 04/0 : 3[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:789258:789720 [3] NCCL INFO Channel 05/0 : 1[3] -> 5[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:789256:789719 [1] NCCL INFO Channel 12/0 : 0[1] -> 4[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:789260:789721 [5] NCCL INFO Channel 09/0 : 2[5] -> 6[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:789262:789722 [7] NCCL INFO Channel 05/0 : 3[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:789258:789720 [3] NCCL INFO Channel 06/0 : 1[3] -> 5[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:789256:789719 [1] NCCL INFO Channel 13/0 : 0[1] -> 4[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:789260:789721 [5] NCCL INFO Channel 10/0 : 2[5] -> 6[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:789262:789722 [7] NCCL INFO Channel 08/0 : 3[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:789256:789719 [1] NCCL INFO Channel 14/0 : 0[1] -> 4[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:789258:789720 [3] NCCL INFO Channel 07/0 : 1[3] -> 5[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:789260:789721 [5] NCCL INFO Channel 11/0 : 2[5] -> 6[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:789262:789722 [7] NCCL INFO Channel 09/0 : 3[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:789256:789719 [1] NCCL INFO Channel 15/0 : 0[1] -> 4[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:789258:789720 [3] NCCL INFO Channel 08/0 : 1[3] -> 5[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:789260:789721 [5] NCCL INFO Channel 14/0 : 2[5] -> 6[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:789262:789722 [7] NCCL INFO Channel 10/0 : 3[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:789258:789720 [3] NCCL INFO Channel 09/0 : 1[3] -> 5[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:789260:789721 [5] NCCL INFO Channel 15/0 : 2[5] -> 6[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:789262:789722 [7] NCCL INFO Channel 11/0 : 3[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:789258:789720 [3] NCCL INFO Channel 12/0 : 1[3] -> 5[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:789262:789722 [7] NCCL INFO Channel 12/0 : 3[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:789258:789720 [3] NCCL INFO Channel 13/0 : 1[3] -> 5[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:789262:789722 [7] NCCL INFO Channel 13/0 : 3[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:789258:789720 [3] NCCL INFO Channel 14/0 : 1[3] -> 5[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:789258:789720 [3] NCCL INFO Channel 15/0 : 1[3] -> 5[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-121:3987801:3988249 [7] NCCL INFO Channel 00/0 : 3[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-121:3987797:3988251 [3] NCCL INFO Channel 00/0 : 1[3] -> 5[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-121:3987797:3988251 [3] NCCL INFO Channel 01/0 : 1[3] -> 5[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-121:3987801:3988249 [7] NCCL INFO Channel 01/0 : 3[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-121:3987795:3988252 [1] NCCL INFO Channel 02/0 : 0[1] -> 4[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-121:3987799:3988250 [5] NCCL INFO Channel 00/0 : 2[5] -> 6[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-121:3987801:3988249 [7] NCCL INFO Channel 02/0 : 3[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-121:3987797:3988251 [3] NCCL INFO Channel 04/0 : 1[3] -> 5[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-121:3987795:3988252 [1] NCCL INFO Channel 03/0 : 0[1] -> 4[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-121:3987799:3988250 [5] NCCL INFO Channel 01/0 : 2[5] -> 6[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-121:3987801:3988249 [7] NCCL INFO Channel 03/0 : 3[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-121:3987797:3988251 [3] NCCL INFO Channel 05/0 : 1[3] -> 5[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-121:3987795:3988252 [1] NCCL INFO Channel 04/0 : 0[1] -> 4[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-121:3987799:3988250 [5] NCCL INFO Channel 02/0 : 2[5] -> 6[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-121:3987801:3988249 [7] NCCL INFO Channel 04/0 : 3[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-121:3987797:3988251 [3] NCCL INFO Channel 06/0 : 1[3] -> 5[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-121:3987795:3988252 [1] NCCL INFO Channel 05/0 : 0[1] -> 4[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-121:3987799:3988250 [5] NCCL INFO Channel 03/0 : 2[5] -> 6[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-121:3987797:3988251 [3] NCCL INFO Channel 07/0 : 1[3] -> 5[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-121:3987801:3988249 [7] NCCL INFO Channel 05/0 : 3[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-121:3987799:3988250 [5] NCCL INFO Channel 06/0 : 2[5] -> 6[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-121:3987795:3988252 [1] NCCL INFO Channel 06/0 : 0[1] -> 4[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-121:3987797:3988251 [3] NCCL INFO Channel 08/0 : 1[3] -> 5[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-121:3987801:3988249 [7] NCCL INFO Channel 08/0 : 3[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-121:3987795:3988252 [1] NCCL INFO Channel 07/0 : 0[1] -> 4[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-121:3987799:3988250 [5] NCCL INFO Channel 07/0 : 2[5] -> 6[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-121:3987797:3988251 [3] NCCL INFO Channel 09/0 : 1[3] -> 5[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-121:3987795:3988252 [1] NCCL INFO Channel 10/0 : 0[1] -> 4[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-121:3987801:3988249 [7] NCCL INFO Channel 09/0 : 3[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-121:3987799:3988250 [5] NCCL INFO Channel 08/0 : 2[5] -> 6[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-121:3987797:3988251 [3] NCCL INFO Channel 12/0 : 1[3] -> 5[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-121:3987795:3988252 [1] NCCL INFO Channel 11/0 : 0[1] -> 4[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-121:3987801:3988249 [7] NCCL INFO Channel 10/0 : 3[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-121:3987799:3988250 [5] NCCL INFO Channel 09/0 : 2[5] -> 6[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-121:3987797:3988251 [3] NCCL INFO Channel 13/0 : 1[3] -> 5[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-121:3987795:3988252 [1] NCCL INFO Channel 12/0 : 0[1] -> 4[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-121:3987799:3988250 [5] NCCL INFO Channel 10/0 : 2[5] -> 6[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-121:3987801:3988249 [7] NCCL INFO Channel 11/0 : 3[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-121:3987797:3988251 [3] NCCL INFO Channel 14/0 : 1[3] -> 5[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-121:3987795:3988252 [1] NCCL INFO Channel 13/0 : 0[1] -> 4[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-121:3987799:3988250 [5] NCCL INFO Channel 11/0 : 2[5] -> 6[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-121:3987801:3988249 [7] NCCL INFO Channel 12/0 : 3[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-121:3987795:3988252 [1] NCCL INFO Channel 14/0 : 0[1] -> 4[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-121:3987797:3988251 [3] NCCL INFO Channel 15/0 : 1[3] -> 5[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-121:3987799:3988250 [5] NCCL INFO Channel 14/0 : 2[5] -> 6[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-121:3987795:3988252 [1] NCCL INFO Channel 15/0 : 0[1] -> 4[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-121:3987801:3988249 [7] NCCL INFO Channel 13/0 : 3[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-121:3987797:3988251 [3] NCCL INFO Channel 00/0 : 5[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-121:3987799:3988250 [5] NCCL INFO Channel 15/0 : 2[5] -> 6[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-121:3987797:3988251 [3] NCCL INFO Channel 01/0 : 5[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-121:3987795:3988252 [1] NCCL INFO Channel 02/0 : 4[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-121:3987801:3988249 [7] NCCL INFO Channel 00/0 : 7[7] -> 3[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-121:3987799:3988250 [5] NCCL INFO Channel 00/0 : 6[5] -> 2[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-121:3987795:3988252 [1] NCCL INFO Channel 03/0 : 4[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-121:3987799:3988250 [5] NCCL INFO Channel 01/0 : 6[5] -> 2[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-121:3987801:3988249 [7] NCCL INFO Channel 01/0 : 7[7] -> 3[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-121:3987797:3988251 [3] NCCL INFO Channel 04/0 : 5[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-121:3987795:3988252 [1] NCCL INFO Channel 04/0 : 4[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-121:3987801:3988249 [7] NCCL INFO Channel 02/0 : 7[7] -> 3[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-121:3987797:3988251 [3] NCCL INFO Channel 05/0 : 5[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-121:3987799:3988250 [5] NCCL INFO Channel 02/0 : 6[5] -> 2[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-121:3987795:3988252 [1] NCCL INFO Channel 05/0 : 4[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-121:3987801:3988249 [7] NCCL INFO Channel 03/0 : 7[7] -> 3[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-121:3987797:3988251 [3] NCCL INFO Channel 06/0 : 5[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-121:3987795:3988252 [1] NCCL INFO Channel 06/0 : 4[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-121:3987799:3988250 [5] NCCL INFO Channel 03/0 : 6[5] -> 2[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-121:3987801:3988249 [7] NCCL INFO Channel 04/0 : 7[7] -> 3[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-121:3987797:3988251 [3] NCCL INFO Channel 07/0 : 5[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-121:3987795:3988252 [1] NCCL INFO Channel 07/0 : 4[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-121:3987799:3988250 [5] NCCL INFO Channel 06/0 : 6[5] -> 2[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-121:3987801:3988249 [7] NCCL INFO Channel 05/0 : 7[7] -> 3[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-121:3987795:3988252 [1] NCCL INFO Channel 10/0 : 4[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-121:3987797:3988251 [3] NCCL INFO Channel 08/0 : 5[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-121:3987799:3988250 [5] NCCL INFO Channel 07/0 : 6[5] -> 2[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-121:3987801:3988249 [7] NCCL INFO Channel 08/0 : 7[7] -> 3[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-121:3987795:3988252 [1] NCCL INFO Channel 11/0 : 4[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-121:3987797:3988251 [3] NCCL INFO Channel 09/0 : 5[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-121:3987799:3988250 [5] NCCL INFO Channel 08/0 : 6[5] -> 2[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-121:3987801:3988249 [7] NCCL INFO Channel 09/0 : 7[7] -> 3[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-121:3987795:3988252 [1] NCCL INFO Channel 12/0 : 4[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-121:3987797:3988251 [3] NCCL INFO Channel 12/0 : 5[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-121:3987799:3988250 [5] NCCL INFO Channel 09/0 : 6[5] -> 2[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-121:3987801:3988249 [7] NCCL INFO Channel 10/0 : 7[7] -> 3[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-121:3987795:3988252 [1] NCCL INFO Channel 13/0 : 4[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-121:3987797:3988251 [3] NCCL INFO Channel 13/0 : 5[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-121:3987801:3988249 [7] NCCL INFO Channel 11/0 : 7[7] -> 3[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-121:3987795:3988252 [1] NCCL INFO Channel 14/0 : 4[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-121:3987799:3988250 [5] NCCL INFO Channel 10/0 : 6[5] -> 2[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-121:3987797:3988251 [3] NCCL INFO Channel 14/0 : 5[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-121:3987801:3988249 [7] NCCL INFO Channel 12/0 : 7[7] -> 3[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-121:3987795:3988252 [1] NCCL INFO Channel 15/0 : 4[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-121:3987799:3988250 [5] NCCL INFO Channel 11/0 : 6[5] -> 2[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-121:3987797:3988251 [3] NCCL INFO Channel 15/0 : 5[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-121:3987801:3988249 [7] NCCL INFO Channel 13/0 : 7[7] -> 3[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-121:3987799:3988250 [5] NCCL INFO Channel 14/0 : 6[5] -> 2[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-121:3987799:3988250 [5] NCCL INFO Channel 15/0 : 6[5] -> 2[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:789257:789715 [2] NCCL INFO Channel 00/0 : 5[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:789259:789717 [4] NCCL INFO Channel 00/0 : 6[4] -> 2[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:789261:789716 [6] NCCL INFO Channel 00/0 : 7[6] -> 3[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:789257:789715 [2] NCCL INFO Channel 01/0 : 5[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:789255:789714 [0] NCCL INFO Channel 02/0 : 4[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:789259:789717 [4] NCCL INFO Channel 01/0 : 6[4] -> 2[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:789257:789715 [2] NCCL INFO Channel 04/0 : 5[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:789261:789716 [6] NCCL INFO Channel 01/0 : 7[6] -> 3[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:789255:789714 [0] NCCL INFO Channel 03/0 : 4[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:789259:789717 [4] NCCL INFO Channel 02/0 : 6[4] -> 2[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:789257:789715 [2] NCCL INFO Channel 05/0 : 5[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:789261:789716 [6] NCCL INFO Channel 02/0 : 7[6] -> 3[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:789255:789714 [0] NCCL INFO Channel 04/0 : 4[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:789259:789717 [4] NCCL INFO Channel 03/0 : 6[4] -> 2[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:789257:789715 [2] NCCL INFO Channel 06/0 : 5[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:789261:789716 [6] NCCL INFO Channel 03/0 : 7[6] -> 3[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:789255:789714 [0] NCCL INFO Channel 05/0 : 4[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:789259:789717 [4] NCCL INFO Channel 06/0 : 6[4] -> 2[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:789261:789716 [6] NCCL INFO Channel 04/0 : 7[6] -> 3[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:789257:789715 [2] NCCL INFO Channel 07/0 : 5[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:789255:789714 [0] NCCL INFO Channel 06/0 : 4[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:789259:789717 [4] NCCL INFO Channel 07/0 : 6[4] -> 2[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:789261:789716 [6] NCCL INFO Channel 05/0 : 7[6] -> 3[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:789257:789715 [2] NCCL INFO Channel 08/0 : 5[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:789255:789714 [0] NCCL INFO Channel 07/0 : 4[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:789259:789717 [4] NCCL INFO Channel 08/0 : 6[4] -> 2[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:789261:789716 [6] NCCL INFO Channel 08/0 : 7[6] -> 3[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:789257:789715 [2] NCCL INFO Channel 09/0 : 5[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:789259:789717 [4] NCCL INFO Channel 09/0 : 6[4] -> 2[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:789255:789714 [0] NCCL INFO Channel 10/0 : 4[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:789261:789716 [6] NCCL INFO Channel 09/0 : 7[6] -> 3[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:789257:789715 [2] NCCL INFO Channel 12/0 : 5[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:789255:789714 [0] NCCL INFO Channel 11/0 : 4[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:789257:789715 [2] NCCL INFO Channel 13/0 : 5[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:789261:789716 [6] NCCL INFO Channel 10/0 : 7[6] -> 3[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:789259:789717 [4] NCCL INFO Channel 10/0 : 6[4] -> 2[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:789255:789714 [0] NCCL INFO Channel 12/0 : 4[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-121:3987794:3988245 [0] NCCL INFO Channel 02/0 : 0[0] -> 4[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:789257:789715 [2] NCCL INFO Channel 14/0 : 5[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:789261:789716 [6] NCCL INFO Channel 11/0 : 7[6] -> 3[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:789259:789717 [4] NCCL INFO Channel 11/0 : 6[4] -> 2[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:789255:789714 [0] NCCL INFO Channel 13/0 : 4[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-121:3987800:3988248 [6] NCCL INFO Channel 00/0 : 3[6] -> 7[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:789257:789715 [2] NCCL INFO Channel 15/0 : 5[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:789261:789716 [6] NCCL INFO Channel 12/0 : 7[6] -> 3[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-121:3987794:3988245 [0] NCCL INFO Channel 03/0 : 0[0] -> 4[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-121:3987800:3988248 [6] NCCL INFO Channel 01/0 : 3[6] -> 7[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:789259:789717 [4] NCCL INFO Channel 14/0 : 6[4] -> 2[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:789255:789714 [0] NCCL INFO Channel 14/0 : 4[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:789257:789715 [2] NCCL INFO Channel 00/0 : 1[2] -> 5[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-121:3987794:3988245 [0] NCCL INFO Channel 04/0 : 0[0] -> 4[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:789261:789716 [6] NCCL INFO Channel 13/0 : 7[6] -> 3[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-121:3987800:3988248 [6] NCCL INFO Channel 02/0 : 3[6] -> 7[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:789259:789717 [4] NCCL INFO Channel 15/0 : 6[4] -> 2[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:789255:789714 [0] NCCL INFO Channel 15/0 : 4[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:789257:789715 [2] NCCL INFO Channel 01/0 : 1[2] -> 5[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-121:3987794:3988245 [0] NCCL INFO Channel 05/0 : 0[0] -> 4[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-121:3987796:3988246 [2] NCCL INFO Channel 00/0 : 1[2] -> 5[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:789261:789716 [6] NCCL INFO Channel 00/0 : 3[6] -> 7[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-121:3987800:3988248 [6] NCCL INFO Channel 03/0 : 3[6] -> 7[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:789259:789717 [4] NCCL INFO Channel 00/0 : 2[4] -> 6[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:789255:789714 [0] NCCL INFO Channel 02/0 : 0[0] -> 4[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:789257:789715 [2] NCCL INFO Channel 04/0 : 1[2] -> 5[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:789261:789716 [6] NCCL INFO Channel 01/0 : 3[6] -> 7[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-121:3987794:3988245 [0] NCCL INFO Channel 06/0 : 0[0] -> 4[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-121:3987798:3988247 [4] NCCL INFO Channel 00/0 : 2[4] -> 6[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-121:3987796:3988246 [2] NCCL INFO Channel 01/0 : 1[2] -> 5[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-121:3987800:3988248 [6] NCCL INFO Channel 04/0 : 3[6] -> 7[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:789259:789717 [4] NCCL INFO Channel 01/0 : 2[4] -> 6[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:789255:789714 [0] NCCL INFO Channel 03/0 : 0[0] -> 4[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:789257:789715 [2] NCCL INFO Channel 05/0 : 1[2] -> 5[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:789261:789716 [6] NCCL INFO Channel 02/0 : 3[6] -> 7[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-121:3987794:3988245 [0] NCCL INFO Channel 07/0 : 0[0] -> 4[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-121:3987798:3988247 [4] NCCL INFO Channel 01/0 : 2[4] -> 6[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:789259:789717 [4] NCCL INFO Channel 02/0 : 2[4] -> 6[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:789255:789714 [0] NCCL INFO Channel 04/0 : 0[0] -> 4[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:789257:789715 [2] NCCL INFO Channel 06/0 : 1[2] -> 5[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:789261:789716 [6] NCCL INFO Channel 03/0 : 3[6] -> 7[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-121:3987800:3988248 [6] NCCL INFO Channel 05/0 : 3[6] -> 7[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-121:3987794:3988245 [0] NCCL INFO Channel 10/0 : 0[0] -> 4[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-121:3987796:3988246 [2] NCCL INFO Channel 04/0 : 1[2] -> 5[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:789259:789717 [4] NCCL INFO Channel 03/0 : 2[4] -> 6[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-121:3987798:3988247 [4] NCCL INFO Channel 02/0 : 2[4] -> 6[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:789255:789714 [0] NCCL INFO Channel 05/0 : 0[0] -> 4[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:789257:789715 [2] NCCL INFO Channel 07/0 : 1[2] -> 5[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:789261:789716 [6] NCCL INFO Channel 04/0 : 3[6] -> 7[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-121:3987800:3988248 [6] NCCL INFO Channel 08/0 : 3[6] -> 7[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:789259:789717 [4] NCCL INFO Channel 06/0 : 2[4] -> 6[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-121:3987794:3988245 [0] NCCL INFO Channel 11/0 : 0[0] -> 4[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:789255:789714 [0] NCCL INFO Channel 06/0 : 0[0] -> 4[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-121:3987796:3988246 [2] NCCL INFO Channel 05/0 : 1[2] -> 5[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-121:3987798:3988247 [4] NCCL INFO Channel 03/0 : 2[4] -> 6[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:789257:789715 [2] NCCL INFO Channel 08/0 : 1[2] -> 5[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:789261:789716 [6] NCCL INFO Channel 05/0 : 3[6] -> 7[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:789259:789717 [4] NCCL INFO Channel 07/0 : 2[4] -> 6[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:789255:789714 [0] NCCL INFO Channel 07/0 : 0[0] -> 4[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-121:3987800:3988248 [6] NCCL INFO Channel 09/0 : 3[6] -> 7[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-121:3987794:3988245 [0] NCCL INFO Channel 12/0 : 0[0] -> 4[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-121:3987796:3988246 [2] NCCL INFO Channel 06/0 : 1[2] -> 5[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-121:3987798:3988247 [4] NCCL INFO Channel 06/0 : 2[4] -> 6[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:789257:789715 [2] NCCL INFO Channel 09/0 : 1[2] -> 5[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:789261:789716 [6] NCCL INFO Channel 08/0 : 3[6] -> 7[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-121:3987800:3988248 [6] NCCL INFO Channel 10/0 : 3[6] -> 7[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-121:3987794:3988245 [0] NCCL INFO Channel 13/0 : 0[0] -> 4[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-121:3987796:3988246 [2] NCCL INFO Channel 07/0 : 1[2] -> 5[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:789255:789714 [0] NCCL INFO Channel 10/0 : 0[0] -> 4[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:789259:789717 [4] NCCL INFO Channel 08/0 : 2[4] -> 6[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-121:3987798:3988247 [4] NCCL INFO Channel 07/0 : 2[4] -> 6[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:789257:789715 [2] NCCL INFO Channel 12/0 : 1[2] -> 5[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:789261:789716 [6] NCCL INFO Channel 09/0 : 3[6] -> 7[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-121:3987794:3988245 [0] NCCL INFO Channel 14/0 : 0[0] -> 4[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-121:3987800:3988248 [6] NCCL INFO Channel 11/0 : 3[6] -> 7[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:789255:789714 [0] NCCL INFO Channel 11/0 : 0[0] -> 4[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:789259:789717 [4] NCCL INFO Channel 09/0 : 2[4] -> 6[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-121:3987796:3988246 [2] NCCL INFO Channel 08/0 : 1[2] -> 5[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-121:3987798:3988247 [4] NCCL INFO Channel 08/0 : 2[4] -> 6[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:789257:789715 [2] NCCL INFO Channel 13/0 : 1[2] -> 5[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:789261:789716 [6] NCCL INFO Channel 10/0 : 3[6] -> 7[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-121:3987794:3988245 [0] NCCL INFO Channel 15/0 : 0[0] -> 4[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-121:3987800:3988248 [6] NCCL INFO Channel 12/0 : 3[6] -> 7[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:789255:789714 [0] NCCL INFO Channel 12/0 : 0[0] -> 4[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:789259:789717 [4] NCCL INFO Channel 10/0 : 2[4] -> 6[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:789257:789715 [2] NCCL INFO Channel 14/0 : 1[2] -> 5[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-121:3987798:3988247 [4] NCCL INFO Channel 09/0 : 2[4] -> 6[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:789261:789716 [6] NCCL INFO Channel 11/0 : 3[6] -> 7[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-121:3987796:3988246 [2] NCCL INFO Channel 09/0 : 1[2] -> 5[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-121:3987794:3988245 [0] NCCL INFO Channel 02/0 : 4[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-121:3987800:3988248 [6] NCCL INFO Channel 13/0 : 3[6] -> 7[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-121:3987798:3988247 [4] NCCL INFO Channel 10/0 : 2[4] -> 6[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-121:3987796:3988246 [2] NCCL INFO Channel 12/0 : 1[2] -> 5[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:789259:789717 [4] NCCL INFO Channel 11/0 : 2[4] -> 6[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:789255:789714 [0] NCCL INFO Channel 13/0 : 0[0] -> 4[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:789257:789715 [2] NCCL INFO Channel 15/0 : 1[2] -> 5[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:789261:789716 [6] NCCL INFO Channel 12/0 : 3[6] -> 7[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-121:3987794:3988245 [0] NCCL INFO Channel 03/0 : 4[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-121:3987800:3988248 [6] NCCL INFO Channel 00/0 : 7[6] -> 3[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-121:3987798:3988247 [4] NCCL INFO Channel 11/0 : 2[4] -> 6[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-121:3987796:3988246 [2] NCCL INFO Channel 13/0 : 1[2] -> 5[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:789259:789717 [4] NCCL INFO Channel 14/0 : 2[4] -> 6[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:789255:789714 [0] NCCL INFO Channel 14/0 : 0[0] -> 4[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:789261:789716 [6] NCCL INFO Channel 13/0 : 3[6] -> 7[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-121:3987794:3988245 [0] NCCL INFO Channel 04/0 : 4[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:789255:789714 [0] NCCL INFO Channel 15/0 : 0[0] -> 4[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:789259:789717 [4] NCCL INFO Channel 15/0 : 2[4] -> 6[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-121:3987798:3988247 [4] NCCL INFO Channel 14/0 : 2[4] -> 6[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-121:3987800:3988248 [6] NCCL INFO Channel 01/0 : 7[6] -> 3[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-121:3987796:3988246 [2] NCCL INFO Channel 14/0 : 1[2] -> 5[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-121:3987794:3988245 [0] NCCL INFO Channel 05/0 : 4[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-121:3987798:3988247 [4] NCCL INFO Channel 15/0 : 2[4] -> 6[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-121:3987800:3988248 [6] NCCL INFO Channel 02/0 : 7[6] -> 3[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-121:3987796:3988246 [2] NCCL INFO Channel 15/0 : 1[2] -> 5[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-121:3987794:3988245 [0] NCCL INFO Channel 06/0 : 4[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-121:3987800:3988248 [6] NCCL INFO Channel 03/0 : 7[6] -> 3[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-121:3987798:3988247 [4] NCCL INFO Channel 00/0 : 6[4] -> 2[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-121:3987794:3988245 [0] NCCL INFO Channel 07/0 : 4[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-121:3987796:3988246 [2] NCCL INFO Channel 00/0 : 5[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-121:3987798:3988247 [4] NCCL INFO Channel 01/0 : 6[4] -> 2[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-121:3987800:3988248 [6] NCCL INFO Channel 04/0 : 7[6] -> 3[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-121:3987794:3988245 [0] NCCL INFO Channel 10/0 : 4[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-121:3987796:3988246 [2] NCCL INFO Channel 01/0 : 5[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-121:3987798:3988247 [4] NCCL INFO Channel 02/0 : 6[4] -> 2[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-121:3987800:3988248 [6] NCCL INFO Channel 05/0 : 7[6] -> 3[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-121:3987794:3988245 [0] NCCL INFO Channel 11/0 : 4[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-121:3987796:3988246 [2] NCCL INFO Channel 04/0 : 5[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-121:3987800:3988248 [6] NCCL INFO Channel 08/0 : 7[6] -> 3[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-121:3987798:3988247 [4] NCCL INFO Channel 03/0 : 6[4] -> 2[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-121:3987794:3988245 [0] NCCL INFO Channel 12/0 : 4[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-121:3987796:3988246 [2] NCCL INFO Channel 05/0 : 5[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-121:3987798:3988247 [4] NCCL INFO Channel 06/0 : 6[4] -> 2[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-121:3987794:3988245 [0] NCCL INFO Channel 13/0 : 4[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-121:3987800:3988248 [6] NCCL INFO Channel 09/0 : 7[6] -> 3[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-121:3987796:3988246 [2] NCCL INFO Channel 06/0 : 5[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-121:3987798:3988247 [4] NCCL INFO Channel 07/0 : 6[4] -> 2[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-121:3987794:3988245 [0] NCCL INFO Channel 14/0 : 4[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-121:3987796:3988246 [2] NCCL INFO Channel 07/0 : 5[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-121:3987800:3988248 [6] NCCL INFO Channel 10/0 : 7[6] -> 3[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-121:3987798:3988247 [4] NCCL INFO Channel 08/0 : 6[4] -> 2[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-121:3987794:3988245 [0] NCCL INFO Channel 15/0 : 4[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-121:3987800:3988248 [6] NCCL INFO Channel 11/0 : 7[6] -> 3[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-121:3987796:3988246 [2] NCCL INFO Channel 08/0 : 5[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-121:3987798:3988247 [4] NCCL INFO Channel 09/0 : 6[4] -> 2[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-121:3987800:3988248 [6] NCCL INFO Channel 12/0 : 7[6] -> 3[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-121:3987796:3988246 [2] NCCL INFO Channel 09/0 : 5[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-121:3987798:3988247 [4] NCCL INFO Channel 10/0 : 6[4] -> 2[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-121:3987800:3988248 [6] NCCL INFO Channel 13/0 : 7[6] -> 3[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-121:3987798:3988247 [4] NCCL INFO Channel 11/0 : 6[4] -> 2[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-121:3987796:3988246 [2] NCCL INFO Channel 12/0 : 5[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-121:3987798:3988247 [4] NCCL INFO Channel 14/0 : 6[4] -> 2[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-121:3987796:3988246 [2] NCCL INFO Channel 13/0 : 5[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-121:3987798:3988247 [4] NCCL INFO Channel 15/0 : 6[4] -> 2[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-121:3987796:3988246 [2] NCCL INFO Channel 14/0 : 5[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-121:3987796:3988246 [2] NCCL INFO Channel 15/0 : 5[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-121:3987795:3988252 [1] NCCL INFO Connected NVLS tree
+ip-26-0-173-121:3987795:3988252 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-173-121:3987795:3988252 [1] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-173-121:3987795:3988252 [1] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-172-142:789256:789719 [1] NCCL INFO Connected NVLS tree
+ip-26-0-172-142:789256:789719 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-142:789256:789719 [1] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-172-142:789256:789719 [1] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-173-121:3987797:3988251 [3] NCCL INFO Connected NVLS tree
+ip-26-0-173-121:3987797:3988251 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-173-121:3987797:3988251 [3] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-173-121:3987797:3988251 [3] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-173-121:3987801:3988249 [7] NCCL INFO Connected NVLS tree
+ip-26-0-173-121:3987801:3988249 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-173-121:3987801:3988249 [7] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-173-121:3987801:3988249 [7] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-172-142:789258:789720 [3] NCCL INFO Connected NVLS tree
+ip-26-0-172-142:789258:789720 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-142:789258:789720 [3] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-172-142:789258:789720 [3] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-172-142:789262:789722 [7] NCCL INFO Connected NVLS tree
+ip-26-0-172-142:789262:789722 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-142:789262:789722 [7] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-172-142:789262:789722 [7] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-173-121:3987799:3988250 [5] NCCL INFO Connected NVLS tree
+ip-26-0-173-121:3987799:3988250 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-173-121:3987799:3988250 [5] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-173-121:3987799:3988250 [5] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-172-142:789260:789721 [5] NCCL INFO Connected NVLS tree
+ip-26-0-172-142:789260:789721 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-142:789260:789721 [5] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-172-142:789260:789721 [5] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-173-121:3987794:3988245 [0] NCCL INFO Connected NVLS tree
+ip-26-0-173-121:3987794:3988245 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-173-121:3987794:3988245 [0] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-173-121:3987794:3988245 [0] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-172-142:789255:789714 [0] NCCL INFO Connected NVLS tree
+ip-26-0-172-142:789255:789714 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-142:789255:789714 [0] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-172-142:789255:789714 [0] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-173-121:3987799:3988250 [5] NCCL INFO comm 0xb10cba0 rank 6 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0x700714da38272a85 - Init COMPLETE
+ip-26-0-173-121:3987795:3988252 [1] NCCL INFO comm 0xb7426f0 rank 4 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0x700714da38272a85 - Init COMPLETE
+ip-26-0-173-121:3987801:3988249 [7] NCCL INFO comm 0xab445e0 rank 7 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0x700714da38272a85 - Init COMPLETE
+ip-26-0-173-121:3987797:3988251 [3] NCCL INFO comm 0xaab04e0 rank 5 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0x700714da38272a85 - Init COMPLETE
+ip-26-0-173-121:3987798:3988247 [4] NCCL INFO Connected NVLS tree
+ip-26-0-173-121:3987798:3988247 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-173-121:3987798:3988247 [4] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-173-121:3987798:3988247 [4] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-172-142:789256:789719 [1] NCCL INFO comm 0xc430a60 rank 0 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0x700714da38272a85 - Init COMPLETE
+ip-26-0-172-142:789260:789721 [5] NCCL INFO comm 0xbfa4730 rank 2 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0x700714da38272a85 - Init COMPLETE
+ip-26-0-172-142:789262:789722 [7] NCCL INFO comm 0xa93d420 rank 3 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0x700714da38272a85 - Init COMPLETE
+ip-26-0-172-142:789258:789720 [3] NCCL INFO comm 0xaa5e860 rank 1 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0x700714da38272a85 - Init COMPLETE
+ip-26-0-173-121:3987800:3988248 [6] NCCL INFO Connected NVLS tree
+ip-26-0-173-121:3987800:3988248 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-173-121:3987800:3988248 [6] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-173-121:3987800:3988248 [6] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-173-121:3987796:3988246 [2] NCCL INFO Connected NVLS tree
+ip-26-0-173-121:3987796:3988246 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-173-121:3987796:3988246 [2] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-173-121:3987796:3988246 [2] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-172-142:789259:789717 [4] NCCL INFO Connected NVLS tree
+ip-26-0-172-142:789259:789717 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-142:789259:789717 [4] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-172-142:789259:789717 [4] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-172-142:789261:789716 [6] NCCL INFO Connected NVLS tree
+ip-26-0-172-142:789261:789716 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-142:789261:789716 [6] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-172-142:789261:789716 [6] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-172-142:789257:789715 [2] NCCL INFO Connected NVLS tree
+ip-26-0-172-142:789257:789715 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-142:789257:789715 [2] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-172-142:789257:789715 [2] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-173-121:3987798:3988247 [4] NCCL INFO comm 0xbb0d530 rank 6 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0xd3c41587a72a5b5b - Init COMPLETE
+ip-26-0-173-121:3987794:3988245 [0] NCCL INFO comm 0xab7ed80 rank 4 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0xd3c41587a72a5b5b - Init COMPLETE
+ip-26-0-173-121:3987800:3988248 [6] NCCL INFO comm 0xa732240 rank 7 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0xd3c41587a72a5b5b - Init COMPLETE
+ip-26-0-173-121:3987796:3988246 [2] NCCL INFO comm 0xafda250 rank 5 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0xd3c41587a72a5b5b - Init COMPLETE
+ip-26-0-172-142:789255:789714 [0] NCCL INFO comm 0xc0e2a70 rank 0 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0xd3c41587a72a5b5b - Init COMPLETE
+ip-26-0-172-142:789259:789717 [4] NCCL INFO comm 0xaf7eae0 rank 2 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0xd3c41587a72a5b5b - Init COMPLETE
+ip-26-0-172-142:789257:789715 [2] NCCL INFO comm 0xb2da4a0 rank 1 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0xd3c41587a72a5b5b - Init COMPLETE
+ip-26-0-172-142:789261:789716 [6] NCCL INFO comm 0xbd202a0 rank 3 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0xd3c41587a72a5b5b - Init COMPLETE
+01/07/2025 00:37:10 [INFO|DP=0|PP=3|TP=0|ip-26-0-172-142]: Local number of parameters: 201M (384.02MiB)
+01/07/2025 00:37:10 [INFO|DP=0|PP=2|TP=0|ip-26-0-172-142]: Local number of parameters: 201M (384.02MiB)
+01/07/2025 00:37:10 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]: Total number of parameters: 1.61G (3072.13MiB)
+01/07/2025 00:37:10 [INFO|DP=0|PP=7|TP=0|ip-26-0-173-121]: Local number of parameters: 268M (512.00MiB)
+01/07/2025 00:37:10 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]: Local number of parameters: 470M (896.02MiB)
+01/07/2025 00:37:10 [INFO|DP=0|PP=1|TP=0|ip-26-0-172-142]: Local number of parameters: 201M (384.02MiB)
+01/07/2025 00:37:10 [INFO|DP=0|PP=5|TP=0|ip-26-0-173-121]: Local number of parameters: 67.1M (128.01MiB)
+01/07/2025 00:37:10 [INFO|DP=0|PP=4|TP=0|ip-26-0-173-121]: Local number of parameters: 134M (256.02MiB)
+01/07/2025 00:37:10 [INFO|DP=0|PP=6|TP=0|ip-26-0-173-121]: Local number of parameters: 67.1M (128.01MiB)
+01/07/2025 00:37:10 [INFO|DP=0|PP=3|TP=0|ip-26-0-172-142]: [After model building] Memory usage: 384.04MiB. Peak allocated: 5504.00MiB Peak reserved: 13698.00MiB
+01/07/2025 00:37:10 [INFO|DP=0|PP=2|TP=0|ip-26-0-172-142]: [After model building] Memory usage: 384.04MiB. Peak allocated: 5504.00MiB Peak reserved: 13698.00MiB
+01/07/2025 00:37:10 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]: [After model building] Memory usage: 896.04MiB. Peak allocated: 5504.00MiB Peak reserved: 13698.00MiB
+01/07/2025 00:37:10 [INFO|DP=0|PP=1|TP=0|ip-26-0-172-142]: [After model building] Memory usage: 384.04MiB. Peak allocated: 5504.00MiB Peak reserved: 13698.00MiB
+01/07/2025 00:37:10 [INFO|DP=0|PP=7|TP=0|ip-26-0-173-121]: [After model building] Memory usage: 512.01MiB. Peak allocated: 5504.00MiB Peak reserved: 13698.00MiB
+01/07/2025 00:37:10 [INFO|DP=0|PP=5|TP=0|ip-26-0-173-121]: [After model building] Memory usage: 128.02MiB. Peak allocated: 5504.00MiB Peak reserved: 13698.00MiB
+01/07/2025 00:37:10 [INFO|DP=0|PP=6|TP=0|ip-26-0-173-121]: [After model building] Memory usage: 128.02MiB. Peak allocated: 5504.00MiB Peak reserved: 13698.00MiB
+01/07/2025 00:37:10 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]: No checkpoint path provided.
+01/07/2025 00:37:10 [INFO|DP=0|PP=4|TP=0|ip-26-0-173-121]: [After model building] Memory usage: 256.03MiB. Peak allocated: 5504.00MiB Peak reserved: 13698.00MiB
+01/07/2025 00:37:10 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]: Parametrizing model parameters using StandardParametrizator
+ip-26-0-172-142:789255:789751 [0] NCCL INFO Using network Libfabric
+ip-26-0-172-142:789256:789752 [1] NCCL INFO Using network Libfabric
+ip-26-0-173-121:3987800:3988280 [6] NCCL INFO Using network Libfabric
+ip-26-0-172-142:789256:789752 [1] NCCL INFO comm 0xc443720 rank 1 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0xcb34a8a225988e45 - Init START
+ip-26-0-172-142:789255:789751 [0] NCCL INFO comm 0xc0f5730 rank 0 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0xcb34a8a225988e45 - Init START
+ip-26-0-173-121:3987796:3988283 [2] NCCL INFO Using network Libfabric
+ip-26-0-172-142:789256:789752 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-173-121:3987801:3988282 [7] NCCL INFO Using network Libfabric
+ip-26-0-172-142:789255:789751 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/205
+ip-26-0-172-142:789262:789755 [7] NCCL INFO Using network Libfabric
+ip-26-0-172-142:789261:789754 [6] NCCL INFO Using network Libfabric
+ip-26-0-173-121:3987797:3988285 [3] NCCL INFO Using network Libfabric
+ip-26-0-172-142:789260:789759 [5] NCCL INFO Using network Libfabric
+ip-26-0-172-142:789259:789758 [4] NCCL INFO Using network Libfabric
+ip-26-0-173-121:3987800:3988280 [6] NCCL INFO comm 0xa8fd180 rank 0 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0x30139afce86f8b55 - Init START
+ip-26-0-173-121:3987801:3988282 [7] NCCL INFO comm 0xad0d1a0 rank 1 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0x30139afce86f8b55 - Init START
+ip-26-0-172-142:789257:789760 [2] NCCL INFO Using network Libfabric
+ip-26-0-172-142:789261:789754 [6] NCCL INFO comm 0xbee8650 rank 0 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0xc64be2bc3243716e - Init START
+ip-26-0-172-142:789262:789755 [7] NCCL INFO comm 0xab05550 rank 1 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0xc64be2bc3243716e - Init START
+ip-26-0-172-142:789258:789761 [3] NCCL INFO Using network Libfabric
+ip-26-0-173-121:3987798:3988286 [4] NCCL INFO Using network Libfabric
+ip-26-0-172-142:789261:789754 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-172-142:789262:789755 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-173-121:3987800:3988280 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-173-121:3987801:3988282 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-173-121:3987796:3988283 [2] NCCL INFO comm 0xb1a2bc0 rank 0 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0xa12c6c01388e3f90 - Init START
+ip-26-0-173-121:3987797:3988285 [3] NCCL INFO comm 0xac794c0 rank 1 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0xa12c6c01388e3f90 - Init START
+ip-26-0-173-121:3987799:3988288 [5] NCCL INFO Using network Libfabric
+ip-26-0-172-142:789259:789758 [4] NCCL INFO comm 0xb146d40 rank 0 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0xa927b92c1cfcdeb3 - Init START
+ip-26-0-172-142:789260:789759 [5] NCCL INFO comm 0xc16c4b0 rank 1 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0xa927b92c1cfcdeb3 - Init START
+ip-26-0-172-142:789259:789758 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-172-142:789260:789759 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-173-121:3987794:3988289 [0] NCCL INFO Using network Libfabric
+ip-26-0-173-121:3987796:3988283 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-173-121:3987797:3988285 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-172-142:789257:789760 [2] NCCL INFO comm 0xb4a29d0 rank 0 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0x322512f564e4938c - Init START
+ip-26-0-172-142:789258:789761 [3] NCCL INFO comm 0xac27860 rank 1 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0x322512f564e4938c - Init START
+ip-26-0-173-121:3987795:3988290 [1] NCCL INFO Using network Libfabric
+ip-26-0-172-142:789257:789760 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-172-142:789258:789761 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-173-121:3987799:3988288 [5] NCCL INFO comm 0xb2d6390 rank 1 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0x11e980de02d689a1 - Init START
+ip-26-0-173-121:3987798:3988286 [4] NCCL INFO comm 0xbcd5370 rank 0 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0x11e980de02d689a1 - Init START
+ip-26-0-173-121:3987798:3988286 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-173-121:3987799:3988288 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-173-121:3987794:3988289 [0] NCCL INFO comm 0xad48930 rank 0 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0x2ae6027f9d90b4c3 - Init START
+ip-26-0-173-121:3987795:3988290 [1] NCCL INFO comm 0xb90bdf0 rank 1 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0x2ae6027f9d90b4c3 - Init START
+ip-26-0-173-121:3987794:3988289 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-173-121:3987795:3988290 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-173-121:3987798:3988286 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-172-142:789259:789758 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-172-142:789255:789751 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-173-121:3987800:3988280 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-173-121:3987801:3988282 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-173-121:3987800:3988280 [6] NCCL INFO Channel 00/24 :    0   1
+ip-26-0-173-121:3987800:3988280 [6] NCCL INFO Channel 01/24 :    0   1
+ip-26-0-173-121:3987800:3988280 [6] NCCL INFO Channel 02/24 :    0   1
+ip-26-0-173-121:3987800:3988280 [6] NCCL INFO Channel 03/24 :    0   1
+ip-26-0-173-121:3987800:3988280 [6] NCCL INFO Channel 04/24 :    0   1
+ip-26-0-173-121:3987800:3988280 [6] NCCL INFO Channel 05/24 :    0   1
+ip-26-0-173-121:3987800:3988280 [6] NCCL INFO Channel 06/24 :    0   1
+ip-26-0-173-121:3987800:3988280 [6] NCCL INFO Channel 07/24 :    0   1
+ip-26-0-173-121:3987800:3988280 [6] NCCL INFO Channel 08/24 :    0   1
+ip-26-0-173-121:3987800:3988280 [6] NCCL INFO Channel 09/24 :    0   1
+ip-26-0-173-121:3987800:3988280 [6] NCCL INFO Channel 10/24 :    0   1
+ip-26-0-173-121:3987801:3988282 [7] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] -1/-1/-1->1->0 [3] -1/-1/-1->1->0 [4] -1/-1/-1->1->0 [5] -1/-1/-1->1->0 [6] 0/-1/-1->1->-1 [7] 0/-1/-1->1->-1 [8] 0/-1/-1->1->-1 [9] 0/-1/-1->1->-1 [10] 0/-1/-1->1->-1 [11] 0/-1/-1->1->-1 [12] -1/-1/-1->1->0 [13] -1/-1/-1->1->0 [14] -1/-1/-1->1->0 [15] -1/-1/-1->1->0 [16] -1/-1/-1->1->0 [17] -1/-1/-1->1->0 [18] 0/-1/-1->1->-1 [19] 0/-1/-1->1->-1 [20] 0/-1/-1->1->-1 [21] 0/-1/-1->1->-1 [22] 0/-1/-1->1->-1 [23] 0/-1/-1->1->-1
+ip-26-0-173-121:3987800:3988280 [6] NCCL INFO Channel 11/24 :    0   1
+ip-26-0-173-121:3987800:3988280 [6] NCCL INFO Channel 12/24 :    0   1
+ip-26-0-173-121:3987801:3988282 [7] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-173-121:3987800:3988280 [6] NCCL INFO Channel 13/24 :    0   1
+ip-26-0-173-121:3987800:3988280 [6] NCCL INFO Channel 14/24 :    0   1
+ip-26-0-173-121:3987800:3988280 [6] NCCL INFO Channel 15/24 :    0   1
+ip-26-0-173-121:3987800:3988280 [6] NCCL INFO Channel 16/24 :    0   1
+ip-26-0-173-121:3987800:3988280 [6] NCCL INFO Channel 17/24 :    0   1
+ip-26-0-173-121:3987800:3988280 [6] NCCL INFO Channel 18/24 :    0   1
+ip-26-0-173-121:3987800:3988280 [6] NCCL INFO Channel 19/24 :    0   1
+ip-26-0-173-121:3987800:3988280 [6] NCCL INFO Channel 20/24 :    0   1
+ip-26-0-173-121:3987800:3988280 [6] NCCL INFO Channel 21/24 :    0   1
+ip-26-0-173-121:3987800:3988280 [6] NCCL INFO Channel 22/24 :    0   1
+ip-26-0-173-121:3987800:3988280 [6] NCCL INFO Channel 23/24 :    0   1
+ip-26-0-173-121:3987800:3988280 [6] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] -1/-1/-1->0->1 [7] -1/-1/-1->0->1 [8] -1/-1/-1->0->1 [9] -1/-1/-1->0->1 [10] -1/-1/-1->0->1 [11] -1/-1/-1->0->1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] -1/-1/-1->0->1 [19] -1/-1/-1->0->1 [20] -1/-1/-1->0->1 [21] -1/-1/-1->0->1 [22] -1/-1/-1->0->1 [23] -1/-1/-1->0->1
+ip-26-0-173-121:3987800:3988280 [6] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-172-142:789256:789752 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-172-142:789255:789751 [0] NCCL INFO Channel 00/24 :    0   1
+ip-26-0-172-142:789255:789751 [0] NCCL INFO Channel 01/24 :    0   1
+ip-26-0-172-142:789256:789752 [1] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] -1/-1/-1->1->0 [3] -1/-1/-1->1->0 [4] -1/-1/-1->1->0 [5] -1/-1/-1->1->0 [6] 0/-1/-1->1->-1 [7] 0/-1/-1->1->-1 [8] 0/-1/-1->1->-1 [9] 0/-1/-1->1->-1 [10] 0/-1/-1->1->-1 [11] 0/-1/-1->1->-1 [12] -1/-1/-1->1->0 [13] -1/-1/-1->1->0 [14] -1/-1/-1->1->0 [15] -1/-1/-1->1->0 [16] -1/-1/-1->1->0 [17] -1/-1/-1->1->0 [18] 0/-1/-1->1->-1 [19] 0/-1/-1->1->-1 [20] 0/-1/-1->1->-1 [21] 0/-1/-1->1->-1 [22] 0/-1/-1->1->-1 [23] 0/-1/-1->1->-1
+ip-26-0-172-142:789255:789751 [0] NCCL INFO Channel 02/24 :    0   1
+ip-26-0-172-142:789256:789752 [1] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-172-142:789255:789751 [0] NCCL INFO Channel 03/24 :    0   1
+ip-26-0-172-142:789255:789751 [0] NCCL INFO Channel 04/24 :    0   1
+ip-26-0-172-142:789255:789751 [0] NCCL INFO Channel 05/24 :    0   1
+ip-26-0-172-142:789255:789751 [0] NCCL INFO Channel 06/24 :    0   1
+ip-26-0-172-142:789255:789751 [0] NCCL INFO Channel 07/24 :    0   1
+ip-26-0-172-142:789255:789751 [0] NCCL INFO Channel 08/24 :    0   1
+ip-26-0-172-142:789255:789751 [0] NCCL INFO Channel 09/24 :    0   1
+ip-26-0-172-142:789255:789751 [0] NCCL INFO Channel 10/24 :    0   1
+ip-26-0-172-142:789255:789751 [0] NCCL INFO Channel 11/24 :    0   1
+ip-26-0-172-142:789255:789751 [0] NCCL INFO Channel 12/24 :    0   1
+ip-26-0-172-142:789255:789751 [0] NCCL INFO Channel 13/24 :    0   1
+ip-26-0-172-142:789255:789751 [0] NCCL INFO Channel 14/24 :    0   1
+ip-26-0-172-142:789255:789751 [0] NCCL INFO Channel 15/24 :    0   1
+ip-26-0-172-142:789255:789751 [0] NCCL INFO Channel 16/24 :    0   1
+ip-26-0-172-142:789255:789751 [0] NCCL INFO Channel 17/24 :    0   1
+ip-26-0-172-142:789255:789751 [0] NCCL INFO Channel 18/24 :    0   1
+ip-26-0-172-142:789255:789751 [0] NCCL INFO Channel 19/24 :    0   1
+ip-26-0-172-142:789255:789751 [0] NCCL INFO Channel 20/24 :    0   1
+ip-26-0-172-142:789255:789751 [0] NCCL INFO Channel 21/24 :    0   1
+ip-26-0-172-142:789255:789751 [0] NCCL INFO Channel 22/24 :    0   1
+ip-26-0-172-142:789255:789751 [0] NCCL INFO Channel 23/24 :    0   1
+ip-26-0-172-142:789255:789751 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] -1/-1/-1->0->1 [7] -1/-1/-1->0->1 [8] -1/-1/-1->0->1 [9] -1/-1/-1->0->1 [10] -1/-1/-1->0->1 [11] -1/-1/-1->0->1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] -1/-1/-1->0->1 [19] -1/-1/-1->0->1 [20] -1/-1/-1->0->1 [21] -1/-1/-1->0->1 [22] -1/-1/-1->0->1 [23] -1/-1/-1->0->1
+ip-26-0-172-142:789255:789751 [0] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-172-142:789262:789755 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-172-142:789260:789759 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-172-142:789259:789758 [4] NCCL INFO Channel 00/24 :    0   1
+ip-26-0-172-142:789259:789758 [4] NCCL INFO Channel 01/24 :    0   1
+ip-26-0-172-142:789259:789758 [4] NCCL INFO Channel 02/24 :    0   1
+ip-26-0-172-142:789259:789758 [4] NCCL INFO Channel 03/24 :    0   1
+ip-26-0-172-142:789259:789758 [4] NCCL INFO Channel 04/24 :    0   1
+ip-26-0-172-142:789259:789758 [4] NCCL INFO Channel 05/24 :    0   1
+ip-26-0-172-142:789259:789758 [4] NCCL INFO Channel 06/24 :    0   1
+ip-26-0-172-142:789259:789758 [4] NCCL INFO Channel 07/24 :    0   1
+ip-26-0-172-142:789260:789759 [5] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] -1/-1/-1->1->0 [3] -1/-1/-1->1->0 [4] -1/-1/-1->1->0 [5] -1/-1/-1->1->0 [6] 0/-1/-1->1->-1 [7] 0/-1/-1->1->-1 [8] 0/-1/-1->1->-1 [9] 0/-1/-1->1->-1 [10] 0/-1/-1->1->-1 [11] 0/-1/-1->1->-1 [12] -1/-1/-1->1->0 [13] -1/-1/-1->1->0 [14] -1/-1/-1->1->0 [15] -1/-1/-1->1->0 [16] -1/-1/-1->1->0 [17] -1/-1/-1->1->0 [18] 0/-1/-1->1->-1 [19] 0/-1/-1->1->-1 [20] 0/-1/-1->1->-1 [21] 0/-1/-1->1->-1 [22] 0/-1/-1->1->-1 [23] 0/-1/-1->1->-1
+ip-26-0-172-142:789259:789758 [4] NCCL INFO Channel 08/24 :    0   1
+ip-26-0-172-142:789260:789759 [5] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-172-142:789259:789758 [4] NCCL INFO Channel 09/24 :    0   1
+ip-26-0-172-142:789259:789758 [4] NCCL INFO Channel 10/24 :    0   1
+ip-26-0-172-142:789259:789758 [4] NCCL INFO Channel 11/24 :    0   1
+ip-26-0-172-142:789259:789758 [4] NCCL INFO Channel 12/24 :    0   1
+ip-26-0-172-142:789259:789758 [4] NCCL INFO Channel 13/24 :    0   1
+ip-26-0-172-142:789259:789758 [4] NCCL INFO Channel 14/24 :    0   1
+ip-26-0-172-142:789259:789758 [4] NCCL INFO Channel 15/24 :    0   1
+ip-26-0-172-142:789259:789758 [4] NCCL INFO Channel 16/24 :    0   1
+ip-26-0-172-142:789259:789758 [4] NCCL INFO Channel 17/24 :    0   1
+ip-26-0-172-142:789259:789758 [4] NCCL INFO Channel 18/24 :    0   1
+ip-26-0-172-142:789259:789758 [4] NCCL INFO Channel 19/24 :    0   1
+ip-26-0-172-142:789259:789758 [4] NCCL INFO Channel 20/24 :    0   1
+ip-26-0-172-142:789259:789758 [4] NCCL INFO Channel 21/24 :    0   1
+ip-26-0-172-142:789259:789758 [4] NCCL INFO Channel 22/24 :    0   1
+ip-26-0-172-142:789259:789758 [4] NCCL INFO Channel 23/24 :    0   1
+ip-26-0-172-142:789259:789758 [4] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] -1/-1/-1->0->1 [7] -1/-1/-1->0->1 [8] -1/-1/-1->0->1 [9] -1/-1/-1->0->1 [10] -1/-1/-1->0->1 [11] -1/-1/-1->0->1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] -1/-1/-1->0->1 [19] -1/-1/-1->0->1 [20] -1/-1/-1->0->1 [21] -1/-1/-1->0->1 [22] -1/-1/-1->0->1 [23] -1/-1/-1->0->1
+ip-26-0-172-142:789259:789758 [4] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-172-142:789258:789761 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-172-142:789261:789754 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-172-142:789261:789754 [6] NCCL INFO Channel 00/24 :    0   1
+ip-26-0-172-142:789261:789754 [6] NCCL INFO Channel 01/24 :    0   1
+ip-26-0-172-142:789261:789754 [6] NCCL INFO Channel 02/24 :    0   1
+ip-26-0-172-142:789261:789754 [6] NCCL INFO Channel 03/24 :    0   1
+ip-26-0-172-142:789261:789754 [6] NCCL INFO Channel 04/24 :    0   1
+ip-26-0-172-142:789261:789754 [6] NCCL INFO Channel 05/24 :    0   1
+ip-26-0-172-142:789261:789754 [6] NCCL INFO Channel 06/24 :    0   1
+ip-26-0-172-142:789261:789754 [6] NCCL INFO Channel 07/24 :    0   1
+ip-26-0-172-142:789261:789754 [6] NCCL INFO Channel 08/24 :    0   1
+ip-26-0-172-142:789261:789754 [6] NCCL INFO Channel 09/24 :    0   1
+ip-26-0-172-142:789262:789755 [7] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] -1/-1/-1->1->0 [3] -1/-1/-1->1->0 [4] -1/-1/-1->1->0 [5] -1/-1/-1->1->0 [6] 0/-1/-1->1->-1 [7] 0/-1/-1->1->-1 [8] 0/-1/-1->1->-1 [9] 0/-1/-1->1->-1 [10] 0/-1/-1->1->-1 [11] 0/-1/-1->1->-1 [12] -1/-1/-1->1->0 [13] -1/-1/-1->1->0 [14] -1/-1/-1->1->0 [15] -1/-1/-1->1->0 [16] -1/-1/-1->1->0 [17] -1/-1/-1->1->0 [18] 0/-1/-1->1->-1 [19] 0/-1/-1->1->-1 [20] 0/-1/-1->1->-1 [21] 0/-1/-1->1->-1 [22] 0/-1/-1->1->-1 [23] 0/-1/-1->1->-1
+ip-26-0-172-142:789261:789754 [6] NCCL INFO Channel 10/24 :    0   1
+ip-26-0-172-142:789261:789754 [6] NCCL INFO Channel 11/24 :    0   1
+ip-26-0-172-142:789262:789755 [7] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-172-142:789261:789754 [6] NCCL INFO Channel 12/24 :    0   1
+ip-26-0-172-142:789261:789754 [6] NCCL INFO Channel 13/24 :    0   1
+ip-26-0-172-142:789261:789754 [6] NCCL INFO Channel 14/24 :    0   1
+ip-26-0-172-142:789261:789754 [6] NCCL INFO Channel 15/24 :    0   1
+ip-26-0-172-142:789261:789754 [6] NCCL INFO Channel 16/24 :    0   1
+ip-26-0-172-142:789261:789754 [6] NCCL INFO Channel 17/24 :    0   1
+ip-26-0-172-142:789261:789754 [6] NCCL INFO Channel 18/24 :    0   1
+ip-26-0-172-142:789261:789754 [6] NCCL INFO Channel 19/24 :    0   1
+ip-26-0-172-142:789261:789754 [6] NCCL INFO Channel 20/24 :    0   1
+ip-26-0-172-142:789261:789754 [6] NCCL INFO Channel 21/24 :    0   1
+ip-26-0-172-142:789261:789754 [6] NCCL INFO Channel 22/24 :    0   1
+ip-26-0-172-142:789261:789754 [6] NCCL INFO Channel 23/24 :    0   1
+ip-26-0-172-142:789261:789754 [6] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] -1/-1/-1->0->1 [7] -1/-1/-1->0->1 [8] -1/-1/-1->0->1 [9] -1/-1/-1->0->1 [10] -1/-1/-1->0->1 [11] -1/-1/-1->0->1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] -1/-1/-1->0->1 [19] -1/-1/-1->0->1 [20] -1/-1/-1->0->1 [21] -1/-1/-1->0->1 [22] -1/-1/-1->0->1 [23] -1/-1/-1->0->1
+ip-26-0-172-142:789261:789754 [6] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-172-142:789257:789760 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-172-142:789257:789760 [2] NCCL INFO Channel 00/24 :    0   1
+ip-26-0-172-142:789257:789760 [2] NCCL INFO Channel 01/24 :    0   1
+ip-26-0-172-142:789257:789760 [2] NCCL INFO Channel 02/24 :    0   1
+ip-26-0-172-142:789257:789760 [2] NCCL INFO Channel 03/24 :    0   1
+ip-26-0-172-142:789257:789760 [2] NCCL INFO Channel 04/24 :    0   1
+ip-26-0-172-142:789257:789760 [2] NCCL INFO Channel 05/24 :    0   1
+ip-26-0-172-142:789257:789760 [2] NCCL INFO Channel 06/24 :    0   1
+ip-26-0-172-142:789257:789760 [2] NCCL INFO Channel 07/24 :    0   1
+ip-26-0-172-142:789257:789760 [2] NCCL INFO Channel 08/24 :    0   1
+ip-26-0-172-142:789257:789760 [2] NCCL INFO Channel 09/24 :    0   1
+ip-26-0-172-142:789258:789761 [3] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] -1/-1/-1->1->0 [3] -1/-1/-1->1->0 [4] -1/-1/-1->1->0 [5] -1/-1/-1->1->0 [6] 0/-1/-1->1->-1 [7] 0/-1/-1->1->-1 [8] 0/-1/-1->1->-1 [9] 0/-1/-1->1->-1 [10] 0/-1/-1->1->-1 [11] 0/-1/-1->1->-1 [12] -1/-1/-1->1->0 [13] -1/-1/-1->1->0 [14] -1/-1/-1->1->0 [15] -1/-1/-1->1->0 [16] -1/-1/-1->1->0 [17] -1/-1/-1->1->0 [18] 0/-1/-1->1->-1 [19] 0/-1/-1->1->-1 [20] 0/-1/-1->1->-1 [21] 0/-1/-1->1->-1 [22] 0/-1/-1->1->-1 [23] 0/-1/-1->1->-1
+ip-26-0-172-142:789257:789760 [2] NCCL INFO Channel 10/24 :    0   1
+ip-26-0-172-142:789258:789761 [3] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-172-142:789257:789760 [2] NCCL INFO Channel 11/24 :    0   1
+ip-26-0-172-142:789257:789760 [2] NCCL INFO Channel 12/24 :    0   1
+ip-26-0-172-142:789257:789760 [2] NCCL INFO Channel 13/24 :    0   1
+ip-26-0-172-142:789257:789760 [2] NCCL INFO Channel 14/24 :    0   1
+ip-26-0-172-142:789257:789760 [2] NCCL INFO Channel 15/24 :    0   1
+ip-26-0-172-142:789257:789760 [2] NCCL INFO Channel 16/24 :    0   1
+ip-26-0-172-142:789257:789760 [2] NCCL INFO Channel 17/24 :    0   1
+ip-26-0-172-142:789257:789760 [2] NCCL INFO Channel 18/24 :    0   1
+ip-26-0-172-142:789257:789760 [2] NCCL INFO Channel 19/24 :    0   1
+ip-26-0-172-142:789257:789760 [2] NCCL INFO Channel 20/24 :    0   1
+ip-26-0-172-142:789257:789760 [2] NCCL INFO Channel 21/24 :    0   1
+ip-26-0-172-142:789257:789760 [2] NCCL INFO Channel 22/24 :    0   1
+ip-26-0-172-142:789257:789760 [2] NCCL INFO Channel 23/24 :    0   1
+ip-26-0-172-142:789257:789760 [2] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] -1/-1/-1->0->1 [7] -1/-1/-1->0->1 [8] -1/-1/-1->0->1 [9] -1/-1/-1->0->1 [10] -1/-1/-1->0->1 [11] -1/-1/-1->0->1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] -1/-1/-1->0->1 [19] -1/-1/-1->0->1 [20] -1/-1/-1->0->1 [21] -1/-1/-1->0->1 [22] -1/-1/-1->0->1 [23] -1/-1/-1->0->1
+ip-26-0-172-142:789257:789760 [2] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-172-142:789256:789752 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789259:789758 [4] NCCL INFO Channel 00/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-172-142:789256:789752 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789259:789758 [4] NCCL INFO Channel 01/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-172-142:789255:789751 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-142:789256:789752 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789259:789758 [4] NCCL INFO Channel 02/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-172-142:789255:789751 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-142:789256:789752 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789259:789758 [4] NCCL INFO Channel 03/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-172-142:789255:789751 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-142:789256:789752 [1] NCCL INFO Channel 04/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789259:789758 [4] NCCL INFO Channel 04/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-172-142:789255:789751 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-142:789259:789758 [4] NCCL INFO Channel 05/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-172-142:789256:789752 [1] NCCL INFO Channel 05/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789261:789754 [6] NCCL INFO Channel 00/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-172-142:789255:789751 [0] NCCL INFO Channel 04/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-142:789256:789752 [1] NCCL INFO Channel 06/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789259:789758 [4] NCCL INFO Channel 06/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-172-142:789261:789754 [6] NCCL INFO Channel 01/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-173-121:3987794:3988289 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-172-142:789255:789751 [0] NCCL INFO Channel 05/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-142:789259:789758 [4] NCCL INFO Channel 07/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-172-142:789256:789752 [1] NCCL INFO Channel 07/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789261:789754 [6] NCCL INFO Channel 02/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-172-142:789260:789759 [5] NCCL INFO Channel 00/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-172-142:789259:789758 [4] NCCL INFO Channel 08/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-172-142:789255:789751 [0] NCCL INFO Channel 06/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-142:789261:789754 [6] NCCL INFO Channel 03/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-172-142:789256:789752 [1] NCCL INFO Channel 08/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789260:789759 [5] NCCL INFO Channel 01/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-172-142:789259:789758 [4] NCCL INFO Channel 09/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-172-142:789255:789751 [0] NCCL INFO Channel 07/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-142:789261:789754 [6] NCCL INFO Channel 04/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-172-142:789256:789752 [1] NCCL INFO Channel 09/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789258:789761 [3] NCCL INFO Channel 00/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-172-142:789260:789759 [5] NCCL INFO Channel 02/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-172-142:789259:789758 [4] NCCL INFO Channel 10/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-172-142:789261:789754 [6] NCCL INFO Channel 05/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-172-142:789255:789751 [0] NCCL INFO Channel 08/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-142:789256:789752 [1] NCCL INFO Channel 10/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789258:789761 [3] NCCL INFO Channel 01/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-172-142:789260:789759 [5] NCCL INFO Channel 03/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-173-121:3987795:3988290 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-172-142:789259:789758 [4] NCCL INFO Channel 11/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-172-142:789261:789754 [6] NCCL INFO Channel 06/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-172-142:789255:789751 [0] NCCL INFO Channel 09/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-142:789256:789752 [1] NCCL INFO Channel 11/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987794:3988289 [0] NCCL INFO Channel 00/24 :    0   1
+ip-26-0-173-121:3987794:3988289 [0] NCCL INFO Channel 01/24 :    0   1
+ip-26-0-173-121:3987794:3988289 [0] NCCL INFO Channel 02/24 :    0   1
+ip-26-0-173-121:3987794:3988289 [0] NCCL INFO Channel 03/24 :    0   1
+ip-26-0-173-121:3987794:3988289 [0] NCCL INFO Channel 04/24 :    0   1
+ip-26-0-173-121:3987794:3988289 [0] NCCL INFO Channel 05/24 :    0   1
+ip-26-0-173-121:3987794:3988289 [0] NCCL INFO Channel 06/24 :    0   1
+ip-26-0-173-121:3987794:3988289 [0] NCCL INFO Channel 07/24 :    0   1
+ip-26-0-173-121:3987794:3988289 [0] NCCL INFO Channel 08/24 :    0   1
+ip-26-0-173-121:3987794:3988289 [0] NCCL INFO Channel 09/24 :    0   1
+ip-26-0-173-121:3987794:3988289 [0] NCCL INFO Channel 10/24 :    0   1
+ip-26-0-173-121:3987795:3988290 [1] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] -1/-1/-1->1->0 [3] -1/-1/-1->1->0 [4] -1/-1/-1->1->0 [5] -1/-1/-1->1->0 [6] 0/-1/-1->1->-1 [7] 0/-1/-1->1->-1 [8] 0/-1/-1->1->-1 [9] 0/-1/-1->1->-1 [10] 0/-1/-1->1->-1 [11] 0/-1/-1->1->-1 [12] -1/-1/-1->1->0 [13] -1/-1/-1->1->0 [14] -1/-1/-1->1->0 [15] -1/-1/-1->1->0 [16] -1/-1/-1->1->0 [17] -1/-1/-1->1->0 [18] 0/-1/-1->1->-1 [19] 0/-1/-1->1->-1 [20] 0/-1/-1->1->-1 [21] 0/-1/-1->1->-1 [22] 0/-1/-1->1->-1 [23] 0/-1/-1->1->-1
+ip-26-0-173-121:3987794:3988289 [0] NCCL INFO Channel 11/24 :    0   1
+ip-26-0-173-121:3987794:3988289 [0] NCCL INFO Channel 12/24 :    0   1
+ip-26-0-173-121:3987795:3988290 [1] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-173-121:3987794:3988289 [0] NCCL INFO Channel 13/24 :    0   1
+ip-26-0-173-121:3987794:3988289 [0] NCCL INFO Channel 14/24 :    0   1
+ip-26-0-173-121:3987794:3988289 [0] NCCL INFO Channel 15/24 :    0   1
+ip-26-0-173-121:3987794:3988289 [0] NCCL INFO Channel 16/24 :    0   1
+ip-26-0-173-121:3987794:3988289 [0] NCCL INFO Channel 17/24 :    0   1
+ip-26-0-173-121:3987794:3988289 [0] NCCL INFO Channel 18/24 :    0   1
+ip-26-0-173-121:3987794:3988289 [0] NCCL INFO Channel 19/24 :    0   1
+ip-26-0-173-121:3987794:3988289 [0] NCCL INFO Channel 20/24 :    0   1
+ip-26-0-173-121:3987794:3988289 [0] NCCL INFO Channel 21/24 :    0   1
+ip-26-0-173-121:3987794:3988289 [0] NCCL INFO Channel 22/24 :    0   1
+ip-26-0-173-121:3987794:3988289 [0] NCCL INFO Channel 23/24 :    0   1
+ip-26-0-173-121:3987794:3988289 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] -1/-1/-1->0->1 [7] -1/-1/-1->0->1 [8] -1/-1/-1->0->1 [9] -1/-1/-1->0->1 [10] -1/-1/-1->0->1 [11] -1/-1/-1->0->1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] -1/-1/-1->0->1 [19] -1/-1/-1->0->1 [20] -1/-1/-1->0->1 [21] -1/-1/-1->0->1 [22] -1/-1/-1->0->1 [23] -1/-1/-1->0->1
+ip-26-0-173-121:3987794:3988289 [0] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-173-121:3987799:3988288 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-172-142:789258:789761 [3] NCCL INFO Channel 02/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-173-121:3987799:3988288 [5] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] -1/-1/-1->1->0 [3] -1/-1/-1->1->0 [4] -1/-1/-1->1->0 [5] -1/-1/-1->1->0 [6] 0/-1/-1->1->-1 [7] 0/-1/-1->1->-1 [8] 0/-1/-1->1->-1 [9] 0/-1/-1->1->-1 [10] 0/-1/-1->1->-1 [11] 0/-1/-1->1->-1 [12] -1/-1/-1->1->0 [13] -1/-1/-1->1->0 [14] -1/-1/-1->1->0 [15] -1/-1/-1->1->0 [16] -1/-1/-1->1->0 [17] -1/-1/-1->1->0 [18] 0/-1/-1->1->-1 [19] 0/-1/-1->1->-1 [20] 0/-1/-1->1->-1 [21] 0/-1/-1->1->-1 [22] 0/-1/-1->1->-1 [23] 0/-1/-1->1->-1
+ip-26-0-173-121:3987799:3988288 [5] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-173-121:3987798:3988286 [4] NCCL INFO Channel 00/24 :    0   1
+ip-26-0-172-142:789260:789759 [5] NCCL INFO Channel 04/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-173-121:3987798:3988286 [4] NCCL INFO Channel 01/24 :    0   1
+ip-26-0-173-121:3987798:3988286 [4] NCCL INFO Channel 02/24 :    0   1
+ip-26-0-173-121:3987798:3988286 [4] NCCL INFO Channel 03/24 :    0   1
+ip-26-0-173-121:3987798:3988286 [4] NCCL INFO Channel 04/24 :    0   1
+ip-26-0-173-121:3987798:3988286 [4] NCCL INFO Channel 05/24 :    0   1
+ip-26-0-173-121:3987798:3988286 [4] NCCL INFO Channel 06/24 :    0   1
+ip-26-0-173-121:3987798:3988286 [4] NCCL INFO Channel 07/24 :    0   1
+ip-26-0-173-121:3987798:3988286 [4] NCCL INFO Channel 08/24 :    0   1
+ip-26-0-173-121:3987798:3988286 [4] NCCL INFO Channel 09/24 :    0   1
+ip-26-0-173-121:3987798:3988286 [4] NCCL INFO Channel 10/24 :    0   1
+ip-26-0-173-121:3987798:3988286 [4] NCCL INFO Channel 11/24 :    0   1
+ip-26-0-173-121:3987798:3988286 [4] NCCL INFO Channel 12/24 :    0   1
+ip-26-0-173-121:3987798:3988286 [4] NCCL INFO Channel 13/24 :    0   1
+ip-26-0-173-121:3987798:3988286 [4] NCCL INFO Channel 14/24 :    0   1
+ip-26-0-173-121:3987798:3988286 [4] NCCL INFO Channel 15/24 :    0   1
+ip-26-0-173-121:3987798:3988286 [4] NCCL INFO Channel 16/24 :    0   1
+ip-26-0-173-121:3987798:3988286 [4] NCCL INFO Channel 17/24 :    0   1
+ip-26-0-173-121:3987798:3988286 [4] NCCL INFO Channel 18/24 :    0   1
+ip-26-0-173-121:3987798:3988286 [4] NCCL INFO Channel 19/24 :    0   1
+ip-26-0-173-121:3987798:3988286 [4] NCCL INFO Channel 20/24 :    0   1
+ip-26-0-173-121:3987798:3988286 [4] NCCL INFO Channel 21/24 :    0   1
+ip-26-0-173-121:3987798:3988286 [4] NCCL INFO Channel 22/24 :    0   1
+ip-26-0-173-121:3987798:3988286 [4] NCCL INFO Channel 23/24 :    0   1
+ip-26-0-173-121:3987798:3988286 [4] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] -1/-1/-1->0->1 [7] -1/-1/-1->0->1 [8] -1/-1/-1->0->1 [9] -1/-1/-1->0->1 [10] -1/-1/-1->0->1 [11] -1/-1/-1->0->1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] -1/-1/-1->0->1 [19] -1/-1/-1->0->1 [20] -1/-1/-1->0->1 [21] -1/-1/-1->0->1 [22] -1/-1/-1->0->1 [23] -1/-1/-1->0->1
+ip-26-0-173-121:3987798:3988286 [4] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-172-142:789261:789754 [6] NCCL INFO Channel 07/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-172-142:789259:789758 [4] NCCL INFO Channel 12/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-172-142:789256:789752 [1] NCCL INFO Channel 12/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789255:789751 [0] NCCL INFO Channel 10/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-142:789258:789761 [3] NCCL INFO Channel 03/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-172-142:789260:789759 [5] NCCL INFO Channel 05/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-172-142:789262:789755 [7] NCCL INFO Channel 00/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-172-142:789259:789758 [4] NCCL INFO Channel 13/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-172-142:789261:789754 [6] NCCL INFO Channel 08/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-172-142:789256:789752 [1] NCCL INFO Channel 13/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789255:789751 [0] NCCL INFO Channel 11/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-142:789258:789761 [3] NCCL INFO Channel 04/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-172-142:789260:789759 [5] NCCL INFO Channel 06/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-172-142:789262:789755 [7] NCCL INFO Channel 01/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-172-142:789259:789758 [4] NCCL INFO Channel 14/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-172-142:789261:789754 [6] NCCL INFO Channel 09/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-172-142:789255:789751 [0] NCCL INFO Channel 12/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-142:789258:789761 [3] NCCL INFO Channel 05/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-172-142:789260:789759 [5] NCCL INFO Channel 07/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-172-142:789262:789755 [7] NCCL INFO Channel 02/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-172-142:789261:789754 [6] NCCL INFO Channel 10/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-172-142:789259:789758 [4] NCCL INFO Channel 15/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-172-142:789258:789761 [3] NCCL INFO Channel 06/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-172-142:789255:789751 [0] NCCL INFO Channel 13/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-142:789260:789759 [5] NCCL INFO Channel 08/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-172-142:789262:789755 [7] NCCL INFO Channel 03/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-172-142:789259:789758 [4] NCCL INFO Channel 16/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-172-142:789261:789754 [6] NCCL INFO Channel 11/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-172-142:789256:789752 [1] NCCL INFO Channel 14/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789255:789751 [0] NCCL INFO Channel 14/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-142:789258:789761 [3] NCCL INFO Channel 07/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-172-142:789260:789759 [5] NCCL INFO Channel 09/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-172-142:789262:789755 [7] NCCL INFO Channel 04/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-172-142:789259:789758 [4] NCCL INFO Channel 17/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-172-142:789261:789754 [6] NCCL INFO Channel 12/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-172-142:789255:789751 [0] NCCL INFO Channel 15/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-142:789260:789759 [5] NCCL INFO Channel 10/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-172-142:789258:789761 [3] NCCL INFO Channel 08/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-172-142:789256:789752 [1] NCCL INFO Channel 15/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789262:789755 [7] NCCL INFO Channel 05/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-172-142:789261:789754 [6] NCCL INFO Channel 13/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-172-142:789259:789758 [4] NCCL INFO Channel 18/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-172-142:789260:789759 [5] NCCL INFO Channel 11/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-172-142:789258:789761 [3] NCCL INFO Channel 09/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-172-142:789256:789752 [1] NCCL INFO Channel 16/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789255:789751 [0] NCCL INFO Channel 16/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-142:789262:789755 [7] NCCL INFO Channel 06/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-172-142:789259:789758 [4] NCCL INFO Channel 19/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-172-142:789261:789754 [6] NCCL INFO Channel 14/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-172-142:789260:789759 [5] NCCL INFO Channel 12/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-172-142:789258:789761 [3] NCCL INFO Channel 10/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-172-142:789256:789752 [1] NCCL INFO Channel 17/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789255:789751 [0] NCCL INFO Channel 17/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-173-121:3987796:3988283 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-172-142:789262:789755 [7] NCCL INFO Channel 07/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-172-142:789259:789758 [4] NCCL INFO Channel 20/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-172-142:789260:789759 [5] NCCL INFO Channel 13/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-172-142:789258:789761 [3] NCCL INFO Channel 11/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-172-142:789261:789754 [6] NCCL INFO Channel 15/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-172-142:789255:789751 [0] NCCL INFO Channel 18/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-142:789256:789752 [1] NCCL INFO Channel 18/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789262:789755 [7] NCCL INFO Channel 08/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-172-142:789260:789759 [5] NCCL INFO Channel 14/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-172-142:789259:789758 [4] NCCL INFO Channel 21/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-172-142:789258:789761 [3] NCCL INFO Channel 12/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-172-142:789261:789754 [6] NCCL INFO Channel 16/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-172-142:789255:789751 [0] NCCL INFO Channel 19/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-142:789256:789752 [1] NCCL INFO Channel 19/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789262:789755 [7] NCCL INFO Channel 09/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-172-142:789260:789759 [5] NCCL INFO Channel 15/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-172-142:789259:789758 [4] NCCL INFO Channel 22/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-172-142:789258:789761 [3] NCCL INFO Channel 13/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-172-142:789261:789754 [6] NCCL INFO Channel 17/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-172-142:789256:789752 [1] NCCL INFO Channel 20/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789260:789759 [5] NCCL INFO Channel 16/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-172-142:789259:789758 [4] NCCL INFO Channel 23/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-172-142:789261:789754 [6] NCCL INFO Channel 18/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-172-142:789258:789761 [3] NCCL INFO Channel 14/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-172-142:789256:789752 [1] NCCL INFO Channel 21/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789255:789751 [0] NCCL INFO Channel 20/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-142:789260:789759 [5] NCCL INFO Channel 17/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-172-142:789258:789761 [3] NCCL INFO Channel 15/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-172-142:789261:789754 [6] NCCL INFO Channel 19/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-172-142:789256:789752 [1] NCCL INFO Channel 22/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789255:789751 [0] NCCL INFO Channel 21/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-142:789260:789759 [5] NCCL INFO Channel 18/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-172-142:789262:789755 [7] NCCL INFO Channel 10/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-172-142:789258:789761 [3] NCCL INFO Channel 16/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-172-142:789261:789754 [6] NCCL INFO Channel 20/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-172-142:789255:789751 [0] NCCL INFO Channel 22/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-142:789256:789752 [1] NCCL INFO Channel 23/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789260:789759 [5] NCCL INFO Channel 19/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-172-142:789262:789755 [7] NCCL INFO Channel 11/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-172-142:789261:789754 [6] NCCL INFO Channel 21/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-172-142:789258:789761 [3] NCCL INFO Channel 17/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-172-142:789255:789751 [0] NCCL INFO Channel 23/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-173-121:3987797:3988285 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-173-121:3987796:3988283 [2] NCCL INFO Channel 00/24 :    0   1
+ip-26-0-173-121:3987796:3988283 [2] NCCL INFO Channel 01/24 :    0   1
+ip-26-0-173-121:3987796:3988283 [2] NCCL INFO Channel 02/24 :    0   1
+ip-26-0-173-121:3987796:3988283 [2] NCCL INFO Channel 03/24 :    0   1
+ip-26-0-173-121:3987796:3988283 [2] NCCL INFO Channel 04/24 :    0   1
+ip-26-0-173-121:3987796:3988283 [2] NCCL INFO Channel 05/24 :    0   1
+ip-26-0-173-121:3987796:3988283 [2] NCCL INFO Channel 06/24 :    0   1
+ip-26-0-173-121:3987796:3988283 [2] NCCL INFO Channel 07/24 :    0   1
+ip-26-0-173-121:3987796:3988283 [2] NCCL INFO Channel 08/24 :    0   1
+ip-26-0-173-121:3987796:3988283 [2] NCCL INFO Channel 09/24 :    0   1
+ip-26-0-173-121:3987796:3988283 [2] NCCL INFO Channel 10/24 :    0   1
+ip-26-0-173-121:3987796:3988283 [2] NCCL INFO Channel 11/24 :    0   1
+ip-26-0-173-121:3987796:3988283 [2] NCCL INFO Channel 12/24 :    0   1
+ip-26-0-173-121:3987797:3988285 [3] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] -1/-1/-1->1->0 [3] -1/-1/-1->1->0 [4] -1/-1/-1->1->0 [5] -1/-1/-1->1->0 [6] 0/-1/-1->1->-1 [7] 0/-1/-1->1->-1 [8] 0/-1/-1->1->-1 [9] 0/-1/-1->1->-1 [10] 0/-1/-1->1->-1 [11] 0/-1/-1->1->-1 [12] -1/-1/-1->1->0 [13] -1/-1/-1->1->0 [14] -1/-1/-1->1->0 [15] -1/-1/-1->1->0 [16] -1/-1/-1->1->0 [17] -1/-1/-1->1->0 [18] 0/-1/-1->1->-1 [19] 0/-1/-1->1->-1 [20] 0/-1/-1->1->-1 [21] 0/-1/-1->1->-1 [22] 0/-1/-1->1->-1 [23] 0/-1/-1->1->-1
+ip-26-0-173-121:3987796:3988283 [2] NCCL INFO Channel 13/24 :    0   1
+ip-26-0-173-121:3987796:3988283 [2] NCCL INFO Channel 14/24 :    0   1
+ip-26-0-173-121:3987797:3988285 [3] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-173-121:3987796:3988283 [2] NCCL INFO Channel 15/24 :    0   1
+ip-26-0-173-121:3987796:3988283 [2] NCCL INFO Channel 16/24 :    0   1
+ip-26-0-173-121:3987796:3988283 [2] NCCL INFO Channel 17/24 :    0   1
+ip-26-0-173-121:3987796:3988283 [2] NCCL INFO Channel 18/24 :    0   1
+ip-26-0-173-121:3987796:3988283 [2] NCCL INFO Channel 19/24 :    0   1
+ip-26-0-173-121:3987796:3988283 [2] NCCL INFO Channel 20/24 :    0   1
+ip-26-0-173-121:3987796:3988283 [2] NCCL INFO Channel 21/24 :    0   1
+ip-26-0-173-121:3987796:3988283 [2] NCCL INFO Channel 22/24 :    0   1
+ip-26-0-173-121:3987796:3988283 [2] NCCL INFO Channel 23/24 :    0   1
+ip-26-0-173-121:3987796:3988283 [2] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] -1/-1/-1->0->1 [7] -1/-1/-1->0->1 [8] -1/-1/-1->0->1 [9] -1/-1/-1->0->1 [10] -1/-1/-1->0->1 [11] -1/-1/-1->0->1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] -1/-1/-1->0->1 [19] -1/-1/-1->0->1 [20] -1/-1/-1->0->1 [21] -1/-1/-1->0->1 [22] -1/-1/-1->0->1 [23] -1/-1/-1->0->1
+ip-26-0-173-121:3987796:3988283 [2] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-172-142:789260:789759 [5] NCCL INFO Channel 20/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-172-142:789262:789755 [7] NCCL INFO Channel 12/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-172-142:789261:789754 [6] NCCL INFO Channel 22/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-172-142:789258:789761 [3] NCCL INFO Channel 18/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-172-142:789260:789759 [5] NCCL INFO Channel 21/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-172-142:789262:789755 [7] NCCL INFO Channel 13/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-172-142:789261:789754 [6] NCCL INFO Channel 23/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-172-142:789258:789761 [3] NCCL INFO Channel 19/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-172-142:789260:789759 [5] NCCL INFO Channel 22/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-172-142:789262:789755 [7] NCCL INFO Channel 14/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-172-142:789258:789761 [3] NCCL INFO Channel 20/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-173-121:3987801:3988282 [7] NCCL INFO Channel 00/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-172-142:789260:789759 [5] NCCL INFO Channel 23/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-172-142:789262:789755 [7] NCCL INFO Channel 15/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-172-142:789258:789761 [3] NCCL INFO Channel 21/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-173-121:3987801:3988282 [7] NCCL INFO Channel 01/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-173-121:3987798:3988286 [4] NCCL INFO Channel 00/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-173-121:3987800:3988280 [6] NCCL INFO Channel 00/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-173-121:3987801:3988282 [7] NCCL INFO Channel 02/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-173-121:3987798:3988286 [4] NCCL INFO Channel 01/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-173-121:3987799:3988288 [5] NCCL INFO Channel 00/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-173-121:3987800:3988280 [6] NCCL INFO Channel 01/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-173-121:3987801:3988282 [7] NCCL INFO Channel 03/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-173-121:3987798:3988286 [4] NCCL INFO Channel 02/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-173-121:3987794:3988289 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-173-121:3987799:3988288 [5] NCCL INFO Channel 01/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-173-121:3987800:3988280 [6] NCCL INFO Channel 02/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-173-121:3987801:3988282 [7] NCCL INFO Channel 04/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-173-121:3987795:3988290 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987798:3988286 [4] NCCL INFO Channel 03/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-173-121:3987794:3988289 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-173-121:3987799:3988288 [5] NCCL INFO Channel 02/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-173-121:3987800:3988280 [6] NCCL INFO Channel 03/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-173-121:3987801:3988282 [7] NCCL INFO Channel 05/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-173-121:3987795:3988290 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987798:3988286 [4] NCCL INFO Channel 04/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-173-121:3987794:3988289 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-173-121:3987799:3988288 [5] NCCL INFO Channel 03/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-173-121:3987800:3988280 [6] NCCL INFO Channel 04/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-173-121:3987801:3988282 [7] NCCL INFO Channel 06/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-173-121:3987795:3988290 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987798:3988286 [4] NCCL INFO Channel 05/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-173-121:3987794:3988289 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-173-121:3987799:3988288 [5] NCCL INFO Channel 04/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-173-121:3987800:3988280 [6] NCCL INFO Channel 05/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-173-121:3987801:3988282 [7] NCCL INFO Channel 07/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-173-121:3987795:3988290 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987798:3988286 [4] NCCL INFO Channel 06/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-173-121:3987794:3988289 [0] NCCL INFO Channel 04/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-142:789258:789761 [3] NCCL INFO Channel 22/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-172-142:789262:789755 [7] NCCL INFO Channel 16/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-173-121:3987799:3988288 [5] NCCL INFO Channel 05/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-173-121:3987800:3988280 [6] NCCL INFO Channel 06/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-173-121:3987801:3988282 [7] NCCL INFO Channel 08/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-173-121:3987800:3988280 [6] NCCL INFO Channel 07/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-173-121:3987795:3988290 [1] NCCL INFO Channel 04/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987798:3988286 [4] NCCL INFO Channel 07/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-173-121:3987800:3988280 [6] NCCL INFO Channel 08/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-173-121:3987799:3988288 [5] NCCL INFO Channel 06/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-173-121:3987798:3988286 [4] NCCL INFO Channel 08/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-173-121:3987800:3988280 [6] NCCL INFO Channel 09/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-173-121:3987794:3988289 [0] NCCL INFO Channel 05/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-173-121:3987799:3988288 [5] NCCL INFO Channel 07/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-173-121:3987798:3988286 [4] NCCL INFO Channel 09/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-173-121:3987800:3988280 [6] NCCL INFO Channel 10/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-173-121:3987795:3988290 [1] NCCL INFO Channel 05/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987794:3988289 [0] NCCL INFO Channel 06/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-173-121:3987799:3988288 [5] NCCL INFO Channel 08/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-172-142:789258:789761 [3] NCCL INFO Channel 23/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-173-121:3987801:3988282 [7] NCCL INFO Channel 09/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-173-121:3987798:3988286 [4] NCCL INFO Channel 10/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-173-121:3987800:3988280 [6] NCCL INFO Channel 11/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-173-121:3987795:3988290 [1] NCCL INFO Channel 06/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987794:3988289 [0] NCCL INFO Channel 07/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-173-121:3987799:3988288 [5] NCCL INFO Channel 09/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-173-121:3987801:3988282 [7] NCCL INFO Channel 10/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-173-121:3987800:3988280 [6] NCCL INFO Channel 12/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-173-121:3987795:3988290 [1] NCCL INFO Channel 07/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987797:3988285 [3] NCCL INFO Channel 00/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-173-121:3987801:3988282 [7] NCCL INFO Channel 11/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-173-121:3987797:3988285 [3] NCCL INFO Channel 01/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-173-121:3987794:3988289 [0] NCCL INFO Channel 08/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-173-121:3987799:3988288 [5] NCCL INFO Channel 10/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-173-121:3987798:3988286 [4] NCCL INFO Channel 11/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-173-121:3987800:3988280 [6] NCCL INFO Channel 13/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-173-121:3987797:3988285 [3] NCCL INFO Channel 02/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-173-121:3987796:3988283 [2] NCCL INFO Channel 00/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-173-121:3987795:3988290 [1] NCCL INFO Channel 08/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987801:3988282 [7] NCCL INFO Channel 12/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-173-121:3987800:3988280 [6] NCCL INFO Channel 14/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-173-121:3987796:3988283 [2] NCCL INFO Channel 01/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-173-121:3987797:3988285 [3] NCCL INFO Channel 03/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-173-121:3987795:3988290 [1] NCCL INFO Channel 09/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987801:3988282 [7] NCCL INFO Channel 13/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-173-121:3987794:3988289 [0] NCCL INFO Channel 09/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-173-121:3987801:3988282 [7] NCCL INFO Channel 14/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-173-121:3987798:3988286 [4] NCCL INFO Channel 12/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-173-121:3987794:3988289 [0] NCCL INFO Channel 10/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-173-121:3987800:3988280 [6] NCCL INFO Channel 15/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-173-121:3987801:3988282 [7] NCCL INFO Channel 15/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-173-121:3987798:3988286 [4] NCCL INFO Channel 13/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-173-121:3987794:3988289 [0] NCCL INFO Channel 11/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-173-121:3987800:3988280 [6] NCCL INFO Channel 16/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-173-121:3987801:3988282 [7] NCCL INFO Channel 16/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-173-121:3987798:3988286 [4] NCCL INFO Channel 14/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-173-121:3987794:3988289 [0] NCCL INFO Channel 12/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-173-121:3987799:3988288 [5] NCCL INFO Channel 11/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-173-121:3987800:3988280 [6] NCCL INFO Channel 17/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-173-121:3987796:3988283 [2] NCCL INFO Channel 02/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-173-121:3987797:3988285 [3] NCCL INFO Channel 04/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-173-121:3987795:3988290 [1] NCCL INFO Channel 10/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987798:3988286 [4] NCCL INFO Channel 15/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-173-121:3987801:3988282 [7] NCCL INFO Channel 17/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-173-121:3987794:3988289 [0] NCCL INFO Channel 13/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-173-121:3987799:3988288 [5] NCCL INFO Channel 12/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-173-121:3987800:3988280 [6] NCCL INFO Channel 18/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-173-121:3987796:3988283 [2] NCCL INFO Channel 03/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-173-121:3987797:3988285 [3] NCCL INFO Channel 05/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-173-121:3987795:3988290 [1] NCCL INFO Channel 11/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987798:3988286 [4] NCCL INFO Channel 16/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-173-121:3987794:3988289 [0] NCCL INFO Channel 14/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-173-121:3987801:3988282 [7] NCCL INFO Channel 18/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-173-121:3987799:3988288 [5] NCCL INFO Channel 13/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-173-121:3987800:3988280 [6] NCCL INFO Channel 19/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-173-121:3987796:3988283 [2] NCCL INFO Channel 04/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-173-121:3987797:3988285 [3] NCCL INFO Channel 06/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-173-121:3987795:3988290 [1] NCCL INFO Channel 12/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987798:3988286 [4] NCCL INFO Channel 17/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-173-121:3987794:3988289 [0] NCCL INFO Channel 15/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-173-121:3987799:3988288 [5] NCCL INFO Channel 14/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-173-121:3987801:3988282 [7] NCCL INFO Channel 19/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-173-121:3987800:3988280 [6] NCCL INFO Channel 20/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-173-121:3987796:3988283 [2] NCCL INFO Channel 05/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-173-121:3987795:3988290 [1] NCCL INFO Channel 13/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987794:3988289 [0] NCCL INFO Channel 16/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-173-121:3987797:3988285 [3] NCCL INFO Channel 07/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-173-121:3987798:3988286 [4] NCCL INFO Channel 18/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-173-121:3987801:3988282 [7] NCCL INFO Channel 20/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-173-121:3987799:3988288 [5] NCCL INFO Channel 15/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-173-121:3987800:3988280 [6] NCCL INFO Channel 21/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-173-121:3987796:3988283 [2] NCCL INFO Channel 06/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-173-121:3987797:3988285 [3] NCCL INFO Channel 08/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-173-121:3987795:3988290 [1] NCCL INFO Channel 14/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987798:3988286 [4] NCCL INFO Channel 19/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-173-121:3987794:3988289 [0] NCCL INFO Channel 17/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-173-121:3987799:3988288 [5] NCCL INFO Channel 16/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-173-121:3987801:3988282 [7] NCCL INFO Channel 21/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-173-121:3987800:3988280 [6] NCCL INFO Channel 22/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-173-121:3987796:3988283 [2] NCCL INFO Channel 07/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-173-121:3987797:3988285 [3] NCCL INFO Channel 09/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-173-121:3987798:3988286 [4] NCCL INFO Channel 20/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-173-121:3987795:3988290 [1] NCCL INFO Channel 15/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987794:3988289 [0] NCCL INFO Channel 18/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-173-121:3987801:3988282 [7] NCCL INFO Channel 22/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-173-121:3987799:3988288 [5] NCCL INFO Channel 17/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-173-121:3987800:3988280 [6] NCCL INFO Channel 23/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-173-121:3987796:3988283 [2] NCCL INFO Channel 08/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-173-121:3987798:3988286 [4] NCCL INFO Channel 21/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-173-121:3987797:3988285 [3] NCCL INFO Channel 10/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-173-121:3987795:3988290 [1] NCCL INFO Channel 16/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987794:3988289 [0] NCCL INFO Channel 19/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-173-121:3987801:3988282 [7] NCCL INFO Channel 23/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-173-121:3987799:3988288 [5] NCCL INFO Channel 18/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-173-121:3987796:3988283 [2] NCCL INFO Channel 09/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-173-121:3987798:3988286 [4] NCCL INFO Channel 22/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-173-121:3987797:3988285 [3] NCCL INFO Channel 11/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-173-121:3987794:3988289 [0] NCCL INFO Channel 20/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-173-121:3987799:3988288 [5] NCCL INFO Channel 19/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-173-121:3987795:3988290 [1] NCCL INFO Channel 17/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987798:3988286 [4] NCCL INFO Channel 23/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-173-121:3987796:3988283 [2] NCCL INFO Channel 10/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-173-121:3987799:3988288 [5] NCCL INFO Channel 20/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-173-121:3987797:3988285 [3] NCCL INFO Channel 12/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-173-121:3987794:3988289 [0] NCCL INFO Channel 21/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-173-121:3987797:3988285 [3] NCCL INFO Channel 13/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-172-142:789262:789755 [7] NCCL INFO Channel 17/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-173-121:3987797:3988285 [3] NCCL INFO Channel 14/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-173-121:3987797:3988285 [3] NCCL INFO Channel 15/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-173-121:3987796:3988283 [2] NCCL INFO Channel 11/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-173-121:3987795:3988290 [1] NCCL INFO Channel 18/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987799:3988288 [5] NCCL INFO Channel 21/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-173-121:3987794:3988289 [0] NCCL INFO Channel 22/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-173-121:3987797:3988285 [3] NCCL INFO Channel 16/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-173-121:3987796:3988283 [2] NCCL INFO Channel 12/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-173-121:3987795:3988290 [1] NCCL INFO Channel 19/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987799:3988288 [5] NCCL INFO Channel 22/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-173-121:3987794:3988289 [0] NCCL INFO Channel 23/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-173-121:3987797:3988285 [3] NCCL INFO Channel 17/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-172-142:789262:789755 [7] NCCL INFO Channel 18/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-173-121:3987796:3988283 [2] NCCL INFO Channel 13/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-173-121:3987799:3988288 [5] NCCL INFO Channel 23/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-173-121:3987797:3988285 [3] NCCL INFO Channel 18/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-173-121:3987796:3988283 [2] NCCL INFO Channel 14/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-173-121:3987795:3988290 [1] NCCL INFO Channel 20/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789262:789755 [7] NCCL INFO Channel 19/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-173-121:3987797:3988285 [3] NCCL INFO Channel 19/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-173-121:3987796:3988283 [2] NCCL INFO Channel 15/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-173-121:3987795:3988290 [1] NCCL INFO Channel 21/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789262:789755 [7] NCCL INFO Channel 20/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-173-121:3987796:3988283 [2] NCCL INFO Channel 16/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-173-121:3987795:3988290 [1] NCCL INFO Channel 22/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789257:789760 [2] NCCL INFO Channel 00/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-172-142:789262:789755 [7] NCCL INFO Channel 21/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-173-121:3987797:3988285 [3] NCCL INFO Channel 20/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-172-142:789262:789755 [7] NCCL INFO Channel 22/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-173-121:3987795:3988290 [1] NCCL INFO Channel 23/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987796:3988283 [2] NCCL INFO Channel 17/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-172-142:789257:789760 [2] NCCL INFO Channel 01/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-172-142:789262:789755 [7] NCCL INFO Channel 23/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-173-121:3987797:3988285 [3] NCCL INFO Channel 21/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-173-121:3987796:3988283 [2] NCCL INFO Channel 18/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-172-142:789257:789760 [2] NCCL INFO Channel 02/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-172-142:789257:789760 [2] NCCL INFO Channel 03/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-173-121:3987796:3988283 [2] NCCL INFO Channel 19/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-173-121:3987797:3988285 [3] NCCL INFO Channel 22/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-172-142:789257:789760 [2] NCCL INFO Channel 04/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-173-121:3987796:3988283 [2] NCCL INFO Channel 20/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-172-142:789257:789760 [2] NCCL INFO Channel 05/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-173-121:3987797:3988285 [3] NCCL INFO Channel 23/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-173-121:3987796:3988283 [2] NCCL INFO Channel 21/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-172-142:789257:789760 [2] NCCL INFO Channel 06/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-173-121:3987796:3988283 [2] NCCL INFO Channel 22/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-172-142:789257:789760 [2] NCCL INFO Channel 07/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-173-121:3987796:3988283 [2] NCCL INFO Channel 23/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-172-142:789257:789760 [2] NCCL INFO Channel 08/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-172-142:789257:789760 [2] NCCL INFO Channel 09/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-172-142:789257:789760 [2] NCCL INFO Channel 10/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-172-142:789257:789760 [2] NCCL INFO Channel 11/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-172-142:789257:789760 [2] NCCL INFO Channel 12/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-172-142:789257:789760 [2] NCCL INFO Channel 13/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-172-142:789257:789760 [2] NCCL INFO Channel 14/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-172-142:789257:789760 [2] NCCL INFO Channel 15/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-172-142:789260:789759 [5] NCCL INFO Connected all rings
+ip-26-0-172-142:789260:789759 [5] NCCL INFO Connected all trees
+ip-26-0-172-142:789260:789759 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-142:789260:789759 [5] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-172-142:789260:789759 [5] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-142:789257:789760 [2] NCCL INFO Channel 16/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-172-142:789255:789751 [0] NCCL INFO Connected all rings
+ip-26-0-172-142:789255:789751 [0] NCCL INFO Connected all trees
+ip-26-0-172-142:789255:789751 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-142:789255:789751 [0] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-172-142:789255:789751 [0] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-142:789257:789760 [2] NCCL INFO Channel 17/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-172-142:789257:789760 [2] NCCL INFO Channel 18/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-172-142:789257:789760 [2] NCCL INFO Channel 19/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-172-142:789256:789752 [1] NCCL INFO Connected all rings
+ip-26-0-172-142:789256:789752 [1] NCCL INFO Connected all trees
+ip-26-0-172-142:789256:789752 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-142:789256:789752 [1] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-172-142:789256:789752 [1] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-142:789259:789758 [4] NCCL INFO Connected all rings
+ip-26-0-172-142:789259:789758 [4] NCCL INFO Connected all trees
+ip-26-0-172-142:789259:789758 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-142:789259:789758 [4] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-172-142:789259:789758 [4] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-142:789257:789760 [2] NCCL INFO Channel 20/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-172-142:789257:789760 [2] NCCL INFO Channel 21/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-172-142:789257:789760 [2] NCCL INFO Channel 22/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-172-142:789257:789760 [2] NCCL INFO Channel 23/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-172-142:789256:789752 [1] NCCL INFO comm 0xc443720 rank 1 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0xcb34a8a225988e45 - Init COMPLETE
+ip-26-0-172-142:789255:789751 [0] NCCL INFO comm 0xc0f5730 rank 0 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0xcb34a8a225988e45 - Init COMPLETE
+ip-26-0-172-142:789260:789759 [5] NCCL INFO comm 0xc16c4b0 rank 1 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0xa927b92c1cfcdeb3 - Init COMPLETE
+ip-26-0-172-142:789259:789758 [4] NCCL INFO comm 0xb146d40 rank 0 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0xa927b92c1cfcdeb3 - Init COMPLETE
+ip-26-0-173-121:3987800:3988280 [6] NCCL INFO Connected all rings
+ip-26-0-173-121:3987800:3988280 [6] NCCL INFO Connected all trees
+ip-26-0-173-121:3987800:3988280 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-173-121:3987800:3988280 [6] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-173-121:3987800:3988280 [6] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-173-121:3987801:3988282 [7] NCCL INFO Connected all rings
+ip-26-0-173-121:3987801:3988282 [7] NCCL INFO Connected all trees
+ip-26-0-173-121:3987801:3988282 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-173-121:3987801:3988282 [7] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-173-121:3987801:3988282 [7] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-142:789262:789755 [7] NCCL INFO Connected all rings
+ip-26-0-172-142:789262:789755 [7] NCCL INFO Connected all trees
+ip-26-0-172-142:789262:789755 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-142:789262:789755 [7] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-172-142:789262:789755 [7] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-142:789261:789754 [6] NCCL INFO Connected all rings
+ip-26-0-172-142:789261:789754 [6] NCCL INFO Connected all trees
+ip-26-0-172-142:789261:789754 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-142:789261:789754 [6] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-172-142:789261:789754 [6] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-142:789262:789755 [7] NCCL INFO comm 0xab05550 rank 1 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0xc64be2bc3243716e - Init COMPLETE
+ip-26-0-172-142:789261:789754 [6] NCCL INFO comm 0xbee8650 rank 0 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0xc64be2bc3243716e - Init COMPLETE
+ip-26-0-173-121:3987800:3988280 [6] NCCL INFO comm 0xa8fd180 rank 0 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0x30139afce86f8b55 - Init COMPLETE
+ip-26-0-173-121:3987801:3988282 [7] NCCL INFO comm 0xad0d1a0 rank 1 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0x30139afce86f8b55 - Init COMPLETE
+ip-26-0-172-142:789255:789781 [0] NCCL INFO Using network Libfabric
+ip-26-0-173-121:3987800:3988305 [6] NCCL INFO Using network Libfabric
+ip-26-0-172-142:789256:789783 [1] NCCL INFO Using network Libfabric
+ip-26-0-173-121:3987801:3988306 [7] NCCL INFO Using network Libfabric
+ip-26-0-173-121:3987800:3988305 [6] NCCL INFO comm 0xa921230 rank 1 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0x8bee938409b2171a - Init START
+ip-26-0-172-142:789255:789781 [0] NCCL INFO comm 0xc11c380 rank 0 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0x8bee938409b2171a - Init START
+ip-26-0-172-142:789255:789781 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/205
+ip-26-0-173-121:3987800:3988305 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-173-121:3987801:3988306 [7] NCCL INFO comm 0xad30bc0 rank 1 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0x3711094657b99ae8 - Init START
+ip-26-0-172-142:789256:789783 [1] NCCL INFO comm 0xc46be60 rank 0 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0x3711094657b99ae8 - Init START
+ip-26-0-172-142:789256:789783 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-173-121:3987801:3988306 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-173-121:3987798:3988286 [4] NCCL INFO Connected all rings
+ip-26-0-173-121:3987798:3988286 [4] NCCL INFO Connected all trees
+ip-26-0-173-121:3987798:3988286 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-173-121:3987798:3988286 [4] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-173-121:3987798:3988286 [4] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-173-121:3987794:3988289 [0] NCCL INFO Connected all rings
+ip-26-0-173-121:3987794:3988289 [0] NCCL INFO Connected all trees
+ip-26-0-173-121:3987794:3988289 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-173-121:3987794:3988289 [0] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-173-121:3987794:3988289 [0] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-173-121:3987799:3988288 [5] NCCL INFO Connected all rings
+ip-26-0-173-121:3987799:3988288 [5] NCCL INFO Connected all trees
+ip-26-0-173-121:3987799:3988288 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-173-121:3987799:3988288 [5] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-173-121:3987799:3988288 [5] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-173-121:3987795:3988290 [1] NCCL INFO Connected all rings
+ip-26-0-173-121:3987795:3988290 [1] NCCL INFO Connected all trees
+ip-26-0-173-121:3987795:3988290 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-173-121:3987795:3988290 [1] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-173-121:3987795:3988290 [1] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-173-121:3987798:3988286 [4] NCCL INFO comm 0xbcd5370 rank 0 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0x11e980de02d689a1 - Init COMPLETE
+ip-26-0-173-121:3987799:3988288 [5] NCCL INFO comm 0xb2d6390 rank 1 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0x11e980de02d689a1 - Init COMPLETE
+ip-26-0-173-121:3987795:3988290 [1] NCCL INFO comm 0xb90bdf0 rank 1 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0x2ae6027f9d90b4c3 - Init COMPLETE
+ip-26-0-173-121:3987794:3988289 [0] NCCL INFO comm 0xad48930 rank 0 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0x2ae6027f9d90b4c3 - Init COMPLETE
+ip-26-0-173-121:3987797:3988285 [3] NCCL INFO Connected all rings
+ip-26-0-173-121:3987797:3988285 [3] NCCL INFO Connected all trees
+ip-26-0-173-121:3987797:3988285 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-173-121:3987797:3988285 [3] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-173-121:3987797:3988285 [3] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-142:789257:789760 [2] NCCL INFO Connected all rings
+ip-26-0-172-142:789257:789760 [2] NCCL INFO Connected all trees
+ip-26-0-172-142:789257:789760 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-142:789257:789760 [2] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-172-142:789257:789760 [2] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-142:789258:789761 [3] NCCL INFO Connected all rings
+ip-26-0-172-142:789258:789761 [3] NCCL INFO Connected all trees
+ip-26-0-172-142:789258:789761 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-142:789258:789761 [3] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-172-142:789258:789761 [3] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-173-121:3987796:3988283 [2] NCCL INFO Connected all rings
+ip-26-0-173-121:3987796:3988283 [2] NCCL INFO Connected all trees
+ip-26-0-173-121:3987796:3988283 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-173-121:3987796:3988283 [2] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-173-121:3987796:3988283 [2] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-142:789257:789760 [2] NCCL INFO comm 0xb4a29d0 rank 0 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0x322512f564e4938c - Init COMPLETE
+ip-26-0-172-142:789258:789761 [3] NCCL INFO comm 0xac27860 rank 1 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0x322512f564e4938c - Init COMPLETE
+ip-26-0-173-121:3987796:3988283 [2] NCCL INFO comm 0xb1a2bc0 rank 0 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0xa12c6c01388e3f90 - Init COMPLETE
+ip-26-0-173-121:3987797:3988285 [3] NCCL INFO comm 0xac794c0 rank 1 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0xa12c6c01388e3f90 - Init COMPLETE
+ip-26-0-172-142:789255:789781 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-172-142:789256:789783 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-173-121:3987801:3988306 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-173-121:3987801:3988306 [7] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-173-121:3987801:3988306 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-142:789256:789783 [1] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-172-142:789256:789783 [1] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-172-142:789256:789783 [1] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-172-142:789256:789783 [1] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-172-142:789256:789783 [1] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-172-142:789256:789783 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-173-121:3987800:3988305 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-173-121:3987800:3988305 [6] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-173-121:3987800:3988305 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-142:789255:789781 [0] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-172-142:789255:789781 [0] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-172-142:789255:789781 [0] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-172-142:789255:789781 [0] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-172-142:789255:789781 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-172-142:789255:789781 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-142:789256:789783 [1] NCCL INFO Channel 00/0 : 1[7] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:789256:789783 [1] NCCL INFO Channel 01/0 : 1[7] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:789256:789783 [1] NCCL INFO Channel 02/0 : 1[7] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:789256:789783 [1] NCCL INFO Channel 03/0 : 1[7] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:789256:789783 [1] NCCL INFO Channel 00/0 : 0[1] -> 1[7] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:789256:789783 [1] NCCL INFO Channel 01/0 : 0[1] -> 1[7] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:789256:789783 [1] NCCL INFO Channel 02/0 : 0[1] -> 1[7] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:789256:789783 [1] NCCL INFO Channel 03/0 : 0[1] -> 1[7] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-121:3987801:3988306 [7] NCCL INFO Channel 00/0 : 0[1] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:789255:789781 [0] NCCL INFO Channel 00/0 : 1[6] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-121:3987801:3988306 [7] NCCL INFO Channel 01/0 : 0[1] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:789255:789781 [0] NCCL INFO Channel 01/0 : 1[6] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-121:3987801:3988306 [7] NCCL INFO Channel 02/0 : 0[1] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:789255:789781 [0] NCCL INFO Channel 02/0 : 1[6] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:789255:789781 [0] NCCL INFO Channel 03/0 : 1[6] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-121:3987801:3988306 [7] NCCL INFO Channel 03/0 : 0[1] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:789255:789781 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[6] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-121:3987801:3988306 [7] NCCL INFO Channel 00/0 : 1[7] -> 0[1] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-121:3987801:3988306 [7] NCCL INFO Channel 01/0 : 1[7] -> 0[1] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:789255:789781 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[6] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-121:3987801:3988306 [7] NCCL INFO Channel 02/0 : 1[7] -> 0[1] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:789255:789781 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[6] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-121:3987801:3988306 [7] NCCL INFO Channel 03/0 : 1[7] -> 0[1] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:789255:789781 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[6] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-121:3987800:3988305 [6] NCCL INFO Channel 00/0 : 0[0] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-121:3987800:3988305 [6] NCCL INFO Channel 01/0 : 0[0] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-121:3987800:3988305 [6] NCCL INFO Channel 02/0 : 0[0] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-121:3987800:3988305 [6] NCCL INFO Channel 03/0 : 0[0] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-121:3987800:3988305 [6] NCCL INFO Channel 00/0 : 1[6] -> 0[0] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-121:3987800:3988305 [6] NCCL INFO Channel 01/0 : 1[6] -> 0[0] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-121:3987800:3988305 [6] NCCL INFO Channel 02/0 : 1[6] -> 0[0] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-121:3987800:3988305 [6] NCCL INFO Channel 03/0 : 1[6] -> 0[0] [send] via NET/Libfabric/6/GDRDMA
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+ip-26-0-172-142:789256:789783 [1] NCCL INFO Connected all rings
+ip-26-0-172-142:789256:789783 [1] NCCL INFO Connected all trees
+ip-26-0-172-142:789256:789783 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-142:789256:789783 [1] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-172-142:789256:789783 [1] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-142:789255:789781 [0] NCCL INFO Connected all rings
+ip-26-0-172-142:789255:789781 [0] NCCL INFO Connected all trees
+ip-26-0-172-142:789255:789781 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-142:789255:789781 [0] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-172-142:789255:789781 [0] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-142:789256:789783 [1] NCCL INFO comm 0xc46be60 rank 0 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0x3711094657b99ae8 - Init COMPLETE
+ip-26-0-172-142:789255:789781 [0] NCCL INFO comm 0xc11c380 rank 0 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0x8bee938409b2171a - Init COMPLETE
+01/07/2025 00:37:12 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]: [Optimizer Building] Using LearningRateForSP as learning rate
+01/07/2025 00:37:12 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]: [ZeRO sharding] Size of optimizer params per rank:
+01/07/2025 00:37:12 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]: [ZeRO sharding] DP Rank 0 has 235M out of 470M (50.00%) params' optimizer states
+01/07/2025 00:37:12 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]: [ZeRO sharding] DP Rank 1 has 235M out of 470M (50.00%) params' optimizer states
+ip-26-0-173-121:3987801:3988306 [7] NCCL INFO Connected all rings
+ip-26-0-173-121:3987801:3988306 [7] NCCL INFO Connected all trees
+ip-26-0-173-121:3987801:3988306 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-173-121:3987801:3988306 [7] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-173-121:3987801:3988306 [7] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-173-121:3987800:3988305 [6] NCCL INFO Connected all rings
+ip-26-0-173-121:3987800:3988305 [6] NCCL INFO Connected all trees
+ip-26-0-173-121:3987800:3988305 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-173-121:3987800:3988305 [6] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-173-121:3987800:3988305 [6] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-173-121:3987801:3988306 [7] NCCL INFO comm 0xad30bc0 rank 1 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0x3711094657b99ae8 - Init COMPLETE
+ip-26-0-173-121:3987800:3988305 [6] NCCL INFO comm 0xa921230 rank 1 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0x8bee938409b2171a - Init COMPLETE
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+01/07/2025 00:37:13 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]: [Training Plan] Stage Stable Training Stage has 99 remaining training steps and has consumed 0 samples
+01/07/2025 00:37:13 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]: Using dummy data generator
+01/07/2025 00:37:13 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]: [Training Plan] There are 1 training stages 
+01/07/2025 00:37:13 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]: [Stage Stable Training Stage] start from step 1 
+01/07/2025 00:37:13 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]: 
+01/07/2025 00:37:13 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]: [Start training] datetime: 2025-01-07 00:37:13.236205 | mbs: 2 | grad_accum: 64 | global_batch_size: 256 | sequence_length: 4096 | train_steps: 100 | start_iteration_step: 0 | consumed_train_samples: 0
+01/07/2025 00:37:13 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]: Resuming training from stage Stable Training Stage, it has trained for 0 samples and has 99 remaining train steps
+01/07/2025 00:37:13 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]:  Memory usage: 3584.11MiB. Peak allocated 5504.00MiB. Peak reserved: 15492.00MiB
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+ip-26-0-172-142:789255:789793 [0] NCCL INFO Using network Libfabric
+ip-26-0-172-142:789257:789794 [2] NCCL INFO Using network Libfabric
+ip-26-0-172-142:789257:789794 [2] NCCL INFO comm 0xb70abb0 rank 1 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0x60f60f4766fdf198 - Init START
+ip-26-0-172-142:789255:789793 [0] NCCL INFO comm 0x179bdb70 rank 0 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0x60f60f4766fdf198 - Init START
+ip-26-0-172-142:789257:789794 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-172-142:789255:789793 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/205
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+ip-26-0-172-142:789256:789796 [1] NCCL INFO Using network Libfabric
+ip-26-0-172-142:789258:789797 [3] NCCL INFO Using network Libfabric
+ip-26-0-172-142:789258:789797 [3] NCCL INFO comm 0xae145d0 rank 1 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0xd05269a24be85487 - Init START
+ip-26-0-172-142:789256:789796 [1] NCCL INFO comm 0x17d11d80 rank 0 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0xd05269a24be85487 - Init START
+ip-26-0-172-142:789258:789797 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-172-142:789256:789796 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-172-142:789257:789794 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-172-142:789255:789793 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-172-142:789255:789793 [0] NCCL INFO Channel 00/24 :    0   1
+ip-26-0-172-142:789255:789793 [0] NCCL INFO Channel 01/24 :    0   1
+ip-26-0-172-142:789255:789793 [0] NCCL INFO Channel 02/24 :    0   1
+ip-26-0-172-142:789255:789793 [0] NCCL INFO Channel 03/24 :    0   1
+ip-26-0-172-142:789255:789793 [0] NCCL INFO Channel 04/24 :    0   1
+ip-26-0-172-142:789255:789793 [0] NCCL INFO Channel 05/24 :    0   1
+ip-26-0-172-142:789255:789793 [0] NCCL INFO Channel 06/24 :    0   1
+ip-26-0-172-142:789255:789793 [0] NCCL INFO Channel 07/24 :    0   1
+ip-26-0-172-142:789255:789793 [0] NCCL INFO Channel 08/24 :    0   1
+ip-26-0-172-142:789255:789793 [0] NCCL INFO Channel 09/24 :    0   1
+ip-26-0-172-142:789255:789793 [0] NCCL INFO Channel 10/24 :    0   1
+ip-26-0-172-142:789255:789793 [0] NCCL INFO Channel 11/24 :    0   1
+ip-26-0-172-142:789255:789793 [0] NCCL INFO Channel 12/24 :    0   1
+ip-26-0-172-142:789255:789793 [0] NCCL INFO Channel 13/24 :    0   1
+ip-26-0-172-142:789255:789793 [0] NCCL INFO Channel 14/24 :    0   1
+ip-26-0-172-142:789257:789794 [2] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] -1/-1/-1->1->0 [3] -1/-1/-1->1->0 [4] -1/-1/-1->1->0 [5] -1/-1/-1->1->0 [6] 0/-1/-1->1->-1 [7] 0/-1/-1->1->-1 [8] 0/-1/-1->1->-1 [9] 0/-1/-1->1->-1 [10] 0/-1/-1->1->-1 [11] 0/-1/-1->1->-1 [12] -1/-1/-1->1->0 [13] -1/-1/-1->1->0 [14] -1/-1/-1->1->0 [15] -1/-1/-1->1->0 [16] -1/-1/-1->1->0 [17] -1/-1/-1->1->0 [18] 0/-1/-1->1->-1 [19] 0/-1/-1->1->-1 [20] 0/-1/-1->1->-1 [21] 0/-1/-1->1->-1 [22] 0/-1/-1->1->-1 [23] 0/-1/-1->1->-1
+ip-26-0-172-142:789255:789793 [0] NCCL INFO Channel 15/24 :    0   1
+ip-26-0-172-142:789255:789793 [0] NCCL INFO Channel 16/24 :    0   1
+ip-26-0-172-142:789255:789793 [0] NCCL INFO Channel 17/24 :    0   1
+ip-26-0-172-142:789255:789793 [0] NCCL INFO Channel 18/24 :    0   1
+ip-26-0-172-142:789255:789793 [0] NCCL INFO Channel 19/24 :    0   1
+ip-26-0-172-142:789257:789794 [2] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-172-142:789255:789793 [0] NCCL INFO Channel 20/24 :    0   1
+ip-26-0-172-142:789255:789793 [0] NCCL INFO Channel 21/24 :    0   1
+ip-26-0-172-142:789255:789793 [0] NCCL INFO Channel 22/24 :    0   1
+ip-26-0-172-142:789255:789793 [0] NCCL INFO Channel 23/24 :    0   1
+ip-26-0-172-142:789255:789793 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] -1/-1/-1->0->1 [7] -1/-1/-1->0->1 [8] -1/-1/-1->0->1 [9] -1/-1/-1->0->1 [10] -1/-1/-1->0->1 [11] -1/-1/-1->0->1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] -1/-1/-1->0->1 [19] -1/-1/-1->0->1 [20] -1/-1/-1->0->1 [21] -1/-1/-1->0->1 [22] -1/-1/-1->0->1 [23] -1/-1/-1->0->1
+ip-26-0-172-142:789255:789793 [0] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-172-142:789258:789797 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-172-142:789256:789796 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-172-142:789256:789796 [1] NCCL INFO Channel 00/24 :    0   1
+ip-26-0-172-142:789256:789796 [1] NCCL INFO Channel 01/24 :    0   1
+ip-26-0-172-142:789256:789796 [1] NCCL INFO Channel 02/24 :    0   1
+ip-26-0-172-142:789256:789796 [1] NCCL INFO Channel 03/24 :    0   1
+ip-26-0-172-142:789256:789796 [1] NCCL INFO Channel 04/24 :    0   1
+ip-26-0-172-142:789256:789796 [1] NCCL INFO Channel 05/24 :    0   1
+ip-26-0-172-142:789256:789796 [1] NCCL INFO Channel 06/24 :    0   1
+ip-26-0-172-142:789256:789796 [1] NCCL INFO Channel 07/24 :    0   1
+ip-26-0-172-142:789256:789796 [1] NCCL INFO Channel 08/24 :    0   1
+ip-26-0-172-142:789256:789796 [1] NCCL INFO Channel 09/24 :    0   1
+ip-26-0-172-142:789256:789796 [1] NCCL INFO Channel 10/24 :    0   1
+ip-26-0-172-142:789256:789796 [1] NCCL INFO Channel 11/24 :    0   1
+ip-26-0-172-142:789256:789796 [1] NCCL INFO Channel 12/24 :    0   1
+ip-26-0-172-142:789258:789797 [3] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] -1/-1/-1->1->0 [3] -1/-1/-1->1->0 [4] -1/-1/-1->1->0 [5] -1/-1/-1->1->0 [6] 0/-1/-1->1->-1 [7] 0/-1/-1->1->-1 [8] 0/-1/-1->1->-1 [9] 0/-1/-1->1->-1 [10] 0/-1/-1->1->-1 [11] 0/-1/-1->1->-1 [12] -1/-1/-1->1->0 [13] -1/-1/-1->1->0 [14] -1/-1/-1->1->0 [15] -1/-1/-1->1->0 [16] -1/-1/-1->1->0 [17] -1/-1/-1->1->0 [18] 0/-1/-1->1->-1 [19] 0/-1/-1->1->-1 [20] 0/-1/-1->1->-1 [21] 0/-1/-1->1->-1 [22] 0/-1/-1->1->-1 [23] 0/-1/-1->1->-1
+ip-26-0-172-142:789256:789796 [1] NCCL INFO Channel 13/24 :    0   1
+ip-26-0-172-142:789256:789796 [1] NCCL INFO Channel 14/24 :    0   1
+ip-26-0-172-142:789258:789797 [3] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-172-142:789256:789796 [1] NCCL INFO Channel 15/24 :    0   1
+ip-26-0-172-142:789256:789796 [1] NCCL INFO Channel 16/24 :    0   1
+ip-26-0-172-142:789256:789796 [1] NCCL INFO Channel 17/24 :    0   1
+ip-26-0-172-142:789256:789796 [1] NCCL INFO Channel 18/24 :    0   1
+ip-26-0-172-142:789256:789796 [1] NCCL INFO Channel 19/24 :    0   1
+ip-26-0-172-142:789256:789796 [1] NCCL INFO Channel 20/24 :    0   1
+ip-26-0-172-142:789256:789796 [1] NCCL INFO Channel 21/24 :    0   1
+ip-26-0-172-142:789256:789796 [1] NCCL INFO Channel 22/24 :    0   1
+ip-26-0-172-142:789256:789796 [1] NCCL INFO Channel 23/24 :    0   1
+ip-26-0-172-142:789256:789796 [1] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] -1/-1/-1->0->1 [7] -1/-1/-1->0->1 [8] -1/-1/-1->0->1 [9] -1/-1/-1->0->1 [10] -1/-1/-1->0->1 [11] -1/-1/-1->0->1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] -1/-1/-1->0->1 [19] -1/-1/-1->0->1 [20] -1/-1/-1->0->1 [21] -1/-1/-1->0->1 [22] -1/-1/-1->0->1 [23] -1/-1/-1->0->1
+ip-26-0-172-142:789256:789796 [1] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-172-142:789255:789793 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-172-142:789257:789794 [2] NCCL INFO Channel 00/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789257:789794 [2] NCCL INFO Channel 01/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789255:789793 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-172-142:789257:789794 [2] NCCL INFO Channel 02/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789257:789794 [2] NCCL INFO Channel 03/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789257:789794 [2] NCCL INFO Channel 04/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789257:789794 [2] NCCL INFO Channel 05/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789257:789794 [2] NCCL INFO Channel 06/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789257:789794 [2] NCCL INFO Channel 07/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789255:789793 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-172-142:789257:789794 [2] NCCL INFO Channel 08/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789255:789793 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-172-142:789257:789794 [2] NCCL INFO Channel 09/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789255:789793 [0] NCCL INFO Channel 04/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-172-142:789257:789794 [2] NCCL INFO Channel 10/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789255:789793 [0] NCCL INFO Channel 05/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-172-142:789257:789794 [2] NCCL INFO Channel 11/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789255:789793 [0] NCCL INFO Channel 06/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-172-142:789257:789794 [2] NCCL INFO Channel 12/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789255:789793 [0] NCCL INFO Channel 07/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-172-142:789257:789794 [2] NCCL INFO Channel 13/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789255:789793 [0] NCCL INFO Channel 08/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-172-142:789257:789794 [2] NCCL INFO Channel 14/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789255:789793 [0] NCCL INFO Channel 09/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-172-142:789257:789794 [2] NCCL INFO Channel 15/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789255:789793 [0] NCCL INFO Channel 10/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-172-142:789257:789794 [2] NCCL INFO Channel 16/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789255:789793 [0] NCCL INFO Channel 11/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-172-142:789257:789794 [2] NCCL INFO Channel 17/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789255:789793 [0] NCCL INFO Channel 12/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-172-142:789256:789796 [1] NCCL INFO Channel 00/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-172-142:789257:789794 [2] NCCL INFO Channel 18/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789255:789793 [0] NCCL INFO Channel 13/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-172-142:789256:789796 [1] NCCL INFO Channel 01/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-172-142:789257:789794 [2] NCCL INFO Channel 19/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789255:789793 [0] NCCL INFO Channel 14/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-172-142:789256:789796 [1] NCCL INFO Channel 02/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-172-142:789257:789794 [2] NCCL INFO Channel 20/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789258:789797 [3] NCCL INFO Channel 00/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-172-142:789255:789793 [0] NCCL INFO Channel 15/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-172-142:789256:789796 [1] NCCL INFO Channel 03/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-172-142:789257:789794 [2] NCCL INFO Channel 21/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789258:789797 [3] NCCL INFO Channel 01/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-172-142:789255:789793 [0] NCCL INFO Channel 16/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-172-142:789258:789797 [3] NCCL INFO Channel 02/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-172-142:789256:789796 [1] NCCL INFO Channel 04/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-172-142:789257:789794 [2] NCCL INFO Channel 22/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789255:789793 [0] NCCL INFO Channel 17/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-172-142:789256:789796 [1] NCCL INFO Channel 05/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-172-142:789257:789794 [2] NCCL INFO Channel 23/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789258:789797 [3] NCCL INFO Channel 03/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-172-142:789255:789793 [0] NCCL INFO Channel 18/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-172-142:789256:789796 [1] NCCL INFO Channel 06/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-172-142:789258:789797 [3] NCCL INFO Channel 04/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-172-142:789255:789793 [0] NCCL INFO Channel 19/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-172-142:789256:789796 [1] NCCL INFO Channel 07/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-172-142:789258:789797 [3] NCCL INFO Channel 05/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-172-142:789255:789793 [0] NCCL INFO Channel 20/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-172-142:789256:789796 [1] NCCL INFO Channel 08/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-172-142:789258:789797 [3] NCCL INFO Channel 06/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-172-142:789255:789793 [0] NCCL INFO Channel 21/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-172-142:789256:789796 [1] NCCL INFO Channel 09/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-172-142:789258:789797 [3] NCCL INFO Channel 07/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-172-142:789255:789793 [0] NCCL INFO Channel 22/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-172-142:789256:789796 [1] NCCL INFO Channel 10/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-172-142:789258:789797 [3] NCCL INFO Channel 08/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-172-142:789255:789793 [0] NCCL INFO Channel 23/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-172-142:789256:789796 [1] NCCL INFO Channel 11/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-172-142:789258:789797 [3] NCCL INFO Channel 09/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-172-142:789256:789796 [1] NCCL INFO Channel 12/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-172-142:789258:789797 [3] NCCL INFO Channel 10/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-172-142:789256:789796 [1] NCCL INFO Channel 13/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-172-142:789258:789797 [3] NCCL INFO Channel 11/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-172-142:789256:789796 [1] NCCL INFO Channel 14/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-172-142:789258:789797 [3] NCCL INFO Channel 12/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-172-142:789256:789796 [1] NCCL INFO Channel 15/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-172-142:789258:789797 [3] NCCL INFO Channel 13/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-172-142:789256:789796 [1] NCCL INFO Channel 16/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-172-142:789258:789797 [3] NCCL INFO Channel 14/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-172-142:789256:789796 [1] NCCL INFO Channel 17/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-172-142:789258:789797 [3] NCCL INFO Channel 15/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-172-142:789258:789797 [3] NCCL INFO Channel 16/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-172-142:789258:789797 [3] NCCL INFO Channel 17/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-172-142:789256:789796 [1] NCCL INFO Channel 18/0 : 0[1] -> 1[3] via P2P/IPC
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+ip-26-0-172-142:789258:789797 [3] NCCL INFO Channel 18/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-172-142:789256:789796 [1] NCCL INFO Channel 19/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-172-142:789258:789797 [3] NCCL INFO Channel 19/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-172-142:789256:789796 [1] NCCL INFO Channel 20/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-172-142:789258:789797 [3] NCCL INFO Channel 20/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-172-142:789256:789796 [1] NCCL INFO Channel 21/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-172-142:789258:789797 [3] NCCL INFO Channel 21/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-172-142:789256:789796 [1] NCCL INFO Channel 22/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-172-142:789258:789797 [3] NCCL INFO Channel 22/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-172-142:789256:789796 [1] NCCL INFO Channel 23/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-172-142:789258:789797 [3] NCCL INFO Channel 23/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-172-142:789257:789794 [2] NCCL INFO Connected all rings
+ip-26-0-172-142:789257:789794 [2] NCCL INFO Connected all trees
+ip-26-0-172-142:789257:789794 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-142:789257:789794 [2] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-172-142:789257:789794 [2] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-142:789255:789793 [0] NCCL INFO Connected all rings
+ip-26-0-172-142:789255:789793 [0] NCCL INFO Connected all trees
+ip-26-0-172-142:789255:789793 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-142:789255:789793 [0] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-172-142:789255:789793 [0] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-142:789255:789793 [0] NCCL INFO comm 0x179bdb70 rank 0 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0x60f60f4766fdf198 - Init COMPLETE
+ip-26-0-172-142:789257:789794 [2] NCCL INFO comm 0xb70abb0 rank 1 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0x60f60f4766fdf198 - Init COMPLETE
+ip-26-0-172-142:789255:789806 [0] NCCL INFO Channel 00/1 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-172-142:789255:789806 [0] NCCL INFO Channel 01/1 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-172-142:789255:789806 [0] NCCL INFO Channel 02/1 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-172-142:789255:789806 [0] NCCL INFO Channel 03/1 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-172-142:789255:789806 [0] NCCL INFO Channel 04/1 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-172-142:789255:789806 [0] NCCL INFO Channel 05/1 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-172-142:789255:789806 [0] NCCL INFO Channel 06/1 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-172-142:789255:789806 [0] NCCL INFO Channel 07/1 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-172-142:789255:789806 [0] NCCL INFO Channel 08/1 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-172-142:789255:789806 [0] NCCL INFO Channel 09/1 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-172-142:789258:789797 [3] NCCL INFO Connected all rings
+ip-26-0-172-142:789258:789797 [3] NCCL INFO Connected all trees
+ip-26-0-172-142:789258:789797 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-142:789258:789797 [3] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-172-142:789258:789797 [3] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-142:789255:789806 [0] NCCL INFO Channel 10/1 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-172-142:789255:789806 [0] NCCL INFO Channel 11/1 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-172-142:789256:789796 [1] NCCL INFO Connected all rings
+ip-26-0-172-142:789256:789796 [1] NCCL INFO Connected all trees
+ip-26-0-172-142:789256:789796 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-142:789256:789796 [1] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-172-142:789256:789796 [1] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-142:789255:789806 [0] NCCL INFO Channel 12/1 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-172-142:789255:789806 [0] NCCL INFO Channel 13/1 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-172-142:789255:789806 [0] NCCL INFO Channel 14/1 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-172-142:789255:789806 [0] NCCL INFO Channel 15/1 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-172-142:789255:789806 [0] NCCL INFO Channel 16/1 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-172-142:789255:789806 [0] NCCL INFO Channel 17/1 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-172-142:789255:789806 [0] NCCL INFO Channel 18/1 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-172-142:789255:789806 [0] NCCL INFO Channel 19/1 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-172-142:789255:789806 [0] NCCL INFO Channel 20/1 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-172-142:789255:789806 [0] NCCL INFO Channel 21/1 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-172-142:789255:789806 [0] NCCL INFO Channel 22/1 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-172-142:789255:789806 [0] NCCL INFO Channel 23/1 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-172-142:789255:789806 [0] NCCL INFO Channel 24/1 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-172-142:789255:789806 [0] NCCL INFO Channel 25/1 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-172-142:789255:789806 [0] NCCL INFO Channel 26/1 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-172-142:789255:789806 [0] NCCL INFO Channel 27/1 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-172-142:789255:789806 [0] NCCL INFO Channel 28/1 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-172-142:789255:789806 [0] NCCL INFO Channel 29/1 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-172-142:789255:789806 [0] NCCL INFO Channel 30/1 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-172-142:789255:789806 [0] NCCL INFO Channel 31/1 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-172-142:789258:789797 [3] NCCL INFO comm 0xae145d0 rank 1 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0xd05269a24be85487 - Init COMPLETE
+ip-26-0-172-142:789256:789796 [1] NCCL INFO comm 0x17d11d80 rank 0 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0xd05269a24be85487 - Init COMPLETE
+ip-26-0-172-142:789256:789810 [1] NCCL INFO Channel 00/1 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-172-142:789256:789810 [1] NCCL INFO Channel 01/1 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-172-142:789256:789810 [1] NCCL INFO Channel 02/1 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-172-142:789256:789810 [1] NCCL INFO Channel 03/1 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-172-142:789256:789810 [1] NCCL INFO Channel 04/1 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-172-142:789256:789810 [1] NCCL INFO Channel 05/1 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-172-142:789256:789810 [1] NCCL INFO Channel 06/1 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-172-142:789256:789810 [1] NCCL INFO Channel 07/1 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-172-142:789256:789810 [1] NCCL INFO Channel 08/1 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-172-142:789256:789810 [1] NCCL INFO Channel 09/1 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-172-142:789256:789810 [1] NCCL INFO Channel 10/1 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-172-142:789256:789810 [1] NCCL INFO Channel 11/1 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-172-142:789256:789810 [1] NCCL INFO Channel 12/1 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-172-142:789256:789810 [1] NCCL INFO Channel 13/1 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-172-142:789256:789810 [1] NCCL INFO Channel 14/1 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-172-142:789256:789810 [1] NCCL INFO Channel 15/1 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-172-142:789256:789810 [1] NCCL INFO Channel 16/1 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-172-142:789256:789810 [1] NCCL INFO Channel 17/1 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-172-142:789256:789810 [1] NCCL INFO Channel 18/1 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-172-142:789256:789810 [1] NCCL INFO Channel 19/1 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-172-142:789256:789810 [1] NCCL INFO Channel 20/1 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-172-142:789256:789810 [1] NCCL INFO Channel 21/1 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-172-142:789256:789810 [1] NCCL INFO Channel 22/1 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-172-142:789256:789810 [1] NCCL INFO Channel 23/1 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-172-142:789256:789810 [1] NCCL INFO Channel 24/1 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-172-142:789256:789810 [1] NCCL INFO Channel 25/1 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-172-142:789256:789810 [1] NCCL INFO Channel 26/1 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-172-142:789256:789810 [1] NCCL INFO Channel 27/1 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-172-142:789256:789810 [1] NCCL INFO Channel 28/1 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-172-142:789256:789810 [1] NCCL INFO Channel 29/1 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-172-142:789256:789810 [1] NCCL INFO Channel 30/1 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-172-142:789256:789810 [1] NCCL INFO Channel 31/1 : 0[1] -> 1[3] via P2P/IPC
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+ip-26-0-172-142:789257:789832 [2] NCCL INFO Using network Libfabric
+ip-26-0-172-142:789259:789833 [4] NCCL INFO Using network Libfabric
+ip-26-0-172-142:789257:789832 [2] NCCL INFO comm 0x129479b0 rank 0 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0xff26070189d2dd18 - Init START
+ip-26-0-172-142:789259:789833 [4] NCCL INFO comm 0xb417b60 rank 1 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0xff26070189d2dd18 - Init START
+ip-26-0-172-142:789259:789833 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-172-142:789257:789832 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+ip-26-0-172-142:789258:789835 [3] NCCL INFO Using network Libfabric
+ip-26-0-172-142:789260:789836 [5] NCCL INFO Using network Libfabric
+ip-26-0-172-142:789260:789836 [5] NCCL INFO comm 0xc43d440 rank 1 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0x248d83db8e600578 - Init START
+ip-26-0-172-142:789258:789835 [3] NCCL INFO comm 0x120cda30 rank 0 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0x248d83db8e600578 - Init START
+ip-26-0-172-142:789260:789836 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-172-142:789258:789835 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-172-142:789259:789833 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-172-142:789257:789832 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-172-142:789259:789833 [4] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] -1/-1/-1->1->0 [3] -1/-1/-1->1->0 [4] -1/-1/-1->1->0 [5] -1/-1/-1->1->0 [6] 0/-1/-1->1->-1 [7] 0/-1/-1->1->-1 [8] 0/-1/-1->1->-1 [9] 0/-1/-1->1->-1 [10] 0/-1/-1->1->-1 [11] 0/-1/-1->1->-1 [12] -1/-1/-1->1->0 [13] -1/-1/-1->1->0 [14] -1/-1/-1->1->0 [15] -1/-1/-1->1->0 [16] -1/-1/-1->1->0 [17] -1/-1/-1->1->0 [18] 0/-1/-1->1->-1 [19] 0/-1/-1->1->-1 [20] 0/-1/-1->1->-1 [21] 0/-1/-1->1->-1 [22] 0/-1/-1->1->-1 [23] 0/-1/-1->1->-1
+ip-26-0-172-142:789257:789832 [2] NCCL INFO Channel 00/24 :    0   1
+ip-26-0-172-142:789259:789833 [4] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-172-142:789257:789832 [2] NCCL INFO Channel 01/24 :    0   1
+ip-26-0-172-142:789257:789832 [2] NCCL INFO Channel 02/24 :    0   1
+ip-26-0-172-142:789257:789832 [2] NCCL INFO Channel 03/24 :    0   1
+ip-26-0-172-142:789257:789832 [2] NCCL INFO Channel 04/24 :    0   1
+ip-26-0-172-142:789257:789832 [2] NCCL INFO Channel 05/24 :    0   1
+ip-26-0-172-142:789257:789832 [2] NCCL INFO Channel 06/24 :    0   1
+ip-26-0-172-142:789257:789832 [2] NCCL INFO Channel 07/24 :    0   1
+ip-26-0-172-142:789257:789832 [2] NCCL INFO Channel 08/24 :    0   1
+ip-26-0-172-142:789257:789832 [2] NCCL INFO Channel 09/24 :    0   1
+ip-26-0-172-142:789257:789832 [2] NCCL INFO Channel 10/24 :    0   1
+ip-26-0-172-142:789257:789832 [2] NCCL INFO Channel 11/24 :    0   1
+ip-26-0-172-142:789257:789832 [2] NCCL INFO Channel 12/24 :    0   1
+ip-26-0-172-142:789257:789832 [2] NCCL INFO Channel 13/24 :    0   1
+ip-26-0-172-142:789257:789832 [2] NCCL INFO Channel 14/24 :    0   1
+ip-26-0-172-142:789257:789832 [2] NCCL INFO Channel 15/24 :    0   1
+ip-26-0-172-142:789257:789832 [2] NCCL INFO Channel 16/24 :    0   1
+ip-26-0-172-142:789257:789832 [2] NCCL INFO Channel 17/24 :    0   1
+ip-26-0-172-142:789257:789832 [2] NCCL INFO Channel 18/24 :    0   1
+ip-26-0-172-142:789257:789832 [2] NCCL INFO Channel 19/24 :    0   1
+ip-26-0-172-142:789257:789832 [2] NCCL INFO Channel 20/24 :    0   1
+ip-26-0-172-142:789257:789832 [2] NCCL INFO Channel 21/24 :    0   1
+ip-26-0-172-142:789257:789832 [2] NCCL INFO Channel 22/24 :    0   1
+ip-26-0-172-142:789257:789832 [2] NCCL INFO Channel 23/24 :    0   1
+ip-26-0-172-142:789257:789832 [2] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] -1/-1/-1->0->1 [7] -1/-1/-1->0->1 [8] -1/-1/-1->0->1 [9] -1/-1/-1->0->1 [10] -1/-1/-1->0->1 [11] -1/-1/-1->0->1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] -1/-1/-1->0->1 [19] -1/-1/-1->0->1 [20] -1/-1/-1->0->1 [21] -1/-1/-1->0->1 [22] -1/-1/-1->0->1 [23] -1/-1/-1->0->1
+ip-26-0-172-142:789257:789832 [2] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-172-142:789257:789832 [2] NCCL INFO Channel 00/0 : 0[2] -> 1[4] via P2P/IPC
+ip-26-0-172-142:789257:789832 [2] NCCL INFO Channel 01/0 : 0[2] -> 1[4] via P2P/IPC
+ip-26-0-172-142:789257:789832 [2] NCCL INFO Channel 02/0 : 0[2] -> 1[4] via P2P/IPC
+ip-26-0-172-142:789259:789833 [4] NCCL INFO Channel 00/0 : 1[4] -> 0[2] via P2P/IPC
+ip-26-0-172-142:789259:789833 [4] NCCL INFO Channel 01/0 : 1[4] -> 0[2] via P2P/IPC
+ip-26-0-172-142:789257:789832 [2] NCCL INFO Channel 03/0 : 0[2] -> 1[4] via P2P/IPC
+ip-26-0-172-142:789259:789833 [4] NCCL INFO Channel 02/0 : 1[4] -> 0[2] via P2P/IPC
+ip-26-0-172-142:789257:789832 [2] NCCL INFO Channel 04/0 : 0[2] -> 1[4] via P2P/IPC
+ip-26-0-172-142:789259:789833 [4] NCCL INFO Channel 03/0 : 1[4] -> 0[2] via P2P/IPC
+ip-26-0-172-142:789257:789832 [2] NCCL INFO Channel 05/0 : 0[2] -> 1[4] via P2P/IPC
+ip-26-0-172-142:789259:789833 [4] NCCL INFO Channel 04/0 : 1[4] -> 0[2] via P2P/IPC
+ip-26-0-172-142:789257:789832 [2] NCCL INFO Channel 06/0 : 0[2] -> 1[4] via P2P/IPC
+ip-26-0-172-142:789259:789833 [4] NCCL INFO Channel 05/0 : 1[4] -> 0[2] via P2P/IPC
+ip-26-0-172-142:789257:789832 [2] NCCL INFO Channel 07/0 : 0[2] -> 1[4] via P2P/IPC
+ip-26-0-172-142:789259:789833 [4] NCCL INFO Channel 06/0 : 1[4] -> 0[2] via P2P/IPC
+ip-26-0-172-142:789257:789832 [2] NCCL INFO Channel 08/0 : 0[2] -> 1[4] via P2P/IPC
+ip-26-0-172-142:789259:789833 [4] NCCL INFO Channel 07/0 : 1[4] -> 0[2] via P2P/IPC
+ip-26-0-172-142:789260:789836 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-172-142:789257:789832 [2] NCCL INFO Channel 09/0 : 0[2] -> 1[4] via P2P/IPC
+ip-26-0-172-142:789259:789833 [4] NCCL INFO Channel 08/0 : 1[4] -> 0[2] via P2P/IPC
+ip-26-0-172-142:789257:789832 [2] NCCL INFO Channel 10/0 : 0[2] -> 1[4] via P2P/IPC
+ip-26-0-172-142:789259:789833 [4] NCCL INFO Channel 09/0 : 1[4] -> 0[2] via P2P/IPC
+ip-26-0-172-142:789257:789832 [2] NCCL INFO Channel 11/0 : 0[2] -> 1[4] via P2P/IPC
+ip-26-0-172-142:789259:789833 [4] NCCL INFO Channel 10/0 : 1[4] -> 0[2] via P2P/IPC
+ip-26-0-172-142:789258:789835 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-172-142:789257:789832 [2] NCCL INFO Channel 12/0 : 0[2] -> 1[4] via P2P/IPC
+ip-26-0-172-142:789259:789833 [4] NCCL INFO Channel 11/0 : 1[4] -> 0[2] via P2P/IPC
+ip-26-0-172-142:789260:789836 [5] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] -1/-1/-1->1->0 [3] -1/-1/-1->1->0 [4] -1/-1/-1->1->0 [5] -1/-1/-1->1->0 [6] 0/-1/-1->1->-1 [7] 0/-1/-1->1->-1 [8] 0/-1/-1->1->-1 [9] 0/-1/-1->1->-1 [10] 0/-1/-1->1->-1 [11] 0/-1/-1->1->-1 [12] -1/-1/-1->1->0 [13] -1/-1/-1->1->0 [14] -1/-1/-1->1->0 [15] -1/-1/-1->1->0 [16] -1/-1/-1->1->0 [17] -1/-1/-1->1->0 [18] 0/-1/-1->1->-1 [19] 0/-1/-1->1->-1 [20] 0/-1/-1->1->-1 [21] 0/-1/-1->1->-1 [22] 0/-1/-1->1->-1 [23] 0/-1/-1->1->-1
+ip-26-0-172-142:789258:789835 [3] NCCL INFO Channel 00/24 :    0   1
+ip-26-0-172-142:789260:789836 [5] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-172-142:789258:789835 [3] NCCL INFO Channel 01/24 :    0   1
+ip-26-0-172-142:789258:789835 [3] NCCL INFO Channel 02/24 :    0   1
+ip-26-0-172-142:789258:789835 [3] NCCL INFO Channel 03/24 :    0   1
+ip-26-0-172-142:789258:789835 [3] NCCL INFO Channel 04/24 :    0   1
+ip-26-0-172-142:789258:789835 [3] NCCL INFO Channel 05/24 :    0   1
+ip-26-0-172-142:789258:789835 [3] NCCL INFO Channel 06/24 :    0   1
+ip-26-0-172-142:789258:789835 [3] NCCL INFO Channel 07/24 :    0   1
+ip-26-0-172-142:789258:789835 [3] NCCL INFO Channel 08/24 :    0   1
+ip-26-0-172-142:789258:789835 [3] NCCL INFO Channel 09/24 :    0   1
+ip-26-0-172-142:789258:789835 [3] NCCL INFO Channel 10/24 :    0   1
+ip-26-0-172-142:789258:789835 [3] NCCL INFO Channel 11/24 :    0   1
+ip-26-0-172-142:789258:789835 [3] NCCL INFO Channel 12/24 :    0   1
+ip-26-0-172-142:789258:789835 [3] NCCL INFO Channel 13/24 :    0   1
+ip-26-0-172-142:789258:789835 [3] NCCL INFO Channel 14/24 :    0   1
+ip-26-0-172-142:789258:789835 [3] NCCL INFO Channel 15/24 :    0   1
+ip-26-0-172-142:789258:789835 [3] NCCL INFO Channel 16/24 :    0   1
+ip-26-0-172-142:789258:789835 [3] NCCL INFO Channel 17/24 :    0   1
+ip-26-0-172-142:789258:789835 [3] NCCL INFO Channel 18/24 :    0   1
+ip-26-0-172-142:789258:789835 [3] NCCL INFO Channel 19/24 :    0   1
+ip-26-0-172-142:789258:789835 [3] NCCL INFO Channel 20/24 :    0   1
+ip-26-0-172-142:789258:789835 [3] NCCL INFO Channel 21/24 :    0   1
+ip-26-0-172-142:789258:789835 [3] NCCL INFO Channel 22/24 :    0   1
+ip-26-0-172-142:789258:789835 [3] NCCL INFO Channel 23/24 :    0   1
+ip-26-0-172-142:789258:789835 [3] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] -1/-1/-1->0->1 [7] -1/-1/-1->0->1 [8] -1/-1/-1->0->1 [9] -1/-1/-1->0->1 [10] -1/-1/-1->0->1 [11] -1/-1/-1->0->1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] -1/-1/-1->0->1 [19] -1/-1/-1->0->1 [20] -1/-1/-1->0->1 [21] -1/-1/-1->0->1 [22] -1/-1/-1->0->1 [23] -1/-1/-1->0->1
+ip-26-0-172-142:789258:789835 [3] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-172-142:789257:789832 [2] NCCL INFO Channel 13/0 : 0[2] -> 1[4] via P2P/IPC
+ip-26-0-172-142:789259:789833 [4] NCCL INFO Channel 12/0 : 1[4] -> 0[2] via P2P/IPC
+ip-26-0-172-142:789257:789832 [2] NCCL INFO Channel 14/0 : 0[2] -> 1[4] via P2P/IPC
+ip-26-0-172-142:789259:789833 [4] NCCL INFO Channel 13/0 : 1[4] -> 0[2] via P2P/IPC
+ip-26-0-172-142:789257:789832 [2] NCCL INFO Channel 15/0 : 0[2] -> 1[4] via P2P/IPC
+ip-26-0-172-142:789259:789833 [4] NCCL INFO Channel 14/0 : 1[4] -> 0[2] via P2P/IPC
+ip-26-0-172-142:789257:789832 [2] NCCL INFO Channel 16/0 : 0[2] -> 1[4] via P2P/IPC
+ip-26-0-172-142:789259:789833 [4] NCCL INFO Channel 15/0 : 1[4] -> 0[2] via P2P/IPC
+ip-26-0-172-142:789257:789832 [2] NCCL INFO Channel 17/0 : 0[2] -> 1[4] via P2P/IPC
+ip-26-0-172-142:789259:789833 [4] NCCL INFO Channel 16/0 : 1[4] -> 0[2] via P2P/IPC
+ip-26-0-172-142:789259:789833 [4] NCCL INFO Channel 17/0 : 1[4] -> 0[2] via P2P/IPC
+ip-26-0-172-142:789257:789832 [2] NCCL INFO Channel 18/0 : 0[2] -> 1[4] via P2P/IPC
+ip-26-0-172-142:789257:789832 [2] NCCL INFO Channel 19/0 : 0[2] -> 1[4] via P2P/IPC
+ip-26-0-172-142:789259:789833 [4] NCCL INFO Channel 18/0 : 1[4] -> 0[2] via P2P/IPC
+ip-26-0-172-142:789257:789832 [2] NCCL INFO Channel 20/0 : 0[2] -> 1[4] via P2P/IPC
+ip-26-0-172-142:789259:789833 [4] NCCL INFO Channel 19/0 : 1[4] -> 0[2] via P2P/IPC
+ip-26-0-172-142:789257:789832 [2] NCCL INFO Channel 21/0 : 0[2] -> 1[4] via P2P/IPC
+ip-26-0-172-142:789259:789833 [4] NCCL INFO Channel 20/0 : 1[4] -> 0[2] via P2P/IPC
+ip-26-0-172-142:789257:789832 [2] NCCL INFO Channel 22/0 : 0[2] -> 1[4] via P2P/IPC
+ip-26-0-172-142:789259:789833 [4] NCCL INFO Channel 21/0 : 1[4] -> 0[2] via P2P/IPC
+ip-26-0-172-142:789257:789832 [2] NCCL INFO Channel 23/0 : 0[2] -> 1[4] via P2P/IPC
+ip-26-0-172-142:789259:789833 [4] NCCL INFO Channel 22/0 : 1[4] -> 0[2] via P2P/IPC
+ip-26-0-172-142:789259:789833 [4] NCCL INFO Channel 23/0 : 1[4] -> 0[2] via P2P/IPC
+ip-26-0-172-142:789260:789836 [5] NCCL INFO Channel 00/0 : 1[5] -> 0[3] via P2P/IPC
+ip-26-0-172-142:789260:789836 [5] NCCL INFO Channel 01/0 : 1[5] -> 0[3] via P2P/IPC
+ip-26-0-172-142:789258:789835 [3] NCCL INFO Channel 00/0 : 0[3] -> 1[5] via P2P/IPC
+ip-26-0-172-142:789260:789836 [5] NCCL INFO Channel 02/0 : 1[5] -> 0[3] via P2P/IPC
+ip-26-0-172-142:789258:789835 [3] NCCL INFO Channel 01/0 : 0[3] -> 1[5] via P2P/IPC
+ip-26-0-172-142:789258:789835 [3] NCCL INFO Channel 02/0 : 0[3] -> 1[5] via P2P/IPC
+ip-26-0-172-142:789260:789836 [5] NCCL INFO Channel 03/0 : 1[5] -> 0[3] via P2P/IPC
+ip-26-0-172-142:789260:789836 [5] NCCL INFO Channel 04/0 : 1[5] -> 0[3] via P2P/IPC
+ip-26-0-172-142:789258:789835 [3] NCCL INFO Channel 03/0 : 0[3] -> 1[5] via P2P/IPC
+ip-26-0-172-142:789260:789836 [5] NCCL INFO Channel 05/0 : 1[5] -> 0[3] via P2P/IPC
+ip-26-0-172-142:789258:789835 [3] NCCL INFO Channel 04/0 : 0[3] -> 1[5] via P2P/IPC
+ip-26-0-172-142:789260:789836 [5] NCCL INFO Channel 06/0 : 1[5] -> 0[3] via P2P/IPC
+ip-26-0-172-142:789258:789835 [3] NCCL INFO Channel 05/0 : 0[3] -> 1[5] via P2P/IPC
+ip-26-0-172-142:789260:789836 [5] NCCL INFO Channel 07/0 : 1[5] -> 0[3] via P2P/IPC
+ip-26-0-172-142:789258:789835 [3] NCCL INFO Channel 06/0 : 0[3] -> 1[5] via P2P/IPC
+ip-26-0-172-142:789260:789836 [5] NCCL INFO Channel 08/0 : 1[5] -> 0[3] via P2P/IPC
+ip-26-0-172-142:789260:789836 [5] NCCL INFO Channel 09/0 : 1[5] -> 0[3] via P2P/IPC
+ip-26-0-172-142:789258:789835 [3] NCCL INFO Channel 07/0 : 0[3] -> 1[5] via P2P/IPC
+ip-26-0-172-142:789260:789836 [5] NCCL INFO Channel 10/0 : 1[5] -> 0[3] via P2P/IPC
+ip-26-0-172-142:789258:789835 [3] NCCL INFO Channel 08/0 : 0[3] -> 1[5] via P2P/IPC
+ip-26-0-172-142:789259:789833 [4] NCCL INFO Connected all rings
+ip-26-0-172-142:789259:789833 [4] NCCL INFO Connected all trees
+ip-26-0-172-142:789259:789833 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-142:789259:789833 [4] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-172-142:789259:789833 [4] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-142:789260:789836 [5] NCCL INFO Channel 11/0 : 1[5] -> 0[3] via P2P/IPC
+ip-26-0-172-142:789258:789835 [3] NCCL INFO Channel 09/0 : 0[3] -> 1[5] via P2P/IPC
+ip-26-0-172-142:789260:789836 [5] NCCL INFO Channel 12/0 : 1[5] -> 0[3] via P2P/IPC
+ip-26-0-172-142:789258:789835 [3] NCCL INFO Channel 10/0 : 0[3] -> 1[5] via P2P/IPC
+ip-26-0-172-142:789260:789836 [5] NCCL INFO Channel 13/0 : 1[5] -> 0[3] via P2P/IPC
+ip-26-0-172-142:789258:789835 [3] NCCL INFO Channel 11/0 : 0[3] -> 1[5] via P2P/IPC
+ip-26-0-172-142:789260:789836 [5] NCCL INFO Channel 14/0 : 1[5] -> 0[3] via P2P/IPC
+ip-26-0-172-142:789258:789835 [3] NCCL INFO Channel 12/0 : 0[3] -> 1[5] via P2P/IPC
+ip-26-0-172-142:789260:789836 [5] NCCL INFO Channel 15/0 : 1[5] -> 0[3] via P2P/IPC
+ip-26-0-172-142:789257:789832 [2] NCCL INFO Connected all rings
+ip-26-0-172-142:789257:789832 [2] NCCL INFO Connected all trees
+ip-26-0-172-142:789257:789832 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-142:789257:789832 [2] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-172-142:789257:789832 [2] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-142:789258:789835 [3] NCCL INFO Channel 13/0 : 0[3] -> 1[5] via P2P/IPC
+ip-26-0-172-142:789260:789836 [5] NCCL INFO Channel 16/0 : 1[5] -> 0[3] via P2P/IPC
+ip-26-0-172-142:789258:789835 [3] NCCL INFO Channel 14/0 : 0[3] -> 1[5] via P2P/IPC
+ip-26-0-172-142:789260:789836 [5] NCCL INFO Channel 17/0 : 1[5] -> 0[3] via P2P/IPC
+ip-26-0-172-142:789258:789835 [3] NCCL INFO Channel 15/0 : 0[3] -> 1[5] via P2P/IPC
+ip-26-0-172-142:789260:789836 [5] NCCL INFO Channel 18/0 : 1[5] -> 0[3] via P2P/IPC
+ip-26-0-172-142:789258:789835 [3] NCCL INFO Channel 16/0 : 0[3] -> 1[5] via P2P/IPC
+ip-26-0-172-142:789260:789836 [5] NCCL INFO Channel 19/0 : 1[5] -> 0[3] via P2P/IPC
+ip-26-0-172-142:789258:789835 [3] NCCL INFO Channel 17/0 : 0[3] -> 1[5] via P2P/IPC
+ip-26-0-172-142:789260:789836 [5] NCCL INFO Channel 20/0 : 1[5] -> 0[3] via P2P/IPC
+ip-26-0-172-142:789258:789835 [3] NCCL INFO Channel 18/0 : 0[3] -> 1[5] via P2P/IPC
+ip-26-0-172-142:789260:789836 [5] NCCL INFO Channel 21/0 : 1[5] -> 0[3] via P2P/IPC
+ip-26-0-172-142:789258:789835 [3] NCCL INFO Channel 19/0 : 0[3] -> 1[5] via P2P/IPC
+ip-26-0-172-142:789260:789836 [5] NCCL INFO Channel 22/0 : 1[5] -> 0[3] via P2P/IPC
+ip-26-0-172-142:789258:789835 [3] NCCL INFO Channel 20/0 : 0[3] -> 1[5] via P2P/IPC
+ip-26-0-172-142:789260:789836 [5] NCCL INFO Channel 23/0 : 1[5] -> 0[3] via P2P/IPC
+ip-26-0-172-142:789258:789835 [3] NCCL INFO Channel 21/0 : 0[3] -> 1[5] via P2P/IPC
+ip-26-0-172-142:789258:789835 [3] NCCL INFO Channel 22/0 : 0[3] -> 1[5] via P2P/IPC
+ip-26-0-172-142:789258:789835 [3] NCCL INFO Channel 23/0 : 0[3] -> 1[5] via P2P/IPC
+ip-26-0-172-142:789259:789833 [4] NCCL INFO comm 0xb417b60 rank 1 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0xff26070189d2dd18 - Init COMPLETE
+ip-26-0-172-142:789257:789832 [2] NCCL INFO comm 0x129479b0 rank 0 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0xff26070189d2dd18 - Init COMPLETE
+ip-26-0-172-142:789257:789844 [2] NCCL INFO Channel 00/1 : 0[2] -> 1[4] via P2P/IPC
+ip-26-0-172-142:789257:789844 [2] NCCL INFO Channel 01/1 : 0[2] -> 1[4] via P2P/IPC
+ip-26-0-172-142:789257:789844 [2] NCCL INFO Channel 02/1 : 0[2] -> 1[4] via P2P/IPC
+ip-26-0-172-142:789257:789844 [2] NCCL INFO Channel 03/1 : 0[2] -> 1[4] via P2P/IPC
+ip-26-0-172-142:789257:789844 [2] NCCL INFO Channel 04/1 : 0[2] -> 1[4] via P2P/IPC
+ip-26-0-172-142:789257:789844 [2] NCCL INFO Channel 05/1 : 0[2] -> 1[4] via P2P/IPC
+ip-26-0-172-142:789257:789844 [2] NCCL INFO Channel 06/1 : 0[2] -> 1[4] via P2P/IPC
+ip-26-0-172-142:789257:789844 [2] NCCL INFO Channel 07/1 : 0[2] -> 1[4] via P2P/IPC
+ip-26-0-172-142:789257:789844 [2] NCCL INFO Channel 08/1 : 0[2] -> 1[4] via P2P/IPC
+ip-26-0-172-142:789257:789844 [2] NCCL INFO Channel 09/1 : 0[2] -> 1[4] via P2P/IPC
+ip-26-0-172-142:789257:789844 [2] NCCL INFO Channel 10/1 : 0[2] -> 1[4] via P2P/IPC
+ip-26-0-172-142:789257:789844 [2] NCCL INFO Channel 11/1 : 0[2] -> 1[4] via P2P/IPC
+ip-26-0-172-142:789257:789844 [2] NCCL INFO Channel 12/1 : 0[2] -> 1[4] via P2P/IPC
+ip-26-0-172-142:789257:789844 [2] NCCL INFO Channel 13/1 : 0[2] -> 1[4] via P2P/IPC
+ip-26-0-172-142:789257:789844 [2] NCCL INFO Channel 14/1 : 0[2] -> 1[4] via P2P/IPC
+ip-26-0-172-142:789257:789844 [2] NCCL INFO Channel 15/1 : 0[2] -> 1[4] via P2P/IPC
+ip-26-0-172-142:789257:789844 [2] NCCL INFO Channel 16/1 : 0[2] -> 1[4] via P2P/IPC
+ip-26-0-172-142:789257:789844 [2] NCCL INFO Channel 17/1 : 0[2] -> 1[4] via P2P/IPC
+ip-26-0-172-142:789257:789844 [2] NCCL INFO Channel 18/1 : 0[2] -> 1[4] via P2P/IPC
+ip-26-0-172-142:789257:789844 [2] NCCL INFO Channel 19/1 : 0[2] -> 1[4] via P2P/IPC
+ip-26-0-172-142:789257:789844 [2] NCCL INFO Channel 20/1 : 0[2] -> 1[4] via P2P/IPC
+ip-26-0-172-142:789257:789844 [2] NCCL INFO Channel 21/1 : 0[2] -> 1[4] via P2P/IPC
+ip-26-0-172-142:789257:789844 [2] NCCL INFO Channel 22/1 : 0[2] -> 1[4] via P2P/IPC
+ip-26-0-172-142:789258:789835 [3] NCCL INFO Connected all rings
+ip-26-0-172-142:789258:789835 [3] NCCL INFO Connected all trees
+ip-26-0-172-142:789258:789835 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-142:789258:789835 [3] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-172-142:789258:789835 [3] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-142:789257:789844 [2] NCCL INFO Channel 23/1 : 0[2] -> 1[4] via P2P/IPC
+ip-26-0-172-142:789257:789844 [2] NCCL INFO Channel 24/1 : 0[2] -> 1[4] via P2P/IPC
+ip-26-0-172-142:789260:789836 [5] NCCL INFO Connected all rings
+ip-26-0-172-142:789260:789836 [5] NCCL INFO Connected all trees
+ip-26-0-172-142:789260:789836 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-142:789260:789836 [5] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-172-142:789260:789836 [5] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-142:789257:789844 [2] NCCL INFO Channel 25/1 : 0[2] -> 1[4] via P2P/IPC
+ip-26-0-172-142:789257:789844 [2] NCCL INFO Channel 26/1 : 0[2] -> 1[4] via P2P/IPC
+ip-26-0-172-142:789257:789844 [2] NCCL INFO Channel 27/1 : 0[2] -> 1[4] via P2P/IPC
+ip-26-0-172-142:789257:789844 [2] NCCL INFO Channel 28/1 : 0[2] -> 1[4] via P2P/IPC
+ip-26-0-172-142:789257:789844 [2] NCCL INFO Channel 29/1 : 0[2] -> 1[4] via P2P/IPC
+ip-26-0-172-142:789257:789844 [2] NCCL INFO Channel 30/1 : 0[2] -> 1[4] via P2P/IPC
+ip-26-0-172-142:789257:789844 [2] NCCL INFO Channel 31/1 : 0[2] -> 1[4] via P2P/IPC
+ip-26-0-172-142:789260:789836 [5] NCCL INFO comm 0xc43d440 rank 1 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0x248d83db8e600578 - Init COMPLETE
+ip-26-0-172-142:789258:789835 [3] NCCL INFO comm 0x120cda30 rank 0 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0x248d83db8e600578 - Init COMPLETE
+ip-26-0-172-142:789258:789849 [3] NCCL INFO Channel 00/1 : 0[3] -> 1[5] via P2P/IPC
+ip-26-0-172-142:789258:789849 [3] NCCL INFO Channel 01/1 : 0[3] -> 1[5] via P2P/IPC
+ip-26-0-172-142:789258:789849 [3] NCCL INFO Channel 02/1 : 0[3] -> 1[5] via P2P/IPC
+ip-26-0-172-142:789258:789849 [3] NCCL INFO Channel 03/1 : 0[3] -> 1[5] via P2P/IPC
+ip-26-0-172-142:789258:789849 [3] NCCL INFO Channel 04/1 : 0[3] -> 1[5] via P2P/IPC
+ip-26-0-172-142:789258:789849 [3] NCCL INFO Channel 05/1 : 0[3] -> 1[5] via P2P/IPC
+ip-26-0-172-142:789258:789849 [3] NCCL INFO Channel 06/1 : 0[3] -> 1[5] via P2P/IPC
+ip-26-0-172-142:789258:789849 [3] NCCL INFO Channel 07/1 : 0[3] -> 1[5] via P2P/IPC
+ip-26-0-172-142:789258:789849 [3] NCCL INFO Channel 08/1 : 0[3] -> 1[5] via P2P/IPC
+ip-26-0-172-142:789258:789849 [3] NCCL INFO Channel 09/1 : 0[3] -> 1[5] via P2P/IPC
+ip-26-0-172-142:789258:789849 [3] NCCL INFO Channel 10/1 : 0[3] -> 1[5] via P2P/IPC
+ip-26-0-172-142:789258:789849 [3] NCCL INFO Channel 11/1 : 0[3] -> 1[5] via P2P/IPC
+ip-26-0-172-142:789258:789849 [3] NCCL INFO Channel 12/1 : 0[3] -> 1[5] via P2P/IPC
+ip-26-0-172-142:789258:789849 [3] NCCL INFO Channel 13/1 : 0[3] -> 1[5] via P2P/IPC
+ip-26-0-172-142:789258:789849 [3] NCCL INFO Channel 14/1 : 0[3] -> 1[5] via P2P/IPC
+ip-26-0-172-142:789258:789849 [3] NCCL INFO Channel 15/1 : 0[3] -> 1[5] via P2P/IPC
+ip-26-0-172-142:789258:789849 [3] NCCL INFO Channel 16/1 : 0[3] -> 1[5] via P2P/IPC
+ip-26-0-172-142:789258:789849 [3] NCCL INFO Channel 17/1 : 0[3] -> 1[5] via P2P/IPC
+ip-26-0-172-142:789258:789849 [3] NCCL INFO Channel 18/1 : 0[3] -> 1[5] via P2P/IPC
+ip-26-0-172-142:789258:789849 [3] NCCL INFO Channel 19/1 : 0[3] -> 1[5] via P2P/IPC
+ip-26-0-172-142:789258:789849 [3] NCCL INFO Channel 20/1 : 0[3] -> 1[5] via P2P/IPC
+ip-26-0-172-142:789258:789849 [3] NCCL INFO Channel 21/1 : 0[3] -> 1[5] via P2P/IPC
+ip-26-0-172-142:789258:789849 [3] NCCL INFO Channel 22/1 : 0[3] -> 1[5] via P2P/IPC
+ip-26-0-172-142:789258:789849 [3] NCCL INFO Channel 23/1 : 0[3] -> 1[5] via P2P/IPC
+ip-26-0-172-142:789258:789849 [3] NCCL INFO Channel 24/1 : 0[3] -> 1[5] via P2P/IPC
+ip-26-0-172-142:789258:789849 [3] NCCL INFO Channel 25/1 : 0[3] -> 1[5] via P2P/IPC
+ip-26-0-172-142:789258:789849 [3] NCCL INFO Channel 26/1 : 0[3] -> 1[5] via P2P/IPC
+ip-26-0-172-142:789258:789849 [3] NCCL INFO Channel 27/1 : 0[3] -> 1[5] via P2P/IPC
+ip-26-0-172-142:789258:789849 [3] NCCL INFO Channel 28/1 : 0[3] -> 1[5] via P2P/IPC
+ip-26-0-172-142:789258:789849 [3] NCCL INFO Channel 29/1 : 0[3] -> 1[5] via P2P/IPC
+ip-26-0-172-142:789258:789849 [3] NCCL INFO Channel 30/1 : 0[3] -> 1[5] via P2P/IPC
+ip-26-0-172-142:789258:789849 [3] NCCL INFO Channel 31/1 : 0[3] -> 1[5] via P2P/IPC
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+ip-26-0-172-142:789259:789871 [4] NCCL INFO Using network Libfabric
+ip-26-0-172-142:789261:789872 [6] NCCL INFO Using network Libfabric
+ip-26-0-172-142:789261:789872 [6] NCCL INFO comm 0xc150440 rank 1 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0x60ad0b2b62f9ba3b - Init START
+ip-26-0-172-142:789259:789871 [4] NCCL INFO comm 0x125eb5b0 rank 0 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0x60ad0b2b62f9ba3b - Init START
+ip-26-0-172-142:789261:789872 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-172-142:789259:789871 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+ip-26-0-172-142:789260:789875 [5] NCCL INFO Using network Libfabric
+ip-26-0-172-142:789262:789876 [7] NCCL INFO Using network Libfabric
+ip-26-0-172-142:789262:789876 [7] NCCL INFO comm 0xadd78e0 rank 1 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0xd8ecbab89c6d4ed3 - Init START
+ip-26-0-172-142:789260:789875 [5] NCCL INFO comm 0x13610200 rank 0 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0xd8ecbab89c6d4ed3 - Init START
+ip-26-0-172-142:789262:789876 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-172-142:789260:789875 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-172-142:789259:789871 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-172-142:789261:789872 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-172-142:789259:789871 [4] NCCL INFO Channel 00/24 :    0   1
+ip-26-0-172-142:789259:789871 [4] NCCL INFO Channel 01/24 :    0   1
+ip-26-0-172-142:789259:789871 [4] NCCL INFO Channel 02/24 :    0   1
+ip-26-0-172-142:789259:789871 [4] NCCL INFO Channel 03/24 :    0   1
+ip-26-0-172-142:789259:789871 [4] NCCL INFO Channel 04/24 :    0   1
+ip-26-0-172-142:789259:789871 [4] NCCL INFO Channel 05/24 :    0   1
+ip-26-0-172-142:789259:789871 [4] NCCL INFO Channel 06/24 :    0   1
+ip-26-0-172-142:789259:789871 [4] NCCL INFO Channel 07/24 :    0   1
+ip-26-0-172-142:789259:789871 [4] NCCL INFO Channel 08/24 :    0   1
+ip-26-0-172-142:789259:789871 [4] NCCL INFO Channel 09/24 :    0   1
+ip-26-0-172-142:789259:789871 [4] NCCL INFO Channel 10/24 :    0   1
+ip-26-0-172-142:789259:789871 [4] NCCL INFO Channel 11/24 :    0   1
+ip-26-0-172-142:789261:789872 [6] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] -1/-1/-1->1->0 [3] -1/-1/-1->1->0 [4] -1/-1/-1->1->0 [5] -1/-1/-1->1->0 [6] 0/-1/-1->1->-1 [7] 0/-1/-1->1->-1 [8] 0/-1/-1->1->-1 [9] 0/-1/-1->1->-1 [10] 0/-1/-1->1->-1 [11] 0/-1/-1->1->-1 [12] -1/-1/-1->1->0 [13] -1/-1/-1->1->0 [14] -1/-1/-1->1->0 [15] -1/-1/-1->1->0 [16] -1/-1/-1->1->0 [17] -1/-1/-1->1->0 [18] 0/-1/-1->1->-1 [19] 0/-1/-1->1->-1 [20] 0/-1/-1->1->-1 [21] 0/-1/-1->1->-1 [22] 0/-1/-1->1->-1 [23] 0/-1/-1->1->-1
+ip-26-0-172-142:789259:789871 [4] NCCL INFO Channel 12/24 :    0   1
+ip-26-0-172-142:789261:789872 [6] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-172-142:789259:789871 [4] NCCL INFO Channel 13/24 :    0   1
+ip-26-0-172-142:789259:789871 [4] NCCL INFO Channel 14/24 :    0   1
+ip-26-0-172-142:789259:789871 [4] NCCL INFO Channel 15/24 :    0   1
+ip-26-0-172-142:789259:789871 [4] NCCL INFO Channel 16/24 :    0   1
+ip-26-0-172-142:789259:789871 [4] NCCL INFO Channel 17/24 :    0   1
+ip-26-0-172-142:789259:789871 [4] NCCL INFO Channel 18/24 :    0   1
+ip-26-0-172-142:789259:789871 [4] NCCL INFO Channel 19/24 :    0   1
+ip-26-0-172-142:789259:789871 [4] NCCL INFO Channel 20/24 :    0   1
+ip-26-0-172-142:789259:789871 [4] NCCL INFO Channel 21/24 :    0   1
+ip-26-0-172-142:789259:789871 [4] NCCL INFO Channel 22/24 :    0   1
+ip-26-0-172-142:789259:789871 [4] NCCL INFO Channel 23/24 :    0   1
+ip-26-0-172-142:789259:789871 [4] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] -1/-1/-1->0->1 [7] -1/-1/-1->0->1 [8] -1/-1/-1->0->1 [9] -1/-1/-1->0->1 [10] -1/-1/-1->0->1 [11] -1/-1/-1->0->1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] -1/-1/-1->0->1 [19] -1/-1/-1->0->1 [20] -1/-1/-1->0->1 [21] -1/-1/-1->0->1 [22] -1/-1/-1->0->1 [23] -1/-1/-1->0->1
+ip-26-0-172-142:789259:789871 [4] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-172-142:789259:789871 [4] NCCL INFO Channel 00/0 : 0[4] -> 1[6] via P2P/IPC
+ip-26-0-172-142:789260:789875 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-172-142:789261:789872 [6] NCCL INFO Channel 00/0 : 1[6] -> 0[4] via P2P/IPC
+ip-26-0-172-142:789259:789871 [4] NCCL INFO Channel 01/0 : 0[4] -> 1[6] via P2P/IPC
+ip-26-0-172-142:789261:789872 [6] NCCL INFO Channel 01/0 : 1[6] -> 0[4] via P2P/IPC
+ip-26-0-172-142:789259:789871 [4] NCCL INFO Channel 02/0 : 0[4] -> 1[6] via P2P/IPC
+ip-26-0-172-142:789261:789872 [6] NCCL INFO Channel 02/0 : 1[6] -> 0[4] via P2P/IPC
+ip-26-0-172-142:789259:789871 [4] NCCL INFO Channel 03/0 : 0[4] -> 1[6] via P2P/IPC
+ip-26-0-172-142:789261:789872 [6] NCCL INFO Channel 03/0 : 1[6] -> 0[4] via P2P/IPC
+ip-26-0-172-142:789259:789871 [4] NCCL INFO Channel 04/0 : 0[4] -> 1[6] via P2P/IPC
+ip-26-0-172-142:789261:789872 [6] NCCL INFO Channel 04/0 : 1[6] -> 0[4] via P2P/IPC
+ip-26-0-172-142:789259:789871 [4] NCCL INFO Channel 05/0 : 0[4] -> 1[6] via P2P/IPC
+ip-26-0-172-142:789261:789872 [6] NCCL INFO Channel 05/0 : 1[6] -> 0[4] via P2P/IPC
+ip-26-0-172-142:789259:789871 [4] NCCL INFO Channel 06/0 : 0[4] -> 1[6] via P2P/IPC
+ip-26-0-172-142:789261:789872 [6] NCCL INFO Channel 06/0 : 1[6] -> 0[4] via P2P/IPC
+ip-26-0-172-142:789259:789871 [4] NCCL INFO Channel 07/0 : 0[4] -> 1[6] via P2P/IPC
+ip-26-0-172-142:789261:789872 [6] NCCL INFO Channel 07/0 : 1[6] -> 0[4] via P2P/IPC
+ip-26-0-172-142:789262:789876 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-172-142:789260:789875 [5] NCCL INFO Channel 00/24 :    0   1
+ip-26-0-172-142:789260:789875 [5] NCCL INFO Channel 01/24 :    0   1
+ip-26-0-172-142:789260:789875 [5] NCCL INFO Channel 02/24 :    0   1
+ip-26-0-172-142:789260:789875 [5] NCCL INFO Channel 03/24 :    0   1
+ip-26-0-172-142:789260:789875 [5] NCCL INFO Channel 04/24 :    0   1
+ip-26-0-172-142:789260:789875 [5] NCCL INFO Channel 05/24 :    0   1
+ip-26-0-172-142:789260:789875 [5] NCCL INFO Channel 06/24 :    0   1
+ip-26-0-172-142:789260:789875 [5] NCCL INFO Channel 07/24 :    0   1
+ip-26-0-172-142:789260:789875 [5] NCCL INFO Channel 08/24 :    0   1
+ip-26-0-172-142:789262:789876 [7] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] -1/-1/-1->1->0 [3] -1/-1/-1->1->0 [4] -1/-1/-1->1->0 [5] -1/-1/-1->1->0 [6] 0/-1/-1->1->-1 [7] 0/-1/-1->1->-1 [8] 0/-1/-1->1->-1 [9] 0/-1/-1->1->-1 [10] 0/-1/-1->1->-1 [11] 0/-1/-1->1->-1 [12] -1/-1/-1->1->0 [13] -1/-1/-1->1->0 [14] -1/-1/-1->1->0 [15] -1/-1/-1->1->0 [16] -1/-1/-1->1->0 [17] -1/-1/-1->1->0 [18] 0/-1/-1->1->-1 [19] 0/-1/-1->1->-1 [20] 0/-1/-1->1->-1 [21] 0/-1/-1->1->-1 [22] 0/-1/-1->1->-1 [23] 0/-1/-1->1->-1
+ip-26-0-172-142:789260:789875 [5] NCCL INFO Channel 09/24 :    0   1
+ip-26-0-172-142:789260:789875 [5] NCCL INFO Channel 10/24 :    0   1
+ip-26-0-172-142:789262:789876 [7] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-172-142:789260:789875 [5] NCCL INFO Channel 11/24 :    0   1
+ip-26-0-172-142:789260:789875 [5] NCCL INFO Channel 12/24 :    0   1
+ip-26-0-172-142:789260:789875 [5] NCCL INFO Channel 13/24 :    0   1
+ip-26-0-172-142:789260:789875 [5] NCCL INFO Channel 14/24 :    0   1
+ip-26-0-172-142:789260:789875 [5] NCCL INFO Channel 15/24 :    0   1
+ip-26-0-172-142:789260:789875 [5] NCCL INFO Channel 16/24 :    0   1
+ip-26-0-172-142:789260:789875 [5] NCCL INFO Channel 17/24 :    0   1
+ip-26-0-172-142:789260:789875 [5] NCCL INFO Channel 18/24 :    0   1
+ip-26-0-172-142:789260:789875 [5] NCCL INFO Channel 19/24 :    0   1
+ip-26-0-172-142:789260:789875 [5] NCCL INFO Channel 20/24 :    0   1
+ip-26-0-172-142:789260:789875 [5] NCCL INFO Channel 21/24 :    0   1
+ip-26-0-172-142:789260:789875 [5] NCCL INFO Channel 22/24 :    0   1
+ip-26-0-172-142:789260:789875 [5] NCCL INFO Channel 23/24 :    0   1
+ip-26-0-172-142:789260:789875 [5] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] -1/-1/-1->0->1 [7] -1/-1/-1->0->1 [8] -1/-1/-1->0->1 [9] -1/-1/-1->0->1 [10] -1/-1/-1->0->1 [11] -1/-1/-1->0->1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] -1/-1/-1->0->1 [19] -1/-1/-1->0->1 [20] -1/-1/-1->0->1 [21] -1/-1/-1->0->1 [22] -1/-1/-1->0->1 [23] -1/-1/-1->0->1
+ip-26-0-172-142:789260:789875 [5] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-172-142:789259:789871 [4] NCCL INFO Channel 08/0 : 0[4] -> 1[6] via P2P/IPC
+ip-26-0-172-142:789261:789872 [6] NCCL INFO Channel 08/0 : 1[6] -> 0[4] via P2P/IPC
+ip-26-0-172-142:789259:789871 [4] NCCL INFO Channel 09/0 : 0[4] -> 1[6] via P2P/IPC
+ip-26-0-172-142:789261:789872 [6] NCCL INFO Channel 09/0 : 1[6] -> 0[4] via P2P/IPC
+ip-26-0-172-142:789259:789871 [4] NCCL INFO Channel 10/0 : 0[4] -> 1[6] via P2P/IPC
+ip-26-0-172-142:789261:789872 [6] NCCL INFO Channel 10/0 : 1[6] -> 0[4] via P2P/IPC
+ip-26-0-172-142:789259:789871 [4] NCCL INFO Channel 11/0 : 0[4] -> 1[6] via P2P/IPC
+ip-26-0-172-142:789261:789872 [6] NCCL INFO Channel 11/0 : 1[6] -> 0[4] via P2P/IPC
+ip-26-0-172-142:789259:789871 [4] NCCL INFO Channel 12/0 : 0[4] -> 1[6] via P2P/IPC
+ip-26-0-172-142:789261:789872 [6] NCCL INFO Channel 12/0 : 1[6] -> 0[4] via P2P/IPC
+ip-26-0-172-142:789259:789871 [4] NCCL INFO Channel 13/0 : 0[4] -> 1[6] via P2P/IPC
+ip-26-0-172-142:789261:789872 [6] NCCL INFO Channel 13/0 : 1[6] -> 0[4] via P2P/IPC
+ip-26-0-172-142:789259:789871 [4] NCCL INFO Channel 14/0 : 0[4] -> 1[6] via P2P/IPC
+ip-26-0-172-142:789261:789872 [6] NCCL INFO Channel 14/0 : 1[6] -> 0[4] via P2P/IPC
+ip-26-0-172-142:789259:789871 [4] NCCL INFO Channel 15/0 : 0[4] -> 1[6] via P2P/IPC
+ip-26-0-172-142:789261:789872 [6] NCCL INFO Channel 15/0 : 1[6] -> 0[4] via P2P/IPC
+ip-26-0-172-142:789259:789871 [4] NCCL INFO Channel 16/0 : 0[4] -> 1[6] via P2P/IPC
+ip-26-0-172-142:789261:789872 [6] NCCL INFO Channel 16/0 : 1[6] -> 0[4] via P2P/IPC
+ip-26-0-172-142:789259:789871 [4] NCCL INFO Channel 17/0 : 0[4] -> 1[6] via P2P/IPC
+ip-26-0-172-142:789261:789872 [6] NCCL INFO Channel 17/0 : 1[6] -> 0[4] via P2P/IPC
+ip-26-0-172-142:789259:789871 [4] NCCL INFO Channel 18/0 : 0[4] -> 1[6] via P2P/IPC
+ip-26-0-172-142:789261:789872 [6] NCCL INFO Channel 18/0 : 1[6] -> 0[4] via P2P/IPC
+ip-26-0-172-142:789259:789871 [4] NCCL INFO Channel 19/0 : 0[4] -> 1[6] via P2P/IPC
+ip-26-0-172-142:789261:789872 [6] NCCL INFO Channel 19/0 : 1[6] -> 0[4] via P2P/IPC
+ip-26-0-172-142:789259:789871 [4] NCCL INFO Channel 20/0 : 0[4] -> 1[6] via P2P/IPC
+ip-26-0-172-142:789261:789872 [6] NCCL INFO Channel 20/0 : 1[6] -> 0[4] via P2P/IPC
+ip-26-0-172-142:789259:789871 [4] NCCL INFO Channel 21/0 : 0[4] -> 1[6] via P2P/IPC
+ip-26-0-172-142:789261:789872 [6] NCCL INFO Channel 21/0 : 1[6] -> 0[4] via P2P/IPC
+ip-26-0-172-142:789259:789871 [4] NCCL INFO Channel 22/0 : 0[4] -> 1[6] via P2P/IPC
+ip-26-0-172-142:789261:789872 [6] NCCL INFO Channel 22/0 : 1[6] -> 0[4] via P2P/IPC
+ip-26-0-172-142:789259:789871 [4] NCCL INFO Channel 23/0 : 0[4] -> 1[6] via P2P/IPC
+ip-26-0-172-142:789261:789872 [6] NCCL INFO Channel 23/0 : 1[6] -> 0[4] via P2P/IPC
+ip-26-0-172-142:789260:789875 [5] NCCL INFO Channel 00/0 : 0[5] -> 1[7] via P2P/IPC
+ip-26-0-172-142:789262:789876 [7] NCCL INFO Channel 00/0 : 1[7] -> 0[5] via P2P/IPC
+ip-26-0-172-142:789260:789875 [5] NCCL INFO Channel 01/0 : 0[5] -> 1[7] via P2P/IPC
+ip-26-0-172-142:789262:789876 [7] NCCL INFO Channel 01/0 : 1[7] -> 0[5] via P2P/IPC
+ip-26-0-172-142:789260:789875 [5] NCCL INFO Channel 02/0 : 0[5] -> 1[7] via P2P/IPC
+ip-26-0-172-142:789262:789876 [7] NCCL INFO Channel 02/0 : 1[7] -> 0[5] via P2P/IPC
+ip-26-0-172-142:789260:789875 [5] NCCL INFO Channel 03/0 : 0[5] -> 1[7] via P2P/IPC
+ip-26-0-172-142:789262:789876 [7] NCCL INFO Channel 03/0 : 1[7] -> 0[5] via P2P/IPC
+ip-26-0-172-142:789260:789875 [5] NCCL INFO Channel 04/0 : 0[5] -> 1[7] via P2P/IPC
+ip-26-0-172-142:789262:789876 [7] NCCL INFO Channel 04/0 : 1[7] -> 0[5] via P2P/IPC
+ip-26-0-172-142:789260:789875 [5] NCCL INFO Channel 05/0 : 0[5] -> 1[7] via P2P/IPC
+ip-26-0-172-142:789262:789876 [7] NCCL INFO Channel 05/0 : 1[7] -> 0[5] via P2P/IPC
+ip-26-0-172-142:789260:789875 [5] NCCL INFO Channel 06/0 : 0[5] -> 1[7] via P2P/IPC
+ip-26-0-172-142:789262:789876 [7] NCCL INFO Channel 06/0 : 1[7] -> 0[5] via P2P/IPC
+ip-26-0-172-142:789260:789875 [5] NCCL INFO Channel 07/0 : 0[5] -> 1[7] via P2P/IPC
+ip-26-0-172-142:789262:789876 [7] NCCL INFO Channel 07/0 : 1[7] -> 0[5] via P2P/IPC
+ip-26-0-172-142:789260:789875 [5] NCCL INFO Channel 08/0 : 0[5] -> 1[7] via P2P/IPC
+ip-26-0-172-142:789262:789876 [7] NCCL INFO Channel 08/0 : 1[7] -> 0[5] via P2P/IPC
+ip-26-0-172-142:789260:789875 [5] NCCL INFO Channel 09/0 : 0[5] -> 1[7] via P2P/IPC
+ip-26-0-172-142:789262:789876 [7] NCCL INFO Channel 09/0 : 1[7] -> 0[5] via P2P/IPC
+ip-26-0-172-142:789260:789875 [5] NCCL INFO Channel 10/0 : 0[5] -> 1[7] via P2P/IPC
+ip-26-0-172-142:789262:789876 [7] NCCL INFO Channel 10/0 : 1[7] -> 0[5] via P2P/IPC
+ip-26-0-172-142:789260:789875 [5] NCCL INFO Channel 11/0 : 0[5] -> 1[7] via P2P/IPC
+ip-26-0-172-142:789262:789876 [7] NCCL INFO Channel 11/0 : 1[7] -> 0[5] via P2P/IPC
+ip-26-0-172-142:789260:789875 [5] NCCL INFO Channel 12/0 : 0[5] -> 1[7] via P2P/IPC
+ip-26-0-172-142:789262:789876 [7] NCCL INFO Channel 12/0 : 1[7] -> 0[5] via P2P/IPC
+ip-26-0-172-142:789260:789875 [5] NCCL INFO Channel 13/0 : 0[5] -> 1[7] via P2P/IPC
+ip-26-0-172-142:789262:789876 [7] NCCL INFO Channel 13/0 : 1[7] -> 0[5] via P2P/IPC
+ip-26-0-172-142:789261:789872 [6] NCCL INFO Connected all rings
+ip-26-0-172-142:789260:789875 [5] NCCL INFO Channel 14/0 : 0[5] -> 1[7] via P2P/IPC
+ip-26-0-172-142:789261:789872 [6] NCCL INFO Connected all trees
+ip-26-0-172-142:789261:789872 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-142:789261:789872 [6] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-172-142:789261:789872 [6] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-142:789262:789876 [7] NCCL INFO Channel 14/0 : 1[7] -> 0[5] via P2P/IPC
+ip-26-0-172-142:789260:789875 [5] NCCL INFO Channel 15/0 : 0[5] -> 1[7] via P2P/IPC
+ip-26-0-172-142:789262:789876 [7] NCCL INFO Channel 15/0 : 1[7] -> 0[5] via P2P/IPC
+ip-26-0-172-142:789260:789875 [5] NCCL INFO Channel 16/0 : 0[5] -> 1[7] via P2P/IPC
+ip-26-0-172-142:789262:789876 [7] NCCL INFO Channel 16/0 : 1[7] -> 0[5] via P2P/IPC
+ip-26-0-172-142:789260:789875 [5] NCCL INFO Channel 17/0 : 0[5] -> 1[7] via P2P/IPC
+ip-26-0-172-142:789262:789876 [7] NCCL INFO Channel 17/0 : 1[7] -> 0[5] via P2P/IPC
+ip-26-0-172-142:789259:789871 [4] NCCL INFO Connected all rings
+ip-26-0-172-142:789259:789871 [4] NCCL INFO Connected all trees
+ip-26-0-172-142:789259:789871 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-142:789259:789871 [4] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-172-142:789259:789871 [4] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-142:789260:789875 [5] NCCL INFO Channel 18/0 : 0[5] -> 1[7] via P2P/IPC
+ip-26-0-172-142:789262:789876 [7] NCCL INFO Channel 18/0 : 1[7] -> 0[5] via P2P/IPC
+ip-26-0-172-142:789260:789875 [5] NCCL INFO Channel 19/0 : 0[5] -> 1[7] via P2P/IPC
+ip-26-0-172-142:789262:789876 [7] NCCL INFO Channel 19/0 : 1[7] -> 0[5] via P2P/IPC
+ip-26-0-172-142:789260:789875 [5] NCCL INFO Channel 20/0 : 0[5] -> 1[7] via P2P/IPC
+ip-26-0-172-142:789262:789876 [7] NCCL INFO Channel 20/0 : 1[7] -> 0[5] via P2P/IPC
+ip-26-0-172-142:789260:789875 [5] NCCL INFO Channel 21/0 : 0[5] -> 1[7] via P2P/IPC
+ip-26-0-172-142:789262:789876 [7] NCCL INFO Channel 21/0 : 1[7] -> 0[5] via P2P/IPC
+ip-26-0-172-142:789260:789875 [5] NCCL INFO Channel 22/0 : 0[5] -> 1[7] via P2P/IPC
+ip-26-0-172-142:789262:789876 [7] NCCL INFO Channel 22/0 : 1[7] -> 0[5] via P2P/IPC
+ip-26-0-172-142:789260:789875 [5] NCCL INFO Channel 23/0 : 0[5] -> 1[7] via P2P/IPC
+ip-26-0-172-142:789262:789876 [7] NCCL INFO Channel 23/0 : 1[7] -> 0[5] via P2P/IPC
+ip-26-0-172-142:789261:789872 [6] NCCL INFO comm 0xc150440 rank 1 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0x60ad0b2b62f9ba3b - Init COMPLETE
+ip-26-0-172-142:789259:789871 [4] NCCL INFO comm 0x125eb5b0 rank 0 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0x60ad0b2b62f9ba3b - Init COMPLETE
+ip-26-0-172-142:789259:789883 [4] NCCL INFO Channel 00/1 : 0[4] -> 1[6] via P2P/IPC
+ip-26-0-172-142:789259:789883 [4] NCCL INFO Channel 01/1 : 0[4] -> 1[6] via P2P/IPC
+ip-26-0-172-142:789259:789883 [4] NCCL INFO Channel 02/1 : 0[4] -> 1[6] via P2P/IPC
+ip-26-0-172-142:789259:789883 [4] NCCL INFO Channel 03/1 : 0[4] -> 1[6] via P2P/IPC
+ip-26-0-172-142:789259:789883 [4] NCCL INFO Channel 04/1 : 0[4] -> 1[6] via P2P/IPC
+ip-26-0-172-142:789259:789883 [4] NCCL INFO Channel 05/1 : 0[4] -> 1[6] via P2P/IPC
+ip-26-0-172-142:789259:789883 [4] NCCL INFO Channel 06/1 : 0[4] -> 1[6] via P2P/IPC
+ip-26-0-172-142:789259:789883 [4] NCCL INFO Channel 07/1 : 0[4] -> 1[6] via P2P/IPC
+ip-26-0-172-142:789259:789883 [4] NCCL INFO Channel 08/1 : 0[4] -> 1[6] via P2P/IPC
+ip-26-0-172-142:789259:789883 [4] NCCL INFO Channel 09/1 : 0[4] -> 1[6] via P2P/IPC
+ip-26-0-172-142:789259:789883 [4] NCCL INFO Channel 10/1 : 0[4] -> 1[6] via P2P/IPC
+ip-26-0-172-142:789259:789883 [4] NCCL INFO Channel 11/1 : 0[4] -> 1[6] via P2P/IPC
+ip-26-0-172-142:789259:789883 [4] NCCL INFO Channel 12/1 : 0[4] -> 1[6] via P2P/IPC
+ip-26-0-172-142:789259:789883 [4] NCCL INFO Channel 13/1 : 0[4] -> 1[6] via P2P/IPC
+ip-26-0-172-142:789259:789883 [4] NCCL INFO Channel 14/1 : 0[4] -> 1[6] via P2P/IPC
+ip-26-0-172-142:789259:789883 [4] NCCL INFO Channel 15/1 : 0[4] -> 1[6] via P2P/IPC
+ip-26-0-172-142:789259:789883 [4] NCCL INFO Channel 16/1 : 0[4] -> 1[6] via P2P/IPC
+ip-26-0-172-142:789259:789883 [4] NCCL INFO Channel 17/1 : 0[4] -> 1[6] via P2P/IPC
+ip-26-0-172-142:789259:789883 [4] NCCL INFO Channel 18/1 : 0[4] -> 1[6] via P2P/IPC
+ip-26-0-172-142:789259:789883 [4] NCCL INFO Channel 19/1 : 0[4] -> 1[6] via P2P/IPC
+ip-26-0-172-142:789259:789883 [4] NCCL INFO Channel 20/1 : 0[4] -> 1[6] via P2P/IPC
+ip-26-0-172-142:789260:789875 [5] NCCL INFO Connected all rings
+ip-26-0-172-142:789260:789875 [5] NCCL INFO Connected all trees
+ip-26-0-172-142:789260:789875 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-142:789260:789875 [5] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-172-142:789260:789875 [5] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-142:789259:789883 [4] NCCL INFO Channel 21/1 : 0[4] -> 1[6] via P2P/IPC
+ip-26-0-172-142:789262:789876 [7] NCCL INFO Connected all rings
+ip-26-0-172-142:789262:789876 [7] NCCL INFO Connected all trees
+ip-26-0-172-142:789262:789876 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-142:789262:789876 [7] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-172-142:789262:789876 [7] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-142:789259:789883 [4] NCCL INFO Channel 22/1 : 0[4] -> 1[6] via P2P/IPC
+ip-26-0-172-142:789259:789883 [4] NCCL INFO Channel 23/1 : 0[4] -> 1[6] via P2P/IPC
+ip-26-0-172-142:789259:789883 [4] NCCL INFO Channel 24/1 : 0[4] -> 1[6] via P2P/IPC
+ip-26-0-172-142:789259:789883 [4] NCCL INFO Channel 25/1 : 0[4] -> 1[6] via P2P/IPC
+ip-26-0-172-142:789259:789883 [4] NCCL INFO Channel 26/1 : 0[4] -> 1[6] via P2P/IPC
+ip-26-0-172-142:789259:789883 [4] NCCL INFO Channel 27/1 : 0[4] -> 1[6] via P2P/IPC
+ip-26-0-172-142:789259:789883 [4] NCCL INFO Channel 28/1 : 0[4] -> 1[6] via P2P/IPC
+ip-26-0-172-142:789259:789883 [4] NCCL INFO Channel 29/1 : 0[4] -> 1[6] via P2P/IPC
+ip-26-0-172-142:789259:789883 [4] NCCL INFO Channel 30/1 : 0[4] -> 1[6] via P2P/IPC
+ip-26-0-172-142:789259:789883 [4] NCCL INFO Channel 31/1 : 0[4] -> 1[6] via P2P/IPC
+ip-26-0-172-142:789260:789875 [5] NCCL INFO comm 0x13610200 rank 0 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0xd8ecbab89c6d4ed3 - Init COMPLETE
+ip-26-0-172-142:789262:789876 [7] NCCL INFO comm 0xadd78e0 rank 1 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0xd8ecbab89c6d4ed3 - Init COMPLETE
+ip-26-0-172-142:789260:789887 [5] NCCL INFO Channel 00/1 : 0[5] -> 1[7] via P2P/IPC
+ip-26-0-172-142:789260:789887 [5] NCCL INFO Channel 01/1 : 0[5] -> 1[7] via P2P/IPC
+ip-26-0-172-142:789260:789887 [5] NCCL INFO Channel 02/1 : 0[5] -> 1[7] via P2P/IPC
+ip-26-0-172-142:789260:789887 [5] NCCL INFO Channel 03/1 : 0[5] -> 1[7] via P2P/IPC
+ip-26-0-172-142:789260:789887 [5] NCCL INFO Channel 04/1 : 0[5] -> 1[7] via P2P/IPC
+ip-26-0-172-142:789260:789887 [5] NCCL INFO Channel 05/1 : 0[5] -> 1[7] via P2P/IPC
+ip-26-0-172-142:789260:789887 [5] NCCL INFO Channel 06/1 : 0[5] -> 1[7] via P2P/IPC
+ip-26-0-172-142:789260:789887 [5] NCCL INFO Channel 07/1 : 0[5] -> 1[7] via P2P/IPC
+ip-26-0-172-142:789260:789887 [5] NCCL INFO Channel 08/1 : 0[5] -> 1[7] via P2P/IPC
+ip-26-0-172-142:789260:789887 [5] NCCL INFO Channel 09/1 : 0[5] -> 1[7] via P2P/IPC
+ip-26-0-172-142:789260:789887 [5] NCCL INFO Channel 10/1 : 0[5] -> 1[7] via P2P/IPC
+ip-26-0-172-142:789260:789887 [5] NCCL INFO Channel 11/1 : 0[5] -> 1[7] via P2P/IPC
+ip-26-0-172-142:789260:789887 [5] NCCL INFO Channel 12/1 : 0[5] -> 1[7] via P2P/IPC
+ip-26-0-172-142:789260:789887 [5] NCCL INFO Channel 13/1 : 0[5] -> 1[7] via P2P/IPC
+ip-26-0-172-142:789260:789887 [5] NCCL INFO Channel 14/1 : 0[5] -> 1[7] via P2P/IPC
+ip-26-0-172-142:789260:789887 [5] NCCL INFO Channel 15/1 : 0[5] -> 1[7] via P2P/IPC
+ip-26-0-172-142:789260:789887 [5] NCCL INFO Channel 16/1 : 0[5] -> 1[7] via P2P/IPC
+ip-26-0-172-142:789260:789887 [5] NCCL INFO Channel 17/1 : 0[5] -> 1[7] via P2P/IPC
+ip-26-0-172-142:789260:789887 [5] NCCL INFO Channel 18/1 : 0[5] -> 1[7] via P2P/IPC
+ip-26-0-172-142:789260:789887 [5] NCCL INFO Channel 19/1 : 0[5] -> 1[7] via P2P/IPC
+ip-26-0-172-142:789260:789887 [5] NCCL INFO Channel 20/1 : 0[5] -> 1[7] via P2P/IPC
+ip-26-0-172-142:789260:789887 [5] NCCL INFO Channel 21/1 : 0[5] -> 1[7] via P2P/IPC
+ip-26-0-172-142:789260:789887 [5] NCCL INFO Channel 22/1 : 0[5] -> 1[7] via P2P/IPC
+ip-26-0-172-142:789260:789887 [5] NCCL INFO Channel 23/1 : 0[5] -> 1[7] via P2P/IPC
+ip-26-0-172-142:789260:789887 [5] NCCL INFO Channel 24/1 : 0[5] -> 1[7] via P2P/IPC
+ip-26-0-172-142:789260:789887 [5] NCCL INFO Channel 25/1 : 0[5] -> 1[7] via P2P/IPC
+ip-26-0-172-142:789260:789887 [5] NCCL INFO Channel 26/1 : 0[5] -> 1[7] via P2P/IPC
+ip-26-0-172-142:789260:789887 [5] NCCL INFO Channel 27/1 : 0[5] -> 1[7] via P2P/IPC
+ip-26-0-172-142:789260:789887 [5] NCCL INFO Channel 28/1 : 0[5] -> 1[7] via P2P/IPC
+ip-26-0-172-142:789260:789887 [5] NCCL INFO Channel 29/1 : 0[5] -> 1[7] via P2P/IPC
+ip-26-0-172-142:789260:789887 [5] NCCL INFO Channel 30/1 : 0[5] -> 1[7] via P2P/IPC
+ip-26-0-172-142:789260:789887 [5] NCCL INFO Channel 31/1 : 0[5] -> 1[7] via P2P/IPC
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+ip-26-0-172-142:789261:789910 [6] NCCL INFO Using network Libfabric
+ip-26-0-173-121:3987794:3988338 [0] NCCL INFO Using network Libfabric
+ip-26-0-173-121:3987794:3988338 [0] NCCL INFO comm 0xaf90200 rank 1 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0xe1818f02b412a5b8 - Init START
+ip-26-0-172-142:789261:789910 [6] NCCL INFO comm 0x1338dc80 rank 0 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0xe1818f02b412a5b8 - Init START
+ip-26-0-172-142:789261:789910 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-173-121:3987794:3988338 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-172-142:789261:789910 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-173-121:3987794:3988338 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-172-142:789261:789910 [6] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-172-142:789261:789910 [6] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-172-142:789261:789910 [6] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-172-142:789261:789910 [6] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-172-142:789261:789910 [6] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-172-142:789261:789910 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-173-121:3987794:3988338 [0] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-173-121:3987794:3988338 [0] NCCL INFO P2P Chunksize set to 131072
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+ip-26-0-172-142:789262:789914 [7] NCCL INFO Using network Libfabric
+ip-26-0-173-121:3987795:3988341 [1] NCCL INFO Using network Libfabric
+ip-26-0-173-121:3987795:3988341 [1] NCCL INFO comm 0xbbc0b60 rank 1 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0x86de10ac4addaef5 - Init START
+ip-26-0-172-142:789262:789914 [7] NCCL INFO comm 0x11fa7b80 rank 0 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0x86de10ac4addaef5 - Init START
+ip-26-0-173-121:3987795:3988341 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-172-142:789262:789914 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-172-142:789261:789910 [6] NCCL INFO Channel 00/0 : 1[0] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:789261:789910 [6] NCCL INFO Channel 01/0 : 1[0] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:789261:789910 [6] NCCL INFO Channel 02/0 : 1[0] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:789261:789910 [6] NCCL INFO Channel 03/0 : 1[0] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:789261:789910 [6] NCCL INFO Channel 00/0 : 0[6] -> 1[0] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:789261:789910 [6] NCCL INFO Channel 01/0 : 0[6] -> 1[0] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:789261:789910 [6] NCCL INFO Channel 02/0 : 0[6] -> 1[0] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:789261:789910 [6] NCCL INFO Channel 03/0 : 0[6] -> 1[0] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-121:3987794:3988338 [0] NCCL INFO Channel 00/0 : 0[6] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-121:3987794:3988338 [0] NCCL INFO Channel 01/0 : 0[6] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-121:3987794:3988338 [0] NCCL INFO Channel 02/0 : 0[6] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-121:3987794:3988338 [0] NCCL INFO Channel 03/0 : 0[6] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-121:3987794:3988338 [0] NCCL INFO Channel 00/0 : 1[0] -> 0[6] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-121:3987794:3988338 [0] NCCL INFO Channel 01/0 : 1[0] -> 0[6] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:789262:789914 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-173-121:3987794:3988338 [0] NCCL INFO Channel 02/0 : 1[0] -> 0[6] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-121:3987794:3988338 [0] NCCL INFO Channel 03/0 : 1[0] -> 0[6] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-121:3987795:3988341 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-172-142:789262:789914 [7] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-172-142:789262:789914 [7] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-172-142:789262:789914 [7] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-172-142:789262:789914 [7] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-173-121:3987795:3988341 [1] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-173-121:3987795:3988341 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-142:789262:789914 [7] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-172-142:789262:789914 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-142:789262:789914 [7] NCCL INFO Channel 00/0 : 1[1] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:789262:789914 [7] NCCL INFO Channel 01/0 : 1[1] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:789262:789914 [7] NCCL INFO Channel 02/0 : 1[1] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:789262:789914 [7] NCCL INFO Channel 03/0 : 1[1] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:789262:789914 [7] NCCL INFO Channel 00/0 : 0[7] -> 1[1] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:789262:789914 [7] NCCL INFO Channel 01/0 : 0[7] -> 1[1] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:789262:789914 [7] NCCL INFO Channel 02/0 : 0[7] -> 1[1] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:789262:789914 [7] NCCL INFO Channel 03/0 : 0[7] -> 1[1] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-121:3987795:3988341 [1] NCCL INFO Channel 00/0 : 0[7] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-121:3987795:3988341 [1] NCCL INFO Channel 01/0 : 0[7] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-121:3987795:3988341 [1] NCCL INFO Channel 02/0 : 0[7] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-121:3987795:3988341 [1] NCCL INFO Channel 03/0 : 0[7] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-121:3987795:3988341 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[7] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-121:3987795:3988341 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[7] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-121:3987795:3988341 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[7] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-121:3987795:3988341 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[7] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:789261:789910 [6] NCCL INFO Connected all rings
+ip-26-0-172-142:789261:789910 [6] NCCL INFO Connected all trees
+ip-26-0-172-142:789261:789910 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-142:789261:789910 [6] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-172-142:789261:789910 [6] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-142:789261:789910 [6] NCCL INFO comm 0x1338dc80 rank 0 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0xe1818f02b412a5b8 - Init COMPLETE
+ip-26-0-172-142:789261:789918 [6] NCCL INFO Channel 02/1 : 0[6] -> 1[0] [send] via NET/Libfabric/6/GDRDMA/Shared
+ip-26-0-172-142:789261:789918 [6] NCCL INFO Channel 03/1 : 0[6] -> 1[0] [send] via NET/Libfabric/6/GDRDMA/Shared
+ip-26-0-173-121:3987794:3988338 [0] NCCL INFO Connected all rings
+ip-26-0-173-121:3987794:3988338 [0] NCCL INFO Connected all trees
+ip-26-0-173-121:3987794:3988338 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-173-121:3987794:3988338 [0] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-173-121:3987794:3988338 [0] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-173-121:3987794:3988338 [0] NCCL INFO comm 0xaf90200 rank 1 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0xe1818f02b412a5b8 - Init COMPLETE
+ip-26-0-173-121:3987794:3988345 [0] NCCL INFO Channel 02/1 : 0[6] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA/Shared
+ip-26-0-173-121:3987794:3988345 [0] NCCL INFO Channel 03/1 : 0[6] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA/Shared
+ip-26-0-172-142:789262:789914 [7] NCCL INFO Connected all rings
+ip-26-0-172-142:789262:789914 [7] NCCL INFO Connected all trees
+ip-26-0-172-142:789262:789914 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-142:789262:789914 [7] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-172-142:789262:789914 [7] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-142:789262:789914 [7] NCCL INFO comm 0x11fa7b80 rank 0 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0x86de10ac4addaef5 - Init COMPLETE
+ip-26-0-172-142:789262:789919 [7] NCCL INFO Channel 02/1 : 0[7] -> 1[1] [send] via NET/Libfabric/7/GDRDMA/Shared
+ip-26-0-172-142:789262:789919 [7] NCCL INFO Channel 03/1 : 0[7] -> 1[1] [send] via NET/Libfabric/7/GDRDMA/Shared
+ip-26-0-173-121:3987795:3988341 [1] NCCL INFO Connected all rings
+ip-26-0-173-121:3987795:3988341 [1] NCCL INFO Connected all trees
+ip-26-0-173-121:3987795:3988341 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-173-121:3987795:3988341 [1] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-173-121:3987795:3988341 [1] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-173-121:3987795:3988341 [1] NCCL INFO comm 0xbbc0b60 rank 1 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0x86de10ac4addaef5 - Init COMPLETE
+ip-26-0-173-121:3987795:3988346 [1] NCCL INFO Channel 02/1 : 0[7] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA/Shared
+ip-26-0-173-121:3987795:3988346 [1] NCCL INFO Channel 03/1 : 0[7] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA/Shared
+ip-26-0-172-142:789261:789928 [6] NCCL INFO Channel 02/1 : 1[0] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA/Shared
+ip-26-0-172-142:789261:789928 [6] NCCL INFO Channel 03/1 : 1[0] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA/Shared
+ip-26-0-172-142:789262:789937 [7] NCCL INFO Channel 02/1 : 1[1] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA/Shared
+ip-26-0-172-142:789262:789937 [7] NCCL INFO Channel 03/1 : 1[1] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA/Shared
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+ip-26-0-173-121:3987794:3988350 [0] NCCL INFO Using network Libfabric
+ip-26-0-173-121:3987796:3988351 [2] NCCL INFO Using network Libfabric
+ip-26-0-173-121:3987796:3988351 [2] NCCL INFO comm 0xb4435b0 rank 1 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0x1316d46c976d6b47 - Init START
+ip-26-0-173-121:3987794:3988350 [0] NCCL INFO comm 0x121c3090 rank 0 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0x1316d46c976d6b47 - Init START
+ip-26-0-173-121:3987796:3988351 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-173-121:3987794:3988350 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+ip-26-0-173-121:3987797:3988354 [3] NCCL INFO Using network Libfabric
+ip-26-0-173-121:3987795:3988353 [1] NCCL INFO Using network Libfabric
+ip-26-0-173-121:3987797:3988354 [3] NCCL INFO comm 0xaea75b0 rank 1 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0x3b2df283cc673b5b - Init START
+ip-26-0-173-121:3987795:3988353 [1] NCCL INFO comm 0x12d866f0 rank 0 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0x3b2df283cc673b5b - Init START
+ip-26-0-173-121:3987797:3988354 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-173-121:3987795:3988353 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-173-121:3987796:3988351 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-173-121:3987794:3988350 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-173-121:3987794:3988350 [0] NCCL INFO Channel 00/24 :    0   1
+ip-26-0-173-121:3987794:3988350 [0] NCCL INFO Channel 01/24 :    0   1
+ip-26-0-173-121:3987794:3988350 [0] NCCL INFO Channel 02/24 :    0   1
+ip-26-0-173-121:3987794:3988350 [0] NCCL INFO Channel 03/24 :    0   1
+ip-26-0-173-121:3987794:3988350 [0] NCCL INFO Channel 04/24 :    0   1
+ip-26-0-173-121:3987794:3988350 [0] NCCL INFO Channel 05/24 :    0   1
+ip-26-0-173-121:3987794:3988350 [0] NCCL INFO Channel 06/24 :    0   1
+ip-26-0-173-121:3987794:3988350 [0] NCCL INFO Channel 07/24 :    0   1
+ip-26-0-173-121:3987794:3988350 [0] NCCL INFO Channel 08/24 :    0   1
+ip-26-0-173-121:3987794:3988350 [0] NCCL INFO Channel 09/24 :    0   1
+ip-26-0-173-121:3987794:3988350 [0] NCCL INFO Channel 10/24 :    0   1
+ip-26-0-173-121:3987794:3988350 [0] NCCL INFO Channel 11/24 :    0   1
+ip-26-0-173-121:3987794:3988350 [0] NCCL INFO Channel 12/24 :    0   1
+ip-26-0-173-121:3987794:3988350 [0] NCCL INFO Channel 13/24 :    0   1
+ip-26-0-173-121:3987794:3988350 [0] NCCL INFO Channel 14/24 :    0   1
+ip-26-0-173-121:3987794:3988350 [0] NCCL INFO Channel 15/24 :    0   1
+ip-26-0-173-121:3987794:3988350 [0] NCCL INFO Channel 16/24 :    0   1
+ip-26-0-173-121:3987794:3988350 [0] NCCL INFO Channel 17/24 :    0   1
+ip-26-0-173-121:3987794:3988350 [0] NCCL INFO Channel 18/24 :    0   1
+ip-26-0-173-121:3987794:3988350 [0] NCCL INFO Channel 19/24 :    0   1
+ip-26-0-173-121:3987794:3988350 [0] NCCL INFO Channel 20/24 :    0   1
+ip-26-0-173-121:3987794:3988350 [0] NCCL INFO Channel 21/24 :    0   1
+ip-26-0-173-121:3987794:3988350 [0] NCCL INFO Channel 22/24 :    0   1
+ip-26-0-173-121:3987794:3988350 [0] NCCL INFO Channel 23/24 :    0   1
+ip-26-0-173-121:3987796:3988351 [2] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] -1/-1/-1->1->0 [3] -1/-1/-1->1->0 [4] -1/-1/-1->1->0 [5] -1/-1/-1->1->0 [6] 0/-1/-1->1->-1 [7] 0/-1/-1->1->-1 [8] 0/-1/-1->1->-1 [9] 0/-1/-1->1->-1 [10] 0/-1/-1->1->-1 [11] 0/-1/-1->1->-1 [12] -1/-1/-1->1->0 [13] -1/-1/-1->1->0 [14] -1/-1/-1->1->0 [15] -1/-1/-1->1->0 [16] -1/-1/-1->1->0 [17] -1/-1/-1->1->0 [18] 0/-1/-1->1->-1 [19] 0/-1/-1->1->-1 [20] 0/-1/-1->1->-1 [21] 0/-1/-1->1->-1 [22] 0/-1/-1->1->-1 [23] 0/-1/-1->1->-1
+ip-26-0-173-121:3987796:3988351 [2] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-173-121:3987794:3988350 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] -1/-1/-1->0->1 [7] -1/-1/-1->0->1 [8] -1/-1/-1->0->1 [9] -1/-1/-1->0->1 [10] -1/-1/-1->0->1 [11] -1/-1/-1->0->1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] -1/-1/-1->0->1 [19] -1/-1/-1->0->1 [20] -1/-1/-1->0->1 [21] -1/-1/-1->0->1 [22] -1/-1/-1->0->1 [23] -1/-1/-1->0->1
+ip-26-0-173-121:3987794:3988350 [0] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-173-121:3987797:3988354 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-173-121:3987796:3988351 [2] NCCL INFO Channel 00/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987794:3988350 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-173-121:3987796:3988351 [2] NCCL INFO Channel 01/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987794:3988350 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-173-121:3987796:3988351 [2] NCCL INFO Channel 02/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987794:3988350 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-173-121:3987796:3988351 [2] NCCL INFO Channel 03/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987794:3988350 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-173-121:3987796:3988351 [2] NCCL INFO Channel 04/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987795:3988353 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-173-121:3987796:3988351 [2] NCCL INFO Channel 05/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987795:3988353 [1] NCCL INFO Channel 00/24 :    0   1
+ip-26-0-173-121:3987795:3988353 [1] NCCL INFO Channel 01/24 :    0   1
+ip-26-0-173-121:3987795:3988353 [1] NCCL INFO Channel 02/24 :    0   1
+ip-26-0-173-121:3987795:3988353 [1] NCCL INFO Channel 03/24 :    0   1
+ip-26-0-173-121:3987795:3988353 [1] NCCL INFO Channel 04/24 :    0   1
+ip-26-0-173-121:3987795:3988353 [1] NCCL INFO Channel 05/24 :    0   1
+ip-26-0-173-121:3987795:3988353 [1] NCCL INFO Channel 06/24 :    0   1
+ip-26-0-173-121:3987795:3988353 [1] NCCL INFO Channel 07/24 :    0   1
+ip-26-0-173-121:3987797:3988354 [3] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] -1/-1/-1->1->0 [3] -1/-1/-1->1->0 [4] -1/-1/-1->1->0 [5] -1/-1/-1->1->0 [6] 0/-1/-1->1->-1 [7] 0/-1/-1->1->-1 [8] 0/-1/-1->1->-1 [9] 0/-1/-1->1->-1 [10] 0/-1/-1->1->-1 [11] 0/-1/-1->1->-1 [12] -1/-1/-1->1->0 [13] -1/-1/-1->1->0 [14] -1/-1/-1->1->0 [15] -1/-1/-1->1->0 [16] -1/-1/-1->1->0 [17] -1/-1/-1->1->0 [18] 0/-1/-1->1->-1 [19] 0/-1/-1->1->-1 [20] 0/-1/-1->1->-1 [21] 0/-1/-1->1->-1 [22] 0/-1/-1->1->-1 [23] 0/-1/-1->1->-1
+ip-26-0-173-121:3987795:3988353 [1] NCCL INFO Channel 08/24 :    0   1
+ip-26-0-173-121:3987795:3988353 [1] NCCL INFO Channel 09/24 :    0   1
+ip-26-0-173-121:3987797:3988354 [3] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-173-121:3987795:3988353 [1] NCCL INFO Channel 10/24 :    0   1
+ip-26-0-173-121:3987795:3988353 [1] NCCL INFO Channel 11/24 :    0   1
+ip-26-0-173-121:3987795:3988353 [1] NCCL INFO Channel 12/24 :    0   1
+ip-26-0-173-121:3987795:3988353 [1] NCCL INFO Channel 13/24 :    0   1
+ip-26-0-173-121:3987795:3988353 [1] NCCL INFO Channel 14/24 :    0   1
+ip-26-0-173-121:3987795:3988353 [1] NCCL INFO Channel 15/24 :    0   1
+ip-26-0-173-121:3987795:3988353 [1] NCCL INFO Channel 16/24 :    0   1
+ip-26-0-173-121:3987795:3988353 [1] NCCL INFO Channel 17/24 :    0   1
+ip-26-0-173-121:3987795:3988353 [1] NCCL INFO Channel 18/24 :    0   1
+ip-26-0-173-121:3987795:3988353 [1] NCCL INFO Channel 19/24 :    0   1
+ip-26-0-173-121:3987795:3988353 [1] NCCL INFO Channel 20/24 :    0   1
+ip-26-0-173-121:3987795:3988353 [1] NCCL INFO Channel 21/24 :    0   1
+ip-26-0-173-121:3987795:3988353 [1] NCCL INFO Channel 22/24 :    0   1
+ip-26-0-173-121:3987795:3988353 [1] NCCL INFO Channel 23/24 :    0   1
+ip-26-0-173-121:3987795:3988353 [1] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] -1/-1/-1->0->1 [7] -1/-1/-1->0->1 [8] -1/-1/-1->0->1 [9] -1/-1/-1->0->1 [10] -1/-1/-1->0->1 [11] -1/-1/-1->0->1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] -1/-1/-1->0->1 [19] -1/-1/-1->0->1 [20] -1/-1/-1->0->1 [21] -1/-1/-1->0->1 [22] -1/-1/-1->0->1 [23] -1/-1/-1->0->1
+ip-26-0-173-121:3987795:3988353 [1] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-173-121:3987796:3988351 [2] NCCL INFO Channel 06/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987794:3988350 [0] NCCL INFO Channel 04/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-173-121:3987796:3988351 [2] NCCL INFO Channel 07/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987794:3988350 [0] NCCL INFO Channel 05/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-173-121:3987796:3988351 [2] NCCL INFO Channel 08/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987794:3988350 [0] NCCL INFO Channel 06/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-173-121:3987796:3988351 [2] NCCL INFO Channel 09/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987794:3988350 [0] NCCL INFO Channel 07/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-173-121:3987796:3988351 [2] NCCL INFO Channel 10/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987794:3988350 [0] NCCL INFO Channel 08/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-173-121:3987796:3988351 [2] NCCL INFO Channel 11/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987794:3988350 [0] NCCL INFO Channel 09/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-173-121:3987796:3988351 [2] NCCL INFO Channel 12/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987794:3988350 [0] NCCL INFO Channel 10/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-173-121:3987796:3988351 [2] NCCL INFO Channel 13/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987794:3988350 [0] NCCL INFO Channel 11/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-173-121:3987796:3988351 [2] NCCL INFO Channel 14/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987794:3988350 [0] NCCL INFO Channel 12/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-173-121:3987796:3988351 [2] NCCL INFO Channel 15/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987794:3988350 [0] NCCL INFO Channel 13/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-173-121:3987796:3988351 [2] NCCL INFO Channel 16/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987794:3988350 [0] NCCL INFO Channel 14/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-173-121:3987796:3988351 [2] NCCL INFO Channel 17/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987794:3988350 [0] NCCL INFO Channel 15/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-173-121:3987796:3988351 [2] NCCL INFO Channel 18/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987794:3988350 [0] NCCL INFO Channel 16/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-173-121:3987796:3988351 [2] NCCL INFO Channel 19/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987794:3988350 [0] NCCL INFO Channel 17/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-173-121:3987796:3988351 [2] NCCL INFO Channel 20/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987794:3988350 [0] NCCL INFO Channel 18/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-173-121:3987796:3988351 [2] NCCL INFO Channel 21/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987794:3988350 [0] NCCL INFO Channel 19/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-173-121:3987796:3988351 [2] NCCL INFO Channel 22/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987794:3988350 [0] NCCL INFO Channel 20/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-173-121:3987796:3988351 [2] NCCL INFO Channel 23/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987794:3988350 [0] NCCL INFO Channel 21/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-173-121:3987794:3988350 [0] NCCL INFO Channel 22/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-173-121:3987794:3988350 [0] NCCL INFO Channel 23/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-173-121:3987795:3988353 [1] NCCL INFO Channel 00/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-173-121:3987795:3988353 [1] NCCL INFO Channel 01/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-173-121:3987797:3988354 [3] NCCL INFO Channel 00/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-173-121:3987795:3988353 [1] NCCL INFO Channel 02/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-173-121:3987795:3988353 [1] NCCL INFO Channel 03/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-173-121:3987797:3988354 [3] NCCL INFO Channel 01/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-173-121:3987795:3988353 [1] NCCL INFO Channel 04/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-173-121:3987797:3988354 [3] NCCL INFO Channel 02/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-173-121:3987795:3988353 [1] NCCL INFO Channel 05/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-173-121:3987797:3988354 [3] NCCL INFO Channel 03/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-173-121:3987795:3988353 [1] NCCL INFO Channel 06/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-173-121:3987797:3988354 [3] NCCL INFO Channel 04/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-173-121:3987797:3988354 [3] NCCL INFO Channel 05/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-173-121:3987795:3988353 [1] NCCL INFO Channel 07/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-173-121:3987797:3988354 [3] NCCL INFO Channel 06/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-173-121:3987795:3988353 [1] NCCL INFO Channel 08/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-173-121:3987797:3988354 [3] NCCL INFO Channel 07/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-173-121:3987795:3988353 [1] NCCL INFO Channel 09/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-173-121:3987797:3988354 [3] NCCL INFO Channel 08/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-173-121:3987795:3988353 [1] NCCL INFO Channel 10/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-173-121:3987797:3988354 [3] NCCL INFO Channel 09/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-173-121:3987795:3988353 [1] NCCL INFO Channel 11/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-173-121:3987797:3988354 [3] NCCL INFO Channel 10/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-173-121:3987795:3988353 [1] NCCL INFO Channel 12/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-173-121:3987797:3988354 [3] NCCL INFO Channel 11/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-173-121:3987795:3988353 [1] NCCL INFO Channel 13/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-173-121:3987797:3988354 [3] NCCL INFO Channel 12/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-173-121:3987795:3988353 [1] NCCL INFO Channel 14/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-173-121:3987797:3988354 [3] NCCL INFO Channel 13/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-173-121:3987795:3988353 [1] NCCL INFO Channel 15/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-173-121:3987797:3988354 [3] NCCL INFO Channel 14/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-173-121:3987795:3988353 [1] NCCL INFO Channel 16/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-173-121:3987797:3988354 [3] NCCL INFO Channel 15/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-173-121:3987795:3988353 [1] NCCL INFO Channel 17/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-173-121:3987797:3988354 [3] NCCL INFO Channel 16/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-173-121:3987795:3988353 [1] NCCL INFO Channel 18/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-173-121:3987797:3988354 [3] NCCL INFO Channel 17/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-173-121:3987795:3988353 [1] NCCL INFO Channel 19/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-173-121:3987797:3988354 [3] NCCL INFO Channel 18/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-173-121:3987795:3988353 [1] NCCL INFO Channel 20/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-173-121:3987794:3988350 [0] NCCL INFO Connected all rings
+ip-26-0-173-121:3987794:3988350 [0] NCCL INFO Connected all trees
+ip-26-0-173-121:3987794:3988350 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-173-121:3987794:3988350 [0] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-173-121:3987794:3988350 [0] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-173-121:3987795:3988353 [1] NCCL INFO Channel 21/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-173-121:3987797:3988354 [3] NCCL INFO Channel 19/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-173-121:3987796:3988351 [2] NCCL INFO Connected all rings
+ip-26-0-173-121:3987796:3988351 [2] NCCL INFO Connected all trees
+ip-26-0-173-121:3987796:3988351 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-173-121:3987796:3988351 [2] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-173-121:3987796:3988351 [2] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-173-121:3987795:3988353 [1] NCCL INFO Channel 22/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-173-121:3987797:3988354 [3] NCCL INFO Channel 20/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-173-121:3987795:3988353 [1] NCCL INFO Channel 23/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-173-121:3987797:3988354 [3] NCCL INFO Channel 21/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-173-121:3987797:3988354 [3] NCCL INFO Channel 22/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-173-121:3987797:3988354 [3] NCCL INFO Channel 23/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-173-121:3987794:3988350 [0] NCCL INFO comm 0x121c3090 rank 0 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0x1316d46c976d6b47 - Init COMPLETE
+ip-26-0-173-121:3987796:3988351 [2] NCCL INFO comm 0xb4435b0 rank 1 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0x1316d46c976d6b47 - Init COMPLETE
+ip-26-0-173-121:3987794:3988362 [0] NCCL INFO Channel 00/1 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-173-121:3987794:3988362 [0] NCCL INFO Channel 01/1 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-173-121:3987794:3988362 [0] NCCL INFO Channel 02/1 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-173-121:3987794:3988362 [0] NCCL INFO Channel 03/1 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-173-121:3987794:3988362 [0] NCCL INFO Channel 04/1 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-173-121:3987794:3988362 [0] NCCL INFO Channel 05/1 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-173-121:3987794:3988362 [0] NCCL INFO Channel 06/1 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-173-121:3987794:3988362 [0] NCCL INFO Channel 07/1 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-173-121:3987794:3988362 [0] NCCL INFO Channel 08/1 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-173-121:3987794:3988362 [0] NCCL INFO Channel 09/1 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-173-121:3987794:3988362 [0] NCCL INFO Channel 10/1 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-173-121:3987794:3988362 [0] NCCL INFO Channel 11/1 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-173-121:3987794:3988362 [0] NCCL INFO Channel 12/1 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-173-121:3987794:3988362 [0] NCCL INFO Channel 13/1 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-173-121:3987794:3988362 [0] NCCL INFO Channel 14/1 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-173-121:3987794:3988362 [0] NCCL INFO Channel 15/1 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-173-121:3987794:3988362 [0] NCCL INFO Channel 16/1 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-173-121:3987794:3988362 [0] NCCL INFO Channel 17/1 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-173-121:3987794:3988362 [0] NCCL INFO Channel 18/1 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-173-121:3987794:3988362 [0] NCCL INFO Channel 19/1 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-173-121:3987794:3988362 [0] NCCL INFO Channel 20/1 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-173-121:3987797:3988354 [3] NCCL INFO Connected all rings
+ip-26-0-173-121:3987797:3988354 [3] NCCL INFO Connected all trees
+ip-26-0-173-121:3987797:3988354 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-173-121:3987797:3988354 [3] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-173-121:3987797:3988354 [3] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-173-121:3987794:3988362 [0] NCCL INFO Channel 21/1 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-173-121:3987794:3988362 [0] NCCL INFO Channel 22/1 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-173-121:3987794:3988362 [0] NCCL INFO Channel 23/1 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-173-121:3987794:3988362 [0] NCCL INFO Channel 24/1 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-173-121:3987794:3988362 [0] NCCL INFO Channel 25/1 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-173-121:3987795:3988353 [1] NCCL INFO Connected all rings
+ip-26-0-173-121:3987795:3988353 [1] NCCL INFO Connected all trees
+ip-26-0-173-121:3987795:3988353 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-173-121:3987795:3988353 [1] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-173-121:3987795:3988353 [1] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-173-121:3987794:3988362 [0] NCCL INFO Channel 26/1 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-173-121:3987794:3988362 [0] NCCL INFO Channel 27/1 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-173-121:3987794:3988362 [0] NCCL INFO Channel 28/1 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-173-121:3987794:3988362 [0] NCCL INFO Channel 29/1 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-173-121:3987794:3988362 [0] NCCL INFO Channel 30/1 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-173-121:3987794:3988362 [0] NCCL INFO Channel 31/1 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-173-121:3987795:3988353 [1] NCCL INFO comm 0x12d866f0 rank 0 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0x3b2df283cc673b5b - Init COMPLETE
+ip-26-0-173-121:3987797:3988354 [3] NCCL INFO comm 0xaea75b0 rank 1 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0x3b2df283cc673b5b - Init COMPLETE
+ip-26-0-173-121:3987795:3988366 [1] NCCL INFO Channel 00/1 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-173-121:3987795:3988366 [1] NCCL INFO Channel 01/1 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-173-121:3987795:3988366 [1] NCCL INFO Channel 02/1 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-173-121:3987795:3988366 [1] NCCL INFO Channel 03/1 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-173-121:3987795:3988366 [1] NCCL INFO Channel 04/1 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-173-121:3987795:3988366 [1] NCCL INFO Channel 05/1 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-173-121:3987795:3988366 [1] NCCL INFO Channel 06/1 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-173-121:3987795:3988366 [1] NCCL INFO Channel 07/1 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-173-121:3987795:3988366 [1] NCCL INFO Channel 08/1 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-173-121:3987795:3988366 [1] NCCL INFO Channel 09/1 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-173-121:3987795:3988366 [1] NCCL INFO Channel 10/1 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-173-121:3987795:3988366 [1] NCCL INFO Channel 11/1 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-173-121:3987795:3988366 [1] NCCL INFO Channel 12/1 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-173-121:3987795:3988366 [1] NCCL INFO Channel 13/1 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-173-121:3987795:3988366 [1] NCCL INFO Channel 14/1 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-173-121:3987795:3988366 [1] NCCL INFO Channel 15/1 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-173-121:3987795:3988366 [1] NCCL INFO Channel 16/1 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-173-121:3987795:3988366 [1] NCCL INFO Channel 17/1 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-173-121:3987795:3988366 [1] NCCL INFO Channel 18/1 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-173-121:3987795:3988366 [1] NCCL INFO Channel 19/1 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-173-121:3987795:3988366 [1] NCCL INFO Channel 20/1 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-173-121:3987795:3988366 [1] NCCL INFO Channel 21/1 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-173-121:3987795:3988366 [1] NCCL INFO Channel 22/1 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-173-121:3987795:3988366 [1] NCCL INFO Channel 23/1 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-173-121:3987795:3988366 [1] NCCL INFO Channel 24/1 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-173-121:3987795:3988366 [1] NCCL INFO Channel 25/1 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-173-121:3987795:3988366 [1] NCCL INFO Channel 26/1 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-173-121:3987795:3988366 [1] NCCL INFO Channel 27/1 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-173-121:3987795:3988366 [1] NCCL INFO Channel 28/1 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-173-121:3987795:3988366 [1] NCCL INFO Channel 29/1 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-173-121:3987795:3988366 [1] NCCL INFO Channel 30/1 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-173-121:3987795:3988366 [1] NCCL INFO Channel 31/1 : 0[1] -> 1[3] via P2P/IPC
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+ip-26-0-173-121:3987796:3988388 [2] NCCL INFO Using network Libfabric
+ip-26-0-173-121:3987798:3988389 [4] NCCL INFO Using network Libfabric
+ip-26-0-173-121:3987796:3988388 [2] NCCL INFO comm 0x125f42f0 rank 0 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0x32b1394e33b5e50 - Init START
+ip-26-0-173-121:3987798:3988389 [4] NCCL INFO comm 0xbf78e20 rank 1 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0x32b1394e33b5e50 - Init START
+ip-26-0-173-121:3987798:3988389 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-173-121:3987796:3988388 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-173-121:3987798:3988389 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-173-121:3987796:3988388 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-173-121:3987796:3988388 [2] NCCL INFO Channel 00/24 :    0   1
+ip-26-0-173-121:3987796:3988388 [2] NCCL INFO Channel 01/24 :    0   1
+ip-26-0-173-121:3987796:3988388 [2] NCCL INFO Channel 02/24 :    0   1
+ip-26-0-173-121:3987796:3988388 [2] NCCL INFO Channel 03/24 :    0   1
+ip-26-0-173-121:3987796:3988388 [2] NCCL INFO Channel 04/24 :    0   1
+ip-26-0-173-121:3987796:3988388 [2] NCCL INFO Channel 05/24 :    0   1
+ip-26-0-173-121:3987796:3988388 [2] NCCL INFO Channel 06/24 :    0   1
+ip-26-0-173-121:3987796:3988388 [2] NCCL INFO Channel 07/24 :    0   1
+ip-26-0-173-121:3987796:3988388 [2] NCCL INFO Channel 08/24 :    0   1
+ip-26-0-173-121:3987796:3988388 [2] NCCL INFO Channel 09/24 :    0   1
+ip-26-0-173-121:3987796:3988388 [2] NCCL INFO Channel 10/24 :    0   1
+ip-26-0-173-121:3987796:3988388 [2] NCCL INFO Channel 11/24 :    0   1
+ip-26-0-173-121:3987796:3988388 [2] NCCL INFO Channel 12/24 :    0   1
+ip-26-0-173-121:3987796:3988388 [2] NCCL INFO Channel 13/24 :    0   1
+ip-26-0-173-121:3987796:3988388 [2] NCCL INFO Channel 14/24 :    0   1
+ip-26-0-173-121:3987796:3988388 [2] NCCL INFO Channel 15/24 :    0   1
+ip-26-0-173-121:3987796:3988388 [2] NCCL INFO Channel 16/24 :    0   1
+ip-26-0-173-121:3987796:3988388 [2] NCCL INFO Channel 17/24 :    0   1
+ip-26-0-173-121:3987796:3988388 [2] NCCL INFO Channel 18/24 :    0   1
+ip-26-0-173-121:3987796:3988388 [2] NCCL INFO Channel 19/24 :    0   1
+ip-26-0-173-121:3987796:3988388 [2] NCCL INFO Channel 20/24 :    0   1
+ip-26-0-173-121:3987796:3988388 [2] NCCL INFO Channel 21/24 :    0   1
+ip-26-0-173-121:3987796:3988388 [2] NCCL INFO Channel 22/24 :    0   1
+ip-26-0-173-121:3987796:3988388 [2] NCCL INFO Channel 23/24 :    0   1
+ip-26-0-173-121:3987796:3988388 [2] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] -1/-1/-1->0->1 [7] -1/-1/-1->0->1 [8] -1/-1/-1->0->1 [9] -1/-1/-1->0->1 [10] -1/-1/-1->0->1 [11] -1/-1/-1->0->1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] -1/-1/-1->0->1 [19] -1/-1/-1->0->1 [20] -1/-1/-1->0->1 [21] -1/-1/-1->0->1 [22] -1/-1/-1->0->1 [23] -1/-1/-1->0->1
+ip-26-0-173-121:3987796:3988388 [2] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-173-121:3987798:3988389 [4] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] -1/-1/-1->1->0 [3] -1/-1/-1->1->0 [4] -1/-1/-1->1->0 [5] -1/-1/-1->1->0 [6] 0/-1/-1->1->-1 [7] 0/-1/-1->1->-1 [8] 0/-1/-1->1->-1 [9] 0/-1/-1->1->-1 [10] 0/-1/-1->1->-1 [11] 0/-1/-1->1->-1 [12] -1/-1/-1->1->0 [13] -1/-1/-1->1->0 [14] -1/-1/-1->1->0 [15] -1/-1/-1->1->0 [16] -1/-1/-1->1->0 [17] -1/-1/-1->1->0 [18] 0/-1/-1->1->-1 [19] 0/-1/-1->1->-1 [20] 0/-1/-1->1->-1 [21] 0/-1/-1->1->-1 [22] 0/-1/-1->1->-1 [23] 0/-1/-1->1->-1
+ip-26-0-173-121:3987798:3988389 [4] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-173-121:3987796:3988388 [2] NCCL INFO Channel 00/0 : 0[2] -> 1[4] via P2P/IPC
+ip-26-0-173-121:3987796:3988388 [2] NCCL INFO Channel 01/0 : 0[2] -> 1[4] via P2P/IPC
+ip-26-0-173-121:3987796:3988388 [2] NCCL INFO Channel 02/0 : 0[2] -> 1[4] via P2P/IPC
+ip-26-0-173-121:3987798:3988389 [4] NCCL INFO Channel 00/0 : 1[4] -> 0[2] via P2P/IPC
+ip-26-0-173-121:3987796:3988388 [2] NCCL INFO Channel 03/0 : 0[2] -> 1[4] via P2P/IPC
+ip-26-0-173-121:3987798:3988389 [4] NCCL INFO Channel 01/0 : 1[4] -> 0[2] via P2P/IPC
+ip-26-0-173-121:3987796:3988388 [2] NCCL INFO Channel 04/0 : 0[2] -> 1[4] via P2P/IPC
+ip-26-0-173-121:3987798:3988389 [4] NCCL INFO Channel 02/0 : 1[4] -> 0[2] via P2P/IPC
+ip-26-0-173-121:3987796:3988388 [2] NCCL INFO Channel 05/0 : 0[2] -> 1[4] via P2P/IPC
+ip-26-0-173-121:3987798:3988389 [4] NCCL INFO Channel 03/0 : 1[4] -> 0[2] via P2P/IPC
+ip-26-0-173-121:3987796:3988388 [2] NCCL INFO Channel 06/0 : 0[2] -> 1[4] via P2P/IPC
+ip-26-0-173-121:3987798:3988389 [4] NCCL INFO Channel 04/0 : 1[4] -> 0[2] via P2P/IPC
+ip-26-0-173-121:3987796:3988388 [2] NCCL INFO Channel 07/0 : 0[2] -> 1[4] via P2P/IPC
+ip-26-0-173-121:3987798:3988389 [4] NCCL INFO Channel 05/0 : 1[4] -> 0[2] via P2P/IPC
+ip-26-0-173-121:3987796:3988388 [2] NCCL INFO Channel 08/0 : 0[2] -> 1[4] via P2P/IPC
+ip-26-0-173-121:3987798:3988389 [4] NCCL INFO Channel 06/0 : 1[4] -> 0[2] via P2P/IPC
+ip-26-0-173-121:3987796:3988388 [2] NCCL INFO Channel 09/0 : 0[2] -> 1[4] via P2P/IPC
+ip-26-0-173-121:3987798:3988389 [4] NCCL INFO Channel 07/0 : 1[4] -> 0[2] via P2P/IPC
+ip-26-0-173-121:3987796:3988388 [2] NCCL INFO Channel 10/0 : 0[2] -> 1[4] via P2P/IPC
+ip-26-0-173-121:3987798:3988389 [4] NCCL INFO Channel 08/0 : 1[4] -> 0[2] via P2P/IPC
+ip-26-0-173-121:3987796:3988388 [2] NCCL INFO Channel 11/0 : 0[2] -> 1[4] via P2P/IPC
+ip-26-0-173-121:3987798:3988389 [4] NCCL INFO Channel 09/0 : 1[4] -> 0[2] via P2P/IPC
+ip-26-0-173-121:3987796:3988388 [2] NCCL INFO Channel 12/0 : 0[2] -> 1[4] via P2P/IPC
+ip-26-0-173-121:3987798:3988389 [4] NCCL INFO Channel 10/0 : 1[4] -> 0[2] via P2P/IPC
+ip-26-0-173-121:3987796:3988388 [2] NCCL INFO Channel 13/0 : 0[2] -> 1[4] via P2P/IPC
+ip-26-0-173-121:3987798:3988389 [4] NCCL INFO Channel 11/0 : 1[4] -> 0[2] via P2P/IPC
+ip-26-0-173-121:3987796:3988388 [2] NCCL INFO Channel 14/0 : 0[2] -> 1[4] via P2P/IPC
+ip-26-0-173-121:3987798:3988389 [4] NCCL INFO Channel 12/0 : 1[4] -> 0[2] via P2P/IPC
+ip-26-0-173-121:3987796:3988388 [2] NCCL INFO Channel 15/0 : 0[2] -> 1[4] via P2P/IPC
+ip-26-0-173-121:3987798:3988389 [4] NCCL INFO Channel 13/0 : 1[4] -> 0[2] via P2P/IPC
+ip-26-0-173-121:3987796:3988388 [2] NCCL INFO Channel 16/0 : 0[2] -> 1[4] via P2P/IPC
+ip-26-0-173-121:3987798:3988389 [4] NCCL INFO Channel 14/0 : 1[4] -> 0[2] via P2P/IPC
+ip-26-0-173-121:3987796:3988388 [2] NCCL INFO Channel 17/0 : 0[2] -> 1[4] via P2P/IPC
+ip-26-0-173-121:3987798:3988389 [4] NCCL INFO Channel 15/0 : 1[4] -> 0[2] via P2P/IPC
+ip-26-0-173-121:3987796:3988388 [2] NCCL INFO Channel 18/0 : 0[2] -> 1[4] via P2P/IPC
+ip-26-0-173-121:3987798:3988389 [4] NCCL INFO Channel 16/0 : 1[4] -> 0[2] via P2P/IPC
+ip-26-0-173-121:3987796:3988388 [2] NCCL INFO Channel 19/0 : 0[2] -> 1[4] via P2P/IPC
+ip-26-0-173-121:3987798:3988389 [4] NCCL INFO Channel 17/0 : 1[4] -> 0[2] via P2P/IPC
+ip-26-0-173-121:3987796:3988388 [2] NCCL INFO Channel 20/0 : 0[2] -> 1[4] via P2P/IPC
+ip-26-0-173-121:3987798:3988389 [4] NCCL INFO Channel 18/0 : 1[4] -> 0[2] via P2P/IPC
+ip-26-0-173-121:3987796:3988388 [2] NCCL INFO Channel 21/0 : 0[2] -> 1[4] via P2P/IPC
+ip-26-0-173-121:3987798:3988389 [4] NCCL INFO Channel 19/0 : 1[4] -> 0[2] via P2P/IPC
+ip-26-0-173-121:3987796:3988388 [2] NCCL INFO Channel 22/0 : 0[2] -> 1[4] via P2P/IPC
+ip-26-0-173-121:3987798:3988389 [4] NCCL INFO Channel 20/0 : 1[4] -> 0[2] via P2P/IPC
+ip-26-0-173-121:3987796:3988388 [2] NCCL INFO Channel 23/0 : 0[2] -> 1[4] via P2P/IPC
+ip-26-0-173-121:3987798:3988389 [4] NCCL INFO Channel 21/0 : 1[4] -> 0[2] via P2P/IPC
+ip-26-0-173-121:3987798:3988389 [4] NCCL INFO Channel 22/0 : 1[4] -> 0[2] via P2P/IPC
+ip-26-0-173-121:3987798:3988389 [4] NCCL INFO Channel 23/0 : 1[4] -> 0[2] via P2P/IPC
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+ip-26-0-173-121:3987797:3988393 [3] NCCL INFO Using network Libfabric
+ip-26-0-173-121:3987799:3988394 [5] NCCL INFO Using network Libfabric
+ip-26-0-173-121:3987799:3988394 [5] NCCL INFO comm 0xb714960 rank 1 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0xf9c0c705ebdc7081 - Init START
+ip-26-0-173-121:3987797:3988393 [3] NCCL INFO comm 0x120c88f0 rank 0 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0xf9c0c705ebdc7081 - Init START
+ip-26-0-173-121:3987799:3988394 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-173-121:3987797:3988393 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-173-121:3987798:3988389 [4] NCCL INFO Connected all rings
+ip-26-0-173-121:3987798:3988389 [4] NCCL INFO Connected all trees
+ip-26-0-173-121:3987798:3988389 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-173-121:3987798:3988389 [4] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-173-121:3987798:3988389 [4] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-173-121:3987796:3988388 [2] NCCL INFO Connected all rings
+ip-26-0-173-121:3987796:3988388 [2] NCCL INFO Connected all trees
+ip-26-0-173-121:3987796:3988388 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-173-121:3987796:3988388 [2] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-173-121:3987796:3988388 [2] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-173-121:3987796:3988388 [2] NCCL INFO comm 0x125f42f0 rank 0 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0x32b1394e33b5e50 - Init COMPLETE
+ip-26-0-173-121:3987798:3988389 [4] NCCL INFO comm 0xbf78e20 rank 1 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0x32b1394e33b5e50 - Init COMPLETE
+ip-26-0-173-121:3987796:3988397 [2] NCCL INFO Channel 00/1 : 0[2] -> 1[4] via P2P/IPC
+ip-26-0-173-121:3987796:3988397 [2] NCCL INFO Channel 01/1 : 0[2] -> 1[4] via P2P/IPC
+ip-26-0-173-121:3987796:3988397 [2] NCCL INFO Channel 02/1 : 0[2] -> 1[4] via P2P/IPC
+ip-26-0-173-121:3987796:3988397 [2] NCCL INFO Channel 03/1 : 0[2] -> 1[4] via P2P/IPC
+ip-26-0-173-121:3987796:3988397 [2] NCCL INFO Channel 04/1 : 0[2] -> 1[4] via P2P/IPC
+ip-26-0-173-121:3987796:3988397 [2] NCCL INFO Channel 05/1 : 0[2] -> 1[4] via P2P/IPC
+ip-26-0-173-121:3987796:3988397 [2] NCCL INFO Channel 06/1 : 0[2] -> 1[4] via P2P/IPC
+ip-26-0-173-121:3987796:3988397 [2] NCCL INFO Channel 07/1 : 0[2] -> 1[4] via P2P/IPC
+ip-26-0-173-121:3987796:3988397 [2] NCCL INFO Channel 08/1 : 0[2] -> 1[4] via P2P/IPC
+ip-26-0-173-121:3987796:3988397 [2] NCCL INFO Channel 09/1 : 0[2] -> 1[4] via P2P/IPC
+ip-26-0-173-121:3987796:3988397 [2] NCCL INFO Channel 10/1 : 0[2] -> 1[4] via P2P/IPC
+ip-26-0-173-121:3987796:3988397 [2] NCCL INFO Channel 11/1 : 0[2] -> 1[4] via P2P/IPC
+ip-26-0-173-121:3987796:3988397 [2] NCCL INFO Channel 12/1 : 0[2] -> 1[4] via P2P/IPC
+ip-26-0-173-121:3987796:3988397 [2] NCCL INFO Channel 13/1 : 0[2] -> 1[4] via P2P/IPC
+ip-26-0-173-121:3987796:3988397 [2] NCCL INFO Channel 14/1 : 0[2] -> 1[4] via P2P/IPC
+ip-26-0-173-121:3987796:3988397 [2] NCCL INFO Channel 15/1 : 0[2] -> 1[4] via P2P/IPC
+ip-26-0-173-121:3987796:3988397 [2] NCCL INFO Channel 16/1 : 0[2] -> 1[4] via P2P/IPC
+ip-26-0-173-121:3987796:3988397 [2] NCCL INFO Channel 17/1 : 0[2] -> 1[4] via P2P/IPC
+ip-26-0-173-121:3987796:3988397 [2] NCCL INFO Channel 18/1 : 0[2] -> 1[4] via P2P/IPC
+ip-26-0-173-121:3987796:3988397 [2] NCCL INFO Channel 19/1 : 0[2] -> 1[4] via P2P/IPC
+ip-26-0-173-121:3987796:3988397 [2] NCCL INFO Channel 20/1 : 0[2] -> 1[4] via P2P/IPC
+ip-26-0-173-121:3987796:3988397 [2] NCCL INFO Channel 21/1 : 0[2] -> 1[4] via P2P/IPC
+ip-26-0-173-121:3987796:3988397 [2] NCCL INFO Channel 22/1 : 0[2] -> 1[4] via P2P/IPC
+ip-26-0-173-121:3987796:3988397 [2] NCCL INFO Channel 23/1 : 0[2] -> 1[4] via P2P/IPC
+ip-26-0-173-121:3987796:3988397 [2] NCCL INFO Channel 24/1 : 0[2] -> 1[4] via P2P/IPC
+ip-26-0-173-121:3987796:3988397 [2] NCCL INFO Channel 25/1 : 0[2] -> 1[4] via P2P/IPC
+ip-26-0-173-121:3987796:3988397 [2] NCCL INFO Channel 26/1 : 0[2] -> 1[4] via P2P/IPC
+ip-26-0-173-121:3987796:3988397 [2] NCCL INFO Channel 27/1 : 0[2] -> 1[4] via P2P/IPC
+ip-26-0-173-121:3987796:3988397 [2] NCCL INFO Channel 28/1 : 0[2] -> 1[4] via P2P/IPC
+ip-26-0-173-121:3987796:3988397 [2] NCCL INFO Channel 29/1 : 0[2] -> 1[4] via P2P/IPC
+ip-26-0-173-121:3987796:3988397 [2] NCCL INFO Channel 30/1 : 0[2] -> 1[4] via P2P/IPC
+ip-26-0-173-121:3987796:3988397 [2] NCCL INFO Channel 31/1 : 0[2] -> 1[4] via P2P/IPC
+ip-26-0-173-121:3987799:3988394 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-173-121:3987797:3988393 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-173-121:3987799:3988394 [5] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] -1/-1/-1->1->0 [3] -1/-1/-1->1->0 [4] -1/-1/-1->1->0 [5] -1/-1/-1->1->0 [6] 0/-1/-1->1->-1 [7] 0/-1/-1->1->-1 [8] 0/-1/-1->1->-1 [9] 0/-1/-1->1->-1 [10] 0/-1/-1->1->-1 [11] 0/-1/-1->1->-1 [12] -1/-1/-1->1->0 [13] -1/-1/-1->1->0 [14] -1/-1/-1->1->0 [15] -1/-1/-1->1->0 [16] -1/-1/-1->1->0 [17] -1/-1/-1->1->0 [18] 0/-1/-1->1->-1 [19] 0/-1/-1->1->-1 [20] 0/-1/-1->1->-1 [21] 0/-1/-1->1->-1 [22] 0/-1/-1->1->-1 [23] 0/-1/-1->1->-1
+ip-26-0-173-121:3987797:3988393 [3] NCCL INFO Channel 00/24 :    0   1
+ip-26-0-173-121:3987799:3988394 [5] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-173-121:3987797:3988393 [3] NCCL INFO Channel 01/24 :    0   1
+ip-26-0-173-121:3987797:3988393 [3] NCCL INFO Channel 02/24 :    0   1
+ip-26-0-173-121:3987797:3988393 [3] NCCL INFO Channel 03/24 :    0   1
+ip-26-0-173-121:3987797:3988393 [3] NCCL INFO Channel 04/24 :    0   1
+ip-26-0-173-121:3987797:3988393 [3] NCCL INFO Channel 05/24 :    0   1
+ip-26-0-173-121:3987797:3988393 [3] NCCL INFO Channel 06/24 :    0   1
+ip-26-0-173-121:3987797:3988393 [3] NCCL INFO Channel 07/24 :    0   1
+ip-26-0-173-121:3987797:3988393 [3] NCCL INFO Channel 08/24 :    0   1
+ip-26-0-173-121:3987797:3988393 [3] NCCL INFO Channel 09/24 :    0   1
+ip-26-0-173-121:3987797:3988393 [3] NCCL INFO Channel 10/24 :    0   1
+ip-26-0-173-121:3987797:3988393 [3] NCCL INFO Channel 11/24 :    0   1
+ip-26-0-173-121:3987797:3988393 [3] NCCL INFO Channel 12/24 :    0   1
+ip-26-0-173-121:3987797:3988393 [3] NCCL INFO Channel 13/24 :    0   1
+ip-26-0-173-121:3987797:3988393 [3] NCCL INFO Channel 14/24 :    0   1
+ip-26-0-173-121:3987797:3988393 [3] NCCL INFO Channel 15/24 :    0   1
+ip-26-0-173-121:3987797:3988393 [3] NCCL INFO Channel 16/24 :    0   1
+ip-26-0-173-121:3987797:3988393 [3] NCCL INFO Channel 17/24 :    0   1
+ip-26-0-173-121:3987797:3988393 [3] NCCL INFO Channel 18/24 :    0   1
+ip-26-0-173-121:3987797:3988393 [3] NCCL INFO Channel 19/24 :    0   1
+ip-26-0-173-121:3987797:3988393 [3] NCCL INFO Channel 20/24 :    0   1
+ip-26-0-173-121:3987797:3988393 [3] NCCL INFO Channel 21/24 :    0   1
+ip-26-0-173-121:3987797:3988393 [3] NCCL INFO Channel 22/24 :    0   1
+ip-26-0-173-121:3987797:3988393 [3] NCCL INFO Channel 23/24 :    0   1
+ip-26-0-173-121:3987797:3988393 [3] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] -1/-1/-1->0->1 [7] -1/-1/-1->0->1 [8] -1/-1/-1->0->1 [9] -1/-1/-1->0->1 [10] -1/-1/-1->0->1 [11] -1/-1/-1->0->1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] -1/-1/-1->0->1 [19] -1/-1/-1->0->1 [20] -1/-1/-1->0->1 [21] -1/-1/-1->0->1 [22] -1/-1/-1->0->1 [23] -1/-1/-1->0->1
+ip-26-0-173-121:3987797:3988393 [3] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-173-121:3987799:3988394 [5] NCCL INFO Channel 00/0 : 1[5] -> 0[3] via P2P/IPC
+ip-26-0-173-121:3987797:3988393 [3] NCCL INFO Channel 00/0 : 0[3] -> 1[5] via P2P/IPC
+ip-26-0-173-121:3987799:3988394 [5] NCCL INFO Channel 01/0 : 1[5] -> 0[3] via P2P/IPC
+ip-26-0-173-121:3987797:3988393 [3] NCCL INFO Channel 01/0 : 0[3] -> 1[5] via P2P/IPC
+ip-26-0-173-121:3987799:3988394 [5] NCCL INFO Channel 02/0 : 1[5] -> 0[3] via P2P/IPC
+ip-26-0-173-121:3987797:3988393 [3] NCCL INFO Channel 02/0 : 0[3] -> 1[5] via P2P/IPC
+ip-26-0-173-121:3987799:3988394 [5] NCCL INFO Channel 03/0 : 1[5] -> 0[3] via P2P/IPC
+ip-26-0-173-121:3987797:3988393 [3] NCCL INFO Channel 03/0 : 0[3] -> 1[5] via P2P/IPC
+ip-26-0-173-121:3987799:3988394 [5] NCCL INFO Channel 04/0 : 1[5] -> 0[3] via P2P/IPC
+ip-26-0-173-121:3987797:3988393 [3] NCCL INFO Channel 04/0 : 0[3] -> 1[5] via P2P/IPC
+ip-26-0-173-121:3987799:3988394 [5] NCCL INFO Channel 05/0 : 1[5] -> 0[3] via P2P/IPC
+ip-26-0-173-121:3987797:3988393 [3] NCCL INFO Channel 05/0 : 0[3] -> 1[5] via P2P/IPC
+ip-26-0-173-121:3987799:3988394 [5] NCCL INFO Channel 06/0 : 1[5] -> 0[3] via P2P/IPC
+ip-26-0-173-121:3987797:3988393 [3] NCCL INFO Channel 06/0 : 0[3] -> 1[5] via P2P/IPC
+ip-26-0-173-121:3987799:3988394 [5] NCCL INFO Channel 07/0 : 1[5] -> 0[3] via P2P/IPC
+ip-26-0-173-121:3987797:3988393 [3] NCCL INFO Channel 07/0 : 0[3] -> 1[5] via P2P/IPC
+ip-26-0-173-121:3987799:3988394 [5] NCCL INFO Channel 08/0 : 1[5] -> 0[3] via P2P/IPC
+ip-26-0-173-121:3987797:3988393 [3] NCCL INFO Channel 08/0 : 0[3] -> 1[5] via P2P/IPC
+ip-26-0-173-121:3987799:3988394 [5] NCCL INFO Channel 09/0 : 1[5] -> 0[3] via P2P/IPC
+ip-26-0-173-121:3987797:3988393 [3] NCCL INFO Channel 09/0 : 0[3] -> 1[5] via P2P/IPC
+ip-26-0-173-121:3987799:3988394 [5] NCCL INFO Channel 10/0 : 1[5] -> 0[3] via P2P/IPC
+ip-26-0-173-121:3987797:3988393 [3] NCCL INFO Channel 10/0 : 0[3] -> 1[5] via P2P/IPC
+ip-26-0-173-121:3987799:3988394 [5] NCCL INFO Channel 11/0 : 1[5] -> 0[3] via P2P/IPC
+ip-26-0-173-121:3987797:3988393 [3] NCCL INFO Channel 11/0 : 0[3] -> 1[5] via P2P/IPC
+ip-26-0-173-121:3987799:3988394 [5] NCCL INFO Channel 12/0 : 1[5] -> 0[3] via P2P/IPC
+ip-26-0-173-121:3987797:3988393 [3] NCCL INFO Channel 12/0 : 0[3] -> 1[5] via P2P/IPC
+ip-26-0-173-121:3987799:3988394 [5] NCCL INFO Channel 13/0 : 1[5] -> 0[3] via P2P/IPC
+ip-26-0-173-121:3987797:3988393 [3] NCCL INFO Channel 13/0 : 0[3] -> 1[5] via P2P/IPC
+ip-26-0-173-121:3987799:3988394 [5] NCCL INFO Channel 14/0 : 1[5] -> 0[3] via P2P/IPC
+ip-26-0-173-121:3987797:3988393 [3] NCCL INFO Channel 14/0 : 0[3] -> 1[5] via P2P/IPC
+ip-26-0-173-121:3987799:3988394 [5] NCCL INFO Channel 15/0 : 1[5] -> 0[3] via P2P/IPC
+ip-26-0-173-121:3987797:3988393 [3] NCCL INFO Channel 15/0 : 0[3] -> 1[5] via P2P/IPC
+ip-26-0-173-121:3987799:3988394 [5] NCCL INFO Channel 16/0 : 1[5] -> 0[3] via P2P/IPC
+ip-26-0-173-121:3987797:3988393 [3] NCCL INFO Channel 16/0 : 0[3] -> 1[5] via P2P/IPC
+ip-26-0-173-121:3987799:3988394 [5] NCCL INFO Channel 17/0 : 1[5] -> 0[3] via P2P/IPC
+ip-26-0-173-121:3987797:3988393 [3] NCCL INFO Channel 17/0 : 0[3] -> 1[5] via P2P/IPC
+ip-26-0-173-121:3987799:3988394 [5] NCCL INFO Channel 18/0 : 1[5] -> 0[3] via P2P/IPC
+ip-26-0-173-121:3987797:3988393 [3] NCCL INFO Channel 18/0 : 0[3] -> 1[5] via P2P/IPC
+ip-26-0-173-121:3987799:3988394 [5] NCCL INFO Channel 19/0 : 1[5] -> 0[3] via P2P/IPC
+ip-26-0-173-121:3987797:3988393 [3] NCCL INFO Channel 19/0 : 0[3] -> 1[5] via P2P/IPC
+ip-26-0-173-121:3987799:3988394 [5] NCCL INFO Channel 20/0 : 1[5] -> 0[3] via P2P/IPC
+ip-26-0-173-121:3987797:3988393 [3] NCCL INFO Channel 20/0 : 0[3] -> 1[5] via P2P/IPC
+ip-26-0-173-121:3987799:3988394 [5] NCCL INFO Channel 21/0 : 1[5] -> 0[3] via P2P/IPC
+ip-26-0-173-121:3987797:3988393 [3] NCCL INFO Channel 21/0 : 0[3] -> 1[5] via P2P/IPC
+ip-26-0-173-121:3987799:3988394 [5] NCCL INFO Channel 22/0 : 1[5] -> 0[3] via P2P/IPC
+ip-26-0-173-121:3987797:3988393 [3] NCCL INFO Channel 22/0 : 0[3] -> 1[5] via P2P/IPC
+ip-26-0-173-121:3987799:3988394 [5] NCCL INFO Channel 23/0 : 1[5] -> 0[3] via P2P/IPC
+ip-26-0-173-121:3987797:3988393 [3] NCCL INFO Channel 23/0 : 0[3] -> 1[5] via P2P/IPC
+ip-26-0-173-121:3987799:3988394 [5] NCCL INFO Connected all rings
+ip-26-0-173-121:3987799:3988394 [5] NCCL INFO Connected all trees
+ip-26-0-173-121:3987799:3988394 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-173-121:3987799:3988394 [5] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-173-121:3987799:3988394 [5] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-173-121:3987797:3988393 [3] NCCL INFO Connected all rings
+ip-26-0-173-121:3987797:3988393 [3] NCCL INFO Connected all trees
+ip-26-0-173-121:3987797:3988393 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-173-121:3987797:3988393 [3] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-173-121:3987797:3988393 [3] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-173-121:3987799:3988394 [5] NCCL INFO comm 0xb714960 rank 1 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0xf9c0c705ebdc7081 - Init COMPLETE
+ip-26-0-173-121:3987797:3988393 [3] NCCL INFO comm 0x120c88f0 rank 0 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0xf9c0c705ebdc7081 - Init COMPLETE
+ip-26-0-173-121:3987797:3988405 [3] NCCL INFO Channel 00/1 : 0[3] -> 1[5] via P2P/IPC
+ip-26-0-173-121:3987797:3988405 [3] NCCL INFO Channel 01/1 : 0[3] -> 1[5] via P2P/IPC
+ip-26-0-173-121:3987797:3988405 [3] NCCL INFO Channel 02/1 : 0[3] -> 1[5] via P2P/IPC
+ip-26-0-173-121:3987797:3988405 [3] NCCL INFO Channel 03/1 : 0[3] -> 1[5] via P2P/IPC
+ip-26-0-173-121:3987797:3988405 [3] NCCL INFO Channel 04/1 : 0[3] -> 1[5] via P2P/IPC
+ip-26-0-173-121:3987797:3988405 [3] NCCL INFO Channel 05/1 : 0[3] -> 1[5] via P2P/IPC
+ip-26-0-173-121:3987797:3988405 [3] NCCL INFO Channel 06/1 : 0[3] -> 1[5] via P2P/IPC
+ip-26-0-173-121:3987797:3988405 [3] NCCL INFO Channel 07/1 : 0[3] -> 1[5] via P2P/IPC
+ip-26-0-173-121:3987797:3988405 [3] NCCL INFO Channel 08/1 : 0[3] -> 1[5] via P2P/IPC
+ip-26-0-173-121:3987797:3988405 [3] NCCL INFO Channel 09/1 : 0[3] -> 1[5] via P2P/IPC
+ip-26-0-173-121:3987797:3988405 [3] NCCL INFO Channel 10/1 : 0[3] -> 1[5] via P2P/IPC
+ip-26-0-173-121:3987797:3988405 [3] NCCL INFO Channel 11/1 : 0[3] -> 1[5] via P2P/IPC
+ip-26-0-173-121:3987797:3988405 [3] NCCL INFO Channel 12/1 : 0[3] -> 1[5] via P2P/IPC
+ip-26-0-173-121:3987797:3988405 [3] NCCL INFO Channel 13/1 : 0[3] -> 1[5] via P2P/IPC
+ip-26-0-173-121:3987797:3988405 [3] NCCL INFO Channel 14/1 : 0[3] -> 1[5] via P2P/IPC
+ip-26-0-173-121:3987797:3988405 [3] NCCL INFO Channel 15/1 : 0[3] -> 1[5] via P2P/IPC
+ip-26-0-173-121:3987797:3988405 [3] NCCL INFO Channel 16/1 : 0[3] -> 1[5] via P2P/IPC
+ip-26-0-173-121:3987797:3988405 [3] NCCL INFO Channel 17/1 : 0[3] -> 1[5] via P2P/IPC
+ip-26-0-173-121:3987797:3988405 [3] NCCL INFO Channel 18/1 : 0[3] -> 1[5] via P2P/IPC
+ip-26-0-173-121:3987797:3988405 [3] NCCL INFO Channel 19/1 : 0[3] -> 1[5] via P2P/IPC
+ip-26-0-173-121:3987797:3988405 [3] NCCL INFO Channel 20/1 : 0[3] -> 1[5] via P2P/IPC
+ip-26-0-173-121:3987797:3988405 [3] NCCL INFO Channel 21/1 : 0[3] -> 1[5] via P2P/IPC
+ip-26-0-173-121:3987797:3988405 [3] NCCL INFO Channel 22/1 : 0[3] -> 1[5] via P2P/IPC
+ip-26-0-173-121:3987797:3988405 [3] NCCL INFO Channel 23/1 : 0[3] -> 1[5] via P2P/IPC
+ip-26-0-173-121:3987797:3988405 [3] NCCL INFO Channel 24/1 : 0[3] -> 1[5] via P2P/IPC
+ip-26-0-173-121:3987797:3988405 [3] NCCL INFO Channel 25/1 : 0[3] -> 1[5] via P2P/IPC
+ip-26-0-173-121:3987797:3988405 [3] NCCL INFO Channel 26/1 : 0[3] -> 1[5] via P2P/IPC
+ip-26-0-173-121:3987797:3988405 [3] NCCL INFO Channel 27/1 : 0[3] -> 1[5] via P2P/IPC
+ip-26-0-173-121:3987797:3988405 [3] NCCL INFO Channel 28/1 : 0[3] -> 1[5] via P2P/IPC
+ip-26-0-173-121:3987797:3988405 [3] NCCL INFO Channel 29/1 : 0[3] -> 1[5] via P2P/IPC
+ip-26-0-173-121:3987797:3988405 [3] NCCL INFO Channel 30/1 : 0[3] -> 1[5] via P2P/IPC
+ip-26-0-173-121:3987797:3988405 [3] NCCL INFO Channel 31/1 : 0[3] -> 1[5] via P2P/IPC
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+ip-26-0-173-121:3987800:3988428 [6] NCCL INFO Using network Libfabric
+ip-26-0-173-121:3987798:3988427 [4] NCCL INFO Using network Libfabric
+ip-26-0-173-121:3987800:3988428 [6] NCCL INFO comm 0xbc27cb0 rank 1 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0x64b21406864cd4fa - Init START
+ip-26-0-173-121:3987798:3988427 [4] NCCL INFO comm 0x13127330 rank 0 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0x64b21406864cd4fa - Init START
+ip-26-0-173-121:3987800:3988428 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-173-121:3987798:3988427 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-173-121:3987798:3988427 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-173-121:3987800:3988428 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-173-121:3987800:3988428 [6] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] -1/-1/-1->1->0 [3] -1/-1/-1->1->0 [4] -1/-1/-1->1->0 [5] -1/-1/-1->1->0 [6] 0/-1/-1->1->-1 [7] 0/-1/-1->1->-1 [8] 0/-1/-1->1->-1 [9] 0/-1/-1->1->-1 [10] 0/-1/-1->1->-1 [11] 0/-1/-1->1->-1 [12] -1/-1/-1->1->0 [13] -1/-1/-1->1->0 [14] -1/-1/-1->1->0 [15] -1/-1/-1->1->0 [16] -1/-1/-1->1->0 [17] -1/-1/-1->1->0 [18] 0/-1/-1->1->-1 [19] 0/-1/-1->1->-1 [20] 0/-1/-1->1->-1 [21] 0/-1/-1->1->-1 [22] 0/-1/-1->1->-1 [23] 0/-1/-1->1->-1
+ip-26-0-173-121:3987800:3988428 [6] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-173-121:3987798:3988427 [4] NCCL INFO Channel 00/24 :    0   1
+ip-26-0-173-121:3987798:3988427 [4] NCCL INFO Channel 01/24 :    0   1
+ip-26-0-173-121:3987798:3988427 [4] NCCL INFO Channel 02/24 :    0   1
+ip-26-0-173-121:3987798:3988427 [4] NCCL INFO Channel 03/24 :    0   1
+ip-26-0-173-121:3987798:3988427 [4] NCCL INFO Channel 04/24 :    0   1
+ip-26-0-173-121:3987798:3988427 [4] NCCL INFO Channel 05/24 :    0   1
+ip-26-0-173-121:3987798:3988427 [4] NCCL INFO Channel 06/24 :    0   1
+ip-26-0-173-121:3987798:3988427 [4] NCCL INFO Channel 07/24 :    0   1
+ip-26-0-173-121:3987798:3988427 [4] NCCL INFO Channel 08/24 :    0   1
+ip-26-0-173-121:3987798:3988427 [4] NCCL INFO Channel 09/24 :    0   1
+ip-26-0-173-121:3987798:3988427 [4] NCCL INFO Channel 10/24 :    0   1
+ip-26-0-173-121:3987798:3988427 [4] NCCL INFO Channel 11/24 :    0   1
+ip-26-0-173-121:3987798:3988427 [4] NCCL INFO Channel 12/24 :    0   1
+ip-26-0-173-121:3987798:3988427 [4] NCCL INFO Channel 13/24 :    0   1
+ip-26-0-173-121:3987798:3988427 [4] NCCL INFO Channel 14/24 :    0   1
+ip-26-0-173-121:3987798:3988427 [4] NCCL INFO Channel 15/24 :    0   1
+ip-26-0-173-121:3987798:3988427 [4] NCCL INFO Channel 16/24 :    0   1
+ip-26-0-173-121:3987798:3988427 [4] NCCL INFO Channel 17/24 :    0   1
+ip-26-0-173-121:3987798:3988427 [4] NCCL INFO Channel 18/24 :    0   1
+ip-26-0-173-121:3987798:3988427 [4] NCCL INFO Channel 19/24 :    0   1
+ip-26-0-173-121:3987798:3988427 [4] NCCL INFO Channel 20/24 :    0   1
+ip-26-0-173-121:3987798:3988427 [4] NCCL INFO Channel 21/24 :    0   1
+ip-26-0-173-121:3987798:3988427 [4] NCCL INFO Channel 22/24 :    0   1
+ip-26-0-173-121:3987798:3988427 [4] NCCL INFO Channel 23/24 :    0   1
+ip-26-0-173-121:3987798:3988427 [4] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] -1/-1/-1->0->1 [7] -1/-1/-1->0->1 [8] -1/-1/-1->0->1 [9] -1/-1/-1->0->1 [10] -1/-1/-1->0->1 [11] -1/-1/-1->0->1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] -1/-1/-1->0->1 [19] -1/-1/-1->0->1 [20] -1/-1/-1->0->1 [21] -1/-1/-1->0->1 [22] -1/-1/-1->0->1 [23] -1/-1/-1->0->1
+ip-26-0-173-121:3987798:3988427 [4] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-173-121:3987798:3988427 [4] NCCL INFO Channel 00/0 : 0[4] -> 1[6] via P2P/IPC
+ip-26-0-173-121:3987800:3988428 [6] NCCL INFO Channel 00/0 : 1[6] -> 0[4] via P2P/IPC
+ip-26-0-173-121:3987798:3988427 [4] NCCL INFO Channel 01/0 : 0[4] -> 1[6] via P2P/IPC
+ip-26-0-173-121:3987800:3988428 [6] NCCL INFO Channel 01/0 : 1[6] -> 0[4] via P2P/IPC
+ip-26-0-173-121:3987798:3988427 [4] NCCL INFO Channel 02/0 : 0[4] -> 1[6] via P2P/IPC
+ip-26-0-173-121:3987800:3988428 [6] NCCL INFO Channel 02/0 : 1[6] -> 0[4] via P2P/IPC
+ip-26-0-173-121:3987798:3988427 [4] NCCL INFO Channel 03/0 : 0[4] -> 1[6] via P2P/IPC
+ip-26-0-173-121:3987800:3988428 [6] NCCL INFO Channel 03/0 : 1[6] -> 0[4] via P2P/IPC
+ip-26-0-173-121:3987798:3988427 [4] NCCL INFO Channel 04/0 : 0[4] -> 1[6] via P2P/IPC
+ip-26-0-173-121:3987800:3988428 [6] NCCL INFO Channel 04/0 : 1[6] -> 0[4] via P2P/IPC
+ip-26-0-173-121:3987798:3988427 [4] NCCL INFO Channel 05/0 : 0[4] -> 1[6] via P2P/IPC
+ip-26-0-173-121:3987800:3988428 [6] NCCL INFO Channel 05/0 : 1[6] -> 0[4] via P2P/IPC
+ip-26-0-173-121:3987798:3988427 [4] NCCL INFO Channel 06/0 : 0[4] -> 1[6] via P2P/IPC
+ip-26-0-173-121:3987800:3988428 [6] NCCL INFO Channel 06/0 : 1[6] -> 0[4] via P2P/IPC
+ip-26-0-173-121:3987798:3988427 [4] NCCL INFO Channel 07/0 : 0[4] -> 1[6] via P2P/IPC
+ip-26-0-173-121:3987800:3988428 [6] NCCL INFO Channel 07/0 : 1[6] -> 0[4] via P2P/IPC
+ip-26-0-173-121:3987798:3988427 [4] NCCL INFO Channel 08/0 : 0[4] -> 1[6] via P2P/IPC
+ip-26-0-173-121:3987800:3988428 [6] NCCL INFO Channel 08/0 : 1[6] -> 0[4] via P2P/IPC
+ip-26-0-173-121:3987798:3988427 [4] NCCL INFO Channel 09/0 : 0[4] -> 1[6] via P2P/IPC
+ip-26-0-173-121:3987800:3988428 [6] NCCL INFO Channel 09/0 : 1[6] -> 0[4] via P2P/IPC
+ip-26-0-173-121:3987798:3988427 [4] NCCL INFO Channel 10/0 : 0[4] -> 1[6] via P2P/IPC
+ip-26-0-173-121:3987800:3988428 [6] NCCL INFO Channel 10/0 : 1[6] -> 0[4] via P2P/IPC
+ip-26-0-173-121:3987798:3988427 [4] NCCL INFO Channel 11/0 : 0[4] -> 1[6] via P2P/IPC
+ip-26-0-173-121:3987800:3988428 [6] NCCL INFO Channel 11/0 : 1[6] -> 0[4] via P2P/IPC
+ip-26-0-173-121:3987798:3988427 [4] NCCL INFO Channel 12/0 : 0[4] -> 1[6] via P2P/IPC
+ip-26-0-173-121:3987800:3988428 [6] NCCL INFO Channel 12/0 : 1[6] -> 0[4] via P2P/IPC
+ip-26-0-173-121:3987798:3988427 [4] NCCL INFO Channel 13/0 : 0[4] -> 1[6] via P2P/IPC
+ip-26-0-173-121:3987800:3988428 [6] NCCL INFO Channel 13/0 : 1[6] -> 0[4] via P2P/IPC
+ip-26-0-173-121:3987798:3988427 [4] NCCL INFO Channel 14/0 : 0[4] -> 1[6] via P2P/IPC
+ip-26-0-173-121:3987800:3988428 [6] NCCL INFO Channel 14/0 : 1[6] -> 0[4] via P2P/IPC
+ip-26-0-173-121:3987798:3988427 [4] NCCL INFO Channel 15/0 : 0[4] -> 1[6] via P2P/IPC
+ip-26-0-173-121:3987800:3988428 [6] NCCL INFO Channel 15/0 : 1[6] -> 0[4] via P2P/IPC
+ip-26-0-173-121:3987798:3988427 [4] NCCL INFO Channel 16/0 : 0[4] -> 1[6] via P2P/IPC
+ip-26-0-173-121:3987800:3988428 [6] NCCL INFO Channel 16/0 : 1[6] -> 0[4] via P2P/IPC
+ip-26-0-173-121:3987798:3988427 [4] NCCL INFO Channel 17/0 : 0[4] -> 1[6] via P2P/IPC
+ip-26-0-173-121:3987800:3988428 [6] NCCL INFO Channel 17/0 : 1[6] -> 0[4] via P2P/IPC
+ip-26-0-173-121:3987798:3988427 [4] NCCL INFO Channel 18/0 : 0[4] -> 1[6] via P2P/IPC
+ip-26-0-173-121:3987800:3988428 [6] NCCL INFO Channel 18/0 : 1[6] -> 0[4] via P2P/IPC
+ip-26-0-173-121:3987798:3988427 [4] NCCL INFO Channel 19/0 : 0[4] -> 1[6] via P2P/IPC
+ip-26-0-173-121:3987800:3988428 [6] NCCL INFO Channel 19/0 : 1[6] -> 0[4] via P2P/IPC
+ip-26-0-173-121:3987798:3988427 [4] NCCL INFO Channel 20/0 : 0[4] -> 1[6] via P2P/IPC
+ip-26-0-173-121:3987800:3988428 [6] NCCL INFO Channel 20/0 : 1[6] -> 0[4] via P2P/IPC
+ip-26-0-173-121:3987798:3988427 [4] NCCL INFO Channel 21/0 : 0[4] -> 1[6] via P2P/IPC
+ip-26-0-173-121:3987800:3988428 [6] NCCL INFO Channel 21/0 : 1[6] -> 0[4] via P2P/IPC
+ip-26-0-173-121:3987798:3988427 [4] NCCL INFO Channel 22/0 : 0[4] -> 1[6] via P2P/IPC
+ip-26-0-173-121:3987800:3988428 [6] NCCL INFO Channel 22/0 : 1[6] -> 0[4] via P2P/IPC
+ip-26-0-173-121:3987798:3988427 [4] NCCL INFO Channel 23/0 : 0[4] -> 1[6] via P2P/IPC
+ip-26-0-173-121:3987800:3988428 [6] NCCL INFO Channel 23/0 : 1[6] -> 0[4] via P2P/IPC
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+ip-26-0-173-121:3987799:3988432 [5] NCCL INFO Using network Libfabric
+ip-26-0-173-121:3987801:3988433 [7] NCCL INFO Using network Libfabric
+ip-26-0-173-121:3987801:3988433 [7] NCCL INFO comm 0xc016200 rank 1 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0x85947c2f8e82efb7 - Init START
+ip-26-0-173-121:3987799:3988432 [5] NCCL INFO comm 0x12724ca0 rank 0 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0x85947c2f8e82efb7 - Init START
+ip-26-0-173-121:3987801:3988433 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-173-121:3987799:3988432 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/184
+ip-26-0-173-121:3987800:3988428 [6] NCCL INFO Connected all rings
+ip-26-0-173-121:3987800:3988428 [6] NCCL INFO Connected all trees
+ip-26-0-173-121:3987800:3988428 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-173-121:3987800:3988428 [6] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-173-121:3987800:3988428 [6] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-173-121:3987798:3988427 [4] NCCL INFO Connected all rings
+ip-26-0-173-121:3987798:3988427 [4] NCCL INFO Connected all trees
+ip-26-0-173-121:3987798:3988427 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-173-121:3987798:3988427 [4] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-173-121:3987798:3988427 [4] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-173-121:3987800:3988428 [6] NCCL INFO comm 0xbc27cb0 rank 1 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0x64b21406864cd4fa - Init COMPLETE
+ip-26-0-173-121:3987798:3988427 [4] NCCL INFO comm 0x13127330 rank 0 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0x64b21406864cd4fa - Init COMPLETE
+ip-26-0-173-121:3987798:3988437 [4] NCCL INFO Channel 00/1 : 0[4] -> 1[6] via P2P/IPC
+ip-26-0-173-121:3987798:3988437 [4] NCCL INFO Channel 01/1 : 0[4] -> 1[6] via P2P/IPC
+ip-26-0-173-121:3987798:3988437 [4] NCCL INFO Channel 02/1 : 0[4] -> 1[6] via P2P/IPC
+ip-26-0-173-121:3987798:3988437 [4] NCCL INFO Channel 03/1 : 0[4] -> 1[6] via P2P/IPC
+ip-26-0-173-121:3987798:3988437 [4] NCCL INFO Channel 04/1 : 0[4] -> 1[6] via P2P/IPC
+ip-26-0-173-121:3987798:3988437 [4] NCCL INFO Channel 05/1 : 0[4] -> 1[6] via P2P/IPC
+ip-26-0-173-121:3987798:3988437 [4] NCCL INFO Channel 06/1 : 0[4] -> 1[6] via P2P/IPC
+ip-26-0-173-121:3987798:3988437 [4] NCCL INFO Channel 07/1 : 0[4] -> 1[6] via P2P/IPC
+ip-26-0-173-121:3987798:3988437 [4] NCCL INFO Channel 08/1 : 0[4] -> 1[6] via P2P/IPC
+ip-26-0-173-121:3987798:3988437 [4] NCCL INFO Channel 09/1 : 0[4] -> 1[6] via P2P/IPC
+ip-26-0-173-121:3987798:3988437 [4] NCCL INFO Channel 10/1 : 0[4] -> 1[6] via P2P/IPC
+ip-26-0-173-121:3987798:3988437 [4] NCCL INFO Channel 11/1 : 0[4] -> 1[6] via P2P/IPC
+ip-26-0-173-121:3987798:3988437 [4] NCCL INFO Channel 12/1 : 0[4] -> 1[6] via P2P/IPC
+ip-26-0-173-121:3987798:3988437 [4] NCCL INFO Channel 13/1 : 0[4] -> 1[6] via P2P/IPC
+ip-26-0-173-121:3987798:3988437 [4] NCCL INFO Channel 14/1 : 0[4] -> 1[6] via P2P/IPC
+ip-26-0-173-121:3987798:3988437 [4] NCCL INFO Channel 15/1 : 0[4] -> 1[6] via P2P/IPC
+ip-26-0-173-121:3987798:3988437 [4] NCCL INFO Channel 16/1 : 0[4] -> 1[6] via P2P/IPC
+ip-26-0-173-121:3987798:3988437 [4] NCCL INFO Channel 17/1 : 0[4] -> 1[6] via P2P/IPC
+ip-26-0-173-121:3987798:3988437 [4] NCCL INFO Channel 18/1 : 0[4] -> 1[6] via P2P/IPC
+ip-26-0-173-121:3987798:3988437 [4] NCCL INFO Channel 19/1 : 0[4] -> 1[6] via P2P/IPC
+ip-26-0-173-121:3987798:3988437 [4] NCCL INFO Channel 20/1 : 0[4] -> 1[6] via P2P/IPC
+ip-26-0-173-121:3987798:3988437 [4] NCCL INFO Channel 21/1 : 0[4] -> 1[6] via P2P/IPC
+ip-26-0-173-121:3987798:3988437 [4] NCCL INFO Channel 22/1 : 0[4] -> 1[6] via P2P/IPC
+ip-26-0-173-121:3987798:3988437 [4] NCCL INFO Channel 23/1 : 0[4] -> 1[6] via P2P/IPC
+ip-26-0-173-121:3987798:3988437 [4] NCCL INFO Channel 24/1 : 0[4] -> 1[6] via P2P/IPC
+ip-26-0-173-121:3987798:3988437 [4] NCCL INFO Channel 25/1 : 0[4] -> 1[6] via P2P/IPC
+ip-26-0-173-121:3987798:3988437 [4] NCCL INFO Channel 26/1 : 0[4] -> 1[6] via P2P/IPC
+ip-26-0-173-121:3987798:3988437 [4] NCCL INFO Channel 27/1 : 0[4] -> 1[6] via P2P/IPC
+ip-26-0-173-121:3987798:3988437 [4] NCCL INFO Channel 28/1 : 0[4] -> 1[6] via P2P/IPC
+ip-26-0-173-121:3987798:3988437 [4] NCCL INFO Channel 29/1 : 0[4] -> 1[6] via P2P/IPC
+ip-26-0-173-121:3987798:3988437 [4] NCCL INFO Channel 30/1 : 0[4] -> 1[6] via P2P/IPC
+ip-26-0-173-121:3987798:3988437 [4] NCCL INFO Channel 31/1 : 0[4] -> 1[6] via P2P/IPC
+ip-26-0-173-121:3987801:3988433 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-173-121:3987799:3988432 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-173-121:3987799:3988432 [5] NCCL INFO Channel 00/24 :    0   1
+ip-26-0-173-121:3987799:3988432 [5] NCCL INFO Channel 01/24 :    0   1
+ip-26-0-173-121:3987799:3988432 [5] NCCL INFO Channel 02/24 :    0   1
+ip-26-0-173-121:3987799:3988432 [5] NCCL INFO Channel 03/24 :    0   1
+ip-26-0-173-121:3987799:3988432 [5] NCCL INFO Channel 04/24 :    0   1
+ip-26-0-173-121:3987799:3988432 [5] NCCL INFO Channel 05/24 :    0   1
+ip-26-0-173-121:3987799:3988432 [5] NCCL INFO Channel 06/24 :    0   1
+ip-26-0-173-121:3987801:3988433 [7] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] -1/-1/-1->1->0 [3] -1/-1/-1->1->0 [4] -1/-1/-1->1->0 [5] -1/-1/-1->1->0 [6] 0/-1/-1->1->-1 [7] 0/-1/-1->1->-1 [8] 0/-1/-1->1->-1 [9] 0/-1/-1->1->-1 [10] 0/-1/-1->1->-1 [11] 0/-1/-1->1->-1 [12] -1/-1/-1->1->0 [13] -1/-1/-1->1->0 [14] -1/-1/-1->1->0 [15] -1/-1/-1->1->0 [16] -1/-1/-1->1->0 [17] -1/-1/-1->1->0 [18] 0/-1/-1->1->-1 [19] 0/-1/-1->1->-1 [20] 0/-1/-1->1->-1 [21] 0/-1/-1->1->-1 [22] 0/-1/-1->1->-1 [23] 0/-1/-1->1->-1
+ip-26-0-173-121:3987799:3988432 [5] NCCL INFO Channel 07/24 :    0   1
+ip-26-0-173-121:3987801:3988433 [7] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-173-121:3987799:3988432 [5] NCCL INFO Channel 08/24 :    0   1
+ip-26-0-173-121:3987799:3988432 [5] NCCL INFO Channel 09/24 :    0   1
+ip-26-0-173-121:3987799:3988432 [5] NCCL INFO Channel 10/24 :    0   1
+ip-26-0-173-121:3987799:3988432 [5] NCCL INFO Channel 11/24 :    0   1
+ip-26-0-173-121:3987799:3988432 [5] NCCL INFO Channel 12/24 :    0   1
+ip-26-0-173-121:3987799:3988432 [5] NCCL INFO Channel 13/24 :    0   1
+ip-26-0-173-121:3987799:3988432 [5] NCCL INFO Channel 14/24 :    0   1
+ip-26-0-173-121:3987799:3988432 [5] NCCL INFO Channel 15/24 :    0   1
+ip-26-0-173-121:3987799:3988432 [5] NCCL INFO Channel 16/24 :    0   1
+ip-26-0-173-121:3987799:3988432 [5] NCCL INFO Channel 17/24 :    0   1
+ip-26-0-173-121:3987799:3988432 [5] NCCL INFO Channel 18/24 :    0   1
+ip-26-0-173-121:3987799:3988432 [5] NCCL INFO Channel 19/24 :    0   1
+ip-26-0-173-121:3987799:3988432 [5] NCCL INFO Channel 20/24 :    0   1
+ip-26-0-173-121:3987799:3988432 [5] NCCL INFO Channel 21/24 :    0   1
+ip-26-0-173-121:3987799:3988432 [5] NCCL INFO Channel 22/24 :    0   1
+ip-26-0-173-121:3987799:3988432 [5] NCCL INFO Channel 23/24 :    0   1
+ip-26-0-173-121:3987799:3988432 [5] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] -1/-1/-1->0->1 [7] -1/-1/-1->0->1 [8] -1/-1/-1->0->1 [9] -1/-1/-1->0->1 [10] -1/-1/-1->0->1 [11] -1/-1/-1->0->1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] -1/-1/-1->0->1 [19] -1/-1/-1->0->1 [20] -1/-1/-1->0->1 [21] -1/-1/-1->0->1 [22] -1/-1/-1->0->1 [23] -1/-1/-1->0->1
+ip-26-0-173-121:3987799:3988432 [5] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-173-121:3987801:3988433 [7] NCCL INFO Channel 00/0 : 1[7] -> 0[5] via P2P/IPC
+ip-26-0-173-121:3987801:3988433 [7] NCCL INFO Channel 01/0 : 1[7] -> 0[5] via P2P/IPC
+ip-26-0-173-121:3987799:3988432 [5] NCCL INFO Channel 00/0 : 0[5] -> 1[7] via P2P/IPC
+ip-26-0-173-121:3987801:3988433 [7] NCCL INFO Channel 02/0 : 1[7] -> 0[5] via P2P/IPC
+ip-26-0-173-121:3987799:3988432 [5] NCCL INFO Channel 01/0 : 0[5] -> 1[7] via P2P/IPC
+ip-26-0-173-121:3987801:3988433 [7] NCCL INFO Channel 03/0 : 1[7] -> 0[5] via P2P/IPC
+ip-26-0-173-121:3987799:3988432 [5] NCCL INFO Channel 02/0 : 0[5] -> 1[7] via P2P/IPC
+ip-26-0-173-121:3987801:3988433 [7] NCCL INFO Channel 04/0 : 1[7] -> 0[5] via P2P/IPC
+ip-26-0-173-121:3987799:3988432 [5] NCCL INFO Channel 03/0 : 0[5] -> 1[7] via P2P/IPC
+ip-26-0-173-121:3987801:3988433 [7] NCCL INFO Channel 05/0 : 1[7] -> 0[5] via P2P/IPC
+ip-26-0-173-121:3987799:3988432 [5] NCCL INFO Channel 04/0 : 0[5] -> 1[7] via P2P/IPC
+ip-26-0-173-121:3987801:3988433 [7] NCCL INFO Channel 06/0 : 1[7] -> 0[5] via P2P/IPC
+ip-26-0-173-121:3987799:3988432 [5] NCCL INFO Channel 05/0 : 0[5] -> 1[7] via P2P/IPC
+ip-26-0-173-121:3987801:3988433 [7] NCCL INFO Channel 07/0 : 1[7] -> 0[5] via P2P/IPC
+ip-26-0-173-121:3987799:3988432 [5] NCCL INFO Channel 06/0 : 0[5] -> 1[7] via P2P/IPC
+ip-26-0-173-121:3987801:3988433 [7] NCCL INFO Channel 08/0 : 1[7] -> 0[5] via P2P/IPC
+ip-26-0-173-121:3987799:3988432 [5] NCCL INFO Channel 07/0 : 0[5] -> 1[7] via P2P/IPC
+ip-26-0-173-121:3987801:3988433 [7] NCCL INFO Channel 09/0 : 1[7] -> 0[5] via P2P/IPC
+ip-26-0-173-121:3987799:3988432 [5] NCCL INFO Channel 08/0 : 0[5] -> 1[7] via P2P/IPC
+ip-26-0-173-121:3987801:3988433 [7] NCCL INFO Channel 10/0 : 1[7] -> 0[5] via P2P/IPC
+ip-26-0-173-121:3987799:3988432 [5] NCCL INFO Channel 09/0 : 0[5] -> 1[7] via P2P/IPC
+ip-26-0-173-121:3987801:3988433 [7] NCCL INFO Channel 11/0 : 1[7] -> 0[5] via P2P/IPC
+ip-26-0-173-121:3987799:3988432 [5] NCCL INFO Channel 10/0 : 0[5] -> 1[7] via P2P/IPC
+ip-26-0-173-121:3987801:3988433 [7] NCCL INFO Channel 12/0 : 1[7] -> 0[5] via P2P/IPC
+ip-26-0-173-121:3987799:3988432 [5] NCCL INFO Channel 11/0 : 0[5] -> 1[7] via P2P/IPC
+ip-26-0-173-121:3987801:3988433 [7] NCCL INFO Channel 13/0 : 1[7] -> 0[5] via P2P/IPC
+ip-26-0-173-121:3987799:3988432 [5] NCCL INFO Channel 12/0 : 0[5] -> 1[7] via P2P/IPC
+ip-26-0-173-121:3987801:3988433 [7] NCCL INFO Channel 14/0 : 1[7] -> 0[5] via P2P/IPC
+ip-26-0-173-121:3987799:3988432 [5] NCCL INFO Channel 13/0 : 0[5] -> 1[7] via P2P/IPC
+ip-26-0-173-121:3987801:3988433 [7] NCCL INFO Channel 15/0 : 1[7] -> 0[5] via P2P/IPC
+ip-26-0-173-121:3987799:3988432 [5] NCCL INFO Channel 14/0 : 0[5] -> 1[7] via P2P/IPC
+ip-26-0-173-121:3987801:3988433 [7] NCCL INFO Channel 16/0 : 1[7] -> 0[5] via P2P/IPC
+ip-26-0-173-121:3987799:3988432 [5] NCCL INFO Channel 15/0 : 0[5] -> 1[7] via P2P/IPC
+ip-26-0-173-121:3987801:3988433 [7] NCCL INFO Channel 17/0 : 1[7] -> 0[5] via P2P/IPC
+ip-26-0-173-121:3987799:3988432 [5] NCCL INFO Channel 16/0 : 0[5] -> 1[7] via P2P/IPC
+ip-26-0-173-121:3987801:3988433 [7] NCCL INFO Channel 18/0 : 1[7] -> 0[5] via P2P/IPC
+ip-26-0-173-121:3987799:3988432 [5] NCCL INFO Channel 17/0 : 0[5] -> 1[7] via P2P/IPC
+ip-26-0-173-121:3987801:3988433 [7] NCCL INFO Channel 19/0 : 1[7] -> 0[5] via P2P/IPC
+ip-26-0-173-121:3987799:3988432 [5] NCCL INFO Channel 18/0 : 0[5] -> 1[7] via P2P/IPC
+ip-26-0-173-121:3987801:3988433 [7] NCCL INFO Channel 20/0 : 1[7] -> 0[5] via P2P/IPC
+ip-26-0-173-121:3987799:3988432 [5] NCCL INFO Channel 19/0 : 0[5] -> 1[7] via P2P/IPC
+ip-26-0-173-121:3987801:3988433 [7] NCCL INFO Channel 21/0 : 1[7] -> 0[5] via P2P/IPC
+ip-26-0-173-121:3987799:3988432 [5] NCCL INFO Channel 20/0 : 0[5] -> 1[7] via P2P/IPC
+ip-26-0-173-121:3987801:3988433 [7] NCCL INFO Channel 22/0 : 1[7] -> 0[5] via P2P/IPC
+ip-26-0-173-121:3987799:3988432 [5] NCCL INFO Channel 21/0 : 0[5] -> 1[7] via P2P/IPC
+ip-26-0-173-121:3987801:3988433 [7] NCCL INFO Channel 23/0 : 1[7] -> 0[5] via P2P/IPC
+ip-26-0-173-121:3987799:3988432 [5] NCCL INFO Channel 22/0 : 0[5] -> 1[7] via P2P/IPC
+ip-26-0-173-121:3987799:3988432 [5] NCCL INFO Channel 23/0 : 0[5] -> 1[7] via P2P/IPC
+ip-26-0-173-121:3987801:3988433 [7] NCCL INFO Connected all rings
+ip-26-0-173-121:3987801:3988433 [7] NCCL INFO Connected all trees
+ip-26-0-173-121:3987801:3988433 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-173-121:3987801:3988433 [7] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-173-121:3987801:3988433 [7] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-173-121:3987799:3988432 [5] NCCL INFO Connected all rings
+ip-26-0-173-121:3987799:3988432 [5] NCCL INFO Connected all trees
+ip-26-0-173-121:3987799:3988432 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-173-121:3987799:3988432 [5] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-173-121:3987799:3988432 [5] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-173-121:3987801:3988433 [7] NCCL INFO comm 0xc016200 rank 1 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0x85947c2f8e82efb7 - Init COMPLETE
+ip-26-0-173-121:3987799:3988432 [5] NCCL INFO comm 0x12724ca0 rank 0 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0x85947c2f8e82efb7 - Init COMPLETE
+ip-26-0-173-121:3987799:3988443 [5] NCCL INFO Channel 00/1 : 0[5] -> 1[7] via P2P/IPC
+ip-26-0-173-121:3987799:3988443 [5] NCCL INFO Channel 01/1 : 0[5] -> 1[7] via P2P/IPC
+ip-26-0-173-121:3987799:3988443 [5] NCCL INFO Channel 02/1 : 0[5] -> 1[7] via P2P/IPC
+ip-26-0-173-121:3987799:3988443 [5] NCCL INFO Channel 03/1 : 0[5] -> 1[7] via P2P/IPC
+ip-26-0-173-121:3987799:3988443 [5] NCCL INFO Channel 04/1 : 0[5] -> 1[7] via P2P/IPC
+ip-26-0-173-121:3987799:3988443 [5] NCCL INFO Channel 05/1 : 0[5] -> 1[7] via P2P/IPC
+ip-26-0-173-121:3987799:3988443 [5] NCCL INFO Channel 06/1 : 0[5] -> 1[7] via P2P/IPC
+ip-26-0-173-121:3987799:3988443 [5] NCCL INFO Channel 07/1 : 0[5] -> 1[7] via P2P/IPC
+ip-26-0-173-121:3987799:3988443 [5] NCCL INFO Channel 08/1 : 0[5] -> 1[7] via P2P/IPC
+ip-26-0-173-121:3987799:3988443 [5] NCCL INFO Channel 09/1 : 0[5] -> 1[7] via P2P/IPC
+ip-26-0-173-121:3987799:3988443 [5] NCCL INFO Channel 10/1 : 0[5] -> 1[7] via P2P/IPC
+ip-26-0-173-121:3987799:3988443 [5] NCCL INFO Channel 11/1 : 0[5] -> 1[7] via P2P/IPC
+ip-26-0-173-121:3987799:3988443 [5] NCCL INFO Channel 12/1 : 0[5] -> 1[7] via P2P/IPC
+ip-26-0-173-121:3987799:3988443 [5] NCCL INFO Channel 13/1 : 0[5] -> 1[7] via P2P/IPC
+ip-26-0-173-121:3987799:3988443 [5] NCCL INFO Channel 14/1 : 0[5] -> 1[7] via P2P/IPC
+ip-26-0-173-121:3987799:3988443 [5] NCCL INFO Channel 15/1 : 0[5] -> 1[7] via P2P/IPC
+ip-26-0-173-121:3987799:3988443 [5] NCCL INFO Channel 16/1 : 0[5] -> 1[7] via P2P/IPC
+ip-26-0-173-121:3987799:3988443 [5] NCCL INFO Channel 17/1 : 0[5] -> 1[7] via P2P/IPC
+ip-26-0-173-121:3987799:3988443 [5] NCCL INFO Channel 18/1 : 0[5] -> 1[7] via P2P/IPC
+ip-26-0-173-121:3987799:3988443 [5] NCCL INFO Channel 19/1 : 0[5] -> 1[7] via P2P/IPC
+ip-26-0-173-121:3987799:3988443 [5] NCCL INFO Channel 20/1 : 0[5] -> 1[7] via P2P/IPC
+ip-26-0-173-121:3987799:3988443 [5] NCCL INFO Channel 21/1 : 0[5] -> 1[7] via P2P/IPC
+ip-26-0-173-121:3987799:3988443 [5] NCCL INFO Channel 22/1 : 0[5] -> 1[7] via P2P/IPC
+ip-26-0-173-121:3987799:3988443 [5] NCCL INFO Channel 23/1 : 0[5] -> 1[7] via P2P/IPC
+ip-26-0-173-121:3987799:3988443 [5] NCCL INFO Channel 24/1 : 0[5] -> 1[7] via P2P/IPC
+ip-26-0-173-121:3987799:3988443 [5] NCCL INFO Channel 25/1 : 0[5] -> 1[7] via P2P/IPC
+ip-26-0-173-121:3987799:3988443 [5] NCCL INFO Channel 26/1 : 0[5] -> 1[7] via P2P/IPC
+ip-26-0-173-121:3987799:3988443 [5] NCCL INFO Channel 27/1 : 0[5] -> 1[7] via P2P/IPC
+ip-26-0-173-121:3987799:3988443 [5] NCCL INFO Channel 28/1 : 0[5] -> 1[7] via P2P/IPC
+ip-26-0-173-121:3987799:3988443 [5] NCCL INFO Channel 29/1 : 0[5] -> 1[7] via P2P/IPC
+ip-26-0-173-121:3987799:3988443 [5] NCCL INFO Channel 30/1 : 0[5] -> 1[7] via P2P/IPC
+ip-26-0-173-121:3987799:3988443 [5] NCCL INFO Channel 31/1 : 0[5] -> 1[7] via P2P/IPC
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+ip-26-0-173-121:3987800:3988486 [6] NCCL INFO Channel 00/1 : 1[6] -> 0[4] via P2P/IPC
+ip-26-0-173-121:3987800:3988486 [6] NCCL INFO Channel 01/1 : 1[6] -> 0[4] via P2P/IPC
+ip-26-0-173-121:3987800:3988486 [6] NCCL INFO Channel 02/1 : 1[6] -> 0[4] via P2P/IPC
+ip-26-0-173-121:3987800:3988486 [6] NCCL INFO Channel 03/1 : 1[6] -> 0[4] via P2P/IPC
+ip-26-0-173-121:3987800:3988486 [6] NCCL INFO Channel 04/1 : 1[6] -> 0[4] via P2P/IPC
+ip-26-0-173-121:3987800:3988486 [6] NCCL INFO Channel 05/1 : 1[6] -> 0[4] via P2P/IPC
+ip-26-0-173-121:3987800:3988486 [6] NCCL INFO Channel 06/1 : 1[6] -> 0[4] via P2P/IPC
+ip-26-0-173-121:3987800:3988486 [6] NCCL INFO Channel 07/1 : 1[6] -> 0[4] via P2P/IPC
+ip-26-0-173-121:3987800:3988486 [6] NCCL INFO Channel 08/1 : 1[6] -> 0[4] via P2P/IPC
+ip-26-0-173-121:3987800:3988486 [6] NCCL INFO Channel 09/1 : 1[6] -> 0[4] via P2P/IPC
+ip-26-0-173-121:3987800:3988486 [6] NCCL INFO Channel 10/1 : 1[6] -> 0[4] via P2P/IPC
+ip-26-0-173-121:3987800:3988486 [6] NCCL INFO Channel 11/1 : 1[6] -> 0[4] via P2P/IPC
+ip-26-0-173-121:3987800:3988486 [6] NCCL INFO Channel 12/1 : 1[6] -> 0[4] via P2P/IPC
+ip-26-0-173-121:3987800:3988486 [6] NCCL INFO Channel 13/1 : 1[6] -> 0[4] via P2P/IPC
+ip-26-0-173-121:3987800:3988486 [6] NCCL INFO Channel 14/1 : 1[6] -> 0[4] via P2P/IPC
+ip-26-0-173-121:3987800:3988486 [6] NCCL INFO Channel 15/1 : 1[6] -> 0[4] via P2P/IPC
+ip-26-0-173-121:3987800:3988486 [6] NCCL INFO Channel 16/1 : 1[6] -> 0[4] via P2P/IPC
+ip-26-0-173-121:3987800:3988486 [6] NCCL INFO Channel 17/1 : 1[6] -> 0[4] via P2P/IPC
+ip-26-0-173-121:3987800:3988486 [6] NCCL INFO Channel 18/1 : 1[6] -> 0[4] via P2P/IPC
+ip-26-0-173-121:3987800:3988486 [6] NCCL INFO Channel 19/1 : 1[6] -> 0[4] via P2P/IPC
+ip-26-0-173-121:3987800:3988486 [6] NCCL INFO Channel 20/1 : 1[6] -> 0[4] via P2P/IPC
+ip-26-0-173-121:3987800:3988486 [6] NCCL INFO Channel 21/1 : 1[6] -> 0[4] via P2P/IPC
+ip-26-0-173-121:3987800:3988486 [6] NCCL INFO Channel 22/1 : 1[6] -> 0[4] via P2P/IPC
+ip-26-0-173-121:3987800:3988486 [6] NCCL INFO Channel 23/1 : 1[6] -> 0[4] via P2P/IPC
+ip-26-0-173-121:3987800:3988486 [6] NCCL INFO Channel 24/1 : 1[6] -> 0[4] via P2P/IPC
+ip-26-0-173-121:3987800:3988486 [6] NCCL INFO Channel 25/1 : 1[6] -> 0[4] via P2P/IPC
+ip-26-0-173-121:3987800:3988486 [6] NCCL INFO Channel 26/1 : 1[6] -> 0[4] via P2P/IPC
+ip-26-0-173-121:3987800:3988486 [6] NCCL INFO Channel 27/1 : 1[6] -> 0[4] via P2P/IPC
+ip-26-0-173-121:3987800:3988486 [6] NCCL INFO Channel 28/1 : 1[6] -> 0[4] via P2P/IPC
+ip-26-0-173-121:3987800:3988486 [6] NCCL INFO Channel 29/1 : 1[6] -> 0[4] via P2P/IPC
+ip-26-0-173-121:3987800:3988486 [6] NCCL INFO Channel 30/1 : 1[6] -> 0[4] via P2P/IPC
+ip-26-0-173-121:3987800:3988486 [6] NCCL INFO Channel 31/1 : 1[6] -> 0[4] via P2P/IPC
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+ip-26-0-173-121:3987801:3988488 [7] NCCL INFO Channel 00/1 : 1[7] -> 0[5] via P2P/IPC
+ip-26-0-173-121:3987801:3988488 [7] NCCL INFO Channel 01/1 : 1[7] -> 0[5] via P2P/IPC
+ip-26-0-173-121:3987801:3988488 [7] NCCL INFO Channel 02/1 : 1[7] -> 0[5] via P2P/IPC
+ip-26-0-173-121:3987801:3988488 [7] NCCL INFO Channel 03/1 : 1[7] -> 0[5] via P2P/IPC
+ip-26-0-173-121:3987801:3988488 [7] NCCL INFO Channel 04/1 : 1[7] -> 0[5] via P2P/IPC
+ip-26-0-173-121:3987801:3988488 [7] NCCL INFO Channel 05/1 : 1[7] -> 0[5] via P2P/IPC
+ip-26-0-173-121:3987801:3988488 [7] NCCL INFO Channel 06/1 : 1[7] -> 0[5] via P2P/IPC
+ip-26-0-173-121:3987801:3988488 [7] NCCL INFO Channel 07/1 : 1[7] -> 0[5] via P2P/IPC
+ip-26-0-173-121:3987801:3988488 [7] NCCL INFO Channel 08/1 : 1[7] -> 0[5] via P2P/IPC
+ip-26-0-173-121:3987801:3988488 [7] NCCL INFO Channel 09/1 : 1[7] -> 0[5] via P2P/IPC
+ip-26-0-173-121:3987801:3988488 [7] NCCL INFO Channel 10/1 : 1[7] -> 0[5] via P2P/IPC
+ip-26-0-173-121:3987801:3988488 [7] NCCL INFO Channel 11/1 : 1[7] -> 0[5] via P2P/IPC
+ip-26-0-173-121:3987801:3988488 [7] NCCL INFO Channel 12/1 : 1[7] -> 0[5] via P2P/IPC
+ip-26-0-173-121:3987801:3988488 [7] NCCL INFO Channel 13/1 : 1[7] -> 0[5] via P2P/IPC
+ip-26-0-173-121:3987801:3988488 [7] NCCL INFO Channel 14/1 : 1[7] -> 0[5] via P2P/IPC
+ip-26-0-173-121:3987801:3988488 [7] NCCL INFO Channel 15/1 : 1[7] -> 0[5] via P2P/IPC
+ip-26-0-173-121:3987801:3988488 [7] NCCL INFO Channel 16/1 : 1[7] -> 0[5] via P2P/IPC
+ip-26-0-173-121:3987801:3988488 [7] NCCL INFO Channel 17/1 : 1[7] -> 0[5] via P2P/IPC
+ip-26-0-173-121:3987801:3988488 [7] NCCL INFO Channel 18/1 : 1[7] -> 0[5] via P2P/IPC
+ip-26-0-173-121:3987801:3988488 [7] NCCL INFO Channel 19/1 : 1[7] -> 0[5] via P2P/IPC
+ip-26-0-173-121:3987801:3988488 [7] NCCL INFO Channel 20/1 : 1[7] -> 0[5] via P2P/IPC
+ip-26-0-173-121:3987801:3988488 [7] NCCL INFO Channel 21/1 : 1[7] -> 0[5] via P2P/IPC
+ip-26-0-173-121:3987801:3988488 [7] NCCL INFO Channel 22/1 : 1[7] -> 0[5] via P2P/IPC
+ip-26-0-173-121:3987801:3988488 [7] NCCL INFO Channel 23/1 : 1[7] -> 0[5] via P2P/IPC
+ip-26-0-173-121:3987801:3988488 [7] NCCL INFO Channel 24/1 : 1[7] -> 0[5] via P2P/IPC
+ip-26-0-173-121:3987801:3988488 [7] NCCL INFO Channel 25/1 : 1[7] -> 0[5] via P2P/IPC
+ip-26-0-173-121:3987801:3988488 [7] NCCL INFO Channel 26/1 : 1[7] -> 0[5] via P2P/IPC
+ip-26-0-173-121:3987801:3988488 [7] NCCL INFO Channel 27/1 : 1[7] -> 0[5] via P2P/IPC
+ip-26-0-173-121:3987801:3988488 [7] NCCL INFO Channel 28/1 : 1[7] -> 0[5] via P2P/IPC
+ip-26-0-173-121:3987801:3988488 [7] NCCL INFO Channel 29/1 : 1[7] -> 0[5] via P2P/IPC
+ip-26-0-173-121:3987801:3988488 [7] NCCL INFO Channel 30/1 : 1[7] -> 0[5] via P2P/IPC
+ip-26-0-173-121:3987801:3988488 [7] NCCL INFO Channel 31/1 : 1[7] -> 0[5] via P2P/IPC
+ip-26-0-173-121:3987798:3988489 [4] NCCL INFO Channel 00/1 : 1[4] -> 0[2] via P2P/IPC
+ip-26-0-173-121:3987798:3988489 [4] NCCL INFO Channel 01/1 : 1[4] -> 0[2] via P2P/IPC
+ip-26-0-173-121:3987798:3988489 [4] NCCL INFO Channel 02/1 : 1[4] -> 0[2] via P2P/IPC
+ip-26-0-173-121:3987798:3988489 [4] NCCL INFO Channel 03/1 : 1[4] -> 0[2] via P2P/IPC
+ip-26-0-173-121:3987798:3988489 [4] NCCL INFO Channel 04/1 : 1[4] -> 0[2] via P2P/IPC
+ip-26-0-173-121:3987798:3988489 [4] NCCL INFO Channel 05/1 : 1[4] -> 0[2] via P2P/IPC
+ip-26-0-173-121:3987798:3988489 [4] NCCL INFO Channel 06/1 : 1[4] -> 0[2] via P2P/IPC
+ip-26-0-173-121:3987798:3988489 [4] NCCL INFO Channel 07/1 : 1[4] -> 0[2] via P2P/IPC
+ip-26-0-173-121:3987798:3988489 [4] NCCL INFO Channel 08/1 : 1[4] -> 0[2] via P2P/IPC
+ip-26-0-173-121:3987798:3988489 [4] NCCL INFO Channel 09/1 : 1[4] -> 0[2] via P2P/IPC
+ip-26-0-173-121:3987798:3988489 [4] NCCL INFO Channel 10/1 : 1[4] -> 0[2] via P2P/IPC
+ip-26-0-173-121:3987798:3988489 [4] NCCL INFO Channel 11/1 : 1[4] -> 0[2] via P2P/IPC
+ip-26-0-173-121:3987798:3988489 [4] NCCL INFO Channel 12/1 : 1[4] -> 0[2] via P2P/IPC
+ip-26-0-173-121:3987798:3988489 [4] NCCL INFO Channel 13/1 : 1[4] -> 0[2] via P2P/IPC
+ip-26-0-173-121:3987798:3988489 [4] NCCL INFO Channel 14/1 : 1[4] -> 0[2] via P2P/IPC
+ip-26-0-173-121:3987798:3988489 [4] NCCL INFO Channel 15/1 : 1[4] -> 0[2] via P2P/IPC
+ip-26-0-173-121:3987798:3988489 [4] NCCL INFO Channel 16/1 : 1[4] -> 0[2] via P2P/IPC
+ip-26-0-173-121:3987798:3988489 [4] NCCL INFO Channel 17/1 : 1[4] -> 0[2] via P2P/IPC
+ip-26-0-173-121:3987798:3988489 [4] NCCL INFO Channel 18/1 : 1[4] -> 0[2] via P2P/IPC
+ip-26-0-173-121:3987798:3988489 [4] NCCL INFO Channel 19/1 : 1[4] -> 0[2] via P2P/IPC
+ip-26-0-173-121:3987798:3988489 [4] NCCL INFO Channel 20/1 : 1[4] -> 0[2] via P2P/IPC
+ip-26-0-173-121:3987798:3988489 [4] NCCL INFO Channel 21/1 : 1[4] -> 0[2] via P2P/IPC
+ip-26-0-173-121:3987798:3988489 [4] NCCL INFO Channel 22/1 : 1[4] -> 0[2] via P2P/IPC
+ip-26-0-173-121:3987798:3988489 [4] NCCL INFO Channel 23/1 : 1[4] -> 0[2] via P2P/IPC
+ip-26-0-173-121:3987798:3988489 [4] NCCL INFO Channel 24/1 : 1[4] -> 0[2] via P2P/IPC
+ip-26-0-173-121:3987798:3988489 [4] NCCL INFO Channel 25/1 : 1[4] -> 0[2] via P2P/IPC
+ip-26-0-173-121:3987798:3988489 [4] NCCL INFO Channel 26/1 : 1[4] -> 0[2] via P2P/IPC
+ip-26-0-173-121:3987798:3988489 [4] NCCL INFO Channel 27/1 : 1[4] -> 0[2] via P2P/IPC
+ip-26-0-173-121:3987798:3988489 [4] NCCL INFO Channel 28/1 : 1[4] -> 0[2] via P2P/IPC
+ip-26-0-173-121:3987798:3988489 [4] NCCL INFO Channel 29/1 : 1[4] -> 0[2] via P2P/IPC
+ip-26-0-173-121:3987798:3988489 [4] NCCL INFO Channel 30/1 : 1[4] -> 0[2] via P2P/IPC
+ip-26-0-173-121:3987798:3988489 [4] NCCL INFO Channel 31/1 : 1[4] -> 0[2] via P2P/IPC
+ip-26-0-173-121:3987799:3988491 [5] NCCL INFO Channel 00/1 : 1[5] -> 0[3] via P2P/IPC
+ip-26-0-173-121:3987799:3988491 [5] NCCL INFO Channel 01/1 : 1[5] -> 0[3] via P2P/IPC
+ip-26-0-173-121:3987799:3988491 [5] NCCL INFO Channel 02/1 : 1[5] -> 0[3] via P2P/IPC
+ip-26-0-173-121:3987799:3988491 [5] NCCL INFO Channel 03/1 : 1[5] -> 0[3] via P2P/IPC
+ip-26-0-173-121:3987799:3988491 [5] NCCL INFO Channel 04/1 : 1[5] -> 0[3] via P2P/IPC
+ip-26-0-173-121:3987799:3988491 [5] NCCL INFO Channel 05/1 : 1[5] -> 0[3] via P2P/IPC
+ip-26-0-173-121:3987799:3988491 [5] NCCL INFO Channel 06/1 : 1[5] -> 0[3] via P2P/IPC
+ip-26-0-173-121:3987799:3988491 [5] NCCL INFO Channel 07/1 : 1[5] -> 0[3] via P2P/IPC
+ip-26-0-173-121:3987799:3988491 [5] NCCL INFO Channel 08/1 : 1[5] -> 0[3] via P2P/IPC
+ip-26-0-173-121:3987799:3988491 [5] NCCL INFO Channel 09/1 : 1[5] -> 0[3] via P2P/IPC
+ip-26-0-173-121:3987799:3988491 [5] NCCL INFO Channel 10/1 : 1[5] -> 0[3] via P2P/IPC
+ip-26-0-173-121:3987799:3988491 [5] NCCL INFO Channel 11/1 : 1[5] -> 0[3] via P2P/IPC
+ip-26-0-173-121:3987799:3988491 [5] NCCL INFO Channel 12/1 : 1[5] -> 0[3] via P2P/IPC
+ip-26-0-173-121:3987799:3988491 [5] NCCL INFO Channel 13/1 : 1[5] -> 0[3] via P2P/IPC
+ip-26-0-173-121:3987799:3988491 [5] NCCL INFO Channel 14/1 : 1[5] -> 0[3] via P2P/IPC
+ip-26-0-173-121:3987799:3988491 [5] NCCL INFO Channel 15/1 : 1[5] -> 0[3] via P2P/IPC
+ip-26-0-173-121:3987799:3988491 [5] NCCL INFO Channel 16/1 : 1[5] -> 0[3] via P2P/IPC
+ip-26-0-173-121:3987799:3988491 [5] NCCL INFO Channel 17/1 : 1[5] -> 0[3] via P2P/IPC
+ip-26-0-173-121:3987799:3988491 [5] NCCL INFO Channel 18/1 : 1[5] -> 0[3] via P2P/IPC
+ip-26-0-173-121:3987799:3988491 [5] NCCL INFO Channel 19/1 : 1[5] -> 0[3] via P2P/IPC
+ip-26-0-173-121:3987799:3988491 [5] NCCL INFO Channel 20/1 : 1[5] -> 0[3] via P2P/IPC
+ip-26-0-173-121:3987799:3988491 [5] NCCL INFO Channel 21/1 : 1[5] -> 0[3] via P2P/IPC
+ip-26-0-173-121:3987799:3988491 [5] NCCL INFO Channel 22/1 : 1[5] -> 0[3] via P2P/IPC
+ip-26-0-173-121:3987799:3988491 [5] NCCL INFO Channel 23/1 : 1[5] -> 0[3] via P2P/IPC
+ip-26-0-173-121:3987799:3988491 [5] NCCL INFO Channel 24/1 : 1[5] -> 0[3] via P2P/IPC
+ip-26-0-173-121:3987799:3988491 [5] NCCL INFO Channel 25/1 : 1[5] -> 0[3] via P2P/IPC
+ip-26-0-173-121:3987799:3988491 [5] NCCL INFO Channel 26/1 : 1[5] -> 0[3] via P2P/IPC
+ip-26-0-173-121:3987799:3988491 [5] NCCL INFO Channel 27/1 : 1[5] -> 0[3] via P2P/IPC
+ip-26-0-173-121:3987799:3988491 [5] NCCL INFO Channel 28/1 : 1[5] -> 0[3] via P2P/IPC
+ip-26-0-173-121:3987799:3988491 [5] NCCL INFO Channel 29/1 : 1[5] -> 0[3] via P2P/IPC
+ip-26-0-173-121:3987799:3988491 [5] NCCL INFO Channel 30/1 : 1[5] -> 0[3] via P2P/IPC
+ip-26-0-173-121:3987799:3988491 [5] NCCL INFO Channel 31/1 : 1[5] -> 0[3] via P2P/IPC
+ip-26-0-173-121:3987796:3988494 [2] NCCL INFO Channel 00/1 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987796:3988494 [2] NCCL INFO Channel 01/1 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987796:3988494 [2] NCCL INFO Channel 02/1 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987796:3988494 [2] NCCL INFO Channel 03/1 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987796:3988494 [2] NCCL INFO Channel 04/1 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987796:3988494 [2] NCCL INFO Channel 05/1 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987796:3988494 [2] NCCL INFO Channel 06/1 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987796:3988494 [2] NCCL INFO Channel 07/1 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987796:3988494 [2] NCCL INFO Channel 08/1 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987796:3988494 [2] NCCL INFO Channel 09/1 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987796:3988494 [2] NCCL INFO Channel 10/1 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987796:3988494 [2] NCCL INFO Channel 11/1 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987796:3988494 [2] NCCL INFO Channel 12/1 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987796:3988494 [2] NCCL INFO Channel 13/1 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987796:3988494 [2] NCCL INFO Channel 14/1 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987796:3988494 [2] NCCL INFO Channel 15/1 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987796:3988494 [2] NCCL INFO Channel 16/1 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987796:3988494 [2] NCCL INFO Channel 17/1 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987796:3988494 [2] NCCL INFO Channel 18/1 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987796:3988494 [2] NCCL INFO Channel 19/1 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987796:3988494 [2] NCCL INFO Channel 20/1 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987796:3988494 [2] NCCL INFO Channel 21/1 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987796:3988494 [2] NCCL INFO Channel 22/1 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987796:3988494 [2] NCCL INFO Channel 23/1 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987796:3988494 [2] NCCL INFO Channel 24/1 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987796:3988494 [2] NCCL INFO Channel 25/1 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987796:3988494 [2] NCCL INFO Channel 26/1 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987796:3988494 [2] NCCL INFO Channel 27/1 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987796:3988494 [2] NCCL INFO Channel 28/1 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987796:3988494 [2] NCCL INFO Channel 29/1 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987796:3988494 [2] NCCL INFO Channel 30/1 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987796:3988494 [2] NCCL INFO Channel 31/1 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-173-121:3987797:3988496 [3] NCCL INFO Channel 00/1 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-173-121:3987797:3988496 [3] NCCL INFO Channel 01/1 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-173-121:3987797:3988496 [3] NCCL INFO Channel 02/1 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-173-121:3987797:3988496 [3] NCCL INFO Channel 03/1 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-173-121:3987797:3988496 [3] NCCL INFO Channel 04/1 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-173-121:3987797:3988496 [3] NCCL INFO Channel 05/1 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-173-121:3987797:3988496 [3] NCCL INFO Channel 06/1 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-173-121:3987797:3988496 [3] NCCL INFO Channel 07/1 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-173-121:3987797:3988496 [3] NCCL INFO Channel 08/1 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-173-121:3987797:3988496 [3] NCCL INFO Channel 09/1 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-173-121:3987797:3988496 [3] NCCL INFO Channel 10/1 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-173-121:3987797:3988496 [3] NCCL INFO Channel 11/1 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-173-121:3987797:3988496 [3] NCCL INFO Channel 12/1 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-173-121:3987797:3988496 [3] NCCL INFO Channel 13/1 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-173-121:3987797:3988496 [3] NCCL INFO Channel 14/1 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-173-121:3987797:3988496 [3] NCCL INFO Channel 15/1 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-173-121:3987797:3988496 [3] NCCL INFO Channel 16/1 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-173-121:3987797:3988496 [3] NCCL INFO Channel 17/1 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-173-121:3987797:3988496 [3] NCCL INFO Channel 18/1 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-173-121:3987797:3988496 [3] NCCL INFO Channel 19/1 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-173-121:3987797:3988496 [3] NCCL INFO Channel 20/1 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-173-121:3987797:3988496 [3] NCCL INFO Channel 21/1 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-173-121:3987797:3988496 [3] NCCL INFO Channel 22/1 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-173-121:3987797:3988496 [3] NCCL INFO Channel 23/1 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-173-121:3987797:3988496 [3] NCCL INFO Channel 24/1 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-173-121:3987797:3988496 [3] NCCL INFO Channel 25/1 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-173-121:3987797:3988496 [3] NCCL INFO Channel 26/1 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-173-121:3987797:3988496 [3] NCCL INFO Channel 27/1 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-173-121:3987797:3988496 [3] NCCL INFO Channel 28/1 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-173-121:3987797:3988496 [3] NCCL INFO Channel 29/1 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-173-121:3987797:3988496 [3] NCCL INFO Channel 30/1 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-173-121:3987797:3988496 [3] NCCL INFO Channel 31/1 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-173-121:3987794:3988497 [0] NCCL INFO Channel 02/1 : 1[0] -> 0[6] [send] via NET/Libfabric/0/GDRDMA/Shared
+ip-26-0-173-121:3987794:3988497 [0] NCCL INFO Channel 03/1 : 1[0] -> 0[6] [send] via NET/Libfabric/0/GDRDMA/Shared
+ip-26-0-173-121:3987795:3988499 [1] NCCL INFO Channel 02/1 : 1[1] -> 0[7] [send] via NET/Libfabric/1/GDRDMA/Shared
+ip-26-0-173-121:3987795:3988499 [1] NCCL INFO Channel 03/1 : 1[1] -> 0[7] [send] via NET/Libfabric/1/GDRDMA/Shared
+ip-26-0-172-142:789261:789958 [6] NCCL INFO Channel 00/1 : 1[6] -> 0[4] via P2P/IPC
+ip-26-0-172-142:789261:789958 [6] NCCL INFO Channel 01/1 : 1[6] -> 0[4] via P2P/IPC
+ip-26-0-172-142:789261:789958 [6] NCCL INFO Channel 02/1 : 1[6] -> 0[4] via P2P/IPC
+ip-26-0-172-142:789261:789958 [6] NCCL INFO Channel 03/1 : 1[6] -> 0[4] via P2P/IPC
+ip-26-0-172-142:789261:789958 [6] NCCL INFO Channel 04/1 : 1[6] -> 0[4] via P2P/IPC
+ip-26-0-172-142:789261:789958 [6] NCCL INFO Channel 05/1 : 1[6] -> 0[4] via P2P/IPC
+ip-26-0-172-142:789261:789958 [6] NCCL INFO Channel 06/1 : 1[6] -> 0[4] via P2P/IPC
+ip-26-0-172-142:789261:789958 [6] NCCL INFO Channel 07/1 : 1[6] -> 0[4] via P2P/IPC
+ip-26-0-172-142:789261:789958 [6] NCCL INFO Channel 08/1 : 1[6] -> 0[4] via P2P/IPC
+ip-26-0-172-142:789261:789958 [6] NCCL INFO Channel 09/1 : 1[6] -> 0[4] via P2P/IPC
+ip-26-0-172-142:789261:789958 [6] NCCL INFO Channel 10/1 : 1[6] -> 0[4] via P2P/IPC
+ip-26-0-172-142:789261:789958 [6] NCCL INFO Channel 11/1 : 1[6] -> 0[4] via P2P/IPC
+ip-26-0-172-142:789261:789958 [6] NCCL INFO Channel 12/1 : 1[6] -> 0[4] via P2P/IPC
+ip-26-0-172-142:789261:789958 [6] NCCL INFO Channel 13/1 : 1[6] -> 0[4] via P2P/IPC
+ip-26-0-172-142:789261:789958 [6] NCCL INFO Channel 14/1 : 1[6] -> 0[4] via P2P/IPC
+ip-26-0-172-142:789261:789958 [6] NCCL INFO Channel 15/1 : 1[6] -> 0[4] via P2P/IPC
+ip-26-0-172-142:789261:789958 [6] NCCL INFO Channel 16/1 : 1[6] -> 0[4] via P2P/IPC
+ip-26-0-172-142:789261:789958 [6] NCCL INFO Channel 17/1 : 1[6] -> 0[4] via P2P/IPC
+ip-26-0-172-142:789261:789958 [6] NCCL INFO Channel 18/1 : 1[6] -> 0[4] via P2P/IPC
+ip-26-0-172-142:789261:789958 [6] NCCL INFO Channel 19/1 : 1[6] -> 0[4] via P2P/IPC
+ip-26-0-172-142:789261:789958 [6] NCCL INFO Channel 20/1 : 1[6] -> 0[4] via P2P/IPC
+ip-26-0-172-142:789261:789958 [6] NCCL INFO Channel 21/1 : 1[6] -> 0[4] via P2P/IPC
+ip-26-0-172-142:789261:789958 [6] NCCL INFO Channel 22/1 : 1[6] -> 0[4] via P2P/IPC
+ip-26-0-172-142:789261:789958 [6] NCCL INFO Channel 23/1 : 1[6] -> 0[4] via P2P/IPC
+ip-26-0-172-142:789261:789958 [6] NCCL INFO Channel 24/1 : 1[6] -> 0[4] via P2P/IPC
+ip-26-0-172-142:789261:789958 [6] NCCL INFO Channel 25/1 : 1[6] -> 0[4] via P2P/IPC
+ip-26-0-172-142:789261:789958 [6] NCCL INFO Channel 26/1 : 1[6] -> 0[4] via P2P/IPC
+ip-26-0-172-142:789261:789958 [6] NCCL INFO Channel 27/1 : 1[6] -> 0[4] via P2P/IPC
+ip-26-0-172-142:789261:789958 [6] NCCL INFO Channel 28/1 : 1[6] -> 0[4] via P2P/IPC
+ip-26-0-172-142:789261:789958 [6] NCCL INFO Channel 29/1 : 1[6] -> 0[4] via P2P/IPC
+ip-26-0-172-142:789261:789958 [6] NCCL INFO Channel 30/1 : 1[6] -> 0[4] via P2P/IPC
+ip-26-0-172-142:789261:789958 [6] NCCL INFO Channel 31/1 : 1[6] -> 0[4] via P2P/IPC
+ip-26-0-172-142:789262:789959 [7] NCCL INFO Channel 00/1 : 1[7] -> 0[5] via P2P/IPC
+ip-26-0-172-142:789262:789959 [7] NCCL INFO Channel 01/1 : 1[7] -> 0[5] via P2P/IPC
+ip-26-0-172-142:789262:789959 [7] NCCL INFO Channel 02/1 : 1[7] -> 0[5] via P2P/IPC
+ip-26-0-172-142:789262:789959 [7] NCCL INFO Channel 03/1 : 1[7] -> 0[5] via P2P/IPC
+ip-26-0-172-142:789262:789959 [7] NCCL INFO Channel 04/1 : 1[7] -> 0[5] via P2P/IPC
+ip-26-0-172-142:789262:789959 [7] NCCL INFO Channel 05/1 : 1[7] -> 0[5] via P2P/IPC
+ip-26-0-172-142:789262:789959 [7] NCCL INFO Channel 06/1 : 1[7] -> 0[5] via P2P/IPC
+ip-26-0-172-142:789262:789959 [7] NCCL INFO Channel 07/1 : 1[7] -> 0[5] via P2P/IPC
+ip-26-0-172-142:789262:789959 [7] NCCL INFO Channel 08/1 : 1[7] -> 0[5] via P2P/IPC
+ip-26-0-172-142:789262:789959 [7] NCCL INFO Channel 09/1 : 1[7] -> 0[5] via P2P/IPC
+ip-26-0-172-142:789262:789959 [7] NCCL INFO Channel 10/1 : 1[7] -> 0[5] via P2P/IPC
+ip-26-0-172-142:789262:789959 [7] NCCL INFO Channel 11/1 : 1[7] -> 0[5] via P2P/IPC
+ip-26-0-172-142:789262:789959 [7] NCCL INFO Channel 12/1 : 1[7] -> 0[5] via P2P/IPC
+ip-26-0-172-142:789262:789959 [7] NCCL INFO Channel 13/1 : 1[7] -> 0[5] via P2P/IPC
+ip-26-0-172-142:789262:789959 [7] NCCL INFO Channel 14/1 : 1[7] -> 0[5] via P2P/IPC
+ip-26-0-172-142:789262:789959 [7] NCCL INFO Channel 15/1 : 1[7] -> 0[5] via P2P/IPC
+ip-26-0-172-142:789262:789959 [7] NCCL INFO Channel 16/1 : 1[7] -> 0[5] via P2P/IPC
+ip-26-0-172-142:789262:789959 [7] NCCL INFO Channel 17/1 : 1[7] -> 0[5] via P2P/IPC
+ip-26-0-172-142:789262:789959 [7] NCCL INFO Channel 18/1 : 1[7] -> 0[5] via P2P/IPC
+ip-26-0-172-142:789262:789959 [7] NCCL INFO Channel 19/1 : 1[7] -> 0[5] via P2P/IPC
+ip-26-0-172-142:789262:789959 [7] NCCL INFO Channel 20/1 : 1[7] -> 0[5] via P2P/IPC
+ip-26-0-172-142:789262:789959 [7] NCCL INFO Channel 21/1 : 1[7] -> 0[5] via P2P/IPC
+ip-26-0-172-142:789262:789959 [7] NCCL INFO Channel 22/1 : 1[7] -> 0[5] via P2P/IPC
+ip-26-0-172-142:789262:789959 [7] NCCL INFO Channel 23/1 : 1[7] -> 0[5] via P2P/IPC
+ip-26-0-172-142:789262:789959 [7] NCCL INFO Channel 24/1 : 1[7] -> 0[5] via P2P/IPC
+ip-26-0-172-142:789262:789959 [7] NCCL INFO Channel 25/1 : 1[7] -> 0[5] via P2P/IPC
+ip-26-0-172-142:789262:789959 [7] NCCL INFO Channel 26/1 : 1[7] -> 0[5] via P2P/IPC
+ip-26-0-172-142:789262:789959 [7] NCCL INFO Channel 27/1 : 1[7] -> 0[5] via P2P/IPC
+ip-26-0-172-142:789262:789959 [7] NCCL INFO Channel 28/1 : 1[7] -> 0[5] via P2P/IPC
+ip-26-0-172-142:789262:789959 [7] NCCL INFO Channel 29/1 : 1[7] -> 0[5] via P2P/IPC
+ip-26-0-172-142:789262:789959 [7] NCCL INFO Channel 30/1 : 1[7] -> 0[5] via P2P/IPC
+ip-26-0-172-142:789262:789959 [7] NCCL INFO Channel 31/1 : 1[7] -> 0[5] via P2P/IPC
+ip-26-0-172-142:789259:789961 [4] NCCL INFO Channel 00/1 : 1[4] -> 0[2] via P2P/IPC
+ip-26-0-172-142:789259:789961 [4] NCCL INFO Channel 01/1 : 1[4] -> 0[2] via P2P/IPC
+ip-26-0-172-142:789259:789961 [4] NCCL INFO Channel 02/1 : 1[4] -> 0[2] via P2P/IPC
+ip-26-0-172-142:789259:789961 [4] NCCL INFO Channel 03/1 : 1[4] -> 0[2] via P2P/IPC
+ip-26-0-172-142:789259:789961 [4] NCCL INFO Channel 04/1 : 1[4] -> 0[2] via P2P/IPC
+ip-26-0-172-142:789259:789961 [4] NCCL INFO Channel 05/1 : 1[4] -> 0[2] via P2P/IPC
+ip-26-0-172-142:789259:789961 [4] NCCL INFO Channel 06/1 : 1[4] -> 0[2] via P2P/IPC
+ip-26-0-172-142:789259:789961 [4] NCCL INFO Channel 07/1 : 1[4] -> 0[2] via P2P/IPC
+ip-26-0-172-142:789259:789961 [4] NCCL INFO Channel 08/1 : 1[4] -> 0[2] via P2P/IPC
+ip-26-0-172-142:789259:789961 [4] NCCL INFO Channel 09/1 : 1[4] -> 0[2] via P2P/IPC
+ip-26-0-172-142:789259:789961 [4] NCCL INFO Channel 10/1 : 1[4] -> 0[2] via P2P/IPC
+ip-26-0-172-142:789259:789961 [4] NCCL INFO Channel 11/1 : 1[4] -> 0[2] via P2P/IPC
+ip-26-0-172-142:789259:789961 [4] NCCL INFO Channel 12/1 : 1[4] -> 0[2] via P2P/IPC
+ip-26-0-172-142:789259:789961 [4] NCCL INFO Channel 13/1 : 1[4] -> 0[2] via P2P/IPC
+ip-26-0-172-142:789259:789961 [4] NCCL INFO Channel 14/1 : 1[4] -> 0[2] via P2P/IPC
+ip-26-0-172-142:789259:789961 [4] NCCL INFO Channel 15/1 : 1[4] -> 0[2] via P2P/IPC
+ip-26-0-172-142:789259:789961 [4] NCCL INFO Channel 16/1 : 1[4] -> 0[2] via P2P/IPC
+ip-26-0-172-142:789259:789961 [4] NCCL INFO Channel 17/1 : 1[4] -> 0[2] via P2P/IPC
+ip-26-0-172-142:789259:789961 [4] NCCL INFO Channel 18/1 : 1[4] -> 0[2] via P2P/IPC
+ip-26-0-172-142:789259:789961 [4] NCCL INFO Channel 19/1 : 1[4] -> 0[2] via P2P/IPC
+ip-26-0-172-142:789259:789961 [4] NCCL INFO Channel 20/1 : 1[4] -> 0[2] via P2P/IPC
+ip-26-0-172-142:789259:789961 [4] NCCL INFO Channel 21/1 : 1[4] -> 0[2] via P2P/IPC
+ip-26-0-172-142:789259:789961 [4] NCCL INFO Channel 22/1 : 1[4] -> 0[2] via P2P/IPC
+ip-26-0-172-142:789259:789961 [4] NCCL INFO Channel 23/1 : 1[4] -> 0[2] via P2P/IPC
+ip-26-0-172-142:789259:789961 [4] NCCL INFO Channel 24/1 : 1[4] -> 0[2] via P2P/IPC
+ip-26-0-172-142:789259:789961 [4] NCCL INFO Channel 25/1 : 1[4] -> 0[2] via P2P/IPC
+ip-26-0-172-142:789259:789961 [4] NCCL INFO Channel 26/1 : 1[4] -> 0[2] via P2P/IPC
+ip-26-0-172-142:789259:789961 [4] NCCL INFO Channel 27/1 : 1[4] -> 0[2] via P2P/IPC
+ip-26-0-172-142:789259:789961 [4] NCCL INFO Channel 28/1 : 1[4] -> 0[2] via P2P/IPC
+ip-26-0-172-142:789259:789961 [4] NCCL INFO Channel 29/1 : 1[4] -> 0[2] via P2P/IPC
+ip-26-0-172-142:789259:789961 [4] NCCL INFO Channel 30/1 : 1[4] -> 0[2] via P2P/IPC
+ip-26-0-172-142:789259:789961 [4] NCCL INFO Channel 31/1 : 1[4] -> 0[2] via P2P/IPC
+ip-26-0-172-142:789260:789962 [5] NCCL INFO Channel 00/1 : 1[5] -> 0[3] via P2P/IPC
+ip-26-0-172-142:789260:789962 [5] NCCL INFO Channel 01/1 : 1[5] -> 0[3] via P2P/IPC
+ip-26-0-172-142:789260:789962 [5] NCCL INFO Channel 02/1 : 1[5] -> 0[3] via P2P/IPC
+ip-26-0-172-142:789260:789962 [5] NCCL INFO Channel 03/1 : 1[5] -> 0[3] via P2P/IPC
+ip-26-0-172-142:789260:789962 [5] NCCL INFO Channel 04/1 : 1[5] -> 0[3] via P2P/IPC
+ip-26-0-172-142:789260:789962 [5] NCCL INFO Channel 05/1 : 1[5] -> 0[3] via P2P/IPC
+ip-26-0-172-142:789260:789962 [5] NCCL INFO Channel 06/1 : 1[5] -> 0[3] via P2P/IPC
+ip-26-0-172-142:789260:789962 [5] NCCL INFO Channel 07/1 : 1[5] -> 0[3] via P2P/IPC
+ip-26-0-172-142:789260:789962 [5] NCCL INFO Channel 08/1 : 1[5] -> 0[3] via P2P/IPC
+ip-26-0-172-142:789260:789962 [5] NCCL INFO Channel 09/1 : 1[5] -> 0[3] via P2P/IPC
+ip-26-0-172-142:789260:789962 [5] NCCL INFO Channel 10/1 : 1[5] -> 0[3] via P2P/IPC
+ip-26-0-172-142:789260:789962 [5] NCCL INFO Channel 11/1 : 1[5] -> 0[3] via P2P/IPC
+ip-26-0-172-142:789260:789962 [5] NCCL INFO Channel 12/1 : 1[5] -> 0[3] via P2P/IPC
+ip-26-0-172-142:789260:789962 [5] NCCL INFO Channel 13/1 : 1[5] -> 0[3] via P2P/IPC
+ip-26-0-172-142:789260:789962 [5] NCCL INFO Channel 14/1 : 1[5] -> 0[3] via P2P/IPC
+ip-26-0-172-142:789260:789962 [5] NCCL INFO Channel 15/1 : 1[5] -> 0[3] via P2P/IPC
+ip-26-0-172-142:789260:789962 [5] NCCL INFO Channel 16/1 : 1[5] -> 0[3] via P2P/IPC
+ip-26-0-172-142:789260:789962 [5] NCCL INFO Channel 17/1 : 1[5] -> 0[3] via P2P/IPC
+ip-26-0-172-142:789260:789962 [5] NCCL INFO Channel 18/1 : 1[5] -> 0[3] via P2P/IPC
+ip-26-0-172-142:789260:789962 [5] NCCL INFO Channel 19/1 : 1[5] -> 0[3] via P2P/IPC
+ip-26-0-172-142:789260:789962 [5] NCCL INFO Channel 20/1 : 1[5] -> 0[3] via P2P/IPC
+ip-26-0-172-142:789260:789962 [5] NCCL INFO Channel 21/1 : 1[5] -> 0[3] via P2P/IPC
+ip-26-0-172-142:789260:789962 [5] NCCL INFO Channel 22/1 : 1[5] -> 0[3] via P2P/IPC
+ip-26-0-172-142:789260:789962 [5] NCCL INFO Channel 23/1 : 1[5] -> 0[3] via P2P/IPC
+ip-26-0-172-142:789260:789962 [5] NCCL INFO Channel 24/1 : 1[5] -> 0[3] via P2P/IPC
+ip-26-0-172-142:789260:789962 [5] NCCL INFO Channel 25/1 : 1[5] -> 0[3] via P2P/IPC
+ip-26-0-172-142:789260:789962 [5] NCCL INFO Channel 26/1 : 1[5] -> 0[3] via P2P/IPC
+ip-26-0-172-142:789260:789962 [5] NCCL INFO Channel 27/1 : 1[5] -> 0[3] via P2P/IPC
+ip-26-0-172-142:789260:789962 [5] NCCL INFO Channel 28/1 : 1[5] -> 0[3] via P2P/IPC
+ip-26-0-172-142:789260:789962 [5] NCCL INFO Channel 29/1 : 1[5] -> 0[3] via P2P/IPC
+ip-26-0-172-142:789260:789962 [5] NCCL INFO Channel 30/1 : 1[5] -> 0[3] via P2P/IPC
+ip-26-0-172-142:789260:789962 [5] NCCL INFO Channel 31/1 : 1[5] -> 0[3] via P2P/IPC
+ip-26-0-172-142:789257:789964 [2] NCCL INFO Channel 00/1 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789257:789964 [2] NCCL INFO Channel 01/1 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789257:789964 [2] NCCL INFO Channel 02/1 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789257:789964 [2] NCCL INFO Channel 03/1 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789257:789964 [2] NCCL INFO Channel 04/1 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789257:789964 [2] NCCL INFO Channel 05/1 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789257:789964 [2] NCCL INFO Channel 06/1 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789257:789964 [2] NCCL INFO Channel 07/1 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789257:789964 [2] NCCL INFO Channel 08/1 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789257:789964 [2] NCCL INFO Channel 09/1 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789257:789964 [2] NCCL INFO Channel 10/1 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789257:789964 [2] NCCL INFO Channel 11/1 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789257:789964 [2] NCCL INFO Channel 12/1 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789257:789964 [2] NCCL INFO Channel 13/1 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789257:789964 [2] NCCL INFO Channel 14/1 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789257:789964 [2] NCCL INFO Channel 15/1 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789257:789964 [2] NCCL INFO Channel 16/1 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789257:789964 [2] NCCL INFO Channel 17/1 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789257:789964 [2] NCCL INFO Channel 18/1 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789257:789964 [2] NCCL INFO Channel 19/1 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789257:789964 [2] NCCL INFO Channel 20/1 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789257:789964 [2] NCCL INFO Channel 21/1 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789257:789964 [2] NCCL INFO Channel 22/1 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789257:789964 [2] NCCL INFO Channel 23/1 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789257:789964 [2] NCCL INFO Channel 24/1 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789257:789964 [2] NCCL INFO Channel 25/1 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789257:789964 [2] NCCL INFO Channel 26/1 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789257:789964 [2] NCCL INFO Channel 27/1 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789257:789964 [2] NCCL INFO Channel 28/1 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789257:789964 [2] NCCL INFO Channel 29/1 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789257:789964 [2] NCCL INFO Channel 30/1 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789257:789964 [2] NCCL INFO Channel 31/1 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-172-142:789258:789966 [3] NCCL INFO Channel 00/1 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-172-142:789258:789966 [3] NCCL INFO Channel 01/1 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-172-142:789258:789966 [3] NCCL INFO Channel 02/1 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-172-142:789258:789966 [3] NCCL INFO Channel 03/1 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-172-142:789258:789966 [3] NCCL INFO Channel 04/1 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-172-142:789258:789966 [3] NCCL INFO Channel 05/1 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-172-142:789258:789966 [3] NCCL INFO Channel 06/1 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-172-142:789258:789966 [3] NCCL INFO Channel 07/1 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-172-142:789258:789966 [3] NCCL INFO Channel 08/1 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-172-142:789258:789966 [3] NCCL INFO Channel 09/1 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-172-142:789258:789966 [3] NCCL INFO Channel 10/1 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-172-142:789258:789966 [3] NCCL INFO Channel 11/1 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-172-142:789258:789966 [3] NCCL INFO Channel 12/1 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-172-142:789258:789966 [3] NCCL INFO Channel 13/1 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-172-142:789258:789966 [3] NCCL INFO Channel 14/1 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-172-142:789258:789966 [3] NCCL INFO Channel 15/1 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-172-142:789258:789966 [3] NCCL INFO Channel 16/1 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-172-142:789258:789966 [3] NCCL INFO Channel 17/1 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-172-142:789258:789966 [3] NCCL INFO Channel 18/1 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-172-142:789258:789966 [3] NCCL INFO Channel 19/1 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-172-142:789258:789966 [3] NCCL INFO Channel 20/1 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-172-142:789258:789966 [3] NCCL INFO Channel 21/1 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-172-142:789258:789966 [3] NCCL INFO Channel 22/1 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-172-142:789258:789966 [3] NCCL INFO Channel 23/1 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-172-142:789258:789966 [3] NCCL INFO Channel 24/1 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-172-142:789258:789966 [3] NCCL INFO Channel 25/1 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-172-142:789258:789966 [3] NCCL INFO Channel 26/1 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-172-142:789258:789966 [3] NCCL INFO Channel 27/1 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-172-142:789258:789966 [3] NCCL INFO Channel 28/1 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-172-142:789258:789966 [3] NCCL INFO Channel 29/1 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-172-142:789258:789966 [3] NCCL INFO Channel 30/1 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-172-142:789258:789966 [3] NCCL INFO Channel 31/1 : 1[3] -> 0[1] via P2P/IPC
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+01/07/2025 00:37:39 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]:  Memory usage: 3649.68MiB. Peak allocated 23995.63MiB. Peak reserved: 25152.00MiB
+01/07/2025 00:37:40 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]:  Memory usage: 5441.73MiB. Peak allocated 7681.79MiB. Peak reserved: 26950.00MiB
+01/07/2025 00:37:40 [INFO|DP=0|PP=7|TP=0|ip-26-0-173-121]: iteration: 1 / 100 | consumed_tokens: 1.05M | elapsed_time_per_iteration_ms: 25.2K | tokens_per_sec: 41.7K | tokens_per_sec_per_gpu: 2.6K | global_batch_size: 256 | lm_loss: 12 | lr: 0.00015 | model_tflops_per_gpu: 25.2 | hardware_tflops_per_gpu: 25.2 | grad_norm: 0.81 | cuda_memory_allocated: 3.29G | cuda_max_memory_reserved: 14G | hd_total_memory_tb: 312G | hd_used_memory_tb: 64.6G | hd_free_memory_tb: 248G
+01/07/2025 00:37:42 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]:  Memory usage: 5441.73MiB. Peak allocated 25787.68MiB. Peak reserved: 26950.00MiB
+01/07/2025 00:37:43 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]:  Memory usage: 5441.73MiB. Peak allocated 7681.79MiB. Peak reserved: 27848.00MiB
+01/07/2025 00:37:43 [INFO|DP=0|PP=7|TP=0|ip-26-0-173-121]: iteration: 2 / 100 | consumed_tokens: 2.1M | elapsed_time_per_iteration_ms: 2.92K | tokens_per_sec: 359K | tokens_per_sec_per_gpu: 22.4K | global_batch_size: 256 | lm_loss: 12 | lr: 0.0003 | model_tflops_per_gpu: 217 | hardware_tflops_per_gpu: 217 | grad_norm: 0.811 | cuda_memory_allocated: 3.29G | cuda_max_memory_reserved: 14G | hd_total_memory_tb: 312G | hd_used_memory_tb: 64.6G | hd_free_memory_tb: 248G
+01/07/2025 00:37:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]:  Memory usage: 5441.73MiB. Peak allocated 25787.68MiB. Peak reserved: 27848.00MiB
+01/07/2025 00:37:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]:  Memory usage: 5441.73MiB. Peak allocated 7681.79MiB. Peak reserved: 27848.00MiB
+01/07/2025 00:37:46 [INFO|DP=0|PP=7|TP=0|ip-26-0-173-121]: iteration: 3 / 100 | consumed_tokens: 3.15M | elapsed_time_per_iteration_ms: 2.93K | tokens_per_sec: 358K | tokens_per_sec_per_gpu: 22.4K | global_batch_size: 256 | lm_loss: 12 | lr: 0.000296 | model_tflops_per_gpu: 216 | hardware_tflops_per_gpu: 216 | grad_norm: 0.802 | cuda_memory_allocated: 3.29G | cuda_max_memory_reserved: 14G | hd_total_memory_tb: 312G | hd_used_memory_tb: 64.6G | hd_free_memory_tb: 248G
+01/07/2025 00:37:48 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]:  Memory usage: 5441.73MiB. Peak allocated 25787.68MiB. Peak reserved: 27848.00MiB
+01/07/2025 00:37:49 [INFO|DP=0|PP=7|TP=0|ip-26-0-173-121]: iteration: 4 / 100 | consumed_tokens: 4.19M | elapsed_time_per_iteration_ms: 2.93K | tokens_per_sec: 358K | tokens_per_sec_per_gpu: 22.4K | global_batch_size: 256 | lm_loss: 12 | lr: 0.000283 | model_tflops_per_gpu: 216 | hardware_tflops_per_gpu: 216 | grad_norm: 0.755 | cuda_memory_allocated: 3.29G | cuda_max_memory_reserved: 14G | hd_total_memory_tb: 312G | hd_used_memory_tb: 64.6G | hd_free_memory_tb: 248G
+01/07/2025 00:37:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]: | job_id   | name                                                           | nodes | seq_len | mbs | batch_accum | gbs | mTFLOPs | hTFLOPs | tok/s/gpu | AllReduce (GB/s) | AllGather (GB/s) | ReduceScatter (GB/s) | AR Intra-node (GB/s) | AG Intra-node (GB/s) | RS Intra-node (GB/s) | Mem Alloc (GB) | Mem Res (GB) | dp | pp | tp | pp_engine                           | tp_mode        | tp_async_comm | hidden_size | hidden_act | num_layers | num_heads | num_kv_heads | max_pos | vocab_size | tie_word_embeddings | dtype          | zero_stage | ddp_bucket_cap_mb | accumulate_grad_in_fp32 | Total Params | Local Params |
+01/07/2025 00:37:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]: | -------- | -------------------------------------------------------------- | ----- | ------- | --- | ----------- | --- | ------- | ------- | --------- | ---------------- | ---------------- | -------------------- | -------------------- | -------------------- | -------------------- | -------------- | ------------ | -- | -- | -- | ----------------------------------- | -------------- | ------------- | ----------- | ---------- | ---------- | --------- | ------------ | ------- | ---------- | ------------------- | -------------- | ---------- | ----------------- | ----------------------- | ------------ | ------------ |
+01/07/2025 00:37:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]: | 14098592 | 1.34G_dp2_tp1_pp8_acc64_mbs2_seq4096_zero1_tpmodeRED_vocab131k | 2     | 4096    | 2   | 64          | 256 | 216.33  | 216.33  | 22386.35  | 347.29           | 172.87           | 169.10               | 426.26               | 242.50               | 242.84               | 7.50           | 27.20        | 2  | 8  | 1  | OneForwardOneBackwardPipelineEngine | REDUCE_SCATTER | True          | 2048        | silu       | 16         | 32        | 32           | 4096    | 131072     | True                | torch.bfloat16 | 1          | 25                | True                    | 1.61G        | 470M         |
+01/07/2025 00:37:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]: Attempting to write benchmark results to CSV file: benchmark/results/bench_final2.csv
+01/07/2025 00:37:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]: Acquired lock for CSV file: benchmark/results/bench_final2.csv
+01/07/2025 00:37:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]: Successfully wrote to CSV file: benchmark/results/bench_final2.csv. Releasing lock...
+01/07/2025 00:37:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-172-142]: Throughput logging complete
+ip-26-0-172-142:789255:789738 [32580] NCCL INFO [Service thread] Connection closed by localRank 2
+ip-26-0-172-142:789261:789737 [32708] NCCL INFO [Service thread] Connection closed by localRank 2
+ip-26-0-172-142:789257:789739 [32511] NCCL INFO [Service thread] Connection closed by localRank 2
+ip-26-0-172-142:789255:789642 [32581] NCCL INFO [Service thread] Connection closed by localRank 4
+ip-26-0-172-142:789261:789602 [32710] NCCL INFO [Service thread] Connection closed by localRank 4
+ip-26-0-172-142:789255:789608 [32583] NCCL INFO [Service thread] Connection closed by localRank 4
+ip-26-0-172-142:789256:789603 [32520] NCCL INFO [Service thread] Connection closed by localRank 4
+ip-26-0-172-142:789257:789606 [32513] NCCL INFO [Service thread] Connection closed by localRank 4
+ip-26-0-172-142:789262:789601 [32752] NCCL INFO [Service thread] Connection closed by localRank 4
+ip-26-0-172-142:789258:789607 [32526] NCCL INFO [Service thread] Connection closed by localRank 4
+ip-26-0-172-142:789261:789438 [32712] NCCL INFO [Service thread] Connection closed by localRank 4
+ip-26-0-172-142:789257:789442 [32514] NCCL INFO [Service thread] Connection closed by localRank 4
+ip-26-0-172-142:789256:789435 [32521] NCCL INFO [Service thread] Connection closed by localRank 4
+ip-26-0-172-142:789255:789440 [32584] NCCL INFO [Service thread] Connection closed by localRank 4
+ip-26-0-172-142:789262:789436 [32753] NCCL INFO [Service thread] Connection closed by localRank 4
+ip-26-0-172-142:789258:789441 [32527] NCCL INFO [Service thread] Connection closed by localRank 4
+ip-26-0-173-121:3987797:3988262 [32597] NCCL INFO [Service thread] Connection closed by localRank 3
+ip-26-0-173-121:3987799:3988261 [32745] NCCL INFO [Service thread] Connection closed by localRank 3
+ip-26-0-173-121:3987795:3988263 [32683] NCCL INFO [Service thread] Connection closed by localRank 3
+ip-26-0-173-121:3987794:3988179 [32594] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-173-121:3987799:3988139 [32747] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-173-121:3987798:3988145 [32636] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-173-121:3987797:3988138 [32599] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-173-121:3987795:3988142 [32685] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-173-121:3987800:3988140 [32728] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-173-121:3987796:3988141 [32557] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-173-121:3987794:3988144 [32596] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-173-121:3987799:3987974 [32749] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-173-121:3987798:3987980 [32638] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-173-121:3987797:3987979 [32600] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-173-121:3987796:3987975 [32558] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-173-121:3987795:3987977 [32686] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-173-121:3987794:3987978 [32598] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-173-121:3987800:3987976 [32729] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-172-142:789256:789731 [32518] NCCL INFO [Service thread] Connection closed by localRank 2
+ip-26-0-172-142:789262:789732 [32750] NCCL INFO [Service thread] Connection closed by localRank 2
+ip-26-0-172-142:789255:789642 [32581] NCCL INFO [Service thread] Connection closed by localRank 5
+ip-26-0-172-142:789262:789601 [32752] NCCL INFO [Service thread] Connection closed by localRank 5
+ip-26-0-172-142:789261:789602 [32710] NCCL INFO [Service thread] Connection closed by localRank 5
+ip-26-0-172-142:789256:789603 [32520] NCCL INFO [Service thread] Connection closed by localRank 5
+ip-26-0-172-142:789255:789608 [32583] NCCL INFO [Service thread] Connection closed by localRank 5
+ip-26-0-172-142:789261:789438 [32712] NCCL INFO [Service thread] Connection closed by localRank 5
+ip-26-0-172-142:789262:789436 [32753] NCCL INFO [Service thread] Connection closed by localRank 5
+ip-26-0-172-142:789256:789435 [32521] NCCL INFO [Service thread] Connection closed by localRank 5
+ip-26-0-172-142:789255:789440 [32584] NCCL INFO [Service thread] Connection closed by localRank 5
+ip-26-0-172-142:789261:789737 [32708] NCCL INFO [Service thread] Connection closed by localRank 1
+ip-26-0-172-142:789255:789738 [32580] NCCL INFO [Service thread] Connection closed by localRank 1
+ip-26-0-172-142:789255:789642 [32581] NCCL INFO [Service thread] Connection closed by localRank 2
+ip-26-0-172-142:789262:789601 [32752] NCCL INFO [Service thread] Connection closed by localRank 2
+ip-26-0-172-142:789261:789602 [32710] NCCL INFO [Service thread] Connection closed by localRank 2
+ip-26-0-172-142:789255:789608 [32583] NCCL INFO [Service thread] Connection closed by localRank 2
+ip-26-0-172-142:789262:789436 [32753] NCCL INFO [Service thread] Connection closed by localRank 2
+ip-26-0-172-142:789261:789438 [32712] NCCL INFO [Service thread] Connection closed by localRank 2
+ip-26-0-172-142:789255:789440 [32584] NCCL INFO [Service thread] Connection closed by localRank 2
+ip-26-0-172-142:789262:789732 [32750] NCCL INFO [Service thread] Connection closed by localRank 1
+ip-26-0-172-142:789255:789642 [32581] NCCL INFO [Service thread] Connection closed by localRank 3
+ip-26-0-172-142:789262:789601 [32752] NCCL INFO [Service thread] Connection closed by localRank 3
+ip-26-0-172-142:789261:789602 [32710] NCCL INFO [Service thread] Connection closed by localRank 3
+ip-26-0-172-142:789255:789608 [32583] NCCL INFO [Service thread] Connection closed by localRank 3
+ip-26-0-172-142:789262:789436 [32753] NCCL INFO [Service thread] Connection closed by localRank 3
+ip-26-0-172-142:789261:789438 [32712] NCCL INFO [Service thread] Connection closed by localRank 3
+ip-26-0-172-142:789255:789440 [32584] NCCL INFO [Service thread] Connection closed by localRank 3
+ip-26-0-172-142:789262:789732 [32750] NCCL INFO [Service thread] Connection closed by localRank 0
+ip-26-0-172-142:789255:789642 [32581] NCCL INFO [Service thread] Connection closed by localRank 1
+ip-26-0-172-142:789262:789601 [32752] NCCL INFO [Service thread] Connection closed by localRank 1
+ip-26-0-172-142:789261:789602 [32710] NCCL INFO [Service thread] Connection closed by localRank 1
+ip-26-0-172-142:789255:789608 [32583] NCCL INFO [Service thread] Connection closed by localRank 1
+ip-26-0-172-142:789262:789436 [32753] NCCL INFO [Service thread] Connection closed by localRank 1
+ip-26-0-172-142:789261:789438 [32712] NCCL INFO [Service thread] Connection closed by localRank 1
+ip-26-0-172-142:789255:789440 [32584] NCCL INFO [Service thread] Connection closed by localRank 1
+ip-26-0-173-121:3987800:3988265 [32729] NCCL INFO [Service thread] Connection closed by localRank 1
+ip-26-0-173-121:3987800:3988140 [32728] NCCL INFO [Service thread] Connection closed by localRank 2
+ip-26-0-173-121:3987795:3988142 [32685] NCCL INFO [Service thread] Connection closed by localRank 2
+ip-26-0-173-121:3987800:3987976 [32729] NCCL INFO [Service thread] Connection closed by localRank 2
+ip-26-0-173-121:3987795:3987977 [32686] NCCL INFO [Service thread] Connection closed by localRank 2
+ip-26-0-173-121:3987795:3988263 [32683] NCCL INFO [Service thread] Connection closed by localRank 1
+ip-26-0-173-121:3987800:3988140 [32728] NCCL INFO [Service thread] Connection closed by localRank 3
+ip-26-0-173-121:3987795:3988142 [32685] NCCL INFO [Service thread] Connection closed by localRank 3
+ip-26-0-173-121:3987800:3987976 [32729] NCCL INFO [Service thread] Connection closed by localRank 3
+ip-26-0-173-121:3987795:3987977 [32686] NCCL INFO [Service thread] Connection closed by localRank 3
+[2025-01-07 00:38:09,919] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3987795 closing signal SIGTERM
+[2025-01-07 00:38:09,920] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3987800 closing signal SIGTERM
+[2025-01-07 00:38:09,928] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 789256 closing signal SIGTERM
+[2025-01-07 00:38:09,929] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 789261 closing signal SIGTERM
+[2025-01-07 00:38:09,929] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 789262 closing signal SIGTERM
+[2025-01-07 00:38:11,007] torch.distributed.elastic.multiprocessing.api: [ERROR] failed (exitcode: 1) local_rank: 0 (pid: 789255) of binary: /fsx/nouamane/miniconda/envs/2-1-cu121/bin/python
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 264, in launch_agent
+    raise ChildFailedError(
+torch.distributed.elastic.multiprocessing.errors.ChildFailedError: 
+============================================================
+/fsx/nouamane/projects/nanotron/run_train.py FAILED
+------------------------------------------------------------
+Failures:
+[1]:
+  time      : 2025-01-07_00:38:09
+  host      : ip-26-0-172-142.ec2.internal
+  rank      : 2 (local_rank: 2)
+  exitcode  : 1 (pid: 789257)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[2]:
+  time      : 2025-01-07_00:38:09
+  host      : ip-26-0-172-142.ec2.internal
+  rank      : 3 (local_rank: 3)
+  exitcode  : 1 (pid: 789258)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[3]:
+  time      : 2025-01-07_00:38:09
+  host      : ip-26-0-172-142.ec2.internal
+  rank      : 4 (local_rank: 4)
+  exitcode  : 1 (pid: 789259)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+[4]:
+  time      : 2025-01-07_00:38:09
+  host      : ip-26-0-172-142.ec2.internal
+  rank      : 5 (local_rank: 5)
+  exitcode  : 1 (pid: 789260)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+------------------------------------------------------------
+Root Cause (first observed failure):
+[0]:
+  time      : 2025-01-07_00:38:09
+  host      : ip-26-0-172-142.ec2.internal
+  rank      : 0 (local_rank: 0)
+  exitcode  : 1 (pid: 789255)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+============================================================
+srun: error: ip-26-0-172-142: task 0: Exited with exit code 1
+srun: launch/slurm: _step_signal: Terminating StepId=14098592.0
+[2025-01-07 00:38:11,330] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-07 00:38:11,330] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3987795 closing signal SIGTERM
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 878, in _invoke_run
+    run_result = self._monitor_workers(self._worker_group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/local_elastic_agent.py", line 307, in _monitor_workers
+    result = self._pcontext.wait(0)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 288, in wait
+    return self._poll()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 668, in _poll
+    self.close()  # terminate all running procs
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 331, in close
+    self._close(death_sig=death_sig, timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 713, in _close
+    handler.proc.wait(time_to_wait)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1209, in wait
+    return self._wait(timeout=timeout)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/subprocess.py", line 1953, in _wait
+    time.sleep(delay)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 3987724 got signal: 15
+srun: error: ip-26-0-173-121: task 1: Exited with exit code 1
diff --git a/logs/14098831-bench_1.34G_dp4_tp2_pp8_acc32_mbs2_seq4096_zero1_tpmodeRED_vocab131k.out b/logs/14098831-bench_1.34G_dp4_tp2_pp8_acc32_mbs2_seq4096_zero1_tpmodeRED_vocab131k.out
new file mode 100644
index 0000000000000000000000000000000000000000..ff04cfc9d5765ab7bb60d9e372d52531443ab145
--- /dev/null
+++ b/logs/14098831-bench_1.34G_dp4_tp2_pp8_acc32_mbs2_seq4096_zero1_tpmodeRED_vocab131k.out
@@ -0,0 +1,37710 @@
++ '[' -z 14098831 ']'
++ source /etc/profile.d/modules.sh
+++ . /usr/share/modules/init/bash
++++ unset _mlshdbg
++++ '[' 0 = 1 ']'
++++ unset _mlre _mlIFS
++++ '[' -n x ']'
++++ _mlIFS=' 	
+'
++++ IFS=' '
++++ '[' -n '' ']'
+++++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash autoinit
++++ _mlcode='module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+test 0;'
++++ _mlret=0
++++ '[' -n x ']'
++++ IFS=' 	
+'
++++ unset _mlIFS
++++ unset _mlre _mlv _mlrv
++++ '[' 0 -eq 0 ']'
++++ eval 'module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+test 0;'
+++++ MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl
+++++ export MODULES_CMD
+++++ ENV=/usr/share/modules/init/profile.sh
+++++ export ENV
+++++ BASH_ENV=/usr/share/modules/init/bash
+++++ export BASH_ENV
+++++ MODULESHOME=/usr/share/modules
+++++ export MODULESHOME
+++++ test 0
++++ '[' 0 = 1 ']'
++++ '[' -t 2 ']'
++++ export -f module
++++ export -f switchml
++++ '[' 5 -ge 3 ']'
++++ [[ ehxB =~ i ]]
++++ [[ ! :/admin/home/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/admin/home/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin: =~ :/usr/bin: ]]
++++ '[' '!' -n x ']'
+++++ manpath
++++ [[ ! :/admin/home/nouamane/miniconda/envs/2-1-cu121/man:/admin/home/nouamane/miniconda/envs/2-1-cu121/share/man:/fsx/nouamane/miniconda/man:/fsx/nouamane/miniconda/share/man:/opt/slurm/share/man:/opt/amazon/openmpi/share/man:/opt/amazon/efa/share/man:/usr/local/man:/usr/local/share/man:/usr/share/man:/admin/home/nouamane/.local/share/man:/admin/home/nouamane/.fzf/man:: =~ :/usr/share/man: ]]
++++ unset _mlcode _mlret
++++ '[' -n '' ']'
++ module load cuda/12.1
++ unset _mlshdbg
++ '[' 0 = 1 ']'
++ unset _mlre _mlIFS
++ '[' -n x ']'
++ _mlIFS=' 	
+'
++ IFS=' '
++ '[' -n '' ']'
+++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash load cuda/12.1
++ eval 'CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include;' export 'CPATH;
+LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/opt/aws-ofi-nccl/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/targets/x86_64-linux/lib/:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/lib:/usr/lib;' export 'LD_LIBRARY_PATH;
+FI_EFA_FORK_SAFE=1;' export 'FI_EFA_FORK_SAFE;
+MANPATH=/usr/local/cuda-12.1/share/man::;' export 'MANPATH;
+LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64;' export 'LIBRARY_PATH;
+_LMFILES_=/usr/share/modules/modulefiles/cuda/12.1;' export '_LMFILES_;
+LOADEDMODULES=cuda/12.1;' export 'LOADEDMODULES;
+MPI_PATH=/opt/amazon/openmpi;' export 'MPI_PATH;
+NCCL_HOME_modshare=/opt/nccl/build:1;' export 'NCCL_HOME_modshare;
+NCCL_PROTO=simple;' export 'NCCL_PROTO;
+MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1;' export 'MANPATH_modshare;
+LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1;' export 'LIBRARY_PATH_modshare;
+NCCL_SOCKET_IFNAME=enp;' export 'NCCL_SOCKET_IFNAME;
+AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl;' export 'AWS_OFI_NCCL_HOME;
+NCCL_HOME=/opt/nccl/build;' export 'NCCL_HOME;
+FI_PROVIDER=efa;' export 'FI_PROVIDER;
+AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1;' export 'AWS_OFI_NCCL_HOME_modshare;
+CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1;' export 'CPATH_modshare;
+LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib/:1:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2:/opt/aws-ofi-nccl/lib:2:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/efa/lib:1:/usr/lib:1;' export 'LD_LIBRARY_PATH_modshare;
+FI_EFA_ENABLE_SHM_TRANSFER=1;' export 'FI_EFA_ENABLE_SHM_TRANSFER;
+_LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1;' export '_LMFILES__modshare;
+LOADEDMODULES_modshare=cuda/12.1:1;' export 'LOADEDMODULES_modshare;
+MPI_PATH_modshare=/opt/amazon/openmpi:1;' export 'MPI_PATH_modshare;
+PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/admin/home/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin;' export 'PATH;
+CUDA_HOME=/usr/local/cuda-12.1;' export 'CUDA_HOME;
+PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/local/cuda-12.1/include:1:/usr/bin:1:/opt/amazon/efa/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/opt/slurm/bin:1:/usr/sbin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:2:/opt/amazon/openmpi/bin:1:/admin/home/nouamane/.fzf/bin:1:/admin/home/nouamane/miniconda/envs/2-1-cu121/bin:1:/usr/local/games:1;' export 'PATH_modshare;
+test' '0;'
+++ CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include
+++ export CPATH
+++ LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/opt/aws-ofi-nccl/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/targets/x86_64-linux/lib/:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/lib:/usr/lib
+++ export LD_LIBRARY_PATH
+++ FI_EFA_FORK_SAFE=1
+++ export FI_EFA_FORK_SAFE
+++ MANPATH=/usr/local/cuda-12.1/share/man::
+++ export MANPATH
+++ LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64
+++ export LIBRARY_PATH
+++ _LMFILES_=/usr/share/modules/modulefiles/cuda/12.1
+++ export _LMFILES_
+++ LOADEDMODULES=cuda/12.1
+++ export LOADEDMODULES
+++ MPI_PATH=/opt/amazon/openmpi
+++ export MPI_PATH
+++ NCCL_HOME_modshare=/opt/nccl/build:1
+++ export NCCL_HOME_modshare
+++ NCCL_PROTO=simple
+++ export NCCL_PROTO
+++ MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1
+++ export MANPATH_modshare
+++ LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1
+++ export LIBRARY_PATH_modshare
+++ NCCL_SOCKET_IFNAME=enp
+++ export NCCL_SOCKET_IFNAME
+++ AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl
+++ export AWS_OFI_NCCL_HOME
+++ NCCL_HOME=/opt/nccl/build
+++ export NCCL_HOME
+++ FI_PROVIDER=efa
+++ export FI_PROVIDER
+++ AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1
+++ export AWS_OFI_NCCL_HOME_modshare
+++ CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1
+++ export CPATH_modshare
+++ LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib/:1:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2:/opt/aws-ofi-nccl/lib:2:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/efa/lib:1:/usr/lib:1
+++ export LD_LIBRARY_PATH_modshare
+++ FI_EFA_ENABLE_SHM_TRANSFER=1
+++ export FI_EFA_ENABLE_SHM_TRANSFER
+++ _LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1
+++ export _LMFILES__modshare
+++ LOADEDMODULES_modshare=cuda/12.1:1
+++ export LOADEDMODULES_modshare
+++ MPI_PATH_modshare=/opt/amazon/openmpi:1
+++ export MPI_PATH_modshare
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/admin/home/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
+++ export PATH
+++ CUDA_HOME=/usr/local/cuda-12.1
+++ export CUDA_HOME
+++ PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/local/cuda-12.1/include:1:/usr/bin:1:/opt/amazon/efa/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/opt/slurm/bin:1:/usr/sbin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:2:/opt/amazon/openmpi/bin:1:/admin/home/nouamane/.fzf/bin:1:/admin/home/nouamane/miniconda/envs/2-1-cu121/bin:1:/usr/local/games:1
+++ export PATH_modshare
+++ test 0
++ _mlstatus=0
++ '[' -n x ']'
++ IFS=' 	
+'
++ unset _mlre _mlv _mlrv _mlIFS
++ '[' -n '' ']'
++ unset _mlshdbg
++ return 0
++ source /fsx/nouamane/miniconda/bin/activate
+++ _CONDA_ROOT=/fsx/nouamane/miniconda
+++ . /fsx/nouamane/miniconda/etc/profile.d/conda.sh
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ '[' -z x ']'
+++ conda activate
+++ local cmd=activate
+++ case "$cmd" in
+++ __conda_activate activate
+++ '[' -n '' ']'
+++ local ask_conda
++++ PS1=
++++ __conda_exe shell.posix activate
++++ /fsx/nouamane/miniconda/bin/conda shell.posix activate
+Error while loading conda entry point: conda-libmamba-solver (module 'libmambapy' has no attribute 'QueryFormat')
+++ ask_conda='. "/admin/home/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/fsx/nouamane/miniconda/bin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/admin/home/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_2='\''/admin/home/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/etc/conda/activate.d/libxml2_activate.sh"'
+++ eval '. "/admin/home/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/fsx/nouamane/miniconda/bin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/admin/home/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_2='\''/admin/home/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/etc/conda/activate.d/libxml2_activate.sh"'
++++ . /admin/home/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh
+++++ test -n ''
+++++ unset XML_CATALOG_FILES
+++++ unset xml_catalog_files_libxml2
++++ PS1='(base) '
++++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/fsx/nouamane/miniconda/bin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/admin/home/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
++++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/fsx/nouamane/miniconda/bin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/admin/home/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
++++ export CONDA_PREFIX=/fsx/nouamane/miniconda
++++ CONDA_PREFIX=/fsx/nouamane/miniconda
++++ export CONDA_SHLVL=3
++++ CONDA_SHLVL=3
++++ export CONDA_DEFAULT_ENV=base
++++ CONDA_DEFAULT_ENV=base
++++ export 'CONDA_PROMPT_MODIFIER=(base) '
++++ CONDA_PROMPT_MODIFIER='(base) '
++++ export CONDA_PREFIX_2=/admin/home/nouamane/miniconda/envs/2-1-cu121
++++ CONDA_PREFIX_2=/admin/home/nouamane/miniconda/envs/2-1-cu121
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ . /fsx/nouamane/miniconda/etc/conda/activate.d/libxml2_activate.sh
+++++ test -n ''
+++++ xml_catalog_files_libxml2=
+++++ XML_CATALOG_FILES=
+++++ conda_catalog_files=
+++++ ifs_libxml2=' 	
+'
+++++ IFS=' '
+++++ rem=/fsx/nouamane/miniconda
+++++ for pre in ${rem}
+++++ test '' = /fsx/nouamane/miniconda
+++++ conda_catalog_files=/fsx/nouamane/miniconda
+++++ rem=
+++++ IFS=' 	
+'
+++++ conda_catalog_files='file:///fsx/nouamane/miniconda/etc/xml/catalog file:///etc/xml/catalog'
+++++ export 'XML_CATALOG_FILES=file:///fsx/nouamane/miniconda/etc/xml/catalog file:///etc/xml/catalog'
+++++ XML_CATALOG_FILES='file:///fsx/nouamane/miniconda/etc/xml/catalog file:///etc/xml/catalog'
+++++ unset conda_catalog_files ifs_libxml2 rem
+++ __conda_hashr
+++ '[' -n '' ']'
+++ '[' -n '' ']'
+++ hash -r
++ conda activate 2-1-cu121
++ local cmd=activate
++ case "$cmd" in
++ __conda_activate activate 2-1-cu121
++ '[' -n '' ']'
++ local ask_conda
+++ PS1='(base) '
+++ __conda_exe shell.posix activate 2-1-cu121
+++ /fsx/nouamane/miniconda/bin/conda shell.posix activate 2-1-cu121
+Error while loading conda entry point: conda-libmamba-solver (module 'libmambapy' has no attribute 'QueryFormat')
++ ask_conda='. "/fsx/nouamane/miniconda/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/admin/home/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''4'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_3='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
++ eval '. "/fsx/nouamane/miniconda/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/admin/home/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''4'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_3='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
+++ . /fsx/nouamane/miniconda/etc/conda/deactivate.d/libxml2_deactivate.sh
++++ test -n ''
++++ unset XML_CATALOG_FILES
++++ unset xml_catalog_files_libxml2
+++ PS1='(2-1-cu121) '
+++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/admin/home/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/admin/home/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
+++ export CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ export CONDA_SHLVL=4
+++ CONDA_SHLVL=4
+++ export CONDA_DEFAULT_ENV=2-1-cu121
+++ CONDA_DEFAULT_ENV=2-1-cu121
+++ export 'CONDA_PROMPT_MODIFIER=(2-1-cu121) '
+++ CONDA_PROMPT_MODIFIER='(2-1-cu121) '
+++ export CONDA_PREFIX_3=/fsx/nouamane/miniconda
+++ CONDA_PREFIX_3=/fsx/nouamane/miniconda
+++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ export _CE_M=
+++ _CE_M=
+++ export _CE_CONDA=
+++ _CE_CONDA=
+++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh
++++ test -n ''
++++ xml_catalog_files_libxml2=
++++ XML_CATALOG_FILES=
++++ conda_catalog_files=
++++ ifs_libxml2=' 	
+'
++++ IFS=' '
++++ rem=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ for pre in ${rem}
++++ test '' = /fsx/nouamane/miniconda/envs/2-1-cu121
++++ conda_catalog_files=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ rem=
++++ IFS=' 	
+'
++++ conda_catalog_files='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ export 'XML_CATALOG_FILES=file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ XML_CATALOG_FILES='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ unset conda_catalog_files ifs_libxml2 rem
++ __conda_hashr
++ '[' -n '' ']'
++ '[' -n '' ']'
++ hash -r
++ export PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/admin/home/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
++ PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/admin/home/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
++ '[' -z '' ']'
+++ scontrol show hostnames 'ip-26-0-162-46,ip-26-0-165-213,ip-26-0-166-[15,36,125,214,244],ip-26-0-167-9'
++ export 'NODELIST=ip-26-0-162-46
+ip-26-0-165-213
+ip-26-0-166-15
+ip-26-0-166-36
+ip-26-0-166-125
+ip-26-0-166-214
+ip-26-0-166-244
+ip-26-0-167-9'
++ NODELIST='ip-26-0-162-46
+ip-26-0-165-213
+ip-26-0-166-15
+ip-26-0-166-36
+ip-26-0-166-125
+ip-26-0-166-214
+ip-26-0-166-244
+ip-26-0-167-9'
+++ head -n1
+++ scontrol show hostnames 'ip-26-0-162-46,ip-26-0-165-213,ip-26-0-166-[15,36,125,214,244],ip-26-0-167-9'
++ export MASTER_NODE=ip-26-0-162-46
++ MASTER_NODE=ip-26-0-162-46
++ export MASTER_PORT=12356
++ MASTER_PORT=12356
++ export NNODES=8
++ NNODES=8
++ export GPUS_PER_NODE=8
++ GPUS_PER_NODE=8
++ export WORLD_SIZE=64
++ WORLD_SIZE=64
++ export CUDA_DEVICE_MAX_CONNECTIONS=1
++ CUDA_DEVICE_MAX_CONNECTIONS=1
++ export NCCL_DEBUG=INFO
++ NCCL_DEBUG=INFO
++ export NANOTRON_BENCHMARK=1
++ NANOTRON_BENCHMARK=1
++ export WANDB_MODE=disabled
++ WANDB_MODE=disabled
++ export TORCH_NCCL_ASYNC_ERROR_HANDLING=1
++ TORCH_NCCL_ASYNC_ERROR_HANDLING=1
++ export TORCH_DISTRIBUTED_DEBUG=DETAIL
++ TORCH_DISTRIBUTED_DEBUG=DETAIL
++ '[' -z '' ']'
++ echo '=== GPU Topology ==='
+=== GPU Topology ===
++ nvidia-smi topo -m
+	GPU0	GPU1	GPU2	GPU3	GPU4	GPU5	GPU6	GPU7	CPU Affinity	NUMA Affinity	GPU NUMA ID
+GPU0	 X 	NV18	NV18	NV18	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU1	NV18	 X 	NV18	NV18	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU2	NV18	NV18	 X 	NV18	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU3	NV18	NV18	NV18	 X 	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU4	NV18	NV18	NV18	NV18	 X 	NV18	NV18	NV18	48-95	1		N/A
+GPU5	NV18	NV18	NV18	NV18	NV18	 X 	NV18	NV18	48-95	1		N/A
+GPU6	NV18	NV18	NV18	NV18	NV18	NV18	 X 	NV18	48-95	1		N/A
+GPU7	NV18	NV18	NV18	NV18	NV18	NV18	NV18	 X 	48-95	1		N/A
+
+Legend:
+
+  X    = Self
+  SYS  = Connection traversing PCIe as well as the SMP interconnect between NUMA nodes (e.g., QPI/UPI)
+  NODE = Connection traversing PCIe as well as the interconnect between PCIe Host Bridges within a NUMA node
+  PHB  = Connection traversing PCIe as well as a PCIe Host Bridge (typically the CPU)
+  PXB  = Connection traversing multiple PCIe bridges (without traversing the PCIe Host Bridge)
+  PIX  = Connection traversing at most a single PCIe bridge
+  NV#  = Connection traversing a bonded set of # NVLinks
++ echo ==================
+==================
++ export SRUN_ALLOC_ARGS=
++ SRUN_ALLOC_ARGS=
++ echo 'Master node: ip-26-0-162-46'
+Master node: ip-26-0-162-46
++ echo 'All nodes: ip-26-0-162-46
+ip-26-0-165-213
+ip-26-0-166-15
+ip-26-0-166-36
+ip-26-0-166-125
+ip-26-0-166-214
+ip-26-0-166-244
+ip-26-0-167-9'
+All nodes: ip-26-0-162-46
+ip-26-0-165-213
+ip-26-0-166-15
+ip-26-0-166-36
+ip-26-0-166-125
+ip-26-0-166-214
+ip-26-0-166-244
+ip-26-0-167-9
++ echo 'World size: 64'
+World size: 64
++ '[' -n '' ']'
++ srun --wait=0 --kill-on-bad-exit=1 torchrun --nnodes=8 --nproc_per_node=8 --rdzv_id=14098831 --rdzv_backend=c10d --rdzv_endpoint=ip-26-0-162-46:12356 --max_restarts 0 --rdzv_conf timeout=60 /fsx/nouamane/projects/nanotron/run_train.py --config-file benchmark/configs/config_1.34G_dp4_tp2_pp8_acc32_mbs2_seq4096_zero1_tpmodeRED_vocab131k.yaml
+[2025-01-07 03:49:11,270] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-07 03:49:11,276] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-07 03:49:11,275] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-07 03:49:11,282] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-07 03:49:11,283] torch.distributed.run: [WARNING] 
+[2025-01-07 03:49:11,283] torch.distributed.run: [WARNING] *****************************************
+[2025-01-07 03:49:11,283] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-07 03:49:11,283] torch.distributed.run: [WARNING] *****************************************
+[2025-01-07 03:49:11,271] torch.distributed.run: [WARNING] 
+[2025-01-07 03:49:11,271] torch.distributed.run: [WARNING] *****************************************
+[2025-01-07 03:49:11,271] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-07 03:49:11,271] torch.distributed.run: [WARNING] *****************************************
+[2025-01-07 03:49:11,276] torch.distributed.run: [WARNING] 
+[2025-01-07 03:49:11,276] torch.distributed.run: [WARNING] *****************************************
+[2025-01-07 03:49:11,276] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-07 03:49:11,276] torch.distributed.run: [WARNING] *****************************************
+[2025-01-07 03:49:11,276] torch.distributed.run: [WARNING] 
+[2025-01-07 03:49:11,276] torch.distributed.run: [WARNING] *****************************************
+[2025-01-07 03:49:11,276] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-07 03:49:11,276] torch.distributed.run: [WARNING] *****************************************
+[2025-01-07 03:49:11,328] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-07 03:49:11,328] torch.distributed.run: [WARNING] 
+[2025-01-07 03:49:11,328] torch.distributed.run: [WARNING] *****************************************
+[2025-01-07 03:49:11,328] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-07 03:49:11,328] torch.distributed.run: [WARNING] *****************************************
+[2025-01-07 03:49:11,338] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-07 03:49:11,357] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-07 03:49:11,357] torch.distributed.run: [WARNING] 
+[2025-01-07 03:49:11,357] torch.distributed.run: [WARNING] *****************************************
+[2025-01-07 03:49:11,357] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-07 03:49:11,357] torch.distributed.run: [WARNING] *****************************************
+[2025-01-07 03:49:11,338] torch.distributed.run: [WARNING] 
+[2025-01-07 03:49:11,338] torch.distributed.run: [WARNING] *****************************************
+[2025-01-07 03:49:11,338] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-07 03:49:11,338] torch.distributed.run: [WARNING] *****************************************
+[2025-01-07 03:49:11,430] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-07 03:49:11,430] torch.distributed.run: [WARNING] 
+[2025-01-07 03:49:11,430] torch.distributed.run: [WARNING] *****************************************
+[2025-01-07 03:49:11,430] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-07 03:49:11,430] torch.distributed.run: [WARNING] *****************************************
+ip-26-0-162-46:702880:702880 [0] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-162-46:702880:702880 [0] NCCL INFO Bootstrap : Using enp71s0:26.0.162.46<0>
+ip-26-0-162-46:702880:702880 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-162-46:702880:702880 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-162-46:702880:702880 [0] NCCL INFO cudaDriverVersion 12020
+NCCL version 2.18.6+cuda12.1
+ip-26-0-162-46:702881:702881 [1] NCCL INFO cudaDriverVersion 12020
+ip-26-0-162-46:702886:702886 [6] NCCL INFO cudaDriverVersion 12020
+ip-26-0-162-46:702885:702885 [5] NCCL INFO cudaDriverVersion 12020
+ip-26-0-162-46:702884:702884 [4] NCCL INFO cudaDriverVersion 12020
+ip-26-0-162-46:702887:702887 [7] NCCL INFO cudaDriverVersion 12020
+ip-26-0-162-46:702882:702882 [2] NCCL INFO cudaDriverVersion 12020
+ip-26-0-162-46:702886:702886 [6] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-162-46:702881:702881 [1] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-162-46:702885:702885 [5] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-162-46:702884:702884 [4] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-167-9:1039537:1039537 [0] NCCL INFO cudaDriverVersion 12020
+ip-26-0-166-125:762818:762818 [3] NCCL INFO cudaDriverVersion 12020
+ip-26-0-166-125:762819:762819 [4] NCCL INFO cudaDriverVersion 12020
+ip-26-0-166-125:762820:762820 [5] NCCL INFO cudaDriverVersion 12020
+ip-26-0-166-125:762822:762822 [7] NCCL INFO cudaDriverVersion 12020
+ip-26-0-167-9:1039543:1039543 [6] NCCL INFO cudaDriverVersion 12020
+ip-26-0-167-9:1039541:1039541 [4] NCCL INFO cudaDriverVersion 12020
+ip-26-0-165-213:9625:9625 [0] NCCL INFO cudaDriverVersion 12020
+ip-26-0-162-46:702887:702887 [7] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-166-125:762816:762816 [1] NCCL INFO cudaDriverVersion 12020
+ip-26-0-162-46:702883:702883 [3] NCCL INFO cudaDriverVersion 12020
+ip-26-0-166-36:104430:104430 [1] NCCL INFO cudaDriverVersion 12020
+ip-26-0-166-36:104431:104431 [2] NCCL INFO cudaDriverVersion 12020
+ip-26-0-166-125:762821:762821 [6] NCCL INFO cudaDriverVersion 12020
+ip-26-0-166-125:762817:762817 [2] NCCL INFO cudaDriverVersion 12020
+ip-26-0-165-213:9632:9632 [7] NCCL INFO cudaDriverVersion 12020
+ip-26-0-166-125:762815:762815 [0] NCCL INFO cudaDriverVersion 12020
+ip-26-0-166-36:104435:104435 [6] NCCL INFO cudaDriverVersion 12020
+ip-26-0-166-244:897243:897243 [3] NCCL INFO cudaDriverVersion 12020
+ip-26-0-165-213:9630:9630 [5] NCCL INFO cudaDriverVersion 12020
+ip-26-0-165-213:9626:9626 [1] NCCL INFO cudaDriverVersion 12020
+ip-26-0-162-46:702882:702882 [2] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-166-36:104436:104436 [7] NCCL INFO cudaDriverVersion 12020
+ip-26-0-166-36:104434:104434 [5] NCCL INFO cudaDriverVersion 12020
+ip-26-0-166-244:897240:897240 [0] NCCL INFO cudaDriverVersion 12020
+ip-26-0-167-9:1039538:1039538 [1] NCCL INFO cudaDriverVersion 12020
+ip-26-0-165-213:9629:9629 [4] NCCL INFO cudaDriverVersion 12020
+ip-26-0-167-9:1039540:1039540 [3] NCCL INFO cudaDriverVersion 12020
+ip-26-0-167-9:1039542:1039542 [5] NCCL INFO cudaDriverVersion 12020
+ip-26-0-166-244:897246:897246 [6] NCCL INFO cudaDriverVersion 12020
+ip-26-0-167-9:1039544:1039544 [7] NCCL INFO cudaDriverVersion 12020
+ip-26-0-166-244:897247:897247 [7] NCCL INFO cudaDriverVersion 12020
+ip-26-0-165-213:9631:9631 [6] NCCL INFO cudaDriverVersion 12020
+ip-26-0-165-213:9627:9627 [2] NCCL INFO cudaDriverVersion 12020
+ip-26-0-165-213:9628:9628 [3] NCCL INFO cudaDriverVersion 12020
+ip-26-0-166-36:104432:104432 [3] NCCL INFO cudaDriverVersion 12020
+ip-26-0-166-36:104433:104433 [4] NCCL INFO cudaDriverVersion 12020
+ip-26-0-167-9:1039539:1039539 [2] NCCL INFO cudaDriverVersion 12020
+ip-26-0-166-244:897241:897241 [1] NCCL INFO cudaDriverVersion 12020
+ip-26-0-166-244:897244:897244 [4] NCCL INFO cudaDriverVersion 12020
+ip-26-0-166-214:624944:624944 [1] NCCL INFO cudaDriverVersion 12020
+ip-26-0-166-214:624950:624950 [7] NCCL INFO cudaDriverVersion 12020
+ip-26-0-166-36:104429:104429 [0] NCCL INFO cudaDriverVersion 12020
+ip-26-0-166-15:127296:127296 [6] NCCL INFO cudaDriverVersion 12020
+ip-26-0-166-15:127297:127297 [7] NCCL INFO cudaDriverVersion 12020
+ip-26-0-166-15:127295:127295 [5] NCCL INFO cudaDriverVersion 12020
+ip-26-0-166-15:127292:127292 [2] NCCL INFO cudaDriverVersion 12020
+ip-26-0-166-244:897245:897245 [5] NCCL INFO cudaDriverVersion 12020
+ip-26-0-166-244:897242:897242 [2] NCCL INFO cudaDriverVersion 12020
+ip-26-0-166-214:624948:624948 [5] NCCL INFO cudaDriverVersion 12020
+ip-26-0-166-214:624947:624947 [4] NCCL INFO cudaDriverVersion 12020
+ip-26-0-166-214:624949:624949 [6] NCCL INFO cudaDriverVersion 12020
+ip-26-0-166-214:624943:624943 [0] NCCL INFO cudaDriverVersion 12020
+ip-26-0-166-15:127290:127290 [0] NCCL INFO cudaDriverVersion 12020
+ip-26-0-166-15:127293:127293 [3] NCCL INFO cudaDriverVersion 12020
+ip-26-0-166-214:624946:624946 [3] NCCL INFO cudaDriverVersion 12020
+ip-26-0-166-214:624945:624945 [2] NCCL INFO cudaDriverVersion 12020
+ip-26-0-166-15:127294:127294 [4] NCCL INFO cudaDriverVersion 12020
+ip-26-0-166-15:127291:127291 [1] NCCL INFO cudaDriverVersion 12020
+ip-26-0-162-46:702883:702883 [3] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-167-9:1039537:1039537 [0] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-167-9:1039540:1039540 [3] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-167-9:1039538:1039538 [1] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-165-213:9632:9632 [7] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-166-125:762815:762815 [0] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-167-9:1039542:1039542 [5] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-167-9:1039539:1039539 [2] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-165-213:9627:9627 [2] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-165-213:9625:9625 [0] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-166-36:104430:104430 [1] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-166-125:762816:762816 [1] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-166-125:762822:762822 [7] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-166-125:762819:762819 [4] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-166-36:104436:104436 [7] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-166-36:104432:104432 [3] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-166-125:762818:762818 [3] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-166-36:104433:104433 [4] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-166-36:104431:104431 [2] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-167-9:1039541:1039541 [4] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-167-9:1039544:1039544 [7] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-166-125:762821:762821 [6] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-166-36:104435:104435 [6] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-166-244:897247:897247 [7] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-166-244:897243:897243 [3] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-166-244:897244:897244 [4] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-166-244:897240:897240 [0] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-166-125:762820:762820 [5] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-167-9:1039543:1039543 [6] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-166-125:762817:762817 [2] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-165-213:9626:9626 [1] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-166-244:897241:897241 [1] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-165-213:9630:9630 [5] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-165-213:9631:9631 [6] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-166-244:897245:897245 [5] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-166-244:897242:897242 [2] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-166-244:897246:897246 [6] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-165-213:9628:9628 [3] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-166-36:104429:104429 [0] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-166-15:127297:127297 [7] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-166-15:127296:127296 [6] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-165-213:9629:9629 [4] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-166-214:624948:624948 [5] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-166-214:624950:624950 [7] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-166-36:104434:104434 [5] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-166-214:624944:624944 [1] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-166-15:127292:127292 [2] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-166-15:127295:127295 [5] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-166-15:127293:127293 [3] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-166-214:624947:624947 [4] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-166-214:624949:624949 [6] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-166-214:624943:624943 [0] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-166-15:127294:127294 [4] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-166-214:624946:624946 [3] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-166-15:127290:127290 [0] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-166-15:127291:127291 [1] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-166-214:624945:624945 [2] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-162-46:702881:702881 [1] NCCL INFO Bootstrap : Using enp71s0:26.0.162.46<0>
+ip-26-0-162-46:702886:702886 [6] NCCL INFO Bootstrap : Using enp71s0:26.0.162.46<0>
+ip-26-0-162-46:702885:702885 [5] NCCL INFO Bootstrap : Using enp71s0:26.0.162.46<0>
+ip-26-0-162-46:702884:702884 [4] NCCL INFO Bootstrap : Using enp71s0:26.0.162.46<0>
+ip-26-0-162-46:702887:702887 [7] NCCL INFO Bootstrap : Using enp71s0:26.0.162.46<0>
+ip-26-0-162-46:702882:702882 [2] NCCL INFO Bootstrap : Using enp71s0:26.0.162.46<0>
+ip-26-0-162-46:702883:702883 [3] NCCL INFO Bootstrap : Using enp71s0:26.0.162.46<0>
+ip-26-0-162-46:702885:702885 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-162-46:702886:702886 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-162-46:702881:702881 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-162-46:702884:702884 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-162-46:702885:702885 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-162-46:702886:702886 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-162-46:702881:702881 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-162-46:702884:702884 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-162-46:702887:702887 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-162-46:702887:702887 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-162-46:702882:702882 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-162-46:702882:702882 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-162-46:702883:702883 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-162-46:702883:702883 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-166-214:624948:624948 [5] NCCL INFO Bootstrap : Using enp71s0:26.0.166.214<0>
+ip-26-0-166-214:624950:624950 [7] NCCL INFO Bootstrap : Using enp71s0:26.0.166.214<0>
+ip-26-0-166-214:624944:624944 [1] NCCL INFO Bootstrap : Using enp71s0:26.0.166.214<0>
+ip-26-0-165-213:9632:9632 [7] NCCL INFO Bootstrap : Using enp71s0:26.0.165.213<0>
+ip-26-0-165-213:9627:9627 [2] NCCL INFO Bootstrap : Using enp71s0:26.0.165.213<0>
+ip-26-0-165-213:9625:9625 [0] NCCL INFO Bootstrap : Using enp71s0:26.0.165.213<0>
+ip-26-0-166-214:624947:624947 [4] NCCL INFO Bootstrap : Using enp71s0:26.0.166.214<0>
+ip-26-0-166-214:624943:624943 [0] NCCL INFO Bootstrap : Using enp71s0:26.0.166.214<0>
+ip-26-0-166-214:624949:624949 [6] NCCL INFO Bootstrap : Using enp71s0:26.0.166.214<0>
+ip-26-0-166-214:624946:624946 [3] NCCL INFO Bootstrap : Using enp71s0:26.0.166.214<0>
+ip-26-0-165-213:9626:9626 [1] NCCL INFO Bootstrap : Using enp71s0:26.0.165.213<0>
+ip-26-0-166-214:624945:624945 [2] NCCL INFO Bootstrap : Using enp71s0:26.0.166.214<0>
+ip-26-0-165-213:9630:9630 [5] NCCL INFO Bootstrap : Using enp71s0:26.0.165.213<0>
+ip-26-0-165-213:9628:9628 [3] NCCL INFO Bootstrap : Using enp71s0:26.0.165.213<0>
+ip-26-0-165-213:9631:9631 [6] NCCL INFO Bootstrap : Using enp71s0:26.0.165.213<0>
+ip-26-0-165-213:9629:9629 [4] NCCL INFO Bootstrap : Using enp71s0:26.0.165.213<0>
+ip-26-0-167-9:1039537:1039537 [0] NCCL INFO Bootstrap : Using enp71s0:26.0.167.9<0>
+ip-26-0-167-9:1039540:1039540 [3] NCCL INFO Bootstrap : Using enp71s0:26.0.167.9<0>
+ip-26-0-167-9:1039542:1039542 [5] NCCL INFO Bootstrap : Using enp71s0:26.0.167.9<0>
+ip-26-0-167-9:1039541:1039541 [4] NCCL INFO Bootstrap : Using enp71s0:26.0.167.9<0>
+ip-26-0-167-9:1039543:1039543 [6] NCCL INFO Bootstrap : Using enp71s0:26.0.167.9<0>
+ip-26-0-167-9:1039544:1039544 [7] NCCL INFO Bootstrap : Using enp71s0:26.0.167.9<0>
+ip-26-0-167-9:1039538:1039538 [1] NCCL INFO Bootstrap : Using enp71s0:26.0.167.9<0>
+ip-26-0-167-9:1039539:1039539 [2] NCCL INFO Bootstrap : Using enp71s0:26.0.167.9<0>
+ip-26-0-166-36:104430:104430 [1] NCCL INFO Bootstrap : Using enp71s0:26.0.166.36<0>
+ip-26-0-166-36:104431:104431 [2] NCCL INFO Bootstrap : Using enp71s0:26.0.166.36<0>
+ip-26-0-166-36:104436:104436 [7] NCCL INFO Bootstrap : Using enp71s0:26.0.166.36<0>
+ip-26-0-166-36:104435:104435 [6] NCCL INFO Bootstrap : Using enp71s0:26.0.166.36<0>
+ip-26-0-166-36:104432:104432 [3] NCCL INFO Bootstrap : Using enp71s0:26.0.166.36<0>
+ip-26-0-166-36:104433:104433 [4] NCCL INFO Bootstrap : Using enp71s0:26.0.166.36<0>
+ip-26-0-166-36:104429:104429 [0] NCCL INFO Bootstrap : Using enp71s0:26.0.166.36<0>
+ip-26-0-166-36:104434:104434 [5] NCCL INFO Bootstrap : Using enp71s0:26.0.166.36<0>
+ip-26-0-166-15:127297:127297 [7] NCCL INFO Bootstrap : Using enp71s0:26.0.166.15<0>
+ip-26-0-166-15:127296:127296 [6] NCCL INFO Bootstrap : Using enp71s0:26.0.166.15<0>
+ip-26-0-166-15:127292:127292 [2] NCCL INFO Bootstrap : Using enp71s0:26.0.166.15<0>
+ip-26-0-166-15:127295:127295 [5] NCCL INFO Bootstrap : Using enp71s0:26.0.166.15<0>
+ip-26-0-166-15:127291:127291 [1] NCCL INFO Bootstrap : Using enp71s0:26.0.166.15<0>
+ip-26-0-166-15:127290:127290 [0] NCCL INFO Bootstrap : Using enp71s0:26.0.166.15<0>
+ip-26-0-166-15:127293:127293 [3] NCCL INFO Bootstrap : Using enp71s0:26.0.166.15<0>
+ip-26-0-166-15:127294:127294 [4] NCCL INFO Bootstrap : Using enp71s0:26.0.166.15<0>
+ip-26-0-166-214:624948:624948 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-166-214:624950:624950 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-166-214:624950:624950 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-166-214:624948:624948 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-165-213:9625:9625 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-165-213:9632:9632 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-165-213:9625:9625 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-165-213:9632:9632 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-165-213:9629:9629 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-165-213:9628:9628 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-165-213:9627:9627 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-165-213:9628:9628 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-165-213:9627:9627 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-165-213:9629:9629 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-166-214:624949:624949 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-166-214:624947:624947 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-166-214:624943:624943 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-166-214:624947:624947 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-166-214:624943:624943 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-166-214:624949:624949 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-165-213:9626:9626 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-165-213:9626:9626 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-166-214:624944:624944 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-166-214:624944:624944 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-165-213:9630:9630 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-165-213:9631:9631 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-165-213:9630:9630 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-165-213:9631:9631 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-167-9:1039537:1039537 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-167-9:1039537:1039537 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-166-214:624946:624946 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-166-214:624945:624945 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-166-214:624946:624946 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-167-9:1039540:1039540 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-167-9:1039540:1039540 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-166-214:624945:624945 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-167-9:1039542:1039542 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-167-9:1039542:1039542 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-167-9:1039539:1039539 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-167-9:1039539:1039539 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-167-9:1039544:1039544 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-167-9:1039543:1039543 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-167-9:1039544:1039544 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-167-9:1039543:1039543 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-166-36:104430:104430 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-166-36:104430:104430 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-167-9:1039541:1039541 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-167-9:1039541:1039541 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-167-9:1039538:1039538 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-167-9:1039538:1039538 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-166-36:104429:104429 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-166-36:104436:104436 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-166-36:104435:104435 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-166-36:104431:104431 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-166-36:104432:104432 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-166-36:104433:104433 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-166-36:104429:104429 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-166-36:104431:104431 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-166-36:104435:104435 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-166-36:104436:104436 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-166-15:127297:127297 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-166-15:127296:127296 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-166-15:127297:127297 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-166-15:127296:127296 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-166-36:104432:104432 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-166-36:104433:104433 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-166-36:104434:104434 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-166-36:104434:104434 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-166-15:127292:127292 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-166-15:127292:127292 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-166-15:127290:127290 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-166-15:127291:127291 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-166-15:127294:127294 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-166-15:127291:127291 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-166-15:127290:127290 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-166-15:127294:127294 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-166-244:897247:897247 [7] NCCL INFO Bootstrap : Using enp71s0:26.0.166.244<0>
+ip-26-0-166-15:127295:127295 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-166-15:127293:127293 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-166-244:897243:897243 [3] NCCL INFO Bootstrap : Using enp71s0:26.0.166.244<0>
+ip-26-0-166-244:897244:897244 [4] NCCL INFO Bootstrap : Using enp71s0:26.0.166.244<0>
+ip-26-0-166-125:762818:762818 [3] NCCL INFO Bootstrap : Using enp72s0:26.0.170.99<0>
+ip-26-0-166-125:762815:762815 [0] NCCL INFO Bootstrap : Using enp72s0:26.0.170.99<0>
+ip-26-0-166-244:897245:897245 [5] NCCL INFO Bootstrap : Using enp71s0:26.0.166.244<0>
+ip-26-0-166-244:897246:897246 [6] NCCL INFO Bootstrap : Using enp71s0:26.0.166.244<0>
+ip-26-0-166-244:897241:897241 [1] NCCL INFO Bootstrap : Using enp71s0:26.0.166.244<0>
+ip-26-0-166-244:897240:897240 [0] NCCL INFO Bootstrap : Using enp71s0:26.0.166.244<0>
+ip-26-0-166-244:897242:897242 [2] NCCL INFO Bootstrap : Using enp71s0:26.0.166.244<0>
+ip-26-0-166-15:127293:127293 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-166-15:127295:127295 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-166-125:762816:762816 [1] NCCL INFO Bootstrap : Using enp72s0:26.0.170.99<0>
+ip-26-0-166-125:762822:762822 [7] NCCL INFO Bootstrap : Using enp72s0:26.0.170.99<0>
+ip-26-0-166-125:762821:762821 [6] NCCL INFO Bootstrap : Using enp72s0:26.0.170.99<0>
+ip-26-0-166-125:762819:762819 [4] NCCL INFO Bootstrap : Using enp72s0:26.0.170.99<0>
+ip-26-0-166-125:762820:762820 [5] NCCL INFO Bootstrap : Using enp72s0:26.0.170.99<0>
+ip-26-0-166-125:762817:762817 [2] NCCL INFO Bootstrap : Using enp72s0:26.0.170.99<0>
+ip-26-0-166-244:897247:897247 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-166-244:897247:897247 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-166-244:897242:897242 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-166-244:897244:897244 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-166-244:897240:897240 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-166-244:897242:897242 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-166-244:897244:897244 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-166-244:897240:897240 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-166-125:762818:762818 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-166-125:762818:762818 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-166-244:897243:897243 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-166-244:897243:897243 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-166-125:762815:762815 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-166-125:762815:762815 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-166-125:762822:762822 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-166-125:762819:762819 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-166-125:762816:762816 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-166-125:762816:762816 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-166-125:762819:762819 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-166-125:762822:762822 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-166-244:897245:897245 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-166-244:897245:897245 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-166-125:762821:762821 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-166-125:762821:762821 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-166-244:897246:897246 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-166-244:897246:897246 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-166-244:897241:897241 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-166-244:897241:897241 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-166-125:762817:762817 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-166-125:762820:762820 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-166-125:762817:762817 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-166-125:762820:762820 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-162-46:702880:703042 [0] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-162-46:702880:703042 [0] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-162-46:702880:703042 [0] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-162-46:702880:703042 [0] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-162-46:702880:703042 [0] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-162-46:702880:703042 [0] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-162-46:702880:703042 [0] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-162-46:702880:703042 [0] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-162-46:702880:703042 [0] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-166-244:897247:897404 [7] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-166-244:897247:897404 [7] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-165-213:9625:9787 [0] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-165-213:9625:9787 [0] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-166-214:624948:625104 [5] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-166-214:624948:625104 [5] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-166-36:104430:104589 [1] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-166-36:104430:104589 [1] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-165-213:9626:9790 [1] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-165-213:9626:9790 [1] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-166-244:897247:897404 [7] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-166-244:897247:897404 [7] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-166-244:897247:897404 [7] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-166-244:897247:897404 [7] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-166-244:897247:897404 [7] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-166-244:897247:897404 [7] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-166-244:897247:897404 [7] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-166-36:104430:104589 [1] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-166-36:104430:104589 [1] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-166-36:104430:104589 [1] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-166-36:104430:104589 [1] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-166-36:104430:104589 [1] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-166-36:104430:104589 [1] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-166-36:104430:104589 [1] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-165-213:9625:9787 [0] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-165-213:9625:9787 [0] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-165-213:9625:9787 [0] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-165-213:9625:9787 [0] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-165-213:9625:9787 [0] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-165-213:9625:9787 [0] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-165-213:9625:9787 [0] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-166-214:624948:625104 [5] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-166-214:624948:625104 [5] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-166-214:624948:625104 [5] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-166-214:624948:625104 [5] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-166-214:624948:625104 [5] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-166-214:624948:625104 [5] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-166-214:624948:625104 [5] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-162-46:702881:703043 [1] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-162-46:702881:703043 [1] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-162-46:702883:703044 [3] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-162-46:702883:703044 [3] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-165-213:9626:9790 [1] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-165-213:9626:9790 [1] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-165-213:9626:9790 [1] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-165-213:9626:9790 [1] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-165-213:9626:9790 [1] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-165-213:9626:9790 [1] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-165-213:9626:9790 [1] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-166-214:624944:625106 [1] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-166-214:624944:625106 [1] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-166-15:127295:127452 [5] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-166-15:127295:127452 [5] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-166-36:104431:104593 [2] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-166-36:104431:104593 [2] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-167-9:1039543:1039699 [6] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-167-9:1039543:1039699 [6] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-166-15:127294:127457 [4] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-166-15:127294:127457 [4] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-162-46:702887:703045 [7] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-162-46:702887:703045 [7] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-166-36:104433:104590 [4] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-166-36:104433:104590 [4] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-166-214:624947:625105 [4] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-166-214:624947:625105 [4] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-167-9:1039543:1039699 [6] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-166-214:624944:625106 [1] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-167-9:1039543:1039699 [6] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-166-214:624944:625106 [1] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-167-9:1039543:1039699 [6] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-167-9:1039543:1039699 [6] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-167-9:1039543:1039699 [6] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-166-214:624944:625106 [1] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-167-9:1039543:1039699 [6] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-167-9:1039543:1039699 [6] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-166-214:624944:625106 [1] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-166-214:624944:625106 [1] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-166-214:624944:625106 [1] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-166-214:624944:625106 [1] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-165-213:9629:9793 [4] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-165-213:9629:9793 [4] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-166-15:127296:127451 [6] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-166-15:127296:127451 [6] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-166-244:897245:897401 [5] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-166-244:897245:897401 [5] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-162-46:702881:703043 [1] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-162-46:702881:703043 [1] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-162-46:702881:703043 [1] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-162-46:702881:703043 [1] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-162-46:702881:703043 [1] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-162-46:702881:703043 [1] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-162-46:702881:703043 [1] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-162-46:702883:703044 [3] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-162-46:702883:703044 [3] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-162-46:702883:703044 [3] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-162-46:702883:703044 [3] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-162-46:702883:703044 [3] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-162-46:702883:703044 [3] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-162-46:702883:703044 [3] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-162-46:702880:703042 [0] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-166-244:897242:897400 [2] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-166-244:897242:897400 [2] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-166-36:104431:104593 [2] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-166-36:104431:104593 [2] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-166-36:104431:104593 [2] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-166-36:104431:104593 [2] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-166-15:127295:127452 [5] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-166-15:127295:127452 [5] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-166-36:104431:104593 [2] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-166-36:104431:104593 [2] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-166-36:104431:104593 [2] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-166-15:127295:127452 [5] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-166-15:127295:127452 [5] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-166-15:127295:127452 [5] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-166-15:127295:127452 [5] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-166-15:127295:127452 [5] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-165-213:9629:9793 [4] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-165-213:9629:9793 [4] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-165-213:9629:9793 [4] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-165-213:9629:9793 [4] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-165-213:9629:9793 [4] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-165-213:9629:9793 [4] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-165-213:9629:9793 [4] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-166-36:104433:104590 [4] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-166-36:104433:104590 [4] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-166-36:104433:104590 [4] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-166-36:104433:104590 [4] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-166-36:104433:104590 [4] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-166-36:104433:104590 [4] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-166-36:104433:104590 [4] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-162-46:702886:703049 [6] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-162-46:702886:703049 [6] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-166-125:762819:762982 [4] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-166-125:762819:762982 [4] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-166-15:127294:127457 [4] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-166-15:127294:127457 [4] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-166-15:127294:127457 [4] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-166-15:127294:127457 [4] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-166-15:127294:127457 [4] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-166-15:127294:127457 [4] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-166-15:127294:127457 [4] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-162-46:702887:703045 [7] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-162-46:702887:703045 [7] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-162-46:702887:703045 [7] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-162-46:702887:703045 [7] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-162-46:702887:703045 [7] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-162-46:702887:703045 [7] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-162-46:702887:703045 [7] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-166-244:897245:897401 [5] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-166-244:897245:897401 [5] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-166-244:897245:897401 [5] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-166-244:897245:897401 [5] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-166-244:897245:897401 [5] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-166-244:897245:897401 [5] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-166-244:897245:897401 [5] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-166-36:104429:104594 [0] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-166-36:104429:104594 [0] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-166-36:104432:104592 [3] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-166-36:104432:104592 [3] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-166-244:897242:897400 [2] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-166-244:897242:897400 [2] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-166-244:897242:897400 [2] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-166-244:897242:897400 [2] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-166-244:897242:897400 [2] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-166-244:897242:897400 [2] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-166-244:897242:897400 [2] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-167-9:1039540:1039705 [3] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-167-9:1039540:1039705 [3] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-166-125:762820:762981 [5] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-166-125:762820:762981 [5] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-166-125:762819:762982 [4] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-166-125:762819:762982 [4] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-166-125:762819:762982 [4] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-166-125:762819:762982 [4] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-166-125:762819:762982 [4] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-166-125:762819:762982 [4] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-166-125:762819:762982 [4] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-162-46:702885:703048 [5] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-162-46:702885:703048 [5] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-166-214:624947:625105 [4] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-166-214:624947:625105 [4] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-166-214:624947:625105 [4] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-166-214:624947:625105 [4] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-166-214:624947:625105 [4] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-166-214:624947:625105 [4] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-166-214:624947:625105 [4] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-166-15:127296:127451 [6] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-166-15:127296:127451 [6] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-166-15:127296:127451 [6] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-166-15:127296:127451 [6] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-166-15:127296:127451 [6] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-166-15:127296:127451 [6] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-166-15:127296:127451 [6] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-166-125:762822:762984 [7] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-166-125:762822:762984 [7] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-166-214:624946:625107 [3] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-166-214:624946:625107 [3] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-165-213:9631:9788 [6] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-165-213:9631:9788 [6] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-166-244:897241:897402 [1] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-166-244:897241:897402 [1] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-166-214:624949:625110 [6] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-166-214:624949:625110 [6] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-165-213:9630:9789 [5] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-165-213:9630:9789 [5] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-162-46:702884:703047 [4] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-162-46:702884:703047 [4] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-165-213:9627:9791 [2] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-165-213:9627:9791 [2] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-166-125:762815:762983 [0] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-166-125:762815:762983 [0] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-166-36:104429:104594 [0] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-166-36:104429:104594 [0] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-166-36:104429:104594 [0] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-166-36:104429:104594 [0] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-166-36:104429:104594 [0] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-166-36:104429:104594 [0] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-166-36:104429:104594 [0] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-166-36:104432:104592 [3] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-166-36:104432:104592 [3] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-166-36:104432:104592 [3] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-166-36:104432:104592 [3] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-166-36:104432:104592 [3] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-166-36:104432:104592 [3] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-166-36:104432:104592 [3] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-166-15:127292:127458 [2] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-166-15:127292:127458 [2] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-166-36:104434:104591 [5] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-166-36:104434:104591 [5] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-166-244:897241:897402 [1] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-166-244:897241:897402 [1] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-162-46:702886:703049 [6] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-166-244:897241:897402 [1] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-166-244:897241:897402 [1] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-162-46:702886:703049 [6] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-162-46:702886:703049 [6] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-162-46:702886:703049 [6] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-162-46:702886:703049 [6] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-166-244:897241:897402 [1] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-162-46:702886:703049 [6] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-162-46:702886:703049 [6] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-166-244:897241:897402 [1] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-166-244:897241:897402 [1] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-166-125:762820:762981 [5] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-166-125:762820:762981 [5] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-166-125:762820:762981 [5] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-166-125:762820:762981 [5] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-166-125:762820:762981 [5] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-166-125:762820:762981 [5] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-166-125:762820:762981 [5] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-167-9:1039544:1039702 [7] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-167-9:1039544:1039702 [7] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-165-213:9631:9788 [6] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-165-213:9631:9788 [6] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-167-9:1039539:1039698 [2] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-167-9:1039539:1039698 [2] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-165-213:9631:9788 [6] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-167-9:1039540:1039705 [3] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-165-213:9631:9788 [6] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-165-213:9631:9788 [6] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-165-213:9631:9788 [6] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-165-213:9631:9788 [6] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-167-9:1039540:1039705 [3] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-167-9:1039538:1039704 [1] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-167-9:1039538:1039704 [1] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-167-9:1039540:1039705 [3] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-167-9:1039540:1039705 [3] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-167-9:1039540:1039705 [3] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-167-9:1039540:1039705 [3] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-167-9:1039540:1039705 [3] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-167-9:1039542:1039701 [5] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-167-9:1039542:1039701 [5] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-166-15:127293:127454 [3] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-166-15:127293:127454 [3] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-166-125:762818:762985 [3] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-166-125:762818:762985 [3] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-166-15:127297:127453 [7] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-166-15:127297:127453 [7] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-166-214:624946:625107 [3] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-166-214:624946:625107 [3] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-166-214:624946:625107 [3] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-166-214:624946:625107 [3] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-166-214:624946:625107 [3] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-166-214:624946:625107 [3] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-166-214:624946:625107 [3] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-166-214:624949:625110 [6] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-166-214:624949:625110 [6] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-166-214:624949:625110 [6] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-166-214:624949:625110 [6] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-166-214:624949:625110 [6] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-166-214:624949:625110 [6] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-166-214:624949:625110 [6] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-162-46:702882:703046 [2] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-162-46:702882:703046 [2] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-166-125:762817:762986 [2] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-166-125:762817:762986 [2] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-166-36:104435:104595 [6] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-166-36:104435:104595 [6] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-166-15:127292:127458 [2] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-166-15:127292:127458 [2] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-166-15:127292:127458 [2] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-166-15:127292:127458 [2] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-166-15:127292:127458 [2] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-165-213:9630:9789 [5] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-165-213:9630:9789 [5] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-166-15:127292:127458 [2] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-166-15:127292:127458 [2] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-165-213:9630:9789 [5] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-165-213:9630:9789 [5] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-165-213:9630:9789 [5] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-165-213:9630:9789 [5] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-165-213:9630:9789 [5] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-166-244:897244:897407 [4] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-166-244:897244:897407 [4] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-166-214:624950:625109 [7] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-166-214:624950:625109 [7] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-165-213:9632:9792 [7] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-165-213:9632:9792 [7] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-166-15:127290:127456 [0] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-166-15:127290:127456 [0] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-166-244:897246:897405 [6] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-166-244:897246:897405 [6] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-166-244:897243:897403 [3] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-166-244:897243:897403 [3] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-166-214:624945:625111 [2] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-166-214:624945:625111 [2] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-165-213:9628:9786 [3] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-165-213:9628:9786 [3] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-167-9:1039537:1039703 [0] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-167-9:1039537:1039703 [0] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-162-46:702885:703048 [5] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-162-46:702885:703048 [5] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-162-46:702885:703048 [5] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-162-46:702885:703048 [5] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-162-46:702885:703048 [5] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-162-46:702885:703048 [5] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-162-46:702885:703048 [5] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-166-125:762822:762984 [7] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-166-125:762822:762984 [7] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-166-125:762822:762984 [7] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-166-36:104436:104596 [7] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-166-36:104436:104596 [7] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-166-125:762822:762984 [7] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-166-125:762822:762984 [7] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-166-125:762822:762984 [7] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-166-125:762822:762984 [7] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-166-125:762816:762987 [1] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-166-125:762816:762987 [1] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-166-15:127293:127454 [3] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-166-15:127293:127454 [3] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-162-46:702884:703047 [4] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-162-46:702884:703047 [4] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-166-15:127293:127454 [3] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-162-46:702884:703047 [4] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-162-46:702884:703047 [4] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-166-15:127293:127454 [3] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-166-15:127293:127454 [3] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-162-46:702884:703047 [4] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-166-15:127293:127454 [3] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-162-46:702884:703047 [4] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-162-46:702884:703047 [4] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-166-15:127293:127454 [3] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-166-36:104434:104591 [5] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-166-36:104434:104591 [5] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-166-36:104434:104591 [5] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-166-36:104434:104591 [5] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-166-36:104434:104591 [5] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-166-36:104434:104591 [5] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-166-36:104434:104591 [5] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-166-244:897240:897406 [0] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-166-244:897240:897406 [0] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-166-15:127291:127455 [1] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-166-15:127291:127455 [1] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-165-213:9627:9791 [2] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-165-213:9627:9791 [2] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-165-213:9627:9791 [2] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-166-244:897244:897407 [4] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-166-244:897244:897407 [4] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-165-213:9627:9791 [2] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-165-213:9627:9791 [2] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-165-213:9627:9791 [2] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-166-244:897244:897407 [4] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-165-213:9627:9791 [2] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-166-244:897244:897407 [4] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-166-244:897244:897407 [4] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-166-244:897244:897407 [4] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-166-244:897244:897407 [4] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-166-244:897243:897403 [3] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-166-244:897243:897403 [3] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-166-244:897243:897403 [3] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-166-244:897243:897403 [3] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-166-244:897243:897403 [3] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-166-244:897243:897403 [3] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-166-244:897243:897403 [3] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-166-15:127297:127453 [7] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-166-15:127297:127453 [7] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-166-15:127297:127453 [7] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-166-15:127297:127453 [7] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-166-15:127297:127453 [7] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-166-15:127297:127453 [7] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-166-15:127297:127453 [7] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-166-125:762815:762983 [0] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-166-125:762815:762983 [0] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-166-125:762815:762983 [0] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-166-125:762815:762983 [0] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-166-125:762815:762983 [0] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-166-125:762815:762983 [0] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-166-125:762815:762983 [0] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-166-36:104435:104595 [6] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-166-36:104435:104595 [6] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-166-36:104435:104595 [6] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-166-36:104435:104595 [6] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-166-36:104435:104595 [6] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-166-36:104435:104595 [6] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-166-36:104435:104595 [6] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-166-214:624950:625109 [7] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-166-214:624950:625109 [7] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-166-214:624950:625109 [7] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-166-214:624950:625109 [7] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-166-214:624950:625109 [7] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-166-214:624950:625109 [7] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-166-214:624950:625109 [7] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-166-214:624945:625111 [2] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-166-214:624945:625111 [2] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-166-214:624945:625111 [2] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-166-214:624945:625111 [2] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-166-214:624945:625111 [2] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-166-214:624945:625111 [2] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-166-214:624945:625111 [2] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-162-46:702882:703046 [2] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-162-46:702882:703046 [2] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-162-46:702882:703046 [2] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-162-46:702882:703046 [2] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-162-46:702882:703046 [2] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-162-46:702882:703046 [2] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-162-46:702882:703046 [2] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-166-36:104436:104596 [7] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-166-36:104436:104596 [7] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-166-36:104436:104596 [7] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-166-36:104436:104596 [7] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-166-36:104436:104596 [7] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-166-36:104436:104596 [7] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-166-36:104436:104596 [7] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-166-15:127290:127456 [0] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-166-15:127290:127456 [0] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-166-15:127290:127456 [0] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-166-15:127290:127456 [0] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-166-15:127290:127456 [0] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-166-15:127290:127456 [0] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-166-15:127290:127456 [0] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-166-125:762818:762985 [3] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-166-125:762818:762985 [3] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-166-125:762818:762985 [3] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-166-125:762818:762985 [3] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-166-125:762818:762985 [3] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-166-125:762818:762985 [3] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-166-125:762818:762985 [3] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-165-213:9632:9792 [7] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-165-213:9632:9792 [7] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-165-213:9632:9792 [7] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-165-213:9632:9792 [7] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-165-213:9632:9792 [7] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-165-213:9632:9792 [7] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-165-213:9632:9792 [7] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-166-125:762817:762986 [2] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-166-125:762817:762986 [2] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-166-125:762817:762986 [2] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-166-125:762817:762986 [2] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-166-125:762817:762986 [2] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-166-125:762817:762986 [2] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-166-125:762817:762986 [2] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-165-213:9628:9786 [3] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-165-213:9628:9786 [3] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-165-213:9628:9786 [3] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-165-213:9628:9786 [3] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-165-213:9628:9786 [3] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-165-213:9628:9786 [3] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-165-213:9628:9786 [3] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-166-244:897240:897406 [0] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-166-244:897240:897406 [0] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-166-244:897240:897406 [0] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-166-244:897240:897406 [0] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-166-244:897240:897406 [0] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-166-244:897240:897406 [0] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-166-244:897240:897406 [0] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-166-244:897246:897405 [6] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-166-244:897246:897405 [6] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-166-244:897246:897405 [6] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-166-244:897246:897405 [6] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-166-244:897246:897405 [6] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-166-244:897246:897405 [6] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-166-244:897246:897405 [6] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-167-9:1039544:1039702 [7] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-167-9:1039544:1039702 [7] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-167-9:1039544:1039702 [7] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-167-9:1039544:1039702 [7] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-167-9:1039544:1039702 [7] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-167-9:1039544:1039702 [7] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-167-9:1039544:1039702 [7] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-166-15:127291:127455 [1] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-166-15:127291:127455 [1] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-166-15:127291:127455 [1] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-166-15:127291:127455 [1] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-166-15:127291:127455 [1] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-166-15:127291:127455 [1] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-166-15:127291:127455 [1] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-166-125:762816:762987 [1] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-167-9:1039539:1039698 [2] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-166-125:762816:762987 [1] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-167-9:1039539:1039698 [2] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-166-125:762816:762987 [1] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-166-125:762816:762987 [1] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-166-125:762816:762987 [1] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-167-9:1039539:1039698 [2] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-167-9:1039539:1039698 [2] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-167-9:1039539:1039698 [2] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-166-125:762816:762987 [1] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-166-125:762816:762987 [1] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-167-9:1039539:1039698 [2] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-167-9:1039539:1039698 [2] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-167-9:1039542:1039701 [5] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-167-9:1039542:1039701 [5] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-167-9:1039542:1039701 [5] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-167-9:1039542:1039701 [5] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-167-9:1039542:1039701 [5] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-167-9:1039538:1039704 [1] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-167-9:1039538:1039704 [1] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-167-9:1039542:1039701 [5] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-167-9:1039542:1039701 [5] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-167-9:1039538:1039704 [1] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-167-9:1039538:1039704 [1] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-167-9:1039538:1039704 [1] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-167-9:1039538:1039704 [1] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-167-9:1039538:1039704 [1] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-167-9:1039537:1039703 [0] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-167-9:1039537:1039703 [0] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-167-9:1039537:1039703 [0] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-167-9:1039537:1039703 [0] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-167-9:1039537:1039703 [0] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-167-9:1039537:1039703 [0] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-167-9:1039537:1039703 [0] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-166-125:762821:762988 [6] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-166-125:762821:762988 [6] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-166-214:624943:625108 [0] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-166-214:624943:625108 [0] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-166-125:762821:762988 [6] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-166-125:762821:762988 [6] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-166-125:762821:762988 [6] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-166-125:762821:762988 [6] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-166-125:762821:762988 [6] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-166-125:762821:762988 [6] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-166-125:762821:762988 [6] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-166-214:624943:625108 [0] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-166-214:624943:625108 [0] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-166-214:624943:625108 [0] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-166-214:624943:625108 [0] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-166-214:624943:625108 [0] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-166-214:624943:625108 [0] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-166-214:624943:625108 [0] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-166-244:897247:897404 [7] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-165-213:9625:9787 [0] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-165-213:9626:9790 [1] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-166-36:104430:104589 [1] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-166-36:104431:104593 [2] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-166-244:897242:897400 [2] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-166-244:897245:897401 [5] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-162-46:702881:703043 [1] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-167-9:1039543:1039699 [6] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-165-213:9629:9793 [4] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-166-214:624948:625104 [5] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-162-46:702887:703045 [7] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-162-46:702883:703044 [3] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-166-15:127294:127457 [4] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-166-15:127295:127452 [5] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-166-15:127296:127451 [6] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-166-36:104429:104594 [0] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-166-36:104432:104592 [3] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-166-244:897241:897402 [1] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-166-36:104433:104590 [4] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-165-213:9630:9789 [5] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-165-213:9631:9788 [6] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-166-15:127292:127458 [2] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-166-214:624946:625107 [3] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-166-214:624949:625110 [6] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-166-214:624944:625106 [1] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-166-125:762819:762982 [4] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-166-214:624947:625105 [4] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-166-125:762822:762984 [7] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-167-9:1039540:1039705 [3] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-167-9:1039544:1039702 [7] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-166-125:762820:762981 [5] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-165-213:9627:9791 [2] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-166-36:104435:104595 [6] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-165-213:9632:9792 [7] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-166-36:104434:104591 [5] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-165-213:9628:9786 [3] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-166-36:104436:104596 [7] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-167-9:1039538:1039704 [1] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-167-9:1039539:1039698 [2] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-167-9:1039537:1039703 [0] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-167-9:1039542:1039701 [5] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-166-125:762815:762983 [0] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-166-125:762817:762986 [2] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-166-125:762816:762987 [1] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-166-125:762818:762985 [3] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-166-214:624950:625109 [7] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-166-214:624943:625108 [0] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-166-214:624945:625111 [2] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-166-125:762821:762988 [6] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-166-244:897243:897403 [3] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-166-244:897246:897405 [6] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-166-244:897240:897406 [0] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-166-244:897244:897407 [4] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-162-46:702880:703042 [0] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-162-46:702880:703042 [0] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-162-46:702880:703042 [0] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-162-46:702880:703042 [0] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-162-46:702880:703042 [0] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-162-46:702880:703042 [0] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-162-46:702880:703042 [0] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-162-46:702880:703042 [0] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-162-46:702880:703042 [0] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-162-46:702880:703042 [0] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-162-46:702880:703042 [0] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-162-46:702880:703042 [0] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-162-46:702880:703042 [0] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-162-46:702880:703042 [0] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-162-46:702880:703042 [0] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-162-46:702880:703042 [0] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-162-46:702880:703042 [0] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-162-46:702880:703042 [0] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-162-46:702880:703042 [0] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-162-46:702880:703042 [0] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-162-46:702880:703042 [0] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-162-46:702880:703042 [0] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-162-46:702880:703042 [0] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-162-46:702880:703042 [0] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-162-46:702880:703042 [0] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-162-46:702880:703042 [0] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-162-46:702880:703042 [0] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-162-46:702880:703042 [0] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-162-46:702880:703042 [0] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-162-46:702880:703042 [0] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-162-46:702880:703042 [0] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-162-46:702880:703042 [0] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-162-46:702880:703042 [0] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/349
+ip-26-0-162-46:702880:703042 [0] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-166-15:127290:127456 [0] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-166-15:127293:127454 [3] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-166-15:127291:127455 [1] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-166-15:127297:127453 [7] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-162-46:702882:703046 [2] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-162-46:702884:703047 [4] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-162-46:702886:703049 [6] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-162-46:702885:703048 [5] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-166-36:104430:104589 [1] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-166-36:104430:104589 [1] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-166-36:104430:104589 [1] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-166-36:104430:104589 [1] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-166-36:104430:104589 [1] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-166-36:104430:104589 [1] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-166-36:104430:104589 [1] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-166-36:104430:104589 [1] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-166-36:104430:104589 [1] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-166-36:104430:104589 [1] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-166-36:104430:104589 [1] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-166-36:104430:104589 [1] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-166-36:104430:104589 [1] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-166-36:104430:104589 [1] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-166-36:104430:104589 [1] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-166-36:104430:104589 [1] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-166-36:104430:104589 [1] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-166-36:104430:104589 [1] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-166-36:104430:104589 [1] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-166-36:104430:104589 [1] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-166-36:104430:104589 [1] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-166-36:104430:104589 [1] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-166-36:104430:104589 [1] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-166-36:104430:104589 [1] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-166-36:104430:104589 [1] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-166-36:104430:104589 [1] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-166-36:104430:104589 [1] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-166-36:104430:104589 [1] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-166-36:104430:104589 [1] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-166-36:104430:104589 [1] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-166-36:104430:104589 [1] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-166-36:104430:104589 [1] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-166-36:104430:104589 [1] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/280
+ip-26-0-166-36:104430:104589 [1] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-166-15:127295:127452 [5] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-166-15:127295:127452 [5] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-166-15:127295:127452 [5] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-166-15:127295:127452 [5] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-166-15:127295:127452 [5] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-166-15:127295:127452 [5] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-166-15:127295:127452 [5] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-166-15:127295:127452 [5] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-166-15:127295:127452 [5] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-166-15:127295:127452 [5] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-166-15:127295:127452 [5] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-166-15:127295:127452 [5] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-166-15:127295:127452 [5] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-166-15:127295:127452 [5] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-166-15:127295:127452 [5] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-166-15:127295:127452 [5] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-166-15:127295:127452 [5] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-166-15:127295:127452 [5] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-166-15:127295:127452 [5] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-166-15:127295:127452 [5] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-166-15:127295:127452 [5] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-166-15:127295:127452 [5] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-166-15:127295:127452 [5] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-166-15:127295:127452 [5] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-166-15:127295:127452 [5] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-166-15:127295:127452 [5] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-166-15:127295:127452 [5] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-166-15:127295:127452 [5] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-166-15:127295:127452 [5] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-166-15:127295:127452 [5] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-166-15:127295:127452 [5] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-166-15:127295:127452 [5] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-166-15:127295:127452 [5] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/280
+ip-26-0-166-15:127295:127452 [5] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-166-244:897245:897401 [5] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-166-244:897245:897401 [5] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-166-244:897245:897401 [5] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-166-244:897245:897401 [5] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-166-244:897245:897401 [5] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-166-244:897245:897401 [5] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-166-244:897245:897401 [5] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-166-244:897245:897401 [5] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-166-244:897245:897401 [5] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-166-244:897245:897401 [5] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-166-244:897245:897401 [5] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-166-244:897245:897401 [5] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-166-244:897245:897401 [5] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-166-244:897245:897401 [5] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-166-244:897245:897401 [5] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-166-244:897245:897401 [5] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-166-244:897245:897401 [5] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-166-244:897245:897401 [5] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-166-244:897245:897401 [5] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-166-244:897245:897401 [5] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-166-244:897245:897401 [5] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-166-244:897245:897401 [5] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-166-244:897245:897401 [5] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-166-244:897245:897401 [5] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-166-244:897245:897401 [5] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-166-244:897245:897401 [5] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-166-244:897245:897401 [5] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-166-244:897245:897401 [5] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-166-244:897245:897401 [5] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-166-244:897245:897401 [5] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-166-244:897245:897401 [5] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-166-244:897245:897401 [5] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-162-46:702880:703042 [0] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-162-46:702880:703042 [0] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-162-46:702880:703042 [0] NCCL INFO Using network Libfabric
+ip-26-0-166-244:897245:897401 [5] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/280
+ip-26-0-166-244:897245:897401 [5] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-162-46:702881:703043 [1] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-162-46:702881:703043 [1] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-162-46:702881:703043 [1] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-162-46:702881:703043 [1] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-162-46:702881:703043 [1] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-162-46:702881:703043 [1] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-162-46:702881:703043 [1] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-162-46:702881:703043 [1] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-162-46:702881:703043 [1] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-162-46:702881:703043 [1] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-162-46:702881:703043 [1] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-162-46:702881:703043 [1] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-162-46:702881:703043 [1] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-162-46:702881:703043 [1] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-162-46:702881:703043 [1] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-162-46:702881:703043 [1] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-162-46:702881:703043 [1] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-162-46:702881:703043 [1] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-162-46:702881:703043 [1] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-162-46:702881:703043 [1] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-162-46:702881:703043 [1] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-162-46:702881:703043 [1] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-162-46:702881:703043 [1] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-162-46:702881:703043 [1] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-162-46:702881:703043 [1] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-162-46:702881:703043 [1] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-162-46:702881:703043 [1] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-162-46:702881:703043 [1] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-162-46:702881:703043 [1] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-162-46:702881:703043 [1] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-162-46:702881:703043 [1] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-162-46:702881:703043 [1] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-162-46:702881:703043 [1] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/280
+ip-26-0-162-46:702881:703043 [1] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-167-9:1039543:1039699 [6] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-167-9:1039543:1039699 [6] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-167-9:1039543:1039699 [6] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-167-9:1039543:1039699 [6] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-167-9:1039543:1039699 [6] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-167-9:1039543:1039699 [6] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-167-9:1039543:1039699 [6] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-167-9:1039543:1039699 [6] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-167-9:1039543:1039699 [6] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-167-9:1039543:1039699 [6] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-167-9:1039543:1039699 [6] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-167-9:1039543:1039699 [6] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-167-9:1039543:1039699 [6] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-167-9:1039543:1039699 [6] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-167-9:1039543:1039699 [6] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-167-9:1039543:1039699 [6] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-167-9:1039543:1039699 [6] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-167-9:1039543:1039699 [6] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-167-9:1039543:1039699 [6] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-167-9:1039543:1039699 [6] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-167-9:1039543:1039699 [6] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-167-9:1039543:1039699 [6] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-167-9:1039543:1039699 [6] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-167-9:1039543:1039699 [6] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-167-9:1039543:1039699 [6] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-167-9:1039543:1039699 [6] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-167-9:1039543:1039699 [6] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-167-9:1039543:1039699 [6] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-167-9:1039543:1039699 [6] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-167-9:1039543:1039699 [6] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-167-9:1039543:1039699 [6] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-167-9:1039543:1039699 [6] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-166-15:127296:127451 [6] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-166-15:127296:127451 [6] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-166-15:127296:127451 [6] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-166-15:127296:127451 [6] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-166-15:127296:127451 [6] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-166-15:127296:127451 [6] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-166-15:127296:127451 [6] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-166-15:127296:127451 [6] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-166-15:127296:127451 [6] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-166-15:127296:127451 [6] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-166-15:127296:127451 [6] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-166-15:127296:127451 [6] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-166-15:127296:127451 [6] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-166-15:127296:127451 [6] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-166-15:127296:127451 [6] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-166-15:127296:127451 [6] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-166-15:127296:127451 [6] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-166-15:127296:127451 [6] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-166-15:127296:127451 [6] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-166-15:127296:127451 [6] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-166-15:127296:127451 [6] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-166-15:127296:127451 [6] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-166-15:127296:127451 [6] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-166-15:127296:127451 [6] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-166-15:127296:127451 [6] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-166-15:127296:127451 [6] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-166-15:127296:127451 [6] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-166-15:127296:127451 [6] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-166-15:127296:127451 [6] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-166-15:127296:127451 [6] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-166-15:127296:127451 [6] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-166-15:127296:127451 [6] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-166-15:127296:127451 [6] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/280
+ip-26-0-166-15:127296:127451 [6] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-167-9:1039543:1039699 [6] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/280
+ip-26-0-167-9:1039543:1039699 [6] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-166-244:897247:897404 [7] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-166-244:897247:897404 [7] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-166-244:897247:897404 [7] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-166-244:897247:897404 [7] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-166-244:897247:897404 [7] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-166-244:897247:897404 [7] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-166-244:897247:897404 [7] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-166-244:897247:897404 [7] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-166-244:897247:897404 [7] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-166-244:897247:897404 [7] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-166-244:897247:897404 [7] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-166-244:897247:897404 [7] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-166-244:897247:897404 [7] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-166-244:897247:897404 [7] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-166-244:897247:897404 [7] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-166-244:897247:897404 [7] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-166-244:897247:897404 [7] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-166-244:897247:897404 [7] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-166-244:897247:897404 [7] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-166-244:897247:897404 [7] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-166-244:897247:897404 [7] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-166-244:897247:897404 [7] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-166-244:897247:897404 [7] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-166-244:897247:897404 [7] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-166-244:897247:897404 [7] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-166-244:897247:897404 [7] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-166-244:897247:897404 [7] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-166-244:897247:897404 [7] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-166-244:897247:897404 [7] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-166-244:897247:897404 [7] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-166-244:897247:897404 [7] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-166-244:897247:897404 [7] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-166-36:104430:104589 [1] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-166-36:104430:104589 [1] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-166-36:104430:104589 [1] NCCL INFO Using network Libfabric
+ip-26-0-162-46:702887:703045 [7] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-162-46:702887:703045 [7] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-162-46:702887:703045 [7] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-162-46:702887:703045 [7] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-162-46:702887:703045 [7] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-162-46:702887:703045 [7] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-162-46:702887:703045 [7] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-162-46:702887:703045 [7] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-162-46:702887:703045 [7] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-162-46:702887:703045 [7] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-162-46:702887:703045 [7] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-162-46:702887:703045 [7] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-162-46:702887:703045 [7] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-162-46:702887:703045 [7] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-162-46:702887:703045 [7] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-162-46:702887:703045 [7] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-162-46:702887:703045 [7] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-162-46:702887:703045 [7] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-162-46:702887:703045 [7] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-162-46:702887:703045 [7] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-162-46:702887:703045 [7] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-162-46:702887:703045 [7] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-162-46:702887:703045 [7] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-162-46:702887:703045 [7] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-162-46:702887:703045 [7] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-162-46:702887:703045 [7] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-162-46:702887:703045 [7] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-162-46:702887:703045 [7] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-162-46:702887:703045 [7] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-162-46:702887:703045 [7] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-162-46:702887:703045 [7] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-162-46:702887:703045 [7] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-162-46:702887:703045 [7] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/280
+ip-26-0-162-46:702887:703045 [7] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-166-15:127294:127457 [4] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-166-15:127294:127457 [4] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-166-15:127294:127457 [4] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-166-15:127294:127457 [4] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-166-15:127294:127457 [4] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-166-15:127294:127457 [4] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-166-15:127294:127457 [4] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-166-15:127294:127457 [4] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-166-15:127294:127457 [4] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-166-15:127294:127457 [4] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-166-15:127294:127457 [4] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-166-15:127294:127457 [4] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-166-15:127294:127457 [4] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-166-15:127294:127457 [4] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-166-15:127294:127457 [4] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-166-15:127294:127457 [4] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-166-15:127294:127457 [4] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-166-15:127294:127457 [4] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-166-15:127294:127457 [4] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-166-15:127294:127457 [4] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-166-15:127294:127457 [4] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-166-15:127294:127457 [4] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-166-15:127294:127457 [4] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-166-15:127294:127457 [4] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-166-15:127294:127457 [4] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-166-15:127294:127457 [4] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-166-15:127294:127457 [4] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-166-15:127294:127457 [4] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-166-15:127294:127457 [4] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-166-15:127294:127457 [4] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-166-15:127294:127457 [4] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-166-15:127294:127457 [4] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-166-15:127294:127457 [4] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/280
+ip-26-0-166-15:127294:127457 [4] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-166-244:897247:897404 [7] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/280
+ip-26-0-166-244:897247:897404 [7] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-166-244:897245:897401 [5] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-166-244:897245:897401 [5] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-166-244:897245:897401 [5] NCCL INFO Using network Libfabric
+ip-26-0-165-213:9625:9787 [0] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-165-213:9625:9787 [0] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-165-213:9625:9787 [0] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-165-213:9625:9787 [0] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-165-213:9625:9787 [0] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-165-213:9625:9787 [0] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-165-213:9625:9787 [0] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-165-213:9625:9787 [0] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-165-213:9625:9787 [0] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-165-213:9625:9787 [0] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-165-213:9625:9787 [0] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-165-213:9625:9787 [0] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-165-213:9625:9787 [0] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-165-213:9625:9787 [0] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-165-213:9625:9787 [0] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-165-213:9625:9787 [0] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-165-213:9625:9787 [0] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-165-213:9625:9787 [0] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-165-213:9625:9787 [0] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-165-213:9625:9787 [0] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-165-213:9625:9787 [0] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-165-213:9625:9787 [0] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-165-213:9625:9787 [0] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-165-213:9625:9787 [0] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-165-213:9625:9787 [0] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-165-213:9625:9787 [0] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-165-213:9625:9787 [0] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-165-213:9625:9787 [0] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-165-213:9625:9787 [0] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-165-213:9625:9787 [0] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-165-213:9625:9787 [0] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-165-213:9625:9787 [0] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-162-46:702883:703044 [3] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-162-46:702883:703044 [3] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-162-46:702883:703044 [3] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-162-46:702883:703044 [3] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-162-46:702883:703044 [3] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-162-46:702883:703044 [3] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-162-46:702883:703044 [3] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-162-46:702883:703044 [3] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-162-46:702883:703044 [3] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-162-46:702883:703044 [3] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-162-46:702883:703044 [3] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-162-46:702883:703044 [3] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-162-46:702883:703044 [3] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-162-46:702883:703044 [3] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-162-46:702883:703044 [3] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-162-46:702883:703044 [3] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-162-46:702883:703044 [3] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-162-46:702883:703044 [3] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-162-46:702883:703044 [3] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-162-46:702883:703044 [3] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-162-46:702883:703044 [3] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-162-46:702883:703044 [3] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-162-46:702883:703044 [3] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-162-46:702883:703044 [3] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-162-46:702883:703044 [3] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-162-46:702883:703044 [3] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-162-46:702883:703044 [3] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-162-46:702883:703044 [3] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-162-46:702883:703044 [3] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-162-46:702883:703044 [3] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-162-46:702883:703044 [3] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-162-46:702883:703044 [3] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-165-213:9625:9787 [0] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/280
+ip-26-0-165-213:9625:9787 [0] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-162-46:702883:703044 [3] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/280
+ip-26-0-162-46:702883:703044 [3] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-166-244:897242:897400 [2] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-166-244:897242:897400 [2] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-166-244:897242:897400 [2] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-166-244:897242:897400 [2] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-166-244:897242:897400 [2] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-166-244:897242:897400 [2] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-166-244:897242:897400 [2] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-166-244:897242:897400 [2] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-166-244:897242:897400 [2] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-166-244:897242:897400 [2] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-166-244:897242:897400 [2] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-166-244:897242:897400 [2] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-166-244:897242:897400 [2] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-166-244:897242:897400 [2] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-166-244:897242:897400 [2] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-166-244:897242:897400 [2] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-166-244:897242:897400 [2] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-166-244:897242:897400 [2] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-166-244:897242:897400 [2] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-166-244:897242:897400 [2] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-166-244:897242:897400 [2] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-166-244:897242:897400 [2] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-166-244:897242:897400 [2] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-166-244:897242:897400 [2] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-166-244:897242:897400 [2] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-166-244:897242:897400 [2] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-166-244:897242:897400 [2] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-166-244:897242:897400 [2] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-166-244:897242:897400 [2] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-166-244:897242:897400 [2] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-166-244:897242:897400 [2] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-166-244:897242:897400 [2] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-166-244:897242:897400 [2] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/280
+ip-26-0-166-244:897242:897400 [2] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-166-125:762820:762981 [5] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-166-125:762820:762981 [5] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-166-125:762820:762981 [5] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-166-125:762820:762981 [5] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-166-125:762820:762981 [5] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-166-125:762820:762981 [5] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-166-125:762820:762981 [5] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-166-125:762820:762981 [5] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-166-125:762820:762981 [5] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-166-125:762820:762981 [5] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-166-125:762820:762981 [5] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-166-125:762820:762981 [5] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-166-125:762820:762981 [5] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-166-125:762820:762981 [5] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-166-125:762820:762981 [5] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-166-125:762820:762981 [5] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-166-125:762820:762981 [5] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-166-125:762820:762981 [5] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-166-125:762820:762981 [5] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-166-125:762820:762981 [5] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-166-125:762820:762981 [5] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-166-125:762820:762981 [5] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-166-125:762820:762981 [5] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-166-125:762820:762981 [5] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-166-125:762820:762981 [5] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-166-125:762820:762981 [5] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-166-125:762820:762981 [5] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-166-125:762820:762981 [5] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-166-125:762820:762981 [5] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-166-125:762820:762981 [5] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-166-125:762820:762981 [5] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-166-125:762820:762981 [5] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-166-125:762820:762981 [5] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/280
+ip-26-0-166-125:762820:762981 [5] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-162-46:702885:703048 [5] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-162-46:702885:703048 [5] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-162-46:702885:703048 [5] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-162-46:702885:703048 [5] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-162-46:702885:703048 [5] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-162-46:702885:703048 [5] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-162-46:702885:703048 [5] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-162-46:702885:703048 [5] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-162-46:702885:703048 [5] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-162-46:702885:703048 [5] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-162-46:702885:703048 [5] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-162-46:702885:703048 [5] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-162-46:702885:703048 [5] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-162-46:702885:703048 [5] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-162-46:702885:703048 [5] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-162-46:702885:703048 [5] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-162-46:702885:703048 [5] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-162-46:702885:703048 [5] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-162-46:702885:703048 [5] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-162-46:702885:703048 [5] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-162-46:702885:703048 [5] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-162-46:702885:703048 [5] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-162-46:702885:703048 [5] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-162-46:702885:703048 [5] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-162-46:702885:703048 [5] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-162-46:702885:703048 [5] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-162-46:702885:703048 [5] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-162-46:702885:703048 [5] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-162-46:702885:703048 [5] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-162-46:702885:703048 [5] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-162-46:702885:703048 [5] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-162-46:702885:703048 [5] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-167-9:1039543:1039699 [6] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-167-9:1039543:1039699 [6] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-167-9:1039543:1039699 [6] NCCL INFO Using network Libfabric
+ip-26-0-166-36:104431:104593 [2] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-166-36:104431:104593 [2] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-166-36:104431:104593 [2] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-166-36:104431:104593 [2] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-166-36:104431:104593 [2] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-166-36:104431:104593 [2] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-166-36:104431:104593 [2] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-166-36:104431:104593 [2] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-166-36:104431:104593 [2] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-166-36:104431:104593 [2] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-166-36:104431:104593 [2] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-166-36:104431:104593 [2] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-166-36:104431:104593 [2] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-166-36:104431:104593 [2] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-166-36:104431:104593 [2] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-166-36:104431:104593 [2] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-166-36:104431:104593 [2] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-166-36:104431:104593 [2] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-166-36:104431:104593 [2] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-166-36:104431:104593 [2] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-166-36:104431:104593 [2] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-166-36:104431:104593 [2] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-166-36:104431:104593 [2] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-166-36:104431:104593 [2] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-166-36:104431:104593 [2] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-166-36:104431:104593 [2] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-166-36:104431:104593 [2] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-166-36:104431:104593 [2] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-166-36:104431:104593 [2] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-166-36:104431:104593 [2] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-166-36:104431:104593 [2] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-166-36:104431:104593 [2] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-166-36:104431:104593 [2] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/280
+ip-26-0-166-36:104431:104593 [2] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-166-244:897241:897402 [1] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-166-244:897241:897402 [1] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-166-244:897241:897402 [1] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-166-244:897241:897402 [1] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-166-244:897241:897402 [1] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-166-244:897241:897402 [1] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-166-244:897241:897402 [1] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-166-244:897241:897402 [1] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-166-244:897241:897402 [1] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-166-244:897241:897402 [1] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-166-244:897241:897402 [1] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-166-244:897241:897402 [1] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-166-244:897241:897402 [1] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-166-244:897241:897402 [1] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-166-244:897241:897402 [1] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-166-244:897241:897402 [1] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-166-244:897241:897402 [1] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-166-244:897241:897402 [1] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-166-244:897241:897402 [1] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-166-244:897241:897402 [1] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-166-244:897241:897402 [1] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-166-244:897241:897402 [1] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-166-244:897241:897402 [1] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-166-244:897241:897402 [1] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-166-244:897241:897402 [1] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-166-244:897241:897402 [1] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-166-244:897241:897402 [1] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-166-244:897241:897402 [1] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-166-244:897241:897402 [1] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-166-244:897241:897402 [1] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-166-244:897241:897402 [1] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-166-244:897241:897402 [1] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-166-244:897241:897402 [1] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/280
+ip-26-0-166-244:897241:897402 [1] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-166-36:104433:104590 [4] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-166-36:104433:104590 [4] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-166-36:104433:104590 [4] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-166-36:104433:104590 [4] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-166-36:104433:104590 [4] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-166-36:104433:104590 [4] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-166-36:104433:104590 [4] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-166-36:104433:104590 [4] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-166-36:104433:104590 [4] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-166-36:104433:104590 [4] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-166-36:104433:104590 [4] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-166-36:104433:104590 [4] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-166-36:104433:104590 [4] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-166-36:104433:104590 [4] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-166-36:104433:104590 [4] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-166-36:104433:104590 [4] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-166-36:104433:104590 [4] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-166-36:104433:104590 [4] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-166-36:104433:104590 [4] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-166-36:104433:104590 [4] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-166-36:104433:104590 [4] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-166-36:104433:104590 [4] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-166-36:104433:104590 [4] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-166-36:104433:104590 [4] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-166-36:104433:104590 [4] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-166-36:104433:104590 [4] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-166-36:104433:104590 [4] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-166-36:104433:104590 [4] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-166-36:104433:104590 [4] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-166-36:104433:104590 [4] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-166-36:104433:104590 [4] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-166-36:104433:104590 [4] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-166-36:104433:104590 [4] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/280
+ip-26-0-166-36:104433:104590 [4] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-166-244:897240:897406 [0] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-166-244:897240:897406 [0] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-166-244:897240:897406 [0] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-166-244:897240:897406 [0] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-166-244:897240:897406 [0] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-166-244:897240:897406 [0] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-166-244:897240:897406 [0] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-166-244:897240:897406 [0] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-166-244:897240:897406 [0] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-166-244:897240:897406 [0] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-166-244:897240:897406 [0] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-166-244:897240:897406 [0] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-166-244:897240:897406 [0] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-166-244:897240:897406 [0] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-166-244:897240:897406 [0] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-166-244:897240:897406 [0] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-166-244:897240:897406 [0] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-166-244:897240:897406 [0] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-166-244:897240:897406 [0] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-166-244:897240:897406 [0] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-166-244:897240:897406 [0] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-166-244:897240:897406 [0] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-166-244:897240:897406 [0] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-166-244:897240:897406 [0] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-166-244:897240:897406 [0] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-166-244:897240:897406 [0] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-166-244:897240:897406 [0] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-166-244:897240:897406 [0] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-166-244:897240:897406 [0] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-166-244:897240:897406 [0] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-166-244:897240:897406 [0] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-166-244:897240:897406 [0] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-166-244:897240:897406 [0] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/280
+ip-26-0-166-244:897240:897406 [0] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-162-46:702885:703048 [5] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/280
+ip-26-0-162-46:702885:703048 [5] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-162-46:702882:703046 [2] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-162-46:702882:703046 [2] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-162-46:702882:703046 [2] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-162-46:702882:703046 [2] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-162-46:702882:703046 [2] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-162-46:702882:703046 [2] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-162-46:702882:703046 [2] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-162-46:702882:703046 [2] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-162-46:702882:703046 [2] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-162-46:702882:703046 [2] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-162-46:702882:703046 [2] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-162-46:702882:703046 [2] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-162-46:702882:703046 [2] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-162-46:702882:703046 [2] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-162-46:702882:703046 [2] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-162-46:702882:703046 [2] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-162-46:702882:703046 [2] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-162-46:702882:703046 [2] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-162-46:702882:703046 [2] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-162-46:702882:703046 [2] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-162-46:702882:703046 [2] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-162-46:702882:703046 [2] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-162-46:702882:703046 [2] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-162-46:702882:703046 [2] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-162-46:702882:703046 [2] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-162-46:702882:703046 [2] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-162-46:702882:703046 [2] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-162-46:702882:703046 [2] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-162-46:702882:703046 [2] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-162-46:702882:703046 [2] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-162-46:702882:703046 [2] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-162-46:702882:703046 [2] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-162-46:702882:703046 [2] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/280
+ip-26-0-162-46:702882:703046 [2] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-162-46:702886:703049 [6] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-162-46:702886:703049 [6] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-162-46:702886:703049 [6] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-162-46:702886:703049 [6] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-162-46:702886:703049 [6] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-162-46:702886:703049 [6] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-162-46:702886:703049 [6] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-162-46:702886:703049 [6] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-162-46:702886:703049 [6] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-162-46:702886:703049 [6] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-162-46:702886:703049 [6] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-162-46:702886:703049 [6] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-162-46:702886:703049 [6] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-162-46:702886:703049 [6] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-162-46:702886:703049 [6] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-162-46:702886:703049 [6] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-162-46:702886:703049 [6] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-162-46:702886:703049 [6] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-162-46:702886:703049 [6] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-162-46:702886:703049 [6] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-162-46:702886:703049 [6] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-162-46:702886:703049 [6] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-162-46:702886:703049 [6] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-162-46:702886:703049 [6] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-162-46:702886:703049 [6] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-162-46:702886:703049 [6] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-162-46:702886:703049 [6] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-162-46:702886:703049 [6] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-162-46:702886:703049 [6] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-162-46:702886:703049 [6] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-162-46:702886:703049 [6] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-162-46:702886:703049 [6] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-162-46:702884:703047 [4] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-162-46:702884:703047 [4] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-162-46:702884:703047 [4] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-162-46:702884:703047 [4] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-162-46:702884:703047 [4] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-162-46:702884:703047 [4] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-162-46:702884:703047 [4] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-162-46:702884:703047 [4] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-162-46:702884:703047 [4] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-162-46:702884:703047 [4] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-162-46:702884:703047 [4] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-162-46:702884:703047 [4] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-162-46:702884:703047 [4] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-162-46:702884:703047 [4] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-162-46:702884:703047 [4] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-162-46:702884:703047 [4] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-162-46:702884:703047 [4] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-162-46:702884:703047 [4] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-162-46:702884:703047 [4] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-162-46:702884:703047 [4] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-162-46:702884:703047 [4] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-162-46:702884:703047 [4] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-162-46:702884:703047 [4] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-162-46:702884:703047 [4] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-162-46:702884:703047 [4] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-162-46:702884:703047 [4] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-162-46:702884:703047 [4] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-162-46:702884:703047 [4] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-162-46:702884:703047 [4] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-162-46:702884:703047 [4] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-162-46:702884:703047 [4] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-162-46:702884:703047 [4] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-162-46:702886:703049 [6] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/280
+ip-26-0-162-46:702886:703049 [6] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-162-46:702884:703047 [4] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/280
+ip-26-0-162-46:702884:703047 [4] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-166-15:127292:127458 [2] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-166-15:127292:127458 [2] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-166-15:127292:127458 [2] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-166-15:127292:127458 [2] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-166-15:127292:127458 [2] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-166-15:127292:127458 [2] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-166-15:127292:127458 [2] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-166-15:127292:127458 [2] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-166-15:127292:127458 [2] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-166-15:127292:127458 [2] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-166-15:127292:127458 [2] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-166-15:127292:127458 [2] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-166-15:127292:127458 [2] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-166-15:127292:127458 [2] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-166-15:127292:127458 [2] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-166-15:127292:127458 [2] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-166-15:127292:127458 [2] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-166-15:127292:127458 [2] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-166-15:127292:127458 [2] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-166-15:127292:127458 [2] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-166-15:127292:127458 [2] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-166-15:127292:127458 [2] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-166-15:127292:127458 [2] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-166-15:127292:127458 [2] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-166-15:127292:127458 [2] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-166-15:127292:127458 [2] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-166-15:127292:127458 [2] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-166-15:127292:127458 [2] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-166-15:127292:127458 [2] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-166-15:127292:127458 [2] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-166-15:127292:127458 [2] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-166-15:127292:127458 [2] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-165-213:9626:9790 [1] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-165-213:9626:9790 [1] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-165-213:9626:9790 [1] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-165-213:9626:9790 [1] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-165-213:9626:9790 [1] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-165-213:9626:9790 [1] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-165-213:9626:9790 [1] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-165-213:9626:9790 [1] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-165-213:9626:9790 [1] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-165-213:9626:9790 [1] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-165-213:9626:9790 [1] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-165-213:9626:9790 [1] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-165-213:9626:9790 [1] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-165-213:9626:9790 [1] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-165-213:9626:9790 [1] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-165-213:9626:9790 [1] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-165-213:9626:9790 [1] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-165-213:9626:9790 [1] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-165-213:9626:9790 [1] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-165-213:9626:9790 [1] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-165-213:9626:9790 [1] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-165-213:9626:9790 [1] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-165-213:9626:9790 [1] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-165-213:9626:9790 [1] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-165-213:9626:9790 [1] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-165-213:9626:9790 [1] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-165-213:9626:9790 [1] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-165-213:9626:9790 [1] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-165-213:9626:9790 [1] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-165-213:9626:9790 [1] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-165-213:9626:9790 [1] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-165-213:9626:9790 [1] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-166-214:624948:625104 [5] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-166-214:624948:625104 [5] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-166-214:624948:625104 [5] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-166-214:624948:625104 [5] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-166-214:624948:625104 [5] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-166-214:624948:625104 [5] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-166-214:624948:625104 [5] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-166-214:624948:625104 [5] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-166-214:624948:625104 [5] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-166-214:624948:625104 [5] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-166-214:624948:625104 [5] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-166-214:624948:625104 [5] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-166-214:624948:625104 [5] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-166-214:624948:625104 [5] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-166-214:624948:625104 [5] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-166-214:624948:625104 [5] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-166-214:624948:625104 [5] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-166-214:624948:625104 [5] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-166-214:624948:625104 [5] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-166-214:624948:625104 [5] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-166-214:624948:625104 [5] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-166-214:624948:625104 [5] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-166-214:624948:625104 [5] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-166-214:624948:625104 [5] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-166-214:624948:625104 [5] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-166-214:624948:625104 [5] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-166-214:624948:625104 [5] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-166-214:624948:625104 [5] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-166-214:624948:625104 [5] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-166-214:624948:625104 [5] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-166-214:624948:625104 [5] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-166-214:624948:625104 [5] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-166-244:897243:897403 [3] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-166-244:897243:897403 [3] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-166-244:897243:897403 [3] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-166-244:897243:897403 [3] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-166-244:897243:897403 [3] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-166-244:897243:897403 [3] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-166-244:897243:897403 [3] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-166-244:897243:897403 [3] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-166-244:897243:897403 [3] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-166-244:897243:897403 [3] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-166-244:897243:897403 [3] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-166-244:897243:897403 [3] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-166-244:897243:897403 [3] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-166-244:897243:897403 [3] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-166-244:897243:897403 [3] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-166-244:897243:897403 [3] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-166-244:897243:897403 [3] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-166-244:897243:897403 [3] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-166-244:897243:897403 [3] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-166-244:897243:897403 [3] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-166-244:897243:897403 [3] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-166-244:897243:897403 [3] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-166-244:897243:897403 [3] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-166-244:897243:897403 [3] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-166-244:897243:897403 [3] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-166-244:897243:897403 [3] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-166-244:897243:897403 [3] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-166-244:897243:897403 [3] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-166-244:897243:897403 [3] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-166-244:897243:897403 [3] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-166-244:897243:897403 [3] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-166-244:897243:897403 [3] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-165-213:9626:9790 [1] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/280
+ip-26-0-165-213:9626:9790 [1] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-166-214:624948:625104 [5] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/280
+ip-26-0-166-214:624948:625104 [5] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-166-244:897243:897403 [3] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/280
+ip-26-0-166-244:897243:897403 [3] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-166-244:897246:897405 [6] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-166-244:897246:897405 [6] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-166-244:897246:897405 [6] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-166-244:897246:897405 [6] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-166-244:897246:897405 [6] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-166-244:897246:897405 [6] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-166-244:897246:897405 [6] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-166-244:897246:897405 [6] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-166-244:897246:897405 [6] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-166-244:897246:897405 [6] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-166-244:897246:897405 [6] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-166-244:897246:897405 [6] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-166-244:897246:897405 [6] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-166-244:897246:897405 [6] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-166-244:897246:897405 [6] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-166-244:897246:897405 [6] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-166-244:897246:897405 [6] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-166-244:897246:897405 [6] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-166-244:897246:897405 [6] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-166-244:897246:897405 [6] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-166-244:897246:897405 [6] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-166-244:897246:897405 [6] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-166-244:897246:897405 [6] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-166-244:897246:897405 [6] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-166-244:897246:897405 [6] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-166-244:897246:897405 [6] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-166-244:897246:897405 [6] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-166-244:897246:897405 [6] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-166-244:897246:897405 [6] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-166-244:897246:897405 [6] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-166-244:897246:897405 [6] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-166-244:897246:897405 [6] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-166-244:897244:897407 [4] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-166-244:897244:897407 [4] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-166-244:897244:897407 [4] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-166-244:897244:897407 [4] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-166-244:897244:897407 [4] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-166-244:897244:897407 [4] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-166-244:897244:897407 [4] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-166-244:897244:897407 [4] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-166-244:897244:897407 [4] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-166-244:897244:897407 [4] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-166-244:897244:897407 [4] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-166-244:897244:897407 [4] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-166-244:897244:897407 [4] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-166-244:897244:897407 [4] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-166-244:897244:897407 [4] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-166-244:897244:897407 [4] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-166-244:897244:897407 [4] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-166-244:897244:897407 [4] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-166-244:897244:897407 [4] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-166-244:897244:897407 [4] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-166-244:897244:897407 [4] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-166-244:897244:897407 [4] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-166-244:897244:897407 [4] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-166-244:897244:897407 [4] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-166-244:897244:897407 [4] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-166-244:897244:897407 [4] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-166-244:897244:897407 [4] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-166-244:897244:897407 [4] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-166-244:897244:897407 [4] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-166-244:897244:897407 [4] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-166-244:897244:897407 [4] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-166-244:897244:897407 [4] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-166-15:127292:127458 [2] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/280
+ip-26-0-166-15:127292:127458 [2] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-166-15:127291:127455 [1] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-166-15:127291:127455 [1] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-166-15:127291:127455 [1] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-166-15:127291:127455 [1] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-166-15:127291:127455 [1] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-166-15:127291:127455 [1] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-166-15:127291:127455 [1] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-166-15:127291:127455 [1] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-166-15:127291:127455 [1] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-166-15:127291:127455 [1] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-166-15:127291:127455 [1] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-166-15:127291:127455 [1] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-166-15:127291:127455 [1] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-166-15:127291:127455 [1] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-166-15:127291:127455 [1] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-166-15:127291:127455 [1] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-166-15:127291:127455 [1] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-166-15:127291:127455 [1] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-166-15:127291:127455 [1] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-166-15:127291:127455 [1] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-166-15:127291:127455 [1] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-166-15:127291:127455 [1] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-166-15:127291:127455 [1] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-166-15:127291:127455 [1] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-166-15:127291:127455 [1] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-166-15:127291:127455 [1] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-166-15:127291:127455 [1] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-166-15:127291:127455 [1] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-166-15:127291:127455 [1] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-166-15:127291:127455 [1] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-166-15:127291:127455 [1] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-166-15:127291:127455 [1] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-166-15:127291:127455 [1] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/280
+ip-26-0-166-15:127291:127455 [1] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-166-15:127290:127456 [0] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-166-15:127290:127456 [0] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-166-15:127290:127456 [0] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-166-15:127290:127456 [0] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-166-15:127290:127456 [0] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-166-15:127290:127456 [0] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-166-15:127290:127456 [0] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-166-15:127290:127456 [0] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-166-15:127290:127456 [0] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-166-15:127290:127456 [0] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-166-15:127290:127456 [0] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-166-15:127290:127456 [0] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-166-15:127290:127456 [0] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-166-15:127290:127456 [0] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-166-15:127290:127456 [0] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-166-15:127290:127456 [0] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-166-15:127290:127456 [0] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-166-15:127290:127456 [0] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-166-15:127290:127456 [0] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-166-15:127290:127456 [0] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-166-15:127290:127456 [0] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-166-15:127290:127456 [0] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-166-15:127290:127456 [0] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-166-15:127290:127456 [0] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-166-15:127290:127456 [0] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-166-15:127290:127456 [0] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-166-15:127290:127456 [0] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-166-15:127290:127456 [0] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-166-15:127290:127456 [0] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-166-15:127290:127456 [0] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-166-15:127290:127456 [0] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-166-15:127290:127456 [0] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-166-15:127290:127456 [0] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/280
+ip-26-0-166-15:127290:127456 [0] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-166-15:127293:127454 [3] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-166-15:127293:127454 [3] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-166-15:127293:127454 [3] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-166-15:127293:127454 [3] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-166-15:127293:127454 [3] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-166-15:127293:127454 [3] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-166-15:127293:127454 [3] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-166-15:127293:127454 [3] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-166-15:127293:127454 [3] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-166-15:127293:127454 [3] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-166-15:127293:127454 [3] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-166-15:127293:127454 [3] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-166-15:127293:127454 [3] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-166-15:127293:127454 [3] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-166-15:127293:127454 [3] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-166-15:127293:127454 [3] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-166-15:127293:127454 [3] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-166-15:127293:127454 [3] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-166-15:127293:127454 [3] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-166-15:127293:127454 [3] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-166-15:127293:127454 [3] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-166-15:127293:127454 [3] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-166-15:127293:127454 [3] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-166-15:127293:127454 [3] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-166-15:127293:127454 [3] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-166-15:127293:127454 [3] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-166-15:127293:127454 [3] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-166-15:127293:127454 [3] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-166-15:127293:127454 [3] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-166-15:127293:127454 [3] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-166-15:127293:127454 [3] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-166-15:127293:127454 [3] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-166-15:127297:127453 [7] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-166-15:127297:127453 [7] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-166-15:127297:127453 [7] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-166-15:127297:127453 [7] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-166-15:127297:127453 [7] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-166-15:127297:127453 [7] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-166-15:127297:127453 [7] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-166-15:127297:127453 [7] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-166-15:127297:127453 [7] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-166-15:127297:127453 [7] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-166-15:127297:127453 [7] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-166-15:127297:127453 [7] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-166-15:127297:127453 [7] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-166-15:127297:127453 [7] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-166-15:127297:127453 [7] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-166-15:127297:127453 [7] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-166-15:127297:127453 [7] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-166-15:127297:127453 [7] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-166-15:127297:127453 [7] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-166-15:127297:127453 [7] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-166-15:127297:127453 [7] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-166-15:127297:127453 [7] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-166-15:127297:127453 [7] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-166-15:127297:127453 [7] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-166-15:127297:127453 [7] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-166-15:127297:127453 [7] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-166-15:127297:127453 [7] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-166-15:127297:127453 [7] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-166-15:127297:127453 [7] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-166-15:127297:127453 [7] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-166-15:127297:127453 [7] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-166-15:127297:127453 [7] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-166-15:127293:127454 [3] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/280
+ip-26-0-166-15:127293:127454 [3] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-166-15:127297:127453 [7] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/280
+ip-26-0-166-15:127297:127453 [7] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-166-36:104432:104592 [3] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-166-36:104432:104592 [3] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-166-36:104432:104592 [3] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-166-36:104432:104592 [3] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-166-36:104432:104592 [3] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-166-36:104432:104592 [3] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-166-36:104432:104592 [3] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-166-36:104432:104592 [3] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-166-36:104432:104592 [3] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-166-36:104432:104592 [3] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-166-36:104432:104592 [3] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-166-36:104432:104592 [3] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-166-36:104432:104592 [3] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-166-36:104432:104592 [3] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-166-36:104432:104592 [3] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-166-36:104432:104592 [3] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-166-36:104432:104592 [3] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-166-36:104432:104592 [3] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-166-36:104432:104592 [3] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-166-36:104432:104592 [3] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-166-36:104432:104592 [3] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-166-36:104432:104592 [3] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-166-36:104432:104592 [3] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-166-36:104432:104592 [3] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-166-36:104432:104592 [3] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-166-36:104432:104592 [3] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-166-36:104432:104592 [3] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-166-36:104432:104592 [3] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-166-36:104432:104592 [3] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-166-36:104432:104592 [3] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-166-36:104432:104592 [3] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-166-36:104432:104592 [3] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-166-36:104432:104592 [3] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/280
+ip-26-0-166-36:104432:104592 [3] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-165-213:9631:9788 [6] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-165-213:9631:9788 [6] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-165-213:9631:9788 [6] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-165-213:9631:9788 [6] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-165-213:9631:9788 [6] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-165-213:9631:9788 [6] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-165-213:9631:9788 [6] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-165-213:9631:9788 [6] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-165-213:9631:9788 [6] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-165-213:9631:9788 [6] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-165-213:9631:9788 [6] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-165-213:9631:9788 [6] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-165-213:9631:9788 [6] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-165-213:9631:9788 [6] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-165-213:9631:9788 [6] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-165-213:9631:9788 [6] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-165-213:9631:9788 [6] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-165-213:9631:9788 [6] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-165-213:9631:9788 [6] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-165-213:9631:9788 [6] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-165-213:9631:9788 [6] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-165-213:9631:9788 [6] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-165-213:9631:9788 [6] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-165-213:9631:9788 [6] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-165-213:9631:9788 [6] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-165-213:9631:9788 [6] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-165-213:9631:9788 [6] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-165-213:9631:9788 [6] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-165-213:9631:9788 [6] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-165-213:9631:9788 [6] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-165-213:9631:9788 [6] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-165-213:9631:9788 [6] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-165-213:9631:9788 [6] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/280
+ip-26-0-165-213:9631:9788 [6] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-166-36:104435:104595 [6] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-166-36:104435:104595 [6] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-166-36:104435:104595 [6] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-166-36:104435:104595 [6] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-166-36:104435:104595 [6] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-166-36:104435:104595 [6] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-166-36:104435:104595 [6] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-166-36:104435:104595 [6] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-166-36:104435:104595 [6] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-166-36:104435:104595 [6] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-166-36:104435:104595 [6] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-166-36:104435:104595 [6] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-166-36:104435:104595 [6] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-166-36:104435:104595 [6] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-166-36:104435:104595 [6] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-166-36:104435:104595 [6] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-166-36:104435:104595 [6] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-166-36:104435:104595 [6] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-166-36:104435:104595 [6] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-166-36:104435:104595 [6] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-166-36:104435:104595 [6] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-166-36:104435:104595 [6] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-166-36:104435:104595 [6] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-166-36:104435:104595 [6] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-166-36:104435:104595 [6] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-166-36:104435:104595 [6] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-166-36:104435:104595 [6] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-166-36:104435:104595 [6] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-166-36:104435:104595 [6] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-166-36:104435:104595 [6] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-166-36:104435:104595 [6] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-166-36:104435:104595 [6] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-166-36:104429:104594 [0] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-166-36:104429:104594 [0] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-166-36:104429:104594 [0] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-166-36:104429:104594 [0] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-166-36:104429:104594 [0] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-166-36:104429:104594 [0] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-166-36:104429:104594 [0] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-166-36:104429:104594 [0] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-166-36:104429:104594 [0] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-166-36:104429:104594 [0] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-166-36:104429:104594 [0] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-166-36:104429:104594 [0] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-166-36:104429:104594 [0] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-166-36:104429:104594 [0] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-166-36:104429:104594 [0] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-166-36:104429:104594 [0] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-166-36:104429:104594 [0] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-166-36:104429:104594 [0] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-166-36:104429:104594 [0] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-166-36:104429:104594 [0] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-166-36:104429:104594 [0] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-166-36:104429:104594 [0] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-166-36:104429:104594 [0] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-166-36:104429:104594 [0] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-166-36:104429:104594 [0] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-166-36:104429:104594 [0] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-166-36:104429:104594 [0] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-166-36:104429:104594 [0] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-166-36:104429:104594 [0] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-166-36:104429:104594 [0] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-166-36:104429:104594 [0] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-166-36:104429:104594 [0] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-167-9:1039542:1039701 [5] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-167-9:1039542:1039701 [5] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-167-9:1039542:1039701 [5] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-167-9:1039542:1039701 [5] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-167-9:1039542:1039701 [5] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-167-9:1039542:1039701 [5] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-167-9:1039542:1039701 [5] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-167-9:1039542:1039701 [5] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-167-9:1039542:1039701 [5] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-167-9:1039542:1039701 [5] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-167-9:1039542:1039701 [5] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-167-9:1039542:1039701 [5] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-167-9:1039542:1039701 [5] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-167-9:1039542:1039701 [5] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-167-9:1039542:1039701 [5] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-167-9:1039542:1039701 [5] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-167-9:1039542:1039701 [5] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-167-9:1039542:1039701 [5] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-167-9:1039542:1039701 [5] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-167-9:1039542:1039701 [5] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-167-9:1039542:1039701 [5] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-167-9:1039542:1039701 [5] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-167-9:1039542:1039701 [5] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-167-9:1039542:1039701 [5] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-167-9:1039542:1039701 [5] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-167-9:1039542:1039701 [5] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-167-9:1039542:1039701 [5] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-167-9:1039542:1039701 [5] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-167-9:1039542:1039701 [5] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-167-9:1039542:1039701 [5] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-167-9:1039542:1039701 [5] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-167-9:1039542:1039701 [5] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-166-36:104435:104595 [6] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/280
+ip-26-0-166-36:104435:104595 [6] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-166-36:104429:104594 [0] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/280
+ip-26-0-166-36:104429:104594 [0] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-165-213:9630:9789 [5] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-165-213:9630:9789 [5] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-165-213:9630:9789 [5] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-165-213:9630:9789 [5] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-165-213:9630:9789 [5] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-165-213:9630:9789 [5] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-165-213:9630:9789 [5] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-165-213:9630:9789 [5] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-165-213:9630:9789 [5] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-165-213:9630:9789 [5] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-165-213:9630:9789 [5] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-165-213:9630:9789 [5] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-165-213:9630:9789 [5] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-165-213:9630:9789 [5] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-165-213:9630:9789 [5] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-165-213:9630:9789 [5] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-165-213:9630:9789 [5] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-165-213:9630:9789 [5] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-165-213:9630:9789 [5] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-165-213:9630:9789 [5] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-165-213:9630:9789 [5] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-165-213:9630:9789 [5] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-165-213:9630:9789 [5] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-165-213:9630:9789 [5] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-165-213:9630:9789 [5] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-165-213:9630:9789 [5] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-165-213:9630:9789 [5] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-165-213:9630:9789 [5] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-165-213:9630:9789 [5] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-165-213:9630:9789 [5] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-165-213:9630:9789 [5] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-165-213:9630:9789 [5] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-167-9:1039542:1039701 [5] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/280
+ip-26-0-167-9:1039542:1039701 [5] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-166-36:104436:104596 [7] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-166-36:104436:104596 [7] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-166-36:104436:104596 [7] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-166-36:104436:104596 [7] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-166-36:104436:104596 [7] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-166-36:104436:104596 [7] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-166-36:104436:104596 [7] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-166-36:104436:104596 [7] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-166-36:104436:104596 [7] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-166-36:104436:104596 [7] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-166-36:104436:104596 [7] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-166-36:104436:104596 [7] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-166-36:104436:104596 [7] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-166-36:104436:104596 [7] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-166-36:104436:104596 [7] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-166-36:104436:104596 [7] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-166-36:104436:104596 [7] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-166-36:104436:104596 [7] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-166-36:104436:104596 [7] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-166-36:104436:104596 [7] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-166-36:104436:104596 [7] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-166-36:104436:104596 [7] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-166-36:104436:104596 [7] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-166-36:104436:104596 [7] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-166-36:104436:104596 [7] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-166-36:104436:104596 [7] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-166-36:104436:104596 [7] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-166-36:104436:104596 [7] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-166-36:104436:104596 [7] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-166-36:104436:104596 [7] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-166-36:104436:104596 [7] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-166-36:104436:104596 [7] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-166-36:104436:104596 [7] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/280
+ip-26-0-166-36:104436:104596 [7] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-166-36:104434:104591 [5] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-166-36:104434:104591 [5] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-166-36:104434:104591 [5] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-166-36:104434:104591 [5] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-166-36:104434:104591 [5] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-166-36:104434:104591 [5] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-166-36:104434:104591 [5] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-166-36:104434:104591 [5] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-166-36:104434:104591 [5] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-166-36:104434:104591 [5] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-166-36:104434:104591 [5] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-166-36:104434:104591 [5] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-166-36:104434:104591 [5] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-166-36:104434:104591 [5] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-166-36:104434:104591 [5] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-166-36:104434:104591 [5] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-166-36:104434:104591 [5] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-166-36:104434:104591 [5] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-166-36:104434:104591 [5] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-166-36:104434:104591 [5] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-166-36:104434:104591 [5] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-166-36:104434:104591 [5] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-166-36:104434:104591 [5] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-166-36:104434:104591 [5] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-166-36:104434:104591 [5] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-166-36:104434:104591 [5] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-166-36:104434:104591 [5] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-166-36:104434:104591 [5] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-166-36:104434:104591 [5] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-166-36:104434:104591 [5] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-166-36:104434:104591 [5] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-166-36:104434:104591 [5] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-167-9:1039539:1039698 [2] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-167-9:1039539:1039698 [2] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-167-9:1039539:1039698 [2] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-167-9:1039539:1039698 [2] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-167-9:1039539:1039698 [2] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-167-9:1039539:1039698 [2] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-167-9:1039539:1039698 [2] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-167-9:1039539:1039698 [2] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-167-9:1039539:1039698 [2] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-167-9:1039539:1039698 [2] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-167-9:1039539:1039698 [2] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-167-9:1039539:1039698 [2] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-167-9:1039539:1039698 [2] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-167-9:1039539:1039698 [2] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-167-9:1039539:1039698 [2] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-167-9:1039539:1039698 [2] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-167-9:1039539:1039698 [2] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-167-9:1039539:1039698 [2] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-167-9:1039539:1039698 [2] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-167-9:1039539:1039698 [2] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-167-9:1039539:1039698 [2] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-167-9:1039539:1039698 [2] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-167-9:1039539:1039698 [2] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-167-9:1039539:1039698 [2] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-167-9:1039539:1039698 [2] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-167-9:1039539:1039698 [2] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-167-9:1039539:1039698 [2] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-167-9:1039539:1039698 [2] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-167-9:1039539:1039698 [2] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-167-9:1039539:1039698 [2] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-167-9:1039539:1039698 [2] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-167-9:1039539:1039698 [2] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-166-125:762817:762986 [2] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-166-125:762817:762986 [2] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-166-125:762817:762986 [2] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-166-125:762817:762986 [2] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-166-125:762817:762986 [2] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-166-125:762817:762986 [2] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-166-125:762817:762986 [2] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-166-125:762817:762986 [2] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-166-125:762817:762986 [2] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-166-125:762817:762986 [2] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-166-125:762817:762986 [2] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-166-125:762817:762986 [2] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-166-125:762817:762986 [2] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-166-125:762817:762986 [2] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-166-125:762817:762986 [2] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-166-125:762817:762986 [2] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-166-125:762817:762986 [2] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-166-125:762817:762986 [2] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-166-125:762817:762986 [2] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-166-125:762817:762986 [2] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-166-125:762817:762986 [2] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-166-125:762817:762986 [2] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-166-125:762817:762986 [2] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-166-125:762817:762986 [2] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-166-125:762817:762986 [2] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-166-125:762817:762986 [2] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-166-125:762817:762986 [2] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-166-125:762817:762986 [2] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-166-125:762817:762986 [2] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-166-125:762817:762986 [2] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-166-125:762817:762986 [2] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-166-125:762817:762986 [2] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-167-9:1039539:1039698 [2] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/280
+ip-26-0-167-9:1039539:1039698 [2] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-166-244:897246:897405 [6] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/280
+ip-26-0-166-244:897246:897405 [6] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-166-244:897244:897407 [4] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/280
+ip-26-0-166-244:897244:897407 [4] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-166-125:762817:762986 [2] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/280
+ip-26-0-166-125:762817:762986 [2] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-167-9:1039540:1039705 [3] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-167-9:1039540:1039705 [3] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-167-9:1039540:1039705 [3] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-167-9:1039540:1039705 [3] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-167-9:1039540:1039705 [3] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-167-9:1039540:1039705 [3] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-167-9:1039540:1039705 [3] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-167-9:1039540:1039705 [3] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-167-9:1039540:1039705 [3] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-167-9:1039540:1039705 [3] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-167-9:1039540:1039705 [3] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-167-9:1039540:1039705 [3] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-167-9:1039540:1039705 [3] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-167-9:1039540:1039705 [3] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-167-9:1039540:1039705 [3] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-167-9:1039540:1039705 [3] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-167-9:1039540:1039705 [3] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-167-9:1039540:1039705 [3] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-167-9:1039540:1039705 [3] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-167-9:1039540:1039705 [3] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-167-9:1039540:1039705 [3] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-167-9:1039540:1039705 [3] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-167-9:1039540:1039705 [3] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-167-9:1039540:1039705 [3] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-167-9:1039540:1039705 [3] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-167-9:1039540:1039705 [3] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-167-9:1039540:1039705 [3] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-167-9:1039540:1039705 [3] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-167-9:1039540:1039705 [3] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-167-9:1039540:1039705 [3] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-167-9:1039540:1039705 [3] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-167-9:1039540:1039705 [3] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-167-9:1039540:1039705 [3] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/280
+ip-26-0-167-9:1039540:1039705 [3] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-165-213:9630:9789 [5] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/280
+ip-26-0-165-213:9630:9789 [5] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-165-213:9625:9787 [0] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-165-213:9625:9787 [0] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-165-213:9625:9787 [0] NCCL INFO Using network Libfabric
+ip-26-0-165-213:9632:9792 [7] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-165-213:9632:9792 [7] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-165-213:9632:9792 [7] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-165-213:9632:9792 [7] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-165-213:9632:9792 [7] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-165-213:9632:9792 [7] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-165-213:9632:9792 [7] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-165-213:9632:9792 [7] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-165-213:9632:9792 [7] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-165-213:9632:9792 [7] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-165-213:9632:9792 [7] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-165-213:9632:9792 [7] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-165-213:9632:9792 [7] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-165-213:9632:9792 [7] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-165-213:9632:9792 [7] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-165-213:9632:9792 [7] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-165-213:9632:9792 [7] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-165-213:9632:9792 [7] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-165-213:9632:9792 [7] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-165-213:9632:9792 [7] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-165-213:9632:9792 [7] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-165-213:9632:9792 [7] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-165-213:9632:9792 [7] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-165-213:9632:9792 [7] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-165-213:9632:9792 [7] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-165-213:9632:9792 [7] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-165-213:9632:9792 [7] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-165-213:9632:9792 [7] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-165-213:9632:9792 [7] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-165-213:9632:9792 [7] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-165-213:9632:9792 [7] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-165-213:9632:9792 [7] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-165-213:9629:9793 [4] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-165-213:9629:9793 [4] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-165-213:9629:9793 [4] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-165-213:9629:9793 [4] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-165-213:9629:9793 [4] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-165-213:9629:9793 [4] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-165-213:9629:9793 [4] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-165-213:9629:9793 [4] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-165-213:9629:9793 [4] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-165-213:9629:9793 [4] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-165-213:9629:9793 [4] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-165-213:9629:9793 [4] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-165-213:9629:9793 [4] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-165-213:9629:9793 [4] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-165-213:9629:9793 [4] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-165-213:9629:9793 [4] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-165-213:9629:9793 [4] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-165-213:9629:9793 [4] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-165-213:9629:9793 [4] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-165-213:9629:9793 [4] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-165-213:9629:9793 [4] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-165-213:9629:9793 [4] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-165-213:9629:9793 [4] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-165-213:9629:9793 [4] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-165-213:9629:9793 [4] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-165-213:9629:9793 [4] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-165-213:9629:9793 [4] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-165-213:9629:9793 [4] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-165-213:9629:9793 [4] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-165-213:9629:9793 [4] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-165-213:9629:9793 [4] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-165-213:9629:9793 [4] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-165-213:9627:9791 [2] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-165-213:9627:9791 [2] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-165-213:9627:9791 [2] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-165-213:9627:9791 [2] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-165-213:9627:9791 [2] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-165-213:9627:9791 [2] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-165-213:9627:9791 [2] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-165-213:9627:9791 [2] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-165-213:9627:9791 [2] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-165-213:9627:9791 [2] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-165-213:9627:9791 [2] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-165-213:9627:9791 [2] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-165-213:9627:9791 [2] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-165-213:9627:9791 [2] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-165-213:9627:9791 [2] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-165-213:9627:9791 [2] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-165-213:9627:9791 [2] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-165-213:9627:9791 [2] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-165-213:9627:9791 [2] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-165-213:9627:9791 [2] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-165-213:9627:9791 [2] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-165-213:9627:9791 [2] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-165-213:9627:9791 [2] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-165-213:9627:9791 [2] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-165-213:9627:9791 [2] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-165-213:9627:9791 [2] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-165-213:9627:9791 [2] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-165-213:9627:9791 [2] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-165-213:9627:9791 [2] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-165-213:9627:9791 [2] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-165-213:9627:9791 [2] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-165-213:9627:9791 [2] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-165-213:9632:9792 [7] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/280
+ip-26-0-165-213:9632:9792 [7] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-166-125:762820:762981 [5] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-166-125:762820:762981 [5] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-165-213:9629:9793 [4] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/280
+ip-26-0-165-213:9629:9793 [4] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-166-125:762820:762981 [5] NCCL INFO Using network Libfabric
+ip-26-0-165-213:9628:9786 [3] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-165-213:9628:9786 [3] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-165-213:9628:9786 [3] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-165-213:9628:9786 [3] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-165-213:9628:9786 [3] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-165-213:9628:9786 [3] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-165-213:9628:9786 [3] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-165-213:9628:9786 [3] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-165-213:9628:9786 [3] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-165-213:9628:9786 [3] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-165-213:9628:9786 [3] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-165-213:9628:9786 [3] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-165-213:9628:9786 [3] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-165-213:9628:9786 [3] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-165-213:9628:9786 [3] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-165-213:9628:9786 [3] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-165-213:9628:9786 [3] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-165-213:9628:9786 [3] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-165-213:9628:9786 [3] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-165-213:9628:9786 [3] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-165-213:9628:9786 [3] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-165-213:9628:9786 [3] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-165-213:9628:9786 [3] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-165-213:9628:9786 [3] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-165-213:9628:9786 [3] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-165-213:9628:9786 [3] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-165-213:9628:9786 [3] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-165-213:9628:9786 [3] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-165-213:9628:9786 [3] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-165-213:9628:9786 [3] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-165-213:9628:9786 [3] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-165-213:9628:9786 [3] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-165-213:9627:9791 [2] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/280
+ip-26-0-165-213:9627:9791 [2] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-165-213:9628:9786 [3] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/280
+ip-26-0-165-213:9628:9786 [3] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-162-46:702881:703043 [1] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-162-46:702881:703043 [1] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-162-46:702881:703043 [1] NCCL INFO Using network Libfabric
+ip-26-0-167-9:1039538:1039704 [1] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-167-9:1039538:1039704 [1] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-167-9:1039538:1039704 [1] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-167-9:1039538:1039704 [1] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-167-9:1039538:1039704 [1] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-167-9:1039538:1039704 [1] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-167-9:1039538:1039704 [1] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-167-9:1039538:1039704 [1] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-167-9:1039538:1039704 [1] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-167-9:1039538:1039704 [1] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-167-9:1039538:1039704 [1] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-167-9:1039538:1039704 [1] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-167-9:1039538:1039704 [1] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-167-9:1039538:1039704 [1] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-167-9:1039538:1039704 [1] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-167-9:1039538:1039704 [1] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-167-9:1039538:1039704 [1] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-167-9:1039538:1039704 [1] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-167-9:1039538:1039704 [1] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-167-9:1039538:1039704 [1] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-167-9:1039538:1039704 [1] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-167-9:1039538:1039704 [1] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-167-9:1039538:1039704 [1] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-167-9:1039538:1039704 [1] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-167-9:1039538:1039704 [1] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-167-9:1039538:1039704 [1] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-167-9:1039538:1039704 [1] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-167-9:1039538:1039704 [1] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-167-9:1039538:1039704 [1] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-167-9:1039538:1039704 [1] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-167-9:1039538:1039704 [1] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-167-9:1039538:1039704 [1] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-166-15:127295:127452 [5] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-166-15:127295:127452 [5] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-166-15:127295:127452 [5] NCCL INFO Using network Libfabric
+ip-26-0-167-9:1039538:1039704 [1] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/280
+ip-26-0-167-9:1039538:1039704 [1] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-167-9:1039544:1039702 [7] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-167-9:1039544:1039702 [7] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-167-9:1039544:1039702 [7] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-167-9:1039544:1039702 [7] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-167-9:1039544:1039702 [7] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-167-9:1039544:1039702 [7] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-167-9:1039544:1039702 [7] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-167-9:1039544:1039702 [7] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-167-9:1039544:1039702 [7] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-167-9:1039544:1039702 [7] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-167-9:1039544:1039702 [7] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-167-9:1039544:1039702 [7] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-167-9:1039544:1039702 [7] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-167-9:1039544:1039702 [7] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-167-9:1039544:1039702 [7] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-167-9:1039544:1039702 [7] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-167-9:1039544:1039702 [7] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-167-9:1039544:1039702 [7] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-167-9:1039544:1039702 [7] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-167-9:1039544:1039702 [7] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-167-9:1039544:1039702 [7] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-167-9:1039544:1039702 [7] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-167-9:1039544:1039702 [7] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-167-9:1039544:1039702 [7] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-167-9:1039544:1039702 [7] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-167-9:1039544:1039702 [7] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-167-9:1039544:1039702 [7] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-167-9:1039544:1039702 [7] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-167-9:1039544:1039702 [7] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-167-9:1039544:1039702 [7] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-167-9:1039544:1039702 [7] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-167-9:1039544:1039702 [7] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-166-15:127296:127451 [6] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-166-15:127296:127451 [6] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-166-15:127296:127451 [6] NCCL INFO Using network Libfabric
+ip-26-0-166-15:127294:127457 [4] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-166-15:127294:127457 [4] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-166-15:127294:127457 [4] NCCL INFO Using network Libfabric
+ip-26-0-166-36:104434:104591 [5] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/280
+ip-26-0-166-36:104434:104591 [5] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-166-214:624949:625110 [6] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-166-214:624949:625110 [6] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-166-214:624949:625110 [6] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-166-214:624949:625110 [6] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-166-214:624949:625110 [6] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-166-214:624949:625110 [6] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-166-214:624949:625110 [6] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-166-214:624949:625110 [6] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-166-214:624949:625110 [6] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-166-214:624949:625110 [6] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-166-214:624949:625110 [6] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-166-214:624949:625110 [6] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-166-214:624949:625110 [6] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-166-214:624949:625110 [6] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-166-214:624949:625110 [6] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-166-214:624949:625110 [6] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-166-214:624949:625110 [6] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-166-214:624949:625110 [6] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-166-214:624949:625110 [6] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-166-214:624949:625110 [6] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-166-214:624949:625110 [6] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-166-214:624949:625110 [6] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-166-214:624949:625110 [6] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-166-214:624949:625110 [6] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-166-214:624949:625110 [6] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-166-214:624949:625110 [6] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-166-214:624949:625110 [6] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-166-214:624949:625110 [6] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-166-214:624949:625110 [6] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-166-214:624949:625110 [6] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-166-214:624949:625110 [6] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-166-214:624949:625110 [6] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-166-214:624949:625110 [6] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/280
+ip-26-0-166-214:624949:625110 [6] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-166-125:762815:762983 [0] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-166-125:762815:762983 [0] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-166-125:762815:762983 [0] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-166-125:762815:762983 [0] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-166-125:762815:762983 [0] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-166-125:762815:762983 [0] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-166-125:762815:762983 [0] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-166-125:762815:762983 [0] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-166-125:762815:762983 [0] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-166-125:762815:762983 [0] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-166-125:762815:762983 [0] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-166-125:762815:762983 [0] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-166-125:762815:762983 [0] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-166-125:762815:762983 [0] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-166-125:762815:762983 [0] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-166-125:762815:762983 [0] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-166-125:762815:762983 [0] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-166-125:762815:762983 [0] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-166-125:762815:762983 [0] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-166-125:762815:762983 [0] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-166-125:762815:762983 [0] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-166-125:762815:762983 [0] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-166-125:762815:762983 [0] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-166-125:762815:762983 [0] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-166-125:762815:762983 [0] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-166-125:762815:762983 [0] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-166-125:762815:762983 [0] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-166-125:762815:762983 [0] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-166-125:762815:762983 [0] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-166-125:762815:762983 [0] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-166-125:762815:762983 [0] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-166-125:762815:762983 [0] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-166-214:624943:625108 [0] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-166-214:624943:625108 [0] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-166-214:624943:625108 [0] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-166-214:624943:625108 [0] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-166-214:624943:625108 [0] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-166-214:624943:625108 [0] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-166-214:624943:625108 [0] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-166-214:624943:625108 [0] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-166-214:624943:625108 [0] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-166-214:624943:625108 [0] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-166-214:624943:625108 [0] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-166-214:624943:625108 [0] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-166-214:624943:625108 [0] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-166-214:624943:625108 [0] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-166-214:624943:625108 [0] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-166-214:624943:625108 [0] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-166-214:624943:625108 [0] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-166-214:624943:625108 [0] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-166-214:624943:625108 [0] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-166-214:624943:625108 [0] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-166-214:624943:625108 [0] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-166-214:624943:625108 [0] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-166-214:624943:625108 [0] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-166-214:624943:625108 [0] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-166-214:624943:625108 [0] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-166-214:624943:625108 [0] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-166-214:624943:625108 [0] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-166-214:624943:625108 [0] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-166-214:624943:625108 [0] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-166-214:624943:625108 [0] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-166-214:624943:625108 [0] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-166-214:624943:625108 [0] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-166-125:762815:762983 [0] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/280
+ip-26-0-166-125:762815:762983 [0] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-162-46:702887:703045 [7] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-162-46:702887:703045 [7] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-162-46:702887:703045 [7] NCCL INFO Using network Libfabric
+ip-26-0-166-214:624943:625108 [0] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/280
+ip-26-0-166-214:624943:625108 [0] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-166-125:762821:762988 [6] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-166-125:762821:762988 [6] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-166-125:762821:762988 [6] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-166-125:762821:762988 [6] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-166-125:762821:762988 [6] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-166-125:762821:762988 [6] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-166-125:762821:762988 [6] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-166-125:762821:762988 [6] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-166-125:762821:762988 [6] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-166-125:762821:762988 [6] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-166-125:762821:762988 [6] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-166-125:762821:762988 [6] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-166-125:762821:762988 [6] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-166-125:762821:762988 [6] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-166-125:762821:762988 [6] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-166-125:762821:762988 [6] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-166-125:762821:762988 [6] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-166-125:762821:762988 [6] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-166-125:762821:762988 [6] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-166-125:762821:762988 [6] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-166-125:762821:762988 [6] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-166-125:762821:762988 [6] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-166-125:762821:762988 [6] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-166-125:762821:762988 [6] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-166-125:762821:762988 [6] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-166-125:762821:762988 [6] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-166-125:762821:762988 [6] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-166-125:762821:762988 [6] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-166-125:762821:762988 [6] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-166-125:762821:762988 [6] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-166-125:762821:762988 [6] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-166-125:762821:762988 [6] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-166-125:762819:762982 [4] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-166-125:762819:762982 [4] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-166-125:762819:762982 [4] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-166-125:762819:762982 [4] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-166-125:762819:762982 [4] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-166-125:762819:762982 [4] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-166-125:762819:762982 [4] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-166-125:762819:762982 [4] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-166-125:762819:762982 [4] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-166-125:762819:762982 [4] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-166-125:762819:762982 [4] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-166-125:762819:762982 [4] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-166-125:762819:762982 [4] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-166-125:762819:762982 [4] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-166-125:762819:762982 [4] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-166-125:762819:762982 [4] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-166-125:762819:762982 [4] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-166-125:762819:762982 [4] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-166-125:762819:762982 [4] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-166-125:762819:762982 [4] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-166-125:762819:762982 [4] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-166-125:762819:762982 [4] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-166-125:762819:762982 [4] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-166-125:762819:762982 [4] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-166-125:762819:762982 [4] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-166-125:762819:762982 [4] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-166-125:762819:762982 [4] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-166-125:762819:762982 [4] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-166-125:762819:762982 [4] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-166-125:762819:762982 [4] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-166-125:762819:762982 [4] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-166-125:762819:762982 [4] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-162-46:702883:703044 [3] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-162-46:702883:703044 [3] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-162-46:702883:703044 [3] NCCL INFO Using network Libfabric
+ip-26-0-166-214:624944:625106 [1] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-166-214:624944:625106 [1] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-166-214:624944:625106 [1] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-166-214:624944:625106 [1] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-166-214:624944:625106 [1] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-166-214:624944:625106 [1] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-166-214:624944:625106 [1] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-166-214:624944:625106 [1] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-166-214:624944:625106 [1] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-166-214:624944:625106 [1] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-166-214:624944:625106 [1] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-166-214:624944:625106 [1] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-166-214:624944:625106 [1] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-166-214:624944:625106 [1] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-166-214:624944:625106 [1] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-166-214:624944:625106 [1] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-166-214:624944:625106 [1] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-166-214:624944:625106 [1] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-166-214:624944:625106 [1] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-166-214:624944:625106 [1] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-166-214:624944:625106 [1] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-166-214:624944:625106 [1] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-166-214:624944:625106 [1] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-166-214:624944:625106 [1] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-166-214:624944:625106 [1] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-166-214:624944:625106 [1] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-166-214:624944:625106 [1] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-166-214:624944:625106 [1] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-166-214:624944:625106 [1] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-166-214:624944:625106 [1] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-166-214:624944:625106 [1] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-166-214:624944:625106 [1] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-166-125:762821:762988 [6] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/280
+ip-26-0-166-125:762821:762988 [6] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-166-125:762819:762982 [4] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/280
+ip-26-0-166-125:762819:762982 [4] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-166-125:762816:762987 [1] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-166-125:762816:762987 [1] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-166-125:762816:762987 [1] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-166-125:762816:762987 [1] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-166-125:762816:762987 [1] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-166-125:762816:762987 [1] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-166-125:762816:762987 [1] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-166-125:762816:762987 [1] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-166-125:762816:762987 [1] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-166-125:762816:762987 [1] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-166-125:762816:762987 [1] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-166-125:762816:762987 [1] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-166-125:762816:762987 [1] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-166-125:762816:762987 [1] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-166-125:762816:762987 [1] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-166-125:762816:762987 [1] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-166-125:762816:762987 [1] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-166-125:762816:762987 [1] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-166-125:762816:762987 [1] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-166-125:762816:762987 [1] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-166-125:762816:762987 [1] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-166-125:762816:762987 [1] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-166-125:762816:762987 [1] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-166-125:762816:762987 [1] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-166-125:762816:762987 [1] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-166-125:762816:762987 [1] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-166-125:762816:762987 [1] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-166-125:762816:762987 [1] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-166-125:762816:762987 [1] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-166-125:762816:762987 [1] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-166-125:762816:762987 [1] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-166-125:762816:762987 [1] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-166-125:762822:762984 [7] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-166-125:762822:762984 [7] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-166-125:762822:762984 [7] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-166-125:762822:762984 [7] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-166-125:762822:762984 [7] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-166-125:762822:762984 [7] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-166-125:762822:762984 [7] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-166-125:762822:762984 [7] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-166-125:762822:762984 [7] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-166-125:762822:762984 [7] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-166-125:762822:762984 [7] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-166-125:762822:762984 [7] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-166-125:762822:762984 [7] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-166-125:762822:762984 [7] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-166-125:762822:762984 [7] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-166-125:762822:762984 [7] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-166-125:762822:762984 [7] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-166-125:762822:762984 [7] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-166-125:762822:762984 [7] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-166-125:762822:762984 [7] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-166-125:762822:762984 [7] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-166-125:762822:762984 [7] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-166-125:762822:762984 [7] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-166-125:762822:762984 [7] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-166-125:762822:762984 [7] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-166-125:762822:762984 [7] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-166-125:762822:762984 [7] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-166-125:762822:762984 [7] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-166-125:762822:762984 [7] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-166-125:762822:762984 [7] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-166-125:762822:762984 [7] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-166-125:762822:762984 [7] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-166-125:762818:762985 [3] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-166-125:762818:762985 [3] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-166-125:762818:762985 [3] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-166-125:762818:762985 [3] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-166-125:762818:762985 [3] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-166-125:762818:762985 [3] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-166-125:762818:762985 [3] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-166-125:762818:762985 [3] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-166-125:762818:762985 [3] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-166-125:762818:762985 [3] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-166-125:762818:762985 [3] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-166-125:762818:762985 [3] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-166-125:762818:762985 [3] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-166-125:762818:762985 [3] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-166-125:762818:762985 [3] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-166-125:762818:762985 [3] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-166-125:762818:762985 [3] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-166-125:762818:762985 [3] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-166-125:762818:762985 [3] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-166-125:762818:762985 [3] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-166-125:762818:762985 [3] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-166-125:762818:762985 [3] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-166-125:762818:762985 [3] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-166-125:762818:762985 [3] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-166-125:762818:762985 [3] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-166-125:762818:762985 [3] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-166-125:762818:762985 [3] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-166-125:762818:762985 [3] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-166-125:762818:762985 [3] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-166-125:762818:762985 [3] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-166-125:762818:762985 [3] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-166-125:762818:762985 [3] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/280
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-167-9:1039544:1039702 [7] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/280
+ip-26-0-167-9:1039544:1039702 [7] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-167-9:1039537:1039703 [0] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-167-9:1039537:1039703 [0] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-167-9:1039537:1039703 [0] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-167-9:1039537:1039703 [0] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-167-9:1039537:1039703 [0] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-167-9:1039537:1039703 [0] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-167-9:1039537:1039703 [0] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-167-9:1039537:1039703 [0] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-167-9:1039537:1039703 [0] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-167-9:1039537:1039703 [0] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-167-9:1039537:1039703 [0] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-167-9:1039537:1039703 [0] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-167-9:1039537:1039703 [0] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-167-9:1039537:1039703 [0] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-167-9:1039537:1039703 [0] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-167-9:1039537:1039703 [0] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-167-9:1039537:1039703 [0] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-167-9:1039537:1039703 [0] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-167-9:1039537:1039703 [0] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-167-9:1039537:1039703 [0] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-167-9:1039537:1039703 [0] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-167-9:1039537:1039703 [0] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-167-9:1039537:1039703 [0] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-167-9:1039537:1039703 [0] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-167-9:1039537:1039703 [0] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-167-9:1039537:1039703 [0] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-167-9:1039537:1039703 [0] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-167-9:1039537:1039703 [0] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-167-9:1039537:1039703 [0] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-167-9:1039537:1039703 [0] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-167-9:1039537:1039703 [0] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-167-9:1039537:1039703 [0] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-167-9:1039537:1039703 [0] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/280
+ip-26-0-167-9:1039537:1039703 [0] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-166-125:762816:762987 [1] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/280
+ip-26-0-166-125:762816:762987 [1] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-166-125:762818:762985 [3] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/280
+ip-26-0-166-125:762818:762985 [3] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-166-125:762822:762984 [7] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/280
+ip-26-0-166-125:762822:762984 [7] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-165-213:9626:9790 [1] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-165-213:9626:9790 [1] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-165-213:9626:9790 [1] NCCL INFO Using network Libfabric
+ip-26-0-166-214:624950:625109 [7] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-166-214:624950:625109 [7] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-166-214:624950:625109 [7] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-166-214:624950:625109 [7] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-166-214:624950:625109 [7] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-166-214:624950:625109 [7] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-166-214:624950:625109 [7] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-166-214:624950:625109 [7] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-166-214:624950:625109 [7] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-166-214:624950:625109 [7] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-166-214:624950:625109 [7] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-166-214:624950:625109 [7] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-166-214:624950:625109 [7] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-166-214:624950:625109 [7] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-166-214:624950:625109 [7] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-166-214:624950:625109 [7] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-166-214:624950:625109 [7] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-166-214:624950:625109 [7] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-166-214:624950:625109 [7] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-166-214:624950:625109 [7] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-166-214:624950:625109 [7] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-166-214:624950:625109 [7] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-166-214:624950:625109 [7] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-166-214:624950:625109 [7] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-166-214:624950:625109 [7] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-166-214:624950:625109 [7] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-166-214:624950:625109 [7] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-166-214:624950:625109 [7] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-166-214:624950:625109 [7] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-166-214:624950:625109 [7] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-166-214:624950:625109 [7] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-166-214:624950:625109 [7] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-166-214:624944:625106 [1] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/280
+ip-26-0-166-214:624944:625106 [1] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-166-214:624950:625109 [7] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/280
+ip-26-0-166-214:624950:625109 [7] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-166-214:624946:625107 [3] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-166-214:624946:625107 [3] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-166-214:624946:625107 [3] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-166-214:624946:625107 [3] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-166-214:624946:625107 [3] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-166-214:624946:625107 [3] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-166-214:624946:625107 [3] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-166-214:624946:625107 [3] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-166-214:624946:625107 [3] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-166-214:624946:625107 [3] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-166-214:624946:625107 [3] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-166-214:624946:625107 [3] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-166-214:624946:625107 [3] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-166-214:624946:625107 [3] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-166-214:624946:625107 [3] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-166-214:624946:625107 [3] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-166-214:624946:625107 [3] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-166-214:624946:625107 [3] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-166-214:624946:625107 [3] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-166-214:624946:625107 [3] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-166-214:624946:625107 [3] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-166-214:624946:625107 [3] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-166-214:624946:625107 [3] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-166-214:624946:625107 [3] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-166-214:624946:625107 [3] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-166-214:624946:625107 [3] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-166-214:624946:625107 [3] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-166-214:624946:625107 [3] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-166-214:624946:625107 [3] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-166-214:624946:625107 [3] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-166-214:624946:625107 [3] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-166-214:624946:625107 [3] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-166-214:624946:625107 [3] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/280
+ip-26-0-166-214:624946:625107 [3] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-166-214:624947:625105 [4] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-166-214:624947:625105 [4] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-166-214:624947:625105 [4] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-166-214:624947:625105 [4] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-166-214:624947:625105 [4] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-166-214:624947:625105 [4] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-166-214:624947:625105 [4] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-166-214:624947:625105 [4] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-166-214:624947:625105 [4] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-166-214:624947:625105 [4] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-166-214:624947:625105 [4] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-166-214:624947:625105 [4] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-166-214:624947:625105 [4] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-166-214:624947:625105 [4] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-166-214:624947:625105 [4] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-166-214:624947:625105 [4] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-166-214:624947:625105 [4] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-166-214:624947:625105 [4] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-166-214:624947:625105 [4] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-166-214:624947:625105 [4] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-166-214:624947:625105 [4] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-166-214:624947:625105 [4] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-166-214:624947:625105 [4] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-166-214:624947:625105 [4] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-166-214:624947:625105 [4] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-166-214:624947:625105 [4] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-166-214:624947:625105 [4] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-166-214:624947:625105 [4] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-166-214:624947:625105 [4] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-166-214:624947:625105 [4] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-166-214:624947:625105 [4] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-166-214:624947:625105 [4] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-166-214:624947:625105 [4] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/280
+ip-26-0-166-214:624947:625105 [4] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-166-214:624945:625111 [2] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-166-214:624945:625111 [2] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-166-214:624945:625111 [2] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-166-214:624945:625111 [2] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-166-214:624945:625111 [2] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-166-214:624945:625111 [2] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-166-214:624945:625111 [2] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-166-214:624945:625111 [2] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-166-214:624945:625111 [2] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-166-214:624945:625111 [2] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-166-214:624945:625111 [2] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-166-214:624945:625111 [2] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-166-214:624945:625111 [2] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-166-214:624945:625111 [2] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-166-214:624945:625111 [2] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-166-214:624945:625111 [2] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-166-214:624945:625111 [2] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-166-214:624945:625111 [2] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-166-214:624945:625111 [2] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-166-214:624945:625111 [2] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-166-214:624945:625111 [2] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-166-214:624945:625111 [2] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-166-214:624945:625111 [2] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-166-214:624945:625111 [2] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-166-214:624945:625111 [2] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-166-214:624945:625111 [2] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-166-214:624945:625111 [2] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-166-214:624945:625111 [2] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-166-214:624945:625111 [2] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-166-214:624945:625111 [2] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-166-214:624945:625111 [2] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-166-214:624945:625111 [2] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-166-214:624948:625104 [5] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-166-214:624948:625104 [5] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-166-214:624948:625104 [5] NCCL INFO Using network Libfabric
+ip-26-0-166-214:624945:625111 [2] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/280
+ip-26-0-166-214:624945:625111 [2] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-162-46:702885:703048 [5] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-162-46:702885:703048 [5] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-162-46:702885:703048 [5] NCCL INFO Using network Libfabric
+ip-26-0-166-244:897247:897404 [7] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-166-244:897247:897404 [7] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-166-244:897247:897404 [7] NCCL INFO Using network Libfabric
+ip-26-0-162-46:702882:703046 [2] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-162-46:702882:703046 [2] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-162-46:702882:703046 [2] NCCL INFO Using network Libfabric
+ip-26-0-166-36:104431:104593 [2] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-166-36:104431:104593 [2] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-166-36:104431:104593 [2] NCCL INFO Using network Libfabric
+ip-26-0-166-36:104433:104590 [4] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-166-36:104433:104590 [4] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-166-36:104433:104590 [4] NCCL INFO Using network Libfabric
+ip-26-0-166-36:104432:104592 [3] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-166-36:104432:104592 [3] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-166-36:104432:104592 [3] NCCL INFO Using network Libfabric
+ip-26-0-166-36:104429:104594 [0] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-166-36:104429:104594 [0] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-166-36:104429:104594 [0] NCCL INFO Using network Libfabric
+ip-26-0-166-36:104435:104595 [6] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-166-36:104435:104595 [6] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-166-36:104435:104595 [6] NCCL INFO Using network Libfabric
+ip-26-0-166-36:104436:104596 [7] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-166-36:104436:104596 [7] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-166-36:104436:104596 [7] NCCL INFO Using network Libfabric
+ip-26-0-166-36:104434:104591 [5] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-166-36:104434:104591 [5] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-166-36:104434:104591 [5] NCCL INFO Using network Libfabric
+ip-26-0-162-46:702886:703049 [6] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-162-46:702886:703049 [6] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-162-46:702886:703049 [6] NCCL INFO Using network Libfabric
+ip-26-0-166-244:897242:897400 [2] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-166-244:897242:897400 [2] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-166-244:897242:897400 [2] NCCL INFO Using network Libfabric
+ip-26-0-166-244:897241:897402 [1] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-166-244:897241:897402 [1] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-166-244:897241:897402 [1] NCCL INFO Using network Libfabric
+ip-26-0-166-15:127292:127458 [2] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-166-15:127292:127458 [2] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-166-15:127292:127458 [2] NCCL INFO Using network Libfabric
+ip-26-0-166-244:897240:897406 [0] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-166-244:897240:897406 [0] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-166-244:897240:897406 [0] NCCL INFO Using network Libfabric
+ip-26-0-162-46:702884:703047 [4] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-162-46:702884:703047 [4] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-162-46:702884:703047 [4] NCCL INFO Using network Libfabric
+ip-26-0-167-9:1039542:1039701 [5] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-167-9:1039542:1039701 [5] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-167-9:1039542:1039701 [5] NCCL INFO Using network Libfabric
+ip-26-0-166-244:897246:897405 [6] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-166-244:897246:897405 [6] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-166-244:897246:897405 [6] NCCL INFO Using network Libfabric
+ip-26-0-167-9:1039540:1039705 [3] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-167-9:1039540:1039705 [3] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-167-9:1039540:1039705 [3] NCCL INFO Using network Libfabric
+ip-26-0-167-9:1039544:1039702 [7] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-167-9:1039544:1039702 [7] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-167-9:1039544:1039702 [7] NCCL INFO Using network Libfabric
+ip-26-0-166-244:897243:897403 [3] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-166-244:897243:897403 [3] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-166-244:897243:897403 [3] NCCL INFO Using network Libfabric
+ip-26-0-167-9:1039537:1039703 [0] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-167-9:1039537:1039703 [0] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-167-9:1039537:1039703 [0] NCCL INFO Using network Libfabric
+ip-26-0-167-9:1039538:1039704 [1] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-167-9:1039538:1039704 [1] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-167-9:1039538:1039704 [1] NCCL INFO Using network Libfabric
+ip-26-0-167-9:1039539:1039698 [2] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-167-9:1039539:1039698 [2] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-167-9:1039539:1039698 [2] NCCL INFO Using network Libfabric
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO Using network Libfabric
+ip-26-0-166-125:762817:762986 [2] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-166-125:762817:762986 [2] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-166-125:762817:762986 [2] NCCL INFO Using network Libfabric
+ip-26-0-166-244:897244:897407 [4] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-166-244:897244:897407 [4] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-166-244:897244:897407 [4] NCCL INFO Using network Libfabric
+ip-26-0-166-125:762815:762983 [0] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-166-125:762815:762983 [0] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-166-125:762815:762983 [0] NCCL INFO Using network Libfabric
+ip-26-0-166-125:762819:762982 [4] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-166-125:762819:762982 [4] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-166-125:762819:762982 [4] NCCL INFO Using network Libfabric
+ip-26-0-166-125:762821:762988 [6] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-166-125:762821:762988 [6] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-166-125:762821:762988 [6] NCCL INFO Using network Libfabric
+ip-26-0-166-125:762822:762984 [7] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-166-125:762822:762984 [7] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-166-125:762822:762984 [7] NCCL INFO Using network Libfabric
+ip-26-0-166-125:762816:762987 [1] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-166-125:762816:762987 [1] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-166-125:762816:762987 [1] NCCL INFO Using network Libfabric
+ip-26-0-166-125:762818:762985 [3] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-166-125:762818:762985 [3] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-166-125:762818:762985 [3] NCCL INFO Using network Libfabric
+ip-26-0-166-214:624949:625110 [6] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-166-214:624949:625110 [6] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-166-214:624949:625110 [6] NCCL INFO Using network Libfabric
+ip-26-0-165-213:9631:9788 [6] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-165-213:9631:9788 [6] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-165-213:9631:9788 [6] NCCL INFO Using network Libfabric
+ip-26-0-165-213:9630:9789 [5] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-165-213:9630:9789 [5] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-165-213:9630:9789 [5] NCCL INFO Using network Libfabric
+ip-26-0-165-213:9628:9786 [3] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-165-213:9628:9786 [3] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-165-213:9628:9786 [3] NCCL INFO Using network Libfabric
+ip-26-0-165-213:9627:9791 [2] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-165-213:9627:9791 [2] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-165-213:9627:9791 [2] NCCL INFO Using network Libfabric
+ip-26-0-165-213:9629:9793 [4] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-165-213:9629:9793 [4] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-165-213:9629:9793 [4] NCCL INFO Using network Libfabric
+ip-26-0-165-213:9632:9792 [7] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-165-213:9632:9792 [7] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-165-213:9632:9792 [7] NCCL INFO Using network Libfabric
+ip-26-0-166-214:624943:625108 [0] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-166-214:624943:625108 [0] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-166-214:624943:625108 [0] NCCL INFO Using network Libfabric
+ip-26-0-166-214:624950:625109 [7] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-166-214:624950:625109 [7] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-166-214:624950:625109 [7] NCCL INFO Using network Libfabric
+ip-26-0-166-214:624944:625106 [1] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-166-214:624944:625106 [1] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-166-214:624944:625106 [1] NCCL INFO Using network Libfabric
+ip-26-0-166-15:127291:127455 [1] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-166-15:127291:127455 [1] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-166-15:127291:127455 [1] NCCL INFO Using network Libfabric
+ip-26-0-166-214:624946:625107 [3] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-166-214:624946:625107 [3] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-166-214:624946:625107 [3] NCCL INFO Using network Libfabric
+ip-26-0-166-15:127290:127456 [0] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-166-15:127290:127456 [0] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-166-15:127290:127456 [0] NCCL INFO Using network Libfabric
+ip-26-0-166-214:624947:625105 [4] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-166-214:624947:625105 [4] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-166-214:624947:625105 [4] NCCL INFO Using network Libfabric
+ip-26-0-166-15:127293:127454 [3] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-166-15:127293:127454 [3] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-166-15:127293:127454 [3] NCCL INFO Using network Libfabric
+ip-26-0-166-15:127297:127453 [7] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-166-15:127297:127453 [7] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-166-15:127297:127453 [7] NCCL INFO Using network Libfabric
+ip-26-0-166-214:624945:625111 [2] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-166-214:624945:625111 [2] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-166-214:624945:625111 [2] NCCL INFO Using network Libfabric
+ip-26-0-166-36:104432:104592 [3] NCCL INFO comm 0x85daf90 rank 51 nranks 64 cudaDev 3 nvmlDev 3 busId 86000 commId 0x8b6df3a32fca5304 - Init START
+ip-26-0-166-36:104429:104594 [0] NCCL INFO comm 0xa2374d0 rank 48 nranks 64 cudaDev 0 nvmlDev 0 busId 53000 commId 0x8b6df3a32fca5304 - Init START
+ip-26-0-166-36:104433:104590 [4] NCCL INFO comm 0xa026de0 rank 52 nranks 64 cudaDev 4 nvmlDev 4 busId 97000 commId 0x8b6df3a32fca5304 - Init START
+ip-26-0-166-36:104436:104596 [7] NCCL INFO comm 0x9507760 rank 55 nranks 64 cudaDev 7 nvmlDev 7 busId ca000 commId 0x8b6df3a32fca5304 - Init START
+ip-26-0-166-244:897247:897404 [7] NCCL INFO comm 0x977c640 rank 47 nranks 64 cudaDev 7 nvmlDev 7 busId ca000 commId 0x8b6df3a32fca5304 - Init START
+ip-26-0-167-9:1039537:1039703 [0] NCCL INFO comm 0x9974360 rank 56 nranks 64 cudaDev 0 nvmlDev 0 busId 53000 commId 0x8b6df3a32fca5304 - Init START
+ip-26-0-167-9:1039544:1039702 [7] NCCL INFO comm 0x93e67f0 rank 63 nranks 64 cudaDev 7 nvmlDev 7 busId ca000 commId 0x8b6df3a32fca5304 - Init START
+ip-26-0-167-9:1039540:1039705 [3] NCCL INFO comm 0x955d240 rank 59 nranks 64 cudaDev 3 nvmlDev 3 busId 86000 commId 0x8b6df3a32fca5304 - Init START
+ip-26-0-167-9:1039539:1039698 [2] NCCL INFO comm 0x8e6cc50 rank 58 nranks 64 cudaDev 2 nvmlDev 2 busId 75000 commId 0x8b6df3a32fca5304 - Init START
+ip-26-0-167-9:1039538:1039704 [1] NCCL INFO comm 0x8e07600 rank 57 nranks 64 cudaDev 1 nvmlDev 1 busId 64000 commId 0x8b6df3a32fca5304 - Init START
+ip-26-0-166-36:104434:104591 [5] NCCL INFO comm 0x8836500 rank 53 nranks 64 cudaDev 5 nvmlDev 5 busId a8000 commId 0x8b6df3a32fca5304 - Init START
+ip-26-0-166-36:104435:104595 [6] NCCL INFO comm 0x8b98830 rank 54 nranks 64 cudaDev 6 nvmlDev 6 busId b9000 commId 0x8b6df3a32fca5304 - Init START
+ip-26-0-166-36:104430:104589 [1] NCCL INFO comm 0x9a479d0 rank 49 nranks 64 cudaDev 1 nvmlDev 1 busId 64000 commId 0x8b6df3a32fca5304 - Init START
+ip-26-0-166-244:897246:897405 [6] NCCL INFO comm 0xa1fc860 rank 46 nranks 64 cudaDev 6 nvmlDev 6 busId b9000 commId 0x8b6df3a32fca5304 - Init START
+ip-26-0-166-214:624950:625109 [7] NCCL INFO comm 0x8b7a600 rank 39 nranks 64 cudaDev 7 nvmlDev 7 busId ca000 commId 0x8b6df3a32fca5304 - Init START
+ip-26-0-166-214:624947:625105 [4] NCCL INFO comm 0x9dbf120 rank 36 nranks 64 cudaDev 4 nvmlDev 4 busId 97000 commId 0x8b6df3a32fca5304 - Init START
+ip-26-0-166-15:127297:127453 [7] NCCL INFO comm 0x9b1a7f0 rank 31 nranks 64 cudaDev 7 nvmlDev 7 busId ca000 commId 0x8b6df3a32fca5304 - Init START
+ip-26-0-166-15:127296:127451 [6] NCCL INFO comm 0x947e710 rank 30 nranks 64 cudaDev 6 nvmlDev 6 busId b9000 commId 0x8b6df3a32fca5304 - Init START
+ip-26-0-166-15:127295:127452 [5] NCCL INFO comm 0x88629b0 rank 29 nranks 64 cudaDev 5 nvmlDev 5 busId a8000 commId 0x8b6df3a32fca5304 - Init START
+ip-26-0-167-9:1039543:1039699 [6] NCCL INFO comm 0x9d28300 rank 62 nranks 64 cudaDev 6 nvmlDev 6 busId b9000 commId 0x8b6df3a32fca5304 - Init START
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO comm 0xa260f20 rank 60 nranks 64 cudaDev 4 nvmlDev 4 busId 97000 commId 0x8b6df3a32fca5304 - Init START
+ip-26-0-167-9:1039542:1039701 [5] NCCL INFO comm 0x8c1a980 rank 61 nranks 64 cudaDev 5 nvmlDev 5 busId a8000 commId 0x8b6df3a32fca5304 - Init START
+ip-26-0-166-36:104431:104593 [2] NCCL INFO comm 0x955bca0 rank 50 nranks 64 cudaDev 2 nvmlDev 2 busId 75000 commId 0x8b6df3a32fca5304 - Init START
+ip-26-0-166-244:897244:897407 [4] NCCL INFO comm 0x8c707d0 rank 44 nranks 64 cudaDev 4 nvmlDev 4 busId 97000 commId 0x8b6df3a32fca5304 - Init START
+ip-26-0-166-244:897245:897401 [5] NCCL INFO comm 0x9819580 rank 45 nranks 64 cudaDev 5 nvmlDev 5 busId a8000 commId 0x8b6df3a32fca5304 - Init START
+ip-26-0-166-214:624949:625110 [6] NCCL INFO comm 0x87cdfa0 rank 38 nranks 64 cudaDev 6 nvmlDev 6 busId b9000 commId 0x8b6df3a32fca5304 - Init START
+ip-26-0-166-214:624946:625107 [3] NCCL INFO comm 0x9da4f90 rank 35 nranks 64 cudaDev 3 nvmlDev 3 busId 86000 commId 0x8b6df3a32fca5304 - Init START
+ip-26-0-166-214:624945:625111 [2] NCCL INFO comm 0x9504970 rank 34 nranks 64 cudaDev 2 nvmlDev 2 busId 75000 commId 0x8b6df3a32fca5304 - Init START
+ip-26-0-166-214:624944:625106 [1] NCCL INFO comm 0x82f1510 rank 33 nranks 64 cudaDev 1 nvmlDev 1 busId 64000 commId 0x8b6df3a32fca5304 - Init START
+ip-26-0-166-214:624943:625108 [0] NCCL INFO comm 0x919eef0 rank 32 nranks 64 cudaDev 0 nvmlDev 0 busId 53000 commId 0x8b6df3a32fca5304 - Init START
+ip-26-0-166-125:762821:762988 [6] NCCL INFO comm 0x85f8640 rank 22 nranks 64 cudaDev 6 nvmlDev 6 busId b9000 commId 0x8b6df3a32fca5304 - Init START
+ip-26-0-166-125:762822:762984 [7] NCCL INFO comm 0x9dbbc60 rank 23 nranks 64 cudaDev 7 nvmlDev 7 busId ca000 commId 0x8b6df3a32fca5304 - Init START
+ip-26-0-166-244:897242:897400 [2] NCCL INFO comm 0x845ae40 rank 42 nranks 64 cudaDev 2 nvmlDev 2 busId 75000 commId 0x8b6df3a32fca5304 - Init START
+ip-26-0-166-244:897243:897403 [3] NCCL INFO comm 0x87e71f0 rank 43 nranks 64 cudaDev 3 nvmlDev 3 busId 86000 commId 0x8b6df3a32fca5304 - Init START
+ip-26-0-166-244:897241:897402 [1] NCCL INFO comm 0x9f8ce70 rank 41 nranks 64 cudaDev 1 nvmlDev 1 busId 64000 commId 0x8b6df3a32fca5304 - Init START
+ip-26-0-166-15:127292:127458 [2] NCCL INFO comm 0x8fd9570 rank 26 nranks 64 cudaDev 2 nvmlDev 2 busId 75000 commId 0x8b6df3a32fca5304 - Init START
+ip-26-0-166-15:127294:127457 [4] NCCL INFO comm 0x8d79aa0 rank 28 nranks 64 cudaDev 4 nvmlDev 4 busId 97000 commId 0x8b6df3a32fca5304 - Init START
+ip-26-0-166-125:762816:762987 [1] NCCL INFO comm 0x93d1180 rank 17 nranks 64 cudaDev 1 nvmlDev 1 busId 64000 commId 0x8b6df3a32fca5304 - Init START
+ip-26-0-166-125:762815:762983 [0] NCCL INFO comm 0x832ff30 rank 16 nranks 64 cudaDev 0 nvmlDev 0 busId 53000 commId 0x8b6df3a32fca5304 - Init START
+ip-26-0-166-244:897240:897406 [0] NCCL INFO comm 0x90efd70 rank 40 nranks 64 cudaDev 0 nvmlDev 0 busId 53000 commId 0x8b6df3a32fca5304 - Init START
+ip-26-0-165-213:9628:9786 [3] NCCL INFO comm 0x90d46e0 rank 11 nranks 64 cudaDev 3 nvmlDev 3 busId 86000 commId 0x8b6df3a32fca5304 - Init START
+ip-26-0-165-213:9627:9791 [2] NCCL INFO comm 0x86f9b60 rank 10 nranks 64 cudaDev 2 nvmlDev 2 busId 75000 commId 0x8b6df3a32fca5304 - Init START
+ip-26-0-165-213:9632:9792 [7] NCCL INFO comm 0x8a49870 rank 15 nranks 64 cudaDev 7 nvmlDev 7 busId ca000 commId 0x8b6df3a32fca5304 - Init START
+ip-26-0-165-213:9626:9790 [1] NCCL INFO comm 0x910f780 rank 9 nranks 64 cudaDev 1 nvmlDev 1 busId 64000 commId 0x8b6df3a32fca5304 - Init START
+ip-26-0-166-214:624948:625104 [5] NCCL INFO comm 0x9e36180 rank 37 nranks 64 cudaDev 5 nvmlDev 5 busId a8000 commId 0x8b6df3a32fca5304 - Init START
+ip-26-0-166-15:127291:127455 [1] NCCL INFO comm 0x9f78530 rank 25 nranks 64 cudaDev 1 nvmlDev 1 busId 64000 commId 0x8b6df3a32fca5304 - Init START
+ip-26-0-166-15:127290:127456 [0] NCCL INFO comm 0x86f0680 rank 24 nranks 64 cudaDev 0 nvmlDev 0 busId 53000 commId 0x8b6df3a32fca5304 - Init START
+ip-26-0-166-125:762819:762982 [4] NCCL INFO comm 0x91ac710 rank 20 nranks 64 cudaDev 4 nvmlDev 4 busId 97000 commId 0x8b6df3a32fca5304 - Init START
+ip-26-0-166-125:762820:762981 [5] NCCL INFO comm 0x90d8cd0 rank 21 nranks 64 cudaDev 5 nvmlDev 5 busId a8000 commId 0x8b6df3a32fca5304 - Init START
+ip-26-0-165-213:9625:9787 [0] NCCL INFO comm 0x8bb7780 rank 8 nranks 64 cudaDev 0 nvmlDev 0 busId 53000 commId 0x8b6df3a32fca5304 - Init START
+ip-26-0-165-213:9629:9793 [4] NCCL INFO comm 0x86eab90 rank 12 nranks 64 cudaDev 4 nvmlDev 4 busId 97000 commId 0x8b6df3a32fca5304 - Init START
+ip-26-0-165-213:9631:9788 [6] NCCL INFO comm 0x98a0a00 rank 14 nranks 64 cudaDev 6 nvmlDev 6 busId b9000 commId 0x8b6df3a32fca5304 - Init START
+ip-26-0-166-15:127293:127454 [3] NCCL INFO comm 0x9ee0590 rank 27 nranks 64 cudaDev 3 nvmlDev 3 busId 86000 commId 0x8b6df3a32fca5304 - Init START
+ip-26-0-166-125:762818:762985 [3] NCCL INFO comm 0x850d180 rank 19 nranks 64 cudaDev 3 nvmlDev 3 busId 86000 commId 0x8b6df3a32fca5304 - Init START
+ip-26-0-166-125:762817:762986 [2] NCCL INFO comm 0xa1ccb40 rank 18 nranks 64 cudaDev 2 nvmlDev 2 busId 75000 commId 0x8b6df3a32fca5304 - Init START
+ip-26-0-165-213:9630:9789 [5] NCCL INFO comm 0x84f2a80 rank 13 nranks 64 cudaDev 5 nvmlDev 5 busId a8000 commId 0x8b6df3a32fca5304 - Init START
+ip-26-0-162-46:702883:703044 [3] NCCL INFO comm 0x834f6a0 rank 3 nranks 64 cudaDev 3 nvmlDev 3 busId 86000 commId 0x8b6df3a32fca5304 - Init START
+ip-26-0-162-46:702880:703042 [0] NCCL INFO comm 0x967b3b0 rank 0 nranks 64 cudaDev 0 nvmlDev 0 busId 53000 commId 0x8b6df3a32fca5304 - Init START
+ip-26-0-162-46:702887:703045 [7] NCCL INFO comm 0x889e490 rank 7 nranks 64 cudaDev 7 nvmlDev 7 busId ca000 commId 0x8b6df3a32fca5304 - Init START
+ip-26-0-162-46:702884:703047 [4] NCCL INFO comm 0x8cdc140 rank 4 nranks 64 cudaDev 4 nvmlDev 4 busId 97000 commId 0x8b6df3a32fca5304 - Init START
+ip-26-0-162-46:702885:703048 [5] NCCL INFO comm 0x88bf5a0 rank 5 nranks 64 cudaDev 5 nvmlDev 5 busId a8000 commId 0x8b6df3a32fca5304 - Init START
+ip-26-0-162-46:702882:703046 [2] NCCL INFO comm 0x958dad0 rank 2 nranks 64 cudaDev 2 nvmlDev 2 busId 75000 commId 0x8b6df3a32fca5304 - Init START
+ip-26-0-162-46:702881:703043 [1] NCCL INFO comm 0x91d1af0 rank 1 nranks 64 cudaDev 1 nvmlDev 1 busId 64000 commId 0x8b6df3a32fca5304 - Init START
+ip-26-0-162-46:702886:703049 [6] NCCL INFO comm 0x8e18ff0 rank 6 nranks 64 cudaDev 6 nvmlDev 6 busId b9000 commId 0x8b6df3a32fca5304 - Init START
+ip-26-0-166-125:762817:762986 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-125:762816:762987 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-125:762815:762983 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-125:762818:762985 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-125:762820:762981 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-125:762822:762984 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-125:762819:762982 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-125:762821:762988 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-15:127290:127456 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-15:127292:127458 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-15:127293:127454 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-15:127295:127452 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-15:127296:127451 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-15:127297:127453 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-15:127291:127455 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-15:127294:127457 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-214:624943:625108 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-36:104433:104590 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-214:624944:625106 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-214:624946:625107 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-214:624945:625111 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-214:624947:625105 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-244:897244:897407 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-244:897245:897401 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-244:897247:897404 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-244:897243:897403 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-244:897246:897405 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-36:104435:104595 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-36:104432:104592 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-36:104436:104596 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-36:104434:104591 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-36:104431:104593 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-36:104429:104594 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-36:104430:104589 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-244:897242:897400 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-244:897241:897402 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-244:897240:897406 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-214:624950:625109 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-214:624948:625104 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-214:624949:625110 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-162-46:702885:703048 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-162-46:702887:703045 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-165-213:9632:9792 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-165-213:9631:9788 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-165-213:9630:9789 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-165-213:9629:9793 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-165-213:9627:9791 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-162-46:702886:703049 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-162-46:702883:703044 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-162-46:702884:703047 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-165-213:9628:9786 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-165-213:9626:9790 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-165-213:9625:9787 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-162-46:702881:703043 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-162-46:702882:703046 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-162-46:702880:703042 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/349
+ip-26-0-167-9:1039540:1039705 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-167-9:1039537:1039703 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-167-9:1039538:1039704 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-167-9:1039539:1039698 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-167-9:1039542:1039701 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-167-9:1039544:1039702 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-167-9:1039543:1039699 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-36:104431:104593 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-166-36:104431:104593 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-162-46:702883:703044 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-162-46:702883:703044 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-166-125:762820:762981 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,fff00000,00000000
+ip-26-0-166-125:762820:762981 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-166-15:127290:127456 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-166-15:127290:127456 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-166-15:127296:127451 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-166-15:127291:127455 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-166-15:127296:127451 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-166-15:127291:127455 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-166-36:104430:104589 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-166-36:104430:104589 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-166-125:762822:762984 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,fff00000,00000000
+ip-26-0-166-125:762822:762984 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-166-15:127297:127453 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-166-15:127297:127453 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-166-244:897244:897407 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-166-244:897244:897407 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-166-125:762816:762987 [1] NCCL INFO Setting affinity for GPU 1 to 0fff,ffffffff
+ip-26-0-166-125:762816:762987 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-167-9:1039539:1039698 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-167-9:1039539:1039698 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-167-9:1039538:1039704 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-167-9:1039538:1039704 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-167-9:1039543:1039699 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-167-9:1039543:1039699 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-166-15:127295:127452 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-166-15:127295:127452 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-166-36:104429:104594 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-166-36:104429:104594 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-166-15:127292:127458 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-166-15:127292:127458 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-166-15:127294:127457 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-166-15:127294:127457 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-166-15:127293:127454 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-166-15:127293:127454 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-166-125:762821:762988 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,fff00000,00000000
+ip-26-0-166-125:762821:762988 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-166-36:104433:104590 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-166-36:104433:104590 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-166-244:897240:897406 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-166-244:897240:897406 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-166-36:104434:104591 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-166-36:104434:104591 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-166-125:762817:762986 [2] NCCL INFO Setting affinity for GPU 2 to 0fff,ffffffff
+ip-26-0-162-46:702880:703042 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-162-46:702880:703042 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-166-125:762817:762986 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-162-46:702886:703049 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-162-46:702886:703049 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-167-9:1039544:1039702 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-167-9:1039544:1039702 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-166-36:104436:104596 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-166-36:104436:104596 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-166-36:104435:104595 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-166-36:104435:104595 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-166-36:104432:104592 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-166-36:104432:104592 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-162-46:702887:703045 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-162-46:702887:703045 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-167-9:1039542:1039701 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-167-9:1039542:1039701 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-166-125:762815:762983 [0] NCCL INFO Setting affinity for GPU 0 to 0fff,ffffffff
+ip-26-0-166-244:897246:897405 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-166-244:897246:897405 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-167-9:1039537:1039703 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-167-9:1039537:1039703 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-166-125:762819:762982 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,fff00000,00000000
+ip-26-0-166-125:762819:762982 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-167-9:1039540:1039705 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-166-125:762818:762985 [3] NCCL INFO Setting affinity for GPU 3 to 0fff,ffffffff
+ip-26-0-166-125:762818:762985 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-167-9:1039540:1039705 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-166-125:762815:762983 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-166-244:897243:897403 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-166-244:897243:897403 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-166-214:624944:625106 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-166-214:624944:625106 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-165-213:9632:9792 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-165-213:9632:9792 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-166-244:897247:897404 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-166-244:897247:897404 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-166-244:897242:897400 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-166-244:897242:897400 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-166-244:897241:897402 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-166-244:897241:897402 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-162-46:702882:703046 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-162-46:702882:703046 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-166-244:897245:897401 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-166-244:897245:897401 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-166-214:624945:625111 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-166-214:624945:625111 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-162-46:702884:703047 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-162-46:702884:703047 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-162-46:702881:703043 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-162-46:702881:703043 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-162-46:702885:703048 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-162-46:702885:703048 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-166-214:624946:625107 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-166-214:624946:625107 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-166-214:624943:625108 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-166-214:624943:625108 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-166-214:624950:625109 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-166-214:624950:625109 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-166-214:624947:625105 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-166-214:624947:625105 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-166-214:624948:625104 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-166-214:624948:625104 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-166-214:624949:625110 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-166-214:624949:625110 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-165-213:9629:9793 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-165-213:9629:9793 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-165-213:9630:9789 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-165-213:9630:9789 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-165-213:9627:9791 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-165-213:9627:9791 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-165-213:9625:9787 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-165-213:9625:9787 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-165-213:9626:9790 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-165-213:9626:9790 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-165-213:9628:9786 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-165-213:9628:9786 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-165-213:9631:9788 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-165-213:9631:9788 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-165-213:9631:9788 [6] NCCL INFO Trees [0] 15/-1/-1->14->13 [1] 15/-1/-1->14->13 [2] 15/-1/-1->14->13 [3] 15/-1/-1->14->13 [4] 15/-1/-1->14->13 [5] 15/-1/-1->14->13 [6] 15/-1/-1->14->22 [7] -1/-1/-1->14->13 [8] 15/-1/-1->14->13 [9] 15/-1/-1->14->13 [10] 15/-1/-1->14->13 [11] 15/-1/-1->14->13 [12] 15/-1/-1->14->13 [13] 15/-1/-1->14->13 [14] 15/22/6->14->30 [15] -1/-1/-1->14->13
+ip-26-0-165-213:9631:9788 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-213:9632:9792 [7] NCCL INFO Trees [0] -1/-1/-1->15->14 [1] 8/-1/-1->15->14 [2] 8/-1/-1->15->14 [3] 8/-1/-1->15->14 [4] 8/-1/-1->15->14 [5] 8/-1/-1->15->14 [6] 8/-1/-1->15->14 [7] 8/-1/-1->15->23 [8] -1/-1/-1->15->14 [9] 8/-1/-1->15->14 [10] 8/-1/-1->15->14 [11] 8/-1/-1->15->14 [12] 8/-1/-1->15->14 [13] 8/-1/-1->15->14 [14] 8/-1/-1->15->14 [15] 8/23/7->15->31
+ip-26-0-165-213:9632:9792 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-125:762815:762983 [0] NCCL INFO Trees [0] 17/8/24->16->32 [1] -1/-1/-1->16->23 [2] 17/-1/-1->16->23 [3] 17/-1/-1->16->23 [4] 17/-1/-1->16->23 [5] 17/-1/-1->16->23 [6] 17/-1/-1->16->23 [7] 17/-1/-1->16->23 [8] 17/-1/-1->16->8 [9] -1/-1/-1->16->23 [10] 17/-1/-1->16->23 [11] 17/-1/-1->16->23 [12] 17/-1/-1->16->23 [13] 17/-1/-1->16->23 [14] 17/-1/-1->16->23 [15] 17/-1/-1->16->23
+ip-26-0-166-125:762815:762983 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-162-46:702887:703045 [7] NCCL INFO NVLS Head  0:  0  8 16 24 32 40 48 56
+ip-26-0-162-46:702885:703048 [5] NCCL INFO NVLS Head  0:  0  8 16 24 32 40 48 56
+ip-26-0-162-46:702887:703045 [7] NCCL INFO NVLS Head  1:  1  9 17 25 33 41 49 57
+ip-26-0-162-46:702885:703048 [5] NCCL INFO NVLS Head  1:  1  9 17 25 33 41 49 57
+ip-26-0-162-46:702887:703045 [7] NCCL INFO NVLS Head  2:  2 10 18 26 34 42 50 58
+ip-26-0-166-125:762817:762986 [2] NCCL INFO Trees [0] 19/-1/-1->18->17 [1] 19/-1/-1->18->17 [2] 19/10/26->18->34 [3] -1/-1/-1->18->17 [4] 19/-1/-1->18->17 [5] 19/-1/-1->18->17 [6] 19/-1/-1->18->17 [7] 19/-1/-1->18->17 [8] 19/-1/-1->18->17 [9] 19/-1/-1->18->17 [10] 19/-1/-1->18->10 [11] -1/-1/-1->18->17 [12] 19/-1/-1->18->17 [13] 19/-1/-1->18->17 [14] 19/-1/-1->18->17 [15] 19/-1/-1->18->17
+ip-26-0-166-125:762817:762986 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-213:9630:9789 [5] NCCL INFO Trees [0] 14/-1/-1->13->12 [1] 14/-1/-1->13->12 [2] 14/-1/-1->13->12 [3] 14/-1/-1->13->12 [4] 14/-1/-1->13->12 [5] 14/-1/-1->13->21 [6] -1/-1/-1->13->12 [7] 14/-1/-1->13->12 [8] 14/-1/-1->13->12 [9] 14/-1/-1->13->12 [10] 14/-1/-1->13->12 [11] 14/-1/-1->13->12 [12] 14/-1/-1->13->12 [13] 14/21/5->13->29 [14] -1/-1/-1->13->12 [15] 14/-1/-1->13->12
+ip-26-0-165-213:9629:9793 [4] NCCL INFO Trees [0] 13/-1/-1->12->11 [1] 13/-1/-1->12->11 [2] 13/-1/-1->12->11 [3] 13/-1/-1->12->11 [4] 13/-1/-1->12->20 [5] -1/-1/-1->12->11 [6] 13/-1/-1->12->11 [7] 13/-1/-1->12->11 [8] 13/-1/-1->12->11 [9] 13/-1/-1->12->11 [10] 13/-1/-1->12->11 [11] 13/-1/-1->12->11 [12] 13/20/4->12->28 [13] -1/-1/-1->12->11 [14] 13/-1/-1->12->11 [15] 13/-1/-1->12->11
+ip-26-0-162-46:702885:703048 [5] NCCL INFO NVLS Head  2:  2 10 18 26 34 42 50 58
+ip-26-0-162-46:702887:703045 [7] NCCL INFO NVLS Head  3:  3 11 19 27 35 43 51 59
+ip-26-0-162-46:702885:703048 [5] NCCL INFO NVLS Head  3:  3 11 19 27 35 43 51 59
+ip-26-0-162-46:702887:703045 [7] NCCL INFO NVLS Head  4:  4 12 20 28 36 44 52 60
+ip-26-0-162-46:702885:703048 [5] NCCL INFO NVLS Head  4:  4 12 20 28 36 44 52 60
+ip-26-0-162-46:702887:703045 [7] NCCL INFO NVLS Head  5:  5 13 21 29 37 45 53 61
+ip-26-0-162-46:702885:703048 [5] NCCL INFO NVLS Head  5:  5 13 21 29 37 45 53 61
+ip-26-0-162-46:702887:703045 [7] NCCL INFO NVLS Head  6:  6 14 22 30 38 46 54 62
+ip-26-0-162-46:702885:703048 [5] NCCL INFO NVLS Head  6:  6 14 22 30 38 46 54 62
+ip-26-0-162-46:702887:703045 [7] NCCL INFO NVLS Head  7:  7 15 23 31 39 47 55 63
+ip-26-0-162-46:702885:703048 [5] NCCL INFO NVLS Head  7:  7 15 23 31 39 47 55 63
+ip-26-0-166-125:762822:762984 [7] NCCL INFO Trees [0] -1/-1/-1->23->22 [1] 16/-1/-1->23->22 [2] 16/-1/-1->23->22 [3] 16/-1/-1->23->22 [4] 16/-1/-1->23->22 [5] 16/-1/-1->23->22 [6] 16/-1/-1->23->22 [7] 16/15/31->23->39 [8] -1/-1/-1->23->22 [9] 16/-1/-1->23->22 [10] 16/-1/-1->23->22 [11] 16/-1/-1->23->22 [12] 16/-1/-1->23->22 [13] 16/-1/-1->23->22 [14] 16/-1/-1->23->22 [15] 16/-1/-1->23->15
+ip-26-0-166-244:897247:897404 [7] NCCL INFO Trees [0] -1/-1/-1->47->46 [1] 40/-1/-1->47->46 [2] 40/-1/-1->47->46 [3] 40/-1/-1->47->46 [4] 40/-1/-1->47->46 [5] 40/-1/-1->47->46 [6] 40/-1/-1->47->46 [7] 40/-1/-1->47->55 [8] -1/-1/-1->47->46 [9] 40/-1/-1->47->46 [10] 40/-1/-1->47->46 [11] 40/-1/-1->47->46 [12] 40/-1/-1->47->46 [13] 40/-1/-1->47->46 [14] 40/-1/-1->47->46 [15] 40/55/39->47->31
+ip-26-0-166-244:897245:897401 [5] NCCL INFO Trees [0] 46/-1/-1->45->44 [1] 46/-1/-1->45->44 [2] 46/-1/-1->45->44 [3] 46/-1/-1->45->44 [4] 46/-1/-1->45->44 [5] 46/-1/-1->45->53 [6] -1/-1/-1->45->44 [7] 46/-1/-1->45->44 [8] 46/-1/-1->45->44 [9] 46/-1/-1->45->44 [10] 46/-1/-1->45->44 [11] 46/-1/-1->45->44 [12] 46/-1/-1->45->44 [13] 46/53/37->45->29 [14] -1/-1/-1->45->44 [15] 46/-1/-1->45->44
+ip-26-0-166-244:897247:897404 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-244:897245:897401 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-214:624950:625109 [7] NCCL INFO Trees [0] -1/-1/-1->39->38 [1] 32/-1/-1->39->38 [2] 32/-1/-1->39->38 [3] 32/-1/-1->39->38 [4] 32/-1/-1->39->38 [5] 32/-1/-1->39->38 [6] 32/-1/-1->39->38 [7] 32/23/55->39->7 [8] -1/-1/-1->39->38 [9] 32/-1/-1->39->38 [10] 32/-1/-1->39->38 [11] 32/-1/-1->39->38 [12] 32/-1/-1->39->38 [13] 32/-1/-1->39->38 [14] 32/-1/-1->39->38 [15] 32/-1/-1->39->47
+ip-26-0-166-214:624950:625109 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-15:127297:127453 [7] NCCL INFO Trees [0] -1/-1/-1->31->30 [1] 24/-1/-1->31->30 [2] 24/-1/-1->31->30 [3] 24/-1/-1->31->30 [4] 24/-1/-1->31->30 [5] 24/-1/-1->31->30 [6] 24/-1/-1->31->30 [7] 24/-1/-1->31->23 [8] -1/-1/-1->31->30 [9] 24/-1/-1->31->30 [10] 24/-1/-1->31->30 [11] 24/-1/-1->31->30 [12] 24/-1/-1->31->30 [13] 24/-1/-1->31->30 [14] 24/-1/-1->31->30 [15] 24/47/15->31->63
+ip-26-0-166-15:127296:127451 [6] NCCL INFO Trees [0] 31/-1/-1->30->29 [1] 31/-1/-1->30->29 [2] 31/-1/-1->30->29 [3] 31/-1/-1->30->29 [4] 31/-1/-1->30->29 [5] 31/-1/-1->30->29 [6] 31/-1/-1->30->22 [7] -1/-1/-1->30->29 [8] 31/-1/-1->30->29 [9] 31/-1/-1->30->29 [10] 31/-1/-1->30->29 [11] 31/-1/-1->30->29 [12] 31/-1/-1->30->29 [13] 31/-1/-1->30->29 [14] 31/46/14->30->62 [15] -1/-1/-1->30->29
+ip-26-0-166-15:127297:127453 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-15:127296:127451 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-162-46:702886:703049 [6] NCCL INFO NVLS Head  0:  0  8 16 24 32 40 48 56
+ip-26-0-162-46:702884:703047 [4] NCCL INFO NVLS Head  0:  0  8 16 24 32 40 48 56
+ip-26-0-162-46:702886:703049 [6] NCCL INFO NVLS Head  1:  1  9 17 25 33 41 49 57
+ip-26-0-162-46:702882:703046 [2] NCCL INFO NVLS Head  0:  0  8 16 24 32 40 48 56
+ip-26-0-162-46:702884:703047 [4] NCCL INFO NVLS Head  1:  1  9 17 25 33 41 49 57
+ip-26-0-162-46:702886:703049 [6] NCCL INFO NVLS Head  2:  2 10 18 26 34 42 50 58
+ip-26-0-162-46:702882:703046 [2] NCCL INFO NVLS Head  1:  1  9 17 25 33 41 49 57
+ip-26-0-162-46:702884:703047 [4] NCCL INFO NVLS Head  2:  2 10 18 26 34 42 50 58
+ip-26-0-162-46:702886:703049 [6] NCCL INFO NVLS Head  3:  3 11 19 27 35 43 51 59
+ip-26-0-162-46:702882:703046 [2] NCCL INFO NVLS Head  2:  2 10 18 26 34 42 50 58
+ip-26-0-162-46:702884:703047 [4] NCCL INFO NVLS Head  3:  3 11 19 27 35 43 51 59
+ip-26-0-162-46:702886:703049 [6] NCCL INFO NVLS Head  4:  4 12 20 28 36 44 52 60
+ip-26-0-166-36:104430:104589 [1] NCCL INFO Trees [0] 50/-1/-1->49->48 [1] 50/41/57->49->33 [2] -1/-1/-1->49->48 [3] 50/-1/-1->49->48 [4] 50/-1/-1->49->48 [5] 50/-1/-1->49->48 [6] 50/-1/-1->49->48 [7] 50/-1/-1->49->48 [8] 50/-1/-1->49->48 [9] 50/-1/-1->49->41 [10] -1/-1/-1->49->48 [11] 50/-1/-1->49->48 [12] 50/-1/-1->49->48 [13] 50/-1/-1->49->48 [14] 50/-1/-1->49->48 [15] 50/-1/-1->49->48
+ip-26-0-166-36:104431:104593 [2] NCCL INFO Trees [0] 51/-1/-1->50->49 [1] 51/-1/-1->50->49 [2] 51/42/58->50->34 [3] -1/-1/-1->50->49 [4] 51/-1/-1->50->49 [5] 51/-1/-1->50->49 [6] 51/-1/-1->50->49 [7] 51/-1/-1->50->49 [8] 51/-1/-1->50->49 [9] 51/-1/-1->50->49 [10] 51/-1/-1->50->42 [11] -1/-1/-1->50->49 [12] 51/-1/-1->50->49 [13] 51/-1/-1->50->49 [14] 51/-1/-1->50->49 [15] 51/-1/-1->50->49
+ip-26-0-166-36:104430:104589 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-36:104431:104593 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-125:762822:762984 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-125:762820:762981 [5] NCCL INFO Trees [0] 22/-1/-1->21->20 [1] 22/-1/-1->21->20 [2] 22/-1/-1->21->20 [3] 22/-1/-1->21->20 [4] 22/-1/-1->21->20 [5] 22/13/29->21->37 [6] -1/-1/-1->21->20 [7] 22/-1/-1->21->20 [8] 22/-1/-1->21->20 [9] 22/-1/-1->21->20 [10] 22/-1/-1->21->20 [11] 22/-1/-1->21->20 [12] 22/-1/-1->21->20 [13] 22/-1/-1->21->13 [14] -1/-1/-1->21->20 [15] 22/-1/-1->21->20
+ip-26-0-166-125:762820:762981 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-244:897244:897407 [4] NCCL INFO Trees [0] 45/-1/-1->44->43 [1] 45/-1/-1->44->43 [2] 45/-1/-1->44->43 [3] 45/-1/-1->44->43 [4] 45/-1/-1->44->52 [5] -1/-1/-1->44->43 [6] 45/-1/-1->44->43 [7] 45/-1/-1->44->43 [8] 45/-1/-1->44->43 [9] 45/-1/-1->44->43 [10] 45/-1/-1->44->43 [11] 45/-1/-1->44->43 [12] 45/52/36->44->28 [13] -1/-1/-1->44->43 [14] 45/-1/-1->44->43 [15] 45/-1/-1->44->43
+ip-26-0-166-244:897244:897407 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-15:127294:127457 [4] NCCL INFO Trees [0] 29/-1/-1->28->27 [1] 29/-1/-1->28->27 [2] 29/-1/-1->28->27 [3] 29/-1/-1->28->27 [4] 29/-1/-1->28->20 [5] -1/-1/-1->28->27 [6] 29/-1/-1->28->27 [7] 29/-1/-1->28->27 [8] 29/-1/-1->28->27 [9] 29/-1/-1->28->27 [10] 29/-1/-1->28->27 [11] 29/-1/-1->28->27 [12] 29/44/12->28->60 [13] -1/-1/-1->28->27 [14] 29/-1/-1->28->27 [15] 29/-1/-1->28->27
+ip-26-0-166-15:127292:127458 [2] NCCL INFO Trees [0] 27/-1/-1->26->25 [1] 27/-1/-1->26->25 [2] 27/-1/-1->26->18 [3] -1/-1/-1->26->25 [4] 27/-1/-1->26->25 [5] 27/-1/-1->26->25 [6] 27/-1/-1->26->25 [7] 27/-1/-1->26->25 [8] 27/-1/-1->26->25 [9] 27/-1/-1->26->25 [10] 27/42/10->26->58 [11] -1/-1/-1->26->25 [12] 27/-1/-1->26->25 [13] 27/-1/-1->26->25 [14] 27/-1/-1->26->25 [15] 27/-1/-1->26->25
+ip-26-0-167-9:1039544:1039702 [7] NCCL INFO Trees [0] -1/-1/-1->63->62 [1] 56/-1/-1->63->62 [2] 56/-1/-1->63->62 [3] 56/-1/-1->63->62 [4] 56/-1/-1->63->62 [5] 56/-1/-1->63->62 [6] 56/-1/-1->63->62 [7] 56/-1/-1->63->55 [8] -1/-1/-1->63->62 [9] 56/-1/-1->63->62 [10] 56/-1/-1->63->62 [11] 56/-1/-1->63->62 [12] 56/-1/-1->63->62 [13] 56/-1/-1->63->62 [14] 56/-1/-1->63->62 [15] 56/31/-1->63->-1
+ip-26-0-167-9:1039544:1039702 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-167-9:1039543:1039699 [6] NCCL INFO Trees [0] 63/-1/-1->62->61 [1] 63/-1/-1->62->61 [2] 63/-1/-1->62->61 [3] 63/-1/-1->62->61 [4] 63/-1/-1->62->61 [5] 63/-1/-1->62->61 [6] 63/-1/-1->62->54 [7] -1/-1/-1->62->61 [8] 63/-1/-1->62->61 [9] 63/-1/-1->62->61 [10] 63/-1/-1->62->61 [11] 63/-1/-1->62->61 [12] 63/-1/-1->62->61 [13] 63/-1/-1->62->61 [14] 63/30/-1->62->-1 [15] -1/-1/-1->62->61
+ip-26-0-167-9:1039543:1039699 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-162-46:702882:703046 [2] NCCL INFO NVLS Head  3:  3 11 19 27 35 43 51 59
+ip-26-0-162-46:702886:703049 [6] NCCL INFO NVLS Head  5:  5 13 21 29 37 45 53 61
+ip-26-0-162-46:702884:703047 [4] NCCL INFO NVLS Head  4:  4 12 20 28 36 44 52 60
+ip-26-0-162-46:702882:703046 [2] NCCL INFO NVLS Head  4:  4 12 20 28 36 44 52 60
+ip-26-0-162-46:702886:703049 [6] NCCL INFO NVLS Head  6:  6 14 22 30 38 46 54 62
+ip-26-0-162-46:702884:703047 [4] NCCL INFO NVLS Head  5:  5 13 21 29 37 45 53 61
+ip-26-0-162-46:702883:703044 [3] NCCL INFO NVLS Head  0:  0  8 16 24 32 40 48 56
+ip-26-0-162-46:702886:703049 [6] NCCL INFO NVLS Head  7:  7 15 23 31 39 47 55 63
+ip-26-0-162-46:702882:703046 [2] NCCL INFO NVLS Head  5:  5 13 21 29 37 45 53 61
+ip-26-0-162-46:702881:703043 [1] NCCL INFO NVLS Head  0:  0  8 16 24 32 40 48 56
+ip-26-0-162-46:702884:703047 [4] NCCL INFO NVLS Head  6:  6 14 22 30 38 46 54 62
+ip-26-0-162-46:702882:703046 [2] NCCL INFO NVLS Head  6:  6 14 22 30 38 46 54 62
+ip-26-0-166-36:104436:104596 [7] NCCL INFO Trees [0] -1/-1/-1->55->54 [1] 48/-1/-1->55->54 [2] 48/-1/-1->55->54 [3] 48/-1/-1->55->54 [4] 48/-1/-1->55->54 [5] 48/-1/-1->55->54 [6] 48/-1/-1->55->54 [7] 48/47/63->55->39 [8] -1/-1/-1->55->54 [9] 48/-1/-1->55->54 [10] 48/-1/-1->55->54 [11] 48/-1/-1->55->54 [12] 48/-1/-1->55->54 [13] 48/-1/-1->55->54 [14] 48/-1/-1->55->54 [15] 48/-1/-1->55->47
+ip-26-0-166-36:104436:104596 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-125:762816:762987 [1] NCCL INFO Trees [0] 18/-1/-1->17->16 [1] 18/9/25->17->33 [2] -1/-1/-1->17->16 [3] 18/-1/-1->17->16 [4] 18/-1/-1->17->16 [5] 18/-1/-1->17->16 [6] 18/-1/-1->17->16 [7] 18/-1/-1->17->16 [8] 18/-1/-1->17->16 [9] 18/-1/-1->17->9 [10] -1/-1/-1->17->16 [11] 18/-1/-1->17->16 [12] 18/-1/-1->17->16 [13] 18/-1/-1->17->16 [14] 18/-1/-1->17->16 [15] 18/-1/-1->17->16
+ip-26-0-166-125:762818:762985 [3] NCCL INFO Trees [0] 20/-1/-1->19->18 [1] 20/-1/-1->19->18 [2] 20/-1/-1->19->18 [3] 20/11/27->19->35 [4] -1/-1/-1->19->18 [5] 20/-1/-1->19->18 [6] 20/-1/-1->19->18 [7] 20/-1/-1->19->18 [8] 20/-1/-1->19->18 [9] 20/-1/-1->19->18 [10] 20/-1/-1->19->18 [11] 20/-1/-1->19->11 [12] -1/-1/-1->19->18 [13] 20/-1/-1->19->18 [14] 20/-1/-1->19->18 [15] 20/-1/-1->19->18
+ip-26-0-166-125:762816:762987 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-213:9626:9790 [1] NCCL INFO Trees [0] 10/-1/-1->9->8 [1] 10/-1/-1->9->17 [2] -1/-1/-1->9->8 [3] 10/-1/-1->9->8 [4] 10/-1/-1->9->8 [5] 10/-1/-1->9->8 [6] 10/-1/-1->9->8 [7] 10/-1/-1->9->8 [8] 10/-1/-1->9->8 [9] 10/17/1->9->25 [10] -1/-1/-1->9->8 [11] 10/-1/-1->9->8 [12] 10/-1/-1->9->8 [13] 10/-1/-1->9->8 [14] 10/-1/-1->9->8 [15] 10/-1/-1->9->8
+ip-26-0-165-213:9630:9789 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-213:9629:9793 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-213:9628:9786 [3] NCCL INFO Trees [0] 12/-1/-1->11->10 [1] 12/-1/-1->11->10 [2] 12/-1/-1->11->10 [3] 12/-1/-1->11->19 [4] -1/-1/-1->11->10 [5] 12/-1/-1->11->10 [6] 12/-1/-1->11->10 [7] 12/-1/-1->11->10 [8] 12/-1/-1->11->10 [9] 12/-1/-1->11->10 [10] 12/-1/-1->11->10 [11] 12/19/3->11->27 [12] -1/-1/-1->11->10 [13] 12/-1/-1->11->10 [14] 12/-1/-1->11->10 [15] 12/-1/-1->11->10
+ip-26-0-165-213:9626:9790 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-214:624949:625110 [6] NCCL INFO Trees [0] 39/-1/-1->38->37 [1] 39/-1/-1->38->37 [2] 39/-1/-1->38->37 [3] 39/-1/-1->38->37 [4] 39/-1/-1->38->37 [5] 39/-1/-1->38->37 [6] 39/22/54->38->6 [7] -1/-1/-1->38->37 [8] 39/-1/-1->38->37 [9] 39/-1/-1->38->37 [10] 39/-1/-1->38->37 [11] 39/-1/-1->38->37 [12] 39/-1/-1->38->37 [13] 39/-1/-1->38->37 [14] 39/-1/-1->38->46 [15] -1/-1/-1->38->37
+ip-26-0-166-214:624949:625110 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-214:624948:625104 [5] NCCL INFO Trees [0] 38/-1/-1->37->36 [1] 38/-1/-1->37->36 [2] 38/-1/-1->37->36 [3] 38/-1/-1->37->36 [4] 38/-1/-1->37->36 [5] 38/21/53->37->5 [6] -1/-1/-1->37->36 [7] 38/-1/-1->37->36 [8] 38/-1/-1->37->36 [9] 38/-1/-1->37->36 [10] 38/-1/-1->37->36 [11] 38/-1/-1->37->36 [12] 38/-1/-1->37->36 [13] 38/-1/-1->37->45 [14] -1/-1/-1->37->36 [15] 38/-1/-1->37->36
+ip-26-0-166-15:127294:127457 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-15:127292:127458 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO Trees [0] 61/-1/-1->60->59 [1] 61/-1/-1->60->59 [2] 61/-1/-1->60->59 [3] 61/-1/-1->60->59 [4] 61/-1/-1->60->52 [5] -1/-1/-1->60->59 [6] 61/-1/-1->60->59 [7] 61/-1/-1->60->59 [8] 61/-1/-1->60->59 [9] 61/-1/-1->60->59 [10] 61/-1/-1->60->59 [11] 61/-1/-1->60->59 [12] 61/28/-1->60->-1 [13] -1/-1/-1->60->59 [14] 61/-1/-1->60->59 [15] 61/-1/-1->60->59
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-167-9:1039542:1039701 [5] NCCL INFO Trees [0] 62/-1/-1->61->60 [1] 62/-1/-1->61->60 [2] 62/-1/-1->61->60 [3] 62/-1/-1->61->60 [4] 62/-1/-1->61->60 [5] 62/-1/-1->61->53 [6] -1/-1/-1->61->60 [7] 62/-1/-1->61->60 [8] 62/-1/-1->61->60 [9] 62/-1/-1->61->60 [10] 62/-1/-1->61->60 [11] 62/-1/-1->61->60 [12] 62/-1/-1->61->60 [13] 62/29/-1->61->-1 [14] -1/-1/-1->61->60 [15] 62/-1/-1->61->60
+ip-26-0-166-36:104435:104595 [6] NCCL INFO Trees [0] 55/-1/-1->54->53 [1] 55/-1/-1->54->53 [2] 55/-1/-1->54->53 [3] 55/-1/-1->54->53 [4] 55/-1/-1->54->53 [5] 55/-1/-1->54->53 [6] 55/46/62->54->38 [7] -1/-1/-1->54->53 [8] 55/-1/-1->54->53 [9] 55/-1/-1->54->53 [10] 55/-1/-1->54->53 [11] 55/-1/-1->54->53 [12] 55/-1/-1->54->53 [13] 55/-1/-1->54->53 [14] 55/-1/-1->54->46 [15] -1/-1/-1->54->53
+ip-26-0-166-36:104435:104595 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-36:104433:104590 [4] NCCL INFO Trees [0] 53/-1/-1->52->51 [1] 53/-1/-1->52->51 [2] 53/-1/-1->52->51 [3] 53/-1/-1->52->51 [4] 53/44/60->52->36 [5] -1/-1/-1->52->51 [6] 53/-1/-1->52->51 [7] 53/-1/-1->52->51 [8] 53/-1/-1->52->51 [9] 53/-1/-1->52->51 [10] 53/-1/-1->52->51 [11] 53/-1/-1->52->51 [12] 53/-1/-1->52->44 [13] -1/-1/-1->52->51 [14] 53/-1/-1->52->51 [15] 53/-1/-1->52->51
+ip-26-0-166-125:762818:762985 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-125:762819:762982 [4] NCCL INFO Trees [0] 21/-1/-1->20->19 [1] 21/-1/-1->20->19 [2] 21/-1/-1->20->19 [3] 21/-1/-1->20->19 [4] 21/12/28->20->36 [5] -1/-1/-1->20->19 [6] 21/-1/-1->20->19 [7] 21/-1/-1->20->19 [8] 21/-1/-1->20->19 [9] 21/-1/-1->20->19 [10] 21/-1/-1->20->19 [11] 21/-1/-1->20->19 [12] 21/-1/-1->20->12 [13] -1/-1/-1->20->19 [14] 21/-1/-1->20->19 [15] 21/-1/-1->20->19
+ip-26-0-166-125:762821:762988 [6] NCCL INFO Trees [0] 23/-1/-1->22->21 [1] 23/-1/-1->22->21 [2] 23/-1/-1->22->21 [3] 23/-1/-1->22->21 [4] 23/-1/-1->22->21 [5] 23/-1/-1->22->21 [6] 23/14/30->22->38 [7] -1/-1/-1->22->21 [8] 23/-1/-1->22->21 [9] 23/-1/-1->22->21 [10] 23/-1/-1->22->21 [11] 23/-1/-1->22->21 [12] 23/-1/-1->22->21 [13] 23/-1/-1->22->21 [14] 23/-1/-1->22->14 [15] -1/-1/-1->22->21
+ip-26-0-166-125:762819:762982 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-244:897242:897400 [2] NCCL INFO Trees [0] 43/-1/-1->42->41 [1] 43/-1/-1->42->41 [2] 43/-1/-1->42->50 [3] -1/-1/-1->42->41 [4] 43/-1/-1->42->41 [5] 43/-1/-1->42->41 [6] 43/-1/-1->42->41 [7] 43/-1/-1->42->41 [8] 43/-1/-1->42->41 [9] 43/-1/-1->42->41 [10] 43/50/34->42->26 [11] -1/-1/-1->42->41 [12] 43/-1/-1->42->41 [13] 43/-1/-1->42->41 [14] 43/-1/-1->42->41 [15] 43/-1/-1->42->41
+ip-26-0-166-244:897242:897400 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-244:897241:897402 [1] NCCL INFO Trees [0] 42/-1/-1->41->40 [1] 42/-1/-1->41->49 [2] -1/-1/-1->41->40 [3] 42/-1/-1->41->40 [4] 42/-1/-1->41->40 [5] 42/-1/-1->41->40 [6] 42/-1/-1->41->40 [7] 42/-1/-1->41->40 [8] 42/-1/-1->41->40 [9] 42/49/33->41->25 [10] -1/-1/-1->41->40 [11] 42/-1/-1->41->40 [12] 42/-1/-1->41->40 [13] 42/-1/-1->41->40 [14] 42/-1/-1->41->40 [15] 42/-1/-1->41->40
+ip-26-0-166-244:897241:897402 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-213:9627:9791 [2] NCCL INFO Trees [0] 11/-1/-1->10->9 [1] 11/-1/-1->10->9 [2] 11/-1/-1->10->18 [3] -1/-1/-1->10->9 [4] 11/-1/-1->10->9 [5] 11/-1/-1->10->9 [6] 11/-1/-1->10->9 [7] 11/-1/-1->10->9 [8] 11/-1/-1->10->9 [9] 11/-1/-1->10->9 [10] 11/18/2->10->26 [11] -1/-1/-1->10->9 [12] 11/-1/-1->10->9 [13] 11/-1/-1->10->9 [14] 11/-1/-1->10->9 [15] 11/-1/-1->10->9
+ip-26-0-165-213:9628:9786 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-213:9627:9791 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-214:624947:625105 [4] NCCL INFO Trees [0] 37/-1/-1->36->35 [1] 37/-1/-1->36->35 [2] 37/-1/-1->36->35 [3] 37/-1/-1->36->35 [4] 37/20/52->36->4 [5] -1/-1/-1->36->35 [6] 37/-1/-1->36->35 [7] 37/-1/-1->36->35 [8] 37/-1/-1->36->35 [9] 37/-1/-1->36->35 [10] 37/-1/-1->36->35 [11] 37/-1/-1->36->35 [12] 37/-1/-1->36->44 [13] -1/-1/-1->36->35 [14] 37/-1/-1->36->35 [15] 37/-1/-1->36->35
+ip-26-0-166-15:127295:127452 [5] NCCL INFO Trees [0] 30/-1/-1->29->28 [1] 30/-1/-1->29->28 [2] 30/-1/-1->29->28 [3] 30/-1/-1->29->28 [4] 30/-1/-1->29->28 [5] 30/-1/-1->29->21 [6] -1/-1/-1->29->28 [7] 30/-1/-1->29->28 [8] 30/-1/-1->29->28 [9] 30/-1/-1->29->28 [10] 30/-1/-1->29->28 [11] 30/-1/-1->29->28 [12] 30/-1/-1->29->28 [13] 30/45/13->29->61 [14] -1/-1/-1->29->28 [15] 30/-1/-1->29->28
+ip-26-0-166-15:127295:127452 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-167-9:1039542:1039701 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-36:104433:104590 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-36:104432:104592 [3] NCCL INFO Trees [0] 52/-1/-1->51->50 [1] 52/-1/-1->51->50 [2] 52/-1/-1->51->50 [3] 52/43/59->51->35 [4] -1/-1/-1->51->50 [5] 52/-1/-1->51->50 [6] 52/-1/-1->51->50 [7] 52/-1/-1->51->50 [8] 52/-1/-1->51->50 [9] 52/-1/-1->51->50 [10] 52/-1/-1->51->50 [11] 52/-1/-1->51->43 [12] -1/-1/-1->51->50 [13] 52/-1/-1->51->50 [14] 52/-1/-1->51->50 [15] 52/-1/-1->51->50
+ip-26-0-166-125:762821:762988 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-244:897240:897406 [0] NCCL INFO Trees [0] 41/-1/-1->40->48 [1] -1/-1/-1->40->47 [2] 41/-1/-1->40->47 [3] 41/-1/-1->40->47 [4] 41/-1/-1->40->47 [5] 41/-1/-1->40->47 [6] 41/-1/-1->40->47 [7] 41/-1/-1->40->47 [8] 41/48/32->40->24 [9] -1/-1/-1->40->47 [10] 41/-1/-1->40->47 [11] 41/-1/-1->40->47 [12] 41/-1/-1->40->47 [13] 41/-1/-1->40->47 [14] 41/-1/-1->40->47 [15] 41/-1/-1->40->47
+ip-26-0-166-244:897246:897405 [6] NCCL INFO Trees [0] 47/-1/-1->46->45 [1] 47/-1/-1->46->45 [2] 47/-1/-1->46->45 [3] 47/-1/-1->46->45 [4] 47/-1/-1->46->45 [5] 47/-1/-1->46->45 [6] 47/-1/-1->46->54 [7] -1/-1/-1->46->45 [8] 47/-1/-1->46->45 [9] 47/-1/-1->46->45 [10] 47/-1/-1->46->45 [11] 47/-1/-1->46->45 [12] 47/-1/-1->46->45 [13] 47/-1/-1->46->45 [14] 47/54/38->46->30 [15] -1/-1/-1->46->45
+ip-26-0-166-244:897240:897406 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-244:897246:897405 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-213:9625:9787 [0] NCCL INFO Trees [0] 9/-1/-1->8->16 [1] -1/-1/-1->8->15 [2] 9/-1/-1->8->15 [3] 9/-1/-1->8->15 [4] 9/-1/-1->8->15 [5] 9/-1/-1->8->15 [6] 9/-1/-1->8->15 [7] 9/-1/-1->8->15 [8] 9/16/0->8->24 [9] -1/-1/-1->8->15 [10] 9/-1/-1->8->15 [11] 9/-1/-1->8->15 [12] 9/-1/-1->8->15 [13] 9/-1/-1->8->15 [14] 9/-1/-1->8->15 [15] 9/-1/-1->8->15
+ip-26-0-166-214:624948:625104 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-214:624947:625105 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-214:624946:625107 [3] NCCL INFO Trees [0] 36/-1/-1->35->34 [1] 36/-1/-1->35->34 [2] 36/-1/-1->35->34 [3] 36/19/51->35->3 [4] -1/-1/-1->35->34 [5] 36/-1/-1->35->34 [6] 36/-1/-1->35->34 [7] 36/-1/-1->35->34 [8] 36/-1/-1->35->34 [9] 36/-1/-1->35->34 [10] 36/-1/-1->35->34 [11] 36/-1/-1->35->43 [12] -1/-1/-1->35->34 [13] 36/-1/-1->35->34 [14] 36/-1/-1->35->34 [15] 36/-1/-1->35->34
+ip-26-0-166-214:624943:625108 [0] NCCL INFO Trees [0] 33/16/48->32->0 [1] -1/-1/-1->32->39 [2] 33/-1/-1->32->39 [3] 33/-1/-1->32->39 [4] 33/-1/-1->32->39 [5] 33/-1/-1->32->39 [6] 33/-1/-1->32->39 [7] 33/-1/-1->32->39 [8] 33/-1/-1->32->40 [9] -1/-1/-1->32->39 [10] 33/-1/-1->32->39 [11] 33/-1/-1->32->39 [12] 33/-1/-1->32->39 [13] 33/-1/-1->32->39 [14] 33/-1/-1->32->39 [15] 33/-1/-1->32->39
+ip-26-0-166-214:624946:625107 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-15:127293:127454 [3] NCCL INFO Trees [0] 28/-1/-1->27->26 [1] 28/-1/-1->27->26 [2] 28/-1/-1->27->26 [3] 28/-1/-1->27->19 [4] -1/-1/-1->27->26 [5] 28/-1/-1->27->26 [6] 28/-1/-1->27->26 [7] 28/-1/-1->27->26 [8] 28/-1/-1->27->26 [9] 28/-1/-1->27->26 [10] 28/-1/-1->27->26 [11] 28/43/11->27->59 [12] -1/-1/-1->27->26 [13] 28/-1/-1->27->26 [14] 28/-1/-1->27->26 [15] 28/-1/-1->27->26
+ip-26-0-167-9:1039537:1039703 [0] NCCL INFO Trees [0] 57/-1/-1->56->48 [1] -1/-1/-1->56->63 [2] 57/-1/-1->56->63 [3] 57/-1/-1->56->63 [4] 57/-1/-1->56->63 [5] 57/-1/-1->56->63 [6] 57/-1/-1->56->63 [7] 57/-1/-1->56->63 [8] 57/24/-1->56->-1 [9] -1/-1/-1->56->63 [10] 57/-1/-1->56->63 [11] 57/-1/-1->56->63 [12] 57/-1/-1->56->63 [13] 57/-1/-1->56->63 [14] 57/-1/-1->56->63 [15] 57/-1/-1->56->63
+ip-26-0-167-9:1039537:1039703 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-36:104429:104594 [0] NCCL INFO Trees [0] 49/40/56->48->32 [1] -1/-1/-1->48->55 [2] 49/-1/-1->48->55 [3] 49/-1/-1->48->55 [4] 49/-1/-1->48->55 [5] 49/-1/-1->48->55 [6] 49/-1/-1->48->55 [7] 49/-1/-1->48->55 [8] 49/-1/-1->48->40 [9] -1/-1/-1->48->55 [10] 49/-1/-1->48->55 [11] 49/-1/-1->48->55 [12] 49/-1/-1->48->55 [13] 49/-1/-1->48->55 [14] 49/-1/-1->48->55 [15] 49/-1/-1->48->55
+ip-26-0-166-36:104432:104592 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-244:897243:897403 [3] NCCL INFO Trees [0] 44/-1/-1->43->42 [1] 44/-1/-1->43->42 [2] 44/-1/-1->43->42 [3] 44/-1/-1->43->51 [4] -1/-1/-1->43->42 [5] 44/-1/-1->43->42 [6] 44/-1/-1->43->42 [7] 44/-1/-1->43->42 [8] 44/-1/-1->43->42 [9] 44/-1/-1->43->42 [10] 44/-1/-1->43->42 [11] 44/51/35->43->27 [12] -1/-1/-1->43->42 [13] 44/-1/-1->43->42 [14] 44/-1/-1->43->42 [15] 44/-1/-1->43->42
+ip-26-0-165-213:9625:9787 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-214:624943:625108 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-214:624944:625106 [1] NCCL INFO Trees [0] 34/-1/-1->33->32 [1] 34/17/49->33->1 [2] -1/-1/-1->33->32 [3] 34/-1/-1->33->32 [4] 34/-1/-1->33->32 [5] 34/-1/-1->33->32 [6] 34/-1/-1->33->32 [7] 34/-1/-1->33->32 [8] 34/-1/-1->33->32 [9] 34/-1/-1->33->41 [10] -1/-1/-1->33->32 [11] 34/-1/-1->33->32 [12] 34/-1/-1->33->32 [13] 34/-1/-1->33->32 [14] 34/-1/-1->33->32 [15] 34/-1/-1->33->32
+ip-26-0-166-214:624945:625111 [2] NCCL INFO Trees [0] 35/-1/-1->34->33 [1] 35/-1/-1->34->33 [2] 35/18/50->34->2 [3] -1/-1/-1->34->33 [4] 35/-1/-1->34->33 [5] 35/-1/-1->34->33 [6] 35/-1/-1->34->33 [7] 35/-1/-1->34->33 [8] 35/-1/-1->34->33 [9] 35/-1/-1->34->33 [10] 35/-1/-1->34->42 [11] -1/-1/-1->34->33 [12] 35/-1/-1->34->33 [13] 35/-1/-1->34->33 [14] 35/-1/-1->34->33 [15] 35/-1/-1->34->33
+ip-26-0-166-214:624944:625106 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-214:624945:625111 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-15:127293:127454 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-15:127290:127456 [0] NCCL INFO Trees [0] 25/-1/-1->24->16 [1] -1/-1/-1->24->31 [2] 25/-1/-1->24->31 [3] 25/-1/-1->24->31 [4] 25/-1/-1->24->31 [5] 25/-1/-1->24->31 [6] 25/-1/-1->24->31 [7] 25/-1/-1->24->31 [8] 25/40/8->24->56 [9] -1/-1/-1->24->31 [10] 25/-1/-1->24->31 [11] 25/-1/-1->24->31 [12] 25/-1/-1->24->31 [13] 25/-1/-1->24->31 [14] 25/-1/-1->24->31 [15] 25/-1/-1->24->31
+ip-26-0-167-9:1039540:1039705 [3] NCCL INFO Trees [0] 60/-1/-1->59->58 [1] 60/-1/-1->59->58 [2] 60/-1/-1->59->58 [3] 60/-1/-1->59->51 [4] -1/-1/-1->59->58 [5] 60/-1/-1->59->58 [6] 60/-1/-1->59->58 [7] 60/-1/-1->59->58 [8] 60/-1/-1->59->58 [9] 60/-1/-1->59->58 [10] 60/-1/-1->59->58 [11] 60/27/-1->59->-1 [12] -1/-1/-1->59->58 [13] 60/-1/-1->59->58 [14] 60/-1/-1->59->58 [15] 60/-1/-1->59->58
+ip-26-0-167-9:1039539:1039698 [2] NCCL INFO Trees [0] 59/-1/-1->58->57 [1] 59/-1/-1->58->57 [2] 59/-1/-1->58->50 [3] -1/-1/-1->58->57 [4] 59/-1/-1->58->57 [5] 59/-1/-1->58->57 [6] 59/-1/-1->58->57 [7] 59/-1/-1->58->57 [8] 59/-1/-1->58->57 [9] 59/-1/-1->58->57 [10] 59/26/-1->58->-1 [11] -1/-1/-1->58->57 [12] 59/-1/-1->58->57 [13] 59/-1/-1->58->57 [14] 59/-1/-1->58->57 [15] 59/-1/-1->58->57
+ip-26-0-166-36:104434:104591 [5] NCCL INFO Trees [0] 54/-1/-1->53->52 [1] 54/-1/-1->53->52 [2] 54/-1/-1->53->52 [3] 54/-1/-1->53->52 [4] 54/-1/-1->53->52 [5] 54/45/61->53->37 [6] -1/-1/-1->53->52 [7] 54/-1/-1->53->52 [8] 54/-1/-1->53->52 [9] 54/-1/-1->53->52 [10] 54/-1/-1->53->52 [11] 54/-1/-1->53->52 [12] 54/-1/-1->53->52 [13] 54/-1/-1->53->45 [14] -1/-1/-1->53->52 [15] 54/-1/-1->53->52
+ip-26-0-166-36:104434:104591 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-244:897243:897403 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-15:127290:127456 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-167-9:1039539:1039698 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-167-9:1039540:1039705 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-36:104429:104594 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-15:127291:127455 [1] NCCL INFO Trees [0] 26/-1/-1->25->24 [1] 26/-1/-1->25->17 [2] -1/-1/-1->25->24 [3] 26/-1/-1->25->24 [4] 26/-1/-1->25->24 [5] 26/-1/-1->25->24 [6] 26/-1/-1->25->24 [7] 26/-1/-1->25->24 [8] 26/-1/-1->25->24 [9] 26/41/9->25->57 [10] -1/-1/-1->25->24 [11] 26/-1/-1->25->24 [12] 26/-1/-1->25->24 [13] 26/-1/-1->25->24 [14] 26/-1/-1->25->24 [15] 26/-1/-1->25->24
+ip-26-0-166-15:127291:127455 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-167-9:1039538:1039704 [1] NCCL INFO Trees [0] 58/-1/-1->57->56 [1] 58/-1/-1->57->49 [2] -1/-1/-1->57->56 [3] 58/-1/-1->57->56 [4] 58/-1/-1->57->56 [5] 58/-1/-1->57->56 [6] 58/-1/-1->57->56 [7] 58/-1/-1->57->56 [8] 58/-1/-1->57->56 [9] 58/25/-1->57->-1 [10] -1/-1/-1->57->56 [11] 58/-1/-1->57->56 [12] 58/-1/-1->57->56 [13] 58/-1/-1->57->56 [14] 58/-1/-1->57->56 [15] 58/-1/-1->57->56
+ip-26-0-167-9:1039538:1039704 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-213:9631:9788 [6] NCCL INFO Channel 00/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-165-213:9631:9788 [6] NCCL INFO Channel 08/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-165-213:9625:9787 [0] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-167-9:1039537:1039703 [0] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-166-244:897245:897401 [5] NCCL INFO Channel 00/0 : 45[5] -> 46[6] via P2P/IPC
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO Channel 00/0 : 60[4] -> 61[5] via P2P/IPC
+ip-26-0-166-244:897240:897406 [0] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-166-214:624943:625108 [0] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-166-244:897245:897401 [5] NCCL INFO Channel 08/0 : 45[5] -> 46[6] via P2P/IPC
+ip-26-0-167-9:1039543:1039699 [6] NCCL INFO Channel 00/0 : 62[6] -> 63[7] via P2P/IPC
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO Channel 08/0 : 60[4] -> 61[5] via P2P/IPC
+ip-26-0-166-125:762815:762983 [0] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-165-213:9630:9789 [5] NCCL INFO Channel 00/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-166-244:897244:897407 [4] NCCL INFO Channel 00/0 : 44[4] -> 45[5] via P2P/IPC
+ip-26-0-166-36:104429:104594 [0] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-166-15:127290:127456 [0] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-165-213:9629:9793 [4] NCCL INFO Channel 00/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-165-213:9626:9790 [1] NCCL INFO Channel 00/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-167-9:1039543:1039699 [6] NCCL INFO Channel 08/0 : 62[6] -> 63[7] via P2P/IPC
+ip-26-0-166-36:104435:104595 [6] NCCL INFO Channel 00/0 : 54[6] -> 55[7] via P2P/IPC
+ip-26-0-165-213:9630:9789 [5] NCCL INFO Channel 08/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-166-15:127294:127457 [4] NCCL INFO Channel 00/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-166-244:897244:897407 [4] NCCL INFO Channel 08/0 : 44[4] -> 45[5] via P2P/IPC
+ip-26-0-167-9:1039542:1039701 [5] NCCL INFO Channel 00/0 : 61[5] -> 62[6] via P2P/IPC
+ip-26-0-165-213:9626:9790 [1] NCCL INFO Channel 08/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-165-213:9629:9793 [4] NCCL INFO Channel 08/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-165-213:9627:9791 [2] NCCL INFO Channel 00/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-165-213:9628:9786 [3] NCCL INFO Channel 00/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-166-36:104430:104589 [1] NCCL INFO Channel 00/0 : 49[1] -> 50[2] via P2P/IPC
+ip-26-0-166-125:762819:762982 [4] NCCL INFO Channel 00/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-166-125:762820:762981 [5] NCCL INFO Channel 00/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-166-15:127295:127452 [5] NCCL INFO Channel 00/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-166-214:624946:625107 [3] NCCL INFO Channel 00/0 : 35[3] -> 36[4] via P2P/IPC
+ip-26-0-167-9:1039542:1039701 [5] NCCL INFO Channel 08/0 : 61[5] -> 62[6] via P2P/IPC
+ip-26-0-166-244:897242:897400 [2] NCCL INFO Channel 00/0 : 42[2] -> 43[3] via P2P/IPC
+ip-26-0-166-36:104435:104595 [6] NCCL INFO Channel 08/0 : 54[6] -> 55[7] via P2P/IPC
+ip-26-0-166-214:624948:625104 [5] NCCL INFO Channel 00/0 : 37[5] -> 38[6] via P2P/IPC
+ip-26-0-167-9:1039539:1039698 [2] NCCL INFO Channel 00/0 : 58[2] -> 59[3] via P2P/IPC
+ip-26-0-165-213:9627:9791 [2] NCCL INFO Channel 08/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-166-15:127296:127451 [6] NCCL INFO Channel 00/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-166-36:104433:104590 [4] NCCL INFO Channel 00/0 : 52[4] -> 53[5] via P2P/IPC
+ip-26-0-165-213:9628:9786 [3] NCCL INFO Channel 08/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-166-214:624947:625105 [4] NCCL INFO Channel 00/0 : 36[4] -> 37[5] via P2P/IPC
+ip-26-0-166-244:897241:897402 [1] NCCL INFO Channel 00/0 : 41[1] -> 42[2] via P2P/IPC
+ip-26-0-166-15:127294:127457 [4] NCCL INFO Channel 08/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-166-244:897246:897405 [6] NCCL INFO Channel 00/0 : 46[6] -> 47[7] via P2P/IPC
+ip-26-0-166-36:104431:104593 [2] NCCL INFO Channel 00/0 : 50[2] -> 51[3] via P2P/IPC
+ip-26-0-166-15:127292:127458 [2] NCCL INFO Channel 00/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-166-125:762820:762981 [5] NCCL INFO Channel 08/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-166-214:624946:625107 [3] NCCL INFO Channel 08/0 : 35[3] -> 36[4] via P2P/IPC
+ip-26-0-166-125:762819:762982 [4] NCCL INFO Channel 08/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-166-214:624945:625111 [2] NCCL INFO Channel 00/0 : 34[2] -> 35[3] via P2P/IPC
+ip-26-0-167-9:1039538:1039704 [1] NCCL INFO Channel 00/0 : 57[1] -> 58[2] via P2P/IPC
+ip-26-0-166-125:762818:762985 [3] NCCL INFO Channel 00/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-166-244:897242:897400 [2] NCCL INFO Channel 08/0 : 42[2] -> 43[3] via P2P/IPC
+ip-26-0-167-9:1039540:1039705 [3] NCCL INFO Channel 00/0 : 59[3] -> 60[4] via P2P/IPC
+ip-26-0-166-36:104430:104589 [1] NCCL INFO Channel 08/0 : 49[1] -> 50[2] via P2P/IPC
+ip-26-0-166-214:624948:625104 [5] NCCL INFO Channel 08/0 : 37[5] -> 38[6] via P2P/IPC
+ip-26-0-167-9:1039539:1039698 [2] NCCL INFO Channel 08/0 : 58[2] -> 59[3] via P2P/IPC
+ip-26-0-166-15:127295:127452 [5] NCCL INFO Channel 08/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-166-214:624947:625105 [4] NCCL INFO Channel 08/0 : 36[4] -> 37[5] via P2P/IPC
+ip-26-0-166-244:897243:897403 [3] NCCL INFO Channel 00/0 : 43[3] -> 44[4] via P2P/IPC
+ip-26-0-165-213:9627:9791 [2] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-165-213:9629:9793 [4] NCCL INFO Channel 05/0 : 12[4] -> 15[7] via P2P/IPC
+ip-26-0-166-244:897241:897402 [1] NCCL INFO Channel 08/0 : 41[1] -> 42[2] via P2P/IPC
+ip-26-0-165-213:9628:9786 [3] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-166-244:897246:897405 [6] NCCL INFO Channel 08/0 : 46[6] -> 47[7] via P2P/IPC
+ip-26-0-166-36:104434:104591 [5] NCCL INFO Channel 00/0 : 53[5] -> 54[6] via P2P/IPC
+ip-26-0-166-15:127296:127451 [6] NCCL INFO Channel 08/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-166-36:104433:104590 [4] NCCL INFO Channel 08/0 : 52[4] -> 53[5] via P2P/IPC
+ip-26-0-166-36:104432:104592 [3] NCCL INFO Channel 00/0 : 51[3] -> 52[4] via P2P/IPC
+ip-26-0-167-9:1039538:1039704 [1] NCCL INFO Channel 08/0 : 57[1] -> 58[2] via P2P/IPC
+ip-26-0-166-214:624945:625111 [2] NCCL INFO Channel 08/0 : 34[2] -> 35[3] via P2P/IPC
+ip-26-0-165-213:9629:9793 [4] NCCL INFO Channel 06/0 : 12[4] -> 15[7] via P2P/IPC
+ip-26-0-166-15:127293:127454 [3] NCCL INFO Channel 00/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-166-36:104431:104593 [2] NCCL INFO Channel 08/0 : 50[2] -> 51[3] via P2P/IPC
+ip-26-0-167-9:1039540:1039705 [3] NCCL INFO Channel 08/0 : 59[3] -> 60[4] via P2P/IPC
+ip-26-0-166-125:762818:762985 [3] NCCL INFO Channel 08/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-166-15:127292:127458 [2] NCCL INFO Channel 08/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-166-125:762816:762987 [1] NCCL INFO Channel 00/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-166-15:127291:127455 [1] NCCL INFO Channel 00/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-165-213:9629:9793 [4] NCCL INFO Channel 13/0 : 12[4] -> 15[7] via P2P/IPC
+ip-26-0-166-125:762817:762986 [2] NCCL INFO Channel 00/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-166-244:897243:897403 [3] NCCL INFO Channel 08/0 : 43[3] -> 44[4] via P2P/IPC
+ip-26-0-165-213:9629:9793 [4] NCCL INFO Channel 14/0 : 12[4] -> 15[7] via P2P/IPC
+ip-26-0-166-36:104434:104591 [5] NCCL INFO Channel 08/0 : 53[5] -> 54[6] via P2P/IPC
+ip-26-0-166-36:104432:104592 [3] NCCL INFO Channel 08/0 : 51[3] -> 52[4] via P2P/IPC
+ip-26-0-166-15:127293:127454 [3] NCCL INFO Channel 08/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-166-125:762816:762987 [1] NCCL INFO Channel 08/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-166-125:762817:762986 [2] NCCL INFO Channel 08/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-166-125:762819:762982 [4] NCCL INFO Channel 05/0 : 20[4] -> 23[7] via P2P/IPC
+ip-26-0-166-15:127291:127455 [1] NCCL INFO Channel 08/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-166-36:104431:104593 [2] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-166-36:104433:104590 [4] NCCL INFO Channel 05/0 : 52[4] -> 55[7] via P2P/IPC
+ip-26-0-166-36:104432:104592 [3] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-166-244:897242:897400 [2] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-166-125:762819:762982 [4] NCCL INFO Channel 06/0 : 20[4] -> 23[7] via P2P/IPC
+ip-26-0-166-15:127294:127457 [4] NCCL INFO Channel 05/0 : 28[4] -> 31[7] via P2P/IPC
+ip-26-0-166-125:762821:762988 [6] NCCL INFO Channel 00/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-166-244:897244:897407 [4] NCCL INFO Channel 05/0 : 44[4] -> 47[7] via P2P/IPC
+ip-26-0-166-15:127292:127458 [2] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-166-15:127293:127454 [3] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-166-36:104433:104590 [4] NCCL INFO Channel 06/0 : 52[4] -> 55[7] via P2P/IPC
+ip-26-0-166-244:897243:897403 [3] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-166-125:762818:762985 [3] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-166-125:762817:762986 [2] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-167-9:1039539:1039698 [2] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-166-214:624949:625110 [6] NCCL INFO Channel 00/0 : 38[6] -> 39[7] via P2P/IPC
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO Channel 05/0 : 60[4] -> 63[7] via P2P/IPC
+ip-26-0-166-125:762819:762982 [4] NCCL INFO Channel 13/0 : 20[4] -> 23[7] via P2P/IPC
+ip-26-0-166-214:624947:625105 [4] NCCL INFO Channel 05/0 : 36[4] -> 39[7] via P2P/IPC
+ip-26-0-166-244:897244:897407 [4] NCCL INFO Channel 06/0 : 44[4] -> 47[7] via P2P/IPC
+ip-26-0-166-36:104433:104590 [4] NCCL INFO Channel 13/0 : 52[4] -> 55[7] via P2P/IPC
+ip-26-0-167-9:1039540:1039705 [3] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-166-125:762821:762988 [6] NCCL INFO Channel 08/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-166-214:624946:625107 [3] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-166-244:897244:897407 [4] NCCL INFO Channel 13/0 : 44[4] -> 47[7] via P2P/IPC
+ip-26-0-166-36:104433:104590 [4] NCCL INFO Channel 14/0 : 52[4] -> 55[7] via P2P/IPC
+ip-26-0-166-125:762819:762982 [4] NCCL INFO Channel 14/0 : 20[4] -> 23[7] via P2P/IPC
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO Channel 06/0 : 60[4] -> 63[7] via P2P/IPC
+ip-26-0-166-214:624949:625110 [6] NCCL INFO Channel 08/0 : 38[6] -> 39[7] via P2P/IPC
+ip-26-0-166-214:624947:625105 [4] NCCL INFO Channel 06/0 : 36[4] -> 39[7] via P2P/IPC
+ip-26-0-166-244:897244:897407 [4] NCCL INFO Channel 14/0 : 44[4] -> 47[7] via P2P/IPC
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO Channel 13/0 : 60[4] -> 63[7] via P2P/IPC
+ip-26-0-166-214:624947:625105 [4] NCCL INFO Channel 13/0 : 36[4] -> 39[7] via P2P/IPC
+ip-26-0-166-15:127294:127457 [4] NCCL INFO Channel 06/0 : 28[4] -> 31[7] via P2P/IPC
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO Channel 14/0 : 60[4] -> 63[7] via P2P/IPC
+ip-26-0-166-214:624947:625105 [4] NCCL INFO Channel 14/0 : 36[4] -> 39[7] via P2P/IPC
+ip-26-0-166-15:127294:127457 [4] NCCL INFO Channel 13/0 : 28[4] -> 31[7] via P2P/IPC
+ip-26-0-166-15:127294:127457 [4] NCCL INFO Channel 14/0 : 28[4] -> 31[7] via P2P/IPC
+ip-26-0-166-214:624944:625106 [1] NCCL INFO Channel 00/0 : 33[1] -> 34[2] via P2P/IPC
+ip-26-0-166-214:624944:625106 [1] NCCL INFO Channel 08/0 : 33[1] -> 34[2] via P2P/IPC
+ip-26-0-166-214:624945:625111 [2] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-162-46:702883:703044 [3] NCCL INFO NVLS Head  1:  1  9 17 25 33 41 49 57
+ip-26-0-162-46:702884:703047 [4] NCCL INFO NVLS Head  7:  7 15 23 31 39 47 55 63
+ip-26-0-162-46:702882:703046 [2] NCCL INFO NVLS Head  7:  7 15 23 31 39 47 55 63
+ip-26-0-162-46:702883:703044 [3] NCCL INFO NVLS Head  2:  2 10 18 26 34 42 50 58
+ip-26-0-162-46:702881:703043 [1] NCCL INFO NVLS Head  1:  1  9 17 25 33 41 49 57
+ip-26-0-162-46:702883:703044 [3] NCCL INFO NVLS Head  3:  3 11 19 27 35 43 51 59
+ip-26-0-162-46:702887:703045 [7] NCCL INFO Trees [0] -1/-1/-1->7->6 [1] 0/-1/-1->7->6 [2] 0/-1/-1->7->6 [3] 0/-1/-1->7->6 [4] 0/-1/-1->7->6 [5] 0/-1/-1->7->6 [6] 0/-1/-1->7->6 [7] 0/39/-1->7->-1 [8] -1/-1/-1->7->6 [9] 0/-1/-1->7->6 [10] 0/-1/-1->7->6 [11] 0/-1/-1->7->6 [12] 0/-1/-1->7->6 [13] 0/-1/-1->7->6 [14] 0/-1/-1->7->6 [15] 0/-1/-1->7->15
+ip-26-0-162-46:702883:703044 [3] NCCL INFO NVLS Head  4:  4 12 20 28 36 44 52 60
+ip-26-0-162-46:702881:703043 [1] NCCL INFO NVLS Head  2:  2 10 18 26 34 42 50 58
+ip-26-0-162-46:702885:703048 [5] NCCL INFO Trees [0] 6/-1/-1->5->4 [1] 6/-1/-1->5->4 [2] 6/-1/-1->5->4 [3] 6/-1/-1->5->4 [4] 6/-1/-1->5->4 [5] 6/37/-1->5->-1 [6] -1/-1/-1->5->4 [7] 6/-1/-1->5->4 [8] 6/-1/-1->5->4 [9] 6/-1/-1->5->4 [10] 6/-1/-1->5->4 [11] 6/-1/-1->5->4 [12] 6/-1/-1->5->4 [13] 6/-1/-1->5->13 [14] -1/-1/-1->5->4 [15] 6/-1/-1->5->4
+ip-26-0-162-46:702883:703044 [3] NCCL INFO NVLS Head  5:  5 13 21 29 37 45 53 61
+ip-26-0-162-46:702887:703045 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-162-46:702880:703042 [0] NCCL INFO NVLS Head  0:  0  8 16 24 32 40 48 56
+ip-26-0-162-46:702881:703043 [1] NCCL INFO NVLS Head  3:  3 11 19 27 35 43 51 59
+ip-26-0-162-46:702885:703048 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-162-46:702883:703044 [3] NCCL INFO NVLS Head  6:  6 14 22 30 38 46 54 62
+ip-26-0-162-46:702881:703043 [1] NCCL INFO NVLS Head  4:  4 12 20 28 36 44 52 60
+ip-26-0-162-46:702880:703042 [0] NCCL INFO NVLS Head  1:  1  9 17 25 33 41 49 57
+ip-26-0-162-46:702883:703044 [3] NCCL INFO NVLS Head  7:  7 15 23 31 39 47 55 63
+ip-26-0-162-46:702881:703043 [1] NCCL INFO NVLS Head  5:  5 13 21 29 37 45 53 61
+ip-26-0-162-46:702880:703042 [0] NCCL INFO NVLS Head  2:  2 10 18 26 34 42 50 58
+ip-26-0-162-46:702881:703043 [1] NCCL INFO NVLS Head  6:  6 14 22 30 38 46 54 62
+ip-26-0-162-46:702880:703042 [0] NCCL INFO NVLS Head  3:  3 11 19 27 35 43 51 59
+ip-26-0-162-46:702881:703043 [1] NCCL INFO NVLS Head  7:  7 15 23 31 39 47 55 63
+ip-26-0-162-46:702880:703042 [0] NCCL INFO NVLS Head  4:  4 12 20 28 36 44 52 60
+ip-26-0-162-46:702882:703046 [2] NCCL INFO Trees [0] 3/-1/-1->2->1 [1] 3/-1/-1->2->1 [2] 3/34/-1->2->-1 [3] -1/-1/-1->2->1 [4] 3/-1/-1->2->1 [5] 3/-1/-1->2->1 [6] 3/-1/-1->2->1 [7] 3/-1/-1->2->1 [8] 3/-1/-1->2->1 [9] 3/-1/-1->2->1 [10] 3/-1/-1->2->10 [11] -1/-1/-1->2->1 [12] 3/-1/-1->2->1 [13] 3/-1/-1->2->1 [14] 3/-1/-1->2->1 [15] 3/-1/-1->2->1
+ip-26-0-162-46:702880:703042 [0] NCCL INFO NVLS Head  5:  5 13 21 29 37 45 53 61
+ip-26-0-162-46:702886:703049 [6] NCCL INFO Trees [0] 7/-1/-1->6->5 [1] 7/-1/-1->6->5 [2] 7/-1/-1->6->5 [3] 7/-1/-1->6->5 [4] 7/-1/-1->6->5 [5] 7/-1/-1->6->5 [6] 7/38/-1->6->-1 [7] -1/-1/-1->6->5 [8] 7/-1/-1->6->5 [9] 7/-1/-1->6->5 [10] 7/-1/-1->6->5 [11] 7/-1/-1->6->5 [12] 7/-1/-1->6->5 [13] 7/-1/-1->6->5 [14] 7/-1/-1->6->14 [15] -1/-1/-1->6->5
+ip-26-0-162-46:702884:703047 [4] NCCL INFO Trees [0] 5/-1/-1->4->3 [1] 5/-1/-1->4->3 [2] 5/-1/-1->4->3 [3] 5/-1/-1->4->3 [4] 5/36/-1->4->-1 [5] -1/-1/-1->4->3 [6] 5/-1/-1->4->3 [7] 5/-1/-1->4->3 [8] 5/-1/-1->4->3 [9] 5/-1/-1->4->3 [10] 5/-1/-1->4->3 [11] 5/-1/-1->4->3 [12] 5/-1/-1->4->12 [13] -1/-1/-1->4->3 [14] 5/-1/-1->4->3 [15] 5/-1/-1->4->3
+ip-26-0-162-46:702880:703042 [0] NCCL INFO NVLS Head  6:  6 14 22 30 38 46 54 62
+ip-26-0-162-46:702882:703046 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-162-46:702880:703042 [0] NCCL INFO NVLS Head  7:  7 15 23 31 39 47 55 63
+ip-26-0-162-46:702886:703049 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-162-46:702884:703047 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-162-46:702880:703042 [0] NCCL INFO Channel 00/16 :    0   1   2   3   4   5   6   7   8   9  10  11  12  13  14  15  16  17  18  19
+ip-26-0-162-46:702880:703042 [0] NCCL INFO Channel 01/16 :    0   7   6   5   4   3   2   9   8  15  14  13  12  11  10  17  16  23  22  21
+ip-26-0-162-46:702880:703042 [0] NCCL INFO Channel 02/16 :    0   7   6   5   4   3  10   9   8  15  14  13  12  11  18  17  16  23  22  21
+ip-26-0-162-46:702880:703042 [0] NCCL INFO Channel 03/16 :    0   7   6   5   4  11  10   9   8  15  14  13  12  19  18  17  16  23  22  21
+ip-26-0-162-46:702881:703043 [1] NCCL INFO Trees [0] 2/-1/-1->1->0 [1] 2/33/-1->1->-1 [2] -1/-1/-1->1->0 [3] 2/-1/-1->1->0 [4] 2/-1/-1->1->0 [5] 2/-1/-1->1->0 [6] 2/-1/-1->1->0 [7] 2/-1/-1->1->0 [8] 2/-1/-1->1->0 [9] 2/-1/-1->1->9 [10] -1/-1/-1->1->0 [11] 2/-1/-1->1->0 [12] 2/-1/-1->1->0 [13] 2/-1/-1->1->0 [14] 2/-1/-1->1->0 [15] 2/-1/-1->1->0
+ip-26-0-162-46:702883:703044 [3] NCCL INFO Trees [0] 4/-1/-1->3->2 [1] 4/-1/-1->3->2 [2] 4/-1/-1->3->2 [3] 4/35/-1->3->-1 [4] -1/-1/-1->3->2 [5] 4/-1/-1->3->2 [6] 4/-1/-1->3->2 [7] 4/-1/-1->3->2 [8] 4/-1/-1->3->2 [9] 4/-1/-1->3->2 [10] 4/-1/-1->3->2 [11] 4/-1/-1->3->11 [12] -1/-1/-1->3->2 [13] 4/-1/-1->3->2 [14] 4/-1/-1->3->2 [15] 4/-1/-1->3->2
+ip-26-0-162-46:702880:703042 [0] NCCL INFO Channel 04/16 :    0   7   6   5  12  11  10   9   8  15  14  13  20  19  18  17  16  23  22  21
+ip-26-0-162-46:702881:703043 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-162-46:702883:703044 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-162-46:702880:703042 [0] NCCL INFO Channel 05/16 :    0   4   7   6  13  11  10   9   8  12  15  14  21  19  18  17  16  20  23  22
+ip-26-0-162-46:702880:703042 [0] NCCL INFO Channel 06/16 :    0   5   4   7  14  11  10   9   8  13  12  15  22  19  18  17  16  21  20  23
+ip-26-0-162-46:702880:703042 [0] NCCL INFO Channel 07/16 :    0   6   5   4  15  11  10   9   8  14  13  12  23  19  18  17  16  22  21  20
+ip-26-0-162-46:702880:703042 [0] NCCL INFO Channel 08/16 :    0   1   2   3   4   5   6   7   8   9  10  11  12  13  14  15  16  17  18  19
+ip-26-0-162-46:702880:703042 [0] NCCL INFO Channel 09/16 :    0   7   6   5   4   3   2   9   8  15  14  13  12  11  10  17  16  23  22  21
+ip-26-0-162-46:702880:703042 [0] NCCL INFO Channel 10/16 :    0   7   6   5   4   3  10   9   8  15  14  13  12  11  18  17  16  23  22  21
+ip-26-0-162-46:702880:703042 [0] NCCL INFO Channel 11/16 :    0   7   6   5   4  11  10   9   8  15  14  13  12  19  18  17  16  23  22  21
+ip-26-0-162-46:702880:703042 [0] NCCL INFO Channel 12/16 :    0   7   6   5  12  11  10   9   8  15  14  13  20  19  18  17  16  23  22  21
+ip-26-0-162-46:702880:703042 [0] NCCL INFO Channel 13/16 :    0   4   7   6  13  11  10   9   8  12  15  14  21  19  18  17  16  20  23  22
+ip-26-0-162-46:702880:703042 [0] NCCL INFO Channel 14/16 :    0   5   4   7  14  11  10   9   8  13  12  15  22  19  18  17  16  21  20  23
+ip-26-0-162-46:702880:703042 [0] NCCL INFO Channel 15/16 :    0   6   5   4  15  11  10   9   8  14  13  12  23  19  18  17  16  22  21  20
+ip-26-0-162-46:702880:703042 [0] NCCL INFO Trees [0] 1/32/-1->0->-1 [1] -1/-1/-1->0->7 [2] 1/-1/-1->0->7 [3] 1/-1/-1->0->7 [4] 1/-1/-1->0->7 [5] 1/-1/-1->0->7 [6] 1/-1/-1->0->7 [7] 1/-1/-1->0->7 [8] 1/-1/-1->0->8 [9] -1/-1/-1->0->7 [10] 1/-1/-1->0->7 [11] 1/-1/-1->0->7 [12] 1/-1/-1->0->7 [13] 1/-1/-1->0->7 [14] 1/-1/-1->0->7 [15] 1/-1/-1->0->7
+ip-26-0-162-46:702880:703042 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-162-46:702880:703042 [0] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-162-46:702882:703046 [2] NCCL INFO Channel 00/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-162-46:702884:703047 [4] NCCL INFO Channel 00/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-162-46:702886:703049 [6] NCCL INFO Channel 00/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-162-46:702883:703044 [3] NCCL INFO Channel 00/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-162-46:702885:703048 [5] NCCL INFO Channel 00/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-162-46:702882:703046 [2] NCCL INFO Channel 08/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-162-46:702881:703043 [1] NCCL INFO Channel 00/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-162-46:702884:703047 [4] NCCL INFO Channel 08/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-162-46:702883:703044 [3] NCCL INFO Channel 08/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-162-46:702886:703049 [6] NCCL INFO Channel 08/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-162-46:702885:703048 [5] NCCL INFO Channel 08/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-162-46:702881:703043 [1] NCCL INFO Channel 08/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-162-46:702884:703047 [4] NCCL INFO Channel 05/0 : 4[4] -> 7[7] via P2P/IPC
+ip-26-0-162-46:702883:703044 [3] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-162-46:702882:703046 [2] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-162-46:702884:703047 [4] NCCL INFO Channel 06/0 : 4[4] -> 7[7] via P2P/IPC
+ip-26-0-162-46:702884:703047 [4] NCCL INFO Channel 13/0 : 4[4] -> 7[7] via P2P/IPC
+ip-26-0-162-46:702884:703047 [4] NCCL INFO Channel 14/0 : 4[4] -> 7[7] via P2P/IPC
+ip-26-0-165-213:9625:9787 [0] NCCL INFO Channel 00/0 : 7[7] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9632:9792 [7] NCCL INFO Channel 00/0 : 15[7] -> 16[0] [send] via NET/Libfabric/0(8)/GDRDMA
+ip-26-0-165-213:9625:9787 [0] NCCL INFO Channel 08/0 : 7[7] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9625:9787 [0] NCCL INFO Channel 00/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-165-213:9632:9792 [7] NCCL INFO Channel 08/0 : 15[7] -> 16[0] [send] via NET/Libfabric/0(8)/GDRDMA
+ip-26-0-166-36:104429:104594 [0] NCCL INFO Channel 00/0 : 47[7] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104436:104596 [7] NCCL INFO Channel 00/0 : 55[7] -> 56[0] [send] via NET/Libfabric/0(48)/GDRDMA
+ip-26-0-166-36:104429:104594 [0] NCCL INFO Channel 08/0 : 47[7] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104429:104594 [0] NCCL INFO Channel 00/0 : 48[0] -> 49[1] via P2P/IPC
+ip-26-0-166-36:104436:104596 [7] NCCL INFO Channel 08/0 : 55[7] -> 56[0] [send] via NET/Libfabric/0(48)/GDRDMA
+ip-26-0-165-213:9625:9787 [0] NCCL INFO Channel 08/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-166-36:104429:104594 [0] NCCL INFO Channel 08/0 : 48[0] -> 49[1] via P2P/IPC
+ip-26-0-165-213:9626:9790 [1] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-166-214:624943:625108 [0] NCCL INFO Channel 00/0 : 31[7] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624950:625109 [7] NCCL INFO Channel 00/0 : 39[7] -> 40[0] [send] via NET/Libfabric/0(32)/GDRDMA
+ip-26-0-166-214:624943:625108 [0] NCCL INFO Channel 08/0 : 31[7] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624943:625108 [0] NCCL INFO Channel 00/0 : 32[0] -> 33[1] via P2P/IPC
+ip-26-0-167-9:1039537:1039703 [0] NCCL INFO Channel 00/0 : 55[7] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-167-9:1039544:1039702 [7] NCCL INFO Channel 00/0 : 63[7] -> 0[0] [send] via NET/Libfabric/0(56)/GDRDMA
+ip-26-0-166-214:624950:625109 [7] NCCL INFO Channel 08/0 : 39[7] -> 40[0] [send] via NET/Libfabric/0(32)/GDRDMA
+ip-26-0-167-9:1039537:1039703 [0] NCCL INFO Channel 08/0 : 55[7] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-167-9:1039537:1039703 [0] NCCL INFO Channel 00/0 : 56[0] -> 57[1] via P2P/IPC
+ip-26-0-167-9:1039544:1039702 [7] NCCL INFO Channel 08/0 : 63[7] -> 0[0] [send] via NET/Libfabric/0(56)/GDRDMA
+ip-26-0-166-36:104430:104589 [1] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-166-15:127290:127456 [0] NCCL INFO Channel 00/0 : 23[7] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127290:127456 [0] NCCL INFO Channel 08/0 : 23[7] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127297:127453 [7] NCCL INFO Channel 00/0 : 31[7] -> 32[0] [send] via NET/Libfabric/0(24)/GDRDMA
+ip-26-0-166-244:897240:897406 [0] NCCL INFO Channel 00/0 : 39[7] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127290:127456 [0] NCCL INFO Channel 00/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-166-244:897247:897404 [7] NCCL INFO Channel 00/0 : 47[7] -> 48[0] [send] via NET/Libfabric/0(40)/GDRDMA
+ip-26-0-166-244:897247:897404 [7] NCCL INFO Channel 08/0 : 47[7] -> 48[0] [send] via NET/Libfabric/0(40)/GDRDMA
+ip-26-0-166-244:897240:897406 [0] NCCL INFO Channel 08/0 : 39[7] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897240:897406 [0] NCCL INFO Channel 00/0 : 40[0] -> 41[1] via P2P/IPC
+ip-26-0-166-214:624943:625108 [0] NCCL INFO Channel 08/0 : 32[0] -> 33[1] via P2P/IPC
+ip-26-0-166-36:104429:104594 [0] NCCL INFO Channel 05/0 : 48[0] -> 52[4] via P2P/IPC
+ip-26-0-166-15:127290:127456 [0] NCCL INFO Channel 08/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-166-15:127297:127453 [7] NCCL INFO Channel 08/0 : 31[7] -> 32[0] [send] via NET/Libfabric/0(24)/GDRDMA
+ip-26-0-166-36:104429:104594 [0] NCCL INFO Channel 13/0 : 48[0] -> 52[4] via P2P/IPC
+ip-26-0-166-214:624944:625106 [1] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-167-9:1039537:1039703 [0] NCCL INFO Channel 08/0 : 56[0] -> 57[1] via P2P/IPC
+ip-26-0-166-214:624943:625108 [0] NCCL INFO Channel 05/0 : 32[0] -> 36[4] via P2P/IPC
+ip-26-0-166-15:127291:127455 [1] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-166-214:624943:625108 [0] NCCL INFO Channel 13/0 : 32[0] -> 36[4] via P2P/IPC
+ip-26-0-166-244:897240:897406 [0] NCCL INFO Channel 08/0 : 40[0] -> 41[1] via P2P/IPC
+ip-26-0-167-9:1039538:1039704 [1] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-167-9:1039537:1039703 [0] NCCL INFO Channel 05/0 : 56[0] -> 60[4] via P2P/IPC
+ip-26-0-166-125:762815:762983 [0] NCCL INFO Channel 00/0 : 15[7] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762822:762984 [7] NCCL INFO Channel 00/0 : 23[7] -> 24[0] [send] via NET/Libfabric/0(16)/GDRDMA
+ip-26-0-166-125:762815:762983 [0] NCCL INFO Channel 08/0 : 15[7] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762822:762984 [7] NCCL INFO Channel 08/0 : 23[7] -> 24[0] [send] via NET/Libfabric/0(16)/GDRDMA
+ip-26-0-167-9:1039537:1039703 [0] NCCL INFO Channel 13/0 : 56[0] -> 60[4] via P2P/IPC
+ip-26-0-166-125:762815:762983 [0] NCCL INFO Channel 00/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-162-46:702880:703042 [0] NCCL INFO Channel 00/0 : 63[7] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127290:127456 [0] NCCL INFO Channel 05/0 : 24[0] -> 28[4] via P2P/IPC
+ip-26-0-162-46:702887:703045 [7] NCCL INFO Channel 00/0 : 7[7] -> 8[0] [send] via NET/Libfabric/0(0)/GDRDMA
+ip-26-0-162-46:702880:703042 [0] NCCL INFO Channel 08/0 : 63[7] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-46:702887:703045 [7] NCCL INFO Channel 08/0 : 7[7] -> 8[0] [send] via NET/Libfabric/0(0)/GDRDMA
+ip-26-0-162-46:702880:703042 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-15:127290:127456 [0] NCCL INFO Channel 13/0 : 24[0] -> 28[4] via P2P/IPC
+ip-26-0-166-125:762815:762983 [0] NCCL INFO Channel 08/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-165-213:9625:9787 [0] NCCL INFO Channel 05/0 : 8[0] -> 12[4] via P2P/IPC
+ip-26-0-165-213:9625:9787 [0] NCCL INFO Channel 13/0 : 8[0] -> 12[4] via P2P/IPC
+ip-26-0-166-15:127292:127458 [2] NCCL INFO Channel 02/0 : 19[3] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897240:897406 [0] NCCL INFO Channel 05/0 : 40[0] -> 44[4] via P2P/IPC
+ip-26-0-166-15:127292:127458 [2] NCCL INFO Channel 10/0 : 19[3] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127294:127457 [4] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-166-15:127290:127456 [0] NCCL INFO Channel 06/0 : 24[0] -> 29[5] via P2P/IPC
+ip-26-0-166-36:104429:104594 [0] NCCL INFO Channel 06/0 : 48[0] -> 53[5] via P2P/IPC
+ip-26-0-162-46:702880:703042 [0] NCCL INFO Channel 08/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-36:104433:104590 [4] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-166-244:897242:897400 [2] NCCL INFO Channel 02/0 : 35[3] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762816:762987 [1] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-166-244:897240:897406 [0] NCCL INFO Channel 13/0 : 40[0] -> 44[4] via P2P/IPC
+ip-26-0-166-15:127290:127456 [0] NCCL INFO Channel 14/0 : 24[0] -> 29[5] via P2P/IPC
+ip-26-0-166-125:762815:762983 [0] NCCL INFO Channel 05/0 : 16[0] -> 20[4] via P2P/IPC
+ip-26-0-166-244:897242:897400 [2] NCCL INFO Channel 10/0 : 35[3] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104429:104594 [0] NCCL INFO Channel 14/0 : 48[0] -> 53[5] via P2P/IPC
+ip-26-0-166-244:897241:897402 [1] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-167-9:1039539:1039698 [2] NCCL INFO Channel 02/0 : 51[3] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762815:762983 [0] NCCL INFO Channel 13/0 : 16[0] -> 20[4] via P2P/IPC
+ip-26-0-167-9:1039539:1039698 [2] NCCL INFO Channel 10/0 : 51[3] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127290:127456 [0] NCCL INFO Channel 07/0 : 24[0] -> 30[6] via P2P/IPC
+ip-26-0-166-36:104429:104594 [0] NCCL INFO Channel 07/0 : 48[0] -> 54[6] via P2P/IPC
+ip-26-0-166-15:127295:127452 [5] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-165-213:9628:9786 [3] NCCL INFO Channel 03/0 : 4[4] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-46:702881:703043 [1] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-166-36:104434:104591 [5] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-165-213:9628:9786 [3] NCCL INFO Channel 11/0 : 4[4] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127293:127454 [3] NCCL INFO Channel 03/0 : 20[4] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-46:702880:703042 [0] NCCL INFO Channel 05/0 : 0[0] -> 4[4] via P2P/IPC
+ip-26-0-166-15:127290:127456 [0] NCCL INFO Channel 15/0 : 24[0] -> 30[6] via P2P/IPC
+ip-26-0-166-244:897243:897403 [3] NCCL INFO Channel 03/0 : 36[4] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127293:127454 [3] NCCL INFO Channel 11/0 : 20[4] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104429:104594 [0] NCCL INFO Channel 15/0 : 48[0] -> 54[6] via P2P/IPC
+ip-26-0-166-244:897244:897407 [4] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-166-244:897240:897406 [0] NCCL INFO Channel 06/0 : 40[0] -> 45[5] via P2P/IPC
+ip-26-0-166-244:897243:897403 [3] NCCL INFO Channel 11/0 : 36[4] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-46:702882:703046 [2] NCCL INFO Channel 02/0 : 59[3] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127293:127454 [3] NCCL INFO Channel 02/0 : 27[3] -> 34[2] [send] via NET/Libfabric/2(26)/GDRDMA
+ip-26-0-166-15:127293:127454 [3] NCCL INFO Channel 10/0 : 27[3] -> 34[2] [send] via NET/Libfabric/2(26)/GDRDMA
+ip-26-0-166-15:127290:127456 [0] NCCL INFO Channel 01/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-166-244:897243:897403 [3] NCCL INFO Channel 02/0 : 43[3] -> 50[2] [send] via NET/Libfabric/2(42)/GDRDMA
+ip-26-0-162-46:702882:703046 [2] NCCL INFO Channel 10/0 : 59[3] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762819:762982 [4] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-166-15:127296:127451 [6] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-162-46:702880:703042 [0] NCCL INFO Channel 13/0 : 0[0] -> 4[4] via P2P/IPC
+ip-26-0-166-125:762815:762983 [0] NCCL INFO Channel 06/0 : 16[0] -> 21[5] via P2P/IPC
+ip-26-0-166-244:897240:897406 [0] NCCL INFO Channel 14/0 : 40[0] -> 45[5] via P2P/IPC
+ip-26-0-166-36:104435:104595 [6] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-166-244:897243:897403 [3] NCCL INFO Channel 10/0 : 43[3] -> 50[2] [send] via NET/Libfabric/2(42)/GDRDMA
+ip-26-0-166-36:104429:104594 [0] NCCL INFO Channel 01/0 : 48[0] -> 55[7] via P2P/IPC
+ip-26-0-166-15:127290:127456 [0] NCCL INFO Channel 02/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-166-125:762815:762983 [0] NCCL INFO Channel 14/0 : 16[0] -> 21[5] via P2P/IPC
+ip-26-0-166-36:104429:104594 [0] NCCL INFO Channel 02/0 : 48[0] -> 55[7] via P2P/IPC
+ip-26-0-166-244:897240:897406 [0] NCCL INFO Channel 07/0 : 40[0] -> 46[6] via P2P/IPC
+ip-26-0-166-15:127290:127456 [0] NCCL INFO Channel 03/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-166-244:897245:897401 [5] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-166-36:104429:104594 [0] NCCL INFO Channel 03/0 : 48[0] -> 55[7] via P2P/IPC
+ip-26-0-166-15:127290:127456 [0] NCCL INFO Channel 04/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-162-46:702884:703047 [4] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-162-46:702880:703042 [0] NCCL INFO Channel 06/0 : 0[0] -> 5[5] via P2P/IPC
+ip-26-0-166-125:762820:762981 [5] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-166-244:897240:897406 [0] NCCL INFO Channel 15/0 : 40[0] -> 46[6] via P2P/IPC
+ip-26-0-165-213:9627:9791 [2] NCCL INFO Channel 02/0 : 3[3] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762815:762983 [0] NCCL INFO Channel 07/0 : 16[0] -> 22[6] via P2P/IPC
+ip-26-0-166-36:104429:104594 [0] NCCL INFO Channel 04/0 : 48[0] -> 55[7] via P2P/IPC
+ip-26-0-166-15:127290:127456 [0] NCCL INFO Channel 09/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-166-244:897242:897400 [2] NCCL INFO Channel 01/0 : 42[2] -> 49[1] [send] via NET/Libfabric/1(41)/GDRDMA
+ip-26-0-166-36:104431:104593 [2] NCCL INFO Channel 02/0 : 43[3] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9628:9786 [3] NCCL INFO Channel 02/0 : 11[3] -> 18[2] [send] via NET/Libfabric/2(10)/GDRDMA
+ip-26-0-166-36:104429:104594 [0] NCCL INFO Channel 09/0 : 48[0] -> 55[7] via P2P/IPC
+ip-26-0-165-213:9627:9791 [2] NCCL INFO Channel 10/0 : 3[3] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104431:104593 [2] NCCL INFO Channel 10/0 : 43[3] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762815:762983 [0] NCCL INFO Channel 15/0 : 16[0] -> 22[6] via P2P/IPC
+ip-26-0-165-213:9628:9786 [3] NCCL INFO Channel 10/0 : 11[3] -> 18[2] [send] via NET/Libfabric/2(10)/GDRDMA
+ip-26-0-162-46:702880:703042 [0] NCCL INFO Channel 14/0 : 0[0] -> 5[5] via P2P/IPC
+ip-26-0-166-244:897240:897406 [0] NCCL INFO Channel 01/0 : 40[0] -> 47[7] via P2P/IPC
+ip-26-0-166-244:897246:897405 [6] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-166-15:127290:127456 [0] NCCL INFO Channel 10/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-166-214:624946:625107 [3] NCCL INFO Channel 03/0 : 28[4] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104429:104594 [0] NCCL INFO Channel 10/0 : 48[0] -> 55[7] via P2P/IPC
+ip-26-0-166-214:624946:625107 [3] NCCL INFO Channel 11/0 : 28[4] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-167-9:1039540:1039705 [3] NCCL INFO Channel 03/0 : 52[4] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897240:897406 [0] NCCL INFO Channel 02/0 : 40[0] -> 47[7] via P2P/IPC
+ip-26-0-166-15:127290:127456 [0] NCCL INFO Channel 11/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-166-214:624943:625108 [0] NCCL INFO Channel 06/0 : 32[0] -> 37[5] via P2P/IPC
+ip-26-0-166-125:762821:762988 [6] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-166-125:762815:762983 [0] NCCL INFO Channel 01/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-167-9:1039537:1039703 [0] NCCL INFO Channel 06/0 : 56[0] -> 61[5] via P2P/IPC
+ip-26-0-167-9:1039540:1039705 [3] NCCL INFO Channel 11/0 : 52[4] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-162-46:702882:703046 [2] NCCL INFO Channel 01/0 : 2[2] -> 9[1] [send] via NET/Libfabric/1(1)/GDRDMA
+ip-26-0-166-36:104429:104594 [0] NCCL INFO Channel 11/0 : 48[0] -> 55[7] via P2P/IPC
+ip-26-0-166-244:897240:897406 [0] NCCL INFO Channel 03/0 : 40[0] -> 47[7] via P2P/IPC
+ip-26-0-166-15:127290:127456 [0] NCCL INFO Channel 12/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-162-46:702883:703044 [3] NCCL INFO Channel 03/0 : 60[4] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-46:702882:703046 [2] NCCL INFO Channel 09/0 : 2[2] -> 9[1] [send] via NET/Libfabric/1(1)/GDRDMA
+ip-26-0-162-46:702880:703042 [0] NCCL INFO Channel 07/0 : 0[0] -> 6[6] via P2P/IPC
+ip-26-0-166-36:104432:104592 [3] NCCL INFO Channel 03/0 : 44[4] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-46:702883:703044 [3] NCCL INFO Channel 11/0 : 60[4] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762815:762983 [0] NCCL INFO Channel 02/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-166-244:897240:897406 [0] NCCL INFO Channel 04/0 : 40[0] -> 47[7] via P2P/IPC
+ip-26-0-166-214:624943:625108 [0] NCCL INFO Channel 14/0 : 32[0] -> 37[5] via P2P/IPC
+ip-26-0-166-15:127297:127453 [7] NCCL INFO Channel 06/0 : 31[7] -> 38[6] [send] via NET/Libfabric/6(30)/GDRDMA
+ip-26-0-167-9:1039540:1039705 [3] NCCL INFO Channel 02/0 : 59[3] -> 2[2] [send] via NET/Libfabric/2(58)/GDRDMA
+ip-26-0-166-36:104429:104594 [0] NCCL INFO Channel 12/0 : 48[0] -> 55[7] via P2P/IPC
+ip-26-0-166-36:104432:104592 [3] NCCL INFO Channel 11/0 : 44[4] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-46:702880:703042 [0] NCCL INFO Channel 15/0 : 0[0] -> 6[6] via P2P/IPC
+ip-26-0-166-36:104436:104596 [7] NCCL INFO Channel 06/0 : 55[7] -> 62[6] [send] via NET/Libfabric/6(54)/GDRDMA
+ip-26-0-166-15:127297:127453 [7] NCCL INFO Channel 14/0 : 31[7] -> 38[6] [send] via NET/Libfabric/6(30)/GDRDMA
+ip-26-0-167-9:1039540:1039705 [3] NCCL INFO Channel 10/0 : 59[3] -> 2[2] [send] via NET/Libfabric/2(58)/GDRDMA
+ip-26-0-167-9:1039537:1039703 [0] NCCL INFO Channel 14/0 : 56[0] -> 61[5] via P2P/IPC
+ip-26-0-166-125:762815:762983 [0] NCCL INFO Channel 03/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-162-46:702883:703044 [3] NCCL INFO Channel 02/0 : 3[3] -> 10[2] [send] via NET/Libfabric/2(2)/GDRDMA
+ip-26-0-166-244:897240:897406 [0] NCCL INFO Channel 09/0 : 40[0] -> 47[7] via P2P/IPC
+ip-26-0-166-36:104432:104592 [3] NCCL INFO Channel 02/0 : 51[3] -> 58[2] [send] via NET/Libfabric/2(50)/GDRDMA
+ip-26-0-165-213:9625:9787 [0] NCCL INFO Channel 06/0 : 8[0] -> 13[5] via P2P/IPC
+ip-26-0-162-46:702883:703044 [3] NCCL INFO Channel 10/0 : 3[3] -> 10[2] [send] via NET/Libfabric/2(2)/GDRDMA
+ip-26-0-166-36:104432:104592 [3] NCCL INFO Channel 10/0 : 51[3] -> 58[2] [send] via NET/Libfabric/2(50)/GDRDMA
+ip-26-0-165-213:9629:9793 [4] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-166-125:762815:762983 [0] NCCL INFO Channel 04/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-166-214:624943:625108 [0] NCCL INFO Channel 07/0 : 32[0] -> 38[6] via P2P/IPC
+ip-26-0-167-9:1039537:1039703 [0] NCCL INFO Channel 07/0 : 56[0] -> 62[6] via P2P/IPC
+ip-26-0-165-213:9625:9787 [0] NCCL INFO Channel 14/0 : 8[0] -> 13[5] via P2P/IPC
+ip-26-0-166-244:897240:897406 [0] NCCL INFO Channel 10/0 : 40[0] -> 47[7] via P2P/IPC
+ip-26-0-162-46:702880:703042 [0] NCCL INFO Channel 01/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-166-125:762815:762983 [0] NCCL INFO Channel 09/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-167-9:1039542:1039701 [5] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-166-244:897240:897406 [0] NCCL INFO Channel 11/0 : 40[0] -> 47[7] via P2P/IPC
+ip-26-0-167-9:1039537:1039703 [0] NCCL INFO Channel 15/0 : 56[0] -> 62[6] via P2P/IPC
+ip-26-0-162-46:702880:703042 [0] NCCL INFO Channel 02/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-166-214:624943:625108 [0] NCCL INFO Channel 15/0 : 32[0] -> 38[6] via P2P/IPC
+ip-26-0-166-244:897247:897404 [7] NCCL INFO Channel 06/0 : 47[7] -> 54[6] [send] via NET/Libfabric/6(46)/GDRDMA
+ip-26-0-165-213:9625:9787 [0] NCCL INFO Channel 07/0 : 8[0] -> 14[6] via P2P/IPC
+ip-26-0-165-213:9630:9789 [5] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-166-244:897240:897406 [0] NCCL INFO Channel 12/0 : 40[0] -> 47[7] via P2P/IPC
+ip-26-0-162-46:702880:703042 [0] NCCL INFO Channel 03/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-165-213:9625:9787 [0] NCCL INFO Channel 15/0 : 8[0] -> 14[6] via P2P/IPC
+ip-26-0-166-125:762815:762983 [0] NCCL INFO Channel 10/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-167-9:1039537:1039703 [0] NCCL INFO Channel 01/0 : 56[0] -> 63[7] via P2P/IPC
+ip-26-0-166-214:624943:625108 [0] NCCL INFO Channel 01/0 : 32[0] -> 39[7] via P2P/IPC
+ip-26-0-167-9:1039543:1039699 [6] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-162-46:702880:703042 [0] NCCL INFO Channel 04/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-166-125:762815:762983 [0] NCCL INFO Channel 11/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-166-125:762817:762986 [2] NCCL INFO Channel 02/0 : 11[3] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-167-9:1039537:1039703 [0] NCCL INFO Channel 02/0 : 56[0] -> 63[7] via P2P/IPC
+ip-26-0-166-125:762822:762984 [7] NCCL INFO Channel 06/0 : 23[7] -> 30[6] [send] via NET/Libfabric/6(22)/GDRDMA
+ip-26-0-165-213:9625:9787 [0] NCCL INFO Channel 01/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-165-213:9631:9788 [6] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-166-125:762817:762986 [2] NCCL INFO Channel 10/0 : 11[3] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-46:702880:703042 [0] NCCL INFO Channel 09/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-166-125:762818:762985 [3] NCCL INFO Channel 03/0 : 12[4] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762815:762983 [0] NCCL INFO Channel 12/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-167-9:1039537:1039703 [0] NCCL INFO Channel 03/0 : 56[0] -> 63[7] via P2P/IPC
+ip-26-0-165-213:9625:9787 [0] NCCL INFO Channel 02/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-166-125:762818:762985 [3] NCCL INFO Channel 11/0 : 12[4] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624949:625110 [6] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-162-46:702885:703048 [5] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-162-46:702886:703049 [6] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-165-213:9625:9787 [0] NCCL INFO Channel 03/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-167-9:1039537:1039703 [0] NCCL INFO Channel 04/0 : 56[0] -> 63[7] via P2P/IPC
+ip-26-0-166-214:624947:625105 [4] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-166-125:762818:762985 [3] NCCL INFO Channel 02/0 : 19[3] -> 26[2] [send] via NET/Libfabric/2(18)/GDRDMA
+ip-26-0-166-125:762818:762985 [3] NCCL INFO Channel 10/0 : 19[3] -> 26[2] [send] via NET/Libfabric/2(18)/GDRDMA
+ip-26-0-162-46:702880:703042 [0] NCCL INFO Channel 10/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-165-213:9625:9787 [0] NCCL INFO Channel 04/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-167-9:1039537:1039703 [0] NCCL INFO Channel 09/0 : 56[0] -> 63[7] via P2P/IPC
+ip-26-0-165-213:9632:9792 [7] NCCL INFO Channel 06/0 : 15[7] -> 22[6] [send] via NET/Libfabric/6(14)/GDRDMA
+ip-26-0-166-214:624943:625108 [0] NCCL INFO Channel 02/0 : 32[0] -> 39[7] via P2P/IPC
+ip-26-0-162-46:702880:703042 [0] NCCL INFO Channel 11/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-165-213:9625:9787 [0] NCCL INFO Channel 09/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-166-214:624948:625104 [5] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-167-9:1039537:1039703 [0] NCCL INFO Channel 10/0 : 56[0] -> 63[7] via P2P/IPC
+ip-26-0-162-46:702887:703045 [7] NCCL INFO Channel 06/0 : 7[7] -> 14[6] [send] via NET/Libfabric/6(6)/GDRDMA
+ip-26-0-166-214:624945:625111 [2] NCCL INFO Channel 02/0 : 27[3] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-46:702880:703042 [0] NCCL INFO Channel 12/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-166-214:624943:625108 [0] NCCL INFO Channel 03/0 : 32[0] -> 39[7] via P2P/IPC
+ip-26-0-167-9:1039537:1039703 [0] NCCL INFO Channel 11/0 : 56[0] -> 63[7] via P2P/IPC
+ip-26-0-166-214:624945:625111 [2] NCCL INFO Channel 10/0 : 27[3] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9625:9787 [0] NCCL INFO Channel 10/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-166-214:624946:625107 [3] NCCL INFO Channel 02/0 : 35[3] -> 42[2] [send] via NET/Libfabric/2(34)/GDRDMA
+ip-26-0-166-214:624946:625107 [3] NCCL INFO Channel 10/0 : 35[3] -> 42[2] [send] via NET/Libfabric/2(34)/GDRDMA
+ip-26-0-166-214:624950:625109 [7] NCCL INFO Channel 06/0 : 39[7] -> 46[6] [send] via NET/Libfabric/6(38)/GDRDMA
+ip-26-0-166-214:624943:625108 [0] NCCL INFO Channel 04/0 : 32[0] -> 39[7] via P2P/IPC
+ip-26-0-167-9:1039537:1039703 [0] NCCL INFO Channel 12/0 : 56[0] -> 63[7] via P2P/IPC
+ip-26-0-165-213:9625:9787 [0] NCCL INFO Channel 11/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-166-214:624943:625108 [0] NCCL INFO Channel 09/0 : 32[0] -> 39[7] via P2P/IPC
+ip-26-0-165-213:9625:9787 [0] NCCL INFO Channel 12/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-166-214:624943:625108 [0] NCCL INFO Channel 10/0 : 32[0] -> 39[7] via P2P/IPC
+ip-26-0-166-214:624943:625108 [0] NCCL INFO Channel 11/0 : 32[0] -> 39[7] via P2P/IPC
+ip-26-0-166-214:624943:625108 [0] NCCL INFO Channel 12/0 : 32[0] -> 39[7] via P2P/IPC
+ip-26-0-166-36:104430:104589 [1] NCCL INFO Channel 01/0 : 42[2] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104430:104589 [1] NCCL INFO Channel 09/0 : 42[2] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104431:104593 [2] NCCL INFO Channel 01/0 : 50[2] -> 57[1] [send] via NET/Libfabric/1(49)/GDRDMA
+ip-26-0-166-36:104431:104593 [2] NCCL INFO Channel 09/0 : 50[2] -> 57[1] [send] via NET/Libfabric/1(49)/GDRDMA
+ip-26-0-165-213:9626:9790 [1] NCCL INFO Channel 01/0 : 2[2] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9626:9790 [1] NCCL INFO Channel 09/0 : 2[2] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9627:9791 [2] NCCL INFO Channel 01/0 : 10[2] -> 17[1] [send] via NET/Libfabric/1(9)/GDRDMA
+ip-26-0-165-213:9627:9791 [2] NCCL INFO Channel 09/0 : 10[2] -> 17[1] [send] via NET/Libfabric/1(9)/GDRDMA
+ip-26-0-166-15:127291:127455 [1] NCCL INFO Channel 01/0 : 18[2] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127291:127455 [1] NCCL INFO Channel 09/0 : 18[2] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127292:127458 [2] NCCL INFO Channel 01/0 : 26[2] -> 33[1] [send] via NET/Libfabric/1(25)/GDRDMA
+ip-26-0-166-36:104433:104590 [4] NCCL INFO Channel 04/0 : 45[5] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127292:127458 [2] NCCL INFO Channel 09/0 : 26[2] -> 33[1] [send] via NET/Libfabric/1(25)/GDRDMA
+ip-26-0-166-36:104433:104590 [4] NCCL INFO Channel 12/0 : 45[5] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104433:104590 [4] NCCL INFO Channel 03/0 : 52[4] -> 59[3] [send] via NET/Libfabric/3(51)/GDRDMA
+ip-26-0-166-36:104433:104590 [4] NCCL INFO Channel 11/0 : 52[4] -> 59[3] [send] via NET/Libfabric/3(51)/GDRDMA
+ip-26-0-166-214:624944:625106 [1] NCCL INFO Channel 01/0 : 26[2] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624944:625106 [1] NCCL INFO Channel 09/0 : 26[2] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624945:625111 [2] NCCL INFO Channel 01/0 : 34[2] -> 41[1] [send] via NET/Libfabric/1(33)/GDRDMA
+ip-26-0-166-214:624945:625111 [2] NCCL INFO Channel 09/0 : 34[2] -> 41[1] [send] via NET/Libfabric/1(33)/GDRDMA
+ip-26-0-167-9:1039538:1039704 [1] NCCL INFO Channel 01/0 : 50[2] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-167-9:1039538:1039704 [1] NCCL INFO Channel 09/0 : 50[2] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-167-9:1039539:1039698 [2] NCCL INFO Channel 01/0 : 58[2] -> 1[1] [send] via NET/Libfabric/1(57)/GDRDMA
+ip-26-0-167-9:1039539:1039698 [2] NCCL INFO Channel 09/0 : 58[2] -> 1[1] [send] via NET/Libfabric/1(57)/GDRDMA
+ip-26-0-166-244:897241:897402 [1] NCCL INFO Channel 01/0 : 34[2] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9626:9790 [1] NCCL INFO Channel 01/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-166-15:127294:127457 [4] NCCL INFO Channel 04/0 : 21[5] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897241:897402 [1] NCCL INFO Channel 09/0 : 34[2] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9626:9790 [1] NCCL INFO Channel 02/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-166-15:127294:127457 [4] NCCL INFO Channel 12/0 : 21[5] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897242:897400 [2] NCCL INFO Channel 09/0 : 42[2] -> 49[1] [send] via NET/Libfabric/1(41)/GDRDMA
+ip-26-0-165-213:9626:9790 [1] NCCL INFO Channel 03/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-166-15:127294:127457 [4] NCCL INFO Channel 03/0 : 28[4] -> 35[3] [send] via NET/Libfabric/3(27)/GDRDMA
+ip-26-0-166-244:897244:897407 [4] NCCL INFO Channel 04/0 : 37[5] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897244:897407 [4] NCCL INFO Channel 12/0 : 37[5] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127294:127457 [4] NCCL INFO Channel 11/0 : 28[4] -> 35[3] [send] via NET/Libfabric/3(27)/GDRDMA
+ip-26-0-165-213:9626:9790 [1] NCCL INFO Channel 04/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-165-213:9626:9790 [1] NCCL INFO Channel 05/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-166-244:897244:897407 [4] NCCL INFO Channel 03/0 : 44[4] -> 51[3] [send] via NET/Libfabric/3(43)/GDRDMA
+ip-26-0-166-244:897244:897407 [4] NCCL INFO Channel 11/0 : 44[4] -> 51[3] [send] via NET/Libfabric/3(43)/GDRDMA
+ip-26-0-165-213:9626:9790 [1] NCCL INFO Channel 06/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-166-36:104434:104591 [5] NCCL INFO Channel 05/0 : 46[6] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762820:762981 [5] NCCL INFO Channel 05/0 : 14[6] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104434:104591 [5] NCCL INFO Channel 13/0 : 46[6] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762820:762981 [5] NCCL INFO Channel 13/0 : 14[6] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9626:9790 [1] NCCL INFO Channel 07/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-166-36:104434:104591 [5] NCCL INFO Channel 04/0 : 53[5] -> 60[4] [send] via NET/Libfabric/4(52)/GDRDMA
+ip-26-0-165-213:9626:9790 [1] NCCL INFO Channel 09/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-166-36:104434:104591 [5] NCCL INFO Channel 12/0 : 53[5] -> 60[4] [send] via NET/Libfabric/4(52)/GDRDMA
+ip-26-0-165-213:9626:9790 [1] NCCL INFO Channel 10/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-166-15:127295:127452 [5] NCCL INFO Channel 05/0 : 22[6] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127295:127452 [5] NCCL INFO Channel 13/0 : 22[6] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9626:9790 [1] NCCL INFO Channel 11/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-166-125:762819:762982 [4] NCCL INFO Channel 04/0 : 13[5] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762819:762982 [4] NCCL INFO Channel 12/0 : 13[5] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127296:127451 [6] NCCL INFO Channel 06/0 : 23[7] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9626:9790 [1] NCCL INFO Channel 12/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-166-125:762820:762981 [5] NCCL INFO Channel 04/0 : 21[5] -> 28[4] [send] via NET/Libfabric/4(20)/GDRDMA
+ip-26-0-166-15:127295:127452 [5] NCCL INFO Channel 04/0 : 29[5] -> 36[4] [send] via NET/Libfabric/4(28)/GDRDMA
+ip-26-0-162-46:702884:703047 [4] NCCL INFO Channel 04/0 : 61[5] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127295:127452 [5] NCCL INFO Channel 12/0 : 29[5] -> 36[4] [send] via NET/Libfabric/4(28)/GDRDMA
+ip-26-0-166-15:127296:127451 [6] NCCL INFO Channel 14/0 : 23[7] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762820:762981 [5] NCCL INFO Channel 12/0 : 21[5] -> 28[4] [send] via NET/Libfabric/4(20)/GDRDMA
+ip-26-0-162-46:702884:703047 [4] NCCL INFO Channel 12/0 : 61[5] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897246:897405 [6] NCCL INFO Channel 06/0 : 39[7] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9626:9790 [1] NCCL INFO Channel 13/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-166-125:762819:762982 [4] NCCL INFO Channel 03/0 : 20[4] -> 27[3] [send] via NET/Libfabric/3(19)/GDRDMA
+ip-26-0-166-15:127296:127451 [6] NCCL INFO Channel 05/0 : 30[6] -> 37[5] [send] via NET/Libfabric/5(29)/GDRDMA
+ip-26-0-166-244:897247:897404 [7] NCCL INFO Channel 14/0 : 47[7] -> 54[6] [send] via NET/Libfabric/6(46)/GDRDMA
+ip-26-0-166-15:127296:127451 [6] NCCL INFO Channel 13/0 : 30[6] -> 37[5] [send] via NET/Libfabric/5(29)/GDRDMA
+ip-26-0-166-125:762819:762982 [4] NCCL INFO Channel 11/0 : 20[4] -> 27[3] [send] via NET/Libfabric/3(19)/GDRDMA
+ip-26-0-165-213:9626:9790 [1] NCCL INFO Channel 14/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO Channel 04/0 : 53[5] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-46:702884:703047 [4] NCCL INFO Channel 03/0 : 4[4] -> 11[3] [send] via NET/Libfabric/3(3)/GDRDMA
+ip-26-0-166-244:897246:897405 [6] NCCL INFO Channel 14/0 : 39[7] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO Channel 12/0 : 53[5] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-46:702882:703046 [2] NCCL INFO Channel 02/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-162-46:702884:703047 [4] NCCL INFO Channel 11/0 : 4[4] -> 11[3] [send] via NET/Libfabric/3(3)/GDRDMA
+ip-26-0-165-213:9626:9790 [1] NCCL INFO Channel 15/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-166-15:127292:127458 [2] NCCL INFO Channel 02/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-167-9:1039538:1039704 [1] NCCL INFO Channel 01/0 : 57[1] -> 56[0] via P2P/IPC
+ip-26-0-162-46:702882:703046 [2] NCCL INFO Channel 03/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-36:104435:104595 [6] NCCL INFO Channel 06/0 : 47[7] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO Channel 03/0 : 60[4] -> 3[3] [send] via NET/Libfabric/3(59)/GDRDMA
+ip-26-0-166-244:897245:897401 [5] NCCL INFO Channel 05/0 : 38[6] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127292:127458 [2] NCCL INFO Channel 03/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO Channel 11/0 : 60[4] -> 3[3] [send] via NET/Libfabric/3(59)/GDRDMA
+ip-26-0-166-36:104436:104596 [7] NCCL INFO Channel 14/0 : 55[7] -> 62[6] [send] via NET/Libfabric/6(54)/GDRDMA
+ip-26-0-167-9:1039538:1039704 [1] NCCL INFO Channel 02/0 : 57[1] -> 56[0] via P2P/IPC
+ip-26-0-166-244:897245:897401 [5] NCCL INFO Channel 13/0 : 38[6] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104435:104595 [6] NCCL INFO Channel 14/0 : 47[7] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897246:897405 [6] NCCL INFO Channel 05/0 : 46[6] -> 53[5] [send] via NET/Libfabric/5(45)/GDRDMA
+ip-26-0-166-214:624949:625110 [6] NCCL INFO Channel 06/0 : 31[7] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897246:897405 [6] NCCL INFO Channel 13/0 : 46[6] -> 53[5] [send] via NET/Libfabric/5(45)/GDRDMA
+ip-26-0-166-36:104435:104595 [6] NCCL INFO Channel 05/0 : 54[6] -> 61[5] [send] via NET/Libfabric/5(53)/GDRDMA
+ip-26-0-167-9:1039538:1039704 [1] NCCL INFO Channel 03/0 : 57[1] -> 56[0] via P2P/IPC
+ip-26-0-166-15:127292:127458 [2] NCCL INFO Channel 04/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-166-36:104435:104595 [6] NCCL INFO Channel 13/0 : 54[6] -> 61[5] [send] via NET/Libfabric/5(53)/GDRDMA
+ip-26-0-166-214:624950:625109 [7] NCCL INFO Channel 14/0 : 39[7] -> 46[6] [send] via NET/Libfabric/6(38)/GDRDMA
+ip-26-0-166-244:897245:897401 [5] NCCL INFO Channel 04/0 : 45[5] -> 52[4] [send] via NET/Libfabric/4(44)/GDRDMA
+ip-26-0-166-214:624949:625110 [6] NCCL INFO Channel 14/0 : 31[7] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897245:897401 [5] NCCL INFO Channel 12/0 : 45[5] -> 52[4] [send] via NET/Libfabric/4(44)/GDRDMA
+ip-26-0-167-9:1039538:1039704 [1] NCCL INFO Channel 04/0 : 57[1] -> 56[0] via P2P/IPC
+ip-26-0-166-214:624944:625106 [1] NCCL INFO Channel 01/0 : 33[1] -> 32[0] via P2P/IPC
+ip-26-0-166-244:897247:897404 [7] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-166-15:127292:127458 [2] NCCL INFO Channel 05/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-167-9:1039538:1039704 [1] NCCL INFO Channel 05/0 : 57[1] -> 56[0] via P2P/IPC
+ip-26-0-166-36:104434:104591 [5] NCCL INFO Channel 05/0 : 53[5] -> 51[3] via P2P/IPC
+ip-26-0-166-125:762816:762987 [1] NCCL INFO Channel 01/0 : 10[2] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127292:127458 [2] NCCL INFO Channel 06/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-167-9:1039538:1039704 [1] NCCL INFO Channel 06/0 : 57[1] -> 56[0] via P2P/IPC
+ip-26-0-166-125:762816:762987 [1] NCCL INFO Channel 09/0 : 10[2] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624944:625106 [1] NCCL INFO Channel 02/0 : 33[1] -> 32[0] via P2P/IPC
+ip-26-0-166-125:762817:762986 [2] NCCL INFO Channel 01/0 : 18[2] -> 25[1] [send] via NET/Libfabric/1(17)/GDRDMA
+ip-26-0-166-125:762817:762986 [2] NCCL INFO Channel 09/0 : 18[2] -> 25[1] [send] via NET/Libfabric/1(17)/GDRDMA
+ip-26-0-166-125:762821:762988 [6] NCCL INFO Channel 06/0 : 15[7] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-167-9:1039538:1039704 [1] NCCL INFO Channel 07/0 : 57[1] -> 56[0] via P2P/IPC
+ip-26-0-166-15:127292:127458 [2] NCCL INFO Channel 07/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-166-125:762822:762984 [7] NCCL INFO Channel 14/0 : 23[7] -> 30[6] [send] via NET/Libfabric/6(22)/GDRDMA
+ip-26-0-162-46:702882:703046 [2] NCCL INFO Channel 04/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-244:897241:897402 [1] NCCL INFO Channel 01/0 : 41[1] -> 40[0] via P2P/IPC
+ip-26-0-166-36:104431:104593 [2] NCCL INFO Channel 02/0 : 50[2] -> 49[1] via P2P/IPC
+ip-26-0-165-213:9629:9793 [4] NCCL INFO Channel 04/0 : 5[5] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624950:625109 [7] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-166-214:624944:625106 [1] NCCL INFO Channel 03/0 : 33[1] -> 32[0] via P2P/IPC
+ip-26-0-166-125:762821:762988 [6] NCCL INFO Channel 14/0 : 15[7] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9629:9793 [4] NCCL INFO Channel 12/0 : 5[5] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127292:127458 [2] NCCL INFO Channel 10/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-167-9:1039538:1039704 [1] NCCL INFO Channel 09/0 : 57[1] -> 56[0] via P2P/IPC
+ip-26-0-162-46:702882:703046 [2] NCCL INFO Channel 05/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-125:762822:762984 [7] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-166-214:624945:625111 [2] NCCL INFO Channel 02/0 : 34[2] -> 33[1] via P2P/IPC
+ip-26-0-166-244:897241:897402 [1] NCCL INFO Channel 02/0 : 41[1] -> 40[0] via P2P/IPC
+ip-26-0-167-9:1039538:1039704 [1] NCCL INFO Channel 10/0 : 57[1] -> 56[0] via P2P/IPC
+ip-26-0-166-15:127294:127457 [4] NCCL INFO Channel 07/0 : 28[4] -> 39[7] [send] via NET/Libfabric/7(31)/GDRDMA
+ip-26-0-166-125:762821:762988 [6] NCCL INFO Channel 05/0 : 22[6] -> 29[5] [send] via NET/Libfabric/5(21)/GDRDMA
+ip-26-0-166-214:624944:625106 [1] NCCL INFO Channel 04/0 : 33[1] -> 32[0] via P2P/IPC
+ip-26-0-166-15:127292:127458 [2] NCCL INFO Channel 11/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-166-125:762821:762988 [6] NCCL INFO Channel 13/0 : 22[6] -> 29[5] [send] via NET/Libfabric/5(21)/GDRDMA
+ip-26-0-162-46:702882:703046 [2] NCCL INFO Channel 06/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-36:104434:104591 [5] NCCL INFO Channel 13/0 : 53[5] -> 51[3] via P2P/IPC
+ip-26-0-166-15:127294:127457 [4] NCCL INFO Channel 15/0 : 28[4] -> 39[7] [send] via NET/Libfabric/7(31)/GDRDMA
+ip-26-0-167-9:1039538:1039704 [1] NCCL INFO Channel 11/0 : 57[1] -> 56[0] via P2P/IPC
+ip-26-0-162-46:702881:703043 [1] NCCL INFO Channel 01/0 : 58[2] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104431:104593 [2] NCCL INFO Channel 03/0 : 50[2] -> 49[1] via P2P/IPC
+ip-26-0-166-214:624945:625111 [2] NCCL INFO Channel 03/0 : 34[2] -> 33[1] via P2P/IPC
+ip-26-0-166-15:127292:127458 [2] NCCL INFO Channel 12/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-166-244:897241:897402 [1] NCCL INFO Channel 03/0 : 41[1] -> 40[0] via P2P/IPC
+ip-26-0-162-46:702881:703043 [1] NCCL INFO Channel 09/0 : 58[2] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-46:702882:703046 [2] NCCL INFO Channel 07/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-214:624944:625106 [1] NCCL INFO Channel 05/0 : 33[1] -> 32[0] via P2P/IPC
+ip-26-0-167-9:1039538:1039704 [1] NCCL INFO Channel 12/0 : 57[1] -> 56[0] via P2P/IPC
+ip-26-0-165-213:9630:9789 [5] NCCL INFO Channel 05/0 : 6[6] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9630:9789 [5] NCCL INFO Channel 13/0 : 6[6] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127292:127458 [2] NCCL INFO Channel 13/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-166-214:624945:625111 [2] NCCL INFO Channel 04/0 : 34[2] -> 33[1] via P2P/IPC
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO Channel 07/0 : 60[4] -> 7[7] [send] via NET/Libfabric/7(63)/GDRDMA
+ip-26-0-162-46:702886:703049 [6] NCCL INFO Channel 06/0 : 63[7] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9629:9793 [4] NCCL INFO Channel 03/0 : 12[4] -> 19[3] [send] via NET/Libfabric/3(11)/GDRDMA
+ip-26-0-166-244:897241:897402 [1] NCCL INFO Channel 04/0 : 41[1] -> 40[0] via P2P/IPC
+ip-26-0-167-9:1039538:1039704 [1] NCCL INFO Channel 13/0 : 57[1] -> 56[0] via P2P/IPC
+ip-26-0-166-214:624944:625106 [1] NCCL INFO Channel 06/0 : 33[1] -> 32[0] via P2P/IPC
+ip-26-0-166-244:897246:897405 [6] NCCL INFO Channel 06/0 : 46[6] -> 43[3] via P2P/IPC
+ip-26-0-165-213:9630:9789 [5] NCCL INFO Channel 04/0 : 13[5] -> 20[4] [send] via NET/Libfabric/4(12)/GDRDMA
+ip-26-0-162-46:702887:703045 [7] NCCL INFO Channel 14/0 : 7[7] -> 14[6] [send] via NET/Libfabric/6(6)/GDRDMA
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO Channel 15/0 : 60[4] -> 7[7] [send] via NET/Libfabric/7(63)/GDRDMA
+ip-26-0-166-36:104431:104593 [2] NCCL INFO Channel 04/0 : 50[2] -> 49[1] via P2P/IPC
+ip-26-0-165-213:9630:9789 [5] NCCL INFO Channel 12/0 : 13[5] -> 20[4] [send] via NET/Libfabric/4(12)/GDRDMA
+ip-26-0-166-15:127292:127458 [2] NCCL INFO Channel 14/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-167-9:1039538:1039704 [1] NCCL INFO Channel 14/0 : 57[1] -> 56[0] via P2P/IPC
+ip-26-0-162-46:702886:703049 [6] NCCL INFO Channel 14/0 : 63[7] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624947:625105 [4] NCCL INFO Channel 04/0 : 29[5] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624945:625111 [2] NCCL INFO Channel 05/0 : 34[2] -> 33[1] via P2P/IPC
+ip-26-0-165-213:9629:9793 [4] NCCL INFO Channel 11/0 : 12[4] -> 19[3] [send] via NET/Libfabric/3(11)/GDRDMA
+ip-26-0-167-9:1039538:1039704 [1] NCCL INFO Channel 15/0 : 57[1] -> 56[0] via P2P/IPC
+ip-26-0-166-214:624947:625105 [4] NCCL INFO Channel 12/0 : 29[5] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624944:625106 [1] NCCL INFO Channel 07/0 : 33[1] -> 32[0] via P2P/IPC
+ip-26-0-166-36:104431:104593 [2] NCCL INFO Channel 05/0 : 50[2] -> 49[1] via P2P/IPC
+ip-26-0-167-9:1039543:1039699 [6] NCCL INFO Channel 06/0 : 55[7] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104433:104590 [4] NCCL INFO Channel 07/0 : 52[4] -> 63[7] [send] via NET/Libfabric/7(55)/GDRDMA
+ip-26-0-166-244:897241:897402 [1] NCCL INFO Channel 05/0 : 41[1] -> 40[0] via P2P/IPC
+ip-26-0-166-15:127292:127458 [2] NCCL INFO Channel 15/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-167-9:1039544:1039702 [7] NCCL INFO Channel 06/0 : 63[7] -> 6[6] [send] via NET/Libfabric/6(62)/GDRDMA
+ip-26-0-166-36:104430:104589 [1] NCCL INFO Channel 01/0 : 49[1] -> 48[0] via P2P/IPC
+ip-26-0-166-36:104433:104590 [4] NCCL INFO Channel 15/0 : 52[4] -> 63[7] [send] via NET/Libfabric/7(55)/GDRDMA
+ip-26-0-165-213:9631:9788 [6] NCCL INFO Channel 06/0 : 7[7] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-167-9:1039543:1039699 [6] NCCL INFO Channel 14/0 : 55[7] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897246:897405 [6] NCCL INFO Channel 14/0 : 46[6] -> 43[3] via P2P/IPC
+ip-26-0-166-214:624947:625105 [4] NCCL INFO Channel 03/0 : 36[4] -> 43[3] [send] via NET/Libfabric/3(35)/GDRDMA
+ip-26-0-166-244:897242:897400 [2] NCCL INFO Channel 02/0 : 42[2] -> 41[1] via P2P/IPC
+ip-26-0-165-213:9632:9792 [7] NCCL INFO Channel 14/0 : 15[7] -> 22[6] [send] via NET/Libfabric/6(14)/GDRDMA
+ip-26-0-167-9:1039542:1039701 [5] NCCL INFO Channel 05/0 : 54[6] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-167-9:1039544:1039702 [7] NCCL INFO Channel 14/0 : 63[7] -> 6[6] [send] via NET/Libfabric/6(62)/GDRDMA
+ip-26-0-166-214:624945:625111 [2] NCCL INFO Channel 06/0 : 34[2] -> 33[1] via P2P/IPC
+ip-26-0-165-213:9631:9788 [6] NCCL INFO Channel 14/0 : 7[7] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-46:702882:703046 [2] NCCL INFO Channel 10/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-214:624944:625106 [1] NCCL INFO Channel 09/0 : 33[1] -> 32[0] via P2P/IPC
+ip-26-0-166-214:624947:625105 [4] NCCL INFO Channel 11/0 : 36[4] -> 43[3] [send] via NET/Libfabric/3(35)/GDRDMA
+ip-26-0-166-36:104431:104593 [2] NCCL INFO Channel 06/0 : 50[2] -> 49[1] via P2P/IPC
+ip-26-0-167-9:1039542:1039701 [5] NCCL INFO Channel 13/0 : 54[6] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104430:104589 [1] NCCL INFO Channel 02/0 : 49[1] -> 48[0] via P2P/IPC
+ip-26-0-162-46:702885:703048 [5] NCCL INFO Channel 05/0 : 62[6] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-167-9:1039543:1039699 [6] NCCL INFO Channel 05/0 : 62[6] -> 5[5] [send] via NET/Libfabric/5(61)/GDRDMA
+ip-26-0-165-213:9631:9788 [6] NCCL INFO Channel 05/0 : 14[6] -> 21[5] [send] via NET/Libfabric/5(13)/GDRDMA
+ip-26-0-166-244:897241:897402 [1] NCCL INFO Channel 06/0 : 41[1] -> 40[0] via P2P/IPC
+ip-26-0-165-213:9632:9792 [7] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-167-9:1039543:1039699 [6] NCCL INFO Channel 13/0 : 62[6] -> 5[5] [send] via NET/Libfabric/5(61)/GDRDMA
+ip-26-0-165-213:9631:9788 [6] NCCL INFO Channel 13/0 : 14[6] -> 21[5] [send] via NET/Libfabric/5(13)/GDRDMA
+ip-26-0-162-46:702886:703049 [6] NCCL INFO Channel 05/0 : 6[6] -> 13[5] [send] via NET/Libfabric/5(5)/GDRDMA
+ip-26-0-162-46:702886:703049 [6] NCCL INFO Channel 13/0 : 6[6] -> 13[5] [send] via NET/Libfabric/5(5)/GDRDMA
+ip-26-0-166-214:624948:625104 [5] NCCL INFO Channel 05/0 : 30[6] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762821:762988 [6] NCCL INFO Channel 06/0 : 22[6] -> 19[3] via P2P/IPC
+ip-26-0-166-15:127295:127452 [5] NCCL INFO Channel 05/0 : 29[5] -> 27[3] via P2P/IPC
+ip-26-0-166-244:897242:897400 [2] NCCL INFO Channel 03/0 : 42[2] -> 41[1] via P2P/IPC
+ip-26-0-167-9:1039542:1039701 [5] NCCL INFO Channel 04/0 : 61[5] -> 4[4] [send] via NET/Libfabric/4(60)/GDRDMA
+ip-26-0-166-36:104431:104593 [2] NCCL INFO Channel 07/0 : 50[2] -> 49[1] via P2P/IPC
+ip-26-0-166-214:624945:625111 [2] NCCL INFO Channel 07/0 : 34[2] -> 33[1] via P2P/IPC
+ip-26-0-166-214:624944:625106 [1] NCCL INFO Channel 10/0 : 33[1] -> 32[0] via P2P/IPC
+ip-26-0-162-46:702885:703048 [5] NCCL INFO Channel 13/0 : 62[6] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-46:702887:703045 [7] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-166-36:104430:104589 [1] NCCL INFO Channel 03/0 : 49[1] -> 48[0] via P2P/IPC
+ip-26-0-167-9:1039542:1039701 [5] NCCL INFO Channel 12/0 : 61[5] -> 4[4] [send] via NET/Libfabric/4(60)/GDRDMA
+ip-26-0-166-214:624949:625110 [6] NCCL INFO Channel 05/0 : 38[6] -> 45[5] [send] via NET/Libfabric/5(37)/GDRDMA
+ip-26-0-166-36:104436:104596 [7] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-166-214:624948:625104 [5] NCCL INFO Channel 13/0 : 30[6] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624949:625110 [6] NCCL INFO Channel 13/0 : 38[6] -> 45[5] [send] via NET/Libfabric/5(37)/GDRDMA
+ip-26-0-167-9:1039544:1039702 [7] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-165-213:9631:9788 [6] NCCL INFO Channel 06/0 : 14[6] -> 11[3] via P2P/IPC
+ip-26-0-166-244:897241:897402 [1] NCCL INFO Channel 07/0 : 41[1] -> 40[0] via P2P/IPC
+ip-26-0-166-125:762820:762981 [5] NCCL INFO Channel 05/0 : 21[5] -> 19[3] via P2P/IPC
+ip-26-0-166-36:104431:104593 [2] NCCL INFO Channel 10/0 : 50[2] -> 49[1] via P2P/IPC
+ip-26-0-166-214:624945:625111 [2] NCCL INFO Channel 10/0 : 34[2] -> 33[1] via P2P/IPC
+ip-26-0-162-46:702885:703048 [5] NCCL INFO Channel 04/0 : 5[5] -> 12[4] [send] via NET/Libfabric/4(4)/GDRDMA
+ip-26-0-162-46:702886:703049 [6] NCCL INFO Channel 06/0 : 6[6] -> 3[3] via P2P/IPC
+ip-26-0-166-36:104435:104595 [6] NCCL INFO Channel 06/0 : 54[6] -> 51[3] via P2P/IPC
+ip-26-0-166-214:624944:625106 [1] NCCL INFO Channel 11/0 : 33[1] -> 32[0] via P2P/IPC
+ip-26-0-165-213:9630:9789 [5] NCCL INFO Channel 05/0 : 13[5] -> 11[3] via P2P/IPC
+ip-26-0-166-244:897242:897400 [2] NCCL INFO Channel 04/0 : 42[2] -> 41[1] via P2P/IPC
+ip-26-0-162-46:702885:703048 [5] NCCL INFO Channel 12/0 : 5[5] -> 12[4] [send] via NET/Libfabric/4(4)/GDRDMA
+ip-26-0-166-36:104430:104589 [1] NCCL INFO Channel 04/0 : 49[1] -> 48[0] via P2P/IPC
+ip-26-0-166-214:624948:625104 [5] NCCL INFO Channel 04/0 : 37[5] -> 44[4] [send] via NET/Libfabric/4(36)/GDRDMA
+ip-26-0-166-36:104431:104593 [2] NCCL INFO Channel 11/0 : 50[2] -> 49[1] via P2P/IPC
+ip-26-0-166-214:624948:625104 [5] NCCL INFO Channel 12/0 : 37[5] -> 44[4] [send] via NET/Libfabric/4(36)/GDRDMA
+ip-26-0-167-9:1039542:1039701 [5] NCCL INFO Channel 05/0 : 61[5] -> 59[3] via P2P/IPC
+ip-26-0-166-244:897241:897402 [1] NCCL INFO Channel 09/0 : 41[1] -> 40[0] via P2P/IPC
+ip-26-0-166-214:624949:625110 [6] NCCL INFO Channel 06/0 : 38[6] -> 35[3] via P2P/IPC
+ip-26-0-166-244:897245:897401 [5] NCCL INFO Channel 05/0 : 45[5] -> 43[3] via P2P/IPC
+ip-26-0-165-213:9631:9788 [6] NCCL INFO Channel 14/0 : 14[6] -> 11[3] via P2P/IPC
+ip-26-0-166-214:624945:625111 [2] NCCL INFO Channel 11/0 : 34[2] -> 33[1] via P2P/IPC
+ip-26-0-166-125:762821:762988 [6] NCCL INFO Channel 14/0 : 22[6] -> 19[3] via P2P/IPC
+ip-26-0-162-46:702886:703049 [6] NCCL INFO Channel 14/0 : 6[6] -> 3[3] via P2P/IPC
+ip-26-0-167-9:1039543:1039699 [6] NCCL INFO Channel 06/0 : 62[6] -> 59[3] via P2P/IPC
+ip-26-0-166-214:624944:625106 [1] NCCL INFO Channel 12/0 : 33[1] -> 32[0] via P2P/IPC
+ip-26-0-166-36:104430:104589 [1] NCCL INFO Channel 05/0 : 49[1] -> 48[0] via P2P/IPC
+ip-26-0-166-15:127296:127451 [6] NCCL INFO Channel 06/0 : 30[6] -> 27[3] via P2P/IPC
+ip-26-0-162-46:702885:703048 [5] NCCL INFO Channel 05/0 : 5[5] -> 3[3] via P2P/IPC
+ip-26-0-166-36:104435:104595 [6] NCCL INFO Channel 14/0 : 54[6] -> 51[3] via P2P/IPC
+ip-26-0-166-36:104431:104593 [2] NCCL INFO Channel 12/0 : 50[2] -> 49[1] via P2P/IPC
+ip-26-0-166-214:624948:625104 [5] NCCL INFO Channel 05/0 : 37[5] -> 35[3] via P2P/IPC
+ip-26-0-165-213:9630:9789 [5] NCCL INFO Channel 13/0 : 13[5] -> 11[3] via P2P/IPC
+ip-26-0-166-214:624949:625110 [6] NCCL INFO Channel 14/0 : 38[6] -> 35[3] via P2P/IPC
+ip-26-0-166-244:897242:897400 [2] NCCL INFO Channel 05/0 : 42[2] -> 41[1] via P2P/IPC
+ip-26-0-166-214:624945:625111 [2] NCCL INFO Channel 12/0 : 34[2] -> 33[1] via P2P/IPC
+ip-26-0-166-244:897241:897402 [1] NCCL INFO Channel 10/0 : 41[1] -> 40[0] via P2P/IPC
+ip-26-0-166-15:127297:127453 [7] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-166-36:104430:104589 [1] NCCL INFO Channel 06/0 : 49[1] -> 48[0] via P2P/IPC
+ip-26-0-162-46:702882:703046 [2] NCCL INFO Channel 11/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-214:624944:625106 [1] NCCL INFO Channel 13/0 : 33[1] -> 32[0] via P2P/IPC
+ip-26-0-166-15:127295:127452 [5] NCCL INFO Channel 13/0 : 29[5] -> 27[3] via P2P/IPC
+ip-26-0-166-36:104431:104593 [2] NCCL INFO Channel 13/0 : 50[2] -> 49[1] via P2P/IPC
+ip-26-0-166-36:104430:104589 [1] NCCL INFO Channel 07/0 : 49[1] -> 48[0] via P2P/IPC
+ip-26-0-166-244:897242:897400 [2] NCCL INFO Channel 06/0 : 42[2] -> 41[1] via P2P/IPC
+ip-26-0-165-213:9627:9791 [2] NCCL INFO Channel 02/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-166-244:897245:897401 [5] NCCL INFO Channel 13/0 : 45[5] -> 43[3] via P2P/IPC
+ip-26-0-166-244:897241:897402 [1] NCCL INFO Channel 11/0 : 41[1] -> 40[0] via P2P/IPC
+ip-26-0-162-46:702885:703048 [5] NCCL INFO Channel 13/0 : 5[5] -> 3[3] via P2P/IPC
+ip-26-0-166-36:104431:104593 [2] NCCL INFO Channel 14/0 : 50[2] -> 49[1] via P2P/IPC
+ip-26-0-166-214:624945:625111 [2] NCCL INFO Channel 13/0 : 34[2] -> 33[1] via P2P/IPC
+ip-26-0-162-46:702882:703046 [2] NCCL INFO Channel 12/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-36:104430:104589 [1] NCCL INFO Channel 09/0 : 49[1] -> 48[0] via P2P/IPC
+ip-26-0-165-213:9627:9791 [2] NCCL INFO Channel 03/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-166-214:624948:625104 [5] NCCL INFO Channel 13/0 : 37[5] -> 35[3] via P2P/IPC
+ip-26-0-166-15:127296:127451 [6] NCCL INFO Channel 14/0 : 30[6] -> 27[3] via P2P/IPC
+ip-26-0-166-214:624944:625106 [1] NCCL INFO Channel 14/0 : 33[1] -> 32[0] via P2P/IPC
+ip-26-0-166-244:897242:897400 [2] NCCL INFO Channel 07/0 : 42[2] -> 41[1] via P2P/IPC
+ip-26-0-166-125:762820:762981 [5] NCCL INFO Channel 13/0 : 21[5] -> 19[3] via P2P/IPC
+ip-26-0-167-9:1039542:1039701 [5] NCCL INFO Channel 13/0 : 61[5] -> 59[3] via P2P/IPC
+ip-26-0-165-213:9627:9791 [2] NCCL INFO Channel 04/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-166-36:104431:104593 [2] NCCL INFO Channel 15/0 : 50[2] -> 49[1] via P2P/IPC
+ip-26-0-166-214:624945:625111 [2] NCCL INFO Channel 14/0 : 34[2] -> 33[1] via P2P/IPC
+ip-26-0-166-36:104430:104589 [1] NCCL INFO Channel 10/0 : 49[1] -> 48[0] via P2P/IPC
+ip-26-0-166-244:897241:897402 [1] NCCL INFO Channel 12/0 : 41[1] -> 40[0] via P2P/IPC
+ip-26-0-167-9:1039543:1039699 [6] NCCL INFO Channel 14/0 : 62[6] -> 59[3] via P2P/IPC
+ip-26-0-162-46:702882:703046 [2] NCCL INFO Channel 13/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-244:897242:897400 [2] NCCL INFO Channel 10/0 : 42[2] -> 41[1] via P2P/IPC
+ip-26-0-165-213:9627:9791 [2] NCCL INFO Channel 05/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-166-214:624944:625106 [1] NCCL INFO Channel 15/0 : 33[1] -> 32[0] via P2P/IPC
+ip-26-0-166-36:104430:104589 [1] NCCL INFO Channel 11/0 : 49[1] -> 48[0] via P2P/IPC
+ip-26-0-166-214:624945:625111 [2] NCCL INFO Channel 15/0 : 34[2] -> 33[1] via P2P/IPC
+ip-26-0-166-244:897241:897402 [1] NCCL INFO Channel 13/0 : 41[1] -> 40[0] via P2P/IPC
+ip-26-0-166-36:104430:104589 [1] NCCL INFO Channel 12/0 : 49[1] -> 48[0] via P2P/IPC
+ip-26-0-165-213:9627:9791 [2] NCCL INFO Channel 06/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-166-244:897242:897400 [2] NCCL INFO Channel 11/0 : 42[2] -> 41[1] via P2P/IPC
+ip-26-0-162-46:702882:703046 [2] NCCL INFO Channel 14/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-36:104430:104589 [1] NCCL INFO Channel 13/0 : 49[1] -> 48[0] via P2P/IPC
+ip-26-0-165-213:9627:9791 [2] NCCL INFO Channel 07/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-166-244:897241:897402 [1] NCCL INFO Channel 14/0 : 41[1] -> 40[0] via P2P/IPC
+ip-26-0-166-36:104430:104589 [1] NCCL INFO Channel 14/0 : 49[1] -> 48[0] via P2P/IPC
+ip-26-0-162-46:702882:703046 [2] NCCL INFO Channel 15/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-165-213:9627:9791 [2] NCCL INFO Channel 10/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-166-36:104430:104589 [1] NCCL INFO Channel 15/0 : 49[1] -> 48[0] via P2P/IPC
+ip-26-0-166-244:897242:897400 [2] NCCL INFO Channel 12/0 : 42[2] -> 41[1] via P2P/IPC
+ip-26-0-165-213:9627:9791 [2] NCCL INFO Channel 11/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-166-244:897241:897402 [1] NCCL INFO Channel 15/0 : 41[1] -> 40[0] via P2P/IPC
+ip-26-0-165-213:9627:9791 [2] NCCL INFO Channel 12/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-166-244:897242:897400 [2] NCCL INFO Channel 13/0 : 42[2] -> 41[1] via P2P/IPC
+ip-26-0-165-213:9627:9791 [2] NCCL INFO Channel 13/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-166-15:127291:127455 [1] NCCL INFO Channel 01/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-166-244:897242:897400 [2] NCCL INFO Channel 14/0 : 42[2] -> 41[1] via P2P/IPC
+ip-26-0-166-15:127291:127455 [1] NCCL INFO Channel 02/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-165-213:9627:9791 [2] NCCL INFO Channel 14/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-167-9:1039539:1039698 [2] NCCL INFO Channel 02/0 : 58[2] -> 57[1] via P2P/IPC
+ip-26-0-166-15:127291:127455 [1] NCCL INFO Channel 03/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-165-213:9627:9791 [2] NCCL INFO Channel 15/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-162-46:702881:703043 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-167-9:1039539:1039698 [2] NCCL INFO Channel 03/0 : 58[2] -> 57[1] via P2P/IPC
+ip-26-0-166-15:127291:127455 [1] NCCL INFO Channel 04/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-162-46:702881:703043 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762816:762987 [1] NCCL INFO Channel 01/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-167-9:1039539:1039698 [2] NCCL INFO Channel 04/0 : 58[2] -> 57[1] via P2P/IPC
+ip-26-0-162-46:702881:703043 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-167-9:1039539:1039698 [2] NCCL INFO Channel 05/0 : 58[2] -> 57[1] via P2P/IPC
+ip-26-0-162-46:702881:703043 [1] NCCL INFO Channel 04/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-167-9:1039539:1039698 [2] NCCL INFO Channel 06/0 : 58[2] -> 57[1] via P2P/IPC
+ip-26-0-166-15:127291:127455 [1] NCCL INFO Channel 05/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-167-9:1039539:1039698 [2] NCCL INFO Channel 07/0 : 58[2] -> 57[1] via P2P/IPC
+ip-26-0-162-46:702881:703043 [1] NCCL INFO Channel 05/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127291:127455 [1] NCCL INFO Channel 06/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-167-9:1039539:1039698 [2] NCCL INFO Channel 10/0 : 58[2] -> 57[1] via P2P/IPC
+ip-26-0-166-125:762816:762987 [1] NCCL INFO Channel 02/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-162-46:702881:703043 [1] NCCL INFO Channel 06/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127291:127455 [1] NCCL INFO Channel 07/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-167-9:1039539:1039698 [2] NCCL INFO Channel 11/0 : 58[2] -> 57[1] via P2P/IPC
+ip-26-0-166-125:762817:762986 [2] NCCL INFO Channel 02/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-166-125:762816:762987 [1] NCCL INFO Channel 03/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-166-15:127291:127455 [1] NCCL INFO Channel 09/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-162-46:702881:703043 [1] NCCL INFO Channel 07/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-167-9:1039539:1039698 [2] NCCL INFO Channel 12/0 : 58[2] -> 57[1] via P2P/IPC
+ip-26-0-166-15:127291:127455 [1] NCCL INFO Channel 10/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-166-125:762816:762987 [1] NCCL INFO Channel 04/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-167-9:1039539:1039698 [2] NCCL INFO Channel 13/0 : 58[2] -> 57[1] via P2P/IPC
+ip-26-0-166-125:762817:762986 [2] NCCL INFO Channel 03/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-162-46:702881:703043 [1] NCCL INFO Channel 09/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127291:127455 [1] NCCL INFO Channel 11/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-166-125:762816:762987 [1] NCCL INFO Channel 05/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-166-125:762817:762986 [2] NCCL INFO Channel 04/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-162-46:702881:703043 [1] NCCL INFO Channel 10/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762816:762987 [1] NCCL INFO Channel 06/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-162-46:702881:703043 [1] NCCL INFO Channel 11/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762817:762986 [2] NCCL INFO Channel 05/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-166-244:897242:897400 [2] NCCL INFO Channel 15/0 : 42[2] -> 41[1] via P2P/IPC
+ip-26-0-166-125:762816:762987 [1] NCCL INFO Channel 07/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-162-46:702881:703043 [1] NCCL INFO Channel 12/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762817:762986 [2] NCCL INFO Channel 06/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-166-125:762816:762987 [1] NCCL INFO Channel 09/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-162-46:702881:703043 [1] NCCL INFO Channel 13/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762817:762986 [2] NCCL INFO Channel 07/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-166-125:762816:762987 [1] NCCL INFO Channel 10/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-162-46:702881:703043 [1] NCCL INFO Channel 14/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127291:127455 [1] NCCL INFO Channel 12/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-166-125:762817:762986 [2] NCCL INFO Channel 10/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-162-46:702881:703043 [1] NCCL INFO Channel 15/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762816:762987 [1] NCCL INFO Channel 11/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-166-15:127291:127455 [1] NCCL INFO Channel 13/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-166-125:762817:762986 [2] NCCL INFO Channel 11/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-166-125:762816:762987 [1] NCCL INFO Channel 12/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-166-15:127291:127455 [1] NCCL INFO Channel 14/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-166-125:762817:762986 [2] NCCL INFO Channel 12/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-166-125:762816:762987 [1] NCCL INFO Channel 13/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-167-9:1039539:1039698 [2] NCCL INFO Channel 14/0 : 58[2] -> 57[1] via P2P/IPC
+ip-26-0-166-125:762817:762986 [2] NCCL INFO Channel 13/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-166-125:762816:762987 [1] NCCL INFO Channel 14/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-166-125:762817:762986 [2] NCCL INFO Channel 14/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-166-125:762816:762987 [1] NCCL INFO Channel 15/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-166-125:762817:762986 [2] NCCL INFO Channel 15/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-166-15:127291:127455 [1] NCCL INFO Channel 15/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-167-9:1039539:1039698 [2] NCCL INFO Channel 15/0 : 58[2] -> 57[1] via P2P/IPC
+ip-26-0-166-244:897247:897404 [7] NCCL INFO Channel 07/0 : 36[4] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897247:897404 [7] NCCL INFO Channel 15/0 : 36[4] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897244:897407 [4] NCCL INFO Channel 07/0 : 44[4] -> 55[7] [send] via NET/Libfabric/7(47)/GDRDMA
+ip-26-0-166-244:897244:897407 [4] NCCL INFO Channel 15/0 : 44[4] -> 55[7] [send] via NET/Libfabric/7(47)/GDRDMA
+ip-26-0-166-125:762822:762984 [7] NCCL INFO Channel 07/0 : 12[4] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762819:762982 [4] NCCL INFO Channel 07/0 : 20[4] -> 31[7] [send] via NET/Libfabric/7(23)/GDRDMA
+ip-26-0-166-125:762822:762984 [7] NCCL INFO Channel 15/0 : 12[4] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762819:762982 [4] NCCL INFO Channel 15/0 : 20[4] -> 31[7] [send] via NET/Libfabric/7(23)/GDRDMA
+ip-26-0-166-36:104436:104596 [7] NCCL INFO Channel 07/0 : 44[4] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104436:104596 [7] NCCL INFO Channel 15/0 : 44[4] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104436:104596 [7] NCCL INFO Channel 07/0 : 55[7] -> 51[3] via P2P/IPC
+ip-26-0-167-9:1039544:1039702 [7] NCCL INFO Channel 07/0 : 52[4] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-167-9:1039544:1039702 [7] NCCL INFO Channel 15/0 : 52[4] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624950:625109 [7] NCCL INFO Channel 07/0 : 28[4] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624950:625109 [7] NCCL INFO Channel 15/0 : 28[4] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624947:625105 [4] NCCL INFO Channel 07/0 : 36[4] -> 47[7] [send] via NET/Libfabric/7(39)/GDRDMA
+ip-26-0-167-9:1039544:1039702 [7] NCCL INFO Channel 07/0 : 63[7] -> 59[3] via P2P/IPC
+ip-26-0-166-214:624947:625105 [4] NCCL INFO Channel 15/0 : 36[4] -> 47[7] [send] via NET/Libfabric/7(39)/GDRDMA
+ip-26-0-166-214:624950:625109 [7] NCCL INFO Channel 07/0 : 39[7] -> 35[3] via P2P/IPC
+ip-26-0-162-46:702887:703045 [7] NCCL INFO Channel 07/0 : 60[4] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-46:702884:703047 [4] NCCL INFO Channel 07/0 : 4[4] -> 15[7] [send] via NET/Libfabric/7(7)/GDRDMA
+ip-26-0-162-46:702887:703045 [7] NCCL INFO Channel 15/0 : 60[4] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897247:897404 [7] NCCL INFO Channel 07/0 : 47[7] -> 43[3] via P2P/IPC
+ip-26-0-162-46:702884:703047 [4] NCCL INFO Channel 15/0 : 4[4] -> 15[7] [send] via NET/Libfabric/7(7)/GDRDMA
+ip-26-0-162-46:702887:703045 [7] NCCL INFO Channel 07/0 : 7[7] -> 3[3] via P2P/IPC
+ip-26-0-166-15:127297:127453 [7] NCCL INFO Channel 07/0 : 20[4] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127297:127453 [7] NCCL INFO Channel 15/0 : 20[4] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127297:127453 [7] NCCL INFO Channel 07/0 : 31[7] -> 27[3] via P2P/IPC
+ip-26-0-165-213:9632:9792 [7] NCCL INFO Channel 07/0 : 4[4] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9629:9793 [4] NCCL INFO Channel 07/0 : 12[4] -> 23[7] [send] via NET/Libfabric/7(15)/GDRDMA
+ip-26-0-165-213:9632:9792 [7] NCCL INFO Channel 15/0 : 4[4] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9629:9793 [4] NCCL INFO Channel 15/0 : 12[4] -> 23[7] [send] via NET/Libfabric/7(15)/GDRDMA
+ip-26-0-165-213:9632:9792 [7] NCCL INFO Channel 07/0 : 15[7] -> 11[3] via P2P/IPC
+ip-26-0-166-125:762822:762984 [7] NCCL INFO Channel 07/0 : 23[7] -> 19[3] via P2P/IPC
+ip-26-0-167-9:1039544:1039702 [7] NCCL INFO Channel 15/0 : 63[7] -> 59[3] via P2P/IPC
+ip-26-0-166-36:104436:104596 [7] NCCL INFO Channel 15/0 : 55[7] -> 51[3] via P2P/IPC
+ip-26-0-166-125:762822:762984 [7] NCCL INFO Channel 15/0 : 23[7] -> 19[3] via P2P/IPC
+ip-26-0-166-214:624950:625109 [7] NCCL INFO Channel 15/0 : 39[7] -> 35[3] via P2P/IPC
+ip-26-0-162-46:702887:703045 [7] NCCL INFO Channel 15/0 : 7[7] -> 3[3] via P2P/IPC
+ip-26-0-166-15:127297:127453 [7] NCCL INFO Channel 15/0 : 31[7] -> 27[3] via P2P/IPC
+ip-26-0-166-244:897247:897404 [7] NCCL INFO Channel 15/0 : 47[7] -> 43[3] via P2P/IPC
+ip-26-0-166-125:762822:762984 [7] NCCL INFO Channel 01/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-167-9:1039544:1039702 [7] NCCL INFO Channel 01/0 : 63[7] -> 62[6] via P2P/IPC
+ip-26-0-162-46:702887:703045 [7] NCCL INFO Channel 01/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-166-36:104436:104596 [7] NCCL INFO Channel 01/0 : 55[7] -> 54[6] via P2P/IPC
+ip-26-0-166-15:127297:127453 [7] NCCL INFO Channel 01/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-166-214:624950:625109 [7] NCCL INFO Channel 01/0 : 39[7] -> 38[6] via P2P/IPC
+ip-26-0-165-213:9632:9792 [7] NCCL INFO Channel 15/0 : 15[7] -> 11[3] via P2P/IPC
+ip-26-0-162-46:702887:703045 [7] NCCL INFO Channel 02/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-167-9:1039544:1039702 [7] NCCL INFO Channel 02/0 : 63[7] -> 62[6] via P2P/IPC
+ip-26-0-166-125:762822:762984 [7] NCCL INFO Channel 02/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-166-15:127297:127453 [7] NCCL INFO Channel 02/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-166-36:104436:104596 [7] NCCL INFO Channel 02/0 : 55[7] -> 54[6] via P2P/IPC
+ip-26-0-165-213:9632:9792 [7] NCCL INFO Channel 01/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-166-15:127297:127453 [7] NCCL INFO Channel 03/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-166-214:624950:625109 [7] NCCL INFO Channel 02/0 : 39[7] -> 38[6] via P2P/IPC
+ip-26-0-162-46:702887:703045 [7] NCCL INFO Channel 03/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-166-125:762822:762984 [7] NCCL INFO Channel 03/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-166-244:897247:897404 [7] NCCL INFO Channel 01/0 : 47[7] -> 46[6] via P2P/IPC
+ip-26-0-165-213:9632:9792 [7] NCCL INFO Channel 02/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-167-9:1039544:1039702 [7] NCCL INFO Channel 03/0 : 63[7] -> 62[6] via P2P/IPC
+ip-26-0-165-213:9632:9792 [7] NCCL INFO Channel 03/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-165-213:9632:9792 [7] NCCL INFO Channel 04/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-166-214:624950:625109 [7] NCCL INFO Channel 03/0 : 39[7] -> 38[6] via P2P/IPC
+ip-26-0-166-15:127297:127453 [7] NCCL INFO Channel 04/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-165-213:9632:9792 [7] NCCL INFO Channel 05/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-166-36:104436:104596 [7] NCCL INFO Channel 03/0 : 55[7] -> 54[6] via P2P/IPC
+ip-26-0-165-213:9632:9792 [7] NCCL INFO Channel 09/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-166-125:762822:762984 [7] NCCL INFO Channel 04/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-166-214:624950:625109 [7] NCCL INFO Channel 04/0 : 39[7] -> 38[6] via P2P/IPC
+ip-26-0-165-213:9632:9792 [7] NCCL INFO Channel 10/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-166-36:104436:104596 [7] NCCL INFO Channel 04/0 : 55[7] -> 54[6] via P2P/IPC
+ip-26-0-166-15:127297:127453 [7] NCCL INFO Channel 05/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-166-36:104436:104596 [7] NCCL INFO Channel 05/0 : 55[7] -> 54[6] via P2P/IPC
+ip-26-0-166-214:624950:625109 [7] NCCL INFO Channel 05/0 : 39[7] -> 38[6] via P2P/IPC
+ip-26-0-165-213:9629:9793 [4] NCCL INFO Channel 01/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-165-213:9632:9792 [7] NCCL INFO Channel 11/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-166-36:104436:104596 [7] NCCL INFO Channel 09/0 : 55[7] -> 54[6] via P2P/IPC
+ip-26-0-166-214:624950:625109 [7] NCCL INFO Channel 09/0 : 39[7] -> 38[6] via P2P/IPC
+ip-26-0-166-15:127297:127453 [7] NCCL INFO Channel 09/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-166-36:104436:104596 [7] NCCL INFO Channel 10/0 : 55[7] -> 54[6] via P2P/IPC
+ip-26-0-165-213:9629:9793 [4] NCCL INFO Channel 02/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-165-213:9632:9792 [7] NCCL INFO Channel 12/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-166-244:897247:897404 [7] NCCL INFO Channel 02/0 : 47[7] -> 46[6] via P2P/IPC
+ip-26-0-166-214:624950:625109 [7] NCCL INFO Channel 10/0 : 39[7] -> 38[6] via P2P/IPC
+ip-26-0-166-214:624950:625109 [7] NCCL INFO Channel 11/0 : 39[7] -> 38[6] via P2P/IPC
+ip-26-0-166-36:104436:104596 [7] NCCL INFO Channel 11/0 : 55[7] -> 54[6] via P2P/IPC
+ip-26-0-166-15:127297:127453 [7] NCCL INFO Channel 10/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-166-244:897247:897404 [7] NCCL INFO Channel 03/0 : 47[7] -> 46[6] via P2P/IPC
+ip-26-0-166-214:624950:625109 [7] NCCL INFO Channel 12/0 : 39[7] -> 38[6] via P2P/IPC
+ip-26-0-166-214:624947:625105 [4] NCCL INFO Channel 01/0 : 36[4] -> 35[3] via P2P/IPC
+ip-26-0-165-213:9629:9793 [4] NCCL INFO Channel 04/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-166-244:897247:897404 [7] NCCL INFO Channel 04/0 : 47[7] -> 46[6] via P2P/IPC
+ip-26-0-165-213:9632:9792 [7] NCCL INFO Channel 13/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-166-125:762822:762984 [7] NCCL INFO Channel 05/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-166-214:624950:625109 [7] NCCL INFO Channel 13/0 : 39[7] -> 38[6] via P2P/IPC
+ip-26-0-166-36:104436:104596 [7] NCCL INFO Channel 12/0 : 55[7] -> 54[6] via P2P/IPC
+ip-26-0-166-15:127297:127453 [7] NCCL INFO Channel 11/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-166-214:624947:625105 [4] NCCL INFO Channel 02/0 : 36[4] -> 35[3] via P2P/IPC
+ip-26-0-166-244:897247:897404 [7] NCCL INFO Channel 05/0 : 47[7] -> 46[6] via P2P/IPC
+ip-26-0-166-36:104433:104590 [4] NCCL INFO Channel 01/0 : 52[4] -> 51[3] via P2P/IPC
+ip-26-0-166-125:762822:762984 [7] NCCL INFO Channel 09/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-166-214:624947:625105 [4] NCCL INFO Channel 04/0 : 36[4] -> 35[3] via P2P/IPC
+ip-26-0-165-213:9629:9793 [4] NCCL INFO Channel 09/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-166-36:104436:104596 [7] NCCL INFO Channel 13/0 : 55[7] -> 54[6] via P2P/IPC
+ip-26-0-166-244:897247:897404 [7] NCCL INFO Channel 09/0 : 47[7] -> 46[6] via P2P/IPC
+ip-26-0-165-213:9628:9786 [3] NCCL INFO Channel 01/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-166-214:624947:625105 [4] NCCL INFO Channel 09/0 : 36[4] -> 35[3] via P2P/IPC
+ip-26-0-162-46:702887:703045 [7] NCCL INFO Channel 04/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-166-36:104433:104590 [4] NCCL INFO Channel 02/0 : 52[4] -> 51[3] via P2P/IPC
+ip-26-0-165-213:9629:9793 [4] NCCL INFO Channel 10/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-165-213:9628:9786 [3] NCCL INFO Channel 03/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-166-15:127297:127453 [7] NCCL INFO Channel 12/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-166-214:624947:625105 [4] NCCL INFO Channel 10/0 : 36[4] -> 35[3] via P2P/IPC
+ip-26-0-166-244:897247:897404 [7] NCCL INFO Channel 10/0 : 47[7] -> 46[6] via P2P/IPC
+ip-26-0-166-15:127294:127457 [4] NCCL INFO Channel 01/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-165-213:9629:9793 [4] NCCL INFO Channel 12/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-166-36:104433:104590 [4] NCCL INFO Channel 04/0 : 52[4] -> 51[3] via P2P/IPC
+ip-26-0-165-213:9628:9786 [3] NCCL INFO Channel 04/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-166-36:104432:104592 [3] NCCL INFO Channel 01/0 : 51[3] -> 50[2] via P2P/IPC
+ip-26-0-166-15:127297:127453 [7] NCCL INFO Channel 13/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-166-125:762822:762984 [7] NCCL INFO Channel 10/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-166-214:624947:625105 [4] NCCL INFO Channel 12/0 : 36[4] -> 35[3] via P2P/IPC
+ip-26-0-166-15:127294:127457 [4] NCCL INFO Channel 02/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-166-36:104433:104590 [4] NCCL INFO Channel 09/0 : 52[4] -> 51[3] via P2P/IPC
+ip-26-0-165-213:9631:9788 [6] NCCL INFO Channel 01/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-166-244:897247:897404 [7] NCCL INFO Channel 11/0 : 47[7] -> 46[6] via P2P/IPC
+ip-26-0-166-36:104432:104592 [3] NCCL INFO Channel 03/0 : 51[3] -> 50[2] via P2P/IPC
+ip-26-0-165-213:9628:9786 [3] NCCL INFO Channel 05/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-166-125:762819:762982 [4] NCCL INFO Channel 01/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-166-125:762822:762984 [7] NCCL INFO Channel 11/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-162-46:702887:703045 [7] NCCL INFO Channel 05/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-165-213:9631:9788 [6] NCCL INFO Channel 02/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-166-36:104433:104590 [4] NCCL INFO Channel 10/0 : 52[4] -> 51[3] via P2P/IPC
+ip-26-0-165-213:9628:9786 [3] NCCL INFO Channel 06/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-166-244:897244:897407 [4] NCCL INFO Channel 01/0 : 44[4] -> 43[3] via P2P/IPC
+ip-26-0-162-46:702887:703045 [7] NCCL INFO Channel 09/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-165-213:9630:9789 [5] NCCL INFO Channel 01/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-166-36:104432:104592 [3] NCCL INFO Channel 04/0 : 51[3] -> 50[2] via P2P/IPC
+ip-26-0-166-244:897247:897404 [7] NCCL INFO Channel 12/0 : 47[7] -> 46[6] via P2P/IPC
+ip-26-0-162-46:702887:703045 [7] NCCL INFO Channel 10/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-165-213:9631:9788 [6] NCCL INFO Channel 03/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-165-213:9628:9786 [3] NCCL INFO Channel 07/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-166-36:104433:104590 [4] NCCL INFO Channel 12/0 : 52[4] -> 51[3] via P2P/IPC
+ip-26-0-165-213:9630:9789 [5] NCCL INFO Channel 02/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-162-46:702887:703045 [7] NCCL INFO Channel 11/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-165-213:9631:9788 [6] NCCL INFO Channel 04/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-166-36:104432:104592 [3] NCCL INFO Channel 05/0 : 51[3] -> 50[2] via P2P/IPC
+ip-26-0-166-214:624949:625110 [6] NCCL INFO Channel 01/0 : 38[6] -> 37[5] via P2P/IPC
+ip-26-0-166-244:897244:897407 [4] NCCL INFO Channel 02/0 : 44[4] -> 43[3] via P2P/IPC
+ip-26-0-165-213:9628:9786 [3] NCCL INFO Channel 09/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-166-244:897247:897404 [7] NCCL INFO Channel 13/0 : 47[7] -> 46[6] via P2P/IPC
+ip-26-0-166-15:127294:127457 [4] NCCL INFO Channel 04/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-165-213:9630:9789 [5] NCCL INFO Channel 03/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-165-213:9631:9788 [6] NCCL INFO Channel 07/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-166-15:127294:127457 [4] NCCL INFO Channel 09/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-165-213:9628:9786 [3] NCCL INFO Channel 11/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-166-36:104432:104592 [3] NCCL INFO Channel 06/0 : 51[3] -> 50[2] via P2P/IPC
+ip-26-0-166-214:624946:625107 [3] NCCL INFO Channel 01/0 : 35[3] -> 34[2] via P2P/IPC
+ip-26-0-166-214:624949:625110 [6] NCCL INFO Channel 02/0 : 38[6] -> 37[5] via P2P/IPC
+ip-26-0-165-213:9630:9789 [5] NCCL INFO Channel 06/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-166-125:762819:762982 [4] NCCL INFO Channel 02/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-166-36:104435:104595 [6] NCCL INFO Channel 01/0 : 54[6] -> 53[5] via P2P/IPC
+ip-26-0-165-213:9631:9788 [6] NCCL INFO Channel 09/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-166-15:127294:127457 [4] NCCL INFO Channel 10/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-166-125:762822:762984 [7] NCCL INFO Channel 12/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-166-244:897244:897407 [4] NCCL INFO Channel 04/0 : 44[4] -> 43[3] via P2P/IPC
+ip-26-0-165-213:9628:9786 [3] NCCL INFO Channel 12/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-162-46:702887:703045 [7] NCCL INFO Channel 12/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-166-36:104432:104592 [3] NCCL INFO Channel 07/0 : 51[3] -> 50[2] via P2P/IPC
+ip-26-0-166-214:624946:625107 [3] NCCL INFO Channel 03/0 : 35[3] -> 34[2] via P2P/IPC
+ip-26-0-165-213:9630:9789 [5] NCCL INFO Channel 07/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-166-214:624949:625110 [6] NCCL INFO Channel 03/0 : 38[6] -> 37[5] via P2P/IPC
+ip-26-0-166-36:104435:104595 [6] NCCL INFO Channel 02/0 : 54[6] -> 53[5] via P2P/IPC
+ip-26-0-166-15:127293:127454 [3] NCCL INFO Channel 01/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-166-244:897243:897403 [3] NCCL INFO Channel 01/0 : 43[3] -> 42[2] via P2P/IPC
+ip-26-0-165-213:9631:9788 [6] NCCL INFO Channel 10/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-166-125:762819:762982 [4] NCCL INFO Channel 04/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-165-213:9628:9786 [3] NCCL INFO Channel 13/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-162-46:702887:703045 [7] NCCL INFO Channel 13/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-166-125:762822:762984 [7] NCCL INFO Channel 13/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-166-36:104432:104592 [3] NCCL INFO Channel 09/0 : 51[3] -> 50[2] via P2P/IPC
+ip-26-0-166-244:897244:897407 [4] NCCL INFO Channel 09/0 : 44[4] -> 43[3] via P2P/IPC
+ip-26-0-166-214:624949:625110 [6] NCCL INFO Channel 04/0 : 38[6] -> 37[5] via P2P/IPC
+ip-26-0-166-36:104435:104595 [6] NCCL INFO Channel 03/0 : 54[6] -> 53[5] via P2P/IPC
+ip-26-0-166-214:624946:625107 [3] NCCL INFO Channel 04/0 : 35[3] -> 34[2] via P2P/IPC
+ip-26-0-165-213:9630:9789 [5] NCCL INFO Channel 09/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-165-213:9631:9788 [6] NCCL INFO Channel 11/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-166-244:897243:897403 [3] NCCL INFO Channel 03/0 : 43[3] -> 42[2] via P2P/IPC
+ip-26-0-165-213:9628:9786 [3] NCCL INFO Channel 14/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-166-36:104432:104592 [3] NCCL INFO Channel 11/0 : 51[3] -> 50[2] via P2P/IPC
+ip-26-0-166-214:624948:625104 [5] NCCL INFO Channel 01/0 : 37[5] -> 36[4] via P2P/IPC
+ip-26-0-165-213:9630:9789 [5] NCCL INFO Channel 10/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-166-214:624949:625110 [6] NCCL INFO Channel 07/0 : 38[6] -> 37[5] via P2P/IPC
+ip-26-0-165-213:9631:9788 [6] NCCL INFO Channel 12/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-166-244:897244:897407 [4] NCCL INFO Channel 10/0 : 44[4] -> 43[3] via P2P/IPC
+ip-26-0-166-214:624946:625107 [3] NCCL INFO Channel 05/0 : 35[3] -> 34[2] via P2P/IPC
+ip-26-0-165-213:9628:9786 [3] NCCL INFO Channel 15/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-166-36:104432:104592 [3] NCCL INFO Channel 12/0 : 51[3] -> 50[2] via P2P/IPC
+ip-26-0-166-244:897243:897403 [3] NCCL INFO Channel 04/0 : 43[3] -> 42[2] via P2P/IPC
+ip-26-0-166-214:624948:625104 [5] NCCL INFO Channel 02/0 : 37[5] -> 36[4] via P2P/IPC
+ip-26-0-166-214:624949:625110 [6] NCCL INFO Channel 09/0 : 38[6] -> 37[5] via P2P/IPC
+ip-26-0-166-36:104432:104592 [3] NCCL INFO Channel 13/0 : 51[3] -> 50[2] via P2P/IPC
+ip-26-0-166-214:624946:625107 [3] NCCL INFO Channel 06/0 : 35[3] -> 34[2] via P2P/IPC
+ip-26-0-166-244:897244:897407 [4] NCCL INFO Channel 12/0 : 44[4] -> 43[3] via P2P/IPC
+ip-26-0-162-46:702884:703047 [4] NCCL INFO Channel 01/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-166-36:104432:104592 [3] NCCL INFO Channel 14/0 : 51[3] -> 50[2] via P2P/IPC
+ip-26-0-166-36:104435:104595 [6] NCCL INFO Channel 04/0 : 54[6] -> 53[5] via P2P/IPC
+ip-26-0-166-214:624948:625104 [5] NCCL INFO Channel 03/0 : 37[5] -> 36[4] via P2P/IPC
+ip-26-0-166-36:104434:104591 [5] NCCL INFO Channel 01/0 : 53[5] -> 52[4] via P2P/IPC
+ip-26-0-166-214:624949:625110 [6] NCCL INFO Channel 10/0 : 38[6] -> 37[5] via P2P/IPC
+ip-26-0-166-214:624946:625107 [3] NCCL INFO Channel 07/0 : 35[3] -> 34[2] via P2P/IPC
+ip-26-0-166-36:104432:104592 [3] NCCL INFO Channel 15/0 : 51[3] -> 50[2] via P2P/IPC
+ip-26-0-166-36:104435:104595 [6] NCCL INFO Channel 07/0 : 54[6] -> 53[5] via P2P/IPC
+ip-26-0-162-46:702884:703047 [4] NCCL INFO Channel 02/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-166-36:104434:104591 [5] NCCL INFO Channel 02/0 : 53[5] -> 52[4] via P2P/IPC
+ip-26-0-166-214:624946:625107 [3] NCCL INFO Channel 09/0 : 35[3] -> 34[2] via P2P/IPC
+ip-26-0-166-214:624946:625107 [3] NCCL INFO Channel 11/0 : 35[3] -> 34[2] via P2P/IPC
+ip-26-0-166-36:104435:104595 [6] NCCL INFO Channel 09/0 : 54[6] -> 53[5] via P2P/IPC
+ip-26-0-166-36:104434:104591 [5] NCCL INFO Channel 03/0 : 53[5] -> 52[4] via P2P/IPC
+ip-26-0-162-46:702884:703047 [4] NCCL INFO Channel 04/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-166-214:624946:625107 [3] NCCL INFO Channel 12/0 : 35[3] -> 34[2] via P2P/IPC
+ip-26-0-166-244:897243:897403 [3] NCCL INFO Channel 05/0 : 43[3] -> 42[2] via P2P/IPC
+ip-26-0-166-214:624948:625104 [5] NCCL INFO Channel 06/0 : 37[5] -> 36[4] via P2P/IPC
+ip-26-0-166-214:624946:625107 [3] NCCL INFO Channel 13/0 : 35[3] -> 34[2] via P2P/IPC
+ip-26-0-166-244:897246:897405 [6] NCCL INFO Channel 01/0 : 46[6] -> 45[5] via P2P/IPC
+ip-26-0-162-46:702884:703047 [4] NCCL INFO Channel 09/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-166-214:624948:625104 [5] NCCL INFO Channel 07/0 : 37[5] -> 36[4] via P2P/IPC
+ip-26-0-166-214:624949:625110 [6] NCCL INFO Channel 11/0 : 38[6] -> 37[5] via P2P/IPC
+ip-26-0-166-214:624946:625107 [3] NCCL INFO Channel 14/0 : 35[3] -> 34[2] via P2P/IPC
+ip-26-0-166-125:762819:762982 [4] NCCL INFO Channel 09/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-166-15:127294:127457 [4] NCCL INFO Channel 12/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-166-214:624948:625104 [5] NCCL INFO Channel 09/0 : 37[5] -> 36[4] via P2P/IPC
+ip-26-0-166-214:624949:625110 [6] NCCL INFO Channel 12/0 : 38[6] -> 37[5] via P2P/IPC
+ip-26-0-162-46:702884:703047 [4] NCCL INFO Channel 10/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-162-46:702883:703044 [3] NCCL INFO Channel 01/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-214:624946:625107 [3] NCCL INFO Channel 15/0 : 35[3] -> 34[2] via P2P/IPC
+ip-26-0-166-15:127293:127454 [3] NCCL INFO Channel 03/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-166-214:624948:625104 [5] NCCL INFO Channel 10/0 : 37[5] -> 36[4] via P2P/IPC
+ip-26-0-166-214:624949:625110 [6] NCCL INFO Channel 15/0 : 38[6] -> 37[5] via P2P/IPC
+ip-26-0-166-125:762819:762982 [4] NCCL INFO Channel 10/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-162-46:702884:703047 [4] NCCL INFO Channel 12/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-167-9:1039544:1039702 [7] NCCL INFO Channel 04/0 : 63[7] -> 62[6] via P2P/IPC
+ip-26-0-162-46:702883:703044 [3] NCCL INFO Channel 03/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-15:127293:127454 [3] NCCL INFO Channel 04/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-166-244:897246:897405 [6] NCCL INFO Channel 02/0 : 46[6] -> 45[5] via P2P/IPC
+ip-26-0-166-244:897243:897403 [3] NCCL INFO Channel 06/0 : 43[3] -> 42[2] via P2P/IPC
+ip-26-0-162-46:702883:703044 [3] NCCL INFO Channel 04/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-125:762819:762982 [4] NCCL INFO Channel 12/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-166-244:897245:897401 [5] NCCL INFO Channel 01/0 : 45[5] -> 44[4] via P2P/IPC
+ip-26-0-166-244:897246:897405 [6] NCCL INFO Channel 03/0 : 46[6] -> 45[5] via P2P/IPC
+ip-26-0-166-244:897243:897403 [3] NCCL INFO Channel 07/0 : 43[3] -> 42[2] via P2P/IPC
+ip-26-0-166-125:762818:762985 [3] NCCL INFO Channel 01/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-162-46:702886:703049 [6] NCCL INFO Channel 01/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-162-46:702883:703044 [3] NCCL INFO Channel 05/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-244:897245:897401 [5] NCCL INFO Channel 02/0 : 45[5] -> 44[4] via P2P/IPC
+ip-26-0-166-244:897246:897405 [6] NCCL INFO Channel 04/0 : 46[6] -> 45[5] via P2P/IPC
+ip-26-0-166-244:897243:897403 [3] NCCL INFO Channel 09/0 : 43[3] -> 42[2] via P2P/IPC
+ip-26-0-162-46:702886:703049 [6] NCCL INFO Channel 02/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-166-125:762821:762988 [6] NCCL INFO Channel 01/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-162-46:702883:703044 [3] NCCL INFO Channel 06/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-125:762818:762985 [3] NCCL INFO Channel 03/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-166-15:127293:127454 [3] NCCL INFO Channel 05/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-167-9:1039544:1039702 [7] NCCL INFO Channel 05/0 : 63[7] -> 62[6] via P2P/IPC
+ip-26-0-166-125:762821:762988 [6] NCCL INFO Channel 02/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-166-125:762818:762985 [3] NCCL INFO Channel 04/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-162-46:702883:703044 [3] NCCL INFO Channel 07/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-162-46:702885:703048 [5] NCCL INFO Channel 01/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-162-46:702883:703044 [3] NCCL INFO Channel 09/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-125:762821:762988 [6] NCCL INFO Channel 03/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-162-46:702886:703049 [6] NCCL INFO Channel 03/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-166-15:127293:127454 [3] NCCL INFO Channel 06/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-166-125:762818:762985 [3] NCCL INFO Channel 05/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-162-46:702885:703048 [5] NCCL INFO Channel 02/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-15:127293:127454 [3] NCCL INFO Channel 07/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-167-9:1039544:1039702 [7] NCCL INFO Channel 09/0 : 63[7] -> 62[6] via P2P/IPC
+ip-26-0-166-125:762821:762988 [6] NCCL INFO Channel 04/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-162-46:702883:703044 [3] NCCL INFO Channel 11/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-125:762818:762985 [3] NCCL INFO Channel 06/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-162-46:702886:703049 [6] NCCL INFO Channel 04/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-166-214:624948:625104 [5] NCCL INFO Channel 11/0 : 37[5] -> 36[4] via P2P/IPC
+ip-26-0-162-46:702885:703048 [5] NCCL INFO Channel 03/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-15:127293:127454 [3] NCCL INFO Channel 09/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-166-125:762820:762981 [5] NCCL INFO Channel 01/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-166-244:897245:897401 [5] NCCL INFO Channel 03/0 : 45[5] -> 44[4] via P2P/IPC
+ip-26-0-166-15:127293:127454 [3] NCCL INFO Channel 11/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-162-46:702883:703044 [3] NCCL INFO Channel 12/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-162-46:702886:703049 [6] NCCL INFO Channel 07/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-162-46:702885:703048 [5] NCCL INFO Channel 06/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-15:127293:127454 [3] NCCL INFO Channel 12/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-167-9:1039544:1039702 [7] NCCL INFO Channel 10/0 : 63[7] -> 62[6] via P2P/IPC
+ip-26-0-162-46:702883:703044 [3] NCCL INFO Channel 13/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-15:127293:127454 [3] NCCL INFO Channel 13/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-166-15:127296:127451 [6] NCCL INFO Channel 01/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-166-36:104435:104595 [6] NCCL INFO Channel 10/0 : 54[6] -> 53[5] via P2P/IPC
+ip-26-0-162-46:702886:703049 [6] NCCL INFO Channel 09/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-162-46:702885:703048 [5] NCCL INFO Channel 07/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-15:127293:127454 [3] NCCL INFO Channel 14/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-162-46:702883:703044 [3] NCCL INFO Channel 14/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-125:762821:762988 [6] NCCL INFO Channel 07/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-166-15:127296:127451 [6] NCCL INFO Channel 02/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-166-125:762818:762985 [3] NCCL INFO Channel 07/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-166-125:762820:762981 [5] NCCL INFO Channel 02/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-167-9:1039544:1039702 [7] NCCL INFO Channel 11/0 : 63[7] -> 62[6] via P2P/IPC
+ip-26-0-162-46:702886:703049 [6] NCCL INFO Channel 10/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-162-46:702885:703048 [5] NCCL INFO Channel 09/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-15:127293:127454 [3] NCCL INFO Channel 15/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-166-125:762821:762988 [6] NCCL INFO Channel 09/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-162-46:702883:703044 [3] NCCL INFO Channel 15/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-244:897246:897405 [6] NCCL INFO Channel 07/0 : 46[6] -> 45[5] via P2P/IPC
+ip-26-0-166-15:127296:127451 [6] NCCL INFO Channel 03/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-166-125:762818:762985 [3] NCCL INFO Channel 09/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-166-125:762820:762981 [5] NCCL INFO Channel 03/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-166-15:127295:127452 [5] NCCL INFO Channel 01/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-166-244:897243:897403 [3] NCCL INFO Channel 11/0 : 43[3] -> 42[2] via P2P/IPC
+ip-26-0-167-9:1039540:1039705 [3] NCCL INFO Channel 01/0 : 59[3] -> 58[2] via P2P/IPC
+ip-26-0-162-46:702886:703049 [6] NCCL INFO Channel 11/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-162-46:702885:703048 [5] NCCL INFO Channel 10/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-125:762821:762988 [6] NCCL INFO Channel 10/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-166-15:127296:127451 [6] NCCL INFO Channel 04/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-166-125:762818:762985 [3] NCCL INFO Channel 11/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO Channel 01/0 : 60[4] -> 59[3] via P2P/IPC
+ip-26-0-166-15:127295:127452 [5] NCCL INFO Channel 02/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-167-9:1039544:1039702 [7] NCCL INFO Channel 12/0 : 63[7] -> 62[6] via P2P/IPC
+ip-26-0-165-213:9630:9789 [5] NCCL INFO Channel 11/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-166-125:762820:762981 [5] NCCL INFO Channel 06/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-162-46:702886:703049 [6] NCCL INFO Channel 12/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-166-125:762821:762988 [6] NCCL INFO Channel 11/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-166-125:762818:762985 [3] NCCL INFO Channel 12/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-167-9:1039540:1039705 [3] NCCL INFO Channel 03/0 : 59[3] -> 58[2] via P2P/IPC
+ip-26-0-166-125:762820:762981 [5] NCCL INFO Channel 07/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-166-125:762821:762988 [6] NCCL INFO Channel 12/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO Channel 02/0 : 60[4] -> 59[3] via P2P/IPC
+ip-26-0-167-9:1039544:1039702 [7] NCCL INFO Channel 13/0 : 63[7] -> 62[6] via P2P/IPC
+ip-26-0-166-125:762818:762985 [3] NCCL INFO Channel 13/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-166-125:762820:762981 [5] NCCL INFO Channel 09/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-166-244:897245:897401 [5] NCCL INFO Channel 06/0 : 45[5] -> 44[4] via P2P/IPC
+ip-26-0-166-244:897246:897405 [6] NCCL INFO Channel 09/0 : 46[6] -> 45[5] via P2P/IPC
+ip-26-0-166-125:762821:762988 [6] NCCL INFO Channel 15/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-167-9:1039540:1039705 [3] NCCL INFO Channel 04/0 : 59[3] -> 58[2] via P2P/IPC
+ip-26-0-167-9:1039543:1039699 [6] NCCL INFO Channel 01/0 : 62[6] -> 61[5] via P2P/IPC
+ip-26-0-166-125:762818:762985 [3] NCCL INFO Channel 14/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-166-244:897243:897403 [3] NCCL INFO Channel 12/0 : 43[3] -> 42[2] via P2P/IPC
+ip-26-0-166-125:762820:762981 [5] NCCL INFO Channel 10/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-166-244:897245:897401 [5] NCCL INFO Channel 07/0 : 45[5] -> 44[4] via P2P/IPC
+ip-26-0-162-46:702885:703048 [5] NCCL INFO Channel 11/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-244:897246:897405 [6] NCCL INFO Channel 10/0 : 46[6] -> 45[5] via P2P/IPC
+ip-26-0-166-36:104434:104591 [5] NCCL INFO Channel 06/0 : 53[5] -> 52[4] via P2P/IPC
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO Channel 04/0 : 60[4] -> 59[3] via P2P/IPC
+ip-26-0-166-125:762818:762985 [3] NCCL INFO Channel 15/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-166-244:897243:897403 [3] NCCL INFO Channel 13/0 : 43[3] -> 42[2] via P2P/IPC
+ip-26-0-167-9:1039540:1039705 [3] NCCL INFO Channel 05/0 : 59[3] -> 58[2] via P2P/IPC
+ip-26-0-166-244:897245:897401 [5] NCCL INFO Channel 09/0 : 45[5] -> 44[4] via P2P/IPC
+ip-26-0-167-9:1039543:1039699 [6] NCCL INFO Channel 02/0 : 62[6] -> 61[5] via P2P/IPC
+ip-26-0-166-244:897246:897405 [6] NCCL INFO Channel 11/0 : 46[6] -> 45[5] via P2P/IPC
+ip-26-0-165-213:9631:9788 [6] NCCL INFO Channel 15/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO Channel 09/0 : 60[4] -> 59[3] via P2P/IPC
+ip-26-0-166-244:897243:897403 [3] NCCL INFO Channel 14/0 : 43[3] -> 42[2] via P2P/IPC
+ip-26-0-166-244:897245:897401 [5] NCCL INFO Channel 10/0 : 45[5] -> 44[4] via P2P/IPC
+ip-26-0-166-244:897246:897405 [6] NCCL INFO Channel 12/0 : 46[6] -> 45[5] via P2P/IPC
+ip-26-0-167-9:1039540:1039705 [3] NCCL INFO Channel 06/0 : 59[3] -> 58[2] via P2P/IPC
+ip-26-0-167-9:1039543:1039699 [6] NCCL INFO Channel 03/0 : 62[6] -> 61[5] via P2P/IPC
+ip-26-0-166-244:897245:897401 [5] NCCL INFO Channel 11/0 : 45[5] -> 44[4] via P2P/IPC
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO Channel 10/0 : 60[4] -> 59[3] via P2P/IPC
+ip-26-0-166-244:897243:897403 [3] NCCL INFO Channel 15/0 : 43[3] -> 42[2] via P2P/IPC
+ip-26-0-166-214:624948:625104 [5] NCCL INFO Channel 14/0 : 37[5] -> 36[4] via P2P/IPC
+ip-26-0-167-9:1039542:1039701 [5] NCCL INFO Channel 01/0 : 61[5] -> 60[4] via P2P/IPC
+ip-26-0-166-244:897246:897405 [6] NCCL INFO Channel 15/0 : 46[6] -> 45[5] via P2P/IPC
+ip-26-0-166-36:104435:104595 [6] NCCL INFO Channel 11/0 : 54[6] -> 53[5] via P2P/IPC
+ip-26-0-167-9:1039540:1039705 [3] NCCL INFO Channel 07/0 : 59[3] -> 58[2] via P2P/IPC
+ip-26-0-167-9:1039543:1039699 [6] NCCL INFO Channel 04/0 : 62[6] -> 61[5] via P2P/IPC
+ip-26-0-166-214:624948:625104 [5] NCCL INFO Channel 15/0 : 37[5] -> 36[4] via P2P/IPC
+ip-26-0-166-15:127296:127451 [6] NCCL INFO Channel 07/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-166-244:897245:897401 [5] NCCL INFO Channel 14/0 : 45[5] -> 44[4] via P2P/IPC
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO Channel 12/0 : 60[4] -> 59[3] via P2P/IPC
+ip-26-0-166-125:762820:762981 [5] NCCL INFO Channel 11/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-167-9:1039540:1039705 [3] NCCL INFO Channel 09/0 : 59[3] -> 58[2] via P2P/IPC
+ip-26-0-167-9:1039540:1039705 [3] NCCL INFO Channel 11/0 : 59[3] -> 58[2] via P2P/IPC
+ip-26-0-167-9:1039540:1039705 [3] NCCL INFO Channel 12/0 : 59[3] -> 58[2] via P2P/IPC
+ip-26-0-166-36:104434:104591 [5] NCCL INFO Channel 07/0 : 53[5] -> 52[4] via P2P/IPC
+ip-26-0-167-9:1039540:1039705 [3] NCCL INFO Channel 13/0 : 59[3] -> 58[2] via P2P/IPC
+ip-26-0-167-9:1039540:1039705 [3] NCCL INFO Channel 14/0 : 59[3] -> 58[2] via P2P/IPC
+ip-26-0-167-9:1039540:1039705 [3] NCCL INFO Channel 15/0 : 59[3] -> 58[2] via P2P/IPC
+ip-26-0-167-9:1039543:1039699 [6] NCCL INFO Channel 07/0 : 62[6] -> 61[5] via P2P/IPC
+ip-26-0-167-9:1039542:1039701 [5] NCCL INFO Channel 02/0 : 61[5] -> 60[4] via P2P/IPC
+ip-26-0-167-9:1039543:1039699 [6] NCCL INFO Channel 09/0 : 62[6] -> 61[5] via P2P/IPC
+ip-26-0-162-46:702885:703048 [5] NCCL INFO Channel 14/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-125:762820:762981 [5] NCCL INFO Channel 14/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-162-46:702885:703048 [5] NCCL INFO Channel 15/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-162-46:702886:703049 [6] NCCL INFO Channel 15/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-166-36:104435:104595 [6] NCCL INFO Channel 12/0 : 54[6] -> 53[5] via P2P/IPC
+ip-26-0-165-213:9630:9789 [5] NCCL INFO Channel 14/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-166-244:897245:897401 [5] NCCL INFO Channel 15/0 : 45[5] -> 44[4] via P2P/IPC
+ip-26-0-167-9:1039542:1039701 [5] NCCL INFO Channel 03/0 : 61[5] -> 60[4] via P2P/IPC
+ip-26-0-167-9:1039542:1039701 [5] NCCL INFO Channel 06/0 : 61[5] -> 60[4] via P2P/IPC
+ip-26-0-166-15:127295:127452 [5] NCCL INFO Channel 03/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-167-9:1039543:1039699 [6] NCCL INFO Channel 10/0 : 62[6] -> 61[5] via P2P/IPC
+ip-26-0-166-36:104434:104591 [5] NCCL INFO Channel 09/0 : 53[5] -> 52[4] via P2P/IPC
+ip-26-0-166-15:127296:127451 [6] NCCL INFO Channel 09/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-166-36:104435:104595 [6] NCCL INFO Channel 15/0 : 54[6] -> 53[5] via P2P/IPC
+ip-26-0-166-15:127295:127452 [5] NCCL INFO Channel 06/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-166-36:104434:104591 [5] NCCL INFO Channel 10/0 : 53[5] -> 52[4] via P2P/IPC
+ip-26-0-167-9:1039543:1039699 [6] NCCL INFO Channel 11/0 : 62[6] -> 61[5] via P2P/IPC
+ip-26-0-167-9:1039542:1039701 [5] NCCL INFO Channel 07/0 : 61[5] -> 60[4] via P2P/IPC
+ip-26-0-167-9:1039542:1039701 [5] NCCL INFO Channel 09/0 : 61[5] -> 60[4] via P2P/IPC
+ip-26-0-166-15:127296:127451 [6] NCCL INFO Channel 10/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-166-15:127295:127452 [5] NCCL INFO Channel 07/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-167-9:1039543:1039699 [6] NCCL INFO Channel 12/0 : 62[6] -> 61[5] via P2P/IPC
+ip-26-0-166-36:104434:104591 [5] NCCL INFO Channel 11/0 : 53[5] -> 52[4] via P2P/IPC
+ip-26-0-167-9:1039543:1039699 [6] NCCL INFO Channel 15/0 : 62[6] -> 61[5] via P2P/IPC
+ip-26-0-165-213:9630:9789 [5] NCCL INFO Channel 15/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-166-125:762820:762981 [5] NCCL INFO Channel 15/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-166-15:127295:127452 [5] NCCL INFO Channel 09/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-166-15:127296:127451 [6] NCCL INFO Channel 11/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-166-36:104434:104591 [5] NCCL INFO Channel 14/0 : 53[5] -> 52[4] via P2P/IPC
+ip-26-0-166-15:127295:127452 [5] NCCL INFO Channel 10/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-166-15:127296:127451 [6] NCCL INFO Channel 12/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-166-15:127295:127452 [5] NCCL INFO Channel 11/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-167-9:1039542:1039701 [5] NCCL INFO Channel 10/0 : 61[5] -> 60[4] via P2P/IPC
+ip-26-0-166-15:127295:127452 [5] NCCL INFO Channel 14/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-166-15:127296:127451 [6] NCCL INFO Channel 15/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-167-9:1039542:1039701 [5] NCCL INFO Channel 11/0 : 61[5] -> 60[4] via P2P/IPC
+ip-26-0-166-36:104434:104591 [5] NCCL INFO Channel 15/0 : 53[5] -> 52[4] via P2P/IPC
+ip-26-0-166-15:127295:127452 [5] NCCL INFO Channel 15/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-167-9:1039542:1039701 [5] NCCL INFO Channel 14/0 : 61[5] -> 60[4] via P2P/IPC
+ip-26-0-167-9:1039542:1039701 [5] NCCL INFO Channel 15/0 : 61[5] -> 60[4] via P2P/IPC
+ip-26-0-166-244:897240:897406 [0] NCCL INFO Connected all rings
+ip-26-0-166-244:897240:897406 [0] NCCL INFO Channel 02/0 : 40[0] -> 41[1] via P2P/IPC
+ip-26-0-166-244:897241:897402 [1] NCCL INFO Connected all rings
+ip-26-0-166-15:127290:127456 [0] NCCL INFO Connected all rings
+ip-26-0-166-15:127290:127456 [0] NCCL INFO Channel 02/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-166-15:127290:127456 [0] NCCL INFO Channel 03/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-166-15:127290:127456 [0] NCCL INFO Channel 04/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-166-15:127290:127456 [0] NCCL INFO Channel 05/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-166-15:127290:127456 [0] NCCL INFO Channel 06/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-166-15:127290:127456 [0] NCCL INFO Channel 07/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-166-15:127290:127456 [0] NCCL INFO Channel 10/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-165-213:9627:9791 [2] NCCL INFO Connected all rings
+ip-26-0-166-15:127290:127456 [0] NCCL INFO Channel 11/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-166-15:127290:127456 [0] NCCL INFO Channel 12/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-166-15:127290:127456 [0] NCCL INFO Channel 13/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-166-15:127290:127456 [0] NCCL INFO Channel 14/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-167-9:1039537:1039703 [0] NCCL INFO Connected all rings
+ip-26-0-167-9:1039537:1039703 [0] NCCL INFO Channel 02/0 : 56[0] -> 57[1] via P2P/IPC
+ip-26-0-166-15:127290:127456 [0] NCCL INFO Channel 15/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-167-9:1039538:1039704 [1] NCCL INFO Connected all rings
+ip-26-0-166-125:762816:762987 [1] NCCL INFO Connected all rings
+ip-26-0-166-36:104432:104592 [3] NCCL INFO Connected all rings
+ip-26-0-166-15:127293:127454 [3] NCCL INFO Connected all rings
+ip-26-0-166-15:127296:127451 [6] NCCL INFO Connected all rings
+ip-26-0-165-213:9625:9787 [0] NCCL INFO Connected all rings
+ip-26-0-165-213:9625:9787 [0] NCCL INFO Channel 02/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-165-213:9625:9787 [0] NCCL INFO Channel 03/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-165-213:9626:9790 [1] NCCL INFO Connected all rings
+ip-26-0-165-213:9625:9787 [0] NCCL INFO Channel 04/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-162-46:702887:703045 [7] NCCL INFO Connected all rings
+ip-26-0-165-213:9625:9787 [0] NCCL INFO Channel 05/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-165-213:9625:9787 [0] NCCL INFO Channel 06/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-165-213:9625:9787 [0] NCCL INFO Channel 07/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-165-213:9625:9787 [0] NCCL INFO Channel 10/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-165-213:9625:9787 [0] NCCL INFO Channel 11/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-165-213:9625:9787 [0] NCCL INFO Channel 12/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-167-9:1039538:1039704 [1] NCCL INFO Channel 01/0 : 57[1] -> 58[2] via P2P/IPC
+ip-26-0-167-9:1039538:1039704 [1] NCCL INFO Channel 03/0 : 57[1] -> 58[2] via P2P/IPC
+ip-26-0-165-213:9625:9787 [0] NCCL INFO Channel 13/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-165-213:9625:9787 [0] NCCL INFO Channel 14/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-165-213:9625:9787 [0] NCCL INFO Channel 15/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-165-213:9626:9790 [1] NCCL INFO Channel 01/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-165-213:9626:9790 [1] NCCL INFO Channel 03/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-167-9:1039540:1039705 [3] NCCL INFO Connected all rings
+ip-26-0-165-213:9626:9790 [1] NCCL INFO Channel 04/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-165-213:9626:9790 [1] NCCL INFO Channel 05/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-165-213:9626:9790 [1] NCCL INFO Channel 06/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-166-36:104429:104594 [0] NCCL INFO Connected all rings
+ip-26-0-166-36:104429:104594 [0] NCCL INFO Channel 02/0 : 48[0] -> 49[1] via P2P/IPC
+ip-26-0-165-213:9626:9790 [1] NCCL INFO Channel 07/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-165-213:9626:9790 [1] NCCL INFO Channel 09/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-165-213:9626:9790 [1] NCCL INFO Channel 11/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-166-125:762822:762984 [7] NCCL INFO Connected all rings
+ip-26-0-165-213:9626:9790 [1] NCCL INFO Channel 12/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-165-213:9626:9790 [1] NCCL INFO Channel 13/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-166-36:104429:104594 [0] NCCL INFO Channel 03/0 : 48[0] -> 49[1] via P2P/IPC
+ip-26-0-165-213:9626:9790 [1] NCCL INFO Channel 14/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-166-36:104429:104594 [0] NCCL INFO Channel 04/0 : 48[0] -> 49[1] via P2P/IPC
+ip-26-0-165-213:9626:9790 [1] NCCL INFO Channel 15/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-165-213:9625:9787 [0] NCCL INFO Channel 05/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-165-213:9625:9787 [0] NCCL INFO Channel 06/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-166-36:104429:104594 [0] NCCL INFO Channel 05/0 : 48[0] -> 49[1] via P2P/IPC
+ip-26-0-165-213:9625:9787 [0] NCCL INFO Channel 07/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-166-36:104429:104594 [0] NCCL INFO Channel 06/0 : 48[0] -> 49[1] via P2P/IPC
+ip-26-0-165-213:9625:9787 [0] NCCL INFO Channel 13/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-166-36:104429:104594 [0] NCCL INFO Channel 07/0 : 48[0] -> 49[1] via P2P/IPC
+ip-26-0-166-36:104429:104594 [0] NCCL INFO Channel 10/0 : 48[0] -> 49[1] via P2P/IPC
+ip-26-0-165-213:9628:9786 [3] NCCL INFO Connected all rings
+ip-26-0-166-36:104430:104589 [1] NCCL INFO Connected all rings
+ip-26-0-166-244:897240:897406 [0] NCCL INFO Channel 03/0 : 40[0] -> 41[1] via P2P/IPC
+ip-26-0-166-36:104429:104594 [0] NCCL INFO Channel 11/0 : 48[0] -> 49[1] via P2P/IPC
+ip-26-0-165-213:9625:9787 [0] NCCL INFO Channel 14/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-166-36:104431:104593 [2] NCCL INFO Connected all rings
+ip-26-0-165-213:9625:9787 [0] NCCL INFO Channel 15/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-166-36:104429:104594 [0] NCCL INFO Channel 12/0 : 48[0] -> 49[1] via P2P/IPC
+ip-26-0-166-15:127292:127458 [2] NCCL INFO Connected all rings
+ip-26-0-166-36:104429:104594 [0] NCCL INFO Channel 13/0 : 48[0] -> 49[1] via P2P/IPC
+ip-26-0-166-36:104429:104594 [0] NCCL INFO Channel 14/0 : 48[0] -> 49[1] via P2P/IPC
+ip-26-0-166-244:897240:897406 [0] NCCL INFO Channel 04/0 : 40[0] -> 41[1] via P2P/IPC
+ip-26-0-166-36:104429:104594 [0] NCCL INFO Channel 15/0 : 48[0] -> 49[1] via P2P/IPC
+ip-26-0-166-244:897240:897406 [0] NCCL INFO Channel 05/0 : 40[0] -> 41[1] via P2P/IPC
+ip-26-0-162-46:702882:703046 [2] NCCL INFO Connected all rings
+ip-26-0-166-244:897240:897406 [0] NCCL INFO Channel 06/0 : 40[0] -> 41[1] via P2P/IPC
+ip-26-0-166-214:624944:625106 [1] NCCL INFO Connected all rings
+ip-26-0-166-244:897240:897406 [0] NCCL INFO Channel 07/0 : 40[0] -> 41[1] via P2P/IPC
+ip-26-0-162-46:702880:703042 [0] NCCL INFO Connected all rings
+ip-26-0-162-46:702880:703042 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-244:897240:897406 [0] NCCL INFO Channel 10/0 : 40[0] -> 41[1] via P2P/IPC
+ip-26-0-166-244:897240:897406 [0] NCCL INFO Channel 11/0 : 40[0] -> 41[1] via P2P/IPC
+ip-26-0-162-46:702880:703042 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-36:104435:104595 [6] NCCL INFO Connected all rings
+ip-26-0-166-36:104430:104589 [1] NCCL INFO Channel 01/0 : 49[1] -> 50[2] via P2P/IPC
+ip-26-0-166-244:897240:897406 [0] NCCL INFO Channel 12/0 : 40[0] -> 41[1] via P2P/IPC
+ip-26-0-162-46:702880:703042 [0] NCCL INFO Channel 04/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-213:9631:9788 [6] NCCL INFO Connected all rings
+ip-26-0-166-36:104430:104589 [1] NCCL INFO Channel 03/0 : 49[1] -> 50[2] via P2P/IPC
+ip-26-0-162-46:702886:703049 [6] NCCL INFO Connected all rings
+ip-26-0-166-244:897240:897406 [0] NCCL INFO Channel 13/0 : 40[0] -> 41[1] via P2P/IPC
+ip-26-0-165-213:9627:9791 [2] NCCL INFO Channel 01/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-162-46:702880:703042 [0] NCCL INFO Channel 05/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-213:9627:9791 [2] NCCL INFO Channel 02/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-165-213:9627:9791 [2] NCCL INFO Channel 04/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-165-213:9627:9791 [2] NCCL INFO Channel 05/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-162-46:702880:703042 [0] NCCL INFO Channel 06/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-244:897242:897400 [2] NCCL INFO Connected all rings
+ip-26-0-162-46:702880:703042 [0] NCCL INFO Channel 07/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-244:897240:897406 [0] NCCL INFO Channel 14/0 : 40[0] -> 41[1] via P2P/IPC
+ip-26-0-166-244:897247:897404 [7] NCCL INFO Connected all rings
+ip-26-0-165-213:9627:9791 [2] NCCL INFO Channel 06/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-162-46:702880:703042 [0] NCCL INFO Channel 10/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-213:9627:9791 [2] NCCL INFO Channel 07/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-165-213:9627:9791 [2] NCCL INFO Channel 09/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-166-125:762815:762983 [0] NCCL INFO Connected all rings
+ip-26-0-166-244:897240:897406 [0] NCCL INFO Channel 15/0 : 40[0] -> 41[1] via P2P/IPC
+ip-26-0-166-125:762815:762983 [0] NCCL INFO Channel 02/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-166-36:104433:104590 [4] NCCL INFO Connected all rings
+ip-26-0-167-9:1039537:1039703 [0] NCCL INFO Channel 03/0 : 56[0] -> 57[1] via P2P/IPC
+ip-26-0-166-36:104430:104589 [1] NCCL INFO Channel 04/0 : 49[1] -> 50[2] via P2P/IPC
+ip-26-0-166-214:624945:625111 [2] NCCL INFO Connected all rings
+ip-26-0-165-213:9627:9791 [2] NCCL INFO Channel 10/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-162-46:702880:703042 [0] NCCL INFO Channel 11/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-244:897241:897402 [1] NCCL INFO Channel 01/0 : 41[1] -> 42[2] via P2P/IPC
+ip-26-0-165-213:9627:9791 [2] NCCL INFO Channel 12/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-166-244:897243:897403 [3] NCCL INFO Connected all rings
+ip-26-0-162-46:702880:703042 [0] NCCL INFO Channel 12/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-36:104431:104593 [2] NCCL INFO Channel 01/0 : 50[2] -> 51[3] via P2P/IPC
+ip-26-0-166-244:897241:897402 [1] NCCL INFO Channel 03/0 : 41[1] -> 42[2] via P2P/IPC
+ip-26-0-165-213:9627:9791 [2] NCCL INFO Channel 13/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-166-125:762815:762983 [0] NCCL INFO Channel 03/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-165-213:9627:9791 [2] NCCL INFO Channel 14/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-162-46:702880:703042 [0] NCCL INFO Channel 13/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-213:9627:9791 [2] NCCL INFO Channel 15/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-166-244:897241:897402 [1] NCCL INFO Channel 04/0 : 41[1] -> 42[2] via P2P/IPC
+ip-26-0-166-125:762815:762983 [0] NCCL INFO Channel 04/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-166-15:127292:127458 [2] NCCL INFO Channel 01/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-162-46:702882:703046 [2] NCCL INFO Channel 01/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-36:104431:104593 [2] NCCL INFO Channel 02/0 : 50[2] -> 51[3] via P2P/IPC
+ip-26-0-162-46:702880:703042 [0] NCCL INFO Channel 14/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-36:104430:104589 [1] NCCL INFO Channel 05/0 : 49[1] -> 50[2] via P2P/IPC
+ip-26-0-166-15:127292:127458 [2] NCCL INFO Channel 02/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-166-214:624944:625106 [1] NCCL INFO Channel 01/0 : 33[1] -> 34[2] via P2P/IPC
+ip-26-0-166-36:104431:104593 [2] NCCL INFO Channel 04/0 : 50[2] -> 51[3] via P2P/IPC
+ip-26-0-166-244:897241:897402 [1] NCCL INFO Channel 05/0 : 41[1] -> 42[2] via P2P/IPC
+ip-26-0-166-214:624944:625106 [1] NCCL INFO Channel 03/0 : 33[1] -> 34[2] via P2P/IPC
+ip-26-0-166-125:762815:762983 [0] NCCL INFO Channel 05/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-166-36:104430:104589 [1] NCCL INFO Channel 06/0 : 49[1] -> 50[2] via P2P/IPC
+ip-26-0-162-46:702882:703046 [2] NCCL INFO Channel 02/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-162-46:702880:703042 [0] NCCL INFO Channel 15/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-214:624944:625106 [1] NCCL INFO Channel 04/0 : 33[1] -> 34[2] via P2P/IPC
+ip-26-0-166-36:104431:104593 [2] NCCL INFO Channel 05/0 : 50[2] -> 51[3] via P2P/IPC
+ip-26-0-167-9:1039537:1039703 [0] NCCL INFO Channel 04/0 : 56[0] -> 57[1] via P2P/IPC
+ip-26-0-166-15:127292:127458 [2] NCCL INFO Channel 04/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-166-36:104430:104589 [1] NCCL INFO Channel 07/0 : 49[1] -> 50[2] via P2P/IPC
+ip-26-0-162-46:702882:703046 [2] NCCL INFO Channel 04/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-244:897245:897401 [5] NCCL INFO Connected all rings
+ip-26-0-167-9:1039537:1039703 [0] NCCL INFO Channel 05/0 : 56[0] -> 57[1] via P2P/IPC
+ip-26-0-165-213:9626:9790 [1] NCCL INFO Channel 09/0 : 1[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9626:9790 [1] NCCL INFO Channel 01/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624944:625106 [1] NCCL INFO Channel 05/0 : 33[1] -> 34[2] via P2P/IPC
+ip-26-0-166-36:104431:104593 [2] NCCL INFO Channel 06/0 : 50[2] -> 51[3] via P2P/IPC
+ip-26-0-165-213:9626:9790 [1] NCCL INFO Channel 09/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104430:104589 [1] NCCL INFO Channel 09/0 : 49[1] -> 50[2] via P2P/IPC
+ip-26-0-166-244:897241:897402 [1] NCCL INFO Channel 06/0 : 41[1] -> 42[2] via P2P/IPC
+ip-26-0-162-46:702882:703046 [2] NCCL INFO Channel 05/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-244:897241:897402 [1] NCCL INFO Channel 07/0 : 41[1] -> 42[2] via P2P/IPC
+ip-26-0-166-36:104431:104593 [2] NCCL INFO Channel 07/0 : 50[2] -> 51[3] via P2P/IPC
+ip-26-0-167-9:1039537:1039703 [0] NCCL INFO Channel 06/0 : 56[0] -> 57[1] via P2P/IPC
+ip-26-0-166-214:624944:625106 [1] NCCL INFO Channel 06/0 : 33[1] -> 34[2] via P2P/IPC
+ip-26-0-166-36:104430:104589 [1] NCCL INFO Channel 11/0 : 49[1] -> 50[2] via P2P/IPC
+ip-26-0-162-46:702882:703046 [2] NCCL INFO Channel 06/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-244:897241:897402 [1] NCCL INFO Channel 09/0 : 41[1] -> 42[2] via P2P/IPC
+ip-26-0-167-9:1039537:1039703 [0] NCCL INFO Channel 07/0 : 56[0] -> 57[1] via P2P/IPC
+ip-26-0-166-15:127292:127458 [2] NCCL INFO Channel 05/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-166-36:104431:104593 [2] NCCL INFO Channel 09/0 : 50[2] -> 51[3] via P2P/IPC
+ip-26-0-166-36:104430:104589 [1] NCCL INFO Channel 12/0 : 49[1] -> 50[2] via P2P/IPC
+ip-26-0-166-244:897241:897402 [1] NCCL INFO Channel 11/0 : 41[1] -> 42[2] via P2P/IPC
+ip-26-0-162-46:702882:703046 [2] NCCL INFO Channel 07/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-214:624944:625106 [1] NCCL INFO Channel 07/0 : 33[1] -> 34[2] via P2P/IPC
+ip-26-0-167-9:1039537:1039703 [0] NCCL INFO Channel 10/0 : 56[0] -> 57[1] via P2P/IPC
+ip-26-0-166-36:104431:104593 [2] NCCL INFO Channel 10/0 : 50[2] -> 51[3] via P2P/IPC
+ip-26-0-166-244:897241:897402 [1] NCCL INFO Channel 12/0 : 41[1] -> 42[2] via P2P/IPC
+ip-26-0-166-214:624948:625104 [5] NCCL INFO Connected all rings
+ip-26-0-162-46:702882:703046 [2] NCCL INFO Channel 09/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-36:104430:104589 [1] NCCL INFO Channel 13/0 : 49[1] -> 50[2] via P2P/IPC
+ip-26-0-166-214:624944:625106 [1] NCCL INFO Channel 09/0 : 33[1] -> 34[2] via P2P/IPC
+ip-26-0-166-36:104431:104593 [2] NCCL INFO Channel 12/0 : 50[2] -> 51[3] via P2P/IPC
+ip-26-0-166-244:897241:897402 [1] NCCL INFO Channel 13/0 : 41[1] -> 42[2] via P2P/IPC
+ip-26-0-167-9:1039537:1039703 [0] NCCL INFO Channel 11/0 : 56[0] -> 57[1] via P2P/IPC
+ip-26-0-166-125:762815:762983 [0] NCCL INFO Channel 06/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-166-244:897242:897400 [2] NCCL INFO Channel 01/0 : 42[2] -> 43[3] via P2P/IPC
+ip-26-0-166-36:104430:104589 [1] NCCL INFO Channel 14/0 : 49[1] -> 50[2] via P2P/IPC
+ip-26-0-166-125:762815:762983 [0] NCCL INFO Channel 07/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-162-46:702886:703049 [6] NCCL INFO Channel 01/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-244:897241:897402 [1] NCCL INFO Channel 14/0 : 41[1] -> 42[2] via P2P/IPC
+ip-26-0-166-15:127292:127458 [2] NCCL INFO Channel 06/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-166-36:104431:104593 [2] NCCL INFO Channel 13/0 : 50[2] -> 51[3] via P2P/IPC
+ip-26-0-162-46:702882:703046 [2] NCCL INFO Channel 10/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-213:9629:9793 [4] NCCL INFO Connected all rings
+ip-26-0-166-214:624944:625106 [1] NCCL INFO Channel 11/0 : 33[1] -> 34[2] via P2P/IPC
+ip-26-0-166-244:897242:897400 [2] NCCL INFO Channel 02/0 : 42[2] -> 43[3] via P2P/IPC
+ip-26-0-166-125:762815:762983 [0] NCCL INFO Channel 10/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-166-36:104430:104589 [1] NCCL INFO Channel 15/0 : 49[1] -> 50[2] via P2P/IPC
+ip-26-0-166-15:127292:127458 [2] NCCL INFO Channel 07/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-166-36:104431:104593 [2] NCCL INFO Channel 14/0 : 50[2] -> 51[3] via P2P/IPC
+ip-26-0-166-244:897241:897402 [1] NCCL INFO Channel 15/0 : 41[1] -> 42[2] via P2P/IPC
+ip-26-0-166-36:104434:104591 [5] NCCL INFO Connected all rings
+ip-26-0-166-125:762815:762983 [0] NCCL INFO Channel 11/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-167-9:1039537:1039703 [0] NCCL INFO Channel 12/0 : 56[0] -> 57[1] via P2P/IPC
+ip-26-0-166-244:897242:897400 [2] NCCL INFO Channel 04/0 : 42[2] -> 43[3] via P2P/IPC
+ip-26-0-166-36:104429:104594 [0] NCCL INFO Channel 05/0 : 48[0] -> 55[7] via P2P/IPC
+ip-26-0-166-15:127292:127458 [2] NCCL INFO Channel 09/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-166-214:624944:625106 [1] NCCL INFO Channel 12/0 : 33[1] -> 34[2] via P2P/IPC
+ip-26-0-166-125:762815:762983 [0] NCCL INFO Channel 12/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-162-46:702884:703047 [4] NCCL INFO Connected all rings
+ip-26-0-166-36:104431:104593 [2] NCCL INFO Channel 15/0 : 50[2] -> 51[3] via P2P/IPC
+ip-26-0-167-9:1039537:1039703 [0] NCCL INFO Channel 13/0 : 56[0] -> 57[1] via P2P/IPC
+ip-26-0-166-15:127292:127458 [2] NCCL INFO Channel 10/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-166-36:104429:104594 [0] NCCL INFO Channel 06/0 : 48[0] -> 55[7] via P2P/IPC
+ip-26-0-166-214:624944:625106 [1] NCCL INFO Channel 13/0 : 33[1] -> 34[2] via P2P/IPC
+ip-26-0-166-36:104432:104592 [3] NCCL INFO Channel 01/0 : 51[3] -> 52[4] via P2P/IPC
+ip-26-0-166-244:897246:897405 [6] NCCL INFO Connected all rings
+ip-26-0-166-125:762815:762983 [0] NCCL INFO Channel 13/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-167-9:1039537:1039703 [0] NCCL INFO Channel 14/0 : 56[0] -> 57[1] via P2P/IPC
+ip-26-0-166-36:104430:104589 [1] NCCL INFO Channel 01/0 : 41[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104430:104589 [1] NCCL INFO Channel 09/0 : 41[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127292:127458 [2] NCCL INFO Channel 12/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-166-36:104433:104590 [4] NCCL INFO Channel 01/0 : 52[4] -> 53[5] via P2P/IPC
+ip-26-0-166-125:762815:762983 [0] NCCL INFO Channel 14/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-166-36:104430:104589 [1] NCCL INFO Channel 01/0 : 49[1] -> 57[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104429:104594 [0] NCCL INFO Channel 07/0 : 48[0] -> 55[7] via P2P/IPC
+ip-26-0-166-36:104432:104592 [3] NCCL INFO Channel 02/0 : 51[3] -> 52[4] via P2P/IPC
+ip-26-0-166-244:897240:897406 [0] NCCL INFO Channel 05/0 : 40[0] -> 47[7] via P2P/IPC
+ip-26-0-162-46:702882:703046 [2] NCCL INFO Channel 12/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-244:897242:897400 [2] NCCL INFO Channel 05/0 : 42[2] -> 43[3] via P2P/IPC
+ip-26-0-165-213:9632:9792 [7] NCCL INFO Connected all rings
+ip-26-0-166-214:624944:625106 [1] NCCL INFO Channel 14/0 : 33[1] -> 34[2] via P2P/IPC
+ip-26-0-166-36:104433:104590 [4] NCCL INFO Channel 02/0 : 52[4] -> 53[5] via P2P/IPC
+ip-26-0-166-36:104429:104594 [0] NCCL INFO Channel 13/0 : 48[0] -> 55[7] via P2P/IPC
+ip-26-0-162-46:702882:703046 [2] NCCL INFO Channel 13/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-36:104432:104592 [3] NCCL INFO Channel 03/0 : 51[3] -> 52[4] via P2P/IPC
+ip-26-0-166-244:897243:897403 [3] NCCL INFO Channel 01/0 : 43[3] -> 44[4] via P2P/IPC
+ip-26-0-166-15:127296:127451 [6] NCCL INFO Channel 01/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-166-15:127292:127458 [2] NCCL INFO Channel 13/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-167-9:1039537:1039703 [0] NCCL INFO Channel 15/0 : 56[0] -> 57[1] via P2P/IPC
+ip-26-0-166-125:762815:762983 [0] NCCL INFO Channel 15/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-166-244:897240:897406 [0] NCCL INFO Channel 06/0 : 40[0] -> 47[7] via P2P/IPC
+ip-26-0-166-244:897242:897400 [2] NCCL INFO Channel 06/0 : 42[2] -> 43[3] via P2P/IPC
+ip-26-0-166-125:762816:762987 [1] NCCL INFO Channel 01/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-166-36:104433:104590 [4] NCCL INFO Channel 03/0 : 52[4] -> 53[5] via P2P/IPC
+ip-26-0-166-214:624943:625108 [0] NCCL INFO Connected all rings
+ip-26-0-166-214:624943:625108 [0] NCCL INFO Channel 02/0 : 32[0] -> 33[1] via P2P/IPC
+ip-26-0-166-36:104429:104594 [0] NCCL INFO Channel 14/0 : 48[0] -> 55[7] via P2P/IPC
+ip-26-0-166-214:624944:625106 [1] NCCL INFO Channel 15/0 : 33[1] -> 34[2] via P2P/IPC
+ip-26-0-162-46:702882:703046 [2] NCCL INFO Channel 14/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-36:104432:104592 [3] NCCL INFO Channel 05/0 : 51[3] -> 52[4] via P2P/IPC
+ip-26-0-162-46:702886:703049 [6] NCCL INFO Channel 02/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-244:897243:897403 [3] NCCL INFO Channel 02/0 : 43[3] -> 44[4] via P2P/IPC
+ip-26-0-166-244:897240:897406 [0] NCCL INFO Channel 07/0 : 40[0] -> 47[7] via P2P/IPC
+ip-26-0-166-214:624946:625107 [3] NCCL INFO Connected all rings
+ip-26-0-162-46:702883:703044 [3] NCCL INFO Connected all rings
+ip-26-0-166-36:104433:104590 [4] NCCL INFO Channel 04/0 : 52[4] -> 53[5] via P2P/IPC
+ip-26-0-166-244:897242:897400 [2] NCCL INFO Channel 07/0 : 42[2] -> 43[3] via P2P/IPC
+ip-26-0-166-36:104429:104594 [0] NCCL INFO Channel 15/0 : 48[0] -> 55[7] via P2P/IPC
+ip-26-0-166-36:104432:104592 [3] NCCL INFO Channel 06/0 : 51[3] -> 52[4] via P2P/IPC
+ip-26-0-162-46:702882:703046 [2] NCCL INFO Channel 15/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-214:624943:625108 [0] NCCL INFO Channel 03/0 : 32[0] -> 33[1] via P2P/IPC
+ip-26-0-166-15:127297:127453 [7] NCCL INFO Connected all rings
+ip-26-0-162-46:702886:703049 [6] NCCL INFO Channel 03/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-244:897243:897403 [3] NCCL INFO Channel 03/0 : 43[3] -> 44[4] via P2P/IPC
+ip-26-0-166-214:624949:625110 [6] NCCL INFO Connected all rings
+ip-26-0-162-46:702885:703048 [5] NCCL INFO Connected all rings
+ip-26-0-166-244:897240:897406 [0] NCCL INFO Channel 13/0 : 40[0] -> 47[7] via P2P/IPC
+ip-26-0-166-244:897242:897400 [2] NCCL INFO Channel 09/0 : 42[2] -> 43[3] via P2P/IPC
+ip-26-0-166-15:127296:127451 [6] NCCL INFO Channel 02/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-166-125:762817:762986 [2] NCCL INFO Connected all rings
+ip-26-0-166-15:127292:127458 [2] NCCL INFO Channel 14/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-162-46:702886:703049 [6] NCCL INFO Channel 04/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-214:624943:625108 [0] NCCL INFO Channel 04/0 : 32[0] -> 33[1] via P2P/IPC
+ip-26-0-166-244:897243:897403 [3] NCCL INFO Channel 05/0 : 43[3] -> 44[4] via P2P/IPC
+ip-26-0-166-214:624945:625111 [2] NCCL INFO Channel 01/0 : 34[2] -> 35[3] via P2P/IPC
+ip-26-0-166-244:897240:897406 [0] NCCL INFO Channel 14/0 : 40[0] -> 47[7] via P2P/IPC
+ip-26-0-166-244:897242:897400 [2] NCCL INFO Channel 10/0 : 42[2] -> 43[3] via P2P/IPC
+ip-26-0-166-125:762816:762987 [1] NCCL INFO Channel 03/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-166-15:127296:127451 [6] NCCL INFO Channel 03/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-166-214:624947:625105 [4] NCCL INFO Connected all rings
+ip-26-0-166-15:127292:127458 [2] NCCL INFO Channel 15/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-166-36:104433:104590 [4] NCCL INFO Channel 06/0 : 52[4] -> 53[5] via P2P/IPC
+ip-26-0-162-46:702886:703049 [6] NCCL INFO Channel 05/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-36:104432:104592 [3] NCCL INFO Channel 07/0 : 51[3] -> 52[4] via P2P/IPC
+ip-26-0-166-244:897243:897403 [3] NCCL INFO Channel 06/0 : 43[3] -> 44[4] via P2P/IPC
+ip-26-0-166-214:624943:625108 [0] NCCL INFO Channel 05/0 : 32[0] -> 33[1] via P2P/IPC
+ip-26-0-166-214:624945:625111 [2] NCCL INFO Channel 02/0 : 34[2] -> 35[3] via P2P/IPC
+ip-26-0-167-9:1039538:1039704 [1] NCCL INFO Channel 04/0 : 57[1] -> 58[2] via P2P/IPC
+ip-26-0-166-244:897240:897406 [0] NCCL INFO Channel 15/0 : 40[0] -> 47[7] via P2P/IPC
+ip-26-0-166-244:897242:897400 [2] NCCL INFO Channel 12/0 : 42[2] -> 43[3] via P2P/IPC
+ip-26-0-166-125:762816:762987 [1] NCCL INFO Channel 04/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-166-214:624950:625109 [7] NCCL INFO Connected all rings
+ip-26-0-166-36:104433:104590 [4] NCCL INFO Channel 07/0 : 52[4] -> 53[5] via P2P/IPC
+ip-26-0-166-36:104432:104592 [3] NCCL INFO Channel 09/0 : 51[3] -> 52[4] via P2P/IPC
+ip-26-0-166-125:762816:762987 [1] NCCL INFO Channel 05/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-166-15:127296:127451 [6] NCCL INFO Channel 04/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-166-244:897243:897403 [3] NCCL INFO Channel 07/0 : 43[3] -> 44[4] via P2P/IPC
+ip-26-0-166-36:104433:104590 [4] NCCL INFO Channel 09/0 : 52[4] -> 53[5] via P2P/IPC
+ip-26-0-166-244:897242:897400 [2] NCCL INFO Channel 13/0 : 42[2] -> 43[3] via P2P/IPC
+ip-26-0-166-36:104432:104592 [3] NCCL INFO Channel 10/0 : 51[3] -> 52[4] via P2P/IPC
+ip-26-0-166-214:624945:625111 [2] NCCL INFO Channel 04/0 : 34[2] -> 35[3] via P2P/IPC
+ip-26-0-166-214:624943:625108 [0] NCCL INFO Channel 06/0 : 32[0] -> 33[1] via P2P/IPC
+ip-26-0-166-244:897243:897403 [3] NCCL INFO Channel 09/0 : 43[3] -> 44[4] via P2P/IPC
+ip-26-0-166-36:104433:104590 [4] NCCL INFO Channel 10/0 : 52[4] -> 53[5] via P2P/IPC
+ip-26-0-166-36:104432:104592 [3] NCCL INFO Channel 11/0 : 51[3] -> 52[4] via P2P/IPC
+ip-26-0-166-244:897242:897400 [2] NCCL INFO Channel 14/0 : 42[2] -> 43[3] via P2P/IPC
+ip-26-0-166-125:762816:762987 [1] NCCL INFO Channel 06/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-167-9:1039539:1039698 [2] NCCL INFO Connected all rings
+ip-26-0-166-125:762816:762987 [1] NCCL INFO Channel 07/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-166-36:104433:104590 [4] NCCL INFO Channel 11/0 : 52[4] -> 53[5] via P2P/IPC
+ip-26-0-162-46:702886:703049 [6] NCCL INFO Channel 06/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-214:624945:625111 [2] NCCL INFO Channel 05/0 : 34[2] -> 35[3] via P2P/IPC
+ip-26-0-166-36:104432:104592 [3] NCCL INFO Channel 13/0 : 51[3] -> 52[4] via P2P/IPC
+ip-26-0-166-214:624943:625108 [0] NCCL INFO Channel 07/0 : 32[0] -> 33[1] via P2P/IPC
+ip-26-0-166-244:897243:897403 [3] NCCL INFO Channel 10/0 : 43[3] -> 44[4] via P2P/IPC
+ip-26-0-166-15:127294:127457 [4] NCCL INFO Connected all rings
+ip-26-0-167-9:1039538:1039704 [1] NCCL INFO Channel 05/0 : 57[1] -> 58[2] via P2P/IPC
+ip-26-0-166-244:897242:897400 [2] NCCL INFO Channel 15/0 : 42[2] -> 43[3] via P2P/IPC
+ip-26-0-166-15:127296:127451 [6] NCCL INFO Channel 05/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-162-46:702886:703049 [6] NCCL INFO Channel 09/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-15:127293:127454 [3] NCCL INFO Channel 01/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-166-36:104432:104592 [3] NCCL INFO Channel 14/0 : 51[3] -> 52[4] via P2P/IPC
+ip-26-0-166-36:104433:104590 [4] NCCL INFO Channel 12/0 : 52[4] -> 53[5] via P2P/IPC
+ip-26-0-166-125:762816:762987 [1] NCCL INFO Channel 09/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-166-214:624945:625111 [2] NCCL INFO Channel 06/0 : 34[2] -> 35[3] via P2P/IPC
+ip-26-0-166-214:624943:625108 [0] NCCL INFO Channel 10/0 : 32[0] -> 33[1] via P2P/IPC
+ip-26-0-166-15:127296:127451 [6] NCCL INFO Channel 06/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-162-46:702886:703049 [6] NCCL INFO Channel 10/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-15:127293:127454 [3] NCCL INFO Channel 02/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-166-125:762816:762987 [1] NCCL INFO Channel 11/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-162-46:702886:703049 [6] NCCL INFO Channel 11/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-15:127296:127451 [6] NCCL INFO Channel 09/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-166-15:127293:127454 [3] NCCL INFO Channel 03/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-166-214:624948:625104 [5] NCCL INFO Channel 01/0 : 37[5] -> 38[6] via P2P/IPC
+ip-26-0-166-214:624945:625111 [2] NCCL INFO Channel 07/0 : 34[2] -> 35[3] via P2P/IPC
+ip-26-0-166-214:624943:625108 [0] NCCL INFO Channel 11/0 : 32[0] -> 33[1] via P2P/IPC
+ip-26-0-166-36:104434:104591 [5] NCCL INFO Channel 01/0 : 53[5] -> 54[6] via P2P/IPC
+ip-26-0-166-244:897244:897407 [4] NCCL INFO Connected all rings
+ip-26-0-166-15:127296:127451 [6] NCCL INFO Channel 10/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-162-46:702886:703049 [6] NCCL INFO Channel 12/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-15:127293:127454 [3] NCCL INFO Channel 05/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-166-36:104432:104592 [3] NCCL INFO Channel 15/0 : 51[3] -> 52[4] via P2P/IPC
+ip-26-0-166-36:104433:104590 [4] NCCL INFO Channel 14/0 : 52[4] -> 53[5] via P2P/IPC
+ip-26-0-166-36:104434:104591 [5] NCCL INFO Channel 02/0 : 53[5] -> 54[6] via P2P/IPC
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO Connected all rings
+ip-26-0-167-9:1039538:1039704 [1] NCCL INFO Channel 06/0 : 57[1] -> 58[2] via P2P/IPC
+ip-26-0-166-125:762816:762987 [1] NCCL INFO Channel 12/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-162-46:702884:703047 [4] NCCL INFO Channel 01/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-166-36:104433:104590 [4] NCCL INFO Channel 15/0 : 52[4] -> 53[5] via P2P/IPC
+ip-26-0-166-36:104434:104591 [5] NCCL INFO Channel 03/0 : 53[5] -> 54[6] via P2P/IPC
+ip-26-0-162-46:702886:703049 [6] NCCL INFO Channel 13/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-125:762816:762987 [1] NCCL INFO Channel 13/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-166-15:127296:127451 [6] NCCL INFO Channel 11/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-166-214:624948:625104 [5] NCCL INFO Channel 02/0 : 37[5] -> 38[6] via P2P/IPC
+ip-26-0-166-36:104431:104593 [2] NCCL INFO Channel 02/0 : 42[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127293:127454 [3] NCCL INFO Channel 06/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-166-214:624945:625111 [2] NCCL INFO Channel 09/0 : 34[2] -> 35[3] via P2P/IPC
+ip-26-0-166-214:624943:625108 [0] NCCL INFO Channel 12/0 : 32[0] -> 33[1] via P2P/IPC
+ip-26-0-166-36:104431:104593 [2] NCCL INFO Channel 10/0 : 42[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104434:104591 [5] NCCL INFO Channel 04/0 : 53[5] -> 54[6] via P2P/IPC
+ip-26-0-166-36:104431:104593 [2] NCCL INFO Channel 02/0 : 50[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-46:702884:703047 [4] NCCL INFO Channel 02/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-166-36:104432:104592 [3] NCCL INFO Channel 03/0 : 43[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104432:104592 [3] NCCL INFO Channel 11/0 : 43[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897243:897403 [3] NCCL INFO Channel 11/0 : 43[3] -> 44[4] via P2P/IPC
+ip-26-0-167-9:1039538:1039704 [1] NCCL INFO Channel 07/0 : 57[1] -> 58[2] via P2P/IPC
+ip-26-0-166-36:104434:104591 [5] NCCL INFO Channel 05/0 : 53[5] -> 54[6] via P2P/IPC
+ip-26-0-162-46:702886:703049 [6] NCCL INFO Channel 14/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-36:104432:104592 [3] NCCL INFO Channel 03/0 : 51[3] -> 59[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127296:127451 [6] NCCL INFO Channel 12/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-166-125:762816:762987 [1] NCCL INFO Channel 14/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-166-15:127293:127454 [3] NCCL INFO Channel 07/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-166-36:104434:104591 [5] NCCL INFO Channel 07/0 : 53[5] -> 54[6] via P2P/IPC
+ip-26-0-166-244:897241:897402 [1] NCCL INFO Channel 09/0 : 33[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897245:897401 [5] NCCL INFO Channel 01/0 : 45[5] -> 46[6] via P2P/IPC
+ip-26-0-166-214:624948:625104 [5] NCCL INFO Channel 03/0 : 37[5] -> 38[6] via P2P/IPC
+ip-26-0-166-214:624945:625111 [2] NCCL INFO Channel 10/0 : 34[2] -> 35[3] via P2P/IPC
+ip-26-0-166-244:897241:897402 [1] NCCL INFO Channel 01/0 : 41[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104434:104591 [5] NCCL INFO Channel 09/0 : 53[5] -> 54[6] via P2P/IPC
+ip-26-0-166-244:897241:897402 [1] NCCL INFO Channel 09/0 : 41[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-167-9:1039544:1039702 [7] NCCL INFO Connected all rings
+ip-26-0-166-214:624943:625108 [0] NCCL INFO Channel 13/0 : 32[0] -> 33[1] via P2P/IPC
+ip-26-0-162-46:702884:703047 [4] NCCL INFO Channel 03/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-166-244:897243:897403 [3] NCCL INFO Channel 13/0 : 43[3] -> 44[4] via P2P/IPC
+ip-26-0-167-9:1039538:1039704 [1] NCCL INFO Channel 09/0 : 57[1] -> 58[2] via P2P/IPC
+ip-26-0-166-15:127296:127451 [6] NCCL INFO Channel 13/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-166-15:127293:127454 [3] NCCL INFO Channel 09/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-166-125:762816:762987 [1] NCCL INFO Channel 15/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-166-36:104434:104591 [5] NCCL INFO Channel 10/0 : 53[5] -> 54[6] via P2P/IPC
+ip-26-0-166-244:897245:897401 [5] NCCL INFO Channel 02/0 : 45[5] -> 46[6] via P2P/IPC
+ip-26-0-166-36:104434:104591 [5] NCCL INFO Channel 11/0 : 53[5] -> 54[6] via P2P/IPC
+ip-26-0-166-125:762818:762985 [3] NCCL INFO Connected all rings
+ip-26-0-166-125:762815:762983 [0] NCCL INFO Channel 05/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-162-46:702884:703047 [4] NCCL INFO Channel 04/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-166-125:762820:762981 [5] NCCL INFO Connected all rings
+ip-26-0-166-15:127296:127451 [6] NCCL INFO Channel 14/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-166-36:104434:104591 [5] NCCL INFO Channel 12/0 : 53[5] -> 54[6] via P2P/IPC
+ip-26-0-166-214:624945:625111 [2] NCCL INFO Channel 12/0 : 34[2] -> 35[3] via P2P/IPC
+ip-26-0-166-214:624948:625104 [5] NCCL INFO Channel 04/0 : 37[5] -> 38[6] via P2P/IPC
+ip-26-0-166-214:624943:625108 [0] NCCL INFO Channel 14/0 : 32[0] -> 33[1] via P2P/IPC
+ip-26-0-166-125:762817:762986 [2] NCCL INFO Channel 01/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-166-15:127293:127454 [3] NCCL INFO Channel 10/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-166-36:104434:104591 [5] NCCL INFO Channel 13/0 : 53[5] -> 54[6] via P2P/IPC
+ip-26-0-162-46:702883:703044 [3] NCCL INFO Channel 01/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-36:104434:104591 [5] NCCL INFO Channel 15/0 : 53[5] -> 54[6] via P2P/IPC
+ip-26-0-162-46:702884:703047 [4] NCCL INFO Channel 06/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-166-125:762815:762983 [0] NCCL INFO Channel 06/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-162-46:702885:703048 [5] NCCL INFO Channel 01/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-166-125:762817:762986 [2] NCCL INFO Channel 02/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-166-244:897243:897403 [3] NCCL INFO Channel 14/0 : 43[3] -> 44[4] via P2P/IPC
+ip-26-0-166-36:104433:104590 [4] NCCL INFO Channel 04/0 : 44[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624948:625104 [5] NCCL INFO Channel 05/0 : 37[5] -> 38[6] via P2P/IPC
+ip-26-0-166-36:104433:104590 [4] NCCL INFO Channel 12/0 : 44[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624945:625111 [2] NCCL INFO Channel 13/0 : 34[2] -> 35[3] via P2P/IPC
+ip-26-0-166-214:624943:625108 [0] NCCL INFO Channel 15/0 : 32[0] -> 33[1] via P2P/IPC
+ip-26-0-162-46:702883:703044 [3] NCCL INFO Channel 02/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-36:104433:104590 [4] NCCL INFO Channel 04/0 : 52[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-46:702884:703047 [4] NCCL INFO Channel 07/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-167-9:1039543:1039699 [6] NCCL INFO Connected all rings
+ip-26-0-166-244:897245:897401 [5] NCCL INFO Channel 03/0 : 45[5] -> 46[6] via P2P/IPC
+ip-26-0-162-46:702885:703048 [5] NCCL INFO Channel 02/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-167-9:1039538:1039704 [1] NCCL INFO Channel 11/0 : 57[1] -> 58[2] via P2P/IPC
+ip-26-0-166-244:897243:897403 [3] NCCL INFO Channel 15/0 : 43[3] -> 44[4] via P2P/IPC
+ip-26-0-166-125:762815:762983 [0] NCCL INFO Channel 07/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-166-244:897245:897401 [5] NCCL INFO Channel 04/0 : 45[5] -> 46[6] via P2P/IPC
+ip-26-0-162-46:702883:703044 [3] NCCL INFO Channel 03/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-125:762817:762986 [2] NCCL INFO Channel 04/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-162-46:702884:703047 [4] NCCL INFO Channel 09/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-166-214:624948:625104 [5] NCCL INFO Channel 07/0 : 37[5] -> 38[6] via P2P/IPC
+ip-26-0-165-213:9630:9789 [5] NCCL INFO Connected all rings
+ip-26-0-166-214:624945:625111 [2] NCCL INFO Channel 14/0 : 34[2] -> 35[3] via P2P/IPC
+ip-26-0-166-15:127295:127452 [5] NCCL INFO Connected all rings
+ip-26-0-162-46:702885:703048 [5] NCCL INFO Channel 03/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-166-15:127293:127454 [3] NCCL INFO Channel 11/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-166-214:624943:625108 [0] NCCL INFO Channel 05/0 : 32[0] -> 39[7] via P2P/IPC
+ip-26-0-166-244:897245:897401 [5] NCCL INFO Channel 05/0 : 45[5] -> 46[6] via P2P/IPC
+ip-26-0-166-244:897242:897400 [2] NCCL INFO Channel 10/0 : 34[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-46:702883:703044 [3] NCCL INFO Channel 05/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-244:897242:897400 [2] NCCL INFO Channel 02/0 : 42[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762815:762983 [0] NCCL INFO Channel 13/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-166-244:897242:897400 [2] NCCL INFO Channel 10/0 : 42[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-167-9:1039538:1039704 [1] NCCL INFO Channel 12/0 : 57[1] -> 58[2] via P2P/IPC
+ip-26-0-166-125:762817:762986 [2] NCCL INFO Channel 05/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-162-46:702885:703048 [5] NCCL INFO Channel 04/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-166-244:897246:897405 [6] NCCL INFO Channel 01/0 : 46[6] -> 47[7] via P2P/IPC
+ip-26-0-166-214:624946:625107 [3] NCCL INFO Channel 01/0 : 35[3] -> 36[4] via P2P/IPC
+ip-26-0-166-244:897245:897401 [5] NCCL INFO Channel 07/0 : 45[5] -> 46[6] via P2P/IPC
+ip-26-0-162-46:702883:703044 [3] NCCL INFO Channel 06/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-15:127293:127454 [3] NCCL INFO Channel 13/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-162-46:702884:703047 [4] NCCL INFO Channel 10/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-166-214:624948:625104 [5] NCCL INFO Channel 09/0 : 37[5] -> 38[6] via P2P/IPC
+ip-26-0-166-214:624945:625111 [2] NCCL INFO Channel 15/0 : 34[2] -> 35[3] via P2P/IPC
+ip-26-0-162-46:702885:703048 [5] NCCL INFO Channel 05/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-166-244:897246:897405 [6] NCCL INFO Channel 02/0 : 46[6] -> 47[7] via P2P/IPC
+ip-26-0-166-214:624943:625108 [0] NCCL INFO Channel 06/0 : 32[0] -> 39[7] via P2P/IPC
+ip-26-0-166-214:624949:625110 [6] NCCL INFO Channel 01/0 : 38[6] -> 39[7] via P2P/IPC
+ip-26-0-166-244:897245:897401 [5] NCCL INFO Channel 09/0 : 45[5] -> 46[6] via P2P/IPC
+ip-26-0-162-46:702883:703044 [3] NCCL INFO Channel 07/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-125:762815:762983 [0] NCCL INFO Channel 14/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-162-46:702884:703047 [4] NCCL INFO Channel 11/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-166-15:127293:127454 [3] NCCL INFO Channel 14/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-166-125:762817:762986 [2] NCCL INFO Channel 06/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-166-244:897246:897405 [6] NCCL INFO Channel 03/0 : 46[6] -> 47[7] via P2P/IPC
+ip-26-0-166-214:624946:625107 [3] NCCL INFO Channel 02/0 : 35[3] -> 36[4] via P2P/IPC
+ip-26-0-162-46:702885:703048 [5] NCCL INFO Channel 07/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-166-244:897245:897401 [5] NCCL INFO Channel 10/0 : 45[5] -> 46[6] via P2P/IPC
+ip-26-0-166-214:624948:625104 [5] NCCL INFO Channel 10/0 : 37[5] -> 38[6] via P2P/IPC
+ip-26-0-162-46:702883:703044 [3] NCCL INFO Channel 09/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-125:762815:762983 [0] NCCL INFO Channel 15/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-162-46:702884:703047 [4] NCCL INFO Channel 12/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-166-15:127293:127454 [3] NCCL INFO Channel 15/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-166-125:762821:762988 [6] NCCL INFO Connected all rings
+ip-26-0-166-214:624943:625108 [0] NCCL INFO Channel 07/0 : 32[0] -> 39[7] via P2P/IPC
+ip-26-0-166-244:897246:897405 [6] NCCL INFO Channel 04/0 : 46[6] -> 47[7] via P2P/IPC
+ip-26-0-166-214:624949:625110 [6] NCCL INFO Channel 02/0 : 38[6] -> 39[7] via P2P/IPC
+ip-26-0-166-125:762817:762986 [2] NCCL INFO Channel 07/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-166-214:624944:625106 [1] NCCL INFO Channel 09/0 : 33[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-46:702885:703048 [5] NCCL INFO Channel 09/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-167-9:1039538:1039704 [1] NCCL INFO Channel 13/0 : 57[1] -> 58[2] via P2P/IPC
+ip-26-0-165-213:9628:9786 [3] NCCL INFO Channel 01/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-166-244:897245:897401 [5] NCCL INFO Channel 11/0 : 45[5] -> 46[6] via P2P/IPC
+ip-26-0-166-214:624944:625106 [1] NCCL INFO Channel 01/0 : 17[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897241:897402 [1] NCCL INFO Channel 09/0 : 25[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-46:702883:703044 [3] NCCL INFO Channel 10/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-162-46:702884:703047 [4] NCCL INFO Channel 14/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-166-214:624946:625107 [3] NCCL INFO Channel 03/0 : 35[3] -> 36[4] via P2P/IPC
+ip-26-0-166-244:897246:897405 [6] NCCL INFO Channel 05/0 : 46[6] -> 47[7] via P2P/IPC
+ip-26-0-166-214:624944:625106 [1] NCCL INFO Channel 01/0 : 33[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-46:702885:703048 [5] NCCL INFO Channel 10/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-166-214:624948:625104 [5] NCCL INFO Channel 11/0 : 37[5] -> 38[6] via P2P/IPC
+ip-26-0-166-214:624947:625105 [4] NCCL INFO Channel 01/0 : 36[4] -> 37[5] via P2P/IPC
+ip-26-0-166-244:897245:897401 [5] NCCL INFO Channel 12/0 : 45[5] -> 46[6] via P2P/IPC
+ip-26-0-166-214:624943:625108 [0] NCCL INFO Channel 13/0 : 32[0] -> 39[7] via P2P/IPC
+ip-26-0-166-214:624949:625110 [6] NCCL INFO Channel 03/0 : 38[6] -> 39[7] via P2P/IPC
+ip-26-0-165-213:9628:9786 [3] NCCL INFO Channel 02/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-162-46:702883:703044 [3] NCCL INFO Channel 11/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-244:897246:897405 [6] NCCL INFO Channel 06/0 : 46[6] -> 47[7] via P2P/IPC
+ip-26-0-162-46:702884:703047 [4] NCCL INFO Channel 15/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-166-244:897245:897401 [5] NCCL INFO Channel 13/0 : 45[5] -> 46[6] via P2P/IPC
+ip-26-0-167-9:1039538:1039704 [1] NCCL INFO Channel 14/0 : 57[1] -> 58[2] via P2P/IPC
+ip-26-0-162-46:702885:703048 [5] NCCL INFO Channel 11/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-166-125:762817:762986 [2] NCCL INFO Channel 09/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-166-214:624946:625107 [3] NCCL INFO Channel 05/0 : 35[3] -> 36[4] via P2P/IPC
+ip-26-0-166-244:897246:897405 [6] NCCL INFO Channel 09/0 : 46[6] -> 47[7] via P2P/IPC
+ip-26-0-166-214:624948:625104 [5] NCCL INFO Channel 12/0 : 37[5] -> 38[6] via P2P/IPC
+ip-26-0-166-244:897245:897401 [5] NCCL INFO Channel 15/0 : 45[5] -> 46[6] via P2P/IPC
+ip-26-0-166-214:624947:625105 [4] NCCL INFO Channel 02/0 : 36[4] -> 37[5] via P2P/IPC
+ip-26-0-165-213:9628:9786 [3] NCCL INFO Channel 03/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-166-214:624943:625108 [0] NCCL INFO Channel 14/0 : 32[0] -> 39[7] via P2P/IPC
+ip-26-0-162-46:702883:703044 [3] NCCL INFO Channel 13/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-214:624949:625110 [6] NCCL INFO Channel 04/0 : 38[6] -> 39[7] via P2P/IPC
+ip-26-0-165-213:9631:9788 [6] NCCL INFO Channel 01/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-166-244:897246:897405 [6] NCCL INFO Channel 10/0 : 46[6] -> 47[7] via P2P/IPC
+ip-26-0-162-46:702885:703048 [5] NCCL INFO Channel 12/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-166-214:624946:625107 [3] NCCL INFO Channel 06/0 : 35[3] -> 36[4] via P2P/IPC
+ip-26-0-167-9:1039538:1039704 [1] NCCL INFO Channel 15/0 : 57[1] -> 58[2] via P2P/IPC
+ip-26-0-162-46:702883:703044 [3] NCCL INFO Channel 14/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-244:897244:897407 [4] NCCL INFO Channel 01/0 : 44[4] -> 45[5] via P2P/IPC
+ip-26-0-166-214:624948:625104 [5] NCCL INFO Channel 13/0 : 37[5] -> 38[6] via P2P/IPC
+ip-26-0-166-214:624947:625105 [4] NCCL INFO Channel 03/0 : 36[4] -> 37[5] via P2P/IPC
+ip-26-0-166-244:897246:897405 [6] NCCL INFO Channel 11/0 : 46[6] -> 47[7] via P2P/IPC
+ip-26-0-166-214:624943:625108 [0] NCCL INFO Channel 15/0 : 32[0] -> 39[7] via P2P/IPC
+ip-26-0-162-46:702885:703048 [5] NCCL INFO Channel 13/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-165-213:9628:9786 [3] NCCL INFO Channel 05/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-166-214:624949:625110 [6] NCCL INFO Channel 05/0 : 38[6] -> 39[7] via P2P/IPC
+ip-26-0-166-244:897244:897407 [4] NCCL INFO Channel 02/0 : 44[4] -> 45[5] via P2P/IPC
+ip-26-0-162-46:702883:703044 [3] NCCL INFO Channel 15/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-244:897246:897405 [6] NCCL INFO Channel 12/0 : 46[6] -> 47[7] via P2P/IPC
+ip-26-0-165-213:9631:9788 [6] NCCL INFO Channel 02/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-162-46:702885:703048 [5] NCCL INFO Channel 15/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-166-125:762819:762982 [4] NCCL INFO Connected all rings
+ip-26-0-166-244:897244:897407 [4] NCCL INFO Channel 03/0 : 44[4] -> 45[5] via P2P/IPC
+ip-26-0-166-214:624946:625107 [3] NCCL INFO Channel 07/0 : 35[3] -> 36[4] via P2P/IPC
+ip-26-0-166-244:897246:897405 [6] NCCL INFO Channel 13/0 : 46[6] -> 47[7] via P2P/IPC
+ip-26-0-166-214:624948:625104 [5] NCCL INFO Channel 15/0 : 37[5] -> 38[6] via P2P/IPC
+ip-26-0-166-244:897244:897407 [4] NCCL INFO Channel 04/0 : 44[4] -> 45[5] via P2P/IPC
+ip-26-0-166-214:624947:625105 [4] NCCL INFO Channel 04/0 : 36[4] -> 37[5] via P2P/IPC
+ip-26-0-165-213:9628:9786 [3] NCCL INFO Channel 06/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-166-244:897246:897405 [6] NCCL INFO Channel 14/0 : 46[6] -> 47[7] via P2P/IPC
+ip-26-0-166-214:624949:625110 [6] NCCL INFO Channel 06/0 : 38[6] -> 39[7] via P2P/IPC
+ip-26-0-167-9:1039537:1039703 [0] NCCL INFO Channel 05/0 : 56[0] -> 63[7] via P2P/IPC
+ip-26-0-166-244:897244:897407 [4] NCCL INFO Channel 06/0 : 44[4] -> 45[5] via P2P/IPC
+ip-26-0-162-46:702883:703044 [3] NCCL INFO Channel 11/0 : 3[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9631:9788 [6] NCCL INFO Channel 03/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-162-46:702884:703047 [4] NCCL INFO Channel 12/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624946:625107 [3] NCCL INFO Channel 09/0 : 35[3] -> 36[4] via P2P/IPC
+ip-26-0-166-36:104436:104596 [7] NCCL INFO Connected all rings
+ip-26-0-162-46:702886:703049 [6] NCCL INFO Channel 14/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762817:762986 [2] NCCL INFO Channel 10/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-162-46:702885:703048 [5] NCCL INFO Channel 13/0 : 5[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897244:897407 [4] NCCL INFO Channel 07/0 : 44[4] -> 45[5] via P2P/IPC
+ip-26-0-166-214:624947:625105 [4] NCCL INFO Channel 06/0 : 36[4] -> 37[5] via P2P/IPC
+ip-26-0-166-15:127291:127455 [1] NCCL INFO Connected all rings
+ip-26-0-166-214:624949:625110 [6] NCCL INFO Channel 09/0 : 38[6] -> 39[7] via P2P/IPC
+ip-26-0-166-244:897244:897407 [4] NCCL INFO Channel 09/0 : 44[4] -> 45[5] via P2P/IPC
+ip-26-0-166-214:624946:625107 [3] NCCL INFO Channel 10/0 : 35[3] -> 36[4] via P2P/IPC
+ip-26-0-165-213:9628:9786 [3] NCCL INFO Channel 07/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-166-244:897246:897405 [6] NCCL INFO Channel 14/0 : 38[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897246:897405 [6] NCCL INFO Channel 06/0 : 46[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-167-9:1039537:1039703 [0] NCCL INFO Channel 06/0 : 56[0] -> 63[7] via P2P/IPC
+ip-26-0-166-244:897246:897405 [6] NCCL INFO Channel 14/0 : 46[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9631:9788 [6] NCCL INFO Channel 04/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-166-214:624947:625105 [4] NCCL INFO Channel 07/0 : 36[4] -> 37[5] via P2P/IPC
+ip-26-0-166-214:624949:625110 [6] NCCL INFO Channel 10/0 : 38[6] -> 39[7] via P2P/IPC
+ip-26-0-166-244:897244:897407 [4] NCCL INFO Channel 10/0 : 44[4] -> 45[5] via P2P/IPC
+ip-26-0-166-214:624946:625107 [3] NCCL INFO Channel 11/0 : 35[3] -> 36[4] via P2P/IPC
+ip-26-0-166-125:762817:762986 [2] NCCL INFO Channel 12/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-165-213:9628:9786 [3] NCCL INFO Channel 09/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-166-214:624947:625105 [4] NCCL INFO Channel 09/0 : 36[4] -> 37[5] via P2P/IPC
+ip-26-0-166-214:624949:625110 [6] NCCL INFO Channel 11/0 : 38[6] -> 39[7] via P2P/IPC
+ip-26-0-166-244:897244:897407 [4] NCCL INFO Channel 11/0 : 44[4] -> 45[5] via P2P/IPC
+ip-26-0-167-9:1039540:1039705 [3] NCCL INFO Channel 01/0 : 59[3] -> 60[4] via P2P/IPC
+ip-26-0-165-213:9631:9788 [6] NCCL INFO Channel 05/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-166-214:624946:625107 [3] NCCL INFO Channel 13/0 : 35[3] -> 36[4] via P2P/IPC
+ip-26-0-167-9:1039537:1039703 [0] NCCL INFO Channel 07/0 : 56[0] -> 63[7] via P2P/IPC
+ip-26-0-166-214:624947:625105 [4] NCCL INFO Channel 10/0 : 36[4] -> 37[5] via P2P/IPC
+ip-26-0-166-244:897244:897407 [4] NCCL INFO Channel 12/0 : 44[4] -> 45[5] via P2P/IPC
+ip-26-0-166-214:624949:625110 [6] NCCL INFO Channel 12/0 : 38[6] -> 39[7] via P2P/IPC
+ip-26-0-165-213:9628:9786 [3] NCCL INFO Channel 10/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-167-9:1039539:1039698 [2] NCCL INFO Channel 01/0 : 58[2] -> 59[3] via P2P/IPC
+ip-26-0-166-214:624946:625107 [3] NCCL INFO Channel 14/0 : 35[3] -> 36[4] via P2P/IPC
+ip-26-0-166-125:762817:762986 [2] NCCL INFO Channel 13/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-165-213:9631:9788 [6] NCCL INFO Channel 06/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-165-213:9629:9793 [4] NCCL INFO Channel 01/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-166-244:897244:897407 [4] NCCL INFO Channel 14/0 : 44[4] -> 45[5] via P2P/IPC
+ip-26-0-162-46:702881:703043 [1] NCCL INFO Connected all rings
+ip-26-0-166-214:624947:625105 [4] NCCL INFO Channel 11/0 : 36[4] -> 37[5] via P2P/IPC
+ip-26-0-166-214:624949:625110 [6] NCCL INFO Channel 13/0 : 38[6] -> 39[7] via P2P/IPC
+ip-26-0-166-214:624946:625107 [3] NCCL INFO Channel 15/0 : 35[3] -> 36[4] via P2P/IPC
+ip-26-0-167-9:1039540:1039705 [3] NCCL INFO Channel 02/0 : 59[3] -> 60[4] via P2P/IPC
+ip-26-0-165-213:9628:9786 [3] NCCL INFO Channel 11/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-166-15:127294:127457 [4] NCCL INFO Channel 01/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-167-9:1039537:1039703 [0] NCCL INFO Channel 13/0 : 56[0] -> 63[7] via P2P/IPC
+ip-26-0-166-244:897244:897407 [4] NCCL INFO Channel 15/0 : 44[4] -> 45[5] via P2P/IPC
+ip-26-0-166-214:624947:625105 [4] NCCL INFO Channel 12/0 : 36[4] -> 37[5] via P2P/IPC
+ip-26-0-166-15:127295:127452 [5] NCCL INFO Channel 01/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-165-213:9631:9788 [6] NCCL INFO Channel 09/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-166-214:624949:625110 [6] NCCL INFO Channel 14/0 : 38[6] -> 39[7] via P2P/IPC
+ip-26-0-165-213:9629:9793 [4] NCCL INFO Channel 02/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-167-9:1039539:1039698 [2] NCCL INFO Channel 02/0 : 58[2] -> 59[3] via P2P/IPC
+ip-26-0-166-15:127294:127457 [4] NCCL INFO Channel 02/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-165-213:9628:9786 [3] NCCL INFO Channel 13/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-166-214:624947:625105 [4] NCCL INFO Channel 14/0 : 36[4] -> 37[5] via P2P/IPC
+ip-26-0-166-125:762817:762986 [2] NCCL INFO Channel 14/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-165-213:9630:9789 [5] NCCL INFO Channel 01/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-166-15:127295:127452 [5] NCCL INFO Channel 02/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-166-244:897240:897406 [0] NCCL INFO Channel 08/0 : 32[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897247:897404 [7] NCCL INFO Channel 15/0 : 39[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624945:625111 [2] NCCL INFO Channel 10/0 : 34[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897243:897403 [3] NCCL INFO Channel 11/0 : 35[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-167-9:1039540:1039705 [3] NCCL INFO Channel 03/0 : 59[3] -> 60[4] via P2P/IPC
+ip-26-0-166-15:127294:127457 [4] NCCL INFO Channel 03/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-165-213:9631:9788 [6] NCCL INFO Channel 10/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-166-214:624947:625105 [4] NCCL INFO Channel 15/0 : 36[4] -> 37[5] via P2P/IPC
+ip-26-0-166-244:897247:897404 [7] NCCL INFO Channel 07/0 : 47[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897240:897406 [0] NCCL INFO Channel 00/0 : 40[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897245:897401 [5] NCCL INFO Channel 13/0 : 37[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9629:9793 [4] NCCL INFO Channel 03/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-167-9:1039537:1039703 [0] NCCL INFO Channel 14/0 : 56[0] -> 63[7] via P2P/IPC
+ip-26-0-166-244:897243:897403 [3] NCCL INFO Channel 03/0 : 43[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897247:897404 [7] NCCL INFO Channel 15/0 : 47[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624945:625111 [2] NCCL INFO Channel 02/0 : 18[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897240:897406 [0] NCCL INFO Channel 08/0 : 40[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897244:897407 [4] NCCL INFO Channel 12/0 : 36[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897245:897401 [5] NCCL INFO Channel 05/0 : 45[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624945:625111 [2] NCCL INFO Channel 02/0 : 34[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897243:897403 [3] NCCL INFO Channel 11/0 : 43[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127295:127452 [5] NCCL INFO Channel 03/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-166-214:624949:625110 [6] NCCL INFO Channel 14/0 : 38[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9628:9786 [3] NCCL INFO Channel 14/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-166-244:897244:897407 [4] NCCL INFO Channel 04/0 : 44[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897245:897401 [5] NCCL INFO Channel 13/0 : 45[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-167-9:1039539:1039698 [2] NCCL INFO Channel 04/0 : 58[2] -> 59[3] via P2P/IPC
+ip-26-0-166-244:897242:897400 [2] NCCL INFO Channel 10/0 : 26[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9630:9789 [5] NCCL INFO Channel 02/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-166-15:127294:127457 [4] NCCL INFO Channel 04/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-166-244:897244:897407 [4] NCCL INFO Channel 12/0 : 44[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624949:625110 [6] NCCL INFO Channel 06/0 : 22[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9631:9788 [6] NCCL INFO Channel 11/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-166-214:624946:625107 [3] NCCL INFO Channel 11/0 : 35[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624949:625110 [6] NCCL INFO Channel 06/0 : 38[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9629:9793 [4] NCCL INFO Channel 04/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-166-125:762817:762986 [2] NCCL INFO Channel 15/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-166-214:624948:625104 [5] NCCL INFO Channel 13/0 : 37[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762818:762985 [3] NCCL INFO Channel 01/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-166-214:624947:625105 [4] NCCL INFO Channel 12/0 : 36[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-167-9:1039540:1039705 [3] NCCL INFO Channel 05/0 : 59[3] -> 60[4] via P2P/IPC
+ip-26-0-166-36:104435:104595 [6] NCCL INFO Channel 01/0 : 54[6] -> 55[7] via P2P/IPC
+ip-26-0-165-213:9628:9786 [3] NCCL INFO Channel 15/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-166-214:624946:625107 [3] NCCL INFO Channel 03/0 : 19[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9630:9789 [5] NCCL INFO Channel 03/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-166-244:897243:897403 [3] NCCL INFO Channel 11/0 : 27[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-167-9:1039539:1039698 [2] NCCL INFO Channel 05/0 : 58[2] -> 59[3] via P2P/IPC
+ip-26-0-166-214:624948:625104 [5] NCCL INFO Channel 05/0 : 21[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624946:625107 [3] NCCL INFO Channel 03/0 : 35[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624947:625105 [4] NCCL INFO Channel 04/0 : 20[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897244:897407 [4] NCCL INFO Channel 12/0 : 28[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624948:625104 [5] NCCL INFO Channel 05/0 : 37[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624947:625105 [4] NCCL INFO Channel 04/0 : 36[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9631:9788 [6] NCCL INFO Channel 12/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-166-36:104435:104595 [6] NCCL INFO Channel 02/0 : 54[6] -> 55[7] via P2P/IPC
+ip-26-0-166-125:762820:762981 [5] NCCL INFO Channel 01/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-165-213:9629:9793 [4] NCCL INFO Channel 06/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-166-36:104435:104595 [6] NCCL INFO Channel 03/0 : 54[6] -> 55[7] via P2P/IPC
+ip-26-0-167-9:1039540:1039705 [3] NCCL INFO Channel 06/0 : 59[3] -> 60[4] via P2P/IPC
+ip-26-0-165-213:9630:9789 [5] NCCL INFO Channel 04/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-167-9:1039539:1039698 [2] NCCL INFO Channel 06/0 : 58[2] -> 59[3] via P2P/IPC
+ip-26-0-166-125:762818:762985 [3] NCCL INFO Channel 02/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-165-213:9631:9788 [6] NCCL INFO Channel 13/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-166-36:104435:104595 [6] NCCL INFO Channel 04/0 : 54[6] -> 55[7] via P2P/IPC
+ip-26-0-165-213:9629:9793 [4] NCCL INFO Channel 07/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-166-15:127295:127452 [5] NCCL INFO Channel 04/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-165-213:9627:9791 [2] NCCL INFO Channel 10/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9630:9789 [5] NCCL INFO Channel 05/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-166-214:624950:625109 [7] NCCL INFO Channel 15/0 : 39[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762816:762987 [1] NCCL INFO Channel 01/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9627:9791 [2] NCCL INFO Channel 02/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624943:625108 [0] NCCL INFO Channel 08/0 : 32[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104435:104595 [6] NCCL INFO Channel 05/0 : 54[6] -> 55[7] via P2P/IPC
+ip-26-0-166-15:127294:127457 [4] NCCL INFO Channel 06/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-165-213:9627:9791 [2] NCCL INFO Channel 10/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762816:762987 [1] NCCL INFO Channel 09/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762820:762981 [5] NCCL INFO Channel 02/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-167-9:1039540:1039705 [3] NCCL INFO Channel 07/0 : 59[3] -> 60[4] via P2P/IPC
+ip-26-0-167-9:1039543:1039699 [6] NCCL INFO Channel 01/0 : 62[6] -> 63[7] via P2P/IPC
+ip-26-0-165-213:9631:9788 [6] NCCL INFO Channel 14/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-165-213:9629:9793 [4] NCCL INFO Channel 09/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-166-125:762816:762987 [1] NCCL INFO Channel 01/0 : 17[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-167-9:1039539:1039698 [2] NCCL INFO Channel 07/0 : 58[2] -> 59[3] via P2P/IPC
+ip-26-0-166-214:624943:625108 [0] NCCL INFO Channel 00/0 : 16[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624950:625109 [7] NCCL INFO Channel 07/0 : 23[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104435:104595 [6] NCCL INFO Channel 06/0 : 54[6] -> 55[7] via P2P/IPC
+ip-26-0-166-15:127295:127452 [5] NCCL INFO Channel 05/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-165-213:9630:9789 [5] NCCL INFO Channel 07/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-166-214:624950:625109 [7] NCCL INFO Channel 07/0 : 39[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762818:762985 [3] NCCL INFO Channel 03/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-166-214:624943:625108 [0] NCCL INFO Channel 00/0 : 32[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127294:127457 [4] NCCL INFO Channel 07/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-166-36:104435:104595 [6] NCCL INFO Channel 09/0 : 54[6] -> 55[7] via P2P/IPC
+ip-26-0-165-213:9629:9793 [4] NCCL INFO Channel 10/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-165-213:9630:9789 [5] NCCL INFO Channel 09/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-166-36:104435:104595 [6] NCCL INFO Channel 10/0 : 54[6] -> 55[7] via P2P/IPC
+ip-26-0-166-15:127295:127452 [5] NCCL INFO Channel 07/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-165-213:9629:9793 [4] NCCL INFO Channel 11/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-166-15:127294:127457 [4] NCCL INFO Channel 09/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-166-36:104435:104595 [6] NCCL INFO Channel 11/0 : 54[6] -> 55[7] via P2P/IPC
+ip-26-0-165-213:9630:9789 [5] NCCL INFO Channel 10/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-166-15:127295:127452 [5] NCCL INFO Channel 09/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-166-36:104435:104595 [6] NCCL INFO Channel 12/0 : 54[6] -> 55[7] via P2P/IPC
+ip-26-0-165-213:9629:9793 [4] NCCL INFO Channel 12/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-165-213:9630:9789 [5] NCCL INFO Channel 11/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-166-36:104435:104595 [6] NCCL INFO Channel 13/0 : 54[6] -> 55[7] via P2P/IPC
+ip-26-0-167-9:1039537:1039703 [0] NCCL INFO Channel 15/0 : 56[0] -> 63[7] via P2P/IPC
+ip-26-0-166-36:104435:104595 [6] NCCL INFO Channel 14/0 : 54[6] -> 55[7] via P2P/IPC
+ip-26-0-167-9:1039543:1039699 [6] NCCL INFO Channel 02/0 : 62[6] -> 63[7] via P2P/IPC
+ip-26-0-165-213:9629:9793 [4] NCCL INFO Channel 14/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-165-213:9630:9789 [5] NCCL INFO Channel 12/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-167-9:1039540:1039705 [3] NCCL INFO Channel 09/0 : 59[3] -> 60[4] via P2P/IPC
+ip-26-0-165-213:9629:9793 [4] NCCL INFO Channel 15/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-166-36:104434:104591 [5] NCCL INFO Channel 05/0 : 45[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9630:9789 [5] NCCL INFO Channel 13/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-167-9:1039543:1039699 [6] NCCL INFO Channel 03/0 : 62[6] -> 63[7] via P2P/IPC
+ip-26-0-166-36:104435:104595 [6] NCCL INFO Channel 06/0 : 46[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104434:104591 [5] NCCL INFO Channel 13/0 : 45[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-167-9:1039539:1039698 [2] NCCL INFO Channel 09/0 : 58[2] -> 59[3] via P2P/IPC
+ip-26-0-166-36:104435:104595 [6] NCCL INFO Channel 14/0 : 46[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127294:127457 [4] NCCL INFO Channel 10/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-166-125:762820:762981 [5] NCCL INFO Channel 03/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-166-36:104434:104591 [5] NCCL INFO Channel 05/0 : 53[5] -> 61[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104435:104595 [6] NCCL INFO Channel 06/0 : 54[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-167-9:1039540:1039705 [3] NCCL INFO Channel 10/0 : 59[3] -> 60[4] via P2P/IPC
+ip-26-0-166-125:762818:762985 [3] NCCL INFO Channel 05/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-165-213:9630:9789 [5] NCCL INFO Channel 15/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-166-15:127295:127452 [5] NCCL INFO Channel 10/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-166-244:897245:897401 [5] NCCL INFO Channel 13/0 : 29[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9628:9786 [3] NCCL INFO Channel 11/0 : 3[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897246:897405 [6] NCCL INFO Channel 14/0 : 30[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9628:9786 [3] NCCL INFO Channel 03/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127294:127457 [4] NCCL INFO Channel 11/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-167-9:1039543:1039699 [6] NCCL INFO Channel 04/0 : 62[6] -> 63[7] via P2P/IPC
+ip-26-0-165-213:9628:9786 [3] NCCL INFO Channel 11/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762820:762981 [5] NCCL INFO Channel 04/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-166-15:127295:127452 [5] NCCL INFO Channel 11/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-167-9:1039539:1039698 [2] NCCL INFO Channel 10/0 : 58[2] -> 59[3] via P2P/IPC
+ip-26-0-167-9:1039540:1039705 [3] NCCL INFO Channel 11/0 : 59[3] -> 60[4] via P2P/IPC
+ip-26-0-167-9:1039543:1039699 [6] NCCL INFO Channel 05/0 : 62[6] -> 63[7] via P2P/IPC
+ip-26-0-166-15:127291:127455 [1] NCCL INFO Channel 01/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-166-125:762818:762985 [3] NCCL INFO Channel 06/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-165-213:9629:9793 [4] NCCL INFO Channel 12/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-46:702883:703044 [3] NCCL INFO Channel 03/0 : 35[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127294:127457 [4] NCCL INFO Channel 12/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-165-213:9631:9788 [6] NCCL INFO Channel 14/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-46:702883:703044 [3] NCCL INFO Channel 03/0 : 3[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9629:9793 [4] NCCL INFO Channel 04/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127295:127452 [5] NCCL INFO Channel 12/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-165-213:9630:9789 [5] NCCL INFO Channel 13/0 : 5[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-167-9:1039539:1039698 [2] NCCL INFO Channel 12/0 : 58[2] -> 59[3] via P2P/IPC
+ip-26-0-165-213:9631:9788 [6] NCCL INFO Channel 06/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9629:9793 [4] NCCL INFO Channel 12/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9630:9789 [5] NCCL INFO Channel 05/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9632:9792 [7] NCCL INFO Channel 15/0 : 7[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762820:762981 [5] NCCL INFO Channel 05/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-167-9:1039540:1039705 [3] NCCL INFO Channel 13/0 : 59[3] -> 60[4] via P2P/IPC
+ip-26-0-165-213:9625:9787 [0] NCCL INFO Channel 08/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9631:9788 [6] NCCL INFO Channel 14/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9630:9789 [5] NCCL INFO Channel 13/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9632:9792 [7] NCCL INFO Channel 07/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127291:127455 [1] NCCL INFO Channel 03/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-166-15:127294:127457 [4] NCCL INFO Channel 14/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-165-213:9625:9787 [0] NCCL INFO Channel 00/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9632:9792 [7] NCCL INFO Channel 15/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9625:9787 [0] NCCL INFO Channel 08/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127295:127452 [5] NCCL INFO Channel 13/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-167-9:1039543:1039699 [6] NCCL INFO Channel 06/0 : 62[6] -> 63[7] via P2P/IPC
+ip-26-0-162-46:702884:703047 [4] NCCL INFO Channel 04/0 : 36[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-46:702886:703049 [6] NCCL INFO Channel 06/0 : 38[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762818:762985 [3] NCCL INFO Channel 07/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-162-46:702884:703047 [4] NCCL INFO Channel 04/0 : 4[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-46:702885:703048 [5] NCCL INFO Channel 05/0 : 37[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104436:104596 [7] NCCL INFO Channel 07/0 : 47[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-46:702886:703049 [6] NCCL INFO Channel 06/0 : 6[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-46:702885:703048 [5] NCCL INFO Channel 05/0 : 5[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-167-9:1039539:1039698 [2] NCCL INFO Channel 13/0 : 58[2] -> 59[3] via P2P/IPC
+ip-26-0-166-36:104436:104596 [7] NCCL INFO Channel 15/0 : 47[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762820:762981 [5] NCCL INFO Channel 07/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-166-15:127291:127455 [1] NCCL INFO Channel 04/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-167-9:1039540:1039705 [3] NCCL INFO Channel 14/0 : 59[3] -> 60[4] via P2P/IPC
+ip-26-0-166-36:104436:104596 [7] NCCL INFO Channel 07/0 : 55[7] -> 63[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127294:127457 [4] NCCL INFO Channel 15/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-166-15:127295:127452 [5] NCCL INFO Channel 15/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-166-15:127291:127455 [1] NCCL INFO Channel 05/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-166-36:104429:104594 [0] NCCL INFO Channel 00/0 : 40[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762818:762985 [3] NCCL INFO Channel 09/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-167-9:1039539:1039698 [2] NCCL INFO Channel 14/0 : 58[2] -> 59[3] via P2P/IPC
+ip-26-0-166-244:897247:897404 [7] NCCL INFO Channel 15/0 : 31[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762821:762988 [6] NCCL INFO Channel 01/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-166-36:104429:104594 [0] NCCL INFO Channel 08/0 : 40[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-167-9:1039540:1039705 [3] NCCL INFO Channel 15/0 : 59[3] -> 60[4] via P2P/IPC
+ip-26-0-167-9:1039543:1039699 [6] NCCL INFO Channel 09/0 : 62[6] -> 63[7] via P2P/IPC
+ip-26-0-166-125:762820:762981 [5] NCCL INFO Channel 09/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-166-36:104429:104594 [0] NCCL INFO Channel 00/0 : 48[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127291:127455 [1] NCCL INFO Channel 06/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-162-46:702881:703043 [1] NCCL INFO Channel 01/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-166-15:127293:127454 [3] NCCL INFO Channel 03/0 : 19[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-46:702881:703043 [1] NCCL INFO Channel 03/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-166-15:127294:127457 [4] NCCL INFO Channel 04/0 : 20[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127296:127451 [6] NCCL INFO Channel 06/0 : 22[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127295:127452 [5] NCCL INFO Channel 05/0 : 21[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762818:762985 [3] NCCL INFO Channel 10/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-167-9:1039539:1039698 [2] NCCL INFO Channel 15/0 : 58[2] -> 59[3] via P2P/IPC
+ip-26-0-166-125:762821:762988 [6] NCCL INFO Channel 02/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-167-9:1039543:1039699 [6] NCCL INFO Channel 10/0 : 62[6] -> 63[7] via P2P/IPC
+ip-26-0-162-46:702881:703043 [1] NCCL INFO Channel 04/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-166-125:762820:762981 [5] NCCL INFO Channel 10/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-162-46:702881:703043 [1] NCCL INFO Channel 05/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-167-9:1039543:1039699 [6] NCCL INFO Channel 11/0 : 62[6] -> 63[7] via P2P/IPC
+ip-26-0-162-46:702881:703043 [1] NCCL INFO Channel 06/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-167-9:1039538:1039704 [1] NCCL INFO Channel 01/0 : 49[1] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-167-9:1039543:1039699 [6] NCCL INFO Channel 12/0 : 62[6] -> 63[7] via P2P/IPC
+ip-26-0-167-9:1039539:1039698 [2] NCCL INFO Channel 02/0 : 50[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-46:702881:703043 [1] NCCL INFO Channel 07/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-167-9:1039538:1039704 [1] NCCL INFO Channel 09/0 : 25[1] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104430:104589 [1] NCCL INFO Channel 01/0 : 33[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127291:127455 [1] NCCL INFO Channel 07/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-167-9:1039543:1039699 [6] NCCL INFO Channel 13/0 : 62[6] -> 63[7] via P2P/IPC
+ip-26-0-167-9:1039539:1039698 [2] NCCL INFO Channel 10/0 : 26[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-167-9:1039538:1039704 [1] NCCL INFO Channel 09/0 : 57[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-46:702881:703043 [1] NCCL INFO Channel 09/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-166-36:104431:104593 [2] NCCL INFO Channel 02/0 : 34[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-167-9:1039539:1039698 [2] NCCL INFO Channel 10/0 : 58[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-167-9:1039543:1039699 [6] NCCL INFO Channel 14/0 : 62[6] -> 63[7] via P2P/IPC
+ip-26-0-166-15:127291:127455 [1] NCCL INFO Channel 09/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-166-36:104430:104589 [1] NCCL INFO Channel 01/0 : 49[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-46:702881:703043 [1] NCCL INFO Channel 11/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-166-36:104431:104593 [2] NCCL INFO Channel 02/0 : 50[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127291:127455 [1] NCCL INFO Channel 11/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-162-46:702881:703043 [1] NCCL INFO Channel 12/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-166-15:127291:127455 [1] NCCL INFO Channel 12/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-162-46:702881:703043 [1] NCCL INFO Channel 13/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-166-15:127291:127455 [1] NCCL INFO Channel 13/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-162-46:702881:703043 [1] NCCL INFO Channel 14/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-166-125:762819:762982 [4] NCCL INFO Channel 01/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-166-15:127291:127455 [1] NCCL INFO Channel 14/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-166-125:762821:762988 [6] NCCL INFO Channel 03/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-162-46:702881:703043 [1] NCCL INFO Channel 15/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-166-15:127291:127455 [1] NCCL INFO Channel 15/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-166-244:897240:897406 [0] NCCL INFO Channel 08/0 : 24[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762820:762981 [5] NCCL INFO Channel 11/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-166-125:762819:762982 [4] NCCL INFO Channel 02/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-162-46:702880:703042 [0] NCCL INFO Channel 05/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-166-125:762821:762988 [6] NCCL INFO Channel 04/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-166-15:127290:127456 [0] NCCL INFO Channel 05/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-162-46:702882:703046 [2] NCCL INFO Channel 10/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-46:702881:703043 [1] NCCL INFO Channel 09/0 : 1[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762818:762985 [3] NCCL INFO Channel 11/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-166-15:127291:127455 [1] NCCL INFO Channel 01/0 : 17[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127292:127458 [2] NCCL INFO Channel 02/0 : 18[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-46:702880:703042 [0] NCCL INFO Channel 06/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-166-125:762820:762981 [5] NCCL INFO Channel 12/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-167-9:1039542:1039701 [5] NCCL INFO Connected all rings
+ip-26-0-162-46:702881:703043 [1] NCCL INFO Channel 01/0 : 33[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-46:702882:703046 [2] NCCL INFO Channel 02/0 : 34[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762819:762982 [4] NCCL INFO Channel 03/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-166-15:127291:127455 [1] NCCL INFO Channel 09/0 : 9[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-46:702881:703043 [1] NCCL INFO Channel 01/0 : 1[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-46:702882:703046 [2] NCCL INFO Channel 02/0 : 2[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762816:762987 [1] NCCL INFO Channel 01/0 : 17[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127291:127455 [1] NCCL INFO Channel 09/0 : 25[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-46:702880:703042 [0] NCCL INFO Channel 07/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-166-125:762821:762988 [6] NCCL INFO Channel 05/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-166-125:762818:762985 [3] NCCL INFO Channel 13/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-166-125:762816:762987 [1] NCCL INFO Channel 01/0 : 33[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-46:702880:703042 [0] NCCL INFO Channel 13/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-166-125:762820:762981 [5] NCCL INFO Channel 13/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-166-214:624944:625106 [1] NCCL INFO Channel 01/0 : 1[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762819:762982 [4] NCCL INFO Channel 04/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-166-214:624944:625106 [1] NCCL INFO Channel 01/0 : 33[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-46:702880:703042 [0] NCCL INFO Channel 14/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-166-125:762821:762988 [6] NCCL INFO Channel 06/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-165-213:9626:9790 [1] NCCL INFO Channel 09/0 : 9[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762818:762985 [3] NCCL INFO Channel 14/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-162-46:702881:703043 [1] NCCL INFO Channel 09/0 : 9[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624944:625106 [1] NCCL INFO Channel 01/0 : 49[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-46:702880:703042 [0] NCCL INFO Channel 15/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-166-214:624944:625106 [1] NCCL INFO Channel 01/0 : 33[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9626:9790 [1] NCCL INFO Channel 09/0 : 25[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127291:127455 [1] NCCL INFO Channel 09/0 : 57[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762820:762981 [5] NCCL INFO Channel 15/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-166-125:762819:762982 [4] NCCL INFO Channel 06/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-166-15:127291:127455 [1] NCCL INFO Channel 09/0 : 25[1] -> 57[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127290:127456 [0] NCCL INFO Channel 06/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-166-36:104430:104589 [1] NCCL INFO Channel 01/0 : 57[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762821:762988 [6] NCCL INFO Channel 09/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-166-36:104430:104589 [1] NCCL INFO Channel 01/0 : 49[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624944:625106 [1] NCCL INFO Channel 09/0 : 41[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762816:762987 [1] NCCL INFO Channel 01/0 : 25[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104430:104589 [1] NCCL INFO Channel 09/0 : 49[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-46:702880:703042 [0] NCCL INFO Channel 08/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-46:702887:703045 [7] NCCL INFO Channel 15/0 : 7[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762818:762985 [3] NCCL INFO Channel 15/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-166-125:762816:762987 [1] NCCL INFO Channel 01/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762816:762987 [1] NCCL INFO Channel 09/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127290:127456 [0] NCCL INFO Channel 07/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-162-46:702880:703042 [0] NCCL INFO Channel 00/0 : 32[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-46:702887:703045 [7] NCCL INFO Channel 07/0 : 39[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127290:127456 [0] NCCL INFO Channel 13/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-162-46:702880:703042 [0] NCCL INFO Channel 00/0 : 0[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762819:762982 [4] NCCL INFO Channel 07/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-162-46:702887:703045 [7] NCCL INFO Channel 07/0 : 7[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762821:762988 [6] NCCL INFO Channel 10/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-166-15:127290:127456 [0] NCCL INFO Channel 14/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-166-15:127290:127456 [0] NCCL INFO Channel 15/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-166-125:762819:762982 [4] NCCL INFO Channel 09/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-166-125:762817:762986 [2] NCCL INFO Channel 02/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762821:762988 [6] NCCL INFO Channel 11/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-166-125:762817:762986 [2] NCCL INFO Channel 10/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762817:762986 [2] NCCL INFO Channel 02/0 : 18[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762819:762982 [4] NCCL INFO Channel 10/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-166-15:127291:127455 [1] NCCL INFO Channel 09/0 : 41[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-167-9:1039538:1039704 [1] NCCL INFO Channel 01/0 : 57[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127297:127453 [7] NCCL INFO Channel 07/0 : 23[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127290:127456 [0] NCCL INFO Channel 00/0 : 16[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127291:127455 [1] NCCL INFO Channel 09/0 : 25[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762821:762988 [6] NCCL INFO Channel 12/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-165-213:9627:9791 [2] NCCL INFO Channel 10/0 : 10[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762819:762982 [4] NCCL INFO Channel 11/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-166-125:762817:762986 [2] NCCL INFO Channel 02/0 : 18[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762821:762988 [6] NCCL INFO Channel 13/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-165-213:9626:9790 [1] NCCL INFO Channel 01/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762819:762982 [4] NCCL INFO Channel 12/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-165-213:9626:9790 [1] NCCL INFO Channel 09/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624945:625111 [2] NCCL INFO Channel 02/0 : 2[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762817:762986 [2] NCCL INFO Channel 02/0 : 34[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9626:9790 [1] NCCL INFO Channel 09/0 : 9[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762821:762988 [6] NCCL INFO Channel 14/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-166-214:624945:625111 [2] NCCL INFO Channel 02/0 : 34[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624945:625111 [2] NCCL INFO Channel 02/0 : 50[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-46:702882:703046 [2] NCCL INFO Channel 10/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624945:625111 [2] NCCL INFO Channel 02/0 : 34[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127292:127458 [2] NCCL INFO Channel 10/0 : 10[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127292:127458 [2] NCCL INFO Channel 10/0 : 26[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624945:625111 [2] NCCL INFO Channel 10/0 : 42[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762817:762986 [2] NCCL INFO Channel 02/0 : 26[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9627:9791 [2] NCCL INFO Channel 10/0 : 26[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762817:762986 [2] NCCL INFO Channel 02/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127292:127458 [2] NCCL INFO Channel 10/0 : 58[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762819:762982 [4] NCCL INFO Channel 14/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-166-244:897242:897400 [2] NCCL INFO Channel 10/0 : 42[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127292:127458 [2] NCCL INFO Channel 10/0 : 26[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762817:762986 [2] NCCL INFO Channel 10/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-167-9:1039544:1039702 [7] NCCL INFO Channel 07/0 : 55[7] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762819:762982 [4] NCCL INFO Channel 15/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-167-9:1039537:1039703 [0] NCCL INFO Channel 00/0 : 48[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127292:127458 [2] NCCL INFO Channel 10/0 : 42[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9626:9790 [1] NCCL INFO Channel 00/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-167-9:1039539:1039698 [2] NCCL INFO Channel 02/0 : 58[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127292:127458 [2] NCCL INFO Channel 10/0 : 26[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-46:702881:703043 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-167-9:1039544:1039702 [7] NCCL INFO Channel 15/0 : 31[7] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104436:104596 [7] NCCL INFO Channel 07/0 : 39[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-167-9:1039537:1039703 [0] NCCL INFO Channel 08/0 : 24[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9626:9790 [1] NCCL INFO Channel 08/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-166-125:762821:762988 [6] NCCL INFO Channel 06/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104429:104594 [0] NCCL INFO Channel 00/0 : 32[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-167-9:1039544:1039702 [7] NCCL INFO Channel 15/0 : 63[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-167-9:1039537:1039703 [0] NCCL INFO Channel 08/0 : 56[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762821:762988 [6] NCCL INFO Channel 14/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-46:702881:703043 [1] NCCL INFO Channel 08/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762821:762988 [6] NCCL INFO Channel 06/0 : 22[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104436:104596 [7] NCCL INFO Channel 07/0 : 55[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-167-9:1039538:1039704 [1] NCCL INFO Channel 00/0 : 57[1] -> 56[0] via P2P/IPC
+ip-26-0-166-125:762818:762985 [3] NCCL INFO Channel 03/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127292:127458 [2] NCCL INFO Channel 02/0 : 26[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9627:9791 [2] NCCL INFO Channel 02/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762820:762981 [5] NCCL INFO Channel 05/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9627:9791 [2] NCCL INFO Channel 10/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762818:762985 [3] NCCL INFO Channel 11/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762819:762982 [4] NCCL INFO Channel 04/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762820:762981 [5] NCCL INFO Channel 13/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9627:9791 [2] NCCL INFO Channel 10/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104429:104594 [0] NCCL INFO Channel 00/0 : 48[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762818:762985 [3] NCCL INFO Channel 03/0 : 19[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762820:762981 [5] NCCL INFO Channel 05/0 : 21[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762819:762982 [4] NCCL INFO Channel 12/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762819:762982 [4] NCCL INFO Channel 04/0 : 20[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9631:9788 [6] NCCL INFO Channel 14/0 : 14[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104431:104593 [2] NCCL INFO Channel 02/0 : 58[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104431:104593 [2] NCCL INFO Channel 02/0 : 50[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-167-9:1039538:1039704 [1] NCCL INFO Channel 08/0 : 57[1] -> 56[0] via P2P/IPC
+ip-26-0-166-36:104431:104593 [2] NCCL INFO Channel 10/0 : 50[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762818:762985 [3] NCCL INFO Channel 03/0 : 19[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9628:9786 [3] NCCL INFO Channel 11/0 : 11[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9630:9789 [5] NCCL INFO Channel 13/0 : 13[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762820:762981 [5] NCCL INFO Channel 05/0 : 21[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9629:9793 [4] NCCL INFO Channel 12/0 : 12[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127293:127454 [3] NCCL INFO Channel 11/0 : 11[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127295:127452 [5] NCCL INFO Channel 13/0 : 13[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762819:762982 [4] NCCL INFO Channel 04/0 : 20[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127293:127454 [3] NCCL INFO Channel 11/0 : 27[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127294:127457 [4] NCCL INFO Channel 12/0 : 12[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127295:127452 [5] NCCL INFO Channel 13/0 : 29[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762818:762985 [3] NCCL INFO Channel 03/0 : 35[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127294:127457 [4] NCCL INFO Channel 12/0 : 28[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762820:762981 [5] NCCL INFO Channel 05/0 : 37[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762819:762982 [4] NCCL INFO Channel 04/0 : 36[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9628:9786 [3] NCCL INFO Channel 11/0 : 27[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9630:9789 [5] NCCL INFO Channel 13/0 : 29[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127292:127458 [2] NCCL INFO Channel 00/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-165-213:9629:9793 [4] NCCL INFO Channel 12/0 : 28[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127293:127454 [3] NCCL INFO Channel 11/0 : 59[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127295:127452 [5] NCCL INFO Channel 13/0 : 61[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127293:127454 [3] NCCL INFO Channel 11/0 : 27[3] -> 59[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897245:897401 [5] NCCL INFO Channel 13/0 : 45[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897243:897403 [3] NCCL INFO Channel 11/0 : 43[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127295:127452 [5] NCCL INFO Channel 13/0 : 29[5] -> 61[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897244:897407 [4] NCCL INFO Channel 12/0 : 44[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127294:127457 [4] NCCL INFO Channel 12/0 : 60[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762822:762984 [7] NCCL INFO Channel 07/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO Channel 01/0 : 60[4] -> 61[5] via P2P/IPC
+ip-26-0-166-15:127292:127458 [2] NCCL INFO Channel 01/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-166-15:127294:127457 [4] NCCL INFO Channel 12/0 : 28[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762822:762984 [7] NCCL INFO Channel 15/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9627:9791 [2] NCCL INFO Channel 00/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-166-244:897241:897402 [1] NCCL INFO Channel 09/0 : 41[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762822:762984 [7] NCCL INFO Channel 07/0 : 23[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-46:702882:703046 [2] NCCL INFO Channel 00/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-125:762817:762986 [2] NCCL INFO Channel 00/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-166-15:127292:127458 [2] NCCL INFO Channel 08/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-165-213:9627:9791 [2] NCCL INFO Channel 01/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-166-244:897241:897402 [1] NCCL INFO Channel 01/0 : 49[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127291:127455 [1] NCCL INFO Channel 01/0 : 25[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO Channel 02/0 : 60[4] -> 61[5] via P2P/IPC
+ip-26-0-166-125:762821:762988 [6] NCCL INFO Channel 06/0 : 22[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-46:702882:703046 [2] NCCL INFO Channel 01/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-244:897241:897402 [1] NCCL INFO Channel 09/0 : 49[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9632:9792 [7] NCCL INFO Channel 15/0 : 15[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127296:127451 [6] NCCL INFO Channel 14/0 : 14[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127292:127458 [2] NCCL INFO Channel 09/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-166-125:762817:762986 [2] NCCL INFO Channel 01/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-166-125:762815:762983 [0] NCCL INFO Channel 00/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897241:897402 [1] NCCL INFO Channel 09/0 : 41[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9627:9791 [2] NCCL INFO Channel 08/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-166-15:127296:127451 [6] NCCL INFO Channel 14/0 : 30[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762822:762984 [7] NCCL INFO Channel 07/0 : 23[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762815:762983 [0] NCCL INFO Channel 08/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127297:127453 [7] NCCL INFO Channel 15/0 : 15[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762815:762983 [0] NCCL INFO Channel 00/0 : 16[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-46:702882:703046 [2] NCCL INFO Channel 08/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-125:762821:762988 [6] NCCL INFO Channel 06/0 : 38[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127297:127453 [7] NCCL INFO Channel 15/0 : 31[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO Channel 03/0 : 60[4] -> 61[5] via P2P/IPC
+ip-26-0-165-213:9627:9791 [2] NCCL INFO Channel 09/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-166-125:762817:762986 [2] NCCL INFO Channel 08/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-166-125:762822:762984 [7] NCCL INFO Channel 07/0 : 39[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-167-9:1039542:1039701 [5] NCCL INFO Channel 01/0 : 61[5] -> 62[6] via P2P/IPC
+ip-26-0-166-214:624950:625109 [7] NCCL INFO Channel 07/0 : 7[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-167-9:1039539:1039698 [2] NCCL INFO Channel 00/0 : 58[2] -> 57[1] via P2P/IPC
+ip-26-0-162-46:702882:703046 [2] NCCL INFO Channel 09/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-244:897242:897400 [2] NCCL INFO Channel 02/0 : 50[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624950:625109 [7] NCCL INFO Channel 07/0 : 39[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9631:9788 [6] NCCL INFO Channel 14/0 : 30[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127296:127451 [6] NCCL INFO Channel 14/0 : 62[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897246:897405 [6] NCCL INFO Channel 14/0 : 46[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9632:9792 [7] NCCL INFO Channel 15/0 : 31[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762817:762986 [2] NCCL INFO Channel 09/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-166-15:127296:127451 [6] NCCL INFO Channel 14/0 : 30[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO Channel 04/0 : 60[4] -> 61[5] via P2P/IPC
+ip-26-0-166-244:897242:897400 [2] NCCL INFO Channel 10/0 : 50[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127290:127456 [0] NCCL INFO Channel 08/0 : 8[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762815:762983 [0] NCCL INFO Channel 00/0 : 16[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897242:897400 [2] NCCL INFO Channel 10/0 : 42[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127297:127453 [7] NCCL INFO Channel 15/0 : 63[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624950:625109 [7] NCCL INFO Channel 07/0 : 55[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897247:897404 [7] NCCL INFO Channel 15/0 : 47[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127290:127456 [0] NCCL INFO Channel 08/0 : 24[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-46:702887:703045 [7] NCCL INFO Channel 15/0 : 15[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127297:127453 [7] NCCL INFO Channel 15/0 : 31[7] -> 63[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-167-9:1039542:1039701 [5] NCCL INFO Channel 02/0 : 61[5] -> 62[6] via P2P/IPC
+ip-26-0-166-214:624950:625109 [7] NCCL INFO Channel 07/0 : 39[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-167-9:1039539:1039698 [2] NCCL INFO Channel 01/0 : 58[2] -> 57[1] via P2P/IPC
+ip-26-0-166-125:762815:762983 [0] NCCL INFO Channel 00/0 : 32[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624943:625108 [0] NCCL INFO Channel 00/0 : 0[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO Channel 06/0 : 60[4] -> 61[5] via P2P/IPC
+ip-26-0-166-15:127297:127453 [7] NCCL INFO Channel 15/0 : 47[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-167-9:1039544:1039702 [7] NCCL INFO Channel 07/0 : 63[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127297:127453 [7] NCCL INFO Channel 15/0 : 31[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624943:625108 [0] NCCL INFO Channel 00/0 : 32[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127291:127455 [1] NCCL INFO Channel 00/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-166-36:104436:104596 [7] NCCL INFO Channel 07/0 : 63[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-167-9:1039542:1039701 [5] NCCL INFO Channel 03/0 : 61[5] -> 62[6] via P2P/IPC
+ip-26-0-166-125:762816:762987 [1] NCCL INFO Channel 00/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-166-36:104436:104596 [7] NCCL INFO Channel 07/0 : 55[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-167-9:1039539:1039698 [2] NCCL INFO Channel 08/0 : 58[2] -> 57[1] via P2P/IPC
+ip-26-0-166-214:624950:625109 [7] NCCL INFO Channel 15/0 : 47[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762822:762984 [7] NCCL INFO Channel 07/0 : 31[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104430:104589 [1] NCCL INFO Channel 00/0 : 49[1] -> 48[0] via P2P/IPC
+ip-26-0-166-36:104436:104596 [7] NCCL INFO Channel 15/0 : 55[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762822:762984 [7] NCCL INFO Channel 07/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO Channel 07/0 : 60[4] -> 61[5] via P2P/IPC
+ip-26-0-166-214:624943:625108 [0] NCCL INFO Channel 00/0 : 48[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127291:127455 [1] NCCL INFO Channel 08/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-162-46:702880:703042 [0] NCCL INFO Channel 08/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762822:762984 [7] NCCL INFO Channel 15/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762816:762987 [1] NCCL INFO Channel 08/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-166-214:624943:625108 [0] NCCL INFO Channel 00/0 : 32[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9632:9792 [7] NCCL INFO Channel 07/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127297:127453 [7] NCCL INFO Channel 07/0 : 31[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-167-9:1039544:1039702 [7] NCCL INFO Channel 01/0 : 63[7] -> 56[0] via P2P/IPC
+ip-26-0-166-36:104430:104589 [1] NCCL INFO Channel 08/0 : 49[1] -> 48[0] via P2P/IPC
+ip-26-0-165-213:9632:9792 [7] NCCL INFO Channel 15/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-167-9:1039542:1039701 [5] NCCL INFO Channel 04/0 : 61[5] -> 62[6] via P2P/IPC
+ip-26-0-165-213:9632:9792 [7] NCCL INFO Channel 15/0 : 15[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127297:127453 [7] NCCL INFO Channel 01/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-167-9:1039539:1039698 [2] NCCL INFO Channel 09/0 : 58[2] -> 57[1] via P2P/IPC
+ip-26-0-166-36:104429:104594 [0] NCCL INFO Channel 00/0 : 56[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO Channel 09/0 : 60[4] -> 61[5] via P2P/IPC
+ip-26-0-166-36:104429:104594 [0] NCCL INFO Channel 00/0 : 48[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762822:762984 [7] NCCL INFO Channel 01/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-166-214:624950:625109 [7] NCCL INFO Channel 01/0 : 39[7] -> 32[0] via P2P/IPC
+ip-26-0-166-36:104429:104594 [0] NCCL INFO Channel 08/0 : 48[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9625:9787 [0] NCCL INFO Channel 08/0 : 8[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624943:625108 [0] NCCL INFO Channel 08/0 : 40[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762815:762983 [0] NCCL INFO Channel 00/0 : 24[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127297:127453 [7] NCCL INFO Channel 02/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-162-46:702887:703045 [7] NCCL INFO Channel 01/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-213:9632:9792 [7] NCCL INFO Channel 01/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-166-125:762815:762983 [0] NCCL INFO Channel 00/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762815:762983 [0] NCCL INFO Channel 08/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624950:625109 [7] NCCL INFO Channel 02/0 : 39[7] -> 32[0] via P2P/IPC
+ip-26-0-167-9:1039544:1039702 [7] NCCL INFO Channel 02/0 : 63[7] -> 56[0] via P2P/IPC
+ip-26-0-167-9:1039542:1039701 [5] NCCL INFO Channel 05/0 : 61[5] -> 62[6] via P2P/IPC
+ip-26-0-166-125:762822:762984 [7] NCCL INFO Channel 02/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-165-213:9625:9787 [0] NCCL INFO Channel 08/0 : 24[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127297:127453 [7] NCCL INFO Channel 03/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-162-46:702887:703045 [7] NCCL INFO Channel 02/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127290:127456 [0] NCCL INFO Channel 08/0 : 56[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762822:762984 [7] NCCL INFO Channel 03/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-166-214:624950:625109 [7] NCCL INFO Channel 03/0 : 39[7] -> 32[0] via P2P/IPC
+ip-26-0-166-15:127290:127456 [0] NCCL INFO Channel 08/0 : 24[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9632:9792 [7] NCCL INFO Channel 02/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO Channel 10/0 : 60[4] -> 61[5] via P2P/IPC
+ip-26-0-166-15:127297:127453 [7] NCCL INFO Channel 04/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-162-46:702887:703045 [7] NCCL INFO Channel 03/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127290:127456 [0] NCCL INFO Channel 08/0 : 40[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762822:762984 [7] NCCL INFO Channel 04/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-166-214:624950:625109 [7] NCCL INFO Channel 04/0 : 39[7] -> 32[0] via P2P/IPC
+ip-26-0-167-9:1039537:1039703 [0] NCCL INFO Channel 00/0 : 56[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104436:104596 [7] NCCL INFO Channel 01/0 : 55[7] -> 48[0] via P2P/IPC
+ip-26-0-165-213:9632:9792 [7] NCCL INFO Channel 03/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-166-15:127290:127456 [0] NCCL INFO Channel 08/0 : 24[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-167-9:1039542:1039701 [5] NCCL INFO Channel 07/0 : 61[5] -> 62[6] via P2P/IPC
+ip-26-0-167-9:1039544:1039702 [7] NCCL INFO Channel 03/0 : 63[7] -> 56[0] via P2P/IPC
+ip-26-0-166-15:127297:127453 [7] NCCL INFO Channel 05/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO Channel 11/0 : 60[4] -> 61[5] via P2P/IPC
+ip-26-0-162-46:702887:703045 [7] NCCL INFO Channel 04/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762822:762984 [7] NCCL INFO Channel 05/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-166-214:624950:625109 [7] NCCL INFO Channel 05/0 : 39[7] -> 32[0] via P2P/IPC
+ip-26-0-166-36:104436:104596 [7] NCCL INFO Channel 02/0 : 55[7] -> 48[0] via P2P/IPC
+ip-26-0-165-213:9632:9792 [7] NCCL INFO Channel 04/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-166-15:127297:127453 [7] NCCL INFO Channel 06/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-162-46:702887:703045 [7] NCCL INFO Channel 05/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127290:127456 [0] NCCL INFO Channel 00/0 : 24[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897240:897406 [0] NCCL INFO Channel 08/0 : 40[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897247:897404 [7] NCCL INFO Channel 07/0 : 55[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897247:897404 [7] NCCL INFO Channel 15/0 : 55[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897247:897404 [7] NCCL INFO Channel 15/0 : 47[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897247:897404 [7] NCCL INFO Channel 01/0 : 47[7] -> 40[0] via P2P/IPC
+ip-26-0-166-244:897240:897406 [0] NCCL INFO Channel 00/0 : 48[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897240:897406 [0] NCCL INFO Channel 08/0 : 48[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762822:762984 [7] NCCL INFO Channel 06/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-166-214:624950:625109 [7] NCCL INFO Channel 06/0 : 39[7] -> 32[0] via P2P/IPC
+ip-26-0-167-9:1039542:1039701 [5] NCCL INFO Channel 09/0 : 61[5] -> 62[6] via P2P/IPC
+ip-26-0-166-244:897240:897406 [0] NCCL INFO Channel 08/0 : 40[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9625:9787 [0] NCCL INFO Channel 00/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127297:127453 [7] NCCL INFO Channel 07/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-167-9:1039544:1039702 [7] NCCL INFO Channel 04/0 : 63[7] -> 56[0] via P2P/IPC
+ip-26-0-165-213:9632:9792 [7] NCCL INFO Channel 05/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-162-46:702887:703045 [7] NCCL INFO Channel 06/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-213:9625:9787 [0] NCCL INFO Channel 08/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104436:104596 [7] NCCL INFO Channel 03/0 : 55[7] -> 48[0] via P2P/IPC
+ip-26-0-166-125:762822:762984 [7] NCCL INFO Channel 07/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-165-213:9625:9787 [0] NCCL INFO Channel 08/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897247:897404 [7] NCCL INFO Channel 02/0 : 47[7] -> 40[0] via P2P/IPC
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO Channel 12/0 : 60[4] -> 61[5] via P2P/IPC
+ip-26-0-166-15:127297:127453 [7] NCCL INFO Channel 09/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-165-213:9632:9792 [7] NCCL INFO Channel 06/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-162-46:702887:703045 [7] NCCL INFO Channel 07/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624950:625109 [7] NCCL INFO Channel 07/0 : 39[7] -> 32[0] via P2P/IPC
+ip-26-0-166-125:762822:762984 [7] NCCL INFO Channel 09/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-166-36:104436:104596 [7] NCCL INFO Channel 04/0 : 55[7] -> 48[0] via P2P/IPC
+ip-26-0-167-9:1039542:1039701 [5] NCCL INFO Channel 10/0 : 61[5] -> 62[6] via P2P/IPC
+ip-26-0-165-213:9632:9792 [7] NCCL INFO Channel 07/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-162-46:702887:703045 [7] NCCL INFO Channel 09/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-167-9:1039544:1039702 [7] NCCL INFO Channel 05/0 : 63[7] -> 56[0] via P2P/IPC
+ip-26-0-166-15:127297:127453 [7] NCCL INFO Channel 10/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-166-244:897247:897404 [7] NCCL INFO Channel 03/0 : 47[7] -> 40[0] via P2P/IPC
+ip-26-0-166-125:762822:762984 [7] NCCL INFO Channel 10/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-165-213:9632:9792 [7] NCCL INFO Channel 09/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-162-46:702887:703045 [7] NCCL INFO Channel 10/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO Channel 14/0 : 60[4] -> 61[5] via P2P/IPC
+ip-26-0-166-244:897241:897402 [1] NCCL INFO Channel 00/0 : 41[1] -> 40[0] via P2P/IPC
+ip-26-0-166-36:104436:104596 [7] NCCL INFO Channel 05/0 : 55[7] -> 48[0] via P2P/IPC
+ip-26-0-166-15:127297:127453 [7] NCCL INFO Channel 11/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-165-213:9632:9792 [7] NCCL INFO Channel 10/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-166-125:762822:762984 [7] NCCL INFO Channel 11/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-162-46:702887:703045 [7] NCCL INFO Channel 11/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-167-9:1039542:1039701 [5] NCCL INFO Channel 11/0 : 61[5] -> 62[6] via P2P/IPC
+ip-26-0-167-9:1039544:1039702 [7] NCCL INFO Channel 06/0 : 63[7] -> 56[0] via P2P/IPC
+ip-26-0-166-244:897247:897404 [7] NCCL INFO Channel 04/0 : 47[7] -> 40[0] via P2P/IPC
+ip-26-0-166-15:127297:127453 [7] NCCL INFO Channel 12/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-166-125:762822:762984 [7] NCCL INFO Channel 12/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-165-213:9632:9792 [7] NCCL INFO Channel 11/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO Channel 15/0 : 60[4] -> 61[5] via P2P/IPC
+ip-26-0-162-46:702887:703045 [7] NCCL INFO Channel 12/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-244:897241:897402 [1] NCCL INFO Channel 08/0 : 41[1] -> 40[0] via P2P/IPC
+ip-26-0-162-46:702887:703045 [7] NCCL INFO Channel 13/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762822:762984 [7] NCCL INFO Channel 13/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-165-213:9632:9792 [7] NCCL INFO Channel 12/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-166-15:127297:127453 [7] NCCL INFO Channel 13/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-167-9:1039544:1039702 [7] NCCL INFO Channel 07/0 : 63[7] -> 56[0] via P2P/IPC
+ip-26-0-167-9:1039542:1039701 [5] NCCL INFO Channel 12/0 : 61[5] -> 62[6] via P2P/IPC
+ip-26-0-162-46:702887:703045 [7] NCCL INFO Channel 14/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762822:762984 [7] NCCL INFO Channel 14/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-165-213:9632:9792 [7] NCCL INFO Channel 13/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-166-214:624950:625109 [7] NCCL INFO Channel 09/0 : 39[7] -> 32[0] via P2P/IPC
+ip-26-0-166-15:127297:127453 [7] NCCL INFO Channel 14/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-167-9:1039544:1039702 [7] NCCL INFO Channel 09/0 : 63[7] -> 56[0] via P2P/IPC
+ip-26-0-166-125:762822:762984 [7] NCCL INFO Channel 15/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-165-213:9632:9792 [7] NCCL INFO Channel 14/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-167-9:1039540:1039705 [3] NCCL INFO Channel 03/0 : 51[3] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-167-9:1039542:1039701 [5] NCCL INFO Channel 13/0 : 61[5] -> 62[6] via P2P/IPC
+ip-26-0-162-46:702887:703045 [7] NCCL INFO Channel 15/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127297:127453 [7] NCCL INFO Channel 15/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-166-36:104436:104596 [7] NCCL INFO Channel 06/0 : 55[7] -> 48[0] via P2P/IPC
+ip-26-0-167-9:1039540:1039705 [3] NCCL INFO Channel 11/0 : 27[3] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897247:897404 [7] NCCL INFO Channel 05/0 : 47[7] -> 40[0] via P2P/IPC
+ip-26-0-165-213:9632:9792 [7] NCCL INFO Channel 15/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-167-9:1039540:1039705 [3] NCCL INFO Channel 11/0 : 59[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624950:625109 [7] NCCL INFO Channel 10/0 : 39[7] -> 32[0] via P2P/IPC
+ip-26-0-167-9:1039544:1039702 [7] NCCL INFO Channel 10/0 : 63[7] -> 56[0] via P2P/IPC
+ip-26-0-167-9:1039542:1039701 [5] NCCL INFO Channel 15/0 : 61[5] -> 62[6] via P2P/IPC
+ip-26-0-166-36:104436:104596 [7] NCCL INFO Channel 07/0 : 55[7] -> 48[0] via P2P/IPC
+ip-26-0-167-9:1039540:1039705 [3] NCCL INFO Channel 03/0 : 59[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127293:127454 [3] NCCL INFO Channel 11/0 : 43[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104436:104596 [7] NCCL INFO Channel 09/0 : 55[7] -> 48[0] via P2P/IPC
+ip-26-0-166-15:127293:127454 [3] NCCL INFO Channel 11/0 : 27[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-167-9:1039544:1039702 [7] NCCL INFO Channel 11/0 : 63[7] -> 56[0] via P2P/IPC
+ip-26-0-166-214:624950:625109 [7] NCCL INFO Channel 11/0 : 39[7] -> 32[0] via P2P/IPC
+ip-26-0-166-36:104436:104596 [7] NCCL INFO Channel 10/0 : 55[7] -> 48[0] via P2P/IPC
+ip-26-0-166-214:624945:625111 [2] NCCL INFO Channel 00/0 : 34[2] -> 33[1] via P2P/IPC
+ip-26-0-167-9:1039544:1039702 [7] NCCL INFO Channel 12/0 : 63[7] -> 56[0] via P2P/IPC
+ip-26-0-166-244:897243:897403 [3] NCCL INFO Channel 03/0 : 51[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127293:127454 [3] NCCL INFO Channel 03/0 : 27[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO Channel 04/0 : 52[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897243:897403 [3] NCCL INFO Channel 11/0 : 51[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104436:104596 [7] NCCL INFO Channel 11/0 : 55[7] -> 48[0] via P2P/IPC
+ip-26-0-165-213:9628:9786 [3] NCCL INFO Channel 03/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-167-9:1039543:1039699 [6] NCCL INFO Channel 06/0 : 54[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897243:897403 [3] NCCL INFO Channel 11/0 : 43[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9628:9786 [3] NCCL INFO Channel 11/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-167-9:1039542:1039701 [5] NCCL INFO Channel 05/0 : 53[5] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9628:9786 [3] NCCL INFO Channel 11/0 : 11[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-167-9:1039544:1039702 [7] NCCL INFO Channel 13/0 : 63[7] -> 56[0] via P2P/IPC
+ip-26-0-166-244:897242:897400 [2] NCCL INFO Channel 00/0 : 42[2] -> 41[1] via P2P/IPC
+ip-26-0-166-214:624950:625109 [7] NCCL INFO Channel 12/0 : 39[7] -> 32[0] via P2P/IPC
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO Channel 12/0 : 28[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104433:104590 [4] NCCL INFO Channel 04/0 : 36[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-167-9:1039543:1039699 [6] NCCL INFO Channel 14/0 : 30[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO Channel 12/0 : 60[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624945:625111 [2] NCCL INFO Channel 01/0 : 34[2] -> 33[1] via P2P/IPC
+ip-26-0-166-36:104436:104596 [7] NCCL INFO Channel 12/0 : 55[7] -> 48[0] via P2P/IPC
+ip-26-0-167-9:1039542:1039701 [5] NCCL INFO Channel 13/0 : 29[5] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104431:104593 [2] NCCL INFO Channel 00/0 : 50[2] -> 49[1] via P2P/IPC
+ip-26-0-166-36:104435:104595 [6] NCCL INFO Channel 06/0 : 38[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897247:897404 [7] NCCL INFO Channel 06/0 : 47[7] -> 40[0] via P2P/IPC
+ip-26-0-167-9:1039543:1039699 [6] NCCL INFO Channel 14/0 : 62[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-167-9:1039542:1039701 [5] NCCL INFO Channel 13/0 : 61[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-167-9:1039544:1039702 [7] NCCL INFO Channel 14/0 : 63[7] -> 56[0] via P2P/IPC
+ip-26-0-166-36:104433:104590 [4] NCCL INFO Channel 04/0 : 52[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO Channel 04/0 : 60[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624947:625105 [4] NCCL INFO Channel 04/0 : 4[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127294:127457 [4] NCCL INFO Channel 12/0 : 44[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104432:104592 [3] NCCL INFO Channel 03/0 : 35[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897242:897400 [2] NCCL INFO Channel 01/0 : 42[2] -> 41[1] via P2P/IPC
+ip-26-0-166-214:624949:625110 [6] NCCL INFO Channel 06/0 : 6[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-167-9:1039542:1039701 [5] NCCL INFO Channel 05/0 : 61[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127294:127457 [4] NCCL INFO Channel 12/0 : 28[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104435:104595 [6] NCCL INFO Channel 06/0 : 54[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104434:104591 [5] NCCL INFO Channel 05/0 : 37[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624947:625105 [4] NCCL INFO Channel 04/0 : 36[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624949:625110 [6] NCCL INFO Channel 06/0 : 38[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127295:127452 [5] NCCL INFO Channel 13/0 : 45[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104436:104596 [7] NCCL INFO Channel 13/0 : 55[7] -> 48[0] via P2P/IPC
+ip-26-0-166-214:624950:625109 [7] NCCL INFO Channel 13/0 : 39[7] -> 32[0] via P2P/IPC
+ip-26-0-166-15:127295:127452 [5] NCCL INFO Channel 13/0 : 29[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104431:104593 [2] NCCL INFO Channel 01/0 : 50[2] -> 49[1] via P2P/IPC
+ip-26-0-166-244:897247:897404 [7] NCCL INFO Channel 07/0 : 47[7] -> 40[0] via P2P/IPC
+ip-26-0-167-9:1039544:1039702 [7] NCCL INFO Channel 15/0 : 63[7] -> 56[0] via P2P/IPC
+ip-26-0-166-214:624945:625111 [2] NCCL INFO Channel 08/0 : 34[2] -> 33[1] via P2P/IPC
+ip-26-0-166-214:624944:625106 [1] NCCL INFO Channel 00/0 : 33[1] -> 32[0] via P2P/IPC
+ip-26-0-166-36:104432:104592 [3] NCCL INFO Channel 03/0 : 51[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104434:104591 [5] NCCL INFO Channel 05/0 : 53[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897244:897407 [4] NCCL INFO Channel 04/0 : 52[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624948:625104 [5] NCCL INFO Channel 05/0 : 5[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104436:104596 [7] NCCL INFO Channel 14/0 : 55[7] -> 48[0] via P2P/IPC
+ip-26-0-162-46:702884:703047 [4] NCCL INFO Channel 12/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127297:127453 [7] NCCL INFO Channel 00/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-166-244:897242:897400 [2] NCCL INFO Channel 08/0 : 42[2] -> 41[1] via P2P/IPC
+ip-26-0-166-214:624949:625110 [6] NCCL INFO Channel 06/0 : 54[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624947:625105 [4] NCCL INFO Channel 04/0 : 52[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624946:625107 [3] NCCL INFO Channel 03/0 : 3[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127294:127457 [4] NCCL INFO Channel 04/0 : 28[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897244:897407 [4] NCCL INFO Channel 12/0 : 52[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897245:897401 [5] NCCL INFO Channel 05/0 : 53[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624948:625104 [5] NCCL INFO Channel 05/0 : 37[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624949:625110 [6] NCCL INFO Channel 06/0 : 38[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624947:625105 [4] NCCL INFO Channel 04/0 : 36[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624950:625109 [7] NCCL INFO Channel 14/0 : 39[7] -> 32[0] via P2P/IPC
+ip-26-0-166-214:624946:625107 [3] NCCL INFO Channel 03/0 : 35[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9629:9793 [4] NCCL INFO Channel 04/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897247:897404 [7] NCCL INFO Channel 09/0 : 47[7] -> 40[0] via P2P/IPC
+ip-26-0-166-244:897244:897407 [4] NCCL INFO Channel 12/0 : 44[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127295:127452 [5] NCCL INFO Channel 05/0 : 29[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9630:9789 [5] NCCL INFO Channel 05/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9629:9793 [4] NCCL INFO Channel 12/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897245:897401 [5] NCCL INFO Channel 13/0 : 53[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104431:104593 [2] NCCL INFO Channel 08/0 : 50[2] -> 49[1] via P2P/IPC
+ip-26-0-165-213:9632:9792 [7] NCCL INFO Channel 00/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-165-213:9630:9789 [5] NCCL INFO Channel 13/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897245:897401 [5] NCCL INFO Channel 13/0 : 45[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624948:625104 [5] NCCL INFO Channel 05/0 : 53[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-46:702887:703045 [7] NCCL INFO Channel 00/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-165-213:9629:9793 [4] NCCL INFO Channel 12/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-46:702885:703048 [5] NCCL INFO Channel 13/0 : 13[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104436:104596 [7] NCCL INFO Channel 15/0 : 55[7] -> 48[0] via P2P/IPC
+ip-26-0-166-36:104435:104595 [6] NCCL INFO Channel 06/0 : 62[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624945:625111 [2] NCCL INFO Channel 09/0 : 34[2] -> 33[1] via P2P/IPC
+ip-26-0-166-15:127297:127453 [7] NCCL INFO Channel 06/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-166-214:624946:625107 [3] NCCL INFO Channel 03/0 : 51[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-46:702883:703044 [3] NCCL INFO Channel 11/0 : 11[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9630:9789 [5] NCCL INFO Channel 13/0 : 13[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104433:104590 [4] NCCL INFO Channel 04/0 : 60[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897242:897400 [2] NCCL INFO Channel 09/0 : 42[2] -> 41[1] via P2P/IPC
+ip-26-0-166-214:624948:625104 [5] NCCL INFO Channel 05/0 : 37[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104435:104595 [6] NCCL INFO Channel 06/0 : 54[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104433:104590 [4] NCCL INFO Channel 04/0 : 52[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762819:762982 [4] NCCL INFO Channel 04/0 : 28[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624944:625106 [1] NCCL INFO Channel 08/0 : 33[1] -> 32[0] via P2P/IPC
+ip-26-0-166-214:624946:625107 [3] NCCL INFO Channel 03/0 : 35[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897247:897404 [7] NCCL INFO Channel 10/0 : 47[7] -> 40[0] via P2P/IPC
+ip-26-0-166-214:624949:625110 [6] NCCL INFO Channel 14/0 : 46[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624947:625105 [4] NCCL INFO Channel 12/0 : 44[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104431:104593 [2] NCCL INFO Channel 09/0 : 50[2] -> 49[1] via P2P/IPC
+ip-26-0-166-36:104435:104595 [6] NCCL INFO Channel 14/0 : 54[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762821:762988 [6] NCCL INFO Channel 06/0 : 30[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624950:625109 [7] NCCL INFO Channel 15/0 : 39[7] -> 32[0] via P2P/IPC
+ip-26-0-166-36:104433:104590 [4] NCCL INFO Channel 12/0 : 52[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762819:762982 [4] NCCL INFO Channel 04/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762821:762988 [6] NCCL INFO Channel 06/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9632:9792 [7] NCCL INFO Channel 06/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-166-36:104434:104591 [5] NCCL INFO Channel 05/0 : 61[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127297:127453 [7] NCCL INFO Channel 08/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-167-9:1039543:1039699 [6] NCCL INFO Channel 06/0 : 62[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762819:762982 [4] NCCL INFO Channel 12/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762821:762988 [6] NCCL INFO Channel 14/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104434:104591 [5] NCCL INFO Channel 05/0 : 53[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624948:625104 [5] NCCL INFO Channel 13/0 : 45[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-46:702887:703045 [7] NCCL INFO Channel 06/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-166-125:762820:762981 [5] NCCL INFO Channel 05/0 : 29[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104432:104592 [3] NCCL INFO Channel 03/0 : 59[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762820:762981 [5] NCCL INFO Channel 05/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104434:104591 [5] NCCL INFO Channel 13/0 : 53[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624946:625107 [3] NCCL INFO Channel 11/0 : 43[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-167-9:1039544:1039702 [7] NCCL INFO Channel 00/0 : 63[7] -> 62[6] via P2P/IPC
+ip-26-0-166-36:104432:104592 [3] NCCL INFO Channel 03/0 : 51[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897247:897404 [7] NCCL INFO Channel 11/0 : 47[7] -> 40[0] via P2P/IPC
+ip-26-0-166-125:762818:762985 [3] NCCL INFO Channel 03/0 : 27[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104432:104592 [3] NCCL INFO Channel 11/0 : 51[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762820:762981 [5] NCCL INFO Channel 13/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762818:762985 [3] NCCL INFO Channel 03/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9632:9792 [7] NCCL INFO Channel 08/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-166-125:762818:762985 [3] NCCL INFO Channel 11/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127296:127451 [6] NCCL INFO Channel 14/0 : 46[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897247:897404 [7] NCCL INFO Channel 12/0 : 47[7] -> 40[0] via P2P/IPC
+ip-26-0-166-15:127296:127451 [6] NCCL INFO Channel 14/0 : 30[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-46:702887:703045 [7] NCCL INFO Channel 08/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-167-9:1039544:1039702 [7] NCCL INFO Channel 06/0 : 63[7] -> 62[6] via P2P/IPC
+ip-26-0-166-244:897247:897404 [7] NCCL INFO Channel 13/0 : 47[7] -> 40[0] via P2P/IPC
+ip-26-0-166-15:127296:127451 [6] NCCL INFO Channel 06/0 : 30[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9631:9788 [6] NCCL INFO Channel 06/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9631:9788 [6] NCCL INFO Channel 14/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9631:9788 [6] NCCL INFO Channel 14/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-46:702886:703049 [6] NCCL INFO Channel 14/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897247:897404 [7] NCCL INFO Channel 14/0 : 47[7] -> 40[0] via P2P/IPC
+ip-26-0-167-9:1039544:1039702 [7] NCCL INFO Channel 08/0 : 63[7] -> 62[6] via P2P/IPC
+ip-26-0-166-244:897247:897404 [7] NCCL INFO Channel 15/0 : 47[7] -> 40[0] via P2P/IPC
+ip-26-0-166-125:762819:762982 [4] NCCL INFO Channel 00/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-166-36:104434:104591 [5] NCCL INFO Channel 00/0 : 53[5] -> 52[4] via P2P/IPC
+ip-26-0-166-244:897246:897405 [6] NCCL INFO Channel 06/0 : 54[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897246:897405 [6] NCCL INFO Channel 14/0 : 54[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897246:897405 [6] NCCL INFO Channel 14/0 : 46[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762820:762981 [5] NCCL INFO Channel 00/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-166-125:762819:762982 [4] NCCL INFO Channel 03/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-166-15:127297:127453 [7] NCCL INFO Channel 14/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-166-36:104434:104591 [5] NCCL INFO Channel 04/0 : 53[5] -> 52[4] via P2P/IPC
+ip-26-0-165-213:9632:9792 [7] NCCL INFO Channel 14/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-166-36:104433:104590 [4] NCCL INFO Channel 00/0 : 52[4] -> 51[3] via P2P/IPC
+ip-26-0-166-36:104434:104591 [5] NCCL INFO Channel 08/0 : 53[5] -> 52[4] via P2P/IPC
+ip-26-0-166-244:897247:897404 [7] NCCL INFO Channel 00/0 : 47[7] -> 46[6] via P2P/IPC
+ip-26-0-166-36:104433:104590 [4] NCCL INFO Channel 03/0 : 52[4] -> 51[3] via P2P/IPC
+ip-26-0-166-36:104434:104591 [5] NCCL INFO Channel 12/0 : 53[5] -> 52[4] via P2P/IPC
+ip-26-0-166-125:762820:762981 [5] NCCL INFO Channel 04/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-166-125:762819:762982 [4] NCCL INFO Channel 05/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-166-214:624948:625104 [5] NCCL INFO Channel 00/0 : 37[5] -> 36[4] via P2P/IPC
+ip-26-0-166-244:897247:897404 [7] NCCL INFO Channel 06/0 : 47[7] -> 46[6] via P2P/IPC
+ip-26-0-162-46:702887:703045 [7] NCCL INFO Channel 14/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-166-36:104433:104590 [4] NCCL INFO Channel 05/0 : 52[4] -> 51[3] via P2P/IPC
+ip-26-0-166-36:104436:104596 [7] NCCL INFO Channel 00/0 : 55[7] -> 54[6] via P2P/IPC
+ip-26-0-166-244:897245:897401 [5] NCCL INFO Channel 00/0 : 45[5] -> 44[4] via P2P/IPC
+ip-26-0-166-214:624947:625105 [4] NCCL INFO Channel 00/0 : 36[4] -> 35[3] via P2P/IPC
+ip-26-0-166-125:762821:762988 [6] NCCL INFO Channel 00/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-166-125:762820:762981 [5] NCCL INFO Channel 08/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-166-125:762819:762982 [4] NCCL INFO Channel 06/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-166-214:624948:625104 [5] NCCL INFO Channel 04/0 : 37[5] -> 36[4] via P2P/IPC
+ip-26-0-166-244:897247:897404 [7] NCCL INFO Channel 08/0 : 47[7] -> 46[6] via P2P/IPC
+ip-26-0-166-36:104433:104590 [4] NCCL INFO Channel 06/0 : 52[4] -> 51[3] via P2P/IPC
+ip-26-0-166-244:897244:897407 [4] NCCL INFO Channel 00/0 : 44[4] -> 43[3] via P2P/IPC
+ip-26-0-166-36:104436:104596 [7] NCCL INFO Channel 06/0 : 55[7] -> 54[6] via P2P/IPC
+ip-26-0-166-214:624947:625105 [4] NCCL INFO Channel 03/0 : 36[4] -> 35[3] via P2P/IPC
+ip-26-0-166-125:762821:762988 [6] NCCL INFO Channel 05/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-166-36:104435:104595 [6] NCCL INFO Channel 00/0 : 54[6] -> 53[5] via P2P/IPC
+ip-26-0-166-214:624948:625104 [5] NCCL INFO Channel 08/0 : 37[5] -> 36[4] via P2P/IPC
+ip-26-0-166-125:762819:762982 [4] NCCL INFO Channel 07/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-166-244:897245:897401 [5] NCCL INFO Channel 04/0 : 45[5] -> 44[4] via P2P/IPC
+ip-26-0-166-36:104433:104590 [4] NCCL INFO Channel 07/0 : 52[4] -> 51[3] via P2P/IPC
+ip-26-0-166-214:624949:625110 [6] NCCL INFO Channel 00/0 : 38[6] -> 37[5] via P2P/IPC
+ip-26-0-166-125:762820:762981 [5] NCCL INFO Channel 12/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-166-125:762821:762988 [6] NCCL INFO Channel 08/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-166-214:624947:625105 [4] NCCL INFO Channel 05/0 : 36[4] -> 35[3] via P2P/IPC
+ip-26-0-166-36:104436:104596 [7] NCCL INFO Channel 08/0 : 55[7] -> 54[6] via P2P/IPC
+ip-26-0-166-214:624948:625104 [5] NCCL INFO Channel 12/0 : 37[5] -> 36[4] via P2P/IPC
+ip-26-0-166-36:104435:104595 [6] NCCL INFO Channel 05/0 : 54[6] -> 53[5] via P2P/IPC
+ip-26-0-166-125:762819:762982 [4] NCCL INFO Channel 08/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-166-36:104433:104590 [4] NCCL INFO Channel 08/0 : 52[4] -> 51[3] via P2P/IPC
+ip-26-0-166-214:624949:625110 [6] NCCL INFO Channel 05/0 : 38[6] -> 37[5] via P2P/IPC
+ip-26-0-166-125:762821:762988 [6] NCCL INFO Channel 13/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-166-214:624950:625109 [7] NCCL INFO Channel 00/0 : 39[7] -> 38[6] via P2P/IPC
+ip-26-0-166-214:624947:625105 [4] NCCL INFO Channel 06/0 : 36[4] -> 35[3] via P2P/IPC
+ip-26-0-166-125:762819:762982 [4] NCCL INFO Channel 11/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-166-214:624949:625110 [6] NCCL INFO Channel 08/0 : 38[6] -> 37[5] via P2P/IPC
+ip-26-0-166-125:762819:762982 [4] NCCL INFO Channel 13/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-166-125:762822:762984 [7] NCCL INFO Channel 00/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-166-214:624950:625109 [7] NCCL INFO Channel 06/0 : 39[7] -> 38[6] via P2P/IPC
+ip-26-0-166-125:762818:762985 [3] NCCL INFO Channel 00/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-166-214:624947:625105 [4] NCCL INFO Channel 07/0 : 36[4] -> 35[3] via P2P/IPC
+ip-26-0-166-214:624949:625110 [6] NCCL INFO Channel 13/0 : 38[6] -> 37[5] via P2P/IPC
+ip-26-0-166-125:762819:762982 [4] NCCL INFO Channel 14/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-167-9:1039544:1039702 [7] NCCL INFO Channel 14/0 : 63[7] -> 62[6] via P2P/IPC
+ip-26-0-166-125:762818:762985 [3] NCCL INFO Channel 02/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-166-125:762822:762984 [7] NCCL INFO Channel 06/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-166-244:897246:897405 [6] NCCL INFO Channel 00/0 : 46[6] -> 45[5] via P2P/IPC
+ip-26-0-166-214:624950:625109 [7] NCCL INFO Channel 08/0 : 39[7] -> 38[6] via P2P/IPC
+ip-26-0-166-214:624947:625105 [4] NCCL INFO Channel 08/0 : 36[4] -> 35[3] via P2P/IPC
+ip-26-0-166-125:762819:762982 [4] NCCL INFO Channel 15/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-166-125:762818:762985 [3] NCCL INFO Channel 08/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-166-125:762822:762984 [7] NCCL INFO Channel 08/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-166-244:897247:897404 [7] NCCL INFO Channel 14/0 : 47[7] -> 46[6] via P2P/IPC
+ip-26-0-166-244:897244:897407 [4] NCCL INFO Channel 03/0 : 44[4] -> 43[3] via P2P/IPC
+ip-26-0-166-36:104436:104596 [7] NCCL INFO Channel 14/0 : 55[7] -> 54[6] via P2P/IPC
+ip-26-0-166-36:104435:104595 [6] NCCL INFO Channel 08/0 : 54[6] -> 53[5] via P2P/IPC
+ip-26-0-166-36:104432:104592 [3] NCCL INFO Channel 00/0 : 51[3] -> 50[2] via P2P/IPC
+ip-26-0-166-15:127295:127452 [5] NCCL INFO Channel 00/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-166-214:624946:625107 [3] NCCL INFO Channel 00/0 : 35[3] -> 34[2] via P2P/IPC
+ip-26-0-166-214:624947:625105 [4] NCCL INFO Channel 11/0 : 36[4] -> 35[3] via P2P/IPC
+ip-26-0-166-15:127295:127452 [5] NCCL INFO Channel 04/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-166-244:897245:897401 [5] NCCL INFO Channel 08/0 : 45[5] -> 44[4] via P2P/IPC
+ip-26-0-162-46:702884:703047 [4] NCCL INFO Channel 00/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-162-46:702885:703048 [5] NCCL INFO Channel 00/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-167-9:1039543:1039699 [6] NCCL INFO Channel 00/0 : 62[6] -> 61[5] via P2P/IPC
+ip-26-0-167-9:1039542:1039701 [5] NCCL INFO Channel 00/0 : 61[5] -> 60[4] via P2P/IPC
+ip-26-0-166-15:127295:127452 [5] NCCL INFO Channel 08/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-166-244:897246:897405 [6] NCCL INFO Channel 05/0 : 46[6] -> 45[5] via P2P/IPC
+ip-26-0-166-36:104433:104590 [4] NCCL INFO Channel 11/0 : 52[4] -> 51[3] via P2P/IPC
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO Channel 00/0 : 60[4] -> 59[3] via P2P/IPC
+ip-26-0-167-9:1039543:1039699 [6] NCCL INFO Channel 05/0 : 62[6] -> 61[5] via P2P/IPC
+ip-26-0-166-244:897244:897407 [4] NCCL INFO Channel 05/0 : 44[4] -> 43[3] via P2P/IPC
+ip-26-0-166-15:127296:127451 [6] NCCL INFO Channel 00/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-166-36:104432:104592 [3] NCCL INFO Channel 02/0 : 51[3] -> 50[2] via P2P/IPC
+ip-26-0-167-9:1039542:1039701 [5] NCCL INFO Channel 04/0 : 61[5] -> 60[4] via P2P/IPC
+ip-26-0-166-244:897245:897401 [5] NCCL INFO Channel 12/0 : 45[5] -> 44[4] via P2P/IPC
+ip-26-0-166-244:897244:897407 [4] NCCL INFO Channel 06/0 : 44[4] -> 43[3] via P2P/IPC
+ip-26-0-166-36:104435:104595 [6] NCCL INFO Channel 13/0 : 54[6] -> 53[5] via P2P/IPC
+ip-26-0-166-244:897243:897403 [3] NCCL INFO Channel 00/0 : 43[3] -> 42[2] via P2P/IPC
+ip-26-0-166-214:624946:625107 [3] NCCL INFO Channel 02/0 : 35[3] -> 34[2] via P2P/IPC
+ip-26-0-166-214:624950:625109 [7] NCCL INFO Channel 14/0 : 39[7] -> 38[6] via P2P/IPC
+ip-26-0-166-36:104433:104590 [4] NCCL INFO Channel 13/0 : 52[4] -> 51[3] via P2P/IPC
+ip-26-0-166-125:762818:762985 [3] NCCL INFO Channel 10/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-166-125:762822:762984 [7] NCCL INFO Channel 14/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-162-46:702886:703049 [6] NCCL INFO Channel 00/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-162-46:702884:703047 [4] NCCL INFO Channel 03/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-162-46:702885:703048 [5] NCCL INFO Channel 04/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-214:624946:625107 [3] NCCL INFO Channel 08/0 : 35[3] -> 34[2] via P2P/IPC
+ip-26-0-166-244:897244:897407 [4] NCCL INFO Channel 07/0 : 44[4] -> 43[3] via P2P/IPC
+ip-26-0-166-36:104432:104592 [3] NCCL INFO Channel 08/0 : 51[3] -> 50[2] via P2P/IPC
+ip-26-0-166-15:127296:127451 [6] NCCL INFO Channel 05/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-167-9:1039543:1039699 [6] NCCL INFO Channel 08/0 : 62[6] -> 61[5] via P2P/IPC
+ip-26-0-166-244:897246:897405 [6] NCCL INFO Channel 08/0 : 46[6] -> 45[5] via P2P/IPC
+ip-26-0-165-213:9629:9793 [4] NCCL INFO Channel 00/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-166-15:127294:127457 [4] NCCL INFO Channel 00/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-165-213:9630:9789 [5] NCCL INFO Channel 00/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-165-213:9630:9789 [5] NCCL INFO Channel 04/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-165-213:9630:9789 [5] NCCL INFO Channel 08/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-165-213:9630:9789 [5] NCCL INFO Channel 12/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-166-15:127296:127451 [6] NCCL INFO Channel 08/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO Channel 03/0 : 60[4] -> 59[3] via P2P/IPC
+ip-26-0-166-214:624947:625105 [4] NCCL INFO Channel 13/0 : 36[4] -> 35[3] via P2P/IPC
+ip-26-0-162-46:702886:703049 [6] NCCL INFO Channel 05/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-162-46:702884:703047 [4] NCCL INFO Channel 05/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-162-46:702885:703048 [5] NCCL INFO Channel 08/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-244:897244:897407 [4] NCCL INFO Channel 08/0 : 44[4] -> 43[3] via P2P/IPC
+ip-26-0-166-214:624946:625107 [3] NCCL INFO Channel 10/0 : 35[3] -> 34[2] via P2P/IPC
+ip-26-0-165-213:9631:9788 [6] NCCL INFO Channel 00/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-162-46:702886:703049 [6] NCCL INFO Channel 08/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-162-46:702884:703047 [4] NCCL INFO Channel 06/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO Channel 05/0 : 60[4] -> 59[3] via P2P/IPC
+ip-26-0-166-244:897246:897405 [6] NCCL INFO Channel 13/0 : 46[6] -> 45[5] via P2P/IPC
+ip-26-0-166-244:897243:897403 [3] NCCL INFO Channel 02/0 : 43[3] -> 42[2] via P2P/IPC
+ip-26-0-165-213:9629:9793 [4] NCCL INFO Channel 03/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-166-36:104432:104592 [3] NCCL INFO Channel 10/0 : 51[3] -> 50[2] via P2P/IPC
+ip-26-0-166-244:897244:897407 [4] NCCL INFO Channel 11/0 : 44[4] -> 43[3] via P2P/IPC
+ip-26-0-162-46:702885:703048 [5] NCCL INFO Channel 12/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-15:127294:127457 [4] NCCL INFO Channel 03/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-166-15:127295:127452 [5] NCCL INFO Channel 12/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-166-15:127296:127451 [6] NCCL INFO Channel 13/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-165-213:9631:9788 [6] NCCL INFO Channel 05/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-166-244:897243:897403 [3] NCCL INFO Channel 08/0 : 43[3] -> 42[2] via P2P/IPC
+ip-26-0-166-36:104433:104590 [4] NCCL INFO Channel 14/0 : 52[4] -> 51[3] via P2P/IPC
+ip-26-0-165-213:9629:9793 [4] NCCL INFO Channel 05/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-166-244:897244:897407 [4] NCCL INFO Channel 13/0 : 44[4] -> 43[3] via P2P/IPC
+ip-26-0-162-46:702883:703044 [3] NCCL INFO Channel 00/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-162-46:702886:703049 [6] NCCL INFO Channel 13/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-162-46:702884:703047 [4] NCCL INFO Channel 07/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-166-15:127294:127457 [4] NCCL INFO Channel 05/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-166-214:624947:625105 [4] NCCL INFO Channel 14/0 : 36[4] -> 35[3] via P2P/IPC
+ip-26-0-167-9:1039542:1039701 [5] NCCL INFO Channel 08/0 : 61[5] -> 60[4] via P2P/IPC
+ip-26-0-162-46:702883:703044 [3] NCCL INFO Channel 02/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-165-213:9631:9788 [6] NCCL INFO Channel 08/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO Channel 06/0 : 60[4] -> 59[3] via P2P/IPC
+ip-26-0-167-9:1039543:1039699 [6] NCCL INFO Channel 13/0 : 62[6] -> 61[5] via P2P/IPC
+ip-26-0-165-213:9629:9793 [4] NCCL INFO Channel 06/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-166-15:127294:127457 [4] NCCL INFO Channel 06/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-167-9:1039542:1039701 [5] NCCL INFO Channel 12/0 : 61[5] -> 60[4] via P2P/IPC
+ip-26-0-162-46:702884:703047 [4] NCCL INFO Channel 08/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-166-244:897243:897403 [3] NCCL INFO Channel 10/0 : 43[3] -> 42[2] via P2P/IPC
+ip-26-0-165-213:9631:9788 [6] NCCL INFO Channel 13/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO Channel 07/0 : 60[4] -> 59[3] via P2P/IPC
+ip-26-0-162-46:702884:703047 [4] NCCL INFO Channel 11/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-165-213:9629:9793 [4] NCCL INFO Channel 07/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-166-15:127294:127457 [4] NCCL INFO Channel 07/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-162-46:702883:703044 [3] NCCL INFO Channel 08/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-162-46:702884:703047 [4] NCCL INFO Channel 13/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-166-244:897244:897407 [4] NCCL INFO Channel 14/0 : 44[4] -> 43[3] via P2P/IPC
+ip-26-0-165-213:9629:9793 [4] NCCL INFO Channel 08/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-162-46:702883:703044 [3] NCCL INFO Channel 10/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO Channel 08/0 : 60[4] -> 59[3] via P2P/IPC
+ip-26-0-166-36:104433:104590 [4] NCCL INFO Channel 15/0 : 52[4] -> 51[3] via P2P/IPC
+ip-26-0-166-214:624947:625105 [4] NCCL INFO Channel 15/0 : 36[4] -> 35[3] via P2P/IPC
+ip-26-0-162-46:702884:703047 [4] NCCL INFO Channel 14/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-166-15:127294:127457 [4] NCCL INFO Channel 08/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-166-244:897244:897407 [4] NCCL INFO Channel 15/0 : 44[4] -> 43[3] via P2P/IPC
+ip-26-0-166-15:127293:127454 [3] NCCL INFO Channel 00/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-165-213:9628:9786 [3] NCCL INFO Channel 00/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-165-213:9629:9793 [4] NCCL INFO Channel 11/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO Channel 11/0 : 60[4] -> 59[3] via P2P/IPC
+ip-26-0-166-15:127293:127454 [3] NCCL INFO Channel 02/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-165-213:9629:9793 [4] NCCL INFO Channel 13/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-162-46:702884:703047 [4] NCCL INFO Channel 15/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-166-15:127294:127457 [4] NCCL INFO Channel 11/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-166-15:127293:127454 [3] NCCL INFO Channel 08/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-165-213:9629:9793 [4] NCCL INFO Channel 14/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-167-9:1039540:1039705 [3] NCCL INFO Channel 00/0 : 59[3] -> 58[2] via P2P/IPC
+ip-26-0-165-213:9628:9786 [3] NCCL INFO Channel 02/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO Channel 13/0 : 60[4] -> 59[3] via P2P/IPC
+ip-26-0-165-213:9629:9793 [4] NCCL INFO Channel 15/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-165-213:9628:9786 [3] NCCL INFO Channel 08/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-166-15:127294:127457 [4] NCCL INFO Channel 13/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO Channel 14/0 : 60[4] -> 59[3] via P2P/IPC
+ip-26-0-167-9:1039540:1039705 [3] NCCL INFO Channel 02/0 : 59[3] -> 58[2] via P2P/IPC
+ip-26-0-166-15:127293:127454 [3] NCCL INFO Channel 10/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO Channel 15/0 : 60[4] -> 59[3] via P2P/IPC
+ip-26-0-166-15:127294:127457 [4] NCCL INFO Channel 14/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-165-213:9628:9786 [3] NCCL INFO Channel 10/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-167-9:1039540:1039705 [3] NCCL INFO Channel 08/0 : 59[3] -> 58[2] via P2P/IPC
+ip-26-0-167-9:1039540:1039705 [3] NCCL INFO Channel 10/0 : 59[3] -> 58[2] via P2P/IPC
+ip-26-0-166-15:127294:127457 [4] NCCL INFO Channel 15/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-162-46:702880:703042 [0] NCCL INFO Connected all trees
+ip-26-0-162-46:702880:703042 [0] NCCL INFO NVLS comm 0x967b3b0 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-162-46:702881:703043 [1] NCCL INFO Connected all trees
+ip-26-0-162-46:702881:703043 [1] NCCL INFO NVLS comm 0x91d1af0 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-167-9:1039537:1039703 [0] NCCL INFO Connected all trees
+ip-26-0-167-9:1039537:1039703 [0] NCCL INFO NVLS comm 0x9974360 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-167-9:1039543:1039699 [6] NCCL INFO Connected all trees
+ip-26-0-162-46:702885:703048 [5] NCCL INFO Connected all trees
+ip-26-0-167-9:1039543:1039699 [6] NCCL INFO NVLS comm 0x9d28300 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-162-46:702885:703048 [5] NCCL INFO NVLS comm 0x88bf5a0 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-167-9:1039542:1039701 [5] NCCL INFO Connected all trees
+ip-26-0-162-46:702883:703044 [3] NCCL INFO Connected all trees
+ip-26-0-167-9:1039542:1039701 [5] NCCL INFO NVLS comm 0x8c1a980 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-162-46:702883:703044 [3] NCCL INFO NVLS comm 0x834f6a0 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-162-46:702886:703049 [6] NCCL INFO Connected all trees
+ip-26-0-162-46:702886:703049 [6] NCCL INFO NVLS comm 0x8e18ff0 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-162-46:702882:703046 [2] NCCL INFO Connected all trees
+ip-26-0-162-46:702882:703046 [2] NCCL INFO NVLS comm 0x958dad0 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-167-9:1039544:1039702 [7] NCCL INFO Connected all trees
+ip-26-0-167-9:1039544:1039702 [7] NCCL INFO NVLS comm 0x93e67f0 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-167-9:1039538:1039704 [1] NCCL INFO Connected all trees
+ip-26-0-167-9:1039538:1039704 [1] NCCL INFO NVLS comm 0x8e07600 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-162-46:702887:703045 [7] NCCL INFO Connected all trees
+ip-26-0-167-9:1039540:1039705 [3] NCCL INFO Connected all trees
+ip-26-0-162-46:702887:703045 [7] NCCL INFO NVLS comm 0x889e490 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-167-9:1039540:1039705 [3] NCCL INFO NVLS comm 0x955d240 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-162-46:702884:703047 [4] NCCL INFO Connected all trees
+ip-26-0-162-46:702884:703047 [4] NCCL INFO NVLS comm 0x8cdc140 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO Connected all trees
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO NVLS comm 0xa260f20 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-167-9:1039539:1039698 [2] NCCL INFO Connected all trees
+ip-26-0-167-9:1039539:1039698 [2] NCCL INFO NVLS comm 0x8e6cc50 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-166-244:897240:897406 [0] NCCL INFO Connected all trees
+ip-26-0-166-36:104429:104594 [0] NCCL INFO Connected all trees
+ip-26-0-166-244:897240:897406 [0] NCCL INFO NVLS comm 0x90efd70 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-166-36:104429:104594 [0] NCCL INFO NVLS comm 0xa2374d0 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-213:9625:9787 [0] NCCL INFO Connected all trees
+ip-26-0-165-213:9625:9787 [0] NCCL INFO NVLS comm 0x8bb7780 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-213:9626:9790 [1] NCCL INFO Connected all trees
+ip-26-0-165-213:9626:9790 [1] NCCL INFO NVLS comm 0x910f780 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-166-244:897246:897405 [6] NCCL INFO Connected all trees
+ip-26-0-166-244:897247:897404 [7] NCCL INFO Connected all trees
+ip-26-0-166-244:897246:897405 [6] NCCL INFO NVLS comm 0xa1fc860 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-166-36:104431:104593 [2] NCCL INFO Connected all trees
+ip-26-0-166-244:897247:897404 [7] NCCL INFO NVLS comm 0x977c640 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-166-36:104431:104593 [2] NCCL INFO NVLS comm 0x955bca0 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-166-15:127297:127453 [7] NCCL INFO Connected all trees
+ip-26-0-165-213:9632:9792 [7] NCCL INFO Connected all trees
+ip-26-0-166-15:127297:127453 [7] NCCL INFO NVLS comm 0x9b1a7f0 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-213:9632:9792 [7] NCCL INFO NVLS comm 0x8a49870 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-166-15:127290:127456 [0] NCCL INFO Connected all trees
+ip-26-0-166-15:127290:127456 [0] NCCL INFO NVLS comm 0x86f0680 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-166-244:897241:897402 [1] NCCL INFO Connected all trees
+ip-26-0-166-125:762822:762984 [7] NCCL INFO Connected all trees
+ip-26-0-166-244:897242:897400 [2] NCCL INFO Connected all trees
+ip-26-0-166-125:762822:762984 [7] NCCL INFO NVLS comm 0x9dbbc60 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-166-244:897241:897402 [1] NCCL INFO NVLS comm 0x9f8ce70 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-166-244:897242:897400 [2] NCCL INFO NVLS comm 0x845ae40 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-166-36:104433:104590 [4] NCCL INFO Connected all trees
+ip-26-0-166-36:104433:104590 [4] NCCL INFO NVLS comm 0xa026de0 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-166-36:104436:104596 [7] NCCL INFO Connected all trees
+ip-26-0-166-36:104436:104596 [7] NCCL INFO NVLS comm 0x9507760 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-166-36:104432:104592 [3] NCCL INFO Connected all trees
+ip-26-0-166-36:104434:104591 [5] NCCL INFO Connected all trees
+ip-26-0-166-36:104435:104595 [6] NCCL INFO Connected all trees
+ip-26-0-166-214:624948:625104 [5] NCCL INFO Connected all trees
+ip-26-0-166-36:104432:104592 [3] NCCL INFO NVLS comm 0x85daf90 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-166-36:104435:104595 [6] NCCL INFO NVLS comm 0x8b98830 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-166-214:624948:625104 [5] NCCL INFO NVLS comm 0x9e36180 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-166-36:104434:104591 [5] NCCL INFO NVLS comm 0x8836500 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-166-125:762821:762988 [6] NCCL INFO Connected all trees
+ip-26-0-166-125:762819:762982 [4] NCCL INFO Connected all trees
+ip-26-0-166-244:897244:897407 [4] NCCL INFO Connected all trees
+ip-26-0-166-125:762821:762988 [6] NCCL INFO NVLS comm 0x85f8640 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-166-125:762819:762982 [4] NCCL INFO NVLS comm 0x91ac710 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-166-244:897244:897407 [4] NCCL INFO NVLS comm 0x8c707d0 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-166-244:897243:897403 [3] NCCL INFO Connected all trees
+ip-26-0-166-244:897243:897403 [3] NCCL INFO NVLS comm 0x87e71f0 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-166-125:762815:762983 [0] NCCL INFO Connected all trees
+ip-26-0-166-125:762816:762987 [1] NCCL INFO Connected all trees
+ip-26-0-166-214:624943:625108 [0] NCCL INFO Connected all trees
+ip-26-0-166-125:762815:762983 [0] NCCL INFO NVLS comm 0x832ff30 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-166-214:624943:625108 [0] NCCL INFO NVLS comm 0x919eef0 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-166-125:762816:762987 [1] NCCL INFO NVLS comm 0x93d1180 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-213:9629:9793 [4] NCCL INFO Connected all trees
+ip-26-0-165-213:9629:9793 [4] NCCL INFO NVLS comm 0x86eab90 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-213:9630:9789 [5] NCCL INFO Connected all trees
+ip-26-0-165-213:9631:9788 [6] NCCL INFO Connected all trees
+ip-26-0-165-213:9630:9789 [5] NCCL INFO NVLS comm 0x84f2a80 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-213:9631:9788 [6] NCCL INFO NVLS comm 0x98a0a00 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-166-15:127291:127455 [1] NCCL INFO Connected all trees
+ip-26-0-166-214:624946:625107 [3] NCCL INFO Connected all trees
+ip-26-0-166-15:127291:127455 [1] NCCL INFO NVLS comm 0x9f78530 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-166-214:624946:625107 [3] NCCL INFO NVLS comm 0x9da4f90 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-166-214:624949:625110 [6] NCCL INFO Connected all trees
+ip-26-0-166-214:624945:625111 [2] NCCL INFO Connected all trees
+ip-26-0-166-214:624949:625110 [6] NCCL INFO NVLS comm 0x87cdfa0 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-213:9627:9791 [2] NCCL INFO Connected all trees
+ip-26-0-166-214:624945:625111 [2] NCCL INFO NVLS comm 0x9504970 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-213:9627:9791 [2] NCCL INFO NVLS comm 0x86f9b60 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-166-125:762817:762986 [2] NCCL INFO Connected all trees
+ip-26-0-166-214:624944:625106 [1] NCCL INFO Connected all trees
+ip-26-0-166-214:624950:625109 [7] NCCL INFO Connected all trees
+ip-26-0-166-125:762817:762986 [2] NCCL INFO NVLS comm 0xa1ccb40 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-166-36:104430:104589 [1] NCCL INFO Connected all trees
+ip-26-0-166-214:624950:625109 [7] NCCL INFO NVLS comm 0x8b7a600 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-166-214:624944:625106 [1] NCCL INFO NVLS comm 0x82f1510 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-166-36:104430:104589 [1] NCCL INFO NVLS comm 0x9a479d0 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-166-244:897245:897401 [5] NCCL INFO Connected all trees
+ip-26-0-166-244:897245:897401 [5] NCCL INFO NVLS comm 0x9819580 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-166-125:762820:762981 [5] NCCL INFO Connected all trees
+ip-26-0-166-125:762820:762981 [5] NCCL INFO NVLS comm 0x90d8cd0 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-166-125:762818:762985 [3] NCCL INFO Connected all trees
+ip-26-0-166-125:762818:762985 [3] NCCL INFO NVLS comm 0x850d180 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-166-214:624947:625105 [4] NCCL INFO Connected all trees
+ip-26-0-162-46:702883:703044 [3] NCCL INFO Channel 01/0 : 3[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624947:625105 [4] NCCL INFO NVLS comm 0x9dbf120 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-162-46:702883:703044 [3] NCCL INFO Channel 03/0 : 3[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-46:702884:703047 [4] NCCL INFO Channel 01/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-46:702883:703044 [3] NCCL INFO Channel 05/0 : 3[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-46:702886:703049 [6] NCCL INFO Channel 01/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-46:702885:703048 [5] NCCL INFO Channel 01/0 : 5[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-46:702883:703044 [3] NCCL INFO Channel 07/0 : 3[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-46:702886:703049 [6] NCCL INFO Channel 03/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-46:702884:703047 [4] NCCL INFO Channel 03/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-46:702887:703045 [7] NCCL INFO Channel 01/0 : 7[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-46:702885:703048 [5] NCCL INFO Channel 03/0 : 5[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-46:702883:703044 [3] NCCL INFO Channel 09/0 : 3[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-46:702886:703049 [6] NCCL INFO Channel 05/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-46:702884:703047 [4] NCCL INFO Channel 05/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-46:702887:703045 [7] NCCL INFO Channel 03/0 : 7[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-46:702880:703042 [0] NCCL INFO Channel 01/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-46:702885:703048 [5] NCCL INFO Channel 05/0 : 5[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-46:702883:703044 [3] NCCL INFO Channel 13/0 : 3[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-46:702886:703049 [6] NCCL INFO Channel 07/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-46:702884:703047 [4] NCCL INFO Channel 07/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-46:702880:703042 [0] NCCL INFO Channel 03/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-46:702887:703045 [7] NCCL INFO Channel 05/0 : 7[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-46:702881:703043 [1] NCCL INFO Channel 01/0 : 1[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-46:702885:703048 [5] NCCL INFO Channel 07/0 : 5[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-46:702883:703044 [3] NCCL INFO Channel 15/0 : 3[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-46:702886:703049 [6] NCCL INFO Channel 09/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-46:702884:703047 [4] NCCL INFO Channel 09/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-46:702880:703042 [0] NCCL INFO Channel 05/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-46:702881:703043 [1] NCCL INFO Channel 03/0 : 1[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-46:702887:703045 [7] NCCL INFO Channel 07/0 : 7[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-46:702885:703048 [5] NCCL INFO Channel 09/0 : 5[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-46:702886:703049 [6] NCCL INFO Channel 11/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-46:702884:703047 [4] NCCL INFO Channel 11/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-46:702880:703042 [0] NCCL INFO Channel 07/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-46:702881:703043 [1] NCCL INFO Channel 05/0 : 1[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-46:702887:703045 [7] NCCL INFO Channel 09/0 : 7[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-46:702884:703047 [4] NCCL INFO Channel 13/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-46:702886:703049 [6] NCCL INFO Channel 13/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-46:702885:703048 [5] NCCL INFO Channel 11/0 : 5[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-46:702880:703042 [0] NCCL INFO Channel 09/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-46:702881:703043 [1] NCCL INFO Channel 07/0 : 1[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-46:702886:703049 [6] NCCL INFO Channel 15/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-46:702887:703045 [7] NCCL INFO Channel 11/0 : 7[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-46:702884:703047 [4] NCCL INFO Channel 15/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-46:702885:703048 [5] NCCL INFO Channel 15/0 : 5[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-46:702880:703042 [0] NCCL INFO Channel 11/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-46:702881:703043 [1] NCCL INFO Channel 11/0 : 1[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-46:702887:703045 [7] NCCL INFO Channel 13/0 : 7[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-46:702881:703043 [1] NCCL INFO Channel 13/0 : 1[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-46:702880:703042 [0] NCCL INFO Channel 13/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-46:702881:703043 [1] NCCL INFO Channel 15/0 : 1[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-46:702880:703042 [0] NCCL INFO Channel 15/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-46:702882:703046 [2] NCCL INFO Channel 01/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-46:702882:703046 [2] NCCL INFO Channel 03/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-46:702882:703046 [2] NCCL INFO Channel 05/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-46:702882:703046 [2] NCCL INFO Channel 07/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-46:702882:703046 [2] NCCL INFO Channel 09/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-46:702882:703046 [2] NCCL INFO Channel 11/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9628:9786 [3] NCCL INFO Connected all trees
+ip-26-0-162-46:702882:703046 [2] NCCL INFO Channel 13/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-46:702882:703046 [2] NCCL INFO Channel 15/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9628:9786 [3] NCCL INFO NVLS comm 0x90d46e0 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-166-15:127296:127451 [6] NCCL INFO Connected all trees
+ip-26-0-166-15:127293:127454 [3] NCCL INFO Connected all trees
+ip-26-0-166-15:127296:127451 [6] NCCL INFO NVLS comm 0x947e710 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-166-15:127293:127454 [3] NCCL INFO NVLS comm 0x9ee0590 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-166-15:127295:127452 [5] NCCL INFO Connected all trees
+ip-26-0-166-15:127295:127452 [5] NCCL INFO NVLS comm 0x88629b0 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-166-15:127292:127458 [2] NCCL INFO Connected all trees
+ip-26-0-166-15:127292:127458 [2] NCCL INFO NVLS comm 0x8fd9570 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-166-15:127294:127457 [4] NCCL INFO Connected all trees
+ip-26-0-166-15:127294:127457 [4] NCCL INFO NVLS comm 0x8d79aa0 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-167-9:1039543:1039699 [6] NCCL INFO Channel 00/0 : 54[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-167-9:1039543:1039699 [6] NCCL INFO Channel 02/0 : 54[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-167-9:1039544:1039702 [7] NCCL INFO Channel 00/0 : 55[7] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-167-9:1039537:1039703 [0] NCCL INFO Channel 02/0 : 48[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO Channel 00/0 : 52[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-167-9:1039543:1039699 [6] NCCL INFO Channel 04/0 : 54[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-167-9:1039544:1039702 [7] NCCL INFO Channel 02/0 : 55[7] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-167-9:1039537:1039703 [0] NCCL INFO Channel 04/0 : 48[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO Channel 02/0 : 52[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-167-9:1039543:1039699 [6] NCCL INFO Channel 08/0 : 54[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-167-9:1039544:1039702 [7] NCCL INFO Channel 04/0 : 55[7] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-167-9:1039537:1039703 [0] NCCL INFO Channel 06/0 : 48[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-167-9:1039542:1039701 [5] NCCL INFO Channel 00/0 : 53[5] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO Channel 06/0 : 52[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-167-9:1039538:1039704 [1] NCCL INFO Channel 00/0 : 49[1] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-167-9:1039543:1039699 [6] NCCL INFO Channel 10/0 : 54[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-167-9:1039539:1039698 [2] NCCL INFO Channel 00/0 : 50[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-167-9:1039544:1039702 [7] NCCL INFO Channel 06/0 : 55[7] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-167-9:1039540:1039705 [3] NCCL INFO Channel 00/0 : 51[3] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-167-9:1039543:1039699 [6] NCCL INFO Channel 12/0 : 54[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-167-9:1039538:1039704 [1] NCCL INFO Channel 02/0 : 49[1] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-167-9:1039537:1039703 [0] NCCL INFO Channel 08/0 : 48[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO Channel 08/0 : 52[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-167-9:1039542:1039701 [5] NCCL INFO Channel 02/0 : 53[5] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-167-9:1039544:1039702 [7] NCCL INFO Channel 08/0 : 55[7] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-167-9:1039540:1039705 [3] NCCL INFO Channel 02/0 : 51[3] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-167-9:1039539:1039698 [2] NCCL INFO Channel 04/0 : 50[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-167-9:1039543:1039699 [6] NCCL INFO Channel 14/0 : 54[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-167-9:1039537:1039703 [0] NCCL INFO Channel 10/0 : 48[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO Channel 10/0 : 52[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-167-9:1039538:1039704 [1] NCCL INFO Channel 04/0 : 49[1] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-167-9:1039544:1039702 [7] NCCL INFO Channel 10/0 : 55[7] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-167-9:1039539:1039698 [2] NCCL INFO Channel 06/0 : 50[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-167-9:1039540:1039705 [3] NCCL INFO Channel 04/0 : 51[3] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-167-9:1039542:1039701 [5] NCCL INFO Channel 04/0 : 53[5] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-167-9:1039537:1039703 [0] NCCL INFO Channel 12/0 : 48[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO Channel 12/0 : 52[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-167-9:1039538:1039704 [1] NCCL INFO Channel 06/0 : 49[1] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-167-9:1039544:1039702 [7] NCCL INFO Channel 12/0 : 55[7] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-167-9:1039539:1039698 [2] NCCL INFO Channel 08/0 : 50[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-167-9:1039540:1039705 [3] NCCL INFO Channel 06/0 : 51[3] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-167-9:1039537:1039703 [0] NCCL INFO Channel 14/0 : 48[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-167-9:1039542:1039701 [5] NCCL INFO Channel 06/0 : 53[5] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO Channel 14/0 : 52[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-167-9:1039544:1039702 [7] NCCL INFO Channel 14/0 : 55[7] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-167-9:1039538:1039704 [1] NCCL INFO Channel 08/0 : 49[1] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-167-9:1039539:1039698 [2] NCCL INFO Channel 10/0 : 50[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-167-9:1039540:1039705 [3] NCCL INFO Channel 08/0 : 51[3] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-167-9:1039542:1039701 [5] NCCL INFO Channel 08/0 : 53[5] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-167-9:1039540:1039705 [3] NCCL INFO Channel 10/0 : 51[3] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-167-9:1039539:1039698 [2] NCCL INFO Channel 12/0 : 50[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-167-9:1039542:1039701 [5] NCCL INFO Channel 10/0 : 53[5] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-167-9:1039538:1039704 [1] NCCL INFO Channel 10/0 : 49[1] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-167-9:1039540:1039705 [3] NCCL INFO Channel 12/0 : 51[3] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-167-9:1039539:1039698 [2] NCCL INFO Channel 14/0 : 50[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-167-9:1039542:1039701 [5] NCCL INFO Channel 12/0 : 53[5] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-167-9:1039538:1039704 [1] NCCL INFO Channel 12/0 : 49[1] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-167-9:1039540:1039705 [3] NCCL INFO Channel 14/0 : 51[3] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-167-9:1039538:1039704 [1] NCCL INFO Channel 14/0 : 49[1] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-167-9:1039542:1039701 [5] NCCL INFO Channel 14/0 : 53[5] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897241:897402 [1] NCCL INFO Channel 01/0 : 33[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897245:897401 [5] NCCL INFO Channel 01/0 : 37[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897241:897402 [1] NCCL INFO Channel 03/0 : 33[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897240:897406 [0] NCCL INFO Channel 01/0 : 32[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897245:897401 [5] NCCL INFO Channel 03/0 : 37[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897241:897402 [1] NCCL INFO Channel 05/0 : 33[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897240:897406 [0] NCCL INFO Channel 03/0 : 32[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897245:897401 [5] NCCL INFO Channel 05/0 : 37[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897241:897402 [1] NCCL INFO Channel 07/0 : 33[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897240:897406 [0] NCCL INFO Channel 05/0 : 32[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897245:897401 [5] NCCL INFO Channel 07/0 : 37[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897241:897402 [1] NCCL INFO Channel 11/0 : 33[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897240:897406 [0] NCCL INFO Channel 07/0 : 32[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897245:897401 [5] NCCL INFO Channel 09/0 : 37[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897241:897402 [1] NCCL INFO Channel 13/0 : 33[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897240:897406 [0] NCCL INFO Channel 09/0 : 32[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897245:897401 [5] NCCL INFO Channel 11/0 : 37[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897241:897402 [1] NCCL INFO Channel 15/0 : 33[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897240:897406 [0] NCCL INFO Channel 11/0 : 32[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897245:897401 [5] NCCL INFO Channel 15/0 : 37[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897241:897402 [1] NCCL INFO Channel 00/0 : 41[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897240:897406 [0] NCCL INFO Channel 13/0 : 32[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897245:897401 [5] NCCL INFO Channel 00/0 : 45[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897241:897402 [1] NCCL INFO Channel 02/0 : 41[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897240:897406 [0] NCCL INFO Channel 15/0 : 32[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897245:897401 [5] NCCL INFO Channel 01/0 : 45[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897240:897406 [0] NCCL INFO Channel 01/0 : 40[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897241:897402 [1] NCCL INFO Channel 03/0 : 41[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897245:897401 [5] NCCL INFO Channel 02/0 : 45[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897240:897406 [0] NCCL INFO Channel 02/0 : 40[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897241:897402 [1] NCCL INFO Channel 04/0 : 41[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897245:897401 [5] NCCL INFO Channel 03/0 : 45[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897240:897406 [0] NCCL INFO Channel 03/0 : 40[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897241:897402 [1] NCCL INFO Channel 05/0 : 41[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897245:897401 [5] NCCL INFO Channel 04/0 : 45[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897240:897406 [0] NCCL INFO Channel 04/0 : 40[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897245:897401 [5] NCCL INFO Channel 06/0 : 45[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897241:897402 [1] NCCL INFO Channel 06/0 : 41[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897240:897406 [0] NCCL INFO Channel 05/0 : 40[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897247:897404 [7] NCCL INFO Channel 01/0 : 39[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897245:897401 [5] NCCL INFO Channel 07/0 : 45[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897240:897406 [0] NCCL INFO Channel 06/0 : 40[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897241:897402 [1] NCCL INFO Channel 07/0 : 41[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897244:897407 [4] NCCL INFO Channel 01/0 : 36[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897247:897404 [7] NCCL INFO Channel 03/0 : 39[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897245:897401 [5] NCCL INFO Channel 08/0 : 45[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897246:897405 [6] NCCL INFO Channel 01/0 : 38[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897240:897406 [0] NCCL INFO Channel 07/0 : 40[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897241:897402 [1] NCCL INFO Channel 08/0 : 41[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897247:897404 [7] NCCL INFO Channel 05/0 : 39[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897244:897407 [4] NCCL INFO Channel 03/0 : 36[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897245:897401 [5] NCCL INFO Channel 09/0 : 45[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897246:897405 [6] NCCL INFO Channel 03/0 : 38[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897240:897406 [0] NCCL INFO Channel 09/0 : 40[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897241:897402 [1] NCCL INFO Channel 10/0 : 41[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897244:897407 [4] NCCL INFO Channel 05/0 : 36[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897245:897401 [5] NCCL INFO Channel 10/0 : 45[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897247:897404 [7] NCCL INFO Channel 07/0 : 39[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897240:897406 [0] NCCL INFO Channel 10/0 : 40[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897246:897405 [6] NCCL INFO Channel 05/0 : 38[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897243:897403 [3] NCCL INFO Channel 01/0 : 35[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897241:897402 [1] NCCL INFO Channel 11/0 : 41[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897242:897400 [2] NCCL INFO Channel 01/0 : 34[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897245:897401 [5] NCCL INFO Channel 11/0 : 45[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897244:897407 [4] NCCL INFO Channel 07/0 : 36[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897247:897404 [7] NCCL INFO Channel 09/0 : 39[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897240:897406 [0] NCCL INFO Channel 11/0 : 40[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897246:897405 [6] NCCL INFO Channel 07/0 : 38[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897241:897402 [1] NCCL INFO Channel 12/0 : 41[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897243:897403 [3] NCCL INFO Channel 03/0 : 35[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897245:897401 [5] NCCL INFO Channel 12/0 : 45[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897242:897400 [2] NCCL INFO Channel 03/0 : 34[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897244:897407 [4] NCCL INFO Channel 09/0 : 36[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897247:897404 [7] NCCL INFO Channel 11/0 : 39[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897240:897406 [0] NCCL INFO Channel 12/0 : 40[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897246:897405 [6] NCCL INFO Channel 09/0 : 38[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897241:897402 [1] NCCL INFO Channel 13/0 : 41[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897245:897401 [5] NCCL INFO Channel 14/0 : 45[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897243:897403 [3] NCCL INFO Channel 05/0 : 35[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897247:897404 [7] NCCL INFO Channel 13/0 : 39[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897244:897407 [4] NCCL INFO Channel 11/0 : 36[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897242:897400 [2] NCCL INFO Channel 05/0 : 34[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897240:897406 [0] NCCL INFO Channel 13/0 : 40[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897246:897405 [6] NCCL INFO Channel 11/0 : 38[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897241:897402 [1] NCCL INFO Channel 14/0 : 41[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897245:897401 [5] NCCL INFO Channel 15/0 : 45[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897247:897404 [7] NCCL INFO Channel 00/0 : 47[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897244:897407 [4] NCCL INFO Channel 13/0 : 36[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897240:897406 [0] NCCL INFO Channel 14/0 : 40[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897246:897405 [6] NCCL INFO Channel 13/0 : 38[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897241:897402 [1] NCCL INFO Channel 15/0 : 41[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897242:897400 [2] NCCL INFO Channel 07/0 : 34[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897243:897403 [3] NCCL INFO Channel 07/0 : 35[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897247:897404 [7] NCCL INFO Channel 01/0 : 47[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897244:897407 [4] NCCL INFO Channel 15/0 : 36[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897240:897406 [0] NCCL INFO Channel 15/0 : 40[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897246:897405 [6] NCCL INFO Channel 15/0 : 38[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897247:897404 [7] NCCL INFO Channel 02/0 : 47[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897242:897400 [2] NCCL INFO Channel 09/0 : 34[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897243:897403 [3] NCCL INFO Channel 09/0 : 35[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897244:897407 [4] NCCL INFO Channel 00/0 : 44[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897247:897404 [7] NCCL INFO Channel 03/0 : 47[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897246:897405 [6] NCCL INFO Channel 00/0 : 46[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897244:897407 [4] NCCL INFO Channel 01/0 : 44[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897242:897400 [2] NCCL INFO Channel 11/0 : 34[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897243:897403 [3] NCCL INFO Channel 13/0 : 35[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897247:897404 [7] NCCL INFO Channel 04/0 : 47[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897246:897405 [6] NCCL INFO Channel 01/0 : 46[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897244:897407 [4] NCCL INFO Channel 02/0 : 44[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897242:897400 [2] NCCL INFO Channel 13/0 : 34[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897243:897403 [3] NCCL INFO Channel 15/0 : 35[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897247:897404 [7] NCCL INFO Channel 05/0 : 47[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897246:897405 [6] NCCL INFO Channel 02/0 : 46[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897244:897407 [4] NCCL INFO Channel 03/0 : 44[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897247:897404 [7] NCCL INFO Channel 06/0 : 47[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897242:897400 [2] NCCL INFO Channel 15/0 : 34[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897243:897403 [3] NCCL INFO Channel 00/0 : 43[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897246:897405 [6] NCCL INFO Channel 03/0 : 46[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897244:897407 [4] NCCL INFO Channel 05/0 : 44[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897247:897404 [7] NCCL INFO Channel 08/0 : 47[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897246:897405 [6] NCCL INFO Channel 04/0 : 46[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897244:897407 [4] NCCL INFO Channel 06/0 : 44[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897242:897400 [2] NCCL INFO Channel 00/0 : 42[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897243:897403 [3] NCCL INFO Channel 01/0 : 43[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897247:897404 [7] NCCL INFO Channel 09/0 : 47[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897246:897405 [6] NCCL INFO Channel 05/0 : 46[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897244:897407 [4] NCCL INFO Channel 07/0 : 44[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897243:897403 [3] NCCL INFO Channel 02/0 : 43[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897242:897400 [2] NCCL INFO Channel 01/0 : 42[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897247:897404 [7] NCCL INFO Channel 10/0 : 47[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897246:897405 [6] NCCL INFO Channel 07/0 : 46[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897244:897407 [4] NCCL INFO Channel 08/0 : 44[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897243:897403 [3] NCCL INFO Channel 04/0 : 43[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897242:897400 [2] NCCL INFO Channel 03/0 : 42[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897247:897404 [7] NCCL INFO Channel 11/0 : 47[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897246:897405 [6] NCCL INFO Channel 08/0 : 46[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897244:897407 [4] NCCL INFO Channel 09/0 : 44[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897247:897404 [7] NCCL INFO Channel 12/0 : 47[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897242:897400 [2] NCCL INFO Channel 04/0 : 42[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897243:897403 [3] NCCL INFO Channel 05/0 : 43[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897246:897405 [6] NCCL INFO Channel 09/0 : 46[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897244:897407 [4] NCCL INFO Channel 10/0 : 44[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897247:897404 [7] NCCL INFO Channel 13/0 : 47[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897242:897400 [2] NCCL INFO Channel 05/0 : 42[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897246:897405 [6] NCCL INFO Channel 10/0 : 46[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897243:897403 [3] NCCL INFO Channel 06/0 : 43[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897244:897407 [4] NCCL INFO Channel 11/0 : 44[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897247:897404 [7] NCCL INFO Channel 14/0 : 47[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897242:897400 [2] NCCL INFO Channel 06/0 : 42[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897246:897405 [6] NCCL INFO Channel 11/0 : 46[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897244:897407 [4] NCCL INFO Channel 13/0 : 44[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897243:897403 [3] NCCL INFO Channel 07/0 : 43[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897242:897400 [2] NCCL INFO Channel 07/0 : 42[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897246:897405 [6] NCCL INFO Channel 12/0 : 46[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897244:897407 [4] NCCL INFO Channel 14/0 : 44[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897243:897403 [3] NCCL INFO Channel 08/0 : 43[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762820:762981 [5] NCCL INFO Channel 00/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897242:897400 [2] NCCL INFO Channel 08/0 : 42[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897246:897405 [6] NCCL INFO Channel 13/0 : 46[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897244:897407 [4] NCCL INFO Channel 15/0 : 44[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897243:897403 [3] NCCL INFO Channel 09/0 : 43[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762820:762981 [5] NCCL INFO Channel 01/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762822:762984 [7] NCCL INFO Channel 00/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897242:897400 [2] NCCL INFO Channel 09/0 : 42[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897246:897405 [6] NCCL INFO Channel 15/0 : 46[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897243:897403 [3] NCCL INFO Channel 10/0 : 43[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897242:897400 [2] NCCL INFO Channel 11/0 : 42[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762816:762987 [1] NCCL INFO Channel 00/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762820:762981 [5] NCCL INFO Channel 02/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897243:897403 [3] NCCL INFO Channel 12/0 : 43[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897242:897400 [2] NCCL INFO Channel 12/0 : 42[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762822:762984 [7] NCCL INFO Channel 01/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897243:897403 [3] NCCL INFO Channel 13/0 : 43[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897242:897400 [2] NCCL INFO Channel 13/0 : 42[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762820:762981 [5] NCCL INFO Channel 03/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762816:762987 [1] NCCL INFO Channel 02/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762822:762984 [7] NCCL INFO Channel 02/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762820:762981 [5] NCCL INFO Channel 04/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762816:762987 [1] NCCL INFO Channel 03/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762822:762984 [7] NCCL INFO Channel 03/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897243:897403 [3] NCCL INFO Channel 14/0 : 43[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897242:897400 [2] NCCL INFO Channel 14/0 : 42[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762820:762981 [5] NCCL INFO Channel 06/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897243:897403 [3] NCCL INFO Channel 15/0 : 43[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897242:897400 [2] NCCL INFO Channel 15/0 : 42[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762816:762987 [1] NCCL INFO Channel 04/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762822:762984 [7] NCCL INFO Channel 04/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762820:762981 [5] NCCL INFO Channel 07/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762816:762987 [1] NCCL INFO Channel 05/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762822:762984 [7] NCCL INFO Channel 05/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762820:762981 [5] NCCL INFO Channel 08/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762816:762987 [1] NCCL INFO Channel 06/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762822:762984 [7] NCCL INFO Channel 06/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762820:762981 [5] NCCL INFO Channel 09/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762816:762987 [1] NCCL INFO Channel 07/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762822:762984 [7] NCCL INFO Channel 08/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762820:762981 [5] NCCL INFO Channel 10/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762816:762987 [1] NCCL INFO Channel 08/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762820:762981 [5] NCCL INFO Channel 11/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762822:762984 [7] NCCL INFO Channel 09/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624950:625109 [7] NCCL INFO Channel 01/0 : 39[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762816:762987 [1] NCCL INFO Channel 10/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762820:762981 [5] NCCL INFO Channel 12/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762816:762987 [1] NCCL INFO Channel 11/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762822:762984 [7] NCCL INFO Channel 10/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624950:625109 [7] NCCL INFO Channel 03/0 : 39[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762820:762981 [5] NCCL INFO Channel 14/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762816:762987 [1] NCCL INFO Channel 12/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762822:762984 [7] NCCL INFO Channel 11/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624948:625104 [5] NCCL INFO Channel 01/0 : 37[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624944:625106 [1] NCCL INFO Channel 01/0 : 33[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762820:762981 [5] NCCL INFO Channel 15/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762816:762987 [1] NCCL INFO Channel 13/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624950:625109 [7] NCCL INFO Channel 05/0 : 39[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624948:625104 [5] NCCL INFO Channel 03/0 : 37[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104436:104596 [7] NCCL INFO Channel 00/0 : 47[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762822:762984 [7] NCCL INFO Channel 12/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762820:762981 [5] NCCL INFO Channel 00/0 : 21[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624944:625106 [1] NCCL INFO Channel 03/0 : 33[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624950:625109 [7] NCCL INFO Channel 07/0 : 39[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104429:104594 [0] NCCL INFO Channel 01/0 : 40[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762816:762987 [1] NCCL INFO Channel 14/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762822:762984 [7] NCCL INFO Channel 13/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762820:762981 [5] NCCL INFO Channel 02/0 : 21[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624948:625104 [5] NCCL INFO Channel 05/0 : 37[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624944:625106 [1] NCCL INFO Channel 05/0 : 33[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624950:625109 [7] NCCL INFO Channel 09/0 : 39[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104436:104596 [7] NCCL INFO Channel 01/0 : 47[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762816:762987 [1] NCCL INFO Channel 15/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762822:762984 [7] NCCL INFO Channel 14/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762820:762981 [5] NCCL INFO Channel 04/0 : 21[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624948:625104 [5] NCCL INFO Channel 07/0 : 37[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104435:104595 [6] NCCL INFO Channel 00/0 : 46[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104434:104591 [5] NCCL INFO Channel 00/0 : 45[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104429:104594 [0] NCCL INFO Channel 02/0 : 40[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104436:104596 [7] NCCL INFO Channel 02/0 : 47[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762816:762987 [1] NCCL INFO Channel 00/0 : 17[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762822:762984 [7] NCCL INFO Channel 00/0 : 23[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762820:762981 [5] NCCL INFO Channel 06/0 : 21[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624944:625106 [1] NCCL INFO Channel 07/0 : 33[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624950:625109 [7] NCCL INFO Channel 11/0 : 39[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104433:104590 [4] NCCL INFO Channel 00/0 : 44[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762819:762982 [4] NCCL INFO Channel 00/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762816:762987 [1] NCCL INFO Channel 02/0 : 17[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762822:762984 [7] NCCL INFO Channel 02/0 : 23[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762820:762981 [5] NCCL INFO Channel 08/0 : 21[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624948:625104 [5] NCCL INFO Channel 09/0 : 37[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624950:625109 [7] NCCL INFO Channel 13/0 : 39[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624944:625106 [1] NCCL INFO Channel 11/0 : 33[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104436:104596 [7] NCCL INFO Channel 03/0 : 47[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762816:762987 [1] NCCL INFO Channel 04/0 : 17[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624948:625104 [5] NCCL INFO Channel 11/0 : 37[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624944:625106 [1] NCCL INFO Channel 13/0 : 33[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104429:104594 [0] NCCL INFO Channel 03/0 : 40[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104435:104595 [6] NCCL INFO Channel 01/0 : 46[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104434:104591 [5] NCCL INFO Channel 01/0 : 45[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762819:762982 [4] NCCL INFO Channel 01/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762822:762984 [7] NCCL INFO Channel 04/0 : 23[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762820:762981 [5] NCCL INFO Channel 10/0 : 21[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762818:762985 [3] NCCL INFO Channel 00/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624948:625104 [5] NCCL INFO Channel 15/0 : 37[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762815:762983 [0] NCCL INFO Channel 01/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762816:762987 [1] NCCL INFO Channel 06/0 : 17[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104432:104592 [3] NCCL INFO Channel 00/0 : 43[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104433:104590 [4] NCCL INFO Channel 01/0 : 44[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104430:104589 [1] NCCL INFO Channel 00/0 : 41[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104436:104596 [7] NCCL INFO Channel 04/0 : 47[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104431:104593 [2] NCCL INFO Channel 00/0 : 42[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104435:104595 [6] NCCL INFO Channel 02/0 : 46[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104429:104594 [0] NCCL INFO Channel 04/0 : 40[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624944:625106 [1] NCCL INFO Channel 15/0 : 33[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762822:762984 [7] NCCL INFO Channel 06/0 : 23[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762819:762982 [4] NCCL INFO Channel 02/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762820:762981 [5] NCCL INFO Channel 12/0 : 21[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762821:762988 [6] NCCL INFO Channel 00/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104434:104591 [5] NCCL INFO Channel 02/0 : 45[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104433:104590 [4] NCCL INFO Channel 02/0 : 44[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104432:104592 [3] NCCL INFO Channel 01/0 : 43[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624950:625109 [7] NCCL INFO Channel 00/0 : 23[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104436:104596 [7] NCCL INFO Channel 05/0 : 47[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104430:104589 [1] NCCL INFO Channel 02/0 : 41[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762818:762985 [3] NCCL INFO Channel 01/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762816:762987 [1] NCCL INFO Channel 08/0 : 17[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762815:762983 [0] NCCL INFO Channel 02/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762822:762984 [7] NCCL INFO Channel 08/0 : 23[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762819:762982 [4] NCCL INFO Channel 03/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762820:762981 [5] NCCL INFO Channel 14/0 : 21[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762821:762988 [6] NCCL INFO Channel 01/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104431:104593 [2] NCCL INFO Channel 01/0 : 42[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624950:625109 [7] NCCL INFO Channel 02/0 : 23[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104435:104595 [6] NCCL INFO Channel 03/0 : 46[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104429:104594 [0] NCCL INFO Channel 05/0 : 40[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104434:104591 [5] NCCL INFO Channel 03/0 : 45[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104433:104590 [4] NCCL INFO Channel 03/0 : 44[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762816:762987 [1] NCCL INFO Channel 10/0 : 17[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762818:762985 [3] NCCL INFO Channel 02/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762815:762983 [0] NCCL INFO Channel 03/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624948:625104 [5] NCCL INFO Channel 00/0 : 21[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762822:762984 [7] NCCL INFO Channel 10/0 : 23[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762819:762982 [4] NCCL INFO Channel 05/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762821:762988 [6] NCCL INFO Channel 02/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104432:104592 [3] NCCL INFO Channel 02/0 : 43[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104436:104596 [7] NCCL INFO Channel 06/0 : 47[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104430:104589 [1] NCCL INFO Channel 03/0 : 41[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104431:104593 [2] NCCL INFO Channel 03/0 : 42[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624950:625109 [7] NCCL INFO Channel 04/0 : 23[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762816:762987 [1] NCCL INFO Channel 12/0 : 17[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624944:625106 [1] NCCL INFO Channel 00/0 : 17[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762817:762986 [2] NCCL INFO Channel 00/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104435:104595 [6] NCCL INFO Channel 04/0 : 46[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104429:104594 [0] NCCL INFO Channel 06/0 : 40[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104434:104591 [5] NCCL INFO Channel 04/0 : 45[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104433:104590 [4] NCCL INFO Channel 05/0 : 44[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104432:104592 [3] NCCL INFO Channel 04/0 : 43[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104436:104596 [7] NCCL INFO Channel 08/0 : 47[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104431:104593 [2] NCCL INFO Channel 04/0 : 42[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624948:625104 [5] NCCL INFO Channel 02/0 : 21[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762818:762985 [3] NCCL INFO Channel 04/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624950:625109 [7] NCCL INFO Channel 06/0 : 23[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762822:762984 [7] NCCL INFO Channel 12/0 : 23[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624944:625106 [1] NCCL INFO Channel 02/0 : 17[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762815:762983 [0] NCCL INFO Channel 04/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762816:762987 [1] NCCL INFO Channel 14/0 : 17[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762819:762982 [4] NCCL INFO Channel 06/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762821:762988 [6] NCCL INFO Channel 03/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104435:104595 [6] NCCL INFO Channel 05/0 : 46[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624948:625104 [5] NCCL INFO Channel 04/0 : 21[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624944:625106 [1] NCCL INFO Channel 04/0 : 17[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762817:762986 [2] NCCL INFO Channel 01/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762818:762985 [3] NCCL INFO Channel 05/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762822:762984 [7] NCCL INFO Channel 14/0 : 23[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104429:104594 [0] NCCL INFO Channel 07/0 : 40[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104434:104591 [5] NCCL INFO Channel 06/0 : 45[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762815:762983 [0] NCCL INFO Channel 05/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104433:104590 [4] NCCL INFO Channel 06/0 : 44[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104430:104589 [1] NCCL INFO Channel 04/0 : 41[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104432:104592 [3] NCCL INFO Channel 05/0 : 43[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104436:104596 [7] NCCL INFO Channel 09/0 : 47[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104431:104593 [2] NCCL INFO Channel 05/0 : 42[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762819:762982 [4] NCCL INFO Channel 07/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762821:762988 [6] NCCL INFO Channel 04/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624950:625109 [7] NCCL INFO Channel 08/0 : 23[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624948:625104 [5] NCCL INFO Channel 06/0 : 21[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104435:104595 [6] NCCL INFO Channel 07/0 : 46[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762818:762985 [3] NCCL INFO Channel 06/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762817:762986 [2] NCCL INFO Channel 03/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762815:762983 [0] NCCL INFO Channel 06/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624944:625106 [1] NCCL INFO Channel 06/0 : 17[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624950:625109 [7] NCCL INFO Channel 10/0 : 23[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624948:625104 [5] NCCL INFO Channel 08/0 : 21[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104429:104594 [0] NCCL INFO Channel 09/0 : 40[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624944:625106 [1] NCCL INFO Channel 08/0 : 17[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104433:104590 [4] NCCL INFO Channel 07/0 : 44[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104434:104591 [5] NCCL INFO Channel 07/0 : 45[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624950:625109 [7] NCCL INFO Channel 12/0 : 23[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104436:104596 [7] NCCL INFO Channel 10/0 : 47[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104430:104589 [1] NCCL INFO Channel 05/0 : 41[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104432:104592 [3] NCCL INFO Channel 06/0 : 43[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762821:762988 [6] NCCL INFO Channel 05/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762819:762982 [4] NCCL INFO Channel 08/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104431:104593 [2] NCCL INFO Channel 06/0 : 42[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762818:762985 [3] NCCL INFO Channel 07/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762817:762986 [2] NCCL INFO Channel 04/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104435:104595 [6] NCCL INFO Channel 08/0 : 46[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624948:625104 [5] NCCL INFO Channel 10/0 : 21[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624944:625106 [1] NCCL INFO Channel 10/0 : 17[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624950:625109 [7] NCCL INFO Channel 14/0 : 23[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104429:104594 [0] NCCL INFO Channel 10/0 : 40[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104433:104590 [4] NCCL INFO Channel 08/0 : 44[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762815:762983 [0] NCCL INFO Channel 07/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762821:762988 [6] NCCL INFO Channel 07/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762819:762982 [4] NCCL INFO Channel 09/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762818:762985 [3] NCCL INFO Channel 08/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762817:762986 [2] NCCL INFO Channel 05/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624948:625104 [5] NCCL INFO Channel 12/0 : 21[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624944:625106 [1] NCCL INFO Channel 12/0 : 17[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104436:104596 [7] NCCL INFO Channel 11/0 : 47[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624950:625109 [7] NCCL INFO Channel 00/0 : 39[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104434:104591 [5] NCCL INFO Channel 08/0 : 45[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104432:104592 [3] NCCL INFO Channel 07/0 : 43[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104435:104595 [6] NCCL INFO Channel 09/0 : 46[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104430:104589 [1] NCCL INFO Channel 06/0 : 41[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104431:104593 [2] NCCL INFO Channel 07/0 : 42[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762815:762983 [0] NCCL INFO Channel 09/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762821:762988 [6] NCCL INFO Channel 08/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762819:762982 [4] NCCL INFO Channel 10/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624949:625110 [6] NCCL INFO Channel 01/0 : 38[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624948:625104 [5] NCCL INFO Channel 14/0 : 21[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104429:104594 [0] NCCL INFO Channel 11/0 : 40[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104433:104590 [4] NCCL INFO Channel 09/0 : 44[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624947:625105 [4] NCCL INFO Channel 01/0 : 36[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104436:104596 [7] NCCL INFO Channel 12/0 : 47[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624944:625106 [1] NCCL INFO Channel 14/0 : 17[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762818:762985 [3] NCCL INFO Channel 09/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762815:762983 [0] NCCL INFO Channel 10/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762817:762986 [2] NCCL INFO Channel 06/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762821:762988 [6] NCCL INFO Channel 09/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624950:625109 [7] NCCL INFO Channel 02/0 : 39[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624946:625107 [3] NCCL INFO Channel 01/0 : 35[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624949:625110 [6] NCCL INFO Channel 03/0 : 38[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104434:104591 [5] NCCL INFO Channel 09/0 : 45[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104432:104592 [3] NCCL INFO Channel 08/0 : 43[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624948:625104 [5] NCCL INFO Channel 00/0 : 37[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624943:625108 [0] NCCL INFO Channel 01/0 : 32[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624947:625105 [4] NCCL INFO Channel 03/0 : 36[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762818:762985 [3] NCCL INFO Channel 10/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762819:762982 [4] NCCL INFO Channel 11/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762815:762983 [0] NCCL INFO Channel 11/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762817:762986 [2] NCCL INFO Channel 07/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762821:762988 [6] NCCL INFO Channel 10/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104435:104595 [6] NCCL INFO Channel 10/0 : 46[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104431:104593 [2] NCCL INFO Channel 08/0 : 42[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104433:104590 [4] NCCL INFO Channel 10/0 : 44[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104430:104589 [1] NCCL INFO Channel 07/0 : 41[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104429:104594 [0] NCCL INFO Channel 12/0 : 40[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104436:104596 [7] NCCL INFO Channel 13/0 : 47[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624944:625106 [1] NCCL INFO Channel 00/0 : 33[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624950:625109 [7] NCCL INFO Channel 04/0 : 39[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762818:762985 [3] NCCL INFO Channel 12/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624945:625111 [2] NCCL INFO Channel 01/0 : 34[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762815:762983 [0] NCCL INFO Channel 12/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762819:762982 [4] NCCL INFO Channel 13/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104434:104591 [5] NCCL INFO Channel 10/0 : 45[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624946:625107 [3] NCCL INFO Channel 03/0 : 35[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624948:625104 [5] NCCL INFO Channel 02/0 : 37[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624949:625110 [6] NCCL INFO Channel 05/0 : 38[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624947:625105 [4] NCCL INFO Channel 05/0 : 36[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624943:625108 [0] NCCL INFO Channel 03/0 : 32[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762821:762988 [6] NCCL INFO Channel 11/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624944:625106 [1] NCCL INFO Channel 02/0 : 33[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624950:625109 [7] NCCL INFO Channel 06/0 : 39[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104435:104595 [6] NCCL INFO Channel 11/0 : 46[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762817:762986 [2] NCCL INFO Channel 08/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624945:625111 [2] NCCL INFO Channel 03/0 : 34[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104432:104592 [3] NCCL INFO Channel 09/0 : 43[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762818:762985 [3] NCCL INFO Channel 13/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624948:625104 [5] NCCL INFO Channel 04/0 : 37[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104431:104593 [2] NCCL INFO Channel 09/0 : 42[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104433:104590 [4] NCCL INFO Channel 11/0 : 44[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104436:104596 [7] NCCL INFO Channel 14/0 : 47[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104430:104589 [1] NCCL INFO Channel 08/0 : 41[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104429:104594 [0] NCCL INFO Channel 13/0 : 40[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104434:104591 [5] NCCL INFO Channel 11/0 : 45[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762815:762983 [0] NCCL INFO Channel 13/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624946:625107 [3] NCCL INFO Channel 05/0 : 35[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624949:625110 [6] NCCL INFO Channel 07/0 : 38[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104435:104595 [6] NCCL INFO Channel 12/0 : 46[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762819:762982 [4] NCCL INFO Channel 14/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624947:625105 [4] NCCL INFO Channel 07/0 : 36[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762821:762988 [6] NCCL INFO Channel 12/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624944:625106 [1] NCCL INFO Channel 04/0 : 33[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762817:762986 [2] NCCL INFO Channel 09/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624950:625109 [7] NCCL INFO Channel 08/0 : 39[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762818:762985 [3] NCCL INFO Channel 14/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624943:625108 [0] NCCL INFO Channel 05/0 : 32[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762815:762983 [0] NCCL INFO Channel 14/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104432:104592 [3] NCCL INFO Channel 10/0 : 43[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104431:104593 [2] NCCL INFO Channel 11/0 : 42[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624945:625111 [2] NCCL INFO Channel 05/0 : 34[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624948:625104 [5] NCCL INFO Channel 06/0 : 37[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624949:625110 [6] NCCL INFO Channel 09/0 : 38[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762819:762982 [4] NCCL INFO Channel 15/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624946:625107 [3] NCCL INFO Channel 07/0 : 35[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104436:104596 [7] NCCL INFO Channel 00/0 : 55[7] -> 63[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762821:762988 [6] NCCL INFO Channel 13/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624947:625105 [4] NCCL INFO Channel 09/0 : 36[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104433:104590 [4] NCCL INFO Channel 13/0 : 44[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104430:104589 [1] NCCL INFO Channel 10/0 : 41[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104429:104594 [0] NCCL INFO Channel 14/0 : 40[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762817:762986 [2] NCCL INFO Channel 11/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762818:762985 [3] NCCL INFO Channel 15/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624944:625106 [1] NCCL INFO Channel 06/0 : 33[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624950:625109 [7] NCCL INFO Channel 10/0 : 39[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104434:104591 [5] NCCL INFO Channel 12/0 : 45[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104435:104595 [6] NCCL INFO Channel 13/0 : 46[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762815:762983 [0] NCCL INFO Channel 15/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104432:104592 [3] NCCL INFO Channel 12/0 : 43[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762819:762982 [4] NCCL INFO Channel 00/0 : 20[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762821:762988 [6] NCCL INFO Channel 15/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624943:625108 [0] NCCL INFO Channel 07/0 : 32[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624948:625104 [5] NCCL INFO Channel 08/0 : 37[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624949:625110 [6] NCCL INFO Channel 11/0 : 38[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104431:104593 [2] NCCL INFO Channel 12/0 : 42[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624945:625111 [2] NCCL INFO Channel 07/0 : 34[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104436:104596 [7] NCCL INFO Channel 02/0 : 55[7] -> 63[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762818:762985 [3] NCCL INFO Channel 00/0 : 19[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624946:625107 [3] NCCL INFO Channel 09/0 : 35[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762815:762983 [0] NCCL INFO Channel 02/0 : 16[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624947:625105 [4] NCCL INFO Channel 11/0 : 36[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762817:762986 [2] NCCL INFO Channel 12/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762819:762982 [4] NCCL INFO Channel 02/0 : 20[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762821:762988 [6] NCCL INFO Channel 00/0 : 22[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762818:762985 [3] NCCL INFO Channel 02/0 : 19[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104433:104590 [4] NCCL INFO Channel 14/0 : 44[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104429:104594 [0] NCCL INFO Channel 15/0 : 40[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624950:625109 [7] NCCL INFO Channel 12/0 : 39[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624944:625106 [1] NCCL INFO Channel 08/0 : 33[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104430:104589 [1] NCCL INFO Channel 11/0 : 41[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624948:625104 [5] NCCL INFO Channel 10/0 : 37[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104434:104591 [5] NCCL INFO Channel 14/0 : 45[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104435:104595 [6] NCCL INFO Channel 15/0 : 46[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104432:104592 [3] NCCL INFO Channel 13/0 : 43[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624949:625110 [6] NCCL INFO Channel 13/0 : 38[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104431:104593 [2] NCCL INFO Channel 13/0 : 42[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624945:625111 [2] NCCL INFO Channel 09/0 : 34[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104436:104596 [7] NCCL INFO Channel 04/0 : 55[7] -> 63[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762815:762983 [0] NCCL INFO Channel 04/0 : 16[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762817:762986 [2] NCCL INFO Channel 13/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762819:762982 [4] NCCL INFO Channel 06/0 : 20[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762821:762988 [6] NCCL INFO Channel 02/0 : 22[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762818:762985 [3] NCCL INFO Channel 04/0 : 19[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762815:762983 [0] NCCL INFO Channel 06/0 : 16[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624946:625107 [3] NCCL INFO Channel 13/0 : 35[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624947:625105 [4] NCCL INFO Channel 13/0 : 36[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104433:104590 [4] NCCL INFO Channel 15/0 : 44[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624943:625108 [0] NCCL INFO Channel 09/0 : 32[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624950:625109 [7] NCCL INFO Channel 14/0 : 39[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624944:625106 [1] NCCL INFO Channel 10/0 : 33[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104429:104594 [0] NCCL INFO Channel 02/0 : 48[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624948:625104 [5] NCCL INFO Channel 12/0 : 37[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624949:625110 [6] NCCL INFO Channel 15/0 : 38[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624945:625111 [2] NCCL INFO Channel 11/0 : 34[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104430:104589 [1] NCCL INFO Channel 12/0 : 41[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104435:104595 [6] NCCL INFO Channel 00/0 : 54[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104434:104591 [5] NCCL INFO Channel 15/0 : 45[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104432:104592 [3] NCCL INFO Channel 14/0 : 43[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624947:625105 [4] NCCL INFO Channel 15/0 : 36[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624946:625107 [3] NCCL INFO Channel 15/0 : 35[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104436:104596 [7] NCCL INFO Channel 06/0 : 55[7] -> 63[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104431:104593 [2] NCCL INFO Channel 14/0 : 42[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104433:104590 [4] NCCL INFO Channel 00/0 : 52[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104429:104594 [0] NCCL INFO Channel 04/0 : 48[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624948:625104 [5] NCCL INFO Channel 14/0 : 37[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624944:625106 [1] NCCL INFO Channel 12/0 : 33[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624945:625111 [2] NCCL INFO Channel 13/0 : 34[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624943:625108 [0] NCCL INFO Channel 11/0 : 32[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104430:104589 [1] NCCL INFO Channel 13/0 : 41[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104435:104595 [6] NCCL INFO Channel 02/0 : 54[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624945:625111 [2] NCCL INFO Channel 15/0 : 34[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624944:625106 [1] NCCL INFO Channel 14/0 : 33[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104432:104592 [3] NCCL INFO Channel 15/0 : 43[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624943:625108 [0] NCCL INFO Channel 13/0 : 32[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104434:104591 [5] NCCL INFO Channel 00/0 : 53[5] -> 61[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104436:104596 [7] NCCL INFO Channel 08/0 : 55[7] -> 63[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104431:104593 [2] NCCL INFO Channel 15/0 : 42[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104433:104590 [4] NCCL INFO Channel 02/0 : 52[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104429:104594 [0] NCCL INFO Channel 06/0 : 48[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624949:625110 [6] NCCL INFO Channel 00/0 : 22[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624943:625108 [0] NCCL INFO Channel 15/0 : 32[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624946:625107 [3] NCCL INFO Channel 00/0 : 19[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104435:104595 [6] NCCL INFO Channel 04/0 : 54[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104430:104589 [1] NCCL INFO Channel 14/0 : 41[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104432:104592 [3] NCCL INFO Channel 00/0 : 51[3] -> 59[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104434:104591 [5] NCCL INFO Channel 02/0 : 53[5] -> 61[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104436:104596 [7] NCCL INFO Channel 10/0 : 55[7] -> 63[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624947:625105 [4] NCCL INFO Channel 00/0 : 20[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624949:625110 [6] NCCL INFO Channel 02/0 : 22[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624946:625107 [3] NCCL INFO Channel 02/0 : 19[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104433:104590 [4] NCCL INFO Channel 06/0 : 52[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104429:104594 [0] NCCL INFO Channel 08/0 : 48[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104431:104593 [2] NCCL INFO Channel 00/0 : 50[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104435:104595 [6] NCCL INFO Channel 08/0 : 54[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104430:104589 [1] NCCL INFO Channel 15/0 : 41[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104432:104592 [3] NCCL INFO Channel 02/0 : 51[3] -> 59[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104434:104591 [5] NCCL INFO Channel 04/0 : 53[5] -> 61[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104436:104596 [7] NCCL INFO Channel 12/0 : 55[7] -> 63[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624947:625105 [4] NCCL INFO Channel 02/0 : 20[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624949:625110 [6] NCCL INFO Channel 04/0 : 22[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624945:625111 [2] NCCL INFO Channel 00/0 : 18[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624946:625107 [3] NCCL INFO Channel 04/0 : 19[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104433:104590 [4] NCCL INFO Channel 08/0 : 52[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624947:625105 [4] NCCL INFO Channel 06/0 : 20[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624949:625110 [6] NCCL INFO Channel 08/0 : 22[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624945:625111 [2] NCCL INFO Channel 04/0 : 18[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624946:625107 [3] NCCL INFO Channel 06/0 : 19[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104431:104593 [2] NCCL INFO Channel 04/0 : 50[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104435:104595 [6] NCCL INFO Channel 10/0 : 54[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104430:104589 [1] NCCL INFO Channel 00/0 : 49[1] -> 57[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9629:9793 [4] NCCL INFO Channel 01/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624947:625105 [4] NCCL INFO Channel 08/0 : 20[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624949:625110 [6] NCCL INFO Channel 10/0 : 22[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624945:625111 [2] NCCL INFO Channel 06/0 : 18[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104434:104591 [5] NCCL INFO Channel 06/0 : 53[5] -> 61[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104436:104596 [7] NCCL INFO Channel 14/0 : 55[7] -> 63[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104432:104592 [3] NCCL INFO Channel 04/0 : 51[3] -> 59[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104433:104590 [4] NCCL INFO Channel 10/0 : 52[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104429:104594 [0] NCCL INFO Channel 10/0 : 48[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624943:625108 [0] NCCL INFO Channel 02/0 : 16[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624946:625107 [3] NCCL INFO Channel 08/0 : 19[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9631:9788 [6] NCCL INFO Channel 01/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104431:104593 [2] NCCL INFO Channel 06/0 : 50[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104435:104595 [6] NCCL INFO Channel 12/0 : 54[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104430:104589 [1] NCCL INFO Channel 02/0 : 49[1] -> 57[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624947:625105 [4] NCCL INFO Channel 10/0 : 20[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624949:625110 [6] NCCL INFO Channel 12/0 : 22[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624945:625111 [2] NCCL INFO Channel 08/0 : 18[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624946:625107 [3] NCCL INFO Channel 10/0 : 19[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624943:625108 [0] NCCL INFO Channel 04/0 : 16[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9625:9787 [0] NCCL INFO Channel 01/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9629:9793 [4] NCCL INFO Channel 03/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104434:104591 [5] NCCL INFO Channel 08/0 : 53[5] -> 61[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104433:104590 [4] NCCL INFO Channel 12/0 : 52[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104432:104592 [3] NCCL INFO Channel 06/0 : 51[3] -> 59[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104429:104594 [0] NCCL INFO Channel 12/0 : 48[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104431:104593 [2] NCCL INFO Channel 08/0 : 50[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104435:104595 [6] NCCL INFO Channel 14/0 : 54[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104430:104589 [1] NCCL INFO Channel 04/0 : 49[1] -> 57[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9631:9788 [6] NCCL INFO Channel 03/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9625:9787 [0] NCCL INFO Channel 03/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9629:9793 [4] NCCL INFO Channel 05/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104433:104590 [4] NCCL INFO Channel 14/0 : 52[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104434:104591 [5] NCCL INFO Channel 10/0 : 53[5] -> 61[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104432:104592 [3] NCCL INFO Channel 08/0 : 51[3] -> 59[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9631:9788 [6] NCCL INFO Channel 05/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104429:104594 [0] NCCL INFO Channel 14/0 : 48[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9625:9787 [0] NCCL INFO Channel 05/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9629:9793 [4] NCCL INFO Channel 07/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104431:104593 [2] NCCL INFO Channel 10/0 : 50[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104430:104589 [1] NCCL INFO Channel 06/0 : 49[1] -> 57[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897247:897404 [7] NCCL INFO Channel 01/0 : 31[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104434:104591 [5] NCCL INFO Channel 12/0 : 53[5] -> 61[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9631:9788 [6] NCCL INFO Channel 07/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104432:104592 [3] NCCL INFO Channel 10/0 : 51[3] -> 59[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9625:9787 [0] NCCL INFO Channel 07/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104430:104589 [1] NCCL INFO Channel 08/0 : 49[1] -> 57[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104431:104593 [2] NCCL INFO Channel 12/0 : 50[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9629:9793 [4] NCCL INFO Channel 09/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897247:897404 [7] NCCL INFO Channel 03/0 : 31[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104434:104591 [5] NCCL INFO Channel 14/0 : 53[5] -> 61[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104430:104589 [1] NCCL INFO Channel 10/0 : 49[1] -> 57[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9631:9788 [6] NCCL INFO Channel 09/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9625:9787 [0] NCCL INFO Channel 09/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9629:9793 [4] NCCL INFO Channel 11/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897246:897405 [6] NCCL INFO Channel 01/0 : 30[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897247:897404 [7] NCCL INFO Channel 05/0 : 31[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104431:104593 [2] NCCL INFO Channel 14/0 : 50[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104436:104596 [7] NCCL INFO Channel 00/0 : 39[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9631:9788 [6] NCCL INFO Channel 11/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9625:9787 [0] NCCL INFO Channel 11/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104432:104592 [3] NCCL INFO Channel 12/0 : 51[3] -> 59[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9629:9793 [4] NCCL INFO Channel 13/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897244:897407 [4] NCCL INFO Channel 01/0 : 28[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-167-9:1039544:1039702 [7] NCCL INFO Channel 01/0 : 31[7] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104430:104589 [1] NCCL INFO Channel 12/0 : 49[1] -> 57[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104436:104596 [7] NCCL INFO Channel 02/0 : 39[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9625:9787 [0] NCCL INFO Channel 13/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9631:9788 [6] NCCL INFO Channel 13/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104432:104592 [3] NCCL INFO Channel 14/0 : 51[3] -> 59[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9629:9793 [4] NCCL INFO Channel 15/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897240:897406 [0] NCCL INFO Channel 01/0 : 24[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897247:897404 [7] NCCL INFO Channel 07/0 : 31[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897246:897405 [6] NCCL INFO Channel 03/0 : 30[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-167-9:1039544:1039702 [7] NCCL INFO Channel 03/0 : 31[7] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104430:104589 [1] NCCL INFO Channel 14/0 : 49[1] -> 57[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104435:104595 [6] NCCL INFO Channel 00/0 : 38[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9625:9787 [0] NCCL INFO Channel 15/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9631:9788 [6] NCCL INFO Channel 15/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104436:104596 [7] NCCL INFO Channel 04/0 : 39[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897244:897407 [4] NCCL INFO Channel 03/0 : 28[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897240:897406 [0] NCCL INFO Channel 03/0 : 24[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9629:9793 [4] NCCL INFO Channel 00/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO Channel 01/0 : 28[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104433:104590 [4] NCCL INFO Channel 00/0 : 36[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104429:104594 [0] NCCL INFO Channel 02/0 : 32[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104435:104595 [6] NCCL INFO Channel 02/0 : 38[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104436:104596 [7] NCCL INFO Channel 06/0 : 39[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897247:897404 [7] NCCL INFO Channel 09/0 : 31[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897246:897405 [6] NCCL INFO Channel 05/0 : 30[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897245:897401 [5] NCCL INFO Channel 01/0 : 29[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9625:9787 [0] NCCL INFO Channel 01/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897244:897407 [4] NCCL INFO Channel 05/0 : 28[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9631:9788 [6] NCCL INFO Channel 00/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9629:9793 [4] NCCL INFO Channel 01/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-167-9:1039543:1039699 [6] NCCL INFO Channel 01/0 : 30[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-167-9:1039544:1039702 [7] NCCL INFO Channel 05/0 : 31[7] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104433:104590 [4] NCCL INFO Channel 02/0 : 36[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-167-9:1039537:1039703 [0] NCCL INFO Channel 01/0 : 24[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104429:104594 [0] NCCL INFO Channel 04/0 : 32[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104435:104595 [6] NCCL INFO Channel 04/0 : 38[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897240:897406 [0] NCCL INFO Channel 05/0 : 24[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897242:897400 [2] NCCL INFO Channel 01/0 : 26[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897247:897404 [7] NCCL INFO Channel 11/0 : 31[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9625:9787 [0] NCCL INFO Channel 02/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9631:9788 [6] NCCL INFO Channel 01/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9629:9793 [4] NCCL INFO Channel 02/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO Channel 03/0 : 28[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-167-9:1039543:1039699 [6] NCCL INFO Channel 03/0 : 30[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104436:104596 [7] NCCL INFO Channel 08/0 : 39[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104434:104591 [5] NCCL INFO Channel 00/0 : 37[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-167-9:1039544:1039702 [7] NCCL INFO Channel 07/0 : 31[7] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104433:104590 [4] NCCL INFO Channel 06/0 : 36[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897246:897405 [6] NCCL INFO Channel 07/0 : 30[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-167-9:1039542:1039701 [5] NCCL INFO Channel 01/0 : 29[5] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104429:104594 [0] NCCL INFO Channel 06/0 : 32[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897245:897401 [5] NCCL INFO Channel 03/0 : 29[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9625:9787 [0] NCCL INFO Channel 03/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-167-9:1039537:1039703 [0] NCCL INFO Channel 03/0 : 24[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104431:104593 [2] NCCL INFO Channel 00/0 : 34[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104435:104595 [6] NCCL INFO Channel 08/0 : 38[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897244:897407 [4] NCCL INFO Channel 07/0 : 28[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9631:9788 [6] NCCL INFO Channel 02/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9629:9793 [4] NCCL INFO Channel 03/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897240:897406 [0] NCCL INFO Channel 07/0 : 24[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897242:897400 [2] NCCL INFO Channel 03/0 : 26[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104436:104596 [7] NCCL INFO Channel 10/0 : 39[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104434:104591 [5] NCCL INFO Channel 02/0 : 37[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104433:104590 [4] NCCL INFO Channel 08/0 : 36[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-167-9:1039539:1039698 [2] NCCL INFO Channel 01/0 : 26[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO Channel 05/0 : 28[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104429:104594 [0] NCCL INFO Channel 08/0 : 32[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9625:9787 [0] NCCL INFO Channel 04/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-167-9:1039543:1039699 [6] NCCL INFO Channel 05/0 : 30[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104431:104593 [2] NCCL INFO Channel 04/0 : 34[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104435:104595 [6] NCCL INFO Channel 10/0 : 38[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897243:897403 [3] NCCL INFO Channel 01/0 : 27[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897247:897404 [7] NCCL INFO Channel 13/0 : 31[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9631:9788 [6] NCCL INFO Channel 03/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9629:9793 [4] NCCL INFO Channel 05/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-167-9:1039544:1039702 [7] NCCL INFO Channel 09/0 : 31[7] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897246:897405 [6] NCCL INFO Channel 09/0 : 30[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-167-9:1039542:1039701 [5] NCCL INFO Channel 03/0 : 29[5] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897241:897402 [1] NCCL INFO Channel 01/0 : 25[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897245:897401 [5] NCCL INFO Channel 05/0 : 29[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897244:897407 [4] NCCL INFO Channel 09/0 : 28[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-167-9:1039537:1039703 [0] NCCL INFO Channel 05/0 : 24[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897240:897406 [0] NCCL INFO Channel 09/0 : 24[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897242:897400 [2] NCCL INFO Channel 05/0 : 26[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9625:9787 [0] NCCL INFO Channel 05/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9631:9788 [6] NCCL INFO Channel 04/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9629:9793 [4] NCCL INFO Channel 06/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9625:9787 [0] NCCL INFO Channel 06/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO Channel 07/0 : 28[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897243:897403 [3] NCCL INFO Channel 03/0 : 27[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897246:897405 [6] NCCL INFO Channel 11/0 : 30[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-167-9:1039539:1039698 [2] NCCL INFO Channel 03/0 : 26[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897245:897401 [5] NCCL INFO Channel 07/0 : 29[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897244:897407 [4] NCCL INFO Channel 11/0 : 28[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-167-9:1039540:1039705 [3] NCCL INFO Channel 01/0 : 27[3] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897241:897402 [1] NCCL INFO Channel 03/0 : 25[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9631:9788 [6] NCCL INFO Channel 05/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9629:9793 [4] NCCL INFO Channel 07/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9625:9787 [0] NCCL INFO Channel 07/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-167-9:1039544:1039702 [7] NCCL INFO Channel 11/0 : 31[7] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-167-9:1039537:1039703 [0] NCCL INFO Channel 07/0 : 24[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-167-9:1039543:1039699 [6] NCCL INFO Channel 07/0 : 30[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-167-9:1039542:1039701 [5] NCCL INFO Channel 05/0 : 29[5] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-167-9:1039538:1039704 [1] NCCL INFO Channel 01/0 : 25[1] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897242:897400 [2] NCCL INFO Channel 07/0 : 26[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897240:897406 [0] NCCL INFO Channel 11/0 : 24[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO Channel 09/0 : 28[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897246:897405 [6] NCCL INFO Channel 13/0 : 30[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-167-9:1039539:1039698 [2] NCCL INFO Channel 05/0 : 26[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897243:897403 [3] NCCL INFO Channel 05/0 : 27[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-167-9:1039540:1039705 [3] NCCL INFO Channel 03/0 : 27[3] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897245:897401 [5] NCCL INFO Channel 09/0 : 29[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9631:9788 [6] NCCL INFO Channel 07/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897244:897407 [4] NCCL INFO Channel 13/0 : 28[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-167-9:1039544:1039702 [7] NCCL INFO Channel 13/0 : 31[7] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-167-9:1039543:1039699 [6] NCCL INFO Channel 09/0 : 30[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9625:9787 [0] NCCL INFO Channel 09/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9629:9793 [4] NCCL INFO Channel 08/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897241:897402 [1] NCCL INFO Channel 05/0 : 25[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9631:9788 [6] NCCL INFO Channel 08/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897246:897405 [6] NCCL INFO Channel 15/0 : 30[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897240:897406 [0] NCCL INFO Channel 13/0 : 24[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897242:897400 [2] NCCL INFO Channel 09/0 : 26[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897243:897403 [3] NCCL INFO Channel 07/0 : 27[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-167-9:1039537:1039703 [0] NCCL INFO Channel 09/0 : 24[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-167-9:1039538:1039704 [1] NCCL INFO Channel 03/0 : 25[1] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-167-9:1039542:1039701 [5] NCCL INFO Channel 07/0 : 29[5] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9625:9787 [0] NCCL INFO Channel 10/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9629:9793 [4] NCCL INFO Channel 09/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-167-9:1039539:1039698 [2] NCCL INFO Channel 07/0 : 26[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO Channel 11/0 : 28[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9631:9788 [6] NCCL INFO Channel 09/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-167-9:1039540:1039705 [3] NCCL INFO Channel 05/0 : 27[3] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-167-9:1039544:1039702 [7] NCCL INFO Channel 01/0 : 63[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897245:897401 [5] NCCL INFO Channel 11/0 : 29[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897244:897407 [4] NCCL INFO Channel 15/0 : 28[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897241:897402 [1] NCCL INFO Channel 07/0 : 25[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897240:897406 [0] NCCL INFO Channel 15/0 : 24[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897242:897400 [2] NCCL INFO Channel 11/0 : 26[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897243:897403 [3] NCCL INFO Channel 09/0 : 27[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897245:897401 [5] NCCL INFO Channel 15/0 : 29[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897241:897402 [1] NCCL INFO Channel 11/0 : 25[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9625:9787 [0] NCCL INFO Channel 11/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-167-9:1039543:1039699 [6] NCCL INFO Channel 11/0 : 30[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-167-9:1039537:1039703 [0] NCCL INFO Channel 11/0 : 24[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9631:9788 [6] NCCL INFO Channel 10/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9629:9793 [4] NCCL INFO Channel 10/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-167-9:1039538:1039704 [1] NCCL INFO Channel 05/0 : 25[1] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-167-9:1039542:1039701 [5] NCCL INFO Channel 09/0 : 29[5] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-167-9:1039539:1039698 [2] NCCL INFO Channel 09/0 : 26[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9625:9787 [0] NCCL INFO Channel 12/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO Channel 13/0 : 28[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9629:9793 [4] NCCL INFO Channel 11/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-167-9:1039544:1039702 [7] NCCL INFO Channel 03/0 : 63[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9631:9788 [6] NCCL INFO Channel 11/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9626:9790 [1] NCCL INFO Channel 01/0 : 1[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-167-9:1039540:1039705 [3] NCCL INFO Channel 07/0 : 27[3] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-167-9:1039537:1039703 [0] NCCL INFO Channel 13/0 : 24[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-167-9:1039543:1039699 [6] NCCL INFO Channel 13/0 : 30[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-167-9:1039538:1039704 [1] NCCL INFO Channel 07/0 : 25[1] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-167-9:1039542:1039701 [5] NCCL INFO Channel 11/0 : 29[5] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9625:9787 [0] NCCL INFO Channel 13/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9627:9791 [2] NCCL INFO Channel 01/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9630:9789 [5] NCCL INFO Channel 01/0 : 5[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9629:9793 [4] NCCL INFO Channel 13/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9631:9788 [6] NCCL INFO Channel 12/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-167-9:1039539:1039698 [2] NCCL INFO Channel 11/0 : 26[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO Channel 15/0 : 28[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-167-9:1039544:1039702 [7] NCCL INFO Channel 05/0 : 63[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-167-9:1039540:1039705 [3] NCCL INFO Channel 09/0 : 27[3] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-167-9:1039537:1039703 [0] NCCL INFO Channel 15/0 : 24[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-167-9:1039543:1039699 [6] NCCL INFO Channel 15/0 : 30[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-167-9:1039538:1039704 [1] NCCL INFO Channel 11/0 : 25[1] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-167-9:1039542:1039701 [5] NCCL INFO Channel 15/0 : 29[5] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9625:9787 [0] NCCL INFO Channel 14/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9626:9790 [1] NCCL INFO Channel 03/0 : 1[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-167-9:1039539:1039698 [2] NCCL INFO Channel 13/0 : 26[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9627:9791 [2] NCCL INFO Channel 03/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9630:9789 [5] NCCL INFO Channel 03/0 : 5[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO Channel 01/0 : 60[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9629:9793 [4] NCCL INFO Channel 14/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9631:9788 [6] NCCL INFO Channel 13/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9625:9787 [0] NCCL INFO Channel 15/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-167-9:1039544:1039702 [7] NCCL INFO Channel 07/0 : 63[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9628:9786 [3] NCCL INFO Channel 01/0 : 3[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-167-9:1039540:1039705 [3] NCCL INFO Channel 13/0 : 27[3] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9626:9790 [1] NCCL INFO Channel 05/0 : 1[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-167-9:1039537:1039703 [0] NCCL INFO Channel 01/0 : 56[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9627:9791 [2] NCCL INFO Channel 05/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-167-9:1039538:1039704 [1] NCCL INFO Channel 13/0 : 25[1] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-167-9:1039543:1039699 [6] NCCL INFO Channel 01/0 : 62[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-167-9:1039542:1039701 [5] NCCL INFO Channel 01/0 : 61[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9630:9789 [5] NCCL INFO Channel 05/0 : 5[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9632:9792 [7] NCCL INFO Channel 01/0 : 7[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9631:9788 [6] NCCL INFO Channel 15/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-167-9:1039539:1039698 [2] NCCL INFO Channel 15/0 : 26[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO Channel 03/0 : 60[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9629:9793 [4] NCCL INFO Channel 15/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-167-9:1039544:1039702 [7] NCCL INFO Channel 09/0 : 63[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9628:9786 [3] NCCL INFO Channel 03/0 : 3[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-167-9:1039537:1039703 [0] NCCL INFO Channel 03/0 : 56[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-167-9:1039540:1039705 [3] NCCL INFO Channel 15/0 : 27[3] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-167-9:1039538:1039704 [1] NCCL INFO Channel 15/0 : 25[1] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-167-9:1039543:1039699 [6] NCCL INFO Channel 03/0 : 62[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9627:9791 [2] NCCL INFO Channel 07/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-167-9:1039542:1039701 [5] NCCL INFO Channel 03/0 : 61[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9626:9790 [1] NCCL INFO Channel 07/0 : 1[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9630:9789 [5] NCCL INFO Channel 07/0 : 5[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9632:9792 [7] NCCL INFO Channel 03/0 : 7[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO Channel 05/0 : 60[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-167-9:1039539:1039698 [2] NCCL INFO Channel 01/0 : 58[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-167-9:1039544:1039702 [7] NCCL INFO Channel 11/0 : 63[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9628:9786 [3] NCCL INFO Channel 05/0 : 3[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-167-9:1039537:1039703 [0] NCCL INFO Channel 05/0 : 56[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9627:9791 [2] NCCL INFO Channel 09/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-167-9:1039540:1039705 [3] NCCL INFO Channel 01/0 : 59[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9630:9789 [5] NCCL INFO Channel 09/0 : 5[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9632:9792 [7] NCCL INFO Channel 05/0 : 7[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9626:9790 [1] NCCL INFO Channel 11/0 : 1[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9628:9786 [3] NCCL INFO Channel 07/0 : 3[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9627:9791 [2] NCCL INFO Channel 11/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-167-9:1039542:1039701 [5] NCCL INFO Channel 05/0 : 61[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-167-9:1039543:1039699 [6] NCCL INFO Channel 05/0 : 62[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-46:702886:703049 [6] NCCL INFO Channel 00/0 : 38[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-46:702884:703047 [4] NCCL INFO Channel 00/0 : 36[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104436:104596 [7] NCCL INFO Channel 12/0 : 39[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104432:104592 [3] NCCL INFO Channel 00/0 : 35[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-167-9:1039538:1039704 [1] NCCL INFO Channel 01/0 : 57[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104434:104591 [5] NCCL INFO Channel 04/0 : 37[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104430:104589 [1] NCCL INFO Channel 00/0 : 33[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104433:104590 [4] NCCL INFO Channel 10/0 : 36[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9630:9789 [5] NCCL INFO Channel 11/0 : 5[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9632:9792 [7] NCCL INFO Channel 07/0 : 7[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104429:104594 [0] NCCL INFO Channel 10/0 : 32[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-46:702886:703049 [6] NCCL INFO Channel 02/0 : 38[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104431:104593 [2] NCCL INFO Channel 06/0 : 34[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104435:104595 [6] NCCL INFO Channel 12/0 : 38[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-46:702884:703047 [4] NCCL INFO Channel 02/0 : 36[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104436:104596 [7] NCCL INFO Channel 14/0 : 39[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-167-9:1039539:1039698 [2] NCCL INFO Channel 03/0 : 58[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104432:104592 [3] NCCL INFO Channel 02/0 : 35[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9627:9791 [2] NCCL INFO Channel 13/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-167-9:1039537:1039703 [0] NCCL INFO Channel 07/0 : 56[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-167-9:1039544:1039702 [7] NCCL INFO Channel 13/0 : 63[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-46:702886:703049 [6] NCCL INFO Channel 04/0 : 38[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104434:104591 [5] NCCL INFO Channel 06/0 : 37[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9628:9786 [3] NCCL INFO Channel 09/0 : 3[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-167-9:1039540:1039705 [3] NCCL INFO Channel 03/0 : 59[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO Channel 07/0 : 60[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104430:104589 [1] NCCL INFO Channel 02/0 : 33[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104433:104590 [4] NCCL INFO Channel 12/0 : 36[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9625:9787 [0] NCCL INFO Channel 01/0 : 8[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104429:104594 [0] NCCL INFO Channel 12/0 : 32[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9626:9790 [1] NCCL INFO Channel 13/0 : 1[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104435:104595 [6] NCCL INFO Channel 14/0 : 38[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104431:104593 [2] NCCL INFO Channel 08/0 : 34[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9632:9792 [7] NCCL INFO Channel 09/0 : 7[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9630:9789 [5] NCCL INFO Channel 15/0 : 5[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104432:104592 [3] NCCL INFO Channel 04/0 : 35[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104434:104591 [5] NCCL INFO Channel 08/0 : 37[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104430:104589 [1] NCCL INFO Channel 04/0 : 33[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-167-9:1039542:1039701 [5] NCCL INFO Channel 07/0 : 61[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-46:702884:703047 [4] NCCL INFO Channel 06/0 : 36[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-46:702880:703042 [0] NCCL INFO Channel 02/0 : 32[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104433:104590 [4] NCCL INFO Channel 14/0 : 36[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-167-9:1039543:1039699 [6] NCCL INFO Channel 07/0 : 62[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-46:702886:703049 [6] NCCL INFO Channel 08/0 : 38[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104431:104593 [2] NCCL INFO Channel 10/0 : 34[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104429:104594 [0] NCCL INFO Channel 14/0 : 32[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104432:104592 [3] NCCL INFO Channel 06/0 : 35[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9627:9791 [2] NCCL INFO Channel 15/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-167-9:1039538:1039704 [1] NCCL INFO Channel 03/0 : 57[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104434:104591 [5] NCCL INFO Channel 10/0 : 37[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104430:104589 [1] NCCL INFO Channel 06/0 : 33[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9629:9793 [4] NCCL INFO Channel 01/0 : 12[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9631:9788 [6] NCCL INFO Channel 01/0 : 14[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-167-9:1039539:1039698 [2] NCCL INFO Channel 05/0 : 58[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-167-9:1039537:1039703 [0] NCCL INFO Channel 09/0 : 56[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104436:104596 [7] NCCL INFO Channel 00/0 : 55[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9628:9786 [3] NCCL INFO Channel 13/0 : 3[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9625:9787 [0] NCCL INFO Channel 03/0 : 8[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-167-9:1039540:1039705 [3] NCCL INFO Channel 05/0 : 59[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO Channel 09/0 : 60[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104431:104593 [2] NCCL INFO Channel 12/0 : 34[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104432:104592 [3] NCCL INFO Channel 08/0 : 35[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-167-9:1039542:1039701 [5] NCCL INFO Channel 09/0 : 61[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104434:104591 [5] NCCL INFO Channel 12/0 : 37[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104430:104589 [1] NCCL INFO Channel 08/0 : 33[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104436:104596 [7] NCCL INFO Channel 02/0 : 55[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104435:104595 [6] NCCL INFO Channel 00/0 : 54[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104431:104593 [2] NCCL INFO Channel 14/0 : 34[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-46:702884:703047 [4] NCCL INFO Channel 08/0 : 36[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-46:702880:703042 [0] NCCL INFO Channel 04/0 : 32[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-46:702886:703049 [6] NCCL INFO Channel 10/0 : 38[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104432:104592 [3] NCCL INFO Channel 10/0 : 35[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104434:104591 [5] NCCL INFO Channel 14/0 : 37[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9632:9792 [7] NCCL INFO Channel 11/0 : 7[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9630:9789 [5] NCCL INFO Channel 00/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9626:9790 [1] NCCL INFO Channel 15/0 : 1[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9627:9791 [2] NCCL INFO Channel 00/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104430:104589 [1] NCCL INFO Channel 10/0 : 33[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9625:9787 [0] NCCL INFO Channel 05/0 : 8[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9631:9788 [6] NCCL INFO Channel 03/0 : 14[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9629:9793 [4] NCCL INFO Channel 03/0 : 12[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-167-9:1039538:1039704 [1] NCCL INFO Channel 05/0 : 57[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104436:104596 [7] NCCL INFO Channel 04/0 : 55[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104433:104590 [4] NCCL INFO Channel 00/0 : 52[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9628:9786 [3] NCCL INFO Channel 15/0 : 3[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-167-9:1039543:1039699 [6] NCCL INFO Channel 09/0 : 62[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104435:104595 [6] NCCL INFO Channel 02/0 : 54[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-167-9:1039539:1039698 [2] NCCL INFO Channel 07/0 : 58[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-167-9:1039537:1039703 [0] NCCL INFO Channel 11/0 : 56[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104432:104592 [3] NCCL INFO Channel 12/0 : 35[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-167-9:1039540:1039705 [3] NCCL INFO Channel 07/0 : 59[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104429:104594 [0] NCCL INFO Channel 02/0 : 48[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO Channel 11/0 : 60[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104430:104589 [1] NCCL INFO Channel 12/0 : 33[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-167-9:1039542:1039701 [5] NCCL INFO Channel 11/0 : 61[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104436:104596 [7] NCCL INFO Channel 06/0 : 55[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104435:104595 [6] NCCL INFO Channel 04/0 : 54[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-167-9:1039538:1039704 [1] NCCL INFO Channel 07/0 : 57[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104433:104590 [4] NCCL INFO Channel 02/0 : 52[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9627:9791 [2] NCCL INFO Channel 01/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104429:104594 [0] NCCL INFO Channel 04/0 : 48[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9630:9789 [5] NCCL INFO Channel 01/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9626:9790 [1] NCCL INFO Channel 00/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9632:9792 [7] NCCL INFO Channel 13/0 : 7[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-46:702886:703049 [6] NCCL INFO Channel 12/0 : 38[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-46:702880:703042 [0] NCCL INFO Channel 06/0 : 32[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104432:104592 [3] NCCL INFO Channel 14/0 : 35[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-46:702884:703047 [4] NCCL INFO Channel 10/0 : 36[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104431:104593 [2] NCCL INFO Channel 00/0 : 50[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104430:104589 [1] NCCL INFO Channel 14/0 : 33[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104436:104596 [7] NCCL INFO Channel 08/0 : 55[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-167-9:1039543:1039699 [6] NCCL INFO Channel 11/0 : 62[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104434:104591 [5] NCCL INFO Channel 00/0 : 53[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9631:9788 [6] NCCL INFO Channel 05/0 : 14[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9625:9787 [0] NCCL INFO Channel 07/0 : 8[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-167-9:1039537:1039703 [0] NCCL INFO Channel 13/0 : 56[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-167-9:1039539:1039698 [2] NCCL INFO Channel 09/0 : 58[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-46:702886:703049 [6] NCCL INFO Channel 14/0 : 38[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104435:104595 [6] NCCL INFO Channel 08/0 : 54[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104433:104590 [4] NCCL INFO Channel 06/0 : 52[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9629:9793 [4] NCCL INFO Channel 05/0 : 12[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9628:9786 [3] NCCL INFO Channel 00/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-167-9:1039540:1039705 [3] NCCL INFO Channel 09/0 : 59[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104429:104594 [0] NCCL INFO Channel 06/0 : 48[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104431:104593 [2] NCCL INFO Channel 04/0 : 50[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-167-9:1039538:1039704 [1] NCCL INFO Channel 11/0 : 57[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-167-9:1039542:1039701 [5] NCCL INFO Channel 15/0 : 61[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO Channel 13/0 : 60[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-167-9:1039543:1039699 [6] NCCL INFO Channel 13/0 : 62[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104436:104596 [7] NCCL INFO Channel 10/0 : 55[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104434:104591 [5] NCCL INFO Channel 02/0 : 53[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104433:104590 [4] NCCL INFO Channel 08/0 : 52[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104435:104595 [6] NCCL INFO Channel 10/0 : 54[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104429:104594 [0] NCCL INFO Channel 08/0 : 48[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-46:702880:703042 [0] NCCL INFO Channel 08/0 : 32[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-46:702884:703047 [4] NCCL INFO Channel 12/0 : 36[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104431:104593 [2] NCCL INFO Channel 06/0 : 50[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9626:9790 [1] NCCL INFO Channel 02/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9630:9789 [5] NCCL INFO Channel 02/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9627:9791 [2] NCCL INFO Channel 03/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9632:9792 [7] NCCL INFO Channel 00/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-46:702886:703049 [6] NCCL INFO Channel 00/0 : 6[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104436:104596 [7] NCCL INFO Channel 12/0 : 55[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9625:9787 [0] NCCL INFO Channel 09/0 : 8[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9631:9788 [6] NCCL INFO Channel 07/0 : 14[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9629:9793 [4] NCCL INFO Channel 07/0 : 12[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9628:9786 [3] NCCL INFO Channel 01/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-167-9:1039537:1039703 [0] NCCL INFO Channel 15/0 : 56[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-46:702880:703042 [0] NCCL INFO Channel 10/0 : 32[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104434:104591 [5] NCCL INFO Channel 04/0 : 53[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-167-9:1039539:1039698 [2] NCCL INFO Channel 11/0 : 58[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104430:104589 [1] NCCL INFO Channel 00/0 : 49[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104432:104592 [3] NCCL INFO Channel 00/0 : 51[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-167-9:1039540:1039705 [3] NCCL INFO Channel 13/0 : 59[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104433:104590 [4] NCCL INFO Channel 10/0 : 52[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-167-9:1039538:1039704 [1] NCCL INFO Channel 13/0 : 57[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104435:104595 [6] NCCL INFO Channel 12/0 : 54[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-167-9:1039543:1039699 [6] NCCL INFO Channel 15/0 : 62[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO Channel 15/0 : 60[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-46:702884:703047 [4] NCCL INFO Channel 14/0 : 36[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-46:702886:703049 [6] NCCL INFO Channel 02/0 : 6[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104429:104594 [0] NCCL INFO Channel 10/0 : 48[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-167-9:1039539:1039698 [2] NCCL INFO Channel 13/0 : 58[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104431:104593 [2] NCCL INFO Channel 08/0 : 50[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9630:9789 [5] NCCL INFO Channel 03/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-167-9:1039540:1039705 [3] NCCL INFO Channel 15/0 : 59[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-46:702880:703042 [0] NCCL INFO Channel 12/0 : 32[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104436:104596 [7] NCCL INFO Channel 14/0 : 55[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9627:9791 [2] NCCL INFO Channel 04/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-46:702886:703049 [6] NCCL INFO Channel 04/0 : 6[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-46:702884:703047 [4] NCCL INFO Channel 00/0 : 4[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104434:104591 [5] NCCL INFO Channel 06/0 : 53[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9626:9790 [1] NCCL INFO Channel 03/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104432:104592 [3] NCCL INFO Channel 02/0 : 51[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104430:104589 [1] NCCL INFO Channel 02/0 : 49[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9632:9792 [7] NCCL INFO Channel 01/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-46:702880:703042 [0] NCCL INFO Channel 14/0 : 32[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104433:104590 [4] NCCL INFO Channel 12/0 : 52[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-167-9:1039538:1039704 [1] NCCL INFO Channel 15/0 : 57[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104435:104595 [6] NCCL INFO Channel 14/0 : 54[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-167-9:1039539:1039698 [2] NCCL INFO Channel 15/0 : 58[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-46:702886:703049 [6] NCCL INFO Channel 08/0 : 6[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104429:104594 [0] NCCL INFO Channel 12/0 : 48[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104431:104593 [2] NCCL INFO Channel 10/0 : 50[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-46:702884:703047 [4] NCCL INFO Channel 02/0 : 4[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104434:104591 [5] NCCL INFO Channel 08/0 : 53[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9625:9787 [0] NCCL INFO Channel 11/0 : 8[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-46:702880:703042 [0] NCCL INFO Channel 02/0 : 0[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104430:104589 [1] NCCL INFO Channel 04/0 : 49[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104432:104592 [3] NCCL INFO Channel 04/0 : 51[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9631:9788 [6] NCCL INFO Channel 09/0 : 14[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9628:9786 [3] NCCL INFO Channel 02/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9629:9793 [4] NCCL INFO Channel 09/0 : 12[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-46:702886:703049 [6] NCCL INFO Channel 10/0 : 6[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9630:9789 [5] NCCL INFO Channel 04/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9627:9791 [2] NCCL INFO Channel 05/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9626:9790 [1] NCCL INFO Channel 04/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-46:702884:703047 [4] NCCL INFO Channel 06/0 : 4[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9632:9792 [7] NCCL INFO Channel 02/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-46:702880:703042 [0] NCCL INFO Channel 04/0 : 0[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104433:104590 [4] NCCL INFO Channel 14/0 : 52[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9625:9787 [0] NCCL INFO Channel 13/0 : 8[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104431:104593 [2] NCCL INFO Channel 12/0 : 50[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104429:104594 [0] NCCL INFO Channel 14/0 : 48[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104434:104591 [5] NCCL INFO Channel 10/0 : 53[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104432:104592 [3] NCCL INFO Channel 06/0 : 51[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104430:104589 [1] NCCL INFO Channel 06/0 : 49[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-46:702886:703049 [6] NCCL INFO Channel 12/0 : 6[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104431:104593 [2] NCCL INFO Channel 14/0 : 50[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104434:104591 [5] NCCL INFO Channel 12/0 : 53[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104432:104592 [3] NCCL INFO Channel 08/0 : 51[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104430:104589 [1] NCCL INFO Channel 08/0 : 49[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104434:104591 [5] NCCL INFO Channel 14/0 : 53[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104430:104589 [1] NCCL INFO Channel 10/0 : 49[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9628:9786 [3] NCCL INFO Channel 04/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9631:9788 [6] NCCL INFO Channel 11/0 : 14[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-46:702884:703047 [4] NCCL INFO Channel 08/0 : 4[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104432:104592 [3] NCCL INFO Channel 10/0 : 51[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9629:9793 [4] NCCL INFO Channel 11/0 : 12[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-46:702880:703042 [0] NCCL INFO Channel 06/0 : 0[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104430:104589 [1] NCCL INFO Channel 12/0 : 49[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9627:9791 [2] NCCL INFO Channel 06/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9630:9789 [5] NCCL INFO Channel 06/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-46:702886:703049 [6] NCCL INFO Channel 14/0 : 6[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104432:104592 [3] NCCL INFO Channel 12/0 : 51[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9626:9790 [1] NCCL INFO Channel 05/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9632:9792 [7] NCCL INFO Channel 03/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9625:9787 [0] NCCL INFO Channel 15/0 : 8[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104430:104589 [1] NCCL INFO Channel 14/0 : 49[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9628:9786 [3] NCCL INFO Channel 05/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104432:104592 [3] NCCL INFO Channel 14/0 : 51[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-46:702884:703047 [4] NCCL INFO Channel 10/0 : 4[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-46:702880:703042 [0] NCCL INFO Channel 08/0 : 0[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9631:9788 [6] NCCL INFO Channel 13/0 : 14[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9629:9793 [4] NCCL INFO Channel 13/0 : 12[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9627:9791 [2] NCCL INFO Channel 07/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9630:9789 [5] NCCL INFO Channel 07/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9626:9790 [1] NCCL INFO Channel 06/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9632:9792 [7] NCCL INFO Channel 04/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9628:9786 [3] NCCL INFO Channel 06/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-46:702880:703042 [0] NCCL INFO Channel 10/0 : 0[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9631:9788 [6] NCCL INFO Channel 15/0 : 14[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-46:702884:703047 [4] NCCL INFO Channel 12/0 : 4[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9629:9793 [4] NCCL INFO Channel 15/0 : 12[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-46:702880:703042 [0] NCCL INFO Channel 12/0 : 0[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-46:702884:703047 [4] NCCL INFO Channel 14/0 : 4[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9627:9791 [2] NCCL INFO Channel 08/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9630:9789 [5] NCCL INFO Channel 08/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9626:9790 [1] NCCL INFO Channel 07/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9628:9786 [3] NCCL INFO Channel 07/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9632:9792 [7] NCCL INFO Channel 05/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-46:702880:703042 [0] NCCL INFO Channel 14/0 : 0[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9627:9791 [2] NCCL INFO Channel 09/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9630:9789 [5] NCCL INFO Channel 09/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9628:9786 [3] NCCL INFO Channel 08/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9626:9790 [1] NCCL INFO Channel 08/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9632:9792 [7] NCCL INFO Channel 06/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9627:9791 [2] NCCL INFO Channel 11/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9630:9789 [5] NCCL INFO Channel 10/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9628:9786 [3] NCCL INFO Channel 09/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9626:9790 [1] NCCL INFO Channel 10/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9627:9791 [2] NCCL INFO Channel 12/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9632:9792 [7] NCCL INFO Channel 08/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9630:9789 [5] NCCL INFO Channel 11/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9628:9786 [3] NCCL INFO Channel 10/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9627:9791 [2] NCCL INFO Channel 13/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9626:9790 [1] NCCL INFO Channel 11/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9632:9792 [7] NCCL INFO Channel 09/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9630:9789 [5] NCCL INFO Channel 12/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9628:9786 [3] NCCL INFO Channel 12/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9627:9791 [2] NCCL INFO Channel 14/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9632:9792 [7] NCCL INFO Channel 10/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9626:9790 [1] NCCL INFO Channel 12/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9630:9789 [5] NCCL INFO Channel 14/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9628:9786 [3] NCCL INFO Channel 13/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9627:9791 [2] NCCL INFO Channel 15/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9632:9792 [7] NCCL INFO Channel 11/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9630:9789 [5] NCCL INFO Channel 15/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9628:9786 [3] NCCL INFO Channel 14/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9626:9790 [1] NCCL INFO Channel 13/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9632:9792 [7] NCCL INFO Channel 12/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9628:9786 [3] NCCL INFO Channel 15/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9626:9790 [1] NCCL INFO Channel 14/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9632:9792 [7] NCCL INFO Channel 13/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9626:9790 [1] NCCL INFO Channel 15/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9632:9792 [7] NCCL INFO Channel 14/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-46:702882:703046 [2] NCCL INFO Channel 00/0 : 34[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9627:9791 [2] NCCL INFO Channel 01/0 : 10[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-46:702885:703048 [5] NCCL INFO Channel 00/0 : 37[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-46:702882:703046 [2] NCCL INFO Channel 04/0 : 34[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9627:9791 [2] NCCL INFO Channel 03/0 : 10[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-46:702885:703048 [5] NCCL INFO Channel 02/0 : 37[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9630:9789 [5] NCCL INFO Channel 01/0 : 13[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-46:702883:703044 [3] NCCL INFO Channel 00/0 : 35[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-46:702882:703046 [2] NCCL INFO Channel 06/0 : 34[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9628:9786 [3] NCCL INFO Channel 01/0 : 11[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-46:702887:703045 [7] NCCL INFO Channel 00/0 : 39[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-46:702885:703048 [5] NCCL INFO Channel 04/0 : 37[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9627:9791 [2] NCCL INFO Channel 05/0 : 10[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9630:9789 [5] NCCL INFO Channel 03/0 : 13[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-46:702881:703043 [1] NCCL INFO Channel 00/0 : 33[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-46:702883:703044 [3] NCCL INFO Channel 02/0 : 35[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-46:702882:703046 [2] NCCL INFO Channel 08/0 : 34[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9628:9786 [3] NCCL INFO Channel 03/0 : 11[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-46:702887:703045 [7] NCCL INFO Channel 02/0 : 39[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-46:702885:703048 [5] NCCL INFO Channel 06/0 : 37[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-46:702883:703044 [3] NCCL INFO Channel 04/0 : 35[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-46:702881:703043 [1] NCCL INFO Channel 02/0 : 33[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-46:702882:703046 [2] NCCL INFO Channel 10/0 : 34[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9627:9791 [2] NCCL INFO Channel 07/0 : 10[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9630:9789 [5] NCCL INFO Channel 05/0 : 13[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9628:9786 [3] NCCL INFO Channel 05/0 : 11[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9632:9792 [7] NCCL INFO Channel 01/0 : 15[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9626:9790 [1] NCCL INFO Channel 01/0 : 9[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9627:9791 [2] NCCL INFO Channel 09/0 : 10[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9630:9789 [5] NCCL INFO Channel 07/0 : 13[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9628:9786 [3] NCCL INFO Channel 07/0 : 11[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-46:702885:703048 [5] NCCL INFO Channel 08/0 : 37[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-46:702883:703044 [3] NCCL INFO Channel 06/0 : 35[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-46:702882:703046 [2] NCCL INFO Channel 12/0 : 34[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-46:702887:703045 [7] NCCL INFO Channel 04/0 : 39[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-46:702881:703043 [1] NCCL INFO Channel 04/0 : 33[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9632:9792 [7] NCCL INFO Channel 03/0 : 15[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-46:702883:703044 [3] NCCL INFO Channel 08/0 : 35[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9626:9790 [1] NCCL INFO Channel 03/0 : 9[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-46:702882:703046 [2] NCCL INFO Channel 14/0 : 34[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9627:9791 [2] NCCL INFO Channel 11/0 : 10[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9630:9789 [5] NCCL INFO Channel 09/0 : 13[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9628:9786 [3] NCCL INFO Channel 09/0 : 11[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-46:702885:703048 [5] NCCL INFO Channel 10/0 : 37[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-46:702881:703043 [1] NCCL INFO Channel 06/0 : 33[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-46:702887:703045 [7] NCCL INFO Channel 06/0 : 39[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-46:702883:703044 [3] NCCL INFO Channel 10/0 : 35[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-46:702882:703046 [2] NCCL INFO Channel 00/0 : 2[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9632:9792 [7] NCCL INFO Channel 05/0 : 15[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9626:9790 [1] NCCL INFO Channel 05/0 : 9[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9627:9791 [2] NCCL INFO Channel 13/0 : 10[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9630:9789 [5] NCCL INFO Channel 11/0 : 13[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9628:9786 [3] NCCL INFO Channel 13/0 : 11[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9632:9792 [7] NCCL INFO Channel 07/0 : 15[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9626:9790 [1] NCCL INFO Channel 07/0 : 9[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-46:702885:703048 [5] NCCL INFO Channel 12/0 : 37[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-46:702881:703043 [1] NCCL INFO Channel 08/0 : 33[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-46:702887:703045 [7] NCCL INFO Channel 08/0 : 39[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-46:702883:703044 [3] NCCL INFO Channel 12/0 : 35[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-46:702882:703046 [2] NCCL INFO Channel 04/0 : 2[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9627:9791 [2] NCCL INFO Channel 15/0 : 10[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-46:702885:703048 [5] NCCL INFO Channel 14/0 : 37[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9630:9789 [5] NCCL INFO Channel 15/0 : 13[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9628:9786 [3] NCCL INFO Channel 15/0 : 11[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9632:9792 [7] NCCL INFO Channel 09/0 : 15[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9626:9790 [1] NCCL INFO Channel 11/0 : 9[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9632:9792 [7] NCCL INFO Channel 11/0 : 15[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9626:9790 [1] NCCL INFO Channel 13/0 : 9[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9632:9792 [7] NCCL INFO Channel 13/0 : 15[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-46:702881:703043 [1] NCCL INFO Channel 10/0 : 33[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-46:702883:703044 [3] NCCL INFO Channel 14/0 : 35[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-46:702882:703046 [2] NCCL INFO Channel 06/0 : 2[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-46:702887:703045 [7] NCCL INFO Channel 10/0 : 39[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-46:702885:703048 [5] NCCL INFO Channel 00/0 : 5[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9626:9790 [1] NCCL INFO Channel 15/0 : 9[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-46:702882:703046 [2] NCCL INFO Channel 08/0 : 2[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-46:702883:703044 [3] NCCL INFO Channel 00/0 : 3[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-46:702881:703043 [1] NCCL INFO Channel 12/0 : 33[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-46:702887:703045 [7] NCCL INFO Channel 12/0 : 39[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-46:702885:703048 [5] NCCL INFO Channel 02/0 : 5[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-46:702883:703044 [3] NCCL INFO Channel 02/0 : 3[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-46:702881:703043 [1] NCCL INFO Channel 14/0 : 33[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-46:702882:703046 [2] NCCL INFO Channel 10/0 : 2[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-46:702885:703048 [5] NCCL INFO Channel 04/0 : 5[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-46:702887:703045 [7] NCCL INFO Channel 14/0 : 39[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-46:702883:703044 [3] NCCL INFO Channel 04/0 : 3[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-46:702881:703043 [1] NCCL INFO Channel 00/0 : 1[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-46:702882:703046 [2] NCCL INFO Channel 12/0 : 2[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-46:702885:703048 [5] NCCL INFO Channel 06/0 : 5[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-46:702887:703045 [7] NCCL INFO Channel 00/0 : 7[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-46:702883:703044 [3] NCCL INFO Channel 06/0 : 3[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-46:702881:703043 [1] NCCL INFO Channel 02/0 : 1[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-46:702885:703048 [5] NCCL INFO Channel 08/0 : 5[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-46:702882:703046 [2] NCCL INFO Channel 14/0 : 2[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-46:702883:703044 [3] NCCL INFO Channel 08/0 : 3[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-46:702887:703045 [7] NCCL INFO Channel 02/0 : 7[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-46:702881:703043 [1] NCCL INFO Channel 04/0 : 1[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-46:702885:703048 [5] NCCL INFO Channel 10/0 : 5[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-46:702881:703043 [1] NCCL INFO Channel 06/0 : 1[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-46:702887:703045 [7] NCCL INFO Channel 04/0 : 7[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-46:702885:703048 [5] NCCL INFO Channel 12/0 : 5[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-46:702883:703044 [3] NCCL INFO Channel 10/0 : 3[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-46:702881:703043 [1] NCCL INFO Channel 08/0 : 1[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-46:702887:703045 [7] NCCL INFO Channel 06/0 : 7[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-46:702885:703048 [5] NCCL INFO Channel 14/0 : 5[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-46:702883:703044 [3] NCCL INFO Channel 12/0 : 3[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-46:702881:703043 [1] NCCL INFO Channel 10/0 : 1[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-46:702887:703045 [7] NCCL INFO Channel 08/0 : 7[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-46:702883:703044 [3] NCCL INFO Channel 14/0 : 3[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-46:702881:703043 [1] NCCL INFO Channel 12/0 : 1[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-46:702887:703045 [7] NCCL INFO Channel 10/0 : 7[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-46:702881:703043 [1] NCCL INFO Channel 14/0 : 1[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-46:702887:703045 [7] NCCL INFO Channel 12/0 : 7[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-46:702887:703045 [7] NCCL INFO Channel 14/0 : 7[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127294:127457 [4] NCCL INFO Channel 00/0 : 20[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127294:127457 [4] NCCL INFO Channel 02/0 : 20[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127294:127457 [4] NCCL INFO Channel 06/0 : 20[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127294:127457 [4] NCCL INFO Channel 08/0 : 20[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127290:127456 [0] NCCL INFO Channel 02/0 : 16[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127294:127457 [4] NCCL INFO Channel 10/0 : 20[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127294:127457 [4] NCCL INFO Channel 12/0 : 20[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127290:127456 [0] NCCL INFO Channel 04/0 : 16[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127294:127457 [4] NCCL INFO Channel 14/0 : 20[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127290:127456 [0] NCCL INFO Channel 06/0 : 16[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127290:127456 [0] NCCL INFO Channel 08/0 : 16[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127290:127456 [0] NCCL INFO Channel 10/0 : 16[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127290:127456 [0] NCCL INFO Channel 12/0 : 16[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127294:127457 [4] NCCL INFO Channel 01/0 : 12[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127294:127457 [4] NCCL INFO Channel 03/0 : 12[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127290:127456 [0] NCCL INFO Channel 14/0 : 16[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127294:127457 [4] NCCL INFO Channel 05/0 : 12[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127294:127457 [4] NCCL INFO Channel 07/0 : 12[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127290:127456 [0] NCCL INFO Channel 01/0 : 8[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127294:127457 [4] NCCL INFO Channel 09/0 : 12[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127294:127457 [4] NCCL INFO Channel 11/0 : 12[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127290:127456 [0] NCCL INFO Channel 03/0 : 8[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127290:127456 [0] NCCL INFO Channel 05/0 : 8[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127294:127457 [4] NCCL INFO Channel 13/0 : 12[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127290:127456 [0] NCCL INFO Channel 07/0 : 8[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127294:127457 [4] NCCL INFO Channel 15/0 : 12[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127290:127456 [0] NCCL INFO Channel 09/0 : 8[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127294:127457 [4] NCCL INFO Channel 01/0 : 28[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127294:127457 [4] NCCL INFO Channel 03/0 : 28[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127290:127456 [0] NCCL INFO Channel 11/0 : 8[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127294:127457 [4] NCCL INFO Channel 05/0 : 28[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127290:127456 [0] NCCL INFO Channel 13/0 : 8[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127294:127457 [4] NCCL INFO Channel 07/0 : 28[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127290:127456 [0] NCCL INFO Channel 15/0 : 8[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127294:127457 [4] NCCL INFO Channel 09/0 : 28[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127290:127456 [0] NCCL INFO Channel 01/0 : 24[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127294:127457 [4] NCCL INFO Channel 11/0 : 28[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127290:127456 [0] NCCL INFO Channel 03/0 : 24[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127294:127457 [4] NCCL INFO Channel 13/0 : 28[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127290:127456 [0] NCCL INFO Channel 05/0 : 24[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127294:127457 [4] NCCL INFO Channel 15/0 : 28[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127290:127456 [0] NCCL INFO Channel 07/0 : 24[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127290:127456 [0] NCCL INFO Channel 09/0 : 24[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127290:127456 [0] NCCL INFO Channel 11/0 : 24[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127290:127456 [0] NCCL INFO Channel 13/0 : 24[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9629:9793 [4] NCCL INFO Channel 01/0 : 28[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127290:127456 [0] NCCL INFO Channel 15/0 : 24[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9629:9793 [4] NCCL INFO Channel 03/0 : 28[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127294:127457 [4] NCCL INFO Channel 01/0 : 60[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127294:127457 [4] NCCL INFO Channel 03/0 : 60[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9629:9793 [4] NCCL INFO Channel 05/0 : 28[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127294:127457 [4] NCCL INFO Channel 05/0 : 60[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9629:9793 [4] NCCL INFO Channel 07/0 : 28[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127294:127457 [4] NCCL INFO Channel 07/0 : 60[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9625:9787 [0] NCCL INFO Channel 01/0 : 24[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9629:9793 [4] NCCL INFO Channel 09/0 : 28[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127294:127457 [4] NCCL INFO Channel 09/0 : 60[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127290:127456 [0] NCCL INFO Channel 01/0 : 56[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9629:9793 [4] NCCL INFO Channel 11/0 : 28[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-46:702884:703047 [4] NCCL INFO Channel 01/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127292:127458 [2] NCCL INFO Channel 00/0 : 18[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127294:127457 [4] NCCL INFO Channel 11/0 : 60[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624947:625105 [4] NCCL INFO Channel 12/0 : 20[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9625:9787 [0] NCCL INFO Channel 03/0 : 24[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624949:625110 [6] NCCL INFO Channel 14/0 : 22[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624945:625111 [2] NCCL INFO Channel 10/0 : 18[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-46:702884:703047 [4] NCCL INFO Channel 03/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624946:625107 [3] NCCL INFO Channel 12/0 : 19[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624943:625108 [0] NCCL INFO Channel 06/0 : 16[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127293:127454 [3] NCCL INFO Channel 00/0 : 19[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127290:127456 [0] NCCL INFO Channel 03/0 : 56[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624949:625110 [6] NCCL INFO Channel 00/0 : 38[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624947:625105 [4] NCCL INFO Channel 14/0 : 20[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127296:127451 [6] NCCL INFO Channel 00/0 : 22[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624945:625111 [2] NCCL INFO Channel 12/0 : 18[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624946:625107 [3] NCCL INFO Channel 14/0 : 19[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127294:127457 [4] NCCL INFO Channel 13/0 : 60[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9629:9793 [4] NCCL INFO Channel 13/0 : 28[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624943:625108 [0] NCCL INFO Channel 08/0 : 16[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9625:9787 [0] NCCL INFO Channel 05/0 : 24[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624949:625110 [6] NCCL INFO Channel 02/0 : 38[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624947:625105 [4] NCCL INFO Channel 00/0 : 36[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-46:702884:703047 [4] NCCL INFO Channel 05/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624945:625111 [2] NCCL INFO Channel 14/0 : 18[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624943:625108 [0] NCCL INFO Channel 10/0 : 16[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624946:625107 [3] NCCL INFO Channel 00/0 : 35[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624949:625110 [6] NCCL INFO Channel 04/0 : 38[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127292:127458 [2] NCCL INFO Channel 04/0 : 18[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624947:625105 [4] NCCL INFO Channel 02/0 : 36[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127290:127456 [0] NCCL INFO Channel 05/0 : 56[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624945:625111 [2] NCCL INFO Channel 00/0 : 34[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624943:625108 [0] NCCL INFO Channel 12/0 : 16[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624946:625107 [3] NCCL INFO Channel 02/0 : 35[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127293:127454 [3] NCCL INFO Channel 02/0 : 19[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9625:9787 [0] NCCL INFO Channel 07/0 : 24[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9629:9793 [4] NCCL INFO Channel 15/0 : 28[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624949:625110 [6] NCCL INFO Channel 08/0 : 38[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624947:625105 [4] NCCL INFO Channel 06/0 : 36[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624945:625111 [2] NCCL INFO Channel 04/0 : 34[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624946:625107 [3] NCCL INFO Channel 04/0 : 35[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624943:625108 [0] NCCL INFO Channel 14/0 : 16[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-46:702884:703047 [4] NCCL INFO Channel 07/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624949:625110 [6] NCCL INFO Channel 10/0 : 38[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624947:625105 [4] NCCL INFO Channel 08/0 : 36[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127296:127451 [6] NCCL INFO Channel 02/0 : 22[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9625:9787 [0] NCCL INFO Channel 09/0 : 24[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624945:625111 [2] NCCL INFO Channel 06/0 : 34[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127294:127457 [4] NCCL INFO Channel 15/0 : 60[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624946:625107 [3] NCCL INFO Channel 06/0 : 35[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624943:625108 [0] NCCL INFO Channel 02/0 : 32[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624949:625110 [6] NCCL INFO Channel 12/0 : 38[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624947:625105 [4] NCCL INFO Channel 10/0 : 36[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-46:702884:703047 [4] NCCL INFO Channel 09/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624945:625111 [2] NCCL INFO Channel 08/0 : 34[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127292:127458 [2] NCCL INFO Channel 06/0 : 18[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127297:127453 [7] NCCL INFO Channel 00/0 : 23[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127290:127456 [0] NCCL INFO Channel 07/0 : 56[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127295:127452 [5] NCCL INFO Channel 00/0 : 21[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9625:9787 [0] NCCL INFO Channel 11/0 : 24[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624946:625107 [3] NCCL INFO Channel 08/0 : 35[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127293:127454 [3] NCCL INFO Channel 04/0 : 19[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624943:625108 [0] NCCL INFO Channel 04/0 : 32[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127296:127451 [6] NCCL INFO Channel 04/0 : 22[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127294:127457 [4] NCCL INFO Channel 01/0 : 28[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624949:625110 [6] NCCL INFO Channel 14/0 : 38[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624947:625105 [4] NCCL INFO Channel 12/0 : 36[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624945:625111 [2] NCCL INFO Channel 10/0 : 34[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-46:702884:703047 [4] NCCL INFO Channel 11/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624946:625107 [3] NCCL INFO Channel 10/0 : 35[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9625:9787 [0] NCCL INFO Channel 13/0 : 24[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624943:625108 [0] NCCL INFO Channel 06/0 : 32[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624947:625105 [4] NCCL INFO Channel 14/0 : 36[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624945:625111 [2] NCCL INFO Channel 12/0 : 34[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127292:127458 [2] NCCL INFO Channel 08/0 : 18[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624946:625107 [3] NCCL INFO Channel 12/0 : 35[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624943:625108 [0] NCCL INFO Channel 08/0 : 32[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127290:127456 [0] NCCL INFO Channel 09/0 : 56[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624945:625111 [2] NCCL INFO Channel 14/0 : 34[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624946:625107 [3] NCCL INFO Channel 14/0 : 35[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624943:625108 [0] NCCL INFO Channel 10/0 : 32[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624943:625108 [0] NCCL INFO Channel 12/0 : 32[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624943:625108 [0] NCCL INFO Channel 14/0 : 32[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624947:625105 [4] NCCL INFO Channel 00/0 : 4[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-46:702884:703047 [4] NCCL INFO Channel 13/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624947:625105 [4] NCCL INFO Channel 02/0 : 4[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9625:9787 [0] NCCL INFO Channel 15/0 : 24[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624947:625105 [4] NCCL INFO Channel 06/0 : 4[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127293:127454 [3] NCCL INFO Channel 06/0 : 19[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127297:127453 [7] NCCL INFO Channel 02/0 : 23[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127291:127455 [1] NCCL INFO Channel 00/0 : 17[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127296:127451 [6] NCCL INFO Channel 08/0 : 22[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624947:625105 [4] NCCL INFO Channel 08/0 : 4[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127294:127457 [4] NCCL INFO Channel 03/0 : 28[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127295:127452 [5] NCCL INFO Channel 02/0 : 21[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624947:625105 [4] NCCL INFO Channel 10/0 : 4[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127290:127456 [0] NCCL INFO Channel 11/0 : 56[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127292:127458 [2] NCCL INFO Channel 10/0 : 18[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624943:625108 [0] NCCL INFO Channel 02/0 : 0[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624947:625105 [4] NCCL INFO Channel 12/0 : 4[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624943:625108 [0] NCCL INFO Channel 04/0 : 0[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624947:625105 [4] NCCL INFO Channel 14/0 : 4[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-46:702884:703047 [4] NCCL INFO Channel 15/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624947:625105 [4] NCCL INFO Channel 00/0 : 36[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624943:625108 [0] NCCL INFO Channel 06/0 : 0[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127293:127454 [3] NCCL INFO Channel 08/0 : 19[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624947:625105 [4] NCCL INFO Channel 02/0 : 36[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624943:625108 [0] NCCL INFO Channel 08/0 : 0[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624947:625105 [4] NCCL INFO Channel 06/0 : 36[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-46:702880:703042 [0] NCCL INFO Channel 01/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624943:625108 [0] NCCL INFO Channel 10/0 : 0[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624947:625105 [4] NCCL INFO Channel 08/0 : 36[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624943:625108 [0] NCCL INFO Channel 12/0 : 0[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127297:127453 [7] NCCL INFO Channel 04/0 : 23[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127291:127455 [1] NCCL INFO Channel 02/0 : 17[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127294:127457 [4] NCCL INFO Channel 05/0 : 28[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624947:625105 [4] NCCL INFO Channel 10/0 : 36[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127296:127451 [6] NCCL INFO Channel 10/0 : 22[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624943:625108 [0] NCCL INFO Channel 14/0 : 0[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127295:127452 [5] NCCL INFO Channel 04/0 : 21[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624947:625105 [4] NCCL INFO Channel 12/0 : 36[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127290:127456 [0] NCCL INFO Channel 13/0 : 56[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-46:702880:703042 [0] NCCL INFO Channel 03/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624943:625108 [0] NCCL INFO Channel 02/0 : 32[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127292:127458 [2] NCCL INFO Channel 12/0 : 18[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624947:625105 [4] NCCL INFO Channel 14/0 : 36[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624943:625108 [0] NCCL INFO Channel 04/0 : 32[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624943:625108 [0] NCCL INFO Channel 06/0 : 32[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624943:625108 [0] NCCL INFO Channel 08/0 : 32[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624947:625105 [4] NCCL INFO Channel 00/0 : 52[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624943:625108 [0] NCCL INFO Channel 10/0 : 32[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624947:625105 [4] NCCL INFO Channel 02/0 : 52[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127293:127454 [3] NCCL INFO Channel 10/0 : 19[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624943:625108 [0] NCCL INFO Channel 12/0 : 32[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127291:127455 [1] NCCL INFO Channel 04/0 : 17[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-46:702880:703042 [0] NCCL INFO Channel 05/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127294:127457 [4] NCCL INFO Channel 07/0 : 28[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127296:127451 [6] NCCL INFO Channel 12/0 : 22[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624947:625105 [4] NCCL INFO Channel 06/0 : 52[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-46:702880:703042 [0] NCCL INFO Channel 07/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624943:625108 [0] NCCL INFO Channel 14/0 : 32[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624947:625105 [4] NCCL INFO Channel 08/0 : 52[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127295:127452 [5] NCCL INFO Channel 06/0 : 21[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127297:127453 [7] NCCL INFO Channel 06/0 : 23[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127292:127458 [2] NCCL INFO Channel 14/0 : 18[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127293:127454 [3] NCCL INFO Channel 12/0 : 19[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624947:625105 [4] NCCL INFO Channel 10/0 : 52[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127290:127456 [0] NCCL INFO Channel 15/0 : 56[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127291:127455 [1] NCCL INFO Channel 06/0 : 17[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624947:625105 [4] NCCL INFO Channel 12/0 : 52[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127294:127457 [4] NCCL INFO Channel 09/0 : 28[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624943:625108 [0] NCCL INFO Channel 02/0 : 48[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127296:127451 [6] NCCL INFO Channel 14/0 : 22[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624947:625105 [4] NCCL INFO Channel 14/0 : 52[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624943:625108 [0] NCCL INFO Channel 04/0 : 48[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624947:625105 [4] NCCL INFO Channel 00/0 : 36[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624943:625108 [0] NCCL INFO Channel 06/0 : 48[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624947:625105 [4] NCCL INFO Channel 02/0 : 36[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-46:702880:703042 [0] NCCL INFO Channel 09/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624943:625108 [0] NCCL INFO Channel 08/0 : 48[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624947:625105 [4] NCCL INFO Channel 06/0 : 36[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127295:127452 [5] NCCL INFO Channel 08/0 : 21[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624943:625108 [0] NCCL INFO Channel 10/0 : 48[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127297:127453 [7] NCCL INFO Channel 08/0 : 23[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624947:625105 [4] NCCL INFO Channel 08/0 : 36[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127290:127456 [0] NCCL INFO Channel 01/0 : 24[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127293:127454 [3] NCCL INFO Channel 14/0 : 19[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624943:625108 [0] NCCL INFO Channel 12/0 : 48[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127291:127455 [1] NCCL INFO Channel 08/0 : 17[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127294:127457 [4] NCCL INFO Channel 11/0 : 28[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624947:625105 [4] NCCL INFO Channel 10/0 : 36[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624943:625108 [0] NCCL INFO Channel 14/0 : 48[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-46:702880:703042 [0] NCCL INFO Channel 11/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624947:625105 [4] NCCL INFO Channel 12/0 : 36[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127295:127452 [5] NCCL INFO Channel 10/0 : 21[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127297:127453 [7] NCCL INFO Channel 10/0 : 23[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127290:127456 [0] NCCL INFO Channel 03/0 : 24[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127291:127455 [1] NCCL INFO Channel 10/0 : 17[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127294:127457 [4] NCCL INFO Channel 13/0 : 28[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624943:625108 [0] NCCL INFO Channel 02/0 : 32[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624947:625105 [4] NCCL INFO Channel 14/0 : 36[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-46:702880:703042 [0] NCCL INFO Channel 13/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624943:625108 [0] NCCL INFO Channel 04/0 : 32[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624943:625108 [0] NCCL INFO Channel 06/0 : 32[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127292:127458 [2] NCCL INFO Channel 01/0 : 10[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127295:127452 [5] NCCL INFO Channel 12/0 : 21[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127296:127451 [6] NCCL INFO Channel 01/0 : 14[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127297:127453 [7] NCCL INFO Channel 12/0 : 23[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127290:127456 [0] NCCL INFO Channel 05/0 : 24[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624943:625108 [0] NCCL INFO Channel 08/0 : 32[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-46:702880:703042 [0] NCCL INFO Channel 15/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127291:127455 [1] NCCL INFO Channel 12/0 : 17[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127294:127457 [4] NCCL INFO Channel 15/0 : 28[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127292:127458 [2] NCCL INFO Channel 03/0 : 10[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104433:104590 [4] NCCL INFO Channel 00/0 : 60[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624943:625108 [0] NCCL INFO Channel 10/0 : 32[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127295:127452 [5] NCCL INFO Channel 14/0 : 21[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127296:127451 [6] NCCL INFO Channel 03/0 : 14[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127290:127456 [0] NCCL INFO Channel 07/0 : 24[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127293:127454 [3] NCCL INFO Channel 01/0 : 11[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127297:127453 [7] NCCL INFO Channel 14/0 : 23[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127291:127455 [1] NCCL INFO Channel 14/0 : 17[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104433:104590 [4] NCCL INFO Channel 02/0 : 60[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127292:127458 [2] NCCL INFO Channel 05/0 : 10[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624947:625105 [4] NCCL INFO Channel 01/0 : 44[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624943:625108 [0] NCCL INFO Channel 12/0 : 32[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104433:104590 [4] NCCL INFO Channel 06/0 : 60[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127296:127451 [6] NCCL INFO Channel 05/0 : 14[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127290:127456 [0] NCCL INFO Channel 09/0 : 24[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127293:127454 [3] NCCL INFO Channel 03/0 : 11[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624947:625105 [4] NCCL INFO Channel 03/0 : 44[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624943:625108 [0] NCCL INFO Channel 14/0 : 32[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127292:127458 [2] NCCL INFO Channel 07/0 : 10[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104433:104590 [4] NCCL INFO Channel 08/0 : 60[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127296:127451 [6] NCCL INFO Channel 07/0 : 14[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624947:625105 [4] NCCL INFO Channel 05/0 : 44[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO Channel 00/0 : 60[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104433:104590 [4] NCCL INFO Channel 10/0 : 60[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127294:127457 [4] NCCL INFO Channel 01/0 : 44[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127292:127458 [2] NCCL INFO Channel 09/0 : 10[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624947:625105 [4] NCCL INFO Channel 07/0 : 44[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO Channel 02/0 : 60[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104433:104590 [4] NCCL INFO Channel 12/0 : 60[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127293:127454 [3] NCCL INFO Channel 05/0 : 11[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127291:127455 [1] NCCL INFO Channel 01/0 : 9[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127290:127456 [0] NCCL INFO Channel 11/0 : 24[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127295:127452 [5] NCCL INFO Channel 01/0 : 13[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127296:127451 [6] NCCL INFO Channel 09/0 : 14[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127297:127453 [7] NCCL INFO Channel 01/0 : 15[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO Channel 06/0 : 60[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624947:625105 [4] NCCL INFO Channel 09/0 : 44[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104433:104590 [4] NCCL INFO Channel 14/0 : 60[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762819:762982 [4] NCCL INFO Channel 08/0 : 20[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762821:762988 [6] NCCL INFO Channel 04/0 : 22[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762818:762985 [3] NCCL INFO Channel 06/0 : 19[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762815:762983 [0] NCCL INFO Channel 08/0 : 16[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762817:762986 [2] NCCL INFO Channel 14/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127292:127458 [2] NCCL INFO Channel 11/0 : 10[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127294:127457 [4] NCCL INFO Channel 03/0 : 44[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762819:762982 [4] NCCL INFO Channel 10/0 : 20[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127293:127454 [3] NCCL INFO Channel 07/0 : 11[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762818:762985 [3] NCCL INFO Channel 08/0 : 19[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624947:625105 [4] NCCL INFO Channel 11/0 : 44[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO Channel 08/0 : 60[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762815:762983 [0] NCCL INFO Channel 10/0 : 16[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762821:762988 [6] NCCL INFO Channel 08/0 : 22[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762817:762986 [2] NCCL INFO Channel 15/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104429:104594 [0] NCCL INFO Channel 02/0 : 56[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762819:762982 [4] NCCL INFO Channel 12/0 : 20[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762818:762985 [3] NCCL INFO Channel 10/0 : 19[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127290:127456 [0] NCCL INFO Channel 13/0 : 24[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127291:127455 [1] NCCL INFO Channel 03/0 : 9[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762815:762983 [0] NCCL INFO Channel 12/0 : 16[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624947:625105 [4] NCCL INFO Channel 13/0 : 44[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624943:625108 [0] NCCL INFO Channel 01/0 : 40[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127295:127452 [5] NCCL INFO Channel 03/0 : 13[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO Channel 10/0 : 60[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104433:104590 [4] NCCL INFO Channel 00/0 : 52[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762817:762986 [2] NCCL INFO Channel 00/0 : 18[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762821:762988 [6] NCCL INFO Channel 10/0 : 22[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127296:127451 [6] NCCL INFO Channel 11/0 : 14[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104429:104594 [0] NCCL INFO Channel 04/0 : 56[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762819:762982 [4] NCCL INFO Channel 14/0 : 20[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127297:127453 [7] NCCL INFO Channel 03/0 : 15[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127292:127458 [2] NCCL INFO Channel 13/0 : 10[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762818:762985 [3] NCCL INFO Channel 12/0 : 19[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127294:127457 [4] NCCL INFO Channel 05/0 : 44[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762815:762983 [0] NCCL INFO Channel 14/0 : 16[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127293:127454 [3] NCCL INFO Channel 09/0 : 11[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762817:762986 [2] NCCL INFO Channel 04/0 : 18[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO Channel 12/0 : 60[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762821:762988 [6] NCCL INFO Channel 12/0 : 22[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624947:625105 [4] NCCL INFO Channel 15/0 : 44[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104433:104590 [4] NCCL INFO Channel 01/0 : 52[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762818:762985 [3] NCCL INFO Channel 14/0 : 19[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104429:104594 [0] NCCL INFO Channel 06/0 : 56[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762817:762986 [2] NCCL INFO Channel 06/0 : 18[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127290:127456 [0] NCCL INFO Channel 15/0 : 24[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127291:127455 [1] NCCL INFO Channel 05/0 : 9[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762821:762988 [6] NCCL INFO Channel 14/0 : 22[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127295:127452 [5] NCCL INFO Channel 05/0 : 13[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762817:762986 [2] NCCL INFO Channel 08/0 : 18[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127296:127451 [6] NCCL INFO Channel 13/0 : 14[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO Channel 14/0 : 60[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762817:762986 [2] NCCL INFO Channel 10/0 : 18[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127297:127453 [7] NCCL INFO Channel 05/0 : 15[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762817:762986 [2] NCCL INFO Channel 12/0 : 18[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624943:625108 [0] NCCL INFO Channel 03/0 : 40[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127292:127458 [2] NCCL INFO Channel 15/0 : 10[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104433:104590 [4] NCCL INFO Channel 02/0 : 52[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762817:762986 [2] NCCL INFO Channel 14/0 : 18[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127293:127454 [3] NCCL INFO Channel 13/0 : 11[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127294:127457 [4] NCCL INFO Channel 07/0 : 44[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127291:127455 [1] NCCL INFO Channel 07/0 : 9[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104429:104594 [0] NCCL INFO Channel 08/0 : 56[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762819:762982 [4] NCCL INFO Channel 00/0 : 20[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127295:127452 [5] NCCL INFO Channel 07/0 : 13[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762819:762982 [4] NCCL INFO Channel 02/0 : 20[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762819:762982 [4] NCCL INFO Channel 06/0 : 20[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624943:625108 [0] NCCL INFO Channel 05/0 : 40[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762819:762982 [4] NCCL INFO Channel 08/0 : 20[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762819:762982 [4] NCCL INFO Channel 10/0 : 20[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104433:104590 [4] NCCL INFO Channel 03/0 : 52[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762815:762983 [0] NCCL INFO Channel 02/0 : 16[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104429:104594 [0] NCCL INFO Channel 10/0 : 56[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104433:104590 [4] NCCL INFO Channel 05/0 : 52[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762819:762982 [4] NCCL INFO Channel 12/0 : 20[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127296:127451 [6] NCCL INFO Channel 15/0 : 14[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762815:762983 [0] NCCL INFO Channel 04/0 : 16[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624943:625108 [0] NCCL INFO Channel 07/0 : 40[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127297:127453 [7] NCCL INFO Channel 07/0 : 15[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-167-9:1039537:1039703 [0] NCCL INFO Channel 02/0 : 56[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762819:762982 [4] NCCL INFO Channel 14/0 : 20[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127292:127458 [2] NCCL INFO Channel 01/0 : 26[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762815:762983 [0] NCCL INFO Channel 06/0 : 16[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762815:762983 [0] NCCL INFO Channel 08/0 : 16[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762815:762983 [0] NCCL INFO Channel 10/0 : 16[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104433:104590 [4] NCCL INFO Channel 06/0 : 52[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762819:762982 [4] NCCL INFO Channel 00/0 : 36[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624943:625108 [0] NCCL INFO Channel 09/0 : 40[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762815:762983 [0] NCCL INFO Channel 12/0 : 16[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127293:127454 [3] NCCL INFO Channel 15/0 : 11[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127291:127455 [1] NCCL INFO Channel 11/0 : 9[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127294:127457 [4] NCCL INFO Channel 09/0 : 44[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-167-9:1039537:1039703 [0] NCCL INFO Channel 04/0 : 56[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762819:762982 [4] NCCL INFO Channel 02/0 : 36[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127295:127452 [5] NCCL INFO Channel 09/0 : 13[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762815:762983 [0] NCCL INFO Channel 14/0 : 16[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127290:127456 [0] NCCL INFO Channel 01/0 : 40[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127296:127451 [6] NCCL INFO Channel 01/0 : 30[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762819:762982 [4] NCCL INFO Channel 06/0 : 36[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127297:127453 [7] NCCL INFO Channel 09/0 : 15[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104429:104594 [0] NCCL INFO Channel 12/0 : 56[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762819:762982 [4] NCCL INFO Channel 08/0 : 36[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127292:127458 [2] NCCL INFO Channel 03/0 : 26[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104433:104590 [4] NCCL INFO Channel 07/0 : 52[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762819:762982 [4] NCCL INFO Channel 10/0 : 36[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762815:762983 [0] NCCL INFO Channel 02/0 : 32[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624949:625110 [6] NCCL INFO Channel 00/0 : 6[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624943:625108 [0] NCCL INFO Channel 11/0 : 40[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-167-9:1039537:1039703 [0] NCCL INFO Channel 06/0 : 56[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762819:762982 [4] NCCL INFO Channel 12/0 : 36[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762815:762983 [0] NCCL INFO Channel 04/0 : 32[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127293:127454 [3] NCCL INFO Channel 01/0 : 27[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127294:127457 [4] NCCL INFO Channel 11/0 : 44[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762819:762982 [4] NCCL INFO Channel 14/0 : 36[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127291:127455 [1] NCCL INFO Channel 13/0 : 9[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104429:104594 [0] NCCL INFO Channel 14/0 : 56[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762815:762983 [0] NCCL INFO Channel 06/0 : 32[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127295:127452 [5] NCCL INFO Channel 11/0 : 13[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104433:104590 [4] NCCL INFO Channel 08/0 : 52[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762815:762983 [0] NCCL INFO Channel 08/0 : 32[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127296:127451 [6] NCCL INFO Channel 03/0 : 30[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127292:127458 [2] NCCL INFO Channel 05/0 : 26[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762815:762983 [0] NCCL INFO Channel 10/0 : 32[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624946:625107 [3] NCCL INFO Channel 00/0 : 3[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127290:127456 [0] NCCL INFO Channel 03/0 : 40[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127297:127453 [7] NCCL INFO Channel 11/0 : 15[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-167-9:1039537:1039703 [0] NCCL INFO Channel 08/0 : 56[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762815:762983 [0] NCCL INFO Channel 12/0 : 32[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762815:762983 [0] NCCL INFO Channel 14/0 : 32[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762817:762986 [2] NCCL INFO Channel 00/0 : 18[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104429:104594 [0] NCCL INFO Channel 01/0 : 48[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762821:762988 [6] NCCL INFO Channel 00/0 : 22[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104433:104590 [4] NCCL INFO Channel 09/0 : 52[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762817:762986 [2] NCCL INFO Channel 04/0 : 18[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624949:625110 [6] NCCL INFO Channel 02/0 : 6[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624943:625108 [0] NCCL INFO Channel 13/0 : 40[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127293:127454 [3] NCCL INFO Channel 03/0 : 27[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762818:762985 [3] NCCL INFO Channel 00/0 : 19[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624946:625107 [3] NCCL INFO Channel 02/0 : 3[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127291:127455 [1] NCCL INFO Channel 15/0 : 9[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762821:762988 [6] NCCL INFO Channel 02/0 : 22[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127294:127457 [4] NCCL INFO Channel 13/0 : 44[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-167-9:1039537:1039703 [0] NCCL INFO Channel 10/0 : 56[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762817:762986 [2] NCCL INFO Channel 06/0 : 18[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104429:104594 [0] NCCL INFO Channel 02/0 : 48[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762818:762985 [3] NCCL INFO Channel 02/0 : 19[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104433:104590 [4] NCCL INFO Channel 10/0 : 52[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762819:762982 [4] NCCL INFO Channel 00/0 : 28[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762817:762986 [2] NCCL INFO Channel 08/0 : 18[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624945:625111 [2] NCCL INFO Channel 00/0 : 2[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762821:762988 [6] NCCL INFO Channel 04/0 : 22[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624949:625110 [6] NCCL INFO Channel 04/0 : 6[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624946:625107 [3] NCCL INFO Channel 04/0 : 3[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624943:625108 [0] NCCL INFO Channel 15/0 : 40[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127295:127452 [5] NCCL INFO Channel 15/0 : 13[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762818:762985 [3] NCCL INFO Channel 04/0 : 19[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127296:127451 [6] NCCL INFO Channel 05/0 : 30[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127292:127458 [2] NCCL INFO Channel 07/0 : 26[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-167-9:1039537:1039703 [0] NCCL INFO Channel 12/0 : 56[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762817:762986 [2] NCCL INFO Channel 10/0 : 18[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127290:127456 [0] NCCL INFO Channel 05/0 : 40[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104429:104594 [0] NCCL INFO Channel 03/0 : 48[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762819:762982 [4] NCCL INFO Channel 02/0 : 28[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127297:127453 [7] NCCL INFO Channel 13/0 : 15[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127293:127454 [3] NCCL INFO Channel 05/0 : 27[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104433:104590 [4] NCCL INFO Channel 11/0 : 52[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762821:762988 [6] NCCL INFO Channel 08/0 : 22[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127291:127455 [1] NCCL INFO Channel 01/0 : 25[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762818:762985 [3] NCCL INFO Channel 06/0 : 19[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624945:625111 [2] NCCL INFO Channel 04/0 : 2[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127294:127457 [4] NCCL INFO Channel 15/0 : 44[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762817:762986 [2] NCCL INFO Channel 12/0 : 18[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127295:127452 [5] NCCL INFO Channel 01/0 : 29[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762819:762982 [4] NCCL INFO Channel 06/0 : 28[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-167-9:1039537:1039703 [0] NCCL INFO Channel 14/0 : 56[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762816:762987 [1] NCCL INFO Channel 00/0 : 17[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104429:104594 [0] NCCL INFO Channel 04/0 : 48[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104433:104590 [4] NCCL INFO Channel 13/0 : 52[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762821:762988 [6] NCCL INFO Channel 10/0 : 22[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762818:762985 [3] NCCL INFO Channel 08/0 : 19[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762820:762981 [5] NCCL INFO Channel 00/0 : 21[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762822:762984 [7] NCCL INFO Channel 00/0 : 23[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624949:625110 [6] NCCL INFO Channel 08/0 : 6[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624946:625107 [3] NCCL INFO Channel 06/0 : 3[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762817:762986 [2] NCCL INFO Channel 14/0 : 18[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624945:625111 [2] NCCL INFO Channel 06/0 : 2[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127296:127451 [6] NCCL INFO Channel 07/0 : 30[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127292:127458 [2] NCCL INFO Channel 09/0 : 26[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127290:127456 [0] NCCL INFO Channel 07/0 : 40[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127293:127454 [3] NCCL INFO Channel 07/0 : 27[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127297:127453 [7] NCCL INFO Channel 01/0 : 31[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127291:127455 [1] NCCL INFO Channel 03/0 : 25[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127294:127457 [4] NCCL INFO Channel 01/0 : 28[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127295:127452 [5] NCCL INFO Channel 03/0 : 29[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104433:104590 [4] NCCL INFO Channel 14/0 : 52[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104429:104594 [0] NCCL INFO Channel 05/0 : 48[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762819:762982 [4] NCCL INFO Channel 08/0 : 28[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624949:625110 [6] NCCL INFO Channel 10/0 : 6[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762816:762987 [1] NCCL INFO Channel 02/0 : 17[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762818:762985 [3] NCCL INFO Channel 10/0 : 19[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127296:127451 [6] NCCL INFO Channel 09/0 : 30[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762821:762988 [6] NCCL INFO Channel 12/0 : 22[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127292:127458 [2] NCCL INFO Channel 11/0 : 26[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127290:127456 [0] NCCL INFO Channel 09/0 : 40[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762822:762984 [7] NCCL INFO Channel 02/0 : 23[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762820:762981 [5] NCCL INFO Channel 02/0 : 21[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762815:762983 [0] NCCL INFO Channel 02/0 : 24[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624946:625107 [3] NCCL INFO Channel 08/0 : 3[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624945:625111 [2] NCCL INFO Channel 08/0 : 2[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104433:104590 [4] NCCL INFO Channel 15/0 : 52[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762819:762982 [4] NCCL INFO Channel 10/0 : 28[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624949:625110 [6] NCCL INFO Channel 12/0 : 6[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762818:762985 [3] NCCL INFO Channel 12/0 : 19[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762816:762987 [1] NCCL INFO Channel 04/0 : 17[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127293:127454 [3] NCCL INFO Channel 09/0 : 27[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762821:762988 [6] NCCL INFO Channel 14/0 : 22[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127291:127455 [1] NCCL INFO Channel 05/0 : 25[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127297:127453 [7] NCCL INFO Channel 03/0 : 31[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127294:127457 [4] NCCL INFO Channel 03/0 : 28[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127295:127452 [5] NCCL INFO Channel 05/0 : 29[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762820:762981 [5] NCCL INFO Channel 04/0 : 21[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762817:762986 [2] NCCL INFO Channel 00/0 : 34[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762815:762983 [0] NCCL INFO Channel 04/0 : 24[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762822:762984 [7] NCCL INFO Channel 04/0 : 23[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127296:127451 [6] NCCL INFO Channel 11/0 : 30[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127292:127458 [2] NCCL INFO Channel 13/0 : 26[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762819:762982 [4] NCCL INFO Channel 12/0 : 28[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762818:762985 [3] NCCL INFO Channel 14/0 : 19[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127290:127456 [0] NCCL INFO Channel 11/0 : 40[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104429:104594 [0] NCCL INFO Channel 06/0 : 48[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762816:762987 [1] NCCL INFO Channel 06/0 : 17[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127293:127454 [3] NCCL INFO Channel 13/0 : 27[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762820:762981 [5] NCCL INFO Channel 06/0 : 21[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762815:762983 [0] NCCL INFO Channel 06/0 : 24[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624946:625107 [3] NCCL INFO Channel 10/0 : 3[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624945:625111 [2] NCCL INFO Channel 10/0 : 2[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762822:762984 [7] NCCL INFO Channel 06/0 : 23[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762817:762986 [2] NCCL INFO Channel 04/0 : 34[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624944:625106 [1] NCCL INFO Channel 00/0 : 1[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762816:762987 [1] NCCL INFO Channel 08/0 : 17[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762819:762982 [4] NCCL INFO Channel 14/0 : 28[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624949:625110 [6] NCCL INFO Channel 14/0 : 6[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762821:762988 [6] NCCL INFO Channel 00/0 : 38[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104429:104594 [0] NCCL INFO Channel 07/0 : 48[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762820:762981 [5] NCCL INFO Channel 08/0 : 21[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762815:762983 [0] NCCL INFO Channel 08/0 : 24[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127291:127455 [1] NCCL INFO Channel 07/0 : 25[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762822:762984 [7] NCCL INFO Channel 08/0 : 23[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762816:762987 [1] NCCL INFO Channel 10/0 : 17[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762818:762985 [3] NCCL INFO Channel 00/0 : 35[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762819:762982 [4] NCCL INFO Channel 00/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624946:625107 [3] NCCL INFO Channel 12/0 : 3[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762817:762986 [2] NCCL INFO Channel 06/0 : 34[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624945:625111 [2] NCCL INFO Channel 12/0 : 2[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104429:104594 [0] NCCL INFO Channel 09/0 : 48[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762821:762988 [6] NCCL INFO Channel 02/0 : 38[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762820:762981 [5] NCCL INFO Channel 10/0 : 21[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624949:625110 [6] NCCL INFO Channel 00/0 : 38[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762815:762983 [0] NCCL INFO Channel 10/0 : 24[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127295:127452 [5] NCCL INFO Channel 07/0 : 29[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127297:127453 [7] NCCL INFO Channel 05/0 : 31[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127294:127457 [4] NCCL INFO Channel 05/0 : 28[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762822:762984 [7] NCCL INFO Channel 10/0 : 23[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127296:127451 [6] NCCL INFO Channel 13/0 : 30[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762816:762987 [1] NCCL INFO Channel 12/0 : 17[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127292:127458 [2] NCCL INFO Channel 15/0 : 26[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127290:127456 [0] NCCL INFO Channel 13/0 : 40[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762818:762985 [3] NCCL INFO Channel 02/0 : 35[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127293:127454 [3] NCCL INFO Channel 15/0 : 27[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127291:127455 [1] NCCL INFO Channel 11/0 : 25[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762817:762986 [2] NCCL INFO Channel 08/0 : 34[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762819:762982 [4] NCCL INFO Channel 01/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104429:104594 [0] NCCL INFO Channel 10/0 : 48[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762815:762983 [0] NCCL INFO Channel 12/0 : 24[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762821:762988 [6] NCCL INFO Channel 04/0 : 38[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762820:762981 [5] NCCL INFO Channel 12/0 : 21[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624944:625106 [1] NCCL INFO Channel 02/0 : 1[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762822:762984 [7] NCCL INFO Channel 12/0 : 23[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762816:762987 [1] NCCL INFO Channel 14/0 : 17[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624946:625107 [3] NCCL INFO Channel 14/0 : 3[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762818:762985 [3] NCCL INFO Channel 04/0 : 35[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624948:625104 [5] NCCL INFO Channel 00/0 : 5[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624945:625111 [2] NCCL INFO Channel 14/0 : 2[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624949:625110 [6] NCCL INFO Channel 02/0 : 38[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624950:625109 [7] NCCL INFO Channel 00/0 : 7[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762819:762982 [4] NCCL INFO Channel 02/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762817:762986 [2] NCCL INFO Channel 10/0 : 34[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104429:104594 [0] NCCL INFO Channel 11/0 : 48[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762820:762981 [5] NCCL INFO Channel 14/0 : 21[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762815:762983 [0] NCCL INFO Channel 14/0 : 24[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762821:762988 [6] NCCL INFO Channel 08/0 : 38[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762822:762984 [7] NCCL INFO Channel 14/0 : 23[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762818:762985 [3] NCCL INFO Channel 06/0 : 35[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127294:127457 [4] NCCL INFO Channel 07/0 : 28[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127297:127453 [7] NCCL INFO Channel 07/0 : 31[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624944:625106 [1] NCCL INFO Channel 04/0 : 1[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624946:625107 [3] NCCL INFO Channel 00/0 : 35[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127295:127452 [5] NCCL INFO Channel 09/0 : 29[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127296:127451 [6] NCCL INFO Channel 15/0 : 30[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127291:127455 [1] NCCL INFO Channel 13/0 : 25[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127290:127456 [0] NCCL INFO Channel 15/0 : 40[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104429:104594 [0] NCCL INFO Channel 12/0 : 48[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762819:762982 [4] NCCL INFO Channel 03/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762817:762986 [2] NCCL INFO Channel 12/0 : 34[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624949:625110 [6] NCCL INFO Channel 04/0 : 38[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762815:762983 [0] NCCL INFO Channel 01/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624945:625111 [2] NCCL INFO Channel 00/0 : 34[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762821:762988 [6] NCCL INFO Channel 10/0 : 38[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624948:625104 [5] NCCL INFO Channel 02/0 : 5[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624950:625109 [7] NCCL INFO Channel 02/0 : 7[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127297:127453 [7] NCCL INFO Channel 09/0 : 31[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127294:127457 [4] NCCL INFO Channel 09/0 : 28[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762816:762987 [1] NCCL INFO Channel 00/0 : 33[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762818:762985 [3] NCCL INFO Channel 08/0 : 35[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624944:625106 [1] NCCL INFO Channel 06/0 : 1[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127291:127455 [1] NCCL INFO Channel 15/0 : 25[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127290:127456 [0] NCCL INFO Channel 01/0 : 24[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104429:104594 [0] NCCL INFO Channel 13/0 : 48[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762819:762982 [4] NCCL INFO Channel 05/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127295:127452 [5] NCCL INFO Channel 11/0 : 29[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762817:762986 [2] NCCL INFO Channel 14/0 : 34[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762815:762983 [0] NCCL INFO Channel 02/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9628:9786 [3] NCCL INFO Channel 01/0 : 27[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762821:762988 [6] NCCL INFO Channel 12/0 : 38[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762818:762985 [3] NCCL INFO Channel 10/0 : 35[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624946:625107 [3] NCCL INFO Channel 02/0 : 35[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104429:104594 [0] NCCL INFO Channel 14/0 : 48[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762816:762987 [1] NCCL INFO Channel 02/0 : 33[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624949:625110 [6] NCCL INFO Channel 08/0 : 38[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762819:762982 [4] NCCL INFO Channel 06/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762820:762981 [5] NCCL INFO Channel 00/0 : 37[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127297:127453 [7] NCCL INFO Channel 11/0 : 31[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762822:762984 [7] NCCL INFO Channel 00/0 : 39[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9627:9791 [2] NCCL INFO Channel 01/0 : 26[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127294:127457 [4] NCCL INFO Channel 11/0 : 28[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762815:762983 [0] NCCL INFO Channel 03/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127295:127452 [5] NCCL INFO Channel 15/0 : 29[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762821:762988 [6] NCCL INFO Channel 14/0 : 38[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127297:127453 [7] NCCL INFO Channel 13/0 : 31[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762818:762985 [3] NCCL INFO Channel 12/0 : 35[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127294:127457 [4] NCCL INFO Channel 13/0 : 28[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104429:104594 [0] NCCL INFO Channel 15/0 : 48[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762816:762987 [1] NCCL INFO Channel 04/0 : 33[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624945:625111 [2] NCCL INFO Channel 04/0 : 34[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624948:625104 [5] NCCL INFO Channel 04/0 : 5[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762819:762982 [4] NCCL INFO Channel 07/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762815:762983 [0] NCCL INFO Channel 04/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762822:762984 [7] NCCL INFO Channel 02/0 : 39[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762820:762981 [5] NCCL INFO Channel 02/0 : 37[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9628:9786 [3] NCCL INFO Channel 03/0 : 27[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624950:625109 [7] NCCL INFO Channel 04/0 : 7[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762818:762985 [3] NCCL INFO Channel 14/0 : 35[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9627:9791 [2] NCCL INFO Channel 03/0 : 26[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624944:625106 [1] NCCL INFO Channel 08/0 : 1[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762816:762987 [1] NCCL INFO Channel 06/0 : 33[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9631:9788 [6] NCCL INFO Channel 01/0 : 30[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624946:625107 [3] NCCL INFO Channel 04/0 : 35[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624949:625110 [6] NCCL INFO Channel 10/0 : 38[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127290:127456 [0] NCCL INFO Channel 03/0 : 24[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127293:127454 [3] NCCL INFO Channel 01/0 : 59[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127292:127458 [2] NCCL INFO Channel 01/0 : 58[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127294:127457 [4] NCCL INFO Channel 15/0 : 28[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762815:762983 [0] NCCL INFO Channel 05/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762819:762982 [4] NCCL INFO Channel 08/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9628:9786 [3] NCCL INFO Channel 05/0 : 27[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762820:762981 [5] NCCL INFO Channel 04/0 : 37[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762822:762984 [7] NCCL INFO Channel 04/0 : 39[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9627:9791 [2] NCCL INFO Channel 05/0 : 26[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624948:625104 [5] NCCL INFO Channel 06/0 : 5[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762816:762987 [1] NCCL INFO Channel 08/0 : 33[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624950:625109 [7] NCCL INFO Channel 06/0 : 7[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624945:625111 [2] NCCL INFO Channel 06/0 : 34[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762815:762983 [0] NCCL INFO Channel 06/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624949:625110 [6] NCCL INFO Channel 12/0 : 38[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624946:625107 [3] NCCL INFO Channel 06/0 : 35[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624944:625106 [1] NCCL INFO Channel 10/0 : 1[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127292:127458 [2] NCCL INFO Channel 03/0 : 58[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762819:762982 [4] NCCL INFO Channel 09/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127290:127456 [0] NCCL INFO Channel 05/0 : 24[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762820:762981 [5] NCCL INFO Channel 06/0 : 37[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127296:127451 [6] NCCL INFO Channel 01/0 : 62[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127293:127454 [3] NCCL INFO Channel 03/0 : 59[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762822:762984 [7] NCCL INFO Channel 06/0 : 39[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9631:9788 [6] NCCL INFO Channel 03/0 : 30[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762816:762987 [1] NCCL INFO Channel 10/0 : 33[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9626:9790 [1] NCCL INFO Channel 01/0 : 25[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762815:762983 [0] NCCL INFO Channel 07/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9628:9786 [3] NCCL INFO Channel 07/0 : 27[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624948:625104 [5] NCCL INFO Channel 08/0 : 5[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762819:762982 [4] NCCL INFO Channel 10/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624950:625109 [7] NCCL INFO Channel 08/0 : 7[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127292:127458 [2] NCCL INFO Channel 05/0 : 58[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127291:127455 [1] NCCL INFO Channel 01/0 : 57[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127290:127456 [0] NCCL INFO Channel 07/0 : 24[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127296:127451 [6] NCCL INFO Channel 03/0 : 62[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9627:9791 [2] NCCL INFO Channel 07/0 : 26[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127292:127458 [2] NCCL INFO Channel 07/0 : 58[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127293:127454 [3] NCCL INFO Channel 05/0 : 59[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762820:762981 [5] NCCL INFO Channel 08/0 : 37[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762816:762987 [1] NCCL INFO Channel 12/0 : 33[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9630:9789 [5] NCCL INFO Channel 01/0 : 29[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9631:9788 [6] NCCL INFO Channel 05/0 : 30[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762822:762984 [7] NCCL INFO Channel 08/0 : 39[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9626:9790 [1] NCCL INFO Channel 03/0 : 25[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624945:625111 [2] NCCL INFO Channel 08/0 : 34[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624949:625110 [6] NCCL INFO Channel 14/0 : 38[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624946:625107 [3] NCCL INFO Channel 08/0 : 35[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762815:762983 [0] NCCL INFO Channel 09/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624944:625106 [1] NCCL INFO Channel 12/0 : 1[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762819:762982 [4] NCCL INFO Channel 11/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624948:625104 [5] NCCL INFO Channel 10/0 : 5[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762816:762987 [1] NCCL INFO Channel 14/0 : 33[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127290:127456 [0] NCCL INFO Channel 09/0 : 24[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9632:9792 [7] NCCL INFO Channel 01/0 : 31[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9628:9786 [3] NCCL INFO Channel 09/0 : 27[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127291:127455 [1] NCCL INFO Channel 03/0 : 57[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127296:127451 [6] NCCL INFO Channel 05/0 : 62[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9627:9791 [2] NCCL INFO Channel 09/0 : 26[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127295:127452 [5] NCCL INFO Channel 01/0 : 61[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9630:9789 [5] NCCL INFO Channel 03/0 : 29[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127292:127458 [2] NCCL INFO Channel 09/0 : 58[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9631:9788 [6] NCCL INFO Channel 07/0 : 30[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624950:625109 [7] NCCL INFO Channel 10/0 : 7[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762820:762981 [5] NCCL INFO Channel 10/0 : 37[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762822:762984 [7] NCCL INFO Channel 10/0 : 39[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762815:762983 [0] NCCL INFO Channel 10/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762819:762982 [4] NCCL INFO Channel 13/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624945:625111 [2] NCCL INFO Channel 10/0 : 34[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762820:762981 [5] NCCL INFO Channel 12/0 : 37[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624944:625106 [1] NCCL INFO Channel 14/0 : 1[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624946:625107 [3] NCCL INFO Channel 10/0 : 35[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624948:625104 [5] NCCL INFO Channel 12/0 : 5[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127297:127453 [7] NCCL INFO Channel 01/0 : 63[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624950:625109 [7] NCCL INFO Channel 12/0 : 7[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127290:127456 [0] NCCL INFO Channel 11/0 : 24[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9626:9790 [1] NCCL INFO Channel 05/0 : 25[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624945:625111 [2] NCCL INFO Channel 12/0 : 34[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127293:127454 [3] NCCL INFO Channel 07/0 : 59[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9632:9792 [7] NCCL INFO Channel 03/0 : 31[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9628:9786 [3] NCCL INFO Channel 13/0 : 27[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9627:9791 [2] NCCL INFO Channel 11/0 : 26[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127291:127455 [1] NCCL INFO Channel 05/0 : 57[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762815:762983 [0] NCCL INFO Channel 11/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9630:9789 [5] NCCL INFO Channel 05/0 : 29[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762819:762982 [4] NCCL INFO Channel 14/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762822:762984 [7] NCCL INFO Channel 12/0 : 39[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762820:762981 [5] NCCL INFO Channel 14/0 : 37[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-46:702886:703049 [6] NCCL INFO Channel 01/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624944:625106 [1] NCCL INFO Channel 00/0 : 33[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127296:127451 [6] NCCL INFO Channel 07/0 : 62[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127295:127452 [5] NCCL INFO Channel 03/0 : 61[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127292:127458 [2] NCCL INFO Channel 11/0 : 58[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9631:9788 [6] NCCL INFO Channel 09/0 : 30[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9629:9793 [4] NCCL INFO Channel 00/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624948:625104 [5] NCCL INFO Channel 14/0 : 5[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127290:127456 [0] NCCL INFO Channel 13/0 : 24[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127294:127457 [4] NCCL INFO Channel 00/0 : 28[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127297:127453 [7] NCCL INFO Channel 03/0 : 63[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9626:9790 [1] NCCL INFO Channel 07/0 : 25[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624946:625107 [3] NCCL INFO Channel 12/0 : 35[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127293:127454 [3] NCCL INFO Channel 09/0 : 59[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127291:127455 [1] NCCL INFO Channel 07/0 : 57[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762815:762983 [0] NCCL INFO Channel 12/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9627:9791 [2] NCCL INFO Channel 13/0 : 26[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624950:625109 [7] NCCL INFO Channel 14/0 : 7[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-46:702886:703049 [6] NCCL INFO Channel 03/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762819:762982 [4] NCCL INFO Channel 15/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762822:762984 [7] NCCL INFO Channel 14/0 : 39[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9628:9786 [3] NCCL INFO Channel 15/0 : 27[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624945:625111 [2] NCCL INFO Channel 14/0 : 34[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9632:9792 [7] NCCL INFO Channel 05/0 : 31[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624949:625110 [6] NCCL INFO Channel 00/0 : 54[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127296:127451 [6] NCCL INFO Channel 09/0 : 62[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624944:625106 [1] NCCL INFO Channel 02/0 : 33[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762815:762983 [0] NCCL INFO Channel 13/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-46:702886:703049 [6] NCCL INFO Channel 05/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762815:762983 [0] NCCL INFO Channel 14/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9630:9789 [5] NCCL INFO Channel 07/0 : 29[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9631:9788 [6] NCCL INFO Channel 11/0 : 30[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9626:9790 [1] NCCL INFO Channel 11/0 : 25[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127295:127452 [5] NCCL INFO Channel 05/0 : 61[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127292:127458 [2] NCCL INFO Channel 13/0 : 58[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9629:9793 [4] NCCL INFO Channel 01/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624948:625104 [5] NCCL INFO Channel 00/0 : 37[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127290:127456 [0] NCCL INFO Channel 15/0 : 24[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9627:9791 [2] NCCL INFO Channel 15/0 : 26[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624946:625107 [3] NCCL INFO Channel 14/0 : 35[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127297:127453 [7] NCCL INFO Channel 05/0 : 63[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127291:127455 [1] NCCL INFO Channel 11/0 : 57[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127294:127457 [4] NCCL INFO Channel 02/0 : 28[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9632:9792 [7] NCCL INFO Channel 07/0 : 31[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9630:9789 [5] NCCL INFO Channel 09/0 : 29[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624950:625109 [7] NCCL INFO Channel 00/0 : 39[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127296:127451 [6] NCCL INFO Channel 11/0 : 62[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-46:702886:703049 [6] NCCL INFO Channel 07/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762815:762983 [0] NCCL INFO Channel 15/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624949:625110 [6] NCCL INFO Channel 02/0 : 54[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624948:625104 [5] NCCL INFO Channel 02/0 : 37[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624944:625106 [1] NCCL INFO Channel 04/0 : 33[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624950:625109 [7] NCCL INFO Channel 02/0 : 39[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9631:9788 [6] NCCL INFO Channel 13/0 : 30[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9626:9790 [1] NCCL INFO Channel 13/0 : 25[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9629:9793 [4] NCCL INFO Channel 02/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127293:127454 [3] NCCL INFO Channel 13/0 : 59[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9632:9792 [7] NCCL INFO Channel 09/0 : 31[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9630:9789 [5] NCCL INFO Channel 11/0 : 29[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127295:127452 [5] NCCL INFO Channel 07/0 : 61[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127292:127458 [2] NCCL INFO Channel 15/0 : 58[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9631:9788 [6] NCCL INFO Channel 15/0 : 30[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127297:127453 [7] NCCL INFO Channel 07/0 : 63[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-46:702882:703046 [2] NCCL INFO Channel 01/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-46:702886:703049 [6] NCCL INFO Channel 09/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624949:625110 [6] NCCL INFO Channel 04/0 : 54[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624948:625104 [5] NCCL INFO Channel 04/0 : 37[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-46:702883:703044 [3] NCCL INFO Channel 01/0 : 11[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624944:625106 [1] NCCL INFO Channel 06/0 : 33[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624950:625109 [7] NCCL INFO Channel 04/0 : 39[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9629:9793 [4] NCCL INFO Channel 03/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9626:9790 [1] NCCL INFO Channel 15/0 : 25[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9632:9792 [7] NCCL INFO Channel 11/0 : 31[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624946:625107 [3] NCCL INFO Channel 00/0 : 51[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127291:127455 [1] NCCL INFO Channel 13/0 : 57[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127296:127451 [6] NCCL INFO Channel 13/0 : 62[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127294:127457 [4] NCCL INFO Channel 06/0 : 28[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9630:9789 [5] NCCL INFO Channel 15/0 : 29[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624945:625111 [2] NCCL INFO Channel 00/0 : 50[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127292:127458 [2] NCCL INFO Channel 01/0 : 26[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-46:702882:703046 [2] NCCL INFO Channel 03/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127295:127452 [5] NCCL INFO Channel 09/0 : 61[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127293:127454 [3] NCCL INFO Channel 15/0 : 59[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-46:702886:703049 [6] NCCL INFO Channel 11/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127297:127453 [7] NCCL INFO Channel 09/0 : 63[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-46:702883:703044 [3] NCCL INFO Channel 03/0 : 11[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9629:9793 [4] NCCL INFO Channel 05/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9632:9792 [7] NCCL INFO Channel 13/0 : 31[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624948:625104 [5] NCCL INFO Channel 06/0 : 37[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624949:625110 [6] NCCL INFO Channel 08/0 : 54[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624950:625109 [7] NCCL INFO Channel 06/0 : 39[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624944:625106 [1] NCCL INFO Channel 08/0 : 33[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127291:127455 [1] NCCL INFO Channel 15/0 : 57[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127296:127451 [6] NCCL INFO Channel 15/0 : 62[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127294:127457 [4] NCCL INFO Channel 08/0 : 28[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-46:702882:703046 [2] NCCL INFO Channel 05/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624946:625107 [3] NCCL INFO Channel 02/0 : 51[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127292:127458 [2] NCCL INFO Channel 03/0 : 26[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127293:127454 [3] NCCL INFO Channel 01/0 : 27[3] -> 59[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9629:9793 [4] NCCL INFO Channel 06/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624945:625111 [2] NCCL INFO Channel 04/0 : 50[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127295:127452 [5] NCCL INFO Channel 11/0 : 61[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-46:702886:703049 [6] NCCL INFO Channel 13/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9625:9787 [0] NCCL INFO Channel 01/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127297:127453 [7] NCCL INFO Channel 11/0 : 63[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-46:702883:703044 [3] NCCL INFO Channel 05/0 : 11[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-46:702882:703046 [2] NCCL INFO Channel 07/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-46:702886:703049 [6] NCCL INFO Channel 15/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624949:625110 [6] NCCL INFO Channel 10/0 : 54[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624948:625104 [5] NCCL INFO Channel 08/0 : 37[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-46:702883:703044 [3] NCCL INFO Channel 07/0 : 11[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9629:9793 [4] NCCL INFO Channel 07/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624950:625109 [7] NCCL INFO Channel 08/0 : 39[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127296:127451 [6] NCCL INFO Channel 01/0 : 30[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9625:9787 [0] NCCL INFO Channel 02/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624944:625106 [1] NCCL INFO Channel 10/0 : 33[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624945:625111 [2] NCCL INFO Channel 06/0 : 50[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624946:625107 [3] NCCL INFO Channel 04/0 : 51[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127294:127457 [4] NCCL INFO Channel 10/0 : 28[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897242:897400 [2] NCCL INFO Channel 13/0 : 26[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9629:9793 [4] NCCL INFO Channel 08/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127291:127455 [1] NCCL INFO Channel 01/0 : 25[1] -> 57[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-46:702882:703046 [2] NCCL INFO Channel 09/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897243:897403 [3] NCCL INFO Channel 13/0 : 27[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897241:897402 [1] NCCL INFO Channel 13/0 : 25[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127292:127458 [2] NCCL INFO Channel 05/0 : 26[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127290:127456 [0] NCCL INFO Channel 02/0 : 24[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127293:127454 [3] NCCL INFO Channel 03/0 : 27[3] -> 59[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897242:897400 [2] NCCL INFO Channel 15/0 : 26[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127295:127452 [5] NCCL INFO Channel 15/0 : 61[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-46:702883:703044 [3] NCCL INFO Channel 09/0 : 11[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897243:897403 [3] NCCL INFO Channel 15/0 : 27[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897241:897402 [1] NCCL INFO Channel 15/0 : 25[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624949:625110 [6] NCCL INFO Channel 12/0 : 54[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624948:625104 [5] NCCL INFO Channel 10/0 : 37[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127297:127453 [7] NCCL INFO Channel 13/0 : 63[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127296:127451 [6] NCCL INFO Channel 03/0 : 30[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897244:897407 [4] NCCL INFO Channel 01/0 : 44[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9625:9787 [0] NCCL INFO Channel 03/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624950:625109 [7] NCCL INFO Channel 10/0 : 39[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-46:702882:703046 [2] NCCL INFO Channel 11/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897244:897407 [4] NCCL INFO Channel 03/0 : 44[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9629:9793 [4] NCCL INFO Channel 09/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624944:625106 [1] NCCL INFO Channel 12/0 : 33[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897244:897407 [4] NCCL INFO Channel 05/0 : 44[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624946:625107 [3] NCCL INFO Channel 06/0 : 51[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897244:897407 [4] NCCL INFO Channel 07/0 : 44[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624945:625111 [2] NCCL INFO Channel 08/0 : 50[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-46:702883:703044 [3] NCCL INFO Channel 13/0 : 11[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897244:897407 [4] NCCL INFO Channel 09/0 : 44[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127294:127457 [4] NCCL INFO Channel 12/0 : 28[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897240:897406 [0] NCCL INFO Channel 01/0 : 40[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127291:127455 [1] NCCL INFO Channel 03/0 : 25[1] -> 57[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897244:897407 [4] NCCL INFO Channel 11/0 : 44[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897240:897406 [0] NCCL INFO Channel 03/0 : 40[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9625:9787 [0] NCCL INFO Channel 04/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127292:127458 [2] NCCL INFO Channel 07/0 : 26[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-46:702882:703046 [2] NCCL INFO Channel 13/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897244:897407 [4] NCCL INFO Channel 13/0 : 44[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897240:897406 [0] NCCL INFO Channel 05/0 : 40[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9629:9793 [4] NCCL INFO Channel 10/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127293:127454 [3] NCCL INFO Channel 05/0 : 27[3] -> 59[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897240:897406 [0] NCCL INFO Channel 07/0 : 40[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897244:897407 [4] NCCL INFO Channel 15/0 : 44[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9625:9787 [0] NCCL INFO Channel 05/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624949:625110 [6] NCCL INFO Channel 14/0 : 54[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624948:625104 [5] NCCL INFO Channel 12/0 : 37[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127295:127452 [5] NCCL INFO Channel 01/0 : 29[5] -> 61[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897240:897406 [0] NCCL INFO Channel 09/0 : 40[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624950:625109 [7] NCCL INFO Channel 12/0 : 39[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127297:127453 [7] NCCL INFO Channel 01/0 : 31[7] -> 63[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127296:127451 [6] NCCL INFO Channel 05/0 : 30[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-46:702883:703044 [3] NCCL INFO Channel 15/0 : 11[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897240:897406 [0] NCCL INFO Channel 11/0 : 40[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624944:625106 [1] NCCL INFO Channel 14/0 : 33[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127290:127456 [0] NCCL INFO Channel 04/0 : 24[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897240:897406 [0] NCCL INFO Channel 13/0 : 40[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624945:625111 [2] NCCL INFO Channel 10/0 : 50[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624946:625107 [3] NCCL INFO Channel 08/0 : 51[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127291:127455 [1] NCCL INFO Channel 05/0 : 25[1] -> 57[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-46:702882:703046 [2] NCCL INFO Channel 15/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897240:897406 [0] NCCL INFO Channel 15/0 : 40[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9629:9793 [4] NCCL INFO Channel 11/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127294:127457 [4] NCCL INFO Channel 14/0 : 28[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897243:897403 [3] NCCL INFO Channel 01/0 : 43[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897242:897400 [2] NCCL INFO Channel 01/0 : 42[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9625:9787 [0] NCCL INFO Channel 06/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127292:127458 [2] NCCL INFO Channel 09/0 : 26[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897242:897400 [2] NCCL INFO Channel 03/0 : 42[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897243:897403 [3] NCCL INFO Channel 03/0 : 43[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897246:897405 [6] NCCL INFO Channel 01/0 : 46[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624949:625110 [6] NCCL INFO Channel 00/0 : 38[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624948:625104 [5] NCCL INFO Channel 14/0 : 37[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897242:897400 [2] NCCL INFO Channel 05/0 : 42[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624950:625109 [7] NCCL INFO Channel 14/0 : 39[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897243:897403 [3] NCCL INFO Channel 05/0 : 43[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624945:625111 [2] NCCL INFO Channel 12/0 : 50[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624946:625107 [3] NCCL INFO Channel 10/0 : 51[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897241:897402 [1] NCCL INFO Channel 01/0 : 41[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9629:9793 [4] NCCL INFO Channel 13/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624949:625110 [6] NCCL INFO Channel 02/0 : 38[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897246:897405 [6] NCCL INFO Channel 03/0 : 46[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897242:897400 [2] NCCL INFO Channel 07/0 : 42[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9625:9787 [0] NCCL INFO Channel 07/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127293:127454 [3] NCCL INFO Channel 07/0 : 27[3] -> 59[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897244:897407 [4] NCCL INFO Channel 00/0 : 52[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127295:127452 [5] NCCL INFO Channel 03/0 : 29[5] -> 61[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897243:897403 [3] NCCL INFO Channel 07/0 : 43[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127296:127451 [6] NCCL INFO Channel 07/0 : 30[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127297:127453 [7] NCCL INFO Channel 03/0 : 31[7] -> 63[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897245:897401 [5] NCCL INFO Channel 01/0 : 45[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127291:127455 [1] NCCL INFO Channel 07/0 : 25[1] -> 57[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897241:897402 [1] NCCL INFO Channel 03/0 : 41[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624945:625111 [2] NCCL INFO Channel 14/0 : 50[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624946:625107 [3] NCCL INFO Channel 12/0 : 51[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127290:127456 [0] NCCL INFO Channel 06/0 : 24[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127293:127454 [3] NCCL INFO Channel 09/0 : 27[3] -> 59[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127292:127458 [2] NCCL INFO Channel 11/0 : 26[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127295:127452 [5] NCCL INFO Channel 05/0 : 29[5] -> 61[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897246:897405 [6] NCCL INFO Channel 05/0 : 46[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9629:9793 [4] NCCL INFO Channel 14/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624944:625106 [1] NCCL INFO Channel 00/0 : 49[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624949:625110 [6] NCCL INFO Channel 04/0 : 38[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127296:127451 [6] NCCL INFO Channel 09/0 : 30[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-46:702881:703043 [1] NCCL INFO Channel 01/0 : 9[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897242:897400 [2] NCCL INFO Channel 09/0 : 42[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897247:897404 [7] NCCL INFO Channel 01/0 : 47[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9625:9787 [0] NCCL INFO Channel 09/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624945:625111 [2] NCCL INFO Channel 00/0 : 34[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897244:897407 [4] NCCL INFO Channel 01/0 : 52[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624946:625107 [3] NCCL INFO Channel 14/0 : 51[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897245:897401 [5] NCCL INFO Channel 03/0 : 45[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-46:702885:703048 [5] NCCL INFO Channel 01/0 : 13[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897246:897405 [6] NCCL INFO Channel 07/0 : 46[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897241:897402 [1] NCCL INFO Channel 05/0 : 41[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897243:897403 [3] NCCL INFO Channel 09/0 : 43[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127291:127455 [1] NCCL INFO Channel 11/0 : 25[1] -> 57[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127297:127453 [7] NCCL INFO Channel 05/0 : 31[7] -> 63[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897247:897404 [7] NCCL INFO Channel 03/0 : 47[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897242:897400 [2] NCCL INFO Channel 11/0 : 42[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127293:127454 [3] NCCL INFO Channel 13/0 : 27[3] -> 59[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-46:702887:703045 [7] NCCL INFO Channel 01/0 : 15[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-46:702881:703043 [1] NCCL INFO Channel 03/0 : 9[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897244:897407 [4] NCCL INFO Channel 02/0 : 52[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9629:9793 [4] NCCL INFO Channel 15/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127292:127458 [2] NCCL INFO Channel 13/0 : 26[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897245:897401 [5] NCCL INFO Channel 05/0 : 45[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9625:9787 [0] NCCL INFO Channel 10/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624949:625110 [6] NCCL INFO Channel 08/0 : 38[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127290:127456 [0] NCCL INFO Channel 08/0 : 24[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127295:127452 [5] NCCL INFO Channel 07/0 : 29[5] -> 61[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127296:127451 [6] NCCL INFO Channel 11/0 : 30[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897246:897405 [6] NCCL INFO Channel 09/0 : 46[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9629:9793 [4] NCCL INFO Channel 01/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624944:625106 [1] NCCL INFO Channel 02/0 : 49[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624948:625104 [5] NCCL INFO Channel 00/0 : 53[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127291:127455 [1] NCCL INFO Channel 13/0 : 25[1] -> 57[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-46:702885:703048 [5] NCCL INFO Channel 03/0 : 13[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897241:897402 [1] NCCL INFO Channel 07/0 : 41[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897243:897403 [3] NCCL INFO Channel 13/0 : 43[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624950:625109 [7] NCCL INFO Channel 00/0 : 55[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624945:625111 [2] NCCL INFO Channel 04/0 : 34[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127297:127453 [7] NCCL INFO Channel 07/0 : 31[7] -> 63[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897247:897404 [7] NCCL INFO Channel 05/0 : 47[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624946:625107 [3] NCCL INFO Channel 00/0 : 35[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624949:625110 [6] NCCL INFO Channel 10/0 : 38[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897242:897400 [2] NCCL INFO Channel 13/0 : 42[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897244:897407 [4] NCCL INFO Channel 03/0 : 52[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897245:897401 [5] NCCL INFO Channel 07/0 : 45[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-46:702887:703045 [7] NCCL INFO Channel 03/0 : 15[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-46:702881:703043 [1] NCCL INFO Channel 05/0 : 9[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897246:897405 [6] NCCL INFO Channel 11/0 : 46[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9625:9787 [0] NCCL INFO Channel 11/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897241:897402 [1] NCCL INFO Channel 11/0 : 41[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9629:9793 [4] NCCL INFO Channel 03/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127293:127454 [3] NCCL INFO Channel 15/0 : 27[3] -> 59[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-46:702885:703048 [5] NCCL INFO Channel 05/0 : 13[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897243:897403 [3] NCCL INFO Channel 15/0 : 43[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624944:625106 [1] NCCL INFO Channel 04/0 : 49[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127292:127458 [2] NCCL INFO Channel 15/0 : 26[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897247:897404 [7] NCCL INFO Channel 07/0 : 47[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624948:625104 [5] NCCL INFO Channel 02/0 : 53[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127290:127456 [0] NCCL INFO Channel 10/0 : 24[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897242:897400 [2] NCCL INFO Channel 15/0 : 42[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897245:897401 [5] NCCL INFO Channel 09/0 : 45[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897246:897405 [6] NCCL INFO Channel 13/0 : 46[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624950:625109 [7] NCCL INFO Channel 02/0 : 55[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127296:127451 [6] NCCL INFO Channel 13/0 : 30[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127295:127452 [5] NCCL INFO Channel 09/0 : 29[5] -> 61[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897244:897407 [4] NCCL INFO Channel 05/0 : 52[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897241:897402 [1] NCCL INFO Channel 13/0 : 41[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897247:897404 [7] NCCL INFO Channel 09/0 : 47[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9625:9787 [0] NCCL INFO Channel 12/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624945:625111 [2] NCCL INFO Channel 06/0 : 34[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127291:127455 [1] NCCL INFO Channel 15/0 : 25[1] -> 57[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127297:127453 [7] NCCL INFO Channel 09/0 : 31[7] -> 63[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-46:702887:703045 [7] NCCL INFO Channel 05/0 : 15[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-46:702881:703043 [1] NCCL INFO Channel 07/0 : 9[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897240:897406 [0] NCCL INFO Channel 01/0 : 48[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9629:9793 [4] NCCL INFO Channel 05/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127290:127456 [0] NCCL INFO Channel 12/0 : 24[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-46:702885:703048 [5] NCCL INFO Channel 07/0 : 13[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897245:897401 [5] NCCL INFO Channel 11/0 : 45[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897246:897405 [6] NCCL INFO Channel 15/0 : 46[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9625:9787 [0] NCCL INFO Channel 13/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897244:897407 [4] NCCL INFO Channel 06/0 : 52[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897247:897404 [7] NCCL INFO Channel 11/0 : 47[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897241:897402 [1] NCCL INFO Channel 15/0 : 41[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624949:625110 [6] NCCL INFO Channel 12/0 : 38[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624946:625107 [3] NCCL INFO Channel 02/0 : 35[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897240:897406 [0] NCCL INFO Channel 02/0 : 48[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624944:625106 [1] NCCL INFO Channel 06/0 : 49[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-167-9:1039540:1039705 [3] NCCL INFO Channel 00/0 : 59[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897245:897401 [5] NCCL INFO Channel 15/0 : 45[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624948:625104 [5] NCCL INFO Channel 04/0 : 53[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-46:702887:703045 [7] NCCL INFO Channel 07/0 : 15[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897244:897407 [4] NCCL INFO Channel 07/0 : 52[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624950:625109 [7] NCCL INFO Channel 04/0 : 55[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127296:127451 [6] NCCL INFO Channel 15/0 : 30[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127295:127452 [5] NCCL INFO Channel 11/0 : 29[5] -> 61[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897247:897404 [7] NCCL INFO Channel 13/0 : 47[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9629:9793 [4] NCCL INFO Channel 07/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127297:127453 [7] NCCL INFO Channel 11/0 : 31[7] -> 63[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-46:702885:703048 [5] NCCL INFO Channel 09/0 : 13[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-46:702881:703043 [1] NCCL INFO Channel 11/0 : 9[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897240:897406 [0] NCCL INFO Channel 03/0 : 48[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9625:9787 [0] NCCL INFO Channel 14/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127290:127456 [0] NCCL INFO Channel 14/0 : 24[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-167-9:1039538:1039704 [1] NCCL INFO Channel 00/0 : 57[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897244:897407 [4] NCCL INFO Channel 08/0 : 52[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9629:9793 [4] NCCL INFO Channel 09/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624949:625110 [6] NCCL INFO Channel 14/0 : 38[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127295:127452 [5] NCCL INFO Channel 15/0 : 29[5] -> 61[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897240:897406 [0] NCCL INFO Channel 04/0 : 48[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624945:625111 [2] NCCL INFO Channel 08/0 : 34[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624946:625107 [3] NCCL INFO Channel 04/0 : 35[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127293:127454 [3] NCCL INFO Channel 01/0 : 43[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127297:127453 [7] NCCL INFO Channel 13/0 : 31[7] -> 63[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-46:702887:703045 [7] NCCL INFO Channel 09/0 : 15[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897244:897407 [4] NCCL INFO Channel 09/0 : 52[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624948:625104 [5] NCCL INFO Channel 06/0 : 53[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624944:625106 [1] NCCL INFO Channel 08/0 : 49[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127292:127458 [2] NCCL INFO Channel 01/0 : 42[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624950:625109 [7] NCCL INFO Channel 06/0 : 55[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-46:702885:703048 [5] NCCL INFO Channel 11/0 : 13[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897240:897406 [0] NCCL INFO Channel 05/0 : 48[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127291:127455 [1] NCCL INFO Channel 01/0 : 41[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897244:897407 [4] NCCL INFO Channel 10/0 : 52[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127293:127454 [3] NCCL INFO Channel 03/0 : 43[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127292:127458 [2] NCCL INFO Channel 03/0 : 42[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-46:702887:703045 [7] NCCL INFO Channel 11/0 : 15[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897240:897406 [0] NCCL INFO Channel 06/0 : 48[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9625:9787 [0] NCCL INFO Channel 15/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624945:625111 [2] NCCL INFO Channel 10/0 : 34[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127291:127455 [1] NCCL INFO Channel 03/0 : 41[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897244:897407 [4] NCCL INFO Channel 11/0 : 52[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9629:9793 [4] NCCL INFO Channel 11/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624946:625107 [3] NCCL INFO Channel 06/0 : 35[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127296:127451 [6] NCCL INFO Channel 01/0 : 46[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-167-9:1039539:1039698 [2] NCCL INFO Channel 00/0 : 58[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-46:702881:703043 [1] NCCL INFO Channel 13/0 : 9[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897240:897406 [0] NCCL INFO Channel 07/0 : 48[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9625:9787 [0] NCCL INFO Channel 01/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624948:625104 [5] NCCL INFO Channel 08/0 : 53[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127293:127454 [3] NCCL INFO Channel 05/0 : 43[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127292:127458 [2] NCCL INFO Channel 05/0 : 42[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-167-9:1039540:1039705 [3] NCCL INFO Channel 02/0 : 59[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-46:702885:703048 [5] NCCL INFO Channel 15/0 : 13[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897244:897407 [4] NCCL INFO Channel 13/0 : 52[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624944:625106 [1] NCCL INFO Channel 10/0 : 49[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127291:127455 [1] NCCL INFO Channel 05/0 : 41[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-167-9:1039538:1039704 [1] NCCL INFO Channel 02/0 : 57[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-46:702887:703045 [7] NCCL INFO Channel 13/0 : 15[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104435:104595 [6] NCCL INFO Channel 00/0 : 62[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897240:897406 [0] NCCL INFO Channel 09/0 : 48[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9629:9793 [4] NCCL INFO Channel 13/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624950:625109 [7] NCCL INFO Channel 08/0 : 55[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897244:897407 [4] NCCL INFO Channel 14/0 : 52[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9625:9787 [0] NCCL INFO Channel 03/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624945:625111 [2] NCCL INFO Channel 12/0 : 34[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897240:897406 [0] NCCL INFO Channel 10/0 : 48[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9629:9793 [4] NCCL INFO Channel 15/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624946:625107 [3] NCCL INFO Channel 08/0 : 35[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127296:127451 [6] NCCL INFO Channel 03/0 : 46[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897244:897407 [4] NCCL INFO Channel 15/0 : 52[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9625:9787 [0] NCCL INFO Channel 05/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127295:127452 [5] NCCL INFO Channel 01/0 : 45[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-167-9:1039540:1039705 [3] NCCL INFO Channel 04/0 : 59[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897240:897406 [0] NCCL INFO Channel 11/0 : 48[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127297:127453 [7] NCCL INFO Channel 01/0 : 47[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-167-9:1039543:1039699 [6] NCCL INFO Channel 00/0 : 62[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762821:762988 [6] NCCL INFO Channel 00/0 : 30[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897244:897407 [4] NCCL INFO Channel 01/0 : 44[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127292:127458 [2] NCCL INFO Channel 07/0 : 42[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-167-9:1039539:1039698 [2] NCCL INFO Channel 04/0 : 58[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104435:104595 [6] NCCL INFO Channel 02/0 : 62[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897240:897406 [0] NCCL INFO Channel 12/0 : 48[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127293:127454 [3] NCCL INFO Channel 07/0 : 43[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127291:127455 [1] NCCL INFO Channel 07/0 : 41[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-167-9:1039538:1039704 [1] NCCL INFO Channel 04/0 : 57[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897244:897407 [4] NCCL INFO Channel 03/0 : 44[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9625:9787 [0] NCCL INFO Channel 07/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624948:625104 [5] NCCL INFO Channel 10/0 : 53[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127296:127451 [6] NCCL INFO Channel 05/0 : 46[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127295:127452 [5] NCCL INFO Channel 03/0 : 45[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-167-9:1039540:1039705 [3] NCCL INFO Channel 06/0 : 59[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-167-9:1039542:1039701 [5] NCCL INFO Channel 00/0 : 61[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897240:897406 [0] NCCL INFO Channel 13/0 : 48[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9625:9787 [0] NCCL INFO Channel 09/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624944:625106 [1] NCCL INFO Channel 12/0 : 49[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127297:127453 [7] NCCL INFO Channel 03/0 : 47[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-167-9:1039543:1039699 [6] NCCL INFO Channel 02/0 : 62[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897244:897407 [4] NCCL INFO Channel 05/0 : 44[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624950:625109 [7] NCCL INFO Channel 10/0 : 55[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-167-9:1039544:1039702 [7] NCCL INFO Channel 00/0 : 63[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762821:762988 [6] NCCL INFO Channel 02/0 : 30[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897240:897406 [0] NCCL INFO Channel 14/0 : 48[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624945:625111 [2] NCCL INFO Channel 14/0 : 34[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-167-9:1039539:1039698 [2] NCCL INFO Channel 06/0 : 58[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104435:104595 [6] NCCL INFO Channel 04/0 : 62[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897244:897407 [4] NCCL INFO Channel 07/0 : 44[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624946:625107 [3] NCCL INFO Channel 10/0 : 35[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-167-9:1039538:1039704 [1] NCCL INFO Channel 06/0 : 57[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897240:897406 [0] NCCL INFO Channel 15/0 : 48[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9625:9787 [0] NCCL INFO Channel 11/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624948:625104 [5] NCCL INFO Channel 12/0 : 53[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127292:127458 [2] NCCL INFO Channel 09/0 : 42[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897244:897407 [4] NCCL INFO Channel 09/0 : 44[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624950:625109 [7] NCCL INFO Channel 12/0 : 55[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624944:625106 [1] NCCL INFO Channel 14/0 : 49[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127293:127454 [3] NCCL INFO Channel 09/0 : 43[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127291:127455 [1] NCCL INFO Channel 11/0 : 41[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897240:897406 [0] NCCL INFO Channel 01/0 : 40[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624946:625107 [3] NCCL INFO Channel 12/0 : 35[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127296:127451 [6] NCCL INFO Channel 07/0 : 46[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127295:127452 [5] NCCL INFO Channel 05/0 : 45[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104435:104595 [6] NCCL INFO Channel 08/0 : 62[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762821:762988 [6] NCCL INFO Channel 04/0 : 30[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897244:897407 [4] NCCL INFO Channel 11/0 : 44[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624949:625110 [6] NCCL INFO Channel 01/0 : 46[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127297:127453 [7] NCCL INFO Channel 05/0 : 47[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-167-9:1039540:1039705 [3] NCCL INFO Channel 08/0 : 59[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897240:897406 [0] NCCL INFO Channel 03/0 : 40[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9625:9787 [0] NCCL INFO Channel 13/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624948:625104 [5] NCCL INFO Channel 14/0 : 53[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127292:127458 [2] NCCL INFO Channel 11/0 : 42[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-167-9:1039542:1039701 [5] NCCL INFO Channel 02/0 : 61[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897244:897407 [4] NCCL INFO Channel 13/0 : 44[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-167-9:1039543:1039699 [6] NCCL INFO Channel 04/0 : 62[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-167-9:1039544:1039702 [7] NCCL INFO Channel 02/0 : 63[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104431:104593 [2] NCCL INFO Channel 00/0 : 58[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-167-9:1039539:1039698 [2] NCCL INFO Channel 08/0 : 58[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-167-9:1039538:1039704 [1] NCCL INFO Channel 08/0 : 57[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762821:762988 [6] NCCL INFO Channel 08/0 : 30[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-167-9:1039540:1039705 [3] NCCL INFO Channel 10/0 : 59[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9625:9787 [0] NCCL INFO Channel 15/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624944:625106 [1] NCCL INFO Channel 00/0 : 33[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-167-9:1039542:1039701 [5] NCCL INFO Channel 04/0 : 61[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624950:625109 [7] NCCL INFO Channel 14/0 : 55[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127293:127454 [3] NCCL INFO Channel 13/0 : 43[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127291:127455 [1] NCCL INFO Channel 13/0 : 41[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-167-9:1039543:1039699 [6] NCCL INFO Channel 08/0 : 62[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897240:897406 [0] NCCL INFO Channel 05/0 : 40[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624946:625107 [3] NCCL INFO Channel 14/0 : 35[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127296:127451 [6] NCCL INFO Channel 09/0 : 46[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127295:127452 [5] NCCL INFO Channel 07/0 : 45[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127297:127453 [7] NCCL INFO Channel 07/0 : 47[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104435:104595 [6] NCCL INFO Channel 10/0 : 62[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897244:897407 [4] NCCL INFO Channel 15/0 : 44[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624949:625110 [6] NCCL INFO Channel 03/0 : 46[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127292:127458 [2] NCCL INFO Channel 13/0 : 42[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104431:104593 [2] NCCL INFO Channel 04/0 : 58[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762817:762986 [2] NCCL INFO Channel 00/0 : 26[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897240:897406 [0] NCCL INFO Channel 07/0 : 40[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624948:625104 [5] NCCL INFO Channel 00/0 : 37[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127291:127455 [1] NCCL INFO Channel 15/0 : 41[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127293:127454 [3] NCCL INFO Channel 15/0 : 43[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104435:104595 [6] NCCL INFO Channel 12/0 : 62[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897240:897406 [0] NCCL INFO Channel 09/0 : 40[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624944:625106 [1] NCCL INFO Channel 02/0 : 33[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-167-9:1039539:1039698 [2] NCCL INFO Channel 10/0 : 58[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897240:897406 [0] NCCL INFO Channel 11/0 : 40[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624950:625109 [7] NCCL INFO Channel 00/0 : 39[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-167-9:1039538:1039704 [1] NCCL INFO Channel 10/0 : 57[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897240:897406 [0] NCCL INFO Channel 13/0 : 40[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624949:625110 [6] NCCL INFO Channel 05/0 : 46[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-167-9:1039544:1039702 [7] NCCL INFO Channel 04/0 : 63[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897240:897406 [0] NCCL INFO Channel 15/0 : 40[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624948:625104 [5] NCCL INFO Channel 02/0 : 37[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-167-9:1039540:1039705 [3] NCCL INFO Channel 12/0 : 59[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-167-9:1039542:1039701 [5] NCCL INFO Channel 06/0 : 61[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762821:762988 [6] NCCL INFO Channel 10/0 : 30[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624944:625106 [1] NCCL INFO Channel 04/0 : 33[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127295:127452 [5] NCCL INFO Channel 09/0 : 45[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127296:127451 [6] NCCL INFO Channel 11/0 : 46[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-167-9:1039543:1039699 [6] NCCL INFO Channel 10/0 : 62[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104431:104593 [2] NCCL INFO Channel 06/0 : 58[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762817:762986 [2] NCCL INFO Channel 04/0 : 26[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624950:625109 [7] NCCL INFO Channel 02/0 : 39[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127292:127458 [2] NCCL INFO Channel 15/0 : 42[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127297:127453 [7] NCCL INFO Channel 09/0 : 47[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-167-9:1039539:1039698 [2] NCCL INFO Channel 12/0 : 58[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104435:104595 [6] NCCL INFO Channel 14/0 : 62[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624949:625110 [6] NCCL INFO Channel 07/0 : 46[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127291:127455 [1] NCCL INFO Channel 01/0 : 25[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127293:127454 [3] NCCL INFO Channel 01/0 : 27[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624945:625111 [2] NCCL INFO Channel 01/0 : 42[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624948:625104 [5] NCCL INFO Channel 04/0 : 37[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127295:127452 [5] NCCL INFO Channel 11/0 : 45[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762821:762988 [6] NCCL INFO Channel 12/0 : 30[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104431:104593 [2] NCCL INFO Channel 08/0 : 58[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762817:762986 [2] NCCL INFO Channel 06/0 : 26[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-167-9:1039538:1039704 [1] NCCL INFO Channel 12/0 : 57[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-167-9:1039544:1039702 [7] NCCL INFO Channel 06/0 : 63[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104432:104592 [3] NCCL INFO Channel 00/0 : 59[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624944:625106 [1] NCCL INFO Channel 06/0 : 33[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127296:127451 [6] NCCL INFO Channel 13/0 : 46[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-167-9:1039540:1039705 [3] NCCL INFO Channel 14/0 : 59[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104435:104595 [6] NCCL INFO Channel 00/0 : 54[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624950:625109 [7] NCCL INFO Channel 04/0 : 39[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127292:127458 [2] NCCL INFO Channel 01/0 : 26[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-167-9:1039542:1039701 [5] NCCL INFO Channel 08/0 : 61[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104431:104593 [2] NCCL INFO Channel 10/0 : 58[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624949:625110 [6] NCCL INFO Channel 09/0 : 46[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127293:127454 [3] NCCL INFO Channel 03/0 : 27[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127291:127455 [1] NCCL INFO Channel 03/0 : 25[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-167-9:1039543:1039699 [6] NCCL INFO Channel 12/0 : 62[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624948:625104 [5] NCCL INFO Channel 06/0 : 37[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127297:127453 [7] NCCL INFO Channel 11/0 : 47[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-167-9:1039539:1039698 [2] NCCL INFO Channel 14/0 : 58[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762818:762985 [3] NCCL INFO Channel 00/0 : 27[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762821:762988 [6] NCCL INFO Channel 14/0 : 30[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624945:625111 [2] NCCL INFO Channel 03/0 : 42[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127295:127452 [5] NCCL INFO Channel 15/0 : 45[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-167-9:1039538:1039704 [1] NCCL INFO Channel 14/0 : 57[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-167-9:1039544:1039702 [7] NCCL INFO Channel 08/0 : 63[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624944:625106 [1] NCCL INFO Channel 08/0 : 33[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127296:127451 [6] NCCL INFO Channel 15/0 : 46[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-167-9:1039543:1039699 [6] NCCL INFO Channel 14/0 : 62[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-167-9:1039542:1039701 [5] NCCL INFO Channel 10/0 : 61[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104435:104595 [6] NCCL INFO Channel 01/0 : 54[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624950:625109 [7] NCCL INFO Channel 06/0 : 39[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104431:104593 [2] NCCL INFO Channel 12/0 : 58[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104432:104592 [3] NCCL INFO Channel 02/0 : 59[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624949:625110 [6] NCCL INFO Channel 11/0 : 46[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624948:625104 [5] NCCL INFO Channel 08/0 : 37[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762817:762986 [2] NCCL INFO Channel 08/0 : 26[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127292:127458 [2] NCCL INFO Channel 03/0 : 26[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762818:762985 [3] NCCL INFO Channel 02/0 : 27[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762821:762988 [6] NCCL INFO Channel 00/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127291:127455 [1] NCCL INFO Channel 05/0 : 25[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127293:127454 [3] NCCL INFO Channel 05/0 : 27[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127297:127453 [7] NCCL INFO Channel 13/0 : 47[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127295:127452 [5] NCCL INFO Channel 01/0 : 29[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-167-9:1039544:1039702 [7] NCCL INFO Channel 10/0 : 63[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127296:127451 [6] NCCL INFO Channel 01/0 : 30[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-167-9:1039542:1039701 [5] NCCL INFO Channel 12/0 : 61[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104435:104595 [6] NCCL INFO Channel 02/0 : 54[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127292:127458 [2] NCCL INFO Channel 05/0 : 26[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-167-9:1039544:1039702 [7] NCCL INFO Channel 12/0 : 63[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104431:104593 [2] NCCL INFO Channel 14/0 : 58[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624950:625109 [7] NCCL INFO Channel 08/0 : 39[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624944:625106 [1] NCCL INFO Channel 10/0 : 33[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-167-9:1039542:1039701 [5] NCCL INFO Channel 14/0 : 61[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762817:762986 [2] NCCL INFO Channel 10/0 : 26[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624945:625111 [2] NCCL INFO Channel 05/0 : 42[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762821:762988 [6] NCCL INFO Channel 01/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762818:762985 [3] NCCL INFO Channel 04/0 : 27[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624946:625107 [3] NCCL INFO Channel 01/0 : 43[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624949:625110 [6] NCCL INFO Channel 13/0 : 46[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624948:625104 [5] NCCL INFO Channel 10/0 : 37[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624950:625109 [7] NCCL INFO Channel 10/0 : 39[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127291:127455 [1] NCCL INFO Channel 07/0 : 25[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104435:104595 [6] NCCL INFO Channel 03/0 : 54[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104432:104592 [3] NCCL INFO Channel 04/0 : 59[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624944:625106 [1] NCCL INFO Channel 12/0 : 33[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624945:625111 [2] NCCL INFO Channel 07/0 : 42[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127293:127454 [3] NCCL INFO Channel 07/0 : 27[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624949:625110 [6] NCCL INFO Channel 15/0 : 46[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624948:625104 [5] NCCL INFO Channel 12/0 : 37[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127295:127452 [5] NCCL INFO Channel 03/0 : 29[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127296:127451 [6] NCCL INFO Channel 03/0 : 30[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127292:127458 [2] NCCL INFO Channel 07/0 : 26[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127297:127453 [7] NCCL INFO Channel 01/0 : 31[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-167-9:1039544:1039702 [7] NCCL INFO Channel 14/0 : 63[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762817:762986 [2] NCCL INFO Channel 12/0 : 26[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104431:104593 [2] NCCL INFO Channel 00/0 : 50[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104435:104595 [6] NCCL INFO Channel 04/0 : 54[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104432:104592 [3] NCCL INFO Channel 06/0 : 59[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624946:625107 [3] NCCL INFO Channel 03/0 : 43[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127291:127455 [1] NCCL INFO Channel 11/0 : 25[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624950:625109 [7] NCCL INFO Channel 12/0 : 39[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127293:127454 [3] NCCL INFO Channel 09/0 : 27[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762821:762988 [6] NCCL INFO Channel 02/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624944:625106 [1] NCCL INFO Channel 14/0 : 33[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624945:625111 [2] NCCL INFO Channel 09/0 : 42[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762818:762985 [3] NCCL INFO Channel 06/0 : 27[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624948:625104 [5] NCCL INFO Channel 14/0 : 37[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624946:625107 [3] NCCL INFO Channel 05/0 : 43[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762817:762986 [2] NCCL INFO Channel 14/0 : 26[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624950:625109 [7] NCCL INFO Channel 14/0 : 39[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104431:104593 [2] NCCL INFO Channel 01/0 : 50[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624945:625111 [2] NCCL INFO Channel 11/0 : 42[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104435:104595 [6] NCCL INFO Channel 05/0 : 54[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127292:127458 [2] NCCL INFO Channel 09/0 : 26[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127295:127452 [5] NCCL INFO Channel 05/0 : 29[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127296:127451 [6] NCCL INFO Channel 05/0 : 30[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127297:127453 [7] NCCL INFO Channel 03/0 : 31[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104432:104592 [3] NCCL INFO Channel 08/0 : 59[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127291:127455 [1] NCCL INFO Channel 13/0 : 25[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127293:127454 [3] NCCL INFO Channel 13/0 : 27[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762821:762988 [6] NCCL INFO Channel 03/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762818:762985 [3] NCCL INFO Channel 08/0 : 27[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762817:762986 [2] NCCL INFO Channel 00/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624946:625107 [3] NCCL INFO Channel 07/0 : 43[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624945:625111 [2] NCCL INFO Channel 13/0 : 42[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624946:625107 [3] NCCL INFO Channel 09/0 : 43[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104431:104593 [2] NCCL INFO Channel 03/0 : 50[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104435:104595 [6] NCCL INFO Channel 07/0 : 54[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104432:104592 [3] NCCL INFO Channel 10/0 : 59[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104431:104593 [2] NCCL INFO Channel 04/0 : 50[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762821:762988 [6] NCCL INFO Channel 04/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127295:127452 [5] NCCL INFO Channel 07/0 : 29[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127292:127458 [2] NCCL INFO Channel 11/0 : 26[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127296:127451 [6] NCCL INFO Channel 07/0 : 30[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762818:762985 [3] NCCL INFO Channel 10/0 : 27[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624945:625111 [2] NCCL INFO Channel 15/0 : 42[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624946:625107 [3] NCCL INFO Channel 13/0 : 43[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104435:104595 [6] NCCL INFO Channel 08/0 : 54[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104432:104592 [3] NCCL INFO Channel 12/0 : 59[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127291:127455 [1] NCCL INFO Channel 15/0 : 25[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127293:127454 [3] NCCL INFO Channel 15/0 : 27[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127297:127453 [7] NCCL INFO Channel 05/0 : 31[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762821:762988 [6] NCCL INFO Channel 05/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762817:762986 [2] NCCL INFO Channel 01/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624946:625107 [3] NCCL INFO Channel 15/0 : 43[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104431:104593 [2] NCCL INFO Channel 05/0 : 50[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762818:762985 [3] NCCL INFO Channel 12/0 : 27[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104430:104589 [1] NCCL INFO Channel 00/0 : 57[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104434:104591 [5] NCCL INFO Channel 00/0 : 61[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127292:127458 [2] NCCL INFO Channel 13/0 : 26[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127295:127452 [5] NCCL INFO Channel 09/0 : 29[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127296:127451 [6] NCCL INFO Channel 09/0 : 30[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104435:104595 [6] NCCL INFO Channel 09/0 : 54[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127297:127453 [7] NCCL INFO Channel 07/0 : 31[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104436:104596 [7] NCCL INFO Channel 00/0 : 63[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624944:625106 [1] NCCL INFO Channel 01/0 : 41[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624948:625104 [5] NCCL INFO Channel 01/0 : 45[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762821:762988 [6] NCCL INFO Channel 07/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127292:127458 [2] NCCL INFO Channel 15/0 : 26[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762817:762986 [2] NCCL INFO Channel 03/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762816:762987 [1] NCCL INFO Channel 00/0 : 25[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104432:104592 [3] NCCL INFO Channel 14/0 : 59[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104431:104593 [2] NCCL INFO Channel 06/0 : 50[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104435:104595 [6] NCCL INFO Channel 10/0 : 54[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624950:625109 [7] NCCL INFO Channel 01/0 : 47[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104434:104591 [5] NCCL INFO Channel 02/0 : 61[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104430:104589 [1] NCCL INFO Channel 02/0 : 57[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624944:625106 [1] NCCL INFO Channel 03/0 : 41[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624948:625104 [5] NCCL INFO Channel 03/0 : 45[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127295:127452 [5] NCCL INFO Channel 11/0 : 29[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127296:127451 [6] NCCL INFO Channel 11/0 : 30[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127297:127453 [7] NCCL INFO Channel 09/0 : 31[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897243:897403 [3] NCCL INFO Channel 00/0 : 51[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762821:762988 [6] NCCL INFO Channel 08/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762820:762981 [5] NCCL INFO Channel 00/0 : 29[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762818:762985 [3] NCCL INFO Channel 14/0 : 27[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762817:762986 [2] NCCL INFO Channel 04/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762822:762984 [7] NCCL INFO Channel 00/0 : 31[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104436:104596 [7] NCCL INFO Channel 02/0 : 63[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762816:762987 [1] NCCL INFO Channel 02/0 : 25[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624950:625109 [7] NCCL INFO Channel 03/0 : 47[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127295:127452 [5] NCCL INFO Channel 15/0 : 29[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104432:104592 [3] NCCL INFO Channel 00/0 : 51[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104431:104593 [2] NCCL INFO Channel 07/0 : 50[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762821:762988 [6] NCCL INFO Channel 09/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897241:897402 [1] NCCL INFO Channel 00/0 : 49[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127297:127453 [7] NCCL INFO Channel 11/0 : 31[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127296:127451 [6] NCCL INFO Channel 13/0 : 30[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104435:104595 [6] NCCL INFO Channel 11/0 : 54[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897243:897403 [3] NCCL INFO Channel 01/0 : 51[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104434:104591 [5] NCCL INFO Channel 04/0 : 61[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104430:104589 [1] NCCL INFO Channel 04/0 : 57[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624948:625104 [5] NCCL INFO Channel 05/0 : 45[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624944:625106 [1] NCCL INFO Channel 05/0 : 41[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624950:625109 [7] NCCL INFO Channel 05/0 : 47[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762820:762981 [5] NCCL INFO Channel 02/0 : 29[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9626:9790 [1] NCCL INFO Channel 00/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9628:9786 [3] NCCL INFO Channel 00/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127293:127454 [3] NCCL INFO Channel 00/0 : 27[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762818:762985 [3] NCCL INFO Channel 00/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897241:897402 [1] NCCL INFO Channel 02/0 : 49[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-46:702881:703043 [1] NCCL INFO Channel 15/0 : 9[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762817:762986 [2] NCCL INFO Channel 05/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762822:762984 [7] NCCL INFO Channel 02/0 : 31[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624948:625104 [5] NCCL INFO Channel 07/0 : 45[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104432:104592 [3] NCCL INFO Channel 01/0 : 51[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762816:762987 [1] NCCL INFO Channel 04/0 : 25[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624944:625106 [1] NCCL INFO Channel 07/0 : 41[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127296:127451 [6] NCCL INFO Channel 15/0 : 30[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127297:127453 [7] NCCL INFO Channel 13/0 : 31[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104431:104593 [2] NCCL INFO Channel 08/0 : 50[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104436:104596 [7] NCCL INFO Channel 04/0 : 63[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104435:104595 [6] NCCL INFO Channel 12/0 : 54[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762821:762988 [6] NCCL INFO Channel 10/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897243:897403 [3] NCCL INFO Channel 02/0 : 51[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9626:9790 [1] NCCL INFO Channel 02/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9628:9786 [3] NCCL INFO Channel 01/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624950:625109 [7] NCCL INFO Channel 07/0 : 47[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127291:127455 [1] NCCL INFO Channel 00/0 : 25[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127293:127454 [3] NCCL INFO Channel 02/0 : 27[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104434:104591 [5] NCCL INFO Channel 06/0 : 61[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762820:762981 [5] NCCL INFO Channel 04/0 : 29[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897242:897400 [2] NCCL INFO Channel 00/0 : 50[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104430:104589 [1] NCCL INFO Channel 06/0 : 57[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897241:897402 [1] NCCL INFO Channel 03/0 : 49[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104432:104592 [3] NCCL INFO Channel 02/0 : 51[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624948:625104 [5] NCCL INFO Channel 09/0 : 45[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127291:127455 [1] NCCL INFO Channel 02/0 : 25[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762818:762985 [3] NCCL INFO Channel 01/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624944:625106 [1] NCCL INFO Channel 11/0 : 41[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624950:625109 [7] NCCL INFO Channel 09/0 : 47[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127293:127454 [3] NCCL INFO Channel 04/0 : 27[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762817:762986 [2] NCCL INFO Channel 06/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897243:897403 [3] NCCL INFO Channel 04/0 : 51[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9626:9790 [1] NCCL INFO Channel 03/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127292:127458 [2] NCCL INFO Channel 00/0 : 26[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104431:104593 [2] NCCL INFO Channel 09/0 : 50[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762822:762984 [7] NCCL INFO Channel 04/0 : 31[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897242:897400 [2] NCCL INFO Channel 01/0 : 50[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104435:104595 [6] NCCL INFO Channel 13/0 : 54[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762821:762988 [6] NCCL INFO Channel 11/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897245:897401 [5] NCCL INFO Channel 00/0 : 53[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897241:897402 [1] NCCL INFO Channel 04/0 : 49[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897243:897403 [3] NCCL INFO Channel 05/0 : 51[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104436:104596 [7] NCCL INFO Channel 06/0 : 63[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762816:762987 [1] NCCL INFO Channel 06/0 : 25[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624948:625104 [5] NCCL INFO Channel 11/0 : 45[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104434:104591 [5] NCCL INFO Channel 08/0 : 61[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762820:762981 [5] NCCL INFO Channel 06/0 : 29[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762818:762985 [3] NCCL INFO Channel 02/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624950:625109 [7] NCCL INFO Channel 11/0 : 47[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624944:625106 [1] NCCL INFO Channel 13/0 : 41[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104432:104592 [3] NCCL INFO Channel 04/0 : 51[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104430:104589 [1] NCCL INFO Channel 08/0 : 57[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762817:762986 [2] NCCL INFO Channel 07/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9628:9786 [3] NCCL INFO Channel 02/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127291:127455 [1] NCCL INFO Channel 04/0 : 25[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104431:104593 [2] NCCL INFO Channel 11/0 : 50[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9627:9791 [2] NCCL INFO Channel 00/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127293:127454 [3] NCCL INFO Channel 06/0 : 27[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127292:127458 [2] NCCL INFO Channel 04/0 : 26[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104435:104595 [6] NCCL INFO Channel 15/0 : 54[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104436:104596 [7] NCCL INFO Channel 08/0 : 63[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624948:625104 [5] NCCL INFO Channel 15/0 : 45[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104434:104591 [5] NCCL INFO Channel 10/0 : 61[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897242:897400 [2] NCCL INFO Channel 03/0 : 50[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104432:104592 [3] NCCL INFO Channel 05/0 : 51[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762821:762988 [6] NCCL INFO Channel 12/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9626:9790 [1] NCCL INFO Channel 04/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762822:762984 [7] NCCL INFO Channel 06/0 : 31[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9627:9791 [2] NCCL INFO Channel 01/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9628:9786 [3] NCCL INFO Channel 04/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9630:9789 [5] NCCL INFO Channel 00/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127295:127452 [5] NCCL INFO Channel 00/0 : 29[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127291:127455 [1] NCCL INFO Channel 06/0 : 25[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762820:762981 [5] NCCL INFO Channel 08/0 : 29[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762816:762987 [1] NCCL INFO Channel 08/0 : 25[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624944:625106 [1] NCCL INFO Channel 15/0 : 41[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624950:625109 [7] NCCL INFO Channel 13/0 : 47[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127292:127458 [2] NCCL INFO Channel 06/0 : 26[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762818:762985 [3] NCCL INFO Channel 04/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762817:762986 [2] NCCL INFO Channel 08/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897245:897401 [5] NCCL INFO Channel 01/0 : 53[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897247:897404 [7] NCCL INFO Channel 00/0 : 55[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897241:897402 [1] NCCL INFO Channel 05/0 : 49[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127293:127454 [3] NCCL INFO Channel 08/0 : 27[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762821:762988 [6] NCCL INFO Channel 13/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897246:897405 [6] NCCL INFO Channel 00/0 : 54[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897242:897400 [2] NCCL INFO Channel 04/0 : 50[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897243:897403 [3] NCCL INFO Channel 06/0 : 51[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104431:104593 [2] NCCL INFO Channel 12/0 : 50[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762822:762984 [7] NCCL INFO Channel 08/0 : 31[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104430:104589 [1] NCCL INFO Channel 10/0 : 57[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762820:762981 [5] NCCL INFO Channel 10/0 : 29[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9626:9790 [1] NCCL INFO Channel 05/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104436:104596 [7] NCCL INFO Channel 10/0 : 63[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104434:104591 [5] NCCL INFO Channel 12/0 : 61[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897245:897401 [5] NCCL INFO Channel 02/0 : 53[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127295:127452 [5] NCCL INFO Channel 02/0 : 29[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127291:127455 [1] NCCL INFO Channel 08/0 : 25[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104432:104592 [3] NCCL INFO Channel 06/0 : 51[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127292:127458 [2] NCCL INFO Channel 08/0 : 26[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104431:104593 [2] NCCL INFO Channel 13/0 : 50[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762816:762987 [1] NCCL INFO Channel 10/0 : 25[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9627:9791 [2] NCCL INFO Channel 03/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127293:127454 [3] NCCL INFO Channel 10/0 : 27[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104430:104589 [1] NCCL INFO Channel 12/0 : 57[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762818:762985 [3] NCCL INFO Channel 05/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9628:9786 [3] NCCL INFO Channel 05/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9630:9789 [5] NCCL INFO Channel 01/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127297:127453 [7] NCCL INFO Channel 00/0 : 31[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104436:104596 [7] NCCL INFO Channel 12/0 : 63[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762817:762986 [2] NCCL INFO Channel 09/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762821:762988 [6] NCCL INFO Channel 15/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897246:897405 [6] NCCL INFO Channel 01/0 : 54[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897241:897402 [1] NCCL INFO Channel 06/0 : 49[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897242:897400 [2] NCCL INFO Channel 05/0 : 50[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897247:897404 [7] NCCL INFO Channel 01/0 : 55[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9632:9792 [7] NCCL INFO Channel 00/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104434:104591 [5] NCCL INFO Channel 14/0 : 61[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762822:762984 [7] NCCL INFO Channel 10/0 : 31[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897243:897403 [3] NCCL INFO Channel 07/0 : 51[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762820:762981 [5] NCCL INFO Channel 12/0 : 29[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762816:762987 [1] NCCL INFO Channel 12/0 : 25[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762818:762985 [3] NCCL INFO Channel 06/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127296:127451 [6] NCCL INFO Channel 00/0 : 30[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762817:762986 [2] NCCL INFO Channel 11/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762822:762984 [7] NCCL INFO Channel 12/0 : 31[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127295:127452 [5] NCCL INFO Channel 04/0 : 29[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127291:127455 [1] NCCL INFO Channel 10/0 : 25[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9631:9788 [6] NCCL INFO Channel 00/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9626:9790 [1] NCCL INFO Channel 06/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9627:9791 [2] NCCL INFO Channel 04/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127292:127458 [2] NCCL INFO Channel 10/0 : 26[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104432:104592 [3] NCCL INFO Channel 07/0 : 51[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104431:104593 [2] NCCL INFO Channel 14/0 : 50[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897245:897401 [5] NCCL INFO Channel 03/0 : 53[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9628:9786 [3] NCCL INFO Channel 06/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9630:9789 [5] NCCL INFO Channel 02/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127293:127454 [3] NCCL INFO Channel 12/0 : 27[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127297:127453 [7] NCCL INFO Channel 02/0 : 31[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104436:104596 [7] NCCL INFO Channel 14/0 : 63[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897246:897405 [6] NCCL INFO Channel 02/0 : 54[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9632:9792 [7] NCCL INFO Channel 01/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127296:127451 [6] NCCL INFO Channel 02/0 : 30[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104430:104589 [1] NCCL INFO Channel 14/0 : 57[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104434:104591 [5] NCCL INFO Channel 00/0 : 53[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762820:762981 [5] NCCL INFO Channel 14/0 : 29[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104431:104593 [2] NCCL INFO Channel 15/0 : 50[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104432:104592 [3] NCCL INFO Channel 08/0 : 51[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762816:762987 [1] NCCL INFO Channel 14/0 : 25[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762818:762985 [3] NCCL INFO Channel 07/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104436:104596 [7] NCCL INFO Channel 00/0 : 55[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762822:762984 [7] NCCL INFO Channel 14/0 : 31[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104430:104589 [1] NCCL INFO Channel 00/0 : 49[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104434:104591 [5] NCCL INFO Channel 01/0 : 53[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762817:762986 [2] NCCL INFO Channel 12/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762820:762981 [5] NCCL INFO Channel 00/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897242:897400 [2] NCCL INFO Channel 06/0 : 50[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897241:897402 [1] NCCL INFO Channel 07/0 : 49[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897243:897403 [3] NCCL INFO Channel 08/0 : 51[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9626:9790 [1] NCCL INFO Channel 07/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9627:9791 [2] NCCL INFO Channel 05/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104432:104592 [3] NCCL INFO Channel 09/0 : 51[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762818:762985 [3] NCCL INFO Channel 08/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897247:897404 [7] NCCL INFO Channel 02/0 : 55[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897245:897401 [5] NCCL INFO Channel 04/0 : 53[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9631:9788 [6] NCCL INFO Channel 01/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9628:9786 [3] NCCL INFO Channel 07/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9630:9789 [5] NCCL INFO Channel 03/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9632:9792 [7] NCCL INFO Channel 02/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127295:127452 [5] NCCL INFO Channel 06/0 : 29[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762816:762987 [1] NCCL INFO Channel 00/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127291:127455 [1] NCCL INFO Channel 12/0 : 25[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127292:127458 [2] NCCL INFO Channel 12/0 : 26[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127293:127454 [3] NCCL INFO Channel 14/0 : 27[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127296:127451 [6] NCCL INFO Channel 04/0 : 30[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127297:127453 [7] NCCL INFO Channel 04/0 : 31[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104436:104596 [7] NCCL INFO Channel 01/0 : 55[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897246:897405 [6] NCCL INFO Channel 03/0 : 54[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104430:104589 [1] NCCL INFO Channel 02/0 : 49[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104434:104591 [5] NCCL INFO Channel 02/0 : 53[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897242:897400 [2] NCCL INFO Channel 07/0 : 50[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897241:897402 [1] NCCL INFO Channel 08/0 : 49[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9627:9791 [2] NCCL INFO Channel 06/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9626:9790 [1] NCCL INFO Channel 08/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104432:104592 [3] NCCL INFO Channel 10/0 : 51[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762822:762984 [7] NCCL INFO Channel 00/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9631:9788 [6] NCCL INFO Channel 02/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104436:104596 [7] NCCL INFO Channel 02/0 : 55[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762820:762981 [5] NCCL INFO Channel 01/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104434:104591 [5] NCCL INFO Channel 03/0 : 53[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104430:104589 [1] NCCL INFO Channel 03/0 : 49[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762818:762985 [3] NCCL INFO Channel 09/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127295:127452 [5] NCCL INFO Channel 08/0 : 29[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104432:104592 [3] NCCL INFO Channel 12/0 : 51[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762817:762986 [2] NCCL INFO Channel 13/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762822:762984 [7] NCCL INFO Channel 01/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127291:127455 [1] NCCL INFO Channel 14/0 : 25[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127292:127458 [2] NCCL INFO Channel 14/0 : 26[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104436:104596 [7] NCCL INFO Channel 03/0 : 55[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762816:762987 [1] NCCL INFO Channel 02/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897243:897403 [3] NCCL INFO Channel 09/0 : 51[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127296:127451 [6] NCCL INFO Channel 08/0 : 30[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127297:127453 [7] NCCL INFO Channel 06/0 : 31[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897245:897401 [5] NCCL INFO Channel 06/0 : 53[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897247:897404 [7] NCCL INFO Channel 03/0 : 55[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9628:9786 [3] NCCL INFO Channel 08/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127295:127452 [5] NCCL INFO Channel 10/0 : 29[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897246:897405 [6] NCCL INFO Channel 04/0 : 54[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9632:9792 [7] NCCL INFO Channel 03/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9630:9789 [5] NCCL INFO Channel 04/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9627:9791 [2] NCCL INFO Channel 07/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9626:9790 [1] NCCL INFO Channel 10/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9631:9788 [6] NCCL INFO Channel 03/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104434:104591 [5] NCCL INFO Channel 04/0 : 53[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762820:762981 [5] NCCL INFO Channel 02/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104430:104589 [1] NCCL INFO Channel 04/0 : 49[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104432:104592 [3] NCCL INFO Channel 13/0 : 51[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762818:762985 [3] NCCL INFO Channel 10/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104436:104596 [7] NCCL INFO Channel 04/0 : 55[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762822:762984 [7] NCCL INFO Channel 02/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897242:897400 [2] NCCL INFO Channel 08/0 : 50[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897241:897402 [1] NCCL INFO Channel 10/0 : 49[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762816:762987 [1] NCCL INFO Channel 03/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897243:897403 [3] NCCL INFO Channel 10/0 : 51[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762817:762986 [2] NCCL INFO Channel 14/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897245:897401 [5] NCCL INFO Channel 07/0 : 53[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897247:897404 [7] NCCL INFO Channel 04/0 : 55[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762820:762981 [5] NCCL INFO Channel 03/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897246:897405 [6] NCCL INFO Channel 05/0 : 54[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897242:897400 [2] NCCL INFO Channel 09/0 : 50[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897241:897402 [1] NCCL INFO Channel 11/0 : 49[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897243:897403 [3] NCCL INFO Channel 12/0 : 51[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104434:104591 [5] NCCL INFO Channel 06/0 : 53[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9628:9786 [3] NCCL INFO Channel 09/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104430:104589 [1] NCCL INFO Channel 05/0 : 49[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104432:104592 [3] NCCL INFO Channel 14/0 : 51[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104436:104596 [7] NCCL INFO Channel 05/0 : 55[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9630:9789 [5] NCCL INFO Channel 06/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9632:9792 [7] NCCL INFO Channel 04/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9627:9791 [2] NCCL INFO Channel 08/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897245:897401 [5] NCCL INFO Channel 08/0 : 53[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762818:762985 [3] NCCL INFO Channel 12/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897246:897405 [6] NCCL INFO Channel 07/0 : 54[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897247:897404 [7] NCCL INFO Channel 05/0 : 55[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104434:104591 [5] NCCL INFO Channel 07/0 : 53[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762822:762984 [7] NCCL INFO Channel 03/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897241:897402 [1] NCCL INFO Channel 12/0 : 49[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897242:897400 [2] NCCL INFO Channel 11/0 : 50[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897243:897403 [3] NCCL INFO Channel 13/0 : 51[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762816:762987 [1] NCCL INFO Channel 04/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897245:897401 [5] NCCL INFO Channel 09/0 : 53[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9631:9788 [6] NCCL INFO Channel 04/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9626:9790 [1] NCCL INFO Channel 11/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762820:762981 [5] NCCL INFO Channel 04/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762817:762986 [2] NCCL INFO Channel 15/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9628:9786 [3] NCCL INFO Channel 10/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762818:762985 [3] NCCL INFO Channel 13/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9630:9789 [5] NCCL INFO Channel 07/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9632:9792 [7] NCCL INFO Channel 05/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9627:9791 [2] NCCL INFO Channel 09/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9631:9788 [6] NCCL INFO Channel 05/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104430:104589 [1] NCCL INFO Channel 06/0 : 49[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104432:104592 [3] NCCL INFO Channel 15/0 : 51[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104436:104596 [7] NCCL INFO Channel 06/0 : 55[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9626:9790 [1] NCCL INFO Channel 12/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104434:104591 [5] NCCL INFO Channel 08/0 : 53[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897246:897405 [6] NCCL INFO Channel 08/0 : 54[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897247:897404 [7] NCCL INFO Channel 06/0 : 55[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762822:762984 [7] NCCL INFO Channel 04/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762816:762987 [1] NCCL INFO Channel 05/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897241:897402 [1] NCCL INFO Channel 13/0 : 49[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762820:762981 [5] NCCL INFO Channel 06/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897242:897400 [2] NCCL INFO Channel 12/0 : 50[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897243:897403 [3] NCCL INFO Channel 14/0 : 51[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762818:762985 [3] NCCL INFO Channel 14/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897245:897401 [5] NCCL INFO Channel 10/0 : 53[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762816:762987 [1] NCCL INFO Channel 06/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762822:762984 [7] NCCL INFO Channel 05/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897246:897405 [6] NCCL INFO Channel 09/0 : 54[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9628:9786 [3] NCCL INFO Channel 12/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9630:9789 [5] NCCL INFO Channel 08/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9632:9792 [7] NCCL INFO Channel 06/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9627:9791 [2] NCCL INFO Channel 11/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104430:104589 [1] NCCL INFO Channel 07/0 : 49[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9631:9788 [6] NCCL INFO Channel 07/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104436:104596 [7] NCCL INFO Channel 08/0 : 55[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104434:104591 [5] NCCL INFO Channel 09/0 : 53[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9626:9790 [1] NCCL INFO Channel 13/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9628:9786 [3] NCCL INFO Channel 13/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762820:762981 [5] NCCL INFO Channel 07/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762818:762985 [3] NCCL INFO Channel 15/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762822:762984 [7] NCCL INFO Channel 06/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897247:897404 [7] NCCL INFO Channel 08/0 : 55[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897241:897402 [1] NCCL INFO Channel 14/0 : 49[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104430:104589 [1] NCCL INFO Channel 08/0 : 49[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897243:897403 [3] NCCL INFO Channel 15/0 : 51[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104434:104591 [5] NCCL INFO Channel 10/0 : 53[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104436:104596 [7] NCCL INFO Channel 09/0 : 55[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897242:897400 [2] NCCL INFO Channel 13/0 : 50[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9630:9789 [5] NCCL INFO Channel 09/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9632:9792 [7] NCCL INFO Channel 08/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897245:897401 [5] NCCL INFO Channel 11/0 : 53[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9627:9791 [2] NCCL INFO Channel 12/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762816:762987 [1] NCCL INFO Channel 07/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762820:762981 [5] NCCL INFO Channel 08/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9631:9788 [6] NCCL INFO Channel 08/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9626:9790 [1] NCCL INFO Channel 14/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9628:9786 [3] NCCL INFO Channel 14/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762822:762984 [7] NCCL INFO Channel 08/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9630:9789 [5] NCCL INFO Channel 10/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104430:104589 [1] NCCL INFO Channel 10/0 : 49[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104436:104596 [7] NCCL INFO Channel 10/0 : 55[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104434:104591 [5] NCCL INFO Channel 11/0 : 53[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897246:897405 [6] NCCL INFO Channel 10/0 : 54[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897247:897404 [7] NCCL INFO Channel 09/0 : 55[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897241:897402 [1] NCCL INFO Channel 15/0 : 49[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762816:762987 [1] NCCL INFO Channel 08/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762820:762981 [5] NCCL INFO Channel 09/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897243:897403 [3] NCCL INFO Channel 01/0 : 43[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762822:762984 [7] NCCL INFO Channel 09/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9632:9792 [7] NCCL INFO Channel 09/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9627:9791 [2] NCCL INFO Channel 13/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9631:9788 [6] NCCL INFO Channel 09/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9626:9790 [1] NCCL INFO Channel 15/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104430:104589 [1] NCCL INFO Channel 11/0 : 49[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9628:9786 [3] NCCL INFO Channel 15/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9630:9789 [5] NCCL INFO Channel 11/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897242:897400 [2] NCCL INFO Channel 14/0 : 50[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897245:897401 [5] NCCL INFO Channel 12/0 : 53[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9632:9792 [7] NCCL INFO Channel 10/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762816:762987 [1] NCCL INFO Channel 10/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897246:897405 [6] NCCL INFO Channel 11/0 : 54[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762820:762981 [5] NCCL INFO Channel 10/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762822:762984 [7] NCCL INFO Channel 10/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897247:897404 [7] NCCL INFO Channel 10/0 : 55[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897241:897402 [1] NCCL INFO Channel 01/0 : 41[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104436:104596 [7] NCCL INFO Channel 11/0 : 55[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104434:104591 [5] NCCL INFO Channel 12/0 : 53[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897243:897403 [3] NCCL INFO Channel 03/0 : 43[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104430:104589 [1] NCCL INFO Channel 12/0 : 49[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9627:9791 [2] NCCL INFO Channel 14/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9626:9790 [1] NCCL INFO Channel 01/0 : 9[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762816:762987 [1] NCCL INFO Channel 11/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9628:9786 [3] NCCL INFO Channel 01/0 : 11[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9630:9789 [5] NCCL INFO Channel 12/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897245:897401 [5] NCCL INFO Channel 14/0 : 53[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9632:9792 [7] NCCL INFO Channel 11/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104434:104591 [5] NCCL INFO Channel 14/0 : 53[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104436:104596 [7] NCCL INFO Channel 12/0 : 55[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897246:897405 [6] NCCL INFO Channel 12/0 : 54[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9627:9791 [2] NCCL INFO Channel 15/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104430:104589 [1] NCCL INFO Channel 13/0 : 49[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897242:897400 [2] NCCL INFO Channel 15/0 : 50[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762820:762981 [5] NCCL INFO Channel 11/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762822:762984 [7] NCCL INFO Channel 11/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897247:897404 [7] NCCL INFO Channel 11/0 : 55[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897241:897402 [1] NCCL INFO Channel 03/0 : 41[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897243:897403 [3] NCCL INFO Channel 05/0 : 43[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897245:897401 [5] NCCL INFO Channel 15/0 : 53[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9626:9790 [1] NCCL INFO Channel 03/0 : 9[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9628:9786 [3] NCCL INFO Channel 03/0 : 11[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104434:104591 [5] NCCL INFO Channel 15/0 : 53[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104436:104596 [7] NCCL INFO Channel 13/0 : 55[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104430:104589 [1] NCCL INFO Channel 14/0 : 49[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9630:9789 [5] NCCL INFO Channel 14/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9627:9791 [2] NCCL INFO Channel 01/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762816:762987 [1] NCCL INFO Channel 12/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9632:9792 [7] NCCL INFO Channel 12/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762820:762981 [5] NCCL INFO Channel 12/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762822:762984 [7] NCCL INFO Channel 12/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9626:9790 [1] NCCL INFO Channel 05/0 : 9[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897246:897405 [6] NCCL INFO Channel 13/0 : 54[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897242:897400 [2] NCCL INFO Channel 01/0 : 42[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897247:897404 [7] NCCL INFO Channel 12/0 : 55[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897241:897402 [1] NCCL INFO Channel 05/0 : 41[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897243:897403 [3] NCCL INFO Channel 07/0 : 43[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897245:897401 [5] NCCL INFO Channel 01/0 : 45[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897246:897405 [6] NCCL INFO Channel 15/0 : 54[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762816:762987 [1] NCCL INFO Channel 13/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762820:762981 [5] NCCL INFO Channel 14/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762822:762984 [7] NCCL INFO Channel 13/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9628:9786 [3] NCCL INFO Channel 05/0 : 11[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104436:104596 [7] NCCL INFO Channel 14/0 : 55[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9630:9789 [5] NCCL INFO Channel 15/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897242:897400 [2] NCCL INFO Channel 03/0 : 42[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897247:897404 [7] NCCL INFO Channel 13/0 : 55[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897245:897401 [5] NCCL INFO Channel 03/0 : 45[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762816:762987 [1] NCCL INFO Channel 14/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897243:897403 [3] NCCL INFO Channel 09/0 : 43[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897241:897402 [1] NCCL INFO Channel 07/0 : 41[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897246:897405 [6] NCCL INFO Channel 01/0 : 46[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104430:104589 [1] NCCL INFO Channel 15/0 : 49[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897242:897400 [2] NCCL INFO Channel 05/0 : 42[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9627:9791 [2] NCCL INFO Channel 03/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9632:9792 [7] NCCL INFO Channel 13/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9628:9786 [3] NCCL INFO Channel 07/0 : 11[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9626:9790 [1] NCCL INFO Channel 07/0 : 9[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9630:9789 [5] NCCL INFO Channel 01/0 : 13[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9627:9791 [2] NCCL INFO Channel 05/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762822:762984 [7] NCCL INFO Channel 14/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762820:762981 [5] NCCL INFO Channel 15/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762816:762987 [1] NCCL INFO Channel 15/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897247:897404 [7] NCCL INFO Channel 14/0 : 55[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897245:897401 [5] NCCL INFO Channel 05/0 : 45[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9628:9786 [3] NCCL INFO Channel 09/0 : 11[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9632:9792 [7] NCCL INFO Channel 14/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9626:9790 [1] NCCL INFO Channel 11/0 : 9[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9630:9789 [5] NCCL INFO Channel 03/0 : 13[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9627:9791 [2] NCCL INFO Channel 07/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897241:897402 [1] NCCL INFO Channel 11/0 : 41[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897246:897405 [6] NCCL INFO Channel 03/0 : 46[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897243:897403 [3] NCCL INFO Channel 13/0 : 43[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897247:897404 [7] NCCL INFO Channel 01/0 : 47[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897242:897400 [2] NCCL INFO Channel 07/0 : 42[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897245:897401 [5] NCCL INFO Channel 07/0 : 45[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9628:9786 [3] NCCL INFO Channel 13/0 : 11[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897246:897405 [6] NCCL INFO Channel 05/0 : 46[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9626:9790 [1] NCCL INFO Channel 13/0 : 9[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9632:9792 [7] NCCL INFO Channel 01/0 : 15[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9630:9789 [5] NCCL INFO Channel 05/0 : 13[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9627:9791 [2] NCCL INFO Channel 09/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897241:897402 [1] NCCL INFO Channel 13/0 : 41[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9628:9786 [3] NCCL INFO Channel 15/0 : 11[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897247:897404 [7] NCCL INFO Channel 03/0 : 47[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897243:897403 [3] NCCL INFO Channel 15/0 : 43[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897242:897400 [2] NCCL INFO Channel 09/0 : 42[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897245:897401 [5] NCCL INFO Channel 09/0 : 45[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9630:9789 [5] NCCL INFO Channel 07/0 : 13[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9632:9792 [7] NCCL INFO Channel 03/0 : 15[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9627:9791 [2] NCCL INFO Channel 11/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9626:9790 [1] NCCL INFO Channel 15/0 : 9[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897246:897405 [6] NCCL INFO Channel 07/0 : 46[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897241:897402 [1] NCCL INFO Channel 15/0 : 41[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9631:9788 [6] NCCL INFO Channel 10/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897247:897404 [7] NCCL INFO Channel 05/0 : 47[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897245:897401 [5] NCCL INFO Channel 11/0 : 45[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897242:897400 [2] NCCL INFO Channel 11/0 : 42[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897246:897405 [6] NCCL INFO Channel 09/0 : 46[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9627:9791 [2] NCCL INFO Channel 13/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9630:9789 [5] NCCL INFO Channel 09/0 : 13[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9632:9792 [7] NCCL INFO Channel 05/0 : 15[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9631:9788 [6] NCCL INFO Channel 11/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897247:897404 [7] NCCL INFO Channel 07/0 : 47[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9627:9791 [2] NCCL INFO Channel 15/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897246:897405 [6] NCCL INFO Channel 11/0 : 46[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897245:897401 [5] NCCL INFO Channel 15/0 : 45[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897242:897400 [2] NCCL INFO Channel 13/0 : 42[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897247:897404 [7] NCCL INFO Channel 09/0 : 47[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9630:9789 [5] NCCL INFO Channel 11/0 : 13[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9632:9792 [7] NCCL INFO Channel 07/0 : 15[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9631:9788 [6] NCCL INFO Channel 12/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897246:897405 [6] NCCL INFO Channel 13/0 : 46[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897242:897400 [2] NCCL INFO Channel 15/0 : 42[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897247:897404 [7] NCCL INFO Channel 11/0 : 47[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9630:9789 [5] NCCL INFO Channel 15/0 : 13[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9632:9792 [7] NCCL INFO Channel 09/0 : 15[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9631:9788 [6] NCCL INFO Channel 13/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897246:897405 [6] NCCL INFO Channel 15/0 : 46[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897247:897404 [7] NCCL INFO Channel 13/0 : 47[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9631:9788 [6] NCCL INFO Channel 15/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9632:9792 [7] NCCL INFO Channel 11/0 : 15[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9631:9788 [6] NCCL INFO Channel 01/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9632:9792 [7] NCCL INFO Channel 13/0 : 15[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9631:9788 [6] NCCL INFO Channel 03/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9631:9788 [6] NCCL INFO Channel 05/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9631:9788 [6] NCCL INFO Channel 07/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9631:9788 [6] NCCL INFO Channel 09/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9631:9788 [6] NCCL INFO Channel 11/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9631:9788 [6] NCCL INFO Channel 13/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9631:9788 [6] NCCL INFO Channel 15/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127296:127451 [6] NCCL INFO Channel 10/0 : 30[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127297:127453 [7] NCCL INFO Channel 08/0 : 31[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127295:127452 [5] NCCL INFO Channel 12/0 : 29[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127296:127451 [6] NCCL INFO Channel 12/0 : 30[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127297:127453 [7] NCCL INFO Channel 10/0 : 31[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127295:127452 [5] NCCL INFO Channel 14/0 : 29[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127296:127451 [6] NCCL INFO Channel 14/0 : 30[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127297:127453 [7] NCCL INFO Channel 12/0 : 31[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127297:127453 [7] NCCL INFO Channel 14/0 : 31[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-46:702886:703049 [6] NCCL INFO Connected NVLS tree
+ip-26-0-162-46:702886:703049 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-162-46:702886:703049 [6] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-162-46:702886:703049 [6] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-162-46:702880:703042 [0] NCCL INFO Connected NVLS tree
+ip-26-0-162-46:702880:703042 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-162-46:702880:703042 [0] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-162-46:702880:703042 [0] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-162-46:702884:703047 [4] NCCL INFO Connected NVLS tree
+ip-26-0-162-46:702884:703047 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-162-46:702884:703047 [4] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-162-46:702884:703047 [4] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-162-46:702881:703043 [1] NCCL INFO Connected NVLS tree
+ip-26-0-162-46:702881:703043 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-162-46:702881:703043 [1] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-162-46:702881:703043 [1] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-167-9:1039539:1039698 [2] NCCL INFO Connected NVLS tree
+ip-26-0-167-9:1039539:1039698 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-167-9:1039539:1039698 [2] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-167-9:1039539:1039698 [2] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-162-46:702885:703048 [5] NCCL INFO Connected NVLS tree
+ip-26-0-162-46:702885:703048 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-162-46:702885:703048 [5] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-162-46:702885:703048 [5] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-167-9:1039540:1039705 [3] NCCL INFO Connected NVLS tree
+ip-26-0-167-9:1039542:1039701 [5] NCCL INFO Connected NVLS tree
+ip-26-0-167-9:1039540:1039705 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-167-9:1039542:1039701 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-167-9:1039540:1039705 [3] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-167-9:1039540:1039705 [3] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-167-9:1039542:1039701 [5] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-167-9:1039542:1039701 [5] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-162-46:702887:703045 [7] NCCL INFO Connected NVLS tree
+ip-26-0-162-46:702887:703045 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-162-46:702887:703045 [7] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-162-46:702887:703045 [7] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO Connected NVLS tree
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-167-9:1039537:1039703 [0] NCCL INFO Connected NVLS tree
+ip-26-0-167-9:1039537:1039703 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-167-9:1039537:1039703 [0] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-167-9:1039537:1039703 [0] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-167-9:1039544:1039702 [7] NCCL INFO Connected NVLS tree
+ip-26-0-167-9:1039544:1039702 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-167-9:1039544:1039702 [7] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-167-9:1039544:1039702 [7] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-165-213:9626:9790 [1] NCCL INFO Connected NVLS tree
+ip-26-0-165-213:9626:9790 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-213:9626:9790 [1] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-165-213:9626:9790 [1] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-167-9:1039543:1039699 [6] NCCL INFO Connected NVLS tree
+ip-26-0-167-9:1039543:1039699 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-167-9:1039543:1039699 [6] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-167-9:1039543:1039699 [6] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-165-213:9632:9792 [7] NCCL INFO Connected NVLS tree
+ip-26-0-165-213:9632:9792 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-213:9632:9792 [7] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-165-213:9632:9792 [7] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-167-9:1039538:1039704 [1] NCCL INFO Connected NVLS tree
+ip-26-0-167-9:1039538:1039704 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-167-9:1039538:1039704 [1] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-167-9:1039538:1039704 [1] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-166-36:104433:104590 [4] NCCL INFO Connected NVLS tree
+ip-26-0-166-36:104433:104590 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-36:104433:104590 [4] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-166-36:104433:104590 [4] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-165-213:9630:9789 [5] NCCL INFO Connected NVLS tree
+ip-26-0-165-213:9630:9789 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-213:9630:9789 [5] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-165-213:9630:9789 [5] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-162-46:702883:703044 [3] NCCL INFO Connected NVLS tree
+ip-26-0-162-46:702883:703044 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-162-46:702883:703044 [3] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-162-46:702883:703044 [3] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-167-9:1039543:1039699 [6] NCCL INFO comm 0x9d28300 rank 62 nranks 64 cudaDev 6 nvmlDev 6 busId b9000 commId 0x8b6df3a32fca5304 - Init COMPLETE
+ip-26-0-167-9:1039541:1039700 [4] NCCL INFO comm 0xa260f20 rank 60 nranks 64 cudaDev 4 nvmlDev 4 busId 97000 commId 0x8b6df3a32fca5304 - Init COMPLETE
+ip-26-0-167-9:1039544:1039702 [7] NCCL INFO comm 0x93e67f0 rank 63 nranks 64 cudaDev 7 nvmlDev 7 busId ca000 commId 0x8b6df3a32fca5304 - Init COMPLETE
+ip-26-0-167-9:1039538:1039704 [1] NCCL INFO comm 0x8e07600 rank 57 nranks 64 cudaDev 1 nvmlDev 1 busId 64000 commId 0x8b6df3a32fca5304 - Init COMPLETE
+ip-26-0-167-9:1039537:1039703 [0] NCCL INFO comm 0x9974360 rank 56 nranks 64 cudaDev 0 nvmlDev 0 busId 53000 commId 0x8b6df3a32fca5304 - Init COMPLETE
+ip-26-0-167-9:1039542:1039701 [5] NCCL INFO comm 0x8c1a980 rank 61 nranks 64 cudaDev 5 nvmlDev 5 busId a8000 commId 0x8b6df3a32fca5304 - Init COMPLETE
+ip-26-0-167-9:1039540:1039705 [3] NCCL INFO comm 0x955d240 rank 59 nranks 64 cudaDev 3 nvmlDev 3 busId 86000 commId 0x8b6df3a32fca5304 - Init COMPLETE
+ip-26-0-167-9:1039539:1039698 [2] NCCL INFO comm 0x8e6cc50 rank 58 nranks 64 cudaDev 2 nvmlDev 2 busId 75000 commId 0x8b6df3a32fca5304 - Init COMPLETE
+ip-26-0-165-213:9628:9786 [3] NCCL INFO Connected NVLS tree
+ip-26-0-165-213:9628:9786 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-213:9628:9786 [3] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-165-213:9628:9786 [3] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-166-36:104435:104595 [6] NCCL INFO Connected NVLS tree
+ip-26-0-166-36:104435:104595 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-36:104435:104595 [6] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-166-36:104435:104595 [6] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-162-46:702882:703046 [2] NCCL INFO Connected NVLS tree
+ip-26-0-162-46:702882:703046 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-162-46:702882:703046 [2] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-162-46:702882:703046 [2] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-166-36:104431:104593 [2] NCCL INFO Connected NVLS tree
+ip-26-0-166-36:104431:104593 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-36:104431:104593 [2] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-166-36:104431:104593 [2] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-162-46:702882:703046 [2] NCCL INFO comm 0x958dad0 rank 2 nranks 64 cudaDev 2 nvmlDev 2 busId 75000 commId 0x8b6df3a32fca5304 - Init COMPLETE
+ip-26-0-162-46:702886:703049 [6] NCCL INFO comm 0x8e18ff0 rank 6 nranks 64 cudaDev 6 nvmlDev 6 busId b9000 commId 0x8b6df3a32fca5304 - Init COMPLETE
+ip-26-0-162-46:702887:703045 [7] NCCL INFO comm 0x889e490 rank 7 nranks 64 cudaDev 7 nvmlDev 7 busId ca000 commId 0x8b6df3a32fca5304 - Init COMPLETE
+ip-26-0-162-46:702883:703044 [3] NCCL INFO comm 0x834f6a0 rank 3 nranks 64 cudaDev 3 nvmlDev 3 busId 86000 commId 0x8b6df3a32fca5304 - Init COMPLETE
+ip-26-0-162-46:702881:703043 [1] NCCL INFO comm 0x91d1af0 rank 1 nranks 64 cudaDev 1 nvmlDev 1 busId 64000 commId 0x8b6df3a32fca5304 - Init COMPLETE
+ip-26-0-162-46:702885:703048 [5] NCCL INFO comm 0x88bf5a0 rank 5 nranks 64 cudaDev 5 nvmlDev 5 busId a8000 commId 0x8b6df3a32fca5304 - Init COMPLETE
+ip-26-0-162-46:702884:703047 [4] NCCL INFO comm 0x8cdc140 rank 4 nranks 64 cudaDev 4 nvmlDev 4 busId 97000 commId 0x8b6df3a32fca5304 - Init COMPLETE
+ip-26-0-162-46:702880:703042 [0] NCCL INFO comm 0x967b3b0 rank 0 nranks 64 cudaDev 0 nvmlDev 0 busId 53000 commId 0x8b6df3a32fca5304 - Init COMPLETE
+ip-26-0-166-244:897244:897407 [4] NCCL INFO Connected NVLS tree
+ip-26-0-166-244:897244:897407 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-244:897244:897407 [4] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-166-244:897244:897407 [4] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-165-213:9631:9788 [6] NCCL INFO Connected NVLS tree
+ip-26-0-165-213:9631:9788 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-213:9631:9788 [6] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-165-213:9627:9791 [2] NCCL INFO Connected NVLS tree
+ip-26-0-165-213:9631:9788 [6] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-165-213:9627:9791 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-213:9627:9791 [2] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-165-213:9627:9791 [2] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-165-213:9629:9793 [4] NCCL INFO Connected NVLS tree
+ip-26-0-165-213:9629:9793 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-213:9629:9793 [4] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-165-213:9629:9793 [4] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-166-244:897246:897405 [6] NCCL INFO Connected NVLS tree
+ip-26-0-166-244:897246:897405 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-244:897246:897405 [6] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-166-244:897246:897405 [6] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-165-213:9625:9787 [0] NCCL INFO Connected NVLS tree
+ip-26-0-165-213:9625:9787 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-213:9625:9787 [0] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-165-213:9625:9787 [0] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-166-36:104429:104594 [0] NCCL INFO Connected NVLS tree
+ip-26-0-166-36:104429:104594 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-36:104429:104594 [0] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-166-36:104429:104594 [0] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-166-244:897240:897406 [0] NCCL INFO Connected NVLS tree
+ip-26-0-166-244:897240:897406 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-244:897240:897406 [0] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-166-244:897240:897406 [0] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-166-15:127291:127455 [1] NCCL INFO Connected NVLS tree
+ip-26-0-166-15:127291:127455 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-15:127291:127455 [1] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-166-15:127291:127455 [1] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-166-15:127297:127453 [7] NCCL INFO Connected NVLS tree
+ip-26-0-166-15:127297:127453 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-15:127297:127453 [7] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-166-15:127297:127453 [7] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-166-15:127295:127452 [5] NCCL INFO Connected NVLS tree
+ip-26-0-166-15:127295:127452 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-15:127295:127452 [5] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-166-15:127295:127452 [5] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-165-213:9632:9792 [7] NCCL INFO comm 0x8a49870 rank 15 nranks 64 cudaDev 7 nvmlDev 7 busId ca000 commId 0x8b6df3a32fca5304 - Init COMPLETE
+ip-26-0-165-213:9628:9786 [3] NCCL INFO comm 0x90d46e0 rank 11 nranks 64 cudaDev 3 nvmlDev 3 busId 86000 commId 0x8b6df3a32fca5304 - Init COMPLETE
+ip-26-0-165-213:9630:9789 [5] NCCL INFO comm 0x84f2a80 rank 13 nranks 64 cudaDev 5 nvmlDev 5 busId a8000 commId 0x8b6df3a32fca5304 - Init COMPLETE
+ip-26-0-165-213:9626:9790 [1] NCCL INFO comm 0x910f780 rank 9 nranks 64 cudaDev 1 nvmlDev 1 busId 64000 commId 0x8b6df3a32fca5304 - Init COMPLETE
+ip-26-0-165-213:9625:9787 [0] NCCL INFO comm 0x8bb7780 rank 8 nranks 64 cudaDev 0 nvmlDev 0 busId 53000 commId 0x8b6df3a32fca5304 - Init COMPLETE
+ip-26-0-165-213:9629:9793 [4] NCCL INFO comm 0x86eab90 rank 12 nranks 64 cudaDev 4 nvmlDev 4 busId 97000 commId 0x8b6df3a32fca5304 - Init COMPLETE
+ip-26-0-165-213:9631:9788 [6] NCCL INFO comm 0x98a0a00 rank 14 nranks 64 cudaDev 6 nvmlDev 6 busId b9000 commId 0x8b6df3a32fca5304 - Init COMPLETE
+ip-26-0-165-213:9627:9791 [2] NCCL INFO comm 0x86f9b60 rank 10 nranks 64 cudaDev 2 nvmlDev 2 busId 75000 commId 0x8b6df3a32fca5304 - Init COMPLETE
+ip-26-0-166-125:762815:762983 [0] NCCL INFO Connected NVLS tree
+ip-26-0-166-125:762815:762983 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-125:762815:762983 [0] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-166-125:762815:762983 [0] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-166-36:104430:104589 [1] NCCL INFO Connected NVLS tree
+ip-26-0-166-36:104430:104589 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-36:104430:104589 [1] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-166-36:104430:104589 [1] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-166-36:104434:104591 [5] NCCL INFO Connected NVLS tree
+ip-26-0-166-36:104434:104591 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-36:104434:104591 [5] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-166-36:104434:104591 [5] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-166-36:104436:104596 [7] NCCL INFO Connected NVLS tree
+ip-26-0-166-36:104436:104596 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-36:104436:104596 [7] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-166-36:104436:104596 [7] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-166-15:127292:127458 [2] NCCL INFO Connected NVLS tree
+ip-26-0-166-15:127292:127458 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-15:127292:127458 [2] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-166-15:127292:127458 [2] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-166-15:127293:127454 [3] NCCL INFO Connected NVLS tree
+ip-26-0-166-15:127293:127454 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-15:127293:127454 [3] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-166-15:127293:127454 [3] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-166-125:762822:762984 [7] NCCL INFO Connected NVLS tree
+ip-26-0-166-125:762822:762984 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-125:762822:762984 [7] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-166-125:762822:762984 [7] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-166-15:127290:127456 [0] NCCL INFO Connected NVLS tree
+ip-26-0-166-15:127290:127456 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-15:127290:127456 [0] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-166-15:127290:127456 [0] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-166-36:104432:104592 [3] NCCL INFO Connected NVLS tree
+ip-26-0-166-36:104432:104592 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-36:104432:104592 [3] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-166-36:104432:104592 [3] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-166-15:127296:127451 [6] NCCL INFO Connected NVLS tree
+ip-26-0-166-15:127296:127451 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-15:127296:127451 [6] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-166-15:127296:127451 [6] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-166-36:104429:104594 [0] NCCL INFO comm 0xa2374d0 rank 48 nranks 64 cudaDev 0 nvmlDev 0 busId 53000 commId 0x8b6df3a32fca5304 - Init COMPLETE
+ip-26-0-166-36:104433:104590 [4] NCCL INFO comm 0xa026de0 rank 52 nranks 64 cudaDev 4 nvmlDev 4 busId 97000 commId 0x8b6df3a32fca5304 - Init COMPLETE
+ip-26-0-166-36:104435:104595 [6] NCCL INFO comm 0x8b98830 rank 54 nranks 64 cudaDev 6 nvmlDev 6 busId b9000 commId 0x8b6df3a32fca5304 - Init COMPLETE
+ip-26-0-166-36:104431:104593 [2] NCCL INFO comm 0x955bca0 rank 50 nranks 64 cudaDev 2 nvmlDev 2 busId 75000 commId 0x8b6df3a32fca5304 - Init COMPLETE
+ip-26-0-166-36:104436:104596 [7] NCCL INFO comm 0x9507760 rank 55 nranks 64 cudaDev 7 nvmlDev 7 busId ca000 commId 0x8b6df3a32fca5304 - Init COMPLETE
+ip-26-0-166-36:104434:104591 [5] NCCL INFO comm 0x8836500 rank 53 nranks 64 cudaDev 5 nvmlDev 5 busId a8000 commId 0x8b6df3a32fca5304 - Init COMPLETE
+ip-26-0-166-36:104432:104592 [3] NCCL INFO comm 0x85daf90 rank 51 nranks 64 cudaDev 3 nvmlDev 3 busId 86000 commId 0x8b6df3a32fca5304 - Init COMPLETE
+ip-26-0-166-36:104430:104589 [1] NCCL INFO comm 0x9a479d0 rank 49 nranks 64 cudaDev 1 nvmlDev 1 busId 64000 commId 0x8b6df3a32fca5304 - Init COMPLETE
+ip-26-0-166-15:127294:127457 [4] NCCL INFO Connected NVLS tree
+ip-26-0-166-15:127294:127457 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-15:127294:127457 [4] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-166-15:127294:127457 [4] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-166-125:762817:762986 [2] NCCL INFO Connected NVLS tree
+ip-26-0-166-125:762817:762986 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-125:762817:762986 [2] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-166-125:762817:762986 [2] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-166-125:762821:762988 [6] NCCL INFO Connected NVLS tree
+ip-26-0-166-125:762821:762988 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-125:762821:762988 [6] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-166-125:762821:762988 [6] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-166-125:762816:762987 [1] NCCL INFO Connected NVLS tree
+ip-26-0-166-125:762816:762987 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-125:762816:762987 [1] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-166-125:762816:762987 [1] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-166-15:127295:127452 [5] NCCL INFO comm 0x88629b0 rank 29 nranks 64 cudaDev 5 nvmlDev 5 busId a8000 commId 0x8b6df3a32fca5304 - Init COMPLETE
+ip-26-0-166-15:127297:127453 [7] NCCL INFO comm 0x9b1a7f0 rank 31 nranks 64 cudaDev 7 nvmlDev 7 busId ca000 commId 0x8b6df3a32fca5304 - Init COMPLETE
+ip-26-0-166-15:127292:127458 [2] NCCL INFO comm 0x8fd9570 rank 26 nranks 64 cudaDev 2 nvmlDev 2 busId 75000 commId 0x8b6df3a32fca5304 - Init COMPLETE
+ip-26-0-166-15:127296:127451 [6] NCCL INFO comm 0x947e710 rank 30 nranks 64 cudaDev 6 nvmlDev 6 busId b9000 commId 0x8b6df3a32fca5304 - Init COMPLETE
+ip-26-0-166-15:127294:127457 [4] NCCL INFO comm 0x8d79aa0 rank 28 nranks 64 cudaDev 4 nvmlDev 4 busId 97000 commId 0x8b6df3a32fca5304 - Init COMPLETE
+ip-26-0-166-15:127291:127455 [1] NCCL INFO comm 0x9f78530 rank 25 nranks 64 cudaDev 1 nvmlDev 1 busId 64000 commId 0x8b6df3a32fca5304 - Init COMPLETE
+ip-26-0-166-15:127293:127454 [3] NCCL INFO comm 0x9ee0590 rank 27 nranks 64 cudaDev 3 nvmlDev 3 busId 86000 commId 0x8b6df3a32fca5304 - Init COMPLETE
+ip-26-0-166-15:127290:127456 [0] NCCL INFO comm 0x86f0680 rank 24 nranks 64 cudaDev 0 nvmlDev 0 busId 53000 commId 0x8b6df3a32fca5304 - Init COMPLETE
+ip-26-0-166-125:762818:762985 [3] NCCL INFO Connected NVLS tree
+ip-26-0-166-125:762818:762985 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-125:762818:762985 [3] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-166-125:762818:762985 [3] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-166-125:762819:762982 [4] NCCL INFO Connected NVLS tree
+ip-26-0-166-125:762819:762982 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-125:762819:762982 [4] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-166-125:762819:762982 [4] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-166-244:897242:897400 [2] NCCL INFO Connected NVLS tree
+ip-26-0-166-244:897242:897400 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-244:897242:897400 [2] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-166-244:897242:897400 [2] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-166-125:762820:762981 [5] NCCL INFO Connected NVLS tree
+ip-26-0-166-125:762820:762981 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-125:762820:762981 [5] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-166-125:762820:762981 [5] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-166-214:624945:625111 [2] NCCL INFO Connected NVLS tree
+ip-26-0-166-214:624945:625111 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-214:624945:625111 [2] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-166-214:624945:625111 [2] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-166-125:762815:762983 [0] NCCL INFO comm 0x832ff30 rank 16 nranks 64 cudaDev 0 nvmlDev 0 busId 53000 commId 0x8b6df3a32fca5304 - Init COMPLETE
+ip-26-0-166-125:762819:762982 [4] NCCL INFO comm 0x91ac710 rank 20 nranks 64 cudaDev 4 nvmlDev 4 busId 97000 commId 0x8b6df3a32fca5304 - Init COMPLETE
+ip-26-0-166-125:762817:762986 [2] NCCL INFO comm 0xa1ccb40 rank 18 nranks 64 cudaDev 2 nvmlDev 2 busId 75000 commId 0x8b6df3a32fca5304 - Init COMPLETE
+ip-26-0-166-125:762821:762988 [6] NCCL INFO comm 0x85f8640 rank 22 nranks 64 cudaDev 6 nvmlDev 6 busId b9000 commId 0x8b6df3a32fca5304 - Init COMPLETE
+ip-26-0-166-125:762816:762987 [1] NCCL INFO comm 0x93d1180 rank 17 nranks 64 cudaDev 1 nvmlDev 1 busId 64000 commId 0x8b6df3a32fca5304 - Init COMPLETE
+ip-26-0-166-125:762820:762981 [5] NCCL INFO comm 0x90d8cd0 rank 21 nranks 64 cudaDev 5 nvmlDev 5 busId a8000 commId 0x8b6df3a32fca5304 - Init COMPLETE
+ip-26-0-166-125:762822:762984 [7] NCCL INFO comm 0x9dbbc60 rank 23 nranks 64 cudaDev 7 nvmlDev 7 busId ca000 commId 0x8b6df3a32fca5304 - Init COMPLETE
+ip-26-0-166-125:762818:762985 [3] NCCL INFO comm 0x850d180 rank 19 nranks 64 cudaDev 3 nvmlDev 3 busId 86000 commId 0x8b6df3a32fca5304 - Init COMPLETE
+ip-26-0-166-214:624949:625110 [6] NCCL INFO Connected NVLS tree
+ip-26-0-166-214:624949:625110 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-214:624949:625110 [6] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-166-214:624949:625110 [6] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-166-214:624943:625108 [0] NCCL INFO Connected NVLS tree
+ip-26-0-166-214:624943:625108 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-214:624943:625108 [0] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-166-214:624943:625108 [0] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-166-244:897243:897403 [3] NCCL INFO Connected NVLS tree
+ip-26-0-166-244:897243:897403 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-244:897243:897403 [3] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-166-244:897243:897403 [3] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-166-214:624946:625107 [3] NCCL INFO Connected NVLS tree
+ip-26-0-166-214:624946:625107 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-214:624946:625107 [3] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-166-214:624946:625107 [3] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-166-214:624947:625105 [4] NCCL INFO Connected NVLS tree
+ip-26-0-166-214:624947:625105 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-214:624947:625105 [4] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-166-214:624947:625105 [4] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-166-244:897247:897404 [7] NCCL INFO Connected NVLS tree
+ip-26-0-166-244:897247:897404 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-244:897247:897404 [7] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-166-244:897247:897404 [7] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-166-214:624950:625109 [7] NCCL INFO Connected NVLS tree
+ip-26-0-166-214:624950:625109 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-214:624950:625109 [7] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-166-214:624950:625109 [7] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-166-244:897245:897401 [5] NCCL INFO Connected NVLS tree
+ip-26-0-166-244:897245:897401 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-244:897245:897401 [5] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-166-244:897245:897401 [5] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-166-244:897241:897402 [1] NCCL INFO Connected NVLS tree
+ip-26-0-166-244:897241:897402 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-244:897241:897402 [1] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-166-244:897241:897402 [1] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-166-214:624944:625106 [1] NCCL INFO Connected NVLS tree
+ip-26-0-166-214:624944:625106 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-214:624944:625106 [1] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-166-214:624944:625106 [1] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-166-214:624948:625104 [5] NCCL INFO Connected NVLS tree
+ip-26-0-166-214:624948:625104 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-214:624948:625104 [5] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-166-214:624948:625104 [5] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-166-244:897245:897401 [5] NCCL INFO comm 0x9819580 rank 45 nranks 64 cudaDev 5 nvmlDev 5 busId a8000 commId 0x8b6df3a32fca5304 - Init COMPLETE
+ip-26-0-166-244:897241:897402 [1] NCCL INFO comm 0x9f8ce70 rank 41 nranks 64 cudaDev 1 nvmlDev 1 busId 64000 commId 0x8b6df3a32fca5304 - Init COMPLETE
+ip-26-0-166-244:897240:897406 [0] NCCL INFO comm 0x90efd70 rank 40 nranks 64 cudaDev 0 nvmlDev 0 busId 53000 commId 0x8b6df3a32fca5304 - Init COMPLETE
+ip-26-0-166-244:897242:897400 [2] NCCL INFO comm 0x845ae40 rank 42 nranks 64 cudaDev 2 nvmlDev 2 busId 75000 commId 0x8b6df3a32fca5304 - Init COMPLETE
+ip-26-0-166-244:897244:897407 [4] NCCL INFO comm 0x8c707d0 rank 44 nranks 64 cudaDev 4 nvmlDev 4 busId 97000 commId 0x8b6df3a32fca5304 - Init COMPLETE
+ip-26-0-166-244:897246:897405 [6] NCCL INFO comm 0xa1fc860 rank 46 nranks 64 cudaDev 6 nvmlDev 6 busId b9000 commId 0x8b6df3a32fca5304 - Init COMPLETE
+ip-26-0-166-244:897243:897403 [3] NCCL INFO comm 0x87e71f0 rank 43 nranks 64 cudaDev 3 nvmlDev 3 busId 86000 commId 0x8b6df3a32fca5304 - Init COMPLETE
+ip-26-0-166-244:897247:897404 [7] NCCL INFO comm 0x977c640 rank 47 nranks 64 cudaDev 7 nvmlDev 7 busId ca000 commId 0x8b6df3a32fca5304 - Init COMPLETE
+ip-26-0-166-214:624947:625105 [4] NCCL INFO comm 0x9dbf120 rank 36 nranks 64 cudaDev 4 nvmlDev 4 busId 97000 commId 0x8b6df3a32fca5304 - Init COMPLETE
+ip-26-0-166-214:624943:625108 [0] NCCL INFO comm 0x919eef0 rank 32 nranks 64 cudaDev 0 nvmlDev 0 busId 53000 commId 0x8b6df3a32fca5304 - Init COMPLETE
+ip-26-0-166-214:624949:625110 [6] NCCL INFO comm 0x87cdfa0 rank 38 nranks 64 cudaDev 6 nvmlDev 6 busId b9000 commId 0x8b6df3a32fca5304 - Init COMPLETE
+ip-26-0-166-214:624945:625111 [2] NCCL INFO comm 0x9504970 rank 34 nranks 64 cudaDev 2 nvmlDev 2 busId 75000 commId 0x8b6df3a32fca5304 - Init COMPLETE
+ip-26-0-166-214:624946:625107 [3] NCCL INFO comm 0x9da4f90 rank 35 nranks 64 cudaDev 3 nvmlDev 3 busId 86000 commId 0x8b6df3a32fca5304 - Init COMPLETE
+ip-26-0-166-214:624950:625109 [7] NCCL INFO comm 0x8b7a600 rank 39 nranks 64 cudaDev 7 nvmlDev 7 busId ca000 commId 0x8b6df3a32fca5304 - Init COMPLETE
+ip-26-0-166-214:624948:625104 [5] NCCL INFO comm 0x9e36180 rank 37 nranks 64 cudaDev 5 nvmlDev 5 busId a8000 commId 0x8b6df3a32fca5304 - Init COMPLETE
+ip-26-0-166-214:624944:625106 [1] NCCL INFO comm 0x82f1510 rank 33 nranks 64 cudaDev 1 nvmlDev 1 busId 64000 commId 0x8b6df3a32fca5304 - Init COMPLETE
+ip-26-0-162-46:702880:703277 [0] NCCL INFO Using network Libfabric
+ip-26-0-166-125:762820:763219 [5] NCCL INFO Using network Libfabric
+ip-26-0-166-125:762819:763221 [4] NCCL INFO Using network Libfabric
+ip-26-0-162-46:702886:703281 [6] NCCL INFO Using network Libfabric
+ip-26-0-162-46:702884:703279 [4] NCCL INFO Using network Libfabric
+ip-26-0-162-46:702881:703280 [1] NCCL INFO Using network Libfabric
+ip-26-0-165-213:9625:10020 [0] NCCL INFO Using network Libfabric
+ip-26-0-162-46:702882:703284 [2] NCCL INFO Using network Libfabric
+ip-26-0-166-125:762818:763220 [3] NCCL INFO Using network Libfabric
+ip-26-0-166-125:762822:763222 [7] NCCL INFO Using network Libfabric
+ip-26-0-166-125:762816:763218 [1] NCCL INFO Using network Libfabric
+ip-26-0-166-125:762821:763224 [6] NCCL INFO Using network Libfabric
+ip-26-0-165-213:9629:10021 [4] NCCL INFO Using network Libfabric
+ip-26-0-166-125:762817:763223 [2] NCCL INFO Using network Libfabric
+ip-26-0-165-213:9631:10022 [6] NCCL INFO Using network Libfabric
+ip-26-0-165-213:9630:10024 [5] NCCL INFO Using network Libfabric
+ip-26-0-165-213:9628:10023 [3] NCCL INFO Using network Libfabric
+ip-26-0-165-213:9626:10025 [1] NCCL INFO Using network Libfabric
+ip-26-0-162-46:702887:703282 [7] NCCL INFO Using network Libfabric
+ip-26-0-162-46:702885:703278 [5] NCCL INFO Using network Libfabric
+ip-26-0-165-213:9627:10027 [2] NCCL INFO Using network Libfabric
+ip-26-0-166-125:762815:763225 [0] NCCL INFO Using network Libfabric
+ip-26-0-165-213:9632:10026 [7] NCCL INFO Using network Libfabric
+ip-26-0-162-46:702883:703283 [3] NCCL INFO Using network Libfabric
+ip-26-0-167-9:1039538:1039934 [1] NCCL INFO Using network Libfabric
+ip-26-0-166-244:897244:897636 [4] NCCL INFO Using network Libfabric
+ip-26-0-166-15:127295:127686 [5] NCCL INFO Using network Libfabric
+ip-26-0-166-15:127290:127685 [0] NCCL INFO Using network Libfabric
+ip-26-0-166-15:127294:127687 [4] NCCL INFO Using network Libfabric
+ip-26-0-167-9:1039544:1039931 [7] NCCL INFO Using network Libfabric
+ip-26-0-167-9:1039537:1039932 [0] NCCL INFO Using network Libfabric
+ip-26-0-167-9:1039539:1039938 [2] NCCL INFO Using network Libfabric
+ip-26-0-167-9:1039543:1039933 [6] NCCL INFO Using network Libfabric
+ip-26-0-167-9:1039542:1039936 [5] NCCL INFO Using network Libfabric
+ip-26-0-167-9:1039541:1039937 [4] NCCL INFO Using network Libfabric
+ip-26-0-166-36:104436:104824 [7] NCCL INFO Using network Libfabric
+ip-26-0-166-244:897243:897637 [3] NCCL INFO Using network Libfabric
+ip-26-0-166-15:127291:127690 [1] NCCL INFO Using network Libfabric
+ip-26-0-166-15:127292:127692 [2] NCCL INFO Using network Libfabric
+ip-26-0-166-244:897241:897635 [1] NCCL INFO Using network Libfabric
+ip-26-0-166-15:127297:127691 [7] NCCL INFO Using network Libfabric
+ip-26-0-166-36:104435:104826 [6] NCCL INFO Using network Libfabric
+ip-26-0-166-36:104432:104827 [3] NCCL INFO Using network Libfabric
+ip-26-0-166-214:624946:625337 [3] NCCL INFO Using network Libfabric
+ip-26-0-166-214:624945:625340 [2] NCCL INFO Using network Libfabric
+ip-26-0-166-36:104429:104829 [0] NCCL INFO Using network Libfabric
+ip-26-0-166-244:897240:897640 [0] NCCL INFO Using network Libfabric
+ip-26-0-166-214:624947:625339 [4] NCCL INFO Using network Libfabric
+ip-26-0-166-214:624950:625342 [7] NCCL INFO Using network Libfabric
+ip-26-0-166-36:104430:104830 [1] NCCL INFO Using network Libfabric
+ip-26-0-166-36:104434:104828 [5] NCCL INFO Using network Libfabric
+ip-26-0-166-214:624944:625341 [1] NCCL INFO Using network Libfabric
+ip-26-0-166-214:624943:625338 [0] NCCL INFO Using network Libfabric
+ip-26-0-166-214:624948:625343 [5] NCCL INFO Using network Libfabric
+ip-26-0-166-214:624949:625344 [6] NCCL INFO Using network Libfabric
+ip-26-0-166-244:897246:897641 [6] NCCL INFO Using network Libfabric
+ip-26-0-166-244:897247:897642 [7] NCCL INFO Using network Libfabric
+ip-26-0-166-244:897242:897639 [2] NCCL INFO Using network Libfabric
+ip-26-0-167-9:1039540:1039935 [3] NCCL INFO Using network Libfabric
+ip-26-0-166-244:897245:897638 [5] NCCL INFO Using network Libfabric
+ip-26-0-166-15:127293:127689 [3] NCCL INFO Using network Libfabric
+ip-26-0-166-15:127296:127688 [6] NCCL INFO Using network Libfabric
+ip-26-0-166-36:104433:104825 [4] NCCL INFO Using network Libfabric
+ip-26-0-166-36:104431:104831 [2] NCCL INFO Using network Libfabric
+ip-26-0-162-46:702883:703283 [3] NCCL INFO comm 0x87b1a30 rank 3 nranks 64 cudaDev 3 nvmlDev 3 busId 86000 commId 0xda4a0eda8742c6eb - Init START
+ip-26-0-162-46:702881:703280 [1] NCCL INFO comm 0x963a9a0 rank 1 nranks 64 cudaDev 1 nvmlDev 1 busId 64000 commId 0xda4a0eda8742c6eb - Init START
+ip-26-0-162-46:702880:703277 [0] NCCL INFO comm 0x9b71eb0 rank 0 nranks 64 cudaDev 0 nvmlDev 0 busId 53000 commId 0xda4a0eda8742c6eb - Init START
+ip-26-0-162-46:702884:703279 [4] NCCL INFO comm 0x9142ba0 rank 4 nranks 64 cudaDev 4 nvmlDev 4 busId 97000 commId 0xda4a0eda8742c6eb - Init START
+ip-26-0-162-46:702882:703284 [2] NCCL INFO comm 0x99f1a00 rank 2 nranks 64 cudaDev 2 nvmlDev 2 busId 75000 commId 0xda4a0eda8742c6eb - Init START
+ip-26-0-162-46:702885:703278 [5] NCCL INFO comm 0x8d96ff0 rank 5 nranks 64 cudaDev 5 nvmlDev 5 busId a8000 commId 0xda4a0eda8742c6eb - Init START
+ip-26-0-165-213:9632:10026 [7] NCCL INFO comm 0x8f2c3f0 rank 15 nranks 64 cudaDev 7 nvmlDev 7 busId ca000 commId 0xda4a0eda8742c6eb - Init START
+ip-26-0-167-9:1039544:1039931 [7] NCCL INFO comm 0x98c6e90 rank 63 nranks 64 cudaDev 7 nvmlDev 7 busId ca000 commId 0xda4a0eda8742c6eb - Init START
+ip-26-0-167-9:1039543:1039933 [6] NCCL INFO comm 0xa191260 rank 62 nranks 64 cudaDev 6 nvmlDev 6 busId b9000 commId 0xda4a0eda8742c6eb - Init START
+ip-26-0-162-46:702887:703282 [7] NCCL INFO comm 0x8d87da0 rank 7 nranks 64 cudaDev 7 nvmlDev 7 busId ca000 commId 0xda4a0eda8742c6eb - Init START
+ip-26-0-162-46:702886:703281 [6] NCCL INFO comm 0x927c830 rank 6 nranks 64 cudaDev 6 nvmlDev 6 busId b9000 commId 0xda4a0eda8742c6eb - Init START
+ip-26-0-166-125:762816:763218 [1] NCCL INFO comm 0x9836d20 rank 17 nranks 64 cudaDev 1 nvmlDev 1 busId 64000 commId 0xda4a0eda8742c6eb - Init START
+ip-26-0-166-125:762817:763223 [2] NCCL INFO comm 0xa633eb0 rank 18 nranks 64 cudaDev 2 nvmlDev 2 busId 75000 commId 0xda4a0eda8742c6eb - Init START
+ip-26-0-166-125:762818:763220 [3] NCCL INFO comm 0x89715f0 rank 19 nranks 64 cudaDev 3 nvmlDev 3 busId 86000 commId 0xda4a0eda8742c6eb - Init START
+ip-26-0-166-125:762820:763219 [5] NCCL INFO comm 0x953c5f0 rank 21 nranks 64 cudaDev 5 nvmlDev 5 busId a8000 commId 0xda4a0eda8742c6eb - Init START
+ip-26-0-166-125:762819:763221 [4] NCCL INFO comm 0x961bfa0 rank 20 nranks 64 cudaDev 4 nvmlDev 4 busId 97000 commId 0xda4a0eda8742c6eb - Init START
+ip-26-0-166-125:762815:763225 [0] NCCL INFO comm 0x88082b0 rank 16 nranks 64 cudaDev 0 nvmlDev 0 busId 53000 commId 0xda4a0eda8742c6eb - Init START
+ip-26-0-166-15:127297:127691 [7] NCCL INFO comm 0x9fff530 rank 31 nranks 64 cudaDev 7 nvmlDev 7 busId ca000 commId 0xda4a0eda8742c6eb - Init START
+ip-26-0-166-15:127296:127688 [6] NCCL INFO comm 0x98e93f0 rank 30 nranks 64 cudaDev 6 nvmlDev 6 busId b9000 commId 0xda4a0eda8742c6eb - Init START
+ip-26-0-166-15:127291:127690 [1] NCCL INFO comm 0xa457d40 rank 25 nranks 64 cudaDev 1 nvmlDev 1 busId 64000 commId 0xda4a0eda8742c6eb - Init START
+ip-26-0-166-15:127295:127686 [5] NCCL INFO comm 0x8cd1ce0 rank 29 nranks 64 cudaDev 5 nvmlDev 5 busId a8000 commId 0xda4a0eda8742c6eb - Init START
+ip-26-0-166-15:127292:127692 [2] NCCL INFO comm 0x94433a0 rank 26 nranks 64 cudaDev 2 nvmlDev 2 busId 75000 commId 0xda4a0eda8742c6eb - Init START
+ip-26-0-166-15:127293:127689 [3] NCCL INFO comm 0xa34b190 rank 27 nranks 64 cudaDev 3 nvmlDev 3 busId 86000 commId 0xda4a0eda8742c6eb - Init START
+ip-26-0-167-9:1039542:1039936 [5] NCCL INFO comm 0x907fcb0 rank 61 nranks 64 cudaDev 5 nvmlDev 5 busId a8000 commId 0xda4a0eda8742c6eb - Init START
+ip-26-0-167-9:1039540:1039935 [3] NCCL INFO comm 0x99bd170 rank 59 nranks 64 cudaDev 3 nvmlDev 3 busId 86000 commId 0xda4a0eda8742c6eb - Init START
+ip-26-0-167-9:1039541:1039937 [4] NCCL INFO comm 0xa6c22f0 rank 60 nranks 64 cudaDev 4 nvmlDev 4 busId 97000 commId 0xda4a0eda8742c6eb - Init START
+ip-26-0-166-36:104436:104824 [7] NCCL INFO comm 0x99716e0 rank 55 nranks 64 cudaDev 7 nvmlDev 7 busId ca000 commId 0xda4a0eda8742c6eb - Init START
+ip-26-0-166-36:104435:104826 [6] NCCL INFO comm 0x90001f0 rank 54 nranks 64 cudaDev 6 nvmlDev 6 busId b9000 commId 0xda4a0eda8742c6eb - Init START
+ip-26-0-166-36:104434:104828 [5] NCCL INFO comm 0x8ca8b60 rank 53 nranks 64 cudaDev 5 nvmlDev 5 busId a8000 commId 0xda4a0eda8742c6eb - Init START
+ip-26-0-166-244:897241:897635 [1] NCCL INFO comm 0xa3fa260 rank 41 nranks 64 cudaDev 1 nvmlDev 1 busId 64000 commId 0xda4a0eda8742c6eb - Init START
+ip-26-0-166-244:897244:897636 [4] NCCL INFO comm 0x90d7750 rank 44 nranks 64 cudaDev 4 nvmlDev 4 busId 97000 commId 0xda4a0eda8742c6eb - Init START
+ip-26-0-166-244:897246:897641 [6] NCCL INFO comm 0xa665d00 rank 46 nranks 64 cudaDev 6 nvmlDev 6 busId b9000 commId 0xda4a0eda8742c6eb - Init START
+ip-26-0-166-244:897247:897642 [7] NCCL INFO comm 0x9be8840 rank 47 nranks 64 cudaDev 7 nvmlDev 7 busId ca000 commId 0xda4a0eda8742c6eb - Init START
+ip-26-0-166-244:897240:897640 [0] NCCL INFO comm 0x95571b0 rank 40 nranks 64 cudaDev 0 nvmlDev 0 busId 53000 commId 0xda4a0eda8742c6eb - Init START
+ip-26-0-166-244:897245:897638 [5] NCCL INFO comm 0x9cf54b0 rank 45 nranks 64 cudaDev 5 nvmlDev 5 busId a8000 commId 0xda4a0eda8742c6eb - Init START
+ip-26-0-165-213:9631:10022 [6] NCCL INFO comm 0x9d025f0 rank 14 nranks 64 cudaDev 6 nvmlDev 6 busId b9000 commId 0xda4a0eda8742c6eb - Init START
+ip-26-0-165-213:9630:10024 [5] NCCL INFO comm 0x89d16b0 rank 13 nranks 64 cudaDev 5 nvmlDev 5 busId a8000 commId 0xda4a0eda8742c6eb - Init START
+ip-26-0-165-213:9625:10020 [0] NCCL INFO comm 0x901a8b0 rank 8 nranks 64 cudaDev 0 nvmlDev 0 busId 53000 commId 0xda4a0eda8742c6eb - Init START
+ip-26-0-165-213:9629:10021 [4] NCCL INFO comm 0x8b51550 rank 12 nranks 64 cudaDev 4 nvmlDev 4 busId 97000 commId 0xda4a0eda8742c6eb - Init START
+ip-26-0-166-214:624950:625342 [7] NCCL INFO comm 0x8fdb6f0 rank 39 nranks 64 cudaDev 7 nvmlDev 7 busId ca000 commId 0xda4a0eda8742c6eb - Init START
+ip-26-0-166-214:624943:625338 [0] NCCL INFO comm 0x967f900 rank 32 nranks 64 cudaDev 0 nvmlDev 0 busId 53000 commId 0xda4a0eda8742c6eb - Init START
+ip-26-0-166-214:624949:625344 [6] NCCL INFO comm 0x8cb8cf0 rank 38 nranks 64 cudaDev 6 nvmlDev 6 busId b9000 commId 0xda4a0eda8742c6eb - Init START
+ip-26-0-166-214:624944:625341 [1] NCCL INFO comm 0x8750fa0 rank 33 nranks 64 cudaDev 1 nvmlDev 1 busId 64000 commId 0xda4a0eda8742c6eb - Init START
+ip-26-0-166-214:624948:625343 [5] NCCL INFO comm 0xa2a1a50 rank 37 nranks 64 cudaDev 5 nvmlDev 5 busId a8000 commId 0xda4a0eda8742c6eb - Init START
+ip-26-0-166-15:127294:127687 [4] NCCL INFO comm 0x91e17f0 rank 28 nranks 64 cudaDev 4 nvmlDev 4 busId 97000 commId 0xda4a0eda8742c6eb - Init START
+ip-26-0-166-15:127290:127685 [0] NCCL INFO comm 0x8bd7da0 rank 24 nranks 64 cudaDev 0 nvmlDev 0 busId 53000 commId 0xda4a0eda8742c6eb - Init START
+ip-26-0-167-9:1039539:1039938 [2] NCCL INFO comm 0x9341460 rank 58 nranks 64 cudaDev 2 nvmlDev 2 busId 75000 commId 0xda4a0eda8742c6eb - Init START
+ip-26-0-167-9:1039538:1039934 [1] NCCL INFO comm 0x926a6f0 rank 57 nranks 64 cudaDev 1 nvmlDev 1 busId 64000 commId 0xda4a0eda8742c6eb - Init START
+ip-26-0-167-9:1039537:1039932 [0] NCCL INFO comm 0x9e4ce90 rank 56 nranks 64 cudaDev 0 nvmlDev 0 busId 53000 commId 0xda4a0eda8742c6eb - Init START
+ip-26-0-165-213:9626:10025 [1] NCCL INFO comm 0x95f5af0 rank 9 nranks 64 cudaDev 1 nvmlDev 1 busId 64000 commId 0xda4a0eda8742c6eb - Init START
+ip-26-0-165-213:9628:10023 [3] NCCL INFO comm 0x9540830 rank 11 nranks 64 cudaDev 3 nvmlDev 3 busId 86000 commId 0xda4a0eda8742c6eb - Init START
+ip-26-0-165-213:9627:10027 [2] NCCL INFO comm 0x8b55300 rank 10 nranks 64 cudaDev 2 nvmlDev 2 busId 75000 commId 0xda4a0eda8742c6eb - Init START
+ip-26-0-166-125:762821:763224 [6] NCCL INFO comm 0x8ae3f70 rank 22 nranks 64 cudaDev 6 nvmlDev 6 busId b9000 commId 0xda4a0eda8742c6eb - Init START
+ip-26-0-166-125:762822:763222 [7] NCCL INFO comm 0xa21d6a0 rank 23 nranks 64 cudaDev 7 nvmlDev 7 busId ca000 commId 0xda4a0eda8742c6eb - Init START
+ip-26-0-166-36:104432:104827 [3] NCCL INFO comm 0x8a395f0 rank 51 nranks 64 cudaDev 3 nvmlDev 3 busId 86000 commId 0xda4a0eda8742c6eb - Init START
+ip-26-0-166-36:104431:104831 [2] NCCL INFO comm 0x9a41180 rank 50 nranks 64 cudaDev 2 nvmlDev 2 busId 75000 commId 0xda4a0eda8742c6eb - Init START
+ip-26-0-166-36:104433:104825 [4] NCCL INFO comm 0xa48b0f0 rank 52 nranks 64 cudaDev 4 nvmlDev 4 busId 97000 commId 0xda4a0eda8742c6eb - Init START
+ip-26-0-166-36:104430:104830 [1] NCCL INFO comm 0x9eb3ae0 rank 49 nranks 64 cudaDev 1 nvmlDev 1 busId 64000 commId 0xda4a0eda8742c6eb - Init START
+ip-26-0-166-36:104429:104829 [0] NCCL INFO comm 0xa69aaf0 rank 48 nranks 64 cudaDev 0 nvmlDev 0 busId 53000 commId 0xda4a0eda8742c6eb - Init START
+ip-26-0-166-244:897242:897639 [2] NCCL INFO comm 0x8943e30 rank 42 nranks 64 cudaDev 2 nvmlDev 2 busId 75000 commId 0xda4a0eda8742c6eb - Init START
+ip-26-0-166-244:897243:897637 [3] NCCL INFO comm 0x8c4b580 rank 43 nranks 64 cudaDev 3 nvmlDev 3 busId 86000 commId 0xda4a0eda8742c6eb - Init START
+ip-26-0-166-214:624945:625340 [2] NCCL INFO comm 0x99ed060 rank 34 nranks 64 cudaDev 2 nvmlDev 2 busId 75000 commId 0xda4a0eda8742c6eb - Init START
+ip-26-0-166-214:624947:625339 [4] NCCL INFO comm 0xa228a50 rank 36 nranks 64 cudaDev 4 nvmlDev 4 busId 97000 commId 0xda4a0eda8742c6eb - Init START
+ip-26-0-166-214:624946:625337 [3] NCCL INFO comm 0xa2847d0 rank 35 nranks 64 cudaDev 3 nvmlDev 3 busId 86000 commId 0xda4a0eda8742c6eb - Init START
+ip-26-0-166-36:104436:104824 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-36:104434:104828 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-214:624950:625342 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-167-9:1039539:1039938 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-167-9:1039538:1039934 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-162-46:702887:703282 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-162-46:702886:703281 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-162-46:702881:703280 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-162-46:702884:703279 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-36:104433:104825 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-36:104432:104827 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-36:104435:104826 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-167-9:1039537:1039932 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-167-9:1039540:1039935 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-167-9:1039541:1039937 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-167-9:1039542:1039936 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-162-46:702882:703284 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-162-46:702885:703278 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-162-46:702883:703283 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-36:104430:104830 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-36:104431:104831 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-36:104429:104829 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-125:762822:763222 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-244:897247:897642 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-244:897244:897636 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-244:897245:897638 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-244:897246:897641 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-244:897241:897635 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-165-213:9631:10022 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-165-213:9632:10026 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-165-213:9628:10023 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-165-213:9630:10024 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-165-213:9626:10025 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-214:624949:625344 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-15:127297:127691 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-167-9:1039543:1039933 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-162-46:702880:703277 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/349
+ip-26-0-166-214:624947:625339 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-15:127294:127687 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-15:127296:127688 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-167-9:1039544:1039931 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-15:127293:127689 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-15:127295:127686 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-15:127292:127692 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-125:762819:763221 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-125:762818:763220 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-244:897242:897639 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-244:897243:897637 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-244:897240:897640 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-165-213:9627:10027 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-15:127291:127690 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-125:762817:763223 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-125:762816:763218 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-125:762820:763219 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-125:762815:763225 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-125:762821:763224 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-165-213:9629:10021 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-214:624948:625343 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-214:624945:625340 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-214:624946:625337 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-214:624944:625341 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-214:624943:625338 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-15:127290:127685 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-165-213:9625:10020 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-244:897243:897637 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-166-244:897243:897637 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-166-214:624943:625338 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-166-214:624943:625338 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-166-214:624950:625342 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-166-214:624950:625342 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-166-244:897241:897635 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-166-244:897241:897635 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-166-15:127291:127690 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-166-15:127291:127690 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-166-244:897244:897636 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-166-244:897244:897636 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-165-213:9625:10020 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-165-213:9625:10020 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-166-214:624946:625337 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-166-214:624946:625337 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-166-214:624949:625344 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-166-214:624949:625344 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-166-214:624944:625341 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-166-214:624944:625341 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-166-244:897246:897641 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-166-244:897246:897641 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-166-214:624948:625343 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-166-214:624948:625343 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-166-244:897240:897640 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-166-244:897240:897640 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-166-15:127297:127691 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-166-15:127297:127691 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-166-214:624947:625339 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-166-214:624947:625339 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-166-214:624945:625340 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-166-214:624945:625340 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-166-244:897242:897639 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-166-244:897242:897639 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-166-244:897247:897642 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-166-244:897247:897642 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-166-244:897245:897638 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-166-244:897245:897638 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-166-15:127293:127689 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-166-15:127293:127689 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-167-9:1039538:1039934 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-167-9:1039538:1039934 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-162-46:702885:703278 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-162-46:702885:703278 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-166-125:762819:763221 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,fff00000,00000000
+ip-26-0-166-125:762819:763221 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-166-125:762816:763218 [1] NCCL INFO Setting affinity for GPU 1 to 0fff,ffffffff
+ip-26-0-166-125:762816:763218 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-167-9:1039542:1039936 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-167-9:1039542:1039936 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-166-15:127296:127688 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-166-15:127296:127688 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-165-213:9632:10026 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-165-213:9632:10026 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-162-46:702887:703282 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-162-46:702887:703282 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-166-15:127295:127686 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-166-15:127295:127686 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-166-125:762817:763223 [2] NCCL INFO Setting affinity for GPU 2 to 0fff,ffffffff
+ip-26-0-166-125:762817:763223 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-167-9:1039543:1039933 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-167-9:1039543:1039933 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-166-15:127290:127685 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-165-213:9628:10023 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-166-15:127290:127685 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-165-213:9628:10023 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-166-15:127294:127687 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-166-15:127294:127687 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-167-9:1039544:1039931 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-167-9:1039544:1039931 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-162-46:702884:703279 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-162-46:702884:703279 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-166-15:127292:127692 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-166-15:127292:127692 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-165-213:9627:10027 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-165-213:9627:10027 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-165-213:9631:10022 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-165-213:9631:10022 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-165-213:9630:10024 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-165-213:9630:10024 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-166-125:762822:763222 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,fff00000,00000000
+ip-26-0-166-125:762822:763222 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-165-213:9626:10025 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-165-213:9626:10025 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-162-46:702886:703281 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-162-46:702886:703281 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-167-9:1039541:1039937 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-167-9:1039541:1039937 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-166-125:762818:763220 [3] NCCL INFO Setting affinity for GPU 3 to 0fff,ffffffff
+ip-26-0-166-125:762818:763220 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-165-213:9629:10021 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-165-213:9629:10021 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-167-9:1039540:1039935 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-167-9:1039540:1039935 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-162-46:702882:703284 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-162-46:702882:703284 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-162-46:702880:703277 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-162-46:702880:703277 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-166-125:762821:763224 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,fff00000,00000000
+ip-26-0-166-125:762821:763224 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-167-9:1039539:1039938 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-167-9:1039539:1039938 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-166-125:762820:763219 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,fff00000,00000000
+ip-26-0-166-125:762820:763219 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-167-9:1039537:1039932 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-167-9:1039537:1039932 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-162-46:702881:703280 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-162-46:702881:703280 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-166-125:762815:763225 [0] NCCL INFO Setting affinity for GPU 0 to 0fff,ffffffff
+ip-26-0-166-125:762815:763225 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-162-46:702883:703283 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-162-46:702883:703283 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-166-36:104435:104826 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-166-36:104435:104826 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-166-36:104431:104831 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-166-36:104431:104831 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-166-36:104434:104828 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-166-36:104434:104828 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-166-36:104430:104830 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-166-36:104430:104830 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-166-36:104436:104824 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-166-36:104436:104824 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-166-36:104432:104827 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-166-36:104432:104827 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-166-36:104429:104829 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-166-36:104429:104829 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-166-36:104433:104825 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-166-36:104433:104825 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-166-36:104433:104825 [4] NCCL INFO Trees [0] 53/-1/-1->52->51 [1] 53/-1/-1->52->51 [2] 53/-1/-1->52->51 [3] 53/-1/-1->52->51 [4] 53/44/60->52->36 [5] -1/-1/-1->52->51 [6] 53/-1/-1->52->51 [7] 53/-1/-1->52->51 [8] 53/-1/-1->52->51 [9] 53/-1/-1->52->51 [10] 53/-1/-1->52->51 [11] 53/-1/-1->52->51 [12] 53/-1/-1->52->44 [13] -1/-1/-1->52->51 [14] 53/-1/-1->52->51 [15] 53/-1/-1->52->51
+ip-26-0-166-36:104433:104825 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-36:104434:104828 [5] NCCL INFO Trees [0] 54/-1/-1->53->52 [1] 54/-1/-1->53->52 [2] 54/-1/-1->53->52 [3] 54/-1/-1->53->52 [4] 54/-1/-1->53->52 [5] 54/45/61->53->37 [6] -1/-1/-1->53->52 [7] 54/-1/-1->53->52 [8] 54/-1/-1->53->52 [9] 54/-1/-1->53->52 [10] 54/-1/-1->53->52 [11] 54/-1/-1->53->52 [12] 54/-1/-1->53->52 [13] 54/-1/-1->53->45 [14] -1/-1/-1->53->52 [15] 54/-1/-1->53->52
+ip-26-0-166-36:104434:104828 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-36:104435:104826 [6] NCCL INFO Trees [0] 55/-1/-1->54->53 [1] 55/-1/-1->54->53 [2] 55/-1/-1->54->53 [3] 55/-1/-1->54->53 [4] 55/-1/-1->54->53 [5] 55/-1/-1->54->53 [6] 55/46/62->54->38 [7] -1/-1/-1->54->53 [8] 55/-1/-1->54->53 [9] 55/-1/-1->54->53 [10] 55/-1/-1->54->53 [11] 55/-1/-1->54->53 [12] 55/-1/-1->54->53 [13] 55/-1/-1->54->53 [14] 55/-1/-1->54->46 [15] -1/-1/-1->54->53
+ip-26-0-166-36:104435:104826 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-36:104436:104824 [7] NCCL INFO Trees [0] -1/-1/-1->55->54 [1] 48/-1/-1->55->54 [2] 48/-1/-1->55->54 [3] 48/-1/-1->55->54 [4] 48/-1/-1->55->54 [5] 48/-1/-1->55->54 [6] 48/-1/-1->55->54 [7] 48/47/63->55->39 [8] -1/-1/-1->55->54 [9] 48/-1/-1->55->54 [10] 48/-1/-1->55->54 [11] 48/-1/-1->55->54 [12] 48/-1/-1->55->54 [13] 48/-1/-1->55->54 [14] 48/-1/-1->55->54 [15] 48/-1/-1->55->47
+ip-26-0-166-36:104436:104824 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-167-9:1039537:1039932 [0] NCCL INFO Trees [0] 57/-1/-1->56->48 [1] -1/-1/-1->56->63 [2] 57/-1/-1->56->63 [3] 57/-1/-1->56->63 [4] 57/-1/-1->56->63 [5] 57/-1/-1->56->63 [6] 57/-1/-1->56->63 [7] 57/-1/-1->56->63 [8] 57/24/-1->56->-1 [9] -1/-1/-1->56->63 [10] 57/-1/-1->56->63 [11] 57/-1/-1->56->63 [12] 57/-1/-1->56->63 [13] 57/-1/-1->56->63 [14] 57/-1/-1->56->63 [15] 57/-1/-1->56->63
+ip-26-0-167-9:1039537:1039932 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-167-9:1039544:1039931 [7] NCCL INFO Trees [0] -1/-1/-1->63->62 [1] 56/-1/-1->63->62 [2] 56/-1/-1->63->62 [3] 56/-1/-1->63->62 [4] 56/-1/-1->63->62 [5] 56/-1/-1->63->62 [6] 56/-1/-1->63->62 [7] 56/-1/-1->63->55 [8] -1/-1/-1->63->62 [9] 56/-1/-1->63->62 [10] 56/-1/-1->63->62 [11] 56/-1/-1->63->62 [12] 56/-1/-1->63->62 [13] 56/-1/-1->63->62 [14] 56/-1/-1->63->62 [15] 56/31/-1->63->-1
+ip-26-0-167-9:1039544:1039931 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-162-46:702880:703277 [0] NCCL INFO NVLS Head  0:  0  8 16 24 32 40 48 56
+ip-26-0-162-46:702881:703280 [1] NCCL INFO NVLS Head  0:  0  8 16 24 32 40 48 56
+ip-26-0-162-46:702881:703280 [1] NCCL INFO NVLS Head  1:  1  9 17 25 33 41 49 57
+ip-26-0-162-46:702880:703277 [0] NCCL INFO NVLS Head  1:  1  9 17 25 33 41 49 57
+ip-26-0-162-46:702881:703280 [1] NCCL INFO NVLS Head  2:  2 10 18 26 34 42 50 58
+ip-26-0-162-46:702880:703277 [0] NCCL INFO NVLS Head  2:  2 10 18 26 34 42 50 58
+ip-26-0-162-46:702887:703282 [7] NCCL INFO NVLS Head  0:  0  8 16 24 32 40 48 56
+ip-26-0-162-46:702881:703280 [1] NCCL INFO NVLS Head  3:  3 11 19 27 35 43 51 59
+ip-26-0-162-46:702880:703277 [0] NCCL INFO NVLS Head  3:  3 11 19 27 35 43 51 59
+ip-26-0-162-46:702887:703282 [7] NCCL INFO NVLS Head  1:  1  9 17 25 33 41 49 57
+ip-26-0-162-46:702881:703280 [1] NCCL INFO NVLS Head  4:  4 12 20 28 36 44 52 60
+ip-26-0-162-46:702880:703277 [0] NCCL INFO NVLS Head  4:  4 12 20 28 36 44 52 60
+ip-26-0-162-46:702887:703282 [7] NCCL INFO NVLS Head  2:  2 10 18 26 34 42 50 58
+ip-26-0-162-46:702886:703281 [6] NCCL INFO NVLS Head  0:  0  8 16 24 32 40 48 56
+ip-26-0-162-46:702880:703277 [0] NCCL INFO NVLS Head  5:  5 13 21 29 37 45 53 61
+ip-26-0-162-46:702881:703280 [1] NCCL INFO NVLS Head  5:  5 13 21 29 37 45 53 61
+ip-26-0-162-46:702887:703282 [7] NCCL INFO NVLS Head  3:  3 11 19 27 35 43 51 59
+ip-26-0-162-46:702886:703281 [6] NCCL INFO NVLS Head  1:  1  9 17 25 33 41 49 57
+ip-26-0-162-46:702881:703280 [1] NCCL INFO NVLS Head  6:  6 14 22 30 38 46 54 62
+ip-26-0-162-46:702880:703277 [0] NCCL INFO NVLS Head  6:  6 14 22 30 38 46 54 62
+ip-26-0-162-46:702887:703282 [7] NCCL INFO NVLS Head  4:  4 12 20 28 36 44 52 60
+ip-26-0-162-46:702885:703278 [5] NCCL INFO NVLS Head  0:  0  8 16 24 32 40 48 56
+ip-26-0-162-46:702881:703280 [1] NCCL INFO NVLS Head  7:  7 15 23 31 39 47 55 63
+ip-26-0-165-213:9625:10020 [0] NCCL INFO Trees [0] 9/-1/-1->8->16 [1] -1/-1/-1->8->15 [2] 9/-1/-1->8->15 [3] 9/-1/-1->8->15 [4] 9/-1/-1->8->15 [5] 9/-1/-1->8->15 [6] 9/-1/-1->8->15 [7] 9/-1/-1->8->15 [8] 9/16/0->8->24 [9] -1/-1/-1->8->15 [10] 9/-1/-1->8->15 [11] 9/-1/-1->8->15 [12] 9/-1/-1->8->15 [13] 9/-1/-1->8->15 [14] 9/-1/-1->8->15 [15] 9/-1/-1->8->15
+ip-26-0-165-213:9625:10020 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-167-9:1039541:1039937 [4] NCCL INFO Trees [0] 61/-1/-1->60->59 [1] 61/-1/-1->60->59 [2] 61/-1/-1->60->59 [3] 61/-1/-1->60->59 [4] 61/-1/-1->60->52 [5] -1/-1/-1->60->59 [6] 61/-1/-1->60->59 [7] 61/-1/-1->60->59 [8] 61/-1/-1->60->59 [9] 61/-1/-1->60->59 [10] 61/-1/-1->60->59 [11] 61/-1/-1->60->59 [12] 61/28/-1->60->-1 [13] -1/-1/-1->60->59 [14] 61/-1/-1->60->59 [15] 61/-1/-1->60->59
+ip-26-0-162-46:702886:703281 [6] NCCL INFO NVLS Head  2:  2 10 18 26 34 42 50 58
+ip-26-0-162-46:702880:703277 [0] NCCL INFO NVLS Head  7:  7 15 23 31 39 47 55 63
+ip-26-0-162-46:702887:703282 [7] NCCL INFO NVLS Head  5:  5 13 21 29 37 45 53 61
+ip-26-0-162-46:702885:703278 [5] NCCL INFO NVLS Head  1:  1  9 17 25 33 41 49 57
+ip-26-0-162-46:702884:703279 [4] NCCL INFO NVLS Head  0:  0  8 16 24 32 40 48 56
+ip-26-0-162-46:702886:703281 [6] NCCL INFO NVLS Head  3:  3 11 19 27 35 43 51 59
+ip-26-0-162-46:702887:703282 [7] NCCL INFO NVLS Head  6:  6 14 22 30 38 46 54 62
+ip-26-0-162-46:702883:703283 [3] NCCL INFO NVLS Head  0:  0  8 16 24 32 40 48 56
+ip-26-0-162-46:702885:703278 [5] NCCL INFO NVLS Head  2:  2 10 18 26 34 42 50 58
+ip-26-0-162-46:702884:703279 [4] NCCL INFO NVLS Head  1:  1  9 17 25 33 41 49 57
+ip-26-0-167-9:1039540:1039935 [3] NCCL INFO Trees [0] 60/-1/-1->59->58 [1] 60/-1/-1->59->58 [2] 60/-1/-1->59->58 [3] 60/-1/-1->59->51 [4] -1/-1/-1->59->58 [5] 60/-1/-1->59->58 [6] 60/-1/-1->59->58 [7] 60/-1/-1->59->58 [8] 60/-1/-1->59->58 [9] 60/-1/-1->59->58 [10] 60/-1/-1->59->58 [11] 60/27/-1->59->-1 [12] -1/-1/-1->59->58 [13] 60/-1/-1->59->58 [14] 60/-1/-1->59->58 [15] 60/-1/-1->59->58
+ip-26-0-167-9:1039541:1039937 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-167-9:1039540:1039935 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-167-9:1039543:1039933 [6] NCCL INFO Trees [0] 63/-1/-1->62->61 [1] 63/-1/-1->62->61 [2] 63/-1/-1->62->61 [3] 63/-1/-1->62->61 [4] 63/-1/-1->62->61 [5] 63/-1/-1->62->61 [6] 63/-1/-1->62->54 [7] -1/-1/-1->62->61 [8] 63/-1/-1->62->61 [9] 63/-1/-1->62->61 [10] 63/-1/-1->62->61 [11] 63/-1/-1->62->61 [12] 63/-1/-1->62->61 [13] 63/-1/-1->62->61 [14] 63/30/-1->62->-1 [15] -1/-1/-1->62->61
+ip-26-0-167-9:1039543:1039933 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-162-46:702880:703277 [0] NCCL INFO Channel 00/16 :    0   1   2   3   4   5   6   7   8   9  10  11  12  13  14  15  16  17  18  19
+ip-26-0-162-46:702887:703282 [7] NCCL INFO NVLS Head  7:  7 15 23 31 39 47 55 63
+ip-26-0-162-46:702885:703278 [5] NCCL INFO NVLS Head  3:  3 11 19 27 35 43 51 59
+ip-26-0-162-46:702886:703281 [6] NCCL INFO NVLS Head  4:  4 12 20 28 36 44 52 60
+ip-26-0-162-46:702884:703279 [4] NCCL INFO NVLS Head  2:  2 10 18 26 34 42 50 58
+ip-26-0-162-46:702883:703283 [3] NCCL INFO NVLS Head  1:  1  9 17 25 33 41 49 57
+ip-26-0-162-46:702882:703284 [2] NCCL INFO NVLS Head  0:  0  8 16 24 32 40 48 56
+ip-26-0-162-46:702885:703278 [5] NCCL INFO NVLS Head  4:  4 12 20 28 36 44 52 60
+ip-26-0-162-46:702886:703281 [6] NCCL INFO NVLS Head  5:  5 13 21 29 37 45 53 61
+ip-26-0-162-46:702883:703283 [3] NCCL INFO NVLS Head  2:  2 10 18 26 34 42 50 58
+ip-26-0-162-46:702880:703277 [0] NCCL INFO Channel 01/16 :    0   7   6   5   4   3   2   9   8  15  14  13  12  11  10  17  16  23  22  21
+ip-26-0-166-36:104431:104831 [2] NCCL INFO Trees [0] 51/-1/-1->50->49 [1] 51/-1/-1->50->49 [2] 51/42/58->50->34 [3] -1/-1/-1->50->49 [4] 51/-1/-1->50->49 [5] 51/-1/-1->50->49 [6] 51/-1/-1->50->49 [7] 51/-1/-1->50->49 [8] 51/-1/-1->50->49 [9] 51/-1/-1->50->49 [10] 51/-1/-1->50->42 [11] -1/-1/-1->50->49 [12] 51/-1/-1->50->49 [13] 51/-1/-1->50->49 [14] 51/-1/-1->50->49 [15] 51/-1/-1->50->49
+ip-26-0-166-36:104429:104829 [0] NCCL INFO Trees [0] 49/40/56->48->32 [1] -1/-1/-1->48->55 [2] 49/-1/-1->48->55 [3] 49/-1/-1->48->55 [4] 49/-1/-1->48->55 [5] 49/-1/-1->48->55 [6] 49/-1/-1->48->55 [7] 49/-1/-1->48->55 [8] 49/-1/-1->48->40 [9] -1/-1/-1->48->55 [10] 49/-1/-1->48->55 [11] 49/-1/-1->48->55 [12] 49/-1/-1->48->55 [13] 49/-1/-1->48->55 [14] 49/-1/-1->48->55 [15] 49/-1/-1->48->55
+ip-26-0-166-36:104431:104831 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-36:104429:104829 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-244:897245:897638 [5] NCCL INFO Trees [0] 46/-1/-1->45->44 [1] 46/-1/-1->45->44 [2] 46/-1/-1->45->44 [3] 46/-1/-1->45->44 [4] 46/-1/-1->45->44 [5] 46/-1/-1->45->53 [6] -1/-1/-1->45->44 [7] 46/-1/-1->45->44 [8] 46/-1/-1->45->44 [9] 46/-1/-1->45->44 [10] 46/-1/-1->45->44 [11] 46/-1/-1->45->44 [12] 46/-1/-1->45->44 [13] 46/53/37->45->29 [14] -1/-1/-1->45->44 [15] 46/-1/-1->45->44
+ip-26-0-166-244:897245:897638 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-244:897246:897641 [6] NCCL INFO Trees [0] 47/-1/-1->46->45 [1] 47/-1/-1->46->45 [2] 47/-1/-1->46->45 [3] 47/-1/-1->46->45 [4] 47/-1/-1->46->45 [5] 47/-1/-1->46->45 [6] 47/-1/-1->46->54 [7] -1/-1/-1->46->45 [8] 47/-1/-1->46->45 [9] 47/-1/-1->46->45 [10] 47/-1/-1->46->45 [11] 47/-1/-1->46->45 [12] 47/-1/-1->46->45 [13] 47/-1/-1->46->45 [14] 47/54/38->46->30 [15] -1/-1/-1->46->45
+ip-26-0-165-213:9626:10025 [1] NCCL INFO Trees [0] 10/-1/-1->9->8 [1] 10/-1/-1->9->17 [2] -1/-1/-1->9->8 [3] 10/-1/-1->9->8 [4] 10/-1/-1->9->8 [5] 10/-1/-1->9->8 [6] 10/-1/-1->9->8 [7] 10/-1/-1->9->8 [8] 10/-1/-1->9->8 [9] 10/17/1->9->25 [10] -1/-1/-1->9->8 [11] 10/-1/-1->9->8 [12] 10/-1/-1->9->8 [13] 10/-1/-1->9->8 [14] 10/-1/-1->9->8 [15] 10/-1/-1->9->8
+ip-26-0-165-213:9626:10025 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-214:624950:625342 [7] NCCL INFO Trees [0] -1/-1/-1->39->38 [1] 32/-1/-1->39->38 [2] 32/-1/-1->39->38 [3] 32/-1/-1->39->38 [4] 32/-1/-1->39->38 [5] 32/-1/-1->39->38 [6] 32/-1/-1->39->38 [7] 32/23/55->39->7 [8] -1/-1/-1->39->38 [9] 32/-1/-1->39->38 [10] 32/-1/-1->39->38 [11] 32/-1/-1->39->38 [12] 32/-1/-1->39->38 [13] 32/-1/-1->39->38 [14] 32/-1/-1->39->38 [15] 32/-1/-1->39->47
+ip-26-0-166-214:624950:625342 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-214:624949:625344 [6] NCCL INFO Trees [0] 39/-1/-1->38->37 [1] 39/-1/-1->38->37 [2] 39/-1/-1->38->37 [3] 39/-1/-1->38->37 [4] 39/-1/-1->38->37 [5] 39/-1/-1->38->37 [6] 39/22/54->38->6 [7] -1/-1/-1->38->37 [8] 39/-1/-1->38->37 [9] 39/-1/-1->38->37 [10] 39/-1/-1->38->37 [11] 39/-1/-1->38->37 [12] 39/-1/-1->38->37 [13] 39/-1/-1->38->37 [14] 39/-1/-1->38->46 [15] -1/-1/-1->38->37
+ip-26-0-166-214:624949:625344 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-167-9:1039538:1039934 [1] NCCL INFO Trees [0] 58/-1/-1->57->56 [1] 58/-1/-1->57->49 [2] -1/-1/-1->57->56 [3] 58/-1/-1->57->56 [4] 58/-1/-1->57->56 [5] 58/-1/-1->57->56 [6] 58/-1/-1->57->56 [7] 58/-1/-1->57->56 [8] 58/-1/-1->57->56 [9] 58/25/-1->57->-1 [10] -1/-1/-1->57->56 [11] 58/-1/-1->57->56 [12] 58/-1/-1->57->56 [13] 58/-1/-1->57->56 [14] 58/-1/-1->57->56 [15] 58/-1/-1->57->56
+ip-26-0-167-9:1039542:1039936 [5] NCCL INFO Trees [0] 62/-1/-1->61->60 [1] 62/-1/-1->61->60 [2] 62/-1/-1->61->60 [3] 62/-1/-1->61->60 [4] 62/-1/-1->61->60 [5] 62/-1/-1->61->53 [6] -1/-1/-1->61->60 [7] 62/-1/-1->61->60 [8] 62/-1/-1->61->60 [9] 62/-1/-1->61->60 [10] 62/-1/-1->61->60 [11] 62/-1/-1->61->60 [12] 62/-1/-1->61->60 [13] 62/29/-1->61->-1 [14] -1/-1/-1->61->60 [15] 62/-1/-1->61->60
+ip-26-0-167-9:1039538:1039934 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-167-9:1039542:1039936 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-162-46:702884:703279 [4] NCCL INFO NVLS Head  3:  3 11 19 27 35 43 51 59
+ip-26-0-162-46:702885:703278 [5] NCCL INFO NVLS Head  5:  5 13 21 29 37 45 53 61
+ip-26-0-162-46:702882:703284 [2] NCCL INFO NVLS Head  1:  1  9 17 25 33 41 49 57
+ip-26-0-162-46:702886:703281 [6] NCCL INFO NVLS Head  6:  6 14 22 30 38 46 54 62
+ip-26-0-162-46:702883:703283 [3] NCCL INFO NVLS Head  3:  3 11 19 27 35 43 51 59
+ip-26-0-162-46:702886:703281 [6] NCCL INFO NVLS Head  7:  7 15 23 31 39 47 55 63
+ip-26-0-162-46:702884:703279 [4] NCCL INFO NVLS Head  4:  4 12 20 28 36 44 52 60
+ip-26-0-162-46:702885:703278 [5] NCCL INFO NVLS Head  6:  6 14 22 30 38 46 54 62
+ip-26-0-162-46:702882:703284 [2] NCCL INFO NVLS Head  2:  2 10 18 26 34 42 50 58
+ip-26-0-162-46:702880:703277 [0] NCCL INFO Channel 02/16 :    0   7   6   5   4   3  10   9   8  15  14  13  12  11  18  17  16  23  22  21
+ip-26-0-162-46:702885:703278 [5] NCCL INFO NVLS Head  7:  7 15 23 31 39 47 55 63
+ip-26-0-166-125:762822:763222 [7] NCCL INFO Trees [0] -1/-1/-1->23->22 [1] 16/-1/-1->23->22 [2] 16/-1/-1->23->22 [3] 16/-1/-1->23->22 [4] 16/-1/-1->23->22 [5] 16/-1/-1->23->22 [6] 16/-1/-1->23->22 [7] 16/15/31->23->39 [8] -1/-1/-1->23->22 [9] 16/-1/-1->23->22 [10] 16/-1/-1->23->22 [11] 16/-1/-1->23->22 [12] 16/-1/-1->23->22 [13] 16/-1/-1->23->22 [14] 16/-1/-1->23->22 [15] 16/-1/-1->23->15
+ip-26-0-166-125:762822:763222 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-125:762818:763220 [3] NCCL INFO Trees [0] 20/-1/-1->19->18 [1] 20/-1/-1->19->18 [2] 20/-1/-1->19->18 [3] 20/11/27->19->35 [4] -1/-1/-1->19->18 [5] 20/-1/-1->19->18 [6] 20/-1/-1->19->18 [7] 20/-1/-1->19->18 [8] 20/-1/-1->19->18 [9] 20/-1/-1->19->18 [10] 20/-1/-1->19->18 [11] 20/-1/-1->19->11 [12] -1/-1/-1->19->18 [13] 20/-1/-1->19->18 [14] 20/-1/-1->19->18 [15] 20/-1/-1->19->18
+ip-26-0-166-125:762818:763220 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-244:897246:897641 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-244:897247:897642 [7] NCCL INFO Trees [0] -1/-1/-1->47->46 [1] 40/-1/-1->47->46 [2] 40/-1/-1->47->46 [3] 40/-1/-1->47->46 [4] 40/-1/-1->47->46 [5] 40/-1/-1->47->46 [6] 40/-1/-1->47->46 [7] 40/-1/-1->47->55 [8] -1/-1/-1->47->46 [9] 40/-1/-1->47->46 [10] 40/-1/-1->47->46 [11] 40/-1/-1->47->46 [12] 40/-1/-1->47->46 [13] 40/-1/-1->47->46 [14] 40/-1/-1->47->46 [15] 40/55/39->47->31
+ip-26-0-165-213:9631:10022 [6] NCCL INFO Trees [0] 15/-1/-1->14->13 [1] 15/-1/-1->14->13 [2] 15/-1/-1->14->13 [3] 15/-1/-1->14->13 [4] 15/-1/-1->14->13 [5] 15/-1/-1->14->13 [6] 15/-1/-1->14->22 [7] -1/-1/-1->14->13 [8] 15/-1/-1->14->13 [9] 15/-1/-1->14->13 [10] 15/-1/-1->14->13 [11] 15/-1/-1->14->13 [12] 15/-1/-1->14->13 [13] 15/-1/-1->14->13 [14] 15/22/6->14->30 [15] -1/-1/-1->14->13
+ip-26-0-165-213:9631:10022 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-213:9629:10021 [4] NCCL INFO Trees [0] 13/-1/-1->12->11 [1] 13/-1/-1->12->11 [2] 13/-1/-1->12->11 [3] 13/-1/-1->12->11 [4] 13/-1/-1->12->20 [5] -1/-1/-1->12->11 [6] 13/-1/-1->12->11 [7] 13/-1/-1->12->11 [8] 13/-1/-1->12->11 [9] 13/-1/-1->12->11 [10] 13/-1/-1->12->11 [11] 13/-1/-1->12->11 [12] 13/20/4->12->28 [13] -1/-1/-1->12->11 [14] 13/-1/-1->12->11 [15] 13/-1/-1->12->11
+ip-26-0-166-15:127296:127688 [6] NCCL INFO Trees [0] 31/-1/-1->30->29 [1] 31/-1/-1->30->29 [2] 31/-1/-1->30->29 [3] 31/-1/-1->30->29 [4] 31/-1/-1->30->29 [5] 31/-1/-1->30->29 [6] 31/-1/-1->30->22 [7] -1/-1/-1->30->29 [8] 31/-1/-1->30->29 [9] 31/-1/-1->30->29 [10] 31/-1/-1->30->29 [11] 31/-1/-1->30->29 [12] 31/-1/-1->30->29 [13] 31/-1/-1->30->29 [14] 31/46/14->30->62 [15] -1/-1/-1->30->29
+ip-26-0-166-15:127296:127688 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-15:127295:127686 [5] NCCL INFO Trees [0] 30/-1/-1->29->28 [1] 30/-1/-1->29->28 [2] 30/-1/-1->29->28 [3] 30/-1/-1->29->28 [4] 30/-1/-1->29->28 [5] 30/-1/-1->29->21 [6] -1/-1/-1->29->28 [7] 30/-1/-1->29->28 [8] 30/-1/-1->29->28 [9] 30/-1/-1->29->28 [10] 30/-1/-1->29->28 [11] 30/-1/-1->29->28 [12] 30/-1/-1->29->28 [13] 30/45/13->29->61 [14] -1/-1/-1->29->28 [15] 30/-1/-1->29->28
+ip-26-0-167-9:1039539:1039938 [2] NCCL INFO Trees [0] 59/-1/-1->58->57 [1] 59/-1/-1->58->57 [2] 59/-1/-1->58->50 [3] -1/-1/-1->58->57 [4] 59/-1/-1->58->57 [5] 59/-1/-1->58->57 [6] 59/-1/-1->58->57 [7] 59/-1/-1->58->57 [8] 59/-1/-1->58->57 [9] 59/-1/-1->58->57 [10] 59/26/-1->58->-1 [11] -1/-1/-1->58->57 [12] 59/-1/-1->58->57 [13] 59/-1/-1->58->57 [14] 59/-1/-1->58->57 [15] 59/-1/-1->58->57
+ip-26-0-167-9:1039539:1039938 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-162-46:702881:703280 [1] NCCL INFO Trees [0] 2/-1/-1->1->0 [1] 2/33/-1->1->-1 [2] -1/-1/-1->1->0 [3] 2/-1/-1->1->0 [4] 2/-1/-1->1->0 [5] 2/-1/-1->1->0 [6] 2/-1/-1->1->0 [7] 2/-1/-1->1->0 [8] 2/-1/-1->1->0 [9] 2/-1/-1->1->9 [10] -1/-1/-1->1->0 [11] 2/-1/-1->1->0 [12] 2/-1/-1->1->0 [13] 2/-1/-1->1->0 [14] 2/-1/-1->1->0 [15] 2/-1/-1->1->0
+ip-26-0-162-46:702883:703283 [3] NCCL INFO NVLS Head  4:  4 12 20 28 36 44 52 60
+ip-26-0-162-46:702884:703279 [4] NCCL INFO NVLS Head  5:  5 13 21 29 37 45 53 61
+ip-26-0-162-46:702881:703280 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-162-46:702883:703283 [3] NCCL INFO NVLS Head  5:  5 13 21 29 37 45 53 61
+ip-26-0-162-46:702882:703284 [2] NCCL INFO NVLS Head  3:  3 11 19 27 35 43 51 59
+ip-26-0-162-46:702880:703277 [0] NCCL INFO Channel 03/16 :    0   7   6   5   4  11  10   9   8  15  14  13  12  19  18  17  16  23  22  21
+ip-26-0-162-46:702884:703279 [4] NCCL INFO NVLS Head  6:  6 14 22 30 38 46 54 62
+ip-26-0-166-244:897247:897642 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-244:897242:897639 [2] NCCL INFO Trees [0] 43/-1/-1->42->41 [1] 43/-1/-1->42->41 [2] 43/-1/-1->42->50 [3] -1/-1/-1->42->41 [4] 43/-1/-1->42->41 [5] 43/-1/-1->42->41 [6] 43/-1/-1->42->41 [7] 43/-1/-1->42->41 [8] 43/-1/-1->42->41 [9] 43/-1/-1->42->41 [10] 43/50/34->42->26 [11] -1/-1/-1->42->41 [12] 43/-1/-1->42->41 [13] 43/-1/-1->42->41 [14] 43/-1/-1->42->41 [15] 43/-1/-1->42->41
+ip-26-0-166-244:897243:897637 [3] NCCL INFO Trees [0] 44/-1/-1->43->42 [1] 44/-1/-1->43->42 [2] 44/-1/-1->43->42 [3] 44/-1/-1->43->51 [4] -1/-1/-1->43->42 [5] 44/-1/-1->43->42 [6] 44/-1/-1->43->42 [7] 44/-1/-1->43->42 [8] 44/-1/-1->43->42 [9] 44/-1/-1->43->42 [10] 44/-1/-1->43->42 [11] 44/51/35->43->27 [12] -1/-1/-1->43->42 [13] 44/-1/-1->43->42 [14] 44/-1/-1->43->42 [15] 44/-1/-1->43->42
+ip-26-0-166-244:897242:897639 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-244:897243:897637 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-213:9627:10027 [2] NCCL INFO Trees [0] 11/-1/-1->10->9 [1] 11/-1/-1->10->9 [2] 11/-1/-1->10->18 [3] -1/-1/-1->10->9 [4] 11/-1/-1->10->9 [5] 11/-1/-1->10->9 [6] 11/-1/-1->10->9 [7] 11/-1/-1->10->9 [8] 11/-1/-1->10->9 [9] 11/-1/-1->10->9 [10] 11/18/2->10->26 [11] -1/-1/-1->10->9 [12] 11/-1/-1->10->9 [13] 11/-1/-1->10->9 [14] 11/-1/-1->10->9 [15] 11/-1/-1->10->9
+ip-26-0-165-213:9629:10021 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-214:624947:625339 [4] NCCL INFO Trees [0] 37/-1/-1->36->35 [1] 37/-1/-1->36->35 [2] 37/-1/-1->36->35 [3] 37/-1/-1->36->35 [4] 37/20/52->36->4 [5] -1/-1/-1->36->35 [6] 37/-1/-1->36->35 [7] 37/-1/-1->36->35 [8] 37/-1/-1->36->35 [9] 37/-1/-1->36->35 [10] 37/-1/-1->36->35 [11] 37/-1/-1->36->35 [12] 37/-1/-1->36->44 [13] -1/-1/-1->36->35 [14] 37/-1/-1->36->35 [15] 37/-1/-1->36->35
+ip-26-0-166-214:624947:625339 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-214:624948:625343 [5] NCCL INFO Trees [0] 38/-1/-1->37->36 [1] 38/-1/-1->37->36 [2] 38/-1/-1->37->36 [3] 38/-1/-1->37->36 [4] 38/-1/-1->37->36 [5] 38/21/53->37->5 [6] -1/-1/-1->37->36 [7] 38/-1/-1->37->36 [8] 38/-1/-1->37->36 [9] 38/-1/-1->37->36 [10] 38/-1/-1->37->36 [11] 38/-1/-1->37->36 [12] 38/-1/-1->37->36 [13] 38/-1/-1->37->45 [14] -1/-1/-1->37->36 [15] 38/-1/-1->37->36
+ip-26-0-166-214:624948:625343 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-162-46:702883:703283 [3] NCCL INFO NVLS Head  6:  6 14 22 30 38 46 54 62
+ip-26-0-162-46:702887:703282 [7] NCCL INFO Trees [0] -1/-1/-1->7->6 [1] 0/-1/-1->7->6 [2] 0/-1/-1->7->6 [3] 0/-1/-1->7->6 [4] 0/-1/-1->7->6 [5] 0/-1/-1->7->6 [6] 0/-1/-1->7->6 [7] 0/39/-1->7->-1 [8] -1/-1/-1->7->6 [9] 0/-1/-1->7->6 [10] 0/-1/-1->7->6 [11] 0/-1/-1->7->6 [12] 0/-1/-1->7->6 [13] 0/-1/-1->7->6 [14] 0/-1/-1->7->6 [15] 0/-1/-1->7->15
+ip-26-0-162-46:702882:703284 [2] NCCL INFO NVLS Head  4:  4 12 20 28 36 44 52 60
+ip-26-0-162-46:702887:703282 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-162-46:702884:703279 [4] NCCL INFO NVLS Head  7:  7 15 23 31 39 47 55 63
+ip-26-0-162-46:702883:703283 [3] NCCL INFO NVLS Head  7:  7 15 23 31 39 47 55 63
+ip-26-0-162-46:702882:703284 [2] NCCL INFO NVLS Head  5:  5 13 21 29 37 45 53 61
+ip-26-0-162-46:702880:703277 [0] NCCL INFO Channel 04/16 :    0   7   6   5  12  11  10   9   8  15  14  13  20  19  18  17  16  23  22  21
+ip-26-0-166-36:104430:104830 [1] NCCL INFO Trees [0] 50/-1/-1->49->48 [1] 50/41/57->49->33 [2] -1/-1/-1->49->48 [3] 50/-1/-1->49->48 [4] 50/-1/-1->49->48 [5] 50/-1/-1->49->48 [6] 50/-1/-1->49->48 [7] 50/-1/-1->49->48 [8] 50/-1/-1->49->48 [9] 50/-1/-1->49->41 [10] -1/-1/-1->49->48 [11] 50/-1/-1->49->48 [12] 50/-1/-1->49->48 [13] 50/-1/-1->49->48 [14] 50/-1/-1->49->48 [15] 50/-1/-1->49->48
+ip-26-0-166-36:104432:104827 [3] NCCL INFO Trees [0] 52/-1/-1->51->50 [1] 52/-1/-1->51->50 [2] 52/-1/-1->51->50 [3] 52/43/59->51->35 [4] -1/-1/-1->51->50 [5] 52/-1/-1->51->50 [6] 52/-1/-1->51->50 [7] 52/-1/-1->51->50 [8] 52/-1/-1->51->50 [9] 52/-1/-1->51->50 [10] 52/-1/-1->51->50 [11] 52/-1/-1->51->43 [12] -1/-1/-1->51->50 [13] 52/-1/-1->51->50 [14] 52/-1/-1->51->50 [15] 52/-1/-1->51->50
+ip-26-0-166-36:104430:104830 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-36:104432:104827 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-125:762816:763218 [1] NCCL INFO Trees [0] 18/-1/-1->17->16 [1] 18/9/25->17->33 [2] -1/-1/-1->17->16 [3] 18/-1/-1->17->16 [4] 18/-1/-1->17->16 [5] 18/-1/-1->17->16 [6] 18/-1/-1->17->16 [7] 18/-1/-1->17->16 [8] 18/-1/-1->17->16 [9] 18/-1/-1->17->9 [10] -1/-1/-1->17->16 [11] 18/-1/-1->17->16 [12] 18/-1/-1->17->16 [13] 18/-1/-1->17->16 [14] 18/-1/-1->17->16 [15] 18/-1/-1->17->16
+ip-26-0-166-125:762821:763224 [6] NCCL INFO Trees [0] 23/-1/-1->22->21 [1] 23/-1/-1->22->21 [2] 23/-1/-1->22->21 [3] 23/-1/-1->22->21 [4] 23/-1/-1->22->21 [5] 23/-1/-1->22->21 [6] 23/14/30->22->38 [7] -1/-1/-1->22->21 [8] 23/-1/-1->22->21 [9] 23/-1/-1->22->21 [10] 23/-1/-1->22->21 [11] 23/-1/-1->22->21 [12] 23/-1/-1->22->21 [13] 23/-1/-1->22->21 [14] 23/-1/-1->22->14 [15] -1/-1/-1->22->21
+ip-26-0-166-244:897244:897636 [4] NCCL INFO Trees [0] 45/-1/-1->44->43 [1] 45/-1/-1->44->43 [2] 45/-1/-1->44->43 [3] 45/-1/-1->44->43 [4] 45/-1/-1->44->52 [5] -1/-1/-1->44->43 [6] 45/-1/-1->44->43 [7] 45/-1/-1->44->43 [8] 45/-1/-1->44->43 [9] 45/-1/-1->44->43 [10] 45/-1/-1->44->43 [11] 45/-1/-1->44->43 [12] 45/52/36->44->28 [13] -1/-1/-1->44->43 [14] 45/-1/-1->44->43 [15] 45/-1/-1->44->43
+ip-26-0-165-213:9627:10027 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-213:9630:10024 [5] NCCL INFO Trees [0] 14/-1/-1->13->12 [1] 14/-1/-1->13->12 [2] 14/-1/-1->13->12 [3] 14/-1/-1->13->12 [4] 14/-1/-1->13->12 [5] 14/-1/-1->13->21 [6] -1/-1/-1->13->12 [7] 14/-1/-1->13->12 [8] 14/-1/-1->13->12 [9] 14/-1/-1->13->12 [10] 14/-1/-1->13->12 [11] 14/-1/-1->13->12 [12] 14/-1/-1->13->12 [13] 14/21/5->13->29 [14] -1/-1/-1->13->12 [15] 14/-1/-1->13->12
+ip-26-0-165-213:9630:10024 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-213:9628:10023 [3] NCCL INFO Trees [0] 12/-1/-1->11->10 [1] 12/-1/-1->11->10 [2] 12/-1/-1->11->10 [3] 12/-1/-1->11->19 [4] -1/-1/-1->11->10 [5] 12/-1/-1->11->10 [6] 12/-1/-1->11->10 [7] 12/-1/-1->11->10 [8] 12/-1/-1->11->10 [9] 12/-1/-1->11->10 [10] 12/-1/-1->11->10 [11] 12/19/3->11->27 [12] -1/-1/-1->11->10 [13] 12/-1/-1->11->10 [14] 12/-1/-1->11->10 [15] 12/-1/-1->11->10
+ip-26-0-165-213:9628:10023 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-214:624946:625337 [3] NCCL INFO Trees [0] 36/-1/-1->35->34 [1] 36/-1/-1->35->34 [2] 36/-1/-1->35->34 [3] 36/19/51->35->3 [4] -1/-1/-1->35->34 [5] 36/-1/-1->35->34 [6] 36/-1/-1->35->34 [7] 36/-1/-1->35->34 [8] 36/-1/-1->35->34 [9] 36/-1/-1->35->34 [10] 36/-1/-1->35->34 [11] 36/-1/-1->35->43 [12] -1/-1/-1->35->34 [13] 36/-1/-1->35->34 [14] 36/-1/-1->35->34 [15] 36/-1/-1->35->34
+ip-26-0-166-214:624944:625341 [1] NCCL INFO Trees [0] 34/-1/-1->33->32 [1] 34/17/49->33->1 [2] -1/-1/-1->33->32 [3] 34/-1/-1->33->32 [4] 34/-1/-1->33->32 [5] 34/-1/-1->33->32 [6] 34/-1/-1->33->32 [7] 34/-1/-1->33->32 [8] 34/-1/-1->33->32 [9] 34/-1/-1->33->41 [10] -1/-1/-1->33->32 [11] 34/-1/-1->33->32 [12] 34/-1/-1->33->32 [13] 34/-1/-1->33->32 [14] 34/-1/-1->33->32 [15] 34/-1/-1->33->32
+ip-26-0-166-214:624946:625337 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-15:127294:127687 [4] NCCL INFO Trees [0] 29/-1/-1->28->27 [1] 29/-1/-1->28->27 [2] 29/-1/-1->28->27 [3] 29/-1/-1->28->27 [4] 29/-1/-1->28->20 [5] -1/-1/-1->28->27 [6] 29/-1/-1->28->27 [7] 29/-1/-1->28->27 [8] 29/-1/-1->28->27 [9] 29/-1/-1->28->27 [10] 29/-1/-1->28->27 [11] 29/-1/-1->28->27 [12] 29/44/12->28->60 [13] -1/-1/-1->28->27 [14] 29/-1/-1->28->27 [15] 29/-1/-1->28->27
+ip-26-0-166-15:127295:127686 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-15:127294:127687 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-162-46:702882:703284 [2] NCCL INFO NVLS Head  6:  6 14 22 30 38 46 54 62
+ip-26-0-162-46:702882:703284 [2] NCCL INFO NVLS Head  7:  7 15 23 31 39 47 55 63
+ip-26-0-162-46:702880:703277 [0] NCCL INFO Channel 05/16 :    0   4   7   6  13  11  10   9   8  12  15  14  21  19  18  17  16  20  23  22
+ip-26-0-162-46:702886:703281 [6] NCCL INFO Trees [0] 7/-1/-1->6->5 [1] 7/-1/-1->6->5 [2] 7/-1/-1->6->5 [3] 7/-1/-1->6->5 [4] 7/-1/-1->6->5 [5] 7/-1/-1->6->5 [6] 7/38/-1->6->-1 [7] -1/-1/-1->6->5 [8] 7/-1/-1->6->5 [9] 7/-1/-1->6->5 [10] 7/-1/-1->6->5 [11] 7/-1/-1->6->5 [12] 7/-1/-1->6->5 [13] 7/-1/-1->6->5 [14] 7/-1/-1->6->14 [15] -1/-1/-1->6->5
+ip-26-0-162-46:702886:703281 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-162-46:702880:703277 [0] NCCL INFO Channel 06/16 :    0   5   4   7  14  11  10   9   8  13  12  15  22  19  18  17  16  21  20  23
+ip-26-0-166-125:762819:763221 [4] NCCL INFO Trees [0] 21/-1/-1->20->19 [1] 21/-1/-1->20->19 [2] 21/-1/-1->20->19 [3] 21/-1/-1->20->19 [4] 21/12/28->20->36 [5] -1/-1/-1->20->19 [6] 21/-1/-1->20->19 [7] 21/-1/-1->20->19 [8] 21/-1/-1->20->19 [9] 21/-1/-1->20->19 [10] 21/-1/-1->20->19 [11] 21/-1/-1->20->19 [12] 21/-1/-1->20->12 [13] -1/-1/-1->20->19 [14] 21/-1/-1->20->19 [15] 21/-1/-1->20->19
+ip-26-0-166-125:762816:763218 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-125:762821:763224 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-125:762819:763221 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-244:897244:897636 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-244:897240:897640 [0] NCCL INFO Trees [0] 41/-1/-1->40->48 [1] -1/-1/-1->40->47 [2] 41/-1/-1->40->47 [3] 41/-1/-1->40->47 [4] 41/-1/-1->40->47 [5] 41/-1/-1->40->47 [6] 41/-1/-1->40->47 [7] 41/-1/-1->40->47 [8] 41/48/32->40->24 [9] -1/-1/-1->40->47 [10] 41/-1/-1->40->47 [11] 41/-1/-1->40->47 [12] 41/-1/-1->40->47 [13] 41/-1/-1->40->47 [14] 41/-1/-1->40->47 [15] 41/-1/-1->40->47
+ip-26-0-166-244:897240:897640 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-244:897241:897635 [1] NCCL INFO Trees [0] 42/-1/-1->41->40 [1] 42/-1/-1->41->49 [2] -1/-1/-1->41->40 [3] 42/-1/-1->41->40 [4] 42/-1/-1->41->40 [5] 42/-1/-1->41->40 [6] 42/-1/-1->41->40 [7] 42/-1/-1->41->40 [8] 42/-1/-1->41->40 [9] 42/49/33->41->25 [10] -1/-1/-1->41->40 [11] 42/-1/-1->41->40 [12] 42/-1/-1->41->40 [13] 42/-1/-1->41->40 [14] 42/-1/-1->41->40 [15] 42/-1/-1->41->40
+ip-26-0-166-244:897241:897635 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-213:9632:10026 [7] NCCL INFO Trees [0] -1/-1/-1->15->14 [1] 8/-1/-1->15->14 [2] 8/-1/-1->15->14 [3] 8/-1/-1->15->14 [4] 8/-1/-1->15->14 [5] 8/-1/-1->15->14 [6] 8/-1/-1->15->14 [7] 8/-1/-1->15->23 [8] -1/-1/-1->15->14 [9] 8/-1/-1->15->14 [10] 8/-1/-1->15->14 [11] 8/-1/-1->15->14 [12] 8/-1/-1->15->14 [13] 8/-1/-1->15->14 [14] 8/-1/-1->15->14 [15] 8/23/7->15->31
+ip-26-0-165-213:9632:10026 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-214:624945:625340 [2] NCCL INFO Trees [0] 35/-1/-1->34->33 [1] 35/-1/-1->34->33 [2] 35/18/50->34->2 [3] -1/-1/-1->34->33 [4] 35/-1/-1->34->33 [5] 35/-1/-1->34->33 [6] 35/-1/-1->34->33 [7] 35/-1/-1->34->33 [8] 35/-1/-1->34->33 [9] 35/-1/-1->34->33 [10] 35/-1/-1->34->42 [11] -1/-1/-1->34->33 [12] 35/-1/-1->34->33 [13] 35/-1/-1->34->33 [14] 35/-1/-1->34->33 [15] 35/-1/-1->34->33
+ip-26-0-166-214:624944:625341 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-214:624945:625340 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-15:127293:127689 [3] NCCL INFO Trees [0] 28/-1/-1->27->26 [1] 28/-1/-1->27->26 [2] 28/-1/-1->27->26 [3] 28/-1/-1->27->19 [4] -1/-1/-1->27->26 [5] 28/-1/-1->27->26 [6] 28/-1/-1->27->26 [7] 28/-1/-1->27->26 [8] 28/-1/-1->27->26 [9] 28/-1/-1->27->26 [10] 28/-1/-1->27->26 [11] 28/43/11->27->59 [12] -1/-1/-1->27->26 [13] 28/-1/-1->27->26 [14] 28/-1/-1->27->26 [15] 28/-1/-1->27->26
+ip-26-0-166-15:127297:127691 [7] NCCL INFO Trees [0] -1/-1/-1->31->30 [1] 24/-1/-1->31->30 [2] 24/-1/-1->31->30 [3] 24/-1/-1->31->30 [4] 24/-1/-1->31->30 [5] 24/-1/-1->31->30 [6] 24/-1/-1->31->30 [7] 24/-1/-1->31->23 [8] -1/-1/-1->31->30 [9] 24/-1/-1->31->30 [10] 24/-1/-1->31->30 [11] 24/-1/-1->31->30 [12] 24/-1/-1->31->30 [13] 24/-1/-1->31->30 [14] 24/-1/-1->31->30 [15] 24/47/15->31->63
+ip-26-0-166-15:127293:127689 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-15:127297:127691 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-162-46:702885:703278 [5] NCCL INFO Trees [0] 6/-1/-1->5->4 [1] 6/-1/-1->5->4 [2] 6/-1/-1->5->4 [3] 6/-1/-1->5->4 [4] 6/-1/-1->5->4 [5] 6/37/-1->5->-1 [6] -1/-1/-1->5->4 [7] 6/-1/-1->5->4 [8] 6/-1/-1->5->4 [9] 6/-1/-1->5->4 [10] 6/-1/-1->5->4 [11] 6/-1/-1->5->4 [12] 6/-1/-1->5->4 [13] 6/-1/-1->5->13 [14] -1/-1/-1->5->4 [15] 6/-1/-1->5->4
+ip-26-0-162-46:702885:703278 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-162-46:702880:703277 [0] NCCL INFO Channel 07/16 :    0   6   5   4  15  11  10   9   8  14  13  12  23  19  18  17  16  22  21  20
+ip-26-0-162-46:702880:703277 [0] NCCL INFO Channel 08/16 :    0   1   2   3   4   5   6   7   8   9  10  11  12  13  14  15  16  17  18  19
+ip-26-0-162-46:702880:703277 [0] NCCL INFO Channel 09/16 :    0   7   6   5   4   3   2   9   8  15  14  13  12  11  10  17  16  23  22  21
+ip-26-0-162-46:702880:703277 [0] NCCL INFO Channel 10/16 :    0   7   6   5   4   3  10   9   8  15  14  13  12  11  18  17  16  23  22  21
+ip-26-0-166-125:762820:763219 [5] NCCL INFO Trees [0] 22/-1/-1->21->20 [1] 22/-1/-1->21->20 [2] 22/-1/-1->21->20 [3] 22/-1/-1->21->20 [4] 22/-1/-1->21->20 [5] 22/13/29->21->37 [6] -1/-1/-1->21->20 [7] 22/-1/-1->21->20 [8] 22/-1/-1->21->20 [9] 22/-1/-1->21->20 [10] 22/-1/-1->21->20 [11] 22/-1/-1->21->20 [12] 22/-1/-1->21->20 [13] 22/-1/-1->21->13 [14] -1/-1/-1->21->20 [15] 22/-1/-1->21->20
+ip-26-0-166-125:762815:763225 [0] NCCL INFO Trees [0] 17/8/24->16->32 [1] -1/-1/-1->16->23 [2] 17/-1/-1->16->23 [3] 17/-1/-1->16->23 [4] 17/-1/-1->16->23 [5] 17/-1/-1->16->23 [6] 17/-1/-1->16->23 [7] 17/-1/-1->16->23 [8] 17/-1/-1->16->8 [9] -1/-1/-1->16->23 [10] 17/-1/-1->16->23 [11] 17/-1/-1->16->23 [12] 17/-1/-1->16->23 [13] 17/-1/-1->16->23 [14] 17/-1/-1->16->23 [15] 17/-1/-1->16->23
+ip-26-0-166-125:762820:763219 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-125:762815:763225 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-214:624943:625338 [0] NCCL INFO Trees [0] 33/16/48->32->0 [1] -1/-1/-1->32->39 [2] 33/-1/-1->32->39 [3] 33/-1/-1->32->39 [4] 33/-1/-1->32->39 [5] 33/-1/-1->32->39 [6] 33/-1/-1->32->39 [7] 33/-1/-1->32->39 [8] 33/-1/-1->32->40 [9] -1/-1/-1->32->39 [10] 33/-1/-1->32->39 [11] 33/-1/-1->32->39 [12] 33/-1/-1->32->39 [13] 33/-1/-1->32->39 [14] 33/-1/-1->32->39 [15] 33/-1/-1->32->39
+ip-26-0-166-214:624943:625338 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-15:127292:127692 [2] NCCL INFO Trees [0] 27/-1/-1->26->25 [1] 27/-1/-1->26->25 [2] 27/-1/-1->26->18 [3] -1/-1/-1->26->25 [4] 27/-1/-1->26->25 [5] 27/-1/-1->26->25 [6] 27/-1/-1->26->25 [7] 27/-1/-1->26->25 [8] 27/-1/-1->26->25 [9] 27/-1/-1->26->25 [10] 27/42/10->26->58 [11] -1/-1/-1->26->25 [12] 27/-1/-1->26->25 [13] 27/-1/-1->26->25 [14] 27/-1/-1->26->25 [15] 27/-1/-1->26->25
+ip-26-0-166-15:127292:127692 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-15:127290:127685 [0] NCCL INFO Trees [0] 25/-1/-1->24->16 [1] -1/-1/-1->24->31 [2] 25/-1/-1->24->31 [3] 25/-1/-1->24->31 [4] 25/-1/-1->24->31 [5] 25/-1/-1->24->31 [6] 25/-1/-1->24->31 [7] 25/-1/-1->24->31 [8] 25/40/8->24->56 [9] -1/-1/-1->24->31 [10] 25/-1/-1->24->31 [11] 25/-1/-1->24->31 [12] 25/-1/-1->24->31 [13] 25/-1/-1->24->31 [14] 25/-1/-1->24->31 [15] 25/-1/-1->24->31
+ip-26-0-166-15:127290:127685 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-162-46:702884:703279 [4] NCCL INFO Trees [0] 5/-1/-1->4->3 [1] 5/-1/-1->4->3 [2] 5/-1/-1->4->3 [3] 5/-1/-1->4->3 [4] 5/36/-1->4->-1 [5] -1/-1/-1->4->3 [6] 5/-1/-1->4->3 [7] 5/-1/-1->4->3 [8] 5/-1/-1->4->3 [9] 5/-1/-1->4->3 [10] 5/-1/-1->4->3 [11] 5/-1/-1->4->3 [12] 5/-1/-1->4->12 [13] -1/-1/-1->4->3 [14] 5/-1/-1->4->3 [15] 5/-1/-1->4->3
+ip-26-0-162-46:702883:703283 [3] NCCL INFO Trees [0] 4/-1/-1->3->2 [1] 4/-1/-1->3->2 [2] 4/-1/-1->3->2 [3] 4/35/-1->3->-1 [4] -1/-1/-1->3->2 [5] 4/-1/-1->3->2 [6] 4/-1/-1->3->2 [7] 4/-1/-1->3->2 [8] 4/-1/-1->3->2 [9] 4/-1/-1->3->2 [10] 4/-1/-1->3->2 [11] 4/-1/-1->3->11 [12] -1/-1/-1->3->2 [13] 4/-1/-1->3->2 [14] 4/-1/-1->3->2 [15] 4/-1/-1->3->2
+ip-26-0-162-46:702884:703279 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-162-46:702883:703283 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-162-46:702880:703277 [0] NCCL INFO Channel 11/16 :    0   7   6   5   4  11  10   9   8  15  14  13  12  19  18  17  16  23  22  21
+ip-26-0-166-125:762817:763223 [2] NCCL INFO Trees [0] 19/-1/-1->18->17 [1] 19/-1/-1->18->17 [2] 19/10/26->18->34 [3] -1/-1/-1->18->17 [4] 19/-1/-1->18->17 [5] 19/-1/-1->18->17 [6] 19/-1/-1->18->17 [7] 19/-1/-1->18->17 [8] 19/-1/-1->18->17 [9] 19/-1/-1->18->17 [10] 19/-1/-1->18->10 [11] -1/-1/-1->18->17 [12] 19/-1/-1->18->17 [13] 19/-1/-1->18->17 [14] 19/-1/-1->18->17 [15] 19/-1/-1->18->17
+ip-26-0-166-15:127291:127690 [1] NCCL INFO Trees [0] 26/-1/-1->25->24 [1] 26/-1/-1->25->17 [2] -1/-1/-1->25->24 [3] 26/-1/-1->25->24 [4] 26/-1/-1->25->24 [5] 26/-1/-1->25->24 [6] 26/-1/-1->25->24 [7] 26/-1/-1->25->24 [8] 26/-1/-1->25->24 [9] 26/41/9->25->57 [10] -1/-1/-1->25->24 [11] 26/-1/-1->25->24 [12] 26/-1/-1->25->24 [13] 26/-1/-1->25->24 [14] 26/-1/-1->25->24 [15] 26/-1/-1->25->24
+ip-26-0-166-15:127291:127690 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-162-46:702882:703284 [2] NCCL INFO Trees [0] 3/-1/-1->2->1 [1] 3/-1/-1->2->1 [2] 3/34/-1->2->-1 [3] -1/-1/-1->2->1 [4] 3/-1/-1->2->1 [5] 3/-1/-1->2->1 [6] 3/-1/-1->2->1 [7] 3/-1/-1->2->1 [8] 3/-1/-1->2->1 [9] 3/-1/-1->2->1 [10] 3/-1/-1->2->10 [11] -1/-1/-1->2->1 [12] 3/-1/-1->2->1 [13] 3/-1/-1->2->1 [14] 3/-1/-1->2->1 [15] 3/-1/-1->2->1
+ip-26-0-162-46:702880:703277 [0] NCCL INFO Channel 12/16 :    0   7   6   5  12  11  10   9   8  15  14  13  20  19  18  17  16  23  22  21
+ip-26-0-162-46:702882:703284 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-162-46:702880:703277 [0] NCCL INFO Channel 13/16 :    0   4   7   6  13  11  10   9   8  12  15  14  21  19  18  17  16  20  23  22
+ip-26-0-162-46:702880:703277 [0] NCCL INFO Channel 14/16 :    0   5   4   7  14  11  10   9   8  13  12  15  22  19  18  17  16  21  20  23
+ip-26-0-162-46:702880:703277 [0] NCCL INFO Channel 15/16 :    0   6   5   4  15  11  10   9   8  14  13  12  23  19  18  17  16  22  21  20
+ip-26-0-166-125:762817:763223 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-162-46:702880:703277 [0] NCCL INFO Trees [0] 1/32/-1->0->-1 [1] -1/-1/-1->0->7 [2] 1/-1/-1->0->7 [3] 1/-1/-1->0->7 [4] 1/-1/-1->0->7 [5] 1/-1/-1->0->7 [6] 1/-1/-1->0->7 [7] 1/-1/-1->0->7 [8] 1/-1/-1->0->8 [9] -1/-1/-1->0->7 [10] 1/-1/-1->0->7 [11] 1/-1/-1->0->7 [12] 1/-1/-1->0->7 [13] 1/-1/-1->0->7 [14] 1/-1/-1->0->7 [15] 1/-1/-1->0->7
+ip-26-0-162-46:702880:703277 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-167-9:1039541:1039937 [4] NCCL INFO Channel 00/0 : 60[4] -> 61[5] via P2P/IPC
+ip-26-0-167-9:1039543:1039933 [6] NCCL INFO Channel 00/0 : 62[6] -> 63[7] via P2P/IPC
+ip-26-0-166-214:624949:625344 [6] NCCL INFO Channel 00/0 : 38[6] -> 39[7] via P2P/IPC
+ip-26-0-166-214:624947:625339 [4] NCCL INFO Channel 00/0 : 36[4] -> 37[5] via P2P/IPC
+ip-26-0-166-125:762820:763219 [5] NCCL INFO Channel 00/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-166-125:762819:763221 [4] NCCL INFO Channel 00/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-166-125:762821:763224 [6] NCCL INFO Channel 00/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-166-214:624948:625343 [5] NCCL INFO Channel 00/0 : 37[5] -> 38[6] via P2P/IPC
+ip-26-0-166-214:624945:625340 [2] NCCL INFO Channel 00/0 : 34[2] -> 35[3] via P2P/IPC
+ip-26-0-166-125:762818:763220 [3] NCCL INFO Channel 00/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-166-214:624944:625341 [1] NCCL INFO Channel 00/0 : 33[1] -> 34[2] via P2P/IPC
+ip-26-0-166-214:624946:625337 [3] NCCL INFO Channel 00/0 : 35[3] -> 36[4] via P2P/IPC
+ip-26-0-167-9:1039538:1039934 [1] NCCL INFO Channel 00/0 : 57[1] -> 58[2] via P2P/IPC
+ip-26-0-167-9:1039539:1039938 [2] NCCL INFO Channel 00/0 : 58[2] -> 59[3] via P2P/IPC
+ip-26-0-167-9:1039541:1039937 [4] NCCL INFO Channel 08/0 : 60[4] -> 61[5] via P2P/IPC
+ip-26-0-167-9:1039542:1039936 [5] NCCL INFO Channel 00/0 : 61[5] -> 62[6] via P2P/IPC
+ip-26-0-166-244:897245:897638 [5] NCCL INFO Channel 00/0 : 45[5] -> 46[6] via P2P/IPC
+ip-26-0-166-125:762816:763218 [1] NCCL INFO Channel 00/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-167-9:1039540:1039935 [3] NCCL INFO Channel 00/0 : 59[3] -> 60[4] via P2P/IPC
+ip-26-0-167-9:1039543:1039933 [6] NCCL INFO Channel 08/0 : 62[6] -> 63[7] via P2P/IPC
+ip-26-0-166-214:624949:625344 [6] NCCL INFO Channel 08/0 : 38[6] -> 39[7] via P2P/IPC
+ip-26-0-166-244:897243:897637 [3] NCCL INFO Channel 00/0 : 43[3] -> 44[4] via P2P/IPC
+ip-26-0-166-15:127296:127688 [6] NCCL INFO Channel 00/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-162-46:702885:703278 [5] NCCL INFO Channel 00/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-166-244:897245:897638 [5] NCCL INFO Channel 08/0 : 45[5] -> 46[6] via P2P/IPC
+ip-26-0-166-244:897246:897641 [6] NCCL INFO Channel 00/0 : 46[6] -> 47[7] via P2P/IPC
+ip-26-0-166-15:127292:127692 [2] NCCL INFO Channel 00/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-166-125:762820:763219 [5] NCCL INFO Channel 08/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-166-214:624947:625339 [4] NCCL INFO Channel 08/0 : 36[4] -> 37[5] via P2P/IPC
+ip-26-0-162-46:702884:703279 [4] NCCL INFO Channel 00/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-162-46:702881:703280 [1] NCCL INFO Channel 00/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-166-125:762817:763223 [2] NCCL INFO Channel 00/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-166-214:624948:625343 [5] NCCL INFO Channel 08/0 : 37[5] -> 38[6] via P2P/IPC
+ip-26-0-166-214:624945:625340 [2] NCCL INFO Channel 08/0 : 34[2] -> 35[3] via P2P/IPC
+ip-26-0-166-244:897243:897637 [3] NCCL INFO Channel 08/0 : 43[3] -> 44[4] via P2P/IPC
+ip-26-0-167-9:1039538:1039934 [1] NCCL INFO Channel 08/0 : 57[1] -> 58[2] via P2P/IPC
+ip-26-0-162-46:702882:703284 [2] NCCL INFO Channel 00/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-125:762819:763221 [4] NCCL INFO Channel 08/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-166-125:762821:763224 [6] NCCL INFO Channel 08/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-166-214:624944:625341 [1] NCCL INFO Channel 08/0 : 33[1] -> 34[2] via P2P/IPC
+ip-26-0-166-125:762818:763220 [3] NCCL INFO Channel 08/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-166-214:624946:625337 [3] NCCL INFO Channel 08/0 : 35[3] -> 36[4] via P2P/IPC
+ip-26-0-167-9:1039539:1039938 [2] NCCL INFO Channel 08/0 : 58[2] -> 59[3] via P2P/IPC
+ip-26-0-167-9:1039542:1039936 [5] NCCL INFO Channel 08/0 : 61[5] -> 62[6] via P2P/IPC
+ip-26-0-166-244:897244:897636 [4] NCCL INFO Channel 00/0 : 44[4] -> 45[5] via P2P/IPC
+ip-26-0-162-46:702883:703283 [3] NCCL INFO Channel 00/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-162-46:702886:703281 [6] NCCL INFO Channel 00/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-125:762816:763218 [1] NCCL INFO Channel 08/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-166-244:897246:897641 [6] NCCL INFO Channel 08/0 : 46[6] -> 47[7] via P2P/IPC
+ip-26-0-167-9:1039540:1039935 [3] NCCL INFO Channel 08/0 : 59[3] -> 60[4] via P2P/IPC
+ip-26-0-166-244:897244:897636 [4] NCCL INFO Channel 08/0 : 44[4] -> 45[5] via P2P/IPC
+ip-26-0-162-46:702885:703278 [5] NCCL INFO Channel 08/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-166-125:762817:763223 [2] NCCL INFO Channel 08/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-166-244:897241:897635 [1] NCCL INFO Channel 00/0 : 41[1] -> 42[2] via P2P/IPC
+ip-26-0-162-46:702884:703279 [4] NCCL INFO Channel 08/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-162-46:702881:703280 [1] NCCL INFO Channel 08/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-167-9:1039541:1039937 [4] NCCL INFO Channel 05/0 : 60[4] -> 63[7] via P2P/IPC
+ip-26-0-166-125:762819:763221 [4] NCCL INFO Channel 05/0 : 20[4] -> 23[7] via P2P/IPC
+ip-26-0-166-214:624947:625339 [4] NCCL INFO Channel 05/0 : 36[4] -> 39[7] via P2P/IPC
+ip-26-0-162-46:702882:703284 [2] NCCL INFO Channel 08/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-162-46:702886:703281 [6] NCCL INFO Channel 08/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-244:897242:897639 [2] NCCL INFO Channel 00/0 : 42[2] -> 43[3] via P2P/IPC
+ip-26-0-162-46:702883:703283 [3] NCCL INFO Channel 08/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-15:127293:127689 [3] NCCL INFO Channel 00/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-166-244:897241:897635 [1] NCCL INFO Channel 08/0 : 41[1] -> 42[2] via P2P/IPC
+ip-26-0-166-244:897244:897636 [4] NCCL INFO Channel 05/0 : 44[4] -> 47[7] via P2P/IPC
+ip-26-0-166-214:624947:625339 [4] NCCL INFO Channel 06/0 : 36[4] -> 39[7] via P2P/IPC
+ip-26-0-166-15:127291:127690 [1] NCCL INFO Channel 00/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-166-125:762819:763221 [4] NCCL INFO Channel 06/0 : 20[4] -> 23[7] via P2P/IPC
+ip-26-0-167-9:1039541:1039937 [4] NCCL INFO Channel 06/0 : 60[4] -> 63[7] via P2P/IPC
+ip-26-0-166-15:127295:127686 [5] NCCL INFO Channel 00/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-166-15:127296:127688 [6] NCCL INFO Channel 08/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-166-244:897242:897639 [2] NCCL INFO Channel 08/0 : 42[2] -> 43[3] via P2P/IPC
+ip-26-0-166-214:624947:625339 [4] NCCL INFO Channel 13/0 : 36[4] -> 39[7] via P2P/IPC
+ip-26-0-166-125:762819:763221 [4] NCCL INFO Channel 13/0 : 20[4] -> 23[7] via P2P/IPC
+ip-26-0-167-9:1039541:1039937 [4] NCCL INFO Channel 13/0 : 60[4] -> 63[7] via P2P/IPC
+ip-26-0-166-125:762819:763221 [4] NCCL INFO Channel 14/0 : 20[4] -> 23[7] via P2P/IPC
+ip-26-0-166-214:624947:625339 [4] NCCL INFO Channel 14/0 : 36[4] -> 39[7] via P2P/IPC
+ip-26-0-166-244:897244:897636 [4] NCCL INFO Channel 06/0 : 44[4] -> 47[7] via P2P/IPC
+ip-26-0-167-9:1039541:1039937 [4] NCCL INFO Channel 14/0 : 60[4] -> 63[7] via P2P/IPC
+ip-26-0-162-46:702884:703279 [4] NCCL INFO Channel 05/0 : 4[4] -> 7[7] via P2P/IPC
+ip-26-0-166-244:897244:897636 [4] NCCL INFO Channel 13/0 : 44[4] -> 47[7] via P2P/IPC
+ip-26-0-166-15:127294:127687 [4] NCCL INFO Channel 00/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-166-15:127293:127689 [3] NCCL INFO Channel 08/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-166-15:127291:127690 [1] NCCL INFO Channel 08/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-166-15:127292:127692 [2] NCCL INFO Channel 08/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-166-244:897244:897636 [4] NCCL INFO Channel 14/0 : 44[4] -> 47[7] via P2P/IPC
+ip-26-0-166-15:127294:127687 [4] NCCL INFO Channel 08/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-162-46:702884:703279 [4] NCCL INFO Channel 06/0 : 4[4] -> 7[7] via P2P/IPC
+ip-26-0-166-15:127295:127686 [5] NCCL INFO Channel 08/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-162-46:702884:703279 [4] NCCL INFO Channel 13/0 : 4[4] -> 7[7] via P2P/IPC
+ip-26-0-162-46:702884:703279 [4] NCCL INFO Channel 14/0 : 4[4] -> 7[7] via P2P/IPC
+ip-26-0-166-15:127294:127687 [4] NCCL INFO Channel 05/0 : 28[4] -> 31[7] via P2P/IPC
+ip-26-0-166-15:127294:127687 [4] NCCL INFO Channel 06/0 : 28[4] -> 31[7] via P2P/IPC
+ip-26-0-166-15:127294:127687 [4] NCCL INFO Channel 13/0 : 28[4] -> 31[7] via P2P/IPC
+ip-26-0-166-15:127294:127687 [4] NCCL INFO Channel 14/0 : 28[4] -> 31[7] via P2P/IPC
+ip-26-0-165-213:9626:10025 [1] NCCL INFO Channel 00/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-165-213:9630:10024 [5] NCCL INFO Channel 00/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-165-213:9629:10021 [4] NCCL INFO Channel 00/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-165-213:9631:10022 [6] NCCL INFO Channel 00/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-165-213:9628:10023 [3] NCCL INFO Channel 00/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-165-213:9627:10027 [2] NCCL INFO Channel 00/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-165-213:9626:10025 [1] NCCL INFO Channel 08/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-165-213:9630:10024 [5] NCCL INFO Channel 08/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-165-213:9631:10022 [6] NCCL INFO Channel 08/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-165-213:9629:10021 [4] NCCL INFO Channel 08/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-165-213:9627:10027 [2] NCCL INFO Channel 08/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-165-213:9628:10023 [3] NCCL INFO Channel 08/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-165-213:9629:10021 [4] NCCL INFO Channel 05/0 : 12[4] -> 15[7] via P2P/IPC
+ip-26-0-165-213:9629:10021 [4] NCCL INFO Channel 06/0 : 12[4] -> 15[7] via P2P/IPC
+ip-26-0-165-213:9629:10021 [4] NCCL INFO Channel 13/0 : 12[4] -> 15[7] via P2P/IPC
+ip-26-0-165-213:9629:10021 [4] NCCL INFO Channel 14/0 : 12[4] -> 15[7] via P2P/IPC
+ip-26-0-166-36:104433:104825 [4] NCCL INFO Channel 00/0 : 52[4] -> 53[5] via P2P/IPC
+ip-26-0-166-36:104433:104825 [4] NCCL INFO Channel 08/0 : 52[4] -> 53[5] via P2P/IPC
+ip-26-0-166-36:104435:104826 [6] NCCL INFO Channel 00/0 : 54[6] -> 55[7] via P2P/IPC
+ip-26-0-166-36:104434:104828 [5] NCCL INFO Channel 00/0 : 53[5] -> 54[6] via P2P/IPC
+ip-26-0-166-36:104431:104831 [2] NCCL INFO Channel 00/0 : 50[2] -> 51[3] via P2P/IPC
+ip-26-0-166-36:104432:104827 [3] NCCL INFO Channel 00/0 : 51[3] -> 52[4] via P2P/IPC
+ip-26-0-166-36:104430:104830 [1] NCCL INFO Channel 00/0 : 49[1] -> 50[2] via P2P/IPC
+ip-26-0-166-36:104435:104826 [6] NCCL INFO Channel 08/0 : 54[6] -> 55[7] via P2P/IPC
+ip-26-0-166-36:104434:104828 [5] NCCL INFO Channel 08/0 : 53[5] -> 54[6] via P2P/IPC
+ip-26-0-166-36:104431:104831 [2] NCCL INFO Channel 08/0 : 50[2] -> 51[3] via P2P/IPC
+ip-26-0-166-36:104432:104827 [3] NCCL INFO Channel 08/0 : 51[3] -> 52[4] via P2P/IPC
+ip-26-0-166-36:104430:104830 [1] NCCL INFO Channel 08/0 : 49[1] -> 50[2] via P2P/IPC
+ip-26-0-166-36:104433:104825 [4] NCCL INFO Channel 05/0 : 52[4] -> 55[7] via P2P/IPC
+ip-26-0-166-36:104433:104825 [4] NCCL INFO Channel 06/0 : 52[4] -> 55[7] via P2P/IPC
+ip-26-0-166-36:104433:104825 [4] NCCL INFO Channel 13/0 : 52[4] -> 55[7] via P2P/IPC
+ip-26-0-166-36:104433:104825 [4] NCCL INFO Channel 14/0 : 52[4] -> 55[7] via P2P/IPC
+ip-26-0-166-214:624943:625338 [0] NCCL INFO Channel 00/0 : 31[7] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624943:625338 [0] NCCL INFO Channel 08/0 : 31[7] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624943:625338 [0] NCCL INFO Channel 00/0 : 32[0] -> 33[1] via P2P/IPC
+ip-26-0-166-214:624950:625342 [7] NCCL INFO Channel 00/0 : 39[7] -> 40[0] [send] via NET/Libfabric/0(32)/GDRDMA
+ip-26-0-165-213:9625:10020 [0] NCCL INFO Channel 00/0 : 7[7] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9625:10020 [0] NCCL INFO Channel 08/0 : 7[7] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9625:10020 [0] NCCL INFO Channel 00/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-165-213:9632:10026 [7] NCCL INFO Channel 00/0 : 15[7] -> 16[0] [send] via NET/Libfabric/0(8)/GDRDMA
+ip-26-0-166-214:624943:625338 [0] NCCL INFO Channel 08/0 : 32[0] -> 33[1] via P2P/IPC
+ip-26-0-166-214:624950:625342 [7] NCCL INFO Channel 08/0 : 39[7] -> 40[0] [send] via NET/Libfabric/0(32)/GDRDMA
+ip-26-0-165-213:9625:10020 [0] NCCL INFO Channel 08/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-165-213:9632:10026 [7] NCCL INFO Channel 08/0 : 15[7] -> 16[0] [send] via NET/Libfabric/0(8)/GDRDMA
+ip-26-0-166-244:897240:897640 [0] NCCL INFO Channel 00/0 : 39[7] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897240:897640 [0] NCCL INFO Channel 08/0 : 39[7] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897240:897640 [0] NCCL INFO Channel 00/0 : 40[0] -> 41[1] via P2P/IPC
+ip-26-0-166-244:897247:897642 [7] NCCL INFO Channel 00/0 : 47[7] -> 48[0] [send] via NET/Libfabric/0(40)/GDRDMA
+ip-26-0-167-9:1039537:1039932 [0] NCCL INFO Channel 00/0 : 55[7] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-167-9:1039537:1039932 [0] NCCL INFO Channel 08/0 : 55[7] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-167-9:1039537:1039932 [0] NCCL INFO Channel 00/0 : 56[0] -> 57[1] via P2P/IPC
+ip-26-0-167-9:1039544:1039931 [7] NCCL INFO Channel 00/0 : 63[7] -> 0[0] [send] via NET/Libfabric/0(56)/GDRDMA
+ip-26-0-166-244:897240:897640 [0] NCCL INFO Channel 08/0 : 40[0] -> 41[1] via P2P/IPC
+ip-26-0-166-244:897247:897642 [7] NCCL INFO Channel 08/0 : 47[7] -> 48[0] [send] via NET/Libfabric/0(40)/GDRDMA
+ip-26-0-167-9:1039537:1039932 [0] NCCL INFO Channel 08/0 : 56[0] -> 57[1] via P2P/IPC
+ip-26-0-167-9:1039544:1039931 [7] NCCL INFO Channel 08/0 : 63[7] -> 0[0] [send] via NET/Libfabric/0(56)/GDRDMA
+ip-26-0-166-125:762815:763225 [0] NCCL INFO Channel 00/0 : 15[7] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897240:897640 [0] NCCL INFO Channel 05/0 : 40[0] -> 44[4] via P2P/IPC
+ip-26-0-166-125:762815:763225 [0] NCCL INFO Channel 08/0 : 15[7] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762815:763225 [0] NCCL INFO Channel 00/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-166-125:762822:763222 [7] NCCL INFO Channel 00/0 : 23[7] -> 24[0] [send] via NET/Libfabric/0(16)/GDRDMA
+ip-26-0-166-15:127290:127685 [0] NCCL INFO Channel 00/0 : 23[7] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-167-9:1039537:1039932 [0] NCCL INFO Channel 05/0 : 56[0] -> 60[4] via P2P/IPC
+ip-26-0-166-15:127290:127685 [0] NCCL INFO Channel 08/0 : 23[7] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897240:897640 [0] NCCL INFO Channel 13/0 : 40[0] -> 44[4] via P2P/IPC
+ip-26-0-166-15:127290:127685 [0] NCCL INFO Channel 00/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-166-15:127297:127691 [7] NCCL INFO Channel 00/0 : 31[7] -> 32[0] [send] via NET/Libfabric/0(24)/GDRDMA
+ip-26-0-166-125:762815:763225 [0] NCCL INFO Channel 08/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-167-9:1039537:1039932 [0] NCCL INFO Channel 13/0 : 56[0] -> 60[4] via P2P/IPC
+ip-26-0-166-125:762822:763222 [7] NCCL INFO Channel 08/0 : 23[7] -> 24[0] [send] via NET/Libfabric/0(16)/GDRDMA
+ip-26-0-162-46:702880:703277 [0] NCCL INFO Channel 00/0 : 63[7] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127290:127685 [0] NCCL INFO Channel 08/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-162-46:702880:703277 [0] NCCL INFO Channel 08/0 : 63[7] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-46:702880:703277 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-15:127297:127691 [7] NCCL INFO Channel 08/0 : 31[7] -> 32[0] [send] via NET/Libfabric/0(24)/GDRDMA
+ip-26-0-162-46:702887:703282 [7] NCCL INFO Channel 00/0 : 7[7] -> 8[0] [send] via NET/Libfabric/0(0)/GDRDMA
+ip-26-0-166-125:762815:763225 [0] NCCL INFO Channel 05/0 : 16[0] -> 20[4] via P2P/IPC
+ip-26-0-166-214:624943:625338 [0] NCCL INFO Channel 05/0 : 32[0] -> 36[4] via P2P/IPC
+ip-26-0-166-125:762815:763225 [0] NCCL INFO Channel 13/0 : 16[0] -> 20[4] via P2P/IPC
+ip-26-0-166-214:624945:625340 [2] NCCL INFO Channel 02/0 : 27[3] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624946:625337 [3] NCCL INFO Channel 03/0 : 28[4] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624945:625340 [2] NCCL INFO Channel 10/0 : 27[3] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624946:625337 [3] NCCL INFO Channel 11/0 : 28[4] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-46:702880:703277 [0] NCCL INFO Channel 08/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-15:127290:127685 [0] NCCL INFO Channel 05/0 : 24[0] -> 28[4] via P2P/IPC
+ip-26-0-162-46:702887:703282 [7] NCCL INFO Channel 08/0 : 7[7] -> 8[0] [send] via NET/Libfabric/0(0)/GDRDMA
+ip-26-0-166-214:624946:625337 [3] NCCL INFO Channel 02/0 : 35[3] -> 42[2] [send] via NET/Libfabric/2(34)/GDRDMA
+ip-26-0-166-214:624943:625338 [0] NCCL INFO Channel 13/0 : 32[0] -> 36[4] via P2P/IPC
+ip-26-0-166-214:624946:625337 [3] NCCL INFO Channel 10/0 : 35[3] -> 42[2] [send] via NET/Libfabric/2(34)/GDRDMA
+ip-26-0-166-244:897240:897640 [0] NCCL INFO Channel 06/0 : 40[0] -> 45[5] via P2P/IPC
+ip-26-0-166-15:127290:127685 [0] NCCL INFO Channel 13/0 : 24[0] -> 28[4] via P2P/IPC
+ip-26-0-165-213:9625:10020 [0] NCCL INFO Channel 05/0 : 8[0] -> 12[4] via P2P/IPC
+ip-26-0-166-244:897240:897640 [0] NCCL INFO Channel 14/0 : 40[0] -> 45[5] via P2P/IPC
+ip-26-0-166-214:624943:625338 [0] NCCL INFO Channel 06/0 : 32[0] -> 37[5] via P2P/IPC
+ip-26-0-162-46:702880:703277 [0] NCCL INFO Channel 05/0 : 0[0] -> 4[4] via P2P/IPC
+ip-26-0-165-213:9625:10020 [0] NCCL INFO Channel 13/0 : 8[0] -> 12[4] via P2P/IPC
+ip-26-0-162-46:702882:703284 [2] NCCL INFO Channel 02/0 : 59[3] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-46:702882:703284 [2] NCCL INFO Channel 10/0 : 59[3] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897240:897640 [0] NCCL INFO Channel 07/0 : 40[0] -> 46[6] via P2P/IPC
+ip-26-0-166-214:624943:625338 [0] NCCL INFO Channel 14/0 : 32[0] -> 37[5] via P2P/IPC
+ip-26-0-166-36:104429:104829 [0] NCCL INFO Channel 00/0 : 47[7] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104429:104829 [0] NCCL INFO Channel 08/0 : 47[7] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104429:104829 [0] NCCL INFO Channel 00/0 : 48[0] -> 49[1] via P2P/IPC
+ip-26-0-166-244:897243:897637 [3] NCCL INFO Channel 03/0 : 36[4] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104436:104824 [7] NCCL INFO Channel 00/0 : 55[7] -> 56[0] [send] via NET/Libfabric/0(48)/GDRDMA
+ip-26-0-166-36:104429:104829 [0] NCCL INFO Channel 08/0 : 48[0] -> 49[1] via P2P/IPC
+ip-26-0-162-46:702880:703277 [0] NCCL INFO Channel 13/0 : 0[0] -> 4[4] via P2P/IPC
+ip-26-0-166-36:104436:104824 [7] NCCL INFO Channel 08/0 : 55[7] -> 56[0] [send] via NET/Libfabric/0(48)/GDRDMA
+ip-26-0-166-36:104429:104829 [0] NCCL INFO Channel 05/0 : 48[0] -> 52[4] via P2P/IPC
+ip-26-0-166-36:104429:104829 [0] NCCL INFO Channel 13/0 : 48[0] -> 52[4] via P2P/IPC
+ip-26-0-166-36:104431:104831 [2] NCCL INFO Channel 02/0 : 43[3] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104431:104831 [2] NCCL INFO Channel 10/0 : 43[3] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104431:104831 [2] NCCL INFO Channel 01/0 : 50[2] -> 57[1] [send] via NET/Libfabric/1(49)/GDRDMA
+ip-26-0-166-244:897243:897637 [3] NCCL INFO Channel 11/0 : 36[4] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104431:104831 [2] NCCL INFO Channel 09/0 : 50[2] -> 57[1] [send] via NET/Libfabric/1(49)/GDRDMA
+ip-26-0-166-36:104432:104827 [3] NCCL INFO Channel 03/0 : 44[4] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897240:897640 [0] NCCL INFO Channel 15/0 : 40[0] -> 46[6] via P2P/IPC
+ip-26-0-166-36:104432:104827 [3] NCCL INFO Channel 11/0 : 44[4] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624943:625338 [0] NCCL INFO Channel 07/0 : 32[0] -> 38[6] via P2P/IPC
+ip-26-0-165-213:9625:10020 [0] NCCL INFO Channel 06/0 : 8[0] -> 13[5] via P2P/IPC
+ip-26-0-167-9:1039540:1039935 [3] NCCL INFO Channel 03/0 : 52[4] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762818:763220 [3] NCCL INFO Channel 03/0 : 12[4] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104432:104827 [3] NCCL INFO Channel 02/0 : 51[3] -> 58[2] [send] via NET/Libfabric/2(50)/GDRDMA
+ip-26-0-166-244:897242:897639 [2] NCCL INFO Channel 02/0 : 35[3] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897240:897640 [0] NCCL INFO Channel 01/0 : 40[0] -> 47[7] via P2P/IPC
+ip-26-0-167-9:1039540:1039935 [3] NCCL INFO Channel 11/0 : 52[4] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104432:104827 [3] NCCL INFO Channel 10/0 : 51[3] -> 58[2] [send] via NET/Libfabric/2(50)/GDRDMA
+ip-26-0-166-125:762818:763220 [3] NCCL INFO Channel 11/0 : 12[4] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897242:897639 [2] NCCL INFO Channel 10/0 : 35[3] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624943:625338 [0] NCCL INFO Channel 15/0 : 32[0] -> 38[6] via P2P/IPC
+ip-26-0-167-9:1039539:1039938 [2] NCCL INFO Channel 02/0 : 51[3] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9625:10020 [0] NCCL INFO Channel 14/0 : 8[0] -> 13[5] via P2P/IPC
+ip-26-0-166-244:897243:897637 [3] NCCL INFO Channel 02/0 : 43[3] -> 50[2] [send] via NET/Libfabric/2(42)/GDRDMA
+ip-26-0-166-244:897243:897637 [3] NCCL INFO Channel 10/0 : 43[3] -> 50[2] [send] via NET/Libfabric/2(42)/GDRDMA
+ip-26-0-167-9:1039539:1039938 [2] NCCL INFO Channel 10/0 : 51[3] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897240:897640 [0] NCCL INFO Channel 02/0 : 40[0] -> 47[7] via P2P/IPC
+ip-26-0-167-9:1039540:1039935 [3] NCCL INFO Channel 02/0 : 59[3] -> 2[2] [send] via NET/Libfabric/2(58)/GDRDMA
+ip-26-0-166-214:624943:625338 [0] NCCL INFO Channel 01/0 : 32[0] -> 39[7] via P2P/IPC
+ip-26-0-167-9:1039540:1039935 [3] NCCL INFO Channel 10/0 : 59[3] -> 2[2] [send] via NET/Libfabric/2(58)/GDRDMA
+ip-26-0-165-213:9625:10020 [0] NCCL INFO Channel 07/0 : 8[0] -> 14[6] via P2P/IPC
+ip-26-0-166-244:897240:897640 [0] NCCL INFO Channel 03/0 : 40[0] -> 47[7] via P2P/IPC
+ip-26-0-166-125:762817:763223 [2] NCCL INFO Channel 02/0 : 11[3] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624943:625338 [0] NCCL INFO Channel 02/0 : 32[0] -> 39[7] via P2P/IPC
+ip-26-0-166-125:762815:763225 [0] NCCL INFO Channel 06/0 : 16[0] -> 21[5] via P2P/IPC
+ip-26-0-162-46:702882:703284 [2] NCCL INFO Channel 01/0 : 2[2] -> 9[1] [send] via NET/Libfabric/1(1)/GDRDMA
+ip-26-0-166-125:762817:763223 [2] NCCL INFO Channel 10/0 : 11[3] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897240:897640 [0] NCCL INFO Channel 04/0 : 40[0] -> 47[7] via P2P/IPC
+ip-26-0-166-125:762818:763220 [3] NCCL INFO Channel 02/0 : 19[3] -> 26[2] [send] via NET/Libfabric/2(18)/GDRDMA
+ip-26-0-162-46:702882:703284 [2] NCCL INFO Channel 09/0 : 2[2] -> 9[1] [send] via NET/Libfabric/1(1)/GDRDMA
+ip-26-0-165-213:9625:10020 [0] NCCL INFO Channel 15/0 : 8[0] -> 14[6] via P2P/IPC
+ip-26-0-166-214:624943:625338 [0] NCCL INFO Channel 03/0 : 32[0] -> 39[7] via P2P/IPC
+ip-26-0-162-46:702883:703283 [3] NCCL INFO Channel 03/0 : 60[4] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762818:763220 [3] NCCL INFO Channel 10/0 : 19[3] -> 26[2] [send] via NET/Libfabric/2(18)/GDRDMA
+ip-26-0-167-9:1039537:1039932 [0] NCCL INFO Channel 06/0 : 56[0] -> 61[5] via P2P/IPC
+ip-26-0-162-46:702883:703283 [3] NCCL INFO Channel 11/0 : 60[4] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762815:763225 [0] NCCL INFO Channel 14/0 : 16[0] -> 21[5] via P2P/IPC
+ip-26-0-166-244:897240:897640 [0] NCCL INFO Channel 09/0 : 40[0] -> 47[7] via P2P/IPC
+ip-26-0-166-214:624943:625338 [0] NCCL INFO Channel 04/0 : 32[0] -> 39[7] via P2P/IPC
+ip-26-0-166-15:127293:127689 [3] NCCL INFO Channel 03/0 : 20[4] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897240:897640 [0] NCCL INFO Channel 10/0 : 40[0] -> 47[7] via P2P/IPC
+ip-26-0-162-46:702883:703283 [3] NCCL INFO Channel 02/0 : 3[3] -> 10[2] [send] via NET/Libfabric/2(2)/GDRDMA
+ip-26-0-165-213:9625:10020 [0] NCCL INFO Channel 01/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-166-214:624943:625338 [0] NCCL INFO Channel 09/0 : 32[0] -> 39[7] via P2P/IPC
+ip-26-0-166-15:127293:127689 [3] NCCL INFO Channel 11/0 : 20[4] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-167-9:1039537:1039932 [0] NCCL INFO Channel 14/0 : 56[0] -> 61[5] via P2P/IPC
+ip-26-0-162-46:702883:703283 [3] NCCL INFO Channel 10/0 : 3[3] -> 10[2] [send] via NET/Libfabric/2(2)/GDRDMA
+ip-26-0-166-125:762815:763225 [0] NCCL INFO Channel 07/0 : 16[0] -> 22[6] via P2P/IPC
+ip-26-0-166-244:897240:897640 [0] NCCL INFO Channel 11/0 : 40[0] -> 47[7] via P2P/IPC
+ip-26-0-165-213:9625:10020 [0] NCCL INFO Channel 02/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-166-214:624943:625338 [0] NCCL INFO Channel 10/0 : 32[0] -> 39[7] via P2P/IPC
+ip-26-0-166-125:762815:763225 [0] NCCL INFO Channel 15/0 : 16[0] -> 22[6] via P2P/IPC
+ip-26-0-166-244:897240:897640 [0] NCCL INFO Channel 12/0 : 40[0] -> 47[7] via P2P/IPC
+ip-26-0-165-213:9625:10020 [0] NCCL INFO Channel 03/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-167-9:1039537:1039932 [0] NCCL INFO Channel 07/0 : 56[0] -> 62[6] via P2P/IPC
+ip-26-0-166-214:624943:625338 [0] NCCL INFO Channel 11/0 : 32[0] -> 39[7] via P2P/IPC
+ip-26-0-166-244:897247:897642 [7] NCCL INFO Channel 06/0 : 47[7] -> 54[6] [send] via NET/Libfabric/6(46)/GDRDMA
+ip-26-0-166-214:624950:625342 [7] NCCL INFO Channel 06/0 : 39[7] -> 46[6] [send] via NET/Libfabric/6(38)/GDRDMA
+ip-26-0-166-125:762815:763225 [0] NCCL INFO Channel 01/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-165-213:9625:10020 [0] NCCL INFO Channel 04/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-166-214:624943:625338 [0] NCCL INFO Channel 12/0 : 32[0] -> 39[7] via P2P/IPC
+ip-26-0-167-9:1039537:1039932 [0] NCCL INFO Channel 15/0 : 56[0] -> 62[6] via P2P/IPC
+ip-26-0-166-125:762815:763225 [0] NCCL INFO Channel 02/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-165-213:9625:10020 [0] NCCL INFO Channel 09/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-166-125:762815:763225 [0] NCCL INFO Channel 03/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-165-213:9625:10020 [0] NCCL INFO Channel 10/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-167-9:1039537:1039932 [0] NCCL INFO Channel 01/0 : 56[0] -> 63[7] via P2P/IPC
+ip-26-0-166-125:762815:763225 [0] NCCL INFO Channel 04/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-165-213:9625:10020 [0] NCCL INFO Channel 11/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-167-9:1039537:1039932 [0] NCCL INFO Channel 02/0 : 56[0] -> 63[7] via P2P/IPC
+ip-26-0-165-213:9625:10020 [0] NCCL INFO Channel 12/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-166-125:762815:763225 [0] NCCL INFO Channel 09/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-167-9:1039537:1039932 [0] NCCL INFO Channel 03/0 : 56[0] -> 63[7] via P2P/IPC
+ip-26-0-166-125:762815:763225 [0] NCCL INFO Channel 10/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-166-15:127297:127691 [7] NCCL INFO Channel 06/0 : 31[7] -> 38[6] [send] via NET/Libfabric/6(30)/GDRDMA
+ip-26-0-167-9:1039537:1039932 [0] NCCL INFO Channel 04/0 : 56[0] -> 63[7] via P2P/IPC
+ip-26-0-166-125:762822:763222 [7] NCCL INFO Channel 06/0 : 23[7] -> 30[6] [send] via NET/Libfabric/6(22)/GDRDMA
+ip-26-0-166-15:127292:127692 [2] NCCL INFO Channel 02/0 : 19[3] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127297:127691 [7] NCCL INFO Channel 14/0 : 31[7] -> 38[6] [send] via NET/Libfabric/6(30)/GDRDMA
+ip-26-0-166-125:762815:763225 [0] NCCL INFO Channel 11/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-166-15:127292:127692 [2] NCCL INFO Channel 10/0 : 19[3] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-167-9:1039537:1039932 [0] NCCL INFO Channel 09/0 : 56[0] -> 63[7] via P2P/IPC
+ip-26-0-166-15:127293:127689 [3] NCCL INFO Channel 02/0 : 27[3] -> 34[2] [send] via NET/Libfabric/2(26)/GDRDMA
+ip-26-0-166-15:127293:127689 [3] NCCL INFO Channel 10/0 : 27[3] -> 34[2] [send] via NET/Libfabric/2(26)/GDRDMA
+ip-26-0-167-9:1039537:1039932 [0] NCCL INFO Channel 10/0 : 56[0] -> 63[7] via P2P/IPC
+ip-26-0-166-125:762815:763225 [0] NCCL INFO Channel 12/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-166-15:127290:127685 [0] NCCL INFO Channel 06/0 : 24[0] -> 29[5] via P2P/IPC
+ip-26-0-167-9:1039537:1039932 [0] NCCL INFO Channel 11/0 : 56[0] -> 63[7] via P2P/IPC
+ip-26-0-166-36:104429:104829 [0] NCCL INFO Channel 06/0 : 48[0] -> 53[5] via P2P/IPC
+ip-26-0-166-15:127290:127685 [0] NCCL INFO Channel 14/0 : 24[0] -> 29[5] via P2P/IPC
+ip-26-0-167-9:1039537:1039932 [0] NCCL INFO Channel 12/0 : 56[0] -> 63[7] via P2P/IPC
+ip-26-0-167-9:1039544:1039931 [7] NCCL INFO Channel 06/0 : 63[7] -> 6[6] [send] via NET/Libfabric/6(62)/GDRDMA
+ip-26-0-166-36:104429:104829 [0] NCCL INFO Channel 14/0 : 48[0] -> 53[5] via P2P/IPC
+ip-26-0-166-15:127290:127685 [0] NCCL INFO Channel 07/0 : 24[0] -> 30[6] via P2P/IPC
+ip-26-0-162-46:702887:703282 [7] NCCL INFO Channel 06/0 : 7[7] -> 14[6] [send] via NET/Libfabric/6(6)/GDRDMA
+ip-26-0-166-15:127290:127685 [0] NCCL INFO Channel 15/0 : 24[0] -> 30[6] via P2P/IPC
+ip-26-0-162-46:702887:703282 [7] NCCL INFO Channel 14/0 : 7[7] -> 14[6] [send] via NET/Libfabric/6(6)/GDRDMA
+ip-26-0-166-36:104429:104829 [0] NCCL INFO Channel 07/0 : 48[0] -> 54[6] via P2P/IPC
+ip-26-0-166-36:104429:104829 [0] NCCL INFO Channel 15/0 : 48[0] -> 54[6] via P2P/IPC
+ip-26-0-166-15:127290:127685 [0] NCCL INFO Channel 01/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-166-15:127290:127685 [0] NCCL INFO Channel 02/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-166-36:104429:104829 [0] NCCL INFO Channel 01/0 : 48[0] -> 55[7] via P2P/IPC
+ip-26-0-166-15:127290:127685 [0] NCCL INFO Channel 03/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-166-36:104429:104829 [0] NCCL INFO Channel 02/0 : 48[0] -> 55[7] via P2P/IPC
+ip-26-0-162-46:702880:703277 [0] NCCL INFO Channel 06/0 : 0[0] -> 5[5] via P2P/IPC
+ip-26-0-166-15:127290:127685 [0] NCCL INFO Channel 04/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-166-36:104429:104829 [0] NCCL INFO Channel 03/0 : 48[0] -> 55[7] via P2P/IPC
+ip-26-0-162-46:702880:703277 [0] NCCL INFO Channel 14/0 : 0[0] -> 5[5] via P2P/IPC
+ip-26-0-166-15:127290:127685 [0] NCCL INFO Channel 09/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-166-36:104429:104829 [0] NCCL INFO Channel 04/0 : 48[0] -> 55[7] via P2P/IPC
+ip-26-0-166-15:127290:127685 [0] NCCL INFO Channel 10/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-162-46:702880:703277 [0] NCCL INFO Channel 07/0 : 0[0] -> 6[6] via P2P/IPC
+ip-26-0-166-36:104429:104829 [0] NCCL INFO Channel 09/0 : 48[0] -> 55[7] via P2P/IPC
+ip-26-0-166-15:127290:127685 [0] NCCL INFO Channel 11/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-162-46:702880:703277 [0] NCCL INFO Channel 15/0 : 0[0] -> 6[6] via P2P/IPC
+ip-26-0-166-36:104429:104829 [0] NCCL INFO Channel 10/0 : 48[0] -> 55[7] via P2P/IPC
+ip-26-0-166-15:127290:127685 [0] NCCL INFO Channel 12/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-166-36:104429:104829 [0] NCCL INFO Channel 11/0 : 48[0] -> 55[7] via P2P/IPC
+ip-26-0-162-46:702880:703277 [0] NCCL INFO Channel 01/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-166-36:104429:104829 [0] NCCL INFO Channel 12/0 : 48[0] -> 55[7] via P2P/IPC
+ip-26-0-166-36:104436:104824 [7] NCCL INFO Channel 06/0 : 55[7] -> 62[6] [send] via NET/Libfabric/6(54)/GDRDMA
+ip-26-0-162-46:702880:703277 [0] NCCL INFO Channel 02/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-162-46:702880:703277 [0] NCCL INFO Channel 03/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-162-46:702880:703277 [0] NCCL INFO Channel 04/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-162-46:702880:703277 [0] NCCL INFO Channel 09/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-162-46:702880:703277 [0] NCCL INFO Channel 10/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-162-46:702880:703277 [0] NCCL INFO Channel 11/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-162-46:702880:703277 [0] NCCL INFO Channel 12/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-165-213:9628:10023 [3] NCCL INFO Channel 03/0 : 4[4] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9628:10023 [3] NCCL INFO Channel 11/0 : 4[4] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9627:10027 [2] NCCL INFO Channel 02/0 : 3[3] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9628:10023 [3] NCCL INFO Channel 02/0 : 11[3] -> 18[2] [send] via NET/Libfabric/2(10)/GDRDMA
+ip-26-0-165-213:9627:10027 [2] NCCL INFO Channel 10/0 : 3[3] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9628:10023 [3] NCCL INFO Channel 10/0 : 11[3] -> 18[2] [send] via NET/Libfabric/2(10)/GDRDMA
+ip-26-0-166-214:624944:625341 [1] NCCL INFO Channel 01/0 : 26[2] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624944:625341 [1] NCCL INFO Channel 09/0 : 26[2] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624945:625340 [2] NCCL INFO Channel 01/0 : 34[2] -> 41[1] [send] via NET/Libfabric/1(33)/GDRDMA
+ip-26-0-166-244:897241:897635 [1] NCCL INFO Channel 01/0 : 34[2] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624945:625340 [2] NCCL INFO Channel 09/0 : 34[2] -> 41[1] [send] via NET/Libfabric/1(33)/GDRDMA
+ip-26-0-166-244:897241:897635 [1] NCCL INFO Channel 09/0 : 34[2] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9626:10025 [1] NCCL INFO Channel 01/0 : 2[2] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897242:897639 [2] NCCL INFO Channel 01/0 : 42[2] -> 49[1] [send] via NET/Libfabric/1(41)/GDRDMA
+ip-26-0-165-213:9627:10027 [2] NCCL INFO Channel 01/0 : 10[2] -> 17[1] [send] via NET/Libfabric/1(9)/GDRDMA
+ip-26-0-165-213:9626:10025 [1] NCCL INFO Channel 09/0 : 2[2] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897242:897639 [2] NCCL INFO Channel 09/0 : 42[2] -> 49[1] [send] via NET/Libfabric/1(41)/GDRDMA
+ip-26-0-165-213:9627:10027 [2] NCCL INFO Channel 09/0 : 10[2] -> 17[1] [send] via NET/Libfabric/1(9)/GDRDMA
+ip-26-0-166-36:104430:104830 [1] NCCL INFO Channel 01/0 : 42[2] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104430:104830 [1] NCCL INFO Channel 09/0 : 42[2] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-167-9:1039538:1039934 [1] NCCL INFO Channel 01/0 : 50[2] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-167-9:1039538:1039934 [1] NCCL INFO Channel 09/0 : 50[2] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-167-9:1039539:1039938 [2] NCCL INFO Channel 01/0 : 58[2] -> 1[1] [send] via NET/Libfabric/1(57)/GDRDMA
+ip-26-0-167-9:1039539:1039938 [2] NCCL INFO Channel 09/0 : 58[2] -> 1[1] [send] via NET/Libfabric/1(57)/GDRDMA
+ip-26-0-166-244:897244:897636 [4] NCCL INFO Channel 04/0 : 37[5] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897244:897636 [4] NCCL INFO Channel 12/0 : 37[5] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9626:10025 [1] NCCL INFO Channel 01/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-166-244:897244:897636 [4] NCCL INFO Channel 03/0 : 44[4] -> 51[3] [send] via NET/Libfabric/3(43)/GDRDMA
+ip-26-0-165-213:9626:10025 [1] NCCL INFO Channel 02/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-166-244:897244:897636 [4] NCCL INFO Channel 11/0 : 44[4] -> 51[3] [send] via NET/Libfabric/3(43)/GDRDMA
+ip-26-0-165-213:9626:10025 [1] NCCL INFO Channel 03/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-166-15:127291:127690 [1] NCCL INFO Channel 01/0 : 18[2] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127291:127690 [1] NCCL INFO Channel 09/0 : 18[2] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9626:10025 [1] NCCL INFO Channel 04/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-166-214:624947:625339 [4] NCCL INFO Channel 04/0 : 29[5] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127292:127692 [2] NCCL INFO Channel 01/0 : 26[2] -> 33[1] [send] via NET/Libfabric/1(25)/GDRDMA
+ip-26-0-166-15:127292:127692 [2] NCCL INFO Channel 09/0 : 26[2] -> 33[1] [send] via NET/Libfabric/1(25)/GDRDMA
+ip-26-0-166-214:624947:625339 [4] NCCL INFO Channel 12/0 : 29[5] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9626:10025 [1] NCCL INFO Channel 05/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-166-214:624947:625339 [4] NCCL INFO Channel 03/0 : 36[4] -> 43[3] [send] via NET/Libfabric/3(35)/GDRDMA
+ip-26-0-165-213:9626:10025 [1] NCCL INFO Channel 06/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-166-125:762816:763218 [1] NCCL INFO Channel 01/0 : 10[2] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624947:625339 [4] NCCL INFO Channel 11/0 : 36[4] -> 43[3] [send] via NET/Libfabric/3(35)/GDRDMA
+ip-26-0-166-125:762816:763218 [1] NCCL INFO Channel 09/0 : 10[2] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762817:763223 [2] NCCL INFO Channel 01/0 : 18[2] -> 25[1] [send] via NET/Libfabric/1(17)/GDRDMA
+ip-26-0-165-213:9626:10025 [1] NCCL INFO Channel 07/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-166-244:897246:897641 [6] NCCL INFO Channel 06/0 : 39[7] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762817:763223 [2] NCCL INFO Channel 09/0 : 18[2] -> 25[1] [send] via NET/Libfabric/1(17)/GDRDMA
+ip-26-0-166-244:897247:897642 [7] NCCL INFO Channel 14/0 : 47[7] -> 54[6] [send] via NET/Libfabric/6(46)/GDRDMA
+ip-26-0-166-244:897246:897641 [6] NCCL INFO Channel 14/0 : 39[7] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-167-9:1039538:1039934 [1] NCCL INFO Channel 01/0 : 57[1] -> 56[0] via P2P/IPC
+ip-26-0-165-213:9626:10025 [1] NCCL INFO Channel 09/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-166-214:624949:625344 [6] NCCL INFO Channel 06/0 : 31[7] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-167-9:1039538:1039934 [1] NCCL INFO Channel 02/0 : 57[1] -> 56[0] via P2P/IPC
+ip-26-0-165-213:9626:10025 [1] NCCL INFO Channel 10/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-166-214:624949:625344 [6] NCCL INFO Channel 14/0 : 31[7] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624950:625342 [7] NCCL INFO Channel 14/0 : 39[7] -> 46[6] [send] via NET/Libfabric/6(38)/GDRDMA
+ip-26-0-167-9:1039538:1039934 [1] NCCL INFO Channel 03/0 : 57[1] -> 56[0] via P2P/IPC
+ip-26-0-165-213:9629:10021 [4] NCCL INFO Channel 04/0 : 5[5] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9629:10021 [4] NCCL INFO Channel 12/0 : 5[5] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624948:625343 [5] NCCL INFO Channel 05/0 : 30[6] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9626:10025 [1] NCCL INFO Channel 11/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-167-9:1039538:1039934 [1] NCCL INFO Channel 04/0 : 57[1] -> 56[0] via P2P/IPC
+ip-26-0-166-214:624949:625344 [6] NCCL INFO Channel 05/0 : 38[6] -> 45[5] [send] via NET/Libfabric/5(37)/GDRDMA
+ip-26-0-165-213:9629:10021 [4] NCCL INFO Channel 03/0 : 12[4] -> 19[3] [send] via NET/Libfabric/3(11)/GDRDMA
+ip-26-0-166-214:624948:625343 [5] NCCL INFO Channel 13/0 : 30[6] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9629:10021 [4] NCCL INFO Channel 11/0 : 12[4] -> 19[3] [send] via NET/Libfabric/3(11)/GDRDMA
+ip-26-0-166-214:624949:625344 [6] NCCL INFO Channel 13/0 : 38[6] -> 45[5] [send] via NET/Libfabric/5(37)/GDRDMA
+ip-26-0-167-9:1039538:1039934 [1] NCCL INFO Channel 05/0 : 57[1] -> 56[0] via P2P/IPC
+ip-26-0-166-214:624948:625343 [5] NCCL INFO Channel 04/0 : 37[5] -> 44[4] [send] via NET/Libfabric/4(36)/GDRDMA
+ip-26-0-165-213:9626:10025 [1] NCCL INFO Channel 12/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-166-214:624948:625343 [5] NCCL INFO Channel 12/0 : 37[5] -> 44[4] [send] via NET/Libfabric/4(36)/GDRDMA
+ip-26-0-166-244:897245:897638 [5] NCCL INFO Channel 05/0 : 38[6] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624945:625340 [2] NCCL INFO Channel 02/0 : 34[2] -> 33[1] via P2P/IPC
+ip-26-0-167-9:1039538:1039934 [1] NCCL INFO Channel 06/0 : 57[1] -> 56[0] via P2P/IPC
+ip-26-0-166-244:897245:897638 [5] NCCL INFO Channel 13/0 : 38[6] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762819:763221 [4] NCCL INFO Channel 04/0 : 13[5] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897246:897641 [6] NCCL INFO Channel 05/0 : 46[6] -> 53[5] [send] via NET/Libfabric/5(45)/GDRDMA
+ip-26-0-165-213:9626:10025 [1] NCCL INFO Channel 13/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-166-125:762819:763221 [4] NCCL INFO Channel 12/0 : 13[5] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-167-9:1039538:1039934 [1] NCCL INFO Channel 07/0 : 57[1] -> 56[0] via P2P/IPC
+ip-26-0-166-244:897246:897641 [6] NCCL INFO Channel 13/0 : 46[6] -> 53[5] [send] via NET/Libfabric/5(45)/GDRDMA
+ip-26-0-166-36:104430:104830 [1] NCCL INFO Channel 01/0 : 49[1] -> 48[0] via P2P/IPC
+ip-26-0-166-244:897245:897638 [5] NCCL INFO Channel 04/0 : 45[5] -> 52[4] [send] via NET/Libfabric/4(44)/GDRDMA
+ip-26-0-167-9:1039541:1039937 [4] NCCL INFO Channel 04/0 : 53[5] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762819:763221 [4] NCCL INFO Channel 03/0 : 20[4] -> 27[3] [send] via NET/Libfabric/3(19)/GDRDMA
+ip-26-0-166-244:897245:897638 [5] NCCL INFO Channel 12/0 : 45[5] -> 52[4] [send] via NET/Libfabric/4(44)/GDRDMA
+ip-26-0-166-125:762820:763219 [5] NCCL INFO Channel 05/0 : 14[6] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-167-9:1039541:1039937 [4] NCCL INFO Channel 12/0 : 53[5] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9626:10025 [1] NCCL INFO Channel 14/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-167-9:1039538:1039934 [1] NCCL INFO Channel 09/0 : 57[1] -> 56[0] via P2P/IPC
+ip-26-0-166-125:762820:763219 [5] NCCL INFO Channel 13/0 : 14[6] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104430:104830 [1] NCCL INFO Channel 02/0 : 49[1] -> 48[0] via P2P/IPC
+ip-26-0-166-125:762819:763221 [4] NCCL INFO Channel 11/0 : 20[4] -> 27[3] [send] via NET/Libfabric/3(19)/GDRDMA
+ip-26-0-167-9:1039538:1039934 [1] NCCL INFO Channel 10/0 : 57[1] -> 56[0] via P2P/IPC
+ip-26-0-166-125:762820:763219 [5] NCCL INFO Channel 04/0 : 21[5] -> 28[4] [send] via NET/Libfabric/4(20)/GDRDMA
+ip-26-0-166-214:624949:625344 [6] NCCL INFO Channel 06/0 : 38[6] -> 35[3] via P2P/IPC
+ip-26-0-167-9:1039541:1039937 [4] NCCL INFO Channel 03/0 : 60[4] -> 3[3] [send] via NET/Libfabric/3(59)/GDRDMA
+ip-26-0-166-125:762820:763219 [5] NCCL INFO Channel 12/0 : 21[5] -> 28[4] [send] via NET/Libfabric/4(20)/GDRDMA
+ip-26-0-167-9:1039541:1039937 [4] NCCL INFO Channel 11/0 : 60[4] -> 3[3] [send] via NET/Libfabric/3(59)/GDRDMA
+ip-26-0-165-213:9626:10025 [1] NCCL INFO Channel 15/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-166-36:104430:104830 [1] NCCL INFO Channel 03/0 : 49[1] -> 48[0] via P2P/IPC
+ip-26-0-167-9:1039538:1039934 [1] NCCL INFO Channel 11/0 : 57[1] -> 56[0] via P2P/IPC
+ip-26-0-166-125:762821:763224 [6] NCCL INFO Channel 06/0 : 15[7] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762822:763222 [7] NCCL INFO Channel 14/0 : 23[7] -> 30[6] [send] via NET/Libfabric/6(22)/GDRDMA
+ip-26-0-166-125:762821:763224 [6] NCCL INFO Channel 14/0 : 15[7] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104431:104831 [2] NCCL INFO Channel 02/0 : 50[2] -> 49[1] via P2P/IPC
+ip-26-0-167-9:1039538:1039934 [1] NCCL INFO Channel 12/0 : 57[1] -> 56[0] via P2P/IPC
+ip-26-0-166-36:104430:104830 [1] NCCL INFO Channel 04/0 : 49[1] -> 48[0] via P2P/IPC
+ip-26-0-166-125:762821:763224 [6] NCCL INFO Channel 05/0 : 22[6] -> 29[5] [send] via NET/Libfabric/5(21)/GDRDMA
+ip-26-0-166-125:762821:763224 [6] NCCL INFO Channel 13/0 : 22[6] -> 29[5] [send] via NET/Libfabric/5(21)/GDRDMA
+ip-26-0-167-9:1039538:1039934 [1] NCCL INFO Channel 13/0 : 57[1] -> 56[0] via P2P/IPC
+ip-26-0-166-36:104433:104825 [4] NCCL INFO Channel 04/0 : 45[5] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104433:104825 [4] NCCL INFO Channel 12/0 : 45[5] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104431:104831 [2] NCCL INFO Channel 03/0 : 50[2] -> 49[1] via P2P/IPC
+ip-26-0-166-36:104430:104830 [1] NCCL INFO Channel 05/0 : 49[1] -> 48[0] via P2P/IPC
+ip-26-0-167-9:1039538:1039934 [1] NCCL INFO Channel 14/0 : 57[1] -> 56[0] via P2P/IPC
+ip-26-0-167-9:1039538:1039934 [1] NCCL INFO Channel 15/0 : 57[1] -> 56[0] via P2P/IPC
+ip-26-0-167-9:1039543:1039933 [6] NCCL INFO Channel 06/0 : 55[7] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127296:127688 [6] NCCL INFO Channel 06/0 : 23[7] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624945:625340 [2] NCCL INFO Channel 03/0 : 34[2] -> 33[1] via P2P/IPC
+ip-26-0-166-15:127296:127688 [6] NCCL INFO Channel 14/0 : 23[7] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-46:702882:703284 [2] NCCL INFO Channel 02/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-165-213:9631:10022 [6] NCCL INFO Channel 06/0 : 7[7] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-167-9:1039544:1039931 [7] NCCL INFO Channel 14/0 : 63[7] -> 6[6] [send] via NET/Libfabric/6(62)/GDRDMA
+ip-26-0-166-244:897244:897636 [4] NCCL INFO Channel 07/0 : 44[4] -> 55[7] [send] via NET/Libfabric/7(47)/GDRDMA
+ip-26-0-166-36:104430:104830 [1] NCCL INFO Channel 06/0 : 49[1] -> 48[0] via P2P/IPC
+ip-26-0-165-213:9632:10026 [7] NCCL INFO Channel 06/0 : 15[7] -> 22[6] [send] via NET/Libfabric/6(14)/GDRDMA
+ip-26-0-167-9:1039543:1039933 [6] NCCL INFO Channel 14/0 : 55[7] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897244:897636 [4] NCCL INFO Channel 15/0 : 44[4] -> 55[7] [send] via NET/Libfabric/7(47)/GDRDMA
+ip-26-0-165-213:9631:10022 [6] NCCL INFO Channel 14/0 : 7[7] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9632:10026 [7] NCCL INFO Channel 14/0 : 15[7] -> 22[6] [send] via NET/Libfabric/6(14)/GDRDMA
+ip-26-0-162-46:702882:703284 [2] NCCL INFO Channel 03/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-36:104430:104830 [1] NCCL INFO Channel 07/0 : 49[1] -> 48[0] via P2P/IPC
+ip-26-0-166-244:897242:897639 [2] NCCL INFO Channel 02/0 : 42[2] -> 41[1] via P2P/IPC
+ip-26-0-165-213:9630:10024 [5] NCCL INFO Channel 05/0 : 6[6] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127292:127692 [2] NCCL INFO Channel 02/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-165-213:9630:10024 [5] NCCL INFO Channel 13/0 : 6[6] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897241:897635 [1] NCCL INFO Channel 01/0 : 41[1] -> 40[0] via P2P/IPC
+ip-26-0-165-213:9631:10022 [6] NCCL INFO Channel 05/0 : 14[6] -> 21[5] [send] via NET/Libfabric/5(13)/GDRDMA
+ip-26-0-162-46:702882:703284 [2] NCCL INFO Channel 04/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-36:104430:104830 [1] NCCL INFO Channel 09/0 : 49[1] -> 48[0] via P2P/IPC
+ip-26-0-167-9:1039542:1039936 [5] NCCL INFO Channel 05/0 : 54[6] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9631:10022 [6] NCCL INFO Channel 13/0 : 14[6] -> 21[5] [send] via NET/Libfabric/5(13)/GDRDMA
+ip-26-0-166-36:104433:104825 [4] NCCL INFO Channel 03/0 : 52[4] -> 59[3] [send] via NET/Libfabric/3(51)/GDRDMA
+ip-26-0-165-213:9630:10024 [5] NCCL INFO Channel 04/0 : 13[5] -> 20[4] [send] via NET/Libfabric/4(12)/GDRDMA
+ip-26-0-167-9:1039543:1039933 [6] NCCL INFO Channel 05/0 : 62[6] -> 5[5] [send] via NET/Libfabric/5(61)/GDRDMA
+ip-26-0-166-244:897242:897639 [2] NCCL INFO Channel 03/0 : 42[2] -> 41[1] via P2P/IPC
+ip-26-0-166-15:127292:127692 [2] NCCL INFO Channel 03/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-166-36:104433:104825 [4] NCCL INFO Channel 11/0 : 52[4] -> 59[3] [send] via NET/Libfabric/3(51)/GDRDMA
+ip-26-0-165-213:9630:10024 [5] NCCL INFO Channel 12/0 : 13[5] -> 20[4] [send] via NET/Libfabric/4(12)/GDRDMA
+ip-26-0-167-9:1039542:1039936 [5] NCCL INFO Channel 13/0 : 54[6] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-46:702881:703280 [1] NCCL INFO Channel 01/0 : 58[2] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-46:702882:703284 [2] NCCL INFO Channel 05/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-15:127291:127690 [1] NCCL INFO Channel 01/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-167-9:1039543:1039933 [6] NCCL INFO Channel 13/0 : 62[6] -> 5[5] [send] via NET/Libfabric/5(61)/GDRDMA
+ip-26-0-162-46:702881:703280 [1] NCCL INFO Channel 09/0 : 58[2] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104430:104830 [1] NCCL INFO Channel 10/0 : 49[1] -> 48[0] via P2P/IPC
+ip-26-0-166-244:897241:897635 [1] NCCL INFO Channel 02/0 : 41[1] -> 40[0] via P2P/IPC
+ip-26-0-166-125:762816:763218 [1] NCCL INFO Channel 01/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-167-9:1039542:1039936 [5] NCCL INFO Channel 04/0 : 61[5] -> 4[4] [send] via NET/Libfabric/4(60)/GDRDMA
+ip-26-0-166-125:762817:763223 [2] NCCL INFO Channel 02/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-167-9:1039542:1039936 [5] NCCL INFO Channel 12/0 : 61[5] -> 4[4] [send] via NET/Libfabric/4(60)/GDRDMA
+ip-26-0-165-213:9631:10022 [6] NCCL INFO Channel 06/0 : 14[6] -> 11[3] via P2P/IPC
+ip-26-0-166-36:104430:104830 [1] NCCL INFO Channel 11/0 : 49[1] -> 48[0] via P2P/IPC
+ip-26-0-166-244:897245:897638 [5] NCCL INFO Channel 05/0 : 45[5] -> 43[3] via P2P/IPC
+ip-26-0-166-15:127292:127692 [2] NCCL INFO Channel 04/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-166-244:897242:897639 [2] NCCL INFO Channel 04/0 : 42[2] -> 41[1] via P2P/IPC
+ip-26-0-166-15:127291:127690 [1] NCCL INFO Channel 02/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-166-36:104431:104831 [2] NCCL INFO Channel 04/0 : 50[2] -> 49[1] via P2P/IPC
+ip-26-0-166-125:762816:763218 [1] NCCL INFO Channel 02/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-166-214:624949:625344 [6] NCCL INFO Channel 14/0 : 38[6] -> 35[3] via P2P/IPC
+ip-26-0-166-125:762817:763223 [2] NCCL INFO Channel 03/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-166-244:897241:897635 [1] NCCL INFO Channel 03/0 : 41[1] -> 40[0] via P2P/IPC
+ip-26-0-166-36:104430:104830 [1] NCCL INFO Channel 12/0 : 49[1] -> 48[0] via P2P/IPC
+ip-26-0-166-15:127295:127686 [5] NCCL INFO Channel 05/0 : 22[6] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104434:104828 [5] NCCL INFO Channel 05/0 : 46[6] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127294:127687 [4] NCCL INFO Channel 04/0 : 21[5] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624945:625340 [2] NCCL INFO Channel 04/0 : 34[2] -> 33[1] via P2P/IPC
+ip-26-0-166-36:104431:104831 [2] NCCL INFO Channel 05/0 : 50[2] -> 49[1] via P2P/IPC
+ip-26-0-166-15:127295:127686 [5] NCCL INFO Channel 13/0 : 22[6] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127292:127692 [2] NCCL INFO Channel 05/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-166-15:127291:127690 [1] NCCL INFO Channel 03/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-166-15:127296:127688 [6] NCCL INFO Channel 05/0 : 30[6] -> 37[5] [send] via NET/Libfabric/5(29)/GDRDMA
+ip-26-0-166-15:127294:127687 [4] NCCL INFO Channel 12/0 : 21[5] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104434:104828 [5] NCCL INFO Channel 13/0 : 46[6] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104430:104830 [1] NCCL INFO Channel 13/0 : 49[1] -> 48[0] via P2P/IPC
+ip-26-0-166-125:762816:763218 [1] NCCL INFO Channel 03/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-166-15:127296:127688 [6] NCCL INFO Channel 13/0 : 30[6] -> 37[5] [send] via NET/Libfabric/5(29)/GDRDMA
+ip-26-0-166-125:762817:763223 [2] NCCL INFO Channel 04/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-166-15:127295:127686 [5] NCCL INFO Channel 04/0 : 29[5] -> 36[4] [send] via NET/Libfabric/4(28)/GDRDMA
+ip-26-0-166-36:104434:104828 [5] NCCL INFO Channel 04/0 : 53[5] -> 60[4] [send] via NET/Libfabric/4(52)/GDRDMA
+ip-26-0-166-15:127295:127686 [5] NCCL INFO Channel 12/0 : 29[5] -> 36[4] [send] via NET/Libfabric/4(28)/GDRDMA
+ip-26-0-166-15:127294:127687 [4] NCCL INFO Channel 03/0 : 28[4] -> 35[3] [send] via NET/Libfabric/3(27)/GDRDMA
+ip-26-0-166-244:897242:897639 [2] NCCL INFO Channel 05/0 : 42[2] -> 41[1] via P2P/IPC
+ip-26-0-166-36:104431:104831 [2] NCCL INFO Channel 06/0 : 50[2] -> 49[1] via P2P/IPC
+ip-26-0-166-244:897245:897638 [5] NCCL INFO Channel 13/0 : 45[5] -> 43[3] via P2P/IPC
+ip-26-0-166-15:127292:127692 [2] NCCL INFO Channel 06/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-166-36:104434:104828 [5] NCCL INFO Channel 12/0 : 53[5] -> 60[4] [send] via NET/Libfabric/4(52)/GDRDMA
+ip-26-0-166-36:104430:104830 [1] NCCL INFO Channel 14/0 : 49[1] -> 48[0] via P2P/IPC
+ip-26-0-166-214:624945:625340 [2] NCCL INFO Channel 05/0 : 34[2] -> 33[1] via P2P/IPC
+ip-26-0-166-244:897241:897635 [1] NCCL INFO Channel 04/0 : 41[1] -> 40[0] via P2P/IPC
+ip-26-0-166-15:127291:127690 [1] NCCL INFO Channel 04/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-166-15:127294:127687 [4] NCCL INFO Channel 11/0 : 28[4] -> 35[3] [send] via NET/Libfabric/3(27)/GDRDMA
+ip-26-0-166-125:762816:763218 [1] NCCL INFO Channel 04/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-166-125:762821:763224 [6] NCCL INFO Channel 06/0 : 22[6] -> 19[3] via P2P/IPC
+ip-26-0-166-125:762817:763223 [2] NCCL INFO Channel 05/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-166-214:624948:625343 [5] NCCL INFO Channel 05/0 : 37[5] -> 35[3] via P2P/IPC
+ip-26-0-166-15:127296:127688 [6] NCCL INFO Channel 06/0 : 30[6] -> 27[3] via P2P/IPC
+ip-26-0-166-36:104431:104831 [2] NCCL INFO Channel 07/0 : 50[2] -> 49[1] via P2P/IPC
+ip-26-0-166-244:897246:897641 [6] NCCL INFO Channel 06/0 : 46[6] -> 43[3] via P2P/IPC
+ip-26-0-166-36:104430:104830 [1] NCCL INFO Channel 15/0 : 49[1] -> 48[0] via P2P/IPC
+ip-26-0-166-125:762820:763219 [5] NCCL INFO Channel 05/0 : 21[5] -> 19[3] via P2P/IPC
+ip-26-0-166-15:127295:127686 [5] NCCL INFO Channel 05/0 : 29[5] -> 27[3] via P2P/IPC
+ip-26-0-166-15:127292:127692 [2] NCCL INFO Channel 07/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-162-46:702884:703279 [4] NCCL INFO Channel 04/0 : 61[5] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127291:127690 [1] NCCL INFO Channel 05/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-165-213:9631:10022 [6] NCCL INFO Channel 14/0 : 14[6] -> 11[3] via P2P/IPC
+ip-26-0-162-46:702884:703279 [4] NCCL INFO Channel 12/0 : 61[5] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104434:104828 [5] NCCL INFO Channel 05/0 : 53[5] -> 51[3] via P2P/IPC
+ip-26-0-166-214:624945:625340 [2] NCCL INFO Channel 06/0 : 34[2] -> 33[1] via P2P/IPC
+ip-26-0-166-244:897242:897639 [2] NCCL INFO Channel 06/0 : 42[2] -> 41[1] via P2P/IPC
+ip-26-0-166-125:762816:763218 [1] NCCL INFO Channel 05/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-166-244:897241:897635 [1] NCCL INFO Channel 05/0 : 41[1] -> 40[0] via P2P/IPC
+ip-26-0-162-46:702884:703279 [4] NCCL INFO Channel 03/0 : 4[4] -> 11[3] [send] via NET/Libfabric/3(3)/GDRDMA
+ip-26-0-166-36:104431:104831 [2] NCCL INFO Channel 10/0 : 50[2] -> 49[1] via P2P/IPC
+ip-26-0-165-213:9627:10027 [2] NCCL INFO Channel 02/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-166-36:104433:104825 [4] NCCL INFO Channel 07/0 : 52[4] -> 63[7] [send] via NET/Libfabric/7(55)/GDRDMA
+ip-26-0-162-46:702884:703279 [4] NCCL INFO Channel 11/0 : 4[4] -> 11[3] [send] via NET/Libfabric/3(3)/GDRDMA
+ip-26-0-166-214:624948:625343 [5] NCCL INFO Channel 13/0 : 37[5] -> 35[3] via P2P/IPC
+ip-26-0-166-36:104433:104825 [4] NCCL INFO Channel 15/0 : 52[4] -> 63[7] [send] via NET/Libfabric/7(55)/GDRDMA
+ip-26-0-166-125:762817:763223 [2] NCCL INFO Channel 06/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-166-125:762819:763221 [4] NCCL INFO Channel 07/0 : 20[4] -> 31[7] [send] via NET/Libfabric/7(23)/GDRDMA
+ip-26-0-166-244:897246:897641 [6] NCCL INFO Channel 14/0 : 46[6] -> 43[3] via P2P/IPC
+ip-26-0-166-125:762821:763224 [6] NCCL INFO Channel 14/0 : 22[6] -> 19[3] via P2P/IPC
+ip-26-0-162-46:702886:703281 [6] NCCL INFO Channel 06/0 : 63[7] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-46:702882:703284 [2] NCCL INFO Channel 06/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-36:104431:104831 [2] NCCL INFO Channel 11/0 : 50[2] -> 49[1] via P2P/IPC
+ip-26-0-166-36:104434:104828 [5] NCCL INFO Channel 13/0 : 53[5] -> 51[3] via P2P/IPC
+ip-26-0-165-213:9627:10027 [2] NCCL INFO Channel 03/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-162-46:702886:703281 [6] NCCL INFO Channel 14/0 : 63[7] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762820:763219 [5] NCCL INFO Channel 13/0 : 21[5] -> 19[3] via P2P/IPC
+ip-26-0-166-244:897242:897639 [2] NCCL INFO Channel 07/0 : 42[2] -> 41[1] via P2P/IPC
+ip-26-0-166-214:624945:625340 [2] NCCL INFO Channel 07/0 : 34[2] -> 33[1] via P2P/IPC
+ip-26-0-166-15:127292:127692 [2] NCCL INFO Channel 10/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-166-15:127291:127690 [1] NCCL INFO Channel 06/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-166-15:127296:127688 [6] NCCL INFO Channel 14/0 : 30[6] -> 27[3] via P2P/IPC
+ip-26-0-162-46:702885:703278 [5] NCCL INFO Channel 05/0 : 62[6] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897241:897635 [1] NCCL INFO Channel 06/0 : 41[1] -> 40[0] via P2P/IPC
+ip-26-0-166-125:762816:763218 [1] NCCL INFO Channel 06/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-166-15:127295:127686 [5] NCCL INFO Channel 13/0 : 29[5] -> 27[3] via P2P/IPC
+ip-26-0-162-46:702885:703278 [5] NCCL INFO Channel 13/0 : 62[6] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9627:10027 [2] NCCL INFO Channel 04/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-162-46:702886:703281 [6] NCCL INFO Channel 05/0 : 6[6] -> 13[5] [send] via NET/Libfabric/5(5)/GDRDMA
+ip-26-0-162-46:702882:703284 [2] NCCL INFO Channel 07/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-36:104431:104831 [2] NCCL INFO Channel 12/0 : 50[2] -> 49[1] via P2P/IPC
+ip-26-0-162-46:702886:703281 [6] NCCL INFO Channel 13/0 : 6[6] -> 13[5] [send] via NET/Libfabric/5(5)/GDRDMA
+ip-26-0-162-46:702885:703278 [5] NCCL INFO Channel 04/0 : 5[5] -> 12[4] [send] via NET/Libfabric/4(4)/GDRDMA
+ip-26-0-166-125:762817:763223 [2] NCCL INFO Channel 07/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-162-46:702885:703278 [5] NCCL INFO Channel 12/0 : 5[5] -> 12[4] [send] via NET/Libfabric/4(4)/GDRDMA
+ip-26-0-166-244:897242:897639 [2] NCCL INFO Channel 10/0 : 42[2] -> 41[1] via P2P/IPC
+ip-26-0-166-15:127291:127690 [1] NCCL INFO Channel 07/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-166-15:127292:127692 [2] NCCL INFO Channel 11/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-166-36:104431:104831 [2] NCCL INFO Channel 13/0 : 50[2] -> 49[1] via P2P/IPC
+ip-26-0-166-244:897241:897635 [1] NCCL INFO Channel 07/0 : 41[1] -> 40[0] via P2P/IPC
+ip-26-0-165-213:9627:10027 [2] NCCL INFO Channel 05/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-162-46:702882:703284 [2] NCCL INFO Channel 10/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-214:624945:625340 [2] NCCL INFO Channel 10/0 : 34[2] -> 33[1] via P2P/IPC
+ip-26-0-166-125:762816:763218 [1] NCCL INFO Channel 07/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-166-36:104431:104831 [2] NCCL INFO Channel 14/0 : 50[2] -> 49[1] via P2P/IPC
+ip-26-0-162-46:702886:703281 [6] NCCL INFO Channel 06/0 : 6[6] -> 3[3] via P2P/IPC
+ip-26-0-165-213:9630:10024 [5] NCCL INFO Channel 05/0 : 13[5] -> 11[3] via P2P/IPC
+ip-26-0-166-244:897242:897639 [2] NCCL INFO Channel 11/0 : 42[2] -> 41[1] via P2P/IPC
+ip-26-0-166-15:127291:127690 [1] NCCL INFO Channel 09/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-166-125:762817:763223 [2] NCCL INFO Channel 10/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-166-244:897241:897635 [1] NCCL INFO Channel 09/0 : 41[1] -> 40[0] via P2P/IPC
+ip-26-0-166-15:127292:127692 [2] NCCL INFO Channel 12/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-162-46:702885:703278 [5] NCCL INFO Channel 05/0 : 5[5] -> 3[3] via P2P/IPC
+ip-26-0-166-214:624945:625340 [2] NCCL INFO Channel 11/0 : 34[2] -> 33[1] via P2P/IPC
+ip-26-0-162-46:702882:703284 [2] NCCL INFO Channel 11/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-36:104431:104831 [2] NCCL INFO Channel 15/0 : 50[2] -> 49[1] via P2P/IPC
+ip-26-0-166-125:762816:763218 [1] NCCL INFO Channel 09/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-167-9:1039541:1039937 [4] NCCL INFO Channel 07/0 : 60[4] -> 7[7] [send] via NET/Libfabric/7(63)/GDRDMA
+ip-26-0-166-15:127291:127690 [1] NCCL INFO Channel 10/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-166-244:897242:897639 [2] NCCL INFO Channel 12/0 : 42[2] -> 41[1] via P2P/IPC
+ip-26-0-166-125:762817:763223 [2] NCCL INFO Channel 11/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-166-15:127292:127692 [2] NCCL INFO Channel 13/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-166-244:897241:897635 [1] NCCL INFO Channel 10/0 : 41[1] -> 40[0] via P2P/IPC
+ip-26-0-166-214:624945:625340 [2] NCCL INFO Channel 12/0 : 34[2] -> 33[1] via P2P/IPC
+ip-26-0-166-125:762816:763218 [1] NCCL INFO Channel 10/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-166-244:897242:897639 [2] NCCL INFO Channel 13/0 : 42[2] -> 41[1] via P2P/IPC
+ip-26-0-162-46:702886:703281 [6] NCCL INFO Channel 14/0 : 6[6] -> 3[3] via P2P/IPC
+ip-26-0-166-15:127291:127690 [1] NCCL INFO Channel 11/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-166-244:897241:897635 [1] NCCL INFO Channel 11/0 : 41[1] -> 40[0] via P2P/IPC
+ip-26-0-162-46:702882:703284 [2] NCCL INFO Channel 12/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-125:762817:763223 [2] NCCL INFO Channel 12/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-162-46:702885:703278 [5] NCCL INFO Channel 13/0 : 5[5] -> 3[3] via P2P/IPC
+ip-26-0-166-125:762816:763218 [1] NCCL INFO Channel 11/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-166-36:104435:104826 [6] NCCL INFO Channel 06/0 : 47[7] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624945:625340 [2] NCCL INFO Channel 13/0 : 34[2] -> 33[1] via P2P/IPC
+ip-26-0-166-36:104436:104824 [7] NCCL INFO Channel 14/0 : 55[7] -> 62[6] [send] via NET/Libfabric/6(54)/GDRDMA
+ip-26-0-166-36:104435:104826 [6] NCCL INFO Channel 14/0 : 47[7] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897242:897639 [2] NCCL INFO Channel 14/0 : 42[2] -> 41[1] via P2P/IPC
+ip-26-0-166-244:897241:897635 [1] NCCL INFO Channel 12/0 : 41[1] -> 40[0] via P2P/IPC
+ip-26-0-166-214:624945:625340 [2] NCCL INFO Channel 14/0 : 34[2] -> 33[1] via P2P/IPC
+ip-26-0-166-125:762817:763223 [2] NCCL INFO Channel 13/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-166-36:104435:104826 [6] NCCL INFO Channel 05/0 : 54[6] -> 61[5] [send] via NET/Libfabric/5(53)/GDRDMA
+ip-26-0-166-125:762816:763218 [1] NCCL INFO Channel 12/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-166-36:104435:104826 [6] NCCL INFO Channel 13/0 : 54[6] -> 61[5] [send] via NET/Libfabric/5(53)/GDRDMA
+ip-26-0-162-46:702882:703284 [2] NCCL INFO Channel 13/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-162-46:702881:703280 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-244:897242:897639 [2] NCCL INFO Channel 15/0 : 42[2] -> 41[1] via P2P/IPC
+ip-26-0-167-9:1039543:1039933 [6] NCCL INFO Channel 06/0 : 62[6] -> 59[3] via P2P/IPC
+ip-26-0-166-244:897241:897635 [1] NCCL INFO Channel 13/0 : 41[1] -> 40[0] via P2P/IPC
+ip-26-0-166-214:624944:625341 [1] NCCL INFO Channel 01/0 : 33[1] -> 32[0] via P2P/IPC
+ip-26-0-166-214:624945:625340 [2] NCCL INFO Channel 15/0 : 34[2] -> 33[1] via P2P/IPC
+ip-26-0-166-125:762817:763223 [2] NCCL INFO Channel 14/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-162-46:702882:703284 [2] NCCL INFO Channel 14/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-162-46:702881:703280 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762816:763218 [1] NCCL INFO Channel 13/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-166-36:104435:104826 [6] NCCL INFO Channel 06/0 : 54[6] -> 51[3] via P2P/IPC
+ip-26-0-166-244:897241:897635 [1] NCCL INFO Channel 14/0 : 41[1] -> 40[0] via P2P/IPC
+ip-26-0-167-9:1039542:1039936 [5] NCCL INFO Channel 05/0 : 61[5] -> 59[3] via P2P/IPC
+ip-26-0-166-214:624944:625341 [1] NCCL INFO Channel 02/0 : 33[1] -> 32[0] via P2P/IPC
+ip-26-0-167-9:1039543:1039933 [6] NCCL INFO Channel 14/0 : 62[6] -> 59[3] via P2P/IPC
+ip-26-0-166-244:897241:897635 [1] NCCL INFO Channel 15/0 : 41[1] -> 40[0] via P2P/IPC
+ip-26-0-162-46:702882:703284 [2] NCCL INFO Channel 15/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-162-46:702881:703280 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762817:763223 [2] NCCL INFO Channel 15/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-166-214:624944:625341 [1] NCCL INFO Channel 03/0 : 33[1] -> 32[0] via P2P/IPC
+ip-26-0-167-9:1039539:1039938 [2] NCCL INFO Channel 02/0 : 58[2] -> 57[1] via P2P/IPC
+ip-26-0-166-36:104435:104826 [6] NCCL INFO Channel 14/0 : 54[6] -> 51[3] via P2P/IPC
+ip-26-0-166-125:762816:763218 [1] NCCL INFO Channel 14/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-166-214:624944:625341 [1] NCCL INFO Channel 04/0 : 33[1] -> 32[0] via P2P/IPC
+ip-26-0-166-125:762816:763218 [1] NCCL INFO Channel 15/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-162-46:702881:703280 [1] NCCL INFO Channel 04/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624944:625341 [1] NCCL INFO Channel 05/0 : 33[1] -> 32[0] via P2P/IPC
+ip-26-0-165-213:9627:10027 [2] NCCL INFO Channel 06/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-162-46:702881:703280 [1] NCCL INFO Channel 05/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-167-9:1039542:1039936 [5] NCCL INFO Channel 13/0 : 61[5] -> 59[3] via P2P/IPC
+ip-26-0-167-9:1039539:1039938 [2] NCCL INFO Channel 03/0 : 58[2] -> 57[1] via P2P/IPC
+ip-26-0-166-214:624944:625341 [1] NCCL INFO Channel 06/0 : 33[1] -> 32[0] via P2P/IPC
+ip-26-0-166-15:127292:127692 [2] NCCL INFO Channel 14/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-166-15:127291:127690 [1] NCCL INFO Channel 12/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-166-15:127292:127692 [2] NCCL INFO Channel 15/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-166-15:127291:127690 [1] NCCL INFO Channel 13/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-166-15:127291:127690 [1] NCCL INFO Channel 14/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-166-15:127291:127690 [1] NCCL INFO Channel 15/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-162-46:702881:703280 [1] NCCL INFO Channel 06/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624944:625341 [1] NCCL INFO Channel 07/0 : 33[1] -> 32[0] via P2P/IPC
+ip-26-0-167-9:1039539:1039938 [2] NCCL INFO Channel 04/0 : 58[2] -> 57[1] via P2P/IPC
+ip-26-0-162-46:702881:703280 [1] NCCL INFO Channel 07/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-213:9630:10024 [5] NCCL INFO Channel 13/0 : 13[5] -> 11[3] via P2P/IPC
+ip-26-0-165-213:9627:10027 [2] NCCL INFO Channel 07/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-162-46:702881:703280 [1] NCCL INFO Channel 09/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624944:625341 [1] NCCL INFO Channel 09/0 : 33[1] -> 32[0] via P2P/IPC
+ip-26-0-162-46:702881:703280 [1] NCCL INFO Channel 10/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624944:625341 [1] NCCL INFO Channel 10/0 : 33[1] -> 32[0] via P2P/IPC
+ip-26-0-162-46:702881:703280 [1] NCCL INFO Channel 11/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-162-46:702881:703280 [1] NCCL INFO Channel 12/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624944:625341 [1] NCCL INFO Channel 11/0 : 33[1] -> 32[0] via P2P/IPC
+ip-26-0-162-46:702881:703280 [1] NCCL INFO Channel 13/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624944:625341 [1] NCCL INFO Channel 12/0 : 33[1] -> 32[0] via P2P/IPC
+ip-26-0-167-9:1039539:1039938 [2] NCCL INFO Channel 05/0 : 58[2] -> 57[1] via P2P/IPC
+ip-26-0-162-46:702881:703280 [1] NCCL INFO Channel 14/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624944:625341 [1] NCCL INFO Channel 13/0 : 33[1] -> 32[0] via P2P/IPC
+ip-26-0-162-46:702881:703280 [1] NCCL INFO Channel 15/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624944:625341 [1] NCCL INFO Channel 14/0 : 33[1] -> 32[0] via P2P/IPC
+ip-26-0-166-214:624944:625341 [1] NCCL INFO Channel 15/0 : 33[1] -> 32[0] via P2P/IPC
+ip-26-0-165-213:9627:10027 [2] NCCL INFO Channel 10/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-167-9:1039539:1039938 [2] NCCL INFO Channel 06/0 : 58[2] -> 57[1] via P2P/IPC
+ip-26-0-167-9:1039539:1039938 [2] NCCL INFO Channel 07/0 : 58[2] -> 57[1] via P2P/IPC
+ip-26-0-167-9:1039539:1039938 [2] NCCL INFO Channel 10/0 : 58[2] -> 57[1] via P2P/IPC
+ip-26-0-167-9:1039539:1039938 [2] NCCL INFO Channel 11/0 : 58[2] -> 57[1] via P2P/IPC
+ip-26-0-167-9:1039539:1039938 [2] NCCL INFO Channel 12/0 : 58[2] -> 57[1] via P2P/IPC
+ip-26-0-167-9:1039539:1039938 [2] NCCL INFO Channel 13/0 : 58[2] -> 57[1] via P2P/IPC
+ip-26-0-167-9:1039539:1039938 [2] NCCL INFO Channel 14/0 : 58[2] -> 57[1] via P2P/IPC
+ip-26-0-167-9:1039539:1039938 [2] NCCL INFO Channel 15/0 : 58[2] -> 57[1] via P2P/IPC
+ip-26-0-165-213:9627:10027 [2] NCCL INFO Channel 11/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-165-213:9627:10027 [2] NCCL INFO Channel 12/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-165-213:9627:10027 [2] NCCL INFO Channel 13/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-165-213:9627:10027 [2] NCCL INFO Channel 14/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-165-213:9627:10027 [2] NCCL INFO Channel 15/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-166-214:624950:625342 [7] NCCL INFO Channel 07/0 : 28[4] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624950:625342 [7] NCCL INFO Channel 15/0 : 28[4] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127297:127691 [7] NCCL INFO Channel 07/0 : 20[4] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624947:625339 [4] NCCL INFO Channel 07/0 : 36[4] -> 47[7] [send] via NET/Libfabric/7(39)/GDRDMA
+ip-26-0-166-214:624947:625339 [4] NCCL INFO Channel 15/0 : 36[4] -> 47[7] [send] via NET/Libfabric/7(39)/GDRDMA
+ip-26-0-166-15:127297:127691 [7] NCCL INFO Channel 15/0 : 20[4] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127294:127687 [4] NCCL INFO Channel 07/0 : 28[4] -> 39[7] [send] via NET/Libfabric/7(31)/GDRDMA
+ip-26-0-166-15:127294:127687 [4] NCCL INFO Channel 15/0 : 28[4] -> 39[7] [send] via NET/Libfabric/7(31)/GDRDMA
+ip-26-0-166-214:624950:625342 [7] NCCL INFO Channel 07/0 : 39[7] -> 35[3] via P2P/IPC
+ip-26-0-162-46:702887:703282 [7] NCCL INFO Channel 07/0 : 60[4] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-46:702887:703282 [7] NCCL INFO Channel 15/0 : 60[4] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-46:702884:703279 [4] NCCL INFO Channel 07/0 : 4[4] -> 15[7] [send] via NET/Libfabric/7(7)/GDRDMA
+ip-26-0-162-46:702884:703279 [4] NCCL INFO Channel 15/0 : 4[4] -> 15[7] [send] via NET/Libfabric/7(7)/GDRDMA
+ip-26-0-165-213:9632:10026 [7] NCCL INFO Channel 07/0 : 4[4] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9632:10026 [7] NCCL INFO Channel 15/0 : 4[4] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9629:10021 [4] NCCL INFO Channel 07/0 : 12[4] -> 23[7] [send] via NET/Libfabric/7(15)/GDRDMA
+ip-26-0-165-213:9629:10021 [4] NCCL INFO Channel 15/0 : 12[4] -> 23[7] [send] via NET/Libfabric/7(15)/GDRDMA
+ip-26-0-165-213:9632:10026 [7] NCCL INFO Channel 07/0 : 15[7] -> 11[3] via P2P/IPC
+ip-26-0-166-125:762822:763222 [7] NCCL INFO Channel 07/0 : 12[4] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762822:763222 [7] NCCL INFO Channel 15/0 : 12[4] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762819:763221 [4] NCCL INFO Channel 15/0 : 20[4] -> 31[7] [send] via NET/Libfabric/7(23)/GDRDMA
+ip-26-0-166-125:762822:763222 [7] NCCL INFO Channel 07/0 : 23[7] -> 19[3] via P2P/IPC
+ip-26-0-166-15:127297:127691 [7] NCCL INFO Channel 07/0 : 31[7] -> 27[3] via P2P/IPC
+ip-26-0-167-9:1039544:1039931 [7] NCCL INFO Channel 07/0 : 52[4] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-167-9:1039541:1039937 [4] NCCL INFO Channel 15/0 : 60[4] -> 7[7] [send] via NET/Libfabric/7(63)/GDRDMA
+ip-26-0-167-9:1039544:1039931 [7] NCCL INFO Channel 15/0 : 52[4] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897247:897642 [7] NCCL INFO Channel 07/0 : 36[4] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-46:702887:703282 [7] NCCL INFO Channel 07/0 : 7[7] -> 3[3] via P2P/IPC
+ip-26-0-167-9:1039544:1039931 [7] NCCL INFO Channel 07/0 : 63[7] -> 59[3] via P2P/IPC
+ip-26-0-166-36:104436:104824 [7] NCCL INFO Channel 07/0 : 44[4] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104436:104824 [7] NCCL INFO Channel 15/0 : 44[4] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897247:897642 [7] NCCL INFO Channel 15/0 : 36[4] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897247:897642 [7] NCCL INFO Channel 07/0 : 47[7] -> 43[3] via P2P/IPC
+ip-26-0-166-36:104436:104824 [7] NCCL INFO Channel 07/0 : 55[7] -> 51[3] via P2P/IPC
+ip-26-0-166-214:624950:625342 [7] NCCL INFO Channel 15/0 : 39[7] -> 35[3] via P2P/IPC
+ip-26-0-165-213:9632:10026 [7] NCCL INFO Channel 15/0 : 15[7] -> 11[3] via P2P/IPC
+ip-26-0-167-9:1039544:1039931 [7] NCCL INFO Channel 15/0 : 63[7] -> 59[3] via P2P/IPC
+ip-26-0-166-125:762822:763222 [7] NCCL INFO Channel 15/0 : 23[7] -> 19[3] via P2P/IPC
+ip-26-0-166-15:127297:127691 [7] NCCL INFO Channel 15/0 : 31[7] -> 27[3] via P2P/IPC
+ip-26-0-162-46:702887:703282 [7] NCCL INFO Channel 15/0 : 7[7] -> 3[3] via P2P/IPC
+ip-26-0-166-36:104436:104824 [7] NCCL INFO Channel 15/0 : 55[7] -> 51[3] via P2P/IPC
+ip-26-0-167-9:1039544:1039931 [7] NCCL INFO Channel 01/0 : 63[7] -> 62[6] via P2P/IPC
+ip-26-0-166-244:897247:897642 [7] NCCL INFO Channel 15/0 : 47[7] -> 43[3] via P2P/IPC
+ip-26-0-166-214:624950:625342 [7] NCCL INFO Channel 01/0 : 39[7] -> 38[6] via P2P/IPC
+ip-26-0-166-15:127297:127691 [7] NCCL INFO Channel 01/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-165-213:9632:10026 [7] NCCL INFO Channel 01/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-166-36:104436:104824 [7] NCCL INFO Channel 01/0 : 55[7] -> 54[6] via P2P/IPC
+ip-26-0-166-125:762822:763222 [7] NCCL INFO Channel 01/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-167-9:1039544:1039931 [7] NCCL INFO Channel 02/0 : 63[7] -> 62[6] via P2P/IPC
+ip-26-0-166-244:897247:897642 [7] NCCL INFO Channel 01/0 : 47[7] -> 46[6] via P2P/IPC
+ip-26-0-166-244:897247:897642 [7] NCCL INFO Channel 02/0 : 47[7] -> 46[6] via P2P/IPC
+ip-26-0-166-36:104436:104824 [7] NCCL INFO Channel 02/0 : 55[7] -> 54[6] via P2P/IPC
+ip-26-0-166-214:624950:625342 [7] NCCL INFO Channel 02/0 : 39[7] -> 38[6] via P2P/IPC
+ip-26-0-166-244:897247:897642 [7] NCCL INFO Channel 03/0 : 47[7] -> 46[6] via P2P/IPC
+ip-26-0-166-15:127297:127691 [7] NCCL INFO Channel 02/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-166-36:104436:104824 [7] NCCL INFO Channel 03/0 : 55[7] -> 54[6] via P2P/IPC
+ip-26-0-166-214:624950:625342 [7] NCCL INFO Channel 03/0 : 39[7] -> 38[6] via P2P/IPC
+ip-26-0-162-46:702887:703282 [7] NCCL INFO Channel 01/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-166-244:897247:897642 [7] NCCL INFO Channel 04/0 : 47[7] -> 46[6] via P2P/IPC
+ip-26-0-166-214:624950:625342 [7] NCCL INFO Channel 04/0 : 39[7] -> 38[6] via P2P/IPC
+ip-26-0-166-36:104436:104824 [7] NCCL INFO Channel 04/0 : 55[7] -> 54[6] via P2P/IPC
+ip-26-0-166-244:897247:897642 [7] NCCL INFO Channel 05/0 : 47[7] -> 46[6] via P2P/IPC
+ip-26-0-166-244:897247:897642 [7] NCCL INFO Channel 09/0 : 47[7] -> 46[6] via P2P/IPC
+ip-26-0-166-214:624950:625342 [7] NCCL INFO Channel 05/0 : 39[7] -> 38[6] via P2P/IPC
+ip-26-0-166-36:104436:104824 [7] NCCL INFO Channel 05/0 : 55[7] -> 54[6] via P2P/IPC
+ip-26-0-166-125:762822:763222 [7] NCCL INFO Channel 02/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-166-214:624950:625342 [7] NCCL INFO Channel 09/0 : 39[7] -> 38[6] via P2P/IPC
+ip-26-0-167-9:1039544:1039931 [7] NCCL INFO Channel 03/0 : 63[7] -> 62[6] via P2P/IPC
+ip-26-0-166-244:897247:897642 [7] NCCL INFO Channel 10/0 : 47[7] -> 46[6] via P2P/IPC
+ip-26-0-166-36:104436:104824 [7] NCCL INFO Channel 09/0 : 55[7] -> 54[6] via P2P/IPC
+ip-26-0-166-125:762822:763222 [7] NCCL INFO Channel 03/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-166-244:897247:897642 [7] NCCL INFO Channel 11/0 : 47[7] -> 46[6] via P2P/IPC
+ip-26-0-166-125:762822:763222 [7] NCCL INFO Channel 04/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-166-214:624950:625342 [7] NCCL INFO Channel 10/0 : 39[7] -> 38[6] via P2P/IPC
+ip-26-0-166-36:104436:104824 [7] NCCL INFO Channel 10/0 : 55[7] -> 54[6] via P2P/IPC
+ip-26-0-166-244:897247:897642 [7] NCCL INFO Channel 12/0 : 47[7] -> 46[6] via P2P/IPC
+ip-26-0-166-15:127297:127691 [7] NCCL INFO Channel 03/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-166-125:762822:763222 [7] NCCL INFO Channel 05/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-166-214:624950:625342 [7] NCCL INFO Channel 11/0 : 39[7] -> 38[6] via P2P/IPC
+ip-26-0-165-213:9632:10026 [7] NCCL INFO Channel 02/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-167-9:1039544:1039931 [7] NCCL INFO Channel 04/0 : 63[7] -> 62[6] via P2P/IPC
+ip-26-0-166-36:104436:104824 [7] NCCL INFO Channel 11/0 : 55[7] -> 54[6] via P2P/IPC
+ip-26-0-166-15:127297:127691 [7] NCCL INFO Channel 04/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-166-244:897247:897642 [7] NCCL INFO Channel 13/0 : 47[7] -> 46[6] via P2P/IPC
+ip-26-0-166-125:762822:763222 [7] NCCL INFO Channel 09/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-166-15:127297:127691 [7] NCCL INFO Channel 05/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-166-214:624950:625342 [7] NCCL INFO Channel 12/0 : 39[7] -> 38[6] via P2P/IPC
+ip-26-0-166-244:897244:897636 [4] NCCL INFO Channel 01/0 : 44[4] -> 43[3] via P2P/IPC
+ip-26-0-166-214:624947:625339 [4] NCCL INFO Channel 01/0 : 36[4] -> 35[3] via P2P/IPC
+ip-26-0-166-36:104436:104824 [7] NCCL INFO Channel 12/0 : 55[7] -> 54[6] via P2P/IPC
+ip-26-0-166-36:104433:104825 [4] NCCL INFO Channel 01/0 : 52[4] -> 51[3] via P2P/IPC
+ip-26-0-166-125:762822:763222 [7] NCCL INFO Channel 10/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-166-15:127297:127691 [7] NCCL INFO Channel 09/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-166-214:624950:625342 [7] NCCL INFO Channel 13/0 : 39[7] -> 38[6] via P2P/IPC
+ip-26-0-166-214:624947:625339 [4] NCCL INFO Channel 02/0 : 36[4] -> 35[3] via P2P/IPC
+ip-26-0-166-36:104436:104824 [7] NCCL INFO Channel 13/0 : 55[7] -> 54[6] via P2P/IPC
+ip-26-0-162-46:702887:703282 [7] NCCL INFO Channel 02/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-166-125:762822:763222 [7] NCCL INFO Channel 11/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-166-15:127297:127691 [7] NCCL INFO Channel 10/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-167-9:1039544:1039931 [7] NCCL INFO Channel 05/0 : 63[7] -> 62[6] via P2P/IPC
+ip-26-0-166-214:624947:625339 [4] NCCL INFO Channel 04/0 : 36[4] -> 35[3] via P2P/IPC
+ip-26-0-166-15:127294:127687 [4] NCCL INFO Channel 01/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-166-214:624946:625337 [3] NCCL INFO Channel 01/0 : 35[3] -> 34[2] via P2P/IPC
+ip-26-0-166-125:762822:763222 [7] NCCL INFO Channel 12/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-166-15:127297:127691 [7] NCCL INFO Channel 11/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-166-125:762819:763221 [4] NCCL INFO Channel 01/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-166-214:624947:625339 [4] NCCL INFO Channel 09/0 : 36[4] -> 35[3] via P2P/IPC
+ip-26-0-166-15:127294:127687 [4] NCCL INFO Channel 02/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-166-244:897243:897637 [3] NCCL INFO Channel 01/0 : 43[3] -> 42[2] via P2P/IPC
+ip-26-0-166-125:762822:763222 [7] NCCL INFO Channel 13/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-166-244:897243:897637 [3] NCCL INFO Channel 03/0 : 43[3] -> 42[2] via P2P/IPC
+ip-26-0-166-36:104433:104825 [4] NCCL INFO Channel 02/0 : 52[4] -> 51[3] via P2P/IPC
+ip-26-0-166-125:762819:763221 [4] NCCL INFO Channel 02/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-162-46:702887:703282 [7] NCCL INFO Channel 03/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-166-244:897243:897637 [3] NCCL INFO Channel 04/0 : 43[3] -> 42[2] via P2P/IPC
+ip-26-0-166-244:897243:897637 [3] NCCL INFO Channel 05/0 : 43[3] -> 42[2] via P2P/IPC
+ip-26-0-166-125:762819:763221 [4] NCCL INFO Channel 04/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-166-214:624947:625339 [4] NCCL INFO Channel 10/0 : 36[4] -> 35[3] via P2P/IPC
+ip-26-0-166-36:104433:104825 [4] NCCL INFO Channel 04/0 : 52[4] -> 51[3] via P2P/IPC
+ip-26-0-162-46:702887:703282 [7] NCCL INFO Channel 04/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-166-214:624946:625337 [3] NCCL INFO Channel 03/0 : 35[3] -> 34[2] via P2P/IPC
+ip-26-0-166-36:104432:104827 [3] NCCL INFO Channel 01/0 : 51[3] -> 50[2] via P2P/IPC
+ip-26-0-166-244:897243:897637 [3] NCCL INFO Channel 06/0 : 43[3] -> 42[2] via P2P/IPC
+ip-26-0-166-125:762819:763221 [4] NCCL INFO Channel 09/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-166-214:624947:625339 [4] NCCL INFO Channel 12/0 : 36[4] -> 35[3] via P2P/IPC
+ip-26-0-166-244:897244:897636 [4] NCCL INFO Channel 02/0 : 44[4] -> 43[3] via P2P/IPC
+ip-26-0-166-214:624946:625337 [3] NCCL INFO Channel 04/0 : 35[3] -> 34[2] via P2P/IPC
+ip-26-0-166-125:762819:763221 [4] NCCL INFO Channel 10/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-166-36:104433:104825 [4] NCCL INFO Channel 09/0 : 52[4] -> 51[3] via P2P/IPC
+ip-26-0-166-244:897243:897637 [3] NCCL INFO Channel 07/0 : 43[3] -> 42[2] via P2P/IPC
+ip-26-0-162-46:702887:703282 [7] NCCL INFO Channel 05/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-166-15:127297:127691 [7] NCCL INFO Channel 12/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-166-15:127294:127687 [4] NCCL INFO Channel 04/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-166-15:127297:127691 [7] NCCL INFO Channel 13/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-166-15:127294:127687 [4] NCCL INFO Channel 09/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-166-15:127294:127687 [4] NCCL INFO Channel 10/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-166-15:127294:127687 [4] NCCL INFO Channel 12/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-166-125:762819:763221 [4] NCCL INFO Channel 12/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-166-36:104432:104827 [3] NCCL INFO Channel 03/0 : 51[3] -> 50[2] via P2P/IPC
+ip-26-0-166-244:897244:897636 [4] NCCL INFO Channel 04/0 : 44[4] -> 43[3] via P2P/IPC
+ip-26-0-167-9:1039544:1039931 [7] NCCL INFO Channel 09/0 : 63[7] -> 62[6] via P2P/IPC
+ip-26-0-166-244:897243:897637 [3] NCCL INFO Channel 09/0 : 43[3] -> 42[2] via P2P/IPC
+ip-26-0-162-46:702887:703282 [7] NCCL INFO Channel 09/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-166-36:104433:104825 [4] NCCL INFO Channel 10/0 : 52[4] -> 51[3] via P2P/IPC
+ip-26-0-166-36:104435:104826 [6] NCCL INFO Channel 01/0 : 54[6] -> 53[5] via P2P/IPC
+ip-26-0-165-213:9632:10026 [7] NCCL INFO Channel 03/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-166-244:897244:897636 [4] NCCL INFO Channel 09/0 : 44[4] -> 43[3] via P2P/IPC
+ip-26-0-166-36:104432:104827 [3] NCCL INFO Channel 04/0 : 51[3] -> 50[2] via P2P/IPC
+ip-26-0-166-125:762818:763220 [3] NCCL INFO Channel 01/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-166-244:897243:897637 [3] NCCL INFO Channel 11/0 : 43[3] -> 42[2] via P2P/IPC
+ip-26-0-166-36:104433:104825 [4] NCCL INFO Channel 12/0 : 52[4] -> 51[3] via P2P/IPC
+ip-26-0-166-125:762821:763224 [6] NCCL INFO Channel 01/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-166-36:104435:104826 [6] NCCL INFO Channel 02/0 : 54[6] -> 53[5] via P2P/IPC
+ip-26-0-166-244:897244:897636 [4] NCCL INFO Channel 10/0 : 44[4] -> 43[3] via P2P/IPC
+ip-26-0-162-46:702887:703282 [7] NCCL INFO Channel 10/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-166-15:127293:127689 [3] NCCL INFO Channel 01/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-166-125:762818:763220 [3] NCCL INFO Channel 03/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-166-36:104432:104827 [3] NCCL INFO Channel 05/0 : 51[3] -> 50[2] via P2P/IPC
+ip-26-0-166-36:104434:104828 [5] NCCL INFO Channel 01/0 : 53[5] -> 52[4] via P2P/IPC
+ip-26-0-166-15:127296:127688 [6] NCCL INFO Channel 01/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-166-125:762821:763224 [6] NCCL INFO Channel 02/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-166-15:127293:127689 [3] NCCL INFO Channel 03/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-166-125:762818:763220 [3] NCCL INFO Channel 04/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-166-244:897244:897636 [4] NCCL INFO Channel 12/0 : 44[4] -> 43[3] via P2P/IPC
+ip-26-0-166-214:624946:625337 [3] NCCL INFO Channel 05/0 : 35[3] -> 34[2] via P2P/IPC
+ip-26-0-166-36:104435:104826 [6] NCCL INFO Channel 03/0 : 54[6] -> 53[5] via P2P/IPC
+ip-26-0-166-36:104432:104827 [3] NCCL INFO Channel 06/0 : 51[3] -> 50[2] via P2P/IPC
+ip-26-0-166-15:127296:127688 [6] NCCL INFO Channel 02/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-166-36:104434:104828 [5] NCCL INFO Channel 02/0 : 53[5] -> 52[4] via P2P/IPC
+ip-26-0-166-125:762821:763224 [6] NCCL INFO Channel 03/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-162-46:702884:703279 [4] NCCL INFO Channel 01/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-166-15:127293:127689 [3] NCCL INFO Channel 04/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-166-125:762818:763220 [3] NCCL INFO Channel 05/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-167-9:1039544:1039931 [7] NCCL INFO Channel 10/0 : 63[7] -> 62[6] via P2P/IPC
+ip-26-0-166-36:104435:104826 [6] NCCL INFO Channel 04/0 : 54[6] -> 53[5] via P2P/IPC
+ip-26-0-166-125:762820:763219 [5] NCCL INFO Channel 01/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-166-15:127296:127688 [6] NCCL INFO Channel 03/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-166-36:104432:104827 [3] NCCL INFO Channel 07/0 : 51[3] -> 50[2] via P2P/IPC
+ip-26-0-162-46:702887:703282 [7] NCCL INFO Channel 11/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-166-36:104434:104828 [5] NCCL INFO Channel 03/0 : 53[5] -> 52[4] via P2P/IPC
+ip-26-0-162-46:702884:703279 [4] NCCL INFO Channel 02/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-166-125:762821:763224 [6] NCCL INFO Channel 04/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-166-244:897246:897641 [6] NCCL INFO Channel 01/0 : 46[6] -> 45[5] via P2P/IPC
+ip-26-0-166-15:127293:127689 [3] NCCL INFO Channel 05/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-166-125:762818:763220 [3] NCCL INFO Channel 06/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-166-36:104435:104826 [6] NCCL INFO Channel 07/0 : 54[6] -> 53[5] via P2P/IPC
+ip-26-0-162-46:702887:703282 [7] NCCL INFO Channel 12/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-166-125:762820:763219 [5] NCCL INFO Channel 02/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-166-15:127296:127688 [6] NCCL INFO Channel 04/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-167-9:1039544:1039931 [7] NCCL INFO Channel 11/0 : 63[7] -> 62[6] via P2P/IPC
+ip-26-0-166-36:104432:104827 [3] NCCL INFO Channel 09/0 : 51[3] -> 50[2] via P2P/IPC
+ip-26-0-166-36:104434:104828 [5] NCCL INFO Channel 06/0 : 53[5] -> 52[4] via P2P/IPC
+ip-26-0-162-46:702884:703279 [4] NCCL INFO Channel 04/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-166-15:127295:127686 [5] NCCL INFO Channel 01/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-166-125:762821:763224 [6] NCCL INFO Channel 07/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-165-213:9632:10026 [7] NCCL INFO Channel 04/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-166-125:762818:763220 [3] NCCL INFO Channel 07/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-166-15:127293:127689 [3] NCCL INFO Channel 06/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-162-46:702887:703282 [7] NCCL INFO Channel 13/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-166-36:104435:104826 [6] NCCL INFO Channel 09/0 : 54[6] -> 53[5] via P2P/IPC
+ip-26-0-166-125:762820:763219 [5] NCCL INFO Channel 03/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-166-214:624949:625344 [6] NCCL INFO Channel 01/0 : 38[6] -> 37[5] via P2P/IPC
+ip-26-0-167-9:1039541:1039937 [4] NCCL INFO Channel 01/0 : 60[4] -> 59[3] via P2P/IPC
+ip-26-0-166-15:127296:127688 [6] NCCL INFO Channel 07/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-162-46:702884:703279 [4] NCCL INFO Channel 09/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-166-214:624946:625337 [3] NCCL INFO Channel 06/0 : 35[3] -> 34[2] via P2P/IPC
+ip-26-0-166-36:104434:104828 [5] NCCL INFO Channel 07/0 : 53[5] -> 52[4] via P2P/IPC
+ip-26-0-166-15:127295:127686 [5] NCCL INFO Channel 02/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-166-36:104432:104827 [3] NCCL INFO Channel 11/0 : 51[3] -> 50[2] via P2P/IPC
+ip-26-0-166-15:127293:127689 [3] NCCL INFO Channel 07/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-166-125:762821:763224 [6] NCCL INFO Channel 09/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-167-9:1039544:1039931 [7] NCCL INFO Channel 12/0 : 63[7] -> 62[6] via P2P/IPC
+ip-26-0-162-46:702884:703279 [4] NCCL INFO Channel 10/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-166-125:762818:763220 [3] NCCL INFO Channel 09/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-166-36:104435:104826 [6] NCCL INFO Channel 10/0 : 54[6] -> 53[5] via P2P/IPC
+ip-26-0-166-125:762820:763219 [5] NCCL INFO Channel 06/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-166-214:624949:625344 [6] NCCL INFO Channel 02/0 : 38[6] -> 37[5] via P2P/IPC
+ip-26-0-166-244:897246:897641 [6] NCCL INFO Channel 02/0 : 46[6] -> 45[5] via P2P/IPC
+ip-26-0-166-244:897245:897638 [5] NCCL INFO Channel 01/0 : 45[5] -> 44[4] via P2P/IPC
+ip-26-0-166-244:897246:897641 [6] NCCL INFO Channel 03/0 : 46[6] -> 45[5] via P2P/IPC
+ip-26-0-166-15:127296:127688 [6] NCCL INFO Channel 09/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-166-244:897245:897638 [5] NCCL INFO Channel 02/0 : 45[5] -> 44[4] via P2P/IPC
+ip-26-0-166-244:897246:897641 [6] NCCL INFO Channel 04/0 : 46[6] -> 45[5] via P2P/IPC
+ip-26-0-166-244:897245:897638 [5] NCCL INFO Channel 03/0 : 45[5] -> 44[4] via P2P/IPC
+ip-26-0-166-244:897246:897641 [6] NCCL INFO Channel 07/0 : 46[6] -> 45[5] via P2P/IPC
+ip-26-0-166-214:624946:625337 [3] NCCL INFO Channel 07/0 : 35[3] -> 34[2] via P2P/IPC
+ip-26-0-162-46:702884:703279 [4] NCCL INFO Channel 12/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-166-36:104434:104828 [5] NCCL INFO Channel 09/0 : 53[5] -> 52[4] via P2P/IPC
+ip-26-0-166-15:127295:127686 [5] NCCL INFO Channel 03/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-166-36:104432:104827 [3] NCCL INFO Channel 12/0 : 51[3] -> 50[2] via P2P/IPC
+ip-26-0-166-15:127293:127689 [3] NCCL INFO Channel 09/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-167-9:1039541:1039937 [4] NCCL INFO Channel 02/0 : 60[4] -> 59[3] via P2P/IPC
+ip-26-0-166-125:762821:763224 [6] NCCL INFO Channel 10/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-166-36:104435:104826 [6] NCCL INFO Channel 11/0 : 54[6] -> 53[5] via P2P/IPC
+ip-26-0-166-15:127296:127688 [6] NCCL INFO Channel 10/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-166-125:762818:763220 [3] NCCL INFO Channel 11/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-166-125:762820:763219 [5] NCCL INFO Channel 07/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-166-214:624949:625344 [6] NCCL INFO Channel 03/0 : 38[6] -> 37[5] via P2P/IPC
+ip-26-0-166-214:624946:625337 [3] NCCL INFO Channel 09/0 : 35[3] -> 34[2] via P2P/IPC
+ip-26-0-166-244:897245:897638 [5] NCCL INFO Channel 06/0 : 45[5] -> 44[4] via P2P/IPC
+ip-26-0-166-244:897246:897641 [6] NCCL INFO Channel 09/0 : 46[6] -> 45[5] via P2P/IPC
+ip-26-0-167-9:1039544:1039931 [7] NCCL INFO Channel 13/0 : 63[7] -> 62[6] via P2P/IPC
+ip-26-0-166-244:897245:897638 [5] NCCL INFO Channel 07/0 : 45[5] -> 44[4] via P2P/IPC
+ip-26-0-166-15:127295:127686 [5] NCCL INFO Channel 06/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-166-244:897246:897641 [6] NCCL INFO Channel 10/0 : 46[6] -> 45[5] via P2P/IPC
+ip-26-0-166-214:624948:625343 [5] NCCL INFO Channel 01/0 : 37[5] -> 36[4] via P2P/IPC
+ip-26-0-166-244:897243:897637 [3] NCCL INFO Channel 12/0 : 43[3] -> 42[2] via P2P/IPC
+ip-26-0-166-244:897245:897638 [5] NCCL INFO Channel 09/0 : 45[5] -> 44[4] via P2P/IPC
+ip-26-0-166-15:127293:127689 [3] NCCL INFO Channel 11/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-166-36:104434:104828 [5] NCCL INFO Channel 10/0 : 53[5] -> 52[4] via P2P/IPC
+ip-26-0-166-36:104432:104827 [3] NCCL INFO Channel 13/0 : 51[3] -> 50[2] via P2P/IPC
+ip-26-0-166-244:897246:897641 [6] NCCL INFO Channel 11/0 : 46[6] -> 45[5] via P2P/IPC
+ip-26-0-166-214:624949:625344 [6] NCCL INFO Channel 04/0 : 38[6] -> 37[5] via P2P/IPC
+ip-26-0-166-15:127296:127688 [6] NCCL INFO Channel 11/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-166-36:104435:104826 [6] NCCL INFO Channel 12/0 : 54[6] -> 53[5] via P2P/IPC
+ip-26-0-166-125:762821:763224 [6] NCCL INFO Channel 11/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-167-9:1039541:1039937 [4] NCCL INFO Channel 04/0 : 60[4] -> 59[3] via P2P/IPC
+ip-26-0-166-214:624946:625337 [3] NCCL INFO Channel 11/0 : 35[3] -> 34[2] via P2P/IPC
+ip-26-0-166-125:762818:763220 [3] NCCL INFO Channel 12/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-166-125:762820:763219 [5] NCCL INFO Channel 09/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-166-214:624948:625343 [5] NCCL INFO Channel 02/0 : 37[5] -> 36[4] via P2P/IPC
+ip-26-0-166-15:127295:127686 [5] NCCL INFO Channel 07/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-166-244:897245:897638 [5] NCCL INFO Channel 10/0 : 45[5] -> 44[4] via P2P/IPC
+ip-26-0-166-214:624949:625344 [6] NCCL INFO Channel 07/0 : 38[6] -> 37[5] via P2P/IPC
+ip-26-0-166-244:897243:897637 [3] NCCL INFO Channel 13/0 : 43[3] -> 42[2] via P2P/IPC
+ip-26-0-162-46:702883:703283 [3] NCCL INFO Channel 01/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-15:127293:127689 [3] NCCL INFO Channel 12/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-166-15:127296:127688 [6] NCCL INFO Channel 12/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-166-244:897246:897641 [6] NCCL INFO Channel 12/0 : 46[6] -> 45[5] via P2P/IPC
+ip-26-0-166-15:127295:127686 [5] NCCL INFO Channel 09/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-166-214:624946:625337 [3] NCCL INFO Channel 12/0 : 35[3] -> 34[2] via P2P/IPC
+ip-26-0-166-125:762821:763224 [6] NCCL INFO Channel 12/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-166-36:104435:104826 [6] NCCL INFO Channel 15/0 : 54[6] -> 53[5] via P2P/IPC
+ip-26-0-166-214:624948:625343 [5] NCCL INFO Channel 03/0 : 37[5] -> 36[4] via P2P/IPC
+ip-26-0-166-125:762818:763220 [3] NCCL INFO Channel 13/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-166-125:762820:763219 [5] NCCL INFO Channel 10/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-166-214:624949:625344 [6] NCCL INFO Channel 09/0 : 38[6] -> 37[5] via P2P/IPC
+ip-26-0-166-244:897245:897638 [5] NCCL INFO Channel 11/0 : 45[5] -> 44[4] via P2P/IPC
+ip-26-0-166-244:897243:897637 [3] NCCL INFO Channel 14/0 : 43[3] -> 42[2] via P2P/IPC
+ip-26-0-162-46:702883:703283 [3] NCCL INFO Channel 03/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-15:127293:127689 [3] NCCL INFO Channel 13/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-166-15:127296:127688 [6] NCCL INFO Channel 15/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-166-15:127295:127686 [5] NCCL INFO Channel 10/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-166-214:624946:625337 [3] NCCL INFO Channel 13/0 : 35[3] -> 34[2] via P2P/IPC
+ip-26-0-166-244:897243:897637 [3] NCCL INFO Channel 15/0 : 43[3] -> 42[2] via P2P/IPC
+ip-26-0-166-214:624948:625343 [5] NCCL INFO Channel 06/0 : 37[5] -> 36[4] via P2P/IPC
+ip-26-0-166-214:624949:625344 [6] NCCL INFO Channel 10/0 : 38[6] -> 37[5] via P2P/IPC
+ip-26-0-166-125:762818:763220 [3] NCCL INFO Channel 14/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-166-125:762821:763224 [6] NCCL INFO Channel 15/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-166-125:762820:763219 [5] NCCL INFO Channel 11/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-166-36:104432:104827 [3] NCCL INFO Channel 14/0 : 51[3] -> 50[2] via P2P/IPC
+ip-26-0-162-46:702883:703283 [3] NCCL INFO Channel 04/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-36:104434:104828 [5] NCCL INFO Channel 11/0 : 53[5] -> 52[4] via P2P/IPC
+ip-26-0-166-214:624946:625337 [3] NCCL INFO Channel 14/0 : 35[3] -> 34[2] via P2P/IPC
+ip-26-0-166-214:624948:625343 [5] NCCL INFO Channel 07/0 : 37[5] -> 36[4] via P2P/IPC
+ip-26-0-166-214:624949:625344 [6] NCCL INFO Channel 11/0 : 38[6] -> 37[5] via P2P/IPC
+ip-26-0-167-9:1039541:1039937 [4] NCCL INFO Channel 09/0 : 60[4] -> 59[3] via P2P/IPC
+ip-26-0-162-46:702883:703283 [3] NCCL INFO Channel 05/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-125:762818:763220 [3] NCCL INFO Channel 15/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-162-46:702886:703281 [6] NCCL INFO Channel 01/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-162-46:702885:703278 [5] NCCL INFO Channel 01/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-214:624946:625337 [3] NCCL INFO Channel 15/0 : 35[3] -> 34[2] via P2P/IPC
+ip-26-0-166-214:624948:625343 [5] NCCL INFO Channel 09/0 : 37[5] -> 36[4] via P2P/IPC
+ip-26-0-166-214:624949:625344 [6] NCCL INFO Channel 12/0 : 38[6] -> 37[5] via P2P/IPC
+ip-26-0-162-46:702883:703283 [3] NCCL INFO Channel 06/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-162-46:702886:703281 [6] NCCL INFO Channel 02/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-167-9:1039541:1039937 [4] NCCL INFO Channel 10/0 : 60[4] -> 59[3] via P2P/IPC
+ip-26-0-162-46:702885:703278 [5] NCCL INFO Channel 02/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-244:897245:897638 [5] NCCL INFO Channel 14/0 : 45[5] -> 44[4] via P2P/IPC
+ip-26-0-166-244:897246:897641 [6] NCCL INFO Channel 15/0 : 46[6] -> 45[5] via P2P/IPC
+ip-26-0-167-9:1039540:1039935 [3] NCCL INFO Channel 01/0 : 59[3] -> 58[2] via P2P/IPC
+ip-26-0-162-46:702883:703283 [3] NCCL INFO Channel 07/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-214:624948:625343 [5] NCCL INFO Channel 10/0 : 37[5] -> 36[4] via P2P/IPC
+ip-26-0-162-46:702886:703281 [6] NCCL INFO Channel 03/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-166-214:624949:625344 [6] NCCL INFO Channel 15/0 : 38[6] -> 37[5] via P2P/IPC
+ip-26-0-162-46:702885:703278 [5] NCCL INFO Channel 03/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-167-9:1039541:1039937 [4] NCCL INFO Channel 12/0 : 60[4] -> 59[3] via P2P/IPC
+ip-26-0-162-46:702883:703283 [3] NCCL INFO Channel 09/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-162-46:702886:703281 [6] NCCL INFO Channel 04/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-162-46:702885:703278 [5] NCCL INFO Channel 06/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-214:624948:625343 [5] NCCL INFO Channel 11/0 : 37[5] -> 36[4] via P2P/IPC
+ip-26-0-166-15:127293:127689 [3] NCCL INFO Channel 14/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-167-9:1039543:1039933 [6] NCCL INFO Channel 01/0 : 62[6] -> 61[5] via P2P/IPC
+ip-26-0-166-125:762820:763219 [5] NCCL INFO Channel 14/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-162-46:702883:703283 [3] NCCL INFO Channel 11/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-167-9:1039543:1039933 [6] NCCL INFO Channel 02/0 : 62[6] -> 61[5] via P2P/IPC
+ip-26-0-162-46:702886:703281 [6] NCCL INFO Channel 07/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-162-46:702885:703278 [5] NCCL INFO Channel 07/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-167-9:1039543:1039933 [6] NCCL INFO Channel 03/0 : 62[6] -> 61[5] via P2P/IPC
+ip-26-0-166-36:104432:104827 [3] NCCL INFO Channel 15/0 : 51[3] -> 50[2] via P2P/IPC
+ip-26-0-166-36:104434:104828 [5] NCCL INFO Channel 14/0 : 53[5] -> 52[4] via P2P/IPC
+ip-26-0-167-9:1039543:1039933 [6] NCCL INFO Channel 04/0 : 62[6] -> 61[5] via P2P/IPC
+ip-26-0-162-46:702883:703283 [3] NCCL INFO Channel 12/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-162-46:702886:703281 [6] NCCL INFO Channel 09/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-162-46:702885:703278 [5] NCCL INFO Channel 09/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-167-9:1039543:1039933 [6] NCCL INFO Channel 07/0 : 62[6] -> 61[5] via P2P/IPC
+ip-26-0-167-9:1039543:1039933 [6] NCCL INFO Channel 09/0 : 62[6] -> 61[5] via P2P/IPC
+ip-26-0-162-46:702883:703283 [3] NCCL INFO Channel 13/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-162-46:702886:703281 [6] NCCL INFO Channel 10/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-162-46:702885:703278 [5] NCCL INFO Channel 10/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-167-9:1039540:1039935 [3] NCCL INFO Channel 03/0 : 59[3] -> 58[2] via P2P/IPC
+ip-26-0-167-9:1039543:1039933 [6] NCCL INFO Channel 10/0 : 62[6] -> 61[5] via P2P/IPC
+ip-26-0-166-15:127295:127686 [5] NCCL INFO Channel 11/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-162-46:702883:703283 [3] NCCL INFO Channel 14/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-162-46:702883:703283 [3] NCCL INFO Channel 15/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-167-9:1039540:1039935 [3] NCCL INFO Channel 04/0 : 59[3] -> 58[2] via P2P/IPC
+ip-26-0-167-9:1039543:1039933 [6] NCCL INFO Channel 11/0 : 62[6] -> 61[5] via P2P/IPC
+ip-26-0-167-9:1039542:1039936 [5] NCCL INFO Channel 01/0 : 61[5] -> 60[4] via P2P/IPC
+ip-26-0-166-15:127293:127689 [3] NCCL INFO Channel 15/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-167-9:1039543:1039933 [6] NCCL INFO Channel 12/0 : 62[6] -> 61[5] via P2P/IPC
+ip-26-0-167-9:1039540:1039935 [3] NCCL INFO Channel 05/0 : 59[3] -> 58[2] via P2P/IPC
+ip-26-0-167-9:1039542:1039936 [5] NCCL INFO Channel 02/0 : 61[5] -> 60[4] via P2P/IPC
+ip-26-0-162-46:702885:703278 [5] NCCL INFO Channel 11/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-162-46:702886:703281 [6] NCCL INFO Channel 11/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-166-125:762820:763219 [5] NCCL INFO Channel 15/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-167-9:1039543:1039933 [6] NCCL INFO Channel 15/0 : 62[6] -> 61[5] via P2P/IPC
+ip-26-0-167-9:1039540:1039935 [3] NCCL INFO Channel 06/0 : 59[3] -> 58[2] via P2P/IPC
+ip-26-0-166-36:104434:104828 [5] NCCL INFO Channel 15/0 : 53[5] -> 52[4] via P2P/IPC
+ip-26-0-167-9:1039542:1039936 [5] NCCL INFO Channel 03/0 : 61[5] -> 60[4] via P2P/IPC
+ip-26-0-167-9:1039540:1039935 [3] NCCL INFO Channel 07/0 : 59[3] -> 58[2] via P2P/IPC
+ip-26-0-166-15:127295:127686 [5] NCCL INFO Channel 14/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-167-9:1039542:1039936 [5] NCCL INFO Channel 06/0 : 61[5] -> 60[4] via P2P/IPC
+ip-26-0-165-213:9632:10026 [7] NCCL INFO Channel 05/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-167-9:1039540:1039935 [3] NCCL INFO Channel 09/0 : 59[3] -> 58[2] via P2P/IPC
+ip-26-0-167-9:1039542:1039936 [5] NCCL INFO Channel 07/0 : 61[5] -> 60[4] via P2P/IPC
+ip-26-0-166-214:624948:625343 [5] NCCL INFO Channel 14/0 : 37[5] -> 36[4] via P2P/IPC
+ip-26-0-167-9:1039540:1039935 [3] NCCL INFO Channel 11/0 : 59[3] -> 58[2] via P2P/IPC
+ip-26-0-167-9:1039542:1039936 [5] NCCL INFO Channel 09/0 : 61[5] -> 60[4] via P2P/IPC
+ip-26-0-167-9:1039540:1039935 [3] NCCL INFO Channel 12/0 : 59[3] -> 58[2] via P2P/IPC
+ip-26-0-167-9:1039542:1039936 [5] NCCL INFO Channel 10/0 : 61[5] -> 60[4] via P2P/IPC
+ip-26-0-166-244:897245:897638 [5] NCCL INFO Channel 15/0 : 45[5] -> 44[4] via P2P/IPC
+ip-26-0-162-46:702885:703278 [5] NCCL INFO Channel 14/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-162-46:702886:703281 [6] NCCL INFO Channel 12/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-167-9:1039540:1039935 [3] NCCL INFO Channel 13/0 : 59[3] -> 58[2] via P2P/IPC
+ip-26-0-167-9:1039542:1039936 [5] NCCL INFO Channel 11/0 : 61[5] -> 60[4] via P2P/IPC
+ip-26-0-166-214:624948:625343 [5] NCCL INFO Channel 15/0 : 37[5] -> 36[4] via P2P/IPC
+ip-26-0-165-213:9632:10026 [7] NCCL INFO Channel 09/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-166-15:127295:127686 [5] NCCL INFO Channel 15/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-167-9:1039540:1039935 [3] NCCL INFO Channel 14/0 : 59[3] -> 58[2] via P2P/IPC
+ip-26-0-167-9:1039542:1039936 [5] NCCL INFO Channel 14/0 : 61[5] -> 60[4] via P2P/IPC
+ip-26-0-162-46:702886:703281 [6] NCCL INFO Channel 15/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-167-9:1039540:1039935 [3] NCCL INFO Channel 15/0 : 59[3] -> 58[2] via P2P/IPC
+ip-26-0-167-9:1039542:1039936 [5] NCCL INFO Channel 15/0 : 61[5] -> 60[4] via P2P/IPC
+ip-26-0-165-213:9632:10026 [7] NCCL INFO Channel 10/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-162-46:702885:703278 [5] NCCL INFO Channel 15/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-213:9632:10026 [7] NCCL INFO Channel 11/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-165-213:9632:10026 [7] NCCL INFO Channel 12/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-165-213:9632:10026 [7] NCCL INFO Channel 13/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-165-213:9629:10021 [4] NCCL INFO Channel 01/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-165-213:9629:10021 [4] NCCL INFO Channel 02/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-165-213:9628:10023 [3] NCCL INFO Channel 01/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-165-213:9629:10021 [4] NCCL INFO Channel 04/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-165-213:9628:10023 [3] NCCL INFO Channel 03/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-165-213:9629:10021 [4] NCCL INFO Channel 09/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-165-213:9628:10023 [3] NCCL INFO Channel 04/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-165-213:9629:10021 [4] NCCL INFO Channel 10/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-165-213:9628:10023 [3] NCCL INFO Channel 05/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-165-213:9631:10022 [6] NCCL INFO Channel 01/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-165-213:9629:10021 [4] NCCL INFO Channel 12/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-165-213:9628:10023 [3] NCCL INFO Channel 06/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-165-213:9631:10022 [6] NCCL INFO Channel 02/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-165-213:9630:10024 [5] NCCL INFO Channel 01/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-165-213:9628:10023 [3] NCCL INFO Channel 07/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-165-213:9631:10022 [6] NCCL INFO Channel 03/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-165-213:9628:10023 [3] NCCL INFO Channel 09/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-165-213:9630:10024 [5] NCCL INFO Channel 02/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-165-213:9631:10022 [6] NCCL INFO Channel 04/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-165-213:9628:10023 [3] NCCL INFO Channel 11/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-165-213:9630:10024 [5] NCCL INFO Channel 03/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-165-213:9631:10022 [6] NCCL INFO Channel 07/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-165-213:9628:10023 [3] NCCL INFO Channel 12/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-165-213:9630:10024 [5] NCCL INFO Channel 06/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-165-213:9631:10022 [6] NCCL INFO Channel 09/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-165-213:9628:10023 [3] NCCL INFO Channel 13/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-165-213:9630:10024 [5] NCCL INFO Channel 07/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-165-213:9631:10022 [6] NCCL INFO Channel 10/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-165-213:9628:10023 [3] NCCL INFO Channel 14/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-165-213:9630:10024 [5] NCCL INFO Channel 09/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-165-213:9631:10022 [6] NCCL INFO Channel 11/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-165-213:9628:10023 [3] NCCL INFO Channel 15/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-165-213:9630:10024 [5] NCCL INFO Channel 10/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-165-213:9631:10022 [6] NCCL INFO Channel 12/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-165-213:9630:10024 [5] NCCL INFO Channel 11/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-165-213:9631:10022 [6] NCCL INFO Channel 15/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-165-213:9630:10024 [5] NCCL INFO Channel 14/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-165-213:9630:10024 [5] NCCL INFO Channel 15/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-166-36:104429:104829 [0] NCCL INFO Connected all rings
+ip-26-0-166-36:104429:104829 [0] NCCL INFO Channel 02/0 : 48[0] -> 49[1] via P2P/IPC
+ip-26-0-166-15:127290:127685 [0] NCCL INFO Connected all rings
+ip-26-0-166-15:127290:127685 [0] NCCL INFO Channel 02/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-166-36:104430:104830 [1] NCCL INFO Connected all rings
+ip-26-0-166-15:127291:127690 [1] NCCL INFO Connected all rings
+ip-26-0-162-46:702880:703277 [0] NCCL INFO Connected all rings
+ip-26-0-162-46:702880:703277 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-36:104434:104828 [5] NCCL INFO Connected all rings
+ip-26-0-166-244:897246:897641 [6] NCCL INFO Connected all rings
+ip-26-0-166-244:897242:897639 [2] NCCL INFO Connected all rings
+ip-26-0-167-9:1039543:1039933 [6] NCCL INFO Connected all rings
+ip-26-0-166-36:104430:104830 [1] NCCL INFO Channel 01/0 : 49[1] -> 50[2] via P2P/IPC
+ip-26-0-166-36:104430:104830 [1] NCCL INFO Channel 03/0 : 49[1] -> 50[2] via P2P/IPC
+ip-26-0-166-36:104430:104830 [1] NCCL INFO Channel 04/0 : 49[1] -> 50[2] via P2P/IPC
+ip-26-0-162-46:702882:703284 [2] NCCL INFO Connected all rings
+ip-26-0-166-36:104430:104830 [1] NCCL INFO Channel 05/0 : 49[1] -> 50[2] via P2P/IPC
+ip-26-0-166-36:104430:104830 [1] NCCL INFO Channel 06/0 : 49[1] -> 50[2] via P2P/IPC
+ip-26-0-166-36:104430:104830 [1] NCCL INFO Channel 07/0 : 49[1] -> 50[2] via P2P/IPC
+ip-26-0-166-36:104430:104830 [1] NCCL INFO Channel 09/0 : 49[1] -> 50[2] via P2P/IPC
+ip-26-0-166-36:104430:104830 [1] NCCL INFO Channel 11/0 : 49[1] -> 50[2] via P2P/IPC
+ip-26-0-166-36:104430:104830 [1] NCCL INFO Channel 12/0 : 49[1] -> 50[2] via P2P/IPC
+ip-26-0-166-36:104430:104830 [1] NCCL INFO Channel 13/0 : 49[1] -> 50[2] via P2P/IPC
+ip-26-0-165-213:9626:10025 [1] NCCL INFO Connected all rings
+ip-26-0-166-36:104430:104830 [1] NCCL INFO Channel 14/0 : 49[1] -> 50[2] via P2P/IPC
+ip-26-0-166-36:104430:104830 [1] NCCL INFO Channel 15/0 : 49[1] -> 50[2] via P2P/IPC
+ip-26-0-162-46:702880:703277 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-214:624943:625338 [0] NCCL INFO Connected all rings
+ip-26-0-166-214:624943:625338 [0] NCCL INFO Channel 02/0 : 32[0] -> 33[1] via P2P/IPC
+ip-26-0-162-46:702880:703277 [0] NCCL INFO Channel 04/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-162-46:702880:703277 [0] NCCL INFO Channel 05/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-167-9:1039537:1039932 [0] NCCL INFO Connected all rings
+ip-26-0-167-9:1039537:1039932 [0] NCCL INFO Channel 02/0 : 56[0] -> 57[1] via P2P/IPC
+ip-26-0-162-46:702880:703277 [0] NCCL INFO Channel 06/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-162-46:702880:703277 [0] NCCL INFO Channel 07/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-167-9:1039537:1039932 [0] NCCL INFO Channel 03/0 : 56[0] -> 57[1] via P2P/IPC
+ip-26-0-162-46:702880:703277 [0] NCCL INFO Channel 10/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-125:762815:763225 [0] NCCL INFO Connected all rings
+ip-26-0-166-125:762815:763225 [0] NCCL INFO Channel 02/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-162-46:702880:703277 [0] NCCL INFO Channel 11/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-167-9:1039537:1039932 [0] NCCL INFO Channel 04/0 : 56[0] -> 57[1] via P2P/IPC
+ip-26-0-166-125:762815:763225 [0] NCCL INFO Channel 03/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-162-46:702880:703277 [0] NCCL INFO Channel 12/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-125:762815:763225 [0] NCCL INFO Channel 04/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-162-46:702880:703277 [0] NCCL INFO Channel 13/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-125:762815:763225 [0] NCCL INFO Channel 05/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-162-46:702880:703277 [0] NCCL INFO Channel 14/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-167-9:1039537:1039932 [0] NCCL INFO Channel 05/0 : 56[0] -> 57[1] via P2P/IPC
+ip-26-0-166-36:104429:104829 [0] NCCL INFO Channel 03/0 : 48[0] -> 49[1] via P2P/IPC
+ip-26-0-166-125:762815:763225 [0] NCCL INFO Channel 06/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-162-46:702880:703277 [0] NCCL INFO Channel 15/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-167-9:1039537:1039932 [0] NCCL INFO Channel 06/0 : 56[0] -> 57[1] via P2P/IPC
+ip-26-0-166-244:897241:897635 [1] NCCL INFO Connected all rings
+ip-26-0-166-125:762815:763225 [0] NCCL INFO Channel 07/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-167-9:1039537:1039932 [0] NCCL INFO Channel 07/0 : 56[0] -> 57[1] via P2P/IPC
+ip-26-0-166-36:104429:104829 [0] NCCL INFO Channel 04/0 : 48[0] -> 49[1] via P2P/IPC
+ip-26-0-166-125:762815:763225 [0] NCCL INFO Channel 10/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-166-36:104429:104829 [0] NCCL INFO Channel 05/0 : 48[0] -> 49[1] via P2P/IPC
+ip-26-0-167-9:1039537:1039932 [0] NCCL INFO Channel 10/0 : 56[0] -> 57[1] via P2P/IPC
+ip-26-0-166-125:762815:763225 [0] NCCL INFO Channel 11/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-166-36:104429:104829 [0] NCCL INFO Channel 06/0 : 48[0] -> 49[1] via P2P/IPC
+ip-26-0-167-9:1039537:1039932 [0] NCCL INFO Channel 11/0 : 56[0] -> 57[1] via P2P/IPC
+ip-26-0-166-125:762815:763225 [0] NCCL INFO Channel 12/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-167-9:1039537:1039932 [0] NCCL INFO Channel 12/0 : 56[0] -> 57[1] via P2P/IPC
+ip-26-0-166-125:762815:763225 [0] NCCL INFO Channel 13/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-166-36:104429:104829 [0] NCCL INFO Channel 07/0 : 48[0] -> 49[1] via P2P/IPC
+ip-26-0-166-125:762815:763225 [0] NCCL INFO Channel 14/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-167-9:1039537:1039932 [0] NCCL INFO Channel 13/0 : 56[0] -> 57[1] via P2P/IPC
+ip-26-0-166-244:897240:897640 [0] NCCL INFO Connected all rings
+ip-26-0-166-244:897240:897640 [0] NCCL INFO Channel 02/0 : 40[0] -> 41[1] via P2P/IPC
+ip-26-0-166-36:104429:104829 [0] NCCL INFO Channel 10/0 : 48[0] -> 49[1] via P2P/IPC
+ip-26-0-166-36:104429:104829 [0] NCCL INFO Channel 11/0 : 48[0] -> 49[1] via P2P/IPC
+ip-26-0-166-125:762815:763225 [0] NCCL INFO Channel 15/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-166-244:897240:897640 [0] NCCL INFO Channel 03/0 : 40[0] -> 41[1] via P2P/IPC
+ip-26-0-166-36:104429:104829 [0] NCCL INFO Channel 12/0 : 48[0] -> 49[1] via P2P/IPC
+ip-26-0-166-36:104429:104829 [0] NCCL INFO Channel 13/0 : 48[0] -> 49[1] via P2P/IPC
+ip-26-0-166-244:897240:897640 [0] NCCL INFO Channel 04/0 : 40[0] -> 41[1] via P2P/IPC
+ip-26-0-166-244:897240:897640 [0] NCCL INFO Channel 05/0 : 40[0] -> 41[1] via P2P/IPC
+ip-26-0-166-36:104429:104829 [0] NCCL INFO Channel 14/0 : 48[0] -> 49[1] via P2P/IPC
+ip-26-0-167-9:1039537:1039932 [0] NCCL INFO Channel 14/0 : 56[0] -> 57[1] via P2P/IPC
+ip-26-0-166-36:104429:104829 [0] NCCL INFO Channel 15/0 : 48[0] -> 49[1] via P2P/IPC
+ip-26-0-167-9:1039537:1039932 [0] NCCL INFO Channel 15/0 : 56[0] -> 57[1] via P2P/IPC
+ip-26-0-166-244:897240:897640 [0] NCCL INFO Channel 06/0 : 40[0] -> 41[1] via P2P/IPC
+ip-26-0-166-125:762820:763219 [5] NCCL INFO Connected all rings
+ip-26-0-166-214:624943:625338 [0] NCCL INFO Channel 03/0 : 32[0] -> 33[1] via P2P/IPC
+ip-26-0-166-36:104429:104829 [0] NCCL INFO Channel 05/0 : 48[0] -> 55[7] via P2P/IPC
+ip-26-0-166-15:127292:127692 [2] NCCL INFO Connected all rings
+ip-26-0-166-244:897240:897640 [0] NCCL INFO Channel 07/0 : 40[0] -> 41[1] via P2P/IPC
+ip-26-0-166-36:104429:104829 [0] NCCL INFO Channel 06/0 : 48[0] -> 55[7] via P2P/IPC
+ip-26-0-166-244:897240:897640 [0] NCCL INFO Channel 10/0 : 40[0] -> 41[1] via P2P/IPC
+ip-26-0-166-244:897241:897635 [1] NCCL INFO Channel 01/0 : 41[1] -> 42[2] via P2P/IPC
+ip-26-0-166-15:127290:127685 [0] NCCL INFO Channel 03/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-166-36:104429:104829 [0] NCCL INFO Channel 07/0 : 48[0] -> 55[7] via P2P/IPC
+ip-26-0-166-214:624943:625338 [0] NCCL INFO Channel 04/0 : 32[0] -> 33[1] via P2P/IPC
+ip-26-0-166-244:897240:897640 [0] NCCL INFO Channel 11/0 : 40[0] -> 41[1] via P2P/IPC
+ip-26-0-166-125:762819:763221 [4] NCCL INFO Connected all rings
+ip-26-0-166-244:897241:897635 [1] NCCL INFO Channel 03/0 : 41[1] -> 42[2] via P2P/IPC
+ip-26-0-166-125:762818:763220 [3] NCCL INFO Connected all rings
+ip-26-0-166-214:624943:625338 [0] NCCL INFO Channel 05/0 : 32[0] -> 33[1] via P2P/IPC
+ip-26-0-167-9:1039540:1039935 [3] NCCL INFO Connected all rings
+ip-26-0-166-244:897240:897640 [0] NCCL INFO Channel 12/0 : 40[0] -> 41[1] via P2P/IPC
+ip-26-0-166-15:127297:127691 [7] NCCL INFO Connected all rings
+ip-26-0-166-244:897241:897635 [1] NCCL INFO Channel 04/0 : 41[1] -> 42[2] via P2P/IPC
+ip-26-0-167-9:1039539:1039938 [2] NCCL INFO Connected all rings
+ip-26-0-166-36:104429:104829 [0] NCCL INFO Channel 13/0 : 48[0] -> 55[7] via P2P/IPC
+ip-26-0-166-15:127290:127685 [0] NCCL INFO Channel 04/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-166-214:624943:625338 [0] NCCL INFO Channel 06/0 : 32[0] -> 33[1] via P2P/IPC
+ip-26-0-166-214:624944:625341 [1] NCCL INFO Connected all rings
+ip-26-0-166-214:624946:625337 [3] NCCL INFO Connected all rings
+ip-26-0-166-15:127290:127685 [0] NCCL INFO Channel 05/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-166-214:624943:625338 [0] NCCL INFO Channel 07/0 : 32[0] -> 33[1] via P2P/IPC
+ip-26-0-166-36:104435:104826 [6] NCCL INFO Connected all rings
+ip-26-0-166-244:897240:897640 [0] NCCL INFO Channel 13/0 : 40[0] -> 41[1] via P2P/IPC
+ip-26-0-166-244:897245:897638 [5] NCCL INFO Connected all rings
+ip-26-0-166-244:897241:897635 [1] NCCL INFO Channel 05/0 : 41[1] -> 42[2] via P2P/IPC
+ip-26-0-166-36:104429:104829 [0] NCCL INFO Channel 14/0 : 48[0] -> 55[7] via P2P/IPC
+ip-26-0-166-125:762817:763223 [2] NCCL INFO Connected all rings
+ip-26-0-166-15:127290:127685 [0] NCCL INFO Channel 06/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-166-214:624943:625338 [0] NCCL INFO Channel 10/0 : 32[0] -> 33[1] via P2P/IPC
+ip-26-0-166-244:897240:897640 [0] NCCL INFO Channel 14/0 : 40[0] -> 41[1] via P2P/IPC
+ip-26-0-166-36:104429:104829 [0] NCCL INFO Channel 15/0 : 48[0] -> 55[7] via P2P/IPC
+ip-26-0-166-15:127290:127685 [0] NCCL INFO Channel 07/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-166-244:897241:897635 [1] NCCL INFO Channel 06/0 : 41[1] -> 42[2] via P2P/IPC
+ip-26-0-166-214:624943:625338 [0] NCCL INFO Channel 11/0 : 32[0] -> 33[1] via P2P/IPC
+ip-26-0-166-244:897240:897640 [0] NCCL INFO Channel 15/0 : 40[0] -> 41[1] via P2P/IPC
+ip-26-0-162-46:702881:703280 [1] NCCL INFO Connected all rings
+ip-26-0-167-9:1039542:1039936 [5] NCCL INFO Connected all rings
+ip-26-0-166-15:127290:127685 [0] NCCL INFO Channel 10/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-166-244:897241:897635 [1] NCCL INFO Channel 07/0 : 41[1] -> 42[2] via P2P/IPC
+ip-26-0-166-214:624943:625338 [0] NCCL INFO Channel 12/0 : 32[0] -> 33[1] via P2P/IPC
+ip-26-0-166-15:127290:127685 [0] NCCL INFO Channel 11/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-166-36:104433:104825 [4] NCCL INFO Connected all rings
+ip-26-0-166-36:104436:104824 [7] NCCL INFO Connected all rings
+ip-26-0-166-214:624943:625338 [0] NCCL INFO Channel 13/0 : 32[0] -> 33[1] via P2P/IPC
+ip-26-0-166-244:897243:897637 [3] NCCL INFO Connected all rings
+ip-26-0-166-15:127290:127685 [0] NCCL INFO Channel 12/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-166-125:762822:763222 [7] NCCL INFO Connected all rings
+ip-26-0-166-214:624943:625338 [0] NCCL INFO Channel 14/0 : 32[0] -> 33[1] via P2P/IPC
+ip-26-0-166-125:762820:763219 [5] NCCL INFO Channel 01/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-166-244:897241:897635 [1] NCCL INFO Channel 09/0 : 41[1] -> 42[2] via P2P/IPC
+ip-26-0-166-214:624943:625338 [0] NCCL INFO Channel 15/0 : 32[0] -> 33[1] via P2P/IPC
+ip-26-0-162-46:702883:703283 [3] NCCL INFO Connected all rings
+ip-26-0-166-125:762820:763219 [5] NCCL INFO Channel 02/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-166-214:624945:625340 [2] NCCL INFO Connected all rings
+ip-26-0-166-244:897241:897635 [1] NCCL INFO Channel 11/0 : 41[1] -> 42[2] via P2P/IPC
+ip-26-0-166-36:104434:104828 [5] NCCL INFO Channel 01/0 : 53[5] -> 54[6] via P2P/IPC
+ip-26-0-166-125:762820:763219 [5] NCCL INFO Channel 03/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-166-244:897241:897635 [1] NCCL INFO Channel 12/0 : 41[1] -> 42[2] via P2P/IPC
+ip-26-0-166-15:127293:127689 [3] NCCL INFO Connected all rings
+ip-26-0-166-15:127290:127685 [0] NCCL INFO Channel 13/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-166-36:104434:104828 [5] NCCL INFO Channel 02/0 : 53[5] -> 54[6] via P2P/IPC
+ip-26-0-166-244:897241:897635 [1] NCCL INFO Channel 13/0 : 41[1] -> 42[2] via P2P/IPC
+ip-26-0-166-125:762820:763219 [5] NCCL INFO Channel 04/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-166-244:897247:897642 [7] NCCL INFO Connected all rings
+ip-26-0-166-214:624950:625342 [7] NCCL INFO Connected all rings
+ip-26-0-167-9:1039539:1039938 [2] NCCL INFO Channel 01/0 : 58[2] -> 59[3] via P2P/IPC
+ip-26-0-167-9:1039540:1039935 [3] NCCL INFO Channel 01/0 : 59[3] -> 60[4] via P2P/IPC
+ip-26-0-166-36:104435:104826 [6] NCCL INFO Channel 01/0 : 54[6] -> 55[7] via P2P/IPC
+ip-26-0-166-36:104434:104828 [5] NCCL INFO Channel 03/0 : 53[5] -> 54[6] via P2P/IPC
+ip-26-0-166-244:897244:897636 [4] NCCL INFO Connected all rings
+ip-26-0-166-125:762818:763220 [3] NCCL INFO Channel 01/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-166-244:897241:897635 [1] NCCL INFO Channel 14/0 : 41[1] -> 42[2] via P2P/IPC
+ip-26-0-166-125:762819:763221 [4] NCCL INFO Channel 01/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-166-125:762820:763219 [5] NCCL INFO Channel 05/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-166-36:104434:104828 [5] NCCL INFO Channel 04/0 : 53[5] -> 54[6] via P2P/IPC
+ip-26-0-166-214:624949:625344 [6] NCCL INFO Connected all rings
+ip-26-0-166-36:104435:104826 [6] NCCL INFO Channel 02/0 : 54[6] -> 55[7] via P2P/IPC
+ip-26-0-166-125:762818:763220 [3] NCCL INFO Channel 02/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-167-9:1039544:1039931 [7] NCCL INFO Connected all rings
+ip-26-0-166-125:762820:763219 [5] NCCL INFO Channel 07/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-167-9:1039539:1039938 [2] NCCL INFO Channel 02/0 : 58[2] -> 59[3] via P2P/IPC
+ip-26-0-167-9:1039540:1039935 [3] NCCL INFO Channel 02/0 : 59[3] -> 60[4] via P2P/IPC
+ip-26-0-166-125:762818:763220 [3] NCCL INFO Channel 03/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-166-36:104432:104827 [3] NCCL INFO Connected all rings
+ip-26-0-166-125:762819:763221 [4] NCCL INFO Channel 02/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-166-125:762820:763219 [5] NCCL INFO Channel 09/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-166-214:624948:625343 [5] NCCL INFO Connected all rings
+ip-26-0-167-9:1039539:1039938 [2] NCCL INFO Channel 04/0 : 58[2] -> 59[3] via P2P/IPC
+ip-26-0-167-9:1039540:1039935 [3] NCCL INFO Channel 03/0 : 59[3] -> 60[4] via P2P/IPC
+ip-26-0-166-214:624946:625337 [3] NCCL INFO Channel 01/0 : 35[3] -> 36[4] via P2P/IPC
+ip-26-0-166-214:624944:625341 [1] NCCL INFO Channel 01/0 : 33[1] -> 34[2] via P2P/IPC
+ip-26-0-166-36:104434:104828 [5] NCCL INFO Channel 05/0 : 53[5] -> 54[6] via P2P/IPC
+ip-26-0-166-36:104435:104826 [6] NCCL INFO Channel 03/0 : 54[6] -> 55[7] via P2P/IPC
+ip-26-0-166-125:762818:763220 [3] NCCL INFO Channel 05/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-167-9:1039539:1039938 [2] NCCL INFO Channel 05/0 : 58[2] -> 59[3] via P2P/IPC
+ip-26-0-167-9:1039540:1039935 [3] NCCL INFO Channel 05/0 : 59[3] -> 60[4] via P2P/IPC
+ip-26-0-167-9:1039542:1039936 [5] NCCL INFO Channel 01/0 : 61[5] -> 62[6] via P2P/IPC
+ip-26-0-166-125:762819:763221 [4] NCCL INFO Channel 03/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-166-125:762820:763219 [5] NCCL INFO Channel 10/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-166-36:104434:104828 [5] NCCL INFO Channel 07/0 : 53[5] -> 54[6] via P2P/IPC
+ip-26-0-166-214:624946:625337 [3] NCCL INFO Channel 02/0 : 35[3] -> 36[4] via P2P/IPC
+ip-26-0-166-214:624944:625341 [1] NCCL INFO Channel 03/0 : 33[1] -> 34[2] via P2P/IPC
+ip-26-0-166-36:104435:104826 [6] NCCL INFO Channel 04/0 : 54[6] -> 55[7] via P2P/IPC
+ip-26-0-167-9:1039543:1039933 [6] NCCL INFO Channel 01/0 : 62[6] -> 63[7] via P2P/IPC
+ip-26-0-166-125:762817:763223 [2] NCCL INFO Channel 01/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-166-125:762818:763220 [3] NCCL INFO Channel 06/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-166-125:762819:763221 [4] NCCL INFO Channel 04/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-167-9:1039539:1039938 [2] NCCL INFO Channel 06/0 : 58[2] -> 59[3] via P2P/IPC
+ip-26-0-166-36:104434:104828 [5] NCCL INFO Channel 09/0 : 53[5] -> 54[6] via P2P/IPC
+ip-26-0-167-9:1039540:1039935 [3] NCCL INFO Channel 06/0 : 59[3] -> 60[4] via P2P/IPC
+ip-26-0-166-36:104435:104826 [6] NCCL INFO Channel 05/0 : 54[6] -> 55[7] via P2P/IPC
+ip-26-0-166-125:762820:763219 [5] NCCL INFO Channel 11/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-167-9:1039542:1039936 [5] NCCL INFO Channel 02/0 : 61[5] -> 62[6] via P2P/IPC
+ip-26-0-166-214:624946:625337 [3] NCCL INFO Channel 03/0 : 35[3] -> 36[4] via P2P/IPC
+ip-26-0-166-214:624944:625341 [1] NCCL INFO Channel 04/0 : 33[1] -> 34[2] via P2P/IPC
+ip-26-0-165-213:9631:10022 [6] NCCL INFO Connected all rings
+ip-26-0-167-9:1039543:1039933 [6] NCCL INFO Channel 02/0 : 62[6] -> 63[7] via P2P/IPC
+ip-26-0-166-214:624947:625339 [4] NCCL INFO Connected all rings
+ip-26-0-162-46:702881:703280 [1] NCCL INFO Channel 01/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-166-125:762818:763220 [3] NCCL INFO Channel 07/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-166-125:762817:763223 [2] NCCL INFO Channel 02/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-162-46:702882:703284 [2] NCCL INFO Channel 01/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-125:762819:763221 [4] NCCL INFO Channel 06/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-166-15:127290:127685 [0] NCCL INFO Channel 14/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-166-36:104434:104828 [5] NCCL INFO Channel 10/0 : 53[5] -> 54[6] via P2P/IPC
+ip-26-0-166-125:762820:763219 [5] NCCL INFO Channel 12/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-166-36:104435:104826 [6] NCCL INFO Channel 06/0 : 54[6] -> 55[7] via P2P/IPC
+ip-26-0-166-36:104433:104825 [4] NCCL INFO Channel 01/0 : 52[4] -> 53[5] via P2P/IPC
+ip-26-0-162-46:702881:703280 [1] NCCL INFO Channel 03/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-166-214:624946:625337 [3] NCCL INFO Channel 05/0 : 35[3] -> 36[4] via P2P/IPC
+ip-26-0-166-214:624944:625341 [1] NCCL INFO Channel 05/0 : 33[1] -> 34[2] via P2P/IPC
+ip-26-0-166-15:127291:127690 [1] NCCL INFO Channel 01/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-166-15:127290:127685 [0] NCCL INFO Channel 15/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-162-46:702882:703284 [2] NCCL INFO Channel 02/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-125:762818:763220 [3] NCCL INFO Channel 09/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-167-9:1039541:1039937 [4] NCCL INFO Connected all rings
+ip-26-0-166-125:762817:763223 [2] NCCL INFO Channel 04/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-166-125:762819:763221 [4] NCCL INFO Channel 07/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-166-36:104434:104828 [5] NCCL INFO Channel 11/0 : 53[5] -> 54[6] via P2P/IPC
+ip-26-0-166-15:127291:127690 [1] NCCL INFO Channel 03/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-166-36:104435:104826 [6] NCCL INFO Channel 09/0 : 54[6] -> 55[7] via P2P/IPC
+ip-26-0-166-36:104433:104825 [4] NCCL INFO Channel 02/0 : 52[4] -> 53[5] via P2P/IPC
+ip-26-0-167-9:1039539:1039938 [2] NCCL INFO Channel 07/0 : 58[2] -> 59[3] via P2P/IPC
+ip-26-0-166-125:762820:763219 [5] NCCL INFO Channel 13/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-167-9:1039540:1039935 [3] NCCL INFO Channel 07/0 : 59[3] -> 60[4] via P2P/IPC
+ip-26-0-166-214:624946:625337 [3] NCCL INFO Channel 06/0 : 35[3] -> 36[4] via P2P/IPC
+ip-26-0-166-214:624944:625341 [1] NCCL INFO Channel 06/0 : 33[1] -> 34[2] via P2P/IPC
+ip-26-0-167-9:1039542:1039936 [5] NCCL INFO Channel 03/0 : 61[5] -> 62[6] via P2P/IPC
+ip-26-0-167-9:1039543:1039933 [6] NCCL INFO Channel 03/0 : 62[6] -> 63[7] via P2P/IPC
+ip-26-0-166-15:127291:127690 [1] NCCL INFO Channel 04/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-166-36:104434:104828 [5] NCCL INFO Channel 12/0 : 53[5] -> 54[6] via P2P/IPC
+ip-26-0-166-36:104435:104826 [6] NCCL INFO Channel 10/0 : 54[6] -> 55[7] via P2P/IPC
+ip-26-0-166-36:104433:104825 [4] NCCL INFO Channel 03/0 : 52[4] -> 53[5] via P2P/IPC
+ip-26-0-166-244:897242:897639 [2] NCCL INFO Channel 01/0 : 42[2] -> 43[3] via P2P/IPC
+ip-26-0-167-9:1039539:1039938 [2] NCCL INFO Channel 09/0 : 58[2] -> 59[3] via P2P/IPC
+ip-26-0-162-46:702881:703280 [1] NCCL INFO Channel 04/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-166-214:624946:625337 [3] NCCL INFO Channel 07/0 : 35[3] -> 36[4] via P2P/IPC
+ip-26-0-166-214:624944:625341 [1] NCCL INFO Channel 07/0 : 33[1] -> 34[2] via P2P/IPC
+ip-26-0-166-15:127291:127690 [1] NCCL INFO Channel 05/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-166-244:897241:897635 [1] NCCL INFO Channel 15/0 : 41[1] -> 42[2] via P2P/IPC
+ip-26-0-166-36:104434:104828 [5] NCCL INFO Channel 13/0 : 53[5] -> 54[6] via P2P/IPC
+ip-26-0-166-36:104435:104826 [6] NCCL INFO Channel 11/0 : 54[6] -> 55[7] via P2P/IPC
+ip-26-0-162-46:702882:703284 [2] NCCL INFO Channel 04/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-36:104433:104825 [4] NCCL INFO Channel 04/0 : 52[4] -> 53[5] via P2P/IPC
+ip-26-0-166-244:897242:897639 [2] NCCL INFO Channel 02/0 : 42[2] -> 43[3] via P2P/IPC
+ip-26-0-166-15:127291:127690 [1] NCCL INFO Channel 06/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-166-214:624946:625337 [3] NCCL INFO Channel 09/0 : 35[3] -> 36[4] via P2P/IPC
+ip-26-0-166-36:104434:104828 [5] NCCL INFO Channel 15/0 : 53[5] -> 54[6] via P2P/IPC
+ip-26-0-166-36:104435:104826 [6] NCCL INFO Channel 12/0 : 54[6] -> 55[7] via P2P/IPC
+ip-26-0-167-9:1039540:1039935 [3] NCCL INFO Channel 09/0 : 59[3] -> 60[4] via P2P/IPC
+ip-26-0-166-214:624944:625341 [1] NCCL INFO Channel 09/0 : 33[1] -> 34[2] via P2P/IPC
+ip-26-0-166-36:104433:104825 [4] NCCL INFO Channel 06/0 : 52[4] -> 53[5] via P2P/IPC
+ip-26-0-166-244:897240:897640 [0] NCCL INFO Channel 05/0 : 40[0] -> 47[7] via P2P/IPC
+ip-26-0-162-46:702881:703280 [1] NCCL INFO Channel 05/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-166-15:127291:127690 [1] NCCL INFO Channel 07/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-162-46:702882:703284 [2] NCCL INFO Channel 05/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-244:897242:897639 [2] NCCL INFO Channel 04/0 : 42[2] -> 43[3] via P2P/IPC
+ip-26-0-166-36:104435:104826 [6] NCCL INFO Channel 13/0 : 54[6] -> 55[7] via P2P/IPC
+ip-26-0-166-36:104433:104825 [4] NCCL INFO Channel 07/0 : 52[4] -> 53[5] via P2P/IPC
+ip-26-0-166-214:624946:625337 [3] NCCL INFO Channel 10/0 : 35[3] -> 36[4] via P2P/IPC
+ip-26-0-166-214:624944:625341 [1] NCCL INFO Channel 11/0 : 33[1] -> 34[2] via P2P/IPC
+ip-26-0-162-46:702881:703280 [1] NCCL INFO Channel 06/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-166-244:897240:897640 [0] NCCL INFO Channel 06/0 : 40[0] -> 47[7] via P2P/IPC
+ip-26-0-166-36:104435:104826 [6] NCCL INFO Channel 14/0 : 54[6] -> 55[7] via P2P/IPC
+ip-26-0-162-46:702882:703284 [2] NCCL INFO Channel 06/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-244:897243:897637 [3] NCCL INFO Channel 01/0 : 43[3] -> 44[4] via P2P/IPC
+ip-26-0-166-36:104433:104825 [4] NCCL INFO Channel 09/0 : 52[4] -> 53[5] via P2P/IPC
+ip-26-0-166-214:624945:625340 [2] NCCL INFO Channel 01/0 : 34[2] -> 35[3] via P2P/IPC
+ip-26-0-166-15:127296:127688 [6] NCCL INFO Connected all rings
+ip-26-0-166-244:897242:897639 [2] NCCL INFO Channel 05/0 : 42[2] -> 43[3] via P2P/IPC
+ip-26-0-167-9:1039542:1039936 [5] NCCL INFO Channel 04/0 : 61[5] -> 62[6] via P2P/IPC
+ip-26-0-166-125:762817:763223 [2] NCCL INFO Channel 05/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-167-9:1039543:1039933 [6] NCCL INFO Channel 04/0 : 62[6] -> 63[7] via P2P/IPC
+ip-26-0-166-125:762818:763220 [3] NCCL INFO Channel 10/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-162-46:702881:703280 [1] NCCL INFO Channel 07/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-166-125:762819:763221 [4] NCCL INFO Channel 09/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-166-36:104433:104825 [4] NCCL INFO Channel 10/0 : 52[4] -> 53[5] via P2P/IPC
+ip-26-0-162-46:702882:703284 [2] NCCL INFO Channel 07/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-244:897240:897640 [0] NCCL INFO Channel 07/0 : 40[0] -> 47[7] via P2P/IPC
+ip-26-0-166-244:897245:897638 [5] NCCL INFO Channel 01/0 : 45[5] -> 46[6] via P2P/IPC
+ip-26-0-166-214:624946:625337 [3] NCCL INFO Channel 11/0 : 35[3] -> 36[4] via P2P/IPC
+ip-26-0-166-214:624944:625341 [1] NCCL INFO Channel 12/0 : 33[1] -> 34[2] via P2P/IPC
+ip-26-0-167-9:1039542:1039936 [5] NCCL INFO Channel 05/0 : 61[5] -> 62[6] via P2P/IPC
+ip-26-0-167-9:1039543:1039933 [6] NCCL INFO Channel 05/0 : 62[6] -> 63[7] via P2P/IPC
+ip-26-0-166-244:897243:897637 [3] NCCL INFO Channel 02/0 : 43[3] -> 44[4] via P2P/IPC
+ip-26-0-166-15:127291:127690 [1] NCCL INFO Channel 09/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-166-36:104435:104826 [6] NCCL INFO Channel 06/0 : 46[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897242:897639 [2] NCCL INFO Channel 06/0 : 42[2] -> 43[3] via P2P/IPC
+ip-26-0-166-36:104433:104825 [4] NCCL INFO Channel 11/0 : 52[4] -> 53[5] via P2P/IPC
+ip-26-0-166-125:762817:763223 [2] NCCL INFO Channel 06/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-166-36:104432:104827 [3] NCCL INFO Channel 01/0 : 51[3] -> 52[4] via P2P/IPC
+ip-26-0-166-36:104435:104826 [6] NCCL INFO Channel 14/0 : 46[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762818:763220 [3] NCCL INFO Channel 11/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-166-36:104435:104826 [6] NCCL INFO Channel 06/0 : 54[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762819:763221 [4] NCCL INFO Channel 10/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-167-9:1039539:1039938 [2] NCCL INFO Channel 10/0 : 58[2] -> 59[3] via P2P/IPC
+ip-26-0-162-46:702881:703280 [1] NCCL INFO Channel 09/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-166-125:762818:763220 [3] NCCL INFO Channel 13/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-166-214:624944:625341 [1] NCCL INFO Channel 13/0 : 33[1] -> 34[2] via P2P/IPC
+ip-26-0-166-36:104433:104825 [4] NCCL INFO Channel 12/0 : 52[4] -> 53[5] via P2P/IPC
+ip-26-0-166-214:624946:625337 [3] NCCL INFO Channel 13/0 : 35[3] -> 36[4] via P2P/IPC
+ip-26-0-162-46:702882:703284 [2] NCCL INFO Channel 09/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-36:104432:104827 [3] NCCL INFO Channel 02/0 : 51[3] -> 52[4] via P2P/IPC
+ip-26-0-167-9:1039542:1039936 [5] NCCL INFO Channel 07/0 : 61[5] -> 62[6] via P2P/IPC
+ip-26-0-167-9:1039543:1039933 [6] NCCL INFO Channel 06/0 : 62[6] -> 63[7] via P2P/IPC
+ip-26-0-166-244:897240:897640 [0] NCCL INFO Channel 13/0 : 40[0] -> 47[7] via P2P/IPC
+ip-26-0-166-244:897245:897638 [5] NCCL INFO Channel 02/0 : 45[5] -> 46[6] via P2P/IPC
+ip-26-0-166-244:897243:897637 [3] NCCL INFO Channel 03/0 : 43[3] -> 44[4] via P2P/IPC
+ip-26-0-167-9:1039540:1039935 [3] NCCL INFO Channel 10/0 : 59[3] -> 60[4] via P2P/IPC
+ip-26-0-166-244:897242:897639 [2] NCCL INFO Channel 07/0 : 42[2] -> 43[3] via P2P/IPC
+ip-26-0-166-36:104433:104825 [4] NCCL INFO Channel 14/0 : 52[4] -> 53[5] via P2P/IPC
+ip-26-0-167-9:1039539:1039938 [2] NCCL INFO Channel 12/0 : 58[2] -> 59[3] via P2P/IPC
+ip-26-0-166-36:104432:104827 [3] NCCL INFO Channel 03/0 : 51[3] -> 52[4] via P2P/IPC
+ip-26-0-166-214:624944:625341 [1] NCCL INFO Channel 14/0 : 33[1] -> 34[2] via P2P/IPC
+ip-26-0-162-46:702881:703280 [1] NCCL INFO Channel 11/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-166-214:624946:625337 [3] NCCL INFO Channel 14/0 : 35[3] -> 36[4] via P2P/IPC
+ip-26-0-166-15:127294:127687 [4] NCCL INFO Connected all rings
+ip-26-0-162-46:702882:703284 [2] NCCL INFO Channel 10/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-167-9:1039542:1039936 [5] NCCL INFO Channel 09/0 : 61[5] -> 62[6] via P2P/IPC
+ip-26-0-166-125:762820:763219 [5] NCCL INFO Channel 15/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-166-15:127291:127690 [1] NCCL INFO Channel 11/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-166-125:762819:763221 [4] NCCL INFO Channel 11/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-166-36:104433:104825 [4] NCCL INFO Channel 15/0 : 52[4] -> 53[5] via P2P/IPC
+ip-26-0-166-214:624949:625344 [6] NCCL INFO Channel 01/0 : 38[6] -> 39[7] via P2P/IPC
+ip-26-0-167-9:1039543:1039933 [6] NCCL INFO Channel 09/0 : 62[6] -> 63[7] via P2P/IPC
+ip-26-0-166-214:624945:625340 [2] NCCL INFO Channel 02/0 : 34[2] -> 35[3] via P2P/IPC
+ip-26-0-166-36:104432:104827 [3] NCCL INFO Channel 05/0 : 51[3] -> 52[4] via P2P/IPC
+ip-26-0-166-244:897240:897640 [0] NCCL INFO Channel 14/0 : 40[0] -> 47[7] via P2P/IPC
+ip-26-0-166-244:897245:897638 [5] NCCL INFO Channel 03/0 : 45[5] -> 46[6] via P2P/IPC
+ip-26-0-166-214:624944:625341 [1] NCCL INFO Channel 15/0 : 33[1] -> 34[2] via P2P/IPC
+ip-26-0-166-244:897243:897637 [3] NCCL INFO Channel 05/0 : 43[3] -> 44[4] via P2P/IPC
+ip-26-0-166-214:624946:625337 [3] NCCL INFO Channel 15/0 : 35[3] -> 36[4] via P2P/IPC
+ip-26-0-166-244:897242:897639 [2] NCCL INFO Channel 09/0 : 42[2] -> 43[3] via P2P/IPC
+ip-26-0-165-213:9627:10027 [2] NCCL INFO Connected all rings
+ip-26-0-167-9:1039540:1039935 [3] NCCL INFO Channel 11/0 : 59[3] -> 60[4] via P2P/IPC
+ip-26-0-166-244:897246:897641 [6] NCCL INFO Channel 01/0 : 46[6] -> 47[7] via P2P/IPC
+ip-26-0-162-46:702881:703280 [1] NCCL INFO Channel 12/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-167-9:1039539:1039938 [2] NCCL INFO Channel 13/0 : 58[2] -> 59[3] via P2P/IPC
+ip-26-0-166-125:762817:763223 [2] NCCL INFO Channel 07/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-166-244:897244:897636 [4] NCCL INFO Channel 01/0 : 44[4] -> 45[5] via P2P/IPC
+ip-26-0-162-46:702882:703284 [2] NCCL INFO Channel 12/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-15:127292:127692 [2] NCCL INFO Channel 01/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-166-125:762819:763221 [4] NCCL INFO Channel 12/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-166-214:624949:625344 [6] NCCL INFO Channel 02/0 : 38[6] -> 39[7] via P2P/IPC
+ip-26-0-166-214:624945:625340 [2] NCCL INFO Channel 04/0 : 34[2] -> 35[3] via P2P/IPC
+ip-26-0-162-46:702886:703281 [6] NCCL INFO Connected all rings
+ip-26-0-167-9:1039542:1039936 [5] NCCL INFO Channel 10/0 : 61[5] -> 62[6] via P2P/IPC
+ip-26-0-166-125:762818:763220 [3] NCCL INFO Channel 14/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-166-244:897245:897638 [5] NCCL INFO Channel 04/0 : 45[5] -> 46[6] via P2P/IPC
+ip-26-0-166-214:624948:625343 [5] NCCL INFO Channel 01/0 : 37[5] -> 38[6] via P2P/IPC
+ip-26-0-166-15:127291:127690 [1] NCCL INFO Channel 12/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-162-46:702881:703280 [1] NCCL INFO Channel 13/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-167-9:1039543:1039933 [6] NCCL INFO Channel 10/0 : 62[6] -> 63[7] via P2P/IPC
+ip-26-0-166-125:762817:763223 [2] NCCL INFO Channel 09/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-166-214:624943:625338 [0] NCCL INFO Channel 05/0 : 32[0] -> 39[7] via P2P/IPC
+ip-26-0-166-244:897240:897640 [0] NCCL INFO Channel 15/0 : 40[0] -> 47[7] via P2P/IPC
+ip-26-0-166-244:897243:897637 [3] NCCL INFO Channel 06/0 : 43[3] -> 44[4] via P2P/IPC
+ip-26-0-166-125:762819:763221 [4] NCCL INFO Channel 14/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-166-244:897242:897639 [2] NCCL INFO Channel 10/0 : 42[2] -> 43[3] via P2P/IPC
+ip-26-0-167-9:1039540:1039935 [3] NCCL INFO Channel 13/0 : 59[3] -> 60[4] via P2P/IPC
+ip-26-0-166-244:897246:897641 [6] NCCL INFO Channel 02/0 : 46[6] -> 47[7] via P2P/IPC
+ip-26-0-167-9:1039539:1039938 [2] NCCL INFO Channel 14/0 : 58[2] -> 59[3] via P2P/IPC
+ip-26-0-162-46:702882:703284 [2] NCCL INFO Channel 13/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-244:897244:897636 [4] NCCL INFO Channel 02/0 : 44[4] -> 45[5] via P2P/IPC
+ip-26-0-166-125:762818:763220 [3] NCCL INFO Channel 15/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-166-125:762817:763223 [2] NCCL INFO Channel 10/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-166-36:104432:104827 [3] NCCL INFO Channel 06/0 : 51[3] -> 52[4] via P2P/IPC
+ip-26-0-166-125:762819:763221 [4] NCCL INFO Channel 15/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-165-213:9625:10020 [0] NCCL INFO Connected all rings
+ip-26-0-166-244:897245:897638 [5] NCCL INFO Channel 05/0 : 45[5] -> 46[6] via P2P/IPC
+ip-26-0-165-213:9625:10020 [0] NCCL INFO Channel 02/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-166-36:104434:104828 [5] NCCL INFO Channel 05/0 : 45[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104432:104827 [3] NCCL INFO Channel 07/0 : 51[3] -> 52[4] via P2P/IPC
+ip-26-0-166-36:104434:104828 [5] NCCL INFO Channel 13/0 : 45[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-167-9:1039542:1039936 [5] NCCL INFO Channel 11/0 : 61[5] -> 62[6] via P2P/IPC
+ip-26-0-166-36:104434:104828 [5] NCCL INFO Channel 05/0 : 53[5] -> 61[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897243:897637 [3] NCCL INFO Channel 07/0 : 43[3] -> 44[4] via P2P/IPC
+ip-26-0-166-244:897242:897639 [2] NCCL INFO Channel 12/0 : 42[2] -> 43[3] via P2P/IPC
+ip-26-0-167-9:1039543:1039933 [6] NCCL INFO Channel 11/0 : 62[6] -> 63[7] via P2P/IPC
+ip-26-0-167-9:1039540:1039935 [3] NCCL INFO Channel 14/0 : 59[3] -> 60[4] via P2P/IPC
+ip-26-0-162-46:702881:703280 [1] NCCL INFO Channel 14/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-166-36:104432:104827 [3] NCCL INFO Channel 09/0 : 51[3] -> 52[4] via P2P/IPC
+ip-26-0-166-244:897246:897641 [6] NCCL INFO Channel 03/0 : 46[6] -> 47[7] via P2P/IPC
+ip-26-0-166-125:762817:763223 [2] NCCL INFO Channel 12/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-166-244:897244:897636 [4] NCCL INFO Channel 03/0 : 44[4] -> 45[5] via P2P/IPC
+ip-26-0-166-15:127295:127686 [5] NCCL INFO Connected all rings
+ip-26-0-167-9:1039539:1039938 [2] NCCL INFO Channel 15/0 : 58[2] -> 59[3] via P2P/IPC
+ip-26-0-166-36:104436:104824 [7] NCCL INFO Channel 07/0 : 47[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624949:625344 [6] NCCL INFO Channel 03/0 : 38[6] -> 39[7] via P2P/IPC
+ip-26-0-166-36:104429:104829 [0] NCCL INFO Channel 00/0 : 40[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104436:104824 [7] NCCL INFO Channel 15/0 : 47[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104429:104829 [0] NCCL INFO Channel 08/0 : 40[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624945:625340 [2] NCCL INFO Channel 05/0 : 34[2] -> 35[3] via P2P/IPC
+ip-26-0-166-36:104436:104824 [7] NCCL INFO Channel 07/0 : 55[7] -> 63[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762817:763223 [2] NCCL INFO Channel 13/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-166-36:104429:104829 [0] NCCL INFO Channel 00/0 : 48[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104432:104827 [3] NCCL INFO Channel 10/0 : 51[3] -> 52[4] via P2P/IPC
+ip-26-0-166-125:762819:763221 [4] NCCL INFO Channel 04/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897245:897638 [5] NCCL INFO Channel 07/0 : 45[5] -> 46[6] via P2P/IPC
+ip-26-0-166-125:762819:763221 [4] NCCL INFO Channel 12/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762819:763221 [4] NCCL INFO Channel 04/0 : 20[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127292:127692 [2] NCCL INFO Channel 02/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-166-214:624949:625344 [6] NCCL INFO Channel 04/0 : 38[6] -> 39[7] via P2P/IPC
+ip-26-0-166-125:762817:763223 [2] NCCL INFO Channel 14/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-166-36:104432:104827 [3] NCCL INFO Channel 11/0 : 51[3] -> 52[4] via P2P/IPC
+ip-26-0-166-244:897243:897637 [3] NCCL INFO Channel 09/0 : 43[3] -> 44[4] via P2P/IPC
+ip-26-0-166-244:897242:897639 [2] NCCL INFO Channel 13/0 : 42[2] -> 43[3] via P2P/IPC
+ip-26-0-166-244:897246:897641 [6] NCCL INFO Channel 04/0 : 46[6] -> 47[7] via P2P/IPC
+ip-26-0-165-213:9625:10020 [0] NCCL INFO Channel 03/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-166-244:897244:897636 [4] NCCL INFO Channel 04/0 : 44[4] -> 45[5] via P2P/IPC
+ip-26-0-166-15:127291:127690 [1] NCCL INFO Channel 13/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-166-214:624945:625340 [2] NCCL INFO Channel 06/0 : 34[2] -> 35[3] via P2P/IPC
+ip-26-0-166-214:624943:625338 [0] NCCL INFO Channel 06/0 : 32[0] -> 39[7] via P2P/IPC
+ip-26-0-166-36:104432:104827 [3] NCCL INFO Channel 13/0 : 51[3] -> 52[4] via P2P/IPC
+ip-26-0-167-9:1039542:1039936 [5] NCCL INFO Channel 12/0 : 61[5] -> 62[6] via P2P/IPC
+ip-26-0-167-9:1039543:1039933 [6] NCCL INFO Channel 12/0 : 62[6] -> 63[7] via P2P/IPC
+ip-26-0-167-9:1039540:1039935 [3] NCCL INFO Channel 15/0 : 59[3] -> 60[4] via P2P/IPC
+ip-26-0-166-125:762817:763223 [2] NCCL INFO Channel 15/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-166-214:624949:625344 [6] NCCL INFO Channel 05/0 : 38[6] -> 39[7] via P2P/IPC
+ip-26-0-166-36:104432:104827 [3] NCCL INFO Channel 14/0 : 51[3] -> 52[4] via P2P/IPC
+ip-26-0-166-244:897245:897638 [5] NCCL INFO Channel 09/0 : 45[5] -> 46[6] via P2P/IPC
+ip-26-0-162-46:702887:703282 [7] NCCL INFO Connected all rings
+ip-26-0-162-46:702882:703284 [2] NCCL INFO Channel 14/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-214:624943:625338 [0] NCCL INFO Channel 07/0 : 32[0] -> 39[7] via P2P/IPC
+ip-26-0-166-36:104432:104827 [3] NCCL INFO Channel 15/0 : 51[3] -> 52[4] via P2P/IPC
+ip-26-0-166-214:624945:625340 [2] NCCL INFO Channel 07/0 : 34[2] -> 35[3] via P2P/IPC
+ip-26-0-166-15:127292:127692 [2] NCCL INFO Channel 04/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-166-244:897243:897637 [3] NCCL INFO Channel 10/0 : 43[3] -> 44[4] via P2P/IPC
+ip-26-0-166-244:897242:897639 [2] NCCL INFO Channel 14/0 : 42[2] -> 43[3] via P2P/IPC
+ip-26-0-166-244:897246:897641 [6] NCCL INFO Channel 05/0 : 46[6] -> 47[7] via P2P/IPC
+ip-26-0-166-244:897244:897636 [4] NCCL INFO Channel 06/0 : 44[4] -> 45[5] via P2P/IPC
+ip-26-0-166-125:762818:763220 [3] NCCL INFO Channel 03/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762818:763220 [3] NCCL INFO Channel 11/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762818:763220 [3] NCCL INFO Channel 03/0 : 19[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-167-9:1039542:1039936 [5] NCCL INFO Channel 13/0 : 61[5] -> 62[6] via P2P/IPC
+ip-26-0-167-9:1039543:1039933 [6] NCCL INFO Channel 13/0 : 62[6] -> 63[7] via P2P/IPC
+ip-26-0-166-15:127291:127690 [1] NCCL INFO Channel 14/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-165-213:9625:10020 [0] NCCL INFO Channel 04/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-166-244:897245:897638 [5] NCCL INFO Channel 10/0 : 45[5] -> 46[6] via P2P/IPC
+ip-26-0-166-36:104433:104825 [4] NCCL INFO Channel 04/0 : 44[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104433:104825 [4] NCCL INFO Channel 12/0 : 44[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104433:104825 [4] NCCL INFO Channel 04/0 : 52[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127293:127689 [3] NCCL INFO Channel 01/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-166-244:897243:897637 [3] NCCL INFO Channel 11/0 : 43[3] -> 44[4] via P2P/IPC
+ip-26-0-166-244:897242:897639 [2] NCCL INFO Channel 15/0 : 42[2] -> 43[3] via P2P/IPC
+ip-26-0-166-244:897244:897636 [4] NCCL INFO Channel 07/0 : 44[4] -> 45[5] via P2P/IPC
+ip-26-0-166-244:897246:897641 [6] NCCL INFO Channel 06/0 : 46[6] -> 47[7] via P2P/IPC
+ip-26-0-167-9:1039542:1039936 [5] NCCL INFO Channel 15/0 : 61[5] -> 62[6] via P2P/IPC
+ip-26-0-166-15:127292:127692 [2] NCCL INFO Channel 05/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-166-214:624948:625343 [5] NCCL INFO Channel 02/0 : 37[5] -> 38[6] via P2P/IPC
+ip-26-0-167-9:1039543:1039933 [6] NCCL INFO Channel 14/0 : 62[6] -> 63[7] via P2P/IPC
+ip-26-0-166-244:897245:897638 [5] NCCL INFO Channel 11/0 : 45[5] -> 46[6] via P2P/IPC
+ip-26-0-166-214:624943:625338 [0] NCCL INFO Channel 13/0 : 32[0] -> 39[7] via P2P/IPC
+ip-26-0-165-213:9625:10020 [0] NCCL INFO Channel 05/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-162-46:702881:703280 [1] NCCL INFO Channel 15/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-162-46:702882:703284 [2] NCCL INFO Channel 15/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-15:127291:127690 [1] NCCL INFO Channel 15/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-166-214:624949:625344 [6] NCCL INFO Channel 06/0 : 38[6] -> 39[7] via P2P/IPC
+ip-26-0-166-244:897243:897637 [3] NCCL INFO Channel 13/0 : 43[3] -> 44[4] via P2P/IPC
+ip-26-0-166-244:897244:897636 [4] NCCL INFO Channel 09/0 : 44[4] -> 45[5] via P2P/IPC
+ip-26-0-166-214:624948:625343 [5] NCCL INFO Channel 03/0 : 37[5] -> 38[6] via P2P/IPC
+ip-26-0-166-244:897246:897641 [6] NCCL INFO Channel 09/0 : 46[6] -> 47[7] via P2P/IPC
+ip-26-0-166-15:127293:127689 [3] NCCL INFO Channel 02/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-162-46:702880:703277 [0] NCCL INFO Channel 05/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-166-244:897245:897638 [5] NCCL INFO Channel 12/0 : 45[5] -> 46[6] via P2P/IPC
+ip-26-0-166-214:624943:625338 [0] NCCL INFO Channel 14/0 : 32[0] -> 39[7] via P2P/IPC
+ip-26-0-166-244:897241:897635 [1] NCCL INFO Channel 09/0 : 33[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897241:897635 [1] NCCL INFO Channel 01/0 : 41[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-167-9:1039541:1039937 [4] NCCL INFO Channel 01/0 : 60[4] -> 61[5] via P2P/IPC
+ip-26-0-162-46:702881:703280 [1] NCCL INFO Channel 09/0 : 1[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897241:897635 [1] NCCL INFO Channel 09/0 : 41[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-167-9:1039543:1039933 [6] NCCL INFO Channel 06/0 : 54[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127292:127692 [2] NCCL INFO Channel 06/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-166-214:624949:625344 [6] NCCL INFO Channel 09/0 : 38[6] -> 39[7] via P2P/IPC
+ip-26-0-167-9:1039541:1039937 [4] NCCL INFO Channel 02/0 : 60[4] -> 61[5] via P2P/IPC
+ip-26-0-166-244:897243:897637 [3] NCCL INFO Channel 14/0 : 43[3] -> 44[4] via P2P/IPC
+ip-26-0-167-9:1039543:1039933 [6] NCCL INFO Channel 14/0 : 30[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9625:10020 [0] NCCL INFO Channel 06/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-166-244:897244:897636 [4] NCCL INFO Channel 10/0 : 44[4] -> 45[5] via P2P/IPC
+ip-26-0-166-244:897246:897641 [6] NCCL INFO Channel 10/0 : 46[6] -> 47[7] via P2P/IPC
+ip-26-0-167-9:1039543:1039933 [6] NCCL INFO Channel 14/0 : 62[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897245:897638 [5] NCCL INFO Channel 13/0 : 45[5] -> 46[6] via P2P/IPC
+ip-26-0-166-214:624945:625340 [2] NCCL INFO Channel 09/0 : 34[2] -> 35[3] via P2P/IPC
+ip-26-0-166-214:624948:625343 [5] NCCL INFO Channel 04/0 : 37[5] -> 38[6] via P2P/IPC
+ip-26-0-166-15:127290:127685 [0] NCCL INFO Channel 05/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-166-214:624943:625338 [0] NCCL INFO Channel 15/0 : 32[0] -> 39[7] via P2P/IPC
+ip-26-0-162-46:702880:703277 [0] NCCL INFO Channel 06/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-165-213:9626:10025 [1] NCCL INFO Channel 01/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-167-9:1039541:1039937 [4] NCCL INFO Channel 03/0 : 60[4] -> 61[5] via P2P/IPC
+ip-26-0-166-15:127293:127689 [3] NCCL INFO Channel 03/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-166-125:762816:763218 [1] NCCL INFO Connected all rings
+ip-26-0-166-214:624949:625344 [6] NCCL INFO Channel 10/0 : 38[6] -> 39[7] via P2P/IPC
+ip-26-0-166-15:127292:127692 [2] NCCL INFO Channel 07/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-166-244:897243:897637 [3] NCCL INFO Channel 15/0 : 43[3] -> 44[4] via P2P/IPC
+ip-26-0-166-244:897244:897636 [4] NCCL INFO Channel 11/0 : 44[4] -> 45[5] via P2P/IPC
+ip-26-0-166-244:897246:897641 [6] NCCL INFO Channel 11/0 : 46[6] -> 47[7] via P2P/IPC
+ip-26-0-162-46:702880:703277 [0] NCCL INFO Channel 07/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-166-244:897245:897638 [5] NCCL INFO Channel 15/0 : 45[5] -> 46[6] via P2P/IPC
+ip-26-0-167-9:1039541:1039937 [4] NCCL INFO Channel 04/0 : 60[4] -> 61[5] via P2P/IPC
+ip-26-0-166-214:624945:625340 [2] NCCL INFO Channel 10/0 : 34[2] -> 35[3] via P2P/IPC
+ip-26-0-166-214:624948:625343 [5] NCCL INFO Channel 05/0 : 37[5] -> 38[6] via P2P/IPC
+ip-26-0-166-214:624947:625339 [4] NCCL INFO Channel 01/0 : 36[4] -> 37[5] via P2P/IPC
+ip-26-0-167-9:1039541:1039937 [4] NCCL INFO Channel 06/0 : 60[4] -> 61[5] via P2P/IPC
+ip-26-0-166-15:127290:127685 [0] NCCL INFO Channel 06/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-165-213:9625:10020 [0] NCCL INFO Channel 07/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-166-244:897244:897636 [4] NCCL INFO Channel 12/0 : 44[4] -> 45[5] via P2P/IPC
+ip-26-0-166-244:897246:897641 [6] NCCL INFO Channel 12/0 : 46[6] -> 47[7] via P2P/IPC
+ip-26-0-166-214:624949:625344 [6] NCCL INFO Channel 11/0 : 38[6] -> 39[7] via P2P/IPC
+ip-26-0-166-15:127293:127689 [3] NCCL INFO Channel 05/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-162-46:702880:703277 [0] NCCL INFO Channel 13/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-167-9:1039541:1039937 [4] NCCL INFO Channel 07/0 : 60[4] -> 61[5] via P2P/IPC
+ip-26-0-166-214:624945:625340 [2] NCCL INFO Channel 12/0 : 34[2] -> 35[3] via P2P/IPC
+ip-26-0-166-244:897242:897639 [2] NCCL INFO Channel 10/0 : 34[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624948:625343 [5] NCCL INFO Channel 07/0 : 37[5] -> 38[6] via P2P/IPC
+ip-26-0-166-214:624947:625339 [4] NCCL INFO Channel 02/0 : 36[4] -> 37[5] via P2P/IPC
+ip-26-0-166-244:897242:897639 [2] NCCL INFO Channel 02/0 : 42[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127292:127692 [2] NCCL INFO Channel 09/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-166-244:897246:897641 [6] NCCL INFO Channel 13/0 : 46[6] -> 47[7] via P2P/IPC
+ip-26-0-166-244:897242:897639 [2] NCCL INFO Channel 10/0 : 42[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9626:10025 [1] NCCL INFO Channel 03/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-166-244:897244:897636 [4] NCCL INFO Channel 14/0 : 44[4] -> 45[5] via P2P/IPC
+ip-26-0-166-214:624949:625344 [6] NCCL INFO Channel 12/0 : 38[6] -> 39[7] via P2P/IPC
+ip-26-0-167-9:1039541:1039937 [4] NCCL INFO Channel 09/0 : 60[4] -> 61[5] via P2P/IPC
+ip-26-0-166-244:897246:897641 [6] NCCL INFO Channel 14/0 : 46[6] -> 47[7] via P2P/IPC
+ip-26-0-166-244:897244:897636 [4] NCCL INFO Channel 15/0 : 44[4] -> 45[5] via P2P/IPC
+ip-26-0-162-46:702880:703277 [0] NCCL INFO Channel 14/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-165-213:9625:10020 [0] NCCL INFO Channel 10/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-166-214:624945:625340 [2] NCCL INFO Channel 13/0 : 34[2] -> 35[3] via P2P/IPC
+ip-26-0-167-9:1039541:1039937 [4] NCCL INFO Channel 10/0 : 60[4] -> 61[5] via P2P/IPC
+ip-26-0-166-15:127290:127685 [0] NCCL INFO Channel 07/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-166-214:624948:625343 [5] NCCL INFO Channel 09/0 : 37[5] -> 38[6] via P2P/IPC
+ip-26-0-166-214:624947:625339 [4] NCCL INFO Channel 03/0 : 36[4] -> 37[5] via P2P/IPC
+ip-26-0-166-15:127293:127689 [3] NCCL INFO Channel 06/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-166-214:624949:625344 [6] NCCL INFO Channel 13/0 : 38[6] -> 39[7] via P2P/IPC
+ip-26-0-167-9:1039541:1039937 [4] NCCL INFO Channel 11/0 : 60[4] -> 61[5] via P2P/IPC
+ip-26-0-165-213:9626:10025 [1] NCCL INFO Channel 04/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-166-15:127292:127692 [2] NCCL INFO Channel 10/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-166-125:762821:763224 [6] NCCL INFO Connected all rings
+ip-26-0-166-244:897243:897637 [3] NCCL INFO Channel 11/0 : 35[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624945:625340 [2] NCCL INFO Channel 14/0 : 34[2] -> 35[3] via P2P/IPC
+ip-26-0-167-9:1039541:1039937 [4] NCCL INFO Channel 12/0 : 60[4] -> 61[5] via P2P/IPC
+ip-26-0-166-244:897246:897641 [6] NCCL INFO Channel 14/0 : 38[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897243:897637 [3] NCCL INFO Channel 03/0 : 43[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897245:897638 [5] NCCL INFO Channel 13/0 : 37[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624948:625343 [5] NCCL INFO Channel 10/0 : 37[5] -> 38[6] via P2P/IPC
+ip-26-0-166-214:624947:625339 [4] NCCL INFO Channel 04/0 : 36[4] -> 37[5] via P2P/IPC
+ip-26-0-166-244:897244:897636 [4] NCCL INFO Channel 12/0 : 36[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624949:625344 [6] NCCL INFO Channel 14/0 : 38[6] -> 39[7] via P2P/IPC
+ip-26-0-166-244:897246:897641 [6] NCCL INFO Channel 06/0 : 46[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897243:897637 [3] NCCL INFO Channel 11/0 : 43[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897245:897638 [5] NCCL INFO Channel 05/0 : 45[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897244:897636 [4] NCCL INFO Channel 04/0 : 44[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897246:897641 [6] NCCL INFO Channel 14/0 : 46[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897245:897638 [5] NCCL INFO Channel 13/0 : 45[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897244:897636 [4] NCCL INFO Channel 12/0 : 44[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-167-9:1039541:1039937 [4] NCCL INFO Channel 14/0 : 60[4] -> 61[5] via P2P/IPC
+ip-26-0-167-9:1039541:1039937 [4] NCCL INFO Channel 15/0 : 60[4] -> 61[5] via P2P/IPC
+ip-26-0-165-213:9625:10020 [0] NCCL INFO Channel 11/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-166-214:624945:625340 [2] NCCL INFO Channel 15/0 : 34[2] -> 35[3] via P2P/IPC
+ip-26-0-166-214:624948:625343 [5] NCCL INFO Channel 11/0 : 37[5] -> 38[6] via P2P/IPC
+ip-26-0-166-36:104435:104826 [6] NCCL INFO Channel 06/0 : 38[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624947:625339 [4] NCCL INFO Channel 06/0 : 36[4] -> 37[5] via P2P/IPC
+ip-26-0-165-213:9626:10025 [1] NCCL INFO Channel 05/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-162-46:702880:703277 [0] NCCL INFO Channel 15/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-167-9:1039540:1039935 [3] NCCL INFO Channel 03/0 : 51[3] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-167-9:1039542:1039936 [5] NCCL INFO Channel 05/0 : 53[5] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-167-9:1039541:1039937 [4] NCCL INFO Channel 04/0 : 52[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624948:625343 [5] NCCL INFO Channel 12/0 : 37[5] -> 38[6] via P2P/IPC
+ip-26-0-166-214:624947:625339 [4] NCCL INFO Channel 07/0 : 36[4] -> 37[5] via P2P/IPC
+ip-26-0-162-46:702886:703281 [6] NCCL INFO Channel 01/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-213:9625:10020 [0] NCCL INFO Channel 12/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-166-214:624944:625341 [1] NCCL INFO Channel 09/0 : 33[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-167-9:1039541:1039937 [4] NCCL INFO Channel 12/0 : 28[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-167-9:1039542:1039936 [5] NCCL INFO Channel 13/0 : 29[5] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897240:897640 [0] NCCL INFO Channel 08/0 : 32[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897247:897642 [7] NCCL INFO Channel 15/0 : 39[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104433:104825 [4] NCCL INFO Channel 04/0 : 36[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104434:104828 [5] NCCL INFO Channel 05/0 : 37[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624945:625340 [2] NCCL INFO Channel 10/0 : 34[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127293:127689 [3] NCCL INFO Channel 07/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-167-9:1039541:1039937 [4] NCCL INFO Channel 12/0 : 60[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897247:897642 [7] NCCL INFO Channel 07/0 : 47[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897240:897640 [0] NCCL INFO Channel 00/0 : 40[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-167-9:1039542:1039936 [5] NCCL INFO Channel 13/0 : 61[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127290:127685 [0] NCCL INFO Channel 13/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-166-244:897247:897642 [7] NCCL INFO Channel 15/0 : 47[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624948:625343 [5] NCCL INFO Channel 13/0 : 37[5] -> 38[6] via P2P/IPC
+ip-26-0-166-214:624947:625339 [4] NCCL INFO Channel 09/0 : 36[4] -> 37[5] via P2P/IPC
+ip-26-0-165-213:9626:10025 [1] NCCL INFO Channel 06/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-166-244:897240:897640 [0] NCCL INFO Channel 08/0 : 40[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624944:625341 [1] NCCL INFO Channel 01/0 : 17[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624945:625340 [2] NCCL INFO Channel 02/0 : 18[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127292:127692 [2] NCCL INFO Channel 12/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-166-214:624944:625341 [1] NCCL INFO Channel 01/0 : 33[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624945:625340 [2] NCCL INFO Channel 02/0 : 34[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762816:763218 [1] NCCL INFO Channel 01/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-162-46:702886:703281 [6] NCCL INFO Channel 02/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-214:624948:625343 [5] NCCL INFO Channel 15/0 : 37[5] -> 38[6] via P2P/IPC
+ip-26-0-166-214:624947:625339 [4] NCCL INFO Channel 10/0 : 36[4] -> 37[5] via P2P/IPC
+ip-26-0-166-36:104431:104831 [2] NCCL INFO Connected all rings
+ip-26-0-166-125:762816:763218 [1] NCCL INFO Channel 03/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-166-15:127293:127689 [3] NCCL INFO Channel 09/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-165-213:9625:10020 [0] NCCL INFO Channel 13/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-166-15:127292:127692 [2] NCCL INFO Channel 13/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-166-15:127290:127685 [0] NCCL INFO Channel 14/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-165-213:9626:10025 [1] NCCL INFO Channel 07/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-166-214:624947:625339 [4] NCCL INFO Channel 11/0 : 36[4] -> 37[5] via P2P/IPC
+ip-26-0-166-125:762816:763218 [1] NCCL INFO Channel 04/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-166-214:624949:625344 [6] NCCL INFO Channel 14/0 : 38[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624947:625339 [4] NCCL INFO Channel 12/0 : 36[4] -> 37[5] via P2P/IPC
+ip-26-0-166-15:127296:127688 [6] NCCL INFO Channel 01/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-166-125:762816:763218 [1] NCCL INFO Channel 05/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-166-214:624949:625344 [6] NCCL INFO Channel 06/0 : 22[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897246:897641 [6] NCCL INFO Channel 14/0 : 30[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127292:127692 [2] NCCL INFO Channel 14/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-166-15:127293:127689 [3] NCCL INFO Channel 10/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-166-214:624949:625344 [6] NCCL INFO Channel 06/0 : 38[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127290:127685 [0] NCCL INFO Channel 15/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-166-214:624947:625339 [4] NCCL INFO Channel 14/0 : 36[4] -> 37[5] via P2P/IPC
+ip-26-0-165-213:9625:10020 [0] NCCL INFO Channel 14/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-166-125:762816:763218 [1] NCCL INFO Channel 06/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-166-214:624947:625339 [4] NCCL INFO Channel 15/0 : 36[4] -> 37[5] via P2P/IPC
+ip-26-0-166-214:624943:625338 [0] NCCL INFO Channel 08/0 : 32[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-46:702885:703278 [5] NCCL INFO Connected all rings
+ip-26-0-166-214:624950:625342 [7] NCCL INFO Channel 15/0 : 39[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104435:104826 [6] NCCL INFO Channel 06/0 : 54[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127296:127688 [6] NCCL INFO Channel 02/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-165-213:9626:10025 [1] NCCL INFO Channel 09/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-166-125:762816:763218 [1] NCCL INFO Channel 07/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-166-214:624943:625338 [0] NCCL INFO Channel 00/0 : 16[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897240:897640 [0] NCCL INFO Channel 08/0 : 24[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624950:625342 [7] NCCL INFO Channel 07/0 : 23[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897247:897642 [7] NCCL INFO Channel 15/0 : 31[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624943:625338 [0] NCCL INFO Channel 00/0 : 32[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624950:625342 [7] NCCL INFO Channel 07/0 : 39[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127292:127692 [2] NCCL INFO Channel 15/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-166-214:624946:625337 [3] NCCL INFO Channel 11/0 : 35[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127293:127689 [3] NCCL INFO Channel 11/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-166-214:624948:625343 [5] NCCL INFO Channel 13/0 : 37[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624947:625339 [4] NCCL INFO Channel 12/0 : 36[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762816:763218 [1] NCCL INFO Channel 09/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-166-214:624946:625337 [3] NCCL INFO Channel 03/0 : 19[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624948:625343 [5] NCCL INFO Channel 05/0 : 21[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897245:897638 [5] NCCL INFO Channel 13/0 : 29[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624946:625337 [3] NCCL INFO Channel 03/0 : 35[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127296:127688 [6] NCCL INFO Channel 03/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-166-214:624947:625339 [4] NCCL INFO Channel 04/0 : 20[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897244:897636 [4] NCCL INFO Channel 12/0 : 28[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9625:10020 [0] NCCL INFO Channel 15/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-166-214:624948:625343 [5] NCCL INFO Channel 05/0 : 37[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624947:625339 [4] NCCL INFO Channel 04/0 : 36[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762816:763218 [1] NCCL INFO Channel 11/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-166-15:127293:127689 [3] NCCL INFO Channel 13/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-165-213:9626:10025 [1] NCCL INFO Channel 11/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-166-36:104434:104828 [5] NCCL INFO Channel 05/0 : 53[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104433:104825 [4] NCCL INFO Channel 04/0 : 52[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127291:127690 [1] NCCL INFO Channel 01/0 : 17[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127296:127688 [6] NCCL INFO Channel 04/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-166-125:762816:763218 [1] NCCL INFO Channel 12/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-162-46:702886:703281 [6] NCCL INFO Channel 03/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-36:104431:104831 [2] NCCL INFO Channel 01/0 : 50[2] -> 51[3] via P2P/IPC
+ip-26-0-166-15:127293:127689 [3] NCCL INFO Channel 14/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-166-125:762821:763224 [6] NCCL INFO Channel 01/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-166-125:762816:763218 [1] NCCL INFO Channel 13/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-166-36:104431:104831 [2] NCCL INFO Channel 02/0 : 50[2] -> 51[3] via P2P/IPC
+ip-26-0-166-15:127294:127687 [4] NCCL INFO Channel 01/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-166-15:127296:127688 [6] NCCL INFO Channel 05/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-162-46:702886:703281 [6] NCCL INFO Channel 04/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-36:104431:104831 [2] NCCL INFO Channel 04/0 : 50[2] -> 51[3] via P2P/IPC
+ip-26-0-166-125:762821:763224 [6] NCCL INFO Channel 02/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-166-125:762816:763218 [1] NCCL INFO Channel 14/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-166-15:127293:127689 [3] NCCL INFO Channel 15/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-166-36:104431:104831 [2] NCCL INFO Channel 05/0 : 50[2] -> 51[3] via P2P/IPC
+ip-26-0-162-46:702886:703281 [6] NCCL INFO Channel 05/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-125:762821:763224 [6] NCCL INFO Channel 03/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-166-36:104431:104831 [2] NCCL INFO Channel 06/0 : 50[2] -> 51[3] via P2P/IPC
+ip-26-0-166-125:762816:763218 [1] NCCL INFO Channel 15/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-166-15:127294:127687 [4] NCCL INFO Channel 02/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-166-15:127296:127688 [6] NCCL INFO Channel 06/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-166-36:104431:104831 [2] NCCL INFO Channel 07/0 : 50[2] -> 51[3] via P2P/IPC
+ip-26-0-162-46:702886:703281 [6] NCCL INFO Channel 06/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-125:762821:763224 [6] NCCL INFO Channel 04/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-166-125:762815:763225 [0] NCCL INFO Channel 05/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-166-36:104431:104831 [2] NCCL INFO Channel 09/0 : 50[2] -> 51[3] via P2P/IPC
+ip-26-0-166-15:127292:127692 [2] NCCL INFO Channel 02/0 : 18[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762821:763224 [6] NCCL INFO Channel 05/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-166-15:127294:127687 [4] NCCL INFO Channel 03/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-166-15:127296:127688 [6] NCCL INFO Channel 09/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-162-46:702886:703281 [6] NCCL INFO Channel 09/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-125:762816:763218 [1] NCCL INFO Channel 01/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104431:104831 [2] NCCL INFO Channel 10/0 : 50[2] -> 51[3] via P2P/IPC
+ip-26-0-166-125:762817:763223 [2] NCCL INFO Channel 02/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762816:763218 [1] NCCL INFO Channel 09/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762817:763223 [2] NCCL INFO Channel 10/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762816:763218 [1] NCCL INFO Channel 01/0 : 17[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762817:763223 [2] NCCL INFO Channel 02/0 : 18[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104431:104831 [2] NCCL INFO Channel 12/0 : 50[2] -> 51[3] via P2P/IPC
+ip-26-0-166-125:762821:763224 [6] NCCL INFO Channel 06/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-166-15:127294:127687 [4] NCCL INFO Channel 04/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-166-125:762815:763225 [0] NCCL INFO Channel 06/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-166-15:127296:127688 [6] NCCL INFO Channel 10/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-162-46:702886:703281 [6] NCCL INFO Channel 10/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-213:9626:10025 [1] NCCL INFO Channel 12/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-166-36:104431:104831 [2] NCCL INFO Channel 13/0 : 50[2] -> 51[3] via P2P/IPC
+ip-26-0-166-125:762821:763224 [6] NCCL INFO Channel 09/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-166-15:127292:127692 [2] NCCL INFO Channel 10/0 : 10[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104431:104831 [2] NCCL INFO Channel 14/0 : 50[2] -> 51[3] via P2P/IPC
+ip-26-0-166-125:762815:763225 [0] NCCL INFO Channel 07/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-166-15:127291:127690 [1] NCCL INFO Channel 09/0 : 9[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-46:702886:703281 [6] NCCL INFO Channel 11/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-15:127292:127692 [2] NCCL INFO Channel 10/0 : 26[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127291:127690 [1] NCCL INFO Channel 09/0 : 25[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127294:127687 [4] NCCL INFO Channel 06/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-166-15:127296:127688 [6] NCCL INFO Channel 11/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-166-15:127295:127686 [5] NCCL INFO Channel 01/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-166-36:104431:104831 [2] NCCL INFO Channel 15/0 : 50[2] -> 51[3] via P2P/IPC
+ip-26-0-166-125:762821:763224 [6] NCCL INFO Channel 10/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-166-125:762815:763225 [0] NCCL INFO Channel 13/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-162-46:702886:703281 [6] NCCL INFO Channel 12/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-15:127294:127687 [4] NCCL INFO Channel 07/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-166-15:127296:127688 [6] NCCL INFO Channel 12/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-166-15:127295:127686 [5] NCCL INFO Channel 02/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-166-125:762821:763224 [6] NCCL INFO Channel 11/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-166-36:104432:104827 [3] NCCL INFO Channel 03/0 : 43[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762815:763225 [0] NCCL INFO Channel 14/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-166-36:104431:104831 [2] NCCL INFO Channel 02/0 : 42[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-46:702886:703281 [6] NCCL INFO Channel 13/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-36:104430:104830 [1] NCCL INFO Channel 01/0 : 41[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104432:104827 [3] NCCL INFO Channel 11/0 : 43[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104431:104831 [2] NCCL INFO Channel 10/0 : 42[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104432:104827 [3] NCCL INFO Channel 03/0 : 51[3] -> 59[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104430:104830 [1] NCCL INFO Channel 09/0 : 41[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104431:104831 [2] NCCL INFO Channel 02/0 : 50[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104430:104830 [1] NCCL INFO Channel 01/0 : 49[1] -> 57[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762821:763224 [6] NCCL INFO Channel 12/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-166-15:127296:127688 [6] NCCL INFO Channel 13/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-166-15:127294:127687 [4] NCCL INFO Channel 09/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-166-125:762815:763225 [0] NCCL INFO Channel 15/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-166-15:127295:127686 [5] NCCL INFO Channel 03/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-166-244:897243:897637 [3] NCCL INFO Channel 11/0 : 27[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897242:897639 [2] NCCL INFO Channel 10/0 : 26[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-46:702886:703281 [6] NCCL INFO Channel 14/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-36:104432:104827 [3] NCCL INFO Channel 03/0 : 35[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897241:897635 [1] NCCL INFO Channel 09/0 : 25[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-167-9:1039540:1039935 [3] NCCL INFO Channel 11/0 : 27[3] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762821:763224 [6] NCCL INFO Channel 13/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-167-9:1039540:1039935 [3] NCCL INFO Channel 11/0 : 59[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897242:897639 [2] NCCL INFO Channel 10/0 : 42[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104432:104827 [3] NCCL INFO Channel 03/0 : 51[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897241:897635 [1] NCCL INFO Channel 09/0 : 41[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127296:127688 [6] NCCL INFO Channel 14/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-166-15:127294:127687 [4] NCCL INFO Channel 10/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-166-15:127295:127686 [5] NCCL INFO Channel 04/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-166-125:762821:763224 [6] NCCL INFO Channel 14/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-166-15:127294:127687 [4] NCCL INFO Channel 11/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-166-15:127295:127686 [5] NCCL INFO Channel 05/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-166-125:762820:763219 [5] NCCL INFO Channel 05/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-46:702885:703278 [5] NCCL INFO Channel 01/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-166-125:762820:763219 [5] NCCL INFO Channel 13/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-167-9:1039538:1039934 [1] NCCL INFO Connected all rings
+ip-26-0-166-125:762821:763224 [6] NCCL INFO Channel 06/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762821:763224 [6] NCCL INFO Channel 14/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762820:763219 [5] NCCL INFO Channel 05/0 : 21[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762821:763224 [6] NCCL INFO Channel 06/0 : 22[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127294:127687 [4] NCCL INFO Channel 12/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-166-15:127295:127686 [5] NCCL INFO Channel 07/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-162-46:702885:703278 [5] NCCL INFO Channel 02/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-166-15:127294:127687 [4] NCCL INFO Channel 14/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-166-15:127295:127686 [5] NCCL INFO Channel 09/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-162-46:702885:703278 [5] NCCL INFO Channel 03/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-166-15:127294:127687 [4] NCCL INFO Channel 15/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-162-46:702885:703278 [5] NCCL INFO Channel 04/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-166-15:127295:127686 [5] NCCL INFO Channel 10/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-166-125:762822:763222 [7] NCCL INFO Channel 07/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762815:763225 [0] NCCL INFO Channel 00/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-46:702885:703278 [5] NCCL INFO Channel 05/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-166-125:762822:763222 [7] NCCL INFO Channel 15/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127295:127686 [5] NCCL INFO Channel 11/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-166-125:762815:763225 [0] NCCL INFO Channel 08/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762822:763222 [7] NCCL INFO Channel 07/0 : 23[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9626:10025 [1] NCCL INFO Channel 13/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-166-125:762815:763225 [0] NCCL INFO Channel 00/0 : 16[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127293:127689 [3] NCCL INFO Channel 03/0 : 19[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127295:127686 [5] NCCL INFO Channel 12/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-162-46:702885:703278 [5] NCCL INFO Channel 07/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-162-46:702887:703282 [7] NCCL INFO Channel 15/0 : 7[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127293:127689 [3] NCCL INFO Channel 11/0 : 11[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127293:127689 [3] NCCL INFO Channel 11/0 : 27[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-46:702885:703278 [5] NCCL INFO Channel 09/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-166-15:127295:127686 [5] NCCL INFO Channel 13/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-166-15:127295:127686 [5] NCCL INFO Channel 15/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-162-46:702885:703278 [5] NCCL INFO Channel 10/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-166-15:127290:127685 [0] NCCL INFO Channel 00/0 : 16[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127297:127691 [7] NCCL INFO Channel 07/0 : 23[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-167-9:1039538:1039934 [1] NCCL INFO Channel 01/0 : 57[1] -> 58[2] via P2P/IPC
+ip-26-0-166-244:897243:897637 [3] NCCL INFO Channel 11/0 : 43[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127290:127685 [0] NCCL INFO Channel 08/0 : 8[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-46:702885:703278 [5] NCCL INFO Channel 11/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-166-15:127297:127691 [7] NCCL INFO Channel 15/0 : 15[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127290:127685 [0] NCCL INFO Channel 08/0 : 24[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127297:127691 [7] NCCL INFO Channel 15/0 : 31[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127294:127687 [4] NCCL INFO Channel 04/0 : 20[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127296:127688 [6] NCCL INFO Channel 06/0 : 22[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-46:702885:703278 [5] NCCL INFO Channel 12/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-166-15:127295:127686 [5] NCCL INFO Channel 05/0 : 21[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127294:127687 [4] NCCL INFO Channel 12/0 : 12[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127294:127687 [4] NCCL INFO Channel 12/0 : 28[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897240:897640 [0] NCCL INFO Channel 08/0 : 40[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9626:10025 [1] NCCL INFO Channel 14/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-166-15:127296:127688 [6] NCCL INFO Channel 14/0 : 14[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-46:702885:703278 [5] NCCL INFO Channel 13/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-166-244:897247:897642 [7] NCCL INFO Channel 15/0 : 47[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127295:127686 [5] NCCL INFO Channel 13/0 : 13[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127296:127688 [6] NCCL INFO Channel 14/0 : 30[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127295:127686 [5] NCCL INFO Channel 13/0 : 29[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-46:702885:703278 [5] NCCL INFO Channel 15/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-165-213:9626:10025 [1] NCCL INFO Channel 15/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-166-244:897244:897636 [4] NCCL INFO Channel 12/0 : 44[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897246:897641 [6] NCCL INFO Channel 14/0 : 46[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897245:897638 [5] NCCL INFO Channel 13/0 : 45[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-167-9:1039538:1039934 [1] NCCL INFO Channel 03/0 : 57[1] -> 58[2] via P2P/IPC
+ip-26-0-165-213:9625:10020 [0] NCCL INFO Channel 05/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-167-9:1039538:1039934 [1] NCCL INFO Channel 04/0 : 57[1] -> 58[2] via P2P/IPC
+ip-26-0-162-46:702886:703281 [6] NCCL INFO Channel 14/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-46:702880:703277 [0] NCCL INFO Channel 08/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-167-9:1039538:1039934 [1] NCCL INFO Channel 05/0 : 57[1] -> 58[2] via P2P/IPC
+ip-26-0-167-9:1039538:1039934 [1] NCCL INFO Channel 06/0 : 57[1] -> 58[2] via P2P/IPC
+ip-26-0-167-9:1039538:1039934 [1] NCCL INFO Channel 07/0 : 57[1] -> 58[2] via P2P/IPC
+ip-26-0-167-9:1039538:1039934 [1] NCCL INFO Channel 09/0 : 57[1] -> 58[2] via P2P/IPC
+ip-26-0-162-46:702884:703279 [4] NCCL INFO Connected all rings
+ip-26-0-165-213:9625:10020 [0] NCCL INFO Channel 06/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-167-9:1039538:1039934 [1] NCCL INFO Channel 11/0 : 57[1] -> 58[2] via P2P/IPC
+ip-26-0-167-9:1039538:1039934 [1] NCCL INFO Channel 12/0 : 57[1] -> 58[2] via P2P/IPC
+ip-26-0-167-9:1039538:1039934 [1] NCCL INFO Channel 13/0 : 57[1] -> 58[2] via P2P/IPC
+ip-26-0-165-213:9625:10020 [0] NCCL INFO Channel 07/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-167-9:1039538:1039934 [1] NCCL INFO Channel 14/0 : 57[1] -> 58[2] via P2P/IPC
+ip-26-0-167-9:1039538:1039934 [1] NCCL INFO Channel 15/0 : 57[1] -> 58[2] via P2P/IPC
+ip-26-0-167-9:1039537:1039932 [0] NCCL INFO Channel 05/0 : 56[0] -> 63[7] via P2P/IPC
+ip-26-0-165-213:9625:10020 [0] NCCL INFO Channel 13/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-167-9:1039539:1039938 [2] NCCL INFO Channel 02/0 : 50[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-167-9:1039538:1039934 [1] NCCL INFO Channel 01/0 : 49[1] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-46:702883:703283 [3] NCCL INFO Channel 01/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-167-9:1039537:1039932 [0] NCCL INFO Channel 06/0 : 56[0] -> 63[7] via P2P/IPC
+ip-26-0-167-9:1039538:1039934 [1] NCCL INFO Channel 09/0 : 25[1] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-167-9:1039539:1039938 [2] NCCL INFO Channel 10/0 : 26[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104431:104831 [2] NCCL INFO Channel 02/0 : 34[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-167-9:1039538:1039934 [1] NCCL INFO Channel 09/0 : 57[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-167-9:1039539:1039938 [2] NCCL INFO Channel 10/0 : 58[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-167-9:1039537:1039932 [0] NCCL INFO Channel 07/0 : 56[0] -> 63[7] via P2P/IPC
+ip-26-0-162-46:702883:703283 [3] NCCL INFO Channel 02/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-213:9625:10020 [0] NCCL INFO Channel 14/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-166-36:104430:104830 [1] NCCL INFO Channel 01/0 : 33[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104431:104831 [2] NCCL INFO Channel 02/0 : 50[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-167-9:1039537:1039932 [0] NCCL INFO Channel 13/0 : 56[0] -> 63[7] via P2P/IPC
+ip-26-0-162-46:702883:703283 [3] NCCL INFO Channel 03/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-36:104430:104830 [1] NCCL INFO Channel 01/0 : 49[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-167-9:1039537:1039932 [0] NCCL INFO Channel 14/0 : 56[0] -> 63[7] via P2P/IPC
+ip-26-0-162-46:702883:703283 [3] NCCL INFO Channel 05/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-167-9:1039537:1039932 [0] NCCL INFO Channel 15/0 : 56[0] -> 63[7] via P2P/IPC
+ip-26-0-162-46:702883:703283 [3] NCCL INFO Channel 06/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-167-9:1039537:1039932 [0] NCCL INFO Channel 00/0 : 48[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-167-9:1039544:1039931 [7] NCCL INFO Channel 07/0 : 55[7] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9625:10020 [0] NCCL INFO Channel 15/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-162-46:702883:703283 [3] NCCL INFO Channel 07/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-167-9:1039537:1039932 [0] NCCL INFO Channel 08/0 : 24[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-167-9:1039544:1039931 [7] NCCL INFO Channel 15/0 : 31[7] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-167-9:1039537:1039932 [0] NCCL INFO Channel 08/0 : 56[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104436:104824 [7] NCCL INFO Channel 07/0 : 39[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-167-9:1039544:1039931 [7] NCCL INFO Channel 15/0 : 63[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-46:702883:703283 [3] NCCL INFO Channel 09/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-36:104436:104824 [7] NCCL INFO Channel 07/0 : 55[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-46:702884:703279 [4] NCCL INFO Channel 01/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-166-36:104429:104829 [0] NCCL INFO Channel 00/0 : 32[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-46:702883:703283 [3] NCCL INFO Channel 10/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-36:104429:104829 [0] NCCL INFO Channel 00/0 : 48[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-46:702884:703279 [4] NCCL INFO Channel 02/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-162-46:702883:703283 [3] NCCL INFO Channel 11/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-162-46:702884:703279 [4] NCCL INFO Channel 03/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-162-46:702883:703283 [3] NCCL INFO Channel 13/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-162-46:702884:703279 [4] NCCL INFO Channel 04/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-162-46:702883:703283 [3] NCCL INFO Channel 14/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-162-46:702884:703279 [4] NCCL INFO Channel 06/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-162-46:702883:703283 [3] NCCL INFO Channel 15/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-162-46:702884:703279 [4] NCCL INFO Channel 07/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-162-46:702884:703279 [4] NCCL INFO Channel 09/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-162-46:702882:703284 [2] NCCL INFO Channel 10/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-46:702884:703279 [4] NCCL INFO Channel 10/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-162-46:702884:703279 [4] NCCL INFO Channel 11/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-162-46:702884:703279 [4] NCCL INFO Channel 12/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-162-46:702884:703279 [4] NCCL INFO Channel 14/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-162-46:702884:703279 [4] NCCL INFO Channel 15/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-162-46:702883:703283 [3] NCCL INFO Channel 11/0 : 3[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-46:702885:703278 [5] NCCL INFO Channel 13/0 : 5[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-46:702884:703279 [4] NCCL INFO Channel 12/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9632:10026 [7] NCCL INFO Connected all rings
+ip-26-0-165-213:9630:10024 [5] NCCL INFO Connected all rings
+ip-26-0-165-213:9627:10027 [2] NCCL INFO Channel 01/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-165-213:9629:10021 [4] NCCL INFO Connected all rings
+ip-26-0-165-213:9627:10027 [2] NCCL INFO Channel 02/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-165-213:9628:10023 [3] NCCL INFO Connected all rings
+ip-26-0-165-213:9627:10027 [2] NCCL INFO Channel 04/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-165-213:9627:10027 [2] NCCL INFO Channel 05/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-165-213:9631:10022 [6] NCCL INFO Channel 01/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-165-213:9627:10027 [2] NCCL INFO Channel 06/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-165-213:9631:10022 [6] NCCL INFO Channel 02/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-165-213:9627:10027 [2] NCCL INFO Channel 07/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-165-213:9631:10022 [6] NCCL INFO Channel 03/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-165-213:9627:10027 [2] NCCL INFO Channel 09/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-165-213:9631:10022 [6] NCCL INFO Channel 04/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-165-213:9627:10027 [2] NCCL INFO Channel 10/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-165-213:9631:10022 [6] NCCL INFO Channel 05/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-165-213:9627:10027 [2] NCCL INFO Channel 12/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-165-213:9631:10022 [6] NCCL INFO Channel 06/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-165-213:9630:10024 [5] NCCL INFO Channel 01/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-165-213:9627:10027 [2] NCCL INFO Channel 13/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-165-213:9630:10024 [5] NCCL INFO Channel 02/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-165-213:9627:10027 [2] NCCL INFO Channel 14/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-165-213:9631:10022 [6] NCCL INFO Channel 09/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-165-213:9630:10024 [5] NCCL INFO Channel 03/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-165-213:9627:10027 [2] NCCL INFO Channel 15/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-165-213:9631:10022 [6] NCCL INFO Channel 10/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-165-213:9630:10024 [5] NCCL INFO Channel 04/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-165-213:9631:10022 [6] NCCL INFO Channel 11/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-165-213:9630:10024 [5] NCCL INFO Channel 05/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-165-213:9629:10021 [4] NCCL INFO Channel 01/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-165-213:9626:10025 [1] NCCL INFO Channel 09/0 : 1[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-46:702881:703280 [1] NCCL INFO Channel 01/0 : 33[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762816:763218 [1] NCCL INFO Channel 01/0 : 17[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9626:10025 [1] NCCL INFO Channel 01/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-46:702881:703280 [1] NCCL INFO Channel 01/0 : 1[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762816:763218 [1] NCCL INFO Channel 01/0 : 33[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9626:10025 [1] NCCL INFO Channel 09/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127291:127690 [1] NCCL INFO Channel 09/0 : 57[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9631:10022 [6] NCCL INFO Channel 12/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-165-213:9630:10024 [5] NCCL INFO Channel 07/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-165-213:9626:10025 [1] NCCL INFO Channel 09/0 : 9[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9629:10021 [4] NCCL INFO Channel 02/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-165-213:9628:10023 [3] NCCL INFO Channel 01/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-165-213:9626:10025 [1] NCCL INFO Channel 09/0 : 25[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624944:625341 [1] NCCL INFO Channel 01/0 : 1[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127291:127690 [1] NCCL INFO Channel 09/0 : 25[1] -> 57[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624944:625341 [1] NCCL INFO Channel 01/0 : 33[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9631:10022 [6] NCCL INFO Channel 13/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-165-213:9630:10024 [5] NCCL INFO Channel 09/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-166-15:127291:127690 [1] NCCL INFO Channel 09/0 : 41[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9629:10021 [4] NCCL INFO Channel 03/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-166-15:127291:127690 [1] NCCL INFO Channel 09/0 : 25[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-167-9:1039538:1039934 [1] NCCL INFO Channel 01/0 : 57[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-46:702881:703280 [1] NCCL INFO Channel 09/0 : 9[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624944:625341 [1] NCCL INFO Channel 01/0 : 49[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9628:10023 [3] NCCL INFO Channel 02/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-166-214:624944:625341 [1] NCCL INFO Channel 01/0 : 33[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127291:127690 [1] NCCL INFO Channel 01/0 : 25[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9626:10025 [1] NCCL INFO Channel 01/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897241:897635 [1] NCCL INFO Channel 01/0 : 49[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104430:104830 [1] NCCL INFO Channel 01/0 : 57[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9631:10022 [6] NCCL INFO Channel 14/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-165-213:9630:10024 [5] NCCL INFO Channel 10/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-166-214:624944:625341 [1] NCCL INFO Channel 09/0 : 41[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104430:104830 [1] NCCL INFO Channel 01/0 : 49[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762816:763218 [1] NCCL INFO Channel 01/0 : 25[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9626:10025 [1] NCCL INFO Channel 09/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897241:897635 [1] NCCL INFO Channel 09/0 : 49[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9626:10025 [1] NCCL INFO Channel 09/0 : 9[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104430:104830 [1] NCCL INFO Channel 09/0 : 49[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762816:763218 [1] NCCL INFO Channel 01/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9629:10021 [4] NCCL INFO Channel 04/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-165-213:9628:10023 [3] NCCL INFO Channel 03/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-166-125:762816:763218 [1] NCCL INFO Channel 09/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9630:10024 [5] NCCL INFO Channel 11/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-165-213:9629:10021 [4] NCCL INFO Channel 06/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-165-213:9628:10023 [3] NCCL INFO Channel 05/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-167-9:1039538:1039934 [1] NCCL INFO Channel 00/0 : 57[1] -> 56[0] via P2P/IPC
+ip-26-0-165-213:9630:10024 [5] NCCL INFO Channel 12/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-167-9:1039538:1039934 [1] NCCL INFO Channel 08/0 : 57[1] -> 56[0] via P2P/IPC
+ip-26-0-165-213:9629:10021 [4] NCCL INFO Channel 07/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-165-213:9628:10023 [3] NCCL INFO Channel 06/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-162-46:702881:703280 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-213:9630:10024 [5] NCCL INFO Channel 13/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-162-46:702881:703280 [1] NCCL INFO Channel 08/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-213:9629:10021 [4] NCCL INFO Channel 09/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-165-213:9628:10023 [3] NCCL INFO Channel 07/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-166-125:762816:763218 [1] NCCL INFO Channel 00/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-165-213:9630:10024 [5] NCCL INFO Channel 15/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-166-125:762816:763218 [1] NCCL INFO Channel 08/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-166-244:897241:897635 [1] NCCL INFO Channel 09/0 : 41[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9629:10021 [4] NCCL INFO Channel 10/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-165-213:9628:10023 [3] NCCL INFO Channel 09/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-166-15:127291:127690 [1] NCCL INFO Channel 00/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-165-213:9626:10025 [1] NCCL INFO Channel 00/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-166-15:127291:127690 [1] NCCL INFO Channel 08/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-165-213:9631:10022 [6] NCCL INFO Channel 14/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9631:10022 [6] NCCL INFO Channel 06/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9631:10022 [6] NCCL INFO Channel 14/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9628:10023 [3] NCCL INFO Channel 10/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-165-213:9629:10021 [4] NCCL INFO Channel 11/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-162-46:702886:703281 [6] NCCL INFO Channel 06/0 : 38[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9631:10022 [6] NCCL INFO Channel 14/0 : 14[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762821:763224 [6] NCCL INFO Channel 06/0 : 22[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9626:10025 [1] NCCL INFO Channel 08/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-166-214:624944:625341 [1] NCCL INFO Channel 00/0 : 33[1] -> 32[0] via P2P/IPC
+ip-26-0-165-213:9631:10022 [6] NCCL INFO Channel 14/0 : 30[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104430:104830 [1] NCCL INFO Channel 00/0 : 49[1] -> 48[0] via P2P/IPC
+ip-26-0-165-213:9628:10023 [3] NCCL INFO Channel 11/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-166-125:762821:763224 [6] NCCL INFO Channel 06/0 : 38[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127296:127688 [6] NCCL INFO Channel 14/0 : 62[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897241:897635 [1] NCCL INFO Channel 00/0 : 41[1] -> 40[0] via P2P/IPC
+ip-26-0-166-214:624944:625341 [1] NCCL INFO Channel 08/0 : 33[1] -> 32[0] via P2P/IPC
+ip-26-0-165-213:9629:10021 [4] NCCL INFO Channel 12/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-166-15:127296:127688 [6] NCCL INFO Channel 14/0 : 30[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624949:625344 [6] NCCL INFO Channel 06/0 : 6[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624949:625344 [6] NCCL INFO Channel 06/0 : 38[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897241:897635 [1] NCCL INFO Channel 08/0 : 41[1] -> 40[0] via P2P/IPC
+ip-26-0-166-15:127296:127688 [6] NCCL INFO Channel 14/0 : 46[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-167-9:1039543:1039933 [6] NCCL INFO Channel 06/0 : 62[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127296:127688 [6] NCCL INFO Channel 14/0 : 30[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9628:10023 [3] NCCL INFO Channel 13/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-165-213:9629:10021 [4] NCCL INFO Channel 14/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-166-244:897246:897641 [6] NCCL INFO Channel 06/0 : 54[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9632:10026 [7] NCCL INFO Channel 15/0 : 7[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9625:10020 [0] NCCL INFO Channel 08/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897246:897641 [6] NCCL INFO Channel 14/0 : 54[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9632:10026 [7] NCCL INFO Channel 07/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127296:127688 [6] NCCL INFO Channel 06/0 : 30[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-46:702886:703281 [6] NCCL INFO Channel 06/0 : 6[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9625:10020 [0] NCCL INFO Channel 00/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897246:897641 [6] NCCL INFO Channel 14/0 : 46[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9631:10022 [6] NCCL INFO Channel 06/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9632:10026 [7] NCCL INFO Channel 15/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9625:10020 [0] NCCL INFO Channel 08/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9631:10022 [6] NCCL INFO Channel 14/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9631:10022 [6] NCCL INFO Channel 14/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9628:10023 [3] NCCL INFO Channel 14/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-162-46:702886:703281 [6] NCCL INFO Channel 14/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9629:10021 [4] NCCL INFO Channel 15/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-166-214:624949:625344 [6] NCCL INFO Channel 06/0 : 54[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-46:702880:703277 [0] NCCL INFO Channel 00/0 : 32[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624949:625344 [6] NCCL INFO Channel 06/0 : 38[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9632:10026 [7] NCCL INFO Channel 15/0 : 15[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-46:702880:703277 [0] NCCL INFO Channel 00/0 : 0[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762822:763222 [7] NCCL INFO Channel 07/0 : 23[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9625:10020 [0] NCCL INFO Channel 08/0 : 8[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-46:702887:703282 [7] NCCL INFO Channel 07/0 : 39[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762815:763225 [0] NCCL INFO Channel 00/0 : 16[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-46:702887:703282 [7] NCCL INFO Channel 07/0 : 7[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9632:10026 [7] NCCL INFO Channel 15/0 : 31[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762822:763222 [7] NCCL INFO Channel 07/0 : 39[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9628:10023 [3] NCCL INFO Channel 15/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-166-15:127297:127691 [7] NCCL INFO Channel 15/0 : 63[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9625:10020 [0] NCCL INFO Channel 08/0 : 24[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624950:625342 [7] NCCL INFO Channel 07/0 : 7[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624949:625344 [6] NCCL INFO Channel 14/0 : 46[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104435:104826 [6] NCCL INFO Channel 06/0 : 62[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127290:127685 [0] NCCL INFO Channel 08/0 : 56[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762821:763224 [6] NCCL INFO Channel 06/0 : 30[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127297:127691 [7] NCCL INFO Channel 15/0 : 31[7] -> 63[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104435:104826 [6] NCCL INFO Channel 06/0 : 54[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624950:625342 [7] NCCL INFO Channel 07/0 : 39[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762815:763225 [0] NCCL INFO Channel 00/0 : 32[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127290:127685 [0] NCCL INFO Channel 08/0 : 24[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104435:104826 [6] NCCL INFO Channel 14/0 : 54[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624943:625338 [0] NCCL INFO Channel 00/0 : 0[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762821:763224 [6] NCCL INFO Channel 06/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9630:10024 [5] NCCL INFO Channel 13/0 : 5[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624943:625338 [0] NCCL INFO Channel 00/0 : 32[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762821:763224 [6] NCCL INFO Channel 14/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9630:10024 [5] NCCL INFO Channel 05/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127297:127691 [7] NCCL INFO Channel 15/0 : 47[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9630:10024 [5] NCCL INFO Channel 13/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127290:127685 [0] NCCL INFO Channel 08/0 : 40[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-46:702887:703282 [7] NCCL INFO Channel 15/0 : 15[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624950:625342 [7] NCCL INFO Channel 07/0 : 55[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-167-9:1039537:1039932 [0] NCCL INFO Channel 00/0 : 56[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127297:127691 [7] NCCL INFO Channel 15/0 : 31[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9627:10027 [2] NCCL INFO Channel 10/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624950:625342 [7] NCCL INFO Channel 07/0 : 39[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-46:702880:703277 [0] NCCL INFO Channel 08/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127290:127685 [0] NCCL INFO Channel 08/0 : 24[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-167-9:1039544:1039931 [7] NCCL INFO Channel 07/0 : 63[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9629:10021 [4] NCCL INFO Channel 12/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624943:625338 [0] NCCL INFO Channel 00/0 : 48[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9627:10027 [2] NCCL INFO Channel 02/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9629:10021 [4] NCCL INFO Channel 04/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9628:10023 [3] NCCL INFO Channel 11/0 : 3[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624943:625338 [0] NCCL INFO Channel 00/0 : 32[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-46:702886:703281 [6] NCCL INFO Channel 00/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-213:9627:10027 [2] NCCL INFO Channel 10/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-46:702885:703278 [5] NCCL INFO Channel 05/0 : 37[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897247:897642 [7] NCCL INFO Channel 07/0 : 55[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897240:897640 [0] NCCL INFO Channel 00/0 : 48[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9629:10021 [4] NCCL INFO Channel 12/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9628:10023 [3] NCCL INFO Channel 03/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-46:702885:703278 [5] NCCL INFO Channel 05/0 : 5[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127297:127691 [7] NCCL INFO Channel 07/0 : 31[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9630:10024 [5] NCCL INFO Channel 13/0 : 13[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9628:10023 [3] NCCL INFO Channel 11/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762820:763219 [5] NCCL INFO Channel 05/0 : 21[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897240:897640 [0] NCCL INFO Channel 08/0 : 48[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9632:10026 [7] NCCL INFO Channel 07/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127290:127685 [0] NCCL INFO Channel 00/0 : 24[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104436:104824 [7] NCCL INFO Channel 07/0 : 63[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897247:897642 [7] NCCL INFO Channel 15/0 : 55[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624950:625342 [7] NCCL INFO Channel 15/0 : 47[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762822:763222 [7] NCCL INFO Channel 07/0 : 31[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897240:897640 [0] NCCL INFO Channel 08/0 : 40[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9625:10020 [0] NCCL INFO Channel 00/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-46:702882:703284 [2] NCCL INFO Channel 02/0 : 34[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104436:104824 [7] NCCL INFO Channel 07/0 : 55[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9632:10026 [7] NCCL INFO Channel 15/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-46:702886:703281 [6] NCCL INFO Channel 05/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-166-36:104429:104829 [0] NCCL INFO Channel 00/0 : 56[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624943:625338 [0] NCCL INFO Channel 08/0 : 40[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762822:763222 [7] NCCL INFO Channel 07/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897247:897642 [7] NCCL INFO Channel 15/0 : 47[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9625:10020 [0] NCCL INFO Channel 08/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-46:702882:703284 [2] NCCL INFO Channel 02/0 : 2[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104436:104824 [7] NCCL INFO Channel 15/0 : 55[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9632:10026 [7] NCCL INFO Channel 15/0 : 15[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762815:763225 [0] NCCL INFO Channel 00/0 : 24[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-46:702884:703279 [4] NCCL INFO Channel 04/0 : 36[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127295:127686 [5] NCCL INFO Channel 13/0 : 61[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-46:702883:703283 [3] NCCL INFO Channel 03/0 : 35[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104429:104829 [0] NCCL INFO Channel 00/0 : 48[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9625:10020 [0] NCCL INFO Channel 08/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762822:763222 [7] NCCL INFO Channel 15/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9630:10024 [5] NCCL INFO Channel 13/0 : 29[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9627:10027 [2] NCCL INFO Channel 10/0 : 10[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-46:702884:703279 [4] NCCL INFO Channel 04/0 : 4[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762817:763223 [2] NCCL INFO Channel 02/0 : 18[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762820:763219 [5] NCCL INFO Channel 05/0 : 37[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624948:625343 [5] NCCL INFO Channel 05/0 : 5[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127295:127686 [5] NCCL INFO Channel 13/0 : 29[5] -> 61[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104429:104829 [0] NCCL INFO Channel 08/0 : 48[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762815:763225 [0] NCCL INFO Channel 00/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-46:702883:703283 [3] NCCL INFO Channel 03/0 : 3[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-46:702886:703281 [6] NCCL INFO Channel 08/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-213:9629:10021 [4] NCCL INFO Channel 12/0 : 12[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9628:10023 [3] NCCL INFO Channel 11/0 : 11[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624949:625344 [6] NCCL INFO Channel 00/0 : 38[6] -> 37[5] via P2P/IPC
+ip-26-0-166-214:624948:625343 [5] NCCL INFO Channel 05/0 : 37[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127297:127691 [7] NCCL INFO Channel 01/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-166-36:104430:104830 [1] NCCL INFO Channel 08/0 : 49[1] -> 48[0] via P2P/IPC
+ip-26-0-166-125:762819:763221 [4] NCCL INFO Channel 04/0 : 20[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-167-9:1039544:1039931 [7] NCCL INFO Channel 01/0 : 63[7] -> 56[0] via P2P/IPC
+ip-26-0-162-46:702887:703282 [7] NCCL INFO Channel 01/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762818:763220 [3] NCCL INFO Channel 03/0 : 19[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762815:763225 [0] NCCL INFO Channel 08/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624950:625342 [7] NCCL INFO Channel 01/0 : 39[7] -> 32[0] via P2P/IPC
+ip-26-0-165-213:9627:10027 [2] NCCL INFO Channel 10/0 : 26[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104436:104824 [7] NCCL INFO Channel 01/0 : 55[7] -> 48[0] via P2P/IPC
+ip-26-0-166-125:762817:763223 [2] NCCL INFO Channel 02/0 : 34[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762822:763222 [7] NCCL INFO Channel 01/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-166-15:127292:127692 [2] NCCL INFO Channel 10/0 : 58[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897247:897642 [7] NCCL INFO Channel 01/0 : 47[7] -> 40[0] via P2P/IPC
+ip-26-0-165-213:9629:10021 [4] NCCL INFO Channel 12/0 : 28[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127295:127686 [5] NCCL INFO Channel 13/0 : 45[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127294:127687 [4] NCCL INFO Channel 12/0 : 60[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9632:10026 [7] NCCL INFO Channel 01/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-166-214:624945:625340 [2] NCCL INFO Channel 02/0 : 2[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9628:10023 [3] NCCL INFO Channel 11/0 : 27[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-167-9:1039542:1039936 [5] NCCL INFO Channel 05/0 : 61[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762819:763221 [4] NCCL INFO Channel 04/0 : 36[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762818:763220 [3] NCCL INFO Channel 03/0 : 35[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127292:127692 [2] NCCL INFO Channel 10/0 : 26[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-46:702885:703278 [5] NCCL INFO Channel 13/0 : 13[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127293:127689 [3] NCCL INFO Channel 11/0 : 59[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9631:10022 [6] NCCL INFO Channel 00/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-166-214:624948:625343 [5] NCCL INFO Channel 05/0 : 53[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-46:702886:703281 [6] NCCL INFO Channel 13/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-166-15:127295:127686 [5] NCCL INFO Channel 13/0 : 29[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127294:127687 [4] NCCL INFO Channel 12/0 : 28[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127296:127688 [6] NCCL INFO Channel 00/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-166-214:624947:625339 [4] NCCL INFO Channel 04/0 : 4[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127293:127689 [3] NCCL INFO Channel 11/0 : 27[3] -> 59[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624945:625340 [2] NCCL INFO Channel 02/0 : 34[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624946:625337 [3] NCCL INFO Channel 03/0 : 3[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624948:625343 [5] NCCL INFO Channel 05/0 : 37[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127297:127691 [7] NCCL INFO Channel 02/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-166-214:624947:625339 [4] NCCL INFO Channel 04/0 : 36[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762821:763224 [6] NCCL INFO Channel 00/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-166-125:762822:763222 [7] NCCL INFO Channel 02/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-166-214:624946:625337 [3] NCCL INFO Channel 03/0 : 35[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9632:10026 [7] NCCL INFO Channel 02/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-166-214:624949:625344 [6] NCCL INFO Channel 05/0 : 38[6] -> 37[5] via P2P/IPC
+ip-26-0-166-15:127292:127692 [2] NCCL INFO Channel 10/0 : 42[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-167-9:1039539:1039938 [2] NCCL INFO Channel 02/0 : 58[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897245:897638 [5] NCCL INFO Channel 05/0 : 53[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127294:127687 [4] NCCL INFO Channel 12/0 : 44[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127292:127692 [2] NCCL INFO Channel 10/0 : 26[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624945:625340 [2] NCCL INFO Channel 02/0 : 50[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127293:127689 [3] NCCL INFO Channel 11/0 : 43[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-167-9:1039541:1039937 [4] NCCL INFO Channel 04/0 : 60[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-46:702882:703284 [2] NCCL INFO Channel 10/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-46:702887:703282 [7] NCCL INFO Channel 02/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624950:625342 [7] NCCL INFO Channel 02/0 : 39[7] -> 32[0] via P2P/IPC
+ip-26-0-166-15:127294:127687 [4] NCCL INFO Channel 12/0 : 28[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104434:104828 [5] NCCL INFO Channel 05/0 : 61[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897245:897638 [5] NCCL INFO Channel 13/0 : 53[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9630:10024 [5] NCCL INFO Channel 05/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-167-9:1039540:1039935 [3] NCCL INFO Channel 03/0 : 59[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-167-9:1039544:1039931 [7] NCCL INFO Channel 02/0 : 63[7] -> 56[0] via P2P/IPC
+ip-26-0-166-15:127295:127686 [5] NCCL INFO Channel 05/0 : 29[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127293:127689 [3] NCCL INFO Channel 11/0 : 27[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9630:10024 [5] NCCL INFO Channel 13/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104434:104828 [5] NCCL INFO Channel 05/0 : 53[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9631:10022 [6] NCCL INFO Channel 05/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-165-213:9630:10024 [5] NCCL INFO Channel 13/0 : 13[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104436:104824 [7] NCCL INFO Channel 02/0 : 55[7] -> 48[0] via P2P/IPC
+ip-26-0-166-244:897242:897639 [2] NCCL INFO Channel 02/0 : 50[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624950:625342 [7] NCCL INFO Channel 03/0 : 39[7] -> 32[0] via P2P/IPC
+ip-26-0-162-46:702887:703282 [7] NCCL INFO Channel 03/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-36:104434:104828 [5] NCCL INFO Channel 13/0 : 53[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897242:897639 [2] NCCL INFO Channel 10/0 : 50[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762822:763222 [7] NCCL INFO Channel 03/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-165-213:9632:10026 [7] NCCL INFO Channel 03/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-166-244:897243:897637 [3] NCCL INFO Channel 03/0 : 51[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762821:763224 [6] NCCL INFO Channel 05/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-166-244:897242:897639 [2] NCCL INFO Channel 10/0 : 42[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127293:127689 [3] NCCL INFO Channel 03/0 : 27[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9628:10023 [3] NCCL INFO Channel 03/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-46:702887:703282 [7] NCCL INFO Channel 04/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-244:897243:897637 [3] NCCL INFO Channel 11/0 : 51[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624950:625342 [7] NCCL INFO Channel 04/0 : 39[7] -> 32[0] via P2P/IPC
+ip-26-0-165-213:9628:10023 [3] NCCL INFO Channel 11/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897247:897642 [7] NCCL INFO Channel 02/0 : 47[7] -> 40[0] via P2P/IPC
+ip-26-0-166-244:897243:897637 [3] NCCL INFO Channel 11/0 : 43[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104436:104824 [7] NCCL INFO Channel 03/0 : 55[7] -> 48[0] via P2P/IPC
+ip-26-0-165-213:9628:10023 [3] NCCL INFO Channel 11/0 : 11[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-167-9:1039544:1039931 [7] NCCL INFO Channel 03/0 : 63[7] -> 56[0] via P2P/IPC
+ip-26-0-165-213:9631:10022 [6] NCCL INFO Channel 08/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-162-46:702887:703282 [7] NCCL INFO Channel 05/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624950:625342 [7] NCCL INFO Channel 05/0 : 39[7] -> 32[0] via P2P/IPC
+ip-26-0-165-213:9632:10026 [7] NCCL INFO Channel 04/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-166-36:104435:104826 [6] NCCL INFO Channel 00/0 : 54[6] -> 53[5] via P2P/IPC
+ip-26-0-166-125:762822:763222 [7] NCCL INFO Channel 04/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-166-125:762821:763224 [6] NCCL INFO Channel 08/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-162-46:702887:703282 [7] NCCL INFO Channel 06/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-244:897247:897642 [7] NCCL INFO Channel 03/0 : 47[7] -> 40[0] via P2P/IPC
+ip-26-0-166-214:624950:625342 [7] NCCL INFO Channel 06/0 : 39[7] -> 32[0] via P2P/IPC
+ip-26-0-166-244:897246:897641 [6] NCCL INFO Channel 00/0 : 46[6] -> 45[5] via P2P/IPC
+ip-26-0-166-36:104436:104824 [7] NCCL INFO Channel 04/0 : 55[7] -> 48[0] via P2P/IPC
+ip-26-0-165-213:9631:10022 [6] NCCL INFO Channel 13/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-166-214:624950:625342 [7] NCCL INFO Channel 07/0 : 39[7] -> 32[0] via P2P/IPC
+ip-26-0-162-46:702887:703282 [7] NCCL INFO Channel 07/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-36:104435:104826 [6] NCCL INFO Channel 05/0 : 54[6] -> 53[5] via P2P/IPC
+ip-26-0-167-9:1039544:1039931 [7] NCCL INFO Channel 04/0 : 63[7] -> 56[0] via P2P/IPC
+ip-26-0-166-125:762820:763219 [5] NCCL INFO Channel 05/0 : 29[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9632:10026 [7] NCCL INFO Channel 05/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-166-214:624948:625343 [5] NCCL INFO Channel 13/0 : 45[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762822:763222 [7] NCCL INFO Channel 05/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-166-125:762820:763219 [5] NCCL INFO Channel 05/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762821:763224 [6] NCCL INFO Channel 13/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-166-214:624949:625344 [6] NCCL INFO Channel 08/0 : 38[6] -> 37[5] via P2P/IPC
+ip-26-0-166-125:762820:763219 [5] NCCL INFO Channel 13/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624945:625340 [2] NCCL INFO Channel 02/0 : 34[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-46:702887:703282 [7] NCCL INFO Channel 09/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624950:625342 [7] NCCL INFO Channel 09/0 : 39[7] -> 32[0] via P2P/IPC
+ip-26-0-166-244:897246:897641 [6] NCCL INFO Channel 05/0 : 46[6] -> 45[5] via P2P/IPC
+ip-26-0-166-244:897247:897642 [7] NCCL INFO Channel 04/0 : 47[7] -> 40[0] via P2P/IPC
+ip-26-0-166-15:127292:127692 [2] NCCL INFO Channel 02/0 : 26[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104436:104824 [7] NCCL INFO Channel 05/0 : 55[7] -> 48[0] via P2P/IPC
+ip-26-0-165-213:9627:10027 [2] NCCL INFO Channel 02/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-167-9:1039543:1039933 [6] NCCL INFO Channel 00/0 : 62[6] -> 61[5] via P2P/IPC
+ip-26-0-165-213:9627:10027 [2] NCCL INFO Channel 10/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-46:702887:703282 [7] NCCL INFO Channel 10/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-36:104435:104826 [6] NCCL INFO Channel 08/0 : 54[6] -> 53[5] via P2P/IPC
+ip-26-0-165-213:9632:10026 [7] NCCL INFO Channel 06/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-166-36:104431:104831 [2] NCCL INFO Channel 02/0 : 58[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762822:763222 [7] NCCL INFO Channel 06/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-165-213:9627:10027 [2] NCCL INFO Channel 10/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624949:625344 [6] NCCL INFO Channel 13/0 : 38[6] -> 37[5] via P2P/IPC
+ip-26-0-166-36:104431:104831 [2] NCCL INFO Channel 02/0 : 50[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762817:763223 [2] NCCL INFO Channel 02/0 : 26[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624950:625342 [7] NCCL INFO Channel 10/0 : 39[7] -> 32[0] via P2P/IPC
+ip-26-0-166-36:104431:104831 [2] NCCL INFO Channel 10/0 : 50[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127296:127688 [6] NCCL INFO Channel 05/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-166-125:762817:763223 [2] NCCL INFO Channel 02/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624945:625340 [2] NCCL INFO Channel 10/0 : 42[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-46:702887:703282 [7] NCCL INFO Channel 11/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127297:127691 [7] NCCL INFO Channel 03/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-166-244:897246:897641 [6] NCCL INFO Channel 08/0 : 46[6] -> 45[5] via P2P/IPC
+ip-26-0-166-36:104436:104824 [7] NCCL INFO Channel 06/0 : 55[7] -> 48[0] via P2P/IPC
+ip-26-0-166-125:762817:763223 [2] NCCL INFO Channel 10/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624947:625339 [4] NCCL INFO Channel 04/0 : 52[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897247:897642 [7] NCCL INFO Channel 05/0 : 47[7] -> 40[0] via P2P/IPC
+ip-26-0-166-214:624947:625339 [4] NCCL INFO Channel 04/0 : 36[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-167-9:1039543:1039933 [6] NCCL INFO Channel 05/0 : 62[6] -> 61[5] via P2P/IPC
+ip-26-0-166-36:104435:104826 [6] NCCL INFO Channel 13/0 : 54[6] -> 53[5] via P2P/IPC
+ip-26-0-162-46:702887:703282 [7] NCCL INFO Channel 12/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-213:9632:10026 [7] NCCL INFO Channel 07/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-166-214:624950:625342 [7] NCCL INFO Channel 11/0 : 39[7] -> 32[0] via P2P/IPC
+ip-26-0-166-125:762822:763222 [7] NCCL INFO Channel 07/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-166-36:104433:104825 [4] NCCL INFO Channel 04/0 : 60[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624947:625339 [4] NCCL INFO Channel 12/0 : 44[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127296:127688 [6] NCCL INFO Channel 08/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-166-36:104433:104825 [4] NCCL INFO Channel 04/0 : 52[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762819:763221 [4] NCCL INFO Channel 04/0 : 28[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127297:127691 [7] NCCL INFO Channel 04/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-166-36:104436:104824 [7] NCCL INFO Channel 07/0 : 55[7] -> 48[0] via P2P/IPC
+ip-26-0-167-9:1039543:1039933 [6] NCCL INFO Channel 08/0 : 62[6] -> 61[5] via P2P/IPC
+ip-26-0-166-214:624946:625337 [3] NCCL INFO Channel 03/0 : 51[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762819:763221 [4] NCCL INFO Channel 04/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-46:702883:703283 [3] NCCL INFO Channel 11/0 : 11[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104433:104825 [4] NCCL INFO Channel 12/0 : 52[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897246:897641 [6] NCCL INFO Channel 13/0 : 46[6] -> 45[5] via P2P/IPC
+ip-26-0-166-244:897247:897642 [7] NCCL INFO Channel 06/0 : 47[7] -> 40[0] via P2P/IPC
+ip-26-0-166-214:624950:625342 [7] NCCL INFO Channel 12/0 : 39[7] -> 32[0] via P2P/IPC
+ip-26-0-166-125:762819:763221 [4] NCCL INFO Channel 12/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-167-9:1039542:1039936 [5] NCCL INFO Channel 00/0 : 61[5] -> 60[4] via P2P/IPC
+ip-26-0-166-214:624946:625337 [3] NCCL INFO Channel 03/0 : 35[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897244:897636 [4] NCCL INFO Channel 04/0 : 52[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9629:10021 [4] NCCL INFO Channel 04/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127294:127687 [4] NCCL INFO Channel 04/0 : 28[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897244:897636 [4] NCCL INFO Channel 12/0 : 52[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9632:10026 [7] NCCL INFO Channel 09/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-165-213:9629:10021 [4] NCCL INFO Channel 12/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897244:897636 [4] NCCL INFO Channel 12/0 : 44[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9629:10021 [4] NCCL INFO Channel 12/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-167-9:1039543:1039933 [6] NCCL INFO Channel 13/0 : 62[6] -> 61[5] via P2P/IPC
+ip-26-0-166-214:624946:625337 [3] NCCL INFO Channel 11/0 : 43[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762818:763220 [3] NCCL INFO Channel 03/0 : 27[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762822:763222 [7] NCCL INFO Channel 09/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-166-214:624950:625342 [7] NCCL INFO Channel 13/0 : 39[7] -> 32[0] via P2P/IPC
+ip-26-0-162-46:702884:703279 [4] NCCL INFO Channel 12/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762818:763220 [3] NCCL INFO Channel 03/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-167-9:1039542:1039936 [5] NCCL INFO Channel 04/0 : 61[5] -> 60[4] via P2P/IPC
+ip-26-0-166-125:762818:763220 [3] NCCL INFO Channel 11/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127296:127688 [6] NCCL INFO Channel 13/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-166-15:127297:127691 [7] NCCL INFO Channel 05/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-166-244:897247:897642 [7] NCCL INFO Channel 07/0 : 47[7] -> 40[0] via P2P/IPC
+ip-26-0-165-213:9632:10026 [7] NCCL INFO Channel 10/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-167-9:1039544:1039931 [7] NCCL INFO Channel 05/0 : 63[7] -> 56[0] via P2P/IPC
+ip-26-0-166-244:897247:897642 [7] NCCL INFO Channel 09/0 : 47[7] -> 40[0] via P2P/IPC
+ip-26-0-166-244:897247:897642 [7] NCCL INFO Channel 10/0 : 47[7] -> 40[0] via P2P/IPC
+ip-26-0-167-9:1039542:1039936 [5] NCCL INFO Channel 08/0 : 61[5] -> 60[4] via P2P/IPC
+ip-26-0-166-36:104436:104824 [7] NCCL INFO Channel 09/0 : 55[7] -> 48[0] via P2P/IPC
+ip-26-0-166-125:762822:763222 [7] NCCL INFO Channel 10/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-166-214:624950:625342 [7] NCCL INFO Channel 14/0 : 39[7] -> 32[0] via P2P/IPC
+ip-26-0-166-36:104432:104827 [3] NCCL INFO Channel 03/0 : 59[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-167-9:1039544:1039931 [7] NCCL INFO Channel 06/0 : 63[7] -> 56[0] via P2P/IPC
+ip-26-0-166-36:104432:104827 [3] NCCL INFO Channel 03/0 : 51[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104432:104827 [3] NCCL INFO Channel 11/0 : 51[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9632:10026 [7] NCCL INFO Channel 11/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-165-213:9630:10024 [5] NCCL INFO Channel 00/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-166-244:897247:897642 [7] NCCL INFO Channel 11/0 : 47[7] -> 40[0] via P2P/IPC
+ip-26-0-166-15:127297:127691 [7] NCCL INFO Channel 06/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-166-36:104436:104824 [7] NCCL INFO Channel 10/0 : 55[7] -> 48[0] via P2P/IPC
+ip-26-0-166-214:624950:625342 [7] NCCL INFO Channel 15/0 : 39[7] -> 32[0] via P2P/IPC
+ip-26-0-167-9:1039542:1039936 [5] NCCL INFO Channel 12/0 : 61[5] -> 60[4] via P2P/IPC
+ip-26-0-166-244:897245:897638 [5] NCCL INFO Channel 13/0 : 45[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624945:625340 [2] NCCL INFO Channel 00/0 : 34[2] -> 33[1] via P2P/IPC
+ip-26-0-167-9:1039544:1039931 [7] NCCL INFO Channel 07/0 : 63[7] -> 56[0] via P2P/IPC
+ip-26-0-162-46:702887:703282 [7] NCCL INFO Channel 13/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-244:897247:897642 [7] NCCL INFO Channel 12/0 : 47[7] -> 40[0] via P2P/IPC
+ip-26-0-166-125:762817:763223 [2] NCCL INFO Channel 00/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-166-36:104436:104824 [7] NCCL INFO Channel 11/0 : 55[7] -> 48[0] via P2P/IPC
+ip-26-0-165-213:9632:10026 [7] NCCL INFO Channel 12/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-165-213:9630:10024 [5] NCCL INFO Channel 04/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-166-125:762817:763223 [2] NCCL INFO Channel 01/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-165-213:9627:10027 [2] NCCL INFO Channel 00/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-166-125:762817:763223 [2] NCCL INFO Channel 08/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-166-15:127295:127686 [5] NCCL INFO Channel 00/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-166-15:127297:127691 [7] NCCL INFO Channel 07/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-166-15:127295:127686 [5] NCCL INFO Channel 04/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-166-15:127292:127692 [2] NCCL INFO Channel 00/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-166-214:624945:625340 [2] NCCL INFO Channel 01/0 : 34[2] -> 33[1] via P2P/IPC
+ip-26-0-167-9:1039544:1039931 [7] NCCL INFO Channel 09/0 : 63[7] -> 56[0] via P2P/IPC
+ip-26-0-166-244:897247:897642 [7] NCCL INFO Channel 13/0 : 47[7] -> 40[0] via P2P/IPC
+ip-26-0-166-125:762817:763223 [2] NCCL INFO Channel 09/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-167-9:1039539:1039938 [2] NCCL INFO Channel 00/0 : 58[2] -> 57[1] via P2P/IPC
+ip-26-0-166-15:127297:127691 [7] NCCL INFO Channel 09/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-166-15:127295:127686 [5] NCCL INFO Channel 08/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-166-125:762820:763219 [5] NCCL INFO Channel 00/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-166-36:104436:104824 [7] NCCL INFO Channel 12/0 : 55[7] -> 48[0] via P2P/IPC
+ip-26-0-165-213:9632:10026 [7] NCCL INFO Channel 13/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-165-213:9630:10024 [5] NCCL INFO Channel 08/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-166-244:897242:897639 [2] NCCL INFO Channel 00/0 : 42[2] -> 41[1] via P2P/IPC
+ip-26-0-166-125:762822:763222 [7] NCCL INFO Channel 11/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-166-15:127292:127692 [2] NCCL INFO Channel 01/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-166-15:127294:127687 [4] NCCL INFO Channel 00/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-165-213:9627:10027 [2] NCCL INFO Channel 01/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-166-214:624945:625340 [2] NCCL INFO Channel 08/0 : 34[2] -> 33[1] via P2P/IPC
+ip-26-0-167-9:1039544:1039931 [7] NCCL INFO Channel 10/0 : 63[7] -> 56[0] via P2P/IPC
+ip-26-0-166-214:624947:625339 [4] NCCL INFO Channel 00/0 : 36[4] -> 35[3] via P2P/IPC
+ip-26-0-166-244:897247:897642 [7] NCCL INFO Channel 14/0 : 47[7] -> 40[0] via P2P/IPC
+ip-26-0-167-9:1039539:1039938 [2] NCCL INFO Channel 01/0 : 58[2] -> 57[1] via P2P/IPC
+ip-26-0-166-125:762820:763219 [5] NCCL INFO Channel 04/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-166-15:127297:127691 [7] NCCL INFO Channel 10/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-166-15:127295:127686 [5] NCCL INFO Channel 12/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-166-36:104436:104824 [7] NCCL INFO Channel 13/0 : 55[7] -> 48[0] via P2P/IPC
+ip-26-0-166-36:104431:104831 [2] NCCL INFO Channel 00/0 : 50[2] -> 49[1] via P2P/IPC
+ip-26-0-166-244:897242:897639 [2] NCCL INFO Channel 01/0 : 42[2] -> 41[1] via P2P/IPC
+ip-26-0-166-125:762822:763222 [7] NCCL INFO Channel 12/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-165-213:9630:10024 [5] NCCL INFO Channel 12/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-165-213:9629:10021 [4] NCCL INFO Channel 00/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-166-36:104433:104825 [4] NCCL INFO Channel 00/0 : 52[4] -> 51[3] via P2P/IPC
+ip-26-0-166-214:624945:625340 [2] NCCL INFO Channel 09/0 : 34[2] -> 33[1] via P2P/IPC
+ip-26-0-165-213:9627:10027 [2] NCCL INFO Channel 08/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-166-244:897247:897642 [7] NCCL INFO Channel 15/0 : 47[7] -> 40[0] via P2P/IPC
+ip-26-0-166-15:127292:127692 [2] NCCL INFO Channel 08/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-167-9:1039544:1039931 [7] NCCL INFO Channel 11/0 : 63[7] -> 56[0] via P2P/IPC
+ip-26-0-166-214:624947:625339 [4] NCCL INFO Channel 03/0 : 36[4] -> 35[3] via P2P/IPC
+ip-26-0-166-15:127294:127687 [4] NCCL INFO Channel 03/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-167-9:1039539:1039938 [2] NCCL INFO Channel 08/0 : 58[2] -> 57[1] via P2P/IPC
+ip-26-0-167-9:1039541:1039937 [4] NCCL INFO Channel 00/0 : 60[4] -> 59[3] via P2P/IPC
+ip-26-0-166-125:762820:763219 [5] NCCL INFO Channel 08/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-166-36:104436:104824 [7] NCCL INFO Channel 14/0 : 55[7] -> 48[0] via P2P/IPC
+ip-26-0-166-244:897244:897636 [4] NCCL INFO Channel 00/0 : 44[4] -> 43[3] via P2P/IPC
+ip-26-0-166-244:897242:897639 [2] NCCL INFO Channel 08/0 : 42[2] -> 41[1] via P2P/IPC
+ip-26-0-162-46:702887:703282 [7] NCCL INFO Channel 14/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127297:127691 [7] NCCL INFO Channel 11/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-166-36:104431:104831 [2] NCCL INFO Channel 01/0 : 50[2] -> 49[1] via P2P/IPC
+ip-26-0-166-125:762819:763221 [4] NCCL INFO Channel 00/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-162-46:702884:703279 [4] NCCL INFO Channel 00/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-166-125:762822:763222 [7] NCCL INFO Channel 13/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-165-213:9629:10021 [4] NCCL INFO Channel 03/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-166-36:104433:104825 [4] NCCL INFO Channel 03/0 : 52[4] -> 51[3] via P2P/IPC
+ip-26-0-165-213:9627:10027 [2] NCCL INFO Channel 09/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-166-214:624947:625339 [4] NCCL INFO Channel 05/0 : 36[4] -> 35[3] via P2P/IPC
+ip-26-0-166-244:897245:897638 [5] NCCL INFO Channel 00/0 : 45[5] -> 44[4] via P2P/IPC
+ip-26-0-165-213:9632:10026 [7] NCCL INFO Channel 14/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-166-244:897247:897642 [7] NCCL INFO Channel 00/0 : 47[7] -> 46[6] via P2P/IPC
+ip-26-0-167-9:1039544:1039931 [7] NCCL INFO Channel 12/0 : 63[7] -> 56[0] via P2P/IPC
+ip-26-0-167-9:1039539:1039938 [2] NCCL INFO Channel 09/0 : 58[2] -> 57[1] via P2P/IPC
+ip-26-0-166-15:127292:127692 [2] NCCL INFO Channel 09/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-167-9:1039541:1039937 [4] NCCL INFO Channel 03/0 : 60[4] -> 59[3] via P2P/IPC
+ip-26-0-166-125:762820:763219 [5] NCCL INFO Channel 12/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-166-244:897242:897639 [2] NCCL INFO Channel 09/0 : 42[2] -> 41[1] via P2P/IPC
+ip-26-0-166-214:624948:625343 [5] NCCL INFO Channel 00/0 : 37[5] -> 36[4] via P2P/IPC
+ip-26-0-166-15:127294:127687 [4] NCCL INFO Channel 05/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-166-244:897244:897636 [4] NCCL INFO Channel 03/0 : 44[4] -> 43[3] via P2P/IPC
+ip-26-0-162-46:702887:703282 [7] NCCL INFO Channel 15/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-162-46:702882:703284 [2] NCCL INFO Channel 00/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-36:104436:104824 [7] NCCL INFO Channel 15/0 : 55[7] -> 48[0] via P2P/IPC
+ip-26-0-162-46:702884:703279 [4] NCCL INFO Channel 03/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-166-36:104434:104828 [5] NCCL INFO Channel 00/0 : 53[5] -> 52[4] via P2P/IPC
+ip-26-0-165-213:9629:10021 [4] NCCL INFO Channel 05/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-166-15:127297:127691 [7] NCCL INFO Channel 12/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-166-214:624947:625339 [4] NCCL INFO Channel 06/0 : 36[4] -> 35[3] via P2P/IPC
+ip-26-0-166-125:762819:763221 [4] NCCL INFO Channel 03/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-166-36:104431:104831 [2] NCCL INFO Channel 08/0 : 50[2] -> 49[1] via P2P/IPC
+ip-26-0-166-36:104433:104825 [4] NCCL INFO Channel 05/0 : 52[4] -> 51[3] via P2P/IPC
+ip-26-0-166-125:762822:763222 [7] NCCL INFO Channel 14/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-165-213:9632:10026 [7] NCCL INFO Channel 15/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-167-9:1039544:1039931 [7] NCCL INFO Channel 13/0 : 63[7] -> 56[0] via P2P/IPC
+ip-26-0-167-9:1039541:1039937 [4] NCCL INFO Channel 05/0 : 60[4] -> 59[3] via P2P/IPC
+ip-26-0-166-214:624948:625343 [5] NCCL INFO Channel 04/0 : 37[5] -> 36[4] via P2P/IPC
+ip-26-0-165-213:9629:10021 [4] NCCL INFO Channel 06/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-166-15:127294:127687 [4] NCCL INFO Channel 06/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-162-46:702882:703284 [2] NCCL INFO Channel 01/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-162-46:702884:703279 [4] NCCL INFO Channel 05/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-166-214:624947:625339 [4] NCCL INFO Channel 07/0 : 36[4] -> 35[3] via P2P/IPC
+ip-26-0-166-125:762819:763221 [4] NCCL INFO Channel 05/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-166-36:104434:104828 [5] NCCL INFO Channel 04/0 : 53[5] -> 52[4] via P2P/IPC
+ip-26-0-166-125:762822:763222 [7] NCCL INFO Channel 15/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-166-36:104436:104824 [7] NCCL INFO Channel 00/0 : 55[7] -> 54[6] via P2P/IPC
+ip-26-0-166-214:624948:625343 [5] NCCL INFO Channel 08/0 : 37[5] -> 36[4] via P2P/IPC
+ip-26-0-166-36:104433:104825 [4] NCCL INFO Channel 06/0 : 52[4] -> 51[3] via P2P/IPC
+ip-26-0-166-36:104431:104831 [2] NCCL INFO Channel 09/0 : 50[2] -> 49[1] via P2P/IPC
+ip-26-0-165-213:9629:10021 [4] NCCL INFO Channel 07/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-166-244:897245:897638 [5] NCCL INFO Channel 04/0 : 45[5] -> 44[4] via P2P/IPC
+ip-26-0-162-46:702882:703284 [2] NCCL INFO Channel 08/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-162-46:702884:703279 [4] NCCL INFO Channel 06/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-166-125:762819:763221 [4] NCCL INFO Channel 06/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-165-213:9632:10026 [7] NCCL INFO Channel 00/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-166-214:624947:625339 [4] NCCL INFO Channel 08/0 : 36[4] -> 35[3] via P2P/IPC
+ip-26-0-165-213:9629:10021 [4] NCCL INFO Channel 08/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-166-15:127297:127691 [7] NCCL INFO Channel 13/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-166-214:624948:625343 [5] NCCL INFO Channel 12/0 : 37[5] -> 36[4] via P2P/IPC
+ip-26-0-166-125:762822:763222 [7] NCCL INFO Channel 00/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-166-36:104434:104828 [5] NCCL INFO Channel 08/0 : 53[5] -> 52[4] via P2P/IPC
+ip-26-0-166-244:897245:897638 [5] NCCL INFO Channel 08/0 : 45[5] -> 44[4] via P2P/IPC
+ip-26-0-166-125:762819:763221 [4] NCCL INFO Channel 07/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-166-244:897244:897636 [4] NCCL INFO Channel 05/0 : 44[4] -> 43[3] via P2P/IPC
+ip-26-0-166-15:127294:127687 [4] NCCL INFO Channel 07/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-167-9:1039540:1039935 [3] NCCL INFO Channel 00/0 : 59[3] -> 58[2] via P2P/IPC
+ip-26-0-162-46:702882:703284 [2] NCCL INFO Channel 09/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-162-46:702884:703279 [4] NCCL INFO Channel 07/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-165-213:9632:10026 [7] NCCL INFO Channel 06/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-165-213:9629:10021 [4] NCCL INFO Channel 11/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-166-214:624946:625337 [3] NCCL INFO Channel 00/0 : 35[3] -> 34[2] via P2P/IPC
+ip-26-0-166-15:127297:127691 [7] NCCL INFO Channel 14/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-166-214:624947:625339 [4] NCCL INFO Channel 11/0 : 36[4] -> 35[3] via P2P/IPC
+ip-26-0-167-9:1039540:1039935 [3] NCCL INFO Channel 02/0 : 59[3] -> 58[2] via P2P/IPC
+ip-26-0-166-125:762822:763222 [7] NCCL INFO Channel 06/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-166-244:897247:897642 [7] NCCL INFO Channel 06/0 : 47[7] -> 46[6] via P2P/IPC
+ip-26-0-166-244:897245:897638 [5] NCCL INFO Channel 12/0 : 45[5] -> 44[4] via P2P/IPC
+ip-26-0-167-9:1039540:1039935 [3] NCCL INFO Channel 08/0 : 59[3] -> 58[2] via P2P/IPC
+ip-26-0-166-15:127294:127687 [4] NCCL INFO Channel 08/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-166-125:762819:763221 [4] NCCL INFO Channel 08/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-166-214:624946:625337 [3] NCCL INFO Channel 02/0 : 35[3] -> 34[2] via P2P/IPC
+ip-26-0-166-214:624947:625339 [4] NCCL INFO Channel 13/0 : 36[4] -> 35[3] via P2P/IPC
+ip-26-0-162-46:702884:703279 [4] NCCL INFO Channel 08/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-167-9:1039541:1039937 [4] NCCL INFO Channel 06/0 : 60[4] -> 59[3] via P2P/IPC
+ip-26-0-167-9:1039540:1039935 [3] NCCL INFO Channel 10/0 : 59[3] -> 58[2] via P2P/IPC
+ip-26-0-166-15:127297:127691 [7] NCCL INFO Channel 15/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-166-125:762822:763222 [7] NCCL INFO Channel 08/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-166-214:624950:625342 [7] NCCL INFO Channel 00/0 : 39[7] -> 38[6] via P2P/IPC
+ip-26-0-165-213:9628:10023 [3] NCCL INFO Channel 00/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-167-9:1039541:1039937 [4] NCCL INFO Channel 07/0 : 60[4] -> 59[3] via P2P/IPC
+ip-26-0-166-15:127294:127687 [4] NCCL INFO Channel 11/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-165-213:9629:10021 [4] NCCL INFO Channel 13/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-166-214:624946:625337 [3] NCCL INFO Channel 08/0 : 35[3] -> 34[2] via P2P/IPC
+ip-26-0-166-36:104433:104825 [4] NCCL INFO Channel 07/0 : 52[4] -> 51[3] via P2P/IPC
+ip-26-0-166-125:762819:763221 [4] NCCL INFO Channel 11/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-165-213:9632:10026 [7] NCCL INFO Channel 08/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-165-213:9628:10023 [3] NCCL INFO Channel 02/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-167-9:1039544:1039931 [7] NCCL INFO Channel 14/0 : 63[7] -> 56[0] via P2P/IPC
+ip-26-0-166-214:624947:625339 [4] NCCL INFO Channel 14/0 : 36[4] -> 35[3] via P2P/IPC
+ip-26-0-167-9:1039541:1039937 [4] NCCL INFO Channel 08/0 : 60[4] -> 59[3] via P2P/IPC
+ip-26-0-162-46:702885:703278 [5] NCCL INFO Channel 00/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-214:624950:625342 [7] NCCL INFO Channel 06/0 : 39[7] -> 38[6] via P2P/IPC
+ip-26-0-166-36:104436:104824 [7] NCCL INFO Channel 06/0 : 55[7] -> 54[6] via P2P/IPC
+ip-26-0-165-213:9629:10021 [4] NCCL INFO Channel 14/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-166-36:104434:104828 [5] NCCL INFO Channel 12/0 : 53[5] -> 52[4] via P2P/IPC
+ip-26-0-166-214:624947:625339 [4] NCCL INFO Channel 15/0 : 36[4] -> 35[3] via P2P/IPC
+ip-26-0-165-213:9628:10023 [3] NCCL INFO Channel 08/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-166-214:624950:625342 [7] NCCL INFO Channel 08/0 : 39[7] -> 38[6] via P2P/IPC
+ip-26-0-162-46:702885:703278 [5] NCCL INFO Channel 04/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-213:9632:10026 [7] NCCL INFO Channel 14/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-165-213:9629:10021 [4] NCCL INFO Channel 15/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-166-244:897244:897636 [4] NCCL INFO Channel 06/0 : 44[4] -> 43[3] via P2P/IPC
+ip-26-0-166-214:624950:625342 [7] NCCL INFO Channel 14/0 : 39[7] -> 38[6] via P2P/IPC
+ip-26-0-162-46:702885:703278 [5] NCCL INFO Channel 08/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-162-46:702884:703279 [4] NCCL INFO Channel 11/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-165-213:9628:10023 [3] NCCL INFO Channel 10/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-166-15:127294:127687 [4] NCCL INFO Channel 13/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-162-46:702885:703278 [5] NCCL INFO Channel 12/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-167-9:1039544:1039931 [7] NCCL INFO Channel 15/0 : 63[7] -> 56[0] via P2P/IPC
+ip-26-0-162-46:702884:703279 [4] NCCL INFO Channel 13/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-162-46:702884:703279 [4] NCCL INFO Channel 14/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-162-46:702887:703282 [7] NCCL INFO Channel 00/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-162-46:702883:703283 [3] NCCL INFO Channel 00/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-167-9:1039541:1039937 [4] NCCL INFO Channel 11/0 : 60[4] -> 59[3] via P2P/IPC
+ip-26-0-166-15:127294:127687 [4] NCCL INFO Channel 14/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-166-125:762822:763222 [7] NCCL INFO Channel 14/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-166-15:127294:127687 [4] NCCL INFO Channel 15/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-166-15:127293:127689 [3] NCCL INFO Channel 00/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-166-15:127297:127691 [7] NCCL INFO Channel 00/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-162-46:702884:703279 [4] NCCL INFO Channel 15/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-166-36:104433:104825 [4] NCCL INFO Channel 08/0 : 52[4] -> 51[3] via P2P/IPC
+ip-26-0-166-36:104436:104824 [7] NCCL INFO Channel 08/0 : 55[7] -> 54[6] via P2P/IPC
+ip-26-0-162-46:702883:703283 [3] NCCL INFO Channel 02/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-125:762818:763220 [3] NCCL INFO Channel 00/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-162-46:702887:703282 [7] NCCL INFO Channel 06/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-166-15:127297:127691 [7] NCCL INFO Channel 06/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-167-9:1039544:1039931 [7] NCCL INFO Channel 00/0 : 63[7] -> 62[6] via P2P/IPC
+ip-26-0-162-46:702883:703283 [3] NCCL INFO Channel 08/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-244:897247:897642 [7] NCCL INFO Channel 08/0 : 47[7] -> 46[6] via P2P/IPC
+ip-26-0-166-244:897247:897642 [7] NCCL INFO Channel 14/0 : 47[7] -> 46[6] via P2P/IPC
+ip-26-0-162-46:702887:703282 [7] NCCL INFO Channel 08/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-166-15:127293:127689 [3] NCCL INFO Channel 02/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-166-214:624946:625337 [3] NCCL INFO Channel 10/0 : 35[3] -> 34[2] via P2P/IPC
+ip-26-0-166-125:762818:763220 [3] NCCL INFO Channel 02/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-166-15:127297:127691 [7] NCCL INFO Channel 08/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-167-9:1039541:1039937 [4] NCCL INFO Channel 13/0 : 60[4] -> 59[3] via P2P/IPC
+ip-26-0-167-9:1039544:1039931 [7] NCCL INFO Channel 06/0 : 63[7] -> 62[6] via P2P/IPC
+ip-26-0-167-9:1039541:1039937 [4] NCCL INFO Channel 14/0 : 60[4] -> 59[3] via P2P/IPC
+ip-26-0-166-15:127297:127691 [7] NCCL INFO Channel 14/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-166-36:104433:104825 [4] NCCL INFO Channel 11/0 : 52[4] -> 51[3] via P2P/IPC
+ip-26-0-166-15:127293:127689 [3] NCCL INFO Channel 08/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-166-244:897244:897636 [4] NCCL INFO Channel 07/0 : 44[4] -> 43[3] via P2P/IPC
+ip-26-0-162-46:702883:703283 [3] NCCL INFO Channel 10/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-36:104433:104825 [4] NCCL INFO Channel 13/0 : 52[4] -> 51[3] via P2P/IPC
+ip-26-0-166-36:104436:104824 [7] NCCL INFO Channel 14/0 : 55[7] -> 54[6] via P2P/IPC
+ip-26-0-167-9:1039544:1039931 [7] NCCL INFO Channel 08/0 : 63[7] -> 62[6] via P2P/IPC
+ip-26-0-167-9:1039541:1039937 [4] NCCL INFO Channel 15/0 : 60[4] -> 59[3] via P2P/IPC
+ip-26-0-162-46:702887:703282 [7] NCCL INFO Channel 14/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-166-36:104432:104827 [3] NCCL INFO Channel 00/0 : 51[3] -> 50[2] via P2P/IPC
+ip-26-0-166-36:104433:104825 [4] NCCL INFO Channel 14/0 : 52[4] -> 51[3] via P2P/IPC
+ip-26-0-166-244:897244:897636 [4] NCCL INFO Channel 08/0 : 44[4] -> 43[3] via P2P/IPC
+ip-26-0-167-9:1039544:1039931 [7] NCCL INFO Channel 14/0 : 63[7] -> 62[6] via P2P/IPC
+ip-26-0-166-125:762818:763220 [3] NCCL INFO Channel 08/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-166-125:762819:763221 [4] NCCL INFO Channel 13/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-166-36:104432:104827 [3] NCCL INFO Channel 02/0 : 51[3] -> 50[2] via P2P/IPC
+ip-26-0-166-15:127293:127689 [3] NCCL INFO Channel 10/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-166-36:104433:104825 [4] NCCL INFO Channel 15/0 : 52[4] -> 51[3] via P2P/IPC
+ip-26-0-166-244:897244:897636 [4] NCCL INFO Channel 11/0 : 44[4] -> 43[3] via P2P/IPC
+ip-26-0-166-36:104432:104827 [3] NCCL INFO Channel 08/0 : 51[3] -> 50[2] via P2P/IPC
+ip-26-0-166-125:762818:763220 [3] NCCL INFO Channel 10/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-166-125:762819:763221 [4] NCCL INFO Channel 14/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-166-244:897244:897636 [4] NCCL INFO Channel 13/0 : 44[4] -> 43[3] via P2P/IPC
+ip-26-0-166-244:897244:897636 [4] NCCL INFO Channel 14/0 : 44[4] -> 43[3] via P2P/IPC
+ip-26-0-166-125:762819:763221 [4] NCCL INFO Channel 15/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-166-244:897244:897636 [4] NCCL INFO Channel 15/0 : 44[4] -> 43[3] via P2P/IPC
+ip-26-0-166-36:104432:104827 [3] NCCL INFO Channel 10/0 : 51[3] -> 50[2] via P2P/IPC
+ip-26-0-166-244:897243:897637 [3] NCCL INFO Channel 00/0 : 43[3] -> 42[2] via P2P/IPC
+ip-26-0-166-244:897243:897637 [3] NCCL INFO Channel 02/0 : 43[3] -> 42[2] via P2P/IPC
+ip-26-0-166-244:897243:897637 [3] NCCL INFO Channel 08/0 : 43[3] -> 42[2] via P2P/IPC
+ip-26-0-166-244:897243:897637 [3] NCCL INFO Channel 10/0 : 43[3] -> 42[2] via P2P/IPC
+ip-26-0-167-9:1039542:1039936 [5] NCCL INFO Connected all trees
+ip-26-0-167-9:1039542:1039936 [5] NCCL INFO NVLS comm 0x907fcb0 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-162-46:702885:703278 [5] NCCL INFO Connected all trees
+ip-26-0-162-46:702885:703278 [5] NCCL INFO NVLS comm 0x8d96ff0 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-162-46:702880:703277 [0] NCCL INFO Connected all trees
+ip-26-0-167-9:1039541:1039937 [4] NCCL INFO Connected all trees
+ip-26-0-162-46:702880:703277 [0] NCCL INFO NVLS comm 0x9b71eb0 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-167-9:1039541:1039937 [4] NCCL INFO NVLS comm 0xa6c22f0 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-167-9:1039539:1039938 [2] NCCL INFO Connected all trees
+ip-26-0-167-9:1039539:1039938 [2] NCCL INFO NVLS comm 0x9341460 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-167-9:1039540:1039935 [3] NCCL INFO Connected all trees
+ip-26-0-167-9:1039540:1039935 [3] NCCL INFO NVLS comm 0x99bd170 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-167-9:1039537:1039932 [0] NCCL INFO Connected all trees
+ip-26-0-167-9:1039537:1039932 [0] NCCL INFO NVLS comm 0x9e4ce90 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-167-9:1039543:1039933 [6] NCCL INFO Connected all trees
+ip-26-0-167-9:1039543:1039933 [6] NCCL INFO NVLS comm 0xa191260 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-162-46:702882:703284 [2] NCCL INFO Connected all trees
+ip-26-0-162-46:702882:703284 [2] NCCL INFO NVLS comm 0x99f1a00 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-162-46:702887:703282 [7] NCCL INFO Connected all trees
+ip-26-0-167-9:1039544:1039931 [7] NCCL INFO Connected all trees
+ip-26-0-162-46:702887:703282 [7] NCCL INFO NVLS comm 0x8d87da0 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-167-9:1039544:1039931 [7] NCCL INFO NVLS comm 0x98c6e90 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-162-46:702886:703281 [6] NCCL INFO Connected all trees
+ip-26-0-162-46:702886:703281 [6] NCCL INFO NVLS comm 0x927c830 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-167-9:1039538:1039934 [1] NCCL INFO Connected all trees
+ip-26-0-162-46:702883:703283 [3] NCCL INFO Connected all trees
+ip-26-0-162-46:702881:703280 [1] NCCL INFO Connected all trees
+ip-26-0-167-9:1039538:1039934 [1] NCCL INFO NVLS comm 0x926a6f0 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-162-46:702883:703283 [3] NCCL INFO NVLS comm 0x87b1a30 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-162-46:702881:703280 [1] NCCL INFO NVLS comm 0x963a9a0 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-162-46:702884:703279 [4] NCCL INFO Connected all trees
+ip-26-0-162-46:702884:703279 [4] NCCL INFO NVLS comm 0x9142ba0 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-166-36:104429:104829 [0] NCCL INFO Connected all trees
+ip-26-0-166-36:104429:104829 [0] NCCL INFO NVLS comm 0xa69aaf0 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-213:9625:10020 [0] NCCL INFO Connected all trees
+ip-26-0-165-213:9625:10020 [0] NCCL INFO NVLS comm 0x901a8b0 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-166-244:897247:897642 [7] NCCL INFO Connected all trees
+ip-26-0-166-244:897247:897642 [7] NCCL INFO NVLS comm 0x9be8840 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-213:9626:10025 [1] NCCL INFO Connected all trees
+ip-26-0-165-213:9626:10025 [1] NCCL INFO NVLS comm 0x95f5af0 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-166-244:897240:897640 [0] NCCL INFO Connected all trees
+ip-26-0-166-244:897240:897640 [0] NCCL INFO NVLS comm 0x95571b0 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-166-36:104436:104824 [7] NCCL INFO Connected all trees
+ip-26-0-166-36:104436:104824 [7] NCCL INFO NVLS comm 0x99716e0 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-166-125:762815:763225 [0] NCCL INFO Connected all trees
+ip-26-0-165-213:9630:10024 [5] NCCL INFO Connected all trees
+ip-26-0-166-125:762815:763225 [0] NCCL INFO NVLS comm 0x88082b0 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-213:9632:10026 [7] NCCL INFO Connected all trees
+ip-26-0-166-244:897246:897641 [6] NCCL INFO Connected all trees
+ip-26-0-165-213:9630:10024 [5] NCCL INFO NVLS comm 0x89d16b0 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-213:9632:10026 [7] NCCL INFO NVLS comm 0x8f2c3f0 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-166-244:897246:897641 [6] NCCL INFO NVLS comm 0xa665d00 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-213:9631:10022 [6] NCCL INFO Connected all trees
+ip-26-0-166-244:897241:897635 [1] NCCL INFO Connected all trees
+ip-26-0-166-15:127291:127690 [1] NCCL INFO Connected all trees
+ip-26-0-165-213:9629:10021 [4] NCCL INFO Connected all trees
+ip-26-0-165-213:9631:10022 [6] NCCL INFO NVLS comm 0x9d025f0 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-166-244:897241:897635 [1] NCCL INFO NVLS comm 0xa3fa260 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-166-15:127291:127690 [1] NCCL INFO NVLS comm 0xa457d40 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-213:9629:10021 [4] NCCL INFO NVLS comm 0x8b51550 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-166-36:104435:104826 [6] NCCL INFO Connected all trees
+ip-26-0-166-36:104435:104826 [6] NCCL INFO NVLS comm 0x90001f0 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-166-36:104431:104831 [2] NCCL INFO Connected all trees
+ip-26-0-166-36:104431:104831 [2] NCCL INFO NVLS comm 0x9a41180 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-166-125:762822:763222 [7] NCCL INFO Connected all trees
+ip-26-0-166-125:762822:763222 [7] NCCL INFO NVLS comm 0xa21d6a0 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-166-15:127297:127691 [7] NCCL INFO Connected all trees
+ip-26-0-166-15:127297:127691 [7] NCCL INFO NVLS comm 0x9fff530 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-166-125:762821:763224 [6] NCCL INFO Connected all trees
+ip-26-0-166-125:762821:763224 [6] NCCL INFO NVLS comm 0x8ae3f70 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-166-36:104430:104830 [1] NCCL INFO Connected all trees
+ip-26-0-166-36:104430:104830 [1] NCCL INFO NVLS comm 0x9eb3ae0 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-166-36:104432:104827 [3] NCCL INFO Connected all trees
+ip-26-0-166-36:104432:104827 [3] NCCL INFO NVLS comm 0x8a395f0 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-166-214:624945:625340 [2] NCCL INFO Connected all trees
+ip-26-0-166-214:624945:625340 [2] NCCL INFO NVLS comm 0x99ed060 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-166-214:624949:625344 [6] NCCL INFO Connected all trees
+ip-26-0-166-214:624949:625344 [6] NCCL INFO NVLS comm 0x8cb8cf0 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-166-125:762818:763220 [3] NCCL INFO Connected all trees
+ip-26-0-165-213:9627:10027 [2] NCCL INFO Connected all trees
+ip-26-0-166-125:762817:763223 [2] NCCL INFO Connected all trees
+ip-26-0-166-244:897244:897636 [4] NCCL INFO Connected all trees
+ip-26-0-166-125:762818:763220 [3] NCCL INFO NVLS comm 0x89715f0 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-166-125:762817:763223 [2] NCCL INFO NVLS comm 0xa633eb0 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-166-244:897244:897636 [4] NCCL INFO NVLS comm 0x90d7750 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-213:9627:10027 [2] NCCL INFO NVLS comm 0x8b55300 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-166-214:624950:625342 [7] NCCL INFO Connected all trees
+ip-26-0-166-125:762816:763218 [1] NCCL INFO Connected all trees
+ip-26-0-166-36:104433:104825 [4] NCCL INFO Connected all trees
+ip-26-0-166-125:762819:763221 [4] NCCL INFO Connected all trees
+ip-26-0-166-214:624950:625342 [7] NCCL INFO NVLS comm 0x8fdb6f0 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-166-36:104433:104825 [4] NCCL INFO NVLS comm 0xa48b0f0 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-166-125:762816:763218 [1] NCCL INFO NVLS comm 0x9836d20 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-166-125:762819:763221 [4] NCCL INFO NVLS comm 0x961bfa0 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-166-36:104434:104828 [5] NCCL INFO Connected all trees
+ip-26-0-166-214:624943:625338 [0] NCCL INFO Connected all trees
+ip-26-0-166-15:127295:127686 [5] NCCL INFO Connected all trees
+ip-26-0-166-36:104434:104828 [5] NCCL INFO NVLS comm 0x8ca8b60 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-166-214:624943:625338 [0] NCCL INFO NVLS comm 0x967f900 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-166-15:127295:127686 [5] NCCL INFO NVLS comm 0x8cd1ce0 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-166-15:127294:127687 [4] NCCL INFO Connected all trees
+ip-26-0-166-125:762820:763219 [5] NCCL INFO Connected all trees
+ip-26-0-166-15:127294:127687 [4] NCCL INFO NVLS comm 0x91e17f0 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-166-125:762820:763219 [5] NCCL INFO NVLS comm 0x953c5f0 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-166-15:127296:127688 [6] NCCL INFO Connected all trees
+ip-26-0-166-15:127296:127688 [6] NCCL INFO NVLS comm 0x98e93f0 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-213:9628:10023 [3] NCCL INFO Connected all trees
+ip-26-0-166-244:897243:897637 [3] NCCL INFO Connected all trees
+ip-26-0-166-244:897243:897637 [3] NCCL INFO NVLS comm 0x8c4b580 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-213:9628:10023 [3] NCCL INFO NVLS comm 0x9540830 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-166-15:127292:127692 [2] NCCL INFO Connected all trees
+ip-26-0-166-15:127292:127692 [2] NCCL INFO NVLS comm 0x94433a0 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-166-15:127290:127685 [0] NCCL INFO Connected all trees
+ip-26-0-166-244:897242:897639 [2] NCCL INFO Connected all trees
+ip-26-0-166-15:127290:127685 [0] NCCL INFO NVLS comm 0x8bd7da0 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-166-244:897245:897638 [5] NCCL INFO Connected all trees
+ip-26-0-166-15:127293:127689 [3] NCCL INFO Connected all trees
+ip-26-0-166-244:897242:897639 [2] NCCL INFO NVLS comm 0x8943e30 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-166-244:897245:897638 [5] NCCL INFO NVLS comm 0x9cf54b0 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-166-15:127293:127689 [3] NCCL INFO NVLS comm 0xa34b190 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-166-214:624947:625339 [4] NCCL INFO Connected all trees
+ip-26-0-166-214:624947:625339 [4] NCCL INFO NVLS comm 0xa228a50 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-166-214:624946:625337 [3] NCCL INFO Connected all trees
+ip-26-0-166-214:624946:625337 [3] NCCL INFO NVLS comm 0xa2847d0 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-166-214:624944:625341 [1] NCCL INFO Connected all trees
+ip-26-0-166-214:624944:625341 [1] NCCL INFO NVLS comm 0x8750fa0 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-167-9:1039538:1039934 [1] NCCL INFO Channel 00/0 : 49[1] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-167-9:1039542:1039936 [5] NCCL INFO Channel 00/0 : 53[5] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-167-9:1039538:1039934 [1] NCCL INFO Channel 02/0 : 49[1] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-167-9:1039540:1039935 [3] NCCL INFO Channel 00/0 : 51[3] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-167-9:1039542:1039936 [5] NCCL INFO Channel 02/0 : 53[5] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-167-9:1039538:1039934 [1] NCCL INFO Channel 04/0 : 49[1] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-167-9:1039540:1039935 [3] NCCL INFO Channel 02/0 : 51[3] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-167-9:1039542:1039936 [5] NCCL INFO Channel 04/0 : 53[5] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-167-9:1039541:1039937 [4] NCCL INFO Channel 00/0 : 52[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-167-9:1039543:1039933 [6] NCCL INFO Channel 00/0 : 54[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-167-9:1039540:1039935 [3] NCCL INFO Channel 04/0 : 51[3] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-167-9:1039538:1039934 [1] NCCL INFO Channel 06/0 : 49[1] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-167-9:1039542:1039936 [5] NCCL INFO Channel 06/0 : 53[5] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-167-9:1039541:1039937 [4] NCCL INFO Channel 02/0 : 52[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-167-9:1039537:1039932 [0] NCCL INFO Channel 02/0 : 48[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-167-9:1039543:1039933 [6] NCCL INFO Channel 02/0 : 54[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-167-9:1039539:1039938 [2] NCCL INFO Channel 00/0 : 50[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-167-9:1039538:1039934 [1] NCCL INFO Channel 08/0 : 49[1] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-167-9:1039540:1039935 [3] NCCL INFO Channel 06/0 : 51[3] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-167-9:1039542:1039936 [5] NCCL INFO Channel 08/0 : 53[5] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-167-9:1039541:1039937 [4] NCCL INFO Channel 06/0 : 52[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-167-9:1039537:1039932 [0] NCCL INFO Channel 04/0 : 48[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-167-9:1039543:1039933 [6] NCCL INFO Channel 04/0 : 54[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-167-9:1039539:1039938 [2] NCCL INFO Channel 04/0 : 50[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-167-9:1039538:1039934 [1] NCCL INFO Channel 10/0 : 49[1] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-167-9:1039540:1039935 [3] NCCL INFO Channel 08/0 : 51[3] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-167-9:1039542:1039936 [5] NCCL INFO Channel 10/0 : 53[5] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-167-9:1039541:1039937 [4] NCCL INFO Channel 08/0 : 52[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-167-9:1039537:1039932 [0] NCCL INFO Channel 06/0 : 48[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-167-9:1039543:1039933 [6] NCCL INFO Channel 08/0 : 54[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-167-9:1039539:1039938 [2] NCCL INFO Channel 06/0 : 50[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-167-9:1039538:1039934 [1] NCCL INFO Channel 12/0 : 49[1] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-167-9:1039540:1039935 [3] NCCL INFO Channel 10/0 : 51[3] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-167-9:1039541:1039937 [4] NCCL INFO Channel 10/0 : 52[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-167-9:1039537:1039932 [0] NCCL INFO Channel 08/0 : 48[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-167-9:1039542:1039936 [5] NCCL INFO Channel 12/0 : 53[5] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-167-9:1039543:1039933 [6] NCCL INFO Channel 10/0 : 54[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-167-9:1039539:1039938 [2] NCCL INFO Channel 08/0 : 50[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-167-9:1039540:1039935 [3] NCCL INFO Channel 12/0 : 51[3] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-167-9:1039538:1039934 [1] NCCL INFO Channel 14/0 : 49[1] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-167-9:1039541:1039937 [4] NCCL INFO Channel 12/0 : 52[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-167-9:1039543:1039933 [6] NCCL INFO Channel 12/0 : 54[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-167-9:1039542:1039936 [5] NCCL INFO Channel 14/0 : 53[5] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-167-9:1039537:1039932 [0] NCCL INFO Channel 10/0 : 48[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-167-9:1039539:1039938 [2] NCCL INFO Channel 10/0 : 50[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-167-9:1039540:1039935 [3] NCCL INFO Channel 14/0 : 51[3] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-167-9:1039541:1039937 [4] NCCL INFO Channel 14/0 : 52[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-167-9:1039543:1039933 [6] NCCL INFO Channel 14/0 : 54[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-167-9:1039537:1039932 [0] NCCL INFO Channel 12/0 : 48[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-167-9:1039539:1039938 [2] NCCL INFO Channel 12/0 : 50[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-167-9:1039537:1039932 [0] NCCL INFO Channel 14/0 : 48[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-167-9:1039539:1039938 [2] NCCL INFO Channel 14/0 : 50[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-167-9:1039544:1039931 [7] NCCL INFO Channel 00/0 : 55[7] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-167-9:1039544:1039931 [7] NCCL INFO Channel 02/0 : 55[7] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-167-9:1039544:1039931 [7] NCCL INFO Channel 04/0 : 55[7] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-167-9:1039544:1039931 [7] NCCL INFO Channel 06/0 : 55[7] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-167-9:1039544:1039931 [7] NCCL INFO Channel 08/0 : 55[7] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-167-9:1039544:1039931 [7] NCCL INFO Channel 10/0 : 55[7] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-167-9:1039544:1039931 [7] NCCL INFO Channel 12/0 : 55[7] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-167-9:1039544:1039931 [7] NCCL INFO Channel 14/0 : 55[7] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624948:625343 [5] NCCL INFO Connected all trees
+ip-26-0-166-214:624948:625343 [5] NCCL INFO NVLS comm 0xa2a1a50 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-162-46:702886:703281 [6] NCCL INFO Channel 01/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-46:702884:703279 [4] NCCL INFO Channel 01/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-46:702886:703281 [6] NCCL INFO Channel 03/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-46:702880:703277 [0] NCCL INFO Channel 01/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-46:702882:703284 [2] NCCL INFO Channel 01/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-46:702884:703279 [4] NCCL INFO Channel 03/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-46:702886:703281 [6] NCCL INFO Channel 05/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-46:702882:703284 [2] NCCL INFO Channel 03/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-46:702884:703279 [4] NCCL INFO Channel 05/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-46:702880:703277 [0] NCCL INFO Channel 03/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-46:702886:703281 [6] NCCL INFO Channel 07/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-46:702882:703284 [2] NCCL INFO Channel 05/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-46:702884:703279 [4] NCCL INFO Channel 07/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-46:702880:703277 [0] NCCL INFO Channel 05/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-46:702886:703281 [6] NCCL INFO Channel 09/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-46:702882:703284 [2] NCCL INFO Channel 07/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-46:702884:703279 [4] NCCL INFO Channel 09/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-46:702881:703280 [1] NCCL INFO Channel 01/0 : 1[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-46:702880:703277 [0] NCCL INFO Channel 07/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-46:702882:703284 [2] NCCL INFO Channel 09/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-46:702886:703281 [6] NCCL INFO Channel 11/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-46:702884:703279 [4] NCCL INFO Channel 11/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-46:702881:703280 [1] NCCL INFO Channel 03/0 : 1[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-46:702880:703277 [0] NCCL INFO Channel 09/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-46:702885:703278 [5] NCCL INFO Channel 01/0 : 5[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-46:702882:703284 [2] NCCL INFO Channel 11/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-46:702886:703281 [6] NCCL INFO Channel 13/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-46:702881:703280 [1] NCCL INFO Channel 05/0 : 1[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-46:702884:703279 [4] NCCL INFO Channel 13/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-46:702883:703283 [3] NCCL INFO Channel 01/0 : 3[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-46:702880:703277 [0] NCCL INFO Channel 11/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-46:702882:703284 [2] NCCL INFO Channel 13/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-46:702885:703278 [5] NCCL INFO Channel 03/0 : 5[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-46:702886:703281 [6] NCCL INFO Channel 15/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-46:702881:703280 [1] NCCL INFO Channel 07/0 : 1[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-46:702883:703283 [3] NCCL INFO Channel 03/0 : 3[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-46:702884:703279 [4] NCCL INFO Channel 15/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-46:702880:703277 [0] NCCL INFO Channel 13/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-46:702882:703284 [2] NCCL INFO Channel 15/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-46:702885:703278 [5] NCCL INFO Channel 05/0 : 5[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-46:702881:703280 [1] NCCL INFO Channel 11/0 : 1[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-46:702883:703283 [3] NCCL INFO Channel 05/0 : 3[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-46:702880:703277 [0] NCCL INFO Channel 15/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-46:702881:703280 [1] NCCL INFO Channel 13/0 : 1[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-46:702885:703278 [5] NCCL INFO Channel 07/0 : 5[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-46:702883:703283 [3] NCCL INFO Channel 07/0 : 3[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-46:702881:703280 [1] NCCL INFO Channel 15/0 : 1[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-46:702885:703278 [5] NCCL INFO Channel 09/0 : 5[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-46:702883:703283 [3] NCCL INFO Channel 09/0 : 3[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-46:702885:703278 [5] NCCL INFO Channel 11/0 : 5[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-46:702883:703283 [3] NCCL INFO Channel 13/0 : 3[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-46:702885:703278 [5] NCCL INFO Channel 15/0 : 5[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-46:702883:703283 [3] NCCL INFO Channel 15/0 : 3[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-46:702887:703282 [7] NCCL INFO Channel 01/0 : 7[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-46:702887:703282 [7] NCCL INFO Channel 03/0 : 7[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-46:702887:703282 [7] NCCL INFO Channel 05/0 : 7[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-46:702887:703282 [7] NCCL INFO Channel 07/0 : 7[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-46:702887:703282 [7] NCCL INFO Channel 09/0 : 7[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-46:702887:703282 [7] NCCL INFO Channel 11/0 : 7[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-46:702887:703282 [7] NCCL INFO Channel 13/0 : 7[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104429:104829 [0] NCCL INFO Channel 01/0 : 40[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104429:104829 [0] NCCL INFO Channel 02/0 : 40[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104429:104829 [0] NCCL INFO Channel 03/0 : 40[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104432:104827 [3] NCCL INFO Channel 00/0 : 43[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104429:104829 [0] NCCL INFO Channel 04/0 : 40[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104433:104825 [4] NCCL INFO Channel 00/0 : 44[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104429:104829 [0] NCCL INFO Channel 05/0 : 40[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104434:104828 [5] NCCL INFO Channel 00/0 : 45[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104432:104827 [3] NCCL INFO Channel 01/0 : 43[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104436:104824 [7] NCCL INFO Channel 00/0 : 47[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104433:104825 [4] NCCL INFO Channel 01/0 : 44[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104431:104831 [2] NCCL INFO Channel 00/0 : 42[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104429:104829 [0] NCCL INFO Channel 06/0 : 40[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104436:104824 [7] NCCL INFO Channel 01/0 : 47[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104434:104828 [5] NCCL INFO Channel 01/0 : 45[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104432:104827 [3] NCCL INFO Channel 02/0 : 43[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104435:104826 [6] NCCL INFO Channel 00/0 : 46[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104430:104830 [1] NCCL INFO Channel 00/0 : 41[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104433:104825 [4] NCCL INFO Channel 02/0 : 44[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104431:104831 [2] NCCL INFO Channel 01/0 : 42[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104429:104829 [0] NCCL INFO Channel 07/0 : 40[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104436:104824 [7] NCCL INFO Channel 02/0 : 47[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104434:104828 [5] NCCL INFO Channel 02/0 : 45[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104435:104826 [6] NCCL INFO Channel 01/0 : 46[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104432:104827 [3] NCCL INFO Channel 04/0 : 43[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104433:104825 [4] NCCL INFO Channel 03/0 : 44[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104430:104830 [1] NCCL INFO Channel 02/0 : 41[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104431:104831 [2] NCCL INFO Channel 03/0 : 42[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104429:104829 [0] NCCL INFO Channel 09/0 : 40[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104434:104828 [5] NCCL INFO Channel 03/0 : 45[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104436:104824 [7] NCCL INFO Channel 03/0 : 47[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104435:104826 [6] NCCL INFO Channel 02/0 : 46[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104429:104829 [0] NCCL INFO Channel 10/0 : 40[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104433:104825 [4] NCCL INFO Channel 05/0 : 44[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104432:104827 [3] NCCL INFO Channel 05/0 : 43[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104431:104831 [2] NCCL INFO Channel 04/0 : 42[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104430:104830 [1] NCCL INFO Channel 03/0 : 41[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104434:104828 [5] NCCL INFO Channel 04/0 : 45[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104436:104824 [7] NCCL INFO Channel 04/0 : 47[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104435:104826 [6] NCCL INFO Channel 03/0 : 46[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104429:104829 [0] NCCL INFO Channel 11/0 : 40[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104433:104825 [4] NCCL INFO Channel 06/0 : 44[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104434:104828 [5] NCCL INFO Channel 06/0 : 45[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104436:104824 [7] NCCL INFO Channel 05/0 : 47[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104431:104831 [2] NCCL INFO Channel 05/0 : 42[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104430:104830 [1] NCCL INFO Channel 04/0 : 41[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104432:104827 [3] NCCL INFO Channel 06/0 : 43[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104435:104826 [6] NCCL INFO Channel 04/0 : 46[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104429:104829 [0] NCCL INFO Channel 12/0 : 40[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104433:104825 [4] NCCL INFO Channel 07/0 : 44[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104434:104828 [5] NCCL INFO Channel 07/0 : 45[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104436:104824 [7] NCCL INFO Channel 06/0 : 47[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104435:104826 [6] NCCL INFO Channel 05/0 : 46[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104430:104830 [1] NCCL INFO Channel 05/0 : 41[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104432:104827 [3] NCCL INFO Channel 07/0 : 43[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104429:104829 [0] NCCL INFO Channel 13/0 : 40[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104433:104825 [4] NCCL INFO Channel 08/0 : 44[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104431:104831 [2] NCCL INFO Channel 06/0 : 42[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104434:104828 [5] NCCL INFO Channel 08/0 : 45[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104436:104824 [7] NCCL INFO Channel 08/0 : 47[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104435:104826 [6] NCCL INFO Channel 07/0 : 46[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104430:104830 [1] NCCL INFO Channel 06/0 : 41[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104429:104829 [0] NCCL INFO Channel 14/0 : 40[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104433:104825 [4] NCCL INFO Channel 09/0 : 44[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104436:104824 [7] NCCL INFO Channel 09/0 : 47[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104432:104827 [3] NCCL INFO Channel 08/0 : 43[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104434:104828 [5] NCCL INFO Channel 09/0 : 45[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104431:104831 [2] NCCL INFO Channel 07/0 : 42[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104435:104826 [6] NCCL INFO Channel 08/0 : 46[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104430:104830 [1] NCCL INFO Channel 07/0 : 41[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104429:104829 [0] NCCL INFO Channel 15/0 : 40[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104433:104825 [4] NCCL INFO Channel 10/0 : 44[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104436:104824 [7] NCCL INFO Channel 10/0 : 47[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104434:104828 [5] NCCL INFO Channel 10/0 : 45[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104432:104827 [3] NCCL INFO Channel 09/0 : 43[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104435:104826 [6] NCCL INFO Channel 09/0 : 46[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104431:104831 [2] NCCL INFO Channel 08/0 : 42[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104430:104830 [1] NCCL INFO Channel 08/0 : 41[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104429:104829 [0] NCCL INFO Channel 02/0 : 48[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104433:104825 [4] NCCL INFO Channel 11/0 : 44[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104436:104824 [7] NCCL INFO Channel 11/0 : 47[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104434:104828 [5] NCCL INFO Channel 11/0 : 45[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104432:104827 [3] NCCL INFO Channel 10/0 : 43[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104435:104826 [6] NCCL INFO Channel 10/0 : 46[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104429:104829 [0] NCCL INFO Channel 04/0 : 48[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104433:104825 [4] NCCL INFO Channel 13/0 : 44[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104436:104824 [7] NCCL INFO Channel 12/0 : 47[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104431:104831 [2] NCCL INFO Channel 09/0 : 42[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104430:104830 [1] NCCL INFO Channel 10/0 : 41[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104434:104828 [5] NCCL INFO Channel 12/0 : 45[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104432:104827 [3] NCCL INFO Channel 12/0 : 43[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104429:104829 [0] NCCL INFO Channel 06/0 : 48[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104433:104825 [4] NCCL INFO Channel 14/0 : 44[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104436:104824 [7] NCCL INFO Channel 13/0 : 47[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104431:104831 [2] NCCL INFO Channel 11/0 : 42[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104434:104828 [5] NCCL INFO Channel 14/0 : 45[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104435:104826 [6] NCCL INFO Channel 11/0 : 46[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104432:104827 [3] NCCL INFO Channel 13/0 : 43[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104429:104829 [0] NCCL INFO Channel 08/0 : 48[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104430:104830 [1] NCCL INFO Channel 11/0 : 41[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104433:104825 [4] NCCL INFO Channel 15/0 : 44[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104436:104824 [7] NCCL INFO Channel 14/0 : 47[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104434:104828 [5] NCCL INFO Channel 15/0 : 45[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104431:104831 [2] NCCL INFO Channel 12/0 : 42[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104435:104826 [6] NCCL INFO Channel 12/0 : 46[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104429:104829 [0] NCCL INFO Channel 10/0 : 48[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104430:104830 [1] NCCL INFO Channel 12/0 : 41[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104432:104827 [3] NCCL INFO Channel 14/0 : 43[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104433:104825 [4] NCCL INFO Channel 00/0 : 52[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104436:104824 [7] NCCL INFO Channel 00/0 : 55[7] -> 63[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104434:104828 [5] NCCL INFO Channel 00/0 : 53[5] -> 61[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104431:104831 [2] NCCL INFO Channel 13/0 : 42[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104429:104829 [0] NCCL INFO Channel 12/0 : 48[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104435:104826 [6] NCCL INFO Channel 13/0 : 46[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104433:104825 [4] NCCL INFO Channel 02/0 : 52[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104432:104827 [3] NCCL INFO Channel 15/0 : 43[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104430:104830 [1] NCCL INFO Channel 13/0 : 41[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104436:104824 [7] NCCL INFO Channel 02/0 : 55[7] -> 63[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104434:104828 [5] NCCL INFO Channel 02/0 : 53[5] -> 61[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104431:104831 [2] NCCL INFO Channel 14/0 : 42[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104429:104829 [0] NCCL INFO Channel 14/0 : 48[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104433:104825 [4] NCCL INFO Channel 06/0 : 52[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104435:104826 [6] NCCL INFO Channel 15/0 : 46[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104436:104824 [7] NCCL INFO Channel 04/0 : 55[7] -> 63[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104434:104828 [5] NCCL INFO Channel 04/0 : 53[5] -> 61[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104432:104827 [3] NCCL INFO Channel 00/0 : 51[3] -> 59[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104430:104830 [1] NCCL INFO Channel 14/0 : 41[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104431:104831 [2] NCCL INFO Channel 15/0 : 42[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104433:104825 [4] NCCL INFO Channel 08/0 : 52[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104435:104826 [6] NCCL INFO Channel 00/0 : 54[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104434:104828 [5] NCCL INFO Channel 06/0 : 53[5] -> 61[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104436:104824 [7] NCCL INFO Channel 06/0 : 55[7] -> 63[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104430:104830 [1] NCCL INFO Channel 15/0 : 41[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104432:104827 [3] NCCL INFO Channel 02/0 : 51[3] -> 59[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104431:104831 [2] NCCL INFO Channel 00/0 : 50[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104433:104825 [4] NCCL INFO Channel 10/0 : 52[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762822:763222 [7] NCCL INFO Channel 00/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104435:104826 [6] NCCL INFO Channel 02/0 : 54[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104434:104828 [5] NCCL INFO Channel 08/0 : 53[5] -> 61[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762820:763219 [5] NCCL INFO Channel 00/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104436:104824 [7] NCCL INFO Channel 08/0 : 55[7] -> 63[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104430:104830 [1] NCCL INFO Channel 00/0 : 49[1] -> 57[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104433:104825 [4] NCCL INFO Channel 12/0 : 52[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104432:104827 [3] NCCL INFO Channel 04/0 : 51[3] -> 59[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104431:104831 [2] NCCL INFO Channel 04/0 : 50[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104435:104826 [6] NCCL INFO Channel 04/0 : 54[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104434:104828 [5] NCCL INFO Channel 10/0 : 53[5] -> 61[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762822:763222 [7] NCCL INFO Channel 01/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104436:104824 [7] NCCL INFO Channel 10/0 : 55[7] -> 63[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762820:763219 [5] NCCL INFO Channel 01/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762822:763222 [7] NCCL INFO Channel 02/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762818:763220 [3] NCCL INFO Channel 00/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104433:104825 [4] NCCL INFO Channel 14/0 : 52[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104432:104827 [3] NCCL INFO Channel 06/0 : 51[3] -> 59[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104430:104830 [1] NCCL INFO Channel 02/0 : 49[1] -> 57[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104435:104826 [6] NCCL INFO Channel 08/0 : 54[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104434:104828 [5] NCCL INFO Channel 12/0 : 53[5] -> 61[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-167-9:1039537:1039932 [0] NCCL INFO Channel 01/0 : 24[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762820:763219 [5] NCCL INFO Channel 02/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762822:763222 [7] NCCL INFO Channel 03/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104431:104831 [2] NCCL INFO Channel 06/0 : 50[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104436:104824 [7] NCCL INFO Channel 12/0 : 55[7] -> 63[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762818:763220 [3] NCCL INFO Channel 01/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762816:763218 [1] NCCL INFO Channel 00/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762820:763219 [5] NCCL INFO Channel 03/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-167-9:1039537:1039932 [0] NCCL INFO Channel 03/0 : 24[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762822:763222 [7] NCCL INFO Channel 04/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104430:104830 [1] NCCL INFO Channel 04/0 : 49[1] -> 57[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104432:104827 [3] NCCL INFO Channel 08/0 : 51[3] -> 59[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762818:763220 [3] NCCL INFO Channel 02/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-167-9:1039537:1039932 [0] NCCL INFO Channel 05/0 : 24[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104434:104828 [5] NCCL INFO Channel 14/0 : 53[5] -> 61[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104435:104826 [6] NCCL INFO Channel 10/0 : 54[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104436:104824 [7] NCCL INFO Channel 14/0 : 55[7] -> 63[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104431:104831 [2] NCCL INFO Channel 08/0 : 50[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762820:763219 [5] NCCL INFO Channel 04/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104432:104827 [3] NCCL INFO Channel 10/0 : 51[3] -> 59[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762816:763218 [1] NCCL INFO Channel 02/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762822:763222 [7] NCCL INFO Channel 05/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-167-9:1039537:1039932 [0] NCCL INFO Channel 07/0 : 24[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104430:104830 [1] NCCL INFO Channel 06/0 : 49[1] -> 57[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104435:104826 [6] NCCL INFO Channel 12/0 : 54[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104431:104831 [2] NCCL INFO Channel 10/0 : 50[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104432:104827 [3] NCCL INFO Channel 12/0 : 51[3] -> 59[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762819:763221 [4] NCCL INFO Channel 00/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762818:763220 [3] NCCL INFO Channel 04/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762820:763219 [5] NCCL INFO Channel 06/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762816:763218 [1] NCCL INFO Channel 03/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762822:763222 [7] NCCL INFO Channel 06/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762817:763223 [2] NCCL INFO Channel 00/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762819:763221 [4] NCCL INFO Channel 01/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-167-9:1039537:1039932 [0] NCCL INFO Channel 09/0 : 24[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104430:104830 [1] NCCL INFO Channel 08/0 : 49[1] -> 57[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104435:104826 [6] NCCL INFO Channel 14/0 : 54[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104431:104831 [2] NCCL INFO Channel 12/0 : 50[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104432:104827 [3] NCCL INFO Channel 14/0 : 51[3] -> 59[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762815:763225 [0] NCCL INFO Channel 01/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762820:763219 [5] NCCL INFO Channel 07/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762818:763220 [3] NCCL INFO Channel 05/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104430:104830 [1] NCCL INFO Channel 10/0 : 49[1] -> 57[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104431:104831 [2] NCCL INFO Channel 14/0 : 50[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-167-9:1039537:1039932 [0] NCCL INFO Channel 11/0 : 24[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762816:763218 [1] NCCL INFO Channel 04/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762821:763224 [6] NCCL INFO Channel 00/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762817:763223 [2] NCCL INFO Channel 01/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762819:763221 [4] NCCL INFO Channel 02/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762822:763222 [7] NCCL INFO Channel 08/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762815:763225 [0] NCCL INFO Channel 02/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104430:104830 [1] NCCL INFO Channel 12/0 : 49[1] -> 57[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-167-9:1039537:1039932 [0] NCCL INFO Channel 13/0 : 24[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762818:763220 [3] NCCL INFO Channel 06/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762820:763219 [5] NCCL INFO Channel 08/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104430:104830 [1] NCCL INFO Channel 14/0 : 49[1] -> 57[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762816:763218 [1] NCCL INFO Channel 05/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-167-9:1039537:1039932 [0] NCCL INFO Channel 15/0 : 24[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762817:763223 [2] NCCL INFO Channel 03/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762821:763224 [6] NCCL INFO Channel 01/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762819:763221 [4] NCCL INFO Channel 03/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762822:763222 [7] NCCL INFO Channel 09/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762815:763225 [0] NCCL INFO Channel 03/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762818:763220 [3] NCCL INFO Channel 07/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762820:763219 [5] NCCL INFO Channel 09/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762816:763218 [1] NCCL INFO Channel 06/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-167-9:1039537:1039932 [0] NCCL INFO Channel 01/0 : 56[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762817:763223 [2] NCCL INFO Channel 04/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762821:763224 [6] NCCL INFO Channel 02/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-167-9:1039544:1039931 [7] NCCL INFO Channel 01/0 : 31[7] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9630:10024 [5] NCCL INFO Channel 01/0 : 5[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762819:763221 [4] NCCL INFO Channel 05/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762822:763222 [7] NCCL INFO Channel 10/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762815:763225 [0] NCCL INFO Channel 04/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762818:763220 [3] NCCL INFO Channel 08/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-167-9:1039537:1039932 [0] NCCL INFO Channel 03/0 : 56[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762816:763218 [1] NCCL INFO Channel 07/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762820:763219 [5] NCCL INFO Channel 10/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762817:763223 [2] NCCL INFO Channel 05/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762821:763224 [6] NCCL INFO Channel 03/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9630:10024 [5] NCCL INFO Channel 03/0 : 5[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-167-9:1039544:1039931 [7] NCCL INFO Channel 03/0 : 31[7] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-167-9:1039543:1039933 [6] NCCL INFO Channel 01/0 : 30[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-167-9:1039537:1039932 [0] NCCL INFO Channel 05/0 : 56[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762819:763221 [4] NCCL INFO Channel 06/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-167-9:1039537:1039932 [0] NCCL INFO Channel 07/0 : 56[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9630:10024 [5] NCCL INFO Channel 05/0 : 5[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762822:763222 [7] NCCL INFO Channel 11/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762815:763225 [0] NCCL INFO Channel 05/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762818:763220 [3] NCCL INFO Channel 09/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762816:763218 [1] NCCL INFO Channel 08/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762820:763219 [5] NCCL INFO Channel 11/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762817:763223 [2] NCCL INFO Channel 06/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-167-9:1039544:1039931 [7] NCCL INFO Channel 05/0 : 31[7] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-167-9:1039543:1039933 [6] NCCL INFO Channel 03/0 : 30[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762821:763224 [6] NCCL INFO Channel 04/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9628:10023 [3] NCCL INFO Channel 01/0 : 3[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762819:763221 [4] NCCL INFO Channel 07/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-167-9:1039541:1039937 [4] NCCL INFO Channel 01/0 : 28[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-167-9:1039537:1039932 [0] NCCL INFO Channel 09/0 : 56[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-167-9:1039540:1039935 [3] NCCL INFO Channel 01/0 : 27[3] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9632:10026 [7] NCCL INFO Channel 01/0 : 7[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9630:10024 [5] NCCL INFO Channel 07/0 : 5[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-167-9:1039544:1039931 [7] NCCL INFO Channel 07/0 : 31[7] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-167-9:1039543:1039933 [6] NCCL INFO Channel 05/0 : 30[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9628:10023 [3] NCCL INFO Channel 03/0 : 3[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762822:763222 [7] NCCL INFO Channel 12/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762815:763225 [0] NCCL INFO Channel 06/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762818:763220 [3] NCCL INFO Channel 10/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762816:763218 [1] NCCL INFO Channel 10/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-167-9:1039538:1039934 [1] NCCL INFO Channel 01/0 : 25[1] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-167-9:1039537:1039932 [0] NCCL INFO Channel 11/0 : 56[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-167-9:1039541:1039937 [4] NCCL INFO Channel 03/0 : 28[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-167-9:1039540:1039935 [3] NCCL INFO Channel 03/0 : 27[3] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9632:10026 [7] NCCL INFO Channel 03/0 : 7[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9630:10024 [5] NCCL INFO Channel 09/0 : 5[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762817:763223 [2] NCCL INFO Channel 07/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762820:763219 [5] NCCL INFO Channel 12/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762819:763221 [4] NCCL INFO Channel 08/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762821:763224 [6] NCCL INFO Channel 05/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762815:763225 [0] NCCL INFO Channel 07/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762822:763222 [7] NCCL INFO Channel 13/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-167-9:1039543:1039933 [6] NCCL INFO Channel 07/0 : 30[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-167-9:1039544:1039931 [7] NCCL INFO Channel 09/0 : 31[7] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-167-9:1039542:1039936 [5] NCCL INFO Channel 01/0 : 29[5] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9628:10023 [3] NCCL INFO Channel 05/0 : 3[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-167-9:1039537:1039932 [0] NCCL INFO Channel 13/0 : 56[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9626:10025 [1] NCCL INFO Channel 01/0 : 1[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9631:10022 [6] NCCL INFO Channel 01/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9632:10026 [7] NCCL INFO Channel 05/0 : 7[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762816:763218 [1] NCCL INFO Channel 11/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762818:763220 [3] NCCL INFO Channel 12/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762817:763223 [2] NCCL INFO Channel 08/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-167-9:1039541:1039937 [4] NCCL INFO Channel 05/0 : 28[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-167-9:1039540:1039935 [3] NCCL INFO Channel 05/0 : 27[3] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762819:763221 [4] NCCL INFO Channel 09/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-167-9:1039538:1039934 [1] NCCL INFO Channel 03/0 : 25[1] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762820:763219 [5] NCCL INFO Channel 14/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762821:763224 [6] NCCL INFO Channel 07/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9630:10024 [5] NCCL INFO Channel 11/0 : 5[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-167-9:1039544:1039931 [7] NCCL INFO Channel 11/0 : 31[7] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762815:763225 [0] NCCL INFO Channel 09/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9629:10021 [4] NCCL INFO Channel 01/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762822:763222 [7] NCCL INFO Channel 14/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9628:10023 [3] NCCL INFO Channel 07/0 : 3[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9626:10025 [1] NCCL INFO Channel 03/0 : 1[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9631:10022 [6] NCCL INFO Channel 03/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9625:10020 [0] NCCL INFO Channel 01/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-167-9:1039543:1039933 [6] NCCL INFO Channel 09/0 : 30[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-167-9:1039537:1039932 [0] NCCL INFO Channel 15/0 : 56[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-167-9:1039539:1039938 [2] NCCL INFO Channel 01/0 : 26[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-167-9:1039542:1039936 [5] NCCL INFO Channel 03/0 : 29[5] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-167-9:1039541:1039937 [4] NCCL INFO Channel 07/0 : 28[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-167-9:1039540:1039935 [3] NCCL INFO Channel 07/0 : 27[3] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762816:763218 [1] NCCL INFO Channel 12/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762818:763220 [3] NCCL INFO Channel 13/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-167-9:1039538:1039934 [1] NCCL INFO Channel 05/0 : 25[1] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9632:10026 [7] NCCL INFO Channel 07/0 : 7[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9630:10024 [5] NCCL INFO Channel 15/0 : 5[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9627:10027 [2] NCCL INFO Channel 01/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9629:10021 [4] NCCL INFO Channel 03/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9626:10025 [1] NCCL INFO Channel 05/0 : 1[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9628:10023 [3] NCCL INFO Channel 09/0 : 3[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9631:10022 [6] NCCL INFO Channel 05/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762817:763223 [2] NCCL INFO Channel 09/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9625:10020 [0] NCCL INFO Channel 03/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-167-9:1039543:1039933 [6] NCCL INFO Channel 11/0 : 30[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762820:763219 [5] NCCL INFO Channel 15/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-167-9:1039539:1039938 [2] NCCL INFO Channel 03/0 : 26[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762819:763221 [4] NCCL INFO Channel 10/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-167-9:1039544:1039931 [7] NCCL INFO Channel 13/0 : 31[7] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762815:763225 [0] NCCL INFO Channel 10/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762821:763224 [6] NCCL INFO Channel 08/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-167-9:1039542:1039936 [5] NCCL INFO Channel 05/0 : 29[5] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762822:763222 [7] NCCL INFO Channel 00/0 : 23[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-167-9:1039541:1039937 [4] NCCL INFO Channel 09/0 : 28[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-167-9:1039540:1039935 [3] NCCL INFO Channel 09/0 : 27[3] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762816:763218 [1] NCCL INFO Channel 13/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762818:763220 [3] NCCL INFO Channel 14/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9632:10026 [7] NCCL INFO Channel 09/0 : 7[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-167-9:1039538:1039934 [1] NCCL INFO Channel 07/0 : 25[1] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9630:10024 [5] NCCL INFO Channel 00/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9627:10027 [2] NCCL INFO Channel 03/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9629:10021 [4] NCCL INFO Channel 05/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9626:10025 [1] NCCL INFO Channel 07/0 : 1[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762817:763223 [2] NCCL INFO Channel 11/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9628:10023 [3] NCCL INFO Channel 13/0 : 3[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762820:763219 [5] NCCL INFO Channel 00/0 : 21[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9631:10022 [6] NCCL INFO Channel 07/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-167-9:1039543:1039933 [6] NCCL INFO Channel 13/0 : 30[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762819:763221 [4] NCCL INFO Channel 11/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-167-9:1039539:1039938 [2] NCCL INFO Channel 05/0 : 26[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762815:763225 [0] NCCL INFO Channel 11/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-167-9:1039544:1039931 [7] NCCL INFO Channel 01/0 : 63[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-167-9:1039542:1039936 [5] NCCL INFO Channel 07/0 : 29[5] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-167-9:1039540:1039935 [3] NCCL INFO Channel 13/0 : 27[3] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-167-9:1039541:1039937 [4] NCCL INFO Channel 11/0 : 28[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9625:10020 [0] NCCL INFO Channel 05/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9632:10026 [7] NCCL INFO Channel 11/0 : 7[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9630:10024 [5] NCCL INFO Channel 01/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762821:763224 [6] NCCL INFO Channel 09/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9627:10027 [2] NCCL INFO Channel 05/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762816:763218 [1] NCCL INFO Channel 14/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762818:763220 [3] NCCL INFO Channel 15/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762822:763222 [7] NCCL INFO Channel 02/0 : 23[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9626:10025 [1] NCCL INFO Channel 11/0 : 1[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9629:10021 [4] NCCL INFO Channel 07/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762817:763223 [2] NCCL INFO Channel 12/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9628:10023 [3] NCCL INFO Channel 15/0 : 3[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-167-9:1039538:1039934 [1] NCCL INFO Channel 11/0 : 25[1] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-167-9:1039539:1039938 [2] NCCL INFO Channel 07/0 : 26[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-167-9:1039543:1039933 [6] NCCL INFO Channel 15/0 : 30[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762820:763219 [5] NCCL INFO Channel 02/0 : 21[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9631:10022 [6] NCCL INFO Channel 09/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-167-9:1039544:1039931 [7] NCCL INFO Channel 03/0 : 63[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762819:763221 [4] NCCL INFO Channel 13/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-167-9:1039542:1039936 [5] NCCL INFO Channel 09/0 : 29[5] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762815:763225 [0] NCCL INFO Channel 12/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-167-9:1039540:1039935 [3] NCCL INFO Channel 15/0 : 27[3] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-167-9:1039541:1039937 [4] NCCL INFO Channel 13/0 : 28[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9632:10026 [7] NCCL INFO Channel 13/0 : 7[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9625:10020 [0] NCCL INFO Channel 07/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9630:10024 [5] NCCL INFO Channel 02/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9627:10027 [2] NCCL INFO Channel 07/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762816:763218 [1] NCCL INFO Channel 15/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762821:763224 [6] NCCL INFO Channel 10/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9626:10025 [1] NCCL INFO Channel 13/0 : 1[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762818:763220 [3] NCCL INFO Channel 00/0 : 19[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9628:10023 [3] NCCL INFO Channel 00/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-167-9:1039538:1039934 [1] NCCL INFO Channel 13/0 : 25[1] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762822:763222 [7] NCCL INFO Channel 04/0 : 23[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762817:763223 [2] NCCL INFO Channel 13/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9629:10021 [4] NCCL INFO Channel 09/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-167-9:1039539:1039938 [2] NCCL INFO Channel 09/0 : 26[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762820:763219 [5] NCCL INFO Channel 04/0 : 21[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9631:10022 [6] NCCL INFO Channel 11/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-167-9:1039543:1039933 [6] NCCL INFO Channel 01/0 : 62[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762819:763221 [4] NCCL INFO Channel 14/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762815:763225 [0] NCCL INFO Channel 13/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9630:10024 [5] NCCL INFO Channel 03/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-167-9:1039544:1039931 [7] NCCL INFO Channel 05/0 : 63[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-167-9:1039542:1039936 [5] NCCL INFO Channel 11/0 : 29[5] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-167-9:1039540:1039935 [3] NCCL INFO Channel 01/0 : 59[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-167-9:1039541:1039937 [4] NCCL INFO Channel 15/0 : 28[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-167-9:1039538:1039934 [1] NCCL INFO Channel 15/0 : 25[1] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762816:763218 [1] NCCL INFO Channel 00/0 : 17[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9625:10020 [0] NCCL INFO Channel 09/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762821:763224 [6] NCCL INFO Channel 11/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9632:10026 [7] NCCL INFO Channel 00/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762818:763220 [3] NCCL INFO Channel 02/0 : 19[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9627:10027 [2] NCCL INFO Channel 09/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9628:10023 [3] NCCL INFO Channel 01/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9626:10025 [1] NCCL INFO Channel 15/0 : 1[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-167-9:1039543:1039933 [6] NCCL INFO Channel 03/0 : 62[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-167-9:1039539:1039938 [2] NCCL INFO Channel 11/0 : 26[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9629:10021 [4] NCCL INFO Channel 11/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9631:10022 [6] NCCL INFO Channel 13/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9630:10024 [5] NCCL INFO Channel 04/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-167-9:1039544:1039931 [7] NCCL INFO Channel 07/0 : 63[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-167-9:1039540:1039935 [3] NCCL INFO Channel 03/0 : 59[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762822:763222 [7] NCCL INFO Channel 06/0 : 23[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-167-9:1039541:1039937 [4] NCCL INFO Channel 01/0 : 60[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-167-9:1039538:1039934 [1] NCCL INFO Channel 01/0 : 57[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-167-9:1039542:1039936 [5] NCCL INFO Channel 15/0 : 29[5] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762817:763223 [2] NCCL INFO Channel 14/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762815:763225 [0] NCCL INFO Channel 14/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762820:763219 [5] NCCL INFO Channel 06/0 : 21[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762819:763221 [4] NCCL INFO Channel 15/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9625:10020 [0] NCCL INFO Channel 11/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762816:763218 [1] NCCL INFO Channel 02/0 : 17[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9632:10026 [7] NCCL INFO Channel 01/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762818:763220 [3] NCCL INFO Channel 04/0 : 19[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762821:763224 [6] NCCL INFO Channel 12/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9627:10027 [2] NCCL INFO Channel 11/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9628:10023 [3] NCCL INFO Channel 02/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-167-9:1039539:1039938 [2] NCCL INFO Channel 13/0 : 26[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-167-9:1039543:1039933 [6] NCCL INFO Channel 05/0 : 62[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9626:10025 [1] NCCL INFO Channel 00/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-167-9:1039544:1039931 [7] NCCL INFO Channel 09/0 : 63[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9631:10022 [6] NCCL INFO Channel 15/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9630:10024 [5] NCCL INFO Channel 06/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9629:10021 [4] NCCL INFO Channel 13/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-167-9:1039540:1039935 [3] NCCL INFO Channel 05/0 : 59[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9625:10020 [0] NCCL INFO Channel 13/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-167-9:1039538:1039934 [1] NCCL INFO Channel 03/0 : 57[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-167-9:1039541:1039937 [4] NCCL INFO Channel 03/0 : 60[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-167-9:1039542:1039936 [5] NCCL INFO Channel 01/0 : 61[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762817:763223 [2] NCCL INFO Channel 15/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762815:763225 [0] NCCL INFO Channel 15/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762822:763222 [7] NCCL INFO Channel 08/0 : 23[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762820:763219 [5] NCCL INFO Channel 08/0 : 21[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762816:763218 [1] NCCL INFO Channel 04/0 : 17[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9632:10026 [7] NCCL INFO Channel 02/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9627:10027 [2] NCCL INFO Channel 13/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-167-9:1039539:1039938 [2] NCCL INFO Channel 15/0 : 26[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-167-9:1039543:1039933 [6] NCCL INFO Channel 07/0 : 62[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9628:10023 [3] NCCL INFO Channel 04/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-167-9:1039544:1039931 [7] NCCL INFO Channel 11/0 : 63[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9626:10025 [1] NCCL INFO Channel 02/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-167-9:1039540:1039935 [3] NCCL INFO Channel 07/0 : 59[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9630:10024 [5] NCCL INFO Channel 07/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-167-9:1039541:1039937 [4] NCCL INFO Channel 05/0 : 60[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-167-9:1039538:1039934 [1] NCCL INFO Channel 05/0 : 57[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762819:763221 [4] NCCL INFO Channel 00/0 : 20[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9629:10021 [4] NCCL INFO Channel 15/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9631:10022 [6] NCCL INFO Channel 00/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762818:763220 [3] NCCL INFO Channel 06/0 : 19[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762821:763224 [6] NCCL INFO Channel 13/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9625:10020 [0] NCCL INFO Channel 15/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762817:763223 [2] NCCL INFO Channel 00/0 : 18[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762822:763222 [7] NCCL INFO Channel 10/0 : 23[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-167-9:1039542:1039936 [5] NCCL INFO Channel 03/0 : 61[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-167-9:1039539:1039938 [2] NCCL INFO Channel 01/0 : 58[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-167-9:1039543:1039933 [6] NCCL INFO Channel 09/0 : 62[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9632:10026 [7] NCCL INFO Channel 03/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9627:10027 [2] NCCL INFO Channel 15/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9628:10023 [3] NCCL INFO Channel 05/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-167-9:1039540:1039935 [3] NCCL INFO Channel 09/0 : 59[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-167-9:1039544:1039931 [7] NCCL INFO Channel 13/0 : 63[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9626:10025 [1] NCCL INFO Channel 03/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-167-9:1039541:1039937 [4] NCCL INFO Channel 07/0 : 60[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-167-9:1039538:1039934 [1] NCCL INFO Channel 07/0 : 57[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762815:763225 [0] NCCL INFO Channel 02/0 : 16[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9630:10024 [5] NCCL INFO Channel 08/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762816:763218 [1] NCCL INFO Channel 06/0 : 17[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762820:763219 [5] NCCL INFO Channel 10/0 : 21[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762818:763220 [3] NCCL INFO Channel 08/0 : 19[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762821:763224 [6] NCCL INFO Channel 15/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762817:763223 [2] NCCL INFO Channel 04/0 : 18[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762819:763221 [4] NCCL INFO Channel 02/0 : 20[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-167-9:1039542:1039936 [5] NCCL INFO Channel 05/0 : 61[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-167-9:1039543:1039933 [6] NCCL INFO Channel 11/0 : 62[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-167-9:1039539:1039938 [2] NCCL INFO Channel 03/0 : 58[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-167-9:1039540:1039935 [3] NCCL INFO Channel 13/0 : 59[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9625:10020 [0] NCCL INFO Channel 01/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9629:10021 [4] NCCL INFO Channel 00/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9631:10022 [6] NCCL INFO Channel 01/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9628:10023 [3] NCCL INFO Channel 06/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9627:10027 [2] NCCL INFO Channel 00/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762822:763222 [7] NCCL INFO Channel 12/0 : 23[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762815:763225 [0] NCCL INFO Channel 04/0 : 16[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9626:10025 [1] NCCL INFO Channel 04/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9632:10026 [7] NCCL INFO Channel 04/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-167-9:1039538:1039934 [1] NCCL INFO Channel 11/0 : 57[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-167-9:1039541:1039937 [4] NCCL INFO Channel 09/0 : 60[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-167-9:1039542:1039936 [5] NCCL INFO Channel 07/0 : 61[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9630:10024 [5] NCCL INFO Channel 09/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762816:763218 [1] NCCL INFO Channel 08/0 : 17[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762820:763219 [5] NCCL INFO Channel 12/0 : 21[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762818:763220 [3] NCCL INFO Channel 10/0 : 19[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-167-9:1039543:1039933 [6] NCCL INFO Channel 13/0 : 62[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-167-9:1039539:1039938 [2] NCCL INFO Channel 05/0 : 58[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-167-9:1039540:1039935 [3] NCCL INFO Channel 15/0 : 59[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-167-9:1039541:1039937 [4] NCCL INFO Channel 11/0 : 60[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-167-9:1039538:1039934 [1] NCCL INFO Channel 13/0 : 57[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-167-9:1039542:1039936 [5] NCCL INFO Channel 09/0 : 61[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9625:10020 [0] NCCL INFO Channel 02/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9629:10021 [4] NCCL INFO Channel 01/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9631:10022 [6] NCCL INFO Channel 02/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9628:10023 [3] NCCL INFO Channel 07/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762817:763223 [2] NCCL INFO Channel 06/0 : 18[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9626:10025 [1] NCCL INFO Channel 05/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762819:763221 [4] NCCL INFO Channel 06/0 : 20[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762821:763224 [6] NCCL INFO Channel 00/0 : 22[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9627:10027 [2] NCCL INFO Channel 01/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-167-9:1039539:1039938 [2] NCCL INFO Channel 07/0 : 58[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-167-9:1039543:1039933 [6] NCCL INFO Channel 15/0 : 62[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762815:763225 [0] NCCL INFO Channel 06/0 : 16[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762822:763222 [7] NCCL INFO Channel 14/0 : 23[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9632:10026 [7] NCCL INFO Channel 05/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9630:10024 [5] NCCL INFO Channel 10/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762816:763218 [1] NCCL INFO Channel 10/0 : 17[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9625:10020 [0] NCCL INFO Channel 03/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9629:10021 [4] NCCL INFO Channel 02/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9631:10022 [6] NCCL INFO Channel 03/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9628:10023 [3] NCCL INFO Channel 08/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9626:10025 [1] NCCL INFO Channel 06/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-167-9:1039538:1039934 [1] NCCL INFO Channel 15/0 : 57[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762818:763220 [3] NCCL INFO Channel 12/0 : 19[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-167-9:1039541:1039937 [4] NCCL INFO Channel 13/0 : 60[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-167-9:1039542:1039936 [5] NCCL INFO Channel 11/0 : 61[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762820:763219 [5] NCCL INFO Channel 14/0 : 21[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-167-9:1039539:1039938 [2] NCCL INFO Channel 09/0 : 58[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762815:763225 [0] NCCL INFO Channel 08/0 : 16[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762817:763223 [2] NCCL INFO Channel 08/0 : 18[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762819:763221 [4] NCCL INFO Channel 08/0 : 20[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762821:763224 [6] NCCL INFO Channel 02/0 : 22[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762816:763218 [1] NCCL INFO Channel 12/0 : 17[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9627:10027 [2] NCCL INFO Channel 03/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762818:763220 [3] NCCL INFO Channel 14/0 : 19[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9625:10020 [0] NCCL INFO Channel 04/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9630:10024 [5] NCCL INFO Channel 11/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9632:10026 [7] NCCL INFO Channel 06/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9631:10022 [6] NCCL INFO Channel 04/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9629:10021 [4] NCCL INFO Channel 03/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-167-9:1039542:1039936 [5] NCCL INFO Channel 15/0 : 61[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-167-9:1039541:1039937 [4] NCCL INFO Channel 15/0 : 60[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-167-9:1039539:1039938 [2] NCCL INFO Channel 11/0 : 58[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9628:10023 [3] NCCL INFO Channel 09/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9626:10025 [1] NCCL INFO Channel 07/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9627:10027 [2] NCCL INFO Channel 04/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9625:10020 [0] NCCL INFO Channel 05/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762815:763225 [0] NCCL INFO Channel 10/0 : 16[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9630:10024 [5] NCCL INFO Channel 12/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762817:763223 [2] NCCL INFO Channel 10/0 : 18[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762819:763221 [4] NCCL INFO Channel 10/0 : 20[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762816:763218 [1] NCCL INFO Channel 14/0 : 17[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762821:763224 [6] NCCL INFO Channel 04/0 : 22[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-167-9:1039539:1039938 [2] NCCL INFO Channel 13/0 : 58[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762815:763225 [0] NCCL INFO Channel 12/0 : 16[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9632:10026 [7] NCCL INFO Channel 08/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9631:10022 [6] NCCL INFO Channel 05/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9628:10023 [3] NCCL INFO Channel 10/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-167-9:1039539:1039938 [2] NCCL INFO Channel 15/0 : 58[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762817:763223 [2] NCCL INFO Channel 12/0 : 18[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762819:763221 [4] NCCL INFO Channel 12/0 : 20[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9627:10027 [2] NCCL INFO Channel 05/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9626:10025 [1] NCCL INFO Channel 08/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762815:763225 [0] NCCL INFO Channel 14/0 : 16[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762821:763224 [6] NCCL INFO Channel 08/0 : 22[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9629:10021 [4] NCCL INFO Channel 05/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9625:10020 [0] NCCL INFO Channel 06/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9630:10024 [5] NCCL INFO Channel 14/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9632:10026 [7] NCCL INFO Channel 09/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9631:10022 [6] NCCL INFO Channel 07/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9628:10023 [3] NCCL INFO Channel 12/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9626:10025 [1] NCCL INFO Channel 10/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762817:763223 [2] NCCL INFO Channel 14/0 : 18[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762819:763221 [4] NCCL INFO Channel 14/0 : 20[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9627:10027 [2] NCCL INFO Channel 06/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9625:10020 [0] NCCL INFO Channel 07/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9629:10021 [4] NCCL INFO Channel 06/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9630:10024 [5] NCCL INFO Channel 15/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9632:10026 [7] NCCL INFO Channel 10/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762821:763224 [6] NCCL INFO Channel 10/0 : 22[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9628:10023 [3] NCCL INFO Channel 13/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9631:10022 [6] NCCL INFO Channel 08/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9626:10025 [1] NCCL INFO Channel 11/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9627:10027 [2] NCCL INFO Channel 07/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9625:10020 [0] NCCL INFO Channel 09/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762821:763224 [6] NCCL INFO Channel 12/0 : 22[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9629:10021 [4] NCCL INFO Channel 07/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762821:763224 [6] NCCL INFO Channel 14/0 : 22[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9632:10026 [7] NCCL INFO Channel 11/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9626:10025 [1] NCCL INFO Channel 12/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9631:10022 [6] NCCL INFO Channel 09/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9627:10027 [2] NCCL INFO Channel 08/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9628:10023 [3] NCCL INFO Channel 14/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9625:10020 [0] NCCL INFO Channel 10/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9629:10021 [4] NCCL INFO Channel 08/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9626:10025 [1] NCCL INFO Channel 13/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9632:10026 [7] NCCL INFO Channel 12/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9631:10022 [6] NCCL INFO Channel 10/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9627:10027 [2] NCCL INFO Channel 09/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9628:10023 [3] NCCL INFO Channel 15/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9625:10020 [0] NCCL INFO Channel 11/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9629:10021 [4] NCCL INFO Channel 09/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9626:10025 [1] NCCL INFO Channel 14/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9627:10027 [2] NCCL INFO Channel 11/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9632:10026 [7] NCCL INFO Channel 13/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9631:10022 [6] NCCL INFO Channel 11/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9629:10021 [4] NCCL INFO Channel 10/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9626:10025 [1] NCCL INFO Channel 15/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9625:10020 [0] NCCL INFO Channel 12/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9630:10024 [5] NCCL INFO Channel 01/0 : 13[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9627:10027 [2] NCCL INFO Channel 12/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9631:10022 [6] NCCL INFO Channel 12/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9632:10026 [7] NCCL INFO Channel 14/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9629:10021 [4] NCCL INFO Channel 11/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9625:10020 [0] NCCL INFO Channel 13/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9630:10024 [5] NCCL INFO Channel 03/0 : 13[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9627:10027 [2] NCCL INFO Channel 13/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9631:10022 [6] NCCL INFO Channel 13/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9629:10021 [4] NCCL INFO Channel 13/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9625:10020 [0] NCCL INFO Channel 14/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9627:10027 [2] NCCL INFO Channel 14/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9630:10024 [5] NCCL INFO Channel 05/0 : 13[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9631:10022 [6] NCCL INFO Channel 15/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9629:10021 [4] NCCL INFO Channel 14/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-46:702881:703280 [1] NCCL INFO Channel 00/0 : 33[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9625:10020 [0] NCCL INFO Channel 15/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9627:10027 [2] NCCL INFO Channel 15/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9630:10024 [5] NCCL INFO Channel 07/0 : 13[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9628:10023 [3] NCCL INFO Channel 01/0 : 11[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9629:10021 [4] NCCL INFO Channel 15/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9626:10025 [1] NCCL INFO Channel 01/0 : 9[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-46:702881:703280 [1] NCCL INFO Channel 02/0 : 33[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9630:10024 [5] NCCL INFO Channel 09/0 : 13[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9628:10023 [3] NCCL INFO Channel 03/0 : 11[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9626:10025 [1] NCCL INFO Channel 03/0 : 9[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-46:702881:703280 [1] NCCL INFO Channel 04/0 : 33[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9632:10026 [7] NCCL INFO Channel 01/0 : 15[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9630:10024 [5] NCCL INFO Channel 11/0 : 13[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9628:10023 [3] NCCL INFO Channel 05/0 : 11[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-46:702881:703280 [1] NCCL INFO Channel 06/0 : 33[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9626:10025 [1] NCCL INFO Channel 05/0 : 9[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9632:10026 [7] NCCL INFO Channel 03/0 : 15[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9630:10024 [5] NCCL INFO Channel 15/0 : 13[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9628:10023 [3] NCCL INFO Channel 07/0 : 11[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9626:10025 [1] NCCL INFO Channel 07/0 : 9[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-46:702881:703280 [1] NCCL INFO Channel 08/0 : 33[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9632:10026 [7] NCCL INFO Channel 05/0 : 15[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9631:10022 [6] NCCL INFO Channel 01/0 : 14[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9628:10023 [3] NCCL INFO Channel 09/0 : 11[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9626:10025 [1] NCCL INFO Channel 11/0 : 9[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-46:702881:703280 [1] NCCL INFO Channel 10/0 : 33[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9625:10020 [0] NCCL INFO Channel 01/0 : 8[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9632:10026 [7] NCCL INFO Channel 07/0 : 15[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9631:10022 [6] NCCL INFO Channel 03/0 : 14[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-46:702881:703280 [1] NCCL INFO Channel 12/0 : 33[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9628:10023 [3] NCCL INFO Channel 13/0 : 11[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9629:10021 [4] NCCL INFO Channel 01/0 : 12[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9626:10025 [1] NCCL INFO Channel 13/0 : 9[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9625:10020 [0] NCCL INFO Channel 03/0 : 8[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-46:702881:703280 [1] NCCL INFO Channel 14/0 : 33[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9632:10026 [7] NCCL INFO Channel 09/0 : 15[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9631:10022 [6] NCCL INFO Channel 05/0 : 14[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9628:10023 [3] NCCL INFO Channel 15/0 : 11[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9629:10021 [4] NCCL INFO Channel 03/0 : 12[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9626:10025 [1] NCCL INFO Channel 15/0 : 9[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-46:702881:703280 [1] NCCL INFO Channel 00/0 : 1[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9625:10020 [0] NCCL INFO Channel 05/0 : 8[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9631:10022 [6] NCCL INFO Channel 07/0 : 14[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9632:10026 [7] NCCL INFO Channel 11/0 : 15[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-46:702881:703280 [1] NCCL INFO Channel 02/0 : 1[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9629:10021 [4] NCCL INFO Channel 05/0 : 12[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9625:10020 [0] NCCL INFO Channel 07/0 : 8[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9631:10022 [6] NCCL INFO Channel 09/0 : 14[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-46:702881:703280 [1] NCCL INFO Channel 04/0 : 1[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9632:10026 [7] NCCL INFO Channel 13/0 : 15[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9629:10021 [4] NCCL INFO Channel 07/0 : 12[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9625:10020 [0] NCCL INFO Channel 09/0 : 8[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9631:10022 [6] NCCL INFO Channel 11/0 : 14[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9629:10021 [4] NCCL INFO Channel 09/0 : 12[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-46:702881:703280 [1] NCCL INFO Channel 06/0 : 1[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9625:10020 [0] NCCL INFO Channel 11/0 : 8[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9631:10022 [6] NCCL INFO Channel 13/0 : 14[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9629:10021 [4] NCCL INFO Channel 11/0 : 12[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-46:702881:703280 [1] NCCL INFO Channel 08/0 : 1[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9625:10020 [0] NCCL INFO Channel 13/0 : 8[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9631:10022 [6] NCCL INFO Channel 15/0 : 14[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9629:10021 [4] NCCL INFO Channel 13/0 : 12[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-46:702881:703280 [1] NCCL INFO Channel 10/0 : 1[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9625:10020 [0] NCCL INFO Channel 15/0 : 8[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9629:10021 [4] NCCL INFO Channel 15/0 : 12[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-46:702881:703280 [1] NCCL INFO Channel 12/0 : 1[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-46:702881:703280 [1] NCCL INFO Channel 14/0 : 1[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9627:10027 [2] NCCL INFO Channel 01/0 : 10[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9627:10027 [2] NCCL INFO Channel 03/0 : 10[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9627:10027 [2] NCCL INFO Channel 05/0 : 10[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9627:10027 [2] NCCL INFO Channel 07/0 : 10[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9627:10027 [2] NCCL INFO Channel 09/0 : 10[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9627:10027 [2] NCCL INFO Channel 11/0 : 10[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9627:10027 [2] NCCL INFO Channel 13/0 : 10[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9627:10027 [2] NCCL INFO Channel 15/0 : 10[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-46:702886:703281 [6] NCCL INFO Channel 00/0 : 38[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-46:702883:703283 [3] NCCL INFO Channel 00/0 : 35[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-46:702886:703281 [6] NCCL INFO Channel 02/0 : 38[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-46:702883:703283 [3] NCCL INFO Channel 02/0 : 35[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-46:702886:703281 [6] NCCL INFO Channel 04/0 : 38[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897243:897637 [3] NCCL INFO Channel 01/0 : 35[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-46:702883:703283 [3] NCCL INFO Channel 04/0 : 35[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-46:702886:703281 [6] NCCL INFO Channel 08/0 : 38[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897243:897637 [3] NCCL INFO Channel 03/0 : 35[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-46:702883:703283 [3] NCCL INFO Channel 06/0 : 35[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897247:897642 [7] NCCL INFO Channel 01/0 : 39[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-46:702886:703281 [6] NCCL INFO Channel 10/0 : 38[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-46:702882:703284 [2] NCCL INFO Channel 00/0 : 34[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897243:897637 [3] NCCL INFO Channel 05/0 : 35[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-46:702883:703283 [3] NCCL INFO Channel 08/0 : 35[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897240:897640 [0] NCCL INFO Channel 01/0 : 32[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897245:897638 [5] NCCL INFO Channel 01/0 : 37[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897247:897642 [7] NCCL INFO Channel 03/0 : 39[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897243:897637 [3] NCCL INFO Channel 07/0 : 35[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-46:702886:703281 [6] NCCL INFO Channel 12/0 : 38[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-46:702882:703284 [2] NCCL INFO Channel 04/0 : 34[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-46:702883:703283 [3] NCCL INFO Channel 10/0 : 35[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897241:897635 [1] NCCL INFO Channel 01/0 : 33[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-46:702886:703281 [6] NCCL INFO Channel 14/0 : 38[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-46:702883:703283 [3] NCCL INFO Channel 12/0 : 35[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-46:702882:703284 [2] NCCL INFO Channel 06/0 : 34[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897245:897638 [5] NCCL INFO Channel 03/0 : 37[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897247:897642 [7] NCCL INFO Channel 05/0 : 39[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897243:897637 [3] NCCL INFO Channel 09/0 : 35[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897240:897640 [0] NCCL INFO Channel 03/0 : 32[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897244:897636 [4] NCCL INFO Channel 01/0 : 36[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-46:702886:703281 [6] NCCL INFO Channel 00/0 : 6[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897241:897635 [1] NCCL INFO Channel 03/0 : 33[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-46:702883:703283 [3] NCCL INFO Channel 14/0 : 35[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-46:702882:703284 [2] NCCL INFO Channel 08/0 : 34[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897247:897642 [7] NCCL INFO Channel 07/0 : 39[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-46:702886:703281 [6] NCCL INFO Channel 02/0 : 6[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897245:897638 [5] NCCL INFO Channel 05/0 : 37[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897243:897637 [3] NCCL INFO Channel 13/0 : 35[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897244:897636 [4] NCCL INFO Channel 03/0 : 36[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897240:897640 [0] NCCL INFO Channel 05/0 : 32[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897241:897635 [1] NCCL INFO Channel 05/0 : 33[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897246:897641 [6] NCCL INFO Channel 01/0 : 38[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-46:702883:703283 [3] NCCL INFO Channel 00/0 : 3[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897247:897642 [7] NCCL INFO Channel 09/0 : 39[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-46:702882:703284 [2] NCCL INFO Channel 10/0 : 34[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-46:702886:703281 [6] NCCL INFO Channel 04/0 : 6[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-46:702883:703283 [3] NCCL INFO Channel 02/0 : 3[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897245:897638 [5] NCCL INFO Channel 07/0 : 37[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-46:702882:703284 [2] NCCL INFO Channel 12/0 : 34[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-46:702886:703281 [6] NCCL INFO Channel 08/0 : 6[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-46:702883:703283 [3] NCCL INFO Channel 04/0 : 3[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897243:897637 [3] NCCL INFO Channel 15/0 : 35[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897241:897635 [1] NCCL INFO Channel 07/0 : 33[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-46:702882:703284 [2] NCCL INFO Channel 14/0 : 34[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897247:897642 [7] NCCL INFO Channel 11/0 : 39[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897244:897636 [4] NCCL INFO Channel 05/0 : 36[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897246:897641 [6] NCCL INFO Channel 03/0 : 38[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897240:897640 [0] NCCL INFO Channel 07/0 : 32[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-46:702886:703281 [6] NCCL INFO Channel 10/0 : 6[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-46:702883:703283 [3] NCCL INFO Channel 06/0 : 3[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897242:897639 [2] NCCL INFO Channel 01/0 : 34[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897245:897638 [5] NCCL INFO Channel 09/0 : 37[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897241:897635 [1] NCCL INFO Channel 11/0 : 33[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-46:702882:703284 [2] NCCL INFO Channel 00/0 : 2[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-46:702886:703281 [6] NCCL INFO Channel 12/0 : 6[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-46:702883:703283 [3] NCCL INFO Channel 08/0 : 3[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897243:897637 [3] NCCL INFO Channel 00/0 : 43[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897247:897642 [7] NCCL INFO Channel 13/0 : 39[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-46:702882:703284 [2] NCCL INFO Channel 04/0 : 2[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897244:897636 [4] NCCL INFO Channel 07/0 : 36[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897246:897641 [6] NCCL INFO Channel 05/0 : 38[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897240:897640 [0] NCCL INFO Channel 09/0 : 32[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-46:702886:703281 [6] NCCL INFO Channel 14/0 : 6[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-46:702883:703283 [3] NCCL INFO Channel 10/0 : 3[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-46:702882:703284 [2] NCCL INFO Channel 06/0 : 2[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897242:897639 [2] NCCL INFO Channel 03/0 : 34[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897245:897638 [5] NCCL INFO Channel 11/0 : 37[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897241:897635 [1] NCCL INFO Channel 13/0 : 33[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-46:702883:703283 [3] NCCL INFO Channel 12/0 : 3[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897243:897637 [3] NCCL INFO Channel 01/0 : 43[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897247:897642 [7] NCCL INFO Channel 00/0 : 47[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897244:897636 [4] NCCL INFO Channel 09/0 : 36[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897246:897641 [6] NCCL INFO Channel 07/0 : 38[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-46:702882:703284 [2] NCCL INFO Channel 08/0 : 2[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-46:702884:703279 [4] NCCL INFO Channel 00/0 : 36[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-46:702883:703283 [3] NCCL INFO Channel 14/0 : 3[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-46:702882:703284 [2] NCCL INFO Channel 10/0 : 2[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897245:897638 [5] NCCL INFO Channel 15/0 : 37[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897241:897635 [1] NCCL INFO Channel 15/0 : 33[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897240:897640 [0] NCCL INFO Channel 11/0 : 32[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897242:897639 [2] NCCL INFO Channel 05/0 : 34[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897247:897642 [7] NCCL INFO Channel 01/0 : 47[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897243:897637 [3] NCCL INFO Channel 02/0 : 43[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897244:897636 [4] NCCL INFO Channel 11/0 : 36[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897246:897641 [6] NCCL INFO Channel 09/0 : 38[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-46:702887:703282 [7] NCCL INFO Channel 00/0 : 39[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-46:702884:703279 [4] NCCL INFO Channel 02/0 : 36[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-46:702882:703284 [2] NCCL INFO Channel 12/0 : 2[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897241:897635 [1] NCCL INFO Channel 00/0 : 41[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897245:897638 [5] NCCL INFO Channel 00/0 : 45[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897240:897640 [0] NCCL INFO Channel 13/0 : 32[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897242:897639 [2] NCCL INFO Channel 07/0 : 34[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897247:897642 [7] NCCL INFO Channel 02/0 : 47[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897243:897637 [3] NCCL INFO Channel 04/0 : 43[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897244:897636 [4] NCCL INFO Channel 13/0 : 36[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897246:897641 [6] NCCL INFO Channel 11/0 : 38[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-46:702882:703284 [2] NCCL INFO Channel 14/0 : 2[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-46:702887:703282 [7] NCCL INFO Channel 02/0 : 39[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-46:702884:703279 [4] NCCL INFO Channel 06/0 : 36[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897241:897635 [1] NCCL INFO Channel 02/0 : 41[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897245:897638 [5] NCCL INFO Channel 01/0 : 45[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897240:897640 [0] NCCL INFO Channel 15/0 : 32[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-46:702887:703282 [7] NCCL INFO Channel 04/0 : 39[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-46:702884:703279 [4] NCCL INFO Channel 08/0 : 36[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897247:897642 [7] NCCL INFO Channel 03/0 : 47[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897243:897637 [3] NCCL INFO Channel 05/0 : 43[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897246:897641 [6] NCCL INFO Channel 13/0 : 38[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897242:897639 [2] NCCL INFO Channel 09/0 : 34[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897244:897636 [4] NCCL INFO Channel 15/0 : 36[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897241:897635 [1] NCCL INFO Channel 03/0 : 41[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897245:897638 [5] NCCL INFO Channel 02/0 : 45[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897240:897640 [0] NCCL INFO Channel 01/0 : 40[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-46:702887:703282 [7] NCCL INFO Channel 06/0 : 39[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-46:702884:703279 [4] NCCL INFO Channel 10/0 : 36[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897247:897642 [7] NCCL INFO Channel 04/0 : 47[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897243:897637 [3] NCCL INFO Channel 06/0 : 43[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897246:897641 [6] NCCL INFO Channel 15/0 : 38[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-46:702887:703282 [7] NCCL INFO Channel 08/0 : 39[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-46:702884:703279 [4] NCCL INFO Channel 12/0 : 36[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897242:897639 [2] NCCL INFO Channel 11/0 : 34[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897244:897636 [4] NCCL INFO Channel 00/0 : 44[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897241:897635 [1] NCCL INFO Channel 04/0 : 41[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897245:897638 [5] NCCL INFO Channel 03/0 : 45[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897240:897640 [0] NCCL INFO Channel 02/0 : 40[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897247:897642 [7] NCCL INFO Channel 05/0 : 47[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-46:702884:703279 [4] NCCL INFO Channel 14/0 : 36[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-46:702887:703282 [7] NCCL INFO Channel 10/0 : 39[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897243:897637 [3] NCCL INFO Channel 07/0 : 43[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-46:702884:703279 [4] NCCL INFO Channel 00/0 : 4[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897246:897641 [6] NCCL INFO Channel 00/0 : 46[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-46:702887:703282 [7] NCCL INFO Channel 12/0 : 39[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897244:897636 [4] NCCL INFO Channel 01/0 : 44[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897242:897639 [2] NCCL INFO Channel 13/0 : 34[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127295:127686 [5] NCCL INFO Channel 00/0 : 21[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897241:897635 [1] NCCL INFO Channel 05/0 : 41[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897245:897638 [5] NCCL INFO Channel 04/0 : 45[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897247:897642 [7] NCCL INFO Channel 06/0 : 47[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897240:897640 [0] NCCL INFO Channel 03/0 : 40[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897243:897637 [3] NCCL INFO Channel 08/0 : 43[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-46:702884:703279 [4] NCCL INFO Channel 02/0 : 4[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-46:702887:703282 [7] NCCL INFO Channel 14/0 : 39[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897246:897641 [6] NCCL INFO Channel 01/0 : 46[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127295:127686 [5] NCCL INFO Channel 02/0 : 21[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-46:702884:703279 [4] NCCL INFO Channel 06/0 : 4[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-46:702887:703282 [7] NCCL INFO Channel 00/0 : 7[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897244:897636 [4] NCCL INFO Channel 02/0 : 44[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897242:897639 [2] NCCL INFO Channel 15/0 : 34[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897241:897635 [1] NCCL INFO Channel 06/0 : 41[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127295:127686 [5] NCCL INFO Channel 04/0 : 21[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897245:897638 [5] NCCL INFO Channel 06/0 : 45[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897247:897642 [7] NCCL INFO Channel 08/0 : 47[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897246:897641 [6] NCCL INFO Channel 02/0 : 46[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897240:897640 [0] NCCL INFO Channel 04/0 : 40[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-46:702884:703279 [4] NCCL INFO Channel 08/0 : 4[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897243:897637 [3] NCCL INFO Channel 09/0 : 43[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897244:897636 [4] NCCL INFO Channel 03/0 : 44[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897241:897635 [1] NCCL INFO Channel 07/0 : 41[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897242:897639 [2] NCCL INFO Channel 00/0 : 42[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-46:702887:703282 [7] NCCL INFO Channel 02/0 : 7[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127295:127686 [5] NCCL INFO Channel 06/0 : 21[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-46:702884:703279 [4] NCCL INFO Channel 10/0 : 4[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897247:897642 [7] NCCL INFO Channel 09/0 : 47[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897245:897638 [5] NCCL INFO Channel 07/0 : 45[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897246:897641 [6] NCCL INFO Channel 03/0 : 46[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-46:702887:703282 [7] NCCL INFO Channel 04/0 : 7[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897244:897636 [4] NCCL INFO Channel 05/0 : 44[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897240:897640 [0] NCCL INFO Channel 05/0 : 40[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897243:897637 [3] NCCL INFO Channel 10/0 : 43[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897241:897635 [1] NCCL INFO Channel 08/0 : 41[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-46:702884:703279 [4] NCCL INFO Channel 12/0 : 4[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897242:897639 [2] NCCL INFO Channel 01/0 : 42[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127295:127686 [5] NCCL INFO Channel 08/0 : 21[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-46:702887:703282 [7] NCCL INFO Channel 06/0 : 7[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897247:897642 [7] NCCL INFO Channel 10/0 : 47[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-46:702884:703279 [4] NCCL INFO Channel 14/0 : 4[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897245:897638 [5] NCCL INFO Channel 08/0 : 45[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-46:702887:703282 [7] NCCL INFO Channel 08/0 : 7[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127294:127687 [4] NCCL INFO Channel 00/0 : 20[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127295:127686 [5] NCCL INFO Channel 10/0 : 21[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897246:897641 [6] NCCL INFO Channel 04/0 : 46[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897244:897636 [4] NCCL INFO Channel 06/0 : 44[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897240:897640 [0] NCCL INFO Channel 06/0 : 40[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-46:702887:703282 [7] NCCL INFO Channel 10/0 : 7[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897241:897635 [1] NCCL INFO Channel 10/0 : 41[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897243:897637 [3] NCCL INFO Channel 12/0 : 43[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897247:897642 [7] NCCL INFO Channel 11/0 : 47[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897242:897639 [2] NCCL INFO Channel 03/0 : 42[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897245:897638 [5] NCCL INFO Channel 09/0 : 45[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127293:127689 [3] NCCL INFO Channel 00/0 : 19[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897246:897641 [6] NCCL INFO Channel 05/0 : 46[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127291:127690 [1] NCCL INFO Channel 00/0 : 17[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-46:702887:703282 [7] NCCL INFO Channel 12/0 : 7[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897244:897636 [4] NCCL INFO Channel 07/0 : 44[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127295:127686 [5] NCCL INFO Channel 12/0 : 21[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127294:127687 [4] NCCL INFO Channel 02/0 : 20[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897240:897640 [0] NCCL INFO Channel 07/0 : 40[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897241:897635 [1] NCCL INFO Channel 11/0 : 41[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897247:897642 [7] NCCL INFO Channel 12/0 : 47[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897243:897637 [3] NCCL INFO Channel 13/0 : 43[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897242:897639 [2] NCCL INFO Channel 04/0 : 42[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127293:127689 [3] NCCL INFO Channel 02/0 : 19[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127296:127688 [6] NCCL INFO Channel 00/0 : 22[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-46:702887:703282 [7] NCCL INFO Channel 14/0 : 7[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897245:897638 [5] NCCL INFO Channel 10/0 : 45[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897246:897641 [6] NCCL INFO Channel 07/0 : 46[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127291:127690 [1] NCCL INFO Channel 02/0 : 17[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897244:897636 [4] NCCL INFO Channel 08/0 : 44[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127297:127691 [7] NCCL INFO Channel 00/0 : 23[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127290:127685 [0] NCCL INFO Channel 02/0 : 16[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897240:897640 [0] NCCL INFO Channel 09/0 : 40[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897241:897635 [1] NCCL INFO Channel 12/0 : 41[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127295:127686 [5] NCCL INFO Channel 14/0 : 21[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127294:127687 [4] NCCL INFO Channel 06/0 : 20[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897247:897642 [7] NCCL INFO Channel 13/0 : 47[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897242:897639 [2] NCCL INFO Channel 05/0 : 42[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897245:897638 [5] NCCL INFO Channel 11/0 : 45[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127292:127692 [2] NCCL INFO Channel 00/0 : 18[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897243:897637 [3] NCCL INFO Channel 14/0 : 43[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897246:897641 [6] NCCL INFO Channel 08/0 : 46[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897244:897636 [4] NCCL INFO Channel 09/0 : 44[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127293:127689 [3] NCCL INFO Channel 04/0 : 19[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127296:127688 [6] NCCL INFO Channel 02/0 : 22[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897241:897635 [1] NCCL INFO Channel 13/0 : 41[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897240:897640 [0] NCCL INFO Channel 10/0 : 40[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127291:127690 [1] NCCL INFO Channel 04/0 : 17[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127297:127691 [7] NCCL INFO Channel 02/0 : 23[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127290:127685 [0] NCCL INFO Channel 04/0 : 16[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127294:127687 [4] NCCL INFO Channel 08/0 : 20[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897247:897642 [7] NCCL INFO Channel 14/0 : 47[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897242:897639 [2] NCCL INFO Channel 06/0 : 42[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127292:127692 [2] NCCL INFO Channel 04/0 : 18[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897246:897641 [6] NCCL INFO Channel 09/0 : 46[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127293:127689 [3] NCCL INFO Channel 06/0 : 19[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127296:127688 [6] NCCL INFO Channel 04/0 : 22[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127291:127690 [1] NCCL INFO Channel 06/0 : 17[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127290:127685 [0] NCCL INFO Channel 06/0 : 16[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762820:763219 [5] NCCL INFO Channel 00/0 : 21[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127297:127691 [7] NCCL INFO Channel 04/0 : 23[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127294:127687 [4] NCCL INFO Channel 10/0 : 20[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897244:897636 [4] NCCL INFO Channel 10/0 : 44[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897245:897638 [5] NCCL INFO Channel 12/0 : 45[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897243:897637 [3] NCCL INFO Channel 15/0 : 43[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897241:897635 [1] NCCL INFO Channel 14/0 : 41[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897240:897640 [0] NCCL INFO Channel 11/0 : 40[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897242:897639 [2] NCCL INFO Channel 07/0 : 42[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897246:897641 [6] NCCL INFO Channel 10/0 : 46[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127295:127686 [5] NCCL INFO Channel 01/0 : 13[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127293:127689 [3] NCCL INFO Channel 08/0 : 19[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127292:127692 [2] NCCL INFO Channel 06/0 : 18[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762820:763219 [5] NCCL INFO Channel 02/0 : 21[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127291:127690 [1] NCCL INFO Channel 08/0 : 17[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897244:897636 [4] NCCL INFO Channel 11/0 : 44[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897245:897638 [5] NCCL INFO Channel 14/0 : 45[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127296:127688 [6] NCCL INFO Channel 08/0 : 22[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897241:897635 [1] NCCL INFO Channel 15/0 : 41[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897240:897640 [0] NCCL INFO Channel 12/0 : 40[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127290:127685 [0] NCCL INFO Channel 08/0 : 16[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762820:763219 [5] NCCL INFO Channel 04/0 : 21[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127297:127691 [7] NCCL INFO Channel 06/0 : 23[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897242:897639 [2] NCCL INFO Channel 08/0 : 42[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897246:897641 [6] NCCL INFO Channel 11/0 : 46[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127294:127687 [4] NCCL INFO Channel 12/0 : 20[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897244:897636 [4] NCCL INFO Channel 13/0 : 44[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127295:127686 [5] NCCL INFO Channel 03/0 : 13[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127292:127692 [2] NCCL INFO Channel 08/0 : 18[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127293:127689 [3] NCCL INFO Channel 10/0 : 19[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897245:897638 [5] NCCL INFO Channel 15/0 : 45[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127291:127690 [1] NCCL INFO Channel 10/0 : 17[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762820:763219 [5] NCCL INFO Channel 06/0 : 21[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897240:897640 [0] NCCL INFO Channel 13/0 : 40[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897246:897641 [6] NCCL INFO Channel 12/0 : 46[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897242:897639 [2] NCCL INFO Channel 09/0 : 42[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897244:897636 [4] NCCL INFO Channel 14/0 : 44[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762820:763219 [5] NCCL INFO Channel 08/0 : 21[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897240:897640 [0] NCCL INFO Channel 14/0 : 40[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104436:104824 [7] NCCL INFO Channel 00/0 : 39[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762820:763219 [5] NCCL INFO Channel 10/0 : 21[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897242:897639 [2] NCCL INFO Channel 11/0 : 42[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897246:897641 [6] NCCL INFO Channel 13/0 : 46[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897244:897636 [4] NCCL INFO Channel 15/0 : 44[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897240:897640 [0] NCCL INFO Channel 15/0 : 40[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104436:104824 [7] NCCL INFO Channel 02/0 : 39[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762820:763219 [5] NCCL INFO Channel 12/0 : 21[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897246:897641 [6] NCCL INFO Channel 15/0 : 46[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897242:897639 [2] NCCL INFO Channel 12/0 : 42[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104432:104827 [3] NCCL INFO Channel 00/0 : 35[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762819:763221 [4] NCCL INFO Channel 00/0 : 20[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897242:897639 [2] NCCL INFO Channel 13/0 : 42[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104436:104824 [7] NCCL INFO Channel 04/0 : 39[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762820:763219 [5] NCCL INFO Channel 14/0 : 21[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104430:104830 [1] NCCL INFO Channel 00/0 : 33[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762819:763221 [4] NCCL INFO Channel 02/0 : 20[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104432:104827 [3] NCCL INFO Channel 02/0 : 35[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897242:897639 [2] NCCL INFO Channel 14/0 : 42[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104436:104824 [7] NCCL INFO Channel 06/0 : 39[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762819:763221 [4] NCCL INFO Channel 06/0 : 20[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104430:104830 [1] NCCL INFO Channel 02/0 : 33[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897242:897639 [2] NCCL INFO Channel 15/0 : 42[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762818:763220 [3] NCCL INFO Channel 00/0 : 19[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762819:763221 [4] NCCL INFO Channel 08/0 : 20[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104432:104827 [3] NCCL INFO Channel 04/0 : 35[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104436:104824 [7] NCCL INFO Channel 08/0 : 39[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104434:104828 [5] NCCL INFO Channel 00/0 : 37[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762815:763225 [0] NCCL INFO Channel 02/0 : 16[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104430:104830 [1] NCCL INFO Channel 04/0 : 33[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104432:104827 [3] NCCL INFO Channel 06/0 : 35[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762816:763218 [1] NCCL INFO Channel 00/0 : 17[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762818:763220 [3] NCCL INFO Channel 02/0 : 19[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104436:104824 [7] NCCL INFO Channel 10/0 : 39[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762819:763221 [4] NCCL INFO Channel 10/0 : 20[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762821:763224 [6] NCCL INFO Channel 00/0 : 22[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-46:702885:703278 [5] NCCL INFO Channel 00/0 : 37[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104434:104828 [5] NCCL INFO Channel 02/0 : 37[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762817:763223 [2] NCCL INFO Channel 00/0 : 18[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-46:702885:703278 [5] NCCL INFO Channel 02/0 : 37[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104433:104825 [4] NCCL INFO Channel 00/0 : 36[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104432:104827 [3] NCCL INFO Channel 08/0 : 35[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104430:104830 [1] NCCL INFO Channel 06/0 : 33[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762822:763222 [7] NCCL INFO Channel 00/0 : 23[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104436:104824 [7] NCCL INFO Channel 12/0 : 39[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762815:763225 [0] NCCL INFO Channel 04/0 : 16[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104434:104828 [5] NCCL INFO Channel 04/0 : 37[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762818:763220 [3] NCCL INFO Channel 04/0 : 19[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104429:104829 [0] NCCL INFO Channel 02/0 : 32[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104435:104826 [6] NCCL INFO Channel 00/0 : 38[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762816:763218 [1] NCCL INFO Channel 02/0 : 17[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-46:702885:703278 [5] NCCL INFO Channel 04/0 : 37[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104433:104825 [4] NCCL INFO Channel 02/0 : 36[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104432:104827 [3] NCCL INFO Channel 10/0 : 35[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104430:104830 [1] NCCL INFO Channel 08/0 : 33[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762819:763221 [4] NCCL INFO Channel 12/0 : 20[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104436:104824 [7] NCCL INFO Channel 14/0 : 39[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762817:763223 [2] NCCL INFO Channel 04/0 : 18[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-46:702885:703278 [5] NCCL INFO Channel 06/0 : 37[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104435:104826 [6] NCCL INFO Channel 02/0 : 38[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104429:104829 [0] NCCL INFO Channel 04/0 : 32[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104434:104828 [5] NCCL INFO Channel 06/0 : 37[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762821:763224 [6] NCCL INFO Channel 02/0 : 22[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762815:763225 [0] NCCL INFO Channel 06/0 : 16[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762822:763222 [7] NCCL INFO Channel 02/0 : 23[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762818:763220 [3] NCCL INFO Channel 06/0 : 19[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762816:763218 [1] NCCL INFO Channel 04/0 : 17[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762819:763221 [4] NCCL INFO Channel 14/0 : 20[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762817:763223 [2] NCCL INFO Channel 06/0 : 18[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762815:763225 [0] NCCL INFO Channel 08/0 : 16[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-46:702885:703278 [5] NCCL INFO Channel 08/0 : 37[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104433:104825 [4] NCCL INFO Channel 06/0 : 36[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104432:104827 [3] NCCL INFO Channel 12/0 : 35[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104430:104830 [1] NCCL INFO Channel 10/0 : 33[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762821:763224 [6] NCCL INFO Channel 04/0 : 22[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104434:104828 [5] NCCL INFO Channel 08/0 : 37[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104431:104831 [2] NCCL INFO Channel 00/0 : 34[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762822:763222 [7] NCCL INFO Channel 04/0 : 23[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762816:763218 [1] NCCL INFO Channel 06/0 : 17[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762818:763220 [3] NCCL INFO Channel 08/0 : 19[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-46:702885:703278 [5] NCCL INFO Channel 10/0 : 37[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104435:104826 [6] NCCL INFO Channel 04/0 : 38[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104429:104829 [0] NCCL INFO Channel 06/0 : 32[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762817:763223 [2] NCCL INFO Channel 08/0 : 18[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104430:104830 [1] NCCL INFO Channel 12/0 : 33[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104432:104827 [3] NCCL INFO Channel 14/0 : 35[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104433:104825 [4] NCCL INFO Channel 08/0 : 36[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762815:763225 [0] NCCL INFO Channel 10/0 : 16[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762821:763224 [6] NCCL INFO Channel 08/0 : 22[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-46:702885:703278 [5] NCCL INFO Channel 12/0 : 37[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104434:104828 [5] NCCL INFO Channel 10/0 : 37[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104431:104831 [2] NCCL INFO Channel 04/0 : 34[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762816:763218 [1] NCCL INFO Channel 08/0 : 17[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762818:763220 [3] NCCL INFO Channel 10/0 : 19[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762817:763223 [2] NCCL INFO Channel 10/0 : 18[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762822:763222 [7] NCCL INFO Channel 06/0 : 23[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762815:763225 [0] NCCL INFO Channel 12/0 : 16[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762821:763224 [6] NCCL INFO Channel 10/0 : 22[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-46:702885:703278 [5] NCCL INFO Channel 14/0 : 37[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104435:104826 [6] NCCL INFO Channel 08/0 : 38[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104429:104829 [0] NCCL INFO Channel 08/0 : 32[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104430:104830 [1] NCCL INFO Channel 14/0 : 33[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104433:104825 [4] NCCL INFO Channel 10/0 : 36[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104434:104828 [5] NCCL INFO Channel 12/0 : 37[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104431:104831 [2] NCCL INFO Channel 06/0 : 34[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762816:763218 [1] NCCL INFO Channel 10/0 : 17[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762818:763220 [3] NCCL INFO Channel 12/0 : 19[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762815:763225 [0] NCCL INFO Channel 14/0 : 16[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-46:702885:703278 [5] NCCL INFO Channel 00/0 : 5[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762817:763223 [2] NCCL INFO Channel 12/0 : 18[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762822:763222 [7] NCCL INFO Channel 08/0 : 23[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762821:763224 [6] NCCL INFO Channel 12/0 : 22[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762816:763218 [1] NCCL INFO Channel 12/0 : 17[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104435:104826 [6] NCCL INFO Channel 10/0 : 38[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762818:763220 [3] NCCL INFO Channel 14/0 : 19[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104433:104825 [4] NCCL INFO Channel 12/0 : 36[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104434:104828 [5] NCCL INFO Channel 14/0 : 37[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104429:104829 [0] NCCL INFO Channel 10/0 : 32[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104431:104831 [2] NCCL INFO Channel 08/0 : 34[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-46:702885:703278 [5] NCCL INFO Channel 02/0 : 5[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762817:763223 [2] NCCL INFO Channel 14/0 : 18[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762822:763222 [7] NCCL INFO Channel 10/0 : 23[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762821:763224 [6] NCCL INFO Channel 14/0 : 22[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762816:763218 [1] NCCL INFO Channel 14/0 : 17[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127290:127685 [0] NCCL INFO Channel 10/0 : 16[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127296:127688 [6] NCCL INFO Channel 10/0 : 22[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127297:127691 [7] NCCL INFO Channel 08/0 : 23[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127294:127687 [4] NCCL INFO Channel 14/0 : 20[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104435:104826 [6] NCCL INFO Channel 12/0 : 38[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127295:127686 [5] NCCL INFO Channel 05/0 : 13[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127293:127689 [3] NCCL INFO Channel 12/0 : 19[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104433:104825 [4] NCCL INFO Channel 14/0 : 36[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127292:127692 [2] NCCL INFO Channel 10/0 : 18[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127291:127690 [1] NCCL INFO Channel 12/0 : 17[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104431:104831 [2] NCCL INFO Channel 10/0 : 34[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104429:104829 [0] NCCL INFO Channel 12/0 : 32[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127290:127685 [0] NCCL INFO Channel 12/0 : 16[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127297:127691 [7] NCCL INFO Channel 10/0 : 23[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-46:702885:703278 [5] NCCL INFO Channel 04/0 : 5[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762822:763222 [7] NCCL INFO Channel 12/0 : 23[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127296:127688 [6] NCCL INFO Channel 12/0 : 22[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762822:763222 [7] NCCL INFO Channel 14/0 : 23[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127295:127686 [5] NCCL INFO Channel 07/0 : 13[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-46:702880:703277 [0] NCCL INFO Channel 02/0 : 32[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104435:104826 [6] NCCL INFO Channel 14/0 : 38[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127293:127689 [3] NCCL INFO Channel 14/0 : 19[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127291:127690 [1] NCCL INFO Channel 14/0 : 17[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127292:127692 [2] NCCL INFO Channel 12/0 : 18[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104431:104831 [2] NCCL INFO Channel 12/0 : 34[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127290:127685 [0] NCCL INFO Channel 14/0 : 16[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127296:127688 [6] NCCL INFO Channel 14/0 : 22[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127297:127691 [7] NCCL INFO Channel 12/0 : 23[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-46:702885:703278 [5] NCCL INFO Channel 06/0 : 5[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127295:127686 [5] NCCL INFO Channel 09/0 : 13[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127294:127687 [4] NCCL INFO Channel 01/0 : 12[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127292:127692 [2] NCCL INFO Channel 14/0 : 18[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127297:127691 [7] NCCL INFO Channel 14/0 : 23[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104429:104829 [0] NCCL INFO Channel 14/0 : 32[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127295:127686 [5] NCCL INFO Channel 11/0 : 13[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127294:127687 [4] NCCL INFO Channel 03/0 : 12[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-46:702880:703277 [0] NCCL INFO Channel 04/0 : 32[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104431:104831 [2] NCCL INFO Channel 14/0 : 34[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127294:127687 [4] NCCL INFO Channel 05/0 : 12[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127295:127686 [5] NCCL INFO Channel 15/0 : 13[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-46:702885:703278 [5] NCCL INFO Channel 08/0 : 5[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127293:127689 [3] NCCL INFO Channel 01/0 : 11[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127291:127690 [1] NCCL INFO Channel 01/0 : 9[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127290:127685 [0] NCCL INFO Channel 01/0 : 8[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127296:127688 [6] NCCL INFO Channel 01/0 : 14[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127294:127687 [4] NCCL INFO Channel 07/0 : 12[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127295:127686 [5] NCCL INFO Channel 01/0 : 29[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127293:127689 [3] NCCL INFO Channel 03/0 : 11[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127292:127692 [2] NCCL INFO Channel 01/0 : 10[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127291:127690 [1] NCCL INFO Channel 03/0 : 9[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-46:702880:703277 [0] NCCL INFO Channel 06/0 : 32[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-46:702885:703278 [5] NCCL INFO Channel 10/0 : 5[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127290:127685 [0] NCCL INFO Channel 03/0 : 8[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127296:127688 [6] NCCL INFO Channel 03/0 : 14[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127297:127691 [7] NCCL INFO Channel 01/0 : 15[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127294:127687 [4] NCCL INFO Channel 09/0 : 12[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127295:127686 [5] NCCL INFO Channel 03/0 : 29[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127293:127689 [3] NCCL INFO Channel 05/0 : 11[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127292:127692 [2] NCCL INFO Channel 03/0 : 10[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127291:127690 [1] NCCL INFO Channel 05/0 : 9[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127290:127685 [0] NCCL INFO Channel 05/0 : 8[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127296:127688 [6] NCCL INFO Channel 05/0 : 14[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127297:127691 [7] NCCL INFO Channel 03/0 : 15[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127294:127687 [4] NCCL INFO Channel 11/0 : 12[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127295:127686 [5] NCCL INFO Channel 05/0 : 29[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127293:127689 [3] NCCL INFO Channel 07/0 : 11[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127291:127690 [1] NCCL INFO Channel 07/0 : 9[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127292:127692 [2] NCCL INFO Channel 05/0 : 10[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127290:127685 [0] NCCL INFO Channel 07/0 : 8[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-46:702885:703278 [5] NCCL INFO Channel 12/0 : 5[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127296:127688 [6] NCCL INFO Channel 07/0 : 14[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127297:127691 [7] NCCL INFO Channel 05/0 : 15[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-46:702880:703277 [0] NCCL INFO Channel 08/0 : 32[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127293:127689 [3] NCCL INFO Channel 09/0 : 11[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127294:127687 [4] NCCL INFO Channel 13/0 : 12[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127295:127686 [5] NCCL INFO Channel 07/0 : 29[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127291:127690 [1] NCCL INFO Channel 11/0 : 9[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127292:127692 [2] NCCL INFO Channel 07/0 : 10[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127290:127685 [0] NCCL INFO Channel 09/0 : 8[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127297:127691 [7] NCCL INFO Channel 07/0 : 15[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127293:127689 [3] NCCL INFO Channel 13/0 : 11[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127296:127688 [6] NCCL INFO Channel 09/0 : 14[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127294:127687 [4] NCCL INFO Channel 15/0 : 12[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-46:702885:703278 [5] NCCL INFO Channel 14/0 : 5[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127295:127686 [5] NCCL INFO Channel 09/0 : 29[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127291:127690 [1] NCCL INFO Channel 13/0 : 9[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127292:127692 [2] NCCL INFO Channel 09/0 : 10[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127290:127685 [0] NCCL INFO Channel 11/0 : 8[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-46:702880:703277 [0] NCCL INFO Channel 10/0 : 32[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127297:127691 [7] NCCL INFO Channel 09/0 : 15[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9630:10024 [5] NCCL INFO Channel 01/0 : 29[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127293:127689 [3] NCCL INFO Channel 15/0 : 11[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127296:127688 [6] NCCL INFO Channel 11/0 : 14[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127294:127687 [4] NCCL INFO Channel 01/0 : 28[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127290:127685 [0] NCCL INFO Channel 13/0 : 8[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127291:127690 [1] NCCL INFO Channel 15/0 : 9[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127295:127686 [5] NCCL INFO Channel 11/0 : 29[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127292:127692 [2] NCCL INFO Channel 11/0 : 10[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127297:127691 [7] NCCL INFO Channel 11/0 : 15[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-46:702880:703277 [0] NCCL INFO Channel 12/0 : 32[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127293:127689 [3] NCCL INFO Channel 01/0 : 27[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127296:127688 [6] NCCL INFO Channel 13/0 : 14[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9630:10024 [5] NCCL INFO Channel 03/0 : 29[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127294:127687 [4] NCCL INFO Channel 03/0 : 28[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-46:702880:703277 [0] NCCL INFO Channel 14/0 : 32[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9630:10024 [5] NCCL INFO Channel 05/0 : 29[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127290:127685 [0] NCCL INFO Channel 15/0 : 8[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127291:127690 [1] NCCL INFO Channel 01/0 : 25[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127292:127692 [2] NCCL INFO Channel 13/0 : 10[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127295:127686 [5] NCCL INFO Channel 15/0 : 29[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127293:127689 [3] NCCL INFO Channel 03/0 : 27[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127297:127691 [7] NCCL INFO Channel 13/0 : 15[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127296:127688 [6] NCCL INFO Channel 15/0 : 14[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127294:127687 [4] NCCL INFO Channel 05/0 : 28[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-46:702880:703277 [0] NCCL INFO Channel 02/0 : 0[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127290:127685 [0] NCCL INFO Channel 01/0 : 24[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127291:127690 [1] NCCL INFO Channel 03/0 : 25[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9630:10024 [5] NCCL INFO Channel 07/0 : 29[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127292:127692 [2] NCCL INFO Channel 15/0 : 10[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127293:127689 [3] NCCL INFO Channel 05/0 : 27[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127297:127691 [7] NCCL INFO Channel 01/0 : 31[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127296:127688 [6] NCCL INFO Channel 01/0 : 30[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127290:127685 [0] NCCL INFO Channel 03/0 : 24[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127294:127687 [4] NCCL INFO Channel 07/0 : 28[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127291:127690 [1] NCCL INFO Channel 05/0 : 25[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127292:127692 [2] NCCL INFO Channel 01/0 : 26[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127293:127689 [3] NCCL INFO Channel 07/0 : 27[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127290:127685 [0] NCCL INFO Channel 05/0 : 24[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127296:127688 [6] NCCL INFO Channel 03/0 : 30[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127294:127687 [4] NCCL INFO Channel 09/0 : 28[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127291:127690 [1] NCCL INFO Channel 07/0 : 25[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127297:127691 [7] NCCL INFO Channel 03/0 : 31[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-46:702880:703277 [0] NCCL INFO Channel 04/0 : 0[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127292:127692 [2] NCCL INFO Channel 03/0 : 26[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127293:127689 [3] NCCL INFO Channel 09/0 : 27[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127296:127688 [6] NCCL INFO Channel 05/0 : 30[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9630:10024 [5] NCCL INFO Channel 09/0 : 29[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-46:702880:703277 [0] NCCL INFO Channel 06/0 : 0[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127291:127690 [1] NCCL INFO Channel 11/0 : 25[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127290:127685 [0] NCCL INFO Channel 07/0 : 24[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127294:127687 [4] NCCL INFO Channel 11/0 : 28[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127292:127692 [2] NCCL INFO Channel 05/0 : 26[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127293:127689 [3] NCCL INFO Channel 13/0 : 27[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9630:10024 [5] NCCL INFO Channel 11/0 : 29[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127297:127691 [7] NCCL INFO Channel 05/0 : 31[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127296:127688 [6] NCCL INFO Channel 07/0 : 30[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127291:127690 [1] NCCL INFO Channel 13/0 : 25[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127290:127685 [0] NCCL INFO Channel 09/0 : 24[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-46:702880:703277 [0] NCCL INFO Channel 08/0 : 0[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127292:127692 [2] NCCL INFO Channel 07/0 : 26[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127293:127689 [3] NCCL INFO Channel 15/0 : 27[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127294:127687 [4] NCCL INFO Channel 13/0 : 28[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127297:127691 [7] NCCL INFO Channel 07/0 : 31[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127296:127688 [6] NCCL INFO Channel 09/0 : 30[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9630:10024 [5] NCCL INFO Channel 15/0 : 29[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-46:702880:703277 [0] NCCL INFO Channel 10/0 : 0[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127291:127690 [1] NCCL INFO Channel 15/0 : 25[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127290:127685 [0] NCCL INFO Channel 11/0 : 24[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127292:127692 [2] NCCL INFO Channel 09/0 : 26[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127294:127687 [4] NCCL INFO Channel 15/0 : 28[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127296:127688 [6] NCCL INFO Channel 11/0 : 30[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127297:127691 [7] NCCL INFO Channel 09/0 : 31[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127290:127685 [0] NCCL INFO Channel 13/0 : 24[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-46:702880:703277 [0] NCCL INFO Channel 12/0 : 0[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127292:127692 [2] NCCL INFO Channel 11/0 : 26[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127296:127688 [6] NCCL INFO Channel 13/0 : 30[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127297:127691 [7] NCCL INFO Channel 11/0 : 31[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9628:10023 [3] NCCL INFO Channel 01/0 : 27[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127290:127685 [0] NCCL INFO Channel 15/0 : 24[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-46:702880:703277 [0] NCCL INFO Channel 14/0 : 0[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9626:10025 [1] NCCL INFO Channel 01/0 : 25[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127296:127688 [6] NCCL INFO Channel 15/0 : 30[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127297:127691 [7] NCCL INFO Channel 13/0 : 31[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127292:127692 [2] NCCL INFO Channel 13/0 : 26[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9628:10023 [3] NCCL INFO Channel 03/0 : 27[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9629:10021 [4] NCCL INFO Channel 01/0 : 28[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9626:10025 [1] NCCL INFO Channel 03/0 : 25[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127292:127692 [2] NCCL INFO Channel 15/0 : 26[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9628:10023 [3] NCCL INFO Channel 05/0 : 27[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9625:10020 [0] NCCL INFO Channel 01/0 : 24[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9626:10025 [1] NCCL INFO Channel 05/0 : 25[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9628:10023 [3] NCCL INFO Channel 07/0 : 27[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9629:10021 [4] NCCL INFO Channel 03/0 : 28[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9632:10026 [7] NCCL INFO Channel 01/0 : 31[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9631:10022 [6] NCCL INFO Channel 01/0 : 30[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9625:10020 [0] NCCL INFO Channel 03/0 : 24[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9626:10025 [1] NCCL INFO Channel 07/0 : 25[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9628:10023 [3] NCCL INFO Channel 09/0 : 27[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9629:10021 [4] NCCL INFO Channel 05/0 : 28[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9627:10027 [2] NCCL INFO Channel 01/0 : 26[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9632:10026 [7] NCCL INFO Channel 03/0 : 31[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9625:10020 [0] NCCL INFO Channel 05/0 : 24[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9626:10025 [1] NCCL INFO Channel 11/0 : 25[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9631:10022 [6] NCCL INFO Channel 03/0 : 30[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9628:10023 [3] NCCL INFO Channel 13/0 : 27[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9629:10021 [4] NCCL INFO Channel 07/0 : 28[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9627:10027 [2] NCCL INFO Channel 03/0 : 26[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9632:10026 [7] NCCL INFO Channel 05/0 : 31[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9625:10020 [0] NCCL INFO Channel 07/0 : 24[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9626:10025 [1] NCCL INFO Channel 13/0 : 25[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9631:10022 [6] NCCL INFO Channel 05/0 : 30[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9628:10023 [3] NCCL INFO Channel 15/0 : 27[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9627:10027 [2] NCCL INFO Channel 05/0 : 26[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9625:10020 [0] NCCL INFO Channel 09/0 : 24[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9629:10021 [4] NCCL INFO Channel 09/0 : 28[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9632:10026 [7] NCCL INFO Channel 07/0 : 31[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9626:10025 [1] NCCL INFO Channel 15/0 : 25[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9631:10022 [6] NCCL INFO Channel 07/0 : 30[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9625:10020 [0] NCCL INFO Channel 11/0 : 24[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9629:10021 [4] NCCL INFO Channel 11/0 : 28[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9632:10026 [7] NCCL INFO Channel 09/0 : 31[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9631:10022 [6] NCCL INFO Channel 09/0 : 30[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9627:10027 [2] NCCL INFO Channel 07/0 : 26[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9625:10020 [0] NCCL INFO Channel 13/0 : 24[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9629:10021 [4] NCCL INFO Channel 13/0 : 28[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9631:10022 [6] NCCL INFO Channel 11/0 : 30[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9632:10026 [7] NCCL INFO Channel 11/0 : 31[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9627:10027 [2] NCCL INFO Channel 09/0 : 26[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9625:10020 [0] NCCL INFO Channel 15/0 : 24[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9629:10021 [4] NCCL INFO Channel 15/0 : 28[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9627:10027 [2] NCCL INFO Channel 11/0 : 26[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9632:10026 [7] NCCL INFO Channel 13/0 : 31[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9631:10022 [6] NCCL INFO Channel 13/0 : 30[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9627:10027 [2] NCCL INFO Channel 13/0 : 26[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9631:10022 [6] NCCL INFO Channel 15/0 : 30[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9627:10027 [2] NCCL INFO Channel 15/0 : 26[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624949:625344 [6] NCCL INFO Channel 01/0 : 38[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624947:625339 [4] NCCL INFO Channel 01/0 : 36[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624945:625340 [2] NCCL INFO Channel 01/0 : 34[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624949:625344 [6] NCCL INFO Channel 03/0 : 38[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624947:625339 [4] NCCL INFO Channel 03/0 : 36[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624945:625340 [2] NCCL INFO Channel 03/0 : 34[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624949:625344 [6] NCCL INFO Channel 05/0 : 38[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624947:625339 [4] NCCL INFO Channel 05/0 : 36[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624943:625338 [0] NCCL INFO Channel 01/0 : 32[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624945:625340 [2] NCCL INFO Channel 05/0 : 34[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624948:625343 [5] NCCL INFO Channel 01/0 : 37[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624947:625339 [4] NCCL INFO Channel 07/0 : 36[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624949:625344 [6] NCCL INFO Channel 07/0 : 38[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624943:625338 [0] NCCL INFO Channel 03/0 : 32[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624945:625340 [2] NCCL INFO Channel 07/0 : 34[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624948:625343 [5] NCCL INFO Channel 03/0 : 37[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624947:625339 [4] NCCL INFO Channel 09/0 : 36[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624949:625344 [6] NCCL INFO Channel 09/0 : 38[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624945:625340 [2] NCCL INFO Channel 09/0 : 34[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624943:625338 [0] NCCL INFO Channel 05/0 : 32[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624944:625341 [1] NCCL INFO Channel 01/0 : 33[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624948:625343 [5] NCCL INFO Channel 05/0 : 37[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624950:625342 [7] NCCL INFO Channel 01/0 : 39[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624946:625337 [3] NCCL INFO Channel 01/0 : 35[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624947:625339 [4] NCCL INFO Channel 11/0 : 36[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624949:625344 [6] NCCL INFO Channel 11/0 : 38[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624945:625340 [2] NCCL INFO Channel 11/0 : 34[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624943:625338 [0] NCCL INFO Channel 07/0 : 32[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624948:625343 [5] NCCL INFO Channel 07/0 : 37[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624944:625341 [1] NCCL INFO Channel 03/0 : 33[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624949:625344 [6] NCCL INFO Channel 13/0 : 38[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624947:625339 [4] NCCL INFO Channel 13/0 : 36[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624950:625342 [7] NCCL INFO Channel 03/0 : 39[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624946:625337 [3] NCCL INFO Channel 03/0 : 35[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624945:625340 [2] NCCL INFO Channel 13/0 : 34[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624948:625343 [5] NCCL INFO Channel 09/0 : 37[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624943:625338 [0] NCCL INFO Channel 09/0 : 32[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624949:625344 [6] NCCL INFO Channel 15/0 : 38[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624947:625339 [4] NCCL INFO Channel 15/0 : 36[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624950:625342 [7] NCCL INFO Channel 05/0 : 39[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624945:625340 [2] NCCL INFO Channel 15/0 : 34[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624946:625337 [3] NCCL INFO Channel 05/0 : 35[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624948:625343 [5] NCCL INFO Channel 11/0 : 37[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624944:625341 [1] NCCL INFO Channel 05/0 : 33[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624943:625338 [0] NCCL INFO Channel 11/0 : 32[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624950:625342 [7] NCCL INFO Channel 07/0 : 39[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624948:625343 [5] NCCL INFO Channel 15/0 : 37[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624944:625341 [1] NCCL INFO Channel 07/0 : 33[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624943:625338 [0] NCCL INFO Channel 13/0 : 32[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624946:625337 [3] NCCL INFO Channel 07/0 : 35[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624950:625342 [7] NCCL INFO Channel 09/0 : 39[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897246:897641 [6] NCCL INFO Channel 01/0 : 30[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897244:897636 [4] NCCL INFO Channel 01/0 : 28[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624949:625344 [6] NCCL INFO Channel 00/0 : 22[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624944:625341 [1] NCCL INFO Channel 11/0 : 33[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897242:897639 [2] NCCL INFO Channel 01/0 : 26[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624946:625337 [3] NCCL INFO Channel 09/0 : 35[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624943:625338 [0] NCCL INFO Channel 15/0 : 32[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624950:625342 [7] NCCL INFO Channel 11/0 : 39[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624947:625339 [4] NCCL INFO Channel 00/0 : 20[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624949:625344 [6] NCCL INFO Channel 02/0 : 22[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624945:625340 [2] NCCL INFO Channel 00/0 : 18[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897246:897641 [6] NCCL INFO Channel 03/0 : 30[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897244:897636 [4] NCCL INFO Channel 03/0 : 28[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624944:625341 [1] NCCL INFO Channel 13/0 : 33[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624946:625337 [3] NCCL INFO Channel 13/0 : 35[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624950:625342 [7] NCCL INFO Channel 13/0 : 39[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897242:897639 [2] NCCL INFO Channel 03/0 : 26[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624947:625339 [4] NCCL INFO Channel 02/0 : 20[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624948:625343 [5] NCCL INFO Channel 00/0 : 21[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624949:625344 [6] NCCL INFO Channel 04/0 : 22[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897245:897638 [5] NCCL INFO Channel 01/0 : 29[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897246:897641 [6] NCCL INFO Channel 05/0 : 30[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897244:897636 [4] NCCL INFO Channel 05/0 : 28[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624945:625340 [2] NCCL INFO Channel 04/0 : 18[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624944:625341 [1] NCCL INFO Channel 15/0 : 33[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624946:625337 [3] NCCL INFO Channel 15/0 : 35[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897242:897639 [2] NCCL INFO Channel 05/0 : 26[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897245:897638 [5] NCCL INFO Channel 03/0 : 29[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897246:897641 [6] NCCL INFO Channel 07/0 : 30[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897244:897636 [4] NCCL INFO Channel 07/0 : 28[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624949:625344 [6] NCCL INFO Channel 08/0 : 22[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624947:625339 [4] NCCL INFO Channel 06/0 : 20[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624948:625343 [5] NCCL INFO Channel 02/0 : 21[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624943:625338 [0] NCCL INFO Channel 02/0 : 16[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624945:625340 [2] NCCL INFO Channel 06/0 : 18[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897240:897640 [0] NCCL INFO Channel 01/0 : 24[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897245:897638 [5] NCCL INFO Channel 05/0 : 29[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624949:625344 [6] NCCL INFO Channel 10/0 : 22[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897242:897639 [2] NCCL INFO Channel 07/0 : 26[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897246:897641 [6] NCCL INFO Channel 09/0 : 30[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897244:897636 [4] NCCL INFO Channel 09/0 : 28[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624948:625343 [5] NCCL INFO Channel 04/0 : 21[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624947:625339 [4] NCCL INFO Channel 08/0 : 20[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897240:897640 [0] NCCL INFO Channel 03/0 : 24[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897245:897638 [5] NCCL INFO Channel 07/0 : 29[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624943:625338 [0] NCCL INFO Channel 04/0 : 16[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624945:625340 [2] NCCL INFO Channel 08/0 : 18[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624950:625342 [7] NCCL INFO Channel 00/0 : 23[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624944:625341 [1] NCCL INFO Channel 00/0 : 17[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624946:625337 [3] NCCL INFO Channel 00/0 : 19[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624949:625344 [6] NCCL INFO Channel 12/0 : 22[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897247:897642 [7] NCCL INFO Channel 01/0 : 31[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897244:897636 [4] NCCL INFO Channel 11/0 : 28[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897242:897639 [2] NCCL INFO Channel 09/0 : 26[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897246:897641 [6] NCCL INFO Channel 11/0 : 30[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624948:625343 [5] NCCL INFO Channel 06/0 : 21[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624947:625339 [4] NCCL INFO Channel 10/0 : 20[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624943:625338 [0] NCCL INFO Channel 06/0 : 16[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897245:897638 [5] NCCL INFO Channel 09/0 : 29[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897240:897640 [0] NCCL INFO Channel 05/0 : 24[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624945:625340 [2] NCCL INFO Channel 10/0 : 18[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624950:625342 [7] NCCL INFO Channel 02/0 : 23[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897241:897635 [1] NCCL INFO Channel 01/0 : 25[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897243:897637 [3] NCCL INFO Channel 01/0 : 27[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624949:625344 [6] NCCL INFO Channel 14/0 : 22[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624944:625341 [1] NCCL INFO Channel 02/0 : 17[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624946:625337 [3] NCCL INFO Channel 02/0 : 19[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624948:625343 [5] NCCL INFO Channel 08/0 : 21[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624947:625339 [4] NCCL INFO Channel 12/0 : 20[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897246:897641 [6] NCCL INFO Channel 13/0 : 30[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897247:897642 [7] NCCL INFO Channel 03/0 : 31[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897244:897636 [4] NCCL INFO Channel 13/0 : 28[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897242:897639 [2] NCCL INFO Channel 11/0 : 26[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624943:625338 [0] NCCL INFO Channel 08/0 : 16[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624950:625342 [7] NCCL INFO Channel 04/0 : 23[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624945:625340 [2] NCCL INFO Channel 12/0 : 18[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624949:625344 [6] NCCL INFO Channel 00/0 : 38[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897245:897638 [5] NCCL INFO Channel 11/0 : 29[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897241:897635 [1] NCCL INFO Channel 03/0 : 25[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897240:897640 [0] NCCL INFO Channel 07/0 : 24[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897243:897637 [3] NCCL INFO Channel 03/0 : 27[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897246:897641 [6] NCCL INFO Channel 15/0 : 30[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624944:625341 [1] NCCL INFO Channel 04/0 : 17[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624946:625337 [3] NCCL INFO Channel 04/0 : 19[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624948:625343 [5] NCCL INFO Channel 10/0 : 21[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624947:625339 [4] NCCL INFO Channel 14/0 : 20[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897247:897642 [7] NCCL INFO Channel 05/0 : 31[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897244:897636 [4] NCCL INFO Channel 15/0 : 28[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897242:897639 [2] NCCL INFO Channel 13/0 : 26[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897245:897638 [5] NCCL INFO Channel 15/0 : 29[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897241:897635 [1] NCCL INFO Channel 05/0 : 25[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897243:897637 [3] NCCL INFO Channel 05/0 : 27[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897240:897640 [0] NCCL INFO Channel 09/0 : 24[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624943:625338 [0] NCCL INFO Channel 10/0 : 16[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624950:625342 [7] NCCL INFO Channel 06/0 : 23[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624945:625340 [2] NCCL INFO Channel 14/0 : 18[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624949:625344 [6] NCCL INFO Channel 02/0 : 38[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624944:625341 [1] NCCL INFO Channel 06/0 : 17[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624946:625337 [3] NCCL INFO Channel 06/0 : 19[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624948:625343 [5] NCCL INFO Channel 12/0 : 21[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897247:897642 [7] NCCL INFO Channel 07/0 : 31[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624947:625339 [4] NCCL INFO Channel 00/0 : 36[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897242:897639 [2] NCCL INFO Channel 15/0 : 26[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897241:897635 [1] NCCL INFO Channel 07/0 : 25[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897243:897637 [3] NCCL INFO Channel 07/0 : 27[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624950:625342 [7] NCCL INFO Channel 08/0 : 23[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624943:625338 [0] NCCL INFO Channel 12/0 : 16[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624949:625344 [6] NCCL INFO Channel 04/0 : 38[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624945:625340 [2] NCCL INFO Channel 00/0 : 34[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897240:897640 [0] NCCL INFO Channel 11/0 : 24[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127296:127688 [6] NCCL INFO Channel 01/0 : 62[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897247:897642 [7] NCCL INFO Channel 09/0 : 31[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897246:897641 [6] NCCL INFO Channel 01/0 : 46[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897241:897635 [1] NCCL INFO Channel 11/0 : 25[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624946:625337 [3] NCCL INFO Channel 08/0 : 19[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624948:625343 [5] NCCL INFO Channel 14/0 : 21[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624947:625339 [4] NCCL INFO Channel 02/0 : 36[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624944:625341 [1] NCCL INFO Channel 08/0 : 17[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624950:625342 [7] NCCL INFO Channel 10/0 : 23[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624943:625338 [0] NCCL INFO Channel 14/0 : 16[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127296:127688 [6] NCCL INFO Channel 03/0 : 62[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897243:897637 [3] NCCL INFO Channel 09/0 : 27[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897240:897640 [0] NCCL INFO Channel 13/0 : 24[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897244:897636 [4] NCCL INFO Channel 01/0 : 44[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897247:897642 [7] NCCL INFO Channel 11/0 : 31[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897246:897641 [6] NCCL INFO Channel 03/0 : 46[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897245:897638 [5] NCCL INFO Channel 01/0 : 45[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624949:625344 [6] NCCL INFO Channel 08/0 : 38[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624945:625340 [2] NCCL INFO Channel 04/0 : 34[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624946:625337 [3] NCCL INFO Channel 10/0 : 19[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624948:625343 [5] NCCL INFO Channel 00/0 : 37[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127294:127687 [4] NCCL INFO Channel 01/0 : 60[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127295:127686 [5] NCCL INFO Channel 01/0 : 61[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897241:897635 [1] NCCL INFO Channel 13/0 : 25[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897243:897637 [3] NCCL INFO Channel 13/0 : 27[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897240:897640 [0] NCCL INFO Channel 15/0 : 24[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897242:897639 [2] NCCL INFO Channel 01/0 : 42[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624947:625339 [4] NCCL INFO Channel 06/0 : 36[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127292:127692 [2] NCCL INFO Channel 01/0 : 58[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897244:897636 [4] NCCL INFO Channel 03/0 : 44[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624944:625341 [1] NCCL INFO Channel 10/0 : 17[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127296:127688 [6] NCCL INFO Channel 05/0 : 62[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897247:897642 [7] NCCL INFO Channel 13/0 : 31[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624950:625342 [7] NCCL INFO Channel 12/0 : 23[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624949:625344 [6] NCCL INFO Channel 10/0 : 38[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624943:625338 [0] NCCL INFO Channel 02/0 : 32[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624945:625340 [2] NCCL INFO Channel 06/0 : 34[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127294:127687 [4] NCCL INFO Channel 03/0 : 60[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897246:897641 [6] NCCL INFO Channel 05/0 : 46[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127295:127686 [5] NCCL INFO Channel 03/0 : 61[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897245:897638 [5] NCCL INFO Channel 03/0 : 45[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624948:625343 [5] NCCL INFO Channel 02/0 : 37[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127292:127692 [2] NCCL INFO Channel 03/0 : 58[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127296:127688 [6] NCCL INFO Channel 07/0 : 62[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897241:897635 [1] NCCL INFO Channel 15/0 : 25[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897243:897637 [3] NCCL INFO Channel 15/0 : 27[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897244:897636 [4] NCCL INFO Channel 05/0 : 44[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624947:625339 [4] NCCL INFO Channel 08/0 : 36[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897242:897639 [2] NCCL INFO Channel 03/0 : 42[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624946:625337 [3] NCCL INFO Channel 12/0 : 19[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897246:897641 [6] NCCL INFO Channel 07/0 : 46[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624944:625341 [1] NCCL INFO Channel 12/0 : 17[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897245:897638 [5] NCCL INFO Channel 05/0 : 45[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624950:625342 [7] NCCL INFO Channel 14/0 : 23[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624949:625344 [6] NCCL INFO Channel 12/0 : 38[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127294:127687 [4] NCCL INFO Channel 05/0 : 60[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127295:127686 [5] NCCL INFO Channel 05/0 : 61[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127296:127688 [6] NCCL INFO Channel 09/0 : 62[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127292:127692 [2] NCCL INFO Channel 05/0 : 58[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897244:897636 [4] NCCL INFO Channel 07/0 : 44[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897242:897639 [2] NCCL INFO Channel 05/0 : 42[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897240:897640 [0] NCCL INFO Channel 01/0 : 40[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897246:897641 [6] NCCL INFO Channel 09/0 : 46[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897245:897638 [5] NCCL INFO Channel 07/0 : 45[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897247:897642 [7] NCCL INFO Channel 01/0 : 47[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624945:625340 [2] NCCL INFO Channel 08/0 : 34[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624943:625338 [0] NCCL INFO Channel 04/0 : 32[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624948:625343 [5] NCCL INFO Channel 04/0 : 37[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624947:625339 [4] NCCL INFO Channel 10/0 : 36[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127290:127685 [0] NCCL INFO Channel 01/0 : 56[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897244:897636 [4] NCCL INFO Channel 09/0 : 44[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897242:897639 [2] NCCL INFO Channel 07/0 : 42[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624946:625337 [3] NCCL INFO Channel 14/0 : 19[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624944:625341 [1] NCCL INFO Channel 14/0 : 17[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127295:127686 [5] NCCL INFO Channel 07/0 : 61[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127296:127688 [6] NCCL INFO Channel 11/0 : 62[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127292:127692 [2] NCCL INFO Channel 07/0 : 58[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127294:127687 [4] NCCL INFO Channel 07/0 : 60[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127297:127691 [7] NCCL INFO Channel 01/0 : 63[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897241:897635 [1] NCCL INFO Channel 01/0 : 41[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897245:897638 [5] NCCL INFO Channel 09/0 : 45[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127290:127685 [0] NCCL INFO Channel 03/0 : 56[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624950:625342 [7] NCCL INFO Channel 00/0 : 39[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624949:625344 [6] NCCL INFO Channel 14/0 : 38[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624945:625340 [2] NCCL INFO Channel 10/0 : 34[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624943:625338 [0] NCCL INFO Channel 06/0 : 32[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624948:625343 [5] NCCL INFO Channel 06/0 : 37[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624947:625339 [4] NCCL INFO Channel 12/0 : 36[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897240:897640 [0] NCCL INFO Channel 03/0 : 40[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624946:625337 [3] NCCL INFO Channel 00/0 : 35[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624944:625341 [1] NCCL INFO Channel 00/0 : 33[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127291:127690 [1] NCCL INFO Channel 01/0 : 57[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897246:897641 [6] NCCL INFO Channel 11/0 : 46[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897243:897637 [3] NCCL INFO Channel 01/0 : 43[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624950:625342 [7] NCCL INFO Channel 02/0 : 39[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127293:127689 [3] NCCL INFO Channel 01/0 : 59[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897247:897642 [7] NCCL INFO Channel 03/0 : 47[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624945:625340 [2] NCCL INFO Channel 12/0 : 34[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127295:127686 [5] NCCL INFO Channel 09/0 : 61[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897244:897636 [4] NCCL INFO Channel 11/0 : 44[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897242:897639 [2] NCCL INFO Channel 09/0 : 42[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624943:625338 [0] NCCL INFO Channel 08/0 : 32[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127292:127692 [2] NCCL INFO Channel 09/0 : 58[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127296:127688 [6] NCCL INFO Channel 13/0 : 62[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897241:897635 [1] NCCL INFO Channel 03/0 : 41[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897245:897638 [5] NCCL INFO Channel 11/0 : 45[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624948:625343 [5] NCCL INFO Channel 08/0 : 37[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127294:127687 [4] NCCL INFO Channel 09/0 : 60[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897240:897640 [0] NCCL INFO Channel 05/0 : 40[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624947:625339 [4] NCCL INFO Channel 14/0 : 36[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127297:127691 [7] NCCL INFO Channel 03/0 : 63[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624944:625341 [1] NCCL INFO Channel 02/0 : 33[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624946:625337 [3] NCCL INFO Channel 02/0 : 35[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624950:625342 [7] NCCL INFO Channel 04/0 : 39[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127291:127690 [1] NCCL INFO Channel 03/0 : 57[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127290:127685 [0] NCCL INFO Channel 05/0 : 56[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127293:127689 [3] NCCL INFO Channel 03/0 : 59[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762821:763224 [6] NCCL INFO Channel 00/0 : 38[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624945:625340 [2] NCCL INFO Channel 14/0 : 34[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624943:625338 [0] NCCL INFO Channel 10/0 : 32[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624948:625343 [5] NCCL INFO Channel 10/0 : 37[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897246:897641 [6] NCCL INFO Channel 13/0 : 46[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897243:897637 [3] NCCL INFO Channel 03/0 : 43[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897247:897642 [7] NCCL INFO Channel 05/0 : 47[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127295:127686 [5] NCCL INFO Channel 11/0 : 61[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127292:127692 [2] NCCL INFO Channel 11/0 : 58[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762821:763224 [6] NCCL INFO Channel 02/0 : 38[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897244:897636 [4] NCCL INFO Channel 13/0 : 44[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897242:897639 [2] NCCL INFO Channel 11/0 : 42[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127296:127688 [6] NCCL INFO Channel 15/0 : 62[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897241:897635 [1] NCCL INFO Channel 05/0 : 41[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624944:625341 [1] NCCL INFO Channel 04/0 : 33[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127294:127687 [4] NCCL INFO Channel 11/0 : 60[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897245:897638 [5] NCCL INFO Channel 15/0 : 45[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624946:625337 [3] NCCL INFO Channel 04/0 : 35[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624950:625342 [7] NCCL INFO Channel 06/0 : 39[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127297:127691 [7] NCCL INFO Channel 05/0 : 63[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897240:897640 [0] NCCL INFO Channel 07/0 : 40[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624948:625343 [5] NCCL INFO Channel 12/0 : 37[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624943:625338 [0] NCCL INFO Channel 12/0 : 32[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762819:763221 [4] NCCL INFO Channel 00/0 : 36[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104435:104826 [6] NCCL INFO Channel 00/0 : 54[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127291:127690 [1] NCCL INFO Channel 05/0 : 57[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127290:127685 [0] NCCL INFO Channel 07/0 : 56[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127293:127689 [3] NCCL INFO Channel 05/0 : 59[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897246:897641 [6] NCCL INFO Channel 15/0 : 46[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127292:127692 [2] NCCL INFO Channel 13/0 : 58[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127295:127686 [5] NCCL INFO Channel 15/0 : 61[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127296:127688 [6] NCCL INFO Channel 01/0 : 30[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762821:763224 [6] NCCL INFO Channel 04/0 : 38[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762817:763223 [2] NCCL INFO Channel 00/0 : 34[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897247:897642 [7] NCCL INFO Channel 07/0 : 47[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897244:897636 [4] NCCL INFO Channel 15/0 : 44[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897243:897637 [3] NCCL INFO Channel 05/0 : 43[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624946:625337 [3] NCCL INFO Channel 06/0 : 35[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127294:127687 [4] NCCL INFO Channel 13/0 : 60[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104435:104826 [6] NCCL INFO Channel 02/0 : 54[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897242:897639 [2] NCCL INFO Channel 13/0 : 42[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897241:897635 [1] NCCL INFO Channel 07/0 : 41[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897240:897640 [0] NCCL INFO Channel 09/0 : 40[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624950:625342 [7] NCCL INFO Channel 08/0 : 39[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624948:625343 [5] NCCL INFO Channel 14/0 : 37[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624949:625344 [6] NCCL INFO Channel 00/0 : 6[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127297:127691 [7] NCCL INFO Channel 07/0 : 63[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624944:625341 [1] NCCL INFO Channel 06/0 : 33[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127290:127685 [0] NCCL INFO Channel 09/0 : 56[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624946:625337 [3] NCCL INFO Channel 08/0 : 35[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624943:625338 [0] NCCL INFO Channel 14/0 : 32[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127291:127690 [1] NCCL INFO Channel 07/0 : 57[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762819:763221 [4] NCCL INFO Channel 02/0 : 36[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104433:104825 [4] NCCL INFO Channel 00/0 : 52[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897247:897642 [7] NCCL INFO Channel 09/0 : 47[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897243:897637 [3] NCCL INFO Channel 07/0 : 43[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897241:897635 [1] NCCL INFO Channel 11/0 : 41[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897242:897639 [2] NCCL INFO Channel 15/0 : 42[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897240:897640 [0] NCCL INFO Channel 11/0 : 40[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127293:127689 [3] NCCL INFO Channel 07/0 : 59[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127292:127692 [2] NCCL INFO Channel 15/0 : 58[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762821:763224 [6] NCCL INFO Channel 08/0 : 38[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624947:625339 [4] NCCL INFO Channel 00/0 : 4[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624949:625344 [6] NCCL INFO Channel 02/0 : 6[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624950:625342 [7] NCCL INFO Channel 10/0 : 39[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624946:625337 [3] NCCL INFO Channel 10/0 : 35[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127296:127688 [6] NCCL INFO Channel 03/0 : 30[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762817:763223 [2] NCCL INFO Channel 04/0 : 34[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127295:127686 [5] NCCL INFO Channel 01/0 : 29[5] -> 61[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104435:104826 [6] NCCL INFO Channel 04/0 : 54[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762819:763221 [4] NCCL INFO Channel 06/0 : 36[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127294:127687 [4] NCCL INFO Channel 15/0 : 60[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127297:127691 [7] NCCL INFO Channel 09/0 : 63[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104431:104831 [2] NCCL INFO Channel 00/0 : 50[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104433:104825 [4] NCCL INFO Channel 02/0 : 52[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127290:127685 [0] NCCL INFO Channel 11/0 : 56[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897247:897642 [7] NCCL INFO Channel 11/0 : 47[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127291:127690 [1] NCCL INFO Channel 11/0 : 57[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127292:127692 [2] NCCL INFO Channel 01/0 : 26[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897243:897637 [3] NCCL INFO Channel 09/0 : 43[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897241:897635 [1] NCCL INFO Channel 13/0 : 41[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897240:897640 [0] NCCL INFO Channel 13/0 : 40[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624944:625341 [1] NCCL INFO Channel 08/0 : 33[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762821:763224 [6] NCCL INFO Channel 10/0 : 38[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624945:625340 [2] NCCL INFO Channel 00/0 : 2[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104435:104826 [6] NCCL INFO Channel 08/0 : 54[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762817:763223 [2] NCCL INFO Channel 06/0 : 34[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762819:763221 [4] NCCL INFO Channel 08/0 : 36[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127293:127689 [3] NCCL INFO Channel 09/0 : 59[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127296:127688 [6] NCCL INFO Channel 05/0 : 30[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897247:897642 [7] NCCL INFO Channel 13/0 : 47[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127295:127686 [5] NCCL INFO Channel 03/0 : 29[5] -> 61[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897243:897637 [3] NCCL INFO Channel 13/0 : 43[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897240:897640 [0] NCCL INFO Channel 15/0 : 40[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897241:897635 [1] NCCL INFO Channel 15/0 : 41[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127294:127687 [4] NCCL INFO Channel 01/0 : 28[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127297:127691 [7] NCCL INFO Channel 11/0 : 63[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127290:127685 [0] NCCL INFO Channel 13/0 : 56[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624947:625339 [4] NCCL INFO Channel 02/0 : 4[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624949:625344 [6] NCCL INFO Channel 04/0 : 6[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127292:127692 [2] NCCL INFO Channel 03/0 : 26[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127291:127690 [1] NCCL INFO Channel 13/0 : 57[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104433:104825 [4] NCCL INFO Channel 06/0 : 52[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624946:625337 [3] NCCL INFO Channel 12/0 : 35[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624950:625342 [7] NCCL INFO Channel 12/0 : 39[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624944:625341 [1] NCCL INFO Channel 10/0 : 33[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624945:625340 [2] NCCL INFO Channel 04/0 : 2[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104431:104831 [2] NCCL INFO Channel 04/0 : 50[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104435:104826 [6] NCCL INFO Channel 10/0 : 54[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897243:897637 [3] NCCL INFO Channel 15/0 : 43[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762820:763219 [5] NCCL INFO Channel 00/0 : 37[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127296:127688 [6] NCCL INFO Channel 07/0 : 30[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762821:763224 [6] NCCL INFO Channel 12/0 : 38[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762815:763225 [0] NCCL INFO Channel 02/0 : 32[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127293:127689 [3] NCCL INFO Channel 13/0 : 59[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127295:127686 [5] NCCL INFO Channel 05/0 : 29[5] -> 61[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762819:763221 [4] NCCL INFO Channel 10/0 : 36[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127294:127687 [4] NCCL INFO Channel 03/0 : 28[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762817:763223 [2] NCCL INFO Channel 08/0 : 34[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127290:127685 [0] NCCL INFO Channel 15/0 : 56[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127297:127691 [7] NCCL INFO Channel 13/0 : 63[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104433:104825 [4] NCCL INFO Channel 08/0 : 52[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127291:127690 [1] NCCL INFO Channel 15/0 : 57[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104431:104831 [2] NCCL INFO Channel 06/0 : 50[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104435:104826 [6] NCCL INFO Channel 12/0 : 54[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624949:625344 [6] NCCL INFO Channel 08/0 : 6[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624947:625339 [4] NCCL INFO Channel 06/0 : 4[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127292:127692 [2] NCCL INFO Channel 05/0 : 26[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762820:763219 [5] NCCL INFO Channel 02/0 : 37[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762815:763225 [0] NCCL INFO Channel 04/0 : 32[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762817:763223 [2] NCCL INFO Channel 10/0 : 34[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104429:104829 [0] NCCL INFO Channel 02/0 : 48[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127296:127688 [6] NCCL INFO Channel 09/0 : 30[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104434:104828 [5] NCCL INFO Channel 00/0 : 53[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104433:104825 [4] NCCL INFO Channel 10/0 : 52[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624944:625341 [1] NCCL INFO Channel 12/0 : 33[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624948:625343 [5] NCCL INFO Channel 00/0 : 5[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624946:625337 [3] NCCL INFO Channel 14/0 : 35[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624950:625342 [7] NCCL INFO Channel 14/0 : 39[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624945:625340 [2] NCCL INFO Channel 06/0 : 2[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127293:127689 [3] NCCL INFO Channel 15/0 : 59[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127295:127686 [5] NCCL INFO Channel 07/0 : 29[5] -> 61[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104435:104826 [6] NCCL INFO Channel 14/0 : 54[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104431:104831 [2] NCCL INFO Channel 08/0 : 50[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624943:625338 [0] NCCL INFO Channel 02/0 : 0[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127294:127687 [4] NCCL INFO Channel 05/0 : 28[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127290:127685 [0] NCCL INFO Channel 01/0 : 24[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127297:127691 [7] NCCL INFO Channel 01/0 : 31[7] -> 63[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762820:763219 [5] NCCL INFO Channel 04/0 : 37[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762815:763225 [0] NCCL INFO Channel 06/0 : 32[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762817:763223 [2] NCCL INFO Channel 12/0 : 34[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127292:127692 [2] NCCL INFO Channel 07/0 : 26[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127291:127690 [1] NCCL INFO Channel 01/0 : 25[1] -> 57[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104429:104829 [0] NCCL INFO Channel 04/0 : 48[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104433:104825 [4] NCCL INFO Channel 12/0 : 52[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104434:104828 [5] NCCL INFO Channel 02/0 : 53[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104431:104831 [2] NCCL INFO Channel 10/0 : 50[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624949:625344 [6] NCCL INFO Channel 10/0 : 6[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624947:625339 [4] NCCL INFO Channel 08/0 : 4[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104429:104829 [0] NCCL INFO Channel 06/0 : 48[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624944:625341 [1] NCCL INFO Channel 14/0 : 33[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624948:625343 [5] NCCL INFO Channel 02/0 : 5[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624943:625338 [0] NCCL INFO Channel 04/0 : 0[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624945:625340 [2] NCCL INFO Channel 08/0 : 2[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127296:127688 [6] NCCL INFO Channel 11/0 : 30[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127293:127689 [3] NCCL INFO Channel 01/0 : 27[3] -> 59[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127295:127686 [5] NCCL INFO Channel 09/0 : 29[5] -> 61[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127294:127687 [4] NCCL INFO Channel 07/0 : 28[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127290:127685 [0] NCCL INFO Channel 03/0 : 24[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762815:763225 [0] NCCL INFO Channel 08/0 : 32[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127297:127691 [7] NCCL INFO Channel 03/0 : 31[7] -> 63[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127292:127692 [2] NCCL INFO Channel 09/0 : 26[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127291:127690 [1] NCCL INFO Channel 03/0 : 25[1] -> 57[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104433:104825 [4] NCCL INFO Channel 14/0 : 52[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104434:104828 [5] NCCL INFO Channel 04/0 : 53[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624949:625344 [6] NCCL INFO Channel 12/0 : 6[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624947:625339 [4] NCCL INFO Channel 10/0 : 4[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104431:104831 [2] NCCL INFO Channel 12/0 : 50[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624948:625343 [5] NCCL INFO Channel 04/0 : 5[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104429:104829 [0] NCCL INFO Channel 08/0 : 48[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624943:625338 [0] NCCL INFO Channel 06/0 : 0[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624945:625340 [2] NCCL INFO Channel 10/0 : 2[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762820:763219 [5] NCCL INFO Channel 06/0 : 37[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762817:763223 [2] NCCL INFO Channel 14/0 : 34[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127296:127688 [6] NCCL INFO Channel 13/0 : 30[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127293:127689 [3] NCCL INFO Channel 03/0 : 27[3] -> 59[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127294:127687 [4] NCCL INFO Channel 09/0 : 28[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104434:104828 [5] NCCL INFO Channel 06/0 : 53[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127295:127686 [5] NCCL INFO Channel 11/0 : 29[5] -> 61[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127290:127685 [0] NCCL INFO Channel 05/0 : 24[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104431:104831 [2] NCCL INFO Channel 14/0 : 50[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127292:127692 [2] NCCL INFO Channel 11/0 : 26[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127291:127690 [1] NCCL INFO Channel 05/0 : 25[1] -> 57[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127297:127691 [7] NCCL INFO Channel 05/0 : 31[7] -> 63[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104429:104829 [0] NCCL INFO Channel 10/0 : 48[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624949:625344 [6] NCCL INFO Channel 14/0 : 6[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624947:625339 [4] NCCL INFO Channel 12/0 : 4[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104434:104828 [5] NCCL INFO Channel 08/0 : 53[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762819:763221 [4] NCCL INFO Channel 12/0 : 36[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762815:763225 [0] NCCL INFO Channel 10/0 : 32[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624948:625343 [5] NCCL INFO Channel 06/0 : 5[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762822:763222 [7] NCCL INFO Channel 00/0 : 39[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624943:625338 [0] NCCL INFO Channel 08/0 : 0[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624945:625340 [2] NCCL INFO Channel 12/0 : 2[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762820:763219 [5] NCCL INFO Channel 08/0 : 37[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762818:763220 [3] NCCL INFO Channel 00/0 : 35[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104429:104829 [0] NCCL INFO Channel 12/0 : 48[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104434:104828 [5] NCCL INFO Channel 10/0 : 53[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624949:625344 [6] NCCL INFO Channel 00/0 : 38[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624947:625339 [4] NCCL INFO Channel 14/0 : 4[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762816:763218 [1] NCCL INFO Channel 00/0 : 33[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624948:625343 [5] NCCL INFO Channel 08/0 : 5[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762819:763221 [4] NCCL INFO Channel 14/0 : 36[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624943:625338 [0] NCCL INFO Channel 10/0 : 0[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624946:625337 [3] NCCL INFO Channel 00/0 : 3[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104429:104829 [0] NCCL INFO Channel 14/0 : 48[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104432:104827 [3] NCCL INFO Channel 00/0 : 51[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104436:104824 [7] NCCL INFO Channel 00/0 : 55[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104434:104828 [5] NCCL INFO Channel 12/0 : 53[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762815:763225 [0] NCCL INFO Channel 12/0 : 32[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762818:763220 [3] NCCL INFO Channel 02/0 : 35[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762820:763219 [5] NCCL INFO Channel 10/0 : 37[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762822:763222 [7] NCCL INFO Channel 02/0 : 39[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624950:625342 [7] NCCL INFO Channel 00/0 : 7[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762816:763218 [1] NCCL INFO Channel 02/0 : 33[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624945:625340 [2] NCCL INFO Channel 14/0 : 2[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624947:625339 [4] NCCL INFO Channel 00/0 : 36[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624949:625344 [6] NCCL INFO Channel 02/0 : 38[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624948:625343 [5] NCCL INFO Channel 10/0 : 5[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104430:104830 [1] NCCL INFO Channel 00/0 : 49[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624943:625338 [0] NCCL INFO Channel 12/0 : 0[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624944:625341 [1] NCCL INFO Channel 00/0 : 1[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624946:625337 [3] NCCL INFO Channel 02/0 : 3[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762821:763224 [6] NCCL INFO Channel 14/0 : 38[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762815:763225 [0] NCCL INFO Channel 14/0 : 32[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-167-9:1039543:1039933 [6] NCCL INFO Channel 00/0 : 62[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104434:104828 [5] NCCL INFO Channel 14/0 : 53[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104436:104824 [7] NCCL INFO Channel 02/0 : 55[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104432:104827 [3] NCCL INFO Channel 02/0 : 51[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762818:763220 [3] NCCL INFO Channel 04/0 : 35[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104430:104830 [1] NCCL INFO Channel 02/0 : 49[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762816:763218 [1] NCCL INFO Channel 04/0 : 33[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762820:763219 [5] NCCL INFO Channel 12/0 : 37[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762822:763222 [7] NCCL INFO Channel 04/0 : 39[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624950:625342 [7] NCCL INFO Channel 02/0 : 7[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624945:625340 [2] NCCL INFO Channel 00/0 : 34[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624949:625344 [6] NCCL INFO Channel 04/0 : 38[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624947:625339 [4] NCCL INFO Channel 02/0 : 36[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624948:625343 [5] NCCL INFO Channel 12/0 : 5[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-167-9:1039543:1039933 [6] NCCL INFO Channel 02/0 : 62[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104436:104824 [7] NCCL INFO Channel 04/0 : 55[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104432:104827 [3] NCCL INFO Channel 04/0 : 51[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762818:763220 [3] NCCL INFO Channel 06/0 : 35[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624943:625338 [0] NCCL INFO Channel 14/0 : 0[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104430:104830 [1] NCCL INFO Channel 04/0 : 49[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762816:763218 [1] NCCL INFO Channel 06/0 : 33[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762820:763219 [5] NCCL INFO Channel 14/0 : 37[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762822:763222 [7] NCCL INFO Channel 06/0 : 39[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-167-9:1039542:1039936 [5] NCCL INFO Channel 00/0 : 61[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624946:625337 [3] NCCL INFO Channel 04/0 : 3[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624950:625342 [7] NCCL INFO Channel 04/0 : 7[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624944:625341 [1] NCCL INFO Channel 02/0 : 1[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104436:104824 [7] NCCL INFO Channel 06/0 : 55[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104432:104827 [3] NCCL INFO Channel 06/0 : 51[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624949:625344 [6] NCCL INFO Channel 08/0 : 38[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624945:625340 [2] NCCL INFO Channel 04/0 : 34[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104430:104830 [1] NCCL INFO Channel 06/0 : 49[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762818:763220 [3] NCCL INFO Channel 08/0 : 35[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624947:625339 [4] NCCL INFO Channel 06/0 : 36[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624948:625343 [5] NCCL INFO Channel 14/0 : 5[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762816:763218 [1] NCCL INFO Channel 08/0 : 33[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624943:625338 [0] NCCL INFO Channel 02/0 : 32[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-167-9:1039543:1039933 [6] NCCL INFO Channel 04/0 : 62[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762822:763222 [7] NCCL INFO Channel 08/0 : 39[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624946:625337 [3] NCCL INFO Channel 06/0 : 3[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624950:625342 [7] NCCL INFO Channel 06/0 : 7[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-167-9:1039539:1039938 [2] NCCL INFO Channel 00/0 : 58[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-167-9:1039542:1039936 [5] NCCL INFO Channel 02/0 : 61[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762818:763220 [3] NCCL INFO Channel 10/0 : 35[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624949:625344 [6] NCCL INFO Channel 10/0 : 38[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104436:104824 [7] NCCL INFO Channel 08/0 : 55[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104432:104827 [3] NCCL INFO Channel 08/0 : 51[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104430:104830 [1] NCCL INFO Channel 08/0 : 49[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-167-9:1039541:1039937 [4] NCCL INFO Channel 00/0 : 60[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762816:763218 [1] NCCL INFO Channel 10/0 : 33[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-167-9:1039543:1039933 [6] NCCL INFO Channel 08/0 : 62[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762822:763222 [7] NCCL INFO Channel 10/0 : 39[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624947:625339 [4] NCCL INFO Channel 08/0 : 36[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762818:763220 [3] NCCL INFO Channel 12/0 : 35[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624945:625340 [2] NCCL INFO Channel 06/0 : 34[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624948:625343 [5] NCCL INFO Channel 00/0 : 37[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104436:104824 [7] NCCL INFO Channel 10/0 : 55[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762816:763218 [1] NCCL INFO Channel 12/0 : 33[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624943:625338 [0] NCCL INFO Channel 04/0 : 32[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104432:104827 [3] NCCL INFO Channel 10/0 : 51[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104430:104830 [1] NCCL INFO Channel 10/0 : 49[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624944:625341 [1] NCCL INFO Channel 04/0 : 1[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624949:625344 [6] NCCL INFO Channel 12/0 : 38[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624950:625342 [7] NCCL INFO Channel 08/0 : 7[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624946:625337 [3] NCCL INFO Channel 08/0 : 3[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-167-9:1039542:1039936 [5] NCCL INFO Channel 04/0 : 61[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-167-9:1039539:1039938 [2] NCCL INFO Channel 04/0 : 58[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624947:625339 [4] NCCL INFO Channel 10/0 : 36[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-167-9:1039541:1039937 [4] NCCL INFO Channel 02/0 : 60[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762822:763222 [7] NCCL INFO Channel 12/0 : 39[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-167-9:1039543:1039933 [6] NCCL INFO Channel 10/0 : 62[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104436:104824 [7] NCCL INFO Channel 12/0 : 55[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762818:763220 [3] NCCL INFO Channel 14/0 : 35[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762816:763218 [1] NCCL INFO Channel 14/0 : 33[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762822:763222 [7] NCCL INFO Channel 14/0 : 39[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624948:625343 [5] NCCL INFO Channel 02/0 : 37[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624945:625340 [2] NCCL INFO Channel 08/0 : 34[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624943:625338 [0] NCCL INFO Channel 06/0 : 32[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624949:625344 [6] NCCL INFO Channel 14/0 : 38[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624944:625341 [1] NCCL INFO Channel 06/0 : 1[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-167-9:1039542:1039936 [5] NCCL INFO Channel 06/0 : 61[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104432:104827 [3] NCCL INFO Channel 12/0 : 51[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104430:104830 [1] NCCL INFO Channel 12/0 : 49[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624947:625339 [4] NCCL INFO Channel 12/0 : 36[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624950:625342 [7] NCCL INFO Channel 10/0 : 7[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624946:625337 [3] NCCL INFO Channel 10/0 : 3[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-167-9:1039539:1039938 [2] NCCL INFO Channel 06/0 : 58[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104436:104824 [7] NCCL INFO Channel 14/0 : 55[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624948:625343 [5] NCCL INFO Channel 04/0 : 37[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-167-9:1039541:1039937 [4] NCCL INFO Channel 06/0 : 60[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-167-9:1039543:1039933 [6] NCCL INFO Channel 12/0 : 62[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-167-9:1039538:1039934 [1] NCCL INFO Channel 00/0 : 57[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104430:104830 [1] NCCL INFO Channel 14/0 : 49[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104432:104827 [3] NCCL INFO Channel 14/0 : 51[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-167-9:1039540:1039935 [3] NCCL INFO Channel 00/0 : 59[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-167-9:1039544:1039931 [7] NCCL INFO Channel 00/0 : 63[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-167-9:1039542:1039936 [5] NCCL INFO Channel 08/0 : 61[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-167-9:1039540:1039935 [3] NCCL INFO Channel 02/0 : 59[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-167-9:1039544:1039931 [7] NCCL INFO Channel 02/0 : 63[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-167-9:1039542:1039936 [5] NCCL INFO Channel 10/0 : 61[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-46:702886:703281 [6] NCCL INFO Channel 01/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-167-9:1039537:1039932 [0] NCCL INFO Channel 02/0 : 56[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-46:702886:703281 [6] NCCL INFO Channel 03/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-167-9:1039540:1039935 [3] NCCL INFO Channel 04/0 : 59[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-167-9:1039544:1039931 [7] NCCL INFO Channel 04/0 : 63[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-167-9:1039542:1039936 [5] NCCL INFO Channel 12/0 : 61[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-46:702886:703281 [6] NCCL INFO Channel 05/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-46:702884:703279 [4] NCCL INFO Channel 01/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-167-9:1039537:1039932 [0] NCCL INFO Channel 04/0 : 56[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-167-9:1039542:1039936 [5] NCCL INFO Channel 14/0 : 61[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-167-9:1039540:1039935 [3] NCCL INFO Channel 06/0 : 59[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-167-9:1039544:1039931 [7] NCCL INFO Channel 06/0 : 63[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-167-9:1039537:1039932 [0] NCCL INFO Channel 06/0 : 56[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-46:702884:703279 [4] NCCL INFO Channel 03/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-46:702886:703281 [6] NCCL INFO Channel 07/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127296:127688 [6] NCCL INFO Channel 15/0 : 30[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127293:127689 [3] NCCL INFO Channel 05/0 : 27[3] -> 59[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127294:127687 [4] NCCL INFO Channel 11/0 : 28[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127290:127685 [0] NCCL INFO Channel 07/0 : 24[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127295:127686 [5] NCCL INFO Channel 15/0 : 29[5] -> 61[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-167-9:1039540:1039935 [3] NCCL INFO Channel 08/0 : 59[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127292:127692 [2] NCCL INFO Channel 13/0 : 26[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127291:127690 [1] NCCL INFO Channel 07/0 : 25[1] -> 57[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127297:127691 [7] NCCL INFO Channel 07/0 : 31[7] -> 63[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-167-9:1039544:1039931 [7] NCCL INFO Channel 08/0 : 63[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127293:127689 [3] NCCL INFO Channel 07/0 : 27[3] -> 59[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-167-9:1039537:1039932 [0] NCCL INFO Channel 08/0 : 56[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127294:127687 [4] NCCL INFO Channel 13/0 : 28[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127290:127685 [0] NCCL INFO Channel 09/0 : 24[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-46:702886:703281 [6] NCCL INFO Channel 09/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-46:702884:703279 [4] NCCL INFO Channel 05/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127292:127692 [2] NCCL INFO Channel 15/0 : 26[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-46:702882:703284 [2] NCCL INFO Channel 01/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127291:127690 [1] NCCL INFO Channel 11/0 : 25[1] -> 57[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127297:127691 [7] NCCL INFO Channel 09/0 : 31[7] -> 63[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127293:127689 [3] NCCL INFO Channel 09/0 : 27[3] -> 59[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-167-9:1039540:1039935 [3] NCCL INFO Channel 10/0 : 59[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127294:127687 [4] NCCL INFO Channel 15/0 : 28[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-167-9:1039544:1039931 [7] NCCL INFO Channel 10/0 : 63[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-46:702880:703277 [0] NCCL INFO Channel 01/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127290:127685 [0] NCCL INFO Channel 11/0 : 24[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127296:127688 [6] NCCL INFO Channel 01/0 : 46[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-46:702886:703281 [6] NCCL INFO Channel 11/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127291:127690 [1] NCCL INFO Channel 13/0 : 25[1] -> 57[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-46:702884:703279 [4] NCCL INFO Channel 07/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127297:127691 [7] NCCL INFO Channel 11/0 : 31[7] -> 63[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127293:127689 [3] NCCL INFO Channel 13/0 : 27[3] -> 59[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-167-9:1039537:1039932 [0] NCCL INFO Channel 10/0 : 56[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127295:127686 [5] NCCL INFO Channel 01/0 : 45[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127296:127688 [6] NCCL INFO Channel 03/0 : 46[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-167-9:1039540:1039935 [3] NCCL INFO Channel 12/0 : 59[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127290:127685 [0] NCCL INFO Channel 13/0 : 24[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-167-9:1039544:1039931 [7] NCCL INFO Channel 12/0 : 63[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127291:127690 [1] NCCL INFO Channel 15/0 : 25[1] -> 57[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127297:127691 [7] NCCL INFO Channel 13/0 : 31[7] -> 63[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-46:702880:703277 [0] NCCL INFO Channel 03/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127293:127689 [3] NCCL INFO Channel 15/0 : 27[3] -> 59[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-46:702886:703281 [6] NCCL INFO Channel 13/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127292:127692 [2] NCCL INFO Channel 01/0 : 42[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127296:127688 [6] NCCL INFO Channel 05/0 : 46[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-167-9:1039537:1039932 [0] NCCL INFO Channel 12/0 : 56[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-46:702882:703284 [2] NCCL INFO Channel 03/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127295:127686 [5] NCCL INFO Channel 03/0 : 45[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-167-9:1039540:1039935 [3] NCCL INFO Channel 14/0 : 59[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-46:702884:703279 [4] NCCL INFO Channel 09/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127294:127687 [4] NCCL INFO Channel 01/0 : 44[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127290:127685 [0] NCCL INFO Channel 15/0 : 24[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-167-9:1039544:1039931 [7] NCCL INFO Channel 14/0 : 63[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127292:127692 [2] NCCL INFO Channel 03/0 : 42[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127296:127688 [6] NCCL INFO Channel 07/0 : 46[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127295:127686 [5] NCCL INFO Channel 05/0 : 45[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127294:127687 [4] NCCL INFO Channel 03/0 : 44[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127292:127692 [2] NCCL INFO Channel 05/0 : 42[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127291:127690 [1] NCCL INFO Channel 01/0 : 41[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127296:127688 [6] NCCL INFO Channel 09/0 : 46[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127295:127686 [5] NCCL INFO Channel 07/0 : 45[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127294:127687 [4] NCCL INFO Channel 05/0 : 44[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-167-9:1039537:1039932 [0] NCCL INFO Channel 14/0 : 56[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127297:127691 [7] NCCL INFO Channel 01/0 : 47[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-46:702886:703281 [6] NCCL INFO Channel 15/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-46:702880:703277 [0] NCCL INFO Channel 05/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127293:127689 [3] NCCL INFO Channel 01/0 : 43[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127290:127685 [0] NCCL INFO Channel 01/0 : 40[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127291:127690 [1] NCCL INFO Channel 03/0 : 41[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-46:702884:703279 [4] NCCL INFO Channel 11/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127292:127692 [2] NCCL INFO Channel 07/0 : 42[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-46:702885:703278 [5] NCCL INFO Channel 01/0 : 13[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-46:702882:703284 [2] NCCL INFO Channel 05/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127296:127688 [6] NCCL INFO Channel 11/0 : 46[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127295:127686 [5] NCCL INFO Channel 09/0 : 45[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127294:127687 [4] NCCL INFO Channel 07/0 : 44[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127297:127691 [7] NCCL INFO Channel 03/0 : 47[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-167-9:1039543:1039933 [6] NCCL INFO Channel 14/0 : 62[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127293:127689 [3] NCCL INFO Channel 03/0 : 43[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127290:127685 [0] NCCL INFO Channel 03/0 : 40[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127291:127690 [1] NCCL INFO Channel 05/0 : 41[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-46:702880:703277 [0] NCCL INFO Channel 07/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127292:127692 [2] NCCL INFO Channel 09/0 : 42[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127296:127688 [6] NCCL INFO Channel 13/0 : 46[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127295:127686 [5] NCCL INFO Channel 11/0 : 45[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127294:127687 [4] NCCL INFO Channel 09/0 : 44[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127297:127691 [7] NCCL INFO Channel 05/0 : 47[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127293:127689 [3] NCCL INFO Channel 05/0 : 43[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-46:702884:703279 [4] NCCL INFO Channel 13/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127290:127685 [0] NCCL INFO Channel 05/0 : 40[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-167-9:1039538:1039934 [1] NCCL INFO Channel 02/0 : 57[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127291:127690 [1] NCCL INFO Channel 07/0 : 41[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-46:702882:703284 [2] NCCL INFO Channel 07/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-46:702885:703278 [5] NCCL INFO Channel 03/0 : 13[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127292:127692 [2] NCCL INFO Channel 11/0 : 42[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127296:127688 [6] NCCL INFO Channel 15/0 : 46[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-46:702880:703277 [0] NCCL INFO Channel 09/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127294:127687 [4] NCCL INFO Channel 11/0 : 44[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127295:127686 [5] NCCL INFO Channel 15/0 : 45[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127293:127689 [3] NCCL INFO Channel 07/0 : 43[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127297:127691 [7] NCCL INFO Channel 07/0 : 47[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-167-9:1039538:1039934 [1] NCCL INFO Channel 04/0 : 57[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-46:702884:703279 [4] NCCL INFO Channel 15/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127290:127685 [0] NCCL INFO Channel 07/0 : 40[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-46:702882:703284 [2] NCCL INFO Channel 09/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127291:127690 [1] NCCL INFO Channel 11/0 : 41[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127292:127692 [2] NCCL INFO Channel 13/0 : 42[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-46:702885:703278 [5] NCCL INFO Channel 05/0 : 13[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127296:127688 [6] NCCL INFO Channel 01/0 : 30[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-46:702880:703277 [0] NCCL INFO Channel 11/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127295:127686 [5] NCCL INFO Channel 01/0 : 29[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127294:127687 [4] NCCL INFO Channel 13/0 : 44[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-167-9:1039538:1039934 [1] NCCL INFO Channel 06/0 : 57[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127293:127689 [3] NCCL INFO Channel 09/0 : 43[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127297:127691 [7] NCCL INFO Channel 09/0 : 47[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127290:127685 [0] NCCL INFO Channel 09/0 : 40[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-46:702882:703284 [2] NCCL INFO Channel 11/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127292:127692 [2] NCCL INFO Channel 15/0 : 42[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127291:127690 [1] NCCL INFO Channel 13/0 : 41[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127296:127688 [6] NCCL INFO Channel 03/0 : 30[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-167-9:1039538:1039934 [1] NCCL INFO Channel 08/0 : 57[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127295:127686 [5] NCCL INFO Channel 03/0 : 29[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-167-9:1039539:1039938 [2] NCCL INFO Channel 08/0 : 58[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127294:127687 [4] NCCL INFO Channel 15/0 : 44[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-46:702880:703277 [0] NCCL INFO Channel 13/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-46:702885:703278 [5] NCCL INFO Channel 07/0 : 13[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127293:127689 [3] NCCL INFO Channel 13/0 : 43[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624945:625340 [2] NCCL INFO Channel 10/0 : 34[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624943:625338 [0] NCCL INFO Channel 08/0 : 32[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624944:625341 [1] NCCL INFO Channel 08/0 : 1[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-46:702882:703284 [2] NCCL INFO Channel 13/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624947:625339 [4] NCCL INFO Channel 14/0 : 36[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624950:625342 [7] NCCL INFO Channel 12/0 : 7[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-167-9:1039538:1039934 [1] NCCL INFO Channel 10/0 : 57[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624946:625337 [3] NCCL INFO Channel 12/0 : 3[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127297:127691 [7] NCCL INFO Channel 11/0 : 47[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127290:127685 [0] NCCL INFO Channel 11/0 : 40[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624948:625343 [5] NCCL INFO Channel 06/0 : 37[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127292:127692 [2] NCCL INFO Channel 01/0 : 26[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-46:702880:703277 [0] NCCL INFO Channel 15/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624943:625338 [0] NCCL INFO Channel 10/0 : 32[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624945:625340 [2] NCCL INFO Channel 12/0 : 34[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624944:625341 [1] NCCL INFO Channel 10/0 : 1[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127291:127690 [1] NCCL INFO Channel 15/0 : 41[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-46:702885:703278 [5] NCCL INFO Channel 09/0 : 13[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624949:625344 [6] NCCL INFO Channel 00/0 : 54[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127296:127688 [6] NCCL INFO Channel 05/0 : 30[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127295:127686 [5] NCCL INFO Channel 05/0 : 29[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624950:625342 [7] NCCL INFO Channel 14/0 : 7[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127294:127687 [4] NCCL INFO Channel 01/0 : 28[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-167-9:1039539:1039938 [2] NCCL INFO Channel 10/0 : 58[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624946:625337 [3] NCCL INFO Channel 14/0 : 3[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127297:127691 [7] NCCL INFO Channel 13/0 : 47[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127290:127685 [0] NCCL INFO Channel 13/0 : 40[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127293:127689 [3] NCCL INFO Channel 15/0 : 43[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127292:127692 [2] NCCL INFO Channel 03/0 : 26[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-167-9:1039538:1039934 [1] NCCL INFO Channel 12/0 : 57[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-46:702882:703284 [2] NCCL INFO Channel 15/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897245:897638 [5] NCCL INFO Channel 00/0 : 53[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624948:625343 [5] NCCL INFO Channel 08/0 : 37[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127291:127690 [1] NCCL INFO Channel 01/0 : 25[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-167-9:1039539:1039938 [2] NCCL INFO Channel 12/0 : 58[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-46:702887:703282 [7] NCCL INFO Channel 01/0 : 15[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624943:625338 [0] NCCL INFO Channel 12/0 : 32[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624944:625341 [1] NCCL INFO Channel 12/0 : 1[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127295:127686 [5] NCCL INFO Channel 07/0 : 29[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127296:127688 [6] NCCL INFO Channel 07/0 : 30[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127294:127687 [4] NCCL INFO Channel 03/0 : 28[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-46:702885:703278 [5] NCCL INFO Channel 11/0 : 13[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624945:625340 [2] NCCL INFO Channel 14/0 : 34[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127297:127691 [7] NCCL INFO Channel 01/0 : 31[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-46:702883:703283 [3] NCCL INFO Channel 01/0 : 11[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624949:625344 [6] NCCL INFO Channel 02/0 : 54[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127292:127692 [2] NCCL INFO Channel 05/0 : 26[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127293:127689 [3] NCCL INFO Channel 01/0 : 27[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127290:127685 [0] NCCL INFO Channel 15/0 : 40[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897245:897638 [5] NCCL INFO Channel 01/0 : 53[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624950:625342 [7] NCCL INFO Channel 00/0 : 39[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127291:127690 [1] NCCL INFO Channel 03/0 : 25[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-167-9:1039538:1039934 [1] NCCL INFO Channel 14/0 : 57[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-46:702887:703282 [7] NCCL INFO Channel 03/0 : 15[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624946:625337 [3] NCCL INFO Channel 00/0 : 35[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127296:127688 [6] NCCL INFO Channel 09/0 : 30[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127294:127687 [4] NCCL INFO Channel 05/0 : 28[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127295:127686 [5] NCCL INFO Channel 09/0 : 29[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-167-9:1039541:1039937 [4] NCCL INFO Channel 08/0 : 60[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-46:702881:703280 [1] NCCL INFO Channel 01/0 : 9[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624948:625343 [5] NCCL INFO Channel 10/0 : 37[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127297:127691 [7] NCCL INFO Channel 03/0 : 31[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-167-9:1039539:1039938 [2] NCCL INFO Channel 14/0 : 58[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-46:702885:703278 [5] NCCL INFO Channel 15/0 : 13[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897246:897641 [6] NCCL INFO Channel 00/0 : 54[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624944:625341 [1] NCCL INFO Channel 14/0 : 1[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624943:625338 [0] NCCL INFO Channel 14/0 : 32[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127293:127689 [3] NCCL INFO Channel 03/0 : 27[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-46:702883:703283 [3] NCCL INFO Channel 03/0 : 11[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897245:897638 [5] NCCL INFO Channel 02/0 : 53[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9630:10024 [5] NCCL INFO Channel 00/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624947:625339 [4] NCCL INFO Channel 00/0 : 52[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127290:127685 [0] NCCL INFO Channel 01/0 : 24[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127292:127692 [2] NCCL INFO Channel 07/0 : 26[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-46:702887:703282 [7] NCCL INFO Channel 05/0 : 15[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624949:625344 [6] NCCL INFO Channel 04/0 : 54[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127291:127690 [1] NCCL INFO Channel 05/0 : 25[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-46:702881:703280 [1] NCCL INFO Channel 03/0 : 9[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624950:625342 [7] NCCL INFO Channel 02/0 : 39[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127296:127688 [6] NCCL INFO Channel 11/0 : 30[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127295:127686 [5] NCCL INFO Channel 11/0 : 29[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127294:127687 [4] NCCL INFO Channel 07/0 : 28[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-167-9:1039541:1039937 [4] NCCL INFO Channel 10/0 : 60[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897246:897641 [6] NCCL INFO Channel 01/0 : 54[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624948:625343 [5] NCCL INFO Channel 12/0 : 37[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624946:625337 [3] NCCL INFO Channel 02/0 : 35[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127297:127691 [7] NCCL INFO Channel 05/0 : 31[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-167-9:1039541:1039937 [4] NCCL INFO Channel 12/0 : 60[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-46:702883:703283 [3] NCCL INFO Channel 05/0 : 11[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-46:702887:703282 [7] NCCL INFO Channel 07/0 : 15[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897245:897638 [5] NCCL INFO Channel 03/0 : 53[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9630:10024 [5] NCCL INFO Channel 01/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624944:625341 [1] NCCL INFO Channel 00/0 : 33[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127293:127689 [3] NCCL INFO Channel 05/0 : 27[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-46:702881:703280 [1] NCCL INFO Channel 05/0 : 9[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624947:625339 [4] NCCL INFO Channel 02/0 : 52[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127290:127685 [0] NCCL INFO Channel 03/0 : 24[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127292:127692 [2] NCCL INFO Channel 09/0 : 26[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624949:625344 [6] NCCL INFO Channel 08/0 : 54[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127291:127690 [1] NCCL INFO Channel 07/0 : 25[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-167-9:1039541:1039937 [4] NCCL INFO Channel 14/0 : 60[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624950:625342 [7] NCCL INFO Channel 04/0 : 39[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127295:127686 [5] NCCL INFO Channel 15/0 : 29[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897246:897641 [6] NCCL INFO Channel 02/0 : 54[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9631:10022 [6] NCCL INFO Channel 00/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624948:625343 [5] NCCL INFO Channel 14/0 : 37[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624946:625337 [3] NCCL INFO Channel 04/0 : 35[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127296:127688 [6] NCCL INFO Channel 13/0 : 30[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-46:702887:703282 [7] NCCL INFO Channel 09/0 : 15[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897245:897638 [5] NCCL INFO Channel 04/0 : 53[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9630:10024 [5] NCCL INFO Channel 02/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624944:625341 [1] NCCL INFO Channel 02/0 : 33[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127294:127687 [4] NCCL INFO Channel 09/0 : 28[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-46:702883:703283 [3] NCCL INFO Channel 07/0 : 11[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-46:702881:703280 [1] NCCL INFO Channel 07/0 : 9[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624947:625339 [4] NCCL INFO Channel 06/0 : 52[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127293:127689 [3] NCCL INFO Channel 07/0 : 27[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127297:127691 [7] NCCL INFO Channel 07/0 : 31[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624949:625344 [6] NCCL INFO Channel 10/0 : 54[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624945:625340 [2] NCCL INFO Channel 00/0 : 50[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127292:127692 [2] NCCL INFO Channel 11/0 : 26[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127290:127685 [0] NCCL INFO Channel 05/0 : 24[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127291:127690 [1] NCCL INFO Channel 11/0 : 25[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624943:625338 [0] NCCL INFO Channel 02/0 : 48[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127296:127688 [6] NCCL INFO Channel 15/0 : 30[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624950:625342 [7] NCCL INFO Channel 06/0 : 39[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127294:127687 [4] NCCL INFO Channel 11/0 : 28[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897241:897635 [1] NCCL INFO Channel 00/0 : 49[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9631:10022 [6] NCCL INFO Channel 01/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624946:625337 [3] NCCL INFO Channel 06/0 : 35[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624944:625341 [1] NCCL INFO Channel 04/0 : 33[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127293:127689 [3] NCCL INFO Channel 09/0 : 27[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-46:702887:703282 [7] NCCL INFO Channel 11/0 : 15[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9630:10024 [5] NCCL INFO Channel 03/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624947:625339 [4] NCCL INFO Channel 08/0 : 52[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624949:625344 [6] NCCL INFO Channel 12/0 : 54[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127297:127691 [7] NCCL INFO Channel 09/0 : 31[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624945:625340 [2] NCCL INFO Channel 04/0 : 50[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624950:625342 [7] NCCL INFO Channel 08/0 : 39[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-46:702881:703280 [1] NCCL INFO Channel 11/0 : 9[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624943:625338 [0] NCCL INFO Channel 04/0 : 48[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624946:625337 [3] NCCL INFO Channel 08/0 : 35[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624944:625341 [1] NCCL INFO Channel 06/0 : 33[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-46:702883:703283 [3] NCCL INFO Channel 09/0 : 11[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897246:897641 [6] NCCL INFO Channel 03/0 : 54[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897244:897636 [4] NCCL INFO Channel 00/0 : 52[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624947:625339 [4] NCCL INFO Channel 10/0 : 52[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-46:702887:703282 [7] NCCL INFO Channel 13/0 : 15[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897245:897638 [5] NCCL INFO Channel 06/0 : 53[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9631:10022 [6] NCCL INFO Channel 02/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624949:625344 [6] NCCL INFO Channel 14/0 : 54[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127291:127690 [1] NCCL INFO Channel 13/0 : 25[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127292:127692 [2] NCCL INFO Channel 13/0 : 26[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127290:127685 [0] NCCL INFO Channel 07/0 : 24[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127294:127687 [4] NCCL INFO Channel 13/0 : 28[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9626:10025 [1] NCCL INFO Channel 00/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624948:625343 [5] NCCL INFO Channel 00/0 : 53[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127293:127689 [3] NCCL INFO Channel 13/0 : 27[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127297:127691 [7] NCCL INFO Channel 11/0 : 31[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624950:625342 [7] NCCL INFO Channel 10/0 : 39[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127291:127690 [1] NCCL INFO Channel 15/0 : 25[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624943:625338 [0] NCCL INFO Channel 06/0 : 48[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624944:625341 [1] NCCL INFO Channel 08/0 : 33[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624945:625340 [2] NCCL INFO Channel 06/0 : 50[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127290:127685 [0] NCCL INFO Channel 09/0 : 24[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-46:702881:703280 [1] NCCL INFO Channel 13/0 : 9[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-46:702883:703283 [3] NCCL INFO Channel 13/0 : 11[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624946:625337 [3] NCCL INFO Channel 10/0 : 35[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127294:127687 [4] NCCL INFO Channel 15/0 : 28[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897247:897642 [7] NCCL INFO Channel 00/0 : 55[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624947:625339 [4] NCCL INFO Channel 12/0 : 52[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624949:625344 [6] NCCL INFO Channel 00/0 : 38[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127293:127689 [3] NCCL INFO Channel 15/0 : 27[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127297:127691 [7] NCCL INFO Channel 13/0 : 31[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897241:897635 [1] NCCL INFO Channel 02/0 : 49[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9630:10024 [5] NCCL INFO Channel 04/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624948:625343 [5] NCCL INFO Channel 02/0 : 53[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127292:127692 [2] NCCL INFO Channel 15/0 : 26[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897243:897637 [3] NCCL INFO Channel 00/0 : 51[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9627:10027 [2] NCCL INFO Channel 00/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9631:10022 [6] NCCL INFO Channel 03/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624950:625342 [7] NCCL INFO Channel 12/0 : 39[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127295:127686 [5] NCCL INFO Channel 00/0 : 29[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127290:127685 [0] NCCL INFO Channel 11/0 : 24[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897246:897641 [6] NCCL INFO Channel 04/0 : 54[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897242:897639 [2] NCCL INFO Channel 00/0 : 50[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9629:10021 [4] NCCL INFO Channel 00/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624943:625338 [0] NCCL INFO Channel 08/0 : 48[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127295:127686 [5] NCCL INFO Channel 02/0 : 29[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897244:897636 [4] NCCL INFO Channel 01/0 : 52[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897245:897638 [5] NCCL INFO Channel 07/0 : 53[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9628:10023 [3] NCCL INFO Channel 00/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624944:625341 [1] NCCL INFO Channel 10/0 : 33[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624945:625340 [2] NCCL INFO Channel 08/0 : 50[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127296:127688 [6] NCCL INFO Channel 00/0 : 30[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9626:10025 [1] NCCL INFO Channel 02/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624946:625337 [3] NCCL INFO Channel 12/0 : 35[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127290:127685 [0] NCCL INFO Channel 13/0 : 24[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-46:702883:703283 [3] NCCL INFO Channel 15/0 : 11[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-46:702881:703280 [1] NCCL INFO Channel 15/0 : 9[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104435:104826 [6] NCCL INFO Channel 00/0 : 62[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624947:625339 [4] NCCL INFO Channel 14/0 : 52[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624949:625344 [6] NCCL INFO Channel 02/0 : 38[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624950:625342 [7] NCCL INFO Channel 14/0 : 39[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127295:127686 [5] NCCL INFO Channel 04/0 : 29[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897247:897642 [7] NCCL INFO Channel 01/0 : 55[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624948:625343 [5] NCCL INFO Channel 04/0 : 53[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624943:625338 [0] NCCL INFO Channel 10/0 : 48[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127296:127688 [6] NCCL INFO Channel 02/0 : 30[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897241:897635 [1] NCCL INFO Channel 03/0 : 49[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624944:625341 [1] NCCL INFO Channel 12/0 : 33[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127290:127685 [0] NCCL INFO Channel 15/0 : 24[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897243:897637 [3] NCCL INFO Channel 01/0 : 51[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9632:10026 [7] NCCL INFO Channel 00/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624945:625340 [2] NCCL INFO Channel 10/0 : 50[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624946:625337 [3] NCCL INFO Channel 14/0 : 35[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127295:127686 [5] NCCL INFO Channel 06/0 : 29[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104435:104826 [6] NCCL INFO Channel 02/0 : 62[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897246:897641 [6] NCCL INFO Channel 05/0 : 54[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9630:10024 [5] NCCL INFO Channel 06/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624949:625344 [6] NCCL INFO Channel 04/0 : 38[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624947:625339 [4] NCCL INFO Channel 00/0 : 36[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624948:625343 [5] NCCL INFO Channel 06/0 : 53[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127296:127688 [6] NCCL INFO Channel 04/0 : 30[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9627:10027 [2] NCCL INFO Channel 01/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624943:625338 [0] NCCL INFO Channel 12/0 : 48[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127291:127690 [1] NCCL INFO Channel 00/0 : 25[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127295:127686 [5] NCCL INFO Channel 08/0 : 29[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9631:10022 [6] NCCL INFO Channel 04/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624944:625341 [1] NCCL INFO Channel 14/0 : 33[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762821:763224 [6] NCCL INFO Channel 00/0 : 30[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624945:625340 [2] NCCL INFO Channel 12/0 : 50[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104435:104826 [6] NCCL INFO Channel 04/0 : 62[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897244:897636 [4] NCCL INFO Channel 02/0 : 52[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897245:897638 [5] NCCL INFO Channel 08/0 : 53[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897242:897639 [2] NCCL INFO Channel 01/0 : 50[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127294:127687 [4] NCCL INFO Channel 00/0 : 28[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897240:897640 [0] NCCL INFO Channel 01/0 : 48[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897247:897642 [7] NCCL INFO Channel 02/0 : 55[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897241:897635 [1] NCCL INFO Channel 04/0 : 49[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897243:897637 [3] NCCL INFO Channel 02/0 : 51[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9629:10021 [4] NCCL INFO Channel 01/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127293:127689 [3] NCCL INFO Channel 00/0 : 27[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127296:127688 [6] NCCL INFO Channel 08/0 : 30[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127292:127692 [2] NCCL INFO Channel 00/0 : 26[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897246:897641 [6] NCCL INFO Channel 07/0 : 54[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9626:10025 [1] NCCL INFO Channel 03/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127297:127691 [7] NCCL INFO Channel 00/0 : 31[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762821:763224 [6] NCCL INFO Channel 02/0 : 30[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9630:10024 [5] NCCL INFO Channel 07/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9628:10023 [3] NCCL INFO Channel 01/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9632:10026 [7] NCCL INFO Channel 01/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624947:625339 [4] NCCL INFO Channel 02/0 : 36[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127291:127690 [1] NCCL INFO Channel 02/0 : 25[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104433:104825 [4] NCCL INFO Channel 00/0 : 60[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9627:10027 [2] NCCL INFO Channel 03/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624949:625344 [6] NCCL INFO Channel 08/0 : 38[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127295:127686 [5] NCCL INFO Channel 10/0 : 29[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104435:104826 [6] NCCL INFO Channel 08/0 : 62[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9631:10022 [6] NCCL INFO Channel 05/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624948:625343 [5] NCCL INFO Channel 08/0 : 53[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624943:625338 [0] NCCL INFO Channel 14/0 : 48[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127294:127687 [4] NCCL INFO Channel 02/0 : 28[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897245:897638 [5] NCCL INFO Channel 09/0 : 53[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9625:10020 [0] NCCL INFO Channel 01/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624950:625342 [7] NCCL INFO Channel 00/0 : 55[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624945:625340 [2] NCCL INFO Channel 14/0 : 50[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127293:127689 [3] NCCL INFO Channel 02/0 : 27[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897244:897636 [4] NCCL INFO Channel 03/0 : 52[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624946:625337 [3] NCCL INFO Channel 00/0 : 51[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127296:127688 [6] NCCL INFO Channel 10/0 : 30[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127292:127692 [2] NCCL INFO Channel 04/0 : 26[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762821:763224 [6] NCCL INFO Channel 04/0 : 30[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897242:897639 [2] NCCL INFO Channel 03/0 : 50[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624947:625339 [4] NCCL INFO Channel 06/0 : 36[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127297:127691 [7] NCCL INFO Channel 02/0 : 31[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897247:897642 [7] NCCL INFO Channel 03/0 : 55[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897241:897635 [1] NCCL INFO Channel 05/0 : 49[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897240:897640 [0] NCCL INFO Channel 02/0 : 48[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897246:897641 [6] NCCL INFO Channel 08/0 : 54[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624949:625344 [6] NCCL INFO Channel 10/0 : 38[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127291:127690 [1] NCCL INFO Channel 04/0 : 25[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127295:127686 [5] NCCL INFO Channel 12/0 : 29[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127294:127687 [4] NCCL INFO Channel 06/0 : 28[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104433:104825 [4] NCCL INFO Channel 02/0 : 60[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897243:897637 [3] NCCL INFO Channel 04/0 : 51[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624948:625343 [5] NCCL INFO Channel 10/0 : 53[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127290:127685 [0] NCCL INFO Channel 02/0 : 24[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104435:104826 [6] NCCL INFO Channel 10/0 : 62[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9629:10021 [4] NCCL INFO Channel 02/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624943:625338 [0] NCCL INFO Channel 02/0 : 32[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624944:625341 [1] NCCL INFO Channel 00/0 : 49[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127293:127689 [3] NCCL INFO Channel 04/0 : 27[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762819:763221 [4] NCCL INFO Channel 00/0 : 28[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9628:10023 [3] NCCL INFO Channel 02/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624945:625340 [2] NCCL INFO Channel 00/0 : 34[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127292:127692 [2] NCCL INFO Channel 06/0 : 26[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127296:127688 [6] NCCL INFO Channel 12/0 : 30[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9630:10024 [5] NCCL INFO Channel 08/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624950:625342 [7] NCCL INFO Channel 02/0 : 55[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127297:127691 [7] NCCL INFO Channel 04/0 : 31[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897245:897638 [5] NCCL INFO Channel 10/0 : 53[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9627:10027 [2] NCCL INFO Channel 04/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624946:625337 [3] NCCL INFO Channel 02/0 : 51[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127291:127690 [1] NCCL INFO Channel 06/0 : 25[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897244:897636 [4] NCCL INFO Channel 05/0 : 52[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9632:10026 [7] NCCL INFO Channel 02/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9626:10025 [1] NCCL INFO Channel 04/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624947:625339 [4] NCCL INFO Channel 08/0 : 36[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624949:625344 [6] NCCL INFO Channel 12/0 : 38[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104433:104825 [4] NCCL INFO Channel 06/0 : 60[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897242:897639 [2] NCCL INFO Channel 04/0 : 50[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9631:10022 [6] NCCL INFO Channel 07/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9625:10020 [0] NCCL INFO Channel 02/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624948:625343 [5] NCCL INFO Channel 12/0 : 53[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104435:104826 [6] NCCL INFO Channel 12/0 : 62[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762821:763224 [6] NCCL INFO Channel 08/0 : 30[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9629:10021 [4] NCCL INFO Channel 03/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624943:625338 [0] NCCL INFO Channel 04/0 : 32[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104433:104825 [4] NCCL INFO Channel 08/0 : 60[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9628:10023 [3] NCCL INFO Channel 04/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624944:625341 [1] NCCL INFO Channel 02/0 : 49[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624945:625340 [2] NCCL INFO Channel 04/0 : 34[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624950:625342 [7] NCCL INFO Channel 04/0 : 55[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127295:127686 [5] NCCL INFO Channel 14/0 : 29[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127294:127687 [4] NCCL INFO Channel 08/0 : 28[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897246:897641 [6] NCCL INFO Channel 09/0 : 54[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897247:897642 [7] NCCL INFO Channel 04/0 : 55[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624949:625344 [6] NCCL INFO Channel 14/0 : 38[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127290:127685 [0] NCCL INFO Channel 04/0 : 24[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104435:104826 [6] NCCL INFO Channel 14/0 : 62[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762819:763221 [4] NCCL INFO Channel 02/0 : 28[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897241:897635 [1] NCCL INFO Channel 06/0 : 49[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897245:897638 [5] NCCL INFO Channel 11/0 : 53[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897243:897637 [3] NCCL INFO Channel 05/0 : 51[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9630:10024 [5] NCCL INFO Channel 09/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624947:625339 [4] NCCL INFO Channel 10/0 : 36[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127293:127689 [3] NCCL INFO Channel 06/0 : 27[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104433:104825 [4] NCCL INFO Channel 10/0 : 60[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762821:763224 [6] NCCL INFO Channel 10/0 : 30[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897240:897640 [0] NCCL INFO Channel 03/0 : 48[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9627:10027 [2] NCCL INFO Channel 05/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624946:625337 [3] NCCL INFO Channel 04/0 : 51[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127292:127692 [2] NCCL INFO Channel 08/0 : 26[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127296:127688 [6] NCCL INFO Channel 14/0 : 30[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897242:897639 [2] NCCL INFO Channel 05/0 : 50[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9626:10025 [1] NCCL INFO Channel 05/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624948:625343 [5] NCCL INFO Channel 14/0 : 53[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624943:625338 [0] NCCL INFO Channel 06/0 : 32[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127297:127691 [7] NCCL INFO Channel 06/0 : 31[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127291:127690 [1] NCCL INFO Channel 08/0 : 25[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897244:897636 [4] NCCL INFO Channel 06/0 : 52[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9631:10022 [6] NCCL INFO Channel 08/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9625:10020 [0] NCCL INFO Channel 03/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624944:625341 [1] NCCL INFO Channel 04/0 : 49[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127290:127685 [0] NCCL INFO Channel 06/0 : 24[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9632:10026 [7] NCCL INFO Channel 03/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624945:625340 [2] NCCL INFO Channel 06/0 : 34[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127294:127687 [4] NCCL INFO Channel 10/0 : 28[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104431:104831 [2] NCCL INFO Channel 00/0 : 58[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9629:10021 [4] NCCL INFO Channel 05/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624950:625342 [7] NCCL INFO Channel 06/0 : 55[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624947:625339 [4] NCCL INFO Channel 12/0 : 36[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127293:127689 [3] NCCL INFO Channel 08/0 : 27[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104435:104826 [6] NCCL INFO Channel 00/0 : 54[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762819:763221 [4] NCCL INFO Channel 06/0 : 28[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9628:10023 [3] NCCL INFO Channel 05/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624946:625337 [3] NCCL INFO Channel 06/0 : 51[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624948:625343 [5] NCCL INFO Channel 00/0 : 37[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127292:127692 [2] NCCL INFO Channel 10/0 : 26[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104433:104825 [4] NCCL INFO Channel 12/0 : 60[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762821:763224 [6] NCCL INFO Channel 12/0 : 30[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9630:10024 [5] NCCL INFO Channel 10/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624943:625338 [0] NCCL INFO Channel 08/0 : 32[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624944:625341 [1] NCCL INFO Channel 06/0 : 49[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624945:625340 [2] NCCL INFO Channel 08/0 : 34[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127291:127690 [1] NCCL INFO Channel 10/0 : 25[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127297:127691 [7] NCCL INFO Channel 08/0 : 31[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104429:104829 [0] NCCL INFO Channel 02/0 : 56[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762819:763221 [4] NCCL INFO Channel 08/0 : 28[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897246:897641 [6] NCCL INFO Channel 10/0 : 54[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9627:10027 [2] NCCL INFO Channel 06/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624950:625342 [7] NCCL INFO Channel 08/0 : 55[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624947:625339 [4] NCCL INFO Channel 14/0 : 36[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127290:127685 [0] NCCL INFO Channel 08/0 : 24[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897247:897642 [7] NCCL INFO Channel 05/0 : 55[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624948:625343 [5] NCCL INFO Channel 02/0 : 37[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624946:625337 [3] NCCL INFO Channel 08/0 : 51[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897245:897638 [5] NCCL INFO Channel 12/0 : 53[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624944:625341 [1] NCCL INFO Channel 08/0 : 49[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624945:625340 [2] NCCL INFO Channel 10/0 : 34[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624943:625338 [0] NCCL INFO Channel 10/0 : 32[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624950:625342 [7] NCCL INFO Channel 10/0 : 55[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897241:897635 [1] NCCL INFO Channel 07/0 : 49[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897243:897637 [3] NCCL INFO Channel 06/0 : 51[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624948:625343 [5] NCCL INFO Channel 04/0 : 37[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897242:897639 [2] NCCL INFO Channel 06/0 : 50[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897240:897640 [0] NCCL INFO Channel 04/0 : 48[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624946:625337 [3] NCCL INFO Channel 10/0 : 51[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104435:104826 [6] NCCL INFO Channel 01/0 : 54[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762821:763224 [6] NCCL INFO Channel 14/0 : 30[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897244:897636 [4] NCCL INFO Channel 07/0 : 52[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624945:625340 [2] NCCL INFO Channel 12/0 : 34[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624943:625338 [0] NCCL INFO Channel 12/0 : 32[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624950:625342 [7] NCCL INFO Channel 12/0 : 55[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104431:104831 [2] NCCL INFO Channel 04/0 : 58[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762817:763223 [2] NCCL INFO Channel 00/0 : 26[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762819:763221 [4] NCCL INFO Channel 10/0 : 28[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9626:10025 [1] NCCL INFO Channel 06/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624944:625341 [1] NCCL INFO Channel 10/0 : 49[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127294:127687 [4] NCCL INFO Channel 12/0 : 28[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127293:127689 [3] NCCL INFO Channel 10/0 : 27[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104433:104825 [4] NCCL INFO Channel 14/0 : 60[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9631:10022 [6] NCCL INFO Channel 09/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624949:625344 [6] NCCL INFO Channel 01/0 : 46[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624948:625343 [5] NCCL INFO Channel 06/0 : 37[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127297:127691 [7] NCCL INFO Channel 10/0 : 31[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104429:104829 [0] NCCL INFO Channel 04/0 : 56[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9628:10023 [3] NCCL INFO Channel 06/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624946:625337 [3] NCCL INFO Channel 12/0 : 51[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127291:127690 [1] NCCL INFO Channel 12/0 : 25[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127290:127685 [0] NCCL INFO Channel 10/0 : 24[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897246:897641 [6] NCCL INFO Channel 11/0 : 54[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897247:897642 [7] NCCL INFO Channel 06/0 : 55[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9625:10020 [0] NCCL INFO Channel 04/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127292:127692 [2] NCCL INFO Channel 12/0 : 26[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897245:897638 [5] NCCL INFO Channel 14/0 : 53[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9627:10027 [2] NCCL INFO Channel 07/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9630:10024 [5] NCCL INFO Channel 11/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9629:10021 [4] NCCL INFO Channel 06/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127294:127687 [4] NCCL INFO Channel 14/0 : 28[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127293:127689 [3] NCCL INFO Channel 12/0 : 27[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762821:763224 [6] NCCL INFO Channel 00/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897241:897635 [1] NCCL INFO Channel 08/0 : 49[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9632:10026 [7] NCCL INFO Channel 04/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127297:127691 [7] NCCL INFO Channel 12/0 : 31[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762815:763225 [0] NCCL INFO Channel 02/0 : 24[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897243:897637 [3] NCCL INFO Channel 07/0 : 51[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9626:10025 [1] NCCL INFO Channel 07/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127291:127690 [1] NCCL INFO Channel 14/0 : 25[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104435:104826 [6] NCCL INFO Channel 02/0 : 54[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897242:897639 [2] NCCL INFO Channel 07/0 : 50[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9631:10022 [6] NCCL INFO Channel 10/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127290:127685 [0] NCCL INFO Channel 12/0 : 24[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104431:104831 [2] NCCL INFO Channel 06/0 : 58[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9628:10023 [3] NCCL INFO Channel 07/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624945:625340 [2] NCCL INFO Channel 14/0 : 34[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127292:127692 [2] NCCL INFO Channel 14/0 : 26[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104433:104825 [4] NCCL INFO Channel 00/0 : 52[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9625:10020 [0] NCCL INFO Channel 05/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624950:625342 [7] NCCL INFO Channel 14/0 : 55[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127293:127689 [3] NCCL INFO Channel 14/0 : 27[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104429:104829 [0] NCCL INFO Channel 06/0 : 56[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104435:104826 [6] NCCL INFO Channel 03/0 : 54[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624944:625341 [1] NCCL INFO Channel 12/0 : 49[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127297:127691 [7] NCCL INFO Channel 14/0 : 31[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762819:763221 [4] NCCL INFO Channel 12/0 : 28[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762817:763223 [2] NCCL INFO Channel 04/0 : 26[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762821:763224 [6] NCCL INFO Channel 01/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624943:625338 [0] NCCL INFO Channel 14/0 : 32[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624948:625343 [5] NCCL INFO Channel 08/0 : 37[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127290:127685 [0] NCCL INFO Channel 14/0 : 24[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897240:897640 [0] NCCL INFO Channel 05/0 : 48[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897244:897636 [4] NCCL INFO Channel 08/0 : 52[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624949:625344 [6] NCCL INFO Channel 03/0 : 46[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897247:897642 [7] NCCL INFO Channel 08/0 : 55[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9627:10027 [2] NCCL INFO Channel 08/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624946:625337 [3] NCCL INFO Channel 14/0 : 51[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897245:897638 [5] NCCL INFO Channel 15/0 : 53[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897246:897641 [6] NCCL INFO Channel 12/0 : 54[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9630:10024 [5] NCCL INFO Channel 12/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624947:625339 [4] NCCL INFO Channel 01/0 : 44[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624950:625342 [7] NCCL INFO Channel 00/0 : 39[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104433:104825 [4] NCCL INFO Channel 01/0 : 52[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104431:104831 [2] NCCL INFO Channel 08/0 : 58[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897241:897635 [1] NCCL INFO Channel 10/0 : 49[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9629:10021 [4] NCCL INFO Channel 07/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624944:625341 [1] NCCL INFO Channel 14/0 : 49[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104435:104826 [6] NCCL INFO Channel 04/0 : 54[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762815:763225 [0] NCCL INFO Channel 04/0 : 24[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897243:897637 [3] NCCL INFO Channel 08/0 : 51[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9626:10025 [1] NCCL INFO Channel 08/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624948:625343 [5] NCCL INFO Channel 10/0 : 37[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104429:104829 [0] NCCL INFO Channel 08/0 : 56[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762819:763221 [4] NCCL INFO Channel 14/0 : 28[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897242:897639 [2] NCCL INFO Channel 08/0 : 50[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9631:10022 [6] NCCL INFO Channel 11/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9632:10026 [7] NCCL INFO Channel 05/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624949:625344 [6] NCCL INFO Channel 05/0 : 46[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104434:104828 [5] NCCL INFO Channel 00/0 : 61[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762821:763224 [6] NCCL INFO Channel 02/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897240:897640 [0] NCCL INFO Channel 06/0 : 48[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9628:10023 [3] NCCL INFO Channel 08/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624946:625337 [3] NCCL INFO Channel 00/0 : 35[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762817:763223 [2] NCCL INFO Channel 06/0 : 26[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9625:10020 [0] NCCL INFO Channel 06/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624947:625339 [4] NCCL INFO Channel 03/0 : 44[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624950:625342 [7] NCCL INFO Channel 02/0 : 39[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9627:10027 [2] NCCL INFO Channel 09/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624948:625343 [5] NCCL INFO Channel 12/0 : 37[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624944:625341 [1] NCCL INFO Channel 00/0 : 33[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9630:10024 [5] NCCL INFO Channel 14/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624949:625344 [6] NCCL INFO Channel 07/0 : 46[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9629:10021 [4] NCCL INFO Channel 08/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624946:625337 [3] NCCL INFO Channel 02/0 : 35[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104433:104825 [4] NCCL INFO Channel 02/0 : 52[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624950:625342 [7] NCCL INFO Channel 04/0 : 39[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624947:625339 [4] NCCL INFO Channel 05/0 : 44[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104431:104831 [2] NCCL INFO Channel 10/0 : 58[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897244:897636 [4] NCCL INFO Channel 09/0 : 52[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624948:625343 [5] NCCL INFO Channel 14/0 : 37[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104435:104826 [6] NCCL INFO Channel 05/0 : 54[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762819:763221 [4] NCCL INFO Channel 00/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897247:897642 [7] NCCL INFO Channel 09/0 : 55[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897246:897641 [6] NCCL INFO Channel 13/0 : 54[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897245:897638 [5] NCCL INFO Channel 01/0 : 45[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624944:625341 [1] NCCL INFO Channel 02/0 : 33[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104429:104829 [0] NCCL INFO Channel 10/0 : 56[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897241:897635 [1] NCCL INFO Channel 11/0 : 49[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624949:625344 [6] NCCL INFO Channel 09/0 : 46[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624946:625337 [3] NCCL INFO Channel 04/0 : 35[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104434:104828 [5] NCCL INFO Channel 02/0 : 61[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897243:897637 [3] NCCL INFO Channel 09/0 : 51[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9628:10023 [3] NCCL INFO Channel 09/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9631:10022 [6] NCCL INFO Channel 12/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624945:625340 [2] NCCL INFO Channel 01/0 : 42[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624950:625342 [7] NCCL INFO Channel 06/0 : 39[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897242:897639 [2] NCCL INFO Channel 09/0 : 50[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9626:10025 [1] NCCL INFO Channel 10/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624947:625339 [4] NCCL INFO Channel 07/0 : 44[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897240:897640 [0] NCCL INFO Channel 07/0 : 48[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9632:10026 [7] NCCL INFO Channel 06/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9625:10020 [0] NCCL INFO Channel 07/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624944:625341 [1] NCCL INFO Channel 04/0 : 33[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624949:625344 [6] NCCL INFO Channel 11/0 : 46[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624946:625337 [3] NCCL INFO Channel 06/0 : 35[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762821:763224 [6] NCCL INFO Channel 03/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762815:763225 [0] NCCL INFO Channel 06/0 : 24[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762817:763223 [2] NCCL INFO Channel 08/0 : 26[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897244:897636 [4] NCCL INFO Channel 10/0 : 52[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9630:10024 [5] NCCL INFO Channel 15/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9627:10027 [2] NCCL INFO Channel 11/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9629:10021 [4] NCCL INFO Channel 09/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624943:625338 [0] NCCL INFO Channel 01/0 : 40[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762820:763219 [5] NCCL INFO Channel 00/0 : 29[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624945:625340 [2] NCCL INFO Channel 03/0 : 42[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104433:104825 [4] NCCL INFO Channel 03/0 : 52[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104431:104831 [2] NCCL INFO Channel 12/0 : 58[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104435:104826 [6] NCCL INFO Channel 07/0 : 54[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104429:104829 [0] NCCL INFO Channel 12/0 : 56[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104434:104828 [5] NCCL INFO Channel 04/0 : 61[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104433:104825 [4] NCCL INFO Channel 05/0 : 52[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897247:897642 [7] NCCL INFO Channel 10/0 : 55[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9628:10023 [3] NCCL INFO Channel 10/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762819:763221 [4] NCCL INFO Channel 01/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897246:897641 [6] NCCL INFO Channel 15/0 : 54[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9631:10022 [6] NCCL INFO Channel 13/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9626:10025 [1] NCCL INFO Channel 11/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762821:763224 [6] NCCL INFO Channel 04/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897245:897638 [5] NCCL INFO Channel 03/0 : 45[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9625:10020 [0] NCCL INFO Channel 09/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624947:625339 [4] NCCL INFO Channel 09/0 : 44[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624950:625342 [7] NCCL INFO Channel 08/0 : 39[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762815:763225 [0] NCCL INFO Channel 08/0 : 24[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897241:897635 [1] NCCL INFO Channel 12/0 : 49[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9632:10026 [7] NCCL INFO Channel 08/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9627:10027 [2] NCCL INFO Channel 12/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624944:625341 [1] NCCL INFO Channel 06/0 : 33[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104431:104831 [2] NCCL INFO Channel 14/0 : 58[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104435:104826 [6] NCCL INFO Channel 08/0 : 54[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762817:763223 [2] NCCL INFO Channel 10/0 : 26[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897243:897637 [3] NCCL INFO Channel 10/0 : 51[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897242:897639 [2] NCCL INFO Channel 11/0 : 50[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897244:897636 [4] NCCL INFO Channel 11/0 : 52[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897240:897640 [0] NCCL INFO Channel 09/0 : 48[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9630:10024 [5] NCCL INFO Channel 01/0 : 13[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624946:625337 [3] NCCL INFO Channel 08/0 : 35[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624949:625344 [6] NCCL INFO Channel 13/0 : 46[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104434:104828 [5] NCCL INFO Channel 06/0 : 61[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104429:104829 [0] NCCL INFO Channel 14/0 : 56[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104433:104825 [4] NCCL INFO Channel 06/0 : 52[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762820:763219 [5] NCCL INFO Channel 02/0 : 29[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762819:763221 [4] NCCL INFO Channel 02/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9629:10021 [4] NCCL INFO Channel 10/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624943:625338 [0] NCCL INFO Channel 03/0 : 40[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9628:10023 [3] NCCL INFO Channel 12/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624945:625340 [2] NCCL INFO Channel 05/0 : 42[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624950:625342 [7] NCCL INFO Channel 10/0 : 39[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624947:625339 [4] NCCL INFO Channel 11/0 : 44[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624944:625341 [1] NCCL INFO Channel 08/0 : 33[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897247:897642 [7] NCCL INFO Channel 11/0 : 55[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624949:625344 [6] NCCL INFO Channel 15/0 : 46[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624946:625337 [3] NCCL INFO Channel 10/0 : 35[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104435:104826 [6] NCCL INFO Channel 09/0 : 54[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104431:104831 [2] NCCL INFO Channel 00/0 : 50[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897246:897641 [6] NCCL INFO Channel 01/0 : 46[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624943:625338 [0] NCCL INFO Channel 05/0 : 40[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104434:104828 [5] NCCL INFO Channel 08/0 : 61[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762821:763224 [6] NCCL INFO Channel 05/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897245:897638 [5] NCCL INFO Channel 05/0 : 45[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9631:10022 [6] NCCL INFO Channel 15/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9626:10025 [1] NCCL INFO Channel 12/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624945:625340 [2] NCCL INFO Channel 07/0 : 42[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104433:104825 [4] NCCL INFO Channel 07/0 : 52[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104429:104829 [0] NCCL INFO Channel 01/0 : 48[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762815:763225 [0] NCCL INFO Channel 10/0 : 24[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897241:897635 [1] NCCL INFO Channel 13/0 : 49[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9625:10020 [0] NCCL INFO Channel 10/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624947:625339 [4] NCCL INFO Channel 13/0 : 44[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762817:763223 [2] NCCL INFO Channel 12/0 : 26[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9632:10026 [7] NCCL INFO Channel 09/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624950:625342 [7] NCCL INFO Channel 12/0 : 39[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624948:625343 [5] NCCL INFO Channel 01/0 : 45[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762819:763221 [4] NCCL INFO Channel 03/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9630:10024 [5] NCCL INFO Channel 03/0 : 13[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9627:10027 [2] NCCL INFO Channel 13/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624944:625341 [1] NCCL INFO Channel 10/0 : 33[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624946:625337 [3] NCCL INFO Channel 12/0 : 35[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762820:763219 [5] NCCL INFO Channel 04/0 : 29[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9629:10021 [4] NCCL INFO Channel 11/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9628:10023 [3] NCCL INFO Channel 13/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624943:625338 [0] NCCL INFO Channel 07/0 : 40[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762821:763224 [6] NCCL INFO Channel 07/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624945:625340 [2] NCCL INFO Channel 09/0 : 42[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104435:104826 [6] NCCL INFO Channel 10/0 : 54[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104431:104831 [2] NCCL INFO Channel 01/0 : 50[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762815:763225 [0] NCCL INFO Channel 12/0 : 24[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897243:897637 [3] NCCL INFO Channel 12/0 : 51[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897244:897636 [4] NCCL INFO Channel 13/0 : 52[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897242:897639 [2] NCCL INFO Channel 12/0 : 50[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624947:625339 [4] NCCL INFO Channel 15/0 : 44[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104434:104828 [5] NCCL INFO Channel 10/0 : 61[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897240:897640 [0] NCCL INFO Channel 10/0 : 48[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624950:625342 [7] NCCL INFO Channel 14/0 : 39[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624948:625343 [5] NCCL INFO Channel 03/0 : 45[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104433:104825 [4] NCCL INFO Channel 08/0 : 52[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897247:897642 [7] NCCL INFO Channel 12/0 : 55[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9626:10025 [1] NCCL INFO Channel 13/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9631:10022 [6] NCCL INFO Channel 01/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9625:10020 [0] NCCL INFO Channel 11/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104429:104829 [0] NCCL INFO Channel 02/0 : 48[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897246:897641 [6] NCCL INFO Channel 03/0 : 46[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897245:897638 [5] NCCL INFO Channel 07/0 : 45[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897241:897635 [1] NCCL INFO Channel 14/0 : 49[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9630:10024 [5] NCCL INFO Channel 05/0 : 13[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762817:763223 [2] NCCL INFO Channel 14/0 : 26[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762819:763221 [4] NCCL INFO Channel 05/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762820:763219 [5] NCCL INFO Channel 06/0 : 29[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762821:763224 [6] NCCL INFO Channel 08/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762815:763225 [0] NCCL INFO Channel 14/0 : 24[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624946:625337 [3] NCCL INFO Channel 14/0 : 35[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624944:625341 [1] NCCL INFO Channel 12/0 : 33[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624943:625338 [0] NCCL INFO Channel 09/0 : 40[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104435:104826 [6] NCCL INFO Channel 11/0 : 54[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9632:10026 [7] NCCL INFO Channel 10/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624945:625340 [2] NCCL INFO Channel 11/0 : 42[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104431:104831 [2] NCCL INFO Channel 03/0 : 50[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104434:104828 [5] NCCL INFO Channel 12/0 : 61[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897244:897636 [4] NCCL INFO Channel 14/0 : 52[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9627:10027 [2] NCCL INFO Channel 14/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624948:625343 [5] NCCL INFO Channel 05/0 : 45[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897243:897637 [3] NCCL INFO Channel 13/0 : 51[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9628:10023 [3] NCCL INFO Channel 14/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624944:625341 [1] NCCL INFO Channel 14/0 : 33[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762817:763223 [2] NCCL INFO Channel 00/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897242:897639 [2] NCCL INFO Channel 13/0 : 50[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9629:10021 [4] NCCL INFO Channel 13/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9626:10025 [1] NCCL INFO Channel 14/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624943:625338 [0] NCCL INFO Channel 11/0 : 40[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897240:897640 [0] NCCL INFO Channel 11/0 : 48[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9625:10020 [0] NCCL INFO Channel 12/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624945:625340 [2] NCCL INFO Channel 13/0 : 42[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624948:625343 [5] NCCL INFO Channel 07/0 : 45[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897247:897642 [7] NCCL INFO Channel 13/0 : 55[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9631:10022 [6] NCCL INFO Channel 03/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624943:625338 [0] NCCL INFO Channel 13/0 : 40[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897245:897638 [5] NCCL INFO Channel 09/0 : 45[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9630:10024 [5] NCCL INFO Channel 07/0 : 13[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624945:625340 [2] NCCL INFO Channel 15/0 : 42[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624948:625343 [5] NCCL INFO Channel 09/0 : 45[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624943:625338 [0] NCCL INFO Channel 15/0 : 40[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104436:104824 [7] NCCL INFO Channel 00/0 : 63[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104432:104827 [3] NCCL INFO Channel 00/0 : 59[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104433:104825 [4] NCCL INFO Channel 09/0 : 52[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104429:104829 [0] NCCL INFO Channel 03/0 : 48[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762815:763225 [0] NCCL INFO Channel 01/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762820:763219 [5] NCCL INFO Channel 08/0 : 29[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762819:763221 [4] NCCL INFO Channel 06/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762821:763224 [6] NCCL INFO Channel 09/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104435:104826 [6] NCCL INFO Channel 12/0 : 54[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762822:763222 [7] NCCL INFO Channel 00/0 : 31[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762817:763223 [2] NCCL INFO Channel 01/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762818:763220 [3] NCCL INFO Channel 00/0 : 27[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897246:897641 [6] NCCL INFO Channel 05/0 : 46[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9627:10027 [2] NCCL INFO Channel 15/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9632:10026 [7] NCCL INFO Channel 11/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897241:897635 [1] NCCL INFO Channel 15/0 : 49[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9628:10023 [3] NCCL INFO Channel 15/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624948:625343 [5] NCCL INFO Channel 11/0 : 45[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104431:104831 [2] NCCL INFO Channel 04/0 : 50[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104434:104828 [5] NCCL INFO Channel 14/0 : 61[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104430:104830 [1] NCCL INFO Channel 00/0 : 57[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897244:897636 [4] NCCL INFO Channel 15/0 : 52[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897243:897637 [3] NCCL INFO Channel 14/0 : 51[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9626:10025 [1] NCCL INFO Channel 15/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624950:625342 [7] NCCL INFO Channel 01/0 : 47[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104436:104824 [7] NCCL INFO Channel 02/0 : 63[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897242:897639 [2] NCCL INFO Channel 14/0 : 50[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9625:10020 [0] NCCL INFO Channel 13/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624946:625337 [3] NCCL INFO Channel 01/0 : 43[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624948:625343 [5] NCCL INFO Channel 15/0 : 45[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104433:104825 [4] NCCL INFO Channel 10/0 : 52[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897247:897642 [7] NCCL INFO Channel 14/0 : 55[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897240:897640 [0] NCCL INFO Channel 12/0 : 48[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9629:10021 [4] NCCL INFO Channel 14/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104429:104829 [0] NCCL INFO Channel 04/0 : 48[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104435:104826 [6] NCCL INFO Channel 13/0 : 54[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104432:104827 [3] NCCL INFO Channel 02/0 : 59[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762821:763224 [6] NCCL INFO Channel 10/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9631:10022 [6] NCCL INFO Channel 05/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762815:763225 [0] NCCL INFO Channel 02/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762820:763219 [5] NCCL INFO Channel 10/0 : 29[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624950:625342 [7] NCCL INFO Channel 03/0 : 47[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897245:897638 [5] NCCL INFO Channel 11/0 : 45[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897246:897641 [6] NCCL INFO Channel 07/0 : 46[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897241:897635 [1] NCCL INFO Channel 01/0 : 41[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624946:625337 [3] NCCL INFO Channel 03/0 : 43[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624944:625341 [1] NCCL INFO Channel 01/0 : 41[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104431:104831 [2] NCCL INFO Channel 05/0 : 50[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897244:897636 [4] NCCL INFO Channel 01/0 : 44[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9630:10024 [5] NCCL INFO Channel 09/0 : 13[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104434:104828 [5] NCCL INFO Channel 00/0 : 53[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897243:897637 [3] NCCL INFO Channel 15/0 : 51[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897242:897639 [2] NCCL INFO Channel 15/0 : 50[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9627:10027 [2] NCCL INFO Channel 01/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9632:10026 [7] NCCL INFO Channel 12/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104436:104824 [7] NCCL INFO Channel 04/0 : 63[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104430:104830 [1] NCCL INFO Channel 02/0 : 57[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762819:763221 [4] NCCL INFO Channel 07/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897247:897642 [7] NCCL INFO Channel 01/0 : 47[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897240:897640 [0] NCCL INFO Channel 13/0 : 48[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9628:10023 [3] NCCL INFO Channel 01/0 : 11[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104433:104825 [4] NCCL INFO Channel 11/0 : 52[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762817:763223 [2] NCCL INFO Channel 03/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9626:10025 [1] NCCL INFO Channel 01/0 : 9[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104429:104829 [0] NCCL INFO Channel 05/0 : 48[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762818:763220 [3] NCCL INFO Channel 02/0 : 27[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624946:625337 [3] NCCL INFO Channel 05/0 : 43[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624950:625342 [7] NCCL INFO Channel 05/0 : 47[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624944:625341 [1] NCCL INFO Channel 03/0 : 41[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104435:104826 [6] NCCL INFO Channel 15/0 : 54[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762816:763218 [1] NCCL INFO Channel 00/0 : 25[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762822:763222 [7] NCCL INFO Channel 02/0 : 31[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762815:763225 [0] NCCL INFO Channel 03/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897245:897638 [5] NCCL INFO Channel 15/0 : 45[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897246:897641 [6] NCCL INFO Channel 09/0 : 46[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897241:897635 [1] NCCL INFO Channel 03/0 : 41[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9625:10020 [0] NCCL INFO Channel 14/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897244:897636 [4] NCCL INFO Channel 03/0 : 44[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897243:897637 [3] NCCL INFO Channel 01/0 : 43[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9631:10022 [6] NCCL INFO Channel 07/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104432:104827 [3] NCCL INFO Channel 04/0 : 59[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9627:10027 [2] NCCL INFO Channel 03/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9629:10021 [4] NCCL INFO Channel 15/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9630:10024 [5] NCCL INFO Channel 11/0 : 13[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624946:625337 [3] NCCL INFO Channel 07/0 : 43[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104434:104828 [5] NCCL INFO Channel 01/0 : 53[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104431:104831 [2] NCCL INFO Channel 06/0 : 50[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762821:763224 [6] NCCL INFO Channel 11/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9632:10026 [7] NCCL INFO Channel 13/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104436:104824 [7] NCCL INFO Channel 06/0 : 63[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762820:763219 [5] NCCL INFO Channel 12/0 : 29[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9626:10025 [1] NCCL INFO Channel 03/0 : 9[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9628:10023 [3] NCCL INFO Channel 03/0 : 11[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104430:104830 [1] NCCL INFO Channel 04/0 : 57[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104433:104825 [4] NCCL INFO Channel 13/0 : 52[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762819:763221 [4] NCCL INFO Channel 08/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104429:104829 [0] NCCL INFO Channel 06/0 : 48[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762817:763223 [2] NCCL INFO Channel 04/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104432:104827 [3] NCCL INFO Channel 06/0 : 59[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762818:763220 [3] NCCL INFO Channel 04/0 : 27[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897247:897642 [7] NCCL INFO Channel 03/0 : 47[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897242:897639 [2] NCCL INFO Channel 01/0 : 42[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897240:897640 [0] NCCL INFO Channel 14/0 : 48[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762816:763218 [1] NCCL INFO Channel 02/0 : 25[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897246:897641 [6] NCCL INFO Channel 11/0 : 46[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624950:625342 [7] NCCL INFO Channel 07/0 : 47[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624944:625341 [1] NCCL INFO Channel 05/0 : 41[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762822:763222 [7] NCCL INFO Channel 04/0 : 31[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897241:897635 [1] NCCL INFO Channel 05/0 : 41[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9625:10020 [0] NCCL INFO Channel 15/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762815:763225 [0] NCCL INFO Channel 04/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9627:10027 [2] NCCL INFO Channel 05/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9631:10022 [6] NCCL INFO Channel 09/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104434:104828 [5] NCCL INFO Channel 02/0 : 53[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9630:10024 [5] NCCL INFO Channel 15/0 : 13[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104431:104831 [2] NCCL INFO Channel 07/0 : 50[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9629:10021 [4] NCCL INFO Channel 01/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104436:104824 [7] NCCL INFO Channel 08/0 : 63[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9632:10026 [7] NCCL INFO Channel 14/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9626:10025 [1] NCCL INFO Channel 05/0 : 9[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9628:10023 [3] NCCL INFO Channel 05/0 : 11[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624946:625337 [3] NCCL INFO Channel 09/0 : 43[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104433:104825 [4] NCCL INFO Channel 14/0 : 52[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9625:10020 [0] NCCL INFO Channel 01/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624950:625342 [7] NCCL INFO Channel 09/0 : 47[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104429:104829 [0] NCCL INFO Channel 07/0 : 48[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762821:763224 [6] NCCL INFO Channel 12/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897244:897636 [4] NCCL INFO Channel 05/0 : 44[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624944:625341 [1] NCCL INFO Channel 07/0 : 41[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104432:104827 [3] NCCL INFO Channel 08/0 : 59[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762820:763219 [5] NCCL INFO Channel 14/0 : 29[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897247:897642 [7] NCCL INFO Channel 05/0 : 47[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897243:897637 [3] NCCL INFO Channel 03/0 : 43[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104434:104828 [5] NCCL INFO Channel 03/0 : 53[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762819:763221 [4] NCCL INFO Channel 09/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762817:763223 [2] NCCL INFO Channel 05/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897242:897639 [2] NCCL INFO Channel 03/0 : 42[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762818:763220 [3] NCCL INFO Channel 06/0 : 27[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762816:763218 [1] NCCL INFO Channel 04/0 : 25[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897240:897640 [0] NCCL INFO Channel 15/0 : 48[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9627:10027 [2] NCCL INFO Channel 07/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762815:763225 [0] NCCL INFO Channel 05/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897246:897641 [6] NCCL INFO Channel 13/0 : 46[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897241:897635 [1] NCCL INFO Channel 07/0 : 41[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9631:10022 [6] NCCL INFO Channel 11/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762822:763222 [7] NCCL INFO Channel 06/0 : 31[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762821:763224 [6] NCCL INFO Channel 13/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897244:897636 [4] NCCL INFO Channel 07/0 : 44[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624946:625337 [3] NCCL INFO Channel 13/0 : 43[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104430:104830 [1] NCCL INFO Channel 06/0 : 57[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762820:763219 [5] NCCL INFO Channel 00/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624950:625342 [7] NCCL INFO Channel 11/0 : 47[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104436:104824 [7] NCCL INFO Channel 10/0 : 63[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104431:104831 [2] NCCL INFO Channel 08/0 : 50[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624944:625341 [1] NCCL INFO Channel 11/0 : 41[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104433:104825 [4] NCCL INFO Channel 15/0 : 52[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9629:10021 [4] NCCL INFO Channel 03/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9626:10025 [1] NCCL INFO Channel 07/0 : 9[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104429:104829 [0] NCCL INFO Channel 09/0 : 48[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9625:10020 [0] NCCL INFO Channel 03/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104432:104827 [3] NCCL INFO Channel 10/0 : 59[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104434:104828 [5] NCCL INFO Channel 04/0 : 53[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9632:10026 [7] NCCL INFO Channel 01/0 : 15[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104430:104830 [1] NCCL INFO Channel 08/0 : 57[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9628:10023 [3] NCCL INFO Channel 07/0 : 11[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104436:104824 [7] NCCL INFO Channel 12/0 : 63[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104431:104831 [2] NCCL INFO Channel 09/0 : 50[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897247:897642 [7] NCCL INFO Channel 07/0 : 47[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9627:10027 [2] NCCL INFO Channel 09/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762817:763223 [2] NCCL INFO Channel 06/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897242:897639 [2] NCCL INFO Channel 05/0 : 42[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9631:10022 [6] NCCL INFO Channel 13/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624946:625337 [3] NCCL INFO Channel 15/0 : 43[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762819:763221 [4] NCCL INFO Channel 10/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897243:897637 [3] NCCL INFO Channel 05/0 : 43[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897240:897640 [0] NCCL INFO Channel 01/0 : 40[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624950:625342 [7] NCCL INFO Channel 13/0 : 47[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762816:763218 [1] NCCL INFO Channel 06/0 : 25[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762818:763220 [3] NCCL INFO Channel 08/0 : 27[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897246:897641 [6] NCCL INFO Channel 15/0 : 46[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624944:625341 [1] NCCL INFO Channel 13/0 : 41[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762815:763225 [0] NCCL INFO Channel 06/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897241:897635 [1] NCCL INFO Channel 11/0 : 41[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897244:897636 [4] NCCL INFO Channel 09/0 : 44[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104429:104829 [0] NCCL INFO Channel 10/0 : 48[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762821:763224 [6] NCCL INFO Channel 15/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897247:897642 [7] NCCL INFO Channel 09/0 : 47[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104434:104828 [5] NCCL INFO Channel 06/0 : 53[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104432:104827 [3] NCCL INFO Channel 12/0 : 59[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762820:763219 [5] NCCL INFO Channel 01/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897242:897639 [2] NCCL INFO Channel 07/0 : 42[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104430:104830 [1] NCCL INFO Channel 10/0 : 57[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762822:763222 [7] NCCL INFO Channel 08/0 : 31[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9626:10025 [1] NCCL INFO Channel 11/0 : 9[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104436:104824 [7] NCCL INFO Channel 14/0 : 63[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104431:104831 [2] NCCL INFO Channel 11/0 : 50[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762817:763223 [2] NCCL INFO Channel 07/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762819:763221 [4] NCCL INFO Channel 11/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9629:10021 [4] NCCL INFO Channel 05/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9625:10020 [0] NCCL INFO Channel 05/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624944:625341 [1] NCCL INFO Channel 15/0 : 41[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104429:104829 [0] NCCL INFO Channel 11/0 : 48[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9632:10026 [7] NCCL INFO Channel 03/0 : 15[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897243:897637 [3] NCCL INFO Channel 07/0 : 43[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897240:897640 [0] NCCL INFO Channel 03/0 : 40[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9627:10027 [2] NCCL INFO Channel 11/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897244:897636 [4] NCCL INFO Channel 11/0 : 44[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897241:897635 [1] NCCL INFO Channel 13/0 : 41[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9628:10023 [3] NCCL INFO Channel 09/0 : 11[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9631:10022 [6] NCCL INFO Channel 15/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762816:763218 [1] NCCL INFO Channel 08/0 : 25[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762818:763220 [3] NCCL INFO Channel 10/0 : 27[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762815:763225 [0] NCCL INFO Channel 07/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897247:897642 [7] NCCL INFO Channel 11/0 : 47[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762820:763219 [5] NCCL INFO Channel 02/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897240:897640 [0] NCCL INFO Channel 05/0 : 40[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897243:897637 [3] NCCL INFO Channel 09/0 : 43[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104434:104828 [5] NCCL INFO Channel 07/0 : 53[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104432:104827 [3] NCCL INFO Channel 14/0 : 59[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762822:763222 [7] NCCL INFO Channel 10/0 : 31[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897242:897639 [2] NCCL INFO Channel 09/0 : 42[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104430:104830 [1] NCCL INFO Channel 12/0 : 57[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762817:763223 [2] NCCL INFO Channel 08/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104436:104824 [7] NCCL INFO Channel 00/0 : 55[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104429:104829 [0] NCCL INFO Channel 12/0 : 48[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104431:104831 [2] NCCL INFO Channel 12/0 : 50[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762819:763221 [4] NCCL INFO Channel 13/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762816:763218 [1] NCCL INFO Channel 10/0 : 25[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762815:763225 [0] NCCL INFO Channel 09/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9626:10025 [1] NCCL INFO Channel 13/0 : 9[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762818:763220 [3] NCCL INFO Channel 12/0 : 27[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9629:10021 [4] NCCL INFO Channel 07/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9625:10020 [0] NCCL INFO Channel 07/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762820:763219 [5] NCCL INFO Channel 03/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9632:10026 [7] NCCL INFO Channel 05/0 : 15[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9628:10023 [3] NCCL INFO Channel 13/0 : 11[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9626:10025 [1] NCCL INFO Channel 15/0 : 9[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9627:10027 [2] NCCL INFO Channel 13/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104434:104828 [5] NCCL INFO Channel 08/0 : 53[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104432:104827 [3] NCCL INFO Channel 00/0 : 51[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104430:104830 [1] NCCL INFO Channel 14/0 : 57[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104436:104824 [7] NCCL INFO Channel 01/0 : 55[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104429:104829 [0] NCCL INFO Channel 13/0 : 48[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897244:897636 [4] NCCL INFO Channel 13/0 : 44[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762817:763223 [2] NCCL INFO Channel 09/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762822:763222 [7] NCCL INFO Channel 12/0 : 31[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762819:763221 [4] NCCL INFO Channel 14/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897241:897635 [1] NCCL INFO Channel 15/0 : 41[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897247:897642 [7] NCCL INFO Channel 13/0 : 47[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762816:763218 [1] NCCL INFO Channel 12/0 : 25[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897240:897640 [0] NCCL INFO Channel 07/0 : 40[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9625:10020 [0] NCCL INFO Channel 09/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762815:763225 [0] NCCL INFO Channel 10/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897243:897637 [3] NCCL INFO Channel 13/0 : 43[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9629:10021 [4] NCCL INFO Channel 09/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762818:763220 [3] NCCL INFO Channel 14/0 : 27[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762820:763219 [5] NCCL INFO Channel 04/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897244:897636 [4] NCCL INFO Channel 15/0 : 44[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9628:10023 [3] NCCL INFO Channel 15/0 : 11[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897242:897639 [2] NCCL INFO Channel 11/0 : 42[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9632:10026 [7] NCCL INFO Channel 07/0 : 15[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104431:104831 [2] NCCL INFO Channel 13/0 : 50[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104432:104827 [3] NCCL INFO Channel 01/0 : 51[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104434:104828 [5] NCCL INFO Channel 09/0 : 53[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104430:104830 [1] NCCL INFO Channel 00/0 : 49[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104436:104824 [7] NCCL INFO Channel 02/0 : 55[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104429:104829 [0] NCCL INFO Channel 14/0 : 48[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762817:763223 [2] NCCL INFO Channel 11/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762819:763221 [4] NCCL INFO Channel 15/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762822:763222 [7] NCCL INFO Channel 14/0 : 31[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897240:897640 [0] NCCL INFO Channel 09/0 : 40[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9627:10027 [2] NCCL INFO Channel 15/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762815:763225 [0] NCCL INFO Channel 11/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897243:897637 [3] NCCL INFO Channel 15/0 : 43[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9625:10020 [0] NCCL INFO Channel 11/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762816:763218 [1] NCCL INFO Channel 14/0 : 25[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897242:897639 [2] NCCL INFO Channel 13/0 : 42[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9629:10021 [4] NCCL INFO Channel 11/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104431:104831 [2] NCCL INFO Channel 14/0 : 50[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762818:763220 [3] NCCL INFO Channel 00/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762820:763219 [5] NCCL INFO Channel 06/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897240:897640 [0] NCCL INFO Channel 11/0 : 40[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9632:10026 [7] NCCL INFO Channel 09/0 : 15[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104434:104828 [5] NCCL INFO Channel 10/0 : 53[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104432:104827 [3] NCCL INFO Channel 02/0 : 51[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897242:897639 [2] NCCL INFO Channel 15/0 : 42[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104430:104830 [1] NCCL INFO Channel 02/0 : 49[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897240:897640 [0] NCCL INFO Channel 13/0 : 40[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104436:104824 [7] NCCL INFO Channel 03/0 : 55[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104429:104829 [0] NCCL INFO Channel 15/0 : 48[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762817:763223 [2] NCCL INFO Channel 12/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762822:763222 [7] NCCL INFO Channel 00/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762815:763225 [0] NCCL INFO Channel 12/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9629:10021 [4] NCCL INFO Channel 13/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9625:10020 [0] NCCL INFO Channel 13/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9632:10026 [7] NCCL INFO Channel 11/0 : 15[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762816:763218 [1] NCCL INFO Channel 00/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9629:10021 [4] NCCL INFO Channel 15/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762818:763220 [3] NCCL INFO Channel 01/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762820:763219 [5] NCCL INFO Channel 07/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897240:897640 [0] NCCL INFO Channel 15/0 : 40[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762817:763223 [2] NCCL INFO Channel 13/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9625:10020 [0] NCCL INFO Channel 15/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9632:10026 [7] NCCL INFO Channel 13/0 : 15[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762822:763222 [7] NCCL INFO Channel 01/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762815:763225 [0] NCCL INFO Channel 13/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762816:763218 [1] NCCL INFO Channel 02/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762818:763220 [3] NCCL INFO Channel 02/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762820:763219 [5] NCCL INFO Channel 08/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762817:763223 [2] NCCL INFO Channel 14/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762822:763222 [7] NCCL INFO Channel 02/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762815:763225 [0] NCCL INFO Channel 14/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762818:763220 [3] NCCL INFO Channel 04/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762817:763223 [2] NCCL INFO Channel 15/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762820:763219 [5] NCCL INFO Channel 09/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762822:763222 [7] NCCL INFO Channel 03/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762815:763225 [0] NCCL INFO Channel 15/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762816:763218 [1] NCCL INFO Channel 03/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762818:763220 [3] NCCL INFO Channel 05/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762820:763219 [5] NCCL INFO Channel 10/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762822:763222 [7] NCCL INFO Channel 04/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762816:763218 [1] NCCL INFO Channel 04/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762818:763220 [3] NCCL INFO Channel 06/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762820:763219 [5] NCCL INFO Channel 11/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762822:763222 [7] NCCL INFO Channel 05/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762816:763218 [1] NCCL INFO Channel 05/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762818:763220 [3] NCCL INFO Channel 07/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762820:763219 [5] NCCL INFO Channel 12/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762816:763218 [1] NCCL INFO Channel 06/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762822:763222 [7] NCCL INFO Channel 06/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762818:763220 [3] NCCL INFO Channel 08/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762820:763219 [5] NCCL INFO Channel 14/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762816:763218 [1] NCCL INFO Channel 07/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762818:763220 [3] NCCL INFO Channel 09/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762822:763222 [7] NCCL INFO Channel 08/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762820:763219 [5] NCCL INFO Channel 15/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762816:763218 [1] NCCL INFO Channel 08/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762818:763220 [3] NCCL INFO Channel 10/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762822:763222 [7] NCCL INFO Channel 09/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762816:763218 [1] NCCL INFO Channel 10/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762818:763220 [3] NCCL INFO Channel 12/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762822:763222 [7] NCCL INFO Channel 10/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762816:763218 [1] NCCL INFO Channel 11/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762818:763220 [3] NCCL INFO Channel 13/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762822:763222 [7] NCCL INFO Channel 11/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762816:763218 [1] NCCL INFO Channel 12/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762818:763220 [3] NCCL INFO Channel 14/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762822:763222 [7] NCCL INFO Channel 12/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762816:763218 [1] NCCL INFO Channel 13/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762818:763220 [3] NCCL INFO Channel 15/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762822:763222 [7] NCCL INFO Channel 13/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762816:763218 [1] NCCL INFO Channel 14/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762822:763222 [7] NCCL INFO Channel 14/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762816:763218 [1] NCCL INFO Channel 15/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104431:104831 [2] NCCL INFO Channel 15/0 : 50[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104432:104827 [3] NCCL INFO Channel 04/0 : 51[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104434:104828 [5] NCCL INFO Channel 11/0 : 53[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104430:104830 [1] NCCL INFO Channel 03/0 : 49[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104436:104824 [7] NCCL INFO Channel 04/0 : 55[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104432:104827 [3] NCCL INFO Channel 05/0 : 51[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104434:104828 [5] NCCL INFO Channel 12/0 : 53[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104436:104824 [7] NCCL INFO Channel 05/0 : 55[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104430:104830 [1] NCCL INFO Channel 04/0 : 49[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104432:104827 [3] NCCL INFO Channel 06/0 : 51[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104436:104824 [7] NCCL INFO Channel 06/0 : 55[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104434:104828 [5] NCCL INFO Channel 14/0 : 53[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104430:104830 [1] NCCL INFO Channel 05/0 : 49[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104432:104827 [3] NCCL INFO Channel 07/0 : 51[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104434:104828 [5] NCCL INFO Channel 15/0 : 53[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104436:104824 [7] NCCL INFO Channel 08/0 : 55[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104432:104827 [3] NCCL INFO Channel 08/0 : 51[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104430:104830 [1] NCCL INFO Channel 06/0 : 49[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104436:104824 [7] NCCL INFO Channel 09/0 : 55[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104432:104827 [3] NCCL INFO Channel 09/0 : 51[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104436:104824 [7] NCCL INFO Channel 10/0 : 55[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104430:104830 [1] NCCL INFO Channel 07/0 : 49[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104432:104827 [3] NCCL INFO Channel 10/0 : 51[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104430:104830 [1] NCCL INFO Channel 08/0 : 49[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104436:104824 [7] NCCL INFO Channel 11/0 : 55[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104432:104827 [3] NCCL INFO Channel 12/0 : 51[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104436:104824 [7] NCCL INFO Channel 12/0 : 55[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104430:104830 [1] NCCL INFO Channel 10/0 : 49[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104432:104827 [3] NCCL INFO Channel 13/0 : 51[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104436:104824 [7] NCCL INFO Channel 13/0 : 55[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104430:104830 [1] NCCL INFO Channel 11/0 : 49[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104432:104827 [3] NCCL INFO Channel 14/0 : 51[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104436:104824 [7] NCCL INFO Channel 14/0 : 55[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104432:104827 [3] NCCL INFO Channel 15/0 : 51[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104430:104830 [1] NCCL INFO Channel 12/0 : 49[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104430:104830 [1] NCCL INFO Channel 13/0 : 49[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104430:104830 [1] NCCL INFO Channel 14/0 : 49[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104430:104830 [1] NCCL INFO Channel 15/0 : 49[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-46:702882:703284 [2] NCCL INFO Connected NVLS tree
+ip-26-0-162-46:702882:703284 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-162-46:702882:703284 [2] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-162-46:702882:703284 [2] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-162-46:702883:703283 [3] NCCL INFO Connected NVLS tree
+ip-26-0-162-46:702883:703283 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-162-46:702883:703283 [3] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-162-46:702883:703283 [3] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-162-46:702887:703282 [7] NCCL INFO Connected NVLS tree
+ip-26-0-162-46:702887:703282 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-162-46:702887:703282 [7] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-162-46:702887:703282 [7] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-162-46:702884:703279 [4] NCCL INFO Connected NVLS tree
+ip-26-0-162-46:702884:703279 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-162-46:702884:703279 [4] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-162-46:702884:703279 [4] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-162-46:702885:703278 [5] NCCL INFO Connected NVLS tree
+ip-26-0-162-46:702885:703278 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-162-46:702885:703278 [5] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-162-46:702885:703278 [5] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-162-46:702886:703281 [6] NCCL INFO Connected NVLS tree
+ip-26-0-162-46:702886:703281 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-162-46:702886:703281 [6] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-162-46:702886:703281 [6] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-162-46:702881:703280 [1] NCCL INFO Connected NVLS tree
+ip-26-0-162-46:702881:703280 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-162-46:702881:703280 [1] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-162-46:702881:703280 [1] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-162-46:702880:703277 [0] NCCL INFO Connected NVLS tree
+ip-26-0-162-46:702880:703277 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-162-46:702880:703277 [0] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-162-46:702880:703277 [0] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-162-46:702882:703284 [2] NCCL INFO comm 0x99f1a00 rank 2 nranks 64 cudaDev 2 nvmlDev 2 busId 75000 commId 0xda4a0eda8742c6eb - Init COMPLETE
+ip-26-0-162-46:702880:703277 [0] NCCL INFO comm 0x9b71eb0 rank 0 nranks 64 cudaDev 0 nvmlDev 0 busId 53000 commId 0xda4a0eda8742c6eb - Init COMPLETE
+ip-26-0-162-46:702884:703279 [4] NCCL INFO comm 0x9142ba0 rank 4 nranks 64 cudaDev 4 nvmlDev 4 busId 97000 commId 0xda4a0eda8742c6eb - Init COMPLETE
+ip-26-0-162-46:702886:703281 [6] NCCL INFO comm 0x927c830 rank 6 nranks 64 cudaDev 6 nvmlDev 6 busId b9000 commId 0xda4a0eda8742c6eb - Init COMPLETE
+ip-26-0-162-46:702881:703280 [1] NCCL INFO comm 0x963a9a0 rank 1 nranks 64 cudaDev 1 nvmlDev 1 busId 64000 commId 0xda4a0eda8742c6eb - Init COMPLETE
+ip-26-0-162-46:702887:703282 [7] NCCL INFO comm 0x8d87da0 rank 7 nranks 64 cudaDev 7 nvmlDev 7 busId ca000 commId 0xda4a0eda8742c6eb - Init COMPLETE
+ip-26-0-162-46:702885:703278 [5] NCCL INFO comm 0x8d96ff0 rank 5 nranks 64 cudaDev 5 nvmlDev 5 busId a8000 commId 0xda4a0eda8742c6eb - Init COMPLETE
+ip-26-0-162-46:702883:703283 [3] NCCL INFO comm 0x87b1a30 rank 3 nranks 64 cudaDev 3 nvmlDev 3 busId 86000 commId 0xda4a0eda8742c6eb - Init COMPLETE
+ip-26-0-167-9:1039539:1039938 [2] NCCL INFO Connected NVLS tree
+ip-26-0-167-9:1039539:1039938 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-167-9:1039539:1039938 [2] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-167-9:1039539:1039938 [2] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-167-9:1039537:1039932 [0] NCCL INFO Connected NVLS tree
+ip-26-0-167-9:1039537:1039932 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-167-9:1039537:1039932 [0] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-167-9:1039537:1039932 [0] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-167-9:1039543:1039933 [6] NCCL INFO Connected NVLS tree
+ip-26-0-167-9:1039543:1039933 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-167-9:1039543:1039933 [6] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-167-9:1039543:1039933 [6] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-167-9:1039541:1039937 [4] NCCL INFO Connected NVLS tree
+ip-26-0-167-9:1039541:1039937 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-167-9:1039541:1039937 [4] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-167-9:1039541:1039937 [4] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-166-36:104433:104825 [4] NCCL INFO Connected NVLS tree
+ip-26-0-166-36:104433:104825 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-36:104433:104825 [4] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-166-36:104433:104825 [4] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-167-9:1039544:1039931 [7] NCCL INFO Connected NVLS tree
+ip-26-0-167-9:1039544:1039931 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-167-9:1039544:1039931 [7] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-167-9:1039544:1039931 [7] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-167-9:1039540:1039935 [3] NCCL INFO Connected NVLS tree
+ip-26-0-167-9:1039540:1039935 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-167-9:1039540:1039935 [3] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-167-9:1039540:1039935 [3] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-166-214:624949:625344 [6] NCCL INFO Connected NVLS tree
+ip-26-0-166-214:624949:625344 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-214:624949:625344 [6] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-166-214:624949:625344 [6] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-166-36:104431:104831 [2] NCCL INFO Connected NVLS tree
+ip-26-0-166-36:104431:104831 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-36:104431:104831 [2] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-166-36:104431:104831 [2] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-166-36:104429:104829 [0] NCCL INFO Connected NVLS tree
+ip-26-0-166-36:104429:104829 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-36:104429:104829 [0] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-166-36:104429:104829 [0] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-167-9:1039542:1039936 [5] NCCL INFO Connected NVLS tree
+ip-26-0-167-9:1039542:1039936 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-167-9:1039542:1039936 [5] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-167-9:1039542:1039936 [5] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-166-36:104435:104826 [6] NCCL INFO Connected NVLS tree
+ip-26-0-166-36:104435:104826 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-36:104435:104826 [6] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-166-36:104435:104826 [6] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-166-36:104436:104824 [7] NCCL INFO Connected NVLS tree
+ip-26-0-166-36:104436:104824 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-36:104436:104824 [7] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-166-36:104436:104824 [7] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-167-9:1039538:1039934 [1] NCCL INFO Connected NVLS tree
+ip-26-0-167-9:1039538:1039934 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-167-9:1039538:1039934 [1] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-167-9:1039538:1039934 [1] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-166-36:104432:104827 [3] NCCL INFO Connected NVLS tree
+ip-26-0-166-36:104432:104827 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-36:104432:104827 [3] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-166-36:104432:104827 [3] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-166-214:624945:625340 [2] NCCL INFO Connected NVLS tree
+ip-26-0-166-214:624945:625340 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-214:624945:625340 [2] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-166-214:624945:625340 [2] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-166-36:104434:104828 [5] NCCL INFO Connected NVLS tree
+ip-26-0-166-36:104434:104828 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-36:104434:104828 [5] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-166-36:104434:104828 [5] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-167-9:1039543:1039933 [6] NCCL INFO comm 0xa191260 rank 62 nranks 64 cudaDev 6 nvmlDev 6 busId b9000 commId 0xda4a0eda8742c6eb - Init COMPLETE
+ip-26-0-167-9:1039537:1039932 [0] NCCL INFO comm 0x9e4ce90 rank 56 nranks 64 cudaDev 0 nvmlDev 0 busId 53000 commId 0xda4a0eda8742c6eb - Init COMPLETE
+ip-26-0-167-9:1039544:1039931 [7] NCCL INFO comm 0x98c6e90 rank 63 nranks 64 cudaDev 7 nvmlDev 7 busId ca000 commId 0xda4a0eda8742c6eb - Init COMPLETE
+ip-26-0-167-9:1039541:1039937 [4] NCCL INFO comm 0xa6c22f0 rank 60 nranks 64 cudaDev 4 nvmlDev 4 busId 97000 commId 0xda4a0eda8742c6eb - Init COMPLETE
+ip-26-0-167-9:1039540:1039935 [3] NCCL INFO comm 0x99bd170 rank 59 nranks 64 cudaDev 3 nvmlDev 3 busId 86000 commId 0xda4a0eda8742c6eb - Init COMPLETE
+ip-26-0-167-9:1039539:1039938 [2] NCCL INFO comm 0x9341460 rank 58 nranks 64 cudaDev 2 nvmlDev 2 busId 75000 commId 0xda4a0eda8742c6eb - Init COMPLETE
+ip-26-0-167-9:1039538:1039934 [1] NCCL INFO comm 0x926a6f0 rank 57 nranks 64 cudaDev 1 nvmlDev 1 busId 64000 commId 0xda4a0eda8742c6eb - Init COMPLETE
+ip-26-0-167-9:1039542:1039936 [5] NCCL INFO comm 0x907fcb0 rank 61 nranks 64 cudaDev 5 nvmlDev 5 busId a8000 commId 0xda4a0eda8742c6eb - Init COMPLETE
+ip-26-0-166-214:624946:625337 [3] NCCL INFO Connected NVLS tree
+ip-26-0-166-214:624946:625337 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-214:624946:625337 [3] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-166-214:624946:625337 [3] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-166-214:624943:625338 [0] NCCL INFO Connected NVLS tree
+ip-26-0-166-214:624943:625338 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-214:624943:625338 [0] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-166-214:624943:625338 [0] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-166-244:897241:897635 [1] NCCL INFO Connected NVLS tree
+ip-26-0-166-244:897241:897635 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-244:897241:897635 [1] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-166-244:897241:897635 [1] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-166-36:104430:104830 [1] NCCL INFO Connected NVLS tree
+ip-26-0-166-36:104430:104830 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-36:104430:104830 [1] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-166-36:104430:104830 [1] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-166-244:897245:897638 [5] NCCL INFO Connected NVLS tree
+ip-26-0-166-244:897245:897638 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-244:897245:897638 [5] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-166-244:897245:897638 [5] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-166-125:762816:763218 [1] NCCL INFO Connected NVLS tree
+ip-26-0-166-125:762816:763218 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-125:762816:763218 [1] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-166-125:762816:763218 [1] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-166-36:104433:104825 [4] NCCL INFO comm 0xa48b0f0 rank 52 nranks 64 cudaDev 4 nvmlDev 4 busId 97000 commId 0xda4a0eda8742c6eb - Init COMPLETE
+ip-26-0-166-36:104431:104831 [2] NCCL INFO comm 0x9a41180 rank 50 nranks 64 cudaDev 2 nvmlDev 2 busId 75000 commId 0xda4a0eda8742c6eb - Init COMPLETE
+ip-26-0-166-36:104432:104827 [3] NCCL INFO comm 0x8a395f0 rank 51 nranks 64 cudaDev 3 nvmlDev 3 busId 86000 commId 0xda4a0eda8742c6eb - Init COMPLETE
+ip-26-0-166-36:104435:104826 [6] NCCL INFO comm 0x90001f0 rank 54 nranks 64 cudaDev 6 nvmlDev 6 busId b9000 commId 0xda4a0eda8742c6eb - Init COMPLETE
+ip-26-0-166-36:104436:104824 [7] NCCL INFO comm 0x99716e0 rank 55 nranks 64 cudaDev 7 nvmlDev 7 busId ca000 commId 0xda4a0eda8742c6eb - Init COMPLETE
+ip-26-0-166-36:104429:104829 [0] NCCL INFO comm 0xa69aaf0 rank 48 nranks 64 cudaDev 0 nvmlDev 0 busId 53000 commId 0xda4a0eda8742c6eb - Init COMPLETE
+ip-26-0-166-36:104430:104830 [1] NCCL INFO comm 0x9eb3ae0 rank 49 nranks 64 cudaDev 1 nvmlDev 1 busId 64000 commId 0xda4a0eda8742c6eb - Init COMPLETE
+ip-26-0-166-36:104434:104828 [5] NCCL INFO comm 0x8ca8b60 rank 53 nranks 64 cudaDev 5 nvmlDev 5 busId a8000 commId 0xda4a0eda8742c6eb - Init COMPLETE
+ip-26-0-166-244:897247:897642 [7] NCCL INFO Connected NVLS tree
+ip-26-0-166-244:897247:897642 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-244:897247:897642 [7] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-166-244:897247:897642 [7] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-165-213:9626:10025 [1] NCCL INFO Connected NVLS tree
+ip-26-0-165-213:9626:10025 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-213:9626:10025 [1] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-165-213:9626:10025 [1] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-166-214:624944:625341 [1] NCCL INFO Connected NVLS tree
+ip-26-0-166-214:624944:625341 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-214:624944:625341 [1] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-166-214:624944:625341 [1] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-166-214:624947:625339 [4] NCCL INFO Connected NVLS tree
+ip-26-0-166-214:624947:625339 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-214:624947:625339 [4] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-166-214:624947:625339 [4] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-166-214:624950:625342 [7] NCCL INFO Connected NVLS tree
+ip-26-0-166-214:624950:625342 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-214:624950:625342 [7] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-166-214:624950:625342 [7] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-165-213:9628:10023 [3] NCCL INFO Connected NVLS tree
+ip-26-0-165-213:9628:10023 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-213:9628:10023 [3] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-165-213:9628:10023 [3] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-166-244:897243:897637 [3] NCCL INFO Connected NVLS tree
+ip-26-0-166-244:897243:897637 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-244:897243:897637 [3] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-166-244:897243:897637 [3] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-165-213:9630:10024 [5] NCCL INFO Connected NVLS tree
+ip-26-0-165-213:9630:10024 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-213:9630:10024 [5] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-165-213:9630:10024 [5] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-166-214:624948:625343 [5] NCCL INFO Connected NVLS tree
+ip-26-0-166-214:624948:625343 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-214:624948:625343 [5] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-166-214:624948:625343 [5] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-165-213:9632:10026 [7] NCCL INFO Connected NVLS tree
+ip-26-0-165-213:9632:10026 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-213:9632:10026 [7] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-165-213:9632:10026 [7] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-166-214:624947:625339 [4] NCCL INFO comm 0xa228a50 rank 36 nranks 64 cudaDev 4 nvmlDev 4 busId 97000 commId 0xda4a0eda8742c6eb - Init COMPLETE
+ip-26-0-166-214:624943:625338 [0] NCCL INFO comm 0x967f900 rank 32 nranks 64 cudaDev 0 nvmlDev 0 busId 53000 commId 0xda4a0eda8742c6eb - Init COMPLETE
+ip-26-0-166-214:624946:625337 [3] NCCL INFO comm 0xa2847d0 rank 35 nranks 64 cudaDev 3 nvmlDev 3 busId 86000 commId 0xda4a0eda8742c6eb - Init COMPLETE
+ip-26-0-166-214:624944:625341 [1] NCCL INFO comm 0x8750fa0 rank 33 nranks 64 cudaDev 1 nvmlDev 1 busId 64000 commId 0xda4a0eda8742c6eb - Init COMPLETE
+ip-26-0-166-214:624950:625342 [7] NCCL INFO comm 0x8fdb6f0 rank 39 nranks 64 cudaDev 7 nvmlDev 7 busId ca000 commId 0xda4a0eda8742c6eb - Init COMPLETE
+ip-26-0-166-214:624949:625344 [6] NCCL INFO comm 0x8cb8cf0 rank 38 nranks 64 cudaDev 6 nvmlDev 6 busId b9000 commId 0xda4a0eda8742c6eb - Init COMPLETE
+ip-26-0-166-214:624948:625343 [5] NCCL INFO comm 0xa2a1a50 rank 37 nranks 64 cudaDev 5 nvmlDev 5 busId a8000 commId 0xda4a0eda8742c6eb - Init COMPLETE
+ip-26-0-166-214:624945:625340 [2] NCCL INFO comm 0x99ed060 rank 34 nranks 64 cudaDev 2 nvmlDev 2 busId 75000 commId 0xda4a0eda8742c6eb - Init COMPLETE
+ip-26-0-166-244:897242:897639 [2] NCCL INFO Connected NVLS tree
+ip-26-0-166-244:897242:897639 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-244:897242:897639 [2] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-166-244:897242:897639 [2] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-165-213:9627:10027 [2] NCCL INFO Connected NVLS tree
+ip-26-0-165-213:9627:10027 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-213:9627:10027 [2] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-165-213:9627:10027 [2] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-166-244:897244:897636 [4] NCCL INFO Connected NVLS tree
+ip-26-0-166-244:897244:897636 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-244:897244:897636 [4] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-166-244:897244:897636 [4] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-165-213:9629:10021 [4] NCCL INFO Connected NVLS tree
+ip-26-0-165-213:9629:10021 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-213:9629:10021 [4] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-165-213:9629:10021 [4] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-166-244:897246:897641 [6] NCCL INFO Connected NVLS tree
+ip-26-0-166-244:897246:897641 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-244:897246:897641 [6] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-166-244:897246:897641 [6] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-165-213:9631:10022 [6] NCCL INFO Connected NVLS tree
+ip-26-0-165-213:9631:10022 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-213:9631:10022 [6] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-165-213:9631:10022 [6] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-166-244:897240:897640 [0] NCCL INFO Connected NVLS tree
+ip-26-0-166-244:897240:897640 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-244:897240:897640 [0] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-166-244:897240:897640 [0] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-165-213:9625:10020 [0] NCCL INFO Connected NVLS tree
+ip-26-0-165-213:9625:10020 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-213:9625:10020 [0] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-165-213:9625:10020 [0] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-166-244:897240:897640 [0] NCCL INFO comm 0x95571b0 rank 40 nranks 64 cudaDev 0 nvmlDev 0 busId 53000 commId 0xda4a0eda8742c6eb - Init COMPLETE
+ip-26-0-166-244:897241:897635 [1] NCCL INFO comm 0xa3fa260 rank 41 nranks 64 cudaDev 1 nvmlDev 1 busId 64000 commId 0xda4a0eda8742c6eb - Init COMPLETE
+ip-26-0-166-244:897243:897637 [3] NCCL INFO comm 0x8c4b580 rank 43 nranks 64 cudaDev 3 nvmlDev 3 busId 86000 commId 0xda4a0eda8742c6eb - Init COMPLETE
+ip-26-0-166-244:897247:897642 [7] NCCL INFO comm 0x9be8840 rank 47 nranks 64 cudaDev 7 nvmlDev 7 busId ca000 commId 0xda4a0eda8742c6eb - Init COMPLETE
+ip-26-0-166-244:897245:897638 [5] NCCL INFO comm 0x9cf54b0 rank 45 nranks 64 cudaDev 5 nvmlDev 5 busId a8000 commId 0xda4a0eda8742c6eb - Init COMPLETE
+ip-26-0-166-244:897246:897641 [6] NCCL INFO comm 0xa665d00 rank 46 nranks 64 cudaDev 6 nvmlDev 6 busId b9000 commId 0xda4a0eda8742c6eb - Init COMPLETE
+ip-26-0-166-244:897242:897639 [2] NCCL INFO comm 0x8943e30 rank 42 nranks 64 cudaDev 2 nvmlDev 2 busId 75000 commId 0xda4a0eda8742c6eb - Init COMPLETE
+ip-26-0-166-244:897244:897636 [4] NCCL INFO comm 0x90d7750 rank 44 nranks 64 cudaDev 4 nvmlDev 4 busId 97000 commId 0xda4a0eda8742c6eb - Init COMPLETE
+ip-26-0-166-15:127297:127691 [7] NCCL INFO Connected NVLS tree
+ip-26-0-166-15:127297:127691 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-15:127297:127691 [7] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-166-15:127297:127691 [7] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-165-213:9625:10020 [0] NCCL INFO comm 0x901a8b0 rank 8 nranks 64 cudaDev 0 nvmlDev 0 busId 53000 commId 0xda4a0eda8742c6eb - Init COMPLETE
+ip-26-0-165-213:9631:10022 [6] NCCL INFO comm 0x9d025f0 rank 14 nranks 64 cudaDev 6 nvmlDev 6 busId b9000 commId 0xda4a0eda8742c6eb - Init COMPLETE
+ip-26-0-165-213:9627:10027 [2] NCCL INFO comm 0x8b55300 rank 10 nranks 64 cudaDev 2 nvmlDev 2 busId 75000 commId 0xda4a0eda8742c6eb - Init COMPLETE
+ip-26-0-165-213:9630:10024 [5] NCCL INFO comm 0x89d16b0 rank 13 nranks 64 cudaDev 5 nvmlDev 5 busId a8000 commId 0xda4a0eda8742c6eb - Init COMPLETE
+ip-26-0-165-213:9626:10025 [1] NCCL INFO comm 0x95f5af0 rank 9 nranks 64 cudaDev 1 nvmlDev 1 busId 64000 commId 0xda4a0eda8742c6eb - Init COMPLETE
+ip-26-0-165-213:9628:10023 [3] NCCL INFO comm 0x9540830 rank 11 nranks 64 cudaDev 3 nvmlDev 3 busId 86000 commId 0xda4a0eda8742c6eb - Init COMPLETE
+ip-26-0-165-213:9632:10026 [7] NCCL INFO comm 0x8f2c3f0 rank 15 nranks 64 cudaDev 7 nvmlDev 7 busId ca000 commId 0xda4a0eda8742c6eb - Init COMPLETE
+ip-26-0-165-213:9629:10021 [4] NCCL INFO comm 0x8b51550 rank 12 nranks 64 cudaDev 4 nvmlDev 4 busId 97000 commId 0xda4a0eda8742c6eb - Init COMPLETE
+ip-26-0-166-125:762822:763222 [7] NCCL INFO Connected NVLS tree
+ip-26-0-166-125:762822:763222 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-125:762822:763222 [7] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-166-125:762822:763222 [7] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-166-15:127291:127690 [1] NCCL INFO Connected NVLS tree
+ip-26-0-166-15:127291:127690 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-15:127291:127690 [1] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-166-15:127291:127690 [1] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-166-125:762818:763220 [3] NCCL INFO Connected NVLS tree
+ip-26-0-166-125:762818:763220 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-125:762818:763220 [3] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-166-125:762818:763220 [3] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-166-15:127293:127689 [3] NCCL INFO Connected NVLS tree
+ip-26-0-166-15:127293:127689 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-15:127293:127689 [3] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-166-15:127293:127689 [3] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-166-15:127295:127686 [5] NCCL INFO Connected NVLS tree
+ip-26-0-166-15:127295:127686 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-15:127295:127686 [5] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-166-15:127295:127686 [5] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-166-125:762820:763219 [5] NCCL INFO Connected NVLS tree
+ip-26-0-166-125:762820:763219 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-125:762820:763219 [5] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-166-125:762820:763219 [5] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-166-15:127292:127692 [2] NCCL INFO Connected NVLS tree
+ip-26-0-166-15:127292:127692 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-15:127292:127692 [2] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-166-15:127292:127692 [2] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-166-125:762817:763223 [2] NCCL INFO Connected NVLS tree
+ip-26-0-166-125:762817:763223 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-125:762817:763223 [2] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-166-125:762817:763223 [2] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-166-15:127296:127688 [6] NCCL INFO Connected NVLS tree
+ip-26-0-166-15:127296:127688 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-15:127296:127688 [6] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-166-15:127296:127688 [6] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-166-15:127294:127687 [4] NCCL INFO Connected NVLS tree
+ip-26-0-166-15:127294:127687 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-15:127294:127687 [4] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-166-15:127294:127687 [4] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-166-125:762821:763224 [6] NCCL INFO Connected NVLS tree
+ip-26-0-166-125:762821:763224 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-125:762821:763224 [6] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-166-125:762821:763224 [6] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-166-15:127290:127685 [0] NCCL INFO Connected NVLS tree
+ip-26-0-166-15:127290:127685 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-15:127290:127685 [0] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-166-15:127290:127685 [0] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-166-125:762815:763225 [0] NCCL INFO Connected NVLS tree
+ip-26-0-166-125:762815:763225 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-125:762815:763225 [0] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-166-125:762815:763225 [0] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-166-125:762819:763221 [4] NCCL INFO Connected NVLS tree
+ip-26-0-166-125:762819:763221 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-125:762819:763221 [4] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-166-125:762819:763221 [4] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-166-15:127295:127686 [5] NCCL INFO comm 0x8cd1ce0 rank 29 nranks 64 cudaDev 5 nvmlDev 5 busId a8000 commId 0xda4a0eda8742c6eb - Init COMPLETE
+ip-26-0-166-15:127293:127689 [3] NCCL INFO comm 0xa34b190 rank 27 nranks 64 cudaDev 3 nvmlDev 3 busId 86000 commId 0xda4a0eda8742c6eb - Init COMPLETE
+ip-26-0-166-15:127297:127691 [7] NCCL INFO comm 0x9fff530 rank 31 nranks 64 cudaDev 7 nvmlDev 7 busId ca000 commId 0xda4a0eda8742c6eb - Init COMPLETE
+ip-26-0-166-15:127296:127688 [6] NCCL INFO comm 0x98e93f0 rank 30 nranks 64 cudaDev 6 nvmlDev 6 busId b9000 commId 0xda4a0eda8742c6eb - Init COMPLETE
+ip-26-0-166-15:127291:127690 [1] NCCL INFO comm 0xa457d40 rank 25 nranks 64 cudaDev 1 nvmlDev 1 busId 64000 commId 0xda4a0eda8742c6eb - Init COMPLETE
+ip-26-0-166-15:127294:127687 [4] NCCL INFO comm 0x91e17f0 rank 28 nranks 64 cudaDev 4 nvmlDev 4 busId 97000 commId 0xda4a0eda8742c6eb - Init COMPLETE
+ip-26-0-166-15:127290:127685 [0] NCCL INFO comm 0x8bd7da0 rank 24 nranks 64 cudaDev 0 nvmlDev 0 busId 53000 commId 0xda4a0eda8742c6eb - Init COMPLETE
+ip-26-0-166-15:127292:127692 [2] NCCL INFO comm 0x94433a0 rank 26 nranks 64 cudaDev 2 nvmlDev 2 busId 75000 commId 0xda4a0eda8742c6eb - Init COMPLETE
+ip-26-0-166-125:762821:763224 [6] NCCL INFO comm 0x8ae3f70 rank 22 nranks 64 cudaDev 6 nvmlDev 6 busId b9000 commId 0xda4a0eda8742c6eb - Init COMPLETE
+ip-26-0-166-125:762819:763221 [4] NCCL INFO comm 0x961bfa0 rank 20 nranks 64 cudaDev 4 nvmlDev 4 busId 97000 commId 0xda4a0eda8742c6eb - Init COMPLETE
+ip-26-0-166-125:762817:763223 [2] NCCL INFO comm 0xa633eb0 rank 18 nranks 64 cudaDev 2 nvmlDev 2 busId 75000 commId 0xda4a0eda8742c6eb - Init COMPLETE
+ip-26-0-166-125:762815:763225 [0] NCCL INFO comm 0x88082b0 rank 16 nranks 64 cudaDev 0 nvmlDev 0 busId 53000 commId 0xda4a0eda8742c6eb - Init COMPLETE
+ip-26-0-166-125:762822:763222 [7] NCCL INFO comm 0xa21d6a0 rank 23 nranks 64 cudaDev 7 nvmlDev 7 busId ca000 commId 0xda4a0eda8742c6eb - Init COMPLETE
+ip-26-0-166-125:762820:763219 [5] NCCL INFO comm 0x953c5f0 rank 21 nranks 64 cudaDev 5 nvmlDev 5 busId a8000 commId 0xda4a0eda8742c6eb - Init COMPLETE
+ip-26-0-166-125:762818:763220 [3] NCCL INFO comm 0x89715f0 rank 19 nranks 64 cudaDev 3 nvmlDev 3 busId 86000 commId 0xda4a0eda8742c6eb - Init COMPLETE
+ip-26-0-166-125:762816:763218 [1] NCCL INFO comm 0x9836d20 rank 17 nranks 64 cudaDev 1 nvmlDev 1 busId 64000 commId 0xda4a0eda8742c6eb - Init COMPLETE
+ip-26-0-162-46:702880:703314 [0] NCCL INFO Using network Libfabric
+ip-26-0-162-46:702886:703315 [6] NCCL INFO Using network Libfabric
+ip-26-0-162-46:702884:703319 [4] NCCL INFO Using network Libfabric
+ip-26-0-162-46:702881:703320 [1] NCCL INFO Using network Libfabric
+ip-26-0-162-46:702882:703321 [2] NCCL INFO Using network Libfabric
+ip-26-0-162-46:702883:703318 [3] NCCL INFO Using network Libfabric
+ip-26-0-162-46:702887:703317 [7] NCCL INFO Using network Libfabric
+ip-26-0-162-46:702885:703316 [5] NCCL INFO Using network Libfabric
+ip-26-0-162-46:702881:703320 [1] NCCL INFO comm 0x9652340 rank 1 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0x65afe7f478768c8 - Init START
+ip-26-0-162-46:702880:703314 [0] NCCL INFO comm 0x9b86160 rank 0 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0x65afe7f478768c8 - Init START
+ip-26-0-162-46:702882:703321 [2] NCCL INFO comm 0x9a08fa0 rank 2 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0x65afe7f478768c8 - Init START
+ip-26-0-162-46:702887:703317 [7] NCCL INFO comm 0x8d9c450 rank 7 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0x65afe7f478768c8 - Init START
+ip-26-0-162-46:702886:703315 [6] NCCL INFO comm 0x9292e40 rank 6 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0x65afe7f478768c8 - Init START
+ip-26-0-162-46:702883:703318 [3] NCCL INFO comm 0x87ca1f0 rank 3 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0x65afe7f478768c8 - Init START
+ip-26-0-162-46:702885:703316 [5] NCCL INFO comm 0x8dac5b0 rank 5 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0x65afe7f478768c8 - Init START
+ip-26-0-162-46:702884:703319 [4] NCCL INFO comm 0x9156360 rank 4 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0x65afe7f478768c8 - Init START
+ip-26-0-162-46:702886:703315 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-162-46:702885:703316 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-162-46:702884:703319 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-162-46:702883:703318 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-162-46:702887:703317 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-162-46:702880:703314 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/349
+ip-26-0-162-46:702882:703321 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-162-46:702881:703320 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+NCCL version 2.18.6+cuda12.1
+ip-26-0-166-214:624943:625375 [0] NCCL INFO Using network Libfabric
+ip-26-0-166-214:624947:625378 [4] NCCL INFO Using network Libfabric
+ip-26-0-166-214:624950:625376 [7] NCCL INFO Using network Libfabric
+ip-26-0-166-214:624945:625380 [2] NCCL INFO Using network Libfabric
+ip-26-0-166-214:624944:625379 [1] NCCL INFO Using network Libfabric
+ip-26-0-166-214:624946:625377 [3] NCCL INFO Using network Libfabric
+ip-26-0-166-214:624949:625381 [6] NCCL INFO Using network Libfabric
+ip-26-0-166-214:624948:625382 [5] NCCL INFO Using network Libfabric
+ip-26-0-166-214:624949:625381 [6] NCCL INFO comm 0x8ccfda0 rank 6 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0xef8d36d72ea6a346 - Init START
+ip-26-0-166-214:624944:625379 [1] NCCL INFO comm 0x8768060 rank 1 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0xef8d36d72ea6a346 - Init START
+ip-26-0-166-214:624945:625380 [2] NCCL INFO comm 0x9a046d0 rank 2 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0xef8d36d72ea6a346 - Init START
+ip-26-0-166-214:624943:625375 [0] NCCL INFO comm 0x9698270 rank 0 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0xef8d36d72ea6a346 - Init START
+ip-26-0-166-214:624950:625376 [7] NCCL INFO comm 0x8ff2880 rank 7 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0xef8d36d72ea6a346 - Init START
+ip-26-0-166-214:624946:625377 [3] NCCL INFO comm 0xa29d720 rank 3 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0xef8d36d72ea6a346 - Init START
+ip-26-0-166-214:624948:625382 [5] NCCL INFO comm 0xa2b4710 rank 5 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0xef8d36d72ea6a346 - Init START
+ip-26-0-166-214:624947:625378 [4] NCCL INFO comm 0xa23cf00 rank 4 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0xef8d36d72ea6a346 - Init START
+ip-26-0-166-214:624943:625375 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-214:624948:625382 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-214:624950:625376 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-214:624946:625377 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-214:624944:625379 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-214:624949:625381 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-214:624945:625380 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-214:624947:625378 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+NCCL version 2.18.6+cuda12.1
+ip-26-0-166-36:104429:104861 [0] NCCL INFO Using network Libfabric
+ip-26-0-166-36:104435:104863 [6] NCCL INFO Using network Libfabric
+ip-26-0-166-36:104432:104866 [3] NCCL INFO Using network Libfabric
+ip-26-0-166-36:104436:104865 [7] NCCL INFO Using network Libfabric
+ip-26-0-166-36:104431:104862 [2] NCCL INFO Using network Libfabric
+ip-26-0-166-36:104433:104864 [4] NCCL INFO Using network Libfabric
+ip-26-0-166-36:104434:104867 [5] NCCL INFO Using network Libfabric
+ip-26-0-166-36:104430:104868 [1] NCCL INFO Using network Libfabric
+NCCL version 2.18.6+cuda12.1
+ip-26-0-166-36:104430:104868 [1] NCCL INFO comm 0x9ecbed0 rank 1 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0xd9fb77963d823d25 - Init START
+ip-26-0-166-36:104436:104865 [7] NCCL INFO comm 0x9987fb0 rank 7 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0xd9fb77963d823d25 - Init START
+ip-26-0-166-36:104429:104861 [0] NCCL INFO comm 0xa6b1af0 rank 0 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0xd9fb77963d823d25 - Init START
+ip-26-0-166-36:104434:104867 [5] NCCL INFO comm 0x8cbe880 rank 5 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0xd9fb77963d823d25 - Init START
+ip-26-0-166-36:104435:104863 [6] NCCL INFO comm 0x9017970 rank 6 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0xd9fb77963d823d25 - Init START
+ip-26-0-166-36:104433:104864 [4] NCCL INFO comm 0xa4a2c80 rank 4 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0xd9fb77963d823d25 - Init START
+ip-26-0-166-36:104431:104862 [2] NCCL INFO comm 0x9a579f0 rank 2 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0xd9fb77963d823d25 - Init START
+ip-26-0-166-36:104432:104866 [3] NCCL INFO comm 0x8a51810 rank 3 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0xd9fb77963d823d25 - Init START
+ip-26-0-166-36:104434:104867 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-36:104435:104863 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-36:104433:104864 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-36:104430:104868 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-36:104429:104861 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-36:104436:104865 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-36:104432:104866 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-36:104431:104862 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-125:762822:763260 [7] NCCL INFO Using network Libfabric
+ip-26-0-166-125:762820:763259 [5] NCCL INFO Using network Libfabric
+ip-26-0-166-125:762815:763258 [0] NCCL INFO Using network Libfabric
+ip-26-0-166-125:762821:763264 [6] NCCL INFO Using network Libfabric
+ip-26-0-166-125:762816:763262 [1] NCCL INFO Using network Libfabric
+ip-26-0-166-125:762817:763265 [2] NCCL INFO Using network Libfabric
+ip-26-0-166-125:762819:763261 [4] NCCL INFO Using network Libfabric
+ip-26-0-166-125:762818:763263 [3] NCCL INFO Using network Libfabric
+NCCL version 2.18.6+cuda12.1
+NCCL version 2.18.6+cuda12.1
+ip-26-0-166-125:762817:763265 [2] NCCL INFO comm 0xa64bcb0 rank 2 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0xe106f35b05bcd308 - Init START
+ip-26-0-166-125:762816:763262 [1] NCCL INFO comm 0x984eea0 rank 1 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0xe106f35b05bcd308 - Init START
+ip-26-0-166-125:762815:763258 [0] NCCL INFO comm 0x881f7d0 rank 0 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0xe106f35b05bcd308 - Init START
+ip-26-0-166-125:762821:763264 [6] NCCL INFO comm 0x8afb540 rank 6 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0xe106f35b05bcd308 - Init START
+ip-26-0-166-125:762822:763260 [7] NCCL INFO comm 0xa233e80 rank 7 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0xe106f35b05bcd308 - Init START
+ip-26-0-166-125:762818:763263 [3] NCCL INFO comm 0x89882d0 rank 3 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0xe106f35b05bcd308 - Init START
+ip-26-0-166-125:762820:763259 [5] NCCL INFO comm 0x95543d0 rank 5 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0xe106f35b05bcd308 - Init START
+ip-26-0-166-125:762819:763261 [4] NCCL INFO comm 0x96329a0 rank 4 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0xe106f35b05bcd308 - Init START
+ip-26-0-166-125:762817:763265 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-125:762816:763262 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-125:762815:763258 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-125:762819:763261 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-125:762820:763259 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-125:762818:763263 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-125:762821:763264 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-125:762822:763260 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-244:897240:897673 [0] NCCL INFO Using network Libfabric
+ip-26-0-165-213:9625:10059 [0] NCCL INFO Using network Libfabric
+ip-26-0-165-213:9626:10061 [1] NCCL INFO Using network Libfabric
+ip-26-0-165-213:9630:10060 [5] NCCL INFO Using network Libfabric
+ip-26-0-166-244:897241:897675 [1] NCCL INFO Using network Libfabric
+ip-26-0-166-244:897247:897677 [7] NCCL INFO Using network Libfabric
+ip-26-0-165-213:9629:10063 [4] NCCL INFO Using network Libfabric
+ip-26-0-165-213:9631:10062 [6] NCCL INFO Using network Libfabric
+ip-26-0-166-244:897244:897674 [4] NCCL INFO Using network Libfabric
+ip-26-0-165-213:9627:10064 [2] NCCL INFO Using network Libfabric
+ip-26-0-165-213:9628:10066 [3] NCCL INFO Using network Libfabric
+ip-26-0-166-244:897242:897679 [2] NCCL INFO Using network Libfabric
+ip-26-0-166-244:897243:897678 [3] NCCL INFO Using network Libfabric
+ip-26-0-166-244:897245:897676 [5] NCCL INFO Using network Libfabric
+ip-26-0-166-244:897246:897680 [6] NCCL INFO Using network Libfabric
+ip-26-0-165-213:9632:10065 [7] NCCL INFO Using network Libfabric
+ip-26-0-166-244:897245:897676 [5] NCCL INFO comm 0x9d0d3b0 rank 5 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0xc427040fef04b119 - Init START
+ip-26-0-166-244:897244:897674 [4] NCCL INFO comm 0x90f0e70 rank 4 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0xc427040fef04b119 - Init START
+ip-26-0-166-244:897241:897675 [1] NCCL INFO comm 0xa412870 rank 1 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0xc427040fef04b119 - Init START
+ip-26-0-166-244:897243:897678 [3] NCCL INFO comm 0x8c64460 rank 3 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0xc427040fef04b119 - Init START
+ip-26-0-166-244:897242:897679 [2] NCCL INFO comm 0x895b4a0 rank 2 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0xc427040fef04b119 - Init START
+ip-26-0-166-244:897240:897673 [0] NCCL INFO comm 0x956e780 rank 0 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0xc427040fef04b119 - Init START
+ip-26-0-166-244:897246:897680 [6] NCCL INFO comm 0xa67d830 rank 6 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0xc427040fef04b119 - Init START
+ip-26-0-166-244:897247:897677 [7] NCCL INFO comm 0x9c01630 rank 7 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0xc427040fef04b119 - Init START
+ip-26-0-166-244:897244:897674 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-244:897243:897678 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-244:897242:897679 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-244:897245:897676 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-244:897241:897675 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-244:897247:897677 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-244:897240:897673 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-244:897246:897680 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-165-213:9631:10062 [6] NCCL INFO comm 0x9d1a1c0 rank 6 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0x98d233044459c3d7 - Init START
+ip-26-0-165-213:9630:10060 [5] NCCL INFO comm 0x89e8c80 rank 5 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0x98d233044459c3d7 - Init START
+ip-26-0-165-213:9629:10063 [4] NCCL INFO comm 0x8b69b00 rank 4 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0x98d233044459c3d7 - Init START
+ip-26-0-165-213:9632:10065 [7] NCCL INFO comm 0x8f449d0 rank 7 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0x98d233044459c3d7 - Init START
+ip-26-0-165-213:9625:10059 [0] NCCL INFO comm 0x9033500 rank 0 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0x98d233044459c3d7 - Init START
+ip-26-0-165-213:9626:10061 [1] NCCL INFO comm 0x960b920 rank 1 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0x98d233044459c3d7 - Init START
+ip-26-0-165-213:9628:10066 [3] NCCL INFO comm 0x9557b40 rank 3 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0x98d233044459c3d7 - Init START
+ip-26-0-165-213:9627:10064 [2] NCCL INFO comm 0x8b6ebb0 rank 2 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0x98d233044459c3d7 - Init START
+ip-26-0-165-213:9631:10062 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-165-213:9632:10065 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-165-213:9630:10060 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-165-213:9625:10059 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-165-213:9629:10063 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-165-213:9626:10061 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-165-213:9628:10066 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-165-213:9627:10064 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+NCCL version 2.18.6+cuda12.1
+NCCL version 2.18.6+cuda12.1
+ip-26-0-167-9:1039537:1039969 [0] NCCL INFO Using network Libfabric
+ip-26-0-166-15:127295:127726 [5] NCCL INFO Using network Libfabric
+ip-26-0-167-9:1039544:1039972 [7] NCCL INFO Using network Libfabric
+ip-26-0-167-9:1039539:1039975 [2] NCCL INFO Using network Libfabric
+ip-26-0-166-15:127294:127725 [4] NCCL INFO Using network Libfabric
+ip-26-0-167-9:1039538:1039973 [1] NCCL INFO Using network Libfabric
+ip-26-0-167-9:1039540:1039974 [3] NCCL INFO Using network Libfabric
+ip-26-0-166-15:127297:127731 [7] NCCL INFO Using network Libfabric
+ip-26-0-166-15:127296:127727 [6] NCCL INFO Using network Libfabric
+ip-26-0-167-9:1039542:1039976 [5] NCCL INFO Using network Libfabric
+ip-26-0-167-9:1039543:1039970 [6] NCCL INFO Using network Libfabric
+ip-26-0-166-15:127292:127729 [2] NCCL INFO Using network Libfabric
+ip-26-0-166-15:127291:127730 [1] NCCL INFO Using network Libfabric
+ip-26-0-166-15:127293:127728 [3] NCCL INFO Using network Libfabric
+ip-26-0-167-9:1039541:1039971 [4] NCCL INFO Using network Libfabric
+ip-26-0-166-15:127290:127724 [0] NCCL INFO Using network Libfabric
+ip-26-0-167-9:1039539:1039975 [2] NCCL INFO comm 0x9358300 rank 2 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0xc6b7614cb294bd51 - Init START
+ip-26-0-167-9:1039541:1039971 [4] NCCL INFO comm 0xa6d9b30 rank 4 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0xc6b7614cb294bd51 - Init START
+ip-26-0-167-9:1039543:1039970 [6] NCCL INFO comm 0xa1956c0 rank 6 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0xc6b7614cb294bd51 - Init START
+ip-26-0-167-9:1039540:1039974 [3] NCCL INFO comm 0x99d3460 rank 3 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0xc6b7614cb294bd51 - Init START
+ip-26-0-167-9:1039544:1039972 [7] NCCL INFO comm 0x98df560 rank 7 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0xc6b7614cb294bd51 - Init START
+ip-26-0-167-9:1039538:1039973 [1] NCCL INFO comm 0x9283370 rank 1 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0xc6b7614cb294bd51 - Init START
+ip-26-0-167-9:1039542:1039976 [5] NCCL INFO comm 0x9095cf0 rank 5 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0xc6b7614cb294bd51 - Init START
+ip-26-0-167-9:1039537:1039969 [0] NCCL INFO comm 0x9e653d0 rank 0 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0xc6b7614cb294bd51 - Init START
+ip-26-0-167-9:1039537:1039969 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-167-9:1039544:1039972 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-167-9:1039540:1039974 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-167-9:1039538:1039973 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-167-9:1039539:1039975 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-167-9:1039541:1039971 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-167-9:1039543:1039970 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-167-9:1039542:1039976 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-15:127292:127729 [2] NCCL INFO comm 0x945a720 rank 2 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0xa88f4e7d7abb936c - Init START
+ip-26-0-166-15:127290:127724 [0] NCCL INFO comm 0x8befd90 rank 0 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0xa88f4e7d7abb936c - Init START
+ip-26-0-166-15:127291:127730 [1] NCCL INFO comm 0xa46fbd0 rank 1 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0xa88f4e7d7abb936c - Init START
+ip-26-0-166-15:127295:127726 [5] NCCL INFO comm 0x8ce9f30 rank 5 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0xa88f4e7d7abb936c - Init START
+ip-26-0-166-15:127293:127728 [3] NCCL INFO comm 0xa35de50 rank 3 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0xa88f4e7d7abb936c - Init START
+ip-26-0-166-15:127294:127725 [4] NCCL INFO comm 0x91f9c40 rank 4 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0xa88f4e7d7abb936c - Init START
+ip-26-0-166-15:127297:127731 [7] NCCL INFO comm 0xa0163c0 rank 7 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0xa88f4e7d7abb936c - Init START
+ip-26-0-166-15:127296:127727 [6] NCCL INFO comm 0x9900580 rank 6 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0xa88f4e7d7abb936c - Init START
+ip-26-0-166-15:127295:127726 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-15:127293:127728 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-15:127296:127727 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-15:127297:127731 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-15:127294:127725 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-15:127292:127729 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-15:127290:127724 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-15:127291:127730 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-125:762817:763265 [2] NCCL INFO Setting affinity for GPU 2 to 0fff,ffffffff
+ip-26-0-166-125:762817:763265 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-166-36:104436:104865 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-166-36:104436:104865 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-166-244:897240:897673 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-166-244:897240:897673 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-162-46:702882:703321 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-162-46:702882:703321 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-166-36:104431:104862 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-166-36:104431:104862 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-166-125:762815:763258 [0] NCCL INFO Setting affinity for GPU 0 to 0fff,ffffffff
+ip-26-0-166-125:762815:763258 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-166-214:624949:625381 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-166-214:624949:625381 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-166-36:104433:104864 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-166-36:104433:104864 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-167-9:1039537:1039969 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-167-9:1039537:1039969 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-167-9:1039539:1039975 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-167-9:1039539:1039975 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-166-125:762820:763259 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,fff00000,00000000
+ip-26-0-166-125:762820:763259 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-166-125:762816:763262 [1] NCCL INFO Setting affinity for GPU 1 to 0fff,ffffffff
+ip-26-0-166-36:104429:104861 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-166-36:104429:104861 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-166-125:762816:763262 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-167-9:1039544:1039972 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-167-9:1039544:1039972 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-166-36:104434:104867 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-166-36:104434:104867 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-162-46:702887:703317 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-162-46:702887:703317 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-165-213:9627:10064 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-165-213:9627:10064 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-166-36:104432:104866 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-166-36:104432:104866 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-166-15:127294:127725 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-166-15:127294:127725 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-167-9:1039543:1039970 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-167-9:1039543:1039970 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-166-15:127295:127726 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-166-15:127295:127726 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-166-214:624944:625379 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-166-214:624944:625379 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-166-214:624950:625376 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-166-214:624950:625376 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-166-15:127293:127728 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-166-15:127293:127728 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-166-36:104430:104868 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-166-36:104430:104868 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-166-214:624946:625377 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-166-214:624946:625377 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-166-36:104435:104863 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-166-36:104435:104863 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-166-36:104435:104863 [6] NCCL INFO Trees [0] 7/-1/-1->6->5 [1] 7/-1/-1->6->5 [2] 7/-1/-1->6->5 [3] 7/-1/-1->6->5 [4] 7/-1/-1->6->5 [5] 7/-1/-1->6->5 [6] 7/-1/-1->6->5 [7] 7/-1/-1->6->5 [8] 7/-1/-1->6->5 [9] 7/-1/-1->6->5 [10] 7/-1/-1->6->5 [11] 7/-1/-1->6->5 [12] 7/-1/-1->6->5 [13] 7/-1/-1->6->5 [14] 7/-1/-1->6->5 [15] 7/-1/-1->6->5 [16] 7/-1/-1->6->5 [17] 7/-1/-1->6->5 [18] 7/-1/-1->6->5 [19] 7/-1/-1->6->5 [20] 7/-1/-1->6->5 [21] 7/-1/-1->6->5 [22] 7/-1/-1->6->5 [23] 7/-1/-1->6->5
+ip-26-0-166-36:104435:104863 [6] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-166-36:104434:104867 [5] NCCL INFO Trees [0] 6/-1/-1->5->4 [1] 6/-1/-1->5->4 [2] 6/-1/-1->5->4 [3] 6/-1/-1->5->4 [4] 6/-1/-1->5->4 [5] 6/-1/-1->5->4 [6] 6/-1/-1->5->4 [7] 6/-1/-1->5->4 [8] 6/-1/-1->5->4 [9] 6/-1/-1->5->4 [10] 6/-1/-1->5->4 [11] 6/-1/-1->5->4 [12] 6/-1/-1->5->4 [13] 6/-1/-1->5->4 [14] 6/-1/-1->5->4 [15] 6/-1/-1->5->4 [16] 6/-1/-1->5->4 [17] 6/-1/-1->5->4 [18] 6/-1/-1->5->4 [19] 6/-1/-1->5->4 [20] 6/-1/-1->5->4 [21] 6/-1/-1->5->4 [22] 6/-1/-1->5->4 [23] 6/-1/-1->5->4
+ip-26-0-166-36:104434:104867 [5] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-166-36:104429:104861 [0] NCCL INFO Channel 00/24 :    0   1   2   3   4   5   6   7
+ip-26-0-166-36:104429:104861 [0] NCCL INFO Channel 01/24 :    0   1   2   3   4   5   6   7
+ip-26-0-166-36:104429:104861 [0] NCCL INFO Channel 02/24 :    0   1   2   3   4   5   6   7
+ip-26-0-166-36:104431:104862 [2] NCCL INFO Trees [0] 3/-1/-1->2->1 [1] 3/-1/-1->2->1 [2] 3/-1/-1->2->1 [3] 3/-1/-1->2->1 [4] 3/-1/-1->2->1 [5] 3/-1/-1->2->1 [6] 3/-1/-1->2->1 [7] 3/-1/-1->2->1 [8] 3/-1/-1->2->1 [9] 3/-1/-1->2->1 [10] 3/-1/-1->2->1 [11] 3/-1/-1->2->1 [12] 3/-1/-1->2->1 [13] 3/-1/-1->2->1 [14] 3/-1/-1->2->1 [15] 3/-1/-1->2->1 [16] 3/-1/-1->2->1 [17] 3/-1/-1->2->1 [18] 3/-1/-1->2->1 [19] 3/-1/-1->2->1 [20] 3/-1/-1->2->1 [21] 3/-1/-1->2->1 [22] 3/-1/-1->2->1 [23] 3/-1/-1->2->1
+ip-26-0-166-36:104429:104861 [0] NCCL INFO Channel 03/24 :    0   1   2   3   4   5   6   7
+ip-26-0-166-36:104432:104866 [3] NCCL INFO Trees [0] 4/-1/-1->3->2 [1] 4/-1/-1->3->2 [2] 4/-1/-1->3->2 [3] 4/-1/-1->3->2 [4] 4/-1/-1->3->2 [5] 4/-1/-1->3->2 [6] 4/-1/-1->3->2 [7] 4/-1/-1->3->2 [8] 4/-1/-1->3->2 [9] 4/-1/-1->3->2 [10] 4/-1/-1->3->2 [11] 4/-1/-1->3->2 [12] 4/-1/-1->3->2 [13] 4/-1/-1->3->2 [14] 4/-1/-1->3->2 [15] 4/-1/-1->3->2 [16] 4/-1/-1->3->2 [17] 4/-1/-1->3->2 [18] 4/-1/-1->3->2 [19] 4/-1/-1->3->2 [20] 4/-1/-1->3->2 [21] 4/-1/-1->3->2 [22] 4/-1/-1->3->2 [23] 4/-1/-1->3->2
+ip-26-0-166-36:104436:104865 [7] NCCL INFO Trees [0] -1/-1/-1->7->6 [1] -1/-1/-1->7->6 [2] -1/-1/-1->7->6 [3] -1/-1/-1->7->6 [4] -1/-1/-1->7->6 [5] -1/-1/-1->7->6 [6] -1/-1/-1->7->6 [7] -1/-1/-1->7->6 [8] -1/-1/-1->7->6 [9] -1/-1/-1->7->6 [10] -1/-1/-1->7->6 [11] -1/-1/-1->7->6 [12] -1/-1/-1->7->6 [13] -1/-1/-1->7->6 [14] -1/-1/-1->7->6 [15] -1/-1/-1->7->6 [16] -1/-1/-1->7->6 [17] -1/-1/-1->7->6 [18] -1/-1/-1->7->6 [19] -1/-1/-1->7->6 [20] -1/-1/-1->7->6 [21] -1/-1/-1->7->6 [22] -1/-1/-1->7->6 [23] -1/-1/-1->7->6
+ip-26-0-166-36:104429:104861 [0] NCCL INFO Channel 04/24 :    0   1   2   3   4   5   6   7
+ip-26-0-166-36:104431:104862 [2] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-166-36:104433:104864 [4] NCCL INFO Trees [0] 5/-1/-1->4->3 [1] 5/-1/-1->4->3 [2] 5/-1/-1->4->3 [3] 5/-1/-1->4->3 [4] 5/-1/-1->4->3 [5] 5/-1/-1->4->3 [6] 5/-1/-1->4->3 [7] 5/-1/-1->4->3 [8] 5/-1/-1->4->3 [9] 5/-1/-1->4->3 [10] 5/-1/-1->4->3 [11] 5/-1/-1->4->3 [12] 5/-1/-1->4->3 [13] 5/-1/-1->4->3 [14] 5/-1/-1->4->3 [15] 5/-1/-1->4->3 [16] 5/-1/-1->4->3 [17] 5/-1/-1->4->3 [18] 5/-1/-1->4->3 [19] 5/-1/-1->4->3 [20] 5/-1/-1->4->3 [21] 5/-1/-1->4->3 [22] 5/-1/-1->4->3 [23] 5/-1/-1->4->3
+ip-26-0-166-36:104432:104866 [3] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-166-36:104430:104868 [1] NCCL INFO Trees [0] 2/-1/-1->1->0 [1] 2/-1/-1->1->0 [2] 2/-1/-1->1->0 [3] 2/-1/-1->1->0 [4] 2/-1/-1->1->0 [5] 2/-1/-1->1->0 [6] 2/-1/-1->1->0 [7] 2/-1/-1->1->0 [8] 2/-1/-1->1->0 [9] 2/-1/-1->1->0 [10] 2/-1/-1->1->0 [11] 2/-1/-1->1->0 [12] 2/-1/-1->1->0 [13] 2/-1/-1->1->0 [14] 2/-1/-1->1->0 [15] 2/-1/-1->1->0 [16] 2/-1/-1->1->0 [17] 2/-1/-1->1->0 [18] 2/-1/-1->1->0 [19] 2/-1/-1->1->0 [20] 2/-1/-1->1->0 [21] 2/-1/-1->1->0 [22] 2/-1/-1->1->0 [23] 2/-1/-1->1->0
+ip-26-0-166-36:104429:104861 [0] NCCL INFO Channel 05/24 :    0   1   2   3   4   5   6   7
+ip-26-0-166-36:104436:104865 [7] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-166-36:104433:104864 [4] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-166-36:104429:104861 [0] NCCL INFO Channel 06/24 :    0   1   2   3   4   5   6   7
+ip-26-0-166-36:104430:104868 [1] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-166-36:104429:104861 [0] NCCL INFO Channel 07/24 :    0   1   2   3   4   5   6   7
+ip-26-0-166-36:104429:104861 [0] NCCL INFO Channel 08/24 :    0   1   2   3   4   5   6   7
+ip-26-0-166-36:104429:104861 [0] NCCL INFO Channel 09/24 :    0   1   2   3   4   5   6   7
+ip-26-0-166-36:104429:104861 [0] NCCL INFO Channel 10/24 :    0   1   2   3   4   5   6   7
+ip-26-0-166-36:104429:104861 [0] NCCL INFO Channel 11/24 :    0   1   2   3   4   5   6   7
+ip-26-0-166-36:104429:104861 [0] NCCL INFO Channel 12/24 :    0   1   2   3   4   5   6   7
+ip-26-0-166-36:104429:104861 [0] NCCL INFO Channel 13/24 :    0   1   2   3   4   5   6   7
+ip-26-0-166-36:104429:104861 [0] NCCL INFO Channel 14/24 :    0   1   2   3   4   5   6   7
+ip-26-0-166-36:104429:104861 [0] NCCL INFO Channel 15/24 :    0   1   2   3   4   5   6   7
+ip-26-0-166-36:104429:104861 [0] NCCL INFO Channel 16/24 :    0   1   2   3   4   5   6   7
+ip-26-0-166-36:104429:104861 [0] NCCL INFO Channel 17/24 :    0   1   2   3   4   5   6   7
+ip-26-0-166-36:104429:104861 [0] NCCL INFO Channel 18/24 :    0   1   2   3   4   5   6   7
+ip-26-0-166-36:104429:104861 [0] NCCL INFO Channel 19/24 :    0   1   2   3   4   5   6   7
+ip-26-0-166-36:104429:104861 [0] NCCL INFO Channel 20/24 :    0   1   2   3   4   5   6   7
+ip-26-0-166-36:104429:104861 [0] NCCL INFO Channel 21/24 :    0   1   2   3   4   5   6   7
+ip-26-0-166-36:104429:104861 [0] NCCL INFO Channel 22/24 :    0   1   2   3   4   5   6   7
+ip-26-0-166-36:104429:104861 [0] NCCL INFO Channel 23/24 :    0   1   2   3   4   5   6   7
+ip-26-0-166-36:104429:104861 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] 1/-1/-1->0->-1 [7] 1/-1/-1->0->-1 [8] 1/-1/-1->0->-1 [9] 1/-1/-1->0->-1 [10] 1/-1/-1->0->-1 [11] 1/-1/-1->0->-1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] 1/-1/-1->0->-1 [19] 1/-1/-1->0->-1 [20] 1/-1/-1->0->-1 [21] 1/-1/-1->0->-1 [22] 1/-1/-1->0->-1 [23] 1/-1/-1->0->-1
+ip-26-0-166-36:104429:104861 [0] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-166-214:624947:625378 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-166-214:624947:625378 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-166-214:624945:625380 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-166-214:624945:625380 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-166-244:897246:897680 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-166-244:897246:897680 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-166-214:624948:625382 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-166-214:624948:625382 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-166-214:624943:625375 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-166-214:624943:625375 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-166-214:624943:625375 [0] NCCL INFO Channel 00/24 :    0   1   2   3   4   5   6   7
+ip-26-0-166-214:624943:625375 [0] NCCL INFO Channel 01/24 :    0   1   2   3   4   5   6   7
+ip-26-0-166-214:624943:625375 [0] NCCL INFO Channel 02/24 :    0   1   2   3   4   5   6   7
+ip-26-0-166-214:624943:625375 [0] NCCL INFO Channel 03/24 :    0   1   2   3   4   5   6   7
+ip-26-0-166-214:624943:625375 [0] NCCL INFO Channel 04/24 :    0   1   2   3   4   5   6   7
+ip-26-0-166-214:624943:625375 [0] NCCL INFO Channel 05/24 :    0   1   2   3   4   5   6   7
+ip-26-0-166-214:624943:625375 [0] NCCL INFO Channel 06/24 :    0   1   2   3   4   5   6   7
+ip-26-0-166-214:624943:625375 [0] NCCL INFO Channel 07/24 :    0   1   2   3   4   5   6   7
+ip-26-0-166-214:624943:625375 [0] NCCL INFO Channel 08/24 :    0   1   2   3   4   5   6   7
+ip-26-0-166-214:624943:625375 [0] NCCL INFO Channel 09/24 :    0   1   2   3   4   5   6   7
+ip-26-0-166-214:624943:625375 [0] NCCL INFO Channel 10/24 :    0   1   2   3   4   5   6   7
+ip-26-0-166-214:624943:625375 [0] NCCL INFO Channel 11/24 :    0   1   2   3   4   5   6   7
+ip-26-0-166-214:624944:625379 [1] NCCL INFO Trees [0] 2/-1/-1->1->0 [1] 2/-1/-1->1->0 [2] 2/-1/-1->1->0 [3] 2/-1/-1->1->0 [4] 2/-1/-1->1->0 [5] 2/-1/-1->1->0 [6] 2/-1/-1->1->0 [7] 2/-1/-1->1->0 [8] 2/-1/-1->1->0 [9] 2/-1/-1->1->0 [10] 2/-1/-1->1->0 [11] 2/-1/-1->1->0 [12] 2/-1/-1->1->0 [13] 2/-1/-1->1->0 [14] 2/-1/-1->1->0 [15] 2/-1/-1->1->0 [16] 2/-1/-1->1->0 [17] 2/-1/-1->1->0 [18] 2/-1/-1->1->0 [19] 2/-1/-1->1->0 [20] 2/-1/-1->1->0 [21] 2/-1/-1->1->0 [22] 2/-1/-1->1->0 [23] 2/-1/-1->1->0
+ip-26-0-166-214:624943:625375 [0] NCCL INFO Channel 12/24 :    0   1   2   3   4   5   6   7
+ip-26-0-166-214:624943:625375 [0] NCCL INFO Channel 13/24 :    0   1   2   3   4   5   6   7
+ip-26-0-166-214:624946:625377 [3] NCCL INFO Trees [0] 4/-1/-1->3->2 [1] 4/-1/-1->3->2 [2] 4/-1/-1->3->2 [3] 4/-1/-1->3->2 [4] 4/-1/-1->3->2 [5] 4/-1/-1->3->2 [6] 4/-1/-1->3->2 [7] 4/-1/-1->3->2 [8] 4/-1/-1->3->2 [9] 4/-1/-1->3->2 [10] 4/-1/-1->3->2 [11] 4/-1/-1->3->2 [12] 4/-1/-1->3->2 [13] 4/-1/-1->3->2 [14] 4/-1/-1->3->2 [15] 4/-1/-1->3->2 [16] 4/-1/-1->3->2 [17] 4/-1/-1->3->2 [18] 4/-1/-1->3->2 [19] 4/-1/-1->3->2 [20] 4/-1/-1->3->2 [21] 4/-1/-1->3->2 [22] 4/-1/-1->3->2 [23] 4/-1/-1->3->2
+ip-26-0-166-214:624945:625380 [2] NCCL INFO Trees [0] 3/-1/-1->2->1 [1] 3/-1/-1->2->1 [2] 3/-1/-1->2->1 [3] 3/-1/-1->2->1 [4] 3/-1/-1->2->1 [5] 3/-1/-1->2->1 [6] 3/-1/-1->2->1 [7] 3/-1/-1->2->1 [8] 3/-1/-1->2->1 [9] 3/-1/-1->2->1 [10] 3/-1/-1->2->1 [11] 3/-1/-1->2->1 [12] 3/-1/-1->2->1 [13] 3/-1/-1->2->1 [14] 3/-1/-1->2->1 [15] 3/-1/-1->2->1 [16] 3/-1/-1->2->1 [17] 3/-1/-1->2->1 [18] 3/-1/-1->2->1 [19] 3/-1/-1->2->1 [20] 3/-1/-1->2->1 [21] 3/-1/-1->2->1 [22] 3/-1/-1->2->1 [23] 3/-1/-1->2->1
+ip-26-0-166-214:624944:625379 [1] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-166-214:624943:625375 [0] NCCL INFO Channel 14/24 :    0   1   2   3   4   5   6   7
+ip-26-0-166-214:624947:625378 [4] NCCL INFO Trees [0] 5/-1/-1->4->3 [1] 5/-1/-1->4->3 [2] 5/-1/-1->4->3 [3] 5/-1/-1->4->3 [4] 5/-1/-1->4->3 [5] 5/-1/-1->4->3 [6] 5/-1/-1->4->3 [7] 5/-1/-1->4->3 [8] 5/-1/-1->4->3 [9] 5/-1/-1->4->3 [10] 5/-1/-1->4->3 [11] 5/-1/-1->4->3 [12] 5/-1/-1->4->3 [13] 5/-1/-1->4->3 [14] 5/-1/-1->4->3 [15] 5/-1/-1->4->3 [16] 5/-1/-1->4->3 [17] 5/-1/-1->4->3 [18] 5/-1/-1->4->3 [19] 5/-1/-1->4->3 [20] 5/-1/-1->4->3 [21] 5/-1/-1->4->3 [22] 5/-1/-1->4->3 [23] 5/-1/-1->4->3
+ip-26-0-166-214:624950:625376 [7] NCCL INFO Trees [0] -1/-1/-1->7->6 [1] -1/-1/-1->7->6 [2] -1/-1/-1->7->6 [3] -1/-1/-1->7->6 [4] -1/-1/-1->7->6 [5] -1/-1/-1->7->6 [6] -1/-1/-1->7->6 [7] -1/-1/-1->7->6 [8] -1/-1/-1->7->6 [9] -1/-1/-1->7->6 [10] -1/-1/-1->7->6 [11] -1/-1/-1->7->6 [12] -1/-1/-1->7->6 [13] -1/-1/-1->7->6 [14] -1/-1/-1->7->6 [15] -1/-1/-1->7->6 [16] -1/-1/-1->7->6 [17] -1/-1/-1->7->6 [18] -1/-1/-1->7->6 [19] -1/-1/-1->7->6 [20] -1/-1/-1->7->6 [21] -1/-1/-1->7->6 [22] -1/-1/-1->7->6 [23] -1/-1/-1->7->6
+ip-26-0-166-214:624948:625382 [5] NCCL INFO Trees [0] 6/-1/-1->5->4 [1] 6/-1/-1->5->4 [2] 6/-1/-1->5->4 [3] 6/-1/-1->5->4 [4] 6/-1/-1->5->4 [5] 6/-1/-1->5->4 [6] 6/-1/-1->5->4 [7] 6/-1/-1->5->4 [8] 6/-1/-1->5->4 [9] 6/-1/-1->5->4 [10] 6/-1/-1->5->4 [11] 6/-1/-1->5->4 [12] 6/-1/-1->5->4 [13] 6/-1/-1->5->4 [14] 6/-1/-1->5->4 [15] 6/-1/-1->5->4 [16] 6/-1/-1->5->4 [17] 6/-1/-1->5->4 [18] 6/-1/-1->5->4 [19] 6/-1/-1->5->4 [20] 6/-1/-1->5->4 [21] 6/-1/-1->5->4 [22] 6/-1/-1->5->4 [23] 6/-1/-1->5->4
+ip-26-0-166-214:624943:625375 [0] NCCL INFO Channel 15/24 :    0   1   2   3   4   5   6   7
+ip-26-0-166-214:624946:625377 [3] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-166-214:624945:625380 [2] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-166-214:624950:625376 [7] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-166-214:624947:625378 [4] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-166-214:624948:625382 [5] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-166-214:624943:625375 [0] NCCL INFO Channel 16/24 :    0   1   2   3   4   5   6   7
+ip-26-0-166-214:624943:625375 [0] NCCL INFO Channel 17/24 :    0   1   2   3   4   5   6   7
+ip-26-0-166-214:624943:625375 [0] NCCL INFO Channel 18/24 :    0   1   2   3   4   5   6   7
+ip-26-0-166-214:624943:625375 [0] NCCL INFO Channel 19/24 :    0   1   2   3   4   5   6   7
+ip-26-0-166-214:624943:625375 [0] NCCL INFO Channel 20/24 :    0   1   2   3   4   5   6   7
+ip-26-0-166-214:624949:625381 [6] NCCL INFO Trees [0] 7/-1/-1->6->5 [1] 7/-1/-1->6->5 [2] 7/-1/-1->6->5 [3] 7/-1/-1->6->5 [4] 7/-1/-1->6->5 [5] 7/-1/-1->6->5 [6] 7/-1/-1->6->5 [7] 7/-1/-1->6->5 [8] 7/-1/-1->6->5 [9] 7/-1/-1->6->5 [10] 7/-1/-1->6->5 [11] 7/-1/-1->6->5 [12] 7/-1/-1->6->5 [13] 7/-1/-1->6->5 [14] 7/-1/-1->6->5 [15] 7/-1/-1->6->5 [16] 7/-1/-1->6->5 [17] 7/-1/-1->6->5 [18] 7/-1/-1->6->5 [19] 7/-1/-1->6->5 [20] 7/-1/-1->6->5 [21] 7/-1/-1->6->5 [22] 7/-1/-1->6->5 [23] 7/-1/-1->6->5
+ip-26-0-166-214:624943:625375 [0] NCCL INFO Channel 21/24 :    0   1   2   3   4   5   6   7
+ip-26-0-166-214:624943:625375 [0] NCCL INFO Channel 22/24 :    0   1   2   3   4   5   6   7
+ip-26-0-166-214:624943:625375 [0] NCCL INFO Channel 23/24 :    0   1   2   3   4   5   6   7
+ip-26-0-166-214:624943:625375 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] 1/-1/-1->0->-1 [7] 1/-1/-1->0->-1 [8] 1/-1/-1->0->-1 [9] 1/-1/-1->0->-1 [10] 1/-1/-1->0->-1 [11] 1/-1/-1->0->-1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] 1/-1/-1->0->-1 [19] 1/-1/-1->0->-1 [20] 1/-1/-1->0->-1 [21] 1/-1/-1->0->-1 [22] 1/-1/-1->0->-1 [23] 1/-1/-1->0->-1
+ip-26-0-166-214:624943:625375 [0] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-166-214:624949:625381 [6] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-167-9:1039540:1039974 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-167-9:1039540:1039974 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-166-125:762819:763261 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,fff00000,00000000
+ip-26-0-166-125:762819:763261 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-162-46:702880:703314 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-162-46:702880:703314 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-162-46:702881:703320 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-162-46:702885:703316 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-162-46:702885:703316 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-166-125:762822:763260 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,fff00000,00000000
+ip-26-0-166-125:762822:763260 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-162-46:702881:703320 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-166-214:624950:625376 [7] NCCL INFO Channel 00/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624950:625376 [7] NCCL INFO Channel 01/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762818:763263 [3] NCCL INFO Setting affinity for GPU 3 to 0fff,ffffffff
+ip-26-0-166-125:762818:763263 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-166-36:104434:104867 [5] NCCL INFO Channel 00/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-166-36:104434:104867 [5] NCCL INFO Channel 01/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-166-36:104436:104865 [7] NCCL INFO Channel 00/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-36:104431:104862 [2] NCCL INFO Channel 00/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-36:104433:104864 [4] NCCL INFO Channel 00/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-166-36:104436:104865 [7] NCCL INFO Channel 01/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-36:104430:104868 [1] NCCL INFO Channel 00/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-166-36:104431:104862 [2] NCCL INFO Channel 01/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-36:104433:104864 [4] NCCL INFO Channel 01/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-166-36:104436:104865 [7] NCCL INFO Channel 02/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-36:104430:104868 [1] NCCL INFO Channel 01/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-166-36:104431:104862 [2] NCCL INFO Channel 02/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-36:104433:104864 [4] NCCL INFO Channel 02/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-166-36:104434:104867 [5] NCCL INFO Channel 02/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-166-36:104436:104865 [7] NCCL INFO Channel 03/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-36:104430:104868 [1] NCCL INFO Channel 02/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-166-36:104431:104862 [2] NCCL INFO Channel 03/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-36:104433:104864 [4] NCCL INFO Channel 03/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-166-36:104432:104866 [3] NCCL INFO Channel 00/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-36:104434:104867 [5] NCCL INFO Channel 03/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-166-36:104436:104865 [7] NCCL INFO Channel 04/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-36:104430:104868 [1] NCCL INFO Channel 03/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-166-36:104429:104861 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-36:104433:104864 [4] NCCL INFO Channel 04/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-166-36:104431:104862 [2] NCCL INFO Channel 04/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-36:104432:104866 [3] NCCL INFO Channel 01/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-36:104434:104867 [5] NCCL INFO Channel 04/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-166-36:104430:104868 [1] NCCL INFO Channel 04/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-166-36:104435:104863 [6] NCCL INFO Channel 00/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-36:104436:104865 [7] NCCL INFO Channel 05/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-36:104429:104861 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-36:104433:104864 [4] NCCL INFO Channel 05/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-166-36:104431:104862 [2] NCCL INFO Channel 05/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-36:104432:104866 [3] NCCL INFO Channel 02/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-36:104434:104867 [5] NCCL INFO Channel 05/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-166-36:104430:104868 [1] NCCL INFO Channel 05/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-166-36:104435:104863 [6] NCCL INFO Channel 01/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-36:104436:104865 [7] NCCL INFO Channel 06/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-36:104429:104861 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-36:104433:104864 [4] NCCL INFO Channel 06/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-166-36:104431:104862 [2] NCCL INFO Channel 06/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-36:104432:104866 [3] NCCL INFO Channel 03/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-36:104430:104868 [1] NCCL INFO Channel 06/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-166-36:104434:104867 [5] NCCL INFO Channel 06/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-166-36:104435:104863 [6] NCCL INFO Channel 02/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-36:104436:104865 [7] NCCL INFO Channel 07/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-36:104429:104861 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-214:624947:625378 [4] NCCL INFO Channel 00/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-166-36:104433:104864 [4] NCCL INFO Channel 07/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-166-36:104431:104862 [2] NCCL INFO Channel 07/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-36:104430:104868 [1] NCCL INFO Channel 07/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-166-36:104434:104867 [5] NCCL INFO Channel 07/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-166-36:104436:104865 [7] NCCL INFO Channel 08/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-36:104432:104866 [3] NCCL INFO Channel 04/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-36:104435:104863 [6] NCCL INFO Channel 03/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-214:624945:625380 [2] NCCL INFO Channel 00/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-214:624950:625376 [7] NCCL INFO Channel 02/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624947:625378 [4] NCCL INFO Channel 01/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-162-46:702884:703319 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-162-46:702884:703319 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-166-214:624945:625380 [2] NCCL INFO Channel 01/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-214:624950:625376 [7] NCCL INFO Channel 03/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762821:763264 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,fff00000,00000000
+ip-26-0-166-125:762821:763264 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-166-125:762821:763264 [6] NCCL INFO Trees [0] 7/-1/-1->6->5 [1] 7/-1/-1->6->5 [2] 7/-1/-1->6->5 [3] 7/-1/-1->6->5 [4] 7/-1/-1->6->5 [5] 7/-1/-1->6->5 [6] 7/-1/-1->6->5 [7] 7/-1/-1->6->5 [8] 7/-1/-1->6->5 [9] 7/-1/-1->6->5 [10] 7/-1/-1->6->5 [11] 7/-1/-1->6->5 [12] 7/-1/-1->6->5 [13] 7/-1/-1->6->5 [14] 7/-1/-1->6->5 [15] 7/-1/-1->6->5 [16] 7/-1/-1->6->5 [17] 7/-1/-1->6->5 [18] 7/-1/-1->6->5 [19] 7/-1/-1->6->5 [20] 7/-1/-1->6->5 [21] 7/-1/-1->6->5 [22] 7/-1/-1->6->5 [23] 7/-1/-1->6->5
+ip-26-0-166-125:762821:763264 [6] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-166-125:762822:763260 [7] NCCL INFO Trees [0] -1/-1/-1->7->6 [1] -1/-1/-1->7->6 [2] -1/-1/-1->7->6 [3] -1/-1/-1->7->6 [4] -1/-1/-1->7->6 [5] -1/-1/-1->7->6 [6] -1/-1/-1->7->6 [7] -1/-1/-1->7->6 [8] -1/-1/-1->7->6 [9] -1/-1/-1->7->6 [10] -1/-1/-1->7->6 [11] -1/-1/-1->7->6 [12] -1/-1/-1->7->6 [13] -1/-1/-1->7->6 [14] -1/-1/-1->7->6 [15] -1/-1/-1->7->6 [16] -1/-1/-1->7->6 [17] -1/-1/-1->7->6 [18] -1/-1/-1->7->6 [19] -1/-1/-1->7->6 [20] -1/-1/-1->7->6 [21] -1/-1/-1->7->6 [22] -1/-1/-1->7->6 [23] -1/-1/-1->7->6
+ip-26-0-166-125:762822:763260 [7] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-166-125:762820:763259 [5] NCCL INFO Trees [0] 6/-1/-1->5->4 [1] 6/-1/-1->5->4 [2] 6/-1/-1->5->4 [3] 6/-1/-1->5->4 [4] 6/-1/-1->5->4 [5] 6/-1/-1->5->4 [6] 6/-1/-1->5->4 [7] 6/-1/-1->5->4 [8] 6/-1/-1->5->4 [9] 6/-1/-1->5->4 [10] 6/-1/-1->5->4 [11] 6/-1/-1->5->4 [12] 6/-1/-1->5->4 [13] 6/-1/-1->5->4 [14] 6/-1/-1->5->4 [15] 6/-1/-1->5->4 [16] 6/-1/-1->5->4 [17] 6/-1/-1->5->4 [18] 6/-1/-1->5->4 [19] 6/-1/-1->5->4 [20] 6/-1/-1->5->4 [21] 6/-1/-1->5->4 [22] 6/-1/-1->5->4 [23] 6/-1/-1->5->4
+ip-26-0-166-125:762819:763261 [4] NCCL INFO Trees [0] 5/-1/-1->4->3 [1] 5/-1/-1->4->3 [2] 5/-1/-1->4->3 [3] 5/-1/-1->4->3 [4] 5/-1/-1->4->3 [5] 5/-1/-1->4->3 [6] 5/-1/-1->4->3 [7] 5/-1/-1->4->3 [8] 5/-1/-1->4->3 [9] 5/-1/-1->4->3 [10] 5/-1/-1->4->3 [11] 5/-1/-1->4->3 [12] 5/-1/-1->4->3 [13] 5/-1/-1->4->3 [14] 5/-1/-1->4->3 [15] 5/-1/-1->4->3 [16] 5/-1/-1->4->3 [17] 5/-1/-1->4->3 [18] 5/-1/-1->4->3 [19] 5/-1/-1->4->3 [20] 5/-1/-1->4->3 [21] 5/-1/-1->4->3 [22] 5/-1/-1->4->3 [23] 5/-1/-1->4->3
+ip-26-0-166-125:762818:763263 [3] NCCL INFO Trees [0] 4/-1/-1->3->2 [1] 4/-1/-1->3->2 [2] 4/-1/-1->3->2 [3] 4/-1/-1->3->2 [4] 4/-1/-1->3->2 [5] 4/-1/-1->3->2 [6] 4/-1/-1->3->2 [7] 4/-1/-1->3->2 [8] 4/-1/-1->3->2 [9] 4/-1/-1->3->2 [10] 4/-1/-1->3->2 [11] 4/-1/-1->3->2 [12] 4/-1/-1->3->2 [13] 4/-1/-1->3->2 [14] 4/-1/-1->3->2 [15] 4/-1/-1->3->2 [16] 4/-1/-1->3->2 [17] 4/-1/-1->3->2 [18] 4/-1/-1->3->2 [19] 4/-1/-1->3->2 [20] 4/-1/-1->3->2 [21] 4/-1/-1->3->2 [22] 4/-1/-1->3->2 [23] 4/-1/-1->3->2
+ip-26-0-166-125:762820:763259 [5] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-166-125:762819:763261 [4] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-166-125:762818:763263 [3] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-166-125:762816:763262 [1] NCCL INFO Trees [0] 2/-1/-1->1->0 [1] 2/-1/-1->1->0 [2] 2/-1/-1->1->0 [3] 2/-1/-1->1->0 [4] 2/-1/-1->1->0 [5] 2/-1/-1->1->0 [6] 2/-1/-1->1->0 [7] 2/-1/-1->1->0 [8] 2/-1/-1->1->0 [9] 2/-1/-1->1->0 [10] 2/-1/-1->1->0 [11] 2/-1/-1->1->0 [12] 2/-1/-1->1->0 [13] 2/-1/-1->1->0 [14] 2/-1/-1->1->0 [15] 2/-1/-1->1->0 [16] 2/-1/-1->1->0 [17] 2/-1/-1->1->0 [18] 2/-1/-1->1->0 [19] 2/-1/-1->1->0 [20] 2/-1/-1->1->0 [21] 2/-1/-1->1->0 [22] 2/-1/-1->1->0 [23] 2/-1/-1->1->0
+ip-26-0-166-125:762817:763265 [2] NCCL INFO Trees [0] 3/-1/-1->2->1 [1] 3/-1/-1->2->1 [2] 3/-1/-1->2->1 [3] 3/-1/-1->2->1 [4] 3/-1/-1->2->1 [5] 3/-1/-1->2->1 [6] 3/-1/-1->2->1 [7] 3/-1/-1->2->1 [8] 3/-1/-1->2->1 [9] 3/-1/-1->2->1 [10] 3/-1/-1->2->1 [11] 3/-1/-1->2->1 [12] 3/-1/-1->2->1 [13] 3/-1/-1->2->1 [14] 3/-1/-1->2->1 [15] 3/-1/-1->2->1 [16] 3/-1/-1->2->1 [17] 3/-1/-1->2->1 [18] 3/-1/-1->2->1 [19] 3/-1/-1->2->1 [20] 3/-1/-1->2->1 [21] 3/-1/-1->2->1 [22] 3/-1/-1->2->1 [23] 3/-1/-1->2->1
+ip-26-0-166-125:762816:763262 [1] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-166-125:762815:763258 [0] NCCL INFO Channel 00/24 :    0   1   2   3   4   5   6   7
+ip-26-0-166-125:762817:763265 [2] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-166-125:762815:763258 [0] NCCL INFO Channel 01/24 :    0   1   2   3   4   5   6   7
+ip-26-0-166-125:762815:763258 [0] NCCL INFO Channel 02/24 :    0   1   2   3   4   5   6   7
+ip-26-0-166-125:762815:763258 [0] NCCL INFO Channel 03/24 :    0   1   2   3   4   5   6   7
+ip-26-0-166-125:762815:763258 [0] NCCL INFO Channel 04/24 :    0   1   2   3   4   5   6   7
+ip-26-0-166-125:762815:763258 [0] NCCL INFO Channel 05/24 :    0   1   2   3   4   5   6   7
+ip-26-0-166-125:762815:763258 [0] NCCL INFO Channel 06/24 :    0   1   2   3   4   5   6   7
+ip-26-0-166-125:762815:763258 [0] NCCL INFO Channel 07/24 :    0   1   2   3   4   5   6   7
+ip-26-0-166-125:762815:763258 [0] NCCL INFO Channel 08/24 :    0   1   2   3   4   5   6   7
+ip-26-0-166-125:762815:763258 [0] NCCL INFO Channel 09/24 :    0   1   2   3   4   5   6   7
+ip-26-0-166-125:762815:763258 [0] NCCL INFO Channel 10/24 :    0   1   2   3   4   5   6   7
+ip-26-0-166-125:762815:763258 [0] NCCL INFO Channel 11/24 :    0   1   2   3   4   5   6   7
+ip-26-0-166-125:762815:763258 [0] NCCL INFO Channel 12/24 :    0   1   2   3   4   5   6   7
+ip-26-0-166-125:762815:763258 [0] NCCL INFO Channel 13/24 :    0   1   2   3   4   5   6   7
+ip-26-0-166-125:762815:763258 [0] NCCL INFO Channel 14/24 :    0   1   2   3   4   5   6   7
+ip-26-0-166-125:762815:763258 [0] NCCL INFO Channel 15/24 :    0   1   2   3   4   5   6   7
+ip-26-0-166-125:762815:763258 [0] NCCL INFO Channel 16/24 :    0   1   2   3   4   5   6   7
+ip-26-0-166-125:762815:763258 [0] NCCL INFO Channel 17/24 :    0   1   2   3   4   5   6   7
+ip-26-0-166-125:762815:763258 [0] NCCL INFO Channel 18/24 :    0   1   2   3   4   5   6   7
+ip-26-0-166-214:624948:625382 [5] NCCL INFO Channel 00/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-166-125:762815:763258 [0] NCCL INFO Channel 19/24 :    0   1   2   3   4   5   6   7
+ip-26-0-166-125:762815:763258 [0] NCCL INFO Channel 20/24 :    0   1   2   3   4   5   6   7
+ip-26-0-166-125:762815:763258 [0] NCCL INFO Channel 21/24 :    0   1   2   3   4   5   6   7
+ip-26-0-166-125:762815:763258 [0] NCCL INFO Channel 22/24 :    0   1   2   3   4   5   6   7
+ip-26-0-166-125:762815:763258 [0] NCCL INFO Channel 23/24 :    0   1   2   3   4   5   6   7
+ip-26-0-166-125:762815:763258 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] 1/-1/-1->0->-1 [7] 1/-1/-1->0->-1 [8] 1/-1/-1->0->-1 [9] 1/-1/-1->0->-1 [10] 1/-1/-1->0->-1 [11] 1/-1/-1->0->-1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] 1/-1/-1->0->-1 [19] 1/-1/-1->0->-1 [20] 1/-1/-1->0->-1 [21] 1/-1/-1->0->-1 [22] 1/-1/-1->0->-1 [23] 1/-1/-1->0->-1
+ip-26-0-166-125:762815:763258 [0] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-166-214:624948:625382 [5] NCCL INFO Channel 01/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-166-214:624948:625382 [5] NCCL INFO Channel 02/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-166-214:624943:625375 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-214:624947:625378 [4] NCCL INFO Channel 02/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-166-214:624945:625380 [2] NCCL INFO Channel 02/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-214:624948:625382 [5] NCCL INFO Channel 03/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-166-214:624943:625375 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-214:624948:625382 [5] NCCL INFO Channel 04/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-166-214:624950:625376 [7] NCCL INFO Channel 04/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624947:625378 [4] NCCL INFO Channel 03/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-166-214:624948:625382 [5] NCCL INFO Channel 05/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-166-214:624950:625376 [7] NCCL INFO Channel 05/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624947:625378 [4] NCCL INFO Channel 04/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-166-214:624948:625382 [5] NCCL INFO Channel 06/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-166-214:624950:625376 [7] NCCL INFO Channel 06/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624947:625378 [4] NCCL INFO Channel 05/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-166-214:624948:625382 [5] NCCL INFO Channel 07/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-166-214:624950:625376 [7] NCCL INFO Channel 07/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624947:625378 [4] NCCL INFO Channel 06/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-166-214:624948:625382 [5] NCCL INFO Channel 08/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-166-214:624950:625376 [7] NCCL INFO Channel 08/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624947:625378 [4] NCCL INFO Channel 07/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-166-214:624946:625377 [3] NCCL INFO Channel 00/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-214:624949:625381 [6] NCCL INFO Channel 00/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-214:624948:625382 [5] NCCL INFO Channel 09/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-166-214:624950:625376 [7] NCCL INFO Channel 09/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624946:625377 [3] NCCL INFO Channel 01/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-214:624945:625380 [2] NCCL INFO Channel 03/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-214:624943:625375 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-214:624946:625377 [3] NCCL INFO Channel 02/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-214:624945:625380 [2] NCCL INFO Channel 04/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-214:624949:625381 [6] NCCL INFO Channel 01/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-214:624946:625377 [3] NCCL INFO Channel 03/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-214:624947:625378 [4] NCCL INFO Channel 08/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-166-214:624948:625382 [5] NCCL INFO Channel 10/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-166-214:624949:625381 [6] NCCL INFO Channel 02/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-214:624946:625377 [3] NCCL INFO Channel 04/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-214:624943:625375 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-214:624947:625378 [4] NCCL INFO Channel 09/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-166-214:624948:625382 [5] NCCL INFO Channel 11/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-166-214:624949:625381 [6] NCCL INFO Channel 03/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-214:624946:625377 [3] NCCL INFO Channel 05/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-214:624943:625375 [0] NCCL INFO Channel 04/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-214:624947:625378 [4] NCCL INFO Channel 10/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-166-214:624948:625382 [5] NCCL INFO Channel 12/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-166-214:624949:625381 [6] NCCL INFO Channel 04/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-214:624946:625377 [3] NCCL INFO Channel 06/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-214:624943:625375 [0] NCCL INFO Channel 05/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-214:624947:625378 [4] NCCL INFO Channel 11/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-166-36:104429:104861 [0] NCCL INFO Channel 04/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-36:104433:104864 [4] NCCL INFO Channel 08/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-166-36:104431:104862 [2] NCCL INFO Channel 08/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-36:104430:104868 [1] NCCL INFO Channel 08/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-166-36:104434:104867 [5] NCCL INFO Channel 08/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-166-36:104436:104865 [7] NCCL INFO Channel 09/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-36:104435:104863 [6] NCCL INFO Channel 04/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-36:104432:104866 [3] NCCL INFO Channel 05/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-36:104429:104861 [0] NCCL INFO Channel 05/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-36:104433:104864 [4] NCCL INFO Channel 09/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-166-36:104431:104862 [2] NCCL INFO Channel 09/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-36:104430:104868 [1] NCCL INFO Channel 09/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-166-36:104434:104867 [5] NCCL INFO Channel 09/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-166-36:104435:104863 [6] NCCL INFO Channel 05/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-36:104436:104865 [7] NCCL INFO Channel 10/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-36:104432:104866 [3] NCCL INFO Channel 06/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-36:104429:104861 [0] NCCL INFO Channel 06/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-36:104433:104864 [4] NCCL INFO Channel 10/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-166-214:624948:625382 [5] NCCL INFO Channel 13/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-166-36:104431:104862 [2] NCCL INFO Channel 10/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-36:104430:104868 [1] NCCL INFO Channel 10/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-166-36:104434:104867 [5] NCCL INFO Channel 10/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-166-36:104435:104863 [6] NCCL INFO Channel 06/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-36:104436:104865 [7] NCCL INFO Channel 11/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-36:104432:104866 [3] NCCL INFO Channel 07/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-36:104429:104861 [0] NCCL INFO Channel 07/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-36:104433:104864 [4] NCCL INFO Channel 11/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-166-36:104431:104862 [2] NCCL INFO Channel 11/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-36:104430:104868 [1] NCCL INFO Channel 11/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-166-36:104434:104867 [5] NCCL INFO Channel 11/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-166-36:104435:104863 [6] NCCL INFO Channel 07/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-36:104436:104865 [7] NCCL INFO Channel 12/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-36:104432:104866 [3] NCCL INFO Channel 08/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-36:104429:104861 [0] NCCL INFO Channel 08/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-36:104433:104864 [4] NCCL INFO Channel 12/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-166-36:104431:104862 [2] NCCL INFO Channel 12/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-36:104430:104868 [1] NCCL INFO Channel 12/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-166-36:104434:104867 [5] NCCL INFO Channel 12/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-166-36:104435:104863 [6] NCCL INFO Channel 08/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-36:104436:104865 [7] NCCL INFO Channel 13/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-36:104432:104866 [3] NCCL INFO Channel 09/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-36:104429:104861 [0] NCCL INFO Channel 09/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-36:104433:104864 [4] NCCL INFO Channel 13/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-166-36:104431:104862 [2] NCCL INFO Channel 13/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-36:104430:104868 [1] NCCL INFO Channel 13/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-166-36:104434:104867 [5] NCCL INFO Channel 13/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-166-36:104435:104863 [6] NCCL INFO Channel 09/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-36:104436:104865 [7] NCCL INFO Channel 14/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-36:104432:104866 [3] NCCL INFO Channel 10/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-36:104429:104861 [0] NCCL INFO Channel 10/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-36:104433:104864 [4] NCCL INFO Channel 14/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-166-36:104431:104862 [2] NCCL INFO Channel 14/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-36:104430:104868 [1] NCCL INFO Channel 14/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-166-36:104434:104867 [5] NCCL INFO Channel 14/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-166-36:104435:104863 [6] NCCL INFO Channel 10/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-36:104436:104865 [7] NCCL INFO Channel 15/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-36:104432:104866 [3] NCCL INFO Channel 11/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-36:104429:104861 [0] NCCL INFO Channel 11/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-36:104431:104862 [2] NCCL INFO Channel 15/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-36:104433:104864 [4] NCCL INFO Channel 15/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-166-36:104434:104867 [5] NCCL INFO Channel 15/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-166-36:104435:104863 [6] NCCL INFO Channel 11/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-36:104430:104868 [1] NCCL INFO Channel 15/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-166-36:104436:104865 [7] NCCL INFO Channel 16/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-36:104432:104866 [3] NCCL INFO Channel 12/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-36:104429:104861 [0] NCCL INFO Channel 12/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-36:104431:104862 [2] NCCL INFO Channel 16/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-36:104433:104864 [4] NCCL INFO Channel 16/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-166-36:104434:104867 [5] NCCL INFO Channel 16/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-166-36:104430:104868 [1] NCCL INFO Channel 16/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-166-36:104435:104863 [6] NCCL INFO Channel 12/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-36:104436:104865 [7] NCCL INFO Channel 17/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-36:104432:104866 [3] NCCL INFO Channel 13/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-214:624949:625381 [6] NCCL INFO Channel 05/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-214:624943:625375 [0] NCCL INFO Channel 06/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-36:104429:104861 [0] NCCL INFO Channel 13/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-36:104431:104862 [2] NCCL INFO Channel 17/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-36:104433:104864 [4] NCCL INFO Channel 17/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-166-36:104434:104867 [5] NCCL INFO Channel 17/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-166-36:104430:104868 [1] NCCL INFO Channel 17/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-166-36:104435:104863 [6] NCCL INFO Channel 13/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-36:104436:104865 [7] NCCL INFO Channel 18/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-36:104432:104866 [3] NCCL INFO Channel 14/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-36:104429:104861 [0] NCCL INFO Channel 14/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-36:104431:104862 [2] NCCL INFO Channel 18/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-214:624946:625377 [3] NCCL INFO Channel 07/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-214:624947:625378 [4] NCCL INFO Channel 12/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-166-36:104433:104864 [4] NCCL INFO Channel 18/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-166-36:104434:104867 [5] NCCL INFO Channel 18/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-166-36:104430:104868 [1] NCCL INFO Channel 18/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-166-36:104435:104863 [6] NCCL INFO Channel 14/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-36:104436:104865 [7] NCCL INFO Channel 19/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-36:104432:104866 [3] NCCL INFO Channel 15/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-36:104429:104861 [0] NCCL INFO Channel 15/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-36:104431:104862 [2] NCCL INFO Channel 19/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-36:104433:104864 [4] NCCL INFO Channel 19/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-166-36:104430:104868 [1] NCCL INFO Channel 19/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-166-36:104434:104867 [5] NCCL INFO Channel 19/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-166-36:104435:104863 [6] NCCL INFO Channel 15/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-36:104436:104865 [7] NCCL INFO Channel 20/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-36:104432:104866 [3] NCCL INFO Channel 16/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-36:104429:104861 [0] NCCL INFO Channel 16/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-36:104431:104862 [2] NCCL INFO Channel 20/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-244:897244:897674 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-166-244:897244:897674 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-166-36:104433:104864 [4] NCCL INFO Channel 20/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-166-214:624948:625382 [5] NCCL INFO Channel 14/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-166-36:104430:104868 [1] NCCL INFO Channel 20/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-166-36:104434:104867 [5] NCCL INFO Channel 20/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-166-36:104436:104865 [7] NCCL INFO Channel 21/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-36:104429:104861 [0] NCCL INFO Channel 17/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-36:104431:104862 [2] NCCL INFO Channel 21/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-36:104432:104866 [3] NCCL INFO Channel 17/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-36:104435:104863 [6] NCCL INFO Channel 16/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-36:104433:104864 [4] NCCL INFO Channel 21/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-167-9:1039542:1039976 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-166-214:624949:625381 [6] NCCL INFO Channel 06/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-214:624943:625375 [0] NCCL INFO Channel 07/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-167-9:1039542:1039976 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-166-214:624947:625378 [4] NCCL INFO Channel 13/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-166-214:624946:625377 [3] NCCL INFO Channel 08/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-214:624948:625382 [5] NCCL INFO Channel 15/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-167-9:1039541:1039971 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-167-9:1039541:1039971 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-166-36:104430:104868 [1] NCCL INFO Channel 21/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-166-214:624949:625381 [6] NCCL INFO Channel 07/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-36:104434:104867 [5] NCCL INFO Channel 21/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-166-36:104436:104865 [7] NCCL INFO Channel 22/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-36:104429:104861 [0] NCCL INFO Channel 18/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-36:104431:104862 [2] NCCL INFO Channel 22/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-36:104435:104863 [6] NCCL INFO Channel 17/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-214:624943:625375 [0] NCCL INFO Channel 08/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-36:104432:104866 [3] NCCL INFO Channel 18/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-36:104433:104864 [4] NCCL INFO Channel 22/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-166-214:624947:625378 [4] NCCL INFO Channel 14/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-166-214:624946:625377 [3] NCCL INFO Channel 09/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-214:624950:625376 [7] NCCL INFO Channel 10/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624945:625380 [2] NCCL INFO Channel 05/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-214:624948:625382 [5] NCCL INFO Channel 16/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-166-214:624943:625375 [0] NCCL INFO Channel 09/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-214:624949:625381 [6] NCCL INFO Channel 08/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-36:104430:104868 [1] NCCL INFO Channel 22/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-166-36:104436:104865 [7] NCCL INFO Channel 23/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-36:104434:104867 [5] NCCL INFO Channel 22/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-166-214:624947:625378 [4] NCCL INFO Channel 15/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-166-36:104431:104862 [2] NCCL INFO Channel 23/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-36:104435:104863 [6] NCCL INFO Channel 18/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-36:104429:104861 [0] NCCL INFO Channel 19/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-36:104432:104866 [3] NCCL INFO Channel 19/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-36:104433:104864 [4] NCCL INFO Channel 23/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-166-214:624946:625377 [3] NCCL INFO Channel 10/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-214:624950:625376 [7] NCCL INFO Channel 11/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624945:625380 [2] NCCL INFO Channel 06/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-214:624948:625382 [5] NCCL INFO Channel 17/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-167-9:1039538:1039973 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-167-9:1039538:1039973 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-166-214:624943:625375 [0] NCCL INFO Channel 10/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-214:624949:625381 [6] NCCL INFO Channel 09/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-214:624947:625378 [4] NCCL INFO Channel 16/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-167-9:1039538:1039973 [1] NCCL INFO Trees [0] 2/-1/-1->1->0 [1] 2/-1/-1->1->0 [2] 2/-1/-1->1->0 [3] 2/-1/-1->1->0 [4] 2/-1/-1->1->0 [5] 2/-1/-1->1->0 [6] 2/-1/-1->1->0 [7] 2/-1/-1->1->0 [8] 2/-1/-1->1->0 [9] 2/-1/-1->1->0 [10] 2/-1/-1->1->0 [11] 2/-1/-1->1->0 [12] 2/-1/-1->1->0 [13] 2/-1/-1->1->0 [14] 2/-1/-1->1->0 [15] 2/-1/-1->1->0 [16] 2/-1/-1->1->0 [17] 2/-1/-1->1->0 [18] 2/-1/-1->1->0 [19] 2/-1/-1->1->0 [20] 2/-1/-1->1->0 [21] 2/-1/-1->1->0 [22] 2/-1/-1->1->0 [23] 2/-1/-1->1->0
+ip-26-0-166-36:104430:104868 [1] NCCL INFO Channel 23/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-167-9:1039538:1039973 [1] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-167-9:1039539:1039975 [2] NCCL INFO Trees [0] 3/-1/-1->2->1 [1] 3/-1/-1->2->1 [2] 3/-1/-1->2->1 [3] 3/-1/-1->2->1 [4] 3/-1/-1->2->1 [5] 3/-1/-1->2->1 [6] 3/-1/-1->2->1 [7] 3/-1/-1->2->1 [8] 3/-1/-1->2->1 [9] 3/-1/-1->2->1 [10] 3/-1/-1->2->1 [11] 3/-1/-1->2->1 [12] 3/-1/-1->2->1 [13] 3/-1/-1->2->1 [14] 3/-1/-1->2->1 [15] 3/-1/-1->2->1 [16] 3/-1/-1->2->1 [17] 3/-1/-1->2->1 [18] 3/-1/-1->2->1 [19] 3/-1/-1->2->1 [20] 3/-1/-1->2->1 [21] 3/-1/-1->2->1 [22] 3/-1/-1->2->1 [23] 3/-1/-1->2->1
+ip-26-0-167-9:1039539:1039975 [2] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-167-9:1039537:1039969 [0] NCCL INFO Channel 00/24 :    0   1   2   3   4   5   6   7
+ip-26-0-167-9:1039540:1039974 [3] NCCL INFO Trees [0] 4/-1/-1->3->2 [1] 4/-1/-1->3->2 [2] 4/-1/-1->3->2 [3] 4/-1/-1->3->2 [4] 4/-1/-1->3->2 [5] 4/-1/-1->3->2 [6] 4/-1/-1->3->2 [7] 4/-1/-1->3->2 [8] 4/-1/-1->3->2 [9] 4/-1/-1->3->2 [10] 4/-1/-1->3->2 [11] 4/-1/-1->3->2 [12] 4/-1/-1->3->2 [13] 4/-1/-1->3->2 [14] 4/-1/-1->3->2 [15] 4/-1/-1->3->2 [16] 4/-1/-1->3->2 [17] 4/-1/-1->3->2 [18] 4/-1/-1->3->2 [19] 4/-1/-1->3->2 [20] 4/-1/-1->3->2 [21] 4/-1/-1->3->2 [22] 4/-1/-1->3->2 [23] 4/-1/-1->3->2
+ip-26-0-167-9:1039537:1039969 [0] NCCL INFO Channel 01/24 :    0   1   2   3   4   5   6   7
+ip-26-0-167-9:1039540:1039974 [3] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-167-9:1039537:1039969 [0] NCCL INFO Channel 02/24 :    0   1   2   3   4   5   6   7
+ip-26-0-167-9:1039537:1039969 [0] NCCL INFO Channel 03/24 :    0   1   2   3   4   5   6   7
+ip-26-0-166-36:104434:104867 [5] NCCL INFO Channel 23/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-166-36:104435:104863 [6] NCCL INFO Channel 19/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-36:104429:104861 [0] NCCL INFO Channel 20/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-167-9:1039544:1039972 [7] NCCL INFO Trees [0] -1/-1/-1->7->6 [1] -1/-1/-1->7->6 [2] -1/-1/-1->7->6 [3] -1/-1/-1->7->6 [4] -1/-1/-1->7->6 [5] -1/-1/-1->7->6 [6] -1/-1/-1->7->6 [7] -1/-1/-1->7->6 [8] -1/-1/-1->7->6 [9] -1/-1/-1->7->6 [10] -1/-1/-1->7->6 [11] -1/-1/-1->7->6 [12] -1/-1/-1->7->6 [13] -1/-1/-1->7->6 [14] -1/-1/-1->7->6 [15] -1/-1/-1->7->6 [16] -1/-1/-1->7->6 [17] -1/-1/-1->7->6 [18] -1/-1/-1->7->6 [19] -1/-1/-1->7->6 [20] -1/-1/-1->7->6 [21] -1/-1/-1->7->6 [22] -1/-1/-1->7->6 [23] -1/-1/-1->7->6
+ip-26-0-167-9:1039537:1039969 [0] NCCL INFO Channel 04/24 :    0   1   2   3   4   5   6   7
+ip-26-0-167-9:1039544:1039972 [7] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-167-9:1039541:1039971 [4] NCCL INFO Trees [0] 5/-1/-1->4->3 [1] 5/-1/-1->4->3 [2] 5/-1/-1->4->3 [3] 5/-1/-1->4->3 [4] 5/-1/-1->4->3 [5] 5/-1/-1->4->3 [6] 5/-1/-1->4->3 [7] 5/-1/-1->4->3 [8] 5/-1/-1->4->3 [9] 5/-1/-1->4->3 [10] 5/-1/-1->4->3 [11] 5/-1/-1->4->3 [12] 5/-1/-1->4->3 [13] 5/-1/-1->4->3 [14] 5/-1/-1->4->3 [15] 5/-1/-1->4->3 [16] 5/-1/-1->4->3 [17] 5/-1/-1->4->3 [18] 5/-1/-1->4->3 [19] 5/-1/-1->4->3 [20] 5/-1/-1->4->3 [21] 5/-1/-1->4->3 [22] 5/-1/-1->4->3 [23] 5/-1/-1->4->3
+ip-26-0-167-9:1039537:1039969 [0] NCCL INFO Channel 05/24 :    0   1   2   3   4   5   6   7
+ip-26-0-167-9:1039537:1039969 [0] NCCL INFO Channel 06/24 :    0   1   2   3   4   5   6   7
+ip-26-0-167-9:1039541:1039971 [4] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-167-9:1039543:1039970 [6] NCCL INFO Trees [0] 7/-1/-1->6->5 [1] 7/-1/-1->6->5 [2] 7/-1/-1->6->5 [3] 7/-1/-1->6->5 [4] 7/-1/-1->6->5 [5] 7/-1/-1->6->5 [6] 7/-1/-1->6->5 [7] 7/-1/-1->6->5 [8] 7/-1/-1->6->5 [9] 7/-1/-1->6->5 [10] 7/-1/-1->6->5 [11] 7/-1/-1->6->5 [12] 7/-1/-1->6->5 [13] 7/-1/-1->6->5 [14] 7/-1/-1->6->5 [15] 7/-1/-1->6->5 [16] 7/-1/-1->6->5 [17] 7/-1/-1->6->5 [18] 7/-1/-1->6->5 [19] 7/-1/-1->6->5 [20] 7/-1/-1->6->5 [21] 7/-1/-1->6->5 [22] 7/-1/-1->6->5 [23] 7/-1/-1->6->5
+ip-26-0-167-9:1039537:1039969 [0] NCCL INFO Channel 07/24 :    0   1   2   3   4   5   6   7
+ip-26-0-167-9:1039542:1039976 [5] NCCL INFO Trees [0] 6/-1/-1->5->4 [1] 6/-1/-1->5->4 [2] 6/-1/-1->5->4 [3] 6/-1/-1->5->4 [4] 6/-1/-1->5->4 [5] 6/-1/-1->5->4 [6] 6/-1/-1->5->4 [7] 6/-1/-1->5->4 [8] 6/-1/-1->5->4 [9] 6/-1/-1->5->4 [10] 6/-1/-1->5->4 [11] 6/-1/-1->5->4 [12] 6/-1/-1->5->4 [13] 6/-1/-1->5->4 [14] 6/-1/-1->5->4 [15] 6/-1/-1->5->4 [16] 6/-1/-1->5->4 [17] 6/-1/-1->5->4 [18] 6/-1/-1->5->4 [19] 6/-1/-1->5->4 [20] 6/-1/-1->5->4 [21] 6/-1/-1->5->4 [22] 6/-1/-1->5->4 [23] 6/-1/-1->5->4
+ip-26-0-167-9:1039537:1039969 [0] NCCL INFO Channel 08/24 :    0   1   2   3   4   5   6   7
+ip-26-0-167-9:1039543:1039970 [6] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-167-9:1039537:1039969 [0] NCCL INFO Channel 09/24 :    0   1   2   3   4   5   6   7
+ip-26-0-167-9:1039542:1039976 [5] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-167-9:1039537:1039969 [0] NCCL INFO Channel 10/24 :    0   1   2   3   4   5   6   7
+ip-26-0-167-9:1039537:1039969 [0] NCCL INFO Channel 11/24 :    0   1   2   3   4   5   6   7
+ip-26-0-166-36:104432:104866 [3] NCCL INFO Channel 20/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-167-9:1039537:1039969 [0] NCCL INFO Channel 12/24 :    0   1   2   3   4   5   6   7
+ip-26-0-166-214:624946:625377 [3] NCCL INFO Channel 11/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-214:624950:625376 [7] NCCL INFO Channel 12/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-167-9:1039537:1039969 [0] NCCL INFO Channel 13/24 :    0   1   2   3   4   5   6   7
+ip-26-0-167-9:1039537:1039969 [0] NCCL INFO Channel 14/24 :    0   1   2   3   4   5   6   7
+ip-26-0-167-9:1039537:1039969 [0] NCCL INFO Channel 15/24 :    0   1   2   3   4   5   6   7
+ip-26-0-167-9:1039537:1039969 [0] NCCL INFO Channel 16/24 :    0   1   2   3   4   5   6   7
+ip-26-0-167-9:1039537:1039969 [0] NCCL INFO Channel 17/24 :    0   1   2   3   4   5   6   7
+ip-26-0-167-9:1039537:1039969 [0] NCCL INFO Channel 18/24 :    0   1   2   3   4   5   6   7
+ip-26-0-167-9:1039537:1039969 [0] NCCL INFO Channel 19/24 :    0   1   2   3   4   5   6   7
+ip-26-0-167-9:1039537:1039969 [0] NCCL INFO Channel 20/24 :    0   1   2   3   4   5   6   7
+ip-26-0-167-9:1039537:1039969 [0] NCCL INFO Channel 21/24 :    0   1   2   3   4   5   6   7
+ip-26-0-167-9:1039537:1039969 [0] NCCL INFO Channel 22/24 :    0   1   2   3   4   5   6   7
+ip-26-0-167-9:1039537:1039969 [0] NCCL INFO Channel 23/24 :    0   1   2   3   4   5   6   7
+ip-26-0-167-9:1039537:1039969 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] 1/-1/-1->0->-1 [7] 1/-1/-1->0->-1 [8] 1/-1/-1->0->-1 [9] 1/-1/-1->0->-1 [10] 1/-1/-1->0->-1 [11] 1/-1/-1->0->-1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] 1/-1/-1->0->-1 [19] 1/-1/-1->0->-1 [20] 1/-1/-1->0->-1 [21] 1/-1/-1->0->-1 [22] 1/-1/-1->0->-1 [23] 1/-1/-1->0->-1
+ip-26-0-167-9:1039537:1039969 [0] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-166-214:624945:625380 [2] NCCL INFO Channel 07/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-214:624948:625382 [5] NCCL INFO Channel 18/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-166-36:104435:104863 [6] NCCL INFO Channel 20/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-36:104429:104861 [0] NCCL INFO Channel 21/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-36:104432:104866 [3] NCCL INFO Channel 21/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-214:624943:625375 [0] NCCL INFO Channel 11/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-214:624949:625381 [6] NCCL INFO Channel 10/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-214:624947:625378 [4] NCCL INFO Channel 17/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-166-214:624950:625376 [7] NCCL INFO Channel 13/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-36:104435:104863 [6] NCCL INFO Channel 21/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-36:104429:104861 [0] NCCL INFO Channel 22/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-214:624946:625377 [3] NCCL INFO Channel 12/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-36:104432:104866 [3] NCCL INFO Channel 22/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-214:624945:625380 [2] NCCL INFO Channel 08/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-214:624948:625382 [5] NCCL INFO Channel 19/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-166-214:624943:625375 [0] NCCL INFO Channel 12/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-214:624949:625381 [6] NCCL INFO Channel 11/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-36:104435:104863 [6] NCCL INFO Channel 22/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-36:104429:104861 [0] NCCL INFO Channel 23/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-214:624947:625378 [4] NCCL INFO Channel 18/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-166-214:624950:625376 [7] NCCL INFO Channel 14/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624946:625377 [3] NCCL INFO Channel 13/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-36:104432:104866 [3] NCCL INFO Channel 23/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-214:624945:625380 [2] NCCL INFO Channel 09/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-36:104435:104863 [6] NCCL INFO Channel 23/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-214:624948:625382 [5] NCCL INFO Channel 20/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-166-214:624943:625375 [0] NCCL INFO Channel 13/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-214:624949:625381 [6] NCCL INFO Channel 12/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-214:624947:625378 [4] NCCL INFO Channel 19/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-166-214:624950:625376 [7] NCCL INFO Channel 15/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624946:625377 [3] NCCL INFO Channel 14/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-214:624945:625380 [2] NCCL INFO Channel 10/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-214:624948:625382 [5] NCCL INFO Channel 21/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-166-214:624943:625375 [0] NCCL INFO Channel 14/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-15:127297:127731 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-166-125:762822:763260 [7] NCCL INFO Channel 00/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127297:127731 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-166-214:624949:625381 [6] NCCL INFO Channel 13/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-214:624947:625378 [4] NCCL INFO Channel 20/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-166-214:624950:625376 [7] NCCL INFO Channel 16/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624946:625377 [3] NCCL INFO Channel 15/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-214:624945:625380 [2] NCCL INFO Channel 11/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-125:762822:763260 [7] NCCL INFO Channel 01/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624948:625382 [5] NCCL INFO Channel 22/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-166-125:762820:763259 [5] NCCL INFO Channel 00/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-166-214:624943:625375 [0] NCCL INFO Channel 15/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-125:762822:763260 [7] NCCL INFO Channel 02/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624949:625381 [6] NCCL INFO Channel 14/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-214:624947:625378 [4] NCCL INFO Channel 21/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-166-15:127290:127724 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-166-15:127290:127724 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-166-214:624950:625376 [7] NCCL INFO Channel 17/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624946:625377 [3] NCCL INFO Channel 16/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-214:624945:625380 [2] NCCL INFO Channel 12/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-214:624944:625379 [1] NCCL INFO Channel 00/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-166-214:624948:625382 [5] NCCL INFO Channel 23/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-166-125:762820:763259 [5] NCCL INFO Channel 01/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-166-214:624943:625375 [0] NCCL INFO Channel 16/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-214:624947:625378 [4] NCCL INFO Channel 22/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-166-214:624949:625381 [6] NCCL INFO Channel 15/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-125:762820:763259 [5] NCCL INFO Channel 02/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-166-214:624950:625376 [7] NCCL INFO Channel 18/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624946:625377 [3] NCCL INFO Channel 17/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-214:624945:625380 [2] NCCL INFO Channel 13/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-214:624944:625379 [1] NCCL INFO Channel 01/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-166-125:762820:763259 [5] NCCL INFO Channel 03/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-166-214:624943:625375 [0] NCCL INFO Channel 17/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-214:624947:625378 [4] NCCL INFO Channel 23/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-166-125:762820:763259 [5] NCCL INFO Channel 04/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-166-214:624949:625381 [6] NCCL INFO Channel 16/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-214:624950:625376 [7] NCCL INFO Channel 19/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762815:763258 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-214:624945:625380 [2] NCCL INFO Channel 14/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-214:624946:625377 [3] NCCL INFO Channel 18/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-214:624944:625379 [1] NCCL INFO Channel 02/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-166-125:762820:763259 [5] NCCL INFO Channel 05/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-166-214:624943:625375 [0] NCCL INFO Channel 18/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-125:762815:763258 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-125:762820:763259 [5] NCCL INFO Channel 06/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-166-214:624949:625381 [6] NCCL INFO Channel 17/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-214:624950:625376 [7] NCCL INFO Channel 20/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624945:625380 [2] NCCL INFO Channel 15/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-214:624944:625379 [1] NCCL INFO Channel 03/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-166-125:762815:763258 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-214:624946:625377 [3] NCCL INFO Channel 19/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-214:624943:625375 [0] NCCL INFO Channel 19/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-125:762820:763259 [5] NCCL INFO Channel 07/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-166-125:762815:763258 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-214:624949:625381 [6] NCCL INFO Channel 18/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-214:624950:625376 [7] NCCL INFO Channel 21/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624945:625380 [2] NCCL INFO Channel 16/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-214:624944:625379 [1] NCCL INFO Channel 04/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-166-214:624946:625377 [3] NCCL INFO Channel 20/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-214:624943:625375 [0] NCCL INFO Channel 20/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-125:762820:763259 [5] NCCL INFO Channel 08/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-166-214:624949:625381 [6] NCCL INFO Channel 19/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-214:624950:625376 [7] NCCL INFO Channel 22/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762822:763260 [7] NCCL INFO Channel 03/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624945:625380 [2] NCCL INFO Channel 17/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-214:624944:625379 [1] NCCL INFO Channel 05/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-166-214:624946:625377 [3] NCCL INFO Channel 21/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-214:624943:625375 [0] NCCL INFO Channel 21/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-125:762820:763259 [5] NCCL INFO Channel 09/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-166-125:762822:763260 [7] NCCL INFO Channel 04/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624949:625381 [6] NCCL INFO Channel 20/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-214:624950:625376 [7] NCCL INFO Channel 23/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624945:625380 [2] NCCL INFO Channel 18/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-214:624944:625379 [1] NCCL INFO Channel 06/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-166-214:624946:625377 [3] NCCL INFO Channel 22/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-214:624943:625375 [0] NCCL INFO Channel 22/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-125:762815:763258 [0] NCCL INFO Channel 04/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-125:762820:763259 [5] NCCL INFO Channel 10/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-166-125:762822:763260 [7] NCCL INFO Channel 05/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762815:763258 [0] NCCL INFO Channel 05/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-214:624949:625381 [6] NCCL INFO Channel 21/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-214:624945:625380 [2] NCCL INFO Channel 19/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-162-46:702883:703318 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-162-46:702883:703318 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-166-125:762820:763259 [5] NCCL INFO Channel 11/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-166-125:762822:763260 [7] NCCL INFO Channel 06/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624944:625379 [1] NCCL INFO Channel 07/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-166-214:624949:625381 [6] NCCL INFO Channel 22/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-214:624945:625380 [2] NCCL INFO Channel 20/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-125:762815:763258 [0] NCCL INFO Channel 06/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-214:624943:625375 [0] NCCL INFO Channel 23/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-214:624944:625379 [1] NCCL INFO Channel 08/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-166-214:624945:625380 [2] NCCL INFO Channel 21/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-214:624949:625381 [6] NCCL INFO Channel 23/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-125:762820:763259 [5] NCCL INFO Channel 12/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-166-125:762817:763265 [2] NCCL INFO Channel 00/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-125:762822:763260 [7] NCCL INFO Channel 07/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762815:763258 [0] NCCL INFO Channel 07/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-125:762817:763265 [2] NCCL INFO Channel 01/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-125:762817:763265 [2] NCCL INFO Channel 02/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-214:624946:625377 [3] NCCL INFO Channel 23/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-125:762818:763263 [3] NCCL INFO Channel 00/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-214:624945:625380 [2] NCCL INFO Channel 22/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-125:762818:763263 [3] NCCL INFO Channel 01/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-125:762817:763265 [2] NCCL INFO Channel 03/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-125:762815:763258 [0] NCCL INFO Channel 08/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-125:762816:763262 [1] NCCL INFO Channel 00/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-166-125:762818:763263 [3] NCCL INFO Channel 02/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-125:762817:763265 [2] NCCL INFO Channel 04/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-125:762819:763261 [4] NCCL INFO Channel 00/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-166-125:762815:763258 [0] NCCL INFO Channel 09/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-125:762816:763262 [1] NCCL INFO Channel 01/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-166-125:762818:763263 [3] NCCL INFO Channel 03/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-125:762817:763265 [2] NCCL INFO Channel 05/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-125:762820:763259 [5] NCCL INFO Channel 13/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-166-125:762819:763261 [4] NCCL INFO Channel 01/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-166-125:762822:763260 [7] NCCL INFO Channel 08/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762815:763258 [0] NCCL INFO Channel 10/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-125:762815:763258 [0] NCCL INFO Channel 11/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-125:762818:763263 [3] NCCL INFO Channel 04/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-125:762817:763265 [2] NCCL INFO Channel 06/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-125:762815:763258 [0] NCCL INFO Channel 12/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-125:762818:763263 [3] NCCL INFO Channel 05/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-125:762817:763265 [2] NCCL INFO Channel 07/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-125:762818:763263 [3] NCCL INFO Channel 06/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-125:762819:763261 [4] NCCL INFO Channel 02/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-166-125:762821:763264 [6] NCCL INFO Channel 00/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-125:762815:763258 [0] NCCL INFO Channel 13/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-125:762817:763265 [2] NCCL INFO Channel 08/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-125:762822:763260 [7] NCCL INFO Channel 09/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762819:763261 [4] NCCL INFO Channel 03/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-166-125:762818:763263 [3] NCCL INFO Channel 07/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-125:762821:763264 [6] NCCL INFO Channel 01/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-125:762816:763262 [1] NCCL INFO Channel 02/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-166-125:762815:763258 [0] NCCL INFO Channel 14/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-125:762817:763265 [2] NCCL INFO Channel 09/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-214:624944:625379 [1] NCCL INFO Channel 09/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-166-125:762820:763259 [5] NCCL INFO Channel 14/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-166-125:762822:763260 [7] NCCL INFO Channel 10/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762819:763261 [4] NCCL INFO Channel 04/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-166-125:762821:763264 [6] NCCL INFO Channel 02/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-125:762816:763262 [1] NCCL INFO Channel 03/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-166-125:762818:763263 [3] NCCL INFO Channel 08/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-125:762817:763265 [2] NCCL INFO Channel 10/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-125:762815:763258 [0] NCCL INFO Channel 15/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-125:762822:763260 [7] NCCL INFO Channel 11/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762819:763261 [4] NCCL INFO Channel 05/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-166-125:762821:763264 [6] NCCL INFO Channel 03/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-125:762818:763263 [3] NCCL INFO Channel 09/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-125:762815:763258 [0] NCCL INFO Channel 16/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-125:762816:763262 [1] NCCL INFO Channel 04/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-166-125:762817:763265 [2] NCCL INFO Channel 11/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-125:762820:763259 [5] NCCL INFO Channel 15/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-162-46:702886:703315 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-162-46:702886:703315 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-166-125:762822:763260 [7] NCCL INFO Channel 12/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762819:763261 [4] NCCL INFO Channel 06/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-166-125:762821:763264 [6] NCCL INFO Channel 04/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-125:762818:763263 [3] NCCL INFO Channel 10/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-125:762816:763262 [1] NCCL INFO Channel 05/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-166-125:762815:763258 [0] NCCL INFO Channel 17/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-125:762817:763265 [2] NCCL INFO Channel 12/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-162-46:702886:703315 [6] NCCL INFO Trees [0] 7/-1/-1->6->5 [1] 7/-1/-1->6->5 [2] 7/-1/-1->6->5 [3] 7/-1/-1->6->5 [4] 7/-1/-1->6->5 [5] 7/-1/-1->6->5 [6] 7/-1/-1->6->5 [7] 7/-1/-1->6->5 [8] 7/-1/-1->6->5 [9] 7/-1/-1->6->5 [10] 7/-1/-1->6->5 [11] 7/-1/-1->6->5 [12] 7/-1/-1->6->5 [13] 7/-1/-1->6->5 [14] 7/-1/-1->6->5 [15] 7/-1/-1->6->5 [16] 7/-1/-1->6->5 [17] 7/-1/-1->6->5 [18] 7/-1/-1->6->5 [19] 7/-1/-1->6->5 [20] 7/-1/-1->6->5 [21] 7/-1/-1->6->5 [22] 7/-1/-1->6->5 [23] 7/-1/-1->6->5
+ip-26-0-162-46:702886:703315 [6] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-162-46:702885:703316 [5] NCCL INFO Trees [0] 6/-1/-1->5->4 [1] 6/-1/-1->5->4 [2] 6/-1/-1->5->4 [3] 6/-1/-1->5->4 [4] 6/-1/-1->5->4 [5] 6/-1/-1->5->4 [6] 6/-1/-1->5->4 [7] 6/-1/-1->5->4 [8] 6/-1/-1->5->4 [9] 6/-1/-1->5->4 [10] 6/-1/-1->5->4 [11] 6/-1/-1->5->4 [12] 6/-1/-1->5->4 [13] 6/-1/-1->5->4 [14] 6/-1/-1->5->4 [15] 6/-1/-1->5->4 [16] 6/-1/-1->5->4 [17] 6/-1/-1->5->4 [18] 6/-1/-1->5->4 [19] 6/-1/-1->5->4 [20] 6/-1/-1->5->4 [21] 6/-1/-1->5->4 [22] 6/-1/-1->5->4 [23] 6/-1/-1->5->4
+ip-26-0-162-46:702883:703318 [3] NCCL INFO Trees [0] 4/-1/-1->3->2 [1] 4/-1/-1->3->2 [2] 4/-1/-1->3->2 [3] 4/-1/-1->3->2 [4] 4/-1/-1->3->2 [5] 4/-1/-1->3->2 [6] 4/-1/-1->3->2 [7] 4/-1/-1->3->2 [8] 4/-1/-1->3->2 [9] 4/-1/-1->3->2 [10] 4/-1/-1->3->2 [11] 4/-1/-1->3->2 [12] 4/-1/-1->3->2 [13] 4/-1/-1->3->2 [14] 4/-1/-1->3->2 [15] 4/-1/-1->3->2 [16] 4/-1/-1->3->2 [17] 4/-1/-1->3->2 [18] 4/-1/-1->3->2 [19] 4/-1/-1->3->2 [20] 4/-1/-1->3->2 [21] 4/-1/-1->3->2 [22] 4/-1/-1->3->2 [23] 4/-1/-1->3->2
+ip-26-0-162-46:702885:703316 [5] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-162-46:702884:703319 [4] NCCL INFO Trees [0] 5/-1/-1->4->3 [1] 5/-1/-1->4->3 [2] 5/-1/-1->4->3 [3] 5/-1/-1->4->3 [4] 5/-1/-1->4->3 [5] 5/-1/-1->4->3 [6] 5/-1/-1->4->3 [7] 5/-1/-1->4->3 [8] 5/-1/-1->4->3 [9] 5/-1/-1->4->3 [10] 5/-1/-1->4->3 [11] 5/-1/-1->4->3 [12] 5/-1/-1->4->3 [13] 5/-1/-1->4->3 [14] 5/-1/-1->4->3 [15] 5/-1/-1->4->3 [16] 5/-1/-1->4->3 [17] 5/-1/-1->4->3 [18] 5/-1/-1->4->3 [19] 5/-1/-1->4->3 [20] 5/-1/-1->4->3 [21] 5/-1/-1->4->3 [22] 5/-1/-1->4->3 [23] 5/-1/-1->4->3
+ip-26-0-162-46:702880:703314 [0] NCCL INFO Channel 00/24 :    0   1   2   3   4   5   6   7
+ip-26-0-162-46:702883:703318 [3] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-162-46:702884:703319 [4] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-162-46:702880:703314 [0] NCCL INFO Channel 01/24 :    0   1   2   3   4   5   6   7
+ip-26-0-162-46:702887:703317 [7] NCCL INFO Trees [0] -1/-1/-1->7->6 [1] -1/-1/-1->7->6 [2] -1/-1/-1->7->6 [3] -1/-1/-1->7->6 [4] -1/-1/-1->7->6 [5] -1/-1/-1->7->6 [6] -1/-1/-1->7->6 [7] -1/-1/-1->7->6 [8] -1/-1/-1->7->6 [9] -1/-1/-1->7->6 [10] -1/-1/-1->7->6 [11] -1/-1/-1->7->6 [12] -1/-1/-1->7->6 [13] -1/-1/-1->7->6 [14] -1/-1/-1->7->6 [15] -1/-1/-1->7->6 [16] -1/-1/-1->7->6 [17] -1/-1/-1->7->6 [18] -1/-1/-1->7->6 [19] -1/-1/-1->7->6 [20] -1/-1/-1->7->6 [21] -1/-1/-1->7->6 [22] -1/-1/-1->7->6 [23] -1/-1/-1->7->6
+ip-26-0-162-46:702882:703321 [2] NCCL INFO Trees [0] 3/-1/-1->2->1 [1] 3/-1/-1->2->1 [2] 3/-1/-1->2->1 [3] 3/-1/-1->2->1 [4] 3/-1/-1->2->1 [5] 3/-1/-1->2->1 [6] 3/-1/-1->2->1 [7] 3/-1/-1->2->1 [8] 3/-1/-1->2->1 [9] 3/-1/-1->2->1 [10] 3/-1/-1->2->1 [11] 3/-1/-1->2->1 [12] 3/-1/-1->2->1 [13] 3/-1/-1->2->1 [14] 3/-1/-1->2->1 [15] 3/-1/-1->2->1 [16] 3/-1/-1->2->1 [17] 3/-1/-1->2->1 [18] 3/-1/-1->2->1 [19] 3/-1/-1->2->1 [20] 3/-1/-1->2->1 [21] 3/-1/-1->2->1 [22] 3/-1/-1->2->1 [23] 3/-1/-1->2->1
+ip-26-0-162-46:702880:703314 [0] NCCL INFO Channel 02/24 :    0   1   2   3   4   5   6   7
+ip-26-0-162-46:702887:703317 [7] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-162-46:702880:703314 [0] NCCL INFO Channel 03/24 :    0   1   2   3   4   5   6   7
+ip-26-0-162-46:702881:703320 [1] NCCL INFO Trees [0] 2/-1/-1->1->0 [1] 2/-1/-1->1->0 [2] 2/-1/-1->1->0 [3] 2/-1/-1->1->0 [4] 2/-1/-1->1->0 [5] 2/-1/-1->1->0 [6] 2/-1/-1->1->0 [7] 2/-1/-1->1->0 [8] 2/-1/-1->1->0 [9] 2/-1/-1->1->0 [10] 2/-1/-1->1->0 [11] 2/-1/-1->1->0 [12] 2/-1/-1->1->0 [13] 2/-1/-1->1->0 [14] 2/-1/-1->1->0 [15] 2/-1/-1->1->0 [16] 2/-1/-1->1->0 [17] 2/-1/-1->1->0 [18] 2/-1/-1->1->0 [19] 2/-1/-1->1->0 [20] 2/-1/-1->1->0 [21] 2/-1/-1->1->0 [22] 2/-1/-1->1->0 [23] 2/-1/-1->1->0
+ip-26-0-162-46:702880:703314 [0] NCCL INFO Channel 04/24 :    0   1   2   3   4   5   6   7
+ip-26-0-162-46:702880:703314 [0] NCCL INFO Channel 05/24 :    0   1   2   3   4   5   6   7
+ip-26-0-162-46:702881:703320 [1] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-162-46:702882:703321 [2] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-162-46:702880:703314 [0] NCCL INFO Channel 06/24 :    0   1   2   3   4   5   6   7
+ip-26-0-162-46:702880:703314 [0] NCCL INFO Channel 07/24 :    0   1   2   3   4   5   6   7
+ip-26-0-162-46:702880:703314 [0] NCCL INFO Channel 08/24 :    0   1   2   3   4   5   6   7
+ip-26-0-162-46:702880:703314 [0] NCCL INFO Channel 09/24 :    0   1   2   3   4   5   6   7
+ip-26-0-162-46:702880:703314 [0] NCCL INFO Channel 10/24 :    0   1   2   3   4   5   6   7
+ip-26-0-162-46:702880:703314 [0] NCCL INFO Channel 11/24 :    0   1   2   3   4   5   6   7
+ip-26-0-162-46:702880:703314 [0] NCCL INFO Channel 12/24 :    0   1   2   3   4   5   6   7
+ip-26-0-162-46:702880:703314 [0] NCCL INFO Channel 13/24 :    0   1   2   3   4   5   6   7
+ip-26-0-162-46:702880:703314 [0] NCCL INFO Channel 14/24 :    0   1   2   3   4   5   6   7
+ip-26-0-162-46:702880:703314 [0] NCCL INFO Channel 15/24 :    0   1   2   3   4   5   6   7
+ip-26-0-162-46:702880:703314 [0] NCCL INFO Channel 16/24 :    0   1   2   3   4   5   6   7
+ip-26-0-162-46:702880:703314 [0] NCCL INFO Channel 17/24 :    0   1   2   3   4   5   6   7
+ip-26-0-162-46:702880:703314 [0] NCCL INFO Channel 18/24 :    0   1   2   3   4   5   6   7
+ip-26-0-162-46:702880:703314 [0] NCCL INFO Channel 19/24 :    0   1   2   3   4   5   6   7
+ip-26-0-162-46:702880:703314 [0] NCCL INFO Channel 20/24 :    0   1   2   3   4   5   6   7
+ip-26-0-162-46:702880:703314 [0] NCCL INFO Channel 21/24 :    0   1   2   3   4   5   6   7
+ip-26-0-162-46:702880:703314 [0] NCCL INFO Channel 22/24 :    0   1   2   3   4   5   6   7
+ip-26-0-162-46:702880:703314 [0] NCCL INFO Channel 23/24 :    0   1   2   3   4   5   6   7
+ip-26-0-162-46:702880:703314 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] 1/-1/-1->0->-1 [7] 1/-1/-1->0->-1 [8] 1/-1/-1->0->-1 [9] 1/-1/-1->0->-1 [10] 1/-1/-1->0->-1 [11] 1/-1/-1->0->-1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] 1/-1/-1->0->-1 [19] 1/-1/-1->0->-1 [20] 1/-1/-1->0->-1 [21] 1/-1/-1->0->-1 [22] 1/-1/-1->0->-1 [23] 1/-1/-1->0->-1
+ip-26-0-162-46:702880:703314 [0] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-166-125:762820:763259 [5] NCCL INFO Channel 16/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-166-125:762819:763261 [4] NCCL INFO Channel 07/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-166-125:762822:763260 [7] NCCL INFO Channel 13/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762818:763263 [3] NCCL INFO Channel 11/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-125:762821:763264 [6] NCCL INFO Channel 05/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-125:762816:763262 [1] NCCL INFO Channel 06/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-166-125:762817:763265 [2] NCCL INFO Channel 13/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-125:762815:763258 [0] NCCL INFO Channel 18/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-125:762820:763259 [5] NCCL INFO Channel 17/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-166-125:762819:763261 [4] NCCL INFO Channel 08/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-166-125:762820:763259 [5] NCCL INFO Channel 18/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-166-125:762822:763260 [7] NCCL INFO Channel 14/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762821:763264 [6] NCCL INFO Channel 06/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-125:762818:763263 [3] NCCL INFO Channel 12/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-125:762815:763258 [0] NCCL INFO Channel 19/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-125:762816:763262 [1] NCCL INFO Channel 07/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-166-125:762817:763265 [2] NCCL INFO Channel 14/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-214:624945:625380 [2] NCCL INFO Channel 23/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-15:127292:127729 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-166-15:127292:127729 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-166-125:762819:763261 [4] NCCL INFO Channel 09/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-166-125:762822:763260 [7] NCCL INFO Channel 15/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762820:763259 [5] NCCL INFO Channel 19/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-166-125:762821:763264 [6] NCCL INFO Channel 07/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-125:762818:763263 [3] NCCL INFO Channel 13/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-125:762815:763258 [0] NCCL INFO Channel 20/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-125:762816:763262 [1] NCCL INFO Channel 08/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-166-125:762817:763265 [2] NCCL INFO Channel 15/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-125:762819:763261 [4] NCCL INFO Channel 10/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-166-125:762822:763260 [7] NCCL INFO Channel 16/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762820:763259 [5] NCCL INFO Channel 20/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-166-125:762815:763258 [0] NCCL INFO Channel 21/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-125:762821:763264 [6] NCCL INFO Channel 08/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-125:762818:763263 [3] NCCL INFO Channel 14/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-125:762816:763262 [1] NCCL INFO Channel 09/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-166-125:762817:763265 [2] NCCL INFO Channel 16/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-125:762819:763261 [4] NCCL INFO Channel 11/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-166-125:762822:763260 [7] NCCL INFO Channel 17/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762821:763264 [6] NCCL INFO Channel 09/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-125:762818:763263 [3] NCCL INFO Channel 15/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-125:762815:763258 [0] NCCL INFO Channel 22/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-125:762816:763262 [1] NCCL INFO Channel 10/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-166-125:762817:763265 [2] NCCL INFO Channel 17/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-214:624944:625379 [1] NCCL INFO Channel 10/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-166-125:762820:763259 [5] NCCL INFO Channel 21/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-166-125:762822:763260 [7] NCCL INFO Channel 18/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762819:763261 [4] NCCL INFO Channel 12/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-166-125:762818:763263 [3] NCCL INFO Channel 16/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-125:762821:763264 [6] NCCL INFO Channel 10/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-125:762816:763262 [1] NCCL INFO Channel 11/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-166-125:762815:763258 [0] NCCL INFO Channel 23/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-125:762817:763265 [2] NCCL INFO Channel 18/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-125:762820:763259 [5] NCCL INFO Channel 22/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-166-125:762822:763260 [7] NCCL INFO Channel 19/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762819:763261 [4] NCCL INFO Channel 13/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-166-125:762821:763264 [6] NCCL INFO Channel 11/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-125:762818:763263 [3] NCCL INFO Channel 17/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-125:762816:763262 [1] NCCL INFO Channel 12/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-166-125:762817:763265 [2] NCCL INFO Channel 19/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-15:127291:127730 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-166-125:762820:763259 [5] NCCL INFO Channel 23/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-167-9:1039544:1039972 [7] NCCL INFO Channel 00/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-167-9:1039544:1039972 [7] NCCL INFO Channel 01/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-167-9:1039537:1039969 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-167-9:1039544:1039972 [7] NCCL INFO Channel 02/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-167-9:1039537:1039969 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-167-9:1039544:1039972 [7] NCCL INFO Channel 03/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-167-9:1039537:1039969 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-167-9:1039544:1039972 [7] NCCL INFO Channel 04/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-167-9:1039537:1039969 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-167-9:1039542:1039976 [5] NCCL INFO Channel 00/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-167-9:1039544:1039972 [7] NCCL INFO Channel 05/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-167-9:1039537:1039969 [0] NCCL INFO Channel 04/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-167-9:1039542:1039976 [5] NCCL INFO Channel 01/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-167-9:1039537:1039969 [0] NCCL INFO Channel 05/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-167-9:1039542:1039976 [5] NCCL INFO Channel 02/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-166-15:127291:127730 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-167-9:1039539:1039975 [2] NCCL INFO Channel 00/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-167-9:1039537:1039969 [0] NCCL INFO Channel 06/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-167-9:1039542:1039976 [5] NCCL INFO Channel 03/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-167-9:1039539:1039975 [2] NCCL INFO Channel 01/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-167-9:1039537:1039969 [0] NCCL INFO Channel 07/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-167-9:1039541:1039971 [4] NCCL INFO Channel 00/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-167-9:1039543:1039970 [6] NCCL INFO Channel 00/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-167-9:1039542:1039976 [5] NCCL INFO Channel 04/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-167-9:1039540:1039974 [3] NCCL INFO Channel 00/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-167-9:1039539:1039975 [2] NCCL INFO Channel 02/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-167-9:1039537:1039969 [0] NCCL INFO Channel 08/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-167-9:1039541:1039971 [4] NCCL INFO Channel 01/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-167-9:1039543:1039970 [6] NCCL INFO Channel 01/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-167-9:1039544:1039972 [7] NCCL INFO Channel 06/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-167-9:1039542:1039976 [5] NCCL INFO Channel 05/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-167-9:1039540:1039974 [3] NCCL INFO Channel 01/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-167-9:1039539:1039975 [2] NCCL INFO Channel 03/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-167-9:1039537:1039969 [0] NCCL INFO Channel 09/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-167-9:1039541:1039971 [4] NCCL INFO Channel 02/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-167-9:1039543:1039970 [6] NCCL INFO Channel 02/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-167-9:1039544:1039972 [7] NCCL INFO Channel 07/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-167-9:1039542:1039976 [5] NCCL INFO Channel 06/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-166-125:762822:763260 [7] NCCL INFO Channel 20/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-167-9:1039540:1039974 [3] NCCL INFO Channel 02/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-167-9:1039539:1039975 [2] NCCL INFO Channel 04/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-125:762819:763261 [4] NCCL INFO Channel 14/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-166-125:762821:763264 [6] NCCL INFO Channel 12/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-125:762818:763263 [3] NCCL INFO Channel 18/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-125:762816:763262 [1] NCCL INFO Channel 13/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-166-125:762817:763265 [2] NCCL INFO Channel 20/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-167-9:1039537:1039969 [0] NCCL INFO Channel 10/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-167-9:1039541:1039971 [4] NCCL INFO Channel 03/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-167-9:1039538:1039973 [1] NCCL INFO Channel 00/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-167-9:1039543:1039970 [6] NCCL INFO Channel 03/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-125:762822:763260 [7] NCCL INFO Channel 21/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-167-9:1039544:1039972 [7] NCCL INFO Channel 08/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762819:763261 [4] NCCL INFO Channel 15/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-166-125:762821:763264 [6] NCCL INFO Channel 13/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-125:762818:763263 [3] NCCL INFO Channel 19/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-125:762816:763262 [1] NCCL INFO Channel 14/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-166-125:762817:763265 [2] NCCL INFO Channel 21/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-167-9:1039542:1039976 [5] NCCL INFO Channel 07/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-167-9:1039540:1039974 [3] NCCL INFO Channel 03/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-167-9:1039539:1039975 [2] NCCL INFO Channel 05/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-125:762822:763260 [7] NCCL INFO Channel 22/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762819:763261 [4] NCCL INFO Channel 16/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-166-125:762818:763263 [3] NCCL INFO Channel 20/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-125:762821:763264 [6] NCCL INFO Channel 14/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-125:762816:763262 [1] NCCL INFO Channel 15/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-166-125:762817:763265 [2] NCCL INFO Channel 22/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-167-9:1039537:1039969 [0] NCCL INFO Channel 11/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-167-9:1039541:1039971 [4] NCCL INFO Channel 04/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-167-9:1039538:1039973 [1] NCCL INFO Channel 01/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-167-9:1039543:1039970 [6] NCCL INFO Channel 04/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-167-9:1039544:1039972 [7] NCCL INFO Channel 09/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-167-9:1039542:1039976 [5] NCCL INFO Channel 08/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-167-9:1039540:1039974 [3] NCCL INFO Channel 04/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-167-9:1039539:1039975 [2] NCCL INFO Channel 06/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-125:762822:763260 [7] NCCL INFO Channel 23/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762819:763261 [4] NCCL INFO Channel 17/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-166-125:762818:763263 [3] NCCL INFO Channel 21/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-125:762816:763262 [1] NCCL INFO Channel 16/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-166-125:762821:763264 [6] NCCL INFO Channel 15/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-125:762817:763265 [2] NCCL INFO Channel 23/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-167-9:1039537:1039969 [0] NCCL INFO Channel 12/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-167-9:1039541:1039971 [4] NCCL INFO Channel 05/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-167-9:1039538:1039973 [1] NCCL INFO Channel 02/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-167-9:1039543:1039970 [6] NCCL INFO Channel 05/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-167-9:1039544:1039972 [7] NCCL INFO Channel 10/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-167-9:1039542:1039976 [5] NCCL INFO Channel 09/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-167-9:1039539:1039975 [2] NCCL INFO Channel 07/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-167-9:1039540:1039974 [3] NCCL INFO Channel 05/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-125:762819:763261 [4] NCCL INFO Channel 18/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-166-125:762818:763263 [3] NCCL INFO Channel 22/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-125:762816:763262 [1] NCCL INFO Channel 17/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-166-125:762821:763264 [6] NCCL INFO Channel 16/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-214:624944:625379 [1] NCCL INFO Channel 11/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-166-125:762819:763261 [4] NCCL INFO Channel 19/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-167-9:1039537:1039969 [0] NCCL INFO Channel 13/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-125:762818:763263 [3] NCCL INFO Channel 23/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-125:762816:763262 [1] NCCL INFO Channel 18/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-166-125:762821:763264 [6] NCCL INFO Channel 17/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-167-9:1039541:1039971 [4] NCCL INFO Channel 06/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-167-9:1039538:1039973 [1] NCCL INFO Channel 03/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-167-9:1039543:1039970 [6] NCCL INFO Channel 06/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-167-9:1039544:1039972 [7] NCCL INFO Channel 11/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-167-9:1039539:1039975 [2] NCCL INFO Channel 08/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-167-9:1039542:1039976 [5] NCCL INFO Channel 10/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-167-9:1039540:1039974 [3] NCCL INFO Channel 06/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-125:762819:763261 [4] NCCL INFO Channel 20/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-166-125:762816:763262 [1] NCCL INFO Channel 19/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-166-125:762821:763264 [6] NCCL INFO Channel 18/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-125:762819:763261 [4] NCCL INFO Channel 21/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-166-125:762816:763262 [1] NCCL INFO Channel 20/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-166-125:762821:763264 [6] NCCL INFO Channel 19/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-15:127296:127727 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-166-15:127296:127727 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-167-9:1039537:1039969 [0] NCCL INFO Channel 14/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-167-9:1039541:1039971 [4] NCCL INFO Channel 07/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-167-9:1039543:1039970 [6] NCCL INFO Channel 07/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-167-9:1039538:1039973 [1] NCCL INFO Channel 04/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-167-9:1039544:1039972 [7] NCCL INFO Channel 12/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127296:127727 [6] NCCL INFO Trees [0] 7/-1/-1->6->5 [1] 7/-1/-1->6->5 [2] 7/-1/-1->6->5 [3] 7/-1/-1->6->5 [4] 7/-1/-1->6->5 [5] 7/-1/-1->6->5 [6] 7/-1/-1->6->5 [7] 7/-1/-1->6->5 [8] 7/-1/-1->6->5 [9] 7/-1/-1->6->5 [10] 7/-1/-1->6->5 [11] 7/-1/-1->6->5 [12] 7/-1/-1->6->5 [13] 7/-1/-1->6->5 [14] 7/-1/-1->6->5 [15] 7/-1/-1->6->5 [16] 7/-1/-1->6->5 [17] 7/-1/-1->6->5 [18] 7/-1/-1->6->5 [19] 7/-1/-1->6->5 [20] 7/-1/-1->6->5 [21] 7/-1/-1->6->5 [22] 7/-1/-1->6->5 [23] 7/-1/-1->6->5
+ip-26-0-166-15:127296:127727 [6] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-166-15:127290:127724 [0] NCCL INFO Channel 00/24 :    0   1   2   3   4   5   6   7
+ip-26-0-166-15:127290:127724 [0] NCCL INFO Channel 01/24 :    0   1   2   3   4   5   6   7
+ip-26-0-166-15:127290:127724 [0] NCCL INFO Channel 02/24 :    0   1   2   3   4   5   6   7
+ip-26-0-166-15:127290:127724 [0] NCCL INFO Channel 03/24 :    0   1   2   3   4   5   6   7
+ip-26-0-166-15:127295:127726 [5] NCCL INFO Trees [0] 6/-1/-1->5->4 [1] 6/-1/-1->5->4 [2] 6/-1/-1->5->4 [3] 6/-1/-1->5->4 [4] 6/-1/-1->5->4 [5] 6/-1/-1->5->4 [6] 6/-1/-1->5->4 [7] 6/-1/-1->5->4 [8] 6/-1/-1->5->4 [9] 6/-1/-1->5->4 [10] 6/-1/-1->5->4 [11] 6/-1/-1->5->4 [12] 6/-1/-1->5->4 [13] 6/-1/-1->5->4 [14] 6/-1/-1->5->4 [15] 6/-1/-1->5->4 [16] 6/-1/-1->5->4 [17] 6/-1/-1->5->4 [18] 6/-1/-1->5->4 [19] 6/-1/-1->5->4 [20] 6/-1/-1->5->4 [21] 6/-1/-1->5->4 [22] 6/-1/-1->5->4 [23] 6/-1/-1->5->4
+ip-26-0-166-15:127290:127724 [0] NCCL INFO Channel 04/24 :    0   1   2   3   4   5   6   7
+ip-26-0-167-9:1039542:1039976 [5] NCCL INFO Channel 11/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-167-9:1039539:1039975 [2] NCCL INFO Channel 09/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-15:127295:127726 [5] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-166-15:127290:127724 [0] NCCL INFO Channel 05/24 :    0   1   2   3   4   5   6   7
+ip-26-0-166-15:127291:127730 [1] NCCL INFO Trees [0] 2/-1/-1->1->0 [1] 2/-1/-1->1->0 [2] 2/-1/-1->1->0 [3] 2/-1/-1->1->0 [4] 2/-1/-1->1->0 [5] 2/-1/-1->1->0 [6] 2/-1/-1->1->0 [7] 2/-1/-1->1->0 [8] 2/-1/-1->1->0 [9] 2/-1/-1->1->0 [10] 2/-1/-1->1->0 [11] 2/-1/-1->1->0 [12] 2/-1/-1->1->0 [13] 2/-1/-1->1->0 [14] 2/-1/-1->1->0 [15] 2/-1/-1->1->0 [16] 2/-1/-1->1->0 [17] 2/-1/-1->1->0 [18] 2/-1/-1->1->0 [19] 2/-1/-1->1->0 [20] 2/-1/-1->1->0 [21] 2/-1/-1->1->0 [22] 2/-1/-1->1->0 [23] 2/-1/-1->1->0
+ip-26-0-166-15:127297:127731 [7] NCCL INFO Trees [0] -1/-1/-1->7->6 [1] -1/-1/-1->7->6 [2] -1/-1/-1->7->6 [3] -1/-1/-1->7->6 [4] -1/-1/-1->7->6 [5] -1/-1/-1->7->6 [6] -1/-1/-1->7->6 [7] -1/-1/-1->7->6 [8] -1/-1/-1->7->6 [9] -1/-1/-1->7->6 [10] -1/-1/-1->7->6 [11] -1/-1/-1->7->6 [12] -1/-1/-1->7->6 [13] -1/-1/-1->7->6 [14] -1/-1/-1->7->6 [15] -1/-1/-1->7->6 [16] -1/-1/-1->7->6 [17] -1/-1/-1->7->6 [18] -1/-1/-1->7->6 [19] -1/-1/-1->7->6 [20] -1/-1/-1->7->6 [21] -1/-1/-1->7->6 [22] -1/-1/-1->7->6 [23] -1/-1/-1->7->6
+ip-26-0-166-15:127292:127729 [2] NCCL INFO Trees [0] 3/-1/-1->2->1 [1] 3/-1/-1->2->1 [2] 3/-1/-1->2->1 [3] 3/-1/-1->2->1 [4] 3/-1/-1->2->1 [5] 3/-1/-1->2->1 [6] 3/-1/-1->2->1 [7] 3/-1/-1->2->1 [8] 3/-1/-1->2->1 [9] 3/-1/-1->2->1 [10] 3/-1/-1->2->1 [11] 3/-1/-1->2->1 [12] 3/-1/-1->2->1 [13] 3/-1/-1->2->1 [14] 3/-1/-1->2->1 [15] 3/-1/-1->2->1 [16] 3/-1/-1->2->1 [17] 3/-1/-1->2->1 [18] 3/-1/-1->2->1 [19] 3/-1/-1->2->1 [20] 3/-1/-1->2->1 [21] 3/-1/-1->2->1 [22] 3/-1/-1->2->1 [23] 3/-1/-1->2->1
+ip-26-0-166-15:127290:127724 [0] NCCL INFO Channel 06/24 :    0   1   2   3   4   5   6   7
+ip-26-0-166-15:127293:127728 [3] NCCL INFO Trees [0] 4/-1/-1->3->2 [1] 4/-1/-1->3->2 [2] 4/-1/-1->3->2 [3] 4/-1/-1->3->2 [4] 4/-1/-1->3->2 [5] 4/-1/-1->3->2 [6] 4/-1/-1->3->2 [7] 4/-1/-1->3->2 [8] 4/-1/-1->3->2 [9] 4/-1/-1->3->2 [10] 4/-1/-1->3->2 [11] 4/-1/-1->3->2 [12] 4/-1/-1->3->2 [13] 4/-1/-1->3->2 [14] 4/-1/-1->3->2 [15] 4/-1/-1->3->2 [16] 4/-1/-1->3->2 [17] 4/-1/-1->3->2 [18] 4/-1/-1->3->2 [19] 4/-1/-1->3->2 [20] 4/-1/-1->3->2 [21] 4/-1/-1->3->2 [22] 4/-1/-1->3->2 [23] 4/-1/-1->3->2
+ip-26-0-166-15:127291:127730 [1] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-166-15:127290:127724 [0] NCCL INFO Channel 07/24 :    0   1   2   3   4   5   6   7
+ip-26-0-166-15:127297:127731 [7] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-166-15:127292:127729 [2] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-166-15:127290:127724 [0] NCCL INFO Channel 08/24 :    0   1   2   3   4   5   6   7
+ip-26-0-167-9:1039540:1039974 [3] NCCL INFO Channel 07/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-15:127290:127724 [0] NCCL INFO Channel 09/24 :    0   1   2   3   4   5   6   7
+ip-26-0-166-15:127294:127725 [4] NCCL INFO Trees [0] 5/-1/-1->4->3 [1] 5/-1/-1->4->3 [2] 5/-1/-1->4->3 [3] 5/-1/-1->4->3 [4] 5/-1/-1->4->3 [5] 5/-1/-1->4->3 [6] 5/-1/-1->4->3 [7] 5/-1/-1->4->3 [8] 5/-1/-1->4->3 [9] 5/-1/-1->4->3 [10] 5/-1/-1->4->3 [11] 5/-1/-1->4->3 [12] 5/-1/-1->4->3 [13] 5/-1/-1->4->3 [14] 5/-1/-1->4->3 [15] 5/-1/-1->4->3 [16] 5/-1/-1->4->3 [17] 5/-1/-1->4->3 [18] 5/-1/-1->4->3 [19] 5/-1/-1->4->3 [20] 5/-1/-1->4->3 [21] 5/-1/-1->4->3 [22] 5/-1/-1->4->3 [23] 5/-1/-1->4->3
+ip-26-0-166-15:127290:127724 [0] NCCL INFO Channel 10/24 :    0   1   2   3   4   5   6   7
+ip-26-0-166-15:127290:127724 [0] NCCL INFO Channel 11/24 :    0   1   2   3   4   5   6   7
+ip-26-0-166-15:127290:127724 [0] NCCL INFO Channel 12/24 :    0   1   2   3   4   5   6   7
+ip-26-0-166-15:127290:127724 [0] NCCL INFO Channel 13/24 :    0   1   2   3   4   5   6   7
+ip-26-0-166-15:127293:127728 [3] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-166-15:127294:127725 [4] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-166-15:127290:127724 [0] NCCL INFO Channel 14/24 :    0   1   2   3   4   5   6   7
+ip-26-0-166-15:127290:127724 [0] NCCL INFO Channel 15/24 :    0   1   2   3   4   5   6   7
+ip-26-0-166-15:127290:127724 [0] NCCL INFO Channel 16/24 :    0   1   2   3   4   5   6   7
+ip-26-0-166-15:127290:127724 [0] NCCL INFO Channel 17/24 :    0   1   2   3   4   5   6   7
+ip-26-0-166-15:127290:127724 [0] NCCL INFO Channel 18/24 :    0   1   2   3   4   5   6   7
+ip-26-0-166-15:127290:127724 [0] NCCL INFO Channel 19/24 :    0   1   2   3   4   5   6   7
+ip-26-0-166-15:127290:127724 [0] NCCL INFO Channel 20/24 :    0   1   2   3   4   5   6   7
+ip-26-0-166-15:127290:127724 [0] NCCL INFO Channel 21/24 :    0   1   2   3   4   5   6   7
+ip-26-0-166-15:127290:127724 [0] NCCL INFO Channel 22/24 :    0   1   2   3   4   5   6   7
+ip-26-0-166-15:127290:127724 [0] NCCL INFO Channel 23/24 :    0   1   2   3   4   5   6   7
+ip-26-0-166-15:127290:127724 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] 1/-1/-1->0->-1 [7] 1/-1/-1->0->-1 [8] 1/-1/-1->0->-1 [9] 1/-1/-1->0->-1 [10] 1/-1/-1->0->-1 [11] 1/-1/-1->0->-1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] 1/-1/-1->0->-1 [19] 1/-1/-1->0->-1 [20] 1/-1/-1->0->-1 [21] 1/-1/-1->0->-1 [22] 1/-1/-1->0->-1 [23] 1/-1/-1->0->-1
+ip-26-0-166-15:127290:127724 [0] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-166-125:762819:763261 [4] NCCL INFO Channel 22/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-166-125:762816:763262 [1] NCCL INFO Channel 21/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-166-125:762821:763264 [6] NCCL INFO Channel 20/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-125:762819:763261 [4] NCCL INFO Channel 23/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-166-125:762816:763262 [1] NCCL INFO Channel 22/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-166-125:762821:763264 [6] NCCL INFO Channel 21/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-167-9:1039537:1039969 [0] NCCL INFO Channel 15/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-167-9:1039541:1039971 [4] NCCL INFO Channel 08/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-167-9:1039543:1039970 [6] NCCL INFO Channel 08/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-167-9:1039544:1039972 [7] NCCL INFO Channel 13/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-167-9:1039538:1039973 [1] NCCL INFO Channel 05/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-167-9:1039539:1039975 [2] NCCL INFO Channel 10/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-167-9:1039542:1039976 [5] NCCL INFO Channel 12/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-167-9:1039540:1039974 [3] NCCL INFO Channel 08/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-125:762816:763262 [1] NCCL INFO Channel 23/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-166-125:762821:763264 [6] NCCL INFO Channel 22/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-167-9:1039541:1039971 [4] NCCL INFO Channel 09/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-167-9:1039537:1039969 [0] NCCL INFO Channel 16/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-167-9:1039543:1039970 [6] NCCL INFO Channel 09/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-167-9:1039544:1039972 [7] NCCL INFO Channel 14/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-167-9:1039538:1039973 [1] NCCL INFO Channel 06/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-167-9:1039539:1039975 [2] NCCL INFO Channel 11/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-167-9:1039542:1039976 [5] NCCL INFO Channel 13/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-167-9:1039540:1039974 [3] NCCL INFO Channel 09/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-244:897242:897679 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-167-9:1039541:1039971 [4] NCCL INFO Channel 10/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-167-9:1039537:1039969 [0] NCCL INFO Channel 17/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-167-9:1039544:1039972 [7] NCCL INFO Channel 15/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-167-9:1039543:1039970 [6] NCCL INFO Channel 10/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-167-9:1039538:1039973 [1] NCCL INFO Channel 07/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-167-9:1039539:1039975 [2] NCCL INFO Channel 12/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-167-9:1039542:1039976 [5] NCCL INFO Channel 14/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-167-9:1039540:1039974 [3] NCCL INFO Channel 10/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-125:762821:763264 [6] NCCL INFO Channel 23/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-244:897242:897679 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-167-9:1039541:1039971 [4] NCCL INFO Channel 11/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-167-9:1039537:1039969 [0] NCCL INFO Channel 18/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-167-9:1039544:1039972 [7] NCCL INFO Channel 16/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-167-9:1039543:1039970 [6] NCCL INFO Channel 11/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-167-9:1039538:1039973 [1] NCCL INFO Channel 08/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-167-9:1039539:1039975 [2] NCCL INFO Channel 13/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-167-9:1039542:1039976 [5] NCCL INFO Channel 15/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-167-9:1039540:1039974 [3] NCCL INFO Channel 11/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-167-9:1039537:1039969 [0] NCCL INFO Channel 19/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-167-9:1039541:1039971 [4] NCCL INFO Channel 12/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-167-9:1039544:1039972 [7] NCCL INFO Channel 17/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-167-9:1039539:1039975 [2] NCCL INFO Channel 14/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-167-9:1039543:1039970 [6] NCCL INFO Channel 12/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-167-9:1039542:1039976 [5] NCCL INFO Channel 16/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-167-9:1039538:1039973 [1] NCCL INFO Channel 09/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-167-9:1039540:1039974 [3] NCCL INFO Channel 12/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-214:624944:625379 [1] NCCL INFO Channel 12/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-167-9:1039537:1039969 [0] NCCL INFO Channel 20/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-167-9:1039541:1039971 [4] NCCL INFO Channel 13/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-167-9:1039544:1039972 [7] NCCL INFO Channel 18/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-167-9:1039539:1039975 [2] NCCL INFO Channel 15/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-167-9:1039543:1039970 [6] NCCL INFO Channel 13/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-167-9:1039542:1039976 [5] NCCL INFO Channel 17/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-167-9:1039538:1039973 [1] NCCL INFO Channel 10/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-167-9:1039540:1039974 [3] NCCL INFO Channel 13/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-244:897241:897675 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-167-9:1039537:1039969 [0] NCCL INFO Channel 21/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-167-9:1039541:1039971 [4] NCCL INFO Channel 14/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-167-9:1039544:1039972 [7] NCCL INFO Channel 19/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-167-9:1039543:1039970 [6] NCCL INFO Channel 14/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-167-9:1039542:1039976 [5] NCCL INFO Channel 18/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-167-9:1039539:1039975 [2] NCCL INFO Channel 16/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-167-9:1039538:1039973 [1] NCCL INFO Channel 11/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-167-9:1039540:1039974 [3] NCCL INFO Channel 14/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-244:897241:897675 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-162-46:702885:703316 [5] NCCL INFO Channel 00/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-162-46:702887:703317 [7] NCCL INFO Channel 00/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-167-9:1039537:1039969 [0] NCCL INFO Channel 22/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-167-9:1039541:1039971 [4] NCCL INFO Channel 15/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-167-9:1039544:1039972 [7] NCCL INFO Channel 20/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-167-9:1039543:1039970 [6] NCCL INFO Channel 15/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-167-9:1039542:1039976 [5] NCCL INFO Channel 19/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-167-9:1039539:1039975 [2] NCCL INFO Channel 17/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-167-9:1039538:1039973 [1] NCCL INFO Channel 12/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-167-9:1039540:1039974 [3] NCCL INFO Channel 15/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-162-46:702880:703314 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-162-46:702885:703316 [5] NCCL INFO Channel 01/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-162-46:702887:703317 [7] NCCL INFO Channel 01/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-162-46:702884:703319 [4] NCCL INFO Channel 00/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-162-46:702880:703314 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-162-46:702885:703316 [5] NCCL INFO Channel 02/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-162-46:702887:703317 [7] NCCL INFO Channel 02/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-167-9:1039541:1039971 [4] NCCL INFO Channel 16/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-167-9:1039537:1039969 [0] NCCL INFO Channel 23/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-167-9:1039544:1039972 [7] NCCL INFO Channel 21/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-167-9:1039543:1039970 [6] NCCL INFO Channel 16/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-167-9:1039542:1039976 [5] NCCL INFO Channel 20/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-167-9:1039538:1039973 [1] NCCL INFO Channel 13/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-167-9:1039539:1039975 [2] NCCL INFO Channel 18/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-167-9:1039540:1039974 [3] NCCL INFO Channel 16/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-162-46:702884:703319 [4] NCCL INFO Channel 01/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-162-46:702880:703314 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-162-46:702885:703316 [5] NCCL INFO Channel 03/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-162-46:702887:703317 [7] NCCL INFO Channel 03/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-162-46:702884:703319 [4] NCCL INFO Channel 02/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-162-46:702880:703314 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-162-46:702885:703316 [5] NCCL INFO Channel 04/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-162-46:702887:703317 [7] NCCL INFO Channel 04/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-167-9:1039541:1039971 [4] NCCL INFO Channel 17/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-167-9:1039544:1039972 [7] NCCL INFO Channel 22/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-167-9:1039543:1039970 [6] NCCL INFO Channel 17/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-167-9:1039542:1039976 [5] NCCL INFO Channel 21/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-167-9:1039539:1039975 [2] NCCL INFO Channel 19/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-167-9:1039538:1039973 [1] NCCL INFO Channel 14/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-167-9:1039540:1039974 [3] NCCL INFO Channel 17/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-162-46:702880:703314 [0] NCCL INFO Channel 04/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-162-46:702884:703319 [4] NCCL INFO Channel 03/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-162-46:702885:703316 [5] NCCL INFO Channel 05/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-162-46:702887:703317 [7] NCCL INFO Channel 05/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-162-46:702880:703314 [0] NCCL INFO Channel 05/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-162-46:702884:703319 [4] NCCL INFO Channel 04/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-166-214:624944:625379 [1] NCCL INFO Channel 13/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-167-9:1039542:1039976 [5] NCCL INFO Channel 22/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-167-9:1039539:1039975 [2] NCCL INFO Channel 20/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-162-46:702885:703316 [5] NCCL INFO Channel 06/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-162-46:702887:703317 [7] NCCL INFO Channel 06/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-167-9:1039540:1039974 [3] NCCL INFO Channel 18/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-167-9:1039538:1039973 [1] NCCL INFO Channel 15/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-162-46:702884:703319 [4] NCCL INFO Channel 05/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-167-9:1039539:1039975 [2] NCCL INFO Channel 21/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-167-9:1039542:1039976 [5] NCCL INFO Channel 23/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-162-46:702880:703314 [0] NCCL INFO Channel 06/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-162-46:702885:703316 [5] NCCL INFO Channel 07/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-162-46:702887:703317 [7] NCCL INFO Channel 07/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-162-46:702886:703315 [6] NCCL INFO Channel 00/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-162-46:702881:703320 [1] NCCL INFO Channel 00/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-167-9:1039538:1039973 [1] NCCL INFO Channel 16/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-162-46:702884:703319 [4] NCCL INFO Channel 06/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-162-46:702880:703314 [0] NCCL INFO Channel 07/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-167-9:1039543:1039970 [6] NCCL INFO Channel 18/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-167-9:1039541:1039971 [4] NCCL INFO Channel 18/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-167-9:1039544:1039972 [7] NCCL INFO Channel 23/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-167-9:1039538:1039973 [1] NCCL INFO Channel 17/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-162-46:702885:703316 [5] NCCL INFO Channel 08/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-162-46:702883:703318 [3] NCCL INFO Channel 00/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-162-46:702887:703317 [7] NCCL INFO Channel 08/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-167-9:1039543:1039970 [6] NCCL INFO Channel 19/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-167-9:1039541:1039971 [4] NCCL INFO Channel 19/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-162-46:702882:703321 [2] NCCL INFO Channel 00/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-167-9:1039538:1039973 [1] NCCL INFO Channel 18/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-162-46:702881:703320 [1] NCCL INFO Channel 01/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-162-46:702886:703315 [6] NCCL INFO Channel 01/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-162-46:702880:703314 [0] NCCL INFO Channel 08/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-167-9:1039540:1039974 [3] NCCL INFO Channel 19/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-167-9:1039543:1039970 [6] NCCL INFO Channel 20/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-162-46:702884:703319 [4] NCCL INFO Channel 07/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-167-9:1039541:1039971 [4] NCCL INFO Channel 20/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-167-9:1039538:1039973 [1] NCCL INFO Channel 19/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-162-46:702885:703316 [5] NCCL INFO Channel 09/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-167-9:1039540:1039974 [3] NCCL INFO Channel 20/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-167-9:1039543:1039970 [6] NCCL INFO Channel 21/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-167-9:1039541:1039971 [4] NCCL INFO Channel 21/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-162-46:702883:703318 [3] NCCL INFO Channel 01/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-162-46:702887:703317 [7] NCCL INFO Channel 09/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-162-46:702881:703320 [1] NCCL INFO Channel 02/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-162-46:702882:703321 [2] NCCL INFO Channel 01/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-167-9:1039538:1039973 [1] NCCL INFO Channel 20/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-162-46:702886:703315 [6] NCCL INFO Channel 02/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-162-46:702880:703314 [0] NCCL INFO Channel 09/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-162-46:702884:703319 [4] NCCL INFO Channel 08/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-167-9:1039540:1039974 [3] NCCL INFO Channel 21/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-167-9:1039541:1039971 [4] NCCL INFO Channel 22/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-167-9:1039543:1039970 [6] NCCL INFO Channel 22/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-167-9:1039539:1039975 [2] NCCL INFO Channel 22/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-162-46:702885:703316 [5] NCCL INFO Channel 10/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-162-46:702887:703317 [7] NCCL INFO Channel 10/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-167-9:1039538:1039973 [1] NCCL INFO Channel 21/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-162-46:702883:703318 [3] NCCL INFO Channel 02/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-167-9:1039540:1039974 [3] NCCL INFO Channel 22/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-167-9:1039541:1039971 [4] NCCL INFO Channel 23/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-167-9:1039543:1039970 [6] NCCL INFO Channel 23/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-167-9:1039539:1039975 [2] NCCL INFO Channel 23/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-162-46:702881:703320 [1] NCCL INFO Channel 03/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-162-46:702882:703321 [2] NCCL INFO Channel 02/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-162-46:702886:703315 [6] NCCL INFO Channel 03/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-162-46:702880:703314 [0] NCCL INFO Channel 10/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-162-46:702884:703319 [4] NCCL INFO Channel 09/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-167-9:1039538:1039973 [1] NCCL INFO Channel 22/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-167-9:1039540:1039974 [3] NCCL INFO Channel 23/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-162-46:702885:703316 [5] NCCL INFO Channel 11/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-162-46:702887:703317 [7] NCCL INFO Channel 11/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-162-46:702883:703318 [3] NCCL INFO Channel 03/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-167-9:1039538:1039973 [1] NCCL INFO Channel 23/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-162-46:702881:703320 [1] NCCL INFO Channel 04/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-162-46:702882:703321 [2] NCCL INFO Channel 03/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-162-46:702886:703315 [6] NCCL INFO Channel 04/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-162-46:702880:703314 [0] NCCL INFO Channel 11/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-162-46:702884:703319 [4] NCCL INFO Channel 10/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-162-46:702885:703316 [5] NCCL INFO Channel 12/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-162-46:702887:703317 [7] NCCL INFO Channel 12/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-162-46:702883:703318 [3] NCCL INFO Channel 04/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-162-46:702881:703320 [1] NCCL INFO Channel 05/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-162-46:702882:703321 [2] NCCL INFO Channel 04/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-162-46:702886:703315 [6] NCCL INFO Channel 05/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-162-46:702880:703314 [0] NCCL INFO Channel 12/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-162-46:702884:703319 [4] NCCL INFO Channel 11/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-162-46:702885:703316 [5] NCCL INFO Channel 13/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-162-46:702887:703317 [7] NCCL INFO Channel 13/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-162-46:702883:703318 [3] NCCL INFO Channel 05/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-162-46:702881:703320 [1] NCCL INFO Channel 06/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-162-46:702886:703315 [6] NCCL INFO Channel 06/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-162-46:702880:703314 [0] NCCL INFO Channel 13/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-162-46:702882:703321 [2] NCCL INFO Channel 05/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-162-46:702884:703319 [4] NCCL INFO Channel 12/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-162-46:702885:703316 [5] NCCL INFO Channel 14/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-162-46:702887:703317 [7] NCCL INFO Channel 14/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-162-46:702883:703318 [3] NCCL INFO Channel 06/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-162-46:702881:703320 [1] NCCL INFO Channel 07/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-162-46:702886:703315 [6] NCCL INFO Channel 07/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-162-46:702880:703314 [0] NCCL INFO Channel 14/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-162-46:702882:703321 [2] NCCL INFO Channel 06/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-162-46:702884:703319 [4] NCCL INFO Channel 13/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-162-46:702885:703316 [5] NCCL INFO Channel 15/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-162-46:702887:703317 [7] NCCL INFO Channel 15/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-162-46:702883:703318 [3] NCCL INFO Channel 07/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-214:624944:625379 [1] NCCL INFO Channel 14/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-162-46:702881:703320 [1] NCCL INFO Channel 08/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-162-46:702880:703314 [0] NCCL INFO Channel 15/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-162-46:702886:703315 [6] NCCL INFO Channel 08/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-162-46:702882:703321 [2] NCCL INFO Channel 07/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-162-46:702884:703319 [4] NCCL INFO Channel 14/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-162-46:702885:703316 [5] NCCL INFO Channel 16/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-162-46:702887:703317 [7] NCCL INFO Channel 16/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-162-46:702883:703318 [3] NCCL INFO Channel 08/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-162-46:702881:703320 [1] NCCL INFO Channel 09/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-162-46:702880:703314 [0] NCCL INFO Channel 16/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-162-46:702882:703321 [2] NCCL INFO Channel 08/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-162-46:702886:703315 [6] NCCL INFO Channel 09/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-15:127295:127726 [5] NCCL INFO Channel 00/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-162-46:702884:703319 [4] NCCL INFO Channel 15/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-162-46:702885:703316 [5] NCCL INFO Channel 17/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-162-46:702887:703317 [7] NCCL INFO Channel 17/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-162-46:702883:703318 [3] NCCL INFO Channel 09/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-15:127294:127725 [4] NCCL INFO Channel 00/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-166-15:127291:127730 [1] NCCL INFO Channel 00/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-166-15:127295:127726 [5] NCCL INFO Channel 01/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-162-46:702881:703320 [1] NCCL INFO Channel 10/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-162-46:702880:703314 [0] NCCL INFO Channel 17/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-162-46:702882:703321 [2] NCCL INFO Channel 09/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-162-46:702886:703315 [6] NCCL INFO Channel 10/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-162-46:702884:703319 [4] NCCL INFO Channel 16/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-162-46:702885:703316 [5] NCCL INFO Channel 18/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-162-46:702887:703317 [7] NCCL INFO Channel 18/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-162-46:702883:703318 [3] NCCL INFO Channel 10/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-15:127294:127725 [4] NCCL INFO Channel 01/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-166-15:127291:127730 [1] NCCL INFO Channel 01/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-166-15:127295:127726 [5] NCCL INFO Channel 02/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-162-46:702881:703320 [1] NCCL INFO Channel 11/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-162-46:702880:703314 [0] NCCL INFO Channel 18/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-162-46:702882:703321 [2] NCCL INFO Channel 10/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-162-46:702886:703315 [6] NCCL INFO Channel 11/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-162-46:702884:703319 [4] NCCL INFO Channel 17/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-162-46:702885:703316 [5] NCCL INFO Channel 19/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-162-46:702887:703317 [7] NCCL INFO Channel 19/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-162-46:702883:703318 [3] NCCL INFO Channel 11/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-15:127297:127731 [7] NCCL INFO Channel 00/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127294:127725 [4] NCCL INFO Channel 02/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-166-15:127292:127729 [2] NCCL INFO Channel 00/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-15:127291:127730 [1] NCCL INFO Channel 02/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-166-15:127295:127726 [5] NCCL INFO Channel 03/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-162-46:702886:703315 [6] NCCL INFO Channel 12/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-162-46:702882:703321 [2] NCCL INFO Channel 11/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-162-46:702881:703320 [1] NCCL INFO Channel 12/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-162-46:702880:703314 [0] NCCL INFO Channel 19/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-162-46:702884:703319 [4] NCCL INFO Channel 18/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-162-46:702885:703316 [5] NCCL INFO Channel 20/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-162-46:702887:703317 [7] NCCL INFO Channel 20/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-162-46:702883:703318 [3] NCCL INFO Channel 12/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-15:127297:127731 [7] NCCL INFO Channel 01/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127294:127725 [4] NCCL INFO Channel 03/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-166-15:127292:127729 [2] NCCL INFO Channel 01/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-15:127291:127730 [1] NCCL INFO Channel 03/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-166-15:127295:127726 [5] NCCL INFO Channel 04/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-162-46:702886:703315 [6] NCCL INFO Channel 13/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-162-46:702882:703321 [2] NCCL INFO Channel 12/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-162-46:702880:703314 [0] NCCL INFO Channel 20/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-162-46:702881:703320 [1] NCCL INFO Channel 13/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-162-46:702884:703319 [4] NCCL INFO Channel 19/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-162-46:702885:703316 [5] NCCL INFO Channel 21/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-162-46:702887:703317 [7] NCCL INFO Channel 21/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-162-46:702883:703318 [3] NCCL INFO Channel 13/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-15:127296:127727 [6] NCCL INFO Channel 00/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-15:127297:127731 [7] NCCL INFO Channel 02/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127294:127725 [4] NCCL INFO Channel 04/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-166-15:127292:127729 [2] NCCL INFO Channel 02/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-15:127291:127730 [1] NCCL INFO Channel 04/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-166-244:897245:897676 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-166-244:897245:897676 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-166-15:127295:127726 [5] NCCL INFO Channel 05/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-166-15:127290:127724 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-162-46:702886:703315 [6] NCCL INFO Channel 14/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-162-46:702882:703321 [2] NCCL INFO Channel 13/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-162-46:702881:703320 [1] NCCL INFO Channel 14/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-162-46:702880:703314 [0] NCCL INFO Channel 21/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-162-46:702887:703317 [7] NCCL INFO Channel 22/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-162-46:702884:703319 [4] NCCL INFO Channel 20/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-162-46:702885:703316 [5] NCCL INFO Channel 22/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-162-46:702883:703318 [3] NCCL INFO Channel 14/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-15:127297:127731 [7] NCCL INFO Channel 03/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127296:127727 [6] NCCL INFO Channel 01/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-15:127294:127725 [4] NCCL INFO Channel 05/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-166-15:127292:127729 [2] NCCL INFO Channel 03/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-15:127295:127726 [5] NCCL INFO Channel 06/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-166-15:127291:127730 [1] NCCL INFO Channel 05/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-166-15:127290:127724 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-162-46:702886:703315 [6] NCCL INFO Channel 15/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-162-46:702882:703321 [2] NCCL INFO Channel 14/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-162-46:702881:703320 [1] NCCL INFO Channel 15/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-162-46:702880:703314 [0] NCCL INFO Channel 22/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-162-46:702884:703319 [4] NCCL INFO Channel 21/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-162-46:702887:703317 [7] NCCL INFO Channel 23/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-162-46:702883:703318 [3] NCCL INFO Channel 15/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-15:127296:127727 [6] NCCL INFO Channel 02/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-15:127297:127731 [7] NCCL INFO Channel 04/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127294:127725 [4] NCCL INFO Channel 06/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-166-15:127292:127729 [2] NCCL INFO Channel 04/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-162-46:702885:703316 [5] NCCL INFO Channel 23/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-166-15:127291:127730 [1] NCCL INFO Channel 06/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-166-15:127295:127726 [5] NCCL INFO Channel 07/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-166-15:127290:127724 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-162-46:702886:703315 [6] NCCL INFO Channel 16/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-162-46:702882:703321 [2] NCCL INFO Channel 15/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-162-46:702881:703320 [1] NCCL INFO Channel 16/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-162-46:702880:703314 [0] NCCL INFO Channel 23/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-162-46:702884:703319 [4] NCCL INFO Channel 22/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-162-46:702883:703318 [3] NCCL INFO Channel 16/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-15:127296:127727 [6] NCCL INFO Channel 03/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-15:127297:127731 [7] NCCL INFO Channel 05/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127293:127728 [3] NCCL INFO Channel 00/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-15:127294:127725 [4] NCCL INFO Channel 07/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-166-15:127292:127729 [2] NCCL INFO Channel 05/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-15:127291:127730 [1] NCCL INFO Channel 07/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-166-15:127295:127726 [5] NCCL INFO Channel 08/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-162-46:702886:703315 [6] NCCL INFO Channel 17/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-15:127290:127724 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-162-46:702882:703321 [2] NCCL INFO Channel 16/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-162-46:702881:703320 [1] NCCL INFO Channel 17/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-162-46:702883:703318 [3] NCCL INFO Channel 17/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-214:624944:625379 [1] NCCL INFO Channel 15/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-162-46:702884:703319 [4] NCCL INFO Channel 23/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-162-46:702886:703315 [6] NCCL INFO Channel 18/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-162-46:702882:703321 [2] NCCL INFO Channel 17/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-162-46:702881:703320 [1] NCCL INFO Channel 18/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-166-15:127296:127727 [6] NCCL INFO Channel 04/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-15:127297:127731 [7] NCCL INFO Channel 06/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127294:127725 [4] NCCL INFO Channel 08/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-166-15:127293:127728 [3] NCCL INFO Channel 01/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-162-46:702883:703318 [3] NCCL INFO Channel 18/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-15:127291:127730 [1] NCCL INFO Channel 08/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-166-15:127292:127729 [2] NCCL INFO Channel 06/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-15:127295:127726 [5] NCCL INFO Channel 09/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-166-15:127290:127724 [0] NCCL INFO Channel 04/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-162-46:702886:703315 [6] NCCL INFO Channel 19/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-162-46:702882:703321 [2] NCCL INFO Channel 18/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-162-46:702881:703320 [1] NCCL INFO Channel 19/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-162-46:702883:703318 [3] NCCL INFO Channel 19/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-162-46:702886:703315 [6] NCCL INFO Channel 20/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-162-46:702882:703321 [2] NCCL INFO Channel 19/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-162-46:702881:703320 [1] NCCL INFO Channel 20/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-166-15:127296:127727 [6] NCCL INFO Channel 05/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-162-46:702883:703318 [3] NCCL INFO Channel 20/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-15:127297:127731 [7] NCCL INFO Channel 07/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127294:127725 [4] NCCL INFO Channel 09/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-166-15:127292:127729 [2] NCCL INFO Channel 07/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-15:127293:127728 [3] NCCL INFO Channel 02/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-15:127295:127726 [5] NCCL INFO Channel 10/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-166-15:127291:127730 [1] NCCL INFO Channel 09/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-166-15:127290:127724 [0] NCCL INFO Channel 05/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-162-46:702882:703321 [2] NCCL INFO Channel 20/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-162-46:702881:703320 [1] NCCL INFO Channel 21/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-162-46:702886:703315 [6] NCCL INFO Channel 21/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-162-46:702883:703318 [3] NCCL INFO Channel 21/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-15:127296:127727 [6] NCCL INFO Channel 06/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-162-46:702881:703320 [1] NCCL INFO Channel 22/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-162-46:702882:703321 [2] NCCL INFO Channel 21/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-162-46:702886:703315 [6] NCCL INFO Channel 22/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-162-46:702883:703318 [3] NCCL INFO Channel 22/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-15:127297:127731 [7] NCCL INFO Channel 08/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127294:127725 [4] NCCL INFO Channel 10/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-166-15:127292:127729 [2] NCCL INFO Channel 08/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-15:127295:127726 [5] NCCL INFO Channel 11/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-166-15:127293:127728 [3] NCCL INFO Channel 03/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-15:127291:127730 [1] NCCL INFO Channel 10/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-166-15:127290:127724 [0] NCCL INFO Channel 06/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-244:897243:897678 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-166-15:127296:127727 [6] NCCL INFO Channel 07/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-15:127297:127731 [7] NCCL INFO Channel 09/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127294:127725 [4] NCCL INFO Channel 11/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-166-15:127292:127729 [2] NCCL INFO Channel 09/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-15:127295:127726 [5] NCCL INFO Channel 12/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-166-15:127293:127728 [3] NCCL INFO Channel 04/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-162-46:702886:703315 [6] NCCL INFO Channel 23/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-162-46:702883:703318 [3] NCCL INFO Channel 23/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-15:127291:127730 [1] NCCL INFO Channel 11/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-166-15:127290:127724 [0] NCCL INFO Channel 07/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-162-46:702881:703320 [1] NCCL INFO Channel 23/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-162-46:702882:703321 [2] NCCL INFO Channel 22/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-244:897243:897678 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-166-214:624944:625379 [1] NCCL INFO Channel 16/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-162-46:702882:703321 [2] NCCL INFO Channel 23/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-15:127296:127727 [6] NCCL INFO Channel 08/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-15:127297:127731 [7] NCCL INFO Channel 10/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127294:127725 [4] NCCL INFO Channel 12/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-166-15:127292:127729 [2] NCCL INFO Channel 10/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-15:127295:127726 [5] NCCL INFO Channel 13/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-166-15:127293:127728 [3] NCCL INFO Channel 05/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-15:127290:127724 [0] NCCL INFO Channel 08/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-15:127291:127730 [1] NCCL INFO Channel 12/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-166-15:127296:127727 [6] NCCL INFO Channel 09/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-213:9626:10061 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-165-213:9626:10061 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-166-15:127297:127731 [7] NCCL INFO Channel 11/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127294:127725 [4] NCCL INFO Channel 13/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-166-15:127292:127729 [2] NCCL INFO Channel 11/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-15:127293:127728 [3] NCCL INFO Channel 06/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-15:127295:127726 [5] NCCL INFO Channel 14/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-166-15:127290:127724 [0] NCCL INFO Channel 09/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-15:127291:127730 [1] NCCL INFO Channel 13/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-166-15:127296:127727 [6] NCCL INFO Channel 10/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-15:127297:127731 [7] NCCL INFO Channel 12/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127294:127725 [4] NCCL INFO Channel 14/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-166-15:127292:127729 [2] NCCL INFO Channel 12/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-15:127293:127728 [3] NCCL INFO Channel 07/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-15:127290:127724 [0] NCCL INFO Channel 10/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-15:127295:127726 [5] NCCL INFO Channel 15/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-166-15:127291:127730 [1] NCCL INFO Channel 14/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-213:9628:10066 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-165-213:9628:10066 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-166-15:127296:127727 [6] NCCL INFO Channel 11/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-15:127297:127731 [7] NCCL INFO Channel 13/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127294:127725 [4] NCCL INFO Channel 15/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-166-15:127292:127729 [2] NCCL INFO Channel 13/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-15:127293:127728 [3] NCCL INFO Channel 08/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-15:127290:127724 [0] NCCL INFO Channel 11/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-15:127295:127726 [5] NCCL INFO Channel 16/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-166-15:127291:127730 [1] NCCL INFO Channel 15/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-166-15:127296:127727 [6] NCCL INFO Channel 12/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-15:127297:127731 [7] NCCL INFO Channel 14/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127294:127725 [4] NCCL INFO Channel 16/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-166-15:127292:127729 [2] NCCL INFO Channel 14/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-15:127293:127728 [3] NCCL INFO Channel 09/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-15:127290:127724 [0] NCCL INFO Channel 12/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-15:127295:127726 [5] NCCL INFO Channel 17/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-166-15:127291:127730 [1] NCCL INFO Channel 16/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-166-214:624944:625379 [1] NCCL INFO Channel 17/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-166-15:127296:127727 [6] NCCL INFO Channel 13/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-15:127294:127725 [4] NCCL INFO Channel 17/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-166-15:127292:127729 [2] NCCL INFO Channel 15/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-15:127297:127731 [7] NCCL INFO Channel 15/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127290:127724 [0] NCCL INFO Channel 13/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-15:127293:127728 [3] NCCL INFO Channel 10/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-15:127295:127726 [5] NCCL INFO Channel 18/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-166-15:127291:127730 [1] NCCL INFO Channel 17/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-166-15:127296:127727 [6] NCCL INFO Channel 14/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-15:127294:127725 [4] NCCL INFO Channel 18/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-166-15:127292:127729 [2] NCCL INFO Channel 16/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-15:127297:127731 [7] NCCL INFO Channel 16/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127290:127724 [0] NCCL INFO Channel 14/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-15:127293:127728 [3] NCCL INFO Channel 11/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-15:127295:127726 [5] NCCL INFO Channel 19/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-166-15:127291:127730 [1] NCCL INFO Channel 18/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-166-214:624944:625379 [1] NCCL INFO Channel 18/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-166-15:127290:127724 [0] NCCL INFO Channel 15/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-15:127295:127726 [5] NCCL INFO Channel 20/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-166-15:127296:127727 [6] NCCL INFO Channel 15/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-15:127293:127728 [3] NCCL INFO Channel 12/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-15:127294:127725 [4] NCCL INFO Channel 19/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-166-15:127292:127729 [2] NCCL INFO Channel 17/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-15:127290:127724 [0] NCCL INFO Channel 16/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-15:127291:127730 [1] NCCL INFO Channel 19/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-166-15:127295:127726 [5] NCCL INFO Channel 21/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-166-15:127296:127727 [6] NCCL INFO Channel 16/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-15:127293:127728 [3] NCCL INFO Channel 13/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-15:127294:127725 [4] NCCL INFO Channel 20/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-166-15:127292:127729 [2] NCCL INFO Channel 18/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-15:127290:127724 [0] NCCL INFO Channel 17/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-15:127291:127730 [1] NCCL INFO Channel 20/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-166-15:127297:127731 [7] NCCL INFO Channel 17/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127295:127726 [5] NCCL INFO Channel 22/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-166-15:127296:127727 [6] NCCL INFO Channel 17/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-15:127293:127728 [3] NCCL INFO Channel 14/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-15:127294:127725 [4] NCCL INFO Channel 21/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-166-15:127292:127729 [2] NCCL INFO Channel 19/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-15:127290:127724 [0] NCCL INFO Channel 18/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-15:127291:127730 [1] NCCL INFO Channel 21/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-166-15:127297:127731 [7] NCCL INFO Channel 18/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127292:127729 [2] NCCL INFO Channel 20/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-15:127294:127725 [4] NCCL INFO Channel 22/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-166-15:127290:127724 [0] NCCL INFO Channel 19/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-15:127297:127731 [7] NCCL INFO Channel 19/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127292:127729 [2] NCCL INFO Channel 21/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-15:127294:127725 [4] NCCL INFO Channel 23/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-166-15:127290:127724 [0] NCCL INFO Channel 20/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-15:127296:127727 [6] NCCL INFO Channel 18/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-15:127297:127731 [7] NCCL INFO Channel 20/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127292:127729 [2] NCCL INFO Channel 22/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-15:127290:127724 [0] NCCL INFO Channel 21/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-15:127297:127731 [7] NCCL INFO Channel 21/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127291:127730 [1] NCCL INFO Channel 22/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-166-15:127295:127726 [5] NCCL INFO Channel 23/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-166-15:127293:127728 [3] NCCL INFO Channel 15/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-15:127297:127731 [7] NCCL INFO Channel 22/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127291:127730 [1] NCCL INFO Channel 23/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-166-15:127293:127728 [3] NCCL INFO Channel 16/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-214:624944:625379 [1] NCCL INFO Channel 19/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-166-15:127296:127727 [6] NCCL INFO Channel 19/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-15:127297:127731 [7] NCCL INFO Channel 23/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127293:127728 [3] NCCL INFO Channel 17/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-15:127296:127727 [6] NCCL INFO Channel 20/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-15:127293:127728 [3] NCCL INFO Channel 18/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-15:127292:127729 [2] NCCL INFO Channel 23/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-15:127296:127727 [6] NCCL INFO Channel 21/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-15:127293:127728 [3] NCCL INFO Channel 19/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-15:127290:127724 [0] NCCL INFO Channel 22/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-15:127296:127727 [6] NCCL INFO Channel 22/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-15:127293:127728 [3] NCCL INFO Channel 20/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-15:127290:127724 [0] NCCL INFO Channel 23/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-15:127296:127727 [6] NCCL INFO Channel 23/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-15:127293:127728 [3] NCCL INFO Channel 21/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-15:127293:127728 [3] NCCL INFO Channel 22/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-214:624944:625379 [1] NCCL INFO Channel 20/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-166-15:127293:127728 [3] NCCL INFO Channel 23/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-213:9632:10065 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-165-213:9632:10065 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-166-244:897247:897677 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-166-244:897247:897677 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-166-244:897247:897677 [7] NCCL INFO Trees [0] -1/-1/-1->7->6 [1] -1/-1/-1->7->6 [2] -1/-1/-1->7->6 [3] -1/-1/-1->7->6 [4] -1/-1/-1->7->6 [5] -1/-1/-1->7->6 [6] -1/-1/-1->7->6 [7] -1/-1/-1->7->6 [8] -1/-1/-1->7->6 [9] -1/-1/-1->7->6 [10] -1/-1/-1->7->6 [11] -1/-1/-1->7->6 [12] -1/-1/-1->7->6 [13] -1/-1/-1->7->6 [14] -1/-1/-1->7->6 [15] -1/-1/-1->7->6 [16] -1/-1/-1->7->6 [17] -1/-1/-1->7->6 [18] -1/-1/-1->7->6 [19] -1/-1/-1->7->6 [20] -1/-1/-1->7->6 [21] -1/-1/-1->7->6 [22] -1/-1/-1->7->6 [23] -1/-1/-1->7->6
+ip-26-0-166-244:897247:897677 [7] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-166-244:897245:897676 [5] NCCL INFO Trees [0] 6/-1/-1->5->4 [1] 6/-1/-1->5->4 [2] 6/-1/-1->5->4 [3] 6/-1/-1->5->4 [4] 6/-1/-1->5->4 [5] 6/-1/-1->5->4 [6] 6/-1/-1->5->4 [7] 6/-1/-1->5->4 [8] 6/-1/-1->5->4 [9] 6/-1/-1->5->4 [10] 6/-1/-1->5->4 [11] 6/-1/-1->5->4 [12] 6/-1/-1->5->4 [13] 6/-1/-1->5->4 [14] 6/-1/-1->5->4 [15] 6/-1/-1->5->4 [16] 6/-1/-1->5->4 [17] 6/-1/-1->5->4 [18] 6/-1/-1->5->4 [19] 6/-1/-1->5->4 [20] 6/-1/-1->5->4 [21] 6/-1/-1->5->4 [22] 6/-1/-1->5->4 [23] 6/-1/-1->5->4
+ip-26-0-166-244:897244:897674 [4] NCCL INFO Trees [0] 5/-1/-1->4->3 [1] 5/-1/-1->4->3 [2] 5/-1/-1->4->3 [3] 5/-1/-1->4->3 [4] 5/-1/-1->4->3 [5] 5/-1/-1->4->3 [6] 5/-1/-1->4->3 [7] 5/-1/-1->4->3 [8] 5/-1/-1->4->3 [9] 5/-1/-1->4->3 [10] 5/-1/-1->4->3 [11] 5/-1/-1->4->3 [12] 5/-1/-1->4->3 [13] 5/-1/-1->4->3 [14] 5/-1/-1->4->3 [15] 5/-1/-1->4->3 [16] 5/-1/-1->4->3 [17] 5/-1/-1->4->3 [18] 5/-1/-1->4->3 [19] 5/-1/-1->4->3 [20] 5/-1/-1->4->3 [21] 5/-1/-1->4->3 [22] 5/-1/-1->4->3 [23] 5/-1/-1->4->3
+ip-26-0-166-244:897246:897680 [6] NCCL INFO Trees [0] 7/-1/-1->6->5 [1] 7/-1/-1->6->5 [2] 7/-1/-1->6->5 [3] 7/-1/-1->6->5 [4] 7/-1/-1->6->5 [5] 7/-1/-1->6->5 [6] 7/-1/-1->6->5 [7] 7/-1/-1->6->5 [8] 7/-1/-1->6->5 [9] 7/-1/-1->6->5 [10] 7/-1/-1->6->5 [11] 7/-1/-1->6->5 [12] 7/-1/-1->6->5 [13] 7/-1/-1->6->5 [14] 7/-1/-1->6->5 [15] 7/-1/-1->6->5 [16] 7/-1/-1->6->5 [17] 7/-1/-1->6->5 [18] 7/-1/-1->6->5 [19] 7/-1/-1->6->5 [20] 7/-1/-1->6->5 [21] 7/-1/-1->6->5 [22] 7/-1/-1->6->5 [23] 7/-1/-1->6->5
+ip-26-0-166-244:897245:897676 [5] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-166-244:897244:897674 [4] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-166-244:897243:897678 [3] NCCL INFO Trees [0] 4/-1/-1->3->2 [1] 4/-1/-1->3->2 [2] 4/-1/-1->3->2 [3] 4/-1/-1->3->2 [4] 4/-1/-1->3->2 [5] 4/-1/-1->3->2 [6] 4/-1/-1->3->2 [7] 4/-1/-1->3->2 [8] 4/-1/-1->3->2 [9] 4/-1/-1->3->2 [10] 4/-1/-1->3->2 [11] 4/-1/-1->3->2 [12] 4/-1/-1->3->2 [13] 4/-1/-1->3->2 [14] 4/-1/-1->3->2 [15] 4/-1/-1->3->2 [16] 4/-1/-1->3->2 [17] 4/-1/-1->3->2 [18] 4/-1/-1->3->2 [19] 4/-1/-1->3->2 [20] 4/-1/-1->3->2 [21] 4/-1/-1->3->2 [22] 4/-1/-1->3->2 [23] 4/-1/-1->3->2
+ip-26-0-166-244:897246:897680 [6] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-166-244:897243:897678 [3] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-166-244:897242:897679 [2] NCCL INFO Trees [0] 3/-1/-1->2->1 [1] 3/-1/-1->2->1 [2] 3/-1/-1->2->1 [3] 3/-1/-1->2->1 [4] 3/-1/-1->2->1 [5] 3/-1/-1->2->1 [6] 3/-1/-1->2->1 [7] 3/-1/-1->2->1 [8] 3/-1/-1->2->1 [9] 3/-1/-1->2->1 [10] 3/-1/-1->2->1 [11] 3/-1/-1->2->1 [12] 3/-1/-1->2->1 [13] 3/-1/-1->2->1 [14] 3/-1/-1->2->1 [15] 3/-1/-1->2->1 [16] 3/-1/-1->2->1 [17] 3/-1/-1->2->1 [18] 3/-1/-1->2->1 [19] 3/-1/-1->2->1 [20] 3/-1/-1->2->1 [21] 3/-1/-1->2->1 [22] 3/-1/-1->2->1 [23] 3/-1/-1->2->1
+ip-26-0-166-244:897240:897673 [0] NCCL INFO Channel 00/24 :    0   1   2   3   4   5   6   7
+ip-26-0-166-244:897241:897675 [1] NCCL INFO Trees [0] 2/-1/-1->1->0 [1] 2/-1/-1->1->0 [2] 2/-1/-1->1->0 [3] 2/-1/-1->1->0 [4] 2/-1/-1->1->0 [5] 2/-1/-1->1->0 [6] 2/-1/-1->1->0 [7] 2/-1/-1->1->0 [8] 2/-1/-1->1->0 [9] 2/-1/-1->1->0 [10] 2/-1/-1->1->0 [11] 2/-1/-1->1->0 [12] 2/-1/-1->1->0 [13] 2/-1/-1->1->0 [14] 2/-1/-1->1->0 [15] 2/-1/-1->1->0 [16] 2/-1/-1->1->0 [17] 2/-1/-1->1->0 [18] 2/-1/-1->1->0 [19] 2/-1/-1->1->0 [20] 2/-1/-1->1->0 [21] 2/-1/-1->1->0 [22] 2/-1/-1->1->0 [23] 2/-1/-1->1->0
+ip-26-0-166-244:897242:897679 [2] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-166-244:897240:897673 [0] NCCL INFO Channel 01/24 :    0   1   2   3   4   5   6   7
+ip-26-0-166-244:897241:897675 [1] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-166-244:897240:897673 [0] NCCL INFO Channel 02/24 :    0   1   2   3   4   5   6   7
+ip-26-0-166-244:897240:897673 [0] NCCL INFO Channel 03/24 :    0   1   2   3   4   5   6   7
+ip-26-0-166-244:897240:897673 [0] NCCL INFO Channel 04/24 :    0   1   2   3   4   5   6   7
+ip-26-0-166-244:897240:897673 [0] NCCL INFO Channel 05/24 :    0   1   2   3   4   5   6   7
+ip-26-0-166-244:897240:897673 [0] NCCL INFO Channel 06/24 :    0   1   2   3   4   5   6   7
+ip-26-0-166-244:897240:897673 [0] NCCL INFO Channel 07/24 :    0   1   2   3   4   5   6   7
+ip-26-0-166-244:897240:897673 [0] NCCL INFO Channel 08/24 :    0   1   2   3   4   5   6   7
+ip-26-0-166-244:897240:897673 [0] NCCL INFO Channel 09/24 :    0   1   2   3   4   5   6   7
+ip-26-0-166-244:897240:897673 [0] NCCL INFO Channel 10/24 :    0   1   2   3   4   5   6   7
+ip-26-0-166-244:897240:897673 [0] NCCL INFO Channel 11/24 :    0   1   2   3   4   5   6   7
+ip-26-0-166-244:897240:897673 [0] NCCL INFO Channel 12/24 :    0   1   2   3   4   5   6   7
+ip-26-0-166-244:897240:897673 [0] NCCL INFO Channel 13/24 :    0   1   2   3   4   5   6   7
+ip-26-0-166-244:897240:897673 [0] NCCL INFO Channel 14/24 :    0   1   2   3   4   5   6   7
+ip-26-0-166-244:897240:897673 [0] NCCL INFO Channel 15/24 :    0   1   2   3   4   5   6   7
+ip-26-0-166-244:897240:897673 [0] NCCL INFO Channel 16/24 :    0   1   2   3   4   5   6   7
+ip-26-0-166-244:897240:897673 [0] NCCL INFO Channel 17/24 :    0   1   2   3   4   5   6   7
+ip-26-0-166-244:897240:897673 [0] NCCL INFO Channel 18/24 :    0   1   2   3   4   5   6   7
+ip-26-0-166-244:897240:897673 [0] NCCL INFO Channel 19/24 :    0   1   2   3   4   5   6   7
+ip-26-0-166-244:897240:897673 [0] NCCL INFO Channel 20/24 :    0   1   2   3   4   5   6   7
+ip-26-0-166-244:897240:897673 [0] NCCL INFO Channel 21/24 :    0   1   2   3   4   5   6   7
+ip-26-0-166-244:897240:897673 [0] NCCL INFO Channel 22/24 :    0   1   2   3   4   5   6   7
+ip-26-0-166-244:897240:897673 [0] NCCL INFO Channel 23/24 :    0   1   2   3   4   5   6   7
+ip-26-0-166-244:897240:897673 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] 1/-1/-1->0->-1 [7] 1/-1/-1->0->-1 [8] 1/-1/-1->0->-1 [9] 1/-1/-1->0->-1 [10] 1/-1/-1->0->-1 [11] 1/-1/-1->0->-1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] 1/-1/-1->0->-1 [19] 1/-1/-1->0->-1 [20] 1/-1/-1->0->-1 [21] 1/-1/-1->0->-1 [22] 1/-1/-1->0->-1 [23] 1/-1/-1->0->-1
+ip-26-0-166-244:897240:897673 [0] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-166-214:624944:625379 [1] NCCL INFO Channel 21/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-166-214:624944:625379 [1] NCCL INFO Channel 22/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-166-214:624944:625379 [1] NCCL INFO Channel 23/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-166-244:897247:897677 [7] NCCL INFO Channel 00/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-244:897246:897680 [6] NCCL INFO Channel 00/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-244:897240:897673 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-244:897241:897675 [1] NCCL INFO Channel 00/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-166-244:897247:897677 [7] NCCL INFO Channel 01/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-244:897244:897674 [4] NCCL INFO Channel 00/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-166-244:897240:897673 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-244:897246:897680 [6] NCCL INFO Channel 01/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-244:897241:897675 [1] NCCL INFO Channel 01/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-166-244:897247:897677 [7] NCCL INFO Channel 02/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-244:897244:897674 [4] NCCL INFO Channel 01/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-166-244:897240:897673 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-244:897246:897680 [6] NCCL INFO Channel 02/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-244:897241:897675 [1] NCCL INFO Channel 02/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-166-244:897247:897677 [7] NCCL INFO Channel 03/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-244:897244:897674 [4] NCCL INFO Channel 02/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-166-244:897246:897680 [6] NCCL INFO Channel 03/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-244:897240:897673 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-244:897241:897675 [1] NCCL INFO Channel 03/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-166-244:897247:897677 [7] NCCL INFO Channel 04/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-244:897244:897674 [4] NCCL INFO Channel 03/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-166-244:897246:897680 [6] NCCL INFO Channel 04/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-244:897240:897673 [0] NCCL INFO Channel 04/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-244:897241:897675 [1] NCCL INFO Channel 04/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-166-244:897247:897677 [7] NCCL INFO Channel 05/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-244:897244:897674 [4] NCCL INFO Channel 04/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-166-244:897246:897680 [6] NCCL INFO Channel 05/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-244:897240:897673 [0] NCCL INFO Channel 05/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-244:897241:897675 [1] NCCL INFO Channel 05/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-166-244:897247:897677 [7] NCCL INFO Channel 06/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-244:897244:897674 [4] NCCL INFO Channel 05/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-166-244:897246:897680 [6] NCCL INFO Channel 06/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-244:897240:897673 [0] NCCL INFO Channel 06/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-244:897241:897675 [1] NCCL INFO Channel 06/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-166-244:897245:897676 [5] NCCL INFO Channel 00/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-166-244:897247:897677 [7] NCCL INFO Channel 07/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-244:897244:897674 [4] NCCL INFO Channel 06/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-166-244:897242:897679 [2] NCCL INFO Channel 00/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-244:897246:897680 [6] NCCL INFO Channel 07/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-244:897240:897673 [0] NCCL INFO Channel 07/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-244:897243:897678 [3] NCCL INFO Channel 00/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-244:897241:897675 [1] NCCL INFO Channel 07/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-166-244:897245:897676 [5] NCCL INFO Channel 01/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-166-244:897247:897677 [7] NCCL INFO Channel 08/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-244:897244:897674 [4] NCCL INFO Channel 07/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-166-244:897242:897679 [2] NCCL INFO Channel 01/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-244:897246:897680 [6] NCCL INFO Channel 08/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-244:897240:897673 [0] NCCL INFO Channel 08/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-244:897243:897678 [3] NCCL INFO Channel 01/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-244:897241:897675 [1] NCCL INFO Channel 08/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-166-244:897247:897677 [7] NCCL INFO Channel 09/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-244:897245:897676 [5] NCCL INFO Channel 02/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-166-244:897244:897674 [4] NCCL INFO Channel 08/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-166-244:897242:897679 [2] NCCL INFO Channel 02/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-244:897240:897673 [0] NCCL INFO Channel 09/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-244:897246:897680 [6] NCCL INFO Channel 09/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-244:897241:897675 [1] NCCL INFO Channel 09/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-166-244:897243:897678 [3] NCCL INFO Channel 02/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-244:897247:897677 [7] NCCL INFO Channel 10/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-244:897245:897676 [5] NCCL INFO Channel 03/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-166-244:897244:897674 [4] NCCL INFO Channel 09/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-166-244:897242:897679 [2] NCCL INFO Channel 03/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-244:897240:897673 [0] NCCL INFO Channel 10/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-244:897246:897680 [6] NCCL INFO Channel 10/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-244:897241:897675 [1] NCCL INFO Channel 10/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-166-244:897243:897678 [3] NCCL INFO Channel 03/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-244:897247:897677 [7] NCCL INFO Channel 11/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-244:897244:897674 [4] NCCL INFO Channel 10/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-166-244:897245:897676 [5] NCCL INFO Channel 04/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-166-244:897242:897679 [2] NCCL INFO Channel 04/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-244:897240:897673 [0] NCCL INFO Channel 11/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-244:897243:897678 [3] NCCL INFO Channel 04/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-244:897246:897680 [6] NCCL INFO Channel 11/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-244:897241:897675 [1] NCCL INFO Channel 11/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-166-244:897247:897677 [7] NCCL INFO Channel 12/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-244:897244:897674 [4] NCCL INFO Channel 11/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-213:9625:10059 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-165-213:9625:10059 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-166-244:897245:897676 [5] NCCL INFO Channel 05/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-166-244:897240:897673 [0] NCCL INFO Channel 12/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-244:897242:897679 [2] NCCL INFO Channel 05/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-244:897246:897680 [6] NCCL INFO Channel 12/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-244:897247:897677 [7] NCCL INFO Channel 13/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-244:897243:897678 [3] NCCL INFO Channel 05/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-244:897241:897675 [1] NCCL INFO Channel 12/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-166-244:897244:897674 [4] NCCL INFO Channel 12/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-166-244:897245:897676 [5] NCCL INFO Channel 06/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-166-244:897242:897679 [2] NCCL INFO Channel 06/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-244:897240:897673 [0] NCCL INFO Channel 13/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-244:897247:897677 [7] NCCL INFO Channel 14/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-244:897246:897680 [6] NCCL INFO Channel 13/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-244:897243:897678 [3] NCCL INFO Channel 06/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-244:897241:897675 [1] NCCL INFO Channel 13/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-166-244:897244:897674 [4] NCCL INFO Channel 13/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-166-244:897245:897676 [5] NCCL INFO Channel 07/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-166-244:897242:897679 [2] NCCL INFO Channel 07/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-244:897240:897673 [0] NCCL INFO Channel 14/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-244:897244:897674 [4] NCCL INFO Channel 14/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-166-244:897246:897680 [6] NCCL INFO Channel 14/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-244:897247:897677 [7] NCCL INFO Channel 15/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-244:897243:897678 [3] NCCL INFO Channel 07/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-244:897241:897675 [1] NCCL INFO Channel 14/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-166-244:897245:897676 [5] NCCL INFO Channel 08/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-166-244:897242:897679 [2] NCCL INFO Channel 08/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-244:897240:897673 [0] NCCL INFO Channel 15/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-244:897244:897674 [4] NCCL INFO Channel 15/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-166-244:897246:897680 [6] NCCL INFO Channel 15/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-244:897247:897677 [7] NCCL INFO Channel 16/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-244:897241:897675 [1] NCCL INFO Channel 15/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-166-244:897243:897678 [3] NCCL INFO Channel 08/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-244:897245:897676 [5] NCCL INFO Channel 09/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-166-244:897242:897679 [2] NCCL INFO Channel 09/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-244:897240:897673 [0] NCCL INFO Channel 16/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-244:897244:897674 [4] NCCL INFO Channel 16/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-166-244:897246:897680 [6] NCCL INFO Channel 16/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-244:897247:897677 [7] NCCL INFO Channel 17/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-244:897241:897675 [1] NCCL INFO Channel 16/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-166-244:897245:897676 [5] NCCL INFO Channel 10/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-166-244:897243:897678 [3] NCCL INFO Channel 09/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-244:897242:897679 [2] NCCL INFO Channel 10/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-244:897240:897673 [0] NCCL INFO Channel 17/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-244:897244:897674 [4] NCCL INFO Channel 17/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-166-244:897247:897677 [7] NCCL INFO Channel 18/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-244:897246:897680 [6] NCCL INFO Channel 17/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-244:897241:897675 [1] NCCL INFO Channel 17/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-166-244:897245:897676 [5] NCCL INFO Channel 11/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-166-244:897243:897678 [3] NCCL INFO Channel 10/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-244:897242:897679 [2] NCCL INFO Channel 11/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-244:897240:897673 [0] NCCL INFO Channel 18/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-244:897244:897674 [4] NCCL INFO Channel 18/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-166-244:897247:897677 [7] NCCL INFO Channel 19/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-244:897246:897680 [6] NCCL INFO Channel 18/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-244:897241:897675 [1] NCCL INFO Channel 18/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-166-244:897245:897676 [5] NCCL INFO Channel 12/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-166-244:897243:897678 [3] NCCL INFO Channel 11/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-244:897242:897679 [2] NCCL INFO Channel 12/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-244:897240:897673 [0] NCCL INFO Channel 19/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-244:897247:897677 [7] NCCL INFO Channel 20/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-244:897244:897674 [4] NCCL INFO Channel 19/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-166-244:897246:897680 [6] NCCL INFO Channel 19/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-244:897245:897676 [5] NCCL INFO Channel 13/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-166-244:897241:897675 [1] NCCL INFO Channel 19/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-166-244:897243:897678 [3] NCCL INFO Channel 12/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-244:897242:897679 [2] NCCL INFO Channel 13/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-244:897240:897673 [0] NCCL INFO Channel 20/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-244:897247:897677 [7] NCCL INFO Channel 21/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-244:897244:897674 [4] NCCL INFO Channel 20/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-166-244:897246:897680 [6] NCCL INFO Channel 20/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-244:897245:897676 [5] NCCL INFO Channel 14/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-166-244:897241:897675 [1] NCCL INFO Channel 20/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-166-244:897243:897678 [3] NCCL INFO Channel 13/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-244:897242:897679 [2] NCCL INFO Channel 14/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-244:897240:897673 [0] NCCL INFO Channel 21/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-244:897244:897674 [4] NCCL INFO Channel 21/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-166-244:897247:897677 [7] NCCL INFO Channel 22/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-244:897246:897680 [6] NCCL INFO Channel 21/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-244:897245:897676 [5] NCCL INFO Channel 15/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-166-244:897241:897675 [1] NCCL INFO Channel 21/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-166-244:897243:897678 [3] NCCL INFO Channel 14/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-244:897242:897679 [2] NCCL INFO Channel 15/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-244:897240:897673 [0] NCCL INFO Channel 22/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-244:897244:897674 [4] NCCL INFO Channel 22/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-166-244:897247:897677 [7] NCCL INFO Channel 23/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-244:897245:897676 [5] NCCL INFO Channel 16/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-166-244:897246:897680 [6] NCCL INFO Channel 22/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-244:897241:897675 [1] NCCL INFO Channel 22/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-166-244:897243:897678 [3] NCCL INFO Channel 15/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-244:897242:897679 [2] NCCL INFO Channel 16/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-244:897240:897673 [0] NCCL INFO Channel 23/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-244:897244:897674 [4] NCCL INFO Channel 23/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-166-244:897245:897676 [5] NCCL INFO Channel 17/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-166-244:897246:897680 [6] NCCL INFO Channel 23/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-244:897241:897675 [1] NCCL INFO Channel 23/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-166-244:897243:897678 [3] NCCL INFO Channel 16/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-244:897242:897679 [2] NCCL INFO Channel 17/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-244:897245:897676 [5] NCCL INFO Channel 18/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-166-244:897243:897678 [3] NCCL INFO Channel 17/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-244:897242:897679 [2] NCCL INFO Channel 18/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-213:9630:10060 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-165-213:9630:10060 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-166-244:897245:897676 [5] NCCL INFO Channel 19/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-166-244:897243:897678 [3] NCCL INFO Channel 18/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-244:897242:897679 [2] NCCL INFO Channel 19/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-244:897245:897676 [5] NCCL INFO Channel 20/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-166-244:897243:897678 [3] NCCL INFO Channel 19/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-244:897242:897679 [2] NCCL INFO Channel 20/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-213:9631:10062 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-165-213:9631:10062 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-165-213:9629:10063 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-165-213:9629:10063 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-165-213:9629:10063 [4] NCCL INFO Trees [0] 5/-1/-1->4->3 [1] 5/-1/-1->4->3 [2] 5/-1/-1->4->3 [3] 5/-1/-1->4->3 [4] 5/-1/-1->4->3 [5] 5/-1/-1->4->3 [6] 5/-1/-1->4->3 [7] 5/-1/-1->4->3 [8] 5/-1/-1->4->3 [9] 5/-1/-1->4->3 [10] 5/-1/-1->4->3 [11] 5/-1/-1->4->3 [12] 5/-1/-1->4->3 [13] 5/-1/-1->4->3 [14] 5/-1/-1->4->3 [15] 5/-1/-1->4->3 [16] 5/-1/-1->4->3 [17] 5/-1/-1->4->3 [18] 5/-1/-1->4->3 [19] 5/-1/-1->4->3 [20] 5/-1/-1->4->3 [21] 5/-1/-1->4->3 [22] 5/-1/-1->4->3 [23] 5/-1/-1->4->3
+ip-26-0-165-213:9629:10063 [4] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-165-213:9630:10060 [5] NCCL INFO Trees [0] 6/-1/-1->5->4 [1] 6/-1/-1->5->4 [2] 6/-1/-1->5->4 [3] 6/-1/-1->5->4 [4] 6/-1/-1->5->4 [5] 6/-1/-1->5->4 [6] 6/-1/-1->5->4 [7] 6/-1/-1->5->4 [8] 6/-1/-1->5->4 [9] 6/-1/-1->5->4 [10] 6/-1/-1->5->4 [11] 6/-1/-1->5->4 [12] 6/-1/-1->5->4 [13] 6/-1/-1->5->4 [14] 6/-1/-1->5->4 [15] 6/-1/-1->5->4 [16] 6/-1/-1->5->4 [17] 6/-1/-1->5->4 [18] 6/-1/-1->5->4 [19] 6/-1/-1->5->4 [20] 6/-1/-1->5->4 [21] 6/-1/-1->5->4 [22] 6/-1/-1->5->4 [23] 6/-1/-1->5->4
+ip-26-0-165-213:9630:10060 [5] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-165-213:9631:10062 [6] NCCL INFO Trees [0] 7/-1/-1->6->5 [1] 7/-1/-1->6->5 [2] 7/-1/-1->6->5 [3] 7/-1/-1->6->5 [4] 7/-1/-1->6->5 [5] 7/-1/-1->6->5 [6] 7/-1/-1->6->5 [7] 7/-1/-1->6->5 [8] 7/-1/-1->6->5 [9] 7/-1/-1->6->5 [10] 7/-1/-1->6->5 [11] 7/-1/-1->6->5 [12] 7/-1/-1->6->5 [13] 7/-1/-1->6->5 [14] 7/-1/-1->6->5 [15] 7/-1/-1->6->5 [16] 7/-1/-1->6->5 [17] 7/-1/-1->6->5 [18] 7/-1/-1->6->5 [19] 7/-1/-1->6->5 [20] 7/-1/-1->6->5 [21] 7/-1/-1->6->5 [22] 7/-1/-1->6->5 [23] 7/-1/-1->6->5
+ip-26-0-165-213:9631:10062 [6] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-165-213:9632:10065 [7] NCCL INFO Trees [0] -1/-1/-1->7->6 [1] -1/-1/-1->7->6 [2] -1/-1/-1->7->6 [3] -1/-1/-1->7->6 [4] -1/-1/-1->7->6 [5] -1/-1/-1->7->6 [6] -1/-1/-1->7->6 [7] -1/-1/-1->7->6 [8] -1/-1/-1->7->6 [9] -1/-1/-1->7->6 [10] -1/-1/-1->7->6 [11] -1/-1/-1->7->6 [12] -1/-1/-1->7->6 [13] -1/-1/-1->7->6 [14] -1/-1/-1->7->6 [15] -1/-1/-1->7->6 [16] -1/-1/-1->7->6 [17] -1/-1/-1->7->6 [18] -1/-1/-1->7->6 [19] -1/-1/-1->7->6 [20] -1/-1/-1->7->6 [21] -1/-1/-1->7->6 [22] -1/-1/-1->7->6 [23] -1/-1/-1->7->6
+ip-26-0-165-213:9628:10066 [3] NCCL INFO Trees [0] 4/-1/-1->3->2 [1] 4/-1/-1->3->2 [2] 4/-1/-1->3->2 [3] 4/-1/-1->3->2 [4] 4/-1/-1->3->2 [5] 4/-1/-1->3->2 [6] 4/-1/-1->3->2 [7] 4/-1/-1->3->2 [8] 4/-1/-1->3->2 [9] 4/-1/-1->3->2 [10] 4/-1/-1->3->2 [11] 4/-1/-1->3->2 [12] 4/-1/-1->3->2 [13] 4/-1/-1->3->2 [14] 4/-1/-1->3->2 [15] 4/-1/-1->3->2 [16] 4/-1/-1->3->2 [17] 4/-1/-1->3->2 [18] 4/-1/-1->3->2 [19] 4/-1/-1->3->2 [20] 4/-1/-1->3->2 [21] 4/-1/-1->3->2 [22] 4/-1/-1->3->2 [23] 4/-1/-1->3->2
+ip-26-0-165-213:9625:10059 [0] NCCL INFO Channel 00/24 :    0   1   2   3   4   5   6   7
+ip-26-0-165-213:9628:10066 [3] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-165-213:9625:10059 [0] NCCL INFO Channel 01/24 :    0   1   2   3   4   5   6   7
+ip-26-0-165-213:9632:10065 [7] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-165-213:9625:10059 [0] NCCL INFO Channel 02/24 :    0   1   2   3   4   5   6   7
+ip-26-0-165-213:9626:10061 [1] NCCL INFO Trees [0] 2/-1/-1->1->0 [1] 2/-1/-1->1->0 [2] 2/-1/-1->1->0 [3] 2/-1/-1->1->0 [4] 2/-1/-1->1->0 [5] 2/-1/-1->1->0 [6] 2/-1/-1->1->0 [7] 2/-1/-1->1->0 [8] 2/-1/-1->1->0 [9] 2/-1/-1->1->0 [10] 2/-1/-1->1->0 [11] 2/-1/-1->1->0 [12] 2/-1/-1->1->0 [13] 2/-1/-1->1->0 [14] 2/-1/-1->1->0 [15] 2/-1/-1->1->0 [16] 2/-1/-1->1->0 [17] 2/-1/-1->1->0 [18] 2/-1/-1->1->0 [19] 2/-1/-1->1->0 [20] 2/-1/-1->1->0 [21] 2/-1/-1->1->0 [22] 2/-1/-1->1->0 [23] 2/-1/-1->1->0
+ip-26-0-165-213:9625:10059 [0] NCCL INFO Channel 03/24 :    0   1   2   3   4   5   6   7
+ip-26-0-165-213:9625:10059 [0] NCCL INFO Channel 04/24 :    0   1   2   3   4   5   6   7
+ip-26-0-165-213:9626:10061 [1] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-165-213:9625:10059 [0] NCCL INFO Channel 05/24 :    0   1   2   3   4   5   6   7
+ip-26-0-165-213:9625:10059 [0] NCCL INFO Channel 06/24 :    0   1   2   3   4   5   6   7
+ip-26-0-165-213:9625:10059 [0] NCCL INFO Channel 07/24 :    0   1   2   3   4   5   6   7
+ip-26-0-165-213:9625:10059 [0] NCCL INFO Channel 08/24 :    0   1   2   3   4   5   6   7
+ip-26-0-165-213:9625:10059 [0] NCCL INFO Channel 09/24 :    0   1   2   3   4   5   6   7
+ip-26-0-165-213:9625:10059 [0] NCCL INFO Channel 10/24 :    0   1   2   3   4   5   6   7
+ip-26-0-165-213:9625:10059 [0] NCCL INFO Channel 11/24 :    0   1   2   3   4   5   6   7
+ip-26-0-165-213:9625:10059 [0] NCCL INFO Channel 12/24 :    0   1   2   3   4   5   6   7
+ip-26-0-165-213:9625:10059 [0] NCCL INFO Channel 13/24 :    0   1   2   3   4   5   6   7
+ip-26-0-165-213:9625:10059 [0] NCCL INFO Channel 14/24 :    0   1   2   3   4   5   6   7
+ip-26-0-165-213:9625:10059 [0] NCCL INFO Channel 15/24 :    0   1   2   3   4   5   6   7
+ip-26-0-165-213:9625:10059 [0] NCCL INFO Channel 16/24 :    0   1   2   3   4   5   6   7
+ip-26-0-165-213:9625:10059 [0] NCCL INFO Channel 17/24 :    0   1   2   3   4   5   6   7
+ip-26-0-165-213:9625:10059 [0] NCCL INFO Channel 18/24 :    0   1   2   3   4   5   6   7
+ip-26-0-165-213:9627:10064 [2] NCCL INFO Trees [0] 3/-1/-1->2->1 [1] 3/-1/-1->2->1 [2] 3/-1/-1->2->1 [3] 3/-1/-1->2->1 [4] 3/-1/-1->2->1 [5] 3/-1/-1->2->1 [6] 3/-1/-1->2->1 [7] 3/-1/-1->2->1 [8] 3/-1/-1->2->1 [9] 3/-1/-1->2->1 [10] 3/-1/-1->2->1 [11] 3/-1/-1->2->1 [12] 3/-1/-1->2->1 [13] 3/-1/-1->2->1 [14] 3/-1/-1->2->1 [15] 3/-1/-1->2->1 [16] 3/-1/-1->2->1 [17] 3/-1/-1->2->1 [18] 3/-1/-1->2->1 [19] 3/-1/-1->2->1 [20] 3/-1/-1->2->1 [21] 3/-1/-1->2->1 [22] 3/-1/-1->2->1 [23] 3/-1/-1->2->1
+ip-26-0-165-213:9625:10059 [0] NCCL INFO Channel 19/24 :    0   1   2   3   4   5   6   7
+ip-26-0-165-213:9625:10059 [0] NCCL INFO Channel 20/24 :    0   1   2   3   4   5   6   7
+ip-26-0-165-213:9625:10059 [0] NCCL INFO Channel 21/24 :    0   1   2   3   4   5   6   7
+ip-26-0-165-213:9625:10059 [0] NCCL INFO Channel 22/24 :    0   1   2   3   4   5   6   7
+ip-26-0-165-213:9625:10059 [0] NCCL INFO Channel 23/24 :    0   1   2   3   4   5   6   7
+ip-26-0-165-213:9625:10059 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] 1/-1/-1->0->-1 [7] 1/-1/-1->0->-1 [8] 1/-1/-1->0->-1 [9] 1/-1/-1->0->-1 [10] 1/-1/-1->0->-1 [11] 1/-1/-1->0->-1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] 1/-1/-1->0->-1 [19] 1/-1/-1->0->-1 [20] 1/-1/-1->0->-1 [21] 1/-1/-1->0->-1 [22] 1/-1/-1->0->-1 [23] 1/-1/-1->0->-1
+ip-26-0-165-213:9625:10059 [0] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-165-213:9627:10064 [2] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-166-244:897245:897676 [5] NCCL INFO Channel 21/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-166-244:897243:897678 [3] NCCL INFO Channel 20/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-244:897242:897679 [2] NCCL INFO Channel 21/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-244:897242:897679 [2] NCCL INFO Channel 22/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-244:897245:897676 [5] NCCL INFO Channel 22/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-166-244:897242:897679 [2] NCCL INFO Channel 23/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-244:897243:897678 [3] NCCL INFO Channel 21/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-166-244:897245:897676 [5] NCCL INFO Channel 23/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-166-244:897243:897678 [3] NCCL INFO Channel 22/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-213:9630:10060 [5] NCCL INFO Channel 00/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-165-213:9626:10061 [1] NCCL INFO Channel 00/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-213:9628:10066 [3] NCCL INFO Channel 00/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-213:9625:10059 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-213:9629:10063 [4] NCCL INFO Channel 00/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-213:9632:10065 [7] NCCL INFO Channel 00/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-213:9630:10060 [5] NCCL INFO Channel 01/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-165-213:9626:10061 [1] NCCL INFO Channel 01/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-213:9627:10064 [2] NCCL INFO Channel 00/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-213:9628:10066 [3] NCCL INFO Channel 01/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-213:9629:10063 [4] NCCL INFO Channel 01/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-213:9625:10059 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-213:9632:10065 [7] NCCL INFO Channel 01/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-166-244:897243:897678 [3] NCCL INFO Channel 23/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-213:9630:10060 [5] NCCL INFO Channel 02/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-165-213:9626:10061 [1] NCCL INFO Channel 02/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-213:9627:10064 [2] NCCL INFO Channel 01/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-213:9628:10066 [3] NCCL INFO Channel 02/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-213:9629:10063 [4] NCCL INFO Channel 02/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-213:9625:10059 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-213:9632:10065 [7] NCCL INFO Channel 02/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-213:9630:10060 [5] NCCL INFO Channel 03/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-165-213:9626:10061 [1] NCCL INFO Channel 03/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-213:9627:10064 [2] NCCL INFO Channel 02/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-213:9628:10066 [3] NCCL INFO Channel 03/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-213:9625:10059 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-213:9629:10063 [4] NCCL INFO Channel 03/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-213:9632:10065 [7] NCCL INFO Channel 03/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-213:9630:10060 [5] NCCL INFO Channel 04/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-165-213:9626:10061 [1] NCCL INFO Channel 04/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-213:9627:10064 [2] NCCL INFO Channel 03/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-213:9628:10066 [3] NCCL INFO Channel 04/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-213:9625:10059 [0] NCCL INFO Channel 04/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-213:9629:10063 [4] NCCL INFO Channel 04/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-213:9630:10060 [5] NCCL INFO Channel 05/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-165-213:9632:10065 [7] NCCL INFO Channel 04/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-213:9626:10061 [1] NCCL INFO Channel 05/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-213:9627:10064 [2] NCCL INFO Channel 04/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-213:9630:10060 [5] NCCL INFO Channel 06/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-165-213:9627:10064 [2] NCCL INFO Channel 05/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-213:9626:10061 [1] NCCL INFO Channel 06/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-213:9631:10062 [6] NCCL INFO Channel 00/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-213:9630:10060 [5] NCCL INFO Channel 07/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-165-213:9627:10064 [2] NCCL INFO Channel 06/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-213:9626:10061 [1] NCCL INFO Channel 07/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-213:9631:10062 [6] NCCL INFO Channel 01/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-213:9630:10060 [5] NCCL INFO Channel 08/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-165-213:9627:10064 [2] NCCL INFO Channel 07/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-213:9626:10061 [1] NCCL INFO Channel 08/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-213:9631:10062 [6] NCCL INFO Channel 02/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-213:9630:10060 [5] NCCL INFO Channel 09/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-165-213:9627:10064 [2] NCCL INFO Channel 08/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-213:9626:10061 [1] NCCL INFO Channel 09/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-213:9631:10062 [6] NCCL INFO Channel 03/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-213:9630:10060 [5] NCCL INFO Channel 10/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-165-213:9628:10066 [3] NCCL INFO Channel 05/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-213:9625:10059 [0] NCCL INFO Channel 05/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-213:9629:10063 [4] NCCL INFO Channel 05/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-213:9627:10064 [2] NCCL INFO Channel 09/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-213:9632:10065 [7] NCCL INFO Channel 05/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-213:9626:10061 [1] NCCL INFO Channel 10/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-213:9631:10062 [6] NCCL INFO Channel 04/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-213:9630:10060 [5] NCCL INFO Channel 11/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-165-213:9629:10063 [4] NCCL INFO Channel 06/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-213:9627:10064 [2] NCCL INFO Channel 10/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-213:9629:10063 [4] NCCL INFO Channel 07/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-213:9632:10065 [7] NCCL INFO Channel 06/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-213:9625:10059 [0] NCCL INFO Channel 06/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-213:9628:10066 [3] NCCL INFO Channel 06/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-213:9629:10063 [4] NCCL INFO Channel 08/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-213:9632:10065 [7] NCCL INFO Channel 07/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-213:9625:10059 [0] NCCL INFO Channel 07/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-213:9629:10063 [4] NCCL INFO Channel 09/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-213:9631:10062 [6] NCCL INFO Channel 05/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-213:9626:10061 [1] NCCL INFO Channel 11/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-213:9630:10060 [5] NCCL INFO Channel 12/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-165-213:9627:10064 [2] NCCL INFO Channel 11/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-213:9631:10062 [6] NCCL INFO Channel 06/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-213:9632:10065 [7] NCCL INFO Channel 08/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-213:9628:10066 [3] NCCL INFO Channel 07/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-213:9625:10059 [0] NCCL INFO Channel 08/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-213:9631:10062 [6] NCCL INFO Channel 07/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-213:9632:10065 [7] NCCL INFO Channel 09/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-213:9628:10066 [3] NCCL INFO Channel 08/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-213:9625:10059 [0] NCCL INFO Channel 09/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-213:9631:10062 [6] NCCL INFO Channel 08/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-213:9632:10065 [7] NCCL INFO Channel 10/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-213:9628:10066 [3] NCCL INFO Channel 09/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-213:9630:10060 [5] NCCL INFO Channel 13/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-165-213:9625:10059 [0] NCCL INFO Channel 10/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-213:9631:10062 [6] NCCL INFO Channel 09/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-213:9632:10065 [7] NCCL INFO Channel 11/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-213:9628:10066 [3] NCCL INFO Channel 10/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-213:9630:10060 [5] NCCL INFO Channel 14/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-165-213:9625:10059 [0] NCCL INFO Channel 11/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-213:9631:10062 [6] NCCL INFO Channel 10/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-213:9632:10065 [7] NCCL INFO Channel 12/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-213:9630:10060 [5] NCCL INFO Channel 15/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-165-213:9628:10066 [3] NCCL INFO Channel 11/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-213:9625:10059 [0] NCCL INFO Channel 12/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-213:9631:10062 [6] NCCL INFO Channel 11/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-213:9632:10065 [7] NCCL INFO Channel 13/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-213:9630:10060 [5] NCCL INFO Channel 16/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-165-213:9628:10066 [3] NCCL INFO Channel 12/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-213:9625:10059 [0] NCCL INFO Channel 13/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-213:9631:10062 [6] NCCL INFO Channel 12/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-213:9632:10065 [7] NCCL INFO Channel 14/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-213:9630:10060 [5] NCCL INFO Channel 17/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-165-213:9625:10059 [0] NCCL INFO Channel 14/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-213:9628:10066 [3] NCCL INFO Channel 13/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-213:9632:10065 [7] NCCL INFO Channel 15/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-213:9631:10062 [6] NCCL INFO Channel 13/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-213:9627:10064 [2] NCCL INFO Channel 12/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-213:9630:10060 [5] NCCL INFO Channel 18/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-165-213:9625:10059 [0] NCCL INFO Channel 15/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-213:9628:10066 [3] NCCL INFO Channel 14/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-213:9629:10063 [4] NCCL INFO Channel 10/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-213:9632:10065 [7] NCCL INFO Channel 16/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-213:9626:10061 [1] NCCL INFO Channel 12/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-213:9631:10062 [6] NCCL INFO Channel 14/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-213:9630:10060 [5] NCCL INFO Channel 19/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-165-213:9627:10064 [2] NCCL INFO Channel 13/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-213:9625:10059 [0] NCCL INFO Channel 16/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-213:9628:10066 [3] NCCL INFO Channel 15/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-213:9629:10063 [4] NCCL INFO Channel 11/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-213:9632:10065 [7] NCCL INFO Channel 17/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-213:9626:10061 [1] NCCL INFO Channel 13/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-213:9631:10062 [6] NCCL INFO Channel 15/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-213:9630:10060 [5] NCCL INFO Channel 20/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-165-213:9627:10064 [2] NCCL INFO Channel 14/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-213:9625:10059 [0] NCCL INFO Channel 17/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-213:9628:10066 [3] NCCL INFO Channel 16/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-213:9629:10063 [4] NCCL INFO Channel 12/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-213:9632:10065 [7] NCCL INFO Channel 18/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-213:9626:10061 [1] NCCL INFO Channel 14/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-213:9631:10062 [6] NCCL INFO Channel 16/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-213:9630:10060 [5] NCCL INFO Channel 21/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-165-213:9627:10064 [2] NCCL INFO Channel 15/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-213:9625:10059 [0] NCCL INFO Channel 18/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-213:9628:10066 [3] NCCL INFO Channel 17/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-213:9629:10063 [4] NCCL INFO Channel 13/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-213:9632:10065 [7] NCCL INFO Channel 19/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-213:9626:10061 [1] NCCL INFO Channel 15/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-213:9631:10062 [6] NCCL INFO Channel 17/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-213:9630:10060 [5] NCCL INFO Channel 22/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-165-213:9627:10064 [2] NCCL INFO Channel 16/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-213:9625:10059 [0] NCCL INFO Channel 19/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-213:9628:10066 [3] NCCL INFO Channel 18/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-213:9629:10063 [4] NCCL INFO Channel 14/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-213:9632:10065 [7] NCCL INFO Channel 20/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-213:9626:10061 [1] NCCL INFO Channel 16/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-213:9631:10062 [6] NCCL INFO Channel 18/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-213:9630:10060 [5] NCCL INFO Channel 23/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-165-213:9627:10064 [2] NCCL INFO Channel 17/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-213:9625:10059 [0] NCCL INFO Channel 20/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-213:9628:10066 [3] NCCL INFO Channel 19/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-213:9629:10063 [4] NCCL INFO Channel 15/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-213:9632:10065 [7] NCCL INFO Channel 21/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-213:9626:10061 [1] NCCL INFO Channel 17/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-213:9631:10062 [6] NCCL INFO Channel 19/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-213:9627:10064 [2] NCCL INFO Channel 18/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-213:9625:10059 [0] NCCL INFO Channel 21/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-213:9628:10066 [3] NCCL INFO Channel 20/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-213:9629:10063 [4] NCCL INFO Channel 16/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-213:9632:10065 [7] NCCL INFO Channel 22/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-213:9626:10061 [1] NCCL INFO Channel 18/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-213:9631:10062 [6] NCCL INFO Channel 20/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-213:9627:10064 [2] NCCL INFO Channel 19/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-213:9625:10059 [0] NCCL INFO Channel 22/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-213:9628:10066 [3] NCCL INFO Channel 21/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-213:9629:10063 [4] NCCL INFO Channel 17/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-213:9632:10065 [7] NCCL INFO Channel 23/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-165-213:9626:10061 [1] NCCL INFO Channel 19/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-213:9631:10062 [6] NCCL INFO Channel 21/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-213:9627:10064 [2] NCCL INFO Channel 20/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-213:9625:10059 [0] NCCL INFO Channel 23/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-213:9628:10066 [3] NCCL INFO Channel 22/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-213:9629:10063 [4] NCCL INFO Channel 18/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-213:9626:10061 [1] NCCL INFO Channel 20/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-213:9631:10062 [6] NCCL INFO Channel 22/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-213:9627:10064 [2] NCCL INFO Channel 21/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-213:9628:10066 [3] NCCL INFO Channel 23/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-165-213:9629:10063 [4] NCCL INFO Channel 19/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-213:9626:10061 [1] NCCL INFO Channel 21/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-213:9631:10062 [6] NCCL INFO Channel 23/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-165-213:9627:10064 [2] NCCL INFO Channel 22/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-213:9626:10061 [1] NCCL INFO Channel 22/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-213:9629:10063 [4] NCCL INFO Channel 20/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-213:9627:10064 [2] NCCL INFO Channel 23/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-213:9626:10061 [1] NCCL INFO Channel 23/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-165-213:9629:10063 [4] NCCL INFO Channel 21/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-213:9629:10063 [4] NCCL INFO Channel 22/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-213:9629:10063 [4] NCCL INFO Channel 23/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-166-214:624949:625381 [6] NCCL INFO Connected all rings
+ip-26-0-166-214:624950:625376 [7] NCCL INFO Connected all rings
+ip-26-0-166-214:624950:625376 [7] NCCL INFO Channel 00/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-166-214:624950:625376 [7] NCCL INFO Channel 01/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-166-214:624950:625376 [7] NCCL INFO Channel 02/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-166-214:624947:625378 [4] NCCL INFO Connected all rings
+ip-26-0-166-36:104431:104862 [2] NCCL INFO Connected all rings
+ip-26-0-166-214:624950:625376 [7] NCCL INFO Channel 03/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-166-214:624946:625377 [3] NCCL INFO Connected all rings
+ip-26-0-166-214:624950:625376 [7] NCCL INFO Channel 04/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-166-36:104429:104861 [0] NCCL INFO Connected all rings
+ip-26-0-166-36:104433:104864 [4] NCCL INFO Connected all rings
+ip-26-0-166-214:624950:625376 [7] NCCL INFO Channel 05/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-166-214:624948:625382 [5] NCCL INFO Connected all rings
+ip-26-0-166-36:104436:104865 [7] NCCL INFO Connected all rings
+ip-26-0-166-36:104436:104865 [7] NCCL INFO Channel 00/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-166-214:624950:625376 [7] NCCL INFO Channel 06/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-166-36:104436:104865 [7] NCCL INFO Channel 01/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-166-214:624950:625376 [7] NCCL INFO Channel 07/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-166-36:104436:104865 [7] NCCL INFO Channel 02/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-166-214:624950:625376 [7] NCCL INFO Channel 08/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-166-36:104436:104865 [7] NCCL INFO Channel 03/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-166-36:104432:104866 [3] NCCL INFO Connected all rings
+ip-26-0-166-125:762820:763259 [5] NCCL INFO Connected all rings
+ip-26-0-166-214:624950:625376 [7] NCCL INFO Channel 09/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-166-36:104436:104865 [7] NCCL INFO Channel 04/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-166-214:624950:625376 [7] NCCL INFO Channel 10/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-166-214:624950:625376 [7] NCCL INFO Channel 11/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-166-36:104436:104865 [7] NCCL INFO Channel 05/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-166-36:104434:104867 [5] NCCL INFO Connected all rings
+ip-26-0-166-36:104430:104868 [1] NCCL INFO Connected all rings
+ip-26-0-166-125:762822:763260 [7] NCCL INFO Connected all rings
+ip-26-0-166-125:762822:763260 [7] NCCL INFO Channel 00/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-166-214:624950:625376 [7] NCCL INFO Channel 12/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-166-36:104436:104865 [7] NCCL INFO Channel 06/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-166-36:104436:104865 [7] NCCL INFO Channel 07/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-166-36:104436:104865 [7] NCCL INFO Channel 08/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-166-36:104436:104865 [7] NCCL INFO Channel 09/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-166-125:762815:763258 [0] NCCL INFO Connected all rings
+ip-26-0-166-36:104436:104865 [7] NCCL INFO Channel 10/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-166-214:624950:625376 [7] NCCL INFO Channel 13/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-162-46:702886:703315 [6] NCCL INFO Connected all rings
+ip-26-0-166-36:104436:104865 [7] NCCL INFO Channel 11/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-166-214:624945:625380 [2] NCCL INFO Connected all rings
+ip-26-0-166-36:104435:104863 [6] NCCL INFO Connected all rings
+ip-26-0-166-36:104436:104865 [7] NCCL INFO Channel 12/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-166-125:762822:763260 [7] NCCL INFO Channel 01/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-166-36:104436:104865 [7] NCCL INFO Channel 13/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-166-36:104436:104865 [7] NCCL INFO Channel 14/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-166-36:104436:104865 [7] NCCL INFO Channel 15/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-166-125:762819:763261 [4] NCCL INFO Connected all rings
+ip-26-0-166-36:104436:104865 [7] NCCL INFO Channel 16/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-166-36:104436:104865 [7] NCCL INFO Channel 17/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-166-125:762822:763260 [7] NCCL INFO Channel 02/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-166-214:624950:625376 [7] NCCL INFO Channel 14/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-166-36:104436:104865 [7] NCCL INFO Channel 18/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-166-36:104436:104865 [7] NCCL INFO Channel 19/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-166-36:104436:104865 [7] NCCL INFO Channel 20/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-166-214:624944:625379 [1] NCCL INFO Connected all rings
+ip-26-0-166-36:104436:104865 [7] NCCL INFO Channel 21/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-166-36:104436:104865 [7] NCCL INFO Channel 22/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-166-36:104436:104865 [7] NCCL INFO Channel 23/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-167-9:1039542:1039976 [5] NCCL INFO Connected all rings
+ip-26-0-166-214:624950:625376 [7] NCCL INFO Channel 15/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-166-214:624950:625376 [7] NCCL INFO Channel 16/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-166-125:762822:763260 [7] NCCL INFO Channel 03/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-166-214:624950:625376 [7] NCCL INFO Channel 17/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-166-214:624950:625376 [7] NCCL INFO Channel 18/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-167-9:1039541:1039971 [4] NCCL INFO Connected all rings
+ip-26-0-166-36:104431:104862 [2] NCCL INFO Channel 00/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-36:104432:104866 [3] NCCL INFO Channel 00/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-125:762816:763262 [1] NCCL INFO Connected all rings
+ip-26-0-166-36:104434:104867 [5] NCCL INFO Channel 00/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-36:104431:104862 [2] NCCL INFO Channel 01/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-36:104432:104866 [3] NCCL INFO Channel 01/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-214:624949:625381 [6] NCCL INFO Channel 00/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-166-36:104433:104864 [4] NCCL INFO Channel 00/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-166-36:104434:104867 [5] NCCL INFO Channel 01/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-36:104431:104862 [2] NCCL INFO Channel 02/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-36:104432:104866 [3] NCCL INFO Channel 02/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-36:104433:104864 [4] NCCL INFO Channel 01/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-166-36:104434:104867 [5] NCCL INFO Channel 02/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-36:104431:104862 [2] NCCL INFO Channel 03/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-214:624950:625376 [7] NCCL INFO Channel 19/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-166-36:104432:104866 [3] NCCL INFO Channel 03/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-36:104433:104864 [4] NCCL INFO Channel 02/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-166-36:104434:104867 [5] NCCL INFO Channel 03/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-36:104431:104862 [2] NCCL INFO Channel 04/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-36:104432:104866 [3] NCCL INFO Channel 04/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-36:104433:104864 [4] NCCL INFO Channel 03/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-166-36:104434:104867 [5] NCCL INFO Channel 04/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-214:624949:625381 [6] NCCL INFO Channel 01/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-166-36:104431:104862 [2] NCCL INFO Channel 05/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-36:104432:104866 [3] NCCL INFO Channel 05/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-36:104433:104864 [4] NCCL INFO Channel 04/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-162-46:702880:703314 [0] NCCL INFO Connected all rings
+ip-26-0-166-36:104434:104867 [5] NCCL INFO Channel 05/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-36:104431:104862 [2] NCCL INFO Channel 06/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-125:762817:763265 [2] NCCL INFO Connected all rings
+ip-26-0-166-36:104432:104866 [3] NCCL INFO Channel 06/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-36:104433:104864 [4] NCCL INFO Channel 05/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-166-214:624950:625376 [7] NCCL INFO Channel 20/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-166-36:104434:104867 [5] NCCL INFO Channel 06/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-36:104431:104862 [2] NCCL INFO Channel 07/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-125:762822:763260 [7] NCCL INFO Channel 04/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-166-36:104432:104866 [3] NCCL INFO Channel 07/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-36:104433:104864 [4] NCCL INFO Channel 06/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-166-36:104434:104867 [5] NCCL INFO Channel 07/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-36:104430:104868 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-36:104431:104862 [2] NCCL INFO Channel 08/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-214:624949:625381 [6] NCCL INFO Channel 02/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-166-36:104433:104864 [4] NCCL INFO Channel 07/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-166-36:104432:104866 [3] NCCL INFO Channel 08/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-36:104434:104867 [5] NCCL INFO Channel 08/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-36:104430:104868 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-36:104431:104862 [2] NCCL INFO Channel 09/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-36:104433:104864 [4] NCCL INFO Channel 08/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-166-36:104434:104867 [5] NCCL INFO Channel 09/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-36:104432:104866 [3] NCCL INFO Channel 09/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-36:104430:104868 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-36:104431:104862 [2] NCCL INFO Channel 10/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-214:624950:625376 [7] NCCL INFO Channel 21/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-162-46:702884:703319 [4] NCCL INFO Connected all rings
+ip-26-0-166-214:624949:625381 [6] NCCL INFO Channel 03/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-166-36:104433:104864 [4] NCCL INFO Channel 09/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-166-36:104434:104867 [5] NCCL INFO Channel 10/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-36:104432:104866 [3] NCCL INFO Channel 10/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-36:104430:104868 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-36:104431:104862 [2] NCCL INFO Channel 11/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-36:104433:104864 [4] NCCL INFO Channel 10/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-166-125:762822:763260 [7] NCCL INFO Channel 05/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-166-36:104434:104867 [5] NCCL INFO Channel 11/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-36:104430:104868 [1] NCCL INFO Channel 04/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-36:104432:104866 [3] NCCL INFO Channel 11/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-36:104431:104862 [2] NCCL INFO Channel 12/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-214:624950:625376 [7] NCCL INFO Channel 22/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-166-36:104433:104864 [4] NCCL INFO Channel 11/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-166-36:104434:104867 [5] NCCL INFO Channel 12/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-36:104430:104868 [1] NCCL INFO Channel 05/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-36:104432:104866 [3] NCCL INFO Channel 12/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-36:104431:104862 [2] NCCL INFO Channel 13/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-167-9:1039537:1039969 [0] NCCL INFO Connected all rings
+ip-26-0-166-36:104433:104864 [4] NCCL INFO Channel 12/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-166-36:104435:104863 [6] NCCL INFO Channel 00/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-166-214:624949:625381 [6] NCCL INFO Channel 04/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-166-36:104434:104867 [5] NCCL INFO Channel 13/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-36:104430:104868 [1] NCCL INFO Channel 06/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-36:104431:104862 [2] NCCL INFO Channel 14/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-36:104432:104866 [3] NCCL INFO Channel 13/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-36:104433:104864 [4] NCCL INFO Channel 13/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-166-36:104435:104863 [6] NCCL INFO Channel 01/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-166-36:104434:104867 [5] NCCL INFO Channel 14/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-36:104430:104868 [1] NCCL INFO Channel 07/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-36:104431:104862 [2] NCCL INFO Channel 15/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-36:104432:104866 [3] NCCL INFO Channel 14/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-36:104433:104864 [4] NCCL INFO Channel 14/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-166-214:624950:625376 [7] NCCL INFO Channel 23/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-166-36:104434:104867 [5] NCCL INFO Channel 15/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-36:104432:104866 [3] NCCL INFO Channel 15/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-36:104435:104863 [6] NCCL INFO Channel 02/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-166-36:104430:104868 [1] NCCL INFO Channel 08/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-36:104431:104862 [2] NCCL INFO Channel 16/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-125:762818:763263 [3] NCCL INFO Connected all rings
+ip-26-0-166-125:762822:763260 [7] NCCL INFO Channel 06/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-166-214:624943:625375 [0] NCCL INFO Connected all rings
+ip-26-0-166-214:624949:625381 [6] NCCL INFO Channel 05/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-166-36:104432:104866 [3] NCCL INFO Channel 16/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-36:104434:104867 [5] NCCL INFO Channel 16/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-36:104435:104863 [6] NCCL INFO Channel 03/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-166-36:104431:104862 [2] NCCL INFO Channel 17/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-36:104430:104868 [1] NCCL INFO Channel 09/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-36:104433:104864 [4] NCCL INFO Channel 15/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-166-214:624947:625378 [4] NCCL INFO Channel 00/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-166-125:762821:763264 [6] NCCL INFO Connected all rings
+ip-26-0-166-214:624949:625381 [6] NCCL INFO Channel 06/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-166-36:104432:104866 [3] NCCL INFO Channel 17/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-36:104434:104867 [5] NCCL INFO Channel 17/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-36:104435:104863 [6] NCCL INFO Channel 04/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-166-36:104431:104862 [2] NCCL INFO Channel 18/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-125:762822:763260 [7] NCCL INFO Channel 07/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-166-214:624947:625378 [4] NCCL INFO Channel 01/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-166-214:624949:625381 [6] NCCL INFO Channel 07/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-166-36:104430:104868 [1] NCCL INFO Channel 10/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-36:104435:104863 [6] NCCL INFO Channel 05/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-166-36:104431:104862 [2] NCCL INFO Channel 19/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-36:104433:104864 [4] NCCL INFO Channel 16/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-166-36:104430:104868 [1] NCCL INFO Channel 11/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762822:763260 [7] NCCL INFO Channel 08/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-166-214:624947:625378 [4] NCCL INFO Channel 02/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-166-214:624949:625381 [6] NCCL INFO Channel 08/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-166-36:104435:104863 [6] NCCL INFO Channel 06/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-166-36:104431:104862 [2] NCCL INFO Channel 20/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-36:104433:104864 [4] NCCL INFO Channel 17/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-166-36:104430:104868 [1] NCCL INFO Channel 12/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-36:104435:104863 [6] NCCL INFO Channel 07/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-166-36:104431:104862 [2] NCCL INFO Channel 21/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-214:624947:625378 [4] NCCL INFO Channel 03/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-166-214:624949:625381 [6] NCCL INFO Channel 09/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-166-125:762822:763260 [7] NCCL INFO Channel 09/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-166-36:104430:104868 [1] NCCL INFO Channel 13/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-36:104433:104864 [4] NCCL INFO Channel 18/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-166-125:762822:763260 [7] NCCL INFO Channel 10/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-166-214:624948:625382 [5] NCCL INFO Channel 00/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-214:624947:625378 [4] NCCL INFO Channel 04/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-166-214:624949:625381 [6] NCCL INFO Channel 10/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-166-125:762822:763260 [7] NCCL INFO Channel 11/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-166-15:127290:127724 [0] NCCL INFO Connected all rings
+ip-26-0-166-36:104433:104864 [4] NCCL INFO Channel 19/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-166-36:104434:104867 [5] NCCL INFO Channel 18/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-214:624948:625382 [5] NCCL INFO Channel 01/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-125:762822:763260 [7] NCCL INFO Channel 12/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-166-36:104432:104866 [3] NCCL INFO Channel 18/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-36:104431:104862 [2] NCCL INFO Channel 22/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-214:624947:625378 [4] NCCL INFO Channel 05/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-166-36:104433:104864 [4] NCCL INFO Channel 20/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-166-36:104434:104867 [5] NCCL INFO Channel 19/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-214:624949:625381 [6] NCCL INFO Channel 11/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-166-125:762822:763260 [7] NCCL INFO Channel 13/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-166-36:104432:104866 [3] NCCL INFO Channel 19/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-36:104431:104862 [2] NCCL INFO Channel 23/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-36:104433:104864 [4] NCCL INFO Channel 21/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-166-36:104434:104867 [5] NCCL INFO Channel 20/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-125:762822:763260 [7] NCCL INFO Channel 14/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-166-36:104432:104866 [3] NCCL INFO Channel 20/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-36:104433:104864 [4] NCCL INFO Channel 22/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-166-36:104434:104867 [5] NCCL INFO Channel 21/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-214:624948:625382 [5] NCCL INFO Channel 02/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-36:104435:104863 [6] NCCL INFO Channel 08/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-166-36:104432:104866 [3] NCCL INFO Channel 21/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-125:762822:763260 [7] NCCL INFO Channel 15/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-166-36:104434:104867 [5] NCCL INFO Channel 22/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-36:104433:104864 [4] NCCL INFO Channel 23/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-166-36:104435:104863 [6] NCCL INFO Channel 09/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-166-125:762822:763260 [7] NCCL INFO Channel 16/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-166-214:624947:625378 [4] NCCL INFO Channel 06/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-166-36:104432:104866 [3] NCCL INFO Channel 22/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-36:104434:104867 [5] NCCL INFO Channel 23/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-214:624946:625377 [3] NCCL INFO Channel 00/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-36:104430:104868 [1] NCCL INFO Channel 14/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624947:625378 [4] NCCL INFO Channel 07/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-166-36:104435:104863 [6] NCCL INFO Channel 10/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-166-36:104432:104866 [3] NCCL INFO Channel 23/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-214:624946:625377 [3] NCCL INFO Channel 01/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-36:104430:104868 [1] NCCL INFO Channel 15/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762822:763260 [7] NCCL INFO Channel 17/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-166-36:104435:104863 [6] NCCL INFO Channel 11/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-166-214:624947:625378 [4] NCCL INFO Channel 08/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-166-36:104430:104868 [1] NCCL INFO Channel 16/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624946:625377 [3] NCCL INFO Channel 02/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-36:104435:104863 [6] NCCL INFO Channel 12/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-166-214:624947:625378 [4] NCCL INFO Channel 09/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-166-125:762822:763260 [7] NCCL INFO Channel 18/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-166-214:624946:625377 [3] NCCL INFO Channel 03/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-214:624947:625378 [4] NCCL INFO Channel 10/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-166-214:624946:625377 [3] NCCL INFO Channel 04/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-125:762822:763260 [7] NCCL INFO Channel 19/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-166-214:624947:625378 [4] NCCL INFO Channel 11/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-166-214:624949:625381 [6] NCCL INFO Channel 12/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-166-214:624948:625382 [5] NCCL INFO Channel 03/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-214:624946:625377 [3] NCCL INFO Channel 05/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-214:624947:625378 [4] NCCL INFO Channel 12/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-166-125:762822:763260 [7] NCCL INFO Channel 20/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-166-36:104430:104868 [1] NCCL INFO Channel 17/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624949:625381 [6] NCCL INFO Channel 13/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-166-214:624948:625382 [5] NCCL INFO Channel 04/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-214:624946:625377 [3] NCCL INFO Channel 06/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-214:624947:625378 [4] NCCL INFO Channel 13/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-166-214:624944:625379 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-167-9:1039544:1039972 [7] NCCL INFO Connected all rings
+ip-26-0-166-214:624949:625381 [6] NCCL INFO Channel 14/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-167-9:1039544:1039972 [7] NCCL INFO Channel 00/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-166-214:624948:625382 [5] NCCL INFO Channel 05/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-214:624946:625377 [3] NCCL INFO Channel 07/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-125:762822:763260 [7] NCCL INFO Channel 21/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-166-214:624947:625378 [4] NCCL INFO Channel 14/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-166-125:762822:763260 [7] NCCL INFO Channel 22/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-166-36:104430:104868 [1] NCCL INFO Channel 18/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624944:625379 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624949:625381 [6] NCCL INFO Channel 15/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-166-214:624948:625382 [5] NCCL INFO Channel 06/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-214:624946:625377 [3] NCCL INFO Channel 08/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-125:762822:763260 [7] NCCL INFO Channel 23/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-166-214:624947:625378 [4] NCCL INFO Channel 15/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-166-36:104435:104863 [6] NCCL INFO Channel 13/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-166-214:624944:625379 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624949:625381 [6] NCCL INFO Channel 16/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-166-214:624948:625382 [5] NCCL INFO Channel 07/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-214:624946:625377 [3] NCCL INFO Channel 09/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-214:624947:625378 [4] NCCL INFO Channel 16/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-167-9:1039543:1039970 [6] NCCL INFO Connected all rings
+ip-26-0-166-214:624944:625379 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624949:625381 [6] NCCL INFO Channel 17/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-166-214:624948:625382 [5] NCCL INFO Channel 08/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-214:624945:625380 [2] NCCL INFO Channel 00/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-214:624946:625377 [3] NCCL INFO Channel 10/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-214:624947:625378 [4] NCCL INFO Channel 17/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-166-36:104430:104868 [1] NCCL INFO Channel 19/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624944:625379 [1] NCCL INFO Channel 04/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624949:625381 [6] NCCL INFO Channel 18/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-166-214:624948:625382 [5] NCCL INFO Channel 09/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-214:624945:625380 [2] NCCL INFO Channel 01/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-214:624946:625377 [3] NCCL INFO Channel 11/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-214:624947:625378 [4] NCCL INFO Channel 18/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-166-125:762820:763259 [5] NCCL INFO Channel 00/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-125:762816:763262 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624944:625379 [1] NCCL INFO Channel 05/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624949:625381 [6] NCCL INFO Channel 19/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-166-214:624948:625382 [5] NCCL INFO Channel 10/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-214:624945:625380 [2] NCCL INFO Channel 02/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-125:762820:763259 [5] NCCL INFO Channel 01/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-214:624946:625377 [3] NCCL INFO Channel 12/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-214:624947:625378 [4] NCCL INFO Channel 19/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-166-125:762816:763262 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624944:625379 [1] NCCL INFO Channel 06/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762820:763259 [5] NCCL INFO Channel 02/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-214:624949:625381 [6] NCCL INFO Channel 20/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-166-214:624948:625382 [5] NCCL INFO Channel 11/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-214:624945:625380 [2] NCCL INFO Channel 03/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-214:624947:625378 [4] NCCL INFO Channel 20/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-166-214:624946:625377 [3] NCCL INFO Channel 13/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-125:762816:763262 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762820:763259 [5] NCCL INFO Channel 03/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-214:624944:625379 [1] NCCL INFO Channel 07/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624949:625381 [6] NCCL INFO Channel 21/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-166-214:624948:625382 [5] NCCL INFO Channel 12/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-214:624945:625380 [2] NCCL INFO Channel 04/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-214:624947:625378 [4] NCCL INFO Channel 21/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-166-214:624946:625377 [3] NCCL INFO Channel 14/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-36:104430:104868 [1] NCCL INFO Channel 20/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-167-9:1039538:1039973 [1] NCCL INFO Connected all rings
+ip-26-0-166-36:104435:104863 [6] NCCL INFO Channel 14/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-166-125:762816:763262 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762820:763259 [5] NCCL INFO Channel 04/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-125:762821:763264 [6] NCCL INFO Channel 00/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-166-214:624944:625379 [1] NCCL INFO Channel 08/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624948:625382 [5] NCCL INFO Channel 13/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-214:624949:625381 [6] NCCL INFO Channel 22/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-166-214:624945:625380 [2] NCCL INFO Channel 05/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-214:624947:625378 [4] NCCL INFO Channel 22/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-166-125:762816:763262 [1] NCCL INFO Channel 04/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624946:625377 [3] NCCL INFO Channel 15/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-125:762820:763259 [5] NCCL INFO Channel 05/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-125:762821:763264 [6] NCCL INFO Channel 01/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-166-125:762816:763262 [1] NCCL INFO Channel 05/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762820:763259 [5] NCCL INFO Channel 06/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-214:624944:625379 [1] NCCL INFO Channel 09/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624948:625382 [5] NCCL INFO Channel 14/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-214:624945:625380 [2] NCCL INFO Channel 06/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-214:624949:625381 [6] NCCL INFO Channel 23/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-166-214:624947:625378 [4] NCCL INFO Channel 23/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-162-46:702885:703316 [5] NCCL INFO Connected all rings
+ip-26-0-166-214:624946:625377 [3] NCCL INFO Channel 16/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-125:762821:763264 [6] NCCL INFO Channel 02/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-166-125:762816:763262 [1] NCCL INFO Channel 06/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762820:763259 [5] NCCL INFO Channel 07/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-214:624944:625379 [1] NCCL INFO Channel 10/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624948:625382 [5] NCCL INFO Channel 15/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-125:762821:763264 [6] NCCL INFO Channel 03/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-166-214:624945:625380 [2] NCCL INFO Channel 07/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-214:624946:625377 [3] NCCL INFO Channel 17/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-36:104430:104868 [1] NCCL INFO Channel 21/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762816:763262 [1] NCCL INFO Channel 07/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762820:763259 [5] NCCL INFO Channel 08/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-125:762821:763264 [6] NCCL INFO Channel 04/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-166-125:762819:763261 [4] NCCL INFO Channel 00/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-167-9:1039544:1039972 [7] NCCL INFO Channel 01/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-166-214:624948:625382 [5] NCCL INFO Channel 16/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-36:104435:104863 [6] NCCL INFO Channel 15/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-166-125:762816:763262 [1] NCCL INFO Channel 08/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624948:625382 [5] NCCL INFO Channel 17/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-125:762820:763259 [5] NCCL INFO Channel 09/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-214:624944:625379 [1] NCCL INFO Channel 11/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624945:625380 [2] NCCL INFO Channel 08/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-125:762821:763264 [6] NCCL INFO Channel 05/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-166-125:762819:763261 [4] NCCL INFO Channel 01/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-166-214:624946:625377 [3] NCCL INFO Channel 18/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-214:624948:625382 [5] NCCL INFO Channel 18/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-214:624945:625380 [2] NCCL INFO Channel 09/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-214:624944:625379 [1] NCCL INFO Channel 12/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762816:763262 [1] NCCL INFO Channel 09/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762820:763259 [5] NCCL INFO Channel 10/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-125:762818:763263 [3] NCCL INFO Channel 00/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-125:762821:763264 [6] NCCL INFO Channel 06/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-166-125:762819:763261 [4] NCCL INFO Channel 02/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-166-214:624946:625377 [3] NCCL INFO Channel 19/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-214:624948:625382 [5] NCCL INFO Channel 19/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-214:624945:625380 [2] NCCL INFO Channel 10/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-214:624944:625379 [1] NCCL INFO Channel 13/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762816:763262 [1] NCCL INFO Channel 10/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762820:763259 [5] NCCL INFO Channel 11/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-125:762818:763263 [3] NCCL INFO Channel 01/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-125:762821:763264 [6] NCCL INFO Channel 07/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-166-214:624946:625377 [3] NCCL INFO Channel 20/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-214:624948:625382 [5] NCCL INFO Channel 20/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-214:624945:625380 [2] NCCL INFO Channel 11/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-125:762819:763261 [4] NCCL INFO Channel 03/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-166-214:624944:625379 [1] NCCL INFO Channel 14/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-36:104430:104868 [1] NCCL INFO Channel 22/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624946:625377 [3] NCCL INFO Channel 21/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-214:624948:625382 [5] NCCL INFO Channel 21/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-214:624945:625380 [2] NCCL INFO Channel 12/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-214:624944:625379 [1] NCCL INFO Channel 15/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762816:763262 [1] NCCL INFO Channel 11/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762818:763263 [3] NCCL INFO Channel 02/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-125:762820:763259 [5] NCCL INFO Channel 12/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-125:762821:763264 [6] NCCL INFO Channel 08/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-166-125:762819:763261 [4] NCCL INFO Channel 04/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-166-214:624948:625382 [5] NCCL INFO Channel 22/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-214:624945:625380 [2] NCCL INFO Channel 13/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-214:624946:625377 [3] NCCL INFO Channel 22/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-36:104435:104863 [6] NCCL INFO Channel 16/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-166-214:624944:625379 [1] NCCL INFO Channel 16/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-167-9:1039540:1039974 [3] NCCL INFO Connected all rings
+ip-26-0-166-125:762816:763262 [1] NCCL INFO Channel 12/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762818:763263 [3] NCCL INFO Channel 03/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-125:762820:763259 [5] NCCL INFO Channel 13/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-125:762819:763261 [4] NCCL INFO Channel 05/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-166-214:624948:625382 [5] NCCL INFO Channel 23/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-214:624945:625380 [2] NCCL INFO Channel 14/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-214:624946:625377 [3] NCCL INFO Channel 23/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-214:624944:625379 [1] NCCL INFO Channel 17/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-167-9:1039539:1039975 [2] NCCL INFO Connected all rings
+ip-26-0-166-125:762816:763262 [1] NCCL INFO Channel 13/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762818:763263 [3] NCCL INFO Channel 04/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-167-9:1039544:1039972 [7] NCCL INFO Channel 02/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-166-125:762820:763259 [5] NCCL INFO Channel 14/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-125:762817:763265 [2] NCCL INFO Channel 00/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-125:762819:763261 [4] NCCL INFO Channel 06/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-166-125:762821:763264 [6] NCCL INFO Channel 09/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-166-125:762818:763263 [3] NCCL INFO Channel 05/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-125:762816:763262 [1] NCCL INFO Channel 14/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762820:763259 [5] NCCL INFO Channel 15/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-125:762817:763265 [2] NCCL INFO Channel 01/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-125:762819:763261 [4] NCCL INFO Channel 07/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-166-214:624944:625379 [1] NCCL INFO Channel 18/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-167-9:1039544:1039972 [7] NCCL INFO Channel 03/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-166-214:624945:625380 [2] NCCL INFO Channel 15/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-125:762821:763264 [6] NCCL INFO Channel 10/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-167-9:1039544:1039972 [7] NCCL INFO Channel 04/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-166-36:104430:104868 [1] NCCL INFO Channel 23/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-36:104435:104863 [6] NCCL INFO Channel 17/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-166-125:762819:763261 [4] NCCL INFO Channel 08/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-166-125:762821:763264 [6] NCCL INFO Channel 11/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-166-125:762818:763263 [3] NCCL INFO Channel 06/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-125:762816:763262 [1] NCCL INFO Channel 15/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762820:763259 [5] NCCL INFO Channel 16/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-167-9:1039544:1039972 [7] NCCL INFO Channel 05/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-166-125:762819:763261 [4] NCCL INFO Channel 09/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-166-214:624944:625379 [1] NCCL INFO Channel 19/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762821:763264 [6] NCCL INFO Channel 12/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-166-125:762818:763263 [3] NCCL INFO Channel 07/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-125:762816:763262 [1] NCCL INFO Channel 16/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762820:763259 [5] NCCL INFO Channel 17/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-167-9:1039544:1039972 [7] NCCL INFO Channel 06/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-166-125:762817:763265 [2] NCCL INFO Channel 02/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-125:762821:763264 [6] NCCL INFO Channel 13/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-166-125:762819:763261 [4] NCCL INFO Channel 10/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-167-9:1039544:1039972 [7] NCCL INFO Channel 07/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-166-125:762816:763262 [1] NCCL INFO Channel 17/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762817:763265 [2] NCCL INFO Channel 03/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-125:762821:763264 [6] NCCL INFO Channel 14/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-166-214:624945:625380 [2] NCCL INFO Channel 16/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-125:762819:763261 [4] NCCL INFO Channel 11/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-166-125:762820:763259 [5] NCCL INFO Channel 18/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-125:762816:763262 [1] NCCL INFO Channel 18/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762817:763265 [2] NCCL INFO Channel 04/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-125:762821:763264 [6] NCCL INFO Channel 15/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-167-9:1039544:1039972 [7] NCCL INFO Channel 08/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-166-36:104435:104863 [6] NCCL INFO Channel 18/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-166-125:762819:763261 [4] NCCL INFO Channel 12/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-166-125:762818:763263 [3] NCCL INFO Channel 08/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-125:762820:763259 [5] NCCL INFO Channel 19/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-125:762816:763262 [1] NCCL INFO Channel 19/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762819:763261 [4] NCCL INFO Channel 13/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-167-9:1039544:1039972 [7] NCCL INFO Channel 09/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-166-125:762818:763263 [3] NCCL INFO Channel 09/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-125:762820:763259 [5] NCCL INFO Channel 20/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-125:762816:763262 [1] NCCL INFO Channel 20/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762819:763261 [4] NCCL INFO Channel 14/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-166-125:762818:763263 [3] NCCL INFO Channel 10/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-125:762820:763259 [5] NCCL INFO Channel 21/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-162-46:702887:703317 [7] NCCL INFO Connected all rings
+ip-26-0-162-46:702887:703317 [7] NCCL INFO Channel 00/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-166-125:762816:763262 [1] NCCL INFO Channel 21/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762819:763261 [4] NCCL INFO Channel 15/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-166-125:762818:763263 [3] NCCL INFO Channel 11/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-125:762820:763259 [5] NCCL INFO Channel 22/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-125:762816:763262 [1] NCCL INFO Channel 22/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762819:763261 [4] NCCL INFO Channel 16/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-166-125:762818:763263 [3] NCCL INFO Channel 12/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-125:762816:763262 [1] NCCL INFO Channel 23/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762820:763259 [5] NCCL INFO Channel 23/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-125:762819:763261 [4] NCCL INFO Channel 17/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-166-125:762818:763263 [3] NCCL INFO Channel 13/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-214:624944:625379 [1] NCCL INFO Channel 20/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762819:763261 [4] NCCL INFO Channel 18/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-166-125:762818:763263 [3] NCCL INFO Channel 14/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-125:762821:763264 [6] NCCL INFO Channel 16/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-167-9:1039544:1039972 [7] NCCL INFO Channel 10/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-167-9:1039544:1039972 [7] NCCL INFO Channel 11/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-167-9:1039544:1039972 [7] NCCL INFO Channel 12/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-167-9:1039544:1039972 [7] NCCL INFO Channel 13/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-167-9:1039544:1039972 [7] NCCL INFO Channel 14/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-167-9:1039544:1039972 [7] NCCL INFO Channel 15/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-162-46:702883:703318 [3] NCCL INFO Connected all rings
+ip-26-0-167-9:1039544:1039972 [7] NCCL INFO Channel 16/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-167-9:1039544:1039972 [7] NCCL INFO Channel 17/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-166-125:762818:763263 [3] NCCL INFO Channel 15/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-125:762817:763265 [2] NCCL INFO Channel 05/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-125:762821:763264 [6] NCCL INFO Channel 17/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-167-9:1039544:1039972 [7] NCCL INFO Channel 18/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-166-36:104435:104863 [6] NCCL INFO Channel 19/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-166-214:624945:625380 [2] NCCL INFO Channel 17/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-125:762819:763261 [4] NCCL INFO Channel 19/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-167-9:1039544:1039972 [7] NCCL INFO Channel 19/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-167-9:1039541:1039971 [4] NCCL INFO Channel 00/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-167-9:1039544:1039972 [7] NCCL INFO Channel 20/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-166-125:762818:763263 [3] NCCL INFO Channel 16/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-15:127295:127726 [5] NCCL INFO Connected all rings
+ip-26-0-166-125:762817:763265 [2] NCCL INFO Channel 06/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-125:762821:763264 [6] NCCL INFO Channel 18/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-167-9:1039541:1039971 [4] NCCL INFO Channel 01/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-162-46:702882:703321 [2] NCCL INFO Connected all rings
+ip-26-0-162-46:702887:703317 [7] NCCL INFO Channel 01/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-167-9:1039544:1039972 [7] NCCL INFO Channel 21/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-166-125:762819:763261 [4] NCCL INFO Channel 20/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-167-9:1039541:1039971 [4] NCCL INFO Channel 02/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-167-9:1039544:1039972 [7] NCCL INFO Channel 22/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-162-46:702881:703320 [1] NCCL INFO Connected all rings
+ip-26-0-167-9:1039541:1039971 [4] NCCL INFO Channel 03/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-167-9:1039544:1039972 [7] NCCL INFO Channel 23/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-167-9:1039541:1039971 [4] NCCL INFO Channel 04/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-166-125:762819:763261 [4] NCCL INFO Channel 21/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-162-46:702887:703317 [7] NCCL INFO Channel 02/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-166-214:624944:625379 [1] NCCL INFO Channel 21/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-167-9:1039541:1039971 [4] NCCL INFO Channel 05/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-166-214:624945:625380 [2] NCCL INFO Channel 18/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-125:762818:763263 [3] NCCL INFO Channel 17/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-162-46:702887:703317 [7] NCCL INFO Channel 03/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-167-9:1039541:1039971 [4] NCCL INFO Channel 06/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-167-9:1039540:1039974 [3] NCCL INFO Channel 00/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-167-9:1039542:1039976 [5] NCCL INFO Channel 00/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-162-46:702887:703317 [7] NCCL INFO Channel 04/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-167-9:1039541:1039971 [4] NCCL INFO Channel 07/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-167-9:1039540:1039974 [3] NCCL INFO Channel 01/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-167-9:1039542:1039976 [5] NCCL INFO Channel 01/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-167-9:1039538:1039973 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762819:763261 [4] NCCL INFO Channel 22/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-167-9:1039541:1039971 [4] NCCL INFO Channel 08/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-167-9:1039540:1039974 [3] NCCL INFO Channel 02/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-167-9:1039542:1039976 [5] NCCL INFO Channel 02/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-167-9:1039538:1039973 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-162-46:702887:703317 [7] NCCL INFO Channel 05/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-166-36:104435:104863 [6] NCCL INFO Channel 20/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-167-9:1039541:1039971 [4] NCCL INFO Channel 09/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-167-9:1039542:1039976 [5] NCCL INFO Channel 03/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-167-9:1039538:1039973 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-167-9:1039540:1039974 [3] NCCL INFO Channel 03/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-167-9:1039541:1039971 [4] NCCL INFO Channel 10/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-162-46:702887:703317 [7] NCCL INFO Channel 06/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-166-214:624944:625379 [1] NCCL INFO Channel 22/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-167-9:1039538:1039973 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-167-9:1039542:1039976 [5] NCCL INFO Channel 04/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-167-9:1039540:1039974 [3] NCCL INFO Channel 04/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-162-46:702887:703317 [7] NCCL INFO Channel 07/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-167-9:1039541:1039971 [4] NCCL INFO Channel 11/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-166-125:762821:763264 [6] NCCL INFO Channel 19/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-166-125:762818:763263 [3] NCCL INFO Channel 18/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-167-9:1039538:1039973 [1] NCCL INFO Channel 04/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-162-46:702887:703317 [7] NCCL INFO Channel 08/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-167-9:1039540:1039974 [3] NCCL INFO Channel 05/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-167-9:1039542:1039976 [5] NCCL INFO Channel 05/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-167-9:1039539:1039975 [2] NCCL INFO Channel 00/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-167-9:1039543:1039970 [6] NCCL INFO Channel 00/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-162-46:702887:703317 [7] NCCL INFO Channel 09/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-167-9:1039541:1039971 [4] NCCL INFO Channel 12/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-166-125:762817:763265 [2] NCCL INFO Channel 07/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-167-9:1039538:1039973 [1] NCCL INFO Channel 05/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-162-46:702887:703317 [7] NCCL INFO Channel 10/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-167-9:1039540:1039974 [3] NCCL INFO Channel 06/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-167-9:1039542:1039976 [5] NCCL INFO Channel 06/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-167-9:1039539:1039975 [2] NCCL INFO Channel 01/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-167-9:1039543:1039970 [6] NCCL INFO Channel 01/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-167-9:1039541:1039971 [4] NCCL INFO Channel 13/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-162-46:702887:703317 [7] NCCL INFO Channel 11/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-167-9:1039538:1039973 [1] NCCL INFO Channel 06/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-167-9:1039540:1039974 [3] NCCL INFO Channel 07/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-167-9:1039542:1039976 [5] NCCL INFO Channel 07/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-167-9:1039539:1039975 [2] NCCL INFO Channel 02/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-162-46:702887:703317 [7] NCCL INFO Channel 12/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-166-125:762821:763264 [6] NCCL INFO Channel 20/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-166-125:762818:763263 [3] NCCL INFO Channel 19/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-167-9:1039543:1039970 [6] NCCL INFO Channel 02/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-166-125:762817:763265 [2] NCCL INFO Channel 08/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-167-9:1039541:1039971 [4] NCCL INFO Channel 14/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-162-46:702887:703317 [7] NCCL INFO Channel 13/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-167-9:1039538:1039973 [1] NCCL INFO Channel 07/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-167-9:1039540:1039974 [3] NCCL INFO Channel 08/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-167-9:1039542:1039976 [5] NCCL INFO Channel 08/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-167-9:1039539:1039975 [2] NCCL INFO Channel 03/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-167-9:1039543:1039970 [6] NCCL INFO Channel 03/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-167-9:1039541:1039971 [4] NCCL INFO Channel 15/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-167-9:1039538:1039973 [1] NCCL INFO Channel 08/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-162-46:702887:703317 [7] NCCL INFO Channel 14/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-167-9:1039540:1039974 [3] NCCL INFO Channel 09/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-167-9:1039542:1039976 [5] NCCL INFO Channel 09/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-167-9:1039539:1039975 [2] NCCL INFO Channel 04/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-15:127297:127731 [7] NCCL INFO Connected all rings
+ip-26-0-166-15:127297:127731 [7] NCCL INFO Channel 00/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-167-9:1039543:1039970 [6] NCCL INFO Channel 04/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-167-9:1039541:1039971 [4] NCCL INFO Channel 16/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-167-9:1039538:1039973 [1] NCCL INFO Channel 09/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-167-9:1039540:1039974 [3] NCCL INFO Channel 10/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-167-9:1039542:1039976 [5] NCCL INFO Channel 10/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-162-46:702887:703317 [7] NCCL INFO Channel 15/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-166-125:762819:763261 [4] NCCL INFO Channel 23/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-166-125:762821:763264 [6] NCCL INFO Channel 21/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-167-9:1039539:1039975 [2] NCCL INFO Channel 05/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-125:762818:763263 [3] NCCL INFO Channel 20/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-125:762817:763265 [2] NCCL INFO Channel 09/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-15:127294:127725 [4] NCCL INFO Connected all rings
+ip-26-0-167-9:1039543:1039970 [6] NCCL INFO Channel 05/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-167-9:1039541:1039971 [4] NCCL INFO Channel 17/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-167-9:1039538:1039973 [1] NCCL INFO Channel 10/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-167-9:1039540:1039974 [3] NCCL INFO Channel 11/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-167-9:1039542:1039976 [5] NCCL INFO Channel 11/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-167-9:1039539:1039975 [2] NCCL INFO Channel 06/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-162-46:702887:703317 [7] NCCL INFO Channel 16/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-167-9:1039543:1039970 [6] NCCL INFO Channel 06/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-167-9:1039541:1039971 [4] NCCL INFO Channel 18/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-166-125:762818:763263 [3] NCCL INFO Channel 21/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-167-9:1039538:1039973 [1] NCCL INFO Channel 11/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-167-9:1039539:1039975 [2] NCCL INFO Channel 07/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-167-9:1039540:1039974 [3] NCCL INFO Channel 12/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-167-9:1039542:1039976 [5] NCCL INFO Channel 12/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-125:762818:763263 [3] NCCL INFO Channel 22/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-162-46:702887:703317 [7] NCCL INFO Channel 17/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-166-125:762818:763263 [3] NCCL INFO Channel 23/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-167-9:1039543:1039970 [6] NCCL INFO Channel 07/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-167-9:1039541:1039971 [4] NCCL INFO Channel 19/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-167-9:1039538:1039973 [1] NCCL INFO Channel 12/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-162-46:702886:703315 [6] NCCL INFO Channel 00/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-167-9:1039539:1039975 [2] NCCL INFO Channel 08/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-167-9:1039542:1039976 [5] NCCL INFO Channel 13/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-167-9:1039540:1039974 [3] NCCL INFO Channel 13/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-15:127293:127728 [3] NCCL INFO Connected all rings
+ip-26-0-162-46:702887:703317 [7] NCCL INFO Channel 18/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-166-125:762817:763265 [2] NCCL INFO Channel 10/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-125:762821:763264 [6] NCCL INFO Channel 22/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-162-46:702886:703315 [6] NCCL INFO Channel 01/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-162-46:702887:703317 [7] NCCL INFO Channel 19/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-166-15:127297:127731 [7] NCCL INFO Channel 01/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-167-9:1039543:1039970 [6] NCCL INFO Channel 08/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-162-46:702886:703315 [6] NCCL INFO Channel 02/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-162-46:702887:703317 [7] NCCL INFO Channel 20/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-167-9:1039542:1039976 [5] NCCL INFO Channel 14/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-167-9:1039539:1039975 [2] NCCL INFO Channel 09/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-167-9:1039541:1039971 [4] NCCL INFO Channel 20/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-167-9:1039543:1039970 [6] NCCL INFO Channel 09/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-167-9:1039542:1039976 [5] NCCL INFO Channel 15/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-167-9:1039539:1039975 [2] NCCL INFO Channel 10/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-162-46:702886:703315 [6] NCCL INFO Channel 03/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-162-46:702887:703317 [7] NCCL INFO Channel 21/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-166-36:104435:104863 [6] NCCL INFO Channel 21/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-167-9:1039541:1039971 [4] NCCL INFO Channel 21/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-167-9:1039543:1039970 [6] NCCL INFO Channel 10/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-166-214:624945:625380 [2] NCCL INFO Channel 19/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-167-9:1039542:1039976 [5] NCCL INFO Channel 16/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-167-9:1039539:1039975 [2] NCCL INFO Channel 11/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-167-9:1039540:1039974 [3] NCCL INFO Channel 14/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-167-9:1039541:1039971 [4] NCCL INFO Channel 22/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-167-9:1039543:1039970 [6] NCCL INFO Channel 11/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-167-9:1039542:1039976 [5] NCCL INFO Channel 17/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-125:762817:763265 [2] NCCL INFO Channel 11/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-167-9:1039539:1039975 [2] NCCL INFO Channel 12/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-125:762821:763264 [6] NCCL INFO Channel 23/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-166-15:127297:127731 [7] NCCL INFO Channel 02/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-167-9:1039540:1039974 [3] NCCL INFO Channel 15/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-162-46:702886:703315 [6] NCCL INFO Channel 04/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-167-9:1039541:1039971 [4] NCCL INFO Channel 23/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-167-9:1039543:1039970 [6] NCCL INFO Channel 12/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-167-9:1039542:1039976 [5] NCCL INFO Channel 18/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-167-9:1039539:1039975 [2] NCCL INFO Channel 13/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-167-9:1039540:1039974 [3] NCCL INFO Channel 16/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-162-46:702886:703315 [6] NCCL INFO Channel 05/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-167-9:1039543:1039970 [6] NCCL INFO Channel 13/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-167-9:1039542:1039976 [5] NCCL INFO Channel 19/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-15:127291:127730 [1] NCCL INFO Connected all rings
+ip-26-0-167-9:1039539:1039975 [2] NCCL INFO Channel 14/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-167-9:1039540:1039974 [3] NCCL INFO Channel 17/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-162-46:702886:703315 [6] NCCL INFO Channel 06/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-167-9:1039543:1039970 [6] NCCL INFO Channel 14/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-167-9:1039542:1039976 [5] NCCL INFO Channel 20/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-167-9:1039540:1039974 [3] NCCL INFO Channel 18/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-167-9:1039538:1039973 [1] NCCL INFO Channel 13/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-167-9:1039539:1039975 [2] NCCL INFO Channel 15/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-167-9:1039543:1039970 [6] NCCL INFO Channel 15/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-167-9:1039542:1039976 [5] NCCL INFO Channel 21/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-162-46:702887:703317 [7] NCCL INFO Channel 22/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-162-46:702884:703319 [4] NCCL INFO Channel 00/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-167-9:1039540:1039974 [3] NCCL INFO Channel 19/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-214:624944:625379 [1] NCCL INFO Channel 23/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-162-46:702887:703317 [7] NCCL INFO Channel 23/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-167-9:1039538:1039973 [1] NCCL INFO Channel 14/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-167-9:1039543:1039970 [6] NCCL INFO Channel 16/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-167-9:1039539:1039975 [2] NCCL INFO Channel 16/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-167-9:1039542:1039976 [5] NCCL INFO Channel 22/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-162-46:702884:703319 [4] NCCL INFO Channel 01/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-167-9:1039540:1039974 [3] NCCL INFO Channel 20/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-162-46:702884:703319 [4] NCCL INFO Channel 02/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-167-9:1039538:1039973 [1] NCCL INFO Channel 15/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-167-9:1039543:1039970 [6] NCCL INFO Channel 17/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-162-46:702882:703321 [2] NCCL INFO Channel 00/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-167-9:1039542:1039976 [5] NCCL INFO Channel 23/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-167-9:1039539:1039975 [2] NCCL INFO Channel 17/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-167-9:1039540:1039974 [3] NCCL INFO Channel 21/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-125:762817:763265 [2] NCCL INFO Channel 12/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-162-46:702884:703319 [4] NCCL INFO Channel 03/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-162-46:702882:703321 [2] NCCL INFO Channel 01/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-15:127297:127731 [7] NCCL INFO Channel 03/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-167-9:1039538:1039973 [1] NCCL INFO Channel 16/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-162-46:702884:703319 [4] NCCL INFO Channel 04/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-167-9:1039543:1039970 [6] NCCL INFO Channel 18/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-167-9:1039539:1039975 [2] NCCL INFO Channel 18/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-167-9:1039540:1039974 [3] NCCL INFO Channel 22/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-162-46:702882:703321 [2] NCCL INFO Channel 02/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-214:624945:625380 [2] NCCL INFO Channel 20/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-162-46:702884:703319 [4] NCCL INFO Channel 05/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-167-9:1039538:1039973 [1] NCCL INFO Channel 17/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-167-9:1039543:1039970 [6] NCCL INFO Channel 19/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-167-9:1039539:1039975 [2] NCCL INFO Channel 19/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-167-9:1039540:1039974 [3] NCCL INFO Channel 23/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-162-46:702882:703321 [2] NCCL INFO Channel 03/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-162-46:702883:703318 [3] NCCL INFO Channel 00/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-15:127296:127727 [6] NCCL INFO Connected all rings
+ip-26-0-162-46:702884:703319 [4] NCCL INFO Channel 06/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-166-36:104435:104863 [6] NCCL INFO Channel 22/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-167-9:1039543:1039970 [6] NCCL INFO Channel 20/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-167-9:1039538:1039973 [1] NCCL INFO Channel 18/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-167-9:1039539:1039975 [2] NCCL INFO Channel 20/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-162-46:702882:703321 [2] NCCL INFO Channel 04/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-162-46:702883:703318 [3] NCCL INFO Channel 01/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-162-46:702884:703319 [4] NCCL INFO Channel 07/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-162-46:702882:703321 [2] NCCL INFO Channel 05/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-162-46:702883:703318 [3] NCCL INFO Channel 02/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-162-46:702884:703319 [4] NCCL INFO Channel 08/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-162-46:702886:703315 [6] NCCL INFO Channel 07/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-167-9:1039539:1039975 [2] NCCL INFO Channel 21/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-167-9:1039543:1039970 [6] NCCL INFO Channel 21/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-162-46:702882:703321 [2] NCCL INFO Channel 06/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-162-46:702883:703318 [3] NCCL INFO Channel 03/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-162-46:702884:703319 [4] NCCL INFO Channel 09/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-162-46:702886:703315 [6] NCCL INFO Channel 08/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-166-15:127292:127729 [2] NCCL INFO Connected all rings
+ip-26-0-166-15:127297:127731 [7] NCCL INFO Channel 04/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-162-46:702882:703321 [2] NCCL INFO Channel 07/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-162-46:702883:703318 [3] NCCL INFO Channel 04/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-167-9:1039538:1039973 [1] NCCL INFO Channel 19/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-162-46:702884:703319 [4] NCCL INFO Channel 10/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-162-46:702885:703316 [5] NCCL INFO Channel 00/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-162-46:702886:703315 [6] NCCL INFO Channel 09/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-166-15:127297:127731 [7] NCCL INFO Channel 05/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-162-46:702883:703318 [3] NCCL INFO Channel 05/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-162-46:702885:703316 [5] NCCL INFO Channel 01/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-162-46:702886:703315 [6] NCCL INFO Channel 10/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-166-15:127297:127731 [7] NCCL INFO Channel 06/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-162-46:702883:703318 [3] NCCL INFO Channel 06/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-167-9:1039539:1039975 [2] NCCL INFO Channel 22/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-167-9:1039543:1039970 [6] NCCL INFO Channel 22/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-162-46:702884:703319 [4] NCCL INFO Channel 11/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-162-46:702885:703316 [5] NCCL INFO Channel 02/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-162-46:702886:703315 [6] NCCL INFO Channel 11/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-162-46:702883:703318 [3] NCCL INFO Channel 07/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-15:127297:127731 [7] NCCL INFO Channel 07/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-162-46:702885:703316 [5] NCCL INFO Channel 03/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-162-46:702884:703319 [4] NCCL INFO Channel 12/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-162-46:702886:703315 [6] NCCL INFO Channel 12/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-162-46:702882:703321 [2] NCCL INFO Channel 08/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-167-9:1039538:1039973 [1] NCCL INFO Channel 20/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-162-46:702883:703318 [3] NCCL INFO Channel 08/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-167-9:1039539:1039975 [2] NCCL INFO Channel 23/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-167-9:1039543:1039970 [6] NCCL INFO Channel 23/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-162-46:702885:703316 [5] NCCL INFO Channel 04/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-162-46:702884:703319 [4] NCCL INFO Channel 13/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-162-46:702886:703315 [6] NCCL INFO Channel 13/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-162-46:702882:703321 [2] NCCL INFO Channel 09/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-162-46:702883:703318 [3] NCCL INFO Channel 09/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-162-46:702885:703316 [5] NCCL INFO Channel 05/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-162-46:702884:703319 [4] NCCL INFO Channel 14/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-162-46:702886:703315 [6] NCCL INFO Channel 14/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-167-9:1039538:1039973 [1] NCCL INFO Channel 21/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-162-46:702882:703321 [2] NCCL INFO Channel 10/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-162-46:702883:703318 [3] NCCL INFO Channel 10/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-125:762817:763265 [2] NCCL INFO Channel 13/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-162-46:702885:703316 [5] NCCL INFO Channel 06/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-162-46:702886:703315 [6] NCCL INFO Channel 15/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-162-46:702884:703319 [4] NCCL INFO Channel 15/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-162-46:702883:703318 [3] NCCL INFO Channel 11/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-162-46:702882:703321 [2] NCCL INFO Channel 11/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-36:104435:104863 [6] NCCL INFO Channel 23/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-166-15:127297:127731 [7] NCCL INFO Channel 08/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-162-46:702885:703316 [5] NCCL INFO Channel 07/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-162-46:702886:703315 [6] NCCL INFO Channel 16/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-162-46:702884:703319 [4] NCCL INFO Channel 16/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-162-46:702883:703318 [3] NCCL INFO Channel 12/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-162-46:702882:703321 [2] NCCL INFO Channel 12/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-15:127297:127731 [7] NCCL INFO Channel 09/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-167-9:1039538:1039973 [1] NCCL INFO Channel 22/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-162-46:702885:703316 [5] NCCL INFO Channel 08/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-162-46:702886:703315 [6] NCCL INFO Channel 17/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-162-46:702883:703318 [3] NCCL INFO Channel 13/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-162-46:702884:703319 [4] NCCL INFO Channel 17/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-162-46:702882:703321 [2] NCCL INFO Channel 13/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-15:127297:127731 [7] NCCL INFO Channel 10/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-162-46:702885:703316 [5] NCCL INFO Channel 09/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-15:127297:127731 [7] NCCL INFO Channel 11/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-162-46:702883:703318 [3] NCCL INFO Channel 14/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-162-46:702886:703315 [6] NCCL INFO Channel 18/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-162-46:702884:703319 [4] NCCL INFO Channel 18/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-162-46:702882:703321 [2] NCCL INFO Channel 14/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-162-46:702881:703320 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127297:127731 [7] NCCL INFO Channel 12/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-162-46:702885:703316 [5] NCCL INFO Channel 10/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-162-46:702883:703318 [3] NCCL INFO Channel 15/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-162-46:702886:703315 [6] NCCL INFO Channel 19/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-162-46:702882:703321 [2] NCCL INFO Channel 15/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-162-46:702884:703319 [4] NCCL INFO Channel 19/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-162-46:702881:703320 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127297:127731 [7] NCCL INFO Channel 13/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-162-46:702885:703316 [5] NCCL INFO Channel 11/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-162-46:702886:703315 [6] NCCL INFO Channel 20/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-162-46:702883:703318 [3] NCCL INFO Channel 16/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-162-46:702882:703321 [2] NCCL INFO Channel 16/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-162-46:702884:703319 [4] NCCL INFO Channel 20/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-166-125:762817:763265 [2] NCCL INFO Channel 14/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-162-46:702881:703320 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127297:127731 [7] NCCL INFO Channel 14/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-162-46:702885:703316 [5] NCCL INFO Channel 12/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-162-46:702882:703321 [2] NCCL INFO Channel 17/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-162-46:702886:703315 [6] NCCL INFO Channel 21/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-162-46:702883:703318 [3] NCCL INFO Channel 17/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-162-46:702884:703319 [4] NCCL INFO Channel 21/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-162-46:702881:703320 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762815:763258 [0] NCCL INFO Connected all trees
+ip-26-0-166-125:762815:763258 [0] NCCL INFO NVLS comm 0x881f7d0 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-162-46:702885:703316 [5] NCCL INFO Channel 13/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-162-46:702886:703315 [6] NCCL INFO Channel 22/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-162-46:702883:703318 [3] NCCL INFO Channel 18/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-162-46:702882:703321 [2] NCCL INFO Channel 18/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-162-46:702884:703319 [4] NCCL INFO Channel 22/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-166-15:127297:127731 [7] NCCL INFO Channel 15/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-162-46:702881:703320 [1] NCCL INFO Channel 04/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-162-46:702885:703316 [5] NCCL INFO Channel 14/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-162-46:702884:703319 [4] NCCL INFO Channel 23/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-162-46:702886:703315 [6] NCCL INFO Channel 23/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-162-46:702883:703318 [3] NCCL INFO Channel 19/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-162-46:702885:703316 [5] NCCL INFO Channel 15/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-15:127297:127731 [7] NCCL INFO Channel 16/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-162-46:702883:703318 [3] NCCL INFO Channel 20/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-162-46:702885:703316 [5] NCCL INFO Channel 16/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-162-46:702882:703321 [2] NCCL INFO Channel 19/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-15:127297:127731 [7] NCCL INFO Channel 17/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-162-46:702881:703320 [1] NCCL INFO Channel 05/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-162-46:702883:703318 [3] NCCL INFO Channel 21/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-162-46:702885:703316 [5] NCCL INFO Channel 17/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-15:127297:127731 [7] NCCL INFO Channel 18/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-166-214:624945:625380 [2] NCCL INFO Channel 21/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-15:127297:127731 [7] NCCL INFO Channel 19/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-166-244:897240:897673 [0] NCCL INFO Connected all rings
+ip-26-0-166-15:127295:127726 [5] NCCL INFO Channel 00/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-15:127297:127731 [7] NCCL INFO Channel 20/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-166-15:127295:127726 [5] NCCL INFO Channel 01/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-15:127297:127731 [7] NCCL INFO Channel 21/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-166-15:127295:127726 [5] NCCL INFO Channel 02/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-15:127297:127731 [7] NCCL INFO Channel 22/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-166-15:127294:127725 [4] NCCL INFO Channel 00/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-162-46:702882:703321 [2] NCCL INFO Channel 20/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-15:127295:127726 [5] NCCL INFO Channel 03/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-162-46:702883:703318 [3] NCCL INFO Channel 22/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-162-46:702885:703316 [5] NCCL INFO Channel 18/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-162-46:702882:703321 [2] NCCL INFO Channel 21/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-15:127297:127731 [7] NCCL INFO Channel 23/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-166-15:127294:127725 [4] NCCL INFO Channel 01/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-166-125:762817:763265 [2] NCCL INFO Channel 15/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-15:127295:127726 [5] NCCL INFO Channel 04/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-15:127294:127725 [4] NCCL INFO Channel 02/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-166-15:127295:127726 [5] NCCL INFO Channel 05/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-15:127293:127728 [3] NCCL INFO Channel 00/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-15:127292:127729 [2] NCCL INFO Channel 00/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-15:127294:127725 [4] NCCL INFO Channel 03/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-166-214:624945:625380 [2] NCCL INFO Channel 22/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-15:127295:127726 [5] NCCL INFO Channel 06/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-15:127293:127728 [3] NCCL INFO Channel 01/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-15:127292:127729 [2] NCCL INFO Channel 01/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-15:127294:127725 [4] NCCL INFO Channel 04/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-162-46:702883:703318 [3] NCCL INFO Channel 23/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-162-46:702885:703316 [5] NCCL INFO Channel 19/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-162-46:702882:703321 [2] NCCL INFO Channel 22/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-15:127295:127726 [5] NCCL INFO Channel 07/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-15:127293:127728 [3] NCCL INFO Channel 02/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-15:127292:127729 [2] NCCL INFO Channel 02/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-15:127294:127725 [4] NCCL INFO Channel 05/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-166-15:127291:127730 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-167-9:1039538:1039973 [1] NCCL INFO Channel 23/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127295:127726 [5] NCCL INFO Channel 08/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-15:127292:127729 [2] NCCL INFO Channel 03/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-15:127293:127728 [3] NCCL INFO Channel 03/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-15:127291:127730 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127294:127725 [4] NCCL INFO Channel 06/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-162-46:702881:703320 [1] NCCL INFO Channel 06/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-162-46:702885:703316 [5] NCCL INFO Channel 20/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-162-46:702882:703321 [2] NCCL INFO Channel 23/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-15:127295:127726 [5] NCCL INFO Channel 09/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-15:127292:127729 [2] NCCL INFO Channel 04/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-15:127293:127728 [3] NCCL INFO Channel 04/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-15:127291:127730 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127294:127725 [4] NCCL INFO Channel 07/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-166-125:762817:763265 [2] NCCL INFO Channel 16/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-162-46:702881:703320 [1] NCCL INFO Channel 07/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127295:127726 [5] NCCL INFO Channel 10/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-244:897247:897677 [7] NCCL INFO Connected all rings
+ip-26-0-166-244:897247:897677 [7] NCCL INFO Channel 00/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-166-15:127294:127725 [4] NCCL INFO Channel 08/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-166-15:127295:127726 [5] NCCL INFO Channel 11/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-15:127291:127730 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-162-46:702885:703316 [5] NCCL INFO Channel 21/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-15:127294:127725 [4] NCCL INFO Channel 09/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-166-15:127295:127726 [5] NCCL INFO Channel 12/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-15:127296:127727 [6] NCCL INFO Channel 00/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-166-15:127292:127729 [2] NCCL INFO Channel 05/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-15:127293:127728 [3] NCCL INFO Channel 05/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-15:127291:127730 [1] NCCL INFO Channel 04/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127294:127725 [4] NCCL INFO Channel 10/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-166-15:127295:127726 [5] NCCL INFO Channel 13/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-15:127296:127727 [6] NCCL INFO Channel 01/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-162-46:702881:703320 [1] NCCL INFO Channel 08/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127292:127729 [2] NCCL INFO Channel 06/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-15:127293:127728 [3] NCCL INFO Channel 06/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-15:127291:127730 [1] NCCL INFO Channel 05/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127294:127725 [4] NCCL INFO Channel 11/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-166-15:127295:127726 [5] NCCL INFO Channel 14/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-15:127296:127727 [6] NCCL INFO Channel 02/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-166-15:127292:127729 [2] NCCL INFO Channel 07/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-15:127293:127728 [3] NCCL INFO Channel 07/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-15:127291:127730 [1] NCCL INFO Channel 06/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127294:127725 [4] NCCL INFO Channel 12/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-166-125:762817:763265 [2] NCCL INFO Channel 17/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-15:127295:127726 [5] NCCL INFO Channel 15/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-15:127296:127727 [6] NCCL INFO Channel 03/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-162-46:702885:703316 [5] NCCL INFO Channel 22/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-15:127292:127729 [2] NCCL INFO Channel 08/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-15:127294:127725 [4] NCCL INFO Channel 13/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-166-15:127291:127730 [1] NCCL INFO Channel 07/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127293:127728 [3] NCCL INFO Channel 08/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-15:127295:127726 [5] NCCL INFO Channel 16/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-15:127296:127727 [6] NCCL INFO Channel 04/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-166-214:624945:625380 [2] NCCL INFO Channel 23/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-15:127292:127729 [2] NCCL INFO Channel 09/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-15:127294:127725 [4] NCCL INFO Channel 14/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-166-15:127293:127728 [3] NCCL INFO Channel 09/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-15:127291:127730 [1] NCCL INFO Channel 08/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127295:127726 [5] NCCL INFO Channel 17/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-15:127296:127727 [6] NCCL INFO Channel 05/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-162-46:702881:703320 [1] NCCL INFO Channel 09/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-162-46:702881:703320 [1] NCCL INFO Channel 10/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-162-46:702881:703320 [1] NCCL INFO Channel 11/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-162-46:702881:703320 [1] NCCL INFO Channel 12/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127292:127729 [2] NCCL INFO Channel 10/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-15:127294:127725 [4] NCCL INFO Channel 15/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-166-15:127291:127730 [1] NCCL INFO Channel 09/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127293:127728 [3] NCCL INFO Channel 10/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-15:127295:127726 [5] NCCL INFO Channel 18/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-15:127296:127727 [6] NCCL INFO Channel 06/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-162-46:702881:703320 [1] NCCL INFO Channel 13/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127292:127729 [2] NCCL INFO Channel 11/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-15:127291:127730 [1] NCCL INFO Channel 10/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127294:127725 [4] NCCL INFO Channel 16/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-166-15:127293:127728 [3] NCCL INFO Channel 11/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-15:127295:127726 [5] NCCL INFO Channel 19/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-15:127296:127727 [6] NCCL INFO Channel 07/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-166-15:127292:127729 [2] NCCL INFO Channel 12/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-15:127291:127730 [1] NCCL INFO Channel 11/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127295:127726 [5] NCCL INFO Channel 20/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-15:127294:127725 [4] NCCL INFO Channel 17/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-166-15:127293:127728 [3] NCCL INFO Channel 12/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-15:127296:127727 [6] NCCL INFO Channel 08/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-166-15:127292:127729 [2] NCCL INFO Channel 13/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-15:127295:127726 [5] NCCL INFO Channel 21/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-15:127293:127728 [3] NCCL INFO Channel 13/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-15:127291:127730 [1] NCCL INFO Channel 12/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127294:127725 [4] NCCL INFO Channel 18/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-166-15:127296:127727 [6] NCCL INFO Channel 09/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-162-46:702885:703316 [5] NCCL INFO Channel 23/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-162-46:702881:703320 [1] NCCL INFO Channel 14/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127295:127726 [5] NCCL INFO Channel 22/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-15:127291:127730 [1] NCCL INFO Channel 13/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127292:127729 [2] NCCL INFO Channel 14/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-15:127293:127728 [3] NCCL INFO Channel 14/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-15:127294:127725 [4] NCCL INFO Channel 19/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-166-15:127296:127727 [6] NCCL INFO Channel 10/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-166-15:127295:127726 [5] NCCL INFO Channel 23/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-15:127291:127730 [1] NCCL INFO Channel 14/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127294:127725 [4] NCCL INFO Channel 20/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-166-15:127293:127728 [3] NCCL INFO Channel 15/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-15:127292:127729 [2] NCCL INFO Channel 15/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-15:127296:127727 [6] NCCL INFO Channel 11/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-166-125:762817:763265 [2] NCCL INFO Channel 18/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-15:127292:127729 [2] NCCL INFO Channel 16/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-15:127291:127730 [1] NCCL INFO Channel 15/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127293:127728 [3] NCCL INFO Channel 16/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-15:127294:127725 [4] NCCL INFO Channel 21/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-166-15:127296:127727 [6] NCCL INFO Channel 12/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-166-15:127292:127729 [2] NCCL INFO Channel 17/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-15:127291:127730 [1] NCCL INFO Channel 16/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127293:127728 [3] NCCL INFO Channel 17/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-162-46:702881:703320 [1] NCCL INFO Channel 15/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127294:127725 [4] NCCL INFO Channel 22/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-166-15:127296:127727 [6] NCCL INFO Channel 13/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-166-15:127292:127729 [2] NCCL INFO Channel 18/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-15:127294:127725 [4] NCCL INFO Channel 23/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-166-15:127291:127730 [1] NCCL INFO Channel 17/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127293:127728 [3] NCCL INFO Channel 18/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-15:127296:127727 [6] NCCL INFO Channel 14/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-166-15:127292:127729 [2] NCCL INFO Channel 19/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-15:127291:127730 [1] NCCL INFO Channel 18/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127293:127728 [3] NCCL INFO Channel 19/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-15:127296:127727 [6] NCCL INFO Channel 15/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-166-15:127292:127729 [2] NCCL INFO Channel 20/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-15:127291:127730 [1] NCCL INFO Channel 19/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127293:127728 [3] NCCL INFO Channel 20/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-15:127296:127727 [6] NCCL INFO Channel 16/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-166-15:127291:127730 [1] NCCL INFO Channel 20/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127293:127728 [3] NCCL INFO Channel 21/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-15:127292:127729 [2] NCCL INFO Channel 21/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-15:127296:127727 [6] NCCL INFO Channel 17/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-166-15:127291:127730 [1] NCCL INFO Channel 21/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127293:127728 [3] NCCL INFO Channel 22/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-15:127292:127729 [2] NCCL INFO Channel 22/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-15:127296:127727 [6] NCCL INFO Channel 18/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-166-15:127291:127730 [1] NCCL INFO Channel 22/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127293:127728 [3] NCCL INFO Channel 23/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-15:127292:127729 [2] NCCL INFO Channel 23/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-162-46:702881:703320 [1] NCCL INFO Channel 16/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127296:127727 [6] NCCL INFO Channel 19/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-166-15:127291:127730 [1] NCCL INFO Channel 23/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127296:127727 [6] NCCL INFO Channel 20/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-166-244:897247:897677 [7] NCCL INFO Channel 01/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-166-125:762817:763265 [2] NCCL INFO Channel 19/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-15:127296:127727 [6] NCCL INFO Channel 21/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-162-46:702881:703320 [1] NCCL INFO Channel 17/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127296:127727 [6] NCCL INFO Channel 22/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-166-15:127296:127727 [6] NCCL INFO Channel 23/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-166-125:762817:763265 [2] NCCL INFO Channel 20/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-125:762817:763265 [2] NCCL INFO Channel 21/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-244:897247:897677 [7] NCCL INFO Channel 02/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-162-46:702881:703320 [1] NCCL INFO Channel 18/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762817:763265 [2] NCCL INFO Channel 22/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-244:897247:897677 [7] NCCL INFO Channel 03/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-166-125:762817:763265 [2] NCCL INFO Channel 23/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-162-46:702881:703320 [1] NCCL INFO Channel 19/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-244:897247:897677 [7] NCCL INFO Channel 04/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-166-244:897246:897680 [6] NCCL INFO Connected all rings
+ip-26-0-165-213:9631:10062 [6] NCCL INFO Connected all rings
+ip-26-0-162-46:702881:703320 [1] NCCL INFO Channel 20/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-213:9625:10059 [0] NCCL INFO Connected all rings
+ip-26-0-166-244:897247:897677 [7] NCCL INFO Channel 05/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-166-125:762822:763260 [7] NCCL INFO Connected all trees
+ip-26-0-166-125:762822:763260 [7] NCCL INFO NVLS comm 0xa233e80 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-213:9627:10064 [2] NCCL INFO Connected all rings
+ip-26-0-165-213:9632:10065 [7] NCCL INFO Connected all rings
+ip-26-0-165-213:9632:10065 [7] NCCL INFO Channel 00/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-162-46:702881:703320 [1] NCCL INFO Channel 21/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-213:9626:10061 [1] NCCL INFO Connected all rings
+ip-26-0-165-213:9632:10065 [7] NCCL INFO Channel 01/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-162-46:702887:703317 [7] NCCL INFO Connected all trees
+ip-26-0-162-46:702887:703317 [7] NCCL INFO NVLS comm 0x8d9c450 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-162-46:702881:703320 [1] NCCL INFO Channel 22/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-36:104429:104861 [0] NCCL INFO Connected all trees
+ip-26-0-166-36:104429:104861 [0] NCCL INFO NVLS comm 0xa6b1af0 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-213:9630:10060 [5] NCCL INFO Connected all rings
+ip-26-0-165-213:9632:10065 [7] NCCL INFO Channel 02/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-165-213:9629:10063 [4] NCCL INFO Connected all rings
+ip-26-0-162-46:702881:703320 [1] NCCL INFO Channel 23/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-213:9628:10066 [3] NCCL INFO Connected all rings
+ip-26-0-166-244:897247:897677 [7] NCCL INFO Channel 06/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-165-213:9632:10065 [7] NCCL INFO Channel 03/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-165-213:9632:10065 [7] NCCL INFO Channel 04/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-165-213:9632:10065 [7] NCCL INFO Channel 05/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-166-244:897245:897676 [5] NCCL INFO Connected all rings
+ip-26-0-165-213:9632:10065 [7] NCCL INFO Channel 06/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-166-244:897241:897675 [1] NCCL INFO Connected all rings
+ip-26-0-165-213:9632:10065 [7] NCCL INFO Channel 07/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-165-213:9632:10065 [7] NCCL INFO Channel 08/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-166-244:897244:897674 [4] NCCL INFO Connected all rings
+ip-26-0-166-244:897247:897677 [7] NCCL INFO Channel 07/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-165-213:9632:10065 [7] NCCL INFO Channel 09/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-165-213:9632:10065 [7] NCCL INFO Channel 10/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-166-214:624950:625376 [7] NCCL INFO Connected all trees
+ip-26-0-166-214:624950:625376 [7] NCCL INFO NVLS comm 0x8ff2880 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-213:9632:10065 [7] NCCL INFO Channel 11/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-165-213:9632:10065 [7] NCCL INFO Channel 12/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-165-213:9632:10065 [7] NCCL INFO Channel 13/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-165-213:9632:10065 [7] NCCL INFO Channel 14/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-166-244:897247:897677 [7] NCCL INFO Channel 08/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-165-213:9632:10065 [7] NCCL INFO Channel 15/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-166-244:897242:897679 [2] NCCL INFO Connected all rings
+ip-26-0-165-213:9632:10065 [7] NCCL INFO Channel 16/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-165-213:9632:10065 [7] NCCL INFO Channel 17/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-166-244:897243:897678 [3] NCCL INFO Connected all rings
+ip-26-0-166-244:897247:897677 [7] NCCL INFO Channel 09/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-165-213:9632:10065 [7] NCCL INFO Channel 18/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-166-244:897247:897677 [7] NCCL INFO Channel 10/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-165-213:9631:10062 [6] NCCL INFO Channel 00/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-213:9632:10065 [7] NCCL INFO Channel 19/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-165-213:9631:10062 [6] NCCL INFO Channel 01/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-166-244:897247:897677 [7] NCCL INFO Channel 11/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-165-213:9632:10065 [7] NCCL INFO Channel 20/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-165-213:9631:10062 [6] NCCL INFO Channel 02/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-166-244:897247:897677 [7] NCCL INFO Channel 12/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-165-213:9632:10065 [7] NCCL INFO Channel 21/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-165-213:9631:10062 [6] NCCL INFO Channel 03/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-166-244:897247:897677 [7] NCCL INFO Channel 13/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-165-213:9632:10065 [7] NCCL INFO Channel 22/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-165-213:9630:10060 [5] NCCL INFO Channel 00/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-213:9631:10062 [6] NCCL INFO Channel 04/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-166-244:897247:897677 [7] NCCL INFO Channel 14/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-165-213:9632:10065 [7] NCCL INFO Channel 23/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-165-213:9629:10063 [4] NCCL INFO Channel 00/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-165-213:9630:10060 [5] NCCL INFO Channel 01/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-213:9627:10064 [2] NCCL INFO Channel 00/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-165-213:9631:10062 [6] NCCL INFO Channel 05/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-166-244:897247:897677 [7] NCCL INFO Channel 15/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-165-213:9629:10063 [4] NCCL INFO Channel 01/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-165-213:9630:10060 [5] NCCL INFO Channel 02/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-213:9627:10064 [2] NCCL INFO Channel 01/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-165-213:9631:10062 [6] NCCL INFO Channel 06/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-213:9629:10063 [4] NCCL INFO Channel 02/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-166-244:897247:897677 [7] NCCL INFO Channel 16/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-165-213:9630:10060 [5] NCCL INFO Channel 03/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-213:9627:10064 [2] NCCL INFO Channel 02/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-165-213:9631:10062 [6] NCCL INFO Channel 07/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-213:9629:10063 [4] NCCL INFO Channel 03/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-165-213:9630:10060 [5] NCCL INFO Channel 04/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-213:9627:10064 [2] NCCL INFO Channel 03/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-165-213:9631:10062 [6] NCCL INFO Channel 08/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-166-244:897247:897677 [7] NCCL INFO Channel 17/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-165-213:9629:10063 [4] NCCL INFO Channel 04/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-165-213:9630:10060 [5] NCCL INFO Channel 05/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-213:9627:10064 [2] NCCL INFO Channel 04/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-165-213:9631:10062 [6] NCCL INFO Channel 09/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-166-244:897247:897677 [7] NCCL INFO Channel 18/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-165-213:9626:10061 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-213:9629:10063 [4] NCCL INFO Channel 05/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-165-213:9631:10062 [6] NCCL INFO Channel 10/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-213:9630:10060 [5] NCCL INFO Channel 06/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-213:9627:10064 [2] NCCL INFO Channel 05/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-244:897247:897677 [7] NCCL INFO Channel 19/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-165-213:9626:10061 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-213:9629:10063 [4] NCCL INFO Channel 06/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-165-213:9631:10062 [6] NCCL INFO Channel 11/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-213:9630:10060 [5] NCCL INFO Channel 07/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-213:9627:10064 [2] NCCL INFO Channel 06/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-165-213:9626:10061 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-244:897247:897677 [7] NCCL INFO Channel 20/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-165-213:9629:10063 [4] NCCL INFO Channel 07/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-165-213:9631:10062 [6] NCCL INFO Channel 12/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-213:9627:10064 [2] NCCL INFO Channel 07/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-165-213:9630:10060 [5] NCCL INFO Channel 08/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-213:9626:10061 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-213:9628:10066 [3] NCCL INFO Channel 00/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-244:897247:897677 [7] NCCL INFO Channel 21/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-165-213:9629:10063 [4] NCCL INFO Channel 08/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-165-213:9627:10064 [2] NCCL INFO Channel 08/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-165-213:9631:10062 [6] NCCL INFO Channel 13/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-213:9630:10060 [5] NCCL INFO Channel 09/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-244:897247:897677 [7] NCCL INFO Channel 22/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-165-213:9626:10061 [1] NCCL INFO Channel 04/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-213:9628:10066 [3] NCCL INFO Channel 01/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-244:897247:897677 [7] NCCL INFO Channel 23/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-165-213:9629:10063 [4] NCCL INFO Channel 09/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-165-213:9627:10064 [2] NCCL INFO Channel 09/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-165-213:9631:10062 [6] NCCL INFO Channel 14/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-213:9630:10060 [5] NCCL INFO Channel 10/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-213:9626:10061 [1] NCCL INFO Channel 05/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-213:9628:10066 [3] NCCL INFO Channel 02/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-165-213:9629:10063 [4] NCCL INFO Channel 10/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-165-213:9627:10064 [2] NCCL INFO Channel 10/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-165-213:9631:10062 [6] NCCL INFO Channel 15/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-213:9630:10060 [5] NCCL INFO Channel 11/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-213:9626:10061 [1] NCCL INFO Channel 06/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-213:9628:10066 [3] NCCL INFO Channel 03/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-165-213:9629:10063 [4] NCCL INFO Channel 11/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-165-213:9627:10064 [2] NCCL INFO Channel 11/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-165-213:9631:10062 [6] NCCL INFO Channel 16/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-213:9630:10060 [5] NCCL INFO Channel 12/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-213:9626:10061 [1] NCCL INFO Channel 07/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-213:9628:10066 [3] NCCL INFO Channel 04/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-165-213:9629:10063 [4] NCCL INFO Channel 12/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-165-213:9627:10064 [2] NCCL INFO Channel 12/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-165-213:9630:10060 [5] NCCL INFO Channel 13/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-213:9631:10062 [6] NCCL INFO Channel 17/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-213:9626:10061 [1] NCCL INFO Channel 08/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-213:9628:10066 [3] NCCL INFO Channel 05/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-165-213:9629:10063 [4] NCCL INFO Channel 13/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-165-213:9627:10064 [2] NCCL INFO Channel 13/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-165-213:9630:10060 [5] NCCL INFO Channel 14/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-213:9626:10061 [1] NCCL INFO Channel 09/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-213:9631:10062 [6] NCCL INFO Channel 18/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-213:9628:10066 [3] NCCL INFO Channel 06/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-165-213:9627:10064 [2] NCCL INFO Channel 14/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-165-213:9629:10063 [4] NCCL INFO Channel 14/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-165-213:9626:10061 [1] NCCL INFO Channel 10/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-213:9630:10060 [5] NCCL INFO Channel 15/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-213:9631:10062 [6] NCCL INFO Channel 19/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-213:9628:10066 [3] NCCL INFO Channel 07/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-165-213:9627:10064 [2] NCCL INFO Channel 15/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-165-213:9629:10063 [4] NCCL INFO Channel 15/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-165-213:9626:10061 [1] NCCL INFO Channel 11/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-213:9630:10060 [5] NCCL INFO Channel 16/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-213:9631:10062 [6] NCCL INFO Channel 20/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-213:9628:10066 [3] NCCL INFO Channel 08/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-244:897241:897675 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-213:9627:10064 [2] NCCL INFO Channel 16/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-165-213:9631:10062 [6] NCCL INFO Channel 21/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-213:9629:10063 [4] NCCL INFO Channel 16/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-165-213:9626:10061 [1] NCCL INFO Channel 12/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-213:9630:10060 [5] NCCL INFO Channel 17/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-213:9628:10066 [3] NCCL INFO Channel 09/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-165-213:9627:10064 [2] NCCL INFO Channel 17/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-165-213:9629:10063 [4] NCCL INFO Channel 17/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-165-213:9631:10062 [6] NCCL INFO Channel 22/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-213:9626:10061 [1] NCCL INFO Channel 13/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-213:9630:10060 [5] NCCL INFO Channel 18/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-213:9628:10066 [3] NCCL INFO Channel 10/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-244:897241:897675 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-244:897246:897680 [6] NCCL INFO Channel 00/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-166-244:897241:897675 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-213:9627:10064 [2] NCCL INFO Channel 18/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-165-213:9629:10063 [4] NCCL INFO Channel 18/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-165-213:9626:10061 [1] NCCL INFO Channel 14/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-213:9631:10062 [6] NCCL INFO Channel 23/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-165-213:9630:10060 [5] NCCL INFO Channel 19/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-213:9628:10066 [3] NCCL INFO Channel 11/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-244:897244:897674 [4] NCCL INFO Channel 00/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-166-244:897246:897680 [6] NCCL INFO Channel 01/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-166-244:897241:897675 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-244:897243:897678 [3] NCCL INFO Channel 00/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-165-213:9627:10064 [2] NCCL INFO Channel 19/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-165-213:9629:10063 [4] NCCL INFO Channel 19/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-165-213:9626:10061 [1] NCCL INFO Channel 15/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-213:9630:10060 [5] NCCL INFO Channel 20/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-213:9628:10066 [3] NCCL INFO Channel 12/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-244:897244:897674 [4] NCCL INFO Channel 01/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-166-244:897246:897680 [6] NCCL INFO Channel 02/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-166-244:897241:897675 [1] NCCL INFO Channel 04/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-244:897243:897678 [3] NCCL INFO Channel 01/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-244:897244:897674 [4] NCCL INFO Channel 02/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-166-244:897246:897680 [6] NCCL INFO Channel 03/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-166-244:897241:897675 [1] NCCL INFO Channel 05/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-244:897243:897678 [3] NCCL INFO Channel 02/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-244:897244:897674 [4] NCCL INFO Channel 03/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-165-213:9627:10064 [2] NCCL INFO Channel 20/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-165-213:9629:10063 [4] NCCL INFO Channel 20/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-165-213:9630:10060 [5] NCCL INFO Channel 21/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-213:9626:10061 [1] NCCL INFO Channel 16/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-213:9628:10066 [3] NCCL INFO Channel 13/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-244:897246:897680 [6] NCCL INFO Channel 04/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-166-244:897241:897675 [1] NCCL INFO Channel 06/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-244:897244:897674 [4] NCCL INFO Channel 04/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-166-244:897243:897678 [3] NCCL INFO Channel 03/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-244:897242:897679 [2] NCCL INFO Channel 00/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-244:897246:897680 [6] NCCL INFO Channel 05/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-166-244:897241:897675 [1] NCCL INFO Channel 07/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-244:897244:897674 [4] NCCL INFO Channel 05/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-166-244:897243:897678 [3] NCCL INFO Channel 04/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-244:897242:897679 [2] NCCL INFO Channel 01/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-244:897246:897680 [6] NCCL INFO Channel 06/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-166-244:897241:897675 [1] NCCL INFO Channel 08/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-244:897244:897674 [4] NCCL INFO Channel 06/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-166-244:897243:897678 [3] NCCL INFO Channel 05/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-165-213:9627:10064 [2] NCCL INFO Channel 21/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-165-213:9629:10063 [4] NCCL INFO Channel 21/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-165-213:9630:10060 [5] NCCL INFO Channel 22/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-213:9626:10061 [1] NCCL INFO Channel 17/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-213:9628:10066 [3] NCCL INFO Channel 14/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-244:897242:897679 [2] NCCL INFO Channel 02/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-244:897246:897680 [6] NCCL INFO Channel 07/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-166-244:897241:897675 [1] NCCL INFO Channel 09/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-244:897244:897674 [4] NCCL INFO Channel 07/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-166-244:897243:897678 [3] NCCL INFO Channel 06/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-244:897246:897680 [6] NCCL INFO Channel 08/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-166-244:897245:897676 [5] NCCL INFO Channel 00/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-244:897241:897675 [1] NCCL INFO Channel 10/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-244:897243:897678 [3] NCCL INFO Channel 07/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-244:897246:897680 [6] NCCL INFO Channel 09/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-166-244:897245:897676 [5] NCCL INFO Channel 01/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-244:897242:897679 [2] NCCL INFO Channel 03/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-244:897241:897675 [1] NCCL INFO Channel 11/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-244:897243:897678 [3] NCCL INFO Channel 08/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-244:897246:897680 [6] NCCL INFO Channel 10/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-166-244:897245:897676 [5] NCCL INFO Channel 02/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-213:9627:10064 [2] NCCL INFO Channel 22/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-165-213:9629:10063 [4] NCCL INFO Channel 22/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-165-213:9630:10060 [5] NCCL INFO Channel 23/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-213:9626:10061 [1] NCCL INFO Channel 18/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-213:9628:10066 [3] NCCL INFO Channel 15/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-244:897242:897679 [2] NCCL INFO Channel 04/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-244:897241:897675 [1] NCCL INFO Channel 12/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-244:897243:897678 [3] NCCL INFO Channel 09/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-244:897246:897680 [6] NCCL INFO Channel 11/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-166-244:897245:897676 [5] NCCL INFO Channel 03/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-244:897242:897679 [2] NCCL INFO Channel 05/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-244:897241:897675 [1] NCCL INFO Channel 13/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-244:897243:897678 [3] NCCL INFO Channel 10/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-244:897245:897676 [5] NCCL INFO Channel 04/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-244:897246:897680 [6] NCCL INFO Channel 12/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-166-244:897242:897679 [2] NCCL INFO Channel 06/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-244:897241:897675 [1] NCCL INFO Channel 14/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-244:897243:897678 [3] NCCL INFO Channel 11/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-244:897245:897676 [5] NCCL INFO Channel 05/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-244:897246:897680 [6] NCCL INFO Channel 13/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-166-244:897242:897679 [2] NCCL INFO Channel 07/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-244:897241:897675 [1] NCCL INFO Channel 15/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-213:9627:10064 [2] NCCL INFO Channel 23/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-165-213:9629:10063 [4] NCCL INFO Channel 23/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-165-213:9626:10061 [1] NCCL INFO Channel 19/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-213:9628:10066 [3] NCCL INFO Channel 16/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-244:897245:897676 [5] NCCL INFO Channel 06/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-244:897243:897678 [3] NCCL INFO Channel 12/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-244:897246:897680 [6] NCCL INFO Channel 14/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-166-244:897241:897675 [1] NCCL INFO Channel 16/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-244:897242:897679 [2] NCCL INFO Channel 08/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-244:897244:897674 [4] NCCL INFO Channel 08/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-166-244:897245:897676 [5] NCCL INFO Channel 07/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-244:897243:897678 [3] NCCL INFO Channel 13/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-244:897246:897680 [6] NCCL INFO Channel 15/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-166-244:897241:897675 [1] NCCL INFO Channel 17/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-244:897242:897679 [2] NCCL INFO Channel 09/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-244:897244:897674 [4] NCCL INFO Channel 09/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-166-244:897245:897676 [5] NCCL INFO Channel 08/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-244:897243:897678 [3] NCCL INFO Channel 14/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-244:897246:897680 [6] NCCL INFO Channel 16/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-166-244:897241:897675 [1] NCCL INFO Channel 18/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-244:897242:897679 [2] NCCL INFO Channel 10/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-244:897244:897674 [4] NCCL INFO Channel 10/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-166-244:897245:897676 [5] NCCL INFO Channel 09/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-244:897246:897680 [6] NCCL INFO Channel 17/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-166-244:897243:897678 [3] NCCL INFO Channel 15/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-244:897241:897675 [1] NCCL INFO Channel 19/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-244:897242:897679 [2] NCCL INFO Channel 11/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-244:897244:897674 [4] NCCL INFO Channel 11/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-166-244:897245:897676 [5] NCCL INFO Channel 10/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-244:897246:897680 [6] NCCL INFO Channel 18/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-166-244:897241:897675 [1] NCCL INFO Channel 20/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-244:897243:897678 [3] NCCL INFO Channel 16/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-244:897242:897679 [2] NCCL INFO Channel 12/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-244:897244:897674 [4] NCCL INFO Channel 12/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-166-244:897245:897676 [5] NCCL INFO Channel 11/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-244:897241:897675 [1] NCCL INFO Channel 21/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-244:897246:897680 [6] NCCL INFO Channel 19/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-166-244:897243:897678 [3] NCCL INFO Channel 17/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-244:897242:897679 [2] NCCL INFO Channel 13/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-244:897244:897674 [4] NCCL INFO Channel 13/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-166-244:897245:897676 [5] NCCL INFO Channel 12/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-244:897241:897675 [1] NCCL INFO Channel 22/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-244:897246:897680 [6] NCCL INFO Channel 20/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-166-244:897242:897679 [2] NCCL INFO Channel 14/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-244:897244:897674 [4] NCCL INFO Channel 14/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-166-244:897243:897678 [3] NCCL INFO Channel 18/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-244:897245:897676 [5] NCCL INFO Channel 13/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-244:897241:897675 [1] NCCL INFO Channel 23/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-244:897246:897680 [6] NCCL INFO Channel 21/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-166-244:897242:897679 [2] NCCL INFO Channel 15/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-244:897244:897674 [4] NCCL INFO Channel 15/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-166-244:897243:897678 [3] NCCL INFO Channel 19/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-244:897245:897676 [5] NCCL INFO Channel 14/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-244:897242:897679 [2] NCCL INFO Channel 16/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-244:897246:897680 [6] NCCL INFO Channel 22/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-166-244:897244:897674 [4] NCCL INFO Channel 16/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-166-244:897243:897678 [3] NCCL INFO Channel 20/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-165-213:9626:10061 [1] NCCL INFO Channel 20/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-213:9628:10066 [3] NCCL INFO Channel 17/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-244:897245:897676 [5] NCCL INFO Channel 15/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-244:897242:897679 [2] NCCL INFO Channel 17/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-244:897246:897680 [6] NCCL INFO Channel 23/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-166-244:897244:897674 [4] NCCL INFO Channel 17/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-166-244:897243:897678 [3] NCCL INFO Channel 21/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-244:897245:897676 [5] NCCL INFO Channel 16/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-244:897245:897676 [5] NCCL INFO Channel 17/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-244:897245:897676 [5] NCCL INFO Channel 18/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-244:897245:897676 [5] NCCL INFO Channel 19/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-244:897245:897676 [5] NCCL INFO Channel 20/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-244:897245:897676 [5] NCCL INFO Channel 21/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-244:897245:897676 [5] NCCL INFO Channel 22/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-244:897245:897676 [5] NCCL INFO Channel 23/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-244:897242:897679 [2] NCCL INFO Channel 18/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-244:897244:897674 [4] NCCL INFO Channel 18/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-166-244:897243:897678 [3] NCCL INFO Channel 22/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-165-213:9626:10061 [1] NCCL INFO Channel 21/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-213:9626:10061 [1] NCCL INFO Channel 22/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624943:625375 [0] NCCL INFO Connected all trees
+ip-26-0-166-244:897242:897679 [2] NCCL INFO Channel 19/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-244:897244:897674 [4] NCCL INFO Channel 19/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-166-244:897243:897678 [3] NCCL INFO Channel 23/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-214:624943:625375 [0] NCCL INFO NVLS comm 0x9698270 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-213:9628:10066 [3] NCCL INFO Channel 18/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-36:104432:104866 [3] NCCL INFO Connected all trees
+ip-26-0-166-36:104432:104866 [3] NCCL INFO NVLS comm 0x8a51810 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-213:9626:10061 [1] NCCL INFO Channel 23/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-244:897242:897679 [2] NCCL INFO Channel 20/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-244:897244:897674 [4] NCCL INFO Channel 20/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-167-9:1039544:1039972 [7] NCCL INFO Connected all trees
+ip-26-0-165-213:9628:10066 [3] NCCL INFO Channel 19/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-167-9:1039544:1039972 [7] NCCL INFO NVLS comm 0x98df560 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-213:9628:10066 [3] NCCL INFO Channel 20/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-244:897242:897679 [2] NCCL INFO Channel 21/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-244:897244:897674 [4] NCCL INFO Channel 21/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-166-244:897242:897679 [2] NCCL INFO Channel 22/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-244:897242:897679 [2] NCCL INFO Channel 23/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-166-244:897244:897674 [4] NCCL INFO Channel 22/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-165-213:9628:10066 [3] NCCL INFO Channel 21/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-36:104436:104865 [7] NCCL INFO Connected all trees
+ip-26-0-165-213:9628:10066 [3] NCCL INFO Channel 22/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-36:104436:104865 [7] NCCL INFO NVLS comm 0x9987fb0 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-166-244:897244:897674 [4] NCCL INFO Channel 23/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-166-36:104433:104864 [4] NCCL INFO Connected all trees
+ip-26-0-166-36:104433:104864 [4] NCCL INFO NVLS comm 0xa4a2c80 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-213:9628:10066 [3] NCCL INFO Channel 23/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-15:127290:127724 [0] NCCL INFO Connected all trees
+ip-26-0-166-15:127290:127724 [0] NCCL INFO NVLS comm 0x8befd90 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-167-9:1039537:1039969 [0] NCCL INFO Connected all trees
+ip-26-0-167-9:1039537:1039969 [0] NCCL INFO NVLS comm 0x9e653d0 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-162-46:702883:703318 [3] NCCL INFO Connected all trees
+ip-26-0-162-46:702883:703318 [3] NCCL INFO NVLS comm 0x87ca1f0 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-166-214:624949:625381 [6] NCCL INFO Connected all trees
+ip-26-0-166-214:624949:625381 [6] NCCL INFO NVLS comm 0x8ccfda0 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-166-15:127297:127731 [7] NCCL INFO Connected all trees
+ip-26-0-166-15:127297:127731 [7] NCCL INFO NVLS comm 0xa0163c0 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-213:9632:10065 [7] NCCL INFO Connected all trees
+ip-26-0-165-213:9632:10065 [7] NCCL INFO NVLS comm 0x8f449d0 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-167-9:1039541:1039971 [4] NCCL INFO Connected all trees
+ip-26-0-167-9:1039541:1039971 [4] NCCL INFO NVLS comm 0xa6d9b30 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-166-36:104430:104868 [1] NCCL INFO Connected all trees
+ip-26-0-166-36:104430:104868 [1] NCCL INFO NVLS comm 0x9ecbed0 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-166-214:624948:625382 [5] NCCL INFO Connected all trees
+ip-26-0-166-214:624948:625382 [5] NCCL INFO NVLS comm 0xa2b4710 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-166-125:762821:763264 [6] NCCL INFO Connected all trees
+ip-26-0-166-125:762821:763264 [6] NCCL INFO NVLS comm 0x8afb540 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-162-46:702880:703314 [0] NCCL INFO Connected all trees
+ip-26-0-166-214:624947:625378 [4] NCCL INFO Connected all trees
+ip-26-0-162-46:702880:703314 [0] NCCL INFO NVLS comm 0x9b86160 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-166-214:624947:625378 [4] NCCL INFO NVLS comm 0xa23cf00 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-166-36:104431:104862 [2] NCCL INFO Connected all trees
+ip-26-0-166-36:104431:104862 [2] NCCL INFO NVLS comm 0x9a579f0 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-166-125:762819:763261 [4] NCCL INFO Connected all trees
+ip-26-0-166-125:762819:763261 [4] NCCL INFO NVLS comm 0x96329a0 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-166-125:762820:763259 [5] NCCL INFO Connected all trees
+ip-26-0-166-125:762820:763259 [5] NCCL INFO NVLS comm 0x95543d0 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-166-36:104434:104867 [5] NCCL INFO Connected all trees
+ip-26-0-166-36:104434:104867 [5] NCCL INFO NVLS comm 0x8cbe880 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-166-36:104435:104863 [6] NCCL INFO Connected all trees
+ip-26-0-166-36:104435:104863 [6] NCCL INFO NVLS comm 0x9017970 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-166-214:624944:625379 [1] NCCL INFO Connected all trees
+ip-26-0-166-214:624944:625379 [1] NCCL INFO NVLS comm 0x8768060 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-166-244:897247:897677 [7] NCCL INFO Connected all trees
+ip-26-0-166-244:897247:897677 [7] NCCL INFO NVLS comm 0x9c01630 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-166-244:897240:897673 [0] NCCL INFO Connected all trees
+ip-26-0-166-244:897240:897673 [0] NCCL INFO NVLS comm 0x956e780 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-162-46:702885:703316 [5] NCCL INFO Connected all trees
+ip-26-0-162-46:702885:703316 [5] NCCL INFO NVLS comm 0x8dac5b0 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-166-214:624946:625377 [3] NCCL INFO Connected all trees
+ip-26-0-166-214:624946:625377 [3] NCCL INFO NVLS comm 0xa29d720 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-166-214:624945:625380 [2] NCCL INFO Connected all trees
+ip-26-0-166-214:624945:625380 [2] NCCL INFO NVLS comm 0x9a046d0 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-162-46:702886:703315 [6] NCCL INFO Connected all trees
+ip-26-0-162-46:702886:703315 [6] NCCL INFO NVLS comm 0x9292e40 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-213:9625:10059 [0] NCCL INFO Connected all trees
+ip-26-0-166-15:127293:127728 [3] NCCL INFO Connected all trees
+ip-26-0-165-213:9625:10059 [0] NCCL INFO NVLS comm 0x9033500 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-166-15:127293:127728 [3] NCCL INFO NVLS comm 0xa35de50 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-166-125:762816:763262 [1] NCCL INFO Connected all trees
+ip-26-0-166-125:762816:763262 [1] NCCL INFO NVLS comm 0x984eea0 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-167-9:1039542:1039976 [5] NCCL INFO Connected all trees
+ip-26-0-167-9:1039542:1039976 [5] NCCL INFO NVLS comm 0x9095cf0 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-166-36:104435:104863 [6] NCCL INFO Connected NVLS tree
+ip-26-0-166-36:104435:104863 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-36:104435:104863 [6] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-166-36:104435:104863 [6] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-166-36:104429:104861 [0] NCCL INFO Connected NVLS tree
+ip-26-0-166-36:104429:104861 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-36:104429:104861 [0] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-166-36:104429:104861 [0] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-166-36:104431:104862 [2] NCCL INFO Connected NVLS tree
+ip-26-0-166-36:104431:104862 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-36:104431:104862 [2] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-166-36:104431:104862 [2] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-166-36:104436:104865 [7] NCCL INFO Connected NVLS tree
+ip-26-0-166-36:104436:104865 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-36:104436:104865 [7] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-166-36:104436:104865 [7] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-166-36:104433:104864 [4] NCCL INFO Connected NVLS tree
+ip-26-0-166-36:104433:104864 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-36:104433:104864 [4] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-166-36:104433:104864 [4] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-166-36:104430:104868 [1] NCCL INFO Connected NVLS tree
+ip-26-0-166-36:104430:104868 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-36:104434:104867 [5] NCCL INFO Connected NVLS tree
+ip-26-0-166-36:104430:104868 [1] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-166-36:104430:104868 [1] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-166-36:104434:104867 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-36:104434:104867 [5] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-166-36:104434:104867 [5] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-166-36:104432:104866 [3] NCCL INFO Connected NVLS tree
+ip-26-0-166-36:104432:104866 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-36:104432:104866 [3] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-166-36:104432:104866 [3] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-167-9:1039538:1039973 [1] NCCL INFO Connected all trees
+ip-26-0-162-46:702884:703319 [4] NCCL INFO Connected all trees
+ip-26-0-162-46:702884:703319 [4] NCCL INFO NVLS comm 0x9156360 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-167-9:1039538:1039973 [1] NCCL INFO NVLS comm 0x9283370 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-162-46:702882:703321 [2] NCCL INFO Connected all trees
+ip-26-0-167-9:1039540:1039974 [3] NCCL INFO Connected all trees
+ip-26-0-162-46:702882:703321 [2] NCCL INFO NVLS comm 0x9a08fa0 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-167-9:1039540:1039974 [3] NCCL INFO NVLS comm 0x99d3460 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-166-125:762817:763265 [2] NCCL INFO Connected all trees
+ip-26-0-166-125:762817:763265 [2] NCCL INFO NVLS comm 0xa64bcb0 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-167-9:1039539:1039975 [2] NCCL INFO Connected all trees
+ip-26-0-167-9:1039539:1039975 [2] NCCL INFO NVLS comm 0x9358300 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-167-9:1039543:1039970 [6] NCCL INFO Connected all trees
+ip-26-0-166-125:762818:763263 [3] NCCL INFO Connected all trees
+ip-26-0-167-9:1039543:1039970 [6] NCCL INFO NVLS comm 0xa1956c0 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-166-125:762818:763263 [3] NCCL INFO NVLS comm 0x89882d0 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-166-36:104436:104865 [7] NCCL INFO comm 0x9987fb0 rank 7 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0xd9fb77963d823d25 - Init COMPLETE
+ip-26-0-166-36:104432:104866 [3] NCCL INFO comm 0x8a51810 rank 3 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0xd9fb77963d823d25 - Init COMPLETE
+ip-26-0-166-36:104430:104868 [1] NCCL INFO comm 0x9ecbed0 rank 1 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0xd9fb77963d823d25 - Init COMPLETE
+ip-26-0-166-36:104434:104867 [5] NCCL INFO comm 0x8cbe880 rank 5 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0xd9fb77963d823d25 - Init COMPLETE
+ip-26-0-166-36:104435:104863 [6] NCCL INFO comm 0x9017970 rank 6 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0xd9fb77963d823d25 - Init COMPLETE
+ip-26-0-166-36:104433:104864 [4] NCCL INFO comm 0xa4a2c80 rank 4 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0xd9fb77963d823d25 - Init COMPLETE
+ip-26-0-166-36:104429:104861 [0] NCCL INFO comm 0xa6b1af0 rank 0 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0xd9fb77963d823d25 - Init COMPLETE
+ip-26-0-166-36:104431:104862 [2] NCCL INFO comm 0x9a579f0 rank 2 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0xd9fb77963d823d25 - Init COMPLETE
+ip-26-0-166-15:127294:127725 [4] NCCL INFO Connected all trees
+ip-26-0-166-15:127294:127725 [4] NCCL INFO NVLS comm 0x91f9c40 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-162-46:702881:703320 [1] NCCL INFO Connected all trees
+ip-26-0-162-46:702881:703320 [1] NCCL INFO NVLS comm 0x9652340 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-166-214:624943:625375 [0] NCCL INFO Connected NVLS tree
+ip-26-0-166-214:624943:625375 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-214:624943:625375 [0] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-166-214:624943:625375 [0] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-166-214:624946:625377 [3] NCCL INFO Connected NVLS tree
+ip-26-0-166-214:624946:625377 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-214:624946:625377 [3] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-166-214:624946:625377 [3] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-166-214:624947:625378 [4] NCCL INFO Connected NVLS tree
+ip-26-0-166-214:624947:625378 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-214:624947:625378 [4] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-166-214:624947:625378 [4] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-166-214:624945:625380 [2] NCCL INFO Connected NVLS tree
+ip-26-0-166-214:624948:625382 [5] NCCL INFO Connected NVLS tree
+ip-26-0-166-214:624945:625380 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-214:624948:625382 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-214:624945:625380 [2] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-166-214:624945:625380 [2] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-166-214:624948:625382 [5] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-166-214:624948:625382 [5] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-166-214:624949:625381 [6] NCCL INFO Connected NVLS tree
+ip-26-0-166-214:624949:625381 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-214:624949:625381 [6] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-166-214:624949:625381 [6] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-166-214:624944:625379 [1] NCCL INFO Connected NVLS tree
+ip-26-0-166-214:624944:625379 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-214:624944:625379 [1] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-166-214:624944:625379 [1] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-166-214:624950:625376 [7] NCCL INFO Connected NVLS tree
+ip-26-0-166-214:624950:625376 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-214:624950:625376 [7] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-166-214:624950:625376 [7] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-166-15:127291:127730 [1] NCCL INFO Connected all trees
+ip-26-0-166-15:127291:127730 [1] NCCL INFO NVLS comm 0xa46fbd0 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-166-214:624945:625380 [2] NCCL INFO comm 0x9a046d0 rank 2 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0xef8d36d72ea6a346 - Init COMPLETE
+ip-26-0-166-214:624946:625377 [3] NCCL INFO comm 0xa29d720 rank 3 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0xef8d36d72ea6a346 - Init COMPLETE
+ip-26-0-166-214:624947:625378 [4] NCCL INFO comm 0xa23cf00 rank 4 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0xef8d36d72ea6a346 - Init COMPLETE
+ip-26-0-166-214:624943:625375 [0] NCCL INFO comm 0x9698270 rank 0 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0xef8d36d72ea6a346 - Init COMPLETE
+ip-26-0-166-214:624950:625376 [7] NCCL INFO comm 0x8ff2880 rank 7 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0xef8d36d72ea6a346 - Init COMPLETE
+ip-26-0-166-214:624949:625381 [6] NCCL INFO comm 0x8ccfda0 rank 6 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0xef8d36d72ea6a346 - Init COMPLETE
+ip-26-0-166-214:624944:625379 [1] NCCL INFO comm 0x8768060 rank 1 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0xef8d36d72ea6a346 - Init COMPLETE
+ip-26-0-166-214:624948:625382 [5] NCCL INFO comm 0xa2b4710 rank 5 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0xef8d36d72ea6a346 - Init COMPLETE
+ip-26-0-166-15:127295:127726 [5] NCCL INFO Connected all trees
+ip-26-0-166-15:127295:127726 [5] NCCL INFO NVLS comm 0x8ce9f30 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-166-15:127292:127729 [2] NCCL INFO Connected all trees
+ip-26-0-166-15:127292:127729 [2] NCCL INFO NVLS comm 0x945a720 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-166-15:127296:127727 [6] NCCL INFO Connected all trees
+ip-26-0-166-15:127296:127727 [6] NCCL INFO NVLS comm 0x9900580 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-166-125:762820:763259 [5] NCCL INFO Connected NVLS tree
+ip-26-0-166-125:762820:763259 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-125:762820:763259 [5] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-166-125:762820:763259 [5] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-166-125:762818:763263 [3] NCCL INFO Connected NVLS tree
+ip-26-0-166-125:762818:763263 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-125:762818:763263 [3] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-166-125:762818:763263 [3] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-166-125:762819:763261 [4] NCCL INFO Connected NVLS tree
+ip-26-0-166-125:762819:763261 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-125:762819:763261 [4] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-166-125:762819:763261 [4] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-166-125:762822:763260 [7] NCCL INFO Connected NVLS tree
+ip-26-0-166-125:762822:763260 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-125:762822:763260 [7] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-166-125:762822:763260 [7] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-166-125:762816:763262 [1] NCCL INFO Connected NVLS tree
+ip-26-0-166-125:762816:763262 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-125:762816:763262 [1] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-166-125:762816:763262 [1] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-166-125:762817:763265 [2] NCCL INFO Connected NVLS tree
+ip-26-0-166-125:762817:763265 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-125:762821:763264 [6] NCCL INFO Connected NVLS tree
+ip-26-0-166-125:762817:763265 [2] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-166-125:762817:763265 [2] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-166-125:762821:763264 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-125:762821:763264 [6] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-166-125:762821:763264 [6] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-166-125:762815:763258 [0] NCCL INFO Connected NVLS tree
+ip-26-0-166-125:762815:763258 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-125:762815:763258 [0] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-166-125:762815:763258 [0] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-167-9:1039543:1039970 [6] NCCL INFO Connected NVLS tree
+ip-26-0-167-9:1039543:1039970 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-167-9:1039543:1039970 [6] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-167-9:1039543:1039970 [6] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-167-9:1039539:1039975 [2] NCCL INFO Connected NVLS tree
+ip-26-0-167-9:1039539:1039975 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-167-9:1039539:1039975 [2] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-167-9:1039539:1039975 [2] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-167-9:1039538:1039973 [1] NCCL INFO Connected NVLS tree
+ip-26-0-167-9:1039538:1039973 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-167-9:1039538:1039973 [1] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-167-9:1039538:1039973 [1] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-167-9:1039541:1039971 [4] NCCL INFO Connected NVLS tree
+ip-26-0-167-9:1039541:1039971 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-167-9:1039541:1039971 [4] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-167-9:1039541:1039971 [4] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-167-9:1039537:1039969 [0] NCCL INFO Connected NVLS tree
+ip-26-0-167-9:1039537:1039969 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-167-9:1039544:1039972 [7] NCCL INFO Connected NVLS tree
+ip-26-0-167-9:1039544:1039972 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-167-9:1039537:1039969 [0] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-167-9:1039537:1039969 [0] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-167-9:1039544:1039972 [7] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-167-9:1039544:1039972 [7] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-167-9:1039540:1039974 [3] NCCL INFO Connected NVLS tree
+ip-26-0-167-9:1039540:1039974 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-167-9:1039540:1039974 [3] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-167-9:1039540:1039974 [3] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-167-9:1039542:1039976 [5] NCCL INFO Connected NVLS tree
+ip-26-0-167-9:1039542:1039976 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-167-9:1039542:1039976 [5] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-167-9:1039542:1039976 [5] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-162-46:702885:703316 [5] NCCL INFO Connected NVLS tree
+ip-26-0-162-46:702885:703316 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-162-46:702885:703316 [5] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-162-46:702885:703316 [5] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-162-46:702887:703317 [7] NCCL INFO Connected NVLS tree
+ip-26-0-162-46:702887:703317 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-162-46:702887:703317 [7] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-162-46:702887:703317 [7] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-162-46:702881:703320 [1] NCCL INFO Connected NVLS tree
+ip-26-0-162-46:702881:703320 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-162-46:702881:703320 [1] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-162-46:702881:703320 [1] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-162-46:702884:703319 [4] NCCL INFO Connected NVLS tree
+ip-26-0-162-46:702884:703319 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-162-46:702884:703319 [4] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-162-46:702884:703319 [4] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-162-46:702883:703318 [3] NCCL INFO Connected NVLS tree
+ip-26-0-162-46:702883:703318 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-162-46:702883:703318 [3] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-162-46:702883:703318 [3] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-162-46:702882:703321 [2] NCCL INFO Connected NVLS tree
+ip-26-0-162-46:702882:703321 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-162-46:702882:703321 [2] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-162-46:702882:703321 [2] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-162-46:702886:703315 [6] NCCL INFO Connected NVLS tree
+ip-26-0-162-46:702880:703314 [0] NCCL INFO Connected NVLS tree
+ip-26-0-162-46:702886:703315 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-162-46:702880:703314 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-162-46:702886:703315 [6] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-162-46:702886:703315 [6] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-162-46:702880:703314 [0] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-162-46:702880:703314 [0] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-165-213:9631:10062 [6] NCCL INFO Connected all trees
+ip-26-0-165-213:9631:10062 [6] NCCL INFO NVLS comm 0x9d1a1c0 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-166-125:762818:763263 [3] NCCL INFO comm 0x89882d0 rank 3 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0xe106f35b05bcd308 - Init COMPLETE
+ip-26-0-166-125:762822:763260 [7] NCCL INFO comm 0xa233e80 rank 7 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0xe106f35b05bcd308 - Init COMPLETE
+ip-26-0-166-125:762820:763259 [5] NCCL INFO comm 0x95543d0 rank 5 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0xe106f35b05bcd308 - Init COMPLETE
+ip-26-0-166-125:762819:763261 [4] NCCL INFO comm 0x96329a0 rank 4 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0xe106f35b05bcd308 - Init COMPLETE
+ip-26-0-166-125:762817:763265 [2] NCCL INFO comm 0xa64bcb0 rank 2 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0xe106f35b05bcd308 - Init COMPLETE
+ip-26-0-166-125:762816:763262 [1] NCCL INFO comm 0x984eea0 rank 1 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0xe106f35b05bcd308 - Init COMPLETE
+ip-26-0-166-125:762815:763258 [0] NCCL INFO comm 0x881f7d0 rank 0 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0xe106f35b05bcd308 - Init COMPLETE
+ip-26-0-166-125:762821:763264 [6] NCCL INFO comm 0x8afb540 rank 6 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0xe106f35b05bcd308 - Init COMPLETE
+ip-26-0-167-9:1039539:1039975 [2] NCCL INFO comm 0x9358300 rank 2 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0xc6b7614cb294bd51 - Init COMPLETE
+ip-26-0-167-9:1039541:1039971 [4] NCCL INFO comm 0xa6d9b30 rank 4 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0xc6b7614cb294bd51 - Init COMPLETE
+ip-26-0-167-9:1039537:1039969 [0] NCCL INFO comm 0x9e653d0 rank 0 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0xc6b7614cb294bd51 - Init COMPLETE
+ip-26-0-167-9:1039543:1039970 [6] NCCL INFO comm 0xa1956c0 rank 6 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0xc6b7614cb294bd51 - Init COMPLETE
+ip-26-0-167-9:1039544:1039972 [7] NCCL INFO comm 0x98df560 rank 7 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0xc6b7614cb294bd51 - Init COMPLETE
+ip-26-0-167-9:1039538:1039973 [1] NCCL INFO comm 0x9283370 rank 1 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0xc6b7614cb294bd51 - Init COMPLETE
+ip-26-0-167-9:1039540:1039974 [3] NCCL INFO comm 0x99d3460 rank 3 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0xc6b7614cb294bd51 - Init COMPLETE
+ip-26-0-167-9:1039542:1039976 [5] NCCL INFO comm 0x9095cf0 rank 5 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0xc6b7614cb294bd51 - Init COMPLETE
+ip-26-0-162-46:702883:703318 [3] NCCL INFO comm 0x87ca1f0 rank 3 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0x65afe7f478768c8 - Init COMPLETE
+ip-26-0-162-46:702881:703320 [1] NCCL INFO comm 0x9652340 rank 1 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0x65afe7f478768c8 - Init COMPLETE
+ip-26-0-162-46:702882:703321 [2] NCCL INFO comm 0x9a08fa0 rank 2 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0x65afe7f478768c8 - Init COMPLETE
+ip-26-0-162-46:702885:703316 [5] NCCL INFO comm 0x8dac5b0 rank 5 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0x65afe7f478768c8 - Init COMPLETE
+ip-26-0-162-46:702887:703317 [7] NCCL INFO comm 0x8d9c450 rank 7 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0x65afe7f478768c8 - Init COMPLETE
+ip-26-0-162-46:702884:703319 [4] NCCL INFO comm 0x9156360 rank 4 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0x65afe7f478768c8 - Init COMPLETE
+ip-26-0-162-46:702880:703314 [0] NCCL INFO comm 0x9b86160 rank 0 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0x65afe7f478768c8 - Init COMPLETE
+ip-26-0-162-46:702886:703315 [6] NCCL INFO comm 0x9292e40 rank 6 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0x65afe7f478768c8 - Init COMPLETE
+ip-26-0-166-15:127291:127730 [1] NCCL INFO Connected NVLS tree
+ip-26-0-166-15:127291:127730 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-15:127291:127730 [1] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-166-15:127291:127730 [1] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-166-15:127293:127728 [3] NCCL INFO Connected NVLS tree
+ip-26-0-166-15:127293:127728 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-15:127293:127728 [3] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-166-15:127293:127728 [3] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-166-15:127295:127726 [5] NCCL INFO Connected NVLS tree
+ip-26-0-166-15:127295:127726 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-15:127294:127725 [4] NCCL INFO Connected NVLS tree
+ip-26-0-166-15:127294:127725 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-15:127295:127726 [5] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-166-15:127295:127726 [5] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-166-15:127294:127725 [4] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-166-15:127294:127725 [4] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-166-15:127297:127731 [7] NCCL INFO Connected NVLS tree
+ip-26-0-166-15:127297:127731 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-15:127297:127731 [7] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-166-15:127297:127731 [7] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-166-15:127296:127727 [6] NCCL INFO Connected NVLS tree
+ip-26-0-166-15:127296:127727 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-15:127296:127727 [6] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-166-15:127296:127727 [6] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-165-213:9630:10060 [5] NCCL INFO Connected all trees
+ip-26-0-165-213:9630:10060 [5] NCCL INFO NVLS comm 0x89e8c80 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-166-244:897246:897680 [6] NCCL INFO Connected all trees
+ip-26-0-166-244:897246:897680 [6] NCCL INFO NVLS comm 0xa67d830 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-213:9629:10063 [4] NCCL INFO Connected all trees
+ip-26-0-165-213:9629:10063 [4] NCCL INFO NVLS comm 0x8b69b00 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-166-15:127290:127724 [0] NCCL INFO Connected NVLS tree
+ip-26-0-166-15:127290:127724 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-15:127290:127724 [0] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-166-15:127290:127724 [0] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-166-15:127292:127729 [2] NCCL INFO Connected NVLS tree
+ip-26-0-166-15:127292:127729 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-15:127292:127729 [2] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-166-15:127292:127729 [2] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-166-15:127295:127726 [5] NCCL INFO comm 0x8ce9f30 rank 5 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0xa88f4e7d7abb936c - Init COMPLETE
+ip-26-0-166-15:127293:127728 [3] NCCL INFO comm 0xa35de50 rank 3 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0xa88f4e7d7abb936c - Init COMPLETE
+ip-26-0-166-15:127291:127730 [1] NCCL INFO comm 0xa46fbd0 rank 1 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0xa88f4e7d7abb936c - Init COMPLETE
+ip-26-0-166-15:127297:127731 [7] NCCL INFO comm 0xa0163c0 rank 7 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0xa88f4e7d7abb936c - Init COMPLETE
+ip-26-0-166-15:127294:127725 [4] NCCL INFO comm 0x91f9c40 rank 4 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0xa88f4e7d7abb936c - Init COMPLETE
+ip-26-0-166-15:127296:127727 [6] NCCL INFO comm 0x9900580 rank 6 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0xa88f4e7d7abb936c - Init COMPLETE
+ip-26-0-166-15:127290:127724 [0] NCCL INFO comm 0x8befd90 rank 0 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0xa88f4e7d7abb936c - Init COMPLETE
+ip-26-0-166-15:127292:127729 [2] NCCL INFO comm 0x945a720 rank 2 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0xa88f4e7d7abb936c - Init COMPLETE
+ip-26-0-166-244:897242:897679 [2] NCCL INFO Connected all trees
+ip-26-0-166-244:897242:897679 [2] NCCL INFO NVLS comm 0x895b4a0 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-213:9626:10061 [1] NCCL INFO Connected all trees
+ip-26-0-165-213:9626:10061 [1] NCCL INFO NVLS comm 0x960b920 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-213:9627:10064 [2] NCCL INFO Connected all trees
+ip-26-0-165-213:9627:10064 [2] NCCL INFO NVLS comm 0x8b6ebb0 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-213:9628:10066 [3] NCCL INFO Connected all trees
+ip-26-0-165-213:9628:10066 [3] NCCL INFO NVLS comm 0x9557b40 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-166-244:897245:897676 [5] NCCL INFO Connected all trees
+ip-26-0-166-244:897245:897676 [5] NCCL INFO NVLS comm 0x9d0d3b0 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-166-244:897241:897675 [1] NCCL INFO Connected all trees
+ip-26-0-166-244:897241:897675 [1] NCCL INFO NVLS comm 0xa412870 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-166-244:897244:897674 [4] NCCL INFO Connected all trees
+ip-26-0-166-244:897244:897674 [4] NCCL INFO NVLS comm 0x90f0e70 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-166-244:897243:897678 [3] NCCL INFO Connected all trees
+ip-26-0-166-244:897243:897678 [3] NCCL INFO NVLS comm 0x8c64460 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-165-213:9631:10062 [6] NCCL INFO Connected NVLS tree
+ip-26-0-165-213:9631:10062 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-213:9628:10066 [3] NCCL INFO Connected NVLS tree
+ip-26-0-165-213:9631:10062 [6] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-165-213:9631:10062 [6] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-165-213:9628:10066 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-213:9628:10066 [3] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-165-213:9628:10066 [3] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-165-213:9629:10063 [4] NCCL INFO Connected NVLS tree
+ip-26-0-165-213:9626:10061 [1] NCCL INFO Connected NVLS tree
+ip-26-0-165-213:9629:10063 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-213:9626:10061 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-213:9629:10063 [4] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-165-213:9626:10061 [1] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-165-213:9629:10063 [4] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-165-213:9626:10061 [1] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-165-213:9625:10059 [0] NCCL INFO Connected NVLS tree
+ip-26-0-165-213:9625:10059 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-213:9627:10064 [2] NCCL INFO Connected NVLS tree
+ip-26-0-165-213:9625:10059 [0] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-165-213:9625:10059 [0] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-165-213:9627:10064 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-213:9627:10064 [2] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-165-213:9627:10064 [2] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-165-213:9630:10060 [5] NCCL INFO Connected NVLS tree
+ip-26-0-165-213:9630:10060 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-213:9630:10060 [5] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-165-213:9630:10060 [5] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-165-213:9632:10065 [7] NCCL INFO Connected NVLS tree
+ip-26-0-165-213:9632:10065 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-213:9632:10065 [7] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-165-213:9632:10065 [7] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-166-244:897244:897674 [4] NCCL INFO Connected NVLS tree
+ip-26-0-166-244:897244:897674 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-244:897244:897674 [4] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-166-244:897244:897674 [4] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-166-244:897242:897679 [2] NCCL INFO Connected NVLS tree
+ip-26-0-166-244:897242:897679 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-244:897242:897679 [2] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-166-244:897242:897679 [2] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-166-244:897246:897680 [6] NCCL INFO Connected NVLS tree
+ip-26-0-166-244:897246:897680 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-244:897246:897680 [6] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-166-244:897246:897680 [6] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-166-244:897245:897676 [5] NCCL INFO Connected NVLS tree
+ip-26-0-166-244:897245:897676 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-244:897245:897676 [5] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-166-244:897245:897676 [5] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-166-244:897247:897677 [7] NCCL INFO Connected NVLS tree
+ip-26-0-166-244:897247:897677 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-244:897247:897677 [7] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-166-244:897247:897677 [7] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-166-244:897241:897675 [1] NCCL INFO Connected NVLS tree
+ip-26-0-166-244:897241:897675 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-244:897241:897675 [1] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-166-244:897241:897675 [1] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-166-244:897240:897673 [0] NCCL INFO Connected NVLS tree
+ip-26-0-166-244:897240:897673 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-244:897240:897673 [0] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-166-244:897240:897673 [0] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-166-244:897243:897678 [3] NCCL INFO Connected NVLS tree
+ip-26-0-166-244:897243:897678 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-244:897243:897678 [3] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-166-244:897243:897678 [3] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-165-213:9629:10063 [4] NCCL INFO comm 0x8b69b00 rank 4 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0x98d233044459c3d7 - Init COMPLETE
+ip-26-0-165-213:9631:10062 [6] NCCL INFO comm 0x9d1a1c0 rank 6 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0x98d233044459c3d7 - Init COMPLETE
+ip-26-0-165-213:9630:10060 [5] NCCL INFO comm 0x89e8c80 rank 5 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0x98d233044459c3d7 - Init COMPLETE
+ip-26-0-165-213:9625:10059 [0] NCCL INFO comm 0x9033500 rank 0 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0x98d233044459c3d7 - Init COMPLETE
+ip-26-0-165-213:9627:10064 [2] NCCL INFO comm 0x8b6ebb0 rank 2 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0x98d233044459c3d7 - Init COMPLETE
+ip-26-0-165-213:9626:10061 [1] NCCL INFO comm 0x960b920 rank 1 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0x98d233044459c3d7 - Init COMPLETE
+ip-26-0-165-213:9628:10066 [3] NCCL INFO comm 0x9557b40 rank 3 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0x98d233044459c3d7 - Init COMPLETE
+ip-26-0-165-213:9632:10065 [7] NCCL INFO comm 0x8f449d0 rank 7 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0x98d233044459c3d7 - Init COMPLETE
+ip-26-0-166-244:897242:897679 [2] NCCL INFO comm 0x895b4a0 rank 2 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0xc427040fef04b119 - Init COMPLETE
+ip-26-0-166-244:897246:897680 [6] NCCL INFO comm 0xa67d830 rank 6 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0xc427040fef04b119 - Init COMPLETE
+ip-26-0-166-244:897240:897673 [0] NCCL INFO comm 0x956e780 rank 0 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0xc427040fef04b119 - Init COMPLETE
+ip-26-0-166-244:897244:897674 [4] NCCL INFO comm 0x90f0e70 rank 4 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0xc427040fef04b119 - Init COMPLETE
+ip-26-0-166-244:897245:897676 [5] NCCL INFO comm 0x9d0d3b0 rank 5 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0xc427040fef04b119 - Init COMPLETE
+ip-26-0-166-244:897241:897675 [1] NCCL INFO comm 0xa412870 rank 1 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0xc427040fef04b119 - Init COMPLETE
+ip-26-0-166-244:897247:897677 [7] NCCL INFO comm 0x9c01630 rank 7 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0xc427040fef04b119 - Init COMPLETE
+ip-26-0-166-244:897243:897678 [3] NCCL INFO comm 0x8c64460 rank 3 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0xc427040fef04b119 - Init COMPLETE
+01/07/2025 03:49:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]: Config:
+01/07/2025 03:49:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]: Config(general=GeneralArgs(project='debug',
+01/07/2025 03:49:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]:                            run='1.34G_dp4_tp2_pp8_acc32_mbs2_seq4096_zero1_tpmodeRED_vocab131k',
+01/07/2025 03:49:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]:                            seed=42,
+01/07/2025 03:49:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]:                            step=None,
+01/07/2025 03:49:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]:                            consumed_train_samples=None,
+01/07/2025 03:49:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]:                            benchmark_csv_path=PosixPath('benchmark/results/bench_final2.csv'),
+01/07/2025 03:49:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]:                            ignore_sanity_checks=True),
+01/07/2025 03:49:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]:        parallelism=ParallelismArgs(dp=4,
+01/07/2025 03:49:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]:                                    pp=8,
+01/07/2025 03:49:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]:                                    tp=2,
+01/07/2025 03:49:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]:                                    pp_engine=<nanotron.parallel.pipeline_parallel.engine.OneForwardOneBackwardPipelineEngine object at 0x7fc68fab8c40>,
+01/07/2025 03:49:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]:                                    tp_mode=<TensorParallelLinearMode.REDUCE_SCATTER: 2>,
+01/07/2025 03:49:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]:                                    tp_linear_async_communication=True,
+01/07/2025 03:49:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]:                                    recompute_layer=False,
+01/07/2025 03:49:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]:                                    tp_recompute_allgather=True,
+01/07/2025 03:49:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]:                                    expert_parallel_size=1),
+01/07/2025 03:49:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]:        model=ModelArgs(model_config=LlamaConfig(bos_token_id=0,
+01/07/2025 03:49:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]:                                                 eos_token_id=0,
+01/07/2025 03:49:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]:                                                 hidden_act='silu',
+01/07/2025 03:49:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]:                                                 hidden_size=2048,
+01/07/2025 03:49:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]:                                                 initializer_range=0.02,
+01/07/2025 03:49:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]:                                                 intermediate_size=8192,
+01/07/2025 03:49:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]:                                                 is_llama_config=True,
+01/07/2025 03:49:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]:                                                 max_position_embeddings=4096,
+01/07/2025 03:49:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]:                                                 num_attention_heads=32,
+01/07/2025 03:49:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]:                                                 num_hidden_layers=16,
+01/07/2025 03:49:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]:                                                 num_key_value_heads=32,
+01/07/2025 03:49:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]:                                                 pad_token_id=None,
+01/07/2025 03:49:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]:                                                 pretraining_tp=1,
+01/07/2025 03:49:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]:                                                 rms_norm_eps=1e-05,
+01/07/2025 03:49:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]:                                                 rope_scaling=None,
+01/07/2025 03:49:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]:                                                 rope_theta=10000.0,
+01/07/2025 03:49:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]:                                                 rope_interleaved=False,
+01/07/2025 03:49:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]:                                                 tie_word_embeddings=True,
+01/07/2025 03:49:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]:                                                 use_cache=True,
+01/07/2025 03:49:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]:                                                 vocab_size=131072),
+01/07/2025 03:49:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]:                        init_method=RandomInit(std=0.02),
+01/07/2025 03:49:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]:                        dtype=torch.bfloat16,
+01/07/2025 03:49:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]:                        make_vocab_size_divisible_by=1,
+01/07/2025 03:49:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]:                        ddp_bucket_cap_mb=25),
+01/07/2025 03:49:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]:        tokenizer=TokenizerArgs(tokenizer_name_or_path='robot-test/dummy-tokenizer-wordlevel',
+01/07/2025 03:49:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]:                                tokenizer_revision=None,
+01/07/2025 03:49:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]:                                tokenizer_max_length=None),
+01/07/2025 03:49:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]:        checkpoints=CheckpointsArgs(checkpoints_path=PosixPath('checkpoints'),
+01/07/2025 03:49:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]:                                    checkpoint_interval=10000,
+01/07/2025 03:49:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]:                                    save_initial_state=False,
+01/07/2025 03:49:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]:                                    save_final_state=False,
+01/07/2025 03:49:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]:                                    resume_checkpoint_path=None,
+01/07/2025 03:49:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]:                                    checkpoints_path_is_shared_file_system=False),
+01/07/2025 03:49:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]:        logging=LoggingArgs(log_level='info',
+01/07/2025 03:49:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]:                            log_level_replica='info',
+01/07/2025 03:49:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]:                            iteration_step_info_interval=1),
+01/07/2025 03:49:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]:        tokens=TokensArgs(sequence_length=4096,
+01/07/2025 03:49:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]:                          train_steps=100,
+01/07/2025 03:49:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]:                          micro_batch_size=2,
+01/07/2025 03:49:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]:                          batch_accumulation_per_replica=32,
+01/07/2025 03:49:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]:                          val_check_interval=100,
+01/07/2025 03:49:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]:                          limit_val_batches=0,
+01/07/2025 03:49:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]:                          limit_test_batches=0),
+01/07/2025 03:49:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]:        optimizer=OptimizerArgs(optimizer_factory=AdamWOptimizerArgs(adam_eps=1e-08,
+01/07/2025 03:49:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]:                                                                     adam_beta1=0.9,
+01/07/2025 03:49:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]:                                                                     adam_beta2=0.95,
+01/07/2025 03:49:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]:                                                                     torch_adam_is_fused=True,
+01/07/2025 03:49:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]:                                                                     name='adamW'),
+01/07/2025 03:49:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]:                                zero_stage=1,
+01/07/2025 03:49:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]:                                weight_decay=0.01,
+01/07/2025 03:49:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]:                                clip_grad=1.0,
+01/07/2025 03:49:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]:                                accumulate_grad_in_fp32=True,
+01/07/2025 03:49:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]:                                learning_rate_scheduler=LRSchedulerArgs(learning_rate=0.0003,
+01/07/2025 03:49:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]:                                                                        lr_warmup_steps=2,
+01/07/2025 03:49:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]:                                                                        lr_warmup_style='linear',
+01/07/2025 03:49:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]:                                                                        lr_decay_style='cosine',
+01/07/2025 03:49:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]:                                                                        lr_decay_steps=13,
+01/07/2025 03:49:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]:                                                                        lr_decay_starting_step=None,
+01/07/2025 03:49:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]:                                                                        min_decay_lr=1e-05)),
+01/07/2025 03:49:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]:        data_stages=[DatasetStageArgs(name='Stable Training Stage',
+01/07/2025 03:49:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]:                                      start_training_step=1,
+01/07/2025 03:49:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]:                                      data=DataArgs(dataset=None,
+01/07/2025 03:49:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]:                                                    seed=42,
+01/07/2025 03:49:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]:                                                    num_loading_workers=1))],
+01/07/2025 03:49:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]:        profiler=None,
+01/07/2025 03:49:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]:        lighteval=None,
+01/07/2025 03:49:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]:        s3_upload=None)
+01/07/2025 03:49:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]: Model Config:
+01/07/2025 03:49:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]: LlamaConfig(bos_token_id=0,
+01/07/2025 03:49:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]:             eos_token_id=0,
+01/07/2025 03:49:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]:             hidden_act='silu',
+01/07/2025 03:49:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]:             hidden_size=2048,
+01/07/2025 03:49:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]:             initializer_range=0.02,
+01/07/2025 03:49:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]:             intermediate_size=8192,
+01/07/2025 03:49:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]:             is_llama_config=True,
+01/07/2025 03:49:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]:             max_position_embeddings=4096,
+01/07/2025 03:49:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]:             num_attention_heads=32,
+01/07/2025 03:49:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]:             num_hidden_layers=16,
+01/07/2025 03:49:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]:             num_key_value_heads=32,
+01/07/2025 03:49:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]:             pad_token_id=None,
+01/07/2025 03:49:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]:             pretraining_tp=1,
+01/07/2025 03:49:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]:             rms_norm_eps=1e-05,
+01/07/2025 03:49:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]:             rope_scaling=None,
+01/07/2025 03:49:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]:             rope_theta=10000.0,
+01/07/2025 03:49:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]:             rope_interleaved=False,
+01/07/2025 03:49:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]:             tie_word_embeddings=True,
+01/07/2025 03:49:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]:             use_cache=True,
+01/07/2025 03:49:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]:             vocab_size=131072)
+01/07/2025 03:49:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]: Building model..
+01/07/2025 03:49:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]: Initialize RoPE Theta = 10000.0
+01/07/2025 03:49:55 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]: Setting PP block ranks...
+01/07/2025 03:49:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]: module_name: model.token_position_embeddings | PP: 0/8 | Block rank: 0
+01/07/2025 03:49:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]: module_name: model.decoder.0 | PP: 0/8 | Block rank: 1
+01/07/2025 03:49:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]: module_name: model.decoder.1 | PP: 0/8 | Block rank: 2
+01/07/2025 03:49:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]: module_name: model.decoder.2 | PP: 0/8 | Block rank: 3
+01/07/2025 03:49:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]: module_name: model.decoder.3 | PP: 1/8 | Block rank: 0
+01/07/2025 03:49:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]: module_name: model.decoder.4 | PP: 1/8 | Block rank: 1
+01/07/2025 03:49:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]: module_name: model.decoder.5 | PP: 1/8 | Block rank: 2
+01/07/2025 03:49:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]: module_name: model.decoder.6 | PP: 2/8 | Block rank: 0
+01/07/2025 03:49:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]: module_name: model.decoder.7 | PP: 2/8 | Block rank: 1
+01/07/2025 03:49:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]: module_name: model.decoder.8 | PP: 2/8 | Block rank: 2
+01/07/2025 03:49:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]: module_name: model.decoder.9 | PP: 3/8 | Block rank: 0
+01/07/2025 03:49:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]: module_name: model.decoder.10 | PP: 3/8 | Block rank: 1
+01/07/2025 03:49:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]: module_name: model.decoder.11 | PP: 3/8 | Block rank: 2
+01/07/2025 03:49:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]: module_name: model.decoder.12 | PP: 4/8 | Block rank: 0
+01/07/2025 03:49:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]: module_name: model.decoder.13 | PP: 4/8 | Block rank: 1
+01/07/2025 03:49:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]: module_name: model.decoder.14 | PP: 5/8 | Block rank: 0
+01/07/2025 03:49:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]: module_name: model.decoder.15 | PP: 6/8 | Block rank: 0
+01/07/2025 03:49:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]: module_name: model.final_layer_norm | PP: 7/8 | Block rank: 0
+01/07/2025 03:49:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]: module_name: model.lm_head | PP: 7/8 | Block rank: 1
+01/07/2025 03:49:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]: module_name: model.cast_to_fp32 | PP: 7/8 | Block rank: 2
+01/07/2025 03:49:56 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]: module_name: loss | PP: 7/8 | Block rank: 3
+NCCL version 2.18.6+cuda12.1
+NCCL version 2.18.6+cuda12.1
+NCCL version 2.18.6+cuda12.1
+NCCL version 2.18.6+cuda12.1
+NCCL version 2.18.6+cuda12.1
+NCCL version 2.18.6+cuda12.1
+NCCL version 2.18.6+cuda12.1
+NCCL version 2.18.6+cuda12.1
+NCCL version 2.18.6+cuda12.1
+NCCL version 2.18.6+cuda12.1
+NCCL version 2.18.6+cuda12.1
+NCCL version 2.18.6+cuda12.1
+NCCL version 2.18.6+cuda12.1
+NCCL version 2.18.6+cuda12.1
+NCCL version 2.18.6+cuda12.1
+NCCL version 2.18.6+cuda12.1
+NCCL version 2.18.6+cuda12.1
+ip-26-0-166-244:897244:897727 [4] NCCL INFO Using network Libfabric
+NCCL version 2.18.6+cuda12.1
+ip-26-0-166-244:897242:897730 [2] NCCL INFO Using network Libfabric
+ip-26-0-166-244:897245:897729 [5] NCCL INFO Using network Libfabric
+ip-26-0-166-244:897243:897731 [3] NCCL INFO Using network Libfabric
+ip-26-0-166-214:624949:625431 [6] NCCL INFO Using network Libfabric
+ip-26-0-166-244:897244:897727 [4] NCCL INFO comm 0xade7470 rank 0 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0x4a1841ee5821a4e0 - Init START
+ip-26-0-166-244:897245:897729 [5] NCCL INFO comm 0xafad090 rank 1 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0x4a1841ee5821a4e0 - Init START
+ip-26-0-166-15:127296:127778 [6] NCCL INFO Using network Libfabric
+ip-26-0-166-214:624947:625432 [4] NCCL INFO Using network Libfabric
+ip-26-0-166-244:897245:897729 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-244:897244:897727 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-165-213:9625:10112 [0] NCCL INFO Using network Libfabric
+ip-26-0-166-244:897247:897736 [7] NCCL INFO Using network Libfabric
+ip-26-0-166-244:897246:897734 [6] NCCL INFO Using network Libfabric
+ip-26-0-166-36:104433:104915 [4] NCCL INFO Using network Libfabric
+ip-26-0-166-214:624950:625434 [7] NCCL INFO Using network Libfabric
+ip-26-0-166-244:897240:897735 [0] NCCL INFO Using network Libfabric
+ip-26-0-166-15:127294:127779 [4] NCCL INFO Using network Libfabric
+ip-26-0-166-36:104431:104917 [2] NCCL INFO Using network Libfabric
+ip-26-0-166-125:762819:763316 [4] NCCL INFO Using network Libfabric
+ip-26-0-166-214:624948:625435 [5] NCCL INFO Using network Libfabric
+ip-26-0-166-244:897243:897731 [3] NCCL INFO comm 0x9f01bc0 rank 1 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0x8e1ccf03aca53ed8 - Init START
+ip-26-0-166-244:897242:897730 [2] NCCL INFO comm 0xa64f180 rank 0 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0x8e1ccf03aca53ed8 - Init START
+ip-26-0-166-15:127297:127781 [7] NCCL INFO Using network Libfabric
+ip-26-0-166-125:762817:763317 [2] NCCL INFO Using network Libfabric
+ip-26-0-166-125:762820:763322 [5] NCCL INFO Using network Libfabric
+ip-26-0-165-213:9629:10113 [4] NCCL INFO Using network Libfabric
+ip-26-0-166-214:624945:625436 [2] NCCL INFO Using network Libfabric
+ip-26-0-166-36:104435:104920 [6] NCCL INFO Using network Libfabric
+ip-26-0-166-214:624943:625438 [0] NCCL INFO Using network Libfabric
+ip-26-0-166-244:897243:897731 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-244:897242:897730 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-125:762821:763318 [6] NCCL INFO Using network Libfabric
+ip-26-0-166-36:104429:104919 [0] NCCL INFO Using network Libfabric
+ip-26-0-165-213:9626:10118 [1] NCCL INFO Using network Libfabric
+ip-26-0-166-125:762815:763320 [0] NCCL INFO Using network Libfabric
+ip-26-0-166-125:762822:763323 [7] NCCL INFO Using network Libfabric
+ip-26-0-165-213:9627:10116 [2] NCCL INFO Using network Libfabric
+ip-26-0-165-213:9631:10117 [6] NCCL INFO Using network Libfabric
+ip-26-0-166-15:127295:127783 [5] NCCL INFO Using network Libfabric
+ip-26-0-166-15:127292:127782 [2] NCCL INFO Using network Libfabric
+ip-26-0-166-36:104434:104921 [5] NCCL INFO Using network Libfabric
+ip-26-0-166-36:104432:104922 [3] NCCL INFO Using network Libfabric
+ip-26-0-166-125:762818:763321 [3] NCCL INFO Using network Libfabric
+ip-26-0-166-214:624944:625440 [1] NCCL INFO Using network Libfabric
+ip-26-0-166-214:624946:625439 [3] NCCL INFO Using network Libfabric
+ip-26-0-166-244:897246:897734 [6] NCCL INFO comm 0xc371140 rank 0 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0xf091798745a84edd - Init START
+ip-26-0-166-244:897247:897736 [7] NCCL INFO comm 0xb8f3eb0 rank 1 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0xf091798745a84edd - Init START
+ip-26-0-165-213:9630:10119 [5] NCCL INFO Using network Libfabric
+ip-26-0-166-214:624948:625435 [5] NCCL INFO comm 0xbfa5b60 rank 1 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0x68fdf1bd57245b59 - Init START
+ip-26-0-166-214:624947:625432 [4] NCCL INFO comm 0xb4dd0e0 rank 0 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0x68fdf1bd57245b59 - Init START
+ip-26-0-166-244:897247:897736 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-244:897246:897734 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-15:127291:127787 [1] NCCL INFO Using network Libfabric
+ip-26-0-166-125:762816:763324 [1] NCCL INFO Using network Libfabric
+ip-26-0-166-214:624949:625431 [6] NCCL INFO comm 0xa9c4a30 rank 0 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0xa0ece3c426d66ea9 - Init START
+ip-26-0-166-214:624950:625434 [7] NCCL INFO comm 0xa28ff10 rank 1 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0xa0ece3c426d66ea9 - Init START
+ip-26-0-166-15:127297:127781 [7] NCCL INFO comm 0xbd0c380 rank 1 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0xaa23d81ecb1fca47 - Init START
+ip-26-0-166-15:127296:127778 [6] NCCL INFO comm 0xb5f4350 rank 0 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0xaa23d81ecb1fca47 - Init START
+ip-26-0-166-125:762819:763316 [4] NCCL INFO comm 0xb32ed70 rank 0 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0x1ea69f320f0c17f - Init START
+ip-26-0-166-125:762820:763322 [5] NCCL INFO comm 0xa7f6080 rank 1 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0x1ea69f320f0c17f - Init START
+ip-26-0-166-214:624947:625432 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-214:624948:625435 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-36:104430:104924 [1] NCCL INFO Using network Libfabric
+ip-26-0-166-15:127293:127786 [3] NCCL INFO Using network Libfabric
+ip-26-0-166-125:762819:763316 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-125:762820:763322 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-214:624950:625434 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-214:624949:625431 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-15:127297:127781 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-15:127296:127778 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-165-213:9625:10112 [0] NCCL INFO comm 0xad2f5c0 rank 0 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0xef34ee973e904713 - Init START
+ip-26-0-165-213:9626:10118 [1] NCCL INFO comm 0xa8af270 rank 1 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0xef34ee973e904713 - Init START
+ip-26-0-166-15:127295:127783 [5] NCCL INFO comm 0x9f8be80 rank 1 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0x41ee8bd379d4b112 - Init START
+ip-26-0-166-15:127294:127779 [4] NCCL INFO comm 0xa49af50 rank 0 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0x41ee8bd379d4b112 - Init START
+ip-26-0-166-125:762821:763318 [6] NCCL INFO comm 0x9d9cf30 rank 0 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0xac2405b00a4fa8a0 - Init START
+ip-26-0-166-125:762822:763323 [7] NCCL INFO comm 0xb4d8070 rank 1 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0xac2405b00a4fa8a0 - Init START
+ip-26-0-166-36:104432:104922 [3] NCCL INFO comm 0x9cee960 rank 1 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0x58fc109f3722ddb0 - Init START
+ip-26-0-166-36:104431:104917 [2] NCCL INFO comm 0xacf6020 rank 0 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0x58fc109f3722ddb0 - Init START
+ip-26-0-166-244:897241:897737 [1] NCCL INFO Using network Libfabric
+ip-26-0-166-15:127290:127785 [0] NCCL INFO Using network Libfabric
+ip-26-0-165-213:9625:10112 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-165-213:9626:10118 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-214:624944:625440 [1] NCCL INFO comm 0xa45b7c0 rank 1 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0x4a9d894e5cf2f81c - Init START
+ip-26-0-166-214:624943:625438 [0] NCCL INFO comm 0xb38d4c0 rank 0 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0x4a9d894e5cf2f81c - Init START
+ip-26-0-166-36:104434:104921 [5] NCCL INFO comm 0x9f5a360 rank 1 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0xbd2039d4027c0ed6 - Init START
+ip-26-0-166-36:104433:104915 [4] NCCL INFO comm 0xb73d350 rank 0 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0xbd2039d4027c0ed6 - Init START
+ip-26-0-166-15:127294:127779 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-15:127295:127783 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-125:762817:763317 [2] NCCL INFO comm 0xc343aa0 rank 0 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0xed427556a7553275 - Init START
+ip-26-0-166-125:762818:763321 [3] NCCL INFO comm 0x9c2c270 rank 1 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0xed427556a7553275 - Init START
+ip-26-0-165-213:9630:10119 [5] NCCL INFO comm 0x9c8b120 rank 1 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0xa0ff12f927ac5392 - Init START
+ip-26-0-165-213:9629:10113 [4] NCCL INFO comm 0x9e0c430 rank 0 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0xa0ff12f927ac5392 - Init START
+ip-26-0-166-36:104432:104922 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-214:624945:625436 [2] NCCL INFO comm 0xaca50f0 rank 0 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0x73d5dfca7b5791eb - Init START
+ip-26-0-166-214:624946:625439 [3] NCCL INFO comm 0xb53cb40 rank 1 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0x73d5dfca7b5791eb - Init START
+ip-26-0-166-36:104431:104917 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-125:762822:763323 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-125:762821:763318 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-165-213:9632:10121 [7] NCCL INFO Using network Libfabric
+ip-26-0-165-213:9628:10120 [3] NCCL INFO Using network Libfabric
+ip-26-0-166-214:624943:625438 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-214:624944:625440 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-214:624946:625439 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-214:624945:625436 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-36:104434:104921 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-36:104433:104915 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-125:762817:763317 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-125:762818:763321 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-165-213:9630:10119 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-165-213:9629:10113 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-125:762816:763324 [1] NCCL INFO comm 0xb545ab0 rank 1 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0x5be10b8c3d993573 - Init START
+ip-26-0-166-125:762815:763320 [0] NCCL INFO comm 0x9ac09f0 rank 0 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0x5be10b8c3d993573 - Init START
+ip-26-0-166-36:104430:104924 [1] NCCL INFO comm 0xbbba9f0 rank 1 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0xfe3a2bff210edb10 - Init START
+ip-26-0-166-36:104429:104919 [0] NCCL INFO comm 0xb951280 rank 0 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0xfe3a2bff210edb10 - Init START
+ip-26-0-166-15:127293:127786 [3] NCCL INFO comm 0xb5fc0c0 rank 1 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0xe182481c705f52fd - Init START
+ip-26-0-166-15:127292:127782 [2] NCCL INFO comm 0xb14f780 rank 0 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0xe182481c705f52fd - Init START
+ip-26-0-166-125:762816:763324 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-125:762815:763320 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-36:104429:104919 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-36:104430:104924 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-15:127292:127782 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-15:127293:127786 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-15:127290:127785 [0] NCCL INFO comm 0xa8e93b0 rank 0 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0x13ca242df769a19e - Init START
+ip-26-0-166-15:127291:127787 [1] NCCL INFO comm 0xc163200 rank 1 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0x13ca242df769a19e - Init START
+ip-26-0-166-36:104436:104923 [7] NCCL INFO Using network Libfabric
+ip-26-0-166-244:897240:897735 [0] NCCL INFO comm 0xa80e050 rank 0 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0xb3cd48026c577012 - Init START
+ip-26-0-166-244:897241:897737 [1] NCCL INFO comm 0xc105c70 rank 1 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0xb3cd48026c577012 - Init START
+ip-26-0-166-15:127291:127787 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-15:127290:127785 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-165-213:9627:10116 [2] NCCL INFO comm 0x9e12000 rank 0 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0x1e66d5e015dcfb4 - Init START
+ip-26-0-165-213:9628:10120 [3] NCCL INFO comm 0xa7fc870 rank 1 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0x1e66d5e015dcfb4 - Init START
+ip-26-0-166-244:897241:897737 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-244:897240:897735 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-165-213:9632:10121 [7] NCCL INFO comm 0xa1e4a30 rank 1 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0x91023a3acfb2d892 - Init START
+ip-26-0-165-213:9631:10117 [6] NCCL INFO comm 0xafbbfa0 rank 0 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0x91023a3acfb2d892 - Init START
+ip-26-0-165-213:9627:10116 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-165-213:9628:10120 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-165-213:9632:10121 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-165-213:9631:10117 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-36:104436:104923 [7] NCCL INFO comm 0xac23ec0 rank 1 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0x3e7d69e4ffee2364 - Init START
+ip-26-0-166-36:104435:104920 [6] NCCL INFO comm 0xa2b2950 rank 0 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0x3e7d69e4ffee2364 - Init START
+ip-26-0-166-36:104435:104920 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-36:104436:104923 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+NCCL version 2.18.6+cuda12.1
+NCCL version 2.18.6+cuda12.1
+NCCL version 2.18.6+cuda12.1
+NCCL version 2.18.6+cuda12.1
+NCCL version 2.18.6+cuda12.1
+NCCL version 2.18.6+cuda12.1
+ip-26-0-162-46:702880:703399 [0] NCCL INFO Using network Libfabric
+ip-26-0-167-9:1039537:1040038 [0] NCCL INFO Using network Libfabric
+ip-26-0-162-46:702885:703404 [5] NCCL INFO Using network Libfabric
+ip-26-0-162-46:702881:703400 [1] NCCL INFO Using network Libfabric
+ip-26-0-162-46:702884:703402 [4] NCCL INFO Using network Libfabric
+ip-26-0-162-46:702887:703406 [7] NCCL INFO Using network Libfabric
+ip-26-0-167-9:1039543:1040043 [6] NCCL INFO Using network Libfabric
+ip-26-0-162-46:702886:703405 [6] NCCL INFO Using network Libfabric
+ip-26-0-167-9:1039538:1040044 [1] NCCL INFO Using network Libfabric
+ip-26-0-162-46:702885:703404 [5] NCCL INFO comm 0xa213dc0 rank 1 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0x6a6e536d58c1344e - Init START
+ip-26-0-162-46:702880:703399 [0] NCCL INFO comm 0xba3bc80 rank 0 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0x6a967c92be24b549 - Init START
+ip-26-0-162-46:702881:703400 [1] NCCL INFO comm 0xb50f5b0 rank 1 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0x6a967c92be24b549 - Init START
+ip-26-0-162-46:702884:703402 [4] NCCL INFO comm 0xb014080 rank 0 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0x6a6e536d58c1344e - Init START
+ip-26-0-167-9:1039544:1040046 [7] NCCL INFO Using network Libfabric
+ip-26-0-162-46:702882:703408 [2] NCCL INFO Using network Libfabric
+ip-26-0-167-9:1039541:1040041 [4] NCCL INFO Using network Libfabric
+ip-26-0-162-46:702885:703404 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-162-46:702880:703399 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/349
+ip-26-0-162-46:702884:703402 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-162-46:702881:703400 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-167-9:1039542:1040047 [5] NCCL INFO Using network Libfabric
+ip-26-0-167-9:1039540:1040048 [3] NCCL INFO Using network Libfabric
+ip-26-0-167-9:1039539:1040045 [2] NCCL INFO Using network Libfabric
+ip-26-0-162-46:702887:703406 [7] NCCL INFO comm 0xa204fc0 rank 1 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0x80f2837194b47716 - Init START
+ip-26-0-162-46:702886:703405 [6] NCCL INFO comm 0xa6f9960 rank 0 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0x80f2837194b47716 - Init START
+ip-26-0-167-9:1039544:1040046 [7] NCCL INFO comm 0xa16c820 rank 1 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0x74262946ed30693c - Init START
+ip-26-0-167-9:1039543:1040043 [6] NCCL INFO comm 0xaa35a60 rank 0 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0x74262946ed30693c - Init START
+ip-26-0-162-46:702883:703409 [3] NCCL INFO Using network Libfabric
+ip-26-0-162-46:702886:703405 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-162-46:702887:703406 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-167-9:1039538:1040044 [1] NCCL INFO comm 0x9b103b0 rank 1 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0x62fffa362beddbec - Init START
+ip-26-0-167-9:1039537:1040038 [0] NCCL INFO comm 0xa6f52b0 rank 0 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0x62fffa362beddbec - Init START
+ip-26-0-167-9:1039544:1040046 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-167-9:1039543:1040043 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-167-9:1039538:1040044 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-167-9:1039537:1040038 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-167-9:1039542:1040047 [5] NCCL INFO comm 0x99250b0 rank 1 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0x8f4515ff2b0a9fd3 - Init START
+ip-26-0-167-9:1039541:1040041 [4] NCCL INFO comm 0xaf66470 rank 0 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0x8f4515ff2b0a9fd3 - Init START
+ip-26-0-167-9:1039540:1040048 [3] NCCL INFO comm 0xaca56b0 rank 1 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0x486d4b076e52af12 - Init START
+ip-26-0-167-9:1039539:1040045 [2] NCCL INFO comm 0x9be8210 rank 0 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0x486d4b076e52af12 - Init START
+ip-26-0-162-46:702883:703409 [3] NCCL INFO comm 0x9c2f910 rank 1 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0x38b4054e8ec636f3 - Init START
+ip-26-0-162-46:702882:703408 [2] NCCL INFO comm 0xae6fb70 rank 0 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0x38b4054e8ec636f3 - Init START
+ip-26-0-167-9:1039541:1040041 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-167-9:1039542:1040047 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-162-46:702883:703409 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-167-9:1039540:1040048 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-167-9:1039539:1040045 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-162-46:702882:703408 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-15:127292:127782 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-166-36:104432:104922 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-165-213:9626:10118 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-162-46:702887:703406 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-166-125:762819:763316 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,fff00000,00000000
+ip-26-0-167-9:1039537:1040038 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-166-15:127297:127781 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-166-125:762818:763321 [3] NCCL INFO Setting affinity for GPU 3 to 0fff,ffffffff
+ip-26-0-166-15:127290:127785 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-165-213:9632:10121 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-167-9:1039542:1040047 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-166-36:104436:104923 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-166-125:762817:763317 [2] NCCL INFO Setting affinity for GPU 2 to 0fff,ffffffff
+ip-26-0-166-125:762817:763317 [2] NCCL INFO Channel 00/24 :    0   1
+ip-26-0-166-125:762818:763321 [3] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] -1/-1/-1->1->0 [3] -1/-1/-1->1->0 [4] -1/-1/-1->1->0 [5] -1/-1/-1->1->0 [6] 0/-1/-1->1->-1 [7] 0/-1/-1->1->-1 [8] 0/-1/-1->1->-1 [9] 0/-1/-1->1->-1 [10] 0/-1/-1->1->-1 [11] 0/-1/-1->1->-1 [12] -1/-1/-1->1->0 [13] -1/-1/-1->1->0 [14] -1/-1/-1->1->0 [15] -1/-1/-1->1->0 [16] -1/-1/-1->1->0 [17] -1/-1/-1->1->0 [18] 0/-1/-1->1->-1 [19] 0/-1/-1->1->-1 [20] 0/-1/-1->1->-1 [21] 0/-1/-1->1->-1 [22] 0/-1/-1->1->-1 [23] 0/-1/-1->1->-1
+ip-26-0-166-125:762817:763317 [2] NCCL INFO Channel 01/24 :    0   1
+ip-26-0-166-125:762818:763321 [3] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-166-125:762817:763317 [2] NCCL INFO Channel 02/24 :    0   1
+ip-26-0-166-125:762817:763317 [2] NCCL INFO Channel 03/24 :    0   1
+ip-26-0-166-125:762817:763317 [2] NCCL INFO Channel 04/24 :    0   1
+ip-26-0-166-125:762817:763317 [2] NCCL INFO Channel 05/24 :    0   1
+ip-26-0-166-125:762817:763317 [2] NCCL INFO Channel 06/24 :    0   1
+ip-26-0-166-125:762817:763317 [2] NCCL INFO Channel 07/24 :    0   1
+ip-26-0-166-125:762817:763317 [2] NCCL INFO Channel 08/24 :    0   1
+ip-26-0-166-125:762817:763317 [2] NCCL INFO Channel 09/24 :    0   1
+ip-26-0-166-125:762817:763317 [2] NCCL INFO Channel 10/24 :    0   1
+ip-26-0-166-125:762817:763317 [2] NCCL INFO Channel 11/24 :    0   1
+ip-26-0-166-125:762817:763317 [2] NCCL INFO Channel 12/24 :    0   1
+ip-26-0-166-125:762817:763317 [2] NCCL INFO Channel 13/24 :    0   1
+ip-26-0-166-125:762817:763317 [2] NCCL INFO Channel 14/24 :    0   1
+ip-26-0-166-125:762817:763317 [2] NCCL INFO Channel 15/24 :    0   1
+ip-26-0-166-125:762817:763317 [2] NCCL INFO Channel 16/24 :    0   1
+ip-26-0-166-125:762817:763317 [2] NCCL INFO Channel 17/24 :    0   1
+ip-26-0-166-125:762817:763317 [2] NCCL INFO Channel 18/24 :    0   1
+ip-26-0-166-125:762817:763317 [2] NCCL INFO Channel 19/24 :    0   1
+ip-26-0-166-125:762817:763317 [2] NCCL INFO Channel 20/24 :    0   1
+ip-26-0-166-125:762817:763317 [2] NCCL INFO Channel 21/24 :    0   1
+ip-26-0-166-125:762817:763317 [2] NCCL INFO Channel 22/24 :    0   1
+ip-26-0-166-125:762817:763317 [2] NCCL INFO Channel 23/24 :    0   1
+ip-26-0-166-125:762817:763317 [2] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] -1/-1/-1->0->1 [7] -1/-1/-1->0->1 [8] -1/-1/-1->0->1 [9] -1/-1/-1->0->1 [10] -1/-1/-1->0->1 [11] -1/-1/-1->0->1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] -1/-1/-1->0->1 [19] -1/-1/-1->0->1 [20] -1/-1/-1->0->1 [21] -1/-1/-1->0->1 [22] -1/-1/-1->0->1 [23] -1/-1/-1->0->1
+ip-26-0-166-125:762817:763317 [2] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-166-36:104430:104924 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-166-36:104435:104920 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-166-36:104435:104920 [6] NCCL INFO Channel 00/24 :    0   1
+ip-26-0-166-36:104435:104920 [6] NCCL INFO Channel 01/24 :    0   1
+ip-26-0-166-36:104435:104920 [6] NCCL INFO Channel 02/24 :    0   1
+ip-26-0-166-36:104435:104920 [6] NCCL INFO Channel 03/24 :    0   1
+ip-26-0-166-36:104435:104920 [6] NCCL INFO Channel 04/24 :    0   1
+ip-26-0-166-36:104435:104920 [6] NCCL INFO Channel 05/24 :    0   1
+ip-26-0-166-36:104435:104920 [6] NCCL INFO Channel 06/24 :    0   1
+ip-26-0-166-36:104435:104920 [6] NCCL INFO Channel 07/24 :    0   1
+ip-26-0-166-36:104435:104920 [6] NCCL INFO Channel 08/24 :    0   1
+ip-26-0-166-36:104435:104920 [6] NCCL INFO Channel 09/24 :    0   1
+ip-26-0-166-36:104435:104920 [6] NCCL INFO Channel 10/24 :    0   1
+ip-26-0-166-36:104435:104920 [6] NCCL INFO Channel 11/24 :    0   1
+ip-26-0-166-36:104435:104920 [6] NCCL INFO Channel 12/24 :    0   1
+ip-26-0-166-36:104435:104920 [6] NCCL INFO Channel 13/24 :    0   1
+ip-26-0-166-36:104435:104920 [6] NCCL INFO Channel 14/24 :    0   1
+ip-26-0-166-36:104435:104920 [6] NCCL INFO Channel 15/24 :    0   1
+ip-26-0-166-36:104435:104920 [6] NCCL INFO Channel 16/24 :    0   1
+ip-26-0-166-36:104436:104923 [7] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] -1/-1/-1->1->0 [3] -1/-1/-1->1->0 [4] -1/-1/-1->1->0 [5] -1/-1/-1->1->0 [6] 0/-1/-1->1->-1 [7] 0/-1/-1->1->-1 [8] 0/-1/-1->1->-1 [9] 0/-1/-1->1->-1 [10] 0/-1/-1->1->-1 [11] 0/-1/-1->1->-1 [12] -1/-1/-1->1->0 [13] -1/-1/-1->1->0 [14] -1/-1/-1->1->0 [15] -1/-1/-1->1->0 [16] -1/-1/-1->1->0 [17] -1/-1/-1->1->0 [18] 0/-1/-1->1->-1 [19] 0/-1/-1->1->-1 [20] 0/-1/-1->1->-1 [21] 0/-1/-1->1->-1 [22] 0/-1/-1->1->-1 [23] 0/-1/-1->1->-1
+ip-26-0-166-36:104435:104920 [6] NCCL INFO Channel 17/24 :    0   1
+ip-26-0-166-36:104436:104923 [7] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-166-36:104435:104920 [6] NCCL INFO Channel 18/24 :    0   1
+ip-26-0-166-36:104435:104920 [6] NCCL INFO Channel 19/24 :    0   1
+ip-26-0-166-36:104435:104920 [6] NCCL INFO Channel 20/24 :    0   1
+ip-26-0-166-36:104435:104920 [6] NCCL INFO Channel 21/24 :    0   1
+ip-26-0-166-36:104435:104920 [6] NCCL INFO Channel 22/24 :    0   1
+ip-26-0-166-36:104435:104920 [6] NCCL INFO Channel 23/24 :    0   1
+ip-26-0-166-36:104435:104920 [6] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] -1/-1/-1->0->1 [7] -1/-1/-1->0->1 [8] -1/-1/-1->0->1 [9] -1/-1/-1->0->1 [10] -1/-1/-1->0->1 [11] -1/-1/-1->0->1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] -1/-1/-1->0->1 [19] -1/-1/-1->0->1 [20] -1/-1/-1->0->1 [21] -1/-1/-1->0->1 [22] -1/-1/-1->0->1 [23] -1/-1/-1->0->1
+ip-26-0-166-36:104435:104920 [6] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-166-15:127293:127786 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-166-15:127292:127782 [2] NCCL INFO Channel 00/24 :    0   1
+ip-26-0-166-15:127292:127782 [2] NCCL INFO Channel 01/24 :    0   1
+ip-26-0-166-15:127292:127782 [2] NCCL INFO Channel 02/24 :    0   1
+ip-26-0-166-15:127292:127782 [2] NCCL INFO Channel 03/24 :    0   1
+ip-26-0-166-15:127292:127782 [2] NCCL INFO Channel 04/24 :    0   1
+ip-26-0-166-15:127292:127782 [2] NCCL INFO Channel 05/24 :    0   1
+ip-26-0-166-15:127292:127782 [2] NCCL INFO Channel 06/24 :    0   1
+ip-26-0-166-15:127293:127786 [3] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] -1/-1/-1->1->0 [3] -1/-1/-1->1->0 [4] -1/-1/-1->1->0 [5] -1/-1/-1->1->0 [6] 0/-1/-1->1->-1 [7] 0/-1/-1->1->-1 [8] 0/-1/-1->1->-1 [9] 0/-1/-1->1->-1 [10] 0/-1/-1->1->-1 [11] 0/-1/-1->1->-1 [12] -1/-1/-1->1->0 [13] -1/-1/-1->1->0 [14] -1/-1/-1->1->0 [15] -1/-1/-1->1->0 [16] -1/-1/-1->1->0 [17] -1/-1/-1->1->0 [18] 0/-1/-1->1->-1 [19] 0/-1/-1->1->-1 [20] 0/-1/-1->1->-1 [21] 0/-1/-1->1->-1 [22] 0/-1/-1->1->-1 [23] 0/-1/-1->1->-1
+ip-26-0-166-15:127292:127782 [2] NCCL INFO Channel 07/24 :    0   1
+ip-26-0-166-15:127292:127782 [2] NCCL INFO Channel 08/24 :    0   1
+ip-26-0-166-15:127292:127782 [2] NCCL INFO Channel 09/24 :    0   1
+ip-26-0-166-15:127293:127786 [3] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-166-15:127292:127782 [2] NCCL INFO Channel 10/24 :    0   1
+ip-26-0-166-15:127292:127782 [2] NCCL INFO Channel 11/24 :    0   1
+ip-26-0-166-15:127292:127782 [2] NCCL INFO Channel 12/24 :    0   1
+ip-26-0-166-15:127292:127782 [2] NCCL INFO Channel 13/24 :    0   1
+ip-26-0-166-15:127292:127782 [2] NCCL INFO Channel 14/24 :    0   1
+ip-26-0-166-15:127292:127782 [2] NCCL INFO Channel 15/24 :    0   1
+ip-26-0-166-15:127292:127782 [2] NCCL INFO Channel 16/24 :    0   1
+ip-26-0-166-15:127292:127782 [2] NCCL INFO Channel 17/24 :    0   1
+ip-26-0-166-15:127292:127782 [2] NCCL INFO Channel 18/24 :    0   1
+ip-26-0-166-15:127292:127782 [2] NCCL INFO Channel 19/24 :    0   1
+ip-26-0-166-15:127292:127782 [2] NCCL INFO Channel 20/24 :    0   1
+ip-26-0-166-15:127292:127782 [2] NCCL INFO Channel 21/24 :    0   1
+ip-26-0-166-15:127292:127782 [2] NCCL INFO Channel 22/24 :    0   1
+ip-26-0-166-15:127292:127782 [2] NCCL INFO Channel 23/24 :    0   1
+ip-26-0-166-15:127292:127782 [2] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] -1/-1/-1->0->1 [7] -1/-1/-1->0->1 [8] -1/-1/-1->0->1 [9] -1/-1/-1->0->1 [10] -1/-1/-1->0->1 [11] -1/-1/-1->0->1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] -1/-1/-1->0->1 [19] -1/-1/-1->0->1 [20] -1/-1/-1->0->1 [21] -1/-1/-1->0->1 [22] -1/-1/-1->0->1 [23] -1/-1/-1->0->1
+ip-26-0-166-15:127292:127782 [2] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-162-46:702882:703408 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-162-46:702880:703399 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-166-125:762821:763318 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,fff00000,00000000
+ip-26-0-166-214:624946:625439 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-162-46:702883:703409 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-162-46:702882:703408 [2] NCCL INFO Channel 00/24 :    0   1
+ip-26-0-162-46:702882:703408 [2] NCCL INFO Channel 01/24 :    0   1
+ip-26-0-162-46:702882:703408 [2] NCCL INFO Channel 02/24 :    0   1
+ip-26-0-162-46:702882:703408 [2] NCCL INFO Channel 03/24 :    0   1
+ip-26-0-162-46:702882:703408 [2] NCCL INFO Channel 04/24 :    0   1
+ip-26-0-162-46:702882:703408 [2] NCCL INFO Channel 05/24 :    0   1
+ip-26-0-162-46:702882:703408 [2] NCCL INFO Channel 06/24 :    0   1
+ip-26-0-162-46:702882:703408 [2] NCCL INFO Channel 07/24 :    0   1
+ip-26-0-162-46:702882:703408 [2] NCCL INFO Channel 08/24 :    0   1
+ip-26-0-162-46:702882:703408 [2] NCCL INFO Channel 09/24 :    0   1
+ip-26-0-162-46:702883:703409 [3] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] -1/-1/-1->1->0 [3] -1/-1/-1->1->0 [4] -1/-1/-1->1->0 [5] -1/-1/-1->1->0 [6] 0/-1/-1->1->-1 [7] 0/-1/-1->1->-1 [8] 0/-1/-1->1->-1 [9] 0/-1/-1->1->-1 [10] 0/-1/-1->1->-1 [11] 0/-1/-1->1->-1 [12] -1/-1/-1->1->0 [13] -1/-1/-1->1->0 [14] -1/-1/-1->1->0 [15] -1/-1/-1->1->0 [16] -1/-1/-1->1->0 [17] -1/-1/-1->1->0 [18] 0/-1/-1->1->-1 [19] 0/-1/-1->1->-1 [20] 0/-1/-1->1->-1 [21] 0/-1/-1->1->-1 [22] 0/-1/-1->1->-1 [23] 0/-1/-1->1->-1
+ip-26-0-162-46:702882:703408 [2] NCCL INFO Channel 10/24 :    0   1
+ip-26-0-162-46:702882:703408 [2] NCCL INFO Channel 11/24 :    0   1
+ip-26-0-162-46:702883:703409 [3] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-162-46:702882:703408 [2] NCCL INFO Channel 12/24 :    0   1
+ip-26-0-162-46:702882:703408 [2] NCCL INFO Channel 13/24 :    0   1
+ip-26-0-162-46:702882:703408 [2] NCCL INFO Channel 14/24 :    0   1
+ip-26-0-162-46:702882:703408 [2] NCCL INFO Channel 15/24 :    0   1
+ip-26-0-162-46:702882:703408 [2] NCCL INFO Channel 16/24 :    0   1
+ip-26-0-162-46:702882:703408 [2] NCCL INFO Channel 17/24 :    0   1
+ip-26-0-162-46:702882:703408 [2] NCCL INFO Channel 18/24 :    0   1
+ip-26-0-162-46:702882:703408 [2] NCCL INFO Channel 19/24 :    0   1
+ip-26-0-162-46:702882:703408 [2] NCCL INFO Channel 20/24 :    0   1
+ip-26-0-162-46:702882:703408 [2] NCCL INFO Channel 21/24 :    0   1
+ip-26-0-162-46:702882:703408 [2] NCCL INFO Channel 22/24 :    0   1
+ip-26-0-162-46:702882:703408 [2] NCCL INFO Channel 23/24 :    0   1
+ip-26-0-165-213:9627:10116 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-162-46:702882:703408 [2] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] -1/-1/-1->0->1 [7] -1/-1/-1->0->1 [8] -1/-1/-1->0->1 [9] -1/-1/-1->0->1 [10] -1/-1/-1->0->1 [11] -1/-1/-1->0->1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] -1/-1/-1->0->1 [19] -1/-1/-1->0->1 [20] -1/-1/-1->0->1 [21] -1/-1/-1->0->1 [22] -1/-1/-1->0->1 [23] -1/-1/-1->0->1
+ip-26-0-162-46:702882:703408 [2] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-162-46:702885:703404 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-162-46:702884:703402 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-162-46:702884:703402 [4] NCCL INFO Channel 00/24 :    0   1
+ip-26-0-162-46:702884:703402 [4] NCCL INFO Channel 01/24 :    0   1
+ip-26-0-162-46:702884:703402 [4] NCCL INFO Channel 02/24 :    0   1
+ip-26-0-162-46:702884:703402 [4] NCCL INFO Channel 03/24 :    0   1
+ip-26-0-162-46:702884:703402 [4] NCCL INFO Channel 04/24 :    0   1
+ip-26-0-162-46:702884:703402 [4] NCCL INFO Channel 05/24 :    0   1
+ip-26-0-162-46:702885:703404 [5] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] -1/-1/-1->1->0 [3] -1/-1/-1->1->0 [4] -1/-1/-1->1->0 [5] -1/-1/-1->1->0 [6] 0/-1/-1->1->-1 [7] 0/-1/-1->1->-1 [8] 0/-1/-1->1->-1 [9] 0/-1/-1->1->-1 [10] 0/-1/-1->1->-1 [11] 0/-1/-1->1->-1 [12] -1/-1/-1->1->0 [13] -1/-1/-1->1->0 [14] -1/-1/-1->1->0 [15] -1/-1/-1->1->0 [16] -1/-1/-1->1->0 [17] -1/-1/-1->1->0 [18] 0/-1/-1->1->-1 [19] 0/-1/-1->1->-1 [20] 0/-1/-1->1->-1 [21] 0/-1/-1->1->-1 [22] 0/-1/-1->1->-1 [23] 0/-1/-1->1->-1
+ip-26-0-162-46:702884:703402 [4] NCCL INFO Channel 06/24 :    0   1
+ip-26-0-162-46:702885:703404 [5] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-162-46:702884:703402 [4] NCCL INFO Channel 07/24 :    0   1
+ip-26-0-162-46:702884:703402 [4] NCCL INFO Channel 08/24 :    0   1
+ip-26-0-162-46:702884:703402 [4] NCCL INFO Channel 09/24 :    0   1
+ip-26-0-162-46:702884:703402 [4] NCCL INFO Channel 10/24 :    0   1
+ip-26-0-162-46:702884:703402 [4] NCCL INFO Channel 11/24 :    0   1
+ip-26-0-162-46:702884:703402 [4] NCCL INFO Channel 12/24 :    0   1
+ip-26-0-162-46:702884:703402 [4] NCCL INFO Channel 13/24 :    0   1
+ip-26-0-162-46:702884:703402 [4] NCCL INFO Channel 14/24 :    0   1
+ip-26-0-162-46:702884:703402 [4] NCCL INFO Channel 15/24 :    0   1
+ip-26-0-162-46:702884:703402 [4] NCCL INFO Channel 16/24 :    0   1
+ip-26-0-162-46:702884:703402 [4] NCCL INFO Channel 17/24 :    0   1
+ip-26-0-162-46:702884:703402 [4] NCCL INFO Channel 18/24 :    0   1
+ip-26-0-162-46:702884:703402 [4] NCCL INFO Channel 19/24 :    0   1
+ip-26-0-162-46:702884:703402 [4] NCCL INFO Channel 20/24 :    0   1
+ip-26-0-162-46:702884:703402 [4] NCCL INFO Channel 21/24 :    0   1
+ip-26-0-162-46:702884:703402 [4] NCCL INFO Channel 22/24 :    0   1
+ip-26-0-162-46:702884:703402 [4] NCCL INFO Channel 23/24 :    0   1
+ip-26-0-162-46:702884:703402 [4] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] -1/-1/-1->0->1 [7] -1/-1/-1->0->1 [8] -1/-1/-1->0->1 [9] -1/-1/-1->0->1 [10] -1/-1/-1->0->1 [11] -1/-1/-1->0->1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] -1/-1/-1->0->1 [19] -1/-1/-1->0->1 [20] -1/-1/-1->0->1 [21] -1/-1/-1->0->1 [22] -1/-1/-1->0->1 [23] -1/-1/-1->0->1
+ip-26-0-162-46:702884:703402 [4] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-166-36:104429:104919 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-166-36:104429:104919 [0] NCCL INFO Channel 00/24 :    0   1
+ip-26-0-166-36:104429:104919 [0] NCCL INFO Channel 01/24 :    0   1
+ip-26-0-166-36:104429:104919 [0] NCCL INFO Channel 02/24 :    0   1
+ip-26-0-166-36:104429:104919 [0] NCCL INFO Channel 03/24 :    0   1
+ip-26-0-166-36:104429:104919 [0] NCCL INFO Channel 04/24 :    0   1
+ip-26-0-166-36:104430:104924 [1] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] -1/-1/-1->1->0 [3] -1/-1/-1->1->0 [4] -1/-1/-1->1->0 [5] -1/-1/-1->1->0 [6] 0/-1/-1->1->-1 [7] 0/-1/-1->1->-1 [8] 0/-1/-1->1->-1 [9] 0/-1/-1->1->-1 [10] 0/-1/-1->1->-1 [11] 0/-1/-1->1->-1 [12] -1/-1/-1->1->0 [13] -1/-1/-1->1->0 [14] -1/-1/-1->1->0 [15] -1/-1/-1->1->0 [16] -1/-1/-1->1->0 [17] -1/-1/-1->1->0 [18] 0/-1/-1->1->-1 [19] 0/-1/-1->1->-1 [20] 0/-1/-1->1->-1 [21] 0/-1/-1->1->-1 [22] 0/-1/-1->1->-1 [23] 0/-1/-1->1->-1
+ip-26-0-166-36:104429:104919 [0] NCCL INFO Channel 05/24 :    0   1
+ip-26-0-166-36:104430:104924 [1] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-166-36:104429:104919 [0] NCCL INFO Channel 06/24 :    0   1
+ip-26-0-166-36:104429:104919 [0] NCCL INFO Channel 07/24 :    0   1
+ip-26-0-166-36:104429:104919 [0] NCCL INFO Channel 08/24 :    0   1
+ip-26-0-166-36:104429:104919 [0] NCCL INFO Channel 09/24 :    0   1
+ip-26-0-166-36:104429:104919 [0] NCCL INFO Channel 10/24 :    0   1
+ip-26-0-166-36:104429:104919 [0] NCCL INFO Channel 11/24 :    0   1
+ip-26-0-166-36:104429:104919 [0] NCCL INFO Channel 12/24 :    0   1
+ip-26-0-166-36:104429:104919 [0] NCCL INFO Channel 13/24 :    0   1
+ip-26-0-166-36:104429:104919 [0] NCCL INFO Channel 14/24 :    0   1
+ip-26-0-166-36:104429:104919 [0] NCCL INFO Channel 15/24 :    0   1
+ip-26-0-166-36:104429:104919 [0] NCCL INFO Channel 16/24 :    0   1
+ip-26-0-166-36:104429:104919 [0] NCCL INFO Channel 17/24 :    0   1
+ip-26-0-166-36:104429:104919 [0] NCCL INFO Channel 18/24 :    0   1
+ip-26-0-166-36:104429:104919 [0] NCCL INFO Channel 19/24 :    0   1
+ip-26-0-166-36:104429:104919 [0] NCCL INFO Channel 20/24 :    0   1
+ip-26-0-166-36:104429:104919 [0] NCCL INFO Channel 21/24 :    0   1
+ip-26-0-166-36:104429:104919 [0] NCCL INFO Channel 22/24 :    0   1
+ip-26-0-166-36:104429:104919 [0] NCCL INFO Channel 23/24 :    0   1
+ip-26-0-166-36:104429:104919 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] -1/-1/-1->0->1 [7] -1/-1/-1->0->1 [8] -1/-1/-1->0->1 [9] -1/-1/-1->0->1 [10] -1/-1/-1->0->1 [11] -1/-1/-1->0->1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] -1/-1/-1->0->1 [19] -1/-1/-1->0->1 [20] -1/-1/-1->0->1 [21] -1/-1/-1->0->1 [22] -1/-1/-1->0->1 [23] -1/-1/-1->0->1
+ip-26-0-166-36:104429:104919 [0] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-167-9:1039543:1040043 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-166-36:104434:104921 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-165-213:9628:10120 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-165-213:9628:10120 [3] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] -1/-1/-1->1->0 [3] -1/-1/-1->1->0 [4] -1/-1/-1->1->0 [5] -1/-1/-1->1->0 [6] 0/-1/-1->1->-1 [7] 0/-1/-1->1->-1 [8] 0/-1/-1->1->-1 [9] 0/-1/-1->1->-1 [10] 0/-1/-1->1->-1 [11] 0/-1/-1->1->-1 [12] -1/-1/-1->1->0 [13] -1/-1/-1->1->0 [14] -1/-1/-1->1->0 [15] -1/-1/-1->1->0 [16] -1/-1/-1->1->0 [17] -1/-1/-1->1->0 [18] 0/-1/-1->1->-1 [19] 0/-1/-1->1->-1 [20] 0/-1/-1->1->-1 [21] 0/-1/-1->1->-1 [22] 0/-1/-1->1->-1 [23] 0/-1/-1->1->-1
+ip-26-0-165-213:9627:10116 [2] NCCL INFO Channel 00/24 :    0   1
+ip-26-0-165-213:9627:10116 [2] NCCL INFO Channel 01/24 :    0   1
+ip-26-0-165-213:9628:10120 [3] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-165-213:9627:10116 [2] NCCL INFO Channel 02/24 :    0   1
+ip-26-0-165-213:9627:10116 [2] NCCL INFO Channel 03/24 :    0   1
+ip-26-0-165-213:9627:10116 [2] NCCL INFO Channel 04/24 :    0   1
+ip-26-0-165-213:9627:10116 [2] NCCL INFO Channel 05/24 :    0   1
+ip-26-0-165-213:9627:10116 [2] NCCL INFO Channel 06/24 :    0   1
+ip-26-0-165-213:9627:10116 [2] NCCL INFO Channel 07/24 :    0   1
+ip-26-0-165-213:9627:10116 [2] NCCL INFO Channel 08/24 :    0   1
+ip-26-0-165-213:9627:10116 [2] NCCL INFO Channel 09/24 :    0   1
+ip-26-0-165-213:9627:10116 [2] NCCL INFO Channel 10/24 :    0   1
+ip-26-0-165-213:9627:10116 [2] NCCL INFO Channel 11/24 :    0   1
+ip-26-0-165-213:9627:10116 [2] NCCL INFO Channel 12/24 :    0   1
+ip-26-0-165-213:9627:10116 [2] NCCL INFO Channel 13/24 :    0   1
+ip-26-0-165-213:9627:10116 [2] NCCL INFO Channel 14/24 :    0   1
+ip-26-0-165-213:9627:10116 [2] NCCL INFO Channel 15/24 :    0   1
+ip-26-0-165-213:9627:10116 [2] NCCL INFO Channel 16/24 :    0   1
+ip-26-0-165-213:9627:10116 [2] NCCL INFO Channel 17/24 :    0   1
+ip-26-0-165-213:9627:10116 [2] NCCL INFO Channel 18/24 :    0   1
+ip-26-0-165-213:9627:10116 [2] NCCL INFO Channel 19/24 :    0   1
+ip-26-0-165-213:9627:10116 [2] NCCL INFO Channel 20/24 :    0   1
+ip-26-0-165-213:9627:10116 [2] NCCL INFO Channel 21/24 :    0   1
+ip-26-0-165-213:9627:10116 [2] NCCL INFO Channel 22/24 :    0   1
+ip-26-0-165-213:9627:10116 [2] NCCL INFO Channel 23/24 :    0   1
+ip-26-0-165-213:9627:10116 [2] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] -1/-1/-1->0->1 [7] -1/-1/-1->0->1 [8] -1/-1/-1->0->1 [9] -1/-1/-1->0->1 [10] -1/-1/-1->0->1 [11] -1/-1/-1->0->1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] -1/-1/-1->0->1 [19] -1/-1/-1->0->1 [20] -1/-1/-1->0->1 [21] -1/-1/-1->0->1 [22] -1/-1/-1->0->1 [23] -1/-1/-1->0->1
+ip-26-0-165-213:9627:10116 [2] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-166-15:127291:127787 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-166-15:127290:127785 [0] NCCL INFO Channel 00/24 :    0   1
+ip-26-0-166-15:127290:127785 [0] NCCL INFO Channel 01/24 :    0   1
+ip-26-0-166-15:127290:127785 [0] NCCL INFO Channel 02/24 :    0   1
+ip-26-0-166-15:127290:127785 [0] NCCL INFO Channel 03/24 :    0   1
+ip-26-0-166-15:127290:127785 [0] NCCL INFO Channel 04/24 :    0   1
+ip-26-0-166-15:127290:127785 [0] NCCL INFO Channel 05/24 :    0   1
+ip-26-0-166-15:127290:127785 [0] NCCL INFO Channel 06/24 :    0   1
+ip-26-0-166-15:127291:127787 [1] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] -1/-1/-1->1->0 [3] -1/-1/-1->1->0 [4] -1/-1/-1->1->0 [5] -1/-1/-1->1->0 [6] 0/-1/-1->1->-1 [7] 0/-1/-1->1->-1 [8] 0/-1/-1->1->-1 [9] 0/-1/-1->1->-1 [10] 0/-1/-1->1->-1 [11] 0/-1/-1->1->-1 [12] -1/-1/-1->1->0 [13] -1/-1/-1->1->0 [14] -1/-1/-1->1->0 [15] -1/-1/-1->1->0 [16] -1/-1/-1->1->0 [17] -1/-1/-1->1->0 [18] 0/-1/-1->1->-1 [19] 0/-1/-1->1->-1 [20] 0/-1/-1->1->-1 [21] 0/-1/-1->1->-1 [22] 0/-1/-1->1->-1 [23] 0/-1/-1->1->-1
+ip-26-0-166-15:127290:127785 [0] NCCL INFO Channel 07/24 :    0   1
+ip-26-0-166-15:127290:127785 [0] NCCL INFO Channel 08/24 :    0   1
+ip-26-0-166-15:127291:127787 [1] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-166-15:127290:127785 [0] NCCL INFO Channel 09/24 :    0   1
+ip-26-0-166-15:127290:127785 [0] NCCL INFO Channel 10/24 :    0   1
+ip-26-0-166-15:127290:127785 [0] NCCL INFO Channel 11/24 :    0   1
+ip-26-0-166-15:127290:127785 [0] NCCL INFO Channel 12/24 :    0   1
+ip-26-0-166-15:127290:127785 [0] NCCL INFO Channel 13/24 :    0   1
+ip-26-0-166-15:127290:127785 [0] NCCL INFO Channel 14/24 :    0   1
+ip-26-0-166-15:127290:127785 [0] NCCL INFO Channel 15/24 :    0   1
+ip-26-0-166-15:127290:127785 [0] NCCL INFO Channel 16/24 :    0   1
+ip-26-0-166-15:127290:127785 [0] NCCL INFO Channel 17/24 :    0   1
+ip-26-0-166-15:127290:127785 [0] NCCL INFO Channel 18/24 :    0   1
+ip-26-0-166-15:127290:127785 [0] NCCL INFO Channel 19/24 :    0   1
+ip-26-0-166-15:127290:127785 [0] NCCL INFO Channel 20/24 :    0   1
+ip-26-0-166-15:127290:127785 [0] NCCL INFO Channel 21/24 :    0   1
+ip-26-0-166-15:127290:127785 [0] NCCL INFO Channel 22/24 :    0   1
+ip-26-0-166-15:127290:127785 [0] NCCL INFO Channel 23/24 :    0   1
+ip-26-0-166-15:127290:127785 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] -1/-1/-1->0->1 [7] -1/-1/-1->0->1 [8] -1/-1/-1->0->1 [9] -1/-1/-1->0->1 [10] -1/-1/-1->0->1 [11] -1/-1/-1->0->1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] -1/-1/-1->0->1 [19] -1/-1/-1->0->1 [20] -1/-1/-1->0->1 [21] -1/-1/-1->0->1 [22] -1/-1/-1->0->1 [23] -1/-1/-1->0->1
+ip-26-0-166-15:127290:127785 [0] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-166-15:127295:127783 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-165-213:9625:10112 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-165-213:9625:10112 [0] NCCL INFO Channel 00/24 :    0   1
+ip-26-0-165-213:9625:10112 [0] NCCL INFO Channel 01/24 :    0   1
+ip-26-0-165-213:9625:10112 [0] NCCL INFO Channel 02/24 :    0   1
+ip-26-0-165-213:9625:10112 [0] NCCL INFO Channel 03/24 :    0   1
+ip-26-0-165-213:9625:10112 [0] NCCL INFO Channel 04/24 :    0   1
+ip-26-0-165-213:9625:10112 [0] NCCL INFO Channel 05/24 :    0   1
+ip-26-0-165-213:9625:10112 [0] NCCL INFO Channel 06/24 :    0   1
+ip-26-0-165-213:9625:10112 [0] NCCL INFO Channel 07/24 :    0   1
+ip-26-0-165-213:9626:10118 [1] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] -1/-1/-1->1->0 [3] -1/-1/-1->1->0 [4] -1/-1/-1->1->0 [5] -1/-1/-1->1->0 [6] 0/-1/-1->1->-1 [7] 0/-1/-1->1->-1 [8] 0/-1/-1->1->-1 [9] 0/-1/-1->1->-1 [10] 0/-1/-1->1->-1 [11] 0/-1/-1->1->-1 [12] -1/-1/-1->1->0 [13] -1/-1/-1->1->0 [14] -1/-1/-1->1->0 [15] -1/-1/-1->1->0 [16] -1/-1/-1->1->0 [17] -1/-1/-1->1->0 [18] 0/-1/-1->1->-1 [19] 0/-1/-1->1->-1 [20] 0/-1/-1->1->-1 [21] 0/-1/-1->1->-1 [22] 0/-1/-1->1->-1 [23] 0/-1/-1->1->-1
+ip-26-0-165-213:9625:10112 [0] NCCL INFO Channel 08/24 :    0   1
+ip-26-0-165-213:9625:10112 [0] NCCL INFO Channel 09/24 :    0   1
+ip-26-0-165-213:9626:10118 [1] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-165-213:9625:10112 [0] NCCL INFO Channel 10/24 :    0   1
+ip-26-0-165-213:9625:10112 [0] NCCL INFO Channel 11/24 :    0   1
+ip-26-0-165-213:9625:10112 [0] NCCL INFO Channel 12/24 :    0   1
+ip-26-0-165-213:9625:10112 [0] NCCL INFO Channel 13/24 :    0   1
+ip-26-0-162-46:702886:703405 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-165-213:9625:10112 [0] NCCL INFO Channel 14/24 :    0   1
+ip-26-0-165-213:9625:10112 [0] NCCL INFO Channel 15/24 :    0   1
+ip-26-0-165-213:9625:10112 [0] NCCL INFO Channel 16/24 :    0   1
+ip-26-0-165-213:9625:10112 [0] NCCL INFO Channel 17/24 :    0   1
+ip-26-0-165-213:9625:10112 [0] NCCL INFO Channel 18/24 :    0   1
+ip-26-0-165-213:9625:10112 [0] NCCL INFO Channel 19/24 :    0   1
+ip-26-0-165-213:9625:10112 [0] NCCL INFO Channel 20/24 :    0   1
+ip-26-0-165-213:9625:10112 [0] NCCL INFO Channel 21/24 :    0   1
+ip-26-0-165-213:9625:10112 [0] NCCL INFO Channel 22/24 :    0   1
+ip-26-0-165-213:9625:10112 [0] NCCL INFO Channel 23/24 :    0   1
+ip-26-0-165-213:9625:10112 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] -1/-1/-1->0->1 [7] -1/-1/-1->0->1 [8] -1/-1/-1->0->1 [9] -1/-1/-1->0->1 [10] -1/-1/-1->0->1 [11] -1/-1/-1->0->1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] -1/-1/-1->0->1 [19] -1/-1/-1->0->1 [20] -1/-1/-1->0->1 [21] -1/-1/-1->0->1 [22] -1/-1/-1->0->1 [23] -1/-1/-1->0->1
+ip-26-0-165-213:9625:10112 [0] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-162-46:702886:703405 [6] NCCL INFO Channel 00/24 :    0   1
+ip-26-0-162-46:702886:703405 [6] NCCL INFO Channel 01/24 :    0   1
+ip-26-0-162-46:702886:703405 [6] NCCL INFO Channel 02/24 :    0   1
+ip-26-0-162-46:702886:703405 [6] NCCL INFO Channel 03/24 :    0   1
+ip-26-0-162-46:702886:703405 [6] NCCL INFO Channel 04/24 :    0   1
+ip-26-0-162-46:702886:703405 [6] NCCL INFO Channel 05/24 :    0   1
+ip-26-0-162-46:702886:703405 [6] NCCL INFO Channel 06/24 :    0   1
+ip-26-0-162-46:702886:703405 [6] NCCL INFO Channel 07/24 :    0   1
+ip-26-0-162-46:702886:703405 [6] NCCL INFO Channel 08/24 :    0   1
+ip-26-0-162-46:702886:703405 [6] NCCL INFO Channel 09/24 :    0   1
+ip-26-0-162-46:702887:703406 [7] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] -1/-1/-1->1->0 [3] -1/-1/-1->1->0 [4] -1/-1/-1->1->0 [5] -1/-1/-1->1->0 [6] 0/-1/-1->1->-1 [7] 0/-1/-1->1->-1 [8] 0/-1/-1->1->-1 [9] 0/-1/-1->1->-1 [10] 0/-1/-1->1->-1 [11] 0/-1/-1->1->-1 [12] -1/-1/-1->1->0 [13] -1/-1/-1->1->0 [14] -1/-1/-1->1->0 [15] -1/-1/-1->1->0 [16] -1/-1/-1->1->0 [17] -1/-1/-1->1->0 [18] 0/-1/-1->1->-1 [19] 0/-1/-1->1->-1 [20] 0/-1/-1->1->-1 [21] 0/-1/-1->1->-1 [22] 0/-1/-1->1->-1 [23] 0/-1/-1->1->-1
+ip-26-0-162-46:702886:703405 [6] NCCL INFO Channel 10/24 :    0   1
+ip-26-0-162-46:702886:703405 [6] NCCL INFO Channel 11/24 :    0   1
+ip-26-0-162-46:702887:703406 [7] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-162-46:702886:703405 [6] NCCL INFO Channel 12/24 :    0   1
+ip-26-0-162-46:702886:703405 [6] NCCL INFO Channel 13/24 :    0   1
+ip-26-0-162-46:702886:703405 [6] NCCL INFO Channel 14/24 :    0   1
+ip-26-0-162-46:702886:703405 [6] NCCL INFO Channel 15/24 :    0   1
+ip-26-0-162-46:702886:703405 [6] NCCL INFO Channel 16/24 :    0   1
+ip-26-0-162-46:702886:703405 [6] NCCL INFO Channel 17/24 :    0   1
+ip-26-0-162-46:702886:703405 [6] NCCL INFO Channel 18/24 :    0   1
+ip-26-0-162-46:702886:703405 [6] NCCL INFO Channel 19/24 :    0   1
+ip-26-0-162-46:702886:703405 [6] NCCL INFO Channel 20/24 :    0   1
+ip-26-0-162-46:702886:703405 [6] NCCL INFO Channel 21/24 :    0   1
+ip-26-0-162-46:702886:703405 [6] NCCL INFO Channel 22/24 :    0   1
+ip-26-0-162-46:702886:703405 [6] NCCL INFO Channel 23/24 :    0   1
+ip-26-0-162-46:702886:703405 [6] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] -1/-1/-1->0->1 [7] -1/-1/-1->0->1 [8] -1/-1/-1->0->1 [9] -1/-1/-1->0->1 [10] -1/-1/-1->0->1 [11] -1/-1/-1->0->1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] -1/-1/-1->0->1 [19] -1/-1/-1->0->1 [20] -1/-1/-1->0->1 [21] -1/-1/-1->0->1 [22] -1/-1/-1->0->1 [23] -1/-1/-1->0->1
+ip-26-0-162-46:702886:703405 [6] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-166-125:762815:763320 [0] NCCL INFO Setting affinity for GPU 0 to 0fff,ffffffff
+ip-26-0-167-9:1039540:1040048 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-166-36:104433:104915 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-166-36:104433:104915 [4] NCCL INFO Channel 00/24 :    0   1
+ip-26-0-166-36:104433:104915 [4] NCCL INFO Channel 01/24 :    0   1
+ip-26-0-166-36:104433:104915 [4] NCCL INFO Channel 02/24 :    0   1
+ip-26-0-166-36:104433:104915 [4] NCCL INFO Channel 03/24 :    0   1
+ip-26-0-166-36:104433:104915 [4] NCCL INFO Channel 04/24 :    0   1
+ip-26-0-166-36:104433:104915 [4] NCCL INFO Channel 05/24 :    0   1
+ip-26-0-166-36:104433:104915 [4] NCCL INFO Channel 06/24 :    0   1
+ip-26-0-166-36:104433:104915 [4] NCCL INFO Channel 07/24 :    0   1
+ip-26-0-166-36:104433:104915 [4] NCCL INFO Channel 08/24 :    0   1
+ip-26-0-166-36:104433:104915 [4] NCCL INFO Channel 09/24 :    0   1
+ip-26-0-166-36:104433:104915 [4] NCCL INFO Channel 10/24 :    0   1
+ip-26-0-166-36:104433:104915 [4] NCCL INFO Channel 11/24 :    0   1
+ip-26-0-166-36:104433:104915 [4] NCCL INFO Channel 12/24 :    0   1
+ip-26-0-166-36:104433:104915 [4] NCCL INFO Channel 13/24 :    0   1
+ip-26-0-166-36:104433:104915 [4] NCCL INFO Channel 14/24 :    0   1
+ip-26-0-166-36:104433:104915 [4] NCCL INFO Channel 15/24 :    0   1
+ip-26-0-166-36:104433:104915 [4] NCCL INFO Channel 16/24 :    0   1
+ip-26-0-166-36:104434:104921 [5] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] -1/-1/-1->1->0 [3] -1/-1/-1->1->0 [4] -1/-1/-1->1->0 [5] -1/-1/-1->1->0 [6] 0/-1/-1->1->-1 [7] 0/-1/-1->1->-1 [8] 0/-1/-1->1->-1 [9] 0/-1/-1->1->-1 [10] 0/-1/-1->1->-1 [11] 0/-1/-1->1->-1 [12] -1/-1/-1->1->0 [13] -1/-1/-1->1->0 [14] -1/-1/-1->1->0 [15] -1/-1/-1->1->0 [16] -1/-1/-1->1->0 [17] -1/-1/-1->1->0 [18] 0/-1/-1->1->-1 [19] 0/-1/-1->1->-1 [20] 0/-1/-1->1->-1 [21] 0/-1/-1->1->-1 [22] 0/-1/-1->1->-1 [23] 0/-1/-1->1->-1
+ip-26-0-166-36:104433:104915 [4] NCCL INFO Channel 17/24 :    0   1
+ip-26-0-166-36:104433:104915 [4] NCCL INFO Channel 18/24 :    0   1
+ip-26-0-166-36:104434:104921 [5] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-166-36:104433:104915 [4] NCCL INFO Channel 19/24 :    0   1
+ip-26-0-166-36:104433:104915 [4] NCCL INFO Channel 20/24 :    0   1
+ip-26-0-166-36:104433:104915 [4] NCCL INFO Channel 21/24 :    0   1
+ip-26-0-166-36:104433:104915 [4] NCCL INFO Channel 22/24 :    0   1
+ip-26-0-166-36:104433:104915 [4] NCCL INFO Channel 23/24 :    0   1
+ip-26-0-166-36:104433:104915 [4] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] -1/-1/-1->0->1 [7] -1/-1/-1->0->1 [8] -1/-1/-1->0->1 [9] -1/-1/-1->0->1 [10] -1/-1/-1->0->1 [11] -1/-1/-1->0->1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] -1/-1/-1->0->1 [19] -1/-1/-1->0->1 [20] -1/-1/-1->0->1 [21] -1/-1/-1->0->1 [22] -1/-1/-1->0->1 [23] -1/-1/-1->0->1
+ip-26-0-166-36:104433:104915 [4] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-166-214:624945:625436 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-166-214:624945:625436 [2] NCCL INFO Channel 00/24 :    0   1
+ip-26-0-166-214:624945:625436 [2] NCCL INFO Channel 01/24 :    0   1
+ip-26-0-166-214:624945:625436 [2] NCCL INFO Channel 02/24 :    0   1
+ip-26-0-166-214:624945:625436 [2] NCCL INFO Channel 03/24 :    0   1
+ip-26-0-166-214:624945:625436 [2] NCCL INFO Channel 04/24 :    0   1
+ip-26-0-166-214:624945:625436 [2] NCCL INFO Channel 05/24 :    0   1
+ip-26-0-166-214:624945:625436 [2] NCCL INFO Channel 06/24 :    0   1
+ip-26-0-166-214:624945:625436 [2] NCCL INFO Channel 07/24 :    0   1
+ip-26-0-166-214:624945:625436 [2] NCCL INFO Channel 08/24 :    0   1
+ip-26-0-166-214:624945:625436 [2] NCCL INFO Channel 09/24 :    0   1
+ip-26-0-166-214:624945:625436 [2] NCCL INFO Channel 10/24 :    0   1
+ip-26-0-166-214:624945:625436 [2] NCCL INFO Channel 11/24 :    0   1
+ip-26-0-166-214:624945:625436 [2] NCCL INFO Channel 12/24 :    0   1
+ip-26-0-166-214:624946:625439 [3] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] -1/-1/-1->1->0 [3] -1/-1/-1->1->0 [4] -1/-1/-1->1->0 [5] -1/-1/-1->1->0 [6] 0/-1/-1->1->-1 [7] 0/-1/-1->1->-1 [8] 0/-1/-1->1->-1 [9] 0/-1/-1->1->-1 [10] 0/-1/-1->1->-1 [11] 0/-1/-1->1->-1 [12] -1/-1/-1->1->0 [13] -1/-1/-1->1->0 [14] -1/-1/-1->1->0 [15] -1/-1/-1->1->0 [16] -1/-1/-1->1->0 [17] -1/-1/-1->1->0 [18] 0/-1/-1->1->-1 [19] 0/-1/-1->1->-1 [20] 0/-1/-1->1->-1 [21] 0/-1/-1->1->-1 [22] 0/-1/-1->1->-1 [23] 0/-1/-1->1->-1
+ip-26-0-166-214:624945:625436 [2] NCCL INFO Channel 13/24 :    0   1
+ip-26-0-166-214:624945:625436 [2] NCCL INFO Channel 14/24 :    0   1
+ip-26-0-166-214:624946:625439 [3] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-166-214:624945:625436 [2] NCCL INFO Channel 15/24 :    0   1
+ip-26-0-166-214:624945:625436 [2] NCCL INFO Channel 16/24 :    0   1
+ip-26-0-166-214:624945:625436 [2] NCCL INFO Channel 17/24 :    0   1
+ip-26-0-166-214:624945:625436 [2] NCCL INFO Channel 18/24 :    0   1
+ip-26-0-166-214:624945:625436 [2] NCCL INFO Channel 19/24 :    0   1
+ip-26-0-166-214:624945:625436 [2] NCCL INFO Channel 20/24 :    0   1
+ip-26-0-166-214:624945:625436 [2] NCCL INFO Channel 21/24 :    0   1
+ip-26-0-166-214:624945:625436 [2] NCCL INFO Channel 22/24 :    0   1
+ip-26-0-166-214:624945:625436 [2] NCCL INFO Channel 23/24 :    0   1
+ip-26-0-166-214:624945:625436 [2] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] -1/-1/-1->0->1 [7] -1/-1/-1->0->1 [8] -1/-1/-1->0->1 [9] -1/-1/-1->0->1 [10] -1/-1/-1->0->1 [11] -1/-1/-1->0->1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] -1/-1/-1->0->1 [19] -1/-1/-1->0->1 [20] -1/-1/-1->0->1 [21] -1/-1/-1->0->1 [22] -1/-1/-1->0->1 [23] -1/-1/-1->0->1
+ip-26-0-166-214:624945:625436 [2] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-165-213:9631:10117 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-165-213:9631:10117 [6] NCCL INFO Channel 00/24 :    0   1
+ip-26-0-165-213:9631:10117 [6] NCCL INFO Channel 01/24 :    0   1
+ip-26-0-165-213:9631:10117 [6] NCCL INFO Channel 02/24 :    0   1
+ip-26-0-165-213:9631:10117 [6] NCCL INFO Channel 03/24 :    0   1
+ip-26-0-165-213:9631:10117 [6] NCCL INFO Channel 04/24 :    0   1
+ip-26-0-165-213:9631:10117 [6] NCCL INFO Channel 05/24 :    0   1
+ip-26-0-165-213:9631:10117 [6] NCCL INFO Channel 06/24 :    0   1
+ip-26-0-165-213:9631:10117 [6] NCCL INFO Channel 07/24 :    0   1
+ip-26-0-165-213:9631:10117 [6] NCCL INFO Channel 08/24 :    0   1
+ip-26-0-165-213:9631:10117 [6] NCCL INFO Channel 09/24 :    0   1
+ip-26-0-165-213:9631:10117 [6] NCCL INFO Channel 10/24 :    0   1
+ip-26-0-165-213:9631:10117 [6] NCCL INFO Channel 11/24 :    0   1
+ip-26-0-165-213:9631:10117 [6] NCCL INFO Channel 12/24 :    0   1
+ip-26-0-165-213:9631:10117 [6] NCCL INFO Channel 13/24 :    0   1
+ip-26-0-165-213:9631:10117 [6] NCCL INFO Channel 14/24 :    0   1
+ip-26-0-165-213:9631:10117 [6] NCCL INFO Channel 15/24 :    0   1
+ip-26-0-165-213:9631:10117 [6] NCCL INFO Channel 16/24 :    0   1
+ip-26-0-165-213:9631:10117 [6] NCCL INFO Channel 17/24 :    0   1
+ip-26-0-165-213:9631:10117 [6] NCCL INFO Channel 18/24 :    0   1
+ip-26-0-165-213:9631:10117 [6] NCCL INFO Channel 19/24 :    0   1
+ip-26-0-165-213:9632:10121 [7] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] -1/-1/-1->1->0 [3] -1/-1/-1->1->0 [4] -1/-1/-1->1->0 [5] -1/-1/-1->1->0 [6] 0/-1/-1->1->-1 [7] 0/-1/-1->1->-1 [8] 0/-1/-1->1->-1 [9] 0/-1/-1->1->-1 [10] 0/-1/-1->1->-1 [11] 0/-1/-1->1->-1 [12] -1/-1/-1->1->0 [13] -1/-1/-1->1->0 [14] -1/-1/-1->1->0 [15] -1/-1/-1->1->0 [16] -1/-1/-1->1->0 [17] -1/-1/-1->1->0 [18] 0/-1/-1->1->-1 [19] 0/-1/-1->1->-1 [20] 0/-1/-1->1->-1 [21] 0/-1/-1->1->-1 [22] 0/-1/-1->1->-1 [23] 0/-1/-1->1->-1
+ip-26-0-165-213:9631:10117 [6] NCCL INFO Channel 20/24 :    0   1
+ip-26-0-165-213:9631:10117 [6] NCCL INFO Channel 21/24 :    0   1
+ip-26-0-165-213:9632:10121 [7] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-165-213:9631:10117 [6] NCCL INFO Channel 22/24 :    0   1
+ip-26-0-165-213:9631:10117 [6] NCCL INFO Channel 23/24 :    0   1
+ip-26-0-165-213:9631:10117 [6] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] -1/-1/-1->0->1 [7] -1/-1/-1->0->1 [8] -1/-1/-1->0->1 [9] -1/-1/-1->0->1 [10] -1/-1/-1->0->1 [11] -1/-1/-1->0->1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] -1/-1/-1->0->1 [19] -1/-1/-1->0->1 [20] -1/-1/-1->0->1 [21] -1/-1/-1->0->1 [22] -1/-1/-1->0->1 [23] -1/-1/-1->0->1
+ip-26-0-165-213:9631:10117 [6] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-166-36:104431:104917 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-162-46:702881:703400 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-166-36:104431:104917 [2] NCCL INFO Channel 00/24 :    0   1
+ip-26-0-166-36:104431:104917 [2] NCCL INFO Channel 01/24 :    0   1
+ip-26-0-166-36:104431:104917 [2] NCCL INFO Channel 02/24 :    0   1
+ip-26-0-166-36:104431:104917 [2] NCCL INFO Channel 03/24 :    0   1
+ip-26-0-166-36:104431:104917 [2] NCCL INFO Channel 04/24 :    0   1
+ip-26-0-166-36:104431:104917 [2] NCCL INFO Channel 05/24 :    0   1
+ip-26-0-166-36:104431:104917 [2] NCCL INFO Channel 06/24 :    0   1
+ip-26-0-166-36:104431:104917 [2] NCCL INFO Channel 07/24 :    0   1
+ip-26-0-166-36:104431:104917 [2] NCCL INFO Channel 08/24 :    0   1
+ip-26-0-166-36:104431:104917 [2] NCCL INFO Channel 09/24 :    0   1
+ip-26-0-166-36:104431:104917 [2] NCCL INFO Channel 10/24 :    0   1
+ip-26-0-166-36:104431:104917 [2] NCCL INFO Channel 11/24 :    0   1
+ip-26-0-166-36:104431:104917 [2] NCCL INFO Channel 12/24 :    0   1
+ip-26-0-166-36:104431:104917 [2] NCCL INFO Channel 13/24 :    0   1
+ip-26-0-162-46:702880:703399 [0] NCCL INFO Channel 00/24 :    0   1
+ip-26-0-162-46:702880:703399 [0] NCCL INFO Channel 01/24 :    0   1
+ip-26-0-162-46:702880:703399 [0] NCCL INFO Channel 02/24 :    0   1
+ip-26-0-162-46:702880:703399 [0] NCCL INFO Channel 03/24 :    0   1
+ip-26-0-162-46:702880:703399 [0] NCCL INFO Channel 04/24 :    0   1
+ip-26-0-162-46:702880:703399 [0] NCCL INFO Channel 05/24 :    0   1
+ip-26-0-162-46:702880:703399 [0] NCCL INFO Channel 06/24 :    0   1
+ip-26-0-162-46:702880:703399 [0] NCCL INFO Channel 07/24 :    0   1
+ip-26-0-162-46:702880:703399 [0] NCCL INFO Channel 08/24 :    0   1
+ip-26-0-162-46:702880:703399 [0] NCCL INFO Channel 09/24 :    0   1
+ip-26-0-166-36:104431:104917 [2] NCCL INFO Channel 14/24 :    0   1
+ip-26-0-166-36:104431:104917 [2] NCCL INFO Channel 15/24 :    0   1
+ip-26-0-166-36:104431:104917 [2] NCCL INFO Channel 16/24 :    0   1
+ip-26-0-166-36:104431:104917 [2] NCCL INFO Channel 17/24 :    0   1
+ip-26-0-166-36:104431:104917 [2] NCCL INFO Channel 18/24 :    0   1
+ip-26-0-166-36:104431:104917 [2] NCCL INFO Channel 19/24 :    0   1
+ip-26-0-166-36:104431:104917 [2] NCCL INFO Channel 20/24 :    0   1
+ip-26-0-162-46:702881:703400 [1] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] -1/-1/-1->1->0 [3] -1/-1/-1->1->0 [4] -1/-1/-1->1->0 [5] -1/-1/-1->1->0 [6] 0/-1/-1->1->-1 [7] 0/-1/-1->1->-1 [8] 0/-1/-1->1->-1 [9] 0/-1/-1->1->-1 [10] 0/-1/-1->1->-1 [11] 0/-1/-1->1->-1 [12] -1/-1/-1->1->0 [13] -1/-1/-1->1->0 [14] -1/-1/-1->1->0 [15] -1/-1/-1->1->0 [16] -1/-1/-1->1->0 [17] -1/-1/-1->1->0 [18] 0/-1/-1->1->-1 [19] 0/-1/-1->1->-1 [20] 0/-1/-1->1->-1 [21] 0/-1/-1->1->-1 [22] 0/-1/-1->1->-1 [23] 0/-1/-1->1->-1
+ip-26-0-162-46:702880:703399 [0] NCCL INFO Channel 10/24 :    0   1
+ip-26-0-162-46:702880:703399 [0] NCCL INFO Channel 11/24 :    0   1
+ip-26-0-162-46:702880:703399 [0] NCCL INFO Channel 12/24 :    0   1
+ip-26-0-162-46:702881:703400 [1] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-162-46:702880:703399 [0] NCCL INFO Channel 13/24 :    0   1
+ip-26-0-162-46:702880:703399 [0] NCCL INFO Channel 14/24 :    0   1
+ip-26-0-162-46:702880:703399 [0] NCCL INFO Channel 15/24 :    0   1
+ip-26-0-166-36:104431:104917 [2] NCCL INFO Channel 21/24 :    0   1
+ip-26-0-166-36:104432:104922 [3] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] -1/-1/-1->1->0 [3] -1/-1/-1->1->0 [4] -1/-1/-1->1->0 [5] -1/-1/-1->1->0 [6] 0/-1/-1->1->-1 [7] 0/-1/-1->1->-1 [8] 0/-1/-1->1->-1 [9] 0/-1/-1->1->-1 [10] 0/-1/-1->1->-1 [11] 0/-1/-1->1->-1 [12] -1/-1/-1->1->0 [13] -1/-1/-1->1->0 [14] -1/-1/-1->1->0 [15] -1/-1/-1->1->0 [16] -1/-1/-1->1->0 [17] -1/-1/-1->1->0 [18] 0/-1/-1->1->-1 [19] 0/-1/-1->1->-1 [20] 0/-1/-1->1->-1 [21] 0/-1/-1->1->-1 [22] 0/-1/-1->1->-1 [23] 0/-1/-1->1->-1
+ip-26-0-166-36:104431:104917 [2] NCCL INFO Channel 22/24 :    0   1
+ip-26-0-166-36:104431:104917 [2] NCCL INFO Channel 23/24 :    0   1
+ip-26-0-166-36:104432:104922 [3] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-162-46:702880:703399 [0] NCCL INFO Channel 16/24 :    0   1
+ip-26-0-166-36:104431:104917 [2] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] -1/-1/-1->0->1 [7] -1/-1/-1->0->1 [8] -1/-1/-1->0->1 [9] -1/-1/-1->0->1 [10] -1/-1/-1->0->1 [11] -1/-1/-1->0->1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] -1/-1/-1->0->1 [19] -1/-1/-1->0->1 [20] -1/-1/-1->0->1 [21] -1/-1/-1->0->1 [22] -1/-1/-1->0->1 [23] -1/-1/-1->0->1
+ip-26-0-166-36:104431:104917 [2] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-162-46:702880:703399 [0] NCCL INFO Channel 17/24 :    0   1
+ip-26-0-162-46:702880:703399 [0] NCCL INFO Channel 18/24 :    0   1
+ip-26-0-162-46:702880:703399 [0] NCCL INFO Channel 19/24 :    0   1
+ip-26-0-162-46:702880:703399 [0] NCCL INFO Channel 20/24 :    0   1
+ip-26-0-162-46:702880:703399 [0] NCCL INFO Channel 21/24 :    0   1
+ip-26-0-162-46:702880:703399 [0] NCCL INFO Channel 22/24 :    0   1
+ip-26-0-162-46:702880:703399 [0] NCCL INFO Channel 23/24 :    0   1
+ip-26-0-162-46:702880:703399 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] -1/-1/-1->0->1 [7] -1/-1/-1->0->1 [8] -1/-1/-1->0->1 [9] -1/-1/-1->0->1 [10] -1/-1/-1->0->1 [11] -1/-1/-1->0->1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] -1/-1/-1->0->1 [19] -1/-1/-1->0->1 [20] -1/-1/-1->0->1 [21] -1/-1/-1->0->1 [22] -1/-1/-1->0->1 [23] -1/-1/-1->0->1
+ip-26-0-162-46:702880:703399 [0] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-165-213:9630:10119 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-166-15:127294:127779 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-166-15:127294:127779 [4] NCCL INFO Channel 00/24 :    0   1
+ip-26-0-166-15:127294:127779 [4] NCCL INFO Channel 01/24 :    0   1
+ip-26-0-166-15:127294:127779 [4] NCCL INFO Channel 02/24 :    0   1
+ip-26-0-166-15:127294:127779 [4] NCCL INFO Channel 03/24 :    0   1
+ip-26-0-166-15:127294:127779 [4] NCCL INFO Channel 04/24 :    0   1
+ip-26-0-166-15:127294:127779 [4] NCCL INFO Channel 05/24 :    0   1
+ip-26-0-166-15:127294:127779 [4] NCCL INFO Channel 06/24 :    0   1
+ip-26-0-166-15:127295:127783 [5] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] -1/-1/-1->1->0 [3] -1/-1/-1->1->0 [4] -1/-1/-1->1->0 [5] -1/-1/-1->1->0 [6] 0/-1/-1->1->-1 [7] 0/-1/-1->1->-1 [8] 0/-1/-1->1->-1 [9] 0/-1/-1->1->-1 [10] 0/-1/-1->1->-1 [11] 0/-1/-1->1->-1 [12] -1/-1/-1->1->0 [13] -1/-1/-1->1->0 [14] -1/-1/-1->1->0 [15] -1/-1/-1->1->0 [16] -1/-1/-1->1->0 [17] -1/-1/-1->1->0 [18] 0/-1/-1->1->-1 [19] 0/-1/-1->1->-1 [20] 0/-1/-1->1->-1 [21] 0/-1/-1->1->-1 [22] 0/-1/-1->1->-1 [23] 0/-1/-1->1->-1
+ip-26-0-166-15:127294:127779 [4] NCCL INFO Channel 07/24 :    0   1
+ip-26-0-166-15:127294:127779 [4] NCCL INFO Channel 08/24 :    0   1
+ip-26-0-166-15:127295:127783 [5] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-166-15:127294:127779 [4] NCCL INFO Channel 09/24 :    0   1
+ip-26-0-166-15:127294:127779 [4] NCCL INFO Channel 10/24 :    0   1
+ip-26-0-166-15:127294:127779 [4] NCCL INFO Channel 11/24 :    0   1
+ip-26-0-166-15:127294:127779 [4] NCCL INFO Channel 12/24 :    0   1
+ip-26-0-166-15:127294:127779 [4] NCCL INFO Channel 13/24 :    0   1
+ip-26-0-166-15:127294:127779 [4] NCCL INFO Channel 14/24 :    0   1
+ip-26-0-166-15:127294:127779 [4] NCCL INFO Channel 15/24 :    0   1
+ip-26-0-166-15:127294:127779 [4] NCCL INFO Channel 16/24 :    0   1
+ip-26-0-166-15:127294:127779 [4] NCCL INFO Channel 17/24 :    0   1
+ip-26-0-166-15:127294:127779 [4] NCCL INFO Channel 18/24 :    0   1
+ip-26-0-166-15:127294:127779 [4] NCCL INFO Channel 19/24 :    0   1
+ip-26-0-166-15:127294:127779 [4] NCCL INFO Channel 20/24 :    0   1
+ip-26-0-166-15:127294:127779 [4] NCCL INFO Channel 21/24 :    0   1
+ip-26-0-166-15:127294:127779 [4] NCCL INFO Channel 22/24 :    0   1
+ip-26-0-166-15:127294:127779 [4] NCCL INFO Channel 23/24 :    0   1
+ip-26-0-166-15:127294:127779 [4] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] -1/-1/-1->0->1 [7] -1/-1/-1->0->1 [8] -1/-1/-1->0->1 [9] -1/-1/-1->0->1 [10] -1/-1/-1->0->1 [11] -1/-1/-1->0->1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] -1/-1/-1->0->1 [19] -1/-1/-1->0->1 [20] -1/-1/-1->0->1 [21] -1/-1/-1->0->1 [22] -1/-1/-1->0->1 [23] -1/-1/-1->0->1
+ip-26-0-166-15:127294:127779 [4] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-165-213:9629:10113 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-165-213:9629:10113 [4] NCCL INFO Channel 00/24 :    0   1
+ip-26-0-165-213:9629:10113 [4] NCCL INFO Channel 01/24 :    0   1
+ip-26-0-165-213:9629:10113 [4] NCCL INFO Channel 02/24 :    0   1
+ip-26-0-165-213:9629:10113 [4] NCCL INFO Channel 03/24 :    0   1
+ip-26-0-165-213:9629:10113 [4] NCCL INFO Channel 04/24 :    0   1
+ip-26-0-165-213:9629:10113 [4] NCCL INFO Channel 05/24 :    0   1
+ip-26-0-165-213:9629:10113 [4] NCCL INFO Channel 06/24 :    0   1
+ip-26-0-165-213:9630:10119 [5] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] -1/-1/-1->1->0 [3] -1/-1/-1->1->0 [4] -1/-1/-1->1->0 [5] -1/-1/-1->1->0 [6] 0/-1/-1->1->-1 [7] 0/-1/-1->1->-1 [8] 0/-1/-1->1->-1 [9] 0/-1/-1->1->-1 [10] 0/-1/-1->1->-1 [11] 0/-1/-1->1->-1 [12] -1/-1/-1->1->0 [13] -1/-1/-1->1->0 [14] -1/-1/-1->1->0 [15] -1/-1/-1->1->0 [16] -1/-1/-1->1->0 [17] -1/-1/-1->1->0 [18] 0/-1/-1->1->-1 [19] 0/-1/-1->1->-1 [20] 0/-1/-1->1->-1 [21] 0/-1/-1->1->-1 [22] 0/-1/-1->1->-1 [23] 0/-1/-1->1->-1
+ip-26-0-165-213:9629:10113 [4] NCCL INFO Channel 07/24 :    0   1
+ip-26-0-165-213:9630:10119 [5] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-165-213:9629:10113 [4] NCCL INFO Channel 08/24 :    0   1
+ip-26-0-165-213:9629:10113 [4] NCCL INFO Channel 09/24 :    0   1
+ip-26-0-165-213:9629:10113 [4] NCCL INFO Channel 10/24 :    0   1
+ip-26-0-165-213:9629:10113 [4] NCCL INFO Channel 11/24 :    0   1
+ip-26-0-165-213:9629:10113 [4] NCCL INFO Channel 12/24 :    0   1
+ip-26-0-165-213:9629:10113 [4] NCCL INFO Channel 13/24 :    0   1
+ip-26-0-165-213:9629:10113 [4] NCCL INFO Channel 14/24 :    0   1
+ip-26-0-165-213:9629:10113 [4] NCCL INFO Channel 15/24 :    0   1
+ip-26-0-165-213:9629:10113 [4] NCCL INFO Channel 16/24 :    0   1
+ip-26-0-165-213:9629:10113 [4] NCCL INFO Channel 17/24 :    0   1
+ip-26-0-165-213:9629:10113 [4] NCCL INFO Channel 18/24 :    0   1
+ip-26-0-165-213:9629:10113 [4] NCCL INFO Channel 19/24 :    0   1
+ip-26-0-165-213:9629:10113 [4] NCCL INFO Channel 20/24 :    0   1
+ip-26-0-165-213:9629:10113 [4] NCCL INFO Channel 21/24 :    0   1
+ip-26-0-165-213:9629:10113 [4] NCCL INFO Channel 22/24 :    0   1
+ip-26-0-165-213:9629:10113 [4] NCCL INFO Channel 23/24 :    0   1
+ip-26-0-165-213:9629:10113 [4] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] -1/-1/-1->0->1 [7] -1/-1/-1->0->1 [8] -1/-1/-1->0->1 [9] -1/-1/-1->0->1 [10] -1/-1/-1->0->1 [11] -1/-1/-1->0->1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] -1/-1/-1->0->1 [19] -1/-1/-1->0->1 [20] -1/-1/-1->0->1 [21] -1/-1/-1->0->1 [22] -1/-1/-1->0->1 [23] -1/-1/-1->0->1
+ip-26-0-165-213:9629:10113 [4] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-167-9:1039538:1040044 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-167-9:1039537:1040038 [0] NCCL INFO Channel 00/24 :    0   1
+ip-26-0-167-9:1039537:1040038 [0] NCCL INFO Channel 01/24 :    0   1
+ip-26-0-167-9:1039537:1040038 [0] NCCL INFO Channel 02/24 :    0   1
+ip-26-0-167-9:1039537:1040038 [0] NCCL INFO Channel 03/24 :    0   1
+ip-26-0-167-9:1039537:1040038 [0] NCCL INFO Channel 04/24 :    0   1
+ip-26-0-167-9:1039537:1040038 [0] NCCL INFO Channel 05/24 :    0   1
+ip-26-0-167-9:1039537:1040038 [0] NCCL INFO Channel 06/24 :    0   1
+ip-26-0-167-9:1039537:1040038 [0] NCCL INFO Channel 07/24 :    0   1
+ip-26-0-167-9:1039537:1040038 [0] NCCL INFO Channel 08/24 :    0   1
+ip-26-0-167-9:1039537:1040038 [0] NCCL INFO Channel 09/24 :    0   1
+ip-26-0-167-9:1039537:1040038 [0] NCCL INFO Channel 10/24 :    0   1
+ip-26-0-167-9:1039537:1040038 [0] NCCL INFO Channel 11/24 :    0   1
+ip-26-0-167-9:1039537:1040038 [0] NCCL INFO Channel 12/24 :    0   1
+ip-26-0-167-9:1039537:1040038 [0] NCCL INFO Channel 13/24 :    0   1
+ip-26-0-167-9:1039537:1040038 [0] NCCL INFO Channel 14/24 :    0   1
+ip-26-0-167-9:1039537:1040038 [0] NCCL INFO Channel 15/24 :    0   1
+ip-26-0-167-9:1039537:1040038 [0] NCCL INFO Channel 16/24 :    0   1
+ip-26-0-167-9:1039537:1040038 [0] NCCL INFO Channel 17/24 :    0   1
+ip-26-0-167-9:1039537:1040038 [0] NCCL INFO Channel 18/24 :    0   1
+ip-26-0-167-9:1039538:1040044 [1] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] -1/-1/-1->1->0 [3] -1/-1/-1->1->0 [4] -1/-1/-1->1->0 [5] -1/-1/-1->1->0 [6] 0/-1/-1->1->-1 [7] 0/-1/-1->1->-1 [8] 0/-1/-1->1->-1 [9] 0/-1/-1->1->-1 [10] 0/-1/-1->1->-1 [11] 0/-1/-1->1->-1 [12] -1/-1/-1->1->0 [13] -1/-1/-1->1->0 [14] -1/-1/-1->1->0 [15] -1/-1/-1->1->0 [16] -1/-1/-1->1->0 [17] -1/-1/-1->1->0 [18] 0/-1/-1->1->-1 [19] 0/-1/-1->1->-1 [20] 0/-1/-1->1->-1 [21] 0/-1/-1->1->-1 [22] 0/-1/-1->1->-1 [23] 0/-1/-1->1->-1
+ip-26-0-167-9:1039537:1040038 [0] NCCL INFO Channel 19/24 :    0   1
+ip-26-0-167-9:1039537:1040038 [0] NCCL INFO Channel 20/24 :    0   1
+ip-26-0-167-9:1039538:1040044 [1] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-167-9:1039537:1040038 [0] NCCL INFO Channel 21/24 :    0   1
+ip-26-0-167-9:1039537:1040038 [0] NCCL INFO Channel 22/24 :    0   1
+ip-26-0-167-9:1039537:1040038 [0] NCCL INFO Channel 23/24 :    0   1
+ip-26-0-167-9:1039537:1040038 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] -1/-1/-1->0->1 [7] -1/-1/-1->0->1 [8] -1/-1/-1->0->1 [9] -1/-1/-1->0->1 [10] -1/-1/-1->0->1 [11] -1/-1/-1->0->1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] -1/-1/-1->0->1 [19] -1/-1/-1->0->1 [20] -1/-1/-1->0->1 [21] -1/-1/-1->0->1 [22] -1/-1/-1->0->1 [23] -1/-1/-1->0->1
+ip-26-0-167-9:1039537:1040038 [0] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-166-15:127296:127778 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-166-15:127296:127778 [6] NCCL INFO Channel 00/24 :    0   1
+ip-26-0-166-15:127296:127778 [6] NCCL INFO Channel 01/24 :    0   1
+ip-26-0-166-15:127296:127778 [6] NCCL INFO Channel 02/24 :    0   1
+ip-26-0-166-15:127296:127778 [6] NCCL INFO Channel 03/24 :    0   1
+ip-26-0-166-15:127296:127778 [6] NCCL INFO Channel 04/24 :    0   1
+ip-26-0-166-15:127296:127778 [6] NCCL INFO Channel 05/24 :    0   1
+ip-26-0-166-15:127296:127778 [6] NCCL INFO Channel 06/24 :    0   1
+ip-26-0-166-15:127296:127778 [6] NCCL INFO Channel 07/24 :    0   1
+ip-26-0-166-15:127296:127778 [6] NCCL INFO Channel 08/24 :    0   1
+ip-26-0-166-15:127296:127778 [6] NCCL INFO Channel 09/24 :    0   1
+ip-26-0-166-15:127296:127778 [6] NCCL INFO Channel 10/24 :    0   1
+ip-26-0-166-15:127296:127778 [6] NCCL INFO Channel 11/24 :    0   1
+ip-26-0-166-15:127296:127778 [6] NCCL INFO Channel 12/24 :    0   1
+ip-26-0-166-15:127296:127778 [6] NCCL INFO Channel 13/24 :    0   1
+ip-26-0-166-15:127296:127778 [6] NCCL INFO Channel 14/24 :    0   1
+ip-26-0-166-15:127296:127778 [6] NCCL INFO Channel 15/24 :    0   1
+ip-26-0-166-15:127296:127778 [6] NCCL INFO Channel 16/24 :    0   1
+ip-26-0-166-15:127296:127778 [6] NCCL INFO Channel 17/24 :    0   1
+ip-26-0-166-15:127296:127778 [6] NCCL INFO Channel 18/24 :    0   1
+ip-26-0-166-15:127296:127778 [6] NCCL INFO Channel 19/24 :    0   1
+ip-26-0-166-15:127297:127781 [7] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] -1/-1/-1->1->0 [3] -1/-1/-1->1->0 [4] -1/-1/-1->1->0 [5] -1/-1/-1->1->0 [6] 0/-1/-1->1->-1 [7] 0/-1/-1->1->-1 [8] 0/-1/-1->1->-1 [9] 0/-1/-1->1->-1 [10] 0/-1/-1->1->-1 [11] 0/-1/-1->1->-1 [12] -1/-1/-1->1->0 [13] -1/-1/-1->1->0 [14] -1/-1/-1->1->0 [15] -1/-1/-1->1->0 [16] -1/-1/-1->1->0 [17] -1/-1/-1->1->0 [18] 0/-1/-1->1->-1 [19] 0/-1/-1->1->-1 [20] 0/-1/-1->1->-1 [21] 0/-1/-1->1->-1 [22] 0/-1/-1->1->-1 [23] 0/-1/-1->1->-1
+ip-26-0-166-15:127296:127778 [6] NCCL INFO Channel 20/24 :    0   1
+ip-26-0-166-15:127296:127778 [6] NCCL INFO Channel 21/24 :    0   1
+ip-26-0-166-15:127297:127781 [7] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-166-15:127296:127778 [6] NCCL INFO Channel 22/24 :    0   1
+ip-26-0-166-15:127296:127778 [6] NCCL INFO Channel 23/24 :    0   1
+ip-26-0-166-15:127296:127778 [6] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] -1/-1/-1->0->1 [7] -1/-1/-1->0->1 [8] -1/-1/-1->0->1 [9] -1/-1/-1->0->1 [10] -1/-1/-1->0->1 [11] -1/-1/-1->0->1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] -1/-1/-1->0->1 [19] -1/-1/-1->0->1 [20] -1/-1/-1->0->1 [21] -1/-1/-1->0->1 [22] -1/-1/-1->0->1 [23] -1/-1/-1->0->1
+ip-26-0-166-15:127296:127778 [6] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-166-36:104435:104920 [6] NCCL INFO Channel 00/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-167-9:1039544:1040046 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-167-9:1039543:1040043 [6] NCCL INFO Channel 00/24 :    0   1
+ip-26-0-167-9:1039543:1040043 [6] NCCL INFO Channel 01/24 :    0   1
+ip-26-0-167-9:1039543:1040043 [6] NCCL INFO Channel 02/24 :    0   1
+ip-26-0-167-9:1039543:1040043 [6] NCCL INFO Channel 03/24 :    0   1
+ip-26-0-167-9:1039543:1040043 [6] NCCL INFO Channel 04/24 :    0   1
+ip-26-0-167-9:1039543:1040043 [6] NCCL INFO Channel 05/24 :    0   1
+ip-26-0-167-9:1039543:1040043 [6] NCCL INFO Channel 06/24 :    0   1
+ip-26-0-167-9:1039543:1040043 [6] NCCL INFO Channel 07/24 :    0   1
+ip-26-0-167-9:1039543:1040043 [6] NCCL INFO Channel 08/24 :    0   1
+ip-26-0-167-9:1039544:1040046 [7] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] -1/-1/-1->1->0 [3] -1/-1/-1->1->0 [4] -1/-1/-1->1->0 [5] -1/-1/-1->1->0 [6] 0/-1/-1->1->-1 [7] 0/-1/-1->1->-1 [8] 0/-1/-1->1->-1 [9] 0/-1/-1->1->-1 [10] 0/-1/-1->1->-1 [11] 0/-1/-1->1->-1 [12] -1/-1/-1->1->0 [13] -1/-1/-1->1->0 [14] -1/-1/-1->1->0 [15] -1/-1/-1->1->0 [16] -1/-1/-1->1->0 [17] -1/-1/-1->1->0 [18] 0/-1/-1->1->-1 [19] 0/-1/-1->1->-1 [20] 0/-1/-1->1->-1 [21] 0/-1/-1->1->-1 [22] 0/-1/-1->1->-1 [23] 0/-1/-1->1->-1
+ip-26-0-167-9:1039543:1040043 [6] NCCL INFO Channel 09/24 :    0   1
+ip-26-0-167-9:1039543:1040043 [6] NCCL INFO Channel 10/24 :    0   1
+ip-26-0-167-9:1039544:1040046 [7] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-167-9:1039543:1040043 [6] NCCL INFO Channel 11/24 :    0   1
+ip-26-0-167-9:1039543:1040043 [6] NCCL INFO Channel 12/24 :    0   1
+ip-26-0-167-9:1039543:1040043 [6] NCCL INFO Channel 13/24 :    0   1
+ip-26-0-167-9:1039543:1040043 [6] NCCL INFO Channel 14/24 :    0   1
+ip-26-0-167-9:1039543:1040043 [6] NCCL INFO Channel 15/24 :    0   1
+ip-26-0-167-9:1039543:1040043 [6] NCCL INFO Channel 16/24 :    0   1
+ip-26-0-167-9:1039543:1040043 [6] NCCL INFO Channel 17/24 :    0   1
+ip-26-0-167-9:1039543:1040043 [6] NCCL INFO Channel 18/24 :    0   1
+ip-26-0-167-9:1039543:1040043 [6] NCCL INFO Channel 19/24 :    0   1
+ip-26-0-167-9:1039543:1040043 [6] NCCL INFO Channel 20/24 :    0   1
+ip-26-0-167-9:1039543:1040043 [6] NCCL INFO Channel 21/24 :    0   1
+ip-26-0-167-9:1039543:1040043 [6] NCCL INFO Channel 22/24 :    0   1
+ip-26-0-167-9:1039543:1040043 [6] NCCL INFO Channel 23/24 :    0   1
+ip-26-0-167-9:1039543:1040043 [6] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] -1/-1/-1->0->1 [7] -1/-1/-1->0->1 [8] -1/-1/-1->0->1 [9] -1/-1/-1->0->1 [10] -1/-1/-1->0->1 [11] -1/-1/-1->0->1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] -1/-1/-1->0->1 [19] -1/-1/-1->0->1 [20] -1/-1/-1->0->1 [21] -1/-1/-1->0->1 [22] -1/-1/-1->0->1 [23] -1/-1/-1->0->1
+ip-26-0-167-9:1039543:1040043 [6] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-166-36:104436:104923 [7] NCCL INFO Channel 00/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-166-36:104435:104920 [6] NCCL INFO Channel 01/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-166-36:104436:104923 [7] NCCL INFO Channel 01/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-166-36:104435:104920 [6] NCCL INFO Channel 02/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-162-46:702883:703409 [3] NCCL INFO Channel 00/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-166-125:762816:763324 [1] NCCL INFO Setting affinity for GPU 1 to 0fff,ffffffff
+ip-26-0-166-36:104436:104923 [7] NCCL INFO Channel 02/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-166-36:104435:104920 [6] NCCL INFO Channel 03/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-166-36:104429:104919 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-125:762822:763323 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,fff00000,00000000
+ip-26-0-166-125:762821:763318 [6] NCCL INFO Channel 00/24 :    0   1
+ip-26-0-166-125:762821:763318 [6] NCCL INFO Channel 01/24 :    0   1
+ip-26-0-162-46:702883:703409 [3] NCCL INFO Channel 01/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-166-125:762821:763318 [6] NCCL INFO Channel 02/24 :    0   1
+ip-26-0-166-125:762821:763318 [6] NCCL INFO Channel 03/24 :    0   1
+ip-26-0-166-125:762821:763318 [6] NCCL INFO Channel 04/24 :    0   1
+ip-26-0-166-125:762821:763318 [6] NCCL INFO Channel 05/24 :    0   1
+ip-26-0-166-125:762821:763318 [6] NCCL INFO Channel 06/24 :    0   1
+ip-26-0-166-125:762821:763318 [6] NCCL INFO Channel 07/24 :    0   1
+ip-26-0-166-125:762822:763323 [7] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] -1/-1/-1->1->0 [3] -1/-1/-1->1->0 [4] -1/-1/-1->1->0 [5] -1/-1/-1->1->0 [6] 0/-1/-1->1->-1 [7] 0/-1/-1->1->-1 [8] 0/-1/-1->1->-1 [9] 0/-1/-1->1->-1 [10] 0/-1/-1->1->-1 [11] 0/-1/-1->1->-1 [12] -1/-1/-1->1->0 [13] -1/-1/-1->1->0 [14] -1/-1/-1->1->0 [15] -1/-1/-1->1->0 [16] -1/-1/-1->1->0 [17] -1/-1/-1->1->0 [18] 0/-1/-1->1->-1 [19] 0/-1/-1->1->-1 [20] 0/-1/-1->1->-1 [21] 0/-1/-1->1->-1 [22] 0/-1/-1->1->-1 [23] 0/-1/-1->1->-1
+ip-26-0-166-125:762821:763318 [6] NCCL INFO Channel 08/24 :    0   1
+ip-26-0-166-125:762822:763323 [7] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-166-125:762821:763318 [6] NCCL INFO Channel 09/24 :    0   1
+ip-26-0-166-125:762821:763318 [6] NCCL INFO Channel 10/24 :    0   1
+ip-26-0-166-125:762821:763318 [6] NCCL INFO Channel 11/24 :    0   1
+ip-26-0-166-125:762821:763318 [6] NCCL INFO Channel 12/24 :    0   1
+ip-26-0-166-125:762821:763318 [6] NCCL INFO Channel 13/24 :    0   1
+ip-26-0-166-125:762821:763318 [6] NCCL INFO Channel 14/24 :    0   1
+ip-26-0-166-125:762821:763318 [6] NCCL INFO Channel 15/24 :    0   1
+ip-26-0-166-125:762821:763318 [6] NCCL INFO Channel 16/24 :    0   1
+ip-26-0-166-125:762821:763318 [6] NCCL INFO Channel 17/24 :    0   1
+ip-26-0-166-125:762821:763318 [6] NCCL INFO Channel 18/24 :    0   1
+ip-26-0-166-125:762821:763318 [6] NCCL INFO Channel 19/24 :    0   1
+ip-26-0-166-125:762821:763318 [6] NCCL INFO Channel 20/24 :    0   1
+ip-26-0-166-125:762821:763318 [6] NCCL INFO Channel 21/24 :    0   1
+ip-26-0-166-125:762821:763318 [6] NCCL INFO Channel 22/24 :    0   1
+ip-26-0-166-125:762821:763318 [6] NCCL INFO Channel 23/24 :    0   1
+ip-26-0-166-125:762821:763318 [6] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] -1/-1/-1->0->1 [7] -1/-1/-1->0->1 [8] -1/-1/-1->0->1 [9] -1/-1/-1->0->1 [10] -1/-1/-1->0->1 [11] -1/-1/-1->0->1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] -1/-1/-1->0->1 [19] -1/-1/-1->0->1 [20] -1/-1/-1->0->1 [21] -1/-1/-1->0->1 [22] -1/-1/-1->0->1 [23] -1/-1/-1->0->1
+ip-26-0-166-125:762821:763318 [6] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-166-125:762815:763320 [0] NCCL INFO Channel 00/24 :    0   1
+ip-26-0-166-125:762815:763320 [0] NCCL INFO Channel 01/24 :    0   1
+ip-26-0-166-125:762815:763320 [0] NCCL INFO Channel 02/24 :    0   1
+ip-26-0-166-125:762815:763320 [0] NCCL INFO Channel 03/24 :    0   1
+ip-26-0-166-125:762815:763320 [0] NCCL INFO Channel 04/24 :    0   1
+ip-26-0-166-125:762815:763320 [0] NCCL INFO Channel 05/24 :    0   1
+ip-26-0-166-125:762816:763324 [1] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] -1/-1/-1->1->0 [3] -1/-1/-1->1->0 [4] -1/-1/-1->1->0 [5] -1/-1/-1->1->0 [6] 0/-1/-1->1->-1 [7] 0/-1/-1->1->-1 [8] 0/-1/-1->1->-1 [9] 0/-1/-1->1->-1 [10] 0/-1/-1->1->-1 [11] 0/-1/-1->1->-1 [12] -1/-1/-1->1->0 [13] -1/-1/-1->1->0 [14] -1/-1/-1->1->0 [15] -1/-1/-1->1->0 [16] -1/-1/-1->1->0 [17] -1/-1/-1->1->0 [18] 0/-1/-1->1->-1 [19] 0/-1/-1->1->-1 [20] 0/-1/-1->1->-1 [21] 0/-1/-1->1->-1 [22] 0/-1/-1->1->-1 [23] 0/-1/-1->1->-1
+ip-26-0-166-125:762815:763320 [0] NCCL INFO Channel 06/24 :    0   1
+ip-26-0-166-125:762815:763320 [0] NCCL INFO Channel 07/24 :    0   1
+ip-26-0-166-125:762816:763324 [1] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-166-125:762815:763320 [0] NCCL INFO Channel 08/24 :    0   1
+ip-26-0-166-125:762815:763320 [0] NCCL INFO Channel 09/24 :    0   1
+ip-26-0-166-125:762815:763320 [0] NCCL INFO Channel 10/24 :    0   1
+ip-26-0-166-125:762815:763320 [0] NCCL INFO Channel 11/24 :    0   1
+ip-26-0-166-125:762815:763320 [0] NCCL INFO Channel 12/24 :    0   1
+ip-26-0-166-125:762815:763320 [0] NCCL INFO Channel 13/24 :    0   1
+ip-26-0-166-125:762815:763320 [0] NCCL INFO Channel 14/24 :    0   1
+ip-26-0-166-125:762815:763320 [0] NCCL INFO Channel 15/24 :    0   1
+ip-26-0-166-125:762815:763320 [0] NCCL INFO Channel 16/24 :    0   1
+ip-26-0-162-46:702884:703402 [4] NCCL INFO Channel 00/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-166-125:762815:763320 [0] NCCL INFO Channel 17/24 :    0   1
+ip-26-0-166-125:762815:763320 [0] NCCL INFO Channel 18/24 :    0   1
+ip-26-0-166-125:762815:763320 [0] NCCL INFO Channel 19/24 :    0   1
+ip-26-0-166-125:762815:763320 [0] NCCL INFO Channel 20/24 :    0   1
+ip-26-0-166-36:104436:104923 [7] NCCL INFO Channel 03/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-166-125:762815:763320 [0] NCCL INFO Channel 21/24 :    0   1
+ip-26-0-166-125:762815:763320 [0] NCCL INFO Channel 22/24 :    0   1
+ip-26-0-166-125:762815:763320 [0] NCCL INFO Channel 23/24 :    0   1
+ip-26-0-166-125:762815:763320 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] -1/-1/-1->0->1 [7] -1/-1/-1->0->1 [8] -1/-1/-1->0->1 [9] -1/-1/-1->0->1 [10] -1/-1/-1->0->1 [11] -1/-1/-1->0->1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] -1/-1/-1->0->1 [19] -1/-1/-1->0->1 [20] -1/-1/-1->0->1 [21] -1/-1/-1->0->1 [22] -1/-1/-1->0->1 [23] -1/-1/-1->0->1
+ip-26-0-166-125:762815:763320 [0] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-166-36:104435:104920 [6] NCCL INFO Channel 04/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-167-9:1039541:1040041 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-166-36:104429:104919 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-167-9:1039541:1040041 [4] NCCL INFO Channel 00/24 :    0   1
+ip-26-0-167-9:1039541:1040041 [4] NCCL INFO Channel 01/24 :    0   1
+ip-26-0-167-9:1039541:1040041 [4] NCCL INFO Channel 02/24 :    0   1
+ip-26-0-167-9:1039541:1040041 [4] NCCL INFO Channel 03/24 :    0   1
+ip-26-0-167-9:1039541:1040041 [4] NCCL INFO Channel 04/24 :    0   1
+ip-26-0-167-9:1039541:1040041 [4] NCCL INFO Channel 05/24 :    0   1
+ip-26-0-167-9:1039541:1040041 [4] NCCL INFO Channel 06/24 :    0   1
+ip-26-0-167-9:1039541:1040041 [4] NCCL INFO Channel 07/24 :    0   1
+ip-26-0-167-9:1039541:1040041 [4] NCCL INFO Channel 08/24 :    0   1
+ip-26-0-167-9:1039541:1040041 [4] NCCL INFO Channel 09/24 :    0   1
+ip-26-0-167-9:1039541:1040041 [4] NCCL INFO Channel 10/24 :    0   1
+ip-26-0-167-9:1039541:1040041 [4] NCCL INFO Channel 11/24 :    0   1
+ip-26-0-167-9:1039541:1040041 [4] NCCL INFO Channel 12/24 :    0   1
+ip-26-0-167-9:1039541:1040041 [4] NCCL INFO Channel 13/24 :    0   1
+ip-26-0-167-9:1039541:1040041 [4] NCCL INFO Channel 14/24 :    0   1
+ip-26-0-167-9:1039541:1040041 [4] NCCL INFO Channel 15/24 :    0   1
+ip-26-0-167-9:1039541:1040041 [4] NCCL INFO Channel 16/24 :    0   1
+ip-26-0-167-9:1039541:1040041 [4] NCCL INFO Channel 17/24 :    0   1
+ip-26-0-167-9:1039541:1040041 [4] NCCL INFO Channel 18/24 :    0   1
+ip-26-0-167-9:1039541:1040041 [4] NCCL INFO Channel 19/24 :    0   1
+ip-26-0-167-9:1039541:1040041 [4] NCCL INFO Channel 20/24 :    0   1
+ip-26-0-167-9:1039541:1040041 [4] NCCL INFO Channel 21/24 :    0   1
+ip-26-0-167-9:1039541:1040041 [4] NCCL INFO Channel 22/24 :    0   1
+ip-26-0-167-9:1039541:1040041 [4] NCCL INFO Channel 23/24 :    0   1
+ip-26-0-167-9:1039542:1040047 [5] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] -1/-1/-1->1->0 [3] -1/-1/-1->1->0 [4] -1/-1/-1->1->0 [5] -1/-1/-1->1->0 [6] 0/-1/-1->1->-1 [7] 0/-1/-1->1->-1 [8] 0/-1/-1->1->-1 [9] 0/-1/-1->1->-1 [10] 0/-1/-1->1->-1 [11] 0/-1/-1->1->-1 [12] -1/-1/-1->1->0 [13] -1/-1/-1->1->0 [14] -1/-1/-1->1->0 [15] -1/-1/-1->1->0 [16] -1/-1/-1->1->0 [17] -1/-1/-1->1->0 [18] 0/-1/-1->1->-1 [19] 0/-1/-1->1->-1 [20] 0/-1/-1->1->-1 [21] 0/-1/-1->1->-1 [22] 0/-1/-1->1->-1 [23] 0/-1/-1->1->-1
+ip-26-0-167-9:1039542:1040047 [5] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-167-9:1039541:1040041 [4] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] -1/-1/-1->0->1 [7] -1/-1/-1->0->1 [8] -1/-1/-1->0->1 [9] -1/-1/-1->0->1 [10] -1/-1/-1->0->1 [11] -1/-1/-1->0->1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] -1/-1/-1->0->1 [19] -1/-1/-1->0->1 [20] -1/-1/-1->0->1 [21] -1/-1/-1->0->1 [22] -1/-1/-1->0->1 [23] -1/-1/-1->0->1
+ip-26-0-167-9:1039541:1040041 [4] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-162-46:702883:703409 [3] NCCL INFO Channel 02/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-166-36:104436:104923 [7] NCCL INFO Channel 04/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-166-36:104435:104920 [6] NCCL INFO Channel 05/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-166-36:104429:104919 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-162-46:702884:703402 [4] NCCL INFO Channel 01/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-166-36:104430:104924 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-36:104436:104923 [7] NCCL INFO Channel 05/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-166-36:104435:104920 [6] NCCL INFO Channel 06/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-162-46:702883:703409 [3] NCCL INFO Channel 03/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-166-36:104429:104919 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-125:762820:763322 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,fff00000,00000000
+ip-26-0-162-46:702887:703406 [7] NCCL INFO Channel 00/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-162-46:702884:703402 [4] NCCL INFO Channel 02/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-166-125:762819:763316 [4] NCCL INFO Channel 00/24 :    0   1
+ip-26-0-166-125:762819:763316 [4] NCCL INFO Channel 01/24 :    0   1
+ip-26-0-166-125:762820:763322 [5] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] -1/-1/-1->1->0 [3] -1/-1/-1->1->0 [4] -1/-1/-1->1->0 [5] -1/-1/-1->1->0 [6] 0/-1/-1->1->-1 [7] 0/-1/-1->1->-1 [8] 0/-1/-1->1->-1 [9] 0/-1/-1->1->-1 [10] 0/-1/-1->1->-1 [11] 0/-1/-1->1->-1 [12] -1/-1/-1->1->0 [13] -1/-1/-1->1->0 [14] -1/-1/-1->1->0 [15] -1/-1/-1->1->0 [16] -1/-1/-1->1->0 [17] -1/-1/-1->1->0 [18] 0/-1/-1->1->-1 [19] 0/-1/-1->1->-1 [20] 0/-1/-1->1->-1 [21] 0/-1/-1->1->-1 [22] 0/-1/-1->1->-1 [23] 0/-1/-1->1->-1
+ip-26-0-166-125:762819:763316 [4] NCCL INFO Channel 02/24 :    0   1
+ip-26-0-166-125:762820:763322 [5] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-166-125:762819:763316 [4] NCCL INFO Channel 03/24 :    0   1
+ip-26-0-166-125:762819:763316 [4] NCCL INFO Channel 04/24 :    0   1
+ip-26-0-166-125:762819:763316 [4] NCCL INFO Channel 05/24 :    0   1
+ip-26-0-166-125:762819:763316 [4] NCCL INFO Channel 06/24 :    0   1
+ip-26-0-166-125:762819:763316 [4] NCCL INFO Channel 07/24 :    0   1
+ip-26-0-166-125:762819:763316 [4] NCCL INFO Channel 08/24 :    0   1
+ip-26-0-166-125:762819:763316 [4] NCCL INFO Channel 09/24 :    0   1
+ip-26-0-166-125:762819:763316 [4] NCCL INFO Channel 10/24 :    0   1
+ip-26-0-166-125:762819:763316 [4] NCCL INFO Channel 11/24 :    0   1
+ip-26-0-166-125:762819:763316 [4] NCCL INFO Channel 12/24 :    0   1
+ip-26-0-166-125:762819:763316 [4] NCCL INFO Channel 13/24 :    0   1
+ip-26-0-166-125:762819:763316 [4] NCCL INFO Channel 14/24 :    0   1
+ip-26-0-166-125:762819:763316 [4] NCCL INFO Channel 15/24 :    0   1
+ip-26-0-166-125:762819:763316 [4] NCCL INFO Channel 16/24 :    0   1
+ip-26-0-166-125:762819:763316 [4] NCCL INFO Channel 17/24 :    0   1
+ip-26-0-166-125:762819:763316 [4] NCCL INFO Channel 18/24 :    0   1
+ip-26-0-166-125:762819:763316 [4] NCCL INFO Channel 19/24 :    0   1
+ip-26-0-166-125:762819:763316 [4] NCCL INFO Channel 20/24 :    0   1
+ip-26-0-166-125:762819:763316 [4] NCCL INFO Channel 21/24 :    0   1
+ip-26-0-166-125:762819:763316 [4] NCCL INFO Channel 22/24 :    0   1
+ip-26-0-166-125:762819:763316 [4] NCCL INFO Channel 23/24 :    0   1
+ip-26-0-166-125:762819:763316 [4] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] -1/-1/-1->0->1 [7] -1/-1/-1->0->1 [8] -1/-1/-1->0->1 [9] -1/-1/-1->0->1 [10] -1/-1/-1->0->1 [11] -1/-1/-1->0->1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] -1/-1/-1->0->1 [19] -1/-1/-1->0->1 [20] -1/-1/-1->0->1 [21] -1/-1/-1->0->1 [22] -1/-1/-1->0->1 [23] -1/-1/-1->0->1
+ip-26-0-166-125:762819:763316 [4] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-162-46:702886:703405 [6] NCCL INFO Channel 00/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-166-36:104430:104924 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-36:104434:104921 [5] NCCL INFO Channel 00/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-166-36:104436:104923 [7] NCCL INFO Channel 06/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-166-36:104435:104920 [6] NCCL INFO Channel 07/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-166-36:104429:104919 [0] NCCL INFO Channel 04/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-162-46:702883:703409 [3] NCCL INFO Channel 04/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-162-46:702887:703406 [7] NCCL INFO Channel 01/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-162-46:702884:703402 [4] NCCL INFO Channel 03/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-162-46:702886:703405 [6] NCCL INFO Channel 01/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-166-36:104434:104921 [5] NCCL INFO Channel 01/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-166-36:104436:104923 [7] NCCL INFO Channel 07/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-166-36:104435:104920 [6] NCCL INFO Channel 08/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-166-36:104429:104919 [0] NCCL INFO Channel 05/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-214:624950:625434 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-166-36:104430:104924 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-162-46:702883:703409 [3] NCCL INFO Channel 05/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-162-46:702884:703402 [4] NCCL INFO Channel 04/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-162-46:702887:703406 [7] NCCL INFO Channel 02/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-166-36:104436:104923 [7] NCCL INFO Channel 08/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-166-36:104434:104921 [5] NCCL INFO Channel 02/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-166-36:104435:104920 [6] NCCL INFO Channel 09/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-162-46:702886:703405 [6] NCCL INFO Channel 02/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-166-36:104429:104919 [0] NCCL INFO Channel 06/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-36:104430:104924 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-162-46:702883:703409 [3] NCCL INFO Channel 06/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-166-36:104436:104923 [7] NCCL INFO Channel 09/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-166-36:104434:104921 [5] NCCL INFO Channel 03/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-166-36:104435:104920 [6] NCCL INFO Channel 10/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-162-46:702884:703402 [4] NCCL INFO Channel 05/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-162-46:702887:703406 [7] NCCL INFO Channel 03/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-166-36:104429:104919 [0] NCCL INFO Channel 07/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-36:104430:104924 [1] NCCL INFO Channel 04/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-162-46:702886:703405 [6] NCCL INFO Channel 03/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-166-36:104436:104923 [7] NCCL INFO Channel 10/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-162-46:702883:703409 [3] NCCL INFO Channel 07/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-166-36:104435:104920 [6] NCCL INFO Channel 11/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-166-36:104434:104921 [5] NCCL INFO Channel 04/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-162-46:702884:703402 [4] NCCL INFO Channel 06/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-162-46:702887:703406 [7] NCCL INFO Channel 04/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-166-36:104429:104919 [0] NCCL INFO Channel 08/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-36:104430:104924 [1] NCCL INFO Channel 05/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-162-46:702886:703405 [6] NCCL INFO Channel 04/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-162-46:702882:703408 [2] NCCL INFO Channel 00/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-166-36:104436:104923 [7] NCCL INFO Channel 11/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-166-36:104435:104920 [6] NCCL INFO Channel 12/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-166-36:104434:104921 [5] NCCL INFO Channel 05/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-162-46:702883:703409 [3] NCCL INFO Channel 08/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-166-36:104429:104919 [0] NCCL INFO Channel 09/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-36:104430:104924 [1] NCCL INFO Channel 06/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-162-46:702884:703402 [4] NCCL INFO Channel 07/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-162-46:702887:703406 [7] NCCL INFO Channel 05/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-162-46:702886:703405 [6] NCCL INFO Channel 05/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-166-36:104436:104923 [7] NCCL INFO Channel 12/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-166-36:104435:104920 [6] NCCL INFO Channel 13/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-166-36:104434:104921 [5] NCCL INFO Channel 06/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-162-46:702882:703408 [2] NCCL INFO Channel 01/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-165-213:9628:10120 [3] NCCL INFO Channel 00/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-166-36:104429:104919 [0] NCCL INFO Channel 10/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-36:104430:104924 [1] NCCL INFO Channel 07/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-162-46:702880:703399 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-213:9625:10112 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-162-46:702883:703409 [3] NCCL INFO Channel 09/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-165-213:9627:10116 [2] NCCL INFO Channel 00/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-165-213:9626:10118 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-162-46:702885:703404 [5] NCCL INFO Channel 00/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-162-46:702887:703406 [7] NCCL INFO Channel 06/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-162-46:702884:703402 [4] NCCL INFO Channel 08/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-166-36:104436:104923 [7] NCCL INFO Channel 13/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-166-36:104435:104920 [6] NCCL INFO Channel 14/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-166-36:104434:104921 [5] NCCL INFO Channel 07/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-162-46:702886:703405 [6] NCCL INFO Channel 06/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-166-36:104429:104919 [0] NCCL INFO Channel 11/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-162-46:702882:703408 [2] NCCL INFO Channel 02/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-166-36:104430:104924 [1] NCCL INFO Channel 08/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-213:9628:10120 [3] NCCL INFO Channel 01/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-162-46:702880:703399 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-162-46:702883:703409 [3] NCCL INFO Channel 10/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-165-213:9625:10112 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-213:9627:10116 [2] NCCL INFO Channel 01/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-166-36:104435:104920 [6] NCCL INFO Channel 15/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-166-36:104436:104923 [7] NCCL INFO Channel 14/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-165-213:9626:10118 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-162-46:702887:703406 [7] NCCL INFO Channel 07/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-162-46:702884:703402 [4] NCCL INFO Channel 09/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-166-36:104434:104921 [5] NCCL INFO Channel 08/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-162-46:702885:703404 [5] NCCL INFO Channel 01/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-165-213:9631:10117 [6] NCCL INFO Channel 00/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-166-36:104429:104919 [0] NCCL INFO Channel 12/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-162-46:702886:703405 [6] NCCL INFO Channel 07/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-166-36:104430:104924 [1] NCCL INFO Channel 09/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-213:9632:10121 [7] NCCL INFO Channel 00/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-162-46:702882:703408 [2] NCCL INFO Channel 03/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-165-213:9628:10120 [3] NCCL INFO Channel 02/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-166-36:104435:104920 [6] NCCL INFO Channel 16/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-166-36:104436:104923 [7] NCCL INFO Channel 15/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-166-36:104434:104921 [5] NCCL INFO Channel 09/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-165-213:9625:10112 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-162-46:702883:703409 [3] NCCL INFO Channel 11/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-162-46:702880:703399 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-213:9627:10116 [2] NCCL INFO Channel 02/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-162-46:702884:703402 [4] NCCL INFO Channel 10/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-166-36:104429:104919 [0] NCCL INFO Channel 13/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-213:9626:10118 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-162-46:702887:703406 [7] NCCL INFO Channel 08/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-166-36:104430:104924 [1] NCCL INFO Channel 10/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-162-46:702885:703404 [5] NCCL INFO Channel 02/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-165-213:9631:10117 [6] NCCL INFO Channel 01/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-162-46:702886:703405 [6] NCCL INFO Channel 08/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-165-213:9632:10121 [7] NCCL INFO Channel 01/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-162-46:702882:703408 [2] NCCL INFO Channel 04/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-166-36:104435:104920 [6] NCCL INFO Channel 17/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-166-36:104436:104923 [7] NCCL INFO Channel 16/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-165-213:9628:10120 [3] NCCL INFO Channel 03/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-166-36:104434:104921 [5] NCCL INFO Channel 10/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-165-213:9625:10112 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-162-46:702883:703409 [3] NCCL INFO Channel 12/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-165-213:9627:10116 [2] NCCL INFO Channel 03/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-165-213:9626:10118 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-36:104429:104919 [0] NCCL INFO Channel 14/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-167-9:1039539:1040045 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-162-46:702880:703399 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-36:104430:104924 [1] NCCL INFO Channel 11/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-162-46:702884:703402 [4] NCCL INFO Channel 11/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-162-46:702887:703406 [7] NCCL INFO Channel 09/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-162-46:702885:703404 [5] NCCL INFO Channel 03/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-165-213:9631:10117 [6] NCCL INFO Channel 02/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-166-36:104436:104923 [7] NCCL INFO Channel 17/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-167-9:1039539:1040045 [2] NCCL INFO Channel 00/24 :    0   1
+ip-26-0-167-9:1039540:1040048 [3] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] -1/-1/-1->1->0 [3] -1/-1/-1->1->0 [4] -1/-1/-1->1->0 [5] -1/-1/-1->1->0 [6] 0/-1/-1->1->-1 [7] 0/-1/-1->1->-1 [8] 0/-1/-1->1->-1 [9] 0/-1/-1->1->-1 [10] 0/-1/-1->1->-1 [11] 0/-1/-1->1->-1 [12] -1/-1/-1->1->0 [13] -1/-1/-1->1->0 [14] -1/-1/-1->1->0 [15] -1/-1/-1->1->0 [16] -1/-1/-1->1->0 [17] -1/-1/-1->1->0 [18] 0/-1/-1->1->-1 [19] 0/-1/-1->1->-1 [20] 0/-1/-1->1->-1 [21] 0/-1/-1->1->-1 [22] 0/-1/-1->1->-1 [23] 0/-1/-1->1->-1
+ip-26-0-167-9:1039539:1040045 [2] NCCL INFO Channel 01/24 :    0   1
+ip-26-0-167-9:1039540:1040048 [3] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-167-9:1039539:1040045 [2] NCCL INFO Channel 02/24 :    0   1
+ip-26-0-167-9:1039539:1040045 [2] NCCL INFO Channel 03/24 :    0   1
+ip-26-0-167-9:1039539:1040045 [2] NCCL INFO Channel 04/24 :    0   1
+ip-26-0-167-9:1039539:1040045 [2] NCCL INFO Channel 05/24 :    0   1
+ip-26-0-162-46:702886:703405 [6] NCCL INFO Channel 09/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-166-36:104435:104920 [6] NCCL INFO Channel 18/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-165-213:9632:10121 [7] NCCL INFO Channel 02/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-167-9:1039539:1040045 [2] NCCL INFO Channel 06/24 :    0   1
+ip-26-0-167-9:1039539:1040045 [2] NCCL INFO Channel 07/24 :    0   1
+ip-26-0-167-9:1039539:1040045 [2] NCCL INFO Channel 08/24 :    0   1
+ip-26-0-167-9:1039539:1040045 [2] NCCL INFO Channel 09/24 :    0   1
+ip-26-0-167-9:1039539:1040045 [2] NCCL INFO Channel 10/24 :    0   1
+ip-26-0-167-9:1039539:1040045 [2] NCCL INFO Channel 11/24 :    0   1
+ip-26-0-167-9:1039539:1040045 [2] NCCL INFO Channel 12/24 :    0   1
+ip-26-0-167-9:1039539:1040045 [2] NCCL INFO Channel 13/24 :    0   1
+ip-26-0-167-9:1039539:1040045 [2] NCCL INFO Channel 14/24 :    0   1
+ip-26-0-167-9:1039539:1040045 [2] NCCL INFO Channel 15/24 :    0   1
+ip-26-0-167-9:1039539:1040045 [2] NCCL INFO Channel 16/24 :    0   1
+ip-26-0-167-9:1039539:1040045 [2] NCCL INFO Channel 17/24 :    0   1
+ip-26-0-167-9:1039539:1040045 [2] NCCL INFO Channel 18/24 :    0   1
+ip-26-0-167-9:1039539:1040045 [2] NCCL INFO Channel 19/24 :    0   1
+ip-26-0-167-9:1039539:1040045 [2] NCCL INFO Channel 20/24 :    0   1
+ip-26-0-167-9:1039539:1040045 [2] NCCL INFO Channel 21/24 :    0   1
+ip-26-0-167-9:1039539:1040045 [2] NCCL INFO Channel 22/24 :    0   1
+ip-26-0-167-9:1039539:1040045 [2] NCCL INFO Channel 23/24 :    0   1
+ip-26-0-167-9:1039539:1040045 [2] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] -1/-1/-1->0->1 [7] -1/-1/-1->0->1 [8] -1/-1/-1->0->1 [9] -1/-1/-1->0->1 [10] -1/-1/-1->0->1 [11] -1/-1/-1->0->1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] -1/-1/-1->0->1 [19] -1/-1/-1->0->1 [20] -1/-1/-1->0->1 [21] -1/-1/-1->0->1 [22] -1/-1/-1->0->1 [23] -1/-1/-1->0->1
+ip-26-0-167-9:1039539:1040045 [2] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-166-36:104434:104921 [5] NCCL INFO Channel 11/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-162-46:702882:703408 [2] NCCL INFO Channel 05/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-165-213:9628:10120 [3] NCCL INFO Channel 04/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-166-36:104429:104919 [0] NCCL INFO Channel 15/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-36:104430:104924 [1] NCCL INFO Channel 12/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-162-46:702883:703409 [3] NCCL INFO Channel 13/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-165-213:9625:10112 [0] NCCL INFO Channel 04/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-213:9627:10116 [2] NCCL INFO Channel 04/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-162-46:702880:703399 [0] NCCL INFO Channel 04/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-213:9626:10118 [1] NCCL INFO Channel 04/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-162-46:702884:703402 [4] NCCL INFO Channel 12/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-162-46:702887:703406 [7] NCCL INFO Channel 10/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-162-46:702885:703404 [5] NCCL INFO Channel 04/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-162-46:702886:703405 [6] NCCL INFO Channel 10/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-166-36:104436:104923 [7] NCCL INFO Channel 18/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-166-36:104435:104920 [6] NCCL INFO Channel 19/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-166-36:104434:104921 [5] NCCL INFO Channel 12/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-165-213:9631:10117 [6] NCCL INFO Channel 03/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-165-213:9632:10121 [7] NCCL INFO Channel 03/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-165-213:9628:10120 [3] NCCL INFO Channel 05/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-162-46:702882:703408 [2] NCCL INFO Channel 06/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-166-36:104429:104919 [0] NCCL INFO Channel 16/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-213:9625:10112 [0] NCCL INFO Channel 05/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-36:104430:104924 [1] NCCL INFO Channel 13/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-162-46:702883:703409 [3] NCCL INFO Channel 14/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-162-46:702880:703399 [0] NCCL INFO Channel 05/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-213:9627:10116 [2] NCCL INFO Channel 05/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-165-213:9626:10118 [1] NCCL INFO Channel 05/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-162-46:702884:703402 [4] NCCL INFO Channel 13/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-162-46:702887:703406 [7] NCCL INFO Channel 11/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-162-46:702885:703404 [5] NCCL INFO Channel 05/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-165-213:9631:10117 [6] NCCL INFO Channel 04/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-166-36:104436:104923 [7] NCCL INFO Channel 19/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-166-36:104435:104920 [6] NCCL INFO Channel 20/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-162-46:702886:703405 [6] NCCL INFO Channel 11/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-166-36:104434:104921 [5] NCCL INFO Channel 13/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-165-213:9632:10121 [7] NCCL INFO Channel 04/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-165-213:9628:10120 [3] NCCL INFO Channel 06/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-166-36:104429:104919 [0] NCCL INFO Channel 17/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-162-46:702882:703408 [2] NCCL INFO Channel 07/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-166-36:104430:104924 [1] NCCL INFO Channel 14/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-162-46:702880:703399 [0] NCCL INFO Channel 06/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-213:9625:10112 [0] NCCL INFO Channel 06/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-213:9627:10116 [2] NCCL INFO Channel 06/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-162-46:702883:703409 [3] NCCL INFO Channel 15/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-165-213:9626:10118 [1] NCCL INFO Channel 06/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-162-46:702884:703402 [4] NCCL INFO Channel 14/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-162-46:702887:703406 [7] NCCL INFO Channel 12/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-166-36:104435:104920 [6] NCCL INFO Channel 21/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-165-213:9629:10113 [4] NCCL INFO Channel 00/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-165-213:9631:10117 [6] NCCL INFO Channel 05/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-162-46:702885:703404 [5] NCCL INFO Channel 06/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-166-36:104436:104923 [7] NCCL INFO Channel 20/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-166-36:104434:104921 [5] NCCL INFO Channel 14/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-162-46:702886:703405 [6] NCCL INFO Channel 12/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-165-213:9632:10121 [7] NCCL INFO Channel 05/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-166-36:104429:104919 [0] NCCL INFO Channel 18/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-213:9628:10120 [3] NCCL INFO Channel 07/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-166-36:104430:104924 [1] NCCL INFO Channel 15/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-162-46:702882:703408 [2] NCCL INFO Channel 08/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-166-125:762817:763317 [2] NCCL INFO Channel 00/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-166-36:104433:104915 [4] NCCL INFO Channel 00/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-165-213:9630:10119 [5] NCCL INFO Channel 00/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-162-46:702880:703399 [0] NCCL INFO Channel 07/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-213:9625:10112 [0] NCCL INFO Channel 07/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-213:9627:10116 [2] NCCL INFO Channel 07/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-165-213:9626:10118 [1] NCCL INFO Channel 07/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-36:104435:104920 [6] NCCL INFO Channel 22/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-162-46:702883:703409 [3] NCCL INFO Channel 16/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-166-36:104436:104923 [7] NCCL INFO Channel 21/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-166-36:104434:104921 [5] NCCL INFO Channel 15/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-162-46:702884:703402 [4] NCCL INFO Channel 15/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-162-46:702885:703404 [5] NCCL INFO Channel 07/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-165-213:9631:10117 [6] NCCL INFO Channel 06/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-166-36:104429:104919 [0] NCCL INFO Channel 19/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-213:9629:10113 [4] NCCL INFO Channel 01/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-162-46:702886:703405 [6] NCCL INFO Channel 13/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-166-36:104430:104924 [1] NCCL INFO Channel 16/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-213:9632:10121 [7] NCCL INFO Channel 06/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-165-213:9628:10120 [3] NCCL INFO Channel 08/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-166-36:104433:104915 [4] NCCL INFO Channel 01/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-162-46:702882:703408 [2] NCCL INFO Channel 09/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-166-36:104435:104920 [6] NCCL INFO Channel 23/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-162-46:702880:703399 [0] NCCL INFO Channel 08/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-36:104436:104923 [7] NCCL INFO Channel 22/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-166-36:104434:104921 [5] NCCL INFO Channel 16/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-165-213:9630:10119 [5] NCCL INFO Channel 01/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-165-213:9625:10112 [0] NCCL INFO Channel 08/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-213:9627:10116 [2] NCCL INFO Channel 08/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-162-46:702883:703409 [3] NCCL INFO Channel 17/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-165-213:9626:10118 [1] NCCL INFO Channel 08/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-36:104429:104919 [0] NCCL INFO Channel 20/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-162-46:702884:703402 [4] NCCL INFO Channel 16/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-162-46:702887:703406 [7] NCCL INFO Channel 13/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-162-46:702885:703404 [5] NCCL INFO Channel 08/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-165-213:9631:10117 [6] NCCL INFO Channel 07/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-166-36:104430:104924 [1] NCCL INFO Channel 17/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-36:104433:104915 [4] NCCL INFO Channel 02/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-162-46:702886:703405 [6] NCCL INFO Channel 14/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-165-213:9629:10113 [4] NCCL INFO Channel 02/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-166-36:104436:104923 [7] NCCL INFO Channel 23/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-166-36:104434:104921 [5] NCCL INFO Channel 17/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-165-213:9632:10121 [7] NCCL INFO Channel 07/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-166-125:762817:763317 [2] NCCL INFO Channel 01/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-165-213:9628:10120 [3] NCCL INFO Channel 09/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-162-46:702882:703408 [2] NCCL INFO Channel 10/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-162-46:702881:703400 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-36:104429:104919 [0] NCCL INFO Channel 21/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-162-46:702880:703399 [0] NCCL INFO Channel 09/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-36:104433:104915 [4] NCCL INFO Channel 03/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-165-213:9630:10119 [5] NCCL INFO Channel 02/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-162-46:702887:703406 [7] NCCL INFO Channel 14/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-165-213:9625:10112 [0] NCCL INFO Channel 09/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-213:9627:10116 [2] NCCL INFO Channel 09/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-166-125:762817:763317 [2] NCCL INFO Channel 02/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-162-46:702883:703409 [3] NCCL INFO Channel 18/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-162-46:702884:703402 [4] NCCL INFO Channel 17/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-165-213:9626:10118 [1] NCCL INFO Channel 09/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-36:104434:104921 [5] NCCL INFO Channel 18/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-162-46:702885:703404 [5] NCCL INFO Channel 09/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-165-213:9631:10117 [6] NCCL INFO Channel 08/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-162-46:702886:703405 [6] NCCL INFO Channel 15/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-165-213:9629:10113 [4] NCCL INFO Channel 03/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-165-213:9632:10121 [7] NCCL INFO Channel 08/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-166-36:104429:104919 [0] NCCL INFO Channel 22/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-125:762817:763317 [2] NCCL INFO Channel 03/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-166-36:104433:104915 [4] NCCL INFO Channel 04/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-165-213:9628:10120 [3] NCCL INFO Channel 10/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-162-46:702882:703408 [2] NCCL INFO Channel 11/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-162-46:702881:703400 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-162-46:702880:703399 [0] NCCL INFO Channel 10/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-213:9630:10119 [5] NCCL INFO Channel 03/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-162-46:702887:703406 [7] NCCL INFO Channel 15/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-165-213:9627:10116 [2] NCCL INFO Channel 10/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-165-213:9625:10112 [0] NCCL INFO Channel 10/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-125:762817:763317 [2] NCCL INFO Channel 04/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-162-46:702883:703409 [3] NCCL INFO Channel 19/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-165-213:9626:10118 [1] NCCL INFO Channel 10/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-162-46:702884:703402 [4] NCCL INFO Channel 18/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-162-46:702885:703404 [5] NCCL INFO Channel 10/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-165-213:9631:10117 [6] NCCL INFO Channel 09/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-162-46:702886:703405 [6] NCCL INFO Channel 16/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-165-213:9629:10113 [4] NCCL INFO Channel 04/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-165-213:9632:10121 [7] NCCL INFO Channel 09/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-165-213:9628:10120 [3] NCCL INFO Channel 11/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-166-36:104430:104924 [1] NCCL INFO Channel 18/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-162-46:702882:703408 [2] NCCL INFO Channel 12/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-162-46:702881:703400 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-162-46:702880:703399 [0] NCCL INFO Channel 11/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-213:9630:10119 [5] NCCL INFO Channel 04/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-162-46:702887:703406 [7] NCCL INFO Channel 16/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-165-213:9627:10116 [2] NCCL INFO Channel 11/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-165-213:9625:10112 [0] NCCL INFO Channel 11/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-162-46:702883:703409 [3] NCCL INFO Channel 20/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-162-46:702884:703402 [4] NCCL INFO Channel 19/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-165-213:9626:10118 [1] NCCL INFO Channel 11/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-162-46:702885:703404 [5] NCCL INFO Channel 11/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-165-213:9631:10117 [6] NCCL INFO Channel 10/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-162-46:702886:703405 [6] NCCL INFO Channel 17/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-165-213:9629:10113 [4] NCCL INFO Channel 05/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-165-213:9632:10121 [7] NCCL INFO Channel 10/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-165-213:9628:10120 [3] NCCL INFO Channel 12/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-162-46:702882:703408 [2] NCCL INFO Channel 13/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-162-46:702881:703400 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-162-46:702880:703399 [0] NCCL INFO Channel 12/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-213:9630:10119 [5] NCCL INFO Channel 05/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-162-46:702887:703406 [7] NCCL INFO Channel 17/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-166-36:104434:104921 [5] NCCL INFO Channel 19/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-165-213:9627:10116 [2] NCCL INFO Channel 12/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-165-213:9625:10112 [0] NCCL INFO Channel 12/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-162-46:702883:703409 [3] NCCL INFO Channel 21/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-165-213:9626:10118 [1] NCCL INFO Channel 12/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-162-46:702884:703402 [4] NCCL INFO Channel 20/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-162-46:702885:703404 [5] NCCL INFO Channel 12/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-165-213:9631:10117 [6] NCCL INFO Channel 11/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-162-46:702886:703405 [6] NCCL INFO Channel 18/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-165-213:9629:10113 [4] NCCL INFO Channel 06/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-165-213:9628:10120 [3] NCCL INFO Channel 13/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-165-213:9632:10121 [7] NCCL INFO Channel 11/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-162-46:702881:703400 [1] NCCL INFO Channel 04/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-162-46:702882:703408 [2] NCCL INFO Channel 14/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-162-46:702880:703399 [0] NCCL INFO Channel 13/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-162-46:702887:703406 [7] NCCL INFO Channel 18/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-165-213:9630:10119 [5] NCCL INFO Channel 06/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-166-125:762818:763321 [3] NCCL INFO Channel 00/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-165-213:9627:10116 [2] NCCL INFO Channel 13/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-162-46:702883:703409 [3] NCCL INFO Channel 22/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-165-213:9625:10112 [0] NCCL INFO Channel 13/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-162-46:702884:703402 [4] NCCL INFO Channel 21/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-165-213:9626:10118 [1] NCCL INFO Channel 13/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-162-46:702885:703404 [5] NCCL INFO Channel 13/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-165-213:9631:10117 [6] NCCL INFO Channel 12/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-162-46:702886:703405 [6] NCCL INFO Channel 19/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-165-213:9629:10113 [4] NCCL INFO Channel 07/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-165-213:9628:10120 [3] NCCL INFO Channel 14/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-165-213:9632:10121 [7] NCCL INFO Channel 12/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-162-46:702881:703400 [1] NCCL INFO Channel 05/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-162-46:702882:703408 [2] NCCL INFO Channel 15/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-166-125:762818:763321 [3] NCCL INFO Channel 01/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-162-46:702880:703399 [0] NCCL INFO Channel 14/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-213:9630:10119 [5] NCCL INFO Channel 07/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-162-46:702887:703406 [7] NCCL INFO Channel 19/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-165-213:9627:10116 [2] NCCL INFO Channel 14/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-162-46:702883:703409 [3] NCCL INFO Channel 23/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-162-46:702885:703404 [5] NCCL INFO Channel 14/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-162-46:702884:703402 [4] NCCL INFO Channel 22/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-165-213:9625:10112 [0] NCCL INFO Channel 14/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-213:9626:10118 [1] NCCL INFO Channel 14/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-213:9631:10117 [6] NCCL INFO Channel 13/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-162-46:702886:703405 [6] NCCL INFO Channel 20/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-165-213:9629:10113 [4] NCCL INFO Channel 08/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-166-125:762818:763321 [3] NCCL INFO Channel 02/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-166-125:762817:763317 [2] NCCL INFO Channel 05/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-162-46:702881:703400 [1] NCCL INFO Channel 06/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-213:9628:10120 [3] NCCL INFO Channel 15/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-165-213:9632:10121 [7] NCCL INFO Channel 13/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-162-46:702882:703408 [2] NCCL INFO Channel 16/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-165-213:9630:10119 [5] NCCL INFO Channel 08/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-162-46:702880:703399 [0] NCCL INFO Channel 15/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-162-46:702887:703406 [7] NCCL INFO Channel 20/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-165-213:9627:10116 [2] NCCL INFO Channel 15/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-162-46:702884:703402 [4] NCCL INFO Channel 23/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-162-46:702885:703404 [5] NCCL INFO Channel 15/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-165-213:9625:10112 [0] NCCL INFO Channel 15/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-213:9626:10118 [1] NCCL INFO Channel 15/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-213:9631:10117 [6] NCCL INFO Channel 14/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-162-46:702886:703405 [6] NCCL INFO Channel 21/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-166-125:762817:763317 [2] NCCL INFO Channel 06/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-166-125:762818:763321 [3] NCCL INFO Channel 03/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-162-46:702881:703400 [1] NCCL INFO Channel 07/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-213:9629:10113 [4] NCCL INFO Channel 09/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-162-46:702882:703408 [2] NCCL INFO Channel 17/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-165-213:9628:10120 [3] NCCL INFO Channel 16/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-162-46:702880:703399 [0] NCCL INFO Channel 16/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-213:9632:10121 [7] NCCL INFO Channel 14/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-162-46:702887:703406 [7] NCCL INFO Channel 21/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-165-213:9630:10119 [5] NCCL INFO Channel 09/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-162-46:702885:703404 [5] NCCL INFO Channel 16/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-165-213:9627:10116 [2] NCCL INFO Channel 16/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-162-46:702886:703405 [6] NCCL INFO Channel 22/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-165-213:9625:10112 [0] NCCL INFO Channel 16/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-213:9626:10118 [1] NCCL INFO Channel 16/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762817:763317 [2] NCCL INFO Channel 07/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-165-213:9631:10117 [6] NCCL INFO Channel 15/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-166-125:762818:763321 [3] NCCL INFO Channel 04/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-162-46:702881:703400 [1] NCCL INFO Channel 08/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-162-46:702882:703408 [2] NCCL INFO Channel 18/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-165-213:9629:10113 [4] NCCL INFO Channel 10/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-165-213:9628:10120 [3] NCCL INFO Channel 17/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-166-125:762820:763322 [5] NCCL INFO Channel 00/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-165-213:9632:10121 [7] NCCL INFO Channel 15/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-165-213:9630:10119 [5] NCCL INFO Channel 10/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-165-213:9627:10116 [2] NCCL INFO Channel 17/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-165-213:9625:10112 [0] NCCL INFO Channel 17/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-125:762817:763317 [2] NCCL INFO Channel 08/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-166-125:762818:763321 [3] NCCL INFO Channel 05/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-165-213:9626:10118 [1] NCCL INFO Channel 17/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762821:763318 [6] NCCL INFO Channel 00/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-165-213:9631:10117 [6] NCCL INFO Channel 16/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-162-46:702880:703399 [0] NCCL INFO Channel 17/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-162-46:702885:703404 [5] NCCL INFO Channel 17/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-166-36:104433:104915 [4] NCCL INFO Channel 05/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-165-213:9629:10113 [4] NCCL INFO Channel 11/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-165-213:9628:10120 [3] NCCL INFO Channel 18/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-166-125:762820:763322 [5] NCCL INFO Channel 01/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-162-46:702886:703405 [6] NCCL INFO Channel 23/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-165-213:9632:10121 [7] NCCL INFO Channel 16/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-165-213:9630:10119 [5] NCCL INFO Channel 11/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-162-46:702881:703400 [1] NCCL INFO Channel 09/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-213:9627:10116 [2] NCCL INFO Channel 18/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-162-46:702880:703399 [0] NCCL INFO Channel 18/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-162-46:702885:703404 [5] NCCL INFO Channel 18/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-165-213:9625:10112 [0] NCCL INFO Channel 18/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-213:9626:10118 [1] NCCL INFO Channel 18/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762818:763321 [3] NCCL INFO Channel 06/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-166-125:762817:763317 [2] NCCL INFO Channel 09/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-165-213:9631:10117 [6] NCCL INFO Channel 17/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-166-125:762821:763318 [6] NCCL INFO Channel 01/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-165-213:9629:10113 [4] NCCL INFO Channel 12/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-166-125:762820:763322 [5] NCCL INFO Channel 02/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-165-213:9628:10120 [3] NCCL INFO Channel 19/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-165-213:9632:10121 [7] NCCL INFO Channel 17/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-166-36:104429:104919 [0] NCCL INFO Channel 23/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-36:104434:104921 [5] NCCL INFO Channel 20/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-165-213:9630:10119 [5] NCCL INFO Channel 12/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-165-213:9627:10116 [2] NCCL INFO Channel 19/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-166-214:624947:625432 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-162-46:702887:703406 [7] NCCL INFO Channel 22/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-162-46:702882:703408 [2] NCCL INFO Channel 19/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-165-213:9625:10112 [0] NCCL INFO Channel 19/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-125:762818:763321 [3] NCCL INFO Channel 07/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-165-213:9626:10118 [1] NCCL INFO Channel 19/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-213:9631:10117 [6] NCCL INFO Channel 18/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-162-46:702881:703400 [1] NCCL INFO Channel 10/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762817:763317 [2] NCCL INFO Channel 10/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-166-125:762821:763318 [6] NCCL INFO Channel 02/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-166-125:762816:763324 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-162-46:702880:703399 [0] NCCL INFO Channel 19/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-213:9629:10113 [4] NCCL INFO Channel 13/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-166-125:762820:763322 [5] NCCL INFO Channel 03/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-165-213:9628:10120 [3] NCCL INFO Channel 20/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-166-214:624949:625431 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-165-213:9632:10121 [7] NCCL INFO Channel 18/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-166-214:624949:625431 [6] NCCL INFO Channel 00/24 :    0   1
+ip-26-0-166-214:624949:625431 [6] NCCL INFO Channel 01/24 :    0   1
+ip-26-0-166-214:624949:625431 [6] NCCL INFO Channel 02/24 :    0   1
+ip-26-0-165-213:9630:10119 [5] NCCL INFO Channel 13/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-166-214:624949:625431 [6] NCCL INFO Channel 03/24 :    0   1
+ip-26-0-166-214:624949:625431 [6] NCCL INFO Channel 04/24 :    0   1
+ip-26-0-166-214:624949:625431 [6] NCCL INFO Channel 05/24 :    0   1
+ip-26-0-166-214:624949:625431 [6] NCCL INFO Channel 06/24 :    0   1
+ip-26-0-166-214:624949:625431 [6] NCCL INFO Channel 07/24 :    0   1
+ip-26-0-166-214:624949:625431 [6] NCCL INFO Channel 08/24 :    0   1
+ip-26-0-166-214:624950:625434 [7] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] -1/-1/-1->1->0 [3] -1/-1/-1->1->0 [4] -1/-1/-1->1->0 [5] -1/-1/-1->1->0 [6] 0/-1/-1->1->-1 [7] 0/-1/-1->1->-1 [8] 0/-1/-1->1->-1 [9] 0/-1/-1->1->-1 [10] 0/-1/-1->1->-1 [11] 0/-1/-1->1->-1 [12] -1/-1/-1->1->0 [13] -1/-1/-1->1->0 [14] -1/-1/-1->1->0 [15] -1/-1/-1->1->0 [16] -1/-1/-1->1->0 [17] -1/-1/-1->1->0 [18] 0/-1/-1->1->-1 [19] 0/-1/-1->1->-1 [20] 0/-1/-1->1->-1 [21] 0/-1/-1->1->-1 [22] 0/-1/-1->1->-1 [23] 0/-1/-1->1->-1
+ip-26-0-166-214:624949:625431 [6] NCCL INFO Channel 09/24 :    0   1
+ip-26-0-166-214:624950:625434 [7] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-166-214:624949:625431 [6] NCCL INFO Channel 10/24 :    0   1
+ip-26-0-166-214:624949:625431 [6] NCCL INFO Channel 11/24 :    0   1
+ip-26-0-166-214:624949:625431 [6] NCCL INFO Channel 12/24 :    0   1
+ip-26-0-166-214:624949:625431 [6] NCCL INFO Channel 13/24 :    0   1
+ip-26-0-166-214:624949:625431 [6] NCCL INFO Channel 14/24 :    0   1
+ip-26-0-166-214:624949:625431 [6] NCCL INFO Channel 15/24 :    0   1
+ip-26-0-166-214:624949:625431 [6] NCCL INFO Channel 16/24 :    0   1
+ip-26-0-166-214:624949:625431 [6] NCCL INFO Channel 17/24 :    0   1
+ip-26-0-166-214:624949:625431 [6] NCCL INFO Channel 18/24 :    0   1
+ip-26-0-166-214:624949:625431 [6] NCCL INFO Channel 19/24 :    0   1
+ip-26-0-166-214:624949:625431 [6] NCCL INFO Channel 20/24 :    0   1
+ip-26-0-166-214:624949:625431 [6] NCCL INFO Channel 21/24 :    0   1
+ip-26-0-166-214:624949:625431 [6] NCCL INFO Channel 22/24 :    0   1
+ip-26-0-166-214:624949:625431 [6] NCCL INFO Channel 23/24 :    0   1
+ip-26-0-166-214:624949:625431 [6] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] -1/-1/-1->0->1 [7] -1/-1/-1->0->1 [8] -1/-1/-1->0->1 [9] -1/-1/-1->0->1 [10] -1/-1/-1->0->1 [11] -1/-1/-1->0->1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] -1/-1/-1->0->1 [19] -1/-1/-1->0->1 [20] -1/-1/-1->0->1 [21] -1/-1/-1->0->1 [22] -1/-1/-1->0->1 [23] -1/-1/-1->0->1
+ip-26-0-166-214:624949:625431 [6] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-166-125:762822:763323 [7] NCCL INFO Channel 00/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-166-15:127291:127787 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127293:127786 [3] NCCL INFO Channel 00/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-166-15:127291:127787 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127293:127786 [3] NCCL INFO Channel 01/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-166-15:127291:127787 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-213:9627:10116 [2] NCCL INFO Channel 20/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-166-15:127293:127786 [3] NCCL INFO Channel 02/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-166-15:127291:127787 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127293:127786 [3] NCCL INFO Channel 03/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-166-15:127291:127787 [1] NCCL INFO Channel 04/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127293:127786 [3] NCCL INFO Channel 04/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-166-15:127291:127787 [1] NCCL INFO Channel 05/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127293:127786 [3] NCCL INFO Channel 05/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-166-15:127291:127787 [1] NCCL INFO Channel 06/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127293:127786 [3] NCCL INFO Channel 06/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-166-15:127292:127782 [2] NCCL INFO Channel 00/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-166-15:127291:127787 [1] NCCL INFO Channel 07/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127293:127786 [3] NCCL INFO Channel 07/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-166-15:127290:127785 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-15:127292:127782 [2] NCCL INFO Channel 01/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-166-15:127291:127787 [1] NCCL INFO Channel 08/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127293:127786 [3] NCCL INFO Channel 08/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-166-15:127290:127785 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-15:127290:127785 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-15:127290:127785 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-15:127290:127785 [0] NCCL INFO Channel 04/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-15:127297:127781 [7] NCCL INFO Channel 00/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-166-15:127290:127785 [0] NCCL INFO Channel 05/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-213:9625:10112 [0] NCCL INFO Channel 20/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-15:127294:127779 [4] NCCL INFO Channel 00/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-166-15:127297:127781 [7] NCCL INFO Channel 01/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-166-15:127290:127785 [0] NCCL INFO Channel 06/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-15:127294:127779 [4] NCCL INFO Channel 01/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-166-15:127297:127781 [7] NCCL INFO Channel 02/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-166-15:127290:127785 [0] NCCL INFO Channel 07/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-15:127294:127779 [4] NCCL INFO Channel 02/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-166-15:127297:127781 [7] NCCL INFO Channel 03/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-166-15:127292:127782 [2] NCCL INFO Channel 02/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-166-15:127290:127785 [0] NCCL INFO Channel 08/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-15:127294:127779 [4] NCCL INFO Channel 03/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-166-15:127297:127781 [7] NCCL INFO Channel 04/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-166-15:127291:127787 [1] NCCL INFO Channel 09/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127293:127786 [3] NCCL INFO Channel 09/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-166-15:127292:127782 [2] NCCL INFO Channel 03/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-166-15:127290:127785 [0] NCCL INFO Channel 09/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-15:127294:127779 [4] NCCL INFO Channel 04/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-166-15:127295:127783 [5] NCCL INFO Channel 00/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-166-125:762818:763321 [3] NCCL INFO Channel 08/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-166-125:762817:763317 [2] NCCL INFO Channel 11/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-165-213:9626:10118 [1] NCCL INFO Channel 20/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127297:127781 [7] NCCL INFO Channel 05/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-166-15:127291:127787 [1] NCCL INFO Channel 10/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127293:127786 [3] NCCL INFO Channel 10/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-166-15:127292:127782 [2] NCCL INFO Channel 04/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-166-15:127290:127785 [0] NCCL INFO Channel 10/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-15:127294:127779 [4] NCCL INFO Channel 05/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-166-15:127296:127778 [6] NCCL INFO Channel 00/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-166-15:127295:127783 [5] NCCL INFO Channel 01/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-166-15:127297:127781 [7] NCCL INFO Channel 06/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-166-15:127291:127787 [1] NCCL INFO Channel 11/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127293:127786 [3] NCCL INFO Channel 11/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-166-15:127292:127782 [2] NCCL INFO Channel 05/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-165-213:9631:10117 [6] NCCL INFO Channel 19/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-166-15:127290:127785 [0] NCCL INFO Channel 11/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-15:127294:127779 [4] NCCL INFO Channel 06/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-166-15:127296:127778 [6] NCCL INFO Channel 01/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-166-125:762821:763318 [6] NCCL INFO Channel 03/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-166-125:762816:763324 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127295:127783 [5] NCCL INFO Channel 02/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-166-15:127297:127781 [7] NCCL INFO Channel 07/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-166-15:127293:127786 [3] NCCL INFO Channel 12/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-166-15:127291:127787 [1] NCCL INFO Channel 12/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127292:127782 [2] NCCL INFO Channel 06/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-166-15:127290:127785 [0] NCCL INFO Channel 12/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-15:127294:127779 [4] NCCL INFO Channel 07/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-166-15:127292:127782 [2] NCCL INFO Channel 07/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-166-15:127297:127781 [7] NCCL INFO Channel 08/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-166-15:127291:127787 [1] NCCL INFO Channel 13/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127294:127779 [4] NCCL INFO Channel 08/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-166-15:127292:127782 [2] NCCL INFO Channel 08/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-166-15:127297:127781 [7] NCCL INFO Channel 09/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-166-15:127291:127787 [1] NCCL INFO Channel 14/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127294:127779 [4] NCCL INFO Channel 09/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-166-15:127292:127782 [2] NCCL INFO Channel 09/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-166-15:127297:127781 [7] NCCL INFO Channel 10/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-166-15:127291:127787 [1] NCCL INFO Channel 15/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127294:127779 [4] NCCL INFO Channel 10/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-166-15:127292:127782 [2] NCCL INFO Channel 10/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-166-15:127297:127781 [7] NCCL INFO Channel 11/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-166-15:127291:127787 [1] NCCL INFO Channel 16/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-162-46:702887:703406 [7] NCCL INFO Channel 23/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-166-15:127294:127779 [4] NCCL INFO Channel 11/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-166-15:127292:127782 [2] NCCL INFO Channel 11/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-166-15:127297:127781 [7] NCCL INFO Channel 12/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-166-15:127291:127787 [1] NCCL INFO Channel 17/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127294:127779 [4] NCCL INFO Channel 12/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-166-15:127297:127781 [7] NCCL INFO Channel 13/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-166-15:127292:127782 [2] NCCL INFO Channel 12/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-166-15:127294:127779 [4] NCCL INFO Channel 13/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-166-15:127291:127787 [1] NCCL INFO Channel 18/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127297:127781 [7] NCCL INFO Channel 14/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-166-15:127292:127782 [2] NCCL INFO Channel 13/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-166-15:127291:127787 [1] NCCL INFO Channel 19/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127294:127779 [4] NCCL INFO Channel 14/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-166-15:127297:127781 [7] NCCL INFO Channel 15/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-165-213:9629:10113 [4] NCCL INFO Channel 14/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-166-15:127292:127782 [2] NCCL INFO Channel 14/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-166-15:127296:127778 [6] NCCL INFO Channel 02/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-166-15:127295:127783 [5] NCCL INFO Channel 03/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-166-15:127294:127779 [4] NCCL INFO Channel 15/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-166-15:127293:127786 [3] NCCL INFO Channel 13/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-166-15:127291:127787 [1] NCCL INFO Channel 20/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127290:127785 [0] NCCL INFO Channel 13/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-125:762820:763322 [5] NCCL INFO Channel 04/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-166-15:127297:127781 [7] NCCL INFO Channel 16/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-166-15:127292:127782 [2] NCCL INFO Channel 15/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-167-9:1039544:1040046 [7] NCCL INFO Channel 00/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-162-46:702885:703404 [5] NCCL INFO Channel 19/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-166-125:762819:763316 [4] NCCL INFO Channel 00/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-165-213:9628:10120 [3] NCCL INFO Channel 21/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-162-46:702881:703400 [1] NCCL INFO Channel 11/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-162-46:702882:703408 [2] NCCL INFO Channel 20/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-165-213:9632:10121 [7] NCCL INFO Channel 19/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-165-213:9630:10119 [5] NCCL INFO Channel 14/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-166-125:762822:763323 [7] NCCL INFO Channel 01/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-166-36:104433:104915 [4] NCCL INFO Channel 06/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-165-213:9627:10116 [2] NCCL INFO Channel 21/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-166-15:127296:127778 [6] NCCL INFO Channel 03/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-166-15:127295:127783 [5] NCCL INFO Channel 04/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-166-15:127294:127779 [4] NCCL INFO Channel 16/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-165-213:9625:10112 [0] NCCL INFO Channel 21/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-213:9626:10118 [1] NCCL INFO Channel 21/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127293:127786 [3] NCCL INFO Channel 14/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-166-125:762817:763317 [2] NCCL INFO Channel 12/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-162-46:702885:703404 [5] NCCL INFO Channel 20/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-166-125:762818:763321 [3] NCCL INFO Channel 09/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-165-213:9631:10117 [6] NCCL INFO Channel 20/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-167-9:1039541:1040041 [4] NCCL INFO Channel 00/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-162-46:702881:703400 [1] NCCL INFO Channel 12/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762816:763324 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762821:763318 [6] NCCL INFO Channel 04/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-166-15:127291:127787 [1] NCCL INFO Channel 21/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127290:127785 [0] NCCL INFO Channel 14/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-213:9629:10113 [4] NCCL INFO Channel 15/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-166-15:127297:127781 [7] NCCL INFO Channel 17/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-166-15:127292:127782 [2] NCCL INFO Channel 16/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-166-125:762820:763322 [5] NCCL INFO Channel 05/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-167-9:1039544:1040046 [7] NCCL INFO Channel 01/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-166-125:762819:763316 [4] NCCL INFO Channel 01/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-166-125:762822:763323 [7] NCCL INFO Channel 02/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-166-15:127295:127783 [5] NCCL INFO Channel 05/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-166-15:127296:127778 [6] NCCL INFO Channel 04/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-166-15:127294:127779 [4] NCCL INFO Channel 17/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-166-15:127293:127786 [3] NCCL INFO Channel 15/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-167-9:1039538:1040044 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762817:763317 [2] NCCL INFO Channel 13/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-167-9:1039541:1040041 [4] NCCL INFO Channel 01/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-166-125:762818:763321 [3] NCCL INFO Channel 10/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-166-125:762816:763324 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762821:763318 [6] NCCL INFO Channel 05/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-166-15:127291:127787 [1] NCCL INFO Channel 22/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127290:127785 [0] NCCL INFO Channel 15/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-213:9632:10121 [7] NCCL INFO Channel 20/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-166-15:127297:127781 [7] NCCL INFO Channel 18/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-166-15:127292:127782 [2] NCCL INFO Channel 17/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-165-213:9625:10112 [0] NCCL INFO Channel 22/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-125:762820:763322 [5] NCCL INFO Channel 06/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-167-9:1039544:1040046 [7] NCCL INFO Channel 02/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-165-213:9631:10117 [6] NCCL INFO Channel 21/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-166-125:762819:763316 [4] NCCL INFO Channel 02/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-165-213:9632:10121 [7] NCCL INFO Channel 21/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-166-125:762822:763323 [7] NCCL INFO Channel 03/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-166-15:127295:127783 [5] NCCL INFO Channel 06/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-166-15:127296:127778 [6] NCCL INFO Channel 05/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-166-15:127294:127779 [4] NCCL INFO Channel 18/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-166-15:127293:127786 [3] NCCL INFO Channel 16/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-167-9:1039538:1040044 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-167-9:1039541:1040041 [4] NCCL INFO Channel 02/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-162-46:702882:703408 [2] NCCL INFO Channel 21/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-166-125:762817:763317 [2] NCCL INFO Channel 14/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-166-125:762816:763324 [1] NCCL INFO Channel 04/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762818:763321 [3] NCCL INFO Channel 11/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-166-15:127291:127787 [1] NCCL INFO Channel 23/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762815:763320 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-125:762821:763318 [6] NCCL INFO Channel 06/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-166-15:127290:127785 [0] NCCL INFO Channel 16/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-15:127297:127781 [7] NCCL INFO Channel 19/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-166-15:127292:127782 [2] NCCL INFO Channel 18/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-167-9:1039544:1040046 [7] NCCL INFO Channel 03/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-166-125:762820:763322 [5] NCCL INFO Channel 07/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-166-125:762819:763316 [4] NCCL INFO Channel 03/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-166-214:624943:625438 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-166-125:762822:763323 [7] NCCL INFO Channel 04/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-166-15:127295:127783 [5] NCCL INFO Channel 07/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-166-15:127294:127779 [4] NCCL INFO Channel 19/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-166-15:127296:127778 [6] NCCL INFO Channel 06/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-167-9:1039538:1040044 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127293:127786 [3] NCCL INFO Channel 17/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-167-9:1039541:1040041 [4] NCCL INFO Channel 03/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-166-15:127290:127785 [0] NCCL INFO Channel 17/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-125:762817:763317 [2] NCCL INFO Channel 15/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-166-125:762816:763324 [1] NCCL INFO Channel 05/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-167-9:1039540:1040048 [3] NCCL INFO Channel 00/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-166-125:762818:763321 [3] NCCL INFO Channel 12/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-166-125:762815:763320 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-15:127297:127781 [7] NCCL INFO Channel 20/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-166-125:762821:763318 [6] NCCL INFO Channel 07/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-166-15:127292:127782 [2] NCCL INFO Channel 19/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-166-125:762820:763322 [5] NCCL INFO Channel 08/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-166-125:762819:763316 [4] NCCL INFO Channel 04/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-165-213:9627:10116 [2] NCCL INFO Channel 22/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-165-213:9626:10118 [1] NCCL INFO Channel 22/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-213:9628:10120 [3] NCCL INFO Channel 22/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-165-213:9630:10119 [5] NCCL INFO Channel 15/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-166-15:127295:127783 [5] NCCL INFO Channel 08/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-166-125:762822:763323 [7] NCCL INFO Channel 05/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-166-15:127294:127779 [4] NCCL INFO Channel 20/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-166-15:127296:127778 [6] NCCL INFO Channel 07/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-166-15:127293:127786 [3] NCCL INFO Channel 18/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-166-15:127290:127785 [0] NCCL INFO Channel 18/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-125:762817:763317 [2] NCCL INFO Channel 16/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-166-15:127297:127781 [7] NCCL INFO Channel 21/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-165-213:9629:10113 [4] NCCL INFO Channel 16/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-166-214:624948:625435 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-166-125:762816:763324 [1] NCCL INFO Channel 06/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762818:763321 [3] NCCL INFO Channel 13/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-166-125:762815:763320 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-15:127292:127782 [2] NCCL INFO Channel 20/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-165-213:9626:10118 [1] NCCL INFO Channel 23/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624947:625432 [4] NCCL INFO Channel 00/24 :    0   1
+ip-26-0-166-214:624947:625432 [4] NCCL INFO Channel 01/24 :    0   1
+ip-26-0-166-214:624947:625432 [4] NCCL INFO Channel 02/24 :    0   1
+ip-26-0-166-214:624947:625432 [4] NCCL INFO Channel 03/24 :    0   1
+ip-26-0-166-214:624947:625432 [4] NCCL INFO Channel 04/24 :    0   1
+ip-26-0-166-214:624947:625432 [4] NCCL INFO Channel 05/24 :    0   1
+ip-26-0-166-214:624947:625432 [4] NCCL INFO Channel 06/24 :    0   1
+ip-26-0-166-214:624947:625432 [4] NCCL INFO Channel 07/24 :    0   1
+ip-26-0-166-214:624947:625432 [4] NCCL INFO Channel 08/24 :    0   1
+ip-26-0-167-9:1039541:1040041 [4] NCCL INFO Channel 04/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-166-125:762821:763318 [6] NCCL INFO Channel 08/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-165-213:9627:10116 [2] NCCL INFO Channel 23/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-165-213:9628:10120 [3] NCCL INFO Channel 23/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-166-214:624947:625432 [4] NCCL INFO Channel 09/24 :    0   1
+ip-26-0-166-214:624947:625432 [4] NCCL INFO Channel 10/24 :    0   1
+ip-26-0-166-214:624948:625435 [5] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] -1/-1/-1->1->0 [3] -1/-1/-1->1->0 [4] -1/-1/-1->1->0 [5] -1/-1/-1->1->0 [6] 0/-1/-1->1->-1 [7] 0/-1/-1->1->-1 [8] 0/-1/-1->1->-1 [9] 0/-1/-1->1->-1 [10] 0/-1/-1->1->-1 [11] 0/-1/-1->1->-1 [12] -1/-1/-1->1->0 [13] -1/-1/-1->1->0 [14] -1/-1/-1->1->0 [15] -1/-1/-1->1->0 [16] -1/-1/-1->1->0 [17] -1/-1/-1->1->0 [18] 0/-1/-1->1->-1 [19] 0/-1/-1->1->-1 [20] 0/-1/-1->1->-1 [21] 0/-1/-1->1->-1 [22] 0/-1/-1->1->-1 [23] 0/-1/-1->1->-1
+ip-26-0-166-214:624947:625432 [4] NCCL INFO Channel 11/24 :    0   1
+ip-26-0-166-214:624947:625432 [4] NCCL INFO Channel 12/24 :    0   1
+ip-26-0-166-214:624948:625435 [5] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-166-214:624947:625432 [4] NCCL INFO Channel 13/24 :    0   1
+ip-26-0-166-214:624947:625432 [4] NCCL INFO Channel 14/24 :    0   1
+ip-26-0-166-214:624947:625432 [4] NCCL INFO Channel 15/24 :    0   1
+ip-26-0-166-214:624947:625432 [4] NCCL INFO Channel 16/24 :    0   1
+ip-26-0-166-214:624947:625432 [4] NCCL INFO Channel 17/24 :    0   1
+ip-26-0-166-214:624947:625432 [4] NCCL INFO Channel 18/24 :    0   1
+ip-26-0-166-214:624947:625432 [4] NCCL INFO Channel 19/24 :    0   1
+ip-26-0-166-214:624947:625432 [4] NCCL INFO Channel 20/24 :    0   1
+ip-26-0-166-214:624947:625432 [4] NCCL INFO Channel 21/24 :    0   1
+ip-26-0-166-214:624947:625432 [4] NCCL INFO Channel 22/24 :    0   1
+ip-26-0-166-214:624947:625432 [4] NCCL INFO Channel 23/24 :    0   1
+ip-26-0-165-213:9630:10119 [5] NCCL INFO Channel 16/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-166-214:624947:625432 [4] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] -1/-1/-1->0->1 [7] -1/-1/-1->0->1 [8] -1/-1/-1->0->1 [9] -1/-1/-1->0->1 [10] -1/-1/-1->0->1 [11] -1/-1/-1->0->1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] -1/-1/-1->0->1 [19] -1/-1/-1->0->1 [20] -1/-1/-1->0->1 [21] -1/-1/-1->0->1 [22] -1/-1/-1->0->1 [23] -1/-1/-1->0->1
+ip-26-0-166-214:624947:625432 [4] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-166-125:762820:763322 [5] NCCL INFO Channel 09/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-162-46:702885:703404 [5] NCCL INFO Channel 21/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-166-125:762819:763316 [4] NCCL INFO Channel 05/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-162-46:702881:703400 [1] NCCL INFO Channel 13/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-167-9:1039538:1040044 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-167-9:1039539:1040045 [2] NCCL INFO Channel 00/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-166-15:127295:127783 [5] NCCL INFO Channel 09/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-167-9:1039543:1040043 [6] NCCL INFO Channel 00/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-166-36:104433:104915 [4] NCCL INFO Channel 07/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-165-213:9629:10113 [4] NCCL INFO Channel 17/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-166-15:127294:127779 [4] NCCL INFO Channel 21/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-167-9:1039541:1040041 [4] NCCL INFO Channel 05/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-166-15:127293:127786 [3] NCCL INFO Channel 19/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-166-15:127296:127778 [6] NCCL INFO Channel 08/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-166-125:762822:763323 [7] NCCL INFO Channel 06/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-166-15:127290:127785 [0] NCCL INFO Channel 19/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-213:9630:10119 [5] NCCL INFO Channel 17/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-166-15:127297:127781 [7] NCCL INFO Channel 22/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-167-9:1039538:1040044 [1] NCCL INFO Channel 04/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-213:9632:10121 [7] NCCL INFO Channel 22/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-166-15:127292:127782 [2] NCCL INFO Channel 21/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-166-125:762817:763317 [2] NCCL INFO Channel 17/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-167-9:1039539:1040045 [2] NCCL INFO Channel 01/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-165-213:9629:10113 [4] NCCL INFO Channel 18/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-167-9:1039543:1040043 [6] NCCL INFO Channel 01/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-166-125:762816:763324 [1] NCCL INFO Channel 07/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762818:763321 [3] NCCL INFO Channel 14/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-167-9:1039541:1040041 [4] NCCL INFO Channel 06/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-166-125:762815:763320 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-125:762821:763318 [6] NCCL INFO Channel 09/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-166-125:762820:763322 [5] NCCL INFO Channel 10/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-166-15:127295:127783 [5] NCCL INFO Channel 10/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-166-125:762819:763316 [4] NCCL INFO Channel 06/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-166-125:762822:763323 [7] NCCL INFO Channel 07/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-166-15:127294:127779 [4] NCCL INFO Channel 22/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-166-15:127293:127786 [3] NCCL INFO Channel 20/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-166-15:127296:127778 [6] NCCL INFO Channel 09/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-162-46:702880:703399 [0] NCCL INFO Channel 20/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-162-46:702882:703408 [2] NCCL INFO Channel 22/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-166-15:127290:127785 [0] NCCL INFO Channel 20/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-162-46:702881:703400 [1] NCCL INFO Channel 14/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127297:127781 [7] NCCL INFO Channel 23/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-166-15:127292:127782 [2] NCCL INFO Channel 22/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-166-36:104433:104915 [4] NCCL INFO Channel 08/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-162-46:702885:703404 [5] NCCL INFO Channel 22/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-166-125:762817:763317 [2] NCCL INFO Channel 18/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-162-46:702880:703399 [0] NCCL INFO Channel 21/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-162-46:702882:703408 [2] NCCL INFO Channel 23/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-166-125:762816:763324 [1] NCCL INFO Channel 08/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-167-9:1039544:1040046 [7] NCCL INFO Channel 04/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-167-9:1039538:1040044 [1] NCCL INFO Channel 05/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762818:763321 [3] NCCL INFO Channel 15/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-166-125:762815:763320 [0] NCCL INFO Channel 04/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-125:762821:763318 [6] NCCL INFO Channel 10/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-166-15:127295:127783 [5] NCCL INFO Channel 11/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-166-15:127294:127779 [4] NCCL INFO Channel 23/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-166-15:127293:127786 [3] NCCL INFO Channel 21/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-166-15:127296:127778 [6] NCCL INFO Channel 10/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-166-125:762820:763322 [5] NCCL INFO Channel 11/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-166-125:762819:763316 [4] NCCL INFO Channel 07/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-166-125:762822:763323 [7] NCCL INFO Channel 08/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-166-15:127290:127785 [0] NCCL INFO Channel 21/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-167-9:1039540:1040048 [3] NCCL INFO Channel 01/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-167-9:1039544:1040046 [7] NCCL INFO Channel 05/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-166-15:127292:127782 [2] NCCL INFO Channel 23/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-167-9:1039538:1040044 [1] NCCL INFO Channel 06/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-36:104434:104921 [5] NCCL INFO Channel 21/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-166-125:762817:763317 [2] NCCL INFO Channel 19/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-166-36:104430:104924 [1] NCCL INFO Channel 19/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-162-46:702881:703400 [1] NCCL INFO Channel 15/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-167-9:1039539:1040045 [2] NCCL INFO Channel 02/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-166-125:762816:763324 [1] NCCL INFO Channel 09/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127295:127783 [5] NCCL INFO Channel 12/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-166-15:127293:127786 [3] NCCL INFO Channel 22/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-166-15:127296:127778 [6] NCCL INFO Channel 11/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-165-213:9625:10112 [0] NCCL INFO Channel 23/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-125:762818:763321 [3] NCCL INFO Channel 16/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-166-125:762815:763320 [0] NCCL INFO Channel 05/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-125:762821:763318 [6] NCCL INFO Channel 11/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-166-15:127290:127785 [0] NCCL INFO Channel 22/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-167-9:1039540:1040048 [3] NCCL INFO Channel 02/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-162-46:702885:703404 [5] NCCL INFO Channel 23/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-167-9:1039537:1040038 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-167-9:1039544:1040046 [7] NCCL INFO Channel 06/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-167-9:1039538:1040044 [1] NCCL INFO Channel 07/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762820:763322 [5] NCCL INFO Channel 12/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-166-125:762819:763316 [4] NCCL INFO Channel 08/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-166-125:762822:763323 [7] NCCL INFO Channel 09/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-166-15:127295:127783 [5] NCCL INFO Channel 13/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-166-15:127293:127786 [3] NCCL INFO Channel 23/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-167-9:1039539:1040045 [2] NCCL INFO Channel 03/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-166-15:127296:127778 [6] NCCL INFO Channel 12/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-166-15:127290:127785 [0] NCCL INFO Channel 23/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-167-9:1039542:1040047 [5] NCCL INFO Channel 00/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-166-125:762817:763317 [2] NCCL INFO Channel 20/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-167-9:1039540:1040048 [3] NCCL INFO Channel 03/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-167-9:1039537:1040038 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-167-9:1039544:1040046 [7] NCCL INFO Channel 07/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-166-125:762816:763324 [1] NCCL INFO Channel 10/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-213:9631:10117 [6] NCCL INFO Channel 22/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-166-15:127295:127783 [5] NCCL INFO Channel 14/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-167-9:1039538:1040044 [1] NCCL INFO Channel 08/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-213:9632:10121 [7] NCCL INFO Channel 23/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-166-125:762818:763321 [3] NCCL INFO Channel 17/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-166-125:762815:763320 [0] NCCL INFO Channel 06/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-125:762821:763318 [6] NCCL INFO Channel 12/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-166-15:127296:127778 [6] NCCL INFO Channel 13/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-165-213:9629:10113 [4] NCCL INFO Channel 19/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-166-125:762820:763322 [5] NCCL INFO Channel 13/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-167-9:1039539:1040045 [2] NCCL INFO Channel 04/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-166-125:762819:763316 [4] NCCL INFO Channel 09/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-166-244:897244:897727 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-162-46:702880:703399 [0] NCCL INFO Channel 22/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-165-213:9631:10117 [6] NCCL INFO Channel 23/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-166-125:762822:763323 [7] NCCL INFO Channel 10/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-162-46:702881:703400 [1] NCCL INFO Channel 16/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-167-9:1039542:1040047 [5] NCCL INFO Channel 01/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-167-9:1039540:1040048 [3] NCCL INFO Channel 04/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-166-125:762817:763317 [2] NCCL INFO Channel 21/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-167-9:1039537:1040038 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-167-9:1039538:1040044 [1] NCCL INFO Channel 09/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-167-9:1039544:1040046 [7] NCCL INFO Channel 08/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-166-125:762816:763324 [1] NCCL INFO Channel 11/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762815:763320 [0] NCCL INFO Channel 07/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-125:762818:763321 [3] NCCL INFO Channel 18/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-166-125:762821:763318 [6] NCCL INFO Channel 13/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-167-9:1039539:1040045 [2] NCCL INFO Channel 05/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-167-9:1039541:1040041 [4] NCCL INFO Channel 07/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-167-9:1039543:1040043 [6] NCCL INFO Channel 02/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-166-36:104433:104915 [4] NCCL INFO Channel 09/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-166-125:762820:763322 [5] NCCL INFO Channel 14/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-166-125:762819:763316 [4] NCCL INFO Channel 10/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-166-125:762822:763323 [7] NCCL INFO Channel 11/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-165-213:9629:10113 [4] NCCL INFO Channel 20/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-167-9:1039542:1040047 [5] NCCL INFO Channel 02/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-166-125:762817:763317 [2] NCCL INFO Channel 22/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-167-9:1039540:1040048 [3] NCCL INFO Channel 05/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-167-9:1039537:1040038 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-167-9:1039538:1040044 [1] NCCL INFO Channel 10/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762818:763321 [3] NCCL INFO Channel 19/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-167-9:1039544:1040046 [7] NCCL INFO Channel 09/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-166-125:762821:763318 [6] NCCL INFO Channel 14/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-166-125:762816:763324 [1] NCCL INFO Channel 12/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-167-9:1039539:1040045 [2] NCCL INFO Channel 06/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-167-9:1039541:1040041 [4] NCCL INFO Channel 08/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-166-125:762815:763320 [0] NCCL INFO Channel 08/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-167-9:1039543:1040043 [6] NCCL INFO Channel 03/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-166-125:762819:763316 [4] NCCL INFO Channel 11/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-166-125:762820:763322 [5] NCCL INFO Channel 15/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-166-125:762822:763323 [7] NCCL INFO Channel 12/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-166-15:127295:127783 [5] NCCL INFO Channel 15/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-167-9:1039542:1040047 [5] NCCL INFO Channel 03/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-166-125:762817:763317 [2] NCCL INFO Channel 23/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-167-9:1039540:1040048 [3] NCCL INFO Channel 06/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-166-214:624944:625440 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-167-9:1039537:1040038 [0] NCCL INFO Channel 04/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-167-9:1039538:1040044 [1] NCCL INFO Channel 11/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-213:9629:10113 [4] NCCL INFO Channel 21/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-166-125:762818:763321 [3] NCCL INFO Channel 20/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-167-9:1039544:1040046 [7] NCCL INFO Channel 10/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-166-125:762815:763320 [0] NCCL INFO Channel 09/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-125:762816:763324 [1] NCCL INFO Channel 13/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-167-9:1039539:1040045 [2] NCCL INFO Channel 07/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-167-9:1039541:1040041 [4] NCCL INFO Channel 09/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-167-9:1039543:1040043 [6] NCCL INFO Channel 04/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-166-36:104434:104921 [5] NCCL INFO Channel 22/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-166-125:762819:763316 [4] NCCL INFO Channel 12/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-166-36:104430:104924 [1] NCCL INFO Channel 20/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762820:763322 [5] NCCL INFO Channel 16/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-166-125:762822:763323 [7] NCCL INFO Channel 13/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-167-9:1039542:1040047 [5] NCCL INFO Channel 04/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-162-46:702881:703400 [1] NCCL INFO Channel 17/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762818:763321 [3] NCCL INFO Channel 21/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-167-9:1039540:1040048 [3] NCCL INFO Channel 07/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-166-125:762815:763320 [0] NCCL INFO Channel 10/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-125:762816:763324 [1] NCCL INFO Channel 14/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-167-9:1039537:1040038 [0] NCCL INFO Channel 05/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-167-9:1039544:1040046 [7] NCCL INFO Channel 11/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-167-9:1039538:1040044 [1] NCCL INFO Channel 12/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762819:763316 [4] NCCL INFO Channel 13/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-167-9:1039539:1040045 [2] NCCL INFO Channel 08/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-166-125:762820:763322 [5] NCCL INFO Channel 17/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-166-125:762822:763323 [7] NCCL INFO Channel 14/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-167-9:1039541:1040041 [4] NCCL INFO Channel 10/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-167-9:1039543:1040043 [6] NCCL INFO Channel 05/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-166-125:762818:763321 [3] NCCL INFO Channel 22/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-166-125:762815:763320 [0] NCCL INFO Channel 11/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-125:762816:763324 [1] NCCL INFO Channel 15/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127296:127778 [6] NCCL INFO Channel 14/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-167-9:1039542:1040047 [5] NCCL INFO Channel 05/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-167-9:1039540:1040048 [3] NCCL INFO Channel 08/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-167-9:1039537:1040038 [0] NCCL INFO Channel 06/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-167-9:1039544:1040046 [7] NCCL INFO Channel 12/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-166-125:762819:763316 [4] NCCL INFO Channel 14/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-166-125:762820:763322 [5] NCCL INFO Channel 18/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-167-9:1039538:1040044 [1] NCCL INFO Channel 13/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762822:763323 [7] NCCL INFO Channel 15/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-167-9:1039539:1040045 [2] NCCL INFO Channel 09/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-167-9:1039541:1040041 [4] NCCL INFO Channel 11/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-167-9:1039543:1040043 [6] NCCL INFO Channel 06/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-166-214:624943:625438 [0] NCCL INFO Channel 00/24 :    0   1
+ip-26-0-166-214:624943:625438 [0] NCCL INFO Channel 01/24 :    0   1
+ip-26-0-166-214:624943:625438 [0] NCCL INFO Channel 02/24 :    0   1
+ip-26-0-166-214:624943:625438 [0] NCCL INFO Channel 03/24 :    0   1
+ip-26-0-166-214:624943:625438 [0] NCCL INFO Channel 04/24 :    0   1
+ip-26-0-166-214:624943:625438 [0] NCCL INFO Channel 05/24 :    0   1
+ip-26-0-166-214:624943:625438 [0] NCCL INFO Channel 06/24 :    0   1
+ip-26-0-166-214:624943:625438 [0] NCCL INFO Channel 07/24 :    0   1
+ip-26-0-166-214:624943:625438 [0] NCCL INFO Channel 08/24 :    0   1
+ip-26-0-166-214:624943:625438 [0] NCCL INFO Channel 09/24 :    0   1
+ip-26-0-166-214:624943:625438 [0] NCCL INFO Channel 10/24 :    0   1
+ip-26-0-166-125:762818:763321 [3] NCCL INFO Channel 23/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-166-214:624944:625440 [1] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] -1/-1/-1->1->0 [3] -1/-1/-1->1->0 [4] -1/-1/-1->1->0 [5] -1/-1/-1->1->0 [6] 0/-1/-1->1->-1 [7] 0/-1/-1->1->-1 [8] 0/-1/-1->1->-1 [9] 0/-1/-1->1->-1 [10] 0/-1/-1->1->-1 [11] 0/-1/-1->1->-1 [12] -1/-1/-1->1->0 [13] -1/-1/-1->1->0 [14] -1/-1/-1->1->0 [15] -1/-1/-1->1->0 [16] -1/-1/-1->1->0 [17] -1/-1/-1->1->0 [18] 0/-1/-1->1->-1 [19] 0/-1/-1->1->-1 [20] 0/-1/-1->1->-1 [21] 0/-1/-1->1->-1 [22] 0/-1/-1->1->-1 [23] 0/-1/-1->1->-1
+ip-26-0-166-214:624943:625438 [0] NCCL INFO Channel 11/24 :    0   1
+ip-26-0-166-214:624943:625438 [0] NCCL INFO Channel 12/24 :    0   1
+ip-26-0-166-214:624943:625438 [0] NCCL INFO Channel 13/24 :    0   1
+ip-26-0-166-214:624944:625440 [1] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-166-214:624943:625438 [0] NCCL INFO Channel 14/24 :    0   1
+ip-26-0-166-214:624943:625438 [0] NCCL INFO Channel 15/24 :    0   1
+ip-26-0-166-214:624943:625438 [0] NCCL INFO Channel 16/24 :    0   1
+ip-26-0-166-214:624943:625438 [0] NCCL INFO Channel 17/24 :    0   1
+ip-26-0-166-214:624943:625438 [0] NCCL INFO Channel 18/24 :    0   1
+ip-26-0-166-214:624943:625438 [0] NCCL INFO Channel 19/24 :    0   1
+ip-26-0-166-214:624943:625438 [0] NCCL INFO Channel 20/24 :    0   1
+ip-26-0-166-214:624943:625438 [0] NCCL INFO Channel 21/24 :    0   1
+ip-26-0-166-214:624943:625438 [0] NCCL INFO Channel 22/24 :    0   1
+ip-26-0-166-214:624943:625438 [0] NCCL INFO Channel 23/24 :    0   1
+ip-26-0-166-214:624943:625438 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] -1/-1/-1->0->1 [7] -1/-1/-1->0->1 [8] -1/-1/-1->0->1 [9] -1/-1/-1->0->1 [10] -1/-1/-1->0->1 [11] -1/-1/-1->0->1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] -1/-1/-1->0->1 [19] -1/-1/-1->0->1 [20] -1/-1/-1->0->1 [21] -1/-1/-1->0->1 [22] -1/-1/-1->0->1 [23] -1/-1/-1->0->1
+ip-26-0-166-214:624943:625438 [0] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-166-125:762815:763320 [0] NCCL INFO Channel 12/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-125:762816:763324 [1] NCCL INFO Channel 16/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-167-9:1039542:1040047 [5] NCCL INFO Channel 06/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-166-125:762819:763316 [4] NCCL INFO Channel 15/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-167-9:1039540:1040048 [3] NCCL INFO Channel 09/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-166-125:762820:763322 [5] NCCL INFO Channel 19/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-166-125:762822:763323 [7] NCCL INFO Channel 16/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-167-9:1039537:1040038 [0] NCCL INFO Channel 07/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-167-9:1039544:1040046 [7] NCCL INFO Channel 13/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-166-36:104433:104915 [4] NCCL INFO Channel 10/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-167-9:1039538:1040044 [1] NCCL INFO Channel 14/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-36:104434:104921 [5] NCCL INFO Channel 23/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-167-9:1039539:1040045 [2] NCCL INFO Channel 10/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-166-125:762815:763320 [0] NCCL INFO Channel 13/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-167-9:1039541:1040041 [4] NCCL INFO Channel 12/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-166-125:762816:763324 [1] NCCL INFO Channel 17/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-167-9:1039543:1040043 [6] NCCL INFO Channel 07/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-165-213:9630:10119 [5] NCCL INFO Channel 18/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-167-9:1039542:1040047 [5] NCCL INFO Channel 07/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-166-125:762820:763322 [5] NCCL INFO Channel 20/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-166-125:762822:763323 [7] NCCL INFO Channel 17/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-167-9:1039540:1040048 [3] NCCL INFO Channel 10/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-167-9:1039544:1040046 [7] NCCL INFO Channel 14/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-167-9:1039537:1040038 [0] NCCL INFO Channel 08/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-167-9:1039538:1040044 [1] NCCL INFO Channel 15/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-36:104430:104924 [1] NCCL INFO Channel 21/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-167-9:1039539:1040045 [2] NCCL INFO Channel 11/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-167-9:1039541:1040041 [4] NCCL INFO Channel 13/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-167-9:1039543:1040043 [6] NCCL INFO Channel 08/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-167-9:1039542:1040047 [5] NCCL INFO Channel 08/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-167-9:1039540:1040048 [3] NCCL INFO Channel 11/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-167-9:1039537:1040038 [0] NCCL INFO Channel 09/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-167-9:1039544:1040046 [7] NCCL INFO Channel 15/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-167-9:1039538:1040044 [1] NCCL INFO Channel 16/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-167-9:1039539:1040045 [2] NCCL INFO Channel 12/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-167-9:1039541:1040041 [4] NCCL INFO Channel 14/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-167-9:1039543:1040043 [6] NCCL INFO Channel 09/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-167-9:1039542:1040047 [5] NCCL INFO Channel 09/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-167-9:1039540:1040048 [3] NCCL INFO Channel 12/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-166-125:762819:763316 [4] NCCL INFO Channel 16/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-167-9:1039537:1040038 [0] NCCL INFO Channel 10/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-167-9:1039544:1040046 [7] NCCL INFO Channel 16/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-165-213:9630:10119 [5] NCCL INFO Channel 19/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-166-15:127295:127783 [5] NCCL INFO Channel 16/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-167-9:1039538:1040044 [1] NCCL INFO Channel 17/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-167-9:1039539:1040045 [2] NCCL INFO Channel 13/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-166-125:762815:763320 [0] NCCL INFO Channel 14/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-167-9:1039541:1040041 [4] NCCL INFO Channel 15/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-166-125:762816:763324 [1] NCCL INFO Channel 18/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-167-9:1039543:1040043 [6] NCCL INFO Channel 10/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-167-9:1039542:1040047 [5] NCCL INFO Channel 10/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-167-9:1039540:1040048 [3] NCCL INFO Channel 13/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-167-9:1039537:1040038 [0] NCCL INFO Channel 11/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-167-9:1039544:1040046 [7] NCCL INFO Channel 17/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-167-9:1039538:1040044 [1] NCCL INFO Channel 18/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-162-46:702881:703400 [1] NCCL INFO Channel 18/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-167-9:1039539:1040045 [2] NCCL INFO Channel 14/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-167-9:1039541:1040041 [4] NCCL INFO Channel 16/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-167-9:1039543:1040043 [6] NCCL INFO Channel 11/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-162-46:702880:703399 [0] NCCL INFO Channel 23/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-244:897245:897729 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-166-244:897244:897727 [4] NCCL INFO Channel 00/24 :    0   1
+ip-26-0-166-244:897244:897727 [4] NCCL INFO Channel 01/24 :    0   1
+ip-26-0-166-244:897244:897727 [4] NCCL INFO Channel 02/24 :    0   1
+ip-26-0-166-244:897244:897727 [4] NCCL INFO Channel 03/24 :    0   1
+ip-26-0-166-244:897244:897727 [4] NCCL INFO Channel 04/24 :    0   1
+ip-26-0-166-244:897244:897727 [4] NCCL INFO Channel 05/24 :    0   1
+ip-26-0-166-244:897244:897727 [4] NCCL INFO Channel 06/24 :    0   1
+ip-26-0-166-244:897244:897727 [4] NCCL INFO Channel 07/24 :    0   1
+ip-26-0-166-244:897244:897727 [4] NCCL INFO Channel 08/24 :    0   1
+ip-26-0-166-244:897245:897729 [5] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] -1/-1/-1->1->0 [3] -1/-1/-1->1->0 [4] -1/-1/-1->1->0 [5] -1/-1/-1->1->0 [6] 0/-1/-1->1->-1 [7] 0/-1/-1->1->-1 [8] 0/-1/-1->1->-1 [9] 0/-1/-1->1->-1 [10] 0/-1/-1->1->-1 [11] 0/-1/-1->1->-1 [12] -1/-1/-1->1->0 [13] -1/-1/-1->1->0 [14] -1/-1/-1->1->0 [15] -1/-1/-1->1->0 [16] -1/-1/-1->1->0 [17] -1/-1/-1->1->0 [18] 0/-1/-1->1->-1 [19] 0/-1/-1->1->-1 [20] 0/-1/-1->1->-1 [21] 0/-1/-1->1->-1 [22] 0/-1/-1->1->-1 [23] 0/-1/-1->1->-1
+ip-26-0-166-244:897244:897727 [4] NCCL INFO Channel 09/24 :    0   1
+ip-26-0-166-244:897244:897727 [4] NCCL INFO Channel 10/24 :    0   1
+ip-26-0-166-244:897245:897729 [5] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-166-244:897244:897727 [4] NCCL INFO Channel 11/24 :    0   1
+ip-26-0-166-244:897244:897727 [4] NCCL INFO Channel 12/24 :    0   1
+ip-26-0-166-244:897244:897727 [4] NCCL INFO Channel 13/24 :    0   1
+ip-26-0-166-244:897244:897727 [4] NCCL INFO Channel 14/24 :    0   1
+ip-26-0-166-244:897244:897727 [4] NCCL INFO Channel 15/24 :    0   1
+ip-26-0-166-244:897244:897727 [4] NCCL INFO Channel 16/24 :    0   1
+ip-26-0-166-244:897244:897727 [4] NCCL INFO Channel 17/24 :    0   1
+ip-26-0-166-244:897244:897727 [4] NCCL INFO Channel 18/24 :    0   1
+ip-26-0-166-244:897244:897727 [4] NCCL INFO Channel 19/24 :    0   1
+ip-26-0-166-244:897244:897727 [4] NCCL INFO Channel 20/24 :    0   1
+ip-26-0-166-244:897244:897727 [4] NCCL INFO Channel 21/24 :    0   1
+ip-26-0-166-244:897244:897727 [4] NCCL INFO Channel 22/24 :    0   1
+ip-26-0-166-244:897244:897727 [4] NCCL INFO Channel 23/24 :    0   1
+ip-26-0-166-244:897244:897727 [4] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] -1/-1/-1->0->1 [7] -1/-1/-1->0->1 [8] -1/-1/-1->0->1 [9] -1/-1/-1->0->1 [10] -1/-1/-1->0->1 [11] -1/-1/-1->0->1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] -1/-1/-1->0->1 [19] -1/-1/-1->0->1 [20] -1/-1/-1->0->1 [21] -1/-1/-1->0->1 [22] -1/-1/-1->0->1 [23] -1/-1/-1->0->1
+ip-26-0-166-244:897244:897727 [4] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-166-36:104433:104915 [4] NCCL INFO Channel 11/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-167-9:1039542:1040047 [5] NCCL INFO Channel 11/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-167-9:1039540:1040048 [3] NCCL INFO Channel 14/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-166-36:104430:104924 [1] NCCL INFO Channel 22/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-167-9:1039537:1040038 [0] NCCL INFO Channel 12/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-167-9:1039544:1040046 [7] NCCL INFO Channel 18/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-167-9:1039538:1040044 [1] NCCL INFO Channel 19/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-167-9:1039539:1040045 [2] NCCL INFO Channel 15/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-167-9:1039541:1040041 [4] NCCL INFO Channel 17/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-166-125:762821:763318 [6] NCCL INFO Channel 15/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-167-9:1039543:1040043 [6] NCCL INFO Channel 12/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-166-125:762820:763322 [5] NCCL INFO Channel 21/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-166-125:762822:763323 [7] NCCL INFO Channel 18/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-162-46:702881:703400 [1] NCCL INFO Channel 19/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-167-9:1039542:1040047 [5] NCCL INFO Channel 12/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-167-9:1039540:1040048 [3] NCCL INFO Channel 15/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-167-9:1039537:1040038 [0] NCCL INFO Channel 13/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-167-9:1039544:1040046 [7] NCCL INFO Channel 19/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-167-9:1039538:1040044 [1] NCCL INFO Channel 20/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-167-9:1039541:1040041 [4] NCCL INFO Channel 18/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-167-9:1039539:1040045 [2] NCCL INFO Channel 16/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-167-9:1039543:1040043 [6] NCCL INFO Channel 13/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-165-213:9629:10113 [4] NCCL INFO Channel 22/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-166-214:624946:625439 [3] NCCL INFO Channel 00/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-166-15:127296:127778 [6] NCCL INFO Channel 15/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-167-9:1039544:1040046 [7] NCCL INFO Channel 20/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-167-9:1039542:1040047 [5] NCCL INFO Channel 13/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-166-125:762819:763316 [4] NCCL INFO Channel 17/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-166-214:624946:625439 [3] NCCL INFO Channel 01/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-165-213:9630:10119 [5] NCCL INFO Channel 20/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-167-9:1039540:1040048 [3] NCCL INFO Channel 16/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-167-9:1039538:1040044 [1] NCCL INFO Channel 21/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-167-9:1039541:1040041 [4] NCCL INFO Channel 19/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-166-214:624946:625439 [3] NCCL INFO Channel 02/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-166-36:104433:104915 [4] NCCL INFO Channel 12/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-166-214:624946:625439 [3] NCCL INFO Channel 03/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-167-9:1039540:1040048 [3] NCCL INFO Channel 17/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-167-9:1039537:1040038 [0] NCCL INFO Channel 14/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-36:104430:104924 [1] NCCL INFO Channel 23/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-167-9:1039540:1040048 [3] NCCL INFO Channel 18/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-167-9:1039543:1040043 [6] NCCL INFO Channel 14/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-167-9:1039537:1040038 [0] NCCL INFO Channel 15/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-214:624950:625434 [7] NCCL INFO Channel 00/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-166-214:624946:625439 [3] NCCL INFO Channel 04/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-167-9:1039539:1040045 [2] NCCL INFO Channel 17/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-167-9:1039544:1040046 [7] NCCL INFO Channel 21/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-167-9:1039540:1040048 [3] NCCL INFO Channel 19/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-167-9:1039543:1040043 [6] NCCL INFO Channel 15/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-167-9:1039537:1040038 [0] NCCL INFO Channel 16/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-167-9:1039539:1040045 [2] NCCL INFO Channel 18/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-167-9:1039540:1040048 [3] NCCL INFO Channel 20/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-167-9:1039544:1040046 [7] NCCL INFO Channel 22/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-166-214:624950:625434 [7] NCCL INFO Channel 01/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-167-9:1039543:1040043 [6] NCCL INFO Channel 16/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-166-125:762815:763320 [0] NCCL INFO Channel 15/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-167-9:1039537:1040038 [0] NCCL INFO Channel 17/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-214:624946:625439 [3] NCCL INFO Channel 05/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-166-125:762821:763318 [6] NCCL INFO Channel 16/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-167-9:1039539:1040045 [2] NCCL INFO Channel 19/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-167-9:1039540:1040048 [3] NCCL INFO Channel 21/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-167-9:1039544:1040046 [7] NCCL INFO Channel 23/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-166-15:127295:127783 [5] NCCL INFO Channel 17/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-167-9:1039543:1040043 [6] NCCL INFO Channel 17/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-167-9:1039537:1040038 [0] NCCL INFO Channel 18/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-214:624950:625434 [7] NCCL INFO Channel 02/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-167-9:1039540:1040048 [3] NCCL INFO Channel 22/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-167-9:1039539:1040045 [2] NCCL INFO Channel 20/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-166-214:624946:625439 [3] NCCL INFO Channel 06/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-166-36:104433:104915 [4] NCCL INFO Channel 13/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-167-9:1039543:1040043 [6] NCCL INFO Channel 18/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-167-9:1039537:1040038 [0] NCCL INFO Channel 19/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-125:762815:763320 [0] NCCL INFO Channel 16/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-167-9:1039540:1040048 [3] NCCL INFO Channel 23/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-167-9:1039539:1040045 [2] NCCL INFO Channel 21/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-166-214:624948:625435 [5] NCCL INFO Channel 00/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-162-46:702881:703400 [1] NCCL INFO Channel 20/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-167-9:1039537:1040038 [0] NCCL INFO Channel 20/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-167-9:1039543:1040043 [6] NCCL INFO Channel 19/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-166-214:624950:625434 [7] NCCL INFO Channel 03/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-167-9:1039539:1040045 [2] NCCL INFO Channel 22/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-166-214:624946:625439 [3] NCCL INFO Channel 07/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-167-9:1039537:1040038 [0] NCCL INFO Channel 21/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-167-9:1039543:1040043 [6] NCCL INFO Channel 20/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-166-214:624948:625435 [5] NCCL INFO Channel 01/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-167-9:1039539:1040045 [2] NCCL INFO Channel 23/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-167-9:1039542:1040047 [5] NCCL INFO Channel 14/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-166-125:762816:763324 [1] NCCL INFO Channel 19/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-167-9:1039538:1040044 [1] NCCL INFO Channel 22/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127296:127778 [6] NCCL INFO Channel 16/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-166-125:762821:763318 [6] NCCL INFO Channel 17/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-166-125:762822:763323 [7] NCCL INFO Channel 19/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-166-214:624950:625434 [7] NCCL INFO Channel 04/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-167-9:1039537:1040038 [0] NCCL INFO Channel 22/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-214:624946:625439 [3] NCCL INFO Channel 08/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-167-9:1039543:1040043 [6] NCCL INFO Channel 21/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-165-213:9629:10113 [4] NCCL INFO Channel 23/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-167-9:1039541:1040041 [4] NCCL INFO Channel 20/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-167-9:1039542:1040047 [5] NCCL INFO Channel 15/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-167-9:1039538:1040044 [1] NCCL INFO Channel 23/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624948:625435 [5] NCCL INFO Channel 02/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-166-36:104431:104917 [2] NCCL INFO Channel 00/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-167-9:1039537:1040038 [0] NCCL INFO Channel 23/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-167-9:1039543:1040043 [6] NCCL INFO Channel 22/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-166-214:624950:625434 [7] NCCL INFO Channel 05/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-167-9:1039541:1040041 [4] NCCL INFO Channel 21/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-166-214:624946:625439 [3] NCCL INFO Channel 09/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-166-125:762820:763322 [5] NCCL INFO Channel 22/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-166-36:104431:104917 [2] NCCL INFO Channel 01/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-166-36:104431:104917 [2] NCCL INFO Channel 02/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-166-214:624948:625435 [5] NCCL INFO Channel 03/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-166-36:104431:104917 [2] NCCL INFO Channel 03/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-166-36:104431:104917 [2] NCCL INFO Channel 04/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-166-214:624950:625434 [7] NCCL INFO Channel 06/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-166-214:624946:625439 [3] NCCL INFO Channel 10/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-166-125:762819:763316 [4] NCCL INFO Channel 18/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-166-125:762815:763320 [0] NCCL INFO Channel 17/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-214:624945:625436 [2] NCCL INFO Channel 00/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-166-214:624948:625435 [5] NCCL INFO Channel 04/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-166-214:624950:625434 [7] NCCL INFO Channel 07/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-165-213:9630:10119 [5] NCCL INFO Channel 21/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-166-214:624946:625439 [3] NCCL INFO Channel 11/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-166-214:624945:625436 [2] NCCL INFO Channel 01/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-166-214:624948:625435 [5] NCCL INFO Channel 05/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-166-15:127296:127778 [6] NCCL INFO Channel 17/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-166-214:624950:625434 [7] NCCL INFO Channel 08/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-166-214:624946:625439 [3] NCCL INFO Channel 12/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-167-9:1039542:1040047 [5] NCCL INFO Channel 16/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-166-214:624945:625436 [2] NCCL INFO Channel 02/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-166-214:624948:625435 [5] NCCL INFO Channel 06/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-166-125:762821:763318 [6] NCCL INFO Channel 18/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-166-125:762820:763322 [5] NCCL INFO Channel 23/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-166-214:624950:625434 [7] NCCL INFO Channel 09/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-166-214:624946:625439 [3] NCCL INFO Channel 13/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-166-214:624945:625436 [2] NCCL INFO Channel 03/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-166-214:624948:625435 [5] NCCL INFO Channel 07/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-167-9:1039543:1040043 [6] NCCL INFO Channel 23/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-167-9:1039541:1040041 [4] NCCL INFO Channel 22/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-166-214:624950:625434 [7] NCCL INFO Channel 10/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-166-214:624946:625439 [3] NCCL INFO Channel 14/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-166-214:624943:625438 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-214:624945:625436 [2] NCCL INFO Channel 04/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-166-214:624948:625435 [5] NCCL INFO Channel 08/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-166-214:624944:625440 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624949:625431 [6] NCCL INFO Channel 00/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-166-214:624950:625434 [7] NCCL INFO Channel 11/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-166-125:762819:763316 [4] NCCL INFO Channel 19/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-166-214:624946:625439 [3] NCCL INFO Channel 15/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-166-125:762815:763320 [0] NCCL INFO Channel 18/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-214:624943:625438 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-214:624945:625436 [2] NCCL INFO Channel 05/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-166-214:624944:625440 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624948:625435 [5] NCCL INFO Channel 09/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-166-214:624949:625431 [6] NCCL INFO Channel 01/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-166-214:624950:625434 [7] NCCL INFO Channel 12/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-166-214:624946:625439 [3] NCCL INFO Channel 16/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-166-214:624947:625432 [4] NCCL INFO Channel 00/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-167-9:1039542:1040047 [5] NCCL INFO Channel 17/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-166-125:762816:763324 [1] NCCL INFO Channel 20/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762821:763318 [6] NCCL INFO Channel 19/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-166-36:104433:104915 [4] NCCL INFO Channel 14/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-166-214:624943:625438 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-214:624945:625436 [2] NCCL INFO Channel 06/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-166-214:624944:625440 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624948:625435 [5] NCCL INFO Channel 10/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-166-214:624949:625431 [6] NCCL INFO Channel 02/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-166-214:624950:625434 [7] NCCL INFO Channel 13/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-166-214:624946:625439 [3] NCCL INFO Channel 17/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-166-214:624947:625432 [4] NCCL INFO Channel 01/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-166-125:762822:763323 [7] NCCL INFO Channel 20/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-166-214:624943:625438 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-214:624945:625436 [2] NCCL INFO Channel 07/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-166-214:624944:625440 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624948:625435 [5] NCCL INFO Channel 11/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-166-214:624949:625431 [6] NCCL INFO Channel 03/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-166-214:624950:625434 [7] NCCL INFO Channel 14/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-166-214:624946:625439 [3] NCCL INFO Channel 18/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-166-214:624947:625432 [4] NCCL INFO Channel 02/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-166-214:624943:625438 [0] NCCL INFO Channel 04/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-214:624944:625440 [1] NCCL INFO Channel 04/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624945:625436 [2] NCCL INFO Channel 08/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-166-214:624949:625431 [6] NCCL INFO Channel 04/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-166-214:624948:625435 [5] NCCL INFO Channel 12/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-166-125:762819:763316 [4] NCCL INFO Channel 20/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-166-214:624950:625434 [7] NCCL INFO Channel 15/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-166-214:624946:625439 [3] NCCL INFO Channel 19/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-166-214:624947:625432 [4] NCCL INFO Channel 03/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-167-9:1039541:1040041 [4] NCCL INFO Channel 23/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-166-36:104433:104915 [4] NCCL INFO Channel 15/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-166-214:624944:625440 [1] NCCL INFO Channel 05/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624943:625438 [0] NCCL INFO Channel 05/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-214:624945:625436 [2] NCCL INFO Channel 09/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-166-214:624949:625431 [6] NCCL INFO Channel 05/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-166-214:624948:625435 [5] NCCL INFO Channel 13/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-166-125:762815:763320 [0] NCCL INFO Channel 19/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-214:624950:625434 [7] NCCL INFO Channel 16/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-166-214:624946:625439 [3] NCCL INFO Channel 20/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-166-125:762816:763324 [1] NCCL INFO Channel 21/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762821:763318 [6] NCCL INFO Channel 20/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-166-125:762822:763323 [7] NCCL INFO Channel 21/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-166-214:624947:625432 [4] NCCL INFO Channel 04/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-166-214:624944:625440 [1] NCCL INFO Channel 06/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624943:625438 [0] NCCL INFO Channel 06/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-214:624945:625436 [2] NCCL INFO Channel 10/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-166-214:624948:625435 [5] NCCL INFO Channel 14/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-166-214:624949:625431 [6] NCCL INFO Channel 06/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-166-214:624950:625434 [7] NCCL INFO Channel 17/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-166-214:624946:625439 [3] NCCL INFO Channel 21/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-166-214:624947:625432 [4] NCCL INFO Channel 05/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-166-214:624944:625440 [1] NCCL INFO Channel 07/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624943:625438 [0] NCCL INFO Channel 07/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-214:624945:625436 [2] NCCL INFO Channel 11/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-166-214:624948:625435 [5] NCCL INFO Channel 15/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-166-214:624949:625431 [6] NCCL INFO Channel 07/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-166-214:624950:625434 [7] NCCL INFO Channel 18/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-165-213:9630:10119 [5] NCCL INFO Channel 22/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-166-214:624946:625439 [3] NCCL INFO Channel 22/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-166-125:762819:763316 [4] NCCL INFO Channel 21/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-166-214:624947:625432 [4] NCCL INFO Channel 06/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-166-214:624944:625440 [1] NCCL INFO Channel 08/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624943:625438 [0] NCCL INFO Channel 08/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-214:624945:625436 [2] NCCL INFO Channel 12/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-166-214:624948:625435 [5] NCCL INFO Channel 16/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-162-46:702881:703400 [1] NCCL INFO Channel 21/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624949:625431 [6] NCCL INFO Channel 08/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-167-9:1039542:1040047 [5] NCCL INFO Channel 18/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-166-214:624950:625434 [7] NCCL INFO Channel 19/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-166-214:624946:625439 [3] NCCL INFO Channel 23/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-166-125:762815:763320 [0] NCCL INFO Channel 20/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-214:624947:625432 [4] NCCL INFO Channel 07/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-166-214:624944:625440 [1] NCCL INFO Channel 09/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624943:625438 [0] NCCL INFO Channel 09/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-214:624945:625436 [2] NCCL INFO Channel 13/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-166-214:624948:625435 [5] NCCL INFO Channel 17/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-166-214:624949:625431 [6] NCCL INFO Channel 09/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-166-214:624950:625434 [7] NCCL INFO Channel 20/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-166-36:104431:104917 [2] NCCL INFO Channel 05/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-166-214:624947:625432 [4] NCCL INFO Channel 08/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-166-214:624944:625440 [1] NCCL INFO Channel 10/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624943:625438 [0] NCCL INFO Channel 10/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-214:624945:625436 [2] NCCL INFO Channel 14/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-166-214:624948:625435 [5] NCCL INFO Channel 18/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-166-214:624949:625431 [6] NCCL INFO Channel 10/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-166-214:624950:625434 [7] NCCL INFO Channel 21/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-166-214:624947:625432 [4] NCCL INFO Channel 09/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-166-214:624944:625440 [1] NCCL INFO Channel 11/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624943:625438 [0] NCCL INFO Channel 11/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-214:624945:625436 [2] NCCL INFO Channel 15/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-166-214:624949:625431 [6] NCCL INFO Channel 11/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-166-214:624948:625435 [5] NCCL INFO Channel 19/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-166-214:624950:625434 [7] NCCL INFO Channel 22/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-166-214:624947:625432 [4] NCCL INFO Channel 10/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-166-214:624944:625440 [1] NCCL INFO Channel 12/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624943:625438 [0] NCCL INFO Channel 12/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-214:624945:625436 [2] NCCL INFO Channel 16/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-166-214:624949:625431 [6] NCCL INFO Channel 12/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-166-214:624948:625435 [5] NCCL INFO Channel 20/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-166-214:624950:625434 [7] NCCL INFO Channel 23/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-166-125:762819:763316 [4] NCCL INFO Channel 22/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-166-125:762815:763320 [0] NCCL INFO Channel 21/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-214:624947:625432 [4] NCCL INFO Channel 11/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-166-214:624944:625440 [1] NCCL INFO Channel 13/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624943:625438 [0] NCCL INFO Channel 13/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-214:624945:625436 [2] NCCL INFO Channel 17/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-166-214:624949:625431 [6] NCCL INFO Channel 13/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-166-214:624948:625435 [5] NCCL INFO Channel 21/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-166-214:624947:625432 [4] NCCL INFO Channel 12/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-166-214:624944:625440 [1] NCCL INFO Channel 14/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624943:625438 [0] NCCL INFO Channel 14/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-214:624949:625431 [6] NCCL INFO Channel 14/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-166-214:624945:625436 [2] NCCL INFO Channel 18/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-166-214:624948:625435 [5] NCCL INFO Channel 22/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-166-36:104431:104917 [2] NCCL INFO Channel 06/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-166-15:127295:127783 [5] NCCL INFO Channel 18/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-166-214:624947:625432 [4] NCCL INFO Channel 13/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-166-214:624944:625440 [1] NCCL INFO Channel 15/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624943:625438 [0] NCCL INFO Channel 15/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-214:624949:625431 [6] NCCL INFO Channel 15/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-166-214:624945:625436 [2] NCCL INFO Channel 19/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-166-214:624948:625435 [5] NCCL INFO Channel 23/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-166-125:762816:763324 [1] NCCL INFO Channel 22/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762821:763318 [6] NCCL INFO Channel 21/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-166-214:624947:625432 [4] NCCL INFO Channel 14/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-166-125:762822:763323 [7] NCCL INFO Channel 22/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-166-214:624944:625440 [1] NCCL INFO Channel 16/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624943:625438 [0] NCCL INFO Channel 16/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-214:624949:625431 [6] NCCL INFO Channel 16/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-166-214:624945:625436 [2] NCCL INFO Channel 20/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-166-214:624947:625432 [4] NCCL INFO Channel 15/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-166-214:624943:625438 [0] NCCL INFO Channel 17/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-214:624944:625440 [1] NCCL INFO Channel 17/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624949:625431 [6] NCCL INFO Channel 17/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-166-214:624945:625436 [2] NCCL INFO Channel 21/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-166-15:127296:127778 [6] NCCL INFO Channel 18/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-166-214:624947:625432 [4] NCCL INFO Channel 16/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-166-214:624943:625438 [0] NCCL INFO Channel 18/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-214:624949:625431 [6] NCCL INFO Channel 18/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-166-214:624944:625440 [1] NCCL INFO Channel 18/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624945:625436 [2] NCCL INFO Channel 22/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-166-214:624947:625432 [4] NCCL INFO Channel 17/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-166-125:762819:763316 [4] NCCL INFO Channel 23/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-166-214:624943:625438 [0] NCCL INFO Channel 19/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-214:624949:625431 [6] NCCL INFO Channel 19/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-166-214:624944:625440 [1] NCCL INFO Channel 19/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762815:763320 [0] NCCL INFO Channel 22/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-214:624945:625436 [2] NCCL INFO Channel 23/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-166-36:104432:104922 [3] NCCL INFO Channel 00/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-166-36:104433:104915 [4] NCCL INFO Channel 16/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-166-214:624947:625432 [4] NCCL INFO Channel 18/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-166-214:624949:625431 [6] NCCL INFO Channel 20/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-166-214:624943:625438 [0] NCCL INFO Channel 20/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-214:624944:625440 [1] NCCL INFO Channel 20/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624947:625432 [4] NCCL INFO Channel 19/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-166-214:624949:625431 [6] NCCL INFO Channel 21/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-165-213:9630:10119 [5] NCCL INFO Channel 23/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-166-36:104431:104917 [2] NCCL INFO Channel 07/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-166-125:762816:763324 [1] NCCL INFO Channel 23/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762821:763318 [6] NCCL INFO Channel 22/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-166-125:762822:763323 [7] NCCL INFO Channel 23/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-166-214:624943:625438 [0] NCCL INFO Channel 21/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-214:624944:625440 [1] NCCL INFO Channel 21/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624947:625432 [4] NCCL INFO Channel 20/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-166-214:624949:625431 [6] NCCL INFO Channel 22/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-166-214:624943:625438 [0] NCCL INFO Channel 22/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-125:762815:763320 [0] NCCL INFO Channel 23/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-214:624944:625440 [1] NCCL INFO Channel 22/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-162-46:702881:703400 [1] NCCL INFO Channel 22/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-36:104432:104922 [3] NCCL INFO Channel 01/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-166-36:104433:104915 [4] NCCL INFO Channel 17/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-166-125:762821:763318 [6] NCCL INFO Channel 23/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-166-36:104431:104917 [2] NCCL INFO Channel 08/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-166-15:127296:127778 [6] NCCL INFO Channel 19/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-166-214:624947:625432 [4] NCCL INFO Channel 21/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-166-214:624949:625431 [6] NCCL INFO Channel 23/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-166-15:127296:127778 [6] NCCL INFO Channel 20/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-166-15:127295:127783 [5] NCCL INFO Channel 19/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-166-214:624944:625440 [1] NCCL INFO Channel 23/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-36:104432:104922 [3] NCCL INFO Channel 02/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-167-9:1039542:1040047 [5] NCCL INFO Channel 19/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-166-214:624947:625432 [4] NCCL INFO Channel 22/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-162-46:702881:703400 [1] NCCL INFO Channel 23/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-36:104433:104915 [4] NCCL INFO Channel 18/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-166-36:104431:104917 [2] NCCL INFO Channel 09/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-166-214:624943:625438 [0] NCCL INFO Channel 23/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-15:127296:127778 [6] NCCL INFO Channel 21/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-166-214:624947:625432 [4] NCCL INFO Channel 23/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-166-15:127295:127783 [5] NCCL INFO Channel 20/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-167-9:1039542:1040047 [5] NCCL INFO Channel 20/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-166-15:127296:127778 [6] NCCL INFO Channel 22/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-166-36:104432:104922 [3] NCCL INFO Channel 03/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-166-36:104433:104915 [4] NCCL INFO Channel 19/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-166-15:127296:127778 [6] NCCL INFO Channel 23/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-166-36:104432:104922 [3] NCCL INFO Channel 04/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-166-36:104433:104915 [4] NCCL INFO Channel 20/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-166-244:897242:897730 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-166-36:104431:104917 [2] NCCL INFO Channel 10/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-167-9:1039542:1040047 [5] NCCL INFO Channel 21/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-166-15:127295:127783 [5] NCCL INFO Channel 21/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-166-36:104433:104915 [4] NCCL INFO Channel 21/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-166-36:104432:104922 [3] NCCL INFO Channel 05/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-166-36:104431:104917 [2] NCCL INFO Channel 11/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-166-15:127295:127783 [5] NCCL INFO Channel 22/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-166-36:104433:104915 [4] NCCL INFO Channel 22/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-166-36:104432:104922 [3] NCCL INFO Channel 06/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-167-9:1039542:1040047 [5] NCCL INFO Channel 22/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-166-36:104431:104917 [2] NCCL INFO Channel 12/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-166-36:104433:104915 [4] NCCL INFO Channel 23/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-166-15:127295:127783 [5] NCCL INFO Channel 23/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-166-36:104432:104922 [3] NCCL INFO Channel 07/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-166-36:104431:104917 [2] NCCL INFO Channel 13/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-167-9:1039542:1040047 [5] NCCL INFO Channel 23/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-166-36:104432:104922 [3] NCCL INFO Channel 08/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-166-36:104431:104917 [2] NCCL INFO Channel 14/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-166-36:104432:104922 [3] NCCL INFO Channel 09/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-166-36:104431:104917 [2] NCCL INFO Channel 15/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-166-36:104432:104922 [3] NCCL INFO Channel 10/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-166-36:104431:104917 [2] NCCL INFO Channel 16/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-166-244:897247:897736 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-166-36:104436:104923 [7] NCCL INFO Connected all rings
+ip-26-0-166-36:104436:104923 [7] NCCL INFO Connected all trees
+ip-26-0-166-36:104436:104923 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-36:104436:104923 [7] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-36:104436:104923 [7] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-166-36:104432:104922 [3] NCCL INFO Channel 11/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-166-36:104431:104917 [2] NCCL INFO Channel 17/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-166-36:104431:104917 [2] NCCL INFO Channel 18/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-166-36:104432:104922 [3] NCCL INFO Channel 12/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-166-36:104432:104922 [3] NCCL INFO Channel 13/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-166-36:104431:104917 [2] NCCL INFO Channel 19/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-166-244:897241:897737 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-166-36:104432:104922 [3] NCCL INFO Channel 14/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-166-36:104431:104917 [2] NCCL INFO Channel 20/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-166-244:897240:897735 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-166-244:897241:897737 [1] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] -1/-1/-1->1->0 [3] -1/-1/-1->1->0 [4] -1/-1/-1->1->0 [5] -1/-1/-1->1->0 [6] 0/-1/-1->1->-1 [7] 0/-1/-1->1->-1 [8] 0/-1/-1->1->-1 [9] 0/-1/-1->1->-1 [10] 0/-1/-1->1->-1 [11] 0/-1/-1->1->-1 [12] -1/-1/-1->1->0 [13] -1/-1/-1->1->0 [14] -1/-1/-1->1->0 [15] -1/-1/-1->1->0 [16] -1/-1/-1->1->0 [17] -1/-1/-1->1->0 [18] 0/-1/-1->1->-1 [19] 0/-1/-1->1->-1 [20] 0/-1/-1->1->-1 [21] 0/-1/-1->1->-1 [22] 0/-1/-1->1->-1 [23] 0/-1/-1->1->-1
+ip-26-0-166-244:897241:897737 [1] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-166-244:897240:897735 [0] NCCL INFO Channel 00/24 :    0   1
+ip-26-0-166-244:897240:897735 [0] NCCL INFO Channel 01/24 :    0   1
+ip-26-0-166-244:897240:897735 [0] NCCL INFO Channel 02/24 :    0   1
+ip-26-0-166-244:897240:897735 [0] NCCL INFO Channel 03/24 :    0   1
+ip-26-0-166-244:897240:897735 [0] NCCL INFO Channel 04/24 :    0   1
+ip-26-0-166-244:897240:897735 [0] NCCL INFO Channel 05/24 :    0   1
+ip-26-0-166-244:897240:897735 [0] NCCL INFO Channel 06/24 :    0   1
+ip-26-0-166-244:897240:897735 [0] NCCL INFO Channel 07/24 :    0   1
+ip-26-0-166-244:897240:897735 [0] NCCL INFO Channel 08/24 :    0   1
+ip-26-0-166-244:897240:897735 [0] NCCL INFO Channel 09/24 :    0   1
+ip-26-0-166-244:897240:897735 [0] NCCL INFO Channel 10/24 :    0   1
+ip-26-0-166-244:897240:897735 [0] NCCL INFO Channel 11/24 :    0   1
+ip-26-0-166-244:897240:897735 [0] NCCL INFO Channel 12/24 :    0   1
+ip-26-0-166-244:897240:897735 [0] NCCL INFO Channel 13/24 :    0   1
+ip-26-0-166-244:897240:897735 [0] NCCL INFO Channel 14/24 :    0   1
+ip-26-0-166-244:897240:897735 [0] NCCL INFO Channel 15/24 :    0   1
+ip-26-0-166-244:897240:897735 [0] NCCL INFO Channel 16/24 :    0   1
+ip-26-0-166-244:897240:897735 [0] NCCL INFO Channel 17/24 :    0   1
+ip-26-0-166-244:897240:897735 [0] NCCL INFO Channel 18/24 :    0   1
+ip-26-0-166-244:897240:897735 [0] NCCL INFO Channel 19/24 :    0   1
+ip-26-0-166-244:897240:897735 [0] NCCL INFO Channel 20/24 :    0   1
+ip-26-0-166-244:897240:897735 [0] NCCL INFO Channel 21/24 :    0   1
+ip-26-0-166-244:897240:897735 [0] NCCL INFO Channel 22/24 :    0   1
+ip-26-0-166-244:897240:897735 [0] NCCL INFO Channel 23/24 :    0   1
+ip-26-0-166-244:897240:897735 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] -1/-1/-1->0->1 [7] -1/-1/-1->0->1 [8] -1/-1/-1->0->1 [9] -1/-1/-1->0->1 [10] -1/-1/-1->0->1 [11] -1/-1/-1->0->1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] -1/-1/-1->0->1 [19] -1/-1/-1->0->1 [20] -1/-1/-1->0->1 [21] -1/-1/-1->0->1 [22] -1/-1/-1->0->1 [23] -1/-1/-1->0->1
+ip-26-0-166-244:897240:897735 [0] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-166-244:897243:897731 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-166-244:897242:897730 [2] NCCL INFO Channel 00/24 :    0   1
+ip-26-0-166-244:897242:897730 [2] NCCL INFO Channel 01/24 :    0   1
+ip-26-0-166-244:897242:897730 [2] NCCL INFO Channel 02/24 :    0   1
+ip-26-0-166-244:897242:897730 [2] NCCL INFO Channel 03/24 :    0   1
+ip-26-0-166-244:897242:897730 [2] NCCL INFO Channel 04/24 :    0   1
+ip-26-0-166-244:897242:897730 [2] NCCL INFO Channel 05/24 :    0   1
+ip-26-0-166-244:897242:897730 [2] NCCL INFO Channel 06/24 :    0   1
+ip-26-0-166-244:897242:897730 [2] NCCL INFO Channel 07/24 :    0   1
+ip-26-0-166-244:897242:897730 [2] NCCL INFO Channel 08/24 :    0   1
+ip-26-0-166-244:897242:897730 [2] NCCL INFO Channel 09/24 :    0   1
+ip-26-0-166-244:897243:897731 [3] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] -1/-1/-1->1->0 [3] -1/-1/-1->1->0 [4] -1/-1/-1->1->0 [5] -1/-1/-1->1->0 [6] 0/-1/-1->1->-1 [7] 0/-1/-1->1->-1 [8] 0/-1/-1->1->-1 [9] 0/-1/-1->1->-1 [10] 0/-1/-1->1->-1 [11] 0/-1/-1->1->-1 [12] -1/-1/-1->1->0 [13] -1/-1/-1->1->0 [14] -1/-1/-1->1->0 [15] -1/-1/-1->1->0 [16] -1/-1/-1->1->0 [17] -1/-1/-1->1->0 [18] 0/-1/-1->1->-1 [19] 0/-1/-1->1->-1 [20] 0/-1/-1->1->-1 [21] 0/-1/-1->1->-1 [22] 0/-1/-1->1->-1 [23] 0/-1/-1->1->-1
+ip-26-0-166-244:897242:897730 [2] NCCL INFO Channel 10/24 :    0   1
+ip-26-0-166-244:897242:897730 [2] NCCL INFO Channel 11/24 :    0   1
+ip-26-0-166-244:897243:897731 [3] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-166-244:897242:897730 [2] NCCL INFO Channel 12/24 :    0   1
+ip-26-0-166-244:897242:897730 [2] NCCL INFO Channel 13/24 :    0   1
+ip-26-0-166-244:897242:897730 [2] NCCL INFO Channel 14/24 :    0   1
+ip-26-0-166-244:897242:897730 [2] NCCL INFO Channel 15/24 :    0   1
+ip-26-0-166-244:897242:897730 [2] NCCL INFO Channel 16/24 :    0   1
+ip-26-0-166-244:897242:897730 [2] NCCL INFO Channel 17/24 :    0   1
+ip-26-0-166-244:897242:897730 [2] NCCL INFO Channel 18/24 :    0   1
+ip-26-0-166-244:897242:897730 [2] NCCL INFO Channel 19/24 :    0   1
+ip-26-0-166-244:897242:897730 [2] NCCL INFO Channel 20/24 :    0   1
+ip-26-0-166-244:897242:897730 [2] NCCL INFO Channel 21/24 :    0   1
+ip-26-0-166-244:897242:897730 [2] NCCL INFO Channel 22/24 :    0   1
+ip-26-0-166-244:897242:897730 [2] NCCL INFO Channel 23/24 :    0   1
+ip-26-0-166-244:897242:897730 [2] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] -1/-1/-1->0->1 [7] -1/-1/-1->0->1 [8] -1/-1/-1->0->1 [9] -1/-1/-1->0->1 [10] -1/-1/-1->0->1 [11] -1/-1/-1->0->1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] -1/-1/-1->0->1 [19] -1/-1/-1->0->1 [20] -1/-1/-1->0->1 [21] -1/-1/-1->0->1 [22] -1/-1/-1->0->1 [23] -1/-1/-1->0->1
+ip-26-0-166-244:897242:897730 [2] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-166-244:897246:897734 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-166-244:897246:897734 [6] NCCL INFO Channel 00/24 :    0   1
+ip-26-0-166-244:897247:897736 [7] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] -1/-1/-1->1->0 [3] -1/-1/-1->1->0 [4] -1/-1/-1->1->0 [5] -1/-1/-1->1->0 [6] 0/-1/-1->1->-1 [7] 0/-1/-1->1->-1 [8] 0/-1/-1->1->-1 [9] 0/-1/-1->1->-1 [10] 0/-1/-1->1->-1 [11] 0/-1/-1->1->-1 [12] -1/-1/-1->1->0 [13] -1/-1/-1->1->0 [14] -1/-1/-1->1->0 [15] -1/-1/-1->1->0 [16] -1/-1/-1->1->0 [17] -1/-1/-1->1->0 [18] 0/-1/-1->1->-1 [19] 0/-1/-1->1->-1 [20] 0/-1/-1->1->-1 [21] 0/-1/-1->1->-1 [22] 0/-1/-1->1->-1 [23] 0/-1/-1->1->-1
+ip-26-0-166-244:897247:897736 [7] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-166-244:897246:897734 [6] NCCL INFO Channel 01/24 :    0   1
+ip-26-0-166-244:897246:897734 [6] NCCL INFO Channel 02/24 :    0   1
+ip-26-0-166-244:897246:897734 [6] NCCL INFO Channel 03/24 :    0   1
+ip-26-0-166-244:897246:897734 [6] NCCL INFO Channel 04/24 :    0   1
+ip-26-0-166-244:897246:897734 [6] NCCL INFO Channel 05/24 :    0   1
+ip-26-0-166-244:897246:897734 [6] NCCL INFO Channel 06/24 :    0   1
+ip-26-0-166-244:897246:897734 [6] NCCL INFO Channel 07/24 :    0   1
+ip-26-0-166-244:897246:897734 [6] NCCL INFO Channel 08/24 :    0   1
+ip-26-0-166-244:897246:897734 [6] NCCL INFO Channel 09/24 :    0   1
+ip-26-0-166-244:897246:897734 [6] NCCL INFO Channel 10/24 :    0   1
+ip-26-0-166-244:897246:897734 [6] NCCL INFO Channel 11/24 :    0   1
+ip-26-0-166-244:897246:897734 [6] NCCL INFO Channel 12/24 :    0   1
+ip-26-0-166-244:897246:897734 [6] NCCL INFO Channel 13/24 :    0   1
+ip-26-0-166-244:897246:897734 [6] NCCL INFO Channel 14/24 :    0   1
+ip-26-0-166-244:897246:897734 [6] NCCL INFO Channel 15/24 :    0   1
+ip-26-0-166-244:897246:897734 [6] NCCL INFO Channel 16/24 :    0   1
+ip-26-0-166-244:897246:897734 [6] NCCL INFO Channel 17/24 :    0   1
+ip-26-0-166-244:897246:897734 [6] NCCL INFO Channel 18/24 :    0   1
+ip-26-0-166-244:897246:897734 [6] NCCL INFO Channel 19/24 :    0   1
+ip-26-0-166-244:897246:897734 [6] NCCL INFO Channel 20/24 :    0   1
+ip-26-0-166-244:897246:897734 [6] NCCL INFO Channel 21/24 :    0   1
+ip-26-0-166-244:897246:897734 [6] NCCL INFO Channel 22/24 :    0   1
+ip-26-0-166-244:897246:897734 [6] NCCL INFO Channel 23/24 :    0   1
+ip-26-0-166-244:897246:897734 [6] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] -1/-1/-1->0->1 [7] -1/-1/-1->0->1 [8] -1/-1/-1->0->1 [9] -1/-1/-1->0->1 [10] -1/-1/-1->0->1 [11] -1/-1/-1->0->1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] -1/-1/-1->0->1 [19] -1/-1/-1->0->1 [20] -1/-1/-1->0->1 [21] -1/-1/-1->0->1 [22] -1/-1/-1->0->1 [23] -1/-1/-1->0->1
+ip-26-0-166-244:897246:897734 [6] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-166-36:104432:104922 [3] NCCL INFO Channel 15/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-166-36:104431:104917 [2] NCCL INFO Channel 21/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-166-244:897244:897727 [4] NCCL INFO Channel 00/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-166-36:104432:104922 [3] NCCL INFO Channel 16/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-166-244:897244:897727 [4] NCCL INFO Channel 01/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-166-244:897244:897727 [4] NCCL INFO Channel 02/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-166-36:104431:104917 [2] NCCL INFO Channel 22/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-166-244:897244:897727 [4] NCCL INFO Channel 03/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-166-244:897245:897729 [5] NCCL INFO Channel 00/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-166-244:897244:897727 [4] NCCL INFO Channel 04/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-166-244:897245:897729 [5] NCCL INFO Channel 01/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-166-244:897244:897727 [4] NCCL INFO Channel 05/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-166-244:897245:897729 [5] NCCL INFO Channel 02/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-166-244:897244:897727 [4] NCCL INFO Channel 06/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-166-244:897245:897729 [5] NCCL INFO Channel 03/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-166-244:897244:897727 [4] NCCL INFO Channel 07/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-166-244:897245:897729 [5] NCCL INFO Channel 04/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-166-244:897244:897727 [4] NCCL INFO Channel 08/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-166-36:104432:104922 [3] NCCL INFO Channel 17/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-166-244:897245:897729 [5] NCCL INFO Channel 05/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-166-244:897244:897727 [4] NCCL INFO Channel 09/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-166-244:897245:897729 [5] NCCL INFO Channel 06/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-166-244:897244:897727 [4] NCCL INFO Channel 10/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-166-244:897242:897730 [2] NCCL INFO Channel 00/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-166-244:897243:897731 [3] NCCL INFO Channel 00/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-166-244:897245:897729 [5] NCCL INFO Channel 07/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-166-244:897244:897727 [4] NCCL INFO Channel 11/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-166-244:897240:897735 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-244:897241:897737 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-36:104431:104917 [2] NCCL INFO Channel 23/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-166-244:897242:897730 [2] NCCL INFO Channel 01/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-166-244:897244:897727 [4] NCCL INFO Channel 12/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-166-244:897243:897731 [3] NCCL INFO Channel 01/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-166-244:897245:897729 [5] NCCL INFO Channel 08/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-166-244:897240:897735 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-244:897241:897737 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-244:897246:897734 [6] NCCL INFO Channel 00/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-166-244:897242:897730 [2] NCCL INFO Channel 02/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-166-244:897244:897727 [4] NCCL INFO Channel 13/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-166-244:897243:897731 [3] NCCL INFO Channel 02/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-166-244:897245:897729 [5] NCCL INFO Channel 09/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-166-244:897240:897735 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-244:897241:897737 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-244:897246:897734 [6] NCCL INFO Channel 01/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-166-244:897242:897730 [2] NCCL INFO Channel 03/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-166-244:897244:897727 [4] NCCL INFO Channel 14/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-166-244:897243:897731 [3] NCCL INFO Channel 03/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-166-244:897245:897729 [5] NCCL INFO Channel 10/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-166-244:897240:897735 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-244:897241:897737 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-244:897246:897734 [6] NCCL INFO Channel 02/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-166-244:897242:897730 [2] NCCL INFO Channel 04/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-166-244:897244:897727 [4] NCCL INFO Channel 15/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-166-244:897243:897731 [3] NCCL INFO Channel 04/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-166-244:897245:897729 [5] NCCL INFO Channel 11/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-166-244:897240:897735 [0] NCCL INFO Channel 04/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-244:897241:897737 [1] NCCL INFO Channel 04/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-36:104435:104920 [6] NCCL INFO Connected all rings
+ip-26-0-166-36:104435:104920 [6] NCCL INFO Connected all trees
+ip-26-0-166-36:104435:104920 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-36:104435:104920 [6] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-36:104435:104920 [6] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-166-244:897246:897734 [6] NCCL INFO Channel 03/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-166-244:897242:897730 [2] NCCL INFO Channel 05/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-166-244:897244:897727 [4] NCCL INFO Channel 16/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-166-244:897243:897731 [3] NCCL INFO Channel 05/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-166-244:897245:897729 [5] NCCL INFO Channel 12/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-166-244:897240:897735 [0] NCCL INFO Channel 05/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-36:104432:104922 [3] NCCL INFO Channel 18/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-166-244:897241:897737 [1] NCCL INFO Channel 05/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-244:897246:897734 [6] NCCL INFO Channel 04/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-166-244:897242:897730 [2] NCCL INFO Channel 06/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-166-244:897244:897727 [4] NCCL INFO Channel 17/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-166-244:897243:897731 [3] NCCL INFO Channel 06/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-166-244:897245:897729 [5] NCCL INFO Channel 13/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-166-244:897240:897735 [0] NCCL INFO Channel 06/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-244:897247:897736 [7] NCCL INFO Channel 00/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-166-244:897241:897737 [1] NCCL INFO Channel 06/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-244:897246:897734 [6] NCCL INFO Channel 05/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-166-244:897242:897730 [2] NCCL INFO Channel 07/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-166-244:897244:897727 [4] NCCL INFO Channel 18/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-166-244:897245:897729 [5] NCCL INFO Channel 14/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-166-244:897240:897735 [0] NCCL INFO Channel 07/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-244:897243:897731 [3] NCCL INFO Channel 07/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-166-244:897247:897736 [7] NCCL INFO Channel 01/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-166-244:897241:897737 [1] NCCL INFO Channel 07/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-244:897246:897734 [6] NCCL INFO Channel 06/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-166-244:897242:897730 [2] NCCL INFO Channel 08/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-166-244:897244:897727 [4] NCCL INFO Channel 19/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-166-244:897245:897729 [5] NCCL INFO Channel 15/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-166-244:897240:897735 [0] NCCL INFO Channel 08/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-244:897243:897731 [3] NCCL INFO Channel 08/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-166-244:897247:897736 [7] NCCL INFO Channel 02/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-166-244:897241:897737 [1] NCCL INFO Channel 08/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-244:897246:897734 [6] NCCL INFO Channel 07/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-166-244:897242:897730 [2] NCCL INFO Channel 09/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-166-244:897244:897727 [4] NCCL INFO Channel 20/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-166-244:897245:897729 [5] NCCL INFO Channel 16/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-166-244:897240:897735 [0] NCCL INFO Channel 09/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-244:897243:897731 [3] NCCL INFO Channel 09/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-166-244:897247:897736 [7] NCCL INFO Channel 03/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-166-244:897241:897737 [1] NCCL INFO Channel 09/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-244:897242:897730 [2] NCCL INFO Channel 10/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-166-244:897246:897734 [6] NCCL INFO Channel 08/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-166-244:897244:897727 [4] NCCL INFO Channel 21/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-166-244:897245:897729 [5] NCCL INFO Channel 17/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-166-244:897240:897735 [0] NCCL INFO Channel 10/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-244:897243:897731 [3] NCCL INFO Channel 10/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-166-244:897247:897736 [7] NCCL INFO Channel 04/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-166-244:897241:897737 [1] NCCL INFO Channel 10/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-244:897242:897730 [2] NCCL INFO Channel 11/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-166-244:897246:897734 [6] NCCL INFO Channel 09/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-166-244:897244:897727 [4] NCCL INFO Channel 22/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-166-244:897245:897729 [5] NCCL INFO Channel 18/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-166-244:897240:897735 [0] NCCL INFO Channel 11/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-244:897243:897731 [3] NCCL INFO Channel 11/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-166-244:897247:897736 [7] NCCL INFO Channel 05/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-166-244:897241:897737 [1] NCCL INFO Channel 11/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-244:897242:897730 [2] NCCL INFO Channel 12/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-166-244:897246:897734 [6] NCCL INFO Channel 10/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-166-244:897244:897727 [4] NCCL INFO Channel 23/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-166-244:897245:897729 [5] NCCL INFO Channel 19/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-166-244:897240:897735 [0] NCCL INFO Channel 12/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-244:897243:897731 [3] NCCL INFO Channel 12/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-166-244:897247:897736 [7] NCCL INFO Channel 06/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-166-244:897241:897737 [1] NCCL INFO Channel 12/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-244:897242:897730 [2] NCCL INFO Channel 13/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-166-244:897246:897734 [6] NCCL INFO Channel 11/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-166-244:897245:897729 [5] NCCL INFO Channel 20/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-166-244:897240:897735 [0] NCCL INFO Channel 13/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-244:897243:897731 [3] NCCL INFO Channel 13/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-166-244:897247:897736 [7] NCCL INFO Channel 07/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-166-244:897241:897737 [1] NCCL INFO Channel 13/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-244:897246:897734 [6] NCCL INFO Channel 12/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-166-244:897242:897730 [2] NCCL INFO Channel 14/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-166-244:897245:897729 [5] NCCL INFO Channel 21/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-166-244:897240:897735 [0] NCCL INFO Channel 14/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-244:897243:897731 [3] NCCL INFO Channel 14/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-166-244:897241:897737 [1] NCCL INFO Channel 14/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-244:897247:897736 [7] NCCL INFO Channel 08/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-166-244:897246:897734 [6] NCCL INFO Channel 13/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-166-244:897242:897730 [2] NCCL INFO Channel 15/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-166-244:897245:897729 [5] NCCL INFO Channel 22/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-166-244:897240:897735 [0] NCCL INFO Channel 15/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-244:897243:897731 [3] NCCL INFO Channel 15/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-166-36:104432:104922 [3] NCCL INFO Channel 19/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-166-244:897247:897736 [7] NCCL INFO Channel 09/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-166-244:897241:897737 [1] NCCL INFO Channel 15/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-244:897246:897734 [6] NCCL INFO Channel 14/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-166-244:897242:897730 [2] NCCL INFO Channel 16/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-166-244:897240:897735 [0] NCCL INFO Channel 16/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-244:897245:897729 [5] NCCL INFO Channel 23/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-166-244:897243:897731 [3] NCCL INFO Channel 16/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-166-244:897247:897736 [7] NCCL INFO Channel 10/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-166-244:897241:897737 [1] NCCL INFO Channel 16/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-244:897246:897734 [6] NCCL INFO Channel 15/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-166-244:897242:897730 [2] NCCL INFO Channel 17/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-166-244:897240:897735 [0] NCCL INFO Channel 17/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-244:897243:897731 [3] NCCL INFO Channel 17/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-166-244:897247:897736 [7] NCCL INFO Channel 11/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-166-244:897241:897737 [1] NCCL INFO Channel 17/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-244:897246:897734 [6] NCCL INFO Channel 16/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-166-244:897242:897730 [2] NCCL INFO Channel 18/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-166-244:897240:897735 [0] NCCL INFO Channel 18/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-244:897243:897731 [3] NCCL INFO Channel 18/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-166-244:897247:897736 [7] NCCL INFO Channel 12/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-166-244:897241:897737 [1] NCCL INFO Channel 18/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-244:897246:897734 [6] NCCL INFO Channel 17/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-166-244:897242:897730 [2] NCCL INFO Channel 19/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-166-244:897240:897735 [0] NCCL INFO Channel 19/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-244:897243:897731 [3] NCCL INFO Channel 19/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-166-244:897247:897736 [7] NCCL INFO Channel 13/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-166-244:897241:897737 [1] NCCL INFO Channel 19/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-244:897246:897734 [6] NCCL INFO Channel 18/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-166-36:104436:104923 [7] NCCL INFO comm 0xac23ec0 rank 1 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0x3e7d69e4ffee2364 - Init COMPLETE
+ip-26-0-166-36:104435:104920 [6] NCCL INFO comm 0xa2b2950 rank 0 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0x3e7d69e4ffee2364 - Init COMPLETE
+ip-26-0-166-244:897242:897730 [2] NCCL INFO Channel 20/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-166-244:897240:897735 [0] NCCL INFO Channel 20/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-36:104432:104922 [3] NCCL INFO Channel 20/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-166-244:897243:897731 [3] NCCL INFO Channel 20/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-166-244:897247:897736 [7] NCCL INFO Channel 14/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-166-244:897241:897737 [1] NCCL INFO Channel 20/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-244:897246:897734 [6] NCCL INFO Channel 19/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-166-244:897242:897730 [2] NCCL INFO Channel 21/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-166-36:104432:104922 [3] NCCL INFO Channel 21/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-166-244:897240:897735 [0] NCCL INFO Channel 21/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-244:897243:897731 [3] NCCL INFO Channel 21/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-166-244:897247:897736 [7] NCCL INFO Channel 15/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-166-244:897241:897737 [1] NCCL INFO Channel 21/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-36:104429:104919 [0] NCCL INFO Connected all rings
+ip-26-0-166-36:104429:104919 [0] NCCL INFO Connected all trees
+ip-26-0-166-36:104429:104919 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-36:104429:104919 [0] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-36:104429:104919 [0] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-166-244:897246:897734 [6] NCCL INFO Channel 20/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-166-244:897242:897730 [2] NCCL INFO Channel 22/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-166-36:104432:104922 [3] NCCL INFO Channel 22/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-166-244:897240:897735 [0] NCCL INFO Channel 22/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-36:104430:104924 [1] NCCL INFO Connected all rings
+ip-26-0-166-36:104430:104924 [1] NCCL INFO Connected all trees
+ip-26-0-166-36:104430:104924 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-36:104430:104924 [1] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-36:104430:104924 [1] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-166-244:897243:897731 [3] NCCL INFO Channel 22/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-166-244:897246:897734 [6] NCCL INFO Channel 21/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-166-244:897242:897730 [2] NCCL INFO Channel 23/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-166-244:897241:897737 [1] NCCL INFO Channel 22/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-244:897243:897731 [3] NCCL INFO Channel 23/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-166-36:104432:104922 [3] NCCL INFO Channel 23/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-166-244:897246:897734 [6] NCCL INFO Channel 22/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-166-244:897241:897737 [1] NCCL INFO Channel 23/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-244:897240:897735 [0] NCCL INFO Channel 23/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-244:897247:897736 [7] NCCL INFO Channel 16/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-166-244:897246:897734 [6] NCCL INFO Channel 23/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-166-244:897247:897736 [7] NCCL INFO Channel 17/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-166-244:897247:897736 [7] NCCL INFO Channel 18/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-166-244:897247:897736 [7] NCCL INFO Channel 19/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-166-36:104430:104924 [1] NCCL INFO comm 0xbbba9f0 rank 1 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0xfe3a2bff210edb10 - Init COMPLETE
+ip-26-0-166-36:104429:104919 [0] NCCL INFO comm 0xb951280 rank 0 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0xfe3a2bff210edb10 - Init COMPLETE
+ip-26-0-166-15:127292:127782 [2] NCCL INFO Connected all rings
+ip-26-0-166-15:127292:127782 [2] NCCL INFO Connected all trees
+ip-26-0-166-15:127292:127782 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-15:127292:127782 [2] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-15:127292:127782 [2] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-166-244:897247:897736 [7] NCCL INFO Channel 20/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-166-244:897247:897736 [7] NCCL INFO Channel 21/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-166-125:762818:763321 [3] NCCL INFO Connected all rings
+ip-26-0-166-125:762818:763321 [3] NCCL INFO Connected all trees
+ip-26-0-166-125:762818:763321 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-125:762818:763321 [3] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-125:762818:763321 [3] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-166-244:897247:897736 [7] NCCL INFO Channel 22/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-162-46:702886:703405 [6] NCCL INFO Connected all rings
+ip-26-0-162-46:702886:703405 [6] NCCL INFO Connected all trees
+ip-26-0-162-46:702886:703405 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-162-46:702886:703405 [6] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-162-46:702886:703405 [6] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-166-244:897247:897736 [7] NCCL INFO Channel 23/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-166-15:127290:127785 [0] NCCL INFO Connected all rings
+ip-26-0-166-15:127290:127785 [0] NCCL INFO Connected all trees
+ip-26-0-166-15:127290:127785 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-15:127290:127785 [0] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-15:127290:127785 [0] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-166-15:127293:127786 [3] NCCL INFO Connected all rings
+ip-26-0-166-15:127293:127786 [3] NCCL INFO Connected all trees
+ip-26-0-166-15:127293:127786 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-15:127293:127786 [3] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-15:127293:127786 [3] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-166-125:762817:763317 [2] NCCL INFO Connected all rings
+ip-26-0-166-125:762817:763317 [2] NCCL INFO Connected all trees
+ip-26-0-166-125:762817:763317 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-125:762817:763317 [2] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-125:762817:763317 [2] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-166-36:104433:104915 [4] NCCL INFO Connected all rings
+ip-26-0-166-36:104433:104915 [4] NCCL INFO Connected all trees
+ip-26-0-166-36:104433:104915 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-36:104433:104915 [4] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-36:104433:104915 [4] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-162-46:702884:703402 [4] NCCL INFO Connected all rings
+ip-26-0-162-46:702884:703402 [4] NCCL INFO Connected all trees
+ip-26-0-162-46:702884:703402 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-162-46:702884:703402 [4] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-162-46:702884:703402 [4] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-162-46:702887:703406 [7] NCCL INFO Connected all rings
+ip-26-0-162-46:702887:703406 [7] NCCL INFO Connected all trees
+ip-26-0-162-46:702887:703406 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-162-46:702887:703406 [7] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-162-46:702887:703406 [7] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-165-213:9632:10121 [7] NCCL INFO Connected all rings
+ip-26-0-165-213:9632:10121 [7] NCCL INFO Connected all trees
+ip-26-0-165-213:9632:10121 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-213:9632:10121 [7] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-165-213:9632:10121 [7] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-166-36:104434:104921 [5] NCCL INFO Connected all rings
+ip-26-0-166-36:104434:104921 [5] NCCL INFO Connected all trees
+ip-26-0-166-36:104434:104921 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-36:104434:104921 [5] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-36:104434:104921 [5] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-166-15:127293:127786 [3] NCCL INFO comm 0xb5fc0c0 rank 1 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0xe182481c705f52fd - Init COMPLETE
+ip-26-0-166-15:127292:127782 [2] NCCL INFO comm 0xb14f780 rank 0 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0xe182481c705f52fd - Init COMPLETE
+ip-26-0-166-125:762818:763321 [3] NCCL INFO comm 0x9c2c270 rank 1 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0xed427556a7553275 - Init COMPLETE
+ip-26-0-166-125:762817:763317 [2] NCCL INFO comm 0xc343aa0 rank 0 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0xed427556a7553275 - Init COMPLETE
+ip-26-0-165-213:9631:10117 [6] NCCL INFO Connected all rings
+ip-26-0-165-213:9631:10117 [6] NCCL INFO Connected all trees
+ip-26-0-165-213:9631:10117 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-213:9631:10117 [6] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-165-213:9631:10117 [6] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-165-213:9627:10116 [2] NCCL INFO Connected all rings
+ip-26-0-165-213:9627:10116 [2] NCCL INFO Connected all trees
+ip-26-0-165-213:9627:10116 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-213:9627:10116 [2] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-165-213:9627:10116 [2] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-165-213:9625:10112 [0] NCCL INFO Connected all rings
+ip-26-0-165-213:9625:10112 [0] NCCL INFO Connected all trees
+ip-26-0-165-213:9625:10112 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-213:9625:10112 [0] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-165-213:9625:10112 [0] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-166-36:104434:104921 [5] NCCL INFO comm 0x9f5a360 rank 1 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0xbd2039d4027c0ed6 - Init COMPLETE
+ip-26-0-166-36:104433:104915 [4] NCCL INFO comm 0xb73d350 rank 0 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0xbd2039d4027c0ed6 - Init COMPLETE
+ip-26-0-166-15:127291:127787 [1] NCCL INFO Connected all rings
+ip-26-0-166-15:127291:127787 [1] NCCL INFO Connected all trees
+ip-26-0-166-15:127291:127787 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-15:127291:127787 [1] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-15:127291:127787 [1] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-165-213:9626:10118 [1] NCCL INFO Connected all rings
+ip-26-0-165-213:9626:10118 [1] NCCL INFO Connected all trees
+ip-26-0-165-213:9626:10118 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-213:9626:10118 [1] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-165-213:9626:10118 [1] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-162-46:702882:703408 [2] NCCL INFO Connected all rings
+ip-26-0-162-46:702882:703408 [2] NCCL INFO Connected all trees
+ip-26-0-162-46:702882:703408 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-162-46:702882:703408 [2] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-162-46:702882:703408 [2] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-166-125:762815:763320 [0] NCCL INFO Connected all rings
+ip-26-0-166-125:762815:763320 [0] NCCL INFO Connected all trees
+ip-26-0-166-125:762815:763320 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-125:762815:763320 [0] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-125:762815:763320 [0] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-162-46:702886:703405 [6] NCCL INFO comm 0xa6f9960 rank 0 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0x80f2837194b47716 - Init COMPLETE
+ip-26-0-162-46:702887:703406 [7] NCCL INFO comm 0xa204fc0 rank 1 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0x80f2837194b47716 - Init COMPLETE
+ip-26-0-162-46:702885:703404 [5] NCCL INFO Connected all rings
+ip-26-0-162-46:702885:703404 [5] NCCL INFO Connected all trees
+ip-26-0-162-46:702885:703404 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-162-46:702885:703404 [5] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-162-46:702885:703404 [5] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-162-46:702883:703409 [3] NCCL INFO Connected all rings
+ip-26-0-162-46:702883:703409 [3] NCCL INFO Connected all trees
+ip-26-0-162-46:702883:703409 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-162-46:702883:703409 [3] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-162-46:702883:703409 [3] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-165-213:9628:10120 [3] NCCL INFO Connected all rings
+ip-26-0-165-213:9628:10120 [3] NCCL INFO Connected all trees
+ip-26-0-165-213:9628:10120 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-213:9628:10120 [3] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-165-213:9628:10120 [3] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-165-213:9632:10121 [7] NCCL INFO comm 0xa1e4a30 rank 1 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0x91023a3acfb2d892 - Init COMPLETE
+ip-26-0-165-213:9631:10117 [6] NCCL INFO comm 0xafbbfa0 rank 0 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0x91023a3acfb2d892 - Init COMPLETE
+ip-26-0-165-213:9625:10112 [0] NCCL INFO comm 0xad2f5c0 rank 0 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0xef34ee973e904713 - Init COMPLETE
+ip-26-0-165-213:9626:10118 [1] NCCL INFO comm 0xa8af270 rank 1 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0xef34ee973e904713 - Init COMPLETE
+ip-26-0-166-125:762822:763323 [7] NCCL INFO Connected all rings
+ip-26-0-166-125:762822:763323 [7] NCCL INFO Connected all trees
+ip-26-0-166-125:762822:763323 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-125:762822:763323 [7] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-125:762822:763323 [7] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-165-213:9629:10113 [4] NCCL INFO Connected all rings
+ip-26-0-165-213:9629:10113 [4] NCCL INFO Connected all trees
+ip-26-0-165-213:9629:10113 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-213:9629:10113 [4] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-165-213:9629:10113 [4] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-166-36:104432:104922 [3] NCCL INFO Connected all rings
+ip-26-0-166-36:104432:104922 [3] NCCL INFO Connected all trees
+ip-26-0-166-36:104432:104922 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-36:104432:104922 [3] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-36:104432:104922 [3] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-166-15:127290:127785 [0] NCCL INFO comm 0xa8e93b0 rank 0 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0x13ca242df769a19e - Init COMPLETE
+ip-26-0-166-15:127291:127787 [1] NCCL INFO comm 0xc163200 rank 1 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0x13ca242df769a19e - Init COMPLETE
+ip-26-0-162-46:702882:703408 [2] NCCL INFO comm 0xae6fb70 rank 0 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0x38b4054e8ec636f3 - Init COMPLETE
+ip-26-0-162-46:702883:703409 [3] NCCL INFO comm 0x9c2f910 rank 1 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0x38b4054e8ec636f3 - Init COMPLETE
+ip-26-0-165-213:9630:10119 [5] NCCL INFO Connected all rings
+ip-26-0-165-213:9630:10119 [5] NCCL INFO Connected all trees
+ip-26-0-165-213:9630:10119 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-213:9630:10119 [5] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-165-213:9630:10119 [5] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-162-46:702881:703400 [1] NCCL INFO Connected all rings
+ip-26-0-162-46:702881:703400 [1] NCCL INFO Connected all trees
+ip-26-0-162-46:702881:703400 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-162-46:702881:703400 [1] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-162-46:702881:703400 [1] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-165-213:9628:10120 [3] NCCL INFO comm 0xa7fc870 rank 1 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0x1e66d5e015dcfb4 - Init COMPLETE
+ip-26-0-165-213:9627:10116 [2] NCCL INFO comm 0x9e12000 rank 0 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0x1e66d5e015dcfb4 - Init COMPLETE
+ip-26-0-166-15:127296:127778 [6] NCCL INFO Connected all rings
+ip-26-0-166-15:127296:127778 [6] NCCL INFO Connected all trees
+ip-26-0-166-15:127296:127778 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-15:127296:127778 [6] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-15:127296:127778 [6] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-166-36:104431:104917 [2] NCCL INFO Connected all rings
+ip-26-0-166-36:104431:104917 [2] NCCL INFO Connected all trees
+ip-26-0-166-36:104431:104917 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-36:104431:104917 [2] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-36:104431:104917 [2] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-162-46:702884:703402 [4] NCCL INFO comm 0xb014080 rank 0 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0x6a6e536d58c1344e - Init COMPLETE
+ip-26-0-162-46:702885:703404 [5] NCCL INFO comm 0xa213dc0 rank 1 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0x6a6e536d58c1344e - Init COMPLETE
+ip-26-0-162-46:702880:703399 [0] NCCL INFO Connected all rings
+ip-26-0-162-46:702880:703399 [0] NCCL INFO Connected all trees
+ip-26-0-162-46:702880:703399 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-162-46:702880:703399 [0] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-162-46:702880:703399 [0] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-166-15:127297:127781 [7] NCCL INFO Connected all rings
+ip-26-0-166-15:127297:127781 [7] NCCL INFO Connected all trees
+ip-26-0-166-15:127297:127781 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-15:127297:127781 [7] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-15:127297:127781 [7] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-165-213:9629:10113 [4] NCCL INFO comm 0x9e0c430 rank 0 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0xa0ff12f927ac5392 - Init COMPLETE
+ip-26-0-165-213:9630:10119 [5] NCCL INFO comm 0x9c8b120 rank 1 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0xa0ff12f927ac5392 - Init COMPLETE
+ip-26-0-166-36:104432:104922 [3] NCCL INFO comm 0x9cee960 rank 1 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0x58fc109f3722ddb0 - Init COMPLETE
+ip-26-0-166-36:104431:104917 [2] NCCL INFO comm 0xacf6020 rank 0 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0x58fc109f3722ddb0 - Init COMPLETE
+ip-26-0-166-125:762816:763324 [1] NCCL INFO Connected all rings
+ip-26-0-166-125:762816:763324 [1] NCCL INFO Connected all trees
+ip-26-0-166-125:762816:763324 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-125:762816:763324 [1] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-125:762816:763324 [1] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-162-46:702881:703400 [1] NCCL INFO comm 0xb50f5b0 rank 1 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0x6a967c92be24b549 - Init COMPLETE
+ip-26-0-162-46:702880:703399 [0] NCCL INFO comm 0xba3bc80 rank 0 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0x6a967c92be24b549 - Init COMPLETE
+ip-26-0-166-125:762819:763316 [4] NCCL INFO Connected all rings
+ip-26-0-166-125:762819:763316 [4] NCCL INFO Connected all trees
+ip-26-0-166-125:762819:763316 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-125:762819:763316 [4] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-125:762819:763316 [4] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-166-15:127295:127783 [5] NCCL INFO Connected all rings
+ip-26-0-166-15:127295:127783 [5] NCCL INFO Connected all trees
+ip-26-0-166-15:127295:127783 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-15:127295:127783 [5] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-15:127295:127783 [5] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-166-15:127296:127778 [6] NCCL INFO comm 0xb5f4350 rank 0 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0xaa23d81ecb1fca47 - Init COMPLETE
+ip-26-0-166-15:127297:127781 [7] NCCL INFO comm 0xbd0c380 rank 1 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0xaa23d81ecb1fca47 - Init COMPLETE
+ip-26-0-166-125:762821:763318 [6] NCCL INFO Connected all rings
+ip-26-0-166-125:762821:763318 [6] NCCL INFO Connected all trees
+ip-26-0-166-125:762821:763318 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-125:762821:763318 [6] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-125:762821:763318 [6] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-166-15:127294:127779 [4] NCCL INFO Connected all rings
+ip-26-0-166-15:127294:127779 [4] NCCL INFO Connected all trees
+ip-26-0-166-15:127294:127779 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-15:127294:127779 [4] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-15:127294:127779 [4] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-166-125:762815:763320 [0] NCCL INFO comm 0x9ac09f0 rank 0 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0x5be10b8c3d993573 - Init COMPLETE
+ip-26-0-166-125:762816:763324 [1] NCCL INFO comm 0xb545ab0 rank 1 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0x5be10b8c3d993573 - Init COMPLETE
+ip-26-0-166-125:762820:763322 [5] NCCL INFO Connected all rings
+ip-26-0-166-125:762820:763322 [5] NCCL INFO Connected all trees
+ip-26-0-166-125:762820:763322 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-125:762820:763322 [5] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-125:762820:763322 [5] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-166-214:624946:625439 [3] NCCL INFO Connected all rings
+ip-26-0-166-214:624946:625439 [3] NCCL INFO Connected all trees
+ip-26-0-166-214:624946:625439 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-214:624946:625439 [3] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-214:624946:625439 [3] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-166-15:127295:127783 [5] NCCL INFO comm 0x9f8be80 rank 1 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0x41ee8bd379d4b112 - Init COMPLETE
+ip-26-0-166-15:127294:127779 [4] NCCL INFO comm 0xa49af50 rank 0 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0x41ee8bd379d4b112 - Init COMPLETE
+ip-26-0-166-125:762822:763323 [7] NCCL INFO comm 0xb4d8070 rank 1 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0xac2405b00a4fa8a0 - Init COMPLETE
+ip-26-0-166-125:762821:763318 [6] NCCL INFO comm 0x9d9cf30 rank 0 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0xac2405b00a4fa8a0 - Init COMPLETE
+ip-26-0-166-125:762820:763322 [5] NCCL INFO comm 0xa7f6080 rank 1 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0x1ea69f320f0c17f - Init COMPLETE
+ip-26-0-166-125:762819:763316 [4] NCCL INFO comm 0xb32ed70 rank 0 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0x1ea69f320f0c17f - Init COMPLETE
+ip-26-0-166-214:624945:625436 [2] NCCL INFO Connected all rings
+ip-26-0-166-214:624945:625436 [2] NCCL INFO Connected all trees
+ip-26-0-166-214:624945:625436 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-214:624945:625436 [2] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-214:624945:625436 [2] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-166-214:624949:625431 [6] NCCL INFO Connected all rings
+ip-26-0-166-214:624949:625431 [6] NCCL INFO Connected all trees
+ip-26-0-166-214:624949:625431 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-214:624949:625431 [6] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-214:624949:625431 [6] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-166-214:624950:625434 [7] NCCL INFO Connected all rings
+ip-26-0-166-214:624950:625434 [7] NCCL INFO Connected all trees
+ip-26-0-166-214:624950:625434 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-214:624950:625434 [7] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-214:624950:625434 [7] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-166-214:624946:625439 [3] NCCL INFO comm 0xb53cb40 rank 1 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0x73d5dfca7b5791eb - Init COMPLETE
+ip-26-0-166-214:624945:625436 [2] NCCL INFO comm 0xaca50f0 rank 0 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0x73d5dfca7b5791eb - Init COMPLETE
+ip-26-0-167-9:1039540:1040048 [3] NCCL INFO Connected all rings
+ip-26-0-167-9:1039540:1040048 [3] NCCL INFO Connected all trees
+ip-26-0-167-9:1039540:1040048 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-167-9:1039540:1040048 [3] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-167-9:1039540:1040048 [3] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-166-214:624944:625440 [1] NCCL INFO Connected all rings
+ip-26-0-166-214:624944:625440 [1] NCCL INFO Connected all trees
+ip-26-0-166-214:624944:625440 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-214:624944:625440 [1] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-214:624944:625440 [1] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-166-214:624943:625438 [0] NCCL INFO Connected all rings
+ip-26-0-166-214:624943:625438 [0] NCCL INFO Connected all trees
+ip-26-0-166-214:624943:625438 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-214:624943:625438 [0] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-214:624943:625438 [0] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-167-9:1039539:1040045 [2] NCCL INFO Connected all rings
+ip-26-0-167-9:1039539:1040045 [2] NCCL INFO Connected all trees
+ip-26-0-167-9:1039539:1040045 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-167-9:1039539:1040045 [2] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-167-9:1039539:1040045 [2] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-166-214:624947:625432 [4] NCCL INFO Connected all rings
+ip-26-0-166-214:624947:625432 [4] NCCL INFO Connected all trees
+ip-26-0-166-214:624947:625432 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-214:624947:625432 [4] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-214:624947:625432 [4] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-166-214:624948:625435 [5] NCCL INFO Connected all rings
+ip-26-0-166-214:624948:625435 [5] NCCL INFO Connected all trees
+ip-26-0-166-214:624948:625435 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-214:624948:625435 [5] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-214:624948:625435 [5] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-166-214:624949:625431 [6] NCCL INFO comm 0xa9c4a30 rank 0 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0xa0ece3c426d66ea9 - Init COMPLETE
+ip-26-0-166-214:624950:625434 [7] NCCL INFO comm 0xa28ff10 rank 1 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0xa0ece3c426d66ea9 - Init COMPLETE
+ip-26-0-167-9:1039538:1040044 [1] NCCL INFO Connected all rings
+ip-26-0-167-9:1039538:1040044 [1] NCCL INFO Connected all trees
+ip-26-0-167-9:1039538:1040044 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-167-9:1039538:1040044 [1] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-167-9:1039538:1040044 [1] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-166-214:624943:625438 [0] NCCL INFO comm 0xb38d4c0 rank 0 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0x4a9d894e5cf2f81c - Init COMPLETE
+ip-26-0-166-214:624944:625440 [1] NCCL INFO comm 0xa45b7c0 rank 1 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0x4a9d894e5cf2f81c - Init COMPLETE
+ip-26-0-166-214:624947:625432 [4] NCCL INFO comm 0xb4dd0e0 rank 0 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0x68fdf1bd57245b59 - Init COMPLETE
+ip-26-0-166-214:624948:625435 [5] NCCL INFO comm 0xbfa5b60 rank 1 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0x68fdf1bd57245b59 - Init COMPLETE
+ip-26-0-166-244:897245:897729 [5] NCCL INFO Connected all rings
+ip-26-0-166-244:897245:897729 [5] NCCL INFO Connected all trees
+ip-26-0-166-244:897245:897729 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-244:897245:897729 [5] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-244:897245:897729 [5] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-166-244:897244:897727 [4] NCCL INFO Connected all rings
+ip-26-0-166-244:897244:897727 [4] NCCL INFO Connected all trees
+ip-26-0-166-244:897244:897727 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-244:897244:897727 [4] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-244:897244:897727 [4] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-167-9:1039539:1040045 [2] NCCL INFO comm 0x9be8210 rank 0 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0x486d4b076e52af12 - Init COMPLETE
+ip-26-0-167-9:1039540:1040048 [3] NCCL INFO comm 0xaca56b0 rank 1 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0x486d4b076e52af12 - Init COMPLETE
+ip-26-0-166-244:897245:897729 [5] NCCL INFO comm 0xafad090 rank 1 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0x4a1841ee5821a4e0 - Init COMPLETE
+ip-26-0-166-244:897244:897727 [4] NCCL INFO comm 0xade7470 rank 0 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0x4a1841ee5821a4e0 - Init COMPLETE
+ip-26-0-166-244:897243:897731 [3] NCCL INFO Connected all rings
+ip-26-0-166-244:897243:897731 [3] NCCL INFO Connected all trees
+ip-26-0-166-244:897243:897731 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-244:897243:897731 [3] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-244:897243:897731 [3] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-166-244:897242:897730 [2] NCCL INFO Connected all rings
+ip-26-0-166-244:897242:897730 [2] NCCL INFO Connected all trees
+ip-26-0-166-244:897242:897730 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-244:897242:897730 [2] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-244:897242:897730 [2] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-166-244:897241:897737 [1] NCCL INFO Connected all rings
+ip-26-0-166-244:897241:897737 [1] NCCL INFO Connected all trees
+ip-26-0-166-244:897241:897737 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-244:897241:897737 [1] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-244:897241:897737 [1] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-167-9:1039543:1040043 [6] NCCL INFO Connected all rings
+ip-26-0-167-9:1039543:1040043 [6] NCCL INFO Connected all trees
+ip-26-0-167-9:1039543:1040043 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-167-9:1039543:1040043 [6] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-167-9:1039543:1040043 [6] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-166-244:897243:897731 [3] NCCL INFO comm 0x9f01bc0 rank 1 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0x8e1ccf03aca53ed8 - Init COMPLETE
+ip-26-0-166-244:897242:897730 [2] NCCL INFO comm 0xa64f180 rank 0 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0x8e1ccf03aca53ed8 - Init COMPLETE
+NCCL version 2.18.6+cuda12.1
+ip-26-0-162-46:702882:703433 [2] NCCL INFO Using network Libfabric
+ip-26-0-166-214:624946:625462 [3] NCCL INFO Using network Libfabric
+ip-26-0-166-125:762818:763344 [3] NCCL INFO Using network Libfabric
+ip-26-0-166-36:104432:104944 [3] NCCL INFO Using network Libfabric
+ip-26-0-166-244:897243:897754 [3] NCCL INFO Using network Libfabric
+ip-26-0-166-244:897240:897735 [0] NCCL INFO Connected all rings
+ip-26-0-166-244:897240:897735 [0] NCCL INFO Connected all trees
+ip-26-0-166-244:897240:897735 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-15:127293:127806 [3] NCCL INFO Using network Libfabric
+ip-26-0-166-244:897240:897735 [0] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-244:897240:897735 [0] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-166-214:624945:625463 [2] NCCL INFO Using network Libfabric
+ip-26-0-165-213:9628:10141 [3] NCCL INFO Using network Libfabric
+ip-26-0-166-15:127292:127807 [2] NCCL INFO Using network Libfabric
+ip-26-0-162-46:702883:703431 [3] NCCL INFO Using network Libfabric
+ip-26-0-166-125:762817:763345 [2] NCCL INFO Using network Libfabric
+ip-26-0-166-244:897242:897755 [2] NCCL INFO Using network Libfabric
+ip-26-0-165-213:9627:10142 [2] NCCL INFO Using network Libfabric
+ip-26-0-167-9:1039540:1040064 [3] NCCL INFO Using network Libfabric
+ip-26-0-167-9:1039539:1040065 [2] NCCL INFO Using network Libfabric
+ip-26-0-166-36:104431:104945 [2] NCCL INFO Using network Libfabric
+ip-26-0-162-46:702882:703433 [2] NCCL INFO comm 0xae72820 rank 0 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0x53075d22960e36e3 - Init START
+ip-26-0-167-9:1039539:1040065 [2] NCCL INFO comm 0x9beaec0 rank 7 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0x53075d22960e36e3 - Init START
+ip-26-0-162-46:702883:703431 [3] NCCL INFO comm 0x9c325c0 rank 0 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0xa976fd02a70fed12 - Init START
+ip-26-0-165-213:9627:10142 [2] NCCL INFO comm 0x9e14cb0 rank 1 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0x53075d22960e36e3 - Init START
+ip-26-0-166-125:762817:763345 [2] NCCL INFO comm 0xb8ed520 rank 2 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0x53075d22960e36e3 - Init START
+ip-26-0-166-244:897242:897755 [2] NCCL INFO comm 0x9bfa1f0 rank 5 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0x53075d22960e36e3 - Init START
+ip-26-0-166-15:127292:127807 [2] NCCL INFO comm 0xa6fa4e0 rank 3 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0x53075d22960e36e3 - Init START
+ip-26-0-166-36:104431:104945 [2] NCCL INFO comm 0xacf8cd0 rank 6 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0x53075d22960e36e3 - Init START
+ip-26-0-165-213:9628:10141 [3] NCCL INFO comm 0xa7ff520 rank 1 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0xa976fd02a70fed12 - Init START
+ip-26-0-166-214:624945:625463 [2] NCCL INFO comm 0xaca7da0 rank 4 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0x53075d22960e36e3 - Init START
+ip-26-0-167-9:1039540:1040064 [3] NCCL INFO comm 0xa261180 rank 7 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0xa976fd02a70fed12 - Init START
+ip-26-0-166-125:762818:763344 [3] NCCL INFO comm 0x9c2ef20 rank 2 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0xa976fd02a70fed12 - Init START
+ip-26-0-166-15:127293:127806 [3] NCCL INFO comm 0xb5fed70 rank 3 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0xa976fd02a70fed12 - Init START
+ip-26-0-166-36:104432:104944 [3] NCCL INFO comm 0x9cf1610 rank 6 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0xa976fd02a70fed12 - Init START
+ip-26-0-166-244:897243:897754 [3] NCCL INFO comm 0x9f04870 rank 5 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0xa976fd02a70fed12 - Init START
+ip-26-0-166-214:624946:625462 [3] NCCL INFO comm 0xb53f7f0 rank 4 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0xa976fd02a70fed12 - Init START
+ip-26-0-162-46:702882:703433 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-36:104431:104945 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-167-9:1039539:1040065 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-125:762817:763345 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-165-213:9627:10142 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-214:624945:625463 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-15:127292:127807 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-162-46:702883:703431 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-36:104432:104944 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-244:897242:897755 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-167-9:1039540:1040064 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-125:762818:763344 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-165-213:9628:10141 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-15:127293:127806 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-214:624946:625462 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-244:897243:897754 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-244:897241:897737 [1] NCCL INFO comm 0xc105c70 rank 1 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0xb3cd48026c577012 - Init COMPLETE
+ip-26-0-166-244:897240:897735 [0] NCCL INFO comm 0xa80e050 rank 0 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0xb3cd48026c577012 - Init COMPLETE
+ip-26-0-166-244:897246:897734 [6] NCCL INFO Connected all rings
+ip-26-0-166-244:897246:897734 [6] NCCL INFO Connected all trees
+ip-26-0-166-244:897246:897734 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-244:897246:897734 [6] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-244:897246:897734 [6] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-167-9:1039544:1040046 [7] NCCL INFO Connected all rings
+ip-26-0-167-9:1039544:1040046 [7] NCCL INFO Connected all trees
+ip-26-0-167-9:1039544:1040046 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-167-9:1039544:1040046 [7] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-167-9:1039544:1040046 [7] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-167-9:1039537:1040038 [0] NCCL INFO Connected all rings
+ip-26-0-167-9:1039537:1040038 [0] NCCL INFO Connected all trees
+ip-26-0-167-9:1039537:1040038 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-167-9:1039537:1040038 [0] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-167-9:1039537:1040038 [0] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-166-244:897247:897736 [7] NCCL INFO Connected all rings
+ip-26-0-166-244:897247:897736 [7] NCCL INFO Connected all trees
+ip-26-0-166-244:897247:897736 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-244:897247:897736 [7] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-244:897247:897736 [7] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-167-9:1039544:1040046 [7] NCCL INFO comm 0xa16c820 rank 1 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0x74262946ed30693c - Init COMPLETE
+ip-26-0-167-9:1039543:1040043 [6] NCCL INFO comm 0xaa35a60 rank 0 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0x74262946ed30693c - Init COMPLETE
+ip-26-0-166-244:897246:897734 [6] NCCL INFO comm 0xc371140 rank 0 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0xf091798745a84edd - Init COMPLETE
+ip-26-0-166-244:897247:897736 [7] NCCL INFO comm 0xb8f3eb0 rank 1 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0xf091798745a84edd - Init COMPLETE
+NCCL version 2.18.6+cuda12.1
+ip-26-0-162-46:702887:703436 [7] NCCL INFO Using network Libfabric
+ip-26-0-166-125:762822:763346 [7] NCCL INFO Using network Libfabric
+ip-26-0-166-244:897247:897759 [7] NCCL INFO Using network Libfabric
+ip-26-0-162-46:702886:703437 [6] NCCL INFO Using network Libfabric
+ip-26-0-166-125:762821:763347 [6] NCCL INFO Using network Libfabric
+ip-26-0-166-214:624950:625464 [7] NCCL INFO Using network Libfabric
+ip-26-0-167-9:1039544:1040068 [7] NCCL INFO Using network Libfabric
+ip-26-0-165-213:9632:10143 [7] NCCL INFO Using network Libfabric
+ip-26-0-166-36:104435:104947 [6] NCCL INFO Using network Libfabric
+ip-26-0-166-214:624949:625465 [6] NCCL INFO Using network Libfabric
+ip-26-0-166-15:127296:127809 [6] NCCL INFO Using network Libfabric
+ip-26-0-166-15:127297:127808 [7] NCCL INFO Using network Libfabric
+ip-26-0-166-36:104436:104946 [7] NCCL INFO Using network Libfabric
+ip-26-0-165-213:9631:10144 [6] NCCL INFO Using network Libfabric
+ip-26-0-167-9:1039543:1040069 [6] NCCL INFO Using network Libfabric
+ip-26-0-166-244:897246:897760 [6] NCCL INFO Using network Libfabric
+ip-26-0-167-9:1039537:1040038 [0] NCCL INFO comm 0xa6f52b0 rank 0 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0x62fffa362beddbec - Init COMPLETE
+ip-26-0-167-9:1039538:1040044 [1] NCCL INFO comm 0x9b103b0 rank 1 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0x62fffa362beddbec - Init COMPLETE
+ip-26-0-167-9:1039544:1040068 [7] NCCL INFO comm 0xa16f4d0 rank 7 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0x30fc6b5205aced7 - Init START
+ip-26-0-162-46:702887:703436 [7] NCCL INFO comm 0xa207c70 rank 0 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0x30fc6b5205aced7 - Init START
+ip-26-0-165-213:9632:10143 [7] NCCL INFO comm 0xa1e76e0 rank 1 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0x30fc6b5205aced7 - Init START
+ip-26-0-166-125:762822:763346 [7] NCCL INFO comm 0xb4dad20 rank 2 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0x30fc6b5205aced7 - Init START
+ip-26-0-166-15:127297:127808 [7] NCCL INFO comm 0xb2b5da0 rank 3 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0x30fc6b5205aced7 - Init START
+ip-26-0-166-214:624950:625464 [7] NCCL INFO comm 0xa292bc0 rank 4 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0x30fc6b5205aced7 - Init START
+ip-26-0-162-46:702886:703437 [6] NCCL INFO comm 0xa6fc610 rank 0 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0xd4a12c6dc0dd1f26 - Init START
+ip-26-0-167-9:1039543:1040069 [6] NCCL INFO comm 0xaa38710 rank 7 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0xd4a12c6dc0dd1f26 - Init START
+ip-26-0-166-36:104436:104946 [7] NCCL INFO comm 0xac26b70 rank 6 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0x30fc6b5205aced7 - Init START
+ip-26-0-166-244:897247:897759 [7] NCCL INFO comm 0xae9e340 rank 5 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0x30fc6b5205aced7 - Init START
+ip-26-0-165-213:9631:10144 [6] NCCL INFO comm 0xafbec50 rank 1 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0xd4a12c6dc0dd1f26 - Init START
+ip-26-0-166-125:762821:763347 [6] NCCL INFO comm 0x9d9fbe0 rank 2 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0xd4a12c6dc0dd1f26 - Init START
+ip-26-0-166-15:127296:127809 [6] NCCL INFO comm 0xab9e5d0 rank 3 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0xd4a12c6dc0dd1f26 - Init START
+ip-26-0-166-214:624949:625465 [6] NCCL INFO comm 0x9f6ead0 rank 4 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0xd4a12c6dc0dd1f26 - Init START
+ip-26-0-166-36:104435:104947 [6] NCCL INFO comm 0xa2b5600 rank 6 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0xd4a12c6dc0dd1f26 - Init START
+ip-26-0-166-244:897246:897760 [6] NCCL INFO comm 0xb91b050 rank 5 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0xd4a12c6dc0dd1f26 - Init START
+ip-26-0-166-36:104436:104946 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-162-46:702887:703436 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-167-9:1039544:1040068 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-165-213:9632:10143 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-125:762822:763346 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-15:127297:127808 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-36:104435:104947 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-214:624950:625464 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-162-46:702886:703437 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-244:897247:897759 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-167-9:1039543:1040069 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-165-213:9631:10144 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-15:127296:127809 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-125:762821:763347 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-214:624949:625465 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-244:897246:897760 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+NCCL version 2.18.6+cuda12.1
+ip-26-0-162-46:702880:703439 [0] NCCL INFO Using network Libfabric
+ip-26-0-165-213:9625:10145 [0] NCCL INFO Using network Libfabric
+ip-26-0-166-125:762815:763348 [0] NCCL INFO Using network Libfabric
+ip-26-0-167-9:1039537:1040070 [0] NCCL INFO Using network Libfabric
+ip-26-0-166-214:624943:625466 [0] NCCL INFO Using network Libfabric
+ip-26-0-166-244:897240:897761 [0] NCCL INFO Using network Libfabric
+ip-26-0-166-15:127290:127810 [0] NCCL INFO Using network Libfabric
+ip-26-0-166-36:104429:104948 [0] NCCL INFO Using network Libfabric
+ip-26-0-162-46:702881:703441 [1] NCCL INFO Using network Libfabric
+ip-26-0-165-213:9626:10146 [1] NCCL INFO Using network Libfabric
+ip-26-0-166-125:762816:763349 [1] NCCL INFO Using network Libfabric
+ip-26-0-166-36:104430:104949 [1] NCCL INFO Using network Libfabric
+ip-26-0-166-15:127291:127811 [1] NCCL INFO Using network Libfabric
+ip-26-0-166-214:624944:625467 [1] NCCL INFO Using network Libfabric
+ip-26-0-166-244:897241:897762 [1] NCCL INFO Using network Libfabric
+ip-26-0-167-9:1039538:1040071 [1] NCCL INFO Using network Libfabric
+ip-26-0-167-9:1039537:1040070 [0] NCCL INFO comm 0xa6f7f60 rank 7 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0x7ada0dd8e53058a4 - Init START
+ip-26-0-162-46:702880:703439 [0] NCCL INFO comm 0xafe5060 rank 0 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0x7ada0dd8e53058a4 - Init START
+ip-26-0-165-213:9625:10145 [0] NCCL INFO comm 0xa2d8ce0 rank 1 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0x7ada0dd8e53058a4 - Init START
+ip-26-0-166-125:762815:763348 [0] NCCL INFO comm 0x9ac36a0 rank 2 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0x7ada0dd8e53058a4 - Init START
+ip-26-0-166-214:624943:625466 [0] NCCL INFO comm 0xa9386c0 rank 4 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0x7ada0dd8e53058a4 - Init START
+ip-26-0-166-15:127290:127810 [0] NCCL INFO comm 0x9e92d70 rank 3 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0x7ada0dd8e53058a4 - Init START
+ip-26-0-166-36:104429:104948 [0] NCCL INFO comm 0xb953f30 rank 6 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0x7ada0dd8e53058a4 - Init START
+ip-26-0-166-244:897240:897761 [0] NCCL INFO comm 0xa810d00 rank 5 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0x7ada0dd8e53058a4 - Init START
+ip-26-0-162-46:702880:703439 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/349
+ip-26-0-166-36:104429:104948 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-167-9:1039537:1040070 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-165-213:9625:10145 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-244:897240:897761 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-125:762815:763348 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-15:127290:127810 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-214:624943:625466 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-167-9:1039538:1040071 [1] NCCL INFO comm 0x9b13060 rank 7 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0x2333452d75ea81bb - Init START
+ip-26-0-162-46:702881:703441 [1] NCCL INFO comm 0xaaba1b0 rank 0 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0x2333452d75ea81bb - Init START
+ip-26-0-165-213:9626:10146 [1] NCCL INFO comm 0xa8b1f20 rank 1 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0x2333452d75ea81bb - Init START
+ip-26-0-166-125:762816:763349 [1] NCCL INFO comm 0xaaf02d0 rank 2 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0x2333452d75ea81bb - Init START
+ip-26-0-166-15:127291:127811 [1] NCCL INFO comm 0xb70d770 rank 3 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0x2333452d75ea81bb - Init START
+ip-26-0-166-214:624944:625467 [1] NCCL INFO comm 0x9a071c0 rank 4 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0x2333452d75ea81bb - Init START
+ip-26-0-166-36:104430:104949 [1] NCCL INFO comm 0xb166740 rank 6 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0x2333452d75ea81bb - Init START
+ip-26-0-166-244:897241:897762 [1] NCCL INFO comm 0xb6b0bc0 rank 5 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0x2333452d75ea81bb - Init START
+ip-26-0-166-36:104430:104949 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-162-46:702881:703441 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-167-9:1039538:1040071 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-165-213:9626:10146 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-125:762816:763349 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-244:897241:897762 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-214:624944:625467 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-15:127291:127811 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-167-9:1039541:1040041 [4] NCCL INFO Connected all rings
+ip-26-0-167-9:1039541:1040041 [4] NCCL INFO Connected all trees
+ip-26-0-167-9:1039541:1040041 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-167-9:1039541:1040041 [4] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-167-9:1039541:1040041 [4] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-167-9:1039542:1040047 [5] NCCL INFO Connected all rings
+ip-26-0-167-9:1039542:1040047 [5] NCCL INFO Connected all trees
+ip-26-0-167-9:1039542:1040047 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-167-9:1039542:1040047 [5] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-167-9:1039542:1040047 [5] NCCL INFO 24 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-166-36:104431:104945 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-162-46:702882:703433 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-166-214:624945:625463 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-165-213:9628:10141 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-162-46:702883:703431 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-166-36:104432:104944 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-167-9:1039542:1040047 [5] NCCL INFO comm 0x99250b0 rank 1 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0x8f4515ff2b0a9fd3 - Init COMPLETE
+ip-26-0-167-9:1039541:1040041 [4] NCCL INFO comm 0xaf66470 rank 0 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0x8f4515ff2b0a9fd3 - Init COMPLETE
+NCCL version 2.18.6+cuda12.1
+ip-26-0-162-46:702884:703443 [4] NCCL INFO Using network Libfabric
+ip-26-0-166-125:762819:763351 [4] NCCL INFO Using network Libfabric
+ip-26-0-165-213:9629:10147 [4] NCCL INFO Using network Libfabric
+ip-26-0-166-15:127294:127812 [4] NCCL INFO Using network Libfabric
+ip-26-0-166-36:104433:104950 [4] NCCL INFO Using network Libfabric
+ip-26-0-166-214:624947:625469 [4] NCCL INFO Using network Libfabric
+ip-26-0-166-244:897244:897763 [4] NCCL INFO Using network Libfabric
+ip-26-0-167-9:1039541:1040074 [4] NCCL INFO Using network Libfabric
+ip-26-0-162-46:702885:703445 [5] NCCL INFO Using network Libfabric
+ip-26-0-166-214:624948:625470 [5] NCCL INFO Using network Libfabric
+ip-26-0-165-213:9630:10148 [5] NCCL INFO Using network Libfabric
+ip-26-0-166-15:127295:127813 [5] NCCL INFO Using network Libfabric
+ip-26-0-166-36:104434:104951 [5] NCCL INFO Using network Libfabric
+ip-26-0-166-125:762820:763352 [5] NCCL INFO Using network Libfabric
+ip-26-0-166-244:897245:897764 [5] NCCL INFO Using network Libfabric
+ip-26-0-167-9:1039542:1040075 [5] NCCL INFO Using network Libfabric
+ip-26-0-167-9:1039541:1040074 [4] NCCL INFO comm 0xaf69120 rank 7 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0x481461516cf15ed6 - Init START
+ip-26-0-162-46:702884:703443 [4] NCCL INFO comm 0xa5bdae0 rank 0 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0x481461516cf15ed6 - Init START
+ip-26-0-166-125:762819:763351 [4] NCCL INFO comm 0xa8d9010 rank 2 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0x481461516cf15ed6 - Init START
+ip-26-0-165-213:9629:10147 [4] NCCL INFO comm 0x9e0f0e0 rank 1 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0x481461516cf15ed6 - Init START
+ip-26-0-166-214:624947:625469 [4] NCCL INFO comm 0xb4dfd90 rank 4 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0x481461516cf15ed6 - Init START
+ip-26-0-166-15:127294:127812 [4] NCCL INFO comm 0xa49dc00 rank 3 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0x481461516cf15ed6 - Init START
+ip-26-0-166-244:897244:897763 [4] NCCL INFO comm 0xa391a00 rank 5 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0x481461516cf15ed6 - Init START
+ip-26-0-166-36:104433:104950 [4] NCCL INFO comm 0xb740000 rank 6 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0x481461516cf15ed6 - Init START
+ip-26-0-166-36:104433:104950 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-167-9:1039541:1040074 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-162-46:702884:703443 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-165-213:9629:10147 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-15:127294:127812 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-214:624947:625469 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-125:762819:763351 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-244:897244:897763 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-167-9:1039542:1040075 [5] NCCL INFO comm 0x9927d60 rank 7 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0xbc5bf908bde42540 - Init START
+ip-26-0-162-46:702885:703445 [5] NCCL INFO comm 0xa216a70 rank 0 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0xbc5bf908bde42540 - Init START
+ip-26-0-165-213:9630:10148 [5] NCCL INFO comm 0x9c8ddd0 rank 1 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0xbc5bf908bde42540 - Init START
+ip-26-0-166-36:104434:104951 [5] NCCL INFO comm 0x9f5d010 rank 6 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0xbc5bf908bde42540 - Init START
+ip-26-0-166-125:762820:763352 [5] NCCL INFO comm 0xa7f8d30 rank 2 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0xbc5bf908bde42540 - Init START
+ip-26-0-166-15:127295:127813 [5] NCCL INFO comm 0x9f8eb30 rank 3 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0xbc5bf908bde42540 - Init START
+ip-26-0-166-214:624948:625470 [5] NCCL INFO comm 0xb5510e0 rank 4 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0xbc5bf908bde42540 - Init START
+ip-26-0-166-244:897245:897764 [5] NCCL INFO comm 0xafafd40 rank 5 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0xbc5bf908bde42540 - Init START
+ip-26-0-166-36:104434:104951 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-167-9:1039542:1040075 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-162-46:702885:703445 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-244:897245:897764 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-165-213:9630:10148 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-125:762820:763352 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-214:624948:625470 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-15:127293:127806 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-166-15:127295:127813 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-214:624946:625462 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-166-15:127292:127807 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-165-213:9627:10142 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-166-125:762818:763344 [3] NCCL INFO Setting affinity for GPU 3 to 0fff,ffffffff
+ip-26-0-166-125:762817:763345 [2] NCCL INFO Setting affinity for GPU 2 to 0fff,ffffffff
+ip-26-0-166-244:897243:897754 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-166-244:897242:897755 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-166-244:897241:897762 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-166-36:104436:104946 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-166-214:624944:625467 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-165-213:9626:10146 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-166-214:624943:625466 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-165-213:9631:10144 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-166-36:104430:104949 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-165-213:9632:10143 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-167-9:1039539:1040065 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-167-9:1039539:1040065 [2] NCCL INFO Trees [0] -1/-1/-1->7->6 [1] -1/-1/-1->7->6 [2] 3/-1/-1->7->-1 [3] 3/-1/-1->7->-1
+ip-26-0-167-9:1039539:1040065 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-162-46:702882:703433 [2] NCCL INFO Channel 00/04 :    0   1   2   3   4   5   6   7
+ip-26-0-162-46:702882:703433 [2] NCCL INFO Channel 01/04 :    0   1   2   3   4   5   6   7
+ip-26-0-162-46:702882:703433 [2] NCCL INFO Channel 02/04 :    0   1   2   3   4   5   6   7
+ip-26-0-162-46:702882:703433 [2] NCCL INFO Channel 03/04 :    0   1   2   3   4   5   6   7
+ip-26-0-162-46:702882:703433 [2] NCCL INFO Trees [0] 4/-1/-1->0->-1 [1] 4/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-162-46:702882:703433 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-213:9627:10142 [2] NCCL INFO Trees [0] -1/-1/-1->1->2 [1] -1/-1/-1->1->2 [2] 2/0/-1->1->3 [3] 2/0/-1->1->3
+ip-26-0-165-213:9627:10142 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-125:762817:763345 [2] NCCL INFO Trees [0] 1/3/-1->2->4 [1] 1/3/-1->2->4 [2] -1/-1/-1->2->1 [3] -1/-1/-1->2->1
+ip-26-0-166-125:762817:763345 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-214:624945:625463 [2] NCCL INFO Trees [0] 2/6/-1->4->0 [1] 2/6/-1->4->0 [2] -1/-1/-1->4->5 [3] -1/-1/-1->4->5
+ip-26-0-166-214:624945:625463 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-15:127292:127807 [2] NCCL INFO Trees [0] -1/-1/-1->3->2 [1] -1/-1/-1->3->2 [2] 5/1/-1->3->7 [3] 5/1/-1->3->7
+ip-26-0-166-15:127292:127807 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-244:897242:897755 [2] NCCL INFO Trees [0] -1/-1/-1->5->6 [1] -1/-1/-1->5->6 [2] 6/4/-1->5->3 [3] 6/4/-1->5->3
+ip-26-0-166-244:897242:897755 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-36:104431:104945 [2] NCCL INFO Trees [0] 5/7/-1->6->4 [1] 5/7/-1->6->4 [2] -1/-1/-1->6->5 [3] -1/-1/-1->6->5
+ip-26-0-166-36:104431:104945 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-167-9:1039540:1040064 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-167-9:1039540:1040064 [3] NCCL INFO Trees [0] -1/-1/-1->7->6 [1] -1/-1/-1->7->6 [2] 3/-1/-1->7->-1 [3] 3/-1/-1->7->-1
+ip-26-0-167-9:1039540:1040064 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-36:104435:104947 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-162-46:702883:703431 [3] NCCL INFO Channel 00/04 :    0   1   2   3   4   5   6   7
+ip-26-0-162-46:702883:703431 [3] NCCL INFO Channel 01/04 :    0   1   2   3   4   5   6   7
+ip-26-0-162-46:702883:703431 [3] NCCL INFO Channel 02/04 :    0   1   2   3   4   5   6   7
+ip-26-0-162-46:702883:703431 [3] NCCL INFO Channel 03/04 :    0   1   2   3   4   5   6   7
+ip-26-0-162-46:702883:703431 [3] NCCL INFO Trees [0] 4/-1/-1->0->-1 [1] 4/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-162-46:702883:703431 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-213:9628:10141 [3] NCCL INFO Trees [0] -1/-1/-1->1->2 [1] -1/-1/-1->1->2 [2] 2/0/-1->1->3 [3] 2/0/-1->1->3
+ip-26-0-165-213:9628:10141 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-125:762818:763344 [3] NCCL INFO Trees [0] 1/3/-1->2->4 [1] 1/3/-1->2->4 [2] -1/-1/-1->2->1 [3] -1/-1/-1->2->1
+ip-26-0-166-125:762818:763344 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-15:127293:127806 [3] NCCL INFO Trees [0] -1/-1/-1->3->2 [1] -1/-1/-1->3->2 [2] 5/1/-1->3->7 [3] 5/1/-1->3->7
+ip-26-0-166-15:127293:127806 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-214:624946:625462 [3] NCCL INFO Trees [0] 2/6/-1->4->0 [1] 2/6/-1->4->0 [2] -1/-1/-1->4->5 [3] -1/-1/-1->4->5
+ip-26-0-166-214:624946:625462 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-244:897243:897754 [3] NCCL INFO Trees [0] -1/-1/-1->5->6 [1] -1/-1/-1->5->6 [2] 6/4/-1->5->3 [3] 6/4/-1->5->3
+ip-26-0-166-244:897243:897754 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-36:104432:104944 [3] NCCL INFO Trees [0] 5/7/-1->6->4 [1] 5/7/-1->6->4 [2] -1/-1/-1->6->5 [3] -1/-1/-1->6->5
+ip-26-0-166-36:104432:104944 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-15:127291:127811 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-166-244:897240:897761 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-165-213:9625:10145 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-166-214:624949:625465 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-166-214:624950:625464 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-166-36:104429:104948 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-166-244:897247:897759 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-166-15:127296:127809 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-166-244:897246:897760 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-166-15:127297:127808 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-166-125:762815:763348 [0] NCCL INFO Setting affinity for GPU 0 to 0fff,ffffffff
+ip-26-0-166-125:762816:763349 [1] NCCL INFO Setting affinity for GPU 1 to 0fff,ffffffff
+ip-26-0-166-15:127290:127810 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-165-213:9629:10147 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-166-214:624947:625469 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-167-9:1039544:1040068 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-165-213:9630:10148 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-166-36:104434:104951 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-166-214:624948:625470 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-167-9:1039543:1040069 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-166-15:127292:127807 [2] NCCL INFO Channel 00/0 : 2[2] -> 3[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127292:127807 [2] NCCL INFO Channel 01/0 : 2[2] -> 3[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762821:763347 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,fff00000,00000000
+ip-26-0-166-15:127292:127807 [2] NCCL INFO Channel 02/0 : 2[2] -> 3[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127292:127807 [2] NCCL INFO Channel 03/0 : 2[2] -> 3[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897242:897755 [2] NCCL INFO Channel 00/0 : 4[2] -> 5[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127292:127807 [2] NCCL INFO Channel 00/0 : 3[2] -> 4[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897242:897755 [2] NCCL INFO Channel 01/0 : 4[2] -> 5[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127292:127807 [2] NCCL INFO Channel 01/0 : 3[2] -> 4[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897242:897755 [2] NCCL INFO Channel 02/0 : 4[2] -> 5[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127292:127807 [2] NCCL INFO Channel 02/0 : 3[2] -> 4[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897242:897755 [2] NCCL INFO Channel 03/0 : 4[2] -> 5[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127292:127807 [2] NCCL INFO Channel 03/0 : 3[2] -> 4[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897242:897755 [2] NCCL INFO Channel 00/0 : 5[2] -> 6[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624945:625463 [2] NCCL INFO Channel 00/0 : 3[2] -> 4[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897242:897755 [2] NCCL INFO Channel 01/0 : 5[2] -> 6[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624945:625463 [2] NCCL INFO Channel 01/0 : 3[2] -> 4[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624945:625463 [2] NCCL INFO Channel 02/0 : 3[2] -> 4[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897242:897755 [2] NCCL INFO Channel 02/0 : 5[2] -> 6[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624945:625463 [2] NCCL INFO Channel 03/0 : 3[2] -> 4[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897242:897755 [2] NCCL INFO Channel 03/0 : 5[2] -> 6[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624945:625463 [2] NCCL INFO Channel 00/0 : 4[2] -> 5[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624945:625463 [2] NCCL INFO Channel 01/0 : 4[2] -> 5[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624945:625463 [2] NCCL INFO Channel 02/0 : 4[2] -> 5[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624945:625463 [2] NCCL INFO Channel 03/0 : 4[2] -> 5[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-46:702882:703433 [2] NCCL INFO Channel 00/0 : 7[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104431:104945 [2] NCCL INFO Channel 00/0 : 5[2] -> 6[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-46:702882:703433 [2] NCCL INFO Channel 01/0 : 7[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104431:104945 [2] NCCL INFO Channel 01/0 : 5[2] -> 6[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-46:702882:703433 [2] NCCL INFO Channel 02/0 : 7[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104431:104945 [2] NCCL INFO Channel 02/0 : 5[2] -> 6[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-46:702882:703433 [2] NCCL INFO Channel 03/0 : 7[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104431:104945 [2] NCCL INFO Channel 03/0 : 5[2] -> 6[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127295:127813 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-162-46:702882:703433 [2] NCCL INFO Channel 00/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104431:104945 [2] NCCL INFO Channel 00/0 : 6[2] -> 7[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-46:702882:703433 [2] NCCL INFO Channel 01/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104431:104945 [2] NCCL INFO Channel 01/0 : 6[2] -> 7[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-46:702882:703433 [2] NCCL INFO Channel 02/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104433:104950 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-162-46:702882:703433 [2] NCCL INFO Channel 03/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104431:104945 [2] NCCL INFO Channel 02/0 : 6[2] -> 7[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104431:104945 [2] NCCL INFO Channel 03/0 : 6[2] -> 7[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762817:763345 [2] NCCL INFO Channel 00/0 : 1[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762817:763345 [2] NCCL INFO Channel 01/0 : 1[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762817:763345 [2] NCCL INFO Channel 02/0 : 1[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762817:763345 [2] NCCL INFO Channel 03/0 : 1[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762817:763345 [2] NCCL INFO Channel 00/0 : 2[2] -> 3[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762817:763345 [2] NCCL INFO Channel 01/0 : 2[2] -> 3[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762817:763345 [2] NCCL INFO Channel 02/0 : 2[2] -> 3[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762817:763345 [2] NCCL INFO Channel 03/0 : 2[2] -> 3[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9627:10142 [2] NCCL INFO Channel 00/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9627:10142 [2] NCCL INFO Channel 01/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9627:10142 [2] NCCL INFO Channel 02/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9627:10142 [2] NCCL INFO Channel 03/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9627:10142 [2] NCCL INFO Channel 00/0 : 1[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9627:10142 [2] NCCL INFO Channel 01/0 : 1[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9627:10142 [2] NCCL INFO Channel 02/0 : 1[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-167-9:1039539:1040065 [2] NCCL INFO Channel 00/0 : 6[2] -> 7[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9627:10142 [2] NCCL INFO Channel 03/0 : 1[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-167-9:1039537:1040070 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-167-9:1039539:1040065 [2] NCCL INFO Channel 01/0 : 6[2] -> 7[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-167-9:1039539:1040065 [2] NCCL INFO Channel 02/0 : 6[2] -> 7[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-167-9:1039539:1040065 [2] NCCL INFO Channel 03/0 : 6[2] -> 7[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-167-9:1039539:1040065 [2] NCCL INFO Channel 00/0 : 7[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-167-9:1039539:1040065 [2] NCCL INFO Channel 01/0 : 7[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-167-9:1039539:1040065 [2] NCCL INFO Channel 02/0 : 7[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-167-9:1039539:1040065 [2] NCCL INFO Channel 03/0 : 7[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897243:897754 [3] NCCL INFO Channel 00/0 : 4[3] -> 5[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897243:897754 [3] NCCL INFO Channel 01/0 : 4[3] -> 5[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897243:897754 [3] NCCL INFO Channel 02/0 : 4[3] -> 5[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897243:897754 [3] NCCL INFO Channel 03/0 : 4[3] -> 5[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897243:897754 [3] NCCL INFO Channel 00/0 : 5[3] -> 6[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897243:897754 [3] NCCL INFO Channel 01/0 : 5[3] -> 6[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624946:625462 [3] NCCL INFO Channel 00/0 : 3[3] -> 4[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897243:897754 [3] NCCL INFO Channel 02/0 : 5[3] -> 6[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624946:625462 [3] NCCL INFO Channel 01/0 : 3[3] -> 4[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897243:897754 [3] NCCL INFO Channel 03/0 : 5[3] -> 6[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624946:625462 [3] NCCL INFO Channel 02/0 : 3[3] -> 4[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624946:625462 [3] NCCL INFO Channel 03/0 : 3[3] -> 4[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624946:625462 [3] NCCL INFO Channel 00/0 : 4[3] -> 5[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127293:127806 [3] NCCL INFO Channel 00/0 : 2[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624946:625462 [3] NCCL INFO Channel 01/0 : 4[3] -> 5[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624946:625462 [3] NCCL INFO Channel 02/0 : 4[3] -> 5[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127293:127806 [3] NCCL INFO Channel 01/0 : 2[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624946:625462 [3] NCCL INFO Channel 03/0 : 4[3] -> 5[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127293:127806 [3] NCCL INFO Channel 02/0 : 2[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127293:127806 [3] NCCL INFO Channel 03/0 : 2[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127293:127806 [3] NCCL INFO Channel 00/0 : 3[3] -> 4[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-167-9:1039540:1040064 [3] NCCL INFO Channel 00/0 : 6[3] -> 7[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127293:127806 [3] NCCL INFO Channel 01/0 : 3[3] -> 4[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-167-9:1039540:1040064 [3] NCCL INFO Channel 01/0 : 6[3] -> 7[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127293:127806 [3] NCCL INFO Channel 02/0 : 3[3] -> 4[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127293:127806 [3] NCCL INFO Channel 03/0 : 3[3] -> 4[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-167-9:1039540:1040064 [3] NCCL INFO Channel 02/0 : 6[3] -> 7[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-167-9:1039540:1040064 [3] NCCL INFO Channel 03/0 : 6[3] -> 7[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-167-9:1039540:1040064 [3] NCCL INFO Channel 00/0 : 7[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-167-9:1039540:1040064 [3] NCCL INFO Channel 01/0 : 7[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-167-9:1039540:1040064 [3] NCCL INFO Channel 02/0 : 7[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-167-9:1039540:1040064 [3] NCCL INFO Channel 03/0 : 7[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762822:763346 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,fff00000,00000000
+ip-26-0-166-125:762818:763344 [3] NCCL INFO Channel 00/0 : 1[3] -> 2[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762818:763344 [3] NCCL INFO Channel 01/0 : 1[3] -> 2[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762818:763344 [3] NCCL INFO Channel 02/0 : 1[3] -> 2[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762818:763344 [3] NCCL INFO Channel 03/0 : 1[3] -> 2[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762818:763344 [3] NCCL INFO Channel 00/0 : 2[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762818:763344 [3] NCCL INFO Channel 01/0 : 2[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762818:763344 [3] NCCL INFO Channel 02/0 : 2[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762818:763344 [3] NCCL INFO Channel 03/0 : 2[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104432:104944 [3] NCCL INFO Channel 00/0 : 5[3] -> 6[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104432:104944 [3] NCCL INFO Channel 01/0 : 5[3] -> 6[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9628:10141 [3] NCCL INFO Channel 00/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-46:702883:703431 [3] NCCL INFO Channel 00/0 : 7[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104432:104944 [3] NCCL INFO Channel 02/0 : 5[3] -> 6[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104432:104944 [3] NCCL INFO Channel 03/0 : 5[3] -> 6[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9628:10141 [3] NCCL INFO Channel 01/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-46:702883:703431 [3] NCCL INFO Channel 01/0 : 7[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104432:104944 [3] NCCL INFO Channel 00/0 : 6[3] -> 7[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9628:10141 [3] NCCL INFO Channel 02/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-46:702883:703431 [3] NCCL INFO Channel 02/0 : 7[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104432:104944 [3] NCCL INFO Channel 01/0 : 6[3] -> 7[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-46:702883:703431 [3] NCCL INFO Channel 03/0 : 7[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9628:10141 [3] NCCL INFO Channel 03/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104432:104944 [3] NCCL INFO Channel 02/0 : 6[3] -> 7[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-46:702883:703431 [3] NCCL INFO Channel 00/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9628:10141 [3] NCCL INFO Channel 00/0 : 1[3] -> 2[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104432:104944 [3] NCCL INFO Channel 03/0 : 6[3] -> 7[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9628:10141 [3] NCCL INFO Channel 01/0 : 1[3] -> 2[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-46:702883:703431 [3] NCCL INFO Channel 01/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9628:10141 [3] NCCL INFO Channel 02/0 : 1[3] -> 2[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-46:702883:703431 [3] NCCL INFO Channel 02/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9628:10141 [3] NCCL INFO Channel 03/0 : 1[3] -> 2[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-46:702883:703431 [3] NCCL INFO Channel 03/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127294:127812 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-166-244:897244:897763 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-166-244:897245:897764 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-166-125:762820:763352 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,fff00000,00000000
+ip-26-0-167-9:1039538:1040071 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-166-125:762819:763351 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,fff00000,00000000
+ip-26-0-167-9:1039542:1040075 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-167-9:1039541:1040074 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-166-214:624945:625463 [2] NCCL INFO Connected all rings
+ip-26-0-166-214:624945:625463 [2] NCCL INFO Channel 00/0 : 2[2] -> 4[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624945:625463 [2] NCCL INFO Channel 01/0 : 2[2] -> 4[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624945:625463 [2] NCCL INFO Channel 00/0 : 4[2] -> 6[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624945:625463 [2] NCCL INFO Channel 01/0 : 4[2] -> 6[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897242:897755 [2] NCCL INFO Connected all rings
+ip-26-0-166-244:897242:897755 [2] NCCL INFO Channel 02/0 : 3[2] -> 5[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897242:897755 [2] NCCL INFO Channel 03/0 : 3[2] -> 5[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127292:127807 [2] NCCL INFO Connected all rings
+ip-26-0-166-15:127292:127807 [2] NCCL INFO Channel 02/0 : 1[2] -> 3[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127292:127807 [2] NCCL INFO Channel 03/0 : 1[2] -> 3[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127292:127807 [2] NCCL INFO Channel 02/0 : 3[2] -> 5[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127292:127807 [2] NCCL INFO Channel 03/0 : 3[2] -> 5[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897242:897755 [2] NCCL INFO Channel 02/0 : 5[2] -> 3[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897242:897755 [2] NCCL INFO Channel 03/0 : 5[2] -> 3[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762817:763345 [2] NCCL INFO Connected all rings
+ip-26-0-166-125:762817:763345 [2] NCCL INFO Channel 00/0 : 2[2] -> 4[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762817:763345 [2] NCCL INFO Channel 01/0 : 2[2] -> 4[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762817:763345 [2] NCCL INFO Channel 00/0 : 4[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762817:763345 [2] NCCL INFO Channel 01/0 : 4[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104432:104944 [3] NCCL INFO Connected all rings
+ip-26-0-166-36:104432:104944 [3] NCCL INFO Channel 00/0 : 4[3] -> 6[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104432:104944 [3] NCCL INFO Channel 01/0 : 4[3] -> 6[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104431:104945 [2] NCCL INFO Connected all rings
+ip-26-0-166-36:104431:104945 [2] NCCL INFO Channel 00/0 : 4[2] -> 6[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104431:104945 [2] NCCL INFO Channel 01/0 : 4[2] -> 6[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104431:104945 [2] NCCL INFO Channel 00/0 : 6[2] -> 4[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104431:104945 [2] NCCL INFO Channel 01/0 : 6[2] -> 4[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624945:625463 [2] NCCL INFO Channel 00/0 : 0[2] -> 4[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624945:625463 [2] NCCL INFO Channel 01/0 : 0[2] -> 4[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624945:625463 [2] NCCL INFO Channel 00/0 : 4[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624945:625463 [2] NCCL INFO Channel 01/0 : 4[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9628:10141 [3] NCCL INFO Connected all rings
+ip-26-0-165-213:9628:10141 [3] NCCL INFO Channel 02/0 : 1[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9628:10141 [3] NCCL INFO Channel 03/0 : 1[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762818:763344 [3] NCCL INFO Connected all rings
+ip-26-0-166-125:762818:763344 [3] NCCL INFO Channel 00/0 : 2[3] -> 4[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762818:763344 [3] NCCL INFO Channel 01/0 : 2[3] -> 4[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127293:127806 [3] NCCL INFO Connected all rings
+ip-26-0-166-15:127293:127806 [3] NCCL INFO Channel 02/0 : 1[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127293:127806 [3] NCCL INFO Channel 03/0 : 1[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127293:127806 [3] NCCL INFO Channel 02/0 : 3[3] -> 5[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127293:127806 [3] NCCL INFO Channel 03/0 : 3[3] -> 5[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9628:10141 [3] NCCL INFO Channel 02/0 : 3[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9628:10141 [3] NCCL INFO Channel 03/0 : 3[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624946:625462 [3] NCCL INFO Connected all rings
+ip-26-0-166-214:624946:625462 [3] NCCL INFO Channel 00/0 : 2[3] -> 4[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624946:625462 [3] NCCL INFO Channel 01/0 : 2[3] -> 4[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624946:625462 [3] NCCL INFO Channel 00/0 : 4[3] -> 6[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624946:625462 [3] NCCL INFO Channel 01/0 : 4[3] -> 6[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762818:763344 [3] NCCL INFO Channel 00/0 : 4[3] -> 2[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762818:763344 [3] NCCL INFO Channel 01/0 : 4[3] -> 2[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624946:625462 [3] NCCL INFO Channel 00/0 : 0[3] -> 4[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104432:104944 [3] NCCL INFO Channel 00/0 : 6[3] -> 4[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624946:625462 [3] NCCL INFO Channel 01/0 : 0[3] -> 4[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104432:104944 [3] NCCL INFO Channel 01/0 : 6[3] -> 4[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624946:625462 [3] NCCL INFO Channel 00/0 : 4[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897243:897754 [3] NCCL INFO Connected all rings
+ip-26-0-166-214:624946:625462 [3] NCCL INFO Channel 01/0 : 4[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897243:897754 [3] NCCL INFO Channel 02/0 : 3[3] -> 5[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897243:897754 [3] NCCL INFO Channel 03/0 : 3[3] -> 5[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897243:897754 [3] NCCL INFO Channel 02/0 : 5[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127293:127806 [3] NCCL INFO Channel 02/0 : 7[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897243:897754 [3] NCCL INFO Channel 03/0 : 5[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127293:127806 [3] NCCL INFO Channel 03/0 : 7[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127293:127806 [3] NCCL INFO Channel 02/0 : 3[3] -> 7[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127293:127806 [3] NCCL INFO Channel 03/0 : 3[3] -> 7[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-46:702880:703439 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-162-46:702880:703439 [0] NCCL INFO Channel 00/04 :    0   1   2   3   4   5   6   7
+ip-26-0-162-46:702880:703439 [0] NCCL INFO Channel 01/04 :    0   1   2   3   4   5   6   7
+ip-26-0-162-46:702880:703439 [0] NCCL INFO Channel 02/04 :    0   1   2   3   4   5   6   7
+ip-26-0-162-46:702880:703439 [0] NCCL INFO Channel 03/04 :    0   1   2   3   4   5   6   7
+ip-26-0-162-46:702880:703439 [0] NCCL INFO Trees [0] 4/-1/-1->0->-1 [1] 4/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-162-46:702880:703439 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-213:9625:10145 [0] NCCL INFO Trees [0] -1/-1/-1->1->2 [1] -1/-1/-1->1->2 [2] 2/0/-1->1->3 [3] 2/0/-1->1->3
+ip-26-0-165-213:9625:10145 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-125:762815:763348 [0] NCCL INFO Trees [0] 1/3/-1->2->4 [1] 1/3/-1->2->4 [2] -1/-1/-1->2->1 [3] -1/-1/-1->2->1
+ip-26-0-166-125:762815:763348 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-15:127290:127810 [0] NCCL INFO Trees [0] -1/-1/-1->3->2 [1] -1/-1/-1->3->2 [2] 5/1/-1->3->7 [3] 5/1/-1->3->7
+ip-26-0-166-15:127290:127810 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-214:624943:625466 [0] NCCL INFO Trees [0] 2/6/-1->4->0 [1] 2/6/-1->4->0 [2] -1/-1/-1->4->5 [3] -1/-1/-1->4->5
+ip-26-0-166-214:624943:625466 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-244:897240:897761 [0] NCCL INFO Trees [0] -1/-1/-1->5->6 [1] -1/-1/-1->5->6 [2] 6/4/-1->5->3 [3] 6/4/-1->5->3
+ip-26-0-166-244:897240:897761 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-36:104429:104948 [0] NCCL INFO Trees [0] 5/7/-1->6->4 [1] 5/7/-1->6->4 [2] -1/-1/-1->6->5 [3] -1/-1/-1->6->5
+ip-26-0-166-36:104429:104948 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-167-9:1039537:1040070 [0] NCCL INFO Trees [0] -1/-1/-1->7->6 [1] -1/-1/-1->7->6 [2] 3/-1/-1->7->-1 [3] 3/-1/-1->7->-1
+ip-26-0-167-9:1039537:1040070 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-162-46:702881:703441 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-162-46:702881:703441 [1] NCCL INFO Channel 00/04 :    0   1   2   3   4   5   6   7
+ip-26-0-162-46:702881:703441 [1] NCCL INFO Channel 01/04 :    0   1   2   3   4   5   6   7
+ip-26-0-162-46:702881:703441 [1] NCCL INFO Channel 02/04 :    0   1   2   3   4   5   6   7
+ip-26-0-162-46:702881:703441 [1] NCCL INFO Channel 03/04 :    0   1   2   3   4   5   6   7
+ip-26-0-162-46:702881:703441 [1] NCCL INFO Trees [0] 4/-1/-1->0->-1 [1] 4/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-162-46:702881:703441 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-213:9626:10146 [1] NCCL INFO Trees [0] -1/-1/-1->1->2 [1] -1/-1/-1->1->2 [2] 2/0/-1->1->3 [3] 2/0/-1->1->3
+ip-26-0-165-213:9626:10146 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-125:762816:763349 [1] NCCL INFO Trees [0] 1/3/-1->2->4 [1] 1/3/-1->2->4 [2] -1/-1/-1->2->1 [3] -1/-1/-1->2->1
+ip-26-0-166-125:762816:763349 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-15:127291:127811 [1] NCCL INFO Trees [0] -1/-1/-1->3->2 [1] -1/-1/-1->3->2 [2] 5/1/-1->3->7 [3] 5/1/-1->3->7
+ip-26-0-166-15:127291:127811 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-214:624944:625467 [1] NCCL INFO Trees [0] 2/6/-1->4->0 [1] 2/6/-1->4->0 [2] -1/-1/-1->4->5 [3] -1/-1/-1->4->5
+ip-26-0-166-214:624944:625467 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-244:897241:897762 [1] NCCL INFO Trees [0] -1/-1/-1->5->6 [1] -1/-1/-1->5->6 [2] 6/4/-1->5->3 [3] 6/4/-1->5->3
+ip-26-0-166-244:897241:897762 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-167-9:1039538:1040071 [1] NCCL INFO Trees [0] -1/-1/-1->7->6 [1] -1/-1/-1->7->6 [2] 3/-1/-1->7->-1 [3] 3/-1/-1->7->-1
+ip-26-0-167-9:1039538:1040071 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-36:104430:104949 [1] NCCL INFO Trees [0] 5/7/-1->6->4 [1] 5/7/-1->6->4 [2] -1/-1/-1->6->5 [3] -1/-1/-1->6->5
+ip-26-0-166-36:104430:104949 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-162-46:702887:703436 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-162-46:702887:703436 [7] NCCL INFO Channel 00/04 :    0   1   2   3   4   5   6   7
+ip-26-0-162-46:702887:703436 [7] NCCL INFO Channel 01/04 :    0   1   2   3   4   5   6   7
+ip-26-0-162-46:702887:703436 [7] NCCL INFO Channel 02/04 :    0   1   2   3   4   5   6   7
+ip-26-0-162-46:702887:703436 [7] NCCL INFO Channel 03/04 :    0   1   2   3   4   5   6   7
+ip-26-0-162-46:702887:703436 [7] NCCL INFO Trees [0] 4/-1/-1->0->-1 [1] 4/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-162-46:702887:703436 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-213:9632:10143 [7] NCCL INFO Trees [0] -1/-1/-1->1->2 [1] -1/-1/-1->1->2 [2] 2/0/-1->1->3 [3] 2/0/-1->1->3
+ip-26-0-165-213:9632:10143 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-125:762822:763346 [7] NCCL INFO Trees [0] 1/3/-1->2->4 [1] 1/3/-1->2->4 [2] -1/-1/-1->2->1 [3] -1/-1/-1->2->1
+ip-26-0-166-125:762822:763346 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-15:127297:127808 [7] NCCL INFO Trees [0] -1/-1/-1->3->2 [1] -1/-1/-1->3->2 [2] 5/1/-1->3->7 [3] 5/1/-1->3->7
+ip-26-0-166-15:127297:127808 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-214:624950:625464 [7] NCCL INFO Trees [0] 2/6/-1->4->0 [1] 2/6/-1->4->0 [2] -1/-1/-1->4->5 [3] -1/-1/-1->4->5
+ip-26-0-166-214:624950:625464 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-244:897247:897759 [7] NCCL INFO Trees [0] -1/-1/-1->5->6 [1] -1/-1/-1->5->6 [2] 6/4/-1->5->3 [3] 6/4/-1->5->3
+ip-26-0-166-244:897247:897759 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-167-9:1039544:1040068 [7] NCCL INFO Trees [0] -1/-1/-1->7->6 [1] -1/-1/-1->7->6 [2] 3/-1/-1->7->-1 [3] 3/-1/-1->7->-1
+ip-26-0-167-9:1039544:1040068 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-36:104436:104946 [7] NCCL INFO Trees [0] 5/7/-1->6->4 [1] 5/7/-1->6->4 [2] -1/-1/-1->6->5 [3] -1/-1/-1->6->5
+ip-26-0-166-36:104436:104946 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-214:624943:625466 [0] NCCL INFO Channel 00/0 : 3[0] -> 4[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624943:625466 [0] NCCL INFO Channel 01/0 : 3[0] -> 4[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104429:104948 [0] NCCL INFO Channel 00/0 : 5[0] -> 6[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624943:625466 [0] NCCL INFO Channel 02/0 : 3[0] -> 4[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104429:104948 [0] NCCL INFO Channel 01/0 : 5[0] -> 6[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624943:625466 [0] NCCL INFO Channel 03/0 : 3[0] -> 4[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104429:104948 [0] NCCL INFO Channel 02/0 : 5[0] -> 6[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624943:625466 [0] NCCL INFO Channel 00/0 : 4[0] -> 5[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104429:104948 [0] NCCL INFO Channel 03/0 : 5[0] -> 6[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624943:625466 [0] NCCL INFO Channel 01/0 : 4[0] -> 5[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104429:104948 [0] NCCL INFO Channel 00/0 : 6[0] -> 7[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624943:625466 [0] NCCL INFO Channel 02/0 : 4[0] -> 5[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624943:625466 [0] NCCL INFO Channel 03/0 : 4[0] -> 5[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104429:104948 [0] NCCL INFO Channel 01/0 : 6[0] -> 7[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104429:104948 [0] NCCL INFO Channel 02/0 : 6[0] -> 7[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104429:104948 [0] NCCL INFO Channel 03/0 : 6[0] -> 7[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9625:10145 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9625:10145 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9625:10145 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9625:10145 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9625:10145 [0] NCCL INFO Channel 00/0 : 1[0] -> 2[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9625:10145 [0] NCCL INFO Channel 01/0 : 1[0] -> 2[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9625:10145 [0] NCCL INFO Channel 02/0 : 1[0] -> 2[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-167-9:1039537:1040070 [0] NCCL INFO Channel 00/0 : 6[0] -> 7[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9625:10145 [0] NCCL INFO Channel 03/0 : 1[0] -> 2[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-167-9:1039537:1040070 [0] NCCL INFO Channel 01/0 : 6[0] -> 7[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-167-9:1039537:1040070 [0] NCCL INFO Channel 02/0 : 6[0] -> 7[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897240:897761 [0] NCCL INFO Channel 00/0 : 4[0] -> 5[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-167-9:1039537:1040070 [0] NCCL INFO Channel 03/0 : 6[0] -> 7[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897240:897761 [0] NCCL INFO Channel 01/0 : 4[0] -> 5[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-167-9:1039537:1040070 [0] NCCL INFO Channel 00/0 : 7[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897240:897761 [0] NCCL INFO Channel 02/0 : 4[0] -> 5[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-167-9:1039537:1040070 [0] NCCL INFO Channel 01/0 : 7[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897240:897761 [0] NCCL INFO Channel 03/0 : 4[0] -> 5[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-167-9:1039537:1040070 [0] NCCL INFO Channel 02/0 : 7[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897240:897761 [0] NCCL INFO Channel 00/0 : 5[0] -> 6[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-167-9:1039537:1040070 [0] NCCL INFO Channel 03/0 : 7[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897240:897761 [0] NCCL INFO Channel 01/0 : 5[0] -> 6[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897240:897761 [0] NCCL INFO Channel 02/0 : 5[0] -> 6[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762815:763348 [0] NCCL INFO Channel 00/0 : 1[0] -> 2[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897240:897761 [0] NCCL INFO Channel 03/0 : 5[0] -> 6[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762815:763348 [0] NCCL INFO Channel 01/0 : 1[0] -> 2[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-46:702880:703439 [0] NCCL INFO Channel 00/0 : 7[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762815:763348 [0] NCCL INFO Channel 02/0 : 1[0] -> 2[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762815:763348 [0] NCCL INFO Channel 03/0 : 1[0] -> 2[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-46:702880:703439 [0] NCCL INFO Channel 01/0 : 7[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762815:763348 [0] NCCL INFO Channel 00/0 : 2[0] -> 3[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-46:702880:703439 [0] NCCL INFO Channel 02/0 : 7[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762815:763348 [0] NCCL INFO Channel 01/0 : 2[0] -> 3[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-46:702880:703439 [0] NCCL INFO Channel 03/0 : 7[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762815:763348 [0] NCCL INFO Channel 02/0 : 2[0] -> 3[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-46:702880:703439 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762815:763348 [0] NCCL INFO Channel 03/0 : 2[0] -> 3[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-46:702880:703439 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-46:702880:703439 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-46:702880:703439 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127290:127810 [0] NCCL INFO Channel 00/0 : 2[0] -> 3[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127290:127810 [0] NCCL INFO Channel 01/0 : 2[0] -> 3[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127290:127810 [0] NCCL INFO Channel 02/0 : 2[0] -> 3[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127290:127810 [0] NCCL INFO Channel 03/0 : 2[0] -> 3[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127290:127810 [0] NCCL INFO Channel 00/0 : 3[0] -> 4[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127290:127810 [0] NCCL INFO Channel 01/0 : 3[0] -> 4[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127290:127810 [0] NCCL INFO Channel 02/0 : 3[0] -> 4[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127290:127810 [0] NCCL INFO Channel 03/0 : 3[0] -> 4[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897241:897762 [1] NCCL INFO Channel 00/0 : 4[1] -> 5[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897241:897762 [1] NCCL INFO Channel 01/0 : 4[1] -> 5[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897241:897762 [1] NCCL INFO Channel 02/0 : 4[1] -> 5[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897241:897762 [1] NCCL INFO Channel 03/0 : 4[1] -> 5[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624944:625467 [1] NCCL INFO Channel 00/0 : 3[1] -> 4[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127291:127811 [1] NCCL INFO Channel 00/0 : 2[1] -> 3[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897241:897762 [1] NCCL INFO Channel 00/0 : 5[1] -> 6[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624944:625467 [1] NCCL INFO Channel 01/0 : 3[1] -> 4[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127291:127811 [1] NCCL INFO Channel 01/0 : 2[1] -> 3[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897241:897762 [1] NCCL INFO Channel 01/0 : 5[1] -> 6[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624944:625467 [1] NCCL INFO Channel 02/0 : 3[1] -> 4[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127291:127811 [1] NCCL INFO Channel 02/0 : 2[1] -> 3[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897241:897762 [1] NCCL INFO Channel 02/0 : 5[1] -> 6[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-46:702881:703441 [1] NCCL INFO Channel 00/0 : 7[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624944:625467 [1] NCCL INFO Channel 03/0 : 3[1] -> 4[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897241:897762 [1] NCCL INFO Channel 03/0 : 5[1] -> 6[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127291:127811 [1] NCCL INFO Channel 03/0 : 2[1] -> 3[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624944:625467 [1] NCCL INFO Channel 00/0 : 4[1] -> 5[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127291:127811 [1] NCCL INFO Channel 00/0 : 3[1] -> 4[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-46:702881:703441 [1] NCCL INFO Channel 01/0 : 7[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127291:127811 [1] NCCL INFO Channel 01/0 : 3[1] -> 4[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624944:625467 [1] NCCL INFO Channel 01/0 : 4[1] -> 5[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-46:702881:703441 [1] NCCL INFO Channel 02/0 : 7[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-167-9:1039538:1040071 [1] NCCL INFO Channel 00/0 : 6[1] -> 7[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127291:127811 [1] NCCL INFO Channel 02/0 : 3[1] -> 4[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624944:625467 [1] NCCL INFO Channel 02/0 : 4[1] -> 5[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-46:702881:703441 [1] NCCL INFO Channel 03/0 : 7[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127291:127811 [1] NCCL INFO Channel 03/0 : 3[1] -> 4[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624944:625467 [1] NCCL INFO Channel 03/0 : 4[1] -> 5[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-167-9:1039538:1040071 [1] NCCL INFO Channel 01/0 : 6[1] -> 7[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-46:702881:703441 [1] NCCL INFO Channel 00/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-167-9:1039538:1040071 [1] NCCL INFO Channel 02/0 : 6[1] -> 7[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-46:702881:703441 [1] NCCL INFO Channel 01/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-167-9:1039538:1040071 [1] NCCL INFO Channel 03/0 : 6[1] -> 7[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-46:702881:703441 [1] NCCL INFO Channel 02/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-167-9:1039538:1040071 [1] NCCL INFO Channel 00/0 : 7[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-46:702881:703441 [1] NCCL INFO Channel 03/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-167-9:1039538:1040071 [1] NCCL INFO Channel 01/0 : 7[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-167-9:1039538:1040071 [1] NCCL INFO Channel 02/0 : 7[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-167-9:1039538:1040071 [1] NCCL INFO Channel 03/0 : 7[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9626:10146 [1] NCCL INFO Channel 00/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762816:763349 [1] NCCL INFO Channel 00/0 : 1[1] -> 2[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9626:10146 [1] NCCL INFO Channel 01/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9626:10146 [1] NCCL INFO Channel 02/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762816:763349 [1] NCCL INFO Channel 01/0 : 1[1] -> 2[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9626:10146 [1] NCCL INFO Channel 03/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762816:763349 [1] NCCL INFO Channel 02/0 : 1[1] -> 2[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9626:10146 [1] NCCL INFO Channel 00/0 : 1[1] -> 2[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762816:763349 [1] NCCL INFO Channel 03/0 : 1[1] -> 2[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9626:10146 [1] NCCL INFO Channel 01/0 : 1[1] -> 2[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762816:763349 [1] NCCL INFO Channel 00/0 : 2[1] -> 3[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9626:10146 [1] NCCL INFO Channel 02/0 : 1[1] -> 2[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762816:763349 [1] NCCL INFO Channel 01/0 : 2[1] -> 3[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9626:10146 [1] NCCL INFO Channel 03/0 : 1[1] -> 2[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762816:763349 [1] NCCL INFO Channel 02/0 : 2[1] -> 3[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762816:763349 [1] NCCL INFO Channel 03/0 : 2[1] -> 3[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104430:104949 [1] NCCL INFO Channel 00/0 : 5[1] -> 6[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104430:104949 [1] NCCL INFO Channel 01/0 : 5[1] -> 6[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104430:104949 [1] NCCL INFO Channel 02/0 : 5[1] -> 6[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104430:104949 [1] NCCL INFO Channel 03/0 : 5[1] -> 6[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104430:104949 [1] NCCL INFO Channel 00/0 : 6[1] -> 7[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104430:104949 [1] NCCL INFO Channel 01/0 : 6[1] -> 7[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104430:104949 [1] NCCL INFO Channel 02/0 : 6[1] -> 7[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104430:104949 [1] NCCL INFO Channel 03/0 : 6[1] -> 7[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-46:702886:703437 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-162-46:702886:703437 [6] NCCL INFO Channel 00/04 :    0   1   2   3   4   5   6   7
+ip-26-0-162-46:702886:703437 [6] NCCL INFO Channel 01/04 :    0   1   2   3   4   5   6   7
+ip-26-0-162-46:702886:703437 [6] NCCL INFO Channel 02/04 :    0   1   2   3   4   5   6   7
+ip-26-0-162-46:702886:703437 [6] NCCL INFO Channel 03/04 :    0   1   2   3   4   5   6   7
+ip-26-0-162-46:702886:703437 [6] NCCL INFO Trees [0] 4/-1/-1->0->-1 [1] 4/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-162-46:702886:703437 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-213:9631:10144 [6] NCCL INFO Trees [0] -1/-1/-1->1->2 [1] -1/-1/-1->1->2 [2] 2/0/-1->1->3 [3] 2/0/-1->1->3
+ip-26-0-165-213:9631:10144 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-125:762821:763347 [6] NCCL INFO Trees [0] 1/3/-1->2->4 [1] 1/3/-1->2->4 [2] -1/-1/-1->2->1 [3] -1/-1/-1->2->1
+ip-26-0-166-125:762821:763347 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-15:127296:127809 [6] NCCL INFO Trees [0] -1/-1/-1->3->2 [1] -1/-1/-1->3->2 [2] 5/1/-1->3->7 [3] 5/1/-1->3->7
+ip-26-0-166-15:127296:127809 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-214:624949:625465 [6] NCCL INFO Trees [0] 2/6/-1->4->0 [1] 2/6/-1->4->0 [2] -1/-1/-1->4->5 [3] -1/-1/-1->4->5
+ip-26-0-166-214:624949:625465 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-244:897246:897760 [6] NCCL INFO Trees [0] -1/-1/-1->5->6 [1] -1/-1/-1->5->6 [2] 6/4/-1->5->3 [3] 6/4/-1->5->3
+ip-26-0-166-244:897246:897760 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-36:104435:104947 [6] NCCL INFO Trees [0] 5/7/-1->6->4 [1] 5/7/-1->6->4 [2] -1/-1/-1->6->5 [3] -1/-1/-1->6->5
+ip-26-0-166-36:104435:104947 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-167-9:1039543:1040069 [6] NCCL INFO Trees [0] -1/-1/-1->7->6 [1] -1/-1/-1->7->6 [2] 3/-1/-1->7->-1 [3] 3/-1/-1->7->-1
+ip-26-0-167-9:1039543:1040069 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-125:762822:763346 [7] NCCL INFO Channel 00/0 : 1[7] -> 2[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762822:763346 [7] NCCL INFO Channel 01/0 : 1[7] -> 2[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762822:763346 [7] NCCL INFO Channel 02/0 : 1[7] -> 2[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-46:702887:703436 [7] NCCL INFO Channel 00/0 : 7[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762822:763346 [7] NCCL INFO Channel 03/0 : 1[7] -> 2[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-46:702887:703436 [7] NCCL INFO Channel 01/0 : 7[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762822:763346 [7] NCCL INFO Channel 00/0 : 2[7] -> 3[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-46:702887:703436 [7] NCCL INFO Channel 02/0 : 7[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762822:763346 [7] NCCL INFO Channel 01/0 : 2[7] -> 3[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-46:702887:703436 [7] NCCL INFO Channel 03/0 : 7[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762822:763346 [7] NCCL INFO Channel 02/0 : 2[7] -> 3[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-46:702887:703436 [7] NCCL INFO Channel 00/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762822:763346 [7] NCCL INFO Channel 03/0 : 2[7] -> 3[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-46:702887:703436 [7] NCCL INFO Channel 01/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-46:702887:703436 [7] NCCL INFO Channel 02/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-46:702887:703436 [7] NCCL INFO Channel 03/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897247:897759 [7] NCCL INFO Channel 00/0 : 4[7] -> 5[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897247:897759 [7] NCCL INFO Channel 01/0 : 4[7] -> 5[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897247:897759 [7] NCCL INFO Channel 02/0 : 4[7] -> 5[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624950:625464 [7] NCCL INFO Channel 00/0 : 3[7] -> 4[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-167-9:1039544:1040068 [7] NCCL INFO Channel 00/0 : 6[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897247:897759 [7] NCCL INFO Channel 03/0 : 4[7] -> 5[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624950:625464 [7] NCCL INFO Channel 01/0 : 3[7] -> 4[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897247:897759 [7] NCCL INFO Channel 00/0 : 5[7] -> 6[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-167-9:1039544:1040068 [7] NCCL INFO Channel 01/0 : 6[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624950:625464 [7] NCCL INFO Channel 02/0 : 3[7] -> 4[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897247:897759 [7] NCCL INFO Channel 01/0 : 5[7] -> 6[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-167-9:1039544:1040068 [7] NCCL INFO Channel 02/0 : 6[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624950:625464 [7] NCCL INFO Channel 03/0 : 3[7] -> 4[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897247:897759 [7] NCCL INFO Channel 02/0 : 5[7] -> 6[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624950:625464 [7] NCCL INFO Channel 00/0 : 4[7] -> 5[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-167-9:1039544:1040068 [7] NCCL INFO Channel 03/0 : 6[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897247:897759 [7] NCCL INFO Channel 03/0 : 5[7] -> 6[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624950:625464 [7] NCCL INFO Channel 01/0 : 4[7] -> 5[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-167-9:1039544:1040068 [7] NCCL INFO Channel 00/0 : 7[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624950:625464 [7] NCCL INFO Channel 02/0 : 4[7] -> 5[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-167-9:1039544:1040068 [7] NCCL INFO Channel 01/0 : 7[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624950:625464 [7] NCCL INFO Channel 03/0 : 4[7] -> 5[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-167-9:1039544:1040068 [7] NCCL INFO Channel 02/0 : 7[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-167-9:1039544:1040068 [7] NCCL INFO Channel 03/0 : 7[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127297:127808 [7] NCCL INFO Channel 00/0 : 2[7] -> 3[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104436:104946 [7] NCCL INFO Channel 00/0 : 5[7] -> 6[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127297:127808 [7] NCCL INFO Channel 01/0 : 2[7] -> 3[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104436:104946 [7] NCCL INFO Channel 01/0 : 5[7] -> 6[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127297:127808 [7] NCCL INFO Channel 02/0 : 2[7] -> 3[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104436:104946 [7] NCCL INFO Channel 02/0 : 5[7] -> 6[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127297:127808 [7] NCCL INFO Channel 03/0 : 2[7] -> 3[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127297:127808 [7] NCCL INFO Channel 00/0 : 3[7] -> 4[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104436:104946 [7] NCCL INFO Channel 03/0 : 5[7] -> 6[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127297:127808 [7] NCCL INFO Channel 01/0 : 3[7] -> 4[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104436:104946 [7] NCCL INFO Channel 00/0 : 6[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127297:127808 [7] NCCL INFO Channel 02/0 : 3[7] -> 4[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104436:104946 [7] NCCL INFO Channel 01/0 : 6[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127297:127808 [7] NCCL INFO Channel 03/0 : 3[7] -> 4[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104436:104946 [7] NCCL INFO Channel 02/0 : 6[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104436:104946 [7] NCCL INFO Channel 03/0 : 6[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9632:10143 [7] NCCL INFO Channel 00/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9632:10143 [7] NCCL INFO Channel 01/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9632:10143 [7] NCCL INFO Channel 02/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9632:10143 [7] NCCL INFO Channel 03/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9632:10143 [7] NCCL INFO Channel 00/0 : 1[7] -> 2[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9632:10143 [7] NCCL INFO Channel 01/0 : 1[7] -> 2[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9632:10143 [7] NCCL INFO Channel 02/0 : 1[7] -> 2[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9632:10143 [7] NCCL INFO Channel 03/0 : 1[7] -> 2[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-46:702884:703443 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-162-46:702884:703443 [4] NCCL INFO Channel 00/04 :    0   1   2   3   4   5   6   7
+ip-26-0-162-46:702884:703443 [4] NCCL INFO Channel 01/04 :    0   1   2   3   4   5   6   7
+ip-26-0-162-46:702884:703443 [4] NCCL INFO Channel 02/04 :    0   1   2   3   4   5   6   7
+ip-26-0-162-46:702884:703443 [4] NCCL INFO Channel 03/04 :    0   1   2   3   4   5   6   7
+ip-26-0-162-46:702884:703443 [4] NCCL INFO Trees [0] 4/-1/-1->0->-1 [1] 4/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-162-46:702884:703443 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-213:9629:10147 [4] NCCL INFO Trees [0] -1/-1/-1->1->2 [1] -1/-1/-1->1->2 [2] 2/0/-1->1->3 [3] 2/0/-1->1->3
+ip-26-0-165-213:9629:10147 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-125:762819:763351 [4] NCCL INFO Trees [0] 1/3/-1->2->4 [1] 1/3/-1->2->4 [2] -1/-1/-1->2->1 [3] -1/-1/-1->2->1
+ip-26-0-166-125:762819:763351 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-15:127294:127812 [4] NCCL INFO Trees [0] -1/-1/-1->3->2 [1] -1/-1/-1->3->2 [2] 5/1/-1->3->7 [3] 5/1/-1->3->7
+ip-26-0-166-15:127294:127812 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-214:624947:625469 [4] NCCL INFO Trees [0] 2/6/-1->4->0 [1] 2/6/-1->4->0 [2] -1/-1/-1->4->5 [3] -1/-1/-1->4->5
+ip-26-0-166-214:624947:625469 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-244:897244:897763 [4] NCCL INFO Trees [0] -1/-1/-1->5->6 [1] -1/-1/-1->5->6 [2] 6/4/-1->5->3 [3] 6/4/-1->5->3
+ip-26-0-166-244:897244:897763 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-36:104433:104950 [4] NCCL INFO Trees [0] 5/7/-1->6->4 [1] 5/7/-1->6->4 [2] -1/-1/-1->6->5 [3] -1/-1/-1->6->5
+ip-26-0-166-36:104433:104950 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-167-9:1039541:1040074 [4] NCCL INFO Trees [0] -1/-1/-1->7->6 [1] -1/-1/-1->7->6 [2] 3/-1/-1->7->-1 [3] 3/-1/-1->7->-1
+ip-26-0-167-9:1039541:1040074 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-162-46:702885:703445 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-162-46:702885:703445 [5] NCCL INFO Channel 00/04 :    0   1   2   3   4   5   6   7
+ip-26-0-162-46:702885:703445 [5] NCCL INFO Channel 01/04 :    0   1   2   3   4   5   6   7
+ip-26-0-162-46:702885:703445 [5] NCCL INFO Channel 02/04 :    0   1   2   3   4   5   6   7
+ip-26-0-162-46:702885:703445 [5] NCCL INFO Channel 03/04 :    0   1   2   3   4   5   6   7
+ip-26-0-162-46:702885:703445 [5] NCCL INFO Trees [0] 4/-1/-1->0->-1 [1] 4/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-162-46:702885:703445 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-213:9630:10148 [5] NCCL INFO Trees [0] -1/-1/-1->1->2 [1] -1/-1/-1->1->2 [2] 2/0/-1->1->3 [3] 2/0/-1->1->3
+ip-26-0-165-213:9630:10148 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-125:762820:763352 [5] NCCL INFO Trees [0] 1/3/-1->2->4 [1] 1/3/-1->2->4 [2] -1/-1/-1->2->1 [3] -1/-1/-1->2->1
+ip-26-0-166-125:762820:763352 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-15:127295:127813 [5] NCCL INFO Trees [0] -1/-1/-1->3->2 [1] -1/-1/-1->3->2 [2] 5/1/-1->3->7 [3] 5/1/-1->3->7
+ip-26-0-166-15:127295:127813 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-214:624948:625470 [5] NCCL INFO Trees [0] 2/6/-1->4->0 [1] 2/6/-1->4->0 [2] -1/-1/-1->4->5 [3] -1/-1/-1->4->5
+ip-26-0-166-214:624948:625470 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-244:897245:897764 [5] NCCL INFO Trees [0] -1/-1/-1->5->6 [1] -1/-1/-1->5->6 [2] 6/4/-1->5->3 [3] 6/4/-1->5->3
+ip-26-0-166-244:897245:897764 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-36:104434:104951 [5] NCCL INFO Trees [0] 5/7/-1->6->4 [1] 5/7/-1->6->4 [2] -1/-1/-1->6->5 [3] -1/-1/-1->6->5
+ip-26-0-166-36:104434:104951 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-167-9:1039542:1040075 [5] NCCL INFO Trees [0] -1/-1/-1->7->6 [1] -1/-1/-1->7->6 [2] 3/-1/-1->7->-1 [3] 3/-1/-1->7->-1
+ip-26-0-167-9:1039542:1040075 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-15:127296:127809 [6] NCCL INFO Channel 00/0 : 2[6] -> 3[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127296:127809 [6] NCCL INFO Channel 01/0 : 2[6] -> 3[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127296:127809 [6] NCCL INFO Channel 02/0 : 2[6] -> 3[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624949:625465 [6] NCCL INFO Channel 00/0 : 3[6] -> 4[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127296:127809 [6] NCCL INFO Channel 03/0 : 2[6] -> 3[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624949:625465 [6] NCCL INFO Channel 01/0 : 3[6] -> 4[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127296:127809 [6] NCCL INFO Channel 00/0 : 3[6] -> 4[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624949:625465 [6] NCCL INFO Channel 02/0 : 3[6] -> 4[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127296:127809 [6] NCCL INFO Channel 01/0 : 3[6] -> 4[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624949:625465 [6] NCCL INFO Channel 03/0 : 3[6] -> 4[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127296:127809 [6] NCCL INFO Channel 02/0 : 3[6] -> 4[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762821:763347 [6] NCCL INFO Channel 00/0 : 1[6] -> 2[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624949:625465 [6] NCCL INFO Channel 00/0 : 4[6] -> 5[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-167-9:1039543:1040069 [6] NCCL INFO Channel 00/0 : 6[6] -> 7[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127296:127809 [6] NCCL INFO Channel 03/0 : 3[6] -> 4[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624949:625465 [6] NCCL INFO Channel 01/0 : 4[6] -> 5[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-46:702886:703437 [6] NCCL INFO Channel 00/0 : 7[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762821:763347 [6] NCCL INFO Channel 01/0 : 1[6] -> 2[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-167-9:1039543:1040069 [6] NCCL INFO Channel 01/0 : 6[6] -> 7[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624949:625465 [6] NCCL INFO Channel 02/0 : 4[6] -> 5[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-46:702886:703437 [6] NCCL INFO Channel 01/0 : 7[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762821:763347 [6] NCCL INFO Channel 02/0 : 1[6] -> 2[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624949:625465 [6] NCCL INFO Channel 03/0 : 4[6] -> 5[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-167-9:1039543:1040069 [6] NCCL INFO Channel 02/0 : 6[6] -> 7[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-46:702886:703437 [6] NCCL INFO Channel 02/0 : 7[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762821:763347 [6] NCCL INFO Channel 03/0 : 1[6] -> 2[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-167-9:1039543:1040069 [6] NCCL INFO Channel 03/0 : 6[6] -> 7[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-46:702886:703437 [6] NCCL INFO Channel 03/0 : 7[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-167-9:1039543:1040069 [6] NCCL INFO Channel 00/0 : 7[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762821:763347 [6] NCCL INFO Channel 00/0 : 2[6] -> 3[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-46:702886:703437 [6] NCCL INFO Channel 00/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-167-9:1039543:1040069 [6] NCCL INFO Channel 01/0 : 7[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104435:104947 [6] NCCL INFO Channel 00/0 : 5[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762821:763347 [6] NCCL INFO Channel 01/0 : 2[6] -> 3[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-46:702886:703437 [6] NCCL INFO Channel 01/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-167-9:1039543:1040069 [6] NCCL INFO Channel 02/0 : 7[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762821:763347 [6] NCCL INFO Channel 02/0 : 2[6] -> 3[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104435:104947 [6] NCCL INFO Channel 01/0 : 5[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9631:10144 [6] NCCL INFO Channel 00/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-46:702886:703437 [6] NCCL INFO Channel 02/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762821:763347 [6] NCCL INFO Channel 03/0 : 2[6] -> 3[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-167-9:1039543:1040069 [6] NCCL INFO Channel 03/0 : 7[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104435:104947 [6] NCCL INFO Channel 02/0 : 5[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-46:702886:703437 [6] NCCL INFO Channel 03/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9631:10144 [6] NCCL INFO Channel 01/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104435:104947 [6] NCCL INFO Channel 03/0 : 5[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9631:10144 [6] NCCL INFO Channel 02/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104435:104947 [6] NCCL INFO Channel 00/0 : 6[6] -> 7[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9631:10144 [6] NCCL INFO Channel 03/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104435:104947 [6] NCCL INFO Channel 01/0 : 6[6] -> 7[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9631:10144 [6] NCCL INFO Channel 00/0 : 1[6] -> 2[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104435:104947 [6] NCCL INFO Channel 02/0 : 6[6] -> 7[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897246:897760 [6] NCCL INFO Channel 00/0 : 4[6] -> 5[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9631:10144 [6] NCCL INFO Channel 01/0 : 1[6] -> 2[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104435:104947 [6] NCCL INFO Channel 03/0 : 6[6] -> 7[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897246:897760 [6] NCCL INFO Channel 01/0 : 4[6] -> 5[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9631:10144 [6] NCCL INFO Channel 02/0 : 1[6] -> 2[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9631:10144 [6] NCCL INFO Channel 03/0 : 1[6] -> 2[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897246:897760 [6] NCCL INFO Channel 02/0 : 4[6] -> 5[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897246:897760 [6] NCCL INFO Channel 03/0 : 4[6] -> 5[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897246:897760 [6] NCCL INFO Channel 00/0 : 5[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897246:897760 [6] NCCL INFO Channel 01/0 : 5[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897246:897760 [6] NCCL INFO Channel 02/0 : 5[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897246:897760 [6] NCCL INFO Channel 03/0 : 5[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-167-9:1039540:1040064 [3] NCCL INFO Connected all rings
+ip-26-0-167-9:1039540:1040064 [3] NCCL INFO Channel 02/0 : 3[3] -> 7[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-167-9:1039540:1040064 [3] NCCL INFO Channel 03/0 : 3[3] -> 7[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-167-9:1039540:1040064 [3] NCCL INFO Channel 02/0 : 7[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-167-9:1039540:1040064 [3] NCCL INFO Channel 03/0 : 7[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-167-9:1039540:1040064 [3] NCCL INFO Channel 00/0 : 7[3] -> 6[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-167-9:1039540:1040064 [3] NCCL INFO Channel 01/0 : 7[3] -> 6[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127293:127806 [3] NCCL INFO Channel 02/0 : 5[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127293:127806 [3] NCCL INFO Channel 03/0 : 5[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127293:127806 [3] NCCL INFO Channel 02/0 : 3[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127293:127806 [3] NCCL INFO Channel 03/0 : 3[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897243:897754 [3] NCCL INFO Channel 00/0 : 6[3] -> 5[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127293:127806 [3] NCCL INFO Channel 00/0 : 3[3] -> 2[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897243:897754 [3] NCCL INFO Channel 01/0 : 6[3] -> 5[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127293:127806 [3] NCCL INFO Channel 01/0 : 3[3] -> 2[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897243:897754 [3] NCCL INFO Channel 02/0 : 6[3] -> 5[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9628:10141 [3] NCCL INFO Channel 00/0 : 2[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897243:897754 [3] NCCL INFO Channel 03/0 : 6[3] -> 5[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9628:10141 [3] NCCL INFO Channel 01/0 : 2[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897243:897754 [3] NCCL INFO Channel 02/0 : 5[3] -> 4[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9628:10141 [3] NCCL INFO Channel 02/0 : 2[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897243:897754 [3] NCCL INFO Channel 03/0 : 5[3] -> 4[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9628:10141 [3] NCCL INFO Channel 03/0 : 2[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9628:10141 [3] NCCL INFO Channel 02/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9628:10141 [3] NCCL INFO Channel 03/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9627:10142 [2] NCCL INFO Connected all rings
+ip-26-0-165-213:9627:10142 [2] NCCL INFO Channel 02/0 : 1[2] -> 3[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9627:10142 [2] NCCL INFO Channel 03/0 : 1[2] -> 3[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104433:104950 [4] NCCL INFO Channel 00/0 : 5[4] -> 6[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9627:10142 [2] NCCL INFO Channel 02/0 : 3[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9627:10142 [2] NCCL INFO Channel 03/0 : 3[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104433:104950 [4] NCCL INFO Channel 01/0 : 5[4] -> 6[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127292:127807 [2] NCCL INFO Channel 02/0 : 7[2] -> 3[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104433:104950 [4] NCCL INFO Channel 02/0 : 5[4] -> 6[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127292:127807 [2] NCCL INFO Channel 03/0 : 7[2] -> 3[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104433:104950 [4] NCCL INFO Channel 03/0 : 5[4] -> 6[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127292:127807 [2] NCCL INFO Channel 02/0 : 3[2] -> 7[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104433:104950 [4] NCCL INFO Channel 00/0 : 6[4] -> 7[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127292:127807 [2] NCCL INFO Channel 03/0 : 3[2] -> 7[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104433:104950 [4] NCCL INFO Channel 01/0 : 6[4] -> 7[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104433:104950 [4] NCCL INFO Channel 02/0 : 6[4] -> 7[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104433:104950 [4] NCCL INFO Channel 03/0 : 6[4] -> 7[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897244:897763 [4] NCCL INFO Channel 00/0 : 4[4] -> 5[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897244:897763 [4] NCCL INFO Channel 01/0 : 4[4] -> 5[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-167-9:1039541:1040074 [4] NCCL INFO Channel 00/0 : 6[4] -> 7[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897244:897763 [4] NCCL INFO Channel 02/0 : 4[4] -> 5[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762819:763351 [4] NCCL INFO Channel 00/0 : 1[4] -> 2[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897244:897763 [4] NCCL INFO Channel 03/0 : 4[4] -> 5[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-167-9:1039541:1040074 [4] NCCL INFO Channel 01/0 : 6[4] -> 7[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-46:702884:703443 [4] NCCL INFO Channel 00/0 : 7[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762819:763351 [4] NCCL INFO Channel 01/0 : 1[4] -> 2[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897244:897763 [4] NCCL INFO Channel 00/0 : 5[4] -> 6[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-167-9:1039541:1040074 [4] NCCL INFO Channel 02/0 : 6[4] -> 7[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762819:763351 [4] NCCL INFO Channel 02/0 : 1[4] -> 2[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-46:702884:703443 [4] NCCL INFO Channel 01/0 : 7[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897244:897763 [4] NCCL INFO Channel 01/0 : 5[4] -> 6[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-167-9:1039541:1040074 [4] NCCL INFO Channel 03/0 : 6[4] -> 7[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624947:625469 [4] NCCL INFO Channel 00/0 : 3[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762819:763351 [4] NCCL INFO Channel 03/0 : 1[4] -> 2[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897244:897763 [4] NCCL INFO Channel 02/0 : 5[4] -> 6[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-46:702884:703443 [4] NCCL INFO Channel 02/0 : 7[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127294:127812 [4] NCCL INFO Channel 00/0 : 2[4] -> 3[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-167-9:1039541:1040074 [4] NCCL INFO Channel 00/0 : 7[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624947:625469 [4] NCCL INFO Channel 01/0 : 3[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762819:763351 [4] NCCL INFO Channel 00/0 : 2[4] -> 3[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897244:897763 [4] NCCL INFO Channel 03/0 : 5[4] -> 6[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-46:702884:703443 [4] NCCL INFO Channel 03/0 : 7[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-167-9:1039541:1040074 [4] NCCL INFO Channel 01/0 : 7[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127294:127812 [4] NCCL INFO Channel 01/0 : 2[4] -> 3[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762819:763351 [4] NCCL INFO Channel 01/0 : 2[4] -> 3[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624947:625469 [4] NCCL INFO Channel 02/0 : 3[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-46:702884:703443 [4] NCCL INFO Channel 00/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-167-9:1039541:1040074 [4] NCCL INFO Channel 02/0 : 7[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127294:127812 [4] NCCL INFO Channel 02/0 : 2[4] -> 3[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-46:702884:703443 [4] NCCL INFO Channel 01/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624947:625469 [4] NCCL INFO Channel 03/0 : 3[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762819:763351 [4] NCCL INFO Channel 02/0 : 2[4] -> 3[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-167-9:1039541:1040074 [4] NCCL INFO Channel 03/0 : 7[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9629:10147 [4] NCCL INFO Channel 00/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-46:702884:703443 [4] NCCL INFO Channel 02/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127294:127812 [4] NCCL INFO Channel 03/0 : 2[4] -> 3[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624947:625469 [4] NCCL INFO Channel 00/0 : 4[4] -> 5[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762819:763351 [4] NCCL INFO Channel 03/0 : 2[4] -> 3[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-46:702884:703443 [4] NCCL INFO Channel 03/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9629:10147 [4] NCCL INFO Channel 01/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624947:625469 [4] NCCL INFO Channel 01/0 : 4[4] -> 5[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127294:127812 [4] NCCL INFO Channel 00/0 : 3[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624947:625469 [4] NCCL INFO Channel 02/0 : 4[4] -> 5[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127294:127812 [4] NCCL INFO Channel 01/0 : 3[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9629:10147 [4] NCCL INFO Channel 02/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624947:625469 [4] NCCL INFO Channel 03/0 : 4[4] -> 5[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127294:127812 [4] NCCL INFO Channel 02/0 : 3[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9629:10147 [4] NCCL INFO Channel 03/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127294:127812 [4] NCCL INFO Channel 03/0 : 3[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9629:10147 [4] NCCL INFO Channel 00/0 : 1[4] -> 2[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9629:10147 [4] NCCL INFO Channel 01/0 : 1[4] -> 2[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9629:10147 [4] NCCL INFO Channel 02/0 : 1[4] -> 2[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9629:10147 [4] NCCL INFO Channel 03/0 : 1[4] -> 2[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-167-9:1039539:1040065 [2] NCCL INFO Connected all rings
+ip-26-0-167-9:1039539:1040065 [2] NCCL INFO Channel 02/0 : 3[2] -> 7[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-167-9:1039539:1040065 [2] NCCL INFO Channel 03/0 : 3[2] -> 7[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-167-9:1039539:1040065 [2] NCCL INFO Channel 02/0 : 7[2] -> 3[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-167-9:1039539:1040065 [2] NCCL INFO Channel 03/0 : 7[2] -> 3[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-167-9:1039539:1040065 [2] NCCL INFO Channel 00/0 : 7[2] -> 6[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-167-9:1039539:1040065 [2] NCCL INFO Channel 01/0 : 7[2] -> 6[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127292:127807 [2] NCCL INFO Channel 02/0 : 5[2] -> 3[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127292:127807 [2] NCCL INFO Channel 03/0 : 5[2] -> 3[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127292:127807 [2] NCCL INFO Channel 02/0 : 3[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127292:127807 [2] NCCL INFO Channel 03/0 : 3[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897242:897755 [2] NCCL INFO Channel 00/0 : 6[2] -> 5[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897242:897755 [2] NCCL INFO Channel 01/0 : 6[2] -> 5[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127292:127807 [2] NCCL INFO Channel 00/0 : 3[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9627:10142 [2] NCCL INFO Channel 00/0 : 2[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762820:763352 [5] NCCL INFO Channel 00/0 : 1[5] -> 2[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897242:897755 [2] NCCL INFO Channel 02/0 : 6[2] -> 5[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127292:127807 [2] NCCL INFO Channel 01/0 : 3[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9627:10142 [2] NCCL INFO Channel 01/0 : 2[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762820:763352 [5] NCCL INFO Channel 01/0 : 1[5] -> 2[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897242:897755 [2] NCCL INFO Channel 03/0 : 6[2] -> 5[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9627:10142 [2] NCCL INFO Channel 02/0 : 2[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897242:897755 [2] NCCL INFO Channel 02/0 : 5[2] -> 4[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9627:10142 [2] NCCL INFO Channel 03/0 : 2[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762820:763352 [5] NCCL INFO Channel 02/0 : 1[5] -> 2[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897242:897755 [2] NCCL INFO Channel 03/0 : 5[2] -> 4[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9627:10142 [2] NCCL INFO Channel 02/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762820:763352 [5] NCCL INFO Channel 03/0 : 1[5] -> 2[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9627:10142 [2] NCCL INFO Channel 03/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762820:763352 [5] NCCL INFO Channel 00/0 : 2[5] -> 3[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762820:763352 [5] NCCL INFO Channel 01/0 : 2[5] -> 3[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762820:763352 [5] NCCL INFO Channel 02/0 : 2[5] -> 3[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762820:763352 [5] NCCL INFO Channel 03/0 : 2[5] -> 3[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104434:104951 [5] NCCL INFO Channel 00/0 : 5[5] -> 6[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104434:104951 [5] NCCL INFO Channel 01/0 : 5[5] -> 6[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104434:104951 [5] NCCL INFO Channel 02/0 : 5[5] -> 6[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104434:104951 [5] NCCL INFO Channel 03/0 : 5[5] -> 6[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-46:702885:703445 [5] NCCL INFO Channel 00/0 : 7[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104434:104951 [5] NCCL INFO Channel 00/0 : 6[5] -> 7[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-46:702885:703445 [5] NCCL INFO Channel 01/0 : 7[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104434:104951 [5] NCCL INFO Channel 01/0 : 6[5] -> 7[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624948:625470 [5] NCCL INFO Channel 00/0 : 3[5] -> 4[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-46:702885:703445 [5] NCCL INFO Channel 02/0 : 7[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104434:104951 [5] NCCL INFO Channel 02/0 : 6[5] -> 7[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624948:625470 [5] NCCL INFO Channel 01/0 : 3[5] -> 4[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-46:702885:703445 [5] NCCL INFO Channel 03/0 : 7[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104434:104951 [5] NCCL INFO Channel 03/0 : 6[5] -> 7[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127295:127813 [5] NCCL INFO Channel 00/0 : 2[5] -> 3[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624948:625470 [5] NCCL INFO Channel 02/0 : 3[5] -> 4[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-46:702885:703445 [5] NCCL INFO Channel 00/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127295:127813 [5] NCCL INFO Channel 01/0 : 2[5] -> 3[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624948:625470 [5] NCCL INFO Channel 03/0 : 3[5] -> 4[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-46:702885:703445 [5] NCCL INFO Channel 01/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624948:625470 [5] NCCL INFO Channel 00/0 : 4[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127295:127813 [5] NCCL INFO Channel 02/0 : 2[5] -> 3[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-46:702885:703445 [5] NCCL INFO Channel 02/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9630:10148 [5] NCCL INFO Channel 00/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624948:625470 [5] NCCL INFO Channel 01/0 : 4[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127295:127813 [5] NCCL INFO Channel 03/0 : 2[5] -> 3[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-46:702885:703445 [5] NCCL INFO Channel 03/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9630:10148 [5] NCCL INFO Channel 01/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624948:625470 [5] NCCL INFO Channel 02/0 : 4[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127295:127813 [5] NCCL INFO Channel 00/0 : 3[5] -> 4[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9630:10148 [5] NCCL INFO Channel 02/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624948:625470 [5] NCCL INFO Channel 03/0 : 4[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127295:127813 [5] NCCL INFO Channel 01/0 : 3[5] -> 4[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9630:10148 [5] NCCL INFO Channel 03/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127295:127813 [5] NCCL INFO Channel 02/0 : 3[5] -> 4[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9630:10148 [5] NCCL INFO Channel 00/0 : 1[5] -> 2[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127295:127813 [5] NCCL INFO Channel 03/0 : 3[5] -> 4[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9630:10148 [5] NCCL INFO Channel 01/0 : 1[5] -> 2[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9630:10148 [5] NCCL INFO Channel 02/0 : 1[5] -> 2[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9630:10148 [5] NCCL INFO Channel 03/0 : 1[5] -> 2[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897245:897764 [5] NCCL INFO Channel 00/0 : 4[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897245:897764 [5] NCCL INFO Channel 01/0 : 4[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897245:897764 [5] NCCL INFO Channel 02/0 : 4[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897245:897764 [5] NCCL INFO Channel 03/0 : 4[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897245:897764 [5] NCCL INFO Channel 00/0 : 5[5] -> 6[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897245:897764 [5] NCCL INFO Channel 01/0 : 5[5] -> 6[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897245:897764 [5] NCCL INFO Channel 02/0 : 5[5] -> 6[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897245:897764 [5] NCCL INFO Channel 03/0 : 5[5] -> 6[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-167-9:1039542:1040075 [5] NCCL INFO Channel 00/0 : 6[5] -> 7[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-167-9:1039542:1040075 [5] NCCL INFO Channel 01/0 : 6[5] -> 7[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-167-9:1039542:1040075 [5] NCCL INFO Channel 02/0 : 6[5] -> 7[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-167-9:1039542:1040075 [5] NCCL INFO Channel 03/0 : 6[5] -> 7[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-167-9:1039542:1040075 [5] NCCL INFO Channel 00/0 : 7[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-167-9:1039542:1040075 [5] NCCL INFO Channel 01/0 : 7[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-167-9:1039542:1040075 [5] NCCL INFO Channel 02/0 : 7[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-167-9:1039542:1040075 [5] NCCL INFO Channel 03/0 : 7[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-46:702882:703433 [2] NCCL INFO Connected all rings
+ip-26-0-162-46:702882:703433 [2] NCCL INFO Channel 00/0 : 4[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-46:702882:703433 [2] NCCL INFO Channel 01/0 : 4[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-46:702882:703433 [2] NCCL INFO Channel 00/0 : 0[2] -> 4[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-46:702882:703433 [2] NCCL INFO Channel 01/0 : 0[2] -> 4[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-46:702882:703433 [2] NCCL INFO Channel 02/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-46:702882:703433 [2] NCCL INFO Channel 03/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624945:625463 [2] NCCL INFO Channel 00/0 : 6[2] -> 4[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624945:625463 [2] NCCL INFO Channel 01/0 : 6[2] -> 4[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624945:625463 [2] NCCL INFO Channel 00/0 : 4[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624945:625463 [2] NCCL INFO Channel 01/0 : 4[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624945:625463 [2] NCCL INFO Channel 02/0 : 5[2] -> 4[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104431:104945 [2] NCCL INFO Channel 00/0 : 7[2] -> 6[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624945:625463 [2] NCCL INFO Channel 03/0 : 5[2] -> 4[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104431:104945 [2] NCCL INFO Channel 01/0 : 7[2] -> 6[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104431:104945 [2] NCCL INFO Channel 00/0 : 6[2] -> 5[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104431:104945 [2] NCCL INFO Channel 01/0 : 6[2] -> 5[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104431:104945 [2] NCCL INFO Channel 02/0 : 6[2] -> 5[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104431:104945 [2] NCCL INFO Channel 03/0 : 6[2] -> 5[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762817:763345 [2] NCCL INFO Channel 00/0 : 3[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762817:763345 [2] NCCL INFO Channel 01/0 : 3[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762817:763345 [2] NCCL INFO Channel 00/0 : 2[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762817:763345 [2] NCCL INFO Channel 01/0 : 2[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762817:763345 [2] NCCL INFO Channel 02/0 : 2[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762817:763345 [2] NCCL INFO Channel 03/0 : 2[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-46:702883:703431 [3] NCCL INFO Connected all rings
+ip-26-0-162-46:702883:703431 [3] NCCL INFO Channel 00/0 : 4[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-46:702883:703431 [3] NCCL INFO Channel 01/0 : 4[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-46:702883:703431 [3] NCCL INFO Channel 00/0 : 0[3] -> 4[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-46:702883:703431 [3] NCCL INFO Channel 01/0 : 0[3] -> 4[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-46:702883:703431 [3] NCCL INFO Channel 02/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-46:702883:703431 [3] NCCL INFO Channel 03/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624946:625462 [3] NCCL INFO Channel 00/0 : 6[3] -> 4[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624946:625462 [3] NCCL INFO Channel 01/0 : 6[3] -> 4[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624946:625462 [3] NCCL INFO Channel 00/0 : 4[3] -> 2[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624946:625462 [3] NCCL INFO Channel 01/0 : 4[3] -> 2[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624946:625462 [3] NCCL INFO Channel 02/0 : 5[3] -> 4[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104432:104944 [3] NCCL INFO Channel 00/0 : 7[3] -> 6[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624946:625462 [3] NCCL INFO Channel 03/0 : 5[3] -> 4[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104432:104944 [3] NCCL INFO Channel 01/0 : 7[3] -> 6[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104432:104944 [3] NCCL INFO Channel 00/0 : 6[3] -> 5[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104432:104944 [3] NCCL INFO Channel 01/0 : 6[3] -> 5[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104432:104944 [3] NCCL INFO Channel 02/0 : 6[3] -> 5[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104432:104944 [3] NCCL INFO Channel 03/0 : 6[3] -> 5[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762818:763344 [3] NCCL INFO Channel 00/0 : 3[3] -> 2[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762818:763344 [3] NCCL INFO Channel 01/0 : 3[3] -> 2[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762818:763344 [3] NCCL INFO Channel 00/0 : 2[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762818:763344 [3] NCCL INFO Channel 01/0 : 2[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762818:763344 [3] NCCL INFO Channel 02/0 : 2[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762818:763344 [3] NCCL INFO Channel 03/0 : 2[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9626:10146 [1] NCCL INFO Connected all rings
+ip-26-0-165-213:9626:10146 [1] NCCL INFO Channel 02/0 : 1[1] -> 3[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9626:10146 [1] NCCL INFO Channel 03/0 : 1[1] -> 3[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104429:104948 [0] NCCL INFO Connected all rings
+ip-26-0-166-36:104429:104948 [0] NCCL INFO Channel 00/0 : 4[0] -> 6[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104429:104948 [0] NCCL INFO Channel 01/0 : 4[0] -> 6[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624943:625466 [0] NCCL INFO Connected all rings
+ip-26-0-166-214:624943:625466 [0] NCCL INFO Channel 00/0 : 2[0] -> 4[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624943:625466 [0] NCCL INFO Channel 01/0 : 2[0] -> 4[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624943:625466 [0] NCCL INFO Channel 00/0 : 4[0] -> 6[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624943:625466 [0] NCCL INFO Channel 01/0 : 4[0] -> 6[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104429:104948 [0] NCCL INFO Channel 00/0 : 6[0] -> 4[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104429:104948 [0] NCCL INFO Channel 01/0 : 6[0] -> 4[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897240:897761 [0] NCCL INFO Connected all rings
+ip-26-0-166-244:897240:897761 [0] NCCL INFO Channel 02/0 : 3[0] -> 5[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897240:897761 [0] NCCL INFO Channel 03/0 : 3[0] -> 5[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9625:10145 [0] NCCL INFO Connected all rings
+ip-26-0-165-213:9625:10145 [0] NCCL INFO Channel 02/0 : 1[0] -> 3[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9625:10145 [0] NCCL INFO Channel 03/0 : 1[0] -> 3[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624944:625467 [1] NCCL INFO Connected all rings
+ip-26-0-166-214:624944:625467 [1] NCCL INFO Channel 00/0 : 2[1] -> 4[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624944:625467 [1] NCCL INFO Channel 01/0 : 2[1] -> 4[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624944:625467 [1] NCCL INFO Channel 00/0 : 4[1] -> 6[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624944:625467 [1] NCCL INFO Channel 01/0 : 4[1] -> 6[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-167-9:1039537:1040070 [0] NCCL INFO Connected all rings
+ip-26-0-167-9:1039537:1040070 [0] NCCL INFO Channel 02/0 : 3[0] -> 7[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-167-9:1039537:1040070 [0] NCCL INFO Channel 03/0 : 3[0] -> 7[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-167-9:1039537:1040070 [0] NCCL INFO Channel 02/0 : 7[0] -> 3[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-167-9:1039537:1040070 [0] NCCL INFO Channel 03/0 : 7[0] -> 3[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127297:127808 [7] NCCL INFO Connected all rings
+ip-26-0-166-15:127297:127808 [7] NCCL INFO Channel 02/0 : 1[7] -> 3[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127297:127808 [7] NCCL INFO Channel 03/0 : 1[7] -> 3[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127297:127808 [7] NCCL INFO Channel 02/0 : 3[7] -> 5[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127297:127808 [7] NCCL INFO Channel 03/0 : 3[7] -> 5[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897247:897759 [7] NCCL INFO Connected all rings
+ip-26-0-166-244:897247:897759 [7] NCCL INFO Channel 02/0 : 3[7] -> 5[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897247:897759 [7] NCCL INFO Channel 03/0 : 3[7] -> 5[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897247:897759 [7] NCCL INFO Channel 02/0 : 5[7] -> 3[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897247:897759 [7] NCCL INFO Channel 03/0 : 5[7] -> 3[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762816:763349 [1] NCCL INFO Connected all rings
+ip-26-0-166-125:762816:763349 [1] NCCL INFO Channel 00/0 : 2[1] -> 4[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762816:763349 [1] NCCL INFO Channel 01/0 : 2[1] -> 4[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762816:763349 [1] NCCL INFO Channel 00/0 : 4[1] -> 2[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762816:763349 [1] NCCL INFO Channel 01/0 : 4[1] -> 2[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127291:127811 [1] NCCL INFO Connected all rings
+ip-26-0-166-15:127291:127811 [1] NCCL INFO Channel 02/0 : 1[1] -> 3[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127291:127811 [1] NCCL INFO Channel 03/0 : 1[1] -> 3[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127291:127811 [1] NCCL INFO Channel 02/0 : 3[1] -> 5[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127291:127811 [1] NCCL INFO Channel 03/0 : 3[1] -> 5[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9626:10146 [1] NCCL INFO Channel 02/0 : 3[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9626:10146 [1] NCCL INFO Channel 03/0 : 3[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897241:897762 [1] NCCL INFO Connected all rings
+ip-26-0-166-244:897241:897762 [1] NCCL INFO Channel 02/0 : 3[1] -> 5[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897241:897762 [1] NCCL INFO Channel 03/0 : 3[1] -> 5[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897241:897762 [1] NCCL INFO Channel 02/0 : 5[1] -> 3[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127291:127811 [1] NCCL INFO Channel 02/0 : 7[1] -> 3[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897241:897762 [1] NCCL INFO Channel 03/0 : 5[1] -> 3[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127291:127811 [1] NCCL INFO Channel 03/0 : 7[1] -> 3[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127291:127811 [1] NCCL INFO Channel 02/0 : 3[1] -> 7[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127291:127811 [1] NCCL INFO Channel 03/0 : 3[1] -> 7[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9632:10143 [7] NCCL INFO Connected all rings
+ip-26-0-167-9:1039544:1040068 [7] NCCL INFO Connected all rings
+ip-26-0-165-213:9632:10143 [7] NCCL INFO Channel 02/0 : 1[7] -> 3[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-167-9:1039544:1040068 [7] NCCL INFO Channel 02/0 : 3[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9632:10143 [7] NCCL INFO Channel 03/0 : 1[7] -> 3[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-167-9:1039544:1040068 [7] NCCL INFO Channel 03/0 : 3[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-167-9:1039544:1040068 [7] NCCL INFO Channel 02/0 : 7[7] -> 3[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-167-9:1039544:1040068 [7] NCCL INFO Channel 03/0 : 7[7] -> 3[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9632:10143 [7] NCCL INFO Channel 02/0 : 3[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127297:127808 [7] NCCL INFO Channel 02/0 : 7[7] -> 3[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9632:10143 [7] NCCL INFO Channel 03/0 : 3[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127297:127808 [7] NCCL INFO Channel 03/0 : 7[7] -> 3[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127297:127808 [7] NCCL INFO Channel 02/0 : 3[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127297:127808 [7] NCCL INFO Channel 03/0 : 3[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127297:127808 [7] NCCL INFO Channel 02/0 : 5[7] -> 3[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-167-9:1039544:1040068 [7] NCCL INFO Channel 00/0 : 7[7] -> 6[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127297:127808 [7] NCCL INFO Channel 03/0 : 5[7] -> 3[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-167-9:1039544:1040068 [7] NCCL INFO Channel 01/0 : 7[7] -> 6[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127297:127808 [7] NCCL INFO Channel 02/0 : 3[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127297:127808 [7] NCCL INFO Channel 03/0 : 3[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897247:897759 [7] NCCL INFO Channel 00/0 : 6[7] -> 5[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897247:897759 [7] NCCL INFO Channel 01/0 : 6[7] -> 5[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127297:127808 [7] NCCL INFO Channel 00/0 : 3[7] -> 2[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9632:10143 [7] NCCL INFO Channel 00/0 : 2[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897247:897759 [7] NCCL INFO Channel 02/0 : 6[7] -> 5[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762815:763348 [0] NCCL INFO Connected all rings
+ip-26-0-165-213:9632:10143 [7] NCCL INFO Channel 01/0 : 2[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127297:127808 [7] NCCL INFO Channel 01/0 : 3[7] -> 2[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897247:897759 [7] NCCL INFO Channel 03/0 : 6[7] -> 5[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762815:763348 [0] NCCL INFO Channel 00/0 : 2[0] -> 4[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9632:10143 [7] NCCL INFO Channel 02/0 : 2[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762815:763348 [0] NCCL INFO Channel 01/0 : 2[0] -> 4[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897247:897759 [7] NCCL INFO Channel 02/0 : 5[7] -> 4[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9632:10143 [7] NCCL INFO Channel 03/0 : 2[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762815:763348 [0] NCCL INFO Channel 00/0 : 4[0] -> 2[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897247:897759 [7] NCCL INFO Channel 03/0 : 5[7] -> 4[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9632:10143 [7] NCCL INFO Channel 02/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624943:625466 [0] NCCL INFO Channel 00/0 : 0[0] -> 4[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762815:763348 [0] NCCL INFO Channel 01/0 : 4[0] -> 2[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9632:10143 [7] NCCL INFO Channel 03/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624943:625466 [0] NCCL INFO Channel 01/0 : 0[0] -> 4[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624943:625466 [0] NCCL INFO Channel 00/0 : 4[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104430:104949 [1] NCCL INFO Connected all rings
+ip-26-0-166-214:624943:625466 [0] NCCL INFO Channel 01/0 : 4[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104430:104949 [1] NCCL INFO Channel 00/0 : 4[1] -> 6[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104430:104949 [1] NCCL INFO Channel 01/0 : 4[1] -> 6[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104430:104949 [1] NCCL INFO Channel 00/0 : 6[1] -> 4[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624944:625467 [1] NCCL INFO Channel 00/0 : 0[1] -> 4[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104430:104949 [1] NCCL INFO Channel 01/0 : 6[1] -> 4[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624944:625467 [1] NCCL INFO Channel 01/0 : 0[1] -> 4[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624944:625467 [1] NCCL INFO Channel 00/0 : 4[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624944:625467 [1] NCCL INFO Channel 01/0 : 4[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-46:702880:703439 [0] NCCL INFO Connected all rings
+ip-26-0-162-46:702880:703439 [0] NCCL INFO Channel 00/0 : 4[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-46:702880:703439 [0] NCCL INFO Channel 01/0 : 4[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-46:702880:703439 [0] NCCL INFO Channel 00/0 : 0[0] -> 4[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-46:702880:703439 [0] NCCL INFO Channel 01/0 : 0[0] -> 4[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-46:702880:703439 [0] NCCL INFO Channel 02/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-46:702880:703439 [0] NCCL INFO Channel 03/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624943:625466 [0] NCCL INFO Channel 00/0 : 6[0] -> 4[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624943:625466 [0] NCCL INFO Channel 01/0 : 6[0] -> 4[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624943:625466 [0] NCCL INFO Channel 00/0 : 4[0] -> 2[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624943:625466 [0] NCCL INFO Channel 01/0 : 4[0] -> 2[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-167-9:1039538:1040071 [1] NCCL INFO Connected all rings
+ip-26-0-167-9:1039538:1040071 [1] NCCL INFO Channel 02/0 : 3[1] -> 7[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-167-9:1039538:1040071 [1] NCCL INFO Channel 03/0 : 3[1] -> 7[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104429:104948 [0] NCCL INFO Channel 00/0 : 7[0] -> 6[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-167-9:1039538:1040071 [1] NCCL INFO Channel 02/0 : 7[1] -> 3[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104429:104948 [0] NCCL INFO Channel 01/0 : 7[0] -> 6[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624943:625466 [0] NCCL INFO Channel 02/0 : 5[0] -> 4[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762815:763348 [0] NCCL INFO Channel 00/0 : 3[0] -> 2[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-167-9:1039538:1040071 [1] NCCL INFO Channel 03/0 : 7[1] -> 3[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104429:104948 [0] NCCL INFO Channel 00/0 : 6[0] -> 5[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624943:625466 [0] NCCL INFO Channel 03/0 : 5[0] -> 4[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762815:763348 [0] NCCL INFO Channel 01/0 : 3[0] -> 2[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104429:104948 [0] NCCL INFO Channel 01/0 : 6[0] -> 5[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762815:763348 [0] NCCL INFO Channel 00/0 : 2[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104429:104948 [0] NCCL INFO Channel 02/0 : 6[0] -> 5[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762815:763348 [0] NCCL INFO Channel 01/0 : 2[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-167-9:1039538:1040071 [1] NCCL INFO Channel 00/0 : 7[1] -> 6[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104429:104948 [0] NCCL INFO Channel 03/0 : 6[0] -> 5[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762815:763348 [0] NCCL INFO Channel 02/0 : 2[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127291:127811 [1] NCCL INFO Channel 02/0 : 5[1] -> 3[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-167-9:1039538:1040071 [1] NCCL INFO Channel 01/0 : 7[1] -> 6[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762815:763348 [0] NCCL INFO Channel 03/0 : 2[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127291:127811 [1] NCCL INFO Channel 03/0 : 5[1] -> 3[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127291:127811 [1] NCCL INFO Channel 02/0 : 3[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127291:127811 [1] NCCL INFO Channel 03/0 : 3[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897241:897762 [1] NCCL INFO Channel 00/0 : 6[1] -> 5[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897241:897762 [1] NCCL INFO Channel 01/0 : 6[1] -> 5[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127291:127811 [1] NCCL INFO Channel 00/0 : 3[1] -> 2[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897241:897762 [1] NCCL INFO Channel 02/0 : 6[1] -> 5[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127291:127811 [1] NCCL INFO Channel 01/0 : 3[1] -> 2[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897241:897762 [1] NCCL INFO Channel 03/0 : 6[1] -> 5[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897241:897762 [1] NCCL INFO Channel 02/0 : 5[1] -> 4[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897241:897762 [1] NCCL INFO Channel 03/0 : 5[1] -> 4[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127290:127810 [0] NCCL INFO Connected all rings
+ip-26-0-166-15:127290:127810 [0] NCCL INFO Channel 02/0 : 1[0] -> 3[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127290:127810 [0] NCCL INFO Channel 03/0 : 1[0] -> 3[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127290:127810 [0] NCCL INFO Channel 02/0 : 3[0] -> 5[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127290:127810 [0] NCCL INFO Channel 03/0 : 3[0] -> 5[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624950:625464 [7] NCCL INFO Connected all rings
+ip-26-0-166-214:624950:625464 [7] NCCL INFO Channel 00/0 : 2[7] -> 4[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762822:763346 [7] NCCL INFO Connected all rings
+ip-26-0-166-214:624950:625464 [7] NCCL INFO Channel 01/0 : 2[7] -> 4[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127290:127810 [0] NCCL INFO Channel 02/0 : 7[0] -> 3[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897240:897761 [0] NCCL INFO Channel 02/0 : 5[0] -> 3[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624950:625464 [7] NCCL INFO Channel 00/0 : 4[7] -> 6[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127290:127810 [0] NCCL INFO Channel 03/0 : 7[0] -> 3[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762822:763346 [7] NCCL INFO Channel 00/0 : 2[7] -> 4[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897240:897761 [0] NCCL INFO Channel 03/0 : 5[0] -> 3[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624950:625464 [7] NCCL INFO Channel 01/0 : 4[7] -> 6[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127290:127810 [0] NCCL INFO Channel 02/0 : 3[0] -> 7[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762822:763346 [7] NCCL INFO Channel 01/0 : 2[7] -> 4[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127290:127810 [0] NCCL INFO Channel 03/0 : 3[0] -> 7[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127290:127810 [0] NCCL INFO Channel 02/0 : 5[0] -> 3[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762822:763346 [7] NCCL INFO Channel 00/0 : 4[7] -> 2[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127290:127810 [0] NCCL INFO Channel 03/0 : 5[0] -> 3[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-167-9:1039537:1040070 [0] NCCL INFO Channel 00/0 : 7[0] -> 6[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762822:763346 [7] NCCL INFO Channel 01/0 : 4[7] -> 2[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-167-9:1039537:1040070 [0] NCCL INFO Channel 01/0 : 7[0] -> 6[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127290:127810 [0] NCCL INFO Channel 02/0 : 3[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127290:127810 [0] NCCL INFO Channel 03/0 : 3[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897240:897761 [0] NCCL INFO Channel 00/0 : 6[0] -> 5[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897240:897761 [0] NCCL INFO Channel 01/0 : 6[0] -> 5[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897240:897761 [0] NCCL INFO Channel 02/0 : 6[0] -> 5[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897240:897761 [0] NCCL INFO Channel 03/0 : 6[0] -> 5[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897240:897761 [0] NCCL INFO Channel 02/0 : 5[0] -> 4[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897240:897761 [0] NCCL INFO Channel 03/0 : 5[0] -> 4[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9625:10145 [0] NCCL INFO Channel 02/0 : 3[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9625:10145 [0] NCCL INFO Channel 03/0 : 3[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127290:127810 [0] NCCL INFO Channel 00/0 : 3[0] -> 2[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127290:127810 [0] NCCL INFO Channel 01/0 : 3[0] -> 2[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9626:10146 [1] NCCL INFO Channel 00/0 : 2[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9626:10146 [1] NCCL INFO Channel 01/0 : 2[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9626:10146 [1] NCCL INFO Channel 02/0 : 2[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9626:10146 [1] NCCL INFO Channel 03/0 : 2[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9626:10146 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9626:10146 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9625:10145 [0] NCCL INFO Channel 00/0 : 2[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9625:10145 [0] NCCL INFO Channel 01/0 : 2[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9625:10145 [0] NCCL INFO Channel 02/0 : 2[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9625:10145 [0] NCCL INFO Channel 03/0 : 2[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9625:10145 [0] NCCL INFO Channel 02/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9625:10145 [0] NCCL INFO Channel 03/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104436:104946 [7] NCCL INFO Connected all rings
+ip-26-0-166-36:104436:104946 [7] NCCL INFO Channel 00/0 : 4[7] -> 6[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104436:104946 [7] NCCL INFO Channel 01/0 : 4[7] -> 6[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104436:104946 [7] NCCL INFO Channel 00/0 : 6[7] -> 4[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624950:625464 [7] NCCL INFO Channel 00/0 : 0[7] -> 4[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104436:104946 [7] NCCL INFO Channel 01/0 : 6[7] -> 4[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624950:625464 [7] NCCL INFO Channel 01/0 : 0[7] -> 4[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624950:625464 [7] NCCL INFO Channel 00/0 : 4[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624950:625464 [7] NCCL INFO Channel 01/0 : 4[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-46:702887:703436 [7] NCCL INFO Connected all rings
+ip-26-0-162-46:702887:703436 [7] NCCL INFO Channel 00/0 : 4[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-46:702887:703436 [7] NCCL INFO Channel 01/0 : 4[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-46:702887:703436 [7] NCCL INFO Channel 00/0 : 0[7] -> 4[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-46:702887:703436 [7] NCCL INFO Channel 01/0 : 0[7] -> 4[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-46:702887:703436 [7] NCCL INFO Channel 02/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624950:625464 [7] NCCL INFO Channel 00/0 : 6[7] -> 4[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-46:702887:703436 [7] NCCL INFO Channel 03/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624950:625464 [7] NCCL INFO Channel 01/0 : 6[7] -> 4[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624950:625464 [7] NCCL INFO Channel 00/0 : 4[7] -> 2[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624950:625464 [7] NCCL INFO Channel 01/0 : 4[7] -> 2[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-46:702881:703441 [1] NCCL INFO Connected all rings
+ip-26-0-166-36:104436:104946 [7] NCCL INFO Channel 00/0 : 7[7] -> 6[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104436:104946 [7] NCCL INFO Channel 01/0 : 7[7] -> 6[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624950:625464 [7] NCCL INFO Channel 02/0 : 5[7] -> 4[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-46:702881:703441 [1] NCCL INFO Channel 00/0 : 4[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762822:763346 [7] NCCL INFO Channel 00/0 : 3[7] -> 2[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104436:104946 [7] NCCL INFO Channel 00/0 : 6[7] -> 5[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624950:625464 [7] NCCL INFO Channel 03/0 : 5[7] -> 4[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762822:763346 [7] NCCL INFO Channel 01/0 : 3[7] -> 2[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-46:702881:703441 [1] NCCL INFO Channel 01/0 : 4[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104436:104946 [7] NCCL INFO Channel 01/0 : 6[7] -> 5[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762822:763346 [7] NCCL INFO Channel 00/0 : 2[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-46:702881:703441 [1] NCCL INFO Channel 00/0 : 0[1] -> 4[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104436:104946 [7] NCCL INFO Channel 02/0 : 6[7] -> 5[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762822:763346 [7] NCCL INFO Channel 01/0 : 2[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-46:702881:703441 [1] NCCL INFO Channel 01/0 : 0[1] -> 4[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104436:104946 [7] NCCL INFO Channel 03/0 : 6[7] -> 5[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762822:763346 [7] NCCL INFO Channel 02/0 : 2[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762822:763346 [7] NCCL INFO Channel 03/0 : 2[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-46:702881:703441 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-46:702881:703441 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624944:625467 [1] NCCL INFO Channel 00/0 : 6[1] -> 4[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624944:625467 [1] NCCL INFO Channel 01/0 : 6[1] -> 4[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624944:625467 [1] NCCL INFO Channel 00/0 : 4[1] -> 2[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624944:625467 [1] NCCL INFO Channel 01/0 : 4[1] -> 2[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104430:104949 [1] NCCL INFO Channel 00/0 : 7[1] -> 6[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624944:625467 [1] NCCL INFO Channel 02/0 : 5[1] -> 4[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104430:104949 [1] NCCL INFO Channel 01/0 : 7[1] -> 6[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762816:763349 [1] NCCL INFO Channel 00/0 : 3[1] -> 2[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624944:625467 [1] NCCL INFO Channel 03/0 : 5[1] -> 4[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104430:104949 [1] NCCL INFO Channel 00/0 : 6[1] -> 5[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762816:763349 [1] NCCL INFO Channel 01/0 : 3[1] -> 2[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104430:104949 [1] NCCL INFO Channel 01/0 : 6[1] -> 5[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762816:763349 [1] NCCL INFO Channel 00/0 : 2[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104430:104949 [1] NCCL INFO Channel 02/0 : 6[1] -> 5[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762816:763349 [1] NCCL INFO Channel 01/0 : 2[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104430:104949 [1] NCCL INFO Channel 03/0 : 6[1] -> 5[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762816:763349 [1] NCCL INFO Channel 02/0 : 2[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762816:763349 [1] NCCL INFO Channel 03/0 : 2[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-167-9:1039539:1040065 [2] NCCL INFO Connected all trees
+ip-26-0-167-9:1039539:1040065 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-167-9:1039539:1040065 [2] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-167-9:1039539:1040065 [2] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-162-46:702882:703433 [2] NCCL INFO Connected all trees
+ip-26-0-162-46:702882:703433 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-162-46:702883:703431 [3] NCCL INFO Connected all trees
+ip-26-0-162-46:702883:703431 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-162-46:702882:703433 [2] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-162-46:702882:703433 [2] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-162-46:702883:703431 [3] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-162-46:702883:703431 [3] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-167-9:1039543:1040069 [6] NCCL INFO Connected all rings
+ip-26-0-167-9:1039543:1040069 [6] NCCL INFO Channel 02/0 : 3[6] -> 7[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-167-9:1039543:1040069 [6] NCCL INFO Channel 03/0 : 3[6] -> 7[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-167-9:1039543:1040069 [6] NCCL INFO Channel 02/0 : 7[6] -> 3[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-167-9:1039543:1040069 [6] NCCL INFO Channel 03/0 : 7[6] -> 3[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762821:763347 [6] NCCL INFO Connected all rings
+ip-26-0-166-125:762821:763347 [6] NCCL INFO Channel 00/0 : 2[6] -> 4[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762821:763347 [6] NCCL INFO Channel 01/0 : 2[6] -> 4[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624949:625465 [6] NCCL INFO Connected all rings
+ip-26-0-166-214:624949:625465 [6] NCCL INFO Channel 00/0 : 2[6] -> 4[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624949:625465 [6] NCCL INFO Channel 01/0 : 2[6] -> 4[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624949:625465 [6] NCCL INFO Channel 00/0 : 4[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624949:625465 [6] NCCL INFO Channel 01/0 : 4[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762821:763347 [6] NCCL INFO Channel 00/0 : 4[6] -> 2[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762821:763347 [6] NCCL INFO Channel 01/0 : 4[6] -> 2[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-167-9:1039539:1040065 [2] NCCL INFO comm 0x9beaec0 rank 7 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0x53075d22960e36e3 - Init COMPLETE
+ip-26-0-165-213:9631:10144 [6] NCCL INFO Connected all rings
+ip-26-0-165-213:9631:10144 [6] NCCL INFO Channel 02/0 : 1[6] -> 3[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9631:10144 [6] NCCL INFO Channel 03/0 : 1[6] -> 3[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897246:897760 [6] NCCL INFO Connected all rings
+ip-26-0-166-244:897246:897760 [6] NCCL INFO Channel 02/0 : 3[6] -> 5[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897246:897760 [6] NCCL INFO Channel 03/0 : 3[6] -> 5[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104435:104947 [6] NCCL INFO Connected all rings
+ip-26-0-166-36:104435:104947 [6] NCCL INFO Channel 00/0 : 4[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104435:104947 [6] NCCL INFO Channel 01/0 : 4[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104435:104947 [6] NCCL INFO Channel 00/0 : 6[6] -> 4[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624949:625465 [6] NCCL INFO Channel 00/0 : 0[6] -> 4[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104435:104947 [6] NCCL INFO Channel 01/0 : 6[6] -> 4[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624949:625465 [6] NCCL INFO Channel 01/0 : 0[6] -> 4[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624949:625465 [6] NCCL INFO Channel 00/0 : 4[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624949:625465 [6] NCCL INFO Channel 01/0 : 4[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-46:702882:703433 [2] NCCL INFO comm 0xae72820 rank 0 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0x53075d22960e36e3 - Init COMPLETE
+ip-26-0-162-46:702883:703431 [3] NCCL INFO comm 0x9c325c0 rank 0 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0xa976fd02a70fed12 - Init COMPLETE
+ip-26-0-162-46:702886:703437 [6] NCCL INFO Connected all rings
+ip-26-0-162-46:702886:703437 [6] NCCL INFO Channel 00/0 : 4[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-46:702886:703437 [6] NCCL INFO Channel 01/0 : 4[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-46:702886:703437 [6] NCCL INFO Channel 00/0 : 0[6] -> 4[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-46:702886:703437 [6] NCCL INFO Channel 01/0 : 0[6] -> 4[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-46:702886:703437 [6] NCCL INFO Channel 02/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624949:625465 [6] NCCL INFO Channel 00/0 : 6[6] -> 4[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-46:702886:703437 [6] NCCL INFO Channel 03/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624949:625465 [6] NCCL INFO Channel 01/0 : 6[6] -> 4[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624949:625465 [6] NCCL INFO Channel 00/0 : 4[6] -> 2[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624949:625465 [6] NCCL INFO Channel 01/0 : 4[6] -> 2[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104435:104947 [6] NCCL INFO Channel 00/0 : 7[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104435:104947 [6] NCCL INFO Channel 01/0 : 7[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624949:625465 [6] NCCL INFO Channel 02/0 : 5[6] -> 4[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-167-9:1039540:1040064 [3] NCCL INFO Connected all trees
+ip-26-0-167-9:1039540:1040064 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-125:762821:763347 [6] NCCL INFO Channel 00/0 : 3[6] -> 2[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104435:104947 [6] NCCL INFO Channel 00/0 : 6[6] -> 5[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-167-9:1039540:1040064 [3] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-167-9:1039540:1040064 [3] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-214:624949:625465 [6] NCCL INFO Channel 03/0 : 5[6] -> 4[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762821:763347 [6] NCCL INFO Channel 01/0 : 3[6] -> 2[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104435:104947 [6] NCCL INFO Channel 01/0 : 6[6] -> 5[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762821:763347 [6] NCCL INFO Channel 00/0 : 2[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104435:104947 [6] NCCL INFO Channel 02/0 : 6[6] -> 5[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104435:104947 [6] NCCL INFO Channel 03/0 : 6[6] -> 5[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762821:763347 [6] NCCL INFO Channel 01/0 : 2[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762821:763347 [6] NCCL INFO Channel 02/0 : 2[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762821:763347 [6] NCCL INFO Channel 03/0 : 2[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-167-9:1039542:1040075 [5] NCCL INFO Connected all rings
+ip-26-0-167-9:1039542:1040075 [5] NCCL INFO Channel 02/0 : 3[5] -> 7[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-167-9:1039542:1040075 [5] NCCL INFO Channel 03/0 : 3[5] -> 7[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-167-9:1039542:1040075 [5] NCCL INFO Channel 02/0 : 7[5] -> 3[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-167-9:1039542:1040075 [5] NCCL INFO Channel 03/0 : 7[5] -> 3[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-167-9:1039541:1040074 [4] NCCL INFO Connected all rings
+ip-26-0-167-9:1039541:1040074 [4] NCCL INFO Channel 02/0 : 3[4] -> 7[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-167-9:1039541:1040074 [4] NCCL INFO Channel 03/0 : 3[4] -> 7[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-167-9:1039541:1040074 [4] NCCL INFO Channel 02/0 : 7[4] -> 3[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-167-9:1039541:1040074 [4] NCCL INFO Channel 03/0 : 7[4] -> 3[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-167-9:1039540:1040064 [3] NCCL INFO comm 0xa261180 rank 7 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0xa976fd02a70fed12 - Init COMPLETE
+ip-26-0-166-125:762819:763351 [4] NCCL INFO Connected all rings
+ip-26-0-166-125:762819:763351 [4] NCCL INFO Channel 00/0 : 2[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762819:763351 [4] NCCL INFO Channel 01/0 : 2[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104433:104950 [4] NCCL INFO Connected all rings
+ip-26-0-166-36:104433:104950 [4] NCCL INFO Channel 00/0 : 4[4] -> 6[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104433:104950 [4] NCCL INFO Channel 01/0 : 4[4] -> 6[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897244:897763 [4] NCCL INFO Connected all rings
+ip-26-0-166-244:897244:897763 [4] NCCL INFO Channel 02/0 : 3[4] -> 5[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897244:897763 [4] NCCL INFO Channel 03/0 : 3[4] -> 5[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762820:763352 [5] NCCL INFO Connected all rings
+ip-26-0-166-125:762820:763352 [5] NCCL INFO Channel 00/0 : 2[5] -> 4[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762820:763352 [5] NCCL INFO Channel 01/0 : 2[5] -> 4[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897245:897764 [5] NCCL INFO Connected all rings
+ip-26-0-166-244:897245:897764 [5] NCCL INFO Channel 02/0 : 3[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897245:897764 [5] NCCL INFO Channel 03/0 : 3[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-46:702884:703443 [4] NCCL INFO Connected all rings
+ip-26-0-165-213:9629:10147 [4] NCCL INFO Connected all rings
+ip-26-0-162-46:702884:703443 [4] NCCL INFO Channel 00/0 : 4[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9629:10147 [4] NCCL INFO Channel 02/0 : 1[4] -> 3[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-46:702884:703443 [4] NCCL INFO Channel 01/0 : 4[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9629:10147 [4] NCCL INFO Channel 03/0 : 1[4] -> 3[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-46:702884:703443 [4] NCCL INFO Channel 00/0 : 0[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-46:702884:703443 [4] NCCL INFO Channel 01/0 : 0[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127296:127809 [6] NCCL INFO Connected all rings
+ip-26-0-166-15:127296:127809 [6] NCCL INFO Channel 02/0 : 1[6] -> 3[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127296:127809 [6] NCCL INFO Channel 03/0 : 1[6] -> 3[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127296:127809 [6] NCCL INFO Channel 02/0 : 3[6] -> 5[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127296:127809 [6] NCCL INFO Channel 03/0 : 3[6] -> 5[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9631:10144 [6] NCCL INFO Channel 02/0 : 3[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127296:127809 [6] NCCL INFO Channel 02/0 : 7[6] -> 3[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9631:10144 [6] NCCL INFO Channel 03/0 : 3[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897246:897760 [6] NCCL INFO Channel 02/0 : 5[6] -> 3[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127296:127809 [6] NCCL INFO Channel 03/0 : 7[6] -> 3[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897246:897760 [6] NCCL INFO Channel 03/0 : 5[6] -> 3[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127296:127809 [6] NCCL INFO Channel 02/0 : 3[6] -> 7[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127296:127809 [6] NCCL INFO Channel 03/0 : 3[6] -> 7[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127296:127809 [6] NCCL INFO Channel 02/0 : 5[6] -> 3[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127296:127809 [6] NCCL INFO Channel 03/0 : 5[6] -> 3[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127296:127809 [6] NCCL INFO Channel 02/0 : 3[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-167-9:1039543:1040069 [6] NCCL INFO Channel 00/0 : 7[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127296:127809 [6] NCCL INFO Channel 03/0 : 3[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-167-9:1039543:1040069 [6] NCCL INFO Channel 01/0 : 7[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897246:897760 [6] NCCL INFO Channel 00/0 : 6[6] -> 5[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897246:897760 [6] NCCL INFO Channel 01/0 : 6[6] -> 5[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9631:10144 [6] NCCL INFO Channel 00/0 : 2[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127296:127809 [6] NCCL INFO Channel 00/0 : 3[6] -> 2[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897246:897760 [6] NCCL INFO Channel 02/0 : 6[6] -> 5[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127296:127809 [6] NCCL INFO Channel 01/0 : 3[6] -> 2[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9631:10144 [6] NCCL INFO Channel 01/0 : 2[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897246:897760 [6] NCCL INFO Channel 03/0 : 6[6] -> 5[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9631:10144 [6] NCCL INFO Channel 02/0 : 2[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897246:897760 [6] NCCL INFO Channel 02/0 : 5[6] -> 4[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9631:10144 [6] NCCL INFO Channel 03/0 : 2[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897246:897760 [6] NCCL INFO Channel 03/0 : 5[6] -> 4[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9631:10144 [6] NCCL INFO Channel 02/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9630:10148 [5] NCCL INFO Connected all rings
+ip-26-0-165-213:9631:10144 [6] NCCL INFO Channel 03/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9630:10148 [5] NCCL INFO Channel 02/0 : 1[5] -> 3[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9630:10148 [5] NCCL INFO Channel 03/0 : 1[5] -> 3[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104431:104945 [2] NCCL INFO Connected all trees
+ip-26-0-166-36:104431:104945 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-36:104431:104945 [2] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-166-36:104431:104945 [2] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-162-46:702885:703445 [5] NCCL INFO Connected all rings
+ip-26-0-162-46:702885:703445 [5] NCCL INFO Channel 00/0 : 4[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-46:702885:703445 [5] NCCL INFO Channel 01/0 : 4[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-46:702885:703445 [5] NCCL INFO Channel 00/0 : 0[5] -> 4[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-46:702885:703445 [5] NCCL INFO Channel 01/0 : 0[5] -> 4[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104434:104951 [5] NCCL INFO Connected all rings
+ip-26-0-166-36:104434:104951 [5] NCCL INFO Channel 00/0 : 4[5] -> 6[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104434:104951 [5] NCCL INFO Channel 01/0 : 4[5] -> 6[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624947:625469 [4] NCCL INFO Connected all rings
+ip-26-0-166-214:624947:625469 [4] NCCL INFO Channel 00/0 : 2[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624947:625469 [4] NCCL INFO Channel 01/0 : 2[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624947:625469 [4] NCCL INFO Channel 00/0 : 4[4] -> 6[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624947:625469 [4] NCCL INFO Channel 01/0 : 4[4] -> 6[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762819:763351 [4] NCCL INFO Channel 00/0 : 4[4] -> 2[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762819:763351 [4] NCCL INFO Channel 01/0 : 4[4] -> 2[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624947:625469 [4] NCCL INFO Channel 00/0 : 0[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624947:625469 [4] NCCL INFO Channel 01/0 : 0[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624947:625469 [4] NCCL INFO Channel 00/0 : 4[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624947:625469 [4] NCCL INFO Channel 01/0 : 4[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624947:625469 [4] NCCL INFO Channel 00/0 : 6[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-46:702884:703443 [4] NCCL INFO Channel 02/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624947:625469 [4] NCCL INFO Channel 01/0 : 6[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-46:702884:703443 [4] NCCL INFO Channel 03/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624947:625469 [4] NCCL INFO Channel 00/0 : 4[4] -> 2[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624947:625469 [4] NCCL INFO Channel 01/0 : 4[4] -> 2[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762819:763351 [4] NCCL INFO Channel 00/0 : 3[4] -> 2[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762819:763351 [4] NCCL INFO Channel 01/0 : 3[4] -> 2[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762819:763351 [4] NCCL INFO Channel 00/0 : 2[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104433:104950 [4] NCCL INFO Channel 00/0 : 6[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762819:763351 [4] NCCL INFO Channel 01/0 : 2[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104433:104950 [4] NCCL INFO Channel 01/0 : 6[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762819:763351 [4] NCCL INFO Channel 02/0 : 2[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762819:763351 [4] NCCL INFO Channel 03/0 : 2[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104433:104950 [4] NCCL INFO Channel 00/0 : 7[4] -> 6[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624947:625469 [4] NCCL INFO Channel 02/0 : 5[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104433:104950 [4] NCCL INFO Channel 01/0 : 7[4] -> 6[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624947:625469 [4] NCCL INFO Channel 03/0 : 5[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104433:104950 [4] NCCL INFO Channel 00/0 : 6[4] -> 5[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104433:104950 [4] NCCL INFO Channel 01/0 : 6[4] -> 5[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104433:104950 [4] NCCL INFO Channel 02/0 : 6[4] -> 5[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104433:104950 [4] NCCL INFO Channel 03/0 : 6[4] -> 5[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104432:104944 [3] NCCL INFO Connected all trees
+ip-26-0-166-36:104432:104944 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-36:104432:104944 [3] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-166-36:104432:104944 [3] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-36:104431:104945 [2] NCCL INFO comm 0xacf8cd0 rank 6 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0x53075d22960e36e3 - Init COMPLETE
+ip-26-0-167-9:1039544:1040068 [7] NCCL INFO Connected all trees
+ip-26-0-167-9:1039544:1040068 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-167-9:1039544:1040068 [7] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-167-9:1039544:1040068 [7] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-214:624948:625470 [5] NCCL INFO Connected all rings
+ip-26-0-166-214:624948:625470 [5] NCCL INFO Channel 00/0 : 2[5] -> 4[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624948:625470 [5] NCCL INFO Channel 01/0 : 2[5] -> 4[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624948:625470 [5] NCCL INFO Channel 00/0 : 4[5] -> 6[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624948:625470 [5] NCCL INFO Channel 01/0 : 4[5] -> 6[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762820:763352 [5] NCCL INFO Channel 00/0 : 4[5] -> 2[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762820:763352 [5] NCCL INFO Channel 01/0 : 4[5] -> 2[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624948:625470 [5] NCCL INFO Channel 00/0 : 0[5] -> 4[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104434:104951 [5] NCCL INFO Channel 00/0 : 6[5] -> 4[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624948:625470 [5] NCCL INFO Channel 01/0 : 0[5] -> 4[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104434:104951 [5] NCCL INFO Channel 01/0 : 6[5] -> 4[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624948:625470 [5] NCCL INFO Channel 00/0 : 4[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127294:127812 [4] NCCL INFO Connected all rings
+ip-26-0-166-214:624948:625470 [5] NCCL INFO Channel 01/0 : 4[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127294:127812 [4] NCCL INFO Channel 02/0 : 1[4] -> 3[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127294:127812 [4] NCCL INFO Channel 03/0 : 1[4] -> 3[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127294:127812 [4] NCCL INFO Channel 02/0 : 3[4] -> 5[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624948:625470 [5] NCCL INFO Channel 00/0 : 6[5] -> 4[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127294:127812 [4] NCCL INFO Channel 03/0 : 3[4] -> 5[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-46:702885:703445 [5] NCCL INFO Channel 02/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624948:625470 [5] NCCL INFO Channel 01/0 : 6[5] -> 4[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-46:702885:703445 [5] NCCL INFO Channel 03/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624948:625470 [5] NCCL INFO Channel 00/0 : 4[5] -> 2[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624948:625470 [5] NCCL INFO Channel 01/0 : 4[5] -> 2[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9629:10147 [4] NCCL INFO Channel 02/0 : 3[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9629:10147 [4] NCCL INFO Channel 03/0 : 3[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127294:127812 [4] NCCL INFO Channel 02/0 : 7[4] -> 3[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104434:104951 [5] NCCL INFO Channel 00/0 : 7[5] -> 6[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897244:897763 [4] NCCL INFO Channel 02/0 : 5[4] -> 3[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127294:127812 [4] NCCL INFO Channel 03/0 : 7[4] -> 3[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897244:897763 [4] NCCL INFO Channel 03/0 : 5[4] -> 3[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104434:104951 [5] NCCL INFO Channel 01/0 : 7[5] -> 6[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624948:625470 [5] NCCL INFO Channel 02/0 : 5[5] -> 4[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127294:127812 [4] NCCL INFO Channel 02/0 : 3[4] -> 7[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762820:763352 [5] NCCL INFO Channel 00/0 : 3[5] -> 2[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104434:104951 [5] NCCL INFO Channel 00/0 : 6[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624948:625470 [5] NCCL INFO Channel 03/0 : 5[5] -> 4[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127294:127812 [4] NCCL INFO Channel 03/0 : 3[4] -> 7[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762820:763352 [5] NCCL INFO Channel 01/0 : 3[5] -> 2[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104434:104951 [5] NCCL INFO Channel 01/0 : 6[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762820:763352 [5] NCCL INFO Channel 00/0 : 2[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104434:104951 [5] NCCL INFO Channel 02/0 : 6[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762820:763352 [5] NCCL INFO Channel 01/0 : 2[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104434:104951 [5] NCCL INFO Channel 03/0 : 6[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127294:127812 [4] NCCL INFO Channel 02/0 : 5[4] -> 3[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9627:10142 [2] NCCL INFO Connected all trees
+ip-26-0-165-213:9627:10142 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-125:762820:763352 [5] NCCL INFO Channel 02/0 : 2[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127294:127812 [4] NCCL INFO Channel 03/0 : 5[4] -> 3[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9627:10142 [2] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-165-213:9627:10142 [2] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-167-9:1039541:1040074 [4] NCCL INFO Channel 00/0 : 7[4] -> 6[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762820:763352 [5] NCCL INFO Channel 03/0 : 2[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127294:127812 [4] NCCL INFO Channel 02/0 : 3[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-167-9:1039541:1040074 [4] NCCL INFO Channel 01/0 : 7[4] -> 6[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127294:127812 [4] NCCL INFO Channel 03/0 : 3[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897244:897763 [4] NCCL INFO Channel 00/0 : 6[4] -> 5[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897244:897763 [4] NCCL INFO Channel 01/0 : 6[4] -> 5[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127294:127812 [4] NCCL INFO Channel 00/0 : 3[4] -> 2[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897244:897763 [4] NCCL INFO Channel 02/0 : 6[4] -> 5[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9629:10147 [4] NCCL INFO Channel 00/0 : 2[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127294:127812 [4] NCCL INFO Channel 01/0 : 3[4] -> 2[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897244:897763 [4] NCCL INFO Channel 03/0 : 6[4] -> 5[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9629:10147 [4] NCCL INFO Channel 01/0 : 2[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897244:897763 [4] NCCL INFO Channel 02/0 : 5[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9629:10147 [4] NCCL INFO Channel 02/0 : 2[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897244:897763 [4] NCCL INFO Channel 03/0 : 5[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9629:10147 [4] NCCL INFO Channel 03/0 : 2[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9629:10147 [4] NCCL INFO Channel 02/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9629:10147 [4] NCCL INFO Channel 03/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127295:127813 [5] NCCL INFO Connected all rings
+ip-26-0-166-15:127295:127813 [5] NCCL INFO Channel 02/0 : 1[5] -> 3[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127295:127813 [5] NCCL INFO Channel 03/0 : 1[5] -> 3[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127295:127813 [5] NCCL INFO Channel 02/0 : 3[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127295:127813 [5] NCCL INFO Channel 03/0 : 3[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9630:10148 [5] NCCL INFO Channel 02/0 : 3[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9630:10148 [5] NCCL INFO Channel 03/0 : 3[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127295:127813 [5] NCCL INFO Channel 02/0 : 7[5] -> 3[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897245:897764 [5] NCCL INFO Channel 02/0 : 5[5] -> 3[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127295:127813 [5] NCCL INFO Channel 03/0 : 7[5] -> 3[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897245:897764 [5] NCCL INFO Channel 03/0 : 5[5] -> 3[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127295:127813 [5] NCCL INFO Channel 02/0 : 3[5] -> 7[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127295:127813 [5] NCCL INFO Channel 03/0 : 3[5] -> 7[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127295:127813 [5] NCCL INFO Channel 02/0 : 5[5] -> 3[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127295:127813 [5] NCCL INFO Channel 03/0 : 5[5] -> 3[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127295:127813 [5] NCCL INFO Channel 02/0 : 3[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127295:127813 [5] NCCL INFO Channel 03/0 : 3[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897245:897764 [5] NCCL INFO Channel 00/0 : 6[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127295:127813 [5] NCCL INFO Channel 00/0 : 3[5] -> 2[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-167-9:1039542:1040075 [5] NCCL INFO Channel 00/0 : 7[5] -> 6[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127295:127813 [5] NCCL INFO Channel 01/0 : 3[5] -> 2[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-167-9:1039542:1040075 [5] NCCL INFO Channel 01/0 : 7[5] -> 6[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897245:897764 [5] NCCL INFO Channel 01/0 : 6[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897245:897764 [5] NCCL INFO Channel 02/0 : 6[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897245:897764 [5] NCCL INFO Channel 03/0 : 6[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104432:104944 [3] NCCL INFO comm 0x9cf1610 rank 6 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0xa976fd02a70fed12 - Init COMPLETE
+ip-26-0-166-244:897245:897764 [5] NCCL INFO Channel 02/0 : 5[5] -> 4[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897245:897764 [5] NCCL INFO Channel 03/0 : 5[5] -> 4[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-167-9:1039544:1040068 [7] NCCL INFO comm 0xa16f4d0 rank 7 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0x30fc6b5205aced7 - Init COMPLETE
+ip-26-0-165-213:9630:10148 [5] NCCL INFO Channel 00/0 : 2[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9630:10148 [5] NCCL INFO Channel 01/0 : 2[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9630:10148 [5] NCCL INFO Channel 02/0 : 2[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9630:10148 [5] NCCL INFO Channel 03/0 : 2[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9630:10148 [5] NCCL INFO Channel 02/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9630:10148 [5] NCCL INFO Channel 03/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9627:10142 [2] NCCL INFO comm 0x9e14cb0 rank 1 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0x53075d22960e36e3 - Init COMPLETE
+ip-26-0-166-125:762817:763345 [2] NCCL INFO Connected all trees
+ip-26-0-166-125:762817:763345 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-125:762817:763345 [2] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-166-125:762817:763345 [2] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-162-46:702880:703439 [0] NCCL INFO Connected all trees
+ip-26-0-162-46:702880:703439 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-162-46:702880:703439 [0] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-162-46:702880:703439 [0] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-165-213:9628:10141 [3] NCCL INFO Connected all trees
+ip-26-0-165-213:9628:10141 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-213:9628:10141 [3] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-165-213:9628:10141 [3] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-125:762817:763345 [2] NCCL INFO comm 0xb8ed520 rank 2 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0x53075d22960e36e3 - Init COMPLETE
+ip-26-0-162-46:702880:703439 [0] NCCL INFO comm 0xafe5060 rank 0 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0x7ada0dd8e53058a4 - Init COMPLETE
+ip-26-0-162-46:702881:703441 [1] NCCL INFO Connected all trees
+ip-26-0-162-46:702881:703441 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-162-46:702881:703441 [1] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-162-46:702881:703441 [1] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-167-9:1039537:1040070 [0] NCCL INFO Connected all trees
+ip-26-0-167-9:1039537:1040070 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-167-9:1039537:1040070 [0] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-167-9:1039537:1040070 [0] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-165-213:9628:10141 [3] NCCL INFO comm 0xa7ff520 rank 1 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0xa976fd02a70fed12 - Init COMPLETE
+ip-26-0-167-9:1039538:1040071 [1] NCCL INFO Connected all trees
+ip-26-0-167-9:1039538:1040071 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-167-9:1039538:1040071 [1] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-167-9:1039538:1040071 [1] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-244:897242:897755 [2] NCCL INFO Connected all trees
+ip-26-0-166-244:897242:897755 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-244:897242:897755 [2] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-166-244:897242:897755 [2] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-214:624945:625463 [2] NCCL INFO Connected all trees
+ip-26-0-166-214:624945:625463 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-214:624945:625463 [2] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-166-214:624945:625463 [2] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-167-9:1039537:1040070 [0] NCCL INFO comm 0xa6f7f60 rank 7 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0x7ada0dd8e53058a4 - Init COMPLETE
+ip-26-0-162-46:702881:703441 [1] NCCL INFO comm 0xaaba1b0 rank 0 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0x2333452d75ea81bb - Init COMPLETE
+ip-26-0-167-9:1039538:1040071 [1] NCCL INFO comm 0x9b13060 rank 7 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0x2333452d75ea81bb - Init COMPLETE
+ip-26-0-162-46:702887:703436 [7] NCCL INFO Connected all trees
+ip-26-0-162-46:702887:703436 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-162-46:702887:703436 [7] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-162-46:702887:703436 [7] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-165-213:9632:10143 [7] NCCL INFO Connected all trees
+ip-26-0-165-213:9632:10143 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-213:9632:10143 [7] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-165-213:9632:10143 [7] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-244:897243:897754 [3] NCCL INFO Connected all trees
+ip-26-0-166-244:897243:897754 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-244:897243:897754 [3] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-166-244:897243:897754 [3] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-125:762818:763344 [3] NCCL INFO Connected all trees
+ip-26-0-166-125:762818:763344 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-125:762818:763344 [3] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-166-125:762818:763344 [3] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-244:897242:897755 [2] NCCL INFO comm 0x9bfa1f0 rank 5 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0x53075d22960e36e3 - Init COMPLETE
+ip-26-0-166-214:624945:625463 [2] NCCL INFO comm 0xaca7da0 rank 4 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0x53075d22960e36e3 - Init COMPLETE
+ip-26-0-166-214:624946:625462 [3] NCCL INFO Connected all trees
+ip-26-0-166-214:624946:625462 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-214:624946:625462 [3] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-166-214:624946:625462 [3] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-15:127292:127807 [2] NCCL INFO Connected all trees
+ip-26-0-166-15:127292:127807 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-15:127292:127807 [2] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-166-15:127292:127807 [2] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-165-213:9632:10143 [7] NCCL INFO comm 0xa1e76e0 rank 1 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0x30fc6b5205aced7 - Init COMPLETE
+ip-26-0-162-46:702887:703436 [7] NCCL INFO comm 0xa207c70 rank 0 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0x30fc6b5205aced7 - Init COMPLETE
+ip-26-0-166-244:897243:897754 [3] NCCL INFO comm 0x9f04870 rank 5 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0xa976fd02a70fed12 - Init COMPLETE
+ip-26-0-166-125:762818:763344 [3] NCCL INFO comm 0x9c2ef20 rank 2 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0xa976fd02a70fed12 - Init COMPLETE
+ip-26-0-166-36:104429:104948 [0] NCCL INFO Connected all trees
+ip-26-0-166-36:104429:104948 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-36:104429:104948 [0] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-166-36:104429:104948 [0] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-214:624946:625462 [3] NCCL INFO comm 0xb53f7f0 rank 4 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0xa976fd02a70fed12 - Init COMPLETE
+ip-26-0-166-15:127292:127807 [2] NCCL INFO comm 0xa6fa4e0 rank 3 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0x53075d22960e36e3 - Init COMPLETE
+ip-26-0-166-36:104429:104948 [0] NCCL INFO comm 0xb953f30 rank 6 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0x7ada0dd8e53058a4 - Init COMPLETE
+ip-26-0-167-9:1039542:1040075 [5] NCCL INFO Connected all trees
+ip-26-0-167-9:1039542:1040075 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-167-9:1039542:1040075 [5] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-167-9:1039542:1040075 [5] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-162-46:702886:703437 [6] NCCL INFO Connected all trees
+ip-26-0-162-46:702886:703437 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-162-46:702886:703437 [6] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-162-46:702886:703437 [6] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-15:127293:127806 [3] NCCL INFO Connected all trees
+ip-26-0-166-15:127293:127806 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-15:127293:127806 [3] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-166-15:127293:127806 [3] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-36:104436:104946 [7] NCCL INFO Connected all trees
+ip-26-0-166-36:104436:104946 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-36:104436:104946 [7] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-166-36:104436:104946 [7] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-167-9:1039542:1040075 [5] NCCL INFO comm 0x9927d60 rank 7 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0xbc5bf908bde42540 - Init COMPLETE
+ip-26-0-166-36:104430:104949 [1] NCCL INFO Connected all trees
+ip-26-0-166-36:104430:104949 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-36:104430:104949 [1] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-166-36:104430:104949 [1] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-162-46:702886:703437 [6] NCCL INFO comm 0xa6fc610 rank 0 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0xd4a12c6dc0dd1f26 - Init COMPLETE
+ip-26-0-166-125:762822:763346 [7] NCCL INFO Connected all trees
+ip-26-0-166-125:762822:763346 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-125:762822:763346 [7] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-166-125:762822:763346 [7] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-15:127293:127806 [3] NCCL INFO comm 0xb5fed70 rank 3 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0xa976fd02a70fed12 - Init COMPLETE
+ip-26-0-166-36:104436:104946 [7] NCCL INFO comm 0xac26b70 rank 6 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0x30fc6b5205aced7 - Init COMPLETE
+ip-26-0-162-46:702884:703443 [4] NCCL INFO Connected all trees
+ip-26-0-162-46:702884:703443 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-162-46:702884:703443 [4] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-162-46:702884:703443 [4] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-36:104430:104949 [1] NCCL INFO comm 0xb166740 rank 6 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0x2333452d75ea81bb - Init COMPLETE
+ip-26-0-166-125:762822:763346 [7] NCCL INFO comm 0xb4dad20 rank 2 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0x30fc6b5205aced7 - Init COMPLETE
+ip-26-0-165-213:9626:10146 [1] NCCL INFO Connected all trees
+ip-26-0-165-213:9626:10146 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-213:9626:10146 [1] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-165-213:9626:10146 [1] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-165-213:9625:10145 [0] NCCL INFO Connected all trees
+ip-26-0-165-213:9625:10145 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-213:9625:10145 [0] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-165-213:9625:10145 [0] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-162-46:702884:703443 [4] NCCL INFO comm 0xa5bdae0 rank 0 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0x481461516cf15ed6 - Init COMPLETE
+ip-26-0-162-46:702885:703445 [5] NCCL INFO Connected all trees
+ip-26-0-162-46:702885:703445 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-162-46:702885:703445 [5] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-162-46:702885:703445 [5] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-165-213:9626:10146 [1] NCCL INFO comm 0xa8b1f20 rank 1 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0x2333452d75ea81bb - Init COMPLETE
+ip-26-0-165-213:9625:10145 [0] NCCL INFO comm 0xa2d8ce0 rank 1 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0x7ada0dd8e53058a4 - Init COMPLETE
+ip-26-0-162-46:702885:703445 [5] NCCL INFO comm 0xa216a70 rank 0 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0xbc5bf908bde42540 - Init COMPLETE
+ip-26-0-166-244:897240:897761 [0] NCCL INFO Connected all trees
+ip-26-0-166-244:897240:897761 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-244:897240:897761 [0] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-166-244:897240:897761 [0] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-125:762815:763348 [0] NCCL INFO Connected all trees
+ip-26-0-166-125:762815:763348 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-125:762815:763348 [0] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-166-125:762815:763348 [0] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-125:762816:763349 [1] NCCL INFO Connected all trees
+ip-26-0-166-125:762816:763349 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-125:762816:763349 [1] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-166-125:762816:763349 [1] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-214:624943:625466 [0] NCCL INFO Connected all trees
+ip-26-0-166-214:624943:625466 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-214:624943:625466 [0] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-166-214:624943:625466 [0] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-244:897247:897759 [7] NCCL INFO Connected all trees
+ip-26-0-166-244:897247:897759 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-244:897247:897759 [7] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-166-244:897247:897759 [7] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-244:897241:897762 [1] NCCL INFO Connected all trees
+ip-26-0-166-244:897241:897762 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-244:897241:897762 [1] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-166-244:897241:897762 [1] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-244:897240:897761 [0] NCCL INFO comm 0xa810d00 rank 5 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0x7ada0dd8e53058a4 - Init COMPLETE
+ip-26-0-166-36:104433:104950 [4] NCCL INFO Connected all trees
+ip-26-0-166-36:104433:104950 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-36:104433:104950 [4] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-166-36:104433:104950 [4] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-125:762815:763348 [0] NCCL INFO comm 0x9ac36a0 rank 2 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0x7ada0dd8e53058a4 - Init COMPLETE
+ip-26-0-166-125:762816:763349 [1] NCCL INFO comm 0xaaf02d0 rank 2 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0x2333452d75ea81bb - Init COMPLETE
+ip-26-0-167-9:1039543:1040069 [6] NCCL INFO Connected all trees
+ip-26-0-167-9:1039543:1040069 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-167-9:1039543:1040069 [6] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-167-9:1039543:1040069 [6] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-214:624943:625466 [0] NCCL INFO comm 0xa9386c0 rank 4 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0x7ada0dd8e53058a4 - Init COMPLETE
+ip-26-0-166-244:897247:897759 [7] NCCL INFO comm 0xae9e340 rank 5 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0x30fc6b5205aced7 - Init COMPLETE
+ip-26-0-166-36:104434:104951 [5] NCCL INFO Connected all trees
+ip-26-0-166-36:104434:104951 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-36:104434:104951 [5] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-166-36:104434:104951 [5] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-244:897241:897762 [1] NCCL INFO comm 0xb6b0bc0 rank 5 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0x2333452d75ea81bb - Init COMPLETE
+ip-26-0-167-9:1039543:1040069 [6] NCCL INFO comm 0xaa38710 rank 7 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0xd4a12c6dc0dd1f26 - Init COMPLETE
+ip-26-0-166-36:104433:104950 [4] NCCL INFO comm 0xb740000 rank 6 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0x481461516cf15ed6 - Init COMPLETE
+ip-26-0-166-214:624950:625464 [7] NCCL INFO Connected all trees
+ip-26-0-166-214:624950:625464 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-214:624950:625464 [7] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-166-214:624950:625464 [7] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-15:127291:127811 [1] NCCL INFO Connected all trees
+ip-26-0-166-15:127291:127811 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-15:127291:127811 [1] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-166-15:127291:127811 [1] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-36:104434:104951 [5] NCCL INFO comm 0x9f5d010 rank 6 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0xbc5bf908bde42540 - Init COMPLETE
+ip-26-0-166-214:624944:625467 [1] NCCL INFO Connected all trees
+ip-26-0-166-214:624944:625467 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-214:624944:625467 [1] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-166-214:624944:625467 [1] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-167-9:1039541:1040074 [4] NCCL INFO Connected all trees
+ip-26-0-167-9:1039541:1040074 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-167-9:1039541:1040074 [4] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-167-9:1039541:1040074 [4] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-15:127290:127810 [0] NCCL INFO Connected all trees
+ip-26-0-166-15:127290:127810 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-15:127290:127810 [0] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-166-15:127290:127810 [0] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-15:127297:127808 [7] NCCL INFO Connected all trees
+ip-26-0-166-15:127297:127808 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-15:127297:127808 [7] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-166-15:127297:127808 [7] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-165-213:9630:10148 [5] NCCL INFO Connected all trees
+ip-26-0-165-213:9630:10148 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-213:9630:10148 [5] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-165-213:9630:10148 [5] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-214:624950:625464 [7] NCCL INFO comm 0xa292bc0 rank 4 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0x30fc6b5205aced7 - Init COMPLETE
+ip-26-0-166-15:127291:127811 [1] NCCL INFO comm 0xb70d770 rank 3 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0x2333452d75ea81bb - Init COMPLETE
+ip-26-0-166-214:624944:625467 [1] NCCL INFO comm 0x9a071c0 rank 4 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0x2333452d75ea81bb - Init COMPLETE
+ip-26-0-167-9:1039541:1040074 [4] NCCL INFO comm 0xaf69120 rank 7 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0x481461516cf15ed6 - Init COMPLETE
+ip-26-0-166-15:127290:127810 [0] NCCL INFO comm 0x9e92d70 rank 3 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0x7ada0dd8e53058a4 - Init COMPLETE
+ip-26-0-166-15:127297:127808 [7] NCCL INFO comm 0xb2b5da0 rank 3 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0x30fc6b5205aced7 - Init COMPLETE
+ip-26-0-165-213:9630:10148 [5] NCCL INFO comm 0x9c8ddd0 rank 1 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0xbc5bf908bde42540 - Init COMPLETE
+ip-26-0-166-36:104435:104947 [6] NCCL INFO Connected all trees
+ip-26-0-166-36:104435:104947 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-36:104435:104947 [6] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-166-36:104435:104947 [6] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-244:897245:897764 [5] NCCL INFO Connected all trees
+ip-26-0-166-244:897245:897764 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-244:897245:897764 [5] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-166-244:897245:897764 [5] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-165-213:9631:10144 [6] NCCL INFO Connected all trees
+ip-26-0-165-213:9631:10144 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-213:9631:10144 [6] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-165-213:9631:10144 [6] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-36:104435:104947 [6] NCCL INFO comm 0xa2b5600 rank 6 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0xd4a12c6dc0dd1f26 - Init COMPLETE
+ip-26-0-166-244:897245:897764 [5] NCCL INFO comm 0xafafd40 rank 5 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0xbc5bf908bde42540 - Init COMPLETE
+ip-26-0-165-213:9631:10144 [6] NCCL INFO comm 0xafbec50 rank 1 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0xd4a12c6dc0dd1f26 - Init COMPLETE
+ip-26-0-165-213:9629:10147 [4] NCCL INFO Connected all trees
+ip-26-0-165-213:9629:10147 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-213:9629:10147 [4] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-165-213:9629:10147 [4] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+01/07/2025 03:50:00 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]: Total number of parameters: 1.61G (3072.26MiB)
+01/07/2025 03:50:00 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]: Local number of parameters: 235M (448.02MiB)
+01/07/2025 03:50:00 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]: [After model building] Memory usage: 448.04MiB. Peak allocated: 5408.00MiB Peak reserved: 11554.00MiB
+01/07/2025 03:50:00 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]: No checkpoint path provided.
+ip-26-0-166-125:762821:763347 [6] NCCL INFO Connected all trees
+ip-26-0-166-125:762821:763347 [6] NCCL INFO NCCL_PROTO set by environment to simple
+01/07/2025 03:50:00 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]: Parametrizing model parameters using StandardParametrizator
+ip-26-0-166-125:762821:763347 [6] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-166-125:762821:763347 [6] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+01/07/2025 03:50:00 [INFO|DP=0|PP=0|TP=1|ip-26-0-162-46]: Local number of parameters: 235M (448.02MiB)
+01/07/2025 03:50:00 [INFO|DP=0|PP=0|TP=1|ip-26-0-162-46]: [After model building] Memory usage: 448.04MiB. Peak allocated: 5408.00MiB Peak reserved: 11554.00MiB
+01/07/2025 03:50:00 [INFO|DP=0|PP=4|TP=0|ip-26-0-166-214]: Local number of parameters: 67.1M (128.02MiB)
+01/07/2025 03:50:00 [INFO|DP=0|PP=6|TP=0|ip-26-0-166-36]: Local number of parameters: 33.6M (64.01MiB)
+01/07/2025 03:50:00 [INFO|DP=0|PP=2|TP=0|ip-26-0-166-125]: Local number of parameters: 101M (192.02MiB)
+01/07/2025 03:50:00 [INFO|DP=0|PP=1|TP=0|ip-26-0-165-213]: Local number of parameters: 101M (192.02MiB)
+01/07/2025 03:50:00 [INFO|DP=0|PP=3|TP=0|ip-26-0-166-15]: Local number of parameters: 101M (192.02MiB)
+01/07/2025 03:50:00 [INFO|DP=0|PP=7|TP=0|ip-26-0-167-9]: Local number of parameters: 134M (256.00MiB)
+01/07/2025 03:50:00 [INFO|DP=0|PP=4|TP=0|ip-26-0-166-214]: [After model building] Memory usage: 128.03MiB. Peak allocated: 5408.00MiB Peak reserved: 11554.00MiB
+01/07/2025 03:50:00 [INFO|DP=0|PP=6|TP=0|ip-26-0-166-36]: [After model building] Memory usage: 64.02MiB. Peak allocated: 5408.00MiB Peak reserved: 11554.00MiB
+01/07/2025 03:50:00 [INFO|DP=0|PP=2|TP=0|ip-26-0-166-125]: [After model building] Memory usage: 192.04MiB. Peak allocated: 5408.00MiB Peak reserved: 11554.00MiB
+01/07/2025 03:50:00 [INFO|DP=0|PP=1|TP=0|ip-26-0-165-213]: [After model building] Memory usage: 192.04MiB. Peak allocated: 5408.00MiB Peak reserved: 11554.00MiB
+01/07/2025 03:50:00 [INFO|DP=0|PP=3|TP=0|ip-26-0-166-15]: [After model building] Memory usage: 192.04MiB. Peak allocated: 5408.00MiB Peak reserved: 11554.00MiB
+01/07/2025 03:50:00 [INFO|DP=0|PP=7|TP=0|ip-26-0-167-9]: [After model building] Memory usage: 256.01MiB. Peak allocated: 5408.00MiB Peak reserved: 11554.00MiB
+01/07/2025 03:50:00 [INFO|DP=0|PP=2|TP=1|ip-26-0-166-125]: Local number of parameters: 101M (192.02MiB)
+01/07/2025 03:50:00 [INFO|DP=0|PP=4|TP=1|ip-26-0-166-214]: Local number of parameters: 67.1M (128.02MiB)
+01/07/2025 03:50:00 [INFO|DP=0|PP=1|TP=1|ip-26-0-165-213]: Local number of parameters: 101M (192.02MiB)
+01/07/2025 03:50:00 [INFO|DP=0|PP=6|TP=1|ip-26-0-166-36]: Local number of parameters: 33.6M (64.01MiB)
+01/07/2025 03:50:00 [INFO|DP=0|PP=7|TP=1|ip-26-0-167-9]: Local number of parameters: 134M (256.00MiB)
+01/07/2025 03:50:00 [INFO|DP=0|PP=3|TP=1|ip-26-0-166-15]: Local number of parameters: 101M (192.02MiB)
+01/07/2025 03:50:00 [INFO|DP=0|PP=5|TP=1|ip-26-0-166-244]: Local number of parameters: 33.6M (64.01MiB)
+01/07/2025 03:50:00 [INFO|DP=0|PP=4|TP=1|ip-26-0-166-214]: [After model building] Memory usage: 128.03MiB. Peak allocated: 5408.00MiB Peak reserved: 11554.00MiB
+01/07/2025 03:50:00 [INFO|DP=0|PP=2|TP=1|ip-26-0-166-125]: [After model building] Memory usage: 192.04MiB. Peak allocated: 5408.00MiB Peak reserved: 11554.00MiB
+01/07/2025 03:50:00 [INFO|DP=0|PP=1|TP=1|ip-26-0-165-213]: [After model building] Memory usage: 192.04MiB. Peak allocated: 5408.00MiB Peak reserved: 11554.00MiB
+01/07/2025 03:50:00 [INFO|DP=0|PP=6|TP=1|ip-26-0-166-36]: [After model building] Memory usage: 64.02MiB. Peak allocated: 5408.00MiB Peak reserved: 11554.00MiB
+01/07/2025 03:50:00 [INFO|DP=0|PP=7|TP=1|ip-26-0-167-9]: [After model building] Memory usage: 256.01MiB. Peak allocated: 5408.00MiB Peak reserved: 11554.00MiB
+01/07/2025 03:50:00 [INFO|DP=0|PP=3|TP=1|ip-26-0-166-15]: [After model building] Memory usage: 192.04MiB. Peak allocated: 5408.00MiB Peak reserved: 11554.00MiB
+01/07/2025 03:50:00 [INFO|DP=0|PP=5|TP=1|ip-26-0-166-244]: [After model building] Memory usage: 64.02MiB. Peak allocated: 5408.00MiB Peak reserved: 11554.00MiB
+ip-26-0-165-213:9629:10147 [4] NCCL INFO comm 0x9e0f0e0 rank 1 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0x481461516cf15ed6 - Init COMPLETE
+01/07/2025 03:50:00 [INFO|DP=0|PP=5|TP=0|ip-26-0-166-244]: Local number of parameters: 33.6M (64.01MiB)
+ip-26-0-166-244:897246:897760 [6] NCCL INFO Connected all trees
+ip-26-0-166-244:897246:897760 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-244:897246:897760 [6] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-166-244:897246:897760 [6] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+01/07/2025 03:50:00 [INFO|DP=0|PP=5|TP=0|ip-26-0-166-244]: [After model building] Memory usage: 64.02MiB. Peak allocated: 5408.00MiB Peak reserved: 11554.00MiB
+ip-26-0-166-125:762821:763347 [6] NCCL INFO comm 0x9d9fbe0 rank 2 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0xd4a12c6dc0dd1f26 - Init COMPLETE
+ip-26-0-166-214:624949:625465 [6] NCCL INFO Connected all trees
+ip-26-0-166-214:624949:625465 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-214:624949:625465 [6] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-166-214:624949:625465 [6] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-125:762819:763351 [4] NCCL INFO Connected all trees
+ip-26-0-166-125:762819:763351 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-125:762819:763351 [4] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-166-125:762819:763351 [4] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-244:897246:897760 [6] NCCL INFO comm 0xb91b050 rank 5 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0xd4a12c6dc0dd1f26 - Init COMPLETE
+ip-26-0-166-214:624949:625465 [6] NCCL INFO comm 0x9f6ead0 rank 4 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0xd4a12c6dc0dd1f26 - Init COMPLETE
+ip-26-0-166-125:762819:763351 [4] NCCL INFO comm 0xa8d9010 rank 2 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0x481461516cf15ed6 - Init COMPLETE
+ip-26-0-166-15:127296:127809 [6] NCCL INFO Connected all trees
+ip-26-0-166-15:127296:127809 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-15:127296:127809 [6] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-166-15:127296:127809 [6] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-244:897244:897763 [4] NCCL INFO Connected all trees
+ip-26-0-166-244:897244:897763 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-244:897244:897763 [4] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-166-244:897244:897763 [4] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-214:624947:625469 [4] NCCL INFO Connected all trees
+ip-26-0-166-214:624947:625469 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-214:624947:625469 [4] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-166-214:624947:625469 [4] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-15:127295:127813 [5] NCCL INFO Connected all trees
+ip-26-0-166-15:127295:127813 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-15:127295:127813 [5] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-166-15:127295:127813 [5] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-15:127296:127809 [6] NCCL INFO comm 0xab9e5d0 rank 3 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0xd4a12c6dc0dd1f26 - Init COMPLETE
+ip-26-0-166-15:127294:127812 [4] NCCL INFO Connected all trees
+ip-26-0-166-15:127294:127812 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-15:127294:127812 [4] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-166-15:127294:127812 [4] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-214:624948:625470 [5] NCCL INFO Connected all trees
+ip-26-0-166-214:624948:625470 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-214:624948:625470 [5] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-166-214:624948:625470 [5] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-125:762820:763352 [5] NCCL INFO Connected all trees
+ip-26-0-166-125:762820:763352 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-125:762820:763352 [5] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-166-125:762820:763352 [5] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-244:897244:897763 [4] NCCL INFO comm 0xa391a00 rank 5 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0x481461516cf15ed6 - Init COMPLETE
+ip-26-0-166-214:624947:625469 [4] NCCL INFO comm 0xb4dfd90 rank 4 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0x481461516cf15ed6 - Init COMPLETE
+ip-26-0-166-15:127295:127813 [5] NCCL INFO comm 0x9f8eb30 rank 3 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0xbc5bf908bde42540 - Init COMPLETE
+ip-26-0-166-15:127294:127812 [4] NCCL INFO comm 0xa49dc00 rank 3 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0x481461516cf15ed6 - Init COMPLETE
+ip-26-0-166-214:624948:625470 [5] NCCL INFO comm 0xb5510e0 rank 4 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0xbc5bf908bde42540 - Init COMPLETE
+ip-26-0-166-125:762820:763352 [5] NCCL INFO comm 0xa7f8d30 rank 2 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0xbc5bf908bde42540 - Init COMPLETE
+ip-26-0-162-46:702881:703465 [1] NCCL INFO Using network Libfabric
+ip-26-0-162-46:702884:703470 [4] NCCL INFO Using network Libfabric
+ip-26-0-162-46:702880:703466 [0] NCCL INFO Using network Libfabric
+ip-26-0-162-46:702887:703467 [7] NCCL INFO Using network Libfabric
+ip-26-0-162-46:702886:703469 [6] NCCL INFO Using network Libfabric
+ip-26-0-162-46:702883:703471 [3] NCCL INFO Using network Libfabric
+ip-26-0-162-46:702885:703468 [5] NCCL INFO Using network Libfabric
+ip-26-0-162-46:702882:703472 [2] NCCL INFO Using network Libfabric
+ip-26-0-162-46:702885:703468 [5] NCCL INFO comm 0xac5faf0 rank 2 nranks 4 cudaDev 5 nvmlDev 5 busId a8000 commId 0x4908ed83bce0c92e - Init START
+ip-26-0-162-46:702883:703471 [3] NCCL INFO comm 0xa67ba70 rank 1 nranks 4 cudaDev 3 nvmlDev 3 busId 86000 commId 0x4908ed83bce0c92e - Init START
+ip-26-0-162-46:702887:703467 [7] NCCL INFO comm 0xac500c0 rank 3 nranks 4 cudaDev 7 nvmlDev 7 busId ca000 commId 0x4908ed83bce0c92e - Init START
+ip-26-0-162-46:702881:703465 [1] NCCL INFO comm 0xb5043f0 rank 0 nranks 4 cudaDev 1 nvmlDev 1 busId 64000 commId 0x4908ed83bce0c92e - Init START
+NCCL version 2.18.6+cuda12.1
+NCCL version 2.18.6+cuda12.1
+NCCL version 2.18.6+cuda12.1
+NCCL version 2.18.6+cuda12.1
+NCCL version 2.18.6+cuda12.1
+ip-26-0-167-9:1039537:1040095 [0] NCCL INFO Using network Libfabric
+NCCL version 2.18.6+cuda12.1
+NCCL version 2.18.6+cuda12.1
+ip-26-0-167-9:1039541:1040097 [4] NCCL INFO Using network Libfabric
+ip-26-0-167-9:1039539:1040100 [2] NCCL INFO Using network Libfabric
+ip-26-0-166-244:897241:897784 [1] NCCL INFO Using network Libfabric
+ip-26-0-167-9:1039543:1040098 [6] NCCL INFO Using network Libfabric
+ip-26-0-167-9:1039538:1040099 [1] NCCL INFO Using network Libfabric
+ip-26-0-165-213:9625:10169 [0] NCCL INFO Using network Libfabric
+ip-26-0-166-244:897240:897786 [0] NCCL INFO Using network Libfabric
+ip-26-0-167-9:1039544:1040101 [7] NCCL INFO Using network Libfabric
+ip-26-0-166-244:897247:897788 [7] NCCL INFO Using network Libfabric
+ip-26-0-167-9:1039542:1040102 [5] NCCL INFO Using network Libfabric
+ip-26-0-167-9:1039540:1040103 [3] NCCL INFO Using network Libfabric
+ip-26-0-166-36:104429:104973 [0] NCCL INFO Using network Libfabric
+ip-26-0-166-244:897245:897787 [5] NCCL INFO Using network Libfabric
+ip-26-0-166-214:624944:625493 [1] NCCL INFO Using network Libfabric
+ip-26-0-165-213:9629:10171 [4] NCCL INFO Using network Libfabric
+ip-26-0-165-213:9631:10172 [6] NCCL INFO Using network Libfabric
+ip-26-0-166-36:104430:104974 [1] NCCL INFO Using network Libfabric
+ip-26-0-166-244:897243:897790 [3] NCCL INFO Using network Libfabric
+ip-26-0-165-213:9626:10173 [1] NCCL INFO Using network Libfabric
+ip-26-0-166-214:624943:625494 [0] NCCL INFO Using network Libfabric
+ip-26-0-166-15:127290:127833 [0] NCCL INFO Using network Libfabric
+ip-26-0-166-244:897244:897789 [4] NCCL INFO Using network Libfabric
+ip-26-0-166-244:897246:897791 [6] NCCL INFO Using network Libfabric
+ip-26-0-166-36:104435:104975 [6] NCCL INFO Using network Libfabric
+ip-26-0-166-125:762815:763371 [0] NCCL INFO Using network Libfabric
+ip-26-0-166-244:897242:897792 [2] NCCL INFO Using network Libfabric
+ip-26-0-166-214:624945:625495 [2] NCCL INFO Using network Libfabric
+ip-26-0-166-214:624950:625499 [7] NCCL INFO Using network Libfabric
+ip-26-0-166-214:624947:625497 [4] NCCL INFO Using network Libfabric
+ip-26-0-167-9:1039543:1040098 [6] NCCL INFO comm 0xb657270 rank 3 nranks 4 cudaDev 6 nvmlDev 6 busId b9000 commId 0xd981cf5f876ca6ba - Init START
+ip-26-0-167-9:1039541:1040097 [4] NCCL INFO comm 0xbb87bb0 rank 2 nranks 4 cudaDev 4 nvmlDev 4 busId 97000 commId 0xd981cf5f876ca6ba - Init START
+ip-26-0-167-9:1039539:1040100 [2] NCCL INFO comm 0xa809100 rank 1 nranks 4 cudaDev 2 nvmlDev 2 busId 75000 commId 0xd981cf5f876ca6ba - Init START
+ip-26-0-167-9:1039537:1040095 [0] NCCL INFO comm 0xb3162d0 rank 0 nranks 4 cudaDev 0 nvmlDev 0 busId 53000 commId 0xd981cf5f876ca6ba - Init START
+ip-26-0-166-36:104433:104976 [4] NCCL INFO Using network Libfabric
+ip-26-0-166-36:104431:104979 [2] NCCL INFO Using network Libfabric
+ip-26-0-166-214:624946:625498 [3] NCCL INFO Using network Libfabric
+ip-26-0-166-15:127294:127835 [4] NCCL INFO Using network Libfabric
+ip-26-0-166-15:127296:127834 [6] NCCL INFO Using network Libfabric
+ip-26-0-166-15:127292:127836 [2] NCCL INFO Using network Libfabric
+ip-26-0-166-214:624949:625496 [6] NCCL INFO Using network Libfabric
+ip-26-0-165-213:9630:10175 [5] NCCL INFO Using network Libfabric
+ip-26-0-166-125:762819:763372 [4] NCCL INFO Using network Libfabric
+ip-26-0-166-36:104436:104978 [7] NCCL INFO Using network Libfabric
+ip-26-0-166-36:104432:104980 [3] NCCL INFO Using network Libfabric
+ip-26-0-167-9:1039543:1040098 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-167-9:1039541:1040097 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-165-213:9628:10177 [3] NCCL INFO Using network Libfabric
+ip-26-0-166-36:104434:104977 [5] NCCL INFO Using network Libfabric
+ip-26-0-166-125:762821:763373 [6] NCCL INFO Using network Libfabric
+ip-26-0-166-214:624948:625500 [5] NCCL INFO Using network Libfabric
+ip-26-0-167-9:1039539:1040100 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-167-9:1039537:1040095 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-165-213:9627:10174 [2] NCCL INFO Using network Libfabric
+ip-26-0-167-9:1039540:1040103 [3] NCCL INFO comm 0xae831b0 rank 1 nranks 4 cudaDev 3 nvmlDev 3 busId 86000 commId 0x689c160a9f6f8f23 - Init START
+ip-26-0-167-9:1039542:1040102 [5] NCCL INFO comm 0xa547f30 rank 2 nranks 4 cudaDev 5 nvmlDev 5 busId a8000 commId 0x689c160a9f6f8f23 - Init START
+ip-26-0-167-9:1039538:1040099 [1] NCCL INFO comm 0xa731d00 rank 0 nranks 4 cudaDev 1 nvmlDev 1 busId 64000 commId 0x689c160a9f6f8f23 - Init START
+ip-26-0-167-9:1039544:1040101 [7] NCCL INFO comm 0xad8eb40 rank 3 nranks 4 cudaDev 7 nvmlDev 7 busId ca000 commId 0x689c160a9f6f8f23 - Init START
+ip-26-0-166-15:127291:127838 [1] NCCL INFO Using network Libfabric
+ip-26-0-166-125:762816:763376 [1] NCCL INFO Using network Libfabric
+ip-26-0-167-9:1039540:1040103 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-167-9:1039542:1040102 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-125:762817:763374 [2] NCCL INFO Using network Libfabric
+ip-26-0-166-244:897247:897788 [7] NCCL INFO comm 0xbacf380 rank 3 nranks 4 cudaDev 7 nvmlDev 7 busId ca000 commId 0x3fa48ab7a0d2f235 - Init START
+ip-26-0-166-244:897243:897790 [3] NCCL INFO comm 0xab319e0 rank 1 nranks 4 cudaDev 3 nvmlDev 3 busId 86000 commId 0x3fa48ab7a0d2f235 - Init START
+ip-26-0-166-244:897245:897787 [5] NCCL INFO comm 0xbbdd3c0 rank 2 nranks 4 cudaDev 5 nvmlDev 5 busId a8000 commId 0x3fa48ab7a0d2f235 - Init START
+ip-26-0-166-244:897241:897784 [1] NCCL INFO comm 0xc2e1ee0 rank 0 nranks 4 cudaDev 1 nvmlDev 1 busId 64000 commId 0x3fa48ab7a0d2f235 - Init START
+ip-26-0-166-15:127295:127839 [5] NCCL INFO Using network Libfabric
+ip-26-0-166-15:127297:127840 [7] NCCL INFO Using network Libfabric
+ip-26-0-167-9:1039544:1040101 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-167-9:1039538:1040099 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-15:127293:127841 [3] NCCL INFO Using network Libfabric
+ip-26-0-166-125:762822:763377 [7] NCCL INFO Using network Libfabric
+ip-26-0-166-244:897247:897788 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-165-213:9632:10176 [7] NCCL INFO Using network Libfabric
+ip-26-0-166-125:762818:763378 [3] NCCL INFO Using network Libfabric
+ip-26-0-166-125:762820:763379 [5] NCCL INFO Using network Libfabric
+ip-26-0-166-36:104431:104979 [2] NCCL INFO comm 0xb9278a0 rank 1 nranks 4 cudaDev 2 nvmlDev 2 busId 75000 commId 0xb03b08d1ced4579 - Init START
+ip-26-0-166-36:104429:104973 [0] NCCL INFO comm 0xc5849c0 rank 0 nranks 4 cudaDev 0 nvmlDev 0 busId 53000 commId 0xb03b08d1ced4579 - Init START
+ip-26-0-166-36:104433:104976 [4] NCCL INFO comm 0xc36e970 rank 2 nranks 4 cudaDev 4 nvmlDev 4 busId 97000 commId 0xb03b08d1ced4579 - Init START
+ip-26-0-166-36:104435:104975 [6] NCCL INFO comm 0xaee4280 rank 3 nranks 4 cudaDev 6 nvmlDev 6 busId b9000 commId 0xb03b08d1ced4579 - Init START
+ip-26-0-166-244:897245:897787 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-244:897243:897790 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-244:897241:897784 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-214:624943:625494 [0] NCCL INFO comm 0xb569b60 rank 0 nranks 4 cudaDev 0 nvmlDev 0 busId 53000 commId 0x63ed815d5d2606 - Init START
+ip-26-0-166-214:624949:625496 [6] NCCL INFO comm 0xaba0260 rank 3 nranks 4 cudaDev 6 nvmlDev 6 busId b9000 commId 0x63ed815d5d2606 - Init START
+ip-26-0-166-214:624945:625495 [2] NCCL INFO comm 0xb8d6660 rank 1 nranks 4 cudaDev 2 nvmlDev 2 busId 75000 commId 0x63ed815d5d2606 - Init START
+ip-26-0-166-214:624947:625497 [4] NCCL INFO comm 0xc10da80 rank 2 nranks 4 cudaDev 4 nvmlDev 4 busId 97000 commId 0x63ed815d5d2606 - Init START
+ip-26-0-166-15:127296:127834 [6] NCCL INFO comm 0xb7cefc0 rank 3 nranks 4 cudaDev 6 nvmlDev 6 busId b9000 commId 0x5b4f6f555ef9c3fa - Init START
+ip-26-0-166-15:127294:127835 [4] NCCL INFO comm 0xb0cc4f0 rank 2 nranks 4 cudaDev 4 nvmlDev 4 busId 97000 commId 0x5b4f6f555ef9c3fa - Init START
+ip-26-0-166-15:127292:127836 [2] NCCL INFO comm 0xb32c4c0 rank 1 nranks 4 cudaDev 2 nvmlDev 2 busId 75000 commId 0x5b4f6f555ef9c3fa - Init START
+ip-26-0-166-15:127290:127833 [0] NCCL INFO comm 0xaac4d80 rank 0 nranks 4 cudaDev 0 nvmlDev 0 busId 53000 commId 0x5b4f6f555ef9c3fa - Init START
+ip-26-0-166-244:897246:897791 [6] NCCL INFO comm 0xc54c400 rank 3 nranks 4 cudaDev 6 nvmlDev 6 busId b9000 commId 0xd3faa027769f7c66 - Init START
+ip-26-0-166-244:897244:897789 [4] NCCL INFO comm 0xafc12f0 rank 2 nranks 4 cudaDev 4 nvmlDev 4 busId 97000 commId 0xd3faa027769f7c66 - Init START
+ip-26-0-166-244:897240:897786 [0] NCCL INFO comm 0xb43e500 rank 0 nranks 4 cudaDev 0 nvmlDev 0 busId 53000 commId 0xd3faa027769f7c66 - Init START
+ip-26-0-166-244:897242:897792 [2] NCCL INFO comm 0xa82b600 rank 1 nranks 4 cudaDev 2 nvmlDev 2 busId 75000 commId 0xd3faa027769f7c66 - Init START
+ip-26-0-166-214:624944:625493 [1] NCCL INFO comm 0xa637bf0 rank 0 nranks 4 cudaDev 1 nvmlDev 1 busId 64000 commId 0x6ca81a321acfffee - Init START
+ip-26-0-166-214:624950:625499 [7] NCCL INFO comm 0xaec1440 rank 3 nranks 4 cudaDev 7 nvmlDev 7 busId ca000 commId 0x6ca81a321acfffee - Init START
+ip-26-0-166-36:104436:104978 [7] NCCL INFO comm 0xb855c60 rank 3 nranks 4 cudaDev 7 nvmlDev 7 busId ca000 commId 0x9a7cb67b8659c59e - Init START
+ip-26-0-166-36:104434:104977 [5] NCCL INFO comm 0xab8bb00 rank 2 nranks 4 cudaDev 5 nvmlDev 5 busId a8000 commId 0x9a7cb67b8659c59e - Init START
+ip-26-0-166-36:104430:104974 [1] NCCL INFO comm 0xbd993c0 rank 0 nranks 4 cudaDev 1 nvmlDev 1 busId 64000 commId 0x9a7cb67b8659c59e - Init START
+ip-26-0-166-36:104432:104980 [3] NCCL INFO comm 0xa91fb80 rank 1 nranks 4 cudaDev 3 nvmlDev 3 busId 86000 commId 0x9a7cb67b8659c59e - Init START
+ip-26-0-166-214:624946:625498 [3] NCCL INFO comm 0xc16d320 rank 1 nranks 4 cudaDev 3 nvmlDev 3 busId 86000 commId 0x6ca81a321acfffee - Init START
+ip-26-0-166-214:624948:625500 [5] NCCL INFO comm 0xc1820e0 rank 2 nranks 4 cudaDev 5 nvmlDev 5 busId a8000 commId 0x6ca81a321acfffee - Init START
+ip-26-0-166-15:127296:127834 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-36:104433:104976 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-36:104435:104975 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-36:104429:104973 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-36:104431:104979 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-214:624947:625497 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-214:624949:625496 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-15:127292:127836 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-15:127294:127835 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-15:127290:127833 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-244:897244:897789 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-244:897242:897792 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-244:897246:897791 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-244:897240:897786 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-165-213:9631:10172 [6] NCCL INFO comm 0xbbecca0 rank 3 nranks 4 cudaDev 6 nvmlDev 6 busId b9000 commId 0x2a4e364c12c0a65b - Init START
+ip-26-0-165-213:9629:10171 [4] NCCL INFO comm 0xaa3ca00 rank 2 nranks 4 cudaDev 4 nvmlDev 4 busId 97000 commId 0x2a4e364c12c0a65b - Init START
+ip-26-0-165-213:9627:10174 [2] NCCL INFO comm 0xaa425a0 rank 1 nranks 4 cudaDev 2 nvmlDev 2 busId 75000 commId 0x2a4e364c12c0a65b - Init START
+ip-26-0-165-213:9625:10169 [0] NCCL INFO comm 0xaf094b0 rank 0 nranks 4 cudaDev 0 nvmlDev 0 busId 53000 commId 0x2a4e364c12c0a65b - Init START
+ip-26-0-166-36:104430:104974 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-36:104436:104978 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-36:104434:104977 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-36:104432:104980 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-214:624943:625494 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-214:624945:625495 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-214:624950:625499 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-214:624948:625500 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-214:624944:625493 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-214:624946:625498 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-125:762819:763372 [4] NCCL INFO comm 0xb508540 rank 2 nranks 4 cudaDev 4 nvmlDev 4 busId 97000 commId 0x16fbe3f8cceea039 - Init START
+ip-26-0-166-125:762821:763373 [6] NCCL INFO comm 0xa9cdcc0 rank 3 nranks 4 cudaDev 6 nvmlDev 6 busId b9000 commId 0x16fbe3f8cceea039 - Init START
+ip-26-0-165-213:9625:10169 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-165-213:9629:10171 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-165-213:9631:10172 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-165-213:9627:10174 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-125:762817:763374 [2] NCCL INFO comm 0xc51e000 rank 1 nranks 4 cudaDev 2 nvmlDev 2 busId 75000 commId 0x16fbe3f8cceea039 - Init START
+ip-26-0-166-125:762815:763371 [0] NCCL INFO comm 0xa6f1020 rank 0 nranks 4 cudaDev 0 nvmlDev 0 busId 53000 commId 0x16fbe3f8cceea039 - Init START
+ip-26-0-166-15:127295:127839 [5] NCCL INFO comm 0xabbd310 rank 2 nranks 4 cudaDev 5 nvmlDev 5 busId a8000 commId 0xaabd8ffb497aa15 - Init START
+ip-26-0-166-125:762819:763372 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-165-213:9632:10176 [7] NCCL INFO comm 0xae15940 rank 3 nranks 4 cudaDev 7 nvmlDev 7 busId ca000 commId 0x394a264b186d9de9 - Init START
+ip-26-0-165-213:9630:10175 [5] NCCL INFO comm 0xa8b9ce0 rank 2 nranks 4 cudaDev 5 nvmlDev 5 busId a8000 commId 0x394a264b186d9de9 - Init START
+ip-26-0-166-15:127297:127840 [7] NCCL INFO comm 0xbee6820 rank 3 nranks 4 cudaDev 7 nvmlDev 7 busId ca000 commId 0xaabd8ffb497aa15 - Init START
+ip-26-0-166-15:127293:127841 [3] NCCL INFO comm 0xc22d550 rank 1 nranks 4 cudaDev 3 nvmlDev 3 busId 86000 commId 0xaabd8ffb497aa15 - Init START
+ip-26-0-166-15:127291:127838 [1] NCCL INFO comm 0xc33fd70 rank 0 nranks 4 cudaDev 1 nvmlDev 1 busId 64000 commId 0xaabd8ffb497aa15 - Init START
+ip-26-0-166-125:762817:763374 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-125:762815:763371 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-165-213:9626:10173 [1] NCCL INFO comm 0xb4e0900 rank 0 nranks 4 cudaDev 1 nvmlDev 1 busId 64000 commId 0x394a264b186d9de9 - Init START
+ip-26-0-165-213:9628:10177 [3] NCCL INFO comm 0xb42cd50 rank 1 nranks 4 cudaDev 3 nvmlDev 3 busId 86000 commId 0x394a264b186d9de9 - Init START
+ip-26-0-166-125:762821:763373 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-125:762816:763376 [1] NCCL INFO comm 0xb71fe00 rank 0 nranks 4 cudaDev 1 nvmlDev 1 busId 64000 commId 0x2244107d9b6a67ed - Init START
+ip-26-0-166-125:762822:763377 [7] NCCL INFO comm 0xc109000 rank 3 nranks 4 cudaDev 7 nvmlDev 7 busId ca000 commId 0x2244107d9b6a67ed - Init START
+ip-26-0-166-125:762818:763378 [3] NCCL INFO comm 0xa85c340 rank 1 nranks 4 cudaDev 3 nvmlDev 3 busId 86000 commId 0x2244107d9b6a67ed - Init START
+ip-26-0-166-125:762820:763379 [5] NCCL INFO comm 0xb4271a0 rank 2 nranks 4 cudaDev 5 nvmlDev 5 busId a8000 commId 0x2244107d9b6a67ed - Init START
+ip-26-0-166-15:127293:127841 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-15:127295:127839 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-15:127291:127838 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-165-213:9632:10176 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-165-213:9628:10177 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-165-213:9630:10175 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-165-213:9626:10173 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-15:127297:127840 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-125:762816:763376 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-125:762822:763377 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-125:762820:763379 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-125:762818:763378 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-162-46:702881:703465 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-162-46:702887:703467 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-162-46:702885:703468 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-162-46:702883:703471 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-162-46:702880:703466 [0] NCCL INFO comm 0xafe7d10 rank 0 nranks 4 cudaDev 0 nvmlDev 0 busId 53000 commId 0x358519269da0b31b - Init START
+ip-26-0-162-46:702882:703472 [2] NCCL INFO comm 0xb8bbb80 rank 1 nranks 4 cudaDev 2 nvmlDev 2 busId 75000 commId 0x358519269da0b31b - Init START
+ip-26-0-162-46:702884:703470 [4] NCCL INFO comm 0xa5c1b20 rank 2 nranks 4 cudaDev 4 nvmlDev 4 busId 97000 commId 0x358519269da0b31b - Init START
+ip-26-0-162-46:702886:703469 [6] NCCL INFO comm 0xb144580 rank 3 nranks 4 cudaDev 6 nvmlDev 6 busId b9000 commId 0x358519269da0b31b - Init START
+ip-26-0-162-46:702880:703466 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/349
+ip-26-0-162-46:702884:703470 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-162-46:702886:703469 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-162-46:702882:703472 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-167-9:1039544:1040101 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-167-9:1039544:1040101 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-166-15:127297:127840 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-166-15:127297:127840 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-166-15:127293:127841 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-166-15:127293:127841 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-166-244:897246:897791 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-166-244:897246:897791 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-167-9:1039538:1040099 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-167-9:1039538:1040099 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-162-46:702886:703469 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-162-46:702886:703469 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-167-9:1039543:1040098 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-167-9:1039543:1040098 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-167-9:1039540:1040103 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-167-9:1039540:1040103 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-166-36:104432:104980 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-166-36:104432:104980 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-162-46:702881:703465 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-162-46:702881:703465 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-167-9:1039541:1040097 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-167-9:1039541:1040097 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-167-9:1039542:1040102 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-167-9:1039542:1040102 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-165-213:9632:10176 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-165-213:9632:10176 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-165-213:9625:10169 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-165-213:9625:10169 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-167-9:1039537:1040095 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-167-9:1039537:1040095 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-167-9:1039538:1040099 [1] NCCL INFO Channel 00/24 :    0   1   2   3
+ip-26-0-167-9:1039542:1040102 [5] NCCL INFO Trees [0] 3/-1/-1->2->1 [1] 3/-1/-1->2->1 [2] 3/-1/-1->2->1 [3] 3/-1/-1->2->1 [4] 3/-1/-1->2->1 [5] 3/-1/-1->2->1 [6] 3/-1/-1->2->1 [7] 3/-1/-1->2->1 [8] 3/-1/-1->2->1 [9] 3/-1/-1->2->1 [10] 3/-1/-1->2->1 [11] 3/-1/-1->2->1 [12] 3/-1/-1->2->1 [13] 3/-1/-1->2->1 [14] 3/-1/-1->2->1 [15] 3/-1/-1->2->1 [16] 3/-1/-1->2->1 [17] 3/-1/-1->2->1 [18] 3/-1/-1->2->1 [19] 3/-1/-1->2->1 [20] 3/-1/-1->2->1 [21] 3/-1/-1->2->1 [22] 3/-1/-1->2->1 [23] 3/-1/-1->2->1
+ip-26-0-167-9:1039538:1040099 [1] NCCL INFO Channel 01/24 :    0   1   2   3
+ip-26-0-167-9:1039542:1040102 [5] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-167-9:1039544:1040101 [7] NCCL INFO Trees [0] -1/-1/-1->3->2 [1] -1/-1/-1->3->2 [2] -1/-1/-1->3->2 [3] -1/-1/-1->3->2 [4] -1/-1/-1->3->2 [5] -1/-1/-1->3->2 [6] -1/-1/-1->3->2 [7] -1/-1/-1->3->2 [8] -1/-1/-1->3->2 [9] -1/-1/-1->3->2 [10] -1/-1/-1->3->2 [11] -1/-1/-1->3->2 [12] -1/-1/-1->3->2 [13] -1/-1/-1->3->2 [14] -1/-1/-1->3->2 [15] -1/-1/-1->3->2 [16] -1/-1/-1->3->2 [17] -1/-1/-1->3->2 [18] -1/-1/-1->3->2 [19] -1/-1/-1->3->2 [20] -1/-1/-1->3->2 [21] -1/-1/-1->3->2 [22] -1/-1/-1->3->2 [23] -1/-1/-1->3->2
+ip-26-0-167-9:1039538:1040099 [1] NCCL INFO Channel 02/24 :    0   1   2   3
+ip-26-0-167-9:1039540:1040103 [3] NCCL INFO Trees [0] 2/-1/-1->1->0 [1] 2/-1/-1->1->0 [2] 2/-1/-1->1->0 [3] 2/-1/-1->1->0 [4] 2/-1/-1->1->0 [5] 2/-1/-1->1->0 [6] 2/-1/-1->1->0 [7] 2/-1/-1->1->0 [8] 2/-1/-1->1->0 [9] 2/-1/-1->1->0 [10] 2/-1/-1->1->0 [11] 2/-1/-1->1->0 [12] 2/-1/-1->1->0 [13] 2/-1/-1->1->0 [14] 2/-1/-1->1->0 [15] 2/-1/-1->1->0 [16] 2/-1/-1->1->0 [17] 2/-1/-1->1->0 [18] 2/-1/-1->1->0 [19] 2/-1/-1->1->0 [20] 2/-1/-1->1->0 [21] 2/-1/-1->1->0 [22] 2/-1/-1->1->0 [23] 2/-1/-1->1->0
+ip-26-0-167-9:1039544:1040101 [7] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-167-9:1039538:1040099 [1] NCCL INFO Channel 03/24 :    0   1   2   3
+ip-26-0-167-9:1039540:1040103 [3] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-167-9:1039538:1040099 [1] NCCL INFO Channel 04/24 :    0   1   2   3
+ip-26-0-167-9:1039538:1040099 [1] NCCL INFO Channel 05/24 :    0   1   2   3
+ip-26-0-167-9:1039538:1040099 [1] NCCL INFO Channel 06/24 :    0   1   2   3
+ip-26-0-167-9:1039538:1040099 [1] NCCL INFO Channel 07/24 :    0   1   2   3
+ip-26-0-167-9:1039538:1040099 [1] NCCL INFO Channel 08/24 :    0   1   2   3
+ip-26-0-167-9:1039538:1040099 [1] NCCL INFO Channel 09/24 :    0   1   2   3
+ip-26-0-167-9:1039538:1040099 [1] NCCL INFO Channel 10/24 :    0   1   2   3
+ip-26-0-167-9:1039538:1040099 [1] NCCL INFO Channel 11/24 :    0   1   2   3
+ip-26-0-167-9:1039538:1040099 [1] NCCL INFO Channel 12/24 :    0   1   2   3
+ip-26-0-167-9:1039538:1040099 [1] NCCL INFO Channel 13/24 :    0   1   2   3
+ip-26-0-167-9:1039538:1040099 [1] NCCL INFO Channel 14/24 :    0   1   2   3
+ip-26-0-167-9:1039538:1040099 [1] NCCL INFO Channel 15/24 :    0   1   2   3
+ip-26-0-167-9:1039538:1040099 [1] NCCL INFO Channel 16/24 :    0   1   2   3
+ip-26-0-167-9:1039538:1040099 [1] NCCL INFO Channel 17/24 :    0   1   2   3
+ip-26-0-167-9:1039538:1040099 [1] NCCL INFO Channel 18/24 :    0   1   2   3
+ip-26-0-167-9:1039538:1040099 [1] NCCL INFO Channel 19/24 :    0   1   2   3
+ip-26-0-167-9:1039538:1040099 [1] NCCL INFO Channel 20/24 :    0   1   2   3
+ip-26-0-167-9:1039538:1040099 [1] NCCL INFO Channel 21/24 :    0   1   2   3
+ip-26-0-167-9:1039538:1040099 [1] NCCL INFO Channel 22/24 :    0   1   2   3
+ip-26-0-167-9:1039538:1040099 [1] NCCL INFO Channel 23/24 :    0   1   2   3
+ip-26-0-167-9:1039538:1040099 [1] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] 1/-1/-1->0->-1 [7] 1/-1/-1->0->-1 [8] 1/-1/-1->0->-1 [9] 1/-1/-1->0->-1 [10] 1/-1/-1->0->-1 [11] 1/-1/-1->0->-1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] 1/-1/-1->0->-1 [19] 1/-1/-1->0->-1 [20] 1/-1/-1->0->-1 [21] 1/-1/-1->0->-1 [22] 1/-1/-1->0->-1 [23] 1/-1/-1->0->-1
+ip-26-0-167-9:1039538:1040099 [1] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-165-213:9628:10177 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-165-213:9628:10177 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-166-15:127295:127839 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-166-15:127295:127839 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-167-9:1039539:1040100 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-167-9:1039539:1040100 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-166-15:127290:127833 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-166-15:127290:127833 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-166-244:897244:897789 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-166-244:897244:897789 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-165-213:9631:10172 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-165-213:9631:10172 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-166-15:127292:127836 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-166-15:127292:127836 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-165-213:9630:10175 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-165-213:9630:10175 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-167-9:1039540:1040103 [3] NCCL INFO Channel 00/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-167-9:1039544:1040101 [7] NCCL INFO Channel 00/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-167-9:1039538:1040099 [1] NCCL INFO Channel 00/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-167-9:1039544:1040101 [7] NCCL INFO Channel 01/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-167-9:1039540:1040103 [3] NCCL INFO Channel 01/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-167-9:1039542:1040102 [5] NCCL INFO Channel 00/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-167-9:1039538:1040099 [1] NCCL INFO Channel 01/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-167-9:1039540:1040103 [3] NCCL INFO Channel 02/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-167-9:1039544:1040101 [7] NCCL INFO Channel 02/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-167-9:1039542:1040102 [5] NCCL INFO Channel 01/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-166-214:624950:625499 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-166-214:624950:625499 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-165-213:9626:10173 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-165-213:9626:10173 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-167-9:1039538:1040099 [1] NCCL INFO Channel 02/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-167-9:1039540:1040103 [3] NCCL INFO Channel 03/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-167-9:1039544:1040101 [7] NCCL INFO Channel 03/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-167-9:1039542:1040102 [5] NCCL INFO Channel 02/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-166-15:127296:127834 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-166-15:127296:127834 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-166-244:897243:897790 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-166-244:897243:897790 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-167-9:1039537:1040095 [0] NCCL INFO Channel 00/24 :    0   1   2   3
+ip-26-0-167-9:1039537:1040095 [0] NCCL INFO Channel 01/24 :    0   1   2   3
+ip-26-0-167-9:1039537:1040095 [0] NCCL INFO Channel 02/24 :    0   1   2   3
+ip-26-0-167-9:1039537:1040095 [0] NCCL INFO Channel 03/24 :    0   1   2   3
+ip-26-0-167-9:1039537:1040095 [0] NCCL INFO Channel 04/24 :    0   1   2   3
+ip-26-0-167-9:1039537:1040095 [0] NCCL INFO Channel 05/24 :    0   1   2   3
+ip-26-0-167-9:1039537:1040095 [0] NCCL INFO Channel 06/24 :    0   1   2   3
+ip-26-0-167-9:1039537:1040095 [0] NCCL INFO Channel 07/24 :    0   1   2   3
+ip-26-0-167-9:1039537:1040095 [0] NCCL INFO Channel 08/24 :    0   1   2   3
+ip-26-0-167-9:1039539:1040100 [2] NCCL INFO Trees [0] 2/-1/-1->1->0 [1] 2/-1/-1->1->0 [2] 2/-1/-1->1->0 [3] 2/-1/-1->1->0 [4] 2/-1/-1->1->0 [5] 2/-1/-1->1->0 [6] 2/-1/-1->1->0 [7] 2/-1/-1->1->0 [8] 2/-1/-1->1->0 [9] 2/-1/-1->1->0 [10] 2/-1/-1->1->0 [11] 2/-1/-1->1->0 [12] 2/-1/-1->1->0 [13] 2/-1/-1->1->0 [14] 2/-1/-1->1->0 [15] 2/-1/-1->1->0 [16] 2/-1/-1->1->0 [17] 2/-1/-1->1->0 [18] 2/-1/-1->1->0 [19] 2/-1/-1->1->0 [20] 2/-1/-1->1->0 [21] 2/-1/-1->1->0 [22] 2/-1/-1->1->0 [23] 2/-1/-1->1->0
+ip-26-0-167-9:1039537:1040095 [0] NCCL INFO Channel 09/24 :    0   1   2   3
+ip-26-0-167-9:1039541:1040097 [4] NCCL INFO Trees [0] 3/-1/-1->2->1 [1] 3/-1/-1->2->1 [2] 3/-1/-1->2->1 [3] 3/-1/-1->2->1 [4] 3/-1/-1->2->1 [5] 3/-1/-1->2->1 [6] 3/-1/-1->2->1 [7] 3/-1/-1->2->1 [8] 3/-1/-1->2->1 [9] 3/-1/-1->2->1 [10] 3/-1/-1->2->1 [11] 3/-1/-1->2->1 [12] 3/-1/-1->2->1 [13] 3/-1/-1->2->1 [14] 3/-1/-1->2->1 [15] 3/-1/-1->2->1 [16] 3/-1/-1->2->1 [17] 3/-1/-1->2->1 [18] 3/-1/-1->2->1 [19] 3/-1/-1->2->1 [20] 3/-1/-1->2->1 [21] 3/-1/-1->2->1 [22] 3/-1/-1->2->1 [23] 3/-1/-1->2->1
+ip-26-0-167-9:1039543:1040098 [6] NCCL INFO Trees [0] -1/-1/-1->3->2 [1] -1/-1/-1->3->2 [2] -1/-1/-1->3->2 [3] -1/-1/-1->3->2 [4] -1/-1/-1->3->2 [5] -1/-1/-1->3->2 [6] -1/-1/-1->3->2 [7] -1/-1/-1->3->2 [8] -1/-1/-1->3->2 [9] -1/-1/-1->3->2 [10] -1/-1/-1->3->2 [11] -1/-1/-1->3->2 [12] -1/-1/-1->3->2 [13] -1/-1/-1->3->2 [14] -1/-1/-1->3->2 [15] -1/-1/-1->3->2 [16] -1/-1/-1->3->2 [17] -1/-1/-1->3->2 [18] -1/-1/-1->3->2 [19] -1/-1/-1->3->2 [20] -1/-1/-1->3->2 [21] -1/-1/-1->3->2 [22] -1/-1/-1->3->2 [23] -1/-1/-1->3->2
+ip-26-0-167-9:1039539:1040100 [2] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-167-9:1039537:1040095 [0] NCCL INFO Channel 10/24 :    0   1   2   3
+ip-26-0-167-9:1039541:1040097 [4] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-167-9:1039543:1040098 [6] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-167-9:1039537:1040095 [0] NCCL INFO Channel 11/24 :    0   1   2   3
+ip-26-0-167-9:1039537:1040095 [0] NCCL INFO Channel 12/24 :    0   1   2   3
+ip-26-0-167-9:1039537:1040095 [0] NCCL INFO Channel 13/24 :    0   1   2   3
+ip-26-0-167-9:1039537:1040095 [0] NCCL INFO Channel 14/24 :    0   1   2   3
+ip-26-0-167-9:1039537:1040095 [0] NCCL INFO Channel 15/24 :    0   1   2   3
+ip-26-0-167-9:1039537:1040095 [0] NCCL INFO Channel 16/24 :    0   1   2   3
+ip-26-0-167-9:1039537:1040095 [0] NCCL INFO Channel 17/24 :    0   1   2   3
+ip-26-0-167-9:1039537:1040095 [0] NCCL INFO Channel 18/24 :    0   1   2   3
+ip-26-0-167-9:1039537:1040095 [0] NCCL INFO Channel 19/24 :    0   1   2   3
+ip-26-0-167-9:1039537:1040095 [0] NCCL INFO Channel 20/24 :    0   1   2   3
+ip-26-0-167-9:1039537:1040095 [0] NCCL INFO Channel 21/24 :    0   1   2   3
+ip-26-0-167-9:1039537:1040095 [0] NCCL INFO Channel 22/24 :    0   1   2   3
+ip-26-0-167-9:1039537:1040095 [0] NCCL INFO Channel 23/24 :    0   1   2   3
+ip-26-0-167-9:1039537:1040095 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] 1/-1/-1->0->-1 [7] 1/-1/-1->0->-1 [8] 1/-1/-1->0->-1 [9] 1/-1/-1->0->-1 [10] 1/-1/-1->0->-1 [11] 1/-1/-1->0->-1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] 1/-1/-1->0->-1 [19] 1/-1/-1->0->-1 [20] 1/-1/-1->0->-1 [21] 1/-1/-1->0->-1 [22] 1/-1/-1->0->-1 [23] 1/-1/-1->0->-1
+ip-26-0-167-9:1039537:1040095 [0] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-167-9:1039538:1040099 [1] NCCL INFO Channel 03/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-167-9:1039540:1040103 [3] NCCL INFO Channel 04/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-167-9:1039544:1040101 [7] NCCL INFO Channel 04/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-167-9:1039542:1040102 [5] NCCL INFO Channel 03/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-167-9:1039538:1040099 [1] NCCL INFO Channel 04/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-167-9:1039544:1040101 [7] NCCL INFO Channel 05/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-167-9:1039540:1040103 [3] NCCL INFO Channel 05/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-167-9:1039542:1040102 [5] NCCL INFO Channel 04/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-162-46:702883:703471 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-162-46:702883:703471 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-167-9:1039538:1040099 [1] NCCL INFO Channel 05/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-167-9:1039540:1040103 [3] NCCL INFO Channel 06/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-167-9:1039544:1040101 [7] NCCL INFO Channel 06/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-167-9:1039542:1040102 [5] NCCL INFO Channel 05/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-166-244:897242:897792 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-166-244:897242:897792 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-167-9:1039538:1040099 [1] NCCL INFO Channel 06/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-167-9:1039540:1040103 [3] NCCL INFO Channel 07/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-167-9:1039542:1040102 [5] NCCL INFO Channel 06/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-167-9:1039544:1040101 [7] NCCL INFO Channel 07/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-166-214:624943:625494 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-166-214:624943:625494 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-167-9:1039538:1040099 [1] NCCL INFO Channel 07/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-167-9:1039542:1040102 [5] NCCL INFO Channel 07/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-167-9:1039540:1040103 [3] NCCL INFO Channel 08/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-167-9:1039542:1040102 [5] NCCL INFO Channel 08/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-167-9:1039540:1040103 [3] NCCL INFO Channel 09/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-167-9:1039542:1040102 [5] NCCL INFO Channel 09/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-167-9:1039540:1040103 [3] NCCL INFO Channel 10/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-165-213:9627:10174 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-165-213:9627:10174 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-167-9:1039544:1040101 [7] NCCL INFO Channel 08/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-167-9:1039542:1040102 [5] NCCL INFO Channel 10/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-166-244:897241:897784 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-166-244:897241:897784 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-167-9:1039538:1040099 [1] NCCL INFO Channel 08/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-167-9:1039544:1040101 [7] NCCL INFO Channel 09/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-167-9:1039542:1040102 [5] NCCL INFO Channel 11/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-166-244:897247:897788 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-166-244:897247:897788 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-167-9:1039542:1040102 [5] NCCL INFO Channel 12/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-167-9:1039540:1040103 [3] NCCL INFO Channel 11/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-167-9:1039540:1040103 [3] NCCL INFO Channel 12/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-167-9:1039542:1040102 [5] NCCL INFO Channel 13/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-165-213:9626:10173 [1] NCCL INFO Channel 00/24 :    0   1   2   3
+ip-26-0-165-213:9626:10173 [1] NCCL INFO Channel 01/24 :    0   1   2   3
+ip-26-0-165-213:9626:10173 [1] NCCL INFO Channel 02/24 :    0   1   2   3
+ip-26-0-165-213:9626:10173 [1] NCCL INFO Channel 03/24 :    0   1   2   3
+ip-26-0-165-213:9626:10173 [1] NCCL INFO Channel 04/24 :    0   1   2   3
+ip-26-0-166-15:127291:127838 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-165-213:9626:10173 [1] NCCL INFO Channel 05/24 :    0   1   2   3
+ip-26-0-165-213:9632:10176 [7] NCCL INFO Trees [0] -1/-1/-1->3->2 [1] -1/-1/-1->3->2 [2] -1/-1/-1->3->2 [3] -1/-1/-1->3->2 [4] -1/-1/-1->3->2 [5] -1/-1/-1->3->2 [6] -1/-1/-1->3->2 [7] -1/-1/-1->3->2 [8] -1/-1/-1->3->2 [9] -1/-1/-1->3->2 [10] -1/-1/-1->3->2 [11] -1/-1/-1->3->2 [12] -1/-1/-1->3->2 [13] -1/-1/-1->3->2 [14] -1/-1/-1->3->2 [15] -1/-1/-1->3->2 [16] -1/-1/-1->3->2 [17] -1/-1/-1->3->2 [18] -1/-1/-1->3->2 [19] -1/-1/-1->3->2 [20] -1/-1/-1->3->2 [21] -1/-1/-1->3->2 [22] -1/-1/-1->3->2 [23] -1/-1/-1->3->2
+ip-26-0-165-213:9626:10173 [1] NCCL INFO Channel 06/24 :    0   1   2   3
+ip-26-0-165-213:9632:10176 [7] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-165-213:9630:10175 [5] NCCL INFO Trees [0] 3/-1/-1->2->1 [1] 3/-1/-1->2->1 [2] 3/-1/-1->2->1 [3] 3/-1/-1->2->1 [4] 3/-1/-1->2->1 [5] 3/-1/-1->2->1 [6] 3/-1/-1->2->1 [7] 3/-1/-1->2->1 [8] 3/-1/-1->2->1 [9] 3/-1/-1->2->1 [10] 3/-1/-1->2->1 [11] 3/-1/-1->2->1 [12] 3/-1/-1->2->1 [13] 3/-1/-1->2->1 [14] 3/-1/-1->2->1 [15] 3/-1/-1->2->1 [16] 3/-1/-1->2->1 [17] 3/-1/-1->2->1 [18] 3/-1/-1->2->1 [19] 3/-1/-1->2->1 [20] 3/-1/-1->2->1 [21] 3/-1/-1->2->1 [22] 3/-1/-1->2->1 [23] 3/-1/-1->2->1
+ip-26-0-165-213:9628:10177 [3] NCCL INFO Trees [0] 2/-1/-1->1->0 [1] 2/-1/-1->1->0 [2] 2/-1/-1->1->0 [3] 2/-1/-1->1->0 [4] 2/-1/-1->1->0 [5] 2/-1/-1->1->0 [6] 2/-1/-1->1->0 [7] 2/-1/-1->1->0 [8] 2/-1/-1->1->0 [9] 2/-1/-1->1->0 [10] 2/-1/-1->1->0 [11] 2/-1/-1->1->0 [12] 2/-1/-1->1->0 [13] 2/-1/-1->1->0 [14] 2/-1/-1->1->0 [15] 2/-1/-1->1->0 [16] 2/-1/-1->1->0 [17] 2/-1/-1->1->0 [18] 2/-1/-1->1->0 [19] 2/-1/-1->1->0 [20] 2/-1/-1->1->0 [21] 2/-1/-1->1->0 [22] 2/-1/-1->1->0 [23] 2/-1/-1->1->0
+ip-26-0-165-213:9626:10173 [1] NCCL INFO Channel 07/24 :    0   1   2   3
+ip-26-0-165-213:9630:10175 [5] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-165-213:9626:10173 [1] NCCL INFO Channel 08/24 :    0   1   2   3
+ip-26-0-165-213:9628:10177 [3] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-165-213:9626:10173 [1] NCCL INFO Channel 09/24 :    0   1   2   3
+ip-26-0-165-213:9626:10173 [1] NCCL INFO Channel 10/24 :    0   1   2   3
+ip-26-0-165-213:9626:10173 [1] NCCL INFO Channel 11/24 :    0   1   2   3
+ip-26-0-165-213:9626:10173 [1] NCCL INFO Channel 12/24 :    0   1   2   3
+ip-26-0-165-213:9626:10173 [1] NCCL INFO Channel 13/24 :    0   1   2   3
+ip-26-0-165-213:9626:10173 [1] NCCL INFO Channel 14/24 :    0   1   2   3
+ip-26-0-165-213:9626:10173 [1] NCCL INFO Channel 15/24 :    0   1   2   3
+ip-26-0-165-213:9626:10173 [1] NCCL INFO Channel 16/24 :    0   1   2   3
+ip-26-0-165-213:9626:10173 [1] NCCL INFO Channel 17/24 :    0   1   2   3
+ip-26-0-165-213:9626:10173 [1] NCCL INFO Channel 18/24 :    0   1   2   3
+ip-26-0-165-213:9626:10173 [1] NCCL INFO Channel 19/24 :    0   1   2   3
+ip-26-0-165-213:9626:10173 [1] NCCL INFO Channel 20/24 :    0   1   2   3
+ip-26-0-165-213:9626:10173 [1] NCCL INFO Channel 21/24 :    0   1   2   3
+ip-26-0-165-213:9626:10173 [1] NCCL INFO Channel 22/24 :    0   1   2   3
+ip-26-0-166-15:127291:127838 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-165-213:9626:10173 [1] NCCL INFO Channel 23/24 :    0   1   2   3
+ip-26-0-165-213:9626:10173 [1] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] 1/-1/-1->0->-1 [7] 1/-1/-1->0->-1 [8] 1/-1/-1->0->-1 [9] 1/-1/-1->0->-1 [10] 1/-1/-1->0->-1 [11] 1/-1/-1->0->-1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] 1/-1/-1->0->-1 [19] 1/-1/-1->0->-1 [20] 1/-1/-1->0->-1 [21] 1/-1/-1->0->-1 [22] 1/-1/-1->0->-1 [23] 1/-1/-1->0->-1
+ip-26-0-165-213:9626:10173 [1] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-167-9:1039540:1040103 [3] NCCL INFO Channel 13/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-167-9:1039542:1040102 [5] NCCL INFO Channel 14/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-165-213:9629:10171 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-165-213:9629:10171 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-167-9:1039540:1040103 [3] NCCL INFO Channel 14/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-167-9:1039542:1040102 [5] NCCL INFO Channel 15/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-167-9:1039540:1040103 [3] NCCL INFO Channel 15/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-167-9:1039542:1040102 [5] NCCL INFO Channel 16/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-167-9:1039540:1040103 [3] NCCL INFO Channel 16/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-167-9:1039542:1040102 [5] NCCL INFO Channel 17/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-166-244:897240:897786 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-167-9:1039538:1040099 [1] NCCL INFO Channel 09/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-166-244:897240:897786 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-167-9:1039544:1040101 [7] NCCL INFO Channel 10/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-166-15:127294:127835 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-166-15:127294:127835 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-167-9:1039540:1040103 [3] NCCL INFO Channel 17/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-167-9:1039542:1040102 [5] NCCL INFO Channel 18/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-167-9:1039544:1040101 [7] NCCL INFO Channel 11/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-167-9:1039538:1040099 [1] NCCL INFO Channel 10/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-167-9:1039540:1040103 [3] NCCL INFO Channel 18/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-167-9:1039542:1040102 [5] NCCL INFO Channel 19/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-167-9:1039544:1040101 [7] NCCL INFO Channel 12/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-167-9:1039538:1040099 [1] NCCL INFO Channel 11/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-167-9:1039540:1040103 [3] NCCL INFO Channel 19/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-167-9:1039542:1040102 [5] NCCL INFO Channel 20/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-167-9:1039544:1040101 [7] NCCL INFO Channel 13/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-167-9:1039538:1040099 [1] NCCL INFO Channel 12/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-166-244:897245:897787 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-166-244:897245:897787 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-167-9:1039540:1040103 [3] NCCL INFO Channel 20/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-167-9:1039542:1040102 [5] NCCL INFO Channel 21/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-167-9:1039544:1040101 [7] NCCL INFO Channel 14/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-167-9:1039538:1040099 [1] NCCL INFO Channel 13/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-167-9:1039540:1040103 [3] NCCL INFO Channel 21/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-167-9:1039542:1040102 [5] NCCL INFO Channel 22/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-167-9:1039544:1040101 [7] NCCL INFO Channel 15/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-167-9:1039538:1040099 [1] NCCL INFO Channel 14/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-167-9:1039542:1040102 [5] NCCL INFO Channel 23/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-167-9:1039540:1040103 [3] NCCL INFO Channel 22/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-167-9:1039538:1040099 [1] NCCL INFO Channel 15/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-167-9:1039544:1040101 [7] NCCL INFO Channel 16/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-166-15:127291:127838 [1] NCCL INFO Channel 00/24 :    0   1   2   3
+ip-26-0-166-15:127291:127838 [1] NCCL INFO Channel 01/24 :    0   1   2   3
+ip-26-0-166-15:127291:127838 [1] NCCL INFO Channel 02/24 :    0   1   2   3
+ip-26-0-166-15:127291:127838 [1] NCCL INFO Channel 03/24 :    0   1   2   3
+ip-26-0-166-15:127291:127838 [1] NCCL INFO Channel 04/24 :    0   1   2   3
+ip-26-0-166-15:127291:127838 [1] NCCL INFO Channel 05/24 :    0   1   2   3
+ip-26-0-166-15:127291:127838 [1] NCCL INFO Channel 06/24 :    0   1   2   3
+ip-26-0-166-15:127291:127838 [1] NCCL INFO Channel 07/24 :    0   1   2   3
+ip-26-0-166-15:127291:127838 [1] NCCL INFO Channel 08/24 :    0   1   2   3
+ip-26-0-166-15:127291:127838 [1] NCCL INFO Channel 09/24 :    0   1   2   3
+ip-26-0-166-15:127291:127838 [1] NCCL INFO Channel 10/24 :    0   1   2   3
+ip-26-0-166-15:127291:127838 [1] NCCL INFO Channel 11/24 :    0   1   2   3
+ip-26-0-166-15:127291:127838 [1] NCCL INFO Channel 12/24 :    0   1   2   3
+ip-26-0-166-15:127291:127838 [1] NCCL INFO Channel 13/24 :    0   1   2   3
+ip-26-0-166-15:127295:127839 [5] NCCL INFO Trees [0] 3/-1/-1->2->1 [1] 3/-1/-1->2->1 [2] 3/-1/-1->2->1 [3] 3/-1/-1->2->1 [4] 3/-1/-1->2->1 [5] 3/-1/-1->2->1 [6] 3/-1/-1->2->1 [7] 3/-1/-1->2->1 [8] 3/-1/-1->2->1 [9] 3/-1/-1->2->1 [10] 3/-1/-1->2->1 [11] 3/-1/-1->2->1 [12] 3/-1/-1->2->1 [13] 3/-1/-1->2->1 [14] 3/-1/-1->2->1 [15] 3/-1/-1->2->1 [16] 3/-1/-1->2->1 [17] 3/-1/-1->2->1 [18] 3/-1/-1->2->1 [19] 3/-1/-1->2->1 [20] 3/-1/-1->2->1 [21] 3/-1/-1->2->1 [22] 3/-1/-1->2->1 [23] 3/-1/-1->2->1
+ip-26-0-166-15:127291:127838 [1] NCCL INFO Channel 14/24 :    0   1   2   3
+ip-26-0-166-15:127295:127839 [5] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-166-15:127291:127838 [1] NCCL INFO Channel 15/24 :    0   1   2   3
+ip-26-0-166-15:127297:127840 [7] NCCL INFO Trees [0] -1/-1/-1->3->2 [1] -1/-1/-1->3->2 [2] -1/-1/-1->3->2 [3] -1/-1/-1->3->2 [4] -1/-1/-1->3->2 [5] -1/-1/-1->3->2 [6] -1/-1/-1->3->2 [7] -1/-1/-1->3->2 [8] -1/-1/-1->3->2 [9] -1/-1/-1->3->2 [10] -1/-1/-1->3->2 [11] -1/-1/-1->3->2 [12] -1/-1/-1->3->2 [13] -1/-1/-1->3->2 [14] -1/-1/-1->3->2 [15] -1/-1/-1->3->2 [16] -1/-1/-1->3->2 [17] -1/-1/-1->3->2 [18] -1/-1/-1->3->2 [19] -1/-1/-1->3->2 [20] -1/-1/-1->3->2 [21] -1/-1/-1->3->2 [22] -1/-1/-1->3->2 [23] -1/-1/-1->3->2
+ip-26-0-166-15:127293:127841 [3] NCCL INFO Trees [0] 2/-1/-1->1->0 [1] 2/-1/-1->1->0 [2] 2/-1/-1->1->0 [3] 2/-1/-1->1->0 [4] 2/-1/-1->1->0 [5] 2/-1/-1->1->0 [6] 2/-1/-1->1->0 [7] 2/-1/-1->1->0 [8] 2/-1/-1->1->0 [9] 2/-1/-1->1->0 [10] 2/-1/-1->1->0 [11] 2/-1/-1->1->0 [12] 2/-1/-1->1->0 [13] 2/-1/-1->1->0 [14] 2/-1/-1->1->0 [15] 2/-1/-1->1->0 [16] 2/-1/-1->1->0 [17] 2/-1/-1->1->0 [18] 2/-1/-1->1->0 [19] 2/-1/-1->1->0 [20] 2/-1/-1->1->0 [21] 2/-1/-1->1->0 [22] 2/-1/-1->1->0 [23] 2/-1/-1->1->0
+ip-26-0-166-15:127291:127838 [1] NCCL INFO Channel 16/24 :    0   1   2   3
+ip-26-0-166-15:127297:127840 [7] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-166-15:127291:127838 [1] NCCL INFO Channel 17/24 :    0   1   2   3
+ip-26-0-166-15:127293:127841 [3] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-166-15:127291:127838 [1] NCCL INFO Channel 18/24 :    0   1   2   3
+ip-26-0-166-15:127291:127838 [1] NCCL INFO Channel 19/24 :    0   1   2   3
+ip-26-0-166-15:127291:127838 [1] NCCL INFO Channel 20/24 :    0   1   2   3
+ip-26-0-166-15:127291:127838 [1] NCCL INFO Channel 21/24 :    0   1   2   3
+ip-26-0-166-15:127291:127838 [1] NCCL INFO Channel 22/24 :    0   1   2   3
+ip-26-0-166-15:127291:127838 [1] NCCL INFO Channel 23/24 :    0   1   2   3
+ip-26-0-166-15:127291:127838 [1] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] 1/-1/-1->0->-1 [7] 1/-1/-1->0->-1 [8] 1/-1/-1->0->-1 [9] 1/-1/-1->0->-1 [10] 1/-1/-1->0->-1 [11] 1/-1/-1->0->-1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] 1/-1/-1->0->-1 [19] 1/-1/-1->0->-1 [20] 1/-1/-1->0->-1 [21] 1/-1/-1->0->-1 [22] 1/-1/-1->0->-1 [23] 1/-1/-1->0->-1
+ip-26-0-166-15:127291:127838 [1] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-167-9:1039540:1040103 [3] NCCL INFO Channel 23/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-165-213:9625:10169 [0] NCCL INFO Channel 00/24 :    0   1   2   3
+ip-26-0-165-213:9625:10169 [0] NCCL INFO Channel 01/24 :    0   1   2   3
+ip-26-0-165-213:9629:10171 [4] NCCL INFO Trees [0] 3/-1/-1->2->1 [1] 3/-1/-1->2->1 [2] 3/-1/-1->2->1 [3] 3/-1/-1->2->1 [4] 3/-1/-1->2->1 [5] 3/-1/-1->2->1 [6] 3/-1/-1->2->1 [7] 3/-1/-1->2->1 [8] 3/-1/-1->2->1 [9] 3/-1/-1->2->1 [10] 3/-1/-1->2->1 [11] 3/-1/-1->2->1 [12] 3/-1/-1->2->1 [13] 3/-1/-1->2->1 [14] 3/-1/-1->2->1 [15] 3/-1/-1->2->1 [16] 3/-1/-1->2->1 [17] 3/-1/-1->2->1 [18] 3/-1/-1->2->1 [19] 3/-1/-1->2->1 [20] 3/-1/-1->2->1 [21] 3/-1/-1->2->1 [22] 3/-1/-1->2->1 [23] 3/-1/-1->2->1
+ip-26-0-165-213:9625:10169 [0] NCCL INFO Channel 02/24 :    0   1   2   3
+ip-26-0-165-213:9629:10171 [4] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-165-213:9627:10174 [2] NCCL INFO Trees [0] 2/-1/-1->1->0 [1] 2/-1/-1->1->0 [2] 2/-1/-1->1->0 [3] 2/-1/-1->1->0 [4] 2/-1/-1->1->0 [5] 2/-1/-1->1->0 [6] 2/-1/-1->1->0 [7] 2/-1/-1->1->0 [8] 2/-1/-1->1->0 [9] 2/-1/-1->1->0 [10] 2/-1/-1->1->0 [11] 2/-1/-1->1->0 [12] 2/-1/-1->1->0 [13] 2/-1/-1->1->0 [14] 2/-1/-1->1->0 [15] 2/-1/-1->1->0 [16] 2/-1/-1->1->0 [17] 2/-1/-1->1->0 [18] 2/-1/-1->1->0 [19] 2/-1/-1->1->0 [20] 2/-1/-1->1->0 [21] 2/-1/-1->1->0 [22] 2/-1/-1->1->0 [23] 2/-1/-1->1->0
+ip-26-0-165-213:9625:10169 [0] NCCL INFO Channel 03/24 :    0   1   2   3
+ip-26-0-165-213:9625:10169 [0] NCCL INFO Channel 04/24 :    0   1   2   3
+ip-26-0-165-213:9627:10174 [2] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-165-213:9625:10169 [0] NCCL INFO Channel 05/24 :    0   1   2   3
+ip-26-0-165-213:9631:10172 [6] NCCL INFO Trees [0] -1/-1/-1->3->2 [1] -1/-1/-1->3->2 [2] -1/-1/-1->3->2 [3] -1/-1/-1->3->2 [4] -1/-1/-1->3->2 [5] -1/-1/-1->3->2 [6] -1/-1/-1->3->2 [7] -1/-1/-1->3->2 [8] -1/-1/-1->3->2 [9] -1/-1/-1->3->2 [10] -1/-1/-1->3->2 [11] -1/-1/-1->3->2 [12] -1/-1/-1->3->2 [13] -1/-1/-1->3->2 [14] -1/-1/-1->3->2 [15] -1/-1/-1->3->2 [16] -1/-1/-1->3->2 [17] -1/-1/-1->3->2 [18] -1/-1/-1->3->2 [19] -1/-1/-1->3->2 [20] -1/-1/-1->3->2 [21] -1/-1/-1->3->2 [22] -1/-1/-1->3->2 [23] -1/-1/-1->3->2
+ip-26-0-165-213:9625:10169 [0] NCCL INFO Channel 06/24 :    0   1   2   3
+ip-26-0-165-213:9631:10172 [6] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-165-213:9625:10169 [0] NCCL INFO Channel 07/24 :    0   1   2   3
+ip-26-0-165-213:9625:10169 [0] NCCL INFO Channel 08/24 :    0   1   2   3
+ip-26-0-165-213:9625:10169 [0] NCCL INFO Channel 09/24 :    0   1   2   3
+ip-26-0-165-213:9625:10169 [0] NCCL INFO Channel 10/24 :    0   1   2   3
+ip-26-0-165-213:9625:10169 [0] NCCL INFO Channel 11/24 :    0   1   2   3
+ip-26-0-165-213:9625:10169 [0] NCCL INFO Channel 12/24 :    0   1   2   3
+ip-26-0-165-213:9625:10169 [0] NCCL INFO Channel 13/24 :    0   1   2   3
+ip-26-0-165-213:9625:10169 [0] NCCL INFO Channel 14/24 :    0   1   2   3
+ip-26-0-165-213:9625:10169 [0] NCCL INFO Channel 15/24 :    0   1   2   3
+ip-26-0-165-213:9625:10169 [0] NCCL INFO Channel 16/24 :    0   1   2   3
+ip-26-0-165-213:9625:10169 [0] NCCL INFO Channel 17/24 :    0   1   2   3
+ip-26-0-165-213:9625:10169 [0] NCCL INFO Channel 18/24 :    0   1   2   3
+ip-26-0-165-213:9625:10169 [0] NCCL INFO Channel 19/24 :    0   1   2   3
+ip-26-0-165-213:9625:10169 [0] NCCL INFO Channel 20/24 :    0   1   2   3
+ip-26-0-165-213:9625:10169 [0] NCCL INFO Channel 21/24 :    0   1   2   3
+ip-26-0-165-213:9625:10169 [0] NCCL INFO Channel 22/24 :    0   1   2   3
+ip-26-0-165-213:9625:10169 [0] NCCL INFO Channel 23/24 :    0   1   2   3
+ip-26-0-165-213:9625:10169 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] 1/-1/-1->0->-1 [7] 1/-1/-1->0->-1 [8] 1/-1/-1->0->-1 [9] 1/-1/-1->0->-1 [10] 1/-1/-1->0->-1 [11] 1/-1/-1->0->-1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] 1/-1/-1->0->-1 [19] 1/-1/-1->0->-1 [20] 1/-1/-1->0->-1 [21] 1/-1/-1->0->-1 [22] 1/-1/-1->0->-1 [23] 1/-1/-1->0->-1
+ip-26-0-165-213:9625:10169 [0] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-167-9:1039544:1040101 [7] NCCL INFO Channel 17/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-167-9:1039538:1040099 [1] NCCL INFO Channel 16/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-167-9:1039544:1040101 [7] NCCL INFO Channel 18/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-167-9:1039538:1040099 [1] NCCL INFO Channel 17/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-167-9:1039544:1040101 [7] NCCL INFO Channel 19/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-167-9:1039538:1040099 [1] NCCL INFO Channel 18/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-165-213:9626:10173 [1] NCCL INFO Channel 00/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-166-244:897240:897786 [0] NCCL INFO Channel 00/24 :    0   1   2   3
+ip-26-0-166-244:897240:897786 [0] NCCL INFO Channel 01/24 :    0   1   2   3
+ip-26-0-166-244:897240:897786 [0] NCCL INFO Channel 02/24 :    0   1   2   3
+ip-26-0-166-244:897240:897786 [0] NCCL INFO Channel 03/24 :    0   1   2   3
+ip-26-0-166-244:897240:897786 [0] NCCL INFO Channel 04/24 :    0   1   2   3
+ip-26-0-166-244:897240:897786 [0] NCCL INFO Channel 05/24 :    0   1   2   3
+ip-26-0-166-244:897240:897786 [0] NCCL INFO Channel 06/24 :    0   1   2   3
+ip-26-0-166-244:897240:897786 [0] NCCL INFO Channel 07/24 :    0   1   2   3
+ip-26-0-166-244:897240:897786 [0] NCCL INFO Channel 08/24 :    0   1   2   3
+ip-26-0-166-244:897246:897791 [6] NCCL INFO Trees [0] -1/-1/-1->3->2 [1] -1/-1/-1->3->2 [2] -1/-1/-1->3->2 [3] -1/-1/-1->3->2 [4] -1/-1/-1->3->2 [5] -1/-1/-1->3->2 [6] -1/-1/-1->3->2 [7] -1/-1/-1->3->2 [8] -1/-1/-1->3->2 [9] -1/-1/-1->3->2 [10] -1/-1/-1->3->2 [11] -1/-1/-1->3->2 [12] -1/-1/-1->3->2 [13] -1/-1/-1->3->2 [14] -1/-1/-1->3->2 [15] -1/-1/-1->3->2 [16] -1/-1/-1->3->2 [17] -1/-1/-1->3->2 [18] -1/-1/-1->3->2 [19] -1/-1/-1->3->2 [20] -1/-1/-1->3->2 [21] -1/-1/-1->3->2 [22] -1/-1/-1->3->2 [23] -1/-1/-1->3->2
+ip-26-0-166-244:897240:897786 [0] NCCL INFO Channel 09/24 :    0   1   2   3
+ip-26-0-166-244:897240:897786 [0] NCCL INFO Channel 10/24 :    0   1   2   3
+ip-26-0-166-244:897246:897791 [6] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-166-244:897240:897786 [0] NCCL INFO Channel 11/24 :    0   1   2   3
+ip-26-0-166-244:897240:897786 [0] NCCL INFO Channel 12/24 :    0   1   2   3
+ip-26-0-166-244:897240:897786 [0] NCCL INFO Channel 13/24 :    0   1   2   3
+ip-26-0-166-244:897240:897786 [0] NCCL INFO Channel 14/24 :    0   1   2   3
+ip-26-0-166-244:897240:897786 [0] NCCL INFO Channel 15/24 :    0   1   2   3
+ip-26-0-166-244:897240:897786 [0] NCCL INFO Channel 16/24 :    0   1   2   3
+ip-26-0-166-244:897240:897786 [0] NCCL INFO Channel 17/24 :    0   1   2   3
+ip-26-0-166-244:897242:897792 [2] NCCL INFO Trees [0] 2/-1/-1->1->0 [1] 2/-1/-1->1->0 [2] 2/-1/-1->1->0 [3] 2/-1/-1->1->0 [4] 2/-1/-1->1->0 [5] 2/-1/-1->1->0 [6] 2/-1/-1->1->0 [7] 2/-1/-1->1->0 [8] 2/-1/-1->1->0 [9] 2/-1/-1->1->0 [10] 2/-1/-1->1->0 [11] 2/-1/-1->1->0 [12] 2/-1/-1->1->0 [13] 2/-1/-1->1->0 [14] 2/-1/-1->1->0 [15] 2/-1/-1->1->0 [16] 2/-1/-1->1->0 [17] 2/-1/-1->1->0 [18] 2/-1/-1->1->0 [19] 2/-1/-1->1->0 [20] 2/-1/-1->1->0 [21] 2/-1/-1->1->0 [22] 2/-1/-1->1->0 [23] 2/-1/-1->1->0
+ip-26-0-166-244:897240:897786 [0] NCCL INFO Channel 18/24 :    0   1   2   3
+ip-26-0-166-244:897240:897786 [0] NCCL INFO Channel 19/24 :    0   1   2   3
+ip-26-0-166-244:897242:897792 [2] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-166-244:897240:897786 [0] NCCL INFO Channel 20/24 :    0   1   2   3
+ip-26-0-166-244:897240:897786 [0] NCCL INFO Channel 21/24 :    0   1   2   3
+ip-26-0-166-244:897240:897786 [0] NCCL INFO Channel 22/24 :    0   1   2   3
+ip-26-0-166-244:897240:897786 [0] NCCL INFO Channel 23/24 :    0   1   2   3
+ip-26-0-166-244:897244:897789 [4] NCCL INFO Trees [0] 3/-1/-1->2->1 [1] 3/-1/-1->2->1 [2] 3/-1/-1->2->1 [3] 3/-1/-1->2->1 [4] 3/-1/-1->2->1 [5] 3/-1/-1->2->1 [6] 3/-1/-1->2->1 [7] 3/-1/-1->2->1 [8] 3/-1/-1->2->1 [9] 3/-1/-1->2->1 [10] 3/-1/-1->2->1 [11] 3/-1/-1->2->1 [12] 3/-1/-1->2->1 [13] 3/-1/-1->2->1 [14] 3/-1/-1->2->1 [15] 3/-1/-1->2->1 [16] 3/-1/-1->2->1 [17] 3/-1/-1->2->1 [18] 3/-1/-1->2->1 [19] 3/-1/-1->2->1 [20] 3/-1/-1->2->1 [21] 3/-1/-1->2->1 [22] 3/-1/-1->2->1 [23] 3/-1/-1->2->1
+ip-26-0-166-244:897244:897789 [4] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-166-244:897240:897786 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] 1/-1/-1->0->-1 [7] 1/-1/-1->0->-1 [8] 1/-1/-1->0->-1 [9] 1/-1/-1->0->-1 [10] 1/-1/-1->0->-1 [11] 1/-1/-1->0->-1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] 1/-1/-1->0->-1 [19] 1/-1/-1->0->-1 [20] 1/-1/-1->0->-1 [21] 1/-1/-1->0->-1 [22] 1/-1/-1->0->-1 [23] 1/-1/-1->0->-1
+ip-26-0-166-244:897240:897786 [0] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-166-15:127290:127833 [0] NCCL INFO Channel 00/24 :    0   1   2   3
+ip-26-0-166-15:127294:127835 [4] NCCL INFO Trees [0] 3/-1/-1->2->1 [1] 3/-1/-1->2->1 [2] 3/-1/-1->2->1 [3] 3/-1/-1->2->1 [4] 3/-1/-1->2->1 [5] 3/-1/-1->2->1 [6] 3/-1/-1->2->1 [7] 3/-1/-1->2->1 [8] 3/-1/-1->2->1 [9] 3/-1/-1->2->1 [10] 3/-1/-1->2->1 [11] 3/-1/-1->2->1 [12] 3/-1/-1->2->1 [13] 3/-1/-1->2->1 [14] 3/-1/-1->2->1 [15] 3/-1/-1->2->1 [16] 3/-1/-1->2->1 [17] 3/-1/-1->2->1 [18] 3/-1/-1->2->1 [19] 3/-1/-1->2->1 [20] 3/-1/-1->2->1 [21] 3/-1/-1->2->1 [22] 3/-1/-1->2->1 [23] 3/-1/-1->2->1
+ip-26-0-166-15:127290:127833 [0] NCCL INFO Channel 01/24 :    0   1   2   3
+ip-26-0-166-15:127294:127835 [4] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-166-15:127290:127833 [0] NCCL INFO Channel 02/24 :    0   1   2   3
+ip-26-0-166-15:127290:127833 [0] NCCL INFO Channel 03/24 :    0   1   2   3
+ip-26-0-166-15:127290:127833 [0] NCCL INFO Channel 04/24 :    0   1   2   3
+ip-26-0-166-15:127290:127833 [0] NCCL INFO Channel 05/24 :    0   1   2   3
+ip-26-0-166-15:127290:127833 [0] NCCL INFO Channel 06/24 :    0   1   2   3
+ip-26-0-166-15:127296:127834 [6] NCCL INFO Trees [0] -1/-1/-1->3->2 [1] -1/-1/-1->3->2 [2] -1/-1/-1->3->2 [3] -1/-1/-1->3->2 [4] -1/-1/-1->3->2 [5] -1/-1/-1->3->2 [6] -1/-1/-1->3->2 [7] -1/-1/-1->3->2 [8] -1/-1/-1->3->2 [9] -1/-1/-1->3->2 [10] -1/-1/-1->3->2 [11] -1/-1/-1->3->2 [12] -1/-1/-1->3->2 [13] -1/-1/-1->3->2 [14] -1/-1/-1->3->2 [15] -1/-1/-1->3->2 [16] -1/-1/-1->3->2 [17] -1/-1/-1->3->2 [18] -1/-1/-1->3->2 [19] -1/-1/-1->3->2 [20] -1/-1/-1->3->2 [21] -1/-1/-1->3->2 [22] -1/-1/-1->3->2 [23] -1/-1/-1->3->2
+ip-26-0-166-15:127290:127833 [0] NCCL INFO Channel 07/24 :    0   1   2   3
+ip-26-0-166-15:127296:127834 [6] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-166-15:127290:127833 [0] NCCL INFO Channel 08/24 :    0   1   2   3
+ip-26-0-166-15:127290:127833 [0] NCCL INFO Channel 09/24 :    0   1   2   3
+ip-26-0-166-15:127292:127836 [2] NCCL INFO Trees [0] 2/-1/-1->1->0 [1] 2/-1/-1->1->0 [2] 2/-1/-1->1->0 [3] 2/-1/-1->1->0 [4] 2/-1/-1->1->0 [5] 2/-1/-1->1->0 [6] 2/-1/-1->1->0 [7] 2/-1/-1->1->0 [8] 2/-1/-1->1->0 [9] 2/-1/-1->1->0 [10] 2/-1/-1->1->0 [11] 2/-1/-1->1->0 [12] 2/-1/-1->1->0 [13] 2/-1/-1->1->0 [14] 2/-1/-1->1->0 [15] 2/-1/-1->1->0 [16] 2/-1/-1->1->0 [17] 2/-1/-1->1->0 [18] 2/-1/-1->1->0 [19] 2/-1/-1->1->0 [20] 2/-1/-1->1->0 [21] 2/-1/-1->1->0 [22] 2/-1/-1->1->0 [23] 2/-1/-1->1->0
+ip-26-0-166-15:127290:127833 [0] NCCL INFO Channel 10/24 :    0   1   2   3
+ip-26-0-166-15:127290:127833 [0] NCCL INFO Channel 11/24 :    0   1   2   3
+ip-26-0-166-15:127292:127836 [2] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-167-9:1039538:1040099 [1] NCCL INFO Channel 19/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-165-213:9632:10176 [7] NCCL INFO Channel 00/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-166-15:127290:127833 [0] NCCL INFO Channel 12/24 :    0   1   2   3
+ip-26-0-166-15:127290:127833 [0] NCCL INFO Channel 13/24 :    0   1   2   3
+ip-26-0-166-15:127290:127833 [0] NCCL INFO Channel 14/24 :    0   1   2   3
+ip-26-0-166-15:127290:127833 [0] NCCL INFO Channel 15/24 :    0   1   2   3
+ip-26-0-166-15:127290:127833 [0] NCCL INFO Channel 16/24 :    0   1   2   3
+ip-26-0-166-15:127290:127833 [0] NCCL INFO Channel 17/24 :    0   1   2   3
+ip-26-0-166-15:127290:127833 [0] NCCL INFO Channel 18/24 :    0   1   2   3
+ip-26-0-166-15:127290:127833 [0] NCCL INFO Channel 19/24 :    0   1   2   3
+ip-26-0-166-15:127290:127833 [0] NCCL INFO Channel 20/24 :    0   1   2   3
+ip-26-0-166-15:127290:127833 [0] NCCL INFO Channel 21/24 :    0   1   2   3
+ip-26-0-166-15:127290:127833 [0] NCCL INFO Channel 22/24 :    0   1   2   3
+ip-26-0-166-15:127290:127833 [0] NCCL INFO Channel 23/24 :    0   1   2   3
+ip-26-0-166-15:127290:127833 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] 1/-1/-1->0->-1 [7] 1/-1/-1->0->-1 [8] 1/-1/-1->0->-1 [9] 1/-1/-1->0->-1 [10] 1/-1/-1->0->-1 [11] 1/-1/-1->0->-1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] 1/-1/-1->0->-1 [19] 1/-1/-1->0->-1 [20] 1/-1/-1->0->-1 [21] 1/-1/-1->0->-1 [22] 1/-1/-1->0->-1 [23] 1/-1/-1->0->-1
+ip-26-0-166-15:127290:127833 [0] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-165-213:9628:10177 [3] NCCL INFO Channel 00/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-165-213:9626:10173 [1] NCCL INFO Channel 01/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-165-213:9630:10175 [5] NCCL INFO Channel 00/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-167-9:1039544:1040101 [7] NCCL INFO Channel 20/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-167-9:1039538:1040099 [1] NCCL INFO Channel 20/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-167-9:1039541:1040097 [4] NCCL INFO Channel 00/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-167-9:1039544:1040101 [7] NCCL INFO Channel 21/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-167-9:1039538:1040099 [1] NCCL INFO Channel 21/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-165-213:9628:10177 [3] NCCL INFO Channel 01/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-167-9:1039541:1040097 [4] NCCL INFO Channel 01/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-167-9:1039543:1040098 [6] NCCL INFO Channel 00/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-167-9:1039544:1040101 [7] NCCL INFO Channel 22/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-167-9:1039538:1040099 [1] NCCL INFO Channel 22/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-165-213:9626:10173 [1] NCCL INFO Channel 02/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-165-213:9628:10177 [3] NCCL INFO Channel 02/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-167-9:1039541:1040097 [4] NCCL INFO Channel 02/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-165-213:9626:10173 [1] NCCL INFO Channel 03/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-165-213:9628:10177 [3] NCCL INFO Channel 03/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-167-9:1039543:1040098 [6] NCCL INFO Channel 01/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-167-9:1039544:1040101 [7] NCCL INFO Channel 23/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-167-9:1039538:1040099 [1] NCCL INFO Channel 23/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-166-244:897241:897784 [1] NCCL INFO Channel 00/24 :    0   1   2   3
+ip-26-0-166-244:897241:897784 [1] NCCL INFO Channel 01/24 :    0   1   2   3
+ip-26-0-166-244:897245:897787 [5] NCCL INFO Trees [0] 3/-1/-1->2->1 [1] 3/-1/-1->2->1 [2] 3/-1/-1->2->1 [3] 3/-1/-1->2->1 [4] 3/-1/-1->2->1 [5] 3/-1/-1->2->1 [6] 3/-1/-1->2->1 [7] 3/-1/-1->2->1 [8] 3/-1/-1->2->1 [9] 3/-1/-1->2->1 [10] 3/-1/-1->2->1 [11] 3/-1/-1->2->1 [12] 3/-1/-1->2->1 [13] 3/-1/-1->2->1 [14] 3/-1/-1->2->1 [15] 3/-1/-1->2->1 [16] 3/-1/-1->2->1 [17] 3/-1/-1->2->1 [18] 3/-1/-1->2->1 [19] 3/-1/-1->2->1 [20] 3/-1/-1->2->1 [21] 3/-1/-1->2->1 [22] 3/-1/-1->2->1 [23] 3/-1/-1->2->1
+ip-26-0-166-244:897241:897784 [1] NCCL INFO Channel 02/24 :    0   1   2   3
+ip-26-0-166-244:897245:897787 [5] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-166-244:897241:897784 [1] NCCL INFO Channel 03/24 :    0   1   2   3
+ip-26-0-166-214:624948:625500 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-166-244:897241:897784 [1] NCCL INFO Channel 04/24 :    0   1   2   3
+ip-26-0-166-244:897241:897784 [1] NCCL INFO Channel 05/24 :    0   1   2   3
+ip-26-0-166-244:897243:897790 [3] NCCL INFO Trees [0] 2/-1/-1->1->0 [1] 2/-1/-1->1->0 [2] 2/-1/-1->1->0 [3] 2/-1/-1->1->0 [4] 2/-1/-1->1->0 [5] 2/-1/-1->1->0 [6] 2/-1/-1->1->0 [7] 2/-1/-1->1->0 [8] 2/-1/-1->1->0 [9] 2/-1/-1->1->0 [10] 2/-1/-1->1->0 [11] 2/-1/-1->1->0 [12] 2/-1/-1->1->0 [13] 2/-1/-1->1->0 [14] 2/-1/-1->1->0 [15] 2/-1/-1->1->0 [16] 2/-1/-1->1->0 [17] 2/-1/-1->1->0 [18] 2/-1/-1->1->0 [19] 2/-1/-1->1->0 [20] 2/-1/-1->1->0 [21] 2/-1/-1->1->0 [22] 2/-1/-1->1->0 [23] 2/-1/-1->1->0
+ip-26-0-166-244:897241:897784 [1] NCCL INFO Channel 06/24 :    0   1   2   3
+ip-26-0-166-244:897247:897788 [7] NCCL INFO Trees [0] -1/-1/-1->3->2 [1] -1/-1/-1->3->2 [2] -1/-1/-1->3->2 [3] -1/-1/-1->3->2 [4] -1/-1/-1->3->2 [5] -1/-1/-1->3->2 [6] -1/-1/-1->3->2 [7] -1/-1/-1->3->2 [8] -1/-1/-1->3->2 [9] -1/-1/-1->3->2 [10] -1/-1/-1->3->2 [11] -1/-1/-1->3->2 [12] -1/-1/-1->3->2 [13] -1/-1/-1->3->2 [14] -1/-1/-1->3->2 [15] -1/-1/-1->3->2 [16] -1/-1/-1->3->2 [17] -1/-1/-1->3->2 [18] -1/-1/-1->3->2 [19] -1/-1/-1->3->2 [20] -1/-1/-1->3->2 [21] -1/-1/-1->3->2 [22] -1/-1/-1->3->2 [23] -1/-1/-1->3->2
+ip-26-0-166-244:897241:897784 [1] NCCL INFO Channel 07/24 :    0   1   2   3
+ip-26-0-166-244:897243:897790 [3] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-166-244:897247:897788 [7] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-166-244:897241:897784 [1] NCCL INFO Channel 08/24 :    0   1   2   3
+ip-26-0-166-244:897241:897784 [1] NCCL INFO Channel 09/24 :    0   1   2   3
+ip-26-0-166-244:897241:897784 [1] NCCL INFO Channel 10/24 :    0   1   2   3
+ip-26-0-166-244:897241:897784 [1] NCCL INFO Channel 11/24 :    0   1   2   3
+ip-26-0-166-244:897241:897784 [1] NCCL INFO Channel 12/24 :    0   1   2   3
+ip-26-0-166-244:897241:897784 [1] NCCL INFO Channel 13/24 :    0   1   2   3
+ip-26-0-166-244:897241:897784 [1] NCCL INFO Channel 14/24 :    0   1   2   3
+ip-26-0-166-244:897241:897784 [1] NCCL INFO Channel 15/24 :    0   1   2   3
+ip-26-0-166-244:897241:897784 [1] NCCL INFO Channel 16/24 :    0   1   2   3
+ip-26-0-166-244:897241:897784 [1] NCCL INFO Channel 17/24 :    0   1   2   3
+ip-26-0-166-244:897241:897784 [1] NCCL INFO Channel 18/24 :    0   1   2   3
+ip-26-0-166-244:897241:897784 [1] NCCL INFO Channel 19/24 :    0   1   2   3
+ip-26-0-166-244:897241:897784 [1] NCCL INFO Channel 20/24 :    0   1   2   3
+ip-26-0-166-244:897241:897784 [1] NCCL INFO Channel 21/24 :    0   1   2   3
+ip-26-0-166-244:897241:897784 [1] NCCL INFO Channel 22/24 :    0   1   2   3
+ip-26-0-166-244:897241:897784 [1] NCCL INFO Channel 23/24 :    0   1   2   3
+ip-26-0-166-244:897241:897784 [1] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] 1/-1/-1->0->-1 [7] 1/-1/-1->0->-1 [8] 1/-1/-1->0->-1 [9] 1/-1/-1->0->-1 [10] 1/-1/-1->0->-1 [11] 1/-1/-1->0->-1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] 1/-1/-1->0->-1 [19] 1/-1/-1->0->-1 [20] 1/-1/-1->0->-1 [21] 1/-1/-1->0->-1 [22] 1/-1/-1->0->-1 [23] 1/-1/-1->0->-1
+ip-26-0-166-244:897241:897784 [1] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-166-214:624948:625500 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-167-9:1039541:1040097 [4] NCCL INFO Channel 03/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-165-213:9632:10176 [7] NCCL INFO Channel 01/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-167-9:1039543:1040098 [6] NCCL INFO Channel 02/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-167-9:1039537:1040095 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-165-213:9630:10175 [5] NCCL INFO Channel 01/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-165-213:9632:10176 [7] NCCL INFO Channel 02/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-167-9:1039541:1040097 [4] NCCL INFO Channel 04/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-165-213:9626:10173 [1] NCCL INFO Channel 04/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-165-213:9630:10175 [5] NCCL INFO Channel 02/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-165-213:9628:10177 [3] NCCL INFO Channel 04/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-165-213:9626:10173 [1] NCCL INFO Channel 05/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-165-213:9630:10175 [5] NCCL INFO Channel 03/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-165-213:9628:10177 [3] NCCL INFO Channel 05/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-165-213:9626:10173 [1] NCCL INFO Channel 06/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-165-213:9630:10175 [5] NCCL INFO Channel 04/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-165-213:9628:10177 [3] NCCL INFO Channel 06/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-165-213:9626:10173 [1] NCCL INFO Channel 07/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-165-213:9630:10175 [5] NCCL INFO Channel 05/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-165-213:9628:10177 [3] NCCL INFO Channel 07/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-165-213:9626:10173 [1] NCCL INFO Channel 08/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-165-213:9630:10175 [5] NCCL INFO Channel 06/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-165-213:9628:10177 [3] NCCL INFO Channel 08/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-165-213:9626:10173 [1] NCCL INFO Channel 09/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-165-213:9630:10175 [5] NCCL INFO Channel 07/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-167-9:1039543:1040098 [6] NCCL INFO Channel 03/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-165-213:9628:10177 [3] NCCL INFO Channel 09/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-165-213:9626:10173 [1] NCCL INFO Channel 10/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-165-213:9630:10175 [5] NCCL INFO Channel 08/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-165-213:9632:10176 [7] NCCL INFO Channel 03/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-165-213:9628:10177 [3] NCCL INFO Channel 10/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-165-213:9626:10173 [1] NCCL INFO Channel 11/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-165-213:9630:10175 [5] NCCL INFO Channel 09/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-166-36:104431:104979 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-166-36:104431:104979 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-165-213:9632:10176 [7] NCCL INFO Channel 04/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-167-9:1039537:1040095 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-165-213:9628:10177 [3] NCCL INFO Channel 11/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-167-9:1039541:1040097 [4] NCCL INFO Channel 05/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-165-213:9628:10177 [3] NCCL INFO Channel 12/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-165-213:9630:10175 [5] NCCL INFO Channel 10/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-165-213:9630:10175 [5] NCCL INFO Channel 11/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-165-213:9628:10177 [3] NCCL INFO Channel 13/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-165-213:9630:10175 [5] NCCL INFO Channel 12/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-165-213:9628:10177 [3] NCCL INFO Channel 14/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-165-213:9630:10175 [5] NCCL INFO Channel 13/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-165-213:9626:10173 [1] NCCL INFO Channel 12/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-165-213:9628:10177 [3] NCCL INFO Channel 15/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-165-213:9632:10176 [7] NCCL INFO Channel 05/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-165-213:9626:10173 [1] NCCL INFO Channel 13/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-166-15:127297:127840 [7] NCCL INFO Channel 00/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-166-15:127293:127841 [3] NCCL INFO Channel 00/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-166-15:127297:127840 [7] NCCL INFO Channel 01/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-165-213:9628:10177 [3] NCCL INFO Channel 16/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-165-213:9630:10175 [5] NCCL INFO Channel 14/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-166-15:127293:127841 [3] NCCL INFO Channel 01/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-165-213:9632:10176 [7] NCCL INFO Channel 06/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-165-213:9628:10177 [3] NCCL INFO Channel 17/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-166-15:127297:127840 [7] NCCL INFO Channel 02/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-167-9:1039543:1040098 [6] NCCL INFO Channel 04/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-167-9:1039537:1040095 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-166-244:897244:897789 [4] NCCL INFO Channel 00/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-166-15:127293:127841 [3] NCCL INFO Channel 02/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-166-15:127297:127840 [7] NCCL INFO Channel 03/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-166-244:897244:897789 [4] NCCL INFO Channel 01/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-165-213:9626:10173 [1] NCCL INFO Channel 14/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-166-15:127293:127841 [3] NCCL INFO Channel 03/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-165-213:9626:10173 [1] NCCL INFO Channel 15/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-167-9:1039541:1040097 [4] NCCL INFO Channel 06/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-166-15:127297:127840 [7] NCCL INFO Channel 04/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-166-244:897244:897789 [4] NCCL INFO Channel 02/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-167-9:1039539:1040100 [2] NCCL INFO Channel 00/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-165-213:9626:10173 [1] NCCL INFO Channel 16/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-165-213:9630:10175 [5] NCCL INFO Channel 15/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-166-15:127293:127841 [3] NCCL INFO Channel 04/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-166-15:127297:127840 [7] NCCL INFO Channel 05/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-165-213:9626:10173 [1] NCCL INFO Channel 17/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-165-213:9630:10175 [5] NCCL INFO Channel 16/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-166-15:127296:127834 [6] NCCL INFO Channel 00/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127293:127841 [3] NCCL INFO Channel 05/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-166-244:897242:897792 [2] NCCL INFO Channel 00/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-166-244:897244:897789 [4] NCCL INFO Channel 03/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-166-15:127297:127840 [7] NCCL INFO Channel 06/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-166-244:897242:897792 [2] NCCL INFO Channel 01/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-165-213:9626:10173 [1] NCCL INFO Channel 18/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-165-213:9630:10175 [5] NCCL INFO Channel 17/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-165-213:9632:10176 [7] NCCL INFO Channel 07/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-166-15:127295:127839 [5] NCCL INFO Channel 00/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-166-244:897244:897789 [4] NCCL INFO Channel 04/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-165-213:9628:10177 [3] NCCL INFO Channel 18/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-166-244:897242:897792 [2] NCCL INFO Channel 02/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-166-15:127296:127834 [6] NCCL INFO Channel 01/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127293:127841 [3] NCCL INFO Channel 06/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-166-15:127292:127836 [2] NCCL INFO Channel 00/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-165-213:9626:10173 [1] NCCL INFO Channel 19/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-166-15:127297:127840 [7] NCCL INFO Channel 07/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-166-125:762816:763376 [1] NCCL INFO Setting affinity for GPU 1 to 0fff,ffffffff
+ip-26-0-165-213:9630:10175 [5] NCCL INFO Channel 18/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-166-125:762816:763376 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-166-15:127290:127833 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-166-15:127291:127838 [1] NCCL INFO Channel 00/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-165-213:9632:10176 [7] NCCL INFO Channel 08/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-166-15:127295:127839 [5] NCCL INFO Channel 01/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-165-213:9628:10177 [3] NCCL INFO Channel 19/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-166-244:897244:897789 [4] NCCL INFO Channel 05/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-166-15:127296:127834 [6] NCCL INFO Channel 02/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127293:127841 [3] NCCL INFO Channel 07/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-166-15:127292:127836 [2] NCCL INFO Channel 01/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-167-9:1039543:1040098 [6] NCCL INFO Channel 05/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127297:127840 [7] NCCL INFO Channel 08/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-166-15:127290:127833 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-166-15:127291:127838 [1] NCCL INFO Channel 01/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-166-15:127295:127839 [5] NCCL INFO Channel 02/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-165-213:9630:10175 [5] NCCL INFO Channel 19/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-166-15:127296:127834 [6] NCCL INFO Channel 03/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-166-244:897242:897792 [2] NCCL INFO Channel 03/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-166-15:127293:127841 [3] NCCL INFO Channel 08/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-166-15:127292:127836 [2] NCCL INFO Channel 02/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-165-213:9630:10175 [5] NCCL INFO Channel 20/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-166-125:762822:763377 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,fff00000,00000000
+ip-26-0-166-15:127297:127840 [7] NCCL INFO Channel 09/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-166-125:762822:763377 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-166-15:127290:127833 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-166-15:127291:127838 [1] NCCL INFO Channel 02/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-166-15:127295:127839 [5] NCCL INFO Channel 03/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-165-213:9630:10175 [5] NCCL INFO Channel 21/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-166-244:897244:897789 [4] NCCL INFO Channel 06/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-165-213:9627:10174 [2] NCCL INFO Channel 00/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-166-15:127296:127834 [6] NCCL INFO Channel 04/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-165-213:9626:10173 [1] NCCL INFO Channel 20/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-166-15:127293:127841 [3] NCCL INFO Channel 09/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-165-213:9630:10175 [5] NCCL INFO Channel 22/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-166-15:127292:127836 [2] NCCL INFO Channel 03/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-166-244:897242:897792 [2] NCCL INFO Channel 04/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-166-244:897247:897788 [7] NCCL INFO Channel 00/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-166-244:897244:897789 [4] NCCL INFO Channel 07/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-166-244:897246:897791 [6] NCCL INFO Channel 00/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624945:625495 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-166-214:624945:625495 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-165-213:9629:10171 [4] NCCL INFO Channel 00/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-166-15:127297:127840 [7] NCCL INFO Channel 10/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-166-15:127290:127833 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-165-213:9627:10174 [2] NCCL INFO Channel 01/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-166-15:127291:127838 [1] NCCL INFO Channel 03/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-166-244:897242:897792 [2] NCCL INFO Channel 05/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-166-244:897247:897788 [7] NCCL INFO Channel 01/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-165-213:9626:10173 [1] NCCL INFO Channel 21/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-166-15:127295:127839 [5] NCCL INFO Channel 04/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-165-213:9630:10175 [5] NCCL INFO Channel 23/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-166-244:897244:897789 [4] NCCL INFO Channel 08/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-166-244:897246:897791 [6] NCCL INFO Channel 01/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127296:127834 [6] NCCL INFO Channel 05/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127293:127841 [3] NCCL INFO Channel 10/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-166-244:897242:897792 [2] NCCL INFO Channel 06/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-166-15:127292:127836 [2] NCCL INFO Channel 04/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-165-213:9629:10171 [4] NCCL INFO Channel 01/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-166-244:897247:897788 [7] NCCL INFO Channel 02/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-165-213:9627:10174 [2] NCCL INFO Channel 02/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-165-213:9626:10173 [1] NCCL INFO Channel 22/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-166-244:897244:897789 [4] NCCL INFO Channel 09/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-166-244:897246:897791 [6] NCCL INFO Channel 02/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127297:127840 [7] NCCL INFO Channel 11/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-166-15:127290:127833 [0] NCCL INFO Channel 04/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-166-15:127291:127838 [1] NCCL INFO Channel 04/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-166-15:127295:127839 [5] NCCL INFO Channel 05/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-165-213:9629:10171 [4] NCCL INFO Channel 02/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-165-213:9626:10173 [1] NCCL INFO Channel 23/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-165-213:9627:10174 [2] NCCL INFO Channel 03/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-165-213:9632:10176 [7] NCCL INFO Channel 09/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-166-15:127296:127834 [6] NCCL INFO Channel 06/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127293:127841 [3] NCCL INFO Channel 11/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-166-15:127292:127836 [2] NCCL INFO Channel 05/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-166-244:897242:897792 [2] NCCL INFO Channel 07/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-165-213:9629:10171 [4] NCCL INFO Channel 03/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-165-213:9625:10169 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-165-213:9627:10174 [2] NCCL INFO Channel 04/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-165-213:9632:10176 [7] NCCL INFO Channel 10/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-166-15:127297:127840 [7] NCCL INFO Channel 12/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-166-15:127291:127838 [1] NCCL INFO Channel 05/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-166-15:127290:127833 [0] NCCL INFO Channel 05/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-165-213:9628:10177 [3] NCCL INFO Channel 20/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-166-15:127295:127839 [5] NCCL INFO Channel 06/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-166-15:127294:127835 [4] NCCL INFO Channel 00/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-166-244:897242:897792 [2] NCCL INFO Channel 08/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-166-244:897245:897787 [5] NCCL INFO Channel 00/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-166-244:897240:897786 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-167-9:1039537:1040095 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-166-15:127296:127834 [6] NCCL INFO Channel 07/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-165-213:9629:10171 [4] NCCL INFO Channel 04/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-165-213:9625:10169 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-165-213:9627:10174 [2] NCCL INFO Channel 05/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-166-244:897244:897789 [4] NCCL INFO Channel 10/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-166-15:127293:127841 [3] NCCL INFO Channel 12/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-166-15:127292:127836 [2] NCCL INFO Channel 06/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-165-213:9632:10176 [7] NCCL INFO Channel 11/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-166-244:897242:897792 [2] NCCL INFO Channel 09/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-166-244:897245:897787 [5] NCCL INFO Channel 01/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-165-213:9628:10177 [3] NCCL INFO Channel 21/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-166-244:897240:897786 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-167-9:1039543:1040098 [6] NCCL INFO Channel 06/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-166-244:897247:897788 [7] NCCL INFO Channel 03/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-166-15:127291:127838 [1] NCCL INFO Channel 06/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-166-15:127290:127833 [0] NCCL INFO Channel 06/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-166-15:127297:127840 [7] NCCL INFO Channel 13/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-166-15:127295:127839 [5] NCCL INFO Channel 07/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-166-15:127294:127835 [4] NCCL INFO Channel 01/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-165-213:9625:10169 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-165-213:9629:10171 [4] NCCL INFO Channel 05/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-165-213:9627:10174 [2] NCCL INFO Channel 06/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-165-213:9631:10172 [6] NCCL INFO Channel 00/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-166-244:897244:897789 [4] NCCL INFO Channel 11/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-165-213:9632:10176 [7] NCCL INFO Channel 12/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-165-213:9628:10177 [3] NCCL INFO Channel 22/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-166-244:897242:897792 [2] NCCL INFO Channel 10/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-166-244:897245:897787 [5] NCCL INFO Channel 02/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-166-244:897243:897790 [3] NCCL INFO Channel 00/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-166-244:897240:897786 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-166-244:897247:897788 [7] NCCL INFO Channel 04/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-167-9:1039537:1040095 [0] NCCL INFO Channel 04/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-166-15:127297:127840 [7] NCCL INFO Channel 14/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-165-213:9625:10169 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-165-213:9629:10171 [4] NCCL INFO Channel 06/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-165-213:9631:10172 [6] NCCL INFO Channel 01/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-165-213:9627:10174 [2] NCCL INFO Channel 07/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-166-15:127292:127836 [2] NCCL INFO Channel 07/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-166-15:127295:127839 [5] NCCL INFO Channel 08/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-166-15:127296:127834 [6] NCCL INFO Channel 08/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-166-244:897244:897789 [4] NCCL INFO Channel 12/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-165-213:9632:10176 [7] NCCL INFO Channel 13/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-165-213:9628:10177 [3] NCCL INFO Channel 23/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-166-15:127297:127840 [7] NCCL INFO Channel 15/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-167-9:1039543:1040098 [6] NCCL INFO Channel 07/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-165-213:9625:10169 [0] NCCL INFO Channel 04/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-165-213:9629:10171 [4] NCCL INFO Channel 07/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-165-213:9631:10172 [6] NCCL INFO Channel 02/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-165-213:9627:10174 [2] NCCL INFO Channel 08/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-166-15:127293:127841 [3] NCCL INFO Channel 13/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-166-15:127295:127839 [5] NCCL INFO Channel 09/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-165-213:9632:10176 [7] NCCL INFO Channel 14/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-166-244:897240:897786 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-166-244:897246:897791 [6] NCCL INFO Channel 03/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-166-244:897243:897790 [3] NCCL INFO Channel 01/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-166-15:127294:127835 [4] NCCL INFO Channel 02/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-166-15:127293:127841 [3] NCCL INFO Channel 14/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-166-15:127295:127839 [5] NCCL INFO Channel 10/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-166-244:897240:897786 [0] NCCL INFO Channel 04/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-166-244:897246:897791 [6] NCCL INFO Channel 04/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624949:625496 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-166-244:897243:897790 [3] NCCL INFO Channel 02/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-166-214:624949:625496 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-165-213:9625:10169 [0] NCCL INFO Channel 05/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-166-15:127290:127833 [0] NCCL INFO Channel 07/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-165-213:9629:10171 [4] NCCL INFO Channel 08/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-165-213:9631:10172 [6] NCCL INFO Channel 03/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-167-9:1039539:1040100 [2] NCCL INFO Channel 01/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-167-9:1039537:1040095 [0] NCCL INFO Channel 05/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-167-9:1039541:1040097 [4] NCCL INFO Channel 07/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-166-244:897241:897784 [1] NCCL INFO Channel 00/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-166-15:127294:127835 [4] NCCL INFO Channel 03/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-166-15:127293:127841 [3] NCCL INFO Channel 15/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-166-15:127295:127839 [5] NCCL INFO Channel 11/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-166-244:897247:897788 [7] NCCL INFO Channel 05/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-162-46:702880:703466 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-166-15:127290:127833 [0] NCCL INFO Channel 08/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-166-244:897240:897786 [0] NCCL INFO Channel 05/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-166-244:897246:897791 [6] NCCL INFO Channel 05/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-166-244:897243:897790 [3] NCCL INFO Channel 03/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-165-213:9627:10174 [2] NCCL INFO Channel 09/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-162-46:702880:703466 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-166-15:127291:127838 [1] NCCL INFO Channel 07/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-165-213:9632:10176 [7] NCCL INFO Channel 15/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-166-15:127294:127835 [4] NCCL INFO Channel 04/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-166-244:897241:897784 [1] NCCL INFO Channel 01/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-166-15:127293:127841 [3] NCCL INFO Channel 16/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-166-15:127295:127839 [5] NCCL INFO Channel 12/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-166-15:127290:127833 [0] NCCL INFO Channel 09/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-166-244:897247:897788 [7] NCCL INFO Channel 06/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-166-244:897242:897792 [2] NCCL INFO Channel 11/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-166-244:897245:897787 [5] NCCL INFO Channel 03/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-166-244:897240:897786 [0] NCCL INFO Channel 06/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-166-244:897246:897791 [6] NCCL INFO Channel 06/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127292:127836 [2] NCCL INFO Channel 08/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-166-244:897243:897790 [3] NCCL INFO Channel 04/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-166-15:127296:127834 [6] NCCL INFO Channel 09/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-166-244:897244:897789 [4] NCCL INFO Channel 13/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-166-15:127297:127840 [7] NCCL INFO Channel 16/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-166-15:127294:127835 [4] NCCL INFO Channel 05/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-166-15:127290:127833 [0] NCCL INFO Channel 10/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-166-15:127295:127839 [5] NCCL INFO Channel 13/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-166-15:127293:127841 [3] NCCL INFO Channel 17/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-166-244:897241:897784 [1] NCCL INFO Channel 02/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-166-244:897247:897788 [7] NCCL INFO Channel 07/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-166-244:897242:897792 [2] NCCL INFO Channel 12/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-166-244:897245:897787 [5] NCCL INFO Channel 04/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-166-244:897240:897786 [0] NCCL INFO Channel 07/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-166-15:127292:127836 [2] NCCL INFO Channel 09/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-166-15:127296:127834 [6] NCCL INFO Channel 10/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-166-244:897246:897791 [6] NCCL INFO Channel 07/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-165-213:9625:10169 [0] NCCL INFO Channel 06/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-165-213:9631:10172 [6] NCCL INFO Channel 04/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-165-213:9629:10171 [4] NCCL INFO Channel 09/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-165-213:9627:10174 [2] NCCL INFO Channel 10/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-166-244:897243:897790 [3] NCCL INFO Channel 05/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-165-213:9632:10176 [7] NCCL INFO Channel 16/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-166-15:127294:127835 [4] NCCL INFO Channel 06/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-166-15:127297:127840 [7] NCCL INFO Channel 17/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-166-15:127295:127839 [5] NCCL INFO Channel 14/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-166-15:127290:127833 [0] NCCL INFO Channel 11/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-162-46:702884:703470 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-162-46:702884:703470 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-166-244:897244:897789 [4] NCCL INFO Channel 14/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-166-244:897241:897784 [1] NCCL INFO Channel 03/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-166-244:897247:897788 [7] NCCL INFO Channel 08/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-166-244:897242:897792 [2] NCCL INFO Channel 13/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-166-244:897245:897787 [5] NCCL INFO Channel 05/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-167-9:1039543:1040098 [6] NCCL INFO Channel 08/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-166-244:897240:897786 [0] NCCL INFO Channel 08/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-166-244:897246:897791 [6] NCCL INFO Channel 08/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-166-244:897243:897790 [3] NCCL INFO Channel 06/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-165-213:9625:10169 [0] NCCL INFO Channel 07/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-166-244:897244:897789 [4] NCCL INFO Channel 15/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-166-15:127296:127834 [6] NCCL INFO Channel 11/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-166-244:897241:897784 [1] NCCL INFO Channel 04/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-166-244:897247:897788 [7] NCCL INFO Channel 09/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-166-244:897242:897792 [2] NCCL INFO Channel 14/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-166-15:127291:127838 [1] NCCL INFO Channel 08/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-166-244:897245:897787 [5] NCCL INFO Channel 06/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-166-244:897240:897786 [0] NCCL INFO Channel 09/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-166-15:127295:127839 [5] NCCL INFO Channel 15/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-166-15:127296:127834 [6] NCCL INFO Channel 12/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127291:127838 [1] NCCL INFO Channel 09/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-165-213:9629:10171 [4] NCCL INFO Channel 10/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-165-213:9631:10172 [6] NCCL INFO Channel 05/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-166-244:897243:897790 [3] NCCL INFO Channel 07/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-165-213:9627:10174 [2] NCCL INFO Channel 11/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-165-213:9632:10176 [7] NCCL INFO Channel 17/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-166-244:897246:897791 [6] NCCL INFO Channel 09/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127297:127840 [7] NCCL INFO Channel 18/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-166-15:127296:127834 [6] NCCL INFO Channel 13/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127295:127839 [5] NCCL INFO Channel 16/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-166-15:127293:127841 [3] NCCL INFO Channel 18/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-166-15:127291:127838 [1] NCCL INFO Channel 10/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-166-15:127292:127836 [2] NCCL INFO Channel 10/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-167-9:1039539:1040100 [2] NCCL INFO Channel 02/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-166-244:897241:897784 [1] NCCL INFO Channel 05/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-166-244:897247:897788 [7] NCCL INFO Channel 10/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-166-15:127297:127840 [7] NCCL INFO Channel 19/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-166-244:897240:897786 [0] NCCL INFO Channel 10/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-166-244:897243:897790 [3] NCCL INFO Channel 08/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-166-15:127296:127834 [6] NCCL INFO Channel 14/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127295:127839 [5] NCCL INFO Channel 17/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-166-15:127294:127835 [4] NCCL INFO Channel 07/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-166-15:127291:127838 [1] NCCL INFO Channel 11/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-166-15:127293:127841 [3] NCCL INFO Channel 19/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-166-15:127290:127833 [0] NCCL INFO Channel 12/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-167-9:1039541:1040097 [4] NCCL INFO Channel 08/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-166-15:127292:127836 [2] NCCL INFO Channel 11/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-166-244:897241:897784 [1] NCCL INFO Channel 06/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-166-244:897247:897788 [7] NCCL INFO Channel 11/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-166-244:897240:897786 [0] NCCL INFO Channel 11/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-166-244:897243:897790 [3] NCCL INFO Channel 09/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-165-213:9625:10169 [0] NCCL INFO Channel 08/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-162-46:702887:703467 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-162-46:702887:703467 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-165-213:9629:10171 [4] NCCL INFO Channel 11/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-165-213:9631:10172 [6] NCCL INFO Channel 06/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-165-213:9627:10174 [2] NCCL INFO Channel 12/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-166-244:897241:897784 [1] NCCL INFO Channel 07/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-165-213:9632:10176 [7] NCCL INFO Channel 18/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-166-244:897247:897788 [7] NCCL INFO Channel 12/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-166-244:897240:897786 [0] NCCL INFO Channel 12/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-166-244:897243:897790 [3] NCCL INFO Channel 10/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-166-15:127294:127835 [4] NCCL INFO Channel 08/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-166-15:127291:127838 [1] NCCL INFO Channel 12/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-166-244:897241:897784 [1] NCCL INFO Channel 08/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-166-244:897247:897788 [7] NCCL INFO Channel 13/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-166-244:897240:897786 [0] NCCL INFO Channel 13/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-166-244:897243:897790 [3] NCCL INFO Channel 11/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-166-244:897241:897784 [1] NCCL INFO Channel 09/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-166-244:897247:897788 [7] NCCL INFO Channel 14/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-166-244:897240:897786 [0] NCCL INFO Channel 14/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-166-244:897243:897790 [3] NCCL INFO Channel 12/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-166-244:897242:897792 [2] NCCL INFO Channel 15/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-166-15:127291:127838 [1] NCCL INFO Channel 13/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-166-15:127290:127833 [0] NCCL INFO Channel 13/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-166-15:127296:127834 [6] NCCL INFO Channel 15/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127295:127839 [5] NCCL INFO Channel 18/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-166-244:897241:897784 [1] NCCL INFO Channel 10/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-166-244:897247:897788 [7] NCCL INFO Channel 15/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-166-244:897240:897786 [0] NCCL INFO Channel 15/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-166-244:897243:897790 [3] NCCL INFO Channel 13/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-166-244:897242:897792 [2] NCCL INFO Channel 16/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-166-15:127290:127833 [0] NCCL INFO Channel 14/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-166-15:127291:127838 [1] NCCL INFO Channel 14/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-167-9:1039543:1040098 [6] NCCL INFO Channel 09/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-165-213:9625:10169 [0] NCCL INFO Channel 09/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-165-213:9631:10172 [6] NCCL INFO Channel 07/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-165-213:9629:10171 [4] NCCL INFO Channel 12/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-165-213:9627:10174 [2] NCCL INFO Channel 13/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-166-244:897245:897787 [5] NCCL INFO Channel 07/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-165-213:9632:10176 [7] NCCL INFO Channel 19/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-166-15:127296:127834 [6] NCCL INFO Channel 16/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127295:127839 [5] NCCL INFO Channel 19/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-166-15:127290:127833 [0] NCCL INFO Channel 15/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-166-15:127291:127838 [1] NCCL INFO Channel 15/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-166-244:897241:897784 [1] NCCL INFO Channel 11/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-166-244:897247:897788 [7] NCCL INFO Channel 16/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-166-244:897243:897790 [3] NCCL INFO Channel 14/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-166-244:897244:897789 [4] NCCL INFO Channel 16/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-166-244:897240:897786 [0] NCCL INFO Channel 16/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-166-244:897242:897792 [2] NCCL INFO Channel 17/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-167-9:1039537:1040095 [0] NCCL INFO Channel 06/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-166-15:127296:127834 [6] NCCL INFO Channel 17/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-166-244:897245:897787 [5] NCCL INFO Channel 08/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-167-9:1039539:1040100 [2] NCCL INFO Channel 03/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-166-244:897246:897791 [6] NCCL INFO Channel 10/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-166-244:897241:897784 [1] NCCL INFO Channel 12/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-166-15:127291:127838 [1] NCCL INFO Channel 16/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-166-244:897243:897790 [3] NCCL INFO Channel 15/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-166-244:897247:897788 [7] NCCL INFO Channel 17/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-166-244:897244:897789 [4] NCCL INFO Channel 17/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-166-244:897240:897786 [0] NCCL INFO Channel 17/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-166-244:897242:897792 [2] NCCL INFO Channel 18/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-166-15:127297:127840 [7] NCCL INFO Channel 20/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-166-15:127293:127841 [3] NCCL INFO Channel 20/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-166-244:897245:897787 [5] NCCL INFO Channel 09/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-165-213:9625:10169 [0] NCCL INFO Channel 10/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-166-15:127292:127836 [2] NCCL INFO Channel 12/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-166-15:127294:127835 [4] NCCL INFO Channel 09/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-165-213:9627:10174 [2] NCCL INFO Channel 14/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-165-213:9631:10172 [6] NCCL INFO Channel 08/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-165-213:9629:10171 [4] NCCL INFO Channel 13/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-165-213:9632:10176 [7] NCCL INFO Channel 20/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-166-15:127291:127838 [1] NCCL INFO Channel 17/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-166-244:897246:897791 [6] NCCL INFO Channel 11/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127297:127840 [7] NCCL INFO Channel 21/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-166-15:127293:127841 [3] NCCL INFO Channel 21/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-166-15:127295:127839 [5] NCCL INFO Channel 20/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-166-214:624944:625493 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-166-214:624944:625493 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-166-244:897241:897784 [1] NCCL INFO Channel 13/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-166-244:897247:897788 [7] NCCL INFO Channel 18/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-166-244:897243:897790 [3] NCCL INFO Channel 16/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-166-244:897244:897789 [4] NCCL INFO Channel 18/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-166-244:897240:897786 [0] NCCL INFO Channel 18/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-166-15:127294:127835 [4] NCCL INFO Channel 10/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-166-15:127292:127836 [2] NCCL INFO Channel 13/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-166-244:897242:897792 [2] NCCL INFO Channel 19/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-166-15:127291:127838 [1] NCCL INFO Channel 18/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-166-15:127290:127833 [0] NCCL INFO Channel 16/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-166-244:897245:897787 [5] NCCL INFO Channel 10/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-166-214:624947:625497 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-166-214:624947:625497 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-167-9:1039541:1040097 [4] NCCL INFO Channel 09/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-166-15:127297:127840 [7] NCCL INFO Channel 22/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-166-15:127293:127841 [3] NCCL INFO Channel 22/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-166-15:127295:127839 [5] NCCL INFO Channel 21/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-166-244:897246:897791 [6] NCCL INFO Channel 12/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-162-46:702882:703472 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-162-46:702882:703472 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-166-244:897241:897784 [1] NCCL INFO Channel 14/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-166-244:897247:897788 [7] NCCL INFO Channel 19/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-166-244:897243:897790 [3] NCCL INFO Channel 17/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-166-244:897240:897786 [0] NCCL INFO Channel 19/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-166-244:897244:897789 [4] NCCL INFO Channel 19/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-166-15:127294:127835 [4] NCCL INFO Channel 11/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-166-244:897242:897792 [2] NCCL INFO Channel 20/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-166-15:127292:127836 [2] NCCL INFO Channel 14/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-166-15:127296:127834 [6] NCCL INFO Channel 18/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-167-9:1039537:1040095 [0] NCCL INFO Channel 07/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-167-9:1039543:1040098 [6] NCCL INFO Channel 10/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-166-244:897245:897787 [5] NCCL INFO Channel 11/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-166-15:127291:127838 [1] NCCL INFO Channel 19/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-166-15:127290:127833 [0] NCCL INFO Channel 17/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-165-213:9627:10174 [2] NCCL INFO Channel 15/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-165-213:9625:10169 [0] NCCL INFO Channel 11/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-165-213:9631:10172 [6] NCCL INFO Channel 09/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-165-213:9629:10171 [4] NCCL INFO Channel 14/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-165-213:9632:10176 [7] NCCL INFO Channel 21/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-166-15:127297:127840 [7] NCCL INFO Channel 23/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-166-15:127293:127841 [3] NCCL INFO Channel 23/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-166-15:127295:127839 [5] NCCL INFO Channel 22/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-166-244:897246:897791 [6] NCCL INFO Channel 13/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-166-244:897241:897784 [1] NCCL INFO Channel 15/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-166-244:897243:897790 [3] NCCL INFO Channel 18/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-166-244:897247:897788 [7] NCCL INFO Channel 20/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-166-244:897240:897786 [0] NCCL INFO Channel 20/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-166-244:897244:897789 [4] NCCL INFO Channel 20/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-166-244:897242:897792 [2] NCCL INFO Channel 21/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-166-15:127294:127835 [4] NCCL INFO Channel 12/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-166-15:127292:127836 [2] NCCL INFO Channel 15/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-166-15:127296:127834 [6] NCCL INFO Channel 19/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127291:127838 [1] NCCL INFO Channel 20/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-166-244:897245:897787 [5] NCCL INFO Channel 12/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-166-15:127290:127833 [0] NCCL INFO Channel 18/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-162-46:702885:703468 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-162-46:702885:703468 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-166-15:127295:127839 [5] NCCL INFO Channel 23/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-166-244:897246:897791 [6] NCCL INFO Channel 14/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127294:127835 [4] NCCL INFO Channel 13/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-166-244:897241:897784 [1] NCCL INFO Channel 16/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-166-244:897243:897790 [3] NCCL INFO Channel 19/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-166-15:127296:127834 [6] NCCL INFO Channel 20/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-166-244:897247:897788 [7] NCCL INFO Channel 21/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-166-244:897240:897786 [0] NCCL INFO Channel 21/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-166-244:897244:897789 [4] NCCL INFO Channel 21/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-166-244:897242:897792 [2] NCCL INFO Channel 22/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-166-15:127292:127836 [2] NCCL INFO Channel 16/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-166-15:127291:127838 [1] NCCL INFO Channel 21/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-166-15:127290:127833 [0] NCCL INFO Channel 19/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-166-244:897245:897787 [5] NCCL INFO Channel 13/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-165-213:9627:10174 [2] NCCL INFO Channel 16/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-165-213:9625:10169 [0] NCCL INFO Channel 12/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-165-213:9629:10171 [4] NCCL INFO Channel 15/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-165-213:9631:10172 [6] NCCL INFO Channel 10/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-165-213:9632:10176 [7] NCCL INFO Channel 22/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-166-15:127294:127835 [4] NCCL INFO Channel 14/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-166-15:127296:127834 [6] NCCL INFO Channel 21/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127291:127838 [1] NCCL INFO Channel 22/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-166-244:897246:897791 [6] NCCL INFO Channel 15/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-166-244:897241:897784 [1] NCCL INFO Channel 17/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-166-244:897243:897790 [3] NCCL INFO Channel 20/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-166-244:897247:897788 [7] NCCL INFO Channel 22/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-166-244:897240:897786 [0] NCCL INFO Channel 22/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-166-244:897244:897789 [4] NCCL INFO Channel 22/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-166-244:897242:897792 [2] NCCL INFO Channel 23/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-166-244:897245:897787 [5] NCCL INFO Channel 14/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-166-15:127292:127836 [2] NCCL INFO Channel 17/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-166-214:624946:625498 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-166-214:624946:625498 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-166-15:127290:127833 [0] NCCL INFO Channel 20/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-166-244:897246:897791 [6] NCCL INFO Channel 16/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-166-244:897241:897784 [1] NCCL INFO Channel 18/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-166-244:897243:897790 [3] NCCL INFO Channel 21/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-166-244:897247:897788 [7] NCCL INFO Channel 23/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-166-244:897240:897786 [0] NCCL INFO Channel 23/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-166-244:897244:897789 [4] NCCL INFO Channel 23/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-166-244:897245:897787 [5] NCCL INFO Channel 15/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-165-213:9627:10174 [2] NCCL INFO Channel 17/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-165-213:9631:10172 [6] NCCL INFO Channel 11/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-165-213:9625:10169 [0] NCCL INFO Channel 13/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-165-213:9632:10176 [7] NCCL INFO Channel 23/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-165-213:9629:10171 [4] NCCL INFO Channel 16/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-166-244:897246:897791 [6] NCCL INFO Channel 17/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-166-244:897241:897784 [1] NCCL INFO Channel 19/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-166-244:897243:897790 [3] NCCL INFO Channel 22/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-166-244:897245:897787 [5] NCCL INFO Channel 16/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-166-15:127294:127835 [4] NCCL INFO Channel 15/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-166-15:127291:127838 [1] NCCL INFO Channel 23/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-167-9:1039539:1040100 [2] NCCL INFO Channel 04/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-165-213:9627:10174 [2] NCCL INFO Channel 18/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-165-213:9625:10169 [0] NCCL INFO Channel 14/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-165-213:9629:10171 [4] NCCL INFO Channel 17/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-165-213:9631:10172 [6] NCCL INFO Channel 12/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127292:127836 [2] NCCL INFO Channel 18/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-167-9:1039537:1040095 [0] NCCL INFO Channel 08/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-166-15:127294:127835 [4] NCCL INFO Channel 16/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-167-9:1039541:1040097 [4] NCCL INFO Channel 10/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-166-244:897243:897790 [3] NCCL INFO Channel 23/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-166-244:897241:897784 [1] NCCL INFO Channel 20/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-167-9:1039543:1040098 [6] NCCL INFO Channel 11/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624947:625497 [4] NCCL INFO Trees [0] 3/-1/-1->2->1 [1] 3/-1/-1->2->1 [2] 3/-1/-1->2->1 [3] 3/-1/-1->2->1 [4] 3/-1/-1->2->1 [5] 3/-1/-1->2->1 [6] 3/-1/-1->2->1 [7] 3/-1/-1->2->1 [8] 3/-1/-1->2->1 [9] 3/-1/-1->2->1 [10] 3/-1/-1->2->1 [11] 3/-1/-1->2->1 [12] 3/-1/-1->2->1 [13] 3/-1/-1->2->1 [14] 3/-1/-1->2->1 [15] 3/-1/-1->2->1 [16] 3/-1/-1->2->1 [17] 3/-1/-1->2->1 [18] 3/-1/-1->2->1 [19] 3/-1/-1->2->1 [20] 3/-1/-1->2->1 [21] 3/-1/-1->2->1 [22] 3/-1/-1->2->1 [23] 3/-1/-1->2->1
+ip-26-0-166-214:624949:625496 [6] NCCL INFO Trees [0] -1/-1/-1->3->2 [1] -1/-1/-1->3->2 [2] -1/-1/-1->3->2 [3] -1/-1/-1->3->2 [4] -1/-1/-1->3->2 [5] -1/-1/-1->3->2 [6] -1/-1/-1->3->2 [7] -1/-1/-1->3->2 [8] -1/-1/-1->3->2 [9] -1/-1/-1->3->2 [10] -1/-1/-1->3->2 [11] -1/-1/-1->3->2 [12] -1/-1/-1->3->2 [13] -1/-1/-1->3->2 [14] -1/-1/-1->3->2 [15] -1/-1/-1->3->2 [16] -1/-1/-1->3->2 [17] -1/-1/-1->3->2 [18] -1/-1/-1->3->2 [19] -1/-1/-1->3->2 [20] -1/-1/-1->3->2 [21] -1/-1/-1->3->2 [22] -1/-1/-1->3->2 [23] -1/-1/-1->3->2
+ip-26-0-166-214:624947:625497 [4] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-166-214:624945:625495 [2] NCCL INFO Trees [0] 2/-1/-1->1->0 [1] 2/-1/-1->1->0 [2] 2/-1/-1->1->0 [3] 2/-1/-1->1->0 [4] 2/-1/-1->1->0 [5] 2/-1/-1->1->0 [6] 2/-1/-1->1->0 [7] 2/-1/-1->1->0 [8] 2/-1/-1->1->0 [9] 2/-1/-1->1->0 [10] 2/-1/-1->1->0 [11] 2/-1/-1->1->0 [12] 2/-1/-1->1->0 [13] 2/-1/-1->1->0 [14] 2/-1/-1->1->0 [15] 2/-1/-1->1->0 [16] 2/-1/-1->1->0 [17] 2/-1/-1->1->0 [18] 2/-1/-1->1->0 [19] 2/-1/-1->1->0 [20] 2/-1/-1->1->0 [21] 2/-1/-1->1->0 [22] 2/-1/-1->1->0 [23] 2/-1/-1->1->0
+ip-26-0-166-214:624949:625496 [6] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-166-214:624945:625495 [2] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-166-214:624943:625494 [0] NCCL INFO Channel 00/24 :    0   1   2   3
+ip-26-0-166-214:624943:625494 [0] NCCL INFO Channel 01/24 :    0   1   2   3
+ip-26-0-166-214:624943:625494 [0] NCCL INFO Channel 02/24 :    0   1   2   3
+ip-26-0-166-214:624943:625494 [0] NCCL INFO Channel 03/24 :    0   1   2   3
+ip-26-0-166-214:624943:625494 [0] NCCL INFO Channel 04/24 :    0   1   2   3
+ip-26-0-166-214:624943:625494 [0] NCCL INFO Channel 05/24 :    0   1   2   3
+ip-26-0-166-214:624943:625494 [0] NCCL INFO Channel 06/24 :    0   1   2   3
+ip-26-0-166-214:624943:625494 [0] NCCL INFO Channel 07/24 :    0   1   2   3
+ip-26-0-166-214:624943:625494 [0] NCCL INFO Channel 08/24 :    0   1   2   3
+ip-26-0-166-214:624943:625494 [0] NCCL INFO Channel 09/24 :    0   1   2   3
+ip-26-0-166-214:624943:625494 [0] NCCL INFO Channel 10/24 :    0   1   2   3
+ip-26-0-166-214:624943:625494 [0] NCCL INFO Channel 11/24 :    0   1   2   3
+ip-26-0-166-214:624943:625494 [0] NCCL INFO Channel 12/24 :    0   1   2   3
+ip-26-0-166-214:624943:625494 [0] NCCL INFO Channel 13/24 :    0   1   2   3
+ip-26-0-166-214:624943:625494 [0] NCCL INFO Channel 14/24 :    0   1   2   3
+ip-26-0-166-214:624943:625494 [0] NCCL INFO Channel 15/24 :    0   1   2   3
+ip-26-0-166-214:624943:625494 [0] NCCL INFO Channel 16/24 :    0   1   2   3
+ip-26-0-166-214:624943:625494 [0] NCCL INFO Channel 17/24 :    0   1   2   3
+ip-26-0-166-214:624943:625494 [0] NCCL INFO Channel 18/24 :    0   1   2   3
+ip-26-0-166-214:624943:625494 [0] NCCL INFO Channel 19/24 :    0   1   2   3
+ip-26-0-166-214:624943:625494 [0] NCCL INFO Channel 20/24 :    0   1   2   3
+ip-26-0-166-214:624943:625494 [0] NCCL INFO Channel 21/24 :    0   1   2   3
+ip-26-0-166-214:624943:625494 [0] NCCL INFO Channel 22/24 :    0   1   2   3
+ip-26-0-166-214:624943:625494 [0] NCCL INFO Channel 23/24 :    0   1   2   3
+ip-26-0-166-214:624943:625494 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] 1/-1/-1->0->-1 [7] 1/-1/-1->0->-1 [8] 1/-1/-1->0->-1 [9] 1/-1/-1->0->-1 [10] 1/-1/-1->0->-1 [11] 1/-1/-1->0->-1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] 1/-1/-1->0->-1 [19] 1/-1/-1->0->-1 [20] 1/-1/-1->0->-1 [21] 1/-1/-1->0->-1 [22] 1/-1/-1->0->-1 [23] 1/-1/-1->0->-1
+ip-26-0-166-214:624943:625494 [0] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-166-244:897245:897787 [5] NCCL INFO Channel 17/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-162-46:702882:703472 [2] NCCL INFO Trees [0] 2/-1/-1->1->0 [1] 2/-1/-1->1->0 [2] 2/-1/-1->1->0 [3] 2/-1/-1->1->0 [4] 2/-1/-1->1->0 [5] 2/-1/-1->1->0 [6] 2/-1/-1->1->0 [7] 2/-1/-1->1->0 [8] 2/-1/-1->1->0 [9] 2/-1/-1->1->0 [10] 2/-1/-1->1->0 [11] 2/-1/-1->1->0 [12] 2/-1/-1->1->0 [13] 2/-1/-1->1->0 [14] 2/-1/-1->1->0 [15] 2/-1/-1->1->0 [16] 2/-1/-1->1->0 [17] 2/-1/-1->1->0 [18] 2/-1/-1->1->0 [19] 2/-1/-1->1->0 [20] 2/-1/-1->1->0 [21] 2/-1/-1->1->0 [22] 2/-1/-1->1->0 [23] 2/-1/-1->1->0
+ip-26-0-162-46:702882:703472 [2] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-162-46:702880:703466 [0] NCCL INFO Channel 00/24 :    0   1   2   3
+ip-26-0-162-46:702880:703466 [0] NCCL INFO Channel 01/24 :    0   1   2   3
+ip-26-0-162-46:702880:703466 [0] NCCL INFO Channel 02/24 :    0   1   2   3
+ip-26-0-162-46:702880:703466 [0] NCCL INFO Channel 03/24 :    0   1   2   3
+ip-26-0-162-46:702886:703469 [6] NCCL INFO Trees [0] -1/-1/-1->3->2 [1] -1/-1/-1->3->2 [2] -1/-1/-1->3->2 [3] -1/-1/-1->3->2 [4] -1/-1/-1->3->2 [5] -1/-1/-1->3->2 [6] -1/-1/-1->3->2 [7] -1/-1/-1->3->2 [8] -1/-1/-1->3->2 [9] -1/-1/-1->3->2 [10] -1/-1/-1->3->2 [11] -1/-1/-1->3->2 [12] -1/-1/-1->3->2 [13] -1/-1/-1->3->2 [14] -1/-1/-1->3->2 [15] -1/-1/-1->3->2 [16] -1/-1/-1->3->2 [17] -1/-1/-1->3->2 [18] -1/-1/-1->3->2 [19] -1/-1/-1->3->2 [20] -1/-1/-1->3->2 [21] -1/-1/-1->3->2 [22] -1/-1/-1->3->2 [23] -1/-1/-1->3->2
+ip-26-0-162-46:702884:703470 [4] NCCL INFO Trees [0] 3/-1/-1->2->1 [1] 3/-1/-1->2->1 [2] 3/-1/-1->2->1 [3] 3/-1/-1->2->1 [4] 3/-1/-1->2->1 [5] 3/-1/-1->2->1 [6] 3/-1/-1->2->1 [7] 3/-1/-1->2->1 [8] 3/-1/-1->2->1 [9] 3/-1/-1->2->1 [10] 3/-1/-1->2->1 [11] 3/-1/-1->2->1 [12] 3/-1/-1->2->1 [13] 3/-1/-1->2->1 [14] 3/-1/-1->2->1 [15] 3/-1/-1->2->1 [16] 3/-1/-1->2->1 [17] 3/-1/-1->2->1 [18] 3/-1/-1->2->1 [19] 3/-1/-1->2->1 [20] 3/-1/-1->2->1 [21] 3/-1/-1->2->1 [22] 3/-1/-1->2->1 [23] 3/-1/-1->2->1
+ip-26-0-162-46:702880:703466 [0] NCCL INFO Channel 04/24 :    0   1   2   3
+ip-26-0-162-46:702886:703469 [6] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-162-46:702884:703470 [4] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-162-46:702880:703466 [0] NCCL INFO Channel 05/24 :    0   1   2   3
+ip-26-0-162-46:702880:703466 [0] NCCL INFO Channel 06/24 :    0   1   2   3
+ip-26-0-162-46:702880:703466 [0] NCCL INFO Channel 07/24 :    0   1   2   3
+ip-26-0-162-46:702880:703466 [0] NCCL INFO Channel 08/24 :    0   1   2   3
+ip-26-0-162-46:702880:703466 [0] NCCL INFO Channel 09/24 :    0   1   2   3
+ip-26-0-162-46:702880:703466 [0] NCCL INFO Channel 10/24 :    0   1   2   3
+ip-26-0-162-46:702880:703466 [0] NCCL INFO Channel 11/24 :    0   1   2   3
+ip-26-0-162-46:702880:703466 [0] NCCL INFO Channel 12/24 :    0   1   2   3
+ip-26-0-162-46:702880:703466 [0] NCCL INFO Channel 13/24 :    0   1   2   3
+ip-26-0-162-46:702880:703466 [0] NCCL INFO Channel 14/24 :    0   1   2   3
+ip-26-0-162-46:702880:703466 [0] NCCL INFO Channel 15/24 :    0   1   2   3
+ip-26-0-162-46:702880:703466 [0] NCCL INFO Channel 16/24 :    0   1   2   3
+ip-26-0-162-46:702880:703466 [0] NCCL INFO Channel 17/24 :    0   1   2   3
+ip-26-0-162-46:702880:703466 [0] NCCL INFO Channel 18/24 :    0   1   2   3
+ip-26-0-162-46:702880:703466 [0] NCCL INFO Channel 19/24 :    0   1   2   3
+ip-26-0-162-46:702880:703466 [0] NCCL INFO Channel 20/24 :    0   1   2   3
+ip-26-0-162-46:702880:703466 [0] NCCL INFO Channel 21/24 :    0   1   2   3
+ip-26-0-162-46:702880:703466 [0] NCCL INFO Channel 22/24 :    0   1   2   3
+ip-26-0-162-46:702880:703466 [0] NCCL INFO Channel 23/24 :    0   1   2   3
+ip-26-0-162-46:702880:703466 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] 1/-1/-1->0->-1 [7] 1/-1/-1->0->-1 [8] 1/-1/-1->0->-1 [9] 1/-1/-1->0->-1 [10] 1/-1/-1->0->-1 [11] 1/-1/-1->0->-1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] 1/-1/-1->0->-1 [19] 1/-1/-1->0->-1 [20] 1/-1/-1->0->-1 [21] 1/-1/-1->0->-1 [22] 1/-1/-1->0->-1 [23] 1/-1/-1->0->-1
+ip-26-0-162-46:702880:703466 [0] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-166-244:897241:897784 [1] NCCL INFO Channel 21/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-165-213:9627:10174 [2] NCCL INFO Channel 19/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-166-244:897245:897787 [5] NCCL INFO Channel 18/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-166-244:897241:897784 [1] NCCL INFO Channel 22/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-166-244:897246:897791 [6] NCCL INFO Channel 18/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-165-213:9629:10171 [4] NCCL INFO Channel 18/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-166-36:104436:104978 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-162-46:702885:703468 [5] NCCL INFO Trees [0] 3/-1/-1->2->1 [1] 3/-1/-1->2->1 [2] 3/-1/-1->2->1 [3] 3/-1/-1->2->1 [4] 3/-1/-1->2->1 [5] 3/-1/-1->2->1 [6] 3/-1/-1->2->1 [7] 3/-1/-1->2->1 [8] 3/-1/-1->2->1 [9] 3/-1/-1->2->1 [10] 3/-1/-1->2->1 [11] 3/-1/-1->2->1 [12] 3/-1/-1->2->1 [13] 3/-1/-1->2->1 [14] 3/-1/-1->2->1 [15] 3/-1/-1->2->1 [16] 3/-1/-1->2->1 [17] 3/-1/-1->2->1 [18] 3/-1/-1->2->1 [19] 3/-1/-1->2->1 [20] 3/-1/-1->2->1 [21] 3/-1/-1->2->1 [22] 3/-1/-1->2->1 [23] 3/-1/-1->2->1
+ip-26-0-162-46:702887:703467 [7] NCCL INFO Trees [0] -1/-1/-1->3->2 [1] -1/-1/-1->3->2 [2] -1/-1/-1->3->2 [3] -1/-1/-1->3->2 [4] -1/-1/-1->3->2 [5] -1/-1/-1->3->2 [6] -1/-1/-1->3->2 [7] -1/-1/-1->3->2 [8] -1/-1/-1->3->2 [9] -1/-1/-1->3->2 [10] -1/-1/-1->3->2 [11] -1/-1/-1->3->2 [12] -1/-1/-1->3->2 [13] -1/-1/-1->3->2 [14] -1/-1/-1->3->2 [15] -1/-1/-1->3->2 [16] -1/-1/-1->3->2 [17] -1/-1/-1->3->2 [18] -1/-1/-1->3->2 [19] -1/-1/-1->3->2 [20] -1/-1/-1->3->2 [21] -1/-1/-1->3->2 [22] -1/-1/-1->3->2 [23] -1/-1/-1->3->2
+ip-26-0-162-46:702881:703465 [1] NCCL INFO Channel 00/24 :    0   1   2   3
+ip-26-0-162-46:702885:703468 [5] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-162-46:702887:703467 [7] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-162-46:702883:703471 [3] NCCL INFO Trees [0] 2/-1/-1->1->0 [1] 2/-1/-1->1->0 [2] 2/-1/-1->1->0 [3] 2/-1/-1->1->0 [4] 2/-1/-1->1->0 [5] 2/-1/-1->1->0 [6] 2/-1/-1->1->0 [7] 2/-1/-1->1->0 [8] 2/-1/-1->1->0 [9] 2/-1/-1->1->0 [10] 2/-1/-1->1->0 [11] 2/-1/-1->1->0 [12] 2/-1/-1->1->0 [13] 2/-1/-1->1->0 [14] 2/-1/-1->1->0 [15] 2/-1/-1->1->0 [16] 2/-1/-1->1->0 [17] 2/-1/-1->1->0 [18] 2/-1/-1->1->0 [19] 2/-1/-1->1->0 [20] 2/-1/-1->1->0 [21] 2/-1/-1->1->0 [22] 2/-1/-1->1->0 [23] 2/-1/-1->1->0
+ip-26-0-162-46:702881:703465 [1] NCCL INFO Channel 01/24 :    0   1   2   3
+ip-26-0-162-46:702883:703471 [3] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-162-46:702881:703465 [1] NCCL INFO Channel 02/24 :    0   1   2   3
+ip-26-0-162-46:702881:703465 [1] NCCL INFO Channel 03/24 :    0   1   2   3
+ip-26-0-166-36:104436:104978 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-162-46:702881:703465 [1] NCCL INFO Channel 04/24 :    0   1   2   3
+ip-26-0-162-46:702881:703465 [1] NCCL INFO Channel 05/24 :    0   1   2   3
+ip-26-0-162-46:702881:703465 [1] NCCL INFO Channel 06/24 :    0   1   2   3
+ip-26-0-162-46:702881:703465 [1] NCCL INFO Channel 07/24 :    0   1   2   3
+ip-26-0-162-46:702881:703465 [1] NCCL INFO Channel 08/24 :    0   1   2   3
+ip-26-0-162-46:702881:703465 [1] NCCL INFO Channel 09/24 :    0   1   2   3
+ip-26-0-162-46:702881:703465 [1] NCCL INFO Channel 10/24 :    0   1   2   3
+ip-26-0-162-46:702881:703465 [1] NCCL INFO Channel 11/24 :    0   1   2   3
+ip-26-0-162-46:702881:703465 [1] NCCL INFO Channel 12/24 :    0   1   2   3
+ip-26-0-162-46:702881:703465 [1] NCCL INFO Channel 13/24 :    0   1   2   3
+ip-26-0-162-46:702881:703465 [1] NCCL INFO Channel 14/24 :    0   1   2   3
+ip-26-0-162-46:702881:703465 [1] NCCL INFO Channel 15/24 :    0   1   2   3
+ip-26-0-162-46:702881:703465 [1] NCCL INFO Channel 16/24 :    0   1   2   3
+ip-26-0-162-46:702881:703465 [1] NCCL INFO Channel 17/24 :    0   1   2   3
+ip-26-0-162-46:702881:703465 [1] NCCL INFO Channel 18/24 :    0   1   2   3
+ip-26-0-162-46:702881:703465 [1] NCCL INFO Channel 19/24 :    0   1   2   3
+ip-26-0-162-46:702881:703465 [1] NCCL INFO Channel 20/24 :    0   1   2   3
+ip-26-0-162-46:702881:703465 [1] NCCL INFO Channel 21/24 :    0   1   2   3
+ip-26-0-162-46:702881:703465 [1] NCCL INFO Channel 22/24 :    0   1   2   3
+ip-26-0-162-46:702881:703465 [1] NCCL INFO Channel 23/24 :    0   1   2   3
+ip-26-0-162-46:702881:703465 [1] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] 1/-1/-1->0->-1 [7] 1/-1/-1->0->-1 [8] 1/-1/-1->0->-1 [9] 1/-1/-1->0->-1 [10] 1/-1/-1->0->-1 [11] 1/-1/-1->0->-1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] 1/-1/-1->0->-1 [19] 1/-1/-1->0->-1 [20] 1/-1/-1->0->-1 [21] 1/-1/-1->0->-1 [22] 1/-1/-1->0->-1 [23] 1/-1/-1->0->-1
+ip-26-0-162-46:702881:703465 [1] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-165-213:9625:10169 [0] NCCL INFO Channel 15/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-165-213:9631:10172 [6] NCCL INFO Channel 13/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127292:127836 [2] NCCL INFO Channel 19/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-166-15:127294:127835 [4] NCCL INFO Channel 17/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-166-15:127296:127834 [6] NCCL INFO Channel 22/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624944:625493 [1] NCCL INFO Channel 00/24 :    0   1   2   3
+ip-26-0-166-214:624944:625493 [1] NCCL INFO Channel 01/24 :    0   1   2   3
+ip-26-0-166-214:624946:625498 [3] NCCL INFO Trees [0] 2/-1/-1->1->0 [1] 2/-1/-1->1->0 [2] 2/-1/-1->1->0 [3] 2/-1/-1->1->0 [4] 2/-1/-1->1->0 [5] 2/-1/-1->1->0 [6] 2/-1/-1->1->0 [7] 2/-1/-1->1->0 [8] 2/-1/-1->1->0 [9] 2/-1/-1->1->0 [10] 2/-1/-1->1->0 [11] 2/-1/-1->1->0 [12] 2/-1/-1->1->0 [13] 2/-1/-1->1->0 [14] 2/-1/-1->1->0 [15] 2/-1/-1->1->0 [16] 2/-1/-1->1->0 [17] 2/-1/-1->1->0 [18] 2/-1/-1->1->0 [19] 2/-1/-1->1->0 [20] 2/-1/-1->1->0 [21] 2/-1/-1->1->0 [22] 2/-1/-1->1->0 [23] 2/-1/-1->1->0
+ip-26-0-166-214:624944:625493 [1] NCCL INFO Channel 02/24 :    0   1   2   3
+ip-26-0-166-214:624946:625498 [3] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-166-214:624944:625493 [1] NCCL INFO Channel 03/24 :    0   1   2   3
+ip-26-0-166-214:624948:625500 [5] NCCL INFO Trees [0] 3/-1/-1->2->1 [1] 3/-1/-1->2->1 [2] 3/-1/-1->2->1 [3] 3/-1/-1->2->1 [4] 3/-1/-1->2->1 [5] 3/-1/-1->2->1 [6] 3/-1/-1->2->1 [7] 3/-1/-1->2->1 [8] 3/-1/-1->2->1 [9] 3/-1/-1->2->1 [10] 3/-1/-1->2->1 [11] 3/-1/-1->2->1 [12] 3/-1/-1->2->1 [13] 3/-1/-1->2->1 [14] 3/-1/-1->2->1 [15] 3/-1/-1->2->1 [16] 3/-1/-1->2->1 [17] 3/-1/-1->2->1 [18] 3/-1/-1->2->1 [19] 3/-1/-1->2->1 [20] 3/-1/-1->2->1 [21] 3/-1/-1->2->1 [22] 3/-1/-1->2->1 [23] 3/-1/-1->2->1
+ip-26-0-166-214:624944:625493 [1] NCCL INFO Channel 04/24 :    0   1   2   3
+ip-26-0-166-214:624950:625499 [7] NCCL INFO Trees [0] -1/-1/-1->3->2 [1] -1/-1/-1->3->2 [2] -1/-1/-1->3->2 [3] -1/-1/-1->3->2 [4] -1/-1/-1->3->2 [5] -1/-1/-1->3->2 [6] -1/-1/-1->3->2 [7] -1/-1/-1->3->2 [8] -1/-1/-1->3->2 [9] -1/-1/-1->3->2 [10] -1/-1/-1->3->2 [11] -1/-1/-1->3->2 [12] -1/-1/-1->3->2 [13] -1/-1/-1->3->2 [14] -1/-1/-1->3->2 [15] -1/-1/-1->3->2 [16] -1/-1/-1->3->2 [17] -1/-1/-1->3->2 [18] -1/-1/-1->3->2 [19] -1/-1/-1->3->2 [20] -1/-1/-1->3->2 [21] -1/-1/-1->3->2 [22] -1/-1/-1->3->2 [23] -1/-1/-1->3->2
+ip-26-0-166-214:624944:625493 [1] NCCL INFO Channel 05/24 :    0   1   2   3
+ip-26-0-166-214:624948:625500 [5] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-166-214:624950:625499 [7] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-166-214:624944:625493 [1] NCCL INFO Channel 06/24 :    0   1   2   3
+ip-26-0-166-214:624944:625493 [1] NCCL INFO Channel 07/24 :    0   1   2   3
+ip-26-0-166-214:624944:625493 [1] NCCL INFO Channel 08/24 :    0   1   2   3
+ip-26-0-166-214:624944:625493 [1] NCCL INFO Channel 09/24 :    0   1   2   3
+ip-26-0-166-214:624944:625493 [1] NCCL INFO Channel 10/24 :    0   1   2   3
+ip-26-0-166-214:624944:625493 [1] NCCL INFO Channel 11/24 :    0   1   2   3
+ip-26-0-166-214:624944:625493 [1] NCCL INFO Channel 12/24 :    0   1   2   3
+ip-26-0-166-214:624944:625493 [1] NCCL INFO Channel 13/24 :    0   1   2   3
+ip-26-0-166-214:624944:625493 [1] NCCL INFO Channel 14/24 :    0   1   2   3
+ip-26-0-166-214:624944:625493 [1] NCCL INFO Channel 15/24 :    0   1   2   3
+ip-26-0-166-214:624944:625493 [1] NCCL INFO Channel 16/24 :    0   1   2   3
+ip-26-0-166-214:624944:625493 [1] NCCL INFO Channel 17/24 :    0   1   2   3
+ip-26-0-166-214:624944:625493 [1] NCCL INFO Channel 18/24 :    0   1   2   3
+ip-26-0-166-214:624944:625493 [1] NCCL INFO Channel 19/24 :    0   1   2   3
+ip-26-0-166-214:624944:625493 [1] NCCL INFO Channel 20/24 :    0   1   2   3
+ip-26-0-166-214:624944:625493 [1] NCCL INFO Channel 21/24 :    0   1   2   3
+ip-26-0-166-214:624944:625493 [1] NCCL INFO Channel 22/24 :    0   1   2   3
+ip-26-0-166-214:624944:625493 [1] NCCL INFO Channel 23/24 :    0   1   2   3
+ip-26-0-166-214:624944:625493 [1] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] 1/-1/-1->0->-1 [7] 1/-1/-1->0->-1 [8] 1/-1/-1->0->-1 [9] 1/-1/-1->0->-1 [10] 1/-1/-1->0->-1 [11] 1/-1/-1->0->-1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] 1/-1/-1->0->-1 [19] 1/-1/-1->0->-1 [20] 1/-1/-1->0->-1 [21] 1/-1/-1->0->-1 [22] 1/-1/-1->0->-1 [23] 1/-1/-1->0->-1
+ip-26-0-166-214:624944:625493 [1] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-167-9:1039539:1040100 [2] NCCL INFO Channel 05/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-166-244:897245:897787 [5] NCCL INFO Channel 19/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-166-244:897241:897784 [1] NCCL INFO Channel 23/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-165-213:9629:10171 [4] NCCL INFO Channel 19/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-165-213:9629:10171 [4] NCCL INFO Channel 20/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-165-213:9629:10171 [4] NCCL INFO Channel 21/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-165-213:9629:10171 [4] NCCL INFO Channel 22/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-167-9:1039537:1040095 [0] NCCL INFO Channel 09/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-167-9:1039541:1040097 [4] NCCL INFO Channel 11/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-167-9:1039543:1040098 [6] NCCL INFO Channel 12/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-166-244:897245:897787 [5] NCCL INFO Channel 20/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-166-15:127290:127833 [0] NCCL INFO Channel 21/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-166-244:897246:897791 [6] NCCL INFO Channel 19/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-165-213:9625:10169 [0] NCCL INFO Channel 16/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-165-213:9631:10172 [6] NCCL INFO Channel 14/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-167-9:1039539:1040100 [2] NCCL INFO Channel 06/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-166-244:897245:897787 [5] NCCL INFO Channel 21/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-165-213:9627:10174 [2] NCCL INFO Channel 20/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-166-244:897246:897791 [6] NCCL INFO Channel 20/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-167-9:1039537:1040095 [0] NCCL INFO Channel 10/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-167-9:1039543:1040098 [6] NCCL INFO Channel 13/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-167-9:1039541:1040097 [4] NCCL INFO Channel 12/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-166-15:127292:127836 [2] NCCL INFO Channel 20/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-166-15:127294:127835 [4] NCCL INFO Channel 18/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-166-15:127296:127834 [6] NCCL INFO Channel 23/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127290:127833 [0] NCCL INFO Channel 22/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-165-213:9631:10172 [6] NCCL INFO Channel 15/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-166-244:897245:897787 [5] NCCL INFO Channel 22/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-166-244:897246:897791 [6] NCCL INFO Channel 21/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-165-213:9629:10171 [4] NCCL INFO Channel 23/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-167-9:1039539:1040100 [2] NCCL INFO Channel 07/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-166-214:624949:625496 [6] NCCL INFO Channel 00/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-165-213:9631:10172 [6] NCCL INFO Channel 16/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-166-244:897245:897787 [5] NCCL INFO Channel 23/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-166-244:897246:897791 [6] NCCL INFO Channel 22/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624949:625496 [6] NCCL INFO Channel 01/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-167-9:1039543:1040098 [6] NCCL INFO Channel 14/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-167-9:1039541:1040097 [4] NCCL INFO Channel 13/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-167-9:1039537:1040095 [0] NCCL INFO Channel 11/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-166-214:624947:625497 [4] NCCL INFO Channel 00/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-166-214:624949:625496 [6] NCCL INFO Channel 02/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-165-213:9625:10169 [0] NCCL INFO Channel 17/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-166-15:127294:127835 [4] NCCL INFO Channel 19/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-166-15:127292:127836 [2] NCCL INFO Channel 21/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-166-214:624943:625494 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-166-214:624949:625496 [6] NCCL INFO Channel 03/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624947:625497 [4] NCCL INFO Channel 01/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-166-214:624943:625494 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-166-214:624947:625497 [4] NCCL INFO Channel 02/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-166-214:624949:625496 [6] NCCL INFO Channel 04/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-165-213:9631:10172 [6] NCCL INFO Channel 17/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624949:625496 [6] NCCL INFO Channel 05/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624947:625497 [4] NCCL INFO Channel 03/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-166-214:624943:625494 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-166-125:762821:763373 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,fff00000,00000000
+ip-26-0-167-9:1039539:1040100 [2] NCCL INFO Channel 08/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-166-125:762821:763373 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-166-214:624949:625496 [6] NCCL INFO Channel 06/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624947:625497 [4] NCCL INFO Channel 04/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-166-244:897246:897791 [6] NCCL INFO Channel 23/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-165-213:9627:10174 [2] NCCL INFO Channel 21/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-166-214:624943:625494 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-166-214:624949:625496 [6] NCCL INFO Channel 07/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624947:625497 [4] NCCL INFO Channel 05/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-167-9:1039543:1040098 [6] NCCL INFO Channel 15/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-167-9:1039541:1040097 [4] NCCL INFO Channel 14/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-167-9:1039537:1040095 [0] NCCL INFO Channel 12/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-166-15:127290:127833 [0] NCCL INFO Channel 23/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-166-214:624943:625494 [0] NCCL INFO Channel 04/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-166-214:624949:625496 [6] NCCL INFO Channel 08/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624947:625497 [4] NCCL INFO Channel 06/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-166-214:624945:625495 [2] NCCL INFO Channel 00/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-166-214:624943:625494 [0] NCCL INFO Channel 05/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-166-214:624947:625497 [4] NCCL INFO Channel 07/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-166-214:624945:625495 [2] NCCL INFO Channel 01/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-166-214:624943:625494 [0] NCCL INFO Channel 06/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-166-15:127294:127835 [4] NCCL INFO Channel 20/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-166-214:624947:625497 [4] NCCL INFO Channel 08/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-166-15:127292:127836 [2] NCCL INFO Channel 22/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-166-214:624945:625495 [2] NCCL INFO Channel 02/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-166-214:624943:625494 [0] NCCL INFO Channel 07/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-166-214:624947:625497 [4] NCCL INFO Channel 09/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-166-214:624945:625495 [2] NCCL INFO Channel 03/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-166-214:624943:625494 [0] NCCL INFO Channel 08/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-166-214:624947:625497 [4] NCCL INFO Channel 10/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-167-9:1039537:1040095 [0] NCCL INFO Channel 13/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-162-46:702886:703469 [6] NCCL INFO Channel 00/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-165-213:9627:10174 [2] NCCL INFO Channel 22/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-162-46:702886:703469 [6] NCCL INFO Channel 01/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-162-46:702884:703470 [4] NCCL INFO Channel 00/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-162-46:702886:703469 [6] NCCL INFO Channel 02/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-162-46:702884:703470 [4] NCCL INFO Channel 01/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-162-46:702886:703469 [6] NCCL INFO Channel 03/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-162-46:702884:703470 [4] NCCL INFO Channel 02/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-162-46:702886:703469 [6] NCCL INFO Channel 04/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-162-46:702885:703468 [5] NCCL INFO Channel 00/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-162-46:702882:703472 [2] NCCL INFO Channel 00/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-162-46:702884:703470 [4] NCCL INFO Channel 03/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-162-46:702885:703468 [5] NCCL INFO Channel 01/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-162-46:702880:703466 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-162-46:702883:703471 [3] NCCL INFO Channel 00/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-162-46:702885:703468 [5] NCCL INFO Channel 02/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-162-46:702880:703466 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-162-46:702881:703465 [1] NCCL INFO Channel 00/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-162-46:702883:703471 [3] NCCL INFO Channel 01/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-162-46:702885:703468 [5] NCCL INFO Channel 03/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-166-214:624945:625495 [2] NCCL INFO Channel 04/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-162-46:702880:703466 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-162-46:702881:703465 [1] NCCL INFO Channel 01/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-162-46:702883:703471 [3] NCCL INFO Channel 02/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-162-46:702885:703468 [5] NCCL INFO Channel 04/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-162-46:702882:703472 [2] NCCL INFO Channel 01/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-162-46:702880:703466 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-162-46:702881:703465 [1] NCCL INFO Channel 02/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-162-46:702883:703471 [3] NCCL INFO Channel 03/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-162-46:702885:703468 [5] NCCL INFO Channel 05/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-162-46:702882:703472 [2] NCCL INFO Channel 02/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-162-46:702880:703466 [0] NCCL INFO Channel 04/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-162-46:702881:703465 [1] NCCL INFO Channel 03/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-162-46:702883:703471 [3] NCCL INFO Channel 04/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-162-46:702886:703469 [6] NCCL INFO Channel 05/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-162-46:702885:703468 [5] NCCL INFO Channel 06/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-162-46:702884:703470 [4] NCCL INFO Channel 04/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-162-46:702880:703466 [0] NCCL INFO Channel 05/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-166-214:624943:625494 [0] NCCL INFO Channel 09/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-167-9:1039543:1040098 [6] NCCL INFO Channel 16/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624947:625497 [4] NCCL INFO Channel 11/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-165-213:9631:10172 [6] NCCL INFO Channel 18/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624949:625496 [6] NCCL INFO Channel 09/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-162-46:702885:703468 [5] NCCL INFO Channel 07/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-166-214:624945:625495 [2] NCCL INFO Channel 05/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-162-46:702882:703472 [2] NCCL INFO Channel 03/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-162-46:702881:703465 [1] NCCL INFO Channel 04/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-162-46:702883:703471 [3] NCCL INFO Channel 05/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-162-46:702886:703469 [6] NCCL INFO Channel 06/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-167-9:1039539:1040100 [2] NCCL INFO Channel 09/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-166-214:624943:625494 [0] NCCL INFO Channel 10/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-166-214:624947:625497 [4] NCCL INFO Channel 12/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-166-214:624949:625496 [6] NCCL INFO Channel 10/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-162-46:702880:703466 [0] NCCL INFO Channel 06/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-166-214:624945:625495 [2] NCCL INFO Channel 06/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-162-46:702885:703468 [5] NCCL INFO Channel 08/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-162-46:702882:703472 [2] NCCL INFO Channel 04/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-162-46:702881:703465 [1] NCCL INFO Channel 05/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-162-46:702883:703471 [3] NCCL INFO Channel 06/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-162-46:702886:703469 [6] NCCL INFO Channel 07/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-165-213:9631:10172 [6] NCCL INFO Channel 19/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624943:625494 [0] NCCL INFO Channel 11/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-166-214:624947:625497 [4] NCCL INFO Channel 13/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-166-214:624949:625496 [6] NCCL INFO Channel 11/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624945:625495 [2] NCCL INFO Channel 07/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-162-46:702884:703470 [4] NCCL INFO Channel 05/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-167-9:1039541:1040097 [4] NCCL INFO Channel 15/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-162-46:702880:703466 [0] NCCL INFO Channel 07/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-162-46:702882:703472 [2] NCCL INFO Channel 05/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-162-46:702881:703465 [1] NCCL INFO Channel 06/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-162-46:702886:703469 [6] NCCL INFO Channel 08/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-162-46:702883:703471 [3] NCCL INFO Channel 07/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-162-46:702887:703467 [7] NCCL INFO Channel 00/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-162-46:702885:703468 [5] NCCL INFO Channel 09/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-166-214:624943:625494 [0] NCCL INFO Channel 12/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-166-214:624947:625497 [4] NCCL INFO Channel 14/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-166-214:624950:625499 [7] NCCL INFO Channel 00/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-166-214:624949:625496 [6] NCCL INFO Channel 12/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624945:625495 [2] NCCL INFO Channel 08/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-162-46:702887:703467 [7] NCCL INFO Channel 01/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-162-46:702885:703468 [5] NCCL INFO Channel 10/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-162-46:702880:703466 [0] NCCL INFO Channel 08/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-162-46:702882:703472 [2] NCCL INFO Channel 06/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-166-214:624943:625494 [0] NCCL INFO Channel 13/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-166-214:624947:625497 [4] NCCL INFO Channel 15/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-166-214:624950:625499 [7] NCCL INFO Channel 01/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-162-46:702887:703467 [7] NCCL INFO Channel 02/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-162-46:702885:703468 [5] NCCL INFO Channel 11/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-162-46:702880:703466 [0] NCCL INFO Channel 09/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-162-46:702882:703472 [2] NCCL INFO Channel 07/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-166-214:624948:625500 [5] NCCL INFO Channel 00/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-166-214:624949:625496 [6] NCCL INFO Channel 13/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624945:625495 [2] NCCL INFO Channel 09/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-162-46:702887:703467 [7] NCCL INFO Channel 03/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-162-46:702885:703468 [5] NCCL INFO Channel 12/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-162-46:702880:703466 [0] NCCL INFO Channel 10/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-162-46:702882:703472 [2] NCCL INFO Channel 08/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-165-213:9625:10169 [0] NCCL INFO Channel 18/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-162-46:702886:703469 [6] NCCL INFO Channel 09/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624943:625494 [0] NCCL INFO Channel 14/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-166-214:624947:625497 [4] NCCL INFO Channel 16/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-166-214:624950:625499 [7] NCCL INFO Channel 02/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-162-46:702887:703467 [7] NCCL INFO Channel 04/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-162-46:702885:703468 [5] NCCL INFO Channel 13/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-162-46:702880:703466 [0] NCCL INFO Channel 11/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-162-46:702882:703472 [2] NCCL INFO Channel 09/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-162-46:702886:703469 [6] NCCL INFO Channel 10/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624948:625500 [5] NCCL INFO Channel 01/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-166-214:624949:625496 [6] NCCL INFO Channel 14/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624945:625495 [2] NCCL INFO Channel 10/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-165-213:9631:10172 [6] NCCL INFO Channel 20/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624943:625494 [0] NCCL INFO Channel 15/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-166-214:624947:625497 [4] NCCL INFO Channel 17/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-162-46:702887:703467 [7] NCCL INFO Channel 05/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-166-214:624950:625499 [7] NCCL INFO Channel 03/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-162-46:702885:703468 [5] NCCL INFO Channel 14/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-162-46:702881:703465 [1] NCCL INFO Channel 07/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-162-46:702883:703471 [3] NCCL INFO Channel 08/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-162-46:702884:703470 [4] NCCL INFO Channel 06/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-166-214:624948:625500 [5] NCCL INFO Channel 02/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-166-214:624949:625496 [6] NCCL INFO Channel 15/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624945:625495 [2] NCCL INFO Channel 11/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-162-46:702880:703466 [0] NCCL INFO Channel 12/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-162-46:702886:703469 [6] NCCL INFO Channel 11/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-162-46:702882:703472 [2] NCCL INFO Channel 10/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-166-214:624946:625498 [3] NCCL INFO Channel 00/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-162-46:702881:703465 [1] NCCL INFO Channel 08/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-166-214:624947:625497 [4] NCCL INFO Channel 18/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-166-214:624943:625494 [0] NCCL INFO Channel 16/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-166-214:624950:625499 [7] NCCL INFO Channel 04/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-162-46:702885:703468 [5] NCCL INFO Channel 15/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-162-46:702880:703466 [0] NCCL INFO Channel 13/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-162-46:702886:703469 [6] NCCL INFO Channel 12/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-162-46:702882:703472 [2] NCCL INFO Channel 11/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-162-46:702881:703465 [1] NCCL INFO Channel 09/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-166-214:624948:625500 [5] NCCL INFO Channel 03/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-166-214:624949:625496 [6] NCCL INFO Channel 16/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624945:625495 [2] NCCL INFO Channel 12/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-166-214:624946:625498 [3] NCCL INFO Channel 01/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-162-46:702884:703470 [4] NCCL INFO Channel 07/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-162-46:702886:703469 [6] NCCL INFO Channel 13/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-162-46:702885:703468 [5] NCCL INFO Channel 16/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-162-46:702880:703466 [0] NCCL INFO Channel 14/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-162-46:702887:703467 [7] NCCL INFO Channel 06/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-162-46:702881:703465 [1] NCCL INFO Channel 10/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-162-46:702882:703472 [2] NCCL INFO Channel 12/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-166-214:624947:625497 [4] NCCL INFO Channel 19/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-166-214:624943:625494 [0] NCCL INFO Channel 17/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-166-214:624950:625499 [7] NCCL INFO Channel 05/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-166-214:624948:625500 [5] NCCL INFO Channel 04/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-166-214:624949:625496 [6] NCCL INFO Channel 17/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624945:625495 [2] NCCL INFO Channel 13/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-166-214:624946:625498 [3] NCCL INFO Channel 02/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-165-213:9627:10174 [2] NCCL INFO Channel 23/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-165-213:9625:10169 [0] NCCL INFO Channel 19/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-162-46:702884:703470 [4] NCCL INFO Channel 08/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-162-46:702880:703466 [0] NCCL INFO Channel 15/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-162-46:702885:703468 [5] NCCL INFO Channel 17/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-162-46:702887:703467 [7] NCCL INFO Channel 07/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-162-46:702881:703465 [1] NCCL INFO Channel 11/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-162-46:702882:703472 [2] NCCL INFO Channel 13/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-166-214:624947:625497 [4] NCCL INFO Channel 20/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-162-46:702886:703469 [6] NCCL INFO Channel 14/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-162-46:702883:703471 [3] NCCL INFO Channel 09/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-166-214:624950:625499 [7] NCCL INFO Channel 06/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-166-15:127294:127835 [4] NCCL INFO Channel 21/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-167-9:1039541:1040097 [4] NCCL INFO Channel 16/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-166-214:624949:625496 [6] NCCL INFO Channel 18/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624948:625500 [5] NCCL INFO Channel 05/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-166-214:624945:625495 [2] NCCL INFO Channel 14/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-166-214:624946:625498 [3] NCCL INFO Channel 03/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-166-214:624947:625497 [4] NCCL INFO Channel 21/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-162-46:702880:703466 [0] NCCL INFO Channel 16/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-162-46:702884:703470 [4] NCCL INFO Channel 09/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-162-46:702885:703468 [5] NCCL INFO Channel 18/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-162-46:702887:703467 [7] NCCL INFO Channel 08/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-162-46:702882:703472 [2] NCCL INFO Channel 14/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-162-46:702881:703465 [1] NCCL INFO Channel 12/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-162-46:702886:703469 [6] NCCL INFO Channel 15/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127292:127836 [2] NCCL INFO Channel 23/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-166-214:624943:625494 [0] NCCL INFO Channel 18/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-166-214:624950:625499 [7] NCCL INFO Channel 07/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-162-46:702883:703471 [3] NCCL INFO Channel 10/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-165-213:9625:10169 [0] NCCL INFO Channel 20/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-166-214:624949:625496 [6] NCCL INFO Channel 19/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624948:625500 [5] NCCL INFO Channel 06/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-166-214:624945:625495 [2] NCCL INFO Channel 15/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-166-214:624946:625498 [3] NCCL INFO Channel 04/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-166-214:624947:625497 [4] NCCL INFO Channel 22/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-162-46:702884:703470 [4] NCCL INFO Channel 10/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-162-46:702880:703466 [0] NCCL INFO Channel 17/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-162-46:702885:703468 [5] NCCL INFO Channel 19/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-162-46:702887:703467 [7] NCCL INFO Channel 09/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-162-46:702882:703472 [2] NCCL INFO Channel 15/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-162-46:702881:703465 [1] NCCL INFO Channel 13/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-162-46:702886:703469 [6] NCCL INFO Channel 16/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624943:625494 [0] NCCL INFO Channel 19/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-162-46:702883:703471 [3] NCCL INFO Channel 11/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-166-214:624950:625499 [7] NCCL INFO Channel 08/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-166-214:624949:625496 [6] NCCL INFO Channel 20/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624948:625500 [5] NCCL INFO Channel 07/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-166-214:624945:625495 [2] NCCL INFO Channel 16/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-166-214:624946:625498 [3] NCCL INFO Channel 05/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-166-214:624947:625497 [4] NCCL INFO Channel 23/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-166-214:624944:625493 [1] NCCL INFO Channel 00/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-162-46:702880:703466 [0] NCCL INFO Channel 18/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-162-46:702884:703470 [4] NCCL INFO Channel 11/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-162-46:702885:703468 [5] NCCL INFO Channel 20/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-162-46:702887:703467 [7] NCCL INFO Channel 10/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-162-46:702881:703465 [1] NCCL INFO Channel 14/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-162-46:702882:703472 [2] NCCL INFO Channel 16/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-162-46:702886:703469 [6] NCCL INFO Channel 17/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-162-46:702883:703471 [3] NCCL INFO Channel 12/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-166-214:624943:625494 [0] NCCL INFO Channel 20/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-166-214:624950:625499 [7] NCCL INFO Channel 09/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-167-9:1039537:1040095 [0] NCCL INFO Channel 14/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-167-9:1039543:1040098 [6] NCCL INFO Channel 17/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624949:625496 [6] NCCL INFO Channel 21/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624948:625500 [5] NCCL INFO Channel 08/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-166-214:624946:625498 [3] NCCL INFO Channel 06/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-166-15:127294:127835 [4] NCCL INFO Channel 22/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-166-214:624944:625493 [1] NCCL INFO Channel 01/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-162-46:702880:703466 [0] NCCL INFO Channel 19/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-162-46:702884:703470 [4] NCCL INFO Channel 12/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-162-46:702887:703467 [7] NCCL INFO Channel 11/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-162-46:702885:703468 [5] NCCL INFO Channel 21/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-162-46:702882:703472 [2] NCCL INFO Channel 17/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-162-46:702881:703465 [1] NCCL INFO Channel 15/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-162-46:702886:703469 [6] NCCL INFO Channel 18/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624943:625494 [0] NCCL INFO Channel 21/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-162-46:702883:703471 [3] NCCL INFO Channel 13/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-166-214:624950:625499 [7] NCCL INFO Channel 10/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-166-214:624945:625495 [2] NCCL INFO Channel 17/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-165-213:9625:10169 [0] NCCL INFO Channel 21/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-166-214:624949:625496 [6] NCCL INFO Channel 22/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624948:625500 [5] NCCL INFO Channel 09/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-166-214:624946:625498 [3] NCCL INFO Channel 07/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-166-214:624944:625493 [1] NCCL INFO Channel 02/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-162-46:702880:703466 [0] NCCL INFO Channel 20/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-162-46:702884:703470 [4] NCCL INFO Channel 13/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-162-46:702887:703467 [7] NCCL INFO Channel 12/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-162-46:702885:703468 [5] NCCL INFO Channel 22/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-162-46:702882:703472 [2] NCCL INFO Channel 18/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-162-46:702881:703465 [1] NCCL INFO Channel 16/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-162-46:702886:703469 [6] NCCL INFO Channel 19/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624943:625494 [0] NCCL INFO Channel 22/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-162-46:702883:703471 [3] NCCL INFO Channel 14/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-166-214:624950:625499 [7] NCCL INFO Channel 11/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-167-9:1039539:1040100 [2] NCCL INFO Channel 10/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-166-214:624949:625496 [6] NCCL INFO Channel 23/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624945:625495 [2] NCCL INFO Channel 18/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-166-214:624946:625498 [3] NCCL INFO Channel 08/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-166-214:624948:625500 [5] NCCL INFO Channel 10/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-166-214:624944:625493 [1] NCCL INFO Channel 03/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-165-213:9631:10172 [6] NCCL INFO Channel 21/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-162-46:702880:703466 [0] NCCL INFO Channel 21/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-162-46:702884:703470 [4] NCCL INFO Channel 14/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-162-46:702887:703467 [7] NCCL INFO Channel 13/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-162-46:702885:703468 [5] NCCL INFO Channel 23/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-162-46:702881:703465 [1] NCCL INFO Channel 17/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-162-46:702882:703472 [2] NCCL INFO Channel 19/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-166-214:624943:625494 [0] NCCL INFO Channel 23/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-162-46:702886:703469 [6] NCCL INFO Channel 20/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624950:625499 [7] NCCL INFO Channel 12/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-166-214:624945:625495 [2] NCCL INFO Channel 19/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-166-214:624946:625498 [3] NCCL INFO Channel 09/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-162-46:702883:703471 [3] NCCL INFO Channel 15/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-166-214:624944:625493 [1] NCCL INFO Channel 04/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-166-214:624948:625500 [5] NCCL INFO Channel 11/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-166-125:762819:763372 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,fff00000,00000000
+ip-26-0-166-125:762819:763372 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-166-214:624950:625499 [7] NCCL INFO Channel 13/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-166-214:624946:625498 [3] NCCL INFO Channel 10/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-166-214:624945:625495 [2] NCCL INFO Channel 20/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-166-214:624944:625493 [1] NCCL INFO Channel 05/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-162-46:702880:703466 [0] NCCL INFO Channel 22/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-166-214:624948:625500 [5] NCCL INFO Channel 12/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-162-46:702884:703470 [4] NCCL INFO Channel 15/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-162-46:702887:703467 [7] NCCL INFO Channel 14/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-162-46:702881:703465 [1] NCCL INFO Channel 18/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-162-46:702882:703472 [2] NCCL INFO Channel 20/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-162-46:702883:703471 [3] NCCL INFO Channel 16/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-167-9:1039537:1040095 [0] NCCL INFO Channel 15/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-167-9:1039541:1040097 [4] NCCL INFO Channel 17/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-162-46:702880:703466 [0] NCCL INFO Channel 23/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-162-46:702887:703467 [7] NCCL INFO Channel 15/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-162-46:702881:703465 [1] NCCL INFO Channel 19/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-162-46:702886:703469 [6] NCCL INFO Channel 21/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-162-46:702882:703472 [2] NCCL INFO Channel 21/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-166-125:762818:763378 [3] NCCL INFO Setting affinity for GPU 3 to 0fff,ffffffff
+ip-26-0-166-125:762818:763378 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-162-46:702883:703471 [3] NCCL INFO Channel 17/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-167-9:1039539:1040100 [2] NCCL INFO Channel 11/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-165-213:9625:10169 [0] NCCL INFO Channel 22/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-167-9:1039537:1040095 [0] NCCL INFO Channel 16/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-167-9:1039541:1040097 [4] NCCL INFO Channel 18/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-166-214:624950:625499 [7] NCCL INFO Channel 14/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-166-214:624946:625498 [3] NCCL INFO Channel 11/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-166-214:624945:625495 [2] NCCL INFO Channel 21/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-166-214:624944:625493 [1] NCCL INFO Channel 06/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-162-46:702882:703472 [2] NCCL INFO Channel 22/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-162-46:702887:703467 [7] NCCL INFO Channel 16/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-162-46:702884:703470 [4] NCCL INFO Channel 16/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-162-46:702886:703469 [6] NCCL INFO Channel 22/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-162-46:702881:703465 [1] NCCL INFO Channel 20/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-162-46:702883:703471 [3] NCCL INFO Channel 18/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-166-214:624948:625500 [5] NCCL INFO Channel 13/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-162-46:702882:703472 [2] NCCL INFO Channel 23/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-162-46:702887:703467 [7] NCCL INFO Channel 17/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-162-46:702886:703469 [6] NCCL INFO Channel 23/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-162-46:702884:703470 [4] NCCL INFO Channel 17/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-162-46:702881:703465 [1] NCCL INFO Channel 21/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-162-46:702883:703471 [3] NCCL INFO Channel 19/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-167-9:1039539:1040100 [2] NCCL INFO Channel 12/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-166-214:624950:625499 [7] NCCL INFO Channel 15/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-162-46:702884:703470 [4] NCCL INFO Channel 18/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-162-46:702887:703467 [7] NCCL INFO Channel 18/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-162-46:702881:703465 [1] NCCL INFO Channel 22/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-162-46:702883:703471 [3] NCCL INFO Channel 20/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-167-9:1039537:1040095 [0] NCCL INFO Channel 17/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-167-9:1039541:1040097 [4] NCCL INFO Channel 19/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-165-213:9631:10172 [6] NCCL INFO Channel 22/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624945:625495 [2] NCCL INFO Channel 22/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-166-214:624946:625498 [3] NCCL INFO Channel 12/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-166-214:624944:625493 [1] NCCL INFO Channel 07/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-166-214:624948:625500 [5] NCCL INFO Channel 14/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-162-46:702884:703470 [4] NCCL INFO Channel 19/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-162-46:702887:703467 [7] NCCL INFO Channel 19/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-162-46:702881:703465 [1] NCCL INFO Channel 23/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-162-46:702883:703471 [3] NCCL INFO Channel 21/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-162-46:702884:703470 [4] NCCL INFO Channel 20/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-166-214:624950:625499 [7] NCCL INFO Channel 16/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-166-214:624945:625495 [2] NCCL INFO Channel 23/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-166-214:624946:625498 [3] NCCL INFO Channel 13/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-166-214:624944:625493 [1] NCCL INFO Channel 08/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-162-46:702883:703471 [3] NCCL INFO Channel 22/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-162-46:702884:703470 [4] NCCL INFO Channel 21/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-162-46:702887:703467 [7] NCCL INFO Channel 20/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-166-214:624948:625500 [5] NCCL INFO Channel 15/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-162-46:702883:703471 [3] NCCL INFO Channel 23/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-162-46:702884:703470 [4] NCCL INFO Channel 22/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-162-46:702887:703467 [7] NCCL INFO Channel 21/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-166-15:127294:127835 [4] NCCL INFO Channel 23/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-162-46:702884:703470 [4] NCCL INFO Channel 23/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-162-46:702887:703467 [7] NCCL INFO Channel 22/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-165-213:9625:10169 [0] NCCL INFO Channel 23/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-166-214:624950:625499 [7] NCCL INFO Channel 17/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-166-214:624946:625498 [3] NCCL INFO Channel 14/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-166-214:624944:625493 [1] NCCL INFO Channel 09/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-166-214:624948:625500 [5] NCCL INFO Channel 16/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-167-9:1039539:1040100 [2] NCCL INFO Channel 13/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-162-46:702887:703467 [7] NCCL INFO Channel 23/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-167-9:1039537:1040095 [0] NCCL INFO Channel 18/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-167-9:1039543:1040098 [6] NCCL INFO Channel 18/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-167-9:1039541:1040097 [4] NCCL INFO Channel 20/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-167-9:1039539:1040100 [2] NCCL INFO Channel 14/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-166-214:624950:625499 [7] NCCL INFO Channel 18/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-166-214:624946:625498 [3] NCCL INFO Channel 15/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-166-214:624944:625493 [1] NCCL INFO Channel 10/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-166-214:624948:625500 [5] NCCL INFO Channel 17/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-165-213:9631:10172 [6] NCCL INFO Channel 23/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-167-9:1039537:1040095 [0] NCCL INFO Channel 19/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-167-9:1039543:1040098 [6] NCCL INFO Channel 19/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-167-9:1039541:1040097 [4] NCCL INFO Channel 21/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-166-214:624950:625499 [7] NCCL INFO Channel 19/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-166-214:624944:625493 [1] NCCL INFO Channel 11/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-167-9:1039539:1040100 [2] NCCL INFO Channel 15/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-166-214:624946:625498 [3] NCCL INFO Channel 16/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-167-9:1039537:1040095 [0] NCCL INFO Channel 20/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-167-9:1039543:1040098 [6] NCCL INFO Channel 20/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-167-9:1039541:1040097 [4] NCCL INFO Channel 22/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-166-214:624946:625498 [3] NCCL INFO Channel 17/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-166-214:624948:625500 [5] NCCL INFO Channel 18/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-167-9:1039539:1040100 [2] NCCL INFO Channel 16/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-166-214:624950:625499 [7] NCCL INFO Channel 20/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-167-9:1039537:1040095 [0] NCCL INFO Channel 21/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-167-9:1039543:1040098 [6] NCCL INFO Channel 21/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624950:625499 [7] NCCL INFO Channel 21/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-167-9:1039541:1040097 [4] NCCL INFO Channel 23/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-166-214:624948:625500 [5] NCCL INFO Channel 19/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-166-214:624944:625493 [1] NCCL INFO Channel 12/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-167-9:1039539:1040100 [2] NCCL INFO Channel 17/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-167-9:1039537:1040095 [0] NCCL INFO Channel 22/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-167-9:1039543:1040098 [6] NCCL INFO Channel 22/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624950:625499 [7] NCCL INFO Channel 22/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-167-9:1039543:1040098 [6] NCCL INFO Channel 23/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-167-9:1039537:1040095 [0] NCCL INFO Channel 23/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-166-125:762817:763374 [2] NCCL INFO Setting affinity for GPU 2 to 0fff,ffffffff
+ip-26-0-166-125:762817:763374 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-166-214:624944:625493 [1] NCCL INFO Channel 13/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-166-214:624948:625500 [5] NCCL INFO Channel 20/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-166-214:624946:625498 [3] NCCL INFO Channel 18/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-166-36:104429:104973 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-166-214:624950:625499 [7] NCCL INFO Channel 23/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-166-36:104429:104973 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-166-214:624944:625493 [1] NCCL INFO Channel 14/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-167-9:1039539:1040100 [2] NCCL INFO Channel 18/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-166-36:104435:104975 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-166-36:104435:104975 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-166-214:624948:625500 [5] NCCL INFO Channel 21/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-166-214:624946:625498 [3] NCCL INFO Channel 19/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-166-36:104430:104974 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-166-36:104430:104974 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-166-214:624944:625493 [1] NCCL INFO Channel 15/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-166-214:624948:625500 [5] NCCL INFO Channel 22/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-166-214:624944:625493 [1] NCCL INFO Channel 16/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-166-214:624946:625498 [3] NCCL INFO Channel 20/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-166-214:624948:625500 [5] NCCL INFO Channel 23/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-167-9:1039539:1040100 [2] NCCL INFO Channel 19/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-166-214:624944:625493 [1] NCCL INFO Channel 17/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-166-125:762820:763379 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,fff00000,00000000
+ip-26-0-166-125:762820:763379 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-167-9:1039539:1040100 [2] NCCL INFO Channel 20/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-166-214:624944:625493 [1] NCCL INFO Channel 18/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-166-214:624946:625498 [3] NCCL INFO Channel 21/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-166-125:762815:763371 [0] NCCL INFO Setting affinity for GPU 0 to 0fff,ffffffff
+ip-26-0-166-125:762815:763371 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-166-214:624944:625493 [1] NCCL INFO Channel 19/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-166-214:624946:625498 [3] NCCL INFO Channel 22/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-166-125:762820:763379 [5] NCCL INFO Trees [0] 3/-1/-1->2->1 [1] 3/-1/-1->2->1 [2] 3/-1/-1->2->1 [3] 3/-1/-1->2->1 [4] 3/-1/-1->2->1 [5] 3/-1/-1->2->1 [6] 3/-1/-1->2->1 [7] 3/-1/-1->2->1 [8] 3/-1/-1->2->1 [9] 3/-1/-1->2->1 [10] 3/-1/-1->2->1 [11] 3/-1/-1->2->1 [12] 3/-1/-1->2->1 [13] 3/-1/-1->2->1 [14] 3/-1/-1->2->1 [15] 3/-1/-1->2->1 [16] 3/-1/-1->2->1 [17] 3/-1/-1->2->1 [18] 3/-1/-1->2->1 [19] 3/-1/-1->2->1 [20] 3/-1/-1->2->1 [21] 3/-1/-1->2->1 [22] 3/-1/-1->2->1 [23] 3/-1/-1->2->1
+ip-26-0-166-125:762820:763379 [5] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-166-125:762822:763377 [7] NCCL INFO Trees [0] -1/-1/-1->3->2 [1] -1/-1/-1->3->2 [2] -1/-1/-1->3->2 [3] -1/-1/-1->3->2 [4] -1/-1/-1->3->2 [5] -1/-1/-1->3->2 [6] -1/-1/-1->3->2 [7] -1/-1/-1->3->2 [8] -1/-1/-1->3->2 [9] -1/-1/-1->3->2 [10] -1/-1/-1->3->2 [11] -1/-1/-1->3->2 [12] -1/-1/-1->3->2 [13] -1/-1/-1->3->2 [14] -1/-1/-1->3->2 [15] -1/-1/-1->3->2 [16] -1/-1/-1->3->2 [17] -1/-1/-1->3->2 [18] -1/-1/-1->3->2 [19] -1/-1/-1->3->2 [20] -1/-1/-1->3->2 [21] -1/-1/-1->3->2 [22] -1/-1/-1->3->2 [23] -1/-1/-1->3->2
+ip-26-0-166-125:762822:763377 [7] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-166-125:762816:763376 [1] NCCL INFO Channel 00/24 :    0   1   2   3
+ip-26-0-166-125:762816:763376 [1] NCCL INFO Channel 01/24 :    0   1   2   3
+ip-26-0-166-125:762816:763376 [1] NCCL INFO Channel 02/24 :    0   1   2   3
+ip-26-0-166-125:762816:763376 [1] NCCL INFO Channel 03/24 :    0   1   2   3
+ip-26-0-166-125:762816:763376 [1] NCCL INFO Channel 04/24 :    0   1   2   3
+ip-26-0-166-125:762816:763376 [1] NCCL INFO Channel 05/24 :    0   1   2   3
+ip-26-0-166-125:762816:763376 [1] NCCL INFO Channel 06/24 :    0   1   2   3
+ip-26-0-166-125:762816:763376 [1] NCCL INFO Channel 07/24 :    0   1   2   3
+ip-26-0-166-125:762816:763376 [1] NCCL INFO Channel 08/24 :    0   1   2   3
+ip-26-0-166-125:762816:763376 [1] NCCL INFO Channel 09/24 :    0   1   2   3
+ip-26-0-166-125:762816:763376 [1] NCCL INFO Channel 10/24 :    0   1   2   3
+ip-26-0-166-125:762816:763376 [1] NCCL INFO Channel 11/24 :    0   1   2   3
+ip-26-0-166-125:762816:763376 [1] NCCL INFO Channel 12/24 :    0   1   2   3
+ip-26-0-166-125:762816:763376 [1] NCCL INFO Channel 13/24 :    0   1   2   3
+ip-26-0-166-125:762816:763376 [1] NCCL INFO Channel 14/24 :    0   1   2   3
+ip-26-0-166-125:762816:763376 [1] NCCL INFO Channel 15/24 :    0   1   2   3
+ip-26-0-166-125:762816:763376 [1] NCCL INFO Channel 16/24 :    0   1   2   3
+ip-26-0-166-125:762816:763376 [1] NCCL INFO Channel 17/24 :    0   1   2   3
+ip-26-0-166-125:762816:763376 [1] NCCL INFO Channel 18/24 :    0   1   2   3
+ip-26-0-166-125:762816:763376 [1] NCCL INFO Channel 19/24 :    0   1   2   3
+ip-26-0-166-125:762816:763376 [1] NCCL INFO Channel 20/24 :    0   1   2   3
+ip-26-0-166-125:762816:763376 [1] NCCL INFO Channel 21/24 :    0   1   2   3
+ip-26-0-166-125:762816:763376 [1] NCCL INFO Channel 22/24 :    0   1   2   3
+ip-26-0-166-125:762816:763376 [1] NCCL INFO Channel 23/24 :    0   1   2   3
+ip-26-0-166-125:762816:763376 [1] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] 1/-1/-1->0->-1 [7] 1/-1/-1->0->-1 [8] 1/-1/-1->0->-1 [9] 1/-1/-1->0->-1 [10] 1/-1/-1->0->-1 [11] 1/-1/-1->0->-1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] 1/-1/-1->0->-1 [19] 1/-1/-1->0->-1 [20] 1/-1/-1->0->-1 [21] 1/-1/-1->0->-1 [22] 1/-1/-1->0->-1 [23] 1/-1/-1->0->-1
+ip-26-0-166-125:762816:763376 [1] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-166-125:762818:763378 [3] NCCL INFO Trees [0] 2/-1/-1->1->0 [1] 2/-1/-1->1->0 [2] 2/-1/-1->1->0 [3] 2/-1/-1->1->0 [4] 2/-1/-1->1->0 [5] 2/-1/-1->1->0 [6] 2/-1/-1->1->0 [7] 2/-1/-1->1->0 [8] 2/-1/-1->1->0 [9] 2/-1/-1->1->0 [10] 2/-1/-1->1->0 [11] 2/-1/-1->1->0 [12] 2/-1/-1->1->0 [13] 2/-1/-1->1->0 [14] 2/-1/-1->1->0 [15] 2/-1/-1->1->0 [16] 2/-1/-1->1->0 [17] 2/-1/-1->1->0 [18] 2/-1/-1->1->0 [19] 2/-1/-1->1->0 [20] 2/-1/-1->1->0 [21] 2/-1/-1->1->0 [22] 2/-1/-1->1->0 [23] 2/-1/-1->1->0
+ip-26-0-166-125:762818:763378 [3] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-166-214:624944:625493 [1] NCCL INFO Channel 20/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-167-9:1039539:1040100 [2] NCCL INFO Channel 21/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-166-125:762815:763371 [0] NCCL INFO Channel 00/24 :    0   1   2   3
+ip-26-0-166-125:762815:763371 [0] NCCL INFO Channel 01/24 :    0   1   2   3
+ip-26-0-166-125:762815:763371 [0] NCCL INFO Channel 02/24 :    0   1   2   3
+ip-26-0-166-125:762815:763371 [0] NCCL INFO Channel 03/24 :    0   1   2   3
+ip-26-0-166-125:762815:763371 [0] NCCL INFO Channel 04/24 :    0   1   2   3
+ip-26-0-166-125:762815:763371 [0] NCCL INFO Channel 05/24 :    0   1   2   3
+ip-26-0-166-125:762815:763371 [0] NCCL INFO Channel 06/24 :    0   1   2   3
+ip-26-0-166-125:762815:763371 [0] NCCL INFO Channel 07/24 :    0   1   2   3
+ip-26-0-166-125:762815:763371 [0] NCCL INFO Channel 08/24 :    0   1   2   3
+ip-26-0-166-125:762815:763371 [0] NCCL INFO Channel 09/24 :    0   1   2   3
+ip-26-0-166-125:762815:763371 [0] NCCL INFO Channel 10/24 :    0   1   2   3
+ip-26-0-166-125:762815:763371 [0] NCCL INFO Channel 11/24 :    0   1   2   3
+ip-26-0-166-125:762815:763371 [0] NCCL INFO Channel 12/24 :    0   1   2   3
+ip-26-0-166-125:762815:763371 [0] NCCL INFO Channel 13/24 :    0   1   2   3
+ip-26-0-166-125:762821:763373 [6] NCCL INFO Trees [0] -1/-1/-1->3->2 [1] -1/-1/-1->3->2 [2] -1/-1/-1->3->2 [3] -1/-1/-1->3->2 [4] -1/-1/-1->3->2 [5] -1/-1/-1->3->2 [6] -1/-1/-1->3->2 [7] -1/-1/-1->3->2 [8] -1/-1/-1->3->2 [9] -1/-1/-1->3->2 [10] -1/-1/-1->3->2 [11] -1/-1/-1->3->2 [12] -1/-1/-1->3->2 [13] -1/-1/-1->3->2 [14] -1/-1/-1->3->2 [15] -1/-1/-1->3->2 [16] -1/-1/-1->3->2 [17] -1/-1/-1->3->2 [18] -1/-1/-1->3->2 [19] -1/-1/-1->3->2 [20] -1/-1/-1->3->2 [21] -1/-1/-1->3->2 [22] -1/-1/-1->3->2 [23] -1/-1/-1->3->2
+ip-26-0-166-125:762815:763371 [0] NCCL INFO Channel 14/24 :    0   1   2   3
+ip-26-0-166-125:762815:763371 [0] NCCL INFO Channel 15/24 :    0   1   2   3
+ip-26-0-166-125:762821:763373 [6] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-166-125:762815:763371 [0] NCCL INFO Channel 16/24 :    0   1   2   3
+ip-26-0-166-125:762815:763371 [0] NCCL INFO Channel 17/24 :    0   1   2   3
+ip-26-0-166-125:762819:763372 [4] NCCL INFO Trees [0] 3/-1/-1->2->1 [1] 3/-1/-1->2->1 [2] 3/-1/-1->2->1 [3] 3/-1/-1->2->1 [4] 3/-1/-1->2->1 [5] 3/-1/-1->2->1 [6] 3/-1/-1->2->1 [7] 3/-1/-1->2->1 [8] 3/-1/-1->2->1 [9] 3/-1/-1->2->1 [10] 3/-1/-1->2->1 [11] 3/-1/-1->2->1 [12] 3/-1/-1->2->1 [13] 3/-1/-1->2->1 [14] 3/-1/-1->2->1 [15] 3/-1/-1->2->1 [16] 3/-1/-1->2->1 [17] 3/-1/-1->2->1 [18] 3/-1/-1->2->1 [19] 3/-1/-1->2->1 [20] 3/-1/-1->2->1 [21] 3/-1/-1->2->1 [22] 3/-1/-1->2->1 [23] 3/-1/-1->2->1
+ip-26-0-166-125:762815:763371 [0] NCCL INFO Channel 18/24 :    0   1   2   3
+ip-26-0-166-125:762815:763371 [0] NCCL INFO Channel 19/24 :    0   1   2   3
+ip-26-0-166-125:762817:763374 [2] NCCL INFO Trees [0] 2/-1/-1->1->0 [1] 2/-1/-1->1->0 [2] 2/-1/-1->1->0 [3] 2/-1/-1->1->0 [4] 2/-1/-1->1->0 [5] 2/-1/-1->1->0 [6] 2/-1/-1->1->0 [7] 2/-1/-1->1->0 [8] 2/-1/-1->1->0 [9] 2/-1/-1->1->0 [10] 2/-1/-1->1->0 [11] 2/-1/-1->1->0 [12] 2/-1/-1->1->0 [13] 2/-1/-1->1->0 [14] 2/-1/-1->1->0 [15] 2/-1/-1->1->0 [16] 2/-1/-1->1->0 [17] 2/-1/-1->1->0 [18] 2/-1/-1->1->0 [19] 2/-1/-1->1->0 [20] 2/-1/-1->1->0 [21] 2/-1/-1->1->0 [22] 2/-1/-1->1->0 [23] 2/-1/-1->1->0
+ip-26-0-166-125:762815:763371 [0] NCCL INFO Channel 20/24 :    0   1   2   3
+ip-26-0-166-125:762815:763371 [0] NCCL INFO Channel 21/24 :    0   1   2   3
+ip-26-0-166-125:762815:763371 [0] NCCL INFO Channel 22/24 :    0   1   2   3
+ip-26-0-166-125:762817:763374 [2] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-166-125:762819:763372 [4] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-166-125:762815:763371 [0] NCCL INFO Channel 23/24 :    0   1   2   3
+ip-26-0-166-125:762815:763371 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] 1/-1/-1->0->-1 [7] 1/-1/-1->0->-1 [8] 1/-1/-1->0->-1 [9] 1/-1/-1->0->-1 [10] 1/-1/-1->0->-1 [11] 1/-1/-1->0->-1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] 1/-1/-1->0->-1 [19] 1/-1/-1->0->-1 [20] 1/-1/-1->0->-1 [21] 1/-1/-1->0->-1 [22] 1/-1/-1->0->-1 [23] 1/-1/-1->0->-1
+ip-26-0-166-125:762815:763371 [0] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-166-214:624944:625493 [1] NCCL INFO Channel 21/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-166-214:624946:625498 [3] NCCL INFO Channel 23/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-166-214:624944:625493 [1] NCCL INFO Channel 22/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-167-9:1039539:1040100 [2] NCCL INFO Channel 22/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-166-36:104434:104977 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-166-36:104434:104977 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-166-125:762820:763379 [5] NCCL INFO Channel 00/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-166-125:762822:763377 [7] NCCL INFO Channel 00/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-166-125:762816:763376 [1] NCCL INFO Channel 00/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-166-125:762820:763379 [5] NCCL INFO Channel 01/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-166-125:762822:763377 [7] NCCL INFO Channel 01/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-166-125:762818:763378 [3] NCCL INFO Channel 00/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-166-125:762816:763376 [1] NCCL INFO Channel 01/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-166-36:104433:104976 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-166-36:104433:104976 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-166-125:762820:763379 [5] NCCL INFO Channel 02/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-166-125:762818:763378 [3] NCCL INFO Channel 01/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-166-125:762822:763377 [7] NCCL INFO Channel 02/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-166-125:762816:763376 [1] NCCL INFO Channel 02/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-166-125:762820:763379 [5] NCCL INFO Channel 03/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-166-125:762822:763377 [7] NCCL INFO Channel 03/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-166-125:762818:763378 [3] NCCL INFO Channel 02/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-166-125:762816:763376 [1] NCCL INFO Channel 03/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-166-125:762820:763379 [5] NCCL INFO Channel 04/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-166-125:762822:763377 [7] NCCL INFO Channel 04/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-166-125:762818:763378 [3] NCCL INFO Channel 03/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-166-125:762816:763376 [1] NCCL INFO Channel 04/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-167-9:1039540:1040103 [3] NCCL INFO Connected all rings
+ip-26-0-166-125:762820:763379 [5] NCCL INFO Channel 05/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-166-125:762822:763377 [7] NCCL INFO Channel 05/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-166-125:762818:763378 [3] NCCL INFO Channel 04/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-166-125:762816:763376 [1] NCCL INFO Channel 05/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-166-125:762820:763379 [5] NCCL INFO Channel 06/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-166-125:762822:763377 [7] NCCL INFO Channel 06/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-166-125:762818:763378 [3] NCCL INFO Channel 05/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-166-125:762816:763376 [1] NCCL INFO Channel 06/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-167-9:1039538:1040099 [1] NCCL INFO Connected all rings
+ip-26-0-166-125:762820:763379 [5] NCCL INFO Channel 07/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-166-125:762822:763377 [7] NCCL INFO Channel 07/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-166-125:762818:763378 [3] NCCL INFO Channel 06/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-166-125:762816:763376 [1] NCCL INFO Channel 07/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-167-9:1039542:1040102 [5] NCCL INFO Connected all rings
+ip-26-0-167-9:1039539:1040100 [2] NCCL INFO Channel 23/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-166-125:762822:763377 [7] NCCL INFO Channel 08/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-166-125:762818:763378 [3] NCCL INFO Channel 07/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-166-125:762816:763376 [1] NCCL INFO Channel 08/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-166-125:762818:763378 [3] NCCL INFO Channel 08/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-166-125:762822:763377 [7] NCCL INFO Channel 09/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-166-125:762816:763376 [1] NCCL INFO Channel 09/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-166-36:104434:104977 [5] NCCL INFO Trees [0] 3/-1/-1->2->1 [1] 3/-1/-1->2->1 [2] 3/-1/-1->2->1 [3] 3/-1/-1->2->1 [4] 3/-1/-1->2->1 [5] 3/-1/-1->2->1 [6] 3/-1/-1->2->1 [7] 3/-1/-1->2->1 [8] 3/-1/-1->2->1 [9] 3/-1/-1->2->1 [10] 3/-1/-1->2->1 [11] 3/-1/-1->2->1 [12] 3/-1/-1->2->1 [13] 3/-1/-1->2->1 [14] 3/-1/-1->2->1 [15] 3/-1/-1->2->1 [16] 3/-1/-1->2->1 [17] 3/-1/-1->2->1 [18] 3/-1/-1->2->1 [19] 3/-1/-1->2->1 [20] 3/-1/-1->2->1 [21] 3/-1/-1->2->1 [22] 3/-1/-1->2->1 [23] 3/-1/-1->2->1
+ip-26-0-166-36:104434:104977 [5] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-166-36:104430:104974 [1] NCCL INFO Channel 00/24 :    0   1   2   3
+ip-26-0-166-36:104430:104974 [1] NCCL INFO Channel 01/24 :    0   1   2   3
+ip-26-0-166-36:104430:104974 [1] NCCL INFO Channel 02/24 :    0   1   2   3
+ip-26-0-166-36:104430:104974 [1] NCCL INFO Channel 03/24 :    0   1   2   3
+ip-26-0-166-36:104430:104974 [1] NCCL INFO Channel 04/24 :    0   1   2   3
+ip-26-0-166-36:104430:104974 [1] NCCL INFO Channel 05/24 :    0   1   2   3
+ip-26-0-166-36:104430:104974 [1] NCCL INFO Channel 06/24 :    0   1   2   3
+ip-26-0-166-36:104430:104974 [1] NCCL INFO Channel 07/24 :    0   1   2   3
+ip-26-0-166-36:104430:104974 [1] NCCL INFO Channel 08/24 :    0   1   2   3
+ip-26-0-166-36:104430:104974 [1] NCCL INFO Channel 09/24 :    0   1   2   3
+ip-26-0-166-36:104430:104974 [1] NCCL INFO Channel 10/24 :    0   1   2   3
+ip-26-0-166-36:104430:104974 [1] NCCL INFO Channel 11/24 :    0   1   2   3
+ip-26-0-166-36:104432:104980 [3] NCCL INFO Trees [0] 2/-1/-1->1->0 [1] 2/-1/-1->1->0 [2] 2/-1/-1->1->0 [3] 2/-1/-1->1->0 [4] 2/-1/-1->1->0 [5] 2/-1/-1->1->0 [6] 2/-1/-1->1->0 [7] 2/-1/-1->1->0 [8] 2/-1/-1->1->0 [9] 2/-1/-1->1->0 [10] 2/-1/-1->1->0 [11] 2/-1/-1->1->0 [12] 2/-1/-1->1->0 [13] 2/-1/-1->1->0 [14] 2/-1/-1->1->0 [15] 2/-1/-1->1->0 [16] 2/-1/-1->1->0 [17] 2/-1/-1->1->0 [18] 2/-1/-1->1->0 [19] 2/-1/-1->1->0 [20] 2/-1/-1->1->0 [21] 2/-1/-1->1->0 [22] 2/-1/-1->1->0 [23] 2/-1/-1->1->0
+ip-26-0-166-36:104430:104974 [1] NCCL INFO Channel 12/24 :    0   1   2   3
+ip-26-0-166-36:104430:104974 [1] NCCL INFO Channel 13/24 :    0   1   2   3
+ip-26-0-166-36:104432:104980 [3] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-166-36:104430:104974 [1] NCCL INFO Channel 14/24 :    0   1   2   3
+ip-26-0-166-36:104430:104974 [1] NCCL INFO Channel 15/24 :    0   1   2   3
+ip-26-0-166-36:104430:104974 [1] NCCL INFO Channel 16/24 :    0   1   2   3
+ip-26-0-166-36:104430:104974 [1] NCCL INFO Channel 17/24 :    0   1   2   3
+ip-26-0-166-36:104430:104974 [1] NCCL INFO Channel 18/24 :    0   1   2   3
+ip-26-0-166-36:104430:104974 [1] NCCL INFO Channel 19/24 :    0   1   2   3
+ip-26-0-166-36:104430:104974 [1] NCCL INFO Channel 20/24 :    0   1   2   3
+ip-26-0-166-36:104430:104974 [1] NCCL INFO Channel 21/24 :    0   1   2   3
+ip-26-0-166-36:104436:104978 [7] NCCL INFO Trees [0] -1/-1/-1->3->2 [1] -1/-1/-1->3->2 [2] -1/-1/-1->3->2 [3] -1/-1/-1->3->2 [4] -1/-1/-1->3->2 [5] -1/-1/-1->3->2 [6] -1/-1/-1->3->2 [7] -1/-1/-1->3->2 [8] -1/-1/-1->3->2 [9] -1/-1/-1->3->2 [10] -1/-1/-1->3->2 [11] -1/-1/-1->3->2 [12] -1/-1/-1->3->2 [13] -1/-1/-1->3->2 [14] -1/-1/-1->3->2 [15] -1/-1/-1->3->2 [16] -1/-1/-1->3->2 [17] -1/-1/-1->3->2 [18] -1/-1/-1->3->2 [19] -1/-1/-1->3->2 [20] -1/-1/-1->3->2 [21] -1/-1/-1->3->2 [22] -1/-1/-1->3->2 [23] -1/-1/-1->3->2
+ip-26-0-166-36:104430:104974 [1] NCCL INFO Channel 22/24 :    0   1   2   3
+ip-26-0-166-36:104430:104974 [1] NCCL INFO Channel 23/24 :    0   1   2   3
+ip-26-0-166-36:104436:104978 [7] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-166-36:104430:104974 [1] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] 1/-1/-1->0->-1 [7] 1/-1/-1->0->-1 [8] 1/-1/-1->0->-1 [9] 1/-1/-1->0->-1 [10] 1/-1/-1->0->-1 [11] 1/-1/-1->0->-1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] 1/-1/-1->0->-1 [19] 1/-1/-1->0->-1 [20] 1/-1/-1->0->-1 [21] 1/-1/-1->0->-1 [22] 1/-1/-1->0->-1 [23] 1/-1/-1->0->-1
+ip-26-0-166-36:104430:104974 [1] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-166-125:762822:763377 [7] NCCL INFO Channel 10/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-166-125:762818:763378 [3] NCCL INFO Channel 09/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-166-125:762816:763376 [1] NCCL INFO Channel 10/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-166-125:762819:763372 [4] NCCL INFO Channel 00/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-166-125:762821:763373 [6] NCCL INFO Channel 00/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624944:625493 [1] NCCL INFO Channel 23/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-166-125:762817:763374 [2] NCCL INFO Channel 00/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-166-125:762822:763377 [7] NCCL INFO Channel 11/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-166-125:762816:763376 [1] NCCL INFO Channel 11/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-166-125:762818:763378 [3] NCCL INFO Channel 10/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-166-125:762815:763371 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-166-125:762819:763372 [4] NCCL INFO Channel 01/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-166-36:104433:104976 [4] NCCL INFO Trees [0] 3/-1/-1->2->1 [1] 3/-1/-1->2->1 [2] 3/-1/-1->2->1 [3] 3/-1/-1->2->1 [4] 3/-1/-1->2->1 [5] 3/-1/-1->2->1 [6] 3/-1/-1->2->1 [7] 3/-1/-1->2->1 [8] 3/-1/-1->2->1 [9] 3/-1/-1->2->1 [10] 3/-1/-1->2->1 [11] 3/-1/-1->2->1 [12] 3/-1/-1->2->1 [13] 3/-1/-1->2->1 [14] 3/-1/-1->2->1 [15] 3/-1/-1->2->1 [16] 3/-1/-1->2->1 [17] 3/-1/-1->2->1 [18] 3/-1/-1->2->1 [19] 3/-1/-1->2->1 [20] 3/-1/-1->2->1 [21] 3/-1/-1->2->1 [22] 3/-1/-1->2->1 [23] 3/-1/-1->2->1
+ip-26-0-166-36:104429:104973 [0] NCCL INFO Channel 00/24 :    0   1   2   3
+ip-26-0-166-36:104433:104976 [4] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-166-36:104429:104973 [0] NCCL INFO Channel 01/24 :    0   1   2   3
+ip-26-0-166-36:104429:104973 [0] NCCL INFO Channel 02/24 :    0   1   2   3
+ip-26-0-166-36:104429:104973 [0] NCCL INFO Channel 03/24 :    0   1   2   3
+ip-26-0-166-36:104429:104973 [0] NCCL INFO Channel 04/24 :    0   1   2   3
+ip-26-0-166-36:104429:104973 [0] NCCL INFO Channel 05/24 :    0   1   2   3
+ip-26-0-166-36:104429:104973 [0] NCCL INFO Channel 06/24 :    0   1   2   3
+ip-26-0-166-36:104429:104973 [0] NCCL INFO Channel 07/24 :    0   1   2   3
+ip-26-0-166-36:104429:104973 [0] NCCL INFO Channel 08/24 :    0   1   2   3
+ip-26-0-166-36:104435:104975 [6] NCCL INFO Trees [0] -1/-1/-1->3->2 [1] -1/-1/-1->3->2 [2] -1/-1/-1->3->2 [3] -1/-1/-1->3->2 [4] -1/-1/-1->3->2 [5] -1/-1/-1->3->2 [6] -1/-1/-1->3->2 [7] -1/-1/-1->3->2 [8] -1/-1/-1->3->2 [9] -1/-1/-1->3->2 [10] -1/-1/-1->3->2 [11] -1/-1/-1->3->2 [12] -1/-1/-1->3->2 [13] -1/-1/-1->3->2 [14] -1/-1/-1->3->2 [15] -1/-1/-1->3->2 [16] -1/-1/-1->3->2 [17] -1/-1/-1->3->2 [18] -1/-1/-1->3->2 [19] -1/-1/-1->3->2 [20] -1/-1/-1->3->2 [21] -1/-1/-1->3->2 [22] -1/-1/-1->3->2 [23] -1/-1/-1->3->2
+ip-26-0-166-36:104429:104973 [0] NCCL INFO Channel 09/24 :    0   1   2   3
+ip-26-0-166-36:104429:104973 [0] NCCL INFO Channel 10/24 :    0   1   2   3
+ip-26-0-166-36:104429:104973 [0] NCCL INFO Channel 11/24 :    0   1   2   3
+ip-26-0-166-36:104429:104973 [0] NCCL INFO Channel 12/24 :    0   1   2   3
+ip-26-0-166-36:104429:104973 [0] NCCL INFO Channel 13/24 :    0   1   2   3
+ip-26-0-166-36:104429:104973 [0] NCCL INFO Channel 14/24 :    0   1   2   3
+ip-26-0-166-36:104429:104973 [0] NCCL INFO Channel 15/24 :    0   1   2   3
+ip-26-0-166-36:104435:104975 [6] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-166-36:104431:104979 [2] NCCL INFO Trees [0] 2/-1/-1->1->0 [1] 2/-1/-1->1->0 [2] 2/-1/-1->1->0 [3] 2/-1/-1->1->0 [4] 2/-1/-1->1->0 [5] 2/-1/-1->1->0 [6] 2/-1/-1->1->0 [7] 2/-1/-1->1->0 [8] 2/-1/-1->1->0 [9] 2/-1/-1->1->0 [10] 2/-1/-1->1->0 [11] 2/-1/-1->1->0 [12] 2/-1/-1->1->0 [13] 2/-1/-1->1->0 [14] 2/-1/-1->1->0 [15] 2/-1/-1->1->0 [16] 2/-1/-1->1->0 [17] 2/-1/-1->1->0 [18] 2/-1/-1->1->0 [19] 2/-1/-1->1->0 [20] 2/-1/-1->1->0 [21] 2/-1/-1->1->0 [22] 2/-1/-1->1->0 [23] 2/-1/-1->1->0
+ip-26-0-166-36:104429:104973 [0] NCCL INFO Channel 16/24 :    0   1   2   3
+ip-26-0-166-36:104429:104973 [0] NCCL INFO Channel 17/24 :    0   1   2   3
+ip-26-0-166-36:104431:104979 [2] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-166-36:104429:104973 [0] NCCL INFO Channel 18/24 :    0   1   2   3
+ip-26-0-166-36:104429:104973 [0] NCCL INFO Channel 19/24 :    0   1   2   3
+ip-26-0-166-36:104429:104973 [0] NCCL INFO Channel 20/24 :    0   1   2   3
+ip-26-0-166-36:104429:104973 [0] NCCL INFO Channel 21/24 :    0   1   2   3
+ip-26-0-166-36:104429:104973 [0] NCCL INFO Channel 22/24 :    0   1   2   3
+ip-26-0-166-36:104429:104973 [0] NCCL INFO Channel 23/24 :    0   1   2   3
+ip-26-0-166-36:104429:104973 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] 1/-1/-1->0->-1 [7] 1/-1/-1->0->-1 [8] 1/-1/-1->0->-1 [9] 1/-1/-1->0->-1 [10] 1/-1/-1->0->-1 [11] 1/-1/-1->0->-1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] 1/-1/-1->0->-1 [19] 1/-1/-1->0->-1 [20] 1/-1/-1->0->-1 [21] 1/-1/-1->0->-1 [22] 1/-1/-1->0->-1 [23] 1/-1/-1->0->-1
+ip-26-0-166-36:104429:104973 [0] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-166-125:762821:763373 [6] NCCL INFO Channel 01/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762817:763374 [2] NCCL INFO Channel 01/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-166-125:762822:763377 [7] NCCL INFO Channel 12/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-166-125:762818:763378 [3] NCCL INFO Channel 11/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-166-125:762816:763376 [1] NCCL INFO Channel 12/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-166-125:762815:763371 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-166-125:762819:763372 [4] NCCL INFO Channel 02/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-166-125:762821:763373 [6] NCCL INFO Channel 02/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762820:763379 [5] NCCL INFO Channel 08/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-166-125:762817:763374 [2] NCCL INFO Channel 02/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-166-125:762822:763377 [7] NCCL INFO Channel 13/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-166-125:762818:763378 [3] NCCL INFO Channel 12/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-166-125:762816:763376 [1] NCCL INFO Channel 13/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-166-125:762815:763371 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-166-125:762819:763372 [4] NCCL INFO Channel 03/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-166-125:762821:763373 [6] NCCL INFO Channel 03/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762820:763379 [5] NCCL INFO Channel 09/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-166-125:762817:763374 [2] NCCL INFO Channel 03/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-166-125:762822:763377 [7] NCCL INFO Channel 14/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-166-125:762815:763371 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-166-125:762816:763376 [1] NCCL INFO Channel 14/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-166-125:762818:763378 [3] NCCL INFO Channel 13/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-166-125:762819:763372 [4] NCCL INFO Channel 04/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-166-125:762821:763373 [6] NCCL INFO Channel 04/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762820:763379 [5] NCCL INFO Channel 10/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-166-125:762817:763374 [2] NCCL INFO Channel 04/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-166-125:762822:763377 [7] NCCL INFO Channel 15/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-166-125:762815:763371 [0] NCCL INFO Channel 04/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-166-125:762818:763378 [3] NCCL INFO Channel 14/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-166-125:762816:763376 [1] NCCL INFO Channel 15/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-166-125:762819:763372 [4] NCCL INFO Channel 05/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-166-125:762821:763373 [6] NCCL INFO Channel 05/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762820:763379 [5] NCCL INFO Channel 11/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-166-125:762817:763374 [2] NCCL INFO Channel 05/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-166-125:762822:763377 [7] NCCL INFO Channel 16/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-166-125:762815:763371 [0] NCCL INFO Channel 05/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-166-125:762818:763378 [3] NCCL INFO Channel 15/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-166-125:762816:763376 [1] NCCL INFO Channel 16/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-166-125:762819:763372 [4] NCCL INFO Channel 06/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-166-125:762821:763373 [6] NCCL INFO Channel 06/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762820:763379 [5] NCCL INFO Channel 12/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-166-125:762817:763374 [2] NCCL INFO Channel 06/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-166-125:762822:763377 [7] NCCL INFO Channel 17/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-166-125:762815:763371 [0] NCCL INFO Channel 06/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-166-125:762816:763376 [1] NCCL INFO Channel 17/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-166-125:762818:763378 [3] NCCL INFO Channel 16/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-166-125:762819:763372 [4] NCCL INFO Channel 07/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-166-125:762821:763373 [6] NCCL INFO Channel 07/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762820:763379 [5] NCCL INFO Channel 13/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-166-125:762817:763374 [2] NCCL INFO Channel 07/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-166-125:762822:763377 [7] NCCL INFO Channel 18/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-166-125:762815:763371 [0] NCCL INFO Channel 07/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-166-125:762816:763376 [1] NCCL INFO Channel 18/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-166-125:762818:763378 [3] NCCL INFO Channel 17/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-166-125:762819:763372 [4] NCCL INFO Channel 08/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-166-125:762821:763373 [6] NCCL INFO Channel 08/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762820:763379 [5] NCCL INFO Channel 14/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-166-125:762817:763374 [2] NCCL INFO Channel 08/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-166-125:762822:763377 [7] NCCL INFO Channel 19/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-166-125:762815:763371 [0] NCCL INFO Channel 08/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-166-125:762818:763378 [3] NCCL INFO Channel 18/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-166-125:762819:763372 [4] NCCL INFO Channel 09/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-166-125:762816:763376 [1] NCCL INFO Channel 19/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-166-125:762821:763373 [6] NCCL INFO Channel 09/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762820:763379 [5] NCCL INFO Channel 15/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-166-125:762817:763374 [2] NCCL INFO Channel 09/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-166-125:762822:763377 [7] NCCL INFO Channel 20/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-166-125:762815:763371 [0] NCCL INFO Channel 09/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-166-125:762818:763378 [3] NCCL INFO Channel 19/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-166-125:762816:763376 [1] NCCL INFO Channel 20/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-166-125:762819:763372 [4] NCCL INFO Channel 10/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-166-125:762821:763373 [6] NCCL INFO Channel 10/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762820:763379 [5] NCCL INFO Channel 16/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-166-125:762817:763374 [2] NCCL INFO Channel 10/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-166-125:762815:763371 [0] NCCL INFO Channel 10/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-166-125:762822:763377 [7] NCCL INFO Channel 21/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-166-125:762818:763378 [3] NCCL INFO Channel 20/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-166-125:762819:763372 [4] NCCL INFO Channel 11/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-166-125:762816:763376 [1] NCCL INFO Channel 21/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-166-125:762821:763373 [6] NCCL INFO Channel 11/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762820:763379 [5] NCCL INFO Channel 17/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-166-125:762817:763374 [2] NCCL INFO Channel 11/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-166-125:762815:763371 [0] NCCL INFO Channel 11/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-166-125:762822:763377 [7] NCCL INFO Channel 22/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-166-125:762818:763378 [3] NCCL INFO Channel 21/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-166-125:762816:763376 [1] NCCL INFO Channel 22/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-166-125:762819:763372 [4] NCCL INFO Channel 12/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-166-125:762821:763373 [6] NCCL INFO Channel 12/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762820:763379 [5] NCCL INFO Channel 18/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-166-125:762817:763374 [2] NCCL INFO Channel 12/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-166-125:762816:763376 [1] NCCL INFO Channel 23/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-166-125:762815:763371 [0] NCCL INFO Channel 12/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-166-125:762819:763372 [4] NCCL INFO Channel 13/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-166-125:762822:763377 [7] NCCL INFO Channel 23/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-166-125:762821:763373 [6] NCCL INFO Channel 13/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762817:763374 [2] NCCL INFO Channel 13/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-166-125:762819:763372 [4] NCCL INFO Channel 14/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-166-125:762815:763371 [0] NCCL INFO Channel 13/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-166-125:762821:763373 [6] NCCL INFO Channel 14/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762817:763374 [2] NCCL INFO Channel 14/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-166-125:762819:763372 [4] NCCL INFO Channel 15/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-166-125:762815:763371 [0] NCCL INFO Channel 14/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-166-125:762821:763373 [6] NCCL INFO Channel 15/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762817:763374 [2] NCCL INFO Channel 15/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-166-36:104434:104977 [5] NCCL INFO Channel 00/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-166-36:104430:104974 [1] NCCL INFO Channel 00/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-166-125:762818:763378 [3] NCCL INFO Channel 22/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-166-125:762820:763379 [5] NCCL INFO Channel 19/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-166-125:762819:763372 [4] NCCL INFO Channel 16/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-166-125:762815:763371 [0] NCCL INFO Channel 15/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-166-125:762821:763373 [6] NCCL INFO Channel 16/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762817:763374 [2] NCCL INFO Channel 16/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-166-36:104436:104978 [7] NCCL INFO Channel 00/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-166-36:104434:104977 [5] NCCL INFO Channel 01/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-166-36:104430:104974 [1] NCCL INFO Channel 01/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-166-125:762820:763379 [5] NCCL INFO Channel 20/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-166-125:762819:763372 [4] NCCL INFO Channel 17/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-166-125:762815:763371 [0] NCCL INFO Channel 16/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-166-125:762818:763378 [3] NCCL INFO Channel 23/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-166-125:762821:763373 [6] NCCL INFO Channel 17/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762817:763374 [2] NCCL INFO Channel 17/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-166-36:104436:104978 [7] NCCL INFO Channel 01/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-166-36:104434:104977 [5] NCCL INFO Channel 02/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-166-36:104430:104974 [1] NCCL INFO Channel 02/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-167-9:1039544:1040101 [7] NCCL INFO Connected all rings
+ip-26-0-167-9:1039544:1040101 [7] NCCL INFO Channel 00/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-166-125:762820:763379 [5] NCCL INFO Channel 21/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-166-125:762819:763372 [4] NCCL INFO Channel 18/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-166-125:762821:763373 [6] NCCL INFO Channel 18/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762815:763371 [0] NCCL INFO Channel 17/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-166-125:762817:763374 [2] NCCL INFO Channel 18/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-166-36:104431:104979 [2] NCCL INFO Channel 00/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-166-36:104436:104978 [7] NCCL INFO Channel 02/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-166-36:104432:104980 [3] NCCL INFO Channel 00/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-166-36:104434:104977 [5] NCCL INFO Channel 03/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-166-36:104430:104974 [1] NCCL INFO Channel 03/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-166-36:104431:104979 [2] NCCL INFO Channel 01/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-166-36:104436:104978 [7] NCCL INFO Channel 03/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-166-36:104434:104977 [5] NCCL INFO Channel 04/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-166-36:104433:104976 [4] NCCL INFO Channel 00/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-166-125:762819:763372 [4] NCCL INFO Channel 19/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-166-125:762815:763371 [0] NCCL INFO Channel 18/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-166-36:104432:104980 [3] NCCL INFO Channel 01/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-166-36:104430:104974 [1] NCCL INFO Channel 04/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-166-125:762821:763373 [6] NCCL INFO Channel 19/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762817:763374 [2] NCCL INFO Channel 19/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-166-125:762819:763372 [4] NCCL INFO Channel 20/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-166-125:762815:763371 [0] NCCL INFO Channel 19/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-166-36:104431:104979 [2] NCCL INFO Channel 02/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-166-36:104434:104977 [5] NCCL INFO Channel 05/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-166-36:104436:104978 [7] NCCL INFO Channel 04/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-166-125:762821:763373 [6] NCCL INFO Channel 20/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-166-36:104433:104976 [4] NCCL INFO Channel 01/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-166-125:762817:763374 [2] NCCL INFO Channel 20/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-166-125:762819:763372 [4] NCCL INFO Channel 21/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-166-125:762815:763371 [0] NCCL INFO Channel 20/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-166-36:104430:104974 [1] NCCL INFO Channel 05/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-166-125:762821:763373 [6] NCCL INFO Channel 21/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-166-36:104432:104980 [3] NCCL INFO Channel 02/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-166-125:762819:763372 [4] NCCL INFO Channel 22/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-166-125:762817:763374 [2] NCCL INFO Channel 21/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-166-125:762815:763371 [0] NCCL INFO Channel 21/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-166-36:104431:104979 [2] NCCL INFO Channel 03/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-166-125:762821:763373 [6] NCCL INFO Channel 22/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-166-36:104434:104977 [5] NCCL INFO Channel 06/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-166-36:104436:104978 [7] NCCL INFO Channel 05/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-166-36:104433:104976 [4] NCCL INFO Channel 02/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-166-36:104430:104974 [1] NCCL INFO Channel 06/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-166-125:762819:763372 [4] NCCL INFO Channel 23/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-166-125:762817:763374 [2] NCCL INFO Channel 22/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-166-125:762815:763371 [0] NCCL INFO Channel 22/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-166-125:762821:763373 [6] NCCL INFO Channel 23/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-166-36:104432:104980 [3] NCCL INFO Channel 03/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-166-125:762820:763379 [5] NCCL INFO Channel 22/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-166-36:104431:104979 [2] NCCL INFO Channel 04/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-166-36:104434:104977 [5] NCCL INFO Channel 07/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-166-36:104436:104978 [7] NCCL INFO Channel 06/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-166-125:762817:763374 [2] NCCL INFO Channel 23/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-166-125:762815:763371 [0] NCCL INFO Channel 23/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-166-125:762820:763379 [5] NCCL INFO Channel 23/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-166-36:104433:104976 [4] NCCL INFO Channel 03/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-166-36:104430:104974 [1] NCCL INFO Channel 07/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-166-36:104435:104975 [6] NCCL INFO Channel 00/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-166-36:104429:104973 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-166-36:104432:104980 [3] NCCL INFO Channel 04/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-166-36:104431:104979 [2] NCCL INFO Channel 05/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-166-36:104434:104977 [5] NCCL INFO Channel 08/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-166-36:104436:104978 [7] NCCL INFO Channel 07/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-166-36:104433:104976 [4] NCCL INFO Channel 04/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-166-36:104430:104974 [1] NCCL INFO Channel 08/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-166-36:104435:104975 [6] NCCL INFO Channel 01/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-166-36:104429:104973 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-166-36:104432:104980 [3] NCCL INFO Channel 05/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-166-36:104431:104979 [2] NCCL INFO Channel 06/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-166-36:104436:104978 [7] NCCL INFO Channel 08/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-166-36:104434:104977 [5] NCCL INFO Channel 09/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-166-36:104430:104974 [1] NCCL INFO Channel 09/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-166-36:104433:104976 [4] NCCL INFO Channel 05/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-165-213:9628:10177 [3] NCCL INFO Connected all rings
+ip-26-0-166-36:104435:104975 [6] NCCL INFO Channel 02/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-166-36:104429:104973 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-166-36:104432:104980 [3] NCCL INFO Channel 06/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-166-36:104431:104979 [2] NCCL INFO Channel 07/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-166-36:104434:104977 [5] NCCL INFO Channel 10/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-166-36:104436:104978 [7] NCCL INFO Channel 09/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-166-36:104430:104974 [1] NCCL INFO Channel 10/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-166-36:104433:104976 [4] NCCL INFO Channel 06/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-166-36:104435:104975 [6] NCCL INFO Channel 03/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-166-36:104429:104973 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-166-36:104432:104980 [3] NCCL INFO Channel 07/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-167-9:1039544:1040101 [7] NCCL INFO Channel 01/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-166-36:104431:104979 [2] NCCL INFO Channel 08/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-166-36:104434:104977 [5] NCCL INFO Channel 11/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-166-36:104436:104978 [7] NCCL INFO Channel 10/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-166-36:104430:104974 [1] NCCL INFO Channel 11/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-166-36:104433:104976 [4] NCCL INFO Channel 07/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-166-36:104435:104975 [6] NCCL INFO Channel 04/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-166-36:104429:104973 [0] NCCL INFO Channel 04/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-166-36:104432:104980 [3] NCCL INFO Channel 08/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-166-36:104431:104979 [2] NCCL INFO Channel 09/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-166-36:104434:104977 [5] NCCL INFO Channel 12/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-166-36:104436:104978 [7] NCCL INFO Channel 11/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-166-36:104430:104974 [1] NCCL INFO Channel 12/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-166-36:104435:104975 [6] NCCL INFO Channel 05/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-166-36:104429:104973 [0] NCCL INFO Channel 05/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-166-36:104433:104976 [4] NCCL INFO Channel 08/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-166-36:104432:104980 [3] NCCL INFO Channel 09/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-166-36:104430:104974 [1] NCCL INFO Channel 13/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-166-36:104432:104980 [3] NCCL INFO Channel 10/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-166-36:104431:104979 [2] NCCL INFO Channel 10/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-166-36:104436:104978 [7] NCCL INFO Channel 12/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-166-36:104430:104974 [1] NCCL INFO Channel 14/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-166-36:104432:104980 [3] NCCL INFO Channel 11/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-166-36:104431:104979 [2] NCCL INFO Channel 11/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-166-36:104436:104978 [7] NCCL INFO Channel 13/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-166-36:104430:104974 [1] NCCL INFO Channel 15/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-166-36:104432:104980 [3] NCCL INFO Channel 12/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-166-36:104436:104978 [7] NCCL INFO Channel 14/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-166-36:104431:104979 [2] NCCL INFO Channel 12/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-166-36:104430:104974 [1] NCCL INFO Channel 16/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-166-36:104432:104980 [3] NCCL INFO Channel 13/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-166-36:104436:104978 [7] NCCL INFO Channel 15/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-166-36:104431:104979 [2] NCCL INFO Channel 13/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-166-36:104430:104974 [1] NCCL INFO Channel 17/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-166-36:104432:104980 [3] NCCL INFO Channel 14/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-166-36:104436:104978 [7] NCCL INFO Channel 16/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-166-36:104431:104979 [2] NCCL INFO Channel 14/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-167-9:1039544:1040101 [7] NCCL INFO Channel 02/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-166-36:104430:104974 [1] NCCL INFO Channel 18/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-166-36:104434:104977 [5] NCCL INFO Channel 13/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-166-36:104432:104980 [3] NCCL INFO Channel 15/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-166-36:104435:104975 [6] NCCL INFO Channel 06/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-166-36:104433:104976 [4] NCCL INFO Channel 09/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-166-36:104429:104973 [0] NCCL INFO Channel 06/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-166-36:104436:104978 [7] NCCL INFO Channel 17/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-166-36:104431:104979 [2] NCCL INFO Channel 15/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-166-36:104430:104974 [1] NCCL INFO Channel 19/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-166-36:104434:104977 [5] NCCL INFO Channel 14/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-166-36:104432:104980 [3] NCCL INFO Channel 16/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-166-36:104435:104975 [6] NCCL INFO Channel 07/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-166-36:104429:104973 [0] NCCL INFO Channel 07/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-166-36:104433:104976 [4] NCCL INFO Channel 10/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-166-36:104436:104978 [7] NCCL INFO Channel 18/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-166-36:104431:104979 [2] NCCL INFO Channel 16/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-166-36:104430:104974 [1] NCCL INFO Channel 20/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-166-36:104434:104977 [5] NCCL INFO Channel 15/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-166-36:104432:104980 [3] NCCL INFO Channel 17/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-166-36:104435:104975 [6] NCCL INFO Channel 08/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-166-36:104429:104973 [0] NCCL INFO Channel 08/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-166-36:104433:104976 [4] NCCL INFO Channel 11/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-166-36:104436:104978 [7] NCCL INFO Channel 19/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-166-36:104431:104979 [2] NCCL INFO Channel 17/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-166-36:104434:104977 [5] NCCL INFO Channel 16/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-166-36:104432:104980 [3] NCCL INFO Channel 18/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-166-36:104429:104973 [0] NCCL INFO Channel 09/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-166-36:104435:104975 [6] NCCL INFO Channel 09/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-166-36:104433:104976 [4] NCCL INFO Channel 12/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-166-36:104436:104978 [7] NCCL INFO Channel 20/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-166-36:104431:104979 [2] NCCL INFO Channel 18/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-166-36:104432:104980 [3] NCCL INFO Channel 19/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-166-36:104434:104977 [5] NCCL INFO Channel 17/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-166-36:104435:104975 [6] NCCL INFO Channel 10/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-166-36:104429:104973 [0] NCCL INFO Channel 10/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-166-36:104433:104976 [4] NCCL INFO Channel 13/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-166-36:104436:104978 [7] NCCL INFO Channel 21/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-166-36:104431:104979 [2] NCCL INFO Channel 19/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-166-36:104434:104977 [5] NCCL INFO Channel 18/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-166-36:104432:104980 [3] NCCL INFO Channel 20/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-166-36:104429:104973 [0] NCCL INFO Channel 11/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-166-36:104435:104975 [6] NCCL INFO Channel 11/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-166-36:104436:104978 [7] NCCL INFO Channel 22/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-166-36:104433:104976 [4] NCCL INFO Channel 14/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-166-36:104431:104979 [2] NCCL INFO Channel 20/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-167-9:1039544:1040101 [7] NCCL INFO Channel 03/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-166-36:104434:104977 [5] NCCL INFO Channel 19/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-166-36:104432:104980 [3] NCCL INFO Channel 21/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-166-36:104429:104973 [0] NCCL INFO Channel 12/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-166-36:104435:104975 [6] NCCL INFO Channel 12/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-166-36:104436:104978 [7] NCCL INFO Channel 23/0 : 3[7] -> 0[1] via P2P/IPC
+ip-26-0-166-36:104433:104976 [4] NCCL INFO Channel 15/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-166-36:104431:104979 [2] NCCL INFO Channel 21/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-166-36:104430:104974 [1] NCCL INFO Channel 21/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-166-36:104434:104977 [5] NCCL INFO Channel 20/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-166-36:104432:104980 [3] NCCL INFO Channel 22/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-166-36:104429:104973 [0] NCCL INFO Channel 13/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-166-36:104435:104975 [6] NCCL INFO Channel 13/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-166-36:104431:104979 [2] NCCL INFO Channel 22/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-166-36:104433:104976 [4] NCCL INFO Channel 16/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-166-36:104430:104974 [1] NCCL INFO Channel 22/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-166-36:104434:104977 [5] NCCL INFO Channel 21/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-166-36:104432:104980 [3] NCCL INFO Channel 23/0 : 1[3] -> 2[5] via P2P/IPC
+ip-26-0-166-36:104429:104973 [0] NCCL INFO Channel 14/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-166-36:104435:104975 [6] NCCL INFO Channel 14/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-166-36:104431:104979 [2] NCCL INFO Channel 23/0 : 1[2] -> 2[4] via P2P/IPC
+ip-26-0-166-36:104433:104976 [4] NCCL INFO Channel 17/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-166-36:104430:104974 [1] NCCL INFO Channel 23/0 : 0[1] -> 1[3] via P2P/IPC
+ip-26-0-166-36:104434:104977 [5] NCCL INFO Channel 22/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-166-36:104429:104973 [0] NCCL INFO Channel 15/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-166-36:104435:104975 [6] NCCL INFO Channel 15/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-166-36:104433:104976 [4] NCCL INFO Channel 18/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-167-9:1039544:1040101 [7] NCCL INFO Channel 04/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-166-36:104429:104973 [0] NCCL INFO Channel 16/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-166-36:104434:104977 [5] NCCL INFO Channel 23/0 : 2[5] -> 3[7] via P2P/IPC
+ip-26-0-166-36:104435:104975 [6] NCCL INFO Channel 16/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-166-36:104433:104976 [4] NCCL INFO Channel 19/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-166-36:104435:104975 [6] NCCL INFO Channel 17/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-166-36:104429:104973 [0] NCCL INFO Channel 17/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-166-36:104435:104975 [6] NCCL INFO Channel 18/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-166-36:104429:104973 [0] NCCL INFO Channel 18/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-166-36:104435:104975 [6] NCCL INFO Channel 19/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-166-36:104433:104976 [4] NCCL INFO Channel 20/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-166-36:104429:104973 [0] NCCL INFO Channel 19/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-167-9:1039544:1040101 [7] NCCL INFO Channel 05/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-166-36:104435:104975 [6] NCCL INFO Channel 20/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-166-36:104429:104973 [0] NCCL INFO Channel 20/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-166-36:104429:104973 [0] NCCL INFO Channel 21/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-166-36:104429:104973 [0] NCCL INFO Channel 22/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-166-36:104429:104973 [0] NCCL INFO Channel 23/0 : 0[0] -> 1[2] via P2P/IPC
+ip-26-0-167-9:1039544:1040101 [7] NCCL INFO Channel 06/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-166-36:104433:104976 [4] NCCL INFO Channel 21/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-166-36:104433:104976 [4] NCCL INFO Channel 22/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-166-244:897242:897792 [2] NCCL INFO Connected all rings
+ip-26-0-167-9:1039544:1040101 [7] NCCL INFO Channel 07/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-166-36:104433:104976 [4] NCCL INFO Channel 23/0 : 2[4] -> 3[6] via P2P/IPC
+ip-26-0-166-36:104435:104975 [6] NCCL INFO Channel 21/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-165-213:9630:10175 [5] NCCL INFO Connected all rings
+ip-26-0-166-36:104435:104975 [6] NCCL INFO Channel 22/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127295:127839 [5] NCCL INFO Connected all rings
+ip-26-0-167-9:1039544:1040101 [7] NCCL INFO Channel 08/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-166-36:104435:104975 [6] NCCL INFO Channel 23/0 : 3[6] -> 0[0] via P2P/IPC
+ip-26-0-167-9:1039544:1040101 [7] NCCL INFO Channel 09/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-167-9:1039544:1040101 [7] NCCL INFO Channel 10/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-167-9:1039544:1040101 [7] NCCL INFO Channel 11/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-167-9:1039544:1040101 [7] NCCL INFO Channel 12/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-167-9:1039544:1040101 [7] NCCL INFO Channel 13/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-167-9:1039544:1040101 [7] NCCL INFO Channel 14/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-166-15:127297:127840 [7] NCCL INFO Connected all rings
+ip-26-0-166-15:127297:127840 [7] NCCL INFO Channel 00/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-166-214:624949:625496 [6] NCCL INFO Connected all rings
+ip-26-0-166-214:624949:625496 [6] NCCL INFO Channel 00/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-167-9:1039540:1040103 [3] NCCL INFO Channel 00/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-167-9:1039540:1040103 [3] NCCL INFO Channel 01/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-167-9:1039544:1040101 [7] NCCL INFO Channel 15/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-166-15:127291:127838 [1] NCCL INFO Connected all rings
+ip-26-0-165-213:9632:10176 [7] NCCL INFO Connected all rings
+ip-26-0-165-213:9632:10176 [7] NCCL INFO Channel 00/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-166-15:127297:127840 [7] NCCL INFO Channel 01/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-167-9:1039540:1040103 [3] NCCL INFO Channel 02/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-167-9:1039544:1040101 [7] NCCL INFO Channel 16/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-166-15:127293:127841 [3] NCCL INFO Connected all rings
+ip-26-0-165-213:9632:10176 [7] NCCL INFO Channel 01/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-166-15:127297:127840 [7] NCCL INFO Channel 02/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-167-9:1039544:1040101 [7] NCCL INFO Channel 17/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-167-9:1039540:1040103 [3] NCCL INFO Channel 03/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-165-213:9626:10173 [1] NCCL INFO Connected all rings
+ip-26-0-166-15:127297:127840 [7] NCCL INFO Channel 03/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-167-9:1039544:1040101 [7] NCCL INFO Channel 18/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-166-214:624949:625496 [6] NCCL INFO Channel 01/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-165-213:9632:10176 [7] NCCL INFO Channel 02/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-167-9:1039542:1040102 [5] NCCL INFO Channel 00/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-166-15:127297:127840 [7] NCCL INFO Channel 04/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-167-9:1039543:1040098 [6] NCCL INFO Connected all rings
+ip-26-0-167-9:1039543:1040098 [6] NCCL INFO Channel 00/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-167-9:1039540:1040103 [3] NCCL INFO Channel 04/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-167-9:1039542:1040102 [5] NCCL INFO Channel 01/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-167-9:1039544:1040101 [7] NCCL INFO Channel 19/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-165-213:9632:10176 [7] NCCL INFO Channel 03/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-167-9:1039542:1040102 [5] NCCL INFO Channel 02/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-167-9:1039543:1040098 [6] NCCL INFO Channel 01/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-166-214:624943:625494 [0] NCCL INFO Connected all rings
+ip-26-0-166-214:624949:625496 [6] NCCL INFO Channel 02/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-166-15:127297:127840 [7] NCCL INFO Channel 05/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-166-214:624947:625497 [4] NCCL INFO Connected all rings
+ip-26-0-165-213:9632:10176 [7] NCCL INFO Channel 04/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-167-9:1039542:1040102 [5] NCCL INFO Channel 03/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-167-9:1039543:1040098 [6] NCCL INFO Channel 02/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-167-9:1039540:1040103 [3] NCCL INFO Channel 05/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-167-9:1039544:1040101 [7] NCCL INFO Channel 20/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-166-15:127297:127840 [7] NCCL INFO Channel 06/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-167-9:1039542:1040102 [5] NCCL INFO Channel 04/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-167-9:1039543:1040098 [6] NCCL INFO Channel 03/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-167-9:1039540:1040103 [3] NCCL INFO Channel 06/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-166-244:897241:897784 [1] NCCL INFO Connected all rings
+ip-26-0-167-9:1039544:1040101 [7] NCCL INFO Channel 21/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-167-9:1039542:1040102 [5] NCCL INFO Channel 05/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-166-214:624949:625496 [6] NCCL INFO Channel 03/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-167-9:1039543:1040098 [6] NCCL INFO Channel 04/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-167-9:1039541:1040097 [4] NCCL INFO Connected all rings
+ip-26-0-167-9:1039540:1040103 [3] NCCL INFO Channel 07/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-166-15:127297:127840 [7] NCCL INFO Channel 07/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-166-214:624945:625495 [2] NCCL INFO Connected all rings
+ip-26-0-167-9:1039544:1040101 [7] NCCL INFO Channel 22/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-167-9:1039542:1040102 [5] NCCL INFO Channel 06/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-167-9:1039543:1040098 [6] NCCL INFO Channel 05/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-166-214:624949:625496 [6] NCCL INFO Channel 04/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-167-9:1039540:1040103 [3] NCCL INFO Channel 08/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-167-9:1039539:1040100 [2] NCCL INFO Connected all rings
+ip-26-0-167-9:1039544:1040101 [7] NCCL INFO Channel 23/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-167-9:1039542:1040102 [5] NCCL INFO Channel 07/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-167-9:1039543:1040098 [6] NCCL INFO Channel 06/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-167-9:1039540:1040103 [3] NCCL INFO Channel 09/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-165-213:9632:10176 [7] NCCL INFO Channel 05/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-167-9:1039542:1040102 [5] NCCL INFO Channel 08/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-166-244:897243:897790 [3] NCCL INFO Connected all rings
+ip-26-0-167-9:1039540:1040103 [3] NCCL INFO Channel 10/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-167-9:1039543:1040098 [6] NCCL INFO Channel 07/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-166-15:127297:127840 [7] NCCL INFO Channel 08/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-167-9:1039542:1040102 [5] NCCL INFO Channel 09/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-167-9:1039540:1040103 [3] NCCL INFO Channel 11/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-167-9:1039543:1040098 [6] NCCL INFO Channel 08/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-167-9:1039543:1040098 [6] NCCL INFO Channel 09/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-167-9:1039540:1040103 [3] NCCL INFO Channel 12/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-167-9:1039542:1040102 [5] NCCL INFO Channel 10/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-166-15:127297:127840 [7] NCCL INFO Channel 09/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-165-213:9625:10169 [0] NCCL INFO Connected all rings
+ip-26-0-165-213:9632:10176 [7] NCCL INFO Channel 06/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-166-214:624949:625496 [6] NCCL INFO Channel 05/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-167-9:1039537:1040095 [0] NCCL INFO Connected all rings
+ip-26-0-167-9:1039542:1040102 [5] NCCL INFO Channel 11/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-167-9:1039540:1040103 [3] NCCL INFO Channel 13/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-167-9:1039543:1040098 [6] NCCL INFO Channel 10/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-167-9:1039542:1040102 [5] NCCL INFO Channel 12/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-167-9:1039540:1040103 [3] NCCL INFO Channel 14/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-167-9:1039543:1040098 [6] NCCL INFO Channel 11/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-167-9:1039542:1040102 [5] NCCL INFO Channel 13/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-167-9:1039540:1040103 [3] NCCL INFO Channel 15/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-167-9:1039543:1040098 [6] NCCL INFO Channel 12/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-167-9:1039542:1040102 [5] NCCL INFO Channel 14/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-167-9:1039540:1040103 [3] NCCL INFO Channel 16/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-167-9:1039543:1040098 [6] NCCL INFO Channel 13/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-166-244:897245:897787 [5] NCCL INFO Connected all rings
+ip-26-0-167-9:1039542:1040102 [5] NCCL INFO Channel 15/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-167-9:1039540:1040103 [3] NCCL INFO Channel 17/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-166-15:127290:127833 [0] NCCL INFO Connected all rings
+ip-26-0-167-9:1039543:1040098 [6] NCCL INFO Channel 14/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-165-213:9632:10176 [7] NCCL INFO Channel 07/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-166-15:127297:127840 [7] NCCL INFO Channel 10/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-167-9:1039542:1040102 [5] NCCL INFO Channel 16/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-166-15:127297:127840 [7] NCCL INFO Channel 11/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-167-9:1039540:1040103 [3] NCCL INFO Channel 18/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-167-9:1039543:1040098 [6] NCCL INFO Channel 15/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-166-214:624949:625496 [6] NCCL INFO Channel 06/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-165-213:9631:10172 [6] NCCL INFO Connected all rings
+ip-26-0-165-213:9631:10172 [6] NCCL INFO Channel 00/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-166-15:127297:127840 [7] NCCL INFO Channel 12/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-167-9:1039542:1040102 [5] NCCL INFO Channel 17/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-165-213:9632:10176 [7] NCCL INFO Channel 08/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-167-9:1039540:1040103 [3] NCCL INFO Channel 19/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-167-9:1039543:1040098 [6] NCCL INFO Channel 16/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-166-15:127297:127840 [7] NCCL INFO Channel 13/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-165-213:9631:10172 [6] NCCL INFO Channel 01/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-165-213:9632:10176 [7] NCCL INFO Channel 09/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-166-15:127297:127840 [7] NCCL INFO Channel 14/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-165-213:9631:10172 [6] NCCL INFO Channel 02/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-167-9:1039542:1040102 [5] NCCL INFO Channel 18/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-167-9:1039540:1040103 [3] NCCL INFO Channel 20/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-167-9:1039543:1040098 [6] NCCL INFO Channel 17/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-165-213:9632:10176 [7] NCCL INFO Channel 10/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-165-213:9631:10172 [6] NCCL INFO Channel 03/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-165-213:9632:10176 [7] NCCL INFO Channel 11/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-167-9:1039542:1040102 [5] NCCL INFO Channel 19/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-167-9:1039540:1040103 [3] NCCL INFO Channel 21/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-167-9:1039543:1040098 [6] NCCL INFO Channel 18/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-165-213:9631:10172 [6] NCCL INFO Channel 04/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-167-9:1039540:1040103 [3] NCCL INFO Channel 22/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-167-9:1039542:1040102 [5] NCCL INFO Channel 20/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-167-9:1039543:1040098 [6] NCCL INFO Channel 19/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-167-9:1039540:1040103 [3] NCCL INFO Channel 23/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-167-9:1039542:1040102 [5] NCCL INFO Channel 21/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-167-9:1039543:1040098 [6] NCCL INFO Channel 20/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-166-244:897246:897791 [6] NCCL INFO Connected all rings
+ip-26-0-166-244:897246:897791 [6] NCCL INFO Channel 00/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-165-213:9632:10176 [7] NCCL INFO Channel 12/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-166-15:127295:127839 [5] NCCL INFO Channel 00/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-166-15:127295:127839 [5] NCCL INFO Channel 01/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-167-9:1039542:1040102 [5] NCCL INFO Channel 22/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-167-9:1039543:1040098 [6] NCCL INFO Channel 21/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-165-213:9628:10177 [3] NCCL INFO Channel 00/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-165-213:9627:10174 [2] NCCL INFO Connected all rings
+ip-26-0-166-15:127295:127839 [5] NCCL INFO Channel 02/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-167-9:1039542:1040102 [5] NCCL INFO Channel 23/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-167-9:1039543:1040098 [6] NCCL INFO Channel 22/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-165-213:9629:10171 [4] NCCL INFO Connected all rings
+ip-26-0-167-9:1039543:1040098 [6] NCCL INFO Channel 23/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-165-213:9631:10172 [6] NCCL INFO Channel 05/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-166-214:624949:625496 [6] NCCL INFO Channel 07/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-165-213:9632:10176 [7] NCCL INFO Channel 13/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-165-213:9628:10177 [3] NCCL INFO Channel 01/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-165-213:9631:10172 [6] NCCL INFO Channel 06/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-165-213:9632:10176 [7] NCCL INFO Channel 14/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-165-213:9628:10177 [3] NCCL INFO Channel 02/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-165-213:9631:10172 [6] NCCL INFO Channel 07/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-165-213:9632:10176 [7] NCCL INFO Channel 15/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-165-213:9628:10177 [3] NCCL INFO Channel 03/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-165-213:9631:10172 [6] NCCL INFO Channel 08/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-165-213:9632:10176 [7] NCCL INFO Channel 16/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-165-213:9628:10177 [3] NCCL INFO Channel 04/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-166-244:897247:897788 [7] NCCL INFO Connected all rings
+ip-26-0-166-244:897247:897788 [7] NCCL INFO Channel 00/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-165-213:9631:10172 [6] NCCL INFO Channel 09/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-166-15:127295:127839 [5] NCCL INFO Channel 03/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-166-15:127297:127840 [7] NCCL INFO Channel 15/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-165-213:9628:10177 [3] NCCL INFO Channel 05/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-165-213:9632:10176 [7] NCCL INFO Channel 17/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-166-244:897244:897789 [4] NCCL INFO Connected all rings
+ip-26-0-165-213:9631:10172 [6] NCCL INFO Channel 10/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-166-244:897246:897791 [6] NCCL INFO Channel 01/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-165-213:9628:10177 [3] NCCL INFO Channel 06/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-165-213:9632:10176 [7] NCCL INFO Channel 18/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-165-213:9631:10172 [6] NCCL INFO Channel 11/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-165-213:9630:10175 [5] NCCL INFO Channel 00/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-166-15:127296:127834 [6] NCCL INFO Connected all rings
+ip-26-0-166-15:127296:127834 [6] NCCL INFO Channel 00/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-166-244:897247:897788 [7] NCCL INFO Channel 01/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-166-214:624949:625496 [6] NCCL INFO Channel 08/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-165-213:9628:10177 [3] NCCL INFO Channel 07/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-165-213:9632:10176 [7] NCCL INFO Channel 19/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-165-213:9630:10175 [5] NCCL INFO Channel 01/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-165-213:9631:10172 [6] NCCL INFO Channel 12/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-167-9:1039541:1040097 [4] NCCL INFO Channel 00/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-166-244:897246:897791 [6] NCCL INFO Channel 02/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-165-213:9628:10177 [3] NCCL INFO Channel 08/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-165-213:9632:10176 [7] NCCL INFO Channel 20/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-165-213:9630:10175 [5] NCCL INFO Channel 02/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-166-15:127295:127839 [5] NCCL INFO Channel 04/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-165-213:9631:10172 [6] NCCL INFO Channel 13/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-166-244:897240:897786 [0] NCCL INFO Connected all rings
+ip-26-0-166-244:897247:897788 [7] NCCL INFO Channel 02/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-165-213:9628:10177 [3] NCCL INFO Channel 09/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-165-213:9632:10176 [7] NCCL INFO Channel 21/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-165-213:9630:10175 [5] NCCL INFO Channel 03/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-166-244:897246:897791 [6] NCCL INFO Channel 03/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-165-213:9631:10172 [6] NCCL INFO Channel 14/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-166-244:897247:897788 [7] NCCL INFO Channel 03/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-166-244:897246:897791 [6] NCCL INFO Channel 04/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-165-213:9628:10177 [3] NCCL INFO Channel 10/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-165-213:9632:10176 [7] NCCL INFO Channel 22/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-165-213:9630:10175 [5] NCCL INFO Channel 04/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-162-46:702881:703465 [1] NCCL INFO Connected all rings
+ip-26-0-165-213:9631:10172 [6] NCCL INFO Channel 15/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-166-244:897247:897788 [7] NCCL INFO Channel 04/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-166-244:897246:897791 [6] NCCL INFO Channel 05/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-165-213:9628:10177 [3] NCCL INFO Channel 11/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-165-213:9632:10176 [7] NCCL INFO Channel 23/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-165-213:9630:10175 [5] NCCL INFO Channel 05/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-166-244:897246:897791 [6] NCCL INFO Channel 06/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-166-244:897247:897788 [7] NCCL INFO Channel 05/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-165-213:9631:10172 [6] NCCL INFO Channel 16/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-166-15:127296:127834 [6] NCCL INFO Channel 01/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-166-244:897246:897791 [6] NCCL INFO Channel 07/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-166-244:897247:897788 [7] NCCL INFO Channel 06/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-166-214:624949:625496 [6] NCCL INFO Channel 09/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-165-213:9630:10175 [5] NCCL INFO Channel 06/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-165-213:9628:10177 [3] NCCL INFO Channel 12/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-165-213:9631:10172 [6] NCCL INFO Channel 17/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-166-244:897246:897791 [6] NCCL INFO Channel 08/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-166-244:897247:897788 [7] NCCL INFO Channel 07/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-166-15:127297:127840 [7] NCCL INFO Channel 16/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-165-213:9630:10175 [5] NCCL INFO Channel 07/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-165-213:9631:10172 [6] NCCL INFO Channel 18/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-166-244:897246:897791 [6] NCCL INFO Channel 09/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-166-244:897247:897788 [7] NCCL INFO Channel 08/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-165-213:9628:10177 [3] NCCL INFO Channel 13/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-166-15:127292:127836 [2] NCCL INFO Connected all rings
+ip-26-0-165-213:9630:10175 [5] NCCL INFO Channel 08/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-166-244:897246:897791 [6] NCCL INFO Channel 10/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-166-244:897247:897788 [7] NCCL INFO Channel 09/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-165-213:9631:10172 [6] NCCL INFO Channel 19/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-165-213:9628:10177 [3] NCCL INFO Channel 14/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-166-15:127296:127834 [6] NCCL INFO Channel 02/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-165-213:9630:10175 [5] NCCL INFO Channel 09/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-165-213:9631:10172 [6] NCCL INFO Channel 20/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-166-244:897246:897791 [6] NCCL INFO Channel 11/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-166-244:897247:897788 [7] NCCL INFO Channel 10/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-165-213:9628:10177 [3] NCCL INFO Channel 15/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-165-213:9630:10175 [5] NCCL INFO Channel 10/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-165-213:9631:10172 [6] NCCL INFO Channel 21/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-166-244:897247:897788 [7] NCCL INFO Channel 11/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-166-244:897246:897791 [6] NCCL INFO Channel 12/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-165-213:9628:10177 [3] NCCL INFO Channel 16/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-166-15:127296:127834 [6] NCCL INFO Channel 03/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-165-213:9630:10175 [5] NCCL INFO Channel 11/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-165-213:9631:10172 [6] NCCL INFO Channel 22/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-165-213:9628:10177 [3] NCCL INFO Channel 17/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-166-244:897247:897788 [7] NCCL INFO Channel 12/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-167-9:1039541:1040097 [4] NCCL INFO Channel 01/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-166-244:897246:897791 [6] NCCL INFO Channel 13/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-165-213:9630:10175 [5] NCCL INFO Channel 12/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-165-213:9631:10172 [6] NCCL INFO Channel 23/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-165-213:9628:10177 [3] NCCL INFO Channel 18/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-166-244:897246:897791 [6] NCCL INFO Channel 14/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-166-244:897247:897788 [7] NCCL INFO Channel 13/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-166-214:624949:625496 [6] NCCL INFO Channel 10/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-165-213:9630:10175 [5] NCCL INFO Channel 13/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-166-15:127295:127839 [5] NCCL INFO Channel 05/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-165-213:9628:10177 [3] NCCL INFO Channel 19/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-166-15:127296:127834 [6] NCCL INFO Channel 04/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-166-15:127297:127840 [7] NCCL INFO Channel 17/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-166-244:897246:897791 [6] NCCL INFO Channel 15/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-166-244:897247:897788 [7] NCCL INFO Channel 14/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-165-213:9630:10175 [5] NCCL INFO Channel 14/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-165-213:9628:10177 [3] NCCL INFO Channel 20/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-166-15:127294:127835 [4] NCCL INFO Connected all rings
+ip-26-0-165-213:9630:10175 [5] NCCL INFO Channel 15/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-165-213:9628:10177 [3] NCCL INFO Channel 21/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-166-244:897246:897791 [6] NCCL INFO Channel 16/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-166-244:897247:897788 [7] NCCL INFO Channel 15/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-165-213:9630:10175 [5] NCCL INFO Channel 16/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-165-213:9628:10177 [3] NCCL INFO Channel 22/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-166-15:127295:127839 [5] NCCL INFO Channel 06/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-166-15:127293:127841 [3] NCCL INFO Channel 00/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-166-15:127296:127834 [6] NCCL INFO Channel 05/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-166-15:127297:127840 [7] NCCL INFO Channel 18/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-166-244:897246:897791 [6] NCCL INFO Channel 17/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-165-213:9630:10175 [5] NCCL INFO Channel 17/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-165-213:9628:10177 [3] NCCL INFO Channel 23/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-166-244:897247:897788 [7] NCCL INFO Channel 16/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-166-15:127295:127839 [5] NCCL INFO Channel 07/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-165-213:9630:10175 [5] NCCL INFO Channel 18/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-165-213:9629:10171 [4] NCCL INFO Channel 00/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-166-15:127293:127841 [3] NCCL INFO Channel 01/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-166-15:127296:127834 [6] NCCL INFO Channel 06/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-166-15:127297:127840 [7] NCCL INFO Channel 19/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-165-213:9627:10174 [2] NCCL INFO Channel 00/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-166-244:897246:897791 [6] NCCL INFO Channel 18/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-166-244:897247:897788 [7] NCCL INFO Channel 17/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-166-244:897242:897792 [2] NCCL INFO Channel 00/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-162-46:702883:703471 [3] NCCL INFO Connected all rings
+ip-26-0-167-9:1039541:1040097 [4] NCCL INFO Channel 02/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-166-15:127295:127839 [5] NCCL INFO Channel 08/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-166-15:127293:127841 [3] NCCL INFO Channel 02/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-166-15:127296:127834 [6] NCCL INFO Channel 07/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-166-15:127297:127840 [7] NCCL INFO Channel 20/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-166-244:897246:897791 [6] NCCL INFO Channel 19/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-165-213:9630:10175 [5] NCCL INFO Channel 19/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-165-213:9629:10171 [4] NCCL INFO Channel 01/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-165-213:9627:10174 [2] NCCL INFO Channel 01/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-166-244:897247:897788 [7] NCCL INFO Channel 18/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-166-244:897242:897792 [2] NCCL INFO Channel 01/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127295:127839 [5] NCCL INFO Channel 09/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-166-15:127293:127841 [3] NCCL INFO Channel 03/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-166-15:127296:127834 [6] NCCL INFO Channel 08/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-166-15:127297:127840 [7] NCCL INFO Channel 21/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-166-244:897246:897791 [6] NCCL INFO Channel 20/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-166-244:897247:897788 [7] NCCL INFO Channel 19/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-166-244:897242:897792 [2] NCCL INFO Channel 02/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-165-213:9630:10175 [5] NCCL INFO Channel 20/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-166-214:624949:625496 [6] NCCL INFO Channel 11/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-166-15:127295:127839 [5] NCCL INFO Channel 10/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-166-15:127293:127841 [3] NCCL INFO Channel 04/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-166-15:127296:127834 [6] NCCL INFO Channel 09/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-166-15:127297:127840 [7] NCCL INFO Channel 22/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-166-244:897246:897791 [6] NCCL INFO Channel 21/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-166-244:897243:897790 [3] NCCL INFO Channel 00/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-166-244:897247:897788 [7] NCCL INFO Channel 20/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-165-213:9629:10171 [4] NCCL INFO Channel 02/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-165-213:9627:10174 [2] NCCL INFO Channel 02/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-166-244:897242:897792 [2] NCCL INFO Channel 03/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127295:127839 [5] NCCL INFO Channel 11/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-166-15:127293:127841 [3] NCCL INFO Channel 05/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-166-15:127296:127834 [6] NCCL INFO Channel 10/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-166-15:127297:127840 [7] NCCL INFO Channel 23/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-166-244:897246:897791 [6] NCCL INFO Channel 22/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-166-244:897243:897790 [3] NCCL INFO Channel 01/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-166-244:897247:897788 [7] NCCL INFO Channel 21/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-166-244:897242:897792 [2] NCCL INFO Channel 04/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127295:127839 [5] NCCL INFO Channel 12/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-166-15:127293:127841 [3] NCCL INFO Channel 06/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-166-15:127296:127834 [6] NCCL INFO Channel 11/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-166-244:897246:897791 [6] NCCL INFO Channel 23/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-166-244:897243:897790 [3] NCCL INFO Channel 02/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-166-244:897245:897787 [5] NCCL INFO Channel 00/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-166-244:897247:897788 [7] NCCL INFO Channel 22/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-166-244:897242:897792 [2] NCCL INFO Channel 05/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127295:127839 [5] NCCL INFO Channel 13/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-165-213:9627:10174 [2] NCCL INFO Channel 03/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127293:127841 [3] NCCL INFO Channel 07/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-166-15:127296:127834 [6] NCCL INFO Channel 12/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-165-213:9629:10171 [4] NCCL INFO Channel 03/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-166-15:127295:127839 [5] NCCL INFO Channel 14/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-166-244:897245:897787 [5] NCCL INFO Channel 01/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-166-244:897243:897790 [3] NCCL INFO Channel 03/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-166-15:127293:127841 [3] NCCL INFO Channel 08/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-166-244:897247:897788 [7] NCCL INFO Channel 23/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-166-15:127296:127834 [6] NCCL INFO Channel 13/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-166-244:897242:897792 [2] NCCL INFO Channel 06/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127295:127839 [5] NCCL INFO Channel 15/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-166-15:127293:127841 [3] NCCL INFO Channel 09/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-166-15:127296:127834 [6] NCCL INFO Channel 14/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-166-244:897245:897787 [5] NCCL INFO Channel 02/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-166-244:897242:897792 [2] NCCL INFO Channel 07/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-165-213:9627:10174 [2] NCCL INFO Channel 04/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-165-213:9630:10175 [5] NCCL INFO Channel 21/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-166-15:127295:127839 [5] NCCL INFO Channel 16/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-166-15:127293:127841 [3] NCCL INFO Channel 10/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-167-9:1039541:1040097 [4] NCCL INFO Channel 03/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-166-15:127296:127834 [6] NCCL INFO Channel 15/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-166-244:897243:897790 [3] NCCL INFO Channel 04/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-166-15:127295:127839 [5] NCCL INFO Channel 17/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-166-15:127293:127841 [3] NCCL INFO Channel 11/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-166-15:127296:127834 [6] NCCL INFO Channel 16/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-166-244:897245:897787 [5] NCCL INFO Channel 03/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-166-244:897242:897792 [2] NCCL INFO Channel 08/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-166-244:897244:897789 [4] NCCL INFO Channel 00/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-165-213:9629:10171 [4] NCCL INFO Channel 04/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-166-15:127295:127839 [5] NCCL INFO Channel 18/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-166-15:127293:127841 [3] NCCL INFO Channel 12/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-166-15:127296:127834 [6] NCCL INFO Channel 17/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-166-244:897242:897792 [2] NCCL INFO Channel 09/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-166-244:897245:897787 [5] NCCL INFO Channel 04/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-166-244:897243:897790 [3] NCCL INFO Channel 05/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-166-244:897244:897789 [4] NCCL INFO Channel 01/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-166-15:127295:127839 [5] NCCL INFO Channel 19/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-166-15:127293:127841 [3] NCCL INFO Channel 13/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-166-15:127296:127834 [6] NCCL INFO Channel 18/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-166-244:897242:897792 [2] NCCL INFO Channel 10/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-166-244:897245:897787 [5] NCCL INFO Channel 05/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-166-244:897243:897790 [3] NCCL INFO Channel 06/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-166-244:897244:897789 [4] NCCL INFO Channel 02/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-166-15:127295:127839 [5] NCCL INFO Channel 20/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-166-15:127293:127841 [3] NCCL INFO Channel 14/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-166-15:127296:127834 [6] NCCL INFO Channel 19/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-166-244:897242:897792 [2] NCCL INFO Channel 11/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-165-213:9629:10171 [4] NCCL INFO Channel 05/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-166-244:897245:897787 [5] NCCL INFO Channel 06/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-166-244:897243:897790 [3] NCCL INFO Channel 07/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-166-244:897244:897789 [4] NCCL INFO Channel 03/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-166-15:127295:127839 [5] NCCL INFO Channel 21/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-166-15:127293:127841 [3] NCCL INFO Channel 15/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-166-15:127296:127834 [6] NCCL INFO Channel 20/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-166-244:897242:897792 [2] NCCL INFO Channel 12/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-166-244:897245:897787 [5] NCCL INFO Channel 07/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-166-244:897243:897790 [3] NCCL INFO Channel 08/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-166-244:897244:897789 [4] NCCL INFO Channel 04/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-166-15:127295:127839 [5] NCCL INFO Channel 22/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-166-15:127293:127841 [3] NCCL INFO Channel 16/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-166-15:127296:127834 [6] NCCL INFO Channel 21/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-166-244:897242:897792 [2] NCCL INFO Channel 13/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-166-244:897243:897790 [3] NCCL INFO Channel 09/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-166-244:897245:897787 [5] NCCL INFO Channel 08/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-166-244:897244:897789 [4] NCCL INFO Channel 05/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-166-15:127295:127839 [5] NCCL INFO Channel 23/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-166-15:127293:127841 [3] NCCL INFO Channel 17/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-166-15:127296:127834 [6] NCCL INFO Channel 22/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-166-244:897242:897792 [2] NCCL INFO Channel 14/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-166-244:897245:897787 [5] NCCL INFO Channel 09/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-166-244:897243:897790 [3] NCCL INFO Channel 10/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-166-244:897244:897789 [4] NCCL INFO Channel 06/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-165-213:9630:10175 [5] NCCL INFO Channel 22/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-167-9:1039541:1040097 [4] NCCL INFO Channel 04/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-166-15:127293:127841 [3] NCCL INFO Channel 18/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-166-15:127296:127834 [6] NCCL INFO Channel 23/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-166-244:897242:897792 [2] NCCL INFO Channel 15/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-166-244:897245:897787 [5] NCCL INFO Channel 10/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-166-244:897243:897790 [3] NCCL INFO Channel 11/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-166-214:624949:625496 [6] NCCL INFO Channel 12/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-166-244:897244:897789 [4] NCCL INFO Channel 07/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-165-213:9629:10171 [4] NCCL INFO Channel 06/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-165-213:9627:10174 [2] NCCL INFO Channel 05/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-166-244:897245:897787 [5] NCCL INFO Channel 11/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-166-244:897242:897792 [2] NCCL INFO Channel 16/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-166-244:897243:897790 [3] NCCL INFO Channel 12/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-166-244:897244:897789 [4] NCCL INFO Channel 08/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-166-244:897245:897787 [5] NCCL INFO Channel 12/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-166-244:897242:897792 [2] NCCL INFO Channel 17/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-166-244:897243:897790 [3] NCCL INFO Channel 13/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-166-244:897244:897789 [4] NCCL INFO Channel 09/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-166-244:897245:897787 [5] NCCL INFO Channel 13/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-166-244:897242:897792 [2] NCCL INFO Channel 18/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-166-244:897243:897790 [3] NCCL INFO Channel 14/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-166-244:897244:897789 [4] NCCL INFO Channel 10/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-166-15:127293:127841 [3] NCCL INFO Channel 19/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-165-213:9629:10171 [4] NCCL INFO Channel 07/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-165-213:9627:10174 [2] NCCL INFO Channel 06/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-165-213:9630:10175 [5] NCCL INFO Channel 23/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-166-244:897245:897787 [5] NCCL INFO Channel 14/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-166-244:897242:897792 [2] NCCL INFO Channel 19/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-166-244:897243:897790 [3] NCCL INFO Channel 15/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-166-244:897244:897789 [4] NCCL INFO Channel 11/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-166-15:127293:127841 [3] NCCL INFO Channel 20/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-166-244:897245:897787 [5] NCCL INFO Channel 15/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-166-244:897242:897792 [2] NCCL INFO Channel 20/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-166-244:897243:897790 [3] NCCL INFO Channel 16/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-166-244:897244:897789 [4] NCCL INFO Channel 12/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-166-244:897245:897787 [5] NCCL INFO Channel 16/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-166-244:897242:897792 [2] NCCL INFO Channel 21/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-166-244:897244:897789 [4] NCCL INFO Channel 13/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-166-244:897243:897790 [3] NCCL INFO Channel 17/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-165-213:9629:10171 [4] NCCL INFO Channel 08/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-165-213:9627:10174 [2] NCCL INFO Channel 07/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-166-244:897245:897787 [5] NCCL INFO Channel 17/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-166-244:897242:897792 [2] NCCL INFO Channel 22/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-166-244:897244:897789 [4] NCCL INFO Channel 14/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-166-244:897243:897790 [3] NCCL INFO Channel 18/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-162-46:702885:703468 [5] NCCL INFO Connected all rings
+ip-26-0-166-244:897245:897787 [5] NCCL INFO Channel 18/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-166-244:897242:897792 [2] NCCL INFO Channel 23/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-166-244:897244:897789 [4] NCCL INFO Channel 15/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-166-244:897243:897790 [3] NCCL INFO Channel 19/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-166-244:897245:897787 [5] NCCL INFO Channel 19/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-166-244:897244:897789 [4] NCCL INFO Channel 16/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-166-244:897243:897790 [3] NCCL INFO Channel 20/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-167-9:1039541:1040097 [4] NCCL INFO Channel 05/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-166-244:897245:897787 [5] NCCL INFO Channel 20/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-166-244:897244:897789 [4] NCCL INFO Channel 17/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-166-244:897243:897790 [3] NCCL INFO Channel 21/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-166-15:127293:127841 [3] NCCL INFO Channel 21/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-166-15:127293:127841 [3] NCCL INFO Channel 22/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-166-244:897245:897787 [5] NCCL INFO Channel 21/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-166-244:897244:897789 [4] NCCL INFO Channel 18/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-166-244:897243:897790 [3] NCCL INFO Channel 22/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-166-15:127294:127835 [4] NCCL INFO Channel 00/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-166-214:624949:625496 [6] NCCL INFO Channel 13/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-166-15:127293:127841 [3] NCCL INFO Channel 23/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-165-213:9627:10174 [2] NCCL INFO Channel 08/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624948:625500 [5] NCCL INFO Connected all rings
+ip-26-0-166-244:897245:897787 [5] NCCL INFO Channel 22/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-162-46:702884:703470 [4] NCCL INFO Connected all rings
+ip-26-0-165-213:9629:10171 [4] NCCL INFO Channel 09/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-167-9:1039541:1040097 [4] NCCL INFO Channel 06/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-166-244:897245:897787 [5] NCCL INFO Channel 23/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-166-244:897244:897789 [4] NCCL INFO Channel 19/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-162-46:702887:703467 [7] NCCL INFO Connected all rings
+ip-26-0-162-46:702887:703467 [7] NCCL INFO Channel 00/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-165-213:9627:10174 [2] NCCL INFO Channel 09/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127294:127835 [4] NCCL INFO Channel 01/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-166-244:897244:897789 [4] NCCL INFO Channel 20/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-166-15:127292:127836 [2] NCCL INFO Channel 00/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127292:127836 [2] NCCL INFO Channel 01/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127292:127836 [2] NCCL INFO Channel 02/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127292:127836 [2] NCCL INFO Channel 03/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-166-244:897243:897790 [3] NCCL INFO Channel 23/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-166-15:127292:127836 [2] NCCL INFO Channel 04/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-165-213:9629:10171 [4] NCCL INFO Channel 10/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-166-15:127292:127836 [2] NCCL INFO Channel 05/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127292:127836 [2] NCCL INFO Channel 06/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127292:127836 [2] NCCL INFO Channel 07/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-162-46:702887:703467 [7] NCCL INFO Channel 01/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-166-15:127292:127836 [2] NCCL INFO Channel 08/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624949:625496 [6] NCCL INFO Channel 14/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-162-46:702886:703469 [6] NCCL INFO Connected all rings
+ip-26-0-162-46:702886:703469 [6] NCCL INFO Channel 00/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-162-46:702887:703467 [7] NCCL INFO Channel 02/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-166-15:127292:127836 [2] NCCL INFO Channel 09/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127294:127835 [4] NCCL INFO Channel 02/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-166-214:624946:625498 [3] NCCL INFO Connected all rings
+ip-26-0-166-15:127294:127835 [4] NCCL INFO Channel 03/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-166-15:127294:127835 [4] NCCL INFO Channel 04/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-167-9:1039541:1040097 [4] NCCL INFO Channel 07/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-162-46:702887:703467 [7] NCCL INFO Channel 03/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-162-46:702886:703469 [6] NCCL INFO Channel 01/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-166-214:624949:625496 [6] NCCL INFO Channel 15/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-166-15:127294:127835 [4] NCCL INFO Channel 05/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-162-46:702880:703466 [0] NCCL INFO Connected all rings
+ip-26-0-165-213:9627:10174 [2] NCCL INFO Channel 10/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-162-46:702887:703467 [7] NCCL INFO Channel 04/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-166-214:624949:625496 [6] NCCL INFO Channel 16/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-162-46:702882:703472 [2] NCCL INFO Connected all rings
+ip-26-0-166-244:897244:897789 [4] NCCL INFO Channel 21/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-162-46:702886:703469 [6] NCCL INFO Channel 02/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-162-46:702887:703467 [7] NCCL INFO Channel 05/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-162-46:702886:703469 [6] NCCL INFO Channel 03/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-165-213:9629:10171 [4] NCCL INFO Channel 11/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-166-214:624949:625496 [6] NCCL INFO Channel 17/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-162-46:702887:703467 [7] NCCL INFO Channel 06/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-162-46:702886:703469 [6] NCCL INFO Channel 04/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-167-9:1039541:1040097 [4] NCCL INFO Channel 08/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-162-46:702887:703467 [7] NCCL INFO Channel 07/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-162-46:702886:703469 [6] NCCL INFO Channel 05/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-162-46:702887:703467 [7] NCCL INFO Channel 08/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-162-46:702886:703469 [6] NCCL INFO Channel 06/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-165-213:9626:10173 [1] NCCL INFO Connected all trees
+ip-26-0-166-15:127292:127836 [2] NCCL INFO Channel 10/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-165-213:9626:10173 [1] NCCL INFO NVLS comm 0xb4e0900 headRank 0 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-162-46:702887:703467 [7] NCCL INFO Channel 09/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-165-213:9627:10174 [2] NCCL INFO Channel 11/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-162-46:702886:703469 [6] NCCL INFO Channel 07/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-162-46:702887:703467 [7] NCCL INFO Channel 10/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-166-214:624949:625496 [6] NCCL INFO Channel 18/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-162-46:702886:703469 [6] NCCL INFO Channel 08/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-162-46:702887:703467 [7] NCCL INFO Channel 11/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-162-46:702886:703469 [6] NCCL INFO Channel 09/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-162-46:702887:703467 [7] NCCL INFO Channel 12/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-166-15:127294:127835 [4] NCCL INFO Channel 06/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-167-9:1039541:1040097 [4] NCCL INFO Channel 09/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-165-213:9629:10171 [4] NCCL INFO Channel 12/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-162-46:702886:703469 [6] NCCL INFO Channel 10/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-162-46:702887:703467 [7] NCCL INFO Channel 13/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-162-46:702887:703467 [7] NCCL INFO Channel 14/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-162-46:702886:703469 [6] NCCL INFO Channel 11/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-166-214:624944:625493 [1] NCCL INFO Connected all rings
+ip-26-0-166-244:897244:897789 [4] NCCL INFO Channel 22/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-162-46:702887:703467 [7] NCCL INFO Channel 15/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-162-46:702886:703469 [6] NCCL INFO Channel 12/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-166-214:624949:625496 [6] NCCL INFO Channel 19/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-165-213:9627:10174 [2] NCCL INFO Channel 12/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-162-46:702887:703467 [7] NCCL INFO Channel 16/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-162-46:702886:703469 [6] NCCL INFO Channel 13/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-166-15:127292:127836 [2] NCCL INFO Channel 11/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624950:625499 [7] NCCL INFO Connected all rings
+ip-26-0-166-214:624950:625499 [7] NCCL INFO Channel 00/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-162-46:702887:703467 [7] NCCL INFO Channel 17/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-166-214:624949:625496 [6] NCCL INFO Channel 20/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-162-46:702886:703469 [6] NCCL INFO Channel 14/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-165-213:9629:10171 [4] NCCL INFO Channel 13/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-166-214:624950:625499 [7] NCCL INFO Channel 01/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-162-46:702887:703467 [7] NCCL INFO Channel 18/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-162-46:702886:703469 [6] NCCL INFO Channel 15/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-166-214:624949:625496 [6] NCCL INFO Channel 21/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-166-214:624950:625499 [7] NCCL INFO Channel 02/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-166-15:127294:127835 [4] NCCL INFO Channel 07/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-166-214:624949:625496 [6] NCCL INFO Channel 22/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-162-46:702887:703467 [7] NCCL INFO Channel 19/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-166-214:624949:625496 [6] NCCL INFO Channel 23/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-166-214:624950:625499 [7] NCCL INFO Channel 03/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-165-213:9627:10174 [2] NCCL INFO Channel 13/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-162-46:702887:703467 [7] NCCL INFO Channel 20/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-166-214:624950:625499 [7] NCCL INFO Channel 04/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-162-46:702887:703467 [7] NCCL INFO Channel 21/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-166-214:624950:625499 [7] NCCL INFO Channel 05/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-167-9:1039541:1040097 [4] NCCL INFO Channel 10/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-162-46:702887:703467 [7] NCCL INFO Channel 22/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-166-214:624950:625499 [7] NCCL INFO Channel 06/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-162-46:702887:703467 [7] NCCL INFO Channel 23/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-165-213:9629:10171 [4] NCCL INFO Channel 14/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-166-214:624950:625499 [7] NCCL INFO Channel 07/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-162-46:702883:703471 [3] NCCL INFO Channel 00/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-166-214:624947:625497 [4] NCCL INFO Channel 00/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-162-46:702883:703471 [3] NCCL INFO Channel 01/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-162-46:702886:703469 [6] NCCL INFO Channel 16/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-167-9:1039539:1040100 [2] NCCL INFO Channel 00/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624950:625499 [7] NCCL INFO Channel 08/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-166-244:897244:897789 [4] NCCL INFO Channel 23/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-166-15:127292:127836 [2] NCCL INFO Channel 12/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624950:625499 [7] NCCL INFO Channel 09/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-166-214:624947:625497 [4] NCCL INFO Channel 01/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-165-213:9627:10174 [2] NCCL INFO Channel 14/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624950:625499 [7] NCCL INFO Channel 10/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-162-46:702885:703468 [5] NCCL INFO Channel 00/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-166-214:624947:625497 [4] NCCL INFO Channel 02/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-166-214:624950:625499 [7] NCCL INFO Channel 11/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-166-214:624947:625497 [4] NCCL INFO Channel 03/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-166-214:624950:625499 [7] NCCL INFO Channel 12/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-162-46:702883:703471 [3] NCCL INFO Channel 02/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-162-46:702886:703469 [6] NCCL INFO Channel 17/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-166-214:624947:625497 [4] NCCL INFO Channel 04/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-166-214:624950:625499 [7] NCCL INFO Channel 13/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-162-46:702883:703471 [3] NCCL INFO Channel 03/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-162-46:702886:703469 [6] NCCL INFO Channel 18/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-166-15:127294:127835 [4] NCCL INFO Channel 08/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-166-214:624947:625497 [4] NCCL INFO Channel 05/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-162-46:702883:703471 [3] NCCL INFO Channel 04/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-162-46:702886:703469 [6] NCCL INFO Channel 19/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-166-214:624950:625499 [7] NCCL INFO Channel 14/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-167-9:1039541:1040097 [4] NCCL INFO Channel 11/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-165-213:9629:10171 [4] NCCL INFO Channel 15/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-162-46:702883:703471 [3] NCCL INFO Channel 05/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-162-46:702886:703469 [6] NCCL INFO Channel 20/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-166-214:624947:625497 [4] NCCL INFO Channel 06/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-162-46:702884:703470 [4] NCCL INFO Channel 00/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-166-214:624950:625499 [7] NCCL INFO Channel 15/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-162-46:702883:703471 [3] NCCL INFO Channel 06/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-162-46:702886:703469 [6] NCCL INFO Channel 21/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-167-9:1039539:1040100 [2] NCCL INFO Channel 01/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-162-46:702884:703470 [4] NCCL INFO Channel 01/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-166-214:624947:625497 [4] NCCL INFO Channel 07/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-166-214:624950:625499 [7] NCCL INFO Channel 16/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-162-46:702883:703471 [3] NCCL INFO Channel 07/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-162-46:702886:703469 [6] NCCL INFO Channel 22/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-162-46:702884:703470 [4] NCCL INFO Channel 02/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-166-214:624947:625497 [4] NCCL INFO Channel 08/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-166-15:127292:127836 [2] NCCL INFO Channel 13/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-162-46:702883:703471 [3] NCCL INFO Channel 08/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-162-46:702886:703469 [6] NCCL INFO Channel 23/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-166-214:624950:625499 [7] NCCL INFO Channel 17/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-162-46:702884:703470 [4] NCCL INFO Channel 03/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-165-213:9627:10174 [2] NCCL INFO Channel 15/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-162-46:702883:703471 [3] NCCL INFO Channel 09/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-166-214:624947:625497 [4] NCCL INFO Channel 09/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-162-46:702884:703470 [4] NCCL INFO Channel 04/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-166-214:624950:625499 [7] NCCL INFO Channel 18/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-162-46:702883:703471 [3] NCCL INFO Channel 10/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-162-46:702885:703468 [5] NCCL INFO Channel 01/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-162-46:702884:703470 [4] NCCL INFO Channel 05/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-166-214:624947:625497 [4] NCCL INFO Channel 10/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-166-214:624950:625499 [7] NCCL INFO Channel 19/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-162-46:702882:703472 [2] NCCL INFO Channel 00/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-162-46:702883:703471 [3] NCCL INFO Channel 11/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-162-46:702885:703468 [5] NCCL INFO Channel 02/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-162-46:702884:703470 [4] NCCL INFO Channel 06/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-166-214:624947:625497 [4] NCCL INFO Channel 11/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-166-214:624946:625498 [3] NCCL INFO Channel 00/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-166-214:624950:625499 [7] NCCL INFO Channel 20/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-167-9:1039544:1040101 [7] NCCL INFO Connected all trees
+ip-26-0-166-214:624945:625495 [2] NCCL INFO Channel 00/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-162-46:702882:703472 [2] NCCL INFO Channel 01/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-167-9:1039544:1040101 [7] NCCL INFO NVLS comm 0xad8eb40 headRank 3 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-162-46:702883:703471 [3] NCCL INFO Channel 12/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-162-46:702884:703470 [4] NCCL INFO Channel 07/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-162-46:702885:703468 [5] NCCL INFO Channel 03/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-162-46:702882:703472 [2] NCCL INFO Channel 02/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-165-213:9629:10171 [4] NCCL INFO Channel 16/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-162-46:702883:703471 [3] NCCL INFO Channel 13/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-162-46:702884:703470 [4] NCCL INFO Channel 08/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-162-46:702885:703468 [5] NCCL INFO Channel 04/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-166-214:624947:625497 [4] NCCL INFO Channel 12/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-162-46:702882:703472 [2] NCCL INFO Channel 03/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624948:625500 [5] NCCL INFO Channel 00/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-167-9:1039538:1040099 [1] NCCL INFO Connected all trees
+ip-26-0-166-214:624945:625495 [2] NCCL INFO Channel 01/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624947:625497 [4] NCCL INFO Channel 13/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-166-214:624948:625500 [5] NCCL INFO Channel 01/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-167-9:1039538:1040099 [1] NCCL INFO NVLS comm 0xa731d00 headRank 0 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-166-15:127294:127835 [4] NCCL INFO Channel 09/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-166-214:624945:625495 [2] NCCL INFO Channel 02/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624947:625497 [4] NCCL INFO Channel 14/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-166-214:624948:625500 [5] NCCL INFO Channel 02/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-166-214:624945:625495 [2] NCCL INFO Channel 03/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624947:625497 [4] NCCL INFO Channel 15/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-166-214:624948:625500 [5] NCCL INFO Channel 03/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-166-214:624950:625499 [7] NCCL INFO Channel 21/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-167-9:1039541:1040097 [4] NCCL INFO Channel 12/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-162-46:702884:703470 [4] NCCL INFO Channel 09/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-162-46:702882:703472 [2] NCCL INFO Channel 04/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624945:625495 [2] NCCL INFO Channel 04/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624947:625497 [4] NCCL INFO Channel 16/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-162-46:702885:703468 [5] NCCL INFO Channel 05/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-166-15:127292:127836 [2] NCCL INFO Channel 14/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624950:625499 [7] NCCL INFO Channel 22/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-162-46:702883:703471 [3] NCCL INFO Channel 14/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-166-214:624948:625500 [5] NCCL INFO Channel 04/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-166-214:624946:625498 [3] NCCL INFO Channel 01/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-162-46:702884:703470 [4] NCCL INFO Channel 10/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-162-46:702885:703468 [5] NCCL INFO Channel 06/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-162-46:702883:703471 [3] NCCL INFO Channel 15/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-162-46:702884:703470 [4] NCCL INFO Channel 11/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-166-214:624945:625495 [2] NCCL INFO Channel 05/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624948:625500 [5] NCCL INFO Channel 05/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-166-214:624950:625499 [7] NCCL INFO Channel 23/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-166-214:624946:625498 [3] NCCL INFO Channel 02/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-167-9:1039539:1040100 [2] NCCL INFO Channel 02/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-162-46:702882:703472 [2] NCCL INFO Channel 05/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-162-46:702885:703468 [5] NCCL INFO Channel 07/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-162-46:702883:703471 [3] NCCL INFO Channel 16/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-162-46:702884:703470 [4] NCCL INFO Channel 12/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-165-213:9627:10174 [2] NCCL INFO Channel 16/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-162-46:702882:703472 [2] NCCL INFO Channel 06/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-162-46:702885:703468 [5] NCCL INFO Channel 08/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-166-214:624946:625498 [3] NCCL INFO Channel 03/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-166-214:624947:625497 [4] NCCL INFO Channel 17/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-166-214:624945:625495 [2] NCCL INFO Channel 06/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-162-46:702884:703470 [4] NCCL INFO Channel 13/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-162-46:702883:703471 [3] NCCL INFO Channel 17/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-166-214:624945:625495 [2] NCCL INFO Channel 07/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624946:625498 [3] NCCL INFO Channel 04/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-162-46:702882:703472 [2] NCCL INFO Channel 07/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-162-46:702885:703468 [5] NCCL INFO Channel 09/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-166-214:624947:625497 [4] NCCL INFO Channel 18/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-166-214:624948:625500 [5] NCCL INFO Channel 06/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-162-46:702884:703470 [4] NCCL INFO Channel 14/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-162-46:702883:703471 [3] NCCL INFO Channel 18/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-162-46:702885:703468 [5] NCCL INFO Channel 10/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-162-46:702882:703472 [2] NCCL INFO Channel 08/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-167-9:1039541:1040097 [4] NCCL INFO Channel 13/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-166-15:127294:127835 [4] NCCL INFO Channel 10/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-166-214:624945:625495 [2] NCCL INFO Channel 08/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624946:625498 [3] NCCL INFO Channel 05/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-166-214:624947:625497 [4] NCCL INFO Channel 19/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-162-46:702884:703470 [4] NCCL INFO Channel 15/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-166-214:624948:625500 [5] NCCL INFO Channel 07/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-162-46:702883:703471 [3] NCCL INFO Channel 19/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-162-46:702885:703468 [5] NCCL INFO Channel 11/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-162-46:702882:703472 [2] NCCL INFO Channel 09/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624945:625495 [2] NCCL INFO Channel 09/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624946:625498 [3] NCCL INFO Channel 06/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-166-214:624947:625497 [4] NCCL INFO Channel 20/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-166-214:624948:625500 [5] NCCL INFO Channel 08/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-162-46:702884:703470 [4] NCCL INFO Channel 16/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-162-46:702883:703471 [3] NCCL INFO Channel 20/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-167-9:1039539:1040100 [2] NCCL INFO Channel 03/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-162-46:702885:703468 [5] NCCL INFO Channel 12/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-162-46:702882:703472 [2] NCCL INFO Channel 10/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624945:625495 [2] NCCL INFO Channel 10/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624946:625498 [3] NCCL INFO Channel 07/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-166-214:624947:625497 [4] NCCL INFO Channel 21/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-166-214:624948:625500 [5] NCCL INFO Channel 09/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-162-46:702884:703470 [4] NCCL INFO Channel 17/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-162-46:702883:703471 [3] NCCL INFO Channel 21/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-162-46:702885:703468 [5] NCCL INFO Channel 13/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-162-46:702882:703472 [2] NCCL INFO Channel 11/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127292:127836 [2] NCCL INFO Channel 15/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624945:625495 [2] NCCL INFO Channel 11/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624946:625498 [3] NCCL INFO Channel 08/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-166-214:624947:625497 [4] NCCL INFO Channel 22/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-166-214:624948:625500 [5] NCCL INFO Channel 10/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-162-46:702884:703470 [4] NCCL INFO Channel 18/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-162-46:702883:703471 [3] NCCL INFO Channel 22/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-162-46:702882:703472 [2] NCCL INFO Channel 12/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-162-46:702885:703468 [5] NCCL INFO Channel 14/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-166-15:127297:127840 [7] NCCL INFO Connected all trees
+ip-26-0-165-213:9629:10171 [4] NCCL INFO Channel 17/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-166-214:624945:625495 [2] NCCL INFO Channel 12/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624946:625498 [3] NCCL INFO Channel 09/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-166-214:624947:625497 [4] NCCL INFO Channel 23/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-166-214:624948:625500 [5] NCCL INFO Channel 11/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-166-15:127297:127840 [7] NCCL INFO NVLS comm 0xbee6820 headRank 3 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-162-46:702884:703470 [4] NCCL INFO Channel 19/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-165-213:9627:10174 [2] NCCL INFO Channel 17/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-162-46:702883:703471 [3] NCCL INFO Channel 23/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-162-46:702882:703472 [2] NCCL INFO Channel 13/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-167-9:1039541:1040097 [4] NCCL INFO Channel 14/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-162-46:702885:703468 [5] NCCL INFO Channel 15/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-162-46:702884:703470 [4] NCCL INFO Channel 20/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-162-46:702882:703472 [2] NCCL INFO Channel 14/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-162-46:702885:703468 [5] NCCL INFO Channel 16/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-166-214:624945:625495 [2] NCCL INFO Channel 13/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624946:625498 [3] NCCL INFO Channel 10/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-167-9:1039539:1040100 [2] NCCL INFO Channel 04/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624948:625500 [5] NCCL INFO Channel 12/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-166-214:624945:625495 [2] NCCL INFO Channel 14/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624946:625498 [3] NCCL INFO Channel 11/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-166-214:624948:625500 [5] NCCL INFO Channel 13/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-162-46:702884:703470 [4] NCCL INFO Channel 21/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-162-46:702882:703472 [2] NCCL INFO Channel 15/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-162-46:702885:703468 [5] NCCL INFO Channel 17/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-166-214:624946:625498 [3] NCCL INFO Channel 12/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-166-214:624945:625495 [2] NCCL INFO Channel 15/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624948:625500 [5] NCCL INFO Channel 14/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-167-9:1039541:1040097 [4] NCCL INFO Channel 15/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-165-213:9629:10171 [4] NCCL INFO Channel 18/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-166-214:624946:625498 [3] NCCL INFO Channel 13/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-166-214:624945:625495 [2] NCCL INFO Channel 16/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624948:625500 [5] NCCL INFO Channel 15/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-166-15:127294:127835 [4] NCCL INFO Channel 11/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-166-214:624946:625498 [3] NCCL INFO Channel 14/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-166-214:624945:625495 [2] NCCL INFO Channel 17/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624948:625500 [5] NCCL INFO Channel 16/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-162-46:702884:703470 [4] NCCL INFO Channel 22/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-162-46:702885:703468 [5] NCCL INFO Channel 18/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-162-46:702882:703472 [2] NCCL INFO Channel 16/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-165-213:9629:10171 [4] NCCL INFO Channel 19/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-167-9:1039541:1040097 [4] NCCL INFO Channel 16/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-167-9:1039539:1040100 [2] NCCL INFO Channel 05/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127292:127836 [2] NCCL INFO Channel 16/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624945:625495 [2] NCCL INFO Channel 18/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624946:625498 [3] NCCL INFO Channel 15/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-162-46:702884:703470 [4] NCCL INFO Channel 23/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-162-46:702882:703472 [2] NCCL INFO Channel 17/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-162-46:702885:703468 [5] NCCL INFO Channel 19/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-167-9:1039541:1040097 [4] NCCL INFO Channel 17/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-166-214:624945:625495 [2] NCCL INFO Channel 19/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624946:625498 [3] NCCL INFO Channel 16/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-166-214:624948:625500 [5] NCCL INFO Channel 17/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-165-213:9629:10171 [4] NCCL INFO Channel 20/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-166-15:127292:127836 [2] NCCL INFO Channel 17/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-162-46:702882:703472 [2] NCCL INFO Channel 18/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-162-46:702885:703468 [5] NCCL INFO Channel 20/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-167-9:1039539:1040100 [2] NCCL INFO Channel 06/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-167-9:1039541:1040097 [4] NCCL INFO Channel 18/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-166-15:127294:127835 [4] NCCL INFO Channel 12/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-162-46:702882:703472 [2] NCCL INFO Channel 19/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-162-46:702885:703468 [5] NCCL INFO Channel 21/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-165-213:9632:10176 [7] NCCL INFO Connected all trees
+ip-26-0-166-214:624946:625498 [3] NCCL INFO Channel 17/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-165-213:9627:10174 [2] NCCL INFO Channel 18/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624945:625495 [2] NCCL INFO Channel 20/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-165-213:9632:10176 [7] NCCL INFO NVLS comm 0xae15940 headRank 3 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-166-15:127292:127836 [2] NCCL INFO Channel 18/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-165-213:9629:10171 [4] NCCL INFO Channel 21/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-166-214:624946:625498 [3] NCCL INFO Channel 18/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-166-214:624945:625495 [2] NCCL INFO Channel 21/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624948:625500 [5] NCCL INFO Channel 18/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-167-9:1039539:1040100 [2] NCCL INFO Channel 07/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-162-46:702882:703472 [2] NCCL INFO Channel 20/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-162-46:702885:703468 [5] NCCL INFO Channel 22/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-166-15:127294:127835 [4] NCCL INFO Channel 13/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-167-9:1039541:1040097 [4] NCCL INFO Channel 19/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-165-213:9627:10174 [2] NCCL INFO Channel 19/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624946:625498 [3] NCCL INFO Channel 19/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-166-214:624945:625495 [2] NCCL INFO Channel 22/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624948:625500 [5] NCCL INFO Channel 19/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-167-9:1039539:1040100 [2] NCCL INFO Channel 08/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-165-213:9629:10171 [4] NCCL INFO Channel 22/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-166-214:624946:625498 [3] NCCL INFO Channel 20/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-166-214:624945:625495 [2] NCCL INFO Channel 23/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624948:625500 [5] NCCL INFO Channel 20/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-166-15:127294:127835 [4] NCCL INFO Channel 14/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-167-9:1039541:1040097 [4] NCCL INFO Channel 20/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-165-213:9627:10174 [2] NCCL INFO Channel 20/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624946:625498 [3] NCCL INFO Channel 21/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-166-214:624948:625500 [5] NCCL INFO Channel 21/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-167-9:1039539:1040100 [2] NCCL INFO Channel 09/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-162-46:702882:703472 [2] NCCL INFO Channel 21/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-162-46:702885:703468 [5] NCCL INFO Channel 23/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-166-15:127292:127836 [2] NCCL INFO Channel 19/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-167-9:1039541:1040097 [4] NCCL INFO Channel 21/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-165-213:9627:10174 [2] NCCL INFO Channel 21/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-165-213:9629:10171 [4] NCCL INFO Channel 23/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-166-214:624948:625500 [5] NCCL INFO Channel 22/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-166-214:624946:625498 [3] NCCL INFO Channel 22/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-166-15:127292:127836 [2] NCCL INFO Channel 20/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-162-46:702882:703472 [2] NCCL INFO Channel 22/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762816:763376 [1] NCCL INFO Connected all rings
+ip-26-0-166-15:127294:127835 [4] NCCL INFO Channel 15/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-165-213:9627:10174 [2] NCCL INFO Channel 22/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-167-9:1039539:1040100 [2] NCCL INFO Channel 10/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-167-9:1039541:1040097 [4] NCCL INFO Channel 22/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-167-9:1039539:1040100 [2] NCCL INFO Channel 11/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-165-213:9627:10174 [2] NCCL INFO Channel 23/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127291:127838 [1] NCCL INFO Connected all trees
+ip-26-0-167-9:1039541:1040097 [4] NCCL INFO Channel 23/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-166-15:127291:127838 [1] NCCL INFO NVLS comm 0xc33fd70 headRank 0 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-166-15:127292:127836 [2] NCCL INFO Channel 21/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-167-9:1039539:1040100 [2] NCCL INFO Channel 12/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-162-46:702882:703472 [2] NCCL INFO Channel 23/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127292:127836 [2] NCCL INFO Channel 22/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127294:127835 [4] NCCL INFO Channel 16/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-166-214:624948:625500 [5] NCCL INFO Channel 23/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-166-214:624946:625498 [3] NCCL INFO Channel 23/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-166-15:127292:127836 [2] NCCL INFO Channel 23/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-167-9:1039539:1040100 [2] NCCL INFO Channel 13/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-167-9:1039539:1040100 [2] NCCL INFO Channel 14/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127294:127835 [4] NCCL INFO Channel 17/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-166-244:897240:897786 [0] NCCL INFO Connected all trees
+ip-26-0-166-244:897240:897786 [0] NCCL INFO NVLS comm 0xb43e500 headRank 0 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-166-15:127294:127835 [4] NCCL INFO Channel 18/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-167-9:1039539:1040100 [2] NCCL INFO Channel 15/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-167-9:1039539:1040100 [2] NCCL INFO Channel 16/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-166-214:624949:625496 [6] NCCL INFO Connected all trees
+ip-26-0-166-214:624949:625496 [6] NCCL INFO NVLS comm 0xaba0260 headRank 3 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-167-9:1039540:1040103 [3] NCCL INFO Connected all trees
+ip-26-0-167-9:1039539:1040100 [2] NCCL INFO Channel 17/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-167-9:1039540:1040103 [3] NCCL INFO NVLS comm 0xae831b0 headRank 1 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-166-125:762815:763371 [0] NCCL INFO Connected all rings
+ip-26-0-166-15:127294:127835 [4] NCCL INFO Channel 19/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-167-9:1039539:1040100 [2] NCCL INFO Channel 18/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-167-9:1039539:1040100 [2] NCCL INFO Channel 19/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127294:127835 [4] NCCL INFO Channel 20/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-166-125:762818:763378 [3] NCCL INFO Connected all rings
+ip-26-0-167-9:1039539:1040100 [2] NCCL INFO Channel 20/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762820:763379 [5] NCCL INFO Connected all rings
+ip-26-0-167-9:1039542:1040102 [5] NCCL INFO Connected all trees
+ip-26-0-167-9:1039542:1040102 [5] NCCL INFO NVLS comm 0xa547f30 headRank 2 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-166-15:127294:127835 [4] NCCL INFO Channel 21/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-166-125:762819:763372 [4] NCCL INFO Connected all rings
+ip-26-0-166-36:104430:104974 [1] NCCL INFO Connected all rings
+ip-26-0-166-125:762822:763377 [7] NCCL INFO Connected all rings
+ip-26-0-166-125:762822:763377 [7] NCCL INFO Channel 00/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-166-36:104436:104978 [7] NCCL INFO Connected all rings
+ip-26-0-166-36:104436:104978 [7] NCCL INFO Channel 00/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-166-125:762817:763374 [2] NCCL INFO Connected all rings
+ip-26-0-166-125:762822:763377 [7] NCCL INFO Channel 01/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-166-15:127294:127835 [4] NCCL INFO Channel 22/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-166-125:762821:763373 [6] NCCL INFO Connected all rings
+ip-26-0-166-125:762821:763373 [6] NCCL INFO Channel 00/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-166-125:762822:763377 [7] NCCL INFO Channel 02/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-166-125:762821:763373 [6] NCCL INFO Channel 01/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-166-36:104434:104977 [5] NCCL INFO Connected all rings
+ip-26-0-166-125:762822:763377 [7] NCCL INFO Channel 03/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-166-125:762821:763373 [6] NCCL INFO Channel 02/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-166-125:762822:763377 [7] NCCL INFO Channel 04/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-166-125:762821:763373 [6] NCCL INFO Channel 03/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-166-15:127294:127835 [4] NCCL INFO Channel 23/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-166-125:762822:763377 [7] NCCL INFO Channel 05/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-162-46:702881:703465 [1] NCCL INFO Connected all trees
+ip-26-0-166-125:762821:763373 [6] NCCL INFO Channel 04/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-162-46:702881:703465 [1] NCCL INFO NVLS comm 0xb5043f0 headRank 0 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-166-125:762822:763377 [7] NCCL INFO Channel 06/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-166-125:762821:763373 [6] NCCL INFO Channel 05/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-166-125:762822:763377 [7] NCCL INFO Channel 07/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-166-125:762821:763373 [6] NCCL INFO Channel 06/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-166-125:762822:763377 [7] NCCL INFO Channel 08/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-166-125:762821:763373 [6] NCCL INFO Channel 07/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-166-125:762822:763377 [7] NCCL INFO Channel 09/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-166-125:762821:763373 [6] NCCL INFO Channel 08/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-166-36:104436:104978 [7] NCCL INFO Channel 01/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-166-125:762822:763377 [7] NCCL INFO Channel 10/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-166-125:762821:763373 [6] NCCL INFO Channel 09/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-166-125:762822:763377 [7] NCCL INFO Channel 11/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-166-125:762821:763373 [6] NCCL INFO Channel 10/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-166-125:762822:763377 [7] NCCL INFO Channel 12/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-166-125:762821:763373 [6] NCCL INFO Channel 11/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-166-125:762822:763377 [7] NCCL INFO Channel 13/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-166-125:762821:763373 [6] NCCL INFO Channel 12/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-166-125:762822:763377 [7] NCCL INFO Channel 14/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-166-125:762821:763373 [6] NCCL INFO Channel 13/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-166-125:762822:763377 [7] NCCL INFO Channel 15/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-166-125:762821:763373 [6] NCCL INFO Channel 14/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-166-125:762822:763377 [7] NCCL INFO Channel 16/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-166-125:762821:763373 [6] NCCL INFO Channel 15/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-166-125:762822:763377 [7] NCCL INFO Channel 17/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-166-125:762821:763373 [6] NCCL INFO Channel 16/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-166-125:762822:763377 [7] NCCL INFO Channel 18/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-166-125:762821:763373 [6] NCCL INFO Channel 17/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-166-36:104436:104978 [7] NCCL INFO Channel 02/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-167-9:1039539:1040100 [2] NCCL INFO Channel 21/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762822:763377 [7] NCCL INFO Channel 19/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-166-125:762821:763373 [6] NCCL INFO Channel 18/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-166-125:762822:763377 [7] NCCL INFO Channel 20/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-166-125:762821:763373 [6] NCCL INFO Channel 19/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-166-125:762822:763377 [7] NCCL INFO Channel 21/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-166-125:762821:763373 [6] NCCL INFO Channel 20/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-166-125:762822:763377 [7] NCCL INFO Channel 22/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-166-125:762821:763373 [6] NCCL INFO Channel 21/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-166-125:762822:763377 [7] NCCL INFO Channel 23/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-166-125:762821:763373 [6] NCCL INFO Channel 22/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-166-125:762818:763378 [3] NCCL INFO Channel 00/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-166-125:762820:763379 [5] NCCL INFO Channel 00/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-166-125:762819:763372 [4] NCCL INFO Channel 00/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-166-125:762821:763373 [6] NCCL INFO Channel 23/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-166-125:762818:763378 [3] NCCL INFO Channel 01/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-166-125:762817:763374 [2] NCCL INFO Channel 00/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762820:763379 [5] NCCL INFO Channel 01/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-166-125:762819:763372 [4] NCCL INFO Channel 01/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-167-9:1039539:1040100 [2] NCCL INFO Channel 22/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762818:763378 [3] NCCL INFO Channel 02/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-166-125:762817:763374 [2] NCCL INFO Channel 01/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762820:763379 [5] NCCL INFO Channel 02/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-166-125:762819:763372 [4] NCCL INFO Channel 02/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-166-125:762818:763378 [3] NCCL INFO Channel 03/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-166-36:104436:104978 [7] NCCL INFO Channel 03/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-166-125:762817:763374 [2] NCCL INFO Channel 02/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762820:763379 [5] NCCL INFO Channel 03/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-166-125:762819:763372 [4] NCCL INFO Channel 03/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-167-9:1039539:1040100 [2] NCCL INFO Channel 23/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762818:763378 [3] NCCL INFO Channel 04/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-166-125:762817:763374 [2] NCCL INFO Channel 03/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762820:763379 [5] NCCL INFO Channel 04/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-166-125:762819:763372 [4] NCCL INFO Channel 04/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-166-125:762818:763378 [3] NCCL INFO Channel 05/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-166-125:762820:763379 [5] NCCL INFO Channel 05/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-166-125:762817:763374 [2] NCCL INFO Channel 04/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762819:763372 [4] NCCL INFO Channel 05/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-166-244:897247:897788 [7] NCCL INFO Connected all trees
+ip-26-0-166-244:897247:897788 [7] NCCL INFO NVLS comm 0xbacf380 headRank 3 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-166-125:762817:763374 [2] NCCL INFO Channel 05/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762818:763378 [3] NCCL INFO Channel 06/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-166-125:762820:763379 [5] NCCL INFO Channel 06/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-166-125:762819:763372 [4] NCCL INFO Channel 06/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-166-125:762818:763378 [3] NCCL INFO Channel 07/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-166-125:762817:763374 [2] NCCL INFO Channel 06/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762820:763379 [5] NCCL INFO Channel 07/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-166-125:762819:763372 [4] NCCL INFO Channel 07/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-166-36:104432:104980 [3] NCCL INFO Connected all rings
+ip-26-0-166-125:762817:763374 [2] NCCL INFO Channel 07/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762820:763379 [5] NCCL INFO Channel 08/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-166-125:762818:763378 [3] NCCL INFO Channel 08/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-166-125:762819:763372 [4] NCCL INFO Channel 08/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-166-125:762820:763379 [5] NCCL INFO Channel 09/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-166-125:762817:763374 [2] NCCL INFO Channel 08/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762818:763378 [3] NCCL INFO Channel 09/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-166-125:762819:763372 [4] NCCL INFO Channel 09/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-166-125:762820:763379 [5] NCCL INFO Channel 10/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-166-125:762817:763374 [2] NCCL INFO Channel 09/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762818:763378 [3] NCCL INFO Channel 10/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-166-125:762819:763372 [4] NCCL INFO Channel 10/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-166-125:762820:763379 [5] NCCL INFO Channel 11/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-166-125:762817:763374 [2] NCCL INFO Channel 10/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762818:763378 [3] NCCL INFO Channel 11/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-166-125:762819:763372 [4] NCCL INFO Channel 11/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-166-125:762820:763379 [5] NCCL INFO Channel 12/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-166-125:762818:763378 [3] NCCL INFO Channel 12/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-166-125:762819:763372 [4] NCCL INFO Channel 12/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-166-125:762817:763374 [2] NCCL INFO Channel 11/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762820:763379 [5] NCCL INFO Channel 13/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-166-125:762818:763378 [3] NCCL INFO Channel 13/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-166-125:762819:763372 [4] NCCL INFO Channel 13/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-166-125:762817:763374 [2] NCCL INFO Channel 12/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-166-244:897241:897784 [1] NCCL INFO Connected all trees
+ip-26-0-166-125:762820:763379 [5] NCCL INFO Channel 14/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-166-244:897241:897784 [1] NCCL INFO NVLS comm 0xc2e1ee0 headRank 0 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-166-125:762818:763378 [3] NCCL INFO Channel 14/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-166-125:762819:763372 [4] NCCL INFO Channel 14/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-166-125:762817:763374 [2] NCCL INFO Channel 13/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762820:763379 [5] NCCL INFO Channel 15/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-166-125:762819:763372 [4] NCCL INFO Channel 15/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-166-125:762818:763378 [3] NCCL INFO Channel 15/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-166-125:762817:763374 [2] NCCL INFO Channel 14/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762820:763379 [5] NCCL INFO Channel 16/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-166-125:762819:763372 [4] NCCL INFO Channel 16/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-166-125:762818:763378 [3] NCCL INFO Channel 16/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-166-125:762817:763374 [2] NCCL INFO Channel 15/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762820:763379 [5] NCCL INFO Channel 17/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-166-125:762819:763372 [4] NCCL INFO Channel 17/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-166-125:762818:763378 [3] NCCL INFO Channel 17/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-166-125:762817:763374 [2] NCCL INFO Channel 16/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762820:763379 [5] NCCL INFO Channel 18/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-166-125:762819:763372 [4] NCCL INFO Channel 18/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-166-125:762818:763378 [3] NCCL INFO Channel 18/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-166-125:762817:763374 [2] NCCL INFO Channel 17/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762819:763372 [4] NCCL INFO Channel 19/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-166-125:762820:763379 [5] NCCL INFO Channel 19/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-166-125:762818:763378 [3] NCCL INFO Channel 19/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-166-125:762817:763374 [2] NCCL INFO Channel 18/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762818:763378 [3] NCCL INFO Channel 20/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-166-125:762819:763372 [4] NCCL INFO Channel 20/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-166-125:762820:763379 [5] NCCL INFO Channel 20/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-166-125:762817:763374 [2] NCCL INFO Channel 19/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762820:763379 [5] NCCL INFO Channel 21/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-166-125:762819:763372 [4] NCCL INFO Channel 21/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-166-125:762818:763378 [3] NCCL INFO Channel 21/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-166-125:762817:763374 [2] NCCL INFO Channel 20/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-166-36:104429:104973 [0] NCCL INFO Connected all rings
+ip-26-0-166-36:104436:104978 [7] NCCL INFO Channel 04/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-166-125:762820:763379 [5] NCCL INFO Channel 22/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-166-125:762818:763378 [3] NCCL INFO Channel 22/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-166-125:762819:763372 [4] NCCL INFO Channel 22/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-166-125:762817:763374 [2] NCCL INFO Channel 21/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762820:763379 [5] NCCL INFO Channel 23/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-166-125:762819:763372 [4] NCCL INFO Channel 23/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-166-125:762818:763378 [3] NCCL INFO Channel 23/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-166-125:762817:763374 [2] NCCL INFO Channel 22/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762817:763374 [2] NCCL INFO Channel 23/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-166-36:104435:104975 [6] NCCL INFO Connected all rings
+ip-26-0-166-36:104435:104975 [6] NCCL INFO Channel 00/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-166-36:104436:104978 [7] NCCL INFO Channel 05/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-167-9:1039538:1040099 [1] NCCL INFO Connected NVLS tree
+ip-26-0-167-9:1039538:1040099 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-167-9:1039538:1040099 [1] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-167-9:1039538:1040099 [1] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-166-36:104433:104976 [4] NCCL INFO Connected all rings
+ip-26-0-167-9:1039542:1040102 [5] NCCL INFO Connected NVLS tree
+ip-26-0-167-9:1039542:1040102 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-167-9:1039542:1040102 [5] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-167-9:1039542:1040102 [5] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-166-36:104431:104979 [2] NCCL INFO Connected all rings
+ip-26-0-167-9:1039544:1040101 [7] NCCL INFO Connected NVLS tree
+ip-26-0-167-9:1039544:1040101 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-167-9:1039544:1040101 [7] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-167-9:1039544:1040101 [7] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-167-9:1039543:1040098 [6] NCCL INFO Connected all trees
+ip-26-0-167-9:1039543:1040098 [6] NCCL INFO NVLS comm 0xb657270 headRank 3 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-166-36:104436:104978 [7] NCCL INFO Channel 06/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-166-36:104435:104975 [6] NCCL INFO Channel 01/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-167-9:1039540:1040103 [3] NCCL INFO Connected NVLS tree
+ip-26-0-167-9:1039540:1040103 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-167-9:1039540:1040103 [3] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-167-9:1039540:1040103 [3] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-166-36:104436:104978 [7] NCCL INFO Channel 07/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-166-36:104435:104975 [6] NCCL INFO Channel 02/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-166-36:104435:104975 [6] NCCL INFO Channel 03/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-166-36:104436:104978 [7] NCCL INFO Channel 08/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-165-213:9628:10177 [3] NCCL INFO Connected all trees
+ip-26-0-166-36:104435:104975 [6] NCCL INFO Channel 04/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-166-36:104436:104978 [7] NCCL INFO Channel 09/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-165-213:9628:10177 [3] NCCL INFO NVLS comm 0xb42cd50 headRank 1 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-166-36:104435:104975 [6] NCCL INFO Channel 05/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-166-36:104436:104978 [7] NCCL INFO Channel 10/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-166-36:104435:104975 [6] NCCL INFO Channel 06/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-166-36:104436:104978 [7] NCCL INFO Channel 11/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-166-36:104436:104978 [7] NCCL INFO Channel 12/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-166-36:104435:104975 [6] NCCL INFO Channel 07/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-166-36:104436:104978 [7] NCCL INFO Channel 13/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-166-36:104435:104975 [6] NCCL INFO Channel 08/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-166-36:104436:104978 [7] NCCL INFO Channel 14/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-166-36:104435:104975 [6] NCCL INFO Channel 09/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-166-36:104436:104978 [7] NCCL INFO Channel 15/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-166-36:104435:104975 [6] NCCL INFO Channel 10/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-166-36:104436:104978 [7] NCCL INFO Channel 16/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-166-36:104435:104975 [6] NCCL INFO Channel 11/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-166-36:104436:104978 [7] NCCL INFO Channel 17/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-166-36:104435:104975 [6] NCCL INFO Channel 12/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-166-36:104436:104978 [7] NCCL INFO Channel 18/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-166-36:104435:104975 [6] NCCL INFO Channel 13/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-166-36:104436:104978 [7] NCCL INFO Channel 19/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-166-36:104435:104975 [6] NCCL INFO Channel 14/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-166-36:104436:104978 [7] NCCL INFO Channel 20/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-166-36:104435:104975 [6] NCCL INFO Channel 15/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-166-36:104436:104978 [7] NCCL INFO Channel 21/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-166-36:104435:104975 [6] NCCL INFO Channel 16/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-166-36:104436:104978 [7] NCCL INFO Channel 22/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-166-36:104435:104975 [6] NCCL INFO Channel 17/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-167-9:1039540:1040103 [3] NCCL INFO comm 0xae831b0 rank 1 nranks 4 cudaDev 3 nvmlDev 3 busId 86000 commId 0x689c160a9f6f8f23 - Init COMPLETE
+ip-26-0-167-9:1039538:1040099 [1] NCCL INFO comm 0xa731d00 rank 0 nranks 4 cudaDev 1 nvmlDev 1 busId 64000 commId 0x689c160a9f6f8f23 - Init COMPLETE
+ip-26-0-167-9:1039542:1040102 [5] NCCL INFO comm 0xa547f30 rank 2 nranks 4 cudaDev 5 nvmlDev 5 busId a8000 commId 0x689c160a9f6f8f23 - Init COMPLETE
+ip-26-0-167-9:1039544:1040101 [7] NCCL INFO comm 0xad8eb40 rank 3 nranks 4 cudaDev 7 nvmlDev 7 busId ca000 commId 0x689c160a9f6f8f23 - Init COMPLETE
+ip-26-0-166-36:104436:104978 [7] NCCL INFO Channel 23/0 : 3[7] -> 2[5] via P2P/IPC
+ip-26-0-166-36:104435:104975 [6] NCCL INFO Channel 18/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-166-36:104434:104977 [5] NCCL INFO Channel 00/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-165-213:9630:10175 [5] NCCL INFO Connected all trees
+ip-26-0-165-213:9630:10175 [5] NCCL INFO NVLS comm 0xa8b9ce0 headRank 2 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-166-36:104434:104977 [5] NCCL INFO Channel 01/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-166-36:104435:104975 [6] NCCL INFO Channel 19/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-166-36:104434:104977 [5] NCCL INFO Channel 02/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-166-15:127295:127839 [5] NCCL INFO Connected all trees
+ip-26-0-166-36:104435:104975 [6] NCCL INFO Channel 20/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-166-15:127295:127839 [5] NCCL INFO NVLS comm 0xabbd310 headRank 2 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-166-36:104434:104977 [5] NCCL INFO Channel 03/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-166-36:104435:104975 [6] NCCL INFO Channel 21/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-166-36:104433:104976 [4] NCCL INFO Channel 00/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-166-36:104434:104977 [5] NCCL INFO Channel 04/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-166-36:104435:104975 [6] NCCL INFO Channel 22/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-166-36:104433:104976 [4] NCCL INFO Channel 01/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-166-36:104432:104980 [3] NCCL INFO Channel 00/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-166-36:104434:104977 [5] NCCL INFO Channel 05/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-166-36:104435:104975 [6] NCCL INFO Channel 23/0 : 3[6] -> 2[4] via P2P/IPC
+ip-26-0-166-36:104433:104976 [4] NCCL INFO Channel 02/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-166-36:104432:104980 [3] NCCL INFO Channel 01/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-162-46:702886:703469 [6] NCCL INFO Connected all trees
+ip-26-0-166-36:104434:104977 [5] NCCL INFO Channel 06/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-162-46:702886:703469 [6] NCCL INFO NVLS comm 0xb144580 headRank 3 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-166-36:104433:104976 [4] NCCL INFO Channel 03/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-166-36:104432:104980 [3] NCCL INFO Channel 02/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-166-36:104434:104977 [5] NCCL INFO Channel 07/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-166-36:104433:104976 [4] NCCL INFO Channel 04/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-166-36:104432:104980 [3] NCCL INFO Channel 03/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-166-36:104434:104977 [5] NCCL INFO Channel 08/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-166-36:104433:104976 [4] NCCL INFO Channel 05/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-166-36:104432:104980 [3] NCCL INFO Channel 04/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-166-36:104434:104977 [5] NCCL INFO Channel 09/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-166-36:104431:104979 [2] NCCL INFO Channel 00/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-166-36:104433:104976 [4] NCCL INFO Channel 06/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-166-36:104432:104980 [3] NCCL INFO Channel 05/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-166-36:104434:104977 [5] NCCL INFO Channel 10/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-166-36:104431:104979 [2] NCCL INFO Channel 01/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-166-36:104433:104976 [4] NCCL INFO Channel 07/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-166-36:104432:104980 [3] NCCL INFO Channel 06/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-166-36:104434:104977 [5] NCCL INFO Channel 11/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-166-36:104431:104979 [2] NCCL INFO Channel 02/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-166-36:104432:104980 [3] NCCL INFO Channel 07/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-166-36:104433:104976 [4] NCCL INFO Channel 08/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-166-36:104434:104977 [5] NCCL INFO Channel 12/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-166-36:104431:104979 [2] NCCL INFO Channel 03/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-166-36:104432:104980 [3] NCCL INFO Channel 08/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-166-36:104433:104976 [4] NCCL INFO Channel 09/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-166-36:104434:104977 [5] NCCL INFO Channel 13/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-166-36:104431:104979 [2] NCCL INFO Channel 04/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-166-36:104432:104980 [3] NCCL INFO Channel 09/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-166-36:104433:104976 [4] NCCL INFO Channel 10/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-166-36:104434:104977 [5] NCCL INFO Channel 14/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-166-36:104431:104979 [2] NCCL INFO Channel 05/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-166-244:897246:897791 [6] NCCL INFO Connected all trees
+ip-26-0-166-15:127293:127841 [3] NCCL INFO Connected all trees
+ip-26-0-166-36:104433:104976 [4] NCCL INFO Channel 11/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-166-36:104432:104980 [3] NCCL INFO Channel 10/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-166-244:897246:897791 [6] NCCL INFO NVLS comm 0xc54c400 headRank 3 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-166-15:127293:127841 [3] NCCL INFO NVLS comm 0xc22d550 headRank 1 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-166-36:104431:104979 [2] NCCL INFO Channel 06/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-166-36:104434:104977 [5] NCCL INFO Channel 15/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-166-36:104433:104976 [4] NCCL INFO Channel 12/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-166-15:127290:127833 [0] NCCL INFO Connected all trees
+ip-26-0-166-36:104432:104980 [3] NCCL INFO Channel 11/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-166-36:104434:104977 [5] NCCL INFO Channel 16/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-166-36:104431:104979 [2] NCCL INFO Channel 07/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127290:127833 [0] NCCL INFO NVLS comm 0xaac4d80 headRank 0 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-166-36:104433:104976 [4] NCCL INFO Channel 13/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-166-36:104434:104977 [5] NCCL INFO Channel 17/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-166-36:104432:104980 [3] NCCL INFO Channel 12/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-166-36:104431:104979 [2] NCCL INFO Channel 08/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-166-36:104433:104976 [4] NCCL INFO Channel 14/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-166-36:104434:104977 [5] NCCL INFO Channel 18/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-166-36:104432:104980 [3] NCCL INFO Channel 13/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-166-36:104431:104979 [2] NCCL INFO Channel 09/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-166-36:104433:104976 [4] NCCL INFO Channel 15/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-166-36:104434:104977 [5] NCCL INFO Channel 19/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-166-36:104432:104980 [3] NCCL INFO Channel 14/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-166-36:104431:104979 [2] NCCL INFO Channel 10/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-166-36:104433:104976 [4] NCCL INFO Channel 16/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-166-36:104432:104980 [3] NCCL INFO Channel 15/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-166-36:104434:104977 [5] NCCL INFO Channel 20/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-166-36:104431:104979 [2] NCCL INFO Channel 11/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-166-36:104433:104976 [4] NCCL INFO Channel 17/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-166-36:104434:104977 [5] NCCL INFO Channel 21/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-166-36:104432:104980 [3] NCCL INFO Channel 16/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-166-36:104431:104979 [2] NCCL INFO Channel 12/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-166-36:104433:104976 [4] NCCL INFO Channel 18/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-166-36:104434:104977 [5] NCCL INFO Channel 22/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-166-36:104432:104980 [3] NCCL INFO Channel 17/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-166-36:104431:104979 [2] NCCL INFO Channel 13/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-166-36:104433:104976 [4] NCCL INFO Channel 19/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-166-36:104434:104977 [5] NCCL INFO Channel 23/0 : 2[5] -> 1[3] via P2P/IPC
+ip-26-0-166-36:104432:104980 [3] NCCL INFO Channel 18/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-166-36:104431:104979 [2] NCCL INFO Channel 14/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-166-36:104432:104980 [3] NCCL INFO Channel 19/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-166-36:104433:104976 [4] NCCL INFO Channel 20/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-166-36:104431:104979 [2] NCCL INFO Channel 15/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-166-36:104432:104980 [3] NCCL INFO Channel 20/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-166-36:104433:104976 [4] NCCL INFO Channel 21/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-166-36:104431:104979 [2] NCCL INFO Channel 16/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-166-36:104433:104976 [4] NCCL INFO Channel 22/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-166-36:104432:104980 [3] NCCL INFO Channel 21/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-166-36:104431:104979 [2] NCCL INFO Channel 17/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-166-36:104431:104979 [2] NCCL INFO Channel 18/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-166-36:104431:104979 [2] NCCL INFO Channel 19/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-166-36:104431:104979 [2] NCCL INFO Channel 20/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-166-36:104431:104979 [2] NCCL INFO Channel 21/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-166-36:104431:104979 [2] NCCL INFO Channel 22/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-166-36:104431:104979 [2] NCCL INFO Channel 23/0 : 1[2] -> 0[0] via P2P/IPC
+ip-26-0-166-36:104432:104980 [3] NCCL INFO Channel 22/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-166-36:104433:104976 [4] NCCL INFO Channel 23/0 : 2[4] -> 1[2] via P2P/IPC
+ip-26-0-166-214:624943:625494 [0] NCCL INFO Connected all trees
+ip-26-0-166-214:624943:625494 [0] NCCL INFO NVLS comm 0xb569b60 headRank 0 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-166-36:104432:104980 [3] NCCL INFO Channel 23/0 : 1[3] -> 0[1] via P2P/IPC
+ip-26-0-165-213:9625:10169 [0] NCCL INFO Connected all trees
+ip-26-0-165-213:9625:10169 [0] NCCL INFO NVLS comm 0xaf094b0 headRank 0 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-162-46:702887:703467 [7] NCCL INFO Connected all trees
+ip-26-0-162-46:702887:703467 [7] NCCL INFO NVLS comm 0xac500c0 headRank 3 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-166-214:624944:625493 [1] NCCL INFO Connected all trees
+ip-26-0-166-214:624944:625493 [1] NCCL INFO NVLS comm 0xa637bf0 headRank 0 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-165-213:9631:10172 [6] NCCL INFO Connected all trees
+ip-26-0-165-213:9631:10172 [6] NCCL INFO NVLS comm 0xbbecca0 headRank 3 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-167-9:1039537:1040095 [0] NCCL INFO Connected all trees
+ip-26-0-167-9:1039537:1040095 [0] NCCL INFO NVLS comm 0xb3162d0 headRank 0 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-165-213:9628:10177 [3] NCCL INFO Connected NVLS tree
+ip-26-0-165-213:9628:10177 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-213:9628:10177 [3] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-165-213:9628:10177 [3] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-165-213:9626:10173 [1] NCCL INFO Connected NVLS tree
+ip-26-0-165-213:9626:10173 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-213:9630:10175 [5] NCCL INFO Connected NVLS tree
+ip-26-0-165-213:9630:10175 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-213:9626:10173 [1] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-165-213:9626:10173 [1] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-165-213:9630:10175 [5] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-165-213:9630:10175 [5] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-165-213:9632:10176 [7] NCCL INFO Connected NVLS tree
+ip-26-0-165-213:9632:10176 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-213:9632:10176 [7] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-165-213:9632:10176 [7] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-166-15:127297:127840 [7] NCCL INFO Connected NVLS tree
+ip-26-0-166-15:127297:127840 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-15:127297:127840 [7] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-166-15:127297:127840 [7] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-166-15:127295:127839 [5] NCCL INFO Connected NVLS tree
+ip-26-0-166-15:127295:127839 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-15:127295:127839 [5] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-166-15:127295:127839 [5] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-166-15:127291:127838 [1] NCCL INFO Connected NVLS tree
+ip-26-0-166-15:127291:127838 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-15:127293:127841 [3] NCCL INFO Connected NVLS tree
+ip-26-0-166-15:127293:127841 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-15:127291:127838 [1] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-166-15:127291:127838 [1] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-166-15:127293:127841 [3] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-166-15:127293:127841 [3] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-165-213:9632:10176 [7] NCCL INFO comm 0xae15940 rank 3 nranks 4 cudaDev 7 nvmlDev 7 busId ca000 commId 0x394a264b186d9de9 - Init COMPLETE
+ip-26-0-165-213:9630:10175 [5] NCCL INFO comm 0xa8b9ce0 rank 2 nranks 4 cudaDev 5 nvmlDev 5 busId a8000 commId 0x394a264b186d9de9 - Init COMPLETE
+ip-26-0-165-213:9628:10177 [3] NCCL INFO comm 0xb42cd50 rank 1 nranks 4 cudaDev 3 nvmlDev 3 busId 86000 commId 0x394a264b186d9de9 - Init COMPLETE
+ip-26-0-165-213:9626:10173 [1] NCCL INFO comm 0xb4e0900 rank 0 nranks 4 cudaDev 1 nvmlDev 1 busId 64000 commId 0x394a264b186d9de9 - Init COMPLETE
+ip-26-0-166-214:624950:625499 [7] NCCL INFO Connected all trees
+ip-26-0-166-214:624950:625499 [7] NCCL INFO NVLS comm 0xaec1440 headRank 3 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-166-244:897245:897787 [5] NCCL INFO Connected all trees
+ip-26-0-166-244:897245:897787 [5] NCCL INFO NVLS comm 0xbbdd3c0 headRank 2 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-166-244:897242:897792 [2] NCCL INFO Connected all trees
+ip-26-0-166-244:897242:897792 [2] NCCL INFO NVLS comm 0xa82b600 headRank 1 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-162-46:702880:703466 [0] NCCL INFO Connected all trees
+ip-26-0-162-46:702880:703466 [0] NCCL INFO NVLS comm 0xafe7d10 headRank 0 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-167-9:1039541:1040097 [4] NCCL INFO Connected all trees
+ip-26-0-167-9:1039541:1040097 [4] NCCL INFO NVLS comm 0xbb87bb0 headRank 2 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-167-9:1039539:1040100 [2] NCCL INFO Connected all trees
+ip-26-0-167-9:1039539:1040100 [2] NCCL INFO NVLS comm 0xa809100 headRank 1 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-166-244:897243:897790 [3] NCCL INFO Connected all trees
+ip-26-0-166-15:127296:127834 [6] NCCL INFO Connected all trees
+ip-26-0-166-244:897243:897790 [3] NCCL INFO NVLS comm 0xab319e0 headRank 1 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-166-15:127296:127834 [6] NCCL INFO NVLS comm 0xb7cefc0 headRank 3 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-166-15:127295:127839 [5] NCCL INFO comm 0xabbd310 rank 2 nranks 4 cudaDev 5 nvmlDev 5 busId a8000 commId 0xaabd8ffb497aa15 - Init COMPLETE
+ip-26-0-166-15:127291:127838 [1] NCCL INFO comm 0xc33fd70 rank 0 nranks 4 cudaDev 1 nvmlDev 1 busId 64000 commId 0xaabd8ffb497aa15 - Init COMPLETE
+ip-26-0-166-15:127293:127841 [3] NCCL INFO comm 0xc22d550 rank 1 nranks 4 cudaDev 3 nvmlDev 3 busId 86000 commId 0xaabd8ffb497aa15 - Init COMPLETE
+ip-26-0-166-15:127297:127840 [7] NCCL INFO comm 0xbee6820 rank 3 nranks 4 cudaDev 7 nvmlDev 7 busId ca000 commId 0xaabd8ffb497aa15 - Init COMPLETE
+ip-26-0-165-213:9627:10174 [2] NCCL INFO Connected all trees
+ip-26-0-165-213:9627:10174 [2] NCCL INFO NVLS comm 0xaa425a0 headRank 1 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-165-213:9629:10171 [4] NCCL INFO Connected all trees
+ip-26-0-165-213:9629:10171 [4] NCCL INFO NVLS comm 0xaa3ca00 headRank 2 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-166-125:762816:763376 [1] NCCL INFO Connected all trees
+ip-26-0-166-125:762816:763376 [1] NCCL INFO NVLS comm 0xb71fe00 headRank 0 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-166-244:897244:897789 [4] NCCL INFO Connected all trees
+ip-26-0-166-244:897244:897789 [4] NCCL INFO NVLS comm 0xafc12f0 headRank 2 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-166-214:624947:625497 [4] NCCL INFO Connected all trees
+ip-26-0-166-214:624947:625497 [4] NCCL INFO NVLS comm 0xc10da80 headRank 2 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-167-9:1039541:1040097 [4] NCCL INFO Connected NVLS tree
+ip-26-0-167-9:1039541:1040097 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-167-9:1039541:1040097 [4] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-167-9:1039541:1040097 [4] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-167-9:1039537:1040095 [0] NCCL INFO Connected NVLS tree
+ip-26-0-167-9:1039537:1040095 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-167-9:1039539:1040100 [2] NCCL INFO Connected NVLS tree
+ip-26-0-167-9:1039537:1040095 [0] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-167-9:1039537:1040095 [0] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-167-9:1039539:1040100 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-167-9:1039539:1040100 [2] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-167-9:1039539:1040100 [2] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-167-9:1039543:1040098 [6] NCCL INFO Connected NVLS tree
+ip-26-0-167-9:1039543:1040098 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-167-9:1039543:1040098 [6] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-167-9:1039543:1040098 [6] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-162-46:702884:703470 [4] NCCL INFO Connected all trees
+ip-26-0-162-46:702884:703470 [4] NCCL INFO NVLS comm 0xa5c1b20 headRank 2 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-166-15:127292:127836 [2] NCCL INFO Connected all trees
+ip-26-0-166-15:127292:127836 [2] NCCL INFO NVLS comm 0xb32c4c0 headRank 1 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-166-15:127294:127835 [4] NCCL INFO Connected all trees
+ip-26-0-166-15:127294:127835 [4] NCCL INFO NVLS comm 0xb0cc4f0 headRank 2 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-162-46:702885:703468 [5] NCCL INFO Connected all trees
+ip-26-0-162-46:702885:703468 [5] NCCL INFO NVLS comm 0xac5faf0 headRank 2 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-167-9:1039537:1040095 [0] NCCL INFO comm 0xb3162d0 rank 0 nranks 4 cudaDev 0 nvmlDev 0 busId 53000 commId 0xd981cf5f876ca6ba - Init COMPLETE
+ip-26-0-167-9:1039541:1040097 [4] NCCL INFO comm 0xbb87bb0 rank 2 nranks 4 cudaDev 4 nvmlDev 4 busId 97000 commId 0xd981cf5f876ca6ba - Init COMPLETE
+ip-26-0-167-9:1039543:1040098 [6] NCCL INFO comm 0xb657270 rank 3 nranks 4 cudaDev 6 nvmlDev 6 busId b9000 commId 0xd981cf5f876ca6ba - Init COMPLETE
+ip-26-0-167-9:1039539:1040100 [2] NCCL INFO comm 0xa809100 rank 1 nranks 4 cudaDev 2 nvmlDev 2 busId 75000 commId 0xd981cf5f876ca6ba - Init COMPLETE
+ip-26-0-165-213:9629:10171 [4] NCCL INFO Connected NVLS tree
+ip-26-0-165-213:9629:10171 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-213:9629:10171 [4] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-165-213:9629:10171 [4] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-165-213:9625:10169 [0] NCCL INFO Connected NVLS tree
+ip-26-0-165-213:9625:10169 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-213:9625:10169 [0] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-165-213:9625:10169 [0] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-165-213:9631:10172 [6] NCCL INFO Connected NVLS tree
+ip-26-0-165-213:9631:10172 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-213:9631:10172 [6] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-165-213:9631:10172 [6] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-165-213:9627:10174 [2] NCCL INFO Connected NVLS tree
+ip-26-0-165-213:9627:10174 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-213:9627:10174 [2] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-165-213:9627:10174 [2] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-166-125:762821:763373 [6] NCCL INFO Connected all trees
+ip-26-0-166-125:762821:763373 [6] NCCL INFO NVLS comm 0xa9cdcc0 headRank 3 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-162-46:702883:703471 [3] NCCL INFO Connected all trees
+ip-26-0-162-46:702883:703471 [3] NCCL INFO NVLS comm 0xa67ba70 headRank 1 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-166-125:762822:763377 [7] NCCL INFO Connected all trees
+ip-26-0-166-125:762822:763377 [7] NCCL INFO NVLS comm 0xc109000 headRank 3 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-166-214:624946:625498 [3] NCCL INFO Connected all trees
+ip-26-0-166-214:624946:625498 [3] NCCL INFO NVLS comm 0xc16d320 headRank 1 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-165-213:9631:10172 [6] NCCL INFO comm 0xbbecca0 rank 3 nranks 4 cudaDev 6 nvmlDev 6 busId b9000 commId 0x2a4e364c12c0a65b - Init COMPLETE
+ip-26-0-165-213:9625:10169 [0] NCCL INFO comm 0xaf094b0 rank 0 nranks 4 cudaDev 0 nvmlDev 0 busId 53000 commId 0x2a4e364c12c0a65b - Init COMPLETE
+ip-26-0-165-213:9629:10171 [4] NCCL INFO comm 0xaa3ca00 rank 2 nranks 4 cudaDev 4 nvmlDev 4 busId 97000 commId 0x2a4e364c12c0a65b - Init COMPLETE
+ip-26-0-165-213:9627:10174 [2] NCCL INFO comm 0xaa425a0 rank 1 nranks 4 cudaDev 2 nvmlDev 2 busId 75000 commId 0x2a4e364c12c0a65b - Init COMPLETE
+ip-26-0-166-125:762815:763371 [0] NCCL INFO Connected all trees
+ip-26-0-166-125:762815:763371 [0] NCCL INFO NVLS comm 0xa6f1020 headRank 0 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-166-244:897245:897787 [5] NCCL INFO Connected NVLS tree
+ip-26-0-166-244:897245:897787 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-244:897245:897787 [5] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-166-244:897245:897787 [5] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-166-244:897241:897784 [1] NCCL INFO Connected NVLS tree
+ip-26-0-166-244:897241:897784 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-244:897247:897788 [7] NCCL INFO Connected NVLS tree
+ip-26-0-166-244:897243:897790 [3] NCCL INFO Connected NVLS tree
+ip-26-0-166-244:897247:897788 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-244:897243:897790 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-244:897241:897784 [1] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-166-244:897241:897784 [1] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-166-244:897247:897788 [7] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-166-244:897247:897788 [7] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-166-244:897243:897790 [3] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-166-244:897243:897790 [3] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-166-214:624945:625495 [2] NCCL INFO Connected all trees
+ip-26-0-166-214:624945:625495 [2] NCCL INFO NVLS comm 0xb8d6660 headRank 1 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-166-36:104429:104973 [0] NCCL INFO Connected all trees
+ip-26-0-166-36:104429:104973 [0] NCCL INFO NVLS comm 0xc5849c0 headRank 0 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-166-244:897247:897788 [7] NCCL INFO comm 0xbacf380 rank 3 nranks 4 cudaDev 7 nvmlDev 7 busId ca000 commId 0x3fa48ab7a0d2f235 - Init COMPLETE
+ip-26-0-166-244:897241:897784 [1] NCCL INFO comm 0xc2e1ee0 rank 0 nranks 4 cudaDev 1 nvmlDev 1 busId 64000 commId 0x3fa48ab7a0d2f235 - Init COMPLETE
+ip-26-0-166-244:897243:897790 [3] NCCL INFO comm 0xab319e0 rank 1 nranks 4 cudaDev 3 nvmlDev 3 busId 86000 commId 0x3fa48ab7a0d2f235 - Init COMPLETE
+ip-26-0-166-244:897245:897787 [5] NCCL INFO comm 0xbbdd3c0 rank 2 nranks 4 cudaDev 5 nvmlDev 5 busId a8000 commId 0x3fa48ab7a0d2f235 - Init COMPLETE
+ip-26-0-166-15:127294:127835 [4] NCCL INFO Connected NVLS tree
+ip-26-0-166-15:127294:127835 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-15:127290:127833 [0] NCCL INFO Connected NVLS tree
+ip-26-0-166-15:127294:127835 [4] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-166-15:127294:127835 [4] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-166-15:127290:127833 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-15:127296:127834 [6] NCCL INFO Connected NVLS tree
+ip-26-0-166-15:127296:127834 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-15:127296:127834 [6] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-166-15:127296:127834 [6] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-166-15:127290:127833 [0] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-166-15:127290:127833 [0] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-166-15:127292:127836 [2] NCCL INFO Connected NVLS tree
+ip-26-0-166-15:127292:127836 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-15:127292:127836 [2] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-166-15:127292:127836 [2] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-162-46:702882:703472 [2] NCCL INFO Connected all trees
+ip-26-0-162-46:702882:703472 [2] NCCL INFO NVLS comm 0xb8bbb80 headRank 1 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-166-36:104436:104978 [7] NCCL INFO Connected all trees
+ip-26-0-166-36:104436:104978 [7] NCCL INFO NVLS comm 0xb855c60 headRank 3 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-166-214:624948:625500 [5] NCCL INFO Connected all trees
+ip-26-0-166-214:624948:625500 [5] NCCL INFO NVLS comm 0xc1820e0 headRank 2 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-166-15:127290:127833 [0] NCCL INFO comm 0xaac4d80 rank 0 nranks 4 cudaDev 0 nvmlDev 0 busId 53000 commId 0x5b4f6f555ef9c3fa - Init COMPLETE
+ip-26-0-166-15:127296:127834 [6] NCCL INFO comm 0xb7cefc0 rank 3 nranks 4 cudaDev 6 nvmlDev 6 busId b9000 commId 0x5b4f6f555ef9c3fa - Init COMPLETE
+ip-26-0-166-15:127292:127836 [2] NCCL INFO comm 0xb32c4c0 rank 1 nranks 4 cudaDev 2 nvmlDev 2 busId 75000 commId 0x5b4f6f555ef9c3fa - Init COMPLETE
+ip-26-0-166-15:127294:127835 [4] NCCL INFO comm 0xb0cc4f0 rank 2 nranks 4 cudaDev 4 nvmlDev 4 busId 97000 commId 0x5b4f6f555ef9c3fa - Init COMPLETE
+ip-26-0-166-244:897244:897789 [4] NCCL INFO Connected NVLS tree
+ip-26-0-166-244:897244:897789 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-244:897244:897789 [4] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-166-244:897244:897789 [4] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-166-244:897242:897792 [2] NCCL INFO Connected NVLS tree
+ip-26-0-166-244:897242:897792 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-244:897242:897792 [2] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-166-244:897242:897792 [2] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-166-244:897246:897791 [6] NCCL INFO Connected NVLS tree
+ip-26-0-166-244:897246:897791 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-244:897246:897791 [6] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-166-244:897246:897791 [6] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-166-244:897240:897786 [0] NCCL INFO Connected NVLS tree
+ip-26-0-166-244:897240:897786 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-244:897240:897786 [0] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-166-244:897240:897786 [0] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-162-46:702883:703471 [3] NCCL INFO Connected NVLS tree
+ip-26-0-162-46:702883:703471 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-162-46:702883:703471 [3] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-162-46:702883:703471 [3] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-162-46:702887:703467 [7] NCCL INFO Connected NVLS tree
+ip-26-0-162-46:702887:703467 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-162-46:702885:703468 [5] NCCL INFO Connected NVLS tree
+ip-26-0-162-46:702887:703467 [7] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-162-46:702887:703467 [7] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-162-46:702885:703468 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-162-46:702885:703468 [5] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-162-46:702885:703468 [5] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-162-46:702881:703465 [1] NCCL INFO Connected NVLS tree
+ip-26-0-162-46:702881:703465 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-162-46:702881:703465 [1] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-162-46:702881:703465 [1] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-166-36:104435:104975 [6] NCCL INFO Connected all trees
+ip-26-0-166-36:104435:104975 [6] NCCL INFO NVLS comm 0xaee4280 headRank 3 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-166-214:624945:625495 [2] NCCL INFO Connected NVLS tree
+ip-26-0-166-214:624945:625495 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-214:624945:625495 [2] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-166-214:624945:625495 [2] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-166-214:624947:625497 [4] NCCL INFO Connected NVLS tree
+ip-26-0-166-214:624947:625497 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-214:624947:625497 [4] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-166-214:624947:625497 [4] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-166-244:897244:897789 [4] NCCL INFO comm 0xafc12f0 rank 2 nranks 4 cudaDev 4 nvmlDev 4 busId 97000 commId 0xd3faa027769f7c66 - Init COMPLETE
+ip-26-0-166-244:897240:897786 [0] NCCL INFO comm 0xb43e500 rank 0 nranks 4 cudaDev 0 nvmlDev 0 busId 53000 commId 0xd3faa027769f7c66 - Init COMPLETE
+ip-26-0-166-244:897242:897792 [2] NCCL INFO comm 0xa82b600 rank 1 nranks 4 cudaDev 2 nvmlDev 2 busId 75000 commId 0xd3faa027769f7c66 - Init COMPLETE
+ip-26-0-166-214:624949:625496 [6] NCCL INFO Connected NVLS tree
+ip-26-0-166-214:624949:625496 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-244:897246:897791 [6] NCCL INFO comm 0xc54c400 rank 3 nranks 4 cudaDev 6 nvmlDev 6 busId b9000 commId 0xd3faa027769f7c66 - Init COMPLETE
+ip-26-0-166-214:624949:625496 [6] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-166-214:624949:625496 [6] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-166-214:624943:625494 [0] NCCL INFO Connected NVLS tree
+ip-26-0-166-214:624943:625494 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-214:624943:625494 [0] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-166-214:624943:625494 [0] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-166-125:762818:763378 [3] NCCL INFO Connected all trees
+ip-26-0-166-125:762818:763378 [3] NCCL INFO NVLS comm 0xa85c340 headRank 1 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-162-46:702883:703471 [3] NCCL INFO comm 0xa67ba70 rank 1 nranks 4 cudaDev 3 nvmlDev 3 busId 86000 commId 0x4908ed83bce0c92e - Init COMPLETE
+ip-26-0-162-46:702887:703467 [7] NCCL INFO comm 0xac500c0 rank 3 nranks 4 cudaDev 7 nvmlDev 7 busId ca000 commId 0x4908ed83bce0c92e - Init COMPLETE
+ip-26-0-162-46:702885:703468 [5] NCCL INFO comm 0xac5faf0 rank 2 nranks 4 cudaDev 5 nvmlDev 5 busId a8000 commId 0x4908ed83bce0c92e - Init COMPLETE
+ip-26-0-162-46:702881:703465 [1] NCCL INFO comm 0xb5043f0 rank 0 nranks 4 cudaDev 1 nvmlDev 1 busId 64000 commId 0x4908ed83bce0c92e - Init COMPLETE
+ip-26-0-166-214:624947:625497 [4] NCCL INFO comm 0xc10da80 rank 2 nranks 4 cudaDev 4 nvmlDev 4 busId 97000 commId 0x63ed815d5d2606 - Init COMPLETE
+ip-26-0-166-214:624945:625495 [2] NCCL INFO comm 0xb8d6660 rank 1 nranks 4 cudaDev 2 nvmlDev 2 busId 75000 commId 0x63ed815d5d2606 - Init COMPLETE
+ip-26-0-166-214:624943:625494 [0] NCCL INFO comm 0xb569b60 rank 0 nranks 4 cudaDev 0 nvmlDev 0 busId 53000 commId 0x63ed815d5d2606 - Init COMPLETE
+ip-26-0-166-214:624949:625496 [6] NCCL INFO comm 0xaba0260 rank 3 nranks 4 cudaDev 6 nvmlDev 6 busId b9000 commId 0x63ed815d5d2606 - Init COMPLETE
+ip-26-0-166-36:104430:104974 [1] NCCL INFO Connected all trees
+ip-26-0-166-36:104430:104974 [1] NCCL INFO NVLS comm 0xbd993c0 headRank 0 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-166-125:762817:763374 [2] NCCL INFO Connected all trees
+ip-26-0-166-125:762817:763374 [2] NCCL INFO NVLS comm 0xc51e000 headRank 1 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-166-125:762819:763372 [4] NCCL INFO Connected all trees
+ip-26-0-166-125:762819:763372 [4] NCCL INFO NVLS comm 0xb508540 headRank 2 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-162-46:702884:703470 [4] NCCL INFO Connected NVLS tree
+ip-26-0-162-46:702884:703470 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-162-46:702884:703470 [4] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-162-46:702884:703470 [4] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-162-46:702880:703466 [0] NCCL INFO Connected NVLS tree
+ip-26-0-162-46:702880:703466 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-162-46:702882:703472 [2] NCCL INFO Connected NVLS tree
+ip-26-0-162-46:702882:703472 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-162-46:702882:703472 [2] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-162-46:702882:703472 [2] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-162-46:702880:703466 [0] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-162-46:702880:703466 [0] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-162-46:702886:703469 [6] NCCL INFO Connected NVLS tree
+ip-26-0-162-46:702886:703469 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-162-46:702886:703469 [6] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-162-46:702886:703469 [6] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-166-125:762820:763379 [5] NCCL INFO Connected all trees
+ip-26-0-166-125:762820:763379 [5] NCCL INFO NVLS comm 0xb4271a0 headRank 2 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-166-214:624946:625498 [3] NCCL INFO Connected NVLS tree
+ip-26-0-166-214:624946:625498 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-214:624946:625498 [3] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-166-214:624946:625498 [3] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-166-214:624950:625499 [7] NCCL INFO Connected NVLS tree
+ip-26-0-166-214:624950:625499 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-214:624950:625499 [7] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-166-214:624950:625499 [7] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-166-214:624948:625500 [5] NCCL INFO Connected NVLS tree
+ip-26-0-166-214:624948:625500 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-214:624948:625500 [5] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-166-214:624948:625500 [5] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-166-214:624944:625493 [1] NCCL INFO Connected NVLS tree
+ip-26-0-166-214:624944:625493 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-214:624944:625493 [1] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-166-214:624944:625493 [1] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-162-46:702882:703472 [2] NCCL INFO comm 0xb8bbb80 rank 1 nranks 4 cudaDev 2 nvmlDev 2 busId 75000 commId 0x358519269da0b31b - Init COMPLETE
+ip-26-0-162-46:702886:703469 [6] NCCL INFO comm 0xb144580 rank 3 nranks 4 cudaDev 6 nvmlDev 6 busId b9000 commId 0x358519269da0b31b - Init COMPLETE
+ip-26-0-162-46:702880:703466 [0] NCCL INFO comm 0xafe7d10 rank 0 nranks 4 cudaDev 0 nvmlDev 0 busId 53000 commId 0x358519269da0b31b - Init COMPLETE
+ip-26-0-162-46:702884:703470 [4] NCCL INFO comm 0xa5c1b20 rank 2 nranks 4 cudaDev 4 nvmlDev 4 busId 97000 commId 0x358519269da0b31b - Init COMPLETE
+ip-26-0-166-214:624946:625498 [3] NCCL INFO comm 0xc16d320 rank 1 nranks 4 cudaDev 3 nvmlDev 3 busId 86000 commId 0x6ca81a321acfffee - Init COMPLETE
+ip-26-0-166-214:624944:625493 [1] NCCL INFO comm 0xa637bf0 rank 0 nranks 4 cudaDev 1 nvmlDev 1 busId 64000 commId 0x6ca81a321acfffee - Init COMPLETE
+ip-26-0-166-214:624950:625499 [7] NCCL INFO comm 0xaec1440 rank 3 nranks 4 cudaDev 7 nvmlDev 7 busId ca000 commId 0x6ca81a321acfffee - Init COMPLETE
+ip-26-0-166-214:624948:625500 [5] NCCL INFO comm 0xc1820e0 rank 2 nranks 4 cudaDev 5 nvmlDev 5 busId a8000 commId 0x6ca81a321acfffee - Init COMPLETE
+ip-26-0-162-46:702886:703497 [6] NCCL INFO Using network Libfabric
+ip-26-0-162-46:702887:703503 [7] NCCL INFO Using network Libfabric
+ip-26-0-162-46:702885:703507 [5] NCCL INFO Using network Libfabric
+ip-26-0-162-46:702884:703504 [4] NCCL INFO Using network Libfabric
+ip-26-0-162-46:702883:703505 [3] NCCL INFO Using network Libfabric
+ip-26-0-162-46:702882:703506 [2] NCCL INFO Using network Libfabric
+ip-26-0-167-9:1039544:1040127 [7] NCCL INFO Using network Libfabric
+ip-26-0-167-9:1039540:1040130 [3] NCCL INFO Using network Libfabric
+ip-26-0-167-9:1039541:1040129 [4] NCCL INFO Using network Libfabric
+ip-26-0-167-9:1039542:1040132 [5] NCCL INFO Using network Libfabric
+ip-26-0-162-46:702881:703510 [1] NCCL INFO Using network Libfabric
+ip-26-0-162-46:702880:703511 [0] NCCL INFO Using network Libfabric
+ip-26-0-167-9:1039543:1040128 [6] NCCL INFO Using network Libfabric
+ip-26-0-167-9:1039538:1040134 [1] NCCL INFO Using network Libfabric
+ip-26-0-167-9:1039539:1040131 [2] NCCL INFO Using network Libfabric
+ip-26-0-167-9:1039537:1040133 [0] NCCL INFO Using network Libfabric
+ip-26-0-162-46:702883:703505 [3] NCCL INFO comm 0xa6c6310 rank 0 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0xabd8fe417432c96e - Init START
+ip-26-0-167-9:1039540:1040130 [3] NCCL INFO comm 0xae98460 rank 1 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0xabd8fe417432c96e - Init START
+ip-26-0-162-46:702887:703503 [7] NCCL INFO comm 0xac9ab50 rank 0 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0x263ef92fd3ed29b3 - Init START
+ip-26-0-167-9:1039544:1040127 [7] NCCL INFO comm 0xada3c80 rank 1 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0x263ef92fd3ed29b3 - Init START
+ip-26-0-162-46:702884:703504 [4] NCCL INFO comm 0xb052d40 rank 0 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0x1dbabbc63ab34d4a - Init START
+ip-26-0-162-46:702883:703505 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-167-9:1039541:1040129 [4] NCCL INFO comm 0xbb9cc40 rank 1 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0x1dbabbc63ab34d4a - Init START
+ip-26-0-162-46:702885:703507 [5] NCCL INFO comm 0xacaa7c0 rank 0 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0xd74ffda726ebe35d - Init START
+ip-26-0-162-46:702887:703503 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-162-46:702884:703504 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-167-9:1039540:1040130 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-167-9:1039542:1040132 [5] NCCL INFO comm 0xa55cea0 rank 1 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0xd74ffda726ebe35d - Init START
+ip-26-0-162-46:702886:703497 [6] NCCL INFO comm 0xb18f0d0 rank 0 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0x56e412e8364bae7c - Init START
+ip-26-0-167-9:1039544:1040127 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-167-9:1039543:1040128 [6] NCCL INFO comm 0xb66c200 rank 1 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0x56e412e8364bae7c - Init START
+ip-26-0-162-46:702880:703511 [0] NCCL INFO comm 0xba7afb0 rank 0 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0xb96eb90f733e8807 - Init START
+ip-26-0-167-9:1039541:1040129 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-162-46:702885:703507 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-162-46:702881:703510 [1] NCCL INFO comm 0xb54fb50 rank 0 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0xee6ba7695144767b - Init START
+ip-26-0-162-46:702882:703506 [2] NCCL INFO comm 0xb906940 rank 0 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0xe6215730505ce7cf - Init START
+ip-26-0-162-46:702886:703497 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-167-9:1039537:1040133 [0] NCCL INFO comm 0xb32bc80 rank 1 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0xb96eb90f733e8807 - Init START
+ip-26-0-167-9:1039538:1040134 [1] NCCL INFO comm 0xa7477c0 rank 1 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0xee6ba7695144767b - Init START
+ip-26-0-167-9:1039542:1040132 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-162-46:702880:703511 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/349
+ip-26-0-166-36:104433:104976 [4] NCCL INFO Connected all trees
+ip-26-0-162-46:702881:703510 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-167-9:1039539:1040131 [2] NCCL INFO comm 0xa81e380 rank 1 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0xe6215730505ce7cf - Init START
+ip-26-0-162-46:702882:703506 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-167-9:1039543:1040128 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-167-9:1039537:1040133 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-167-9:1039538:1040134 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-167-9:1039539:1040131 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-36:104433:104976 [4] NCCL INFO NVLS comm 0xc36e970 headRank 2 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-166-36:104432:104980 [3] NCCL INFO Connected all trees
+ip-26-0-166-36:104432:104980 [3] NCCL INFO NVLS comm 0xa91fb80 headRank 1 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-166-125:762817:763374 [2] NCCL INFO Connected NVLS tree
+ip-26-0-166-125:762817:763374 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-125:762819:763372 [4] NCCL INFO Connected NVLS tree
+ip-26-0-166-125:762819:763372 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-125:762819:763372 [4] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-166-125:762819:763372 [4] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-166-125:762817:763374 [2] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-166-125:762817:763374 [2] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-166-125:762821:763373 [6] NCCL INFO Connected NVLS tree
+ip-26-0-166-125:762821:763373 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-125:762821:763373 [6] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-166-125:762821:763373 [6] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-166-36:104434:104977 [5] NCCL INFO Connected all trees
+ip-26-0-166-125:762815:763371 [0] NCCL INFO Connected NVLS tree
+ip-26-0-166-125:762815:763371 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-125:762815:763371 [0] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-166-125:762815:763371 [0] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-166-36:104434:104977 [5] NCCL INFO NVLS comm 0xab8bb00 headRank 2 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-166-125:762821:763373 [6] NCCL INFO comm 0xa9cdcc0 rank 3 nranks 4 cudaDev 6 nvmlDev 6 busId b9000 commId 0x16fbe3f8cceea039 - Init COMPLETE
+ip-26-0-166-125:762817:763374 [2] NCCL INFO comm 0xc51e000 rank 1 nranks 4 cudaDev 2 nvmlDev 2 busId 75000 commId 0x16fbe3f8cceea039 - Init COMPLETE
+ip-26-0-166-125:762819:763372 [4] NCCL INFO comm 0xb508540 rank 2 nranks 4 cudaDev 4 nvmlDev 4 busId 97000 commId 0x16fbe3f8cceea039 - Init COMPLETE
+ip-26-0-166-125:762815:763371 [0] NCCL INFO comm 0xa6f1020 rank 0 nranks 4 cudaDev 0 nvmlDev 0 busId 53000 commId 0x16fbe3f8cceea039 - Init COMPLETE
+ip-26-0-166-36:104431:104979 [2] NCCL INFO Connected all trees
+ip-26-0-166-36:104431:104979 [2] NCCL INFO NVLS comm 0xb9278a0 headRank 1 nHeads 4 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 805306368
+ip-26-0-166-125:762816:763376 [1] NCCL INFO Connected NVLS tree
+ip-26-0-166-125:762816:763376 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-125:762820:763379 [5] NCCL INFO Connected NVLS tree
+ip-26-0-166-125:762820:763379 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-125:762820:763379 [5] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-166-125:762820:763379 [5] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-166-125:762816:763376 [1] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-166-125:762816:763376 [1] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-166-125:762822:763377 [7] NCCL INFO Connected NVLS tree
+ip-26-0-166-125:762822:763377 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-125:762822:763377 [7] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-166-125:762822:763377 [7] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-166-125:762818:763378 [3] NCCL INFO Connected NVLS tree
+ip-26-0-166-125:762818:763378 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-125:762818:763378 [3] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-166-125:762818:763378 [3] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-166-36:104432:104980 [3] NCCL INFO Connected NVLS tree
+ip-26-0-166-36:104432:104980 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-36:104432:104980 [3] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-166-36:104432:104980 [3] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-166-36:104436:104978 [7] NCCL INFO Connected NVLS tree
+ip-26-0-166-36:104436:104978 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-36:104436:104978 [7] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-166-36:104436:104978 [7] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-166-36:104434:104977 [5] NCCL INFO Connected NVLS tree
+ip-26-0-166-36:104434:104977 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-36:104434:104977 [5] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-166-36:104434:104977 [5] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-166-125:762816:763376 [1] NCCL INFO comm 0xb71fe00 rank 0 nranks 4 cudaDev 1 nvmlDev 1 busId 64000 commId 0x2244107d9b6a67ed - Init COMPLETE
+ip-26-0-166-125:762820:763379 [5] NCCL INFO comm 0xb4271a0 rank 2 nranks 4 cudaDev 5 nvmlDev 5 busId a8000 commId 0x2244107d9b6a67ed - Init COMPLETE
+ip-26-0-166-125:762822:763377 [7] NCCL INFO comm 0xc109000 rank 3 nranks 4 cudaDev 7 nvmlDev 7 busId ca000 commId 0x2244107d9b6a67ed - Init COMPLETE
+ip-26-0-166-125:762818:763378 [3] NCCL INFO comm 0xa85c340 rank 1 nranks 4 cudaDev 3 nvmlDev 3 busId 86000 commId 0x2244107d9b6a67ed - Init COMPLETE
+ip-26-0-166-36:104430:104974 [1] NCCL INFO Connected NVLS tree
+ip-26-0-166-36:104430:104974 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-36:104430:104974 [1] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-166-36:104430:104974 [1] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-166-36:104432:104980 [3] NCCL INFO comm 0xa91fb80 rank 1 nranks 4 cudaDev 3 nvmlDev 3 busId 86000 commId 0x9a7cb67b8659c59e - Init COMPLETE
+ip-26-0-166-36:104436:104978 [7] NCCL INFO comm 0xb855c60 rank 3 nranks 4 cudaDev 7 nvmlDev 7 busId ca000 commId 0x9a7cb67b8659c59e - Init COMPLETE
+ip-26-0-166-36:104434:104977 [5] NCCL INFO comm 0xab8bb00 rank 2 nranks 4 cudaDev 5 nvmlDev 5 busId a8000 commId 0x9a7cb67b8659c59e - Init COMPLETE
+ip-26-0-166-36:104430:104974 [1] NCCL INFO comm 0xbd993c0 rank 0 nranks 4 cudaDev 1 nvmlDev 1 busId 64000 commId 0x9a7cb67b8659c59e - Init COMPLETE
+ip-26-0-166-36:104431:104979 [2] NCCL INFO Connected NVLS tree
+ip-26-0-166-36:104431:104979 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-36:104431:104979 [2] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-166-36:104431:104979 [2] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-166-36:104433:104976 [4] NCCL INFO Connected NVLS tree
+ip-26-0-166-36:104433:104976 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-36:104433:104976 [4] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-166-36:104433:104976 [4] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-166-36:104435:104975 [6] NCCL INFO Connected NVLS tree
+ip-26-0-166-36:104435:104975 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-36:104435:104975 [6] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-166-36:104435:104975 [6] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-166-36:104429:104973 [0] NCCL INFO Connected NVLS tree
+ip-26-0-166-36:104429:104973 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-36:104429:104973 [0] NCCL INFO threadThresholds 8/8/64 | 32/8/64 | 512 | 512
+ip-26-0-166-36:104429:104973 [0] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-166-36:104431:104979 [2] NCCL INFO comm 0xb9278a0 rank 1 nranks 4 cudaDev 2 nvmlDev 2 busId 75000 commId 0xb03b08d1ced4579 - Init COMPLETE
+ip-26-0-166-36:104435:104975 [6] NCCL INFO comm 0xaee4280 rank 3 nranks 4 cudaDev 6 nvmlDev 6 busId b9000 commId 0xb03b08d1ced4579 - Init COMPLETE
+ip-26-0-166-36:104433:104976 [4] NCCL INFO comm 0xc36e970 rank 2 nranks 4 cudaDev 4 nvmlDev 4 busId 97000 commId 0xb03b08d1ced4579 - Init COMPLETE
+ip-26-0-166-36:104429:104973 [0] NCCL INFO comm 0xc5849c0 rank 0 nranks 4 cudaDev 0 nvmlDev 0 busId 53000 commId 0xb03b08d1ced4579 - Init COMPLETE
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+ip-26-0-162-46:702881:703510 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+ip-26-0-162-46:702886:703497 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+ip-26-0-167-9:1039539:1040131 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+ip-26-0-167-9:1039542:1040132 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-162-46:702885:703507 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-162-46:702885:703507 [5] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-162-46:702885:703507 [5] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-162-46:702885:703507 [5] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-162-46:702885:703507 [5] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-162-46:702885:703507 [5] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-162-46:702885:703507 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-167-9:1039542:1040132 [5] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-167-9:1039542:1040132 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-167-9:1039537:1040133 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-167-9:1039540:1040130 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-162-46:702880:703511 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-162-46:702880:703511 [0] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-162-46:702880:703511 [0] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-162-46:702880:703511 [0] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-162-46:702880:703511 [0] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-162-46:702880:703511 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-162-46:702880:703511 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-167-9:1039537:1040133 [0] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-167-9:1039537:1040133 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-162-46:702883:703505 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-167-9:1039544:1040127 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-162-46:702883:703505 [3] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-162-46:702883:703505 [3] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-162-46:702883:703505 [3] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-162-46:702883:703505 [3] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-162-46:702883:703505 [3] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-162-46:702883:703505 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-167-9:1039540:1040130 [3] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-167-9:1039540:1040130 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-167-9:1039541:1040129 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-162-46:702884:703504 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-162-46:702884:703504 [4] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-162-46:702884:703504 [4] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-162-46:702884:703504 [4] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-162-46:702884:703504 [4] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-162-46:702884:703504 [4] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-162-46:702884:703504 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-167-9:1039541:1040129 [4] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-167-9:1039541:1040129 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-167-9:1039538:1040134 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-167-9:1039543:1040128 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-167-9:1039538:1040134 [1] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-167-9:1039538:1040134 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-162-46:702881:703510 [1] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-162-46:702881:703510 [1] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-162-46:702881:703510 [1] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-162-46:702881:703510 [1] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-162-46:702881:703510 [1] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-162-46:702881:703510 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-167-9:1039543:1040128 [6] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-167-9:1039543:1040128 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-162-46:702886:703497 [6] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-162-46:702886:703497 [6] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-162-46:702886:703497 [6] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-162-46:702886:703497 [6] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-162-46:702886:703497 [6] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-162-46:702886:703497 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-162-46:702882:703506 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-162-46:702887:703503 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-162-46:702882:703506 [2] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-162-46:702882:703506 [2] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-162-46:702882:703506 [2] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-162-46:702882:703506 [2] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-162-46:702882:703506 [2] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-162-46:702882:703506 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-167-9:1039539:1040131 [2] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-167-9:1039539:1040131 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-162-46:702887:703503 [7] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-162-46:702887:703503 [7] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-162-46:702887:703503 [7] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-162-46:702887:703503 [7] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-162-46:702887:703503 [7] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-162-46:702887:703503 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-167-9:1039544:1040127 [7] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-167-9:1039544:1040127 [7] NCCL INFO P2P Chunksize set to 131072
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+ip-26-0-162-46:702885:703507 [5] NCCL INFO Channel 00/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-46:702885:703507 [5] NCCL INFO Channel 01/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-46:702885:703507 [5] NCCL INFO Channel 02/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-46:702885:703507 [5] NCCL INFO Channel 03/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-46:702885:703507 [5] NCCL INFO Channel 00/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-46:702885:703507 [5] NCCL INFO Channel 01/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-46:702885:703507 [5] NCCL INFO Channel 02/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-46:702885:703507 [5] NCCL INFO Channel 03/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-167-9:1039542:1040132 [5] NCCL INFO Channel 00/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-167-9:1039542:1040132 [5] NCCL INFO Channel 01/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-167-9:1039542:1040132 [5] NCCL INFO Channel 02/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-167-9:1039542:1040132 [5] NCCL INFO Channel 03/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-167-9:1039542:1040132 [5] NCCL INFO Channel 00/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-167-9:1039542:1040132 [5] NCCL INFO Channel 01/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-167-9:1039542:1040132 [5] NCCL INFO Channel 02/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-167-9:1039542:1040132 [5] NCCL INFO Channel 03/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-167-9:1039537:1040133 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-167-9:1039537:1040133 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-167-9:1039537:1040133 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-167-9:1039537:1040133 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-167-9:1039537:1040133 [0] NCCL INFO Channel 00/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-167-9:1039537:1040133 [0] NCCL INFO Channel 01/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-167-9:1039537:1040133 [0] NCCL INFO Channel 02/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-167-9:1039537:1040133 [0] NCCL INFO Channel 03/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-167-9:1039540:1040130 [3] NCCL INFO Channel 00/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-167-9:1039540:1040130 [3] NCCL INFO Channel 01/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-167-9:1039540:1040130 [3] NCCL INFO Channel 02/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+ip-26-0-167-9:1039540:1040130 [3] NCCL INFO Channel 03/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-167-9:1039540:1040130 [3] NCCL INFO Channel 00/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-46:702880:703511 [0] NCCL INFO Channel 00/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-167-9:1039540:1040130 [3] NCCL INFO Channel 01/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-46:702880:703511 [0] NCCL INFO Channel 01/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-46:702880:703511 [0] NCCL INFO Channel 02/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-167-9:1039540:1040130 [3] NCCL INFO Channel 02/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-46:702880:703511 [0] NCCL INFO Channel 03/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-167-9:1039540:1040130 [3] NCCL INFO Channel 03/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-46:702883:703505 [3] NCCL INFO Channel 00/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-46:702880:703511 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-46:702880:703511 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-46:702883:703505 [3] NCCL INFO Channel 01/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-46:702880:703511 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-46:702883:703505 [3] NCCL INFO Channel 02/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-46:702880:703511 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-46:702883:703505 [3] NCCL INFO Channel 03/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-46:702883:703505 [3] NCCL INFO Channel 00/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-46:702883:703505 [3] NCCL INFO Channel 01/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-46:702883:703505 [3] NCCL INFO Channel 02/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-46:702883:703505 [3] NCCL INFO Channel 03/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+ip-26-0-167-9:1039541:1040129 [4] NCCL INFO Channel 00/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-167-9:1039541:1040129 [4] NCCL INFO Channel 01/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-167-9:1039541:1040129 [4] NCCL INFO Channel 02/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-167-9:1039541:1040129 [4] NCCL INFO Channel 03/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-167-9:1039541:1040129 [4] NCCL INFO Channel 00/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-167-9:1039541:1040129 [4] NCCL INFO Channel 01/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-46:702884:703504 [4] NCCL INFO Channel 00/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-167-9:1039541:1040129 [4] NCCL INFO Channel 02/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-46:702884:703504 [4] NCCL INFO Channel 01/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-167-9:1039541:1040129 [4] NCCL INFO Channel 03/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-46:702884:703504 [4] NCCL INFO Channel 02/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-46:702884:703504 [4] NCCL INFO Channel 03/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-46:702884:703504 [4] NCCL INFO Channel 00/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-46:702884:703504 [4] NCCL INFO Channel 01/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-46:702884:703504 [4] NCCL INFO Channel 02/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-46:702884:703504 [4] NCCL INFO Channel 03/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+ip-26-0-167-9:1039538:1040134 [1] NCCL INFO Channel 00/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-167-9:1039538:1040134 [1] NCCL INFO Channel 01/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-167-9:1039538:1040134 [1] NCCL INFO Channel 02/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-167-9:1039538:1040134 [1] NCCL INFO Channel 03/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-46:702881:703510 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-167-9:1039538:1040134 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+ip-26-0-167-9:1039538:1040134 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-46:702881:703510 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-46:702881:703510 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-167-9:1039538:1040134 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-46:702881:703510 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-167-9:1039538:1040134 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-46:702881:703510 [1] NCCL INFO Channel 00/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-46:702881:703510 [1] NCCL INFO Channel 01/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-46:702881:703510 [1] NCCL INFO Channel 02/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-46:702881:703510 [1] NCCL INFO Channel 03/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-46:702882:703506 [2] NCCL INFO Channel 00/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-46:702882:703506 [2] NCCL INFO Channel 01/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-46:702882:703506 [2] NCCL INFO Channel 02/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-46:702882:703506 [2] NCCL INFO Channel 03/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-46:702882:703506 [2] NCCL INFO Channel 00/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-46:702882:703506 [2] NCCL INFO Channel 01/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-46:702882:703506 [2] NCCL INFO Channel 02/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-46:702882:703506 [2] NCCL INFO Channel 03/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-46:702886:703497 [6] NCCL INFO Channel 00/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-46:702886:703497 [6] NCCL INFO Channel 01/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-46:702886:703497 [6] NCCL INFO Channel 02/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-46:702886:703497 [6] NCCL INFO Channel 03/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-46:702886:703497 [6] NCCL INFO Channel 00/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-167-9:1039539:1040131 [2] NCCL INFO Channel 00/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-46:702886:703497 [6] NCCL INFO Channel 01/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-167-9:1039543:1040128 [6] NCCL INFO Channel 00/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-46:702886:703497 [6] NCCL INFO Channel 02/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-167-9:1039543:1040128 [6] NCCL INFO Channel 01/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-167-9:1039539:1040131 [2] NCCL INFO Channel 01/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-46:702886:703497 [6] NCCL INFO Channel 03/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-167-9:1039543:1040128 [6] NCCL INFO Channel 02/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-167-9:1039539:1040131 [2] NCCL INFO Channel 02/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-167-9:1039543:1040128 [6] NCCL INFO Channel 03/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-167-9:1039539:1040131 [2] NCCL INFO Channel 03/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-167-9:1039539:1040131 [2] NCCL INFO Channel 00/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-167-9:1039543:1040128 [6] NCCL INFO Channel 00/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-167-9:1039539:1040131 [2] NCCL INFO Channel 01/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-167-9:1039543:1040128 [6] NCCL INFO Channel 01/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-167-9:1039539:1040131 [2] NCCL INFO Channel 02/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-167-9:1039543:1040128 [6] NCCL INFO Channel 02/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-167-9:1039539:1040131 [2] NCCL INFO Channel 03/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-167-9:1039543:1040128 [6] NCCL INFO Channel 03/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-46:702887:703503 [7] NCCL INFO Channel 00/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-46:702887:703503 [7] NCCL INFO Channel 01/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-46:702887:703503 [7] NCCL INFO Channel 02/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-46:702887:703503 [7] NCCL INFO Channel 03/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-46:702887:703503 [7] NCCL INFO Channel 00/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-46:702887:703503 [7] NCCL INFO Channel 01/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-46:702887:703503 [7] NCCL INFO Channel 02/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-46:702887:703503 [7] NCCL INFO Channel 03/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-167-9:1039544:1040127 [7] NCCL INFO Channel 00/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-167-9:1039544:1040127 [7] NCCL INFO Channel 01/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-167-9:1039544:1040127 [7] NCCL INFO Channel 02/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-167-9:1039544:1040127 [7] NCCL INFO Channel 03/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-167-9:1039544:1040127 [7] NCCL INFO Channel 00/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-167-9:1039544:1040127 [7] NCCL INFO Channel 01/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-167-9:1039544:1040127 [7] NCCL INFO Channel 02/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-167-9:1039544:1040127 [7] NCCL INFO Channel 03/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+ip-26-0-167-9:1039542:1040132 [5] NCCL INFO Connected all rings
+ip-26-0-167-9:1039542:1040132 [5] NCCL INFO Connected all trees
+ip-26-0-167-9:1039542:1040132 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-167-9:1039542:1040132 [5] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-167-9:1039542:1040132 [5] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-167-9:1039542:1040132 [5] NCCL INFO comm 0xa55cea0 rank 1 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0xd74ffda726ebe35d - Init COMPLETE
+ip-26-0-162-46:702885:703507 [5] NCCL INFO Connected all rings
+ip-26-0-162-46:702885:703507 [5] NCCL INFO Connected all trees
+ip-26-0-162-46:702885:703507 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-162-46:702885:703507 [5] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-162-46:702885:703507 [5] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-162-46:702885:703507 [5] NCCL INFO comm 0xacaa7c0 rank 0 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0xd74ffda726ebe35d - Init COMPLETE
+ip-26-0-162-46:702883:703505 [3] NCCL INFO Connected all rings
+ip-26-0-162-46:702883:703505 [3] NCCL INFO Connected all trees
+ip-26-0-162-46:702883:703505 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-162-46:702883:703505 [3] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-162-46:702883:703505 [3] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-167-9:1039540:1040130 [3] NCCL INFO Connected all rings
+ip-26-0-167-9:1039540:1040130 [3] NCCL INFO Connected all trees
+ip-26-0-167-9:1039540:1040130 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-167-9:1039540:1040130 [3] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-167-9:1039540:1040130 [3] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-167-9:1039541:1040129 [4] NCCL INFO Connected all rings
+ip-26-0-167-9:1039541:1040129 [4] NCCL INFO Connected all trees
+ip-26-0-167-9:1039541:1040129 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-167-9:1039541:1040129 [4] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-167-9:1039541:1040129 [4] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-162-46:702884:703504 [4] NCCL INFO Connected all rings
+ip-26-0-162-46:702884:703504 [4] NCCL INFO Connected all trees
+ip-26-0-162-46:702884:703504 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-162-46:702884:703504 [4] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-162-46:702884:703504 [4] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-162-46:702887:703503 [7] NCCL INFO Connected all rings
+ip-26-0-162-46:702887:703503 [7] NCCL INFO Connected all trees
+ip-26-0-162-46:702887:703503 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-162-46:702887:703503 [7] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-162-46:702887:703503 [7] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-167-9:1039537:1040133 [0] NCCL INFO Connected all rings
+ip-26-0-167-9:1039537:1040133 [0] NCCL INFO Connected all trees
+ip-26-0-167-9:1039537:1040133 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-167-9:1039537:1040133 [0] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-167-9:1039537:1040133 [0] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-167-9:1039544:1040127 [7] NCCL INFO Connected all rings
+ip-26-0-167-9:1039544:1040127 [7] NCCL INFO Connected all trees
+ip-26-0-167-9:1039544:1040127 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-167-9:1039544:1040127 [7] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-167-9:1039544:1040127 [7] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-162-46:702880:703511 [0] NCCL INFO Connected all rings
+ip-26-0-162-46:702880:703511 [0] NCCL INFO Connected all trees
+ip-26-0-162-46:702880:703511 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-162-46:702880:703511 [0] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-162-46:702880:703511 [0] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-162-46:702886:703497 [6] NCCL INFO Connected all rings
+ip-26-0-162-46:702886:703497 [6] NCCL INFO Connected all trees
+ip-26-0-162-46:702886:703497 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-162-46:702886:703497 [6] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-162-46:702886:703497 [6] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-162-46:702883:703505 [3] NCCL INFO comm 0xa6c6310 rank 0 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0xabd8fe417432c96e - Init COMPLETE
+ip-26-0-167-9:1039543:1040128 [6] NCCL INFO Connected all rings
+ip-26-0-167-9:1039543:1040128 [6] NCCL INFO Connected all trees
+ip-26-0-167-9:1039543:1040128 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-167-9:1039543:1040128 [6] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-167-9:1039543:1040128 [6] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-167-9:1039538:1040134 [1] NCCL INFO Connected all rings
+ip-26-0-167-9:1039538:1040134 [1] NCCL INFO Connected all trees
+ip-26-0-167-9:1039538:1040134 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-167-9:1039538:1040134 [1] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-167-9:1039538:1040134 [1] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-167-9:1039539:1040131 [2] NCCL INFO Connected all rings
+ip-26-0-167-9:1039539:1040131 [2] NCCL INFO Connected all trees
+ip-26-0-167-9:1039539:1040131 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-162-46:702882:703506 [2] NCCL INFO Connected all rings
+ip-26-0-162-46:702882:703506 [2] NCCL INFO Connected all trees
+ip-26-0-162-46:702882:703506 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-167-9:1039539:1040131 [2] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-167-9:1039539:1040131 [2] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-162-46:702882:703506 [2] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-162-46:702882:703506 [2] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-167-9:1039540:1040130 [3] NCCL INFO comm 0xae98460 rank 1 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0xabd8fe417432c96e - Init COMPLETE
+ip-26-0-162-46:702881:703510 [1] NCCL INFO Connected all rings
+ip-26-0-162-46:702881:703510 [1] NCCL INFO Connected all trees
+ip-26-0-162-46:702881:703510 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-162-46:702881:703510 [1] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-162-46:702881:703510 [1] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-167-9:1039541:1040129 [4] NCCL INFO comm 0xbb9cc40 rank 1 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0x1dbabbc63ab34d4a - Init COMPLETE
+ip-26-0-162-46:702884:703504 [4] NCCL INFO comm 0xb052d40 rank 0 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0x1dbabbc63ab34d4a - Init COMPLETE
+ip-26-0-162-46:702887:703503 [7] NCCL INFO comm 0xac9ab50 rank 0 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0x263ef92fd3ed29b3 - Init COMPLETE
+ip-26-0-167-9:1039537:1040133 [0] NCCL INFO comm 0xb32bc80 rank 1 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0xb96eb90f733e8807 - Init COMPLETE
+ip-26-0-167-9:1039544:1040127 [7] NCCL INFO comm 0xada3c80 rank 1 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0x263ef92fd3ed29b3 - Init COMPLETE
+ip-26-0-162-46:702880:703511 [0] NCCL INFO comm 0xba7afb0 rank 0 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0xb96eb90f733e8807 - Init COMPLETE
+ip-26-0-162-46:702886:703497 [6] NCCL INFO comm 0xb18f0d0 rank 0 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0x56e412e8364bae7c - Init COMPLETE
+ip-26-0-167-9:1039543:1040128 [6] NCCL INFO comm 0xb66c200 rank 1 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0x56e412e8364bae7c - Init COMPLETE
+01/07/2025 03:50:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]: [Optimizer Building] Using LearningRateForSP as learning rate
+ip-26-0-167-9:1039538:1040134 [1] NCCL INFO comm 0xa7477c0 rank 1 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0xee6ba7695144767b - Init COMPLETE
+ip-26-0-167-9:1039539:1040131 [2] NCCL INFO comm 0xa81e380 rank 1 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0xe6215730505ce7cf - Init COMPLETE
+ip-26-0-162-46:702882:703506 [2] NCCL INFO comm 0xb906940 rank 0 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0xe6215730505ce7cf - Init COMPLETE
+01/07/2025 03:50:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]: [ZeRO sharding] Size of optimizer params per rank:
+01/07/2025 03:50:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]: [ZeRO sharding] DP Rank 0 has 58.7M out of 235M (25.00%) params' optimizer states
+01/07/2025 03:50:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]: [ZeRO sharding] DP Rank 1 has 58.7M out of 235M (25.00%) params' optimizer states
+01/07/2025 03:50:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]: [ZeRO sharding] DP Rank 2 has 58.7M out of 235M (25.00%) params' optimizer states
+01/07/2025 03:50:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]: [ZeRO sharding] DP Rank 3 has 58.7M out of 235M (25.00%) params' optimizer states
+ip-26-0-162-46:702881:703510 [1] NCCL INFO comm 0xb54fb50 rank 0 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0xee6ba7695144767b - Init COMPLETE
+01/07/2025 03:50:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]: [Training Plan] Stage Stable Training Stage has 99 remaining training steps and has consumed 0 samples
+01/07/2025 03:50:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]: Using dummy data generator
+01/07/2025 03:50:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]: [Training Plan] There are 1 training stages 
+01/07/2025 03:50:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]: [Stage Stable Training Stage] start from step 1 
+01/07/2025 03:50:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]: 
+01/07/2025 03:50:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]: [Start training] datetime: 2025-01-07 03:50:05.642664 | mbs: 2 | grad_accum: 32 | global_batch_size: 256 | sequence_length: 4096 | train_steps: 100 | start_iteration_step: 0 | consumed_train_samples: 0
+01/07/2025 03:50:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]: Resuming training from stage Stable Training Stage, it has trained for 0 samples and has 99 remaining train steps
+01/07/2025 03:50:05 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]:  Memory usage: 1568.09MiB. Peak allocated 5408.00MiB. Peak reserved: 11554.00MiB
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+ip-26-0-162-46:702884:703534 [4] NCCL INFO Using network Libfabric
+ip-26-0-162-46:702881:703538 [1] NCCL INFO Using network Libfabric
+ip-26-0-162-46:702885:703536 [5] NCCL INFO Using network Libfabric
+ip-26-0-162-46:702880:703539 [0] NCCL INFO Using network Libfabric
+ip-26-0-165-213:9630:10205 [5] NCCL INFO Using network Libfabric
+ip-26-0-165-213:9625:10208 [0] NCCL INFO Using network Libfabric
+ip-26-0-165-213:9629:10206 [4] NCCL INFO Using network Libfabric
+ip-26-0-165-213:9626:10207 [1] NCCL INFO Using network Libfabric
+ip-26-0-165-213:9626:10207 [1] NCCL INFO comm 0xb9403c0 rank 1 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0x5d4e907b2e49529b - Init START
+ip-26-0-162-46:702881:703538 [1] NCCL INFO comm 0x1bf45300 rank 0 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0x5d4e907b2e49529b - Init START
+ip-26-0-162-46:702880:703539 [0] NCCL INFO comm 0x1c472390 rank 0 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0x38cb7e7c3e6712ad - Init START
+ip-26-0-162-46:702884:703534 [4] NCCL INFO comm 0x1ba54ce0 rank 0 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0xf7a17e9c5701b2a3 - Init START
+ip-26-0-162-46:702885:703536 [5] NCCL INFO comm 0x1b6a2030 rank 0 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0x9e98cc82a4e3e115 - Init START
+ip-26-0-165-213:9629:10206 [4] NCCL INFO comm 0xac16920 rank 1 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0xf7a17e9c5701b2a3 - Init START
+ip-26-0-165-213:9630:10205 [5] NCCL INFO comm 0xaa92090 rank 1 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0x9e98cc82a4e3e115 - Init START
+ip-26-0-162-46:702881:703538 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-162-46:702884:703534 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-165-213:9625:10208 [0] NCCL INFO comm 0xb1a2e70 rank 1 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0x38cb7e7c3e6712ad - Init START
+ip-26-0-162-46:702885:703536 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-162-46:702880:703539 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/349
+ip-26-0-165-213:9626:10207 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-165-213:9625:10208 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-165-213:9630:10205 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-165-213:9629:10206 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+ip-26-0-162-46:702883:703543 [3] NCCL INFO Using network Libfabric
+ip-26-0-162-46:702882:703542 [2] NCCL INFO Using network Libfabric
+ip-26-0-165-213:9627:10209 [2] NCCL INFO Using network Libfabric
+ip-26-0-165-213:9628:10210 [3] NCCL INFO Using network Libfabric
+ip-26-0-162-46:702882:703542 [2] NCCL INFO comm 0x1c2fd420 rank 0 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0xf9e132832d0ec61b - Init START
+ip-26-0-165-213:9627:10209 [2] NCCL INFO comm 0xac1c710 rank 1 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0xf9e132832d0ec61b - Init START
+ip-26-0-162-46:702883:703543 [3] NCCL INFO comm 0x1b0bcc30 rank 0 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0x53e8c79cced5760e - Init START
+ip-26-0-165-213:9628:10210 [3] NCCL INFO comm 0xb88fbd0 rank 1 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0x53e8c79cced5760e - Init START
+ip-26-0-162-46:702882:703542 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-165-213:9627:10209 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-162-46:702883:703543 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-165-213:9628:10210 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+ip-26-0-162-46:702886:703545 [6] NCCL INFO Using network Libfabric
+ip-26-0-162-46:702887:703547 [7] NCCL INFO Using network Libfabric
+ip-26-0-165-213:9632:10212 [7] NCCL INFO Using network Libfabric
+ip-26-0-165-213:9631:10211 [6] NCCL INFO Using network Libfabric
+ip-26-0-162-46:702886:703545 [6] NCCL INFO comm 0x1bb86170 rank 0 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0xba6c9e2b4095f929 - Init START
+ip-26-0-162-46:702887:703547 [7] NCCL INFO comm 0x1b690380 rank 0 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0xf82b03744fa9a6e6 - Init START
+ip-26-0-165-213:9631:10211 [6] NCCL INFO comm 0xbdc7590 rank 1 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0xba6c9e2b4095f929 - Init START
+ip-26-0-162-46:702886:703545 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-165-213:9632:10212 [7] NCCL INFO comm 0xafebf10 rank 1 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0xf82b03744fa9a6e6 - Init START
+ip-26-0-162-46:702887:703547 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-165-213:9631:10211 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-165-213:9632:10212 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-165-213:9626:10207 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-165-213:9629:10206 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-162-46:702880:703539 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-162-46:702885:703536 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-165-213:9625:10208 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-165-213:9625:10208 [0] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-165-213:9625:10208 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-162-46:702880:703539 [0] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-162-46:702880:703539 [0] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-162-46:702880:703539 [0] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-162-46:702880:703539 [0] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-162-46:702880:703539 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-162-46:702880:703539 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-162-46:702884:703534 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-162-46:702884:703534 [4] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-162-46:702884:703534 [4] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-162-46:702884:703534 [4] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-162-46:702884:703534 [4] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-162-46:702884:703534 [4] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-162-46:702884:703534 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-213:9629:10206 [4] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-165-213:9629:10206 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-213:9630:10205 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-165-213:9630:10205 [5] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-165-213:9630:10205 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-162-46:702885:703536 [5] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-162-46:702885:703536 [5] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-162-46:702885:703536 [5] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-162-46:702885:703536 [5] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-162-46:702885:703536 [5] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-162-46:702885:703536 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-162-46:702881:703538 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-162-46:702881:703538 [1] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-162-46:702881:703538 [1] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-162-46:702881:703538 [1] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-162-46:702881:703538 [1] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-162-46:702881:703538 [1] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-162-46:702881:703538 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-213:9626:10207 [1] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-165-213:9626:10207 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-213:9625:10208 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9625:10208 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9625:10208 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9625:10208 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9625:10208 [0] NCCL INFO Channel 00/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9625:10208 [0] NCCL INFO Channel 01/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9625:10208 [0] NCCL INFO Channel 02/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9625:10208 [0] NCCL INFO Channel 03/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-46:702880:703539 [0] NCCL INFO Channel 00/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-46:702880:703539 [0] NCCL INFO Channel 01/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-46:702880:703539 [0] NCCL INFO Channel 02/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-46:702880:703539 [0] NCCL INFO Channel 03/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-46:702880:703539 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-46:702880:703539 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-46:702880:703539 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-46:702880:703539 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9627:10209 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-162-46:702886:703545 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-162-46:702884:703534 [4] NCCL INFO Channel 00/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-46:702884:703534 [4] NCCL INFO Channel 01/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-46:702884:703534 [4] NCCL INFO Channel 02/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-46:702884:703534 [4] NCCL INFO Channel 03/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-46:702884:703534 [4] NCCL INFO Channel 00/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-46:702884:703534 [4] NCCL INFO Channel 01/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-46:702884:703534 [4] NCCL INFO Channel 02/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-46:702884:703534 [4] NCCL INFO Channel 03/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9629:10206 [4] NCCL INFO Channel 00/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9629:10206 [4] NCCL INFO Channel 01/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9629:10206 [4] NCCL INFO Channel 02/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9629:10206 [4] NCCL INFO Channel 03/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9629:10206 [4] NCCL INFO Channel 00/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9629:10206 [4] NCCL INFO Channel 01/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9629:10206 [4] NCCL INFO Channel 02/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9629:10206 [4] NCCL INFO Channel 03/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-46:702885:703536 [5] NCCL INFO Channel 00/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-46:702885:703536 [5] NCCL INFO Channel 01/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-46:702885:703536 [5] NCCL INFO Channel 02/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-46:702885:703536 [5] NCCL INFO Channel 03/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-46:702885:703536 [5] NCCL INFO Channel 00/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-46:702885:703536 [5] NCCL INFO Channel 01/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-46:702885:703536 [5] NCCL INFO Channel 02/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9630:10205 [5] NCCL INFO Channel 00/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-46:702885:703536 [5] NCCL INFO Channel 03/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9630:10205 [5] NCCL INFO Channel 01/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9630:10205 [5] NCCL INFO Channel 02/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9630:10205 [5] NCCL INFO Channel 03/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9630:10205 [5] NCCL INFO Channel 00/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9630:10205 [5] NCCL INFO Channel 01/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9630:10205 [5] NCCL INFO Channel 02/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9630:10205 [5] NCCL INFO Channel 03/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9628:10210 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-162-46:702887:703547 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-162-46:702882:703542 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+ip-26-0-162-46:702882:703542 [2] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-162-46:702882:703542 [2] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-162-46:702882:703542 [2] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-162-46:702882:703542 [2] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-162-46:702882:703542 [2] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-162-46:702882:703542 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-213:9627:10209 [2] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-165-213:9627:10209 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-162-46:702883:703543 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-162-46:702883:703543 [3] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-162-46:702883:703543 [3] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-162-46:702883:703543 [3] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-162-46:702883:703543 [3] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-162-46:702883:703543 [3] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-162-46:702883:703543 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-213:9628:10210 [3] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-165-213:9628:10210 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-213:9632:10212 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-165-213:9632:10212 [7] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-165-213:9632:10212 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-162-46:702887:703547 [7] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-162-46:702887:703547 [7] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-162-46:702887:703547 [7] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-162-46:702887:703547 [7] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-162-46:702887:703547 [7] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-162-46:702887:703547 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-213:9626:10207 [1] NCCL INFO Channel 00/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9626:10207 [1] NCCL INFO Channel 01/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9626:10207 [1] NCCL INFO Channel 02/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9626:10207 [1] NCCL INFO Channel 03/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9626:10207 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9626:10207 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9626:10207 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9626:10207 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9631:10211 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-162-46:702881:703538 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-46:702881:703538 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-46:702881:703538 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-46:702881:703538 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-46:702881:703538 [1] NCCL INFO Channel 00/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9631:10211 [6] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-165-213:9631:10211 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-162-46:702886:703545 [6] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-162-46:702881:703538 [1] NCCL INFO Channel 01/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-46:702886:703545 [6] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-162-46:702886:703545 [6] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-162-46:702886:703545 [6] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-162-46:702886:703545 [6] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-162-46:702886:703545 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-162-46:702881:703538 [1] NCCL INFO Channel 02/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-46:702881:703538 [1] NCCL INFO Channel 03/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-46:702882:703542 [2] NCCL INFO Channel 00/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-46:702882:703542 [2] NCCL INFO Channel 01/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-46:702882:703542 [2] NCCL INFO Channel 02/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-46:702882:703542 [2] NCCL INFO Channel 03/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-46:702882:703542 [2] NCCL INFO Channel 00/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-46:702882:703542 [2] NCCL INFO Channel 01/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-46:702882:703542 [2] NCCL INFO Channel 02/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-46:702882:703542 [2] NCCL INFO Channel 03/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9627:10209 [2] NCCL INFO Channel 00/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9627:10209 [2] NCCL INFO Channel 01/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9627:10209 [2] NCCL INFO Channel 02/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9627:10209 [2] NCCL INFO Channel 03/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9627:10209 [2] NCCL INFO Channel 00/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9627:10209 [2] NCCL INFO Channel 01/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9627:10209 [2] NCCL INFO Channel 02/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9627:10209 [2] NCCL INFO Channel 03/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-46:702887:703547 [7] NCCL INFO Channel 00/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-46:702883:703543 [3] NCCL INFO Channel 00/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-46:702887:703547 [7] NCCL INFO Channel 01/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-46:702883:703543 [3] NCCL INFO Channel 01/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-46:702883:703543 [3] NCCL INFO Channel 02/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-46:702887:703547 [7] NCCL INFO Channel 02/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-46:702883:703543 [3] NCCL INFO Channel 03/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-46:702887:703547 [7] NCCL INFO Channel 03/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-46:702883:703543 [3] NCCL INFO Channel 00/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-46:702887:703547 [7] NCCL INFO Channel 00/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-46:702883:703543 [3] NCCL INFO Channel 01/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-46:702887:703547 [7] NCCL INFO Channel 01/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9628:10210 [3] NCCL INFO Channel 00/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-46:702883:703543 [3] NCCL INFO Channel 02/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-46:702887:703547 [7] NCCL INFO Channel 02/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9628:10210 [3] NCCL INFO Channel 01/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-46:702883:703543 [3] NCCL INFO Channel 03/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-46:702887:703547 [7] NCCL INFO Channel 03/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9628:10210 [3] NCCL INFO Channel 02/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9628:10210 [3] NCCL INFO Channel 03/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9628:10210 [3] NCCL INFO Channel 00/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9628:10210 [3] NCCL INFO Channel 01/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9628:10210 [3] NCCL INFO Channel 02/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9628:10210 [3] NCCL INFO Channel 03/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9632:10212 [7] NCCL INFO Channel 00/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9632:10212 [7] NCCL INFO Channel 01/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9632:10212 [7] NCCL INFO Channel 02/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9632:10212 [7] NCCL INFO Channel 03/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9632:10212 [7] NCCL INFO Channel 00/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9632:10212 [7] NCCL INFO Channel 01/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9632:10212 [7] NCCL INFO Channel 02/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9632:10212 [7] NCCL INFO Channel 03/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-46:702886:703545 [6] NCCL INFO Channel 00/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9631:10211 [6] NCCL INFO Channel 00/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-46:702886:703545 [6] NCCL INFO Channel 01/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9631:10211 [6] NCCL INFO Channel 01/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-46:702886:703545 [6] NCCL INFO Channel 02/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9631:10211 [6] NCCL INFO Channel 02/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-46:702886:703545 [6] NCCL INFO Channel 03/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9631:10211 [6] NCCL INFO Channel 03/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-46:702886:703545 [6] NCCL INFO Channel 00/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9631:10211 [6] NCCL INFO Channel 00/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9631:10211 [6] NCCL INFO Channel 01/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-46:702886:703545 [6] NCCL INFO Channel 01/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-46:702886:703545 [6] NCCL INFO Channel 02/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9631:10211 [6] NCCL INFO Channel 02/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9631:10211 [6] NCCL INFO Channel 03/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-46:702886:703545 [6] NCCL INFO Channel 03/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9625:10208 [0] NCCL INFO Connected all rings
+ip-26-0-165-213:9625:10208 [0] NCCL INFO Connected all trees
+ip-26-0-165-213:9625:10208 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-213:9625:10208 [0] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-165-213:9625:10208 [0] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-165-213:9629:10206 [4] NCCL INFO Connected all rings
+ip-26-0-165-213:9629:10206 [4] NCCL INFO Connected all trees
+ip-26-0-165-213:9629:10206 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-213:9629:10206 [4] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-165-213:9629:10206 [4] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-162-46:702880:703539 [0] NCCL INFO Connected all rings
+ip-26-0-162-46:702880:703539 [0] NCCL INFO Connected all trees
+ip-26-0-162-46:702880:703539 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-162-46:702880:703539 [0] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-162-46:702880:703539 [0] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-165-213:9625:10208 [0] NCCL INFO comm 0xb1a2e70 rank 1 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0x38cb7e7c3e6712ad - Init COMPLETE
+ip-26-0-165-213:9625:10229 [0] NCCL INFO Channel 02/1 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA/Shared
+ip-26-0-165-213:9625:10229 [0] NCCL INFO Channel 03/1 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA/Shared
+ip-26-0-165-213:9629:10206 [4] NCCL INFO comm 0xac16920 rank 1 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0xf7a17e9c5701b2a3 - Init COMPLETE
+ip-26-0-165-213:9629:10230 [4] NCCL INFO Channel 02/1 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA/Shared
+ip-26-0-165-213:9629:10230 [4] NCCL INFO Channel 03/1 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA/Shared
+ip-26-0-162-46:702884:703534 [4] NCCL INFO Connected all rings
+ip-26-0-162-46:702884:703534 [4] NCCL INFO Connected all trees
+ip-26-0-162-46:702884:703534 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-162-46:702884:703534 [4] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-162-46:702884:703534 [4] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-162-46:702880:703539 [0] NCCL INFO comm 0x1c472390 rank 0 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0x38cb7e7c3e6712ad - Init COMPLETE
+ip-26-0-162-46:702880:703565 [0] NCCL INFO Channel 02/1 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA/Shared
+ip-26-0-162-46:702880:703565 [0] NCCL INFO Channel 03/1 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA/Shared
+ip-26-0-162-46:702884:703534 [4] NCCL INFO comm 0x1ba54ce0 rank 0 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0xf7a17e9c5701b2a3 - Init COMPLETE
+ip-26-0-162-46:702884:703566 [4] NCCL INFO Channel 02/1 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA/Shared
+ip-26-0-162-46:702884:703566 [4] NCCL INFO Channel 03/1 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA/Shared
+ip-26-0-165-213:9630:10205 [5] NCCL INFO Connected all rings
+ip-26-0-165-213:9630:10205 [5] NCCL INFO Connected all trees
+ip-26-0-165-213:9630:10205 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-213:9630:10205 [5] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-165-213:9630:10205 [5] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-162-46:702885:703536 [5] NCCL INFO Connected all rings
+ip-26-0-162-46:702885:703536 [5] NCCL INFO Connected all trees
+ip-26-0-162-46:702885:703536 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-162-46:702885:703536 [5] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-162-46:702885:703536 [5] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-165-213:9630:10205 [5] NCCL INFO comm 0xaa92090 rank 1 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0x9e98cc82a4e3e115 - Init COMPLETE
+ip-26-0-165-213:9630:10231 [5] NCCL INFO Channel 02/1 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA/Shared
+ip-26-0-165-213:9630:10231 [5] NCCL INFO Channel 03/1 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA/Shared
+ip-26-0-162-46:702885:703536 [5] NCCL INFO comm 0x1b6a2030 rank 0 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0x9e98cc82a4e3e115 - Init COMPLETE
+ip-26-0-162-46:702885:703567 [5] NCCL INFO Channel 02/1 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA/Shared
+ip-26-0-162-46:702885:703567 [5] NCCL INFO Channel 03/1 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA/Shared
+ip-26-0-165-213:9626:10207 [1] NCCL INFO Connected all rings
+ip-26-0-165-213:9626:10207 [1] NCCL INFO Connected all trees
+ip-26-0-165-213:9626:10207 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-213:9626:10207 [1] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-165-213:9626:10207 [1] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-162-46:702881:703538 [1] NCCL INFO Connected all rings
+ip-26-0-162-46:702881:703538 [1] NCCL INFO Connected all trees
+ip-26-0-162-46:702881:703538 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-162-46:702881:703538 [1] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-162-46:702881:703538 [1] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-165-213:9626:10207 [1] NCCL INFO comm 0xb9403c0 rank 1 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0x5d4e907b2e49529b - Init COMPLETE
+ip-26-0-165-213:9626:10232 [1] NCCL INFO Channel 02/1 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA/Shared
+ip-26-0-165-213:9626:10232 [1] NCCL INFO Channel 03/1 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA/Shared
+ip-26-0-162-46:702881:703538 [1] NCCL INFO comm 0x1bf45300 rank 0 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0x5d4e907b2e49529b - Init COMPLETE
+ip-26-0-162-46:702881:703568 [1] NCCL INFO Channel 02/1 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA/Shared
+ip-26-0-162-46:702881:703568 [1] NCCL INFO Channel 03/1 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA/Shared
+ip-26-0-162-46:702882:703542 [2] NCCL INFO Connected all rings
+ip-26-0-162-46:702882:703542 [2] NCCL INFO Connected all trees
+ip-26-0-162-46:702882:703542 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-162-46:702882:703542 [2] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-162-46:702882:703542 [2] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-165-213:9627:10209 [2] NCCL INFO Connected all rings
+ip-26-0-165-213:9627:10209 [2] NCCL INFO Connected all trees
+ip-26-0-165-213:9627:10209 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-213:9627:10209 [2] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-165-213:9627:10209 [2] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-162-46:702883:703543 [3] NCCL INFO Connected all rings
+ip-26-0-162-46:702883:703543 [3] NCCL INFO Connected all trees
+ip-26-0-162-46:702883:703543 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-162-46:702883:703543 [3] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-162-46:702883:703543 [3] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-162-46:702887:703547 [7] NCCL INFO Connected all rings
+ip-26-0-162-46:702887:703547 [7] NCCL INFO Connected all trees
+ip-26-0-162-46:702887:703547 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-162-46:702887:703547 [7] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-162-46:702887:703547 [7] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-165-213:9632:10212 [7] NCCL INFO Connected all rings
+ip-26-0-165-213:9632:10212 [7] NCCL INFO Connected all trees
+ip-26-0-165-213:9632:10212 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-213:9632:10212 [7] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-165-213:9632:10212 [7] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-165-213:9628:10210 [3] NCCL INFO Connected all rings
+ip-26-0-165-213:9628:10210 [3] NCCL INFO Connected all trees
+ip-26-0-165-213:9628:10210 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-213:9628:10210 [3] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-165-213:9628:10210 [3] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-162-46:702882:703542 [2] NCCL INFO comm 0x1c2fd420 rank 0 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0xf9e132832d0ec61b - Init COMPLETE
+ip-26-0-162-46:702882:703569 [2] NCCL INFO Channel 02/1 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA/Shared
+ip-26-0-162-46:702882:703569 [2] NCCL INFO Channel 03/1 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA/Shared
+ip-26-0-162-46:702886:703545 [6] NCCL INFO Connected all rings
+ip-26-0-162-46:702886:703545 [6] NCCL INFO Connected all trees
+ip-26-0-162-46:702886:703545 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-162-46:702886:703545 [6] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-162-46:702886:703545 [6] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-165-213:9631:10211 [6] NCCL INFO Connected all rings
+ip-26-0-165-213:9631:10211 [6] NCCL INFO Connected all trees
+ip-26-0-165-213:9631:10211 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-213:9631:10211 [6] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-165-213:9631:10211 [6] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-162-46:702883:703543 [3] NCCL INFO comm 0x1b0bcc30 rank 0 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0x53e8c79cced5760e - Init COMPLETE
+ip-26-0-165-213:9627:10209 [2] NCCL INFO comm 0xac1c710 rank 1 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0xf9e132832d0ec61b - Init COMPLETE
+ip-26-0-162-46:702883:703570 [3] NCCL INFO Channel 02/1 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA/Shared
+ip-26-0-162-46:702883:703570 [3] NCCL INFO Channel 03/1 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA/Shared
+ip-26-0-165-213:9627:10234 [2] NCCL INFO Channel 02/1 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA/Shared
+ip-26-0-165-213:9627:10234 [2] NCCL INFO Channel 03/1 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA/Shared
+ip-26-0-165-213:9632:10212 [7] NCCL INFO comm 0xafebf10 rank 1 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0xf82b03744fa9a6e6 - Init COMPLETE
+ip-26-0-162-46:702887:703547 [7] NCCL INFO comm 0x1b690380 rank 0 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0xf82b03744fa9a6e6 - Init COMPLETE
+ip-26-0-165-213:9632:10235 [7] NCCL INFO Channel 02/1 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA/Shared
+ip-26-0-165-213:9632:10235 [7] NCCL INFO Channel 03/1 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA/Shared
+ip-26-0-162-46:702887:703571 [7] NCCL INFO Channel 02/1 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA/Shared
+ip-26-0-162-46:702887:703571 [7] NCCL INFO Channel 03/1 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA/Shared
+ip-26-0-165-213:9628:10210 [3] NCCL INFO comm 0xb88fbd0 rank 1 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0x53e8c79cced5760e - Init COMPLETE
+ip-26-0-165-213:9628:10236 [3] NCCL INFO Channel 02/1 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA/Shared
+ip-26-0-165-213:9628:10236 [3] NCCL INFO Channel 03/1 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA/Shared
+ip-26-0-162-46:702886:703545 [6] NCCL INFO comm 0x1bb86170 rank 0 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0xba6c9e2b4095f929 - Init COMPLETE
+ip-26-0-162-46:702886:703572 [6] NCCL INFO Channel 02/1 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA/Shared
+ip-26-0-162-46:702886:703572 [6] NCCL INFO Channel 03/1 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA/Shared
+ip-26-0-165-213:9631:10211 [6] NCCL INFO comm 0xbdc7590 rank 1 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0xba6c9e2b4095f929 - Init COMPLETE
+ip-26-0-165-213:9631:10237 [6] NCCL INFO Channel 02/1 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA/Shared
+ip-26-0-165-213:9631:10237 [6] NCCL INFO Channel 03/1 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA/Shared
+ip-26-0-162-46:702885:703589 [5] NCCL INFO Channel 02/1 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA/Shared
+ip-26-0-162-46:702885:703589 [5] NCCL INFO Channel 03/1 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA/Shared
+ip-26-0-162-46:702884:703590 [4] NCCL INFO Channel 02/1 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA/Shared
+ip-26-0-162-46:702884:703590 [4] NCCL INFO Channel 03/1 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA/Shared
+ip-26-0-162-46:702880:703601 [0] NCCL INFO Channel 02/1 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA/Shared
+ip-26-0-162-46:702880:703601 [0] NCCL INFO Channel 03/1 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA/Shared
+ip-26-0-162-46:702881:703608 [1] NCCL INFO Channel 02/1 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA/Shared
+ip-26-0-162-46:702881:703608 [1] NCCL INFO Channel 03/1 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA/Shared
+ip-26-0-162-46:702882:703637 [2] NCCL INFO Channel 02/1 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA/Shared
+ip-26-0-162-46:702883:703641 [3] NCCL INFO Channel 02/1 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA/Shared
+ip-26-0-162-46:702882:703637 [2] NCCL INFO Channel 03/1 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA/Shared
+ip-26-0-162-46:702883:703641 [3] NCCL INFO Channel 03/1 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA/Shared
+ip-26-0-162-46:702886:703643 [6] NCCL INFO Channel 02/1 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA/Shared
+ip-26-0-162-46:702887:703644 [7] NCCL INFO Channel 02/1 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA/Shared
+ip-26-0-162-46:702886:703643 [6] NCCL INFO Channel 03/1 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA/Shared
+ip-26-0-162-46:702887:703644 [7] NCCL INFO Channel 03/1 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA/Shared
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+NCCL version 2.18.6+cuda12.1
+ip-26-0-165-213:9630:10243 [5] NCCL INFO Using network Libfabric
+ip-26-0-165-213:9629:10242 [4] NCCL INFO Using network Libfabric
+ip-26-0-166-125:762819:763425 [4] NCCL INFO Using network Libfabric
+ip-26-0-166-125:762820:763426 [5] NCCL INFO Using network Libfabric
+ip-26-0-166-125:762820:763426 [5] NCCL INFO comm 0xb6eccc0 rank 1 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0xe9585f6fdf5c131 - Init START
+ip-26-0-165-213:9630:10243 [5] NCCL INFO comm 0x135eca50 rank 0 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0xe9585f6fdf5c131 - Init START
+ip-26-0-166-125:762820:763426 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-165-213:9630:10243 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-125:762819:763425 [4] NCCL INFO comm 0xb6dcff0 rank 1 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0x9e5f5724852a0060 - Init START
+ip-26-0-165-213:9629:10242 [4] NCCL INFO comm 0x1376f2c0 rank 0 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0x9e5f5724852a0060 - Init START
+ip-26-0-165-213:9629:10242 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-125:762819:763425 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+ip-26-0-165-213:9625:10246 [0] NCCL INFO Using network Libfabric
+ip-26-0-165-213:9626:10247 [1] NCCL INFO Using network Libfabric
+ip-26-0-166-125:762815:763427 [0] NCCL INFO Using network Libfabric
+ip-26-0-166-125:762816:763428 [1] NCCL INFO Using network Libfabric
+ip-26-0-166-125:762815:763427 [0] NCCL INFO comm 0xa97cea0 rank 1 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0x937261382f4d117a - Init START
+ip-26-0-165-213:9625:10246 [0] NCCL INFO comm 0x13c3d540 rank 0 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0x937261382f4d117a - Init START
+ip-26-0-166-125:762816:763428 [1] NCCL INFO comm 0xb9eac40 rank 1 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0x4f3f8ce10223be04 - Init START
+ip-26-0-165-213:9626:10247 [1] NCCL INFO comm 0x14212190 rank 0 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0x4f3f8ce10223be04 - Init START
+ip-26-0-166-125:762815:763427 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-165-213:9625:10246 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-165-213:9626:10247 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-125:762816:763428 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-125:762820:763426 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,fff00000,00000000
+ip-26-0-165-213:9630:10243 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-165-213:9630:10243 [5] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-165-213:9630:10243 [5] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-166-125:762820:763426 [5] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-166-125:762820:763426 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-213:9630:10243 [5] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-165-213:9630:10243 [5] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-165-213:9630:10243 [5] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-165-213:9630:10243 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-125:762819:763425 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,fff00000,00000000
+ip-26-0-165-213:9629:10242 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-165-213:9629:10242 [4] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-165-213:9629:10242 [4] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-166-125:762819:763425 [4] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-166-125:762819:763425 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-213:9629:10242 [4] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-165-213:9629:10242 [4] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-165-213:9629:10242 [4] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-165-213:9629:10242 [4] NCCL INFO P2P Chunksize set to 131072
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+NCCL version 2.18.6+cuda12.1
+ip-26-0-165-213:9627:10253 [2] NCCL INFO Using network Libfabric
+ip-26-0-166-125:762817:763433 [2] NCCL INFO Using network Libfabric
+ip-26-0-166-125:762818:763434 [3] NCCL INFO Using network Libfabric
+ip-26-0-165-213:9628:10255 [3] NCCL INFO Using network Libfabric
+ip-26-0-166-125:762817:763433 [2] NCCL INFO comm 0xc7dc860 rank 1 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0x396224c66001810a - Init START
+ip-26-0-165-213:9627:10253 [2] NCCL INFO comm 0x13775a80 rank 0 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0x396224c66001810a - Init START
+ip-26-0-166-125:762817:763433 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-165-213:9627:10253 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-125:762818:763434 [3] NCCL INFO comm 0xab23300 rank 1 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0x207e7c0e85972fff - Init START
+ip-26-0-165-213:9628:10255 [3] NCCL INFO comm 0x1415c080 rank 0 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0x207e7c0e85972fff - Init START
+ip-26-0-166-125:762818:763434 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-165-213:9628:10255 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-125:762820:763426 [5] NCCL INFO Channel 00/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762820:763426 [5] NCCL INFO Channel 01/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762820:763426 [5] NCCL INFO Channel 02/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762820:763426 [5] NCCL INFO Channel 03/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762820:763426 [5] NCCL INFO Channel 00/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762820:763426 [5] NCCL INFO Channel 01/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762820:763426 [5] NCCL INFO Channel 02/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762820:763426 [5] NCCL INFO Channel 03/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9630:10243 [5] NCCL INFO Channel 00/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9630:10243 [5] NCCL INFO Channel 01/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9630:10243 [5] NCCL INFO Channel 02/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9630:10243 [5] NCCL INFO Channel 03/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9630:10243 [5] NCCL INFO Channel 00/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9630:10243 [5] NCCL INFO Channel 01/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9630:10243 [5] NCCL INFO Channel 02/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9630:10243 [5] NCCL INFO Channel 03/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9629:10242 [4] NCCL INFO Channel 00/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762819:763425 [4] NCCL INFO Channel 00/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9629:10242 [4] NCCL INFO Channel 01/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762819:763425 [4] NCCL INFO Channel 01/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9629:10242 [4] NCCL INFO Channel 02/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762819:763425 [4] NCCL INFO Channel 02/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9629:10242 [4] NCCL INFO Channel 03/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762819:763425 [4] NCCL INFO Channel 03/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9629:10242 [4] NCCL INFO Channel 00/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762819:763425 [4] NCCL INFO Channel 00/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9629:10242 [4] NCCL INFO Channel 01/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762819:763425 [4] NCCL INFO Channel 01/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9629:10242 [4] NCCL INFO Channel 02/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762819:763425 [4] NCCL INFO Channel 02/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9629:10242 [4] NCCL INFO Channel 03/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762819:763425 [4] NCCL INFO Channel 03/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+ip-26-0-165-213:9631:10257 [6] NCCL INFO Using network Libfabric
+ip-26-0-166-125:762821:763435 [6] NCCL INFO Using network Libfabric
+ip-26-0-166-125:762821:763435 [6] NCCL INFO comm 0xac8d140 rank 1 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0x4a1373ad47fec75c - Init START
+ip-26-0-165-213:9631:10257 [6] NCCL INFO comm 0x14920a00 rank 0 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0x4a1373ad47fec75c - Init START
+ip-26-0-166-125:762821:763435 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-165-213:9631:10257 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+NCCL version 2.18.6+cuda12.1
+ip-26-0-165-213:9632:10259 [7] NCCL INFO Using network Libfabric
+ip-26-0-166-125:762822:763436 [7] NCCL INFO Using network Libfabric
+ip-26-0-166-125:762822:763436 [7] NCCL INFO comm 0xc3cad90 rank 1 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0x2bcecbf67b328445 - Init START
+ip-26-0-165-213:9632:10259 [7] NCCL INFO comm 0x13b48f90 rank 0 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0x2bcecbf67b328445 - Init START
+ip-26-0-166-125:762822:763436 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-165-213:9632:10259 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-165-213:9626:10247 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-166-125:762815:763427 [0] NCCL INFO Setting affinity for GPU 0 to 0fff,ffffffff
+ip-26-0-166-125:762816:763428 [1] NCCL INFO Setting affinity for GPU 1 to 0fff,ffffffff
+ip-26-0-166-125:762816:763428 [1] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-166-125:762816:763428 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-213:9626:10247 [1] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-165-213:9626:10247 [1] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-165-213:9626:10247 [1] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-165-213:9626:10247 [1] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-165-213:9626:10247 [1] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-165-213:9626:10247 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-125:762816:763428 [1] NCCL INFO Channel 00/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762816:763428 [1] NCCL INFO Channel 01/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762816:763428 [1] NCCL INFO Channel 02/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762816:763428 [1] NCCL INFO Channel 03/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762816:763428 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762816:763428 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762816:763428 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762816:763428 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9626:10247 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9626:10247 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9626:10247 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9626:10247 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9626:10247 [1] NCCL INFO Channel 00/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9626:10247 [1] NCCL INFO Channel 01/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9626:10247 [1] NCCL INFO Channel 02/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9626:10247 [1] NCCL INFO Channel 03/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9625:10246 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-165-213:9625:10246 [0] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-165-213:9625:10246 [0] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-165-213:9625:10246 [0] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-165-213:9625:10246 [0] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-165-213:9625:10246 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-165-213:9625:10246 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-125:762815:763427 [0] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-166-125:762815:763427 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-125:762817:763433 [2] NCCL INFO Setting affinity for GPU 2 to 0fff,ffffffff
+ip-26-0-165-213:9628:10255 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-165-213:9625:10246 [0] NCCL INFO Channel 00/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9625:10246 [0] NCCL INFO Channel 01/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9625:10246 [0] NCCL INFO Channel 02/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9625:10246 [0] NCCL INFO Channel 03/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9625:10246 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9625:10246 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9625:10246 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9625:10246 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762815:763427 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762815:763427 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762815:763427 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762815:763427 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762815:763427 [0] NCCL INFO Channel 00/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762815:763427 [0] NCCL INFO Channel 01/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762815:763427 [0] NCCL INFO Channel 02/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762815:763427 [0] NCCL INFO Channel 03/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9627:10253 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-165-213:9627:10253 [2] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-165-213:9627:10253 [2] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-165-213:9627:10253 [2] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-165-213:9627:10253 [2] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-165-213:9627:10253 [2] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-165-213:9627:10253 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-125:762817:763433 [2] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-166-125:762817:763433 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-125:762818:763434 [3] NCCL INFO Setting affinity for GPU 3 to 0fff,ffffffff
+ip-26-0-166-125:762818:763434 [3] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-166-125:762818:763434 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-213:9628:10255 [3] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-165-213:9628:10255 [3] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-165-213:9628:10255 [3] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-165-213:9628:10255 [3] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-165-213:9628:10255 [3] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-165-213:9628:10255 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-125:762821:763435 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,fff00000,00000000
+ip-26-0-165-213:9632:10259 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-166-125:762822:763436 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,fff00000,00000000
+ip-26-0-166-125:762822:763436 [7] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-166-125:762822:763436 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-213:9632:10259 [7] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-165-213:9632:10259 [7] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-165-213:9632:10259 [7] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-165-213:9632:10259 [7] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-165-213:9632:10259 [7] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-165-213:9632:10259 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-213:9631:10257 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-165-213:9631:10257 [6] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-165-213:9631:10257 [6] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-165-213:9631:10257 [6] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-166-125:762821:763435 [6] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-166-125:762821:763435 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-213:9631:10257 [6] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-165-213:9631:10257 [6] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-165-213:9631:10257 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-125:762817:763433 [2] NCCL INFO Channel 00/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762817:763433 [2] NCCL INFO Channel 01/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762817:763433 [2] NCCL INFO Channel 02/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762817:763433 [2] NCCL INFO Channel 03/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762817:763433 [2] NCCL INFO Channel 00/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9627:10253 [2] NCCL INFO Channel 00/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762817:763433 [2] NCCL INFO Channel 01/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762817:763433 [2] NCCL INFO Channel 02/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9627:10253 [2] NCCL INFO Channel 01/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762817:763433 [2] NCCL INFO Channel 03/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9627:10253 [2] NCCL INFO Channel 02/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9627:10253 [2] NCCL INFO Channel 03/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9627:10253 [2] NCCL INFO Channel 00/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9627:10253 [2] NCCL INFO Channel 01/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9627:10253 [2] NCCL INFO Channel 02/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9627:10253 [2] NCCL INFO Channel 03/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762818:763434 [3] NCCL INFO Channel 00/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762818:763434 [3] NCCL INFO Channel 01/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762818:763434 [3] NCCL INFO Channel 02/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762818:763434 [3] NCCL INFO Channel 03/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762818:763434 [3] NCCL INFO Channel 00/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762818:763434 [3] NCCL INFO Channel 01/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762818:763434 [3] NCCL INFO Channel 02/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762818:763434 [3] NCCL INFO Channel 03/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9628:10255 [3] NCCL INFO Channel 00/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9628:10255 [3] NCCL INFO Channel 01/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9628:10255 [3] NCCL INFO Channel 02/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9628:10255 [3] NCCL INFO Channel 03/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9628:10255 [3] NCCL INFO Channel 00/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9628:10255 [3] NCCL INFO Channel 01/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9628:10255 [3] NCCL INFO Channel 02/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9628:10255 [3] NCCL INFO Channel 03/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762822:763436 [7] NCCL INFO Channel 00/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762822:763436 [7] NCCL INFO Channel 01/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762822:763436 [7] NCCL INFO Channel 02/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762822:763436 [7] NCCL INFO Channel 03/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762822:763436 [7] NCCL INFO Channel 00/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762822:763436 [7] NCCL INFO Channel 01/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762822:763436 [7] NCCL INFO Channel 02/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762822:763436 [7] NCCL INFO Channel 03/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762821:763435 [6] NCCL INFO Channel 00/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762821:763435 [6] NCCL INFO Channel 01/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762821:763435 [6] NCCL INFO Channel 02/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762821:763435 [6] NCCL INFO Channel 03/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762821:763435 [6] NCCL INFO Channel 00/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762821:763435 [6] NCCL INFO Channel 01/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762821:763435 [6] NCCL INFO Channel 02/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762821:763435 [6] NCCL INFO Channel 03/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9632:10259 [7] NCCL INFO Channel 00/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9632:10259 [7] NCCL INFO Channel 01/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9632:10259 [7] NCCL INFO Channel 02/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9632:10259 [7] NCCL INFO Channel 03/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9632:10259 [7] NCCL INFO Channel 00/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9632:10259 [7] NCCL INFO Channel 01/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9632:10259 [7] NCCL INFO Channel 02/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9632:10259 [7] NCCL INFO Channel 03/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9631:10257 [6] NCCL INFO Channel 00/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9631:10257 [6] NCCL INFO Channel 01/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9631:10257 [6] NCCL INFO Channel 02/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9631:10257 [6] NCCL INFO Channel 03/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9631:10257 [6] NCCL INFO Channel 00/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9631:10257 [6] NCCL INFO Channel 01/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9631:10257 [6] NCCL INFO Channel 02/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9631:10257 [6] NCCL INFO Channel 03/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9629:10242 [4] NCCL INFO Connected all rings
+ip-26-0-165-213:9629:10242 [4] NCCL INFO Connected all trees
+ip-26-0-165-213:9629:10242 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-213:9629:10242 [4] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-165-213:9629:10242 [4] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-165-213:9630:10243 [5] NCCL INFO Connected all rings
+ip-26-0-165-213:9630:10243 [5] NCCL INFO Connected all trees
+ip-26-0-165-213:9630:10243 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-213:9630:10243 [5] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-165-213:9630:10243 [5] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-125:762819:763425 [4] NCCL INFO Connected all rings
+ip-26-0-166-125:762819:763425 [4] NCCL INFO Connected all trees
+ip-26-0-166-125:762819:763425 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-125:762819:763425 [4] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-125:762819:763425 [4] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-125:762820:763426 [5] NCCL INFO Connected all rings
+ip-26-0-166-125:762820:763426 [5] NCCL INFO Connected all trees
+ip-26-0-166-125:762820:763426 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-125:762820:763426 [5] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-125:762820:763426 [5] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-165-213:9629:10242 [4] NCCL INFO comm 0x1376f2c0 rank 0 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0x9e5f5724852a0060 - Init COMPLETE
+ip-26-0-165-213:9629:10272 [4] NCCL INFO Channel 02/1 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA/Shared
+ip-26-0-165-213:9629:10272 [4] NCCL INFO Channel 03/1 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA/Shared
+ip-26-0-165-213:9630:10243 [5] NCCL INFO comm 0x135eca50 rank 0 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0xe9585f6fdf5c131 - Init COMPLETE
+ip-26-0-165-213:9630:10273 [5] NCCL INFO Channel 02/1 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA/Shared
+ip-26-0-165-213:9630:10273 [5] NCCL INFO Channel 03/1 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA/Shared
+ip-26-0-165-213:9626:10247 [1] NCCL INFO Connected all rings
+ip-26-0-165-213:9626:10247 [1] NCCL INFO Connected all trees
+ip-26-0-165-213:9626:10247 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-213:9626:10247 [1] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-165-213:9626:10247 [1] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-125:762819:763425 [4] NCCL INFO comm 0xb6dcff0 rank 1 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0x9e5f5724852a0060 - Init COMPLETE
+ip-26-0-166-125:762819:763450 [4] NCCL INFO Channel 02/1 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA/Shared
+ip-26-0-166-125:762820:763426 [5] NCCL INFO comm 0xb6eccc0 rank 1 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0xe9585f6fdf5c131 - Init COMPLETE
+ip-26-0-166-125:762819:763450 [4] NCCL INFO Channel 03/1 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA/Shared
+ip-26-0-166-125:762820:763451 [5] NCCL INFO Channel 02/1 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA/Shared
+ip-26-0-166-125:762820:763451 [5] NCCL INFO Channel 03/1 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA/Shared
+ip-26-0-166-125:762816:763428 [1] NCCL INFO Connected all rings
+ip-26-0-166-125:762816:763428 [1] NCCL INFO Connected all trees
+ip-26-0-166-125:762816:763428 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-125:762816:763428 [1] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-125:762816:763428 [1] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-165-213:9626:10247 [1] NCCL INFO comm 0x14212190 rank 0 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0x4f3f8ce10223be04 - Init COMPLETE
+ip-26-0-165-213:9626:10274 [1] NCCL INFO Channel 02/1 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA/Shared
+ip-26-0-165-213:9626:10274 [1] NCCL INFO Channel 03/1 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA/Shared
+ip-26-0-166-125:762816:763428 [1] NCCL INFO comm 0xb9eac40 rank 1 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0x4f3f8ce10223be04 - Init COMPLETE
+ip-26-0-166-125:762816:763452 [1] NCCL INFO Channel 02/1 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA/Shared
+ip-26-0-166-125:762816:763452 [1] NCCL INFO Channel 03/1 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA/Shared
+ip-26-0-165-213:9625:10246 [0] NCCL INFO Connected all rings
+ip-26-0-165-213:9625:10246 [0] NCCL INFO Connected all trees
+ip-26-0-165-213:9625:10246 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-213:9625:10246 [0] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-165-213:9625:10246 [0] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-165-213:9625:10246 [0] NCCL INFO comm 0x13c3d540 rank 0 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0x937261382f4d117a - Init COMPLETE
+ip-26-0-165-213:9625:10275 [0] NCCL INFO Channel 02/1 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA/Shared
+ip-26-0-165-213:9625:10275 [0] NCCL INFO Channel 03/1 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA/Shared
+ip-26-0-166-125:762815:763427 [0] NCCL INFO Connected all rings
+ip-26-0-166-125:762815:763427 [0] NCCL INFO Connected all trees
+ip-26-0-166-125:762815:763427 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-125:762815:763427 [0] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-125:762815:763427 [0] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-125:762815:763427 [0] NCCL INFO comm 0xa97cea0 rank 1 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0x937261382f4d117a - Init COMPLETE
+ip-26-0-166-125:762815:763453 [0] NCCL INFO Channel 02/1 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA/Shared
+ip-26-0-166-125:762815:763453 [0] NCCL INFO Channel 03/1 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA/Shared
+ip-26-0-166-125:762818:763434 [3] NCCL INFO Connected all rings
+ip-26-0-166-125:762818:763434 [3] NCCL INFO Connected all trees
+ip-26-0-166-125:762818:763434 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-125:762818:763434 [3] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-125:762818:763434 [3] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-165-213:9627:10253 [2] NCCL INFO Connected all rings
+ip-26-0-165-213:9627:10253 [2] NCCL INFO Connected all trees
+ip-26-0-165-213:9627:10253 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-213:9627:10253 [2] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-165-213:9627:10253 [2] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-125:762817:763433 [2] NCCL INFO Connected all rings
+ip-26-0-166-125:762817:763433 [2] NCCL INFO Connected all trees
+ip-26-0-166-125:762817:763433 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-125:762817:763433 [2] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-125:762817:763433 [2] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-165-213:9628:10255 [3] NCCL INFO Connected all rings
+ip-26-0-165-213:9628:10255 [3] NCCL INFO Connected all trees
+ip-26-0-165-213:9628:10255 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-213:9628:10255 [3] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-165-213:9628:10255 [3] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-125:762818:763434 [3] NCCL INFO comm 0xab23300 rank 1 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0x207e7c0e85972fff - Init COMPLETE
+ip-26-0-166-125:762818:763454 [3] NCCL INFO Channel 02/1 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA/Shared
+ip-26-0-165-213:9627:10253 [2] NCCL INFO comm 0x13775a80 rank 0 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0x396224c66001810a - Init COMPLETE
+ip-26-0-166-125:762818:763454 [3] NCCL INFO Channel 03/1 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA/Shared
+ip-26-0-165-213:9627:10276 [2] NCCL INFO Channel 02/1 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA/Shared
+ip-26-0-165-213:9627:10276 [2] NCCL INFO Channel 03/1 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA/Shared
+ip-26-0-166-125:762817:763433 [2] NCCL INFO comm 0xc7dc860 rank 1 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0x396224c66001810a - Init COMPLETE
+ip-26-0-166-125:762817:763455 [2] NCCL INFO Channel 02/1 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA/Shared
+ip-26-0-166-125:762817:763455 [2] NCCL INFO Channel 03/1 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA/Shared
+ip-26-0-165-213:9628:10255 [3] NCCL INFO comm 0x1415c080 rank 0 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0x207e7c0e85972fff - Init COMPLETE
+ip-26-0-165-213:9628:10277 [3] NCCL INFO Channel 02/1 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA/Shared
+ip-26-0-165-213:9628:10277 [3] NCCL INFO Channel 03/1 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA/Shared
+ip-26-0-166-125:762821:763435 [6] NCCL INFO Connected all rings
+ip-26-0-166-125:762821:763435 [6] NCCL INFO Connected all trees
+ip-26-0-166-125:762821:763435 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-125:762821:763435 [6] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-125:762821:763435 [6] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-125:762821:763435 [6] NCCL INFO comm 0xac8d140 rank 1 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0x4a1373ad47fec75c - Init COMPLETE
+ip-26-0-166-125:762821:763456 [6] NCCL INFO Channel 02/1 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA/Shared
+ip-26-0-166-125:762821:763456 [6] NCCL INFO Channel 03/1 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA/Shared
+ip-26-0-166-125:762822:763436 [7] NCCL INFO Connected all rings
+ip-26-0-166-125:762822:763436 [7] NCCL INFO Connected all trees
+ip-26-0-166-125:762822:763436 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-125:762822:763436 [7] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-125:762822:763436 [7] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-165-213:9632:10259 [7] NCCL INFO Connected all rings
+ip-26-0-165-213:9632:10259 [7] NCCL INFO Connected all trees
+ip-26-0-165-213:9632:10259 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-213:9632:10259 [7] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-165-213:9632:10259 [7] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-165-213:9631:10257 [6] NCCL INFO Connected all rings
+ip-26-0-165-213:9631:10257 [6] NCCL INFO Connected all trees
+ip-26-0-165-213:9631:10257 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-213:9631:10257 [6] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-165-213:9631:10257 [6] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-125:762822:763436 [7] NCCL INFO comm 0xc3cad90 rank 1 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0x2bcecbf67b328445 - Init COMPLETE
+ip-26-0-166-125:762822:763457 [7] NCCL INFO Channel 02/1 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA/Shared
+ip-26-0-166-125:762822:763457 [7] NCCL INFO Channel 03/1 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA/Shared
+ip-26-0-165-213:9632:10259 [7] NCCL INFO comm 0x13b48f90 rank 0 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0x2bcecbf67b328445 - Init COMPLETE
+ip-26-0-165-213:9632:10278 [7] NCCL INFO Channel 02/1 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA/Shared
+ip-26-0-165-213:9632:10278 [7] NCCL INFO Channel 03/1 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA/Shared
+ip-26-0-165-213:9631:10257 [6] NCCL INFO comm 0x14920a00 rank 0 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0x4a1373ad47fec75c - Init COMPLETE
+ip-26-0-165-213:9631:10279 [6] NCCL INFO Channel 02/1 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA/Shared
+ip-26-0-165-213:9631:10279 [6] NCCL INFO Channel 03/1 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA/Shared
+ip-26-0-165-213:9629:10297 [4] NCCL INFO Channel 02/1 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA/Shared
+ip-26-0-165-213:9630:10298 [5] NCCL INFO Channel 02/1 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA/Shared
+ip-26-0-165-213:9629:10297 [4] NCCL INFO Channel 03/1 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA/Shared
+ip-26-0-165-213:9630:10298 [5] NCCL INFO Channel 03/1 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA/Shared
+ip-26-0-165-213:9626:10315 [1] NCCL INFO Channel 02/1 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA/Shared
+ip-26-0-165-213:9625:10316 [0] NCCL INFO Channel 02/1 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA/Shared
+ip-26-0-165-213:9626:10315 [1] NCCL INFO Channel 03/1 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA/Shared
+ip-26-0-165-213:9625:10316 [0] NCCL INFO Channel 03/1 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA/Shared
+ip-26-0-165-213:9627:10333 [2] NCCL INFO Channel 02/1 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA/Shared
+ip-26-0-165-213:9627:10333 [2] NCCL INFO Channel 03/1 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA/Shared
+ip-26-0-165-213:9628:10334 [3] NCCL INFO Channel 02/1 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA/Shared
+ip-26-0-165-213:9628:10334 [3] NCCL INFO Channel 03/1 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA/Shared
+ip-26-0-165-213:9631:10352 [6] NCCL INFO Channel 02/1 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA/Shared
+ip-26-0-165-213:9632:10351 [7] NCCL INFO Channel 02/1 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA/Shared
+ip-26-0-165-213:9631:10352 [6] NCCL INFO Channel 03/1 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA/Shared
+ip-26-0-165-213:9632:10351 [7] NCCL INFO Channel 03/1 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA/Shared
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+NCCL version 2.18.6+cuda12.1
+ip-26-0-166-125:762819:763461 [4] NCCL INFO Using network Libfabric
+ip-26-0-166-125:762820:763462 [5] NCCL INFO Using network Libfabric
+ip-26-0-166-15:127295:127874 [5] NCCL INFO Using network Libfabric
+ip-26-0-166-15:127294:127875 [4] NCCL INFO Using network Libfabric
+ip-26-0-166-15:127295:127874 [5] NCCL INFO comm 0xae0ea00 rank 1 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0x79de6ffa55f6be9d - Init START
+ip-26-0-166-125:762820:763462 [5] NCCL INFO comm 0x14159ec0 rank 0 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0x79de6ffa55f6be9d - Init START
+ip-26-0-166-15:127294:127875 [4] NCCL INFO comm 0xb2f8fe0 rank 1 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0xdbf8660b36dca753 - Init START
+ip-26-0-166-125:762819:763461 [4] NCCL INFO comm 0x14238520 rank 0 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0xdbf8660b36dca753 - Init START
+ip-26-0-166-15:127295:127874 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-125:762820:763462 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-15:127294:127875 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-125:762819:763461 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+ip-26-0-166-125:762816:763465 [1] NCCL INFO Using network Libfabric
+ip-26-0-166-125:762815:763467 [0] NCCL INFO Using network Libfabric
+ip-26-0-166-15:127291:127876 [1] NCCL INFO Using network Libfabric
+ip-26-0-166-15:127290:127877 [0] NCCL INFO Using network Libfabric
+ip-26-0-166-15:127290:127877 [0] NCCL INFO comm 0xaf28780 rank 1 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0x169415bf4ad33309 - Init START
+ip-26-0-166-125:762815:763467 [0] NCCL INFO comm 0x13425100 rank 0 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0x169415bf4ad33309 - Init START
+ip-26-0-166-15:127290:127877 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-125:762815:763467 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-15:127291:127876 [1] NCCL INFO comm 0xc5da7a0 rank 1 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0x394ef85110bc3661 - Init START
+ip-26-0-166-125:762816:763465 [1] NCCL INFO comm 0x14456100 rank 0 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0x394ef85110bc3661 - Init START
+ip-26-0-166-15:127291:127876 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-125:762816:763465 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-15:127295:127874 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-166-15:127294:127875 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-166-125:762819:763461 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,fff00000,00000000
+ip-26-0-166-125:762819:763461 [4] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-166-125:762819:763461 [4] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-166-125:762819:763461 [4] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-166-125:762819:763461 [4] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-166-15:127294:127875 [4] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-166-15:127294:127875 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-125:762819:763461 [4] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-166-125:762819:763461 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-125:762820:763462 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,fff00000,00000000
+ip-26-0-166-125:762820:763462 [5] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-166-125:762820:763462 [5] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-166-125:762820:763462 [5] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-166-125:762820:763462 [5] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-166-125:762820:763462 [5] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-166-125:762820:763462 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-15:127295:127874 [5] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-166-15:127295:127874 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-125:762819:763461 [4] NCCL INFO Channel 00/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762819:763461 [4] NCCL INFO Channel 01/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762819:763461 [4] NCCL INFO Channel 02/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762819:763461 [4] NCCL INFO Channel 03/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762819:763461 [4] NCCL INFO Channel 00/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762819:763461 [4] NCCL INFO Channel 01/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762819:763461 [4] NCCL INFO Channel 02/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762819:763461 [4] NCCL INFO Channel 03/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127294:127875 [4] NCCL INFO Channel 00/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127294:127875 [4] NCCL INFO Channel 01/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127294:127875 [4] NCCL INFO Channel 02/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127294:127875 [4] NCCL INFO Channel 03/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127294:127875 [4] NCCL INFO Channel 00/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127294:127875 [4] NCCL INFO Channel 01/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127294:127875 [4] NCCL INFO Channel 02/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127294:127875 [4] NCCL INFO Channel 03/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127295:127874 [5] NCCL INFO Channel 00/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127295:127874 [5] NCCL INFO Channel 01/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127295:127874 [5] NCCL INFO Channel 02/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127295:127874 [5] NCCL INFO Channel 03/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127295:127874 [5] NCCL INFO Channel 00/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127295:127874 [5] NCCL INFO Channel 01/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127295:127874 [5] NCCL INFO Channel 02/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127295:127874 [5] NCCL INFO Channel 03/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762820:763462 [5] NCCL INFO Channel 00/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762820:763462 [5] NCCL INFO Channel 01/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762820:763462 [5] NCCL INFO Channel 02/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762820:763462 [5] NCCL INFO Channel 03/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762820:763462 [5] NCCL INFO Channel 00/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762820:763462 [5] NCCL INFO Channel 01/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762820:763462 [5] NCCL INFO Channel 02/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762820:763462 [5] NCCL INFO Channel 03/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127290:127877 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-166-15:127291:127876 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-166-125:762815:763467 [0] NCCL INFO Setting affinity for GPU 0 to 0fff,ffffffff
+ip-26-0-166-125:762815:763467 [0] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-166-125:762815:763467 [0] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-166-125:762815:763467 [0] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-166-15:127290:127877 [0] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-166-15:127290:127877 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-125:762815:763467 [0] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-166-125:762815:763467 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-166-125:762815:763467 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-125:762816:763465 [1] NCCL INFO Setting affinity for GPU 1 to 0fff,ffffffff
+ip-26-0-166-125:762816:763465 [1] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-166-125:762816:763465 [1] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-166-125:762816:763465 [1] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-166-125:762816:763465 [1] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-166-125:762816:763465 [1] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-166-125:762816:763465 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-15:127291:127876 [1] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-166-15:127291:127876 [1] NCCL INFO P2P Chunksize set to 131072
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+NCCL version 2.18.6+cuda12.1
+ip-26-0-166-125:762817:763478 [2] NCCL INFO Using network Libfabric
+ip-26-0-166-125:762818:763479 [3] NCCL INFO Using network Libfabric
+ip-26-0-166-15:127292:127887 [2] NCCL INFO Using network Libfabric
+ip-26-0-166-15:127293:127886 [3] NCCL INFO Using network Libfabric
+ip-26-0-166-15:127292:127887 [2] NCCL INFO comm 0xb57e240 rank 1 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0x94d6c206686962eb - Init START
+ip-26-0-166-125:762817:763478 [2] NCCL INFO comm 0x15252410 rank 0 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0x94d6c206686962eb - Init START
+ip-26-0-166-15:127293:127886 [3] NCCL INFO comm 0xc47f230 rank 1 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0x802584e5feb4e98a - Init START
+ip-26-0-166-125:762818:763479 [3] NCCL INFO comm 0x1358f380 rank 0 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0x802584e5feb4e98a - Init START
+ip-26-0-166-15:127292:127887 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-125:762817:763478 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-15:127293:127886 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-125:762818:763479 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-15:127290:127877 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127290:127877 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127290:127877 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127290:127877 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127290:127877 [0] NCCL INFO Channel 00/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127290:127877 [0] NCCL INFO Channel 01/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127290:127877 [0] NCCL INFO Channel 02/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127290:127877 [0] NCCL INFO Channel 03/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762815:763467 [0] NCCL INFO Channel 00/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762815:763467 [0] NCCL INFO Channel 01/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762815:763467 [0] NCCL INFO Channel 02/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762815:763467 [0] NCCL INFO Channel 03/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762815:763467 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762815:763467 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762815:763467 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762815:763467 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127291:127876 [1] NCCL INFO Channel 00/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127291:127876 [1] NCCL INFO Channel 01/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127291:127876 [1] NCCL INFO Channel 02/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127291:127876 [1] NCCL INFO Channel 03/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127291:127876 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127291:127876 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127291:127876 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127291:127876 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762816:763465 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762816:763465 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762816:763465 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762816:763465 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762816:763465 [1] NCCL INFO Channel 00/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762816:763465 [1] NCCL INFO Channel 01/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762816:763465 [1] NCCL INFO Channel 02/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762816:763465 [1] NCCL INFO Channel 03/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+ip-26-0-166-125:762821:763481 [6] NCCL INFO Using network Libfabric
+NCCL version 2.18.6+cuda12.1
+ip-26-0-166-15:127296:127889 [6] NCCL INFO Using network Libfabric
+ip-26-0-166-125:762822:763483 [7] NCCL INFO Using network Libfabric
+ip-26-0-166-15:127296:127889 [6] NCCL INFO comm 0xba205a0 rank 1 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0xb814955b7173143c - Init START
+ip-26-0-166-125:762821:763481 [6] NCCL INFO comm 0x136fcea0 rank 0 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0xb814955b7173143c - Init START
+ip-26-0-166-125:762821:763481 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-15:127297:127890 [7] NCCL INFO Using network Libfabric
+ip-26-0-166-15:127296:127889 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-15:127297:127890 [7] NCCL INFO comm 0xc13e470 rank 1 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0xdfcf3744600a38a8 - Init START
+ip-26-0-166-125:762822:763483 [7] NCCL INFO comm 0x14e365e0 rank 0 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0xdfcf3744600a38a8 - Init START
+ip-26-0-166-15:127297:127890 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-125:762822:763483 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-125:762817:763478 [2] NCCL INFO Setting affinity for GPU 2 to 0fff,ffffffff
+ip-26-0-166-15:127293:127886 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-166-125:762818:763479 [3] NCCL INFO Setting affinity for GPU 3 to 0fff,ffffffff
+ip-26-0-166-15:127293:127886 [3] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-166-15:127293:127886 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-125:762818:763479 [3] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-166-125:762818:763479 [3] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-166-125:762818:763479 [3] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-166-125:762818:763479 [3] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-166-125:762818:763479 [3] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-166-125:762818:763479 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-15:127292:127887 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-166-15:127292:127887 [2] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-166-15:127292:127887 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-125:762817:763478 [2] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-166-125:762817:763478 [2] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-166-125:762817:763478 [2] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-166-125:762817:763478 [2] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-166-125:762817:763478 [2] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-166-125:762817:763478 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-15:127293:127886 [3] NCCL INFO Channel 00/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127293:127886 [3] NCCL INFO Channel 01/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127293:127886 [3] NCCL INFO Channel 02/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127293:127886 [3] NCCL INFO Channel 03/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127293:127886 [3] NCCL INFO Channel 00/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127293:127886 [3] NCCL INFO Channel 01/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127293:127886 [3] NCCL INFO Channel 02/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127293:127886 [3] NCCL INFO Channel 03/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762818:763479 [3] NCCL INFO Channel 00/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762818:763479 [3] NCCL INFO Channel 01/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762818:763479 [3] NCCL INFO Channel 02/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762818:763479 [3] NCCL INFO Channel 03/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762818:763479 [3] NCCL INFO Channel 00/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762818:763479 [3] NCCL INFO Channel 01/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762818:763479 [3] NCCL INFO Channel 02/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762818:763479 [3] NCCL INFO Channel 03/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127292:127887 [2] NCCL INFO Channel 00/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127292:127887 [2] NCCL INFO Channel 01/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127292:127887 [2] NCCL INFO Channel 02/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127292:127887 [2] NCCL INFO Channel 03/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127292:127887 [2] NCCL INFO Channel 00/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127292:127887 [2] NCCL INFO Channel 01/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127292:127887 [2] NCCL INFO Channel 02/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127292:127887 [2] NCCL INFO Channel 03/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762817:763478 [2] NCCL INFO Channel 00/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762817:763478 [2] NCCL INFO Channel 01/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762817:763478 [2] NCCL INFO Channel 02/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762817:763478 [2] NCCL INFO Channel 03/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762817:763478 [2] NCCL INFO Channel 00/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762817:763478 [2] NCCL INFO Channel 01/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762817:763478 [2] NCCL INFO Channel 02/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762817:763478 [2] NCCL INFO Channel 03/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762822:763483 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,fff00000,00000000
+ip-26-0-166-125:762821:763481 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,fff00000,00000000
+ip-26-0-166-15:127297:127890 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-166-125:762822:763483 [7] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-166-125:762822:763483 [7] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-166-125:762822:763483 [7] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-166-125:762822:763483 [7] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-166-15:127297:127890 [7] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-166-15:127297:127890 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-125:762822:763483 [7] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-166-125:762822:763483 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-15:127296:127889 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-166-15:127296:127889 [6] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-166-15:127296:127889 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-125:762821:763481 [6] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-166-125:762821:763481 [6] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-166-125:762821:763481 [6] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-166-125:762821:763481 [6] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-166-125:762821:763481 [6] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-166-125:762821:763481 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-125:762819:763461 [4] NCCL INFO Connected all rings
+ip-26-0-166-125:762819:763461 [4] NCCL INFO Connected all trees
+ip-26-0-166-125:762819:763461 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-125:762819:763461 [4] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-125:762819:763461 [4] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-125:762819:763461 [4] NCCL INFO comm 0x14238520 rank 0 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0xdbf8660b36dca753 - Init COMPLETE
+ip-26-0-166-125:762819:763493 [4] NCCL INFO Channel 02/1 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA/Shared
+ip-26-0-166-125:762819:763493 [4] NCCL INFO Channel 03/1 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA/Shared
+ip-26-0-166-15:127294:127875 [4] NCCL INFO Connected all rings
+ip-26-0-166-15:127294:127875 [4] NCCL INFO Connected all trees
+ip-26-0-166-15:127294:127875 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-15:127294:127875 [4] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-15:127294:127875 [4] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-15:127294:127875 [4] NCCL INFO comm 0xb2f8fe0 rank 1 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0xdbf8660b36dca753 - Init COMPLETE
+ip-26-0-166-15:127294:127900 [4] NCCL INFO Channel 02/1 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA/Shared
+ip-26-0-166-15:127294:127900 [4] NCCL INFO Channel 03/1 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA/Shared
+ip-26-0-166-15:127297:127890 [7] NCCL INFO Channel 00/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127297:127890 [7] NCCL INFO Channel 01/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127297:127890 [7] NCCL INFO Channel 02/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127297:127890 [7] NCCL INFO Channel 03/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127297:127890 [7] NCCL INFO Channel 00/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127297:127890 [7] NCCL INFO Channel 01/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127297:127890 [7] NCCL INFO Channel 02/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127297:127890 [7] NCCL INFO Channel 03/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762820:763462 [5] NCCL INFO Connected all rings
+ip-26-0-166-125:762820:763462 [5] NCCL INFO Connected all trees
+ip-26-0-166-125:762820:763462 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-125:762820:763462 [5] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-125:762820:763462 [5] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-125:762822:763483 [7] NCCL INFO Channel 00/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762822:763483 [7] NCCL INFO Channel 01/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762822:763483 [7] NCCL INFO Channel 02/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762822:763483 [7] NCCL INFO Channel 03/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762822:763483 [7] NCCL INFO Channel 00/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762822:763483 [7] NCCL INFO Channel 01/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762822:763483 [7] NCCL INFO Channel 02/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762822:763483 [7] NCCL INFO Channel 03/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127296:127889 [6] NCCL INFO Channel 00/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127296:127889 [6] NCCL INFO Channel 01/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127296:127889 [6] NCCL INFO Channel 02/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127296:127889 [6] NCCL INFO Channel 03/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127296:127889 [6] NCCL INFO Channel 00/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127296:127889 [6] NCCL INFO Channel 01/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127296:127889 [6] NCCL INFO Channel 02/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127296:127889 [6] NCCL INFO Channel 03/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762821:763481 [6] NCCL INFO Channel 00/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762821:763481 [6] NCCL INFO Channel 01/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762821:763481 [6] NCCL INFO Channel 02/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762821:763481 [6] NCCL INFO Channel 03/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762821:763481 [6] NCCL INFO Channel 00/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762821:763481 [6] NCCL INFO Channel 01/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762821:763481 [6] NCCL INFO Channel 02/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762821:763481 [6] NCCL INFO Channel 03/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762820:763462 [5] NCCL INFO comm 0x14159ec0 rank 0 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0x79de6ffa55f6be9d - Init COMPLETE
+ip-26-0-166-125:762820:763494 [5] NCCL INFO Channel 02/1 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA/Shared
+ip-26-0-166-125:762820:763494 [5] NCCL INFO Channel 03/1 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA/Shared
+ip-26-0-166-125:762815:763467 [0] NCCL INFO Connected all rings
+ip-26-0-166-125:762815:763467 [0] NCCL INFO Connected all trees
+ip-26-0-166-125:762815:763467 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-125:762815:763467 [0] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-125:762815:763467 [0] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-125:762816:763465 [1] NCCL INFO Connected all rings
+ip-26-0-166-125:762816:763465 [1] NCCL INFO Connected all trees
+ip-26-0-166-125:762816:763465 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-125:762816:763465 [1] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-125:762816:763465 [1] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-15:127295:127874 [5] NCCL INFO Connected all rings
+ip-26-0-166-15:127295:127874 [5] NCCL INFO Connected all trees
+ip-26-0-166-15:127295:127874 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-15:127295:127874 [5] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-15:127295:127874 [5] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-125:762815:763467 [0] NCCL INFO comm 0x13425100 rank 0 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0x169415bf4ad33309 - Init COMPLETE
+ip-26-0-166-125:762816:763465 [1] NCCL INFO comm 0x14456100 rank 0 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0x394ef85110bc3661 - Init COMPLETE
+ip-26-0-166-125:762816:763497 [1] NCCL INFO Channel 02/1 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA/Shared
+ip-26-0-166-125:762815:763496 [0] NCCL INFO Channel 02/1 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA/Shared
+ip-26-0-166-125:762816:763497 [1] NCCL INFO Channel 03/1 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA/Shared
+ip-26-0-166-125:762815:763496 [0] NCCL INFO Channel 03/1 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA/Shared
+ip-26-0-166-15:127295:127874 [5] NCCL INFO comm 0xae0ea00 rank 1 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0x79de6ffa55f6be9d - Init COMPLETE
+ip-26-0-166-15:127295:127901 [5] NCCL INFO Channel 02/1 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA/Shared
+ip-26-0-166-15:127295:127901 [5] NCCL INFO Channel 03/1 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA/Shared
+ip-26-0-166-15:127290:127877 [0] NCCL INFO Connected all rings
+ip-26-0-166-15:127290:127877 [0] NCCL INFO Connected all trees
+ip-26-0-166-15:127290:127877 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-15:127290:127877 [0] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-15:127290:127877 [0] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-15:127291:127876 [1] NCCL INFO Connected all rings
+ip-26-0-166-15:127291:127876 [1] NCCL INFO Connected all trees
+ip-26-0-166-15:127291:127876 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-15:127291:127876 [1] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-15:127291:127876 [1] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-15:127290:127877 [0] NCCL INFO comm 0xaf28780 rank 1 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0x169415bf4ad33309 - Init COMPLETE
+ip-26-0-166-15:127290:127902 [0] NCCL INFO Channel 02/1 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA/Shared
+ip-26-0-166-15:127290:127902 [0] NCCL INFO Channel 03/1 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA/Shared
+ip-26-0-166-15:127291:127876 [1] NCCL INFO comm 0xc5da7a0 rank 1 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0x394ef85110bc3661 - Init COMPLETE
+ip-26-0-166-15:127291:127903 [1] NCCL INFO Channel 02/1 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA/Shared
+ip-26-0-166-15:127291:127903 [1] NCCL INFO Channel 03/1 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA/Shared
+ip-26-0-166-125:762818:763479 [3] NCCL INFO Connected all rings
+ip-26-0-166-125:762818:763479 [3] NCCL INFO Connected all trees
+ip-26-0-166-125:762818:763479 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-125:762818:763479 [3] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-125:762818:763479 [3] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-125:762817:763478 [2] NCCL INFO Connected all rings
+ip-26-0-166-125:762817:763478 [2] NCCL INFO Connected all trees
+ip-26-0-166-125:762817:763478 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-125:762817:763478 [2] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-125:762817:763478 [2] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-125:762818:763479 [3] NCCL INFO comm 0x1358f380 rank 0 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0x802584e5feb4e98a - Init COMPLETE
+ip-26-0-166-125:762818:763498 [3] NCCL INFO Channel 02/1 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA/Shared
+ip-26-0-166-125:762818:763498 [3] NCCL INFO Channel 03/1 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA/Shared
+ip-26-0-166-15:127293:127886 [3] NCCL INFO Connected all rings
+ip-26-0-166-15:127293:127886 [3] NCCL INFO Connected all trees
+ip-26-0-166-15:127293:127886 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-15:127293:127886 [3] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-15:127293:127886 [3] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-15:127292:127887 [2] NCCL INFO Connected all rings
+ip-26-0-166-15:127292:127887 [2] NCCL INFO Connected all trees
+ip-26-0-166-15:127292:127887 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-15:127292:127887 [2] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-15:127292:127887 [2] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-125:762817:763478 [2] NCCL INFO comm 0x15252410 rank 0 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0x94d6c206686962eb - Init COMPLETE
+ip-26-0-166-125:762817:763499 [2] NCCL INFO Channel 02/1 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA/Shared
+ip-26-0-166-125:762817:763499 [2] NCCL INFO Channel 03/1 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA/Shared
+ip-26-0-166-15:127292:127887 [2] NCCL INFO comm 0xb57e240 rank 1 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0x94d6c206686962eb - Init COMPLETE
+ip-26-0-166-15:127292:127904 [2] NCCL INFO Channel 02/1 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA/Shared
+ip-26-0-166-15:127292:127904 [2] NCCL INFO Channel 03/1 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA/Shared
+ip-26-0-166-15:127293:127886 [3] NCCL INFO comm 0xc47f230 rank 1 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0x802584e5feb4e98a - Init COMPLETE
+ip-26-0-166-15:127293:127905 [3] NCCL INFO Channel 02/1 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA/Shared
+ip-26-0-166-15:127293:127905 [3] NCCL INFO Channel 03/1 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA/Shared
+ip-26-0-166-15:127297:127890 [7] NCCL INFO Connected all rings
+ip-26-0-166-15:127297:127890 [7] NCCL INFO Connected all trees
+ip-26-0-166-15:127297:127890 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-15:127297:127890 [7] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-15:127297:127890 [7] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-15:127297:127890 [7] NCCL INFO comm 0xc13e470 rank 1 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0xdfcf3744600a38a8 - Init COMPLETE
+ip-26-0-166-15:127297:127906 [7] NCCL INFO Channel 02/1 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA/Shared
+ip-26-0-166-15:127297:127906 [7] NCCL INFO Channel 03/1 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA/Shared
+ip-26-0-166-125:762822:763483 [7] NCCL INFO Connected all rings
+ip-26-0-166-125:762822:763483 [7] NCCL INFO Connected all trees
+ip-26-0-166-125:762822:763483 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-125:762822:763483 [7] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-125:762822:763483 [7] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-15:127296:127889 [6] NCCL INFO Connected all rings
+ip-26-0-166-15:127296:127889 [6] NCCL INFO Connected all trees
+ip-26-0-166-15:127296:127889 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-15:127296:127889 [6] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-15:127296:127889 [6] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-125:762822:763483 [7] NCCL INFO comm 0x14e365e0 rank 0 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0xdfcf3744600a38a8 - Init COMPLETE
+ip-26-0-166-15:127296:127889 [6] NCCL INFO comm 0xba205a0 rank 1 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0xb814955b7173143c - Init COMPLETE
+ip-26-0-166-125:762822:763500 [7] NCCL INFO Channel 02/1 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA/Shared
+ip-26-0-166-125:762822:763500 [7] NCCL INFO Channel 03/1 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA/Shared
+ip-26-0-166-125:762821:763481 [6] NCCL INFO Connected all rings
+ip-26-0-166-125:762821:763481 [6] NCCL INFO Connected all trees
+ip-26-0-166-125:762821:763481 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-125:762821:763481 [6] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-125:762821:763481 [6] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-15:127296:127907 [6] NCCL INFO Channel 02/1 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA/Shared
+ip-26-0-166-15:127296:127907 [6] NCCL INFO Channel 03/1 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA/Shared
+ip-26-0-166-125:762821:763481 [6] NCCL INFO comm 0x136fcea0 rank 0 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0xb814955b7173143c - Init COMPLETE
+ip-26-0-166-125:762821:763501 [6] NCCL INFO Channel 02/1 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA/Shared
+ip-26-0-166-125:762821:763501 [6] NCCL INFO Channel 03/1 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA/Shared
+ip-26-0-166-125:762819:763518 [4] NCCL INFO Channel 02/1 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA/Shared
+ip-26-0-166-125:762819:763518 [4] NCCL INFO Channel 03/1 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA/Shared
+ip-26-0-166-125:762820:763519 [5] NCCL INFO Channel 02/1 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA/Shared
+ip-26-0-166-125:762820:763519 [5] NCCL INFO Channel 03/1 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA/Shared
+ip-26-0-166-125:762818:763536 [3] NCCL INFO Channel 02/1 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA/Shared
+ip-26-0-166-125:762817:763537 [2] NCCL INFO Channel 02/1 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA/Shared
+ip-26-0-166-125:762818:763536 [3] NCCL INFO Channel 03/1 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA/Shared
+ip-26-0-166-125:762817:763537 [2] NCCL INFO Channel 03/1 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA/Shared
+ip-26-0-166-125:762816:763554 [1] NCCL INFO Channel 02/1 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA/Shared
+ip-26-0-166-125:762816:763554 [1] NCCL INFO Channel 03/1 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA/Shared
+ip-26-0-166-125:762815:763555 [0] NCCL INFO Channel 02/1 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA/Shared
+ip-26-0-166-125:762815:763555 [0] NCCL INFO Channel 03/1 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA/Shared
+ip-26-0-166-125:762822:763572 [7] NCCL INFO Channel 02/1 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA/Shared
+ip-26-0-166-125:762822:763572 [7] NCCL INFO Channel 03/1 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA/Shared
+ip-26-0-166-125:762821:763573 [6] NCCL INFO Channel 02/1 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA/Shared
+ip-26-0-166-125:762821:763573 [6] NCCL INFO Channel 03/1 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA/Shared
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+NCCL version 2.18.6+cuda12.1
+ip-26-0-166-15:127295:127911 [5] NCCL INFO Using network Libfabric
+ip-26-0-166-15:127294:127912 [4] NCCL INFO Using network Libfabric
+ip-26-0-166-214:624948:625553 [5] NCCL INFO Using network Libfabric
+ip-26-0-166-214:624947:625552 [4] NCCL INFO Using network Libfabric
+ip-26-0-166-214:624947:625552 [4] NCCL INFO comm 0xc5567b0 rank 1 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0x60a7ae1b36b9c99e - Init START
+ip-26-0-166-15:127294:127912 [4] NCCL INFO comm 0x13dff710 rank 0 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0x60a7ae1b36b9c99e - Init START
+ip-26-0-166-214:624947:625552 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-15:127294:127912 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-15:127295:127911 [5] NCCL INFO comm 0x138f0950 rank 0 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0xe781f047205b9a46 - Init START
+ip-26-0-166-214:624948:625553 [5] NCCL INFO comm 0xc3bb5c0 rank 1 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0xe781f047205b9a46 - Init START
+ip-26-0-166-214:624948:625553 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-15:127295:127911 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+ip-26-0-166-15:127290:127915 [0] NCCL INFO Using network Libfabric
+ip-26-0-166-15:127291:127916 [1] NCCL INFO Using network Libfabric
+ip-26-0-166-214:624944:625554 [1] NCCL INFO Using network Libfabric
+ip-26-0-166-214:624943:625555 [0] NCCL INFO Using network Libfabric
+ip-26-0-166-214:624944:625554 [1] NCCL INFO comm 0xaa81800 rank 1 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0xc82e6eb2c39465ca - Init START
+ip-26-0-166-15:127291:127916 [1] NCCL INFO comm 0x15074e10 rank 0 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0xc82e6eb2c39465ca - Init START
+ip-26-0-166-214:624944:625554 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-15:127291:127916 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-214:624943:625555 [0] NCCL INFO comm 0xb7a7340 rank 1 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0x8ebf9cd225756019 - Init START
+ip-26-0-166-15:127290:127915 [0] NCCL INFO comm 0x137f8350 rank 0 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0x8ebf9cd225756019 - Init START
+ip-26-0-166-15:127290:127915 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-214:624943:625555 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-15:127294:127912 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-166-214:624947:625552 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-166-214:624947:625552 [4] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-166-214:624947:625552 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-15:127294:127912 [4] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-166-15:127294:127912 [4] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-166-15:127294:127912 [4] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-166-15:127294:127912 [4] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-166-15:127294:127912 [4] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-166-15:127294:127912 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-214:624948:625553 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-166-214:624943:625555 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-166-214:624944:625554 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-166-15:127295:127911 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-166-15:127295:127911 [5] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-166-15:127295:127911 [5] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-166-214:624948:625553 [5] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-166-214:624948:625553 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-15:127295:127911 [5] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-166-15:127295:127911 [5] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-166-15:127295:127911 [5] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-166-15:127295:127911 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-15:127290:127915 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-166-15:127290:127915 [0] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-166-15:127290:127915 [0] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-166-15:127290:127915 [0] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-166-15:127290:127915 [0] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-166-15:127290:127915 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-166-15:127290:127915 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-214:624943:625555 [0] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-166-214:624943:625555 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-15:127291:127916 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-166-214:624944:625554 [1] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-166-214:624944:625554 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-15:127291:127916 [1] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-166-15:127291:127916 [1] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-166-15:127291:127916 [1] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-166-15:127291:127916 [1] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-166-15:127291:127916 [1] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-166-15:127291:127916 [1] NCCL INFO P2P Chunksize set to 131072
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+NCCL version 2.18.6+cuda12.1
+ip-26-0-166-15:127292:127927 [2] NCCL INFO Using network Libfabric
+ip-26-0-166-15:127293:127928 [3] NCCL INFO Using network Libfabric
+ip-26-0-166-214:624945:625565 [2] NCCL INFO Using network Libfabric
+ip-26-0-166-214:624946:625566 [3] NCCL INFO Using network Libfabric
+ip-26-0-166-214:624945:625565 [2] NCCL INFO comm 0xbb57610 rank 1 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0xca0c852d0b45451d - Init START
+ip-26-0-166-15:127292:127927 [2] NCCL INFO comm 0x1405ed40 rank 0 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0xca0c852d0b45451d - Init START
+ip-26-0-166-214:624945:625565 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-15:127292:127927 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-214:624946:625566 [3] NCCL INFO comm 0xc3aae40 rank 1 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0x1682e5293ebcf526 - Init START
+ip-26-0-166-15:127293:127928 [3] NCCL INFO comm 0x14f60e00 rank 0 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0x1682e5293ebcf526 - Init START
+ip-26-0-166-214:624946:625566 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-15:127293:127928 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-15:127294:127912 [4] NCCL INFO Channel 00/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127294:127912 [4] NCCL INFO Channel 01/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127294:127912 [4] NCCL INFO Channel 02/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127294:127912 [4] NCCL INFO Channel 03/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127294:127912 [4] NCCL INFO Channel 00/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127294:127912 [4] NCCL INFO Channel 01/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127294:127912 [4] NCCL INFO Channel 02/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127294:127912 [4] NCCL INFO Channel 03/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624947:625552 [4] NCCL INFO Channel 00/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624947:625552 [4] NCCL INFO Channel 01/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624947:625552 [4] NCCL INFO Channel 02/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624947:625552 [4] NCCL INFO Channel 03/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624947:625552 [4] NCCL INFO Channel 00/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624947:625552 [4] NCCL INFO Channel 01/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624947:625552 [4] NCCL INFO Channel 02/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624947:625552 [4] NCCL INFO Channel 03/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624943:625555 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127295:127911 [5] NCCL INFO Channel 00/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624943:625555 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127295:127911 [5] NCCL INFO Channel 01/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624943:625555 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127295:127911 [5] NCCL INFO Channel 02/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624943:625555 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127295:127911 [5] NCCL INFO Channel 03/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624943:625555 [0] NCCL INFO Channel 00/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127295:127911 [5] NCCL INFO Channel 00/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624943:625555 [0] NCCL INFO Channel 01/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127295:127911 [5] NCCL INFO Channel 01/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624943:625555 [0] NCCL INFO Channel 02/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127295:127911 [5] NCCL INFO Channel 02/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624943:625555 [0] NCCL INFO Channel 03/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127295:127911 [5] NCCL INFO Channel 03/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624948:625553 [5] NCCL INFO Channel 00/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624948:625553 [5] NCCL INFO Channel 01/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624948:625553 [5] NCCL INFO Channel 02/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624948:625553 [5] NCCL INFO Channel 03/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624948:625553 [5] NCCL INFO Channel 00/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624948:625553 [5] NCCL INFO Channel 01/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624948:625553 [5] NCCL INFO Channel 02/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624948:625553 [5] NCCL INFO Channel 03/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127290:127915 [0] NCCL INFO Channel 00/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127290:127915 [0] NCCL INFO Channel 01/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624944:625554 [1] NCCL INFO Channel 00/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127290:127915 [0] NCCL INFO Channel 02/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624944:625554 [1] NCCL INFO Channel 01/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127290:127915 [0] NCCL INFO Channel 03/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624944:625554 [1] NCCL INFO Channel 02/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127290:127915 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624944:625554 [1] NCCL INFO Channel 03/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127290:127915 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624944:625554 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127290:127915 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624944:625554 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127290:127915 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624944:625554 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624944:625554 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127291:127916 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127291:127916 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127291:127916 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127291:127916 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127291:127916 [1] NCCL INFO Channel 00/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127291:127916 [1] NCCL INFO Channel 01/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127291:127916 [1] NCCL INFO Channel 02/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127291:127916 [1] NCCL INFO Channel 03/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624946:625566 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-166-214:624945:625565 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+NCCL version 2.18.6+cuda12.1
+ip-26-0-166-15:127297:127932 [7] NCCL INFO Using network Libfabric
+ip-26-0-166-214:624950:625568 [7] NCCL INFO Using network Libfabric
+ip-26-0-166-15:127292:127927 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-166-214:624949:625569 [6] NCCL INFO Using network Libfabric
+ip-26-0-166-15:127296:127931 [6] NCCL INFO Using network Libfabric
+ip-26-0-166-214:624950:625568 [7] NCCL INFO comm 0xb0d7bb0 rank 1 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0x8f40e652d823a866 - Init START
+ip-26-0-166-15:127297:127932 [7] NCCL INFO comm 0x14c1d310 rank 0 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0x8f40e652d823a866 - Init START
+ip-26-0-166-214:624950:625568 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-15:127297:127932 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-214:624949:625569 [6] NCCL INFO comm 0xadd8040 rank 1 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0x15021c49659a8335 - Init START
+ip-26-0-166-15:127296:127931 [6] NCCL INFO comm 0x14503600 rank 0 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0x15021c49659a8335 - Init START
+ip-26-0-166-214:624949:625569 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-15:127296:127931 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-15:127292:127927 [2] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-166-15:127292:127927 [2] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-166-15:127292:127927 [2] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-166-15:127292:127927 [2] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-166-15:127292:127927 [2] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-166-15:127292:127927 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-214:624945:625565 [2] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-166-214:624945:625565 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-15:127293:127928 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-166-15:127293:127928 [3] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-166-15:127293:127928 [3] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-166-15:127293:127928 [3] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-166-15:127293:127928 [3] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-166-15:127293:127928 [3] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-166-15:127293:127928 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-214:624946:625566 [3] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-166-214:624946:625566 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-214:624945:625565 [2] NCCL INFO Channel 00/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624945:625565 [2] NCCL INFO Channel 01/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624945:625565 [2] NCCL INFO Channel 02/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624945:625565 [2] NCCL INFO Channel 03/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624945:625565 [2] NCCL INFO Channel 00/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624945:625565 [2] NCCL INFO Channel 01/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624945:625565 [2] NCCL INFO Channel 02/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624945:625565 [2] NCCL INFO Channel 03/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127292:127927 [2] NCCL INFO Channel 00/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127292:127927 [2] NCCL INFO Channel 01/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127292:127927 [2] NCCL INFO Channel 02/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127292:127927 [2] NCCL INFO Channel 03/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127292:127927 [2] NCCL INFO Channel 00/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127292:127927 [2] NCCL INFO Channel 01/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127292:127927 [2] NCCL INFO Channel 02/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127292:127927 [2] NCCL INFO Channel 03/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624946:625566 [3] NCCL INFO Channel 00/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624946:625566 [3] NCCL INFO Channel 01/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127293:127928 [3] NCCL INFO Channel 00/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624946:625566 [3] NCCL INFO Channel 02/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624946:625566 [3] NCCL INFO Channel 03/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127293:127928 [3] NCCL INFO Channel 01/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624946:625566 [3] NCCL INFO Channel 00/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127293:127928 [3] NCCL INFO Channel 02/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624946:625566 [3] NCCL INFO Channel 01/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127293:127928 [3] NCCL INFO Channel 03/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624946:625566 [3] NCCL INFO Channel 02/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127293:127928 [3] NCCL INFO Channel 00/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624946:625566 [3] NCCL INFO Channel 03/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127293:127928 [3] NCCL INFO Channel 01/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127293:127928 [3] NCCL INFO Channel 02/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127293:127928 [3] NCCL INFO Channel 03/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624949:625569 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-166-15:127297:127932 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-166-214:624950:625568 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-166-214:624950:625568 [7] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-166-214:624950:625568 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-15:127297:127932 [7] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-166-15:127297:127932 [7] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-166-15:127297:127932 [7] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-166-15:127297:127932 [7] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-166-15:127297:127932 [7] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-166-15:127297:127932 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-15:127296:127931 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-166-15:127296:127931 [6] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-166-15:127296:127931 [6] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-166-15:127296:127931 [6] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-166-15:127296:127931 [6] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-166-15:127296:127931 [6] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-166-214:624949:625569 [6] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-166-214:624949:625569 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-15:127296:127931 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-214:624950:625568 [7] NCCL INFO Channel 00/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624950:625568 [7] NCCL INFO Channel 01/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624950:625568 [7] NCCL INFO Channel 02/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624950:625568 [7] NCCL INFO Channel 03/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624950:625568 [7] NCCL INFO Channel 00/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624950:625568 [7] NCCL INFO Channel 01/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624950:625568 [7] NCCL INFO Channel 02/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624950:625568 [7] NCCL INFO Channel 03/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127297:127932 [7] NCCL INFO Channel 00/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127297:127932 [7] NCCL INFO Channel 01/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127297:127932 [7] NCCL INFO Channel 02/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127297:127932 [7] NCCL INFO Channel 03/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127297:127932 [7] NCCL INFO Channel 00/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127297:127932 [7] NCCL INFO Channel 01/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127297:127932 [7] NCCL INFO Channel 02/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127297:127932 [7] NCCL INFO Channel 03/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624949:625569 [6] NCCL INFO Channel 00/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624949:625569 [6] NCCL INFO Channel 01/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624949:625569 [6] NCCL INFO Channel 02/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624949:625569 [6] NCCL INFO Channel 03/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624949:625569 [6] NCCL INFO Channel 00/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624949:625569 [6] NCCL INFO Channel 01/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624949:625569 [6] NCCL INFO Channel 02/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624949:625569 [6] NCCL INFO Channel 03/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127296:127931 [6] NCCL INFO Channel 00/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127296:127931 [6] NCCL INFO Channel 01/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127296:127931 [6] NCCL INFO Channel 02/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127296:127931 [6] NCCL INFO Channel 03/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127296:127931 [6] NCCL INFO Channel 00/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127296:127931 [6] NCCL INFO Channel 01/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127296:127931 [6] NCCL INFO Channel 02/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127296:127931 [6] NCCL INFO Channel 03/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624944:625554 [1] NCCL INFO Connected all rings
+ip-26-0-166-214:624944:625554 [1] NCCL INFO Connected all trees
+ip-26-0-166-214:624944:625554 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-214:624944:625554 [1] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-214:624944:625554 [1] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-214:624943:625555 [0] NCCL INFO Connected all rings
+ip-26-0-166-214:624943:625555 [0] NCCL INFO Connected all trees
+ip-26-0-166-214:624943:625555 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-214:624943:625555 [0] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-214:624943:625555 [0] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-214:624948:625553 [5] NCCL INFO Connected all rings
+ip-26-0-166-214:624948:625553 [5] NCCL INFO Connected all trees
+ip-26-0-166-214:624948:625553 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-214:624948:625553 [5] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-214:624948:625553 [5] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-214:624944:625554 [1] NCCL INFO comm 0xaa81800 rank 1 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0xc82e6eb2c39465ca - Init COMPLETE
+ip-26-0-166-214:624944:625578 [1] NCCL INFO Channel 02/1 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA/Shared
+ip-26-0-166-214:624944:625578 [1] NCCL INFO Channel 03/1 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA/Shared
+ip-26-0-166-214:624947:625552 [4] NCCL INFO Connected all rings
+ip-26-0-166-214:624947:625552 [4] NCCL INFO Connected all trees
+ip-26-0-166-214:624947:625552 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-214:624947:625552 [4] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-214:624947:625552 [4] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-214:624948:625553 [5] NCCL INFO comm 0xc3bb5c0 rank 1 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0xe781f047205b9a46 - Init COMPLETE
+ip-26-0-166-214:624943:625555 [0] NCCL INFO comm 0xb7a7340 rank 1 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0x8ebf9cd225756019 - Init COMPLETE
+ip-26-0-166-214:624948:625579 [5] NCCL INFO Channel 02/1 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA/Shared
+ip-26-0-166-214:624948:625579 [5] NCCL INFO Channel 03/1 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA/Shared
+ip-26-0-166-214:624943:625580 [0] NCCL INFO Channel 02/1 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA/Shared
+ip-26-0-166-214:624943:625580 [0] NCCL INFO Channel 03/1 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA/Shared
+ip-26-0-166-15:127294:127912 [4] NCCL INFO Connected all rings
+ip-26-0-166-15:127294:127912 [4] NCCL INFO Connected all trees
+ip-26-0-166-15:127294:127912 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-15:127294:127912 [4] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-15:127294:127912 [4] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-15:127291:127916 [1] NCCL INFO Connected all rings
+ip-26-0-166-15:127291:127916 [1] NCCL INFO Connected all trees
+ip-26-0-166-15:127291:127916 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-15:127291:127916 [1] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-15:127291:127916 [1] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-214:624947:625552 [4] NCCL INFO comm 0xc5567b0 rank 1 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0x60a7ae1b36b9c99e - Init COMPLETE
+ip-26-0-166-214:624947:625581 [4] NCCL INFO Channel 02/1 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA/Shared
+ip-26-0-166-214:624947:625581 [4] NCCL INFO Channel 03/1 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA/Shared
+ip-26-0-166-15:127290:127915 [0] NCCL INFO Connected all rings
+ip-26-0-166-15:127290:127915 [0] NCCL INFO Connected all trees
+ip-26-0-166-15:127290:127915 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-15:127290:127915 [0] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-15:127290:127915 [0] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-15:127294:127912 [4] NCCL INFO comm 0x13dff710 rank 0 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0x60a7ae1b36b9c99e - Init COMPLETE
+ip-26-0-166-15:127294:127942 [4] NCCL INFO Channel 02/1 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA/Shared
+ip-26-0-166-15:127294:127942 [4] NCCL INFO Channel 03/1 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA/Shared
+ip-26-0-166-15:127295:127911 [5] NCCL INFO Connected all rings
+ip-26-0-166-15:127295:127911 [5] NCCL INFO Connected all trees
+ip-26-0-166-15:127295:127911 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-15:127295:127911 [5] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-15:127295:127911 [5] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-15:127291:127916 [1] NCCL INFO comm 0x15074e10 rank 0 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0xc82e6eb2c39465ca - Init COMPLETE
+ip-26-0-166-15:127291:127943 [1] NCCL INFO Channel 02/1 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA/Shared
+ip-26-0-166-15:127291:127943 [1] NCCL INFO Channel 03/1 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA/Shared
+ip-26-0-166-15:127290:127915 [0] NCCL INFO comm 0x137f8350 rank 0 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0x8ebf9cd225756019 - Init COMPLETE
+ip-26-0-166-214:624945:625565 [2] NCCL INFO Connected all rings
+ip-26-0-166-214:624945:625565 [2] NCCL INFO Connected all trees
+ip-26-0-166-214:624945:625565 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-214:624945:625565 [2] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-214:624945:625565 [2] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-15:127290:127944 [0] NCCL INFO Channel 02/1 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA/Shared
+ip-26-0-166-15:127290:127944 [0] NCCL INFO Channel 03/1 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA/Shared
+ip-26-0-166-214:624946:625566 [3] NCCL INFO Connected all rings
+ip-26-0-166-214:624946:625566 [3] NCCL INFO Connected all trees
+ip-26-0-166-214:624946:625566 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-214:624946:625566 [3] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-214:624946:625566 [3] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-15:127295:127911 [5] NCCL INFO comm 0x138f0950 rank 0 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0xe781f047205b9a46 - Init COMPLETE
+ip-26-0-166-15:127295:127945 [5] NCCL INFO Channel 02/1 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA/Shared
+ip-26-0-166-15:127295:127945 [5] NCCL INFO Channel 03/1 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA/Shared
+ip-26-0-166-214:624945:625565 [2] NCCL INFO comm 0xbb57610 rank 1 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0xca0c852d0b45451d - Init COMPLETE
+ip-26-0-166-214:624945:625582 [2] NCCL INFO Channel 02/1 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA/Shared
+ip-26-0-166-214:624945:625582 [2] NCCL INFO Channel 03/1 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA/Shared
+ip-26-0-166-214:624946:625566 [3] NCCL INFO comm 0xc3aae40 rank 1 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0x1682e5293ebcf526 - Init COMPLETE
+ip-26-0-166-214:624946:625583 [3] NCCL INFO Channel 02/1 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA/Shared
+ip-26-0-166-214:624946:625583 [3] NCCL INFO Channel 03/1 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA/Shared
+ip-26-0-166-15:127292:127927 [2] NCCL INFO Connected all rings
+ip-26-0-166-15:127292:127927 [2] NCCL INFO Connected all trees
+ip-26-0-166-15:127292:127927 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-15:127292:127927 [2] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-15:127292:127927 [2] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-15:127293:127928 [3] NCCL INFO Connected all rings
+ip-26-0-166-15:127293:127928 [3] NCCL INFO Connected all trees
+ip-26-0-166-15:127293:127928 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-15:127293:127928 [3] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-15:127293:127928 [3] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-15:127292:127927 [2] NCCL INFO comm 0x1405ed40 rank 0 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0xca0c852d0b45451d - Init COMPLETE
+ip-26-0-166-15:127292:127946 [2] NCCL INFO Channel 02/1 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA/Shared
+ip-26-0-166-15:127292:127946 [2] NCCL INFO Channel 03/1 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA/Shared
+ip-26-0-166-15:127293:127928 [3] NCCL INFO comm 0x14f60e00 rank 0 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0x1682e5293ebcf526 - Init COMPLETE
+ip-26-0-166-15:127293:127947 [3] NCCL INFO Channel 02/1 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA/Shared
+ip-26-0-166-15:127293:127947 [3] NCCL INFO Channel 03/1 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA/Shared
+ip-26-0-166-214:624950:625568 [7] NCCL INFO Connected all rings
+ip-26-0-166-214:624950:625568 [7] NCCL INFO Connected all trees
+ip-26-0-166-214:624950:625568 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-214:624950:625568 [7] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-214:624950:625568 [7] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-214:624949:625569 [6] NCCL INFO Connected all rings
+ip-26-0-166-214:624949:625569 [6] NCCL INFO Connected all trees
+ip-26-0-166-214:624949:625569 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-214:624949:625569 [6] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-214:624949:625569 [6] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-214:624950:625568 [7] NCCL INFO comm 0xb0d7bb0 rank 1 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0x8f40e652d823a866 - Init COMPLETE
+ip-26-0-166-214:624950:625584 [7] NCCL INFO Channel 02/1 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA/Shared
+ip-26-0-166-214:624950:625584 [7] NCCL INFO Channel 03/1 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA/Shared
+ip-26-0-166-214:624949:625569 [6] NCCL INFO comm 0xadd8040 rank 1 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0x15021c49659a8335 - Init COMPLETE
+ip-26-0-166-214:624949:625585 [6] NCCL INFO Channel 02/1 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA/Shared
+ip-26-0-166-214:624949:625585 [6] NCCL INFO Channel 03/1 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA/Shared
+ip-26-0-166-15:127297:127932 [7] NCCL INFO Connected all rings
+ip-26-0-166-15:127297:127932 [7] NCCL INFO Connected all trees
+ip-26-0-166-15:127297:127932 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-15:127297:127932 [7] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-15:127297:127932 [7] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-15:127296:127931 [6] NCCL INFO Connected all rings
+ip-26-0-166-15:127296:127931 [6] NCCL INFO Connected all trees
+ip-26-0-166-15:127296:127931 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-15:127296:127931 [6] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-15:127296:127931 [6] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-15:127297:127932 [7] NCCL INFO comm 0x14c1d310 rank 0 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0x8f40e652d823a866 - Init COMPLETE
+ip-26-0-166-15:127297:127948 [7] NCCL INFO Channel 02/1 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA/Shared
+ip-26-0-166-15:127297:127948 [7] NCCL INFO Channel 03/1 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA/Shared
+ip-26-0-166-15:127296:127931 [6] NCCL INFO comm 0x14503600 rank 0 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0x15021c49659a8335 - Init COMPLETE
+ip-26-0-166-15:127296:127949 [6] NCCL INFO Channel 02/1 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA/Shared
+ip-26-0-166-15:127296:127949 [6] NCCL INFO Channel 03/1 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA/Shared
+ip-26-0-166-15:127291:127966 [1] NCCL INFO Channel 02/1 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA/Shared
+ip-26-0-166-15:127291:127966 [1] NCCL INFO Channel 03/1 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA/Shared
+ip-26-0-166-15:127290:127967 [0] NCCL INFO Channel 02/1 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA/Shared
+ip-26-0-166-15:127290:127967 [0] NCCL INFO Channel 03/1 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA/Shared
+ip-26-0-166-15:127294:127984 [4] NCCL INFO Channel 02/1 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA/Shared
+ip-26-0-166-15:127294:127984 [4] NCCL INFO Channel 03/1 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA/Shared
+ip-26-0-166-15:127295:127985 [5] NCCL INFO Channel 02/1 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA/Shared
+ip-26-0-166-15:127295:127985 [5] NCCL INFO Channel 03/1 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA/Shared
+ip-26-0-166-15:127293:128003 [3] NCCL INFO Channel 02/1 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA/Shared
+ip-26-0-166-15:127293:128003 [3] NCCL INFO Channel 03/1 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA/Shared
+ip-26-0-166-15:127292:128002 [2] NCCL INFO Channel 02/1 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA/Shared
+ip-26-0-166-15:127292:128002 [2] NCCL INFO Channel 03/1 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA/Shared
+ip-26-0-166-15:127297:128015 [7] NCCL INFO Channel 02/1 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA/Shared
+ip-26-0-166-15:127297:128015 [7] NCCL INFO Channel 03/1 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA/Shared
+ip-26-0-166-15:127296:128021 [6] NCCL INFO Channel 02/1 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA/Shared
+ip-26-0-166-15:127296:128021 [6] NCCL INFO Channel 03/1 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA/Shared
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+ip-26-0-166-214:624944:625591 [1] NCCL INFO Using network Libfabric
+ip-26-0-166-214:624943:625592 [0] NCCL INFO Using network Libfabric
+ip-26-0-166-244:897240:897833 [0] NCCL INFO Using network Libfabric
+ip-26-0-166-244:897241:897834 [1] NCCL INFO Using network Libfabric
+ip-26-0-166-244:897240:897833 [0] NCCL INFO comm 0xb86c020 rank 1 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0xd97c7a410eecb43c - Init START
+ip-26-0-166-214:624943:625592 [0] NCCL INFO comm 0x14270d50 rank 0 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0xd97c7a410eecb43c - Init START
+ip-26-0-166-244:897241:897834 [1] NCCL INFO comm 0xc574e80 rank 1 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0x28c2fd43c7dcc21d - Init START
+ip-26-0-166-214:624944:625591 [1] NCCL INFO comm 0x1333bb40 rank 0 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0x28c2fd43c7dcc21d - Init START
+ip-26-0-166-244:897240:897833 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-214:624943:625592 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-244:897241:897834 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-214:624944:625591 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+NCCL version 2.18.6+cuda12.1
+ip-26-0-166-214:624947:625594 [4] NCCL INFO Using network Libfabric
+ip-26-0-166-214:624948:625596 [5] NCCL INFO Using network Libfabric
+ip-26-0-166-244:897245:897836 [5] NCCL INFO Using network Libfabric
+ip-26-0-166-244:897244:897835 [4] NCCL INFO Using network Libfabric
+ip-26-0-166-244:897245:897836 [5] NCCL INFO comm 0xbe6fef0 rank 1 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0x51cd0f910a67264f - Init START
+ip-26-0-166-214:624948:625596 [5] NCCL INFO comm 0x14e86150 rank 0 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0x51cd0f910a67264f - Init START
+ip-26-0-166-244:897245:897836 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-214:624948:625596 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-244:897244:897835 [4] NCCL INFO comm 0xb227b50 rank 1 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0x798d13d37276b739 - Init START
+ip-26-0-166-214:624947:625594 [4] NCCL INFO comm 0x14e11080 rank 0 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0x798d13d37276b739 - Init START
+ip-26-0-166-244:897244:897835 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-214:624947:625594 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+NCCL version 2.18.6+cuda12.1
+ip-26-0-166-214:624946:625599 [3] NCCL INFO Using network Libfabric
+ip-26-0-166-214:624945:625600 [2] NCCL INFO Using network Libfabric
+ip-26-0-166-244:897243:897837 [3] NCCL INFO Using network Libfabric
+ip-26-0-166-244:897242:897838 [2] NCCL INFO Using network Libfabric
+ip-26-0-166-244:897243:897837 [3] NCCL INFO comm 0xadc2bf0 rank 1 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0x8ae135b20b129cf2 - Init START
+ip-26-0-166-214:624946:625599 [3] NCCL INFO comm 0x14e747f0 rank 0 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0x8ae135b20b129cf2 - Init START
+ip-26-0-166-244:897243:897837 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-214:624946:625599 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-244:897242:897838 [2] NCCL INFO comm 0xaab7390 rank 1 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0xf7cacdb6031fb045 - Init START
+ip-26-0-166-214:624945:625600 [2] NCCL INFO comm 0x145d9170 rank 0 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0xf7cacdb6031fb045 - Init START
+ip-26-0-166-244:897242:897838 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-214:624945:625600 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-214:624944:625591 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-166-214:624943:625592 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-166-244:897240:897833 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-166-244:897240:897833 [0] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-166-244:897240:897833 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-214:624943:625592 [0] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-166-214:624943:625592 [0] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-166-214:624943:625592 [0] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-166-214:624943:625592 [0] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-166-214:624943:625592 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-166-214:624943:625592 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-244:897241:897834 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-166-244:897241:897834 [1] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-166-244:897241:897834 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-214:624944:625591 [1] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-166-214:624944:625591 [1] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-166-214:624944:625591 [1] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-166-214:624944:625591 [1] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-166-214:624944:625591 [1] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-166-214:624944:625591 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-214:624948:625596 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-166-214:624947:625594 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+NCCL version 2.18.6+cuda12.1
+ip-26-0-166-214:624950:625608 [7] NCCL INFO Using network Libfabric
+ip-26-0-166-214:624949:625609 [6] NCCL INFO Using network Libfabric
+ip-26-0-166-244:897247:897845 [7] NCCL INFO Using network Libfabric
+ip-26-0-166-244:897246:897844 [6] NCCL INFO Using network Libfabric
+ip-26-0-166-244:897247:897845 [7] NCCL INFO comm 0xbd5c080 rank 1 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0x70fd6a2771ea6673 - Init START
+ip-26-0-166-214:624950:625608 [7] NCCL INFO comm 0x13bc7c00 rank 0 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0x70fd6a2771ea6673 - Init START
+ip-26-0-166-244:897247:897845 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-214:624950:625608 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-244:897246:897844 [6] NCCL INFO comm 0xc979e20 rank 1 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0xe5fc12103ec98adb - Init START
+ip-26-0-166-214:624949:625609 [6] NCCL INFO comm 0x138a3ad0 rank 0 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0xe5fc12103ec98adb - Init START
+ip-26-0-166-244:897246:897844 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-214:624949:625609 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-244:897244:897835 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-166-214:624947:625594 [4] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-166-214:624947:625594 [4] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-166-214:624947:625594 [4] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-166-214:624947:625594 [4] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-166-244:897244:897835 [4] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-166-244:897244:897835 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-214:624947:625594 [4] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-166-214:624947:625594 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-244:897240:897833 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897240:897833 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897240:897833 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897240:897833 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897240:897833 [0] NCCL INFO Channel 00/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897240:897833 [0] NCCL INFO Channel 01/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897240:897833 [0] NCCL INFO Channel 02/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897240:897833 [0] NCCL INFO Channel 03/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624943:625592 [0] NCCL INFO Channel 00/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624943:625592 [0] NCCL INFO Channel 01/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624943:625592 [0] NCCL INFO Channel 02/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624943:625592 [0] NCCL INFO Channel 03/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624943:625592 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624943:625592 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624943:625592 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624943:625592 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897241:897834 [1] NCCL INFO Channel 00/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897241:897834 [1] NCCL INFO Channel 01/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897241:897834 [1] NCCL INFO Channel 02/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897241:897834 [1] NCCL INFO Channel 03/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897241:897834 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897241:897834 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897241:897834 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897241:897834 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624944:625591 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624944:625591 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624944:625591 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624944:625591 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624944:625591 [1] NCCL INFO Channel 00/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624944:625591 [1] NCCL INFO Channel 01/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624944:625591 [1] NCCL INFO Channel 02/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624944:625591 [1] NCCL INFO Channel 03/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897245:897836 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-166-244:897245:897836 [5] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-166-244:897245:897836 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-214:624948:625596 [5] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-166-214:624948:625596 [5] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-166-214:624948:625596 [5] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-166-214:624948:625596 [5] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-166-214:624948:625596 [5] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-166-214:624948:625596 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-214:624946:625599 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-166-214:624947:625594 [4] NCCL INFO Channel 00/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624947:625594 [4] NCCL INFO Channel 01/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624947:625594 [4] NCCL INFO Channel 02/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624947:625594 [4] NCCL INFO Channel 03/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624947:625594 [4] NCCL INFO Channel 00/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624947:625594 [4] NCCL INFO Channel 01/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624947:625594 [4] NCCL INFO Channel 02/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624947:625594 [4] NCCL INFO Channel 03/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897244:897835 [4] NCCL INFO Channel 00/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897244:897835 [4] NCCL INFO Channel 01/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897244:897835 [4] NCCL INFO Channel 02/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897244:897835 [4] NCCL INFO Channel 03/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897244:897835 [4] NCCL INFO Channel 00/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897244:897835 [4] NCCL INFO Channel 01/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897244:897835 [4] NCCL INFO Channel 02/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897244:897835 [4] NCCL INFO Channel 03/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624945:625600 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-166-244:897243:897837 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-166-214:624946:625599 [3] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-166-214:624946:625599 [3] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-166-214:624946:625599 [3] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-166-214:624946:625599 [3] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-166-214:624946:625599 [3] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-166-214:624946:625599 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-244:897243:897837 [3] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-166-244:897243:897837 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-214:624948:625596 [5] NCCL INFO Channel 00/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897245:897836 [5] NCCL INFO Channel 00/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624948:625596 [5] NCCL INFO Channel 01/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624948:625596 [5] NCCL INFO Channel 02/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897245:897836 [5] NCCL INFO Channel 01/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624948:625596 [5] NCCL INFO Channel 03/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897245:897836 [5] NCCL INFO Channel 02/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624948:625596 [5] NCCL INFO Channel 00/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897245:897836 [5] NCCL INFO Channel 03/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624948:625596 [5] NCCL INFO Channel 01/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897245:897836 [5] NCCL INFO Channel 00/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624948:625596 [5] NCCL INFO Channel 02/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897245:897836 [5] NCCL INFO Channel 01/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624948:625596 [5] NCCL INFO Channel 03/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897245:897836 [5] NCCL INFO Channel 02/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897245:897836 [5] NCCL INFO Channel 03/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897242:897838 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-166-244:897242:897838 [2] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-166-244:897242:897838 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-214:624945:625600 [2] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-166-214:624945:625600 [2] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-166-214:624945:625600 [2] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-166-214:624945:625600 [2] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-166-214:624945:625600 [2] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-166-214:624945:625600 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-244:897243:897837 [3] NCCL INFO Channel 00/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897243:897837 [3] NCCL INFO Channel 01/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897243:897837 [3] NCCL INFO Channel 02/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897243:897837 [3] NCCL INFO Channel 03/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897243:897837 [3] NCCL INFO Channel 00/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897243:897837 [3] NCCL INFO Channel 01/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897243:897837 [3] NCCL INFO Channel 02/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897243:897837 [3] NCCL INFO Channel 03/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624946:625599 [3] NCCL INFO Channel 00/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624946:625599 [3] NCCL INFO Channel 01/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624946:625599 [3] NCCL INFO Channel 02/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624946:625599 [3] NCCL INFO Channel 03/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624946:625599 [3] NCCL INFO Channel 00/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624946:625599 [3] NCCL INFO Channel 01/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624946:625599 [3] NCCL INFO Channel 02/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624946:625599 [3] NCCL INFO Channel 03/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624949:625609 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-166-214:624950:625608 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-166-244:897242:897838 [2] NCCL INFO Channel 00/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897242:897838 [2] NCCL INFO Channel 01/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897242:897838 [2] NCCL INFO Channel 02/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897242:897838 [2] NCCL INFO Channel 03/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897242:897838 [2] NCCL INFO Channel 00/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897242:897838 [2] NCCL INFO Channel 01/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897242:897838 [2] NCCL INFO Channel 02/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897242:897838 [2] NCCL INFO Channel 03/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624945:625600 [2] NCCL INFO Channel 00/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624945:625600 [2] NCCL INFO Channel 01/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624945:625600 [2] NCCL INFO Channel 02/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624945:625600 [2] NCCL INFO Channel 03/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624945:625600 [2] NCCL INFO Channel 00/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624945:625600 [2] NCCL INFO Channel 01/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624945:625600 [2] NCCL INFO Channel 02/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624945:625600 [2] NCCL INFO Channel 03/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897247:897845 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-166-244:897247:897845 [7] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-166-244:897247:897845 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-214:624950:625608 [7] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-166-214:624950:625608 [7] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-166-214:624950:625608 [7] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-166-214:624950:625608 [7] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-166-214:624950:625608 [7] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-166-214:624950:625608 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-244:897246:897844 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-166-244:897246:897844 [6] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-166-244:897246:897844 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-214:624949:625609 [6] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-166-214:624949:625609 [6] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-166-214:624949:625609 [6] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-166-214:624949:625609 [6] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-166-214:624949:625609 [6] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-166-214:624949:625609 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-244:897247:897845 [7] NCCL INFO Channel 00/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897247:897845 [7] NCCL INFO Channel 01/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897247:897845 [7] NCCL INFO Channel 02/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897247:897845 [7] NCCL INFO Channel 03/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897247:897845 [7] NCCL INFO Channel 00/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897247:897845 [7] NCCL INFO Channel 01/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897247:897845 [7] NCCL INFO Channel 02/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897247:897845 [7] NCCL INFO Channel 03/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624950:625608 [7] NCCL INFO Channel 00/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624950:625608 [7] NCCL INFO Channel 01/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624950:625608 [7] NCCL INFO Channel 02/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624950:625608 [7] NCCL INFO Channel 03/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624950:625608 [7] NCCL INFO Channel 00/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624950:625608 [7] NCCL INFO Channel 01/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624950:625608 [7] NCCL INFO Channel 02/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624950:625608 [7] NCCL INFO Channel 03/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624949:625609 [6] NCCL INFO Channel 00/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624949:625609 [6] NCCL INFO Channel 01/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624949:625609 [6] NCCL INFO Channel 02/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624949:625609 [6] NCCL INFO Channel 03/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624949:625609 [6] NCCL INFO Channel 00/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624949:625609 [6] NCCL INFO Channel 01/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624949:625609 [6] NCCL INFO Channel 02/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624949:625609 [6] NCCL INFO Channel 03/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897246:897844 [6] NCCL INFO Channel 00/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897246:897844 [6] NCCL INFO Channel 01/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897246:897844 [6] NCCL INFO Channel 02/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897246:897844 [6] NCCL INFO Channel 03/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897246:897844 [6] NCCL INFO Channel 00/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897246:897844 [6] NCCL INFO Channel 01/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897246:897844 [6] NCCL INFO Channel 02/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897246:897844 [6] NCCL INFO Channel 03/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624944:625591 [1] NCCL INFO Connected all rings
+ip-26-0-166-214:624944:625591 [1] NCCL INFO Connected all trees
+ip-26-0-166-214:624944:625591 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-214:624944:625591 [1] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-214:624944:625591 [1] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-214:624943:625592 [0] NCCL INFO Connected all rings
+ip-26-0-166-214:624943:625592 [0] NCCL INFO Connected all trees
+ip-26-0-166-214:624943:625592 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-214:624943:625592 [0] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-214:624943:625592 [0] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-214:624944:625591 [1] NCCL INFO comm 0x1333bb40 rank 0 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0x28c2fd43c7dcc21d - Init COMPLETE
+ip-26-0-166-214:624944:625622 [1] NCCL INFO Channel 02/1 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA/Shared
+ip-26-0-166-214:624944:625622 [1] NCCL INFO Channel 03/1 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA/Shared
+ip-26-0-166-214:624948:625596 [5] NCCL INFO Connected all rings
+ip-26-0-166-214:624948:625596 [5] NCCL INFO Connected all trees
+ip-26-0-166-214:624948:625596 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-214:624948:625596 [5] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-214:624948:625596 [5] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-214:624943:625592 [0] NCCL INFO comm 0x14270d50 rank 0 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0xd97c7a410eecb43c - Init COMPLETE
+ip-26-0-166-214:624943:625623 [0] NCCL INFO Channel 02/1 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA/Shared
+ip-26-0-166-214:624943:625623 [0] NCCL INFO Channel 03/1 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA/Shared
+ip-26-0-166-214:624947:625594 [4] NCCL INFO Connected all rings
+ip-26-0-166-214:624947:625594 [4] NCCL INFO Connected all trees
+ip-26-0-166-214:624947:625594 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-214:624947:625594 [4] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-214:624947:625594 [4] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-244:897241:897834 [1] NCCL INFO Connected all rings
+ip-26-0-166-244:897241:897834 [1] NCCL INFO Connected all trees
+ip-26-0-166-244:897241:897834 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-244:897241:897834 [1] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-244:897241:897834 [1] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-214:624948:625596 [5] NCCL INFO comm 0x14e86150 rank 0 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0x51cd0f910a67264f - Init COMPLETE
+ip-26-0-166-214:624948:625624 [5] NCCL INFO Channel 02/1 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA/Shared
+ip-26-0-166-214:624948:625624 [5] NCCL INFO Channel 03/1 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA/Shared
+ip-26-0-166-244:897244:897835 [4] NCCL INFO Connected all rings
+ip-26-0-166-244:897244:897835 [4] NCCL INFO Connected all trees
+ip-26-0-166-244:897244:897835 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-244:897244:897835 [4] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-244:897244:897835 [4] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-244:897240:897833 [0] NCCL INFO Connected all rings
+ip-26-0-166-244:897240:897833 [0] NCCL INFO Connected all trees
+ip-26-0-166-244:897240:897833 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-244:897240:897833 [0] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-244:897240:897833 [0] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-214:624947:625594 [4] NCCL INFO comm 0x14e11080 rank 0 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0x798d13d37276b739 - Init COMPLETE
+ip-26-0-166-214:624947:625625 [4] NCCL INFO Channel 02/1 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA/Shared
+ip-26-0-166-214:624947:625625 [4] NCCL INFO Channel 03/1 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA/Shared
+ip-26-0-166-244:897241:897834 [1] NCCL INFO comm 0xc574e80 rank 1 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0x28c2fd43c7dcc21d - Init COMPLETE
+ip-26-0-166-244:897241:897859 [1] NCCL INFO Channel 02/1 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA/Shared
+ip-26-0-166-244:897241:897859 [1] NCCL INFO Channel 03/1 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA/Shared
+ip-26-0-166-244:897240:897833 [0] NCCL INFO comm 0xb86c020 rank 1 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0xd97c7a410eecb43c - Init COMPLETE
+ip-26-0-166-244:897240:897860 [0] NCCL INFO Channel 02/1 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA/Shared
+ip-26-0-166-244:897240:897860 [0] NCCL INFO Channel 03/1 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA/Shared
+ip-26-0-166-244:897245:897836 [5] NCCL INFO Connected all rings
+ip-26-0-166-244:897245:897836 [5] NCCL INFO Connected all trees
+ip-26-0-166-244:897245:897836 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-244:897245:897836 [5] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-244:897245:897836 [5] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-244:897244:897835 [4] NCCL INFO comm 0xb227b50 rank 1 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0x798d13d37276b739 - Init COMPLETE
+ip-26-0-166-244:897244:897861 [4] NCCL INFO Channel 02/1 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA/Shared
+ip-26-0-166-244:897244:897861 [4] NCCL INFO Channel 03/1 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA/Shared
+ip-26-0-166-214:624946:625599 [3] NCCL INFO Connected all rings
+ip-26-0-166-214:624946:625599 [3] NCCL INFO Connected all trees
+ip-26-0-166-214:624946:625599 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-214:624946:625599 [3] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-214:624946:625599 [3] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-244:897245:897836 [5] NCCL INFO comm 0xbe6fef0 rank 1 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0x51cd0f910a67264f - Init COMPLETE
+ip-26-0-166-244:897245:897862 [5] NCCL INFO Channel 02/1 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA/Shared
+ip-26-0-166-244:897245:897862 [5] NCCL INFO Channel 03/1 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA/Shared
+ip-26-0-166-214:624946:625599 [3] NCCL INFO comm 0x14e747f0 rank 0 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0x8ae135b20b129cf2 - Init COMPLETE
+ip-26-0-166-214:624946:625626 [3] NCCL INFO Channel 02/1 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA/Shared
+ip-26-0-166-214:624946:625626 [3] NCCL INFO Channel 03/1 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA/Shared
+ip-26-0-166-244:897243:897837 [3] NCCL INFO Connected all rings
+ip-26-0-166-244:897243:897837 [3] NCCL INFO Connected all trees
+ip-26-0-166-244:897243:897837 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-244:897243:897837 [3] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-244:897243:897837 [3] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-244:897243:897837 [3] NCCL INFO comm 0xadc2bf0 rank 1 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0x8ae135b20b129cf2 - Init COMPLETE
+ip-26-0-166-244:897243:897864 [3] NCCL INFO Channel 02/1 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA/Shared
+ip-26-0-166-244:897243:897864 [3] NCCL INFO Channel 03/1 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA/Shared
+ip-26-0-166-214:624945:625600 [2] NCCL INFO Connected all rings
+ip-26-0-166-214:624945:625600 [2] NCCL INFO Connected all trees
+ip-26-0-166-214:624945:625600 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-214:624945:625600 [2] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-214:624945:625600 [2] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-244:897242:897838 [2] NCCL INFO Connected all rings
+ip-26-0-166-244:897242:897838 [2] NCCL INFO Connected all trees
+ip-26-0-166-244:897242:897838 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-244:897242:897838 [2] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-244:897242:897838 [2] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-214:624945:625600 [2] NCCL INFO comm 0x145d9170 rank 0 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0xf7cacdb6031fb045 - Init COMPLETE
+ip-26-0-166-214:624945:625627 [2] NCCL INFO Channel 02/1 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA/Shared
+ip-26-0-166-214:624945:625627 [2] NCCL INFO Channel 03/1 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA/Shared
+ip-26-0-166-244:897242:897838 [2] NCCL INFO comm 0xaab7390 rank 1 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0xf7cacdb6031fb045 - Init COMPLETE
+ip-26-0-166-244:897242:897865 [2] NCCL INFO Channel 02/1 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA/Shared
+ip-26-0-166-244:897242:897865 [2] NCCL INFO Channel 03/1 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA/Shared
+ip-26-0-166-214:624950:625608 [7] NCCL INFO Connected all rings
+ip-26-0-166-214:624950:625608 [7] NCCL INFO Connected all trees
+ip-26-0-166-214:624950:625608 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-214:624950:625608 [7] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-214:624950:625608 [7] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-244:897247:897845 [7] NCCL INFO Connected all rings
+ip-26-0-166-244:897247:897845 [7] NCCL INFO Connected all trees
+ip-26-0-166-244:897247:897845 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-244:897247:897845 [7] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-244:897247:897845 [7] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-244:897246:897844 [6] NCCL INFO Connected all rings
+ip-26-0-166-244:897246:897844 [6] NCCL INFO Connected all trees
+ip-26-0-166-244:897246:897844 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-244:897246:897844 [6] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-244:897246:897844 [6] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-214:624949:625609 [6] NCCL INFO Connected all rings
+ip-26-0-166-214:624949:625609 [6] NCCL INFO Connected all trees
+ip-26-0-166-214:624949:625609 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-214:624949:625609 [6] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-214:624949:625609 [6] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-214:624950:625608 [7] NCCL INFO comm 0x13bc7c00 rank 0 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0x70fd6a2771ea6673 - Init COMPLETE
+ip-26-0-166-214:624950:625629 [7] NCCL INFO Channel 02/1 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA/Shared
+ip-26-0-166-214:624950:625629 [7] NCCL INFO Channel 03/1 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA/Shared
+ip-26-0-166-244:897247:897845 [7] NCCL INFO comm 0xbd5c080 rank 1 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0x70fd6a2771ea6673 - Init COMPLETE
+ip-26-0-166-244:897247:897866 [7] NCCL INFO Channel 02/1 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA/Shared
+ip-26-0-166-244:897247:897866 [7] NCCL INFO Channel 03/1 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA/Shared
+ip-26-0-166-244:897246:897844 [6] NCCL INFO comm 0xc979e20 rank 1 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0xe5fc12103ec98adb - Init COMPLETE
+ip-26-0-166-244:897246:897867 [6] NCCL INFO Channel 02/1 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA/Shared
+ip-26-0-166-244:897246:897867 [6] NCCL INFO Channel 03/1 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA/Shared
+ip-26-0-166-214:624949:625609 [6] NCCL INFO comm 0x138a3ad0 rank 0 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0xe5fc12103ec98adb - Init COMPLETE
+ip-26-0-166-214:624949:625630 [6] NCCL INFO Channel 02/1 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA/Shared
+ip-26-0-166-214:624949:625630 [6] NCCL INFO Channel 03/1 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA/Shared
+ip-26-0-166-214:624944:625647 [1] NCCL INFO Channel 02/1 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA/Shared
+ip-26-0-166-214:624943:625648 [0] NCCL INFO Channel 02/1 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA/Shared
+ip-26-0-166-214:624944:625647 [1] NCCL INFO Channel 03/1 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA/Shared
+ip-26-0-166-214:624943:625648 [0] NCCL INFO Channel 03/1 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA/Shared
+ip-26-0-166-214:624947:625665 [4] NCCL INFO Channel 02/1 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA/Shared
+ip-26-0-166-214:624947:625665 [4] NCCL INFO Channel 03/1 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA/Shared
+ip-26-0-166-214:624948:625666 [5] NCCL INFO Channel 02/1 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA/Shared
+ip-26-0-166-214:624948:625666 [5] NCCL INFO Channel 03/1 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA/Shared
+ip-26-0-166-214:624945:625684 [2] NCCL INFO Channel 02/1 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA/Shared
+ip-26-0-166-214:624946:625685 [3] NCCL INFO Channel 02/1 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA/Shared
+ip-26-0-166-214:624945:625684 [2] NCCL INFO Channel 03/1 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA/Shared
+ip-26-0-166-214:624946:625685 [3] NCCL INFO Channel 03/1 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA/Shared
+ip-26-0-166-214:624950:625702 [7] NCCL INFO Channel 02/1 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA/Shared
+ip-26-0-166-214:624950:625702 [7] NCCL INFO Channel 03/1 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA/Shared
+ip-26-0-166-214:624949:625703 [6] NCCL INFO Channel 02/1 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA/Shared
+ip-26-0-166-214:624949:625703 [6] NCCL INFO Channel 03/1 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA/Shared
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+ip-26-0-166-244:897240:897871 [0] NCCL INFO Using network Libfabric
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+ip-26-0-166-36:104429:105020 [0] NCCL INFO Using network Libfabric
+ip-26-0-166-36:104429:105020 [0] NCCL INFO comm 0xc811370 rank 1 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0x829a5a5003cf4fdb - Init START
+ip-26-0-166-244:897240:897871 [0] NCCL INFO comm 0x14119020 rank 0 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0x829a5a5003cf4fdb - Init START
+ip-26-0-166-36:104429:105020 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-244:897240:897871 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-244:897241:897873 [1] NCCL INFO Using network Libfabric
+ip-26-0-166-36:104430:105021 [1] NCCL INFO Using network Libfabric
+ip-26-0-166-36:104430:105021 [1] NCCL INFO comm 0xbfb8680 rank 1 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0x852a737d30d6c6b7 - Init START
+ip-26-0-166-244:897241:897873 [1] NCCL INFO comm 0x14fbfd40 rank 0 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0x852a737d30d6c6b7 - Init START
+ip-26-0-166-36:104430:105021 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-244:897241:897873 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+NCCL version 2.18.6+cuda12.1
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+ip-26-0-166-244:897245:897875 [5] NCCL INFO Using network Libfabric
+ip-26-0-166-244:897244:897877 [4] NCCL INFO Using network Libfabric
+ip-26-0-166-36:104434:105022 [5] NCCL INFO Using network Libfabric
+ip-26-0-166-36:104433:105023 [4] NCCL INFO Using network Libfabric
+ip-26-0-166-36:104434:105022 [5] NCCL INFO comm 0xafb93d0 rank 1 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0x4aca07c1fffe5f34 - Init START
+ip-26-0-166-244:897245:897875 [5] NCCL INFO comm 0x148b8d50 rank 0 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0x4aca07c1fffe5f34 - Init START
+ip-26-0-166-36:104434:105022 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-244:897245:897875 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-36:104433:105023 [4] NCCL INFO comm 0xc517e80 rank 1 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0xa727c3bcbb6064de - Init START
+ip-26-0-166-244:897244:897877 [4] NCCL INFO comm 0x13c9b9e0 rank 0 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0xa727c3bcbb6064de - Init START
+ip-26-0-166-36:104433:105023 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-244:897244:897877 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+NCCL version 2.18.6+cuda12.1
+ip-26-0-166-244:897243:897880 [3] NCCL INFO Using network Libfabric
+ip-26-0-166-244:897242:897881 [2] NCCL INFO Using network Libfabric
+ip-26-0-166-36:104432:105024 [3] NCCL INFO Using network Libfabric
+ip-26-0-166-36:104431:105025 [2] NCCL INFO Using network Libfabric
+ip-26-0-166-36:104432:105024 [3] NCCL INFO comm 0xad4e0c0 rank 1 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0xf552a17783bb308 - Init START
+ip-26-0-166-244:897243:897880 [3] NCCL INFO comm 0x1380dee0 rank 0 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0xf552a17783bb308 - Init START
+ip-26-0-166-36:104432:105024 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-244:897243:897880 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-36:104431:105025 [2] NCCL INFO comm 0xbbb92e0 rank 1 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0x9929739d002d583e - Init START
+ip-26-0-166-244:897242:897881 [2] NCCL INFO comm 0x13505950 rank 0 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0x9929739d002d583e - Init START
+ip-26-0-166-244:897242:897881 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-36:104431:105025 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-244:897240:897871 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-166-36:104429:105020 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-166-36:104429:105020 [0] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-166-36:104429:105020 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-244:897240:897871 [0] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-166-244:897240:897871 [0] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-166-244:897240:897871 [0] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-166-244:897240:897871 [0] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-166-244:897240:897871 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-166-244:897240:897871 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-244:897241:897873 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-166-36:104430:105021 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-166-36:104430:105021 [1] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-166-36:104430:105021 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-244:897241:897873 [1] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-166-244:897241:897873 [1] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-166-244:897241:897873 [1] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-166-244:897241:897873 [1] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-166-244:897241:897873 [1] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-166-244:897241:897873 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-36:104434:105022 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-166-244:897245:897875 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-166-36:104433:105023 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-166-36:104434:105022 [5] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-166-36:104434:105022 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-244:897245:897875 [5] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-166-244:897245:897875 [5] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-166-244:897245:897875 [5] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-166-244:897245:897875 [5] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-166-244:897245:897875 [5] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-166-244:897245:897875 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-244:897244:897877 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-166-244:897244:897877 [4] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-166-244:897244:897877 [4] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-166-244:897244:897877 [4] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-166-36:104433:105023 [4] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-166-36:104433:105023 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-244:897244:897877 [4] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-166-244:897244:897877 [4] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-166-244:897244:897877 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-36:104429:105020 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104429:105020 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104429:105020 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104429:105020 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104429:105020 [0] NCCL INFO Channel 00/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104429:105020 [0] NCCL INFO Channel 01/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104429:105020 [0] NCCL INFO Channel 02/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104429:105020 [0] NCCL INFO Channel 03/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897240:897871 [0] NCCL INFO Channel 00/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897240:897871 [0] NCCL INFO Channel 01/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897240:897871 [0] NCCL INFO Channel 02/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897240:897871 [0] NCCL INFO Channel 03/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897240:897871 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897240:897871 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897240:897871 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897240:897871 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897242:897881 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-166-36:104432:105024 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-166-36:104431:105025 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-166-244:897243:897880 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-166-36:104431:105025 [2] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-166-36:104431:105025 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-244:897242:897881 [2] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-166-244:897242:897881 [2] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-166-244:897242:897881 [2] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-166-244:897242:897881 [2] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-166-244:897242:897881 [2] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-166-244:897242:897881 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-244:897243:897880 [3] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-166-244:897243:897880 [3] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-166-244:897243:897880 [3] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-166-244:897243:897880 [3] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-166-244:897243:897880 [3] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-166-244:897243:897880 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-36:104432:105024 [3] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-166-36:104432:105024 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-244:897241:897873 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897241:897873 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897241:897873 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897241:897873 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897241:897873 [1] NCCL INFO Channel 00/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897241:897873 [1] NCCL INFO Channel 01/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897241:897873 [1] NCCL INFO Channel 02/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897241:897873 [1] NCCL INFO Channel 03/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104430:105021 [1] NCCL INFO Channel 00/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104430:105021 [1] NCCL INFO Channel 01/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104430:105021 [1] NCCL INFO Channel 02/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104430:105021 [1] NCCL INFO Channel 03/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104430:105021 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104430:105021 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104430:105021 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104430:105021 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897245:897875 [5] NCCL INFO Channel 00/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897245:897875 [5] NCCL INFO Channel 01/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897245:897875 [5] NCCL INFO Channel 02/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897245:897875 [5] NCCL INFO Channel 03/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897245:897875 [5] NCCL INFO Channel 00/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897245:897875 [5] NCCL INFO Channel 01/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897245:897875 [5] NCCL INFO Channel 02/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897245:897875 [5] NCCL INFO Channel 03/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104434:105022 [5] NCCL INFO Channel 00/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104434:105022 [5] NCCL INFO Channel 01/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104433:105023 [4] NCCL INFO Channel 00/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104434:105022 [5] NCCL INFO Channel 02/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104433:105023 [4] NCCL INFO Channel 01/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104434:105022 [5] NCCL INFO Channel 03/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104433:105023 [4] NCCL INFO Channel 02/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104434:105022 [5] NCCL INFO Channel 00/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104433:105023 [4] NCCL INFO Channel 03/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104434:105022 [5] NCCL INFO Channel 01/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104433:105023 [4] NCCL INFO Channel 00/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104434:105022 [5] NCCL INFO Channel 02/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104433:105023 [4] NCCL INFO Channel 01/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104434:105022 [5] NCCL INFO Channel 03/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104433:105023 [4] NCCL INFO Channel 02/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104433:105023 [4] NCCL INFO Channel 03/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897244:897877 [4] NCCL INFO Channel 00/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897244:897877 [4] NCCL INFO Channel 01/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897244:897877 [4] NCCL INFO Channel 02/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897244:897877 [4] NCCL INFO Channel 03/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897244:897877 [4] NCCL INFO Channel 00/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897244:897877 [4] NCCL INFO Channel 01/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897244:897877 [4] NCCL INFO Channel 02/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897244:897877 [4] NCCL INFO Channel 03/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104431:105025 [2] NCCL INFO Channel 00/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104431:105025 [2] NCCL INFO Channel 01/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104431:105025 [2] NCCL INFO Channel 02/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104431:105025 [2] NCCL INFO Channel 03/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104431:105025 [2] NCCL INFO Channel 00/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104431:105025 [2] NCCL INFO Channel 01/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104431:105025 [2] NCCL INFO Channel 02/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104431:105025 [2] NCCL INFO Channel 03/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897242:897881 [2] NCCL INFO Channel 00/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897242:897881 [2] NCCL INFO Channel 01/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897242:897881 [2] NCCL INFO Channel 02/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897242:897881 [2] NCCL INFO Channel 03/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897242:897881 [2] NCCL INFO Channel 00/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897242:897881 [2] NCCL INFO Channel 01/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897242:897881 [2] NCCL INFO Channel 02/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897242:897881 [2] NCCL INFO Channel 03/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897243:897880 [3] NCCL INFO Channel 00/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897243:897880 [3] NCCL INFO Channel 01/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897243:897880 [3] NCCL INFO Channel 02/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897243:897880 [3] NCCL INFO Channel 03/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897243:897880 [3] NCCL INFO Channel 00/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897243:897880 [3] NCCL INFO Channel 01/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897243:897880 [3] NCCL INFO Channel 02/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897243:897880 [3] NCCL INFO Channel 03/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104432:105024 [3] NCCL INFO Channel 00/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104432:105024 [3] NCCL INFO Channel 01/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104432:105024 [3] NCCL INFO Channel 02/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104432:105024 [3] NCCL INFO Channel 03/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104432:105024 [3] NCCL INFO Channel 00/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104432:105024 [3] NCCL INFO Channel 01/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104432:105024 [3] NCCL INFO Channel 02/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104432:105024 [3] NCCL INFO Channel 03/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+NCCL version 2.18.6+cuda12.1
+ip-26-0-166-244:897246:897896 [6] NCCL INFO Using network Libfabric
+ip-26-0-166-244:897247:897898 [7] NCCL INFO Using network Libfabric
+ip-26-0-166-36:104435:105039 [6] NCCL INFO Using network Libfabric
+ip-26-0-166-36:104436:105040 [7] NCCL INFO Using network Libfabric
+ip-26-0-166-36:104435:105039 [6] NCCL INFO comm 0xb175c80 rank 1 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0x3d478d20852d788d - Init START
+ip-26-0-166-244:897246:897896 [6] NCCL INFO comm 0x15228320 rank 0 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0x3d478d20852d788d - Init START
+ip-26-0-166-36:104435:105039 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-244:897246:897896 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-36:104436:105040 [7] NCCL INFO comm 0xbaeacb0 rank 1 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0x6a566fb09d3bd505 - Init START
+ip-26-0-166-244:897247:897898 [7] NCCL INFO comm 0x147aa4a0 rank 0 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0x6a566fb09d3bd505 - Init START
+ip-26-0-166-244:897247:897898 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-36:104436:105040 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-36:104435:105039 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-166-36:104436:105040 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-166-244:897247:897898 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-166-244:897247:897898 [7] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-166-244:897247:897898 [7] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-166-244:897247:897898 [7] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-166-244:897247:897898 [7] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-166-244:897247:897898 [7] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-166-244:897247:897898 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-36:104436:105040 [7] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-166-36:104436:105040 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-36:104435:105039 [6] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-166-36:104435:105039 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-244:897246:897896 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-166-244:897246:897896 [6] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-166-244:897246:897896 [6] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-166-244:897246:897896 [6] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-166-244:897246:897896 [6] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-166-244:897246:897896 [6] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-166-244:897246:897896 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-36:104436:105040 [7] NCCL INFO Channel 00/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104436:105040 [7] NCCL INFO Channel 01/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104436:105040 [7] NCCL INFO Channel 02/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104436:105040 [7] NCCL INFO Channel 03/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104436:105040 [7] NCCL INFO Channel 00/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104436:105040 [7] NCCL INFO Channel 01/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104436:105040 [7] NCCL INFO Channel 02/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104436:105040 [7] NCCL INFO Channel 03/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897247:897898 [7] NCCL INFO Channel 00/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897247:897898 [7] NCCL INFO Channel 01/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897247:897898 [7] NCCL INFO Channel 02/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897247:897898 [7] NCCL INFO Channel 03/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897247:897898 [7] NCCL INFO Channel 00/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897247:897898 [7] NCCL INFO Channel 01/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897247:897898 [7] NCCL INFO Channel 02/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897247:897898 [7] NCCL INFO Channel 03/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104435:105039 [6] NCCL INFO Channel 00/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104435:105039 [6] NCCL INFO Channel 01/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104435:105039 [6] NCCL INFO Channel 02/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897246:897896 [6] NCCL INFO Channel 00/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104435:105039 [6] NCCL INFO Channel 03/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897246:897896 [6] NCCL INFO Channel 01/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104435:105039 [6] NCCL INFO Channel 00/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897246:897896 [6] NCCL INFO Channel 02/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104435:105039 [6] NCCL INFO Channel 01/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897246:897896 [6] NCCL INFO Channel 03/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104435:105039 [6] NCCL INFO Channel 02/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897246:897896 [6] NCCL INFO Channel 00/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104435:105039 [6] NCCL INFO Channel 03/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897246:897896 [6] NCCL INFO Channel 01/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897246:897896 [6] NCCL INFO Channel 02/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897246:897896 [6] NCCL INFO Channel 03/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104429:105020 [0] NCCL INFO Connected all rings
+ip-26-0-166-36:104429:105020 [0] NCCL INFO Connected all trees
+ip-26-0-166-36:104429:105020 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-36:104429:105020 [0] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-36:104429:105020 [0] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-36:104429:105020 [0] NCCL INFO comm 0xc811370 rank 1 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0x829a5a5003cf4fdb - Init COMPLETE
+ip-26-0-166-36:104429:105046 [0] NCCL INFO Channel 02/1 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA/Shared
+ip-26-0-166-36:104429:105046 [0] NCCL INFO Channel 03/1 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA/Shared
+ip-26-0-166-244:897240:897871 [0] NCCL INFO Connected all rings
+ip-26-0-166-244:897240:897871 [0] NCCL INFO Connected all trees
+ip-26-0-166-244:897240:897871 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-244:897240:897871 [0] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-244:897240:897871 [0] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-244:897240:897871 [0] NCCL INFO comm 0x14119020 rank 0 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0x829a5a5003cf4fdb - Init COMPLETE
+ip-26-0-166-244:897240:897906 [0] NCCL INFO Channel 02/1 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA/Shared
+ip-26-0-166-244:897240:897906 [0] NCCL INFO Channel 03/1 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA/Shared
+ip-26-0-166-36:104430:105021 [1] NCCL INFO Connected all rings
+ip-26-0-166-36:104430:105021 [1] NCCL INFO Connected all trees
+ip-26-0-166-36:104430:105021 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-36:104430:105021 [1] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-36:104430:105021 [1] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-36:104433:105023 [4] NCCL INFO Connected all rings
+ip-26-0-166-36:104433:105023 [4] NCCL INFO Connected all trees
+ip-26-0-166-36:104433:105023 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-36:104433:105023 [4] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-36:104433:105023 [4] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-36:104434:105022 [5] NCCL INFO Connected all rings
+ip-26-0-166-36:104434:105022 [5] NCCL INFO Connected all trees
+ip-26-0-166-36:104434:105022 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-36:104434:105022 [5] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-36:104434:105022 [5] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-36:104431:105025 [2] NCCL INFO Connected all rings
+ip-26-0-166-36:104431:105025 [2] NCCL INFO Connected all trees
+ip-26-0-166-36:104431:105025 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-36:104431:105025 [2] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-36:104431:105025 [2] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-36:104430:105021 [1] NCCL INFO comm 0xbfb8680 rank 1 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0x852a737d30d6c6b7 - Init COMPLETE
+ip-26-0-166-36:104432:105024 [3] NCCL INFO Connected all rings
+ip-26-0-166-36:104432:105024 [3] NCCL INFO Connected all trees
+ip-26-0-166-36:104432:105024 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-36:104432:105024 [3] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-36:104432:105024 [3] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-36:104433:105023 [4] NCCL INFO comm 0xc517e80 rank 1 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0xa727c3bcbb6064de - Init COMPLETE
+ip-26-0-166-36:104430:105047 [1] NCCL INFO Channel 02/1 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA/Shared
+ip-26-0-166-36:104430:105047 [1] NCCL INFO Channel 03/1 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA/Shared
+ip-26-0-166-36:104433:105048 [4] NCCL INFO Channel 02/1 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA/Shared
+ip-26-0-166-36:104433:105048 [4] NCCL INFO Channel 03/1 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA/Shared
+ip-26-0-166-36:104434:105022 [5] NCCL INFO comm 0xafb93d0 rank 1 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0x4aca07c1fffe5f34 - Init COMPLETE
+ip-26-0-166-36:104434:105049 [5] NCCL INFO Channel 02/1 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA/Shared
+ip-26-0-166-36:104434:105049 [5] NCCL INFO Channel 03/1 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA/Shared
+ip-26-0-166-36:104431:105025 [2] NCCL INFO comm 0xbbb92e0 rank 1 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0x9929739d002d583e - Init COMPLETE
+ip-26-0-166-36:104431:105050 [2] NCCL INFO Channel 02/1 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA/Shared
+ip-26-0-166-36:104431:105050 [2] NCCL INFO Channel 03/1 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA/Shared
+ip-26-0-166-36:104432:105024 [3] NCCL INFO comm 0xad4e0c0 rank 1 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0xf552a17783bb308 - Init COMPLETE
+ip-26-0-166-36:104432:105051 [3] NCCL INFO Channel 02/1 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA/Shared
+ip-26-0-166-36:104432:105051 [3] NCCL INFO Channel 03/1 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA/Shared
+ip-26-0-166-244:897241:897873 [1] NCCL INFO Connected all rings
+ip-26-0-166-244:897241:897873 [1] NCCL INFO Connected all trees
+ip-26-0-166-244:897241:897873 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-244:897241:897873 [1] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-244:897241:897873 [1] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-244:897244:897877 [4] NCCL INFO Connected all rings
+ip-26-0-166-244:897244:897877 [4] NCCL INFO Connected all trees
+ip-26-0-166-244:897244:897877 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-244:897244:897877 [4] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-244:897244:897877 [4] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-36:104435:105039 [6] NCCL INFO Connected all rings
+ip-26-0-166-36:104435:105039 [6] NCCL INFO Connected all trees
+ip-26-0-166-36:104435:105039 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-36:104435:105039 [6] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-36:104435:105039 [6] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-244:897245:897875 [5] NCCL INFO Connected all rings
+ip-26-0-166-244:897245:897875 [5] NCCL INFO Connected all trees
+ip-26-0-166-244:897245:897875 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-244:897245:897875 [5] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-244:897245:897875 [5] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-36:104436:105040 [7] NCCL INFO Connected all rings
+ip-26-0-166-36:104436:105040 [7] NCCL INFO Connected all trees
+ip-26-0-166-36:104436:105040 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-36:104436:105040 [7] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-36:104436:105040 [7] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-244:897241:897873 [1] NCCL INFO comm 0x14fbfd40 rank 0 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0x852a737d30d6c6b7 - Init COMPLETE
+ip-26-0-166-244:897241:897907 [1] NCCL INFO Channel 02/1 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA/Shared
+ip-26-0-166-244:897241:897907 [1] NCCL INFO Channel 03/1 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA/Shared
+ip-26-0-166-244:897244:897877 [4] NCCL INFO comm 0x13c9b9e0 rank 0 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0xa727c3bcbb6064de - Init COMPLETE
+ip-26-0-166-244:897244:897908 [4] NCCL INFO Channel 02/1 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA/Shared
+ip-26-0-166-244:897244:897908 [4] NCCL INFO Channel 03/1 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA/Shared
+ip-26-0-166-244:897242:897881 [2] NCCL INFO Connected all rings
+ip-26-0-166-244:897242:897881 [2] NCCL INFO Connected all trees
+ip-26-0-166-244:897242:897881 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-244:897242:897881 [2] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-244:897242:897881 [2] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-244:897243:897880 [3] NCCL INFO Connected all rings
+ip-26-0-166-244:897243:897880 [3] NCCL INFO Connected all trees
+ip-26-0-166-244:897243:897880 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-244:897243:897880 [3] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-244:897243:897880 [3] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-36:104435:105039 [6] NCCL INFO comm 0xb175c80 rank 1 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0x3d478d20852d788d - Init COMPLETE
+ip-26-0-166-36:104435:105052 [6] NCCL INFO Channel 02/1 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA/Shared
+ip-26-0-166-36:104435:105052 [6] NCCL INFO Channel 03/1 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA/Shared
+ip-26-0-166-36:104436:105040 [7] NCCL INFO comm 0xbaeacb0 rank 1 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0x6a566fb09d3bd505 - Init COMPLETE
+ip-26-0-166-36:104436:105053 [7] NCCL INFO Channel 02/1 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA/Shared
+ip-26-0-166-36:104436:105053 [7] NCCL INFO Channel 03/1 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA/Shared
+ip-26-0-166-244:897245:897875 [5] NCCL INFO comm 0x148b8d50 rank 0 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0x4aca07c1fffe5f34 - Init COMPLETE
+ip-26-0-166-244:897245:897909 [5] NCCL INFO Channel 02/1 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA/Shared
+ip-26-0-166-244:897245:897909 [5] NCCL INFO Channel 03/1 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA/Shared
+ip-26-0-166-244:897242:897881 [2] NCCL INFO comm 0x13505950 rank 0 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0x9929739d002d583e - Init COMPLETE
+ip-26-0-166-244:897243:897880 [3] NCCL INFO comm 0x1380dee0 rank 0 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0xf552a17783bb308 - Init COMPLETE
+ip-26-0-166-244:897243:897910 [3] NCCL INFO Channel 02/1 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA/Shared
+ip-26-0-166-244:897242:897911 [2] NCCL INFO Channel 02/1 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA/Shared
+ip-26-0-166-244:897243:897910 [3] NCCL INFO Channel 03/1 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA/Shared
+ip-26-0-166-244:897242:897911 [2] NCCL INFO Channel 03/1 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA/Shared
+ip-26-0-166-244:897246:897896 [6] NCCL INFO Connected all rings
+ip-26-0-166-244:897246:897896 [6] NCCL INFO Connected all trees
+ip-26-0-166-244:897246:897896 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-244:897246:897896 [6] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-244:897246:897896 [6] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-244:897247:897898 [7] NCCL INFO Connected all rings
+ip-26-0-166-244:897247:897898 [7] NCCL INFO Connected all trees
+ip-26-0-166-244:897247:897898 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-244:897247:897898 [7] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-244:897247:897898 [7] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-244:897246:897896 [6] NCCL INFO comm 0x15228320 rank 0 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0x3d478d20852d788d - Init COMPLETE
+ip-26-0-166-244:897246:897912 [6] NCCL INFO Channel 02/1 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA/Shared
+ip-26-0-166-244:897246:897912 [6] NCCL INFO Channel 03/1 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA/Shared
+ip-26-0-166-244:897247:897898 [7] NCCL INFO comm 0x147aa4a0 rank 0 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0x6a566fb09d3bd505 - Init COMPLETE
+ip-26-0-166-244:897247:897913 [7] NCCL INFO Channel 02/1 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA/Shared
+ip-26-0-166-244:897247:897913 [7] NCCL INFO Channel 03/1 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA/Shared
+ip-26-0-166-244:897240:897930 [0] NCCL INFO Channel 02/1 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA/Shared
+ip-26-0-166-244:897241:897931 [1] NCCL INFO Channel 02/1 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA/Shared
+ip-26-0-166-244:897240:897930 [0] NCCL INFO Channel 03/1 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA/Shared
+ip-26-0-166-244:897241:897931 [1] NCCL INFO Channel 03/1 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA/Shared
+ip-26-0-166-244:897244:897949 [4] NCCL INFO Channel 02/1 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA/Shared
+ip-26-0-166-244:897245:897948 [5] NCCL INFO Channel 02/1 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA/Shared
+ip-26-0-166-244:897244:897949 [4] NCCL INFO Channel 03/1 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA/Shared
+ip-26-0-166-244:897245:897948 [5] NCCL INFO Channel 03/1 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA/Shared
+ip-26-0-166-244:897247:897966 [7] NCCL INFO Channel 02/1 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA/Shared
+ip-26-0-166-244:897247:897966 [7] NCCL INFO Channel 03/1 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA/Shared
+ip-26-0-166-244:897246:897967 [6] NCCL INFO Channel 02/1 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA/Shared
+ip-26-0-166-244:897246:897967 [6] NCCL INFO Channel 03/1 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA/Shared
+ip-26-0-166-244:897243:897984 [3] NCCL INFO Channel 02/1 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA/Shared
+ip-26-0-166-244:897242:897985 [2] NCCL INFO Channel 02/1 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA/Shared
+ip-26-0-166-244:897243:897984 [3] NCCL INFO Channel 03/1 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA/Shared
+ip-26-0-166-244:897242:897985 [2] NCCL INFO Channel 03/1 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA/Shared
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+ip-26-0-166-36:104429:105058 [0] NCCL INFO Using network Libfabric
+ip-26-0-166-36:104430:105057 [1] NCCL INFO Using network Libfabric
+ip-26-0-167-9:1039537:1040181 [0] NCCL INFO Using network Libfabric
+ip-26-0-167-9:1039537:1040181 [0] NCCL INFO comm 0xc630ff0 rank 1 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0x600087803401a0c5 - Init START
+ip-26-0-166-36:104429:105058 [0] NCCL INFO comm 0x1525eea0 rank 0 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0x600087803401a0c5 - Init START
+ip-26-0-167-9:1039538:1040182 [1] NCCL INFO Using network Libfabric
+ip-26-0-166-36:104429:105058 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-167-9:1039537:1040181 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-167-9:1039538:1040182 [1] NCCL INFO comm 0xba2e2a0 rank 1 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0x489b8266c05991e9 - Init START
+ip-26-0-166-36:104430:105057 [1] NCCL INFO comm 0x14a74fa0 rank 0 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0x489b8266c05991e9 - Init START
+ip-26-0-166-36:104430:105057 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-167-9:1039538:1040182 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+NCCL version 2.18.6+cuda12.1
+ip-26-0-166-36:104433:105060 [4] NCCL INFO Using network Libfabric
+ip-26-0-167-9:1039541:1040183 [4] NCCL INFO Using network Libfabric
+ip-26-0-167-9:1039542:1040184 [5] NCCL INFO Using network Libfabric
+ip-26-0-166-36:104434:105062 [5] NCCL INFO Using network Libfabric
+ip-26-0-167-9:1039541:1040183 [4] NCCL INFO comm 0xce81720 rank 1 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0x422bf6f52362d66c - Init START
+ip-26-0-166-36:104433:105060 [4] NCCL INFO comm 0x1504c3e0 rank 0 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0x422bf6f52362d66c - Init START
+ip-26-0-167-9:1039542:1040184 [5] NCCL INFO comm 0xb842a80 rank 1 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0x5a0d02cdc630de22 - Init START
+ip-26-0-167-9:1039541:1040183 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-36:104433:105060 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-36:104434:105062 [5] NCCL INFO comm 0x13866660 rank 0 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0x5a0d02cdc630de22 - Init START
+ip-26-0-167-9:1039542:1040184 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-36:104434:105062 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+NCCL version 2.18.6+cuda12.1
+ip-26-0-166-36:104431:105064 [2] NCCL INFO Using network Libfabric
+ip-26-0-166-36:104432:105066 [3] NCCL INFO Using network Libfabric
+ip-26-0-167-9:1039540:1040186 [3] NCCL INFO Using network Libfabric
+ip-26-0-167-9:1039539:1040185 [2] NCCL INFO Using network Libfabric
+ip-26-0-167-9:1039540:1040186 [3] NCCL INFO comm 0xc17f580 rank 1 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0x2fa074fb70716c2c - Init START
+ip-26-0-166-36:104432:105066 [3] NCCL INFO comm 0x135fd710 rank 0 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0x2fa074fb70716c2c - Init START
+ip-26-0-166-36:104431:105064 [2] NCCL INFO comm 0x146043e0 rank 0 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0x2f9d04618ca6d246 - Init START
+ip-26-0-167-9:1039539:1040185 [2] NCCL INFO comm 0xbb055e0 rank 1 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0x2f9d04618ca6d246 - Init START
+ip-26-0-166-36:104432:105066 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-167-9:1039540:1040186 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-36:104431:105064 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-167-9:1039539:1040185 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+NCCL version 2.18.6+cuda12.1
+ip-26-0-166-36:104435:105069 [6] NCCL INFO Using network Libfabric
+ip-26-0-166-36:104436:105070 [7] NCCL INFO Using network Libfabric
+ip-26-0-167-9:1039543:1040187 [6] NCCL INFO Using network Libfabric
+ip-26-0-167-9:1039544:1040188 [7] NCCL INFO Using network Libfabric
+ip-26-0-166-36:104436:105070 [7] NCCL INFO comm 0x14534600 rank 0 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0xc4b6f822cb496b78 - Init START
+ip-26-0-167-9:1039544:1040188 [7] NCCL INFO comm 0xc08bc20 rank 1 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0xc4b6f822cb496b78 - Init START
+ip-26-0-166-36:104435:105069 [6] NCCL INFO comm 0x13bc1410 rank 0 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0xa54e21ce8d7ec3fe - Init START
+ip-26-0-167-9:1039543:1040187 [6] NCCL INFO comm 0xc951f40 rank 1 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0xa54e21ce8d7ec3fe - Init START
+ip-26-0-166-36:104436:105070 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-167-9:1039544:1040188 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-36:104435:105069 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-167-9:1039543:1040187 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-36:104430:105057 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-167-9:1039537:1040181 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-167-9:1039538:1040182 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-167-9:1039538:1040182 [1] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-167-9:1039538:1040182 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-36:104430:105057 [1] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-166-36:104430:105057 [1] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-166-36:104430:105057 [1] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-166-36:104430:105057 [1] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-166-36:104430:105057 [1] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-166-36:104430:105057 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-167-9:1039542:1040184 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-166-36:104429:105058 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-166-36:104429:105058 [0] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-166-36:104429:105058 [0] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-166-36:104429:105058 [0] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-166-36:104429:105058 [0] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-166-36:104429:105058 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-166-36:104429:105058 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-167-9:1039537:1040181 [0] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-167-9:1039537:1040181 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-36:104433:105060 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-166-36:104434:105062 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-166-36:104434:105062 [5] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-166-36:104434:105062 [5] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-166-36:104434:105062 [5] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-166-36:104434:105062 [5] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-166-36:104434:105062 [5] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-166-36:104434:105062 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-167-9:1039542:1040184 [5] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-167-9:1039542:1040184 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-167-9:1039541:1040183 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-167-9:1039539:1040185 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-167-9:1039541:1040183 [4] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-167-9:1039541:1040183 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-36:104433:105060 [4] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-166-36:104433:105060 [4] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-166-36:104433:105060 [4] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-166-36:104433:105060 [4] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-166-36:104433:105060 [4] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-166-36:104433:105060 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-167-9:1039544:1040188 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-166-36:104430:105057 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104430:105057 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104430:105057 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104430:105057 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104430:105057 [1] NCCL INFO Channel 00/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104430:105057 [1] NCCL INFO Channel 01/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104430:105057 [1] NCCL INFO Channel 02/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104430:105057 [1] NCCL INFO Channel 03/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-167-9:1039538:1040182 [1] NCCL INFO Channel 00/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-167-9:1039538:1040182 [1] NCCL INFO Channel 01/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-167-9:1039538:1040182 [1] NCCL INFO Channel 02/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-167-9:1039538:1040182 [1] NCCL INFO Channel 03/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-167-9:1039538:1040182 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-167-9:1039538:1040182 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-167-9:1039538:1040182 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-167-9:1039538:1040182 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-167-9:1039540:1040186 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-167-9:1039543:1040187 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-166-36:104436:105070 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-166-36:104436:105070 [7] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-166-36:104436:105070 [7] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-166-36:104436:105070 [7] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-166-36:104436:105070 [7] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-166-36:104436:105070 [7] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-166-36:104436:105070 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-167-9:1039544:1040188 [7] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-167-9:1039544:1040188 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-36:104429:105058 [0] NCCL INFO Channel 00/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104429:105058 [0] NCCL INFO Channel 01/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104429:105058 [0] NCCL INFO Channel 02/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104429:105058 [0] NCCL INFO Channel 03/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104429:105058 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104429:105058 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104429:105058 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104429:105058 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-167-9:1039537:1040181 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-167-9:1039537:1040181 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-167-9:1039537:1040181 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-167-9:1039537:1040181 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-167-9:1039537:1040181 [0] NCCL INFO Channel 00/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-167-9:1039537:1040181 [0] NCCL INFO Channel 01/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-167-9:1039537:1040181 [0] NCCL INFO Channel 02/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-167-9:1039537:1040181 [0] NCCL INFO Channel 03/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104435:105069 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-166-36:104435:105069 [6] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-166-36:104435:105069 [6] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-166-36:104435:105069 [6] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-166-36:104435:105069 [6] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-166-36:104435:105069 [6] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-166-36:104435:105069 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-167-9:1039543:1040187 [6] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-167-9:1039543:1040187 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-36:104431:105064 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-166-36:104431:105064 [2] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-166-36:104431:105064 [2] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-166-36:104431:105064 [2] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-166-36:104431:105064 [2] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-166-36:104431:105064 [2] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-166-36:104431:105064 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-167-9:1039539:1040185 [2] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-167-9:1039539:1040185 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-36:104432:105066 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-166-36:104432:105066 [3] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-166-36:104432:105066 [3] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-166-36:104432:105066 [3] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-166-36:104432:105066 [3] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-166-36:104432:105066 [3] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-166-36:104432:105066 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-167-9:1039540:1040186 [3] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-167-9:1039540:1040186 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-36:104434:105062 [5] NCCL INFO Channel 00/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104434:105062 [5] NCCL INFO Channel 01/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104434:105062 [5] NCCL INFO Channel 02/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104434:105062 [5] NCCL INFO Channel 03/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104434:105062 [5] NCCL INFO Channel 00/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104434:105062 [5] NCCL INFO Channel 01/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104434:105062 [5] NCCL INFO Channel 02/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104434:105062 [5] NCCL INFO Channel 03/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104433:105060 [4] NCCL INFO Channel 00/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104433:105060 [4] NCCL INFO Channel 01/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104433:105060 [4] NCCL INFO Channel 02/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104433:105060 [4] NCCL INFO Channel 03/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104433:105060 [4] NCCL INFO Channel 00/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104433:105060 [4] NCCL INFO Channel 01/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-167-9:1039542:1040184 [5] NCCL INFO Channel 00/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104433:105060 [4] NCCL INFO Channel 02/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-167-9:1039542:1040184 [5] NCCL INFO Channel 01/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104433:105060 [4] NCCL INFO Channel 03/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-167-9:1039542:1040184 [5] NCCL INFO Channel 02/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-167-9:1039542:1040184 [5] NCCL INFO Channel 03/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-167-9:1039542:1040184 [5] NCCL INFO Channel 00/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-167-9:1039542:1040184 [5] NCCL INFO Channel 01/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-167-9:1039542:1040184 [5] NCCL INFO Channel 02/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-167-9:1039542:1040184 [5] NCCL INFO Channel 03/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-167-9:1039541:1040183 [4] NCCL INFO Channel 00/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-167-9:1039541:1040183 [4] NCCL INFO Channel 01/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-167-9:1039541:1040183 [4] NCCL INFO Channel 02/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-167-9:1039541:1040183 [4] NCCL INFO Channel 03/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-167-9:1039541:1040183 [4] NCCL INFO Channel 00/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-167-9:1039541:1040183 [4] NCCL INFO Channel 01/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-167-9:1039541:1040183 [4] NCCL INFO Channel 02/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-167-9:1039541:1040183 [4] NCCL INFO Channel 03/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104436:105070 [7] NCCL INFO Channel 00/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104436:105070 [7] NCCL INFO Channel 01/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104436:105070 [7] NCCL INFO Channel 02/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104436:105070 [7] NCCL INFO Channel 03/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104436:105070 [7] NCCL INFO Channel 00/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104436:105070 [7] NCCL INFO Channel 01/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104436:105070 [7] NCCL INFO Channel 02/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104436:105070 [7] NCCL INFO Channel 03/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-167-9:1039544:1040188 [7] NCCL INFO Channel 00/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-167-9:1039544:1040188 [7] NCCL INFO Channel 01/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-167-9:1039544:1040188 [7] NCCL INFO Channel 02/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-167-9:1039544:1040188 [7] NCCL INFO Channel 03/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-167-9:1039544:1040188 [7] NCCL INFO Channel 00/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-167-9:1039544:1040188 [7] NCCL INFO Channel 01/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-167-9:1039544:1040188 [7] NCCL INFO Channel 02/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-167-9:1039544:1040188 [7] NCCL INFO Channel 03/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-167-9:1039543:1040187 [6] NCCL INFO Channel 00/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-167-9:1039543:1040187 [6] NCCL INFO Channel 01/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-167-9:1039543:1040187 [6] NCCL INFO Channel 02/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-167-9:1039543:1040187 [6] NCCL INFO Channel 03/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-167-9:1039543:1040187 [6] NCCL INFO Channel 00/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-167-9:1039543:1040187 [6] NCCL INFO Channel 01/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-167-9:1039543:1040187 [6] NCCL INFO Channel 02/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-167-9:1039543:1040187 [6] NCCL INFO Channel 03/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104435:105069 [6] NCCL INFO Channel 00/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104435:105069 [6] NCCL INFO Channel 01/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104435:105069 [6] NCCL INFO Channel 02/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104435:105069 [6] NCCL INFO Channel 03/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104435:105069 [6] NCCL INFO Channel 00/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104435:105069 [6] NCCL INFO Channel 01/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104435:105069 [6] NCCL INFO Channel 02/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104435:105069 [6] NCCL INFO Channel 03/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-167-9:1039539:1040185 [2] NCCL INFO Channel 00/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-167-9:1039539:1040185 [2] NCCL INFO Channel 01/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104431:105064 [2] NCCL INFO Channel 00/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-167-9:1039539:1040185 [2] NCCL INFO Channel 02/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104431:105064 [2] NCCL INFO Channel 01/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-167-9:1039539:1040185 [2] NCCL INFO Channel 03/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104431:105064 [2] NCCL INFO Channel 02/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-167-9:1039539:1040185 [2] NCCL INFO Channel 00/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104431:105064 [2] NCCL INFO Channel 03/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-167-9:1039539:1040185 [2] NCCL INFO Channel 01/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104431:105064 [2] NCCL INFO Channel 00/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-167-9:1039539:1040185 [2] NCCL INFO Channel 02/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104431:105064 [2] NCCL INFO Channel 01/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-167-9:1039539:1040185 [2] NCCL INFO Channel 03/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104431:105064 [2] NCCL INFO Channel 02/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104431:105064 [2] NCCL INFO Channel 03/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-167-9:1039540:1040186 [3] NCCL INFO Channel 00/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-167-9:1039540:1040186 [3] NCCL INFO Channel 01/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-167-9:1039540:1040186 [3] NCCL INFO Channel 02/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-167-9:1039540:1040186 [3] NCCL INFO Channel 03/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-167-9:1039540:1040186 [3] NCCL INFO Channel 00/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-167-9:1039540:1040186 [3] NCCL INFO Channel 01/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-167-9:1039540:1040186 [3] NCCL INFO Channel 02/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-167-9:1039540:1040186 [3] NCCL INFO Channel 03/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104432:105066 [3] NCCL INFO Channel 00/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104432:105066 [3] NCCL INFO Channel 01/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104432:105066 [3] NCCL INFO Channel 02/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104432:105066 [3] NCCL INFO Channel 03/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104432:105066 [3] NCCL INFO Channel 00/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104432:105066 [3] NCCL INFO Channel 01/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104432:105066 [3] NCCL INFO Channel 02/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104432:105066 [3] NCCL INFO Channel 03/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104430:105057 [1] NCCL INFO Connected all rings
+ip-26-0-166-36:104430:105057 [1] NCCL INFO Connected all trees
+ip-26-0-166-36:104430:105057 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-36:104430:105057 [1] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-36:104430:105057 [1] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-167-9:1039538:1040182 [1] NCCL INFO Connected all rings
+ip-26-0-167-9:1039538:1040182 [1] NCCL INFO Connected all trees
+ip-26-0-167-9:1039538:1040182 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-167-9:1039538:1040182 [1] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-167-9:1039538:1040182 [1] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-36:104430:105057 [1] NCCL INFO comm 0x14a74fa0 rank 0 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0x489b8266c05991e9 - Init COMPLETE
+ip-26-0-166-36:104430:105088 [1] NCCL INFO Channel 02/1 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA/Shared
+ip-26-0-166-36:104430:105088 [1] NCCL INFO Channel 03/1 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA/Shared
+ip-26-0-167-9:1039538:1040182 [1] NCCL INFO comm 0xba2e2a0 rank 1 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0x489b8266c05991e9 - Init COMPLETE
+ip-26-0-167-9:1039538:1040206 [1] NCCL INFO Channel 02/1 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA/Shared
+ip-26-0-167-9:1039538:1040206 [1] NCCL INFO Channel 03/1 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA/Shared
+ip-26-0-167-9:1039537:1040181 [0] NCCL INFO Connected all rings
+ip-26-0-167-9:1039537:1040181 [0] NCCL INFO Connected all trees
+ip-26-0-167-9:1039537:1040181 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-167-9:1039537:1040181 [0] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-167-9:1039537:1040181 [0] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-36:104429:105058 [0] NCCL INFO Connected all rings
+ip-26-0-166-36:104429:105058 [0] NCCL INFO Connected all trees
+ip-26-0-166-36:104429:105058 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-36:104429:105058 [0] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-36:104429:105058 [0] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-36:104429:105058 [0] NCCL INFO comm 0x1525eea0 rank 0 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0x600087803401a0c5 - Init COMPLETE
+ip-26-0-167-9:1039537:1040181 [0] NCCL INFO comm 0xc630ff0 rank 1 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0x600087803401a0c5 - Init COMPLETE
+ip-26-0-166-36:104429:105089 [0] NCCL INFO Channel 02/1 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA/Shared
+ip-26-0-166-36:104429:105089 [0] NCCL INFO Channel 03/1 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA/Shared
+ip-26-0-167-9:1039537:1040207 [0] NCCL INFO Channel 02/1 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA/Shared
+ip-26-0-167-9:1039537:1040207 [0] NCCL INFO Channel 03/1 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA/Shared
+ip-26-0-166-36:104433:105060 [4] NCCL INFO Connected all rings
+ip-26-0-166-36:104433:105060 [4] NCCL INFO Connected all trees
+ip-26-0-166-36:104433:105060 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-36:104433:105060 [4] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-36:104433:105060 [4] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-167-9:1039541:1040183 [4] NCCL INFO Connected all rings
+ip-26-0-167-9:1039541:1040183 [4] NCCL INFO Connected all trees
+ip-26-0-167-9:1039541:1040183 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-167-9:1039541:1040183 [4] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-167-9:1039541:1040183 [4] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-36:104434:105062 [5] NCCL INFO Connected all rings
+ip-26-0-166-36:104434:105062 [5] NCCL INFO Connected all trees
+ip-26-0-166-36:104434:105062 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-36:104434:105062 [5] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-36:104434:105062 [5] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-167-9:1039542:1040184 [5] NCCL INFO Connected all rings
+ip-26-0-167-9:1039542:1040184 [5] NCCL INFO Connected all trees
+ip-26-0-167-9:1039542:1040184 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-167-9:1039542:1040184 [5] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-167-9:1039542:1040184 [5] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-36:104433:105060 [4] NCCL INFO comm 0x1504c3e0 rank 0 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0x422bf6f52362d66c - Init COMPLETE
+ip-26-0-167-9:1039541:1040183 [4] NCCL INFO comm 0xce81720 rank 1 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0x422bf6f52362d66c - Init COMPLETE
+ip-26-0-166-36:104433:105090 [4] NCCL INFO Channel 02/1 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA/Shared
+ip-26-0-166-36:104433:105090 [4] NCCL INFO Channel 03/1 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA/Shared
+ip-26-0-167-9:1039541:1040208 [4] NCCL INFO Channel 02/1 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA/Shared
+ip-26-0-167-9:1039541:1040208 [4] NCCL INFO Channel 03/1 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA/Shared
+ip-26-0-166-36:104434:105062 [5] NCCL INFO comm 0x13866660 rank 0 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0x5a0d02cdc630de22 - Init COMPLETE
+ip-26-0-166-36:104434:105091 [5] NCCL INFO Channel 02/1 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA/Shared
+ip-26-0-166-36:104434:105091 [5] NCCL INFO Channel 03/1 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA/Shared
+ip-26-0-166-36:104436:105070 [7] NCCL INFO Connected all rings
+ip-26-0-166-36:104436:105070 [7] NCCL INFO Connected all trees
+ip-26-0-166-36:104436:105070 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-36:104436:105070 [7] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-36:104436:105070 [7] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-167-9:1039542:1040184 [5] NCCL INFO comm 0xb842a80 rank 1 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0x5a0d02cdc630de22 - Init COMPLETE
+ip-26-0-167-9:1039542:1040209 [5] NCCL INFO Channel 02/1 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA/Shared
+ip-26-0-167-9:1039542:1040209 [5] NCCL INFO Channel 03/1 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA/Shared
+ip-26-0-167-9:1039544:1040188 [7] NCCL INFO Connected all rings
+ip-26-0-167-9:1039544:1040188 [7] NCCL INFO Connected all trees
+ip-26-0-167-9:1039544:1040188 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-167-9:1039544:1040188 [7] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-167-9:1039544:1040188 [7] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-36:104436:105070 [7] NCCL INFO comm 0x14534600 rank 0 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0xc4b6f822cb496b78 - Init COMPLETE
+ip-26-0-166-36:104436:105092 [7] NCCL INFO Channel 02/1 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA/Shared
+ip-26-0-166-36:104436:105092 [7] NCCL INFO Channel 03/1 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA/Shared
+ip-26-0-166-36:104431:105064 [2] NCCL INFO Connected all rings
+ip-26-0-166-36:104431:105064 [2] NCCL INFO Connected all trees
+ip-26-0-166-36:104431:105064 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-36:104431:105064 [2] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-36:104431:105064 [2] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-167-9:1039544:1040188 [7] NCCL INFO comm 0xc08bc20 rank 1 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0xc4b6f822cb496b78 - Init COMPLETE
+ip-26-0-167-9:1039544:1040210 [7] NCCL INFO Channel 02/1 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA/Shared
+ip-26-0-167-9:1039544:1040210 [7] NCCL INFO Channel 03/1 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA/Shared
+ip-26-0-166-36:104432:105066 [3] NCCL INFO Connected all rings
+ip-26-0-166-36:104432:105066 [3] NCCL INFO Connected all trees
+ip-26-0-166-36:104432:105066 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-36:104432:105066 [3] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-36:104432:105066 [3] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-36:104435:105069 [6] NCCL INFO Connected all rings
+ip-26-0-166-36:104435:105069 [6] NCCL INFO Connected all trees
+ip-26-0-166-36:104435:105069 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-36:104435:105069 [6] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-166-36:104435:105069 [6] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-167-9:1039539:1040185 [2] NCCL INFO Connected all rings
+ip-26-0-167-9:1039539:1040185 [2] NCCL INFO Connected all trees
+ip-26-0-167-9:1039539:1040185 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-167-9:1039539:1040185 [2] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-167-9:1039539:1040185 [2] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-36:104431:105064 [2] NCCL INFO comm 0x146043e0 rank 0 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0x2f9d04618ca6d246 - Init COMPLETE
+ip-26-0-166-36:104431:105093 [2] NCCL INFO Channel 02/1 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA/Shared
+ip-26-0-166-36:104431:105093 [2] NCCL INFO Channel 03/1 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA/Shared
+ip-26-0-167-9:1039543:1040187 [6] NCCL INFO Connected all rings
+ip-26-0-167-9:1039543:1040187 [6] NCCL INFO Connected all trees
+ip-26-0-167-9:1039543:1040187 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-167-9:1039543:1040187 [6] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-167-9:1039543:1040187 [6] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-167-9:1039540:1040186 [3] NCCL INFO Connected all rings
+ip-26-0-167-9:1039540:1040186 [3] NCCL INFO Connected all trees
+ip-26-0-167-9:1039540:1040186 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-167-9:1039540:1040186 [3] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-167-9:1039540:1040186 [3] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-166-36:104432:105066 [3] NCCL INFO comm 0x135fd710 rank 0 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0x2fa074fb70716c2c - Init COMPLETE
+ip-26-0-166-36:104435:105069 [6] NCCL INFO comm 0x13bc1410 rank 0 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0xa54e21ce8d7ec3fe - Init COMPLETE
+ip-26-0-166-36:104432:105095 [3] NCCL INFO Channel 02/1 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA/Shared
+ip-26-0-166-36:104435:105094 [6] NCCL INFO Channel 02/1 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA/Shared
+ip-26-0-166-36:104432:105095 [3] NCCL INFO Channel 03/1 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA/Shared
+ip-26-0-166-36:104435:105094 [6] NCCL INFO Channel 03/1 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA/Shared
+ip-26-0-167-9:1039539:1040185 [2] NCCL INFO comm 0xbb055e0 rank 1 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0x2f9d04618ca6d246 - Init COMPLETE
+ip-26-0-167-9:1039539:1040211 [2] NCCL INFO Channel 02/1 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA/Shared
+ip-26-0-167-9:1039539:1040211 [2] NCCL INFO Channel 03/1 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA/Shared
+ip-26-0-167-9:1039543:1040187 [6] NCCL INFO comm 0xc951f40 rank 1 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0xa54e21ce8d7ec3fe - Init COMPLETE
+ip-26-0-167-9:1039543:1040212 [6] NCCL INFO Channel 02/1 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA/Shared
+ip-26-0-167-9:1039543:1040212 [6] NCCL INFO Channel 03/1 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA/Shared
+ip-26-0-167-9:1039540:1040186 [3] NCCL INFO comm 0xc17f580 rank 1 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0x2fa074fb70716c2c - Init COMPLETE
+ip-26-0-167-9:1039540:1040213 [3] NCCL INFO Channel 02/1 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA/Shared
+ip-26-0-167-9:1039540:1040213 [3] NCCL INFO Channel 03/1 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA/Shared
+ip-26-0-166-36:104430:105112 [1] NCCL INFO Channel 02/1 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA/Shared
+ip-26-0-166-36:104430:105112 [1] NCCL INFO Channel 03/1 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA/Shared
+ip-26-0-166-36:104429:105113 [0] NCCL INFO Channel 02/1 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA/Shared
+ip-26-0-166-36:104429:105113 [0] NCCL INFO Channel 03/1 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA/Shared
+ip-26-0-166-36:104433:105130 [4] NCCL INFO Channel 02/1 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA/Shared
+ip-26-0-166-36:104433:105130 [4] NCCL INFO Channel 03/1 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA/Shared
+ip-26-0-166-36:104434:105131 [5] NCCL INFO Channel 02/1 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA/Shared
+ip-26-0-166-36:104434:105131 [5] NCCL INFO Channel 03/1 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA/Shared
+ip-26-0-166-36:104435:105149 [6] NCCL INFO Channel 02/1 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA/Shared
+ip-26-0-166-36:104435:105149 [6] NCCL INFO Channel 03/1 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA/Shared
+ip-26-0-166-36:104436:105150 [7] NCCL INFO Channel 02/1 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA/Shared
+ip-26-0-166-36:104436:105150 [7] NCCL INFO Channel 03/1 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA/Shared
+ip-26-0-166-36:104431:105168 [2] NCCL INFO Channel 02/1 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA/Shared
+ip-26-0-166-36:104432:105167 [3] NCCL INFO Channel 02/1 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA/Shared
+ip-26-0-166-36:104432:105167 [3] NCCL INFO Channel 03/1 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA/Shared
+ip-26-0-166-36:104431:105168 [2] NCCL INFO Channel 03/1 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA/Shared
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+ip-26-0-167-9:1039538:1040298 [1] NCCL INFO Channel 02/1 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA/Shared
+ip-26-0-167-9:1039538:1040298 [1] NCCL INFO Channel 03/1 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA/Shared
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+ip-26-0-167-9:1039537:1040299 [0] NCCL INFO Channel 02/1 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA/Shared
+ip-26-0-167-9:1039537:1040299 [0] NCCL INFO Channel 03/1 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA/Shared
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+ip-26-0-167-9:1039542:1040300 [5] NCCL INFO Channel 02/1 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA/Shared
+ip-26-0-167-9:1039542:1040300 [5] NCCL INFO Channel 03/1 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA/Shared
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+ip-26-0-167-9:1039541:1040301 [4] NCCL INFO Channel 02/1 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA/Shared
+ip-26-0-167-9:1039541:1040301 [4] NCCL INFO Channel 03/1 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA/Shared
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+ip-26-0-167-9:1039544:1040302 [7] NCCL INFO Channel 02/1 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA/Shared
+ip-26-0-167-9:1039544:1040302 [7] NCCL INFO Channel 03/1 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA/Shared
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+ip-26-0-167-9:1039540:1040303 [3] NCCL INFO Channel 02/1 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA/Shared
+ip-26-0-167-9:1039543:1040304 [6] NCCL INFO Channel 02/1 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA/Shared
+ip-26-0-167-9:1039540:1040303 [3] NCCL INFO Channel 03/1 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA/Shared
+ip-26-0-167-9:1039543:1040304 [6] NCCL INFO Channel 03/1 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA/Shared
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+ip-26-0-167-9:1039539:1040305 [2] NCCL INFO Channel 02/1 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA/Shared
+ip-26-0-167-9:1039539:1040305 [2] NCCL INFO Channel 03/1 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA/Shared
+ip-26-0-166-36:104430:105173 [1] NCCL INFO Channel 02/1 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA/Shared
+ip-26-0-166-36:104430:105173 [1] NCCL INFO Channel 03/1 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA/Shared
+ip-26-0-166-36:104429:105174 [0] NCCL INFO Channel 02/1 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA/Shared
+ip-26-0-166-36:104429:105174 [0] NCCL INFO Channel 03/1 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA/Shared
+ip-26-0-166-36:104434:105175 [5] NCCL INFO Channel 02/1 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA/Shared
+ip-26-0-166-36:104434:105175 [5] NCCL INFO Channel 03/1 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA/Shared
+ip-26-0-166-36:104433:105176 [4] NCCL INFO Channel 02/1 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA/Shared
+ip-26-0-166-36:104433:105176 [4] NCCL INFO Channel 03/1 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA/Shared
+ip-26-0-166-36:104431:105178 [2] NCCL INFO Channel 02/1 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA/Shared
+ip-26-0-166-36:104432:105179 [3] NCCL INFO Channel 02/1 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA/Shared
+ip-26-0-166-36:104431:105178 [2] NCCL INFO Channel 03/1 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA/Shared
+ip-26-0-166-36:104432:105179 [3] NCCL INFO Channel 03/1 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA/Shared
+ip-26-0-166-36:104435:105180 [6] NCCL INFO Channel 02/1 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA/Shared
+ip-26-0-166-36:104435:105180 [6] NCCL INFO Channel 03/1 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA/Shared
+ip-26-0-166-36:104436:105181 [7] NCCL INFO Channel 02/1 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA/Shared
+ip-26-0-166-36:104436:105181 [7] NCCL INFO Channel 03/1 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA/Shared
+ip-26-0-166-244:897241:898002 [1] NCCL INFO Channel 02/1 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA/Shared
+ip-26-0-166-244:897241:898002 [1] NCCL INFO Channel 03/1 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA/Shared
+ip-26-0-166-244:897240:898003 [0] NCCL INFO Channel 02/1 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA/Shared
+ip-26-0-166-244:897240:898003 [0] NCCL INFO Channel 03/1 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA/Shared
+ip-26-0-166-244:897245:898004 [5] NCCL INFO Channel 02/1 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA/Shared
+ip-26-0-166-244:897245:898004 [5] NCCL INFO Channel 03/1 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA/Shared
+ip-26-0-166-244:897244:898005 [4] NCCL INFO Channel 02/1 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA/Shared
+ip-26-0-166-244:897244:898005 [4] NCCL INFO Channel 03/1 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA/Shared
+ip-26-0-166-244:897242:898006 [2] NCCL INFO Channel 02/1 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA/Shared
+ip-26-0-166-244:897242:898006 [2] NCCL INFO Channel 03/1 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA/Shared
+ip-26-0-166-244:897243:898007 [3] NCCL INFO Channel 02/1 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA/Shared
+ip-26-0-166-244:897243:898007 [3] NCCL INFO Channel 03/1 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA/Shared
+ip-26-0-166-244:897247:898008 [7] NCCL INFO Channel 02/1 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA/Shared
+ip-26-0-166-244:897247:898008 [7] NCCL INFO Channel 03/1 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA/Shared
+ip-26-0-166-244:897246:898009 [6] NCCL INFO Channel 02/1 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA/Shared
+ip-26-0-166-244:897246:898009 [6] NCCL INFO Channel 03/1 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA/Shared
+ip-26-0-166-214:624943:625723 [0] NCCL INFO Channel 02/1 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA/Shared
+ip-26-0-166-214:624943:625723 [0] NCCL INFO Channel 03/1 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA/Shared
+ip-26-0-166-214:624944:625724 [1] NCCL INFO Channel 02/1 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA/Shared
+ip-26-0-166-214:624944:625724 [1] NCCL INFO Channel 03/1 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA/Shared
+ip-26-0-166-214:624948:625725 [5] NCCL INFO Channel 02/1 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA/Shared
+ip-26-0-166-214:624947:625726 [4] NCCL INFO Channel 02/1 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA/Shared
+ip-26-0-166-214:624948:625725 [5] NCCL INFO Channel 03/1 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA/Shared
+ip-26-0-166-214:624947:625726 [4] NCCL INFO Channel 03/1 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA/Shared
+ip-26-0-166-214:624946:625727 [3] NCCL INFO Channel 02/1 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA/Shared
+ip-26-0-166-214:624946:625727 [3] NCCL INFO Channel 03/1 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA/Shared
+ip-26-0-166-214:624945:625728 [2] NCCL INFO Channel 02/1 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA/Shared
+ip-26-0-166-214:624945:625728 [2] NCCL INFO Channel 03/1 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA/Shared
+ip-26-0-166-214:624950:625729 [7] NCCL INFO Channel 02/1 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA/Shared
+ip-26-0-166-214:624950:625729 [7] NCCL INFO Channel 03/1 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA/Shared
+ip-26-0-166-214:624949:625730 [6] NCCL INFO Channel 02/1 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA/Shared
+ip-26-0-166-214:624949:625730 [6] NCCL INFO Channel 03/1 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA/Shared
+ip-26-0-166-15:127291:128039 [1] NCCL INFO Channel 02/1 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA/Shared
+ip-26-0-166-15:127291:128039 [1] NCCL INFO Channel 03/1 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA/Shared
+ip-26-0-166-15:127290:128040 [0] NCCL INFO Channel 02/1 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA/Shared
+ip-26-0-166-15:127290:128040 [0] NCCL INFO Channel 03/1 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA/Shared
+ip-26-0-166-15:127294:128041 [4] NCCL INFO Channel 02/1 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA/Shared
+ip-26-0-166-15:127294:128041 [4] NCCL INFO Channel 03/1 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA/Shared
+ip-26-0-166-15:127295:128042 [5] NCCL INFO Channel 02/1 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA/Shared
+ip-26-0-166-15:127295:128042 [5] NCCL INFO Channel 03/1 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA/Shared
+ip-26-0-166-15:127292:128044 [2] NCCL INFO Channel 02/1 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA/Shared
+ip-26-0-166-15:127292:128044 [2] NCCL INFO Channel 03/1 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA/Shared
+ip-26-0-166-15:127293:128045 [3] NCCL INFO Channel 02/1 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA/Shared
+ip-26-0-166-15:127293:128045 [3] NCCL INFO Channel 03/1 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA/Shared
+ip-26-0-166-15:127296:128047 [6] NCCL INFO Channel 02/1 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA/Shared
+ip-26-0-166-15:127297:128046 [7] NCCL INFO Channel 02/1 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA/Shared
+ip-26-0-166-15:127297:128046 [7] NCCL INFO Channel 03/1 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA/Shared
+ip-26-0-166-15:127296:128047 [6] NCCL INFO Channel 03/1 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA/Shared
+ip-26-0-166-125:762816:763599 [1] NCCL INFO Channel 02/1 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA/Shared
+ip-26-0-166-125:762816:763599 [1] NCCL INFO Channel 03/1 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA/Shared
+ip-26-0-166-125:762815:763600 [0] NCCL INFO Channel 02/1 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA/Shared
+ip-26-0-166-125:762815:763600 [0] NCCL INFO Channel 03/1 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA/Shared
+ip-26-0-166-125:762819:763601 [4] NCCL INFO Channel 02/1 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA/Shared
+ip-26-0-166-125:762819:763601 [4] NCCL INFO Channel 03/1 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA/Shared
+ip-26-0-166-125:762820:763602 [5] NCCL INFO Channel 02/1 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA/Shared
+ip-26-0-166-125:762820:763602 [5] NCCL INFO Channel 03/1 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA/Shared
+ip-26-0-166-125:762821:763603 [6] NCCL INFO Channel 02/1 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA/Shared
+ip-26-0-166-125:762821:763603 [6] NCCL INFO Channel 03/1 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA/Shared
+ip-26-0-166-125:762822:763604 [7] NCCL INFO Channel 02/1 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA/Shared
+ip-26-0-166-125:762822:763604 [7] NCCL INFO Channel 03/1 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA/Shared
+ip-26-0-166-125:762818:763605 [3] NCCL INFO Channel 02/1 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA/Shared
+ip-26-0-166-125:762818:763605 [3] NCCL INFO Channel 03/1 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA/Shared
+ip-26-0-166-125:762817:763606 [2] NCCL INFO Channel 02/1 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA/Shared
+ip-26-0-166-125:762817:763606 [2] NCCL INFO Channel 03/1 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA/Shared
+ip-26-0-165-213:9626:10378 [1] NCCL INFO Channel 02/1 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA/Shared
+ip-26-0-165-213:9626:10378 [1] NCCL INFO Channel 03/1 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA/Shared
+ip-26-0-165-213:9625:10379 [0] NCCL INFO Channel 02/1 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA/Shared
+ip-26-0-165-213:9625:10379 [0] NCCL INFO Channel 03/1 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA/Shared
+ip-26-0-165-213:9629:10381 [4] NCCL INFO Channel 02/1 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA/Shared
+ip-26-0-165-213:9630:10382 [5] NCCL INFO Channel 02/1 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA/Shared
+ip-26-0-165-213:9629:10381 [4] NCCL INFO Channel 03/1 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA/Shared
+ip-26-0-165-213:9630:10382 [5] NCCL INFO Channel 03/1 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA/Shared
+ip-26-0-165-213:9632:10383 [7] NCCL INFO Channel 02/1 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA/Shared
+ip-26-0-165-213:9631:10384 [6] NCCL INFO Channel 02/1 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA/Shared
+ip-26-0-165-213:9631:10384 [6] NCCL INFO Channel 03/1 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA/Shared
+ip-26-0-165-213:9632:10383 [7] NCCL INFO Channel 03/1 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA/Shared
+ip-26-0-165-213:9627:10385 [2] NCCL INFO Channel 02/1 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA/Shared
+ip-26-0-165-213:9627:10385 [2] NCCL INFO Channel 03/1 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA/Shared
+ip-26-0-165-213:9628:10386 [3] NCCL INFO Channel 02/1 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA/Shared
+ip-26-0-165-213:9628:10386 [3] NCCL INFO Channel 03/1 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA/Shared
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+01/07/2025 03:50:29 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]:  Memory usage: 1697.67MiB. Peak allocated 11870.99MiB. Peak reserved: 14070.00MiB
+ip-26-0-162-46:702880:703676 [0] NCCL INFO Using network Libfabric
+ip-26-0-162-46:702881:703677 [1] NCCL INFO Using network Libfabric
+ip-26-0-167-9:1039537:1040316 [0] NCCL INFO Using network Libfabric
+ip-26-0-166-125:762816:763612 [1] NCCL INFO Using network Libfabric
+ip-26-0-166-36:104430:105190 [1] NCCL INFO Using network Libfabric
+ip-26-0-165-213:9625:10390 [0] NCCL INFO Using network Libfabric
+ip-26-0-166-214:624943:625737 [0] NCCL INFO Using network Libfabric
+ip-26-0-166-244:897240:898030 [0] NCCL INFO Using network Libfabric
+ip-26-0-165-213:9626:10389 [1] NCCL INFO Using network Libfabric
+ip-26-0-166-214:624944:625738 [1] NCCL INFO Using network Libfabric
+ip-26-0-166-125:762815:763613 [0] NCCL INFO Using network Libfabric
+ip-26-0-166-244:897241:898029 [1] NCCL INFO Using network Libfabric
+ip-26-0-166-36:104429:105191 [0] NCCL INFO Using network Libfabric
+ip-26-0-166-15:127290:128054 [0] NCCL INFO Using network Libfabric
+ip-26-0-167-9:1039538:1040317 [1] NCCL INFO Using network Libfabric
+ip-26-0-162-46:702884:703679 [4] NCCL INFO Using network Libfabric
+ip-26-0-162-46:702885:703680 [5] NCCL INFO Using network Libfabric
+ip-26-0-166-125:762819:763614 [4] NCCL INFO Using network Libfabric
+ip-26-0-166-36:104434:105192 [5] NCCL INFO Using network Libfabric
+ip-26-0-165-213:9630:10392 [5] NCCL INFO Using network Libfabric
+ip-26-0-166-125:762820:763615 [5] NCCL INFO Using network Libfabric
+ip-26-0-166-214:624947:625739 [4] NCCL INFO Using network Libfabric
+ip-26-0-166-15:127295:128055 [5] NCCL INFO Using network Libfabric
+ip-26-0-166-244:897244:898032 [4] NCCL INFO Using network Libfabric
+ip-26-0-167-9:1039541:1040318 [4] NCCL INFO Using network Libfabric
+ip-26-0-166-15:127294:128056 [4] NCCL INFO Using network Libfabric
+ip-26-0-166-244:897245:898031 [5] NCCL INFO Using network Libfabric
+ip-26-0-165-213:9629:10391 [4] NCCL INFO Using network Libfabric
+ip-26-0-166-36:104433:105193 [4] NCCL INFO Using network Libfabric
+ip-26-0-166-214:624948:625740 [5] NCCL INFO Using network Libfabric
+ip-26-0-166-15:127291:128053 [1] NCCL INFO Using network Libfabric
+ip-26-0-167-9:1039542:1040319 [5] NCCL INFO Using network Libfabric
+ip-26-0-162-46:702882:703682 [2] NCCL INFO Using network Libfabric
+ip-26-0-167-9:1039540:1040320 [3] NCCL INFO Using network Libfabric
+ip-26-0-166-214:624945:625741 [2] NCCL INFO Using network Libfabric
+ip-26-0-162-46:702883:703683 [3] NCCL INFO Using network Libfabric
+ip-26-0-165-213:9627:10393 [2] NCCL INFO Using network Libfabric
+ip-26-0-166-214:624946:625742 [3] NCCL INFO Using network Libfabric
+ip-26-0-166-36:104432:105194 [3] NCCL INFO Using network Libfabric
+ip-26-0-166-36:104431:105195 [2] NCCL INFO Using network Libfabric
+ip-26-0-165-213:9628:10394 [3] NCCL INFO Using network Libfabric
+ip-26-0-166-15:127292:128057 [2] NCCL INFO Using network Libfabric
+ip-26-0-167-9:1039539:1040321 [2] NCCL INFO Using network Libfabric
+ip-26-0-166-244:897243:898033 [3] NCCL INFO Using network Libfabric
+ip-26-0-166-125:762818:763616 [3] NCCL INFO Using network Libfabric
+ip-26-0-166-125:762817:763617 [2] NCCL INFO Using network Libfabric
+ip-26-0-166-15:127293:128058 [3] NCCL INFO Using network Libfabric
+ip-26-0-162-46:702886:703685 [6] NCCL INFO Using network Libfabric
+ip-26-0-166-244:897242:898034 [2] NCCL INFO Using network Libfabric
+ip-26-0-162-46:702887:703686 [7] NCCL INFO Using network Libfabric
+ip-26-0-165-213:9632:10395 [7] NCCL INFO Using network Libfabric
+ip-26-0-167-9:1039544:1040322 [7] NCCL INFO Using network Libfabric
+ip-26-0-166-36:104435:105196 [6] NCCL INFO Using network Libfabric
+ip-26-0-166-125:762822:763618 [7] NCCL INFO Using network Libfabric
+ip-26-0-166-214:624950:625743 [7] NCCL INFO Using network Libfabric
+ip-26-0-167-9:1039543:1040323 [6] NCCL INFO Using network Libfabric
+ip-26-0-166-244:897247:898035 [7] NCCL INFO Using network Libfabric
+ip-26-0-166-125:762821:763619 [6] NCCL INFO Using network Libfabric
+ip-26-0-166-15:127297:128060 [7] NCCL INFO Using network Libfabric
+ip-26-0-166-244:897246:898036 [6] NCCL INFO Using network Libfabric
+ip-26-0-165-213:9631:10396 [6] NCCL INFO Using network Libfabric
+ip-26-0-166-15:127296:128059 [6] NCCL INFO Using network Libfabric
+ip-26-0-166-36:104436:105197 [7] NCCL INFO Using network Libfabric
+ip-26-0-166-214:624949:625744 [6] NCCL INFO Using network Libfabric
+ip-26-0-162-46:702880:703676 [0] NCCL INFO comm 0x1d9e9040 rank 0 nranks 16 cudaDev 0 nvmlDev 0 busId 53000 commId 0x27beb3196ace4a07 - Init START
+ip-26-0-162-46:702881:703677 [1] NCCL INFO comm 0x1f3f8120 rank 1 nranks 16 cudaDev 1 nvmlDev 1 busId 64000 commId 0x27beb3196ace4a07 - Init START
+ip-26-0-167-9:1039538:1040317 [1] NCCL INFO comm 0x1b80ccb0 rank 15 nranks 16 cudaDev 1 nvmlDev 1 busId 64000 commId 0x27beb3196ace4a07 - Init START
+ip-26-0-162-46:702885:703680 [5] NCCL INFO comm 0x1eb5c6a0 rank 1 nranks 16 cudaDev 5 nvmlDev 5 busId a8000 commId 0xddcb46505a6ea727 - Init START
+ip-26-0-162-46:702884:703679 [4] NCCL INFO comm 0x1cfccc90 rank 0 nranks 16 cudaDev 4 nvmlDev 4 busId 97000 commId 0xddcb46505a6ea727 - Init START
+ip-26-0-166-36:104434:105192 [5] NCCL INFO comm 0x16ba8110 rank 13 nranks 16 cudaDev 5 nvmlDev 5 busId a8000 commId 0xddcb46505a6ea727 - Init START
+ip-26-0-166-36:104433:105193 [4] NCCL INFO comm 0x1838ccd0 rank 12 nranks 16 cudaDev 4 nvmlDev 4 busId 97000 commId 0xddcb46505a6ea727 - Init START
+ip-26-0-166-125:762815:763613 [0] NCCL INFO comm 0x168bbc20 rank 4 nranks 16 cudaDev 0 nvmlDev 0 busId 53000 commId 0x27beb3196ace4a07 - Init START
+ip-26-0-166-125:762816:763612 [1] NCCL INFO comm 0x178ea100 rank 5 nranks 16 cudaDev 1 nvmlDev 1 busId 64000 commId 0x27beb3196ace4a07 - Init START
+ip-26-0-165-213:9625:10390 [0] NCCL INFO comm 0x171022d0 rank 2 nranks 16 cudaDev 0 nvmlDev 0 busId 53000 commId 0x27beb3196ace4a07 - Init START
+ip-26-0-165-213:9626:10389 [1] NCCL INFO comm 0x176d8980 rank 3 nranks 16 cudaDev 1 nvmlDev 1 busId 64000 commId 0x27beb3196ace4a07 - Init START
+ip-26-0-166-15:127291:128053 [1] NCCL INFO comm 0x184d7020 rank 7 nranks 16 cudaDev 1 nvmlDev 1 busId 64000 commId 0x27beb3196ace4a07 - Init START
+ip-26-0-166-15:127290:128054 [0] NCCL INFO comm 0x16c594f0 rank 6 nranks 16 cudaDev 0 nvmlDev 0 busId 53000 commId 0x27beb3196ace4a07 - Init START
+ip-26-0-167-9:1039542:1040319 [5] NCCL INFO comm 0x1b6228e0 rank 15 nranks 16 cudaDev 5 nvmlDev 5 busId a8000 commId 0xddcb46505a6ea727 - Init START
+ip-26-0-166-36:104430:105190 [1] NCCL INFO comm 0x17da6700 rank 13 nranks 16 cudaDev 1 nvmlDev 1 busId 64000 commId 0x27beb3196ace4a07 - Init START
+ip-26-0-166-36:104429:105191 [0] NCCL INFO comm 0x185904e0 rank 12 nranks 16 cudaDev 0 nvmlDev 0 busId 53000 commId 0x27beb3196ace4a07 - Init START
+ip-26-0-166-244:897245:898031 [5] NCCL INFO comm 0x17bfad90 rank 11 nranks 16 cudaDev 5 nvmlDev 5 busId a8000 commId 0xddcb46505a6ea727 - Init START
+ip-26-0-166-214:624944:625738 [1] NCCL INFO comm 0x166ec9b0 rank 9 nranks 16 cudaDev 1 nvmlDev 1 busId 64000 commId 0x27beb3196ace4a07 - Init START
+ip-26-0-166-214:624943:625737 [0] NCCL INFO comm 0x17641680 rank 8 nranks 16 cudaDev 0 nvmlDev 0 busId 53000 commId 0x27beb3196ace4a07 - Init START
+ip-26-0-166-214:624948:625740 [5] NCCL INFO comm 0x18255f80 rank 9 nranks 16 cudaDev 5 nvmlDev 5 busId a8000 commId 0xddcb46505a6ea727 - Init START
+ip-26-0-167-9:1039541:1040318 [4] NCCL INFO comm 0x216daf60 rank 14 nranks 16 cudaDev 4 nvmlDev 4 busId 97000 commId 0xddcb46505a6ea727 - Init START
+ip-26-0-166-125:762820:763615 [5] NCCL INFO comm 0x175bbef0 rank 5 nranks 16 cudaDev 5 nvmlDev 5 busId a8000 commId 0xddcb46505a6ea727 - Init START
+ip-26-0-166-125:762819:763614 [4] NCCL INFO comm 0x1769afa0 rank 4 nranks 16 cudaDev 4 nvmlDev 4 busId 97000 commId 0xddcb46505a6ea727 - Init START
+ip-26-0-165-213:9629:10391 [4] NCCL INFO comm 0x16c33e40 rank 2 nranks 16 cudaDev 4 nvmlDev 4 busId 97000 commId 0xddcb46505a6ea727 - Init START
+ip-26-0-165-213:9630:10392 [5] NCCL INFO comm 0x16ab1430 rank 3 nranks 16 cudaDev 5 nvmlDev 5 busId a8000 commId 0xddcb46505a6ea727 - Init START
+ip-26-0-166-15:127295:128055 [5] NCCL INFO comm 0x16d56070 rank 7 nranks 16 cudaDev 5 nvmlDev 5 busId a8000 commId 0xddcb46505a6ea727 - Init START
+ip-26-0-166-15:127294:128056 [4] NCCL INFO comm 0x172652a0 rank 6 nranks 16 cudaDev 4 nvmlDev 4 busId 97000 commId 0xddcb46505a6ea727 - Init START
+ip-26-0-166-244:897241:898029 [1] NCCL INFO comm 0x18302ab0 rank 11 nranks 16 cudaDev 1 nvmlDev 1 busId 64000 commId 0x27beb3196ace4a07 - Init START
+ip-26-0-166-244:897244:898032 [4] NCCL INFO comm 0x16fdcbf0 rank 10 nranks 16 cudaDev 4 nvmlDev 4 busId 97000 commId 0xddcb46505a6ea727 - Init START
+ip-26-0-166-244:897240:898030 [0] NCCL INFO comm 0x1745a980 rank 10 nranks 16 cudaDev 0 nvmlDev 0 busId 53000 commId 0x27beb3196ace4a07 - Init START
+ip-26-0-166-214:624947:625739 [4] NCCL INFO comm 0x181df0e0 rank 8 nranks 16 cudaDev 4 nvmlDev 4 busId 97000 commId 0xddcb46505a6ea727 - Init START
+ip-26-0-167-9:1039537:1040316 [0] NCCL INFO comm 0x20e82de0 rank 14 nranks 16 cudaDev 0 nvmlDev 0 busId 53000 commId 0x27beb3196ace4a07 - Init START
+ip-26-0-166-244:897245:898031 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-167-9:1039537:1040316 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-162-46:702880:703676 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/349
+ip-26-0-162-46:702881:703677 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-162-46:702885:703680 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-36:104430:105190 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-125:762815:763613 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-125:762816:763612 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-162-46:702884:703679 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-244:897244:898032 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-165-213:9626:10389 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-165-213:9625:10390 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-214:624948:625740 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-15:127295:128055 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-15:127291:128053 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-167-9:1039538:1040317 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-36:104434:105192 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-36:104433:105193 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-244:897240:898030 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-244:897241:898029 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-167-9:1039541:1040318 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-36:104429:105191 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-125:762820:763615 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-125:762819:763614 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-165-213:9630:10392 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-165-213:9629:10391 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-214:624944:625738 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-15:127290:128054 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-214:624947:625739 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-214:624943:625737 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-15:127294:128056 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-167-9:1039542:1040319 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-162-46:702883:703683 [3] NCCL INFO comm 0x1e5674b0 rank 1 nranks 16 cudaDev 3 nvmlDev 3 busId 86000 commId 0xbb8f0b4ef9cdd09a - Init START
+ip-26-0-162-46:702882:703682 [2] NCCL INFO comm 0x1d872dd0 rank 0 nranks 16 cudaDev 2 nvmlDev 2 busId 75000 commId 0xbb8f0b4ef9cdd09a - Init START
+ip-26-0-167-9:1039540:1040320 [3] NCCL INFO comm 0x1bf57ad0 rank 15 nranks 16 cudaDev 3 nvmlDev 3 busId 86000 commId 0xbb8f0b4ef9cdd09a - Init START
+ip-26-0-165-213:9628:10394 [3] NCCL INFO comm 0x17627d70 rank 3 nranks 16 cudaDev 3 nvmlDev 3 busId 86000 commId 0xbb8f0b4ef9cdd09a - Init START
+ip-26-0-165-213:9627:10393 [2] NCCL INFO comm 0x16c3bbb0 rank 2 nranks 16 cudaDev 2 nvmlDev 2 busId 75000 commId 0xbb8f0b4ef9cdd09a - Init START
+ip-26-0-166-125:762817:763617 [2] NCCL INFO comm 0x186bc3a0 rank 4 nranks 16 cudaDev 2 nvmlDev 2 busId 75000 commId 0xbb8f0b4ef9cdd09a - Init START
+ip-26-0-166-125:762818:763616 [3] NCCL INFO comm 0x169f9380 rank 5 nranks 16 cudaDev 3 nvmlDev 3 busId 86000 commId 0xbb8f0b4ef9cdd09a - Init START
+ip-26-0-166-36:104432:105194 [3] NCCL INFO comm 0x1692d450 rank 13 nranks 16 cudaDev 3 nvmlDev 3 busId 86000 commId 0xbb8f0b4ef9cdd09a - Init START
+ip-26-0-166-15:127293:128058 [3] NCCL INFO comm 0x183c6690 rank 7 nranks 16 cudaDev 3 nvmlDev 3 busId 86000 commId 0xbb8f0b4ef9cdd09a - Init START
+ip-26-0-166-15:127292:128057 [2] NCCL INFO comm 0x174c51c0 rank 6 nranks 16 cudaDev 2 nvmlDev 2 busId 75000 commId 0xbb8f0b4ef9cdd09a - Init START
+ip-26-0-167-9:1039539:1040321 [2] NCCL INFO comm 0x2034e5e0 rank 14 nranks 16 cudaDev 2 nvmlDev 2 busId 75000 commId 0xbb8f0b4ef9cdd09a - Init START
+ip-26-0-166-244:897243:898033 [3] NCCL INFO comm 0x16b51e40 rank 11 nranks 16 cudaDev 3 nvmlDev 3 busId 86000 commId 0xbb8f0b4ef9cdd09a - Init START
+ip-26-0-166-214:624945:625741 [2] NCCL INFO comm 0x1798aed0 rank 8 nranks 16 cudaDev 2 nvmlDev 2 busId 75000 commId 0xbb8f0b4ef9cdd09a - Init START
+ip-26-0-166-214:624946:625742 [3] NCCL INFO comm 0x18227c50 rank 9 nranks 16 cudaDev 3 nvmlDev 3 busId 86000 commId 0xbb8f0b4ef9cdd09a - Init START
+ip-26-0-166-36:104431:105195 [2] NCCL INFO comm 0x17934380 rank 12 nranks 16 cudaDev 2 nvmlDev 2 busId 75000 commId 0xbb8f0b4ef9cdd09a - Init START
+ip-26-0-166-244:897242:898034 [2] NCCL INFO comm 0x16847d90 rank 10 nranks 16 cudaDev 2 nvmlDev 2 busId 75000 commId 0xbb8f0b4ef9cdd09a - Init START
+ip-26-0-162-46:702883:703683 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-162-46:702882:703682 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-165-213:9628:10394 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-167-9:1039539:1040321 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-36:104432:105194 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-36:104431:105195 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-125:762818:763616 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-244:897243:898033 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-214:624946:625742 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-15:127293:128058 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-15:127292:128057 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-165-213:9627:10393 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-167-9:1039540:1040320 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-125:762817:763617 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-244:897242:898034 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-214:624945:625741 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-162-46:702886:703685 [6] NCCL INFO comm 0x1d0fd800 rank 0 nranks 16 cudaDev 6 nvmlDev 6 busId b9000 commId 0x30acb1707865c266 - Init START
+ip-26-0-162-46:702887:703686 [7] NCCL INFO comm 0x1eb4c5a0 rank 1 nranks 16 cudaDev 7 nvmlDev 7 busId ca000 commId 0x30acb1707865c266 - Init START
+ip-26-0-167-9:1039544:1040322 [7] NCCL INFO comm 0x1be68400 rank 15 nranks 16 cudaDev 7 nvmlDev 7 busId ca000 commId 0x30acb1707865c266 - Init START
+ip-26-0-166-125:762822:763618 [7] NCCL INFO comm 0x182cd440 rank 5 nranks 16 cudaDev 7 nvmlDev 7 busId ca000 commId 0x30acb1707865c266 - Init START
+ip-26-0-165-213:9631:10396 [6] NCCL INFO comm 0x17de4a70 rank 2 nranks 16 cudaDev 6 nvmlDev 6 busId b9000 commId 0x30acb1707865c266 - Init START
+ip-26-0-165-213:9632:10395 [7] NCCL INFO comm 0x1700ed20 rank 3 nranks 16 cudaDev 7 nvmlDev 7 busId ca000 commId 0x30acb1707865c266 - Init START
+ip-26-0-166-15:127297:128060 [7] NCCL INFO comm 0x18052140 rank 7 nranks 16 cudaDev 7 nvmlDev 7 busId ca000 commId 0x30acb1707865c266 - Init START
+ip-26-0-166-15:127296:128059 [6] NCCL INFO comm 0x1793a8e0 rank 6 nranks 16 cudaDev 6 nvmlDev 6 busId b9000 commId 0x30acb1707865c266 - Init START
+ip-26-0-166-36:104436:105197 [7] NCCL INFO comm 0x178626f0 rank 13 nranks 16 cudaDev 7 nvmlDev 7 busId ca000 commId 0x30acb1707865c266 - Init START
+ip-26-0-166-125:762821:763619 [6] NCCL INFO comm 0x16b951b0 rank 4 nranks 16 cudaDev 6 nvmlDev 6 busId b9000 commId 0x30acb1707865c266 - Init START
+ip-26-0-166-244:897246:898036 [6] NCCL INFO comm 0x185776b0 rank 10 nranks 16 cudaDev 6 nvmlDev 6 busId b9000 commId 0x30acb1707865c266 - Init START
+ip-26-0-166-244:897247:898035 [7] NCCL INFO comm 0x17afb910 rank 11 nranks 16 cudaDev 7 nvmlDev 7 busId ca000 commId 0x30acb1707865c266 - Init START
+ip-26-0-166-214:624950:625743 [7] NCCL INFO comm 0x16f730c0 rank 9 nranks 16 cudaDev 7 nvmlDev 7 busId ca000 commId 0x30acb1707865c266 - Init START
+ip-26-0-166-214:624949:625744 [6] NCCL INFO comm 0x16c4eaf0 rank 8 nranks 16 cudaDev 6 nvmlDev 6 busId b9000 commId 0x30acb1707865c266 - Init START
+ip-26-0-167-9:1039543:1040323 [6] NCCL INFO comm 0x211a8520 rank 14 nranks 16 cudaDev 6 nvmlDev 6 busId b9000 commId 0x30acb1707865c266 - Init START
+ip-26-0-166-36:104435:105196 [6] NCCL INFO comm 0x16ef07b0 rank 12 nranks 16 cudaDev 6 nvmlDev 6 busId b9000 commId 0x30acb1707865c266 - Init START
+ip-26-0-167-9:1039543:1040323 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-162-46:702887:703686 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-162-46:702886:703685 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-165-213:9632:10395 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-165-213:9631:10396 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-36:104436:105197 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-125:762822:763618 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-125:762821:763619 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-214:624950:625743 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-167-9:1039544:1040322 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-244:897247:898035 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-244:897246:898036 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-15:127297:128060 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-15:127296:128059 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-36:104435:105196 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-166-214:624949:625744 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-167-9:1039538:1040317 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-166-15:127293:128058 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-167-9:1039539:1040321 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-166-125:762819:763614 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,fff00000,00000000
+ip-26-0-166-15:127295:128055 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-167-9:1039541:1040318 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-162-46:702882:703682 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-167-9:1039544:1040322 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-166-36:104436:105197 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-167-9:1039543:1040323 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-166-36:104435:105196 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-166-244:897241:898029 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-166-15:127291:128053 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-167-9:1039537:1040316 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-166-125:762815:763613 [0] NCCL INFO Setting affinity for GPU 0 to 0fff,ffffffff
+ip-26-0-167-9:1039540:1040320 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-166-15:127292:128057 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-166-125:762818:763616 [3] NCCL INFO Setting affinity for GPU 3 to 0fff,ffffffff
+ip-26-0-167-9:1039542:1040319 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-166-15:127294:128056 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-166-244:897243:898033 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-166-125:762821:763619 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,fff00000,00000000
+ip-26-0-166-15:127297:128060 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-166-125:762822:763618 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,fff00000,00000000
+ip-26-0-166-15:127290:128054 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-166-15:127296:128059 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-162-46:702886:703685 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-166-125:762816:763612 [1] NCCL INFO Setting affinity for GPU 1 to 0fff,ffffffff
+ip-26-0-162-46:702883:703683 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-166-244:897240:898030 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-166-125:762820:763615 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,fff00000,00000000
+ip-26-0-162-46:702884:703679 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-166-125:762817:763617 [2] NCCL INFO Setting affinity for GPU 2 to 0fff,ffffffff
+ip-26-0-166-36:104429:105191 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-166-36:104432:105194 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-162-46:702885:703680 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-166-36:104431:105195 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-162-46:702887:703686 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-166-36:104434:105192 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-166-36:104433:105193 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-166-244:897247:898035 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-166-244:897246:898036 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-166-36:104430:105190 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-162-46:702881:703677 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-162-46:702880:703676 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-166-244:897245:898031 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-166-244:897242:898034 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-166-244:897244:898032 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-166-214:624943:625737 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-166-214:624949:625744 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-166-214:624950:625743 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-166-214:624944:625738 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-166-214:624946:625742 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-166-214:624945:625741 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-166-214:624948:625740 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-166-214:624947:625739 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-165-213:9631:10396 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-165-213:9630:10392 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-165-213:9625:10390 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-165-213:9628:10394 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-165-213:9626:10389 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-165-213:9626:10389 [1] NCCL INFO Trees [0] -1/-1/-1->3->2 [1] -1/-1/-1->3->2 [2] 2/-1/-1->3->5 [3] 2/-1/-1->3->5 [4] -1/-1/-1->3->2 [5] -1/-1/-1->3->2 [6] 2/5/1->3->7 [7] 2/5/1->3->7
+ip-26-0-165-213:9626:10389 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-125:762816:763612 [1] NCCL INFO Trees [0] -1/-1/-1->5->4 [1] -1/-1/-1->5->4 [2] 4/3/7->5->9 [3] 4/3/7->5->9 [4] -1/-1/-1->5->4 [5] -1/-1/-1->5->4 [6] 4/-1/-1->5->3 [7] 4/-1/-1->5->3
+ip-26-0-166-125:762815:763613 [0] NCCL INFO Trees [0] 5/2/6->4->8 [1] 5/2/6->4->8 [2] -1/-1/-1->4->5 [3] -1/-1/-1->4->5 [4] 5/-1/-1->4->2 [5] 5/-1/-1->4->2 [6] -1/-1/-1->4->5 [7] -1/-1/-1->4->5
+ip-26-0-166-125:762816:763612 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-125:762815:763613 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-15:127290:128054 [0] NCCL INFO Trees [0] 7/-1/-1->6->4 [1] 7/-1/-1->6->4 [2] -1/-1/-1->6->7 [3] -1/-1/-1->6->7 [4] 7/10/2->6->14 [5] 7/10/2->6->14 [6] -1/-1/-1->6->7 [7] -1/-1/-1->6->7
+ip-26-0-166-15:127290:128054 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-15:127291:128053 [1] NCCL INFO Trees [0] -1/-1/-1->7->6 [1] -1/-1/-1->7->6 [2] 6/-1/-1->7->5 [3] 6/-1/-1->7->5 [4] -1/-1/-1->7->6 [5] -1/-1/-1->7->6 [6] 6/11/3->7->15 [7] 6/11/3->7->15
+ip-26-0-166-15:127291:128053 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-214:624943:625737 [0] NCCL INFO Trees [0] 9/4/12->8->0 [1] 9/4/12->8->0 [2] -1/-1/-1->8->9 [3] -1/-1/-1->8->9 [4] 9/-1/-1->8->10 [5] 9/-1/-1->8->10 [6] -1/-1/-1->8->9 [7] -1/-1/-1->8->9
+ip-26-0-166-214:624943:625737 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-214:624944:625738 [1] NCCL INFO Trees [0] -1/-1/-1->9->8 [1] -1/-1/-1->9->8 [2] 8/5/13->9->1 [3] 8/5/13->9->1 [4] -1/-1/-1->9->8 [5] -1/-1/-1->9->8 [6] 8/-1/-1->9->11 [7] 8/-1/-1->9->11
+ip-26-0-166-214:624944:625738 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-244:897241:898029 [1] NCCL INFO Trees [0] -1/-1/-1->11->10 [1] -1/-1/-1->11->10 [2] 10/-1/-1->11->13 [3] 10/-1/-1->11->13 [4] -1/-1/-1->11->10 [5] -1/-1/-1->11->10 [6] 10/13/9->11->7 [7] 10/13/9->11->7
+ip-26-0-166-244:897240:898030 [0] NCCL INFO Trees [0] 11/-1/-1->10->12 [1] 11/-1/-1->10->12 [2] -1/-1/-1->10->11 [3] -1/-1/-1->10->11 [4] 11/12/8->10->6 [5] 11/12/8->10->6 [6] -1/-1/-1->10->11 [7] -1/-1/-1->10->11
+ip-26-0-166-244:897241:898029 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-244:897240:898030 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-36:104429:105191 [0] NCCL INFO Trees [0] 13/10/14->12->8 [1] 13/10/14->12->8 [2] -1/-1/-1->12->13 [3] -1/-1/-1->12->13 [4] 13/-1/-1->12->10 [5] 13/-1/-1->12->10 [6] -1/-1/-1->12->13 [7] -1/-1/-1->12->13
+ip-26-0-166-36:104429:105191 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-36:104430:105190 [1] NCCL INFO Trees [0] -1/-1/-1->13->12 [1] -1/-1/-1->13->12 [2] 12/11/15->13->9 [3] 12/11/15->13->9 [4] -1/-1/-1->13->12 [5] -1/-1/-1->13->12 [6] 12/-1/-1->13->11 [7] 12/-1/-1->13->11
+ip-26-0-166-36:104430:105190 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-167-9:1039537:1040316 [0] NCCL INFO Trees [0] 15/-1/-1->14->12 [1] 15/-1/-1->14->12 [2] -1/-1/-1->14->15 [3] -1/-1/-1->14->15 [4] 15/6/-1->14->-1 [5] 15/6/-1->14->-1 [6] -1/-1/-1->14->15 [7] -1/-1/-1->14->15
+ip-26-0-167-9:1039537:1040316 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-213:9625:10390 [0] NCCL INFO Trees [0] 3/-1/-1->2->4 [1] 3/-1/-1->2->4 [2] -1/-1/-1->2->3 [3] -1/-1/-1->2->3 [4] 3/4/0->2->6 [5] 3/4/0->2->6 [6] -1/-1/-1->2->3 [7] -1/-1/-1->2->3
+ip-26-0-165-213:9625:10390 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-167-9:1039538:1040317 [1] NCCL INFO Trees [0] -1/-1/-1->15->14 [1] -1/-1/-1->15->14 [2] 14/-1/-1->15->13 [3] 14/-1/-1->15->13 [4] -1/-1/-1->15->14 [5] -1/-1/-1->15->14 [6] 14/7/-1->15->-1 [7] 14/7/-1->15->-1
+ip-26-0-167-9:1039538:1040317 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-162-46:702880:703676 [0] NCCL INFO Channel 00/08 :    0   1   2   3   4   5   6   7   8   9  10  11  12  13  14  15
+ip-26-0-162-46:702881:703677 [1] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/9/-1->1->-1 [3] 0/9/-1->1->-1 [4] -1/-1/-1->1->0 [5] -1/-1/-1->1->0 [6] 0/-1/-1->1->3 [7] 0/-1/-1->1->3
+ip-26-0-162-46:702880:703676 [0] NCCL INFO Channel 01/08 :    0   3   2   5   4   7   6   9   8  11  10  13  12  15  14   1
+ip-26-0-162-46:702881:703677 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-162-46:702880:703676 [0] NCCL INFO Channel 02/08 :    0   1   2   3   4   5   6   7   8   9  10  11  12  13  14  15
+ip-26-0-162-46:702880:703676 [0] NCCL INFO Channel 03/08 :    0   3   2   5   4   7   6   9   8  11  10  13  12  15  14   1
+ip-26-0-162-46:702880:703676 [0] NCCL INFO Channel 04/08 :    0   1   2   3   4   5   6   7   8   9  10  11  12  13  14  15
+ip-26-0-162-46:702880:703676 [0] NCCL INFO Channel 05/08 :    0   3   2   5   4   7   6   9   8  11  10  13  12  15  14   1
+ip-26-0-162-46:702880:703676 [0] NCCL INFO Channel 06/08 :    0   1   2   3   4   5   6   7   8   9  10  11  12  13  14  15
+ip-26-0-162-46:702880:703676 [0] NCCL INFO Channel 07/08 :    0   3   2   5   4   7   6   9   8  11  10  13  12  15  14   1
+ip-26-0-162-46:702880:703676 [0] NCCL INFO Trees [0] 1/8/-1->0->-1 [1] 1/8/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1 [4] 1/-1/-1->0->2 [5] 1/-1/-1->0->2 [6] -1/-1/-1->0->1 [7] -1/-1/-1->0->1
+ip-26-0-162-46:702880:703676 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-213:9629:10391 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-165-213:9632:10395 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-165-213:9629:10391 [4] NCCL INFO Trees [0] 3/-1/-1->2->4 [1] 3/-1/-1->2->4 [2] -1/-1/-1->2->3 [3] -1/-1/-1->2->3 [4] 3/4/0->2->6 [5] 3/4/0->2->6 [6] -1/-1/-1->2->3 [7] -1/-1/-1->2->3
+ip-26-0-165-213:9629:10391 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-213:9630:10392 [5] NCCL INFO Trees [0] -1/-1/-1->3->2 [1] -1/-1/-1->3->2 [2] 2/-1/-1->3->5 [3] 2/-1/-1->3->5 [4] -1/-1/-1->3->2 [5] -1/-1/-1->3->2 [6] 2/5/1->3->7 [7] 2/5/1->3->7
+ip-26-0-165-213:9630:10392 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-125:762819:763614 [4] NCCL INFO Trees [0] 5/2/6->4->8 [1] 5/2/6->4->8 [2] -1/-1/-1->4->5 [3] -1/-1/-1->4->5 [4] 5/-1/-1->4->2 [5] 5/-1/-1->4->2 [6] -1/-1/-1->4->5 [7] -1/-1/-1->4->5
+ip-26-0-166-125:762819:763614 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-125:762820:763615 [5] NCCL INFO Trees [0] -1/-1/-1->5->4 [1] -1/-1/-1->5->4 [2] 4/3/7->5->9 [3] 4/3/7->5->9 [4] -1/-1/-1->5->4 [5] -1/-1/-1->5->4 [6] 4/-1/-1->5->3 [7] 4/-1/-1->5->3
+ip-26-0-166-125:762820:763615 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-214:624948:625740 [5] NCCL INFO Trees [0] -1/-1/-1->9->8 [1] -1/-1/-1->9->8 [2] 8/5/13->9->1 [3] 8/5/13->9->1 [4] -1/-1/-1->9->8 [5] -1/-1/-1->9->8 [6] 8/-1/-1->9->11 [7] 8/-1/-1->9->11
+ip-26-0-166-214:624947:625739 [4] NCCL INFO Trees [0] 9/4/12->8->0 [1] 9/4/12->8->0 [2] -1/-1/-1->8->9 [3] -1/-1/-1->8->9 [4] 9/-1/-1->8->10 [5] 9/-1/-1->8->10 [6] -1/-1/-1->8->9 [7] -1/-1/-1->8->9
+ip-26-0-166-214:624948:625740 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-214:624947:625739 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-15:127295:128055 [5] NCCL INFO Trees [0] -1/-1/-1->7->6 [1] -1/-1/-1->7->6 [2] 6/-1/-1->7->5 [3] 6/-1/-1->7->5 [4] -1/-1/-1->7->6 [5] -1/-1/-1->7->6 [6] 6/11/3->7->15 [7] 6/11/3->7->15
+ip-26-0-166-15:127294:128056 [4] NCCL INFO Trees [0] 7/-1/-1->6->4 [1] 7/-1/-1->6->4 [2] -1/-1/-1->6->7 [3] -1/-1/-1->6->7 [4] 7/10/2->6->14 [5] 7/10/2->6->14 [6] -1/-1/-1->6->7 [7] -1/-1/-1->6->7
+ip-26-0-166-15:127294:128056 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-15:127295:128055 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-244:897245:898031 [5] NCCL INFO Trees [0] -1/-1/-1->11->10 [1] -1/-1/-1->11->10 [2] 10/-1/-1->11->13 [3] 10/-1/-1->11->13 [4] -1/-1/-1->11->10 [5] -1/-1/-1->11->10 [6] 10/13/9->11->7 [7] 10/13/9->11->7
+ip-26-0-166-244:897244:898032 [4] NCCL INFO Trees [0] 11/-1/-1->10->12 [1] 11/-1/-1->10->12 [2] -1/-1/-1->10->11 [3] -1/-1/-1->10->11 [4] 11/12/8->10->6 [5] 11/12/8->10->6 [6] -1/-1/-1->10->11 [7] -1/-1/-1->10->11
+ip-26-0-166-244:897245:898031 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-244:897244:898032 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-36:104433:105193 [4] NCCL INFO Trees [0] 13/10/14->12->8 [1] 13/10/14->12->8 [2] -1/-1/-1->12->13 [3] -1/-1/-1->12->13 [4] 13/-1/-1->12->10 [5] 13/-1/-1->12->10 [6] -1/-1/-1->12->13 [7] -1/-1/-1->12->13
+ip-26-0-166-36:104434:105192 [5] NCCL INFO Trees [0] -1/-1/-1->13->12 [1] -1/-1/-1->13->12 [2] 12/11/15->13->9 [3] 12/11/15->13->9 [4] -1/-1/-1->13->12 [5] -1/-1/-1->13->12 [6] 12/-1/-1->13->11 [7] 12/-1/-1->13->11
+ip-26-0-166-36:104433:105193 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-36:104434:105192 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-167-9:1039541:1040318 [4] NCCL INFO Trees [0] 15/-1/-1->14->12 [1] 15/-1/-1->14->12 [2] -1/-1/-1->14->15 [3] -1/-1/-1->14->15 [4] 15/6/-1->14->-1 [5] 15/6/-1->14->-1 [6] -1/-1/-1->14->15 [7] -1/-1/-1->14->15
+ip-26-0-167-9:1039541:1040318 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-162-46:702885:703680 [5] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/9/-1->1->-1 [3] 0/9/-1->1->-1 [4] -1/-1/-1->1->0 [5] -1/-1/-1->1->0 [6] 0/-1/-1->1->3 [7] 0/-1/-1->1->3
+ip-26-0-162-46:702884:703679 [4] NCCL INFO Channel 00/08 :    0   1   2   3   4   5   6   7   8   9  10  11  12  13  14  15
+ip-26-0-162-46:702885:703680 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-162-46:702884:703679 [4] NCCL INFO Channel 01/08 :    0   3   2   5   4   7   6   9   8  11  10  13  12  15  14   1
+ip-26-0-162-46:702884:703679 [4] NCCL INFO Channel 02/08 :    0   1   2   3   4   5   6   7   8   9  10  11  12  13  14  15
+ip-26-0-162-46:702884:703679 [4] NCCL INFO Channel 03/08 :    0   3   2   5   4   7   6   9   8  11  10  13  12  15  14   1
+ip-26-0-162-46:702884:703679 [4] NCCL INFO Channel 04/08 :    0   1   2   3   4   5   6   7   8   9  10  11  12  13  14  15
+ip-26-0-162-46:702884:703679 [4] NCCL INFO Channel 05/08 :    0   3   2   5   4   7   6   9   8  11  10  13  12  15  14   1
+ip-26-0-162-46:702884:703679 [4] NCCL INFO Channel 06/08 :    0   1   2   3   4   5   6   7   8   9  10  11  12  13  14  15
+ip-26-0-162-46:702884:703679 [4] NCCL INFO Channel 07/08 :    0   3   2   5   4   7   6   9   8  11  10  13  12  15  14   1
+ip-26-0-162-46:702884:703679 [4] NCCL INFO Trees [0] 1/8/-1->0->-1 [1] 1/8/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1 [4] 1/-1/-1->0->2 [5] 1/-1/-1->0->2 [6] -1/-1/-1->0->1 [7] -1/-1/-1->0->1
+ip-26-0-162-46:702884:703679 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-167-9:1039542:1040319 [5] NCCL INFO Trees [0] -1/-1/-1->15->14 [1] -1/-1/-1->15->14 [2] 14/-1/-1->15->13 [3] 14/-1/-1->15->13 [4] -1/-1/-1->15->14 [5] -1/-1/-1->15->14 [6] 14/7/-1->15->-1 [7] 14/7/-1->15->-1
+ip-26-0-167-9:1039542:1040319 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-213:9627:10393 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-165-213:9632:10395 [7] NCCL INFO Trees [0] -1/-1/-1->3->2 [1] -1/-1/-1->3->2 [2] 2/-1/-1->3->5 [3] 2/-1/-1->3->5 [4] -1/-1/-1->3->2 [5] -1/-1/-1->3->2 [6] 2/5/1->3->7 [7] 2/5/1->3->7
+ip-26-0-165-213:9632:10395 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-125:762821:763619 [6] NCCL INFO Trees [0] 5/2/6->4->8 [1] 5/2/6->4->8 [2] -1/-1/-1->4->5 [3] -1/-1/-1->4->5 [4] 5/-1/-1->4->2 [5] 5/-1/-1->4->2 [6] -1/-1/-1->4->5 [7] -1/-1/-1->4->5
+ip-26-0-166-125:762821:763619 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-125:762822:763618 [7] NCCL INFO Trees [0] -1/-1/-1->5->4 [1] -1/-1/-1->5->4 [2] 4/3/7->5->9 [3] 4/3/7->5->9 [4] -1/-1/-1->5->4 [5] -1/-1/-1->5->4 [6] 4/-1/-1->5->3 [7] 4/-1/-1->5->3
+ip-26-0-166-125:762822:763618 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-214:624950:625743 [7] NCCL INFO Trees [0] -1/-1/-1->9->8 [1] -1/-1/-1->9->8 [2] 8/5/13->9->1 [3] 8/5/13->9->1 [4] -1/-1/-1->9->8 [5] -1/-1/-1->9->8 [6] 8/-1/-1->9->11 [7] 8/-1/-1->9->11
+ip-26-0-166-214:624950:625743 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-214:624949:625744 [6] NCCL INFO Trees [0] 9/4/12->8->0 [1] 9/4/12->8->0 [2] -1/-1/-1->8->9 [3] -1/-1/-1->8->9 [4] 9/-1/-1->8->10 [5] 9/-1/-1->8->10 [6] -1/-1/-1->8->9 [7] -1/-1/-1->8->9
+ip-26-0-166-214:624949:625744 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-15:127296:128059 [6] NCCL INFO Trees [0] 7/-1/-1->6->4 [1] 7/-1/-1->6->4 [2] -1/-1/-1->6->7 [3] -1/-1/-1->6->7 [4] 7/10/2->6->14 [5] 7/10/2->6->14 [6] -1/-1/-1->6->7 [7] -1/-1/-1->6->7
+ip-26-0-166-15:127296:128059 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-15:127297:128060 [7] NCCL INFO Trees [0] -1/-1/-1->7->6 [1] -1/-1/-1->7->6 [2] 6/-1/-1->7->5 [3] 6/-1/-1->7->5 [4] -1/-1/-1->7->6 [5] -1/-1/-1->7->6 [6] 6/11/3->7->15 [7] 6/11/3->7->15
+ip-26-0-166-15:127297:128060 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-244:897247:898035 [7] NCCL INFO Trees [0] -1/-1/-1->11->10 [1] -1/-1/-1->11->10 [2] 10/-1/-1->11->13 [3] 10/-1/-1->11->13 [4] -1/-1/-1->11->10 [5] -1/-1/-1->11->10 [6] 10/13/9->11->7 [7] 10/13/9->11->7
+ip-26-0-166-244:897247:898035 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-244:897246:898036 [6] NCCL INFO Trees [0] 11/-1/-1->10->12 [1] 11/-1/-1->10->12 [2] -1/-1/-1->10->11 [3] -1/-1/-1->10->11 [4] 11/12/8->10->6 [5] 11/12/8->10->6 [6] -1/-1/-1->10->11 [7] -1/-1/-1->10->11
+ip-26-0-166-244:897246:898036 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-36:104436:105197 [7] NCCL INFO Trees [0] -1/-1/-1->13->12 [1] -1/-1/-1->13->12 [2] 12/11/15->13->9 [3] 12/11/15->13->9 [4] -1/-1/-1->13->12 [5] -1/-1/-1->13->12 [6] 12/-1/-1->13->11 [7] 12/-1/-1->13->11
+ip-26-0-166-36:104435:105196 [6] NCCL INFO Trees [0] 13/10/14->12->8 [1] 13/10/14->12->8 [2] -1/-1/-1->12->13 [3] -1/-1/-1->12->13 [4] 13/-1/-1->12->10 [5] 13/-1/-1->12->10 [6] -1/-1/-1->12->13 [7] -1/-1/-1->12->13
+ip-26-0-166-36:104436:105197 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-36:104435:105196 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-167-9:1039544:1040322 [7] NCCL INFO Trees [0] -1/-1/-1->15->14 [1] -1/-1/-1->15->14 [2] 14/-1/-1->15->13 [3] 14/-1/-1->15->13 [4] -1/-1/-1->15->14 [5] -1/-1/-1->15->14 [6] 14/7/-1->15->-1 [7] 14/7/-1->15->-1
+ip-26-0-167-9:1039544:1040322 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-162-46:702887:703686 [7] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/9/-1->1->-1 [3] 0/9/-1->1->-1 [4] -1/-1/-1->1->0 [5] -1/-1/-1->1->0 [6] 0/-1/-1->1->3 [7] 0/-1/-1->1->3
+ip-26-0-162-46:702887:703686 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-162-46:702886:703685 [6] NCCL INFO Channel 00/08 :    0   1   2   3   4   5   6   7   8   9  10  11  12  13  14  15
+ip-26-0-162-46:702886:703685 [6] NCCL INFO Channel 01/08 :    0   3   2   5   4   7   6   9   8  11  10  13  12  15  14   1
+ip-26-0-162-46:702886:703685 [6] NCCL INFO Channel 02/08 :    0   1   2   3   4   5   6   7   8   9  10  11  12  13  14  15
+ip-26-0-162-46:702886:703685 [6] NCCL INFO Channel 03/08 :    0   3   2   5   4   7   6   9   8  11  10  13  12  15  14   1
+ip-26-0-162-46:702886:703685 [6] NCCL INFO Channel 04/08 :    0   1   2   3   4   5   6   7   8   9  10  11  12  13  14  15
+ip-26-0-162-46:702886:703685 [6] NCCL INFO Channel 05/08 :    0   3   2   5   4   7   6   9   8  11  10  13  12  15  14   1
+ip-26-0-162-46:702886:703685 [6] NCCL INFO Channel 06/08 :    0   1   2   3   4   5   6   7   8   9  10  11  12  13  14  15
+ip-26-0-162-46:702886:703685 [6] NCCL INFO Channel 07/08 :    0   3   2   5   4   7   6   9   8  11  10  13  12  15  14   1
+ip-26-0-167-9:1039543:1040323 [6] NCCL INFO Trees [0] 15/-1/-1->14->12 [1] 15/-1/-1->14->12 [2] -1/-1/-1->14->15 [3] -1/-1/-1->14->15 [4] 15/6/-1->14->-1 [5] 15/6/-1->14->-1 [6] -1/-1/-1->14->15 [7] -1/-1/-1->14->15
+ip-26-0-167-9:1039543:1040323 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-162-46:702886:703685 [6] NCCL INFO Trees [0] 1/8/-1->0->-1 [1] 1/8/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1 [4] 1/-1/-1->0->2 [5] 1/-1/-1->0->2 [6] -1/-1/-1->0->1 [7] -1/-1/-1->0->1
+ip-26-0-162-46:702886:703685 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-213:9631:10396 [6] NCCL INFO Trees [0] 3/-1/-1->2->4 [1] 3/-1/-1->2->4 [2] -1/-1/-1->2->3 [3] -1/-1/-1->2->3 [4] 3/4/0->2->6 [5] 3/4/0->2->6 [6] -1/-1/-1->2->3 [7] -1/-1/-1->2->3
+ip-26-0-165-213:9631:10396 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-125:762818:763616 [3] NCCL INFO Trees [0] -1/-1/-1->5->4 [1] -1/-1/-1->5->4 [2] 4/3/7->5->9 [3] 4/3/7->5->9 [4] -1/-1/-1->5->4 [5] -1/-1/-1->5->4 [6] 4/-1/-1->5->3 [7] 4/-1/-1->5->3
+ip-26-0-166-125:762817:763617 [2] NCCL INFO Trees [0] 5/2/6->4->8 [1] 5/2/6->4->8 [2] -1/-1/-1->4->5 [3] -1/-1/-1->4->5 [4] 5/-1/-1->4->2 [5] 5/-1/-1->4->2 [6] -1/-1/-1->4->5 [7] -1/-1/-1->4->5
+ip-26-0-166-125:762818:763616 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-125:762817:763617 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-213:9627:10393 [2] NCCL INFO Trees [0] 3/-1/-1->2->4 [1] 3/-1/-1->2->4 [2] -1/-1/-1->2->3 [3] -1/-1/-1->2->3 [4] 3/4/0->2->6 [5] 3/4/0->2->6 [6] -1/-1/-1->2->3 [7] -1/-1/-1->2->3
+ip-26-0-165-213:9627:10393 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-165-213:9628:10394 [3] NCCL INFO Trees [0] -1/-1/-1->3->2 [1] -1/-1/-1->3->2 [2] 2/-1/-1->3->5 [3] 2/-1/-1->3->5 [4] -1/-1/-1->3->2 [5] -1/-1/-1->3->2 [6] 2/5/1->3->7 [7] 2/5/1->3->7
+ip-26-0-165-213:9628:10394 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-244:897242:898034 [2] NCCL INFO Trees [0] 11/-1/-1->10->12 [1] 11/-1/-1->10->12 [2] -1/-1/-1->10->11 [3] -1/-1/-1->10->11 [4] 11/12/8->10->6 [5] 11/12/8->10->6 [6] -1/-1/-1->10->11 [7] -1/-1/-1->10->11
+ip-26-0-166-244:897242:898034 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-36:104432:105194 [3] NCCL INFO Trees [0] -1/-1/-1->13->12 [1] -1/-1/-1->13->12 [2] 12/11/15->13->9 [3] 12/11/15->13->9 [4] -1/-1/-1->13->12 [5] -1/-1/-1->13->12 [6] 12/-1/-1->13->11 [7] 12/-1/-1->13->11
+ip-26-0-166-36:104431:105195 [2] NCCL INFO Trees [0] 13/10/14->12->8 [1] 13/10/14->12->8 [2] -1/-1/-1->12->13 [3] -1/-1/-1->12->13 [4] 13/-1/-1->12->10 [5] 13/-1/-1->12->10 [6] -1/-1/-1->12->13 [7] -1/-1/-1->12->13
+ip-26-0-166-36:104432:105194 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-36:104431:105195 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-15:127292:128057 [2] NCCL INFO Trees [0] 7/-1/-1->6->4 [1] 7/-1/-1->6->4 [2] -1/-1/-1->6->7 [3] -1/-1/-1->6->7 [4] 7/10/2->6->14 [5] 7/10/2->6->14 [6] -1/-1/-1->6->7 [7] -1/-1/-1->6->7
+ip-26-0-166-15:127292:128057 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-15:127293:128058 [3] NCCL INFO Trees [0] -1/-1/-1->7->6 [1] -1/-1/-1->7->6 [2] 6/-1/-1->7->5 [3] 6/-1/-1->7->5 [4] -1/-1/-1->7->6 [5] -1/-1/-1->7->6 [6] 6/11/3->7->15 [7] 6/11/3->7->15
+ip-26-0-166-15:127293:128058 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-162-46:702882:703682 [2] NCCL INFO Channel 00/08 :    0   1   2   3   4   5   6   7   8   9  10  11  12  13  14  15
+ip-26-0-162-46:702882:703682 [2] NCCL INFO Channel 01/08 :    0   3   2   5   4   7   6   9   8  11  10  13  12  15  14   1
+ip-26-0-162-46:702882:703682 [2] NCCL INFO Channel 02/08 :    0   1   2   3   4   5   6   7   8   9  10  11  12  13  14  15
+ip-26-0-162-46:702882:703682 [2] NCCL INFO Channel 03/08 :    0   3   2   5   4   7   6   9   8  11  10  13  12  15  14   1
+ip-26-0-162-46:702882:703682 [2] NCCL INFO Channel 04/08 :    0   1   2   3   4   5   6   7   8   9  10  11  12  13  14  15
+ip-26-0-162-46:702882:703682 [2] NCCL INFO Channel 05/08 :    0   3   2   5   4   7   6   9   8  11  10  13  12  15  14   1
+ip-26-0-162-46:702882:703682 [2] NCCL INFO Channel 06/08 :    0   1   2   3   4   5   6   7   8   9  10  11  12  13  14  15
+ip-26-0-162-46:702882:703682 [2] NCCL INFO Channel 07/08 :    0   3   2   5   4   7   6   9   8  11  10  13  12  15  14   1
+ip-26-0-167-9:1039540:1040320 [3] NCCL INFO Trees [0] -1/-1/-1->15->14 [1] -1/-1/-1->15->14 [2] 14/-1/-1->15->13 [3] 14/-1/-1->15->13 [4] -1/-1/-1->15->14 [5] -1/-1/-1->15->14 [6] 14/7/-1->15->-1 [7] 14/7/-1->15->-1
+ip-26-0-167-9:1039540:1040320 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-162-46:702883:703683 [3] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/9/-1->1->-1 [3] 0/9/-1->1->-1 [4] -1/-1/-1->1->0 [5] -1/-1/-1->1->0 [6] 0/-1/-1->1->3 [7] 0/-1/-1->1->3
+ip-26-0-162-46:702882:703682 [2] NCCL INFO Trees [0] 1/8/-1->0->-1 [1] 1/8/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1 [4] 1/-1/-1->0->2 [5] 1/-1/-1->0->2 [6] -1/-1/-1->0->1 [7] -1/-1/-1->0->1
+ip-26-0-162-46:702882:703682 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-162-46:702883:703683 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-244:897243:898033 [3] NCCL INFO Trees [0] -1/-1/-1->11->10 [1] -1/-1/-1->11->10 [2] 10/-1/-1->11->13 [3] 10/-1/-1->11->13 [4] -1/-1/-1->11->10 [5] -1/-1/-1->11->10 [6] 10/13/9->11->7 [7] 10/13/9->11->7
+ip-26-0-166-244:897243:898033 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-167-9:1039539:1040321 [2] NCCL INFO Trees [0] 15/-1/-1->14->12 [1] 15/-1/-1->14->12 [2] -1/-1/-1->14->15 [3] -1/-1/-1->14->15 [4] 15/6/-1->14->-1 [5] 15/6/-1->14->-1 [6] -1/-1/-1->14->15 [7] -1/-1/-1->14->15
+ip-26-0-167-9:1039539:1040321 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-214:624946:625742 [3] NCCL INFO Trees [0] -1/-1/-1->9->8 [1] -1/-1/-1->9->8 [2] 8/5/13->9->1 [3] 8/5/13->9->1 [4] -1/-1/-1->9->8 [5] -1/-1/-1->9->8 [6] 8/-1/-1->9->11 [7] 8/-1/-1->9->11
+ip-26-0-166-214:624946:625742 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-214:624945:625741 [2] NCCL INFO Trees [0] 9/4/12->8->0 [1] 9/4/12->8->0 [2] -1/-1/-1->8->9 [3] -1/-1/-1->8->9 [4] 9/-1/-1->8->10 [5] 9/-1/-1->8->10 [6] -1/-1/-1->8->9 [7] -1/-1/-1->8->9
+ip-26-0-166-214:624945:625741 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-166-36:104429:105191 [0] NCCL INFO Channel 00/0 : 11[1] -> 12[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104429:105191 [0] NCCL INFO Channel 02/0 : 11[1] -> 12[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104430:105190 [1] NCCL INFO Channel 00/0 : 13[1] -> 14[0] [send] via NET/Libfabric/0(12)/GDRDMA
+ip-26-0-166-36:104429:105191 [0] NCCL INFO Channel 04/0 : 11[1] -> 12[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104430:105190 [1] NCCL INFO Channel 02/0 : 13[1] -> 14[0] [send] via NET/Libfabric/0(12)/GDRDMA
+ip-26-0-166-36:104429:105191 [0] NCCL INFO Channel 06/0 : 11[1] -> 12[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104429:105191 [0] NCCL INFO Channel 00/0 : 12[0] -> 13[1] via P2P/IPC
+ip-26-0-166-36:104430:105190 [1] NCCL INFO Channel 04/0 : 13[1] -> 14[0] [send] via NET/Libfabric/0(12)/GDRDMA
+ip-26-0-166-36:104429:105191 [0] NCCL INFO Channel 02/0 : 12[0] -> 13[1] via P2P/IPC
+ip-26-0-166-36:104430:105190 [1] NCCL INFO Channel 06/0 : 13[1] -> 14[0] [send] via NET/Libfabric/0(12)/GDRDMA
+ip-26-0-166-36:104429:105191 [0] NCCL INFO Channel 04/0 : 12[0] -> 13[1] via P2P/IPC
+ip-26-0-166-125:762815:763613 [0] NCCL INFO Channel 00/0 : 3[1] -> 4[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104429:105191 [0] NCCL INFO Channel 06/0 : 12[0] -> 13[1] via P2P/IPC
+ip-26-0-166-125:762815:763613 [0] NCCL INFO Channel 02/0 : 3[1] -> 4[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762816:763612 [1] NCCL INFO Channel 00/0 : 5[1] -> 6[0] [send] via NET/Libfabric/0(4)/GDRDMA
+ip-26-0-166-125:762815:763613 [0] NCCL INFO Channel 04/0 : 3[1] -> 4[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762816:763612 [1] NCCL INFO Channel 02/0 : 5[1] -> 6[0] [send] via NET/Libfabric/0(4)/GDRDMA
+ip-26-0-166-125:762815:763613 [0] NCCL INFO Channel 06/0 : 3[1] -> 4[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762815:763613 [0] NCCL INFO Channel 00/0 : 4[0] -> 5[1] via P2P/IPC
+ip-26-0-166-125:762816:763612 [1] NCCL INFO Channel 04/0 : 5[1] -> 6[0] [send] via NET/Libfabric/0(4)/GDRDMA
+ip-26-0-165-213:9625:10390 [0] NCCL INFO Channel 00/0 : 1[1] -> 2[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9625:10390 [0] NCCL INFO Channel 02/0 : 1[1] -> 2[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9626:10389 [1] NCCL INFO Channel 00/0 : 3[1] -> 4[0] [send] via NET/Libfabric/0(2)/GDRDMA
+ip-26-0-165-213:9625:10390 [0] NCCL INFO Channel 04/0 : 1[1] -> 2[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9626:10389 [1] NCCL INFO Channel 02/0 : 3[1] -> 4[0] [send] via NET/Libfabric/0(2)/GDRDMA
+ip-26-0-165-213:9625:10390 [0] NCCL INFO Channel 06/0 : 1[1] -> 2[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9625:10390 [0] NCCL INFO Channel 00/0 : 2[0] -> 3[1] via P2P/IPC
+ip-26-0-166-125:762815:763613 [0] NCCL INFO Channel 02/0 : 4[0] -> 5[1] via P2P/IPC
+ip-26-0-165-213:9626:10389 [1] NCCL INFO Channel 04/0 : 3[1] -> 4[0] [send] via NET/Libfabric/0(2)/GDRDMA
+ip-26-0-166-125:762816:763612 [1] NCCL INFO Channel 06/0 : 5[1] -> 6[0] [send] via NET/Libfabric/0(4)/GDRDMA
+ip-26-0-166-244:897240:898030 [0] NCCL INFO Channel 00/0 : 9[1] -> 10[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897240:898030 [0] NCCL INFO Channel 02/0 : 9[1] -> 10[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127290:128054 [0] NCCL INFO Channel 00/0 : 5[1] -> 6[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897241:898029 [1] NCCL INFO Channel 00/0 : 11[1] -> 12[0] [send] via NET/Libfabric/0(10)/GDRDMA
+ip-26-0-166-15:127290:128054 [0] NCCL INFO Channel 02/0 : 5[1] -> 6[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897240:898030 [0] NCCL INFO Channel 04/0 : 9[1] -> 10[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897241:898029 [1] NCCL INFO Channel 02/0 : 11[1] -> 12[0] [send] via NET/Libfabric/0(10)/GDRDMA
+ip-26-0-167-9:1039537:1040316 [0] NCCL INFO Channel 00/0 : 13[1] -> 14[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762815:763613 [0] NCCL INFO Channel 04/0 : 4[0] -> 5[1] via P2P/IPC
+ip-26-0-166-244:897240:898030 [0] NCCL INFO Channel 06/0 : 9[1] -> 10[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127291:128053 [1] NCCL INFO Channel 00/0 : 7[1] -> 8[0] [send] via NET/Libfabric/0(6)/GDRDMA
+ip-26-0-166-244:897240:898030 [0] NCCL INFO Channel 00/0 : 10[0] -> 11[1] via P2P/IPC
+ip-26-0-166-15:127290:128054 [0] NCCL INFO Channel 04/0 : 5[1] -> 6[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-167-9:1039537:1040316 [0] NCCL INFO Channel 02/0 : 13[1] -> 14[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897241:898029 [1] NCCL INFO Channel 04/0 : 11[1] -> 12[0] [send] via NET/Libfabric/0(10)/GDRDMA
+ip-26-0-166-15:127291:128053 [1] NCCL INFO Channel 02/0 : 7[1] -> 8[0] [send] via NET/Libfabric/0(6)/GDRDMA
+ip-26-0-167-9:1039538:1040317 [1] NCCL INFO Channel 00/0 : 15[1] -> 0[0] [send] via NET/Libfabric/0(14)/GDRDMA
+ip-26-0-166-15:127290:128054 [0] NCCL INFO Channel 06/0 : 5[1] -> 6[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127290:128054 [0] NCCL INFO Channel 00/0 : 6[0] -> 7[1] via P2P/IPC
+ip-26-0-167-9:1039537:1040316 [0] NCCL INFO Channel 04/0 : 13[1] -> 14[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127291:128053 [1] NCCL INFO Channel 04/0 : 7[1] -> 8[0] [send] via NET/Libfabric/0(6)/GDRDMA
+ip-26-0-167-9:1039538:1040317 [1] NCCL INFO Channel 02/0 : 15[1] -> 0[0] [send] via NET/Libfabric/0(14)/GDRDMA
+ip-26-0-166-125:762815:763613 [0] NCCL INFO Channel 06/0 : 4[0] -> 5[1] via P2P/IPC
+ip-26-0-167-9:1039537:1040316 [0] NCCL INFO Channel 06/0 : 13[1] -> 14[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-167-9:1039537:1040316 [0] NCCL INFO Channel 00/0 : 14[0] -> 15[1] via P2P/IPC
+ip-26-0-167-9:1039538:1040317 [1] NCCL INFO Channel 04/0 : 15[1] -> 0[0] [send] via NET/Libfabric/0(14)/GDRDMA
+ip-26-0-165-213:9625:10390 [0] NCCL INFO Channel 02/0 : 2[0] -> 3[1] via P2P/IPC
+ip-26-0-165-213:9626:10389 [1] NCCL INFO Channel 06/0 : 3[1] -> 4[0] [send] via NET/Libfabric/0(2)/GDRDMA
+ip-26-0-166-244:897240:898030 [0] NCCL INFO Channel 02/0 : 10[0] -> 11[1] via P2P/IPC
+ip-26-0-166-244:897241:898029 [1] NCCL INFO Channel 06/0 : 11[1] -> 12[0] [send] via NET/Libfabric/0(10)/GDRDMA
+ip-26-0-165-213:9625:10390 [0] NCCL INFO Channel 04/0 : 2[0] -> 3[1] via P2P/IPC
+ip-26-0-167-9:1039537:1040316 [0] NCCL INFO Channel 02/0 : 14[0] -> 15[1] via P2P/IPC
+ip-26-0-166-244:897240:898030 [0] NCCL INFO Channel 04/0 : 10[0] -> 11[1] via P2P/IPC
+ip-26-0-166-15:127290:128054 [0] NCCL INFO Channel 02/0 : 6[0] -> 7[1] via P2P/IPC
+ip-26-0-167-9:1039538:1040317 [1] NCCL INFO Channel 06/0 : 15[1] -> 0[0] [send] via NET/Libfabric/0(14)/GDRDMA
+ip-26-0-165-213:9625:10390 [0] NCCL INFO Channel 06/0 : 2[0] -> 3[1] via P2P/IPC
+ip-26-0-166-15:127291:128053 [1] NCCL INFO Channel 06/0 : 7[1] -> 8[0] [send] via NET/Libfabric/0(6)/GDRDMA
+ip-26-0-166-244:897240:898030 [0] NCCL INFO Channel 06/0 : 10[0] -> 11[1] via P2P/IPC
+ip-26-0-167-9:1039537:1040316 [0] NCCL INFO Channel 04/0 : 14[0] -> 15[1] via P2P/IPC
+ip-26-0-166-15:127290:128054 [0] NCCL INFO Channel 04/0 : 6[0] -> 7[1] via P2P/IPC
+ip-26-0-167-9:1039537:1040316 [0] NCCL INFO Channel 06/0 : 14[0] -> 15[1] via P2P/IPC
+ip-26-0-162-46:702880:703676 [0] NCCL INFO Channel 00/0 : 15[1] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127290:128054 [0] NCCL INFO Channel 06/0 : 6[0] -> 7[1] via P2P/IPC
+ip-26-0-162-46:702880:703676 [0] NCCL INFO Channel 02/0 : 15[1] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-46:702881:703677 [1] NCCL INFO Channel 00/0 : 1[1] -> 2[0] [send] via NET/Libfabric/0(0)/GDRDMA
+ip-26-0-162-46:702880:703676 [0] NCCL INFO Channel 04/0 : 15[1] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-46:702881:703677 [1] NCCL INFO Channel 02/0 : 1[1] -> 2[0] [send] via NET/Libfabric/0(0)/GDRDMA
+ip-26-0-162-46:702880:703676 [0] NCCL INFO Channel 06/0 : 15[1] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-46:702880:703676 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-162-46:702881:703677 [1] NCCL INFO Channel 04/0 : 1[1] -> 2[0] [send] via NET/Libfabric/0(0)/GDRDMA
+ip-26-0-166-36:104429:105191 [0] NCCL INFO Channel 01/0 : 12[0] -> 15[1] [send] via NET/Libfabric/1(13)/GDRDMA
+ip-26-0-162-46:702880:703676 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-162-46:702881:703677 [1] NCCL INFO Channel 06/0 : 1[1] -> 2[0] [send] via NET/Libfabric/0(0)/GDRDMA
+ip-26-0-166-125:762815:763613 [0] NCCL INFO Channel 01/0 : 4[0] -> 7[1] [send] via NET/Libfabric/1(5)/GDRDMA
+ip-26-0-162-46:702880:703676 [0] NCCL INFO Channel 04/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-125:762815:763613 [0] NCCL INFO Channel 03/0 : 4[0] -> 7[1] [send] via NET/Libfabric/1(5)/GDRDMA
+ip-26-0-166-125:762815:763613 [0] NCCL INFO Channel 05/0 : 4[0] -> 7[1] [send] via NET/Libfabric/1(5)/GDRDMA
+ip-26-0-162-46:702880:703676 [0] NCCL INFO Channel 06/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-167-9:1039537:1040316 [0] NCCL INFO Channel 01/0 : 14[0] -> 1[1] [send] via NET/Libfabric/1(15)/GDRDMA
+ip-26-0-167-9:1039537:1040316 [0] NCCL INFO Channel 03/0 : 14[0] -> 1[1] [send] via NET/Libfabric/1(15)/GDRDMA
+ip-26-0-167-9:1039537:1040316 [0] NCCL INFO Channel 05/0 : 14[0] -> 1[1] [send] via NET/Libfabric/1(15)/GDRDMA
+ip-26-0-167-9:1039537:1040316 [0] NCCL INFO Channel 07/0 : 14[0] -> 1[1] [send] via NET/Libfabric/1(15)/GDRDMA
+ip-26-0-166-36:104433:105193 [4] NCCL INFO Channel 00/0 : 11[5] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104433:105193 [4] NCCL INFO Channel 02/0 : 11[5] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104434:105192 [5] NCCL INFO Channel 00/0 : 13[5] -> 14[4] [send] via NET/Libfabric/4(12)/GDRDMA
+ip-26-0-166-36:104433:105193 [4] NCCL INFO Channel 04/0 : 11[5] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104434:105192 [5] NCCL INFO Channel 02/0 : 13[5] -> 14[4] [send] via NET/Libfabric/4(12)/GDRDMA
+ip-26-0-166-36:104433:105193 [4] NCCL INFO Channel 06/0 : 11[5] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104433:105193 [4] NCCL INFO Channel 00/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-166-36:104434:105192 [5] NCCL INFO Channel 04/0 : 13[5] -> 14[4] [send] via NET/Libfabric/4(12)/GDRDMA
+ip-26-0-166-36:104433:105193 [4] NCCL INFO Channel 02/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-166-36:104434:105192 [5] NCCL INFO Channel 06/0 : 13[5] -> 14[4] [send] via NET/Libfabric/4(12)/GDRDMA
+ip-26-0-166-244:897244:898032 [4] NCCL INFO Channel 00/0 : 9[5] -> 10[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897245:898031 [5] NCCL INFO Channel 00/0 : 11[5] -> 12[4] [send] via NET/Libfabric/4(10)/GDRDMA
+ip-26-0-166-244:897244:898032 [4] NCCL INFO Channel 02/0 : 9[5] -> 10[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104433:105193 [4] NCCL INFO Channel 04/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-166-244:897245:898031 [5] NCCL INFO Channel 02/0 : 11[5] -> 12[4] [send] via NET/Libfabric/4(10)/GDRDMA
+ip-26-0-166-244:897244:898032 [4] NCCL INFO Channel 04/0 : 9[5] -> 10[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897245:898031 [5] NCCL INFO Channel 04/0 : 11[5] -> 12[4] [send] via NET/Libfabric/4(10)/GDRDMA
+ip-26-0-166-244:897244:898032 [4] NCCL INFO Channel 06/0 : 9[5] -> 10[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897244:898032 [4] NCCL INFO Channel 00/0 : 10[4] -> 11[5] via P2P/IPC
+ip-26-0-162-46:702884:703679 [4] NCCL INFO Channel 00/0 : 15[5] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-46:702885:703680 [5] NCCL INFO Channel 00/0 : 1[5] -> 2[4] [send] via NET/Libfabric/4(0)/GDRDMA
+ip-26-0-166-244:897245:898031 [5] NCCL INFO Channel 06/0 : 11[5] -> 12[4] [send] via NET/Libfabric/4(10)/GDRDMA
+ip-26-0-166-36:104433:105193 [4] NCCL INFO Channel 06/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-166-125:762819:763614 [4] NCCL INFO Channel 00/0 : 3[5] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-46:702884:703679 [4] NCCL INFO Channel 02/0 : 15[5] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-46:702885:703680 [5] NCCL INFO Channel 02/0 : 1[5] -> 2[4] [send] via NET/Libfabric/4(0)/GDRDMA
+ip-26-0-166-125:762819:763614 [4] NCCL INFO Channel 02/0 : 3[5] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-46:702884:703679 [4] NCCL INFO Channel 04/0 : 15[5] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-46:702885:703680 [5] NCCL INFO Channel 04/0 : 1[5] -> 2[4] [send] via NET/Libfabric/4(0)/GDRDMA
+ip-26-0-166-125:762820:763615 [5] NCCL INFO Channel 00/0 : 5[5] -> 6[4] [send] via NET/Libfabric/4(4)/GDRDMA
+ip-26-0-162-46:702884:703679 [4] NCCL INFO Channel 06/0 : 15[5] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-46:702884:703679 [4] NCCL INFO Channel 00/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-162-46:702885:703680 [5] NCCL INFO Channel 06/0 : 1[5] -> 2[4] [send] via NET/Libfabric/4(0)/GDRDMA
+ip-26-0-167-9:1039541:1040318 [4] NCCL INFO Channel 00/0 : 13[5] -> 14[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762819:763614 [4] NCCL INFO Channel 04/0 : 3[5] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762820:763615 [5] NCCL INFO Channel 02/0 : 5[5] -> 6[4] [send] via NET/Libfabric/4(4)/GDRDMA
+ip-26-0-166-125:762819:763614 [4] NCCL INFO Channel 06/0 : 3[5] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762819:763614 [4] NCCL INFO Channel 00/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-167-9:1039541:1040318 [4] NCCL INFO Channel 02/0 : 13[5] -> 14[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762820:763615 [5] NCCL INFO Channel 04/0 : 5[5] -> 6[4] [send] via NET/Libfabric/4(4)/GDRDMA
+ip-26-0-167-9:1039542:1040319 [5] NCCL INFO Channel 00/0 : 15[5] -> 0[4] [send] via NET/Libfabric/4(14)/GDRDMA
+ip-26-0-167-9:1039541:1040318 [4] NCCL INFO Channel 04/0 : 13[5] -> 14[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897244:898032 [4] NCCL INFO Channel 02/0 : 10[4] -> 11[5] via P2P/IPC
+ip-26-0-167-9:1039542:1040319 [5] NCCL INFO Channel 02/0 : 15[5] -> 0[4] [send] via NET/Libfabric/4(14)/GDRDMA
+ip-26-0-166-15:127294:128056 [4] NCCL INFO Channel 00/0 : 5[5] -> 6[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-167-9:1039541:1040318 [4] NCCL INFO Channel 06/0 : 13[5] -> 14[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-167-9:1039541:1040318 [4] NCCL INFO Channel 00/0 : 14[4] -> 15[5] via P2P/IPC
+ip-26-0-166-15:127294:128056 [4] NCCL INFO Channel 02/0 : 5[5] -> 6[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-167-9:1039542:1040319 [5] NCCL INFO Channel 04/0 : 15[5] -> 0[4] [send] via NET/Libfabric/4(14)/GDRDMA
+ip-26-0-166-15:127295:128055 [5] NCCL INFO Channel 00/0 : 7[5] -> 8[4] [send] via NET/Libfabric/4(6)/GDRDMA
+ip-26-0-166-244:897244:898032 [4] NCCL INFO Channel 04/0 : 10[4] -> 11[5] via P2P/IPC
+ip-26-0-166-15:127294:128056 [4] NCCL INFO Channel 04/0 : 5[5] -> 6[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127295:128055 [5] NCCL INFO Channel 02/0 : 7[5] -> 8[4] [send] via NET/Libfabric/4(6)/GDRDMA
+ip-26-0-166-15:127294:128056 [4] NCCL INFO Channel 06/0 : 5[5] -> 6[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127294:128056 [4] NCCL INFO Channel 00/0 : 6[4] -> 7[5] via P2P/IPC
+ip-26-0-165-213:9629:10391 [4] NCCL INFO Channel 00/0 : 1[5] -> 2[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127295:128055 [5] NCCL INFO Channel 04/0 : 7[5] -> 8[4] [send] via NET/Libfabric/4(6)/GDRDMA
+ip-26-0-165-213:9629:10391 [4] NCCL INFO Channel 02/0 : 1[5] -> 2[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897244:898032 [4] NCCL INFO Channel 06/0 : 10[4] -> 11[5] via P2P/IPC
+ip-26-0-165-213:9630:10392 [5] NCCL INFO Channel 00/0 : 3[5] -> 4[4] [send] via NET/Libfabric/4(2)/GDRDMA
+ip-26-0-162-46:702884:703679 [4] NCCL INFO Channel 02/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-166-125:762819:763614 [4] NCCL INFO Channel 02/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-165-213:9629:10391 [4] NCCL INFO Channel 04/0 : 1[5] -> 2[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9630:10392 [5] NCCL INFO Channel 02/0 : 3[5] -> 4[4] [send] via NET/Libfabric/4(2)/GDRDMA
+ip-26-0-165-213:9629:10391 [4] NCCL INFO Channel 06/0 : 1[5] -> 2[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762820:763615 [5] NCCL INFO Channel 06/0 : 5[5] -> 6[4] [send] via NET/Libfabric/4(4)/GDRDMA
+ip-26-0-165-213:9629:10391 [4] NCCL INFO Channel 00/0 : 2[4] -> 3[5] via P2P/IPC
+ip-26-0-165-213:9630:10392 [5] NCCL INFO Channel 04/0 : 3[5] -> 4[4] [send] via NET/Libfabric/4(2)/GDRDMA
+ip-26-0-167-9:1039541:1040318 [4] NCCL INFO Channel 02/0 : 14[4] -> 15[5] via P2P/IPC
+ip-26-0-167-9:1039542:1040319 [5] NCCL INFO Channel 06/0 : 15[5] -> 0[4] [send] via NET/Libfabric/4(14)/GDRDMA
+ip-26-0-162-46:702884:703679 [4] NCCL INFO Channel 04/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-166-125:762819:763614 [4] NCCL INFO Channel 04/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-167-9:1039541:1040318 [4] NCCL INFO Channel 04/0 : 14[4] -> 15[5] via P2P/IPC
+ip-26-0-166-15:127294:128056 [4] NCCL INFO Channel 02/0 : 6[4] -> 7[5] via P2P/IPC
+ip-26-0-166-125:762819:763614 [4] NCCL INFO Channel 06/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-162-46:702884:703679 [4] NCCL INFO Channel 06/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-166-15:127295:128055 [5] NCCL INFO Channel 06/0 : 7[5] -> 8[4] [send] via NET/Libfabric/4(6)/GDRDMA
+ip-26-0-167-9:1039541:1040318 [4] NCCL INFO Channel 06/0 : 14[4] -> 15[5] via P2P/IPC
+ip-26-0-166-15:127294:128056 [4] NCCL INFO Channel 04/0 : 6[4] -> 7[5] via P2P/IPC
+ip-26-0-165-213:9629:10391 [4] NCCL INFO Channel 02/0 : 2[4] -> 3[5] via P2P/IPC
+ip-26-0-165-213:9630:10392 [5] NCCL INFO Channel 06/0 : 3[5] -> 4[4] [send] via NET/Libfabric/4(2)/GDRDMA
+ip-26-0-166-36:104433:105193 [4] NCCL INFO Channel 01/0 : 12[4] -> 15[5] [send] via NET/Libfabric/5(13)/GDRDMA
+ip-26-0-166-36:104433:105193 [4] NCCL INFO Channel 03/0 : 12[4] -> 15[5] [send] via NET/Libfabric/5(13)/GDRDMA
+ip-26-0-166-15:127294:128056 [4] NCCL INFO Channel 06/0 : 6[4] -> 7[5] via P2P/IPC
+ip-26-0-166-36:104433:105193 [4] NCCL INFO Channel 05/0 : 12[4] -> 15[5] [send] via NET/Libfabric/5(13)/GDRDMA
+ip-26-0-165-213:9629:10391 [4] NCCL INFO Channel 04/0 : 2[4] -> 3[5] via P2P/IPC
+ip-26-0-165-213:9629:10391 [4] NCCL INFO Channel 06/0 : 2[4] -> 3[5] via P2P/IPC
+ip-26-0-167-9:1039541:1040318 [4] NCCL INFO Channel 01/0 : 14[4] -> 1[5] [send] via NET/Libfabric/5(15)/GDRDMA
+ip-26-0-162-46:702884:703679 [4] NCCL INFO Channel 01/0 : 0[4] -> 3[5] [send] via NET/Libfabric/5(1)/GDRDMA
+ip-26-0-166-125:762819:763614 [4] NCCL INFO Channel 01/0 : 4[4] -> 7[5] [send] via NET/Libfabric/5(5)/GDRDMA
+ip-26-0-166-15:127296:128059 [6] NCCL INFO Channel 00/0 : 5[7] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127296:128059 [6] NCCL INFO Channel 02/0 : 5[7] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127297:128060 [7] NCCL INFO Channel 00/0 : 7[7] -> 8[6] [send] via NET/Libfabric/6(6)/GDRDMA
+ip-26-0-166-15:127296:128059 [6] NCCL INFO Channel 04/0 : 5[7] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127297:128060 [7] NCCL INFO Channel 02/0 : 7[7] -> 8[6] [send] via NET/Libfabric/6(6)/GDRDMA
+ip-26-0-166-15:127296:128059 [6] NCCL INFO Channel 06/0 : 5[7] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127296:128059 [6] NCCL INFO Channel 00/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-15:127297:128060 [7] NCCL INFO Channel 04/0 : 7[7] -> 8[6] [send] via NET/Libfabric/6(6)/GDRDMA
+ip-26-0-166-36:104435:105196 [6] NCCL INFO Channel 00/0 : 11[7] -> 12[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-46:702886:703685 [6] NCCL INFO Channel 00/0 : 15[7] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-46:702886:703685 [6] NCCL INFO Channel 02/0 : 15[7] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897246:898036 [6] NCCL INFO Channel 00/0 : 9[7] -> 10[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-46:702887:703686 [7] NCCL INFO Channel 00/0 : 1[7] -> 2[6] [send] via NET/Libfabric/6(0)/GDRDMA
+ip-26-0-166-36:104436:105197 [7] NCCL INFO Channel 00/0 : 13[7] -> 14[6] [send] via NET/Libfabric/6(12)/GDRDMA
+ip-26-0-162-46:702886:703685 [6] NCCL INFO Channel 04/0 : 15[7] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762821:763619 [6] NCCL INFO Channel 00/0 : 3[7] -> 4[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897247:898035 [7] NCCL INFO Channel 00/0 : 11[7] -> 12[6] [send] via NET/Libfabric/6(10)/GDRDMA
+ip-26-0-162-46:702887:703686 [7] NCCL INFO Channel 02/0 : 1[7] -> 2[6] [send] via NET/Libfabric/6(0)/GDRDMA
+ip-26-0-166-36:104435:105196 [6] NCCL INFO Channel 02/0 : 11[7] -> 12[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-46:702886:703685 [6] NCCL INFO Channel 06/0 : 15[7] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104436:105197 [7] NCCL INFO Channel 02/0 : 13[7] -> 14[6] [send] via NET/Libfabric/6(12)/GDRDMA
+ip-26-0-162-46:702886:703685 [6] NCCL INFO Channel 00/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-166-36:104435:105196 [6] NCCL INFO Channel 04/0 : 11[7] -> 12[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762822:763618 [7] NCCL INFO Channel 00/0 : 5[7] -> 6[6] [send] via NET/Libfabric/6(4)/GDRDMA
+ip-26-0-166-244:897246:898036 [6] NCCL INFO Channel 02/0 : 9[7] -> 10[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-46:702887:703686 [7] NCCL INFO Channel 04/0 : 1[7] -> 2[6] [send] via NET/Libfabric/6(0)/GDRDMA
+ip-26-0-166-244:897247:898035 [7] NCCL INFO Channel 02/0 : 11[7] -> 12[6] [send] via NET/Libfabric/6(10)/GDRDMA
+ip-26-0-166-244:897246:898036 [6] NCCL INFO Channel 04/0 : 9[7] -> 10[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104436:105197 [7] NCCL INFO Channel 04/0 : 13[7] -> 14[6] [send] via NET/Libfabric/6(12)/GDRDMA
+ip-26-0-166-125:762821:763619 [6] NCCL INFO Channel 02/0 : 3[7] -> 4[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104435:105196 [6] NCCL INFO Channel 06/0 : 11[7] -> 12[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897247:898035 [7] NCCL INFO Channel 04/0 : 11[7] -> 12[6] [send] via NET/Libfabric/6(10)/GDRDMA
+ip-26-0-166-244:897246:898036 [6] NCCL INFO Channel 06/0 : 9[7] -> 10[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762822:763618 [7] NCCL INFO Channel 02/0 : 5[7] -> 6[6] [send] via NET/Libfabric/6(4)/GDRDMA
+ip-26-0-166-36:104435:105196 [6] NCCL INFO Channel 00/0 : 12[6] -> 13[7] via P2P/IPC
+ip-26-0-166-15:127296:128059 [6] NCCL INFO Channel 02/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-244:897246:898036 [6] NCCL INFO Channel 00/0 : 10[6] -> 11[7] via P2P/IPC
+ip-26-0-166-244:897247:898035 [7] NCCL INFO Channel 06/0 : 11[7] -> 12[6] [send] via NET/Libfabric/6(10)/GDRDMA
+ip-26-0-166-125:762821:763619 [6] NCCL INFO Channel 04/0 : 3[7] -> 4[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104436:105197 [7] NCCL INFO Channel 06/0 : 13[7] -> 14[6] [send] via NET/Libfabric/6(12)/GDRDMA
+ip-26-0-166-125:762822:763618 [7] NCCL INFO Channel 04/0 : 5[7] -> 6[6] [send] via NET/Libfabric/6(4)/GDRDMA
+ip-26-0-166-15:127297:128060 [7] NCCL INFO Channel 06/0 : 7[7] -> 8[6] [send] via NET/Libfabric/6(6)/GDRDMA
+ip-26-0-166-125:762821:763619 [6] NCCL INFO Channel 06/0 : 3[7] -> 4[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762821:763619 [6] NCCL INFO Channel 00/0 : 4[6] -> 5[7] via P2P/IPC
+ip-26-0-166-125:762822:763618 [7] NCCL INFO Channel 06/0 : 5[7] -> 6[6] [send] via NET/Libfabric/6(4)/GDRDMA
+ip-26-0-166-15:127296:128059 [6] NCCL INFO Channel 04/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-167-9:1039543:1040323 [6] NCCL INFO Channel 00/0 : 13[7] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-167-9:1039543:1040323 [6] NCCL INFO Channel 02/0 : 13[7] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897246:898036 [6] NCCL INFO Channel 02/0 : 10[6] -> 11[7] via P2P/IPC
+ip-26-0-167-9:1039544:1040322 [7] NCCL INFO Channel 00/0 : 15[7] -> 0[6] [send] via NET/Libfabric/6(14)/GDRDMA
+ip-26-0-166-15:127296:128059 [6] NCCL INFO Channel 06/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-162-46:702886:703685 [6] NCCL INFO Channel 02/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-167-9:1039543:1040323 [6] NCCL INFO Channel 04/0 : 13[7] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104435:105196 [6] NCCL INFO Channel 02/0 : 12[6] -> 13[7] via P2P/IPC
+ip-26-0-162-46:702887:703686 [7] NCCL INFO Channel 06/0 : 1[7] -> 2[6] [send] via NET/Libfabric/6(0)/GDRDMA
+ip-26-0-167-9:1039544:1040322 [7] NCCL INFO Channel 02/0 : 15[7] -> 0[6] [send] via NET/Libfabric/6(14)/GDRDMA
+ip-26-0-167-9:1039543:1040323 [6] NCCL INFO Channel 06/0 : 13[7] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-167-9:1039543:1040323 [6] NCCL INFO Channel 00/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-167-9:1039544:1040322 [7] NCCL INFO Channel 04/0 : 15[7] -> 0[6] [send] via NET/Libfabric/6(14)/GDRDMA
+ip-26-0-166-244:897246:898036 [6] NCCL INFO Channel 04/0 : 10[6] -> 11[7] via P2P/IPC
+ip-26-0-166-125:762821:763619 [6] NCCL INFO Channel 02/0 : 4[6] -> 5[7] via P2P/IPC
+ip-26-0-162-46:702886:703685 [6] NCCL INFO Channel 04/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-166-244:897246:898036 [6] NCCL INFO Channel 06/0 : 10[6] -> 11[7] via P2P/IPC
+ip-26-0-166-36:104435:105196 [6] NCCL INFO Channel 04/0 : 12[6] -> 13[7] via P2P/IPC
+ip-26-0-166-125:762821:763619 [6] NCCL INFO Channel 04/0 : 4[6] -> 5[7] via P2P/IPC
+ip-26-0-162-46:702886:703685 [6] NCCL INFO Channel 06/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-167-9:1039543:1040323 [6] NCCL INFO Channel 02/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-166-36:104435:105196 [6] NCCL INFO Channel 06/0 : 12[6] -> 13[7] via P2P/IPC
+ip-26-0-167-9:1039544:1040322 [7] NCCL INFO Channel 06/0 : 15[7] -> 0[6] [send] via NET/Libfabric/6(14)/GDRDMA
+ip-26-0-166-125:762821:763619 [6] NCCL INFO Channel 06/0 : 4[6] -> 5[7] via P2P/IPC
+ip-26-0-167-9:1039543:1040323 [6] NCCL INFO Channel 04/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-165-213:9631:10396 [6] NCCL INFO Channel 00/0 : 1[7] -> 2[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9631:10396 [6] NCCL INFO Channel 02/0 : 1[7] -> 2[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-167-9:1039543:1040323 [6] NCCL INFO Channel 06/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-165-213:9632:10395 [7] NCCL INFO Channel 00/0 : 3[7] -> 4[6] [send] via NET/Libfabric/6(2)/GDRDMA
+ip-26-0-165-213:9631:10396 [6] NCCL INFO Channel 04/0 : 1[7] -> 2[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9632:10395 [7] NCCL INFO Channel 02/0 : 3[7] -> 4[6] [send] via NET/Libfabric/6(2)/GDRDMA
+ip-26-0-165-213:9631:10396 [6] NCCL INFO Channel 06/0 : 1[7] -> 2[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9631:10396 [6] NCCL INFO Channel 00/0 : 2[6] -> 3[7] via P2P/IPC
+ip-26-0-165-213:9632:10395 [7] NCCL INFO Channel 04/0 : 3[7] -> 4[6] [send] via NET/Libfabric/6(2)/GDRDMA
+ip-26-0-166-15:127296:128059 [6] NCCL INFO Channel 01/0 : 6[6] -> 9[7] [send] via NET/Libfabric/7(7)/GDRDMA
+ip-26-0-165-213:9631:10396 [6] NCCL INFO Channel 02/0 : 2[6] -> 3[7] via P2P/IPC
+ip-26-0-165-213:9632:10395 [7] NCCL INFO Channel 06/0 : 3[7] -> 4[6] [send] via NET/Libfabric/6(2)/GDRDMA
+ip-26-0-166-244:897246:898036 [6] NCCL INFO Channel 01/0 : 10[6] -> 13[7] [send] via NET/Libfabric/7(11)/GDRDMA
+ip-26-0-166-244:897242:898034 [2] NCCL INFO Channel 00/0 : 9[3] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897242:898034 [2] NCCL INFO Channel 02/0 : 9[3] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-46:702882:703682 [2] NCCL INFO Channel 00/0 : 15[3] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897243:898033 [3] NCCL INFO Channel 00/0 : 11[3] -> 12[2] [send] via NET/Libfabric/2(10)/GDRDMA
+ip-26-0-165-213:9631:10396 [6] NCCL INFO Channel 04/0 : 2[6] -> 3[7] via P2P/IPC
+ip-26-0-166-244:897242:898034 [2] NCCL INFO Channel 04/0 : 9[3] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-46:702882:703682 [2] NCCL INFO Channel 02/0 : 15[3] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897243:898033 [3] NCCL INFO Channel 02/0 : 11[3] -> 12[2] [send] via NET/Libfabric/2(10)/GDRDMA
+ip-26-0-166-244:897242:898034 [2] NCCL INFO Channel 06/0 : 9[3] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897242:898034 [2] NCCL INFO Channel 00/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-162-46:702883:703683 [3] NCCL INFO Channel 00/0 : 1[3] -> 2[2] [send] via NET/Libfabric/2(0)/GDRDMA
+ip-26-0-162-46:702882:703682 [2] NCCL INFO Channel 04/0 : 15[3] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-167-9:1039539:1040321 [2] NCCL INFO Channel 00/0 : 13[3] -> 14[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897243:898033 [3] NCCL INFO Channel 04/0 : 11[3] -> 12[2] [send] via NET/Libfabric/2(10)/GDRDMA
+ip-26-0-162-46:702886:703685 [6] NCCL INFO Channel 01/0 : 0[6] -> 3[7] [send] via NET/Libfabric/7(1)/GDRDMA
+ip-26-0-162-46:702883:703683 [3] NCCL INFO Channel 02/0 : 1[3] -> 2[2] [send] via NET/Libfabric/2(0)/GDRDMA
+ip-26-0-166-36:104435:105196 [6] NCCL INFO Channel 01/0 : 12[6] -> 15[7] [send] via NET/Libfabric/7(13)/GDRDMA
+ip-26-0-162-46:702882:703682 [2] NCCL INFO Channel 06/0 : 15[3] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-46:702882:703682 [2] NCCL INFO Channel 00/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-165-213:9631:10396 [6] NCCL INFO Channel 06/0 : 2[6] -> 3[7] via P2P/IPC
+ip-26-0-167-9:1039539:1040321 [2] NCCL INFO Channel 02/0 : 13[3] -> 14[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-46:702886:703685 [6] NCCL INFO Channel 03/0 : 0[6] -> 3[7] [send] via NET/Libfabric/7(1)/GDRDMA
+ip-26-0-162-46:702883:703683 [3] NCCL INFO Channel 04/0 : 1[3] -> 2[2] [send] via NET/Libfabric/2(0)/GDRDMA
+ip-26-0-167-9:1039540:1040320 [3] NCCL INFO Channel 00/0 : 15[3] -> 0[2] [send] via NET/Libfabric/2(14)/GDRDMA
+ip-26-0-162-46:702886:703685 [6] NCCL INFO Channel 05/0 : 0[6] -> 3[7] [send] via NET/Libfabric/7(1)/GDRDMA
+ip-26-0-167-9:1039539:1040321 [2] NCCL INFO Channel 04/0 : 13[3] -> 14[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-46:702886:703685 [6] NCCL INFO Channel 07/0 : 0[6] -> 3[7] [send] via NET/Libfabric/7(1)/GDRDMA
+ip-26-0-167-9:1039540:1040320 [3] NCCL INFO Channel 02/0 : 15[3] -> 0[2] [send] via NET/Libfabric/2(14)/GDRDMA
+ip-26-0-167-9:1039539:1040321 [2] NCCL INFO Channel 06/0 : 13[3] -> 14[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-167-9:1039539:1040321 [2] NCCL INFO Channel 00/0 : 14[2] -> 15[3] via P2P/IPC
+ip-26-0-167-9:1039540:1040320 [3] NCCL INFO Channel 04/0 : 15[3] -> 0[2] [send] via NET/Libfabric/2(14)/GDRDMA
+ip-26-0-166-15:127292:128057 [2] NCCL INFO Channel 00/0 : 5[3] -> 6[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127292:128057 [2] NCCL INFO Channel 02/0 : 5[3] -> 6[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127293:128058 [3] NCCL INFO Channel 00/0 : 7[3] -> 8[2] [send] via NET/Libfabric/2(6)/GDRDMA
+ip-26-0-166-15:127292:128057 [2] NCCL INFO Channel 04/0 : 5[3] -> 6[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897242:898034 [2] NCCL INFO Channel 02/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-166-15:127293:128058 [3] NCCL INFO Channel 02/0 : 7[3] -> 8[2] [send] via NET/Libfabric/2(6)/GDRDMA
+ip-26-0-166-15:127292:128057 [2] NCCL INFO Channel 06/0 : 5[3] -> 6[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127292:128057 [2] NCCL INFO Channel 00/0 : 6[2] -> 7[3] via P2P/IPC
+ip-26-0-166-244:897243:898033 [3] NCCL INFO Channel 06/0 : 11[3] -> 12[2] [send] via NET/Libfabric/2(10)/GDRDMA
+ip-26-0-166-15:127293:128058 [3] NCCL INFO Channel 04/0 : 7[3] -> 8[2] [send] via NET/Libfabric/2(6)/GDRDMA
+ip-26-0-166-244:897242:898034 [2] NCCL INFO Channel 04/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-166-36:104431:105195 [2] NCCL INFO Channel 00/0 : 11[3] -> 12[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762817:763617 [2] NCCL INFO Channel 00/0 : 3[3] -> 4[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-46:702882:703682 [2] NCCL INFO Channel 02/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-166-36:104432:105194 [3] NCCL INFO Channel 00/0 : 13[3] -> 14[2] [send] via NET/Libfabric/2(12)/GDRDMA
+ip-26-0-166-125:762817:763617 [2] NCCL INFO Channel 02/0 : 3[3] -> 4[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-46:702883:703683 [3] NCCL INFO Channel 06/0 : 1[3] -> 2[2] [send] via NET/Libfabric/2(0)/GDRDMA
+ip-26-0-167-9:1039539:1040321 [2] NCCL INFO Channel 02/0 : 14[2] -> 15[3] via P2P/IPC
+ip-26-0-166-125:762818:763616 [3] NCCL INFO Channel 00/0 : 5[3] -> 6[2] [send] via NET/Libfabric/2(4)/GDRDMA
+ip-26-0-166-125:762817:763617 [2] NCCL INFO Channel 04/0 : 3[3] -> 4[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-167-9:1039540:1040320 [3] NCCL INFO Channel 06/0 : 15[3] -> 0[2] [send] via NET/Libfabric/2(14)/GDRDMA
+ip-26-0-166-125:762818:763616 [3] NCCL INFO Channel 02/0 : 5[3] -> 6[2] [send] via NET/Libfabric/2(4)/GDRDMA
+ip-26-0-166-125:762817:763617 [2] NCCL INFO Channel 06/0 : 3[3] -> 4[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762817:763617 [2] NCCL INFO Channel 00/0 : 4[2] -> 5[3] via P2P/IPC
+ip-26-0-166-125:762818:763616 [3] NCCL INFO Channel 04/0 : 5[3] -> 6[2] [send] via NET/Libfabric/2(4)/GDRDMA
+ip-26-0-167-9:1039539:1040321 [2] NCCL INFO Channel 04/0 : 14[2] -> 15[3] via P2P/IPC
+ip-26-0-162-46:702882:703682 [2] NCCL INFO Channel 04/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-166-15:127292:128057 [2] NCCL INFO Channel 02/0 : 6[2] -> 7[3] via P2P/IPC
+ip-26-0-166-36:104431:105195 [2] NCCL INFO Channel 02/0 : 11[3] -> 12[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104432:105194 [3] NCCL INFO Channel 02/0 : 13[3] -> 14[2] [send] via NET/Libfabric/2(12)/GDRDMA
+ip-26-0-166-36:104431:105195 [2] NCCL INFO Channel 04/0 : 11[3] -> 12[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104432:105194 [3] NCCL INFO Channel 04/0 : 13[3] -> 14[2] [send] via NET/Libfabric/2(12)/GDRDMA
+ip-26-0-166-36:104431:105195 [2] NCCL INFO Channel 06/0 : 11[3] -> 12[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104431:105195 [2] NCCL INFO Channel 00/0 : 12[2] -> 13[3] via P2P/IPC
+ip-26-0-166-36:104432:105194 [3] NCCL INFO Channel 06/0 : 13[3] -> 14[2] [send] via NET/Libfabric/2(12)/GDRDMA
+ip-26-0-166-244:897242:898034 [2] NCCL INFO Channel 06/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-166-15:127293:128058 [3] NCCL INFO Channel 06/0 : 7[3] -> 8[2] [send] via NET/Libfabric/2(6)/GDRDMA
+ip-26-0-167-9:1039539:1040321 [2] NCCL INFO Channel 06/0 : 14[2] -> 15[3] via P2P/IPC
+ip-26-0-162-46:702882:703682 [2] NCCL INFO Channel 06/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-166-15:127292:128057 [2] NCCL INFO Channel 04/0 : 6[2] -> 7[3] via P2P/IPC
+ip-26-0-166-15:127292:128057 [2] NCCL INFO Channel 06/0 : 6[2] -> 7[3] via P2P/IPC
+ip-26-0-166-125:762817:763617 [2] NCCL INFO Channel 02/0 : 4[2] -> 5[3] via P2P/IPC
+ip-26-0-166-125:762818:763616 [3] NCCL INFO Channel 06/0 : 5[3] -> 6[2] [send] via NET/Libfabric/2(4)/GDRDMA
+ip-26-0-166-125:762817:763617 [2] NCCL INFO Channel 04/0 : 4[2] -> 5[3] via P2P/IPC
+ip-26-0-166-125:762817:763617 [2] NCCL INFO Channel 06/0 : 4[2] -> 5[3] via P2P/IPC
+ip-26-0-165-213:9627:10393 [2] NCCL INFO Channel 00/0 : 1[3] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9627:10393 [2] NCCL INFO Channel 02/0 : 1[3] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9628:10394 [3] NCCL INFO Channel 00/0 : 3[3] -> 4[2] [send] via NET/Libfabric/2(2)/GDRDMA
+ip-26-0-165-213:9627:10393 [2] NCCL INFO Channel 04/0 : 1[3] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9628:10394 [3] NCCL INFO Channel 02/0 : 3[3] -> 4[2] [send] via NET/Libfabric/2(2)/GDRDMA
+ip-26-0-165-213:9627:10393 [2] NCCL INFO Channel 06/0 : 1[3] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9627:10393 [2] NCCL INFO Channel 00/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-213:9628:10394 [3] NCCL INFO Channel 04/0 : 3[3] -> 4[2] [send] via NET/Libfabric/2(2)/GDRDMA
+ip-26-0-167-9:1039539:1040321 [2] NCCL INFO Channel 01/0 : 14[2] -> 1[3] [send] via NET/Libfabric/3(15)/GDRDMA
+ip-26-0-166-244:897242:898034 [2] NCCL INFO Channel 01/0 : 10[2] -> 13[3] [send] via NET/Libfabric/3(11)/GDRDMA
+ip-26-0-162-46:702882:703682 [2] NCCL INFO Channel 01/0 : 0[2] -> 3[3] [send] via NET/Libfabric/3(1)/GDRDMA
+ip-26-0-162-46:702882:703682 [2] NCCL INFO Channel 03/0 : 0[2] -> 3[3] [send] via NET/Libfabric/3(1)/GDRDMA
+ip-26-0-166-15:127292:128057 [2] NCCL INFO Channel 01/0 : 6[2] -> 9[3] [send] via NET/Libfabric/3(7)/GDRDMA
+ip-26-0-165-213:9627:10393 [2] NCCL INFO Channel 02/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-162-46:702882:703682 [2] NCCL INFO Channel 05/0 : 0[2] -> 3[3] [send] via NET/Libfabric/3(1)/GDRDMA
+ip-26-0-165-213:9628:10394 [3] NCCL INFO Channel 06/0 : 3[3] -> 4[2] [send] via NET/Libfabric/2(2)/GDRDMA
+ip-26-0-162-46:702882:703682 [2] NCCL INFO Channel 07/0 : 0[2] -> 3[3] [send] via NET/Libfabric/3(1)/GDRDMA
+ip-26-0-165-213:9627:10393 [2] NCCL INFO Channel 04/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-165-213:9627:10393 [2] NCCL INFO Channel 06/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-214:624943:625737 [0] NCCL INFO Channel 00/0 : 7[1] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624943:625737 [0] NCCL INFO Channel 02/0 : 7[1] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624944:625738 [1] NCCL INFO Channel 00/0 : 9[1] -> 10[0] [send] via NET/Libfabric/0(8)/GDRDMA
+ip-26-0-166-214:624943:625737 [0] NCCL INFO Channel 04/0 : 7[1] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624944:625738 [1] NCCL INFO Channel 02/0 : 9[1] -> 10[0] [send] via NET/Libfabric/0(8)/GDRDMA
+ip-26-0-166-214:624943:625737 [0] NCCL INFO Channel 06/0 : 7[1] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624943:625737 [0] NCCL INFO Channel 00/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-166-214:624944:625738 [1] NCCL INFO Channel 04/0 : 9[1] -> 10[0] [send] via NET/Libfabric/0(8)/GDRDMA
+ip-26-0-166-214:624943:625737 [0] NCCL INFO Channel 02/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-166-214:624944:625738 [1] NCCL INFO Channel 06/0 : 9[1] -> 10[0] [send] via NET/Libfabric/0(8)/GDRDMA
+ip-26-0-166-214:624943:625737 [0] NCCL INFO Channel 04/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-166-214:624943:625737 [0] NCCL INFO Channel 06/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-166-214:624943:625737 [0] NCCL INFO Channel 01/0 : 8[0] -> 11[1] [send] via NET/Libfabric/1(9)/GDRDMA
+ip-26-0-166-214:624947:625739 [4] NCCL INFO Channel 00/0 : 7[5] -> 8[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624948:625740 [5] NCCL INFO Channel 00/0 : 9[5] -> 10[4] [send] via NET/Libfabric/4(8)/GDRDMA
+ip-26-0-166-214:624947:625739 [4] NCCL INFO Channel 02/0 : 7[5] -> 8[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624948:625740 [5] NCCL INFO Channel 02/0 : 9[5] -> 10[4] [send] via NET/Libfabric/4(8)/GDRDMA
+ip-26-0-166-214:624947:625739 [4] NCCL INFO Channel 04/0 : 7[5] -> 8[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624948:625740 [5] NCCL INFO Channel 04/0 : 9[5] -> 10[4] [send] via NET/Libfabric/4(8)/GDRDMA
+ip-26-0-166-214:624947:625739 [4] NCCL INFO Channel 06/0 : 7[5] -> 8[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624947:625739 [4] NCCL INFO Channel 00/0 : 8[4] -> 9[5] via P2P/IPC
+ip-26-0-166-214:624948:625740 [5] NCCL INFO Channel 06/0 : 9[5] -> 10[4] [send] via NET/Libfabric/4(8)/GDRDMA
+ip-26-0-166-214:624947:625739 [4] NCCL INFO Channel 02/0 : 8[4] -> 9[5] via P2P/IPC
+ip-26-0-166-214:624947:625739 [4] NCCL INFO Channel 04/0 : 8[4] -> 9[5] via P2P/IPC
+ip-26-0-166-214:624947:625739 [4] NCCL INFO Channel 06/0 : 8[4] -> 9[5] via P2P/IPC
+ip-26-0-166-214:624949:625744 [6] NCCL INFO Channel 00/0 : 7[7] -> 8[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624950:625743 [7] NCCL INFO Channel 00/0 : 9[7] -> 10[6] [send] via NET/Libfabric/6(8)/GDRDMA
+ip-26-0-166-214:624949:625744 [6] NCCL INFO Channel 02/0 : 7[7] -> 8[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624950:625743 [7] NCCL INFO Channel 02/0 : 9[7] -> 10[6] [send] via NET/Libfabric/6(8)/GDRDMA
+ip-26-0-166-214:624949:625744 [6] NCCL INFO Channel 04/0 : 7[7] -> 8[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624950:625743 [7] NCCL INFO Channel 04/0 : 9[7] -> 10[6] [send] via NET/Libfabric/6(8)/GDRDMA
+ip-26-0-166-214:624949:625744 [6] NCCL INFO Channel 06/0 : 7[7] -> 8[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624949:625744 [6] NCCL INFO Channel 00/0 : 8[6] -> 9[7] via P2P/IPC
+ip-26-0-166-214:624950:625743 [7] NCCL INFO Channel 06/0 : 9[7] -> 10[6] [send] via NET/Libfabric/6(8)/GDRDMA
+ip-26-0-166-214:624949:625744 [6] NCCL INFO Channel 02/0 : 8[6] -> 9[7] via P2P/IPC
+ip-26-0-166-214:624949:625744 [6] NCCL INFO Channel 04/0 : 8[6] -> 9[7] via P2P/IPC
+ip-26-0-166-214:624949:625744 [6] NCCL INFO Channel 06/0 : 8[6] -> 9[7] via P2P/IPC
+ip-26-0-166-214:624945:625741 [2] NCCL INFO Channel 00/0 : 7[3] -> 8[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624945:625741 [2] NCCL INFO Channel 02/0 : 7[3] -> 8[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624946:625742 [3] NCCL INFO Channel 00/0 : 9[3] -> 10[2] [send] via NET/Libfabric/2(8)/GDRDMA
+ip-26-0-166-214:624945:625741 [2] NCCL INFO Channel 04/0 : 7[3] -> 8[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624946:625742 [3] NCCL INFO Channel 02/0 : 9[3] -> 10[2] [send] via NET/Libfabric/2(8)/GDRDMA
+ip-26-0-166-214:624945:625741 [2] NCCL INFO Channel 06/0 : 7[3] -> 8[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624945:625741 [2] NCCL INFO Channel 00/0 : 8[2] -> 9[3] via P2P/IPC
+ip-26-0-166-214:624946:625742 [3] NCCL INFO Channel 04/0 : 9[3] -> 10[2] [send] via NET/Libfabric/2(8)/GDRDMA
+ip-26-0-166-214:624945:625741 [2] NCCL INFO Channel 02/0 : 8[2] -> 9[3] via P2P/IPC
+ip-26-0-166-214:624946:625742 [3] NCCL INFO Channel 06/0 : 9[3] -> 10[2] [send] via NET/Libfabric/2(8)/GDRDMA
+ip-26-0-166-214:624945:625741 [2] NCCL INFO Channel 04/0 : 8[2] -> 9[3] via P2P/IPC
+ip-26-0-166-214:624945:625741 [2] NCCL INFO Channel 06/0 : 8[2] -> 9[3] via P2P/IPC
+ip-26-0-166-36:104431:105195 [2] NCCL INFO Channel 02/0 : 12[2] -> 13[3] via P2P/IPC
+ip-26-0-166-36:104431:105195 [2] NCCL INFO Channel 04/0 : 12[2] -> 13[3] via P2P/IPC
+ip-26-0-166-36:104431:105195 [2] NCCL INFO Channel 06/0 : 12[2] -> 13[3] via P2P/IPC
+ip-26-0-166-36:104430:105190 [1] NCCL INFO Channel 01/0 : 10[0] -> 13[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104429:105191 [0] NCCL INFO Channel 03/0 : 12[0] -> 15[1] [send] via NET/Libfabric/1(13)/GDRDMA
+ip-26-0-166-36:104430:105190 [1] NCCL INFO Channel 03/0 : 10[0] -> 13[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104429:105191 [0] NCCL INFO Channel 05/0 : 12[0] -> 15[1] [send] via NET/Libfabric/1(13)/GDRDMA
+ip-26-0-166-36:104430:105190 [1] NCCL INFO Channel 05/0 : 10[0] -> 13[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104429:105191 [0] NCCL INFO Channel 07/0 : 12[0] -> 15[1] [send] via NET/Libfabric/1(13)/GDRDMA
+ip-26-0-166-214:624944:625738 [1] NCCL INFO Channel 01/0 : 6[0] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104430:105190 [1] NCCL INFO Channel 07/0 : 10[0] -> 13[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624943:625737 [0] NCCL INFO Channel 03/0 : 8[0] -> 11[1] [send] via NET/Libfabric/1(9)/GDRDMA
+ip-26-0-166-214:624944:625738 [1] NCCL INFO Channel 03/0 : 6[0] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624943:625737 [0] NCCL INFO Channel 05/0 : 8[0] -> 11[1] [send] via NET/Libfabric/1(9)/GDRDMA
+ip-26-0-166-214:624944:625738 [1] NCCL INFO Channel 05/0 : 6[0] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624943:625737 [0] NCCL INFO Channel 07/0 : 8[0] -> 11[1] [send] via NET/Libfabric/1(9)/GDRDMA
+ip-26-0-166-214:624944:625738 [1] NCCL INFO Channel 07/0 : 6[0] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897241:898029 [1] NCCL INFO Channel 01/0 : 8[0] -> 11[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897240:898030 [0] NCCL INFO Channel 01/0 : 10[0] -> 13[1] [send] via NET/Libfabric/1(11)/GDRDMA
+ip-26-0-166-244:897241:898029 [1] NCCL INFO Channel 03/0 : 8[0] -> 11[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897240:898030 [0] NCCL INFO Channel 03/0 : 10[0] -> 13[1] [send] via NET/Libfabric/1(11)/GDRDMA
+ip-26-0-166-244:897241:898029 [1] NCCL INFO Channel 05/0 : 8[0] -> 11[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897240:898030 [0] NCCL INFO Channel 05/0 : 10[0] -> 13[1] [send] via NET/Libfabric/1(11)/GDRDMA
+ip-26-0-166-244:897241:898029 [1] NCCL INFO Channel 07/0 : 8[0] -> 11[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897240:898030 [0] NCCL INFO Channel 07/0 : 10[0] -> 13[1] [send] via NET/Libfabric/1(11)/GDRDMA
+ip-26-0-166-244:897241:898029 [1] NCCL INFO Channel 01/0 : 11[1] -> 10[0] via P2P/IPC
+ip-26-0-166-125:762816:763612 [1] NCCL INFO Channel 01/0 : 2[0] -> 5[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104430:105190 [1] NCCL INFO Channel 01/0 : 13[1] -> 12[0] via P2P/IPC
+ip-26-0-166-125:762815:763613 [0] NCCL INFO Channel 07/0 : 4[0] -> 7[1] [send] via NET/Libfabric/1(5)/GDRDMA
+ip-26-0-166-125:762816:763612 [1] NCCL INFO Channel 03/0 : 2[0] -> 5[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762816:763612 [1] NCCL INFO Channel 05/0 : 2[0] -> 5[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762816:763612 [1] NCCL INFO Channel 07/0 : 2[0] -> 5[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104430:105190 [1] NCCL INFO Channel 03/0 : 13[1] -> 12[0] via P2P/IPC
+ip-26-0-166-244:897241:898029 [1] NCCL INFO Channel 03/0 : 11[1] -> 10[0] via P2P/IPC
+ip-26-0-166-36:104430:105190 [1] NCCL INFO Channel 05/0 : 13[1] -> 12[0] via P2P/IPC
+ip-26-0-166-36:104430:105190 [1] NCCL INFO Channel 07/0 : 13[1] -> 12[0] via P2P/IPC
+ip-26-0-165-213:9626:10389 [1] NCCL INFO Channel 01/0 : 0[0] -> 3[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9626:10389 [1] NCCL INFO Channel 03/0 : 0[0] -> 3[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9625:10390 [0] NCCL INFO Channel 01/0 : 2[0] -> 5[1] [send] via NET/Libfabric/1(3)/GDRDMA
+ip-26-0-165-213:9626:10389 [1] NCCL INFO Channel 05/0 : 0[0] -> 3[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9625:10390 [0] NCCL INFO Channel 03/0 : 2[0] -> 5[1] [send] via NET/Libfabric/1(3)/GDRDMA
+ip-26-0-165-213:9626:10389 [1] NCCL INFO Channel 07/0 : 0[0] -> 3[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9625:10390 [0] NCCL INFO Channel 05/0 : 2[0] -> 5[1] [send] via NET/Libfabric/1(3)/GDRDMA
+ip-26-0-165-213:9625:10390 [0] NCCL INFO Channel 07/0 : 2[0] -> 5[1] [send] via NET/Libfabric/1(3)/GDRDMA
+ip-26-0-166-15:127291:128053 [1] NCCL INFO Channel 01/0 : 4[0] -> 7[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127290:128054 [0] NCCL INFO Channel 01/0 : 6[0] -> 9[1] [send] via NET/Libfabric/1(7)/GDRDMA
+ip-26-0-166-15:127291:128053 [1] NCCL INFO Channel 03/0 : 4[0] -> 7[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127290:128054 [0] NCCL INFO Channel 03/0 : 6[0] -> 9[1] [send] via NET/Libfabric/1(7)/GDRDMA
+ip-26-0-166-15:127291:128053 [1] NCCL INFO Channel 05/0 : 4[0] -> 7[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-167-9:1039538:1040317 [1] NCCL INFO Channel 01/0 : 12[0] -> 15[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762816:763612 [1] NCCL INFO Channel 01/0 : 5[1] -> 4[0] via P2P/IPC
+ip-26-0-166-15:127290:128054 [0] NCCL INFO Channel 05/0 : 6[0] -> 9[1] [send] via NET/Libfabric/1(7)/GDRDMA
+ip-26-0-167-9:1039538:1040317 [1] NCCL INFO Channel 03/0 : 12[0] -> 15[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127291:128053 [1] NCCL INFO Channel 07/0 : 4[0] -> 7[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-167-9:1039538:1040317 [1] NCCL INFO Channel 05/0 : 12[0] -> 15[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127290:128054 [0] NCCL INFO Channel 07/0 : 6[0] -> 9[1] [send] via NET/Libfabric/1(7)/GDRDMA
+ip-26-0-166-15:127291:128053 [1] NCCL INFO Channel 01/0 : 7[1] -> 6[0] via P2P/IPC
+ip-26-0-167-9:1039538:1040317 [1] NCCL INFO Channel 07/0 : 12[0] -> 15[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624944:625738 [1] NCCL INFO Channel 01/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-166-125:762816:763612 [1] NCCL INFO Channel 03/0 : 5[1] -> 4[0] via P2P/IPC
+ip-26-0-167-9:1039538:1040317 [1] NCCL INFO Channel 01/0 : 15[1] -> 14[0] via P2P/IPC
+ip-26-0-166-244:897241:898029 [1] NCCL INFO Channel 05/0 : 11[1] -> 10[0] via P2P/IPC
+ip-26-0-162-46:702881:703677 [1] NCCL INFO Channel 01/0 : 14[0] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-46:702880:703676 [0] NCCL INFO Channel 01/0 : 0[0] -> 3[1] [send] via NET/Libfabric/1(1)/GDRDMA
+ip-26-0-166-214:624944:625738 [1] NCCL INFO Channel 03/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-162-46:702881:703677 [1] NCCL INFO Channel 03/0 : 14[0] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762816:763612 [1] NCCL INFO Channel 05/0 : 5[1] -> 4[0] via P2P/IPC
+ip-26-0-166-15:127291:128053 [1] NCCL INFO Channel 03/0 : 7[1] -> 6[0] via P2P/IPC
+ip-26-0-162-46:702880:703676 [0] NCCL INFO Channel 03/0 : 0[0] -> 3[1] [send] via NET/Libfabric/1(1)/GDRDMA
+ip-26-0-162-46:702881:703677 [1] NCCL INFO Channel 05/0 : 14[0] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-46:702880:703676 [0] NCCL INFO Channel 05/0 : 0[0] -> 3[1] [send] via NET/Libfabric/1(1)/GDRDMA
+ip-26-0-162-46:702881:703677 [1] NCCL INFO Channel 07/0 : 14[0] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624944:625738 [1] NCCL INFO Channel 05/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-167-9:1039538:1040317 [1] NCCL INFO Channel 03/0 : 15[1] -> 14[0] via P2P/IPC
+ip-26-0-162-46:702880:703676 [0] NCCL INFO Channel 07/0 : 0[0] -> 3[1] [send] via NET/Libfabric/1(1)/GDRDMA
+ip-26-0-166-125:762816:763612 [1] NCCL INFO Channel 07/0 : 5[1] -> 4[0] via P2P/IPC
+ip-26-0-162-46:702881:703677 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127291:128053 [1] NCCL INFO Channel 05/0 : 7[1] -> 6[0] via P2P/IPC
+ip-26-0-167-9:1039538:1040317 [1] NCCL INFO Channel 05/0 : 15[1] -> 14[0] via P2P/IPC
+ip-26-0-166-214:624944:625738 [1] NCCL INFO Channel 07/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-165-213:9626:10389 [1] NCCL INFO Channel 01/0 : 3[1] -> 2[0] via P2P/IPC
+ip-26-0-167-9:1039538:1040317 [1] NCCL INFO Channel 07/0 : 15[1] -> 14[0] via P2P/IPC
+ip-26-0-166-15:127291:128053 [1] NCCL INFO Channel 07/0 : 7[1] -> 6[0] via P2P/IPC
+ip-26-0-165-213:9626:10389 [1] NCCL INFO Channel 03/0 : 3[1] -> 2[0] via P2P/IPC
+ip-26-0-165-213:9626:10389 [1] NCCL INFO Channel 05/0 : 3[1] -> 2[0] via P2P/IPC
+ip-26-0-165-213:9626:10389 [1] NCCL INFO Channel 07/0 : 3[1] -> 2[0] via P2P/IPC
+ip-26-0-166-244:897241:898029 [1] NCCL INFO Channel 07/0 : 11[1] -> 10[0] via P2P/IPC
+ip-26-0-162-46:702881:703677 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-162-46:702881:703677 [1] NCCL INFO Channel 05/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-162-46:702881:703677 [1] NCCL INFO Channel 07/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-36:104434:105192 [5] NCCL INFO Channel 01/0 : 10[4] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104433:105193 [4] NCCL INFO Channel 07/0 : 12[4] -> 15[5] [send] via NET/Libfabric/5(13)/GDRDMA
+ip-26-0-166-36:104434:105192 [5] NCCL INFO Channel 03/0 : 10[4] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104434:105192 [5] NCCL INFO Channel 05/0 : 10[4] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104434:105192 [5] NCCL INFO Channel 07/0 : 10[4] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897245:898031 [5] NCCL INFO Channel 01/0 : 8[4] -> 11[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897244:898032 [4] NCCL INFO Channel 01/0 : 10[4] -> 13[5] [send] via NET/Libfabric/5(11)/GDRDMA
+ip-26-0-166-244:897245:898031 [5] NCCL INFO Channel 03/0 : 8[4] -> 11[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897244:898032 [4] NCCL INFO Channel 03/0 : 10[4] -> 13[5] [send] via NET/Libfabric/5(11)/GDRDMA
+ip-26-0-166-244:897245:898031 [5] NCCL INFO Channel 05/0 : 8[4] -> 11[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897244:898032 [4] NCCL INFO Channel 05/0 : 10[4] -> 13[5] [send] via NET/Libfabric/5(11)/GDRDMA
+ip-26-0-166-244:897245:898031 [5] NCCL INFO Channel 07/0 : 8[4] -> 11[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897244:898032 [4] NCCL INFO Channel 07/0 : 10[4] -> 13[5] [send] via NET/Libfabric/5(11)/GDRDMA
+ip-26-0-166-36:104434:105192 [5] NCCL INFO Channel 01/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-166-214:624948:625740 [5] NCCL INFO Channel 01/0 : 6[4] -> 9[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624947:625739 [4] NCCL INFO Channel 01/0 : 8[4] -> 11[5] [send] via NET/Libfabric/5(9)/GDRDMA
+ip-26-0-166-125:762820:763615 [5] NCCL INFO Channel 01/0 : 2[4] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624948:625740 [5] NCCL INFO Channel 03/0 : 6[4] -> 9[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624948:625740 [5] NCCL INFO Channel 05/0 : 6[4] -> 9[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762819:763614 [4] NCCL INFO Channel 03/0 : 4[4] -> 7[5] [send] via NET/Libfabric/5(5)/GDRDMA
+ip-26-0-166-214:624947:625739 [4] NCCL INFO Channel 03/0 : 8[4] -> 11[5] [send] via NET/Libfabric/5(9)/GDRDMA
+ip-26-0-166-125:762820:763615 [5] NCCL INFO Channel 03/0 : 2[4] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624948:625740 [5] NCCL INFO Channel 07/0 : 6[4] -> 9[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762819:763614 [4] NCCL INFO Channel 05/0 : 4[4] -> 7[5] [send] via NET/Libfabric/5(5)/GDRDMA
+ip-26-0-166-214:624947:625739 [4] NCCL INFO Channel 05/0 : 8[4] -> 11[5] [send] via NET/Libfabric/5(9)/GDRDMA
+ip-26-0-166-125:762820:763615 [5] NCCL INFO Channel 05/0 : 2[4] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762819:763614 [4] NCCL INFO Channel 07/0 : 4[4] -> 7[5] [send] via NET/Libfabric/5(5)/GDRDMA
+ip-26-0-166-214:624947:625739 [4] NCCL INFO Channel 07/0 : 8[4] -> 11[5] [send] via NET/Libfabric/5(9)/GDRDMA
+ip-26-0-166-125:762820:763615 [5] NCCL INFO Channel 07/0 : 2[4] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897245:898031 [5] NCCL INFO Channel 01/0 : 11[5] -> 10[4] via P2P/IPC
+ip-26-0-162-46:702885:703680 [5] NCCL INFO Channel 01/0 : 14[4] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-46:702884:703679 [4] NCCL INFO Channel 03/0 : 0[4] -> 3[5] [send] via NET/Libfabric/5(1)/GDRDMA
+ip-26-0-162-46:702885:703680 [5] NCCL INFO Channel 03/0 : 14[4] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-46:702884:703679 [4] NCCL INFO Channel 05/0 : 0[4] -> 3[5] [send] via NET/Libfabric/5(1)/GDRDMA
+ip-26-0-162-46:702885:703680 [5] NCCL INFO Channel 05/0 : 14[4] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127295:128055 [5] NCCL INFO Channel 01/0 : 4[4] -> 7[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-46:702884:703679 [4] NCCL INFO Channel 07/0 : 0[4] -> 3[5] [send] via NET/Libfabric/5(1)/GDRDMA
+ip-26-0-162-46:702885:703680 [5] NCCL INFO Channel 07/0 : 14[4] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127294:128056 [4] NCCL INFO Channel 01/0 : 6[4] -> 9[5] [send] via NET/Libfabric/5(7)/GDRDMA
+ip-26-0-166-15:127295:128055 [5] NCCL INFO Channel 03/0 : 4[4] -> 7[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127294:128056 [4] NCCL INFO Channel 03/0 : 6[4] -> 9[5] [send] via NET/Libfabric/5(7)/GDRDMA
+ip-26-0-166-15:127295:128055 [5] NCCL INFO Channel 05/0 : 4[4] -> 7[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127294:128056 [4] NCCL INFO Channel 05/0 : 6[4] -> 9[5] [send] via NET/Libfabric/5(7)/GDRDMA
+ip-26-0-166-15:127295:128055 [5] NCCL INFO Channel 07/0 : 4[4] -> 7[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127294:128056 [4] NCCL INFO Channel 07/0 : 6[4] -> 9[5] [send] via NET/Libfabric/5(7)/GDRDMA
+ip-26-0-166-15:127295:128055 [5] NCCL INFO Channel 01/0 : 7[5] -> 6[4] via P2P/IPC
+ip-26-0-166-214:624948:625740 [5] NCCL INFO Channel 01/0 : 9[5] -> 8[4] via P2P/IPC
+ip-26-0-167-9:1039542:1040319 [5] NCCL INFO Channel 01/0 : 12[4] -> 15[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9630:10392 [5] NCCL INFO Channel 01/0 : 0[4] -> 3[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-167-9:1039541:1040318 [4] NCCL INFO Channel 03/0 : 14[4] -> 1[5] [send] via NET/Libfabric/5(15)/GDRDMA
+ip-26-0-165-213:9629:10391 [4] NCCL INFO Channel 01/0 : 2[4] -> 5[5] [send] via NET/Libfabric/5(3)/GDRDMA
+ip-26-0-167-9:1039542:1040319 [5] NCCL INFO Channel 03/0 : 12[4] -> 15[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-167-9:1039541:1040318 [4] NCCL INFO Channel 05/0 : 14[4] -> 1[5] [send] via NET/Libfabric/5(15)/GDRDMA
+ip-26-0-165-213:9630:10392 [5] NCCL INFO Channel 03/0 : 0[4] -> 3[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-167-9:1039542:1040319 [5] NCCL INFO Channel 05/0 : 12[4] -> 15[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9629:10391 [4] NCCL INFO Channel 03/0 : 2[4] -> 5[5] [send] via NET/Libfabric/5(3)/GDRDMA
+ip-26-0-167-9:1039541:1040318 [4] NCCL INFO Channel 07/0 : 14[4] -> 1[5] [send] via NET/Libfabric/5(15)/GDRDMA
+ip-26-0-165-213:9630:10392 [5] NCCL INFO Channel 05/0 : 0[4] -> 3[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-167-9:1039542:1040319 [5] NCCL INFO Channel 07/0 : 12[4] -> 15[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9629:10391 [4] NCCL INFO Channel 05/0 : 2[4] -> 5[5] [send] via NET/Libfabric/5(3)/GDRDMA
+ip-26-0-165-213:9630:10392 [5] NCCL INFO Channel 07/0 : 0[4] -> 3[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-46:702885:703680 [5] NCCL INFO Channel 01/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-167-9:1039542:1040319 [5] NCCL INFO Channel 01/0 : 15[5] -> 14[4] via P2P/IPC
+ip-26-0-165-213:9629:10391 [4] NCCL INFO Channel 07/0 : 2[4] -> 5[5] [send] via NET/Libfabric/5(3)/GDRDMA
+ip-26-0-165-213:9630:10392 [5] NCCL INFO Channel 01/0 : 3[5] -> 2[4] via P2P/IPC
+ip-26-0-166-125:762820:763615 [5] NCCL INFO Channel 01/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-36:104434:105192 [5] NCCL INFO Channel 03/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-166-244:897245:898031 [5] NCCL INFO Channel 03/0 : 11[5] -> 10[4] via P2P/IPC
+ip-26-0-166-15:127295:128055 [5] NCCL INFO Channel 03/0 : 7[5] -> 6[4] via P2P/IPC
+ip-26-0-166-214:624948:625740 [5] NCCL INFO Channel 03/0 : 9[5] -> 8[4] via P2P/IPC
+ip-26-0-166-15:127297:128060 [7] NCCL INFO Channel 01/0 : 4[6] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127296:128059 [6] NCCL INFO Channel 03/0 : 6[6] -> 9[7] [send] via NET/Libfabric/7(7)/GDRDMA
+ip-26-0-166-15:127297:128060 [7] NCCL INFO Channel 03/0 : 4[6] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127296:128059 [6] NCCL INFO Channel 05/0 : 6[6] -> 9[7] [send] via NET/Libfabric/7(7)/GDRDMA
+ip-26-0-166-15:127297:128060 [7] NCCL INFO Channel 05/0 : 4[6] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9630:10392 [5] NCCL INFO Channel 03/0 : 3[5] -> 2[4] via P2P/IPC
+ip-26-0-166-15:127296:128059 [6] NCCL INFO Channel 07/0 : 6[6] -> 9[7] [send] via NET/Libfabric/7(7)/GDRDMA
+ip-26-0-166-15:127297:128060 [7] NCCL INFO Channel 07/0 : 4[6] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624950:625743 [7] NCCL INFO Channel 01/0 : 6[6] -> 9[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624949:625744 [6] NCCL INFO Channel 01/0 : 8[6] -> 11[7] [send] via NET/Libfabric/7(9)/GDRDMA
+ip-26-0-166-244:897247:898035 [7] NCCL INFO Channel 01/0 : 8[6] -> 11[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624950:625743 [7] NCCL INFO Channel 03/0 : 6[6] -> 9[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624949:625744 [6] NCCL INFO Channel 03/0 : 8[6] -> 11[7] [send] via NET/Libfabric/7(9)/GDRDMA
+ip-26-0-166-214:624950:625743 [7] NCCL INFO Channel 05/0 : 6[6] -> 9[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897246:898036 [6] NCCL INFO Channel 03/0 : 10[6] -> 13[7] [send] via NET/Libfabric/7(11)/GDRDMA
+ip-26-0-166-214:624949:625744 [6] NCCL INFO Channel 05/0 : 8[6] -> 11[7] [send] via NET/Libfabric/7(9)/GDRDMA
+ip-26-0-166-244:897247:898035 [7] NCCL INFO Channel 03/0 : 8[6] -> 11[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624950:625743 [7] NCCL INFO Channel 07/0 : 6[6] -> 9[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104436:105197 [7] NCCL INFO Channel 01/0 : 10[6] -> 13[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897246:898036 [6] NCCL INFO Channel 05/0 : 10[6] -> 13[7] [send] via NET/Libfabric/7(11)/GDRDMA
+ip-26-0-166-244:897247:898035 [7] NCCL INFO Channel 05/0 : 8[6] -> 11[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624949:625744 [6] NCCL INFO Channel 07/0 : 8[6] -> 11[7] [send] via NET/Libfabric/7(9)/GDRDMA
+ip-26-0-166-244:897246:898036 [6] NCCL INFO Channel 07/0 : 10[6] -> 13[7] [send] via NET/Libfabric/7(11)/GDRDMA
+ip-26-0-166-36:104435:105196 [6] NCCL INFO Channel 03/0 : 12[6] -> 15[7] [send] via NET/Libfabric/7(13)/GDRDMA
+ip-26-0-166-244:897247:898035 [7] NCCL INFO Channel 07/0 : 8[6] -> 11[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624950:625743 [7] NCCL INFO Channel 01/0 : 9[7] -> 8[6] via P2P/IPC
+ip-26-0-166-36:104436:105197 [7] NCCL INFO Channel 03/0 : 10[6] -> 13[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-46:702887:703686 [7] NCCL INFO Channel 01/0 : 14[6] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104435:105196 [6] NCCL INFO Channel 05/0 : 12[6] -> 15[7] [send] via NET/Libfabric/7(13)/GDRDMA
+ip-26-0-166-36:104436:105197 [7] NCCL INFO Channel 05/0 : 10[6] -> 13[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-46:702887:703686 [7] NCCL INFO Channel 03/0 : 14[6] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897247:898035 [7] NCCL INFO Channel 01/0 : 11[7] -> 10[6] via P2P/IPC
+ip-26-0-166-36:104435:105196 [6] NCCL INFO Channel 07/0 : 12[6] -> 15[7] [send] via NET/Libfabric/7(13)/GDRDMA
+ip-26-0-162-46:702887:703686 [7] NCCL INFO Channel 05/0 : 14[6] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104436:105197 [7] NCCL INFO Channel 07/0 : 10[6] -> 13[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-46:702887:703686 [7] NCCL INFO Channel 07/0 : 14[6] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762822:763618 [7] NCCL INFO Channel 01/0 : 2[6] -> 5[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104436:105197 [7] NCCL INFO Channel 01/0 : 13[7] -> 12[6] via P2P/IPC
+ip-26-0-166-125:762821:763619 [6] NCCL INFO Channel 01/0 : 4[6] -> 7[7] [send] via NET/Libfabric/7(5)/GDRDMA
+ip-26-0-166-125:762822:763618 [7] NCCL INFO Channel 03/0 : 2[6] -> 5[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762821:763619 [6] NCCL INFO Channel 03/0 : 4[6] -> 7[7] [send] via NET/Libfabric/7(5)/GDRDMA
+ip-26-0-166-125:762822:763618 [7] NCCL INFO Channel 05/0 : 2[6] -> 5[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104434:105192 [5] NCCL INFO Channel 05/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-166-125:762821:763619 [6] NCCL INFO Channel 05/0 : 4[6] -> 7[7] [send] via NET/Libfabric/7(5)/GDRDMA
+ip-26-0-166-125:762820:763615 [5] NCCL INFO Channel 03/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-125:762822:763618 [7] NCCL INFO Channel 07/0 : 2[6] -> 5[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762821:763619 [6] NCCL INFO Channel 07/0 : 4[6] -> 7[7] [send] via NET/Libfabric/7(5)/GDRDMA
+ip-26-0-166-15:127297:128060 [7] NCCL INFO Channel 01/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-167-9:1039542:1040319 [5] NCCL INFO Channel 03/0 : 15[5] -> 14[4] via P2P/IPC
+ip-26-0-166-244:897245:898031 [5] NCCL INFO Channel 05/0 : 11[5] -> 10[4] via P2P/IPC
+ip-26-0-166-244:897243:898033 [3] NCCL INFO Channel 01/0 : 8[2] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-167-9:1039544:1040322 [7] NCCL INFO Channel 01/0 : 12[6] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-167-9:1039543:1040323 [6] NCCL INFO Channel 01/0 : 14[6] -> 1[7] [send] via NET/Libfabric/7(15)/GDRDMA
+ip-26-0-166-244:897242:898034 [2] NCCL INFO Channel 03/0 : 10[2] -> 13[3] [send] via NET/Libfabric/3(11)/GDRDMA
+ip-26-0-167-9:1039544:1040322 [7] NCCL INFO Channel 03/0 : 12[6] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897243:898033 [3] NCCL INFO Channel 03/0 : 8[2] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-167-9:1039543:1040323 [6] NCCL INFO Channel 03/0 : 14[6] -> 1[7] [send] via NET/Libfabric/7(15)/GDRDMA
+ip-26-0-166-244:897242:898034 [2] NCCL INFO Channel 05/0 : 10[2] -> 13[3] [send] via NET/Libfabric/3(11)/GDRDMA
+ip-26-0-167-9:1039544:1040322 [7] NCCL INFO Channel 05/0 : 12[6] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897243:898033 [3] NCCL INFO Channel 05/0 : 8[2] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-167-9:1039543:1040323 [6] NCCL INFO Channel 05/0 : 14[6] -> 1[7] [send] via NET/Libfabric/7(15)/GDRDMA
+ip-26-0-167-9:1039544:1040322 [7] NCCL INFO Channel 07/0 : 12[6] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897242:898034 [2] NCCL INFO Channel 07/0 : 10[2] -> 13[3] [send] via NET/Libfabric/3(11)/GDRDMA
+ip-26-0-166-244:897243:898033 [3] NCCL INFO Channel 07/0 : 8[2] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-167-9:1039543:1040323 [6] NCCL INFO Channel 07/0 : 14[6] -> 1[7] [send] via NET/Libfabric/7(15)/GDRDMA
+ip-26-0-167-9:1039544:1040322 [7] NCCL INFO Channel 01/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-162-46:702887:703686 [7] NCCL INFO Channel 01/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-166-36:104436:105197 [7] NCCL INFO Channel 03/0 : 13[7] -> 12[6] via P2P/IPC
+ip-26-0-166-15:127293:128058 [3] NCCL INFO Channel 01/0 : 4[2] -> 7[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127292:128057 [2] NCCL INFO Channel 03/0 : 6[2] -> 9[3] [send] via NET/Libfabric/3(7)/GDRDMA
+ip-26-0-166-15:127293:128058 [3] NCCL INFO Channel 03/0 : 4[2] -> 7[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-167-9:1039540:1040320 [3] NCCL INFO Channel 01/0 : 12[2] -> 15[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127292:128057 [2] NCCL INFO Channel 05/0 : 6[2] -> 9[3] [send] via NET/Libfabric/3(7)/GDRDMA
+ip-26-0-166-15:127293:128058 [3] NCCL INFO Channel 05/0 : 4[2] -> 7[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127292:128057 [2] NCCL INFO Channel 07/0 : 6[2] -> 9[3] [send] via NET/Libfabric/3(7)/GDRDMA
+ip-26-0-167-9:1039539:1040321 [2] NCCL INFO Channel 03/0 : 14[2] -> 1[3] [send] via NET/Libfabric/3(15)/GDRDMA
+ip-26-0-166-15:127293:128058 [3] NCCL INFO Channel 07/0 : 4[2] -> 7[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-167-9:1039540:1040320 [3] NCCL INFO Channel 03/0 : 12[2] -> 15[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-167-9:1039539:1040321 [2] NCCL INFO Channel 05/0 : 14[2] -> 1[3] [send] via NET/Libfabric/3(15)/GDRDMA
+ip-26-0-167-9:1039540:1040320 [3] NCCL INFO Channel 05/0 : 12[2] -> 15[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-167-9:1039539:1040321 [2] NCCL INFO Channel 07/0 : 14[2] -> 1[3] [send] via NET/Libfabric/3(15)/GDRDMA
+ip-26-0-167-9:1039540:1040320 [3] NCCL INFO Channel 07/0 : 12[2] -> 15[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127297:128060 [7] NCCL INFO Channel 03/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-166-15:127295:128055 [5] NCCL INFO Channel 05/0 : 7[5] -> 6[4] via P2P/IPC
+ip-26-0-166-244:897247:898035 [7] NCCL INFO Channel 03/0 : 11[7] -> 10[6] via P2P/IPC
+ip-26-0-166-214:624946:625742 [3] NCCL INFO Channel 01/0 : 6[2] -> 9[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624945:625741 [2] NCCL INFO Channel 01/0 : 8[2] -> 11[3] [send] via NET/Libfabric/3(9)/GDRDMA
+ip-26-0-166-214:624946:625742 [3] NCCL INFO Channel 03/0 : 6[2] -> 9[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624945:625741 [2] NCCL INFO Channel 03/0 : 8[2] -> 11[3] [send] via NET/Libfabric/3(9)/GDRDMA
+ip-26-0-166-214:624946:625742 [3] NCCL INFO Channel 05/0 : 6[2] -> 9[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624945:625741 [2] NCCL INFO Channel 05/0 : 8[2] -> 11[3] [send] via NET/Libfabric/3(9)/GDRDMA
+ip-26-0-166-214:624946:625742 [3] NCCL INFO Channel 07/0 : 6[2] -> 9[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624945:625741 [2] NCCL INFO Channel 07/0 : 8[2] -> 11[3] [send] via NET/Libfabric/3(9)/GDRDMA
+ip-26-0-166-214:624948:625740 [5] NCCL INFO Channel 05/0 : 9[5] -> 8[4] via P2P/IPC
+ip-26-0-166-214:624950:625743 [7] NCCL INFO Channel 03/0 : 9[7] -> 8[6] via P2P/IPC
+ip-26-0-166-214:624946:625742 [3] NCCL INFO Channel 01/0 : 9[3] -> 8[2] via P2P/IPC
+ip-26-0-166-244:897243:898033 [3] NCCL INFO Channel 01/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-166-15:127297:128060 [7] NCCL INFO Channel 05/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-165-213:9632:10395 [7] NCCL INFO Channel 01/0 : 0[6] -> 3[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127295:128055 [5] NCCL INFO Channel 07/0 : 7[5] -> 6[4] via P2P/IPC
+ip-26-0-165-213:9631:10396 [6] NCCL INFO Channel 01/0 : 2[6] -> 5[7] [send] via NET/Libfabric/7(3)/GDRDMA
+ip-26-0-165-213:9630:10392 [5] NCCL INFO Channel 05/0 : 3[5] -> 2[4] via P2P/IPC
+ip-26-0-165-213:9632:10395 [7] NCCL INFO Channel 03/0 : 0[6] -> 3[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9631:10396 [6] NCCL INFO Channel 03/0 : 2[6] -> 5[7] [send] via NET/Libfabric/7(3)/GDRDMA
+ip-26-0-165-213:9632:10395 [7] NCCL INFO Channel 05/0 : 0[6] -> 3[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-46:702885:703680 [5] NCCL INFO Channel 03/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-166-125:762820:763615 [5] NCCL INFO Channel 05/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-213:9631:10396 [6] NCCL INFO Channel 05/0 : 2[6] -> 5[7] [send] via NET/Libfabric/7(3)/GDRDMA
+ip-26-0-165-213:9632:10395 [7] NCCL INFO Channel 07/0 : 0[6] -> 3[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9631:10396 [6] NCCL INFO Channel 07/0 : 2[6] -> 5[7] [send] via NET/Libfabric/7(3)/GDRDMA
+ip-26-0-166-125:762818:763616 [3] NCCL INFO Channel 01/0 : 2[2] -> 5[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9632:10395 [7] NCCL INFO Channel 01/0 : 3[7] -> 2[6] via P2P/IPC
+ip-26-0-166-36:104434:105192 [5] NCCL INFO Channel 07/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-166-15:127297:128060 [7] NCCL INFO Channel 07/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-167-9:1039542:1040319 [5] NCCL INFO Channel 05/0 : 15[5] -> 14[4] via P2P/IPC
+ip-26-0-166-15:127293:128058 [3] NCCL INFO Channel 01/0 : 7[3] -> 6[2] via P2P/IPC
+ip-26-0-166-36:104432:105194 [3] NCCL INFO Channel 01/0 : 10[2] -> 13[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762817:763617 [2] NCCL INFO Channel 01/0 : 4[2] -> 7[3] [send] via NET/Libfabric/3(5)/GDRDMA
+ip-26-0-166-125:762818:763616 [3] NCCL INFO Channel 03/0 : 2[2] -> 5[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762817:763617 [2] NCCL INFO Channel 03/0 : 4[2] -> 7[3] [send] via NET/Libfabric/3(5)/GDRDMA
+ip-26-0-166-125:762822:763618 [7] NCCL INFO Channel 01/0 : 5[7] -> 4[6] via P2P/IPC
+ip-26-0-166-125:762818:763616 [3] NCCL INFO Channel 05/0 : 2[2] -> 5[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762820:763615 [5] NCCL INFO Channel 07/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-166-125:762817:763617 [2] NCCL INFO Channel 05/0 : 4[2] -> 7[3] [send] via NET/Libfabric/3(5)/GDRDMA
+ip-26-0-166-125:762818:763616 [3] NCCL INFO Channel 07/0 : 2[2] -> 5[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762817:763617 [2] NCCL INFO Channel 07/0 : 4[2] -> 7[3] [send] via NET/Libfabric/3(5)/GDRDMA
+ip-26-0-166-36:104431:105195 [2] NCCL INFO Channel 01/0 : 12[2] -> 15[3] [send] via NET/Libfabric/3(13)/GDRDMA
+ip-26-0-166-36:104432:105194 [3] NCCL INFO Channel 03/0 : 10[2] -> 13[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104431:105195 [2] NCCL INFO Channel 03/0 : 12[2] -> 15[3] [send] via NET/Libfabric/3(13)/GDRDMA
+ip-26-0-166-36:104432:105194 [3] NCCL INFO Channel 05/0 : 10[2] -> 13[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104431:105195 [2] NCCL INFO Channel 05/0 : 12[2] -> 15[3] [send] via NET/Libfabric/3(13)/GDRDMA
+ip-26-0-166-125:762822:763618 [7] NCCL INFO Channel 03/0 : 5[7] -> 4[6] via P2P/IPC
+ip-26-0-166-36:104432:105194 [3] NCCL INFO Channel 07/0 : 10[2] -> 13[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104431:105195 [2] NCCL INFO Channel 07/0 : 12[2] -> 15[3] [send] via NET/Libfabric/3(13)/GDRDMA
+ip-26-0-166-36:104432:105194 [3] NCCL INFO Channel 01/0 : 13[3] -> 12[2] via P2P/IPC
+ip-26-0-167-9:1039540:1040320 [3] NCCL INFO Channel 01/0 : 15[3] -> 14[2] via P2P/IPC
+ip-26-0-166-244:897245:898031 [5] NCCL INFO Channel 07/0 : 11[5] -> 10[4] via P2P/IPC
+ip-26-0-165-213:9630:10392 [5] NCCL INFO Channel 07/0 : 3[5] -> 2[4] via P2P/IPC
+ip-26-0-165-213:9632:10395 [7] NCCL INFO Channel 03/0 : 3[7] -> 2[6] via P2P/IPC
+ip-26-0-166-214:624950:625743 [7] NCCL INFO Channel 05/0 : 9[7] -> 8[6] via P2P/IPC
+ip-26-0-166-214:624948:625740 [5] NCCL INFO Channel 07/0 : 9[5] -> 8[4] via P2P/IPC
+ip-26-0-166-214:624946:625742 [3] NCCL INFO Channel 03/0 : 9[3] -> 8[2] via P2P/IPC
+ip-26-0-165-213:9632:10395 [7] NCCL INFO Channel 05/0 : 3[7] -> 2[6] via P2P/IPC
+ip-26-0-162-46:702883:703683 [3] NCCL INFO Channel 01/0 : 14[2] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624950:625743 [7] NCCL INFO Channel 07/0 : 9[7] -> 8[6] via P2P/IPC
+ip-26-0-162-46:702883:703683 [3] NCCL INFO Channel 03/0 : 14[2] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624946:625742 [3] NCCL INFO Channel 05/0 : 9[3] -> 8[2] via P2P/IPC
+ip-26-0-162-46:702883:703683 [3] NCCL INFO Channel 05/0 : 14[2] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104436:105197 [7] NCCL INFO Channel 05/0 : 13[7] -> 12[6] via P2P/IPC
+ip-26-0-162-46:702883:703683 [3] NCCL INFO Channel 07/0 : 14[2] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-167-9:1039544:1040322 [7] NCCL INFO Channel 03/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-162-46:702883:703683 [3] NCCL INFO Channel 01/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-166-244:897247:898035 [7] NCCL INFO Channel 05/0 : 11[7] -> 10[6] via P2P/IPC
+ip-26-0-166-244:897243:898033 [3] NCCL INFO Channel 03/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-165-213:9628:10394 [3] NCCL INFO Channel 01/0 : 0[2] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9627:10393 [2] NCCL INFO Channel 01/0 : 2[2] -> 5[3] [send] via NET/Libfabric/3(3)/GDRDMA
+ip-26-0-165-213:9628:10394 [3] NCCL INFO Channel 03/0 : 0[2] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9627:10393 [2] NCCL INFO Channel 03/0 : 2[2] -> 5[3] [send] via NET/Libfabric/3(3)/GDRDMA
+ip-26-0-165-213:9628:10394 [3] NCCL INFO Channel 05/0 : 0[2] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9627:10393 [2] NCCL INFO Channel 05/0 : 2[2] -> 5[3] [send] via NET/Libfabric/3(3)/GDRDMA
+ip-26-0-165-213:9628:10394 [3] NCCL INFO Channel 07/0 : 0[2] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9627:10393 [2] NCCL INFO Channel 07/0 : 2[2] -> 5[3] [send] via NET/Libfabric/3(3)/GDRDMA
+ip-26-0-166-36:104432:105194 [3] NCCL INFO Channel 03/0 : 13[3] -> 12[2] via P2P/IPC
+ip-26-0-165-213:9628:10394 [3] NCCL INFO Channel 01/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-36:104436:105197 [7] NCCL INFO Channel 07/0 : 13[7] -> 12[6] via P2P/IPC
+ip-26-0-166-125:762818:763616 [3] NCCL INFO Channel 01/0 : 5[3] -> 4[2] via P2P/IPC
+ip-26-0-166-36:104432:105194 [3] NCCL INFO Channel 05/0 : 13[3] -> 12[2] via P2P/IPC
+ip-26-0-165-213:9632:10395 [7] NCCL INFO Channel 07/0 : 3[7] -> 2[6] via P2P/IPC
+ip-26-0-166-214:624946:625742 [3] NCCL INFO Channel 07/0 : 9[3] -> 8[2] via P2P/IPC
+ip-26-0-167-9:1039540:1040320 [3] NCCL INFO Channel 03/0 : 15[3] -> 14[2] via P2P/IPC
+ip-26-0-166-244:897247:898035 [7] NCCL INFO Channel 07/0 : 11[7] -> 10[6] via P2P/IPC
+ip-26-0-166-244:897243:898033 [3] NCCL INFO Channel 05/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-167-9:1039544:1040322 [7] NCCL INFO Channel 05/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-162-46:702887:703686 [7] NCCL INFO Channel 03/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-167-9:1039540:1040320 [3] NCCL INFO Channel 05/0 : 15[3] -> 14[2] via P2P/IPC
+ip-26-0-167-9:1039544:1040322 [7] NCCL INFO Channel 07/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-167-9:1039540:1040320 [3] NCCL INFO Channel 07/0 : 15[3] -> 14[2] via P2P/IPC
+ip-26-0-167-9:1039542:1040319 [5] NCCL INFO Channel 07/0 : 15[5] -> 14[4] via P2P/IPC
+ip-26-0-166-244:897243:898033 [3] NCCL INFO Channel 07/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-166-125:762818:763616 [3] NCCL INFO Channel 03/0 : 5[3] -> 4[2] via P2P/IPC
+ip-26-0-166-125:762822:763618 [7] NCCL INFO Channel 05/0 : 5[7] -> 4[6] via P2P/IPC
+ip-26-0-165-213:9628:10394 [3] NCCL INFO Channel 03/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-36:104432:105194 [3] NCCL INFO Channel 07/0 : 13[3] -> 12[2] via P2P/IPC
+ip-26-0-162-46:702885:703680 [5] NCCL INFO Channel 05/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-162-46:702883:703683 [3] NCCL INFO Channel 03/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-162-46:702887:703686 [7] NCCL INFO Channel 05/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-166-15:127293:128058 [3] NCCL INFO Channel 03/0 : 7[3] -> 6[2] via P2P/IPC
+ip-26-0-162-46:702885:703680 [5] NCCL INFO Channel 07/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-162-46:702883:703683 [3] NCCL INFO Channel 05/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-162-46:702887:703686 [7] NCCL INFO Channel 07/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-162-46:702883:703683 [3] NCCL INFO Channel 07/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-166-125:762818:763616 [3] NCCL INFO Channel 05/0 : 5[3] -> 4[2] via P2P/IPC
+ip-26-0-166-125:762822:763618 [7] NCCL INFO Channel 07/0 : 5[7] -> 4[6] via P2P/IPC
+ip-26-0-165-213:9628:10394 [3] NCCL INFO Channel 05/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-125:762818:763616 [3] NCCL INFO Channel 07/0 : 5[3] -> 4[2] via P2P/IPC
+ip-26-0-166-15:127293:128058 [3] NCCL INFO Channel 05/0 : 7[3] -> 6[2] via P2P/IPC
+ip-26-0-165-213:9628:10394 [3] NCCL INFO Channel 07/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-15:127293:128058 [3] NCCL INFO Channel 07/0 : 7[3] -> 6[2] via P2P/IPC
+ip-26-0-167-9:1039538:1040317 [1] NCCL INFO Connected all rings
+ip-26-0-166-244:897240:898030 [0] NCCL INFO Connected all rings
+ip-26-0-166-244:897240:898030 [0] NCCL INFO Channel 01/0 : 10[0] -> 11[1] via P2P/IPC
+ip-26-0-165-213:9626:10389 [1] NCCL INFO Connected all rings
+ip-26-0-166-244:897240:898030 [0] NCCL INFO Channel 03/0 : 10[0] -> 11[1] via P2P/IPC
+ip-26-0-166-244:897240:898030 [0] NCCL INFO Channel 05/0 : 10[0] -> 11[1] via P2P/IPC
+ip-26-0-167-9:1039537:1040316 [0] NCCL INFO Connected all rings
+ip-26-0-167-9:1039537:1040316 [0] NCCL INFO Channel 01/0 : 14[0] -> 15[1] via P2P/IPC
+ip-26-0-167-9:1039537:1040316 [0] NCCL INFO Channel 03/0 : 14[0] -> 15[1] via P2P/IPC
+ip-26-0-167-9:1039537:1040316 [0] NCCL INFO Channel 05/0 : 14[0] -> 15[1] via P2P/IPC
+ip-26-0-167-9:1039537:1040316 [0] NCCL INFO Channel 07/0 : 14[0] -> 15[1] via P2P/IPC
+ip-26-0-166-214:624944:625738 [1] NCCL INFO Connected all rings
+ip-26-0-166-15:127291:128053 [1] NCCL INFO Connected all rings
+ip-26-0-167-9:1039538:1040317 [1] NCCL INFO Channel 02/0 : 13[1] -> 15[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-167-9:1039537:1040316 [0] NCCL INFO Channel 00/0 : 12[0] -> 14[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-167-9:1039538:1040317 [1] NCCL INFO Channel 03/0 : 13[1] -> 15[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-167-9:1039537:1040316 [0] NCCL INFO Channel 01/0 : 12[0] -> 14[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897240:898030 [0] NCCL INFO Channel 07/0 : 10[0] -> 11[1] via P2P/IPC
+ip-26-0-162-46:702885:703680 [5] NCCL INFO Connected all rings
+ip-26-0-166-15:127295:128055 [5] NCCL INFO Connected all rings
+ip-26-0-166-244:897241:898029 [1] NCCL INFO Connected all rings
+ip-26-0-167-9:1039541:1040318 [4] NCCL INFO Connected all rings
+ip-26-0-167-9:1039541:1040318 [4] NCCL INFO Channel 01/0 : 14[4] -> 15[5] via P2P/IPC
+ip-26-0-166-125:762815:763613 [0] NCCL INFO Connected all rings
+ip-26-0-166-125:762815:763613 [0] NCCL INFO Channel 01/0 : 4[0] -> 5[1] via P2P/IPC
+ip-26-0-166-244:897241:898029 [1] NCCL INFO Channel 06/0 : 9[1] -> 11[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897240:898030 [0] NCCL INFO Channel 04/0 : 8[0] -> 10[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897241:898029 [1] NCCL INFO Channel 07/0 : 9[1] -> 11[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897240:898030 [0] NCCL INFO Channel 05/0 : 8[0] -> 10[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897241:898029 [1] NCCL INFO Channel 02/0 : 11[1] -> 13[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897240:898030 [0] NCCL INFO Channel 00/0 : 10[0] -> 12[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897241:898029 [1] NCCL INFO Channel 03/0 : 11[1] -> 13[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897240:898030 [0] NCCL INFO Channel 01/0 : 10[0] -> 12[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-167-9:1039541:1040318 [4] NCCL INFO Channel 03/0 : 14[4] -> 15[5] via P2P/IPC
+ip-26-0-166-244:897241:898029 [1] NCCL INFO Channel 06/0 : 11[1] -> 13[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897240:898030 [0] NCCL INFO Channel 04/0 : 10[0] -> 12[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897241:898029 [1] NCCL INFO Channel 07/0 : 11[1] -> 13[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897240:898030 [0] NCCL INFO Channel 05/0 : 10[0] -> 12[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762815:763613 [0] NCCL INFO Channel 03/0 : 4[0] -> 5[1] via P2P/IPC
+ip-26-0-167-9:1039541:1040318 [4] NCCL INFO Channel 05/0 : 14[4] -> 15[5] via P2P/IPC
+ip-26-0-167-9:1039541:1040318 [4] NCCL INFO Channel 07/0 : 14[4] -> 15[5] via P2P/IPC
+ip-26-0-166-125:762815:763613 [0] NCCL INFO Channel 05/0 : 4[0] -> 5[1] via P2P/IPC
+ip-26-0-166-36:104429:105191 [0] NCCL INFO Connected all rings
+ip-26-0-166-36:104429:105191 [0] NCCL INFO Channel 01/0 : 12[0] -> 13[1] via P2P/IPC
+ip-26-0-166-15:127290:128054 [0] NCCL INFO Connected all rings
+ip-26-0-166-15:127290:128054 [0] NCCL INFO Channel 01/0 : 6[0] -> 7[1] via P2P/IPC
+ip-26-0-166-36:104429:105191 [0] NCCL INFO Channel 03/0 : 12[0] -> 13[1] via P2P/IPC
+ip-26-0-165-213:9625:10390 [0] NCCL INFO Connected all rings
+ip-26-0-165-213:9625:10390 [0] NCCL INFO Channel 01/0 : 2[0] -> 3[1] via P2P/IPC
+ip-26-0-166-15:127290:128054 [0] NCCL INFO Channel 03/0 : 6[0] -> 7[1] via P2P/IPC
+ip-26-0-166-214:624943:625737 [0] NCCL INFO Connected all rings
+ip-26-0-166-214:624943:625737 [0] NCCL INFO Channel 01/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-166-125:762815:763613 [0] NCCL INFO Channel 07/0 : 4[0] -> 5[1] via P2P/IPC
+ip-26-0-166-36:104429:105191 [0] NCCL INFO Channel 05/0 : 12[0] -> 13[1] via P2P/IPC
+ip-26-0-166-15:127290:128054 [0] NCCL INFO Channel 05/0 : 6[0] -> 7[1] via P2P/IPC
+ip-26-0-166-15:127292:128057 [2] NCCL INFO Connected all rings
+ip-26-0-166-15:127292:128057 [2] NCCL INFO Channel 01/0 : 6[2] -> 7[3] via P2P/IPC
+ip-26-0-166-36:104429:105191 [0] NCCL INFO Channel 07/0 : 12[0] -> 13[1] via P2P/IPC
+ip-26-0-165-213:9625:10390 [0] NCCL INFO Channel 03/0 : 2[0] -> 3[1] via P2P/IPC
+ip-26-0-166-214:624943:625737 [0] NCCL INFO Channel 03/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-166-15:127290:128054 [0] NCCL INFO Channel 07/0 : 6[0] -> 7[1] via P2P/IPC
+ip-26-0-165-213:9625:10390 [0] NCCL INFO Channel 05/0 : 2[0] -> 3[1] via P2P/IPC
+ip-26-0-166-214:624943:625737 [0] NCCL INFO Channel 05/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-165-213:9625:10390 [0] NCCL INFO Channel 07/0 : 2[0] -> 3[1] via P2P/IPC
+ip-26-0-166-15:127290:128054 [0] NCCL INFO Channel 00/0 : 4[0] -> 6[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127291:128053 [1] NCCL INFO Channel 02/0 : 5[1] -> 7[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127290:128054 [0] NCCL INFO Channel 01/0 : 4[0] -> 6[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127291:128053 [1] NCCL INFO Channel 03/0 : 5[1] -> 7[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624943:625737 [0] NCCL INFO Channel 07/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-165-213:9626:10389 [1] NCCL INFO Channel 06/0 : 1[1] -> 3[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9625:10390 [0] NCCL INFO Channel 04/0 : 0[0] -> 2[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9625:10390 [0] NCCL INFO Channel 05/0 : 0[0] -> 2[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9626:10389 [1] NCCL INFO Channel 07/0 : 1[1] -> 3[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9625:10390 [0] NCCL INFO Channel 00/0 : 2[0] -> 4[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9626:10389 [1] NCCL INFO Channel 02/0 : 3[1] -> 5[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9625:10390 [0] NCCL INFO Channel 01/0 : 2[0] -> 4[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9626:10389 [1] NCCL INFO Channel 03/0 : 3[1] -> 5[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9625:10390 [0] NCCL INFO Channel 04/0 : 2[0] -> 4[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9626:10389 [1] NCCL INFO Channel 06/0 : 3[1] -> 5[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9625:10390 [0] NCCL INFO Channel 05/0 : 2[0] -> 4[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9626:10389 [1] NCCL INFO Channel 07/0 : 3[1] -> 5[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624944:625738 [1] NCCL INFO Channel 06/0 : 9[1] -> 11[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624943:625737 [0] NCCL INFO Channel 04/0 : 8[0] -> 10[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624943:625737 [0] NCCL INFO Channel 05/0 : 8[0] -> 10[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624944:625738 [1] NCCL INFO Channel 07/0 : 9[1] -> 11[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624944:625738 [1] NCCL INFO Channel 02/0 : 5[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624943:625737 [0] NCCL INFO Channel 00/0 : 4[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624944:625738 [1] NCCL INFO Channel 03/0 : 5[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624943:625737 [0] NCCL INFO Channel 01/0 : 4[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624943:625737 [0] NCCL INFO Channel 00/0 : 8[0] -> 12[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624944:625738 [1] NCCL INFO Channel 02/0 : 9[1] -> 13[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624943:625737 [0] NCCL INFO Channel 01/0 : 8[0] -> 12[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624944:625738 [1] NCCL INFO Channel 03/0 : 9[1] -> 13[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762816:763612 [1] NCCL INFO Connected all rings
+ip-26-0-165-213:9629:10391 [4] NCCL INFO Connected all rings
+ip-26-0-165-213:9629:10391 [4] NCCL INFO Channel 01/0 : 2[4] -> 3[5] via P2P/IPC
+ip-26-0-166-125:762815:763613 [0] NCCL INFO Channel 00/0 : 2[0] -> 4[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762816:763612 [1] NCCL INFO Channel 02/0 : 3[1] -> 5[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762815:763613 [0] NCCL INFO Channel 01/0 : 2[0] -> 4[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762816:763612 [1] NCCL INFO Channel 03/0 : 3[1] -> 5[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762815:763613 [0] NCCL INFO Channel 04/0 : 2[0] -> 4[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897244:898032 [4] NCCL INFO Connected all rings
+ip-26-0-166-244:897244:898032 [4] NCCL INFO Channel 01/0 : 10[4] -> 11[5] via P2P/IPC
+ip-26-0-166-125:762816:763612 [1] NCCL INFO Channel 06/0 : 3[1] -> 5[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762815:763613 [0] NCCL INFO Channel 05/0 : 2[0] -> 4[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762816:763612 [1] NCCL INFO Channel 07/0 : 3[1] -> 5[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762815:763613 [0] NCCL INFO Channel 00/0 : 4[0] -> 6[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762816:763612 [1] NCCL INFO Channel 02/0 : 5[1] -> 7[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762815:763613 [0] NCCL INFO Channel 01/0 : 4[0] -> 6[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762816:763612 [1] NCCL INFO Channel 03/0 : 5[1] -> 7[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762815:763613 [0] NCCL INFO Channel 00/0 : 4[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762816:763612 [1] NCCL INFO Channel 02/0 : 5[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127290:128054 [0] NCCL INFO Channel 04/0 : 2[0] -> 6[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127291:128053 [1] NCCL INFO Channel 06/0 : 3[1] -> 7[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762815:763613 [0] NCCL INFO Channel 01/0 : 4[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762816:763612 [1] NCCL INFO Channel 03/0 : 5[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127290:128054 [0] NCCL INFO Channel 05/0 : 2[0] -> 6[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127290:128054 [0] NCCL INFO Channel 04/0 : 6[0] -> 10[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127291:128053 [1] NCCL INFO Channel 07/0 : 3[1] -> 7[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127290:128054 [0] NCCL INFO Channel 05/0 : 6[0] -> 10[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127291:128053 [1] NCCL INFO Channel 06/0 : 7[1] -> 11[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762815:763613 [0] NCCL INFO Channel 00/0 : 8[0] -> 4[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127291:128053 [1] NCCL INFO Channel 07/0 : 7[1] -> 11[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762816:763612 [1] NCCL INFO Channel 02/0 : 9[1] -> 5[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762815:763613 [0] NCCL INFO Channel 01/0 : 8[0] -> 4[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762816:763612 [1] NCCL INFO Channel 03/0 : 9[1] -> 5[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897244:898032 [4] NCCL INFO Channel 03/0 : 10[4] -> 11[5] via P2P/IPC
+ip-26-0-166-244:897244:898032 [4] NCCL INFO Channel 05/0 : 10[4] -> 11[5] via P2P/IPC
+ip-26-0-166-244:897244:898032 [4] NCCL INFO Channel 07/0 : 10[4] -> 11[5] via P2P/IPC
+ip-26-0-167-9:1039544:1040322 [7] NCCL INFO Connected all rings
+ip-26-0-162-46:702880:703676 [0] NCCL INFO Connected all rings
+ip-26-0-162-46:702880:703676 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-166-15:127296:128059 [6] NCCL INFO Connected all rings
+ip-26-0-166-15:127296:128059 [6] NCCL INFO Channel 01/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-162-46:702880:703676 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-162-46:702880:703676 [0] NCCL INFO Channel 05/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-162-46:702880:703676 [0] NCCL INFO Channel 07/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-167-9:1039540:1040320 [3] NCCL INFO Connected all rings
+ip-26-0-166-125:762822:763618 [7] NCCL INFO Connected all rings
+ip-26-0-166-214:624947:625739 [4] NCCL INFO Connected all rings
+ip-26-0-166-214:624947:625739 [4] NCCL INFO Channel 01/0 : 8[4] -> 9[5] via P2P/IPC
+ip-26-0-166-214:624947:625739 [4] NCCL INFO Channel 03/0 : 8[4] -> 9[5] via P2P/IPC
+ip-26-0-166-214:624947:625739 [4] NCCL INFO Channel 05/0 : 8[4] -> 9[5] via P2P/IPC
+ip-26-0-166-214:624947:625739 [4] NCCL INFO Channel 07/0 : 8[4] -> 9[5] via P2P/IPC
+ip-26-0-166-36:104433:105193 [4] NCCL INFO Connected all rings
+ip-26-0-166-36:104433:105193 [4] NCCL INFO Channel 01/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-166-214:624950:625743 [7] NCCL INFO Connected all rings
+ip-26-0-166-125:762820:763615 [5] NCCL INFO Connected all rings
+ip-26-0-165-213:9629:10391 [4] NCCL INFO Channel 03/0 : 2[4] -> 3[5] via P2P/IPC
+ip-26-0-166-36:104430:105190 [1] NCCL INFO Connected all rings
+ip-26-0-162-46:702881:703677 [1] NCCL INFO Connected all rings
+ip-26-0-165-213:9629:10391 [4] NCCL INFO Channel 05/0 : 2[4] -> 3[5] via P2P/IPC
+ip-26-0-166-244:897245:898031 [5] NCCL INFO Connected all rings
+ip-26-0-165-213:9629:10391 [4] NCCL INFO Channel 07/0 : 2[4] -> 3[5] via P2P/IPC
+ip-26-0-166-36:104430:105190 [1] NCCL INFO Channel 02/0 : 11[1] -> 13[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104429:105191 [0] NCCL INFO Channel 00/0 : 10[0] -> 12[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104430:105190 [1] NCCL INFO Channel 03/0 : 11[1] -> 13[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104429:105191 [0] NCCL INFO Channel 01/0 : 10[0] -> 12[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104430:105190 [1] NCCL INFO Channel 06/0 : 11[1] -> 13[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104429:105191 [0] NCCL INFO Channel 04/0 : 10[0] -> 12[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104430:105190 [1] NCCL INFO Channel 07/0 : 11[1] -> 13[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104430:105190 [1] NCCL INFO Channel 02/0 : 13[1] -> 15[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104429:105191 [0] NCCL INFO Channel 05/0 : 10[0] -> 12[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104430:105190 [1] NCCL INFO Channel 03/0 : 13[1] -> 15[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897241:898029 [1] NCCL INFO Channel 06/0 : 7[1] -> 11[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104430:105190 [1] NCCL INFO Channel 02/0 : 9[1] -> 13[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897241:898029 [1] NCCL INFO Channel 07/0 : 7[1] -> 11[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104430:105190 [1] NCCL INFO Channel 03/0 : 9[1] -> 13[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-167-9:1039538:1040317 [1] NCCL INFO Channel 06/0 : 7[1] -> 15[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-167-9:1039538:1040317 [1] NCCL INFO Channel 07/0 : 7[1] -> 15[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104430:105190 [1] NCCL INFO Channel 02/0 : 13[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-167-9:1039538:1040317 [1] NCCL INFO Channel 06/0 : 15[1] -> 7[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104430:105190 [1] NCCL INFO Channel 03/0 : 13[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-167-9:1039538:1040317 [1] NCCL INFO Channel 07/0 : 15[1] -> 7[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897245:898031 [5] NCCL INFO Channel 06/0 : 9[5] -> 11[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624944:625738 [1] NCCL INFO Channel 02/0 : 1[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897245:898031 [5] NCCL INFO Channel 07/0 : 9[5] -> 11[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624944:625738 [1] NCCL INFO Channel 03/0 : 1[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897245:898031 [5] NCCL INFO Channel 02/0 : 11[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624944:625738 [1] NCCL INFO Channel 02/0 : 9[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104431:105195 [2] NCCL INFO Connected all rings
+ip-26-0-166-244:897245:898031 [5] NCCL INFO Channel 03/0 : 11[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104431:105195 [2] NCCL INFO Channel 01/0 : 12[2] -> 13[3] via P2P/IPC
+ip-26-0-166-214:624944:625738 [1] NCCL INFO Channel 03/0 : 9[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897245:898031 [5] NCCL INFO Channel 06/0 : 11[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897245:898031 [5] NCCL INFO Channel 07/0 : 11[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-46:702882:703682 [2] NCCL INFO Connected all rings
+ip-26-0-162-46:702882:703682 [2] NCCL INFO Channel 01/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-166-36:104431:105195 [2] NCCL INFO Channel 03/0 : 12[2] -> 13[3] via P2P/IPC
+ip-26-0-166-244:897243:898033 [3] NCCL INFO Connected all rings
+ip-26-0-166-36:104431:105195 [2] NCCL INFO Channel 05/0 : 12[2] -> 13[3] via P2P/IPC
+ip-26-0-166-244:897246:898036 [6] NCCL INFO Connected all rings
+ip-26-0-166-244:897246:898036 [6] NCCL INFO Channel 01/0 : 10[6] -> 11[7] via P2P/IPC
+ip-26-0-162-46:702880:703676 [0] NCCL INFO Channel 04/0 : 0[0] -> 2[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-46:702881:703677 [1] NCCL INFO Channel 06/0 : 1[1] -> 3[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-46:702880:703676 [0] NCCL INFO Channel 05/0 : 0[0] -> 2[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-46:702881:703677 [1] NCCL INFO Channel 07/0 : 1[1] -> 3[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897244:898032 [4] NCCL INFO Channel 04/0 : 8[4] -> 10[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104433:105193 [4] NCCL INFO Channel 03/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-166-244:897244:898032 [4] NCCL INFO Channel 05/0 : 8[4] -> 10[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897244:898032 [4] NCCL INFO Channel 00/0 : 10[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897244:898032 [4] NCCL INFO Channel 01/0 : 10[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-46:702881:703677 [1] NCCL INFO Channel 02/0 : 9[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-46:702880:703676 [0] NCCL INFO Channel 00/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897244:898032 [4] NCCL INFO Channel 04/0 : 10[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-46:702881:703677 [1] NCCL INFO Channel 03/0 : 9[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-46:702880:703676 [0] NCCL INFO Channel 01/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897244:898032 [4] NCCL INFO Channel 05/0 : 10[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-46:702881:703677 [1] NCCL INFO Channel 02/0 : 1[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-46:702880:703676 [0] NCCL INFO Channel 00/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897246:898036 [6] NCCL INFO Channel 03/0 : 10[6] -> 11[7] via P2P/IPC
+ip-26-0-162-46:702881:703677 [1] NCCL INFO Channel 03/0 : 1[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-46:702880:703676 [0] NCCL INFO Channel 01/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897241:898029 [1] NCCL INFO Channel 06/0 : 11[1] -> 7[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897241:898029 [1] NCCL INFO Channel 07/0 : 11[1] -> 7[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-162-46:702881:703677 [1] NCCL INFO Channel 06/0 : 3[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104431:105195 [2] NCCL INFO Channel 07/0 : 12[2] -> 13[3] via P2P/IPC
+ip-26-0-162-46:702881:703677 [1] NCCL INFO Channel 07/0 : 3[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624944:625738 [1] NCCL INFO Channel 02/0 : 13[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897246:898036 [6] NCCL INFO Channel 05/0 : 10[6] -> 11[7] via P2P/IPC
+ip-26-0-166-214:624944:625738 [1] NCCL INFO Channel 03/0 : 13[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624944:625738 [1] NCCL INFO Channel 02/0 : 9[1] -> 5[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624944:625738 [1] NCCL INFO Channel 03/0 : 9[1] -> 5[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897246:898036 [6] NCCL INFO Channel 07/0 : 10[6] -> 11[7] via P2P/IPC
+ip-26-0-166-36:104430:105190 [1] NCCL INFO Channel 02/0 : 15[1] -> 13[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104429:105191 [0] NCCL INFO Channel 00/0 : 12[0] -> 14[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104430:105190 [1] NCCL INFO Channel 03/0 : 15[1] -> 13[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624944:625738 [1] NCCL INFO Channel 06/0 : 11[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104429:105191 [0] NCCL INFO Channel 01/0 : 12[0] -> 14[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104430:105190 [1] NCCL INFO Channel 02/0 : 13[1] -> 11[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762816:763612 [1] NCCL INFO Channel 02/0 : 7[1] -> 5[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624946:625742 [3] NCCL INFO Connected all rings
+ip-26-0-166-36:104430:105190 [1] NCCL INFO Channel 03/0 : 13[1] -> 11[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762816:763612 [1] NCCL INFO Channel 03/0 : 7[1] -> 5[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104430:105190 [1] NCCL INFO Channel 06/0 : 13[1] -> 11[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762816:763612 [1] NCCL INFO Channel 02/0 : 5[1] -> 3[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104430:105190 [1] NCCL INFO Channel 07/0 : 13[1] -> 11[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762816:763612 [1] NCCL INFO Channel 03/0 : 5[1] -> 3[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762816:763612 [1] NCCL INFO Channel 06/0 : 5[1] -> 3[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-125:762816:763612 [1] NCCL INFO Channel 07/0 : 5[1] -> 3[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-36:104429:105191 [0] NCCL INFO Channel 00/0 : 8[0] -> 12[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104429:105191 [0] NCCL INFO Channel 01/0 : 8[0] -> 12[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-167-9:1039537:1040316 [0] NCCL INFO Channel 04/0 : 6[0] -> 14[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-167-9:1039537:1040316 [0] NCCL INFO Channel 05/0 : 6[0] -> 14[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9625:10390 [0] NCCL INFO Channel 04/0 : 2[0] -> 6[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104429:105191 [0] NCCL INFO Channel 00/0 : 12[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9625:10390 [0] NCCL INFO Channel 05/0 : 2[0] -> 6[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-167-9:1039537:1040316 [0] NCCL INFO Channel 04/0 : 14[0] -> 6[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127294:128056 [4] NCCL INFO Connected all rings
+ip-26-0-166-15:127294:128056 [4] NCCL INFO Channel 01/0 : 6[4] -> 7[5] via P2P/IPC
+ip-26-0-166-36:104429:105191 [0] NCCL INFO Channel 01/0 : 12[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9626:10389 [1] NCCL INFO Channel 06/0 : 3[1] -> 7[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-167-9:1039537:1040316 [0] NCCL INFO Channel 05/0 : 14[0] -> 6[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9626:10389 [1] NCCL INFO Channel 07/0 : 3[1] -> 7[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9625:10390 [0] NCCL INFO Channel 04/0 : 6[0] -> 2[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9628:10394 [3] NCCL INFO Connected all rings
+ip-26-0-165-213:9625:10390 [0] NCCL INFO Channel 05/0 : 6[0] -> 2[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9626:10389 [1] NCCL INFO Channel 06/0 : 7[1] -> 3[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624944:625738 [1] NCCL INFO Channel 07/0 : 11[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9626:10389 [1] NCCL INFO Channel 07/0 : 7[1] -> 3[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-214:624948:625740 [5] NCCL INFO Connected all rings
+ip-26-0-166-15:127296:128059 [6] NCCL INFO Channel 03/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-15:127294:128056 [4] NCCL INFO Channel 03/0 : 6[4] -> 7[5] via P2P/IPC
+ip-26-0-166-244:897240:898030 [0] NCCL INFO Channel 04/0 : 6[0] -> 10[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897240:898030 [0] NCCL INFO Channel 05/0 : 6[0] -> 10[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624943:625737 [0] NCCL INFO Channel 00/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624943:625737 [0] NCCL INFO Channel 01/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897240:898030 [0] NCCL INFO Channel 04/0 : 10[0] -> 6[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624943:625737 [0] NCCL INFO Channel 00/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897240:898030 [0] NCCL INFO Channel 05/0 : 10[0] -> 6[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624943:625737 [0] NCCL INFO Channel 01/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624943:625737 [0] NCCL INFO Channel 00/0 : 12[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-46:702880:703676 [0] NCCL INFO Channel 04/0 : 2[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624943:625737 [0] NCCL INFO Channel 01/0 : 12[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-46:702880:703676 [0] NCCL INFO Channel 05/0 : 2[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104433:105193 [4] NCCL INFO Channel 05/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-166-214:624943:625737 [0] NCCL INFO Channel 00/0 : 8[0] -> 4[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624943:625737 [0] NCCL INFO Channel 01/0 : 8[0] -> 4[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104429:105191 [0] NCCL INFO Channel 00/0 : 14[0] -> 12[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104433:105193 [4] NCCL INFO Channel 07/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-166-36:104429:105191 [0] NCCL INFO Channel 01/0 : 14[0] -> 12[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624943:625737 [0] NCCL INFO Channel 04/0 : 10[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104429:105191 [0] NCCL INFO Channel 00/0 : 12[0] -> 10[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104429:105191 [0] NCCL INFO Channel 01/0 : 12[0] -> 10[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624947:625739 [4] NCCL INFO Channel 04/0 : 8[4] -> 10[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762815:763613 [0] NCCL INFO Channel 00/0 : 6[0] -> 4[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104429:105191 [0] NCCL INFO Channel 04/0 : 12[0] -> 10[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624943:625737 [0] NCCL INFO Channel 05/0 : 10[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624948:625740 [5] NCCL INFO Channel 06/0 : 9[5] -> 11[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104429:105191 [0] NCCL INFO Channel 05/0 : 12[0] -> 10[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762815:763613 [0] NCCL INFO Channel 01/0 : 6[0] -> 4[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624947:625739 [4] NCCL INFO Channel 05/0 : 8[4] -> 10[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624948:625740 [5] NCCL INFO Channel 07/0 : 9[5] -> 11[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762815:763613 [0] NCCL INFO Channel 00/0 : 4[0] -> 2[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762815:763613 [0] NCCL INFO Channel 01/0 : 4[0] -> 2[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762815:763613 [0] NCCL INFO Channel 04/0 : 4[0] -> 2[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624947:625739 [4] NCCL INFO Channel 00/0 : 4[4] -> 8[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624948:625740 [5] NCCL INFO Channel 02/0 : 5[5] -> 9[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762815:763613 [0] NCCL INFO Channel 05/0 : 4[0] -> 2[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127296:128059 [6] NCCL INFO Channel 05/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-15:127294:128056 [4] NCCL INFO Channel 05/0 : 6[4] -> 7[5] via P2P/IPC
+ip-26-0-166-214:624948:625740 [5] NCCL INFO Channel 03/0 : 5[5] -> 9[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624947:625739 [4] NCCL INFO Channel 01/0 : 4[4] -> 8[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624948:625740 [5] NCCL INFO Channel 02/0 : 9[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624947:625739 [4] NCCL INFO Channel 00/0 : 8[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624948:625740 [5] NCCL INFO Channel 03/0 : 9[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624947:625739 [4] NCCL INFO Channel 01/0 : 8[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-167-9:1039543:1040323 [6] NCCL INFO Connected all rings
+ip-26-0-167-9:1039543:1040323 [6] NCCL INFO Channel 01/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-166-15:127294:128056 [4] NCCL INFO Channel 07/0 : 6[4] -> 7[5] via P2P/IPC
+ip-26-0-165-213:9630:10392 [5] NCCL INFO Connected all rings
+ip-26-0-166-15:127291:128053 [1] NCCL INFO Channel 06/0 : 15[1] -> 7[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127291:128053 [1] NCCL INFO Channel 07/0 : 15[1] -> 7[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127291:128053 [1] NCCL INFO Channel 06/0 : 7[1] -> 15[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127296:128059 [6] NCCL INFO Channel 07/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-166-15:127291:128053 [1] NCCL INFO Channel 07/0 : 7[1] -> 15[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127291:128053 [1] NCCL INFO Channel 06/0 : 11[1] -> 7[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9632:10395 [7] NCCL INFO Connected all rings
+ip-26-0-166-15:127291:128053 [1] NCCL INFO Channel 07/0 : 11[1] -> 7[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127291:128053 [1] NCCL INFO Channel 06/0 : 7[1] -> 3[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-167-9:1039542:1040319 [5] NCCL INFO Connected all rings
+ip-26-0-166-15:127291:128053 [1] NCCL INFO Channel 07/0 : 7[1] -> 3[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-167-9:1039543:1040323 [6] NCCL INFO Channel 03/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-166-244:897241:898029 [1] NCCL INFO Channel 02/0 : 13[1] -> 11[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127291:128053 [1] NCCL INFO Channel 02/0 : 7[1] -> 5[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897241:898029 [1] NCCL INFO Channel 03/0 : 13[1] -> 11[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9626:10389 [1] NCCL INFO Channel 02/0 : 5[1] -> 3[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127291:128053 [1] NCCL INFO Channel 03/0 : 7[1] -> 5[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897241:898029 [1] NCCL INFO Channel 06/0 : 13[1] -> 11[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9626:10389 [1] NCCL INFO Channel 03/0 : 5[1] -> 3[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127294:128056 [4] NCCL INFO Channel 00/0 : 4[4] -> 6[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104434:105192 [5] NCCL INFO Connected all rings
+ip-26-0-166-244:897241:898029 [1] NCCL INFO Channel 07/0 : 13[1] -> 11[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127294:128056 [4] NCCL INFO Channel 01/0 : 4[4] -> 6[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-167-9:1039543:1040323 [6] NCCL INFO Channel 05/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-165-213:9626:10389 [1] NCCL INFO Channel 06/0 : 5[1] -> 3[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127291:128053 [1] NCCL INFO Channel 00/0 : 7[1] -> 6[0] via P2P/IPC
+ip-26-0-166-244:897241:898029 [1] NCCL INFO Channel 06/0 : 11[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9626:10389 [1] NCCL INFO Channel 07/0 : 5[1] -> 3[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897241:898029 [1] NCCL INFO Channel 07/0 : 11[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-165-213:9626:10389 [1] NCCL INFO Channel 06/0 : 3[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127295:128055 [5] NCCL INFO Channel 02/0 : 5[5] -> 7[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9626:10389 [1] NCCL INFO Channel 07/0 : 3[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-15:127295:128055 [5] NCCL INFO Channel 03/0 : 5[5] -> 7[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897241:898029 [1] NCCL INFO Channel 00/0 : 11[1] -> 10[0] via P2P/IPC
+ip-26-0-166-125:762816:763612 [1] NCCL INFO Channel 00/0 : 5[1] -> 4[0] via P2P/IPC
+ip-26-0-167-9:1039543:1040323 [6] NCCL INFO Channel 07/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-166-214:624944:625738 [1] NCCL INFO Channel 00/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-165-213:9626:10389 [1] NCCL INFO Channel 00/0 : 3[1] -> 2[0] via P2P/IPC
+ip-26-0-162-46:702881:703677 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127291:128053 [1] NCCL INFO Channel 02/0 : 7[1] -> 6[0] via P2P/IPC
+ip-26-0-165-213:9626:10389 [1] NCCL INFO Channel 02/0 : 3[1] -> 2[0] via P2P/IPC
+ip-26-0-167-9:1039544:1040322 [7] NCCL INFO Channel 02/0 : 13[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-167-9:1039543:1040323 [6] NCCL INFO Channel 00/0 : 12[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127291:128053 [1] NCCL INFO Channel 04/0 : 7[1] -> 6[0] via P2P/IPC
+ip-26-0-165-213:9630:10392 [5] NCCL INFO Channel 06/0 : 1[5] -> 3[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624944:625738 [1] NCCL INFO Channel 02/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-167-9:1039544:1040322 [7] NCCL INFO Channel 03/0 : 13[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9629:10391 [4] NCCL INFO Channel 04/0 : 0[4] -> 2[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9630:10392 [5] NCCL INFO Channel 07/0 : 1[5] -> 3[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-167-9:1039543:1040323 [6] NCCL INFO Channel 01/0 : 12[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9629:10391 [4] NCCL INFO Channel 05/0 : 0[4] -> 2[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9630:10392 [5] NCCL INFO Channel 02/0 : 3[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9629:10391 [4] NCCL INFO Channel 00/0 : 2[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-167-9:1039542:1040319 [5] NCCL INFO Channel 02/0 : 13[5] -> 15[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9630:10392 [5] NCCL INFO Channel 03/0 : 3[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-167-9:1039541:1040318 [4] NCCL INFO Channel 00/0 : 12[4] -> 14[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9629:10391 [4] NCCL INFO Channel 01/0 : 2[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9630:10392 [5] NCCL INFO Channel 06/0 : 3[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-167-9:1039542:1040319 [5] NCCL INFO Channel 03/0 : 13[5] -> 15[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-167-9:1039538:1040317 [1] NCCL INFO Channel 02/0 : 15[1] -> 13[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-167-9:1039541:1040318 [4] NCCL INFO Channel 01/0 : 12[4] -> 14[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127290:128054 [0] NCCL INFO Channel 04/0 : 14[0] -> 6[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9630:10392 [5] NCCL INFO Channel 07/0 : 3[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9629:10391 [4] NCCL INFO Channel 04/0 : 2[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-167-9:1039538:1040317 [1] NCCL INFO Channel 03/0 : 15[1] -> 13[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-166-244:897241:898029 [1] NCCL INFO Channel 02/0 : 11[1] -> 10[0] via P2P/IPC
+ip-26-0-166-15:127290:128054 [0] NCCL INFO Channel 05/0 : 14[0] -> 6[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9629:10391 [4] NCCL INFO Channel 05/0 : 2[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9626:10389 [1] NCCL INFO Channel 04/0 : 3[1] -> 2[0] via P2P/IPC
+ip-26-0-166-15:127290:128054 [0] NCCL INFO Channel 04/0 : 6[0] -> 14[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-167-9:1039538:1040317 [1] NCCL INFO Channel 00/0 : 15[1] -> 14[0] via P2P/IPC
+ip-26-0-166-15:127291:128053 [1] NCCL INFO Channel 06/0 : 7[1] -> 6[0] via P2P/IPC
+ip-26-0-166-36:104430:105190 [1] NCCL INFO Channel 00/0 : 13[1] -> 12[0] via P2P/IPC
+ip-26-0-162-46:702881:703677 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-15:127290:128054 [0] NCCL INFO Channel 05/0 : 6[0] -> 14[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-244:897241:898029 [1] NCCL INFO Channel 04/0 : 11[1] -> 10[0] via P2P/IPC
+ip-26-0-166-36:104433:105193 [4] NCCL INFO Channel 00/0 : 10[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104434:105192 [5] NCCL INFO Channel 02/0 : 11[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104433:105193 [4] NCCL INFO Channel 01/0 : 10[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104434:105192 [5] NCCL INFO Channel 03/0 : 11[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104433:105193 [4] NCCL INFO Channel 04/0 : 10[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104434:105192 [5] NCCL INFO Channel 06/0 : 11[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9626:10389 [1] NCCL INFO Channel 06/0 : 3[1] -> 2[0] via P2P/IPC
+ip-26-0-166-15:127290:128054 [0] NCCL INFO Channel 04/0 : 10[0] -> 6[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104433:105193 [4] NCCL INFO Channel 05/0 : 10[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104434:105192 [5] NCCL INFO Channel 07/0 : 11[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-167-9:1039537:1040316 [0] NCCL INFO Channel 00/0 : 14[0] -> 12[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-15:127290:128054 [0] NCCL INFO Channel 05/0 : 10[0] -> 6[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104433:105193 [4] NCCL INFO Channel 00/0 : 12[4] -> 14[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-167-9:1039538:1040317 [1] NCCL INFO Channel 02/0 : 15[1] -> 14[0] via P2P/IPC
+ip-26-0-166-125:762819:763614 [4] NCCL INFO Connected all rings
+ip-26-0-166-15:127290:128054 [0] NCCL INFO Channel 04/0 : 6[0] -> 2[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-167-9:1039537:1040316 [0] NCCL INFO Channel 01/0 : 14[0] -> 12[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104434:105192 [5] NCCL INFO Channel 02/0 : 13[5] -> 15[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897241:898029 [1] NCCL INFO Channel 06/0 : 11[1] -> 10[0] via P2P/IPC
+ip-26-0-166-36:104430:105190 [1] NCCL INFO Channel 02/0 : 13[1] -> 12[0] via P2P/IPC
+ip-26-0-166-125:762819:763614 [4] NCCL INFO Channel 01/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-166-36:104433:105193 [4] NCCL INFO Channel 01/0 : 12[4] -> 14[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127290:128054 [0] NCCL INFO Channel 05/0 : 6[0] -> 2[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104434:105192 [5] NCCL INFO Channel 03/0 : 13[5] -> 15[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-167-9:1039538:1040317 [1] NCCL INFO Channel 04/0 : 15[1] -> 14[0] via P2P/IPC
+ip-26-0-166-36:104430:105190 [1] NCCL INFO Channel 04/0 : 13[1] -> 12[0] via P2P/IPC
+ip-26-0-166-244:897244:898032 [4] NCCL INFO Channel 04/0 : 6[4] -> 10[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897245:898031 [5] NCCL INFO Channel 06/0 : 7[5] -> 11[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897240:898030 [0] NCCL INFO Channel 00/0 : 12[0] -> 10[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624944:625738 [1] NCCL INFO Channel 04/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-166-36:104433:105193 [4] NCCL INFO Channel 00/0 : 8[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127290:128054 [0] NCCL INFO Channel 00/0 : 6[0] -> 4[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762816:763612 [1] NCCL INFO Channel 02/0 : 5[1] -> 4[0] via P2P/IPC
+ip-26-0-166-244:897244:898032 [4] NCCL INFO Channel 05/0 : 6[4] -> 10[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-167-9:1039541:1040318 [4] NCCL INFO Channel 04/0 : 6[4] -> 14[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897245:898031 [5] NCCL INFO Channel 07/0 : 7[5] -> 11[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897240:898030 [0] NCCL INFO Channel 01/0 : 12[0] -> 10[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104434:105192 [5] NCCL INFO Channel 02/0 : 9[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127290:128054 [0] NCCL INFO Channel 01/0 : 6[0] -> 4[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-167-9:1039542:1040319 [5] NCCL INFO Channel 06/0 : 7[5] -> 15[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897240:898030 [0] NCCL INFO Channel 04/0 : 12[0] -> 10[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104433:105193 [4] NCCL INFO Channel 01/0 : 8[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-167-9:1039542:1040319 [5] NCCL INFO Channel 07/0 : 7[5] -> 15[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-167-9:1039541:1040318 [4] NCCL INFO Channel 05/0 : 6[4] -> 14[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104434:105192 [5] NCCL INFO Channel 03/0 : 9[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897247:898035 [7] NCCL INFO Connected all rings
+ip-26-0-167-9:1039538:1040317 [1] NCCL INFO Channel 06/0 : 15[1] -> 14[0] via P2P/IPC
+ip-26-0-166-244:897240:898030 [0] NCCL INFO Channel 05/0 : 12[0] -> 10[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104433:105193 [4] NCCL INFO Channel 00/0 : 12[4] -> 8[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-167-9:1039542:1040319 [5] NCCL INFO Channel 06/0 : 15[5] -> 7[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-167-9:1039541:1040318 [4] NCCL INFO Channel 04/0 : 14[4] -> 6[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897240:898030 [0] NCCL INFO Channel 04/0 : 10[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104434:105192 [5] NCCL INFO Channel 02/0 : 13[5] -> 9[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-167-9:1039542:1040319 [5] NCCL INFO Channel 07/0 : 15[5] -> 7[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-167-9:1039541:1040318 [4] NCCL INFO Channel 05/0 : 14[4] -> 6[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897240:898030 [0] NCCL INFO Channel 05/0 : 10[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-36:104433:105193 [4] NCCL INFO Channel 01/0 : 12[4] -> 8[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104430:105190 [1] NCCL INFO Channel 06/0 : 13[1] -> 12[0] via P2P/IPC
+ip-26-0-166-36:104434:105192 [5] NCCL INFO Channel 03/0 : 13[5] -> 9[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762819:763614 [4] NCCL INFO Channel 03/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-162-46:702881:703677 [1] NCCL INFO Channel 04/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-165-213:9625:10390 [0] NCCL INFO Channel 00/0 : 4[0] -> 2[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9625:10390 [0] NCCL INFO Channel 01/0 : 4[0] -> 2[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-214:624944:625738 [1] NCCL INFO Channel 06/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-165-213:9625:10390 [0] NCCL INFO Channel 04/0 : 4[0] -> 2[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-162-46:702882:703682 [2] NCCL INFO Channel 03/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-165-213:9625:10390 [0] NCCL INFO Channel 05/0 : 4[0] -> 2[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9625:10390 [0] NCCL INFO Channel 04/0 : 2[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-165-213:9625:10390 [0] NCCL INFO Channel 05/0 : 2[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-166-125:762816:763612 [1] NCCL INFO Channel 04/0 : 5[1] -> 4[0] via P2P/IPC
+ip-26-0-166-125:762819:763614 [4] NCCL INFO Channel 05/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-166-214:624945:625741 [2] NCCL INFO Connected all rings
+ip-26-0-166-214:624945:625741 [2] NCCL INFO Channel 01/0 : 8[2] -> 9[3] via P2P/IPC
+ip-26-0-166-244:897246:898036 [6] NCCL INFO Channel 04/0 : 8[6] -> 10[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897247:898035 [7] NCCL INFO Channel 06/0 : 9[7] -> 11[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897246:898036 [6] NCCL INFO Channel 05/0 : 8[6] -> 10[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897247:898035 [7] NCCL INFO Channel 07/0 : 9[7] -> 11[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762816:763612 [1] NCCL INFO Channel 06/0 : 5[1] -> 4[0] via P2P/IPC
+ip-26-0-166-244:897246:898036 [6] NCCL INFO Channel 00/0 : 10[6] -> 12[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897247:898035 [7] NCCL INFO Channel 02/0 : 11[7] -> 13[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897246:898036 [6] NCCL INFO Channel 01/0 : 10[6] -> 12[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897247:898035 [7] NCCL INFO Channel 03/0 : 11[7] -> 13[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897246:898036 [6] NCCL INFO Channel 04/0 : 10[6] -> 12[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897247:898035 [7] NCCL INFO Channel 06/0 : 11[7] -> 13[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897246:898036 [6] NCCL INFO Channel 05/0 : 10[6] -> 12[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897247:898035 [7] NCCL INFO Channel 07/0 : 11[7] -> 13[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762819:763614 [4] NCCL INFO Channel 07/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-166-214:624945:625741 [2] NCCL INFO Channel 03/0 : 8[2] -> 9[3] via P2P/IPC
+ip-26-0-166-36:104435:105196 [6] NCCL INFO Connected all rings
+ip-26-0-166-36:104435:105196 [6] NCCL INFO Channel 01/0 : 12[6] -> 13[7] via P2P/IPC
+ip-26-0-166-125:762820:763615 [5] NCCL INFO Channel 02/0 : 3[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762819:763614 [4] NCCL INFO Channel 00/0 : 2[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762820:763615 [5] NCCL INFO Channel 03/0 : 3[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762819:763614 [4] NCCL INFO Channel 01/0 : 2[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762820:763615 [5] NCCL INFO Channel 06/0 : 3[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762819:763614 [4] NCCL INFO Channel 04/0 : 2[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762820:763615 [5] NCCL INFO Channel 07/0 : 3[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762819:763614 [4] NCCL INFO Channel 05/0 : 2[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762820:763615 [5] NCCL INFO Channel 02/0 : 5[5] -> 7[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762819:763614 [4] NCCL INFO Channel 00/0 : 4[4] -> 6[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762820:763615 [5] NCCL INFO Channel 03/0 : 5[5] -> 7[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762819:763614 [4] NCCL INFO Channel 01/0 : 4[4] -> 6[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624949:625744 [6] NCCL INFO Connected all rings
+ip-26-0-166-214:624949:625744 [6] NCCL INFO Channel 01/0 : 8[6] -> 9[7] via P2P/IPC
+ip-26-0-166-125:762820:763615 [5] NCCL INFO Channel 02/0 : 5[5] -> 9[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762819:763614 [4] NCCL INFO Channel 00/0 : 4[4] -> 8[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127295:128055 [5] NCCL INFO Channel 06/0 : 3[5] -> 7[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762820:763615 [5] NCCL INFO Channel 03/0 : 5[5] -> 9[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127295:128055 [5] NCCL INFO Channel 07/0 : 3[5] -> 7[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127294:128056 [4] NCCL INFO Channel 04/0 : 2[4] -> 6[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762819:763614 [4] NCCL INFO Channel 01/0 : 4[4] -> 8[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127295:128055 [5] NCCL INFO Channel 06/0 : 7[5] -> 11[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127294:128056 [4] NCCL INFO Channel 05/0 : 2[4] -> 6[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762820:763615 [5] NCCL INFO Channel 02/0 : 9[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127295:128055 [5] NCCL INFO Channel 07/0 : 7[5] -> 11[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127294:128056 [4] NCCL INFO Channel 04/0 : 6[4] -> 10[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-46:702881:703677 [1] NCCL INFO Channel 06/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-166-125:762819:763614 [4] NCCL INFO Channel 00/0 : 8[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127294:128056 [4] NCCL INFO Channel 05/0 : 6[4] -> 10[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624948:625740 [5] NCCL INFO Channel 02/0 : 1[5] -> 9[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762820:763615 [5] NCCL INFO Channel 03/0 : 9[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624947:625739 [4] NCCL INFO Channel 00/0 : 0[4] -> 8[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624945:625741 [2] NCCL INFO Channel 05/0 : 8[2] -> 9[3] via P2P/IPC
+ip-26-0-166-125:762819:763614 [4] NCCL INFO Channel 01/0 : 8[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-46:702882:703682 [2] NCCL INFO Channel 05/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-166-214:624948:625740 [5] NCCL INFO Channel 03/0 : 1[5] -> 9[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624947:625739 [4] NCCL INFO Channel 01/0 : 0[4] -> 8[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624948:625740 [5] NCCL INFO Channel 02/0 : 9[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624947:625739 [4] NCCL INFO Channel 00/0 : 8[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624948:625740 [5] NCCL INFO Channel 03/0 : 9[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897245:898031 [5] NCCL INFO Channel 06/0 : 11[5] -> 7[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624949:625744 [6] NCCL INFO Channel 03/0 : 8[6] -> 9[7] via P2P/IPC
+ip-26-0-166-214:624947:625739 [4] NCCL INFO Channel 01/0 : 8[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897245:898031 [5] NCCL INFO Channel 07/0 : 11[5] -> 7[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897244:898032 [4] NCCL INFO Channel 04/0 : 10[4] -> 6[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897244:898032 [4] NCCL INFO Channel 05/0 : 10[4] -> 6[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624945:625741 [2] NCCL INFO Channel 07/0 : 8[2] -> 9[3] via P2P/IPC
+ip-26-0-166-214:624949:625744 [6] NCCL INFO Channel 05/0 : 8[6] -> 9[7] via P2P/IPC
+ip-26-0-162-46:702882:703682 [2] NCCL INFO Channel 07/0 : 0[2] -> 1[3] via P2P/IPC
+ip-26-0-166-214:624949:625744 [6] NCCL INFO Channel 07/0 : 8[6] -> 9[7] via P2P/IPC
+ip-26-0-166-214:624946:625742 [3] NCCL INFO Channel 06/0 : 9[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624945:625741 [2] NCCL INFO Channel 04/0 : 8[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624946:625742 [3] NCCL INFO Channel 07/0 : 9[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624945:625741 [2] NCCL INFO Channel 05/0 : 8[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104435:105196 [6] NCCL INFO Channel 03/0 : 12[6] -> 13[7] via P2P/IPC
+ip-26-0-166-214:624949:625744 [6] NCCL INFO Channel 04/0 : 8[6] -> 10[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624950:625743 [7] NCCL INFO Channel 06/0 : 9[7] -> 11[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624949:625744 [6] NCCL INFO Channel 05/0 : 8[6] -> 10[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624950:625743 [7] NCCL INFO Channel 07/0 : 9[7] -> 11[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624949:625744 [6] NCCL INFO Channel 00/0 : 4[6] -> 8[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624950:625743 [7] NCCL INFO Channel 02/0 : 5[7] -> 9[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624949:625744 [6] NCCL INFO Channel 01/0 : 4[6] -> 8[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624949:625744 [6] NCCL INFO Channel 00/0 : 8[6] -> 12[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624949:625744 [6] NCCL INFO Channel 01/0 : 8[6] -> 12[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624950:625743 [7] NCCL INFO Channel 03/0 : 5[7] -> 9[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624950:625743 [7] NCCL INFO Channel 02/0 : 9[7] -> 13[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624950:625743 [7] NCCL INFO Channel 03/0 : 9[7] -> 13[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-46:702886:703685 [6] NCCL INFO Connected all rings
+ip-26-0-162-46:702886:703685 [6] NCCL INFO Channel 01/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-162-46:702886:703685 [6] NCCL INFO Channel 03/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-162-46:702886:703685 [6] NCCL INFO Channel 05/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-162-46:702886:703685 [6] NCCL INFO Channel 07/0 : 0[6] -> 1[7] via P2P/IPC
+ip-26-0-166-36:104435:105196 [6] NCCL INFO Channel 05/0 : 12[6] -> 13[7] via P2P/IPC
+ip-26-0-162-46:702884:703679 [4] NCCL INFO Connected all rings
+ip-26-0-162-46:702884:703679 [4] NCCL INFO Channel 01/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-166-36:104435:105196 [6] NCCL INFO Channel 07/0 : 12[6] -> 13[7] via P2P/IPC
+ip-26-0-162-46:702884:703679 [4] NCCL INFO Channel 03/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-162-46:702884:703679 [4] NCCL INFO Channel 05/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-166-36:104436:105197 [7] NCCL INFO Connected all rings
+ip-26-0-162-46:702884:703679 [4] NCCL INFO Channel 07/0 : 0[4] -> 1[5] via P2P/IPC
+ip-26-0-162-46:702884:703679 [4] NCCL INFO Channel 04/0 : 0[4] -> 2[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-46:702885:703680 [5] NCCL INFO Channel 06/0 : 1[5] -> 3[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104435:105196 [6] NCCL INFO Channel 00/0 : 10[6] -> 12[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104436:105197 [7] NCCL INFO Channel 02/0 : 11[7] -> 13[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104435:105196 [6] NCCL INFO Channel 01/0 : 10[6] -> 12[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-46:702885:703680 [5] NCCL INFO Channel 07/0 : 1[5] -> 3[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-46:702884:703679 [4] NCCL INFO Channel 05/0 : 0[4] -> 2[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104436:105197 [7] NCCL INFO Channel 03/0 : 11[7] -> 13[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104435:105196 [6] NCCL INFO Channel 04/0 : 10[6] -> 12[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104436:105197 [7] NCCL INFO Channel 06/0 : 11[7] -> 13[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104435:105196 [6] NCCL INFO Channel 05/0 : 10[6] -> 12[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104436:105197 [7] NCCL INFO Channel 07/0 : 11[7] -> 13[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104435:105196 [6] NCCL INFO Channel 00/0 : 12[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-46:702884:703679 [4] NCCL INFO Channel 00/0 : 8[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-46:702885:703680 [5] NCCL INFO Channel 02/0 : 9[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104436:105197 [7] NCCL INFO Channel 02/0 : 13[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104435:105196 [6] NCCL INFO Channel 01/0 : 12[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9630:10392 [5] NCCL INFO Channel 06/0 : 3[5] -> 7[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-46:702884:703679 [4] NCCL INFO Channel 01/0 : 8[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-46:702885:703680 [5] NCCL INFO Channel 03/0 : 9[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104436:105197 [7] NCCL INFO Channel 03/0 : 13[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9629:10391 [4] NCCL INFO Channel 04/0 : 2[4] -> 6[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9630:10392 [5] NCCL INFO Channel 07/0 : 3[5] -> 7[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-162-46:702884:703679 [4] NCCL INFO Channel 00/0 : 0[4] -> 8[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-46:702885:703680 [5] NCCL INFO Channel 02/0 : 1[5] -> 9[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9629:10391 [4] NCCL INFO Channel 05/0 : 2[4] -> 6[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-46:702884:703679 [4] NCCL INFO Channel 01/0 : 0[4] -> 8[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-46:702885:703680 [5] NCCL INFO Channel 03/0 : 1[5] -> 9[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897246:898036 [6] NCCL INFO Channel 04/0 : 6[6] -> 10[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9630:10392 [5] NCCL INFO Channel 06/0 : 7[5] -> 3[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897247:898035 [7] NCCL INFO Channel 06/0 : 7[7] -> 11[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104435:105196 [6] NCCL INFO Channel 00/0 : 8[6] -> 12[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9629:10391 [4] NCCL INFO Channel 04/0 : 6[4] -> 2[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127295:128055 [5] NCCL INFO Channel 06/0 : 15[5] -> 7[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897246:898036 [6] NCCL INFO Channel 05/0 : 6[6] -> 10[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897247:898035 [7] NCCL INFO Channel 07/0 : 7[7] -> 11[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104436:105197 [7] NCCL INFO Channel 02/0 : 9[7] -> 13[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9630:10392 [5] NCCL INFO Channel 07/0 : 7[5] -> 3[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9629:10391 [4] NCCL INFO Channel 05/0 : 6[4] -> 2[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127294:128056 [4] NCCL INFO Channel 04/0 : 14[4] -> 6[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-167-9:1039543:1040323 [6] NCCL INFO Channel 04/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-46:702884:703679 [4] NCCL INFO Channel 04/0 : 2[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-46:702885:703680 [5] NCCL INFO Channel 06/0 : 3[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104435:105196 [6] NCCL INFO Channel 01/0 : 8[6] -> 12[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104436:105197 [7] NCCL INFO Channel 03/0 : 9[7] -> 13[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127295:128055 [5] NCCL INFO Channel 07/0 : 15[5] -> 7[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127294:128056 [4] NCCL INFO Channel 05/0 : 14[4] -> 6[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-167-9:1039544:1040322 [7] NCCL INFO Channel 06/0 : 7[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-167-9:1039543:1040323 [6] NCCL INFO Channel 05/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-46:702884:703679 [4] NCCL INFO Channel 05/0 : 2[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-162-46:702885:703680 [5] NCCL INFO Channel 07/0 : 3[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624947:625739 [4] NCCL INFO Channel 00/0 : 12[4] -> 8[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127295:128055 [5] NCCL INFO Channel 06/0 : 7[5] -> 15[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127294:128056 [4] NCCL INFO Channel 04/0 : 6[4] -> 14[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-167-9:1039544:1040322 [7] NCCL INFO Channel 07/0 : 7[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624947:625739 [4] NCCL INFO Channel 01/0 : 12[4] -> 8[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127295:128055 [5] NCCL INFO Channel 07/0 : 7[5] -> 15[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-167-9:1039543:1040323 [6] NCCL INFO Channel 04/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127294:128056 [4] NCCL INFO Channel 05/0 : 6[4] -> 14[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-167-9:1039544:1040322 [7] NCCL INFO Channel 06/0 : 15[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104435:105196 [6] NCCL INFO Channel 00/0 : 12[6] -> 8[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104436:105197 [7] NCCL INFO Channel 02/0 : 13[7] -> 9[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624947:625739 [4] NCCL INFO Channel 00/0 : 8[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-167-9:1039543:1040323 [6] NCCL INFO Channel 05/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-167-9:1039544:1040322 [7] NCCL INFO Channel 07/0 : 15[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104435:105196 [6] NCCL INFO Channel 01/0 : 12[6] -> 8[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104436:105197 [7] NCCL INFO Channel 03/0 : 13[7] -> 9[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624947:625739 [4] NCCL INFO Channel 01/0 : 8[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127295:128055 [5] NCCL INFO Channel 06/0 : 11[5] -> 7[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-214:624948:625740 [5] NCCL INFO Channel 02/0 : 13[5] -> 9[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127294:128056 [4] NCCL INFO Channel 04/0 : 10[4] -> 6[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624948:625740 [5] NCCL INFO Channel 03/0 : 13[5] -> 9[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127295:128055 [5] NCCL INFO Channel 07/0 : 11[5] -> 7[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-167-9:1039542:1040319 [5] NCCL INFO Channel 02/0 : 15[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-167-9:1039541:1040318 [4] NCCL INFO Channel 00/0 : 14[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127294:128056 [4] NCCL INFO Channel 05/0 : 10[4] -> 6[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624948:625740 [5] NCCL INFO Channel 02/0 : 9[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127295:128055 [5] NCCL INFO Channel 06/0 : 7[5] -> 3[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127294:128056 [4] NCCL INFO Channel 04/0 : 6[4] -> 2[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-167-9:1039542:1040319 [5] NCCL INFO Channel 03/0 : 15[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-167-9:1039541:1040318 [4] NCCL INFO Channel 01/0 : 14[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624948:625740 [5] NCCL INFO Channel 03/0 : 9[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127295:128055 [5] NCCL INFO Channel 07/0 : 7[5] -> 3[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127294:128056 [4] NCCL INFO Channel 05/0 : 6[4] -> 2[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104433:105193 [4] NCCL INFO Channel 00/0 : 14[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104433:105193 [4] NCCL INFO Channel 01/0 : 14[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624947:625739 [4] NCCL INFO Channel 04/0 : 10[4] -> 8[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104433:105193 [4] NCCL INFO Channel 00/0 : 12[4] -> 10[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762819:763614 [4] NCCL INFO Channel 00/0 : 6[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624947:625739 [4] NCCL INFO Channel 05/0 : 10[4] -> 8[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104433:105193 [4] NCCL INFO Channel 01/0 : 12[4] -> 10[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762819:763614 [4] NCCL INFO Channel 01/0 : 6[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897245:898031 [5] NCCL INFO Channel 02/0 : 13[5] -> 11[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897244:898032 [4] NCCL INFO Channel 00/0 : 12[4] -> 10[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104434:105192 [5] NCCL INFO Channel 02/0 : 15[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762819:763614 [4] NCCL INFO Channel 00/0 : 4[4] -> 2[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-15:127294:128056 [4] NCCL INFO Channel 00/0 : 6[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104433:105193 [4] NCCL INFO Channel 04/0 : 12[4] -> 10[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9629:10391 [4] NCCL INFO Channel 00/0 : 4[4] -> 2[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897244:898032 [4] NCCL INFO Channel 01/0 : 12[4] -> 10[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897245:898031 [5] NCCL INFO Channel 03/0 : 13[5] -> 11[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762819:763614 [4] NCCL INFO Channel 01/0 : 4[4] -> 2[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624948:625740 [5] NCCL INFO Channel 06/0 : 11[5] -> 9[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127295:128055 [5] NCCL INFO Channel 02/0 : 7[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104434:105192 [5] NCCL INFO Channel 03/0 : 15[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104433:105193 [4] NCCL INFO Channel 05/0 : 12[4] -> 10[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9630:10392 [5] NCCL INFO Channel 02/0 : 5[5] -> 3[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897244:898032 [4] NCCL INFO Channel 04/0 : 12[4] -> 10[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762820:763615 [5] NCCL INFO Channel 02/0 : 7[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127294:128056 [4] NCCL INFO Channel 01/0 : 6[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-214:624948:625740 [5] NCCL INFO Channel 07/0 : 11[5] -> 9[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-15:127295:128055 [5] NCCL INFO Channel 03/0 : 7[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104434:105192 [5] NCCL INFO Channel 02/0 : 13[5] -> 11[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9629:10391 [4] NCCL INFO Channel 01/0 : 4[4] -> 2[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897245:898031 [5] NCCL INFO Channel 06/0 : 13[5] -> 11[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9630:10392 [5] NCCL INFO Channel 03/0 : 5[5] -> 3[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762819:763614 [4] NCCL INFO Channel 04/0 : 4[4] -> 2[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762820:763615 [5] NCCL INFO Channel 03/0 : 7[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-36:104434:105192 [5] NCCL INFO Channel 03/0 : 13[5] -> 11[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897244:898032 [4] NCCL INFO Channel 05/0 : 12[4] -> 10[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897242:898034 [2] NCCL INFO Connected all rings
+ip-26-0-166-244:897245:898031 [5] NCCL INFO Channel 07/0 : 13[5] -> 11[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9629:10391 [4] NCCL INFO Channel 04/0 : 4[4] -> 2[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897242:898034 [2] NCCL INFO Channel 01/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-165-213:9630:10392 [5] NCCL INFO Channel 06/0 : 5[5] -> 3[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762821:763619 [6] NCCL INFO Connected all rings
+ip-26-0-166-125:762819:763614 [4] NCCL INFO Channel 05/0 : 4[4] -> 2[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-165-213:9629:10391 [4] NCCL INFO Channel 05/0 : 4[4] -> 2[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762821:763619 [6] NCCL INFO Channel 01/0 : 4[6] -> 5[7] via P2P/IPC
+ip-26-0-166-36:104434:105192 [5] NCCL INFO Channel 06/0 : 13[5] -> 11[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762820:763615 [5] NCCL INFO Channel 02/0 : 5[5] -> 3[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897244:898032 [4] NCCL INFO Channel 04/0 : 10[4] -> 8[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897245:898031 [5] NCCL INFO Channel 06/0 : 11[5] -> 9[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9630:10392 [5] NCCL INFO Channel 07/0 : 5[5] -> 3[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9629:10391 [4] NCCL INFO Channel 04/0 : 2[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-36:104434:105192 [5] NCCL INFO Channel 07/0 : 13[5] -> 11[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762820:763615 [5] NCCL INFO Channel 03/0 : 5[5] -> 3[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-244:897244:898032 [4] NCCL INFO Channel 05/0 : 10[4] -> 8[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-244:897245:898031 [5] NCCL INFO Channel 07/0 : 11[5] -> 9[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9630:10392 [5] NCCL INFO Channel 06/0 : 3[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9629:10391 [4] NCCL INFO Channel 05/0 : 2[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-166-125:762820:763615 [5] NCCL INFO Channel 06/0 : 5[5] -> 3[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-165-213:9630:10392 [5] NCCL INFO Channel 07/0 : 3[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-166-125:762820:763615 [5] NCCL INFO Channel 07/0 : 5[5] -> 3[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-167-9:1039542:1040319 [5] NCCL INFO Channel 00/0 : 15[5] -> 14[4] via P2P/IPC
+ip-26-0-166-15:127295:128055 [5] NCCL INFO Channel 00/0 : 7[5] -> 6[4] via P2P/IPC
+ip-26-0-166-214:624948:625740 [5] NCCL INFO Channel 00/0 : 9[5] -> 8[4] via P2P/IPC
+ip-26-0-162-46:702885:703680 [5] NCCL INFO Channel 00/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-166-125:762820:763615 [5] NCCL INFO Channel 00/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-213:9630:10392 [5] NCCL INFO Channel 00/0 : 3[5] -> 2[4] via P2P/IPC
+ip-26-0-166-244:897242:898034 [2] NCCL INFO Channel 03/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-166-244:897242:898034 [2] NCCL INFO Channel 05/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-166-15:127295:128055 [5] NCCL INFO Channel 02/0 : 7[5] -> 6[4] via P2P/IPC
+ip-26-0-166-244:897242:898034 [2] NCCL INFO Channel 07/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-166-125:762821:763619 [6] NCCL INFO Channel 03/0 : 4[6] -> 5[7] via P2P/IPC
+ip-26-0-166-244:897242:898034 [2] NCCL INFO Channel 04/0 : 8[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897242:898034 [2] NCCL INFO Channel 05/0 : 8[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9630:10392 [5] NCCL INFO Channel 02/0 : 3[5] -> 2[4] via P2P/IPC
+ip-26-0-166-244:897243:898033 [3] NCCL INFO Channel 06/0 : 9[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897242:898034 [2] NCCL INFO Channel 00/0 : 10[2] -> 12[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897243:898033 [3] NCCL INFO Channel 07/0 : 9[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897242:898034 [2] NCCL INFO Channel 01/0 : 10[2] -> 12[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897242:898034 [2] NCCL INFO Channel 04/0 : 10[2] -> 12[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897243:898033 [3] NCCL INFO Channel 02/0 : 11[3] -> 13[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897242:898034 [2] NCCL INFO Channel 05/0 : 10[2] -> 12[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897243:898033 [3] NCCL INFO Channel 03/0 : 11[3] -> 13[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897243:898033 [3] NCCL INFO Channel 06/0 : 11[3] -> 13[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897243:898033 [3] NCCL INFO Channel 07/0 : 11[3] -> 13[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104432:105194 [3] NCCL INFO Connected all rings
+ip-26-0-166-214:624945:625741 [2] NCCL INFO Channel 00/0 : 4[2] -> 8[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624945:625741 [2] NCCL INFO Channel 01/0 : 4[2] -> 8[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624946:625742 [3] NCCL INFO Channel 02/0 : 5[3] -> 9[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624945:625741 [2] NCCL INFO Channel 00/0 : 8[2] -> 12[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624946:625742 [3] NCCL INFO Channel 03/0 : 5[3] -> 9[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762821:763619 [6] NCCL INFO Channel 05/0 : 4[6] -> 5[7] via P2P/IPC
+ip-26-0-166-214:624946:625742 [3] NCCL INFO Channel 02/0 : 9[3] -> 13[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624945:625741 [2] NCCL INFO Channel 01/0 : 8[2] -> 12[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624946:625742 [3] NCCL INFO Channel 03/0 : 9[3] -> 13[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624948:625740 [5] NCCL INFO Channel 02/0 : 9[5] -> 8[4] via P2P/IPC
+ip-26-0-166-125:762820:763615 [5] NCCL INFO Channel 02/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-167-9:1039542:1040319 [5] NCCL INFO Channel 02/0 : 15[5] -> 14[4] via P2P/IPC
+ip-26-0-166-125:762821:763619 [6] NCCL INFO Channel 07/0 : 4[6] -> 5[7] via P2P/IPC
+ip-26-0-166-36:104434:105192 [5] NCCL INFO Channel 00/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-166-244:897245:898031 [5] NCCL INFO Channel 00/0 : 11[5] -> 10[4] via P2P/IPC
+ip-26-0-166-125:762820:763615 [5] NCCL INFO Channel 04/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-167-9:1039542:1040319 [5] NCCL INFO Channel 04/0 : 15[5] -> 14[4] via P2P/IPC
+ip-26-0-166-244:897245:898031 [5] NCCL INFO Channel 02/0 : 11[5] -> 10[4] via P2P/IPC
+ip-26-0-166-36:104434:105192 [5] NCCL INFO Channel 02/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-166-36:104432:105194 [3] NCCL INFO Channel 02/0 : 11[3] -> 13[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897245:898031 [5] NCCL INFO Channel 04/0 : 11[5] -> 10[4] via P2P/IPC
+ip-26-0-167-9:1039542:1040319 [5] NCCL INFO Channel 06/0 : 15[5] -> 14[4] via P2P/IPC
+ip-26-0-166-36:104432:105194 [3] NCCL INFO Channel 03/0 : 11[3] -> 13[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624948:625740 [5] NCCL INFO Channel 04/0 : 9[5] -> 8[4] via P2P/IPC
+ip-26-0-166-36:104432:105194 [3] NCCL INFO Channel 06/0 : 11[3] -> 13[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104432:105194 [3] NCCL INFO Channel 07/0 : 11[3] -> 13[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897245:898031 [5] NCCL INFO Channel 06/0 : 11[5] -> 10[4] via P2P/IPC
+ip-26-0-166-36:104432:105194 [3] NCCL INFO Channel 02/0 : 13[3] -> 15[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-46:702885:703680 [5] NCCL INFO Channel 02/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-166-36:104434:105192 [5] NCCL INFO Channel 04/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-166-15:127297:128060 [7] NCCL INFO Connected all rings
+ip-26-0-166-36:104432:105194 [3] NCCL INFO Channel 03/0 : 13[3] -> 15[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762820:763615 [5] NCCL INFO Channel 06/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-165-213:9630:10392 [5] NCCL INFO Channel 04/0 : 3[5] -> 2[4] via P2P/IPC
+ip-26-0-166-36:104434:105192 [5] NCCL INFO Channel 06/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-166-244:897243:898033 [3] NCCL INFO Channel 06/0 : 7[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897243:898033 [3] NCCL INFO Channel 07/0 : 7[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127295:128055 [5] NCCL INFO Channel 04/0 : 7[5] -> 6[4] via P2P/IPC
+ip-26-0-166-125:762822:763618 [7] NCCL INFO Channel 02/0 : 3[7] -> 5[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762821:763619 [6] NCCL INFO Channel 00/0 : 2[6] -> 4[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9630:10392 [5] NCCL INFO Channel 06/0 : 3[5] -> 2[4] via P2P/IPC
+ip-26-0-166-125:762822:763618 [7] NCCL INFO Channel 03/0 : 3[7] -> 5[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762821:763619 [6] NCCL INFO Channel 01/0 : 2[6] -> 4[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762822:763618 [7] NCCL INFO Channel 06/0 : 3[7] -> 5[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762821:763619 [6] NCCL INFO Channel 04/0 : 2[6] -> 4[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762822:763618 [7] NCCL INFO Channel 07/0 : 3[7] -> 5[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762821:763619 [6] NCCL INFO Channel 05/0 : 2[6] -> 4[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762822:763618 [7] NCCL INFO Channel 02/0 : 5[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762821:763619 [6] NCCL INFO Channel 00/0 : 4[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762822:763618 [7] NCCL INFO Channel 03/0 : 5[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762821:763619 [6] NCCL INFO Channel 01/0 : 4[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762817:763617 [2] NCCL INFO Connected all rings
+ip-26-0-166-125:762817:763617 [2] NCCL INFO Channel 01/0 : 4[2] -> 5[3] via P2P/IPC
+ip-26-0-162-46:702885:703680 [5] NCCL INFO Channel 04/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-166-36:104431:105195 [2] NCCL INFO Channel 00/0 : 10[2] -> 12[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104431:105195 [2] NCCL INFO Channel 01/0 : 10[2] -> 12[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104431:105195 [2] NCCL INFO Channel 04/0 : 10[2] -> 12[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104431:105195 [2] NCCL INFO Channel 05/0 : 10[2] -> 12[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104431:105195 [2] NCCL INFO Channel 00/0 : 12[2] -> 14[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104431:105195 [2] NCCL INFO Channel 01/0 : 12[2] -> 14[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762817:763617 [2] NCCL INFO Channel 03/0 : 4[2] -> 5[3] via P2P/IPC
+ip-26-0-166-244:897242:898034 [2] NCCL INFO Channel 04/0 : 6[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897242:898034 [2] NCCL INFO Channel 05/0 : 6[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127295:128055 [5] NCCL INFO Channel 06/0 : 7[5] -> 6[4] via P2P/IPC
+ip-26-0-165-213:9631:10396 [6] NCCL INFO Connected all rings
+ip-26-0-165-213:9631:10396 [6] NCCL INFO Channel 01/0 : 2[6] -> 3[7] via P2P/IPC
+ip-26-0-166-125:762817:763617 [2] NCCL INFO Channel 05/0 : 4[2] -> 5[3] via P2P/IPC
+ip-26-0-166-15:127292:128057 [2] NCCL INFO Channel 03/0 : 6[2] -> 7[3] via P2P/IPC
+ip-26-0-166-214:624948:625740 [5] NCCL INFO Channel 06/0 : 9[5] -> 8[4] via P2P/IPC
+ip-26-0-162-46:702885:703680 [5] NCCL INFO Channel 06/0 : 1[5] -> 0[4] via P2P/IPC
+ip-26-0-165-213:9631:10396 [6] NCCL INFO Channel 03/0 : 2[6] -> 3[7] via P2P/IPC
+ip-26-0-166-15:127292:128057 [2] NCCL INFO Channel 05/0 : 6[2] -> 7[3] via P2P/IPC
+ip-26-0-166-15:127297:128060 [7] NCCL INFO Channel 02/0 : 5[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127296:128059 [6] NCCL INFO Channel 00/0 : 4[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127297:128060 [7] NCCL INFO Channel 03/0 : 5[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127296:128059 [6] NCCL INFO Channel 01/0 : 4[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127297:128060 [7] NCCL INFO Channel 06/0 : 3[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127296:128059 [6] NCCL INFO Channel 04/0 : 2[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127297:128060 [7] NCCL INFO Channel 07/0 : 3[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127296:128059 [6] NCCL INFO Channel 05/0 : 2[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127297:128060 [7] NCCL INFO Channel 06/0 : 7[7] -> 11[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127296:128059 [6] NCCL INFO Channel 04/0 : 6[6] -> 10[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127297:128060 [7] NCCL INFO Channel 07/0 : 7[7] -> 11[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127296:128059 [6] NCCL INFO Channel 05/0 : 6[6] -> 10[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897247:898035 [7] NCCL INFO Channel 06/0 : 11[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897247:898035 [7] NCCL INFO Channel 07/0 : 11[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897246:898036 [6] NCCL INFO Channel 04/0 : 10[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897246:898036 [6] NCCL INFO Channel 05/0 : 10[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127292:128057 [2] NCCL INFO Channel 07/0 : 6[2] -> 7[3] via P2P/IPC
+ip-26-0-165-213:9631:10396 [6] NCCL INFO Channel 05/0 : 2[6] -> 3[7] via P2P/IPC
+ip-26-0-166-15:127293:128058 [3] NCCL INFO Connected all rings
+ip-26-0-166-125:762817:763617 [2] NCCL INFO Channel 07/0 : 4[2] -> 5[3] via P2P/IPC
+ip-26-0-165-213:9631:10396 [6] NCCL INFO Channel 07/0 : 2[6] -> 3[7] via P2P/IPC
+ip-26-0-165-213:9631:10396 [6] NCCL INFO Channel 04/0 : 0[6] -> 2[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9632:10395 [7] NCCL INFO Channel 06/0 : 1[7] -> 3[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9631:10396 [6] NCCL INFO Channel 05/0 : 0[6] -> 2[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9632:10395 [7] NCCL INFO Channel 07/0 : 1[7] -> 3[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9631:10396 [6] NCCL INFO Channel 00/0 : 2[6] -> 4[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9632:10395 [7] NCCL INFO Channel 02/0 : 3[7] -> 5[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9631:10396 [6] NCCL INFO Channel 01/0 : 2[6] -> 4[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9632:10395 [7] NCCL INFO Channel 03/0 : 3[7] -> 5[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9631:10396 [6] NCCL INFO Channel 04/0 : 2[6] -> 4[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9632:10395 [7] NCCL INFO Channel 06/0 : 3[7] -> 5[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9631:10396 [6] NCCL INFO Channel 05/0 : 2[6] -> 4[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9632:10395 [7] NCCL INFO Channel 07/0 : 3[7] -> 5[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762821:763619 [6] NCCL INFO Channel 00/0 : 4[6] -> 8[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762822:763618 [7] NCCL INFO Channel 02/0 : 5[7] -> 9[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762821:763619 [6] NCCL INFO Channel 01/0 : 4[6] -> 8[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762822:763618 [7] NCCL INFO Channel 03/0 : 5[7] -> 9[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762821:763619 [6] NCCL INFO Channel 00/0 : 8[6] -> 4[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762822:763618 [7] NCCL INFO Channel 02/0 : 9[7] -> 5[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762821:763619 [6] NCCL INFO Channel 01/0 : 8[6] -> 4[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762822:763618 [7] NCCL INFO Channel 03/0 : 9[7] -> 5[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624950:625743 [7] NCCL INFO Channel 02/0 : 1[7] -> 9[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624950:625743 [7] NCCL INFO Channel 03/0 : 1[7] -> 9[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624950:625743 [7] NCCL INFO Channel 02/0 : 9[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624949:625744 [6] NCCL INFO Channel 00/0 : 0[6] -> 8[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624950:625743 [7] NCCL INFO Channel 03/0 : 9[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-46:702887:703686 [7] NCCL INFO Connected all rings
+ip-26-0-166-214:624949:625744 [6] NCCL INFO Channel 01/0 : 0[6] -> 8[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624949:625744 [6] NCCL INFO Channel 00/0 : 8[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624949:625744 [6] NCCL INFO Channel 01/0 : 8[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127293:128058 [3] NCCL INFO Channel 02/0 : 5[3] -> 7[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127292:128057 [2] NCCL INFO Channel 00/0 : 4[2] -> 6[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127293:128058 [3] NCCL INFO Channel 03/0 : 5[3] -> 7[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127292:128057 [2] NCCL INFO Channel 01/0 : 4[2] -> 6[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-167-9:1039539:1040321 [2] NCCL INFO Connected all rings
+ip-26-0-167-9:1039539:1040321 [2] NCCL INFO Channel 01/0 : 14[2] -> 15[3] via P2P/IPC
+ip-26-0-167-9:1039539:1040321 [2] NCCL INFO Channel 03/0 : 14[2] -> 15[3] via P2P/IPC
+ip-26-0-167-9:1039539:1040321 [2] NCCL INFO Channel 05/0 : 14[2] -> 15[3] via P2P/IPC
+ip-26-0-167-9:1039539:1040321 [2] NCCL INFO Channel 07/0 : 14[2] -> 15[3] via P2P/IPC
+ip-26-0-167-9:1039539:1040321 [2] NCCL INFO Channel 00/0 : 12[2] -> 14[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-167-9:1039539:1040321 [2] NCCL INFO Channel 01/0 : 12[2] -> 14[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-167-9:1039540:1040320 [3] NCCL INFO Channel 02/0 : 13[3] -> 15[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-167-9:1039540:1040320 [3] NCCL INFO Channel 03/0 : 13[3] -> 15[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-167-9:1039539:1040321 [2] NCCL INFO Channel 04/0 : 6[2] -> 14[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-167-9:1039539:1040321 [2] NCCL INFO Channel 05/0 : 6[2] -> 14[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-167-9:1039540:1040320 [3] NCCL INFO Channel 06/0 : 7[3] -> 15[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-167-9:1039539:1040321 [2] NCCL INFO Channel 04/0 : 14[2] -> 6[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-167-9:1039540:1040320 [3] NCCL INFO Channel 07/0 : 7[3] -> 15[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-167-9:1039539:1040321 [2] NCCL INFO Channel 05/0 : 14[2] -> 6[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-167-9:1039540:1040320 [3] NCCL INFO Channel 06/0 : 15[3] -> 7[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-167-9:1039540:1040320 [3] NCCL INFO Channel 07/0 : 15[3] -> 7[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104431:105195 [2] NCCL INFO Channel 00/0 : 8[2] -> 12[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104431:105195 [2] NCCL INFO Channel 01/0 : 8[2] -> 12[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104431:105195 [2] NCCL INFO Channel 00/0 : 12[2] -> 8[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104432:105194 [3] NCCL INFO Channel 02/0 : 9[3] -> 13[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762818:763616 [3] NCCL INFO Connected all rings
+ip-26-0-166-36:104431:105195 [2] NCCL INFO Channel 01/0 : 12[2] -> 8[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104432:105194 [3] NCCL INFO Channel 03/0 : 9[3] -> 13[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104432:105194 [3] NCCL INFO Channel 02/0 : 13[3] -> 9[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104432:105194 [3] NCCL INFO Channel 03/0 : 13[3] -> 9[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-46:702887:703686 [7] NCCL INFO Channel 06/0 : 1[7] -> 3[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-46:702886:703685 [6] NCCL INFO Channel 04/0 : 0[6] -> 2[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-46:702887:703686 [7] NCCL INFO Channel 07/0 : 1[7] -> 3[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-46:702886:703685 [6] NCCL INFO Channel 05/0 : 0[6] -> 2[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-46:702887:703686 [7] NCCL INFO Channel 02/0 : 9[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-46:702886:703685 [6] NCCL INFO Channel 00/0 : 8[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-46:702887:703686 [7] NCCL INFO Channel 03/0 : 9[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-46:702886:703685 [6] NCCL INFO Channel 01/0 : 8[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9632:10395 [7] NCCL INFO Channel 06/0 : 3[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9631:10396 [6] NCCL INFO Channel 04/0 : 2[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-46:702887:703686 [7] NCCL INFO Channel 02/0 : 1[7] -> 9[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-46:702886:703685 [6] NCCL INFO Channel 00/0 : 0[6] -> 8[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9632:10395 [7] NCCL INFO Channel 07/0 : 3[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-46:702887:703686 [7] NCCL INFO Channel 03/0 : 1[7] -> 9[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9631:10396 [6] NCCL INFO Channel 05/0 : 2[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-46:702886:703685 [6] NCCL INFO Channel 01/0 : 0[6] -> 8[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9632:10395 [7] NCCL INFO Channel 06/0 : 7[7] -> 3[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9631:10396 [6] NCCL INFO Channel 04/0 : 6[6] -> 2[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-46:702887:703686 [7] NCCL INFO Channel 06/0 : 3[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-46:702886:703685 [6] NCCL INFO Channel 04/0 : 2[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9632:10395 [7] NCCL INFO Channel 07/0 : 7[7] -> 3[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9631:10396 [6] NCCL INFO Channel 05/0 : 6[6] -> 2[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-162-46:702887:703686 [7] NCCL INFO Channel 07/0 : 3[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-46:702886:703685 [6] NCCL INFO Channel 05/0 : 2[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624950:625743 [7] NCCL INFO Channel 02/0 : 13[7] -> 9[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624950:625743 [7] NCCL INFO Channel 03/0 : 13[7] -> 9[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624950:625743 [7] NCCL INFO Channel 02/0 : 9[7] -> 5[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624950:625743 [7] NCCL INFO Channel 03/0 : 9[7] -> 5[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9627:10393 [2] NCCL INFO Connected all rings
+ip-26-0-165-213:9627:10393 [2] NCCL INFO Channel 01/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-125:762818:763616 [3] NCCL INFO Channel 02/0 : 3[3] -> 5[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624949:625744 [6] NCCL INFO Channel 00/0 : 12[6] -> 8[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762817:763617 [2] NCCL INFO Channel 00/0 : 2[2] -> 4[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104436:105197 [7] NCCL INFO Channel 02/0 : 15[7] -> 13[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762818:763616 [3] NCCL INFO Channel 03/0 : 3[3] -> 5[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624949:625744 [6] NCCL INFO Channel 01/0 : 12[6] -> 8[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762817:763617 [2] NCCL INFO Channel 01/0 : 2[2] -> 4[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104436:105197 [7] NCCL INFO Channel 03/0 : 15[7] -> 13[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624950:625743 [7] NCCL INFO Channel 06/0 : 11[7] -> 9[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762818:763616 [3] NCCL INFO Channel 06/0 : 3[3] -> 5[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762817:763617 [2] NCCL INFO Channel 04/0 : 2[2] -> 4[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624949:625744 [6] NCCL INFO Channel 00/0 : 8[6] -> 4[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104436:105197 [7] NCCL INFO Channel 02/0 : 13[7] -> 11[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762818:763616 [3] NCCL INFO Channel 07/0 : 3[3] -> 5[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762822:763618 [7] NCCL INFO Channel 02/0 : 7[7] -> 5[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762817:763617 [2] NCCL INFO Channel 05/0 : 2[2] -> 4[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624950:625743 [7] NCCL INFO Channel 07/0 : 11[7] -> 9[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-214:624949:625744 [6] NCCL INFO Channel 01/0 : 8[6] -> 4[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104436:105197 [7] NCCL INFO Channel 03/0 : 13[7] -> 11[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762818:763616 [3] NCCL INFO Channel 02/0 : 5[3] -> 7[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104436:105197 [7] NCCL INFO Channel 06/0 : 13[7] -> 11[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762822:763618 [7] NCCL INFO Channel 03/0 : 7[7] -> 5[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762817:763617 [2] NCCL INFO Channel 00/0 : 4[2] -> 6[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762818:763616 [3] NCCL INFO Channel 03/0 : 5[3] -> 7[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762822:763618 [7] NCCL INFO Channel 02/0 : 5[7] -> 3[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762817:763617 [2] NCCL INFO Channel 01/0 : 4[2] -> 6[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104436:105197 [7] NCCL INFO Channel 07/0 : 13[7] -> 11[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762822:763618 [7] NCCL INFO Channel 03/0 : 5[7] -> 3[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104435:105196 [6] NCCL INFO Channel 00/0 : 14[6] -> 12[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762822:763618 [7] NCCL INFO Channel 06/0 : 5[7] -> 3[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104435:105196 [6] NCCL INFO Channel 01/0 : 14[6] -> 12[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624949:625744 [6] NCCL INFO Channel 04/0 : 10[6] -> 8[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762822:763618 [7] NCCL INFO Channel 07/0 : 5[7] -> 3[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-36:104435:105196 [6] NCCL INFO Channel 00/0 : 12[6] -> 10[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-214:624949:625744 [6] NCCL INFO Channel 05/0 : 10[6] -> 8[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762821:763619 [6] NCCL INFO Channel 00/0 : 6[6] -> 4[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104435:105196 [6] NCCL INFO Channel 01/0 : 12[6] -> 10[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127293:128058 [3] NCCL INFO Channel 06/0 : 3[3] -> 7[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762821:763619 [6] NCCL INFO Channel 01/0 : 6[6] -> 4[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104435:105196 [6] NCCL INFO Channel 04/0 : 12[6] -> 10[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127292:128057 [2] NCCL INFO Channel 04/0 : 2[2] -> 6[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762821:763619 [6] NCCL INFO Channel 00/0 : 4[6] -> 2[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127293:128058 [3] NCCL INFO Channel 07/0 : 3[3] -> 7[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127292:128057 [2] NCCL INFO Channel 05/0 : 2[2] -> 6[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104435:105196 [6] NCCL INFO Channel 05/0 : 12[6] -> 10[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-125:762821:763619 [6] NCCL INFO Channel 01/0 : 4[6] -> 2[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127293:128058 [3] NCCL INFO Channel 06/0 : 7[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762821:763619 [6] NCCL INFO Channel 04/0 : 4[6] -> 2[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127292:128057 [2] NCCL INFO Channel 04/0 : 6[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762821:763619 [6] NCCL INFO Channel 05/0 : 4[6] -> 2[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127293:128058 [3] NCCL INFO Channel 07/0 : 7[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127292:128057 [2] NCCL INFO Channel 05/0 : 6[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9627:10393 [2] NCCL INFO Channel 03/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-244:897242:898034 [2] NCCL INFO Channel 04/0 : 10[2] -> 6[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897243:898033 [3] NCCL INFO Channel 06/0 : 11[3] -> 7[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897242:898034 [2] NCCL INFO Channel 05/0 : 10[2] -> 6[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897243:898033 [3] NCCL INFO Channel 07/0 : 11[3] -> 7[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9627:10393 [2] NCCL INFO Channel 05/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-15:127296:128059 [6] NCCL INFO Channel 04/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127297:128060 [7] NCCL INFO Channel 06/0 : 15[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9627:10393 [2] NCCL INFO Channel 07/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-166-15:127296:128059 [6] NCCL INFO Channel 05/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127297:128060 [7] NCCL INFO Channel 07/0 : 15[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127296:128059 [6] NCCL INFO Channel 04/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127296:128059 [6] NCCL INFO Channel 05/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127297:128060 [7] NCCL INFO Channel 06/0 : 7[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127297:128060 [7] NCCL INFO Channel 07/0 : 7[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127296:128059 [6] NCCL INFO Channel 04/0 : 10[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127296:128059 [6] NCCL INFO Channel 05/0 : 10[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127297:128060 [7] NCCL INFO Channel 06/0 : 11[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127296:128059 [6] NCCL INFO Channel 04/0 : 6[6] -> 2[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127297:128060 [7] NCCL INFO Channel 07/0 : 11[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127297:128060 [7] NCCL INFO Channel 06/0 : 7[7] -> 3[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127296:128059 [6] NCCL INFO Channel 05/0 : 6[6] -> 2[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-15:127297:128060 [7] NCCL INFO Channel 07/0 : 7[7] -> 3[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-162-46:702883:703683 [3] NCCL INFO Connected all rings
+ip-26-0-165-213:9628:10394 [3] NCCL INFO Channel 06/0 : 1[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897246:898036 [6] NCCL INFO Channel 00/0 : 12[6] -> 10[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897247:898035 [7] NCCL INFO Channel 02/0 : 13[7] -> 11[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-15:127296:128059 [6] NCCL INFO Channel 00/0 : 6[6] -> 4[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9627:10393 [2] NCCL INFO Channel 04/0 : 0[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9628:10394 [3] NCCL INFO Channel 07/0 : 1[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127296:128059 [6] NCCL INFO Channel 01/0 : 6[6] -> 4[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9631:10396 [6] NCCL INFO Channel 00/0 : 4[6] -> 2[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897247:898035 [7] NCCL INFO Channel 03/0 : 13[7] -> 11[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897246:898036 [6] NCCL INFO Channel 01/0 : 12[6] -> 10[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9628:10394 [3] NCCL INFO Channel 02/0 : 3[3] -> 5[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9627:10393 [2] NCCL INFO Channel 05/0 : 0[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127297:128060 [7] NCCL INFO Channel 02/0 : 7[7] -> 5[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897247:898035 [7] NCCL INFO Channel 06/0 : 13[7] -> 11[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9631:10396 [6] NCCL INFO Channel 01/0 : 4[6] -> 2[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9628:10394 [3] NCCL INFO Channel 03/0 : 3[3] -> 5[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127297:128060 [7] NCCL INFO Channel 03/0 : 7[7] -> 5[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9632:10395 [7] NCCL INFO Channel 02/0 : 5[7] -> 3[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9627:10393 [2] NCCL INFO Channel 00/0 : 2[2] -> 4[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897246:898036 [6] NCCL INFO Channel 04/0 : 12[6] -> 10[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-244:897247:898035 [7] NCCL INFO Channel 07/0 : 13[7] -> 11[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897246:898036 [6] NCCL INFO Channel 05/0 : 12[6] -> 10[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9631:10396 [6] NCCL INFO Channel 04/0 : 4[6] -> 2[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9628:10394 [3] NCCL INFO Channel 06/0 : 3[3] -> 5[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9632:10395 [7] NCCL INFO Channel 03/0 : 5[7] -> 3[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9627:10393 [2] NCCL INFO Channel 01/0 : 2[2] -> 4[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127297:128060 [7] NCCL INFO Channel 00/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-166-244:897247:898035 [7] NCCL INFO Channel 06/0 : 11[7] -> 9[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897246:898036 [6] NCCL INFO Channel 04/0 : 10[6] -> 8[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9631:10396 [6] NCCL INFO Channel 05/0 : 4[6] -> 2[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9628:10394 [3] NCCL INFO Channel 07/0 : 3[3] -> 5[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9632:10395 [7] NCCL INFO Channel 06/0 : 5[7] -> 3[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897247:898035 [7] NCCL INFO Channel 07/0 : 11[7] -> 9[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9631:10396 [6] NCCL INFO Channel 04/0 : 2[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9627:10393 [2] NCCL INFO Channel 04/0 : 2[2] -> 4[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9632:10395 [7] NCCL INFO Channel 07/0 : 5[7] -> 3[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-244:897246:898036 [6] NCCL INFO Channel 05/0 : 10[6] -> 8[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9631:10396 [6] NCCL INFO Channel 05/0 : 2[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-165-213:9627:10393 [2] NCCL INFO Channel 05/0 : 2[2] -> 4[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9632:10395 [7] NCCL INFO Channel 06/0 : 3[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-165-213:9632:10395 [7] NCCL INFO Channel 07/0 : 3[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-167-9:1039544:1040322 [7] NCCL INFO Channel 02/0 : 15[7] -> 13[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-167-9:1039544:1040322 [7] NCCL INFO Channel 03/0 : 15[7] -> 13[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-166-125:762818:763616 [3] NCCL INFO Channel 02/0 : 5[3] -> 9[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762822:763618 [7] NCCL INFO Channel 00/0 : 5[7] -> 4[6] via P2P/IPC
+ip-26-0-166-125:762818:763616 [3] NCCL INFO Channel 03/0 : 5[3] -> 9[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127297:128060 [7] NCCL INFO Channel 02/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-165-213:9632:10395 [7] NCCL INFO Channel 00/0 : 3[7] -> 2[6] via P2P/IPC
+ip-26-0-166-125:762817:763617 [2] NCCL INFO Channel 00/0 : 4[2] -> 8[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-167-9:1039544:1040322 [7] NCCL INFO Channel 00/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-162-46:702887:703686 [7] NCCL INFO Channel 00/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-166-125:762817:763617 [2] NCCL INFO Channel 01/0 : 4[2] -> 8[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762818:763616 [3] NCCL INFO Channel 02/0 : 9[3] -> 5[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104436:105197 [7] NCCL INFO Channel 00/0 : 13[7] -> 12[6] via P2P/IPC
+ip-26-0-166-125:762818:763616 [3] NCCL INFO Channel 03/0 : 9[3] -> 5[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624946:625742 [3] NCCL INFO Channel 02/0 : 1[3] -> 9[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762817:763617 [2] NCCL INFO Channel 00/0 : 8[2] -> 4[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-167-9:1039544:1040322 [7] NCCL INFO Channel 02/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-166-125:762817:763617 [2] NCCL INFO Channel 01/0 : 8[2] -> 4[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-167-9:1039543:1040323 [6] NCCL INFO Channel 00/0 : 14[6] -> 12[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-167-9:1039543:1040323 [6] NCCL INFO Channel 01/0 : 14[6] -> 12[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-166-36:104436:105197 [7] NCCL INFO Channel 02/0 : 13[7] -> 12[6] via P2P/IPC
+ip-26-0-167-9:1039544:1040322 [7] NCCL INFO Channel 04/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-166-15:127297:128060 [7] NCCL INFO Channel 04/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-162-46:702887:703686 [7] NCCL INFO Channel 02/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-165-213:9632:10395 [7] NCCL INFO Channel 02/0 : 3[7] -> 2[6] via P2P/IPC
+ip-26-0-166-244:897247:898035 [7] NCCL INFO Channel 00/0 : 11[7] -> 10[6] via P2P/IPC
+ip-26-0-166-214:624950:625743 [7] NCCL INFO Channel 00/0 : 9[7] -> 8[6] via P2P/IPC
+ip-26-0-167-9:1039544:1040322 [7] NCCL INFO Channel 06/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-166-15:127297:128060 [7] NCCL INFO Channel 06/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-166-125:762822:763618 [7] NCCL INFO Channel 02/0 : 5[7] -> 4[6] via P2P/IPC
+ip-26-0-166-214:624946:625742 [3] NCCL INFO Channel 03/0 : 1[3] -> 9[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624946:625742 [3] NCCL INFO Channel 02/0 : 9[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624946:625742 [3] NCCL INFO Channel 03/0 : 9[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104436:105197 [7] NCCL INFO Channel 04/0 : 13[7] -> 12[6] via P2P/IPC
+ip-26-0-162-46:702887:703686 [7] NCCL INFO Channel 04/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-165-213:9632:10395 [7] NCCL INFO Channel 04/0 : 3[7] -> 2[6] via P2P/IPC
+ip-26-0-166-244:897247:898035 [7] NCCL INFO Channel 02/0 : 11[7] -> 10[6] via P2P/IPC
+ip-26-0-162-46:702882:703682 [2] NCCL INFO Channel 04/0 : 0[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-46:702883:703683 [3] NCCL INFO Channel 06/0 : 1[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-46:702882:703682 [2] NCCL INFO Channel 05/0 : 0[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-46:702883:703683 [3] NCCL INFO Channel 07/0 : 1[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762822:763618 [7] NCCL INFO Channel 04/0 : 5[7] -> 4[6] via P2P/IPC
+ip-26-0-166-214:624945:625741 [2] NCCL INFO Channel 00/0 : 0[2] -> 8[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-46:702883:703683 [3] NCCL INFO Channel 02/0 : 9[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624950:625743 [7] NCCL INFO Channel 02/0 : 9[7] -> 8[6] via P2P/IPC
+ip-26-0-162-46:702882:703682 [2] NCCL INFO Channel 00/0 : 8[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-46:702883:703683 [3] NCCL INFO Channel 03/0 : 9[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-46:702882:703682 [2] NCCL INFO Channel 01/0 : 8[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9628:10394 [3] NCCL INFO Channel 06/0 : 3[3] -> 7[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9627:10393 [2] NCCL INFO Channel 04/0 : 2[2] -> 6[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624945:625741 [2] NCCL INFO Channel 01/0 : 0[2] -> 8[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-46:702883:703683 [3] NCCL INFO Channel 02/0 : 1[3] -> 9[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624945:625741 [2] NCCL INFO Channel 00/0 : 8[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-46:702882:703682 [2] NCCL INFO Channel 00/0 : 0[2] -> 8[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624945:625741 [2] NCCL INFO Channel 01/0 : 8[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104436:105197 [7] NCCL INFO Channel 06/0 : 13[7] -> 12[6] via P2P/IPC
+ip-26-0-165-213:9628:10394 [3] NCCL INFO Channel 07/0 : 3[3] -> 7[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-46:702883:703683 [3] NCCL INFO Channel 03/0 : 1[3] -> 9[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-46:702882:703682 [2] NCCL INFO Channel 01/0 : 0[2] -> 8[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9627:10393 [2] NCCL INFO Channel 05/0 : 2[2] -> 6[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9632:10395 [7] NCCL INFO Channel 06/0 : 3[7] -> 2[6] via P2P/IPC
+ip-26-0-166-214:624950:625743 [7] NCCL INFO Channel 04/0 : 9[7] -> 8[6] via P2P/IPC
+ip-26-0-162-46:702883:703683 [3] NCCL INFO Channel 06/0 : 3[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9628:10394 [3] NCCL INFO Channel 06/0 : 7[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9627:10393 [2] NCCL INFO Channel 04/0 : 6[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-46:702882:703682 [2] NCCL INFO Channel 04/0 : 2[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624946:625742 [3] NCCL INFO Channel 02/0 : 13[3] -> 9[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127293:128058 [3] NCCL INFO Channel 06/0 : 15[3] -> 7[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127292:128057 [2] NCCL INFO Channel 04/0 : 14[2] -> 6[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-162-46:702883:703683 [3] NCCL INFO Channel 07/0 : 3[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9628:10394 [3] NCCL INFO Channel 07/0 : 7[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9627:10393 [2] NCCL INFO Channel 05/0 : 6[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897247:898035 [7] NCCL INFO Channel 04/0 : 11[7] -> 10[6] via P2P/IPC
+ip-26-0-166-214:624945:625741 [2] NCCL INFO Channel 00/0 : 12[2] -> 8[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624946:625742 [3] NCCL INFO Channel 03/0 : 13[3] -> 9[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-162-46:702882:703682 [2] NCCL INFO Channel 05/0 : 2[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127293:128058 [3] NCCL INFO Channel 07/0 : 15[3] -> 7[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127292:128057 [2] NCCL INFO Channel 05/0 : 14[2] -> 6[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624945:625741 [2] NCCL INFO Channel 01/0 : 12[2] -> 8[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624946:625742 [3] NCCL INFO Channel 02/0 : 9[3] -> 5[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127293:128058 [3] NCCL INFO Channel 06/0 : 7[3] -> 15[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127292:128057 [2] NCCL INFO Channel 04/0 : 6[2] -> 14[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762822:763618 [7] NCCL INFO Channel 06/0 : 5[7] -> 4[6] via P2P/IPC
+ip-26-0-166-214:624946:625742 [3] NCCL INFO Channel 03/0 : 9[3] -> 5[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624945:625741 [2] NCCL INFO Channel 00/0 : 8[2] -> 4[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127292:128057 [2] NCCL INFO Channel 05/0 : 6[2] -> 14[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127293:128058 [3] NCCL INFO Channel 07/0 : 7[3] -> 15[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624945:625741 [2] NCCL INFO Channel 01/0 : 8[2] -> 4[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127293:128058 [3] NCCL INFO Channel 06/0 : 11[3] -> 7[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127292:128057 [2] NCCL INFO Channel 04/0 : 10[2] -> 6[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-167-9:1039540:1040320 [3] NCCL INFO Channel 02/0 : 15[3] -> 13[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624946:625742 [3] NCCL INFO Channel 06/0 : 11[3] -> 9[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127293:128058 [3] NCCL INFO Channel 07/0 : 11[3] -> 7[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897247:898035 [7] NCCL INFO Channel 06/0 : 11[7] -> 10[6] via P2P/IPC
+ip-26-0-167-9:1039539:1040321 [2] NCCL INFO Channel 00/0 : 14[2] -> 12[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762818:763616 [3] NCCL INFO Channel 02/0 : 7[3] -> 5[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624950:625743 [7] NCCL INFO Channel 06/0 : 9[7] -> 8[6] via P2P/IPC
+ip-26-0-162-46:702887:703686 [7] NCCL INFO Channel 06/0 : 1[7] -> 0[6] via P2P/IPC
+ip-26-0-166-15:127292:128057 [2] NCCL INFO Channel 05/0 : 10[2] -> 6[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-167-9:1039540:1040320 [3] NCCL INFO Channel 03/0 : 15[3] -> 13[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-167-9:1039539:1040321 [2] NCCL INFO Channel 01/0 : 14[2] -> 12[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762818:763616 [3] NCCL INFO Channel 03/0 : 7[3] -> 5[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624946:625742 [3] NCCL INFO Channel 07/0 : 11[3] -> 9[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-214:624945:625741 [2] NCCL INFO Channel 04/0 : 10[2] -> 8[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127293:128058 [3] NCCL INFO Channel 06/0 : 7[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127292:128057 [2] NCCL INFO Channel 04/0 : 6[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762817:763617 [2] NCCL INFO Channel 00/0 : 6[2] -> 4[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624945:625741 [2] NCCL INFO Channel 05/0 : 10[2] -> 8[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127293:128058 [3] NCCL INFO Channel 07/0 : 7[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127292:128057 [2] NCCL INFO Channel 05/0 : 6[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762818:763616 [3] NCCL INFO Channel 02/0 : 5[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762817:763617 [2] NCCL INFO Channel 01/0 : 6[2] -> 4[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762818:763616 [3] NCCL INFO Channel 03/0 : 5[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762817:763617 [2] NCCL INFO Channel 00/0 : 4[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-125:762818:763616 [3] NCCL INFO Channel 06/0 : 5[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762817:763617 [2] NCCL INFO Channel 01/0 : 4[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897243:898033 [3] NCCL INFO Channel 02/0 : 13[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762818:763616 [3] NCCL INFO Channel 07/0 : 5[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762817:763617 [2] NCCL INFO Channel 04/0 : 4[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897242:898034 [2] NCCL INFO Channel 00/0 : 12[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897243:898033 [3] NCCL INFO Channel 03/0 : 13[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-125:762817:763617 [2] NCCL INFO Channel 05/0 : 4[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127292:128057 [2] NCCL INFO Channel 00/0 : 6[2] -> 4[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9628:10394 [3] NCCL INFO Channel 02/0 : 5[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9627:10393 [2] NCCL INFO Channel 00/0 : 4[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897242:898034 [2] NCCL INFO Channel 01/0 : 12[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127293:128058 [3] NCCL INFO Channel 02/0 : 7[3] -> 5[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897243:898033 [3] NCCL INFO Channel 06/0 : 13[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9628:10394 [3] NCCL INFO Channel 03/0 : 5[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897242:898034 [2] NCCL INFO Channel 04/0 : 12[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-15:127293:128058 [3] NCCL INFO Channel 03/0 : 7[3] -> 5[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9627:10393 [2] NCCL INFO Channel 01/0 : 4[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897243:898033 [3] NCCL INFO Channel 07/0 : 13[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897242:898034 [2] NCCL INFO Channel 05/0 : 12[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9628:10394 [3] NCCL INFO Channel 06/0 : 5[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9627:10393 [2] NCCL INFO Channel 04/0 : 4[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897243:898033 [3] NCCL INFO Channel 06/0 : 11[3] -> 9[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127293:128058 [3] NCCL INFO Channel 00/0 : 7[3] -> 6[2] via P2P/IPC
+ip-26-0-165-213:9628:10394 [3] NCCL INFO Channel 07/0 : 5[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897242:898034 [2] NCCL INFO Channel 04/0 : 10[2] -> 8[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9627:10393 [2] NCCL INFO Channel 05/0 : 4[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9628:10394 [3] NCCL INFO Channel 06/0 : 3[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-244:897243:898033 [3] NCCL INFO Channel 07/0 : 11[3] -> 9[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-15:127292:128057 [2] NCCL INFO Channel 01/0 : 6[2] -> 4[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-244:897242:898034 [2] NCCL INFO Channel 05/0 : 10[2] -> 8[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9628:10394 [3] NCCL INFO Channel 07/0 : 3[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-165-213:9627:10393 [2] NCCL INFO Channel 04/0 : 2[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-165-213:9627:10393 [2] NCCL INFO Channel 05/0 : 2[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-214:624946:625742 [3] NCCL INFO Channel 00/0 : 9[3] -> 8[2] via P2P/IPC
+ip-26-0-166-125:762818:763616 [3] NCCL INFO Channel 00/0 : 5[3] -> 4[2] via P2P/IPC
+ip-26-0-165-213:9628:10394 [3] NCCL INFO Channel 00/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-162-46:702883:703683 [3] NCCL INFO Channel 00/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-166-36:104431:105195 [2] NCCL INFO Channel 00/0 : 14[2] -> 12[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104431:105195 [2] NCCL INFO Channel 01/0 : 14[2] -> 12[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104432:105194 [3] NCCL INFO Channel 02/0 : 15[3] -> 13[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104431:105195 [2] NCCL INFO Channel 00/0 : 12[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104432:105194 [3] NCCL INFO Channel 03/0 : 15[3] -> 13[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104431:105195 [2] NCCL INFO Channel 01/0 : 12[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104432:105194 [3] NCCL INFO Channel 02/0 : 13[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104431:105195 [2] NCCL INFO Channel 04/0 : 12[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104432:105194 [3] NCCL INFO Channel 03/0 : 13[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104431:105195 [2] NCCL INFO Channel 05/0 : 12[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-166-36:104432:105194 [3] NCCL INFO Channel 06/0 : 13[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-166-36:104432:105194 [3] NCCL INFO Channel 07/0 : 13[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-167-9:1039540:1040320 [3] NCCL INFO Channel 00/0 : 15[3] -> 14[2] via P2P/IPC
+ip-26-0-166-36:104432:105194 [3] NCCL INFO Channel 00/0 : 13[3] -> 12[2] via P2P/IPC
+ip-26-0-166-125:762818:763616 [3] NCCL INFO Channel 02/0 : 5[3] -> 4[2] via P2P/IPC
+ip-26-0-166-244:897243:898033 [3] NCCL INFO Channel 00/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-162-46:702883:703683 [3] NCCL INFO Channel 02/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-162-46:702883:703683 [3] NCCL INFO Channel 04/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-162-46:702883:703683 [3] NCCL INFO Channel 06/0 : 1[3] -> 0[2] via P2P/IPC
+ip-26-0-166-15:127293:128058 [3] NCCL INFO Channel 02/0 : 7[3] -> 6[2] via P2P/IPC
+ip-26-0-166-15:127293:128058 [3] NCCL INFO Channel 04/0 : 7[3] -> 6[2] via P2P/IPC
+ip-26-0-166-15:127293:128058 [3] NCCL INFO Channel 06/0 : 7[3] -> 6[2] via P2P/IPC
+ip-26-0-165-213:9628:10394 [3] NCCL INFO Channel 02/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-125:762818:763616 [3] NCCL INFO Channel 04/0 : 5[3] -> 4[2] via P2P/IPC
+ip-26-0-166-214:624946:625742 [3] NCCL INFO Channel 02/0 : 9[3] -> 8[2] via P2P/IPC
+ip-26-0-166-36:104432:105194 [3] NCCL INFO Channel 02/0 : 13[3] -> 12[2] via P2P/IPC
+ip-26-0-167-9:1039540:1040320 [3] NCCL INFO Channel 02/0 : 15[3] -> 14[2] via P2P/IPC
+ip-26-0-166-244:897243:898033 [3] NCCL INFO Channel 02/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-166-214:624946:625742 [3] NCCL INFO Channel 04/0 : 9[3] -> 8[2] via P2P/IPC
+ip-26-0-166-125:762818:763616 [3] NCCL INFO Channel 06/0 : 5[3] -> 4[2] via P2P/IPC
+ip-26-0-166-214:624946:625742 [3] NCCL INFO Channel 06/0 : 9[3] -> 8[2] via P2P/IPC
+ip-26-0-166-244:897243:898033 [3] NCCL INFO Channel 04/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-167-9:1039540:1040320 [3] NCCL INFO Channel 04/0 : 15[3] -> 14[2] via P2P/IPC
+ip-26-0-165-213:9628:10394 [3] NCCL INFO Channel 04/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-166-36:104432:105194 [3] NCCL INFO Channel 04/0 : 13[3] -> 12[2] via P2P/IPC
+ip-26-0-167-9:1039540:1040320 [3] NCCL INFO Channel 06/0 : 15[3] -> 14[2] via P2P/IPC
+ip-26-0-166-244:897243:898033 [3] NCCL INFO Channel 06/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-166-36:104432:105194 [3] NCCL INFO Channel 06/0 : 13[3] -> 12[2] via P2P/IPC
+ip-26-0-165-213:9628:10394 [3] NCCL INFO Channel 06/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-167-9:1039542:1040319 [5] NCCL INFO Connected all trees
+ip-26-0-167-9:1039542:1040319 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-167-9:1039542:1040319 [5] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+ip-26-0-167-9:1039542:1040319 [5] NCCL INFO 8 coll channels, 0 nvls channels, 8 p2p channels, 2 p2p channels per peer
+ip-26-0-162-46:702881:703677 [1] NCCL INFO Connected all trees
+ip-26-0-162-46:702881:703677 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-162-46:702881:703677 [1] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+ip-26-0-162-46:702881:703677 [1] NCCL INFO 8 coll channels, 0 nvls channels, 8 p2p channels, 2 p2p channels per peer
+ip-26-0-167-9:1039537:1040316 [0] NCCL INFO Connected all trees
+ip-26-0-167-9:1039537:1040316 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-167-9:1039537:1040316 [0] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+ip-26-0-167-9:1039537:1040316 [0] NCCL INFO 8 coll channels, 0 nvls channels, 8 p2p channels, 2 p2p channels per peer
+ip-26-0-162-46:702886:703685 [6] NCCL INFO Connected all trees
+ip-26-0-162-46:702886:703685 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-162-46:702886:703685 [6] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+ip-26-0-162-46:702886:703685 [6] NCCL INFO 8 coll channels, 0 nvls channels, 8 p2p channels, 2 p2p channels per peer
+ip-26-0-167-9:1039538:1040317 [1] NCCL INFO Connected all trees
+ip-26-0-167-9:1039538:1040317 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-167-9:1039538:1040317 [1] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+ip-26-0-167-9:1039538:1040317 [1] NCCL INFO 8 coll channels, 0 nvls channels, 8 p2p channels, 2 p2p channels per peer
+ip-26-0-167-9:1039537:1040316 [0] NCCL INFO comm 0x20e82de0 rank 14 nranks 16 cudaDev 0 nvmlDev 0 busId 53000 commId 0x27beb3196ace4a07 - Init COMPLETE
+ip-26-0-167-9:1039538:1040317 [1] NCCL INFO comm 0x1b80ccb0 rank 15 nranks 16 cudaDev 1 nvmlDev 1 busId 64000 commId 0x27beb3196ace4a07 - Init COMPLETE
+ip-26-0-167-9:1039541:1040318 [4] NCCL INFO Connected all trees
+ip-26-0-167-9:1039541:1040318 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-167-9:1039541:1040318 [4] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+ip-26-0-167-9:1039541:1040318 [4] NCCL INFO 8 coll channels, 0 nvls channels, 8 p2p channels, 2 p2p channels per peer
+ip-26-0-167-9:1039541:1040318 [4] NCCL INFO comm 0x216daf60 rank 14 nranks 16 cudaDev 4 nvmlDev 4 busId 97000 commId 0xddcb46505a6ea727 - Init COMPLETE
+ip-26-0-167-9:1039542:1040319 [5] NCCL INFO comm 0x1b6228e0 rank 15 nranks 16 cudaDev 5 nvmlDev 5 busId a8000 commId 0xddcb46505a6ea727 - Init COMPLETE
+ip-26-0-167-9:1039539:1040321 [2] NCCL INFO Connected all trees
+ip-26-0-167-9:1039539:1040321 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-167-9:1039539:1040321 [2] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+ip-26-0-167-9:1039539:1040321 [2] NCCL INFO 8 coll channels, 0 nvls channels, 8 p2p channels, 2 p2p channels per peer
+ip-26-0-167-9:1039544:1040322 [7] NCCL INFO Connected all trees
+ip-26-0-167-9:1039544:1040322 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-167-9:1039544:1040322 [7] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+ip-26-0-167-9:1039544:1040322 [7] NCCL INFO 8 coll channels, 0 nvls channels, 8 p2p channels, 2 p2p channels per peer
+ip-26-0-162-46:702885:703680 [5] NCCL INFO Connected all trees
+ip-26-0-162-46:702885:703680 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-162-46:702885:703680 [5] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+ip-26-0-162-46:702885:703680 [5] NCCL INFO 8 coll channels, 0 nvls channels, 8 p2p channels, 2 p2p channels per peer
+ip-26-0-167-9:1039540:1040320 [3] NCCL INFO Connected all trees
+ip-26-0-167-9:1039540:1040320 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-167-9:1039540:1040320 [3] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+ip-26-0-167-9:1039540:1040320 [3] NCCL INFO 8 coll channels, 0 nvls channels, 8 p2p channels, 2 p2p channels per peer
+ip-26-0-162-46:702883:703683 [3] NCCL INFO Connected all trees
+ip-26-0-162-46:702883:703683 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-162-46:702883:703683 [3] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+ip-26-0-162-46:702883:703683 [3] NCCL INFO 8 coll channels, 0 nvls channels, 8 p2p channels, 2 p2p channels per peer
+ip-26-0-167-9:1039540:1040320 [3] NCCL INFO comm 0x1bf57ad0 rank 15 nranks 16 cudaDev 3 nvmlDev 3 busId 86000 commId 0xbb8f0b4ef9cdd09a - Init COMPLETE
+ip-26-0-167-9:1039539:1040321 [2] NCCL INFO comm 0x2034e5e0 rank 14 nranks 16 cudaDev 2 nvmlDev 2 busId 75000 commId 0xbb8f0b4ef9cdd09a - Init COMPLETE
+ip-26-0-162-46:702884:703679 [4] NCCL INFO Connected all trees
+ip-26-0-162-46:702884:703679 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-162-46:702884:703679 [4] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+ip-26-0-162-46:702884:703679 [4] NCCL INFO 8 coll channels, 0 nvls channels, 8 p2p channels, 2 p2p channels per peer
+ip-26-0-167-9:1039543:1040323 [6] NCCL INFO Connected all trees
+ip-26-0-167-9:1039543:1040323 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-167-9:1039543:1040323 [6] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+ip-26-0-167-9:1039543:1040323 [6] NCCL INFO 8 coll channels, 0 nvls channels, 8 p2p channels, 2 p2p channels per peer
+ip-26-0-162-46:702887:703686 [7] NCCL INFO Connected all trees
+ip-26-0-162-46:702887:703686 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-162-46:702887:703686 [7] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+ip-26-0-162-46:702887:703686 [7] NCCL INFO 8 coll channels, 0 nvls channels, 8 p2p channels, 2 p2p channels per peer
+ip-26-0-162-46:702884:703679 [4] NCCL INFO comm 0x1cfccc90 rank 0 nranks 16 cudaDev 4 nvmlDev 4 busId 97000 commId 0xddcb46505a6ea727 - Init COMPLETE
+ip-26-0-162-46:702885:703680 [5] NCCL INFO comm 0x1eb5c6a0 rank 1 nranks 16 cudaDev 5 nvmlDev 5 busId a8000 commId 0xddcb46505a6ea727 - Init COMPLETE
+ip-26-0-167-9:1039543:1040323 [6] NCCL INFO comm 0x211a8520 rank 14 nranks 16 cudaDev 6 nvmlDev 6 busId b9000 commId 0x30acb1707865c266 - Init COMPLETE
+ip-26-0-167-9:1039544:1040322 [7] NCCL INFO comm 0x1be68400 rank 15 nranks 16 cudaDev 7 nvmlDev 7 busId ca000 commId 0x30acb1707865c266 - Init COMPLETE
+ip-26-0-162-46:702886:703685 [6] NCCL INFO comm 0x1d0fd800 rank 0 nranks 16 cudaDev 6 nvmlDev 6 busId b9000 commId 0x30acb1707865c266 - Init COMPLETE
+ip-26-0-162-46:702887:703686 [7] NCCL INFO comm 0x1eb4c5a0 rank 1 nranks 16 cudaDev 7 nvmlDev 7 busId ca000 commId 0x30acb1707865c266 - Init COMPLETE
+ip-26-0-162-46:702880:703676 [0] NCCL INFO Connected all trees
+ip-26-0-162-46:702880:703676 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-162-46:702880:703676 [0] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+ip-26-0-162-46:702880:703676 [0] NCCL INFO 8 coll channels, 0 nvls channels, 8 p2p channels, 2 p2p channels per peer
+ip-26-0-162-46:702882:703682 [2] NCCL INFO Connected all trees
+ip-26-0-162-46:702882:703682 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-162-46:702882:703682 [2] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+ip-26-0-162-46:702882:703682 [2] NCCL INFO 8 coll channels, 0 nvls channels, 8 p2p channels, 2 p2p channels per peer
+ip-26-0-162-46:702881:703677 [1] NCCL INFO comm 0x1f3f8120 rank 1 nranks 16 cudaDev 1 nvmlDev 1 busId 64000 commId 0x27beb3196ace4a07 - Init COMPLETE
+ip-26-0-162-46:702880:703676 [0] NCCL INFO comm 0x1d9e9040 rank 0 nranks 16 cudaDev 0 nvmlDev 0 busId 53000 commId 0x27beb3196ace4a07 - Init COMPLETE
+ip-26-0-162-46:702883:703683 [3] NCCL INFO comm 0x1e5674b0 rank 1 nranks 16 cudaDev 3 nvmlDev 3 busId 86000 commId 0xbb8f0b4ef9cdd09a - Init COMPLETE
+ip-26-0-162-46:702882:703682 [2] NCCL INFO comm 0x1d872dd0 rank 0 nranks 16 cudaDev 2 nvmlDev 2 busId 75000 commId 0xbb8f0b4ef9cdd09a - Init COMPLETE
+ip-26-0-166-36:104430:105190 [1] NCCL INFO Connected all trees
+ip-26-0-166-36:104430:105190 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-36:104430:105190 [1] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+ip-26-0-166-36:104430:105190 [1] NCCL INFO 8 coll channels, 0 nvls channels, 8 p2p channels, 2 p2p channels per peer
+ip-26-0-165-213:9625:10390 [0] NCCL INFO Connected all trees
+ip-26-0-165-213:9625:10390 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-213:9625:10390 [0] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+ip-26-0-165-213:9625:10390 [0] NCCL INFO 8 coll channels, 0 nvls channels, 8 p2p channels, 2 p2p channels per peer
+ip-26-0-166-244:897244:898032 [4] NCCL INFO Connected all trees
+ip-26-0-166-244:897244:898032 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-244:897244:898032 [4] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+ip-26-0-166-244:897244:898032 [4] NCCL INFO 8 coll channels, 0 nvls channels, 8 p2p channels, 2 p2p channels per peer
+ip-26-0-166-244:897241:898029 [1] NCCL INFO Connected all trees
+ip-26-0-166-244:897241:898029 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-244:897241:898029 [1] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+ip-26-0-166-244:897241:898029 [1] NCCL INFO 8 coll channels, 0 nvls channels, 8 p2p channels, 2 p2p channels per peer
+ip-26-0-166-36:104434:105192 [5] NCCL INFO Connected all trees
+ip-26-0-166-36:104434:105192 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-36:104434:105192 [5] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+ip-26-0-166-36:104434:105192 [5] NCCL INFO 8 coll channels, 0 nvls channels, 8 p2p channels, 2 p2p channels per peer
+ip-26-0-166-214:624943:625737 [0] NCCL INFO Connected all trees
+ip-26-0-166-214:624943:625737 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-214:624943:625737 [0] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+ip-26-0-166-214:624943:625737 [0] NCCL INFO 8 coll channels, 0 nvls channels, 8 p2p channels, 2 p2p channels per peer
+ip-26-0-166-214:624944:625738 [1] NCCL INFO Connected all trees
+ip-26-0-166-214:624944:625738 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-214:624944:625738 [1] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+ip-26-0-166-214:624944:625738 [1] NCCL INFO 8 coll channels, 0 nvls channels, 8 p2p channels, 2 p2p channels per peer
+ip-26-0-165-213:9629:10391 [4] NCCL INFO Connected all trees
+ip-26-0-165-213:9629:10391 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-213:9629:10391 [4] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+ip-26-0-165-213:9629:10391 [4] NCCL INFO 8 coll channels, 0 nvls channels, 8 p2p channels, 2 p2p channels per peer
+ip-26-0-165-213:9626:10389 [1] NCCL INFO Connected all trees
+ip-26-0-165-213:9626:10389 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-213:9626:10389 [1] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+ip-26-0-165-213:9626:10389 [1] NCCL INFO 8 coll channels, 0 nvls channels, 8 p2p channels, 2 p2p channels per peer
+ip-26-0-166-36:104433:105193 [4] NCCL INFO Connected all trees
+ip-26-0-166-36:104433:105193 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-36:104433:105193 [4] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+ip-26-0-166-36:104433:105193 [4] NCCL INFO 8 coll channels, 0 nvls channels, 8 p2p channels, 2 p2p channels per peer
+ip-26-0-166-36:104429:105191 [0] NCCL INFO Connected all trees
+ip-26-0-166-36:104429:105191 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-36:104429:105191 [0] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+ip-26-0-166-36:104429:105191 [0] NCCL INFO 8 coll channels, 0 nvls channels, 8 p2p channels, 2 p2p channels per peer
+ip-26-0-166-244:897240:898030 [0] NCCL INFO Connected all trees
+ip-26-0-166-244:897240:898030 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-244:897240:898030 [0] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+ip-26-0-166-244:897240:898030 [0] NCCL INFO 8 coll channels, 0 nvls channels, 8 p2p channels, 2 p2p channels per peer
+ip-26-0-166-244:897245:898031 [5] NCCL INFO Connected all trees
+ip-26-0-166-244:897245:898031 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-244:897245:898031 [5] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+ip-26-0-166-244:897245:898031 [5] NCCL INFO 8 coll channels, 0 nvls channels, 8 p2p channels, 2 p2p channels per peer
+ip-26-0-166-15:127290:128054 [0] NCCL INFO Connected all trees
+ip-26-0-166-15:127290:128054 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-15:127290:128054 [0] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+ip-26-0-166-15:127290:128054 [0] NCCL INFO 8 coll channels, 0 nvls channels, 8 p2p channels, 2 p2p channels per peer
+ip-26-0-166-125:762819:763614 [4] NCCL INFO Connected all trees
+ip-26-0-166-125:762819:763614 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-125:762819:763614 [4] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+ip-26-0-166-125:762819:763614 [4] NCCL INFO 8 coll channels, 0 nvls channels, 8 p2p channels, 2 p2p channels per peer
+ip-26-0-165-213:9630:10392 [5] NCCL INFO Connected all trees
+ip-26-0-165-213:9630:10392 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-213:9630:10392 [5] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+ip-26-0-165-213:9630:10392 [5] NCCL INFO 8 coll channels, 0 nvls channels, 8 p2p channels, 2 p2p channels per peer
+ip-26-0-166-214:624943:625737 [0] NCCL INFO comm 0x17641680 rank 8 nranks 16 cudaDev 0 nvmlDev 0 busId 53000 commId 0x27beb3196ace4a07 - Init COMPLETE
+ip-26-0-166-214:624944:625738 [1] NCCL INFO comm 0x166ec9b0 rank 9 nranks 16 cudaDev 1 nvmlDev 1 busId 64000 commId 0x27beb3196ace4a07 - Init COMPLETE
+ip-26-0-166-36:104429:105191 [0] NCCL INFO comm 0x185904e0 rank 12 nranks 16 cudaDev 0 nvmlDev 0 busId 53000 commId 0x27beb3196ace4a07 - Init COMPLETE
+ip-26-0-166-36:104430:105190 [1] NCCL INFO comm 0x17da6700 rank 13 nranks 16 cudaDev 1 nvmlDev 1 busId 64000 commId 0x27beb3196ace4a07 - Init COMPLETE
+ip-26-0-166-214:624947:625739 [4] NCCL INFO Connected all trees
+ip-26-0-166-214:624947:625739 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-214:624947:625739 [4] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+ip-26-0-166-214:624947:625739 [4] NCCL INFO 8 coll channels, 0 nvls channels, 8 p2p channels, 2 p2p channels per peer
+ip-26-0-165-213:9626:10389 [1] NCCL INFO comm 0x176d8980 rank 3 nranks 16 cudaDev 1 nvmlDev 1 busId 64000 commId 0x27beb3196ace4a07 - Init COMPLETE
+ip-26-0-165-213:9625:10390 [0] NCCL INFO comm 0x171022d0 rank 2 nranks 16 cudaDev 0 nvmlDev 0 busId 53000 commId 0x27beb3196ace4a07 - Init COMPLETE
+ip-26-0-166-214:624948:625740 [5] NCCL INFO Connected all trees
+ip-26-0-166-214:624948:625740 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-214:624948:625740 [5] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+ip-26-0-166-214:624948:625740 [5] NCCL INFO 8 coll channels, 0 nvls channels, 8 p2p channels, 2 p2p channels per peer
+ip-26-0-166-36:104434:105192 [5] NCCL INFO comm 0x16ba8110 rank 13 nranks 16 cudaDev 5 nvmlDev 5 busId a8000 commId 0xddcb46505a6ea727 - Init COMPLETE
+ip-26-0-166-36:104433:105193 [4] NCCL INFO comm 0x1838ccd0 rank 12 nranks 16 cudaDev 4 nvmlDev 4 busId 97000 commId 0xddcb46505a6ea727 - Init COMPLETE
+ip-26-0-166-15:127295:128055 [5] NCCL INFO Connected all trees
+ip-26-0-166-15:127295:128055 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-15:127295:128055 [5] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+ip-26-0-166-15:127295:128055 [5] NCCL INFO 8 coll channels, 0 nvls channels, 8 p2p channels, 2 p2p channels per peer
+ip-26-0-166-125:762815:763613 [0] NCCL INFO Connected all trees
+ip-26-0-166-125:762815:763613 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-125:762815:763613 [0] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+ip-26-0-166-125:762815:763613 [0] NCCL INFO 8 coll channels, 0 nvls channels, 8 p2p channels, 2 p2p channels per peer
+ip-26-0-166-214:624950:625743 [7] NCCL INFO Connected all trees
+ip-26-0-166-214:624950:625743 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-214:624950:625743 [7] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+ip-26-0-166-214:624950:625743 [7] NCCL INFO 8 coll channels, 0 nvls channels, 8 p2p channels, 2 p2p channels per peer
+ip-26-0-166-36:104431:105195 [2] NCCL INFO Connected all trees
+ip-26-0-166-36:104431:105195 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-36:104431:105195 [2] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+ip-26-0-166-36:104431:105195 [2] NCCL INFO 8 coll channels, 0 nvls channels, 8 p2p channels, 2 p2p channels per peer
+ip-26-0-166-15:127291:128053 [1] NCCL INFO Connected all trees
+ip-26-0-166-15:127291:128053 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-15:127291:128053 [1] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+ip-26-0-166-15:127291:128053 [1] NCCL INFO 8 coll channels, 0 nvls channels, 8 p2p channels, 2 p2p channels per peer
+ip-26-0-166-36:104432:105194 [3] NCCL INFO Connected all trees
+ip-26-0-166-36:104432:105194 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-36:104432:105194 [3] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+ip-26-0-166-36:104432:105194 [3] NCCL INFO 8 coll channels, 0 nvls channels, 8 p2p channels, 2 p2p channels per peer
+ip-26-0-166-36:104435:105196 [6] NCCL INFO Connected all trees
+ip-26-0-166-36:104435:105196 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-36:104435:105196 [6] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+ip-26-0-166-36:104435:105196 [6] NCCL INFO 8 coll channels, 0 nvls channels, 8 p2p channels, 2 p2p channels per peer
+ip-26-0-166-125:762820:763615 [5] NCCL INFO Connected all trees
+ip-26-0-166-125:762820:763615 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-125:762820:763615 [5] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+ip-26-0-166-125:762820:763615 [5] NCCL INFO 8 coll channels, 0 nvls channels, 8 p2p channels, 2 p2p channels per peer
+ip-26-0-166-214:624947:625739 [4] NCCL INFO comm 0x181df0e0 rank 8 nranks 16 cudaDev 4 nvmlDev 4 busId 97000 commId 0xddcb46505a6ea727 - Init COMPLETE
+ip-26-0-166-214:624948:625740 [5] NCCL INFO comm 0x18255f80 rank 9 nranks 16 cudaDev 5 nvmlDev 5 busId a8000 commId 0xddcb46505a6ea727 - Init COMPLETE
+ip-26-0-166-244:897241:898029 [1] NCCL INFO comm 0x18302ab0 rank 11 nranks 16 cudaDev 1 nvmlDev 1 busId 64000 commId 0x27beb3196ace4a07 - Init COMPLETE
+ip-26-0-166-244:897240:898030 [0] NCCL INFO comm 0x1745a980 rank 10 nranks 16 cudaDev 0 nvmlDev 0 busId 53000 commId 0x27beb3196ace4a07 - Init COMPLETE
+ip-26-0-166-15:127294:128056 [4] NCCL INFO Connected all trees
+ip-26-0-166-15:127294:128056 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-15:127294:128056 [4] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+ip-26-0-166-15:127294:128056 [4] NCCL INFO 8 coll channels, 0 nvls channels, 8 p2p channels, 2 p2p channels per peer
+ip-26-0-166-125:762816:763612 [1] NCCL INFO Connected all trees
+ip-26-0-166-125:762816:763612 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-125:762816:763612 [1] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+ip-26-0-166-125:762816:763612 [1] NCCL INFO 8 coll channels, 0 nvls channels, 8 p2p channels, 2 p2p channels per peer
+ip-26-0-166-214:624945:625741 [2] NCCL INFO Connected all trees
+ip-26-0-166-214:624945:625741 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-214:624945:625741 [2] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+ip-26-0-166-214:624945:625741 [2] NCCL INFO 8 coll channels, 0 nvls channels, 8 p2p channels, 2 p2p channels per peer
+ip-26-0-166-244:897244:898032 [4] NCCL INFO comm 0x16fdcbf0 rank 10 nranks 16 cudaDev 4 nvmlDev 4 busId 97000 commId 0xddcb46505a6ea727 - Init COMPLETE
+ip-26-0-166-244:897245:898031 [5] NCCL INFO comm 0x17bfad90 rank 11 nranks 16 cudaDev 5 nvmlDev 5 busId a8000 commId 0xddcb46505a6ea727 - Init COMPLETE
+ip-26-0-166-244:897246:898036 [6] NCCL INFO Connected all trees
+ip-26-0-166-244:897246:898036 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-244:897246:898036 [6] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+ip-26-0-166-244:897246:898036 [6] NCCL INFO 8 coll channels, 0 nvls channels, 8 p2p channels, 2 p2p channels per peer
+ip-26-0-165-213:9629:10391 [4] NCCL INFO comm 0x16c33e40 rank 2 nranks 16 cudaDev 4 nvmlDev 4 busId 97000 commId 0xddcb46505a6ea727 - Init COMPLETE
+ip-26-0-165-213:9630:10392 [5] NCCL INFO comm 0x16ab1430 rank 3 nranks 16 cudaDev 5 nvmlDev 5 busId a8000 commId 0xddcb46505a6ea727 - Init COMPLETE
+ip-26-0-166-15:127297:128060 [7] NCCL INFO Connected all trees
+ip-26-0-166-15:127297:128060 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-15:127297:128060 [7] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+ip-26-0-166-15:127297:128060 [7] NCCL INFO 8 coll channels, 0 nvls channels, 8 p2p channels, 2 p2p channels per peer
+ip-26-0-165-213:9628:10394 [3] NCCL INFO Connected all trees
+ip-26-0-165-213:9628:10394 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-213:9628:10394 [3] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+ip-26-0-165-213:9628:10394 [3] NCCL INFO 8 coll channels, 0 nvls channels, 8 p2p channels, 2 p2p channels per peer
+ip-26-0-166-125:762822:763618 [7] NCCL INFO Connected all trees
+ip-26-0-166-125:762822:763618 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-125:762822:763618 [7] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+ip-26-0-166-125:762822:763618 [7] NCCL INFO 8 coll channels, 0 nvls channels, 8 p2p channels, 2 p2p channels per peer
+ip-26-0-165-213:9631:10396 [6] NCCL INFO Connected all trees
+ip-26-0-165-213:9631:10396 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-213:9631:10396 [6] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+ip-26-0-165-213:9631:10396 [6] NCCL INFO 8 coll channels, 0 nvls channels, 8 p2p channels, 2 p2p channels per peer
+ip-26-0-166-214:624946:625742 [3] NCCL INFO Connected all trees
+ip-26-0-166-214:624946:625742 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-214:624946:625742 [3] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+ip-26-0-166-214:624946:625742 [3] NCCL INFO 8 coll channels, 0 nvls channels, 8 p2p channels, 2 p2p channels per peer
+ip-26-0-165-213:9627:10393 [2] NCCL INFO Connected all trees
+ip-26-0-165-213:9627:10393 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-213:9627:10393 [2] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+ip-26-0-165-213:9627:10393 [2] NCCL INFO 8 coll channels, 0 nvls channels, 8 p2p channels, 2 p2p channels per peer
+ip-26-0-166-214:624949:625744 [6] NCCL INFO Connected all trees
+ip-26-0-166-214:624949:625744 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-214:624949:625744 [6] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+ip-26-0-166-214:624949:625744 [6] NCCL INFO 8 coll channels, 0 nvls channels, 8 p2p channels, 2 p2p channels per peer
+ip-26-0-165-213:9632:10395 [7] NCCL INFO Connected all trees
+ip-26-0-165-213:9632:10395 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-165-213:9632:10395 [7] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+ip-26-0-165-213:9632:10395 [7] NCCL INFO 8 coll channels, 0 nvls channels, 8 p2p channels, 2 p2p channels per peer
+ip-26-0-166-36:104431:105195 [2] NCCL INFO comm 0x17934380 rank 12 nranks 16 cudaDev 2 nvmlDev 2 busId 75000 commId 0xbb8f0b4ef9cdd09a - Init COMPLETE
+ip-26-0-166-36:104432:105194 [3] NCCL INFO comm 0x1692d450 rank 13 nranks 16 cudaDev 3 nvmlDev 3 busId 86000 commId 0xbb8f0b4ef9cdd09a - Init COMPLETE
+ip-26-0-166-15:127290:128054 [0] NCCL INFO comm 0x16c594f0 rank 6 nranks 16 cudaDev 0 nvmlDev 0 busId 53000 commId 0x27beb3196ace4a07 - Init COMPLETE
+ip-26-0-166-15:127291:128053 [1] NCCL INFO comm 0x184d7020 rank 7 nranks 16 cudaDev 1 nvmlDev 1 busId 64000 commId 0x27beb3196ace4a07 - Init COMPLETE
+ip-26-0-166-244:897243:898033 [3] NCCL INFO Connected all trees
+ip-26-0-166-244:897243:898033 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-244:897243:898033 [3] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+ip-26-0-166-244:897243:898033 [3] NCCL INFO 8 coll channels, 0 nvls channels, 8 p2p channels, 2 p2p channels per peer
+ip-26-0-166-36:104436:105197 [7] NCCL INFO Connected all trees
+ip-26-0-166-36:104436:105197 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-36:104436:105197 [7] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+ip-26-0-166-36:104436:105197 [7] NCCL INFO 8 coll channels, 0 nvls channels, 8 p2p channels, 2 p2p channels per peer
+ip-26-0-166-125:762819:763614 [4] NCCL INFO comm 0x1769afa0 rank 4 nranks 16 cudaDev 4 nvmlDev 4 busId 97000 commId 0xddcb46505a6ea727 - Init COMPLETE
+ip-26-0-166-125:762820:763615 [5] NCCL INFO comm 0x175bbef0 rank 5 nranks 16 cudaDev 5 nvmlDev 5 busId a8000 commId 0xddcb46505a6ea727 - Init COMPLETE
+ip-26-0-166-125:762817:763617 [2] NCCL INFO Connected all trees
+ip-26-0-166-125:762817:763617 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-125:762817:763617 [2] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+ip-26-0-166-125:762817:763617 [2] NCCL INFO 8 coll channels, 0 nvls channels, 8 p2p channels, 2 p2p channels per peer
+ip-26-0-166-244:897242:898034 [2] NCCL INFO Connected all trees
+ip-26-0-166-244:897242:898034 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-244:897242:898034 [2] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+ip-26-0-166-244:897242:898034 [2] NCCL INFO 8 coll channels, 0 nvls channels, 8 p2p channels, 2 p2p channels per peer
+ip-26-0-166-244:897247:898035 [7] NCCL INFO Connected all trees
+ip-26-0-166-244:897247:898035 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-244:897247:898035 [7] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+ip-26-0-166-244:897247:898035 [7] NCCL INFO 8 coll channels, 0 nvls channels, 8 p2p channels, 2 p2p channels per peer
+ip-26-0-166-15:127292:128057 [2] NCCL INFO Connected all trees
+ip-26-0-166-15:127292:128057 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-15:127292:128057 [2] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+ip-26-0-166-15:127292:128057 [2] NCCL INFO 8 coll channels, 0 nvls channels, 8 p2p channels, 2 p2p channels per peer
+ip-26-0-166-15:127295:128055 [5] NCCL INFO comm 0x16d56070 rank 7 nranks 16 cudaDev 5 nvmlDev 5 busId a8000 commId 0xddcb46505a6ea727 - Init COMPLETE
+ip-26-0-166-15:127294:128056 [4] NCCL INFO comm 0x172652a0 rank 6 nranks 16 cudaDev 4 nvmlDev 4 busId 97000 commId 0xddcb46505a6ea727 - Init COMPLETE
+ip-26-0-166-125:762815:763613 [0] NCCL INFO comm 0x168bbc20 rank 4 nranks 16 cudaDev 0 nvmlDev 0 busId 53000 commId 0x27beb3196ace4a07 - Init COMPLETE
+ip-26-0-166-125:762816:763612 [1] NCCL INFO comm 0x178ea100 rank 5 nranks 16 cudaDev 1 nvmlDev 1 busId 64000 commId 0x27beb3196ace4a07 - Init COMPLETE
+ip-26-0-166-214:624945:625741 [2] NCCL INFO comm 0x1798aed0 rank 8 nranks 16 cudaDev 2 nvmlDev 2 busId 75000 commId 0xbb8f0b4ef9cdd09a - Init COMPLETE
+ip-26-0-166-214:624946:625742 [3] NCCL INFO comm 0x18227c50 rank 9 nranks 16 cudaDev 3 nvmlDev 3 busId 86000 commId 0xbb8f0b4ef9cdd09a - Init COMPLETE
+ip-26-0-166-15:127296:128059 [6] NCCL INFO Connected all trees
+ip-26-0-166-15:127296:128059 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-15:127296:128059 [6] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+ip-26-0-166-15:127296:128059 [6] NCCL INFO 8 coll channels, 0 nvls channels, 8 p2p channels, 2 p2p channels per peer
+ip-26-0-165-213:9628:10394 [3] NCCL INFO comm 0x17627d70 rank 3 nranks 16 cudaDev 3 nvmlDev 3 busId 86000 commId 0xbb8f0b4ef9cdd09a - Init COMPLETE
+ip-26-0-165-213:9627:10393 [2] NCCL INFO comm 0x16c3bbb0 rank 2 nranks 16 cudaDev 2 nvmlDev 2 busId 75000 commId 0xbb8f0b4ef9cdd09a - Init COMPLETE
+ip-26-0-166-36:104435:105196 [6] NCCL INFO comm 0x16ef07b0 rank 12 nranks 16 cudaDev 6 nvmlDev 6 busId b9000 commId 0x30acb1707865c266 - Init COMPLETE
+ip-26-0-166-36:104436:105197 [7] NCCL INFO comm 0x178626f0 rank 13 nranks 16 cudaDev 7 nvmlDev 7 busId ca000 commId 0x30acb1707865c266 - Init COMPLETE
+ip-26-0-166-214:624950:625743 [7] NCCL INFO comm 0x16f730c0 rank 9 nranks 16 cudaDev 7 nvmlDev 7 busId ca000 commId 0x30acb1707865c266 - Init COMPLETE
+ip-26-0-166-214:624949:625744 [6] NCCL INFO comm 0x16c4eaf0 rank 8 nranks 16 cudaDev 6 nvmlDev 6 busId b9000 commId 0x30acb1707865c266 - Init COMPLETE
+ip-26-0-166-15:127293:128058 [3] NCCL INFO Connected all trees
+ip-26-0-166-15:127293:128058 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-15:127293:128058 [3] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+ip-26-0-166-15:127293:128058 [3] NCCL INFO 8 coll channels, 0 nvls channels, 8 p2p channels, 2 p2p channels per peer
+ip-26-0-166-125:762818:763616 [3] NCCL INFO Connected all trees
+ip-26-0-166-125:762818:763616 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-125:762818:763616 [3] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+ip-26-0-166-125:762818:763616 [3] NCCL INFO 8 coll channels, 0 nvls channels, 8 p2p channels, 2 p2p channels per peer
+ip-26-0-166-125:762821:763619 [6] NCCL INFO Connected all trees
+ip-26-0-166-125:762821:763619 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-166-125:762821:763619 [6] NCCL INFO threadThresholds 8/8/64 | 128/8/64 | 512 | 512
+ip-26-0-166-125:762821:763619 [6] NCCL INFO 8 coll channels, 0 nvls channels, 8 p2p channels, 2 p2p channels per peer
+ip-26-0-165-213:9632:10395 [7] NCCL INFO comm 0x1700ed20 rank 3 nranks 16 cudaDev 7 nvmlDev 7 busId ca000 commId 0x30acb1707865c266 - Init COMPLETE
+ip-26-0-165-213:9631:10396 [6] NCCL INFO comm 0x17de4a70 rank 2 nranks 16 cudaDev 6 nvmlDev 6 busId b9000 commId 0x30acb1707865c266 - Init COMPLETE
+ip-26-0-166-244:897243:898033 [3] NCCL INFO comm 0x16b51e40 rank 11 nranks 16 cudaDev 3 nvmlDev 3 busId 86000 commId 0xbb8f0b4ef9cdd09a - Init COMPLETE
+ip-26-0-166-244:897242:898034 [2] NCCL INFO comm 0x16847d90 rank 10 nranks 16 cudaDev 2 nvmlDev 2 busId 75000 commId 0xbb8f0b4ef9cdd09a - Init COMPLETE
+ip-26-0-166-244:897246:898036 [6] NCCL INFO comm 0x185776b0 rank 10 nranks 16 cudaDev 6 nvmlDev 6 busId b9000 commId 0x30acb1707865c266 - Init COMPLETE
+ip-26-0-166-244:897247:898035 [7] NCCL INFO comm 0x17afb910 rank 11 nranks 16 cudaDev 7 nvmlDev 7 busId ca000 commId 0x30acb1707865c266 - Init COMPLETE
+ip-26-0-166-15:127297:128060 [7] NCCL INFO comm 0x18052140 rank 7 nranks 16 cudaDev 7 nvmlDev 7 busId ca000 commId 0x30acb1707865c266 - Init COMPLETE
+ip-26-0-166-15:127296:128059 [6] NCCL INFO comm 0x1793a8e0 rank 6 nranks 16 cudaDev 6 nvmlDev 6 busId b9000 commId 0x30acb1707865c266 - Init COMPLETE
+ip-26-0-166-15:127292:128057 [2] NCCL INFO comm 0x174c51c0 rank 6 nranks 16 cudaDev 2 nvmlDev 2 busId 75000 commId 0xbb8f0b4ef9cdd09a - Init COMPLETE
+ip-26-0-166-15:127293:128058 [3] NCCL INFO comm 0x183c6690 rank 7 nranks 16 cudaDev 3 nvmlDev 3 busId 86000 commId 0xbb8f0b4ef9cdd09a - Init COMPLETE
+ip-26-0-166-125:762817:763617 [2] NCCL INFO comm 0x186bc3a0 rank 4 nranks 16 cudaDev 2 nvmlDev 2 busId 75000 commId 0xbb8f0b4ef9cdd09a - Init COMPLETE
+ip-26-0-166-125:762818:763616 [3] NCCL INFO comm 0x169f9380 rank 5 nranks 16 cudaDev 3 nvmlDev 3 busId 86000 commId 0xbb8f0b4ef9cdd09a - Init COMPLETE
+ip-26-0-166-125:762822:763618 [7] NCCL INFO comm 0x182cd440 rank 5 nranks 16 cudaDev 7 nvmlDev 7 busId ca000 commId 0x30acb1707865c266 - Init COMPLETE
+ip-26-0-166-125:762821:763619 [6] NCCL INFO comm 0x16b951b0 rank 4 nranks 16 cudaDev 6 nvmlDev 6 busId b9000 commId 0x30acb1707865c266 - Init COMPLETE
+01/07/2025 03:50:32 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]:  Memory usage: 2145.70MiB. Peak allocated 3153.75MiB. Peak reserved: 14070.00MiB
+01/07/2025 03:50:32 [INFO|DP=0|PP=7|TP=0|ip-26-0-167-9]: iteration: 1 / 100 | consumed_tokens: 1.05M | elapsed_time_per_iteration_ms: 25.2K | tokens_per_sec: 41.7K | tokens_per_sec_per_gpu: 651 | global_batch_size: 256 | lm_loss: 12 | lr: 0.00015 | model_tflops_per_gpu: 6.29 | hardware_tflops_per_gpu: 6.29 | grad_norm: 0.608 | cuda_memory_allocated: 1.31G | cuda_max_memory_reserved: 8.85G | hd_total_memory_tb: 312G | hd_used_memory_tb: 69.8G | hd_free_memory_tb: 242G
+01/07/2025 03:50:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]:  Memory usage: 2145.70MiB. Peak allocated 12319.02MiB. Peak reserved: 14582.00MiB
+01/07/2025 03:50:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]:  Memory usage: 2145.70MiB. Peak allocated 3153.75MiB. Peak reserved: 14582.00MiB
+01/07/2025 03:50:33 [INFO|DP=0|PP=7|TP=0|ip-26-0-167-9]: iteration: 2 / 100 | consumed_tokens: 2.1M | elapsed_time_per_iteration_ms: 949 | tokens_per_sec: 1.11M | tokens_per_sec_per_gpu: 17.3K | global_batch_size: 256 | lm_loss: 12 | lr: 0.0003 | model_tflops_per_gpu: 167 | hardware_tflops_per_gpu: 167 | grad_norm: 0.608 | cuda_memory_allocated: 1.31G | cuda_max_memory_reserved: 8.85G | hd_total_memory_tb: 312G | hd_used_memory_tb: 69.8G | hd_free_memory_tb: 242G
+01/07/2025 03:50:34 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]:  Memory usage: 2145.70MiB. Peak allocated 12319.02MiB. Peak reserved: 14838.00MiB
+01/07/2025 03:50:34 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]:  Memory usage: 2145.70MiB. Peak allocated 3153.75MiB. Peak reserved: 14838.00MiB
+01/07/2025 03:50:34 [INFO|DP=0|PP=7|TP=0|ip-26-0-167-9]: iteration: 3 / 100 | consumed_tokens: 3.15M | elapsed_time_per_iteration_ms: 938 | tokens_per_sec: 1.12M | tokens_per_sec_per_gpu: 17.5K | global_batch_size: 256 | lm_loss: 12 | lr: 0.000296 | model_tflops_per_gpu: 169 | hardware_tflops_per_gpu: 169 | grad_norm: 0.604 | cuda_memory_allocated: 1.31G | cuda_max_memory_reserved: 8.85G | hd_total_memory_tb: 312G | hd_used_memory_tb: 69.8G | hd_free_memory_tb: 242G
+01/07/2025 03:50:35 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]:  Memory usage: 2145.70MiB. Peak allocated 12319.02MiB. Peak reserved: 14838.00MiB
+01/07/2025 03:50:35 [INFO|DP=0|PP=7|TP=0|ip-26-0-167-9]: iteration: 4 / 100 | consumed_tokens: 4.19M | elapsed_time_per_iteration_ms: 933 | tokens_per_sec: 1.12M | tokens_per_sec_per_gpu: 17.6K | global_batch_size: 256 | lm_loss: 12 | lr: 0.000283 | model_tflops_per_gpu: 170 | hardware_tflops_per_gpu: 170 | grad_norm: 0.577 | cuda_memory_allocated: 1.31G | cuda_max_memory_reserved: 8.85G | hd_total_memory_tb: 312G | hd_used_memory_tb: 69.8G | hd_free_memory_tb: 242G
+01/07/2025 03:50:35 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]: | job_id   | name                                                           | nodes | seq_len | mbs | batch_accum | gbs | mTFLOPs | hTFLOPs | tok/s/gpu | AllReduce (GB/s) | AllGather (GB/s) | ReduceScatter (GB/s) | AR Intra-node (GB/s) | AG Intra-node (GB/s) | RS Intra-node (GB/s) | Mem Alloc (GB) | Mem Res (GB) | dp | pp | tp | pp_engine                           | tp_mode        | tp_async_comm | hidden_size | hidden_act | num_layers | num_heads | num_kv_heads | max_pos | vocab_size | tie_word_embeddings | dtype          | zero_stage | ddp_bucket_cap_mb | accumulate_grad_in_fp32 | Total Params | Local Params |
+01/07/2025 03:50:35 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]: | -------- | -------------------------------------------------------------- | ----- | ------- | --- | ----------- | --- | ------- | ------- | --------- | ---------------- | ---------------- | -------------------- | -------------------- | -------------------- | -------------------- | -------------- | ------------ | -- | -- | -- | ----------------------------------- | -------------- | ------------- | ----------- | ---------- | ---------- | --------- | ------------ | ------- | ---------- | ------------------- | -------------- | ---------- | ----------------- | ----------------------- | ------------ | ------------ |
+01/07/2025 03:50:35 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]: | 14098831 | 1.34G_dp4_tp2_pp8_acc32_mbs2_seq4096_zero1_tpmodeRED_vocab131k | 8     | 4096    | 2   | 32          | 256 | 169.71  | 169.71  | 17561.75  | 98.40            | 57.11            | 55.76                | 426.69               | 241.39               | 243.00               | 3.08           | 14.49        | 4  | 8  | 2  | OneForwardOneBackwardPipelineEngine | REDUCE_SCATTER | True          | 2048        | silu       | 16         | 32        | 32           | 4096    | 131072     | True                | torch.bfloat16 | 1          | 25                | True                    | 1.61G        | 235M         |
+01/07/2025 03:50:35 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]: Attempting to write benchmark results to CSV file: benchmark/results/bench_final2.csv
+01/07/2025 03:50:35 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]: Acquired lock for CSV file: benchmark/results/bench_final2.csv
+01/07/2025 03:50:35 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]: Successfully wrote to CSV file: benchmark/results/bench_final2.csv. Releasing lock...
+01/07/2025 03:50:35 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-46]: Throughput logging complete
+ip-26-0-167-9:1039540:1039950 [32644] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-167-9:1039540:1039722 [32646] NCCL INFO [Service thread] Connection closed by localRank 7
+[E TraceUtils.h:35] Store is down while updating #2415 with key NCCL_1_trace_end
+[E TraceUtils.h:35] Store is down while updating #3648 with key NCCL_0_trace_end
+[E TraceUtils.h:35] Store is down while updating #3647 with key NCCL_1_trace_end
+[E TraceUtils.h:35] Store is down while updating #3644 with key NCCL_0_trace_end
+[E TraceUtils.h:35] Store is down while updating #3616 with key NCCL_1_trace_start
+[E TraceUtils.h:35] Store is down while updating #3566 with key NCCL_0_trace_start
+[E TraceUtils.h:35] Store is down while updating #3778 with key NCCL_1_trace_start
+[E TraceUtils.h:35] Store is down while updating #3827 with key NCCL_0_trace_start
+[E TraceUtils.h:35] Store is down while updating #3645 with key NCCL_1_trace_start
+[E TraceUtils.h:35] Store is down while updating #3822 with key NCCL_1_trace_start
+[E TraceUtils.h:35] Store is down while updating #3562 with key NCCL_1_trace_start
+[E TraceUtils.h:35] Store is down while updating #3780 with key NCCL_0_trace_start
+[E TraceUtils.h:35] Store is down while updating #3624 with key NCCL_0_trace_start
+ip-26-0-165-213:9625:10400 [32636] NCCL INFO [Service thread] Connection closed by localRank 1
+ip-26-0-165-213:9625:10076 [32639] NCCL INFO [Service thread] Connection closed by localRank 1
+ip-26-0-165-213:9632:10044 [32664] NCCL INFO [Service thread] Connection closed by localRank 1
+ip-26-0-165-213:9630:10039 [32521] NCCL INFO [Service thread] Connection closed by localRank 1
+ip-26-0-165-213:9631:10041 [32580] NCCL INFO [Service thread] Connection closed by localRank 1
+ip-26-0-165-213:9629:10040 [32528] NCCL INFO [Service thread] Connection closed by localRank 1
+ip-26-0-165-213:9628:10043 [32523] NCCL INFO [Service thread] Connection closed by localRank 1
+ip-26-0-165-213:9625:10037 [32641] NCCL INFO [Service thread] Connection closed by localRank 1
+ip-26-0-165-213:9627:10042 [32649] NCCL INFO [Service thread] Connection closed by localRank 1
+ip-26-0-165-213:9632:9807 [32666] NCCL INFO [Service thread] Connection closed by localRank 1
+ip-26-0-165-213:9631:9806 [32582] NCCL INFO [Service thread] Connection closed by localRank 1
+ip-26-0-165-213:9630:9808 [32523] NCCL INFO [Service thread] Connection closed by localRank 1
+ip-26-0-165-213:9629:9810 [32530] NCCL INFO [Service thread] Connection closed by localRank 1
+ip-26-0-165-213:9625:9812 [32643] NCCL INFO [Service thread] Connection closed by localRank 1
+ip-26-0-165-213:9627:9811 [32651] NCCL INFO [Service thread] Connection closed by localRank 1
+ip-26-0-165-213:9628:9813 [32525] NCCL INFO [Service thread] Connection closed by localRank 1
+ip-26-0-166-244:897246:898050 [32652] NCCL INFO [Service thread] Connection closed by localRank 1
+ip-26-0-166-244:897241:897804 [32634] NCCL INFO [Service thread] Connection closed by localRank 3
+ip-26-0-166-244:897240:897693 [32749] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-166-244:897245:897652 [32706] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-166-244:897244:897659 [32534] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-166-244:897243:897654 [32748] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-166-244:897242:897655 [32695] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-166-244:897241:897656 [32640] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-166-244:897240:897657 [32750] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-166-244:897246:897653 [32653] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-166-244:897245:897421 [32707] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-166-244:897244:897422 [32536] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-166-244:897243:897428 [32750] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-166-244:897242:897425 [32697] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-166-244:897241:897424 [32641] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-166-244:897246:897427 [32655] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-166-244:897240:897426 [32754] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-165-213:9631:10041 [32580] NCCL INFO [Service thread] Connection closed by localRank 0
+ip-26-0-165-213:9630:10039 [32521] NCCL INFO [Service thread] Connection closed by localRank 0
+ip-26-0-165-213:9629:10040 [32528] NCCL INFO [Service thread] Connection closed by localRank 0
+ip-26-0-165-213:9628:10043 [32523] NCCL INFO [Service thread] Connection closed by localRank 0
+ip-26-0-165-213:9627:10042 [32649] NCCL INFO [Service thread] Connection closed by localRank 0
+ip-26-0-165-213:9631:9806 [32582] NCCL INFO [Service thread] Connection closed by localRank 0
+ip-26-0-165-213:9630:9808 [32523] NCCL INFO [Service thread] Connection closed by localRank 0
+ip-26-0-165-213:9629:9810 [32530] NCCL INFO [Service thread] Connection closed by localRank 0
+ip-26-0-165-213:9627:9811 [32651] NCCL INFO [Service thread] Connection closed by localRank 0
+ip-26-0-165-213:9628:9813 [32525] NCCL INFO [Service thread] Connection closed by localRank 0
+ip-26-0-165-213:9631:10406 [32575] NCCL INFO [Service thread] Connection closed by localRank 1
+ip-26-0-165-213:9628:10043 [32523] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-165-213:9627:10042 [32649] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-165-213:9628:9813 [32525] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-165-213:9627:9811 [32651] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-165-213:9627:10042 [32649] NCCL INFO [Service thread] Connection closed by localRank 4
+ip-26-0-165-213:9628:10043 [32523] NCCL INFO [Service thread] Connection closed by localRank 4
+ip-26-0-165-213:9627:9811 [32651] NCCL INFO [Service thread] Connection closed by localRank 4
+ip-26-0-165-213:9628:9813 [32525] NCCL INFO [Service thread] Connection closed by localRank 4
+[2025-01-07 03:50:58,699] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1039537 closing signal SIGTERM
+[2025-01-07 03:50:58,700] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1039538 closing signal SIGTERM
+[2025-01-07 03:50:58,700] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1039539 closing signal SIGTERM
+[2025-01-07 03:50:58,700] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1039540 closing signal SIGTERM
+[2025-01-07 03:50:58,700] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1039541 closing signal SIGTERM
+[2025-01-07 03:50:58,700] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1039542 closing signal SIGTERM
+[2025-01-07 03:50:58,700] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1039543 closing signal SIGTERM
+[2025-01-07 03:51:01,996] torch.distributed.elastic.multiprocessing.api: [ERROR] failed (exitcode: 1) local_rank: 7 (pid: 1039544) of binary: /fsx/nouamane/miniconda/envs/2-1-cu121/bin/python
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 264, in launch_agent
+    raise ChildFailedError(
+torch.distributed.elastic.multiprocessing.errors.ChildFailedError: 
+============================================================
+/fsx/nouamane/projects/nanotron/run_train.py FAILED
+------------------------------------------------------------
+Failures:
+  <NO_OTHER_FAILURES>
+------------------------------------------------------------
+Root Cause (first observed failure):
+[0]:
+  time      : 2025-01-07_03:50:58
+  host      : ip-26-0-167-9.ec2.internal
+  rank      : 63 (local_rank: 7)
+  exitcode  : 1 (pid: 1039544)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+============================================================
+srun: error: ip-26-0-167-9: task 7: Exited with exit code 1
+srun: launch/slurm: _step_signal: Terminating StepId=14098831.0
+slurmstepd: error: *** STEP 14098831.0 ON ip-26-0-162-46 CANCELLED AT 2025-01-07T03:51:02 ***
+[2025-01-07 03:51:02,296] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-07 03:51:02,297] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 702881 closing signal SIGTERM
+[2025-01-07 03:51:02,297] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 702882 closing signal SIGTERM
+[2025-01-07 03:51:02,296] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-07 03:51:02,297] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 702883 closing signal SIGTERM
+[2025-01-07 03:51:02,297] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 702884 closing signal SIGTERM
+[2025-01-07 03:51:02,296] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 762816 closing signal SIGTERM
+[2025-01-07 03:51:02,297] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 702885 closing signal SIGTERM
+[2025-01-07 03:51:02,296] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 762817 closing signal SIGTERM
+[2025-01-07 03:51:02,296] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 762818 closing signal SIGTERM
+[2025-01-07 03:51:02,296] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 762819 closing signal SIGTERM
+[2025-01-07 03:51:02,296] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 762820 closing signal SIGTERM
+[2025-01-07 03:51:02,297] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-07 03:51:02,296] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 762822 closing signal SIGTERM
+[2025-01-07 03:51:02,297] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-07 03:51:02,297] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-07 03:51:02,297] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 104430 closing signal SIGTERM
+[2025-01-07 03:51:02,297] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 127290 closing signal SIGTERM
+[2025-01-07 03:51:02,297] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 624943 closing signal SIGTERM
+[2025-01-07 03:51:02,297] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 624944 closing signal SIGTERM
+[2025-01-07 03:51:02,297] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 127291 closing signal SIGTERM
+[2025-01-07 03:51:02,296] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-07 03:51:02,297] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 624945 closing signal SIGTERM
+[2025-01-07 03:51:02,297] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 104431 closing signal SIGTERM
+[2025-01-07 03:51:02,297] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 624946 closing signal SIGTERM
+[2025-01-07 03:51:02,297] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 127292 closing signal SIGTERM
+[2025-01-07 03:51:02,298] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-07 03:51:02,297] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 624947 closing signal SIGTERM
+[2025-01-07 03:51:02,297] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 624948 closing signal SIGTERM
+[2025-01-07 03:51:02,297] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 127293 closing signal SIGTERM
+[2025-01-07 03:51:02,296] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 9625 closing signal SIGTERM
+[2025-01-07 03:51:02,297] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 127294 closing signal SIGTERM
+[2025-01-07 03:51:02,297] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 104432 closing signal SIGTERM
+[2025-01-07 03:51:02,297] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 624949 closing signal SIGTERM
+[2025-01-07 03:51:02,297] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 624950 closing signal SIGTERM
+[2025-01-07 03:51:02,297] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 127295 closing signal SIGTERM
+[2025-01-07 03:51:02,297] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 127296 closing signal SIGTERM
+[2025-01-07 03:51:02,298] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 897240 closing signal SIGTERM
+[2025-01-07 03:51:02,297] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 104433 closing signal SIGTERM
+[2025-01-07 03:51:02,299] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 897241 closing signal SIGTERM
+[2025-01-07 03:51:02,296] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 9627 closing signal SIGTERM
+[2025-01-07 03:51:02,299] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 897242 closing signal SIGTERM
+[2025-01-07 03:51:02,299] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 897243 closing signal SIGTERM
+[2025-01-07 03:51:02,297] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 104434 closing signal SIGTERM
+[2025-01-07 03:51:02,299] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 897244 closing signal SIGTERM
+[2025-01-07 03:51:02,296] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 9628 closing signal SIGTERM
+[2025-01-07 03:51:02,299] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 897245 closing signal SIGTERM
+[2025-01-07 03:51:02,297] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 104435 closing signal SIGTERM
+[2025-01-07 03:51:02,299] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 897246 closing signal SIGTERM
+[2025-01-07 03:51:02,296] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 9629 closing signal SIGTERM
+[2025-01-07 03:51:02,296] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 9630 closing signal SIGTERM
+[2025-01-07 03:51:02,296] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 9631 closing signal SIGTERM
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 702806 got signal: 15
+[2025-01-07 03:51:03,547] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-166-214.ec2.internal_624870_0' has failed to send a keep-alive heartbeat to the rendezvous '14098831' due to an error of type RendezvousConnectionError.
+[2025-01-07 03:51:03,573] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-166-15.ec2.internal_127219_0' has failed to send a keep-alive heartbeat to the rendezvous '14098831' due to an error of type RendezvousConnectionError.
+srun: error: ip-26-0-162-46: task 0: Exited with exit code 1
+[2025-01-07 03:51:03,634] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-166-125.ec2.internal_762743_0' has failed to send a keep-alive heartbeat to the rendezvous '14098831' due to an error of type RendezvousConnectionError.
+[2025-01-07 03:51:03,659] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-166-36.ec2.internal_104357_0' has failed to send a keep-alive heartbeat to the rendezvous '14098831' due to an error of type RendezvousConnectionError.
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 9554 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 127219 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 897167 got signal: 15
+srun: error: ip-26-0-166-15: task 2: Exited with exit code 1
+srun: error: ip-26-0-166-244: task 6: Exited with exit code 1
+srun: error: ip-26-0-165-213: task 1: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 762743 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 624870 got signal: 15
+srun: error: ip-26-0-166-125: task 4: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 104357 got signal: 15
+srun: error: ip-26-0-166-214: task 5: Exited with exit code 1
+srun: error: ip-26-0-166-36: task 3: Exited with exit code 1
diff --git a/logs/14098868-bench_3.57G_dp8_tp1_pp8_acc16_mbs2_seq4096_zero1_tpmodeRED_vocab131k.out b/logs/14098868-bench_3.57G_dp8_tp1_pp8_acc16_mbs2_seq4096_zero1_tpmodeRED_vocab131k.out
new file mode 100644
index 0000000000000000000000000000000000000000..48254ea21fa2206e78ebcefaedc2d8ed16d3f15c
--- /dev/null
+++ b/logs/14098868-bench_3.57G_dp8_tp1_pp8_acc16_mbs2_seq4096_zero1_tpmodeRED_vocab131k.out
@@ -0,0 +1,30589 @@
++ '[' -z 14098868 ']'
++ source /etc/profile.d/modules.sh
+++ . /usr/share/modules/init/bash
++++ unset _mlshdbg
++++ '[' 0 = 1 ']'
++++ unset _mlre _mlIFS
++++ '[' -n x ']'
++++ _mlIFS=' 	
+'
++++ IFS=' '
++++ '[' -n '' ']'
+++++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash autoinit
++++ _mlcode='module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+test 0;'
++++ _mlret=0
++++ '[' -n x ']'
++++ IFS=' 	
+'
++++ unset _mlIFS
++++ unset _mlre _mlv _mlrv
++++ '[' 0 -eq 0 ']'
++++ eval 'module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+test 0;'
+++++ MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl
+++++ export MODULES_CMD
+++++ ENV=/usr/share/modules/init/profile.sh
+++++ export ENV
+++++ BASH_ENV=/usr/share/modules/init/bash
+++++ export BASH_ENV
+++++ MODULESHOME=/usr/share/modules
+++++ export MODULESHOME
+++++ test 0
++++ '[' 0 = 1 ']'
++++ '[' -t 2 ']'
++++ export -f module
++++ export -f switchml
++++ '[' 5 -ge 3 ']'
++++ [[ ehxB =~ i ]]
++++ [[ ! :/admin/home/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/admin/home/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin: =~ :/usr/bin: ]]
++++ '[' '!' -n x ']'
+++++ manpath
++++ [[ ! :/admin/home/nouamane/miniconda/envs/2-1-cu121/man:/admin/home/nouamane/miniconda/envs/2-1-cu121/share/man:/fsx/nouamane/miniconda/man:/fsx/nouamane/miniconda/share/man:/opt/slurm/share/man:/opt/amazon/openmpi/share/man:/opt/amazon/efa/share/man:/usr/local/man:/usr/local/share/man:/usr/share/man:/admin/home/nouamane/.local/share/man:/admin/home/nouamane/.fzf/man:: =~ :/usr/share/man: ]]
++++ unset _mlcode _mlret
++++ '[' -n '' ']'
++ module load cuda/12.1
++ unset _mlshdbg
++ '[' 0 = 1 ']'
++ unset _mlre _mlIFS
++ '[' -n x ']'
++ _mlIFS=' 	
+'
++ IFS=' '
++ '[' -n '' ']'
+++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash load cuda/12.1
++ eval 'CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include;' export 'CPATH;
+LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/opt/aws-ofi-nccl/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/targets/x86_64-linux/lib/:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/lib:/usr/lib;' export 'LD_LIBRARY_PATH;
+FI_EFA_FORK_SAFE=1;' export 'FI_EFA_FORK_SAFE;
+MANPATH=/usr/local/cuda-12.1/share/man::;' export 'MANPATH;
+LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64;' export 'LIBRARY_PATH;
+_LMFILES_=/usr/share/modules/modulefiles/cuda/12.1;' export '_LMFILES_;
+LOADEDMODULES=cuda/12.1;' export 'LOADEDMODULES;
+MPI_PATH=/opt/amazon/openmpi;' export 'MPI_PATH;
+NCCL_HOME_modshare=/opt/nccl/build:1;' export 'NCCL_HOME_modshare;
+NCCL_PROTO=simple;' export 'NCCL_PROTO;
+MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1;' export 'MANPATH_modshare;
+LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1;' export 'LIBRARY_PATH_modshare;
+NCCL_SOCKET_IFNAME=enp;' export 'NCCL_SOCKET_IFNAME;
+AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl;' export 'AWS_OFI_NCCL_HOME;
+NCCL_HOME=/opt/nccl/build;' export 'NCCL_HOME;
+FI_PROVIDER=efa;' export 'FI_PROVIDER;
+AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1;' export 'AWS_OFI_NCCL_HOME_modshare;
+CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1;' export 'CPATH_modshare;
+LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib/:1:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2:/opt/aws-ofi-nccl/lib:2:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/efa/lib:1:/usr/lib:1;' export 'LD_LIBRARY_PATH_modshare;
+FI_EFA_ENABLE_SHM_TRANSFER=1;' export 'FI_EFA_ENABLE_SHM_TRANSFER;
+_LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1;' export '_LMFILES__modshare;
+LOADEDMODULES_modshare=cuda/12.1:1;' export 'LOADEDMODULES_modshare;
+MPI_PATH_modshare=/opt/amazon/openmpi:1;' export 'MPI_PATH_modshare;
+PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/admin/home/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin;' export 'PATH;
+CUDA_HOME=/usr/local/cuda-12.1;' export 'CUDA_HOME;
+PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/local/cuda-12.1/include:1:/usr/bin:1:/opt/amazon/efa/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/opt/slurm/bin:1:/usr/sbin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:2:/opt/amazon/openmpi/bin:1:/admin/home/nouamane/.fzf/bin:1:/admin/home/nouamane/miniconda/envs/2-1-cu121/bin:1:/usr/local/games:1;' export 'PATH_modshare;
+test' '0;'
+++ CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include
+++ export CPATH
+++ LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/opt/aws-ofi-nccl/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/targets/x86_64-linux/lib/:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/lib:/usr/lib
+++ export LD_LIBRARY_PATH
+++ FI_EFA_FORK_SAFE=1
+++ export FI_EFA_FORK_SAFE
+++ MANPATH=/usr/local/cuda-12.1/share/man::
+++ export MANPATH
+++ LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64
+++ export LIBRARY_PATH
+++ _LMFILES_=/usr/share/modules/modulefiles/cuda/12.1
+++ export _LMFILES_
+++ LOADEDMODULES=cuda/12.1
+++ export LOADEDMODULES
+++ MPI_PATH=/opt/amazon/openmpi
+++ export MPI_PATH
+++ NCCL_HOME_modshare=/opt/nccl/build:1
+++ export NCCL_HOME_modshare
+++ NCCL_PROTO=simple
+++ export NCCL_PROTO
+++ MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1
+++ export MANPATH_modshare
+++ LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1
+++ export LIBRARY_PATH_modshare
+++ NCCL_SOCKET_IFNAME=enp
+++ export NCCL_SOCKET_IFNAME
+++ AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl
+++ export AWS_OFI_NCCL_HOME
+++ NCCL_HOME=/opt/nccl/build
+++ export NCCL_HOME
+++ FI_PROVIDER=efa
+++ export FI_PROVIDER
+++ AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1
+++ export AWS_OFI_NCCL_HOME_modshare
+++ CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1
+++ export CPATH_modshare
+++ LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib/:1:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2:/opt/aws-ofi-nccl/lib:2:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/efa/lib:1:/usr/lib:1
+++ export LD_LIBRARY_PATH_modshare
+++ FI_EFA_ENABLE_SHM_TRANSFER=1
+++ export FI_EFA_ENABLE_SHM_TRANSFER
+++ _LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1
+++ export _LMFILES__modshare
+++ LOADEDMODULES_modshare=cuda/12.1:1
+++ export LOADEDMODULES_modshare
+++ MPI_PATH_modshare=/opt/amazon/openmpi:1
+++ export MPI_PATH_modshare
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/admin/home/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/admin/home/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
+++ export PATH
+++ CUDA_HOME=/usr/local/cuda-12.1
+++ export CUDA_HOME
+++ PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/local/cuda-12.1/include:1:/usr/bin:1:/opt/amazon/efa/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/opt/slurm/bin:1:/usr/sbin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:2:/opt/amazon/openmpi/bin:1:/admin/home/nouamane/.fzf/bin:1:/admin/home/nouamane/miniconda/envs/2-1-cu121/bin:1:/usr/local/games:1
+++ export PATH_modshare
+++ test 0
++ _mlstatus=0
++ '[' -n x ']'
++ IFS=' 	
+'
++ unset _mlre _mlv _mlrv _mlIFS
++ '[' -n '' ']'
++ unset _mlshdbg
++ return 0
++ source /fsx/nouamane/miniconda/bin/activate
+++ _CONDA_ROOT=/fsx/nouamane/miniconda
+++ . /fsx/nouamane/miniconda/etc/profile.d/conda.sh
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ '[' -z x ']'
+++ conda activate
+++ local cmd=activate
+++ case "$cmd" in
+++ __conda_activate activate
+++ '[' -n '' ']'
+++ local ask_conda
++++ PS1=
++++ __conda_exe shell.posix activate
++++ /fsx/nouamane/miniconda/bin/conda shell.posix activate
+Error while loading conda entry point: conda-libmamba-solver (module 'libmambapy' has no attribute 'QueryFormat')
+++ ask_conda='. "/admin/home/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/fsx/nouamane/miniconda/bin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/admin/home/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_2='\''/admin/home/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/etc/conda/activate.d/libxml2_activate.sh"'
+++ eval '. "/admin/home/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/fsx/nouamane/miniconda/bin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/admin/home/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_2='\''/admin/home/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/etc/conda/activate.d/libxml2_activate.sh"'
++++ . /admin/home/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh
+++++ test -n ''
+++++ unset XML_CATALOG_FILES
+++++ unset xml_catalog_files_libxml2
++++ PS1='(base) '
++++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/fsx/nouamane/miniconda/bin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/admin/home/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
++++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/fsx/nouamane/miniconda/bin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/admin/home/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
++++ export CONDA_PREFIX=/fsx/nouamane/miniconda
++++ CONDA_PREFIX=/fsx/nouamane/miniconda
++++ export CONDA_SHLVL=3
++++ CONDA_SHLVL=3
++++ export CONDA_DEFAULT_ENV=base
++++ CONDA_DEFAULT_ENV=base
++++ export 'CONDA_PROMPT_MODIFIER=(base) '
++++ CONDA_PROMPT_MODIFIER='(base) '
++++ export CONDA_PREFIX_2=/admin/home/nouamane/miniconda/envs/2-1-cu121
++++ CONDA_PREFIX_2=/admin/home/nouamane/miniconda/envs/2-1-cu121
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ . /fsx/nouamane/miniconda/etc/conda/activate.d/libxml2_activate.sh
+++++ test -n ''
+++++ xml_catalog_files_libxml2=
+++++ XML_CATALOG_FILES=
+++++ conda_catalog_files=
+++++ ifs_libxml2=' 	
+'
+++++ IFS=' '
+++++ rem=/fsx/nouamane/miniconda
+++++ for pre in ${rem}
+++++ test '' = /fsx/nouamane/miniconda
+++++ conda_catalog_files=/fsx/nouamane/miniconda
+++++ rem=
+++++ IFS=' 	
+'
+++++ conda_catalog_files='file:///fsx/nouamane/miniconda/etc/xml/catalog file:///etc/xml/catalog'
+++++ export 'XML_CATALOG_FILES=file:///fsx/nouamane/miniconda/etc/xml/catalog file:///etc/xml/catalog'
+++++ XML_CATALOG_FILES='file:///fsx/nouamane/miniconda/etc/xml/catalog file:///etc/xml/catalog'
+++++ unset conda_catalog_files ifs_libxml2 rem
+++ __conda_hashr
+++ '[' -n '' ']'
+++ '[' -n '' ']'
+++ hash -r
++ conda activate 2-1-cu121
++ local cmd=activate
++ case "$cmd" in
++ __conda_activate activate 2-1-cu121
++ '[' -n '' ']'
++ local ask_conda
+++ PS1='(base) '
+++ __conda_exe shell.posix activate 2-1-cu121
+++ /fsx/nouamane/miniconda/bin/conda shell.posix activate 2-1-cu121
+Error while loading conda entry point: conda-libmamba-solver (module 'libmambapy' has no attribute 'QueryFormat')
++ ask_conda='. "/fsx/nouamane/miniconda/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/admin/home/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''4'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_3='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
++ eval '. "/fsx/nouamane/miniconda/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/admin/home/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''4'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_3='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
+++ . /fsx/nouamane/miniconda/etc/conda/deactivate.d/libxml2_deactivate.sh
++++ test -n ''
++++ unset XML_CATALOG_FILES
++++ unset xml_catalog_files_libxml2
+++ PS1='(2-1-cu121) '
+++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/admin/home/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/admin/home/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
+++ export CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ export CONDA_SHLVL=4
+++ CONDA_SHLVL=4
+++ export CONDA_DEFAULT_ENV=2-1-cu121
+++ CONDA_DEFAULT_ENV=2-1-cu121
+++ export 'CONDA_PROMPT_MODIFIER=(2-1-cu121) '
+++ CONDA_PROMPT_MODIFIER='(2-1-cu121) '
+++ export CONDA_PREFIX_3=/fsx/nouamane/miniconda
+++ CONDA_PREFIX_3=/fsx/nouamane/miniconda
+++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ export _CE_M=
+++ _CE_M=
+++ export _CE_CONDA=
+++ _CE_CONDA=
+++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh
++++ test -n ''
++++ xml_catalog_files_libxml2=
++++ XML_CATALOG_FILES=
++++ conda_catalog_files=
++++ ifs_libxml2=' 	
+'
++++ IFS=' '
++++ rem=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ for pre in ${rem}
++++ test '' = /fsx/nouamane/miniconda/envs/2-1-cu121
++++ conda_catalog_files=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ rem=
++++ IFS=' 	
+'
++++ conda_catalog_files='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ export 'XML_CATALOG_FILES=file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ XML_CATALOG_FILES='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ unset conda_catalog_files ifs_libxml2 rem
++ __conda_hashr
++ '[' -n '' ']'
++ '[' -n '' ']'
++ hash -r
++ export PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/admin/home/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
++ PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/admin/home/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/opt/slurm/bin:/opt/amazon/openmpi/bin:/opt/amazon/efa/bin:/usr/local/cuda-12.1/bin:/usr/local/cuda-12.1/include:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
++ '[' -z '' ']'
+++ scontrol show hostnames 'ip-26-0-160-225,ip-26-0-172-[57,73,116,142,147,252],ip-26-0-173-7'
++ export 'NODELIST=ip-26-0-160-225
+ip-26-0-172-57
+ip-26-0-172-73
+ip-26-0-172-116
+ip-26-0-172-142
+ip-26-0-172-147
+ip-26-0-172-252
+ip-26-0-173-7'
++ NODELIST='ip-26-0-160-225
+ip-26-0-172-57
+ip-26-0-172-73
+ip-26-0-172-116
+ip-26-0-172-142
+ip-26-0-172-147
+ip-26-0-172-252
+ip-26-0-173-7'
+++ head -n1
+++ scontrol show hostnames 'ip-26-0-160-225,ip-26-0-172-[57,73,116,142,147,252],ip-26-0-173-7'
++ export MASTER_NODE=ip-26-0-160-225
++ MASTER_NODE=ip-26-0-160-225
++ export MASTER_PORT=12356
++ MASTER_PORT=12356
++ export NNODES=8
++ NNODES=8
++ export GPUS_PER_NODE=8
++ GPUS_PER_NODE=8
++ export WORLD_SIZE=64
++ WORLD_SIZE=64
++ export CUDA_DEVICE_MAX_CONNECTIONS=1
++ CUDA_DEVICE_MAX_CONNECTIONS=1
++ export NCCL_DEBUG=INFO
++ NCCL_DEBUG=INFO
++ export NANOTRON_BENCHMARK=1
++ NANOTRON_BENCHMARK=1
++ export WANDB_MODE=disabled
++ WANDB_MODE=disabled
++ export TORCH_NCCL_ASYNC_ERROR_HANDLING=1
++ TORCH_NCCL_ASYNC_ERROR_HANDLING=1
++ export TORCH_DISTRIBUTED_DEBUG=DETAIL
++ TORCH_DISTRIBUTED_DEBUG=DETAIL
++ '[' -z '' ']'
++ echo '=== GPU Topology ==='
+=== GPU Topology ===
++ nvidia-smi topo -m
+	GPU0	GPU1	GPU2	GPU3	GPU4	GPU5	GPU6	GPU7	CPU Affinity	NUMA Affinity	GPU NUMA ID
+GPU0	 X 	NV18	NV18	NV18	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU1	NV18	 X 	NV18	NV18	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU2	NV18	NV18	 X 	NV18	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU3	NV18	NV18	NV18	 X 	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU4	NV18	NV18	NV18	NV18	 X 	NV18	NV18	NV18	48-95	1		N/A
+GPU5	NV18	NV18	NV18	NV18	NV18	 X 	NV18	NV18	48-95	1		N/A
+GPU6	NV18	NV18	NV18	NV18	NV18	NV18	 X 	NV18	48-95	1		N/A
+GPU7	NV18	NV18	NV18	NV18	NV18	NV18	NV18	 X 	48-95	1		N/A
+
+Legend:
+
+  X    = Self
+  SYS  = Connection traversing PCIe as well as the SMP interconnect between NUMA nodes (e.g., QPI/UPI)
+  NODE = Connection traversing PCIe as well as the interconnect between PCIe Host Bridges within a NUMA node
+  PHB  = Connection traversing PCIe as well as a PCIe Host Bridge (typically the CPU)
+  PXB  = Connection traversing multiple PCIe bridges (without traversing the PCIe Host Bridge)
+  PIX  = Connection traversing at most a single PCIe bridge
+  NV#  = Connection traversing a bonded set of # NVLinks
++ echo ==================
+==================
++ export SRUN_ALLOC_ARGS=
++ SRUN_ALLOC_ARGS=
++ echo 'Master node: ip-26-0-160-225'
+Master node: ip-26-0-160-225
++ echo 'All nodes: ip-26-0-160-225
+ip-26-0-172-57
+ip-26-0-172-73
+ip-26-0-172-116
+ip-26-0-172-142
+ip-26-0-172-147
+ip-26-0-172-252
+ip-26-0-173-7'
+All nodes: ip-26-0-160-225
+ip-26-0-172-57
+ip-26-0-172-73
+ip-26-0-172-116
+ip-26-0-172-142
+ip-26-0-172-147
+ip-26-0-172-252
+ip-26-0-173-7
++ echo 'World size: 64'
+World size: 64
++ '[' -n '' ']'
++ srun --wait=0 --kill-on-bad-exit=1 torchrun --nnodes=8 --nproc_per_node=8 --rdzv_id=14098868 --rdzv_backend=c10d --rdzv_endpoint=ip-26-0-160-225:12356 --max_restarts 0 --rdzv_conf timeout=60 /fsx/nouamane/projects/nanotron/run_train.py --config-file benchmark/configs/config_3.57G_dp8_tp1_pp8_acc16_mbs2_seq4096_zero1_tpmodeRED_vocab131k.yaml
+[2025-01-07 05:26:34,824] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-07 05:26:34,824] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-07 05:26:34,825] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-07 05:26:34,825] torch.distributed.run: [WARNING] 
+[2025-01-07 05:26:34,825] torch.distributed.run: [WARNING] *****************************************
+[2025-01-07 05:26:34,825] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-07 05:26:34,825] torch.distributed.run: [WARNING] *****************************************
+[2025-01-07 05:26:34,824] torch.distributed.run: [WARNING] 
+[2025-01-07 05:26:34,824] torch.distributed.run: [WARNING] *****************************************
+[2025-01-07 05:26:34,824] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-07 05:26:34,824] torch.distributed.run: [WARNING] *****************************************
+[2025-01-07 05:26:34,869] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-07 05:26:34,826] torch.distributed.run: [WARNING] 
+[2025-01-07 05:26:34,826] torch.distributed.run: [WARNING] *****************************************
+[2025-01-07 05:26:34,826] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-07 05:26:34,826] torch.distributed.run: [WARNING] *****************************************
+[2025-01-07 05:26:34,876] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-07 05:26:34,890] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-07 05:26:34,869] torch.distributed.run: [WARNING] 
+[2025-01-07 05:26:34,869] torch.distributed.run: [WARNING] *****************************************
+[2025-01-07 05:26:34,869] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-07 05:26:34,869] torch.distributed.run: [WARNING] *****************************************
+[2025-01-07 05:26:34,876] torch.distributed.run: [WARNING] 
+[2025-01-07 05:26:34,876] torch.distributed.run: [WARNING] *****************************************
+[2025-01-07 05:26:34,876] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-07 05:26:34,876] torch.distributed.run: [WARNING] *****************************************
+[2025-01-07 05:26:34,923] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-07 05:26:34,890] torch.distributed.run: [WARNING] 
+[2025-01-07 05:26:34,890] torch.distributed.run: [WARNING] *****************************************
+[2025-01-07 05:26:34,890] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-07 05:26:34,890] torch.distributed.run: [WARNING] *****************************************
+[2025-01-07 05:26:34,923] torch.distributed.run: [WARNING] 
+[2025-01-07 05:26:34,923] torch.distributed.run: [WARNING] *****************************************
+[2025-01-07 05:26:34,923] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-07 05:26:34,923] torch.distributed.run: [WARNING] *****************************************
+[2025-01-07 05:26:35,006] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-07 05:26:35,007] torch.distributed.run: [WARNING] 
+[2025-01-07 05:26:35,007] torch.distributed.run: [WARNING] *****************************************
+[2025-01-07 05:26:35,007] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-07 05:26:35,007] torch.distributed.run: [WARNING] *****************************************
+ip-26-0-160-225:94342:94342 [0] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-160-225:94342:94342 [0] NCCL INFO Bootstrap : Using enp71s0:26.0.160.225<0>
+ip-26-0-160-225:94342:94342 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-160-225:94342:94342 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-160-225:94342:94342 [0] NCCL INFO cudaDriverVersion 12020
+NCCL version 2.18.6+cuda12.1
+ip-26-0-160-225:94346:94346 [4] NCCL INFO cudaDriverVersion 12020
+ip-26-0-160-225:94345:94345 [3] NCCL INFO cudaDriverVersion 12020
+ip-26-0-160-225:94343:94343 [1] NCCL INFO cudaDriverVersion 12020
+ip-26-0-160-225:94347:94347 [5] NCCL INFO cudaDriverVersion 12020
+ip-26-0-160-225:94344:94344 [2] NCCL INFO cudaDriverVersion 12020
+ip-26-0-160-225:94349:94349 [7] NCCL INFO cudaDriverVersion 12020
+ip-26-0-160-225:94348:94348 [6] NCCL INFO cudaDriverVersion 12020
+ip-26-0-160-225:94346:94346 [4] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-160-225:94345:94345 [3] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-172-73:926601:926601 [7] NCCL INFO cudaDriverVersion 12020
+ip-26-0-172-73:926596:926596 [2] NCCL INFO cudaDriverVersion 12020
+ip-26-0-160-225:94343:94343 [1] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-172-73:926599:926599 [5] NCCL INFO cudaDriverVersion 12020
+ip-26-0-160-225:94347:94347 [5] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-172-73:926597:926597 [3] NCCL INFO cudaDriverVersion 12020
+ip-26-0-172-73:926598:926598 [4] NCCL INFO cudaDriverVersion 12020
+ip-26-0-172-73:926600:926600 [6] NCCL INFO cudaDriverVersion 12020
+ip-26-0-172-73:926594:926594 [0] NCCL INFO cudaDriverVersion 12020
+ip-26-0-172-73:926595:926595 [1] NCCL INFO cudaDriverVersion 12020
+ip-26-0-172-57:942615:942615 [4] NCCL INFO cudaDriverVersion 12020
+ip-26-0-172-57:942612:942612 [1] NCCL INFO cudaDriverVersion 12020
+ip-26-0-172-57:942618:942618 [7] NCCL INFO cudaDriverVersion 12020
+ip-26-0-172-116:1019295:1019295 [7] NCCL INFO cudaDriverVersion 12020
+ip-26-0-172-116:1019293:1019293 [5] NCCL INFO cudaDriverVersion 12020
+ip-26-0-160-225:94349:94349 [7] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-160-225:94344:94344 [2] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-173-7:811248:811248 [0] NCCL INFO cudaDriverVersion 12020
+ip-26-0-173-7:811249:811249 [1] NCCL INFO cudaDriverVersion 12020
+ip-26-0-173-7:811250:811250 [2] NCCL INFO cudaDriverVersion 12020
+ip-26-0-172-116:1019291:1019291 [3] NCCL INFO cudaDriverVersion 12020
+ip-26-0-172-116:1019294:1019294 [6] NCCL INFO cudaDriverVersion 12020
+ip-26-0-172-57:942614:942614 [3] NCCL INFO cudaDriverVersion 12020
+ip-26-0-172-116:1019290:1019290 [2] NCCL INFO cudaDriverVersion 12020
+ip-26-0-172-57:942616:942616 [5] NCCL INFO cudaDriverVersion 12020
+ip-26-0-172-57:942611:942611 [0] NCCL INFO cudaDriverVersion 12020
+ip-26-0-172-57:942617:942617 [6] NCCL INFO cudaDriverVersion 12020
+ip-26-0-172-116:1019288:1019288 [0] NCCL INFO cudaDriverVersion 12020
+ip-26-0-172-116:1019292:1019292 [4] NCCL INFO cudaDriverVersion 12020
+ip-26-0-160-225:94348:94348 [6] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-173-7:811252:811252 [4] NCCL INFO cudaDriverVersion 12020
+ip-26-0-173-7:811253:811253 [5] NCCL INFO cudaDriverVersion 12020
+ip-26-0-172-57:942613:942613 [2] NCCL INFO cudaDriverVersion 12020
+ip-26-0-172-142:880686:880686 [7] NCCL INFO cudaDriverVersion 12020
+ip-26-0-172-142:880683:880683 [4] NCCL INFO cudaDriverVersion 12020
+ip-26-0-172-147:618584:618584 [0] NCCL INFO cudaDriverVersion 12020
+ip-26-0-172-147:618591:618591 [7] NCCL INFO cudaDriverVersion 12020
+ip-26-0-172-142:880679:880679 [0] NCCL INFO cudaDriverVersion 12020
+ip-26-0-172-142:880684:880684 [5] NCCL INFO cudaDriverVersion 12020
+ip-26-0-172-252:688201:688201 [1] NCCL INFO cudaDriverVersion 12020
+ip-26-0-172-252:688207:688207 [7] NCCL INFO cudaDriverVersion 12020
+ip-26-0-172-252:688202:688202 [2] NCCL INFO cudaDriverVersion 12020
+ip-26-0-172-252:688200:688200 [0] NCCL INFO cudaDriverVersion 12020
+ip-26-0-173-7:811254:811254 [6] NCCL INFO cudaDriverVersion 12020
+ip-26-0-172-116:1019289:1019289 [1] NCCL INFO cudaDriverVersion 12020
+ip-26-0-172-142:880681:880681 [2] NCCL INFO cudaDriverVersion 12020
+ip-26-0-172-142:880682:880682 [3] NCCL INFO cudaDriverVersion 12020
+ip-26-0-172-142:880685:880685 [6] NCCL INFO cudaDriverVersion 12020
+ip-26-0-172-142:880680:880680 [1] NCCL INFO cudaDriverVersion 12020
+ip-26-0-172-147:618585:618585 [1] NCCL INFO cudaDriverVersion 12020
+ip-26-0-173-7:811255:811255 [7] NCCL INFO cudaDriverVersion 12020
+ip-26-0-172-252:688205:688205 [5] NCCL INFO cudaDriverVersion 12020
+ip-26-0-173-7:811251:811251 [3] NCCL INFO cudaDriverVersion 12020
+ip-26-0-172-252:688204:688204 [4] NCCL INFO cudaDriverVersion 12020
+ip-26-0-172-147:618587:618587 [3] NCCL INFO cudaDriverVersion 12020
+ip-26-0-172-147:618588:618588 [4] NCCL INFO cudaDriverVersion 12020
+ip-26-0-172-147:618590:618590 [6] NCCL INFO cudaDriverVersion 12020
+ip-26-0-172-147:618586:618586 [2] NCCL INFO cudaDriverVersion 12020
+ip-26-0-172-147:618589:618589 [5] NCCL INFO cudaDriverVersion 12020
+ip-26-0-172-252:688203:688203 [3] NCCL INFO cudaDriverVersion 12020
+ip-26-0-172-252:688206:688206 [6] NCCL INFO cudaDriverVersion 12020
+ip-26-0-172-73:926594:926594 [0] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-172-73:926601:926601 [7] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-172-116:1019295:1019295 [7] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-172-116:1019294:1019294 [6] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-172-73:926600:926600 [6] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-172-73:926597:926597 [3] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-172-73:926598:926598 [4] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-172-73:926595:926595 [1] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-172-116:1019292:1019292 [4] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-172-116:1019288:1019288 [0] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-172-142:880686:880686 [7] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-172-142:880682:880682 [3] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-173-7:811248:811248 [0] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-172-116:1019289:1019289 [1] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-172-73:926599:926599 [5] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-173-7:811251:811251 [3] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-172-116:1019291:1019291 [3] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-172-57:942614:942614 [3] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-172-142:880683:880683 [4] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-173-7:811254:811254 [6] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-172-116:1019290:1019290 [2] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-172-116:1019293:1019293 [5] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-172-57:942612:942612 [1] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-172-142:880684:880684 [5] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-172-57:942615:942615 [4] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-173-7:811253:811253 [5] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-173-7:811255:811255 [7] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-172-73:926596:926596 [2] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-172-57:942616:942616 [5] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-172-142:880680:880680 [1] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-173-7:811250:811250 [2] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-172-252:688207:688207 [7] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-173-7:811252:811252 [4] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-172-57:942617:942617 [6] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-172-147:618584:618584 [0] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-172-147:618591:618591 [7] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-172-142:880679:880679 [0] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-172-142:880681:880681 [2] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-172-142:880685:880685 [6] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-172-252:688206:688206 [6] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-172-57:942611:942611 [0] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-172-57:942618:942618 [7] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-172-147:618587:618587 [3] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-172-147:618585:618585 [1] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-172-252:688201:688201 [1] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-172-147:618588:618588 [4] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-172-57:942613:942613 [2] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-173-7:811249:811249 [1] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-172-252:688200:688200 [0] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-172-147:618586:618586 [2] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-172-252:688202:688202 [2] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-172-252:688204:688204 [4] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-172-147:618589:618589 [5] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-172-147:618590:618590 [6] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-172-252:688203:688203 [3] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-172-252:688205:688205 [5] NCCL INFO NCCL_SOCKET_IFNAME set by environment to enp
+ip-26-0-160-225:94346:94346 [4] NCCL INFO Bootstrap : Using enp71s0:26.0.160.225<0>
+ip-26-0-160-225:94345:94345 [3] NCCL INFO Bootstrap : Using enp71s0:26.0.160.225<0>
+ip-26-0-160-225:94343:94343 [1] NCCL INFO Bootstrap : Using enp71s0:26.0.160.225<0>
+ip-26-0-160-225:94349:94349 [7] NCCL INFO Bootstrap : Using enp71s0:26.0.160.225<0>
+ip-26-0-160-225:94344:94344 [2] NCCL INFO Bootstrap : Using enp71s0:26.0.160.225<0>
+ip-26-0-160-225:94347:94347 [5] NCCL INFO Bootstrap : Using enp71s0:26.0.160.225<0>
+ip-26-0-160-225:94348:94348 [6] NCCL INFO Bootstrap : Using enp71s0:26.0.160.225<0>
+ip-26-0-160-225:94346:94346 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-160-225:94346:94346 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-160-225:94345:94345 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-160-225:94345:94345 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-160-225:94343:94343 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-160-225:94343:94343 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-160-225:94348:94348 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-160-225:94348:94348 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-160-225:94349:94349 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-160-225:94349:94349 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-160-225:94344:94344 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-160-225:94344:94344 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-160-225:94347:94347 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-160-225:94347:94347 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-172-57:942614:942614 [3] NCCL INFO Bootstrap : Using enp71s0:26.0.172.57<0>
+ip-26-0-172-57:942615:942615 [4] NCCL INFO Bootstrap : Using enp71s0:26.0.172.57<0>
+ip-26-0-172-57:942616:942616 [5] NCCL INFO Bootstrap : Using enp71s0:26.0.172.57<0>
+ip-26-0-172-57:942617:942617 [6] NCCL INFO Bootstrap : Using enp71s0:26.0.172.57<0>
+ip-26-0-172-57:942612:942612 [1] NCCL INFO Bootstrap : Using enp71s0:26.0.172.57<0>
+ip-26-0-172-57:942611:942611 [0] NCCL INFO Bootstrap : Using enp71s0:26.0.172.57<0>
+ip-26-0-172-57:942618:942618 [7] NCCL INFO Bootstrap : Using enp71s0:26.0.172.57<0>
+ip-26-0-172-57:942613:942613 [2] NCCL INFO Bootstrap : Using enp71s0:26.0.172.57<0>
+ip-26-0-172-142:880682:880682 [3] NCCL INFO Bootstrap : Using enp71s0:26.0.172.142<0>
+ip-26-0-172-142:880683:880683 [4] NCCL INFO Bootstrap : Using enp71s0:26.0.172.142<0>
+ip-26-0-172-142:880685:880685 [6] NCCL INFO Bootstrap : Using enp71s0:26.0.172.142<0>
+ip-26-0-172-142:880686:880686 [7] NCCL INFO Bootstrap : Using enp71s0:26.0.172.142<0>
+ip-26-0-172-142:880679:880679 [0] NCCL INFO Bootstrap : Using enp71s0:26.0.172.142<0>
+ip-26-0-172-142:880684:880684 [5] NCCL INFO Bootstrap : Using enp71s0:26.0.172.142<0>
+ip-26-0-172-142:880681:880681 [2] NCCL INFO Bootstrap : Using enp71s0:26.0.172.142<0>
+ip-26-0-172-142:880680:880680 [1] NCCL INFO Bootstrap : Using enp71s0:26.0.172.142<0>
+ip-26-0-172-147:618584:618584 [0] NCCL INFO Bootstrap : Using enp71s0:26.0.172.147<0>
+ip-26-0-172-147:618591:618591 [7] NCCL INFO Bootstrap : Using enp71s0:26.0.172.147<0>
+ip-26-0-172-147:618587:618587 [3] NCCL INFO Bootstrap : Using enp71s0:26.0.172.147<0>
+ip-26-0-172-147:618586:618586 [2] NCCL INFO Bootstrap : Using enp71s0:26.0.172.147<0>
+ip-26-0-172-147:618589:618589 [5] NCCL INFO Bootstrap : Using enp71s0:26.0.172.147<0>
+ip-26-0-172-147:618590:618590 [6] NCCL INFO Bootstrap : Using enp71s0:26.0.172.147<0>
+ip-26-0-172-147:618585:618585 [1] NCCL INFO Bootstrap : Using enp71s0:26.0.172.147<0>
+ip-26-0-172-147:618588:618588 [4] NCCL INFO Bootstrap : Using enp71s0:26.0.172.147<0>
+ip-26-0-172-252:688207:688207 [7] NCCL INFO Bootstrap : Using enp71s0:26.0.172.252<0>
+ip-26-0-172-252:688206:688206 [6] NCCL INFO Bootstrap : Using enp71s0:26.0.172.252<0>
+ip-26-0-172-116:1019295:1019295 [7] NCCL INFO Bootstrap : Using enp71s0:26.0.172.116<0>
+ip-26-0-172-116:1019288:1019288 [0] NCCL INFO Bootstrap : Using enp71s0:26.0.172.116<0>
+ip-26-0-172-252:688200:688200 [0] NCCL INFO Bootstrap : Using enp71s0:26.0.172.252<0>
+ip-26-0-172-73:926601:926601 [7] NCCL INFO Bootstrap : Using enp71s0:26.0.172.73<0>
+ip-26-0-172-73:926594:926594 [0] NCCL INFO Bootstrap : Using enp71s0:26.0.172.73<0>
+ip-26-0-172-252:688202:688202 [2] NCCL INFO Bootstrap : Using enp71s0:26.0.172.252<0>
+ip-26-0-172-252:688205:688205 [5] NCCL INFO Bootstrap : Using enp71s0:26.0.172.252<0>
+ip-26-0-172-252:688203:688203 [3] NCCL INFO Bootstrap : Using enp71s0:26.0.172.252<0>
+ip-26-0-172-252:688204:688204 [4] NCCL INFO Bootstrap : Using enp71s0:26.0.172.252<0>
+ip-26-0-172-252:688201:688201 [1] NCCL INFO Bootstrap : Using enp71s0:26.0.172.252<0>
+ip-26-0-172-116:1019292:1019292 [4] NCCL INFO Bootstrap : Using enp71s0:26.0.172.116<0>
+ip-26-0-172-116:1019293:1019293 [5] NCCL INFO Bootstrap : Using enp71s0:26.0.172.116<0>
+ip-26-0-172-116:1019294:1019294 [6] NCCL INFO Bootstrap : Using enp71s0:26.0.172.116<0>
+ip-26-0-172-116:1019289:1019289 [1] NCCL INFO Bootstrap : Using enp71s0:26.0.172.116<0>
+ip-26-0-172-116:1019290:1019290 [2] NCCL INFO Bootstrap : Using enp71s0:26.0.172.116<0>
+ip-26-0-172-116:1019291:1019291 [3] NCCL INFO Bootstrap : Using enp71s0:26.0.172.116<0>
+ip-26-0-172-73:926598:926598 [4] NCCL INFO Bootstrap : Using enp71s0:26.0.172.73<0>
+ip-26-0-172-73:926600:926600 [6] NCCL INFO Bootstrap : Using enp71s0:26.0.172.73<0>
+ip-26-0-172-73:926597:926597 [3] NCCL INFO Bootstrap : Using enp71s0:26.0.172.73<0>
+ip-26-0-172-73:926595:926595 [1] NCCL INFO Bootstrap : Using enp71s0:26.0.172.73<0>
+ip-26-0-172-73:926599:926599 [5] NCCL INFO Bootstrap : Using enp71s0:26.0.172.73<0>
+ip-26-0-172-73:926596:926596 [2] NCCL INFO Bootstrap : Using enp71s0:26.0.172.73<0>
+ip-26-0-172-57:942617:942617 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-172-57:942616:942616 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-172-57:942612:942612 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-172-57:942615:942615 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-172-57:942611:942611 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-172-57:942614:942614 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-172-57:942612:942612 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-172-57:942617:942617 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-172-57:942616:942616 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-172-57:942611:942611 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-172-57:942614:942614 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-172-57:942615:942615 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-172-57:942613:942613 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-172-57:942613:942613 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-173-7:811255:811255 [7] NCCL INFO Bootstrap : Using enp71s0:26.0.173.7<0>
+ip-26-0-173-7:811248:811248 [0] NCCL INFO Bootstrap : Using enp71s0:26.0.173.7<0>
+ip-26-0-173-7:811251:811251 [3] NCCL INFO Bootstrap : Using enp71s0:26.0.173.7<0>
+ip-26-0-173-7:811254:811254 [6] NCCL INFO Bootstrap : Using enp71s0:26.0.173.7<0>
+ip-26-0-173-7:811253:811253 [5] NCCL INFO Bootstrap : Using enp71s0:26.0.173.7<0>
+ip-26-0-173-7:811252:811252 [4] NCCL INFO Bootstrap : Using enp71s0:26.0.173.7<0>
+ip-26-0-173-7:811249:811249 [1] NCCL INFO Bootstrap : Using enp71s0:26.0.173.7<0>
+ip-26-0-172-57:942618:942618 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-172-57:942618:942618 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-173-7:811250:811250 [2] NCCL INFO Bootstrap : Using enp71s0:26.0.173.7<0>
+ip-26-0-172-142:880681:880681 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-172-142:880679:880679 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-172-142:880682:880682 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-172-142:880685:880685 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-172-142:880681:880681 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-172-142:880679:880679 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-172-142:880682:880682 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-172-142:880685:880685 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-172-142:880683:880683 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-172-142:880683:880683 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-172-147:618585:618585 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-172-147:618584:618584 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-172-147:618587:618587 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-172-147:618588:618588 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-172-147:618591:618591 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-172-147:618584:618584 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-172-147:618585:618585 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-172-147:618588:618588 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-172-147:618587:618587 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-172-147:618591:618591 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-172-142:880686:880686 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-172-142:880684:880684 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-172-142:880686:880686 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-172-142:880684:880684 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-172-142:880680:880680 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-172-142:880680:880680 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-172-252:688202:688202 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-172-252:688207:688207 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-172-252:688202:688202 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-172-252:688207:688207 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-172-252:688206:688206 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-172-252:688200:688200 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-172-252:688206:688206 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-172-252:688200:688200 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-172-73:926597:926597 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-172-73:926594:926594 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-172-73:926598:926598 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-172-73:926596:926596 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-172-73:926594:926594 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-172-73:926597:926597 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-172-73:926596:926596 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-172-73:926598:926598 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-172-116:1019293:1019293 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-172-116:1019288:1019288 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-172-116:1019290:1019290 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-172-116:1019292:1019292 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-172-252:688204:688204 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-172-252:688204:688204 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-172-116:1019293:1019293 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-172-116:1019290:1019290 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-172-116:1019292:1019292 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-172-116:1019288:1019288 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-172-73:926601:926601 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-172-73:926601:926601 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-173-7:811255:811255 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-173-7:811255:811255 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-172-147:618589:618589 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-172-147:618590:618590 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-172-147:618589:618589 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-172-147:618586:618586 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-172-147:618590:618590 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-172-73:926600:926600 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-172-73:926600:926600 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-172-116:1019295:1019295 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-172-116:1019295:1019295 [7] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-172-147:618586:618586 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-172-252:688205:688205 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-172-252:688205:688205 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-172-116:1019289:1019289 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-172-116:1019289:1019289 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-173-7:811248:811248 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-173-7:811248:811248 [0] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-172-116:1019291:1019291 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-172-116:1019294:1019294 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-172-116:1019291:1019291 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-172-116:1019294:1019294 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-173-7:811252:811252 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-173-7:811253:811253 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-173-7:811252:811252 [4] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-173-7:811253:811253 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-173-7:811250:811250 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-173-7:811250:811250 [2] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-172-252:688203:688203 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-172-252:688203:688203 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-172-73:926599:926599 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-172-73:926595:926595 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-172-73:926599:926599 [5] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-172-73:926595:926595 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-172-252:688201:688201 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-172-252:688201:688201 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-173-7:811251:811251 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-173-7:811254:811254 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-173-7:811251:811251 [3] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-173-7:811254:811254 [6] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-173-7:811249:811249 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin_v6 symbol.
+ip-26-0-173-7:811249:811249 [1] NCCL INFO NET/Plugin: Failed to find ncclCollNetPlugin symbol (v4 or v5).
+ip-26-0-160-225:94342:94504 [0] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-160-225:94342:94504 [0] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-160-225:94342:94504 [0] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-160-225:94342:94504 [0] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-160-225:94342:94504 [0] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-160-225:94342:94504 [0] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-160-225:94342:94504 [0] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-160-225:94342:94504 [0] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-160-225:94342:94504 [0] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-160-225:94347:94507 [5] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-160-225:94347:94507 [5] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-160-225:94342:94504 [0] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-160-225:94347:94507 [5] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-160-225:94347:94507 [5] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-160-225:94347:94507 [5] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-160-225:94347:94507 [5] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-160-225:94347:94507 [5] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-160-225:94347:94507 [5] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-160-225:94347:94507 [5] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-172-142:880684:880842 [5] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-172-142:880684:880842 [5] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-160-225:94344:94510 [2] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-160-225:94344:94510 [2] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-160-225:94349:94505 [7] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-160-225:94349:94505 [7] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-172-142:880684:880842 [5] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-172-142:880684:880842 [5] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-172-142:880684:880842 [5] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-172-142:880684:880842 [5] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-172-142:880684:880842 [5] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-172-142:880684:880842 [5] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-172-142:880684:880842 [5] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-160-225:94343:94508 [1] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-160-225:94343:94508 [1] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-172-142:880683:880844 [4] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-172-142:880683:880844 [4] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-160-225:94344:94510 [2] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-160-225:94344:94510 [2] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-160-225:94344:94510 [2] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-160-225:94344:94510 [2] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-160-225:94344:94510 [2] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-160-225:94344:94510 [2] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-160-225:94344:94510 [2] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-160-225:94348:94506 [6] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-160-225:94348:94506 [6] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-172-57:942611:942772 [0] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-172-57:942611:942772 [0] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-172-252:688203:688362 [3] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-172-252:688203:688362 [3] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-172-142:880683:880844 [4] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-172-142:880683:880844 [4] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-172-142:880683:880844 [4] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-172-142:880683:880844 [4] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-172-142:880683:880844 [4] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-172-142:880683:880844 [4] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-172-142:880683:880844 [4] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-173-7:811251:811409 [3] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-173-7:811251:811409 [3] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-172-57:942611:942772 [0] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-172-57:942611:942772 [0] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-172-57:942611:942772 [0] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-172-57:942611:942772 [0] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-172-57:942611:942772 [0] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-172-57:942611:942772 [0] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-172-57:942611:942772 [0] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-172-252:688203:688362 [3] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-172-252:688203:688362 [3] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-172-252:688203:688362 [3] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-172-252:688203:688362 [3] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-172-252:688203:688362 [3] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-172-252:688203:688362 [3] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-172-252:688203:688362 [3] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-160-225:94349:94505 [7] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-160-225:94349:94505 [7] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-160-225:94349:94505 [7] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-160-225:94349:94505 [7] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-160-225:94349:94505 [7] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-160-225:94349:94505 [7] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-160-225:94349:94505 [7] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-172-147:618589:618744 [5] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-172-147:618589:618744 [5] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-160-225:94346:94511 [4] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-160-225:94346:94511 [4] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-173-7:811251:811409 [3] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-173-7:811251:811409 [3] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-173-7:811251:811409 [3] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-173-7:811251:811409 [3] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-173-7:811251:811409 [3] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-173-7:811251:811409 [3] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-173-7:811251:811409 [3] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-172-73:926600:926757 [6] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-172-73:926600:926757 [6] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-172-147:618589:618744 [5] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-172-147:618589:618744 [5] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-172-147:618589:618744 [5] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-172-252:688204:688367 [4] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-172-252:688204:688367 [4] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-172-147:618589:618744 [5] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-172-147:618589:618744 [5] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-172-147:618589:618744 [5] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-172-147:618589:618744 [5] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-172-73:926600:926757 [6] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-172-73:926600:926757 [6] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-172-73:926600:926757 [6] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-172-73:926600:926757 [6] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-172-73:926600:926757 [6] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-172-73:926597:926758 [3] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-172-73:926597:926758 [3] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-172-73:926600:926757 [6] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-172-73:926600:926757 [6] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-172-252:688204:688367 [4] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-172-252:688204:688367 [4] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-172-252:688204:688367 [4] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-172-252:688204:688367 [4] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-172-252:688204:688367 [4] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-172-252:688204:688367 [4] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-172-252:688204:688367 [4] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-172-73:926596:926756 [2] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-172-73:926596:926756 [2] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-172-252:688206:688363 [6] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-172-252:688206:688363 [6] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-160-225:94343:94508 [1] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-172-57:942614:942773 [3] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-172-57:942614:942773 [3] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-160-225:94343:94508 [1] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-160-225:94343:94508 [1] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-160-225:94343:94508 [1] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-160-225:94343:94508 [1] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-160-225:94343:94508 [1] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-160-225:94343:94508 [1] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-173-7:811248:811415 [0] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-173-7:811248:811415 [0] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-160-225:94348:94506 [6] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-160-225:94348:94506 [6] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-160-225:94348:94506 [6] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-160-225:94348:94506 [6] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-160-225:94348:94506 [6] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-160-225:94348:94506 [6] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-160-225:94348:94506 [6] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-172-147:618591:618746 [7] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-172-147:618591:618746 [7] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-172-147:618587:618751 [3] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-172-147:618587:618751 [3] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-172-57:942614:942773 [3] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-172-57:942614:942773 [3] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-172-57:942614:942773 [3] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-172-57:942614:942773 [3] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-172-57:942614:942773 [3] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-172-57:942614:942773 [3] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-172-57:942614:942773 [3] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-172-142:880682:880840 [3] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-172-142:880682:880840 [3] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-160-225:94346:94511 [4] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-160-225:94346:94511 [4] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-160-225:94346:94511 [4] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-160-225:94346:94511 [4] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-160-225:94346:94511 [4] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-160-225:94346:94511 [4] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-160-225:94346:94511 [4] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-160-225:94345:94509 [3] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-160-225:94345:94509 [3] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-172-73:926598:926761 [4] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-172-73:926598:926761 [4] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-172-252:688206:688363 [6] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-172-252:688206:688363 [6] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-172-252:688206:688363 [6] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-172-252:688206:688363 [6] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-172-252:688206:688363 [6] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-172-252:688206:688363 [6] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-172-252:688206:688363 [6] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-172-73:926597:926758 [3] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-172-73:926597:926758 [3] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-172-73:926597:926758 [3] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-172-73:926597:926758 [3] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-172-73:926597:926758 [3] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-172-73:926597:926758 [3] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-172-73:926597:926758 [3] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-172-252:688201:688364 [1] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-172-252:688201:688364 [1] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-172-142:880682:880840 [3] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-172-142:880682:880840 [3] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-172-142:880682:880840 [3] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-172-142:880682:880840 [3] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-172-142:880682:880840 [3] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-172-142:880682:880840 [3] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-172-142:880682:880840 [3] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-173-7:811252:811408 [4] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-173-7:811252:811408 [4] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-172-147:618584:618747 [0] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-172-147:618584:618747 [0] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-173-7:811255:811411 [7] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-173-7:811255:811411 [7] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-173-7:811254:811414 [6] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-173-7:811254:811414 [6] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-172-147:618590:618749 [6] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-172-147:618590:618749 [6] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-172-147:618591:618746 [7] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-172-147:618591:618746 [7] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-173-7:811250:811412 [2] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-173-7:811250:811412 [2] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-172-147:618591:618746 [7] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-172-147:618591:618746 [7] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-172-147:618591:618746 [7] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-172-147:618591:618746 [7] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-172-147:618591:618746 [7] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-172-73:926599:926755 [5] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-172-73:926599:926755 [5] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-172-147:618588:618750 [4] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-172-147:618588:618750 [4] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-172-252:688207:688366 [7] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-172-252:688207:688366 [7] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-172-73:926596:926756 [2] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-172-73:926596:926756 [2] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-172-73:926596:926756 [2] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-172-73:926596:926756 [2] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-172-73:926596:926756 [2] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-172-73:926596:926756 [2] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-172-73:926596:926756 [2] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-173-7:811249:811413 [1] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-173-7:811249:811413 [1] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-172-147:618587:618751 [3] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-172-147:618587:618751 [3] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-173-7:811248:811415 [0] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-172-147:618587:618751 [3] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-172-147:618587:618751 [3] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-173-7:811248:811415 [0] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-172-147:618587:618751 [3] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-172-147:618587:618751 [3] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-172-147:618587:618751 [3] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-173-7:811248:811415 [0] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-173-7:811248:811415 [0] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-173-7:811248:811415 [0] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-173-7:811248:811415 [0] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-173-7:811248:811415 [0] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-172-57:942612:942779 [1] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-172-57:942612:942779 [1] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-172-57:942615:942775 [4] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-172-57:942615:942775 [4] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-172-147:618586:618745 [2] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-172-147:618586:618745 [2] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-172-142:880680:880843 [1] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-172-142:880680:880843 [1] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-172-142:880679:880846 [0] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-172-142:880679:880846 [0] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-172-73:926601:926759 [7] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-172-73:926601:926759 [7] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-172-57:942617:942774 [6] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-172-57:942617:942774 [6] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-172-147:618585:618748 [1] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-172-147:618585:618748 [1] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-172-252:688201:688364 [1] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-172-252:688201:688364 [1] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-172-252:688201:688364 [1] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-172-252:688201:688364 [1] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-172-252:688201:688364 [1] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-172-252:688201:688364 [1] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-172-252:688201:688364 [1] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-160-225:94345:94509 [3] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-160-225:94345:94509 [3] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-160-225:94345:94509 [3] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-160-225:94345:94509 [3] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-160-225:94345:94509 [3] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-160-225:94345:94509 [3] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-160-225:94345:94509 [3] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-172-252:688207:688366 [7] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-172-57:942613:942776 [2] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-172-57:942613:942776 [2] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-172-252:688207:688366 [7] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-172-252:688207:688366 [7] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-172-252:688207:688366 [7] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-172-252:688207:688366 [7] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-172-252:688207:688366 [7] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-172-252:688207:688366 [7] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-172-57:942618:942778 [7] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-172-57:942618:942778 [7] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-172-252:688205:688365 [5] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-172-252:688205:688365 [5] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-172-142:880680:880843 [1] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-172-142:880680:880843 [1] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-172-142:880680:880843 [1] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-172-142:880680:880843 [1] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-172-142:880680:880843 [1] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-172-142:880680:880843 [1] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-172-142:880680:880843 [1] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-172-142:880679:880846 [0] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-172-142:880679:880846 [0] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-172-73:926594:926762 [0] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-172-73:926598:926761 [4] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-172-73:926594:926762 [0] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-172-73:926598:926761 [4] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-172-142:880686:880841 [7] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-172-142:880686:880841 [7] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-172-142:880679:880846 [0] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-172-142:880679:880846 [0] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-172-142:880679:880846 [0] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-172-73:926598:926761 [4] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-172-73:926598:926761 [4] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-172-142:880679:880846 [0] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-172-142:880679:880846 [0] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-172-73:926598:926761 [4] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-172-73:926598:926761 [4] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-172-73:926598:926761 [4] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-172-142:880681:880845 [2] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-172-142:880681:880845 [2] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-172-57:942612:942779 [1] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-172-57:942612:942779 [1] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-172-57:942612:942779 [1] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-172-57:942612:942779 [1] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-172-57:942612:942779 [1] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-172-57:942612:942779 [1] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-172-57:942612:942779 [1] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-172-73:926595:926760 [1] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-172-73:926595:926760 [1] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-172-57:942615:942775 [4] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-172-57:942615:942775 [4] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-172-57:942615:942775 [4] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-172-57:942615:942775 [4] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-172-57:942615:942775 [4] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-172-57:942615:942775 [4] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-172-57:942615:942775 [4] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-172-252:688200:688369 [0] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-172-252:688200:688369 [0] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-173-7:811252:811408 [4] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-173-7:811252:811408 [4] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-172-57:942616:942777 [5] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-172-57:942616:942777 [5] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-173-7:811252:811408 [4] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-173-7:811252:811408 [4] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-173-7:811252:811408 [4] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-173-7:811252:811408 [4] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-173-7:811252:811408 [4] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-172-142:880685:880847 [6] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-172-142:880685:880847 [6] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-172-73:926599:926755 [5] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-172-73:926599:926755 [5] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-173-7:811254:811414 [6] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-172-73:926599:926755 [5] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-172-73:926599:926755 [5] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-172-73:926599:926755 [5] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-173-7:811254:811414 [6] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-173-7:811254:811414 [6] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-173-7:811254:811414 [6] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-173-7:811254:811414 [6] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-172-73:926599:926755 [5] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-172-73:926599:926755 [5] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-173-7:811254:811414 [6] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-173-7:811254:811414 [6] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-173-7:811255:811411 [7] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-173-7:811255:811411 [7] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-173-7:811255:811411 [7] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-173-7:811255:811411 [7] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-173-7:811255:811411 [7] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-173-7:811255:811411 [7] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-173-7:811255:811411 [7] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-173-7:811250:811412 [2] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-173-7:811250:811412 [2] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-173-7:811250:811412 [2] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-173-7:811250:811412 [2] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-173-7:811250:811412 [2] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-173-7:811250:811412 [2] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-173-7:811250:811412 [2] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-172-147:618584:618747 [0] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-172-147:618584:618747 [0] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-172-252:688205:688365 [5] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-172-147:618584:618747 [0] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-172-147:618584:618747 [0] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-172-147:618584:618747 [0] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-172-252:688205:688365 [5] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-172-147:618584:618747 [0] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-172-147:618584:618747 [0] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-172-252:688205:688365 [5] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-172-252:688205:688365 [5] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-172-252:688205:688365 [5] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-172-252:688205:688365 [5] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-172-252:688205:688365 [5] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-173-7:811249:811413 [1] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-173-7:811249:811413 [1] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-173-7:811249:811413 [1] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-173-7:811249:811413 [1] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-173-7:811249:811413 [1] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-173-7:811249:811413 [1] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-173-7:811249:811413 [1] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-172-252:688202:688368 [2] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-172-147:618590:618749 [6] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-172-252:688202:688368 [2] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-172-147:618590:618749 [6] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-172-147:618590:618749 [6] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-172-147:618590:618749 [6] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-172-147:618590:618749 [6] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-172-147:618590:618749 [6] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-172-147:618590:618749 [6] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-172-252:688200:688369 [0] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-172-252:688200:688369 [0] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-172-252:688200:688369 [0] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-172-252:688200:688369 [0] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-172-252:688200:688369 [0] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-172-252:688200:688369 [0] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-172-252:688200:688369 [0] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-172-57:942617:942774 [6] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-172-57:942617:942774 [6] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-172-57:942617:942774 [6] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-172-57:942617:942774 [6] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-172-57:942617:942774 [6] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-172-147:618588:618750 [4] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-172-147:618588:618750 [4] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-172-57:942617:942774 [6] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-172-57:942617:942774 [6] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-172-147:618588:618750 [4] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-172-147:618588:618750 [4] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-172-73:926601:926759 [7] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-172-73:926601:926759 [7] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-172-147:618588:618750 [4] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-172-147:618588:618750 [4] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-172-147:618588:618750 [4] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-172-73:926601:926759 [7] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-172-73:926601:926759 [7] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-172-73:926601:926759 [7] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-172-73:926601:926759 [7] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-172-73:926601:926759 [7] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-172-142:880686:880841 [7] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-172-142:880686:880841 [7] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-172-142:880686:880841 [7] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-172-142:880686:880841 [7] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-172-142:880686:880841 [7] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-172-142:880686:880841 [7] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-172-142:880686:880841 [7] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-172-57:942613:942776 [2] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-172-57:942613:942776 [2] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-172-57:942613:942776 [2] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-172-57:942613:942776 [2] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-172-57:942613:942776 [2] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-172-57:942613:942776 [2] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-172-57:942613:942776 [2] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-172-57:942618:942778 [7] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-172-57:942618:942778 [7] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-172-57:942618:942778 [7] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-172-57:942618:942778 [7] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-172-57:942618:942778 [7] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-172-57:942618:942778 [7] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-172-57:942618:942778 [7] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-172-142:880681:880845 [2] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-172-142:880681:880845 [2] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-172-142:880681:880845 [2] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-172-142:880681:880845 [2] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-172-142:880681:880845 [2] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-172-142:880681:880845 [2] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-172-142:880681:880845 [2] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-172-147:618586:618745 [2] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-172-147:618586:618745 [2] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-172-147:618586:618745 [2] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-172-147:618586:618745 [2] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-172-147:618586:618745 [2] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-172-147:618586:618745 [2] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-172-147:618586:618745 [2] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-172-73:926594:926762 [0] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-172-73:926594:926762 [0] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-172-73:926594:926762 [0] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-172-73:926594:926762 [0] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-172-73:926594:926762 [0] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-172-73:926594:926762 [0] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-172-73:926594:926762 [0] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-172-142:880685:880847 [6] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-172-142:880685:880847 [6] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-172-142:880685:880847 [6] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-172-142:880685:880847 [6] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-172-142:880685:880847 [6] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-172-142:880685:880847 [6] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-172-142:880685:880847 [6] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-172-73:926595:926760 [1] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-172-73:926595:926760 [1] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-172-73:926595:926760 [1] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-172-73:926595:926760 [1] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-172-73:926595:926760 [1] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-172-73:926595:926760 [1] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-172-73:926595:926760 [1] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-172-252:688202:688368 [2] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-172-252:688202:688368 [2] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-172-252:688202:688368 [2] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-172-252:688202:688368 [2] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-172-252:688202:688368 [2] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-172-252:688202:688368 [2] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-172-252:688202:688368 [2] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-173-7:811253:811410 [5] NCCL INFO NET/OFI Initializing aws-ofi-nccl 1.13.2-aws
+ip-26-0-173-7:811253:811410 [5] NCCL INFO NET/OFI Using Libfabric version 1.22
+ip-26-0-172-57:942616:942777 [5] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-172-57:942616:942777 [5] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-172-57:942616:942777 [5] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-172-57:942616:942777 [5] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-172-57:942616:942777 [5] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-172-57:942616:942777 [5] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-172-57:942616:942777 [5] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-173-7:811253:811410 [5] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-173-7:811253:811410 [5] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-173-7:811253:811410 [5] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-173-7:811253:811410 [5] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-173-7:811253:811410 [5] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-173-7:811253:811410 [5] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-173-7:811253:811410 [5] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-172-147:618585:618748 [1] NCCL INFO NET/OFI Using CUDA driver version 12020 with runtime 12020
+ip-26-0-172-147:618585:618748 [1] NCCL INFO NET/OFI Configuring AWS-specific options
+ip-26-0-172-147:618585:618748 [1] NCCL INFO NET/OFI Setting NCCL_NET_FORCE_FLUSH=0 for Hopper GPUs
+ip-26-0-172-147:618585:618748 [1] NCCL INFO NET/OFI Setting NCCL_NVLSTREE_MAX_CHUNKSIZE to 512KiB
+ip-26-0-172-147:618585:618748 [1] NCCL INFO NET/OFI Setting NCCL_NVLS_CHUNKSIZE to 512KiB
+ip-26-0-172-147:618585:618748 [1] NCCL INFO NET/OFI Internode latency set at 75.0 us
+ip-26-0-172-147:618585:618748 [1] NCCL INFO NET/OFI Using transport protocol RDMA (platform set)
+ip-26-0-160-225:94347:94507 [5] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-172-142:880684:880842 [5] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-172-142:880683:880844 [4] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-172-57:942611:942772 [0] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-172-252:688203:688362 [3] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-172-142:880682:880840 [3] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-173-7:811251:811409 [3] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-172-252:688204:688367 [4] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-172-147:618589:618744 [5] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-172-73:926596:926756 [2] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-172-73:926600:926757 [6] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-172-57:942612:942779 [1] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-172-57:942615:942775 [4] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-172-73:926597:926758 [3] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-172-57:942614:942773 [3] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-172-252:688206:688363 [6] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-172-252:688201:688364 [1] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-172-252:688207:688366 [7] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-173-7:811252:811408 [4] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-173-7:811254:811414 [6] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-173-7:811248:811415 [0] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-172-147:618591:618746 [7] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-173-7:811249:811413 [1] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-172-142:880680:880843 [1] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-172-147:618587:618751 [3] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-172-142:880679:880846 [0] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-173-7:811250:811412 [2] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-172-73:926598:926761 [4] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-173-7:811253:811410 [5] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-172-73:926595:926760 [1] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-172-73:926601:926759 [7] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-172-73:926594:926762 [0] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-172-73:926599:926755 [5] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-160-225:94342:94504 [0] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-160-225:94342:94504 [0] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-160-225:94342:94504 [0] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-160-225:94342:94504 [0] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-160-225:94342:94504 [0] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-160-225:94342:94504 [0] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-160-225:94342:94504 [0] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-160-225:94342:94504 [0] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-160-225:94342:94504 [0] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-160-225:94342:94504 [0] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-160-225:94342:94504 [0] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-160-225:94342:94504 [0] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-160-225:94342:94504 [0] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-160-225:94342:94504 [0] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-160-225:94342:94504 [0] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-160-225:94342:94504 [0] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-160-225:94342:94504 [0] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-160-225:94342:94504 [0] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-160-225:94342:94504 [0] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-160-225:94342:94504 [0] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-160-225:94342:94504 [0] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-160-225:94342:94504 [0] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-160-225:94342:94504 [0] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-160-225:94342:94504 [0] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-160-225:94342:94504 [0] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-160-225:94342:94504 [0] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-160-225:94342:94504 [0] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-160-225:94342:94504 [0] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-160-225:94342:94504 [0] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-160-225:94342:94504 [0] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-160-225:94342:94504 [0] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-160-225:94342:94504 [0] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-172-57:942618:942778 [7] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-172-57:942617:942774 [6] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-160-225:94342:94504 [0] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/349
+ip-26-0-160-225:94342:94504 [0] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-172-57:942616:942777 [5] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-173-7:811255:811411 [7] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-172-57:942613:942776 [2] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-172-252:688200:688369 [0] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-172-252:688205:688365 [5] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-172-252:688202:688368 [2] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-160-225:94349:94505 [7] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-160-225:94344:94510 [2] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-172-142:880685:880847 [6] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-172-142:880686:880841 [7] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-172-142:880681:880845 [2] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-172-147:618584:618747 [0] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-172-147:618588:618750 [4] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-172-147:618585:618748 [1] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-172-147:618590:618749 [6] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-172-147:618586:618745 [2] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-160-225:94346:94511 [4] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-160-225:94345:94509 [3] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-160-225:94348:94506 [6] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-160-225:94343:94508 [1] NCCL INFO NET/OFI Selected Provider is efa (found 32 nics)
+ip-26-0-172-57:942611:942772 [0] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-172-57:942611:942772 [0] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-172-57:942611:942772 [0] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-172-57:942611:942772 [0] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-172-57:942611:942772 [0] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-172-57:942611:942772 [0] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-172-57:942611:942772 [0] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-172-57:942611:942772 [0] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-172-57:942611:942772 [0] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-172-57:942611:942772 [0] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-172-57:942611:942772 [0] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-172-57:942611:942772 [0] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-172-57:942611:942772 [0] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-172-57:942611:942772 [0] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-172-57:942611:942772 [0] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-172-57:942611:942772 [0] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-172-57:942611:942772 [0] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-172-57:942611:942772 [0] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-172-57:942611:942772 [0] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-172-57:942611:942772 [0] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-172-57:942611:942772 [0] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-172-57:942611:942772 [0] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-172-57:942611:942772 [0] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-172-57:942611:942772 [0] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-172-57:942611:942772 [0] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-172-57:942611:942772 [0] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-172-57:942611:942772 [0] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-172-57:942611:942772 [0] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-172-57:942611:942772 [0] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-172-57:942611:942772 [0] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-172-57:942611:942772 [0] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-172-57:942611:942772 [0] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-160-225:94347:94507 [5] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-160-225:94347:94507 [5] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-160-225:94347:94507 [5] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-160-225:94347:94507 [5] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-160-225:94347:94507 [5] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-160-225:94347:94507 [5] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-160-225:94347:94507 [5] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-160-225:94347:94507 [5] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-160-225:94347:94507 [5] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-160-225:94347:94507 [5] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-160-225:94347:94507 [5] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-160-225:94347:94507 [5] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-160-225:94347:94507 [5] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-160-225:94347:94507 [5] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-160-225:94347:94507 [5] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-160-225:94347:94507 [5] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-160-225:94347:94507 [5] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-160-225:94347:94507 [5] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-160-225:94347:94507 [5] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-160-225:94347:94507 [5] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-160-225:94347:94507 [5] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-160-225:94347:94507 [5] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-160-225:94347:94507 [5] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-160-225:94347:94507 [5] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-160-225:94347:94507 [5] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-160-225:94347:94507 [5] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-160-225:94347:94507 [5] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-160-225:94347:94507 [5] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-160-225:94347:94507 [5] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-160-225:94347:94507 [5] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-160-225:94347:94507 [5] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-160-225:94347:94507 [5] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-160-225:94347:94507 [5] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/280
+ip-26-0-160-225:94347:94507 [5] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/280
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-160-225:94342:94504 [0] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-160-225:94342:94504 [0] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-160-225:94342:94504 [0] NCCL INFO Using network Libfabric
+ip-26-0-172-57:942611:942772 [0] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/280
+ip-26-0-172-57:942611:942772 [0] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-172-142:880684:880842 [5] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-172-142:880684:880842 [5] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-172-142:880684:880842 [5] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-172-142:880684:880842 [5] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-172-142:880684:880842 [5] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-172-142:880684:880842 [5] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-172-142:880684:880842 [5] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-172-142:880684:880842 [5] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-172-142:880684:880842 [5] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-172-142:880684:880842 [5] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-172-142:880684:880842 [5] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-172-142:880684:880842 [5] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-172-142:880684:880842 [5] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-172-142:880684:880842 [5] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-172-142:880684:880842 [5] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-172-142:880684:880842 [5] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-172-142:880684:880842 [5] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-172-142:880684:880842 [5] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-172-142:880684:880842 [5] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-172-142:880684:880842 [5] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-172-142:880684:880842 [5] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-172-142:880684:880842 [5] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-172-142:880684:880842 [5] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-172-142:880684:880842 [5] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-172-142:880684:880842 [5] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-172-142:880684:880842 [5] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-172-142:880684:880842 [5] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-172-142:880684:880842 [5] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-172-142:880684:880842 [5] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-172-142:880684:880842 [5] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-172-142:880684:880842 [5] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-172-142:880684:880842 [5] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-172-142:880684:880842 [5] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/280
+ip-26-0-172-142:880684:880842 [5] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-172-142:880683:880844 [4] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-172-142:880683:880844 [4] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-172-142:880683:880844 [4] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-172-142:880683:880844 [4] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-172-142:880683:880844 [4] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-172-142:880683:880844 [4] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-172-142:880683:880844 [4] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-172-142:880683:880844 [4] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-172-142:880683:880844 [4] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-172-142:880683:880844 [4] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-172-142:880683:880844 [4] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-172-142:880683:880844 [4] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-172-142:880683:880844 [4] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-172-142:880683:880844 [4] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-172-142:880683:880844 [4] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-172-142:880683:880844 [4] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-172-142:880683:880844 [4] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-172-142:880683:880844 [4] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-172-142:880683:880844 [4] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-172-142:880683:880844 [4] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-172-142:880683:880844 [4] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-172-142:880683:880844 [4] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-172-142:880683:880844 [4] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-172-142:880683:880844 [4] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-172-142:880683:880844 [4] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-172-142:880683:880844 [4] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-172-142:880683:880844 [4] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-172-142:880683:880844 [4] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-172-142:880683:880844 [4] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-172-142:880683:880844 [4] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-172-142:880683:880844 [4] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-172-142:880683:880844 [4] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-172-57:942611:942772 [0] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-172-57:942611:942772 [0] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-172-57:942611:942772 [0] NCCL INFO Using network Libfabric
+ip-26-0-172-142:880683:880844 [4] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/280
+ip-26-0-172-142:880683:880844 [4] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO Using network Libfabric
+ip-26-0-172-73:926597:926758 [3] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-172-73:926597:926758 [3] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-172-73:926597:926758 [3] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-172-73:926597:926758 [3] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-172-73:926597:926758 [3] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-172-73:926597:926758 [3] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-172-73:926597:926758 [3] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-172-73:926597:926758 [3] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-172-73:926597:926758 [3] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-172-73:926597:926758 [3] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-172-73:926597:926758 [3] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-172-73:926597:926758 [3] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-172-73:926597:926758 [3] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-172-73:926597:926758 [3] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-172-73:926597:926758 [3] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-172-73:926597:926758 [3] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-172-73:926597:926758 [3] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-172-73:926597:926758 [3] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-172-73:926597:926758 [3] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-172-73:926597:926758 [3] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-172-73:926597:926758 [3] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-172-73:926597:926758 [3] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-172-73:926597:926758 [3] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-172-73:926597:926758 [3] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-172-73:926597:926758 [3] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-172-73:926597:926758 [3] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-172-73:926597:926758 [3] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-172-73:926597:926758 [3] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-172-73:926597:926758 [3] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-172-73:926597:926758 [3] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-172-73:926597:926758 [3] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-172-73:926597:926758 [3] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-172-73:926597:926758 [3] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/280
+ip-26-0-172-73:926597:926758 [3] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-172-142:880684:880842 [5] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-172-142:880684:880842 [5] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-172-142:880684:880842 [5] NCCL INFO Using network Libfabric
+ip-26-0-172-57:942615:942775 [4] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-172-57:942615:942775 [4] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-172-57:942615:942775 [4] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-172-57:942615:942775 [4] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-172-57:942615:942775 [4] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-172-57:942615:942775 [4] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-172-57:942615:942775 [4] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-172-57:942615:942775 [4] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-172-57:942615:942775 [4] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-172-57:942615:942775 [4] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-172-57:942615:942775 [4] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-172-57:942615:942775 [4] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-172-57:942615:942775 [4] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-172-57:942615:942775 [4] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-172-57:942615:942775 [4] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-172-57:942615:942775 [4] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-172-57:942615:942775 [4] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-172-57:942615:942775 [4] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-172-57:942615:942775 [4] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-172-57:942615:942775 [4] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-172-57:942615:942775 [4] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-172-57:942615:942775 [4] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-172-57:942615:942775 [4] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-172-57:942615:942775 [4] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-172-57:942615:942775 [4] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-172-57:942615:942775 [4] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-172-57:942615:942775 [4] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-172-57:942615:942775 [4] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-172-57:942615:942775 [4] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-172-57:942615:942775 [4] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-172-57:942615:942775 [4] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-172-57:942615:942775 [4] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-172-57:942615:942775 [4] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/280
+ip-26-0-172-57:942615:942775 [4] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-160-225:94345:94509 [3] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-160-225:94345:94509 [3] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-160-225:94345:94509 [3] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-160-225:94345:94509 [3] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-160-225:94345:94509 [3] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-160-225:94345:94509 [3] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-160-225:94345:94509 [3] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-160-225:94345:94509 [3] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-160-225:94345:94509 [3] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-160-225:94345:94509 [3] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-160-225:94345:94509 [3] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-160-225:94345:94509 [3] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-160-225:94345:94509 [3] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-160-225:94345:94509 [3] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-160-225:94345:94509 [3] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-160-225:94345:94509 [3] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-160-225:94345:94509 [3] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-160-225:94345:94509 [3] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-160-225:94345:94509 [3] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-160-225:94345:94509 [3] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-160-225:94345:94509 [3] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-160-225:94345:94509 [3] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-160-225:94345:94509 [3] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-160-225:94345:94509 [3] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-160-225:94345:94509 [3] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-160-225:94345:94509 [3] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-160-225:94345:94509 [3] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-160-225:94345:94509 [3] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-160-225:94345:94509 [3] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-160-225:94345:94509 [3] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-160-225:94345:94509 [3] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-160-225:94345:94509 [3] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/280
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-172-57:942612:942779 [1] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-172-57:942612:942779 [1] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-172-57:942612:942779 [1] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-172-57:942612:942779 [1] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-172-57:942612:942779 [1] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-172-57:942612:942779 [1] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-172-57:942612:942779 [1] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-172-57:942612:942779 [1] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-172-57:942612:942779 [1] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-172-57:942612:942779 [1] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-172-57:942612:942779 [1] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-172-57:942612:942779 [1] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-172-57:942612:942779 [1] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-172-57:942612:942779 [1] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-172-57:942612:942779 [1] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-172-57:942612:942779 [1] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-172-57:942612:942779 [1] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-172-57:942612:942779 [1] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-172-57:942612:942779 [1] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-172-57:942612:942779 [1] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-172-57:942612:942779 [1] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-172-57:942612:942779 [1] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-172-57:942612:942779 [1] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-172-57:942612:942779 [1] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-172-57:942612:942779 [1] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-172-57:942612:942779 [1] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-172-57:942612:942779 [1] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-172-57:942612:942779 [1] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-172-57:942612:942779 [1] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-172-57:942612:942779 [1] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-172-57:942612:942779 [1] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-172-57:942612:942779 [1] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-172-57:942612:942779 [1] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/280
+ip-26-0-172-57:942612:942779 [1] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-173-7:811251:811409 [3] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-173-7:811251:811409 [3] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-173-7:811251:811409 [3] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-173-7:811251:811409 [3] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-173-7:811251:811409 [3] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-173-7:811251:811409 [3] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-173-7:811251:811409 [3] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-173-7:811251:811409 [3] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-173-7:811251:811409 [3] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-173-7:811251:811409 [3] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-173-7:811251:811409 [3] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-173-7:811251:811409 [3] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-173-7:811251:811409 [3] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-173-7:811251:811409 [3] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-173-7:811251:811409 [3] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-173-7:811251:811409 [3] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-173-7:811251:811409 [3] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-173-7:811251:811409 [3] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-173-7:811251:811409 [3] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-173-7:811251:811409 [3] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-173-7:811251:811409 [3] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-173-7:811251:811409 [3] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-173-7:811251:811409 [3] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-173-7:811251:811409 [3] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-173-7:811251:811409 [3] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-173-7:811251:811409 [3] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-173-7:811251:811409 [3] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-173-7:811251:811409 [3] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-173-7:811251:811409 [3] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-173-7:811251:811409 [3] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-173-7:811251:811409 [3] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-173-7:811251:811409 [3] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-173-7:811251:811409 [3] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/280
+ip-26-0-173-7:811251:811409 [3] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-172-147:618587:618751 [3] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-172-147:618587:618751 [3] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-172-147:618587:618751 [3] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-172-147:618587:618751 [3] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-172-147:618587:618751 [3] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-172-147:618587:618751 [3] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-172-147:618587:618751 [3] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-172-147:618587:618751 [3] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-172-147:618587:618751 [3] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-172-147:618587:618751 [3] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-172-147:618587:618751 [3] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-172-147:618587:618751 [3] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-172-147:618587:618751 [3] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-172-147:618587:618751 [3] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-172-147:618587:618751 [3] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-172-147:618587:618751 [3] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-172-147:618587:618751 [3] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-172-147:618587:618751 [3] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-172-147:618587:618751 [3] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-172-147:618587:618751 [3] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-172-147:618587:618751 [3] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-172-147:618587:618751 [3] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-172-147:618587:618751 [3] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-172-147:618587:618751 [3] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-172-147:618587:618751 [3] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-172-147:618587:618751 [3] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-172-147:618587:618751 [3] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-172-147:618587:618751 [3] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-172-147:618587:618751 [3] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-172-147:618587:618751 [3] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-172-147:618587:618751 [3] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-172-147:618587:618751 [3] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-172-147:618587:618751 [3] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/280
+ip-26-0-172-147:618587:618751 [3] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-160-225:94345:94509 [3] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/280
+ip-26-0-160-225:94345:94509 [3] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-160-225:94349:94505 [7] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-160-225:94349:94505 [7] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-160-225:94349:94505 [7] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-160-225:94349:94505 [7] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-160-225:94349:94505 [7] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-160-225:94349:94505 [7] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-160-225:94349:94505 [7] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-160-225:94349:94505 [7] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-160-225:94349:94505 [7] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-160-225:94349:94505 [7] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-160-225:94349:94505 [7] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-160-225:94349:94505 [7] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-160-225:94349:94505 [7] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-160-225:94349:94505 [7] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-160-225:94349:94505 [7] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-160-225:94349:94505 [7] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-160-225:94349:94505 [7] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-160-225:94349:94505 [7] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-160-225:94349:94505 [7] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-160-225:94349:94505 [7] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-160-225:94349:94505 [7] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-160-225:94349:94505 [7] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-160-225:94349:94505 [7] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-160-225:94349:94505 [7] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-160-225:94349:94505 [7] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-160-225:94349:94505 [7] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-160-225:94349:94505 [7] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-160-225:94349:94505 [7] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-160-225:94349:94505 [7] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-160-225:94349:94505 [7] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-160-225:94349:94505 [7] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-160-225:94349:94505 [7] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-160-225:94349:94505 [7] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/280
+ip-26-0-160-225:94349:94505 [7] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-160-225:94348:94506 [6] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-160-225:94348:94506 [6] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-160-225:94348:94506 [6] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-160-225:94348:94506 [6] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-160-225:94348:94506 [6] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-160-225:94348:94506 [6] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-160-225:94348:94506 [6] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-160-225:94348:94506 [6] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-160-225:94348:94506 [6] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-160-225:94348:94506 [6] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-160-225:94348:94506 [6] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-160-225:94348:94506 [6] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-160-225:94348:94506 [6] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-160-225:94348:94506 [6] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-160-225:94348:94506 [6] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-160-225:94348:94506 [6] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-160-225:94348:94506 [6] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-160-225:94348:94506 [6] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-160-225:94348:94506 [6] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-160-225:94348:94506 [6] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-160-225:94348:94506 [6] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-160-225:94348:94506 [6] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-160-225:94348:94506 [6] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-160-225:94348:94506 [6] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-160-225:94348:94506 [6] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-160-225:94348:94506 [6] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-160-225:94348:94506 [6] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-160-225:94348:94506 [6] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-160-225:94348:94506 [6] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-160-225:94348:94506 [6] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-160-225:94348:94506 [6] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-160-225:94348:94506 [6] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-160-225:94347:94507 [5] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-160-225:94347:94507 [5] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-160-225:94347:94507 [5] NCCL INFO Using network Libfabric
+ip-26-0-160-225:94348:94506 [6] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/280
+ip-26-0-160-225:94348:94506 [6] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-172-142:880682:880840 [3] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-172-142:880682:880840 [3] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-172-142:880682:880840 [3] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-172-142:880682:880840 [3] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-172-142:880682:880840 [3] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-172-142:880682:880840 [3] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-172-142:880682:880840 [3] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-172-142:880682:880840 [3] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-172-142:880682:880840 [3] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-172-142:880682:880840 [3] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-172-142:880682:880840 [3] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-172-142:880682:880840 [3] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-172-142:880682:880840 [3] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-172-142:880682:880840 [3] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-172-142:880682:880840 [3] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-172-142:880682:880840 [3] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-172-142:880682:880840 [3] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-172-142:880682:880840 [3] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-172-142:880682:880840 [3] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-172-142:880682:880840 [3] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-172-142:880682:880840 [3] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-172-142:880682:880840 [3] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-172-142:880682:880840 [3] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-172-142:880682:880840 [3] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-172-142:880682:880840 [3] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-172-142:880682:880840 [3] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-172-142:880682:880840 [3] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-172-142:880682:880840 [3] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-172-142:880682:880840 [3] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-172-142:880682:880840 [3] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-172-142:880682:880840 [3] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-172-142:880682:880840 [3] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-172-142:880682:880840 [3] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/280
+ip-26-0-172-142:880682:880840 [3] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-172-57:942614:942773 [3] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-172-57:942614:942773 [3] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-172-57:942614:942773 [3] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-172-57:942614:942773 [3] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-172-57:942614:942773 [3] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-172-57:942614:942773 [3] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-172-57:942614:942773 [3] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-172-57:942614:942773 [3] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-172-57:942614:942773 [3] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-172-57:942614:942773 [3] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-172-57:942614:942773 [3] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-172-57:942614:942773 [3] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-172-57:942614:942773 [3] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-172-57:942614:942773 [3] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-172-57:942614:942773 [3] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-172-57:942614:942773 [3] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-172-57:942614:942773 [3] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-172-57:942614:942773 [3] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-172-57:942614:942773 [3] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-172-57:942614:942773 [3] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-172-57:942614:942773 [3] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-172-57:942614:942773 [3] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-172-57:942614:942773 [3] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-172-57:942614:942773 [3] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-172-57:942614:942773 [3] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-172-57:942614:942773 [3] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-172-57:942614:942773 [3] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-172-57:942614:942773 [3] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-172-57:942614:942773 [3] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-172-57:942614:942773 [3] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-172-57:942614:942773 [3] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-172-57:942614:942773 [3] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-172-57:942614:942773 [3] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/280
+ip-26-0-172-142:880679:880846 [0] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-172-142:880679:880846 [0] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-172-142:880679:880846 [0] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-172-142:880679:880846 [0] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-172-142:880679:880846 [0] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-172-142:880679:880846 [0] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-172-142:880679:880846 [0] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-172-142:880679:880846 [0] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-172-142:880679:880846 [0] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-172-142:880679:880846 [0] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-172-142:880679:880846 [0] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-172-57:942614:942773 [3] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-172-142:880679:880846 [0] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-172-142:880679:880846 [0] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-172-142:880679:880846 [0] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-172-142:880679:880846 [0] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-172-142:880679:880846 [0] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-172-142:880679:880846 [0] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-172-142:880679:880846 [0] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-172-142:880679:880846 [0] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-172-142:880679:880846 [0] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-172-142:880679:880846 [0] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-172-142:880679:880846 [0] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-172-142:880679:880846 [0] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-172-142:880679:880846 [0] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-172-142:880679:880846 [0] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-172-142:880679:880846 [0] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-172-142:880679:880846 [0] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-172-142:880679:880846 [0] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-172-142:880679:880846 [0] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-172-142:880679:880846 [0] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-172-142:880679:880846 [0] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-172-142:880679:880846 [0] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-172-142:880679:880846 [0] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/280
+ip-26-0-172-142:880679:880846 [0] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-172-73:926597:926758 [3] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-172-73:926597:926758 [3] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-172-73:926597:926758 [3] NCCL INFO Using network Libfabric
+ip-26-0-172-142:880683:880844 [4] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-172-142:880683:880844 [4] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-172-142:880683:880844 [4] NCCL INFO Using network Libfabric
+ip-26-0-172-142:880681:880845 [2] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-172-142:880681:880845 [2] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-172-142:880681:880845 [2] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-172-142:880681:880845 [2] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-172-142:880681:880845 [2] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-172-142:880681:880845 [2] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-172-142:880681:880845 [2] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-172-142:880681:880845 [2] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-172-142:880681:880845 [2] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-172-142:880681:880845 [2] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-172-142:880681:880845 [2] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-172-142:880681:880845 [2] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-172-142:880681:880845 [2] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-172-142:880681:880845 [2] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-172-142:880681:880845 [2] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-172-142:880681:880845 [2] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-172-142:880681:880845 [2] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-172-142:880681:880845 [2] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-172-142:880681:880845 [2] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-172-142:880681:880845 [2] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-172-142:880681:880845 [2] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-172-142:880681:880845 [2] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-172-142:880681:880845 [2] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-172-142:880681:880845 [2] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-172-142:880681:880845 [2] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-172-142:880681:880845 [2] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-172-142:880681:880845 [2] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-172-142:880681:880845 [2] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-172-142:880681:880845 [2] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-172-142:880681:880845 [2] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-172-142:880681:880845 [2] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-172-142:880681:880845 [2] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-172-142:880681:880845 [2] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/280
+ip-26-0-172-142:880681:880845 [2] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-172-57:942617:942774 [6] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-172-57:942617:942774 [6] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-172-57:942617:942774 [6] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-172-57:942617:942774 [6] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-172-57:942617:942774 [6] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-172-57:942617:942774 [6] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-172-57:942617:942774 [6] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-172-57:942617:942774 [6] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-172-57:942617:942774 [6] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-172-57:942617:942774 [6] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-172-57:942617:942774 [6] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-172-57:942617:942774 [6] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-172-57:942617:942774 [6] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-172-57:942617:942774 [6] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-172-57:942617:942774 [6] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-172-57:942617:942774 [6] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-172-57:942617:942774 [6] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-172-57:942617:942774 [6] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-172-57:942617:942774 [6] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-172-57:942617:942774 [6] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-172-57:942617:942774 [6] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-172-57:942617:942774 [6] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-172-57:942617:942774 [6] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-172-57:942617:942774 [6] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-172-57:942617:942774 [6] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-172-57:942617:942774 [6] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-172-57:942617:942774 [6] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-172-57:942617:942774 [6] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-172-57:942617:942774 [6] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-172-57:942617:942774 [6] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-172-57:942617:942774 [6] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-172-57:942617:942774 [6] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-172-142:880686:880841 [7] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-172-142:880686:880841 [7] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-172-142:880686:880841 [7] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-172-142:880686:880841 [7] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-172-142:880686:880841 [7] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-172-142:880686:880841 [7] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-172-142:880686:880841 [7] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-172-142:880686:880841 [7] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-172-142:880686:880841 [7] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-172-142:880686:880841 [7] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-172-142:880686:880841 [7] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-172-142:880686:880841 [7] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-172-142:880686:880841 [7] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-172-142:880686:880841 [7] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-172-142:880686:880841 [7] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-172-142:880686:880841 [7] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-172-142:880686:880841 [7] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-172-142:880686:880841 [7] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-172-142:880686:880841 [7] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-172-142:880686:880841 [7] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-172-142:880686:880841 [7] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-172-142:880686:880841 [7] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-172-142:880686:880841 [7] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-172-142:880686:880841 [7] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-172-142:880686:880841 [7] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-172-142:880686:880841 [7] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-172-142:880686:880841 [7] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-172-142:880686:880841 [7] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-172-142:880686:880841 [7] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-172-142:880686:880841 [7] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-172-142:880686:880841 [7] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-172-142:880686:880841 [7] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-172-57:942617:942774 [6] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/280
+ip-26-0-172-57:942617:942774 [6] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-172-142:880685:880847 [6] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-172-142:880685:880847 [6] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-172-142:880685:880847 [6] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-172-142:880685:880847 [6] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-172-142:880685:880847 [6] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-172-142:880685:880847 [6] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-172-142:880685:880847 [6] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-172-142:880685:880847 [6] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-172-142:880685:880847 [6] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-172-142:880685:880847 [6] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-172-142:880685:880847 [6] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-172-142:880685:880847 [6] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-172-142:880685:880847 [6] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-172-142:880685:880847 [6] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-172-142:880685:880847 [6] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-172-142:880685:880847 [6] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-172-142:880685:880847 [6] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-172-142:880685:880847 [6] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-172-142:880685:880847 [6] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-172-142:880685:880847 [6] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-172-142:880685:880847 [6] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-172-142:880685:880847 [6] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-172-142:880685:880847 [6] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-172-142:880685:880847 [6] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-172-142:880685:880847 [6] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-172-142:880685:880847 [6] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-172-142:880685:880847 [6] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-172-142:880685:880847 [6] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-172-142:880685:880847 [6] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-172-142:880685:880847 [6] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-172-142:880685:880847 [6] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-172-142:880685:880847 [6] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-172-142:880680:880843 [1] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-172-142:880680:880843 [1] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-172-142:880680:880843 [1] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-172-142:880680:880843 [1] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-172-142:880680:880843 [1] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-172-142:880680:880843 [1] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-172-142:880686:880841 [7] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/280
+ip-26-0-172-142:880680:880843 [1] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-172-142:880680:880843 [1] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-172-142:880680:880843 [1] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-172-142:880680:880843 [1] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-172-142:880680:880843 [1] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-172-142:880680:880843 [1] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-172-142:880680:880843 [1] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-172-142:880680:880843 [1] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-172-142:880680:880843 [1] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-172-142:880680:880843 [1] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-172-142:880680:880843 [1] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-172-142:880680:880843 [1] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-172-142:880680:880843 [1] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-172-142:880680:880843 [1] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-172-142:880680:880843 [1] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-172-142:880680:880843 [1] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-172-142:880680:880843 [1] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-172-142:880680:880843 [1] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-172-142:880686:880841 [7] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-172-142:880680:880843 [1] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-172-142:880680:880843 [1] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-172-142:880680:880843 [1] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-172-142:880680:880843 [1] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-172-142:880680:880843 [1] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-172-142:880680:880843 [1] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-172-142:880680:880843 [1] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-172-142:880680:880843 [1] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/280
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/280
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/280
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-172-73:926594:926762 [0] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-172-73:926594:926762 [0] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-172-73:926594:926762 [0] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-172-73:926594:926762 [0] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-172-73:926594:926762 [0] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-172-73:926594:926762 [0] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-172-73:926594:926762 [0] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-172-73:926594:926762 [0] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-172-73:926594:926762 [0] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-172-73:926594:926762 [0] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-172-73:926594:926762 [0] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-172-73:926594:926762 [0] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-172-73:926594:926762 [0] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-172-73:926594:926762 [0] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-172-73:926594:926762 [0] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-172-73:926594:926762 [0] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-172-73:926594:926762 [0] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-172-73:926594:926762 [0] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-172-73:926594:926762 [0] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-172-73:926594:926762 [0] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-172-73:926594:926762 [0] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-172-73:926594:926762 [0] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-172-73:926594:926762 [0] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-172-73:926594:926762 [0] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-172-73:926594:926762 [0] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-172-73:926594:926762 [0] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-172-73:926594:926762 [0] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-172-73:926594:926762 [0] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-172-73:926594:926762 [0] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-172-73:926594:926762 [0] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-172-73:926594:926762 [0] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-172-73:926594:926762 [0] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-172-57:942616:942777 [5] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-172-57:942616:942777 [5] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-172-57:942616:942777 [5] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-172-57:942616:942777 [5] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-172-57:942616:942777 [5] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-172-57:942616:942777 [5] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-172-57:942616:942777 [5] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-172-57:942616:942777 [5] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-172-57:942616:942777 [5] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-172-57:942616:942777 [5] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-172-57:942616:942777 [5] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-172-57:942616:942777 [5] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-172-57:942616:942777 [5] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-172-57:942616:942777 [5] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-172-57:942616:942777 [5] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-172-57:942616:942777 [5] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-172-57:942616:942777 [5] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-172-57:942616:942777 [5] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-172-57:942616:942777 [5] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-172-57:942616:942777 [5] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-172-57:942616:942777 [5] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-172-57:942616:942777 [5] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-172-57:942616:942777 [5] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-172-57:942616:942777 [5] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-172-57:942616:942777 [5] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-172-57:942616:942777 [5] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-172-57:942616:942777 [5] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-172-57:942616:942777 [5] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-172-57:942616:942777 [5] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-172-57:942616:942777 [5] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-172-57:942616:942777 [5] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-172-57:942616:942777 [5] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/280
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/280
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/280
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-172-73:926594:926762 [0] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/280
+ip-26-0-172-73:926594:926762 [0] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-172-57:942616:942777 [5] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/280
+ip-26-0-172-57:942616:942777 [5] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-172-73:926596:926756 [2] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-172-73:926596:926756 [2] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-172-73:926596:926756 [2] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-172-73:926596:926756 [2] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-172-73:926596:926756 [2] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-172-73:926596:926756 [2] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-172-73:926596:926756 [2] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-172-73:926596:926756 [2] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-172-73:926596:926756 [2] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-172-73:926596:926756 [2] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-172-73:926596:926756 [2] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-172-73:926596:926756 [2] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-172-73:926596:926756 [2] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-172-73:926596:926756 [2] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-172-73:926596:926756 [2] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-172-73:926596:926756 [2] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-172-73:926596:926756 [2] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-172-73:926596:926756 [2] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-172-73:926596:926756 [2] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-172-73:926596:926756 [2] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-172-73:926596:926756 [2] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-172-73:926596:926756 [2] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-172-73:926596:926756 [2] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-172-73:926596:926756 [2] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-172-73:926596:926756 [2] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-172-73:926596:926756 [2] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-172-73:926596:926756 [2] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-172-73:926596:926756 [2] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-172-73:926596:926756 [2] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-172-73:926596:926756 [2] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-172-73:926596:926756 [2] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-172-73:926596:926756 [2] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-172-147:618589:618744 [5] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-172-147:618589:618744 [5] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-172-147:618589:618744 [5] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-172-147:618589:618744 [5] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-172-147:618589:618744 [5] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-172-147:618589:618744 [5] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-172-147:618589:618744 [5] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-172-147:618589:618744 [5] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-172-147:618589:618744 [5] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-172-147:618589:618744 [5] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-172-147:618589:618744 [5] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-172-147:618589:618744 [5] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-172-147:618589:618744 [5] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-172-147:618589:618744 [5] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-172-147:618589:618744 [5] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-172-147:618589:618744 [5] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-172-147:618589:618744 [5] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-172-147:618589:618744 [5] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-172-147:618589:618744 [5] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-172-147:618589:618744 [5] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-172-147:618589:618744 [5] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-172-147:618589:618744 [5] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-172-147:618589:618744 [5] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-172-147:618589:618744 [5] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-172-147:618589:618744 [5] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-172-147:618589:618744 [5] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-172-147:618589:618744 [5] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-172-147:618589:618744 [5] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-172-147:618589:618744 [5] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-172-147:618589:618744 [5] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-172-147:618589:618744 [5] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-172-147:618589:618744 [5] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-172-73:926596:926756 [2] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/280
+ip-26-0-172-73:926596:926756 [2] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-172-147:618589:618744 [5] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/280
+ip-26-0-172-147:618589:618744 [5] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-172-57:942618:942778 [7] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-172-57:942618:942778 [7] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-172-57:942618:942778 [7] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-172-57:942618:942778 [7] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-172-57:942618:942778 [7] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-172-57:942618:942778 [7] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-172-57:942618:942778 [7] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-172-57:942618:942778 [7] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-172-57:942618:942778 [7] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-172-57:942618:942778 [7] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-172-57:942618:942778 [7] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-172-57:942618:942778 [7] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-172-57:942618:942778 [7] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-172-57:942618:942778 [7] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-172-57:942618:942778 [7] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-172-57:942618:942778 [7] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-172-57:942618:942778 [7] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-172-57:942618:942778 [7] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-172-57:942618:942778 [7] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-172-57:942618:942778 [7] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-172-57:942618:942778 [7] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-172-57:942618:942778 [7] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-172-57:942618:942778 [7] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-172-57:942618:942778 [7] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-172-57:942618:942778 [7] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-172-57:942618:942778 [7] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-172-57:942618:942778 [7] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-172-57:942618:942778 [7] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-172-57:942618:942778 [7] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-172-57:942618:942778 [7] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-172-57:942618:942778 [7] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-172-57:942618:942778 [7] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-172-57:942613:942776 [2] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-172-57:942613:942776 [2] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-172-57:942613:942776 [2] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-172-57:942613:942776 [2] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-172-57:942613:942776 [2] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-172-57:942613:942776 [2] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-172-57:942613:942776 [2] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-172-57:942613:942776 [2] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-172-57:942613:942776 [2] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-172-57:942613:942776 [2] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-172-57:942613:942776 [2] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-172-57:942613:942776 [2] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-172-57:942613:942776 [2] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-172-57:942613:942776 [2] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-172-57:942613:942776 [2] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-172-57:942613:942776 [2] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-172-57:942613:942776 [2] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-172-57:942613:942776 [2] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-172-57:942613:942776 [2] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-172-57:942613:942776 [2] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-172-57:942613:942776 [2] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-172-57:942613:942776 [2] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-172-57:942613:942776 [2] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-172-57:942613:942776 [2] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-172-57:942613:942776 [2] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-172-57:942613:942776 [2] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-172-57:942613:942776 [2] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-172-57:942613:942776 [2] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-172-57:942613:942776 [2] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-172-57:942613:942776 [2] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-172-57:942613:942776 [2] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-172-57:942613:942776 [2] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-172-57:942618:942778 [7] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/280
+ip-26-0-172-57:942618:942778 [7] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-172-57:942613:942776 [2] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/280
+ip-26-0-172-57:942613:942776 [2] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-160-225:94344:94510 [2] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-160-225:94344:94510 [2] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-160-225:94344:94510 [2] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-160-225:94344:94510 [2] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-160-225:94344:94510 [2] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-160-225:94346:94511 [4] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-160-225:94344:94510 [2] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-160-225:94346:94511 [4] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-160-225:94344:94510 [2] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-160-225:94346:94511 [4] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-160-225:94344:94510 [2] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-160-225:94346:94511 [4] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-160-225:94344:94510 [2] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-160-225:94346:94511 [4] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-160-225:94344:94510 [2] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-160-225:94346:94511 [4] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-160-225:94344:94510 [2] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-160-225:94346:94511 [4] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-160-225:94344:94510 [2] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-160-225:94346:94511 [4] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-160-225:94344:94510 [2] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-160-225:94346:94511 [4] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-160-225:94344:94510 [2] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-160-225:94346:94511 [4] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-160-225:94344:94510 [2] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-160-225:94346:94511 [4] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-160-225:94344:94510 [2] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-160-225:94346:94511 [4] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-160-225:94344:94510 [2] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-160-225:94346:94511 [4] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-160-225:94344:94510 [2] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-160-225:94346:94511 [4] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-160-225:94344:94510 [2] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-160-225:94346:94511 [4] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-160-225:94344:94510 [2] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-160-225:94346:94511 [4] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-160-225:94344:94510 [2] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-160-225:94346:94511 [4] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-160-225:94344:94510 [2] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-160-225:94346:94511 [4] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-160-225:94344:94510 [2] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-160-225:94346:94511 [4] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-160-225:94344:94510 [2] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-160-225:94346:94511 [4] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-160-225:94344:94510 [2] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-160-225:94346:94511 [4] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-160-225:94344:94510 [2] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-160-225:94346:94511 [4] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-160-225:94344:94510 [2] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-160-225:94346:94511 [4] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-173-7:811255:811411 [7] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-173-7:811255:811411 [7] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-173-7:811255:811411 [7] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-173-7:811255:811411 [7] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-173-7:811255:811411 [7] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-173-7:811255:811411 [7] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-173-7:811255:811411 [7] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-173-7:811255:811411 [7] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-173-7:811255:811411 [7] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-173-7:811255:811411 [7] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-173-7:811255:811411 [7] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-160-225:94344:94510 [2] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-160-225:94346:94511 [4] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-160-225:94344:94510 [2] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-160-225:94346:94511 [4] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-160-225:94344:94510 [2] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-160-225:94346:94511 [4] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-160-225:94344:94510 [2] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-160-225:94346:94511 [4] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-160-225:94344:94510 [2] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-160-225:94346:94511 [4] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-160-225:94346:94511 [4] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-160-225:94346:94511 [4] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-173-7:811255:811411 [7] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-173-7:811255:811411 [7] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-173-7:811255:811411 [7] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-173-7:811255:811411 [7] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-173-7:811255:811411 [7] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-173-7:811255:811411 [7] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-173-7:811255:811411 [7] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-173-7:811255:811411 [7] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-173-7:811255:811411 [7] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-173-7:811255:811411 [7] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-173-7:811255:811411 [7] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-173-7:811255:811411 [7] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-160-225:94346:94511 [4] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-160-225:94346:94511 [4] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-173-7:811255:811411 [7] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-173-7:811255:811411 [7] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-173-7:811255:811411 [7] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-173-7:811255:811411 [7] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-173-7:811255:811411 [7] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-173-7:811255:811411 [7] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-173-7:811255:811411 [7] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-173-7:811255:811411 [7] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-173-7:811255:811411 [7] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-160-225:94344:94510 [2] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/280
+ip-26-0-160-225:94344:94510 [2] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-160-225:94346:94511 [4] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/280
+ip-26-0-160-225:94346:94511 [4] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-160-225:94343:94508 [1] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-160-225:94343:94508 [1] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-160-225:94343:94508 [1] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-160-225:94343:94508 [1] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-160-225:94343:94508 [1] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-160-225:94343:94508 [1] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-160-225:94343:94508 [1] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-160-225:94343:94508 [1] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-160-225:94343:94508 [1] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-160-225:94343:94508 [1] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-160-225:94343:94508 [1] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-160-225:94343:94508 [1] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-160-225:94343:94508 [1] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-160-225:94343:94508 [1] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-160-225:94343:94508 [1] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-160-225:94343:94508 [1] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-160-225:94343:94508 [1] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-160-225:94343:94508 [1] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-160-225:94343:94508 [1] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-160-225:94343:94508 [1] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-160-225:94343:94508 [1] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-160-225:94343:94508 [1] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-160-225:94343:94508 [1] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-160-225:94343:94508 [1] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-160-225:94343:94508 [1] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-160-225:94343:94508 [1] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-160-225:94343:94508 [1] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-160-225:94343:94508 [1] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-160-225:94343:94508 [1] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-160-225:94343:94508 [1] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-160-225:94343:94508 [1] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-160-225:94343:94508 [1] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-160-225:94343:94508 [1] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/280
+ip-26-0-160-225:94343:94508 [1] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-173-7:811255:811411 [7] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/280
+ip-26-0-173-7:811255:811411 [7] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-172-252:688207:688366 [7] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-172-252:688207:688366 [7] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-172-252:688207:688366 [7] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-172-252:688207:688366 [7] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-172-252:688207:688366 [7] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-172-252:688207:688366 [7] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-172-252:688207:688366 [7] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-172-252:688207:688366 [7] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-172-252:688207:688366 [7] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-172-252:688207:688366 [7] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-172-252:688207:688366 [7] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-172-252:688207:688366 [7] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-172-252:688207:688366 [7] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-172-252:688207:688366 [7] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-172-252:688207:688366 [7] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-172-252:688207:688366 [7] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-172-252:688207:688366 [7] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-172-252:688207:688366 [7] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-172-252:688207:688366 [7] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-172-252:688207:688366 [7] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-172-252:688207:688366 [7] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-172-252:688207:688366 [7] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-172-252:688207:688366 [7] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-172-252:688207:688366 [7] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-172-252:688207:688366 [7] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-172-252:688207:688366 [7] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-172-252:688207:688366 [7] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-172-252:688207:688366 [7] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-172-252:688207:688366 [7] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-172-252:688207:688366 [7] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-172-252:688207:688366 [7] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-172-252:688207:688366 [7] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-172-252:688207:688366 [7] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/280
+ip-26-0-172-252:688207:688366 [7] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-172-73:926598:926761 [4] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-172-73:926598:926761 [4] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-172-73:926598:926761 [4] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-172-73:926598:926761 [4] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-172-73:926598:926761 [4] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-172-73:926598:926761 [4] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-172-73:926598:926761 [4] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-172-73:926598:926761 [4] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-172-73:926598:926761 [4] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-172-73:926598:926761 [4] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-172-73:926598:926761 [4] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-172-73:926598:926761 [4] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-172-73:926598:926761 [4] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-172-73:926598:926761 [4] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-172-73:926598:926761 [4] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-172-73:926598:926761 [4] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-172-73:926598:926761 [4] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-172-73:926598:926761 [4] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-172-73:926598:926761 [4] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-172-73:926598:926761 [4] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-172-73:926598:926761 [4] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-172-73:926598:926761 [4] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-172-73:926598:926761 [4] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-172-73:926598:926761 [4] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-172-73:926598:926761 [4] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-172-73:926598:926761 [4] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-172-73:926598:926761 [4] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-172-73:926598:926761 [4] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-172-73:926598:926761 [4] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-172-73:926598:926761 [4] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-172-73:926598:926761 [4] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-172-73:926598:926761 [4] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-172-73:926598:926761 [4] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/280
+ip-26-0-172-73:926598:926761 [4] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-172-142:880685:880847 [6] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/280
+ip-26-0-172-142:880685:880847 [6] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-172-142:880680:880843 [1] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/280
+ip-26-0-172-142:880680:880843 [1] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-172-73:926600:926757 [6] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-172-73:926600:926757 [6] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-172-73:926600:926757 [6] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-172-73:926600:926757 [6] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-172-73:926600:926757 [6] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-172-73:926600:926757 [6] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-172-73:926600:926757 [6] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-172-73:926600:926757 [6] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-172-73:926600:926757 [6] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-172-73:926600:926757 [6] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-172-73:926600:926757 [6] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-172-73:926600:926757 [6] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-172-73:926600:926757 [6] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-172-73:926600:926757 [6] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-172-73:926600:926757 [6] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-172-73:926600:926757 [6] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-172-73:926600:926757 [6] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-172-73:926600:926757 [6] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-172-73:926600:926757 [6] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-172-73:926600:926757 [6] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-172-73:926600:926757 [6] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-172-73:926600:926757 [6] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-172-73:926600:926757 [6] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-172-73:926600:926757 [6] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-172-73:926600:926757 [6] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-172-73:926600:926757 [6] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-172-73:926600:926757 [6] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-172-73:926600:926757 [6] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-172-73:926600:926757 [6] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-172-73:926600:926757 [6] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-172-73:926600:926757 [6] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-172-73:926600:926757 [6] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-172-73:926599:926755 [5] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-172-73:926599:926755 [5] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-172-73:926599:926755 [5] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-172-73:926599:926755 [5] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-172-73:926599:926755 [5] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-172-73:926599:926755 [5] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-172-73:926599:926755 [5] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-172-73:926599:926755 [5] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-172-73:926599:926755 [5] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-172-73:926599:926755 [5] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-172-73:926599:926755 [5] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-172-73:926599:926755 [5] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-172-73:926599:926755 [5] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-172-73:926599:926755 [5] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-172-73:926599:926755 [5] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-172-73:926599:926755 [5] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-172-73:926599:926755 [5] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-172-147:618591:618746 [7] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-172-147:618591:618746 [7] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-172-147:618591:618746 [7] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-172-147:618591:618746 [7] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-172-147:618591:618746 [7] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-172-147:618591:618746 [7] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-172-147:618591:618746 [7] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-172-73:926599:926755 [5] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-172-73:926599:926755 [5] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-172-73:926599:926755 [5] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-172-73:926599:926755 [5] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-172-73:926599:926755 [5] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-172-73:926599:926755 [5] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-172-73:926599:926755 [5] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-172-73:926599:926755 [5] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-172-73:926599:926755 [5] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-172-73:926599:926755 [5] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-172-73:926599:926755 [5] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-172-73:926599:926755 [5] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-172-73:926599:926755 [5] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-172-73:926599:926755 [5] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-172-73:926599:926755 [5] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-172-147:618591:618746 [7] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-172-147:618591:618746 [7] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-172-147:618591:618746 [7] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-172-147:618591:618746 [7] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-172-147:618591:618746 [7] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-172-147:618591:618746 [7] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-172-147:618591:618746 [7] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-172-147:618591:618746 [7] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-172-147:618591:618746 [7] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-172-147:618591:618746 [7] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-172-147:618591:618746 [7] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-172-147:618591:618746 [7] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-172-147:618591:618746 [7] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-172-147:618591:618746 [7] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-172-147:618591:618746 [7] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-172-147:618591:618746 [7] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-172-147:618591:618746 [7] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-172-147:618591:618746 [7] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-172-147:618591:618746 [7] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-172-147:618591:618746 [7] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-172-147:618591:618746 [7] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-172-147:618591:618746 [7] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-172-147:618591:618746 [7] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-172-147:618591:618746 [7] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-172-147:618591:618746 [7] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-172-73:926595:926760 [1] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-172-73:926595:926760 [1] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-172-73:926595:926760 [1] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-172-73:926595:926760 [1] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-172-73:926595:926760 [1] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-172-73:926595:926760 [1] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-172-73:926595:926760 [1] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-172-73:926595:926760 [1] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-172-73:926595:926760 [1] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-172-73:926595:926760 [1] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-172-73:926595:926760 [1] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-172-73:926595:926760 [1] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-172-73:926595:926760 [1] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-172-73:926595:926760 [1] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-172-73:926595:926760 [1] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-172-73:926595:926760 [1] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-172-73:926595:926760 [1] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-172-73:926595:926760 [1] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-172-73:926595:926760 [1] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-172-73:926595:926760 [1] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-172-73:926595:926760 [1] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-172-73:926595:926760 [1] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-172-73:926595:926760 [1] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-172-73:926595:926760 [1] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-172-73:926595:926760 [1] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-172-73:926595:926760 [1] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-172-73:926595:926760 [1] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-172-73:926595:926760 [1] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-172-73:926595:926760 [1] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-172-73:926595:926760 [1] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-172-73:926595:926760 [1] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-172-73:926595:926760 [1] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-172-73:926600:926757 [6] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/280
+ip-26-0-172-73:926601:926759 [7] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-172-73:926601:926759 [7] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-172-73:926601:926759 [7] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-172-73:926601:926759 [7] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-172-73:926601:926759 [7] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-172-73:926600:926757 [6] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-172-73:926601:926759 [7] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-172-73:926601:926759 [7] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-172-73:926601:926759 [7] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-172-73:926601:926759 [7] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-172-73:926601:926759 [7] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-172-73:926601:926759 [7] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-172-73:926601:926759 [7] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-172-73:926601:926759 [7] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-172-73:926601:926759 [7] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-172-73:926601:926759 [7] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-172-73:926601:926759 [7] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-172-73:926601:926759 [7] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-172-73:926601:926759 [7] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-172-73:926601:926759 [7] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-172-73:926601:926759 [7] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-172-73:926601:926759 [7] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-172-73:926601:926759 [7] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-172-73:926601:926759 [7] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-172-73:926601:926759 [7] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-172-73:926601:926759 [7] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-172-73:926601:926759 [7] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-172-73:926601:926759 [7] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-172-73:926601:926759 [7] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-172-73:926601:926759 [7] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-172-73:926601:926759 [7] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-172-73:926601:926759 [7] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-172-73:926601:926759 [7] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-172-147:618591:618746 [7] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/280
+ip-26-0-172-147:618591:618746 [7] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-173-7:811254:811414 [6] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-173-7:811254:811414 [6] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-173-7:811254:811414 [6] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-173-7:811254:811414 [6] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-173-7:811254:811414 [6] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-173-7:811254:811414 [6] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-173-7:811254:811414 [6] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-173-7:811254:811414 [6] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-173-7:811254:811414 [6] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-173-7:811254:811414 [6] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-173-7:811254:811414 [6] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-173-7:811254:811414 [6] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-173-7:811254:811414 [6] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-173-7:811254:811414 [6] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-173-7:811254:811414 [6] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-173-7:811254:811414 [6] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-173-7:811254:811414 [6] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-173-7:811254:811414 [6] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-173-7:811254:811414 [6] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-173-7:811254:811414 [6] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-173-7:811254:811414 [6] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-173-7:811254:811414 [6] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-173-7:811254:811414 [6] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-173-7:811254:811414 [6] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-173-7:811254:811414 [6] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-173-7:811254:811414 [6] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-173-7:811254:811414 [6] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-173-7:811254:811414 [6] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-173-7:811254:811414 [6] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-173-7:811254:811414 [6] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-173-7:811254:811414 [6] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-173-7:811254:811414 [6] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-173-7:811250:811412 [2] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-173-7:811250:811412 [2] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-173-7:811250:811412 [2] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-173-7:811250:811412 [2] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-173-7:811250:811412 [2] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-173-7:811250:811412 [2] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-173-7:811250:811412 [2] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-173-7:811250:811412 [2] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-173-7:811250:811412 [2] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-173-7:811250:811412 [2] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-173-7:811250:811412 [2] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-173-7:811250:811412 [2] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-173-7:811250:811412 [2] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-173-7:811250:811412 [2] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-173-7:811250:811412 [2] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-173-7:811250:811412 [2] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-173-7:811250:811412 [2] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-173-7:811250:811412 [2] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-173-7:811250:811412 [2] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-173-7:811250:811412 [2] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-173-7:811250:811412 [2] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-173-7:811250:811412 [2] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-173-7:811250:811412 [2] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-173-7:811250:811412 [2] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-173-7:811250:811412 [2] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-173-7:811250:811412 [2] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-173-7:811250:811412 [2] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-173-7:811250:811412 [2] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-173-7:811250:811412 [2] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-173-7:811250:811412 [2] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-173-7:811250:811412 [2] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-173-7:811250:811412 [2] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-173-7:811254:811414 [6] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/280
+ip-26-0-173-7:811254:811414 [6] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-172-147:618588:618750 [4] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-172-147:618588:618750 [4] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-172-147:618588:618750 [4] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-172-147:618588:618750 [4] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-172-147:618588:618750 [4] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-172-147:618588:618750 [4] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-172-147:618588:618750 [4] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-172-147:618588:618750 [4] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-172-147:618588:618750 [4] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-172-147:618588:618750 [4] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-172-147:618588:618750 [4] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-172-147:618588:618750 [4] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-172-147:618588:618750 [4] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-172-147:618588:618750 [4] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-172-147:618588:618750 [4] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-172-147:618588:618750 [4] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-172-147:618588:618750 [4] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-172-147:618588:618750 [4] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-172-147:618588:618750 [4] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-172-147:618588:618750 [4] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-172-147:618588:618750 [4] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-172-147:618588:618750 [4] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-172-147:618588:618750 [4] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-172-147:618588:618750 [4] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-172-147:618588:618750 [4] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-172-147:618588:618750 [4] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-172-147:618588:618750 [4] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-172-147:618588:618750 [4] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-172-147:618588:618750 [4] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-172-147:618588:618750 [4] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-172-147:618588:618750 [4] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-172-147:618588:618750 [4] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-173-7:811250:811412 [2] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/280
+ip-26-0-173-7:811250:811412 [2] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-172-147:618588:618750 [4] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/280
+ip-26-0-172-147:618588:618750 [4] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-173-7:811252:811408 [4] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-173-7:811252:811408 [4] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-173-7:811252:811408 [4] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-173-7:811252:811408 [4] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-173-7:811252:811408 [4] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-173-7:811252:811408 [4] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-173-7:811252:811408 [4] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-173-7:811252:811408 [4] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-173-7:811252:811408 [4] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-173-7:811252:811408 [4] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-173-7:811252:811408 [4] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-173-7:811252:811408 [4] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-173-7:811252:811408 [4] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-173-7:811252:811408 [4] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-173-7:811252:811408 [4] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-173-7:811252:811408 [4] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-173-7:811252:811408 [4] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-173-7:811252:811408 [4] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-173-7:811252:811408 [4] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-173-7:811252:811408 [4] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-173-7:811252:811408 [4] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-173-7:811252:811408 [4] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-173-7:811252:811408 [4] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-173-7:811252:811408 [4] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-173-7:811252:811408 [4] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-173-7:811252:811408 [4] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-173-7:811252:811408 [4] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-173-7:811252:811408 [4] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-173-7:811252:811408 [4] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-173-7:811252:811408 [4] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-173-7:811252:811408 [4] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-173-7:811252:811408 [4] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-173-7:811253:811410 [5] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-173-7:811253:811410 [5] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-173-7:811253:811410 [5] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-173-7:811253:811410 [5] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-173-7:811253:811410 [5] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-173-7:811253:811410 [5] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-173-7:811253:811410 [5] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-173-7:811253:811410 [5] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-173-7:811253:811410 [5] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-173-7:811253:811410 [5] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-173-7:811253:811410 [5] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-173-7:811253:811410 [5] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-173-7:811253:811410 [5] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-173-7:811253:811410 [5] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-173-7:811253:811410 [5] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-173-7:811253:811410 [5] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-173-7:811253:811410 [5] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-173-7:811253:811410 [5] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-173-7:811253:811410 [5] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-173-7:811253:811410 [5] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-173-7:811253:811410 [5] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-173-7:811253:811410 [5] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-173-7:811253:811410 [5] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-173-7:811253:811410 [5] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-173-7:811253:811410 [5] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-173-7:811253:811410 [5] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-173-7:811253:811410 [5] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-173-7:811253:811410 [5] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-173-7:811253:811410 [5] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-173-7:811253:811410 [5] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-173-7:811253:811410 [5] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-173-7:811253:811410 [5] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-172-147:618590:618749 [6] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-172-147:618590:618749 [6] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-172-147:618590:618749 [6] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-172-147:618590:618749 [6] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-172-147:618590:618749 [6] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-172-147:618590:618749 [6] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-172-147:618590:618749 [6] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-172-147:618590:618749 [6] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-172-147:618590:618749 [6] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-172-147:618590:618749 [6] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-172-147:618590:618749 [6] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-172-147:618590:618749 [6] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-172-147:618590:618749 [6] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-172-147:618590:618749 [6] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-172-147:618590:618749 [6] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-172-147:618590:618749 [6] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-172-147:618590:618749 [6] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-172-147:618590:618749 [6] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-172-147:618590:618749 [6] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-172-147:618590:618749 [6] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-172-147:618590:618749 [6] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-172-147:618590:618749 [6] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-172-147:618590:618749 [6] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-172-147:618590:618749 [6] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-172-147:618590:618749 [6] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-172-147:618590:618749 [6] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-172-147:618590:618749 [6] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-172-147:618590:618749 [6] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-172-147:618590:618749 [6] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-172-147:618590:618749 [6] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-172-147:618590:618749 [6] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-172-147:618590:618749 [6] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-173-7:811252:811408 [4] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/280
+ip-26-0-173-7:811252:811408 [4] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-172-252:688201:688364 [1] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-172-252:688201:688364 [1] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-172-252:688201:688364 [1] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-172-252:688201:688364 [1] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-172-252:688201:688364 [1] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-172-252:688201:688364 [1] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-172-252:688201:688364 [1] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-172-252:688201:688364 [1] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-172-252:688201:688364 [1] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-172-252:688201:688364 [1] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-172-252:688201:688364 [1] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-172-252:688201:688364 [1] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-172-252:688201:688364 [1] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-172-252:688201:688364 [1] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-172-252:688201:688364 [1] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-172-252:688201:688364 [1] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-172-252:688201:688364 [1] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-172-252:688201:688364 [1] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-172-252:688201:688364 [1] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-172-252:688201:688364 [1] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-172-252:688201:688364 [1] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-172-252:688201:688364 [1] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-172-252:688201:688364 [1] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-172-252:688201:688364 [1] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-172-252:688201:688364 [1] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-173-7:811253:811410 [5] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/280
+ip-26-0-172-252:688201:688364 [1] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-172-252:688201:688364 [1] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-172-252:688201:688364 [1] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-172-252:688201:688364 [1] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-172-252:688201:688364 [1] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-172-252:688201:688364 [1] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-172-252:688201:688364 [1] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-173-7:811253:811410 [5] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-172-147:618590:618749 [6] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/280
+ip-26-0-172-147:618590:618749 [6] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-172-252:688201:688364 [1] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/280
+ip-26-0-172-252:688201:688364 [1] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-173-7:811249:811413 [1] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-173-7:811249:811413 [1] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-173-7:811249:811413 [1] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-173-7:811249:811413 [1] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-173-7:811249:811413 [1] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-173-7:811249:811413 [1] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-173-7:811249:811413 [1] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-173-7:811249:811413 [1] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-173-7:811249:811413 [1] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-173-7:811249:811413 [1] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-173-7:811249:811413 [1] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-173-7:811249:811413 [1] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-173-7:811249:811413 [1] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-173-7:811249:811413 [1] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-173-7:811249:811413 [1] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-173-7:811249:811413 [1] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-173-7:811249:811413 [1] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-173-7:811249:811413 [1] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-173-7:811249:811413 [1] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-173-7:811249:811413 [1] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-173-7:811249:811413 [1] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-173-7:811249:811413 [1] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-173-7:811249:811413 [1] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-173-7:811249:811413 [1] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-173-7:811249:811413 [1] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-173-7:811249:811413 [1] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-173-7:811249:811413 [1] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-173-7:811249:811413 [1] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-173-7:811249:811413 [1] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-173-7:811249:811413 [1] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-173-7:811249:811413 [1] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-173-7:811249:811413 [1] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-173-7:811249:811413 [1] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/280
+ip-26-0-172-147:618585:618748 [1] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-172-147:618585:618748 [1] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-172-147:618585:618748 [1] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-172-147:618585:618748 [1] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-172-147:618585:618748 [1] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-172-147:618585:618748 [1] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-172-147:618585:618748 [1] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-172-147:618585:618748 [1] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-172-147:618585:618748 [1] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-172-147:618585:618748 [1] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-172-147:618585:618748 [1] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-173-7:811249:811413 [1] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-172-147:618585:618748 [1] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-172-147:618585:618748 [1] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-172-147:618585:618748 [1] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-172-147:618585:618748 [1] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-172-147:618585:618748 [1] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-172-147:618585:618748 [1] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-172-147:618585:618748 [1] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-172-147:618585:618748 [1] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-172-147:618585:618748 [1] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-172-147:618585:618748 [1] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-172-147:618585:618748 [1] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-172-147:618585:618748 [1] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-172-147:618585:618748 [1] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-172-147:618585:618748 [1] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-172-147:618585:618748 [1] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-172-147:618585:618748 [1] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-172-147:618585:618748 [1] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-172-147:618585:618748 [1] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-172-147:618585:618748 [1] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-172-147:618585:618748 [1] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-172-147:618585:618748 [1] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-173-7:811248:811415 [0] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-173-7:811248:811415 [0] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-173-7:811248:811415 [0] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-173-7:811248:811415 [0] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-173-7:811248:811415 [0] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-173-7:811248:811415 [0] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-173-7:811248:811415 [0] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-173-7:811248:811415 [0] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-173-7:811248:811415 [0] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-173-7:811248:811415 [0] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-173-7:811248:811415 [0] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-173-7:811248:811415 [0] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-173-7:811248:811415 [0] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-173-7:811248:811415 [0] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-173-7:811248:811415 [0] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-173-7:811248:811415 [0] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-173-7:811248:811415 [0] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-173-7:811248:811415 [0] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-173-7:811248:811415 [0] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-173-7:811248:811415 [0] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-173-7:811248:811415 [0] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-173-7:811248:811415 [0] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-173-7:811248:811415 [0] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-173-7:811248:811415 [0] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-173-7:811248:811415 [0] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-173-7:811248:811415 [0] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-173-7:811248:811415 [0] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-173-7:811248:811415 [0] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-173-7:811248:811415 [0] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-173-7:811248:811415 [0] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-173-7:811248:811415 [0] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-173-7:811248:811415 [0] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-173-7:811248:811415 [0] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/280
+ip-26-0-172-147:618585:618748 [1] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/280
+ip-26-0-173-7:811248:811415 [0] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-172-147:618585:618748 [1] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-172-147:618584:618747 [0] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-172-147:618584:618747 [0] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-172-147:618584:618747 [0] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-172-147:618584:618747 [0] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-172-147:618584:618747 [0] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-172-147:618584:618747 [0] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-172-147:618584:618747 [0] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-172-147:618584:618747 [0] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-172-147:618584:618747 [0] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-172-147:618584:618747 [0] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-172-147:618584:618747 [0] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-172-147:618584:618747 [0] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-172-147:618584:618747 [0] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-172-147:618584:618747 [0] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-172-147:618584:618747 [0] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-172-147:618584:618747 [0] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-172-147:618584:618747 [0] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-172-147:618584:618747 [0] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-172-147:618584:618747 [0] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-172-147:618584:618747 [0] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-172-147:618584:618747 [0] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-172-147:618584:618747 [0] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-172-147:618584:618747 [0] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-172-147:618584:618747 [0] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-172-147:618584:618747 [0] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-172-147:618584:618747 [0] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-172-147:618584:618747 [0] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-172-147:618584:618747 [0] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-172-147:618584:618747 [0] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-172-147:618584:618747 [0] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-172-147:618584:618747 [0] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-172-147:618584:618747 [0] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-173-7:811251:811409 [3] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-173-7:811251:811409 [3] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-173-7:811251:811409 [3] NCCL INFO Using network Libfabric
+ip-26-0-172-147:618584:618747 [0] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/280
+ip-26-0-172-147:618584:618747 [0] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-172-147:618586:618745 [2] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-172-147:618586:618745 [2] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-172-147:618586:618745 [2] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-172-147:618586:618745 [2] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-172-147:618586:618745 [2] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-172-147:618586:618745 [2] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-172-147:618586:618745 [2] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-172-147:618586:618745 [2] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-172-147:618586:618745 [2] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-172-147:618586:618745 [2] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-172-147:618586:618745 [2] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-172-147:618586:618745 [2] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-172-147:618586:618745 [2] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-172-147:618586:618745 [2] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-172-147:618586:618745 [2] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-172-147:618586:618745 [2] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-172-147:618586:618745 [2] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-172-147:618586:618745 [2] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-172-147:618586:618745 [2] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-172-147:618586:618745 [2] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-172-147:618586:618745 [2] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-172-147:618586:618745 [2] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-172-147:618586:618745 [2] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-172-147:618586:618745 [2] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-172-147:618586:618745 [2] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-172-147:618586:618745 [2] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-172-147:618586:618745 [2] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-172-147:618586:618745 [2] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-172-147:618586:618745 [2] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-172-147:618586:618745 [2] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-172-147:618586:618745 [2] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-172-147:618586:618745 [2] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-172-147:618586:618745 [2] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/280
+ip-26-0-172-147:618586:618745 [2] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-172-147:618587:618751 [3] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-172-147:618587:618751 [3] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-172-147:618587:618751 [3] NCCL INFO Using network Libfabric
+ip-26-0-172-252:688205:688365 [5] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-172-252:688204:688367 [4] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-172-252:688204:688367 [4] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-172-252:688205:688365 [5] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-172-252:688204:688367 [4] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-172-252:688205:688365 [5] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-172-252:688204:688367 [4] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-172-252:688205:688365 [5] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-172-252:688204:688367 [4] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-172-252:688205:688365 [5] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-172-252:688204:688367 [4] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-172-252:688205:688365 [5] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-172-252:688204:688367 [4] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-172-252:688205:688365 [5] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-172-252:688204:688367 [4] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-172-252:688205:688365 [5] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-172-252:688204:688367 [4] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-172-252:688205:688365 [5] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-172-252:688204:688367 [4] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-172-252:688205:688365 [5] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-172-252:688204:688367 [4] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-172-252:688205:688365 [5] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-172-252:688204:688367 [4] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-172-252:688205:688365 [5] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-172-252:688204:688367 [4] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-172-252:688205:688365 [5] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-172-252:688204:688367 [4] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-172-252:688205:688365 [5] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-172-252:688204:688367 [4] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-172-252:688205:688365 [5] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-172-252:688204:688367 [4] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-172-252:688205:688365 [5] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-172-252:688204:688367 [4] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-172-252:688205:688365 [5] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-172-252:688204:688367 [4] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-172-252:688205:688365 [5] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-172-252:688204:688367 [4] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-172-252:688205:688365 [5] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-172-252:688204:688367 [4] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-172-252:688205:688365 [5] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-172-252:688204:688367 [4] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-172-252:688205:688365 [5] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-172-252:688204:688367 [4] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-172-252:688205:688365 [5] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-172-252:688204:688367 [4] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-172-252:688205:688365 [5] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-172-252:688204:688367 [4] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-172-252:688205:688365 [5] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-172-252:688204:688367 [4] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-172-73:926599:926755 [5] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/280
+ip-26-0-172-252:688205:688365 [5] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-172-252:688204:688367 [4] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-172-252:688205:688365 [5] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-172-252:688204:688367 [4] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-172-252:688205:688365 [5] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-172-252:688204:688367 [4] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-172-252:688205:688365 [5] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-172-252:688204:688367 [4] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-172-252:688205:688365 [5] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-172-252:688204:688367 [4] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-172-252:688205:688365 [5] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-172-73:926599:926755 [5] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-172-252:688204:688367 [4] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-172-252:688205:688365 [5] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-172-252:688204:688367 [4] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-172-252:688205:688365 [5] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-172-73:926595:926760 [1] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/280
+ip-26-0-172-73:926595:926760 [1] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-172-73:926601:926759 [7] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/280
+ip-26-0-172-73:926601:926759 [7] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-172-252:688205:688365 [5] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/280
+ip-26-0-172-252:688205:688365 [5] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-172-252:688204:688367 [4] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/280
+ip-26-0-172-252:688204:688367 [4] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-172-252:688206:688363 [6] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-172-252:688206:688363 [6] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-172-252:688206:688363 [6] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-172-252:688206:688363 [6] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-172-252:688206:688363 [6] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-172-252:688206:688363 [6] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-172-252:688206:688363 [6] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-172-252:688206:688363 [6] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-172-252:688206:688363 [6] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-172-252:688206:688363 [6] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-172-252:688206:688363 [6] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-172-252:688206:688363 [6] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-172-252:688206:688363 [6] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-172-252:688206:688363 [6] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-172-252:688206:688363 [6] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-172-252:688206:688363 [6] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-172-252:688206:688363 [6] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-172-252:688206:688363 [6] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-172-252:688206:688363 [6] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-172-252:688206:688363 [6] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-172-252:688206:688363 [6] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-172-252:688206:688363 [6] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-172-252:688206:688363 [6] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-172-252:688206:688363 [6] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-172-252:688206:688363 [6] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-172-252:688206:688363 [6] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-172-252:688206:688363 [6] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-172-252:688206:688363 [6] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-172-252:688206:688363 [6] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-172-252:688206:688363 [6] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-172-252:688206:688363 [6] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-172-252:688206:688363 [6] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-172-252:688206:688363 [6] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/280
+ip-26-0-172-252:688206:688363 [6] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-172-252:688202:688368 [2] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-172-252:688202:688368 [2] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-172-252:688202:688368 [2] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-172-252:688202:688368 [2] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-172-252:688202:688368 [2] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-172-252:688202:688368 [2] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-172-252:688202:688368 [2] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-172-252:688202:688368 [2] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-172-252:688202:688368 [2] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-172-252:688202:688368 [2] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-172-252:688202:688368 [2] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-172-252:688202:688368 [2] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-172-252:688202:688368 [2] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-172-252:688202:688368 [2] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-172-252:688202:688368 [2] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-172-252:688202:688368 [2] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-172-252:688202:688368 [2] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-172-252:688202:688368 [2] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-172-252:688202:688368 [2] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-172-252:688202:688368 [2] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-172-252:688202:688368 [2] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-172-252:688202:688368 [2] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-172-252:688202:688368 [2] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-172-252:688202:688368 [2] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-172-252:688202:688368 [2] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-172-252:688202:688368 [2] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-172-252:688202:688368 [2] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-172-252:688202:688368 [2] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-172-252:688202:688368 [2] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-172-252:688202:688368 [2] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-172-252:688202:688368 [2] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-172-252:688202:688368 [2] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-172-252:688202:688368 [2] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/280
+ip-26-0-172-252:688202:688368 [2] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-172-252:688200:688369 [0] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-172-252:688200:688369 [0] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-172-252:688200:688369 [0] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-172-252:688200:688369 [0] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-172-252:688200:688369 [0] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-172-252:688200:688369 [0] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-172-252:688200:688369 [0] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-172-252:688200:688369 [0] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-172-252:688200:688369 [0] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-172-252:688200:688369 [0] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-172-252:688200:688369 [0] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-172-252:688200:688369 [0] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-172-252:688200:688369 [0] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-172-252:688200:688369 [0] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-172-252:688200:688369 [0] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-172-252:688200:688369 [0] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-172-252:688200:688369 [0] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-172-252:688200:688369 [0] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-172-252:688200:688369 [0] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-172-252:688203:688362 [3] NCCL INFO NET/OFI NIC group 0 device #0 0000:52:00.0
+ip-26-0-172-252:688200:688369 [0] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-172-252:688200:688369 [0] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-172-252:688203:688362 [3] NCCL INFO NET/OFI NIC group 0 device #1 0000:51:00.0
+ip-26-0-172-252:688200:688369 [0] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-172-252:688203:688362 [3] NCCL INFO NET/OFI NIC group 0 device #2 0000:50:00.0
+ip-26-0-172-252:688200:688369 [0] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-172-252:688203:688362 [3] NCCL INFO NET/OFI NIC group 0 device #3 0000:4f:00.0
+ip-26-0-172-252:688200:688369 [0] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-172-252:688203:688362 [3] NCCL INFO NET/OFI NIC group 1 device #0 0000:63:00.0
+ip-26-0-172-252:688200:688369 [0] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-172-252:688203:688362 [3] NCCL INFO NET/OFI NIC group 1 device #1 0000:62:00.0
+ip-26-0-172-252:688200:688369 [0] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-172-252:688203:688362 [3] NCCL INFO NET/OFI NIC group 1 device #2 0000:61:00.0
+ip-26-0-172-252:688200:688369 [0] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-172-252:688203:688362 [3] NCCL INFO NET/OFI NIC group 1 device #3 0000:60:00.0
+ip-26-0-172-252:688200:688369 [0] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-172-252:688203:688362 [3] NCCL INFO NET/OFI NIC group 2 device #0 0000:74:00.0
+ip-26-0-172-252:688200:688369 [0] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-172-252:688203:688362 [3] NCCL INFO NET/OFI NIC group 2 device #1 0000:73:00.0
+ip-26-0-172-252:688200:688369 [0] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-172-252:688203:688362 [3] NCCL INFO NET/OFI NIC group 2 device #2 0000:72:00.0
+ip-26-0-172-252:688200:688369 [0] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-172-252:688203:688362 [3] NCCL INFO NET/OFI NIC group 2 device #3 0000:71:00.0
+ip-26-0-172-252:688200:688369 [0] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-172-252:688203:688362 [3] NCCL INFO NET/OFI NIC group 3 device #0 0000:85:00.0
+ip-26-0-172-252:688203:688362 [3] NCCL INFO NET/OFI NIC group 3 device #1 0000:84:00.0
+ip-26-0-172-252:688203:688362 [3] NCCL INFO NET/OFI NIC group 3 device #2 0000:83:00.0
+ip-26-0-172-252:688203:688362 [3] NCCL INFO NET/OFI NIC group 3 device #3 0000:82:00.0
+ip-26-0-172-252:688203:688362 [3] NCCL INFO NET/OFI NIC group 4 device #0 0000:96:00.0
+ip-26-0-172-252:688203:688362 [3] NCCL INFO NET/OFI NIC group 4 device #1 0000:95:00.0
+ip-26-0-172-252:688203:688362 [3] NCCL INFO NET/OFI NIC group 4 device #2 0000:94:00.0
+ip-26-0-172-252:688203:688362 [3] NCCL INFO NET/OFI NIC group 4 device #3 0000:93:00.0
+ip-26-0-172-252:688203:688362 [3] NCCL INFO NET/OFI NIC group 5 device #0 0000:a7:00.0
+ip-26-0-172-252:688203:688362 [3] NCCL INFO NET/OFI NIC group 5 device #1 0000:a6:00.0
+ip-26-0-172-252:688203:688362 [3] NCCL INFO NET/OFI NIC group 5 device #2 0000:a5:00.0
+ip-26-0-172-252:688203:688362 [3] NCCL INFO NET/OFI NIC group 5 device #3 0000:a4:00.0
+ip-26-0-172-252:688203:688362 [3] NCCL INFO NET/OFI NIC group 6 device #0 0000:b8:00.0
+ip-26-0-172-252:688203:688362 [3] NCCL INFO NET/OFI NIC group 6 device #1 0000:b7:00.0
+ip-26-0-172-252:688203:688362 [3] NCCL INFO NET/OFI NIC group 6 device #2 0000:b6:00.0
+ip-26-0-172-252:688203:688362 [3] NCCL INFO NET/OFI NIC group 6 device #3 0000:b5:00.0
+ip-26-0-172-252:688203:688362 [3] NCCL INFO NET/OFI NIC group 7 device #0 0000:c9:00.0
+ip-26-0-172-252:688203:688362 [3] NCCL INFO NET/OFI NIC group 7 device #1 0000:c8:00.0
+ip-26-0-172-252:688203:688362 [3] NCCL INFO NET/OFI NIC group 7 device #2 0000:c7:00.0
+ip-26-0-172-252:688203:688362 [3] NCCL INFO NET/OFI NIC group 7 device #3 0000:c6:00.0
+ip-26-0-172-252:688200:688369 [0] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/280
+ip-26-0-172-252:688200:688369 [0] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-172-252:688203:688362 [3] NCCL INFO NET/OFI Setting NCCL_TOPO_FILE environment variable to /proc/self/fd/280
+ip-26-0-172-252:688203:688362 [3] NCCL INFO NET/OFI Creating one domain per process
+ip-26-0-172-147:618589:618744 [5] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-172-147:618589:618744 [5] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-172-147:618589:618744 [5] NCCL INFO Using network Libfabric
+ip-26-0-172-57:942615:942775 [4] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-172-57:942615:942775 [4] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-172-57:942615:942775 [4] NCCL INFO Using network Libfabric
+ip-26-0-172-57:942612:942779 [1] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-172-57:942612:942779 [1] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-172-57:942612:942779 [1] NCCL INFO Using network Libfabric
+ip-26-0-173-7:811255:811411 [7] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-173-7:811255:811411 [7] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-173-7:811255:811411 [7] NCCL INFO Using network Libfabric
+ip-26-0-172-252:688207:688366 [7] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-172-252:688207:688366 [7] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-172-252:688207:688366 [7] NCCL INFO Using network Libfabric
+ip-26-0-172-57:942614:942773 [3] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-172-57:942614:942773 [3] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-172-57:942614:942773 [3] NCCL INFO Using network Libfabric
+ip-26-0-172-57:942617:942774 [6] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-172-57:942617:942774 [6] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-172-57:942617:942774 [6] NCCL INFO Using network Libfabric
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO Using network Libfabric
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO Using network Libfabric
+ip-26-0-172-57:942616:942777 [5] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-172-57:942616:942777 [5] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO Using network Libfabric
+ip-26-0-172-57:942616:942777 [5] NCCL INFO Using network Libfabric
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO Using network Libfabric
+ip-26-0-172-57:942618:942778 [7] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-172-57:942618:942778 [7] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-172-57:942618:942778 [7] NCCL INFO Using network Libfabric
+ip-26-0-172-57:942613:942776 [2] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-172-57:942613:942776 [2] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-172-57:942613:942776 [2] NCCL INFO Using network Libfabric
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO Using network Libfabric
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO Using network Libfabric
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO Using network Libfabric
+ip-26-0-160-225:94345:94509 [3] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-160-225:94345:94509 [3] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-160-225:94345:94509 [3] NCCL INFO Using network Libfabric
+ip-26-0-172-252:688201:688364 [1] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-172-252:688201:688364 [1] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-172-252:688201:688364 [1] NCCL INFO Using network Libfabric
+ip-26-0-160-225:94349:94505 [7] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-160-225:94349:94505 [7] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-160-225:94349:94505 [7] NCCL INFO Using network Libfabric
+ip-26-0-160-225:94348:94506 [6] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-160-225:94348:94506 [6] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-160-225:94348:94506 [6] NCCL INFO Using network Libfabric
+ip-26-0-160-225:94343:94508 [1] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-160-225:94343:94508 [1] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-160-225:94343:94508 [1] NCCL INFO Using network Libfabric
+ip-26-0-160-225:94346:94511 [4] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-160-225:94346:94511 [4] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-160-225:94346:94511 [4] NCCL INFO Using network Libfabric
+ip-26-0-160-225:94344:94510 [2] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-160-225:94344:94510 [2] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-160-225:94344:94510 [2] NCCL INFO Using network Libfabric
+ip-26-0-172-73:926594:926762 [0] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-172-73:926594:926762 [0] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-172-73:926594:926762 [0] NCCL INFO Using network Libfabric
+ip-26-0-172-73:926596:926756 [2] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-172-73:926596:926756 [2] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-172-73:926596:926756 [2] NCCL INFO Using network Libfabric
+ip-26-0-172-142:880680:880843 [1] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-172-142:880680:880843 [1] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-172-142:880680:880843 [1] NCCL INFO Using network Libfabric
+ip-26-0-172-142:880682:880840 [3] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-172-142:880682:880840 [3] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-172-142:880682:880840 [3] NCCL INFO Using network Libfabric
+ip-26-0-172-73:926598:926761 [4] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-172-73:926598:926761 [4] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-172-73:926598:926761 [4] NCCL INFO Using network Libfabric
+ip-26-0-172-73:926595:926760 [1] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-172-73:926595:926760 [1] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-172-73:926595:926760 [1] NCCL INFO Using network Libfabric
+ip-26-0-172-73:926600:926757 [6] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-172-73:926600:926757 [6] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-172-73:926600:926757 [6] NCCL INFO Using network Libfabric
+ip-26-0-172-73:926601:926759 [7] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-172-73:926601:926759 [7] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-172-73:926601:926759 [7] NCCL INFO Using network Libfabric
+ip-26-0-172-73:926599:926755 [5] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-172-73:926599:926755 [5] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-172-73:926599:926755 [5] NCCL INFO Using network Libfabric
+ip-26-0-172-142:880681:880845 [2] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-172-142:880681:880845 [2] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-172-142:880681:880845 [2] NCCL INFO Using network Libfabric
+ip-26-0-172-142:880679:880846 [0] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-172-142:880679:880846 [0] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-172-142:880679:880846 [0] NCCL INFO Using network Libfabric
+ip-26-0-172-142:880685:880847 [6] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-172-142:880685:880847 [6] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-172-142:880685:880847 [6] NCCL INFO Using network Libfabric
+ip-26-0-172-142:880686:880841 [7] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-172-142:880686:880841 [7] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-172-142:880686:880841 [7] NCCL INFO Using network Libfabric
+ip-26-0-172-147:618591:618746 [7] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-172-147:618591:618746 [7] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-172-147:618591:618746 [7] NCCL INFO Using network Libfabric
+ip-26-0-173-7:811254:811414 [6] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-173-7:811254:811414 [6] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-173-7:811254:811414 [6] NCCL INFO Using network Libfabric
+ip-26-0-172-252:688204:688367 [4] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-172-252:688204:688367 [4] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-172-252:688204:688367 [4] NCCL INFO Using network Libfabric
+ip-26-0-173-7:811250:811412 [2] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-173-7:811250:811412 [2] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-173-7:811250:811412 [2] NCCL INFO Using network Libfabric
+ip-26-0-173-7:811252:811408 [4] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-173-7:811252:811408 [4] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-173-7:811252:811408 [4] NCCL INFO Using network Libfabric
+ip-26-0-173-7:811253:811410 [5] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-173-7:811253:811410 [5] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-173-7:811253:811410 [5] NCCL INFO Using network Libfabric
+ip-26-0-172-147:618588:618750 [4] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-172-147:618588:618750 [4] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-172-147:618588:618750 [4] NCCL INFO Using network Libfabric
+ip-26-0-173-7:811249:811413 [1] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-173-7:811249:811413 [1] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-173-7:811249:811413 [1] NCCL INFO Using network Libfabric
+ip-26-0-173-7:811248:811415 [0] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-173-7:811248:811415 [0] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-173-7:811248:811415 [0] NCCL INFO Using network Libfabric
+ip-26-0-172-147:618590:618749 [6] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-172-147:618590:618749 [6] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-172-147:618590:618749 [6] NCCL INFO Using network Libfabric
+ip-26-0-172-147:618584:618747 [0] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-172-147:618584:618747 [0] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-172-147:618584:618747 [0] NCCL INFO Using network Libfabric
+ip-26-0-172-252:688205:688365 [5] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-172-252:688205:688365 [5] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-172-252:688205:688365 [5] NCCL INFO Using network Libfabric
+ip-26-0-172-147:618585:618748 [1] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-172-147:618585:618748 [1] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-172-147:618585:618748 [1] NCCL INFO Using network Libfabric
+ip-26-0-172-147:618586:618745 [2] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-172-147:618586:618745 [2] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-172-147:618586:618745 [2] NCCL INFO Using network Libfabric
+ip-26-0-172-252:688206:688363 [6] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-172-252:688206:688363 [6] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-172-252:688206:688363 [6] NCCL INFO Using network Libfabric
+ip-26-0-172-252:688200:688369 [0] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-172-252:688200:688369 [0] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-172-252:688200:688369 [0] NCCL INFO Using network Libfabric
+ip-26-0-172-252:688203:688362 [3] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-172-252:688203:688362 [3] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-172-252:688203:688362 [3] NCCL INFO Using network Libfabric
+ip-26-0-172-252:688202:688368 [2] NCCL INFO NET/OFI Support for global registrations: true
+ip-26-0-172-252:688202:688368 [2] NCCL INFO NET/OFI Support for DMA-BUF registrations: false
+ip-26-0-172-252:688202:688368 [2] NCCL INFO Using network Libfabric
+ip-26-0-160-225:94344:94510 [2] NCCL INFO comm 0x9047390 rank 2 nranks 64 cudaDev 2 nvmlDev 2 busId 75000 commId 0x3f4d485835722cf1 - Init START
+ip-26-0-173-7:811252:811408 [4] NCCL INFO comm 0x8c2ebd0 rank 60 nranks 64 cudaDev 4 nvmlDev 4 busId 97000 commId 0x3f4d485835722cf1 - Init START
+ip-26-0-173-7:811251:811409 [3] NCCL INFO comm 0x8b1a600 rank 59 nranks 64 cudaDev 3 nvmlDev 3 busId 86000 commId 0x3f4d485835722cf1 - Init START
+ip-26-0-173-7:811254:811414 [6] NCCL INFO comm 0x89213d0 rank 62 nranks 64 cudaDev 6 nvmlDev 6 busId b9000 commId 0x3f4d485835722cf1 - Init START
+ip-26-0-160-225:94342:94504 [0] NCCL INFO comm 0x8eae1b0 rank 0 nranks 64 cudaDev 0 nvmlDev 0 busId 53000 commId 0x3f4d485835722cf1 - Init START
+ip-26-0-160-225:94349:94505 [7] NCCL INFO comm 0x89eb7d0 rank 7 nranks 64 cudaDev 7 nvmlDev 7 busId ca000 commId 0x3f4d485835722cf1 - Init START
+ip-26-0-160-225:94347:94507 [5] NCCL INFO comm 0x90b4a60 rank 5 nranks 64 cudaDev 5 nvmlDev 5 busId a8000 commId 0x3f4d485835722cf1 - Init START
+ip-26-0-173-7:811255:811411 [7] NCCL INFO comm 0x9bdb100 rank 63 nranks 64 cudaDev 7 nvmlDev 7 busId ca000 commId 0x3f4d485835722cf1 - Init START
+ip-26-0-173-7:811253:811410 [5] NCCL INFO comm 0x95fcf60 rank 61 nranks 64 cudaDev 5 nvmlDev 5 busId a8000 commId 0x3f4d485835722cf1 - Init START
+ip-26-0-173-7:811250:811412 [2] NCCL INFO comm 0x958ce00 rank 58 nranks 64 cudaDev 2 nvmlDev 2 busId 75000 commId 0x3f4d485835722cf1 - Init START
+ip-26-0-160-225:94346:94511 [4] NCCL INFO comm 0xa00d090 rank 4 nranks 64 cudaDev 4 nvmlDev 4 busId 97000 commId 0x3f4d485835722cf1 - Init START
+ip-26-0-160-225:94343:94508 [1] NCCL INFO comm 0x82f6cb0 rank 1 nranks 64 cudaDev 1 nvmlDev 1 busId 64000 commId 0x3f4d485835722cf1 - Init START
+ip-26-0-160-225:94345:94509 [3] NCCL INFO comm 0xa1ae2d0 rank 3 nranks 64 cudaDev 3 nvmlDev 3 busId 86000 commId 0x3f4d485835722cf1 - Init START
+ip-26-0-160-225:94348:94506 [6] NCCL INFO comm 0x97f3050 rank 6 nranks 64 cudaDev 6 nvmlDev 6 busId b9000 commId 0x3f4d485835722cf1 - Init START
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO comm 0x84277a0 rank 9 nranks 64 cudaDev 1 nvmlDev 1 busId 64000 commId 0x3f4d485835722cf1 - Init START
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO comm 0x89c15f0 rank 8 nranks 64 cudaDev 0 nvmlDev 0 busId 53000 commId 0x3f4d485835722cf1 - Init START
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO comm 0xa1a18a0 rank 11 nranks 64 cudaDev 3 nvmlDev 3 busId 86000 commId 0x3f4d485835722cf1 - Init START
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO comm 0x9e252e0 rank 10 nranks 64 cudaDev 2 nvmlDev 2 busId 75000 commId 0x3f4d485835722cf1 - Init START
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO comm 0xa199350 rank 12 nranks 64 cudaDev 4 nvmlDev 4 busId 97000 commId 0x3f4d485835722cf1 - Init START
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO comm 0xa1c3a50 rank 13 nranks 64 cudaDev 5 nvmlDev 5 busId a8000 commId 0x3f4d485835722cf1 - Init START
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO comm 0x83bb440 rank 14 nranks 64 cudaDev 6 nvmlDev 6 busId b9000 commId 0x3f4d485835722cf1 - Init START
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO comm 0x97f7620 rank 15 nranks 64 cudaDev 7 nvmlDev 7 busId ca000 commId 0x3f4d485835722cf1 - Init START
+ip-26-0-172-142:880679:880846 [0] NCCL INFO comm 0x9ed6ec0 rank 16 nranks 64 cudaDev 0 nvmlDev 0 busId 53000 commId 0x3f4d485835722cf1 - Init START
+ip-26-0-172-142:880680:880843 [1] NCCL INFO comm 0x8da1b60 rank 17 nranks 64 cudaDev 1 nvmlDev 1 busId 64000 commId 0x3f4d485835722cf1 - Init START
+ip-26-0-172-142:880682:880840 [3] NCCL INFO comm 0x928d640 rank 19 nranks 64 cudaDev 3 nvmlDev 3 busId 86000 commId 0x3f4d485835722cf1 - Init START
+ip-26-0-172-142:880681:880845 [2] NCCL INFO comm 0x8e71310 rank 18 nranks 64 cudaDev 2 nvmlDev 2 busId 75000 commId 0x3f4d485835722cf1 - Init START
+ip-26-0-172-142:880683:880844 [4] NCCL INFO comm 0x9a8c230 rank 20 nranks 64 cudaDev 4 nvmlDev 4 busId 97000 commId 0x3f4d485835722cf1 - Init START
+ip-26-0-172-142:880684:880842 [5] NCCL INFO comm 0x88ad900 rank 21 nranks 64 cudaDev 5 nvmlDev 5 busId a8000 commId 0x3f4d485835722cf1 - Init START
+ip-26-0-172-142:880686:880841 [7] NCCL INFO comm 0x94e5950 rank 23 nranks 64 cudaDev 7 nvmlDev 7 busId ca000 commId 0x3f4d485835722cf1 - Init START
+ip-26-0-172-142:880685:880847 [6] NCCL INFO comm 0x876ac20 rank 22 nranks 64 cudaDev 6 nvmlDev 6 busId b9000 commId 0x3f4d485835722cf1 - Init START
+ip-26-0-172-147:618584:618747 [0] NCCL INFO comm 0x9905f70 rank 24 nranks 64 cudaDev 0 nvmlDev 0 busId 53000 commId 0x3f4d485835722cf1 - Init START
+ip-26-0-172-147:618585:618748 [1] NCCL INFO comm 0x9e17fb0 rank 25 nranks 64 cudaDev 1 nvmlDev 1 busId 64000 commId 0x3f4d485835722cf1 - Init START
+ip-26-0-172-147:618586:618745 [2] NCCL INFO comm 0x89463c0 rank 26 nranks 64 cudaDev 2 nvmlDev 2 busId 75000 commId 0x3f4d485835722cf1 - Init START
+ip-26-0-172-147:618588:618750 [4] NCCL INFO comm 0x8654810 rank 28 nranks 64 cudaDev 4 nvmlDev 4 busId 97000 commId 0x3f4d485835722cf1 - Init START
+ip-26-0-172-147:618589:618744 [5] NCCL INFO comm 0x962f830 rank 29 nranks 64 cudaDev 5 nvmlDev 5 busId a8000 commId 0x3f4d485835722cf1 - Init START
+ip-26-0-172-147:618590:618749 [6] NCCL INFO comm 0x9f398d0 rank 30 nranks 64 cudaDev 6 nvmlDev 6 busId b9000 commId 0x3f4d485835722cf1 - Init START
+ip-26-0-172-147:618587:618751 [3] NCCL INFO comm 0x8914d00 rank 27 nranks 64 cudaDev 3 nvmlDev 3 busId 86000 commId 0x3f4d485835722cf1 - Init START
+ip-26-0-172-147:618591:618746 [7] NCCL INFO comm 0x91d1750 rank 31 nranks 64 cudaDev 7 nvmlDev 7 busId ca000 commId 0x3f4d485835722cf1 - Init START
+ip-26-0-172-252:688200:688369 [0] NCCL INFO comm 0x920bad0 rank 32 nranks 64 cudaDev 0 nvmlDev 0 busId 53000 commId 0x3f4d485835722cf1 - Init START
+ip-26-0-172-252:688203:688362 [3] NCCL INFO comm 0x8600940 rank 35 nranks 64 cudaDev 3 nvmlDev 3 busId 86000 commId 0x3f4d485835722cf1 - Init START
+ip-26-0-172-252:688202:688368 [2] NCCL INFO comm 0x900a490 rank 34 nranks 64 cudaDev 2 nvmlDev 2 busId 75000 commId 0x3f4d485835722cf1 - Init START
+ip-26-0-172-252:688204:688367 [4] NCCL INFO comm 0x9286a60 rank 36 nranks 64 cudaDev 4 nvmlDev 4 busId 97000 commId 0x3f4d485835722cf1 - Init START
+ip-26-0-172-252:688205:688365 [5] NCCL INFO comm 0x8c6c940 rank 37 nranks 64 cudaDev 5 nvmlDev 5 busId a8000 commId 0x3f4d485835722cf1 - Init START
+ip-26-0-172-252:688201:688364 [1] NCCL INFO comm 0x91aa790 rank 33 nranks 64 cudaDev 1 nvmlDev 1 busId 64000 commId 0x3f4d485835722cf1 - Init START
+ip-26-0-172-252:688206:688363 [6] NCCL INFO comm 0xa251280 rank 38 nranks 64 cudaDev 6 nvmlDev 6 busId b9000 commId 0x3f4d485835722cf1 - Init START
+ip-26-0-172-252:688207:688366 [7] NCCL INFO comm 0x8cf40a0 rank 39 nranks 64 cudaDev 7 nvmlDev 7 busId ca000 commId 0x3f4d485835722cf1 - Init START
+ip-26-0-172-57:942611:942772 [0] NCCL INFO comm 0x9b57e50 rank 40 nranks 64 cudaDev 0 nvmlDev 0 busId 53000 commId 0x3f4d485835722cf1 - Init START
+ip-26-0-172-57:942615:942775 [4] NCCL INFO comm 0x9ebb790 rank 44 nranks 64 cudaDev 4 nvmlDev 4 busId 97000 commId 0x3f4d485835722cf1 - Init START
+ip-26-0-172-57:942617:942774 [6] NCCL INFO comm 0x959be20 rank 46 nranks 64 cudaDev 6 nvmlDev 6 busId b9000 commId 0x3f4d485835722cf1 - Init START
+ip-26-0-172-57:942613:942776 [2] NCCL INFO comm 0x995bf70 rank 42 nranks 64 cudaDev 2 nvmlDev 2 busId 75000 commId 0x3f4d485835722cf1 - Init START
+ip-26-0-172-57:942618:942778 [7] NCCL INFO comm 0x8e28120 rank 47 nranks 64 cudaDev 7 nvmlDev 7 busId ca000 commId 0x3f4d485835722cf1 - Init START
+ip-26-0-172-73:926594:926762 [0] NCCL INFO comm 0x920c450 rank 48 nranks 64 cudaDev 0 nvmlDev 0 busId 53000 commId 0x3f4d485835722cf1 - Init START
+ip-26-0-173-7:811248:811415 [0] NCCL INFO comm 0xa104f70 rank 56 nranks 64 cudaDev 0 nvmlDev 0 busId 53000 commId 0x3f4d485835722cf1 - Init START
+ip-26-0-173-7:811249:811413 [1] NCCL INFO comm 0x8794560 rank 57 nranks 64 cudaDev 1 nvmlDev 1 busId 64000 commId 0x3f4d485835722cf1 - Init START
+ip-26-0-172-57:942612:942779 [1] NCCL INFO comm 0x891af90 rank 41 nranks 64 cudaDev 1 nvmlDev 1 busId 64000 commId 0x3f4d485835722cf1 - Init START
+ip-26-0-172-57:942614:942773 [3] NCCL INFO comm 0x89687c0 rank 43 nranks 64 cudaDev 3 nvmlDev 3 busId 86000 commId 0x3f4d485835722cf1 - Init START
+ip-26-0-172-57:942616:942777 [5] NCCL INFO comm 0x84435f0 rank 45 nranks 64 cudaDev 5 nvmlDev 5 busId a8000 commId 0x3f4d485835722cf1 - Init START
+ip-26-0-172-73:926595:926760 [1] NCCL INFO comm 0x82893c0 rank 49 nranks 64 cudaDev 1 nvmlDev 1 busId 64000 commId 0x3f4d485835722cf1 - Init START
+ip-26-0-172-73:926601:926759 [7] NCCL INFO comm 0x8367300 rank 55 nranks 64 cudaDev 7 nvmlDev 7 busId ca000 commId 0x3f4d485835722cf1 - Init START
+ip-26-0-172-73:926600:926757 [6] NCCL INFO comm 0x9e0a1c0 rank 54 nranks 64 cudaDev 6 nvmlDev 6 busId b9000 commId 0x3f4d485835722cf1 - Init START
+ip-26-0-172-73:926599:926755 [5] NCCL INFO comm 0x9df4700 rank 53 nranks 64 cudaDev 5 nvmlDev 5 busId a8000 commId 0x3f4d485835722cf1 - Init START
+ip-26-0-172-73:926598:926761 [4] NCCL INFO comm 0x96d54b0 rank 52 nranks 64 cudaDev 4 nvmlDev 4 busId 97000 commId 0x3f4d485835722cf1 - Init START
+ip-26-0-172-73:926596:926756 [2] NCCL INFO comm 0x8bdc2a0 rank 50 nranks 64 cudaDev 2 nvmlDev 2 busId 75000 commId 0x3f4d485835722cf1 - Init START
+ip-26-0-172-73:926597:926758 [3] NCCL INFO comm 0x8ae3590 rank 51 nranks 64 cudaDev 3 nvmlDev 3 busId 86000 commId 0x3f4d485835722cf1 - Init START
+ip-26-0-172-147:618584:618747 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-147:618589:618744 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-147:618586:618745 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-142:880684:880842 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-142:880686:880841 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-142:880685:880847 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-147:618590:618749 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-147:618591:618746 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-57:942618:942778 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-57:942617:942774 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-57:942616:942777 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-57:942615:942775 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-147:618588:618750 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-147:618585:618748 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-147:618587:618751 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-57:942614:942773 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-142:880682:880840 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-142:880683:880844 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-142:880681:880845 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-142:880679:880846 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-57:942613:942776 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-142:880680:880843 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-173-7:811249:811413 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-173-7:811250:811412 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-173-7:811251:811409 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-173-7:811252:811408 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-173-7:811253:811410 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-160-225:94349:94505 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-160-225:94346:94511 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-160-225:94348:94506 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-160-225:94342:94504 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/349
+ip-26-0-160-225:94347:94507 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-160-225:94344:94510 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-160-225:94343:94508 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-173-7:811254:811414 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-160-225:94345:94509 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-173-7:811255:811411 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-252:688200:688369 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-57:942612:942779 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-73:926594:926762 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-73:926595:926760 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-57:942611:942772 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-252:688207:688366 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-73:926596:926756 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-73:926597:926758 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-173-7:811248:811415 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-73:926598:926761 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-73:926599:926755 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-73:926600:926757 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-73:926601:926759 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-252:688201:688364 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-252:688206:688363 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-252:688202:688368 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-252:688203:688362 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-252:688205:688365 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-252:688204:688367 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-172-142:880684:880842 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-172-142:880684:880842 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-160-225:94349:94505 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-160-225:94349:94505 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-172-57:942616:942777 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-172-57:942616:942777 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-172-252:688207:688366 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-172-252:688207:688366 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-172-252:688200:688369 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-172-252:688200:688369 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-172-252:688203:688362 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-172-252:688203:688362 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-173-7:811255:811411 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-173-7:811255:811411 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-173-7:811250:811412 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-173-7:811250:811412 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-173-7:811252:811408 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-173-7:811252:811408 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-173-7:811253:811410 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-173-7:811253:811410 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-173-7:811254:811414 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-173-7:811254:811414 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-172-142:880683:880844 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-172-142:880683:880844 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-172-142:880685:880847 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-172-142:880685:880847 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-172-142:880682:880840 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-172-142:880682:880840 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-172-252:688205:688365 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-172-252:688205:688365 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-172-73:926595:926760 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-172-73:926595:926760 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-172-73:926597:926758 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-172-73:926597:926758 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-172-142:880686:880841 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-172-142:880686:880841 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-172-252:688201:688364 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-172-252:688201:688364 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-173-7:811251:811409 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-173-7:811251:811409 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-172-142:880679:880846 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-172-142:880679:880846 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-172-252:688204:688367 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-172-252:688204:688367 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-172-73:926599:926755 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-172-73:926599:926755 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-172-142:880681:880845 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-172-252:688202:688368 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-172-252:688202:688368 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-172-142:880681:880845 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-172-252:688206:688363 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-172-252:688206:688363 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-160-225:94343:94508 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-160-225:94343:94508 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-172-142:880680:880843 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-172-73:926601:926759 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-172-142:880680:880843 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-172-73:926598:926761 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-172-73:926598:926761 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-172-73:926596:926756 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-172-73:926601:926759 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-172-73:926596:926756 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-172-73:926594:926762 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-172-73:926600:926757 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-172-73:926600:926757 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-172-73:926594:926762 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-172-57:942612:942779 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-172-57:942612:942779 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-173-7:811249:811413 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-173-7:811249:811413 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-173-7:811248:811415 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-173-7:811248:811415 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-172-57:942613:942776 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-172-57:942613:942776 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-172-147:618587:618751 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-172-147:618587:618751 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-172-57:942614:942773 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-172-57:942614:942773 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-172-147:618589:618744 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-172-147:618589:618744 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-172-147:618590:618749 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-172-147:618590:618749 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-172-147:618585:618748 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-172-147:618585:618748 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-172-147:618591:618746 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-172-147:618591:618746 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-172-147:618584:618747 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-172-147:618584:618747 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-172-147:618586:618745 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-172-147:618586:618745 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-172-147:618588:618750 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-172-147:618588:618750 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-172-57:942615:942775 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-172-57:942615:942775 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-172-57:942611:942772 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-172-57:942617:942774 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-172-57:942617:942774 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-172-57:942611:942772 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-172-57:942618:942778 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-172-57:942618:942778 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-160-225:94346:94511 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-160-225:94346:94511 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-160-225:94348:94506 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-160-225:94348:94506 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-160-225:94345:94509 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-160-225:94345:94509 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-160-225:94344:94510 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-160-225:94344:94510 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-160-225:94342:94504 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-160-225:94342:94504 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-160-225:94347:94507 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-160-225:94347:94507 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-160-225:94347:94507 [5] NCCL INFO NVLS Head  0:  0  8 16 24 32 40 48 56
+ip-26-0-160-225:94347:94507 [5] NCCL INFO NVLS Head  1:  1  9 17 25 33 41 49 57
+ip-26-0-160-225:94347:94507 [5] NCCL INFO NVLS Head  2:  2 10 18 26 34 42 50 58
+ip-26-0-160-225:94347:94507 [5] NCCL INFO NVLS Head  3:  3 11 19 27 35 43 51 59
+ip-26-0-160-225:94347:94507 [5] NCCL INFO NVLS Head  4:  4 12 20 28 36 44 52 60
+ip-26-0-160-225:94347:94507 [5] NCCL INFO NVLS Head  5:  5 13 21 29 37 45 53 61
+ip-26-0-160-225:94347:94507 [5] NCCL INFO NVLS Head  6:  6 14 22 30 38 46 54 62
+ip-26-0-160-225:94347:94507 [5] NCCL INFO NVLS Head  7:  7 15 23 31 39 47 55 63
+ip-26-0-160-225:94347:94507 [5] NCCL INFO Trees [0] 6/-1/-1->5->4 [1] 6/-1/-1->5->4 [2] 6/-1/-1->5->4 [3] 6/-1/-1->5->4 [4] 6/-1/-1->5->4 [5] 6/37/-1->5->-1 [6] -1/-1/-1->5->4 [7] 6/-1/-1->5->4 [8] 6/-1/-1->5->4 [9] 6/-1/-1->5->4 [10] 6/-1/-1->5->4 [11] 6/-1/-1->5->4 [12] 6/-1/-1->5->4 [13] 6/-1/-1->5->13 [14] -1/-1/-1->5->4 [15] 6/-1/-1->5->4
+ip-26-0-160-225:94347:94507 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-225:94348:94506 [6] NCCL INFO NVLS Head  0:  0  8 16 24 32 40 48 56
+ip-26-0-160-225:94348:94506 [6] NCCL INFO NVLS Head  1:  1  9 17 25 33 41 49 57
+ip-26-0-160-225:94348:94506 [6] NCCL INFO NVLS Head  2:  2 10 18 26 34 42 50 58
+ip-26-0-160-225:94348:94506 [6] NCCL INFO NVLS Head  3:  3 11 19 27 35 43 51 59
+ip-26-0-160-225:94348:94506 [6] NCCL INFO NVLS Head  4:  4 12 20 28 36 44 52 60
+ip-26-0-160-225:94348:94506 [6] NCCL INFO NVLS Head  5:  5 13 21 29 37 45 53 61
+ip-26-0-160-225:94348:94506 [6] NCCL INFO NVLS Head  6:  6 14 22 30 38 46 54 62
+ip-26-0-160-225:94348:94506 [6] NCCL INFO NVLS Head  7:  7 15 23 31 39 47 55 63
+ip-26-0-160-225:94348:94506 [6] NCCL INFO Trees [0] 7/-1/-1->6->5 [1] 7/-1/-1->6->5 [2] 7/-1/-1->6->5 [3] 7/-1/-1->6->5 [4] 7/-1/-1->6->5 [5] 7/-1/-1->6->5 [6] 7/38/-1->6->-1 [7] -1/-1/-1->6->5 [8] 7/-1/-1->6->5 [9] 7/-1/-1->6->5 [10] 7/-1/-1->6->5 [11] 7/-1/-1->6->5 [12] 7/-1/-1->6->5 [13] 7/-1/-1->6->5 [14] 7/-1/-1->6->14 [15] -1/-1/-1->6->5
+ip-26-0-160-225:94348:94506 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-225:94349:94505 [7] NCCL INFO NVLS Head  0:  0  8 16 24 32 40 48 56
+ip-26-0-160-225:94349:94505 [7] NCCL INFO NVLS Head  1:  1  9 17 25 33 41 49 57
+ip-26-0-160-225:94349:94505 [7] NCCL INFO NVLS Head  2:  2 10 18 26 34 42 50 58
+ip-26-0-160-225:94349:94505 [7] NCCL INFO NVLS Head  3:  3 11 19 27 35 43 51 59
+ip-26-0-160-225:94349:94505 [7] NCCL INFO NVLS Head  4:  4 12 20 28 36 44 52 60
+ip-26-0-160-225:94349:94505 [7] NCCL INFO NVLS Head  5:  5 13 21 29 37 45 53 61
+ip-26-0-160-225:94349:94505 [7] NCCL INFO NVLS Head  6:  6 14 22 30 38 46 54 62
+ip-26-0-160-225:94349:94505 [7] NCCL INFO NVLS Head  7:  7 15 23 31 39 47 55 63
+ip-26-0-160-225:94349:94505 [7] NCCL INFO Trees [0] -1/-1/-1->7->6 [1] 0/-1/-1->7->6 [2] 0/-1/-1->7->6 [3] 0/-1/-1->7->6 [4] 0/-1/-1->7->6 [5] 0/-1/-1->7->6 [6] 0/-1/-1->7->6 [7] 0/39/-1->7->-1 [8] -1/-1/-1->7->6 [9] 0/-1/-1->7->6 [10] 0/-1/-1->7->6 [11] 0/-1/-1->7->6 [12] 0/-1/-1->7->6 [13] 0/-1/-1->7->6 [14] 0/-1/-1->7->6 [15] 0/-1/-1->7->15
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO Trees [0] 9/-1/-1->8->16 [1] -1/-1/-1->8->15 [2] 9/-1/-1->8->15 [3] 9/-1/-1->8->15 [4] 9/-1/-1->8->15 [5] 9/-1/-1->8->15 [6] 9/-1/-1->8->15 [7] 9/-1/-1->8->15 [8] 9/16/0->8->24 [9] -1/-1/-1->8->15 [10] 9/-1/-1->8->15 [11] 9/-1/-1->8->15 [12] 9/-1/-1->8->15 [13] 9/-1/-1->8->15 [14] 9/-1/-1->8->15 [15] 9/-1/-1->8->15
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-225:94349:94505 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO Trees [0] 10/-1/-1->9->8 [1] 10/-1/-1->9->17 [2] -1/-1/-1->9->8 [3] 10/-1/-1->9->8 [4] 10/-1/-1->9->8 [5] 10/-1/-1->9->8 [6] 10/-1/-1->9->8 [7] 10/-1/-1->9->8 [8] 10/-1/-1->9->8 [9] 10/17/1->9->25 [10] -1/-1/-1->9->8 [11] 10/-1/-1->9->8 [12] 10/-1/-1->9->8 [13] 10/-1/-1->9->8 [14] 10/-1/-1->9->8 [15] 10/-1/-1->9->8
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO Trees [0] 11/-1/-1->10->9 [1] 11/-1/-1->10->9 [2] 11/-1/-1->10->18 [3] -1/-1/-1->10->9 [4] 11/-1/-1->10->9 [5] 11/-1/-1->10->9 [6] 11/-1/-1->10->9 [7] 11/-1/-1->10->9 [8] 11/-1/-1->10->9 [9] 11/-1/-1->10->9 [10] 11/18/2->10->26 [11] -1/-1/-1->10->9 [12] 11/-1/-1->10->9 [13] 11/-1/-1->10->9 [14] 11/-1/-1->10->9 [15] 11/-1/-1->10->9
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO Trees [0] 13/-1/-1->12->11 [1] 13/-1/-1->12->11 [2] 13/-1/-1->12->11 [3] 13/-1/-1->12->11 [4] 13/-1/-1->12->20 [5] -1/-1/-1->12->11 [6] 13/-1/-1->12->11 [7] 13/-1/-1->12->11 [8] 13/-1/-1->12->11 [9] 13/-1/-1->12->11 [10] 13/-1/-1->12->11 [11] 13/-1/-1->12->11 [12] 13/20/4->12->28 [13] -1/-1/-1->12->11 [14] 13/-1/-1->12->11 [15] 13/-1/-1->12->11
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO Trees [0] 12/-1/-1->11->10 [1] 12/-1/-1->11->10 [2] 12/-1/-1->11->10 [3] 12/-1/-1->11->19 [4] -1/-1/-1->11->10 [5] 12/-1/-1->11->10 [6] 12/-1/-1->11->10 [7] 12/-1/-1->11->10 [8] 12/-1/-1->11->10 [9] 12/-1/-1->11->10 [10] 12/-1/-1->11->10 [11] 12/19/3->11->27 [12] -1/-1/-1->11->10 [13] 12/-1/-1->11->10 [14] 12/-1/-1->11->10 [15] 12/-1/-1->11->10
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO Trees [0] -1/-1/-1->15->14 [1] 8/-1/-1->15->14 [2] 8/-1/-1->15->14 [3] 8/-1/-1->15->14 [4] 8/-1/-1->15->14 [5] 8/-1/-1->15->14 [6] 8/-1/-1->15->14 [7] 8/-1/-1->15->23 [8] -1/-1/-1->15->14 [9] 8/-1/-1->15->14 [10] 8/-1/-1->15->14 [11] 8/-1/-1->15->14 [12] 8/-1/-1->15->14 [13] 8/-1/-1->15->14 [14] 8/-1/-1->15->14 [15] 8/23/7->15->31
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-142:880680:880843 [1] NCCL INFO Trees [0] 18/-1/-1->17->16 [1] 18/9/25->17->33 [2] -1/-1/-1->17->16 [3] 18/-1/-1->17->16 [4] 18/-1/-1->17->16 [5] 18/-1/-1->17->16 [6] 18/-1/-1->17->16 [7] 18/-1/-1->17->16 [8] 18/-1/-1->17->16 [9] 18/-1/-1->17->9 [10] -1/-1/-1->17->16 [11] 18/-1/-1->17->16 [12] 18/-1/-1->17->16 [13] 18/-1/-1->17->16 [14] 18/-1/-1->17->16 [15] 18/-1/-1->17->16
+ip-26-0-172-142:880685:880847 [6] NCCL INFO Trees [0] 23/-1/-1->22->21 [1] 23/-1/-1->22->21 [2] 23/-1/-1->22->21 [3] 23/-1/-1->22->21 [4] 23/-1/-1->22->21 [5] 23/-1/-1->22->21 [6] 23/14/30->22->38 [7] -1/-1/-1->22->21 [8] 23/-1/-1->22->21 [9] 23/-1/-1->22->21 [10] 23/-1/-1->22->21 [11] 23/-1/-1->22->21 [12] 23/-1/-1->22->21 [13] 23/-1/-1->22->21 [14] 23/-1/-1->22->14 [15] -1/-1/-1->22->21
+ip-26-0-172-147:618591:618746 [7] NCCL INFO Trees [0] -1/-1/-1->31->30 [1] 24/-1/-1->31->30 [2] 24/-1/-1->31->30 [3] 24/-1/-1->31->30 [4] 24/-1/-1->31->30 [5] 24/-1/-1->31->30 [6] 24/-1/-1->31->30 [7] 24/-1/-1->31->23 [8] -1/-1/-1->31->30 [9] 24/-1/-1->31->30 [10] 24/-1/-1->31->30 [11] 24/-1/-1->31->30 [12] 24/-1/-1->31->30 [13] 24/-1/-1->31->30 [14] 24/-1/-1->31->30 [15] 24/47/15->31->63
+ip-26-0-172-147:618591:618746 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-225:94342:94504 [0] NCCL INFO NVLS Head  0:  0  8 16 24 32 40 48 56
+ip-26-0-160-225:94345:94509 [3] NCCL INFO NVLS Head  0:  0  8 16 24 32 40 48 56
+ip-26-0-160-225:94346:94511 [4] NCCL INFO NVLS Head  0:  0  8 16 24 32 40 48 56
+ip-26-0-160-225:94342:94504 [0] NCCL INFO NVLS Head  1:  1  9 17 25 33 41 49 57
+ip-26-0-160-225:94345:94509 [3] NCCL INFO NVLS Head  1:  1  9 17 25 33 41 49 57
+ip-26-0-160-225:94346:94511 [4] NCCL INFO NVLS Head  1:  1  9 17 25 33 41 49 57
+ip-26-0-160-225:94342:94504 [0] NCCL INFO NVLS Head  2:  2 10 18 26 34 42 50 58
+ip-26-0-160-225:94345:94509 [3] NCCL INFO NVLS Head  2:  2 10 18 26 34 42 50 58
+ip-26-0-160-225:94346:94511 [4] NCCL INFO NVLS Head  2:  2 10 18 26 34 42 50 58
+ip-26-0-160-225:94342:94504 [0] NCCL INFO NVLS Head  3:  3 11 19 27 35 43 51 59
+ip-26-0-160-225:94345:94509 [3] NCCL INFO NVLS Head  3:  3 11 19 27 35 43 51 59
+ip-26-0-160-225:94346:94511 [4] NCCL INFO NVLS Head  3:  3 11 19 27 35 43 51 59
+ip-26-0-160-225:94342:94504 [0] NCCL INFO NVLS Head  4:  4 12 20 28 36 44 52 60
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO Trees [0] 15/-1/-1->14->13 [1] 15/-1/-1->14->13 [2] 15/-1/-1->14->13 [3] 15/-1/-1->14->13 [4] 15/-1/-1->14->13 [5] 15/-1/-1->14->13 [6] 15/-1/-1->14->22 [7] -1/-1/-1->14->13 [8] 15/-1/-1->14->13 [9] 15/-1/-1->14->13 [10] 15/-1/-1->14->13 [11] 15/-1/-1->14->13 [12] 15/-1/-1->14->13 [13] 15/-1/-1->14->13 [14] 15/22/6->14->30 [15] -1/-1/-1->14->13
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO Trees [0] 14/-1/-1->13->12 [1] 14/-1/-1->13->12 [2] 14/-1/-1->13->12 [3] 14/-1/-1->13->12 [4] 14/-1/-1->13->12 [5] 14/-1/-1->13->21 [6] -1/-1/-1->13->12 [7] 14/-1/-1->13->12 [8] 14/-1/-1->13->12 [9] 14/-1/-1->13->12 [10] 14/-1/-1->13->12 [11] 14/-1/-1->13->12 [12] 14/-1/-1->13->12 [13] 14/21/5->13->29 [14] -1/-1/-1->13->12 [15] 14/-1/-1->13->12
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-252:688200:688369 [0] NCCL INFO Trees [0] 33/16/48->32->0 [1] -1/-1/-1->32->39 [2] 33/-1/-1->32->39 [3] 33/-1/-1->32->39 [4] 33/-1/-1->32->39 [5] 33/-1/-1->32->39 [6] 33/-1/-1->32->39 [7] 33/-1/-1->32->39 [8] 33/-1/-1->32->40 [9] -1/-1/-1->32->39 [10] 33/-1/-1->32->39 [11] 33/-1/-1->32->39 [12] 33/-1/-1->32->39 [13] 33/-1/-1->32->39 [14] 33/-1/-1->32->39 [15] 33/-1/-1->32->39
+ip-26-0-172-252:688200:688369 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-225:94345:94509 [3] NCCL INFO NVLS Head  4:  4 12 20 28 36 44 52 60
+ip-26-0-160-225:94344:94510 [2] NCCL INFO NVLS Head  0:  0  8 16 24 32 40 48 56
+ip-26-0-160-225:94343:94508 [1] NCCL INFO NVLS Head  0:  0  8 16 24 32 40 48 56
+ip-26-0-160-225:94346:94511 [4] NCCL INFO NVLS Head  4:  4 12 20 28 36 44 52 60
+ip-26-0-160-225:94345:94509 [3] NCCL INFO NVLS Head  5:  5 13 21 29 37 45 53 61
+ip-26-0-160-225:94342:94504 [0] NCCL INFO NVLS Head  5:  5 13 21 29 37 45 53 61
+ip-26-0-160-225:94346:94511 [4] NCCL INFO NVLS Head  5:  5 13 21 29 37 45 53 61
+ip-26-0-160-225:94344:94510 [2] NCCL INFO NVLS Head  1:  1  9 17 25 33 41 49 57
+ip-26-0-160-225:94342:94504 [0] NCCL INFO NVLS Head  6:  6 14 22 30 38 46 54 62
+ip-26-0-160-225:94343:94508 [1] NCCL INFO NVLS Head  1:  1  9 17 25 33 41 49 57
+ip-26-0-160-225:94345:94509 [3] NCCL INFO NVLS Head  6:  6 14 22 30 38 46 54 62
+ip-26-0-160-225:94346:94511 [4] NCCL INFO NVLS Head  6:  6 14 22 30 38 46 54 62
+ip-26-0-173-7:811255:811411 [7] NCCL INFO Trees [0] -1/-1/-1->63->62 [1] 56/-1/-1->63->62 [2] 56/-1/-1->63->62 [3] 56/-1/-1->63->62 [4] 56/-1/-1->63->62 [5] 56/-1/-1->63->62 [6] 56/-1/-1->63->62 [7] 56/-1/-1->63->55 [8] -1/-1/-1->63->62 [9] 56/-1/-1->63->62 [10] 56/-1/-1->63->62 [11] 56/-1/-1->63->62 [12] 56/-1/-1->63->62 [13] 56/-1/-1->63->62 [14] 56/-1/-1->63->62 [15] 56/31/-1->63->-1
+ip-26-0-173-7:811255:811411 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-142:880686:880841 [7] NCCL INFO Trees [0] -1/-1/-1->23->22 [1] 16/-1/-1->23->22 [2] 16/-1/-1->23->22 [3] 16/-1/-1->23->22 [4] 16/-1/-1->23->22 [5] 16/-1/-1->23->22 [6] 16/-1/-1->23->22 [7] 16/15/31->23->39 [8] -1/-1/-1->23->22 [9] 16/-1/-1->23->22 [10] 16/-1/-1->23->22 [11] 16/-1/-1->23->22 [12] 16/-1/-1->23->22 [13] 16/-1/-1->23->22 [14] 16/-1/-1->23->22 [15] 16/-1/-1->23->15
+ip-26-0-172-142:880680:880843 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-142:880685:880847 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-142:880686:880841 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-147:618586:618745 [2] NCCL INFO Trees [0] 27/-1/-1->26->25 [1] 27/-1/-1->26->25 [2] 27/-1/-1->26->18 [3] -1/-1/-1->26->25 [4] 27/-1/-1->26->25 [5] 27/-1/-1->26->25 [6] 27/-1/-1->26->25 [7] 27/-1/-1->26->25 [8] 27/-1/-1->26->25 [9] 27/-1/-1->26->25 [10] 27/42/10->26->58 [11] -1/-1/-1->26->25 [12] 27/-1/-1->26->25 [13] 27/-1/-1->26->25 [14] 27/-1/-1->26->25 [15] 27/-1/-1->26->25
+ip-26-0-172-147:618589:618744 [5] NCCL INFO Trees [0] 30/-1/-1->29->28 [1] 30/-1/-1->29->28 [2] 30/-1/-1->29->28 [3] 30/-1/-1->29->28 [4] 30/-1/-1->29->28 [5] 30/-1/-1->29->21 [6] -1/-1/-1->29->28 [7] 30/-1/-1->29->28 [8] 30/-1/-1->29->28 [9] 30/-1/-1->29->28 [10] 30/-1/-1->29->28 [11] 30/-1/-1->29->28 [12] 30/-1/-1->29->28 [13] 30/45/13->29->61 [14] -1/-1/-1->29->28 [15] 30/-1/-1->29->28
+ip-26-0-172-147:618586:618745 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-147:618589:618744 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-252:688201:688364 [1] NCCL INFO Trees [0] 34/-1/-1->33->32 [1] 34/17/49->33->1 [2] -1/-1/-1->33->32 [3] 34/-1/-1->33->32 [4] 34/-1/-1->33->32 [5] 34/-1/-1->33->32 [6] 34/-1/-1->33->32 [7] 34/-1/-1->33->32 [8] 34/-1/-1->33->32 [9] 34/-1/-1->33->41 [10] -1/-1/-1->33->32 [11] 34/-1/-1->33->32 [12] 34/-1/-1->33->32 [13] 34/-1/-1->33->32 [14] 34/-1/-1->33->32 [15] 34/-1/-1->33->32
+ip-26-0-172-252:688201:688364 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-225:94344:94510 [2] NCCL INFO NVLS Head  2:  2 10 18 26 34 42 50 58
+ip-26-0-160-225:94342:94504 [0] NCCL INFO NVLS Head  7:  7 15 23 31 39 47 55 63
+ip-26-0-160-225:94343:94508 [1] NCCL INFO NVLS Head  2:  2 10 18 26 34 42 50 58
+ip-26-0-160-225:94346:94511 [4] NCCL INFO NVLS Head  7:  7 15 23 31 39 47 55 63
+ip-26-0-160-225:94345:94509 [3] NCCL INFO NVLS Head  7:  7 15 23 31 39 47 55 63
+ip-26-0-160-225:94344:94510 [2] NCCL INFO NVLS Head  3:  3 11 19 27 35 43 51 59
+ip-26-0-160-225:94343:94508 [1] NCCL INFO NVLS Head  3:  3 11 19 27 35 43 51 59
+ip-26-0-160-225:94343:94508 [1] NCCL INFO NVLS Head  4:  4 12 20 28 36 44 52 60
+ip-26-0-160-225:94344:94510 [2] NCCL INFO NVLS Head  4:  4 12 20 28 36 44 52 60
+ip-26-0-172-142:880683:880844 [4] NCCL INFO Trees [0] 21/-1/-1->20->19 [1] 21/-1/-1->20->19 [2] 21/-1/-1->20->19 [3] 21/-1/-1->20->19 [4] 21/12/28->20->36 [5] -1/-1/-1->20->19 [6] 21/-1/-1->20->19 [7] 21/-1/-1->20->19 [8] 21/-1/-1->20->19 [9] 21/-1/-1->20->19 [10] 21/-1/-1->20->19 [11] 21/-1/-1->20->19 [12] 21/-1/-1->20->12 [13] -1/-1/-1->20->19 [14] 21/-1/-1->20->19 [15] 21/-1/-1->20->19
+ip-26-0-172-142:880683:880844 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-142:880682:880840 [3] NCCL INFO Trees [0] 20/-1/-1->19->18 [1] 20/-1/-1->19->18 [2] 20/-1/-1->19->18 [3] 20/11/27->19->35 [4] -1/-1/-1->19->18 [5] 20/-1/-1->19->18 [6] 20/-1/-1->19->18 [7] 20/-1/-1->19->18 [8] 20/-1/-1->19->18 [9] 20/-1/-1->19->18 [10] 20/-1/-1->19->18 [11] 20/-1/-1->19->11 [12] -1/-1/-1->19->18 [13] 20/-1/-1->19->18 [14] 20/-1/-1->19->18 [15] 20/-1/-1->19->18
+ip-26-0-172-147:618590:618749 [6] NCCL INFO Trees [0] 31/-1/-1->30->29 [1] 31/-1/-1->30->29 [2] 31/-1/-1->30->29 [3] 31/-1/-1->30->29 [4] 31/-1/-1->30->29 [5] 31/-1/-1->30->29 [6] 31/-1/-1->30->22 [7] -1/-1/-1->30->29 [8] 31/-1/-1->30->29 [9] 31/-1/-1->30->29 [10] 31/-1/-1->30->29 [11] 31/-1/-1->30->29 [12] 31/-1/-1->30->29 [13] 31/-1/-1->30->29 [14] 31/46/14->30->62 [15] -1/-1/-1->30->29
+ip-26-0-172-147:618590:618749 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-225:94342:94504 [0] NCCL INFO Channel 00/16 :    0   1   2   3   4   5   6   7   8   9  10  11  12  13  14  15  16  17  18  19
+ip-26-0-160-225:94343:94508 [1] NCCL INFO NVLS Head  5:  5 13 21 29 37 45 53 61
+ip-26-0-160-225:94344:94510 [2] NCCL INFO NVLS Head  5:  5 13 21 29 37 45 53 61
+ip-26-0-160-225:94343:94508 [1] NCCL INFO NVLS Head  6:  6 14 22 30 38 46 54 62
+ip-26-0-160-225:94344:94510 [2] NCCL INFO NVLS Head  6:  6 14 22 30 38 46 54 62
+ip-26-0-160-225:94342:94504 [0] NCCL INFO Channel 01/16 :    0   7   6   5   4   3   2   9   8  15  14  13  12  11  10  17  16  23  22  21
+ip-26-0-160-225:94343:94508 [1] NCCL INFO NVLS Head  7:  7 15 23 31 39 47 55 63
+ip-26-0-160-225:94344:94510 [2] NCCL INFO NVLS Head  7:  7 15 23 31 39 47 55 63
+ip-26-0-160-225:94342:94504 [0] NCCL INFO Channel 02/16 :    0   7   6   5   4   3  10   9   8  15  14  13  12  11  18  17  16  23  22  21
+ip-26-0-173-7:811248:811415 [0] NCCL INFO Trees [0] 57/-1/-1->56->48 [1] -1/-1/-1->56->63 [2] 57/-1/-1->56->63 [3] 57/-1/-1->56->63 [4] 57/-1/-1->56->63 [5] 57/-1/-1->56->63 [6] 57/-1/-1->56->63 [7] 57/-1/-1->56->63 [8] 57/24/-1->56->-1 [9] -1/-1/-1->56->63 [10] 57/-1/-1->56->63 [11] 57/-1/-1->56->63 [12] 57/-1/-1->56->63 [13] 57/-1/-1->56->63 [14] 57/-1/-1->56->63 [15] 57/-1/-1->56->63
+ip-26-0-173-7:811248:811415 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-173-7:811254:811414 [6] NCCL INFO Trees [0] 63/-1/-1->62->61 [1] 63/-1/-1->62->61 [2] 63/-1/-1->62->61 [3] 63/-1/-1->62->61 [4] 63/-1/-1->62->61 [5] 63/-1/-1->62->61 [6] 63/-1/-1->62->54 [7] -1/-1/-1->62->61 [8] 63/-1/-1->62->61 [9] 63/-1/-1->62->61 [10] 63/-1/-1->62->61 [11] 63/-1/-1->62->61 [12] 63/-1/-1->62->61 [13] 63/-1/-1->62->61 [14] 63/30/-1->62->-1 [15] -1/-1/-1->62->61
+ip-26-0-172-142:880681:880845 [2] NCCL INFO Trees [0] 19/-1/-1->18->17 [1] 19/-1/-1->18->17 [2] 19/10/26->18->34 [3] -1/-1/-1->18->17 [4] 19/-1/-1->18->17 [5] 19/-1/-1->18->17 [6] 19/-1/-1->18->17 [7] 19/-1/-1->18->17 [8] 19/-1/-1->18->17 [9] 19/-1/-1->18->17 [10] 19/-1/-1->18->10 [11] -1/-1/-1->18->17 [12] 19/-1/-1->18->17 [13] 19/-1/-1->18->17 [14] 19/-1/-1->18->17 [15] 19/-1/-1->18->17
+ip-26-0-172-142:880682:880840 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-142:880681:880845 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-147:618585:618748 [1] NCCL INFO Trees [0] 26/-1/-1->25->24 [1] 26/-1/-1->25->17 [2] -1/-1/-1->25->24 [3] 26/-1/-1->25->24 [4] 26/-1/-1->25->24 [5] 26/-1/-1->25->24 [6] 26/-1/-1->25->24 [7] 26/-1/-1->25->24 [8] 26/-1/-1->25->24 [9] 26/41/9->25->57 [10] -1/-1/-1->25->24 [11] 26/-1/-1->25->24 [12] 26/-1/-1->25->24 [13] 26/-1/-1->25->24 [14] 26/-1/-1->25->24 [15] 26/-1/-1->25->24
+ip-26-0-172-147:618588:618750 [4] NCCL INFO Trees [0] 29/-1/-1->28->27 [1] 29/-1/-1->28->27 [2] 29/-1/-1->28->27 [3] 29/-1/-1->28->27 [4] 29/-1/-1->28->20 [5] -1/-1/-1->28->27 [6] 29/-1/-1->28->27 [7] 29/-1/-1->28->27 [8] 29/-1/-1->28->27 [9] 29/-1/-1->28->27 [10] 29/-1/-1->28->27 [11] 29/-1/-1->28->27 [12] 29/44/12->28->60 [13] -1/-1/-1->28->27 [14] 29/-1/-1->28->27 [15] 29/-1/-1->28->27
+ip-26-0-172-147:618585:618748 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-225:94342:94504 [0] NCCL INFO Channel 03/16 :    0   7   6   5   4  11  10   9   8  15  14  13  12  19  18  17  16  23  22  21
+ip-26-0-160-225:94342:94504 [0] NCCL INFO Channel 04/16 :    0   7   6   5  12  11  10   9   8  15  14  13  20  19  18  17  16  23  22  21
+ip-26-0-172-142:880679:880846 [0] NCCL INFO Trees [0] 17/8/24->16->32 [1] -1/-1/-1->16->23 [2] 17/-1/-1->16->23 [3] 17/-1/-1->16->23 [4] 17/-1/-1->16->23 [5] 17/-1/-1->16->23 [6] 17/-1/-1->16->23 [7] 17/-1/-1->16->23 [8] 17/-1/-1->16->8 [9] -1/-1/-1->16->23 [10] 17/-1/-1->16->23 [11] 17/-1/-1->16->23 [12] 17/-1/-1->16->23 [13] 17/-1/-1->16->23 [14] 17/-1/-1->16->23 [15] 17/-1/-1->16->23
+ip-26-0-172-142:880679:880846 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-147:618587:618751 [3] NCCL INFO Trees [0] 28/-1/-1->27->26 [1] 28/-1/-1->27->26 [2] 28/-1/-1->27->26 [3] 28/-1/-1->27->19 [4] -1/-1/-1->27->26 [5] 28/-1/-1->27->26 [6] 28/-1/-1->27->26 [7] 28/-1/-1->27->26 [8] 28/-1/-1->27->26 [9] 28/-1/-1->27->26 [10] 28/-1/-1->27->26 [11] 28/43/11->27->59 [12] -1/-1/-1->27->26 [13] 28/-1/-1->27->26 [14] 28/-1/-1->27->26 [15] 28/-1/-1->27->26
+ip-26-0-172-147:618588:618750 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-147:618587:618751 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-225:94342:94504 [0] NCCL INFO Channel 05/16 :    0   4   7   6  13  11  10   9   8  12  15  14  21  19  18  17  16  20  23  22
+ip-26-0-160-225:94345:94509 [3] NCCL INFO Trees [0] 4/-1/-1->3->2 [1] 4/-1/-1->3->2 [2] 4/-1/-1->3->2 [3] 4/35/-1->3->-1 [4] -1/-1/-1->3->2 [5] 4/-1/-1->3->2 [6] 4/-1/-1->3->2 [7] 4/-1/-1->3->2 [8] 4/-1/-1->3->2 [9] 4/-1/-1->3->2 [10] 4/-1/-1->3->2 [11] 4/-1/-1->3->11 [12] -1/-1/-1->3->2 [13] 4/-1/-1->3->2 [14] 4/-1/-1->3->2 [15] 4/-1/-1->3->2
+ip-26-0-160-225:94342:94504 [0] NCCL INFO Channel 06/16 :    0   5   4   7  14  11  10   9   8  13  12  15  22  19  18  17  16  21  20  23
+ip-26-0-160-225:94346:94511 [4] NCCL INFO Trees [0] 5/-1/-1->4->3 [1] 5/-1/-1->4->3 [2] 5/-1/-1->4->3 [3] 5/-1/-1->4->3 [4] 5/36/-1->4->-1 [5] -1/-1/-1->4->3 [6] 5/-1/-1->4->3 [7] 5/-1/-1->4->3 [8] 5/-1/-1->4->3 [9] 5/-1/-1->4->3 [10] 5/-1/-1->4->3 [11] 5/-1/-1->4->3 [12] 5/-1/-1->4->12 [13] -1/-1/-1->4->3 [14] 5/-1/-1->4->3 [15] 5/-1/-1->4->3
+ip-26-0-172-142:880684:880842 [5] NCCL INFO Trees [0] 22/-1/-1->21->20 [1] 22/-1/-1->21->20 [2] 22/-1/-1->21->20 [3] 22/-1/-1->21->20 [4] 22/-1/-1->21->20 [5] 22/13/29->21->37 [6] -1/-1/-1->21->20 [7] 22/-1/-1->21->20 [8] 22/-1/-1->21->20 [9] 22/-1/-1->21->20 [10] 22/-1/-1->21->20 [11] 22/-1/-1->21->20 [12] 22/-1/-1->21->20 [13] 22/-1/-1->21->13 [14] -1/-1/-1->21->20 [15] 22/-1/-1->21->20
+ip-26-0-172-73:926598:926761 [4] NCCL INFO Trees [0] 53/-1/-1->52->51 [1] 53/-1/-1->52->51 [2] 53/-1/-1->52->51 [3] 53/-1/-1->52->51 [4] 53/44/60->52->36 [5] -1/-1/-1->52->51 [6] 53/-1/-1->52->51 [7] 53/-1/-1->52->51 [8] 53/-1/-1->52->51 [9] 53/-1/-1->52->51 [10] 53/-1/-1->52->51 [11] 53/-1/-1->52->51 [12] 53/-1/-1->52->44 [13] -1/-1/-1->52->51 [14] 53/-1/-1->52->51 [15] 53/-1/-1->52->51
+ip-26-0-172-73:926598:926761 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-147:618584:618747 [0] NCCL INFO Trees [0] 25/-1/-1->24->16 [1] -1/-1/-1->24->31 [2] 25/-1/-1->24->31 [3] 25/-1/-1->24->31 [4] 25/-1/-1->24->31 [5] 25/-1/-1->24->31 [6] 25/-1/-1->24->31 [7] 25/-1/-1->24->31 [8] 25/40/8->24->56 [9] -1/-1/-1->24->31 [10] 25/-1/-1->24->31 [11] 25/-1/-1->24->31 [12] 25/-1/-1->24->31 [13] 25/-1/-1->24->31 [14] 25/-1/-1->24->31 [15] 25/-1/-1->24->31
+ip-26-0-172-147:618584:618747 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-252:688203:688362 [3] NCCL INFO Trees [0] 36/-1/-1->35->34 [1] 36/-1/-1->35->34 [2] 36/-1/-1->35->34 [3] 36/19/51->35->3 [4] -1/-1/-1->35->34 [5] 36/-1/-1->35->34 [6] 36/-1/-1->35->34 [7] 36/-1/-1->35->34 [8] 36/-1/-1->35->34 [9] 36/-1/-1->35->34 [10] 36/-1/-1->35->34 [11] 36/-1/-1->35->43 [12] -1/-1/-1->35->34 [13] 36/-1/-1->35->34 [14] 36/-1/-1->35->34 [15] 36/-1/-1->35->34
+ip-26-0-172-252:688203:688362 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-225:94345:94509 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-225:94346:94511 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-225:94342:94504 [0] NCCL INFO Channel 07/16 :    0   6   5   4  15  11  10   9   8  14  13  12  23  19  18  17  16  22  21  20
+ip-26-0-173-7:811251:811409 [3] NCCL INFO Trees [0] 60/-1/-1->59->58 [1] 60/-1/-1->59->58 [2] 60/-1/-1->59->58 [3] 60/-1/-1->59->51 [4] -1/-1/-1->59->58 [5] 60/-1/-1->59->58 [6] 60/-1/-1->59->58 [7] 60/-1/-1->59->58 [8] 60/-1/-1->59->58 [9] 60/-1/-1->59->58 [10] 60/-1/-1->59->58 [11] 60/27/-1->59->-1 [12] -1/-1/-1->59->58 [13] 60/-1/-1->59->58 [14] 60/-1/-1->59->58 [15] 60/-1/-1->59->58
+ip-26-0-172-142:880684:880842 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-73:926601:926759 [7] NCCL INFO Trees [0] -1/-1/-1->55->54 [1] 48/-1/-1->55->54 [2] 48/-1/-1->55->54 [3] 48/-1/-1->55->54 [4] 48/-1/-1->55->54 [5] 48/-1/-1->55->54 [6] 48/-1/-1->55->54 [7] 48/47/63->55->39 [8] -1/-1/-1->55->54 [9] 48/-1/-1->55->54 [10] 48/-1/-1->55->54 [11] 48/-1/-1->55->54 [12] 48/-1/-1->55->54 [13] 48/-1/-1->55->54 [14] 48/-1/-1->55->54 [15] 48/-1/-1->55->47
+ip-26-0-172-73:926600:926757 [6] NCCL INFO Trees [0] 55/-1/-1->54->53 [1] 55/-1/-1->54->53 [2] 55/-1/-1->54->53 [3] 55/-1/-1->54->53 [4] 55/-1/-1->54->53 [5] 55/-1/-1->54->53 [6] 55/46/62->54->38 [7] -1/-1/-1->54->53 [8] 55/-1/-1->54->53 [9] 55/-1/-1->54->53 [10] 55/-1/-1->54->53 [11] 55/-1/-1->54->53 [12] 55/-1/-1->54->53 [13] 55/-1/-1->54->53 [14] 55/-1/-1->54->46 [15] -1/-1/-1->54->53
+ip-26-0-172-73:926601:926759 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-73:926600:926757 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-225:94342:94504 [0] NCCL INFO Channel 08/16 :    0   1   2   3   4   5   6   7   8   9  10  11  12  13  14  15  16  17  18  19
+ip-26-0-160-225:94343:94508 [1] NCCL INFO Trees [0] 2/-1/-1->1->0 [1] 2/33/-1->1->-1 [2] -1/-1/-1->1->0 [3] 2/-1/-1->1->0 [4] 2/-1/-1->1->0 [5] 2/-1/-1->1->0 [6] 2/-1/-1->1->0 [7] 2/-1/-1->1->0 [8] 2/-1/-1->1->0 [9] 2/-1/-1->1->9 [10] -1/-1/-1->1->0 [11] 2/-1/-1->1->0 [12] 2/-1/-1->1->0 [13] 2/-1/-1->1->0 [14] 2/-1/-1->1->0 [15] 2/-1/-1->1->0
+ip-26-0-160-225:94344:94510 [2] NCCL INFO Trees [0] 3/-1/-1->2->1 [1] 3/-1/-1->2->1 [2] 3/34/-1->2->-1 [3] -1/-1/-1->2->1 [4] 3/-1/-1->2->1 [5] 3/-1/-1->2->1 [6] 3/-1/-1->2->1 [7] 3/-1/-1->2->1 [8] 3/-1/-1->2->1 [9] 3/-1/-1->2->1 [10] 3/-1/-1->2->10 [11] -1/-1/-1->2->1 [12] 3/-1/-1->2->1 [13] 3/-1/-1->2->1 [14] 3/-1/-1->2->1 [15] 3/-1/-1->2->1
+ip-26-0-160-225:94342:94504 [0] NCCL INFO Channel 09/16 :    0   7   6   5   4   3   2   9   8  15  14  13  12  11  10  17  16  23  22  21
+ip-26-0-173-7:811254:811414 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-173-7:811251:811409 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-173-7:811252:811408 [4] NCCL INFO Trees [0] 61/-1/-1->60->59 [1] 61/-1/-1->60->59 [2] 61/-1/-1->60->59 [3] 61/-1/-1->60->59 [4] 61/-1/-1->60->52 [5] -1/-1/-1->60->59 [6] 61/-1/-1->60->59 [7] 61/-1/-1->60->59 [8] 61/-1/-1->60->59 [9] 61/-1/-1->60->59 [10] 61/-1/-1->60->59 [11] 61/-1/-1->60->59 [12] 61/28/-1->60->-1 [13] -1/-1/-1->60->59 [14] 61/-1/-1->60->59 [15] 61/-1/-1->60->59
+ip-26-0-173-7:811252:811408 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-73:926599:926755 [5] NCCL INFO Trees [0] 54/-1/-1->53->52 [1] 54/-1/-1->53->52 [2] 54/-1/-1->53->52 [3] 54/-1/-1->53->52 [4] 54/-1/-1->53->52 [5] 54/45/61->53->37 [6] -1/-1/-1->53->52 [7] 54/-1/-1->53->52 [8] 54/-1/-1->53->52 [9] 54/-1/-1->53->52 [10] 54/-1/-1->53->52 [11] 54/-1/-1->53->52 [12] 54/-1/-1->53->52 [13] 54/-1/-1->53->45 [14] -1/-1/-1->53->52 [15] 54/-1/-1->53->52
+ip-26-0-172-252:688202:688368 [2] NCCL INFO Trees [0] 35/-1/-1->34->33 [1] 35/-1/-1->34->33 [2] 35/18/50->34->2 [3] -1/-1/-1->34->33 [4] 35/-1/-1->34->33 [5] 35/-1/-1->34->33 [6] 35/-1/-1->34->33 [7] 35/-1/-1->34->33 [8] 35/-1/-1->34->33 [9] 35/-1/-1->34->33 [10] 35/-1/-1->34->42 [11] -1/-1/-1->34->33 [12] 35/-1/-1->34->33 [13] 35/-1/-1->34->33 [14] 35/-1/-1->34->33 [15] 35/-1/-1->34->33
+ip-26-0-172-252:688207:688366 [7] NCCL INFO Trees [0] -1/-1/-1->39->38 [1] 32/-1/-1->39->38 [2] 32/-1/-1->39->38 [3] 32/-1/-1->39->38 [4] 32/-1/-1->39->38 [5] 32/-1/-1->39->38 [6] 32/-1/-1->39->38 [7] 32/23/55->39->7 [8] -1/-1/-1->39->38 [9] 32/-1/-1->39->38 [10] 32/-1/-1->39->38 [11] 32/-1/-1->39->38 [12] 32/-1/-1->39->38 [13] 32/-1/-1->39->38 [14] 32/-1/-1->39->38 [15] 32/-1/-1->39->47
+ip-26-0-160-225:94343:94508 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-225:94344:94510 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-225:94342:94504 [0] NCCL INFO Channel 10/16 :    0   7   6   5   4   3  10   9   8  15  14  13  12  11  18  17  16  23  22  21
+ip-26-0-173-7:811253:811410 [5] NCCL INFO Trees [0] 62/-1/-1->61->60 [1] 62/-1/-1->61->60 [2] 62/-1/-1->61->60 [3] 62/-1/-1->61->60 [4] 62/-1/-1->61->60 [5] 62/-1/-1->61->53 [6] -1/-1/-1->61->60 [7] 62/-1/-1->61->60 [8] 62/-1/-1->61->60 [9] 62/-1/-1->61->60 [10] 62/-1/-1->61->60 [11] 62/-1/-1->61->60 [12] 62/-1/-1->61->60 [13] 62/29/-1->61->-1 [14] -1/-1/-1->61->60 [15] 62/-1/-1->61->60
+ip-26-0-173-7:811249:811413 [1] NCCL INFO Trees [0] 58/-1/-1->57->56 [1] 58/-1/-1->57->49 [2] -1/-1/-1->57->56 [3] 58/-1/-1->57->56 [4] 58/-1/-1->57->56 [5] 58/-1/-1->57->56 [6] 58/-1/-1->57->56 [7] 58/-1/-1->57->56 [8] 58/-1/-1->57->56 [9] 58/25/-1->57->-1 [10] -1/-1/-1->57->56 [11] 58/-1/-1->57->56 [12] 58/-1/-1->57->56 [13] 58/-1/-1->57->56 [14] 58/-1/-1->57->56 [15] 58/-1/-1->57->56
+ip-26-0-173-7:811253:811410 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-57:942611:942772 [0] NCCL INFO Trees [0] 41/-1/-1->40->48 [1] -1/-1/-1->40->47 [2] 41/-1/-1->40->47 [3] 41/-1/-1->40->47 [4] 41/-1/-1->40->47 [5] 41/-1/-1->40->47 [6] 41/-1/-1->40->47 [7] 41/-1/-1->40->47 [8] 41/48/32->40->24 [9] -1/-1/-1->40->47 [10] 41/-1/-1->40->47 [11] 41/-1/-1->40->47 [12] 41/-1/-1->40->47 [13] 41/-1/-1->40->47 [14] 41/-1/-1->40->47 [15] 41/-1/-1->40->47
+ip-26-0-172-57:942616:942777 [5] NCCL INFO Trees [0] 46/-1/-1->45->44 [1] 46/-1/-1->45->44 [2] 46/-1/-1->45->44 [3] 46/-1/-1->45->44 [4] 46/-1/-1->45->44 [5] 46/-1/-1->45->53 [6] -1/-1/-1->45->44 [7] 46/-1/-1->45->44 [8] 46/-1/-1->45->44 [9] 46/-1/-1->45->44 [10] 46/-1/-1->45->44 [11] 46/-1/-1->45->44 [12] 46/-1/-1->45->44 [13] 46/53/37->45->29 [14] -1/-1/-1->45->44 [15] 46/-1/-1->45->44
+ip-26-0-172-57:942611:942772 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-57:942616:942777 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-73:926599:926755 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-252:688205:688365 [5] NCCL INFO Trees [0] 38/-1/-1->37->36 [1] 38/-1/-1->37->36 [2] 38/-1/-1->37->36 [3] 38/-1/-1->37->36 [4] 38/-1/-1->37->36 [5] 38/21/53->37->5 [6] -1/-1/-1->37->36 [7] 38/-1/-1->37->36 [8] 38/-1/-1->37->36 [9] 38/-1/-1->37->36 [10] 38/-1/-1->37->36 [11] 38/-1/-1->37->36 [12] 38/-1/-1->37->36 [13] 38/-1/-1->37->45 [14] -1/-1/-1->37->36 [15] 38/-1/-1->37->36
+ip-26-0-172-252:688207:688366 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-252:688205:688365 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-252:688202:688368 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-225:94342:94504 [0] NCCL INFO Channel 11/16 :    0   7   6   5   4  11  10   9   8  15  14  13  12  19  18  17  16  23  22  21
+ip-26-0-160-225:94342:94504 [0] NCCL INFO Channel 12/16 :    0   7   6   5  12  11  10   9   8  15  14  13  20  19  18  17  16  23  22  21
+ip-26-0-160-225:94342:94504 [0] NCCL INFO Channel 13/16 :    0   4   7   6  13  11  10   9   8  12  15  14  21  19  18  17  16  20  23  22
+ip-26-0-173-7:811250:811412 [2] NCCL INFO Trees [0] 59/-1/-1->58->57 [1] 59/-1/-1->58->57 [2] 59/-1/-1->58->50 [3] -1/-1/-1->58->57 [4] 59/-1/-1->58->57 [5] 59/-1/-1->58->57 [6] 59/-1/-1->58->57 [7] 59/-1/-1->58->57 [8] 59/-1/-1->58->57 [9] 59/-1/-1->58->57 [10] 59/26/-1->58->-1 [11] -1/-1/-1->58->57 [12] 59/-1/-1->58->57 [13] 59/-1/-1->58->57 [14] 59/-1/-1->58->57 [15] 59/-1/-1->58->57
+ip-26-0-172-73:926594:926762 [0] NCCL INFO Trees [0] 49/40/56->48->32 [1] -1/-1/-1->48->55 [2] 49/-1/-1->48->55 [3] 49/-1/-1->48->55 [4] 49/-1/-1->48->55 [5] 49/-1/-1->48->55 [6] 49/-1/-1->48->55 [7] 49/-1/-1->48->55 [8] 49/-1/-1->48->40 [9] -1/-1/-1->48->55 [10] 49/-1/-1->48->55 [11] 49/-1/-1->48->55 [12] 49/-1/-1->48->55 [13] 49/-1/-1->48->55 [14] 49/-1/-1->48->55 [15] 49/-1/-1->48->55
+ip-26-0-172-73:926597:926758 [3] NCCL INFO Trees [0] 52/-1/-1->51->50 [1] 52/-1/-1->51->50 [2] 52/-1/-1->51->50 [3] 52/43/59->51->35 [4] -1/-1/-1->51->50 [5] 52/-1/-1->51->50 [6] 52/-1/-1->51->50 [7] 52/-1/-1->51->50 [8] 52/-1/-1->51->50 [9] 52/-1/-1->51->50 [10] 52/-1/-1->51->50 [11] 52/-1/-1->51->43 [12] -1/-1/-1->51->50 [13] 52/-1/-1->51->50 [14] 52/-1/-1->51->50 [15] 52/-1/-1->51->50
+ip-26-0-172-73:926594:926762 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-73:926597:926758 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-252:688204:688367 [4] NCCL INFO Trees [0] 37/-1/-1->36->35 [1] 37/-1/-1->36->35 [2] 37/-1/-1->36->35 [3] 37/-1/-1->36->35 [4] 37/20/52->36->4 [5] -1/-1/-1->36->35 [6] 37/-1/-1->36->35 [7] 37/-1/-1->36->35 [8] 37/-1/-1->36->35 [9] 37/-1/-1->36->35 [10] 37/-1/-1->36->35 [11] 37/-1/-1->36->35 [12] 37/-1/-1->36->44 [13] -1/-1/-1->36->35 [14] 37/-1/-1->36->35 [15] 37/-1/-1->36->35
+ip-26-0-172-252:688206:688363 [6] NCCL INFO Trees [0] 39/-1/-1->38->37 [1] 39/-1/-1->38->37 [2] 39/-1/-1->38->37 [3] 39/-1/-1->38->37 [4] 39/-1/-1->38->37 [5] 39/-1/-1->38->37 [6] 39/22/54->38->6 [7] -1/-1/-1->38->37 [8] 39/-1/-1->38->37 [9] 39/-1/-1->38->37 [10] 39/-1/-1->38->37 [11] 39/-1/-1->38->37 [12] 39/-1/-1->38->37 [13] 39/-1/-1->38->37 [14] 39/-1/-1->38->46 [15] -1/-1/-1->38->37
+ip-26-0-172-252:688204:688367 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-252:688206:688363 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-225:94342:94504 [0] NCCL INFO Channel 14/16 :    0   5   4   7  14  11  10   9   8  13  12  15  22  19  18  17  16  21  20  23
+ip-26-0-160-225:94342:94504 [0] NCCL INFO Channel 15/16 :    0   6   5   4  15  11  10   9   8  14  13  12  23  19  18  17  16  22  21  20
+ip-26-0-173-7:811249:811413 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-173-7:811250:811412 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-57:942615:942775 [4] NCCL INFO Trees [0] 45/-1/-1->44->43 [1] 45/-1/-1->44->43 [2] 45/-1/-1->44->43 [3] 45/-1/-1->44->43 [4] 45/-1/-1->44->52 [5] -1/-1/-1->44->43 [6] 45/-1/-1->44->43 [7] 45/-1/-1->44->43 [8] 45/-1/-1->44->43 [9] 45/-1/-1->44->43 [10] 45/-1/-1->44->43 [11] 45/-1/-1->44->43 [12] 45/52/36->44->28 [13] -1/-1/-1->44->43 [14] 45/-1/-1->44->43 [15] 45/-1/-1->44->43
+ip-26-0-172-57:942618:942778 [7] NCCL INFO Trees [0] -1/-1/-1->47->46 [1] 40/-1/-1->47->46 [2] 40/-1/-1->47->46 [3] 40/-1/-1->47->46 [4] 40/-1/-1->47->46 [5] 40/-1/-1->47->46 [6] 40/-1/-1->47->46 [7] 40/-1/-1->47->55 [8] -1/-1/-1->47->46 [9] 40/-1/-1->47->46 [10] 40/-1/-1->47->46 [11] 40/-1/-1->47->46 [12] 40/-1/-1->47->46 [13] 40/-1/-1->47->46 [14] 40/-1/-1->47->46 [15] 40/55/39->47->31
+ip-26-0-172-57:942615:942775 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-73:926596:926756 [2] NCCL INFO Trees [0] 51/-1/-1->50->49 [1] 51/-1/-1->50->49 [2] 51/42/58->50->34 [3] -1/-1/-1->50->49 [4] 51/-1/-1->50->49 [5] 51/-1/-1->50->49 [6] 51/-1/-1->50->49 [7] 51/-1/-1->50->49 [8] 51/-1/-1->50->49 [9] 51/-1/-1->50->49 [10] 51/-1/-1->50->42 [11] -1/-1/-1->50->49 [12] 51/-1/-1->50->49 [13] 51/-1/-1->50->49 [14] 51/-1/-1->50->49 [15] 51/-1/-1->50->49
+ip-26-0-160-225:94342:94504 [0] NCCL INFO Trees [0] 1/32/-1->0->-1 [1] -1/-1/-1->0->7 [2] 1/-1/-1->0->7 [3] 1/-1/-1->0->7 [4] 1/-1/-1->0->7 [5] 1/-1/-1->0->7 [6] 1/-1/-1->0->7 [7] 1/-1/-1->0->7 [8] 1/-1/-1->0->8 [9] -1/-1/-1->0->7 [10] 1/-1/-1->0->7 [11] 1/-1/-1->0->7 [12] 1/-1/-1->0->7 [13] 1/-1/-1->0->7 [14] 1/-1/-1->0->7 [15] 1/-1/-1->0->7
+ip-26-0-172-57:942613:942776 [2] NCCL INFO Trees [0] 43/-1/-1->42->41 [1] 43/-1/-1->42->41 [2] 43/-1/-1->42->50 [3] -1/-1/-1->42->41 [4] 43/-1/-1->42->41 [5] 43/-1/-1->42->41 [6] 43/-1/-1->42->41 [7] 43/-1/-1->42->41 [8] 43/-1/-1->42->41 [9] 43/-1/-1->42->41 [10] 43/50/34->42->26 [11] -1/-1/-1->42->41 [12] 43/-1/-1->42->41 [13] 43/-1/-1->42->41 [14] 43/-1/-1->42->41 [15] 43/-1/-1->42->41
+ip-26-0-172-57:942618:942778 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-57:942612:942779 [1] NCCL INFO Trees [0] 42/-1/-1->41->40 [1] 42/-1/-1->41->49 [2] -1/-1/-1->41->40 [3] 42/-1/-1->41->40 [4] 42/-1/-1->41->40 [5] 42/-1/-1->41->40 [6] 42/-1/-1->41->40 [7] 42/-1/-1->41->40 [8] 42/-1/-1->41->40 [9] 42/49/33->41->25 [10] -1/-1/-1->41->40 [11] 42/-1/-1->41->40 [12] 42/-1/-1->41->40 [13] 42/-1/-1->41->40 [14] 42/-1/-1->41->40 [15] 42/-1/-1->41->40
+ip-26-0-172-57:942613:942776 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-73:926596:926756 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-225:94342:94504 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-57:942617:942774 [6] NCCL INFO Trees [0] 47/-1/-1->46->45 [1] 47/-1/-1->46->45 [2] 47/-1/-1->46->45 [3] 47/-1/-1->46->45 [4] 47/-1/-1->46->45 [5] 47/-1/-1->46->45 [6] 47/-1/-1->46->54 [7] -1/-1/-1->46->45 [8] 47/-1/-1->46->45 [9] 47/-1/-1->46->45 [10] 47/-1/-1->46->45 [11] 47/-1/-1->46->45 [12] 47/-1/-1->46->45 [13] 47/-1/-1->46->45 [14] 47/54/38->46->30 [15] -1/-1/-1->46->45
+ip-26-0-172-57:942612:942779 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-57:942617:942774 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-73:926595:926760 [1] NCCL INFO Trees [0] 50/-1/-1->49->48 [1] 50/41/57->49->33 [2] -1/-1/-1->49->48 [3] 50/-1/-1->49->48 [4] 50/-1/-1->49->48 [5] 50/-1/-1->49->48 [6] 50/-1/-1->49->48 [7] 50/-1/-1->49->48 [8] 50/-1/-1->49->48 [9] 50/-1/-1->49->41 [10] -1/-1/-1->49->48 [11] 50/-1/-1->49->48 [12] 50/-1/-1->49->48 [13] 50/-1/-1->49->48 [14] 50/-1/-1->49->48 [15] 50/-1/-1->49->48
+ip-26-0-172-73:926595:926760 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-57:942614:942773 [3] NCCL INFO Trees [0] 44/-1/-1->43->42 [1] 44/-1/-1->43->42 [2] 44/-1/-1->43->42 [3] 44/-1/-1->43->51 [4] -1/-1/-1->43->42 [5] 44/-1/-1->43->42 [6] 44/-1/-1->43->42 [7] 44/-1/-1->43->42 [8] 44/-1/-1->43->42 [9] 44/-1/-1->43->42 [10] 44/-1/-1->43->42 [11] 44/51/35->43->27 [12] -1/-1/-1->43->42 [13] 44/-1/-1->43->42 [14] 44/-1/-1->43->42 [15] 44/-1/-1->43->42
+ip-26-0-172-57:942614:942773 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-172-252:688200:688369 [0] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-160-225:94348:94506 [6] NCCL INFO Channel 00/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-173-7:811248:811415 [0] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-172-252:688201:688364 [1] NCCL INFO Channel 00/0 : 33[1] -> 34[2] via P2P/IPC
+ip-26-0-160-225:94342:94504 [0] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-172-142:880685:880847 [6] NCCL INFO Channel 00/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-172-142:880683:880844 [4] NCCL INFO Channel 00/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-172-252:688201:688364 [1] NCCL INFO Channel 08/0 : 33[1] -> 34[2] via P2P/IPC
+ip-26-0-172-142:880685:880847 [6] NCCL INFO Channel 08/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-172-142:880679:880846 [0] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-172-142:880683:880844 [4] NCCL INFO Channel 08/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-172-147:618584:618747 [0] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-172-57:942611:942772 [0] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-160-225:94348:94506 [6] NCCL INFO Channel 08/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-57:942617:942774 [6] NCCL INFO Channel 00/0 : 46[6] -> 47[7] via P2P/IPC
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO Channel 00/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-172-147:618589:618744 [5] NCCL INFO Channel 00/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO Channel 00/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-160-225:94349:94505 [7] NCCL INFO Channel 00/0 : 7[7] -> 8[0] [send] via NET/Libfabric/0(0)/GDRDMA
+ip-26-0-172-57:942616:942777 [5] NCCL INFO Channel 00/0 : 45[5] -> 46[6] via P2P/IPC
+ip-26-0-172-57:942617:942774 [6] NCCL INFO Channel 08/0 : 46[6] -> 47[7] via P2P/IPC
+ip-26-0-172-57:942615:942775 [4] NCCL INFO Channel 00/0 : 44[4] -> 45[5] via P2P/IPC
+ip-26-0-172-147:618589:618744 [5] NCCL INFO Channel 08/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-172-147:618588:618750 [4] NCCL INFO Channel 00/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO Channel 08/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO Channel 08/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-172-252:688203:688362 [3] NCCL INFO Channel 00/0 : 35[3] -> 36[4] via P2P/IPC
+ip-26-0-172-147:618586:618745 [2] NCCL INFO Channel 00/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-172-142:880680:880843 [1] NCCL INFO Channel 00/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO Channel 00/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-172-252:688204:688367 [4] NCCL INFO Channel 00/0 : 36[4] -> 37[5] via P2P/IPC
+ip-26-0-172-252:688205:688365 [5] NCCL INFO Channel 00/0 : 37[5] -> 38[6] via P2P/IPC
+ip-26-0-172-252:688206:688363 [6] NCCL INFO Channel 00/0 : 38[6] -> 39[7] via P2P/IPC
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO Channel 00/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-172-142:880681:880845 [2] NCCL INFO Channel 00/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-172-142:880682:880840 [3] NCCL INFO Channel 00/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-160-225:94346:94511 [4] NCCL INFO Channel 00/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-252:688203:688362 [3] NCCL INFO Channel 08/0 : 35[3] -> 36[4] via P2P/IPC
+ip-26-0-172-147:618588:618750 [4] NCCL INFO Channel 08/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-172-57:942616:942777 [5] NCCL INFO Channel 08/0 : 45[5] -> 46[6] via P2P/IPC
+ip-26-0-172-142:880684:880842 [5] NCCL INFO Channel 00/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-172-252:688202:688368 [2] NCCL INFO Channel 00/0 : 34[2] -> 35[3] via P2P/IPC
+ip-26-0-172-142:880680:880843 [1] NCCL INFO Channel 08/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-173-7:811254:811414 [6] NCCL INFO Channel 00/0 : 62[6] -> 63[7] via P2P/IPC
+ip-26-0-172-57:942615:942775 [4] NCCL INFO Channel 08/0 : 44[4] -> 45[5] via P2P/IPC
+ip-26-0-172-252:688204:688367 [4] NCCL INFO Channel 08/0 : 36[4] -> 37[5] via P2P/IPC
+ip-26-0-172-57:942613:942776 [2] NCCL INFO Channel 00/0 : 42[2] -> 43[3] via P2P/IPC
+ip-26-0-172-252:688205:688365 [5] NCCL INFO Channel 08/0 : 37[5] -> 38[6] via P2P/IPC
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO Channel 08/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO Channel 00/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO Channel 00/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-172-252:688206:688363 [6] NCCL INFO Channel 08/0 : 38[6] -> 39[7] via P2P/IPC
+ip-26-0-173-7:811251:811409 [3] NCCL INFO Channel 00/0 : 59[3] -> 60[4] via P2P/IPC
+ip-26-0-172-142:880682:880840 [3] NCCL INFO Channel 08/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-173-7:811249:811413 [1] NCCL INFO Channel 00/0 : 57[1] -> 58[2] via P2P/IPC
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO Channel 08/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-172-142:880681:880845 [2] NCCL INFO Channel 08/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-160-225:94345:94509 [3] NCCL INFO Channel 00/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-225:94347:94507 [5] NCCL INFO Channel 00/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-225:94343:94508 [1] NCCL INFO Channel 00/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-225:94346:94511 [4] NCCL INFO Channel 08/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-173-7:811254:811414 [6] NCCL INFO Channel 08/0 : 62[6] -> 63[7] via P2P/IPC
+ip-26-0-172-142:880684:880842 [5] NCCL INFO Channel 08/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-173-7:811253:811410 [5] NCCL INFO Channel 00/0 : 61[5] -> 62[6] via P2P/IPC
+ip-26-0-172-57:942612:942779 [1] NCCL INFO Channel 00/0 : 41[1] -> 42[2] via P2P/IPC
+ip-26-0-172-57:942614:942773 [3] NCCL INFO Channel 00/0 : 43[3] -> 44[4] via P2P/IPC
+ip-26-0-172-252:688202:688368 [2] NCCL INFO Channel 08/0 : 34[2] -> 35[3] via P2P/IPC
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO Channel 08/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO Channel 08/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-173-7:811251:811409 [3] NCCL INFO Channel 08/0 : 59[3] -> 60[4] via P2P/IPC
+ip-26-0-173-7:811249:811413 [1] NCCL INFO Channel 08/0 : 57[1] -> 58[2] via P2P/IPC
+ip-26-0-172-57:942613:942776 [2] NCCL INFO Channel 08/0 : 42[2] -> 43[3] via P2P/IPC
+ip-26-0-172-142:880682:880840 [3] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-172-252:688204:688367 [4] NCCL INFO Channel 05/0 : 36[4] -> 39[7] via P2P/IPC
+ip-26-0-172-142:880681:880845 [2] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-172-142:880683:880844 [4] NCCL INFO Channel 05/0 : 20[4] -> 23[7] via P2P/IPC
+ip-26-0-172-57:942612:942779 [1] NCCL INFO Channel 08/0 : 41[1] -> 42[2] via P2P/IPC
+ip-26-0-160-225:94347:94507 [5] NCCL INFO Channel 08/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-173-7:811253:811410 [5] NCCL INFO Channel 08/0 : 61[5] -> 62[6] via P2P/IPC
+ip-26-0-172-57:942614:942773 [3] NCCL INFO Channel 08/0 : 43[3] -> 44[4] via P2P/IPC
+ip-26-0-172-252:688203:688362 [3] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-160-225:94343:94508 [1] NCCL INFO Channel 08/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-147:618587:618751 [3] NCCL INFO Channel 00/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-160-225:94345:94509 [3] NCCL INFO Channel 08/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-147:618590:618749 [6] NCCL INFO Channel 00/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-172-252:688202:688368 [2] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-160-225:94344:94510 [2] NCCL INFO Channel 00/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-142:880683:880844 [4] NCCL INFO Channel 06/0 : 20[4] -> 23[7] via P2P/IPC
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO Channel 05/0 : 12[4] -> 15[7] via P2P/IPC
+ip-26-0-172-252:688204:688367 [4] NCCL INFO Channel 06/0 : 36[4] -> 39[7] via P2P/IPC
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-172-147:618586:618745 [2] NCCL INFO Channel 08/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-172-147:618585:618748 [1] NCCL INFO Channel 00/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-172-142:880683:880844 [4] NCCL INFO Channel 13/0 : 20[4] -> 23[7] via P2P/IPC
+ip-26-0-173-7:811252:811408 [4] NCCL INFO Channel 00/0 : 60[4] -> 61[5] via P2P/IPC
+ip-26-0-172-147:618590:618749 [6] NCCL INFO Channel 08/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-172-252:688204:688367 [4] NCCL INFO Channel 13/0 : 36[4] -> 39[7] via P2P/IPC
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO Channel 06/0 : 12[4] -> 15[7] via P2P/IPC
+ip-26-0-172-57:942613:942776 [2] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-172-142:880683:880844 [4] NCCL INFO Channel 14/0 : 20[4] -> 23[7] via P2P/IPC
+ip-26-0-160-225:94344:94510 [2] NCCL INFO Channel 08/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-57:942615:942775 [4] NCCL INFO Channel 05/0 : 44[4] -> 47[7] via P2P/IPC
+ip-26-0-172-252:688204:688367 [4] NCCL INFO Channel 14/0 : 36[4] -> 39[7] via P2P/IPC
+ip-26-0-172-57:942614:942773 [3] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-173-7:811252:811408 [4] NCCL INFO Channel 08/0 : 60[4] -> 61[5] via P2P/IPC
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO Channel 13/0 : 12[4] -> 15[7] via P2P/IPC
+ip-26-0-160-225:94346:94511 [4] NCCL INFO Channel 05/0 : 4[4] -> 7[7] via P2P/IPC
+ip-26-0-173-7:811250:811412 [2] NCCL INFO Channel 00/0 : 58[2] -> 59[3] via P2P/IPC
+ip-26-0-172-147:618585:618748 [1] NCCL INFO Channel 08/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO Channel 14/0 : 12[4] -> 15[7] via P2P/IPC
+ip-26-0-172-57:942615:942775 [4] NCCL INFO Channel 06/0 : 44[4] -> 47[7] via P2P/IPC
+ip-26-0-173-7:811250:811412 [2] NCCL INFO Channel 08/0 : 58[2] -> 59[3] via P2P/IPC
+ip-26-0-172-57:942615:942775 [4] NCCL INFO Channel 13/0 : 44[4] -> 47[7] via P2P/IPC
+ip-26-0-173-7:811252:811408 [4] NCCL INFO Channel 05/0 : 60[4] -> 63[7] via P2P/IPC
+ip-26-0-172-57:942615:942775 [4] NCCL INFO Channel 14/0 : 44[4] -> 47[7] via P2P/IPC
+ip-26-0-173-7:811251:811409 [3] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-173-7:811250:811412 [2] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-173-7:811252:811408 [4] NCCL INFO Channel 06/0 : 60[4] -> 63[7] via P2P/IPC
+ip-26-0-173-7:811252:811408 [4] NCCL INFO Channel 13/0 : 60[4] -> 63[7] via P2P/IPC
+ip-26-0-172-147:618587:618751 [3] NCCL INFO Channel 08/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-160-225:94345:94509 [3] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-160-225:94346:94511 [4] NCCL INFO Channel 06/0 : 4[4] -> 7[7] via P2P/IPC
+ip-26-0-160-225:94344:94510 [2] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-173-7:811252:811408 [4] NCCL INFO Channel 14/0 : 60[4] -> 63[7] via P2P/IPC
+ip-26-0-160-225:94346:94511 [4] NCCL INFO Channel 13/0 : 4[4] -> 7[7] via P2P/IPC
+ip-26-0-172-147:618586:618745 [2] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-172-147:618588:618750 [4] NCCL INFO Channel 05/0 : 28[4] -> 31[7] via P2P/IPC
+ip-26-0-160-225:94346:94511 [4] NCCL INFO Channel 14/0 : 4[4] -> 7[7] via P2P/IPC
+ip-26-0-172-147:618587:618751 [3] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-172-147:618588:618750 [4] NCCL INFO Channel 06/0 : 28[4] -> 31[7] via P2P/IPC
+ip-26-0-172-147:618588:618750 [4] NCCL INFO Channel 13/0 : 28[4] -> 31[7] via P2P/IPC
+ip-26-0-172-147:618588:618750 [4] NCCL INFO Channel 14/0 : 28[4] -> 31[7] via P2P/IPC
+ip-26-0-172-73:926594:926762 [0] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-172-73:926598:926761 [4] NCCL INFO Channel 00/0 : 52[4] -> 53[5] via P2P/IPC
+ip-26-0-172-73:926595:926760 [1] NCCL INFO Channel 00/0 : 49[1] -> 50[2] via P2P/IPC
+ip-26-0-172-73:926600:926757 [6] NCCL INFO Channel 00/0 : 54[6] -> 55[7] via P2P/IPC
+ip-26-0-172-73:926598:926761 [4] NCCL INFO Channel 08/0 : 52[4] -> 53[5] via P2P/IPC
+ip-26-0-172-73:926600:926757 [6] NCCL INFO Channel 08/0 : 54[6] -> 55[7] via P2P/IPC
+ip-26-0-172-73:926595:926760 [1] NCCL INFO Channel 08/0 : 49[1] -> 50[2] via P2P/IPC
+ip-26-0-172-73:926599:926755 [5] NCCL INFO Channel 00/0 : 53[5] -> 54[6] via P2P/IPC
+ip-26-0-172-73:926599:926755 [5] NCCL INFO Channel 08/0 : 53[5] -> 54[6] via P2P/IPC
+ip-26-0-172-73:926596:926756 [2] NCCL INFO Channel 00/0 : 50[2] -> 51[3] via P2P/IPC
+ip-26-0-172-73:926597:926758 [3] NCCL INFO Channel 00/0 : 51[3] -> 52[4] via P2P/IPC
+ip-26-0-172-73:926596:926756 [2] NCCL INFO Channel 08/0 : 50[2] -> 51[3] via P2P/IPC
+ip-26-0-172-73:926597:926758 [3] NCCL INFO Channel 08/0 : 51[3] -> 52[4] via P2P/IPC
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO Channel 00/0 : 7[7] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO Channel 08/0 : 7[7] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO Channel 00/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO Channel 00/0 : 15[7] -> 16[0] [send] via NET/Libfabric/0(8)/GDRDMA
+ip-26-0-160-225:94342:94504 [0] NCCL INFO Channel 00/0 : 63[7] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:94349:94505 [7] NCCL INFO Channel 08/0 : 7[7] -> 8[0] [send] via NET/Libfabric/0(0)/GDRDMA
+ip-26-0-160-225:94342:94504 [0] NCCL INFO Channel 08/0 : 63[7] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:94342:94504 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO Channel 08/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO Channel 08/0 : 15[7] -> 16[0] [send] via NET/Libfabric/0(8)/GDRDMA
+ip-26-0-160-225:94342:94504 [0] NCCL INFO Channel 08/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-252:688200:688369 [0] NCCL INFO Channel 00/0 : 31[7] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO Channel 05/0 : 8[0] -> 12[4] via P2P/IPC
+ip-26-0-172-252:688207:688366 [7] NCCL INFO Channel 00/0 : 39[7] -> 40[0] [send] via NET/Libfabric/0(32)/GDRDMA
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-172-252:688200:688369 [0] NCCL INFO Channel 08/0 : 31[7] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688200:688369 [0] NCCL INFO Channel 00/0 : 32[0] -> 33[1] via P2P/IPC
+ip-26-0-172-252:688207:688366 [7] NCCL INFO Channel 08/0 : 39[7] -> 40[0] [send] via NET/Libfabric/0(32)/GDRDMA
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO Channel 13/0 : 8[0] -> 12[4] via P2P/IPC
+ip-26-0-172-142:880679:880846 [0] NCCL INFO Channel 00/0 : 15[7] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:94343:94508 [1] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-172-142:880679:880846 [0] NCCL INFO Channel 08/0 : 15[7] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880686:880841 [7] NCCL INFO Channel 00/0 : 23[7] -> 24[0] [send] via NET/Libfabric/0(16)/GDRDMA
+ip-26-0-172-142:880679:880846 [0] NCCL INFO Channel 00/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-172-252:688200:688369 [0] NCCL INFO Channel 08/0 : 32[0] -> 33[1] via P2P/IPC
+ip-26-0-173-7:811248:811415 [0] NCCL INFO Channel 00/0 : 55[7] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-7:811248:811415 [0] NCCL INFO Channel 08/0 : 55[7] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-7:811248:811415 [0] NCCL INFO Channel 00/0 : 56[0] -> 57[1] via P2P/IPC
+ip-26-0-173-7:811255:811411 [7] NCCL INFO Channel 00/0 : 63[7] -> 0[0] [send] via NET/Libfabric/0(56)/GDRDMA
+ip-26-0-172-142:880679:880846 [0] NCCL INFO Channel 08/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-172-142:880686:880841 [7] NCCL INFO Channel 08/0 : 23[7] -> 24[0] [send] via NET/Libfabric/0(16)/GDRDMA
+ip-26-0-172-252:688201:688364 [1] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-172-147:618584:618747 [0] NCCL INFO Channel 00/0 : 23[7] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-7:811248:811415 [0] NCCL INFO Channel 08/0 : 56[0] -> 57[1] via P2P/IPC
+ip-26-0-172-147:618591:618746 [7] NCCL INFO Channel 00/0 : 31[7] -> 32[0] [send] via NET/Libfabric/0(24)/GDRDMA
+ip-26-0-173-7:811255:811411 [7] NCCL INFO Channel 08/0 : 63[7] -> 0[0] [send] via NET/Libfabric/0(56)/GDRDMA
+ip-26-0-172-147:618584:618747 [0] NCCL INFO Channel 08/0 : 23[7] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618584:618747 [0] NCCL INFO Channel 00/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-172-147:618591:618746 [7] NCCL INFO Channel 08/0 : 31[7] -> 32[0] [send] via NET/Libfabric/0(24)/GDRDMA
+ip-26-0-160-225:94342:94504 [0] NCCL INFO Channel 05/0 : 0[0] -> 4[4] via P2P/IPC
+ip-26-0-172-142:880680:880843 [1] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-172-252:688200:688369 [0] NCCL INFO Channel 05/0 : 32[0] -> 36[4] via P2P/IPC
+ip-26-0-172-142:880679:880846 [0] NCCL INFO Channel 05/0 : 16[0] -> 20[4] via P2P/IPC
+ip-26-0-173-7:811249:811413 [1] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-160-225:94342:94504 [0] NCCL INFO Channel 13/0 : 0[0] -> 4[4] via P2P/IPC
+ip-26-0-172-252:688200:688369 [0] NCCL INFO Channel 13/0 : 32[0] -> 36[4] via P2P/IPC
+ip-26-0-172-142:880679:880846 [0] NCCL INFO Channel 13/0 : 16[0] -> 20[4] via P2P/IPC
+ip-26-0-172-57:942611:942772 [0] NCCL INFO Channel 00/0 : 39[7] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618584:618747 [0] NCCL INFO Channel 08/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-172-57:942611:942772 [0] NCCL INFO Channel 08/0 : 39[7] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:94346:94511 [4] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-160-225:94342:94504 [0] NCCL INFO Channel 06/0 : 0[0] -> 5[5] via P2P/IPC
+ip-26-0-172-57:942611:942772 [0] NCCL INFO Channel 00/0 : 40[0] -> 41[1] via P2P/IPC
+ip-26-0-172-57:942618:942778 [7] NCCL INFO Channel 00/0 : 47[7] -> 48[0] [send] via NET/Libfabric/0(40)/GDRDMA
+ip-26-0-160-225:94342:94504 [0] NCCL INFO Channel 14/0 : 0[0] -> 5[5] via P2P/IPC
+ip-26-0-160-225:94342:94504 [0] NCCL INFO Channel 07/0 : 0[0] -> 6[6] via P2P/IPC
+ip-26-0-172-57:942611:942772 [0] NCCL INFO Channel 08/0 : 40[0] -> 41[1] via P2P/IPC
+ip-26-0-172-147:618584:618747 [0] NCCL INFO Channel 05/0 : 24[0] -> 28[4] via P2P/IPC
+ip-26-0-172-57:942618:942778 [7] NCCL INFO Channel 08/0 : 47[7] -> 48[0] [send] via NET/Libfabric/0(40)/GDRDMA
+ip-26-0-160-225:94342:94504 [0] NCCL INFO Channel 15/0 : 0[0] -> 6[6] via P2P/IPC
+ip-26-0-172-147:618585:618748 [1] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-173-7:811251:811409 [3] NCCL INFO Channel 03/0 : 52[4] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618584:618747 [0] NCCL INFO Channel 13/0 : 24[0] -> 28[4] via P2P/IPC
+ip-26-0-173-7:811251:811409 [3] NCCL INFO Channel 11/0 : 52[4] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:94342:94504 [0] NCCL INFO Channel 01/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-172-57:942612:942779 [1] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-172-57:942611:942772 [0] NCCL INFO Channel 05/0 : 40[0] -> 44[4] via P2P/IPC
+ip-26-0-160-225:94342:94504 [0] NCCL INFO Channel 02/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO Channel 03/0 : 4[4] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO Channel 06/0 : 8[0] -> 13[5] via P2P/IPC
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO Channel 11/0 : 4[4] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:94348:94506 [6] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-160-225:94342:94504 [0] NCCL INFO Channel 03/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-172-57:942611:942772 [0] NCCL INFO Channel 13/0 : 40[0] -> 44[4] via P2P/IPC
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO Channel 14/0 : 8[0] -> 13[5] via P2P/IPC
+ip-26-0-172-147:618584:618747 [0] NCCL INFO Channel 06/0 : 24[0] -> 29[5] via P2P/IPC
+ip-26-0-160-225:94342:94504 [0] NCCL INFO Channel 04/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-172-147:618588:618750 [4] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO Channel 02/0 : 3[3] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688202:688368 [2] NCCL INFO Channel 02/0 : 27[3] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:94349:94505 [7] NCCL INFO Channel 06/0 : 7[7] -> 14[6] [send] via NET/Libfabric/6(6)/GDRDMA
+ip-26-0-160-225:94342:94504 [0] NCCL INFO Channel 09/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO Channel 07/0 : 8[0] -> 14[6] via P2P/IPC
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO Channel 10/0 : 3[3] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688202:688368 [2] NCCL INFO Channel 10/0 : 27[3] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:94344:94510 [2] NCCL INFO Channel 02/0 : 59[3] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO Channel 02/0 : 11[3] -> 18[2] [send] via NET/Libfabric/2(10)/GDRDMA
+ip-26-0-172-147:618584:618747 [0] NCCL INFO Channel 14/0 : 24[0] -> 29[5] via P2P/IPC
+ip-26-0-160-225:94347:94507 [5] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-160-225:94344:94510 [2] NCCL INFO Channel 10/0 : 59[3] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO Channel 10/0 : 11[3] -> 18[2] [send] via NET/Libfabric/2(10)/GDRDMA
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO Channel 15/0 : 8[0] -> 14[6] via P2P/IPC
+ip-26-0-172-142:880681:880845 [2] NCCL INFO Channel 02/0 : 11[3] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:94342:94504 [0] NCCL INFO Channel 10/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-172-142:880681:880845 [2] NCCL INFO Channel 10/0 : 11[3] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618584:618747 [0] NCCL INFO Channel 07/0 : 24[0] -> 30[6] via P2P/IPC
+ip-26-0-160-225:94342:94504 [0] NCCL INFO Channel 11/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-173-7:811250:811412 [2] NCCL INFO Channel 02/0 : 51[3] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880679:880846 [0] NCCL INFO Channel 06/0 : 16[0] -> 21[5] via P2P/IPC
+ip-26-0-173-7:811250:811412 [2] NCCL INFO Channel 10/0 : 51[3] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880683:880844 [4] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-160-225:94342:94504 [0] NCCL INFO Channel 12/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO Channel 01/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-173-7:811251:811409 [3] NCCL INFO Channel 02/0 : 59[3] -> 2[2] [send] via NET/Libfabric/2(58)/GDRDMA
+ip-26-0-172-147:618584:618747 [0] NCCL INFO Channel 15/0 : 24[0] -> 30[6] via P2P/IPC
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-173-7:811251:811409 [3] NCCL INFO Channel 10/0 : 59[3] -> 2[2] [send] via NET/Libfabric/2(58)/GDRDMA
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-172-252:688200:688369 [0] NCCL INFO Channel 06/0 : 32[0] -> 37[5] via P2P/IPC
+ip-26-0-160-225:94345:94509 [3] NCCL INFO Channel 03/0 : 60[4] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880679:880846 [0] NCCL INFO Channel 14/0 : 16[0] -> 21[5] via P2P/IPC
+ip-26-0-172-252:688204:688367 [4] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-160-225:94345:94509 [3] NCCL INFO Channel 11/0 : 60[4] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO Channel 02/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-172-147:618584:618747 [0] NCCL INFO Channel 01/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-160-225:94345:94509 [3] NCCL INFO Channel 02/0 : 3[3] -> 10[2] [send] via NET/Libfabric/2(2)/GDRDMA
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO Channel 03/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-172-147:618590:618749 [6] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-172-252:688200:688369 [0] NCCL INFO Channel 14/0 : 32[0] -> 37[5] via P2P/IPC
+ip-26-0-172-142:880679:880846 [0] NCCL INFO Channel 07/0 : 16[0] -> 22[6] via P2P/IPC
+ip-26-0-172-147:618589:618744 [5] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-160-225:94345:94509 [3] NCCL INFO Channel 10/0 : 3[3] -> 10[2] [send] via NET/Libfabric/2(2)/GDRDMA
+ip-26-0-172-142:880682:880840 [3] NCCL INFO Channel 03/0 : 12[4] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880684:880842 [5] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-172-147:618584:618747 [0] NCCL INFO Channel 02/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO Channel 04/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-172-142:880682:880840 [3] NCCL INFO Channel 11/0 : 12[4] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880679:880846 [0] NCCL INFO Channel 15/0 : 16[0] -> 22[6] via P2P/IPC
+ip-26-0-172-252:688200:688369 [0] NCCL INFO Channel 07/0 : 32[0] -> 38[6] via P2P/IPC
+ip-26-0-172-147:618584:618747 [0] NCCL INFO Channel 03/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO Channel 09/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-172-142:880682:880840 [3] NCCL INFO Channel 02/0 : 19[3] -> 26[2] [send] via NET/Libfabric/2(18)/GDRDMA
+ip-26-0-172-252:688205:688365 [5] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-172-252:688203:688362 [3] NCCL INFO Channel 03/0 : 28[4] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880682:880840 [3] NCCL INFO Channel 10/0 : 19[3] -> 26[2] [send] via NET/Libfabric/2(18)/GDRDMA
+ip-26-0-172-252:688203:688362 [3] NCCL INFO Channel 11/0 : 28[4] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-7:811255:811411 [7] NCCL INFO Channel 06/0 : 63[7] -> 6[6] [send] via NET/Libfabric/6(62)/GDRDMA
+ip-26-0-172-147:618584:618747 [0] NCCL INFO Channel 04/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-172-252:688200:688369 [0] NCCL INFO Channel 15/0 : 32[0] -> 38[6] via P2P/IPC
+ip-26-0-173-7:811255:811411 [7] NCCL INFO Channel 14/0 : 63[7] -> 6[6] [send] via NET/Libfabric/6(62)/GDRDMA
+ip-26-0-172-252:688203:688362 [3] NCCL INFO Channel 02/0 : 35[3] -> 42[2] [send] via NET/Libfabric/2(34)/GDRDMA
+ip-26-0-172-142:880679:880846 [0] NCCL INFO Channel 01/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-172-252:688203:688362 [3] NCCL INFO Channel 10/0 : 35[3] -> 42[2] [send] via NET/Libfabric/2(34)/GDRDMA
+ip-26-0-172-147:618584:618747 [0] NCCL INFO Channel 09/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO Channel 10/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-172-57:942614:942773 [3] NCCL INFO Channel 03/0 : 36[4] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942613:942776 [2] NCCL INFO Channel 02/0 : 35[3] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880679:880846 [0] NCCL INFO Channel 02/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-172-57:942614:942773 [3] NCCL INFO Channel 11/0 : 36[4] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO Channel 11/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-172-57:942613:942776 [2] NCCL INFO Channel 10/0 : 35[3] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688200:688369 [0] NCCL INFO Channel 01/0 : 32[0] -> 39[7] via P2P/IPC
+ip-26-0-172-142:880679:880846 [0] NCCL INFO Channel 03/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-172-252:688206:688363 [6] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-172-57:942614:942773 [3] NCCL INFO Channel 02/0 : 43[3] -> 50[2] [send] via NET/Libfabric/2(42)/GDRDMA
+ip-26-0-172-147:618584:618747 [0] NCCL INFO Channel 10/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO Channel 12/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-172-57:942614:942773 [3] NCCL INFO Channel 10/0 : 43[3] -> 50[2] [send] via NET/Libfabric/2(42)/GDRDMA
+ip-26-0-172-252:688200:688369 [0] NCCL INFO Channel 02/0 : 32[0] -> 39[7] via P2P/IPC
+ip-26-0-172-142:880679:880846 [0] NCCL INFO Channel 04/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-172-147:618584:618747 [0] NCCL INFO Channel 11/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-172-142:880685:880847 [6] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-172-252:688200:688369 [0] NCCL INFO Channel 03/0 : 32[0] -> 39[7] via P2P/IPC
+ip-26-0-172-142:880679:880846 [0] NCCL INFO Channel 09/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-172-147:618584:618747 [0] NCCL INFO Channel 12/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-172-252:688200:688369 [0] NCCL INFO Channel 04/0 : 32[0] -> 39[7] via P2P/IPC
+ip-26-0-172-252:688200:688369 [0] NCCL INFO Channel 09/0 : 32[0] -> 39[7] via P2P/IPC
+ip-26-0-172-142:880679:880846 [0] NCCL INFO Channel 10/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-172-73:926594:926762 [0] NCCL INFO Channel 00/0 : 47[7] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926594:926762 [0] NCCL INFO Channel 08/0 : 47[7] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926594:926762 [0] NCCL INFO Channel 00/0 : 48[0] -> 49[1] via P2P/IPC
+ip-26-0-172-142:880679:880846 [0] NCCL INFO Channel 11/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-172-73:926601:926759 [7] NCCL INFO Channel 00/0 : 55[7] -> 56[0] [send] via NET/Libfabric/0(48)/GDRDMA
+ip-26-0-172-252:688200:688369 [0] NCCL INFO Channel 10/0 : 32[0] -> 39[7] via P2P/IPC
+ip-26-0-172-142:880686:880841 [7] NCCL INFO Channel 06/0 : 23[7] -> 30[6] [send] via NET/Libfabric/6(22)/GDRDMA
+ip-26-0-172-142:880679:880846 [0] NCCL INFO Channel 12/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-172-252:688200:688369 [0] NCCL INFO Channel 11/0 : 32[0] -> 39[7] via P2P/IPC
+ip-26-0-172-252:688207:688366 [7] NCCL INFO Channel 06/0 : 39[7] -> 46[6] [send] via NET/Libfabric/6(38)/GDRDMA
+ip-26-0-172-252:688200:688369 [0] NCCL INFO Channel 12/0 : 32[0] -> 39[7] via P2P/IPC
+ip-26-0-172-252:688207:688366 [7] NCCL INFO Channel 14/0 : 39[7] -> 46[6] [send] via NET/Libfabric/6(38)/GDRDMA
+ip-26-0-172-147:618587:618751 [3] NCCL INFO Channel 03/0 : 20[4] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618587:618751 [3] NCCL INFO Channel 11/0 : 20[4] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618586:618745 [2] NCCL INFO Channel 02/0 : 19[3] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618586:618745 [2] NCCL INFO Channel 10/0 : 19[3] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618587:618751 [3] NCCL INFO Channel 02/0 : 27[3] -> 34[2] [send] via NET/Libfabric/2(26)/GDRDMA
+ip-26-0-172-147:618587:618751 [3] NCCL INFO Channel 10/0 : 27[3] -> 34[2] [send] via NET/Libfabric/2(26)/GDRDMA
+ip-26-0-172-73:926596:926756 [2] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-172-73:926598:926761 [4] NCCL INFO Channel 05/0 : 52[4] -> 55[7] via P2P/IPC
+ip-26-0-172-73:926597:926758 [3] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-172-73:926598:926761 [4] NCCL INFO Channel 06/0 : 52[4] -> 55[7] via P2P/IPC
+ip-26-0-172-73:926594:926762 [0] NCCL INFO Channel 08/0 : 48[0] -> 49[1] via P2P/IPC
+ip-26-0-172-73:926601:926759 [7] NCCL INFO Channel 08/0 : 55[7] -> 56[0] [send] via NET/Libfabric/0(48)/GDRDMA
+ip-26-0-173-7:811248:811415 [0] NCCL INFO Channel 05/0 : 56[0] -> 60[4] via P2P/IPC
+ip-26-0-173-7:811248:811415 [0] NCCL INFO Channel 13/0 : 56[0] -> 60[4] via P2P/IPC
+ip-26-0-172-73:926598:926761 [4] NCCL INFO Channel 13/0 : 52[4] -> 55[7] via P2P/IPC
+ip-26-0-173-7:811248:811415 [0] NCCL INFO Channel 06/0 : 56[0] -> 61[5] via P2P/IPC
+ip-26-0-173-7:811252:811408 [4] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-173-7:811248:811415 [0] NCCL INFO Channel 14/0 : 56[0] -> 61[5] via P2P/IPC
+ip-26-0-173-7:811248:811415 [0] NCCL INFO Channel 07/0 : 56[0] -> 62[6] via P2P/IPC
+ip-26-0-173-7:811253:811410 [5] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-173-7:811248:811415 [0] NCCL INFO Channel 15/0 : 56[0] -> 62[6] via P2P/IPC
+ip-26-0-173-7:811248:811415 [0] NCCL INFO Channel 01/0 : 56[0] -> 63[7] via P2P/IPC
+ip-26-0-173-7:811254:811414 [6] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-173-7:811248:811415 [0] NCCL INFO Channel 02/0 : 56[0] -> 63[7] via P2P/IPC
+ip-26-0-173-7:811248:811415 [0] NCCL INFO Channel 03/0 : 56[0] -> 63[7] via P2P/IPC
+ip-26-0-173-7:811248:811415 [0] NCCL INFO Channel 04/0 : 56[0] -> 63[7] via P2P/IPC
+ip-26-0-173-7:811248:811415 [0] NCCL INFO Channel 09/0 : 56[0] -> 63[7] via P2P/IPC
+ip-26-0-173-7:811248:811415 [0] NCCL INFO Channel 10/0 : 56[0] -> 63[7] via P2P/IPC
+ip-26-0-173-7:811248:811415 [0] NCCL INFO Channel 11/0 : 56[0] -> 63[7] via P2P/IPC
+ip-26-0-173-7:811248:811415 [0] NCCL INFO Channel 12/0 : 56[0] -> 63[7] via P2P/IPC
+ip-26-0-172-73:926594:926762 [0] NCCL INFO Channel 05/0 : 48[0] -> 52[4] via P2P/IPC
+ip-26-0-172-73:926595:926760 [1] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-172-73:926598:926761 [4] NCCL INFO Channel 14/0 : 52[4] -> 55[7] via P2P/IPC
+ip-26-0-172-57:942611:942772 [0] NCCL INFO Channel 06/0 : 40[0] -> 45[5] via P2P/IPC
+ip-26-0-172-57:942615:942775 [4] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-172-57:942611:942772 [0] NCCL INFO Channel 14/0 : 40[0] -> 45[5] via P2P/IPC
+ip-26-0-172-57:942611:942772 [0] NCCL INFO Channel 07/0 : 40[0] -> 46[6] via P2P/IPC
+ip-26-0-172-57:942616:942777 [5] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-172-57:942611:942772 [0] NCCL INFO Channel 15/0 : 40[0] -> 46[6] via P2P/IPC
+ip-26-0-172-57:942611:942772 [0] NCCL INFO Channel 01/0 : 40[0] -> 47[7] via P2P/IPC
+ip-26-0-172-57:942617:942774 [6] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-172-57:942611:942772 [0] NCCL INFO Channel 02/0 : 40[0] -> 47[7] via P2P/IPC
+ip-26-0-172-57:942611:942772 [0] NCCL INFO Channel 03/0 : 40[0] -> 47[7] via P2P/IPC
+ip-26-0-172-57:942611:942772 [0] NCCL INFO Channel 04/0 : 40[0] -> 47[7] via P2P/IPC
+ip-26-0-172-57:942611:942772 [0] NCCL INFO Channel 09/0 : 40[0] -> 47[7] via P2P/IPC
+ip-26-0-172-57:942611:942772 [0] NCCL INFO Channel 10/0 : 40[0] -> 47[7] via P2P/IPC
+ip-26-0-172-57:942611:942772 [0] NCCL INFO Channel 11/0 : 40[0] -> 47[7] via P2P/IPC
+ip-26-0-172-57:942611:942772 [0] NCCL INFO Channel 12/0 : 40[0] -> 47[7] via P2P/IPC
+ip-26-0-172-57:942618:942778 [7] NCCL INFO Channel 06/0 : 47[7] -> 54[6] [send] via NET/Libfabric/6(46)/GDRDMA
+ip-26-0-160-225:94343:94508 [1] NCCL INFO Channel 01/0 : 58[2] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:94343:94508 [1] NCCL INFO Channel 09/0 : 58[2] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:94344:94510 [2] NCCL INFO Channel 01/0 : 2[2] -> 9[1] [send] via NET/Libfabric/1(1)/GDRDMA
+ip-26-0-160-225:94344:94510 [2] NCCL INFO Channel 09/0 : 2[2] -> 9[1] [send] via NET/Libfabric/1(1)/GDRDMA
+ip-26-0-172-142:880680:880843 [1] NCCL INFO Channel 01/0 : 10[2] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880680:880843 [1] NCCL INFO Channel 09/0 : 10[2] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880681:880845 [2] NCCL INFO Channel 01/0 : 18[2] -> 25[1] [send] via NET/Libfabric/1(17)/GDRDMA
+ip-26-0-172-142:880681:880845 [2] NCCL INFO Channel 09/0 : 18[2] -> 25[1] [send] via NET/Libfabric/1(17)/GDRDMA
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO Channel 01/0 : 2[2] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO Channel 09/0 : 2[2] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO Channel 01/0 : 10[2] -> 17[1] [send] via NET/Libfabric/1(9)/GDRDMA
+ip-26-0-173-7:811249:811413 [1] NCCL INFO Channel 01/0 : 50[2] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-7:811249:811413 [1] NCCL INFO Channel 09/0 : 50[2] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO Channel 09/0 : 10[2] -> 17[1] [send] via NET/Libfabric/1(9)/GDRDMA
+ip-26-0-173-7:811250:811412 [2] NCCL INFO Channel 01/0 : 58[2] -> 1[1] [send] via NET/Libfabric/1(57)/GDRDMA
+ip-26-0-172-252:688201:688364 [1] NCCL INFO Channel 01/0 : 26[2] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-7:811250:811412 [2] NCCL INFO Channel 09/0 : 58[2] -> 1[1] [send] via NET/Libfabric/1(57)/GDRDMA
+ip-26-0-172-252:688201:688364 [1] NCCL INFO Channel 09/0 : 26[2] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688202:688368 [2] NCCL INFO Channel 01/0 : 34[2] -> 41[1] [send] via NET/Libfabric/1(33)/GDRDMA
+ip-26-0-172-252:688202:688368 [2] NCCL INFO Channel 09/0 : 34[2] -> 41[1] [send] via NET/Libfabric/1(33)/GDRDMA
+ip-26-0-160-225:94346:94511 [4] NCCL INFO Channel 04/0 : 61[5] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:94346:94511 [4] NCCL INFO Channel 12/0 : 61[5] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:94348:94506 [6] NCCL INFO Channel 06/0 : 63[7] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618585:618748 [1] NCCL INFO Channel 01/0 : 18[2] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618585:618748 [1] NCCL INFO Channel 09/0 : 18[2] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:94349:94505 [7] NCCL INFO Channel 14/0 : 7[7] -> 14[6] [send] via NET/Libfabric/6(6)/GDRDMA
+ip-26-0-160-225:94346:94511 [4] NCCL INFO Channel 03/0 : 4[4] -> 11[3] [send] via NET/Libfabric/3(3)/GDRDMA
+ip-26-0-172-147:618586:618745 [2] NCCL INFO Channel 01/0 : 26[2] -> 33[1] [send] via NET/Libfabric/1(25)/GDRDMA
+ip-26-0-160-225:94348:94506 [6] NCCL INFO Channel 14/0 : 63[7] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618586:618745 [2] NCCL INFO Channel 09/0 : 26[2] -> 33[1] [send] via NET/Libfabric/1(25)/GDRDMA
+ip-26-0-160-225:94346:94511 [4] NCCL INFO Channel 11/0 : 4[4] -> 11[3] [send] via NET/Libfabric/3(3)/GDRDMA
+ip-26-0-172-147:618588:618750 [4] NCCL INFO Channel 04/0 : 21[5] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:94347:94507 [5] NCCL INFO Channel 05/0 : 62[6] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:94348:94506 [6] NCCL INFO Channel 05/0 : 6[6] -> 13[5] [send] via NET/Libfabric/5(5)/GDRDMA
+ip-26-0-172-147:618588:618750 [4] NCCL INFO Channel 12/0 : 21[5] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:94347:94507 [5] NCCL INFO Channel 13/0 : 62[6] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:94348:94506 [6] NCCL INFO Channel 13/0 : 6[6] -> 13[5] [send] via NET/Libfabric/5(5)/GDRDMA
+ip-26-0-160-225:94347:94507 [5] NCCL INFO Channel 04/0 : 5[5] -> 12[4] [send] via NET/Libfabric/4(4)/GDRDMA
+ip-26-0-172-147:618588:618750 [4] NCCL INFO Channel 03/0 : 28[4] -> 35[3] [send] via NET/Libfabric/3(27)/GDRDMA
+ip-26-0-160-225:94347:94507 [5] NCCL INFO Channel 12/0 : 5[5] -> 12[4] [send] via NET/Libfabric/4(4)/GDRDMA
+ip-26-0-172-147:618588:618750 [4] NCCL INFO Channel 11/0 : 28[4] -> 35[3] [send] via NET/Libfabric/3(27)/GDRDMA
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO Channel 04/0 : 5[5] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO Channel 12/0 : 5[5] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO Channel 03/0 : 12[4] -> 19[3] [send] via NET/Libfabric/3(11)/GDRDMA
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO Channel 06/0 : 7[7] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO Channel 11/0 : 12[4] -> 19[3] [send] via NET/Libfabric/3(11)/GDRDMA
+ip-26-0-172-142:880683:880844 [4] NCCL INFO Channel 04/0 : 13[5] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-7:811255:811411 [7] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO Channel 06/0 : 15[7] -> 22[6] [send] via NET/Libfabric/6(14)/GDRDMA
+ip-26-0-172-142:880683:880844 [4] NCCL INFO Channel 12/0 : 13[5] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO Channel 14/0 : 7[7] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO Channel 14/0 : 15[7] -> 22[6] [send] via NET/Libfabric/6(14)/GDRDMA
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO Channel 01/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-172-142:880683:880844 [4] NCCL INFO Channel 03/0 : 20[4] -> 27[3] [send] via NET/Libfabric/3(19)/GDRDMA
+ip-26-0-172-142:880683:880844 [4] NCCL INFO Channel 11/0 : 20[4] -> 27[3] [send] via NET/Libfabric/3(19)/GDRDMA
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO Channel 05/0 : 6[6] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:94344:94510 [2] NCCL INFO Channel 02/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO Channel 13/0 : 6[6] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO Channel 05/0 : 14[6] -> 21[5] [send] via NET/Libfabric/5(13)/GDRDMA
+ip-26-0-172-142:880685:880847 [6] NCCL INFO Channel 06/0 : 15[7] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO Channel 02/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-172-142:880685:880847 [6] NCCL INFO Channel 14/0 : 15[7] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO Channel 13/0 : 14[6] -> 21[5] [send] via NET/Libfabric/5(13)/GDRDMA
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO Channel 04/0 : 13[5] -> 20[4] [send] via NET/Libfabric/4(12)/GDRDMA
+ip-26-0-172-142:880686:880841 [7] NCCL INFO Channel 14/0 : 23[7] -> 30[6] [send] via NET/Libfabric/6(22)/GDRDMA
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO Channel 02/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-172-57:942612:942779 [1] NCCL INFO Channel 01/0 : 34[2] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO Channel 12/0 : 13[5] -> 20[4] [send] via NET/Libfabric/4(12)/GDRDMA
+ip-26-0-160-225:94343:94508 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-225:94344:94510 [2] NCCL INFO Channel 03/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO Channel 03/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-160-225:94349:94505 [7] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-172-57:942613:942776 [2] NCCL INFO Channel 01/0 : 42[2] -> 49[1] [send] via NET/Libfabric/1(41)/GDRDMA
+ip-26-0-172-57:942612:942779 [1] NCCL INFO Channel 09/0 : 34[2] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880684:880842 [5] NCCL INFO Channel 05/0 : 14[6] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942613:942776 [2] NCCL INFO Channel 09/0 : 42[2] -> 49[1] [send] via NET/Libfabric/1(41)/GDRDMA
+ip-26-0-160-225:94343:94508 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO Channel 03/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-172-252:688201:688364 [1] NCCL INFO Channel 01/0 : 33[1] -> 32[0] via P2P/IPC
+ip-26-0-172-142:880684:880842 [5] NCCL INFO Channel 13/0 : 14[6] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:94348:94506 [6] NCCL INFO Channel 06/0 : 6[6] -> 3[3] via P2P/IPC
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO Channel 04/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-160-225:94344:94510 [2] NCCL INFO Channel 04/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-142:880680:880843 [1] NCCL INFO Channel 01/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-172-142:880685:880847 [6] NCCL INFO Channel 05/0 : 22[6] -> 29[5] [send] via NET/Libfabric/5(21)/GDRDMA
+ip-26-0-172-252:688204:688367 [4] NCCL INFO Channel 04/0 : 29[5] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO Channel 05/0 : 13[5] -> 11[3] via P2P/IPC
+ip-26-0-172-142:880685:880847 [6] NCCL INFO Channel 13/0 : 22[6] -> 29[5] [send] via NET/Libfabric/5(21)/GDRDMA
+ip-26-0-172-142:880684:880842 [5] NCCL INFO Channel 04/0 : 21[5] -> 28[4] [send] via NET/Libfabric/4(20)/GDRDMA
+ip-26-0-172-252:688204:688367 [4] NCCL INFO Channel 12/0 : 29[5] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:94343:94508 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-142:880684:880842 [5] NCCL INFO Channel 12/0 : 21[5] -> 28[4] [send] via NET/Libfabric/4(20)/GDRDMA
+ip-26-0-172-252:688201:688364 [1] NCCL INFO Channel 02/0 : 33[1] -> 32[0] via P2P/IPC
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO Channel 04/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO Channel 05/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-160-225:94344:94510 [2] NCCL INFO Channel 05/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-142:880680:880843 [1] NCCL INFO Channel 02/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-160-225:94343:94508 [1] NCCL INFO Channel 04/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-225:94348:94506 [6] NCCL INFO Channel 14/0 : 6[6] -> 3[3] via P2P/IPC
+ip-26-0-172-252:688201:688364 [1] NCCL INFO Channel 03/0 : 33[1] -> 32[0] via P2P/IPC
+ip-26-0-160-225:94344:94510 [2] NCCL INFO Channel 06/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-252:688205:688365 [5] NCCL INFO Channel 05/0 : 30[6] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688205:688365 [5] NCCL INFO Channel 13/0 : 30[6] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO Channel 05/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-172-142:880680:880843 [1] NCCL INFO Channel 03/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-172-147:618586:618745 [2] NCCL INFO Channel 02/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-172-252:688201:688364 [1] NCCL INFO Channel 04/0 : 33[1] -> 32[0] via P2P/IPC
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO Channel 13/0 : 13[5] -> 11[3] via P2P/IPC
+ip-26-0-160-225:94343:94508 [1] NCCL INFO Channel 05/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO Channel 06/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-172-147:618589:618744 [5] NCCL INFO Channel 05/0 : 22[6] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO Channel 07/0 : 12[4] -> 23[7] [send] via NET/Libfabric/7(15)/GDRDMA
+ip-26-0-160-225:94344:94510 [2] NCCL INFO Channel 07/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-147:618585:618748 [1] NCCL INFO Channel 01/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-172-252:688205:688365 [5] NCCL INFO Channel 04/0 : 37[5] -> 44[4] [send] via NET/Libfabric/4(36)/GDRDMA
+ip-26-0-172-147:618590:618749 [6] NCCL INFO Channel 06/0 : 23[7] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618591:618746 [7] NCCL INFO Channel 06/0 : 31[7] -> 38[6] [send] via NET/Libfabric/6(30)/GDRDMA
+ip-26-0-172-252:688205:688365 [5] NCCL INFO Channel 12/0 : 37[5] -> 44[4] [send] via NET/Libfabric/4(36)/GDRDMA
+ip-26-0-172-147:618586:618745 [2] NCCL INFO Channel 03/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-172-147:618589:618744 [5] NCCL INFO Channel 13/0 : 22[6] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:94343:94508 [1] NCCL INFO Channel 06/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-147:618591:618746 [7] NCCL INFO Channel 14/0 : 31[7] -> 38[6] [send] via NET/Libfabric/6(30)/GDRDMA
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO Channel 06/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-172-142:880680:880843 [1] NCCL INFO Channel 04/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-172-252:688201:688364 [1] NCCL INFO Channel 05/0 : 33[1] -> 32[0] via P2P/IPC
+ip-26-0-172-147:618589:618744 [5] NCCL INFO Channel 04/0 : 29[5] -> 36[4] [send] via NET/Libfabric/4(28)/GDRDMA
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO Channel 07/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-160-225:94344:94510 [2] NCCL INFO Channel 10/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-147:618590:618749 [6] NCCL INFO Channel 14/0 : 23[7] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618585:618748 [1] NCCL INFO Channel 02/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-172-252:688204:688367 [4] NCCL INFO Channel 03/0 : 36[4] -> 43[3] [send] via NET/Libfabric/3(35)/GDRDMA
+ip-26-0-172-147:618589:618744 [5] NCCL INFO Channel 12/0 : 29[5] -> 36[4] [send] via NET/Libfabric/4(28)/GDRDMA
+ip-26-0-172-147:618586:618745 [2] NCCL INFO Channel 04/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO Channel 07/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-172-252:688204:688367 [4] NCCL INFO Channel 11/0 : 36[4] -> 43[3] [send] via NET/Libfabric/3(35)/GDRDMA
+ip-26-0-172-252:688201:688364 [1] NCCL INFO Channel 06/0 : 33[1] -> 32[0] via P2P/IPC
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO Channel 09/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-160-225:94343:94508 [1] NCCL INFO Channel 07/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-147:618585:618748 [1] NCCL INFO Channel 03/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-172-142:880681:880845 [2] NCCL INFO Channel 02/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-172-147:618590:618749 [6] NCCL INFO Channel 05/0 : 30[6] -> 37[5] [send] via NET/Libfabric/5(29)/GDRDMA
+ip-26-0-160-225:94344:94510 [2] NCCL INFO Channel 11/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-142:880680:880843 [1] NCCL INFO Channel 05/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO Channel 06/0 : 14[6] -> 11[3] via P2P/IPC
+ip-26-0-172-147:618590:618749 [6] NCCL INFO Channel 13/0 : 30[6] -> 37[5] [send] via NET/Libfabric/5(29)/GDRDMA
+ip-26-0-172-147:618586:618745 [2] NCCL INFO Channel 05/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-172-142:880685:880847 [6] NCCL INFO Channel 06/0 : 22[6] -> 19[3] via P2P/IPC
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO Channel 10/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-172-252:688206:688363 [6] NCCL INFO Channel 06/0 : 31[7] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880683:880844 [4] NCCL INFO Channel 07/0 : 20[4] -> 31[7] [send] via NET/Libfabric/7(23)/GDRDMA
+ip-26-0-172-142:880684:880842 [5] NCCL INFO Channel 05/0 : 21[5] -> 19[3] via P2P/IPC
+ip-26-0-172-147:618589:618744 [5] NCCL INFO Channel 05/0 : 29[5] -> 27[3] via P2P/IPC
+ip-26-0-172-142:880683:880844 [4] NCCL INFO Channel 15/0 : 20[4] -> 31[7] [send] via NET/Libfabric/7(23)/GDRDMA
+ip-26-0-172-252:688201:688364 [1] NCCL INFO Channel 07/0 : 33[1] -> 32[0] via P2P/IPC
+ip-26-0-172-147:618585:618748 [1] NCCL INFO Channel 04/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-172-252:688206:688363 [6] NCCL INFO Channel 14/0 : 31[7] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:94343:94508 [1] NCCL INFO Channel 09/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-142:880686:880841 [7] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO Channel 10/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-160-225:94344:94510 [2] NCCL INFO Channel 12/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-147:618586:618745 [2] NCCL INFO Channel 06/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-172-252:688206:688363 [6] NCCL INFO Channel 05/0 : 38[6] -> 45[5] [send] via NET/Libfabric/5(37)/GDRDMA
+ip-26-0-172-252:688201:688364 [1] NCCL INFO Channel 09/0 : 33[1] -> 32[0] via P2P/IPC
+ip-26-0-172-252:688206:688363 [6] NCCL INFO Channel 13/0 : 38[6] -> 45[5] [send] via NET/Libfabric/5(37)/GDRDMA
+ip-26-0-172-142:880681:880845 [2] NCCL INFO Channel 03/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-172-147:618590:618749 [6] NCCL INFO Channel 06/0 : 30[6] -> 27[3] via P2P/IPC
+ip-26-0-160-225:94343:94508 [1] NCCL INFO Channel 10/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-147:618585:618748 [1] NCCL INFO Channel 05/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-172-142:880680:880843 [1] NCCL INFO Channel 06/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-160-225:94344:94510 [2] NCCL INFO Channel 13/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-147:618589:618744 [5] NCCL INFO Channel 13/0 : 29[5] -> 27[3] via P2P/IPC
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO Channel 11/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-172-147:618586:618745 [2] NCCL INFO Channel 07/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-172-142:880685:880847 [6] NCCL INFO Channel 14/0 : 22[6] -> 19[3] via P2P/IPC
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO Channel 14/0 : 14[6] -> 11[3] via P2P/IPC
+ip-26-0-172-252:688201:688364 [1] NCCL INFO Channel 10/0 : 33[1] -> 32[0] via P2P/IPC
+ip-26-0-172-147:618591:618746 [7] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO Channel 11/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-160-225:94343:94508 [1] NCCL INFO Channel 11/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-252:688201:688364 [1] NCCL INFO Channel 11/0 : 33[1] -> 32[0] via P2P/IPC
+ip-26-0-172-147:618585:618748 [1] NCCL INFO Channel 06/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-160-225:94344:94510 [2] NCCL INFO Channel 14/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO Channel 12/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-172-142:880684:880842 [5] NCCL INFO Channel 13/0 : 21[5] -> 19[3] via P2P/IPC
+ip-26-0-172-142:880681:880845 [2] NCCL INFO Channel 04/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-160-225:94343:94508 [1] NCCL INFO Channel 12/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-252:688201:688364 [1] NCCL INFO Channel 12/0 : 33[1] -> 32[0] via P2P/IPC
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO Channel 12/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-172-142:880680:880843 [1] NCCL INFO Channel 07/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-160-225:94344:94510 [2] NCCL INFO Channel 15/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-147:618586:618745 [2] NCCL INFO Channel 10/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO Channel 13/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-172-252:688201:688364 [1] NCCL INFO Channel 13/0 : 33[1] -> 32[0] via P2P/IPC
+ip-26-0-160-225:94343:94508 [1] NCCL INFO Channel 13/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-147:618585:618748 [1] NCCL INFO Channel 07/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-172-57:942612:942779 [1] NCCL INFO Channel 01/0 : 41[1] -> 40[0] via P2P/IPC
+ip-26-0-172-147:618590:618749 [6] NCCL INFO Channel 14/0 : 30[6] -> 27[3] via P2P/IPC
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO Channel 13/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-172-142:880681:880845 [2] NCCL INFO Channel 05/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-172-147:618588:618750 [4] NCCL INFO Channel 07/0 : 28[4] -> 39[7] [send] via NET/Libfabric/7(31)/GDRDMA
+ip-26-0-172-252:688201:688364 [1] NCCL INFO Channel 14/0 : 33[1] -> 32[0] via P2P/IPC
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO Channel 14/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-172-57:942612:942779 [1] NCCL INFO Channel 02/0 : 41[1] -> 40[0] via P2P/IPC
+ip-26-0-172-142:880680:880843 [1] NCCL INFO Channel 09/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-160-225:94343:94508 [1] NCCL INFO Channel 14/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-147:618586:618745 [2] NCCL INFO Channel 11/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO Channel 14/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-172-252:688201:688364 [1] NCCL INFO Channel 15/0 : 33[1] -> 32[0] via P2P/IPC
+ip-26-0-172-57:942612:942779 [1] NCCL INFO Channel 03/0 : 41[1] -> 40[0] via P2P/IPC
+ip-26-0-172-142:880681:880845 [2] NCCL INFO Channel 06/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-172-147:618585:618748 [1] NCCL INFO Channel 09/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-172-252:688204:688367 [4] NCCL INFO Channel 07/0 : 36[4] -> 47[7] [send] via NET/Libfabric/7(39)/GDRDMA
+ip-26-0-160-225:94343:94508 [1] NCCL INFO Channel 15/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-252:688204:688367 [4] NCCL INFO Channel 15/0 : 36[4] -> 47[7] [send] via NET/Libfabric/7(39)/GDRDMA
+ip-26-0-172-57:942612:942779 [1] NCCL INFO Channel 04/0 : 41[1] -> 40[0] via P2P/IPC
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO Channel 15/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-172-142:880680:880843 [1] NCCL INFO Channel 10/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-172-252:688202:688368 [2] NCCL INFO Channel 02/0 : 34[2] -> 33[1] via P2P/IPC
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO Channel 15/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-172-142:880681:880845 [2] NCCL INFO Channel 07/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-172-147:618586:618745 [2] NCCL INFO Channel 12/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-172-57:942612:942779 [1] NCCL INFO Channel 05/0 : 41[1] -> 40[0] via P2P/IPC
+ip-26-0-172-147:618585:618748 [1] NCCL INFO Channel 10/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-172-142:880680:880843 [1] NCCL INFO Channel 11/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-172-57:942612:942779 [1] NCCL INFO Channel 06/0 : 41[1] -> 40[0] via P2P/IPC
+ip-26-0-172-142:880681:880845 [2] NCCL INFO Channel 10/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-172-57:942612:942779 [1] NCCL INFO Channel 07/0 : 41[1] -> 40[0] via P2P/IPC
+ip-26-0-172-142:880680:880843 [1] NCCL INFO Channel 12/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-172-147:618586:618745 [2] NCCL INFO Channel 13/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-172-147:618585:618748 [1] NCCL INFO Channel 11/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-172-57:942612:942779 [1] NCCL INFO Channel 09/0 : 41[1] -> 40[0] via P2P/IPC
+ip-26-0-172-142:880681:880845 [2] NCCL INFO Channel 11/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-172-147:618586:618745 [2] NCCL INFO Channel 14/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-172-142:880680:880843 [1] NCCL INFO Channel 13/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-172-57:942612:942779 [1] NCCL INFO Channel 10/0 : 41[1] -> 40[0] via P2P/IPC
+ip-26-0-172-147:618585:618748 [1] NCCL INFO Channel 12/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-172-57:942612:942779 [1] NCCL INFO Channel 11/0 : 41[1] -> 40[0] via P2P/IPC
+ip-26-0-172-142:880681:880845 [2] NCCL INFO Channel 12/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-172-147:618586:618745 [2] NCCL INFO Channel 15/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-172-142:880680:880843 [1] NCCL INFO Channel 14/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-172-57:942612:942779 [1] NCCL INFO Channel 12/0 : 41[1] -> 40[0] via P2P/IPC
+ip-26-0-172-147:618585:618748 [1] NCCL INFO Channel 13/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-172-142:880681:880845 [2] NCCL INFO Channel 13/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-172-57:942612:942779 [1] NCCL INFO Channel 13/0 : 41[1] -> 40[0] via P2P/IPC
+ip-26-0-172-142:880680:880843 [1] NCCL INFO Channel 15/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-172-57:942612:942779 [1] NCCL INFO Channel 14/0 : 41[1] -> 40[0] via P2P/IPC
+ip-26-0-172-147:618585:618748 [1] NCCL INFO Channel 14/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-172-142:880681:880845 [2] NCCL INFO Channel 14/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-172-57:942612:942779 [1] NCCL INFO Channel 15/0 : 41[1] -> 40[0] via P2P/IPC
+ip-26-0-172-147:618585:618748 [1] NCCL INFO Channel 15/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-172-142:880681:880845 [2] NCCL INFO Channel 15/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-172-73:926594:926762 [0] NCCL INFO Channel 13/0 : 48[0] -> 52[4] via P2P/IPC
+ip-26-0-172-73:926597:926758 [3] NCCL INFO Channel 03/0 : 44[4] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926597:926758 [3] NCCL INFO Channel 11/0 : 44[4] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688202:688368 [2] NCCL INFO Channel 03/0 : 34[2] -> 33[1] via P2P/IPC
+ip-26-0-172-73:926596:926756 [2] NCCL INFO Channel 02/0 : 43[3] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926596:926756 [2] NCCL INFO Channel 10/0 : 43[3] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926597:926758 [3] NCCL INFO Channel 02/0 : 51[3] -> 58[2] [send] via NET/Libfabric/2(50)/GDRDMA
+ip-26-0-172-252:688202:688368 [2] NCCL INFO Channel 04/0 : 34[2] -> 33[1] via P2P/IPC
+ip-26-0-172-73:926597:926758 [3] NCCL INFO Channel 10/0 : 51[3] -> 58[2] [send] via NET/Libfabric/2(50)/GDRDMA
+ip-26-0-172-252:688202:688368 [2] NCCL INFO Channel 05/0 : 34[2] -> 33[1] via P2P/IPC
+ip-26-0-172-252:688202:688368 [2] NCCL INFO Channel 06/0 : 34[2] -> 33[1] via P2P/IPC
+ip-26-0-172-252:688202:688368 [2] NCCL INFO Channel 07/0 : 34[2] -> 33[1] via P2P/IPC
+ip-26-0-172-252:688202:688368 [2] NCCL INFO Channel 10/0 : 34[2] -> 33[1] via P2P/IPC
+ip-26-0-172-252:688202:688368 [2] NCCL INFO Channel 11/0 : 34[2] -> 33[1] via P2P/IPC
+ip-26-0-173-7:811250:811412 [2] NCCL INFO Channel 02/0 : 58[2] -> 57[1] via P2P/IPC
+ip-26-0-173-7:811250:811412 [2] NCCL INFO Channel 03/0 : 58[2] -> 57[1] via P2P/IPC
+ip-26-0-172-252:688202:688368 [2] NCCL INFO Channel 12/0 : 34[2] -> 33[1] via P2P/IPC
+ip-26-0-173-7:811252:811408 [4] NCCL INFO Channel 04/0 : 53[5] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-7:811250:811412 [2] NCCL INFO Channel 04/0 : 58[2] -> 57[1] via P2P/IPC
+ip-26-0-173-7:811252:811408 [4] NCCL INFO Channel 12/0 : 53[5] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-7:811250:811412 [2] NCCL INFO Channel 05/0 : 58[2] -> 57[1] via P2P/IPC
+ip-26-0-173-7:811252:811408 [4] NCCL INFO Channel 03/0 : 60[4] -> 3[3] [send] via NET/Libfabric/3(59)/GDRDMA
+ip-26-0-173-7:811250:811412 [2] NCCL INFO Channel 06/0 : 58[2] -> 57[1] via P2P/IPC
+ip-26-0-173-7:811252:811408 [4] NCCL INFO Channel 11/0 : 60[4] -> 3[3] [send] via NET/Libfabric/3(59)/GDRDMA
+ip-26-0-173-7:811250:811412 [2] NCCL INFO Channel 07/0 : 58[2] -> 57[1] via P2P/IPC
+ip-26-0-173-7:811250:811412 [2] NCCL INFO Channel 10/0 : 58[2] -> 57[1] via P2P/IPC
+ip-26-0-172-252:688202:688368 [2] NCCL INFO Channel 13/0 : 34[2] -> 33[1] via P2P/IPC
+ip-26-0-173-7:811254:811414 [6] NCCL INFO Channel 06/0 : 55[7] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688202:688368 [2] NCCL INFO Channel 14/0 : 34[2] -> 33[1] via P2P/IPC
+ip-26-0-173-7:811254:811414 [6] NCCL INFO Channel 14/0 : 55[7] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-7:811253:811410 [5] NCCL INFO Channel 05/0 : 54[6] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-7:811253:811410 [5] NCCL INFO Channel 13/0 : 54[6] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-7:811254:811414 [6] NCCL INFO Channel 05/0 : 62[6] -> 5[5] [send] via NET/Libfabric/5(61)/GDRDMA
+ip-26-0-173-7:811254:811414 [6] NCCL INFO Channel 13/0 : 62[6] -> 5[5] [send] via NET/Libfabric/5(61)/GDRDMA
+ip-26-0-173-7:811253:811410 [5] NCCL INFO Channel 04/0 : 61[5] -> 4[4] [send] via NET/Libfabric/4(60)/GDRDMA
+ip-26-0-173-7:811253:811410 [5] NCCL INFO Channel 12/0 : 61[5] -> 4[4] [send] via NET/Libfabric/4(60)/GDRDMA
+ip-26-0-173-7:811250:811412 [2] NCCL INFO Channel 11/0 : 58[2] -> 57[1] via P2P/IPC
+ip-26-0-160-225:94347:94507 [5] NCCL INFO Channel 05/0 : 5[5] -> 3[3] via P2P/IPC
+ip-26-0-173-7:811250:811412 [2] NCCL INFO Channel 12/0 : 58[2] -> 57[1] via P2P/IPC
+ip-26-0-173-7:811250:811412 [2] NCCL INFO Channel 13/0 : 58[2] -> 57[1] via P2P/IPC
+ip-26-0-173-7:811250:811412 [2] NCCL INFO Channel 14/0 : 58[2] -> 57[1] via P2P/IPC
+ip-26-0-173-7:811250:811412 [2] NCCL INFO Channel 15/0 : 58[2] -> 57[1] via P2P/IPC
+ip-26-0-172-252:688202:688368 [2] NCCL INFO Channel 15/0 : 34[2] -> 33[1] via P2P/IPC
+ip-26-0-160-225:94347:94507 [5] NCCL INFO Channel 13/0 : 5[5] -> 3[3] via P2P/IPC
+ip-26-0-172-73:926598:926761 [4] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-172-73:926594:926762 [0] NCCL INFO Channel 06/0 : 48[0] -> 53[5] via P2P/IPC
+ip-26-0-172-73:926594:926762 [0] NCCL INFO Channel 14/0 : 48[0] -> 53[5] via P2P/IPC
+ip-26-0-172-73:926594:926762 [0] NCCL INFO Channel 07/0 : 48[0] -> 54[6] via P2P/IPC
+ip-26-0-172-73:926599:926755 [5] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-172-73:926594:926762 [0] NCCL INFO Channel 15/0 : 48[0] -> 54[6] via P2P/IPC
+ip-26-0-172-73:926594:926762 [0] NCCL INFO Channel 01/0 : 48[0] -> 55[7] via P2P/IPC
+ip-26-0-172-73:926600:926757 [6] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-172-73:926594:926762 [0] NCCL INFO Channel 02/0 : 48[0] -> 55[7] via P2P/IPC
+ip-26-0-172-73:926594:926762 [0] NCCL INFO Channel 03/0 : 48[0] -> 55[7] via P2P/IPC
+ip-26-0-172-73:926594:926762 [0] NCCL INFO Channel 04/0 : 48[0] -> 55[7] via P2P/IPC
+ip-26-0-172-73:926594:926762 [0] NCCL INFO Channel 09/0 : 48[0] -> 55[7] via P2P/IPC
+ip-26-0-172-73:926594:926762 [0] NCCL INFO Channel 10/0 : 48[0] -> 55[7] via P2P/IPC
+ip-26-0-172-73:926594:926762 [0] NCCL INFO Channel 11/0 : 48[0] -> 55[7] via P2P/IPC
+ip-26-0-172-73:926594:926762 [0] NCCL INFO Channel 12/0 : 48[0] -> 55[7] via P2P/IPC
+ip-26-0-172-73:926595:926760 [1] NCCL INFO Channel 01/0 : 42[2] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926595:926760 [1] NCCL INFO Channel 09/0 : 42[2] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926596:926756 [2] NCCL INFO Channel 01/0 : 50[2] -> 57[1] [send] via NET/Libfabric/1(49)/GDRDMA
+ip-26-0-172-73:926596:926756 [2] NCCL INFO Channel 09/0 : 50[2] -> 57[1] [send] via NET/Libfabric/1(49)/GDRDMA
+ip-26-0-172-57:942615:942775 [4] NCCL INFO Channel 04/0 : 37[5] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942615:942775 [4] NCCL INFO Channel 12/0 : 37[5] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942615:942775 [4] NCCL INFO Channel 03/0 : 44[4] -> 51[3] [send] via NET/Libfabric/3(43)/GDRDMA
+ip-26-0-172-57:942615:942775 [4] NCCL INFO Channel 11/0 : 44[4] -> 51[3] [send] via NET/Libfabric/3(43)/GDRDMA
+ip-26-0-172-252:688205:688365 [5] NCCL INFO Channel 05/0 : 37[5] -> 35[3] via P2P/IPC
+ip-26-0-172-57:942617:942774 [6] NCCL INFO Channel 06/0 : 39[7] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942617:942774 [6] NCCL INFO Channel 14/0 : 39[7] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942618:942778 [7] NCCL INFO Channel 14/0 : 47[7] -> 54[6] [send] via NET/Libfabric/6(46)/GDRDMA
+ip-26-0-172-57:942616:942777 [5] NCCL INFO Channel 05/0 : 38[6] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942616:942777 [5] NCCL INFO Channel 13/0 : 38[6] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942617:942774 [6] NCCL INFO Channel 05/0 : 46[6] -> 53[5] [send] via NET/Libfabric/5(45)/GDRDMA
+ip-26-0-172-57:942617:942774 [6] NCCL INFO Channel 13/0 : 46[6] -> 53[5] [send] via NET/Libfabric/5(45)/GDRDMA
+ip-26-0-172-73:926595:926760 [1] NCCL INFO Channel 01/0 : 49[1] -> 48[0] via P2P/IPC
+ip-26-0-172-57:942616:942777 [5] NCCL INFO Channel 04/0 : 45[5] -> 52[4] [send] via NET/Libfabric/4(44)/GDRDMA
+ip-26-0-172-57:942616:942777 [5] NCCL INFO Channel 12/0 : 45[5] -> 52[4] [send] via NET/Libfabric/4(44)/GDRDMA
+ip-26-0-172-252:688207:688366 [7] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-172-73:926595:926760 [1] NCCL INFO Channel 02/0 : 49[1] -> 48[0] via P2P/IPC
+ip-26-0-172-252:688206:688363 [6] NCCL INFO Channel 06/0 : 38[6] -> 35[3] via P2P/IPC
+ip-26-0-172-73:926595:926760 [1] NCCL INFO Channel 03/0 : 49[1] -> 48[0] via P2P/IPC
+ip-26-0-172-57:942615:942775 [4] NCCL INFO Channel 07/0 : 44[4] -> 55[7] [send] via NET/Libfabric/7(47)/GDRDMA
+ip-26-0-172-57:942613:942776 [2] NCCL INFO Channel 02/0 : 42[2] -> 41[1] via P2P/IPC
+ip-26-0-172-57:942615:942775 [4] NCCL INFO Channel 15/0 : 44[4] -> 55[7] [send] via NET/Libfabric/7(47)/GDRDMA
+ip-26-0-172-73:926596:926756 [2] NCCL INFO Channel 02/0 : 50[2] -> 49[1] via P2P/IPC
+ip-26-0-172-73:926595:926760 [1] NCCL INFO Channel 04/0 : 49[1] -> 48[0] via P2P/IPC
+ip-26-0-172-57:942613:942776 [2] NCCL INFO Channel 03/0 : 42[2] -> 41[1] via P2P/IPC
+ip-26-0-172-73:926596:926756 [2] NCCL INFO Channel 03/0 : 50[2] -> 49[1] via P2P/IPC
+ip-26-0-172-73:926595:926760 [1] NCCL INFO Channel 05/0 : 49[1] -> 48[0] via P2P/IPC
+ip-26-0-173-7:811249:811413 [1] NCCL INFO Channel 01/0 : 57[1] -> 56[0] via P2P/IPC
+ip-26-0-172-57:942613:942776 [2] NCCL INFO Channel 04/0 : 42[2] -> 41[1] via P2P/IPC
+ip-26-0-172-73:926596:926756 [2] NCCL INFO Channel 04/0 : 50[2] -> 49[1] via P2P/IPC
+ip-26-0-172-73:926595:926760 [1] NCCL INFO Channel 06/0 : 49[1] -> 48[0] via P2P/IPC
+ip-26-0-173-7:811249:811413 [1] NCCL INFO Channel 02/0 : 57[1] -> 56[0] via P2P/IPC
+ip-26-0-172-57:942613:942776 [2] NCCL INFO Channel 05/0 : 42[2] -> 41[1] via P2P/IPC
+ip-26-0-172-73:926596:926756 [2] NCCL INFO Channel 05/0 : 50[2] -> 49[1] via P2P/IPC
+ip-26-0-172-73:926595:926760 [1] NCCL INFO Channel 07/0 : 49[1] -> 48[0] via P2P/IPC
+ip-26-0-173-7:811249:811413 [1] NCCL INFO Channel 03/0 : 57[1] -> 56[0] via P2P/IPC
+ip-26-0-172-57:942613:942776 [2] NCCL INFO Channel 06/0 : 42[2] -> 41[1] via P2P/IPC
+ip-26-0-172-73:926596:926756 [2] NCCL INFO Channel 06/0 : 50[2] -> 49[1] via P2P/IPC
+ip-26-0-172-73:926595:926760 [1] NCCL INFO Channel 09/0 : 49[1] -> 48[0] via P2P/IPC
+ip-26-0-173-7:811249:811413 [1] NCCL INFO Channel 04/0 : 57[1] -> 56[0] via P2P/IPC
+ip-26-0-172-57:942613:942776 [2] NCCL INFO Channel 07/0 : 42[2] -> 41[1] via P2P/IPC
+ip-26-0-172-73:926596:926756 [2] NCCL INFO Channel 07/0 : 50[2] -> 49[1] via P2P/IPC
+ip-26-0-173-7:811249:811413 [1] NCCL INFO Channel 05/0 : 57[1] -> 56[0] via P2P/IPC
+ip-26-0-172-73:926595:926760 [1] NCCL INFO Channel 10/0 : 49[1] -> 48[0] via P2P/IPC
+ip-26-0-172-57:942613:942776 [2] NCCL INFO Channel 10/0 : 42[2] -> 41[1] via P2P/IPC
+ip-26-0-173-7:811249:811413 [1] NCCL INFO Channel 06/0 : 57[1] -> 56[0] via P2P/IPC
+ip-26-0-172-73:926596:926756 [2] NCCL INFO Channel 10/0 : 50[2] -> 49[1] via P2P/IPC
+ip-26-0-172-73:926595:926760 [1] NCCL INFO Channel 11/0 : 49[1] -> 48[0] via P2P/IPC
+ip-26-0-173-7:811249:811413 [1] NCCL INFO Channel 07/0 : 57[1] -> 56[0] via P2P/IPC
+ip-26-0-172-57:942613:942776 [2] NCCL INFO Channel 11/0 : 42[2] -> 41[1] via P2P/IPC
+ip-26-0-172-73:926596:926756 [2] NCCL INFO Channel 11/0 : 50[2] -> 49[1] via P2P/IPC
+ip-26-0-172-73:926595:926760 [1] NCCL INFO Channel 12/0 : 49[1] -> 48[0] via P2P/IPC
+ip-26-0-173-7:811249:811413 [1] NCCL INFO Channel 09/0 : 57[1] -> 56[0] via P2P/IPC
+ip-26-0-172-73:926596:926756 [2] NCCL INFO Channel 12/0 : 50[2] -> 49[1] via P2P/IPC
+ip-26-0-172-57:942613:942776 [2] NCCL INFO Channel 12/0 : 42[2] -> 41[1] via P2P/IPC
+ip-26-0-172-73:926595:926760 [1] NCCL INFO Channel 13/0 : 49[1] -> 48[0] via P2P/IPC
+ip-26-0-173-7:811249:811413 [1] NCCL INFO Channel 10/0 : 57[1] -> 56[0] via P2P/IPC
+ip-26-0-172-57:942613:942776 [2] NCCL INFO Channel 13/0 : 42[2] -> 41[1] via P2P/IPC
+ip-26-0-172-73:926596:926756 [2] NCCL INFO Channel 13/0 : 50[2] -> 49[1] via P2P/IPC
+ip-26-0-173-7:811249:811413 [1] NCCL INFO Channel 11/0 : 57[1] -> 56[0] via P2P/IPC
+ip-26-0-172-73:926595:926760 [1] NCCL INFO Channel 14/0 : 49[1] -> 48[0] via P2P/IPC
+ip-26-0-173-7:811249:811413 [1] NCCL INFO Channel 12/0 : 57[1] -> 56[0] via P2P/IPC
+ip-26-0-172-73:926596:926756 [2] NCCL INFO Channel 14/0 : 50[2] -> 49[1] via P2P/IPC
+ip-26-0-172-252:688205:688365 [5] NCCL INFO Channel 13/0 : 37[5] -> 35[3] via P2P/IPC
+ip-26-0-172-57:942613:942776 [2] NCCL INFO Channel 14/0 : 42[2] -> 41[1] via P2P/IPC
+ip-26-0-172-73:926595:926760 [1] NCCL INFO Channel 15/0 : 49[1] -> 48[0] via P2P/IPC
+ip-26-0-173-7:811249:811413 [1] NCCL INFO Channel 13/0 : 57[1] -> 56[0] via P2P/IPC
+ip-26-0-172-73:926596:926756 [2] NCCL INFO Channel 15/0 : 50[2] -> 49[1] via P2P/IPC
+ip-26-0-173-7:811249:811413 [1] NCCL INFO Channel 14/0 : 57[1] -> 56[0] via P2P/IPC
+ip-26-0-173-7:811249:811413 [1] NCCL INFO Channel 15/0 : 57[1] -> 56[0] via P2P/IPC
+ip-26-0-172-252:688206:688363 [6] NCCL INFO Channel 14/0 : 38[6] -> 35[3] via P2P/IPC
+ip-26-0-172-57:942613:942776 [2] NCCL INFO Channel 15/0 : 42[2] -> 41[1] via P2P/IPC
+ip-26-0-173-7:811255:811411 [7] NCCL INFO Channel 07/0 : 52[4] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-7:811255:811411 [7] NCCL INFO Channel 15/0 : 52[4] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:94349:94505 [7] NCCL INFO Channel 07/0 : 60[4] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:94349:94505 [7] NCCL INFO Channel 15/0 : 60[4] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:94346:94511 [4] NCCL INFO Channel 07/0 : 4[4] -> 15[7] [send] via NET/Libfabric/7(7)/GDRDMA
+ip-26-0-160-225:94346:94511 [4] NCCL INFO Channel 15/0 : 4[4] -> 15[7] [send] via NET/Libfabric/7(7)/GDRDMA
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO Channel 07/0 : 4[4] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO Channel 15/0 : 12[4] -> 23[7] [send] via NET/Libfabric/7(15)/GDRDMA
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO Channel 15/0 : 4[4] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO Channel 07/0 : 15[7] -> 11[3] via P2P/IPC
+ip-26-0-172-147:618591:618746 [7] NCCL INFO Channel 07/0 : 20[4] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618591:618746 [7] NCCL INFO Channel 15/0 : 20[4] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880686:880841 [7] NCCL INFO Channel 07/0 : 12[4] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618588:618750 [4] NCCL INFO Channel 15/0 : 28[4] -> 39[7] [send] via NET/Libfabric/7(31)/GDRDMA
+ip-26-0-172-142:880686:880841 [7] NCCL INFO Channel 15/0 : 12[4] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618591:618746 [7] NCCL INFO Channel 07/0 : 31[7] -> 27[3] via P2P/IPC
+ip-26-0-172-142:880686:880841 [7] NCCL INFO Channel 07/0 : 23[7] -> 19[3] via P2P/IPC
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO Channel 15/0 : 15[7] -> 11[3] via P2P/IPC
+ip-26-0-172-147:618591:618746 [7] NCCL INFO Channel 15/0 : 31[7] -> 27[3] via P2P/IPC
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO Channel 01/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-172-147:618591:618746 [7] NCCL INFO Channel 01/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-172-142:880686:880841 [7] NCCL INFO Channel 15/0 : 23[7] -> 19[3] via P2P/IPC
+ip-26-0-172-142:880686:880841 [7] NCCL INFO Channel 01/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO Channel 02/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO Channel 03/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO Channel 04/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO Channel 05/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-172-147:618591:618746 [7] NCCL INFO Channel 02/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO Channel 09/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO Channel 10/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO Channel 11/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-160-225:94346:94511 [4] NCCL INFO Channel 01/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-225:94346:94511 [4] NCCL INFO Channel 02/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO Channel 12/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-172-142:880686:880841 [7] NCCL INFO Channel 02/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-160-225:94346:94511 [4] NCCL INFO Channel 04/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-225:94346:94511 [4] NCCL INFO Channel 09/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO Channel 13/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-160-225:94346:94511 [4] NCCL INFO Channel 10/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-147:618591:618746 [7] NCCL INFO Channel 03/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-160-225:94346:94511 [4] NCCL INFO Channel 12/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO Channel 01/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-172-142:880686:880841 [7] NCCL INFO Channel 03/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-172-142:880686:880841 [7] NCCL INFO Channel 04/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO Channel 02/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-172-147:618591:618746 [7] NCCL INFO Channel 04/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-172-142:880686:880841 [7] NCCL INFO Channel 05/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-172-147:618591:618746 [7] NCCL INFO Channel 05/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO Channel 04/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-172-142:880686:880841 [7] NCCL INFO Channel 09/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-172-147:618591:618746 [7] NCCL INFO Channel 09/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-172-147:618591:618746 [7] NCCL INFO Channel 10/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO Channel 09/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-172-147:618591:618746 [7] NCCL INFO Channel 11/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-172-142:880686:880841 [7] NCCL INFO Channel 10/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-172-147:618591:618746 [7] NCCL INFO Channel 12/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO Channel 10/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-172-147:618591:618746 [7] NCCL INFO Channel 13/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-172-142:880686:880841 [7] NCCL INFO Channel 11/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO Channel 01/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO Channel 12/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO Channel 03/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-172-142:880686:880841 [7] NCCL INFO Channel 12/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO Channel 01/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO Channel 04/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO Channel 02/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-172-142:880686:880841 [7] NCCL INFO Channel 13/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-172-142:880683:880844 [4] NCCL INFO Channel 01/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-172-142:880683:880844 [4] NCCL INFO Channel 02/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-172-142:880683:880844 [4] NCCL INFO Channel 04/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-172-142:880682:880840 [3] NCCL INFO Channel 01/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO Channel 03/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-172-142:880683:880844 [4] NCCL INFO Channel 09/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-172-142:880682:880840 [3] NCCL INFO Channel 03/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO Channel 04/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO Channel 05/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO Channel 07/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-172-142:880683:880844 [4] NCCL INFO Channel 10/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO Channel 06/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-172-142:880682:880840 [3] NCCL INFO Channel 04/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO Channel 09/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO Channel 07/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO Channel 10/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-172-142:880683:880844 [4] NCCL INFO Channel 12/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-172-142:880682:880840 [3] NCCL INFO Channel 05/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO Channel 09/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO Channel 11/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-172-142:880685:880847 [6] NCCL INFO Channel 01/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO Channel 11/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO Channel 12/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-172-142:880682:880840 [3] NCCL INFO Channel 06/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO Channel 12/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-172-142:880685:880847 [6] NCCL INFO Channel 02/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO Channel 15/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-172-142:880682:880840 [3] NCCL INFO Channel 07/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-172-142:880685:880847 [6] NCCL INFO Channel 03/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO Channel 13/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-172-142:880684:880842 [5] NCCL INFO Channel 01/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO Channel 01/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-172-142:880682:880840 [3] NCCL INFO Channel 09/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-172-142:880685:880847 [6] NCCL INFO Channel 04/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-172-142:880684:880842 [5] NCCL INFO Channel 02/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-172-142:880682:880840 [3] NCCL INFO Channel 11/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-172-142:880685:880847 [6] NCCL INFO Channel 07/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-172-142:880684:880842 [5] NCCL INFO Channel 03/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-172-142:880682:880840 [3] NCCL INFO Channel 12/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-172-142:880685:880847 [6] NCCL INFO Channel 09/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-172-142:880684:880842 [5] NCCL INFO Channel 06/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-172-142:880682:880840 [3] NCCL INFO Channel 13/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO Channel 14/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-172-142:880685:880847 [6] NCCL INFO Channel 10/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-172-147:618587:618751 [3] NCCL INFO Channel 01/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO Channel 02/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-172-142:880684:880842 [5] NCCL INFO Channel 07/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-172-142:880682:880840 [3] NCCL INFO Channel 14/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-172-142:880685:880847 [6] NCCL INFO Channel 11/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-172-147:618587:618751 [3] NCCL INFO Channel 03/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-172-142:880684:880842 [5] NCCL INFO Channel 09/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-172-142:880682:880840 [3] NCCL INFO Channel 15/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-172-142:880685:880847 [6] NCCL INFO Channel 12/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-172-147:618587:618751 [3] NCCL INFO Channel 04/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-172-142:880684:880842 [5] NCCL INFO Channel 10/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-172-147:618589:618744 [5] NCCL INFO Channel 01/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-172-147:618587:618751 [3] NCCL INFO Channel 05/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-172-147:618589:618744 [5] NCCL INFO Channel 02/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-172-147:618587:618751 [3] NCCL INFO Channel 06/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-172-147:618589:618744 [5] NCCL INFO Channel 03/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-172-147:618590:618749 [6] NCCL INFO Channel 01/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-172-147:618587:618751 [3] NCCL INFO Channel 07/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-172-147:618589:618744 [5] NCCL INFO Channel 06/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO Channel 15/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-172-147:618590:618749 [6] NCCL INFO Channel 02/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-172-147:618587:618751 [3] NCCL INFO Channel 09/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-172-147:618589:618744 [5] NCCL INFO Channel 07/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO Channel 03/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-172-147:618590:618749 [6] NCCL INFO Channel 03/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-172-147:618587:618751 [3] NCCL INFO Channel 11/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-172-147:618589:618744 [5] NCCL INFO Channel 09/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-172-147:618590:618749 [6] NCCL INFO Channel 04/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-172-147:618587:618751 [3] NCCL INFO Channel 12/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-172-147:618589:618744 [5] NCCL INFO Channel 10/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-172-147:618590:618749 [6] NCCL INFO Channel 07/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-172-147:618587:618751 [3] NCCL INFO Channel 13/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-172-147:618589:618744 [5] NCCL INFO Channel 11/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-172-147:618590:618749 [6] NCCL INFO Channel 09/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-172-147:618587:618751 [3] NCCL INFO Channel 14/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-172-142:880685:880847 [6] NCCL INFO Channel 15/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-172-147:618589:618744 [5] NCCL INFO Channel 14/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-172-142:880684:880842 [5] NCCL INFO Channel 11/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-172-147:618590:618749 [6] NCCL INFO Channel 10/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-172-147:618587:618751 [3] NCCL INFO Channel 15/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-172-147:618589:618744 [5] NCCL INFO Channel 15/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-172-147:618590:618749 [6] NCCL INFO Channel 11/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO Channel 06/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-172-147:618590:618749 [6] NCCL INFO Channel 12/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-172-147:618590:618749 [6] NCCL INFO Channel 15/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-172-73:926598:926761 [4] NCCL INFO Channel 04/0 : 45[5] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926598:926761 [4] NCCL INFO Channel 12/0 : 45[5] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926598:926761 [4] NCCL INFO Channel 03/0 : 52[4] -> 59[3] [send] via NET/Libfabric/3(51)/GDRDMA
+ip-26-0-172-73:926598:926761 [4] NCCL INFO Channel 11/0 : 52[4] -> 59[3] [send] via NET/Libfabric/3(51)/GDRDMA
+ip-26-0-172-142:880684:880842 [5] NCCL INFO Channel 14/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-172-57:942616:942777 [5] NCCL INFO Channel 05/0 : 45[5] -> 43[3] via P2P/IPC
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO Channel 07/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-172-73:926599:926755 [5] NCCL INFO Channel 05/0 : 46[6] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926598:926761 [4] NCCL INFO Channel 07/0 : 52[4] -> 63[7] [send] via NET/Libfabric/7(55)/GDRDMA
+ip-26-0-172-73:926599:926755 [5] NCCL INFO Channel 13/0 : 46[6] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926598:926761 [4] NCCL INFO Channel 15/0 : 52[4] -> 63[7] [send] via NET/Libfabric/7(55)/GDRDMA
+ip-26-0-172-73:926599:926755 [5] NCCL INFO Channel 04/0 : 53[5] -> 60[4] [send] via NET/Libfabric/4(52)/GDRDMA
+ip-26-0-172-73:926599:926755 [5] NCCL INFO Channel 12/0 : 53[5] -> 60[4] [send] via NET/Libfabric/4(52)/GDRDMA
+ip-26-0-172-57:942617:942774 [6] NCCL INFO Channel 06/0 : 46[6] -> 43[3] via P2P/IPC
+ip-26-0-172-73:926599:926755 [5] NCCL INFO Channel 05/0 : 53[5] -> 51[3] via P2P/IPC
+ip-26-0-172-73:926600:926757 [6] NCCL INFO Channel 06/0 : 47[7] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926601:926759 [7] NCCL INFO Channel 06/0 : 55[7] -> 62[6] [send] via NET/Libfabric/6(54)/GDRDMA
+ip-26-0-172-73:926600:926757 [6] NCCL INFO Channel 14/0 : 47[7] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926601:926759 [7] NCCL INFO Channel 14/0 : 55[7] -> 62[6] [send] via NET/Libfabric/6(54)/GDRDMA
+ip-26-0-173-7:811255:811411 [7] NCCL INFO Channel 07/0 : 63[7] -> 59[3] via P2P/IPC
+ip-26-0-173-7:811252:811408 [4] NCCL INFO Channel 07/0 : 60[4] -> 7[7] [send] via NET/Libfabric/7(63)/GDRDMA
+ip-26-0-172-73:926601:926759 [7] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-173-7:811254:811414 [6] NCCL INFO Channel 06/0 : 62[6] -> 59[3] via P2P/IPC
+ip-26-0-172-73:926599:926755 [5] NCCL INFO Channel 13/0 : 53[5] -> 51[3] via P2P/IPC
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO Channel 09/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-172-73:926600:926757 [6] NCCL INFO Channel 05/0 : 54[6] -> 61[5] [send] via NET/Libfabric/5(53)/GDRDMA
+ip-26-0-172-73:926600:926757 [6] NCCL INFO Channel 13/0 : 54[6] -> 61[5] [send] via NET/Libfabric/5(53)/GDRDMA
+ip-26-0-172-57:942618:942778 [7] NCCL INFO NCCL_NET_FORCE_FLUSH set by environment to 0.
+ip-26-0-172-73:926600:926757 [6] NCCL INFO Channel 06/0 : 54[6] -> 51[3] via P2P/IPC
+ip-26-0-173-7:811253:811410 [5] NCCL INFO Channel 05/0 : 61[5] -> 59[3] via P2P/IPC
+ip-26-0-172-73:926600:926757 [6] NCCL INFO Channel 14/0 : 54[6] -> 51[3] via P2P/IPC
+ip-26-0-172-57:942617:942774 [6] NCCL INFO Channel 14/0 : 46[6] -> 43[3] via P2P/IPC
+ip-26-0-172-57:942616:942777 [5] NCCL INFO Channel 13/0 : 45[5] -> 43[3] via P2P/IPC
+ip-26-0-172-73:926598:926761 [4] NCCL INFO Channel 01/0 : 52[4] -> 51[3] via P2P/IPC
+ip-26-0-172-73:926598:926761 [4] NCCL INFO Channel 02/0 : 52[4] -> 51[3] via P2P/IPC
+ip-26-0-172-73:926598:926761 [4] NCCL INFO Channel 04/0 : 52[4] -> 51[3] via P2P/IPC
+ip-26-0-172-73:926598:926761 [4] NCCL INFO Channel 09/0 : 52[4] -> 51[3] via P2P/IPC
+ip-26-0-172-73:926598:926761 [4] NCCL INFO Channel 10/0 : 52[4] -> 51[3] via P2P/IPC
+ip-26-0-172-73:926598:926761 [4] NCCL INFO Channel 12/0 : 52[4] -> 51[3] via P2P/IPC
+ip-26-0-173-7:811254:811414 [6] NCCL INFO Channel 14/0 : 62[6] -> 59[3] via P2P/IPC
+ip-26-0-173-7:811255:811411 [7] NCCL INFO Channel 15/0 : 63[7] -> 59[3] via P2P/IPC
+ip-26-0-173-7:811252:811408 [4] NCCL INFO Channel 15/0 : 60[4] -> 7[7] [send] via NET/Libfabric/7(63)/GDRDMA
+ip-26-0-160-225:94349:94505 [7] NCCL INFO Channel 07/0 : 7[7] -> 3[3] via P2P/IPC
+ip-26-0-172-142:880684:880842 [5] NCCL INFO Channel 15/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO Channel 10/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-160-225:94349:94505 [7] NCCL INFO Channel 15/0 : 7[7] -> 3[3] via P2P/IPC
+ip-26-0-160-225:94349:94505 [7] NCCL INFO Channel 01/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-225:94349:94505 [7] NCCL INFO Channel 02/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-225:94349:94505 [7] NCCL INFO Channel 03/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-225:94349:94505 [7] NCCL INFO Channel 04/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-173-7:811253:811410 [5] NCCL INFO Channel 13/0 : 61[5] -> 59[3] via P2P/IPC
+ip-26-0-160-225:94349:94505 [7] NCCL INFO Channel 05/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-225:94349:94505 [7] NCCL INFO Channel 09/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-225:94349:94505 [7] NCCL INFO Channel 10/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-252:688207:688366 [7] NCCL INFO Channel 07/0 : 28[4] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688207:688366 [7] NCCL INFO Channel 15/0 : 28[4] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:94349:94505 [7] NCCL INFO Channel 11/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-252:688207:688366 [7] NCCL INFO Channel 07/0 : 39[7] -> 35[3] via P2P/IPC
+ip-26-0-160-225:94349:94505 [7] NCCL INFO Channel 12/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-252:688207:688366 [7] NCCL INFO Channel 15/0 : 39[7] -> 35[3] via P2P/IPC
+ip-26-0-160-225:94349:94505 [7] NCCL INFO Channel 13/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-252:688207:688366 [7] NCCL INFO Channel 01/0 : 39[7] -> 38[6] via P2P/IPC
+ip-26-0-172-252:688207:688366 [7] NCCL INFO Channel 02/0 : 39[7] -> 38[6] via P2P/IPC
+ip-26-0-172-252:688207:688366 [7] NCCL INFO Channel 03/0 : 39[7] -> 38[6] via P2P/IPC
+ip-26-0-172-252:688207:688366 [7] NCCL INFO Channel 04/0 : 39[7] -> 38[6] via P2P/IPC
+ip-26-0-172-252:688207:688366 [7] NCCL INFO Channel 05/0 : 39[7] -> 38[6] via P2P/IPC
+ip-26-0-172-252:688207:688366 [7] NCCL INFO Channel 09/0 : 39[7] -> 38[6] via P2P/IPC
+ip-26-0-172-252:688207:688366 [7] NCCL INFO Channel 10/0 : 39[7] -> 38[6] via P2P/IPC
+ip-26-0-160-225:94345:94509 [3] NCCL INFO Channel 01/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-252:688207:688366 [7] NCCL INFO Channel 11/0 : 39[7] -> 38[6] via P2P/IPC
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO Channel 11/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-172-252:688207:688366 [7] NCCL INFO Channel 12/0 : 39[7] -> 38[6] via P2P/IPC
+ip-26-0-172-252:688207:688366 [7] NCCL INFO Channel 13/0 : 39[7] -> 38[6] via P2P/IPC
+ip-26-0-173-7:811255:811411 [7] NCCL INFO Channel 01/0 : 63[7] -> 62[6] via P2P/IPC
+ip-26-0-160-225:94345:94509 [3] NCCL INFO Channel 03/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-252:688203:688362 [3] NCCL INFO Channel 01/0 : 35[3] -> 34[2] via P2P/IPC
+ip-26-0-160-225:94345:94509 [3] NCCL INFO Channel 04/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-225:94348:94506 [6] NCCL INFO Channel 01/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-252:688203:688362 [3] NCCL INFO Channel 03/0 : 35[3] -> 34[2] via P2P/IPC
+ip-26-0-172-252:688206:688363 [6] NCCL INFO Channel 01/0 : 38[6] -> 37[5] via P2P/IPC
+ip-26-0-172-252:688203:688362 [3] NCCL INFO Channel 04/0 : 35[3] -> 34[2] via P2P/IPC
+ip-26-0-160-225:94345:94509 [3] NCCL INFO Channel 05/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-225:94348:94506 [6] NCCL INFO Channel 02/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-252:688206:688363 [6] NCCL INFO Channel 02/0 : 38[6] -> 37[5] via P2P/IPC
+ip-26-0-172-252:688203:688362 [3] NCCL INFO Channel 05/0 : 35[3] -> 34[2] via P2P/IPC
+ip-26-0-172-252:688206:688363 [6] NCCL INFO Channel 03/0 : 38[6] -> 37[5] via P2P/IPC
+ip-26-0-172-252:688203:688362 [3] NCCL INFO Channel 06/0 : 35[3] -> 34[2] via P2P/IPC
+ip-26-0-172-252:688206:688363 [6] NCCL INFO Channel 04/0 : 38[6] -> 37[5] via P2P/IPC
+ip-26-0-172-252:688203:688362 [3] NCCL INFO Channel 07/0 : 35[3] -> 34[2] via P2P/IPC
+ip-26-0-172-252:688206:688363 [6] NCCL INFO Channel 07/0 : 38[6] -> 37[5] via P2P/IPC
+ip-26-0-172-252:688203:688362 [3] NCCL INFO Channel 09/0 : 35[3] -> 34[2] via P2P/IPC
+ip-26-0-160-225:94345:94509 [3] NCCL INFO Channel 06/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-252:688206:688363 [6] NCCL INFO Channel 09/0 : 38[6] -> 37[5] via P2P/IPC
+ip-26-0-160-225:94348:94506 [6] NCCL INFO Channel 03/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-173-7:811255:811411 [7] NCCL INFO Channel 02/0 : 63[7] -> 62[6] via P2P/IPC
+ip-26-0-160-225:94345:94509 [3] NCCL INFO Channel 07/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-225:94348:94506 [6] NCCL INFO Channel 04/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-225:94347:94507 [5] NCCL INFO Channel 01/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-225:94345:94509 [3] NCCL INFO Channel 09/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-252:688206:688363 [6] NCCL INFO Channel 10/0 : 38[6] -> 37[5] via P2P/IPC
+ip-26-0-172-252:688206:688363 [6] NCCL INFO Channel 11/0 : 38[6] -> 37[5] via P2P/IPC
+ip-26-0-172-252:688205:688365 [5] NCCL INFO Channel 01/0 : 37[5] -> 36[4] via P2P/IPC
+ip-26-0-172-252:688206:688363 [6] NCCL INFO Channel 12/0 : 38[6] -> 37[5] via P2P/IPC
+ip-26-0-172-252:688203:688362 [3] NCCL INFO Channel 11/0 : 35[3] -> 34[2] via P2P/IPC
+ip-26-0-160-225:94345:94509 [3] NCCL INFO Channel 11/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-252:688205:688365 [5] NCCL INFO Channel 02/0 : 37[5] -> 36[4] via P2P/IPC
+ip-26-0-172-252:688206:688363 [6] NCCL INFO Channel 15/0 : 38[6] -> 37[5] via P2P/IPC
+ip-26-0-160-225:94345:94509 [3] NCCL INFO Channel 12/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-252:688203:688362 [3] NCCL INFO Channel 12/0 : 35[3] -> 34[2] via P2P/IPC
+ip-26-0-172-252:688205:688365 [5] NCCL INFO Channel 03/0 : 37[5] -> 36[4] via P2P/IPC
+ip-26-0-160-225:94347:94507 [5] NCCL INFO Channel 02/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-225:94345:94509 [3] NCCL INFO Channel 13/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-225:94348:94506 [6] NCCL INFO Channel 07/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-252:688203:688362 [3] NCCL INFO Channel 13/0 : 35[3] -> 34[2] via P2P/IPC
+ip-26-0-160-225:94347:94507 [5] NCCL INFO Channel 03/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-225:94345:94509 [3] NCCL INFO Channel 14/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-225:94348:94506 [6] NCCL INFO Channel 09/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-225:94347:94507 [5] NCCL INFO Channel 06/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-225:94345:94509 [3] NCCL INFO Channel 15/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-225:94348:94506 [6] NCCL INFO Channel 10/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-225:94347:94507 [5] NCCL INFO Channel 07/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-225:94348:94506 [6] NCCL INFO Channel 11/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-225:94347:94507 [5] NCCL INFO Channel 09/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-252:688205:688365 [5] NCCL INFO Channel 06/0 : 37[5] -> 36[4] via P2P/IPC
+ip-26-0-160-225:94348:94506 [6] NCCL INFO Channel 12/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-173-7:811255:811411 [7] NCCL INFO Channel 03/0 : 63[7] -> 62[6] via P2P/IPC
+ip-26-0-172-252:688203:688362 [3] NCCL INFO Channel 14/0 : 35[3] -> 34[2] via P2P/IPC
+ip-26-0-172-252:688205:688365 [5] NCCL INFO Channel 07/0 : 37[5] -> 36[4] via P2P/IPC
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO Channel 14/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-160-225:94347:94507 [5] NCCL INFO Channel 10/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-252:688203:688362 [3] NCCL INFO Channel 15/0 : 35[3] -> 34[2] via P2P/IPC
+ip-26-0-173-7:811255:811411 [7] NCCL INFO Channel 04/0 : 63[7] -> 62[6] via P2P/IPC
+ip-26-0-172-252:688205:688365 [5] NCCL INFO Channel 09/0 : 37[5] -> 36[4] via P2P/IPC
+ip-26-0-173-7:811255:811411 [7] NCCL INFO Channel 05/0 : 63[7] -> 62[6] via P2P/IPC
+ip-26-0-173-7:811255:811411 [7] NCCL INFO Channel 09/0 : 63[7] -> 62[6] via P2P/IPC
+ip-26-0-160-225:94348:94506 [6] NCCL INFO Channel 15/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-173-7:811255:811411 [7] NCCL INFO Channel 10/0 : 63[7] -> 62[6] via P2P/IPC
+ip-26-0-173-7:811255:811411 [7] NCCL INFO Channel 11/0 : 63[7] -> 62[6] via P2P/IPC
+ip-26-0-173-7:811255:811411 [7] NCCL INFO Channel 12/0 : 63[7] -> 62[6] via P2P/IPC
+ip-26-0-173-7:811255:811411 [7] NCCL INFO Channel 13/0 : 63[7] -> 62[6] via P2P/IPC
+ip-26-0-173-7:811252:811408 [4] NCCL INFO Channel 01/0 : 60[4] -> 59[3] via P2P/IPC
+ip-26-0-160-225:94347:94507 [5] NCCL INFO Channel 11/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-173-7:811252:811408 [4] NCCL INFO Channel 02/0 : 60[4] -> 59[3] via P2P/IPC
+ip-26-0-173-7:811252:811408 [4] NCCL INFO Channel 04/0 : 60[4] -> 59[3] via P2P/IPC
+ip-26-0-172-252:688205:688365 [5] NCCL INFO Channel 10/0 : 37[5] -> 36[4] via P2P/IPC
+ip-26-0-173-7:811252:811408 [4] NCCL INFO Channel 09/0 : 60[4] -> 59[3] via P2P/IPC
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO Channel 15/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-173-7:811252:811408 [4] NCCL INFO Channel 10/0 : 60[4] -> 59[3] via P2P/IPC
+ip-26-0-173-7:811252:811408 [4] NCCL INFO Channel 12/0 : 60[4] -> 59[3] via P2P/IPC
+ip-26-0-173-7:811251:811409 [3] NCCL INFO Channel 01/0 : 59[3] -> 58[2] via P2P/IPC
+ip-26-0-160-225:94347:94507 [5] NCCL INFO Channel 14/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-173-7:811251:811409 [3] NCCL INFO Channel 03/0 : 59[3] -> 58[2] via P2P/IPC
+ip-26-0-173-7:811254:811414 [6] NCCL INFO Channel 01/0 : 62[6] -> 61[5] via P2P/IPC
+ip-26-0-173-7:811251:811409 [3] NCCL INFO Channel 04/0 : 59[3] -> 58[2] via P2P/IPC
+ip-26-0-173-7:811251:811409 [3] NCCL INFO Channel 05/0 : 59[3] -> 58[2] via P2P/IPC
+ip-26-0-173-7:811251:811409 [3] NCCL INFO Channel 06/0 : 59[3] -> 58[2] via P2P/IPC
+ip-26-0-173-7:811251:811409 [3] NCCL INFO Channel 07/0 : 59[3] -> 58[2] via P2P/IPC
+ip-26-0-173-7:811251:811409 [3] NCCL INFO Channel 09/0 : 59[3] -> 58[2] via P2P/IPC
+ip-26-0-173-7:811251:811409 [3] NCCL INFO Channel 11/0 : 59[3] -> 58[2] via P2P/IPC
+ip-26-0-173-7:811254:811414 [6] NCCL INFO Channel 02/0 : 62[6] -> 61[5] via P2P/IPC
+ip-26-0-172-252:688205:688365 [5] NCCL INFO Channel 11/0 : 37[5] -> 36[4] via P2P/IPC
+ip-26-0-173-7:811251:811409 [3] NCCL INFO Channel 12/0 : 59[3] -> 58[2] via P2P/IPC
+ip-26-0-173-7:811254:811414 [6] NCCL INFO Channel 03/0 : 62[6] -> 61[5] via P2P/IPC
+ip-26-0-173-7:811251:811409 [3] NCCL INFO Channel 13/0 : 59[3] -> 58[2] via P2P/IPC
+ip-26-0-173-7:811253:811410 [5] NCCL INFO Channel 01/0 : 61[5] -> 60[4] via P2P/IPC
+ip-26-0-173-7:811254:811414 [6] NCCL INFO Channel 04/0 : 62[6] -> 61[5] via P2P/IPC
+ip-26-0-173-7:811251:811409 [3] NCCL INFO Channel 14/0 : 59[3] -> 58[2] via P2P/IPC
+ip-26-0-173-7:811253:811410 [5] NCCL INFO Channel 02/0 : 61[5] -> 60[4] via P2P/IPC
+ip-26-0-173-7:811254:811414 [6] NCCL INFO Channel 07/0 : 62[6] -> 61[5] via P2P/IPC
+ip-26-0-173-7:811253:811410 [5] NCCL INFO Channel 03/0 : 61[5] -> 60[4] via P2P/IPC
+ip-26-0-173-7:811251:811409 [3] NCCL INFO Channel 15/0 : 59[3] -> 58[2] via P2P/IPC
+ip-26-0-173-7:811254:811414 [6] NCCL INFO Channel 09/0 : 62[6] -> 61[5] via P2P/IPC
+ip-26-0-173-7:811253:811410 [5] NCCL INFO Channel 06/0 : 61[5] -> 60[4] via P2P/IPC
+ip-26-0-173-7:811254:811414 [6] NCCL INFO Channel 10/0 : 62[6] -> 61[5] via P2P/IPC
+ip-26-0-172-252:688205:688365 [5] NCCL INFO Channel 14/0 : 37[5] -> 36[4] via P2P/IPC
+ip-26-0-160-225:94347:94507 [5] NCCL INFO Channel 15/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-173-7:811253:811410 [5] NCCL INFO Channel 07/0 : 61[5] -> 60[4] via P2P/IPC
+ip-26-0-173-7:811254:811414 [6] NCCL INFO Channel 11/0 : 62[6] -> 61[5] via P2P/IPC
+ip-26-0-173-7:811253:811410 [5] NCCL INFO Channel 09/0 : 61[5] -> 60[4] via P2P/IPC
+ip-26-0-173-7:811253:811410 [5] NCCL INFO Channel 10/0 : 61[5] -> 60[4] via P2P/IPC
+ip-26-0-172-252:688205:688365 [5] NCCL INFO Channel 15/0 : 37[5] -> 36[4] via P2P/IPC
+ip-26-0-173-7:811254:811414 [6] NCCL INFO Channel 12/0 : 62[6] -> 61[5] via P2P/IPC
+ip-26-0-173-7:811253:811410 [5] NCCL INFO Channel 11/0 : 61[5] -> 60[4] via P2P/IPC
+ip-26-0-173-7:811254:811414 [6] NCCL INFO Channel 15/0 : 62[6] -> 61[5] via P2P/IPC
+ip-26-0-173-7:811253:811410 [5] NCCL INFO Channel 14/0 : 61[5] -> 60[4] via P2P/IPC
+ip-26-0-173-7:811253:811410 [5] NCCL INFO Channel 15/0 : 61[5] -> 60[4] via P2P/IPC
+ip-26-0-172-73:926601:926759 [7] NCCL INFO Channel 07/0 : 44[4] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926601:926759 [7] NCCL INFO Channel 15/0 : 44[4] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926601:926759 [7] NCCL INFO Channel 07/0 : 55[7] -> 51[3] via P2P/IPC
+ip-26-0-172-57:942618:942778 [7] NCCL INFO Channel 07/0 : 36[4] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942618:942778 [7] NCCL INFO Channel 15/0 : 36[4] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942618:942778 [7] NCCL INFO Channel 07/0 : 47[7] -> 43[3] via P2P/IPC
+ip-26-0-172-57:942618:942778 [7] NCCL INFO Channel 15/0 : 47[7] -> 43[3] via P2P/IPC
+ip-26-0-172-57:942618:942778 [7] NCCL INFO Channel 01/0 : 47[7] -> 46[6] via P2P/IPC
+ip-26-0-172-57:942618:942778 [7] NCCL INFO Channel 02/0 : 47[7] -> 46[6] via P2P/IPC
+ip-26-0-172-57:942618:942778 [7] NCCL INFO Channel 03/0 : 47[7] -> 46[6] via P2P/IPC
+ip-26-0-172-57:942618:942778 [7] NCCL INFO Channel 04/0 : 47[7] -> 46[6] via P2P/IPC
+ip-26-0-172-73:926601:926759 [7] NCCL INFO Channel 15/0 : 55[7] -> 51[3] via P2P/IPC
+ip-26-0-172-57:942618:942778 [7] NCCL INFO Channel 05/0 : 47[7] -> 46[6] via P2P/IPC
+ip-26-0-172-57:942618:942778 [7] NCCL INFO Channel 09/0 : 47[7] -> 46[6] via P2P/IPC
+ip-26-0-172-57:942618:942778 [7] NCCL INFO Channel 10/0 : 47[7] -> 46[6] via P2P/IPC
+ip-26-0-172-57:942618:942778 [7] NCCL INFO Channel 11/0 : 47[7] -> 46[6] via P2P/IPC
+ip-26-0-172-57:942618:942778 [7] NCCL INFO Channel 12/0 : 47[7] -> 46[6] via P2P/IPC
+ip-26-0-172-57:942618:942778 [7] NCCL INFO Channel 13/0 : 47[7] -> 46[6] via P2P/IPC
+ip-26-0-172-57:942615:942775 [4] NCCL INFO Channel 01/0 : 44[4] -> 43[3] via P2P/IPC
+ip-26-0-172-57:942615:942775 [4] NCCL INFO Channel 02/0 : 44[4] -> 43[3] via P2P/IPC
+ip-26-0-172-73:926601:926759 [7] NCCL INFO Channel 01/0 : 55[7] -> 54[6] via P2P/IPC
+ip-26-0-172-57:942615:942775 [4] NCCL INFO Channel 04/0 : 44[4] -> 43[3] via P2P/IPC
+ip-26-0-172-57:942615:942775 [4] NCCL INFO Channel 09/0 : 44[4] -> 43[3] via P2P/IPC
+ip-26-0-172-57:942615:942775 [4] NCCL INFO Channel 10/0 : 44[4] -> 43[3] via P2P/IPC
+ip-26-0-172-147:618588:618750 [4] NCCL INFO Channel 01/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-172-57:942615:942775 [4] NCCL INFO Channel 12/0 : 44[4] -> 43[3] via P2P/IPC
+ip-26-0-172-57:942614:942773 [3] NCCL INFO Channel 01/0 : 43[3] -> 42[2] via P2P/IPC
+ip-26-0-172-57:942614:942773 [3] NCCL INFO Channel 03/0 : 43[3] -> 42[2] via P2P/IPC
+ip-26-0-172-57:942617:942774 [6] NCCL INFO Channel 01/0 : 46[6] -> 45[5] via P2P/IPC
+ip-26-0-172-57:942614:942773 [3] NCCL INFO Channel 04/0 : 43[3] -> 42[2] via P2P/IPC
+ip-26-0-172-57:942617:942774 [6] NCCL INFO Channel 02/0 : 46[6] -> 45[5] via P2P/IPC
+ip-26-0-172-57:942614:942773 [3] NCCL INFO Channel 05/0 : 43[3] -> 42[2] via P2P/IPC
+ip-26-0-172-57:942617:942774 [6] NCCL INFO Channel 03/0 : 46[6] -> 45[5] via P2P/IPC
+ip-26-0-172-57:942614:942773 [3] NCCL INFO Channel 06/0 : 43[3] -> 42[2] via P2P/IPC
+ip-26-0-172-147:618588:618750 [4] NCCL INFO Channel 02/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-172-73:926601:926759 [7] NCCL INFO Channel 02/0 : 55[7] -> 54[6] via P2P/IPC
+ip-26-0-172-57:942614:942773 [3] NCCL INFO Channel 07/0 : 43[3] -> 42[2] via P2P/IPC
+ip-26-0-172-57:942614:942773 [3] NCCL INFO Channel 09/0 : 43[3] -> 42[2] via P2P/IPC
+ip-26-0-172-57:942614:942773 [3] NCCL INFO Channel 11/0 : 43[3] -> 42[2] via P2P/IPC
+ip-26-0-172-57:942616:942777 [5] NCCL INFO Channel 01/0 : 45[5] -> 44[4] via P2P/IPC
+ip-26-0-172-57:942617:942774 [6] NCCL INFO Channel 04/0 : 46[6] -> 45[5] via P2P/IPC
+ip-26-0-172-57:942614:942773 [3] NCCL INFO Channel 12/0 : 43[3] -> 42[2] via P2P/IPC
+ip-26-0-172-57:942616:942777 [5] NCCL INFO Channel 02/0 : 45[5] -> 44[4] via P2P/IPC
+ip-26-0-172-57:942617:942774 [6] NCCL INFO Channel 07/0 : 46[6] -> 45[5] via P2P/IPC
+ip-26-0-172-57:942614:942773 [3] NCCL INFO Channel 13/0 : 43[3] -> 42[2] via P2P/IPC
+ip-26-0-172-57:942616:942777 [5] NCCL INFO Channel 03/0 : 45[5] -> 44[4] via P2P/IPC
+ip-26-0-172-57:942617:942774 [6] NCCL INFO Channel 09/0 : 46[6] -> 45[5] via P2P/IPC
+ip-26-0-172-147:618588:618750 [4] NCCL INFO Channel 04/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-172-57:942614:942773 [3] NCCL INFO Channel 14/0 : 43[3] -> 42[2] via P2P/IPC
+ip-26-0-172-57:942616:942777 [5] NCCL INFO Channel 06/0 : 45[5] -> 44[4] via P2P/IPC
+ip-26-0-172-57:942617:942774 [6] NCCL INFO Channel 10/0 : 46[6] -> 45[5] via P2P/IPC
+ip-26-0-172-57:942614:942773 [3] NCCL INFO Channel 15/0 : 43[3] -> 42[2] via P2P/IPC
+ip-26-0-172-57:942616:942777 [5] NCCL INFO Channel 07/0 : 45[5] -> 44[4] via P2P/IPC
+ip-26-0-172-57:942617:942774 [6] NCCL INFO Channel 11/0 : 46[6] -> 45[5] via P2P/IPC
+ip-26-0-172-57:942616:942777 [5] NCCL INFO Channel 09/0 : 45[5] -> 44[4] via P2P/IPC
+ip-26-0-172-57:942617:942774 [6] NCCL INFO Channel 12/0 : 46[6] -> 45[5] via P2P/IPC
+ip-26-0-172-73:926601:926759 [7] NCCL INFO Channel 03/0 : 55[7] -> 54[6] via P2P/IPC
+ip-26-0-172-73:926601:926759 [7] NCCL INFO Channel 04/0 : 55[7] -> 54[6] via P2P/IPC
+ip-26-0-172-73:926601:926759 [7] NCCL INFO Channel 05/0 : 55[7] -> 54[6] via P2P/IPC
+ip-26-0-172-73:926601:926759 [7] NCCL INFO Channel 09/0 : 55[7] -> 54[6] via P2P/IPC
+ip-26-0-172-73:926601:926759 [7] NCCL INFO Channel 10/0 : 55[7] -> 54[6] via P2P/IPC
+ip-26-0-172-73:926601:926759 [7] NCCL INFO Channel 11/0 : 55[7] -> 54[6] via P2P/IPC
+ip-26-0-172-73:926601:926759 [7] NCCL INFO Channel 12/0 : 55[7] -> 54[6] via P2P/IPC
+ip-26-0-172-73:926601:926759 [7] NCCL INFO Channel 13/0 : 55[7] -> 54[6] via P2P/IPC
+ip-26-0-172-147:618588:618750 [4] NCCL INFO Channel 09/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-172-73:926597:926758 [3] NCCL INFO Channel 01/0 : 51[3] -> 50[2] via P2P/IPC
+ip-26-0-172-73:926597:926758 [3] NCCL INFO Channel 03/0 : 51[3] -> 50[2] via P2P/IPC
+ip-26-0-172-73:926597:926758 [3] NCCL INFO Channel 04/0 : 51[3] -> 50[2] via P2P/IPC
+ip-26-0-172-73:926600:926757 [6] NCCL INFO Channel 01/0 : 54[6] -> 53[5] via P2P/IPC
+ip-26-0-172-73:926597:926758 [3] NCCL INFO Channel 05/0 : 51[3] -> 50[2] via P2P/IPC
+ip-26-0-172-73:926600:926757 [6] NCCL INFO Channel 02/0 : 54[6] -> 53[5] via P2P/IPC
+ip-26-0-172-73:926597:926758 [3] NCCL INFO Channel 06/0 : 51[3] -> 50[2] via P2P/IPC
+ip-26-0-172-73:926599:926755 [5] NCCL INFO Channel 01/0 : 53[5] -> 52[4] via P2P/IPC
+ip-26-0-172-73:926600:926757 [6] NCCL INFO Channel 03/0 : 54[6] -> 53[5] via P2P/IPC
+ip-26-0-172-73:926597:926758 [3] NCCL INFO Channel 07/0 : 51[3] -> 50[2] via P2P/IPC
+ip-26-0-172-73:926599:926755 [5] NCCL INFO Channel 02/0 : 53[5] -> 52[4] via P2P/IPC
+ip-26-0-172-73:926600:926757 [6] NCCL INFO Channel 04/0 : 54[6] -> 53[5] via P2P/IPC
+ip-26-0-172-73:926597:926758 [3] NCCL INFO Channel 09/0 : 51[3] -> 50[2] via P2P/IPC
+ip-26-0-172-73:926599:926755 [5] NCCL INFO Channel 03/0 : 53[5] -> 52[4] via P2P/IPC
+ip-26-0-172-73:926600:926757 [6] NCCL INFO Channel 07/0 : 54[6] -> 53[5] via P2P/IPC
+ip-26-0-172-73:926597:926758 [3] NCCL INFO Channel 11/0 : 51[3] -> 50[2] via P2P/IPC
+ip-26-0-172-73:926599:926755 [5] NCCL INFO Channel 06/0 : 53[5] -> 52[4] via P2P/IPC
+ip-26-0-172-73:926600:926757 [6] NCCL INFO Channel 09/0 : 54[6] -> 53[5] via P2P/IPC
+ip-26-0-172-57:942616:942777 [5] NCCL INFO Channel 10/0 : 45[5] -> 44[4] via P2P/IPC
+ip-26-0-172-57:942617:942774 [6] NCCL INFO Channel 15/0 : 46[6] -> 45[5] via P2P/IPC
+ip-26-0-172-73:926597:926758 [3] NCCL INFO Channel 12/0 : 51[3] -> 50[2] via P2P/IPC
+ip-26-0-172-73:926599:926755 [5] NCCL INFO Channel 07/0 : 53[5] -> 52[4] via P2P/IPC
+ip-26-0-172-147:618588:618750 [4] NCCL INFO Channel 10/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-172-73:926600:926757 [6] NCCL INFO Channel 10/0 : 54[6] -> 53[5] via P2P/IPC
+ip-26-0-172-73:926597:926758 [3] NCCL INFO Channel 13/0 : 51[3] -> 50[2] via P2P/IPC
+ip-26-0-172-73:926599:926755 [5] NCCL INFO Channel 09/0 : 53[5] -> 52[4] via P2P/IPC
+ip-26-0-172-73:926600:926757 [6] NCCL INFO Channel 11/0 : 54[6] -> 53[5] via P2P/IPC
+ip-26-0-172-73:926597:926758 [3] NCCL INFO Channel 14/0 : 51[3] -> 50[2] via P2P/IPC
+ip-26-0-172-73:926599:926755 [5] NCCL INFO Channel 10/0 : 53[5] -> 52[4] via P2P/IPC
+ip-26-0-172-73:926600:926757 [6] NCCL INFO Channel 12/0 : 54[6] -> 53[5] via P2P/IPC
+ip-26-0-172-73:926597:926758 [3] NCCL INFO Channel 15/0 : 51[3] -> 50[2] via P2P/IPC
+ip-26-0-172-57:942616:942777 [5] NCCL INFO Channel 11/0 : 45[5] -> 44[4] via P2P/IPC
+ip-26-0-172-73:926599:926755 [5] NCCL INFO Channel 11/0 : 53[5] -> 52[4] via P2P/IPC
+ip-26-0-172-147:618588:618750 [4] NCCL INFO Channel 12/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-172-73:926600:926757 [6] NCCL INFO Channel 15/0 : 54[6] -> 53[5] via P2P/IPC
+ip-26-0-172-73:926599:926755 [5] NCCL INFO Channel 14/0 : 53[5] -> 52[4] via P2P/IPC
+ip-26-0-172-57:942616:942777 [5] NCCL INFO Channel 14/0 : 45[5] -> 44[4] via P2P/IPC
+ip-26-0-172-57:942616:942777 [5] NCCL INFO Channel 15/0 : 45[5] -> 44[4] via P2P/IPC
+ip-26-0-172-73:926599:926755 [5] NCCL INFO Channel 15/0 : 53[5] -> 52[4] via P2P/IPC
+ip-26-0-172-252:688204:688367 [4] NCCL INFO Channel 01/0 : 36[4] -> 35[3] via P2P/IPC
+ip-26-0-172-252:688204:688367 [4] NCCL INFO Channel 02/0 : 36[4] -> 35[3] via P2P/IPC
+ip-26-0-172-252:688204:688367 [4] NCCL INFO Channel 04/0 : 36[4] -> 35[3] via P2P/IPC
+ip-26-0-172-252:688204:688367 [4] NCCL INFO Channel 09/0 : 36[4] -> 35[3] via P2P/IPC
+ip-26-0-172-252:688204:688367 [4] NCCL INFO Channel 10/0 : 36[4] -> 35[3] via P2P/IPC
+ip-26-0-172-252:688204:688367 [4] NCCL INFO Channel 12/0 : 36[4] -> 35[3] via P2P/IPC
+ip-26-0-172-142:880679:880846 [0] NCCL INFO Connected all rings
+ip-26-0-172-142:880679:880846 [0] NCCL INFO Channel 02/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-172-142:880681:880845 [2] NCCL INFO Connected all rings
+ip-26-0-172-147:618585:618748 [1] NCCL INFO Connected all rings
+ip-26-0-172-142:880685:880847 [6] NCCL INFO Connected all rings
+ip-26-0-172-142:880683:880844 [4] NCCL INFO Connected all rings
+ip-26-0-172-142:880679:880846 [0] NCCL INFO Channel 03/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-172-252:688201:688364 [1] NCCL INFO Connected all rings
+ip-26-0-172-142:880679:880846 [0] NCCL INFO Channel 04/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-172-142:880679:880846 [0] NCCL INFO Channel 05/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-172-142:880679:880846 [0] NCCL INFO Channel 06/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-172-142:880679:880846 [0] NCCL INFO Channel 07/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-172-142:880679:880846 [0] NCCL INFO Channel 10/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-172-142:880679:880846 [0] NCCL INFO Channel 11/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-172-252:688200:688369 [0] NCCL INFO Connected all rings
+ip-26-0-172-252:688200:688369 [0] NCCL INFO Channel 02/0 : 32[0] -> 33[1] via P2P/IPC
+ip-26-0-172-142:880679:880846 [0] NCCL INFO Channel 12/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-172-252:688200:688369 [0] NCCL INFO Channel 03/0 : 32[0] -> 33[1] via P2P/IPC
+ip-26-0-172-252:688200:688369 [0] NCCL INFO Channel 04/0 : 32[0] -> 33[1] via P2P/IPC
+ip-26-0-172-142:880679:880846 [0] NCCL INFO Channel 13/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-172-252:688200:688369 [0] NCCL INFO Channel 05/0 : 32[0] -> 33[1] via P2P/IPC
+ip-26-0-172-252:688200:688369 [0] NCCL INFO Channel 06/0 : 32[0] -> 33[1] via P2P/IPC
+ip-26-0-160-225:94342:94504 [0] NCCL INFO Connected all rings
+ip-26-0-160-225:94342:94504 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-225:94343:94508 [1] NCCL INFO Connected all rings
+ip-26-0-160-225:94342:94504 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-252:688200:688369 [0] NCCL INFO Channel 07/0 : 32[0] -> 33[1] via P2P/IPC
+ip-26-0-172-252:688200:688369 [0] NCCL INFO Channel 10/0 : 32[0] -> 33[1] via P2P/IPC
+ip-26-0-172-252:688200:688369 [0] NCCL INFO Channel 11/0 : 32[0] -> 33[1] via P2P/IPC
+ip-26-0-172-142:880680:880843 [1] NCCL INFO Connected all rings
+ip-26-0-172-252:688200:688369 [0] NCCL INFO Channel 12/0 : 32[0] -> 33[1] via P2P/IPC
+ip-26-0-172-252:688200:688369 [0] NCCL INFO Channel 13/0 : 32[0] -> 33[1] via P2P/IPC
+ip-26-0-172-252:688200:688369 [0] NCCL INFO Channel 14/0 : 32[0] -> 33[1] via P2P/IPC
+ip-26-0-172-252:688201:688364 [1] NCCL INFO Channel 01/0 : 33[1] -> 34[2] via P2P/IPC
+ip-26-0-172-252:688200:688369 [0] NCCL INFO Channel 15/0 : 32[0] -> 33[1] via P2P/IPC
+ip-26-0-172-147:618591:618746 [7] NCCL INFO Connected all rings
+ip-26-0-172-252:688201:688364 [1] NCCL INFO Channel 03/0 : 33[1] -> 34[2] via P2P/IPC
+ip-26-0-172-252:688201:688364 [1] NCCL INFO Channel 04/0 : 33[1] -> 34[2] via P2P/IPC
+ip-26-0-172-252:688201:688364 [1] NCCL INFO Channel 05/0 : 33[1] -> 34[2] via P2P/IPC
+ip-26-0-172-142:880680:880843 [1] NCCL INFO Channel 01/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-172-252:688201:688364 [1] NCCL INFO Channel 06/0 : 33[1] -> 34[2] via P2P/IPC
+ip-26-0-172-252:688201:688364 [1] NCCL INFO Channel 07/0 : 33[1] -> 34[2] via P2P/IPC
+ip-26-0-172-252:688201:688364 [1] NCCL INFO Channel 09/0 : 33[1] -> 34[2] via P2P/IPC
+ip-26-0-172-252:688201:688364 [1] NCCL INFO Channel 11/0 : 33[1] -> 34[2] via P2P/IPC
+ip-26-0-172-252:688201:688364 [1] NCCL INFO Channel 12/0 : 33[1] -> 34[2] via P2P/IPC
+ip-26-0-172-252:688201:688364 [1] NCCL INFO Channel 13/0 : 33[1] -> 34[2] via P2P/IPC
+ip-26-0-172-252:688201:688364 [1] NCCL INFO Channel 14/0 : 33[1] -> 34[2] via P2P/IPC
+ip-26-0-172-142:880680:880843 [1] NCCL INFO Channel 03/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-172-252:688201:688364 [1] NCCL INFO Channel 15/0 : 33[1] -> 34[2] via P2P/IPC
+ip-26-0-172-252:688200:688369 [0] NCCL INFO Channel 05/0 : 32[0] -> 39[7] via P2P/IPC
+ip-26-0-172-252:688200:688369 [0] NCCL INFO Channel 06/0 : 32[0] -> 39[7] via P2P/IPC
+ip-26-0-160-225:94348:94506 [6] NCCL INFO Connected all rings
+ip-26-0-172-142:880680:880843 [1] NCCL INFO Channel 04/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-172-252:688200:688369 [0] NCCL INFO Channel 07/0 : 32[0] -> 39[7] via P2P/IPC
+ip-26-0-172-147:618586:618745 [2] NCCL INFO Connected all rings
+ip-26-0-172-142:880680:880843 [1] NCCL INFO Channel 05/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-172-142:880680:880843 [1] NCCL INFO Channel 06/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-172-142:880679:880846 [0] NCCL INFO Channel 14/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-172-142:880680:880843 [1] NCCL INFO Channel 07/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-172-252:688200:688369 [0] NCCL INFO Channel 13/0 : 32[0] -> 39[7] via P2P/IPC
+ip-26-0-172-252:688200:688369 [0] NCCL INFO Channel 14/0 : 32[0] -> 39[7] via P2P/IPC
+ip-26-0-172-252:688200:688369 [0] NCCL INFO Channel 15/0 : 32[0] -> 39[7] via P2P/IPC
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO Connected all rings
+ip-26-0-172-142:880679:880846 [0] NCCL INFO Channel 15/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-172-142:880680:880843 [1] NCCL INFO Channel 09/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-172-142:880680:880843 [1] NCCL INFO Channel 11/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-172-142:880680:880843 [1] NCCL INFO Channel 12/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-172-142:880680:880843 [1] NCCL INFO Channel 13/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-172-142:880681:880845 [2] NCCL INFO Channel 01/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-172-142:880680:880843 [1] NCCL INFO Channel 14/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-172-142:880681:880845 [2] NCCL INFO Channel 02/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-172-142:880680:880843 [1] NCCL INFO Channel 15/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-172-142:880681:880845 [2] NCCL INFO Channel 04/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-172-142:880681:880845 [2] NCCL INFO Channel 05/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-172-147:618584:618747 [0] NCCL INFO Connected all rings
+ip-26-0-172-142:880679:880846 [0] NCCL INFO Channel 05/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-172-147:618584:618747 [0] NCCL INFO Channel 02/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-172-142:880681:880845 [2] NCCL INFO Channel 06/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-172-142:880679:880846 [0] NCCL INFO Channel 06/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO Connected all rings
+ip-26-0-172-142:880681:880845 [2] NCCL INFO Channel 07/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-172-142:880679:880846 [0] NCCL INFO Channel 07/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-172-142:880681:880845 [2] NCCL INFO Channel 09/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-172-142:880679:880846 [0] NCCL INFO Channel 13/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-172-142:880681:880845 [2] NCCL INFO Channel 10/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-172-147:618589:618744 [5] NCCL INFO Connected all rings
+ip-26-0-172-142:880679:880846 [0] NCCL INFO Channel 14/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-172-142:880681:880845 [2] NCCL INFO Channel 12/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-172-147:618584:618747 [0] NCCL INFO Channel 03/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-172-142:880679:880846 [0] NCCL INFO Channel 15/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-172-147:618584:618747 [0] NCCL INFO Channel 04/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-172-142:880681:880845 [2] NCCL INFO Channel 13/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-172-147:618585:618748 [1] NCCL INFO Channel 01/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-172-147:618584:618747 [0] NCCL INFO Channel 05/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-172-142:880686:880841 [7] NCCL INFO Connected all rings
+ip-26-0-172-142:880681:880845 [2] NCCL INFO Channel 14/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-172-142:880685:880847 [6] NCCL INFO Channel 01/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-172-142:880681:880845 [2] NCCL INFO Channel 15/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-172-147:618585:618748 [1] NCCL INFO Channel 03/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-172-147:618584:618747 [0] NCCL INFO Channel 06/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-172-147:618585:618748 [1] NCCL INFO Channel 04/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-172-147:618584:618747 [0] NCCL INFO Channel 07/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-172-147:618585:618748 [1] NCCL INFO Channel 05/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-172-147:618584:618747 [0] NCCL INFO Channel 10/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-172-142:880685:880847 [6] NCCL INFO Channel 02/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-172-142:880680:880843 [1] NCCL INFO Channel 01/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880680:880843 [1] NCCL INFO Channel 09/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880685:880847 [6] NCCL INFO Channel 03/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-172-142:880680:880843 [1] NCCL INFO Channel 01/0 : 17[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618585:618748 [1] NCCL INFO Channel 06/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-172-147:618584:618747 [0] NCCL INFO Channel 11/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-172-142:880685:880847 [6] NCCL INFO Channel 04/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-172-147:618585:618748 [1] NCCL INFO Channel 07/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-172-147:618584:618747 [0] NCCL INFO Channel 12/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-172-147:618585:618748 [1] NCCL INFO Channel 09/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-172-142:880685:880847 [6] NCCL INFO Channel 05/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-172-147:618584:618747 [0] NCCL INFO Channel 13/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-172-147:618585:618748 [1] NCCL INFO Channel 11/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-172-147:618584:618747 [0] NCCL INFO Channel 14/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-172-147:618585:618748 [1] NCCL INFO Channel 12/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-172-147:618584:618747 [0] NCCL INFO Channel 15/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-172-147:618585:618748 [1] NCCL INFO Channel 13/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-172-142:880685:880847 [6] NCCL INFO Channel 06/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-172-147:618585:618748 [1] NCCL INFO Channel 14/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-172-147:618585:618748 [1] NCCL INFO Channel 15/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-172-142:880685:880847 [6] NCCL INFO Channel 09/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-172-147:618584:618747 [0] NCCL INFO Channel 05/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-172-142:880685:880847 [6] NCCL INFO Channel 10/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-172-252:688205:688365 [5] NCCL INFO Connected all rings
+ip-26-0-172-147:618584:618747 [0] NCCL INFO Channel 06/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-172-142:880685:880847 [6] NCCL INFO Channel 11/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-172-147:618584:618747 [0] NCCL INFO Channel 07/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-172-142:880685:880847 [6] NCCL INFO Channel 12/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-172-147:618584:618747 [0] NCCL INFO Channel 13/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-172-142:880685:880847 [6] NCCL INFO Channel 13/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-172-147:618584:618747 [0] NCCL INFO Channel 14/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-172-142:880685:880847 [6] NCCL INFO Channel 14/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-172-147:618584:618747 [0] NCCL INFO Channel 15/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-172-142:880683:880844 [4] NCCL INFO Channel 01/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-172-142:880683:880844 [4] NCCL INFO Channel 02/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-172-142:880683:880844 [4] NCCL INFO Channel 03/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-172-142:880683:880844 [4] NCCL INFO Channel 04/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-172-142:880683:880844 [4] NCCL INFO Channel 06/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-172-142:880686:880841 [7] NCCL INFO Channel 07/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880679:880846 [0] NCCL INFO Channel 00/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880686:880841 [7] NCCL INFO Channel 15/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880679:880846 [0] NCCL INFO Channel 08/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880683:880844 [4] NCCL INFO Channel 07/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-172-142:880686:880841 [7] NCCL INFO Channel 07/0 : 23[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880679:880846 [0] NCCL INFO Channel 00/0 : 16[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880683:880844 [4] NCCL INFO Channel 09/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-172-142:880683:880844 [4] NCCL INFO Channel 10/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-172-142:880683:880844 [4] NCCL INFO Channel 11/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-172-252:688205:688365 [5] NCCL INFO Channel 01/0 : 37[5] -> 38[6] via P2P/IPC
+ip-26-0-172-252:688205:688365 [5] NCCL INFO Channel 02/0 : 37[5] -> 38[6] via P2P/IPC
+ip-26-0-172-142:880683:880844 [4] NCCL INFO Channel 12/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-172-142:880683:880844 [4] NCCL INFO Channel 14/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-172-252:688205:688365 [5] NCCL INFO Channel 03/0 : 37[5] -> 38[6] via P2P/IPC
+ip-26-0-172-142:880683:880844 [4] NCCL INFO Channel 15/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-172-252:688205:688365 [5] NCCL INFO Channel 04/0 : 37[5] -> 38[6] via P2P/IPC
+ip-26-0-172-252:688205:688365 [5] NCCL INFO Channel 05/0 : 37[5] -> 38[6] via P2P/IPC
+ip-26-0-172-147:618590:618749 [6] NCCL INFO Connected all rings
+ip-26-0-160-225:94344:94510 [2] NCCL INFO Connected all rings
+ip-26-0-172-252:688205:688365 [5] NCCL INFO Channel 07/0 : 37[5] -> 38[6] via P2P/IPC
+ip-26-0-172-252:688205:688365 [5] NCCL INFO Channel 09/0 : 37[5] -> 38[6] via P2P/IPC
+ip-26-0-172-252:688205:688365 [5] NCCL INFO Channel 10/0 : 37[5] -> 38[6] via P2P/IPC
+ip-26-0-172-252:688205:688365 [5] NCCL INFO Channel 11/0 : 37[5] -> 38[6] via P2P/IPC
+ip-26-0-172-252:688205:688365 [5] NCCL INFO Channel 12/0 : 37[5] -> 38[6] via P2P/IPC
+ip-26-0-172-252:688205:688365 [5] NCCL INFO Channel 13/0 : 37[5] -> 38[6] via P2P/IPC
+ip-26-0-172-252:688205:688365 [5] NCCL INFO Channel 15/0 : 37[5] -> 38[6] via P2P/IPC
+ip-26-0-160-225:94342:94504 [0] NCCL INFO Channel 04/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-142:880684:880842 [5] NCCL INFO Connected all rings
+ip-26-0-172-147:618589:618744 [5] NCCL INFO Channel 01/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-160-225:94342:94504 [0] NCCL INFO Channel 05/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-225:94342:94504 [0] NCCL INFO Channel 06/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-225:94342:94504 [0] NCCL INFO Channel 07/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-147:618589:618744 [5] NCCL INFO Channel 02/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-160-225:94342:94504 [0] NCCL INFO Channel 10/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-225:94343:94508 [1] NCCL INFO Channel 01/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-147:618590:618749 [6] NCCL INFO Channel 01/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-172-147:618589:618744 [5] NCCL INFO Channel 03/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-172-147:618590:618749 [6] NCCL INFO Channel 02/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-172-147:618589:618744 [5] NCCL INFO Channel 04/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-172-142:880684:880842 [5] NCCL INFO Channel 01/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-160-225:94342:94504 [0] NCCL INFO Channel 11/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-142:880684:880842 [5] NCCL INFO Channel 02/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-172-147:618590:618749 [6] NCCL INFO Channel 03/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-160-225:94343:94508 [1] NCCL INFO Channel 03/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-142:880684:880842 [5] NCCL INFO Channel 03/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-160-225:94342:94504 [0] NCCL INFO Channel 12/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-142:880684:880842 [5] NCCL INFO Channel 04/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-172-142:880684:880842 [5] NCCL INFO Channel 05/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-160-225:94343:94508 [1] NCCL INFO Channel 04/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-147:618589:618744 [5] NCCL INFO Channel 05/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-172-142:880684:880842 [5] NCCL INFO Channel 07/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-172-147:618590:618749 [6] NCCL INFO Channel 04/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-172-147:618589:618744 [5] NCCL INFO Channel 07/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-172-147:618590:618749 [6] NCCL INFO Channel 05/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-172-142:880684:880842 [5] NCCL INFO Channel 09/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-160-225:94342:94504 [0] NCCL INFO Channel 13/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-147:618589:618744 [5] NCCL INFO Channel 09/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-172-142:880684:880842 [5] NCCL INFO Channel 10/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-172-142:880684:880842 [5] NCCL INFO Channel 11/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-172-147:618590:618749 [6] NCCL INFO Channel 06/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-172-142:880684:880842 [5] NCCL INFO Channel 12/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-172-147:618589:618744 [5] NCCL INFO Channel 10/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-172-142:880684:880842 [5] NCCL INFO Channel 13/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO Connected all rings
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO Channel 02/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-172-147:618590:618749 [6] NCCL INFO Channel 09/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-160-225:94345:94509 [3] NCCL INFO Connected all rings
+ip-26-0-172-147:618589:618744 [5] NCCL INFO Channel 11/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-172-142:880684:880842 [5] NCCL INFO Channel 15/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-160-225:94343:94508 [1] NCCL INFO Channel 05/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-147:618590:618749 [6] NCCL INFO Channel 10/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-172-142:880685:880847 [6] NCCL INFO Channel 06/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880684:880842 [5] NCCL INFO Channel 05/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:94342:94504 [0] NCCL INFO Channel 14/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-142:880685:880847 [6] NCCL INFO Channel 14/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880684:880842 [5] NCCL INFO Channel 13/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618589:618744 [5] NCCL INFO Channel 12/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-172-142:880685:880847 [6] NCCL INFO Channel 06/0 : 22[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880684:880842 [5] NCCL INFO Channel 05/0 : 21[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO Channel 03/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-172-147:618590:618749 [6] NCCL INFO Channel 11/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-172-147:618589:618744 [5] NCCL INFO Channel 13/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-160-225:94343:94508 [1] NCCL INFO Channel 06/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-147:618590:618749 [6] NCCL INFO Channel 12/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-160-225:94342:94504 [0] NCCL INFO Channel 15/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-147:618589:618744 [5] NCCL INFO Channel 15/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-160-225:94343:94508 [1] NCCL INFO Channel 07/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO Channel 04/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO Connected all rings
+ip-26-0-160-225:94343:94508 [1] NCCL INFO Channel 09/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-147:618590:618749 [6] NCCL INFO Channel 13/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-160-225:94347:94507 [5] NCCL INFO Connected all rings
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO Channel 05/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-160-225:94343:94508 [1] NCCL INFO Channel 11/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO Channel 06/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-172-147:618587:618751 [3] NCCL INFO Connected all rings
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO Channel 07/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-160-225:94343:94508 [1] NCCL INFO Channel 12/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-225:94346:94511 [4] NCCL INFO Connected all rings
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO Connected all rings
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO Channel 10/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-172-147:618590:618749 [6] NCCL INFO Channel 14/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-172-142:880682:880840 [3] NCCL INFO Connected all rings
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO Channel 11/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-172-147:618590:618749 [6] NCCL INFO Channel 06/0 : 22[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688203:688362 [3] NCCL INFO Connected all rings
+ip-26-0-172-147:618590:618749 [6] NCCL INFO Channel 14/0 : 14[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO Channel 12/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-172-147:618590:618749 [6] NCCL INFO Channel 14/0 : 30[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:94349:94505 [7] NCCL INFO Connected all rings
+ip-26-0-160-225:94343:94508 [1] NCCL INFO Channel 13/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO Channel 13/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-172-147:618586:618745 [2] NCCL INFO Channel 01/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-160-225:94343:94508 [1] NCCL INFO Channel 14/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-147:618586:618745 [2] NCCL INFO Channel 02/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO Channel 14/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-172-147:618586:618745 [2] NCCL INFO Channel 04/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-173-7:811252:811408 [4] NCCL INFO Connected all rings
+ip-26-0-172-142:880682:880840 [3] NCCL INFO Channel 01/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-172-142:880682:880840 [3] NCCL INFO Channel 02/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-172-147:618586:618745 [2] NCCL INFO Channel 05/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-160-225:94343:94508 [1] NCCL INFO Channel 15/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-142:880682:880840 [3] NCCL INFO Channel 03/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-172-142:880682:880840 [3] NCCL INFO Channel 05/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-172-147:618586:618745 [2] NCCL INFO Channel 06/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-172-142:880682:880840 [3] NCCL INFO Channel 06/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-172-147:618586:618745 [2] NCCL INFO Channel 07/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-172-142:880682:880840 [3] NCCL INFO Channel 07/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-172-142:880682:880840 [3] NCCL INFO Channel 09/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO Connected all rings
+ip-26-0-172-147:618586:618745 [2] NCCL INFO Channel 09/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-160-225:94342:94504 [0] NCCL INFO Channel 05/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-172-142:880682:880840 [3] NCCL INFO Channel 10/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO Channel 15/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-172-142:880682:880840 [3] NCCL INFO Channel 11/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-172-147:618586:618745 [2] NCCL INFO Channel 10/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-172-142:880682:880840 [3] NCCL INFO Channel 13/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-172-142:880682:880840 [3] NCCL INFO Channel 14/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-172-142:880682:880840 [3] NCCL INFO Channel 15/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-160-225:94342:94504 [0] NCCL INFO Channel 06/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-172-147:618586:618745 [2] NCCL INFO Channel 12/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-160-225:94344:94510 [2] NCCL INFO Channel 01/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-142:880683:880844 [4] NCCL INFO Channel 04/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880682:880840 [3] NCCL INFO Channel 03/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880683:880844 [4] NCCL INFO Channel 12/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880682:880840 [3] NCCL INFO Channel 11/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880683:880844 [4] NCCL INFO Channel 04/0 : 20[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880682:880840 [3] NCCL INFO Channel 03/0 : 19[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880681:880845 [2] NCCL INFO Channel 02/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:94342:94504 [0] NCCL INFO Channel 07/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-172-252:688204:688367 [4] NCCL INFO Connected all rings
+ip-26-0-172-142:880681:880845 [2] NCCL INFO Channel 10/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:94344:94510 [2] NCCL INFO Channel 02/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-142:880681:880845 [2] NCCL INFO Channel 02/0 : 18[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-7:811252:811408 [4] NCCL INFO Channel 01/0 : 60[4] -> 61[5] via P2P/IPC
+ip-26-0-160-225:94345:94509 [3] NCCL INFO Channel 01/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO Channel 01/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-173-7:811252:811408 [4] NCCL INFO Channel 02/0 : 60[4] -> 61[5] via P2P/IPC
+ip-26-0-172-147:618586:618745 [2] NCCL INFO Channel 13/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-160-225:94342:94504 [0] NCCL INFO Channel 13/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-173-7:811252:811408 [4] NCCL INFO Channel 03/0 : 60[4] -> 61[5] via P2P/IPC
+ip-26-0-160-225:94344:94510 [2] NCCL INFO Channel 04/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO Channel 03/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-160-225:94345:94509 [3] NCCL INFO Channel 02/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-147:618586:618745 [2] NCCL INFO Channel 14/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-173-7:811252:811408 [4] NCCL INFO Channel 04/0 : 60[4] -> 61[5] via P2P/IPC
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO Channel 04/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-172-147:618586:618745 [2] NCCL INFO Channel 15/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-173-7:811252:811408 [4] NCCL INFO Channel 06/0 : 60[4] -> 61[5] via P2P/IPC
+ip-26-0-173-7:811252:811408 [4] NCCL INFO Channel 07/0 : 60[4] -> 61[5] via P2P/IPC
+ip-26-0-160-225:94342:94504 [0] NCCL INFO Channel 14/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-160-225:94344:94510 [2] NCCL INFO Channel 05/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-225:94345:94509 [3] NCCL INFO Channel 03/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-173-7:811252:811408 [4] NCCL INFO Channel 09/0 : 60[4] -> 61[5] via P2P/IPC
+ip-26-0-172-147:618585:618748 [1] NCCL INFO Channel 01/0 : 17[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688203:688362 [3] NCCL INFO Channel 01/0 : 35[3] -> 36[4] via P2P/IPC
+ip-26-0-173-7:811252:811408 [4] NCCL INFO Channel 10/0 : 60[4] -> 61[5] via P2P/IPC
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO Channel 05/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-172-147:618585:618748 [1] NCCL INFO Channel 09/0 : 9[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688203:688362 [3] NCCL INFO Channel 02/0 : 35[3] -> 36[4] via P2P/IPC
+ip-26-0-172-147:618585:618748 [1] NCCL INFO Channel 09/0 : 25[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-7:811252:811408 [4] NCCL INFO Channel 11/0 : 60[4] -> 61[5] via P2P/IPC
+ip-26-0-172-252:688203:688362 [3] NCCL INFO Channel 03/0 : 35[3] -> 36[4] via P2P/IPC
+ip-26-0-160-225:94344:94510 [2] NCCL INFO Channel 06/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-225:94342:94504 [0] NCCL INFO Channel 15/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-160-225:94345:94509 [3] NCCL INFO Channel 05/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-173-7:811252:811408 [4] NCCL INFO Channel 12/0 : 60[4] -> 61[5] via P2P/IPC
+ip-26-0-172-252:688203:688362 [3] NCCL INFO Channel 05/0 : 35[3] -> 36[4] via P2P/IPC
+ip-26-0-173-7:811252:811408 [4] NCCL INFO Channel 14/0 : 60[4] -> 61[5] via P2P/IPC
+ip-26-0-172-252:688203:688362 [3] NCCL INFO Channel 06/0 : 35[3] -> 36[4] via P2P/IPC
+ip-26-0-173-7:811252:811408 [4] NCCL INFO Channel 15/0 : 60[4] -> 61[5] via P2P/IPC
+ip-26-0-172-252:688203:688362 [3] NCCL INFO Channel 07/0 : 35[3] -> 36[4] via P2P/IPC
+ip-26-0-172-147:618587:618751 [3] NCCL INFO Channel 01/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-160-225:94344:94510 [2] NCCL INFO Channel 07/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-225:94345:94509 [3] NCCL INFO Channel 06/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-225:94347:94507 [5] NCCL INFO Channel 01/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO Connected all rings
+ip-26-0-172-252:688203:688362 [3] NCCL INFO Channel 09/0 : 35[3] -> 36[4] via P2P/IPC
+ip-26-0-172-252:688204:688367 [4] NCCL INFO Channel 01/0 : 36[4] -> 37[5] via P2P/IPC
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO Channel 06/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-172-252:688203:688362 [3] NCCL INFO Channel 10/0 : 35[3] -> 36[4] via P2P/IPC
+ip-26-0-172-147:618587:618751 [3] NCCL INFO Channel 02/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-173-7:811255:811411 [7] NCCL INFO Connected all rings
+ip-26-0-172-252:688204:688367 [4] NCCL INFO Channel 02/0 : 36[4] -> 37[5] via P2P/IPC
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO Channel 01/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-160-225:94344:94510 [2] NCCL INFO Channel 09/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-252:688203:688362 [3] NCCL INFO Channel 11/0 : 35[3] -> 36[4] via P2P/IPC
+ip-26-0-160-225:94345:94509 [3] NCCL INFO Channel 07/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-225:94348:94506 [6] NCCL INFO Channel 01/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-225:94347:94507 [5] NCCL INFO Channel 02/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-252:688204:688367 [4] NCCL INFO Channel 03/0 : 36[4] -> 37[5] via P2P/IPC
+ip-26-0-172-252:688203:688362 [3] NCCL INFO Channel 13/0 : 35[3] -> 36[4] via P2P/IPC
+ip-26-0-160-225:94346:94511 [4] NCCL INFO Channel 01/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-252:688204:688367 [4] NCCL INFO Channel 04/0 : 36[4] -> 37[5] via P2P/IPC
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO Channel 07/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-172-252:688203:688362 [3] NCCL INFO Channel 14/0 : 35[3] -> 36[4] via P2P/IPC
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO Channel 01/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO Connected all rings
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO Channel 02/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-172-252:688204:688367 [4] NCCL INFO Channel 06/0 : 36[4] -> 37[5] via P2P/IPC
+ip-26-0-160-225:94344:94510 [2] NCCL INFO Channel 10/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-225:94348:94506 [6] NCCL INFO Channel 02/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-147:618588:618750 [4] NCCL INFO Connected all rings
+ip-26-0-160-225:94347:94507 [5] NCCL INFO Channel 03/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-252:688203:688362 [3] NCCL INFO Channel 15/0 : 35[3] -> 36[4] via P2P/IPC
+ip-26-0-160-225:94345:94509 [3] NCCL INFO Channel 09/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-252:688204:688367 [4] NCCL INFO Channel 07/0 : 36[4] -> 37[5] via P2P/IPC
+ip-26-0-160-225:94346:94511 [4] NCCL INFO Channel 02/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-147:618587:618751 [3] NCCL INFO Channel 03/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-172-252:688204:688367 [4] NCCL INFO Channel 09/0 : 36[4] -> 37[5] via P2P/IPC
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO Channel 09/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO Channel 02/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-160-225:94344:94510 [2] NCCL INFO Channel 12/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-225:94348:94506 [6] NCCL INFO Channel 03/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-252:688204:688367 [4] NCCL INFO Channel 10/0 : 36[4] -> 37[5] via P2P/IPC
+ip-26-0-160-225:94347:94507 [5] NCCL INFO Channel 04/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-147:618587:618751 [3] NCCL INFO Channel 05/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-160-225:94345:94509 [3] NCCL INFO Channel 10/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO Channel 04/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-160-225:94346:94511 [4] NCCL INFO Channel 03/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-252:688204:688367 [4] NCCL INFO Channel 11/0 : 36[4] -> 37[5] via P2P/IPC
+ip-26-0-172-252:688204:688367 [4] NCCL INFO Channel 12/0 : 36[4] -> 37[5] via P2P/IPC
+ip-26-0-172-147:618587:618751 [3] NCCL INFO Channel 06/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-160-225:94348:94506 [6] NCCL INFO Channel 04/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-225:94347:94507 [5] NCCL INFO Channel 05/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-225:94345:94509 [3] NCCL INFO Channel 11/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-225:94344:94510 [2] NCCL INFO Channel 13/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO Channel 11/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-160-225:94346:94511 [4] NCCL INFO Channel 04/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-147:618587:618751 [3] NCCL INFO Channel 07/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO Channel 03/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-172-252:688204:688367 [4] NCCL INFO Channel 14/0 : 36[4] -> 37[5] via P2P/IPC
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO Channel 05/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-172-147:618587:618751 [3] NCCL INFO Channel 09/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-160-225:94347:94507 [5] NCCL INFO Channel 07/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-225:94348:94506 [6] NCCL INFO Channel 05/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO Channel 01/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-160-225:94345:94509 [3] NCCL INFO Channel 13/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-225:94344:94510 [2] NCCL INFO Channel 14/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-225:94346:94511 [4] NCCL INFO Channel 06/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-147:618587:618751 [3] NCCL INFO Channel 10/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO Channel 12/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-160-225:94347:94507 [5] NCCL INFO Channel 09/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-225:94348:94506 [6] NCCL INFO Channel 06/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO Channel 05/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO Channel 06/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-172-147:618587:618751 [3] NCCL INFO Channel 11/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-160-225:94345:94509 [3] NCCL INFO Channel 14/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-225:94344:94510 [2] NCCL INFO Channel 15/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-225:94346:94511 [4] NCCL INFO Channel 07/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO Channel 02/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-172-147:618591:618746 [7] NCCL INFO Channel 07/0 : 23[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:94347:94507 [5] NCCL INFO Channel 10/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-252:688204:688367 [4] NCCL INFO Channel 15/0 : 36[4] -> 37[5] via P2P/IPC
+ip-26-0-172-147:618587:618751 [3] NCCL INFO Channel 13/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-160-225:94348:94506 [6] NCCL INFO Channel 09/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-147:618591:618746 [7] NCCL INFO Channel 15/0 : 15[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO Channel 13/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-172-147:618591:618746 [7] NCCL INFO Channel 15/0 : 31[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:94345:94509 [3] NCCL INFO Channel 15/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-147:618587:618751 [3] NCCL INFO Channel 14/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-160-225:94346:94511 [4] NCCL INFO Channel 09/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO Channel 06/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO Channel 07/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-160-225:94347:94507 [5] NCCL INFO Channel 11/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO Channel 01/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-172-252:688204:688367 [4] NCCL INFO Channel 12/0 : 36[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926595:926760 [1] NCCL INFO Connected all rings
+ip-26-0-172-73:926594:926762 [0] NCCL INFO Connected all rings
+ip-26-0-172-147:618587:618751 [3] NCCL INFO Channel 15/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-160-225:94348:94506 [6] NCCL INFO Channel 10/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO Channel 03/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-172-73:926594:926762 [0] NCCL INFO Channel 02/0 : 48[0] -> 49[1] via P2P/IPC
+ip-26-0-160-225:94346:94511 [4] NCCL INFO Channel 10/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-147:618588:618750 [4] NCCL INFO Channel 01/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-160-225:94347:94507 [5] NCCL INFO Channel 12/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-225:94344:94510 [2] NCCL INFO Channel 10/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:94348:94506 [6] NCCL INFO Channel 11/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO Channel 14/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-160-225:94346:94511 [4] NCCL INFO Channel 11/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-147:618586:618745 [2] NCCL INFO Channel 02/0 : 18[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618588:618750 [4] NCCL INFO Channel 02/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-160-225:94343:94508 [1] NCCL INFO Channel 09/0 : 1[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO Channel 07/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO Channel 09/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-172-147:618586:618745 [2] NCCL INFO Channel 10/0 : 10[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:94347:94507 [5] NCCL INFO Channel 13/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-225:94348:94506 [6] NCCL INFO Channel 12/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-147:618588:618750 [4] NCCL INFO Channel 03/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-160-225:94346:94511 [4] NCCL INFO Channel 12/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO Channel 02/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-172-147:618586:618745 [2] NCCL INFO Channel 10/0 : 26[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO Channel 04/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-172-147:618588:618750 [4] NCCL INFO Channel 04/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-160-225:94347:94507 [5] NCCL INFO Channel 15/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-225:94348:94506 [6] NCCL INFO Channel 13/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-225:94346:94511 [4] NCCL INFO Channel 14/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-147:618588:618750 [4] NCCL INFO Channel 06/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO Channel 15/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-172-147:618584:618747 [0] NCCL INFO Channel 00/0 : 16[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618588:618750 [4] NCCL INFO Channel 07/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO Channel 09/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-160-225:94348:94506 [6] NCCL INFO Channel 14/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-225:94346:94511 [4] NCCL INFO Channel 15/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO Channel 10/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-172-147:618584:618747 [0] NCCL INFO Channel 08/0 : 8[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618588:618750 [4] NCCL INFO Channel 09/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO Channel 03/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-172-147:618584:618747 [0] NCCL INFO Channel 08/0 : 24[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO Channel 05/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-172-147:618588:618750 [4] NCCL INFO Channel 10/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-172-147:618588:618750 [4] NCCL INFO Channel 11/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-160-225:94345:94509 [3] NCCL INFO Channel 11/0 : 3[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:94348:94506 [6] NCCL INFO Channel 14/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:94347:94507 [5] NCCL INFO Channel 13/0 : 5[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:94346:94511 [4] NCCL INFO Channel 12/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO Channel 05/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-172-147:618588:618750 [4] NCCL INFO Channel 12/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO Channel 10/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-172-147:618588:618750 [4] NCCL INFO Channel 14/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO Channel 12/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO Channel 04/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO Channel 07/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-172-147:618588:618750 [4] NCCL INFO Channel 15/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO Channel 06/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-172-147:618587:618751 [3] NCCL INFO Channel 03/0 : 19[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO Channel 11/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-172-147:618589:618744 [5] NCCL INFO Channel 05/0 : 21[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618588:618750 [4] NCCL INFO Channel 04/0 : 20[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO Channel 13/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO Channel 05/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-172-147:618587:618751 [3] NCCL INFO Channel 11/0 : 11[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:94349:94505 [7] NCCL INFO Channel 15/0 : 7[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO Channel 09/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-160-225:94342:94504 [0] NCCL INFO Channel 08/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618588:618750 [4] NCCL INFO Channel 12/0 : 12[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618587:618751 [3] NCCL INFO Channel 11/0 : 27[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618589:618744 [5] NCCL INFO Channel 13/0 : 13[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618588:618750 [4] NCCL INFO Channel 12/0 : 28[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618589:618744 [5] NCCL INFO Channel 13/0 : 29[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO Channel 07/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO Channel 13/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO Channel 14/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO Channel 06/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO Channel 10/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO Channel 13/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO Channel 14/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-172-57:942613:942776 [2] NCCL INFO Connected all rings
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO Channel 15/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO Channel 09/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO Channel 01/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO Channel 11/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO Channel 14/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO Channel 15/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO Channel 10/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO Channel 09/0 : 1[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO Channel 02/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO Channel 12/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO Channel 01/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO Channel 09/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO Channel 15/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-160-225:94343:94508 [1] NCCL INFO Channel 01/0 : 33[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:94343:94508 [1] NCCL INFO Channel 01/0 : 1[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO Channel 09/0 : 9[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO Channel 11/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO Channel 10/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO Channel 02/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880680:880843 [1] NCCL INFO Channel 01/0 : 17[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO Channel 03/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO Channel 13/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO Channel 09/0 : 25[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO Channel 10/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO Channel 12/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-160-225:94344:94510 [2] NCCL INFO Channel 02/0 : 34[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:94344:94510 [2] NCCL INFO Channel 02/0 : 2[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO Channel 04/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO Channel 10/0 : 10[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO Channel 15/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-172-142:880681:880845 [2] NCCL INFO Channel 02/0 : 18[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO Channel 13/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO Channel 10/0 : 26[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO Channel 06/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO Channel 14/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO Channel 07/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO Channel 09/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO Channel 14/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO Channel 10/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO Channel 06/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO Channel 14/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO Channel 11/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-160-225:94348:94506 [6] NCCL INFO Channel 06/0 : 38[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO Channel 12/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-160-225:94348:94506 [6] NCCL INFO Channel 06/0 : 6[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880685:880847 [6] NCCL INFO Channel 06/0 : 22[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO Channel 14/0 : 14[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO Channel 14/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-172-57:942618:942778 [7] NCCL INFO Connected all rings
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO Channel 14/0 : 30[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688202:688368 [2] NCCL INFO Connected all rings
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO Channel 15/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-172-73:926595:926760 [1] NCCL INFO Channel 01/0 : 49[1] -> 50[2] via P2P/IPC
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO Channel 11/0 : 3[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO Channel 13/0 : 5[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO Channel 15/0 : 7[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO Channel 12/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO Channel 08/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO Channel 03/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO Channel 07/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO Channel 05/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO Channel 04/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO Channel 00/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO Channel 11/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO Channel 15/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926595:926760 [1] NCCL INFO Channel 03/0 : 49[1] -> 50[2] via P2P/IPC
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO Channel 13/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO Channel 12/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO Channel 08/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:94345:94509 [3] NCCL INFO Channel 03/0 : 35[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO Channel 11/0 : 11[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:94347:94507 [5] NCCL INFO Channel 05/0 : 37[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:94342:94504 [0] NCCL INFO Channel 00/0 : 32[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:94349:94505 [7] NCCL INFO Channel 07/0 : 39[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:94346:94511 [4] NCCL INFO Channel 04/0 : 36[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:94345:94509 [3] NCCL INFO Channel 03/0 : 3[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:94347:94507 [5] NCCL INFO Channel 05/0 : 5[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO Channel 15/0 : 15[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880682:880840 [3] NCCL INFO Channel 03/0 : 19[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:94349:94505 [7] NCCL INFO Channel 07/0 : 7[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:94342:94504 [0] NCCL INFO Channel 00/0 : 0[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:94346:94511 [4] NCCL INFO Channel 04/0 : 4[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO Channel 08/0 : 8[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO Channel 12/0 : 12[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO Channel 13/0 : 13[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880679:880846 [0] NCCL INFO Channel 00/0 : 16[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880683:880844 [4] NCCL INFO Channel 04/0 : 20[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO Channel 11/0 : 27[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO Channel 15/0 : 31[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO Channel 08/0 : 24[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO Channel 12/0 : 28[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO Channel 13/0 : 29[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880684:880842 [5] NCCL INFO Channel 05/0 : 21[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880686:880841 [7] NCCL INFO Channel 07/0 : 23[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926595:926760 [1] NCCL INFO Channel 04/0 : 49[1] -> 50[2] via P2P/IPC
+ip-26-0-172-73:926595:926760 [1] NCCL INFO Channel 05/0 : 49[1] -> 50[2] via P2P/IPC
+ip-26-0-172-252:688202:688368 [2] NCCL INFO Channel 01/0 : 34[2] -> 35[3] via P2P/IPC
+ip-26-0-172-252:688202:688368 [2] NCCL INFO Channel 02/0 : 34[2] -> 35[3] via P2P/IPC
+ip-26-0-172-252:688202:688368 [2] NCCL INFO Channel 04/0 : 34[2] -> 35[3] via P2P/IPC
+ip-26-0-172-252:688202:688368 [2] NCCL INFO Channel 05/0 : 34[2] -> 35[3] via P2P/IPC
+ip-26-0-172-252:688202:688368 [2] NCCL INFO Channel 06/0 : 34[2] -> 35[3] via P2P/IPC
+ip-26-0-172-252:688202:688368 [2] NCCL INFO Channel 07/0 : 34[2] -> 35[3] via P2P/IPC
+ip-26-0-172-252:688202:688368 [2] NCCL INFO Channel 09/0 : 34[2] -> 35[3] via P2P/IPC
+ip-26-0-172-252:688202:688368 [2] NCCL INFO Channel 10/0 : 34[2] -> 35[3] via P2P/IPC
+ip-26-0-172-252:688202:688368 [2] NCCL INFO Channel 12/0 : 34[2] -> 35[3] via P2P/IPC
+ip-26-0-172-252:688202:688368 [2] NCCL INFO Channel 13/0 : 34[2] -> 35[3] via P2P/IPC
+ip-26-0-172-252:688202:688368 [2] NCCL INFO Channel 14/0 : 34[2] -> 35[3] via P2P/IPC
+ip-26-0-172-73:926594:926762 [0] NCCL INFO Channel 03/0 : 48[0] -> 49[1] via P2P/IPC
+ip-26-0-172-252:688202:688368 [2] NCCL INFO Channel 15/0 : 34[2] -> 35[3] via P2P/IPC
+ip-26-0-172-252:688201:688364 [1] NCCL INFO Channel 09/0 : 33[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688203:688362 [3] NCCL INFO Channel 11/0 : 35[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688202:688368 [2] NCCL INFO Channel 10/0 : 34[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926594:926762 [0] NCCL INFO Channel 04/0 : 48[0] -> 49[1] via P2P/IPC
+ip-26-0-172-73:926594:926762 [0] NCCL INFO Channel 05/0 : 48[0] -> 49[1] via P2P/IPC
+ip-26-0-172-73:926594:926762 [0] NCCL INFO Channel 06/0 : 48[0] -> 49[1] via P2P/IPC
+ip-26-0-172-73:926594:926762 [0] NCCL INFO Channel 07/0 : 48[0] -> 49[1] via P2P/IPC
+ip-26-0-173-7:811250:811412 [2] NCCL INFO Connected all rings
+ip-26-0-173-7:811254:811414 [6] NCCL INFO Connected all rings
+ip-26-0-173-7:811253:811410 [5] NCCL INFO Connected all rings
+ip-26-0-172-252:688206:688363 [6] NCCL INFO Connected all rings
+ip-26-0-172-57:942612:942779 [1] NCCL INFO Connected all rings
+ip-26-0-173-7:811251:811409 [3] NCCL INFO Connected all rings
+ip-26-0-173-7:811250:811412 [2] NCCL INFO Channel 01/0 : 58[2] -> 59[3] via P2P/IPC
+ip-26-0-173-7:811250:811412 [2] NCCL INFO Channel 02/0 : 58[2] -> 59[3] via P2P/IPC
+ip-26-0-173-7:811250:811412 [2] NCCL INFO Channel 04/0 : 58[2] -> 59[3] via P2P/IPC
+ip-26-0-172-252:688207:688366 [7] NCCL INFO Connected all rings
+ip-26-0-173-7:811250:811412 [2] NCCL INFO Channel 05/0 : 58[2] -> 59[3] via P2P/IPC
+ip-26-0-173-7:811250:811412 [2] NCCL INFO Channel 06/0 : 58[2] -> 59[3] via P2P/IPC
+ip-26-0-173-7:811250:811412 [2] NCCL INFO Channel 07/0 : 58[2] -> 59[3] via P2P/IPC
+ip-26-0-172-252:688206:688363 [6] NCCL INFO Channel 01/0 : 38[6] -> 39[7] via P2P/IPC
+ip-26-0-173-7:811250:811412 [2] NCCL INFO Channel 09/0 : 58[2] -> 59[3] via P2P/IPC
+ip-26-0-172-252:688206:688363 [6] NCCL INFO Channel 02/0 : 38[6] -> 39[7] via P2P/IPC
+ip-26-0-173-7:811250:811412 [2] NCCL INFO Channel 10/0 : 58[2] -> 59[3] via P2P/IPC
+ip-26-0-172-252:688206:688363 [6] NCCL INFO Channel 03/0 : 38[6] -> 39[7] via P2P/IPC
+ip-26-0-172-252:688206:688363 [6] NCCL INFO Channel 04/0 : 38[6] -> 39[7] via P2P/IPC
+ip-26-0-172-252:688206:688363 [6] NCCL INFO Channel 05/0 : 38[6] -> 39[7] via P2P/IPC
+ip-26-0-173-7:811253:811410 [5] NCCL INFO Channel 01/0 : 61[5] -> 62[6] via P2P/IPC
+ip-26-0-172-252:688206:688363 [6] NCCL INFO Channel 06/0 : 38[6] -> 39[7] via P2P/IPC
+ip-26-0-173-7:811250:811412 [2] NCCL INFO Channel 12/0 : 58[2] -> 59[3] via P2P/IPC
+ip-26-0-173-7:811254:811414 [6] NCCL INFO Channel 01/0 : 62[6] -> 63[7] via P2P/IPC
+ip-26-0-173-7:811253:811410 [5] NCCL INFO Channel 02/0 : 61[5] -> 62[6] via P2P/IPC
+ip-26-0-173-7:811250:811412 [2] NCCL INFO Channel 13/0 : 58[2] -> 59[3] via P2P/IPC
+ip-26-0-173-7:811254:811414 [6] NCCL INFO Channel 02/0 : 62[6] -> 63[7] via P2P/IPC
+ip-26-0-173-7:811253:811410 [5] NCCL INFO Channel 03/0 : 61[5] -> 62[6] via P2P/IPC
+ip-26-0-173-7:811250:811412 [2] NCCL INFO Channel 14/0 : 58[2] -> 59[3] via P2P/IPC
+ip-26-0-173-7:811254:811414 [6] NCCL INFO Channel 03/0 : 62[6] -> 63[7] via P2P/IPC
+ip-26-0-172-252:688206:688363 [6] NCCL INFO Channel 09/0 : 38[6] -> 39[7] via P2P/IPC
+ip-26-0-172-252:688206:688363 [6] NCCL INFO Channel 10/0 : 38[6] -> 39[7] via P2P/IPC
+ip-26-0-172-252:688206:688363 [6] NCCL INFO Channel 11/0 : 38[6] -> 39[7] via P2P/IPC
+ip-26-0-172-252:688206:688363 [6] NCCL INFO Channel 12/0 : 38[6] -> 39[7] via P2P/IPC
+ip-26-0-172-252:688206:688363 [6] NCCL INFO Channel 13/0 : 38[6] -> 39[7] via P2P/IPC
+ip-26-0-172-252:688206:688363 [6] NCCL INFO Channel 14/0 : 38[6] -> 39[7] via P2P/IPC
+ip-26-0-173-7:811253:811410 [5] NCCL INFO Channel 04/0 : 61[5] -> 62[6] via P2P/IPC
+ip-26-0-172-252:688205:688365 [5] NCCL INFO Channel 13/0 : 37[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-7:811254:811414 [6] NCCL INFO Channel 04/0 : 62[6] -> 63[7] via P2P/IPC
+ip-26-0-172-252:688206:688363 [6] NCCL INFO Channel 14/0 : 38[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-7:811251:811409 [3] NCCL INFO Channel 01/0 : 59[3] -> 60[4] via P2P/IPC
+ip-26-0-173-7:811250:811412 [2] NCCL INFO Channel 15/0 : 58[2] -> 59[3] via P2P/IPC
+ip-26-0-173-7:811253:811410 [5] NCCL INFO Channel 05/0 : 61[5] -> 62[6] via P2P/IPC
+ip-26-0-173-7:811254:811414 [6] NCCL INFO Channel 05/0 : 62[6] -> 63[7] via P2P/IPC
+ip-26-0-173-7:811248:811415 [0] NCCL INFO Connected all rings
+ip-26-0-173-7:811248:811415 [0] NCCL INFO Channel 02/0 : 56[0] -> 57[1] via P2P/IPC
+ip-26-0-173-7:811251:811409 [3] NCCL INFO Channel 02/0 : 59[3] -> 60[4] via P2P/IPC
+ip-26-0-172-57:942614:942773 [3] NCCL INFO Connected all rings
+ip-26-0-172-57:942612:942779 [1] NCCL INFO Channel 01/0 : 41[1] -> 42[2] via P2P/IPC
+ip-26-0-173-7:811253:811410 [5] NCCL INFO Channel 07/0 : 61[5] -> 62[6] via P2P/IPC
+ip-26-0-173-7:811254:811414 [6] NCCL INFO Channel 06/0 : 62[6] -> 63[7] via P2P/IPC
+ip-26-0-172-252:688207:688366 [7] NCCL INFO Channel 15/0 : 39[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688200:688369 [0] NCCL INFO Channel 08/0 : 32[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942612:942779 [1] NCCL INFO Channel 03/0 : 41[1] -> 42[2] via P2P/IPC
+ip-26-0-173-7:811248:811415 [0] NCCL INFO Channel 03/0 : 56[0] -> 57[1] via P2P/IPC
+ip-26-0-172-57:942612:942779 [1] NCCL INFO Channel 04/0 : 41[1] -> 42[2] via P2P/IPC
+ip-26-0-173-7:811251:811409 [3] NCCL INFO Channel 03/0 : 59[3] -> 60[4] via P2P/IPC
+ip-26-0-172-57:942611:942772 [0] NCCL INFO Connected all rings
+ip-26-0-173-7:811253:811410 [5] NCCL INFO Channel 09/0 : 61[5] -> 62[6] via P2P/IPC
+ip-26-0-172-57:942611:942772 [0] NCCL INFO Channel 02/0 : 40[0] -> 41[1] via P2P/IPC
+ip-26-0-173-7:811248:811415 [0] NCCL INFO Channel 04/0 : 56[0] -> 57[1] via P2P/IPC
+ip-26-0-173-7:811251:811409 [3] NCCL INFO Channel 05/0 : 59[3] -> 60[4] via P2P/IPC
+ip-26-0-173-7:811254:811414 [6] NCCL INFO Channel 09/0 : 62[6] -> 63[7] via P2P/IPC
+ip-26-0-173-7:811253:811410 [5] NCCL INFO Channel 10/0 : 61[5] -> 62[6] via P2P/IPC
+ip-26-0-173-7:811248:811415 [0] NCCL INFO Channel 05/0 : 56[0] -> 57[1] via P2P/IPC
+ip-26-0-172-57:942612:942779 [1] NCCL INFO Channel 05/0 : 41[1] -> 42[2] via P2P/IPC
+ip-26-0-173-7:811251:811409 [3] NCCL INFO Channel 06/0 : 59[3] -> 60[4] via P2P/IPC
+ip-26-0-172-57:942611:942772 [0] NCCL INFO Channel 03/0 : 40[0] -> 41[1] via P2P/IPC
+ip-26-0-173-7:811253:811410 [5] NCCL INFO Channel 11/0 : 61[5] -> 62[6] via P2P/IPC
+ip-26-0-173-7:811254:811414 [6] NCCL INFO Channel 10/0 : 62[6] -> 63[7] via P2P/IPC
+ip-26-0-173-7:811248:811415 [0] NCCL INFO Channel 06/0 : 56[0] -> 57[1] via P2P/IPC
+ip-26-0-172-57:942612:942779 [1] NCCL INFO Channel 06/0 : 41[1] -> 42[2] via P2P/IPC
+ip-26-0-172-57:942611:942772 [0] NCCL INFO Channel 04/0 : 40[0] -> 41[1] via P2P/IPC
+ip-26-0-173-7:811251:811409 [3] NCCL INFO Channel 07/0 : 59[3] -> 60[4] via P2P/IPC
+ip-26-0-172-57:942612:942779 [1] NCCL INFO Channel 07/0 : 41[1] -> 42[2] via P2P/IPC
+ip-26-0-173-7:811254:811414 [6] NCCL INFO Channel 11/0 : 62[6] -> 63[7] via P2P/IPC
+ip-26-0-173-7:811253:811410 [5] NCCL INFO Channel 12/0 : 61[5] -> 62[6] via P2P/IPC
+ip-26-0-173-7:811251:811409 [3] NCCL INFO Channel 09/0 : 59[3] -> 60[4] via P2P/IPC
+ip-26-0-172-57:942611:942772 [0] NCCL INFO Channel 05/0 : 40[0] -> 41[1] via P2P/IPC
+ip-26-0-173-7:811253:811410 [5] NCCL INFO Channel 13/0 : 61[5] -> 62[6] via P2P/IPC
+ip-26-0-172-57:942612:942779 [1] NCCL INFO Channel 09/0 : 41[1] -> 42[2] via P2P/IPC
+ip-26-0-173-7:811254:811414 [6] NCCL INFO Channel 12/0 : 62[6] -> 63[7] via P2P/IPC
+ip-26-0-173-7:811251:811409 [3] NCCL INFO Channel 10/0 : 59[3] -> 60[4] via P2P/IPC
+ip-26-0-173-7:811248:811415 [0] NCCL INFO Channel 07/0 : 56[0] -> 57[1] via P2P/IPC
+ip-26-0-173-7:811253:811410 [5] NCCL INFO Channel 15/0 : 61[5] -> 62[6] via P2P/IPC
+ip-26-0-172-57:942611:942772 [0] NCCL INFO Channel 06/0 : 40[0] -> 41[1] via P2P/IPC
+ip-26-0-173-7:811254:811414 [6] NCCL INFO Channel 13/0 : 62[6] -> 63[7] via P2P/IPC
+ip-26-0-172-73:926594:926762 [0] NCCL INFO Channel 10/0 : 48[0] -> 49[1] via P2P/IPC
+ip-26-0-173-7:811251:811409 [3] NCCL INFO Channel 11/0 : 59[3] -> 60[4] via P2P/IPC
+ip-26-0-172-57:942612:942779 [1] NCCL INFO Channel 11/0 : 41[1] -> 42[2] via P2P/IPC
+ip-26-0-173-7:811248:811415 [0] NCCL INFO Channel 10/0 : 56[0] -> 57[1] via P2P/IPC
+ip-26-0-172-57:942611:942772 [0] NCCL INFO Channel 07/0 : 40[0] -> 41[1] via P2P/IPC
+ip-26-0-173-7:811254:811414 [6] NCCL INFO Channel 14/0 : 62[6] -> 63[7] via P2P/IPC
+ip-26-0-173-7:811251:811409 [3] NCCL INFO Channel 13/0 : 59[3] -> 60[4] via P2P/IPC
+ip-26-0-173-7:811248:811415 [0] NCCL INFO Channel 11/0 : 56[0] -> 57[1] via P2P/IPC
+ip-26-0-172-57:942612:942779 [1] NCCL INFO Channel 12/0 : 41[1] -> 42[2] via P2P/IPC
+ip-26-0-173-7:811251:811409 [3] NCCL INFO Channel 14/0 : 59[3] -> 60[4] via P2P/IPC
+ip-26-0-173-7:811248:811415 [0] NCCL INFO Channel 12/0 : 56[0] -> 57[1] via P2P/IPC
+ip-26-0-173-7:811253:811410 [5] NCCL INFO Channel 05/0 : 53[5] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-7:811254:811414 [6] NCCL INFO Channel 06/0 : 54[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942611:942772 [0] NCCL INFO Channel 10/0 : 40[0] -> 41[1] via P2P/IPC
+ip-26-0-173-7:811251:811409 [3] NCCL INFO Channel 15/0 : 59[3] -> 60[4] via P2P/IPC
+ip-26-0-173-7:811248:811415 [0] NCCL INFO Channel 13/0 : 56[0] -> 57[1] via P2P/IPC
+ip-26-0-172-73:926594:926762 [0] NCCL INFO Channel 11/0 : 48[0] -> 49[1] via P2P/IPC
+ip-26-0-172-57:942612:942779 [1] NCCL INFO Channel 13/0 : 41[1] -> 42[2] via P2P/IPC
+ip-26-0-173-7:811248:811415 [0] NCCL INFO Channel 14/0 : 56[0] -> 57[1] via P2P/IPC
+ip-26-0-173-7:811252:811408 [4] NCCL INFO Channel 04/0 : 52[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-7:811251:811409 [3] NCCL INFO Channel 03/0 : 51[3] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-7:811249:811413 [1] NCCL INFO Connected all rings
+ip-26-0-173-7:811248:811415 [0] NCCL INFO Channel 15/0 : 56[0] -> 57[1] via P2P/IPC
+ip-26-0-172-73:926594:926762 [0] NCCL INFO Channel 12/0 : 48[0] -> 49[1] via P2P/IPC
+ip-26-0-172-73:926594:926762 [0] NCCL INFO Channel 13/0 : 48[0] -> 49[1] via P2P/IPC
+ip-26-0-172-57:942611:942772 [0] NCCL INFO Channel 11/0 : 40[0] -> 41[1] via P2P/IPC
+ip-26-0-172-57:942613:942776 [2] NCCL INFO Channel 01/0 : 42[2] -> 43[3] via P2P/IPC
+ip-26-0-172-57:942612:942779 [1] NCCL INFO Channel 14/0 : 41[1] -> 42[2] via P2P/IPC
+ip-26-0-172-73:926594:926762 [0] NCCL INFO Channel 14/0 : 48[0] -> 49[1] via P2P/IPC
+ip-26-0-172-57:942611:942772 [0] NCCL INFO Channel 12/0 : 40[0] -> 41[1] via P2P/IPC
+ip-26-0-172-57:942617:942774 [6] NCCL INFO Connected all rings
+ip-26-0-172-57:942613:942776 [2] NCCL INFO Channel 02/0 : 42[2] -> 43[3] via P2P/IPC
+ip-26-0-172-57:942612:942779 [1] NCCL INFO Channel 15/0 : 41[1] -> 42[2] via P2P/IPC
+ip-26-0-172-73:926594:926762 [0] NCCL INFO Channel 15/0 : 48[0] -> 49[1] via P2P/IPC
+ip-26-0-172-57:942616:942777 [5] NCCL INFO Connected all rings
+ip-26-0-172-57:942611:942772 [0] NCCL INFO Channel 13/0 : 40[0] -> 41[1] via P2P/IPC
+ip-26-0-173-7:811249:811413 [1] NCCL INFO Channel 01/0 : 57[1] -> 58[2] via P2P/IPC
+ip-26-0-173-7:811249:811413 [1] NCCL INFO Channel 03/0 : 57[1] -> 58[2] via P2P/IPC
+ip-26-0-173-7:811249:811413 [1] NCCL INFO Channel 04/0 : 57[1] -> 58[2] via P2P/IPC
+ip-26-0-173-7:811249:811413 [1] NCCL INFO Channel 05/0 : 57[1] -> 58[2] via P2P/IPC
+ip-26-0-172-57:942613:942776 [2] NCCL INFO Channel 04/0 : 42[2] -> 43[3] via P2P/IPC
+ip-26-0-173-7:811249:811413 [1] NCCL INFO Channel 06/0 : 57[1] -> 58[2] via P2P/IPC
+ip-26-0-173-7:811249:811413 [1] NCCL INFO Channel 07/0 : 57[1] -> 58[2] via P2P/IPC
+ip-26-0-172-57:942611:942772 [0] NCCL INFO Channel 14/0 : 40[0] -> 41[1] via P2P/IPC
+ip-26-0-172-57:942613:942776 [2] NCCL INFO Channel 05/0 : 42[2] -> 43[3] via P2P/IPC
+ip-26-0-173-7:811249:811413 [1] NCCL INFO Channel 09/0 : 57[1] -> 58[2] via P2P/IPC
+ip-26-0-172-57:942611:942772 [0] NCCL INFO Channel 15/0 : 40[0] -> 41[1] via P2P/IPC
+ip-26-0-173-7:811249:811413 [1] NCCL INFO Channel 11/0 : 57[1] -> 58[2] via P2P/IPC
+ip-26-0-172-57:942613:942776 [2] NCCL INFO Channel 06/0 : 42[2] -> 43[3] via P2P/IPC
+ip-26-0-173-7:811249:811413 [1] NCCL INFO Channel 12/0 : 57[1] -> 58[2] via P2P/IPC
+ip-26-0-173-7:811249:811413 [1] NCCL INFO Channel 13/0 : 57[1] -> 58[2] via P2P/IPC
+ip-26-0-173-7:811249:811413 [1] NCCL INFO Channel 14/0 : 57[1] -> 58[2] via P2P/IPC
+ip-26-0-173-7:811249:811413 [1] NCCL INFO Channel 15/0 : 57[1] -> 58[2] via P2P/IPC
+ip-26-0-173-7:811248:811415 [0] NCCL INFO Channel 05/0 : 56[0] -> 63[7] via P2P/IPC
+ip-26-0-172-73:926599:926755 [5] NCCL INFO Connected all rings
+ip-26-0-173-7:811250:811412 [2] NCCL INFO Channel 02/0 : 50[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-7:811249:811413 [1] NCCL INFO Channel 01/0 : 49[1] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-7:811248:811415 [0] NCCL INFO Channel 06/0 : 56[0] -> 63[7] via P2P/IPC
+ip-26-0-173-7:811248:811415 [0] NCCL INFO Channel 07/0 : 56[0] -> 63[7] via P2P/IPC
+ip-26-0-173-7:811248:811415 [0] NCCL INFO Channel 13/0 : 56[0] -> 63[7] via P2P/IPC
+ip-26-0-173-7:811248:811415 [0] NCCL INFO Channel 14/0 : 56[0] -> 63[7] via P2P/IPC
+ip-26-0-172-57:942611:942772 [0] NCCL INFO Channel 05/0 : 40[0] -> 47[7] via P2P/IPC
+ip-26-0-173-7:811248:811415 [0] NCCL INFO Channel 15/0 : 56[0] -> 63[7] via P2P/IPC
+ip-26-0-173-7:811248:811415 [0] NCCL INFO Channel 00/0 : 48[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-7:811255:811411 [7] NCCL INFO Channel 07/0 : 55[7] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926600:926757 [6] NCCL INFO Connected all rings
+ip-26-0-172-57:942613:942776 [2] NCCL INFO Channel 07/0 : 42[2] -> 43[3] via P2P/IPC
+ip-26-0-172-57:942613:942776 [2] NCCL INFO Channel 09/0 : 42[2] -> 43[3] via P2P/IPC
+ip-26-0-172-57:942613:942776 [2] NCCL INFO Channel 10/0 : 42[2] -> 43[3] via P2P/IPC
+ip-26-0-172-73:926597:926758 [3] NCCL INFO Connected all rings
+ip-26-0-172-57:942615:942775 [4] NCCL INFO Connected all rings
+ip-26-0-172-57:942614:942773 [3] NCCL INFO Channel 01/0 : 43[3] -> 44[4] via P2P/IPC
+ip-26-0-172-57:942613:942776 [2] NCCL INFO Channel 12/0 : 42[2] -> 43[3] via P2P/IPC
+ip-26-0-172-73:926598:926761 [4] NCCL INFO Connected all rings
+ip-26-0-172-57:942614:942773 [3] NCCL INFO Channel 02/0 : 43[3] -> 44[4] via P2P/IPC
+ip-26-0-172-57:942611:942772 [0] NCCL INFO Channel 06/0 : 40[0] -> 47[7] via P2P/IPC
+ip-26-0-172-57:942614:942773 [3] NCCL INFO Channel 03/0 : 43[3] -> 44[4] via P2P/IPC
+ip-26-0-172-57:942613:942776 [2] NCCL INFO Channel 13/0 : 42[2] -> 43[3] via P2P/IPC
+ip-26-0-172-57:942611:942772 [0] NCCL INFO Channel 07/0 : 40[0] -> 47[7] via P2P/IPC
+ip-26-0-172-57:942613:942776 [2] NCCL INFO Channel 14/0 : 42[2] -> 43[3] via P2P/IPC
+ip-26-0-172-57:942614:942773 [3] NCCL INFO Channel 05/0 : 43[3] -> 44[4] via P2P/IPC
+ip-26-0-172-57:942616:942777 [5] NCCL INFO Channel 01/0 : 45[5] -> 46[6] via P2P/IPC
+ip-26-0-172-57:942611:942772 [0] NCCL INFO Channel 13/0 : 40[0] -> 47[7] via P2P/IPC
+ip-26-0-172-57:942613:942776 [2] NCCL INFO Channel 15/0 : 42[2] -> 43[3] via P2P/IPC
+ip-26-0-172-57:942614:942773 [3] NCCL INFO Channel 06/0 : 43[3] -> 44[4] via P2P/IPC
+ip-26-0-172-57:942616:942777 [5] NCCL INFO Channel 02/0 : 45[5] -> 46[6] via P2P/IPC
+ip-26-0-172-57:942611:942772 [0] NCCL INFO Channel 14/0 : 40[0] -> 47[7] via P2P/IPC
+ip-26-0-172-57:942614:942773 [3] NCCL INFO Channel 07/0 : 43[3] -> 44[4] via P2P/IPC
+ip-26-0-172-57:942616:942777 [5] NCCL INFO Channel 03/0 : 45[5] -> 46[6] via P2P/IPC
+ip-26-0-172-57:942612:942779 [1] NCCL INFO Channel 09/0 : 33[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942612:942779 [1] NCCL INFO Channel 01/0 : 41[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942612:942779 [1] NCCL INFO Channel 09/0 : 41[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942611:942772 [0] NCCL INFO Channel 15/0 : 40[0] -> 47[7] via P2P/IPC
+ip-26-0-172-57:942617:942774 [6] NCCL INFO Channel 01/0 : 46[6] -> 47[7] via P2P/IPC
+ip-26-0-172-57:942614:942773 [3] NCCL INFO Channel 09/0 : 43[3] -> 44[4] via P2P/IPC
+ip-26-0-172-73:926595:926760 [1] NCCL INFO Channel 06/0 : 49[1] -> 50[2] via P2P/IPC
+ip-26-0-172-57:942616:942777 [5] NCCL INFO Channel 04/0 : 45[5] -> 46[6] via P2P/IPC
+ip-26-0-172-252:688201:688364 [1] NCCL INFO Channel 01/0 : 17[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688201:688364 [1] NCCL INFO Channel 01/0 : 33[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942617:942774 [6] NCCL INFO Channel 02/0 : 46[6] -> 47[7] via P2P/IPC
+ip-26-0-172-57:942614:942773 [3] NCCL INFO Channel 10/0 : 43[3] -> 44[4] via P2P/IPC
+ip-26-0-172-57:942616:942777 [5] NCCL INFO Channel 05/0 : 45[5] -> 46[6] via P2P/IPC
+ip-26-0-172-142:880680:880843 [1] NCCL INFO Channel 01/0 : 33[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942614:942773 [3] NCCL INFO Channel 11/0 : 43[3] -> 44[4] via P2P/IPC
+ip-26-0-172-57:942616:942777 [5] NCCL INFO Channel 07/0 : 45[5] -> 46[6] via P2P/IPC
+ip-26-0-172-57:942617:942774 [6] NCCL INFO Channel 03/0 : 46[6] -> 47[7] via P2P/IPC
+ip-26-0-172-73:926596:926756 [2] NCCL INFO Connected all rings
+ip-26-0-172-57:942614:942773 [3] NCCL INFO Channel 13/0 : 43[3] -> 44[4] via P2P/IPC
+ip-26-0-172-57:942616:942777 [5] NCCL INFO Channel 09/0 : 45[5] -> 46[6] via P2P/IPC
+ip-26-0-172-57:942617:942774 [6] NCCL INFO Channel 04/0 : 46[6] -> 47[7] via P2P/IPC
+ip-26-0-172-57:942614:942773 [3] NCCL INFO Channel 14/0 : 43[3] -> 44[4] via P2P/IPC
+ip-26-0-172-57:942616:942777 [5] NCCL INFO Channel 10/0 : 45[5] -> 46[6] via P2P/IPC
+ip-26-0-172-57:942617:942774 [6] NCCL INFO Channel 05/0 : 46[6] -> 47[7] via P2P/IPC
+ip-26-0-172-57:942614:942773 [3] NCCL INFO Channel 15/0 : 43[3] -> 44[4] via P2P/IPC
+ip-26-0-172-57:942617:942774 [6] NCCL INFO Channel 06/0 : 46[6] -> 47[7] via P2P/IPC
+ip-26-0-172-57:942616:942777 [5] NCCL INFO Channel 11/0 : 45[5] -> 46[6] via P2P/IPC
+ip-26-0-172-57:942617:942774 [6] NCCL INFO Channel 09/0 : 46[6] -> 47[7] via P2P/IPC
+ip-26-0-172-57:942616:942777 [5] NCCL INFO Channel 12/0 : 45[5] -> 46[6] via P2P/IPC
+ip-26-0-172-57:942613:942776 [2] NCCL INFO Channel 10/0 : 34[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942613:942776 [2] NCCL INFO Channel 02/0 : 42[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942613:942776 [2] NCCL INFO Channel 10/0 : 42[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926599:926755 [5] NCCL INFO Channel 01/0 : 53[5] -> 54[6] via P2P/IPC
+ip-26-0-172-57:942617:942774 [6] NCCL INFO Channel 10/0 : 46[6] -> 47[7] via P2P/IPC
+ip-26-0-172-57:942616:942777 [5] NCCL INFO Channel 13/0 : 45[5] -> 46[6] via P2P/IPC
+ip-26-0-172-252:688202:688368 [2] NCCL INFO Channel 02/0 : 18[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688202:688368 [2] NCCL INFO Channel 02/0 : 34[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942617:942774 [6] NCCL INFO Channel 11/0 : 46[6] -> 47[7] via P2P/IPC
+ip-26-0-172-57:942616:942777 [5] NCCL INFO Channel 15/0 : 45[5] -> 46[6] via P2P/IPC
+ip-26-0-172-73:926599:926755 [5] NCCL INFO Channel 02/0 : 53[5] -> 54[6] via P2P/IPC
+ip-26-0-172-142:880681:880845 [2] NCCL INFO Channel 02/0 : 34[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942617:942774 [6] NCCL INFO Channel 12/0 : 46[6] -> 47[7] via P2P/IPC
+ip-26-0-172-57:942617:942774 [6] NCCL INFO Channel 13/0 : 46[6] -> 47[7] via P2P/IPC
+ip-26-0-172-57:942617:942774 [6] NCCL INFO Channel 14/0 : 46[6] -> 47[7] via P2P/IPC
+ip-26-0-172-73:926599:926755 [5] NCCL INFO Channel 03/0 : 53[5] -> 54[6] via P2P/IPC
+ip-26-0-172-73:926595:926760 [1] NCCL INFO Channel 07/0 : 49[1] -> 50[2] via P2P/IPC
+ip-26-0-172-57:942615:942775 [4] NCCL INFO Channel 01/0 : 44[4] -> 45[5] via P2P/IPC
+ip-26-0-172-57:942615:942775 [4] NCCL INFO Channel 02/0 : 44[4] -> 45[5] via P2P/IPC
+ip-26-0-172-57:942617:942774 [6] NCCL INFO Channel 14/0 : 38[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926597:926758 [3] NCCL INFO Channel 01/0 : 51[3] -> 52[4] via P2P/IPC
+ip-26-0-172-57:942617:942774 [6] NCCL INFO Channel 06/0 : 46[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942617:942774 [6] NCCL INFO Channel 14/0 : 46[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942615:942775 [4] NCCL INFO Channel 03/0 : 44[4] -> 45[5] via P2P/IPC
+ip-26-0-172-73:926599:926755 [5] NCCL INFO Channel 04/0 : 53[5] -> 54[6] via P2P/IPC
+ip-26-0-172-73:926595:926760 [1] NCCL INFO Channel 09/0 : 49[1] -> 50[2] via P2P/IPC
+ip-26-0-172-57:942615:942775 [4] NCCL INFO Channel 04/0 : 44[4] -> 45[5] via P2P/IPC
+ip-26-0-172-57:942615:942775 [4] NCCL INFO Channel 06/0 : 44[4] -> 45[5] via P2P/IPC
+ip-26-0-172-142:880685:880847 [6] NCCL INFO Channel 06/0 : 38[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926597:926758 [3] NCCL INFO Channel 02/0 : 51[3] -> 52[4] via P2P/IPC
+ip-26-0-172-57:942615:942775 [4] NCCL INFO Channel 07/0 : 44[4] -> 45[5] via P2P/IPC
+ip-26-0-172-57:942615:942775 [4] NCCL INFO Channel 09/0 : 44[4] -> 45[5] via P2P/IPC
+ip-26-0-172-73:926599:926755 [5] NCCL INFO Channel 05/0 : 53[5] -> 54[6] via P2P/IPC
+ip-26-0-172-73:926595:926760 [1] NCCL INFO Channel 11/0 : 49[1] -> 50[2] via P2P/IPC
+ip-26-0-172-57:942615:942775 [4] NCCL INFO Channel 10/0 : 44[4] -> 45[5] via P2P/IPC
+ip-26-0-172-73:926598:926761 [4] NCCL INFO Channel 01/0 : 52[4] -> 53[5] via P2P/IPC
+ip-26-0-172-252:688206:688363 [6] NCCL INFO Channel 06/0 : 22[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942615:942775 [4] NCCL INFO Channel 11/0 : 44[4] -> 45[5] via P2P/IPC
+ip-26-0-172-73:926597:926758 [3] NCCL INFO Channel 03/0 : 51[3] -> 52[4] via P2P/IPC
+ip-26-0-172-57:942618:942778 [7] NCCL INFO Channel 15/0 : 39[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942611:942772 [0] NCCL INFO Channel 08/0 : 32[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942615:942775 [4] NCCL INFO Channel 12/0 : 44[4] -> 45[5] via P2P/IPC
+ip-26-0-172-57:942611:942772 [0] NCCL INFO Channel 00/0 : 40[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942618:942778 [7] NCCL INFO Channel 07/0 : 47[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942611:942772 [0] NCCL INFO Channel 08/0 : 40[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942618:942778 [7] NCCL INFO Channel 15/0 : 47[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942615:942775 [4] NCCL INFO Channel 14/0 : 44[4] -> 45[5] via P2P/IPC
+ip-26-0-172-73:926595:926760 [1] NCCL INFO Channel 12/0 : 49[1] -> 50[2] via P2P/IPC
+ip-26-0-172-73:926599:926755 [5] NCCL INFO Channel 07/0 : 53[5] -> 54[6] via P2P/IPC
+ip-26-0-172-57:942615:942775 [4] NCCL INFO Channel 15/0 : 44[4] -> 45[5] via P2P/IPC
+ip-26-0-172-73:926598:926761 [4] NCCL INFO Channel 02/0 : 52[4] -> 53[5] via P2P/IPC
+ip-26-0-172-73:926597:926758 [3] NCCL INFO Channel 05/0 : 51[3] -> 52[4] via P2P/IPC
+ip-26-0-172-57:942614:942773 [3] NCCL INFO Channel 11/0 : 35[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880686:880841 [7] NCCL INFO Channel 07/0 : 39[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926600:926757 [6] NCCL INFO Channel 01/0 : 54[6] -> 55[7] via P2P/IPC
+ip-26-0-172-57:942616:942777 [5] NCCL INFO Channel 13/0 : 37[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880679:880846 [0] NCCL INFO Channel 00/0 : 32[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926601:926759 [7] NCCL INFO Connected all rings
+ip-26-0-172-57:942614:942773 [3] NCCL INFO Channel 03/0 : 43[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942615:942775 [4] NCCL INFO Channel 12/0 : 36[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942616:942777 [5] NCCL INFO Channel 05/0 : 45[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942614:942773 [3] NCCL INFO Channel 11/0 : 43[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942615:942775 [4] NCCL INFO Channel 04/0 : 44[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942616:942777 [5] NCCL INFO Channel 13/0 : 45[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942615:942775 [4] NCCL INFO Channel 12/0 : 44[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926595:926760 [1] NCCL INFO Channel 13/0 : 49[1] -> 50[2] via P2P/IPC
+ip-26-0-172-73:926599:926755 [5] NCCL INFO Channel 09/0 : 53[5] -> 54[6] via P2P/IPC
+ip-26-0-172-73:926598:926761 [4] NCCL INFO Channel 03/0 : 52[4] -> 53[5] via P2P/IPC
+ip-26-0-172-252:688206:688363 [6] NCCL INFO Channel 06/0 : 38[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926597:926758 [3] NCCL INFO Channel 06/0 : 51[3] -> 52[4] via P2P/IPC
+ip-26-0-172-142:880684:880842 [5] NCCL INFO Channel 05/0 : 37[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880682:880840 [3] NCCL INFO Channel 03/0 : 35[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880683:880844 [4] NCCL INFO Channel 04/0 : 36[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926600:926757 [6] NCCL INFO Channel 02/0 : 54[6] -> 55[7] via P2P/IPC
+ip-26-0-172-73:926599:926755 [5] NCCL INFO Channel 10/0 : 53[5] -> 54[6] via P2P/IPC
+ip-26-0-172-73:926595:926760 [1] NCCL INFO Channel 14/0 : 49[1] -> 50[2] via P2P/IPC
+ip-26-0-172-252:688200:688369 [0] NCCL INFO Channel 00/0 : 16[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688207:688366 [7] NCCL INFO Channel 07/0 : 23[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688200:688369 [0] NCCL INFO Channel 00/0 : 32[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926598:926761 [4] NCCL INFO Channel 04/0 : 52[4] -> 53[5] via P2P/IPC
+ip-26-0-172-252:688207:688366 [7] NCCL INFO Channel 07/0 : 39[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688205:688365 [5] NCCL INFO Channel 05/0 : 21[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688203:688362 [3] NCCL INFO Channel 03/0 : 19[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688204:688367 [4] NCCL INFO Channel 04/0 : 20[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688205:688365 [5] NCCL INFO Channel 05/0 : 37[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688203:688362 [3] NCCL INFO Channel 03/0 : 35[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688204:688367 [4] NCCL INFO Channel 04/0 : 36[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926597:926758 [3] NCCL INFO Channel 07/0 : 51[3] -> 52[4] via P2P/IPC
+ip-26-0-172-73:926600:926757 [6] NCCL INFO Channel 03/0 : 54[6] -> 55[7] via P2P/IPC
+ip-26-0-172-73:926599:926755 [5] NCCL INFO Channel 11/0 : 53[5] -> 54[6] via P2P/IPC
+ip-26-0-172-73:926595:926760 [1] NCCL INFO Channel 15/0 : 49[1] -> 50[2] via P2P/IPC
+ip-26-0-172-73:926598:926761 [4] NCCL INFO Channel 06/0 : 52[4] -> 53[5] via P2P/IPC
+ip-26-0-172-73:926596:926756 [2] NCCL INFO Channel 01/0 : 50[2] -> 51[3] via P2P/IPC
+ip-26-0-172-73:926597:926758 [3] NCCL INFO Channel 09/0 : 51[3] -> 52[4] via P2P/IPC
+ip-26-0-172-73:926600:926757 [6] NCCL INFO Channel 04/0 : 54[6] -> 55[7] via P2P/IPC
+ip-26-0-172-73:926599:926755 [5] NCCL INFO Channel 12/0 : 53[5] -> 54[6] via P2P/IPC
+ip-26-0-172-73:926598:926761 [4] NCCL INFO Channel 07/0 : 52[4] -> 53[5] via P2P/IPC
+ip-26-0-172-73:926594:926762 [0] NCCL INFO Channel 05/0 : 48[0] -> 55[7] via P2P/IPC
+ip-26-0-172-73:926596:926756 [2] NCCL INFO Channel 02/0 : 50[2] -> 51[3] via P2P/IPC
+ip-26-0-172-73:926597:926758 [3] NCCL INFO Channel 10/0 : 51[3] -> 52[4] via P2P/IPC
+ip-26-0-172-73:926600:926757 [6] NCCL INFO Channel 05/0 : 54[6] -> 55[7] via P2P/IPC
+ip-26-0-172-73:926599:926755 [5] NCCL INFO Channel 13/0 : 53[5] -> 54[6] via P2P/IPC
+ip-26-0-172-73:926598:926761 [4] NCCL INFO Channel 09/0 : 52[4] -> 53[5] via P2P/IPC
+ip-26-0-172-73:926596:926756 [2] NCCL INFO Channel 04/0 : 50[2] -> 51[3] via P2P/IPC
+ip-26-0-172-73:926594:926762 [0] NCCL INFO Channel 06/0 : 48[0] -> 55[7] via P2P/IPC
+ip-26-0-172-73:926597:926758 [3] NCCL INFO Channel 11/0 : 51[3] -> 52[4] via P2P/IPC
+ip-26-0-172-73:926600:926757 [6] NCCL INFO Channel 06/0 : 54[6] -> 55[7] via P2P/IPC
+ip-26-0-172-73:926599:926755 [5] NCCL INFO Channel 15/0 : 53[5] -> 54[6] via P2P/IPC
+ip-26-0-172-73:926598:926761 [4] NCCL INFO Channel 10/0 : 52[4] -> 53[5] via P2P/IPC
+ip-26-0-172-73:926596:926756 [2] NCCL INFO Channel 05/0 : 50[2] -> 51[3] via P2P/IPC
+ip-26-0-172-73:926594:926762 [0] NCCL INFO Channel 07/0 : 48[0] -> 55[7] via P2P/IPC
+ip-26-0-172-73:926597:926758 [3] NCCL INFO Channel 13/0 : 51[3] -> 52[4] via P2P/IPC
+ip-26-0-172-73:926600:926757 [6] NCCL INFO Channel 09/0 : 54[6] -> 55[7] via P2P/IPC
+ip-26-0-172-73:926598:926761 [4] NCCL INFO Channel 11/0 : 52[4] -> 53[5] via P2P/IPC
+ip-26-0-172-73:926596:926756 [2] NCCL INFO Channel 06/0 : 50[2] -> 51[3] via P2P/IPC
+ip-26-0-172-73:926597:926758 [3] NCCL INFO Channel 14/0 : 51[3] -> 52[4] via P2P/IPC
+ip-26-0-172-73:926600:926757 [6] NCCL INFO Channel 10/0 : 54[6] -> 55[7] via P2P/IPC
+ip-26-0-172-73:926598:926761 [4] NCCL INFO Channel 12/0 : 52[4] -> 53[5] via P2P/IPC
+ip-26-0-172-73:926596:926756 [2] NCCL INFO Channel 07/0 : 50[2] -> 51[3] via P2P/IPC
+ip-26-0-172-73:926597:926758 [3] NCCL INFO Channel 15/0 : 51[3] -> 52[4] via P2P/IPC
+ip-26-0-172-73:926600:926757 [6] NCCL INFO Channel 11/0 : 54[6] -> 55[7] via P2P/IPC
+ip-26-0-172-73:926598:926761 [4] NCCL INFO Channel 14/0 : 52[4] -> 53[5] via P2P/IPC
+ip-26-0-172-73:926596:926756 [2] NCCL INFO Channel 09/0 : 50[2] -> 51[3] via P2P/IPC
+ip-26-0-172-73:926600:926757 [6] NCCL INFO Channel 12/0 : 54[6] -> 55[7] via P2P/IPC
+ip-26-0-172-73:926598:926761 [4] NCCL INFO Channel 15/0 : 52[4] -> 53[5] via P2P/IPC
+ip-26-0-172-73:926596:926756 [2] NCCL INFO Channel 10/0 : 50[2] -> 51[3] via P2P/IPC
+ip-26-0-172-73:926600:926757 [6] NCCL INFO Channel 13/0 : 54[6] -> 55[7] via P2P/IPC
+ip-26-0-172-73:926596:926756 [2] NCCL INFO Channel 12/0 : 50[2] -> 51[3] via P2P/IPC
+ip-26-0-172-73:926600:926757 [6] NCCL INFO Channel 14/0 : 54[6] -> 55[7] via P2P/IPC
+ip-26-0-172-73:926598:926761 [4] NCCL INFO Channel 04/0 : 44[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926596:926756 [2] NCCL INFO Channel 13/0 : 50[2] -> 51[3] via P2P/IPC
+ip-26-0-172-73:926598:926761 [4] NCCL INFO Channel 12/0 : 44[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926598:926761 [4] NCCL INFO Channel 04/0 : 52[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926596:926756 [2] NCCL INFO Channel 14/0 : 50[2] -> 51[3] via P2P/IPC
+ip-26-0-172-57:942615:942775 [4] NCCL INFO Channel 12/0 : 28[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926599:926755 [5] NCCL INFO Channel 05/0 : 45[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926600:926757 [6] NCCL INFO Channel 06/0 : 46[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926596:926756 [2] NCCL INFO Channel 15/0 : 50[2] -> 51[3] via P2P/IPC
+ip-26-0-172-73:926598:926761 [4] NCCL INFO Channel 04/0 : 36[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926599:926755 [5] NCCL INFO Channel 13/0 : 45[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-7:811252:811408 [4] NCCL INFO Channel 12/0 : 28[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942615:942775 [4] NCCL INFO Channel 12/0 : 44[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926600:926757 [6] NCCL INFO Channel 14/0 : 46[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-7:811252:811408 [4] NCCL INFO Channel 12/0 : 60[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926599:926755 [5] NCCL INFO Channel 05/0 : 53[5] -> 61[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926600:926757 [6] NCCL INFO Channel 06/0 : 54[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618588:618750 [4] NCCL INFO Channel 12/0 : 60[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926598:926761 [4] NCCL INFO Channel 04/0 : 52[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618588:618750 [4] NCCL INFO Channel 12/0 : 28[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688204:688367 [4] NCCL INFO Channel 04/0 : 4[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688204:688367 [4] NCCL INFO Channel 04/0 : 36[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942616:942777 [5] NCCL INFO Channel 13/0 : 29[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942617:942774 [6] NCCL INFO Channel 14/0 : 30[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926599:926755 [5] NCCL INFO Channel 05/0 : 37[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618588:618750 [4] NCCL INFO Channel 12/0 : 44[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-7:811252:811408 [4] NCCL INFO Channel 04/0 : 60[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926600:926757 [6] NCCL INFO Channel 06/0 : 38[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618588:618750 [4] NCCL INFO Channel 12/0 : 28[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926595:926760 [1] NCCL INFO Channel 01/0 : 41[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926596:926756 [2] NCCL INFO Channel 02/0 : 42[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688204:688367 [4] NCCL INFO Channel 04/0 : 52[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942616:942777 [5] NCCL INFO Channel 13/0 : 45[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-7:811254:811414 [6] NCCL INFO Channel 14/0 : 30[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926597:926758 [3] NCCL INFO Channel 03/0 : 43[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926595:926760 [1] NCCL INFO Channel 09/0 : 41[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618589:618744 [5] NCCL INFO Channel 13/0 : 61[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942617:942774 [6] NCCL INFO Channel 14/0 : 46[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:94346:94511 [4] NCCL INFO Channel 12/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-7:811254:811414 [6] NCCL INFO Channel 14/0 : 62[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926596:926756 [2] NCCL INFO Channel 10/0 : 42[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926597:926758 [3] NCCL INFO Channel 11/0 : 43[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926600:926757 [6] NCCL INFO Channel 06/0 : 54[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926595:926760 [1] NCCL INFO Channel 01/0 : 49[1] -> 57[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618589:618744 [5] NCCL INFO Channel 13/0 : 29[5] -> 61[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926596:926756 [2] NCCL INFO Channel 02/0 : 50[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926599:926755 [5] NCCL INFO Channel 05/0 : 53[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942615:942775 [4] NCCL INFO Channel 04/0 : 52[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618590:618749 [6] NCCL INFO Channel 14/0 : 62[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926597:926758 [3] NCCL INFO Channel 03/0 : 51[3] -> 59[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942615:942775 [4] NCCL INFO Channel 12/0 : 52[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618590:618749 [6] NCCL INFO Channel 14/0 : 30[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942615:942775 [4] NCCL INFO Channel 12/0 : 44[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618588:618750 [4] NCCL INFO Channel 04/0 : 28[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO Channel 04/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942613:942776 [2] NCCL INFO Channel 10/0 : 26[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942614:942773 [3] NCCL INFO Channel 11/0 : 27[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO Channel 12/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926596:926756 [2] NCCL INFO Channel 02/0 : 34[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO Channel 12/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926595:926760 [1] NCCL INFO Channel 01/0 : 33[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618590:618749 [6] NCCL INFO Channel 14/0 : 46[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926597:926758 [3] NCCL INFO Channel 03/0 : 35[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942613:942776 [2] NCCL INFO Channel 10/0 : 42[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-7:811251:811409 [3] NCCL INFO Channel 11/0 : 27[3] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-7:811254:811414 [6] NCCL INFO Channel 06/0 : 62[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-7:811250:811412 [2] NCCL INFO Channel 10/0 : 26[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618590:618749 [6] NCCL INFO Channel 14/0 : 30[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942614:942773 [3] NCCL INFO Channel 11/0 : 43[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-7:811250:811412 [2] NCCL INFO Channel 10/0 : 58[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-7:811251:811409 [3] NCCL INFO Channel 11/0 : 59[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926596:926756 [2] NCCL INFO Channel 02/0 : 50[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618586:618745 [2] NCCL INFO Channel 10/0 : 58[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926595:926760 [1] NCCL INFO Channel 01/0 : 49[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618587:618751 [3] NCCL INFO Channel 11/0 : 59[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618586:618745 [2] NCCL INFO Channel 10/0 : 26[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926597:926758 [3] NCCL INFO Channel 03/0 : 51[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-7:811249:811413 [1] NCCL INFO Channel 09/0 : 25[1] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618587:618751 [3] NCCL INFO Channel 11/0 : 27[3] -> 59[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926594:926762 [0] NCCL INFO Channel 13/0 : 48[0] -> 55[7] via P2P/IPC
+ip-26-0-172-57:942617:942774 [6] NCCL INFO Channel 06/0 : 54[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-7:811249:811413 [1] NCCL INFO Channel 09/0 : 57[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942617:942774 [6] NCCL INFO Channel 14/0 : 54[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-7:811250:811412 [2] NCCL INFO Channel 02/0 : 58[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618586:618745 [2] NCCL INFO Channel 10/0 : 42[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942617:942774 [6] NCCL INFO Channel 14/0 : 46[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618590:618749 [6] NCCL INFO Channel 06/0 : 30[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-7:811251:811409 [3] NCCL INFO Channel 03/0 : 59[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO Channel 06/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618587:618751 [3] NCCL INFO Channel 11/0 : 43[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618586:618745 [2] NCCL INFO Channel 10/0 : 26[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618587:618751 [3] NCCL INFO Channel 11/0 : 27[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-7:811253:811410 [5] NCCL INFO Channel 13/0 : 29[5] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO Channel 14/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926594:926762 [0] NCCL INFO Channel 14/0 : 48[0] -> 55[7] via P2P/IPC
+ip-26-0-173-7:811253:811410 [5] NCCL INFO Channel 13/0 : 61[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO Channel 14/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942613:942776 [2] NCCL INFO Channel 02/0 : 50[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942614:942773 [3] NCCL INFO Channel 03/0 : 51[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926594:926762 [0] NCCL INFO Channel 15/0 : 48[0] -> 55[7] via P2P/IPC
+ip-26-0-172-147:618586:618745 [2] NCCL INFO Channel 02/0 : 26[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942613:942776 [2] NCCL INFO Channel 10/0 : 50[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-7:811253:811410 [5] NCCL INFO Channel 05/0 : 61[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618587:618751 [3] NCCL INFO Channel 03/0 : 27[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942614:942773 [3] NCCL INFO Channel 11/0 : 51[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618589:618744 [5] NCCL INFO Channel 13/0 : 45[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO Channel 03/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618589:618744 [5] NCCL INFO Channel 13/0 : 29[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO Channel 11/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO Channel 02/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO Channel 11/0 : 11[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO Channel 10/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926601:926759 [7] NCCL INFO Channel 07/0 : 47[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO Channel 10/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926594:926762 [0] NCCL INFO Channel 00/0 : 40[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618589:618744 [5] NCCL INFO Channel 05/0 : 29[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926601:926759 [7] NCCL INFO Channel 15/0 : 47[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926594:926762 [0] NCCL INFO Channel 08/0 : 40[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO Channel 05/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926601:926759 [7] NCCL INFO Channel 07/0 : 55[7] -> 63[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO Channel 13/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926594:926762 [0] NCCL INFO Channel 00/0 : 48[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO Channel 13/0 : 13[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926594:926762 [0] NCCL INFO Channel 00/0 : 32[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-7:811248:811415 [0] NCCL INFO Channel 08/0 : 24[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926601:926759 [7] NCCL INFO Channel 07/0 : 39[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-7:811255:811411 [7] NCCL INFO Channel 15/0 : 31[7] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-7:811248:811415 [0] NCCL INFO Channel 08/0 : 56[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-7:811255:811411 [7] NCCL INFO Channel 15/0 : 63[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926594:926762 [0] NCCL INFO Channel 00/0 : 48[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618591:618746 [7] NCCL INFO Channel 15/0 : 63[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926601:926759 [7] NCCL INFO Channel 07/0 : 55[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618591:618746 [7] NCCL INFO Channel 15/0 : 31[7] -> 63[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618591:618746 [7] NCCL INFO Channel 15/0 : 47[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-7:811255:811411 [7] NCCL INFO Channel 07/0 : 63[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942613:942776 [2] NCCL INFO Channel 10/0 : 42[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618591:618746 [7] NCCL INFO Channel 15/0 : 31[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942614:942773 [3] NCCL INFO Channel 11/0 : 43[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926595:926760 [1] NCCL INFO Channel 01/0 : 57[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942616:942777 [5] NCCL INFO Channel 05/0 : 53[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942616:942777 [5] NCCL INFO Channel 13/0 : 53[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942616:942777 [5] NCCL INFO Channel 13/0 : 45[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942611:942772 [0] NCCL INFO Channel 08/0 : 24[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942618:942778 [7] NCCL INFO Channel 15/0 : 31[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942611:942772 [0] NCCL INFO Channel 08/0 : 40[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942618:942778 [7] NCCL INFO Channel 15/0 : 47[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926595:926760 [1] NCCL INFO Channel 01/0 : 49[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688201:688364 [1] NCCL INFO Channel 01/0 : 1[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880680:880843 [1] NCCL INFO Channel 01/0 : 25[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688201:688364 [1] NCCL INFO Channel 01/0 : 33[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926595:926760 [1] NCCL INFO Channel 09/0 : 49[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688201:688364 [1] NCCL INFO Channel 01/0 : 49[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688201:688364 [1] NCCL INFO Channel 01/0 : 33[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:94346:94511 [4] NCCL INFO Channel 00/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-142:880680:880843 [1] NCCL INFO Channel 01/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:94343:94508 [1] NCCL INFO Channel 09/0 : 9[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880680:880843 [1] NCCL INFO Channel 09/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO Channel 07/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO Channel 15/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:94346:94511 [4] NCCL INFO Channel 03/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO Channel 15/0 : 15[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688201:688364 [1] NCCL INFO Channel 09/0 : 41[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:94346:94511 [4] NCCL INFO Channel 05/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-225:94346:94511 [4] NCCL INFO Channel 06/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-252:688200:688369 [0] NCCL INFO Channel 00/0 : 0[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618591:618746 [7] NCCL INFO Channel 07/0 : 31[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942618:942778 [7] NCCL INFO Channel 07/0 : 55[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:94346:94511 [4] NCCL INFO Channel 07/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-252:688200:688369 [0] NCCL INFO Channel 00/0 : 32[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942618:942778 [7] NCCL INFO Channel 15/0 : 55[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688203:688362 [3] NCCL INFO Channel 03/0 : 3[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942618:942778 [7] NCCL INFO Channel 15/0 : 47[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688203:688362 [3] NCCL INFO Channel 03/0 : 35[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:94346:94511 [4] NCCL INFO Channel 08/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-147:618584:618747 [0] NCCL INFO Channel 08/0 : 56[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688200:688369 [0] NCCL INFO Channel 00/0 : 48[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:94342:94504 [0] NCCL INFO Channel 08/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618584:618747 [0] NCCL INFO Channel 08/0 : 24[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688200:688369 [0] NCCL INFO Channel 00/0 : 32[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:94346:94511 [4] NCCL INFO Channel 11/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-252:688207:688366 [7] NCCL INFO Channel 07/0 : 7[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688203:688362 [3] NCCL INFO Channel 03/0 : 51[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688207:688366 [7] NCCL INFO Channel 07/0 : 39[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:94345:94509 [3] NCCL INFO Channel 11/0 : 11[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618584:618747 [0] NCCL INFO Channel 08/0 : 40[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688203:688362 [3] NCCL INFO Channel 03/0 : 35[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-7:811248:811415 [0] NCCL INFO Channel 00/0 : 56[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926594:926762 [0] NCCL INFO Channel 00/0 : 56[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942612:942779 [1] NCCL INFO Channel 09/0 : 25[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:94346:94511 [4] NCCL INFO Channel 13/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-147:618584:618747 [0] NCCL INFO Channel 08/0 : 24[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688204:688367 [4] NCCL INFO Channel 04/0 : 36[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926594:926762 [0] NCCL INFO Channel 00/0 : 48[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688202:688368 [2] NCCL INFO Channel 02/0 : 2[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880679:880846 [0] NCCL INFO Channel 00/0 : 24[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688205:688365 [5] NCCL INFO Channel 05/0 : 5[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688200:688369 [0] NCCL INFO Channel 08/0 : 40[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880679:880846 [0] NCCL INFO Channel 00/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926594:926762 [0] NCCL INFO Channel 08/0 : 48[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926597:926758 [3] NCCL INFO Channel 03/0 : 59[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:94349:94505 [7] NCCL INFO Channel 15/0 : 15[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942612:942779 [1] NCCL INFO Channel 09/0 : 41[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688202:688368 [2] NCCL INFO Channel 02/0 : 34[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688205:688365 [5] NCCL INFO Channel 05/0 : 37[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688207:688366 [7] NCCL INFO Channel 07/0 : 55[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880679:880846 [0] NCCL INFO Channel 08/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926598:926761 [4] NCCL INFO Channel 04/0 : 60[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:94346:94511 [4] NCCL INFO Channel 14/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-252:688207:688366 [7] NCCL INFO Channel 07/0 : 39[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942611:942772 [0] NCCL INFO Channel 00/0 : 48[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880682:880840 [3] NCCL INFO Channel 03/0 : 27[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926597:926758 [3] NCCL INFO Channel 03/0 : 51[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:94349:94505 [7] NCCL INFO Channel 01/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-57:942611:942772 [0] NCCL INFO Channel 08/0 : 48[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618585:618748 [1] NCCL INFO Channel 09/0 : 57[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880683:880844 [4] NCCL INFO Channel 04/0 : 28[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926598:926761 [4] NCCL INFO Channel 04/0 : 52[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO Channel 00/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926597:926758 [3] NCCL INFO Channel 11/0 : 51[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926598:926761 [4] NCCL INFO Channel 12/0 : 52[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618584:618747 [0] NCCL INFO Channel 00/0 : 24[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618585:618748 [1] NCCL INFO Channel 09/0 : 25[1] -> 57[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:94344:94510 [2] NCCL INFO Channel 10/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:94347:94507 [5] NCCL INFO Channel 13/0 : 13[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880682:880840 [3] NCCL INFO Channel 03/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942611:942772 [0] NCCL INFO Channel 08/0 : 40[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880683:880844 [4] NCCL INFO Channel 04/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO Channel 08/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926601:926759 [7] NCCL INFO Channel 07/0 : 63[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880682:880840 [3] NCCL INFO Channel 11/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926601:926759 [7] NCCL INFO Channel 07/0 : 55[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO Channel 08/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880683:880844 [4] NCCL INFO Channel 12/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926601:926759 [7] NCCL INFO Channel 15/0 : 55[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880686:880841 [7] NCCL INFO Channel 07/0 : 31[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618585:618748 [1] NCCL INFO Channel 09/0 : 41[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880686:880841 [7] NCCL INFO Channel 07/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688203:688362 [3] NCCL INFO Channel 11/0 : 43[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926599:926755 [5] NCCL INFO Channel 05/0 : 61[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926596:926756 [2] NCCL INFO Channel 02/0 : 58[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688204:688367 [4] NCCL INFO Channel 12/0 : 44[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-7:811249:811413 [1] NCCL INFO Channel 01/0 : 57[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618585:618748 [1] NCCL INFO Channel 09/0 : 25[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880686:880841 [7] NCCL INFO Channel 15/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688202:688368 [2] NCCL INFO Channel 02/0 : 50[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688205:688365 [5] NCCL INFO Channel 05/0 : 53[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688202:688368 [2] NCCL INFO Channel 02/0 : 34[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926599:926755 [5] NCCL INFO Channel 05/0 : 53[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926596:926756 [2] NCCL INFO Channel 02/0 : 50[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688205:688365 [5] NCCL INFO Channel 05/0 : 37[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688207:688366 [7] NCCL INFO Channel 15/0 : 47[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688207:688366 [7] NCCL INFO Channel 01/0 : 39[7] -> 32[0] via P2P/IPC
+ip-26-0-172-252:688206:688363 [6] NCCL INFO Channel 06/0 : 6[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880684:880842 [5] NCCL INFO Channel 05/0 : 29[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-7:811255:811411 [7] NCCL INFO Channel 01/0 : 63[7] -> 56[0] via P2P/IPC
+ip-26-0-172-252:688205:688365 [5] NCCL INFO Channel 13/0 : 45[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688202:688368 [2] NCCL INFO Channel 10/0 : 42[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880681:880845 [2] NCCL INFO Channel 02/0 : 26[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926596:926756 [2] NCCL INFO Channel 10/0 : 50[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926601:926759 [7] NCCL INFO Channel 01/0 : 55[7] -> 48[0] via P2P/IPC
+ip-26-0-172-73:926599:926755 [5] NCCL INFO Channel 13/0 : 53[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942618:942778 [7] NCCL INFO Channel 01/0 : 47[7] -> 40[0] via P2P/IPC
+ip-26-0-172-147:618591:618746 [7] NCCL INFO Channel 01/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-172-252:688206:688363 [6] NCCL INFO Channel 06/0 : 38[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880684:880842 [5] NCCL INFO Channel 05/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880681:880845 [2] NCCL INFO Channel 02/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942612:942779 [1] NCCL INFO Channel 01/0 : 49[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880684:880842 [5] NCCL INFO Channel 13/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880681:880845 [2] NCCL INFO Channel 10/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942612:942779 [1] NCCL INFO Channel 09/0 : 49[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880686:880841 [7] NCCL INFO Channel 01/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-172-252:688206:688363 [6] NCCL INFO Channel 06/0 : 54[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618585:618748 [1] NCCL INFO Channel 01/0 : 25[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:94348:94506 [6] NCCL INFO Channel 14/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942612:942779 [1] NCCL INFO Channel 09/0 : 41[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688206:688363 [6] NCCL INFO Channel 06/0 : 38[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926600:926757 [6] NCCL INFO Channel 06/0 : 62[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:94346:94511 [4] NCCL INFO Channel 15/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-225:94349:94505 [7] NCCL INFO Channel 02/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-252:688207:688366 [7] NCCL INFO Channel 02/0 : 39[7] -> 32[0] via P2P/IPC
+ip-26-0-172-73:926601:926759 [7] NCCL INFO Channel 02/0 : 55[7] -> 48[0] via P2P/IPC
+ip-26-0-172-252:688206:688363 [6] NCCL INFO Channel 14/0 : 46[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926600:926757 [6] NCCL INFO Channel 06/0 : 54[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880685:880847 [6] NCCL INFO Channel 06/0 : 30[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926600:926757 [6] NCCL INFO Channel 14/0 : 54[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880685:880847 [6] NCCL INFO Channel 06/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880685:880847 [6] NCCL INFO Channel 14/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618591:618746 [7] NCCL INFO Channel 02/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-172-57:942618:942778 [7] NCCL INFO Channel 02/0 : 47[7] -> 40[0] via P2P/IPC
+ip-26-0-172-142:880686:880841 [7] NCCL INFO Channel 02/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO Channel 01/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO Channel 09/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO Channel 09/0 : 9[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO Channel 01/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-160-225:94349:94505 [7] NCCL INFO Channel 03/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-73:926601:926759 [7] NCCL INFO Channel 03/0 : 55[7] -> 48[0] via P2P/IPC
+ip-26-0-172-142:880686:880841 [7] NCCL INFO Channel 03/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-172-252:688207:688366 [7] NCCL INFO Channel 03/0 : 39[7] -> 32[0] via P2P/IPC
+ip-26-0-173-7:811255:811411 [7] NCCL INFO Channel 02/0 : 63[7] -> 56[0] via P2P/IPC
+ip-26-0-173-7:811253:811410 [5] NCCL INFO Channel 00/0 : 61[5] -> 60[4] via P2P/IPC
+ip-26-0-172-147:618591:618746 [7] NCCL INFO Channel 03/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-172-57:942615:942775 [4] NCCL INFO Channel 00/0 : 44[4] -> 43[3] via P2P/IPC
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO Channel 02/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-160-225:94349:94505 [7] NCCL INFO Channel 04/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-252:688207:688366 [7] NCCL INFO Channel 04/0 : 39[7] -> 32[0] via P2P/IPC
+ip-26-0-173-7:811255:811411 [7] NCCL INFO Channel 03/0 : 63[7] -> 56[0] via P2P/IPC
+ip-26-0-172-73:926601:926759 [7] NCCL INFO Channel 04/0 : 55[7] -> 48[0] via P2P/IPC
+ip-26-0-172-57:942615:942775 [4] NCCL INFO Channel 03/0 : 44[4] -> 43[3] via P2P/IPC
+ip-26-0-173-7:811253:811410 [5] NCCL INFO Channel 04/0 : 61[5] -> 60[4] via P2P/IPC
+ip-26-0-172-73:926598:926761 [4] NCCL INFO Channel 00/0 : 52[4] -> 51[3] via P2P/IPC
+ip-26-0-172-142:880686:880841 [7] NCCL INFO Channel 04/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-172-57:942615:942775 [4] NCCL INFO Channel 05/0 : 44[4] -> 43[3] via P2P/IPC
+ip-26-0-172-57:942617:942774 [6] NCCL INFO Channel 00/0 : 46[6] -> 45[5] via P2P/IPC
+ip-26-0-172-252:688204:688367 [4] NCCL INFO Channel 00/0 : 36[4] -> 35[3] via P2P/IPC
+ip-26-0-172-147:618591:618746 [7] NCCL INFO Channel 04/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-172-142:880683:880844 [4] NCCL INFO Channel 00/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-173-7:811249:811413 [1] NCCL INFO Channel 00/0 : 57[1] -> 56[0] via P2P/IPC
+ip-26-0-160-225:94349:94505 [7] NCCL INFO Channel 05/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-252:688207:688366 [7] NCCL INFO Channel 05/0 : 39[7] -> 32[0] via P2P/IPC
+ip-26-0-160-225:94344:94510 [2] NCCL INFO Channel 00/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-225:94347:94507 [5] NCCL INFO Channel 00/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO Channel 03/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-173-7:811255:811411 [7] NCCL INFO Channel 04/0 : 63[7] -> 56[0] via P2P/IPC
+ip-26-0-172-147:618588:618750 [4] NCCL INFO Channel 00/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-172-73:926601:926759 [7] NCCL INFO Channel 05/0 : 55[7] -> 48[0] via P2P/IPC
+ip-26-0-173-7:811253:811410 [5] NCCL INFO Channel 08/0 : 61[5] -> 60[4] via P2P/IPC
+ip-26-0-172-252:688202:688368 [2] NCCL INFO Channel 00/0 : 34[2] -> 33[1] via P2P/IPC
+ip-26-0-172-73:926598:926761 [4] NCCL INFO Channel 03/0 : 52[4] -> 51[3] via P2P/IPC
+ip-26-0-172-142:880686:880841 [7] NCCL INFO Channel 05/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-172-252:688205:688365 [5] NCCL INFO Channel 00/0 : 37[5] -> 36[4] via P2P/IPC
+ip-26-0-172-252:688204:688367 [4] NCCL INFO Channel 03/0 : 36[4] -> 35[3] via P2P/IPC
+ip-26-0-173-7:811250:811412 [2] NCCL INFO Channel 00/0 : 58[2] -> 57[1] via P2P/IPC
+ip-26-0-160-225:94349:94505 [7] NCCL INFO Channel 06/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-73:926596:926756 [2] NCCL INFO Channel 00/0 : 50[2] -> 49[1] via P2P/IPC
+ip-26-0-172-142:880683:880844 [4] NCCL INFO Channel 03/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-172-57:942615:942775 [4] NCCL INFO Channel 06/0 : 44[4] -> 43[3] via P2P/IPC
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO Channel 00/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-172-73:926599:926755 [5] NCCL INFO Channel 00/0 : 53[5] -> 52[4] via P2P/IPC
+ip-26-0-160-225:94344:94510 [2] NCCL INFO Channel 01/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-57:942617:942774 [6] NCCL INFO Channel 05/0 : 46[6] -> 45[5] via P2P/IPC
+ip-26-0-172-57:942615:942775 [4] NCCL INFO Channel 07/0 : 44[4] -> 43[3] via P2P/IPC
+ip-26-0-172-57:942617:942774 [6] NCCL INFO Channel 08/0 : 46[6] -> 45[5] via P2P/IPC
+ip-26-0-172-57:942615:942775 [4] NCCL INFO Channel 08/0 : 44[4] -> 43[3] via P2P/IPC
+ip-26-0-160-225:94347:94507 [5] NCCL INFO Channel 04/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-57:942616:942777 [5] NCCL INFO Channel 00/0 : 45[5] -> 44[4] via P2P/IPC
+ip-26-0-173-7:811249:811413 [1] NCCL INFO Channel 08/0 : 57[1] -> 56[0] via P2P/IPC
+ip-26-0-172-252:688207:688366 [7] NCCL INFO Channel 06/0 : 39[7] -> 32[0] via P2P/IPC
+ip-26-0-172-57:942617:942774 [6] NCCL INFO Channel 13/0 : 46[6] -> 45[5] via P2P/IPC
+ip-26-0-160-225:94348:94506 [6] NCCL INFO Channel 00/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-142:880684:880842 [5] NCCL INFO Channel 00/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO Channel 04/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-172-252:688202:688368 [2] NCCL INFO Channel 01/0 : 34[2] -> 33[1] via P2P/IPC
+ip-26-0-172-57:942615:942775 [4] NCCL INFO Channel 11/0 : 44[4] -> 43[3] via P2P/IPC
+ip-26-0-172-73:926595:926760 [1] NCCL INFO Channel 00/0 : 49[1] -> 48[0] via P2P/IPC
+ip-26-0-172-57:942616:942777 [5] NCCL INFO Channel 04/0 : 45[5] -> 44[4] via P2P/IPC
+ip-26-0-172-142:880681:880845 [2] NCCL INFO Channel 00/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-172-73:926601:926759 [7] NCCL INFO Channel 06/0 : 55[7] -> 48[0] via P2P/IPC
+ip-26-0-173-7:811255:811411 [7] NCCL INFO Channel 05/0 : 63[7] -> 56[0] via P2P/IPC
+ip-26-0-173-7:811253:811410 [5] NCCL INFO Channel 12/0 : 61[5] -> 60[4] via P2P/IPC
+ip-26-0-172-73:926598:926761 [4] NCCL INFO Channel 05/0 : 52[4] -> 51[3] via P2P/IPC
+ip-26-0-172-252:688205:688365 [5] NCCL INFO Channel 04/0 : 37[5] -> 36[4] via P2P/IPC
+ip-26-0-173-7:811252:811408 [4] NCCL INFO Channel 00/0 : 60[4] -> 59[3] via P2P/IPC
+ip-26-0-172-252:688204:688367 [4] NCCL INFO Channel 05/0 : 36[4] -> 35[3] via P2P/IPC
+ip-26-0-160-225:94349:94505 [7] NCCL INFO Channel 07/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-173-7:811250:811412 [2] NCCL INFO Channel 01/0 : 58[2] -> 57[1] via P2P/IPC
+ip-26-0-172-142:880686:880841 [7] NCCL INFO Channel 06/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-172-147:618589:618744 [5] NCCL INFO Channel 00/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO Channel 00/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-172-57:942615:942775 [4] NCCL INFO Channel 13/0 : 44[4] -> 43[3] via P2P/IPC
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO Channel 00/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-172-57:942616:942777 [5] NCCL INFO Channel 08/0 : 45[5] -> 44[4] via P2P/IPC
+ip-26-0-172-147:618585:618748 [1] NCCL INFO Channel 00/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-172-147:618590:618749 [6] NCCL INFO Channel 00/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-160-225:94344:94510 [2] NCCL INFO Channel 08/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-142:880683:880844 [4] NCCL INFO Channel 05/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-172-73:926600:926757 [6] NCCL INFO Channel 00/0 : 54[6] -> 53[5] via P2P/IPC
+ip-26-0-172-252:688201:688364 [1] NCCL INFO Channel 00/0 : 33[1] -> 32[0] via P2P/IPC
+ip-26-0-172-147:618588:618750 [4] NCCL INFO Channel 03/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-160-225:94347:94507 [5] NCCL INFO Channel 08/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-252:688207:688366 [7] NCCL INFO Channel 07/0 : 39[7] -> 32[0] via P2P/IPC
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO Channel 05/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-172-73:926599:926755 [5] NCCL INFO Channel 04/0 : 53[5] -> 52[4] via P2P/IPC
+ip-26-0-160-225:94348:94506 [6] NCCL INFO Channel 05/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-142:880685:880847 [6] NCCL INFO Channel 00/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-172-147:618589:618744 [5] NCCL INFO Channel 04/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-172-252:688202:688368 [2] NCCL INFO Channel 08/0 : 34[2] -> 33[1] via P2P/IPC
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO Channel 05/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-172-147:618586:618745 [2] NCCL INFO Channel 00/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-172-142:880684:880842 [5] NCCL INFO Channel 04/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO Channel 00/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-172-142:880680:880843 [1] NCCL INFO Channel 00/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-172-73:926595:926760 [1] NCCL INFO Channel 08/0 : 49[1] -> 48[0] via P2P/IPC
+ip-26-0-173-7:811255:811411 [7] NCCL INFO Channel 06/0 : 63[7] -> 56[0] via P2P/IPC
+ip-26-0-172-147:618591:618746 [7] NCCL INFO Channel 05/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-172-73:926601:926759 [7] NCCL INFO Channel 07/0 : 55[7] -> 48[0] via P2P/IPC
+ip-26-0-172-142:880681:880845 [2] NCCL INFO Channel 01/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-172-252:688205:688365 [5] NCCL INFO Channel 08/0 : 37[5] -> 36[4] via P2P/IPC
+ip-26-0-172-57:942615:942775 [4] NCCL INFO Channel 14/0 : 44[4] -> 43[3] via P2P/IPC
+ip-26-0-172-57:942616:942777 [5] NCCL INFO Channel 12/0 : 45[5] -> 44[4] via P2P/IPC
+ip-26-0-172-252:688204:688367 [4] NCCL INFO Channel 06/0 : 36[4] -> 35[3] via P2P/IPC
+ip-26-0-172-73:926598:926761 [4] NCCL INFO Channel 06/0 : 52[4] -> 51[3] via P2P/IPC
+ip-26-0-160-225:94349:94505 [7] NCCL INFO Channel 09/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-173-7:811252:811408 [4] NCCL INFO Channel 03/0 : 60[4] -> 59[3] via P2P/IPC
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO Channel 03/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-173-7:811254:811414 [6] NCCL INFO Channel 00/0 : 62[6] -> 61[5] via P2P/IPC
+ip-26-0-173-7:811250:811412 [2] NCCL INFO Channel 08/0 : 58[2] -> 57[1] via P2P/IPC
+ip-26-0-172-147:618585:618748 [1] NCCL INFO Channel 08/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-172-142:880686:880841 [7] NCCL INFO Channel 07/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-172-147:618590:618749 [6] NCCL INFO Channel 05/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-160-225:94344:94510 [2] NCCL INFO Channel 09/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-142:880683:880844 [4] NCCL INFO Channel 06/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO Channel 08/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-160-225:94347:94507 [5] NCCL INFO Channel 12/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-225:94343:94508 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-252:688201:688364 [1] NCCL INFO Channel 08/0 : 33[1] -> 32[0] via P2P/IPC
+ip-26-0-172-73:926600:926757 [6] NCCL INFO Channel 05/0 : 54[6] -> 53[5] via P2P/IPC
+ip-26-0-172-252:688207:688366 [7] NCCL INFO Channel 09/0 : 39[7] -> 32[0] via P2P/IPC
+ip-26-0-172-73:926599:926755 [5] NCCL INFO Channel 08/0 : 53[5] -> 52[4] via P2P/IPC
+ip-26-0-172-147:618588:618750 [4] NCCL INFO Channel 05/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO Channel 00/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-160-225:94348:94506 [6] NCCL INFO Channel 08/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO Channel 08/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-173-7:811255:811411 [7] NCCL INFO Channel 07/0 : 63[7] -> 56[0] via P2P/IPC
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO Channel 06/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-172-57:942615:942775 [4] NCCL INFO Channel 15/0 : 44[4] -> 43[3] via P2P/IPC
+ip-26-0-172-252:688202:688368 [2] NCCL INFO Channel 09/0 : 34[2] -> 33[1] via P2P/IPC
+ip-26-0-172-147:618589:618744 [5] NCCL INFO Channel 08/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-172-73:926601:926759 [7] NCCL INFO Channel 09/0 : 55[7] -> 48[0] via P2P/IPC
+ip-26-0-173-7:811252:811408 [4] NCCL INFO Channel 05/0 : 60[4] -> 59[3] via P2P/IPC
+ip-26-0-173-7:811254:811414 [6] NCCL INFO Channel 05/0 : 62[6] -> 61[5] via P2P/IPC
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO Channel 04/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-172-57:942613:942776 [2] NCCL INFO Channel 00/0 : 42[2] -> 41[1] via P2P/IPC
+ip-26-0-172-147:618586:618745 [2] NCCL INFO Channel 01/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-173-7:811250:811412 [2] NCCL INFO Channel 09/0 : 58[2] -> 57[1] via P2P/IPC
+ip-26-0-172-57:942618:942778 [7] NCCL INFO Channel 03/0 : 47[7] -> 40[0] via P2P/IPC
+ip-26-0-172-73:926598:926761 [4] NCCL INFO Channel 07/0 : 52[4] -> 51[3] via P2P/IPC
+ip-26-0-172-252:688205:688365 [5] NCCL INFO Channel 12/0 : 37[5] -> 36[4] via P2P/IPC
+ip-26-0-172-252:688204:688367 [4] NCCL INFO Channel 07/0 : 36[4] -> 35[3] via P2P/IPC
+ip-26-0-172-147:618591:618746 [7] NCCL INFO Channel 06/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-172-142:880681:880845 [2] NCCL INFO Channel 08/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-172-73:926600:926757 [6] NCCL INFO Channel 08/0 : 54[6] -> 53[5] via P2P/IPC
+ip-26-0-172-57:942612:942779 [1] NCCL INFO Channel 00/0 : 41[1] -> 40[0] via P2P/IPC
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO Channel 05/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-172-73:926599:926755 [5] NCCL INFO Channel 12/0 : 53[5] -> 52[4] via P2P/IPC
+ip-26-0-160-225:94343:94508 [1] NCCL INFO Channel 08/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-225:94348:94506 [6] NCCL INFO Channel 13/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-173-7:811255:811411 [7] NCCL INFO Channel 09/0 : 63[7] -> 56[0] via P2P/IPC
+ip-26-0-172-147:618590:618749 [6] NCCL INFO Channel 08/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-172-252:688207:688366 [7] NCCL INFO Channel 10/0 : 39[7] -> 32[0] via P2P/IPC
+ip-26-0-172-142:880683:880844 [4] NCCL INFO Channel 07/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-172-73:926601:926759 [7] NCCL INFO Channel 10/0 : 55[7] -> 48[0] via P2P/IPC
+ip-26-0-173-7:811252:811408 [4] NCCL INFO Channel 06/0 : 60[4] -> 59[3] via P2P/IPC
+ip-26-0-172-142:880681:880845 [2] NCCL INFO Channel 09/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-172-147:618588:618750 [4] NCCL INFO Channel 06/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-172-142:880680:880843 [1] NCCL INFO Channel 08/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-172-142:880683:880844 [4] NCCL INFO Channel 08/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-172-57:942613:942776 [2] NCCL INFO Channel 01/0 : 42[2] -> 41[1] via P2P/IPC
+ip-26-0-172-73:926598:926761 [4] NCCL INFO Channel 08/0 : 52[4] -> 51[3] via P2P/IPC
+ip-26-0-173-7:811254:811414 [6] NCCL INFO Channel 08/0 : 62[6] -> 61[5] via P2P/IPC
+ip-26-0-172-142:880685:880847 [6] NCCL INFO Channel 05/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-160-225:94349:94505 [7] NCCL INFO Channel 10/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-57:942618:942778 [7] NCCL INFO Channel 04/0 : 47[7] -> 40[0] via P2P/IPC
+ip-26-0-172-57:942614:942773 [3] NCCL INFO Channel 00/0 : 43[3] -> 42[2] via P2P/IPC
+ip-26-0-172-147:618589:618744 [5] NCCL INFO Channel 12/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-172-142:880684:880842 [5] NCCL INFO Channel 08/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-160-225:94345:94509 [3] NCCL INFO Channel 00/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-73:926600:926757 [6] NCCL INFO Channel 13/0 : 54[6] -> 53[5] via P2P/IPC
+ip-26-0-172-252:688204:688367 [4] NCCL INFO Channel 08/0 : 36[4] -> 35[3] via P2P/IPC
+ip-26-0-172-142:880683:880844 [4] NCCL INFO Channel 11/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-172-147:618586:618745 [2] NCCL INFO Channel 08/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-172-142:880686:880841 [7] NCCL INFO Channel 09/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-173-7:811255:811411 [7] NCCL INFO Channel 10/0 : 63[7] -> 56[0] via P2P/IPC
+ip-26-0-160-225:94349:94505 [7] NCCL INFO Channel 11/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-147:618591:618746 [7] NCCL INFO Channel 07/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-172-252:688207:688366 [7] NCCL INFO Channel 11/0 : 39[7] -> 32[0] via P2P/IPC
+ip-26-0-172-73:926601:926759 [7] NCCL INFO Channel 11/0 : 55[7] -> 48[0] via P2P/IPC
+ip-26-0-172-73:926597:926758 [3] NCCL INFO Channel 00/0 : 51[3] -> 50[2] via P2P/IPC
+ip-26-0-172-142:880685:880847 [6] NCCL INFO Channel 08/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-172-147:618590:618749 [6] NCCL INFO Channel 13/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-172-73:926598:926761 [4] NCCL INFO Channel 11/0 : 52[4] -> 51[3] via P2P/IPC
+ip-26-0-173-7:811255:811411 [7] NCCL INFO Channel 11/0 : 63[7] -> 56[0] via P2P/IPC
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO Channel 00/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-172-73:926596:926756 [2] NCCL INFO Channel 01/0 : 50[2] -> 49[1] via P2P/IPC
+ip-26-0-172-252:688204:688367 [4] NCCL INFO Channel 11/0 : 36[4] -> 35[3] via P2P/IPC
+ip-26-0-172-142:880684:880842 [5] NCCL INFO Channel 12/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-173-7:811255:811411 [7] NCCL INFO Channel 12/0 : 63[7] -> 56[0] via P2P/IPC
+ip-26-0-172-252:688203:688362 [3] NCCL INFO Channel 00/0 : 35[3] -> 34[2] via P2P/IPC
+ip-26-0-172-147:618588:618750 [4] NCCL INFO Channel 07/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO Channel 02/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-172-142:880683:880844 [4] NCCL INFO Channel 13/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-172-142:880686:880841 [7] NCCL INFO Channel 10/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-172-252:688207:688366 [7] NCCL INFO Channel 12/0 : 39[7] -> 32[0] via P2P/IPC
+ip-26-0-173-7:811255:811411 [7] NCCL INFO Channel 13/0 : 63[7] -> 56[0] via P2P/IPC
+ip-26-0-172-73:926598:926761 [4] NCCL INFO Channel 13/0 : 52[4] -> 51[3] via P2P/IPC
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO Channel 08/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-172-147:618586:618745 [2] NCCL INFO Channel 09/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-172-57:942612:942779 [1] NCCL INFO Channel 08/0 : 41[1] -> 40[0] via P2P/IPC
+ip-26-0-172-73:926596:926756 [2] NCCL INFO Channel 08/0 : 50[2] -> 49[1] via P2P/IPC
+ip-26-0-173-7:811255:811411 [7] NCCL INFO Channel 14/0 : 63[7] -> 56[0] via P2P/IPC
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO Channel 10/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-172-142:880685:880847 [6] NCCL INFO Channel 13/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-172-147:618591:618746 [7] NCCL INFO Channel 09/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-173-7:811255:811411 [7] NCCL INFO Channel 15/0 : 63[7] -> 56[0] via P2P/IPC
+ip-26-0-172-73:926601:926759 [7] NCCL INFO Channel 12/0 : 55[7] -> 48[0] via P2P/IPC
+ip-26-0-172-252:688206:688363 [6] NCCL INFO Channel 00/0 : 38[6] -> 37[5] via P2P/IPC
+ip-26-0-172-73:926598:926761 [4] NCCL INFO Channel 14/0 : 52[4] -> 51[3] via P2P/IPC
+ip-26-0-172-252:688207:688366 [7] NCCL INFO Channel 13/0 : 39[7] -> 32[0] via P2P/IPC
+ip-26-0-173-7:811254:811414 [6] NCCL INFO Channel 13/0 : 62[6] -> 61[5] via P2P/IPC
+ip-26-0-172-147:618588:618750 [4] NCCL INFO Channel 08/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-172-142:880686:880841 [7] NCCL INFO Channel 11/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-172-73:926596:926756 [2] NCCL INFO Channel 09/0 : 50[2] -> 49[1] via P2P/IPC
+ip-26-0-160-225:94345:94509 [3] NCCL INFO Channel 02/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-142:880683:880844 [4] NCCL INFO Channel 14/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-172-252:688206:688363 [6] NCCL INFO Channel 05/0 : 38[6] -> 37[5] via P2P/IPC
+ip-26-0-172-142:880682:880840 [3] NCCL INFO Channel 00/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-172-252:688207:688366 [7] NCCL INFO Channel 14/0 : 39[7] -> 32[0] via P2P/IPC
+ip-26-0-172-73:926601:926759 [7] NCCL INFO Channel 13/0 : 55[7] -> 48[0] via P2P/IPC
+ip-26-0-172-147:618591:618746 [7] NCCL INFO Channel 10/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-172-252:688206:688363 [6] NCCL INFO Channel 08/0 : 38[6] -> 37[5] via P2P/IPC
+ip-26-0-173-7:811252:811408 [4] NCCL INFO Channel 07/0 : 60[4] -> 59[3] via P2P/IPC
+ip-26-0-172-252:688207:688366 [7] NCCL INFO Channel 15/0 : 39[7] -> 32[0] via P2P/IPC
+ip-26-0-172-73:926598:926761 [4] NCCL INFO Channel 15/0 : 52[4] -> 51[3] via P2P/IPC
+ip-26-0-172-147:618587:618751 [3] NCCL INFO Channel 00/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-172-142:880686:880841 [7] NCCL INFO Channel 12/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-172-147:618588:618750 [4] NCCL INFO Channel 11/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-172-252:688206:688363 [6] NCCL INFO Channel 13/0 : 38[6] -> 37[5] via P2P/IPC
+ip-26-0-172-142:880683:880844 [4] NCCL INFO Channel 15/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-173-7:811251:811409 [3] NCCL INFO Channel 00/0 : 59[3] -> 58[2] via P2P/IPC
+ip-26-0-172-142:880682:880840 [3] NCCL INFO Channel 02/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-172-73:926601:926759 [7] NCCL INFO Channel 14/0 : 55[7] -> 48[0] via P2P/IPC
+ip-26-0-172-252:688204:688367 [4] NCCL INFO Channel 13/0 : 36[4] -> 35[3] via P2P/IPC
+ip-26-0-173-7:811252:811408 [4] NCCL INFO Channel 08/0 : 60[4] -> 59[3] via P2P/IPC
+ip-26-0-172-73:926597:926758 [3] NCCL INFO Channel 02/0 : 51[3] -> 50[2] via P2P/IPC
+ip-26-0-172-252:688203:688362 [3] NCCL INFO Channel 02/0 : 35[3] -> 34[2] via P2P/IPC
+ip-26-0-172-147:618591:618746 [7] NCCL INFO Channel 11/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-172-147:618587:618751 [3] NCCL INFO Channel 02/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-173-7:811251:811409 [3] NCCL INFO Channel 02/0 : 59[3] -> 58[2] via P2P/IPC
+ip-26-0-172-57:942613:942776 [2] NCCL INFO Channel 08/0 : 42[2] -> 41[1] via P2P/IPC
+ip-26-0-172-147:618588:618750 [4] NCCL INFO Channel 13/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-172-142:880686:880841 [7] NCCL INFO Channel 13/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-172-73:926601:926759 [7] NCCL INFO Channel 15/0 : 55[7] -> 48[0] via P2P/IPC
+ip-26-0-173-7:811252:811408 [4] NCCL INFO Channel 11/0 : 60[4] -> 59[3] via P2P/IPC
+ip-26-0-172-142:880682:880840 [3] NCCL INFO Channel 08/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-172-252:688204:688367 [4] NCCL INFO Channel 14/0 : 36[4] -> 35[3] via P2P/IPC
+ip-26-0-173-7:811251:811409 [3] NCCL INFO Channel 08/0 : 59[3] -> 58[2] via P2P/IPC
+ip-26-0-172-147:618591:618746 [7] NCCL INFO Channel 12/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-172-147:618587:618751 [3] NCCL INFO Channel 08/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-172-252:688203:688362 [3] NCCL INFO Channel 08/0 : 35[3] -> 34[2] via P2P/IPC
+ip-26-0-173-7:811252:811408 [4] NCCL INFO Channel 13/0 : 60[4] -> 59[3] via P2P/IPC
+ip-26-0-172-147:618588:618750 [4] NCCL INFO Channel 14/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-172-252:688204:688367 [4] NCCL INFO Channel 15/0 : 36[4] -> 35[3] via P2P/IPC
+ip-26-0-173-7:811251:811409 [3] NCCL INFO Channel 10/0 : 59[3] -> 58[2] via P2P/IPC
+ip-26-0-173-7:811255:811411 [7] NCCL INFO Channel 00/0 : 63[7] -> 62[6] via P2P/IPC
+ip-26-0-173-7:811252:811408 [4] NCCL INFO Channel 14/0 : 60[4] -> 59[3] via P2P/IPC
+ip-26-0-172-147:618591:618746 [7] NCCL INFO Channel 13/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO Channel 13/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-172-252:688203:688362 [3] NCCL INFO Channel 10/0 : 35[3] -> 34[2] via P2P/IPC
+ip-26-0-160-225:94349:94505 [7] NCCL INFO Channel 12/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-147:618587:618751 [3] NCCL INFO Channel 10/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-172-147:618588:618750 [4] NCCL INFO Channel 15/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-172-252:688207:688366 [7] NCCL INFO Channel 00/0 : 39[7] -> 38[6] via P2P/IPC
+ip-26-0-172-57:942618:942778 [7] NCCL INFO Channel 05/0 : 47[7] -> 40[0] via P2P/IPC
+ip-26-0-172-147:618591:618746 [7] NCCL INFO Channel 14/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-172-57:942614:942773 [3] NCCL INFO Channel 02/0 : 43[3] -> 42[2] via P2P/IPC
+ip-26-0-173-7:811252:811408 [4] NCCL INFO Channel 15/0 : 60[4] -> 59[3] via P2P/IPC
+ip-26-0-172-73:926597:926758 [3] NCCL INFO Channel 08/0 : 51[3] -> 50[2] via P2P/IPC
+ip-26-0-160-225:94345:94509 [3] NCCL INFO Channel 08/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-142:880686:880841 [7] NCCL INFO Channel 14/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-172-147:618591:618746 [7] NCCL INFO Channel 15/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-172-142:880682:880840 [3] NCCL INFO Channel 10/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO Channel 06/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO Channel 07/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO Channel 07/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-173-7:811255:811411 [7] NCCL INFO Channel 06/0 : 63[7] -> 62[6] via P2P/IPC
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO Channel 08/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO Channel 08/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO Channel 09/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-160-225:94349:94505 [7] NCCL INFO Channel 13/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-73:926601:926759 [7] NCCL INFO Channel 00/0 : 55[7] -> 54[6] via P2P/IPC
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO Channel 12/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO Channel 11/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO Channel 10/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO Channel 13/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO Channel 11/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO Channel 14/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-172-73:926597:926758 [3] NCCL INFO Channel 10/0 : 51[3] -> 50[2] via P2P/IPC
+ip-26-0-173-7:811255:811411 [7] NCCL INFO Channel 08/0 : 63[7] -> 62[6] via P2P/IPC
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO Channel 01/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO Channel 12/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO Channel 15/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-172-142:880686:880841 [7] NCCL INFO Channel 15/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO Channel 08/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO Channel 13/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-172-57:942613:942776 [2] NCCL INFO Channel 09/0 : 42[2] -> 41[1] via P2P/IPC
+ip-26-0-172-73:926601:926759 [7] NCCL INFO Channel 06/0 : 55[7] -> 54[6] via P2P/IPC
+ip-26-0-172-57:942618:942778 [7] NCCL INFO Channel 06/0 : 47[7] -> 40[0] via P2P/IPC
+ip-26-0-172-57:942614:942773 [3] NCCL INFO Channel 08/0 : 43[3] -> 42[2] via P2P/IPC
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO Channel 09/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO Channel 14/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-160-225:94345:94509 [3] NCCL INFO Channel 10/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-147:618591:618746 [7] NCCL INFO Channel 00/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-172-57:942618:942778 [7] NCCL INFO Channel 07/0 : 47[7] -> 40[0] via P2P/IPC
+ip-26-0-160-225:94349:94505 [7] NCCL INFO Channel 14/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-57:942614:942773 [3] NCCL INFO Channel 10/0 : 43[3] -> 42[2] via P2P/IPC
+ip-26-0-172-73:926601:926759 [7] NCCL INFO Channel 08/0 : 55[7] -> 54[6] via P2P/IPC
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO Channel 15/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-172-142:880686:880841 [7] NCCL INFO Channel 00/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-160-225:94349:94505 [7] NCCL INFO Channel 15/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-147:618591:618746 [7] NCCL INFO Channel 06/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-172-57:942618:942778 [7] NCCL INFO Channel 09/0 : 47[7] -> 40[0] via P2P/IPC
+ip-26-0-172-57:942618:942778 [7] NCCL INFO Channel 10/0 : 47[7] -> 40[0] via P2P/IPC
+ip-26-0-172-252:688207:688366 [7] NCCL INFO Channel 06/0 : 39[7] -> 38[6] via P2P/IPC
+ip-26-0-173-7:811255:811411 [7] NCCL INFO Channel 14/0 : 63[7] -> 62[6] via P2P/IPC
+ip-26-0-172-147:618591:618746 [7] NCCL INFO Channel 08/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-172-142:880686:880841 [7] NCCL INFO Channel 06/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-172-142:880686:880841 [7] NCCL INFO Channel 08/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO Channel 00/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-172-57:942618:942778 [7] NCCL INFO Channel 11/0 : 47[7] -> 40[0] via P2P/IPC
+ip-26-0-172-147:618591:618746 [7] NCCL INFO Channel 14/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-172-142:880686:880841 [7] NCCL INFO Channel 14/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-160-225:94349:94505 [7] NCCL INFO Channel 00/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-252:688207:688366 [7] NCCL INFO Channel 08/0 : 39[7] -> 38[6] via P2P/IPC
+ip-26-0-172-57:942618:942778 [7] NCCL INFO Channel 12/0 : 47[7] -> 40[0] via P2P/IPC
+ip-26-0-172-73:926601:926759 [7] NCCL INFO Channel 14/0 : 55[7] -> 54[6] via P2P/IPC
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO Channel 06/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-160-225:94349:94505 [7] NCCL INFO Channel 06/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-252:688207:688366 [7] NCCL INFO Channel 14/0 : 39[7] -> 38[6] via P2P/IPC
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO Channel 08/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-172-57:942618:942778 [7] NCCL INFO Channel 13/0 : 47[7] -> 40[0] via P2P/IPC
+ip-26-0-160-225:94349:94505 [7] NCCL INFO Channel 08/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO Channel 14/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-160-225:94349:94505 [7] NCCL INFO Channel 14/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-57:942618:942778 [7] NCCL INFO Channel 14/0 : 47[7] -> 40[0] via P2P/IPC
+ip-26-0-172-57:942618:942778 [7] NCCL INFO Channel 15/0 : 47[7] -> 40[0] via P2P/IPC
+ip-26-0-172-57:942618:942778 [7] NCCL INFO Channel 00/0 : 47[7] -> 46[6] via P2P/IPC
+ip-26-0-172-57:942618:942778 [7] NCCL INFO Channel 06/0 : 47[7] -> 46[6] via P2P/IPC
+ip-26-0-172-57:942618:942778 [7] NCCL INFO Channel 08/0 : 47[7] -> 46[6] via P2P/IPC
+ip-26-0-172-57:942618:942778 [7] NCCL INFO Channel 14/0 : 47[7] -> 46[6] via P2P/IPC
+ip-26-0-173-7:811255:811411 [7] NCCL INFO Connected all trees
+ip-26-0-173-7:811255:811411 [7] NCCL INFO NVLS comm 0x9bdb100 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-225:94343:94508 [1] NCCL INFO Connected all trees
+ip-26-0-160-225:94343:94508 [1] NCCL INFO NVLS comm 0x82f6cb0 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-225:94342:94504 [0] NCCL INFO Connected all trees
+ip-26-0-160-225:94342:94504 [0] NCCL INFO NVLS comm 0x8eae1b0 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-142:880684:880842 [5] NCCL INFO Connected all trees
+ip-26-0-172-142:880684:880842 [5] NCCL INFO NVLS comm 0x88ad900 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-225:94347:94507 [5] NCCL INFO Connected all trees
+ip-26-0-160-225:94347:94507 [5] NCCL INFO NVLS comm 0x90b4a60 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-142:880682:880840 [3] NCCL INFO Connected all trees
+ip-26-0-172-142:880682:880840 [3] NCCL INFO NVLS comm 0x928d640 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-142:880680:880843 [1] NCCL INFO Connected all trees
+ip-26-0-172-142:880680:880843 [1] NCCL INFO NVLS comm 0x8da1b60 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-73:926600:926757 [6] NCCL INFO Connected all trees
+ip-26-0-172-73:926600:926757 [6] NCCL INFO NVLS comm 0x9e0a1c0 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-142:880685:880847 [6] NCCL INFO Connected all trees
+ip-26-0-172-142:880685:880847 [6] NCCL INFO NVLS comm 0x876ac20 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-225:94348:94506 [6] NCCL INFO Connected all trees
+ip-26-0-160-225:94348:94506 [6] NCCL INFO NVLS comm 0x97f3050 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-73:926595:926760 [1] NCCL INFO Connected all trees
+ip-26-0-172-73:926595:926760 [1] NCCL INFO NVLS comm 0x82893c0 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-73:926594:926762 [0] NCCL INFO Connected all trees
+ip-26-0-172-73:926594:926762 [0] NCCL INFO NVLS comm 0x920c450 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-73:926601:926759 [7] NCCL INFO Connected all trees
+ip-26-0-172-73:926601:926759 [7] NCCL INFO NVLS comm 0x8367300 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-57:942613:942776 [2] NCCL INFO Connected all trees
+ip-26-0-172-57:942613:942776 [2] NCCL INFO NVLS comm 0x995bf70 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-173-7:811251:811409 [3] NCCL INFO Connected all trees
+ip-26-0-173-7:811253:811410 [5] NCCL INFO Connected all trees
+ip-26-0-173-7:811253:811410 [5] NCCL INFO NVLS comm 0x95fcf60 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-173-7:811251:811409 [3] NCCL INFO NVLS comm 0x8b1a600 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-173-7:811254:811414 [6] NCCL INFO Connected all trees
+ip-26-0-173-7:811249:811413 [1] NCCL INFO Connected all trees
+ip-26-0-173-7:811254:811414 [6] NCCL INFO NVLS comm 0x89213d0 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-173-7:811252:811408 [4] NCCL INFO Connected all trees
+ip-26-0-173-7:811249:811413 [1] NCCL INFO NVLS comm 0x8794560 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-173-7:811252:811408 [4] NCCL INFO NVLS comm 0x8c2ebd0 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-57:942616:942777 [5] NCCL INFO Connected all trees
+ip-26-0-172-57:942616:942777 [5] NCCL INFO NVLS comm 0x84435f0 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-57:942614:942773 [3] NCCL INFO Connected all trees
+ip-26-0-172-57:942614:942773 [3] NCCL INFO NVLS comm 0x89687c0 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-57:942615:942775 [4] NCCL INFO Connected all trees
+ip-26-0-172-57:942615:942775 [4] NCCL INFO NVLS comm 0x9ebb790 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-57:942612:942779 [1] NCCL INFO Connected all trees
+ip-26-0-172-57:942612:942779 [1] NCCL INFO NVLS comm 0x891af90 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-225:94344:94510 [2] NCCL INFO Connected all trees
+ip-26-0-160-225:94344:94510 [2] NCCL INFO NVLS comm 0x9047390 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-57:942617:942774 [6] NCCL INFO Connected all trees
+ip-26-0-172-57:942617:942774 [6] NCCL INFO NVLS comm 0x959be20 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-225:94346:94511 [4] NCCL INFO Connected all trees
+ip-26-0-160-225:94346:94511 [4] NCCL INFO NVLS comm 0xa00d090 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-173-7:811248:811415 [0] NCCL INFO Connected all trees
+ip-26-0-173-7:811248:811415 [0] NCCL INFO NVLS comm 0xa104f70 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-225:94345:94509 [3] NCCL INFO Connected all trees
+ip-26-0-173-7:811250:811412 [2] NCCL INFO Connected all trees
+ip-26-0-160-225:94345:94509 [3] NCCL INFO NVLS comm 0xa1ae2d0 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-173-7:811250:811412 [2] NCCL INFO NVLS comm 0x958ce00 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-73:926596:926756 [2] NCCL INFO Connected all trees
+ip-26-0-172-73:926596:926756 [2] NCCL INFO NVLS comm 0x8bdc2a0 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO Connected all trees
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO Connected all trees
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO Connected all trees
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO NVLS comm 0x83bb440 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO NVLS comm 0xa1a18a0 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO NVLS comm 0xa1c3a50 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-225:94349:94505 [7] NCCL INFO Connected all trees
+ip-26-0-160-225:94349:94505 [7] NCCL INFO NVLS comm 0x89eb7d0 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO Connected all trees
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO NVLS comm 0x84277a0 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO Connected all trees
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO NVLS comm 0xa199350 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO Connected all trees
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO Connected all trees
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO NVLS comm 0x89c15f0 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO NVLS comm 0x9e252e0 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-73:926598:926761 [4] NCCL INFO Connected all trees
+ip-26-0-172-73:926598:926761 [4] NCCL INFO NVLS comm 0x96d54b0 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-73:926597:926758 [3] NCCL INFO Connected all trees
+ip-26-0-172-73:926597:926758 [3] NCCL INFO NVLS comm 0x8ae3590 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-57:942611:942772 [0] NCCL INFO Connected all trees
+ip-26-0-172-57:942611:942772 [0] NCCL INFO NVLS comm 0x9b57e50 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-57:942618:942778 [7] NCCL INFO Connected all trees
+ip-26-0-172-57:942618:942778 [7] NCCL INFO NVLS comm 0x8e28120 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO Connected all trees
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO NVLS comm 0x97f7620 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-73:926599:926755 [5] NCCL INFO Connected all trees
+ip-26-0-172-73:926599:926755 [5] NCCL INFO NVLS comm 0x9df4700 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-252:688201:688364 [1] NCCL INFO Connected all trees
+ip-26-0-172-252:688201:688364 [1] NCCL INFO NVLS comm 0x91aa790 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-252:688206:688363 [6] NCCL INFO Connected all trees
+ip-26-0-172-252:688206:688363 [6] NCCL INFO NVLS comm 0xa251280 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-252:688203:688362 [3] NCCL INFO Connected all trees
+ip-26-0-172-252:688203:688362 [3] NCCL INFO NVLS comm 0x8600940 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-252:688205:688365 [5] NCCL INFO Connected all trees
+ip-26-0-172-252:688205:688365 [5] NCCL INFO NVLS comm 0x8c6c940 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-252:688202:688368 [2] NCCL INFO Connected all trees
+ip-26-0-172-252:688202:688368 [2] NCCL INFO NVLS comm 0x900a490 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-252:688200:688369 [0] NCCL INFO Connected all trees
+ip-26-0-172-252:688200:688369 [0] NCCL INFO NVLS comm 0x920bad0 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-252:688207:688366 [7] NCCL INFO Connected all trees
+ip-26-0-172-252:688207:688366 [7] NCCL INFO NVLS comm 0x8cf40a0 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-252:688204:688367 [4] NCCL INFO Connected all trees
+ip-26-0-172-252:688204:688367 [4] NCCL INFO NVLS comm 0x9286a60 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-147:618586:618745 [2] NCCL INFO Connected all trees
+ip-26-0-172-147:618586:618745 [2] NCCL INFO NVLS comm 0x89463c0 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-147:618584:618747 [0] NCCL INFO Connected all trees
+ip-26-0-173-7:811252:811408 [4] NCCL INFO Channel 00/0 : 52[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-7:811248:811415 [0] NCCL INFO Channel 02/0 : 48[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-7:811252:811408 [4] NCCL INFO Channel 02/0 : 52[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-7:811254:811414 [6] NCCL INFO Channel 00/0 : 54[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-7:811248:811415 [0] NCCL INFO Channel 04/0 : 48[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-7:811252:811408 [4] NCCL INFO Channel 06/0 : 52[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-7:811254:811414 [6] NCCL INFO Channel 02/0 : 54[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-7:811252:811408 [4] NCCL INFO Channel 08/0 : 52[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-7:811248:811415 [0] NCCL INFO Channel 06/0 : 48[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-7:811254:811414 [6] NCCL INFO Channel 04/0 : 54[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618584:618747 [0] NCCL INFO NVLS comm 0x9905f70 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-173-7:811248:811415 [0] NCCL INFO Channel 08/0 : 48[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-7:811252:811408 [4] NCCL INFO Channel 10/0 : 52[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-7:811254:811414 [6] NCCL INFO Channel 08/0 : 54[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-7:811252:811408 [4] NCCL INFO Channel 12/0 : 52[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-7:811254:811414 [6] NCCL INFO Channel 10/0 : 54[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-7:811248:811415 [0] NCCL INFO Channel 10/0 : 48[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-7:811252:811408 [4] NCCL INFO Channel 14/0 : 52[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-7:811248:811415 [0] NCCL INFO Channel 12/0 : 48[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-7:811254:811414 [6] NCCL INFO Channel 12/0 : 54[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-7:811248:811415 [0] NCCL INFO Channel 14/0 : 48[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-7:811254:811414 [6] NCCL INFO Channel 14/0 : 54[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618590:618749 [6] NCCL INFO Connected all trees
+ip-26-0-173-7:811255:811411 [7] NCCL INFO Channel 00/0 : 55[7] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-7:811250:811412 [2] NCCL INFO Channel 00/0 : 50[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-7:811253:811410 [5] NCCL INFO Channel 00/0 : 53[5] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-7:811255:811411 [7] NCCL INFO Channel 02/0 : 55[7] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618590:618749 [6] NCCL INFO NVLS comm 0x9f398d0 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-173-7:811250:811412 [2] NCCL INFO Channel 04/0 : 50[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-7:811253:811410 [5] NCCL INFO Channel 02/0 : 53[5] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-7:811255:811411 [7] NCCL INFO Channel 04/0 : 55[7] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-7:811249:811413 [1] NCCL INFO Channel 00/0 : 49[1] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-7:811250:811412 [2] NCCL INFO Channel 06/0 : 50[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-7:811253:811410 [5] NCCL INFO Channel 04/0 : 53[5] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-7:811251:811409 [3] NCCL INFO Channel 00/0 : 51[3] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-7:811249:811413 [1] NCCL INFO Channel 02/0 : 49[1] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-7:811255:811411 [7] NCCL INFO Channel 06/0 : 55[7] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-7:811250:811412 [2] NCCL INFO Channel 08/0 : 50[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-7:811251:811409 [3] NCCL INFO Channel 02/0 : 51[3] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-7:811253:811410 [5] NCCL INFO Channel 06/0 : 53[5] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-7:811249:811413 [1] NCCL INFO Channel 04/0 : 49[1] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-7:811255:811411 [7] NCCL INFO Channel 08/0 : 55[7] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-7:811250:811412 [2] NCCL INFO Channel 10/0 : 50[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-7:811251:811409 [3] NCCL INFO Channel 04/0 : 51[3] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-7:811249:811413 [1] NCCL INFO Channel 06/0 : 49[1] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-7:811253:811410 [5] NCCL INFO Channel 08/0 : 53[5] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-7:811255:811411 [7] NCCL INFO Channel 10/0 : 55[7] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-7:811250:811412 [2] NCCL INFO Channel 12/0 : 50[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-7:811251:811409 [3] NCCL INFO Channel 06/0 : 51[3] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-7:811249:811413 [1] NCCL INFO Channel 08/0 : 49[1] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-7:811253:811410 [5] NCCL INFO Channel 10/0 : 53[5] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-7:811250:811412 [2] NCCL INFO Channel 14/0 : 50[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-7:811255:811411 [7] NCCL INFO Channel 12/0 : 55[7] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-7:811251:811409 [3] NCCL INFO Channel 08/0 : 51[3] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-7:811249:811413 [1] NCCL INFO Channel 10/0 : 49[1] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-7:811253:811410 [5] NCCL INFO Channel 12/0 : 53[5] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-7:811255:811411 [7] NCCL INFO Channel 14/0 : 55[7] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-7:811251:811409 [3] NCCL INFO Channel 10/0 : 51[3] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-7:811249:811413 [1] NCCL INFO Channel 12/0 : 49[1] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-7:811253:811410 [5] NCCL INFO Channel 14/0 : 53[5] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-7:811249:811413 [1] NCCL INFO Channel 14/0 : 49[1] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-7:811251:811409 [3] NCCL INFO Channel 12/0 : 51[3] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-7:811251:811409 [3] NCCL INFO Channel 14/0 : 51[3] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618589:618744 [5] NCCL INFO Connected all trees
+ip-26-0-172-147:618589:618744 [5] NCCL INFO NVLS comm 0x962f830 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-225:94346:94511 [4] NCCL INFO Channel 01/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:94348:94506 [6] NCCL INFO Channel 01/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:94346:94511 [4] NCCL INFO Channel 03/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:94342:94504 [0] NCCL INFO Channel 01/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:94348:94506 [6] NCCL INFO Channel 03/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618591:618746 [7] NCCL INFO Connected all trees
+ip-26-0-160-225:94346:94511 [4] NCCL INFO Channel 05/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:94342:94504 [0] NCCL INFO Channel 03/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:94348:94506 [6] NCCL INFO Channel 05/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:94346:94511 [4] NCCL INFO Channel 07/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:94342:94504 [0] NCCL INFO Channel 05/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:94348:94506 [6] NCCL INFO Channel 07/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:94346:94511 [4] NCCL INFO Channel 09/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:94342:94504 [0] NCCL INFO Channel 07/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:94348:94506 [6] NCCL INFO Channel 09/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:94342:94504 [0] NCCL INFO Channel 09/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:94346:94511 [4] NCCL INFO Channel 11/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:94348:94506 [6] NCCL INFO Channel 11/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:94346:94511 [4] NCCL INFO Channel 13/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:94348:94506 [6] NCCL INFO Channel 13/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:94342:94504 [0] NCCL INFO Channel 11/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:94346:94511 [4] NCCL INFO Channel 15/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:94348:94506 [6] NCCL INFO Channel 15/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618591:618746 [7] NCCL INFO NVLS comm 0x91d1750 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-225:94342:94504 [0] NCCL INFO Channel 13/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:94342:94504 [0] NCCL INFO Channel 15/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:94343:94508 [1] NCCL INFO Channel 01/0 : 1[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:94349:94505 [7] NCCL INFO Channel 01/0 : 7[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:94343:94508 [1] NCCL INFO Channel 03/0 : 1[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:94349:94505 [7] NCCL INFO Channel 03/0 : 7[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:94343:94508 [1] NCCL INFO Channel 05/0 : 1[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:94349:94505 [7] NCCL INFO Channel 05/0 : 7[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:94343:94508 [1] NCCL INFO Channel 07/0 : 1[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:94349:94505 [7] NCCL INFO Channel 07/0 : 7[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:94347:94507 [5] NCCL INFO Channel 01/0 : 5[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:94344:94510 [2] NCCL INFO Channel 01/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:94343:94508 [1] NCCL INFO Channel 11/0 : 1[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:94349:94505 [7] NCCL INFO Channel 09/0 : 7[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:94347:94507 [5] NCCL INFO Channel 03/0 : 5[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:94344:94510 [2] NCCL INFO Channel 03/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:94343:94508 [1] NCCL INFO Channel 13/0 : 1[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:94345:94509 [3] NCCL INFO Channel 01/0 : 3[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:94349:94505 [7] NCCL INFO Channel 11/0 : 7[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:94347:94507 [5] NCCL INFO Channel 05/0 : 5[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:94344:94510 [2] NCCL INFO Channel 05/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:94343:94508 [1] NCCL INFO Channel 15/0 : 1[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:94345:94509 [3] NCCL INFO Channel 03/0 : 3[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:94349:94505 [7] NCCL INFO Channel 13/0 : 7[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:94347:94507 [5] NCCL INFO Channel 07/0 : 5[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:94344:94510 [2] NCCL INFO Channel 07/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:94345:94509 [3] NCCL INFO Channel 05/0 : 3[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:94347:94507 [5] NCCL INFO Channel 09/0 : 5[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:94344:94510 [2] NCCL INFO Channel 09/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:94345:94509 [3] NCCL INFO Channel 07/0 : 3[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:94347:94507 [5] NCCL INFO Channel 11/0 : 5[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:94344:94510 [2] NCCL INFO Channel 11/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:94345:94509 [3] NCCL INFO Channel 09/0 : 3[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:94347:94507 [5] NCCL INFO Channel 15/0 : 5[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:94344:94510 [2] NCCL INFO Channel 13/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:94345:94509 [3] NCCL INFO Channel 13/0 : 3[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:94345:94509 [3] NCCL INFO Channel 15/0 : 3[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:94344:94510 [2] NCCL INFO Channel 15/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618588:618750 [4] NCCL INFO Connected all trees
+ip-26-0-172-147:618588:618750 [4] NCCL INFO NVLS comm 0x8654810 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-147:618587:618751 [3] NCCL INFO Connected all trees
+ip-26-0-172-147:618587:618751 [3] NCCL INFO NVLS comm 0x8914d00 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-147:618585:618748 [1] NCCL INFO Connected all trees
+ip-26-0-172-147:618585:618748 [1] NCCL INFO NVLS comm 0x9e17fb0 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-57:942617:942774 [6] NCCL INFO Channel 01/0 : 38[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942615:942775 [4] NCCL INFO Channel 01/0 : 36[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942617:942774 [6] NCCL INFO Channel 03/0 : 38[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942617:942774 [6] NCCL INFO Channel 05/0 : 38[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942615:942775 [4] NCCL INFO Channel 03/0 : 36[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942617:942774 [6] NCCL INFO Channel 07/0 : 38[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942615:942775 [4] NCCL INFO Channel 05/0 : 36[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942611:942772 [0] NCCL INFO Channel 01/0 : 32[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942617:942774 [6] NCCL INFO Channel 09/0 : 38[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942615:942775 [4] NCCL INFO Channel 07/0 : 36[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942611:942772 [0] NCCL INFO Channel 03/0 : 32[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942617:942774 [6] NCCL INFO Channel 11/0 : 38[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942615:942775 [4] NCCL INFO Channel 09/0 : 36[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942617:942774 [6] NCCL INFO Channel 13/0 : 38[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942611:942772 [0] NCCL INFO Channel 05/0 : 32[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942615:942775 [4] NCCL INFO Channel 11/0 : 36[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942617:942774 [6] NCCL INFO Channel 15/0 : 38[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942611:942772 [0] NCCL INFO Channel 07/0 : 32[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942615:942775 [4] NCCL INFO Channel 13/0 : 36[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942617:942774 [6] NCCL INFO Channel 00/0 : 46[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942611:942772 [0] NCCL INFO Channel 09/0 : 32[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942615:942775 [4] NCCL INFO Channel 15/0 : 36[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942617:942774 [6] NCCL INFO Channel 01/0 : 46[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942611:942772 [0] NCCL INFO Channel 11/0 : 32[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942615:942775 [4] NCCL INFO Channel 00/0 : 44[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942617:942774 [6] NCCL INFO Channel 02/0 : 46[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942611:942772 [0] NCCL INFO Channel 13/0 : 32[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942615:942775 [4] NCCL INFO Channel 01/0 : 44[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942617:942774 [6] NCCL INFO Channel 03/0 : 46[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942611:942772 [0] NCCL INFO Channel 15/0 : 32[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942615:942775 [4] NCCL INFO Channel 02/0 : 44[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942617:942774 [6] NCCL INFO Channel 04/0 : 46[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942611:942772 [0] NCCL INFO Channel 01/0 : 40[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942615:942775 [4] NCCL INFO Channel 03/0 : 44[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942617:942774 [6] NCCL INFO Channel 05/0 : 46[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942611:942772 [0] NCCL INFO Channel 02/0 : 40[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942617:942774 [6] NCCL INFO Channel 07/0 : 46[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942615:942775 [4] NCCL INFO Channel 05/0 : 44[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942611:942772 [0] NCCL INFO Channel 03/0 : 40[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942617:942774 [6] NCCL INFO Channel 08/0 : 46[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942615:942775 [4] NCCL INFO Channel 06/0 : 44[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942611:942772 [0] NCCL INFO Channel 04/0 : 40[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942617:942774 [6] NCCL INFO Channel 09/0 : 46[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942615:942775 [4] NCCL INFO Channel 07/0 : 44[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942611:942772 [0] NCCL INFO Channel 05/0 : 40[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942617:942774 [6] NCCL INFO Channel 10/0 : 46[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942618:942778 [7] NCCL INFO Channel 01/0 : 39[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942615:942775 [4] NCCL INFO Channel 08/0 : 44[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942611:942772 [0] NCCL INFO Channel 06/0 : 40[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942612:942779 [1] NCCL INFO Channel 01/0 : 33[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942616:942777 [5] NCCL INFO Channel 01/0 : 37[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942617:942774 [6] NCCL INFO Channel 11/0 : 46[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942615:942775 [4] NCCL INFO Channel 09/0 : 44[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942618:942778 [7] NCCL INFO Channel 03/0 : 39[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942611:942772 [0] NCCL INFO Channel 07/0 : 40[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942613:942776 [2] NCCL INFO Channel 01/0 : 34[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942612:942779 [1] NCCL INFO Channel 03/0 : 33[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942614:942773 [3] NCCL INFO Channel 01/0 : 35[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942617:942774 [6] NCCL INFO Channel 12/0 : 46[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942616:942777 [5] NCCL INFO Channel 03/0 : 37[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942615:942775 [4] NCCL INFO Channel 10/0 : 44[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942611:942772 [0] NCCL INFO Channel 09/0 : 40[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942618:942778 [7] NCCL INFO Channel 05/0 : 39[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942613:942776 [2] NCCL INFO Channel 03/0 : 34[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942612:942779 [1] NCCL INFO Channel 05/0 : 33[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942617:942774 [6] NCCL INFO Channel 13/0 : 46[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942614:942773 [3] NCCL INFO Channel 03/0 : 35[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942616:942777 [5] NCCL INFO Channel 05/0 : 37[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942615:942775 [4] NCCL INFO Channel 11/0 : 44[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942611:942772 [0] NCCL INFO Channel 10/0 : 40[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942618:942778 [7] NCCL INFO Channel 07/0 : 39[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942613:942776 [2] NCCL INFO Channel 05/0 : 34[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942612:942779 [1] NCCL INFO Channel 07/0 : 33[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942617:942774 [6] NCCL INFO Channel 15/0 : 46[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942614:942773 [3] NCCL INFO Channel 05/0 : 35[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942616:942777 [5] NCCL INFO Channel 07/0 : 37[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942611:942772 [0] NCCL INFO Channel 11/0 : 40[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942615:942775 [4] NCCL INFO Channel 13/0 : 44[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942618:942778 [7] NCCL INFO Channel 09/0 : 39[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942612:942779 [1] NCCL INFO Channel 11/0 : 33[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942613:942776 [2] NCCL INFO Channel 07/0 : 34[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942614:942773 [3] NCCL INFO Channel 07/0 : 35[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942611:942772 [0] NCCL INFO Channel 12/0 : 40[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942615:942775 [4] NCCL INFO Channel 14/0 : 44[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942616:942777 [5] NCCL INFO Channel 09/0 : 37[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942612:942779 [1] NCCL INFO Channel 13/0 : 33[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942618:942778 [7] NCCL INFO Channel 11/0 : 39[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942613:942776 [2] NCCL INFO Channel 09/0 : 34[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942611:942772 [0] NCCL INFO Channel 13/0 : 40[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942614:942773 [3] NCCL INFO Channel 09/0 : 35[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942616:942777 [5] NCCL INFO Channel 11/0 : 37[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942615:942775 [4] NCCL INFO Channel 15/0 : 44[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942618:942778 [7] NCCL INFO Channel 13/0 : 39[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942613:942776 [2] NCCL INFO Channel 11/0 : 34[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942612:942779 [1] NCCL INFO Channel 15/0 : 33[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942611:942772 [0] NCCL INFO Channel 14/0 : 40[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942614:942773 [3] NCCL INFO Channel 13/0 : 35[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942616:942777 [5] NCCL INFO Channel 15/0 : 37[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942618:942778 [7] NCCL INFO Channel 00/0 : 47[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942612:942779 [1] NCCL INFO Channel 00/0 : 41[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942611:942772 [0] NCCL INFO Channel 15/0 : 40[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942613:942776 [2] NCCL INFO Channel 13/0 : 34[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942614:942773 [3] NCCL INFO Channel 15/0 : 35[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942616:942777 [5] NCCL INFO Channel 00/0 : 45[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942618:942778 [7] NCCL INFO Channel 01/0 : 47[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942612:942779 [1] NCCL INFO Channel 02/0 : 41[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942613:942776 [2] NCCL INFO Channel 15/0 : 34[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942614:942773 [3] NCCL INFO Channel 00/0 : 43[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942616:942777 [5] NCCL INFO Channel 01/0 : 45[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942618:942778 [7] NCCL INFO Channel 02/0 : 47[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942612:942779 [1] NCCL INFO Channel 03/0 : 41[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942613:942776 [2] NCCL INFO Channel 00/0 : 42[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942616:942777 [5] NCCL INFO Channel 02/0 : 45[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942614:942773 [3] NCCL INFO Channel 01/0 : 43[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942612:942779 [1] NCCL INFO Channel 04/0 : 41[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942618:942778 [7] NCCL INFO Channel 03/0 : 47[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942613:942776 [2] NCCL INFO Channel 01/0 : 42[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942616:942777 [5] NCCL INFO Channel 03/0 : 45[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942614:942773 [3] NCCL INFO Channel 02/0 : 43[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942612:942779 [1] NCCL INFO Channel 05/0 : 41[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942613:942776 [2] NCCL INFO Channel 03/0 : 42[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942618:942778 [7] NCCL INFO Channel 04/0 : 47[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942616:942777 [5] NCCL INFO Channel 04/0 : 45[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942614:942773 [3] NCCL INFO Channel 04/0 : 43[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942612:942779 [1] NCCL INFO Channel 06/0 : 41[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942613:942776 [2] NCCL INFO Channel 04/0 : 42[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942618:942778 [7] NCCL INFO Channel 05/0 : 47[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942616:942777 [5] NCCL INFO Channel 06/0 : 45[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942614:942773 [3] NCCL INFO Channel 05/0 : 43[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942612:942779 [1] NCCL INFO Channel 07/0 : 41[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942613:942776 [2] NCCL INFO Channel 05/0 : 42[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942618:942778 [7] NCCL INFO Channel 06/0 : 47[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942616:942777 [5] NCCL INFO Channel 07/0 : 45[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942614:942773 [3] NCCL INFO Channel 06/0 : 43[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942612:942779 [1] NCCL INFO Channel 08/0 : 41[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942613:942776 [2] NCCL INFO Channel 06/0 : 42[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942618:942778 [7] NCCL INFO Channel 08/0 : 47[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942614:942773 [3] NCCL INFO Channel 07/0 : 43[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942616:942777 [5] NCCL INFO Channel 08/0 : 45[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942612:942779 [1] NCCL INFO Channel 10/0 : 41[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942613:942776 [2] NCCL INFO Channel 07/0 : 42[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942618:942778 [7] NCCL INFO Channel 09/0 : 47[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942614:942773 [3] NCCL INFO Channel 08/0 : 43[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942612:942779 [1] NCCL INFO Channel 11/0 : 41[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942616:942777 [5] NCCL INFO Channel 09/0 : 45[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942613:942776 [2] NCCL INFO Channel 08/0 : 42[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942618:942778 [7] NCCL INFO Channel 10/0 : 47[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942614:942773 [3] NCCL INFO Channel 09/0 : 43[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942612:942779 [1] NCCL INFO Channel 12/0 : 41[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942616:942777 [5] NCCL INFO Channel 10/0 : 45[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942618:942778 [7] NCCL INFO Channel 11/0 : 47[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942613:942776 [2] NCCL INFO Channel 09/0 : 42[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942614:942773 [3] NCCL INFO Channel 10/0 : 43[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942612:942779 [1] NCCL INFO Channel 13/0 : 41[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942616:942777 [5] NCCL INFO Channel 11/0 : 45[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942618:942778 [7] NCCL INFO Channel 12/0 : 47[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942613:942776 [2] NCCL INFO Channel 11/0 : 42[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942614:942773 [3] NCCL INFO Channel 12/0 : 43[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942612:942779 [1] NCCL INFO Channel 14/0 : 41[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942616:942777 [5] NCCL INFO Channel 12/0 : 45[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942618:942778 [7] NCCL INFO Channel 13/0 : 47[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO Channel 01/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942613:942776 [2] NCCL INFO Channel 12/0 : 42[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942614:942773 [3] NCCL INFO Channel 13/0 : 43[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942612:942779 [1] NCCL INFO Channel 15/0 : 41[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942616:942777 [5] NCCL INFO Channel 14/0 : 45[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942618:942778 [7] NCCL INFO Channel 14/0 : 47[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO Channel 03/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942613:942776 [2] NCCL INFO Channel 13/0 : 42[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942614:942773 [3] NCCL INFO Channel 14/0 : 43[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942616:942777 [5] NCCL INFO Channel 15/0 : 45[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO Channel 01/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942613:942776 [2] NCCL INFO Channel 14/0 : 42[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942614:942773 [3] NCCL INFO Channel 15/0 : 43[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO Channel 05/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942613:942776 [2] NCCL INFO Channel 15/0 : 42[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO Channel 03/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO Channel 07/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO Channel 05/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO Channel 09/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO Channel 07/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO Channel 11/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO Channel 09/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO Channel 13/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO Channel 11/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO Channel 15/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO Channel 13/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO Channel 00/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO Channel 01/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO Channel 15/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO Channel 02/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO Channel 01/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO Channel 03/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO Channel 02/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO Channel 04/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO Channel 03/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO Channel 05/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO Channel 04/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO Channel 07/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO Channel 05/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO Channel 08/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO Channel 06/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO Channel 09/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO Channel 07/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO Channel 10/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO Channel 11/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO Channel 09/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO Channel 01/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO Channel 12/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO Channel 10/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO Channel 01/0 : 1[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO Channel 03/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO Channel 13/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO Channel 11/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO Channel 01/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO Channel 01/0 : 7[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO Channel 05/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO Channel 03/0 : 1[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO Channel 15/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO Channel 12/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO Channel 01/0 : 5[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO Channel 03/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO Channel 01/0 : 3[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO Channel 03/0 : 7[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO Channel 07/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO Channel 05/0 : 1[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO Channel 13/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO Channel 03/0 : 5[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO Channel 05/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO Channel 05/0 : 7[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO Channel 09/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO Channel 03/0 : 3[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO Channel 14/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO Channel 07/0 : 1[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO Channel 07/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO Channel 05/0 : 5[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:94348:94506 [6] NCCL INFO Channel 00/0 : 38[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO Channel 07/0 : 7[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO Channel 11/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:94348:94506 [6] NCCL INFO Channel 02/0 : 38[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO Channel 15/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO Channel 05/0 : 3[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO Channel 11/0 : 1[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO Channel 09/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO Channel 07/0 : 5[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO Channel 09/0 : 7[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:94348:94506 [6] NCCL INFO Channel 04/0 : 38[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO Channel 13/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO Channel 11/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO Channel 13/0 : 1[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO Channel 07/0 : 3[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO Channel 09/0 : 5[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:94348:94506 [6] NCCL INFO Channel 08/0 : 38[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO Channel 11/0 : 7[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO Channel 15/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO Channel 13/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:94348:94506 [6] NCCL INFO Channel 10/0 : 38[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO Channel 15/0 : 1[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO Channel 11/0 : 5[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO Channel 09/0 : 3[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:94342:94504 [0] NCCL INFO Channel 02/0 : 32[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO Channel 00/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO Channel 13/0 : 7[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:94348:94506 [6] NCCL INFO Channel 12/0 : 38[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:94342:94504 [0] NCCL INFO Channel 04/0 : 32[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO Channel 15/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO Channel 00/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO Channel 15/0 : 5[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:94348:94506 [6] NCCL INFO Channel 14/0 : 38[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO Channel 13/0 : 3[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:94342:94504 [0] NCCL INFO Channel 06/0 : 32[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO Channel 01/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO Channel 00/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO Channel 00/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO Channel 02/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:94348:94506 [6] NCCL INFO Channel 00/0 : 6[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO Channel 00/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO Channel 15/0 : 3[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:94342:94504 [0] NCCL INFO Channel 08/0 : 32[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO Channel 02/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:94348:94506 [6] NCCL INFO Channel 02/0 : 6[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO Channel 01/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO Channel 01/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:94342:94504 [0] NCCL INFO Channel 10/0 : 32[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO Channel 03/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO Channel 01/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:94348:94506 [6] NCCL INFO Channel 04/0 : 6[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO Channel 03/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO Channel 00/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:94342:94504 [0] NCCL INFO Channel 12/0 : 32[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:94348:94506 [6] NCCL INFO Channel 08/0 : 6[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO Channel 03/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO Channel 02/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:94342:94504 [0] NCCL INFO Channel 14/0 : 32[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO Channel 04/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO Channel 02/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO Channel 05/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO Channel 01/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO Channel 04/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:94348:94506 [6] NCCL INFO Channel 10/0 : 6[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO Channel 03/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:94342:94504 [0] NCCL INFO Channel 02/0 : 0[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO Channel 03/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO Channel 05/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO Channel 06/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:94348:94506 [6] NCCL INFO Channel 12/0 : 6[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO Channel 02/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO Channel 05/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO Channel 04/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:94342:94504 [0] NCCL INFO Channel 04/0 : 0[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO Channel 04/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO Channel 06/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:94348:94506 [6] NCCL INFO Channel 14/0 : 6[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:94342:94504 [0] NCCL INFO Channel 06/0 : 0[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO Channel 07/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO Channel 06/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO Channel 05/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO Channel 04/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO Channel 06/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO Channel 07/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO Channel 08/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:94342:94504 [0] NCCL INFO Channel 08/0 : 0[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO Channel 07/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO Channel 05/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO Channel 06/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO Channel 07/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:94342:94504 [0] NCCL INFO Channel 10/0 : 0[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO Channel 08/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO Channel 09/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO Channel 08/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:94342:94504 [0] NCCL INFO Channel 12/0 : 0[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:94342:94504 [0] NCCL INFO Channel 14/0 : 0[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:94346:94511 [4] NCCL INFO Channel 00/0 : 36[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:94346:94511 [4] NCCL INFO Channel 02/0 : 36[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:94346:94511 [4] NCCL INFO Channel 06/0 : 36[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:94343:94508 [1] NCCL INFO Channel 00/0 : 33[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:94347:94507 [5] NCCL INFO Channel 00/0 : 37[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:94349:94505 [7] NCCL INFO Channel 00/0 : 39[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:94344:94510 [2] NCCL INFO Channel 00/0 : 34[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:94346:94511 [4] NCCL INFO Channel 08/0 : 36[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:94343:94508 [1] NCCL INFO Channel 02/0 : 33[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:94347:94507 [5] NCCL INFO Channel 02/0 : 37[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:94349:94505 [7] NCCL INFO Channel 02/0 : 39[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:94344:94510 [2] NCCL INFO Channel 04/0 : 34[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:94346:94511 [4] NCCL INFO Channel 10/0 : 36[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:94343:94508 [1] NCCL INFO Channel 04/0 : 33[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:94345:94509 [3] NCCL INFO Channel 00/0 : 35[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:94347:94507 [5] NCCL INFO Channel 04/0 : 37[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:94349:94505 [7] NCCL INFO Channel 04/0 : 39[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:94346:94511 [4] NCCL INFO Channel 12/0 : 36[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:94344:94510 [2] NCCL INFO Channel 06/0 : 34[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:94343:94508 [1] NCCL INFO Channel 06/0 : 33[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:94345:94509 [3] NCCL INFO Channel 02/0 : 35[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:94349:94505 [7] NCCL INFO Channel 06/0 : 39[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:94347:94507 [5] NCCL INFO Channel 06/0 : 37[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:94346:94511 [4] NCCL INFO Channel 14/0 : 36[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:94343:94508 [1] NCCL INFO Channel 08/0 : 33[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:94344:94510 [2] NCCL INFO Channel 08/0 : 34[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:94345:94509 [3] NCCL INFO Channel 04/0 : 35[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:94347:94507 [5] NCCL INFO Channel 08/0 : 37[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:94349:94505 [7] NCCL INFO Channel 08/0 : 39[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:94346:94511 [4] NCCL INFO Channel 00/0 : 4[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:94343:94508 [1] NCCL INFO Channel 10/0 : 33[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:94344:94510 [2] NCCL INFO Channel 10/0 : 34[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:94345:94509 [3] NCCL INFO Channel 06/0 : 35[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:94347:94507 [5] NCCL INFO Channel 10/0 : 37[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:94349:94505 [7] NCCL INFO Channel 10/0 : 39[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:94346:94511 [4] NCCL INFO Channel 02/0 : 4[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:94343:94508 [1] NCCL INFO Channel 12/0 : 33[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:94344:94510 [2] NCCL INFO Channel 12/0 : 34[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:94345:94509 [3] NCCL INFO Channel 08/0 : 35[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:94347:94507 [5] NCCL INFO Channel 12/0 : 37[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:94346:94511 [4] NCCL INFO Channel 06/0 : 4[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:94349:94505 [7] NCCL INFO Channel 12/0 : 39[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:94343:94508 [1] NCCL INFO Channel 14/0 : 33[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:94344:94510 [2] NCCL INFO Channel 14/0 : 34[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:94345:94509 [3] NCCL INFO Channel 10/0 : 35[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:94347:94507 [5] NCCL INFO Channel 14/0 : 37[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:94349:94505 [7] NCCL INFO Channel 14/0 : 39[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:94346:94511 [4] NCCL INFO Channel 08/0 : 4[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:94343:94508 [1] NCCL INFO Channel 00/0 : 1[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:94344:94510 [2] NCCL INFO Channel 00/0 : 2[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:94347:94507 [5] NCCL INFO Channel 00/0 : 5[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:94345:94509 [3] NCCL INFO Channel 12/0 : 35[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:94349:94505 [7] NCCL INFO Channel 00/0 : 7[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:94346:94511 [4] NCCL INFO Channel 10/0 : 4[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:94343:94508 [1] NCCL INFO Channel 02/0 : 1[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:94344:94510 [2] NCCL INFO Channel 04/0 : 2[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:94347:94507 [5] NCCL INFO Channel 02/0 : 5[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:94345:94509 [3] NCCL INFO Channel 14/0 : 35[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:94349:94505 [7] NCCL INFO Channel 02/0 : 7[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:94346:94511 [4] NCCL INFO Channel 12/0 : 4[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:94343:94508 [1] NCCL INFO Channel 04/0 : 1[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:94344:94510 [2] NCCL INFO Channel 06/0 : 2[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:94347:94507 [5] NCCL INFO Channel 04/0 : 5[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:94345:94509 [3] NCCL INFO Channel 00/0 : 3[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:94349:94505 [7] NCCL INFO Channel 04/0 : 7[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:94346:94511 [4] NCCL INFO Channel 14/0 : 4[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:94343:94508 [1] NCCL INFO Channel 06/0 : 1[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:94344:94510 [2] NCCL INFO Channel 08/0 : 2[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:94347:94507 [5] NCCL INFO Channel 06/0 : 5[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:94345:94509 [3] NCCL INFO Channel 02/0 : 3[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:94349:94505 [7] NCCL INFO Channel 06/0 : 7[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:94343:94508 [1] NCCL INFO Channel 08/0 : 1[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:94344:94510 [2] NCCL INFO Channel 10/0 : 2[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:94347:94507 [5] NCCL INFO Channel 08/0 : 5[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:94349:94505 [7] NCCL INFO Channel 08/0 : 7[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:94345:94509 [3] NCCL INFO Channel 04/0 : 3[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:94343:94508 [1] NCCL INFO Channel 10/0 : 1[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:94344:94510 [2] NCCL INFO Channel 12/0 : 2[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:94347:94507 [5] NCCL INFO Channel 10/0 : 5[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:94349:94505 [7] NCCL INFO Channel 10/0 : 7[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:94345:94509 [3] NCCL INFO Channel 06/0 : 3[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:94343:94508 [1] NCCL INFO Channel 12/0 : 1[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:94344:94510 [2] NCCL INFO Channel 14/0 : 2[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:94347:94507 [5] NCCL INFO Channel 12/0 : 5[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:94349:94505 [7] NCCL INFO Channel 12/0 : 7[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:94345:94509 [3] NCCL INFO Channel 08/0 : 3[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:94343:94508 [1] NCCL INFO Channel 14/0 : 1[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:94347:94507 [5] NCCL INFO Channel 14/0 : 5[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:94349:94505 [7] NCCL INFO Channel 14/0 : 7[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:94345:94509 [3] NCCL INFO Channel 10/0 : 3[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:94345:94509 [3] NCCL INFO Channel 12/0 : 3[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:94345:94509 [3] NCCL INFO Channel 14/0 : 3[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926598:926761 [4] NCCL INFO Channel 00/0 : 44[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926598:926761 [4] NCCL INFO Channel 01/0 : 44[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926600:926757 [6] NCCL INFO Channel 00/0 : 46[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926598:926761 [4] NCCL INFO Channel 02/0 : 44[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926600:926757 [6] NCCL INFO Channel 01/0 : 46[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926598:926761 [4] NCCL INFO Channel 03/0 : 44[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926600:926757 [6] NCCL INFO Channel 02/0 : 46[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926598:926761 [4] NCCL INFO Channel 05/0 : 44[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926600:926757 [6] NCCL INFO Channel 03/0 : 46[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926598:926761 [4] NCCL INFO Channel 06/0 : 44[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926598:926761 [4] NCCL INFO Channel 07/0 : 44[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926600:926757 [6] NCCL INFO Channel 04/0 : 46[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926598:926761 [4] NCCL INFO Channel 08/0 : 44[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926600:926757 [6] NCCL INFO Channel 05/0 : 46[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926598:926761 [4] NCCL INFO Channel 09/0 : 44[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926600:926757 [6] NCCL INFO Channel 07/0 : 46[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926598:926761 [4] NCCL INFO Channel 10/0 : 44[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926600:926757 [6] NCCL INFO Channel 08/0 : 46[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926598:926761 [4] NCCL INFO Channel 11/0 : 44[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926600:926757 [6] NCCL INFO Channel 09/0 : 46[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926598:926761 [4] NCCL INFO Channel 13/0 : 44[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926600:926757 [6] NCCL INFO Channel 10/0 : 46[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926598:926761 [4] NCCL INFO Channel 14/0 : 44[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926600:926757 [6] NCCL INFO Channel 11/0 : 46[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926598:926761 [4] NCCL INFO Channel 15/0 : 44[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926600:926757 [6] NCCL INFO Channel 12/0 : 46[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926598:926761 [4] NCCL INFO Channel 00/0 : 52[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926600:926757 [6] NCCL INFO Channel 13/0 : 46[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926598:926761 [4] NCCL INFO Channel 02/0 : 52[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926600:926757 [6] NCCL INFO Channel 15/0 : 46[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926598:926761 [4] NCCL INFO Channel 06/0 : 52[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926600:926757 [6] NCCL INFO Channel 00/0 : 54[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926598:926761 [4] NCCL INFO Channel 08/0 : 52[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926600:926757 [6] NCCL INFO Channel 02/0 : 54[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926598:926761 [4] NCCL INFO Channel 10/0 : 52[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926600:926757 [6] NCCL INFO Channel 04/0 : 54[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926598:926761 [4] NCCL INFO Channel 12/0 : 52[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926600:926757 [6] NCCL INFO Channel 08/0 : 54[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926595:926760 [1] NCCL INFO Channel 00/0 : 41[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926598:926761 [4] NCCL INFO Channel 14/0 : 52[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926600:926757 [6] NCCL INFO Channel 10/0 : 54[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926601:926759 [7] NCCL INFO Channel 00/0 : 47[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926595:926760 [1] NCCL INFO Channel 02/0 : 41[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926600:926757 [6] NCCL INFO Channel 12/0 : 54[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926599:926755 [5] NCCL INFO Channel 00/0 : 45[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926595:926760 [1] NCCL INFO Channel 03/0 : 41[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926600:926757 [6] NCCL INFO Channel 14/0 : 54[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926601:926759 [7] NCCL INFO Channel 01/0 : 47[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926599:926755 [5] NCCL INFO Channel 01/0 : 45[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926597:926758 [3] NCCL INFO Channel 00/0 : 43[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926595:926760 [1] NCCL INFO Channel 04/0 : 41[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926594:926762 [0] NCCL INFO Channel 01/0 : 40[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926601:926759 [7] NCCL INFO Channel 02/0 : 47[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926596:926756 [2] NCCL INFO Channel 00/0 : 42[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926599:926755 [5] NCCL INFO Channel 02/0 : 45[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926595:926760 [1] NCCL INFO Channel 05/0 : 41[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926597:926758 [3] NCCL INFO Channel 01/0 : 43[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926598:926761 [4] NCCL INFO Channel 00/0 : 36[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-7:811252:811408 [4] NCCL INFO Channel 01/0 : 28[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926601:926759 [7] NCCL INFO Channel 03/0 : 47[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-7:811252:811408 [4] NCCL INFO Channel 03/0 : 28[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926596:926756 [2] NCCL INFO Channel 01/0 : 42[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926594:926762 [0] NCCL INFO Channel 02/0 : 40[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926595:926760 [1] NCCL INFO Channel 06/0 : 41[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926599:926755 [5] NCCL INFO Channel 03/0 : 45[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926598:926761 [4] NCCL INFO Channel 02/0 : 36[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926597:926758 [3] NCCL INFO Channel 02/0 : 43[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926601:926759 [7] NCCL INFO Channel 04/0 : 47[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-7:811252:811408 [4] NCCL INFO Channel 05/0 : 28[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926596:926756 [2] NCCL INFO Channel 03/0 : 42[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-7:811252:811408 [4] NCCL INFO Channel 07/0 : 28[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926594:926762 [0] NCCL INFO Channel 03/0 : 40[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926595:926760 [1] NCCL INFO Channel 07/0 : 41[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926601:926759 [7] NCCL INFO Channel 05/0 : 47[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926598:926761 [4] NCCL INFO Channel 06/0 : 36[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926599:926755 [5] NCCL INFO Channel 04/0 : 45[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926596:926756 [2] NCCL INFO Channel 04/0 : 42[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926597:926758 [3] NCCL INFO Channel 04/0 : 43[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926600:926757 [6] NCCL INFO Channel 00/0 : 38[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926594:926762 [0] NCCL INFO Channel 04/0 : 40[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-7:811254:811414 [6] NCCL INFO Channel 01/0 : 30[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-7:811252:811408 [4] NCCL INFO Channel 09/0 : 28[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926595:926760 [1] NCCL INFO Channel 08/0 : 41[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926601:926759 [7] NCCL INFO Channel 06/0 : 47[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926598:926761 [4] NCCL INFO Channel 08/0 : 36[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-7:811252:811408 [4] NCCL INFO Channel 11/0 : 28[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926596:926756 [2] NCCL INFO Channel 05/0 : 42[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-7:811254:811414 [6] NCCL INFO Channel 03/0 : 30[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926599:926755 [5] NCCL INFO Channel 06/0 : 45[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926595:926760 [1] NCCL INFO Channel 10/0 : 41[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926600:926757 [6] NCCL INFO Channel 02/0 : 38[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926594:926762 [0] NCCL INFO Channel 05/0 : 40[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926597:926758 [3] NCCL INFO Channel 05/0 : 43[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-7:811252:811408 [4] NCCL INFO Channel 13/0 : 28[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926601:926759 [7] NCCL INFO Channel 08/0 : 47[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926598:926761 [4] NCCL INFO Channel 10/0 : 36[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-7:811254:811414 [6] NCCL INFO Channel 05/0 : 30[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926596:926756 [2] NCCL INFO Channel 06/0 : 42[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926599:926755 [5] NCCL INFO Channel 07/0 : 45[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-7:811252:811408 [4] NCCL INFO Channel 15/0 : 28[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-7:811254:811414 [6] NCCL INFO Channel 07/0 : 30[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926595:926760 [1] NCCL INFO Channel 11/0 : 41[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-7:811252:811408 [4] NCCL INFO Channel 01/0 : 60[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-7:811254:811414 [6] NCCL INFO Channel 09/0 : 30[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926594:926762 [0] NCCL INFO Channel 06/0 : 40[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926600:926757 [6] NCCL INFO Channel 04/0 : 38[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926601:926759 [7] NCCL INFO Channel 09/0 : 47[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926598:926761 [4] NCCL INFO Channel 12/0 : 36[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926597:926758 [3] NCCL INFO Channel 06/0 : 43[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926596:926756 [2] NCCL INFO Channel 07/0 : 42[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926599:926755 [5] NCCL INFO Channel 08/0 : 45[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-7:811252:811408 [4] NCCL INFO Channel 03/0 : 60[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-7:811254:811414 [6] NCCL INFO Channel 11/0 : 30[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926595:926760 [1] NCCL INFO Channel 12/0 : 41[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-7:811252:811408 [4] NCCL INFO Channel 05/0 : 60[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-7:811254:811414 [6] NCCL INFO Channel 13/0 : 30[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926600:926757 [6] NCCL INFO Channel 08/0 : 38[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926601:926759 [7] NCCL INFO Channel 10/0 : 47[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926598:926761 [4] NCCL INFO Channel 14/0 : 36[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926594:926762 [0] NCCL INFO Channel 07/0 : 40[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926596:926756 [2] NCCL INFO Channel 08/0 : 42[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926597:926758 [3] NCCL INFO Channel 07/0 : 43[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926595:926760 [1] NCCL INFO Channel 13/0 : 41[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-7:811252:811408 [4] NCCL INFO Channel 07/0 : 60[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-7:811254:811414 [6] NCCL INFO Channel 15/0 : 30[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926599:926755 [5] NCCL INFO Channel 09/0 : 45[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926601:926759 [7] NCCL INFO Channel 11/0 : 47[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926600:926757 [6] NCCL INFO Channel 10/0 : 38[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926594:926762 [0] NCCL INFO Channel 09/0 : 40[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-7:811252:811408 [4] NCCL INFO Channel 09/0 : 60[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-7:811254:811414 [6] NCCL INFO Channel 01/0 : 62[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926596:926756 [2] NCCL INFO Channel 09/0 : 42[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926595:926760 [1] NCCL INFO Channel 14/0 : 41[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-7:811252:811408 [4] NCCL INFO Channel 11/0 : 60[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926599:926755 [5] NCCL INFO Channel 10/0 : 45[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926597:926758 [3] NCCL INFO Channel 08/0 : 43[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-7:811254:811414 [6] NCCL INFO Channel 03/0 : 62[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926601:926759 [7] NCCL INFO Channel 12/0 : 47[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926600:926757 [6] NCCL INFO Channel 12/0 : 38[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926594:926762 [0] NCCL INFO Channel 10/0 : 40[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-7:811252:811408 [4] NCCL INFO Channel 13/0 : 60[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926596:926756 [2] NCCL INFO Channel 11/0 : 42[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926595:926760 [1] NCCL INFO Channel 15/0 : 41[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926599:926755 [5] NCCL INFO Channel 11/0 : 45[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-7:811254:811414 [6] NCCL INFO Channel 05/0 : 62[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-7:811252:811408 [4] NCCL INFO Channel 15/0 : 60[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926601:926759 [7] NCCL INFO Channel 13/0 : 47[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-7:811254:811414 [6] NCCL INFO Channel 07/0 : 62[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926597:926758 [3] NCCL INFO Channel 09/0 : 43[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926600:926757 [6] NCCL INFO Channel 14/0 : 38[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926594:926762 [0] NCCL INFO Channel 11/0 : 40[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926595:926760 [1] NCCL INFO Channel 00/0 : 49[1] -> 57[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926596:926756 [2] NCCL INFO Channel 12/0 : 42[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926599:926755 [5] NCCL INFO Channel 12/0 : 45[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926601:926759 [7] NCCL INFO Channel 14/0 : 47[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-7:811254:811414 [6] NCCL INFO Channel 09/0 : 62[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926595:926760 [1] NCCL INFO Channel 02/0 : 49[1] -> 57[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926596:926756 [2] NCCL INFO Channel 13/0 : 42[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-7:811254:811414 [6] NCCL INFO Channel 11/0 : 62[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926599:926755 [5] NCCL INFO Channel 14/0 : 45[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926594:926762 [0] NCCL INFO Channel 12/0 : 40[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926601:926759 [7] NCCL INFO Channel 00/0 : 55[7] -> 63[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926597:926758 [3] NCCL INFO Channel 10/0 : 43[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-7:811254:811414 [6] NCCL INFO Channel 13/0 : 62[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926595:926760 [1] NCCL INFO Channel 04/0 : 49[1] -> 57[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926596:926756 [2] NCCL INFO Channel 14/0 : 42[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-7:811254:811414 [6] NCCL INFO Channel 15/0 : 62[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926599:926755 [5] NCCL INFO Channel 15/0 : 45[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926601:926759 [7] NCCL INFO Channel 02/0 : 55[7] -> 63[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926594:926762 [0] NCCL INFO Channel 13/0 : 40[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926597:926758 [3] NCCL INFO Channel 12/0 : 43[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926595:926760 [1] NCCL INFO Channel 06/0 : 49[1] -> 57[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926596:926756 [2] NCCL INFO Channel 15/0 : 42[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926599:926755 [5] NCCL INFO Channel 00/0 : 53[5] -> 61[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926601:926759 [7] NCCL INFO Channel 04/0 : 55[7] -> 63[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926595:926760 [1] NCCL INFO Channel 08/0 : 49[1] -> 57[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926594:926762 [0] NCCL INFO Channel 14/0 : 40[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926596:926756 [2] NCCL INFO Channel 00/0 : 50[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926599:926755 [5] NCCL INFO Channel 02/0 : 53[5] -> 61[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926597:926758 [3] NCCL INFO Channel 13/0 : 43[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926601:926759 [7] NCCL INFO Channel 06/0 : 55[7] -> 63[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926595:926760 [1] NCCL INFO Channel 10/0 : 49[1] -> 57[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926596:926756 [2] NCCL INFO Channel 04/0 : 50[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926594:926762 [0] NCCL INFO Channel 15/0 : 40[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926599:926755 [5] NCCL INFO Channel 04/0 : 53[5] -> 61[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926601:926759 [7] NCCL INFO Channel 08/0 : 55[7] -> 63[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926597:926758 [3] NCCL INFO Channel 14/0 : 43[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926595:926760 [1] NCCL INFO Channel 12/0 : 49[1] -> 57[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926596:926756 [2] NCCL INFO Channel 06/0 : 50[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926594:926762 [0] NCCL INFO Channel 02/0 : 48[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926599:926755 [5] NCCL INFO Channel 06/0 : 53[5] -> 61[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926601:926759 [7] NCCL INFO Channel 10/0 : 55[7] -> 63[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926595:926760 [1] NCCL INFO Channel 14/0 : 49[1] -> 57[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926597:926758 [3] NCCL INFO Channel 15/0 : 43[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-7:811251:811409 [3] NCCL INFO Channel 01/0 : 27[3] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-7:811251:811409 [3] NCCL INFO Channel 03/0 : 27[3] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-7:811253:811410 [5] NCCL INFO Channel 01/0 : 29[5] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-7:811251:811409 [3] NCCL INFO Channel 05/0 : 27[3] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-7:811253:811410 [5] NCCL INFO Channel 03/0 : 29[5] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-7:811251:811409 [3] NCCL INFO Channel 07/0 : 27[3] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-7:811253:811410 [5] NCCL INFO Channel 05/0 : 29[5] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-7:811251:811409 [3] NCCL INFO Channel 09/0 : 27[3] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-7:811251:811409 [3] NCCL INFO Channel 13/0 : 27[3] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-7:811253:811410 [5] NCCL INFO Channel 07/0 : 29[5] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-7:811251:811409 [3] NCCL INFO Channel 15/0 : 27[3] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-7:811253:811410 [5] NCCL INFO Channel 09/0 : 29[5] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-7:811251:811409 [3] NCCL INFO Channel 01/0 : 59[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-7:811253:811410 [5] NCCL INFO Channel 11/0 : 29[5] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-7:811251:811409 [3] NCCL INFO Channel 03/0 : 59[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-7:811253:811410 [5] NCCL INFO Channel 15/0 : 29[5] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-7:811251:811409 [3] NCCL INFO Channel 05/0 : 59[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-7:811253:811410 [5] NCCL INFO Channel 01/0 : 61[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-7:811251:811409 [3] NCCL INFO Channel 07/0 : 59[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-7:811253:811410 [5] NCCL INFO Channel 03/0 : 61[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-7:811251:811409 [3] NCCL INFO Channel 09/0 : 59[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-7:811253:811410 [5] NCCL INFO Channel 05/0 : 61[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-7:811251:811409 [3] NCCL INFO Channel 13/0 : 59[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-7:811253:811410 [5] NCCL INFO Channel 07/0 : 61[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-7:811251:811409 [3] NCCL INFO Channel 15/0 : 59[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-7:811253:811410 [5] NCCL INFO Channel 09/0 : 61[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-7:811253:811410 [5] NCCL INFO Channel 11/0 : 61[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-7:811253:811410 [5] NCCL INFO Channel 15/0 : 61[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-7:811250:811412 [2] NCCL INFO Channel 01/0 : 26[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-7:811249:811413 [1] NCCL INFO Channel 01/0 : 25[1] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-7:811250:811412 [2] NCCL INFO Channel 03/0 : 26[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-7:811255:811411 [7] NCCL INFO Channel 01/0 : 31[7] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-7:811249:811413 [1] NCCL INFO Channel 03/0 : 25[1] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-7:811250:811412 [2] NCCL INFO Channel 05/0 : 26[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-7:811255:811411 [7] NCCL INFO Channel 03/0 : 31[7] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-7:811249:811413 [1] NCCL INFO Channel 05/0 : 25[1] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-7:811255:811411 [7] NCCL INFO Channel 05/0 : 31[7] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-7:811250:811412 [2] NCCL INFO Channel 07/0 : 26[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-7:811249:811413 [1] NCCL INFO Channel 07/0 : 25[1] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-7:811255:811411 [7] NCCL INFO Channel 07/0 : 31[7] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-7:811250:811412 [2] NCCL INFO Channel 09/0 : 26[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-7:811248:811415 [0] NCCL INFO Channel 01/0 : 24[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-7:811249:811413 [1] NCCL INFO Channel 11/0 : 25[1] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-7:811255:811411 [7] NCCL INFO Channel 09/0 : 31[7] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-7:811250:811412 [2] NCCL INFO Channel 11/0 : 26[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-7:811249:811413 [1] NCCL INFO Channel 13/0 : 25[1] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-7:811248:811415 [0] NCCL INFO Channel 03/0 : 24[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-7:811255:811411 [7] NCCL INFO Channel 11/0 : 31[7] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-7:811250:811412 [2] NCCL INFO Channel 13/0 : 26[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-7:811249:811413 [1] NCCL INFO Channel 15/0 : 25[1] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-7:811248:811415 [0] NCCL INFO Channel 05/0 : 24[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-7:811250:811412 [2] NCCL INFO Channel 15/0 : 26[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-7:811255:811411 [7] NCCL INFO Channel 13/0 : 31[7] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-7:811249:811413 [1] NCCL INFO Channel 01/0 : 57[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-7:811248:811415 [0] NCCL INFO Channel 07/0 : 24[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-7:811255:811411 [7] NCCL INFO Channel 01/0 : 63[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-7:811250:811412 [2] NCCL INFO Channel 01/0 : 58[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-7:811249:811413 [1] NCCL INFO Channel 03/0 : 57[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-7:811248:811415 [0] NCCL INFO Channel 09/0 : 24[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-7:811255:811411 [7] NCCL INFO Channel 03/0 : 63[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-7:811250:811412 [2] NCCL INFO Channel 03/0 : 58[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-7:811249:811413 [1] NCCL INFO Channel 05/0 : 57[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-7:811248:811415 [0] NCCL INFO Channel 11/0 : 24[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-7:811255:811411 [7] NCCL INFO Channel 05/0 : 63[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-7:811250:811412 [2] NCCL INFO Channel 05/0 : 58[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-7:811249:811413 [1] NCCL INFO Channel 07/0 : 57[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-7:811248:811415 [0] NCCL INFO Channel 13/0 : 24[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-7:811250:811412 [2] NCCL INFO Channel 07/0 : 58[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-7:811255:811411 [7] NCCL INFO Channel 07/0 : 63[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-7:811249:811413 [1] NCCL INFO Channel 11/0 : 57[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-7:811248:811415 [0] NCCL INFO Channel 15/0 : 24[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-7:811250:811412 [2] NCCL INFO Channel 09/0 : 58[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-7:811255:811411 [7] NCCL INFO Channel 09/0 : 63[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-7:811249:811413 [1] NCCL INFO Channel 13/0 : 57[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-7:811248:811415 [0] NCCL INFO Channel 01/0 : 56[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-7:811250:811412 [2] NCCL INFO Channel 11/0 : 58[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-7:811255:811411 [7] NCCL INFO Channel 11/0 : 63[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-7:811248:811415 [0] NCCL INFO Channel 03/0 : 56[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-7:811249:811413 [1] NCCL INFO Channel 15/0 : 57[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-7:811250:811412 [2] NCCL INFO Channel 13/0 : 58[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-7:811255:811411 [7] NCCL INFO Channel 13/0 : 63[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-7:811248:811415 [0] NCCL INFO Channel 05/0 : 56[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-7:811250:811412 [2] NCCL INFO Channel 15/0 : 58[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-7:811248:811415 [0] NCCL INFO Channel 07/0 : 56[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-7:811248:811415 [0] NCCL INFO Channel 09/0 : 56[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-7:811248:811415 [0] NCCL INFO Channel 11/0 : 56[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-7:811248:811415 [0] NCCL INFO Channel 13/0 : 56[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-7:811248:811415 [0] NCCL INFO Channel 15/0 : 56[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880681:880845 [2] NCCL INFO Connected all trees
+ip-26-0-172-142:880681:880845 [2] NCCL INFO NVLS comm 0x8e71310 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-252:688205:688365 [5] NCCL INFO Channel 01/0 : 37[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688207:688366 [7] NCCL INFO Channel 01/0 : 39[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688205:688365 [5] NCCL INFO Channel 03/0 : 37[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688207:688366 [7] NCCL INFO Channel 03/0 : 39[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688205:688365 [5] NCCL INFO Channel 05/0 : 37[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688205:688365 [5] NCCL INFO Channel 07/0 : 37[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688207:688366 [7] NCCL INFO Channel 05/0 : 39[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688205:688365 [5] NCCL INFO Channel 09/0 : 37[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688207:688366 [7] NCCL INFO Channel 07/0 : 39[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688205:688365 [5] NCCL INFO Channel 11/0 : 37[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688207:688366 [7] NCCL INFO Channel 09/0 : 39[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688205:688365 [5] NCCL INFO Channel 15/0 : 37[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688207:688366 [7] NCCL INFO Channel 11/0 : 39[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688207:688366 [7] NCCL INFO Channel 13/0 : 39[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688205:688365 [5] NCCL INFO Channel 00/0 : 21[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688205:688365 [5] NCCL INFO Channel 02/0 : 21[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942616:942777 [5] NCCL INFO Channel 01/0 : 29[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688207:688366 [7] NCCL INFO Channel 00/0 : 23[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942618:942778 [7] NCCL INFO Channel 01/0 : 31[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942616:942777 [5] NCCL INFO Channel 03/0 : 29[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688205:688365 [5] NCCL INFO Channel 04/0 : 21[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688207:688366 [7] NCCL INFO Channel 02/0 : 23[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942618:942778 [7] NCCL INFO Channel 03/0 : 31[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688205:688365 [5] NCCL INFO Channel 06/0 : 21[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942616:942777 [5] NCCL INFO Channel 05/0 : 29[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688207:688366 [7] NCCL INFO Channel 04/0 : 23[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688205:688365 [5] NCCL INFO Channel 08/0 : 21[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942618:942778 [7] NCCL INFO Channel 05/0 : 31[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942616:942777 [5] NCCL INFO Channel 07/0 : 29[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688207:688366 [7] NCCL INFO Channel 06/0 : 23[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688205:688365 [5] NCCL INFO Channel 10/0 : 21[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942618:942778 [7] NCCL INFO Channel 07/0 : 31[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688207:688366 [7] NCCL INFO Channel 08/0 : 23[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688205:688365 [5] NCCL INFO Channel 12/0 : 21[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942616:942777 [5] NCCL INFO Channel 09/0 : 29[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688207:688366 [7] NCCL INFO Channel 10/0 : 23[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942618:942778 [7] NCCL INFO Channel 09/0 : 31[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942616:942777 [5] NCCL INFO Channel 11/0 : 29[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688205:688365 [5] NCCL INFO Channel 14/0 : 21[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688204:688367 [4] NCCL INFO Channel 01/0 : 36[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942618:942778 [7] NCCL INFO Channel 11/0 : 31[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942616:942777 [5] NCCL INFO Channel 15/0 : 29[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688207:688366 [7] NCCL INFO Channel 12/0 : 23[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942618:942778 [7] NCCL INFO Channel 13/0 : 31[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688203:688362 [3] NCCL INFO Channel 01/0 : 35[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688205:688365 [5] NCCL INFO Channel 00/0 : 37[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688206:688363 [6] NCCL INFO Channel 01/0 : 38[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688207:688366 [7] NCCL INFO Channel 14/0 : 23[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688204:688367 [4] NCCL INFO Channel 03/0 : 36[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688205:688365 [5] NCCL INFO Channel 02/0 : 37[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688203:688362 [3] NCCL INFO Channel 03/0 : 35[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688207:688366 [7] NCCL INFO Channel 00/0 : 39[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688204:688367 [4] NCCL INFO Channel 05/0 : 36[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688206:688363 [6] NCCL INFO Channel 03/0 : 38[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688205:688365 [5] NCCL INFO Channel 04/0 : 37[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688203:688362 [3] NCCL INFO Channel 05/0 : 35[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688200:688369 [0] NCCL INFO Channel 01/0 : 32[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688207:688366 [7] NCCL INFO Channel 02/0 : 39[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688204:688367 [4] NCCL INFO Channel 07/0 : 36[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688205:688365 [5] NCCL INFO Channel 06/0 : 37[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688206:688363 [6] NCCL INFO Channel 05/0 : 38[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688203:688362 [3] NCCL INFO Channel 07/0 : 35[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688207:688366 [7] NCCL INFO Channel 04/0 : 39[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688201:688364 [1] NCCL INFO Channel 01/0 : 33[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688204:688367 [4] NCCL INFO Channel 09/0 : 36[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688200:688369 [0] NCCL INFO Channel 03/0 : 32[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688202:688368 [2] NCCL INFO Channel 01/0 : 34[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688205:688365 [5] NCCL INFO Channel 08/0 : 37[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688207:688366 [7] NCCL INFO Channel 06/0 : 39[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688206:688363 [6] NCCL INFO Channel 07/0 : 38[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688203:688362 [3] NCCL INFO Channel 09/0 : 35[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688204:688367 [4] NCCL INFO Channel 11/0 : 36[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688205:688365 [5] NCCL INFO Channel 10/0 : 37[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688201:688364 [1] NCCL INFO Channel 03/0 : 33[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688207:688366 [7] NCCL INFO Channel 08/0 : 39[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688200:688369 [0] NCCL INFO Channel 05/0 : 32[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688202:688368 [2] NCCL INFO Channel 03/0 : 34[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688206:688363 [6] NCCL INFO Channel 09/0 : 38[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688203:688362 [3] NCCL INFO Channel 13/0 : 35[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688204:688367 [4] NCCL INFO Channel 13/0 : 36[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688205:688365 [5] NCCL INFO Channel 12/0 : 37[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688207:688366 [7] NCCL INFO Channel 10/0 : 39[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688201:688364 [1] NCCL INFO Channel 05/0 : 33[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688200:688369 [0] NCCL INFO Channel 07/0 : 32[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688202:688368 [2] NCCL INFO Channel 05/0 : 34[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688206:688363 [6] NCCL INFO Channel 11/0 : 38[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688203:688362 [3] NCCL INFO Channel 15/0 : 35[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688204:688367 [4] NCCL INFO Channel 15/0 : 36[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688205:688365 [5] NCCL INFO Channel 14/0 : 37[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688207:688366 [7] NCCL INFO Channel 12/0 : 39[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688201:688364 [1] NCCL INFO Channel 07/0 : 33[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688200:688369 [0] NCCL INFO Channel 09/0 : 32[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688206:688363 [6] NCCL INFO Channel 13/0 : 38[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688202:688368 [2] NCCL INFO Channel 07/0 : 34[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688207:688366 [7] NCCL INFO Channel 14/0 : 39[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688201:688364 [1] NCCL INFO Channel 11/0 : 33[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688200:688369 [0] NCCL INFO Channel 11/0 : 32[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688206:688363 [6] NCCL INFO Channel 15/0 : 38[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688202:688368 [2] NCCL INFO Channel 09/0 : 34[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688201:688364 [1] NCCL INFO Channel 13/0 : 33[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688203:688362 [3] NCCL INFO Channel 00/0 : 19[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688200:688369 [0] NCCL INFO Channel 13/0 : 32[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688204:688367 [4] NCCL INFO Channel 00/0 : 20[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688202:688368 [2] NCCL INFO Channel 11/0 : 34[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688201:688364 [1] NCCL INFO Channel 15/0 : 33[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942614:942773 [3] NCCL INFO Channel 01/0 : 27[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942615:942775 [4] NCCL INFO Channel 01/0 : 28[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688200:688369 [0] NCCL INFO Channel 15/0 : 32[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688203:688362 [3] NCCL INFO Channel 02/0 : 19[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688204:688367 [4] NCCL INFO Channel 02/0 : 20[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688202:688368 [2] NCCL INFO Channel 13/0 : 34[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688206:688363 [6] NCCL INFO Channel 00/0 : 22[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942614:942773 [3] NCCL INFO Channel 03/0 : 27[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942615:942775 [4] NCCL INFO Channel 03/0 : 28[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688204:688367 [4] NCCL INFO Channel 06/0 : 20[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688202:688368 [2] NCCL INFO Channel 15/0 : 34[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942617:942774 [6] NCCL INFO Channel 01/0 : 30[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688206:688363 [6] NCCL INFO Channel 02/0 : 22[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688203:688362 [3] NCCL INFO Channel 04/0 : 19[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942614:942773 [3] NCCL INFO Channel 05/0 : 27[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688201:688364 [1] NCCL INFO Channel 00/0 : 17[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942615:942775 [4] NCCL INFO Channel 05/0 : 28[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688204:688367 [4] NCCL INFO Channel 08/0 : 20[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942617:942774 [6] NCCL INFO Channel 03/0 : 30[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942612:942779 [1] NCCL INFO Channel 01/0 : 25[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688206:688363 [6] NCCL INFO Channel 04/0 : 22[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688200:688369 [0] NCCL INFO Channel 02/0 : 16[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688203:688362 [3] NCCL INFO Channel 06/0 : 19[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688201:688364 [1] NCCL INFO Channel 02/0 : 17[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942611:942772 [0] NCCL INFO Channel 01/0 : 24[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942615:942775 [4] NCCL INFO Channel 07/0 : 28[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942614:942773 [3] NCCL INFO Channel 07/0 : 27[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942617:942774 [6] NCCL INFO Channel 05/0 : 30[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688204:688367 [4] NCCL INFO Channel 10/0 : 20[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688206:688363 [6] NCCL INFO Channel 08/0 : 22[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926596:926756 [2] NCCL INFO Channel 08/0 : 50[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926594:926762 [0] NCCL INFO Channel 04/0 : 48[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688201:688364 [1] NCCL INFO Channel 04/0 : 17[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688203:688362 [3] NCCL INFO Channel 08/0 : 19[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926599:926755 [5] NCCL INFO Channel 08/0 : 53[5] -> 61[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926601:926759 [7] NCCL INFO Channel 12/0 : 55[7] -> 63[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942612:942779 [1] NCCL INFO Channel 03/0 : 25[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926597:926758 [3] NCCL INFO Channel 00/0 : 51[3] -> 59[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942611:942772 [0] NCCL INFO Channel 03/0 : 24[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926596:926756 [2] NCCL INFO Channel 10/0 : 50[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926594:926762 [0] NCCL INFO Channel 06/0 : 48[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926601:926759 [7] NCCL INFO Channel 14/0 : 55[7] -> 63[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688202:688368 [2] NCCL INFO Channel 00/0 : 18[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688200:688369 [0] NCCL INFO Channel 04/0 : 16[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926599:926755 [5] NCCL INFO Channel 10/0 : 53[5] -> 61[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688204:688367 [4] NCCL INFO Channel 12/0 : 20[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926597:926758 [3] NCCL INFO Channel 02/0 : 51[3] -> 59[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926596:926756 [2] NCCL INFO Channel 12/0 : 50[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688206:688363 [6] NCCL INFO Channel 10/0 : 22[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926594:926762 [0] NCCL INFO Channel 08/0 : 48[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942614:942773 [3] NCCL INFO Channel 09/0 : 27[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942615:942775 [4] NCCL INFO Channel 09/0 : 28[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942617:942774 [6] NCCL INFO Channel 07/0 : 30[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926599:926755 [5] NCCL INFO Channel 12/0 : 53[5] -> 61[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942612:942779 [1] NCCL INFO Channel 05/0 : 25[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926597:926758 [3] NCCL INFO Channel 04/0 : 51[3] -> 59[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926596:926756 [2] NCCL INFO Channel 14/0 : 50[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926594:926762 [0] NCCL INFO Channel 10/0 : 48[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926599:926755 [5] NCCL INFO Channel 14/0 : 53[5] -> 61[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926597:926758 [3] NCCL INFO Channel 06/0 : 51[3] -> 59[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688201:688364 [1] NCCL INFO Channel 06/0 : 17[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926594:926762 [0] NCCL INFO Channel 12/0 : 48[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688203:688362 [3] NCCL INFO Channel 10/0 : 19[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926595:926760 [1] NCCL INFO Channel 00/0 : 33[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926597:926758 [3] NCCL INFO Channel 08/0 : 51[3] -> 59[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688200:688369 [0] NCCL INFO Channel 06/0 : 16[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688202:688368 [2] NCCL INFO Channel 04/0 : 18[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942613:942776 [2] NCCL INFO Channel 01/0 : 26[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942611:942772 [0] NCCL INFO Channel 05/0 : 24[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926594:926762 [0] NCCL INFO Channel 14/0 : 48[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688204:688367 [4] NCCL INFO Channel 14/0 : 20[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942614:942773 [3] NCCL INFO Channel 13/0 : 27[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926595:926760 [1] NCCL INFO Channel 02/0 : 33[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926597:926758 [3] NCCL INFO Channel 10/0 : 51[3] -> 59[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688206:688363 [6] NCCL INFO Channel 12/0 : 22[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926595:926760 [1] NCCL INFO Channel 04/0 : 33[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688201:688364 [1] NCCL INFO Channel 08/0 : 17[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926597:926758 [3] NCCL INFO Channel 12/0 : 51[3] -> 59[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926595:926760 [1] NCCL INFO Channel 06/0 : 33[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926597:926758 [3] NCCL INFO Channel 14/0 : 51[3] -> 59[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926599:926755 [5] NCCL INFO Channel 00/0 : 37[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926595:926760 [1] NCCL INFO Channel 08/0 : 33[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942617:942774 [6] NCCL INFO Channel 09/0 : 30[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926594:926762 [0] NCCL INFO Channel 02/0 : 32[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926599:926755 [5] NCCL INFO Channel 02/0 : 37[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942615:942775 [4] NCCL INFO Channel 11/0 : 28[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942612:942779 [1] NCCL INFO Channel 07/0 : 25[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926595:926760 [1] NCCL INFO Channel 10/0 : 33[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926594:926762 [0] NCCL INFO Channel 04/0 : 32[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688203:688362 [3] NCCL INFO Channel 12/0 : 19[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926599:926755 [5] NCCL INFO Channel 04/0 : 37[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688200:688369 [0] NCCL INFO Channel 08/0 : 16[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926595:926760 [1] NCCL INFO Channel 12/0 : 33[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688202:688368 [2] NCCL INFO Channel 06/0 : 18[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926599:926755 [5] NCCL INFO Channel 06/0 : 37[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926594:926762 [0] NCCL INFO Channel 06/0 : 32[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926595:926760 [1] NCCL INFO Channel 14/0 : 33[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688204:688367 [4] NCCL INFO Channel 00/0 : 36[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688206:688363 [6] NCCL INFO Channel 14/0 : 22[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942611:942772 [0] NCCL INFO Channel 07/0 : 24[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942613:942776 [2] NCCL INFO Channel 03/0 : 26[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942614:942773 [3] NCCL INFO Channel 15/0 : 27[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926597:926758 [3] NCCL INFO Channel 00/0 : 35[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688201:688364 [1] NCCL INFO Channel 10/0 : 17[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942617:942774 [6] NCCL INFO Channel 11/0 : 30[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926599:926755 [5] NCCL INFO Channel 08/0 : 37[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688203:688362 [3] NCCL INFO Channel 14/0 : 19[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926594:926762 [0] NCCL INFO Channel 08/0 : 32[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926597:926758 [3] NCCL INFO Channel 02/0 : 35[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688200:688369 [0] NCCL INFO Channel 10/0 : 16[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926599:926755 [5] NCCL INFO Channel 10/0 : 37[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926597:926758 [3] NCCL INFO Channel 04/0 : 35[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926594:926762 [0] NCCL INFO Channel 10/0 : 32[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926599:926755 [5] NCCL INFO Channel 12/0 : 37[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942612:942779 [1] NCCL INFO Channel 11/0 : 25[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926594:926762 [0] NCCL INFO Channel 12/0 : 32[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926597:926758 [3] NCCL INFO Channel 06/0 : 35[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942615:942775 [4] NCCL INFO Channel 13/0 : 28[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926599:926755 [5] NCCL INFO Channel 14/0 : 37[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942611:942772 [0] NCCL INFO Channel 09/0 : 24[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942613:942776 [2] NCCL INFO Channel 05/0 : 26[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942617:942774 [6] NCCL INFO Channel 13/0 : 30[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926594:926762 [0] NCCL INFO Channel 14/0 : 32[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926597:926758 [3] NCCL INFO Channel 08/0 : 35[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688202:688368 [2] NCCL INFO Channel 08/0 : 18[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926597:926758 [3] NCCL INFO Channel 10/0 : 35[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688204:688367 [4] NCCL INFO Channel 02/0 : 36[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926597:926758 [3] NCCL INFO Channel 12/0 : 35[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688206:688363 [6] NCCL INFO Channel 00/0 : 38[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926597:926758 [3] NCCL INFO Channel 14/0 : 35[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688201:688364 [1] NCCL INFO Channel 12/0 : 17[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688203:688362 [3] NCCL INFO Channel 00/0 : 35[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688200:688369 [0] NCCL INFO Channel 12/0 : 16[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926601:926759 [7] NCCL INFO Channel 00/0 : 39[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688202:688368 [2] NCCL INFO Channel 10/0 : 18[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926601:926759 [7] NCCL INFO Channel 02/0 : 39[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688204:688367 [4] NCCL INFO Channel 06/0 : 36[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926596:926756 [2] NCCL INFO Channel 00/0 : 34[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926601:926759 [7] NCCL INFO Channel 04/0 : 39[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926596:926756 [2] NCCL INFO Channel 04/0 : 34[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926601:926759 [7] NCCL INFO Channel 06/0 : 39[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926596:926756 [2] NCCL INFO Channel 06/0 : 34[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926601:926759 [7] NCCL INFO Channel 08/0 : 39[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926596:926756 [2] NCCL INFO Channel 08/0 : 34[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926601:926759 [7] NCCL INFO Channel 10/0 : 39[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926596:926756 [2] NCCL INFO Channel 10/0 : 34[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688206:688363 [6] NCCL INFO Channel 02/0 : 38[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926601:926759 [7] NCCL INFO Channel 12/0 : 39[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688201:688364 [1] NCCL INFO Channel 14/0 : 17[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926596:926756 [2] NCCL INFO Channel 12/0 : 34[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688200:688369 [0] NCCL INFO Channel 14/0 : 16[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926601:926759 [7] NCCL INFO Channel 14/0 : 39[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688202:688368 [2] NCCL INFO Channel 12/0 : 18[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926596:926756 [2] NCCL INFO Channel 14/0 : 34[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688204:688367 [4] NCCL INFO Channel 08/0 : 36[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688206:688363 [6] NCCL INFO Channel 04/0 : 38[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926599:926755 [5] NCCL INFO Channel 00/0 : 53[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688203:688362 [3] NCCL INFO Channel 02/0 : 35[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926601:926759 [7] NCCL INFO Channel 00/0 : 55[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688201:688364 [1] NCCL INFO Channel 00/0 : 33[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688200:688369 [0] NCCL INFO Channel 02/0 : 32[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926599:926755 [5] NCCL INFO Channel 02/0 : 53[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926601:926759 [7] NCCL INFO Channel 02/0 : 55[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926599:926755 [5] NCCL INFO Channel 04/0 : 53[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926601:926759 [7] NCCL INFO Channel 04/0 : 55[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926599:926755 [5] NCCL INFO Channel 06/0 : 53[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926601:926759 [7] NCCL INFO Channel 06/0 : 55[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688202:688368 [2] NCCL INFO Channel 14/0 : 18[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688204:688367 [4] NCCL INFO Channel 10/0 : 36[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688206:688363 [6] NCCL INFO Channel 08/0 : 38[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688201:688364 [1] NCCL INFO Channel 02/0 : 33[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688200:688369 [0] NCCL INFO Channel 04/0 : 32[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688203:688362 [3] NCCL INFO Channel 04/0 : 35[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688202:688368 [2] NCCL INFO Channel 00/0 : 34[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688204:688367 [4] NCCL INFO Channel 12/0 : 36[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688206:688363 [6] NCCL INFO Channel 10/0 : 38[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926599:926755 [5] NCCL INFO Channel 08/0 : 53[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926601:926759 [7] NCCL INFO Channel 08/0 : 55[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926599:926755 [5] NCCL INFO Channel 10/0 : 53[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926601:926759 [7] NCCL INFO Channel 10/0 : 55[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926599:926755 [5] NCCL INFO Channel 12/0 : 53[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926601:926759 [7] NCCL INFO Channel 12/0 : 55[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688201:688364 [1] NCCL INFO Channel 04/0 : 33[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926599:926755 [5] NCCL INFO Channel 14/0 : 53[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688200:688369 [0] NCCL INFO Channel 06/0 : 32[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926601:926759 [7] NCCL INFO Channel 14/0 : 55[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688202:688368 [2] NCCL INFO Channel 04/0 : 34[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688203:688362 [3] NCCL INFO Channel 06/0 : 35[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688204:688367 [4] NCCL INFO Channel 14/0 : 36[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688206:688363 [6] NCCL INFO Channel 12/0 : 38[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688201:688364 [1] NCCL INFO Channel 06/0 : 33[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688200:688369 [0] NCCL INFO Channel 08/0 : 32[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688202:688368 [2] NCCL INFO Channel 06/0 : 34[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688203:688362 [3] NCCL INFO Channel 08/0 : 35[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688206:688363 [6] NCCL INFO Channel 14/0 : 38[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688201:688364 [1] NCCL INFO Channel 08/0 : 33[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688200:688369 [0] NCCL INFO Channel 10/0 : 32[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688202:688368 [2] NCCL INFO Channel 08/0 : 34[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688203:688362 [3] NCCL INFO Channel 10/0 : 35[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688201:688364 [1] NCCL INFO Channel 10/0 : 33[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688200:688369 [0] NCCL INFO Channel 12/0 : 32[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688202:688368 [2] NCCL INFO Channel 10/0 : 34[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688203:688362 [3] NCCL INFO Channel 12/0 : 35[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688201:688364 [1] NCCL INFO Channel 12/0 : 33[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688200:688369 [0] NCCL INFO Channel 14/0 : 32[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688202:688368 [2] NCCL INFO Channel 12/0 : 34[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688203:688362 [3] NCCL INFO Channel 14/0 : 35[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688201:688364 [1] NCCL INFO Channel 14/0 : 33[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688202:688368 [2] NCCL INFO Channel 14/0 : 34[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926598:926761 [4] NCCL INFO Channel 00/0 : 52[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926600:926757 [6] NCCL INFO Channel 00/0 : 54[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926598:926761 [4] NCCL INFO Channel 02/0 : 52[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926600:926757 [6] NCCL INFO Channel 02/0 : 54[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926598:926761 [4] NCCL INFO Channel 06/0 : 52[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926600:926757 [6] NCCL INFO Channel 04/0 : 54[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926598:926761 [4] NCCL INFO Channel 08/0 : 52[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926594:926762 [0] NCCL INFO Channel 02/0 : 48[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926595:926760 [1] NCCL INFO Channel 00/0 : 49[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926596:926756 [2] NCCL INFO Channel 00/0 : 50[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926597:926758 [3] NCCL INFO Channel 00/0 : 51[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926600:926757 [6] NCCL INFO Channel 08/0 : 54[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926598:926761 [4] NCCL INFO Channel 10/0 : 52[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926594:926762 [0] NCCL INFO Channel 04/0 : 48[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926595:926760 [1] NCCL INFO Channel 02/0 : 49[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926596:926756 [2] NCCL INFO Channel 04/0 : 50[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926597:926758 [3] NCCL INFO Channel 02/0 : 51[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926598:926761 [4] NCCL INFO Channel 12/0 : 52[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926600:926757 [6] NCCL INFO Channel 10/0 : 54[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926594:926762 [0] NCCL INFO Channel 06/0 : 48[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926595:926760 [1] NCCL INFO Channel 04/0 : 49[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926596:926756 [2] NCCL INFO Channel 06/0 : 50[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926597:926758 [3] NCCL INFO Channel 04/0 : 51[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926598:926761 [4] NCCL INFO Channel 14/0 : 52[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926594:926762 [0] NCCL INFO Channel 08/0 : 48[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926600:926757 [6] NCCL INFO Channel 12/0 : 54[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926595:926760 [1] NCCL INFO Channel 06/0 : 49[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926596:926756 [2] NCCL INFO Channel 08/0 : 50[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926597:926758 [3] NCCL INFO Channel 06/0 : 51[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926594:926762 [0] NCCL INFO Channel 10/0 : 48[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926600:926757 [6] NCCL INFO Channel 14/0 : 54[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926596:926756 [2] NCCL INFO Channel 10/0 : 50[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926595:926760 [1] NCCL INFO Channel 08/0 : 49[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926597:926758 [3] NCCL INFO Channel 08/0 : 51[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926594:926762 [0] NCCL INFO Channel 12/0 : 48[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926596:926756 [2] NCCL INFO Channel 12/0 : 50[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926597:926758 [3] NCCL INFO Channel 10/0 : 51[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926595:926760 [1] NCCL INFO Channel 10/0 : 49[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926596:926756 [2] NCCL INFO Channel 14/0 : 50[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926594:926762 [0] NCCL INFO Channel 14/0 : 48[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942612:942779 [1] NCCL INFO Channel 13/0 : 25[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942615:942775 [4] NCCL INFO Channel 15/0 : 28[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926595:926760 [1] NCCL INFO Channel 12/0 : 49[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926597:926758 [3] NCCL INFO Channel 12/0 : 51[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942611:942772 [0] NCCL INFO Channel 11/0 : 24[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942613:942776 [2] NCCL INFO Channel 07/0 : 26[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942617:942774 [6] NCCL INFO Channel 15/0 : 30[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942612:942779 [1] NCCL INFO Channel 15/0 : 25[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942611:942772 [0] NCCL INFO Channel 13/0 : 24[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942613:942776 [2] NCCL INFO Channel 09/0 : 26[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942611:942772 [0] NCCL INFO Channel 15/0 : 24[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942613:942776 [2] NCCL INFO Channel 11/0 : 26[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926595:926760 [1] NCCL INFO Channel 14/0 : 49[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942613:942776 [2] NCCL INFO Channel 13/0 : 26[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926597:926758 [3] NCCL INFO Channel 14/0 : 51[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942613:942776 [2] NCCL INFO Channel 15/0 : 26[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880686:880841 [7] NCCL INFO Connected all trees
+ip-26-0-172-142:880686:880841 [7] NCCL INFO NVLS comm 0x94e5950 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-142:880683:880844 [4] NCCL INFO Connected all trees
+ip-26-0-172-142:880683:880844 [4] NCCL INFO NVLS comm 0x9a8c230 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-142:880679:880846 [0] NCCL INFO Connected all trees
+ip-26-0-172-142:880679:880846 [0] NCCL INFO NVLS comm 0x9ed6ec0 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO Channel 08/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO Channel 08/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO Channel 06/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO Channel 10/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO Channel 10/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO Channel 09/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO Channel 09/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO Channel 09/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO Channel 11/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO Channel 07/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO Channel 11/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO Channel 11/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO Channel 10/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO Channel 10/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO Channel 13/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO Channel 08/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO Channel 12/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO Channel 12/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO Channel 11/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO Channel 11/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO Channel 14/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO Channel 09/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO Channel 13/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO Channel 13/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO Channel 12/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO Channel 12/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO Channel 15/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO Channel 10/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO Channel 14/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO Channel 14/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO Channel 13/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO Channel 14/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO Channel 15/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO Channel 12/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO Channel 15/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO Channel 14/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO Channel 15/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO Channel 13/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO Channel 14/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO Channel 15/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618591:618746 [7] NCCL INFO Channel 00/0 : 23[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618589:618744 [5] NCCL INFO Channel 00/0 : 21[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618585:618748 [1] NCCL INFO Channel 00/0 : 17[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618589:618744 [5] NCCL INFO Channel 02/0 : 21[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618585:618748 [1] NCCL INFO Channel 02/0 : 17[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618591:618746 [7] NCCL INFO Channel 02/0 : 23[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618585:618748 [1] NCCL INFO Channel 04/0 : 17[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618591:618746 [7] NCCL INFO Channel 04/0 : 23[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618589:618744 [5] NCCL INFO Channel 04/0 : 21[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618585:618748 [1] NCCL INFO Channel 06/0 : 17[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618589:618744 [5] NCCL INFO Channel 06/0 : 21[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618591:618746 [7] NCCL INFO Channel 06/0 : 23[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618585:618748 [1] NCCL INFO Channel 08/0 : 17[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618589:618744 [5] NCCL INFO Channel 08/0 : 21[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618591:618746 [7] NCCL INFO Channel 08/0 : 23[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618585:618748 [1] NCCL INFO Channel 10/0 : 17[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618589:618744 [5] NCCL INFO Channel 10/0 : 21[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618585:618748 [1] NCCL INFO Channel 12/0 : 17[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618591:618746 [7] NCCL INFO Channel 10/0 : 23[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618589:618744 [5] NCCL INFO Channel 12/0 : 21[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618585:618748 [1] NCCL INFO Channel 14/0 : 17[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618591:618746 [7] NCCL INFO Channel 12/0 : 23[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618589:618744 [5] NCCL INFO Channel 14/0 : 21[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618591:618746 [7] NCCL INFO Channel 14/0 : 23[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618584:618747 [0] NCCL INFO Channel 02/0 : 16[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618588:618750 [4] NCCL INFO Channel 00/0 : 20[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618587:618751 [3] NCCL INFO Channel 00/0 : 19[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618584:618747 [0] NCCL INFO Channel 04/0 : 16[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618588:618750 [4] NCCL INFO Channel 02/0 : 20[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618590:618749 [6] NCCL INFO Channel 00/0 : 22[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618586:618745 [2] NCCL INFO Channel 00/0 : 18[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618587:618751 [3] NCCL INFO Channel 02/0 : 19[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618584:618747 [0] NCCL INFO Channel 06/0 : 16[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618588:618750 [4] NCCL INFO Channel 06/0 : 20[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618586:618745 [2] NCCL INFO Channel 04/0 : 18[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618590:618749 [6] NCCL INFO Channel 02/0 : 22[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618587:618751 [3] NCCL INFO Channel 04/0 : 19[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618584:618747 [0] NCCL INFO Channel 08/0 : 16[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618588:618750 [4] NCCL INFO Channel 08/0 : 20[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618587:618751 [3] NCCL INFO Channel 06/0 : 19[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618590:618749 [6] NCCL INFO Channel 04/0 : 22[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618586:618745 [2] NCCL INFO Channel 06/0 : 18[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618584:618747 [0] NCCL INFO Channel 10/0 : 16[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618588:618750 [4] NCCL INFO Channel 10/0 : 20[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618587:618751 [3] NCCL INFO Channel 08/0 : 19[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618590:618749 [6] NCCL INFO Channel 08/0 : 22[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618588:618750 [4] NCCL INFO Channel 12/0 : 20[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618587:618751 [3] NCCL INFO Channel 10/0 : 19[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618586:618745 [2] NCCL INFO Channel 08/0 : 18[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618584:618747 [0] NCCL INFO Channel 12/0 : 16[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618590:618749 [6] NCCL INFO Channel 10/0 : 22[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618588:618750 [4] NCCL INFO Channel 14/0 : 20[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618587:618751 [3] NCCL INFO Channel 12/0 : 19[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618590:618749 [6] NCCL INFO Channel 12/0 : 22[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618586:618745 [2] NCCL INFO Channel 10/0 : 18[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618584:618747 [0] NCCL INFO Channel 14/0 : 16[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618587:618751 [3] NCCL INFO Channel 14/0 : 19[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618590:618749 [6] NCCL INFO Channel 14/0 : 22[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618586:618745 [2] NCCL INFO Channel 12/0 : 18[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618586:618745 [2] NCCL INFO Channel 14/0 : 18[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880680:880843 [1] NCCL INFO Channel 00/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880680:880843 [1] NCCL INFO Channel 02/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880680:880843 [1] NCCL INFO Channel 03/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880680:880843 [1] NCCL INFO Channel 04/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880680:880843 [1] NCCL INFO Channel 05/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880680:880843 [1] NCCL INFO Channel 06/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880680:880843 [1] NCCL INFO Channel 07/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880680:880843 [1] NCCL INFO Channel 08/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880680:880843 [1] NCCL INFO Channel 10/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880680:880843 [1] NCCL INFO Channel 11/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880680:880843 [1] NCCL INFO Channel 12/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880680:880843 [1] NCCL INFO Channel 13/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880680:880843 [1] NCCL INFO Channel 14/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880680:880843 [1] NCCL INFO Channel 15/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880680:880843 [1] NCCL INFO Channel 00/0 : 17[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880686:880841 [7] NCCL INFO Channel 00/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880680:880843 [1] NCCL INFO Channel 02/0 : 17[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880680:880843 [1] NCCL INFO Channel 04/0 : 17[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880686:880841 [7] NCCL INFO Channel 01/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880679:880846 [0] NCCL INFO Channel 01/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880680:880843 [1] NCCL INFO Channel 06/0 : 17[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880684:880842 [5] NCCL INFO Channel 00/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880685:880847 [6] NCCL INFO Channel 00/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880686:880841 [7] NCCL INFO Channel 02/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880680:880843 [1] NCCL INFO Channel 08/0 : 17[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880679:880846 [0] NCCL INFO Channel 02/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880685:880847 [6] NCCL INFO Channel 01/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880686:880841 [7] NCCL INFO Channel 03/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880680:880843 [1] NCCL INFO Channel 10/0 : 17[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880684:880842 [5] NCCL INFO Channel 01/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880679:880846 [0] NCCL INFO Channel 03/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880685:880847 [6] NCCL INFO Channel 02/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880686:880841 [7] NCCL INFO Channel 04/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880680:880843 [1] NCCL INFO Channel 12/0 : 17[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880684:880842 [5] NCCL INFO Channel 02/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880679:880846 [0] NCCL INFO Channel 04/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880685:880847 [6] NCCL INFO Channel 03/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880686:880841 [7] NCCL INFO Channel 05/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880680:880843 [1] NCCL INFO Channel 14/0 : 17[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880679:880846 [0] NCCL INFO Channel 05/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880684:880842 [5] NCCL INFO Channel 03/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880686:880841 [7] NCCL INFO Channel 06/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880685:880847 [6] NCCL INFO Channel 04/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880684:880842 [5] NCCL INFO Channel 04/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880679:880846 [0] NCCL INFO Channel 06/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880685:880847 [6] NCCL INFO Channel 05/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880686:880841 [7] NCCL INFO Channel 08/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880684:880842 [5] NCCL INFO Channel 06/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880679:880846 [0] NCCL INFO Channel 07/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880686:880841 [7] NCCL INFO Channel 09/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880685:880847 [6] NCCL INFO Channel 07/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880684:880842 [5] NCCL INFO Channel 07/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880679:880846 [0] NCCL INFO Channel 09/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880686:880841 [7] NCCL INFO Channel 10/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880685:880847 [6] NCCL INFO Channel 08/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880684:880842 [5] NCCL INFO Channel 08/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880679:880846 [0] NCCL INFO Channel 10/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO Channel 01/0 : 9[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880686:880841 [7] NCCL INFO Channel 11/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880685:880847 [6] NCCL INFO Channel 09/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO Channel 03/0 : 9[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880684:880842 [5] NCCL INFO Channel 09/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880679:880846 [0] NCCL INFO Channel 11/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880680:880843 [1] NCCL INFO Channel 00/0 : 17[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO Channel 05/0 : 9[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618585:618748 [1] NCCL INFO Channel 01/0 : 9[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880685:880847 [6] NCCL INFO Channel 10/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880686:880841 [7] NCCL INFO Channel 12/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880684:880842 [5] NCCL INFO Channel 10/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO Channel 07/0 : 9[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618585:618748 [1] NCCL INFO Channel 03/0 : 9[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880679:880846 [0] NCCL INFO Channel 12/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880680:880843 [1] NCCL INFO Channel 02/0 : 17[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880685:880847 [6] NCCL INFO Channel 11/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880686:880841 [7] NCCL INFO Channel 13/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880684:880842 [5] NCCL INFO Channel 11/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618585:618748 [1] NCCL INFO Channel 05/0 : 9[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO Channel 11/0 : 9[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880679:880846 [0] NCCL INFO Channel 13/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880680:880843 [1] NCCL INFO Channel 04/0 : 17[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880685:880847 [6] NCCL INFO Channel 12/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880686:880841 [7] NCCL INFO Channel 14/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880684:880842 [5] NCCL INFO Channel 12/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO Channel 13/0 : 9[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618585:618748 [1] NCCL INFO Channel 07/0 : 9[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880679:880846 [0] NCCL INFO Channel 14/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880680:880843 [1] NCCL INFO Channel 06/0 : 17[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880685:880847 [6] NCCL INFO Channel 13/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880686:880841 [7] NCCL INFO Channel 00/0 : 23[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880684:880842 [5] NCCL INFO Channel 14/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO Channel 15/0 : 9[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618585:618748 [1] NCCL INFO Channel 11/0 : 9[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880679:880846 [0] NCCL INFO Channel 15/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880680:880843 [1] NCCL INFO Channel 08/0 : 17[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880685:880847 [6] NCCL INFO Channel 15/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880686:880841 [7] NCCL INFO Channel 02/0 : 23[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880684:880842 [5] NCCL INFO Channel 15/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618585:618748 [1] NCCL INFO Channel 13/0 : 9[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880679:880846 [0] NCCL INFO Channel 02/0 : 16[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880680:880843 [1] NCCL INFO Channel 10/0 : 17[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880685:880847 [6] NCCL INFO Channel 00/0 : 22[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880686:880841 [7] NCCL INFO Channel 04/0 : 23[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880684:880842 [5] NCCL INFO Channel 00/0 : 21[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618585:618748 [1] NCCL INFO Channel 15/0 : 9[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880683:880844 [4] NCCL INFO Channel 00/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880679:880846 [0] NCCL INFO Channel 04/0 : 16[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880680:880843 [1] NCCL INFO Channel 12/0 : 17[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880685:880847 [6] NCCL INFO Channel 02/0 : 22[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880686:880841 [7] NCCL INFO Channel 06/0 : 23[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880684:880842 [5] NCCL INFO Channel 02/0 : 21[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618585:618748 [1] NCCL INFO Channel 01/0 : 25[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880679:880846 [0] NCCL INFO Channel 06/0 : 16[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880680:880843 [1] NCCL INFO Channel 14/0 : 17[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880683:880844 [4] NCCL INFO Channel 01/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880685:880847 [6] NCCL INFO Channel 04/0 : 22[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880686:880841 [7] NCCL INFO Channel 08/0 : 23[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880684:880842 [5] NCCL INFO Channel 04/0 : 21[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618585:618748 [1] NCCL INFO Channel 03/0 : 25[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618585:618748 [1] NCCL INFO Channel 05/0 : 25[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618585:618748 [1] NCCL INFO Channel 07/0 : 25[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618585:618748 [1] NCCL INFO Channel 11/0 : 25[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618585:618748 [1] NCCL INFO Channel 13/0 : 25[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688201:688364 [1] NCCL INFO Channel 00/0 : 1[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688201:688364 [1] NCCL INFO Channel 02/0 : 1[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618585:618748 [1] NCCL INFO Channel 15/0 : 25[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688201:688364 [1] NCCL INFO Channel 04/0 : 1[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688201:688364 [1] NCCL INFO Channel 06/0 : 1[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688201:688364 [1] NCCL INFO Channel 08/0 : 1[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO Channel 01/0 : 15[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688201:688364 [1] NCCL INFO Channel 10/0 : 1[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO Channel 01/0 : 25[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO Channel 03/0 : 15[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618591:618746 [7] NCCL INFO Channel 01/0 : 15[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688201:688364 [1] NCCL INFO Channel 12/0 : 1[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO Channel 01/0 : 8[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO Channel 03/0 : 25[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618585:618748 [1] NCCL INFO Channel 01/0 : 57[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688201:688364 [1] NCCL INFO Channel 14/0 : 1[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO Channel 01/0 : 14[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618591:618746 [7] NCCL INFO Channel 03/0 : 15[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO Channel 05/0 : 15[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO Channel 03/0 : 8[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO Channel 05/0 : 25[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688201:688364 [1] NCCL INFO Channel 00/0 : 33[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618584:618747 [0] NCCL INFO Channel 01/0 : 8[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618590:618749 [6] NCCL INFO Channel 01/0 : 14[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618585:618748 [1] NCCL INFO Channel 03/0 : 57[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO Channel 03/0 : 14[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO Channel 01/0 : 13[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618591:618746 [7] NCCL INFO Channel 05/0 : 15[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688201:688364 [1] NCCL INFO Channel 02/0 : 33[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO Channel 05/0 : 8[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO Channel 07/0 : 15[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO Channel 07/0 : 25[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618589:618744 [5] NCCL INFO Channel 01/0 : 13[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618584:618747 [0] NCCL INFO Channel 03/0 : 8[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618585:618748 [1] NCCL INFO Channel 05/0 : 57[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO Channel 03/0 : 13[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO Channel 05/0 : 14[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618590:618749 [6] NCCL INFO Channel 03/0 : 14[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688201:688364 [1] NCCL INFO Channel 04/0 : 33[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO Channel 07/0 : 8[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO Channel 09/0 : 15[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO Channel 11/0 : 25[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618591:618746 [7] NCCL INFO Channel 07/0 : 15[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO Channel 05/0 : 13[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688201:688364 [1] NCCL INFO Channel 06/0 : 33[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618585:618748 [1] NCCL INFO Channel 07/0 : 57[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO Channel 07/0 : 14[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO Channel 09/0 : 8[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO Channel 11/0 : 15[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO Channel 13/0 : 25[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618589:618744 [5] NCCL INFO Channel 03/0 : 13[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688201:688364 [1] NCCL INFO Channel 08/0 : 33[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618584:618747 [0] NCCL INFO Channel 05/0 : 8[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618590:618749 [6] NCCL INFO Channel 05/0 : 14[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618591:618746 [7] NCCL INFO Channel 09/0 : 15[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO Channel 07/0 : 13[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO Channel 09/0 : 14[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO Channel 11/0 : 8[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618585:618748 [1] NCCL INFO Channel 11/0 : 57[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618589:618744 [5] NCCL INFO Channel 05/0 : 13[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688201:688364 [1] NCCL INFO Channel 10/0 : 33[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880681:880845 [2] NCCL INFO Channel 00/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO Channel 13/0 : 15[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880679:880846 [0] NCCL INFO Channel 08/0 : 16[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO Channel 15/0 : 25[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880683:880844 [4] NCCL INFO Channel 02/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO Channel 09/0 : 13[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880685:880847 [6] NCCL INFO Channel 08/0 : 22[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO Channel 11/0 : 14[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880686:880841 [7] NCCL INFO Channel 10/0 : 23[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO Channel 13/0 : 8[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880684:880842 [5] NCCL INFO Channel 06/0 : 21[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618584:618747 [0] NCCL INFO Channel 07/0 : 8[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880681:880845 [2] NCCL INFO Channel 01/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880679:880846 [0] NCCL INFO Channel 10/0 : 16[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618591:618746 [7] NCCL INFO Channel 11/0 : 15[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880683:880844 [4] NCCL INFO Channel 03/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618590:618749 [6] NCCL INFO Channel 07/0 : 14[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688201:688364 [1] NCCL INFO Channel 12/0 : 33[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880685:880847 [6] NCCL INFO Channel 10/0 : 22[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618585:618748 [1] NCCL INFO Channel 13/0 : 57[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880686:880841 [7] NCCL INFO Channel 12/0 : 23[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO Channel 11/0 : 13[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880684:880842 [5] NCCL INFO Channel 08/0 : 21[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880681:880845 [2] NCCL INFO Channel 03/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880679:880846 [0] NCCL INFO Channel 12/0 : 16[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880680:880843 [1] NCCL INFO Channel 00/0 : 33[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880685:880847 [6] NCCL INFO Channel 12/0 : 22[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880683:880844 [4] NCCL INFO Channel 05/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880686:880841 [7] NCCL INFO Channel 14/0 : 23[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880684:880842 [5] NCCL INFO Channel 10/0 : 21[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880682:880840 [3] NCCL INFO Channel 00/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880681:880845 [2] NCCL INFO Channel 04/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618589:618744 [5] NCCL INFO Channel 07/0 : 13[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688201:688364 [1] NCCL INFO Channel 14/0 : 33[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880679:880846 [0] NCCL INFO Channel 14/0 : 16[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO Channel 13/0 : 14[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO Channel 15/0 : 8[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880680:880843 [1] NCCL INFO Channel 02/0 : 33[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO Channel 15/0 : 13[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880685:880847 [6] NCCL INFO Channel 14/0 : 22[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880683:880844 [4] NCCL INFO Channel 06/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880684:880842 [5] NCCL INFO Channel 12/0 : 21[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880682:880840 [3] NCCL INFO Channel 01/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880681:880845 [2] NCCL INFO Channel 05/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880680:880843 [1] NCCL INFO Channel 04/0 : 33[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618590:618749 [6] NCCL INFO Channel 09/0 : 14[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618584:618747 [0] NCCL INFO Channel 09/0 : 8[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618591:618746 [7] NCCL INFO Channel 13/0 : 15[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618585:618748 [1] NCCL INFO Channel 15/0 : 57[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880683:880844 [4] NCCL INFO Channel 07/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880684:880842 [5] NCCL INFO Channel 14/0 : 21[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618589:618744 [5] NCCL INFO Channel 09/0 : 13[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880682:880840 [3] NCCL INFO Channel 02/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO Channel 15/0 : 14[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880681:880845 [2] NCCL INFO Channel 06/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880680:880843 [1] NCCL INFO Channel 06/0 : 33[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880683:880844 [4] NCCL INFO Channel 08/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880682:880840 [3] NCCL INFO Channel 04/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880680:880843 [1] NCCL INFO Channel 08/0 : 33[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880683:880844 [4] NCCL INFO Channel 09/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880681:880845 [2] NCCL INFO Channel 07/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618590:618749 [6] NCCL INFO Channel 11/0 : 14[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618591:618746 [7] NCCL INFO Channel 01/0 : 31[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618585:618748 [1] NCCL INFO Channel 01/0 : 25[1] -> 57[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880682:880840 [3] NCCL INFO Channel 05/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880686:880841 [7] NCCL INFO Channel 00/0 : 23[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880683:880844 [4] NCCL INFO Channel 10/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880680:880843 [1] NCCL INFO Channel 10/0 : 33[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942612:942779 [1] NCCL INFO Channel 01/0 : 41[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880681:880845 [2] NCCL INFO Channel 08/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880686:880841 [7] NCCL INFO Channel 02/0 : 23[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880682:880840 [3] NCCL INFO Channel 06/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618589:618744 [5] NCCL INFO Channel 11/0 : 13[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618584:618747 [0] NCCL INFO Channel 11/0 : 8[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880685:880847 [6] NCCL INFO Channel 00/0 : 22[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880679:880846 [0] NCCL INFO Channel 02/0 : 16[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880683:880844 [4] NCCL INFO Channel 11/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880680:880843 [1] NCCL INFO Channel 12/0 : 33[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880681:880845 [2] NCCL INFO Channel 09/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880686:880841 [7] NCCL INFO Channel 04/0 : 23[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618590:618749 [6] NCCL INFO Channel 13/0 : 14[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688201:688364 [1] NCCL INFO Channel 00/0 : 49[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880682:880840 [3] NCCL INFO Channel 07/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880684:880842 [5] NCCL INFO Channel 00/0 : 21[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880685:880847 [6] NCCL INFO Channel 02/0 : 22[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880680:880843 [1] NCCL INFO Channel 14/0 : 33[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880683:880844 [4] NCCL INFO Channel 13/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880679:880846 [0] NCCL INFO Channel 04/0 : 16[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880681:880845 [2] NCCL INFO Channel 11/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880686:880841 [7] NCCL INFO Channel 06/0 : 23[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880682:880840 [3] NCCL INFO Channel 08/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880684:880842 [5] NCCL INFO Channel 02/0 : 21[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618591:618746 [7] NCCL INFO Channel 03/0 : 31[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618585:618748 [1] NCCL INFO Channel 03/0 : 25[1] -> 57[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880685:880847 [6] NCCL INFO Channel 04/0 : 22[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618589:618744 [5] NCCL INFO Channel 15/0 : 13[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618584:618747 [0] NCCL INFO Channel 13/0 : 8[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880679:880846 [0] NCCL INFO Channel 06/0 : 16[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880681:880845 [2] NCCL INFO Channel 12/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880683:880844 [4] NCCL INFO Channel 14/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880682:880840 [3] NCCL INFO Channel 09/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880686:880841 [7] NCCL INFO Channel 08/0 : 23[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880685:880847 [6] NCCL INFO Channel 08/0 : 22[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880684:880842 [5] NCCL INFO Channel 04/0 : 21[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688201:688364 [1] NCCL INFO Channel 02/0 : 49[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880679:880846 [0] NCCL INFO Channel 08/0 : 16[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880681:880845 [2] NCCL INFO Channel 13/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880683:880844 [4] NCCL INFO Channel 15/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618590:618749 [6] NCCL INFO Channel 15/0 : 14[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880682:880840 [3] NCCL INFO Channel 10/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880685:880847 [6] NCCL INFO Channel 10/0 : 22[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880686:880841 [7] NCCL INFO Channel 10/0 : 23[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618591:618746 [7] NCCL INFO Channel 05/0 : 31[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880684:880842 [5] NCCL INFO Channel 06/0 : 21[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618585:618748 [1] NCCL INFO Channel 05/0 : 25[1] -> 57[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880679:880846 [0] NCCL INFO Channel 10/0 : 16[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880681:880845 [2] NCCL INFO Channel 14/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880683:880844 [4] NCCL INFO Channel 00/0 : 20[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688201:688364 [1] NCCL INFO Channel 04/0 : 49[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618589:618744 [5] NCCL INFO Channel 01/0 : 29[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618584:618747 [0] NCCL INFO Channel 15/0 : 8[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618590:618749 [6] NCCL INFO Channel 01/0 : 30[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618591:618746 [7] NCCL INFO Channel 07/0 : 31[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618585:618748 [1] NCCL INFO Channel 07/0 : 25[1] -> 57[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880682:880840 [3] NCCL INFO Channel 12/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688206:688363 [6] NCCL INFO Channel 00/0 : 6[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880685:880847 [6] NCCL INFO Channel 12/0 : 22[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880686:880841 [7] NCCL INFO Channel 12/0 : 23[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880679:880846 [0] NCCL INFO Channel 12/0 : 16[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880684:880842 [5] NCCL INFO Channel 08/0 : 21[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880681:880845 [2] NCCL INFO Channel 15/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880683:880844 [4] NCCL INFO Channel 02/0 : 20[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618589:618744 [5] NCCL INFO Channel 03/0 : 29[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880682:880840 [3] NCCL INFO Channel 13/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618584:618747 [0] NCCL INFO Channel 01/0 : 24[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688201:688364 [1] NCCL INFO Channel 06/0 : 49[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688207:688366 [7] NCCL INFO Channel 00/0 : 7[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880685:880847 [6] NCCL INFO Channel 14/0 : 22[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618590:618749 [6] NCCL INFO Channel 03/0 : 30[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688206:688363 [6] NCCL INFO Channel 02/0 : 6[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880686:880841 [7] NCCL INFO Channel 14/0 : 23[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618591:618746 [7] NCCL INFO Channel 09/0 : 31[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880679:880846 [0] NCCL INFO Channel 14/0 : 16[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618585:618748 [1] NCCL INFO Channel 11/0 : 25[1] -> 57[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880683:880844 [4] NCCL INFO Channel 06/0 : 20[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880684:880842 [5] NCCL INFO Channel 10/0 : 21[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618589:618744 [5] NCCL INFO Channel 05/0 : 29[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880681:880845 [2] NCCL INFO Channel 00/0 : 18[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880682:880840 [3] NCCL INFO Channel 14/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880683:880844 [4] NCCL INFO Channel 08/0 : 20[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688201:688364 [1] NCCL INFO Channel 08/0 : 49[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880684:880842 [5] NCCL INFO Channel 12/0 : 21[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880681:880845 [2] NCCL INFO Channel 04/0 : 18[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880682:880840 [3] NCCL INFO Channel 15/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618584:618747 [0] NCCL INFO Channel 03/0 : 24[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880685:880847 [6] NCCL INFO Channel 00/0 : 38[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618590:618749 [6] NCCL INFO Channel 05/0 : 30[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880683:880844 [4] NCCL INFO Channel 10/0 : 20[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618591:618746 [7] NCCL INFO Channel 11/0 : 31[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880684:880842 [5] NCCL INFO Channel 14/0 : 21[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618585:618748 [1] NCCL INFO Channel 13/0 : 25[1] -> 57[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880681:880845 [2] NCCL INFO Channel 06/0 : 18[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618589:618744 [5] NCCL INFO Channel 07/0 : 29[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688207:688366 [7] NCCL INFO Channel 02/0 : 7[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688206:688363 [6] NCCL INFO Channel 04/0 : 6[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880682:880840 [3] NCCL INFO Channel 00/0 : 19[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688201:688364 [1] NCCL INFO Channel 10/0 : 49[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880679:880846 [0] NCCL INFO Channel 02/0 : 32[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880686:880841 [7] NCCL INFO Channel 00/0 : 39[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880685:880847 [6] NCCL INFO Channel 02/0 : 38[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880683:880844 [4] NCCL INFO Channel 12/0 : 20[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618584:618747 [0] NCCL INFO Channel 05/0 : 24[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618590:618749 [6] NCCL INFO Channel 07/0 : 30[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618591:618746 [7] NCCL INFO Channel 13/0 : 31[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618585:618748 [1] NCCL INFO Channel 15/0 : 25[1] -> 57[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618589:618744 [5] NCCL INFO Channel 09/0 : 29[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688206:688363 [6] NCCL INFO Channel 08/0 : 6[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688207:688366 [7] NCCL INFO Channel 04/0 : 7[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688201:688364 [1] NCCL INFO Channel 12/0 : 49[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880682:880840 [3] NCCL INFO Channel 02/0 : 19[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880681:880845 [2] NCCL INFO Channel 08/0 : 18[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880686:880841 [7] NCCL INFO Channel 02/0 : 39[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880679:880846 [0] NCCL INFO Channel 04/0 : 32[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880685:880847 [6] NCCL INFO Channel 04/0 : 38[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880683:880844 [4] NCCL INFO Channel 14/0 : 20[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880682:880840 [3] NCCL INFO Channel 04/0 : 19[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880681:880845 [2] NCCL INFO Channel 10/0 : 18[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880679:880846 [0] NCCL INFO Channel 06/0 : 32[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688205:688365 [5] NCCL INFO Channel 00/0 : 5[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880686:880841 [7] NCCL INFO Channel 04/0 : 39[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880685:880847 [6] NCCL INFO Channel 08/0 : 38[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880682:880840 [3] NCCL INFO Channel 06/0 : 19[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880684:880842 [5] NCCL INFO Channel 00/0 : 37[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880681:880845 [2] NCCL INFO Channel 12/0 : 18[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880679:880846 [0] NCCL INFO Channel 08/0 : 32[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO Channel 01/0 : 12[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688201:688364 [1] NCCL INFO Channel 14/0 : 49[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688207:688366 [7] NCCL INFO Channel 06/0 : 7[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688206:688363 [6] NCCL INFO Channel 10/0 : 6[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880686:880841 [7] NCCL INFO Channel 06/0 : 39[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880685:880847 [6] NCCL INFO Channel 10/0 : 38[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880682:880840 [3] NCCL INFO Channel 08/0 : 19[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880681:880845 [2] NCCL INFO Channel 14/0 : 18[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880684:880842 [5] NCCL INFO Channel 02/0 : 37[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO Channel 03/0 : 12[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880679:880846 [0] NCCL INFO Channel 10/0 : 32[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880682:880840 [3] NCCL INFO Channel 10/0 : 19[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880685:880847 [6] NCCL INFO Channel 12/0 : 38[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688205:688365 [5] NCCL INFO Channel 02/0 : 5[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880686:880841 [7] NCCL INFO Channel 08/0 : 39[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880684:880842 [5] NCCL INFO Channel 04/0 : 37[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688201:688364 [1] NCCL INFO Channel 00/0 : 33[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO Channel 01/0 : 31[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880683:880844 [4] NCCL INFO Channel 00/0 : 20[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880679:880846 [0] NCCL INFO Channel 12/0 : 32[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688207:688366 [7] NCCL INFO Channel 08/0 : 7[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688206:688363 [6] NCCL INFO Channel 12/0 : 6[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688205:688365 [5] NCCL INFO Channel 04/0 : 5[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880685:880847 [6] NCCL INFO Channel 14/0 : 38[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688201:688364 [1] NCCL INFO Channel 02/0 : 33[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO Channel 05/0 : 12[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880686:880841 [7] NCCL INFO Channel 10/0 : 39[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880684:880842 [5] NCCL INFO Channel 06/0 : 37[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880682:880840 [3] NCCL INFO Channel 12/0 : 19[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688206:688363 [6] NCCL INFO Channel 14/0 : 6[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880683:880844 [4] NCCL INFO Channel 02/0 : 20[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880679:880846 [0] NCCL INFO Channel 14/0 : 32[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO Channel 03/0 : 31[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO Channel 07/0 : 12[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-7:811249:811413 [1] NCCL INFO Channel 00/0 : 57[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688205:688365 [5] NCCL INFO Channel 06/0 : 5[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688207:688366 [7] NCCL INFO Channel 10/0 : 7[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688201:688364 [1] NCCL INFO Channel 04/0 : 33[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688206:688363 [6] NCCL INFO Channel 00/0 : 38[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880686:880841 [7] NCCL INFO Channel 12/0 : 39[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880684:880842 [5] NCCL INFO Channel 08/0 : 37[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO Channel 01/0 : 29[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880682:880840 [3] NCCL INFO Channel 14/0 : 19[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-7:811249:811413 [1] NCCL INFO Channel 02/0 : 57[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880683:880844 [4] NCCL INFO Channel 06/0 : 20[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688201:688364 [1] NCCL INFO Channel 06/0 : 33[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO Channel 05/0 : 31[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO Channel 09/0 : 12[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-7:811249:811413 [1] NCCL INFO Channel 04/0 : 57[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880686:880841 [7] NCCL INFO Channel 14/0 : 39[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880684:880842 [5] NCCL INFO Channel 10/0 : 37[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880683:880844 [4] NCCL INFO Channel 08/0 : 20[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880681:880845 [2] NCCL INFO Channel 00/0 : 18[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688207:688366 [7] NCCL INFO Channel 12/0 : 7[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688205:688365 [5] NCCL INFO Channel 08/0 : 5[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO Channel 03/0 : 29[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO Channel 01/0 : 10[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688206:688363 [6] NCCL INFO Channel 02/0 : 38[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688201:688364 [1] NCCL INFO Channel 08/0 : 33[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-7:811249:811413 [1] NCCL INFO Channel 06/0 : 57[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880684:880842 [5] NCCL INFO Channel 12/0 : 37[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO Channel 07/0 : 31[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880683:880844 [4] NCCL INFO Channel 10/0 : 20[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688207:688366 [7] NCCL INFO Channel 14/0 : 7[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO Channel 11/0 : 12[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880681:880845 [2] NCCL INFO Channel 04/0 : 18[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688206:688363 [6] NCCL INFO Channel 04/0 : 38[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO Channel 05/0 : 29[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880684:880842 [5] NCCL INFO Channel 14/0 : 37[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688205:688365 [5] NCCL INFO Channel 10/0 : 5[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688201:688364 [1] NCCL INFO Channel 10/0 : 33[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-7:811249:811413 [1] NCCL INFO Channel 08/0 : 57[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO Channel 01/0 : 30[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO Channel 03/0 : 10[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO Channel 09/0 : 31[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880683:880844 [4] NCCL INFO Channel 12/0 : 20[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO Channel 13/0 : 12[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880681:880845 [2] NCCL INFO Channel 06/0 : 18[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880683:880844 [4] NCCL INFO Channel 14/0 : 20[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688207:688366 [7] NCCL INFO Channel 00/0 : 39[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-7:811249:811413 [1] NCCL INFO Channel 10/0 : 57[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688206:688363 [6] NCCL INFO Channel 08/0 : 38[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688205:688365 [5] NCCL INFO Channel 12/0 : 5[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688201:688364 [1] NCCL INFO Channel 12/0 : 33[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO Channel 01/0 : 24[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO Channel 07/0 : 29[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880681:880845 [2] NCCL INFO Channel 08/0 : 18[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-7:811249:811413 [1] NCCL INFO Channel 12/0 : 57[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688207:688366 [7] NCCL INFO Channel 02/0 : 39[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688206:688363 [6] NCCL INFO Channel 10/0 : 38[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO Channel 03/0 : 30[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO Channel 05/0 : 10[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688201:688364 [1] NCCL INFO Channel 14/0 : 33[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO Channel 11/0 : 31[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880682:880840 [3] NCCL INFO Channel 00/0 : 19[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880681:880845 [2] NCCL INFO Channel 10/0 : 18[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688205:688365 [5] NCCL INFO Channel 14/0 : 5[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO Channel 15/0 : 12[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688207:688366 [7] NCCL INFO Channel 04/0 : 39[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO Channel 03/0 : 24[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO Channel 01/0 : 11[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688206:688363 [6] NCCL INFO Channel 12/0 : 38[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-7:811249:811413 [1] NCCL INFO Channel 14/0 : 57[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO Channel 09/0 : 29[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880681:880845 [2] NCCL INFO Channel 12/0 : 18[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880682:880840 [3] NCCL INFO Channel 02/0 : 19[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688207:688366 [7] NCCL INFO Channel 06/0 : 39[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688205:688365 [5] NCCL INFO Channel 00/0 : 37[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688206:688363 [6] NCCL INFO Channel 14/0 : 38[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO Channel 07/0 : 10[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO Channel 05/0 : 30[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO Channel 13/0 : 31[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO Channel 05/0 : 24[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO Channel 11/0 : 29[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO Channel 03/0 : 11[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880683:880844 [4] NCCL INFO Channel 00/0 : 36[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880681:880845 [2] NCCL INFO Channel 14/0 : 18[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688207:688366 [7] NCCL INFO Channel 08/0 : 39[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688205:688365 [5] NCCL INFO Channel 02/0 : 37[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO Channel 07/0 : 30[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO Channel 09/0 : 10[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880682:880840 [3] NCCL INFO Channel 04/0 : 19[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880683:880844 [4] NCCL INFO Channel 02/0 : 36[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688204:688367 [4] NCCL INFO Channel 00/0 : 4[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688207:688366 [7] NCCL INFO Channel 10/0 : 39[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688205:688365 [5] NCCL INFO Channel 04/0 : 37[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO Channel 15/0 : 29[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO Channel 07/0 : 24[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO Channel 05/0 : 11[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO Channel 09/0 : 30[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO Channel 11/0 : 10[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880682:880840 [3] NCCL INFO Channel 06/0 : 19[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880683:880844 [4] NCCL INFO Channel 06/0 : 36[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO Channel 07/0 : 11[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688204:688367 [4] NCCL INFO Channel 02/0 : 4[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO Channel 11/0 : 30[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO Channel 09/0 : 24[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO Channel 13/0 : 10[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880682:880840 [3] NCCL INFO Channel 08/0 : 19[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880680:880843 [1] NCCL INFO Channel 00/0 : 25[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880683:880844 [4] NCCL INFO Channel 08/0 : 36[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880681:880845 [2] NCCL INFO Channel 00/0 : 34[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO Channel 09/0 : 11[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688207:688366 [7] NCCL INFO Channel 12/0 : 39[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO Channel 13/0 : 30[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688205:688365 [5] NCCL INFO Channel 06/0 : 37[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO Channel 15/0 : 10[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688201:688364 [1] NCCL INFO Channel 01/0 : 41[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688206:688363 [6] NCCL INFO Channel 00/0 : 54[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688204:688367 [4] NCCL INFO Channel 06/0 : 4[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688207:688366 [7] NCCL INFO Channel 14/0 : 39[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688205:688365 [5] NCCL INFO Channel 08/0 : 37[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880682:880840 [3] NCCL INFO Channel 10/0 : 19[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO Channel 11/0 : 24[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO Channel 13/0 : 11[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO Channel 15/0 : 30[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688201:688364 [1] NCCL INFO Channel 03/0 : 41[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688206:688363 [6] NCCL INFO Channel 02/0 : 54[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880680:880843 [1] NCCL INFO Channel 02/0 : 25[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880683:880844 [4] NCCL INFO Channel 10/0 : 36[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688204:688367 [4] NCCL INFO Channel 08/0 : 4[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880681:880845 [2] NCCL INFO Channel 04/0 : 34[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688205:688365 [5] NCCL INFO Channel 10/0 : 37[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO Channel 13/0 : 24[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO Channel 15/0 : 11[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942612:942779 [1] NCCL INFO Channel 03/0 : 41[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942612:942779 [1] NCCL INFO Channel 05/0 : 41[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880682:880840 [3] NCCL INFO Channel 12/0 : 19[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO Channel 15/0 : 24[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942612:942779 [1] NCCL INFO Channel 07/0 : 41[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880680:880843 [1] NCCL INFO Channel 04/0 : 25[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688201:688364 [1] NCCL INFO Channel 05/0 : 41[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942612:942779 [1] NCCL INFO Channel 11/0 : 41[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880681:880845 [2] NCCL INFO Channel 06/0 : 34[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880683:880844 [4] NCCL INFO Channel 12/0 : 36[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688206:688363 [6] NCCL INFO Channel 04/0 : 54[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942612:942779 [1] NCCL INFO Channel 13/0 : 41[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688202:688368 [2] NCCL INFO Channel 00/0 : 2[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942612:942779 [1] NCCL INFO Channel 15/0 : 41[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688205:688365 [5] NCCL INFO Channel 12/0 : 37[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688204:688367 [4] NCCL INFO Channel 10/0 : 4[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688201:688364 [1] NCCL INFO Channel 07/0 : 41[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880682:880840 [3] NCCL INFO Channel 14/0 : 19[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880680:880843 [1] NCCL INFO Channel 06/0 : 25[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926595:926760 [1] NCCL INFO Channel 00/0 : 57[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688206:688363 [6] NCCL INFO Channel 08/0 : 54[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688205:688365 [5] NCCL INFO Channel 14/0 : 37[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688204:688367 [4] NCCL INFO Channel 12/0 : 4[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688202:688368 [2] NCCL INFO Channel 04/0 : 2[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880681:880845 [2] NCCL INFO Channel 08/0 : 34[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880683:880844 [4] NCCL INFO Channel 14/0 : 36[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688207:688366 [7] NCCL INFO Channel 00/0 : 55[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688201:688364 [1] NCCL INFO Channel 11/0 : 41[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688206:688363 [6] NCCL INFO Channel 10/0 : 54[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880680:880843 [1] NCCL INFO Channel 08/0 : 25[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880681:880845 [2] NCCL INFO Channel 10/0 : 34[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688202:688368 [2] NCCL INFO Channel 06/0 : 2[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880680:880843 [1] NCCL INFO Channel 10/0 : 25[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880681:880845 [2] NCCL INFO Channel 12/0 : 34[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618584:618747 [0] NCCL INFO Channel 07/0 : 24[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688204:688367 [4] NCCL INFO Channel 14/0 : 4[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618590:618749 [6] NCCL INFO Channel 09/0 : 30[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688207:688366 [7] NCCL INFO Channel 02/0 : 55[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688206:688363 [6] NCCL INFO Channel 12/0 : 54[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688201:688364 [1] NCCL INFO Channel 13/0 : 41[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618589:618744 [5] NCCL INFO Channel 11/0 : 29[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618590:618749 [6] NCCL INFO Channel 11/0 : 30[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618584:618747 [0] NCCL INFO Channel 09/0 : 24[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618589:618744 [5] NCCL INFO Channel 15/0 : 29[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880681:880845 [2] NCCL INFO Channel 14/0 : 34[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618590:618749 [6] NCCL INFO Channel 13/0 : 30[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618585:618748 [1] NCCL INFO Channel 01/0 : 41[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618584:618747 [0] NCCL INFO Channel 11/0 : 24[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618590:618749 [6] NCCL INFO Channel 15/0 : 30[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688202:688368 [2] NCCL INFO Channel 08/0 : 2[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618588:618750 [4] NCCL INFO Channel 01/0 : 12[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688205:688365 [5] NCCL INFO Channel 00/0 : 53[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618584:618747 [0] NCCL INFO Channel 13/0 : 24[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618585:618748 [1] NCCL INFO Channel 03/0 : 41[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880680:880843 [1] NCCL INFO Channel 12/0 : 25[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618591:618746 [7] NCCL INFO Channel 01/0 : 63[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880682:880840 [3] NCCL INFO Channel 00/0 : 35[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618584:618747 [0] NCCL INFO Channel 15/0 : 24[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618588:618750 [4] NCCL INFO Channel 03/0 : 12[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618591:618746 [7] NCCL INFO Channel 03/0 : 63[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618585:618748 [1] NCCL INFO Channel 05/0 : 41[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:94349:94505 [7] NCCL INFO Channel 01/0 : 15[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618588:618750 [4] NCCL INFO Channel 05/0 : 12[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688206:688363 [6] NCCL INFO Channel 14/0 : 54[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688207:688366 [7] NCCL INFO Channel 04/0 : 55[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618591:618746 [7] NCCL INFO Channel 05/0 : 63[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688201:688364 [1] NCCL INFO Channel 15/0 : 41[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880680:880843 [1] NCCL INFO Channel 14/0 : 25[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618589:618744 [5] NCCL INFO Channel 01/0 : 61[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880682:880840 [3] NCCL INFO Channel 02/0 : 35[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618585:618748 [1] NCCL INFO Channel 07/0 : 41[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618588:618750 [4] NCCL INFO Channel 07/0 : 12[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618586:618745 [2] NCCL INFO Channel 01/0 : 10[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618591:618746 [7] NCCL INFO Channel 07/0 : 63[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618589:618744 [5] NCCL INFO Channel 03/0 : 61[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618588:618750 [4] NCCL INFO Channel 09/0 : 12[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618585:618748 [1] NCCL INFO Channel 11/0 : 41[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688203:688362 [3] NCCL INFO Channel 00/0 : 3[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688205:688365 [5] NCCL INFO Channel 02/0 : 53[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618591:618746 [7] NCCL INFO Channel 09/0 : 63[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618586:618745 [2] NCCL INFO Channel 03/0 : 10[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618589:618744 [5] NCCL INFO Channel 05/0 : 61[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618590:618749 [6] NCCL INFO Channel 01/0 : 62[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688202:688368 [2] NCCL INFO Channel 10/0 : 2[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:94349:94505 [7] NCCL INFO Channel 03/0 : 15[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880680:880843 [1] NCCL INFO Channel 00/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618584:618747 [0] NCCL INFO Channel 01/0 : 56[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688206:688363 [6] NCCL INFO Channel 00/0 : 38[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618588:618750 [4] NCCL INFO Channel 11/0 : 12[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618585:618748 [1] NCCL INFO Channel 13/0 : 41[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688207:688366 [7] NCCL INFO Channel 06/0 : 55[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618586:618745 [2] NCCL INFO Channel 05/0 : 10[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618591:618746 [7] NCCL INFO Channel 11/0 : 63[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618590:618749 [6] NCCL INFO Channel 03/0 : 62[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618589:618744 [5] NCCL INFO Channel 07/0 : 61[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618584:618747 [0] NCCL INFO Channel 03/0 : 56[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618587:618751 [3] NCCL INFO Channel 01/0 : 11[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618585:618748 [1] NCCL INFO Channel 15/0 : 41[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618588:618750 [4] NCCL INFO Channel 13/0 : 12[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618586:618745 [2] NCCL INFO Channel 07/0 : 10[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618590:618749 [6] NCCL INFO Channel 05/0 : 62[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880682:880840 [3] NCCL INFO Channel 04/0 : 35[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618587:618751 [3] NCCL INFO Channel 03/0 : 11[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618584:618747 [0] NCCL INFO Channel 05/0 : 56[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618591:618746 [7] NCCL INFO Channel 13/0 : 63[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618589:618744 [5] NCCL INFO Channel 09/0 : 61[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:94349:94505 [7] NCCL INFO Channel 05/0 : 15[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880680:880843 [1] NCCL INFO Channel 02/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618585:618748 [1] NCCL INFO Channel 01/0 : 25[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688205:688365 [5] NCCL INFO Channel 04/0 : 53[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618588:618750 [4] NCCL INFO Channel 15/0 : 12[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688204:688367 [4] NCCL INFO Channel 00/0 : 36[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618586:618745 [2] NCCL INFO Channel 09/0 : 10[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688202:688368 [2] NCCL INFO Channel 12/0 : 2[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618590:618749 [6] NCCL INFO Channel 07/0 : 62[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688206:688363 [6] NCCL INFO Channel 02/0 : 38[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618587:618751 [3] NCCL INFO Channel 05/0 : 11[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618589:618744 [5] NCCL INFO Channel 11/0 : 61[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880682:880840 [3] NCCL INFO Channel 06/0 : 35[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618591:618746 [7] NCCL INFO Channel 01/0 : 31[7] -> 63[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618584:618747 [0] NCCL INFO Channel 07/0 : 56[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618585:618748 [1] NCCL INFO Channel 03/0 : 25[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880680:880843 [1] NCCL INFO Channel 03/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618588:618750 [4] NCCL INFO Channel 01/0 : 28[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618586:618745 [2] NCCL INFO Channel 11/0 : 10[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:94349:94505 [7] NCCL INFO Channel 07/0 : 15[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618590:618749 [6] NCCL INFO Channel 09/0 : 62[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618587:618751 [3] NCCL INFO Channel 07/0 : 11[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688207:688366 [7] NCCL INFO Channel 08/0 : 55[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618589:618744 [5] NCCL INFO Channel 15/0 : 61[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688203:688362 [3] NCCL INFO Channel 02/0 : 3[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688205:688365 [5] NCCL INFO Channel 06/0 : 53[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880682:880840 [3] NCCL INFO Channel 08/0 : 35[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618585:618748 [1] NCCL INFO Channel 05/0 : 25[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688204:688367 [4] NCCL INFO Channel 02/0 : 36[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880680:880843 [1] NCCL INFO Channel 04/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618591:618746 [7] NCCL INFO Channel 03/0 : 31[7] -> 63[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618584:618747 [0] NCCL INFO Channel 09/0 : 56[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688202:688368 [2] NCCL INFO Channel 14/0 : 2[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618586:618745 [2] NCCL INFO Channel 13/0 : 10[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618588:618750 [4] NCCL INFO Channel 03/0 : 28[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688206:688363 [6] NCCL INFO Channel 04/0 : 38[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618590:618749 [6] NCCL INFO Channel 11/0 : 62[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618587:618751 [3] NCCL INFO Channel 09/0 : 11[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618589:618744 [5] NCCL INFO Channel 01/0 : 29[5] -> 61[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:94349:94505 [7] NCCL INFO Channel 09/0 : 15[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880682:880840 [3] NCCL INFO Channel 10/0 : 35[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618585:618748 [1] NCCL INFO Channel 07/0 : 25[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880680:880843 [1] NCCL INFO Channel 05/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618591:618746 [7] NCCL INFO Channel 05/0 : 31[7] -> 63[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618584:618747 [0] NCCL INFO Channel 11/0 : 56[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688207:688366 [7] NCCL INFO Channel 10/0 : 55[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618586:618745 [2] NCCL INFO Channel 15/0 : 10[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618587:618751 [3] NCCL INFO Channel 13/0 : 11[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618588:618750 [4] NCCL INFO Channel 05/0 : 28[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688204:688367 [4] NCCL INFO Channel 06/0 : 36[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688205:688365 [5] NCCL INFO Channel 08/0 : 53[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618590:618749 [6] NCCL INFO Channel 13/0 : 62[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688203:688362 [3] NCCL INFO Channel 04/0 : 3[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618589:618744 [5] NCCL INFO Channel 03/0 : 29[5] -> 61[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688202:688368 [2] NCCL INFO Channel 00/0 : 34[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618585:618748 [1] NCCL INFO Channel 11/0 : 25[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688206:688363 [6] NCCL INFO Channel 08/0 : 38[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:94349:94505 [7] NCCL INFO Channel 11/0 : 15[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880682:880840 [3] NCCL INFO Channel 12/0 : 35[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880680:880843 [1] NCCL INFO Channel 06/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688207:688366 [7] NCCL INFO Channel 12/0 : 55[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618591:618746 [7] NCCL INFO Channel 07/0 : 31[7] -> 63[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618584:618747 [0] NCCL INFO Channel 13/0 : 56[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688204:688367 [4] NCCL INFO Channel 08/0 : 36[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618586:618745 [2] NCCL INFO Channel 01/0 : 26[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688205:688365 [5] NCCL INFO Channel 10/0 : 53[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880680:880843 [1] NCCL INFO Channel 07/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618588:618750 [4] NCCL INFO Channel 07/0 : 28[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618587:618751 [3] NCCL INFO Channel 15/0 : 11[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688203:688362 [3] NCCL INFO Channel 06/0 : 3[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688202:688368 [2] NCCL INFO Channel 04/0 : 34[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618590:618749 [6] NCCL INFO Channel 15/0 : 62[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688206:688363 [6] NCCL INFO Channel 10/0 : 38[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688207:688366 [7] NCCL INFO Channel 14/0 : 55[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618585:618748 [1] NCCL INFO Channel 13/0 : 25[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618589:618744 [5] NCCL INFO Channel 05/0 : 29[5] -> 61[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688204:688367 [4] NCCL INFO Channel 10/0 : 36[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618591:618746 [7] NCCL INFO Channel 09/0 : 31[7] -> 63[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688205:688365 [5] NCCL INFO Channel 12/0 : 53[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:94347:94507 [5] NCCL INFO Channel 01/0 : 13[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:94349:94505 [7] NCCL INFO Channel 13/0 : 15[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618584:618747 [0] NCCL INFO Channel 15/0 : 56[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880682:880840 [3] NCCL INFO Channel 14/0 : 35[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618586:618745 [2] NCCL INFO Channel 03/0 : 26[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618587:618751 [3] NCCL INFO Channel 01/0 : 27[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618588:618750 [4] NCCL INFO Channel 09/0 : 28[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618590:618749 [6] NCCL INFO Channel 01/0 : 30[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618589:618744 [5] NCCL INFO Channel 07/0 : 29[5] -> 61[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688202:688368 [2] NCCL INFO Channel 06/0 : 34[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688203:688362 [3] NCCL INFO Channel 08/0 : 3[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618585:618748 [1] NCCL INFO Channel 15/0 : 25[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688206:688363 [6] NCCL INFO Channel 12/0 : 38[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618584:618747 [0] NCCL INFO Channel 01/0 : 24[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618591:618746 [7] NCCL INFO Channel 11/0 : 31[7] -> 63[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618586:618745 [2] NCCL INFO Channel 05/0 : 26[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688207:688366 [7] NCCL INFO Channel 00/0 : 39[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618587:618751 [3] NCCL INFO Channel 03/0 : 27[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688204:688367 [4] NCCL INFO Channel 12/0 : 36[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688205:688365 [5] NCCL INFO Channel 14/0 : 53[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880680:880843 [1] NCCL INFO Channel 08/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618590:618749 [6] NCCL INFO Channel 03/0 : 30[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618588:618750 [4] NCCL INFO Channel 11/0 : 28[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:94347:94507 [5] NCCL INFO Channel 03/0 : 13[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO Channel 00/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618589:618744 [5] NCCL INFO Channel 09/0 : 29[5] -> 61[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618586:618745 [2] NCCL INFO Channel 07/0 : 26[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880680:880843 [1] NCCL INFO Channel 10/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618587:618751 [3] NCCL INFO Channel 05/0 : 27[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688203:688362 [3] NCCL INFO Channel 10/0 : 3[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618584:618747 [0] NCCL INFO Channel 03/0 : 24[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618591:618746 [7] NCCL INFO Channel 13/0 : 31[7] -> 63[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688206:688363 [6] NCCL INFO Channel 14/0 : 38[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618588:618750 [4] NCCL INFO Channel 13/0 : 28[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618589:618744 [5] NCCL INFO Channel 11/0 : 29[5] -> 61[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618590:618749 [6] NCCL INFO Channel 05/0 : 30[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688207:688366 [7] NCCL INFO Channel 02/0 : 39[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-7:811255:811411 [7] NCCL INFO Channel 00/0 : 63[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO Channel 02/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618586:618745 [2] NCCL INFO Channel 09/0 : 26[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880680:880843 [1] NCCL INFO Channel 11/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:94347:94507 [5] NCCL INFO Channel 05/0 : 13[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688204:688367 [4] NCCL INFO Channel 14/0 : 36[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO Channel 03/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618587:618751 [3] NCCL INFO Channel 07/0 : 27[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688205:688365 [5] NCCL INFO Channel 00/0 : 37[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-7:811255:811411 [7] NCCL INFO Channel 02/0 : 63[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880680:880843 [1] NCCL INFO Channel 12/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618584:618747 [0] NCCL INFO Channel 05/0 : 24[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688203:688362 [3] NCCL INFO Channel 12/0 : 3[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618589:618744 [5] NCCL INFO Channel 15/0 : 29[5] -> 61[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688202:688368 [2] NCCL INFO Channel 08/0 : 34[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:94347:94507 [5] NCCL INFO Channel 07/0 : 13[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942618:942778 [7] NCCL INFO Channel 01/0 : 47[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618588:618750 [4] NCCL INFO Channel 15/0 : 28[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618590:618749 [6] NCCL INFO Channel 07/0 : 30[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688207:688366 [7] NCCL INFO Channel 04/0 : 39[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO Channel 01/0 : 28[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942618:942778 [7] NCCL INFO Channel 03/0 : 47[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618586:618745 [2] NCCL INFO Channel 11/0 : 26[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688205:688365 [5] NCCL INFO Channel 02/0 : 37[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO Channel 04/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942616:942777 [5] NCCL INFO Channel 01/0 : 45[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880680:880843 [1] NCCL INFO Channel 13/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618584:618747 [0] NCCL INFO Channel 07/0 : 24[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618587:618751 [3] NCCL INFO Channel 09/0 : 27[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-7:811255:811411 [7] NCCL INFO Channel 04/0 : 63[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942618:942778 [7] NCCL INFO Channel 05/0 : 47[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618585:618748 [1] NCCL INFO Channel 00/0 : 25[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942616:942777 [5] NCCL INFO Channel 03/0 : 45[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618590:618749 [6] NCCL INFO Channel 09/0 : 30[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618586:618745 [2] NCCL INFO Channel 13/0 : 26[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:94347:94507 [5] NCCL INFO Channel 09/0 : 13[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942618:942778 [7] NCCL INFO Channel 07/0 : 47[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618584:618747 [0] NCCL INFO Channel 09/0 : 24[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618587:618751 [3] NCCL INFO Channel 13/0 : 27[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618591:618746 [7] NCCL INFO Channel 01/0 : 47[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688202:688368 [2] NCCL INFO Channel 10/0 : 34[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688203:688362 [3] NCCL INFO Channel 14/0 : 3[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO Channel 03/0 : 28[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942616:942777 [5] NCCL INFO Channel 05/0 : 45[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880680:880843 [1] NCCL INFO Channel 14/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618585:618748 [1] NCCL INFO Channel 02/0 : 25[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688207:688366 [7] NCCL INFO Channel 06/0 : 39[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-7:811255:811411 [7] NCCL INFO Channel 06/0 : 63[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942618:942778 [7] NCCL INFO Channel 09/0 : 47[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688205:688365 [5] NCCL INFO Channel 04/0 : 37[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942616:942777 [5] NCCL INFO Channel 07/0 : 45[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688202:688368 [2] NCCL INFO Channel 12/0 : 34[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688203:688362 [3] NCCL INFO Channel 00/0 : 35[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:94347:94507 [5] NCCL INFO Channel 11/0 : 13[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942617:942774 [6] NCCL INFO Channel 01/0 : 46[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688207:688366 [7] NCCL INFO Channel 08/0 : 39[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO Channel 05/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942618:942778 [7] NCCL INFO Channel 11/0 : 47[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880680:880843 [1] NCCL INFO Channel 15/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618586:618745 [2] NCCL INFO Channel 15/0 : 26[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688205:688365 [5] NCCL INFO Channel 06/0 : 37[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-7:811255:811411 [7] NCCL INFO Channel 08/0 : 63[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO Channel 05/0 : 28[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942616:942777 [5] NCCL INFO Channel 09/0 : 45[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926595:926760 [1] NCCL INFO Channel 02/0 : 57[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618590:618749 [6] NCCL INFO Channel 11/0 : 30[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688202:688368 [2] NCCL INFO Channel 14/0 : 34[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:94346:94511 [4] NCCL INFO Channel 01/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942617:942774 [6] NCCL INFO Channel 03/0 : 46[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926595:926760 [1] NCCL INFO Channel 04/0 : 57[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618587:618751 [3] NCCL INFO Channel 15/0 : 27[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:94347:94507 [5] NCCL INFO Channel 15/0 : 13[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942618:942778 [7] NCCL INFO Channel 13/0 : 47[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880685:880847 [6] NCCL INFO Channel 00/0 : 30[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926595:926760 [1] NCCL INFO Channel 06/0 : 57[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618584:618747 [0] NCCL INFO Channel 11/0 : 24[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-7:811255:811411 [7] NCCL INFO Channel 10/0 : 63[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO Channel 06/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942611:942772 [0] NCCL INFO Channel 01/0 : 40[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942616:942777 [5] NCCL INFO Channel 11/0 : 45[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942617:942774 [6] NCCL INFO Channel 05/0 : 46[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926595:926760 [1] NCCL INFO Channel 08/0 : 57[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618591:618746 [7] NCCL INFO Channel 03/0 : 47[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688203:688362 [3] NCCL INFO Channel 02/0 : 35[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:94346:94511 [4] NCCL INFO Channel 03/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO Channel 07/0 : 28[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942616:942777 [5] NCCL INFO Channel 15/0 : 45[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926595:926760 [1] NCCL INFO Channel 10/0 : 57[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618589:618744 [5] NCCL INFO Channel 01/0 : 45[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618585:618748 [1] NCCL INFO Channel 04/0 : 25[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688207:688366 [7] NCCL INFO Channel 10/0 : 39[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO Channel 01/0 : 26[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO Channel 01/0 : 27[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO Channel 07/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942617:942774 [6] NCCL INFO Channel 07/0 : 46[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880685:880847 [6] NCCL INFO Channel 02/0 : 30[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926595:926760 [1] NCCL INFO Channel 12/0 : 57[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618590:618749 [6] NCCL INFO Channel 13/0 : 30[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688204:688367 [4] NCCL INFO Channel 00/0 : 52[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-7:811255:811411 [7] NCCL INFO Channel 12/0 : 63[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942611:942772 [0] NCCL INFO Channel 03/0 : 40[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618584:618747 [0] NCCL INFO Channel 13/0 : 24[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688205:688365 [5] NCCL INFO Channel 08/0 : 37[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942617:942774 [6] NCCL INFO Channel 09/0 : 46[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618591:618746 [7] NCCL INFO Channel 05/0 : 47[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688203:688362 [3] NCCL INFO Channel 04/0 : 35[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:94344:94510 [2] NCCL INFO Channel 01/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942611:942772 [0] NCCL INFO Channel 05/0 : 40[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618585:618748 [1] NCCL INFO Channel 06/0 : 25[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618589:618744 [5] NCCL INFO Channel 03/0 : 45[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688207:688366 [7] NCCL INFO Channel 12/0 : 39[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:94346:94511 [4] NCCL INFO Channel 05/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO Channel 09/0 : 28[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942617:942774 [6] NCCL INFO Channel 11/0 : 46[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880685:880847 [6] NCCL INFO Channel 04/0 : 30[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926595:926760 [1] NCCL INFO Channel 14/0 : 57[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618590:618749 [6] NCCL INFO Channel 15/0 : 30[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688206:688363 [6] NCCL INFO Channel 01/0 : 46[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-7:811255:811411 [7] NCCL INFO Channel 14/0 : 63[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-7:811253:811410 [5] NCCL INFO Channel 00/0 : 61[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO Channel 08/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO Channel 03/0 : 26[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942611:942772 [0] NCCL INFO Channel 07/0 : 40[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926595:926760 [1] NCCL INFO Channel 00/0 : 49[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618584:618747 [0] NCCL INFO Channel 15/0 : 24[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688204:688367 [4] NCCL INFO Channel 02/0 : 52[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688205:688365 [5] NCCL INFO Channel 10/0 : 37[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO Channel 03/0 : 27[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942617:942774 [6] NCCL INFO Channel 13/0 : 46[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926595:926760 [1] NCCL INFO Channel 02/0 : 49[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618588:618750 [4] NCCL INFO Channel 01/0 : 60[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688203:688362 [3] NCCL INFO Channel 06/0 : 35[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942611:942772 [0] NCCL INFO Channel 09/0 : 40[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926595:926760 [1] NCCL INFO Channel 03/0 : 49[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618591:618746 [7] NCCL INFO Channel 07/0 : 47[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688207:688366 [7] NCCL INFO Channel 14/0 : 39[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:94344:94510 [2] NCCL INFO Channel 03/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942617:942774 [6] NCCL INFO Channel 15/0 : 46[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880685:880847 [6] NCCL INFO Channel 08/0 : 30[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926595:926760 [1] NCCL INFO Channel 04/0 : 49[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688206:688363 [6] NCCL INFO Channel 03/0 : 46[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:94346:94511 [4] NCCL INFO Channel 07/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-7:811254:811414 [6] NCCL INFO Channel 00/0 : 62[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-7:811248:811415 [0] NCCL INFO Channel 02/0 : 56[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO Channel 11/0 : 28[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942611:942772 [0] NCCL INFO Channel 11/0 : 40[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926595:926760 [1] NCCL INFO Channel 05/0 : 49[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO Channel 10/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942611:942772 [0] NCCL INFO Channel 13/0 : 40[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926595:926760 [1] NCCL INFO Channel 06/0 : 49[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618589:618744 [5] NCCL INFO Channel 05/0 : 45[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618585:618748 [1] NCCL INFO Channel 08/0 : 25[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO Channel 05/0 : 26[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO Channel 05/0 : 27[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942611:942772 [0] NCCL INFO Channel 15/0 : 40[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926595:926760 [1] NCCL INFO Channel 07/0 : 49[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618588:618750 [4] NCCL INFO Channel 03/0 : 60[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO Channel 13/0 : 28[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880685:880847 [6] NCCL INFO Channel 10/0 : 30[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926595:926760 [1] NCCL INFO Channel 08/0 : 49[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618591:618746 [7] NCCL INFO Channel 09/0 : 47[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:94344:94510 [2] NCCL INFO Channel 05/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926595:926760 [1] NCCL INFO Channel 10/0 : 49[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618586:618745 [2] NCCL INFO Channel 01/0 : 58[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688205:688365 [5] NCCL INFO Channel 12/0 : 37[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688204:688367 [4] NCCL INFO Channel 06/0 : 52[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:94346:94511 [4] NCCL INFO Channel 09/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-7:811253:811410 [5] NCCL INFO Channel 02/0 : 61[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926595:926760 [1] NCCL INFO Channel 11/0 : 49[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618589:618744 [5] NCCL INFO Channel 07/0 : 45[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688202:688368 [2] NCCL INFO Channel 00/0 : 50[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO Channel 11/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO Channel 07/0 : 26[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880685:880847 [6] NCCL INFO Channel 12/0 : 30[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926595:926760 [1] NCCL INFO Channel 12/0 : 49[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618588:618750 [4] NCCL INFO Channel 05/0 : 60[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688203:688362 [3] NCCL INFO Channel 08/0 : 35[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO Channel 07/0 : 27[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926595:926760 [1] NCCL INFO Channel 13/0 : 49[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618585:618748 [1] NCCL INFO Channel 10/0 : 25[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618591:618746 [7] NCCL INFO Channel 11/0 : 47[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688206:688363 [6] NCCL INFO Channel 05/0 : 46[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:94344:94510 [2] NCCL INFO Channel 07/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO Channel 15/0 : 28[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926595:926760 [1] NCCL INFO Channel 14/0 : 49[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618587:618751 [3] NCCL INFO Channel 01/0 : 59[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688204:688367 [4] NCCL INFO Channel 08/0 : 52[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-7:811248:811415 [0] NCCL INFO Channel 04/0 : 56[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-7:811254:811414 [6] NCCL INFO Channel 02/0 : 62[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926595:926760 [1] NCCL INFO Channel 15/0 : 49[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618586:618745 [2] NCCL INFO Channel 03/0 : 58[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688202:688368 [2] NCCL INFO Channel 04/0 : 50[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:94346:94511 [4] NCCL INFO Channel 11/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-7:811253:811410 [5] NCCL INFO Channel 04/0 : 61[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926600:926757 [6] NCCL INFO Channel 00/0 : 62[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618584:618747 [0] NCCL INFO Channel 01/0 : 40[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618590:618749 [6] NCCL INFO Channel 01/0 : 46[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618589:618744 [5] NCCL INFO Channel 09/0 : 45[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688205:688365 [5] NCCL INFO Channel 14/0 : 37[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880685:880847 [6] NCCL INFO Channel 14/0 : 30[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618588:618750 [4] NCCL INFO Channel 07/0 : 60[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688203:688362 [3] NCCL INFO Channel 10/0 : 35[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO Channel 09/0 : 26[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688206:688363 [6] NCCL INFO Channel 07/0 : 46[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:94344:94510 [2] NCCL INFO Channel 09/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO Channel 12/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688204:688367 [4] NCCL INFO Channel 10/0 : 52[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO Channel 09/0 : 27[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688202:688368 [2] NCCL INFO Channel 06/0 : 50[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-7:811248:811415 [0] NCCL INFO Channel 06/0 : 56[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-7:811254:811414 [6] NCCL INFO Channel 04/0 : 62[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688203:688362 [3] NCCL INFO Channel 12/0 : 35[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:94346:94511 [4] NCCL INFO Channel 13/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-7:811253:811410 [5] NCCL INFO Channel 06/0 : 61[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880686:880841 [7] NCCL INFO Channel 00/0 : 31[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618585:618748 [1] NCCL INFO Channel 12/0 : 25[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688206:688363 [6] NCCL INFO Channel 09/0 : 46[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926599:926755 [5] NCCL INFO Channel 00/0 : 61[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618587:618751 [3] NCCL INFO Channel 03/0 : 59[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618591:618746 [7] NCCL INFO Channel 13/0 : 47[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO Channel 11/0 : 26[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO Channel 13/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618586:618745 [2] NCCL INFO Channel 05/0 : 58[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:94344:94510 [2] NCCL INFO Channel 11/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO Channel 13/0 : 27[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618584:618747 [0] NCCL INFO Channel 03/0 : 40[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:94346:94511 [4] NCCL INFO Channel 15/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-7:811248:811415 [0] NCCL INFO Channel 08/0 : 56[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO Channel 13/0 : 26[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880685:880847 [6] NCCL INFO Channel 00/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618589:618744 [5] NCCL INFO Channel 11/0 : 45[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618590:618749 [6] NCCL INFO Channel 03/0 : 46[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618588:618750 [4] NCCL INFO Channel 09/0 : 60[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688204:688367 [4] NCCL INFO Channel 12/0 : 52[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880686:880841 [7] NCCL INFO Channel 02/0 : 31[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926599:926755 [5] NCCL INFO Channel 02/0 : 61[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618585:618748 [1] NCCL INFO Channel 14/0 : 25[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688200:688369 [0] NCCL INFO Channel 02/0 : 0[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618591:618746 [7] NCCL INFO Channel 01/0 : 31[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618587:618751 [3] NCCL INFO Channel 05/0 : 59[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688207:688366 [7] NCCL INFO Channel 01/0 : 47[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618586:618745 [2] NCCL INFO Channel 07/0 : 58[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688200:688369 [0] NCCL INFO Channel 04/0 : 0[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-7:811254:811414 [6] NCCL INFO Channel 08/0 : 62[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-7:811253:811410 [5] NCCL INFO Channel 08/0 : 61[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618584:618747 [0] NCCL INFO Channel 05/0 : 40[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688207:688366 [7] NCCL INFO Channel 03/0 : 47[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:94344:94510 [2] NCCL INFO Channel 13/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688200:688369 [0] NCCL INFO Channel 06/0 : 0[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO Channel 14/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880685:880847 [6] NCCL INFO Channel 01/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:94343:94508 [1] NCCL INFO Channel 01/0 : 9[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO Channel 15/0 : 27[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880686:880841 [7] NCCL INFO Channel 04/0 : 31[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926599:926755 [5] NCCL INFO Channel 04/0 : 61[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-7:811248:811415 [0] NCCL INFO Channel 10/0 : 56[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO Channel 15/0 : 26[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880684:880842 [5] NCCL INFO Channel 00/0 : 29[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618589:618744 [5] NCCL INFO Channel 15/0 : 45[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618590:618749 [6] NCCL INFO Channel 05/0 : 46[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618588:618750 [4] NCCL INFO Channel 11/0 : 60[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-7:811254:811414 [6] NCCL INFO Channel 10/0 : 62[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-7:811253:811410 [5] NCCL INFO Channel 10/0 : 61[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO Channel 15/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880685:880847 [6] NCCL INFO Channel 02/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618591:618746 [7] NCCL INFO Channel 03/0 : 31[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688207:688366 [7] NCCL INFO Channel 05/0 : 47[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:94344:94510 [2] NCCL INFO Channel 15/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618586:618745 [2] NCCL INFO Channel 09/0 : 58[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688205:688365 [5] NCCL INFO Channel 01/0 : 45[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926600:926757 [6] NCCL INFO Channel 02/0 : 62[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618587:618751 [3] NCCL INFO Channel 07/0 : 59[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688200:688369 [0] NCCL INFO Channel 08/0 : 0[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO Channel 01/0 : 9[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618584:618747 [0] NCCL INFO Channel 07/0 : 40[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688207:688366 [7] NCCL INFO Channel 07/0 : 47[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:94343:94508 [1] NCCL INFO Channel 03/0 : 9[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-7:811248:811415 [0] NCCL INFO Channel 12/0 : 56[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO Channel 03/0 : 9[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880686:880841 [7] NCCL INFO Channel 06/0 : 31[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618589:618744 [5] NCCL INFO Channel 01/0 : 29[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688205:688365 [5] NCCL INFO Channel 03/0 : 45[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-7:811253:811410 [5] NCCL INFO Channel 12/0 : 61[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-7:811254:811414 [6] NCCL INFO Channel 12/0 : 62[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880684:880842 [5] NCCL INFO Channel 02/0 : 29[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618588:618750 [4] NCCL INFO Channel 13/0 : 60[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618590:618749 [6] NCCL INFO Channel 07/0 : 46[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880685:880847 [6] NCCL INFO Channel 03/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926599:926755 [5] NCCL INFO Channel 06/0 : 61[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618591:618746 [7] NCCL INFO Channel 05/0 : 31[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880686:880841 [7] NCCL INFO Channel 08/0 : 31[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926600:926757 [6] NCCL INFO Channel 04/0 : 62[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618586:618745 [2] NCCL INFO Channel 11/0 : 58[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618587:618751 [3] NCCL INFO Channel 09/0 : 59[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:94343:94508 [1] NCCL INFO Channel 05/0 : 9[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO Channel 05/0 : 9[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618584:618747 [0] NCCL INFO Channel 09/0 : 40[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688200:688369 [0] NCCL INFO Channel 10/0 : 0[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-7:811248:811415 [0] NCCL INFO Channel 14/0 : 56[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618589:618744 [5] NCCL INFO Channel 03/0 : 29[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688207:688366 [7] NCCL INFO Channel 09/0 : 47[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688205:688365 [5] NCCL INFO Channel 05/0 : 45[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-7:811253:811410 [5] NCCL INFO Channel 14/0 : 61[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-7:811254:811414 [6] NCCL INFO Channel 14/0 : 62[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880684:880842 [5] NCCL INFO Channel 04/0 : 29[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880685:880847 [6] NCCL INFO Channel 04/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688200:688369 [0] NCCL INFO Channel 12/0 : 0[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880686:880841 [7] NCCL INFO Channel 10/0 : 31[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926599:926755 [5] NCCL INFO Channel 08/0 : 61[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO Channel 07/0 : 9[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926600:926757 [6] NCCL INFO Channel 08/0 : 62[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:94343:94508 [1] NCCL INFO Channel 07/0 : 9[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618588:618750 [4] NCCL INFO Channel 15/0 : 60[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618590:618749 [6] NCCL INFO Channel 09/0 : 46[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618591:618746 [7] NCCL INFO Channel 07/0 : 31[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688207:688366 [7] NCCL INFO Channel 11/0 : 47[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688205:688365 [5] NCCL INFO Channel 07/0 : 45[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880684:880842 [5] NCCL INFO Channel 06/0 : 29[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880685:880847 [6] NCCL INFO Channel 05/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618586:618745 [2] NCCL INFO Channel 13/0 : 58[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618587:618751 [3] NCCL INFO Channel 13/0 : 59[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688200:688369 [0] NCCL INFO Channel 14/0 : 0[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO Channel 11/0 : 9[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880686:880841 [7] NCCL INFO Channel 12/0 : 31[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926599:926755 [5] NCCL INFO Channel 10/0 : 61[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618584:618747 [0] NCCL INFO Channel 11/0 : 40[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926600:926757 [6] NCCL INFO Channel 10/0 : 62[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618589:618744 [5] NCCL INFO Channel 05/0 : 29[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926601:926759 [7] NCCL INFO Channel 00/0 : 63[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618588:618750 [4] NCCL INFO Channel 01/0 : 28[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618590:618749 [6] NCCL INFO Channel 11/0 : 46[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:94343:94508 [1] NCCL INFO Channel 11/0 : 9[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618591:618746 [7] NCCL INFO Channel 09/0 : 31[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618586:618745 [2] NCCL INFO Channel 15/0 : 58[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688207:688366 [7] NCCL INFO Channel 13/0 : 47[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688205:688365 [5] NCCL INFO Channel 09/0 : 45[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO Channel 13/0 : 9[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880685:880847 [6] NCCL INFO Channel 07/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618587:618751 [3] NCCL INFO Channel 15/0 : 59[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688200:688369 [0] NCCL INFO Channel 02/0 : 32[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880684:880842 [5] NCCL INFO Channel 08/0 : 29[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926599:926755 [5] NCCL INFO Channel 12/0 : 61[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880686:880841 [7] NCCL INFO Channel 14/0 : 31[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926600:926757 [6] NCCL INFO Channel 12/0 : 62[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:94343:94508 [1] NCCL INFO Channel 13/0 : 9[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO Channel 15/0 : 9[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618589:618744 [5] NCCL INFO Channel 07/0 : 29[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618584:618747 [0] NCCL INFO Channel 13/0 : 40[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688205:688365 [5] NCCL INFO Channel 11/0 : 45[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688200:688369 [0] NCCL INFO Channel 04/0 : 32[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618588:618750 [4] NCCL INFO Channel 03/0 : 28[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880685:880847 [6] NCCL INFO Channel 08/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618590:618749 [6] NCCL INFO Channel 13/0 : 46[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618591:618746 [7] NCCL INFO Channel 11/0 : 31[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880684:880842 [5] NCCL INFO Channel 10/0 : 29[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926601:926759 [7] NCCL INFO Channel 02/0 : 63[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618586:618745 [2] NCCL INFO Channel 01/0 : 26[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880686:880841 [7] NCCL INFO Channel 00/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926599:926755 [5] NCCL INFO Channel 14/0 : 61[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926600:926757 [6] NCCL INFO Channel 14/0 : 62[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618587:618751 [3] NCCL INFO Channel 01/0 : 27[3] -> 59[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:94343:94508 [1] NCCL INFO Channel 15/0 : 9[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880685:880847 [6] NCCL INFO Channel 09/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926601:926759 [7] NCCL INFO Channel 04/0 : 63[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618589:618744 [5] NCCL INFO Channel 09/0 : 29[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618584:618747 [0] NCCL INFO Channel 15/0 : 40[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688205:688365 [5] NCCL INFO Channel 15/0 : 45[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618588:618750 [4] NCCL INFO Channel 05/0 : 28[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688200:688369 [0] NCCL INFO Channel 06/0 : 32[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618591:618746 [7] NCCL INFO Channel 13/0 : 31[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618586:618745 [2] NCCL INFO Channel 03/0 : 26[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618590:618749 [6] NCCL INFO Channel 15/0 : 46[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618587:618751 [3] NCCL INFO Channel 03/0 : 27[3] -> 59[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880684:880842 [5] NCCL INFO Channel 12/0 : 29[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880686:880841 [7] NCCL INFO Channel 01/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926599:926755 [5] NCCL INFO Channel 00/0 : 53[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926600:926757 [6] NCCL INFO Channel 00/0 : 54[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880685:880847 [6] NCCL INFO Channel 10/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926601:926759 [7] NCCL INFO Channel 06/0 : 63[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688200:688369 [0] NCCL INFO Channel 08/0 : 32[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880684:880842 [5] NCCL INFO Channel 14/0 : 29[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880686:880841 [7] NCCL INFO Channel 02/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618589:618744 [5] NCCL INFO Channel 11/0 : 29[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618584:618747 [0] NCCL INFO Channel 01/0 : 24[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880685:880847 [6] NCCL INFO Channel 11/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926600:926757 [6] NCCL INFO Channel 01/0 : 54[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926599:926755 [5] NCCL INFO Channel 01/0 : 53[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618588:618750 [4] NCCL INFO Channel 07/0 : 28[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688200:688369 [0] NCCL INFO Channel 10/0 : 32[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618587:618751 [3] NCCL INFO Channel 05/0 : 27[3] -> 59[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618590:618749 [6] NCCL INFO Channel 01/0 : 30[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618586:618745 [2] NCCL INFO Channel 05/0 : 26[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618584:618747 [0] NCCL INFO Channel 03/0 : 24[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618589:618744 [5] NCCL INFO Channel 15/0 : 29[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618588:618750 [4] NCCL INFO Channel 09/0 : 28[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880684:880842 [5] NCCL INFO Channel 00/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880686:880841 [7] NCCL INFO Channel 03/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880685:880847 [6] NCCL INFO Channel 12/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926601:926759 [7] NCCL INFO Channel 08/0 : 63[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688200:688369 [0] NCCL INFO Channel 12/0 : 32[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618587:618751 [3] NCCL INFO Channel 07/0 : 27[3] -> 59[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618586:618745 [2] NCCL INFO Channel 07/0 : 26[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618590:618749 [6] NCCL INFO Channel 03/0 : 30[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618584:618747 [0] NCCL INFO Channel 05/0 : 24[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880684:880842 [5] NCCL INFO Channel 01/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880686:880841 [7] NCCL INFO Channel 04/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880685:880847 [6] NCCL INFO Channel 13/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618588:618750 [4] NCCL INFO Channel 11/0 : 28[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926600:926757 [6] NCCL INFO Channel 02/0 : 54[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926599:926755 [5] NCCL INFO Channel 02/0 : 53[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688200:688369 [0] NCCL INFO Channel 14/0 : 32[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926601:926759 [7] NCCL INFO Channel 10/0 : 63[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618586:618745 [2] NCCL INFO Channel 09/0 : 26[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880686:880841 [7] NCCL INFO Channel 05/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926600:926757 [6] NCCL INFO Channel 03/0 : 54[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926599:926755 [5] NCCL INFO Channel 03/0 : 53[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926601:926759 [7] NCCL INFO Channel 12/0 : 63[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618584:618747 [0] NCCL INFO Channel 07/0 : 24[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618590:618749 [6] NCCL INFO Channel 05/0 : 30[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO Channel 00/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618588:618750 [4] NCCL INFO Channel 13/0 : 28[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880685:880847 [6] NCCL INFO Channel 15/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880684:880842 [5] NCCL INFO Channel 02/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618587:618751 [3] NCCL INFO Channel 09/0 : 27[3] -> 59[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618591:618746 [7] NCCL INFO Channel 00/0 : 31[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926600:926757 [6] NCCL INFO Channel 04/0 : 54[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926599:926755 [5] NCCL INFO Channel 04/0 : 53[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926601:926759 [7] NCCL INFO Channel 14/0 : 63[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO Channel 01/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880686:880841 [7] NCCL INFO Channel 06/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880684:880842 [5] NCCL INFO Channel 03/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618586:618745 [2] NCCL INFO Channel 11/0 : 26[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618584:618747 [0] NCCL INFO Channel 09/0 : 24[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926600:926757 [6] NCCL INFO Channel 05/0 : 54[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926599:926755 [5] NCCL INFO Channel 06/0 : 53[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO Channel 00/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926601:926759 [7] NCCL INFO Channel 00/0 : 55[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688200:688369 [0] NCCL INFO Channel 02/0 : 48[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880686:880841 [7] NCCL INFO Channel 08/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880684:880842 [5] NCCL INFO Channel 04/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618590:618749 [6] NCCL INFO Channel 07/0 : 30[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618588:618750 [4] NCCL INFO Channel 15/0 : 28[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618587:618751 [3] NCCL INFO Channel 13/0 : 27[3] -> 59[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO Channel 02/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926600:926757 [6] NCCL INFO Channel 07/0 : 54[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618586:618745 [2] NCCL INFO Channel 13/0 : 26[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926599:926755 [5] NCCL INFO Channel 07/0 : 53[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618591:618746 [7] NCCL INFO Channel 02/0 : 31[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618589:618744 [5] NCCL INFO Channel 00/0 : 29[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880686:880841 [7] NCCL INFO Channel 09/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688200:688369 [0] NCCL INFO Channel 04/0 : 48[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880684:880842 [5] NCCL INFO Channel 06/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO Channel 01/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO Channel 03/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926601:926759 [7] NCCL INFO Channel 01/0 : 55[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926600:926757 [6] NCCL INFO Channel 08/0 : 54[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618584:618747 [0] NCCL INFO Channel 11/0 : 24[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926599:926755 [5] NCCL INFO Channel 08/0 : 53[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618590:618749 [6] NCCL INFO Channel 09/0 : 30[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618587:618751 [3] NCCL INFO Channel 15/0 : 27[3] -> 59[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880686:880841 [7] NCCL INFO Channel 10/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618586:618745 [2] NCCL INFO Channel 15/0 : 26[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688200:688369 [0] NCCL INFO Channel 06/0 : 48[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO Channel 02/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880684:880842 [5] NCCL INFO Channel 07/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:94342:94504 [0] NCCL INFO Channel 01/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO Channel 04/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926601:926759 [7] NCCL INFO Channel 02/0 : 55[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880686:880841 [7] NCCL INFO Channel 11/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926600:926757 [6] NCCL INFO Channel 09/0 : 54[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618591:618746 [7] NCCL INFO Channel 04/0 : 31[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688200:688369 [0] NCCL INFO Channel 08/0 : 48[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO Channel 03/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880684:880842 [5] NCCL INFO Channel 08/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926599:926755 [5] NCCL INFO Channel 09/0 : 53[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618589:618744 [5] NCCL INFO Channel 02/0 : 29[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926601:926759 [7] NCCL INFO Channel 03/0 : 55[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618584:618747 [0] NCCL INFO Channel 13/0 : 24[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618590:618749 [6] NCCL INFO Channel 11/0 : 30[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618591:618746 [7] NCCL INFO Channel 06/0 : 31[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-7:811252:811408 [4] NCCL INFO Channel 00/0 : 60[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:94342:94504 [0] NCCL INFO Channel 03/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO Channel 05/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880686:880841 [7] NCCL INFO Channel 12/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880684:880842 [5] NCCL INFO Channel 09/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926600:926757 [6] NCCL INFO Channel 10/0 : 54[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688200:688369 [0] NCCL INFO Channel 10/0 : 48[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926599:926755 [5] NCCL INFO Channel 10/0 : 53[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618589:618744 [5] NCCL INFO Channel 04/0 : 29[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618588:618750 [4] NCCL INFO Channel 01/0 : 44[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618590:618749 [6] NCCL INFO Channel 13/0 : 30[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618584:618747 [0] NCCL INFO Channel 15/0 : 24[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926601:926759 [7] NCCL INFO Channel 04/0 : 55[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618587:618751 [3] NCCL INFO Channel 01/0 : 43[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926600:926757 [6] NCCL INFO Channel 11/0 : 54[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618591:618746 [7] NCCL INFO Channel 08/0 : 31[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-7:811252:811408 [4] NCCL INFO Channel 02/0 : 60[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO Channel 04/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO Channel 06/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880686:880841 [7] NCCL INFO Channel 13/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:94342:94504 [0] NCCL INFO Channel 05/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880684:880842 [5] NCCL INFO Channel 10/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688200:688369 [0] NCCL INFO Channel 12/0 : 48[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926599:926755 [5] NCCL INFO Channel 11/0 : 53[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926601:926759 [7] NCCL INFO Channel 05/0 : 55[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618586:618745 [2] NCCL INFO Channel 01/0 : 42[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-7:811250:811412 [2] NCCL INFO Channel 00/0 : 58[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926600:926757 [6] NCCL INFO Channel 12/0 : 54[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618589:618744 [5] NCCL INFO Channel 06/0 : 29[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618590:618749 [6] NCCL INFO Channel 15/0 : 30[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618588:618750 [4] NCCL INFO Channel 03/0 : 44[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO Channel 06/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618587:618751 [3] NCCL INFO Channel 03/0 : 43[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO Channel 08/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618591:618746 [7] NCCL INFO Channel 10/0 : 31[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880686:880841 [7] NCCL INFO Channel 14/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618586:618745 [2] NCCL INFO Channel 03/0 : 42[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880684:880842 [5] NCCL INFO Channel 11/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688200:688369 [0] NCCL INFO Channel 14/0 : 48[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:94342:94504 [0] NCCL INFO Channel 07/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-7:811251:811409 [3] NCCL INFO Channel 00/0 : 59[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-7:811252:811408 [4] NCCL INFO Channel 06/0 : 60[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926599:926755 [5] NCCL INFO Channel 12/0 : 53[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO Channel 07/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926601:926759 [7] NCCL INFO Channel 06/0 : 55[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618589:618744 [5] NCCL INFO Channel 08/0 : 29[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO Channel 09/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880684:880842 [5] NCCL INFO Channel 12/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926600:926757 [6] NCCL INFO Channel 13/0 : 54[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618588:618750 [4] NCCL INFO Channel 05/0 : 44[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688200:688369 [0] NCCL INFO Channel 02/0 : 32[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618587:618751 [3] NCCL INFO Channel 05/0 : 43[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-7:811250:811412 [2] NCCL INFO Channel 04/0 : 58[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618591:618746 [7] NCCL INFO Channel 12/0 : 31[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:94342:94504 [0] NCCL INFO Channel 09/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-7:811252:811408 [4] NCCL INFO Channel 08/0 : 60[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-7:811251:811409 [3] NCCL INFO Channel 02/0 : 59[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO Channel 08/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880684:880842 [5] NCCL INFO Channel 14/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926599:926755 [5] NCCL INFO Channel 14/0 : 53[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926601:926759 [7] NCCL INFO Channel 08/0 : 55[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO Channel 10/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926600:926757 [6] NCCL INFO Channel 15/0 : 54[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618586:618745 [2] NCCL INFO Channel 05/0 : 42[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688200:688369 [0] NCCL INFO Channel 04/0 : 32[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926599:926755 [5] NCCL INFO Channel 15/0 : 53[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618589:618744 [5] NCCL INFO Channel 10/0 : 29[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618588:618750 [4] NCCL INFO Channel 07/0 : 44[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926601:926759 [7] NCCL INFO Channel 09/0 : 55[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618587:618751 [3] NCCL INFO Channel 07/0 : 43[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-7:811250:811412 [2] NCCL INFO Channel 06/0 : 58[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618591:618746 [7] NCCL INFO Channel 14/0 : 31[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-7:811252:811408 [4] NCCL INFO Channel 10/0 : 60[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880684:880842 [5] NCCL INFO Channel 15/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618586:618745 [2] NCCL INFO Channel 07/0 : 42[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:94342:94504 [0] NCCL INFO Channel 11/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-7:811251:811409 [3] NCCL INFO Channel 04/0 : 59[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO Channel 09/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO Channel 11/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926601:926759 [7] NCCL INFO Channel 10/0 : 55[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688200:688369 [0] NCCL INFO Channel 06/0 : 32[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618589:618744 [5] NCCL INFO Channel 12/0 : 29[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618588:618750 [4] NCCL INFO Channel 09/0 : 44[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618584:618747 [0] NCCL INFO Channel 02/0 : 24[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-7:811250:811412 [2] NCCL INFO Channel 08/0 : 58[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618587:618751 [3] NCCL INFO Channel 09/0 : 43[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-7:811252:811408 [4] NCCL INFO Channel 12/0 : 60[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618586:618745 [2] NCCL INFO Channel 09/0 : 42[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-7:811251:811409 [3] NCCL INFO Channel 06/0 : 59[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO Channel 01/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926601:926759 [7] NCCL INFO Channel 11/0 : 55[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618589:618744 [5] NCCL INFO Channel 14/0 : 29[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:94348:94506 [6] NCCL INFO Channel 01/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-7:811250:811412 [2] NCCL INFO Channel 10/0 : 58[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO Channel 10/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688200:688369 [0] NCCL INFO Channel 08/0 : 32[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:94342:94504 [0] NCCL INFO Channel 13/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO Channel 12/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942612:942779 [1] NCCL INFO Channel 00/0 : 49[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926601:926759 [7] NCCL INFO Channel 12/0 : 55[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618588:618750 [4] NCCL INFO Channel 11/0 : 44[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942612:942779 [1] NCCL INFO Channel 02/0 : 49[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618590:618749 [6] NCCL INFO Channel 00/0 : 30[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618587:618751 [3] NCCL INFO Channel 13/0 : 43[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-7:811252:811408 [4] NCCL INFO Channel 14/0 : 60[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942612:942779 [1] NCCL INFO Channel 03/0 : 49[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618584:618747 [0] NCCL INFO Channel 04/0 : 24[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618586:618745 [2] NCCL INFO Channel 11/0 : 42[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:94348:94506 [6] NCCL INFO Channel 03/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-7:811251:811409 [3] NCCL INFO Channel 08/0 : 59[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942612:942779 [1] NCCL INFO Channel 04/0 : 49[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688200:688369 [0] NCCL INFO Channel 10/0 : 32[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:94342:94504 [0] NCCL INFO Channel 15/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-7:811250:811412 [2] NCCL INFO Channel 12/0 : 58[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO Channel 02/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942612:942779 [1] NCCL INFO Channel 05/0 : 49[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926601:926759 [7] NCCL INFO Channel 13/0 : 55[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO Channel 00/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO Channel 11/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942612:942779 [1] NCCL INFO Channel 06/0 : 49[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618588:618750 [4] NCCL INFO Channel 13/0 : 44[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO Channel 13/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942612:942779 [1] NCCL INFO Channel 07/0 : 49[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:94348:94506 [6] NCCL INFO Channel 05/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO Channel 03/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942615:942775 [4] NCCL INFO Channel 01/0 : 44[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688200:688369 [0] NCCL INFO Channel 12/0 : 32[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-7:811251:811409 [3] NCCL INFO Channel 10/0 : 59[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942612:942779 [1] NCCL INFO Channel 08/0 : 49[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-7:811250:811412 [2] NCCL INFO Channel 14/0 : 58[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942615:942775 [4] NCCL INFO Channel 03/0 : 44[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926601:926759 [7] NCCL INFO Channel 14/0 : 55[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618586:618745 [2] NCCL INFO Channel 13/0 : 42[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618584:618747 [0] NCCL INFO Channel 06/0 : 24[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618590:618749 [6] NCCL INFO Channel 02/0 : 30[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-7:811251:811409 [3] NCCL INFO Channel 12/0 : 59[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO Channel 12/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO Channel 01/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942612:942779 [1] NCCL INFO Channel 10/0 : 49[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618588:618750 [4] NCCL INFO Channel 15/0 : 44[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942615:942775 [4] NCCL INFO Channel 05/0 : 44[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618587:618751 [3] NCCL INFO Channel 15/0 : 43[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688200:688369 [0] NCCL INFO Channel 14/0 : 32[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:94348:94506 [6] NCCL INFO Channel 07/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942612:942779 [1] NCCL INFO Channel 11/0 : 49[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942613:942776 [2] NCCL INFO Channel 01/0 : 42[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942615:942775 [4] NCCL INFO Channel 07/0 : 44[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-7:811251:811409 [3] NCCL INFO Channel 14/0 : 59[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942614:942773 [3] NCCL INFO Channel 01/0 : 43[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942612:942779 [1] NCCL INFO Channel 12/0 : 49[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942613:942776 [2] NCCL INFO Channel 03/0 : 42[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618584:618747 [0] NCCL INFO Channel 08/0 : 24[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618586:618745 [2] NCCL INFO Channel 15/0 : 42[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942615:942775 [4] NCCL INFO Channel 09/0 : 44[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:94348:94506 [6] NCCL INFO Channel 09/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942614:942773 [3] NCCL INFO Channel 03/0 : 43[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942612:942779 [1] NCCL INFO Channel 13/0 : 49[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942613:942776 [2] NCCL INFO Channel 05/0 : 42[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942615:942775 [4] NCCL INFO Channel 11/0 : 44[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618590:618749 [6] NCCL INFO Channel 04/0 : 30[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618588:618750 [4] NCCL INFO Channel 01/0 : 28[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618587:618751 [3] NCCL INFO Channel 01/0 : 27[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942614:942773 [3] NCCL INFO Channel 05/0 : 43[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942612:942779 [1] NCCL INFO Channel 14/0 : 49[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942613:942776 [2] NCCL INFO Channel 07/0 : 42[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942615:942775 [4] NCCL INFO Channel 13/0 : 44[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942612:942779 [1] NCCL INFO Channel 15/0 : 49[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942614:942773 [3] NCCL INFO Channel 07/0 : 43[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942613:942776 [2] NCCL INFO Channel 09/0 : 42[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:94348:94506 [6] NCCL INFO Channel 11/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942612:942779 [1] NCCL INFO Channel 01/0 : 41[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942615:942775 [4] NCCL INFO Channel 15/0 : 44[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942614:942773 [3] NCCL INFO Channel 09/0 : 43[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942613:942776 [2] NCCL INFO Channel 11/0 : 42[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618586:618745 [2] NCCL INFO Channel 01/0 : 26[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942612:942779 [1] NCCL INFO Channel 03/0 : 41[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942614:942773 [3] NCCL INFO Channel 13/0 : 43[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942613:942776 [2] NCCL INFO Channel 13/0 : 42[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942612:942779 [1] NCCL INFO Channel 05/0 : 41[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942614:942773 [3] NCCL INFO Channel 15/0 : 43[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942613:942776 [2] NCCL INFO Channel 15/0 : 42[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942612:942779 [1] NCCL INFO Channel 07/0 : 41[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:94348:94506 [6] NCCL INFO Channel 13/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942612:942779 [1] NCCL INFO Channel 11/0 : 41[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618584:618747 [0] NCCL INFO Channel 10/0 : 24[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942612:942779 [1] NCCL INFO Channel 13/0 : 41[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618587:618751 [3] NCCL INFO Channel 03/0 : 27[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618590:618749 [6] NCCL INFO Channel 08/0 : 30[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618588:618750 [4] NCCL INFO Channel 03/0 : 28[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942612:942779 [1] NCCL INFO Channel 15/0 : 41[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942618:942778 [7] NCCL INFO Channel 00/0 : 55[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942618:942778 [7] NCCL INFO Channel 01/0 : 55[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942616:942777 [5] NCCL INFO Channel 00/0 : 53[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926594:926762 [0] NCCL INFO Channel 02/0 : 56[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942618:942778 [7] NCCL INFO Channel 02/0 : 55[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942618:942778 [7] NCCL INFO Channel 03/0 : 55[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942616:942777 [5] NCCL INFO Channel 01/0 : 53[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618586:618745 [2] NCCL INFO Channel 03/0 : 26[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:94348:94506 [6] NCCL INFO Channel 15/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942618:942778 [7] NCCL INFO Channel 04/0 : 55[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942616:942777 [5] NCCL INFO Channel 02/0 : 53[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688200:688369 [0] NCCL INFO Channel 01/0 : 40[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942618:942778 [7] NCCL INFO Channel 05/0 : 55[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926594:926762 [0] NCCL INFO Channel 04/0 : 56[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942616:942777 [5] NCCL INFO Channel 03/0 : 53[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618584:618747 [0] NCCL INFO Channel 12/0 : 24[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942618:942778 [7] NCCL INFO Channel 06/0 : 55[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942616:942777 [5] NCCL INFO Channel 04/0 : 53[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880679:880846 [0] NCCL INFO Channel 02/0 : 24[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942618:942778 [7] NCCL INFO Channel 08/0 : 55[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942616:942777 [5] NCCL INFO Channel 06/0 : 53[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688200:688369 [0] NCCL INFO Channel 03/0 : 40[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942618:942778 [7] NCCL INFO Channel 09/0 : 55[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618587:618751 [3] NCCL INFO Channel 05/0 : 27[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618590:618749 [6] NCCL INFO Channel 10/0 : 30[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942616:942777 [5] NCCL INFO Channel 07/0 : 53[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926594:926762 [0] NCCL INFO Channel 06/0 : 56[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618586:618745 [2] NCCL INFO Channel 05/0 : 26[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942618:942778 [7] NCCL INFO Channel 10/0 : 55[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618584:618747 [0] NCCL INFO Channel 14/0 : 24[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942616:942777 [5] NCCL INFO Channel 08/0 : 53[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880679:880846 [0] NCCL INFO Channel 04/0 : 24[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942618:942778 [7] NCCL INFO Channel 11/0 : 55[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942616:942777 [5] NCCL INFO Channel 09/0 : 53[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942618:942778 [7] NCCL INFO Channel 12/0 : 55[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688200:688369 [0] NCCL INFO Channel 05/0 : 40[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942611:942772 [0] NCCL INFO Channel 01/0 : 48[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618587:618751 [3] NCCL INFO Channel 07/0 : 27[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942616:942777 [5] NCCL INFO Channel 10/0 : 53[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926594:926762 [0] NCCL INFO Channel 08/0 : 56[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618586:618745 [2] NCCL INFO Channel 07/0 : 26[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618590:618749 [6] NCCL INFO Channel 12/0 : 30[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO Channel 14/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942618:942778 [7] NCCL INFO Channel 13/0 : 55[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942611:942772 [0] NCCL INFO Channel 02/0 : 48[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942616:942777 [5] NCCL INFO Channel 11/0 : 53[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942618:942778 [7] NCCL INFO Channel 14/0 : 55[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880679:880846 [0] NCCL INFO Channel 06/0 : 24[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942617:942774 [6] NCCL INFO Channel 00/0 : 54[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688200:688369 [0] NCCL INFO Channel 07/0 : 40[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942611:942772 [0] NCCL INFO Channel 03/0 : 48[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942616:942777 [5] NCCL INFO Channel 12/0 : 53[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618587:618751 [3] NCCL INFO Channel 09/0 : 27[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO Channel 04/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942618:942778 [7] NCCL INFO Channel 01/0 : 47[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942617:942774 [6] NCCL INFO Channel 01/0 : 54[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926594:926762 [0] NCCL INFO Channel 10/0 : 56[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942611:942772 [0] NCCL INFO Channel 04/0 : 48[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880679:880846 [0] NCCL INFO Channel 08/0 : 24[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942616:942777 [5] NCCL INFO Channel 14/0 : 53[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942618:942778 [7] NCCL INFO Channel 03/0 : 47[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942617:942774 [6] NCCL INFO Channel 02/0 : 54[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618588:618750 [4] NCCL INFO Channel 05/0 : 28[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688200:688369 [0] NCCL INFO Channel 09/0 : 40[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO Channel 01/0 : 15[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618586:618745 [2] NCCL INFO Channel 09/0 : 26[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926594:926762 [0] NCCL INFO Channel 12/0 : 56[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618590:618749 [6] NCCL INFO Channel 14/0 : 30[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618587:618751 [3] NCCL INFO Channel 13/0 : 27[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942611:942772 [0] NCCL INFO Channel 05/0 : 48[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942616:942777 [5] NCCL INFO Channel 15/0 : 53[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942618:942778 [7] NCCL INFO Channel 05/0 : 47[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880679:880846 [0] NCCL INFO Channel 10/0 : 24[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688200:688369 [0] NCCL INFO Channel 11/0 : 40[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO Channel 03/0 : 15[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942617:942774 [6] NCCL INFO Channel 03/0 : 54[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942611:942772 [0] NCCL INFO Channel 06/0 : 48[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942616:942777 [5] NCCL INFO Channel 01/0 : 45[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618588:618750 [4] NCCL INFO Channel 07/0 : 28[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942618:942778 [7] NCCL INFO Channel 07/0 : 47[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618586:618745 [2] NCCL INFO Channel 11/0 : 26[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942617:942774 [6] NCCL INFO Channel 04/0 : 54[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618587:618751 [3] NCCL INFO Channel 15/0 : 27[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942611:942772 [0] NCCL INFO Channel 07/0 : 48[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926594:926762 [0] NCCL INFO Channel 14/0 : 56[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO Channel 05/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942616:942777 [5] NCCL INFO Channel 03/0 : 45[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688200:688369 [0] NCCL INFO Channel 13/0 : 40[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO Channel 05/0 : 15[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942618:942778 [7] NCCL INFO Channel 09/0 : 47[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942617:942774 [6] NCCL INFO Channel 05/0 : 54[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880679:880846 [0] NCCL INFO Channel 12/0 : 24[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618588:618750 [4] NCCL INFO Channel 09/0 : 28[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942611:942772 [0] NCCL INFO Channel 09/0 : 48[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618586:618745 [2] NCCL INFO Channel 13/0 : 26[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942616:942777 [5] NCCL INFO Channel 05/0 : 45[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926594:926762 [0] NCCL INFO Channel 01/0 : 48[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942618:942778 [7] NCCL INFO Channel 11/0 : 47[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688200:688369 [0] NCCL INFO Channel 15/0 : 40[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942617:942774 [6] NCCL INFO Channel 07/0 : 54[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942611:942772 [0] NCCL INFO Channel 10/0 : 48[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942616:942777 [5] NCCL INFO Channel 07/0 : 45[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618586:618745 [2] NCCL INFO Channel 15/0 : 26[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942618:942778 [7] NCCL INFO Channel 13/0 : 47[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880679:880846 [0] NCCL INFO Channel 14/0 : 24[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618588:618750 [4] NCCL INFO Channel 11/0 : 28[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO Channel 06/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO Channel 07/0 : 15[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942617:942774 [6] NCCL INFO Channel 08/0 : 54[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942611:942772 [0] NCCL INFO Channel 11/0 : 48[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926594:926762 [0] NCCL INFO Channel 02/0 : 48[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942616:942777 [5] NCCL INFO Channel 09/0 : 45[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942617:942774 [6] NCCL INFO Channel 09/0 : 54[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942611:942772 [0] NCCL INFO Channel 12/0 : 48[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942616:942777 [5] NCCL INFO Channel 11/0 : 45[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942617:942774 [6] NCCL INFO Channel 10/0 : 54[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880679:880846 [0] NCCL INFO Channel 01/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618588:618750 [4] NCCL INFO Channel 13/0 : 28[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO Channel 09/0 : 15[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942611:942772 [0] NCCL INFO Channel 13/0 : 48[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO Channel 07/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942616:942777 [5] NCCL INFO Channel 15/0 : 45[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942617:942774 [6] NCCL INFO Channel 11/0 : 54[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926594:926762 [0] NCCL INFO Channel 03/0 : 48[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942611:942772 [0] NCCL INFO Channel 14/0 : 48[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618588:618750 [4] NCCL INFO Channel 15/0 : 28[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942617:942774 [6] NCCL INFO Channel 12/0 : 54[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942611:942772 [0] NCCL INFO Channel 15/0 : 48[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880679:880846 [0] NCCL INFO Channel 02/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO Channel 11/0 : 15[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942617:942774 [6] NCCL INFO Channel 13/0 : 54[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO Channel 09/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942611:942772 [0] NCCL INFO Channel 01/0 : 40[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618587:618751 [3] NCCL INFO Channel 00/0 : 27[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926594:926762 [0] NCCL INFO Channel 04/0 : 48[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880679:880846 [0] NCCL INFO Channel 03/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO Channel 13/0 : 15[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942617:942774 [6] NCCL INFO Channel 15/0 : 54[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO Channel 10/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942611:942772 [0] NCCL INFO Channel 03/0 : 40[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942617:942774 [6] NCCL INFO Channel 01/0 : 46[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618587:618751 [3] NCCL INFO Channel 02/0 : 27[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942611:942772 [0] NCCL INFO Channel 05/0 : 40[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942617:942774 [6] NCCL INFO Channel 03/0 : 46[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926594:926762 [0] NCCL INFO Channel 05/0 : 48[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942611:942772 [0] NCCL INFO Channel 07/0 : 40[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880679:880846 [0] NCCL INFO Channel 04/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO Channel 11/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942617:942774 [6] NCCL INFO Channel 05/0 : 46[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942611:942772 [0] NCCL INFO Channel 09/0 : 40[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942614:942773 [3] NCCL INFO Channel 00/0 : 51[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942617:942774 [6] NCCL INFO Channel 07/0 : 46[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618587:618751 [3] NCCL INFO Channel 04/0 : 27[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942611:942772 [0] NCCL INFO Channel 11/0 : 40[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942614:942773 [3] NCCL INFO Channel 01/0 : 51[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942617:942774 [6] NCCL INFO Channel 09/0 : 46[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926594:926762 [0] NCCL INFO Channel 06/0 : 48[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880679:880846 [0] NCCL INFO Channel 05/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO Channel 02/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618586:618745 [2] NCCL INFO Channel 00/0 : 26[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942611:942772 [0] NCCL INFO Channel 13/0 : 40[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942613:942776 [2] NCCL INFO Channel 00/0 : 50[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942614:942773 [3] NCCL INFO Channel 02/0 : 51[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942617:942774 [6] NCCL INFO Channel 11/0 : 46[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942611:942772 [0] NCCL INFO Channel 15/0 : 40[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926594:926762 [0] NCCL INFO Channel 07/0 : 48[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942613:942776 [2] NCCL INFO Channel 01/0 : 50[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO Channel 12/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942617:942774 [6] NCCL INFO Channel 13/0 : 46[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942614:942773 [3] NCCL INFO Channel 04/0 : 51[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880679:880846 [0] NCCL INFO Channel 06/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618587:618751 [3] NCCL INFO Channel 06/0 : 27[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO Channel 13/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942615:942775 [4] NCCL INFO Channel 00/0 : 52[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942613:942776 [2] NCCL INFO Channel 03/0 : 50[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926594:926762 [0] NCCL INFO Channel 09/0 : 48[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880679:880846 [0] NCCL INFO Channel 07/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618586:618745 [2] NCCL INFO Channel 04/0 : 26[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO Channel 14/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618587:618751 [3] NCCL INFO Channel 08/0 : 27[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618588:618750 [4] NCCL INFO Channel 00/0 : 28[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO Channel 03/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942614:942773 [3] NCCL INFO Channel 05/0 : 51[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942617:942774 [6] NCCL INFO Channel 15/0 : 46[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942613:942776 [2] NCCL INFO Channel 04/0 : 50[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942615:942775 [4] NCCL INFO Channel 01/0 : 52[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926594:926762 [0] NCCL INFO Channel 10/0 : 48[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO Channel 00/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO Channel 14/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880679:880846 [0] NCCL INFO Channel 09/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618586:618745 [2] NCCL INFO Channel 06/0 : 26[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO Channel 15/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942614:942773 [3] NCCL INFO Channel 06/0 : 51[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO Channel 04/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926594:926762 [0] NCCL INFO Channel 11/0 : 48[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880679:880846 [0] NCCL INFO Channel 10/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618587:618751 [3] NCCL INFO Channel 10/0 : 27[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618588:618750 [4] NCCL INFO Channel 02/0 : 28[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618586:618745 [2] NCCL INFO Channel 08/0 : 26[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942615:942775 [4] NCCL INFO Channel 02/0 : 52[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO Channel 01/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942613:942776 [2] NCCL INFO Channel 05/0 : 50[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942614:942773 [3] NCCL INFO Channel 07/0 : 51[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO Channel 15/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO Channel 01/0 : 13[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO Channel 05/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880679:880846 [0] NCCL INFO Channel 11/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926594:926762 [0] NCCL INFO Channel 12/0 : 48[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618586:618745 [2] NCCL INFO Channel 10/0 : 26[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942615:942775 [4] NCCL INFO Channel 03/0 : 52[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO Channel 02/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO Channel 01/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO Channel 03/0 : 13[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO Channel 07/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880679:880846 [0] NCCL INFO Channel 12/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942613:942776 [2] NCCL INFO Channel 06/0 : 50[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942614:942773 [3] NCCL INFO Channel 08/0 : 51[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926594:926762 [0] NCCL INFO Channel 13/0 : 48[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618587:618751 [3] NCCL INFO Channel 12/0 : 27[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618588:618750 [4] NCCL INFO Channel 06/0 : 28[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO Channel 03/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO Channel 00/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942615:942775 [4] NCCL INFO Channel 05/0 : 52[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880679:880846 [0] NCCL INFO Channel 13/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926594:926762 [0] NCCL INFO Channel 14/0 : 48[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618586:618745 [2] NCCL INFO Channel 12/0 : 26[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618587:618751 [3] NCCL INFO Channel 14/0 : 27[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618588:618750 [4] NCCL INFO Channel 08/0 : 28[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO Channel 05/0 : 13[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO Channel 03/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942613:942776 [2] NCCL INFO Channel 07/0 : 50[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942614:942773 [3] NCCL INFO Channel 09/0 : 51[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO Channel 08/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880679:880846 [0] NCCL INFO Channel 14/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO Channel 00/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926594:926762 [0] NCCL INFO Channel 15/0 : 48[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618586:618745 [2] NCCL INFO Channel 14/0 : 26[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618588:618750 [4] NCCL INFO Channel 10/0 : 28[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942613:942776 [2] NCCL INFO Channel 08/0 : 50[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942614:942773 [3] NCCL INFO Channel 10/0 : 51[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880679:880846 [0] NCCL INFO Channel 15/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO Channel 05/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO Channel 01/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618588:618750 [4] NCCL INFO Channel 12/0 : 28[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO Channel 07/0 : 13[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO Channel 09/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO Channel 05/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO Channel 01/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942615:942775 [4] NCCL INFO Channel 06/0 : 52[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618588:618750 [4] NCCL INFO Channel 14/0 : 28[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942614:942773 [3] NCCL INFO Channel 12/0 : 51[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942613:942776 [2] NCCL INFO Channel 09/0 : 50[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO Channel 06/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942615:942775 [4] NCCL INFO Channel 07/0 : 52[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO Channel 09/0 : 13[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO Channel 10/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO Channel 03/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO Channel 07/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942614:942773 [3] NCCL INFO Channel 13/0 : 51[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO Channel 02/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942613:942776 [2] NCCL INFO Channel 11/0 : 50[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942615:942775 [4] NCCL INFO Channel 08/0 : 52[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO Channel 07/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942614:942773 [3] NCCL INFO Channel 14/0 : 51[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942613:942776 [2] NCCL INFO Channel 12/0 : 50[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942615:942775 [4] NCCL INFO Channel 09/0 : 52[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO Channel 11/0 : 13[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO Channel 11/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO Channel 09/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO Channel 04/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO Channel 04/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO Channel 08/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942614:942773 [3] NCCL INFO Channel 15/0 : 51[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO Channel 15/0 : 13[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO Channel 12/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942613:942776 [2] NCCL INFO Channel 13/0 : 50[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942615:942775 [4] NCCL INFO Channel 10/0 : 52[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942614:942773 [3] NCCL INFO Channel 01/0 : 43[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO Channel 11/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO Channel 05/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO Channel 05/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942613:942776 [2] NCCL INFO Channel 14/0 : 50[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO Channel 09/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942615:942775 [4] NCCL INFO Channel 11/0 : 52[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942614:942773 [3] NCCL INFO Channel 03/0 : 43[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO Channel 13/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO Channel 13/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO Channel 06/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO Channel 10/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO Channel 06/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942613:942776 [2] NCCL INFO Channel 15/0 : 50[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO Channel 15/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO Channel 15/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO Channel 07/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942614:942773 [3] NCCL INFO Channel 05/0 : 43[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942615:942775 [4] NCCL INFO Channel 13/0 : 52[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942613:942776 [2] NCCL INFO Channel 01/0 : 42[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO Channel 11/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO Channel 07/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO Channel 01/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942614:942773 [3] NCCL INFO Channel 07/0 : 43[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942615:942775 [4] NCCL INFO Channel 14/0 : 52[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO Channel 08/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO Channel 13/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942613:942776 [2] NCCL INFO Channel 03/0 : 42[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO Channel 03/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO Channel 08/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942614:942773 [3] NCCL INFO Channel 09/0 : 43[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942613:942776 [2] NCCL INFO Channel 05/0 : 42[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942615:942775 [4] NCCL INFO Channel 15/0 : 52[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO Channel 14/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO Channel 09/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO Channel 05/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO Channel 09/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942614:942773 [3] NCCL INFO Channel 13/0 : 43[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO Channel 15/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO Channel 11/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO Channel 07/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942613:942776 [2] NCCL INFO Channel 07/0 : 42[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942615:942775 [4] NCCL INFO Channel 01/0 : 44[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO Channel 10/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO Channel 01/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942614:942773 [3] NCCL INFO Channel 15/0 : 43[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO Channel 12/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO Channel 09/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942613:942776 [2] NCCL INFO Channel 09/0 : 42[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942615:942775 [4] NCCL INFO Channel 03/0 : 44[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO Channel 12/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO Channel 03/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942613:942776 [2] NCCL INFO Channel 11/0 : 42[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO Channel 11/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO Channel 13/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO Channel 13/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO Channel 05/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942615:942775 [4] NCCL INFO Channel 05/0 : 44[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688204:688367 [4] NCCL INFO Channel 14/0 : 52[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO Channel 13/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942613:942776 [2] NCCL INFO Channel 13/0 : 42[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO Channel 14/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942615:942775 [4] NCCL INFO Channel 07/0 : 44[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688204:688367 [4] NCCL INFO Channel 00/0 : 36[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO Channel 07/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO Channel 14/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942613:942776 [2] NCCL INFO Channel 15/0 : 42[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO Channel 15/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942615:942775 [4] NCCL INFO Channel 09/0 : 44[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO Channel 15/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688202:688368 [2] NCCL INFO Channel 08/0 : 50[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942615:942775 [4] NCCL INFO Channel 11/0 : 44[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO Channel 09/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO Channel 15/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO Channel 01/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688204:688367 [4] NCCL INFO Channel 02/0 : 36[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942615:942775 [4] NCCL INFO Channel 13/0 : 44[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO Channel 11/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO Channel 01/0 : 11[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO Channel 03/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942615:942775 [4] NCCL INFO Channel 15/0 : 44[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688204:688367 [4] NCCL INFO Channel 06/0 : 36[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO Channel 13/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688204:688367 [4] NCCL INFO Channel 08/0 : 36[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688202:688368 [2] NCCL INFO Channel 10/0 : 50[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO Channel 03/0 : 11[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO Channel 05/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688204:688367 [4] NCCL INFO Channel 10/0 : 36[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO Channel 15/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO Channel 07/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO Channel 05/0 : 11[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688206:688363 [6] NCCL INFO Channel 11/0 : 46[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688202:688368 [2] NCCL INFO Channel 12/0 : 50[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO Channel 09/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO Channel 07/0 : 11[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688204:688367 [4] NCCL INFO Channel 12/0 : 36[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688206:688363 [6] NCCL INFO Channel 13/0 : 46[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO Channel 11/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688202:688368 [2] NCCL INFO Channel 14/0 : 50[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688204:688367 [4] NCCL INFO Channel 14/0 : 36[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO Channel 09/0 : 11[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688206:688363 [6] NCCL INFO Channel 15/0 : 46[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO Channel 13/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO Channel 13/0 : 11[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688202:688368 [2] NCCL INFO Channel 00/0 : 34[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO Channel 15/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688202:688368 [2] NCCL INFO Channel 04/0 : 34[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO Channel 15/0 : 11[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688202:688368 [2] NCCL INFO Channel 06/0 : 34[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926598:926761 [4] NCCL INFO Channel 00/0 : 60[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688202:688368 [2] NCCL INFO Channel 08/0 : 34[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880683:880844 [4] NCCL INFO Channel 00/0 : 28[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926598:926761 [4] NCCL INFO Channel 02/0 : 60[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688204:688367 [4] NCCL INFO Channel 01/0 : 44[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880683:880844 [4] NCCL INFO Channel 02/0 : 28[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926598:926761 [4] NCCL INFO Channel 06/0 : 60[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688202:688368 [2] NCCL INFO Channel 10/0 : 34[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688204:688367 [4] NCCL INFO Channel 03/0 : 44[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926598:926761 [4] NCCL INFO Channel 08/0 : 60[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688202:688368 [2] NCCL INFO Channel 12/0 : 34[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688204:688367 [4] NCCL INFO Channel 05/0 : 44[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880683:880844 [4] NCCL INFO Channel 06/0 : 28[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926598:926761 [4] NCCL INFO Channel 10/0 : 60[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688202:688368 [2] NCCL INFO Channel 14/0 : 34[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688204:688367 [4] NCCL INFO Channel 07/0 : 44[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880683:880844 [4] NCCL INFO Channel 08/0 : 28[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926598:926761 [4] NCCL INFO Channel 12/0 : 60[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688204:688367 [4] NCCL INFO Channel 09/0 : 44[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880683:880844 [4] NCCL INFO Channel 10/0 : 28[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688204:688367 [4] NCCL INFO Channel 11/0 : 44[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926598:926761 [4] NCCL INFO Channel 14/0 : 60[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688204:688367 [4] NCCL INFO Channel 13/0 : 44[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926598:926761 [4] NCCL INFO Channel 00/0 : 52[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880683:880844 [4] NCCL INFO Channel 12/0 : 28[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688204:688367 [4] NCCL INFO Channel 15/0 : 44[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880683:880844 [4] NCCL INFO Channel 14/0 : 28[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926598:926761 [4] NCCL INFO Channel 01/0 : 52[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926598:926761 [4] NCCL INFO Channel 02/0 : 52[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880683:880844 [4] NCCL INFO Channel 00/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688202:688368 [2] NCCL INFO Channel 01/0 : 42[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880681:880845 [2] NCCL INFO Channel 00/0 : 26[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926596:926756 [2] NCCL INFO Channel 00/0 : 58[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880683:880844 [4] NCCL INFO Channel 01/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880681:880845 [2] NCCL INFO Channel 04/0 : 26[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926598:926761 [4] NCCL INFO Channel 03/0 : 52[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688202:688368 [2] NCCL INFO Channel 03/0 : 42[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880683:880844 [4] NCCL INFO Channel 02/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926596:926756 [2] NCCL INFO Channel 04/0 : 58[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926598:926761 [4] NCCL INFO Channel 05/0 : 52[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688202:688368 [2] NCCL INFO Channel 05/0 : 42[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880681:880845 [2] NCCL INFO Channel 06/0 : 26[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688203:688362 [3] NCCL INFO Channel 14/0 : 35[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880683:880844 [4] NCCL INFO Channel 03/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880681:880845 [2] NCCL INFO Channel 08/0 : 26[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688202:688368 [2] NCCL INFO Channel 07/0 : 42[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880683:880844 [4] NCCL INFO Channel 05/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880681:880845 [2] NCCL INFO Channel 10/0 : 26[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688202:688368 [2] NCCL INFO Channel 09/0 : 42[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880683:880844 [4] NCCL INFO Channel 06/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688202:688368 [2] NCCL INFO Channel 11/0 : 42[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880681:880845 [2] NCCL INFO Channel 12/0 : 26[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880683:880844 [4] NCCL INFO Channel 07/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688202:688368 [2] NCCL INFO Channel 13/0 : 42[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880681:880845 [2] NCCL INFO Channel 14/0 : 26[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880683:880844 [4] NCCL INFO Channel 08/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688203:688362 [3] NCCL INFO Channel 00/0 : 51[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880681:880845 [2] NCCL INFO Channel 00/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:94345:94509 [3] NCCL INFO Channel 01/0 : 11[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688202:688368 [2] NCCL INFO Channel 15/0 : 42[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688203:688362 [3] NCCL INFO Channel 02/0 : 51[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880683:880844 [4] NCCL INFO Channel 09/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880681:880845 [2] NCCL INFO Channel 01/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688203:688362 [3] NCCL INFO Channel 04/0 : 51[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:94345:94509 [3] NCCL INFO Channel 03/0 : 11[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880683:880844 [4] NCCL INFO Channel 10/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688203:688362 [3] NCCL INFO Channel 06/0 : 51[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:94345:94509 [3] NCCL INFO Channel 05/0 : 11[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880681:880845 [2] NCCL INFO Channel 03/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880683:880844 [4] NCCL INFO Channel 11/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688203:688362 [3] NCCL INFO Channel 08/0 : 51[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880681:880845 [2] NCCL INFO Channel 04/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880683:880844 [4] NCCL INFO Channel 13/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:94345:94509 [3] NCCL INFO Channel 07/0 : 11[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880681:880845 [2] NCCL INFO Channel 05/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880683:880844 [4] NCCL INFO Channel 14/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880681:880845 [2] NCCL INFO Channel 06/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:94345:94509 [3] NCCL INFO Channel 09/0 : 11[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688203:688362 [3] NCCL INFO Channel 10/0 : 51[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880683:880844 [4] NCCL INFO Channel 15/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880681:880845 [2] NCCL INFO Channel 07/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688203:688362 [3] NCCL INFO Channel 12/0 : 51[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:94345:94509 [3] NCCL INFO Channel 13/0 : 11[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880681:880845 [2] NCCL INFO Channel 08/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688203:688362 [3] NCCL INFO Channel 14/0 : 51[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:94345:94509 [3] NCCL INFO Channel 15/0 : 11[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880681:880845 [2] NCCL INFO Channel 09/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688203:688362 [3] NCCL INFO Channel 00/0 : 35[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880681:880845 [2] NCCL INFO Channel 11/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688203:688362 [3] NCCL INFO Channel 02/0 : 35[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880681:880845 [2] NCCL INFO Channel 12/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688203:688362 [3] NCCL INFO Channel 04/0 : 35[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880681:880845 [2] NCCL INFO Channel 13/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688203:688362 [3] NCCL INFO Channel 06/0 : 35[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880681:880845 [2] NCCL INFO Channel 14/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688203:688362 [3] NCCL INFO Channel 08/0 : 35[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880681:880845 [2] NCCL INFO Channel 15/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688203:688362 [3] NCCL INFO Channel 10/0 : 35[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688203:688362 [3] NCCL INFO Channel 12/0 : 35[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688203:688362 [3] NCCL INFO Channel 14/0 : 35[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926597:926758 [3] NCCL INFO Channel 00/0 : 59[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688203:688362 [3] NCCL INFO Channel 01/0 : 43[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926598:926761 [4] NCCL INFO Channel 06/0 : 52[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880682:880840 [3] NCCL INFO Channel 00/0 : 27[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688203:688362 [3] NCCL INFO Channel 03/0 : 43[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926597:926758 [3] NCCL INFO Channel 02/0 : 59[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880682:880840 [3] NCCL INFO Channel 02/0 : 27[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688203:688362 [3] NCCL INFO Channel 05/0 : 43[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880682:880840 [3] NCCL INFO Channel 04/0 : 27[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688203:688362 [3] NCCL INFO Channel 07/0 : 43[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880682:880840 [3] NCCL INFO Channel 06/0 : 27[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688203:688362 [3] NCCL INFO Channel 09/0 : 43[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880682:880840 [3] NCCL INFO Channel 08/0 : 27[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688203:688362 [3] NCCL INFO Channel 13/0 : 43[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880682:880840 [3] NCCL INFO Channel 10/0 : 27[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688203:688362 [3] NCCL INFO Channel 15/0 : 43[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880682:880840 [3] NCCL INFO Channel 12/0 : 27[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880682:880840 [3] NCCL INFO Channel 14/0 : 27[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880682:880840 [3] NCCL INFO Channel 00/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880682:880840 [3] NCCL INFO Channel 01/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880682:880840 [3] NCCL INFO Channel 02/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880682:880840 [3] NCCL INFO Channel 04/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880682:880840 [3] NCCL INFO Channel 05/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880682:880840 [3] NCCL INFO Channel 06/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880682:880840 [3] NCCL INFO Channel 07/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880682:880840 [3] NCCL INFO Channel 08/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880682:880840 [3] NCCL INFO Channel 09/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926598:926761 [4] NCCL INFO Channel 07/0 : 52[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926596:926756 [2] NCCL INFO Channel 06/0 : 58[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926597:926758 [3] NCCL INFO Channel 04/0 : 59[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880682:880840 [3] NCCL INFO Channel 10/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926598:926761 [4] NCCL INFO Channel 08/0 : 52[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926596:926756 [2] NCCL INFO Channel 08/0 : 58[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926597:926758 [3] NCCL INFO Channel 06/0 : 59[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926598:926761 [4] NCCL INFO Channel 09/0 : 52[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926596:926756 [2] NCCL INFO Channel 10/0 : 58[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926597:926758 [3] NCCL INFO Channel 08/0 : 59[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926598:926761 [4] NCCL INFO Channel 10/0 : 52[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926596:926756 [2] NCCL INFO Channel 12/0 : 58[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926597:926758 [3] NCCL INFO Channel 10/0 : 59[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926598:926761 [4] NCCL INFO Channel 11/0 : 52[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880682:880840 [3] NCCL INFO Channel 12/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926596:926756 [2] NCCL INFO Channel 14/0 : 58[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926597:926758 [3] NCCL INFO Channel 12/0 : 59[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926598:926761 [4] NCCL INFO Channel 13/0 : 52[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926596:926756 [2] NCCL INFO Channel 00/0 : 50[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926597:926758 [3] NCCL INFO Channel 14/0 : 59[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926598:926761 [4] NCCL INFO Channel 14/0 : 52[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926596:926756 [2] NCCL INFO Channel 01/0 : 50[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926597:926758 [3] NCCL INFO Channel 00/0 : 51[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926598:926761 [4] NCCL INFO Channel 15/0 : 52[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926596:926756 [2] NCCL INFO Channel 03/0 : 50[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926597:926758 [3] NCCL INFO Channel 01/0 : 51[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880682:880840 [3] NCCL INFO Channel 13/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926596:926756 [2] NCCL INFO Channel 04/0 : 50[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926597:926758 [3] NCCL INFO Channel 02/0 : 51[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926596:926756 [2] NCCL INFO Channel 05/0 : 50[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926597:926758 [3] NCCL INFO Channel 04/0 : 51[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926596:926756 [2] NCCL INFO Channel 06/0 : 50[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926597:926758 [3] NCCL INFO Channel 05/0 : 51[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926596:926756 [2] NCCL INFO Channel 07/0 : 50[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926597:926758 [3] NCCL INFO Channel 06/0 : 51[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926596:926756 [2] NCCL INFO Channel 08/0 : 50[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926597:926758 [3] NCCL INFO Channel 07/0 : 51[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926596:926756 [2] NCCL INFO Channel 09/0 : 50[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880682:880840 [3] NCCL INFO Channel 14/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926597:926758 [3] NCCL INFO Channel 08/0 : 51[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926596:926756 [2] NCCL INFO Channel 11/0 : 50[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926597:926758 [3] NCCL INFO Channel 09/0 : 51[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926596:926756 [2] NCCL INFO Channel 12/0 : 50[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880682:880840 [3] NCCL INFO Channel 15/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926596:926756 [2] NCCL INFO Channel 13/0 : 50[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926597:926758 [3] NCCL INFO Channel 10/0 : 51[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926596:926756 [2] NCCL INFO Channel 14/0 : 50[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926597:926758 [3] NCCL INFO Channel 12/0 : 51[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926596:926756 [2] NCCL INFO Channel 15/0 : 50[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926597:926758 [3] NCCL INFO Channel 13/0 : 51[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926597:926758 [3] NCCL INFO Channel 14/0 : 51[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926597:926758 [3] NCCL INFO Channel 15/0 : 51[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-7:811252:811408 [4] NCCL INFO Connected NVLS tree
+ip-26-0-173-7:811252:811408 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-173-7:811252:811408 [4] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-173-7:811252:811408 [4] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-173-7:811250:811412 [2] NCCL INFO Connected NVLS tree
+ip-26-0-173-7:811250:811412 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-173-7:811250:811412 [2] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-173-7:811250:811412 [2] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-173-7:811254:811414 [6] NCCL INFO Connected NVLS tree
+ip-26-0-173-7:811254:811414 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-173-7:811254:811414 [6] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-173-7:811254:811414 [6] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-173-7:811248:811415 [0] NCCL INFO Connected NVLS tree
+ip-26-0-173-7:811248:811415 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-173-7:811248:811415 [0] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-173-7:811248:811415 [0] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-173-7:811255:811411 [7] NCCL INFO Connected NVLS tree
+ip-26-0-173-7:811255:811411 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-173-7:811255:811411 [7] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-173-7:811255:811411 [7] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-173-7:811253:811410 [5] NCCL INFO Connected NVLS tree
+ip-26-0-173-7:811253:811410 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-173-7:811253:811410 [5] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-173-7:811253:811410 [5] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-173-7:811249:811413 [1] NCCL INFO Connected NVLS tree
+ip-26-0-173-7:811249:811413 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-173-7:811249:811413 [1] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-173-7:811249:811413 [1] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-173-7:811251:811409 [3] NCCL INFO Connected NVLS tree
+ip-26-0-173-7:811251:811409 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-173-7:811251:811409 [3] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-173-7:811251:811409 [3] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-160-225:94349:94505 [7] NCCL INFO Connected NVLS tree
+ip-26-0-160-225:94349:94505 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-225:94349:94505 [7] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-160-225:94349:94505 [7] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-173-7:811248:811415 [0] NCCL INFO comm 0xa104f70 rank 56 nranks 64 cudaDev 0 nvmlDev 0 busId 53000 commId 0x3f4d485835722cf1 - Init COMPLETE
+ip-26-0-173-7:811250:811412 [2] NCCL INFO comm 0x958ce00 rank 58 nranks 64 cudaDev 2 nvmlDev 2 busId 75000 commId 0x3f4d485835722cf1 - Init COMPLETE
+ip-26-0-173-7:811253:811410 [5] NCCL INFO comm 0x95fcf60 rank 61 nranks 64 cudaDev 5 nvmlDev 5 busId a8000 commId 0x3f4d485835722cf1 - Init COMPLETE
+ip-26-0-173-7:811254:811414 [6] NCCL INFO comm 0x89213d0 rank 62 nranks 64 cudaDev 6 nvmlDev 6 busId b9000 commId 0x3f4d485835722cf1 - Init COMPLETE
+ip-26-0-173-7:811252:811408 [4] NCCL INFO comm 0x8c2ebd0 rank 60 nranks 64 cudaDev 4 nvmlDev 4 busId 97000 commId 0x3f4d485835722cf1 - Init COMPLETE
+ip-26-0-173-7:811249:811413 [1] NCCL INFO comm 0x8794560 rank 57 nranks 64 cudaDev 1 nvmlDev 1 busId 64000 commId 0x3f4d485835722cf1 - Init COMPLETE
+ip-26-0-173-7:811255:811411 [7] NCCL INFO comm 0x9bdb100 rank 63 nranks 64 cudaDev 7 nvmlDev 7 busId ca000 commId 0x3f4d485835722cf1 - Init COMPLETE
+ip-26-0-173-7:811251:811409 [3] NCCL INFO comm 0x8b1a600 rank 59 nranks 64 cudaDev 3 nvmlDev 3 busId 86000 commId 0x3f4d485835722cf1 - Init COMPLETE
+ip-26-0-160-225:94345:94509 [3] NCCL INFO Connected NVLS tree
+ip-26-0-160-225:94345:94509 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-225:94345:94509 [3] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-160-225:94345:94509 [3] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-172-252:688201:688364 [1] NCCL INFO Connected NVLS tree
+ip-26-0-172-252:688201:688364 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-252:688201:688364 [1] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-172-252:688201:688364 [1] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-172-142:880682:880840 [3] NCCL INFO Connected NVLS tree
+ip-26-0-172-142:880682:880840 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-142:880682:880840 [3] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-172-142:880682:880840 [3] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-172-252:688203:688362 [3] NCCL INFO Connected NVLS tree
+ip-26-0-172-252:688203:688362 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-252:688203:688362 [3] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-172-252:688203:688362 [3] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-172-73:926597:926758 [3] NCCL INFO Connected NVLS tree
+ip-26-0-172-73:926597:926758 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-73:926597:926758 [3] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-172-73:926597:926758 [3] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-160-225:94347:94507 [5] NCCL INFO Connected NVLS tree
+ip-26-0-160-225:94347:94507 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-225:94347:94507 [5] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-160-225:94347:94507 [5] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-172-142:880686:880841 [7] NCCL INFO Connected NVLS tree
+ip-26-0-172-142:880684:880842 [5] NCCL INFO Connected NVLS tree
+ip-26-0-172-142:880684:880842 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-142:880686:880841 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-142:880684:880842 [5] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-172-142:880684:880842 [5] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-172-142:880686:880841 [7] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-172-142:880686:880841 [7] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO Connected NVLS tree
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-160-225:94343:94508 [1] NCCL INFO Connected NVLS tree
+ip-26-0-160-225:94343:94508 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-225:94343:94508 [1] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-160-225:94343:94508 [1] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO Connected NVLS tree
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO Connected NVLS tree
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-172-147:618589:618744 [5] NCCL INFO Connected NVLS tree
+ip-26-0-172-147:618589:618744 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-147:618589:618744 [5] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-172-147:618589:618744 [5] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-172-147:618585:618748 [1] NCCL INFO Connected NVLS tree
+ip-26-0-172-147:618585:618748 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-147:618585:618748 [1] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-172-147:618585:618748 [1] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-172-142:880680:880843 [1] NCCL INFO Connected NVLS tree
+ip-26-0-172-142:880680:880843 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-142:880680:880843 [1] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-172-142:880680:880843 [1] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO Connected NVLS tree
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-172-252:688207:688366 [7] NCCL INFO Connected NVLS tree
+ip-26-0-172-252:688207:688366 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-252:688207:688366 [7] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-172-252:688207:688366 [7] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-172-147:618587:618751 [3] NCCL INFO Connected NVLS tree
+ip-26-0-172-147:618587:618751 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-147:618587:618751 [3] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-172-147:618587:618751 [3] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-172-73:926601:926759 [7] NCCL INFO Connected NVLS tree
+ip-26-0-172-73:926601:926759 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-73:926601:926759 [7] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-172-73:926601:926759 [7] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-172-142:880685:880847 [6] NCCL INFO Connected NVLS tree
+ip-26-0-172-142:880685:880847 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-142:880685:880847 [6] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-172-142:880685:880847 [6] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-160-225:94348:94506 [6] NCCL INFO Connected NVLS tree
+ip-26-0-160-225:94348:94506 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-225:94348:94506 [6] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-160-225:94348:94506 [6] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-172-147:618591:618746 [7] NCCL INFO Connected NVLS tree
+ip-26-0-172-147:618591:618746 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-147:618591:618746 [7] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-172-147:618591:618746 [7] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-172-147:618590:618749 [6] NCCL INFO Connected NVLS tree
+ip-26-0-172-147:618590:618749 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-147:618590:618749 [6] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-172-147:618590:618749 [6] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-160-225:94346:94511 [4] NCCL INFO Connected NVLS tree
+ip-26-0-160-225:94346:94511 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-225:94346:94511 [4] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-160-225:94346:94511 [4] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-160-225:94342:94504 [0] NCCL INFO Connected NVLS tree
+ip-26-0-160-225:94342:94504 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-225:94342:94504 [0] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-160-225:94342:94504 [0] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-172-252:688205:688365 [5] NCCL INFO Connected NVLS tree
+ip-26-0-172-252:688205:688365 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-252:688205:688365 [5] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-172-252:688205:688365 [5] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-160-225:94344:94510 [2] NCCL INFO Connected NVLS tree
+ip-26-0-160-225:94344:94510 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-225:94344:94510 [2] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-160-225:94344:94510 [2] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-172-73:926599:926755 [5] NCCL INFO Connected NVLS tree
+ip-26-0-172-73:926599:926755 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-73:926599:926755 [5] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-172-73:926599:926755 [5] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-172-147:618588:618750 [4] NCCL INFO Connected NVLS tree
+ip-26-0-172-147:618588:618750 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-147:618588:618750 [4] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-172-147:618588:618750 [4] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-172-57:942618:942778 [7] NCCL INFO Connected NVLS tree
+ip-26-0-172-57:942618:942778 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-57:942618:942778 [7] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-172-57:942618:942778 [7] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO Connected NVLS tree
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO Connected NVLS tree
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO Connected NVLS tree
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-172-147:618584:618747 [0] NCCL INFO Connected NVLS tree
+ip-26-0-172-147:618584:618747 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-147:618584:618747 [0] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-172-147:618584:618747 [0] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-172-73:926595:926760 [1] NCCL INFO Connected NVLS tree
+ip-26-0-172-73:926595:926760 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-73:926595:926760 [1] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-172-73:926595:926760 [1] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO Connected NVLS tree
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-160-225:94344:94510 [2] NCCL INFO comm 0x9047390 rank 2 nranks 64 cudaDev 2 nvmlDev 2 busId 75000 commId 0x3f4d485835722cf1 - Init COMPLETE
+ip-26-0-160-225:94342:94504 [0] NCCL INFO comm 0x8eae1b0 rank 0 nranks 64 cudaDev 0 nvmlDev 0 busId 53000 commId 0x3f4d485835722cf1 - Init COMPLETE
+ip-26-0-160-225:94347:94507 [5] NCCL INFO comm 0x90b4a60 rank 5 nranks 64 cudaDev 5 nvmlDev 5 busId a8000 commId 0x3f4d485835722cf1 - Init COMPLETE
+ip-26-0-160-225:94349:94505 [7] NCCL INFO comm 0x89eb7d0 rank 7 nranks 64 cudaDev 7 nvmlDev 7 busId ca000 commId 0x3f4d485835722cf1 - Init COMPLETE
+ip-26-0-160-225:94346:94511 [4] NCCL INFO comm 0xa00d090 rank 4 nranks 64 cudaDev 4 nvmlDev 4 busId 97000 commId 0x3f4d485835722cf1 - Init COMPLETE
+ip-26-0-160-225:94343:94508 [1] NCCL INFO comm 0x82f6cb0 rank 1 nranks 64 cudaDev 1 nvmlDev 1 busId 64000 commId 0x3f4d485835722cf1 - Init COMPLETE
+ip-26-0-160-225:94348:94506 [6] NCCL INFO comm 0x97f3050 rank 6 nranks 64 cudaDev 6 nvmlDev 6 busId b9000 commId 0x3f4d485835722cf1 - Init COMPLETE
+ip-26-0-160-225:94345:94509 [3] NCCL INFO comm 0xa1ae2d0 rank 3 nranks 64 cudaDev 3 nvmlDev 3 busId 86000 commId 0x3f4d485835722cf1 - Init COMPLETE
+ip-26-0-172-57:942612:942779 [1] NCCL INFO Connected NVLS tree
+ip-26-0-172-57:942612:942779 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-57:942612:942779 [1] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-172-57:942612:942779 [1] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-172-147:618586:618745 [2] NCCL INFO Connected NVLS tree
+ip-26-0-172-147:618586:618745 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-147:618586:618745 [2] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-172-147:618586:618745 [2] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-172-57:942616:942777 [5] NCCL INFO Connected NVLS tree
+ip-26-0-172-57:942616:942777 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-57:942616:942777 [5] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-172-57:942616:942777 [5] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-172-252:688204:688367 [4] NCCL INFO Connected NVLS tree
+ip-26-0-172-252:688204:688367 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-252:688204:688367 [4] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-172-252:688204:688367 [4] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-172-73:926598:926761 [4] NCCL INFO Connected NVLS tree
+ip-26-0-172-73:926598:926761 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-73:926598:926761 [4] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-172-73:926598:926761 [4] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-172-116:1019289:1019452 [1] NCCL INFO comm 0x84277a0 rank 9 nranks 64 cudaDev 1 nvmlDev 1 busId 64000 commId 0x3f4d485835722cf1 - Init COMPLETE
+ip-26-0-172-116:1019291:1019450 [3] NCCL INFO comm 0xa1a18a0 rank 11 nranks 64 cudaDev 3 nvmlDev 3 busId 86000 commId 0x3f4d485835722cf1 - Init COMPLETE
+ip-26-0-172-116:1019293:1019455 [5] NCCL INFO comm 0xa1c3a50 rank 13 nranks 64 cudaDev 5 nvmlDev 5 busId a8000 commId 0x3f4d485835722cf1 - Init COMPLETE
+ip-26-0-172-116:1019294:1019451 [6] NCCL INFO comm 0x83bb440 rank 14 nranks 64 cudaDev 6 nvmlDev 6 busId b9000 commId 0x3f4d485835722cf1 - Init COMPLETE
+ip-26-0-172-116:1019292:1019454 [4] NCCL INFO comm 0xa199350 rank 12 nranks 64 cudaDev 4 nvmlDev 4 busId 97000 commId 0x3f4d485835722cf1 - Init COMPLETE
+ip-26-0-172-116:1019295:1019449 [7] NCCL INFO comm 0x97f7620 rank 15 nranks 64 cudaDev 7 nvmlDev 7 busId ca000 commId 0x3f4d485835722cf1 - Init COMPLETE
+ip-26-0-172-116:1019288:1019453 [0] NCCL INFO comm 0x89c15f0 rank 8 nranks 64 cudaDev 0 nvmlDev 0 busId 53000 commId 0x3f4d485835722cf1 - Init COMPLETE
+ip-26-0-172-116:1019290:1019456 [2] NCCL INFO comm 0x9e252e0 rank 10 nranks 64 cudaDev 2 nvmlDev 2 busId 75000 commId 0x3f4d485835722cf1 - Init COMPLETE
+ip-26-0-172-57:942614:942773 [3] NCCL INFO Connected NVLS tree
+ip-26-0-172-57:942614:942773 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-57:942614:942773 [3] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-172-57:942614:942773 [3] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-172-142:880681:880845 [2] NCCL INFO Connected NVLS tree
+ip-26-0-172-142:880681:880845 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-142:880681:880845 [2] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-172-142:880681:880845 [2] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-172-147:618589:618744 [5] NCCL INFO comm 0x962f830 rank 29 nranks 64 cudaDev 5 nvmlDev 5 busId a8000 commId 0x3f4d485835722cf1 - Init COMPLETE
+ip-26-0-172-147:618587:618751 [3] NCCL INFO comm 0x8914d00 rank 27 nranks 64 cudaDev 3 nvmlDev 3 busId 86000 commId 0x3f4d485835722cf1 - Init COMPLETE
+ip-26-0-172-147:618585:618748 [1] NCCL INFO comm 0x9e17fb0 rank 25 nranks 64 cudaDev 1 nvmlDev 1 busId 64000 commId 0x3f4d485835722cf1 - Init COMPLETE
+ip-26-0-172-147:618590:618749 [6] NCCL INFO comm 0x9f398d0 rank 30 nranks 64 cudaDev 6 nvmlDev 6 busId b9000 commId 0x3f4d485835722cf1 - Init COMPLETE
+ip-26-0-172-147:618588:618750 [4] NCCL INFO comm 0x8654810 rank 28 nranks 64 cudaDev 4 nvmlDev 4 busId 97000 commId 0x3f4d485835722cf1 - Init COMPLETE
+ip-26-0-172-147:618584:618747 [0] NCCL INFO comm 0x9905f70 rank 24 nranks 64 cudaDev 0 nvmlDev 0 busId 53000 commId 0x3f4d485835722cf1 - Init COMPLETE
+ip-26-0-172-147:618586:618745 [2] NCCL INFO comm 0x89463c0 rank 26 nranks 64 cudaDev 2 nvmlDev 2 busId 75000 commId 0x3f4d485835722cf1 - Init COMPLETE
+ip-26-0-172-73:926600:926757 [6] NCCL INFO Connected NVLS tree
+ip-26-0-172-73:926600:926757 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-73:926600:926757 [6] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-172-73:926600:926757 [6] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-172-142:880683:880844 [4] NCCL INFO Connected NVLS tree
+ip-26-0-172-142:880683:880844 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-142:880683:880844 [4] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-172-142:880683:880844 [4] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-172-252:688206:688363 [6] NCCL INFO Connected NVLS tree
+ip-26-0-172-252:688206:688363 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-252:688206:688363 [6] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-172-252:688206:688363 [6] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-172-147:618591:618746 [7] NCCL INFO comm 0x91d1750 rank 31 nranks 64 cudaDev 7 nvmlDev 7 busId ca000 commId 0x3f4d485835722cf1 - Init COMPLETE
+ip-26-0-172-57:942615:942775 [4] NCCL INFO Connected NVLS tree
+ip-26-0-172-57:942615:942775 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-57:942615:942775 [4] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-172-57:942615:942775 [4] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-172-142:880679:880846 [0] NCCL INFO Connected NVLS tree
+ip-26-0-172-142:880679:880846 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-142:880679:880846 [0] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-172-142:880679:880846 [0] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-172-252:688202:688368 [2] NCCL INFO Connected NVLS tree
+ip-26-0-172-252:688202:688368 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-252:688202:688368 [2] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-172-252:688202:688368 [2] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-172-73:926596:926756 [2] NCCL INFO Connected NVLS tree
+ip-26-0-172-73:926596:926756 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-73:926596:926756 [2] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-172-73:926596:926756 [2] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-172-252:688200:688369 [0] NCCL INFO Connected NVLS tree
+ip-26-0-172-252:688200:688369 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-252:688200:688369 [0] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-172-252:688200:688369 [0] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-172-73:926594:926762 [0] NCCL INFO Connected NVLS tree
+ip-26-0-172-73:926594:926762 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-73:926594:926762 [0] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-172-73:926594:926762 [0] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-172-57:942617:942774 [6] NCCL INFO Connected NVLS tree
+ip-26-0-172-57:942617:942774 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-57:942617:942774 [6] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-172-57:942617:942774 [6] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-172-142:880684:880842 [5] NCCL INFO comm 0x88ad900 rank 21 nranks 64 cudaDev 5 nvmlDev 5 busId a8000 commId 0x3f4d485835722cf1 - Init COMPLETE
+ip-26-0-172-142:880680:880843 [1] NCCL INFO comm 0x8da1b60 rank 17 nranks 64 cudaDev 1 nvmlDev 1 busId 64000 commId 0x3f4d485835722cf1 - Init COMPLETE
+ip-26-0-172-142:880682:880840 [3] NCCL INFO comm 0x928d640 rank 19 nranks 64 cudaDev 3 nvmlDev 3 busId 86000 commId 0x3f4d485835722cf1 - Init COMPLETE
+ip-26-0-172-142:880685:880847 [6] NCCL INFO comm 0x876ac20 rank 22 nranks 64 cudaDev 6 nvmlDev 6 busId b9000 commId 0x3f4d485835722cf1 - Init COMPLETE
+ip-26-0-172-142:880681:880845 [2] NCCL INFO comm 0x8e71310 rank 18 nranks 64 cudaDev 2 nvmlDev 2 busId 75000 commId 0x3f4d485835722cf1 - Init COMPLETE
+ip-26-0-172-142:880683:880844 [4] NCCL INFO comm 0x9a8c230 rank 20 nranks 64 cudaDev 4 nvmlDev 4 busId 97000 commId 0x3f4d485835722cf1 - Init COMPLETE
+ip-26-0-172-142:880679:880846 [0] NCCL INFO comm 0x9ed6ec0 rank 16 nranks 64 cudaDev 0 nvmlDev 0 busId 53000 commId 0x3f4d485835722cf1 - Init COMPLETE
+ip-26-0-172-142:880686:880841 [7] NCCL INFO comm 0x94e5950 rank 23 nranks 64 cudaDev 7 nvmlDev 7 busId ca000 commId 0x3f4d485835722cf1 - Init COMPLETE
+ip-26-0-172-57:942611:942772 [0] NCCL INFO Connected NVLS tree
+ip-26-0-172-57:942611:942772 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-57:942611:942772 [0] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-172-57:942611:942772 [0] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-172-57:942613:942776 [2] NCCL INFO Connected NVLS tree
+ip-26-0-172-57:942613:942776 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-57:942613:942776 [2] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-172-57:942613:942776 [2] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-172-252:688202:688368 [2] NCCL INFO comm 0x900a490 rank 34 nranks 64 cudaDev 2 nvmlDev 2 busId 75000 commId 0x3f4d485835722cf1 - Init COMPLETE
+ip-26-0-172-252:688200:688369 [0] NCCL INFO comm 0x920bad0 rank 32 nranks 64 cudaDev 0 nvmlDev 0 busId 53000 commId 0x3f4d485835722cf1 - Init COMPLETE
+ip-26-0-172-252:688206:688363 [6] NCCL INFO comm 0xa251280 rank 38 nranks 64 cudaDev 6 nvmlDev 6 busId b9000 commId 0x3f4d485835722cf1 - Init COMPLETE
+ip-26-0-172-252:688201:688364 [1] NCCL INFO comm 0x91aa790 rank 33 nranks 64 cudaDev 1 nvmlDev 1 busId 64000 commId 0x3f4d485835722cf1 - Init COMPLETE
+ip-26-0-172-252:688203:688362 [3] NCCL INFO comm 0x8600940 rank 35 nranks 64 cudaDev 3 nvmlDev 3 busId 86000 commId 0x3f4d485835722cf1 - Init COMPLETE
+ip-26-0-172-252:688207:688366 [7] NCCL INFO comm 0x8cf40a0 rank 39 nranks 64 cudaDev 7 nvmlDev 7 busId ca000 commId 0x3f4d485835722cf1 - Init COMPLETE
+ip-26-0-172-252:688205:688365 [5] NCCL INFO comm 0x8c6c940 rank 37 nranks 64 cudaDev 5 nvmlDev 5 busId a8000 commId 0x3f4d485835722cf1 - Init COMPLETE
+ip-26-0-172-252:688204:688367 [4] NCCL INFO comm 0x9286a60 rank 36 nranks 64 cudaDev 4 nvmlDev 4 busId 97000 commId 0x3f4d485835722cf1 - Init COMPLETE
+ip-26-0-172-73:926600:926757 [6] NCCL INFO comm 0x9e0a1c0 rank 54 nranks 64 cudaDev 6 nvmlDev 6 busId b9000 commId 0x3f4d485835722cf1 - Init COMPLETE
+ip-26-0-172-73:926601:926759 [7] NCCL INFO comm 0x8367300 rank 55 nranks 64 cudaDev 7 nvmlDev 7 busId ca000 commId 0x3f4d485835722cf1 - Init COMPLETE
+ip-26-0-172-73:926595:926760 [1] NCCL INFO comm 0x82893c0 rank 49 nranks 64 cudaDev 1 nvmlDev 1 busId 64000 commId 0x3f4d485835722cf1 - Init COMPLETE
+ip-26-0-172-73:926597:926758 [3] NCCL INFO comm 0x8ae3590 rank 51 nranks 64 cudaDev 3 nvmlDev 3 busId 86000 commId 0x3f4d485835722cf1 - Init COMPLETE
+ip-26-0-172-73:926596:926756 [2] NCCL INFO comm 0x8bdc2a0 rank 50 nranks 64 cudaDev 2 nvmlDev 2 busId 75000 commId 0x3f4d485835722cf1 - Init COMPLETE
+ip-26-0-172-73:926599:926755 [5] NCCL INFO comm 0x9df4700 rank 53 nranks 64 cudaDev 5 nvmlDev 5 busId a8000 commId 0x3f4d485835722cf1 - Init COMPLETE
+ip-26-0-172-73:926598:926761 [4] NCCL INFO comm 0x96d54b0 rank 52 nranks 64 cudaDev 4 nvmlDev 4 busId 97000 commId 0x3f4d485835722cf1 - Init COMPLETE
+ip-26-0-172-73:926594:926762 [0] NCCL INFO comm 0x920c450 rank 48 nranks 64 cudaDev 0 nvmlDev 0 busId 53000 commId 0x3f4d485835722cf1 - Init COMPLETE
+ip-26-0-172-57:942618:942778 [7] NCCL INFO comm 0x8e28120 rank 47 nranks 64 cudaDev 7 nvmlDev 7 busId ca000 commId 0x3f4d485835722cf1 - Init COMPLETE
+ip-26-0-172-57:942617:942774 [6] NCCL INFO comm 0x959be20 rank 46 nranks 64 cudaDev 6 nvmlDev 6 busId b9000 commId 0x3f4d485835722cf1 - Init COMPLETE
+ip-26-0-172-57:942616:942777 [5] NCCL INFO comm 0x84435f0 rank 45 nranks 64 cudaDev 5 nvmlDev 5 busId a8000 commId 0x3f4d485835722cf1 - Init COMPLETE
+ip-26-0-172-57:942612:942779 [1] NCCL INFO comm 0x891af90 rank 41 nranks 64 cudaDev 1 nvmlDev 1 busId 64000 commId 0x3f4d485835722cf1 - Init COMPLETE
+ip-26-0-172-57:942615:942775 [4] NCCL INFO comm 0x9ebb790 rank 44 nranks 64 cudaDev 4 nvmlDev 4 busId 97000 commId 0x3f4d485835722cf1 - Init COMPLETE
+ip-26-0-172-57:942611:942772 [0] NCCL INFO comm 0x9b57e50 rank 40 nranks 64 cudaDev 0 nvmlDev 0 busId 53000 commId 0x3f4d485835722cf1 - Init COMPLETE
+ip-26-0-172-57:942614:942773 [3] NCCL INFO comm 0x89687c0 rank 43 nranks 64 cudaDev 3 nvmlDev 3 busId 86000 commId 0x3f4d485835722cf1 - Init COMPLETE
+ip-26-0-172-57:942613:942776 [2] NCCL INFO comm 0x995bf70 rank 42 nranks 64 cudaDev 2 nvmlDev 2 busId 75000 commId 0x3f4d485835722cf1 - Init COMPLETE
+ip-26-0-172-142:880685:880977 [6] NCCL INFO Using network Libfabric
+ip-26-0-160-225:94349:94645 [7] NCCL INFO Using network Libfabric
+ip-26-0-160-225:94343:94647 [1] NCCL INFO Using network Libfabric
+ip-26-0-160-225:94346:94644 [4] NCCL INFO Using network Libfabric
+ip-26-0-160-225:94347:94646 [5] NCCL INFO Using network Libfabric
+ip-26-0-160-225:94344:94648 [2] NCCL INFO Using network Libfabric
+ip-26-0-172-142:880682:880978 [3] NCCL INFO Using network Libfabric
+ip-26-0-172-252:688203:688501 [3] NCCL INFO Using network Libfabric
+ip-26-0-172-142:880686:880979 [7] NCCL INFO Using network Libfabric
+ip-26-0-172-252:688207:688504 [7] NCCL INFO Using network Libfabric
+ip-26-0-172-252:688202:688503 [2] NCCL INFO Using network Libfabric
+ip-26-0-172-252:688204:688502 [4] NCCL INFO Using network Libfabric
+ip-26-0-172-252:688206:688506 [6] NCCL INFO Using network Libfabric
+ip-26-0-172-57:942611:942911 [0] NCCL INFO Using network Libfabric
+ip-26-0-172-142:880683:880982 [4] NCCL INFO Using network Libfabric
+ip-26-0-172-116:1019293:1019592 [5] NCCL INFO Using network Libfabric
+ip-26-0-172-116:1019288:1019588 [0] NCCL INFO Using network Libfabric
+ip-26-0-172-73:926598:926894 [4] NCCL INFO Using network Libfabric
+ip-26-0-172-252:688201:688500 [1] NCCL INFO Using network Libfabric
+ip-26-0-172-116:1019292:1019590 [4] NCCL INFO Using network Libfabric
+ip-26-0-172-116:1019291:1019594 [3] NCCL INFO Using network Libfabric
+ip-26-0-172-116:1019295:1019591 [7] NCCL INFO Using network Libfabric
+ip-26-0-172-142:880684:880981 [5] NCCL INFO Using network Libfabric
+ip-26-0-172-147:618589:618883 [5] NCCL INFO Using network Libfabric
+ip-26-0-160-225:94348:94650 [6] NCCL INFO Using network Libfabric
+ip-26-0-172-57:942612:942914 [1] NCCL INFO Using network Libfabric
+ip-26-0-172-57:942614:942915 [3] NCCL INFO Using network Libfabric
+ip-26-0-172-142:880681:880983 [2] NCCL INFO Using network Libfabric
+ip-26-0-172-73:926599:926895 [5] NCCL INFO Using network Libfabric
+ip-26-0-172-73:926600:926893 [6] NCCL INFO Using network Libfabric
+ip-26-0-172-116:1019289:1019593 [1] NCCL INFO Using network Libfabric
+ip-26-0-172-147:618588:618888 [4] NCCL INFO Using network Libfabric
+ip-26-0-172-147:618585:618884 [1] NCCL INFO Using network Libfabric
+ip-26-0-172-147:618584:618882 [0] NCCL INFO Using network Libfabric
+ip-26-0-172-57:942615:942912 [4] NCCL INFO Using network Libfabric
+ip-26-0-173-7:811250:811555 [2] NCCL INFO Using network Libfabric
+ip-26-0-173-7:811251:811556 [3] NCCL INFO Using network Libfabric
+ip-26-0-172-147:618586:618886 [2] NCCL INFO Using network Libfabric
+ip-26-0-172-147:618587:618887 [3] NCCL INFO Using network Libfabric
+ip-26-0-172-142:880680:880984 [1] NCCL INFO Using network Libfabric
+ip-26-0-172-147:618591:618885 [7] NCCL INFO Using network Libfabric
+ip-26-0-172-147:618590:618889 [6] NCCL INFO Using network Libfabric
+ip-26-0-172-252:688205:688505 [5] NCCL INFO Using network Libfabric
+ip-26-0-173-7:811249:811559 [1] NCCL INFO Using network Libfabric
+ip-26-0-173-7:811248:811554 [0] NCCL INFO Using network Libfabric
+ip-26-0-172-57:942617:942916 [6] NCCL INFO Using network Libfabric
+ip-26-0-173-7:811254:811557 [6] NCCL INFO Using network Libfabric
+ip-26-0-172-57:942618:942918 [7] NCCL INFO Using network Libfabric
+ip-26-0-160-225:94342:94643 [0] NCCL INFO Using network Libfabric
+ip-26-0-172-73:926596:926899 [2] NCCL INFO Using network Libfabric
+ip-26-0-172-252:688200:688507 [0] NCCL INFO Using network Libfabric
+ip-26-0-160-225:94345:94649 [3] NCCL INFO Using network Libfabric
+ip-26-0-172-57:942613:942917 [2] NCCL INFO Using network Libfabric
+ip-26-0-172-73:926597:926900 [3] NCCL INFO Using network Libfabric
+ip-26-0-173-7:811253:811558 [5] NCCL INFO Using network Libfabric
+ip-26-0-172-116:1019290:1019595 [2] NCCL INFO Using network Libfabric
+ip-26-0-172-73:926594:926898 [0] NCCL INFO Using network Libfabric
+ip-26-0-172-73:926595:926897 [1] NCCL INFO Using network Libfabric
+ip-26-0-172-57:942616:942913 [5] NCCL INFO Using network Libfabric
+ip-26-0-172-142:880679:880980 [0] NCCL INFO Using network Libfabric
+ip-26-0-173-7:811255:811561 [7] NCCL INFO Using network Libfabric
+ip-26-0-172-73:926601:926896 [7] NCCL INFO Using network Libfabric
+ip-26-0-173-7:811252:811560 [4] NCCL INFO Using network Libfabric
+ip-26-0-172-116:1019294:1019589 [6] NCCL INFO Using network Libfabric
+ip-26-0-160-225:94342:94643 [0] NCCL INFO comm 0x9322400 rank 0 nranks 64 cudaDev 0 nvmlDev 0 busId 53000 commId 0x6ca6bcab597fda79 - Init START
+ip-26-0-160-225:94349:94645 [7] NCCL INFO comm 0x8e4ac60 rank 7 nranks 64 cudaDev 7 nvmlDev 7 busId ca000 commId 0x6ca6bcab597fda79 - Init START
+ip-26-0-160-225:94343:94647 [1] NCCL INFO comm 0x874a1e0 rank 1 nranks 64 cudaDev 1 nvmlDev 1 busId 64000 commId 0x6ca6bcab597fda79 - Init START
+ip-26-0-160-225:94345:94649 [3] NCCL INFO comm 0xa67f5a0 rank 3 nranks 64 cudaDev 3 nvmlDev 3 busId 86000 commId 0x6ca6bcab597fda79 - Init START
+ip-26-0-160-225:94346:94644 [4] NCCL INFO comm 0xa4e59b0 rank 4 nranks 64 cudaDev 4 nvmlDev 4 busId 97000 commId 0x6ca6bcab597fda79 - Init START
+ip-26-0-160-225:94348:94650 [6] NCCL INFO comm 0x9c46720 rank 6 nranks 64 cudaDev 6 nvmlDev 6 busId b9000 commId 0x6ca6bcab597fda79 - Init START
+ip-26-0-160-225:94344:94648 [2] NCCL INFO comm 0x949ee50 rank 2 nranks 64 cudaDev 2 nvmlDev 2 busId 75000 commId 0x6ca6bcab597fda79 - Init START
+ip-26-0-173-7:811255:811561 [7] NCCL INFO comm 0xa0349b0 rank 63 nranks 64 cudaDev 7 nvmlDev 7 busId ca000 commId 0x6ca6bcab597fda79 - Init START
+ip-26-0-160-225:94347:94646 [5] NCCL INFO comm 0x95790b0 rank 5 nranks 64 cudaDev 5 nvmlDev 5 busId a8000 commId 0x6ca6bcab597fda79 - Init START
+ip-26-0-173-7:811254:811557 [6] NCCL INFO comm 0x8df0080 rank 62 nranks 64 cudaDev 6 nvmlDev 6 busId b9000 commId 0x6ca6bcab597fda79 - Init START
+ip-26-0-173-7:811253:811558 [5] NCCL INFO comm 0x9a582e0 rank 61 nranks 64 cudaDev 5 nvmlDev 5 busId a8000 commId 0x6ca6bcab597fda79 - Init START
+ip-26-0-173-7:811251:811556 [3] NCCL INFO comm 0x8fe5b80 rank 59 nranks 64 cudaDev 3 nvmlDev 3 busId 86000 commId 0x6ca6bcab597fda79 - Init START
+ip-26-0-173-7:811252:811560 [4] NCCL INFO comm 0x9103640 rank 60 nranks 64 cudaDev 4 nvmlDev 4 busId 97000 commId 0x6ca6bcab597fda79 - Init START
+ip-26-0-172-116:1019293:1019592 [5] NCCL INFO comm 0xa68afb0 rank 13 nranks 64 cudaDev 5 nvmlDev 5 busId a8000 commId 0x6ca6bcab597fda79 - Init START
+ip-26-0-172-116:1019294:1019589 [6] NCCL INFO comm 0x8886e00 rank 14 nranks 64 cudaDev 6 nvmlDev 6 busId b9000 commId 0x6ca6bcab597fda79 - Init START
+ip-26-0-172-116:1019292:1019590 [4] NCCL INFO comm 0xa5e9420 rank 12 nranks 64 cudaDev 4 nvmlDev 4 busId 97000 commId 0x6ca6bcab597fda79 - Init START
+ip-26-0-172-116:1019288:1019588 [0] NCCL INFO comm 0x8e0d8a0 rank 8 nranks 64 cudaDev 0 nvmlDev 0 busId 53000 commId 0x6ca6bcab597fda79 - Init START
+ip-26-0-172-116:1019295:1019591 [7] NCCL INFO comm 0x9c544f0 rank 15 nranks 64 cudaDev 7 nvmlDev 7 busId ca000 commId 0x6ca6bcab597fda79 - Init START
+ip-26-0-172-116:1019291:1019594 [3] NCCL INFO comm 0xa5fbcf0 rank 11 nranks 64 cudaDev 3 nvmlDev 3 busId 86000 commId 0x6ca6bcab597fda79 - Init START
+ip-26-0-172-57:942618:942918 [7] NCCL INFO comm 0x92847f0 rank 47 nranks 64 cudaDev 7 nvmlDev 7 busId ca000 commId 0x6ca6bcab597fda79 - Init START
+ip-26-0-172-142:880685:880977 [6] NCCL INFO comm 0x8bbe030 rank 22 nranks 64 cudaDev 6 nvmlDev 6 busId b9000 commId 0x6ca6bcab597fda79 - Init START
+ip-26-0-172-142:880686:880979 [7] NCCL INFO comm 0x99b27a0 rank 23 nranks 64 cudaDev 7 nvmlDev 7 busId ca000 commId 0x6ca6bcab597fda79 - Init START
+ip-26-0-172-142:880684:880981 [5] NCCL INFO comm 0x8d065a0 rank 21 nranks 64 cudaDev 5 nvmlDev 5 busId a8000 commId 0x6ca6bcab597fda79 - Init START
+ip-26-0-172-142:880683:880982 [4] NCCL INFO comm 0x9ee12e0 rank 20 nranks 64 cudaDev 4 nvmlDev 4 busId 97000 commId 0x6ca6bcab597fda79 - Init START
+ip-26-0-172-142:880682:880978 [3] NCCL INFO comm 0x9765da0 rank 19 nranks 64 cudaDev 3 nvmlDev 3 busId 86000 commId 0x6ca6bcab597fda79 - Init START
+ip-26-0-172-73:926601:926896 [7] NCCL INFO comm 0x8839590 rank 55 nranks 64 cudaDev 7 nvmlDev 7 busId ca000 commId 0x6ca6bcab597fda79 - Init START
+ip-26-0-172-73:926600:926893 [6] NCCL INFO comm 0xa25f780 rank 54 nranks 64 cudaDev 6 nvmlDev 6 busId b9000 commId 0x6ca6bcab597fda79 - Init START
+ip-26-0-172-147:618585:618884 [1] NCCL INFO comm 0xa2724f0 rank 25 nranks 64 cudaDev 1 nvmlDev 1 busId 64000 commId 0x6ca6bcab597fda79 - Init START
+ip-26-0-172-147:618584:618882 [0] NCCL INFO comm 0x9d5f5f0 rank 24 nranks 64 cudaDev 0 nvmlDev 0 busId 53000 commId 0x6ca6bcab597fda79 - Init START
+ip-26-0-172-147:618591:618885 [7] NCCL INFO comm 0x9627a20 rank 31 nranks 64 cudaDev 7 nvmlDev 7 busId ca000 commId 0x6ca6bcab597fda79 - Init START
+ip-26-0-172-147:618589:618883 [5] NCCL INFO comm 0x9a820e0 rank 29 nranks 64 cudaDev 5 nvmlDev 5 busId a8000 commId 0x6ca6bcab597fda79 - Init START
+ip-26-0-172-147:618590:618889 [6] NCCL INFO comm 0xa40dfa0 rank 30 nranks 64 cudaDev 6 nvmlDev 6 busId b9000 commId 0x6ca6bcab597fda79 - Init START
+ip-26-0-172-147:618588:618888 [4] NCCL INFO comm 0x8ab30c0 rank 28 nranks 64 cudaDev 4 nvmlDev 4 busId 97000 commId 0x6ca6bcab597fda79 - Init START
+ip-26-0-172-252:688200:688507 [0] NCCL INFO comm 0x965a660 rank 32 nranks 64 cudaDev 0 nvmlDev 0 busId 53000 commId 0x6ca6bcab597fda79 - Init START
+ip-26-0-172-252:688201:688500 [1] NCCL INFO comm 0x96030f0 rank 33 nranks 64 cudaDev 1 nvmlDev 1 busId 64000 commId 0x6ca6bcab597fda79 - Init START
+ip-26-0-172-252:688202:688503 [2] NCCL INFO comm 0x94e7bb0 rank 34 nranks 64 cudaDev 2 nvmlDev 2 busId 75000 commId 0x6ca6bcab597fda79 - Init START
+ip-26-0-172-252:688207:688504 [7] NCCL INFO comm 0x91ccc70 rank 39 nranks 64 cudaDev 7 nvmlDev 7 busId ca000 commId 0x6ca6bcab597fda79 - Init START
+ip-26-0-172-252:688205:688505 [5] NCCL INFO comm 0x90c5ab0 rank 37 nranks 64 cudaDev 5 nvmlDev 5 busId a8000 commId 0x6ca6bcab597fda79 - Init START
+ip-26-0-172-252:688204:688502 [4] NCCL INFO comm 0x9756c40 rank 36 nranks 64 cudaDev 4 nvmlDev 4 busId 97000 commId 0x6ca6bcab597fda79 - Init START
+ip-26-0-173-7:811250:811555 [2] NCCL INFO comm 0x99e6c20 rank 58 nranks 64 cudaDev 2 nvmlDev 2 busId 75000 commId 0x6ca6bcab597fda79 - Init START
+ip-26-0-173-7:811249:811559 [1] NCCL INFO comm 0x8be9fa0 rank 57 nranks 64 cudaDev 1 nvmlDev 1 busId 64000 commId 0x6ca6bcab597fda79 - Init START
+ip-26-0-173-7:811248:811554 [0] NCCL INFO comm 0xa5d9ba0 rank 56 nranks 64 cudaDev 0 nvmlDev 0 busId 53000 commId 0x6ca6bcab597fda79 - Init START
+ip-26-0-172-116:1019290:1019595 [2] NCCL INFO comm 0xa2736a0 rank 10 nranks 64 cudaDev 2 nvmlDev 2 busId 75000 commId 0x6ca6bcab597fda79 - Init START
+ip-26-0-172-116:1019289:1019593 [1] NCCL INFO comm 0x8887c60 rank 9 nranks 64 cudaDev 1 nvmlDev 1 busId 64000 commId 0x6ca6bcab597fda79 - Init START
+ip-26-0-172-57:942617:942916 [6] NCCL INFO comm 0x99f7c90 rank 46 nranks 64 cudaDev 6 nvmlDev 6 busId b9000 commId 0x6ca6bcab597fda79 - Init START
+ip-26-0-172-57:942616:942913 [5] NCCL INFO comm 0x89176f0 rank 45 nranks 64 cudaDev 5 nvmlDev 5 busId a8000 commId 0x6ca6bcab597fda79 - Init START
+ip-26-0-172-142:880679:880980 [0] NCCL INFO comm 0xa3a4d80 rank 16 nranks 64 cudaDev 0 nvmlDev 0 busId 53000 commId 0x6ca6bcab597fda79 - Init START
+ip-26-0-172-142:880681:880983 [2] NCCL INFO comm 0x92c70b0 rank 18 nranks 64 cudaDev 2 nvmlDev 2 busId 75000 commId 0x6ca6bcab597fda79 - Init START
+ip-26-0-172-142:880680:880984 [1] NCCL INFO comm 0x91fe360 rank 17 nranks 64 cudaDev 1 nvmlDev 1 busId 64000 commId 0x6ca6bcab597fda79 - Init START
+ip-26-0-172-73:926597:926900 [3] NCCL INFO comm 0x8f37a80 rank 51 nranks 64 cudaDev 3 nvmlDev 3 busId 86000 commId 0x6ca6bcab597fda79 - Init START
+ip-26-0-172-73:926599:926895 [5] NCCL INFO comm 0xa251780 rank 53 nranks 64 cudaDev 5 nvmlDev 5 busId a8000 commId 0x6ca6bcab597fda79 - Init START
+ip-26-0-172-73:926598:926894 [4] NCCL INFO comm 0x9ba3630 rank 52 nranks 64 cudaDev 4 nvmlDev 4 busId 97000 commId 0x6ca6bcab597fda79 - Init START
+ip-26-0-172-57:942615:942912 [4] NCCL INFO comm 0xa394dc0 rank 44 nranks 64 cudaDev 4 nvmlDev 4 busId 97000 commId 0x6ca6bcab597fda79 - Init START
+ip-26-0-172-57:942612:942914 [1] NCCL INFO comm 0x8de9ae0 rank 41 nranks 64 cudaDev 1 nvmlDev 1 busId 64000 commId 0x6ca6bcab597fda79 - Init START
+ip-26-0-172-57:942613:942917 [2] NCCL INFO comm 0x9dacfa0 rank 42 nranks 64 cudaDev 2 nvmlDev 2 busId 75000 commId 0x6ca6bcab597fda79 - Init START
+ip-26-0-172-57:942614:942915 [3] NCCL INFO comm 0x8c1a590 rank 43 nranks 64 cudaDev 3 nvmlDev 3 busId 86000 commId 0x6ca6bcab597fda79 - Init START
+ip-26-0-172-73:926596:926899 [2] NCCL INFO comm 0x90ace20 rank 50 nranks 64 cudaDev 2 nvmlDev 2 busId 75000 commId 0x6ca6bcab597fda79 - Init START
+ip-26-0-172-73:926595:926897 [1] NCCL INFO comm 0x875b3b0 rank 49 nranks 64 cudaDev 1 nvmlDev 1 busId 64000 commId 0x6ca6bcab597fda79 - Init START
+ip-26-0-172-147:618586:618886 [2] NCCL INFO comm 0x8da0660 rank 26 nranks 64 cudaDev 2 nvmlDev 2 busId 75000 commId 0x6ca6bcab597fda79 - Init START
+ip-26-0-172-147:618587:618887 [3] NCCL INFO comm 0x8dedc20 rank 27 nranks 64 cudaDev 3 nvmlDev 3 busId 86000 commId 0x6ca6bcab597fda79 - Init START
+ip-26-0-172-252:688206:688506 [6] NCCL INFO comm 0xa6af5e0 rank 38 nranks 64 cudaDev 6 nvmlDev 6 busId b9000 commId 0x6ca6bcab597fda79 - Init START
+ip-26-0-172-252:688203:688501 [3] NCCL INFO comm 0x8a56220 rank 35 nranks 64 cudaDev 3 nvmlDev 3 busId 86000 commId 0x6ca6bcab597fda79 - Init START
+ip-26-0-172-57:942611:942911 [0] NCCL INFO comm 0xa02b330 rank 40 nranks 64 cudaDev 0 nvmlDev 0 busId 53000 commId 0x6ca6bcab597fda79 - Init START
+ip-26-0-172-73:926594:926898 [0] NCCL INFO comm 0x9665ce0 rank 48 nranks 64 cudaDev 0 nvmlDev 0 busId 53000 commId 0x6ca6bcab597fda79 - Init START
+ip-26-0-172-57:942616:942913 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-57:942615:942912 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-160-225:94342:94643 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/349
+ip-26-0-160-225:94346:94644 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-160-225:94348:94650 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-160-225:94343:94647 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-160-225:94349:94645 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-160-225:94347:94646 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-160-225:94344:94648 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-160-225:94345:94649 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-252:688207:688504 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-173-7:811255:811561 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-173-7:811248:811554 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-173-7:811254:811557 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-173-7:811249:811559 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-173-7:811252:811560 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-116:1019294:1019589 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-116:1019292:1019590 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-116:1019291:1019594 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-116:1019288:1019588 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-116:1019290:1019595 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-116:1019293:1019592 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-116:1019289:1019593 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-57:942614:942915 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-57:942613:942917 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-142:880684:880981 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-142:880683:880982 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-142:880685:880977 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-142:880686:880979 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-73:926597:926900 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-73:926600:926893 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-73:926596:926899 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-73:926598:926894 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-73:926595:926897 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-73:926601:926896 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-73:926599:926895 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-73:926594:926898 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-173-7:811251:811556 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-173-7:811250:811555 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-173-7:811253:811558 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-57:942618:942918 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-252:688206:688506 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-57:942612:942914 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-57:942617:942916 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-147:618591:618885 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-147:618590:618889 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-147:618588:618888 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-147:618587:618887 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-147:618589:618883 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-147:618586:618886 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-252:688205:688505 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-116:1019295:1019591 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-57:942611:942911 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-142:880682:880978 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-142:880679:880980 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-142:880680:880984 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-142:880681:880983 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-252:688203:688501 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-252:688204:688502 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-252:688202:688503 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-252:688201:688500 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-252:688200:688507 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-147:618585:618884 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-147:618584:618882 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-116:1019289:1019593 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-172-116:1019289:1019593 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-172-57:942611:942911 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-172-57:942611:942911 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-172-57:942618:942918 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-172-57:942618:942918 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-172-142:880683:880982 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-172-142:880683:880982 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-172-73:926598:926894 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-172-73:926598:926894 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-172-73:926595:926897 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-172-73:926595:926897 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-173-7:811249:811559 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-173-7:811249:811559 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-172-57:942612:942914 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-172-57:942612:942914 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-172-116:1019295:1019591 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-172-116:1019295:1019591 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-172-57:942615:942912 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-172-57:942615:942912 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-172-142:880682:880978 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-172-142:880682:880978 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-172-142:880680:880984 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-172-142:880680:880984 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-172-57:942614:942915 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-172-57:942614:942915 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-172-116:1019294:1019589 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-172-116:1019294:1019589 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-172-57:942613:942917 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-172-57:942613:942917 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-172-57:942616:942913 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-172-57:942617:942916 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-172-57:942617:942916 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-172-57:942616:942913 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-173-7:811254:811557 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-173-7:811254:811557 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-172-116:1019291:1019594 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-172-116:1019291:1019594 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-172-116:1019293:1019592 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-172-116:1019293:1019592 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-173-7:811251:811556 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-173-7:811251:811556 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-172-142:880679:880980 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-172-142:880679:880980 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-172-116:1019288:1019588 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-172-116:1019288:1019588 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-173-7:811255:811561 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-173-7:811255:811561 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-172-116:1019290:1019595 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-172-116:1019290:1019595 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-172-116:1019292:1019590 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-172-116:1019292:1019590 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-172-147:618589:618883 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-172-147:618589:618883 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-172-142:880681:880983 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-172-142:880681:880983 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-172-73:926594:926898 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-172-73:926594:926898 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-172-142:880685:880977 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-172-142:880685:880977 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-172-142:880684:880981 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-172-142:880684:880981 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-172-73:926597:926900 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-172-73:926599:926895 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-172-73:926599:926895 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-172-73:926597:926900 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-172-142:880686:880979 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-172-142:880686:880979 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-172-73:926601:926896 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-172-73:926601:926896 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-172-73:926600:926893 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-172-73:926600:926893 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-172-73:926596:926899 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-172-73:926596:926899 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-160-225:94342:94643 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-160-225:94342:94643 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-172-147:618588:618888 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-172-147:618588:618888 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-172-252:688203:688501 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-172-252:688203:688501 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-172-147:618590:618889 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-172-147:618590:618889 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-172-147:618591:618885 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-172-147:618591:618885 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-172-252:688204:688502 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-172-252:688204:688502 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-172-252:688205:688505 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-172-252:688205:688505 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-173-7:811250:811555 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-173-7:811250:811555 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-172-147:618586:618886 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-172-147:618586:618886 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-173-7:811252:811560 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-173-7:811252:811560 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-173-7:811253:811558 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-173-7:811253:811558 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-172-147:618584:618882 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-172-147:618584:618882 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-172-252:688201:688500 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-172-252:688201:688500 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-160-225:94344:94648 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-160-225:94344:94648 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-172-147:618587:618887 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-172-147:618587:618887 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-173-7:811248:811554 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-173-7:811248:811554 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-172-252:688202:688503 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-172-252:688202:688503 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-172-252:688200:688507 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-172-252:688200:688507 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-172-252:688206:688506 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-172-252:688206:688506 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-172-147:618585:618884 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-172-147:618585:618884 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-160-225:94343:94647 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-160-225:94343:94647 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-172-252:688207:688504 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-172-252:688207:688504 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-160-225:94345:94649 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-160-225:94345:94649 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-160-225:94346:94644 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-160-225:94346:94644 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-160-225:94347:94646 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-160-225:94347:94646 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-160-225:94348:94650 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-160-225:94348:94650 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-160-225:94349:94645 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-160-225:94349:94645 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-160-225:94349:94645 [7] NCCL INFO NVLS Head  0:  0  8 16 24 32 40 48 56
+ip-26-0-160-225:94349:94645 [7] NCCL INFO NVLS Head  1:  1  9 17 25 33 41 49 57
+ip-26-0-160-225:94349:94645 [7] NCCL INFO NVLS Head  2:  2 10 18 26 34 42 50 58
+ip-26-0-160-225:94349:94645 [7] NCCL INFO NVLS Head  3:  3 11 19 27 35 43 51 59
+ip-26-0-160-225:94349:94645 [7] NCCL INFO NVLS Head  4:  4 12 20 28 36 44 52 60
+ip-26-0-160-225:94349:94645 [7] NCCL INFO NVLS Head  5:  5 13 21 29 37 45 53 61
+ip-26-0-160-225:94349:94645 [7] NCCL INFO NVLS Head  6:  6 14 22 30 38 46 54 62
+ip-26-0-160-225:94349:94645 [7] NCCL INFO NVLS Head  7:  7 15 23 31 39 47 55 63
+ip-26-0-160-225:94349:94645 [7] NCCL INFO Trees [0] -1/-1/-1->7->6 [1] 0/-1/-1->7->6 [2] 0/-1/-1->7->6 [3] 0/-1/-1->7->6 [4] 0/-1/-1->7->6 [5] 0/-1/-1->7->6 [6] 0/-1/-1->7->6 [7] 0/39/-1->7->-1 [8] -1/-1/-1->7->6 [9] 0/-1/-1->7->6 [10] 0/-1/-1->7->6 [11] 0/-1/-1->7->6 [12] 0/-1/-1->7->6 [13] 0/-1/-1->7->6 [14] 0/-1/-1->7->6 [15] 0/-1/-1->7->15
+ip-26-0-160-225:94349:94645 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-116:1019288:1019588 [0] NCCL INFO Trees [0] 9/-1/-1->8->16 [1] -1/-1/-1->8->15 [2] 9/-1/-1->8->15 [3] 9/-1/-1->8->15 [4] 9/-1/-1->8->15 [5] 9/-1/-1->8->15 [6] 9/-1/-1->8->15 [7] 9/-1/-1->8->15 [8] 9/16/0->8->24 [9] -1/-1/-1->8->15 [10] 9/-1/-1->8->15 [11] 9/-1/-1->8->15 [12] 9/-1/-1->8->15 [13] 9/-1/-1->8->15 [14] 9/-1/-1->8->15 [15] 9/-1/-1->8->15
+ip-26-0-172-116:1019288:1019588 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-116:1019289:1019593 [1] NCCL INFO Trees [0] 10/-1/-1->9->8 [1] 10/-1/-1->9->17 [2] -1/-1/-1->9->8 [3] 10/-1/-1->9->8 [4] 10/-1/-1->9->8 [5] 10/-1/-1->9->8 [6] 10/-1/-1->9->8 [7] 10/-1/-1->9->8 [8] 10/-1/-1->9->8 [9] 10/17/1->9->25 [10] -1/-1/-1->9->8 [11] 10/-1/-1->9->8 [12] 10/-1/-1->9->8 [13] 10/-1/-1->9->8 [14] 10/-1/-1->9->8 [15] 10/-1/-1->9->8
+ip-26-0-172-116:1019289:1019593 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-116:1019290:1019595 [2] NCCL INFO Trees [0] 11/-1/-1->10->9 [1] 11/-1/-1->10->9 [2] 11/-1/-1->10->18 [3] -1/-1/-1->10->9 [4] 11/-1/-1->10->9 [5] 11/-1/-1->10->9 [6] 11/-1/-1->10->9 [7] 11/-1/-1->10->9 [8] 11/-1/-1->10->9 [9] 11/-1/-1->10->9 [10] 11/18/2->10->26 [11] -1/-1/-1->10->9 [12] 11/-1/-1->10->9 [13] 11/-1/-1->10->9 [14] 11/-1/-1->10->9 [15] 11/-1/-1->10->9
+ip-26-0-172-116:1019290:1019595 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-116:1019292:1019590 [4] NCCL INFO Trees [0] 13/-1/-1->12->11 [1] 13/-1/-1->12->11 [2] 13/-1/-1->12->11 [3] 13/-1/-1->12->11 [4] 13/-1/-1->12->20 [5] -1/-1/-1->12->11 [6] 13/-1/-1->12->11 [7] 13/-1/-1->12->11 [8] 13/-1/-1->12->11 [9] 13/-1/-1->12->11 [10] 13/-1/-1->12->11 [11] 13/-1/-1->12->11 [12] 13/20/4->12->28 [13] -1/-1/-1->12->11 [14] 13/-1/-1->12->11 [15] 13/-1/-1->12->11
+ip-26-0-172-116:1019291:1019594 [3] NCCL INFO Trees [0] 12/-1/-1->11->10 [1] 12/-1/-1->11->10 [2] 12/-1/-1->11->10 [3] 12/-1/-1->11->19 [4] -1/-1/-1->11->10 [5] 12/-1/-1->11->10 [6] 12/-1/-1->11->10 [7] 12/-1/-1->11->10 [8] 12/-1/-1->11->10 [9] 12/-1/-1->11->10 [10] 12/-1/-1->11->10 [11] 12/19/3->11->27 [12] -1/-1/-1->11->10 [13] 12/-1/-1->11->10 [14] 12/-1/-1->11->10 [15] 12/-1/-1->11->10
+ip-26-0-172-116:1019292:1019590 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-116:1019291:1019594 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-142:880679:880980 [0] NCCL INFO Trees [0] 17/8/24->16->32 [1] -1/-1/-1->16->23 [2] 17/-1/-1->16->23 [3] 17/-1/-1->16->23 [4] 17/-1/-1->16->23 [5] 17/-1/-1->16->23 [6] 17/-1/-1->16->23 [7] 17/-1/-1->16->23 [8] 17/-1/-1->16->8 [9] -1/-1/-1->16->23 [10] 17/-1/-1->16->23 [11] 17/-1/-1->16->23 [12] 17/-1/-1->16->23 [13] 17/-1/-1->16->23 [14] 17/-1/-1->16->23 [15] 17/-1/-1->16->23
+ip-26-0-172-142:880679:880980 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-116:1019293:1019592 [5] NCCL INFO Trees [0] 14/-1/-1->13->12 [1] 14/-1/-1->13->12 [2] 14/-1/-1->13->12 [3] 14/-1/-1->13->12 [4] 14/-1/-1->13->12 [5] 14/-1/-1->13->21 [6] -1/-1/-1->13->12 [7] 14/-1/-1->13->12 [8] 14/-1/-1->13->12 [9] 14/-1/-1->13->12 [10] 14/-1/-1->13->12 [11] 14/-1/-1->13->12 [12] 14/-1/-1->13->12 [13] 14/21/5->13->29 [14] -1/-1/-1->13->12 [15] 14/-1/-1->13->12
+ip-26-0-172-116:1019293:1019592 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-116:1019295:1019591 [7] NCCL INFO Trees [0] -1/-1/-1->15->14 [1] 8/-1/-1->15->14 [2] 8/-1/-1->15->14 [3] 8/-1/-1->15->14 [4] 8/-1/-1->15->14 [5] 8/-1/-1->15->14 [6] 8/-1/-1->15->14 [7] 8/-1/-1->15->23 [8] -1/-1/-1->15->14 [9] 8/-1/-1->15->14 [10] 8/-1/-1->15->14 [11] 8/-1/-1->15->14 [12] 8/-1/-1->15->14 [13] 8/-1/-1->15->14 [14] 8/-1/-1->15->14 [15] 8/23/7->15->31
+ip-26-0-172-116:1019295:1019591 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-116:1019294:1019589 [6] NCCL INFO Trees [0] 15/-1/-1->14->13 [1] 15/-1/-1->14->13 [2] 15/-1/-1->14->13 [3] 15/-1/-1->14->13 [4] 15/-1/-1->14->13 [5] 15/-1/-1->14->13 [6] 15/-1/-1->14->22 [7] -1/-1/-1->14->13 [8] 15/-1/-1->14->13 [9] 15/-1/-1->14->13 [10] 15/-1/-1->14->13 [11] 15/-1/-1->14->13 [12] 15/-1/-1->14->13 [13] 15/-1/-1->14->13 [14] 15/22/6->14->30 [15] -1/-1/-1->14->13
+ip-26-0-172-116:1019294:1019589 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-142:880680:880984 [1] NCCL INFO Trees [0] 18/-1/-1->17->16 [1] 18/9/25->17->33 [2] -1/-1/-1->17->16 [3] 18/-1/-1->17->16 [4] 18/-1/-1->17->16 [5] 18/-1/-1->17->16 [6] 18/-1/-1->17->16 [7] 18/-1/-1->17->16 [8] 18/-1/-1->17->16 [9] 18/-1/-1->17->9 [10] -1/-1/-1->17->16 [11] 18/-1/-1->17->16 [12] 18/-1/-1->17->16 [13] 18/-1/-1->17->16 [14] 18/-1/-1->17->16 [15] 18/-1/-1->17->16
+ip-26-0-172-142:880680:880984 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-142:880681:880983 [2] NCCL INFO Trees [0] 19/-1/-1->18->17 [1] 19/-1/-1->18->17 [2] 19/10/26->18->34 [3] -1/-1/-1->18->17 [4] 19/-1/-1->18->17 [5] 19/-1/-1->18->17 [6] 19/-1/-1->18->17 [7] 19/-1/-1->18->17 [8] 19/-1/-1->18->17 [9] 19/-1/-1->18->17 [10] 19/-1/-1->18->10 [11] -1/-1/-1->18->17 [12] 19/-1/-1->18->17 [13] 19/-1/-1->18->17 [14] 19/-1/-1->18->17 [15] 19/-1/-1->18->17
+ip-26-0-172-142:880681:880983 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-142:880682:880978 [3] NCCL INFO Trees [0] 20/-1/-1->19->18 [1] 20/-1/-1->19->18 [2] 20/-1/-1->19->18 [3] 20/11/27->19->35 [4] -1/-1/-1->19->18 [5] 20/-1/-1->19->18 [6] 20/-1/-1->19->18 [7] 20/-1/-1->19->18 [8] 20/-1/-1->19->18 [9] 20/-1/-1->19->18 [10] 20/-1/-1->19->18 [11] 20/-1/-1->19->11 [12] -1/-1/-1->19->18 [13] 20/-1/-1->19->18 [14] 20/-1/-1->19->18 [15] 20/-1/-1->19->18
+ip-26-0-172-147:618586:618886 [2] NCCL INFO Trees [0] 27/-1/-1->26->25 [1] 27/-1/-1->26->25 [2] 27/-1/-1->26->18 [3] -1/-1/-1->26->25 [4] 27/-1/-1->26->25 [5] 27/-1/-1->26->25 [6] 27/-1/-1->26->25 [7] 27/-1/-1->26->25 [8] 27/-1/-1->26->25 [9] 27/-1/-1->26->25 [10] 27/42/10->26->58 [11] -1/-1/-1->26->25 [12] 27/-1/-1->26->25 [13] 27/-1/-1->26->25 [14] 27/-1/-1->26->25 [15] 27/-1/-1->26->25
+ip-26-0-172-147:618586:618886 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-147:618585:618884 [1] NCCL INFO Trees [0] 26/-1/-1->25->24 [1] 26/-1/-1->25->17 [2] -1/-1/-1->25->24 [3] 26/-1/-1->25->24 [4] 26/-1/-1->25->24 [5] 26/-1/-1->25->24 [6] 26/-1/-1->25->24 [7] 26/-1/-1->25->24 [8] 26/-1/-1->25->24 [9] 26/41/9->25->57 [10] -1/-1/-1->25->24 [11] 26/-1/-1->25->24 [12] 26/-1/-1->25->24 [13] 26/-1/-1->25->24 [14] 26/-1/-1->25->24 [15] 26/-1/-1->25->24
+ip-26-0-172-142:880682:880978 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-142:880683:880982 [4] NCCL INFO Trees [0] 21/-1/-1->20->19 [1] 21/-1/-1->20->19 [2] 21/-1/-1->20->19 [3] 21/-1/-1->20->19 [4] 21/12/28->20->36 [5] -1/-1/-1->20->19 [6] 21/-1/-1->20->19 [7] 21/-1/-1->20->19 [8] 21/-1/-1->20->19 [9] 21/-1/-1->20->19 [10] 21/-1/-1->20->19 [11] 21/-1/-1->20->19 [12] 21/-1/-1->20->12 [13] -1/-1/-1->20->19 [14] 21/-1/-1->20->19 [15] 21/-1/-1->20->19
+ip-26-0-172-142:880685:880977 [6] NCCL INFO Trees [0] 23/-1/-1->22->21 [1] 23/-1/-1->22->21 [2] 23/-1/-1->22->21 [3] 23/-1/-1->22->21 [4] 23/-1/-1->22->21 [5] 23/-1/-1->22->21 [6] 23/14/30->22->38 [7] -1/-1/-1->22->21 [8] 23/-1/-1->22->21 [9] 23/-1/-1->22->21 [10] 23/-1/-1->22->21 [11] 23/-1/-1->22->21 [12] 23/-1/-1->22->21 [13] 23/-1/-1->22->21 [14] 23/-1/-1->22->14 [15] -1/-1/-1->22->21
+ip-26-0-172-142:880683:880982 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-142:880685:880977 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-142:880686:880979 [7] NCCL INFO Trees [0] -1/-1/-1->23->22 [1] 16/-1/-1->23->22 [2] 16/-1/-1->23->22 [3] 16/-1/-1->23->22 [4] 16/-1/-1->23->22 [5] 16/-1/-1->23->22 [6] 16/-1/-1->23->22 [7] 16/15/31->23->39 [8] -1/-1/-1->23->22 [9] 16/-1/-1->23->22 [10] 16/-1/-1->23->22 [11] 16/-1/-1->23->22 [12] 16/-1/-1->23->22 [13] 16/-1/-1->23->22 [14] 16/-1/-1->23->22 [15] 16/-1/-1->23->15
+ip-26-0-172-142:880684:880981 [5] NCCL INFO Trees [0] 22/-1/-1->21->20 [1] 22/-1/-1->21->20 [2] 22/-1/-1->21->20 [3] 22/-1/-1->21->20 [4] 22/-1/-1->21->20 [5] 22/13/29->21->37 [6] -1/-1/-1->21->20 [7] 22/-1/-1->21->20 [8] 22/-1/-1->21->20 [9] 22/-1/-1->21->20 [10] 22/-1/-1->21->20 [11] 22/-1/-1->21->20 [12] 22/-1/-1->21->20 [13] 22/-1/-1->21->13 [14] -1/-1/-1->21->20 [15] 22/-1/-1->21->20
+ip-26-0-172-142:880686:880979 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-142:880684:880981 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-147:618585:618884 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-147:618584:618882 [0] NCCL INFO Trees [0] 25/-1/-1->24->16 [1] -1/-1/-1->24->31 [2] 25/-1/-1->24->31 [3] 25/-1/-1->24->31 [4] 25/-1/-1->24->31 [5] 25/-1/-1->24->31 [6] 25/-1/-1->24->31 [7] 25/-1/-1->24->31 [8] 25/40/8->24->56 [9] -1/-1/-1->24->31 [10] 25/-1/-1->24->31 [11] 25/-1/-1->24->31 [12] 25/-1/-1->24->31 [13] 25/-1/-1->24->31 [14] 25/-1/-1->24->31 [15] 25/-1/-1->24->31
+ip-26-0-172-147:618584:618882 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-147:618587:618887 [3] NCCL INFO Trees [0] 28/-1/-1->27->26 [1] 28/-1/-1->27->26 [2] 28/-1/-1->27->26 [3] 28/-1/-1->27->19 [4] -1/-1/-1->27->26 [5] 28/-1/-1->27->26 [6] 28/-1/-1->27->26 [7] 28/-1/-1->27->26 [8] 28/-1/-1->27->26 [9] 28/-1/-1->27->26 [10] 28/-1/-1->27->26 [11] 28/43/11->27->59 [12] -1/-1/-1->27->26 [13] 28/-1/-1->27->26 [14] 28/-1/-1->27->26 [15] 28/-1/-1->27->26
+ip-26-0-172-147:618587:618887 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-147:618588:618888 [4] NCCL INFO Trees [0] 29/-1/-1->28->27 [1] 29/-1/-1->28->27 [2] 29/-1/-1->28->27 [3] 29/-1/-1->28->27 [4] 29/-1/-1->28->20 [5] -1/-1/-1->28->27 [6] 29/-1/-1->28->27 [7] 29/-1/-1->28->27 [8] 29/-1/-1->28->27 [9] 29/-1/-1->28->27 [10] 29/-1/-1->28->27 [11] 29/-1/-1->28->27 [12] 29/44/12->28->60 [13] -1/-1/-1->28->27 [14] 29/-1/-1->28->27 [15] 29/-1/-1->28->27
+ip-26-0-172-147:618588:618888 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-252:688206:688506 [6] NCCL INFO Trees [0] 39/-1/-1->38->37 [1] 39/-1/-1->38->37 [2] 39/-1/-1->38->37 [3] 39/-1/-1->38->37 [4] 39/-1/-1->38->37 [5] 39/-1/-1->38->37 [6] 39/22/54->38->6 [7] -1/-1/-1->38->37 [8] 39/-1/-1->38->37 [9] 39/-1/-1->38->37 [10] 39/-1/-1->38->37 [11] 39/-1/-1->38->37 [12] 39/-1/-1->38->37 [13] 39/-1/-1->38->37 [14] 39/-1/-1->38->46 [15] -1/-1/-1->38->37
+ip-26-0-172-252:688207:688504 [7] NCCL INFO Trees [0] -1/-1/-1->39->38 [1] 32/-1/-1->39->38 [2] 32/-1/-1->39->38 [3] 32/-1/-1->39->38 [4] 32/-1/-1->39->38 [5] 32/-1/-1->39->38 [6] 32/-1/-1->39->38 [7] 32/23/55->39->7 [8] -1/-1/-1->39->38 [9] 32/-1/-1->39->38 [10] 32/-1/-1->39->38 [11] 32/-1/-1->39->38 [12] 32/-1/-1->39->38 [13] 32/-1/-1->39->38 [14] 32/-1/-1->39->38 [15] 32/-1/-1->39->47
+ip-26-0-172-147:618590:618889 [6] NCCL INFO Trees [0] 31/-1/-1->30->29 [1] 31/-1/-1->30->29 [2] 31/-1/-1->30->29 [3] 31/-1/-1->30->29 [4] 31/-1/-1->30->29 [5] 31/-1/-1->30->29 [6] 31/-1/-1->30->22 [7] -1/-1/-1->30->29 [8] 31/-1/-1->30->29 [9] 31/-1/-1->30->29 [10] 31/-1/-1->30->29 [11] 31/-1/-1->30->29 [12] 31/-1/-1->30->29 [13] 31/-1/-1->30->29 [14] 31/46/14->30->62 [15] -1/-1/-1->30->29
+ip-26-0-172-147:618589:618883 [5] NCCL INFO Trees [0] 30/-1/-1->29->28 [1] 30/-1/-1->29->28 [2] 30/-1/-1->29->28 [3] 30/-1/-1->29->28 [4] 30/-1/-1->29->28 [5] 30/-1/-1->29->21 [6] -1/-1/-1->29->28 [7] 30/-1/-1->29->28 [8] 30/-1/-1->29->28 [9] 30/-1/-1->29->28 [10] 30/-1/-1->29->28 [11] 30/-1/-1->29->28 [12] 30/-1/-1->29->28 [13] 30/45/13->29->61 [14] -1/-1/-1->29->28 [15] 30/-1/-1->29->28
+ip-26-0-172-147:618590:618889 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-147:618589:618883 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-57:942613:942917 [2] NCCL INFO Trees [0] 43/-1/-1->42->41 [1] 43/-1/-1->42->41 [2] 43/-1/-1->42->50 [3] -1/-1/-1->42->41 [4] 43/-1/-1->42->41 [5] 43/-1/-1->42->41 [6] 43/-1/-1->42->41 [7] 43/-1/-1->42->41 [8] 43/-1/-1->42->41 [9] 43/-1/-1->42->41 [10] 43/50/34->42->26 [11] -1/-1/-1->42->41 [12] 43/-1/-1->42->41 [13] 43/-1/-1->42->41 [14] 43/-1/-1->42->41 [15] 43/-1/-1->42->41
+ip-26-0-172-57:942613:942917 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-147:618591:618885 [7] NCCL INFO Trees [0] -1/-1/-1->31->30 [1] 24/-1/-1->31->30 [2] 24/-1/-1->31->30 [3] 24/-1/-1->31->30 [4] 24/-1/-1->31->30 [5] 24/-1/-1->31->30 [6] 24/-1/-1->31->30 [7] 24/-1/-1->31->23 [8] -1/-1/-1->31->30 [9] 24/-1/-1->31->30 [10] 24/-1/-1->31->30 [11] 24/-1/-1->31->30 [12] 24/-1/-1->31->30 [13] 24/-1/-1->31->30 [14] 24/-1/-1->31->30 [15] 24/47/15->31->63
+ip-26-0-172-147:618591:618885 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-225:94343:94647 [1] NCCL INFO NVLS Head  0:  0  8 16 24 32 40 48 56
+ip-26-0-160-225:94343:94647 [1] NCCL INFO NVLS Head  1:  1  9 17 25 33 41 49 57
+ip-26-0-160-225:94343:94647 [1] NCCL INFO NVLS Head  2:  2 10 18 26 34 42 50 58
+ip-26-0-160-225:94343:94647 [1] NCCL INFO NVLS Head  3:  3 11 19 27 35 43 51 59
+ip-26-0-160-225:94343:94647 [1] NCCL INFO NVLS Head  4:  4 12 20 28 36 44 52 60
+ip-26-0-160-225:94344:94648 [2] NCCL INFO NVLS Head  0:  0  8 16 24 32 40 48 56
+ip-26-0-160-225:94343:94647 [1] NCCL INFO NVLS Head  5:  5 13 21 29 37 45 53 61
+ip-26-0-172-252:688201:688500 [1] NCCL INFO Trees [0] 34/-1/-1->33->32 [1] 34/17/49->33->1 [2] -1/-1/-1->33->32 [3] 34/-1/-1->33->32 [4] 34/-1/-1->33->32 [5] 34/-1/-1->33->32 [6] 34/-1/-1->33->32 [7] 34/-1/-1->33->32 [8] 34/-1/-1->33->32 [9] 34/-1/-1->33->41 [10] -1/-1/-1->33->32 [11] 34/-1/-1->33->32 [12] 34/-1/-1->33->32 [13] 34/-1/-1->33->32 [14] 34/-1/-1->33->32 [15] 34/-1/-1->33->32
+ip-26-0-172-252:688207:688504 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-252:688206:688506 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-252:688201:688500 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-252:688202:688503 [2] NCCL INFO Trees [0] 35/-1/-1->34->33 [1] 35/-1/-1->34->33 [2] 35/18/50->34->2 [3] -1/-1/-1->34->33 [4] 35/-1/-1->34->33 [5] 35/-1/-1->34->33 [6] 35/-1/-1->34->33 [7] 35/-1/-1->34->33 [8] 35/-1/-1->34->33 [9] 35/-1/-1->34->33 [10] 35/-1/-1->34->42 [11] -1/-1/-1->34->33 [12] 35/-1/-1->34->33 [13] 35/-1/-1->34->33 [14] 35/-1/-1->34->33 [15] 35/-1/-1->34->33
+ip-26-0-160-225:94343:94647 [1] NCCL INFO NVLS Head  6:  6 14 22 30 38 46 54 62
+ip-26-0-160-225:94348:94650 [6] NCCL INFO NVLS Head  0:  0  8 16 24 32 40 48 56
+ip-26-0-160-225:94344:94648 [2] NCCL INFO NVLS Head  1:  1  9 17 25 33 41 49 57
+ip-26-0-160-225:94343:94647 [1] NCCL INFO NVLS Head  7:  7 15 23 31 39 47 55 63
+ip-26-0-160-225:94345:94649 [3] NCCL INFO NVLS Head  0:  0  8 16 24 32 40 48 56
+ip-26-0-160-225:94348:94650 [6] NCCL INFO NVLS Head  1:  1  9 17 25 33 41 49 57
+ip-26-0-160-225:94344:94648 [2] NCCL INFO NVLS Head  2:  2 10 18 26 34 42 50 58
+ip-26-0-160-225:94346:94644 [4] NCCL INFO NVLS Head  0:  0  8 16 24 32 40 48 56
+ip-26-0-160-225:94345:94649 [3] NCCL INFO NVLS Head  1:  1  9 17 25 33 41 49 57
+ip-26-0-160-225:94344:94648 [2] NCCL INFO NVLS Head  3:  3 11 19 27 35 43 51 59
+ip-26-0-160-225:94348:94650 [6] NCCL INFO NVLS Head  2:  2 10 18 26 34 42 50 58
+ip-26-0-160-225:94347:94646 [5] NCCL INFO NVLS Head  0:  0  8 16 24 32 40 48 56
+ip-26-0-172-57:942612:942914 [1] NCCL INFO Trees [0] 42/-1/-1->41->40 [1] 42/-1/-1->41->49 [2] -1/-1/-1->41->40 [3] 42/-1/-1->41->40 [4] 42/-1/-1->41->40 [5] 42/-1/-1->41->40 [6] 42/-1/-1->41->40 [7] 42/-1/-1->41->40 [8] 42/-1/-1->41->40 [9] 42/49/33->41->25 [10] -1/-1/-1->41->40 [11] 42/-1/-1->41->40 [12] 42/-1/-1->41->40 [13] 42/-1/-1->41->40 [14] 42/-1/-1->41->40 [15] 42/-1/-1->41->40
+ip-26-0-172-252:688202:688503 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-173-7:811255:811561 [7] NCCL INFO Trees [0] -1/-1/-1->63->62 [1] 56/-1/-1->63->62 [2] 56/-1/-1->63->62 [3] 56/-1/-1->63->62 [4] 56/-1/-1->63->62 [5] 56/-1/-1->63->62 [6] 56/-1/-1->63->62 [7] 56/-1/-1->63->55 [8] -1/-1/-1->63->62 [9] 56/-1/-1->63->62 [10] 56/-1/-1->63->62 [11] 56/-1/-1->63->62 [12] 56/-1/-1->63->62 [13] 56/-1/-1->63->62 [14] 56/-1/-1->63->62 [15] 56/31/-1->63->-1
+ip-26-0-173-7:811255:811561 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-57:942612:942914 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-73:926600:926893 [6] NCCL INFO Trees [0] 55/-1/-1->54->53 [1] 55/-1/-1->54->53 [2] 55/-1/-1->54->53 [3] 55/-1/-1->54->53 [4] 55/-1/-1->54->53 [5] 55/-1/-1->54->53 [6] 55/46/62->54->38 [7] -1/-1/-1->54->53 [8] 55/-1/-1->54->53 [9] 55/-1/-1->54->53 [10] 55/-1/-1->54->53 [11] 55/-1/-1->54->53 [12] 55/-1/-1->54->53 [13] 55/-1/-1->54->53 [14] 55/-1/-1->54->46 [15] -1/-1/-1->54->53
+ip-26-0-172-73:926600:926893 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-73:926601:926896 [7] NCCL INFO Trees [0] -1/-1/-1->55->54 [1] 48/-1/-1->55->54 [2] 48/-1/-1->55->54 [3] 48/-1/-1->55->54 [4] 48/-1/-1->55->54 [5] 48/-1/-1->55->54 [6] 48/-1/-1->55->54 [7] 48/47/63->55->39 [8] -1/-1/-1->55->54 [9] 48/-1/-1->55->54 [10] 48/-1/-1->55->54 [11] 48/-1/-1->55->54 [12] 48/-1/-1->55->54 [13] 48/-1/-1->55->54 [14] 48/-1/-1->55->54 [15] 48/-1/-1->55->47
+ip-26-0-172-252:688200:688507 [0] NCCL INFO Trees [0] 33/16/48->32->0 [1] -1/-1/-1->32->39 [2] 33/-1/-1->32->39 [3] 33/-1/-1->32->39 [4] 33/-1/-1->32->39 [5] 33/-1/-1->32->39 [6] 33/-1/-1->32->39 [7] 33/-1/-1->32->39 [8] 33/-1/-1->32->40 [9] -1/-1/-1->32->39 [10] 33/-1/-1->32->39 [11] 33/-1/-1->32->39 [12] 33/-1/-1->32->39 [13] 33/-1/-1->32->39 [14] 33/-1/-1->32->39 [15] 33/-1/-1->32->39
+ip-26-0-172-252:688204:688502 [4] NCCL INFO Trees [0] 37/-1/-1->36->35 [1] 37/-1/-1->36->35 [2] 37/-1/-1->36->35 [3] 37/-1/-1->36->35 [4] 37/20/52->36->4 [5] -1/-1/-1->36->35 [6] 37/-1/-1->36->35 [7] 37/-1/-1->36->35 [8] 37/-1/-1->36->35 [9] 37/-1/-1->36->35 [10] 37/-1/-1->36->35 [11] 37/-1/-1->36->35 [12] 37/-1/-1->36->44 [13] -1/-1/-1->36->35 [14] 37/-1/-1->36->35 [15] 37/-1/-1->36->35
+ip-26-0-160-225:94345:94649 [3] NCCL INFO NVLS Head  2:  2 10 18 26 34 42 50 58
+ip-26-0-160-225:94344:94648 [2] NCCL INFO NVLS Head  4:  4 12 20 28 36 44 52 60
+ip-26-0-160-225:94346:94644 [4] NCCL INFO NVLS Head  1:  1  9 17 25 33 41 49 57
+ip-26-0-160-225:94342:94643 [0] NCCL INFO NVLS Head  0:  0  8 16 24 32 40 48 56
+ip-26-0-160-225:94345:94649 [3] NCCL INFO NVLS Head  3:  3 11 19 27 35 43 51 59
+ip-26-0-160-225:94347:94646 [5] NCCL INFO NVLS Head  1:  1  9 17 25 33 41 49 57
+ip-26-0-160-225:94348:94650 [6] NCCL INFO NVLS Head  3:  3 11 19 27 35 43 51 59
+ip-26-0-160-225:94344:94648 [2] NCCL INFO NVLS Head  5:  5 13 21 29 37 45 53 61
+ip-26-0-160-225:94346:94644 [4] NCCL INFO NVLS Head  2:  2 10 18 26 34 42 50 58
+ip-26-0-160-225:94342:94643 [0] NCCL INFO NVLS Head  1:  1  9 17 25 33 41 49 57
+ip-26-0-160-225:94345:94649 [3] NCCL INFO NVLS Head  4:  4 12 20 28 36 44 52 60
+ip-26-0-160-225:94344:94648 [2] NCCL INFO NVLS Head  6:  6 14 22 30 38 46 54 62
+ip-26-0-172-57:942611:942911 [0] NCCL INFO Trees [0] 41/-1/-1->40->48 [1] -1/-1/-1->40->47 [2] 41/-1/-1->40->47 [3] 41/-1/-1->40->47 [4] 41/-1/-1->40->47 [5] 41/-1/-1->40->47 [6] 41/-1/-1->40->47 [7] 41/-1/-1->40->47 [8] 41/48/32->40->24 [9] -1/-1/-1->40->47 [10] 41/-1/-1->40->47 [11] 41/-1/-1->40->47 [12] 41/-1/-1->40->47 [13] 41/-1/-1->40->47 [14] 41/-1/-1->40->47 [15] 41/-1/-1->40->47
+ip-26-0-172-57:942611:942911 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-252:688203:688501 [3] NCCL INFO Trees [0] 36/-1/-1->35->34 [1] 36/-1/-1->35->34 [2] 36/-1/-1->35->34 [3] 36/19/51->35->3 [4] -1/-1/-1->35->34 [5] 36/-1/-1->35->34 [6] 36/-1/-1->35->34 [7] 36/-1/-1->35->34 [8] 36/-1/-1->35->34 [9] 36/-1/-1->35->34 [10] 36/-1/-1->35->34 [11] 36/-1/-1->35->43 [12] -1/-1/-1->35->34 [13] 36/-1/-1->35->34 [14] 36/-1/-1->35->34 [15] 36/-1/-1->35->34
+ip-26-0-172-252:688200:688507 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-252:688204:688502 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-252:688203:688501 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-252:688205:688505 [5] NCCL INFO Trees [0] 38/-1/-1->37->36 [1] 38/-1/-1->37->36 [2] 38/-1/-1->37->36 [3] 38/-1/-1->37->36 [4] 38/-1/-1->37->36 [5] 38/21/53->37->5 [6] -1/-1/-1->37->36 [7] 38/-1/-1->37->36 [8] 38/-1/-1->37->36 [9] 38/-1/-1->37->36 [10] 38/-1/-1->37->36 [11] 38/-1/-1->37->36 [12] 38/-1/-1->37->36 [13] 38/-1/-1->37->45 [14] -1/-1/-1->37->36 [15] 38/-1/-1->37->36
+ip-26-0-160-225:94348:94650 [6] NCCL INFO NVLS Head  4:  4 12 20 28 36 44 52 60
+ip-26-0-160-225:94347:94646 [5] NCCL INFO NVLS Head  2:  2 10 18 26 34 42 50 58
+ip-26-0-160-225:94346:94644 [4] NCCL INFO NVLS Head  3:  3 11 19 27 35 43 51 59
+ip-26-0-160-225:94348:94650 [6] NCCL INFO NVLS Head  5:  5 13 21 29 37 45 53 61
+ip-26-0-160-225:94347:94646 [5] NCCL INFO NVLS Head  3:  3 11 19 27 35 43 51 59
+ip-26-0-160-225:94348:94650 [6] NCCL INFO NVLS Head  6:  6 14 22 30 38 46 54 62
+ip-26-0-160-225:94347:94646 [5] NCCL INFO NVLS Head  4:  4 12 20 28 36 44 52 60
+ip-26-0-160-225:94345:94649 [3] NCCL INFO NVLS Head  5:  5 13 21 29 37 45 53 61
+ip-26-0-160-225:94344:94648 [2] NCCL INFO NVLS Head  7:  7 15 23 31 39 47 55 63
+ip-26-0-160-225:94346:94644 [4] NCCL INFO NVLS Head  4:  4 12 20 28 36 44 52 60
+ip-26-0-173-7:811249:811559 [1] NCCL INFO Trees [0] 58/-1/-1->57->56 [1] 58/-1/-1->57->49 [2] -1/-1/-1->57->56 [3] 58/-1/-1->57->56 [4] 58/-1/-1->57->56 [5] 58/-1/-1->57->56 [6] 58/-1/-1->57->56 [7] 58/-1/-1->57->56 [8] 58/-1/-1->57->56 [9] 58/25/-1->57->-1 [10] -1/-1/-1->57->56 [11] 58/-1/-1->57->56 [12] 58/-1/-1->57->56 [13] 58/-1/-1->57->56 [14] 58/-1/-1->57->56 [15] 58/-1/-1->57->56
+ip-26-0-173-7:811248:811554 [0] NCCL INFO Trees [0] 57/-1/-1->56->48 [1] -1/-1/-1->56->63 [2] 57/-1/-1->56->63 [3] 57/-1/-1->56->63 [4] 57/-1/-1->56->63 [5] 57/-1/-1->56->63 [6] 57/-1/-1->56->63 [7] 57/-1/-1->56->63 [8] 57/24/-1->56->-1 [9] -1/-1/-1->56->63 [10] 57/-1/-1->56->63 [11] 57/-1/-1->56->63 [12] 57/-1/-1->56->63 [13] 57/-1/-1->56->63 [14] 57/-1/-1->56->63 [15] 57/-1/-1->56->63
+ip-26-0-173-7:811249:811559 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-173-7:811248:811554 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-57:942617:942916 [6] NCCL INFO Trees [0] 47/-1/-1->46->45 [1] 47/-1/-1->46->45 [2] 47/-1/-1->46->45 [3] 47/-1/-1->46->45 [4] 47/-1/-1->46->45 [5] 47/-1/-1->46->45 [6] 47/-1/-1->46->54 [7] -1/-1/-1->46->45 [8] 47/-1/-1->46->45 [9] 47/-1/-1->46->45 [10] 47/-1/-1->46->45 [11] 47/-1/-1->46->45 [12] 47/-1/-1->46->45 [13] 47/-1/-1->46->45 [14] 47/54/38->46->30 [15] -1/-1/-1->46->45
+ip-26-0-172-57:942618:942918 [7] NCCL INFO Trees [0] -1/-1/-1->47->46 [1] 40/-1/-1->47->46 [2] 40/-1/-1->47->46 [3] 40/-1/-1->47->46 [4] 40/-1/-1->47->46 [5] 40/-1/-1->47->46 [6] 40/-1/-1->47->46 [7] 40/-1/-1->47->55 [8] -1/-1/-1->47->46 [9] 40/-1/-1->47->46 [10] 40/-1/-1->47->46 [11] 40/-1/-1->47->46 [12] 40/-1/-1->47->46 [13] 40/-1/-1->47->46 [14] 40/-1/-1->47->46 [15] 40/55/39->47->31
+ip-26-0-172-57:942617:942916 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-57:942618:942918 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-73:926601:926896 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-73:926599:926895 [5] NCCL INFO Trees [0] 54/-1/-1->53->52 [1] 54/-1/-1->53->52 [2] 54/-1/-1->53->52 [3] 54/-1/-1->53->52 [4] 54/-1/-1->53->52 [5] 54/45/61->53->37 [6] -1/-1/-1->53->52 [7] 54/-1/-1->53->52 [8] 54/-1/-1->53->52 [9] 54/-1/-1->53->52 [10] 54/-1/-1->53->52 [11] 54/-1/-1->53->52 [12] 54/-1/-1->53->52 [13] 54/-1/-1->53->45 [14] -1/-1/-1->53->52 [15] 54/-1/-1->53->52
+ip-26-0-172-73:926599:926895 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-252:688205:688505 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-225:94343:94647 [1] NCCL INFO Trees [0] 2/-1/-1->1->0 [1] 2/33/-1->1->-1 [2] -1/-1/-1->1->0 [3] 2/-1/-1->1->0 [4] 2/-1/-1->1->0 [5] 2/-1/-1->1->0 [6] 2/-1/-1->1->0 [7] 2/-1/-1->1->0 [8] 2/-1/-1->1->0 [9] 2/-1/-1->1->9 [10] -1/-1/-1->1->0 [11] 2/-1/-1->1->0 [12] 2/-1/-1->1->0 [13] 2/-1/-1->1->0 [14] 2/-1/-1->1->0 [15] 2/-1/-1->1->0
+ip-26-0-173-7:811254:811557 [6] NCCL INFO Trees [0] 63/-1/-1->62->61 [1] 63/-1/-1->62->61 [2] 63/-1/-1->62->61 [3] 63/-1/-1->62->61 [4] 63/-1/-1->62->61 [5] 63/-1/-1->62->61 [6] 63/-1/-1->62->54 [7] -1/-1/-1->62->61 [8] 63/-1/-1->62->61 [9] 63/-1/-1->62->61 [10] 63/-1/-1->62->61 [11] 63/-1/-1->62->61 [12] 63/-1/-1->62->61 [13] 63/-1/-1->62->61 [14] 63/30/-1->62->-1 [15] -1/-1/-1->62->61
+ip-26-0-173-7:811254:811557 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-57:942616:942913 [5] NCCL INFO Trees [0] 46/-1/-1->45->44 [1] 46/-1/-1->45->44 [2] 46/-1/-1->45->44 [3] 46/-1/-1->45->44 [4] 46/-1/-1->45->44 [5] 46/-1/-1->45->53 [6] -1/-1/-1->45->44 [7] 46/-1/-1->45->44 [8] 46/-1/-1->45->44 [9] 46/-1/-1->45->44 [10] 46/-1/-1->45->44 [11] 46/-1/-1->45->44 [12] 46/-1/-1->45->44 [13] 46/53/37->45->29 [14] -1/-1/-1->45->44 [15] 46/-1/-1->45->44
+ip-26-0-172-57:942614:942915 [3] NCCL INFO Trees [0] 44/-1/-1->43->42 [1] 44/-1/-1->43->42 [2] 44/-1/-1->43->42 [3] 44/-1/-1->43->51 [4] -1/-1/-1->43->42 [5] 44/-1/-1->43->42 [6] 44/-1/-1->43->42 [7] 44/-1/-1->43->42 [8] 44/-1/-1->43->42 [9] 44/-1/-1->43->42 [10] 44/-1/-1->43->42 [11] 44/51/35->43->27 [12] -1/-1/-1->43->42 [13] 44/-1/-1->43->42 [14] 44/-1/-1->43->42 [15] 44/-1/-1->43->42
+ip-26-0-172-73:926595:926897 [1] NCCL INFO Trees [0] 50/-1/-1->49->48 [1] 50/41/57->49->33 [2] -1/-1/-1->49->48 [3] 50/-1/-1->49->48 [4] 50/-1/-1->49->48 [5] 50/-1/-1->49->48 [6] 50/-1/-1->49->48 [7] 50/-1/-1->49->48 [8] 50/-1/-1->49->48 [9] 50/-1/-1->49->41 [10] -1/-1/-1->49->48 [11] 50/-1/-1->49->48 [12] 50/-1/-1->49->48 [13] 50/-1/-1->49->48 [14] 50/-1/-1->49->48 [15] 50/-1/-1->49->48
+ip-26-0-160-225:94348:94650 [6] NCCL INFO NVLS Head  7:  7 15 23 31 39 47 55 63
+ip-26-0-160-225:94347:94646 [5] NCCL INFO NVLS Head  5:  5 13 21 29 37 45 53 61
+ip-26-0-160-225:94345:94649 [3] NCCL INFO NVLS Head  6:  6 14 22 30 38 46 54 62
+ip-26-0-160-225:94343:94647 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-225:94346:94644 [4] NCCL INFO NVLS Head  5:  5 13 21 29 37 45 53 61
+ip-26-0-160-225:94342:94643 [0] NCCL INFO NVLS Head  2:  2 10 18 26 34 42 50 58
+ip-26-0-160-225:94347:94646 [5] NCCL INFO NVLS Head  6:  6 14 22 30 38 46 54 62
+ip-26-0-160-225:94345:94649 [3] NCCL INFO NVLS Head  7:  7 15 23 31 39 47 55 63
+ip-26-0-160-225:94346:94644 [4] NCCL INFO NVLS Head  6:  6 14 22 30 38 46 54 62
+ip-26-0-160-225:94342:94643 [0] NCCL INFO NVLS Head  3:  3 11 19 27 35 43 51 59
+ip-26-0-160-225:94347:94646 [5] NCCL INFO NVLS Head  7:  7 15 23 31 39 47 55 63
+ip-26-0-160-225:94346:94644 [4] NCCL INFO NVLS Head  7:  7 15 23 31 39 47 55 63
+ip-26-0-173-7:811250:811555 [2] NCCL INFO Trees [0] 59/-1/-1->58->57 [1] 59/-1/-1->58->57 [2] 59/-1/-1->58->50 [3] -1/-1/-1->58->57 [4] 59/-1/-1->58->57 [5] 59/-1/-1->58->57 [6] 59/-1/-1->58->57 [7] 59/-1/-1->58->57 [8] 59/-1/-1->58->57 [9] 59/-1/-1->58->57 [10] 59/26/-1->58->-1 [11] -1/-1/-1->58->57 [12] 59/-1/-1->58->57 [13] 59/-1/-1->58->57 [14] 59/-1/-1->58->57 [15] 59/-1/-1->58->57
+ip-26-0-173-7:811252:811560 [4] NCCL INFO Trees [0] 61/-1/-1->60->59 [1] 61/-1/-1->60->59 [2] 61/-1/-1->60->59 [3] 61/-1/-1->60->59 [4] 61/-1/-1->60->52 [5] -1/-1/-1->60->59 [6] 61/-1/-1->60->59 [7] 61/-1/-1->60->59 [8] 61/-1/-1->60->59 [9] 61/-1/-1->60->59 [10] 61/-1/-1->60->59 [11] 61/-1/-1->60->59 [12] 61/28/-1->60->-1 [13] -1/-1/-1->60->59 [14] 61/-1/-1->60->59 [15] 61/-1/-1->60->59
+ip-26-0-172-57:942615:942912 [4] NCCL INFO Trees [0] 45/-1/-1->44->43 [1] 45/-1/-1->44->43 [2] 45/-1/-1->44->43 [3] 45/-1/-1->44->43 [4] 45/-1/-1->44->52 [5] -1/-1/-1->44->43 [6] 45/-1/-1->44->43 [7] 45/-1/-1->44->43 [8] 45/-1/-1->44->43 [9] 45/-1/-1->44->43 [10] 45/-1/-1->44->43 [11] 45/-1/-1->44->43 [12] 45/52/36->44->28 [13] -1/-1/-1->44->43 [14] 45/-1/-1->44->43 [15] 45/-1/-1->44->43
+ip-26-0-172-57:942616:942913 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-57:942614:942915 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-57:942615:942912 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-73:926595:926897 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-225:94342:94643 [0] NCCL INFO NVLS Head  4:  4 12 20 28 36 44 52 60
+ip-26-0-160-225:94342:94643 [0] NCCL INFO NVLS Head  5:  5 13 21 29 37 45 53 61
+ip-26-0-160-225:94342:94643 [0] NCCL INFO NVLS Head  6:  6 14 22 30 38 46 54 62
+ip-26-0-160-225:94342:94643 [0] NCCL INFO NVLS Head  7:  7 15 23 31 39 47 55 63
+ip-26-0-160-225:94342:94643 [0] NCCL INFO Channel 00/16 :    0   1   2   3   4   5   6   7   8   9  10  11  12  13  14  15  16  17  18  19
+ip-26-0-160-225:94344:94648 [2] NCCL INFO Trees [0] 3/-1/-1->2->1 [1] 3/-1/-1->2->1 [2] 3/34/-1->2->-1 [3] -1/-1/-1->2->1 [4] 3/-1/-1->2->1 [5] 3/-1/-1->2->1 [6] 3/-1/-1->2->1 [7] 3/-1/-1->2->1 [8] 3/-1/-1->2->1 [9] 3/-1/-1->2->1 [10] 3/-1/-1->2->10 [11] -1/-1/-1->2->1 [12] 3/-1/-1->2->1 [13] 3/-1/-1->2->1 [14] 3/-1/-1->2->1 [15] 3/-1/-1->2->1
+ip-26-0-160-225:94342:94643 [0] NCCL INFO Channel 01/16 :    0   7   6   5   4   3   2   9   8  15  14  13  12  11  10  17  16  23  22  21
+ip-26-0-173-7:811253:811558 [5] NCCL INFO Trees [0] 62/-1/-1->61->60 [1] 62/-1/-1->61->60 [2] 62/-1/-1->61->60 [3] 62/-1/-1->61->60 [4] 62/-1/-1->61->60 [5] 62/-1/-1->61->53 [6] -1/-1/-1->61->60 [7] 62/-1/-1->61->60 [8] 62/-1/-1->61->60 [9] 62/-1/-1->61->60 [10] 62/-1/-1->61->60 [11] 62/-1/-1->61->60 [12] 62/-1/-1->61->60 [13] 62/29/-1->61->-1 [14] -1/-1/-1->61->60 [15] 62/-1/-1->61->60
+ip-26-0-173-7:811250:811555 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-173-7:811252:811560 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-173-7:811253:811558 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-73:926598:926894 [4] NCCL INFO Trees [0] 53/-1/-1->52->51 [1] 53/-1/-1->52->51 [2] 53/-1/-1->52->51 [3] 53/-1/-1->52->51 [4] 53/44/60->52->36 [5] -1/-1/-1->52->51 [6] 53/-1/-1->52->51 [7] 53/-1/-1->52->51 [8] 53/-1/-1->52->51 [9] 53/-1/-1->52->51 [10] 53/-1/-1->52->51 [11] 53/-1/-1->52->51 [12] 53/-1/-1->52->44 [13] -1/-1/-1->52->51 [14] 53/-1/-1->52->51 [15] 53/-1/-1->52->51
+ip-26-0-160-225:94348:94650 [6] NCCL INFO Trees [0] 7/-1/-1->6->5 [1] 7/-1/-1->6->5 [2] 7/-1/-1->6->5 [3] 7/-1/-1->6->5 [4] 7/-1/-1->6->5 [5] 7/-1/-1->6->5 [6] 7/38/-1->6->-1 [7] -1/-1/-1->6->5 [8] 7/-1/-1->6->5 [9] 7/-1/-1->6->5 [10] 7/-1/-1->6->5 [11] 7/-1/-1->6->5 [12] 7/-1/-1->6->5 [13] 7/-1/-1->6->5 [14] 7/-1/-1->6->14 [15] -1/-1/-1->6->5
+ip-26-0-160-225:94344:94648 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-225:94348:94650 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-225:94342:94643 [0] NCCL INFO Channel 02/16 :    0   7   6   5   4   3  10   9   8  15  14  13  12  11  18  17  16  23  22  21
+ip-26-0-160-225:94345:94649 [3] NCCL INFO Trees [0] 4/-1/-1->3->2 [1] 4/-1/-1->3->2 [2] 4/-1/-1->3->2 [3] 4/35/-1->3->-1 [4] -1/-1/-1->3->2 [5] 4/-1/-1->3->2 [6] 4/-1/-1->3->2 [7] 4/-1/-1->3->2 [8] 4/-1/-1->3->2 [9] 4/-1/-1->3->2 [10] 4/-1/-1->3->2 [11] 4/-1/-1->3->11 [12] -1/-1/-1->3->2 [13] 4/-1/-1->3->2 [14] 4/-1/-1->3->2 [15] 4/-1/-1->3->2
+ip-26-0-173-7:811251:811556 [3] NCCL INFO Trees [0] 60/-1/-1->59->58 [1] 60/-1/-1->59->58 [2] 60/-1/-1->59->58 [3] 60/-1/-1->59->51 [4] -1/-1/-1->59->58 [5] 60/-1/-1->59->58 [6] 60/-1/-1->59->58 [7] 60/-1/-1->59->58 [8] 60/-1/-1->59->58 [9] 60/-1/-1->59->58 [10] 60/-1/-1->59->58 [11] 60/27/-1->59->-1 [12] -1/-1/-1->59->58 [13] 60/-1/-1->59->58 [14] 60/-1/-1->59->58 [15] 60/-1/-1->59->58
+ip-26-0-173-7:811251:811556 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-73:926598:926894 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-73:926596:926899 [2] NCCL INFO Trees [0] 51/-1/-1->50->49 [1] 51/-1/-1->50->49 [2] 51/42/58->50->34 [3] -1/-1/-1->50->49 [4] 51/-1/-1->50->49 [5] 51/-1/-1->50->49 [6] 51/-1/-1->50->49 [7] 51/-1/-1->50->49 [8] 51/-1/-1->50->49 [9] 51/-1/-1->50->49 [10] 51/-1/-1->50->42 [11] -1/-1/-1->50->49 [12] 51/-1/-1->50->49 [13] 51/-1/-1->50->49 [14] 51/-1/-1->50->49 [15] 51/-1/-1->50->49
+ip-26-0-160-225:94342:94643 [0] NCCL INFO Channel 03/16 :    0   7   6   5   4  11  10   9   8  15  14  13  12  19  18  17  16  23  22  21
+ip-26-0-160-225:94345:94649 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-73:926596:926899 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-73:926594:926898 [0] NCCL INFO Trees [0] 49/40/56->48->32 [1] -1/-1/-1->48->55 [2] 49/-1/-1->48->55 [3] 49/-1/-1->48->55 [4] 49/-1/-1->48->55 [5] 49/-1/-1->48->55 [6] 49/-1/-1->48->55 [7] 49/-1/-1->48->55 [8] 49/-1/-1->48->40 [9] -1/-1/-1->48->55 [10] 49/-1/-1->48->55 [11] 49/-1/-1->48->55 [12] 49/-1/-1->48->55 [13] 49/-1/-1->48->55 [14] 49/-1/-1->48->55 [15] 49/-1/-1->48->55
+ip-26-0-172-73:926597:926900 [3] NCCL INFO Trees [0] 52/-1/-1->51->50 [1] 52/-1/-1->51->50 [2] 52/-1/-1->51->50 [3] 52/43/59->51->35 [4] -1/-1/-1->51->50 [5] 52/-1/-1->51->50 [6] 52/-1/-1->51->50 [7] 52/-1/-1->51->50 [8] 52/-1/-1->51->50 [9] 52/-1/-1->51->50 [10] 52/-1/-1->51->50 [11] 52/-1/-1->51->43 [12] -1/-1/-1->51->50 [13] 52/-1/-1->51->50 [14] 52/-1/-1->51->50 [15] 52/-1/-1->51->50
+ip-26-0-160-225:94347:94646 [5] NCCL INFO Trees [0] 6/-1/-1->5->4 [1] 6/-1/-1->5->4 [2] 6/-1/-1->5->4 [3] 6/-1/-1->5->4 [4] 6/-1/-1->5->4 [5] 6/37/-1->5->-1 [6] -1/-1/-1->5->4 [7] 6/-1/-1->5->4 [8] 6/-1/-1->5->4 [9] 6/-1/-1->5->4 [10] 6/-1/-1->5->4 [11] 6/-1/-1->5->4 [12] 6/-1/-1->5->4 [13] 6/-1/-1->5->13 [14] -1/-1/-1->5->4 [15] 6/-1/-1->5->4
+ip-26-0-160-225:94346:94644 [4] NCCL INFO Trees [0] 5/-1/-1->4->3 [1] 5/-1/-1->4->3 [2] 5/-1/-1->4->3 [3] 5/-1/-1->4->3 [4] 5/36/-1->4->-1 [5] -1/-1/-1->4->3 [6] 5/-1/-1->4->3 [7] 5/-1/-1->4->3 [8] 5/-1/-1->4->3 [9] 5/-1/-1->4->3 [10] 5/-1/-1->4->3 [11] 5/-1/-1->4->3 [12] 5/-1/-1->4->12 [13] -1/-1/-1->4->3 [14] 5/-1/-1->4->3 [15] 5/-1/-1->4->3
+ip-26-0-160-225:94342:94643 [0] NCCL INFO Channel 04/16 :    0   7   6   5  12  11  10   9   8  15  14  13  20  19  18  17  16  23  22  21
+ip-26-0-160-225:94347:94646 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-225:94346:94644 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-73:926594:926898 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-73:926597:926900 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-225:94342:94643 [0] NCCL INFO Channel 05/16 :    0   4   7   6  13  11  10   9   8  12  15  14  21  19  18  17  16  20  23  22
+ip-26-0-160-225:94342:94643 [0] NCCL INFO Channel 06/16 :    0   5   4   7  14  11  10   9   8  13  12  15  22  19  18  17  16  21  20  23
+ip-26-0-160-225:94342:94643 [0] NCCL INFO Channel 07/16 :    0   6   5   4  15  11  10   9   8  14  13  12  23  19  18  17  16  22  21  20
+ip-26-0-160-225:94342:94643 [0] NCCL INFO Channel 08/16 :    0   1   2   3   4   5   6   7   8   9  10  11  12  13  14  15  16  17  18  19
+ip-26-0-160-225:94342:94643 [0] NCCL INFO Channel 09/16 :    0   7   6   5   4   3   2   9   8  15  14  13  12  11  10  17  16  23  22  21
+ip-26-0-160-225:94342:94643 [0] NCCL INFO Channel 10/16 :    0   7   6   5   4   3  10   9   8  15  14  13  12  11  18  17  16  23  22  21
+ip-26-0-160-225:94342:94643 [0] NCCL INFO Channel 11/16 :    0   7   6   5   4  11  10   9   8  15  14  13  12  19  18  17  16  23  22  21
+ip-26-0-160-225:94342:94643 [0] NCCL INFO Channel 12/16 :    0   7   6   5  12  11  10   9   8  15  14  13  20  19  18  17  16  23  22  21
+ip-26-0-160-225:94342:94643 [0] NCCL INFO Channel 13/16 :    0   4   7   6  13  11  10   9   8  12  15  14  21  19  18  17  16  20  23  22
+ip-26-0-160-225:94342:94643 [0] NCCL INFO Channel 14/16 :    0   5   4   7  14  11  10   9   8  13  12  15  22  19  18  17  16  21  20  23
+ip-26-0-160-225:94342:94643 [0] NCCL INFO Channel 15/16 :    0   6   5   4  15  11  10   9   8  14  13  12  23  19  18  17  16  22  21  20
+ip-26-0-160-225:94342:94643 [0] NCCL INFO Trees [0] 1/32/-1->0->-1 [1] -1/-1/-1->0->7 [2] 1/-1/-1->0->7 [3] 1/-1/-1->0->7 [4] 1/-1/-1->0->7 [5] 1/-1/-1->0->7 [6] 1/-1/-1->0->7 [7] 1/-1/-1->0->7 [8] 1/-1/-1->0->8 [9] -1/-1/-1->0->7 [10] 1/-1/-1->0->7 [11] 1/-1/-1->0->7 [12] 1/-1/-1->0->7 [13] 1/-1/-1->0->7 [14] 1/-1/-1->0->7 [15] 1/-1/-1->0->7
+ip-26-0-160-225:94342:94643 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-225:94343:94647 [1] NCCL INFO Channel 00/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-225:94349:94645 [7] NCCL INFO Channel 00/0 : 7[7] -> 8[0] [send] via NET/Libfabric/0(0)/GDRDMA
+ip-26-0-160-225:94348:94650 [6] NCCL INFO Channel 00/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-225:94346:94644 [4] NCCL INFO Channel 00/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-225:94343:94647 [1] NCCL INFO Channel 08/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-252:688206:688506 [6] NCCL INFO Channel 00/0 : 38[6] -> 39[7] via P2P/IPC
+ip-26-0-172-147:618586:618886 [2] NCCL INFO Channel 00/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-172-116:1019289:1019593 [1] NCCL INFO Channel 00/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-172-147:618588:618888 [4] NCCL INFO Channel 00/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-172-73:926600:926893 [6] NCCL INFO Channel 00/0 : 54[6] -> 55[7] via P2P/IPC
+ip-26-0-160-225:94344:94648 [2] NCCL INFO Channel 00/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-252:688204:688502 [4] NCCL INFO Channel 00/0 : 36[4] -> 37[5] via P2P/IPC
+ip-26-0-160-225:94345:94649 [3] NCCL INFO Channel 00/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-116:1019294:1019589 [6] NCCL INFO Channel 00/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-172-57:942615:942912 [4] NCCL INFO Channel 00/0 : 44[4] -> 45[5] via P2P/IPC
+ip-26-0-172-116:1019292:1019590 [4] NCCL INFO Channel 00/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-172-252:688203:688501 [3] NCCL INFO Channel 00/0 : 35[3] -> 36[4] via P2P/IPC
+ip-26-0-172-252:688205:688505 [5] NCCL INFO Channel 00/0 : 37[5] -> 38[6] via P2P/IPC
+ip-26-0-173-7:811254:811557 [6] NCCL INFO Channel 00/0 : 62[6] -> 63[7] via P2P/IPC
+ip-26-0-173-7:811253:811558 [5] NCCL INFO Channel 00/0 : 61[5] -> 62[6] via P2P/IPC
+ip-26-0-172-116:1019293:1019592 [5] NCCL INFO Channel 00/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-172-57:942616:942913 [5] NCCL INFO Channel 00/0 : 45[5] -> 46[6] via P2P/IPC
+ip-26-0-172-116:1019291:1019594 [3] NCCL INFO Channel 00/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-172-147:618585:618884 [1] NCCL INFO Channel 00/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-173-7:811252:811560 [4] NCCL INFO Channel 00/0 : 60[4] -> 61[5] via P2P/IPC
+ip-26-0-172-116:1019290:1019595 [2] NCCL INFO Channel 00/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-172-252:688202:688503 [2] NCCL INFO Channel 00/0 : 34[2] -> 35[3] via P2P/IPC
+ip-26-0-172-147:618587:618887 [3] NCCL INFO Channel 00/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-172-252:688206:688506 [6] NCCL INFO Channel 08/0 : 38[6] -> 39[7] via P2P/IPC
+ip-26-0-172-147:618590:618889 [6] NCCL INFO Channel 00/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-172-73:926596:926899 [2] NCCL INFO Channel 00/0 : 50[2] -> 51[3] via P2P/IPC
+ip-26-0-160-225:94346:94644 [4] NCCL INFO Channel 08/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-57:942617:942916 [6] NCCL INFO Channel 00/0 : 46[6] -> 47[7] via P2P/IPC
+ip-26-0-172-147:618586:618886 [2] NCCL INFO Channel 08/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-172-57:942613:942917 [2] NCCL INFO Channel 00/0 : 42[2] -> 43[3] via P2P/IPC
+ip-26-0-172-57:942612:942914 [1] NCCL INFO Channel 00/0 : 41[1] -> 42[2] via P2P/IPC
+ip-26-0-172-73:926598:926894 [4] NCCL INFO Channel 00/0 : 52[4] -> 53[5] via P2P/IPC
+ip-26-0-173-7:811251:811556 [3] NCCL INFO Channel 00/0 : 59[3] -> 60[4] via P2P/IPC
+ip-26-0-173-7:811249:811559 [1] NCCL INFO Channel 00/0 : 57[1] -> 58[2] via P2P/IPC
+ip-26-0-160-225:94348:94650 [6] NCCL INFO Channel 08/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-57:942614:942915 [3] NCCL INFO Channel 00/0 : 43[3] -> 44[4] via P2P/IPC
+ip-26-0-172-147:618588:618888 [4] NCCL INFO Channel 08/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-172-252:688204:688502 [4] NCCL INFO Channel 08/0 : 36[4] -> 37[5] via P2P/IPC
+ip-26-0-173-7:811250:811555 [2] NCCL INFO Channel 00/0 : 58[2] -> 59[3] via P2P/IPC
+ip-26-0-172-116:1019289:1019593 [1] NCCL INFO Channel 08/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-172-116:1019294:1019589 [6] NCCL INFO Channel 08/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-172-73:926597:926900 [3] NCCL INFO Channel 00/0 : 51[3] -> 52[4] via P2P/IPC
+ip-26-0-172-252:688205:688505 [5] NCCL INFO Channel 08/0 : 37[5] -> 38[6] via P2P/IPC
+ip-26-0-160-225:94344:94648 [2] NCCL INFO Channel 08/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-225:94345:94649 [3] NCCL INFO Channel 08/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-73:926595:926897 [1] NCCL INFO Channel 00/0 : 49[1] -> 50[2] via P2P/IPC
+ip-26-0-172-73:926599:926895 [5] NCCL INFO Channel 00/0 : 53[5] -> 54[6] via P2P/IPC
+ip-26-0-172-57:942615:942912 [4] NCCL INFO Channel 08/0 : 44[4] -> 45[5] via P2P/IPC
+ip-26-0-172-252:688203:688501 [3] NCCL INFO Channel 08/0 : 35[3] -> 36[4] via P2P/IPC
+ip-26-0-173-7:811253:811558 [5] NCCL INFO Channel 08/0 : 61[5] -> 62[6] via P2P/IPC
+ip-26-0-172-73:926600:926893 [6] NCCL INFO Channel 08/0 : 54[6] -> 55[7] via P2P/IPC
+ip-26-0-173-7:811254:811557 [6] NCCL INFO Channel 08/0 : 62[6] -> 63[7] via P2P/IPC
+ip-26-0-172-116:1019292:1019590 [4] NCCL INFO Channel 08/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-172-147:618585:618884 [1] NCCL INFO Channel 08/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-172-57:942616:942913 [5] NCCL INFO Channel 08/0 : 45[5] -> 46[6] via P2P/IPC
+ip-26-0-173-7:811252:811560 [4] NCCL INFO Channel 08/0 : 60[4] -> 61[5] via P2P/IPC
+ip-26-0-172-252:688202:688503 [2] NCCL INFO Channel 08/0 : 34[2] -> 35[3] via P2P/IPC
+ip-26-0-172-147:618587:618887 [3] NCCL INFO Channel 08/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-172-116:1019290:1019595 [2] NCCL INFO Channel 08/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-172-116:1019293:1019592 [5] NCCL INFO Channel 08/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-172-73:926596:926899 [2] NCCL INFO Channel 08/0 : 50[2] -> 51[3] via P2P/IPC
+ip-26-0-172-116:1019291:1019594 [3] NCCL INFO Channel 08/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-172-147:618590:618889 [6] NCCL INFO Channel 08/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-172-73:926598:926894 [4] NCCL INFO Channel 08/0 : 52[4] -> 53[5] via P2P/IPC
+ip-26-0-172-57:942617:942916 [6] NCCL INFO Channel 08/0 : 46[6] -> 47[7] via P2P/IPC
+ip-26-0-173-7:811249:811559 [1] NCCL INFO Channel 08/0 : 57[1] -> 58[2] via P2P/IPC
+ip-26-0-160-225:94347:94646 [5] NCCL INFO Channel 00/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-57:942612:942914 [1] NCCL INFO Channel 08/0 : 41[1] -> 42[2] via P2P/IPC
+ip-26-0-173-7:811250:811555 [2] NCCL INFO Channel 08/0 : 58[2] -> 59[3] via P2P/IPC
+ip-26-0-172-142:880684:880981 [5] NCCL INFO Channel 00/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-172-57:942613:942917 [2] NCCL INFO Channel 08/0 : 42[2] -> 43[3] via P2P/IPC
+ip-26-0-173-7:811251:811556 [3] NCCL INFO Channel 08/0 : 59[3] -> 60[4] via P2P/IPC
+ip-26-0-172-57:942614:942915 [3] NCCL INFO Channel 08/0 : 43[3] -> 44[4] via P2P/IPC
+ip-26-0-172-142:880685:880977 [6] NCCL INFO Channel 00/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-172-142:880683:880982 [4] NCCL INFO Channel 00/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-172-252:688204:688502 [4] NCCL INFO Channel 05/0 : 36[4] -> 39[7] via P2P/IPC
+ip-26-0-172-73:926597:926900 [3] NCCL INFO Channel 08/0 : 51[3] -> 52[4] via P2P/IPC
+ip-26-0-172-73:926599:926895 [5] NCCL INFO Channel 08/0 : 53[5] -> 54[6] via P2P/IPC
+ip-26-0-172-142:880681:880983 [2] NCCL INFO Channel 00/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-172-73:926595:926897 [1] NCCL INFO Channel 08/0 : 49[1] -> 50[2] via P2P/IPC
+ip-26-0-160-225:94347:94646 [5] NCCL INFO Channel 08/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-252:688201:688500 [1] NCCL INFO Channel 00/0 : 33[1] -> 34[2] via P2P/IPC
+ip-26-0-172-252:688204:688502 [4] NCCL INFO Channel 06/0 : 36[4] -> 39[7] via P2P/IPC
+ip-26-0-172-116:1019292:1019590 [4] NCCL INFO Channel 05/0 : 12[4] -> 15[7] via P2P/IPC
+ip-26-0-160-225:94346:94644 [4] NCCL INFO Channel 05/0 : 4[4] -> 7[7] via P2P/IPC
+ip-26-0-172-142:880680:880984 [1] NCCL INFO Channel 00/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-172-147:618589:618883 [5] NCCL INFO Channel 00/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-173-7:811252:811560 [4] NCCL INFO Channel 05/0 : 60[4] -> 63[7] via P2P/IPC
+ip-26-0-172-142:880684:880981 [5] NCCL INFO Channel 08/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-172-57:942615:942912 [4] NCCL INFO Channel 05/0 : 44[4] -> 47[7] via P2P/IPC
+ip-26-0-172-116:1019292:1019590 [4] NCCL INFO Channel 06/0 : 12[4] -> 15[7] via P2P/IPC
+ip-26-0-172-252:688204:688502 [4] NCCL INFO Channel 13/0 : 36[4] -> 39[7] via P2P/IPC
+ip-26-0-172-73:926598:926894 [4] NCCL INFO Channel 05/0 : 52[4] -> 55[7] via P2P/IPC
+ip-26-0-172-252:688201:688500 [1] NCCL INFO Channel 08/0 : 33[1] -> 34[2] via P2P/IPC
+ip-26-0-172-147:618589:618883 [5] NCCL INFO Channel 08/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-172-57:942615:942912 [4] NCCL INFO Channel 06/0 : 44[4] -> 47[7] via P2P/IPC
+ip-26-0-173-7:811252:811560 [4] NCCL INFO Channel 06/0 : 60[4] -> 63[7] via P2P/IPC
+ip-26-0-172-116:1019292:1019590 [4] NCCL INFO Channel 13/0 : 12[4] -> 15[7] via P2P/IPC
+ip-26-0-172-252:688204:688502 [4] NCCL INFO Channel 14/0 : 36[4] -> 39[7] via P2P/IPC
+ip-26-0-172-73:926598:926894 [4] NCCL INFO Channel 06/0 : 52[4] -> 55[7] via P2P/IPC
+ip-26-0-172-57:942615:942912 [4] NCCL INFO Channel 13/0 : 44[4] -> 47[7] via P2P/IPC
+ip-26-0-173-7:811252:811560 [4] NCCL INFO Channel 13/0 : 60[4] -> 63[7] via P2P/IPC
+ip-26-0-172-116:1019292:1019590 [4] NCCL INFO Channel 14/0 : 12[4] -> 15[7] via P2P/IPC
+ip-26-0-172-147:618588:618888 [4] NCCL INFO Channel 05/0 : 28[4] -> 31[7] via P2P/IPC
+ip-26-0-160-225:94346:94644 [4] NCCL INFO Channel 06/0 : 4[4] -> 7[7] via P2P/IPC
+ip-26-0-172-57:942615:942912 [4] NCCL INFO Channel 14/0 : 44[4] -> 47[7] via P2P/IPC
+ip-26-0-173-7:811252:811560 [4] NCCL INFO Channel 14/0 : 60[4] -> 63[7] via P2P/IPC
+ip-26-0-172-142:880681:880983 [2] NCCL INFO Channel 08/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-172-73:926598:926894 [4] NCCL INFO Channel 13/0 : 52[4] -> 55[7] via P2P/IPC
+ip-26-0-172-142:880682:880978 [3] NCCL INFO Channel 00/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-172-142:880683:880982 [4] NCCL INFO Channel 08/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-160-225:94346:94644 [4] NCCL INFO Channel 13/0 : 4[4] -> 7[7] via P2P/IPC
+ip-26-0-172-147:618588:618888 [4] NCCL INFO Channel 06/0 : 28[4] -> 31[7] via P2P/IPC
+ip-26-0-172-142:880685:880977 [6] NCCL INFO Channel 08/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-172-73:926598:926894 [4] NCCL INFO Channel 14/0 : 52[4] -> 55[7] via P2P/IPC
+ip-26-0-160-225:94346:94644 [4] NCCL INFO Channel 14/0 : 4[4] -> 7[7] via P2P/IPC
+ip-26-0-172-147:618588:618888 [4] NCCL INFO Channel 13/0 : 28[4] -> 31[7] via P2P/IPC
+ip-26-0-172-147:618588:618888 [4] NCCL INFO Channel 14/0 : 28[4] -> 31[7] via P2P/IPC
+ip-26-0-172-142:880680:880984 [1] NCCL INFO Channel 08/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-172-142:880682:880978 [3] NCCL INFO Channel 08/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-172-142:880683:880982 [4] NCCL INFO Channel 05/0 : 20[4] -> 23[7] via P2P/IPC
+ip-26-0-172-142:880683:880982 [4] NCCL INFO Channel 06/0 : 20[4] -> 23[7] via P2P/IPC
+ip-26-0-172-142:880683:880982 [4] NCCL INFO Channel 13/0 : 20[4] -> 23[7] via P2P/IPC
+ip-26-0-172-142:880683:880982 [4] NCCL INFO Channel 14/0 : 20[4] -> 23[7] via P2P/IPC
+ip-26-0-172-142:880679:880980 [0] NCCL INFO Channel 00/0 : 15[7] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880679:880980 [0] NCCL INFO Channel 08/0 : 15[7] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880679:880980 [0] NCCL INFO Channel 00/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-172-142:880686:880979 [7] NCCL INFO Channel 00/0 : 23[7] -> 24[0] [send] via NET/Libfabric/0(16)/GDRDMA
+ip-26-0-160-225:94342:94643 [0] NCCL INFO Channel 00/0 : 63[7] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:94349:94645 [7] NCCL INFO Channel 08/0 : 7[7] -> 8[0] [send] via NET/Libfabric/0(0)/GDRDMA
+ip-26-0-160-225:94342:94643 [0] NCCL INFO Channel 08/0 : 63[7] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:94342:94643 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-142:880679:880980 [0] NCCL INFO Channel 08/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-172-142:880686:880979 [7] NCCL INFO Channel 08/0 : 23[7] -> 24[0] [send] via NET/Libfabric/0(16)/GDRDMA
+ip-26-0-172-147:618584:618882 [0] NCCL INFO Channel 00/0 : 23[7] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688200:688507 [0] NCCL INFO Channel 00/0 : 31[7] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618584:618882 [0] NCCL INFO Channel 08/0 : 23[7] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688200:688507 [0] NCCL INFO Channel 08/0 : 31[7] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618584:618882 [0] NCCL INFO Channel 00/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-172-147:618591:618885 [7] NCCL INFO Channel 00/0 : 31[7] -> 32[0] [send] via NET/Libfabric/0(24)/GDRDMA
+ip-26-0-172-252:688200:688507 [0] NCCL INFO Channel 00/0 : 32[0] -> 33[1] via P2P/IPC
+ip-26-0-172-252:688207:688504 [7] NCCL INFO Channel 00/0 : 39[7] -> 40[0] [send] via NET/Libfabric/0(32)/GDRDMA
+ip-26-0-160-225:94342:94643 [0] NCCL INFO Channel 08/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-252:688200:688507 [0] NCCL INFO Channel 08/0 : 32[0] -> 33[1] via P2P/IPC
+ip-26-0-172-147:618584:618882 [0] NCCL INFO Channel 08/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-172-252:688207:688504 [7] NCCL INFO Channel 08/0 : 39[7] -> 40[0] [send] via NET/Libfabric/0(32)/GDRDMA
+ip-26-0-172-147:618591:618885 [7] NCCL INFO Channel 08/0 : 31[7] -> 32[0] [send] via NET/Libfabric/0(24)/GDRDMA
+ip-26-0-172-252:688200:688507 [0] NCCL INFO Channel 05/0 : 32[0] -> 36[4] via P2P/IPC
+ip-26-0-173-7:811248:811554 [0] NCCL INFO Channel 00/0 : 55[7] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618584:618882 [0] NCCL INFO Channel 05/0 : 24[0] -> 28[4] via P2P/IPC
+ip-26-0-173-7:811248:811554 [0] NCCL INFO Channel 08/0 : 55[7] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-7:811248:811554 [0] NCCL INFO Channel 00/0 : 56[0] -> 57[1] via P2P/IPC
+ip-26-0-173-7:811255:811561 [7] NCCL INFO Channel 00/0 : 63[7] -> 0[0] [send] via NET/Libfabric/0(56)/GDRDMA
+ip-26-0-172-252:688200:688507 [0] NCCL INFO Channel 13/0 : 32[0] -> 36[4] via P2P/IPC
+ip-26-0-160-225:94345:94649 [3] NCCL INFO Channel 03/0 : 60[4] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618584:618882 [0] NCCL INFO Channel 13/0 : 24[0] -> 28[4] via P2P/IPC
+ip-26-0-160-225:94345:94649 [3] NCCL INFO Channel 11/0 : 60[4] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-7:811248:811554 [0] NCCL INFO Channel 08/0 : 56[0] -> 57[1] via P2P/IPC
+ip-26-0-160-225:94344:94648 [2] NCCL INFO Channel 02/0 : 59[3] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:94345:94649 [3] NCCL INFO Channel 02/0 : 3[3] -> 10[2] [send] via NET/Libfabric/2(2)/GDRDMA
+ip-26-0-160-225:94344:94648 [2] NCCL INFO Channel 10/0 : 59[3] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-7:811255:811561 [7] NCCL INFO Channel 08/0 : 63[7] -> 0[0] [send] via NET/Libfabric/0(56)/GDRDMA
+ip-26-0-172-116:1019288:1019588 [0] NCCL INFO Channel 00/0 : 7[7] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:94345:94649 [3] NCCL INFO Channel 10/0 : 3[3] -> 10[2] [send] via NET/Libfabric/2(2)/GDRDMA
+ip-26-0-172-116:1019288:1019588 [0] NCCL INFO Channel 08/0 : 7[7] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019288:1019588 [0] NCCL INFO Channel 00/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-172-116:1019295:1019591 [7] NCCL INFO Channel 00/0 : 15[7] -> 16[0] [send] via NET/Libfabric/0(8)/GDRDMA
+ip-26-0-173-7:811251:811556 [3] NCCL INFO Channel 03/0 : 52[4] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:94342:94643 [0] NCCL INFO Channel 05/0 : 0[0] -> 4[4] via P2P/IPC
+ip-26-0-173-7:811251:811556 [3] NCCL INFO Channel 11/0 : 52[4] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942611:942911 [0] NCCL INFO Channel 00/0 : 39[7] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:94342:94643 [0] NCCL INFO Channel 13/0 : 0[0] -> 4[4] via P2P/IPC
+ip-26-0-172-116:1019288:1019588 [0] NCCL INFO Channel 08/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-172-57:942611:942911 [0] NCCL INFO Channel 08/0 : 39[7] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019295:1019591 [7] NCCL INFO Channel 08/0 : 15[7] -> 16[0] [send] via NET/Libfabric/0(8)/GDRDMA
+ip-26-0-172-57:942618:942918 [7] NCCL INFO Channel 00/0 : 47[7] -> 48[0] [send] via NET/Libfabric/0(40)/GDRDMA
+ip-26-0-172-57:942611:942911 [0] NCCL INFO Channel 00/0 : 40[0] -> 41[1] via P2P/IPC
+ip-26-0-172-73:926594:926898 [0] NCCL INFO Channel 00/0 : 47[7] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926594:926898 [0] NCCL INFO Channel 08/0 : 47[7] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926594:926898 [0] NCCL INFO Channel 00/0 : 48[0] -> 49[1] via P2P/IPC
+ip-26-0-172-73:926601:926896 [7] NCCL INFO Channel 00/0 : 55[7] -> 56[0] [send] via NET/Libfabric/0(48)/GDRDMA
+ip-26-0-172-142:880679:880980 [0] NCCL INFO Channel 05/0 : 16[0] -> 20[4] via P2P/IPC
+ip-26-0-172-147:618584:618882 [0] NCCL INFO Channel 06/0 : 24[0] -> 29[5] via P2P/IPC
+ip-26-0-172-116:1019288:1019588 [0] NCCL INFO Channel 05/0 : 8[0] -> 12[4] via P2P/IPC
+ip-26-0-173-7:811250:811555 [2] NCCL INFO Channel 02/0 : 51[3] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-7:811250:811555 [2] NCCL INFO Channel 10/0 : 51[3] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942611:942911 [0] NCCL INFO Channel 08/0 : 40[0] -> 41[1] via P2P/IPC
+ip-26-0-173-7:811251:811556 [3] NCCL INFO Channel 02/0 : 59[3] -> 2[2] [send] via NET/Libfabric/2(58)/GDRDMA
+ip-26-0-172-57:942618:942918 [7] NCCL INFO Channel 08/0 : 47[7] -> 48[0] [send] via NET/Libfabric/0(40)/GDRDMA
+ip-26-0-172-73:926594:926898 [0] NCCL INFO Channel 08/0 : 48[0] -> 49[1] via P2P/IPC
+ip-26-0-173-7:811251:811556 [3] NCCL INFO Channel 10/0 : 59[3] -> 2[2] [send] via NET/Libfabric/2(58)/GDRDMA
+ip-26-0-172-142:880679:880980 [0] NCCL INFO Channel 13/0 : 16[0] -> 20[4] via P2P/IPC
+ip-26-0-172-73:926601:926896 [7] NCCL INFO Channel 08/0 : 55[7] -> 56[0] [send] via NET/Libfabric/0(48)/GDRDMA
+ip-26-0-172-147:618584:618882 [0] NCCL INFO Channel 14/0 : 24[0] -> 29[5] via P2P/IPC
+ip-26-0-172-252:688203:688501 [3] NCCL INFO Channel 03/0 : 28[4] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019288:1019588 [0] NCCL INFO Channel 13/0 : 8[0] -> 12[4] via P2P/IPC
+ip-26-0-172-252:688203:688501 [3] NCCL INFO Channel 11/0 : 28[4] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-7:811248:811554 [0] NCCL INFO Channel 05/0 : 56[0] -> 60[4] via P2P/IPC
+ip-26-0-172-252:688202:688503 [2] NCCL INFO Channel 02/0 : 27[3] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880679:880980 [0] NCCL INFO Channel 06/0 : 16[0] -> 21[5] via P2P/IPC
+ip-26-0-172-147:618584:618882 [0] NCCL INFO Channel 07/0 : 24[0] -> 30[6] via P2P/IPC
+ip-26-0-172-252:688202:688503 [2] NCCL INFO Channel 10/0 : 27[3] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942611:942911 [0] NCCL INFO Channel 05/0 : 40[0] -> 44[4] via P2P/IPC
+ip-26-0-172-252:688203:688501 [3] NCCL INFO Channel 02/0 : 35[3] -> 42[2] [send] via NET/Libfabric/2(34)/GDRDMA
+ip-26-0-172-73:926594:926898 [0] NCCL INFO Channel 05/0 : 48[0] -> 52[4] via P2P/IPC
+ip-26-0-172-147:618587:618887 [3] NCCL INFO Channel 03/0 : 20[4] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688203:688501 [3] NCCL INFO Channel 10/0 : 35[3] -> 42[2] [send] via NET/Libfabric/2(34)/GDRDMA
+ip-26-0-172-116:1019291:1019594 [3] NCCL INFO Channel 03/0 : 4[4] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618586:618886 [2] NCCL INFO Channel 02/0 : 19[3] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880679:880980 [0] NCCL INFO Channel 14/0 : 16[0] -> 21[5] via P2P/IPC
+ip-26-0-172-147:618587:618887 [3] NCCL INFO Channel 11/0 : 20[4] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-7:811248:811554 [0] NCCL INFO Channel 13/0 : 56[0] -> 60[4] via P2P/IPC
+ip-26-0-172-116:1019291:1019594 [3] NCCL INFO Channel 11/0 : 4[4] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618586:618886 [2] NCCL INFO Channel 10/0 : 19[3] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618584:618882 [0] NCCL INFO Channel 15/0 : 24[0] -> 30[6] via P2P/IPC
+ip-26-0-172-57:942614:942915 [3] NCCL INFO Channel 03/0 : 36[4] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926594:926898 [0] NCCL INFO Channel 13/0 : 48[0] -> 52[4] via P2P/IPC
+ip-26-0-172-57:942614:942915 [3] NCCL INFO Channel 11/0 : 36[4] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019290:1019595 [2] NCCL INFO Channel 02/0 : 3[3] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618587:618887 [3] NCCL INFO Channel 02/0 : 27[3] -> 34[2] [send] via NET/Libfabric/2(26)/GDRDMA
+ip-26-0-172-147:618587:618887 [3] NCCL INFO Channel 10/0 : 27[3] -> 34[2] [send] via NET/Libfabric/2(26)/GDRDMA
+ip-26-0-172-57:942611:942911 [0] NCCL INFO Channel 13/0 : 40[0] -> 44[4] via P2P/IPC
+ip-26-0-172-142:880679:880980 [0] NCCL INFO Channel 07/0 : 16[0] -> 22[6] via P2P/IPC
+ip-26-0-172-116:1019290:1019595 [2] NCCL INFO Channel 10/0 : 3[3] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019291:1019594 [3] NCCL INFO Channel 02/0 : 11[3] -> 18[2] [send] via NET/Libfabric/2(10)/GDRDMA
+ip-26-0-173-7:811248:811554 [0] NCCL INFO Channel 06/0 : 56[0] -> 61[5] via P2P/IPC
+ip-26-0-172-116:1019291:1019594 [3] NCCL INFO Channel 10/0 : 11[3] -> 18[2] [send] via NET/Libfabric/2(10)/GDRDMA
+ip-26-0-172-142:880679:880980 [0] NCCL INFO Channel 15/0 : 16[0] -> 22[6] via P2P/IPC
+ip-26-0-172-147:618584:618882 [0] NCCL INFO Channel 01/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-173-7:811248:811554 [0] NCCL INFO Channel 14/0 : 56[0] -> 61[5] via P2P/IPC
+ip-26-0-172-147:618584:618882 [0] NCCL INFO Channel 02/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-160-225:94342:94643 [0] NCCL INFO Channel 06/0 : 0[0] -> 5[5] via P2P/IPC
+ip-26-0-172-142:880679:880980 [0] NCCL INFO Channel 01/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-172-57:942613:942917 [2] NCCL INFO Channel 02/0 : 35[3] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942613:942917 [2] NCCL INFO Channel 10/0 : 35[3] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618584:618882 [0] NCCL INFO Channel 03/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-172-142:880679:880980 [0] NCCL INFO Channel 02/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-172-57:942614:942915 [3] NCCL INFO Channel 02/0 : 43[3] -> 50[2] [send] via NET/Libfabric/2(42)/GDRDMA
+ip-26-0-173-7:811248:811554 [0] NCCL INFO Channel 07/0 : 56[0] -> 62[6] via P2P/IPC
+ip-26-0-160-225:94342:94643 [0] NCCL INFO Channel 14/0 : 0[0] -> 5[5] via P2P/IPC
+ip-26-0-172-57:942614:942915 [3] NCCL INFO Channel 10/0 : 43[3] -> 50[2] [send] via NET/Libfabric/2(42)/GDRDMA
+ip-26-0-172-142:880679:880980 [0] NCCL INFO Channel 03/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-172-147:618584:618882 [0] NCCL INFO Channel 04/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-172-73:926597:926900 [3] NCCL INFO Channel 03/0 : 44[4] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-7:811248:811554 [0] NCCL INFO Channel 15/0 : 56[0] -> 62[6] via P2P/IPC
+ip-26-0-172-73:926597:926900 [3] NCCL INFO Channel 11/0 : 44[4] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880679:880980 [0] NCCL INFO Channel 04/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-172-147:618584:618882 [0] NCCL INFO Channel 09/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-160-225:94342:94643 [0] NCCL INFO Channel 07/0 : 0[0] -> 6[6] via P2P/IPC
+ip-26-0-172-57:942613:942917 [2] NCCL INFO Channel 01/0 : 42[2] -> 49[1] [send] via NET/Libfabric/1(41)/GDRDMA
+ip-26-0-172-57:942613:942917 [2] NCCL INFO Channel 09/0 : 42[2] -> 49[1] [send] via NET/Libfabric/1(41)/GDRDMA
+ip-26-0-172-142:880679:880980 [0] NCCL INFO Channel 09/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-173-7:811248:811554 [0] NCCL INFO Channel 01/0 : 56[0] -> 63[7] via P2P/IPC
+ip-26-0-172-147:618584:618882 [0] NCCL INFO Channel 10/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-160-225:94342:94643 [0] NCCL INFO Channel 15/0 : 0[0] -> 6[6] via P2P/IPC
+ip-26-0-172-142:880679:880980 [0] NCCL INFO Channel 10/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-173-7:811248:811554 [0] NCCL INFO Channel 02/0 : 56[0] -> 63[7] via P2P/IPC
+ip-26-0-172-147:618584:618882 [0] NCCL INFO Channel 11/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-172-142:880686:880979 [7] NCCL INFO Channel 06/0 : 23[7] -> 30[6] [send] via NET/Libfabric/6(22)/GDRDMA
+ip-26-0-172-142:880679:880980 [0] NCCL INFO Channel 11/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-172-147:618591:618885 [7] NCCL INFO Channel 06/0 : 31[7] -> 38[6] [send] via NET/Libfabric/6(30)/GDRDMA
+ip-26-0-173-7:811248:811554 [0] NCCL INFO Channel 03/0 : 56[0] -> 63[7] via P2P/IPC
+ip-26-0-160-225:94342:94643 [0] NCCL INFO Channel 01/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-172-147:618584:618882 [0] NCCL INFO Channel 12/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-173-7:811248:811554 [0] NCCL INFO Channel 04/0 : 56[0] -> 63[7] via P2P/IPC
+ip-26-0-172-142:880679:880980 [0] NCCL INFO Channel 12/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-160-225:94342:94643 [0] NCCL INFO Channel 02/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-172-142:880682:880978 [3] NCCL INFO Channel 03/0 : 12[4] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880682:880978 [3] NCCL INFO Channel 11/0 : 12[4] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-7:811248:811554 [0] NCCL INFO Channel 09/0 : 56[0] -> 63[7] via P2P/IPC
+ip-26-0-160-225:94342:94643 [0] NCCL INFO Channel 03/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-172-73:926596:926899 [2] NCCL INFO Channel 02/0 : 43[3] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880681:880983 [2] NCCL INFO Channel 02/0 : 11[3] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926596:926899 [2] NCCL INFO Channel 10/0 : 43[3] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-7:811248:811554 [0] NCCL INFO Channel 10/0 : 56[0] -> 63[7] via P2P/IPC
+ip-26-0-172-142:880681:880983 [2] NCCL INFO Channel 10/0 : 11[3] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926597:926900 [3] NCCL INFO Channel 02/0 : 51[3] -> 58[2] [send] via NET/Libfabric/2(50)/GDRDMA
+ip-26-0-172-252:688200:688507 [0] NCCL INFO Channel 06/0 : 32[0] -> 37[5] via P2P/IPC
+ip-26-0-172-142:880682:880978 [3] NCCL INFO Channel 02/0 : 19[3] -> 26[2] [send] via NET/Libfabric/2(18)/GDRDMA
+ip-26-0-160-225:94342:94643 [0] NCCL INFO Channel 04/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-172-73:926597:926900 [3] NCCL INFO Channel 10/0 : 51[3] -> 58[2] [send] via NET/Libfabric/2(50)/GDRDMA
+ip-26-0-173-7:811248:811554 [0] NCCL INFO Channel 11/0 : 56[0] -> 63[7] via P2P/IPC
+ip-26-0-172-142:880682:880978 [3] NCCL INFO Channel 10/0 : 19[3] -> 26[2] [send] via NET/Libfabric/2(18)/GDRDMA
+ip-26-0-172-73:926596:926899 [2] NCCL INFO Channel 01/0 : 50[2] -> 57[1] [send] via NET/Libfabric/1(49)/GDRDMA
+ip-26-0-160-225:94342:94643 [0] NCCL INFO Channel 09/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-172-73:926596:926899 [2] NCCL INFO Channel 09/0 : 50[2] -> 57[1] [send] via NET/Libfabric/1(49)/GDRDMA
+ip-26-0-172-252:688200:688507 [0] NCCL INFO Channel 14/0 : 32[0] -> 37[5] via P2P/IPC
+ip-26-0-173-7:811248:811554 [0] NCCL INFO Channel 12/0 : 56[0] -> 63[7] via P2P/IPC
+ip-26-0-173-7:811255:811561 [7] NCCL INFO Channel 06/0 : 63[7] -> 6[6] [send] via NET/Libfabric/6(62)/GDRDMA
+ip-26-0-160-225:94342:94643 [0] NCCL INFO Channel 10/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-172-252:688200:688507 [0] NCCL INFO Channel 07/0 : 32[0] -> 38[6] via P2P/IPC
+ip-26-0-160-225:94342:94643 [0] NCCL INFO Channel 11/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-172-252:688200:688507 [0] NCCL INFO Channel 15/0 : 32[0] -> 38[6] via P2P/IPC
+ip-26-0-172-116:1019295:1019591 [7] NCCL INFO Channel 06/0 : 15[7] -> 22[6] [send] via NET/Libfabric/6(14)/GDRDMA
+ip-26-0-160-225:94342:94643 [0] NCCL INFO Channel 12/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-172-116:1019295:1019591 [7] NCCL INFO Channel 14/0 : 15[7] -> 22[6] [send] via NET/Libfabric/6(14)/GDRDMA
+ip-26-0-160-225:94349:94645 [7] NCCL INFO Channel 06/0 : 7[7] -> 14[6] [send] via NET/Libfabric/6(6)/GDRDMA
+ip-26-0-172-116:1019288:1019588 [0] NCCL INFO Channel 06/0 : 8[0] -> 13[5] via P2P/IPC
+ip-26-0-160-225:94349:94645 [7] NCCL INFO Channel 14/0 : 7[7] -> 14[6] [send] via NET/Libfabric/6(6)/GDRDMA
+ip-26-0-172-252:688200:688507 [0] NCCL INFO Channel 01/0 : 32[0] -> 39[7] via P2P/IPC
+ip-26-0-172-252:688200:688507 [0] NCCL INFO Channel 02/0 : 32[0] -> 39[7] via P2P/IPC
+ip-26-0-172-116:1019288:1019588 [0] NCCL INFO Channel 14/0 : 8[0] -> 13[5] via P2P/IPC
+ip-26-0-172-252:688200:688507 [0] NCCL INFO Channel 03/0 : 32[0] -> 39[7] via P2P/IPC
+ip-26-0-172-252:688207:688504 [7] NCCL INFO Channel 06/0 : 39[7] -> 46[6] [send] via NET/Libfabric/6(38)/GDRDMA
+ip-26-0-172-116:1019288:1019588 [0] NCCL INFO Channel 07/0 : 8[0] -> 14[6] via P2P/IPC
+ip-26-0-172-73:926601:926896 [7] NCCL INFO Channel 06/0 : 55[7] -> 62[6] [send] via NET/Libfabric/6(54)/GDRDMA
+ip-26-0-172-73:926601:926896 [7] NCCL INFO Channel 14/0 : 55[7] -> 62[6] [send] via NET/Libfabric/6(54)/GDRDMA
+ip-26-0-172-252:688200:688507 [0] NCCL INFO Channel 04/0 : 32[0] -> 39[7] via P2P/IPC
+ip-26-0-172-116:1019288:1019588 [0] NCCL INFO Channel 15/0 : 8[0] -> 14[6] via P2P/IPC
+ip-26-0-172-252:688200:688507 [0] NCCL INFO Channel 09/0 : 32[0] -> 39[7] via P2P/IPC
+ip-26-0-172-116:1019288:1019588 [0] NCCL INFO Channel 01/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-172-252:688200:688507 [0] NCCL INFO Channel 10/0 : 32[0] -> 39[7] via P2P/IPC
+ip-26-0-172-57:942618:942918 [7] NCCL INFO Channel 06/0 : 47[7] -> 54[6] [send] via NET/Libfabric/6(46)/GDRDMA
+ip-26-0-172-57:942618:942918 [7] NCCL INFO Channel 14/0 : 47[7] -> 54[6] [send] via NET/Libfabric/6(46)/GDRDMA
+ip-26-0-172-116:1019288:1019588 [0] NCCL INFO Channel 02/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-172-252:688200:688507 [0] NCCL INFO Channel 11/0 : 32[0] -> 39[7] via P2P/IPC
+ip-26-0-172-116:1019288:1019588 [0] NCCL INFO Channel 03/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-172-252:688200:688507 [0] NCCL INFO Channel 12/0 : 32[0] -> 39[7] via P2P/IPC
+ip-26-0-172-116:1019288:1019588 [0] NCCL INFO Channel 04/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-172-116:1019288:1019588 [0] NCCL INFO Channel 09/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-172-116:1019288:1019588 [0] NCCL INFO Channel 10/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-172-116:1019288:1019588 [0] NCCL INFO Channel 11/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-172-116:1019288:1019588 [0] NCCL INFO Channel 12/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-172-57:942611:942911 [0] NCCL INFO Channel 06/0 : 40[0] -> 45[5] via P2P/IPC
+ip-26-0-172-73:926594:926898 [0] NCCL INFO Channel 06/0 : 48[0] -> 53[5] via P2P/IPC
+ip-26-0-172-57:942611:942911 [0] NCCL INFO Channel 14/0 : 40[0] -> 45[5] via P2P/IPC
+ip-26-0-172-73:926594:926898 [0] NCCL INFO Channel 14/0 : 48[0] -> 53[5] via P2P/IPC
+ip-26-0-172-57:942611:942911 [0] NCCL INFO Channel 07/0 : 40[0] -> 46[6] via P2P/IPC
+ip-26-0-172-73:926594:926898 [0] NCCL INFO Channel 07/0 : 48[0] -> 54[6] via P2P/IPC
+ip-26-0-172-57:942611:942911 [0] NCCL INFO Channel 15/0 : 40[0] -> 46[6] via P2P/IPC
+ip-26-0-172-73:926594:926898 [0] NCCL INFO Channel 15/0 : 48[0] -> 54[6] via P2P/IPC
+ip-26-0-172-57:942611:942911 [0] NCCL INFO Channel 01/0 : 40[0] -> 47[7] via P2P/IPC
+ip-26-0-172-73:926594:926898 [0] NCCL INFO Channel 01/0 : 48[0] -> 55[7] via P2P/IPC
+ip-26-0-172-57:942611:942911 [0] NCCL INFO Channel 02/0 : 40[0] -> 47[7] via P2P/IPC
+ip-26-0-172-73:926594:926898 [0] NCCL INFO Channel 02/0 : 48[0] -> 55[7] via P2P/IPC
+ip-26-0-172-57:942611:942911 [0] NCCL INFO Channel 03/0 : 40[0] -> 47[7] via P2P/IPC
+ip-26-0-172-73:926594:926898 [0] NCCL INFO Channel 03/0 : 48[0] -> 55[7] via P2P/IPC
+ip-26-0-172-57:942611:942911 [0] NCCL INFO Channel 04/0 : 40[0] -> 47[7] via P2P/IPC
+ip-26-0-172-73:926594:926898 [0] NCCL INFO Channel 04/0 : 48[0] -> 55[7] via P2P/IPC
+ip-26-0-172-57:942611:942911 [0] NCCL INFO Channel 09/0 : 40[0] -> 47[7] via P2P/IPC
+ip-26-0-172-73:926594:926898 [0] NCCL INFO Channel 09/0 : 48[0] -> 55[7] via P2P/IPC
+ip-26-0-172-57:942611:942911 [0] NCCL INFO Channel 10/0 : 40[0] -> 47[7] via P2P/IPC
+ip-26-0-172-73:926594:926898 [0] NCCL INFO Channel 10/0 : 48[0] -> 55[7] via P2P/IPC
+ip-26-0-172-57:942611:942911 [0] NCCL INFO Channel 11/0 : 40[0] -> 47[7] via P2P/IPC
+ip-26-0-172-73:926594:926898 [0] NCCL INFO Channel 11/0 : 48[0] -> 55[7] via P2P/IPC
+ip-26-0-172-57:942611:942911 [0] NCCL INFO Channel 12/0 : 40[0] -> 47[7] via P2P/IPC
+ip-26-0-172-73:926594:926898 [0] NCCL INFO Channel 12/0 : 48[0] -> 55[7] via P2P/IPC
+ip-26-0-172-142:880680:880984 [1] NCCL INFO Channel 01/0 : 10[2] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880680:880984 [1] NCCL INFO Channel 09/0 : 10[2] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880681:880983 [2] NCCL INFO Channel 01/0 : 18[2] -> 25[1] [send] via NET/Libfabric/1(17)/GDRDMA
+ip-26-0-172-142:880681:880983 [2] NCCL INFO Channel 09/0 : 18[2] -> 25[1] [send] via NET/Libfabric/1(17)/GDRDMA
+ip-26-0-160-225:94343:94647 [1] NCCL INFO Channel 01/0 : 58[2] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:94344:94648 [2] NCCL INFO Channel 01/0 : 2[2] -> 9[1] [send] via NET/Libfabric/1(1)/GDRDMA
+ip-26-0-160-225:94343:94647 [1] NCCL INFO Channel 09/0 : 58[2] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:94344:94648 [2] NCCL INFO Channel 09/0 : 2[2] -> 9[1] [send] via NET/Libfabric/1(1)/GDRDMA
+ip-26-0-172-252:688201:688500 [1] NCCL INFO Channel 01/0 : 26[2] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618585:618884 [1] NCCL INFO Channel 01/0 : 18[2] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688201:688500 [1] NCCL INFO Channel 09/0 : 26[2] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688202:688503 [2] NCCL INFO Channel 01/0 : 34[2] -> 41[1] [send] via NET/Libfabric/1(33)/GDRDMA
+ip-26-0-172-147:618586:618886 [2] NCCL INFO Channel 01/0 : 26[2] -> 33[1] [send] via NET/Libfabric/1(25)/GDRDMA
+ip-26-0-172-252:688202:688503 [2] NCCL INFO Channel 09/0 : 34[2] -> 41[1] [send] via NET/Libfabric/1(33)/GDRDMA
+ip-26-0-172-147:618585:618884 [1] NCCL INFO Channel 09/0 : 18[2] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618586:618886 [2] NCCL INFO Channel 09/0 : 26[2] -> 33[1] [send] via NET/Libfabric/1(25)/GDRDMA
+ip-26-0-173-7:811249:811559 [1] NCCL INFO Channel 01/0 : 50[2] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-7:811250:811555 [2] NCCL INFO Channel 01/0 : 58[2] -> 1[1] [send] via NET/Libfabric/1(57)/GDRDMA
+ip-26-0-173-7:811249:811559 [1] NCCL INFO Channel 09/0 : 50[2] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-7:811250:811555 [2] NCCL INFO Channel 09/0 : 58[2] -> 1[1] [send] via NET/Libfabric/1(57)/GDRDMA
+ip-26-0-173-7:811252:811560 [4] NCCL INFO Channel 04/0 : 53[5] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-7:811252:811560 [4] NCCL INFO Channel 12/0 : 53[5] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-7:811252:811560 [4] NCCL INFO Channel 03/0 : 60[4] -> 3[3] [send] via NET/Libfabric/3(59)/GDRDMA
+ip-26-0-173-7:811252:811560 [4] NCCL INFO Channel 11/0 : 60[4] -> 3[3] [send] via NET/Libfabric/3(59)/GDRDMA
+ip-26-0-172-142:880683:880982 [4] NCCL INFO Channel 04/0 : 13[5] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880683:880982 [4] NCCL INFO Channel 12/0 : 13[5] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880683:880982 [4] NCCL INFO Channel 03/0 : 20[4] -> 27[3] [send] via NET/Libfabric/3(19)/GDRDMA
+ip-26-0-172-142:880683:880982 [4] NCCL INFO Channel 11/0 : 20[4] -> 27[3] [send] via NET/Libfabric/3(19)/GDRDMA
+ip-26-0-172-116:1019289:1019593 [1] NCCL INFO Channel 01/0 : 2[2] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019289:1019593 [1] NCCL INFO Channel 09/0 : 2[2] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019290:1019595 [2] NCCL INFO Channel 01/0 : 10[2] -> 17[1] [send] via NET/Libfabric/1(9)/GDRDMA
+ip-26-0-172-116:1019290:1019595 [2] NCCL INFO Channel 09/0 : 10[2] -> 17[1] [send] via NET/Libfabric/1(9)/GDRDMA
+ip-26-0-160-225:94346:94644 [4] NCCL INFO Channel 04/0 : 61[5] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880681:880983 [2] NCCL INFO Channel 02/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-160-225:94346:94644 [4] NCCL INFO Channel 12/0 : 61[5] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880685:880977 [6] NCCL INFO Channel 06/0 : 15[7] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880685:880977 [6] NCCL INFO Channel 14/0 : 15[7] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880686:880979 [7] NCCL INFO Channel 14/0 : 23[7] -> 30[6] [send] via NET/Libfabric/6(22)/GDRDMA
+ip-26-0-172-142:880681:880983 [2] NCCL INFO Channel 03/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-160-225:94346:94644 [4] NCCL INFO Channel 03/0 : 4[4] -> 11[3] [send] via NET/Libfabric/3(3)/GDRDMA
+ip-26-0-172-252:688201:688500 [1] NCCL INFO Channel 01/0 : 33[1] -> 32[0] via P2P/IPC
+ip-26-0-172-142:880681:880983 [2] NCCL INFO Channel 04/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-172-252:688201:688500 [1] NCCL INFO Channel 02/0 : 33[1] -> 32[0] via P2P/IPC
+ip-26-0-172-252:688201:688500 [1] NCCL INFO Channel 03/0 : 33[1] -> 32[0] via P2P/IPC
+ip-26-0-172-142:880681:880983 [2] NCCL INFO Channel 05/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-172-147:618589:618883 [5] NCCL INFO Channel 05/0 : 22[6] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688201:688500 [1] NCCL INFO Channel 04/0 : 33[1] -> 32[0] via P2P/IPC
+ip-26-0-172-147:618589:618883 [5] NCCL INFO Channel 13/0 : 22[6] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:94346:94644 [4] NCCL INFO Channel 11/0 : 4[4] -> 11[3] [send] via NET/Libfabric/3(3)/GDRDMA
+ip-26-0-172-142:880681:880983 [2] NCCL INFO Channel 06/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-172-147:618585:618884 [1] NCCL INFO Channel 01/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-172-252:688201:688500 [1] NCCL INFO Channel 05/0 : 33[1] -> 32[0] via P2P/IPC
+ip-26-0-172-147:618588:618888 [4] NCCL INFO Channel 04/0 : 21[5] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618586:618886 [2] NCCL INFO Channel 02/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-172-147:618588:618888 [4] NCCL INFO Channel 12/0 : 21[5] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618589:618883 [5] NCCL INFO Channel 04/0 : 29[5] -> 36[4] [send] via NET/Libfabric/4(28)/GDRDMA
+ip-26-0-172-252:688201:688500 [1] NCCL INFO Channel 06/0 : 33[1] -> 32[0] via P2P/IPC
+ip-26-0-172-147:618585:618884 [1] NCCL INFO Channel 02/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-172-147:618589:618883 [5] NCCL INFO Channel 12/0 : 29[5] -> 36[4] [send] via NET/Libfabric/4(28)/GDRDMA
+ip-26-0-172-147:618588:618888 [4] NCCL INFO Channel 03/0 : 28[4] -> 35[3] [send] via NET/Libfabric/3(27)/GDRDMA
+ip-26-0-172-147:618586:618886 [2] NCCL INFO Channel 03/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-172-142:880684:880981 [5] NCCL INFO Channel 05/0 : 14[6] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926596:926899 [2] NCCL INFO Channel 02/0 : 50[2] -> 49[1] via P2P/IPC
+ip-26-0-172-252:688201:688500 [1] NCCL INFO Channel 07/0 : 33[1] -> 32[0] via P2P/IPC
+ip-26-0-172-147:618588:618888 [4] NCCL INFO Channel 11/0 : 28[4] -> 35[3] [send] via NET/Libfabric/3(27)/GDRDMA
+ip-26-0-172-147:618585:618884 [1] NCCL INFO Channel 03/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-172-147:618590:618889 [6] NCCL INFO Channel 06/0 : 23[7] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880684:880981 [5] NCCL INFO Channel 13/0 : 14[6] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880681:880983 [2] NCCL INFO Channel 07/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-172-142:880685:880977 [6] NCCL INFO Channel 05/0 : 22[6] -> 29[5] [send] via NET/Libfabric/5(21)/GDRDMA
+ip-26-0-172-73:926596:926899 [2] NCCL INFO Channel 03/0 : 50[2] -> 49[1] via P2P/IPC
+ip-26-0-172-252:688201:688500 [1] NCCL INFO Channel 09/0 : 33[1] -> 32[0] via P2P/IPC
+ip-26-0-172-147:618591:618885 [7] NCCL INFO Channel 14/0 : 31[7] -> 38[6] [send] via NET/Libfabric/6(30)/GDRDMA
+ip-26-0-173-7:811253:811558 [5] NCCL INFO Channel 05/0 : 54[6] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618590:618889 [6] NCCL INFO Channel 14/0 : 23[7] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618586:618886 [2] NCCL INFO Channel 04/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-172-142:880685:880977 [6] NCCL INFO Channel 13/0 : 22[6] -> 29[5] [send] via NET/Libfabric/5(21)/GDRDMA
+ip-26-0-172-147:618585:618884 [1] NCCL INFO Channel 04/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-172-142:880684:880981 [5] NCCL INFO Channel 04/0 : 21[5] -> 28[4] [send] via NET/Libfabric/4(20)/GDRDMA
+ip-26-0-173-7:811253:811558 [5] NCCL INFO Channel 13/0 : 54[6] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880684:880981 [5] NCCL INFO Channel 12/0 : 21[5] -> 28[4] [send] via NET/Libfabric/4(20)/GDRDMA
+ip-26-0-172-73:926596:926899 [2] NCCL INFO Channel 04/0 : 50[2] -> 49[1] via P2P/IPC
+ip-26-0-172-252:688201:688500 [1] NCCL INFO Channel 10/0 : 33[1] -> 32[0] via P2P/IPC
+ip-26-0-160-225:94347:94646 [5] NCCL INFO Channel 05/0 : 62[6] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618590:618889 [6] NCCL INFO Channel 05/0 : 30[6] -> 37[5] [send] via NET/Libfabric/5(29)/GDRDMA
+ip-26-0-172-147:618590:618889 [6] NCCL INFO Channel 13/0 : 30[6] -> 37[5] [send] via NET/Libfabric/5(29)/GDRDMA
+ip-26-0-172-142:880681:880983 [2] NCCL INFO Channel 10/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-160-225:94347:94646 [5] NCCL INFO Channel 13/0 : 62[6] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-7:811253:811558 [5] NCCL INFO Channel 04/0 : 61[5] -> 4[4] [send] via NET/Libfabric/4(60)/GDRDMA
+ip-26-0-173-7:811249:811559 [1] NCCL INFO Channel 01/0 : 57[1] -> 56[0] via P2P/IPC
+ip-26-0-173-7:811254:811557 [6] NCCL INFO Channel 06/0 : 55[7] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:94343:94647 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-173-7:811253:811558 [5] NCCL INFO Channel 12/0 : 61[5] -> 4[4] [send] via NET/Libfabric/4(60)/GDRDMA
+ip-26-0-172-142:880685:880977 [6] NCCL INFO Channel 06/0 : 22[6] -> 19[3] via P2P/IPC
+ip-26-0-172-147:618586:618886 [2] NCCL INFO Channel 05/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-172-252:688201:688500 [1] NCCL INFO Channel 11/0 : 33[1] -> 32[0] via P2P/IPC
+ip-26-0-160-225:94347:94646 [5] NCCL INFO Channel 04/0 : 5[5] -> 12[4] [send] via NET/Libfabric/4(4)/GDRDMA
+ip-26-0-173-7:811255:811561 [7] NCCL INFO Channel 14/0 : 63[7] -> 6[6] [send] via NET/Libfabric/6(62)/GDRDMA
+ip-26-0-173-7:811254:811557 [6] NCCL INFO Channel 14/0 : 55[7] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618585:618884 [1] NCCL INFO Channel 05/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-160-225:94347:94646 [5] NCCL INFO Channel 12/0 : 5[5] -> 12[4] [send] via NET/Libfabric/4(4)/GDRDMA
+ip-26-0-172-142:880681:880983 [2] NCCL INFO Channel 11/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-160-225:94348:94650 [6] NCCL INFO Channel 06/0 : 63[7] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-7:811249:811559 [1] NCCL INFO Channel 02/0 : 57[1] -> 56[0] via P2P/IPC
+ip-26-0-160-225:94348:94650 [6] NCCL INFO Channel 14/0 : 63[7] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-7:811254:811557 [6] NCCL INFO Channel 05/0 : 62[6] -> 5[5] [send] via NET/Libfabric/5(61)/GDRDMA
+ip-26-0-173-7:811250:811555 [2] NCCL INFO Channel 02/0 : 58[2] -> 57[1] via P2P/IPC
+ip-26-0-172-252:688201:688500 [1] NCCL INFO Channel 12/0 : 33[1] -> 32[0] via P2P/IPC
+ip-26-0-160-225:94343:94647 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-173-7:811254:811557 [6] NCCL INFO Channel 13/0 : 62[6] -> 5[5] [send] via NET/Libfabric/5(61)/GDRDMA
+ip-26-0-160-225:94348:94650 [6] NCCL INFO Channel 05/0 : 6[6] -> 13[5] [send] via NET/Libfabric/5(5)/GDRDMA
+ip-26-0-172-147:618586:618886 [2] NCCL INFO Channel 06/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-172-147:618585:618884 [1] NCCL INFO Channel 06/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-160-225:94348:94650 [6] NCCL INFO Channel 13/0 : 6[6] -> 13[5] [send] via NET/Libfabric/5(5)/GDRDMA
+ip-26-0-172-252:688201:688500 [1] NCCL INFO Channel 13/0 : 33[1] -> 32[0] via P2P/IPC
+ip-26-0-172-142:880685:880977 [6] NCCL INFO Channel 14/0 : 22[6] -> 19[3] via P2P/IPC
+ip-26-0-160-225:94343:94647 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-142:880681:880983 [2] NCCL INFO Channel 12/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-173-7:811249:811559 [1] NCCL INFO Channel 03/0 : 57[1] -> 56[0] via P2P/IPC
+ip-26-0-173-7:811254:811557 [6] NCCL INFO Channel 06/0 : 62[6] -> 59[3] via P2P/IPC
+ip-26-0-172-252:688201:688500 [1] NCCL INFO Channel 14/0 : 33[1] -> 32[0] via P2P/IPC
+ip-26-0-173-7:811250:811555 [2] NCCL INFO Channel 03/0 : 58[2] -> 57[1] via P2P/IPC
+ip-26-0-172-73:926596:926899 [2] NCCL INFO Channel 05/0 : 50[2] -> 49[1] via P2P/IPC
+ip-26-0-172-147:618586:618886 [2] NCCL INFO Channel 07/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-172-147:618585:618884 [1] NCCL INFO Channel 07/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-172-252:688201:688500 [1] NCCL INFO Channel 15/0 : 33[1] -> 32[0] via P2P/IPC
+ip-26-0-160-225:94343:94647 [1] NCCL INFO Channel 04/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-116:1019294:1019589 [6] NCCL INFO Channel 06/0 : 7[7] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926596:926899 [2] NCCL INFO Channel 06/0 : 50[2] -> 49[1] via P2P/IPC
+ip-26-0-172-116:1019294:1019589 [6] NCCL INFO Channel 14/0 : 7[7] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880681:880983 [2] NCCL INFO Channel 13/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-172-252:688206:688506 [6] NCCL INFO Channel 06/0 : 31[7] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688207:688504 [7] NCCL INFO Channel 14/0 : 39[7] -> 46[6] [send] via NET/Libfabric/6(38)/GDRDMA
+ip-26-0-173-7:811249:811559 [1] NCCL INFO Channel 04/0 : 57[1] -> 56[0] via P2P/IPC
+ip-26-0-160-225:94343:94647 [1] NCCL INFO Channel 05/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-73:926596:926899 [2] NCCL INFO Channel 07/0 : 50[2] -> 49[1] via P2P/IPC
+ip-26-0-172-252:688206:688506 [6] NCCL INFO Channel 14/0 : 31[7] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-7:811254:811557 [6] NCCL INFO Channel 14/0 : 62[6] -> 59[3] via P2P/IPC
+ip-26-0-173-7:811250:811555 [2] NCCL INFO Channel 04/0 : 58[2] -> 57[1] via P2P/IPC
+ip-26-0-172-147:618585:618884 [1] NCCL INFO Channel 09/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-172-147:618586:618886 [2] NCCL INFO Channel 10/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-172-142:880681:880983 [2] NCCL INFO Channel 14/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-172-73:926596:926899 [2] NCCL INFO Channel 10/0 : 50[2] -> 49[1] via P2P/IPC
+ip-26-0-160-225:94343:94647 [1] NCCL INFO Channel 06/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-147:618588:618888 [4] NCCL INFO Channel 07/0 : 28[4] -> 39[7] [send] via NET/Libfabric/7(31)/GDRDMA
+ip-26-0-160-225:94344:94648 [2] NCCL INFO Channel 02/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-252:688204:688502 [4] NCCL INFO Channel 04/0 : 29[5] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019289:1019593 [1] NCCL INFO Channel 01/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-172-147:618588:618888 [4] NCCL INFO Channel 15/0 : 28[4] -> 39[7] [send] via NET/Libfabric/7(31)/GDRDMA
+ip-26-0-172-73:926596:926899 [2] NCCL INFO Channel 11/0 : 50[2] -> 49[1] via P2P/IPC
+ip-26-0-172-252:688204:688502 [4] NCCL INFO Channel 12/0 : 29[5] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:94346:94644 [4] NCCL INFO Channel 07/0 : 4[4] -> 15[7] [send] via NET/Libfabric/7(7)/GDRDMA
+ip-26-0-173-7:811249:811559 [1] NCCL INFO Channel 05/0 : 57[1] -> 56[0] via P2P/IPC
+ip-26-0-172-147:618585:618884 [1] NCCL INFO Channel 10/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-172-147:618586:618886 [2] NCCL INFO Channel 11/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-172-116:1019290:1019595 [2] NCCL INFO Channel 02/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-173-7:811250:811555 [2] NCCL INFO Channel 05/0 : 58[2] -> 57[1] via P2P/IPC
+ip-26-0-160-225:94346:94644 [4] NCCL INFO Channel 15/0 : 4[4] -> 15[7] [send] via NET/Libfabric/7(7)/GDRDMA
+ip-26-0-172-142:880681:880983 [2] NCCL INFO Channel 15/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-160-225:94344:94648 [2] NCCL INFO Channel 03/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-73:926596:926899 [2] NCCL INFO Channel 12/0 : 50[2] -> 49[1] via P2P/IPC
+ip-26-0-160-225:94343:94647 [1] NCCL INFO Channel 07/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-116:1019289:1019593 [1] NCCL INFO Channel 02/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-172-252:688205:688505 [5] NCCL INFO Channel 05/0 : 30[6] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688204:688502 [4] NCCL INFO Channel 03/0 : 36[4] -> 43[3] [send] via NET/Libfabric/3(35)/GDRDMA
+ip-26-0-172-252:688205:688505 [5] NCCL INFO Channel 13/0 : 30[6] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926596:926899 [2] NCCL INFO Channel 13/0 : 50[2] -> 49[1] via P2P/IPC
+ip-26-0-172-116:1019290:1019595 [2] NCCL INFO Channel 03/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-172-252:688206:688506 [6] NCCL INFO Channel 05/0 : 38[6] -> 45[5] [send] via NET/Libfabric/5(37)/GDRDMA
+ip-26-0-172-116:1019289:1019593 [1] NCCL INFO Channel 03/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-172-57:942612:942914 [1] NCCL INFO Channel 01/0 : 34[2] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688206:688506 [6] NCCL INFO Channel 13/0 : 38[6] -> 45[5] [send] via NET/Libfabric/5(37)/GDRDMA
+ip-26-0-172-147:618585:618884 [1] NCCL INFO Channel 11/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-160-225:94344:94648 [2] NCCL INFO Channel 04/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-225:94343:94647 [1] NCCL INFO Channel 09/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-57:942612:942914 [1] NCCL INFO Channel 09/0 : 34[2] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688205:688505 [5] NCCL INFO Channel 04/0 : 37[5] -> 44[4] [send] via NET/Libfabric/4(36)/GDRDMA
+ip-26-0-172-147:618586:618886 [2] NCCL INFO Channel 12/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-172-73:926596:926899 [2] NCCL INFO Channel 14/0 : 50[2] -> 49[1] via P2P/IPC
+ip-26-0-172-252:688205:688505 [5] NCCL INFO Channel 12/0 : 37[5] -> 44[4] [send] via NET/Libfabric/4(36)/GDRDMA
+ip-26-0-172-252:688204:688502 [4] NCCL INFO Channel 11/0 : 36[4] -> 43[3] [send] via NET/Libfabric/3(35)/GDRDMA
+ip-26-0-172-116:1019290:1019595 [2] NCCL INFO Channel 04/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-172-116:1019289:1019593 [1] NCCL INFO Channel 04/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-172-73:926596:926899 [2] NCCL INFO Channel 15/0 : 50[2] -> 49[1] via P2P/IPC
+ip-26-0-160-225:94344:94648 [2] NCCL INFO Channel 05/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-225:94343:94647 [1] NCCL INFO Channel 10/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-147:618589:618883 [5] NCCL INFO Channel 05/0 : 29[5] -> 27[3] via P2P/IPC
+ip-26-0-172-147:618590:618889 [6] NCCL INFO Channel 06/0 : 30[6] -> 27[3] via P2P/IPC
+ip-26-0-172-116:1019292:1019590 [4] NCCL INFO Channel 04/0 : 5[5] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618585:618884 [1] NCCL INFO Channel 12/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-172-147:618586:618886 [2] NCCL INFO Channel 13/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-172-116:1019293:1019592 [5] NCCL INFO Channel 05/0 : 6[6] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019292:1019590 [4] NCCL INFO Channel 12/0 : 5[5] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019290:1019595 [2] NCCL INFO Channel 05/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-172-116:1019293:1019592 [5] NCCL INFO Channel 13/0 : 6[6] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019289:1019593 [1] NCCL INFO Channel 05/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-172-116:1019294:1019589 [6] NCCL INFO Channel 05/0 : 14[6] -> 21[5] [send] via NET/Libfabric/5(13)/GDRDMA
+ip-26-0-160-225:94344:94648 [2] NCCL INFO Channel 06/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-225:94343:94647 [1] NCCL INFO Channel 11/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-116:1019294:1019589 [6] NCCL INFO Channel 13/0 : 14[6] -> 21[5] [send] via NET/Libfabric/5(13)/GDRDMA
+ip-26-0-172-116:1019292:1019590 [4] NCCL INFO Channel 03/0 : 12[4] -> 19[3] [send] via NET/Libfabric/3(11)/GDRDMA
+ip-26-0-172-116:1019293:1019592 [5] NCCL INFO Channel 04/0 : 13[5] -> 20[4] [send] via NET/Libfabric/4(12)/GDRDMA
+ip-26-0-172-142:880680:880984 [1] NCCL INFO Channel 01/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-172-116:1019292:1019590 [4] NCCL INFO Channel 11/0 : 12[4] -> 19[3] [send] via NET/Libfabric/3(11)/GDRDMA
+ip-26-0-172-116:1019293:1019592 [5] NCCL INFO Channel 12/0 : 13[5] -> 20[4] [send] via NET/Libfabric/4(12)/GDRDMA
+ip-26-0-160-225:94344:94648 [2] NCCL INFO Channel 07/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-225:94343:94647 [1] NCCL INFO Channel 12/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-142:880680:880984 [1] NCCL INFO Channel 02/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-160-225:94347:94646 [5] NCCL INFO Channel 05/0 : 5[5] -> 3[3] via P2P/IPC
+ip-26-0-172-116:1019294:1019589 [6] NCCL INFO Channel 06/0 : 14[6] -> 11[3] via P2P/IPC
+ip-26-0-172-142:880684:880981 [5] NCCL INFO Channel 05/0 : 21[5] -> 19[3] via P2P/IPC
+ip-26-0-160-225:94348:94650 [6] NCCL INFO Channel 06/0 : 6[6] -> 3[3] via P2P/IPC
+ip-26-0-172-147:618590:618889 [6] NCCL INFO Channel 14/0 : 30[6] -> 27[3] via P2P/IPC
+ip-26-0-172-147:618585:618884 [1] NCCL INFO Channel 13/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-172-147:618589:618883 [5] NCCL INFO Channel 13/0 : 29[5] -> 27[3] via P2P/IPC
+ip-26-0-172-147:618586:618886 [2] NCCL INFO Channel 14/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-172-116:1019293:1019592 [5] NCCL INFO Channel 05/0 : 13[5] -> 11[3] via P2P/IPC
+ip-26-0-172-142:880680:880984 [1] NCCL INFO Channel 03/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-160-225:94344:94648 [2] NCCL INFO Channel 10/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-116:1019290:1019595 [2] NCCL INFO Channel 06/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-172-116:1019289:1019593 [1] NCCL INFO Channel 06/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-160-225:94343:94647 [1] NCCL INFO Channel 13/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-147:618585:618884 [1] NCCL INFO Channel 14/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-172-147:618586:618886 [2] NCCL INFO Channel 15/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-160-225:94348:94650 [6] NCCL INFO Channel 14/0 : 6[6] -> 3[3] via P2P/IPC
+ip-26-0-172-142:880684:880981 [5] NCCL INFO Channel 13/0 : 21[5] -> 19[3] via P2P/IPC
+ip-26-0-160-225:94347:94646 [5] NCCL INFO Channel 13/0 : 5[5] -> 3[3] via P2P/IPC
+ip-26-0-172-252:688204:688502 [4] NCCL INFO Channel 07/0 : 36[4] -> 47[7] [send] via NET/Libfabric/7(39)/GDRDMA
+ip-26-0-172-142:880680:880984 [1] NCCL INFO Channel 04/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-172-252:688204:688502 [4] NCCL INFO Channel 15/0 : 36[4] -> 47[7] [send] via NET/Libfabric/7(39)/GDRDMA
+ip-26-0-160-225:94344:94648 [2] NCCL INFO Channel 11/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-225:94343:94647 [1] NCCL INFO Channel 14/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-147:618585:618884 [1] NCCL INFO Channel 15/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-172-116:1019294:1019589 [6] NCCL INFO Channel 14/0 : 14[6] -> 11[3] via P2P/IPC
+ip-26-0-172-116:1019290:1019595 [2] NCCL INFO Channel 07/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-172-57:942615:942912 [4] NCCL INFO Channel 04/0 : 37[5] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019289:1019593 [1] NCCL INFO Channel 07/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-172-116:1019293:1019592 [5] NCCL INFO Channel 13/0 : 13[5] -> 11[3] via P2P/IPC
+ip-26-0-172-57:942615:942912 [4] NCCL INFO Channel 12/0 : 37[5] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:94344:94648 [2] NCCL INFO Channel 12/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-225:94343:94647 [1] NCCL INFO Channel 15/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-57:942615:942912 [4] NCCL INFO Channel 03/0 : 44[4] -> 51[3] [send] via NET/Libfabric/3(43)/GDRDMA
+ip-26-0-172-142:880680:880984 [1] NCCL INFO Channel 05/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-172-57:942615:942912 [4] NCCL INFO Channel 11/0 : 44[4] -> 51[3] [send] via NET/Libfabric/3(43)/GDRDMA
+ip-26-0-160-225:94344:94648 [2] NCCL INFO Channel 13/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-116:1019290:1019595 [2] NCCL INFO Channel 10/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-172-116:1019289:1019593 [1] NCCL INFO Channel 09/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-172-142:880680:880984 [1] NCCL INFO Channel 06/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-173-7:811250:811555 [2] NCCL INFO Channel 06/0 : 58[2] -> 57[1] via P2P/IPC
+ip-26-0-172-252:688205:688505 [5] NCCL INFO Channel 05/0 : 37[5] -> 35[3] via P2P/IPC
+ip-26-0-173-7:811249:811559 [1] NCCL INFO Channel 06/0 : 57[1] -> 56[0] via P2P/IPC
+ip-26-0-172-57:942612:942914 [1] NCCL INFO Channel 01/0 : 41[1] -> 40[0] via P2P/IPC
+ip-26-0-160-225:94344:94648 [2] NCCL INFO Channel 14/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-116:1019290:1019595 [2] NCCL INFO Channel 11/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-172-142:880680:880984 [1] NCCL INFO Channel 07/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-172-57:942612:942914 [1] NCCL INFO Channel 02/0 : 41[1] -> 40[0] via P2P/IPC
+ip-26-0-172-116:1019289:1019593 [1] NCCL INFO Channel 10/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-172-57:942612:942914 [1] NCCL INFO Channel 03/0 : 41[1] -> 40[0] via P2P/IPC
+ip-26-0-173-7:811250:811555 [2] NCCL INFO Channel 07/0 : 58[2] -> 57[1] via P2P/IPC
+ip-26-0-173-7:811249:811559 [1] NCCL INFO Channel 07/0 : 57[1] -> 56[0] via P2P/IPC
+ip-26-0-172-252:688205:688505 [5] NCCL INFO Channel 13/0 : 37[5] -> 35[3] via P2P/IPC
+ip-26-0-172-57:942612:942914 [1] NCCL INFO Channel 04/0 : 41[1] -> 40[0] via P2P/IPC
+ip-26-0-172-73:926595:926897 [1] NCCL INFO Channel 01/0 : 42[2] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926595:926897 [1] NCCL INFO Channel 09/0 : 42[2] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-7:811250:811555 [2] NCCL INFO Channel 10/0 : 58[2] -> 57[1] via P2P/IPC
+ip-26-0-172-57:942612:942914 [1] NCCL INFO Channel 05/0 : 41[1] -> 40[0] via P2P/IPC
+ip-26-0-173-7:811249:811559 [1] NCCL INFO Channel 09/0 : 57[1] -> 56[0] via P2P/IPC
+ip-26-0-172-57:942612:942914 [1] NCCL INFO Channel 06/0 : 41[1] -> 40[0] via P2P/IPC
+ip-26-0-172-116:1019290:1019595 [2] NCCL INFO Channel 12/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-173-7:811250:811555 [2] NCCL INFO Channel 11/0 : 58[2] -> 57[1] via P2P/IPC
+ip-26-0-172-116:1019289:1019593 [1] NCCL INFO Channel 11/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-172-142:880680:880984 [1] NCCL INFO Channel 09/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-173-7:811249:811559 [1] NCCL INFO Channel 10/0 : 57[1] -> 56[0] via P2P/IPC
+ip-26-0-172-57:942612:942914 [1] NCCL INFO Channel 07/0 : 41[1] -> 40[0] via P2P/IPC
+ip-26-0-172-142:880680:880984 [1] NCCL INFO Channel 10/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-172-116:1019290:1019595 [2] NCCL INFO Channel 13/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-172-116:1019289:1019593 [1] NCCL INFO Channel 12/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-173-7:811250:811555 [2] NCCL INFO Channel 12/0 : 58[2] -> 57[1] via P2P/IPC
+ip-26-0-173-7:811249:811559 [1] NCCL INFO Channel 11/0 : 57[1] -> 56[0] via P2P/IPC
+ip-26-0-172-57:942612:942914 [1] NCCL INFO Channel 09/0 : 41[1] -> 40[0] via P2P/IPC
+ip-26-0-172-142:880680:880984 [1] NCCL INFO Channel 11/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-172-57:942617:942916 [6] NCCL INFO Channel 06/0 : 39[7] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942615:942912 [4] NCCL INFO Channel 07/0 : 44[4] -> 55[7] [send] via NET/Libfabric/7(47)/GDRDMA
+ip-26-0-172-57:942615:942912 [4] NCCL INFO Channel 15/0 : 44[4] -> 55[7] [send] via NET/Libfabric/7(47)/GDRDMA
+ip-26-0-172-116:1019290:1019595 [2] NCCL INFO Channel 14/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-172-57:942617:942916 [6] NCCL INFO Channel 14/0 : 39[7] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019289:1019593 [1] NCCL INFO Channel 13/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-172-57:942612:942914 [1] NCCL INFO Channel 10/0 : 41[1] -> 40[0] via P2P/IPC
+ip-26-0-173-7:811250:811555 [2] NCCL INFO Channel 13/0 : 58[2] -> 57[1] via P2P/IPC
+ip-26-0-173-7:811249:811559 [1] NCCL INFO Channel 12/0 : 57[1] -> 56[0] via P2P/IPC
+ip-26-0-172-57:942612:942914 [1] NCCL INFO Channel 11/0 : 41[1] -> 40[0] via P2P/IPC
+ip-26-0-172-116:1019290:1019595 [2] NCCL INFO Channel 15/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-172-116:1019289:1019593 [1] NCCL INFO Channel 14/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-160-225:94344:94648 [2] NCCL INFO Channel 15/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-57:942612:942914 [1] NCCL INFO Channel 12/0 : 41[1] -> 40[0] via P2P/IPC
+ip-26-0-173-7:811249:811559 [1] NCCL INFO Channel 13/0 : 57[1] -> 56[0] via P2P/IPC
+ip-26-0-173-7:811250:811555 [2] NCCL INFO Channel 14/0 : 58[2] -> 57[1] via P2P/IPC
+ip-26-0-172-57:942616:942913 [5] NCCL INFO Channel 05/0 : 38[6] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019289:1019593 [1] NCCL INFO Channel 15/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-172-57:942616:942913 [5] NCCL INFO Channel 13/0 : 38[6] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942612:942914 [1] NCCL INFO Channel 13/0 : 41[1] -> 40[0] via P2P/IPC
+ip-26-0-172-57:942617:942916 [6] NCCL INFO Channel 05/0 : 46[6] -> 53[5] [send] via NET/Libfabric/5(45)/GDRDMA
+ip-26-0-172-73:926598:926894 [4] NCCL INFO Channel 04/0 : 45[5] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-7:811249:811559 [1] NCCL INFO Channel 14/0 : 57[1] -> 56[0] via P2P/IPC
+ip-26-0-172-73:926598:926894 [4] NCCL INFO Channel 12/0 : 45[5] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-7:811250:811555 [2] NCCL INFO Channel 15/0 : 58[2] -> 57[1] via P2P/IPC
+ip-26-0-172-73:926598:926894 [4] NCCL INFO Channel 03/0 : 52[4] -> 59[3] [send] via NET/Libfabric/3(51)/GDRDMA
+ip-26-0-172-252:688206:688506 [6] NCCL INFO Channel 06/0 : 38[6] -> 35[3] via P2P/IPC
+ip-26-0-172-73:926598:926894 [4] NCCL INFO Channel 11/0 : 52[4] -> 59[3] [send] via NET/Libfabric/3(51)/GDRDMA
+ip-26-0-173-7:811249:811559 [1] NCCL INFO Channel 15/0 : 57[1] -> 56[0] via P2P/IPC
+ip-26-0-172-73:926600:926893 [6] NCCL INFO Channel 06/0 : 47[7] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926599:926895 [5] NCCL INFO Channel 05/0 : 46[6] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926600:926893 [6] NCCL INFO Channel 14/0 : 47[7] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926599:926895 [5] NCCL INFO Channel 13/0 : 46[6] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880680:880984 [1] NCCL INFO Channel 12/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-172-73:926600:926893 [6] NCCL INFO Channel 05/0 : 54[6] -> 61[5] [send] via NET/Libfabric/5(53)/GDRDMA
+ip-26-0-172-73:926595:926897 [1] NCCL INFO Channel 01/0 : 49[1] -> 48[0] via P2P/IPC
+ip-26-0-172-73:926599:926895 [5] NCCL INFO Channel 04/0 : 53[5] -> 60[4] [send] via NET/Libfabric/4(52)/GDRDMA
+ip-26-0-172-73:926600:926893 [6] NCCL INFO Channel 13/0 : 54[6] -> 61[5] [send] via NET/Libfabric/5(53)/GDRDMA
+ip-26-0-172-73:926599:926895 [5] NCCL INFO Channel 12/0 : 53[5] -> 60[4] [send] via NET/Libfabric/4(52)/GDRDMA
+ip-26-0-172-142:880680:880984 [1] NCCL INFO Channel 13/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-172-73:926595:926897 [1] NCCL INFO Channel 02/0 : 49[1] -> 48[0] via P2P/IPC
+ip-26-0-172-142:880680:880984 [1] NCCL INFO Channel 14/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-172-73:926600:926893 [6] NCCL INFO Channel 06/0 : 54[6] -> 51[3] via P2P/IPC
+ip-26-0-173-7:811253:811558 [5] NCCL INFO Channel 05/0 : 61[5] -> 59[3] via P2P/IPC
+ip-26-0-172-73:926599:926895 [5] NCCL INFO Channel 05/0 : 53[5] -> 51[3] via P2P/IPC
+ip-26-0-172-142:880680:880984 [1] NCCL INFO Channel 15/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-172-252:688206:688506 [6] NCCL INFO Channel 14/0 : 38[6] -> 35[3] via P2P/IPC
+ip-26-0-172-57:942617:942916 [6] NCCL INFO Channel 13/0 : 46[6] -> 53[5] [send] via NET/Libfabric/5(45)/GDRDMA
+ip-26-0-172-57:942616:942913 [5] NCCL INFO Channel 04/0 : 45[5] -> 52[4] [send] via NET/Libfabric/4(44)/GDRDMA
+ip-26-0-172-57:942612:942914 [1] NCCL INFO Channel 14/0 : 41[1] -> 40[0] via P2P/IPC
+ip-26-0-172-57:942616:942913 [5] NCCL INFO Channel 12/0 : 45[5] -> 52[4] [send] via NET/Libfabric/4(44)/GDRDMA
+ip-26-0-172-57:942612:942914 [1] NCCL INFO Channel 15/0 : 41[1] -> 40[0] via P2P/IPC
+ip-26-0-172-57:942616:942913 [5] NCCL INFO Channel 05/0 : 45[5] -> 43[3] via P2P/IPC
+ip-26-0-172-57:942617:942916 [6] NCCL INFO Channel 06/0 : 46[6] -> 43[3] via P2P/IPC
+ip-26-0-172-73:926595:926897 [1] NCCL INFO Channel 03/0 : 49[1] -> 48[0] via P2P/IPC
+ip-26-0-172-73:926600:926893 [6] NCCL INFO Channel 14/0 : 54[6] -> 51[3] via P2P/IPC
+ip-26-0-172-73:926599:926895 [5] NCCL INFO Channel 13/0 : 53[5] -> 51[3] via P2P/IPC
+ip-26-0-172-73:926595:926897 [1] NCCL INFO Channel 04/0 : 49[1] -> 48[0] via P2P/IPC
+ip-26-0-172-252:688202:688503 [2] NCCL INFO Channel 02/0 : 34[2] -> 33[1] via P2P/IPC
+ip-26-0-172-73:926595:926897 [1] NCCL INFO Channel 05/0 : 49[1] -> 48[0] via P2P/IPC
+ip-26-0-172-73:926595:926897 [1] NCCL INFO Channel 06/0 : 49[1] -> 48[0] via P2P/IPC
+ip-26-0-172-57:942616:942913 [5] NCCL INFO Channel 13/0 : 45[5] -> 43[3] via P2P/IPC
+ip-26-0-172-57:942617:942916 [6] NCCL INFO Channel 14/0 : 46[6] -> 43[3] via P2P/IPC
+ip-26-0-172-73:926595:926897 [1] NCCL INFO Channel 07/0 : 49[1] -> 48[0] via P2P/IPC
+ip-26-0-172-73:926595:926897 [1] NCCL INFO Channel 09/0 : 49[1] -> 48[0] via P2P/IPC
+ip-26-0-173-7:811253:811558 [5] NCCL INFO Channel 13/0 : 61[5] -> 59[3] via P2P/IPC
+ip-26-0-172-73:926595:926897 [1] NCCL INFO Channel 10/0 : 49[1] -> 48[0] via P2P/IPC
+ip-26-0-172-73:926595:926897 [1] NCCL INFO Channel 11/0 : 49[1] -> 48[0] via P2P/IPC
+ip-26-0-172-73:926595:926897 [1] NCCL INFO Channel 12/0 : 49[1] -> 48[0] via P2P/IPC
+ip-26-0-172-73:926595:926897 [1] NCCL INFO Channel 13/0 : 49[1] -> 48[0] via P2P/IPC
+ip-26-0-172-73:926595:926897 [1] NCCL INFO Channel 14/0 : 49[1] -> 48[0] via P2P/IPC
+ip-26-0-172-73:926595:926897 [1] NCCL INFO Channel 15/0 : 49[1] -> 48[0] via P2P/IPC
+ip-26-0-172-252:688202:688503 [2] NCCL INFO Channel 03/0 : 34[2] -> 33[1] via P2P/IPC
+ip-26-0-172-252:688202:688503 [2] NCCL INFO Channel 04/0 : 34[2] -> 33[1] via P2P/IPC
+ip-26-0-172-252:688202:688503 [2] NCCL INFO Channel 05/0 : 34[2] -> 33[1] via P2P/IPC
+ip-26-0-172-252:688202:688503 [2] NCCL INFO Channel 06/0 : 34[2] -> 33[1] via P2P/IPC
+ip-26-0-172-252:688202:688503 [2] NCCL INFO Channel 07/0 : 34[2] -> 33[1] via P2P/IPC
+ip-26-0-172-252:688202:688503 [2] NCCL INFO Channel 10/0 : 34[2] -> 33[1] via P2P/IPC
+ip-26-0-172-252:688202:688503 [2] NCCL INFO Channel 11/0 : 34[2] -> 33[1] via P2P/IPC
+ip-26-0-172-252:688202:688503 [2] NCCL INFO Channel 12/0 : 34[2] -> 33[1] via P2P/IPC
+ip-26-0-172-252:688202:688503 [2] NCCL INFO Channel 13/0 : 34[2] -> 33[1] via P2P/IPC
+ip-26-0-172-57:942613:942917 [2] NCCL INFO Channel 02/0 : 42[2] -> 41[1] via P2P/IPC
+ip-26-0-172-252:688202:688503 [2] NCCL INFO Channel 14/0 : 34[2] -> 33[1] via P2P/IPC
+ip-26-0-172-57:942613:942917 [2] NCCL INFO Channel 03/0 : 42[2] -> 41[1] via P2P/IPC
+ip-26-0-172-57:942613:942917 [2] NCCL INFO Channel 04/0 : 42[2] -> 41[1] via P2P/IPC
+ip-26-0-172-252:688202:688503 [2] NCCL INFO Channel 15/0 : 34[2] -> 33[1] via P2P/IPC
+ip-26-0-172-57:942613:942917 [2] NCCL INFO Channel 05/0 : 42[2] -> 41[1] via P2P/IPC
+ip-26-0-172-57:942613:942917 [2] NCCL INFO Channel 06/0 : 42[2] -> 41[1] via P2P/IPC
+ip-26-0-172-57:942613:942917 [2] NCCL INFO Channel 07/0 : 42[2] -> 41[1] via P2P/IPC
+ip-26-0-172-57:942613:942917 [2] NCCL INFO Channel 10/0 : 42[2] -> 41[1] via P2P/IPC
+ip-26-0-172-57:942613:942917 [2] NCCL INFO Channel 11/0 : 42[2] -> 41[1] via P2P/IPC
+ip-26-0-172-57:942613:942917 [2] NCCL INFO Channel 12/0 : 42[2] -> 41[1] via P2P/IPC
+ip-26-0-172-57:942613:942917 [2] NCCL INFO Channel 13/0 : 42[2] -> 41[1] via P2P/IPC
+ip-26-0-172-57:942613:942917 [2] NCCL INFO Channel 14/0 : 42[2] -> 41[1] via P2P/IPC
+ip-26-0-172-57:942613:942917 [2] NCCL INFO Channel 15/0 : 42[2] -> 41[1] via P2P/IPC
+ip-26-0-173-7:811255:811561 [7] NCCL INFO Channel 07/0 : 52[4] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-7:811255:811561 [7] NCCL INFO Channel 15/0 : 52[4] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-7:811252:811560 [4] NCCL INFO Channel 07/0 : 60[4] -> 7[7] [send] via NET/Libfabric/7(63)/GDRDMA
+ip-26-0-173-7:811252:811560 [4] NCCL INFO Channel 15/0 : 60[4] -> 7[7] [send] via NET/Libfabric/7(63)/GDRDMA
+ip-26-0-172-142:880686:880979 [7] NCCL INFO Channel 07/0 : 12[4] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880686:880979 [7] NCCL INFO Channel 15/0 : 12[4] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880683:880982 [4] NCCL INFO Channel 07/0 : 20[4] -> 31[7] [send] via NET/Libfabric/7(23)/GDRDMA
+ip-26-0-172-142:880683:880982 [4] NCCL INFO Channel 15/0 : 20[4] -> 31[7] [send] via NET/Libfabric/7(23)/GDRDMA
+ip-26-0-160-225:94349:94645 [7] NCCL INFO Channel 07/0 : 60[4] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:94349:94645 [7] NCCL INFO Channel 15/0 : 60[4] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019295:1019591 [7] NCCL INFO Channel 07/0 : 4[4] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019295:1019591 [7] NCCL INFO Channel 15/0 : 4[4] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019292:1019590 [4] NCCL INFO Channel 07/0 : 12[4] -> 23[7] [send] via NET/Libfabric/7(15)/GDRDMA
+ip-26-0-160-225:94349:94645 [7] NCCL INFO Channel 07/0 : 7[7] -> 3[3] via P2P/IPC
+ip-26-0-172-116:1019292:1019590 [4] NCCL INFO Channel 15/0 : 12[4] -> 23[7] [send] via NET/Libfabric/7(15)/GDRDMA
+ip-26-0-172-116:1019295:1019591 [7] NCCL INFO Channel 07/0 : 15[7] -> 11[3] via P2P/IPC
+ip-26-0-172-142:880686:880979 [7] NCCL INFO Channel 07/0 : 23[7] -> 19[3] via P2P/IPC
+ip-26-0-172-73:926601:926896 [7] NCCL INFO Channel 07/0 : 44[4] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926601:926896 [7] NCCL INFO Channel 15/0 : 44[4] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926598:926894 [4] NCCL INFO Channel 07/0 : 52[4] -> 63[7] [send] via NET/Libfabric/7(55)/GDRDMA
+ip-26-0-172-73:926598:926894 [4] NCCL INFO Channel 15/0 : 52[4] -> 63[7] [send] via NET/Libfabric/7(55)/GDRDMA
+ip-26-0-172-73:926601:926896 [7] NCCL INFO Channel 07/0 : 55[7] -> 51[3] via P2P/IPC
+ip-26-0-172-147:618591:618885 [7] NCCL INFO Channel 07/0 : 20[4] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-7:811255:811561 [7] NCCL INFO Channel 07/0 : 63[7] -> 59[3] via P2P/IPC
+ip-26-0-172-147:618591:618885 [7] NCCL INFO Channel 15/0 : 20[4] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618591:618885 [7] NCCL INFO Channel 07/0 : 31[7] -> 27[3] via P2P/IPC
+ip-26-0-160-225:94349:94645 [7] NCCL INFO Channel 15/0 : 7[7] -> 3[3] via P2P/IPC
+ip-26-0-172-142:880686:880979 [7] NCCL INFO Channel 15/0 : 23[7] -> 19[3] via P2P/IPC
+ip-26-0-173-7:811255:811561 [7] NCCL INFO Channel 15/0 : 63[7] -> 59[3] via P2P/IPC
+ip-26-0-172-147:618591:618885 [7] NCCL INFO Channel 15/0 : 31[7] -> 27[3] via P2P/IPC
+ip-26-0-172-57:942618:942918 [7] NCCL INFO Channel 07/0 : 36[4] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942618:942918 [7] NCCL INFO Channel 15/0 : 36[4] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942618:942918 [7] NCCL INFO Channel 07/0 : 47[7] -> 43[3] via P2P/IPC
+ip-26-0-172-252:688207:688504 [7] NCCL INFO Channel 07/0 : 28[4] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688207:688504 [7] NCCL INFO Channel 15/0 : 28[4] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:94349:94645 [7] NCCL INFO Channel 01/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-252:688207:688504 [7] NCCL INFO Channel 07/0 : 39[7] -> 35[3] via P2P/IPC
+ip-26-0-173-7:811255:811561 [7] NCCL INFO Channel 01/0 : 63[7] -> 62[6] via P2P/IPC
+ip-26-0-172-147:618591:618885 [7] NCCL INFO Channel 01/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-172-142:880686:880979 [7] NCCL INFO Channel 01/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-173-7:811255:811561 [7] NCCL INFO Channel 02/0 : 63[7] -> 62[6] via P2P/IPC
+ip-26-0-172-116:1019295:1019591 [7] NCCL INFO Channel 15/0 : 15[7] -> 11[3] via P2P/IPC
+ip-26-0-173-7:811255:811561 [7] NCCL INFO Channel 03/0 : 63[7] -> 62[6] via P2P/IPC
+ip-26-0-172-252:688207:688504 [7] NCCL INFO Channel 15/0 : 39[7] -> 35[3] via P2P/IPC
+ip-26-0-172-116:1019295:1019591 [7] NCCL INFO Channel 01/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-173-7:811255:811561 [7] NCCL INFO Channel 04/0 : 63[7] -> 62[6] via P2P/IPC
+ip-26-0-172-57:942618:942918 [7] NCCL INFO Channel 15/0 : 47[7] -> 43[3] via P2P/IPC
+ip-26-0-172-73:926601:926896 [7] NCCL INFO Channel 15/0 : 55[7] -> 51[3] via P2P/IPC
+ip-26-0-160-225:94349:94645 [7] NCCL INFO Channel 02/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-116:1019295:1019591 [7] NCCL INFO Channel 02/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-173-7:811255:811561 [7] NCCL INFO Channel 05/0 : 63[7] -> 62[6] via P2P/IPC
+ip-26-0-172-116:1019295:1019591 [7] NCCL INFO Channel 03/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-173-7:811255:811561 [7] NCCL INFO Channel 09/0 : 63[7] -> 62[6] via P2P/IPC
+ip-26-0-172-147:618591:618885 [7] NCCL INFO Channel 02/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-172-116:1019295:1019591 [7] NCCL INFO Channel 04/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-173-7:811255:811561 [7] NCCL INFO Channel 10/0 : 63[7] -> 62[6] via P2P/IPC
+ip-26-0-172-147:618591:618885 [7] NCCL INFO Channel 03/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-172-116:1019295:1019591 [7] NCCL INFO Channel 05/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-173-7:811255:811561 [7] NCCL INFO Channel 11/0 : 63[7] -> 62[6] via P2P/IPC
+ip-26-0-172-147:618591:618885 [7] NCCL INFO Channel 04/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-172-252:688207:688504 [7] NCCL INFO Channel 01/0 : 39[7] -> 38[6] via P2P/IPC
+ip-26-0-172-116:1019295:1019591 [7] NCCL INFO Channel 09/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-172-57:942618:942918 [7] NCCL INFO Channel 01/0 : 47[7] -> 46[6] via P2P/IPC
+ip-26-0-173-7:811255:811561 [7] NCCL INFO Channel 12/0 : 63[7] -> 62[6] via P2P/IPC
+ip-26-0-160-225:94349:94645 [7] NCCL INFO Channel 03/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-173-7:811252:811560 [4] NCCL INFO Channel 01/0 : 60[4] -> 59[3] via P2P/IPC
+ip-26-0-172-147:618591:618885 [7] NCCL INFO Channel 05/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-172-116:1019292:1019590 [4] NCCL INFO Channel 01/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-172-116:1019295:1019591 [7] NCCL INFO Channel 10/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-173-7:811255:811561 [7] NCCL INFO Channel 13/0 : 63[7] -> 62[6] via P2P/IPC
+ip-26-0-172-147:618591:618885 [7] NCCL INFO Channel 09/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-173-7:811252:811560 [4] NCCL INFO Channel 02/0 : 60[4] -> 59[3] via P2P/IPC
+ip-26-0-172-142:880686:880979 [7] NCCL INFO Channel 02/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-172-116:1019292:1019590 [4] NCCL INFO Channel 02/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-172-116:1019295:1019591 [7] NCCL INFO Channel 11/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-172-142:880686:880979 [7] NCCL INFO Channel 03/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-173-7:811252:811560 [4] NCCL INFO Channel 04/0 : 60[4] -> 59[3] via P2P/IPC
+ip-26-0-172-116:1019292:1019590 [4] NCCL INFO Channel 04/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-172-147:618591:618885 [7] NCCL INFO Channel 10/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-160-225:94349:94645 [7] NCCL INFO Channel 04/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-142:880686:880979 [7] NCCL INFO Channel 04/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-172-116:1019295:1019591 [7] NCCL INFO Channel 12/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-173-7:811252:811560 [4] NCCL INFO Channel 09/0 : 60[4] -> 59[3] via P2P/IPC
+ip-26-0-172-147:618591:618885 [7] NCCL INFO Channel 11/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-172-116:1019292:1019590 [4] NCCL INFO Channel 09/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-172-147:618591:618885 [7] NCCL INFO Channel 12/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-172-116:1019295:1019591 [7] NCCL INFO Channel 13/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-172-252:688207:688504 [7] NCCL INFO Channel 02/0 : 39[7] -> 38[6] via P2P/IPC
+ip-26-0-160-225:94349:94645 [7] NCCL INFO Channel 05/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-142:880686:880979 [7] NCCL INFO Channel 05/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-173-7:811251:811556 [3] NCCL INFO Channel 01/0 : 59[3] -> 58[2] via P2P/IPC
+ip-26-0-172-147:618591:618885 [7] NCCL INFO Channel 13/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-172-116:1019292:1019590 [4] NCCL INFO Channel 10/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-173-7:811252:811560 [4] NCCL INFO Channel 10/0 : 60[4] -> 59[3] via P2P/IPC
+ip-26-0-172-73:926601:926896 [7] NCCL INFO Channel 01/0 : 55[7] -> 54[6] via P2P/IPC
+ip-26-0-172-57:942618:942918 [7] NCCL INFO Channel 02/0 : 47[7] -> 46[6] via P2P/IPC
+ip-26-0-160-225:94349:94645 [7] NCCL INFO Channel 09/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-142:880686:880979 [7] NCCL INFO Channel 09/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-172-116:1019292:1019590 [4] NCCL INFO Channel 12/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-173-7:811252:811560 [4] NCCL INFO Channel 12/0 : 60[4] -> 59[3] via P2P/IPC
+ip-26-0-173-7:811251:811556 [3] NCCL INFO Channel 03/0 : 59[3] -> 58[2] via P2P/IPC
+ip-26-0-173-7:811251:811556 [3] NCCL INFO Channel 04/0 : 59[3] -> 58[2] via P2P/IPC
+ip-26-0-160-225:94349:94645 [7] NCCL INFO Channel 10/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-142:880686:880979 [7] NCCL INFO Channel 10/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-172-252:688207:688504 [7] NCCL INFO Channel 03/0 : 39[7] -> 38[6] via P2P/IPC
+ip-26-0-173-7:811251:811556 [3] NCCL INFO Channel 05/0 : 59[3] -> 58[2] via P2P/IPC
+ip-26-0-173-7:811254:811557 [6] NCCL INFO Channel 01/0 : 62[6] -> 61[5] via P2P/IPC
+ip-26-0-173-7:811253:811558 [5] NCCL INFO Channel 01/0 : 61[5] -> 60[4] via P2P/IPC
+ip-26-0-160-225:94349:94645 [7] NCCL INFO Channel 11/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-142:880686:880979 [7] NCCL INFO Channel 11/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-172-57:942618:942918 [7] NCCL INFO Channel 03/0 : 47[7] -> 46[6] via P2P/IPC
+ip-26-0-173-7:811251:811556 [3] NCCL INFO Channel 06/0 : 59[3] -> 58[2] via P2P/IPC
+ip-26-0-173-7:811254:811557 [6] NCCL INFO Channel 02/0 : 62[6] -> 61[5] via P2P/IPC
+ip-26-0-160-225:94349:94645 [7] NCCL INFO Channel 12/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-173-7:811253:811558 [5] NCCL INFO Channel 02/0 : 61[5] -> 60[4] via P2P/IPC
+ip-26-0-172-116:1019291:1019594 [3] NCCL INFO Channel 01/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-172-147:618588:618888 [4] NCCL INFO Channel 01/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-160-225:94349:94645 [7] NCCL INFO Channel 13/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-173-7:811251:811556 [3] NCCL INFO Channel 07/0 : 59[3] -> 58[2] via P2P/IPC
+ip-26-0-173-7:811254:811557 [6] NCCL INFO Channel 03/0 : 62[6] -> 61[5] via P2P/IPC
+ip-26-0-173-7:811253:811558 [5] NCCL INFO Channel 03/0 : 61[5] -> 60[4] via P2P/IPC
+ip-26-0-172-116:1019291:1019594 [3] NCCL INFO Channel 03/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-172-142:880686:880979 [7] NCCL INFO Channel 12/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-172-142:880683:880982 [4] NCCL INFO Channel 01/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-173-7:811251:811556 [3] NCCL INFO Channel 09/0 : 59[3] -> 58[2] via P2P/IPC
+ip-26-0-172-147:618588:618888 [4] NCCL INFO Channel 02/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-173-7:811254:811557 [6] NCCL INFO Channel 04/0 : 62[6] -> 61[5] via P2P/IPC
+ip-26-0-172-147:618587:618887 [3] NCCL INFO Channel 01/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-173-7:811253:811558 [5] NCCL INFO Channel 06/0 : 61[5] -> 60[4] via P2P/IPC
+ip-26-0-172-116:1019291:1019594 [3] NCCL INFO Channel 04/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-172-252:688207:688504 [7] NCCL INFO Channel 04/0 : 39[7] -> 38[6] via P2P/IPC
+ip-26-0-173-7:811251:811556 [3] NCCL INFO Channel 11/0 : 59[3] -> 58[2] via P2P/IPC
+ip-26-0-173-7:811254:811557 [6] NCCL INFO Channel 07/0 : 62[6] -> 61[5] via P2P/IPC
+ip-26-0-172-142:880686:880979 [7] NCCL INFO Channel 13/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-172-142:880683:880982 [4] NCCL INFO Channel 02/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-173-7:811253:811558 [5] NCCL INFO Channel 07/0 : 61[5] -> 60[4] via P2P/IPC
+ip-26-0-172-116:1019291:1019594 [3] NCCL INFO Channel 05/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-172-147:618588:618888 [4] NCCL INFO Channel 04/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-172-147:618587:618887 [3] NCCL INFO Channel 03/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-173-7:811251:811556 [3] NCCL INFO Channel 12/0 : 59[3] -> 58[2] via P2P/IPC
+ip-26-0-172-116:1019291:1019594 [3] NCCL INFO Channel 06/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-173-7:811254:811557 [6] NCCL INFO Channel 09/0 : 62[6] -> 61[5] via P2P/IPC
+ip-26-0-172-116:1019294:1019589 [6] NCCL INFO Channel 01/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-173-7:811253:811558 [5] NCCL INFO Channel 09/0 : 61[5] -> 60[4] via P2P/IPC
+ip-26-0-172-147:618588:618888 [4] NCCL INFO Channel 09/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-160-225:94346:94644 [4] NCCL INFO Channel 01/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-142:880683:880982 [4] NCCL INFO Channel 04/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-172-147:618587:618887 [3] NCCL INFO Channel 04/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-172-116:1019291:1019594 [3] NCCL INFO Channel 07/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-160-225:94345:94649 [3] NCCL INFO Channel 01/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-116:1019294:1019589 [6] NCCL INFO Channel 02/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-173-7:811251:811556 [3] NCCL INFO Channel 13/0 : 59[3] -> 58[2] via P2P/IPC
+ip-26-0-173-7:811254:811557 [6] NCCL INFO Channel 10/0 : 62[6] -> 61[5] via P2P/IPC
+ip-26-0-173-7:811253:811558 [5] NCCL INFO Channel 10/0 : 61[5] -> 60[4] via P2P/IPC
+ip-26-0-160-225:94346:94644 [4] NCCL INFO Channel 02/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-116:1019291:1019594 [3] NCCL INFO Channel 09/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-172-147:618590:618889 [6] NCCL INFO Channel 01/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-172-142:880683:880982 [4] NCCL INFO Channel 09/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-172-116:1019294:1019589 [6] NCCL INFO Channel 03/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-172-57:942618:942918 [7] NCCL INFO Channel 04/0 : 47[7] -> 46[6] via P2P/IPC
+ip-26-0-172-147:618588:618888 [4] NCCL INFO Channel 10/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-172-142:880682:880978 [3] NCCL INFO Channel 01/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-160-225:94348:94650 [6] NCCL INFO Channel 01/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-147:618587:618887 [3] NCCL INFO Channel 05/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-160-225:94345:94649 [3] NCCL INFO Channel 03/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-173-7:811251:811556 [3] NCCL INFO Channel 14/0 : 59[3] -> 58[2] via P2P/IPC
+ip-26-0-173-7:811254:811557 [6] NCCL INFO Channel 11/0 : 62[6] -> 61[5] via P2P/IPC
+ip-26-0-173-7:811253:811558 [5] NCCL INFO Channel 11/0 : 61[5] -> 60[4] via P2P/IPC
+ip-26-0-160-225:94346:94644 [4] NCCL INFO Channel 04/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-73:926601:926896 [7] NCCL INFO Channel 02/0 : 55[7] -> 54[6] via P2P/IPC
+ip-26-0-172-142:880683:880982 [4] NCCL INFO Channel 10/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-172-116:1019291:1019594 [3] NCCL INFO Channel 11/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-172-116:1019294:1019589 [6] NCCL INFO Channel 04/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-172-57:942615:942912 [4] NCCL INFO Channel 01/0 : 44[4] -> 43[3] via P2P/IPC
+ip-26-0-172-147:618588:618888 [4] NCCL INFO Channel 12/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-172-147:618590:618889 [6] NCCL INFO Channel 02/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-172-142:880683:880982 [4] NCCL INFO Channel 12/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-172-147:618587:618887 [3] NCCL INFO Channel 06/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-160-225:94348:94650 [6] NCCL INFO Channel 02/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-225:94345:94649 [3] NCCL INFO Channel 04/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-173-7:811254:811557 [6] NCCL INFO Channel 12/0 : 62[6] -> 61[5] via P2P/IPC
+ip-26-0-173-7:811251:811556 [3] NCCL INFO Channel 15/0 : 59[3] -> 58[2] via P2P/IPC
+ip-26-0-173-7:811253:811558 [5] NCCL INFO Channel 14/0 : 61[5] -> 60[4] via P2P/IPC
+ip-26-0-172-116:1019291:1019594 [3] NCCL INFO Channel 12/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-172-116:1019294:1019589 [6] NCCL INFO Channel 07/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-160-225:94346:94644 [4] NCCL INFO Channel 09/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-116:1019293:1019592 [5] NCCL INFO Channel 01/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-172-147:618589:618883 [5] NCCL INFO Channel 01/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-172-252:688207:688504 [7] NCCL INFO Channel 05/0 : 39[7] -> 38[6] via P2P/IPC
+ip-26-0-160-225:94348:94650 [6] NCCL INFO Channel 03/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-173-7:811254:811557 [6] NCCL INFO Channel 15/0 : 62[6] -> 61[5] via P2P/IPC
+ip-26-0-172-147:618590:618889 [6] NCCL INFO Channel 03/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-172-116:1019291:1019594 [3] NCCL INFO Channel 13/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-172-142:880685:880977 [6] NCCL INFO Channel 01/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-160-225:94345:94649 [3] NCCL INFO Channel 05/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-147:618587:618887 [3] NCCL INFO Channel 07/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-173-7:811253:811558 [5] NCCL INFO Channel 15/0 : 61[5] -> 60[4] via P2P/IPC
+ip-26-0-172-116:1019294:1019589 [6] NCCL INFO Channel 09/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-172-116:1019293:1019592 [5] NCCL INFO Channel 02/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-160-225:94346:94644 [4] NCCL INFO Channel 10/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-142:880682:880978 [3] NCCL INFO Channel 03/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-172-147:618589:618883 [5] NCCL INFO Channel 02/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-172-57:942618:942918 [7] NCCL INFO Channel 05/0 : 47[7] -> 46[6] via P2P/IPC
+ip-26-0-172-252:688207:688504 [7] NCCL INFO Channel 09/0 : 39[7] -> 38[6] via P2P/IPC
+ip-26-0-172-142:880685:880977 [6] NCCL INFO Channel 02/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-172-116:1019291:1019594 [3] NCCL INFO Channel 14/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-172-147:618590:618889 [6] NCCL INFO Channel 04/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-160-225:94348:94650 [6] NCCL INFO Channel 04/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-116:1019294:1019589 [6] NCCL INFO Channel 10/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-172-116:1019293:1019592 [5] NCCL INFO Channel 03/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-172-142:880682:880978 [3] NCCL INFO Channel 04/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-160-225:94345:94649 [3] NCCL INFO Channel 06/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-147:618590:618889 [6] NCCL INFO Channel 07/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-160-225:94346:94644 [4] NCCL INFO Channel 12/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-142:880685:880977 [6] NCCL INFO Channel 03/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-172-116:1019291:1019594 [3] NCCL INFO Channel 15/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-160-225:94348:94650 [6] NCCL INFO Channel 07/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-147:618590:618889 [6] NCCL INFO Channel 09/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-172-142:880682:880978 [3] NCCL INFO Channel 05/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-172-116:1019293:1019592 [5] NCCL INFO Channel 06/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-172-116:1019294:1019589 [6] NCCL INFO Channel 11/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-172-252:688207:688504 [7] NCCL INFO Channel 10/0 : 39[7] -> 38[6] via P2P/IPC
+ip-26-0-172-142:880685:880977 [6] NCCL INFO Channel 04/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-160-225:94345:94649 [3] NCCL INFO Channel 07/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-147:618590:618889 [6] NCCL INFO Channel 10/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-172-116:1019293:1019592 [5] NCCL INFO Channel 07/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-172-116:1019294:1019589 [6] NCCL INFO Channel 12/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-172-142:880682:880978 [3] NCCL INFO Channel 06/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-160-225:94347:94646 [5] NCCL INFO Channel 01/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-147:618587:618887 [3] NCCL INFO Channel 09/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-160-225:94348:94650 [6] NCCL INFO Channel 09/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-142:880685:880977 [6] NCCL INFO Channel 07/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-172-147:618590:618889 [6] NCCL INFO Channel 11/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-160-225:94345:94649 [3] NCCL INFO Channel 09/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-147:618589:618883 [5] NCCL INFO Channel 03/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-172-57:942615:942912 [4] NCCL INFO Channel 02/0 : 44[4] -> 43[3] via P2P/IPC
+ip-26-0-160-225:94348:94650 [6] NCCL INFO Channel 10/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-225:94347:94646 [5] NCCL INFO Channel 02/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-142:880682:880978 [3] NCCL INFO Channel 07/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-172-147:618587:618887 [3] NCCL INFO Channel 11/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-172-147:618590:618889 [6] NCCL INFO Channel 12/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-160-225:94345:94649 [3] NCCL INFO Channel 11/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-142:880684:880981 [5] NCCL INFO Channel 01/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-172-142:880685:880977 [6] NCCL INFO Channel 09/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-172-147:618589:618883 [5] NCCL INFO Channel 06/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-160-225:94348:94650 [6] NCCL INFO Channel 11/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-225:94345:94649 [3] NCCL INFO Channel 12/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-225:94347:94646 [5] NCCL INFO Channel 03/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-142:880682:880978 [3] NCCL INFO Channel 09/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-172-142:880684:880981 [5] NCCL INFO Channel 02/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-172-147:618587:618887 [3] NCCL INFO Channel 12/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-172-147:618590:618889 [6] NCCL INFO Channel 15/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-172-147:618589:618883 [5] NCCL INFO Channel 07/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-160-225:94348:94650 [6] NCCL INFO Channel 12/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-73:926601:926896 [7] NCCL INFO Channel 03/0 : 55[7] -> 54[6] via P2P/IPC
+ip-26-0-160-225:94345:94649 [3] NCCL INFO Channel 13/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-225:94347:94646 [5] NCCL INFO Channel 06/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-57:942618:942918 [7] NCCL INFO Channel 09/0 : 47[7] -> 46[6] via P2P/IPC
+ip-26-0-172-73:926601:926896 [7] NCCL INFO Channel 04/0 : 55[7] -> 54[6] via P2P/IPC
+ip-26-0-160-225:94348:94650 [6] NCCL INFO Channel 15/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-147:618587:618887 [3] NCCL INFO Channel 13/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-172-147:618589:618883 [5] NCCL INFO Channel 09/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-172-252:688207:688504 [7] NCCL INFO Channel 11/0 : 39[7] -> 38[6] via P2P/IPC
+ip-26-0-160-225:94345:94649 [3] NCCL INFO Channel 14/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-225:94347:94646 [5] NCCL INFO Channel 07/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-73:926601:926896 [7] NCCL INFO Channel 05/0 : 55[7] -> 54[6] via P2P/IPC
+ip-26-0-172-252:688204:688502 [4] NCCL INFO Channel 01/0 : 36[4] -> 35[3] via P2P/IPC
+ip-26-0-172-142:880682:880978 [3] NCCL INFO Channel 11/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-172-252:688207:688504 [7] NCCL INFO Channel 12/0 : 39[7] -> 38[6] via P2P/IPC
+ip-26-0-172-142:880684:880981 [5] NCCL INFO Channel 03/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-172-73:926601:926896 [7] NCCL INFO Channel 09/0 : 55[7] -> 54[6] via P2P/IPC
+ip-26-0-172-116:1019294:1019589 [6] NCCL INFO Channel 15/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-172-142:880682:880978 [3] NCCL INFO Channel 12/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-172-142:880684:880981 [5] NCCL INFO Channel 06/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-172-142:880685:880977 [6] NCCL INFO Channel 10/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-172-57:942615:942912 [4] NCCL INFO Channel 04/0 : 44[4] -> 43[3] via P2P/IPC
+ip-26-0-172-252:688204:688502 [4] NCCL INFO Channel 02/0 : 36[4] -> 35[3] via P2P/IPC
+ip-26-0-172-252:688207:688504 [7] NCCL INFO Channel 13/0 : 39[7] -> 38[6] via P2P/IPC
+ip-26-0-172-142:880682:880978 [3] NCCL INFO Channel 13/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-172-142:880684:880981 [5] NCCL INFO Channel 07/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-172-147:618587:618887 [3] NCCL INFO Channel 14/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-172-147:618589:618883 [5] NCCL INFO Channel 10/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-172-73:926601:926896 [7] NCCL INFO Channel 10/0 : 55[7] -> 54[6] via P2P/IPC
+ip-26-0-172-57:942618:942918 [7] NCCL INFO Channel 10/0 : 47[7] -> 46[6] via P2P/IPC
+ip-26-0-172-252:688203:688501 [3] NCCL INFO Channel 01/0 : 35[3] -> 34[2] via P2P/IPC
+ip-26-0-160-225:94347:94646 [5] NCCL INFO Channel 09/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-225:94345:94649 [3] NCCL INFO Channel 15/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-73:926601:926896 [7] NCCL INFO Channel 11/0 : 55[7] -> 54[6] via P2P/IPC
+ip-26-0-172-73:926601:926896 [7] NCCL INFO Channel 12/0 : 55[7] -> 54[6] via P2P/IPC
+ip-26-0-172-252:688204:688502 [4] NCCL INFO Channel 04/0 : 36[4] -> 35[3] via P2P/IPC
+ip-26-0-172-73:926598:926894 [4] NCCL INFO Channel 01/0 : 52[4] -> 51[3] via P2P/IPC
+ip-26-0-172-252:688204:688502 [4] NCCL INFO Channel 09/0 : 36[4] -> 35[3] via P2P/IPC
+ip-26-0-172-73:926601:926896 [7] NCCL INFO Channel 13/0 : 55[7] -> 54[6] via P2P/IPC
+ip-26-0-172-142:880685:880977 [6] NCCL INFO Channel 11/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-172-73:926598:926894 [4] NCCL INFO Channel 02/0 : 52[4] -> 51[3] via P2P/IPC
+ip-26-0-172-252:688204:688502 [4] NCCL INFO Channel 10/0 : 36[4] -> 35[3] via P2P/IPC
+ip-26-0-172-142:880682:880978 [3] NCCL INFO Channel 14/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-172-252:688204:688502 [4] NCCL INFO Channel 12/0 : 36[4] -> 35[3] via P2P/IPC
+ip-26-0-172-142:880685:880977 [6] NCCL INFO Channel 12/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-160-225:94347:94646 [5] NCCL INFO Channel 10/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-57:942615:942912 [4] NCCL INFO Channel 09/0 : 44[4] -> 43[3] via P2P/IPC
+ip-26-0-172-73:926598:926894 [4] NCCL INFO Channel 04/0 : 52[4] -> 51[3] via P2P/IPC
+ip-26-0-172-252:688203:688501 [3] NCCL INFO Channel 03/0 : 35[3] -> 34[2] via P2P/IPC
+ip-26-0-172-252:688206:688506 [6] NCCL INFO Channel 01/0 : 38[6] -> 37[5] via P2P/IPC
+ip-26-0-172-73:926598:926894 [4] NCCL INFO Channel 09/0 : 52[4] -> 51[3] via P2P/IPC
+ip-26-0-172-252:688203:688501 [3] NCCL INFO Channel 04/0 : 35[3] -> 34[2] via P2P/IPC
+ip-26-0-172-252:688206:688506 [6] NCCL INFO Channel 02/0 : 38[6] -> 37[5] via P2P/IPC
+ip-26-0-172-147:618589:618883 [5] NCCL INFO Channel 11/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-172-147:618587:618887 [3] NCCL INFO Channel 15/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-172-142:880684:880981 [5] NCCL INFO Channel 09/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-172-142:880682:880978 [3] NCCL INFO Channel 15/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-172-142:880685:880977 [6] NCCL INFO Channel 15/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-172-73:926598:926894 [4] NCCL INFO Channel 10/0 : 52[4] -> 51[3] via P2P/IPC
+ip-26-0-172-57:942618:942918 [7] NCCL INFO Channel 11/0 : 47[7] -> 46[6] via P2P/IPC
+ip-26-0-172-252:688203:688501 [3] NCCL INFO Channel 05/0 : 35[3] -> 34[2] via P2P/IPC
+ip-26-0-172-142:880684:880981 [5] NCCL INFO Channel 10/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-172-252:688206:688506 [6] NCCL INFO Channel 03/0 : 38[6] -> 37[5] via P2P/IPC
+ip-26-0-172-73:926598:926894 [4] NCCL INFO Channel 12/0 : 52[4] -> 51[3] via P2P/IPC
+ip-26-0-172-252:688203:688501 [3] NCCL INFO Channel 06/0 : 35[3] -> 34[2] via P2P/IPC
+ip-26-0-172-142:880684:880981 [5] NCCL INFO Channel 11/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-172-252:688206:688506 [6] NCCL INFO Channel 04/0 : 38[6] -> 37[5] via P2P/IPC
+ip-26-0-160-225:94347:94646 [5] NCCL INFO Channel 11/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-57:942615:942912 [4] NCCL INFO Channel 10/0 : 44[4] -> 43[3] via P2P/IPC
+ip-26-0-172-252:688203:688501 [3] NCCL INFO Channel 07/0 : 35[3] -> 34[2] via P2P/IPC
+ip-26-0-172-57:942618:942918 [7] NCCL INFO Channel 12/0 : 47[7] -> 46[6] via P2P/IPC
+ip-26-0-172-252:688206:688506 [6] NCCL INFO Channel 07/0 : 38[6] -> 37[5] via P2P/IPC
+ip-26-0-172-142:880684:880981 [5] NCCL INFO Channel 14/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-172-147:618589:618883 [5] NCCL INFO Channel 14/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-172-252:688203:688501 [3] NCCL INFO Channel 09/0 : 35[3] -> 34[2] via P2P/IPC
+ip-26-0-172-252:688206:688506 [6] NCCL INFO Channel 09/0 : 38[6] -> 37[5] via P2P/IPC
+ip-26-0-172-252:688203:688501 [3] NCCL INFO Channel 11/0 : 35[3] -> 34[2] via P2P/IPC
+ip-26-0-172-252:688205:688505 [5] NCCL INFO Channel 01/0 : 37[5] -> 36[4] via P2P/IPC
+ip-26-0-172-252:688206:688506 [6] NCCL INFO Channel 10/0 : 38[6] -> 37[5] via P2P/IPC
+ip-26-0-172-116:1019293:1019592 [5] NCCL INFO Channel 09/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-172-73:926597:926900 [3] NCCL INFO Channel 01/0 : 51[3] -> 50[2] via P2P/IPC
+ip-26-0-172-57:942615:942912 [4] NCCL INFO Channel 12/0 : 44[4] -> 43[3] via P2P/IPC
+ip-26-0-172-252:688203:688501 [3] NCCL INFO Channel 12/0 : 35[3] -> 34[2] via P2P/IPC
+ip-26-0-160-225:94347:94646 [5] NCCL INFO Channel 14/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-73:926600:926893 [6] NCCL INFO Channel 01/0 : 54[6] -> 53[5] via P2P/IPC
+ip-26-0-172-252:688205:688505 [5] NCCL INFO Channel 02/0 : 37[5] -> 36[4] via P2P/IPC
+ip-26-0-172-252:688206:688506 [6] NCCL INFO Channel 11/0 : 38[6] -> 37[5] via P2P/IPC
+ip-26-0-172-73:926597:926900 [3] NCCL INFO Channel 03/0 : 51[3] -> 50[2] via P2P/IPC
+ip-26-0-172-57:942618:942918 [7] NCCL INFO Channel 13/0 : 47[7] -> 46[6] via P2P/IPC
+ip-26-0-172-252:688203:688501 [3] NCCL INFO Channel 13/0 : 35[3] -> 34[2] via P2P/IPC
+ip-26-0-172-252:688205:688505 [5] NCCL INFO Channel 03/0 : 37[5] -> 36[4] via P2P/IPC
+ip-26-0-172-73:926600:926893 [6] NCCL INFO Channel 02/0 : 54[6] -> 53[5] via P2P/IPC
+ip-26-0-172-252:688206:688506 [6] NCCL INFO Channel 12/0 : 38[6] -> 37[5] via P2P/IPC
+ip-26-0-172-73:926597:926900 [3] NCCL INFO Channel 04/0 : 51[3] -> 50[2] via P2P/IPC
+ip-26-0-172-252:688203:688501 [3] NCCL INFO Channel 14/0 : 35[3] -> 34[2] via P2P/IPC
+ip-26-0-172-73:926600:926893 [6] NCCL INFO Channel 03/0 : 54[6] -> 53[5] via P2P/IPC
+ip-26-0-172-252:688205:688505 [5] NCCL INFO Channel 06/0 : 37[5] -> 36[4] via P2P/IPC
+ip-26-0-172-252:688206:688506 [6] NCCL INFO Channel 15/0 : 38[6] -> 37[5] via P2P/IPC
+ip-26-0-172-73:926600:926893 [6] NCCL INFO Channel 04/0 : 54[6] -> 53[5] via P2P/IPC
+ip-26-0-172-73:926600:926893 [6] NCCL INFO Channel 07/0 : 54[6] -> 53[5] via P2P/IPC
+ip-26-0-172-73:926597:926900 [3] NCCL INFO Channel 05/0 : 51[3] -> 50[2] via P2P/IPC
+ip-26-0-172-252:688203:688501 [3] NCCL INFO Channel 15/0 : 35[3] -> 34[2] via P2P/IPC
+ip-26-0-172-57:942617:942916 [6] NCCL INFO Channel 01/0 : 46[6] -> 45[5] via P2P/IPC
+ip-26-0-172-73:926600:926893 [6] NCCL INFO Channel 09/0 : 54[6] -> 53[5] via P2P/IPC
+ip-26-0-172-57:942617:942916 [6] NCCL INFO Channel 02/0 : 46[6] -> 45[5] via P2P/IPC
+ip-26-0-172-73:926597:926900 [3] NCCL INFO Channel 06/0 : 51[3] -> 50[2] via P2P/IPC
+ip-26-0-172-73:926599:926895 [5] NCCL INFO Channel 01/0 : 53[5] -> 52[4] via P2P/IPC
+ip-26-0-172-57:942617:942916 [6] NCCL INFO Channel 03/0 : 46[6] -> 45[5] via P2P/IPC
+ip-26-0-172-57:942617:942916 [6] NCCL INFO Channel 04/0 : 46[6] -> 45[5] via P2P/IPC
+ip-26-0-172-73:926600:926893 [6] NCCL INFO Channel 10/0 : 54[6] -> 53[5] via P2P/IPC
+ip-26-0-172-73:926597:926900 [3] NCCL INFO Channel 07/0 : 51[3] -> 50[2] via P2P/IPC
+ip-26-0-172-73:926599:926895 [5] NCCL INFO Channel 02/0 : 53[5] -> 52[4] via P2P/IPC
+ip-26-0-172-252:688205:688505 [5] NCCL INFO Channel 07/0 : 37[5] -> 36[4] via P2P/IPC
+ip-26-0-172-57:942614:942915 [3] NCCL INFO Channel 01/0 : 43[3] -> 42[2] via P2P/IPC
+ip-26-0-172-57:942617:942916 [6] NCCL INFO Channel 07/0 : 46[6] -> 45[5] via P2P/IPC
+ip-26-0-172-57:942614:942915 [3] NCCL INFO Channel 03/0 : 43[3] -> 42[2] via P2P/IPC
+ip-26-0-172-57:942617:942916 [6] NCCL INFO Channel 09/0 : 46[6] -> 45[5] via P2P/IPC
+ip-26-0-172-73:926600:926893 [6] NCCL INFO Channel 11/0 : 54[6] -> 53[5] via P2P/IPC
+ip-26-0-172-73:926597:926900 [3] NCCL INFO Channel 09/0 : 51[3] -> 50[2] via P2P/IPC
+ip-26-0-172-73:926599:926895 [5] NCCL INFO Channel 03/0 : 53[5] -> 52[4] via P2P/IPC
+ip-26-0-172-57:942614:942915 [3] NCCL INFO Channel 04/0 : 43[3] -> 42[2] via P2P/IPC
+ip-26-0-172-57:942617:942916 [6] NCCL INFO Channel 10/0 : 46[6] -> 45[5] via P2P/IPC
+ip-26-0-172-73:926600:926893 [6] NCCL INFO Channel 12/0 : 54[6] -> 53[5] via P2P/IPC
+ip-26-0-172-73:926597:926900 [3] NCCL INFO Channel 11/0 : 51[3] -> 50[2] via P2P/IPC
+ip-26-0-172-73:926599:926895 [5] NCCL INFO Channel 06/0 : 53[5] -> 52[4] via P2P/IPC
+ip-26-0-172-142:880684:880981 [5] NCCL INFO Channel 15/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-172-57:942614:942915 [3] NCCL INFO Channel 05/0 : 43[3] -> 42[2] via P2P/IPC
+ip-26-0-172-57:942617:942916 [6] NCCL INFO Channel 11/0 : 46[6] -> 45[5] via P2P/IPC
+ip-26-0-172-73:926600:926893 [6] NCCL INFO Channel 15/0 : 54[6] -> 53[5] via P2P/IPC
+ip-26-0-172-73:926597:926900 [3] NCCL INFO Channel 12/0 : 51[3] -> 50[2] via P2P/IPC
+ip-26-0-172-73:926599:926895 [5] NCCL INFO Channel 07/0 : 53[5] -> 52[4] via P2P/IPC
+ip-26-0-172-57:942614:942915 [3] NCCL INFO Channel 06/0 : 43[3] -> 42[2] via P2P/IPC
+ip-26-0-172-57:942617:942916 [6] NCCL INFO Channel 12/0 : 46[6] -> 45[5] via P2P/IPC
+ip-26-0-172-73:926597:926900 [3] NCCL INFO Channel 13/0 : 51[3] -> 50[2] via P2P/IPC
+ip-26-0-172-57:942617:942916 [6] NCCL INFO Channel 15/0 : 46[6] -> 45[5] via P2P/IPC
+ip-26-0-172-57:942614:942915 [3] NCCL INFO Channel 07/0 : 43[3] -> 42[2] via P2P/IPC
+ip-26-0-172-73:926599:926895 [5] NCCL INFO Channel 09/0 : 53[5] -> 52[4] via P2P/IPC
+ip-26-0-160-225:94347:94646 [5] NCCL INFO Channel 15/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-57:942614:942915 [3] NCCL INFO Channel 09/0 : 43[3] -> 42[2] via P2P/IPC
+ip-26-0-172-57:942616:942913 [5] NCCL INFO Channel 01/0 : 45[5] -> 44[4] via P2P/IPC
+ip-26-0-172-57:942614:942915 [3] NCCL INFO Channel 11/0 : 43[3] -> 42[2] via P2P/IPC
+ip-26-0-172-116:1019293:1019592 [5] NCCL INFO Channel 10/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-172-57:942614:942915 [3] NCCL INFO Channel 12/0 : 43[3] -> 42[2] via P2P/IPC
+ip-26-0-172-57:942616:942913 [5] NCCL INFO Channel 02/0 : 45[5] -> 44[4] via P2P/IPC
+ip-26-0-172-147:618589:618883 [5] NCCL INFO Channel 15/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-172-57:942614:942915 [3] NCCL INFO Channel 13/0 : 43[3] -> 42[2] via P2P/IPC
+ip-26-0-172-57:942616:942913 [5] NCCL INFO Channel 03/0 : 45[5] -> 44[4] via P2P/IPC
+ip-26-0-172-73:926597:926900 [3] NCCL INFO Channel 14/0 : 51[3] -> 50[2] via P2P/IPC
+ip-26-0-172-73:926599:926895 [5] NCCL INFO Channel 10/0 : 53[5] -> 52[4] via P2P/IPC
+ip-26-0-172-57:942614:942915 [3] NCCL INFO Channel 14/0 : 43[3] -> 42[2] via P2P/IPC
+ip-26-0-172-57:942616:942913 [5] NCCL INFO Channel 06/0 : 45[5] -> 44[4] via P2P/IPC
+ip-26-0-172-57:942614:942915 [3] NCCL INFO Channel 15/0 : 43[3] -> 42[2] via P2P/IPC
+ip-26-0-172-57:942616:942913 [5] NCCL INFO Channel 07/0 : 45[5] -> 44[4] via P2P/IPC
+ip-26-0-172-252:688205:688505 [5] NCCL INFO Channel 09/0 : 37[5] -> 36[4] via P2P/IPC
+ip-26-0-172-73:926597:926900 [3] NCCL INFO Channel 15/0 : 51[3] -> 50[2] via P2P/IPC
+ip-26-0-172-116:1019293:1019592 [5] NCCL INFO Channel 11/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-172-252:688205:688505 [5] NCCL INFO Channel 10/0 : 37[5] -> 36[4] via P2P/IPC
+ip-26-0-172-252:688205:688505 [5] NCCL INFO Channel 11/0 : 37[5] -> 36[4] via P2P/IPC
+ip-26-0-172-73:926599:926895 [5] NCCL INFO Channel 11/0 : 53[5] -> 52[4] via P2P/IPC
+ip-26-0-172-57:942616:942913 [5] NCCL INFO Channel 09/0 : 45[5] -> 44[4] via P2P/IPC
+ip-26-0-172-252:688205:688505 [5] NCCL INFO Channel 14/0 : 37[5] -> 36[4] via P2P/IPC
+ip-26-0-172-57:942616:942913 [5] NCCL INFO Channel 10/0 : 45[5] -> 44[4] via P2P/IPC
+ip-26-0-172-73:926599:926895 [5] NCCL INFO Channel 14/0 : 53[5] -> 52[4] via P2P/IPC
+ip-26-0-172-252:688205:688505 [5] NCCL INFO Channel 15/0 : 37[5] -> 36[4] via P2P/IPC
+ip-26-0-172-116:1019293:1019592 [5] NCCL INFO Channel 14/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-172-116:1019293:1019592 [5] NCCL INFO Channel 15/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-172-73:926599:926895 [5] NCCL INFO Channel 15/0 : 53[5] -> 52[4] via P2P/IPC
+ip-26-0-172-57:942616:942913 [5] NCCL INFO Channel 11/0 : 45[5] -> 44[4] via P2P/IPC
+ip-26-0-172-57:942616:942913 [5] NCCL INFO Channel 14/0 : 45[5] -> 44[4] via P2P/IPC
+ip-26-0-172-57:942616:942913 [5] NCCL INFO Channel 15/0 : 45[5] -> 44[4] via P2P/IPC
+ip-26-0-160-225:94342:94643 [0] NCCL INFO Connected all rings
+ip-26-0-160-225:94342:94643 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-57:942611:942911 [0] NCCL INFO Connected all rings
+ip-26-0-172-57:942611:942911 [0] NCCL INFO Channel 02/0 : 40[0] -> 41[1] via P2P/IPC
+ip-26-0-172-57:942611:942911 [0] NCCL INFO Channel 03/0 : 40[0] -> 41[1] via P2P/IPC
+ip-26-0-172-57:942611:942911 [0] NCCL INFO Channel 04/0 : 40[0] -> 41[1] via P2P/IPC
+ip-26-0-172-147:618584:618882 [0] NCCL INFO Connected all rings
+ip-26-0-172-147:618584:618882 [0] NCCL INFO Channel 02/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-172-142:880680:880984 [1] NCCL INFO Connected all rings
+ip-26-0-172-57:942611:942911 [0] NCCL INFO Channel 05/0 : 40[0] -> 41[1] via P2P/IPC
+ip-26-0-160-225:94343:94647 [1] NCCL INFO Connected all rings
+ip-26-0-160-225:94347:94646 [5] NCCL INFO Connected all rings
+ip-26-0-160-225:94345:94649 [3] NCCL INFO Connected all rings
+ip-26-0-172-142:880683:880982 [4] NCCL INFO Connected all rings
+ip-26-0-173-7:811250:811555 [2] NCCL INFO Connected all rings
+ip-26-0-172-73:926595:926897 [1] NCCL INFO Connected all rings
+ip-26-0-172-147:618587:618887 [3] NCCL INFO Connected all rings
+ip-26-0-172-252:688205:688505 [5] NCCL INFO Connected all rings
+ip-26-0-172-57:942615:942912 [4] NCCL INFO Connected all rings
+ip-26-0-173-7:811248:811554 [0] NCCL INFO Connected all rings
+ip-26-0-173-7:811248:811554 [0] NCCL INFO Channel 02/0 : 56[0] -> 57[1] via P2P/IPC
+ip-26-0-173-7:811248:811554 [0] NCCL INFO Channel 03/0 : 56[0] -> 57[1] via P2P/IPC
+ip-26-0-172-252:688202:688503 [2] NCCL INFO Connected all rings
+ip-26-0-173-7:811248:811554 [0] NCCL INFO Channel 04/0 : 56[0] -> 57[1] via P2P/IPC
+ip-26-0-172-73:926597:926900 [3] NCCL INFO Connected all rings
+ip-26-0-173-7:811248:811554 [0] NCCL INFO Channel 05/0 : 56[0] -> 57[1] via P2P/IPC
+ip-26-0-173-7:811248:811554 [0] NCCL INFO Channel 06/0 : 56[0] -> 57[1] via P2P/IPC
+ip-26-0-173-7:811248:811554 [0] NCCL INFO Channel 07/0 : 56[0] -> 57[1] via P2P/IPC
+ip-26-0-173-7:811248:811554 [0] NCCL INFO Channel 10/0 : 56[0] -> 57[1] via P2P/IPC
+ip-26-0-160-225:94342:94643 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-173-7:811248:811554 [0] NCCL INFO Channel 11/0 : 56[0] -> 57[1] via P2P/IPC
+ip-26-0-173-7:811248:811554 [0] NCCL INFO Channel 12/0 : 56[0] -> 57[1] via P2P/IPC
+ip-26-0-173-7:811248:811554 [0] NCCL INFO Channel 13/0 : 56[0] -> 57[1] via P2P/IPC
+ip-26-0-160-225:94342:94643 [0] NCCL INFO Channel 04/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-225:94343:94647 [1] NCCL INFO Channel 01/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-225:94342:94643 [0] NCCL INFO Channel 05/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-225:94343:94647 [1] NCCL INFO Channel 03/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-142:880679:880980 [0] NCCL INFO Connected all rings
+ip-26-0-172-142:880679:880980 [0] NCCL INFO Channel 02/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-160-225:94342:94643 [0] NCCL INFO Channel 06/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-225:94343:94647 [1] NCCL INFO Channel 04/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-225:94342:94643 [0] NCCL INFO Channel 07/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-173-7:811248:811554 [0] NCCL INFO Channel 14/0 : 56[0] -> 57[1] via P2P/IPC
+ip-26-0-160-225:94343:94647 [1] NCCL INFO Channel 05/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-225:94342:94643 [0] NCCL INFO Channel 10/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-225:94343:94647 [1] NCCL INFO Channel 06/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-173-7:811248:811554 [0] NCCL INFO Channel 15/0 : 56[0] -> 57[1] via P2P/IPC
+ip-26-0-160-225:94342:94643 [0] NCCL INFO Channel 11/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-142:880679:880980 [0] NCCL INFO Channel 03/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-160-225:94343:94647 [1] NCCL INFO Channel 07/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-252:688200:688507 [0] NCCL INFO Connected all rings
+ip-26-0-172-252:688200:688507 [0] NCCL INFO Channel 02/0 : 32[0] -> 33[1] via P2P/IPC
+ip-26-0-160-225:94342:94643 [0] NCCL INFO Channel 12/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-142:880679:880980 [0] NCCL INFO Channel 04/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-172-142:880679:880980 [0] NCCL INFO Channel 05/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-173-7:811254:811557 [6] NCCL INFO Connected all rings
+ip-26-0-172-252:688200:688507 [0] NCCL INFO Channel 03/0 : 32[0] -> 33[1] via P2P/IPC
+ip-26-0-160-225:94343:94647 [1] NCCL INFO Channel 09/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-225:94342:94643 [0] NCCL INFO Channel 13/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-142:880679:880980 [0] NCCL INFO Channel 06/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-160-225:94343:94647 [1] NCCL INFO Channel 11/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-252:688207:688504 [7] NCCL INFO Connected all rings
+ip-26-0-173-7:811253:811558 [5] NCCL INFO Connected all rings
+ip-26-0-160-225:94342:94643 [0] NCCL INFO Channel 14/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-142:880679:880980 [0] NCCL INFO Channel 07/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-172-57:942611:942911 [0] NCCL INFO Channel 06/0 : 40[0] -> 41[1] via P2P/IPC
+ip-26-0-172-252:688200:688507 [0] NCCL INFO Channel 04/0 : 32[0] -> 33[1] via P2P/IPC
+ip-26-0-160-225:94343:94647 [1] NCCL INFO Channel 12/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-173-7:811249:811559 [1] NCCL INFO Connected all rings
+ip-26-0-160-225:94342:94643 [0] NCCL INFO Channel 15/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-142:880679:880980 [0] NCCL INFO Channel 10/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-160-225:94343:94647 [1] NCCL INFO Channel 13/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-252:688200:688507 [0] NCCL INFO Channel 05/0 : 32[0] -> 33[1] via P2P/IPC
+ip-26-0-172-142:880679:880980 [0] NCCL INFO Channel 11/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-160-225:94343:94647 [1] NCCL INFO Channel 14/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-73:926601:926896 [7] NCCL INFO Connected all rings
+ip-26-0-172-252:688200:688507 [0] NCCL INFO Channel 06/0 : 32[0] -> 33[1] via P2P/IPC
+ip-26-0-172-57:942611:942911 [0] NCCL INFO Channel 07/0 : 40[0] -> 41[1] via P2P/IPC
+ip-26-0-173-7:811250:811555 [2] NCCL INFO Channel 01/0 : 58[2] -> 59[3] via P2P/IPC
+ip-26-0-172-252:688200:688507 [0] NCCL INFO Channel 07/0 : 32[0] -> 33[1] via P2P/IPC
+ip-26-0-172-57:942612:942914 [1] NCCL INFO Connected all rings
+ip-26-0-172-147:618584:618882 [0] NCCL INFO Channel 03/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-172-57:942611:942911 [0] NCCL INFO Channel 10/0 : 40[0] -> 41[1] via P2P/IPC
+ip-26-0-160-225:94348:94650 [6] NCCL INFO Connected all rings
+ip-26-0-173-7:811250:811555 [2] NCCL INFO Channel 02/0 : 58[2] -> 59[3] via P2P/IPC
+ip-26-0-160-225:94343:94647 [1] NCCL INFO Channel 15/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-252:688200:688507 [0] NCCL INFO Channel 10/0 : 32[0] -> 33[1] via P2P/IPC
+ip-26-0-173-7:811250:811555 [2] NCCL INFO Channel 04/0 : 58[2] -> 59[3] via P2P/IPC
+ip-26-0-172-252:688200:688507 [0] NCCL INFO Channel 11/0 : 32[0] -> 33[1] via P2P/IPC
+ip-26-0-160-225:94342:94643 [0] NCCL INFO Channel 05/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-172-142:880681:880983 [2] NCCL INFO Connected all rings
+ip-26-0-172-147:618585:618884 [1] NCCL INFO Connected all rings
+ip-26-0-172-252:688200:688507 [0] NCCL INFO Channel 12/0 : 32[0] -> 33[1] via P2P/IPC
+ip-26-0-173-7:811250:811555 [2] NCCL INFO Channel 05/0 : 58[2] -> 59[3] via P2P/IPC
+ip-26-0-160-225:94342:94643 [0] NCCL INFO Channel 06/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-172-57:942611:942911 [0] NCCL INFO Channel 11/0 : 40[0] -> 41[1] via P2P/IPC
+ip-26-0-172-142:880679:880980 [0] NCCL INFO Channel 12/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-172-252:688201:688500 [1] NCCL INFO Connected all rings
+ip-26-0-172-57:942611:942911 [0] NCCL INFO Channel 12/0 : 40[0] -> 41[1] via P2P/IPC
+ip-26-0-173-7:811250:811555 [2] NCCL INFO Channel 06/0 : 58[2] -> 59[3] via P2P/IPC
+ip-26-0-172-252:688200:688507 [0] NCCL INFO Channel 13/0 : 32[0] -> 33[1] via P2P/IPC
+ip-26-0-173-7:811252:811560 [4] NCCL INFO Connected all rings
+ip-26-0-172-142:880679:880980 [0] NCCL INFO Channel 13/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-160-225:94349:94645 [7] NCCL INFO Connected all rings
+ip-26-0-173-7:811250:811555 [2] NCCL INFO Channel 07/0 : 58[2] -> 59[3] via P2P/IPC
+ip-26-0-173-7:811255:811561 [7] NCCL INFO Connected all rings
+ip-26-0-172-142:880679:880980 [0] NCCL INFO Channel 14/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-160-225:94342:94643 [0] NCCL INFO Channel 07/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-172-147:618584:618882 [0] NCCL INFO Channel 04/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-172-252:688200:688507 [0] NCCL INFO Channel 14/0 : 32[0] -> 33[1] via P2P/IPC
+ip-26-0-173-7:811250:811555 [2] NCCL INFO Channel 09/0 : 58[2] -> 59[3] via P2P/IPC
+ip-26-0-160-225:94342:94643 [0] NCCL INFO Channel 13/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-160-225:94344:94648 [2] NCCL INFO Connected all rings
+ip-26-0-172-147:618584:618882 [0] NCCL INFO Channel 05/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-160-225:94342:94643 [0] NCCL INFO Channel 14/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-172-142:880679:880980 [0] NCCL INFO Channel 15/0 : 16[0] -> 17[1] via P2P/IPC
+ip-26-0-173-7:811250:811555 [2] NCCL INFO Channel 10/0 : 58[2] -> 59[3] via P2P/IPC
+ip-26-0-173-7:811254:811557 [6] NCCL INFO Channel 01/0 : 62[6] -> 63[7] via P2P/IPC
+ip-26-0-160-225:94342:94643 [0] NCCL INFO Channel 15/0 : 0[0] -> 7[7] via P2P/IPC
+ip-26-0-173-7:811250:811555 [2] NCCL INFO Channel 12/0 : 58[2] -> 59[3] via P2P/IPC
+ip-26-0-173-7:811254:811557 [6] NCCL INFO Channel 02/0 : 62[6] -> 63[7] via P2P/IPC
+ip-26-0-173-7:811250:811555 [2] NCCL INFO Channel 13/0 : 58[2] -> 59[3] via P2P/IPC
+ip-26-0-172-252:688200:688507 [0] NCCL INFO Channel 15/0 : 32[0] -> 33[1] via P2P/IPC
+ip-26-0-173-7:811254:811557 [6] NCCL INFO Channel 03/0 : 62[6] -> 63[7] via P2P/IPC
+ip-26-0-160-225:94347:94646 [5] NCCL INFO Channel 01/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-173-7:811253:811558 [5] NCCL INFO Channel 01/0 : 61[5] -> 62[6] via P2P/IPC
+ip-26-0-173-7:811250:811555 [2] NCCL INFO Channel 14/0 : 58[2] -> 59[3] via P2P/IPC
+ip-26-0-160-225:94347:94646 [5] NCCL INFO Channel 02/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-173-7:811249:811559 [1] NCCL INFO Channel 01/0 : 57[1] -> 58[2] via P2P/IPC
+ip-26-0-173-7:811254:811557 [6] NCCL INFO Channel 04/0 : 62[6] -> 63[7] via P2P/IPC
+ip-26-0-173-7:811253:811558 [5] NCCL INFO Channel 02/0 : 61[5] -> 62[6] via P2P/IPC
+ip-26-0-160-225:94347:94646 [5] NCCL INFO Channel 03/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-173-7:811250:811555 [2] NCCL INFO Channel 15/0 : 58[2] -> 59[3] via P2P/IPC
+ip-26-0-172-116:1019288:1019588 [0] NCCL INFO Connected all rings
+ip-26-0-172-116:1019288:1019588 [0] NCCL INFO Channel 02/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-173-7:811249:811559 [1] NCCL INFO Channel 03/0 : 57[1] -> 58[2] via P2P/IPC
+ip-26-0-173-7:811254:811557 [6] NCCL INFO Channel 05/0 : 62[6] -> 63[7] via P2P/IPC
+ip-26-0-160-225:94347:94646 [5] NCCL INFO Channel 04/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-173-7:811253:811558 [5] NCCL INFO Channel 03/0 : 61[5] -> 62[6] via P2P/IPC
+ip-26-0-160-225:94348:94650 [6] NCCL INFO Channel 01/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-147:618584:618882 [0] NCCL INFO Channel 06/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-172-116:1019289:1019593 [1] NCCL INFO Connected all rings
+ip-26-0-172-116:1019288:1019588 [0] NCCL INFO Channel 03/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-172-142:880682:880978 [3] NCCL INFO Connected all rings
+ip-26-0-173-7:811254:811557 [6] NCCL INFO Channel 06/0 : 62[6] -> 63[7] via P2P/IPC
+ip-26-0-173-7:811249:811559 [1] NCCL INFO Channel 04/0 : 57[1] -> 58[2] via P2P/IPC
+ip-26-0-173-7:811253:811558 [5] NCCL INFO Channel 04/0 : 61[5] -> 62[6] via P2P/IPC
+ip-26-0-172-116:1019290:1019595 [2] NCCL INFO Connected all rings
+ip-26-0-172-116:1019294:1019589 [6] NCCL INFO Connected all rings
+ip-26-0-172-116:1019288:1019588 [0] NCCL INFO Channel 04/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-173-7:811254:811557 [6] NCCL INFO Channel 09/0 : 62[6] -> 63[7] via P2P/IPC
+ip-26-0-172-147:618584:618882 [0] NCCL INFO Channel 07/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-173-7:811251:811556 [3] NCCL INFO Connected all rings
+ip-26-0-173-7:811249:811559 [1] NCCL INFO Channel 05/0 : 57[1] -> 58[2] via P2P/IPC
+ip-26-0-173-7:811253:811558 [5] NCCL INFO Channel 05/0 : 61[5] -> 62[6] via P2P/IPC
+ip-26-0-172-252:688203:688501 [3] NCCL INFO Connected all rings
+ip-26-0-172-142:880680:880984 [1] NCCL INFO Channel 01/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-172-73:926594:926898 [0] NCCL INFO Connected all rings
+ip-26-0-172-73:926594:926898 [0] NCCL INFO Channel 02/0 : 48[0] -> 49[1] via P2P/IPC
+ip-26-0-173-7:811254:811557 [6] NCCL INFO Channel 10/0 : 62[6] -> 63[7] via P2P/IPC
+ip-26-0-172-116:1019288:1019588 [0] NCCL INFO Channel 05/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-172-147:618584:618882 [0] NCCL INFO Channel 10/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-173-7:811249:811559 [1] NCCL INFO Channel 06/0 : 57[1] -> 58[2] via P2P/IPC
+ip-26-0-160-225:94347:94646 [5] NCCL INFO Channel 05/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-173-7:811253:811558 [5] NCCL INFO Channel 07/0 : 61[5] -> 62[6] via P2P/IPC
+ip-26-0-160-225:94348:94650 [6] NCCL INFO Channel 02/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-173-7:811254:811557 [6] NCCL INFO Channel 11/0 : 62[6] -> 63[7] via P2P/IPC
+ip-26-0-160-225:94347:94646 [5] NCCL INFO Channel 07/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-116:1019288:1019588 [0] NCCL INFO Channel 06/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-173-7:811249:811559 [1] NCCL INFO Channel 07/0 : 57[1] -> 58[2] via P2P/IPC
+ip-26-0-173-7:811253:811558 [5] NCCL INFO Channel 09/0 : 61[5] -> 62[6] via P2P/IPC
+ip-26-0-160-225:94348:94650 [6] NCCL INFO Channel 03/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-225:94347:94646 [5] NCCL INFO Channel 09/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-173-7:811254:811557 [6] NCCL INFO Channel 12/0 : 62[6] -> 63[7] via P2P/IPC
+ip-26-0-172-73:926594:926898 [0] NCCL INFO Channel 03/0 : 48[0] -> 49[1] via P2P/IPC
+ip-26-0-172-147:618584:618882 [0] NCCL INFO Channel 11/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-172-142:880680:880984 [1] NCCL INFO Channel 03/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-160-225:94348:94650 [6] NCCL INFO Channel 04/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-116:1019288:1019588 [0] NCCL INFO Channel 07/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-172-73:926594:926898 [0] NCCL INFO Channel 04/0 : 48[0] -> 49[1] via P2P/IPC
+ip-26-0-172-147:618585:618884 [1] NCCL INFO Channel 01/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-173-7:811249:811559 [1] NCCL INFO Channel 09/0 : 57[1] -> 58[2] via P2P/IPC
+ip-26-0-173-7:811253:811558 [5] NCCL INFO Channel 10/0 : 61[5] -> 62[6] via P2P/IPC
+ip-26-0-160-225:94347:94646 [5] NCCL INFO Channel 10/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-173-7:811252:811560 [4] NCCL INFO Channel 01/0 : 60[4] -> 61[5] via P2P/IPC
+ip-26-0-160-225:94345:94649 [3] NCCL INFO Channel 01/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-147:618584:618882 [0] NCCL INFO Channel 12/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-172-57:942611:942911 [0] NCCL INFO Channel 13/0 : 40[0] -> 41[1] via P2P/IPC
+ip-26-0-160-225:94344:94648 [2] NCCL INFO Channel 01/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-173-7:811254:811557 [6] NCCL INFO Channel 13/0 : 62[6] -> 63[7] via P2P/IPC
+ip-26-0-172-147:618586:618886 [2] NCCL INFO Connected all rings
+ip-26-0-160-225:94348:94650 [6] NCCL INFO Channel 05/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-147:618585:618884 [1] NCCL INFO Channel 03/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-172-142:880680:880984 [1] NCCL INFO Channel 04/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-160-225:94347:94646 [5] NCCL INFO Channel 11/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-225:94345:94649 [3] NCCL INFO Channel 02/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-173-7:811249:811559 [1] NCCL INFO Channel 11/0 : 57[1] -> 58[2] via P2P/IPC
+ip-26-0-173-7:811253:811558 [5] NCCL INFO Channel 11/0 : 61[5] -> 62[6] via P2P/IPC
+ip-26-0-172-147:618584:618882 [0] NCCL INFO Channel 13/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-172-57:942611:942911 [0] NCCL INFO Channel 14/0 : 40[0] -> 41[1] via P2P/IPC
+ip-26-0-173-7:811252:811560 [4] NCCL INFO Channel 02/0 : 60[4] -> 61[5] via P2P/IPC
+ip-26-0-160-225:94344:94648 [2] NCCL INFO Channel 02/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-116:1019288:1019588 [0] NCCL INFO Channel 10/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-173-7:811254:811557 [6] NCCL INFO Channel 14/0 : 62[6] -> 63[7] via P2P/IPC
+ip-26-0-160-225:94348:94650 [6] NCCL INFO Channel 06/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-147:618585:618884 [1] NCCL INFO Channel 04/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-160-225:94347:94646 [5] NCCL INFO Channel 12/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-225:94345:94649 [3] NCCL INFO Channel 03/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-173-7:811249:811559 [1] NCCL INFO Channel 12/0 : 57[1] -> 58[2] via P2P/IPC
+ip-26-0-173-7:811253:811558 [5] NCCL INFO Channel 12/0 : 61[5] -> 62[6] via P2P/IPC
+ip-26-0-173-7:811252:811560 [4] NCCL INFO Channel 03/0 : 60[4] -> 61[5] via P2P/IPC
+ip-26-0-172-147:618584:618882 [0] NCCL INFO Channel 14/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-172-142:880684:880981 [5] NCCL INFO Connected all rings
+ip-26-0-172-73:926594:926898 [0] NCCL INFO Channel 05/0 : 48[0] -> 49[1] via P2P/IPC
+ip-26-0-172-116:1019288:1019588 [0] NCCL INFO Channel 11/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-172-57:942611:942911 [0] NCCL INFO Channel 15/0 : 40[0] -> 41[1] via P2P/IPC
+ip-26-0-173-7:811249:811559 [1] NCCL INFO Channel 13/0 : 57[1] -> 58[2] via P2P/IPC
+ip-26-0-173-7:811253:811558 [5] NCCL INFO Channel 13/0 : 61[5] -> 62[6] via P2P/IPC
+ip-26-0-172-142:880680:880984 [1] NCCL INFO Channel 05/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-173-7:811252:811560 [4] NCCL INFO Channel 04/0 : 60[4] -> 61[5] via P2P/IPC
+ip-26-0-172-73:926594:926898 [0] NCCL INFO Channel 06/0 : 48[0] -> 49[1] via P2P/IPC
+ip-26-0-160-225:94348:94650 [6] NCCL INFO Channel 09/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-225:94344:94648 [2] NCCL INFO Channel 04/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-116:1019288:1019588 [0] NCCL INFO Channel 12/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-173-7:811249:811559 [1] NCCL INFO Channel 14/0 : 57[1] -> 58[2] via P2P/IPC
+ip-26-0-173-7:811253:811558 [5] NCCL INFO Channel 15/0 : 61[5] -> 62[6] via P2P/IPC
+ip-26-0-173-7:811252:811560 [4] NCCL INFO Channel 06/0 : 60[4] -> 61[5] via P2P/IPC
+ip-26-0-172-142:880680:880984 [1] NCCL INFO Channel 06/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-172-252:688201:688500 [1] NCCL INFO Channel 01/0 : 33[1] -> 34[2] via P2P/IPC
+ip-26-0-172-147:618585:618884 [1] NCCL INFO Channel 05/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-173-7:811249:811559 [1] NCCL INFO Channel 15/0 : 57[1] -> 58[2] via P2P/IPC
+ip-26-0-172-116:1019288:1019588 [0] NCCL INFO Channel 13/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-160-225:94347:94646 [5] NCCL INFO Channel 13/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-173-7:811252:811560 [4] NCCL INFO Channel 07/0 : 60[4] -> 61[5] via P2P/IPC
+ip-26-0-173-7:811254:811557 [6] NCCL INFO Channel 06/0 : 54[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688201:688500 [1] NCCL INFO Channel 03/0 : 33[1] -> 34[2] via P2P/IPC
+ip-26-0-172-142:880680:880984 [1] NCCL INFO Channel 07/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-172-73:926594:926898 [0] NCCL INFO Channel 07/0 : 48[0] -> 49[1] via P2P/IPC
+ip-26-0-160-225:94347:94646 [5] NCCL INFO Channel 15/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-173-7:811252:811560 [4] NCCL INFO Channel 09/0 : 60[4] -> 61[5] via P2P/IPC
+ip-26-0-173-7:811248:811554 [0] NCCL INFO Channel 05/0 : 56[0] -> 63[7] via P2P/IPC
+ip-26-0-172-73:926594:926898 [0] NCCL INFO Channel 10/0 : 48[0] -> 49[1] via P2P/IPC
+ip-26-0-172-116:1019288:1019588 [0] NCCL INFO Channel 14/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-173-7:811249:811559 [1] NCCL INFO Channel 01/0 : 49[1] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688201:688500 [1] NCCL INFO Channel 04/0 : 33[1] -> 34[2] via P2P/IPC
+ip-26-0-172-116:1019292:1019590 [4] NCCL INFO Connected all rings
+ip-26-0-173-7:811252:811560 [4] NCCL INFO Channel 10/0 : 60[4] -> 61[5] via P2P/IPC
+ip-26-0-172-57:942612:942914 [1] NCCL INFO Channel 01/0 : 41[1] -> 42[2] via P2P/IPC
+ip-26-0-173-7:811248:811554 [0] NCCL INFO Channel 06/0 : 56[0] -> 63[7] via P2P/IPC
+ip-26-0-172-57:942612:942914 [1] NCCL INFO Channel 03/0 : 41[1] -> 42[2] via P2P/IPC
+ip-26-0-172-142:880685:880977 [6] NCCL INFO Connected all rings
+ip-26-0-172-57:942614:942915 [3] NCCL INFO Connected all rings
+ip-26-0-160-225:94345:94649 [3] NCCL INFO Channel 05/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-173-7:811252:811560 [4] NCCL INFO Channel 11/0 : 60[4] -> 61[5] via P2P/IPC
+ip-26-0-172-73:926594:926898 [0] NCCL INFO Channel 11/0 : 48[0] -> 49[1] via P2P/IPC
+ip-26-0-172-252:688201:688500 [1] NCCL INFO Channel 05/0 : 33[1] -> 34[2] via P2P/IPC
+ip-26-0-173-7:811248:811554 [0] NCCL INFO Channel 07/0 : 56[0] -> 63[7] via P2P/IPC
+ip-26-0-172-116:1019288:1019588 [0] NCCL INFO Channel 15/0 : 8[0] -> 9[1] via P2P/IPC
+ip-26-0-172-142:880686:880979 [7] NCCL INFO Connected all rings
+ip-26-0-173-7:811252:811560 [4] NCCL INFO Channel 12/0 : 60[4] -> 61[5] via P2P/IPC
+ip-26-0-173-7:811251:811556 [3] NCCL INFO Channel 01/0 : 59[3] -> 60[4] via P2P/IPC
+ip-26-0-172-142:880680:880984 [1] NCCL INFO Channel 09/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-172-252:688201:688500 [1] NCCL INFO Channel 06/0 : 33[1] -> 34[2] via P2P/IPC
+ip-26-0-172-142:880681:880983 [2] NCCL INFO Channel 01/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-172-73:926594:926898 [0] NCCL INFO Channel 12/0 : 48[0] -> 49[1] via P2P/IPC
+ip-26-0-172-57:942612:942914 [1] NCCL INFO Channel 04/0 : 41[1] -> 42[2] via P2P/IPC
+ip-26-0-173-7:811252:811560 [4] NCCL INFO Channel 14/0 : 60[4] -> 61[5] via P2P/IPC
+ip-26-0-173-7:811251:811556 [3] NCCL INFO Channel 02/0 : 59[3] -> 60[4] via P2P/IPC
+ip-26-0-172-252:688202:688503 [2] NCCL INFO Channel 01/0 : 34[2] -> 35[3] via P2P/IPC
+ip-26-0-172-57:942613:942917 [2] NCCL INFO Connected all rings
+ip-26-0-173-7:811252:811560 [4] NCCL INFO Channel 15/0 : 60[4] -> 61[5] via P2P/IPC
+ip-26-0-173-7:811251:811556 [3] NCCL INFO Channel 03/0 : 59[3] -> 60[4] via P2P/IPC
+ip-26-0-172-57:942612:942914 [1] NCCL INFO Channel 05/0 : 41[1] -> 42[2] via P2P/IPC
+ip-26-0-172-252:688201:688500 [1] NCCL INFO Channel 07/0 : 33[1] -> 34[2] via P2P/IPC
+ip-26-0-172-73:926594:926898 [0] NCCL INFO Channel 13/0 : 48[0] -> 49[1] via P2P/IPC
+ip-26-0-172-142:880680:880984 [1] NCCL INFO Channel 11/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-173-7:811251:811556 [3] NCCL INFO Channel 05/0 : 59[3] -> 60[4] via P2P/IPC
+ip-26-0-172-142:880681:880983 [2] NCCL INFO Channel 02/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-160-225:94348:94650 [6] NCCL INFO Channel 10/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-225:94345:94649 [3] NCCL INFO Channel 06/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-252:688202:688503 [2] NCCL INFO Channel 02/0 : 34[2] -> 35[3] via P2P/IPC
+ip-26-0-173-7:811251:811556 [3] NCCL INFO Channel 06/0 : 59[3] -> 60[4] via P2P/IPC
+ip-26-0-173-7:811253:811558 [5] NCCL INFO Channel 05/0 : 53[5] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:94344:94648 [2] NCCL INFO Channel 05/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-173-7:811251:811556 [3] NCCL INFO Channel 07/0 : 59[3] -> 60[4] via P2P/IPC
+ip-26-0-172-57:942612:942914 [1] NCCL INFO Channel 06/0 : 41[1] -> 42[2] via P2P/IPC
+ip-26-0-173-7:811251:811556 [3] NCCL INFO Channel 09/0 : 59[3] -> 60[4] via P2P/IPC
+ip-26-0-160-225:94348:94650 [6] NCCL INFO Channel 11/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-225:94345:94649 [3] NCCL INFO Channel 07/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-225:94344:94648 [2] NCCL INFO Channel 06/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-173-7:811251:811556 [3] NCCL INFO Channel 10/0 : 59[3] -> 60[4] via P2P/IPC
+ip-26-0-172-252:688201:688500 [1] NCCL INFO Channel 09/0 : 33[1] -> 34[2] via P2P/IPC
+ip-26-0-172-142:880680:880984 [1] NCCL INFO Channel 12/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-172-142:880681:880983 [2] NCCL INFO Channel 04/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-173-7:811251:811556 [3] NCCL INFO Channel 11/0 : 59[3] -> 60[4] via P2P/IPC
+ip-26-0-172-116:1019289:1019593 [1] NCCL INFO Channel 01/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-172-252:688202:688503 [2] NCCL INFO Channel 04/0 : 34[2] -> 35[3] via P2P/IPC
+ip-26-0-172-57:942612:942914 [1] NCCL INFO Channel 07/0 : 41[1] -> 42[2] via P2P/IPC
+ip-26-0-173-7:811251:811556 [3] NCCL INFO Channel 13/0 : 59[3] -> 60[4] via P2P/IPC
+ip-26-0-173-7:811251:811556 [3] NCCL INFO Channel 14/0 : 59[3] -> 60[4] via P2P/IPC
+ip-26-0-172-73:926594:926898 [0] NCCL INFO Channel 14/0 : 48[0] -> 49[1] via P2P/IPC
+ip-26-0-172-116:1019289:1019593 [1] NCCL INFO Channel 03/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-172-147:618589:618883 [5] NCCL INFO Connected all rings
+ip-26-0-173-7:811251:811556 [3] NCCL INFO Channel 15/0 : 59[3] -> 60[4] via P2P/IPC
+ip-26-0-172-57:942618:942918 [7] NCCL INFO Connected all rings
+ip-26-0-172-57:942612:942914 [1] NCCL INFO Channel 09/0 : 41[1] -> 42[2] via P2P/IPC
+ip-26-0-172-142:880680:880984 [1] NCCL INFO Channel 13/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-172-142:880681:880983 [2] NCCL INFO Channel 05/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-160-225:94345:94649 [3] NCCL INFO Channel 09/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-225:94344:94648 [2] NCCL INFO Channel 07/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-173-7:811250:811555 [2] NCCL INFO Channel 02/0 : 50[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019289:1019593 [1] NCCL INFO Channel 04/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-173-7:811252:811560 [4] NCCL INFO Channel 04/0 : 52[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-7:811251:811556 [3] NCCL INFO Channel 03/0 : 51[3] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:94348:94650 [6] NCCL INFO Channel 12/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-116:1019294:1019589 [6] NCCL INFO Channel 01/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-172-57:942612:942914 [1] NCCL INFO Channel 11/0 : 41[1] -> 42[2] via P2P/IPC
+ip-26-0-160-225:94345:94649 [3] NCCL INFO Channel 10/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-225:94344:94648 [2] NCCL INFO Channel 09/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-116:1019291:1019594 [3] NCCL INFO Connected all rings
+ip-26-0-160-225:94348:94650 [6] NCCL INFO Channel 13/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-252:688201:688500 [1] NCCL INFO Channel 11/0 : 33[1] -> 34[2] via P2P/IPC
+ip-26-0-172-142:880680:880984 [1] NCCL INFO Channel 14/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-172-142:880681:880983 [2] NCCL INFO Channel 06/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-172-73:926594:926898 [0] NCCL INFO Channel 15/0 : 48[0] -> 49[1] via P2P/IPC
+ip-26-0-172-142:880682:880978 [3] NCCL INFO Channel 01/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-172-116:1019289:1019593 [1] NCCL INFO Channel 05/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-172-116:1019294:1019589 [6] NCCL INFO Channel 02/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-160-225:94345:94649 [3] NCCL INFO Channel 11/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-225:94344:94648 [2] NCCL INFO Channel 10/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-252:688202:688503 [2] NCCL INFO Channel 05/0 : 34[2] -> 35[3] via P2P/IPC
+ip-26-0-172-57:942612:942914 [1] NCCL INFO Channel 12/0 : 41[1] -> 42[2] via P2P/IPC
+ip-26-0-160-225:94348:94650 [6] NCCL INFO Channel 14/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-225:94344:94648 [2] NCCL INFO Channel 12/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-225:94345:94649 [3] NCCL INFO Channel 13/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-252:688201:688500 [1] NCCL INFO Channel 12/0 : 33[1] -> 34[2] via P2P/IPC
+ip-26-0-172-252:688202:688503 [2] NCCL INFO Channel 06/0 : 34[2] -> 35[3] via P2P/IPC
+ip-26-0-172-116:1019294:1019589 [6] NCCL INFO Channel 03/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-172-116:1019289:1019593 [1] NCCL INFO Channel 06/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-172-116:1019290:1019595 [2] NCCL INFO Channel 01/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-172-142:880680:880984 [1] NCCL INFO Channel 15/0 : 17[1] -> 18[2] via P2P/IPC
+ip-26-0-160-225:94344:94648 [2] NCCL INFO Channel 13/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-225:94345:94649 [3] NCCL INFO Channel 14/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-142:880681:880983 [2] NCCL INFO Channel 07/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-172-142:880682:880978 [3] NCCL INFO Channel 02/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-172-57:942612:942914 [1] NCCL INFO Channel 13/0 : 41[1] -> 42[2] via P2P/IPC
+ip-26-0-160-225:94344:94648 [2] NCCL INFO Channel 14/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-225:94345:94649 [3] NCCL INFO Channel 15/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-147:618584:618882 [0] NCCL INFO Channel 15/0 : 24[0] -> 25[1] via P2P/IPC
+ip-26-0-172-252:688201:688500 [1] NCCL INFO Channel 13/0 : 33[1] -> 34[2] via P2P/IPC
+ip-26-0-172-252:688202:688503 [2] NCCL INFO Channel 07/0 : 34[2] -> 35[3] via P2P/IPC
+ip-26-0-172-116:1019294:1019589 [6] NCCL INFO Channel 04/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-172-116:1019289:1019593 [1] NCCL INFO Channel 07/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-172-147:618585:618884 [1] NCCL INFO Channel 06/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-160-225:94344:94648 [2] NCCL INFO Channel 15/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-116:1019290:1019595 [2] NCCL INFO Channel 02/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-173-7:811248:811554 [0] NCCL INFO Channel 13/0 : 56[0] -> 63[7] via P2P/IPC
+ip-26-0-172-142:880681:880983 [2] NCCL INFO Channel 09/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-172-142:880682:880978 [3] NCCL INFO Channel 03/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-172-142:880679:880980 [0] NCCL INFO Channel 05/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-173-7:811248:811554 [0] NCCL INFO Channel 14/0 : 56[0] -> 63[7] via P2P/IPC
+ip-26-0-172-252:688203:688501 [3] NCCL INFO Channel 01/0 : 35[3] -> 36[4] via P2P/IPC
+ip-26-0-160-225:94343:94647 [1] NCCL INFO Channel 09/0 : 1[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880683:880982 [4] NCCL INFO Channel 01/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-160-225:94344:94648 [2] NCCL INFO Channel 10/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618591:618885 [7] NCCL INFO Connected all rings
+ip-26-0-172-147:618585:618884 [1] NCCL INFO Channel 07/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-172-57:942616:942913 [5] NCCL INFO Connected all rings
+ip-26-0-172-116:1019294:1019589 [6] NCCL INFO Channel 05/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-173-7:811248:811554 [0] NCCL INFO Channel 15/0 : 56[0] -> 63[7] via P2P/IPC
+ip-26-0-172-116:1019289:1019593 [1] NCCL INFO Channel 09/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-172-73:926595:926897 [1] NCCL INFO Channel 01/0 : 49[1] -> 50[2] via P2P/IPC
+ip-26-0-172-116:1019290:1019595 [2] NCCL INFO Channel 04/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-172-142:880681:880983 [2] NCCL INFO Channel 10/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-172-142:880682:880978 [3] NCCL INFO Channel 05/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-173-7:811248:811554 [0] NCCL INFO Channel 00/0 : 48[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880679:880980 [0] NCCL INFO Channel 06/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-172-73:926595:926897 [1] NCCL INFO Channel 03/0 : 49[1] -> 50[2] via P2P/IPC
+ip-26-0-173-7:811255:811561 [7] NCCL INFO Channel 07/0 : 55[7] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880683:880982 [4] NCCL INFO Channel 02/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-172-57:942612:942914 [1] NCCL INFO Channel 14/0 : 41[1] -> 42[2] via P2P/IPC
+ip-26-0-172-73:926600:926893 [6] NCCL INFO Connected all rings
+ip-26-0-172-252:688201:688500 [1] NCCL INFO Channel 14/0 : 33[1] -> 34[2] via P2P/IPC
+ip-26-0-172-252:688202:688503 [2] NCCL INFO Channel 09/0 : 34[2] -> 35[3] via P2P/IPC
+ip-26-0-172-252:688203:688501 [3] NCCL INFO Channel 02/0 : 35[3] -> 36[4] via P2P/IPC
+ip-26-0-172-116:1019294:1019589 [6] NCCL INFO Channel 06/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-172-116:1019289:1019593 [1] NCCL INFO Channel 11/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-172-116:1019290:1019595 [2] NCCL INFO Channel 05/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-172-142:880681:880983 [2] NCCL INFO Channel 12/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-172-73:926596:926899 [2] NCCL INFO Connected all rings
+ip-26-0-172-252:688201:688500 [1] NCCL INFO Channel 15/0 : 33[1] -> 34[2] via P2P/IPC
+ip-26-0-172-142:880682:880978 [3] NCCL INFO Channel 06/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-172-252:688202:688503 [2] NCCL INFO Channel 10/0 : 34[2] -> 35[3] via P2P/IPC
+ip-26-0-172-142:880679:880980 [0] NCCL INFO Channel 07/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-172-57:942612:942914 [1] NCCL INFO Channel 15/0 : 41[1] -> 42[2] via P2P/IPC
+ip-26-0-172-147:618590:618889 [6] NCCL INFO Connected all rings
+ip-26-0-172-142:880683:880982 [4] NCCL INFO Channel 03/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-172-147:618585:618884 [1] NCCL INFO Channel 09/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-172-252:688203:688501 [3] NCCL INFO Channel 03/0 : 35[3] -> 36[4] via P2P/IPC
+ip-26-0-172-116:1019294:1019589 [6] NCCL INFO Channel 09/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-172-116:1019289:1019593 [1] NCCL INFO Channel 12/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-172-116:1019290:1019595 [2] NCCL INFO Channel 06/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-172-73:926595:926897 [1] NCCL INFO Channel 04/0 : 49[1] -> 50[2] via P2P/IPC
+ip-26-0-172-142:880684:880981 [5] NCCL INFO Channel 01/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-172-142:880681:880983 [2] NCCL INFO Channel 13/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-172-142:880682:880978 [3] NCCL INFO Channel 07/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-172-142:880683:880982 [4] NCCL INFO Channel 04/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-172-252:688202:688503 [2] NCCL INFO Channel 12/0 : 34[2] -> 35[3] via P2P/IPC
+ip-26-0-172-252:688200:688507 [0] NCCL INFO Channel 05/0 : 32[0] -> 39[7] via P2P/IPC
+ip-26-0-172-252:688203:688501 [3] NCCL INFO Channel 05/0 : 35[3] -> 36[4] via P2P/IPC
+ip-26-0-172-116:1019294:1019589 [6] NCCL INFO Channel 10/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-172-147:618585:618884 [1] NCCL INFO Channel 11/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-172-116:1019289:1019593 [1] NCCL INFO Channel 13/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-172-252:688204:688502 [4] NCCL INFO Connected all rings
+ip-26-0-172-116:1019290:1019595 [2] NCCL INFO Channel 07/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-172-116:1019292:1019590 [4] NCCL INFO Channel 01/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-172-73:926595:926897 [1] NCCL INFO Channel 05/0 : 49[1] -> 50[2] via P2P/IPC
+ip-26-0-172-142:880684:880981 [5] NCCL INFO Channel 02/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-172-142:880681:880983 [2] NCCL INFO Channel 14/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-172-142:880682:880978 [3] NCCL INFO Channel 09/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-172-142:880683:880982 [4] NCCL INFO Channel 06/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-160-225:94346:94644 [4] NCCL INFO Connected all rings
+ip-26-0-172-142:880685:880977 [6] NCCL INFO Channel 01/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-172-116:1019294:1019589 [6] NCCL INFO Channel 11/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-172-252:688206:688506 [6] NCCL INFO Connected all rings
+ip-26-0-172-142:880679:880980 [0] NCCL INFO Channel 13/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-172-116:1019289:1019593 [1] NCCL INFO Channel 14/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-172-116:1019292:1019590 [4] NCCL INFO Channel 02/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-172-116:1019290:1019595 [2] NCCL INFO Channel 09/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-172-147:618588:618888 [4] NCCL INFO Connected all rings
+ip-26-0-172-142:880684:880981 [5] NCCL INFO Channel 03/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-172-142:880681:880983 [2] NCCL INFO Channel 15/0 : 18[2] -> 19[3] via P2P/IPC
+ip-26-0-172-142:880682:880978 [3] NCCL INFO Channel 10/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-172-142:880683:880982 [4] NCCL INFO Channel 07/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-172-252:688202:688503 [2] NCCL INFO Channel 13/0 : 34[2] -> 35[3] via P2P/IPC
+ip-26-0-172-252:688200:688507 [0] NCCL INFO Channel 06/0 : 32[0] -> 39[7] via P2P/IPC
+ip-26-0-172-116:1019294:1019589 [6] NCCL INFO Channel 12/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-172-142:880685:880977 [6] NCCL INFO Channel 02/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-172-252:688203:688501 [3] NCCL INFO Channel 06/0 : 35[3] -> 36[4] via P2P/IPC
+ip-26-0-172-73:926598:926894 [4] NCCL INFO Connected all rings
+ip-26-0-172-116:1019289:1019593 [1] NCCL INFO Channel 15/0 : 9[1] -> 10[2] via P2P/IPC
+ip-26-0-172-142:880679:880980 [0] NCCL INFO Channel 14/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-172-116:1019292:1019590 [4] NCCL INFO Channel 03/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-172-147:618585:618884 [1] NCCL INFO Channel 12/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-172-116:1019290:1019595 [2] NCCL INFO Channel 10/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-172-142:880684:880981 [5] NCCL INFO Channel 04/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-172-142:880683:880982 [4] NCCL INFO Channel 09/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-172-142:880682:880978 [3] NCCL INFO Channel 11/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-172-142:880680:880984 [1] NCCL INFO Channel 01/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019294:1019589 [6] NCCL INFO Channel 13/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-172-252:688202:688503 [2] NCCL INFO Channel 14/0 : 34[2] -> 35[3] via P2P/IPC
+ip-26-0-172-142:880680:880984 [1] NCCL INFO Channel 09/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880685:880977 [6] NCCL INFO Channel 03/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-172-252:688200:688507 [0] NCCL INFO Channel 07/0 : 32[0] -> 39[7] via P2P/IPC
+ip-26-0-172-142:880680:880984 [1] NCCL INFO Channel 01/0 : 17[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880679:880980 [0] NCCL INFO Channel 15/0 : 16[0] -> 23[7] via P2P/IPC
+ip-26-0-172-73:926595:926897 [1] NCCL INFO Channel 06/0 : 49[1] -> 50[2] via P2P/IPC
+ip-26-0-172-252:688203:688501 [3] NCCL INFO Channel 07/0 : 35[3] -> 36[4] via P2P/IPC
+ip-26-0-172-116:1019288:1019588 [0] NCCL INFO Channel 05/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-172-116:1019292:1019590 [4] NCCL INFO Channel 04/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-172-116:1019290:1019595 [2] NCCL INFO Channel 12/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-172-57:942611:942911 [0] NCCL INFO Channel 05/0 : 40[0] -> 47[7] via P2P/IPC
+ip-26-0-172-147:618585:618884 [1] NCCL INFO Channel 13/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-172-142:880684:880981 [5] NCCL INFO Channel 05/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-172-142:880683:880982 [4] NCCL INFO Channel 10/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-172-142:880682:880978 [3] NCCL INFO Channel 13/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-172-142:880685:880977 [6] NCCL INFO Channel 04/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-172-57:942617:942916 [6] NCCL INFO Connected all rings
+ip-26-0-172-116:1019294:1019589 [6] NCCL INFO Channel 14/0 : 14[6] -> 15[7] via P2P/IPC
+ip-26-0-172-252:688202:688503 [2] NCCL INFO Channel 15/0 : 34[2] -> 35[3] via P2P/IPC
+ip-26-0-172-252:688200:688507 [0] NCCL INFO Channel 13/0 : 32[0] -> 39[7] via P2P/IPC
+ip-26-0-172-116:1019288:1019588 [0] NCCL INFO Channel 06/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-172-116:1019292:1019590 [4] NCCL INFO Channel 06/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-172-142:880684:880981 [5] NCCL INFO Channel 07/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-172-142:880683:880982 [4] NCCL INFO Channel 11/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-172-116:1019290:1019595 [2] NCCL INFO Channel 13/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-172-116:1019293:1019592 [5] NCCL INFO Connected all rings
+ip-26-0-172-142:880682:880978 [3] NCCL INFO Channel 14/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-172-252:688203:688501 [3] NCCL INFO Channel 09/0 : 35[3] -> 36[4] via P2P/IPC
+ip-26-0-172-73:926595:926897 [1] NCCL INFO Channel 07/0 : 49[1] -> 50[2] via P2P/IPC
+ip-26-0-172-142:880685:880977 [6] NCCL INFO Channel 05/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-172-147:618585:618884 [1] NCCL INFO Channel 14/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-160-225:94348:94650 [6] NCCL INFO Channel 14/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880683:880982 [4] NCCL INFO Channel 12/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-172-142:880682:880978 [3] NCCL INFO Channel 15/0 : 19[3] -> 20[4] via P2P/IPC
+ip-26-0-172-142:880684:880981 [5] NCCL INFO Channel 09/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-172-142:880685:880977 [6] NCCL INFO Channel 06/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-172-73:926595:926897 [1] NCCL INFO Channel 09/0 : 49[1] -> 50[2] via P2P/IPC
+ip-26-0-160-225:94346:94644 [4] NCCL INFO Channel 01/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-252:688200:688507 [0] NCCL INFO Channel 14/0 : 32[0] -> 39[7] via P2P/IPC
+ip-26-0-172-252:688201:688500 [1] NCCL INFO Channel 09/0 : 33[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688203:688501 [3] NCCL INFO Channel 10/0 : 35[3] -> 36[4] via P2P/IPC
+ip-26-0-172-116:1019292:1019590 [4] NCCL INFO Channel 07/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-172-116:1019288:1019588 [0] NCCL INFO Channel 07/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-172-116:1019291:1019594 [3] NCCL INFO Channel 01/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-172-116:1019290:1019595 [2] NCCL INFO Channel 14/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-172-147:618585:618884 [1] NCCL INFO Channel 15/0 : 25[1] -> 26[2] via P2P/IPC
+ip-26-0-160-225:94346:94644 [4] NCCL INFO Channel 02/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-142:880684:880981 [5] NCCL INFO Channel 10/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-172-142:880683:880982 [4] NCCL INFO Channel 14/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-172-142:880685:880977 [6] NCCL INFO Channel 09/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-160-225:94346:94644 [4] NCCL INFO Channel 03/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-142:880681:880983 [2] NCCL INFO Channel 02/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880681:880983 [2] NCCL INFO Channel 10/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880681:880983 [2] NCCL INFO Channel 02/0 : 18[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926595:926897 [1] NCCL INFO Channel 11/0 : 49[1] -> 50[2] via P2P/IPC
+ip-26-0-172-252:688200:688507 [0] NCCL INFO Channel 15/0 : 32[0] -> 39[7] via P2P/IPC
+ip-26-0-172-142:880684:880981 [5] NCCL INFO Channel 11/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-172-116:1019292:1019590 [4] NCCL INFO Channel 09/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-172-142:880683:880982 [4] NCCL INFO Channel 15/0 : 20[4] -> 21[5] via P2P/IPC
+ip-26-0-172-252:688203:688501 [3] NCCL INFO Channel 11/0 : 35[3] -> 36[4] via P2P/IPC
+ip-26-0-172-142:880685:880977 [6] NCCL INFO Channel 10/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-160-225:94346:94644 [4] NCCL INFO Channel 04/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-116:1019288:1019588 [0] NCCL INFO Channel 13/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-172-116:1019290:1019595 [2] NCCL INFO Channel 15/0 : 10[2] -> 11[3] via P2P/IPC
+ip-26-0-172-116:1019291:1019594 [3] NCCL INFO Channel 02/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-172-147:618586:618886 [2] NCCL INFO Channel 01/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-172-147:618584:618882 [0] NCCL INFO Channel 05/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-160-225:94346:94644 [4] NCCL INFO Channel 06/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-142:880684:880981 [5] NCCL INFO Channel 12/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-172-142:880685:880977 [6] NCCL INFO Channel 11/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-160-225:94349:94645 [7] NCCL INFO Channel 15/0 : 7[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:94342:94643 [0] NCCL INFO Channel 08/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688203:688501 [3] NCCL INFO Channel 13/0 : 35[3] -> 36[4] via P2P/IPC
+ip-26-0-160-225:94346:94644 [4] NCCL INFO Channel 07/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-142:880682:880978 [3] NCCL INFO Channel 03/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880682:880978 [3] NCCL INFO Channel 11/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:94346:94644 [4] NCCL INFO Channel 09/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-116:1019292:1019590 [4] NCCL INFO Channel 10/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-172-116:1019288:1019588 [0] NCCL INFO Channel 14/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-172-116:1019291:1019594 [3] NCCL INFO Channel 03/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-172-142:880684:880981 [5] NCCL INFO Channel 13/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-172-142:880685:880977 [6] NCCL INFO Channel 12/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-172-142:880682:880978 [3] NCCL INFO Channel 03/0 : 19[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618586:618886 [2] NCCL INFO Channel 02/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-172-73:926599:926895 [5] NCCL INFO Connected all rings
+ip-26-0-172-252:688203:688501 [3] NCCL INFO Channel 14/0 : 35[3] -> 36[4] via P2P/IPC
+ip-26-0-160-225:94346:94644 [4] NCCL INFO Channel 10/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-116:1019289:1019593 [1] NCCL INFO Channel 09/0 : 1[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019289:1019593 [1] NCCL INFO Channel 01/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880684:880981 [5] NCCL INFO Channel 15/0 : 21[5] -> 22[6] via P2P/IPC
+ip-26-0-172-147:618584:618882 [0] NCCL INFO Channel 06/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-172-116:1019289:1019593 [1] NCCL INFO Channel 09/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880685:880977 [6] NCCL INFO Channel 13/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-160-225:94346:94644 [4] NCCL INFO Channel 11/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-57:942611:942911 [0] NCCL INFO Channel 06/0 : 40[0] -> 47[7] via P2P/IPC
+ip-26-0-172-116:1019292:1019590 [4] NCCL INFO Channel 11/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-172-116:1019291:1019594 [3] NCCL INFO Channel 05/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-172-116:1019288:1019588 [0] NCCL INFO Channel 15/0 : 8[0] -> 15[7] via P2P/IPC
+ip-26-0-160-225:94346:94644 [4] NCCL INFO Channel 12/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-142:880685:880977 [6] NCCL INFO Channel 14/0 : 22[6] -> 23[7] via P2P/IPC
+ip-26-0-172-116:1019289:1019593 [1] NCCL INFO Channel 09/0 : 9[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688203:688501 [3] NCCL INFO Channel 15/0 : 35[3] -> 36[4] via P2P/IPC
+ip-26-0-160-225:94346:94644 [4] NCCL INFO Channel 14/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-147:618586:618886 [2] NCCL INFO Channel 04/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-172-142:880683:880982 [4] NCCL INFO Channel 04/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926595:926897 [1] NCCL INFO Channel 12/0 : 49[1] -> 50[2] via P2P/IPC
+ip-26-0-172-142:880683:880982 [4] NCCL INFO Channel 12/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880683:880982 [4] NCCL INFO Channel 04/0 : 20[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942611:942911 [0] NCCL INFO Channel 07/0 : 40[0] -> 47[7] via P2P/IPC
+ip-26-0-172-142:880684:880981 [5] NCCL INFO Channel 05/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:94346:94644 [4] NCCL INFO Channel 15/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-147:618584:618882 [0] NCCL INFO Channel 07/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-172-142:880684:880981 [5] NCCL INFO Channel 13/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880685:880977 [6] NCCL INFO Channel 06/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880684:880981 [5] NCCL INFO Channel 05/0 : 21[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880685:880977 [6] NCCL INFO Channel 14/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019292:1019590 [4] NCCL INFO Channel 12/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-172-116:1019291:1019594 [3] NCCL INFO Channel 06/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-172-142:880685:880977 [6] NCCL INFO Channel 06/0 : 22[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:94345:94649 [3] NCCL INFO Channel 11/0 : 3[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942614:942915 [3] NCCL INFO Channel 01/0 : 43[3] -> 44[4] via P2P/IPC
+ip-26-0-160-225:94346:94644 [4] NCCL INFO Channel 12/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:94347:94646 [5] NCCL INFO Channel 13/0 : 5[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688202:688503 [2] NCCL INFO Channel 10/0 : 34[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926595:926897 [1] NCCL INFO Channel 13/0 : 49[1] -> 50[2] via P2P/IPC
+ip-26-0-172-147:618586:618886 [2] NCCL INFO Channel 05/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-172-116:1019292:1019590 [4] NCCL INFO Channel 14/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-172-116:1019291:1019594 [3] NCCL INFO Channel 07/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-172-147:618584:618882 [0] NCCL INFO Channel 13/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-172-252:688205:688505 [5] NCCL INFO Channel 01/0 : 37[5] -> 38[6] via P2P/IPC
+ip-26-0-172-116:1019292:1019590 [4] NCCL INFO Channel 15/0 : 12[4] -> 13[5] via P2P/IPC
+ip-26-0-172-116:1019291:1019594 [3] NCCL INFO Channel 09/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-172-73:926595:926897 [1] NCCL INFO Channel 14/0 : 49[1] -> 50[2] via P2P/IPC
+ip-26-0-172-252:688204:688502 [4] NCCL INFO Channel 01/0 : 36[4] -> 37[5] via P2P/IPC
+ip-26-0-172-142:880686:880979 [7] NCCL INFO Channel 07/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880686:880979 [7] NCCL INFO Channel 15/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618586:618886 [2] NCCL INFO Channel 06/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-172-252:688206:688506 [6] NCCL INFO Channel 01/0 : 38[6] -> 39[7] via P2P/IPC
+ip-26-0-172-142:880686:880979 [7] NCCL INFO Channel 07/0 : 23[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688205:688505 [5] NCCL INFO Channel 02/0 : 37[5] -> 38[6] via P2P/IPC
+ip-26-0-172-252:688204:688502 [4] NCCL INFO Channel 02/0 : 36[4] -> 37[5] via P2P/IPC
+ip-26-0-172-116:1019291:1019594 [3] NCCL INFO Channel 10/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-172-147:618584:618882 [0] NCCL INFO Channel 14/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-172-252:688206:688506 [6] NCCL INFO Channel 02/0 : 38[6] -> 39[7] via P2P/IPC
+ip-26-0-172-147:618589:618883 [5] NCCL INFO Channel 01/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-172-252:688205:688505 [5] NCCL INFO Channel 03/0 : 37[5] -> 38[6] via P2P/IPC
+ip-26-0-172-116:1019291:1019594 [3] NCCL INFO Channel 11/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-172-252:688204:688502 [4] NCCL INFO Channel 03/0 : 36[4] -> 37[5] via P2P/IPC
+ip-26-0-172-73:926595:926897 [1] NCCL INFO Channel 15/0 : 49[1] -> 50[2] via P2P/IPC
+ip-26-0-172-73:926597:926900 [3] NCCL INFO Channel 01/0 : 51[3] -> 52[4] via P2P/IPC
+ip-26-0-172-147:618586:618886 [2] NCCL INFO Channel 07/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-172-252:688206:688506 [6] NCCL INFO Channel 03/0 : 38[6] -> 39[7] via P2P/IPC
+ip-26-0-172-116:1019291:1019594 [3] NCCL INFO Channel 13/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-172-252:688205:688505 [5] NCCL INFO Channel 04/0 : 37[5] -> 38[6] via P2P/IPC
+ip-26-0-172-252:688204:688502 [4] NCCL INFO Channel 04/0 : 36[4] -> 37[5] via P2P/IPC
+ip-26-0-160-225:94343:94647 [1] NCCL INFO Channel 01/0 : 33[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618584:618882 [0] NCCL INFO Channel 15/0 : 24[0] -> 31[7] via P2P/IPC
+ip-26-0-160-225:94343:94647 [1] NCCL INFO Channel 01/0 : 1[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926600:926893 [6] NCCL INFO Channel 01/0 : 54[6] -> 55[7] via P2P/IPC
+ip-26-0-172-116:1019291:1019594 [3] NCCL INFO Channel 14/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-172-147:618589:618883 [5] NCCL INFO Channel 02/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-172-252:688206:688506 [6] NCCL INFO Channel 04/0 : 38[6] -> 39[7] via P2P/IPC
+ip-26-0-172-73:926594:926898 [0] NCCL INFO Channel 05/0 : 48[0] -> 55[7] via P2P/IPC
+ip-26-0-172-252:688205:688505 [5] NCCL INFO Channel 05/0 : 37[5] -> 38[6] via P2P/IPC
+ip-26-0-172-252:688204:688502 [4] NCCL INFO Channel 06/0 : 36[4] -> 37[5] via P2P/IPC
+ip-26-0-172-147:618590:618889 [6] NCCL INFO Channel 01/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-172-73:926597:926900 [3] NCCL INFO Channel 02/0 : 51[3] -> 52[4] via P2P/IPC
+ip-26-0-172-147:618586:618886 [2] NCCL INFO Channel 09/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-172-116:1019291:1019594 [3] NCCL INFO Channel 15/0 : 11[3] -> 12[4] via P2P/IPC
+ip-26-0-172-116:1019293:1019592 [5] NCCL INFO Channel 01/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-172-252:688206:688506 [6] NCCL INFO Channel 05/0 : 38[6] -> 39[7] via P2P/IPC
+ip-26-0-172-252:688205:688505 [5] NCCL INFO Channel 07/0 : 37[5] -> 38[6] via P2P/IPC
+ip-26-0-172-73:926600:926893 [6] NCCL INFO Channel 02/0 : 54[6] -> 55[7] via P2P/IPC
+ip-26-0-172-252:688204:688502 [4] NCCL INFO Channel 07/0 : 36[4] -> 37[5] via P2P/IPC
+ip-26-0-172-147:618589:618883 [5] NCCL INFO Channel 03/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-172-116:1019293:1019592 [5] NCCL INFO Channel 02/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-172-73:926594:926898 [0] NCCL INFO Channel 06/0 : 48[0] -> 55[7] via P2P/IPC
+ip-26-0-172-147:618590:618889 [6] NCCL INFO Channel 02/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-172-73:926597:926900 [3] NCCL INFO Channel 03/0 : 51[3] -> 52[4] via P2P/IPC
+ip-26-0-172-116:1019290:1019595 [2] NCCL INFO Channel 10/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618586:618886 [2] NCCL INFO Channel 10/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-172-116:1019293:1019592 [5] NCCL INFO Channel 03/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-172-116:1019290:1019595 [2] NCCL INFO Channel 02/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019291:1019594 [3] NCCL INFO Channel 11/0 : 3[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688205:688505 [5] NCCL INFO Channel 09/0 : 37[5] -> 38[6] via P2P/IPC
+ip-26-0-172-116:1019290:1019595 [2] NCCL INFO Channel 10/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019291:1019594 [3] NCCL INFO Channel 03/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688206:688506 [6] NCCL INFO Channel 06/0 : 38[6] -> 39[7] via P2P/IPC
+ip-26-0-172-252:688204:688502 [4] NCCL INFO Channel 09/0 : 36[4] -> 37[5] via P2P/IPC
+ip-26-0-172-116:1019291:1019594 [3] NCCL INFO Channel 11/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019293:1019592 [5] NCCL INFO Channel 04/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-172-73:926600:926893 [6] NCCL INFO Channel 03/0 : 54[6] -> 55[7] via P2P/IPC
+ip-26-0-160-225:94344:94648 [2] NCCL INFO Channel 02/0 : 34[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:94344:94648 [2] NCCL INFO Channel 02/0 : 2[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:94345:94649 [3] NCCL INFO Channel 03/0 : 35[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019290:1019595 [2] NCCL INFO Channel 10/0 : 10[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:94345:94649 [3] NCCL INFO Channel 03/0 : 3[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019293:1019592 [5] NCCL INFO Channel 05/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-172-147:618589:618883 [5] NCCL INFO Channel 04/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-172-116:1019291:1019594 [3] NCCL INFO Channel 11/0 : 11[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618588:618888 [4] NCCL INFO Channel 01/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-172-252:688205:688505 [5] NCCL INFO Channel 10/0 : 37[5] -> 38[6] via P2P/IPC
+ip-26-0-172-252:688206:688506 [6] NCCL INFO Channel 09/0 : 38[6] -> 39[7] via P2P/IPC
+ip-26-0-172-252:688204:688502 [4] NCCL INFO Channel 10/0 : 36[4] -> 37[5] via P2P/IPC
+ip-26-0-172-147:618590:618889 [6] NCCL INFO Channel 03/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-172-73:926594:926898 [0] NCCL INFO Channel 07/0 : 48[0] -> 55[7] via P2P/IPC
+ip-26-0-172-142:880679:880980 [0] NCCL INFO Channel 00/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618586:618886 [2] NCCL INFO Channel 12/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-172-116:1019293:1019592 [5] NCCL INFO Channel 07/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-172-142:880679:880980 [0] NCCL INFO Channel 08/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926597:926900 [3] NCCL INFO Channel 05/0 : 51[3] -> 52[4] via P2P/IPC
+ip-26-0-172-142:880679:880980 [0] NCCL INFO Channel 00/0 : 16[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942611:942911 [0] NCCL INFO Channel 13/0 : 40[0] -> 47[7] via P2P/IPC
+ip-26-0-172-116:1019293:1019592 [5] NCCL INFO Channel 09/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-172-73:926600:926893 [6] NCCL INFO Channel 04/0 : 54[6] -> 55[7] via P2P/IPC
+ip-26-0-172-252:688205:688505 [5] NCCL INFO Channel 11/0 : 37[5] -> 38[6] via P2P/IPC
+ip-26-0-172-252:688206:688506 [6] NCCL INFO Channel 10/0 : 38[6] -> 39[7] via P2P/IPC
+ip-26-0-172-252:688204:688502 [4] NCCL INFO Channel 11/0 : 36[4] -> 37[5] via P2P/IPC
+ip-26-0-172-116:1019295:1019591 [7] NCCL INFO Connected all rings
+ip-26-0-172-116:1019293:1019592 [5] NCCL INFO Channel 10/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-172-147:618589:618883 [5] NCCL INFO Channel 05/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-172-147:618588:618888 [4] NCCL INFO Channel 02/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-172-116:1019293:1019592 [5] NCCL INFO Channel 11/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-172-147:618590:618889 [6] NCCL INFO Channel 04/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-172-147:618586:618886 [2] NCCL INFO Channel 13/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-172-73:926594:926898 [0] NCCL INFO Channel 13/0 : 48[0] -> 55[7] via P2P/IPC
+ip-26-0-172-252:688205:688505 [5] NCCL INFO Channel 12/0 : 37[5] -> 38[6] via P2P/IPC
+ip-26-0-172-252:688206:688506 [6] NCCL INFO Channel 11/0 : 38[6] -> 39[7] via P2P/IPC
+ip-26-0-172-252:688204:688502 [4] NCCL INFO Channel 12/0 : 36[4] -> 37[5] via P2P/IPC
+ip-26-0-172-73:926597:926900 [3] NCCL INFO Channel 06/0 : 51[3] -> 52[4] via P2P/IPC
+ip-26-0-172-116:1019293:1019592 [5] NCCL INFO Channel 12/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-172-73:926600:926893 [6] NCCL INFO Channel 05/0 : 54[6] -> 55[7] via P2P/IPC
+ip-26-0-172-147:618589:618883 [5] NCCL INFO Channel 07/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-172-116:1019293:1019592 [5] NCCL INFO Channel 13/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-172-147:618588:618888 [4] NCCL INFO Channel 03/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-172-252:688205:688505 [5] NCCL INFO Channel 13/0 : 37[5] -> 38[6] via P2P/IPC
+ip-26-0-172-252:688206:688506 [6] NCCL INFO Channel 12/0 : 38[6] -> 39[7] via P2P/IPC
+ip-26-0-172-252:688204:688502 [4] NCCL INFO Channel 14/0 : 36[4] -> 37[5] via P2P/IPC
+ip-26-0-172-147:618590:618889 [6] NCCL INFO Channel 05/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-172-147:618586:618886 [2] NCCL INFO Channel 14/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-172-116:1019293:1019592 [5] NCCL INFO Channel 15/0 : 13[5] -> 14[6] via P2P/IPC
+ip-26-0-172-73:926594:926898 [0] NCCL INFO Channel 14/0 : 48[0] -> 55[7] via P2P/IPC
+ip-26-0-172-73:926597:926900 [3] NCCL INFO Channel 07/0 : 51[3] -> 52[4] via P2P/IPC
+ip-26-0-172-252:688205:688505 [5] NCCL INFO Channel 15/0 : 37[5] -> 38[6] via P2P/IPC
+ip-26-0-172-57:942611:942911 [0] NCCL INFO Channel 14/0 : 40[0] -> 47[7] via P2P/IPC
+ip-26-0-172-57:942611:942911 [0] NCCL INFO Channel 15/0 : 40[0] -> 47[7] via P2P/IPC
+ip-26-0-172-252:688206:688506 [6] NCCL INFO Channel 13/0 : 38[6] -> 39[7] via P2P/IPC
+ip-26-0-172-57:942613:942917 [2] NCCL INFO Channel 01/0 : 42[2] -> 43[3] via P2P/IPC
+ip-26-0-172-252:688204:688502 [4] NCCL INFO Channel 15/0 : 36[4] -> 37[5] via P2P/IPC
+ip-26-0-172-73:926600:926893 [6] NCCL INFO Channel 06/0 : 54[6] -> 55[7] via P2P/IPC
+ip-26-0-172-147:618589:618883 [5] NCCL INFO Channel 09/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-172-147:618588:618888 [4] NCCL INFO Channel 04/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-172-116:1019292:1019590 [4] NCCL INFO Channel 12/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019292:1019590 [4] NCCL INFO Channel 04/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019293:1019592 [5] NCCL INFO Channel 13/0 : 5[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019292:1019590 [4] NCCL INFO Channel 12/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019293:1019592 [5] NCCL INFO Channel 05/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618590:618889 [6] NCCL INFO Channel 06/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-172-116:1019293:1019592 [5] NCCL INFO Channel 13/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618586:618886 [2] NCCL INFO Channel 15/0 : 26[2] -> 27[3] via P2P/IPC
+ip-26-0-172-252:688206:688506 [6] NCCL INFO Channel 14/0 : 38[6] -> 39[7] via P2P/IPC
+ip-26-0-160-225:94346:94644 [4] NCCL INFO Channel 04/0 : 36[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:94346:94644 [4] NCCL INFO Channel 04/0 : 4[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688203:688501 [3] NCCL INFO Channel 11/0 : 35[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019292:1019590 [4] NCCL INFO Channel 12/0 : 12[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688204:688502 [4] NCCL INFO Channel 12/0 : 36[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618588:618888 [4] NCCL INFO Channel 06/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-172-147:618589:618883 [5] NCCL INFO Channel 10/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-172-147:618590:618889 [6] NCCL INFO Channel 09/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-172-73:926594:926898 [0] NCCL INFO Channel 15/0 : 48[0] -> 55[7] via P2P/IPC
+ip-26-0-172-73:926597:926900 [3] NCCL INFO Channel 09/0 : 51[3] -> 52[4] via P2P/IPC
+ip-26-0-172-73:926598:926894 [4] NCCL INFO Channel 01/0 : 52[4] -> 53[5] via P2P/IPC
+ip-26-0-172-147:618585:618884 [1] NCCL INFO Channel 01/0 : 17[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942614:942915 [3] NCCL INFO Channel 02/0 : 43[3] -> 44[4] via P2P/IPC
+ip-26-0-172-57:942615:942912 [4] NCCL INFO Channel 01/0 : 44[4] -> 45[5] via P2P/IPC
+ip-26-0-172-57:942613:942917 [2] NCCL INFO Channel 02/0 : 42[2] -> 43[3] via P2P/IPC
+ip-26-0-172-57:942614:942915 [3] NCCL INFO Channel 03/0 : 43[3] -> 44[4] via P2P/IPC
+ip-26-0-172-57:942615:942912 [4] NCCL INFO Channel 02/0 : 44[4] -> 45[5] via P2P/IPC
+ip-26-0-172-57:942613:942917 [2] NCCL INFO Channel 04/0 : 42[2] -> 43[3] via P2P/IPC
+ip-26-0-172-57:942616:942913 [5] NCCL INFO Channel 01/0 : 45[5] -> 46[6] via P2P/IPC
+ip-26-0-172-57:942614:942915 [3] NCCL INFO Channel 05/0 : 43[3] -> 44[4] via P2P/IPC
+ip-26-0-172-57:942615:942912 [4] NCCL INFO Channel 03/0 : 44[4] -> 45[5] via P2P/IPC
+ip-26-0-172-57:942613:942917 [2] NCCL INFO Channel 05/0 : 42[2] -> 43[3] via P2P/IPC
+ip-26-0-172-252:688205:688505 [5] NCCL INFO Channel 13/0 : 37[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618588:618888 [4] NCCL INFO Channel 07/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-172-147:618589:618883 [5] NCCL INFO Channel 11/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-172-147:618585:618884 [1] NCCL INFO Channel 09/0 : 9[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688206:688506 [6] NCCL INFO Channel 14/0 : 38[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880680:880984 [1] NCCL INFO Channel 01/0 : 17[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618590:618889 [6] NCCL INFO Channel 10/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-172-147:618585:618884 [1] NCCL INFO Channel 09/0 : 25[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019294:1019589 [6] NCCL INFO Channel 14/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019294:1019589 [6] NCCL INFO Channel 06/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019293:1019592 [5] NCCL INFO Channel 13/0 : 13[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019294:1019589 [6] NCCL INFO Channel 14/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926597:926900 [3] NCCL INFO Channel 10/0 : 51[3] -> 52[4] via P2P/IPC
+ip-26-0-172-73:926598:926894 [4] NCCL INFO Channel 02/0 : 52[4] -> 53[5] via P2P/IPC
+ip-26-0-172-116:1019289:1019593 [1] NCCL INFO Channel 09/0 : 25[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:94348:94650 [6] NCCL INFO Channel 06/0 : 38[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:94348:94650 [6] NCCL INFO Channel 06/0 : 6[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618588:618888 [4] NCCL INFO Channel 09/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-172-116:1019294:1019589 [6] NCCL INFO Channel 14/0 : 14[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926600:926893 [6] NCCL INFO Channel 09/0 : 54[6] -> 55[7] via P2P/IPC
+ip-26-0-172-147:618589:618883 [5] NCCL INFO Channel 12/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-172-147:618587:618887 [3] NCCL INFO Channel 01/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-172-147:618590:618889 [6] NCCL INFO Channel 11/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-172-147:618588:618888 [4] NCCL INFO Channel 10/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-172-147:618589:618883 [5] NCCL INFO Channel 13/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-172-57:942616:942913 [5] NCCL INFO Channel 02/0 : 45[5] -> 46[6] via P2P/IPC
+ip-26-0-172-57:942615:942912 [4] NCCL INFO Channel 04/0 : 44[4] -> 45[5] via P2P/IPC
+ip-26-0-172-57:942614:942915 [3] NCCL INFO Channel 06/0 : 43[3] -> 44[4] via P2P/IPC
+ip-26-0-172-57:942613:942917 [2] NCCL INFO Channel 06/0 : 42[2] -> 43[3] via P2P/IPC
+ip-26-0-172-57:942616:942913 [5] NCCL INFO Channel 03/0 : 45[5] -> 46[6] via P2P/IPC
+ip-26-0-172-57:942615:942912 [4] NCCL INFO Channel 06/0 : 44[4] -> 45[5] via P2P/IPC
+ip-26-0-172-57:942614:942915 [3] NCCL INFO Channel 07/0 : 43[3] -> 44[4] via P2P/IPC
+ip-26-0-172-57:942613:942917 [2] NCCL INFO Channel 07/0 : 42[2] -> 43[3] via P2P/IPC
+ip-26-0-172-147:618587:618887 [3] NCCL INFO Channel 02/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-172-147:618590:618889 [6] NCCL INFO Channel 12/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-172-57:942616:942913 [5] NCCL INFO Channel 04/0 : 45[5] -> 46[6] via P2P/IPC
+ip-26-0-172-116:1019295:1019591 [7] NCCL INFO Channel 15/0 : 7[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688207:688504 [7] NCCL INFO Channel 15/0 : 39[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019295:1019591 [7] NCCL INFO Channel 07/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688200:688507 [0] NCCL INFO Channel 08/0 : 32[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019295:1019591 [7] NCCL INFO Channel 15/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618588:618888 [4] NCCL INFO Channel 11/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-160-225:94349:94645 [7] NCCL INFO Channel 07/0 : 39[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926597:926900 [3] NCCL INFO Channel 11/0 : 51[3] -> 52[4] via P2P/IPC
+ip-26-0-160-225:94349:94645 [7] NCCL INFO Channel 07/0 : 7[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926598:926894 [4] NCCL INFO Channel 03/0 : 52[4] -> 53[5] via P2P/IPC
+ip-26-0-172-116:1019295:1019591 [7] NCCL INFO Channel 15/0 : 15[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926599:926895 [5] NCCL INFO Channel 01/0 : 53[5] -> 54[6] via P2P/IPC
+ip-26-0-172-147:618589:618883 [5] NCCL INFO Channel 15/0 : 29[5] -> 30[6] via P2P/IPC
+ip-26-0-172-147:618590:618889 [6] NCCL INFO Channel 13/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-172-147:618587:618887 [3] NCCL INFO Channel 03/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-172-116:1019288:1019588 [0] NCCL INFO Channel 08/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019288:1019588 [0] NCCL INFO Channel 00/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019288:1019588 [0] NCCL INFO Channel 08/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618588:618888 [4] NCCL INFO Channel 12/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-160-225:94342:94643 [0] NCCL INFO Channel 00/0 : 32[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926600:926893 [6] NCCL INFO Channel 10/0 : 54[6] -> 55[7] via P2P/IPC
+ip-26-0-172-147:618590:618889 [6] NCCL INFO Channel 14/0 : 30[6] -> 31[7] via P2P/IPC
+ip-26-0-160-225:94342:94643 [0] NCCL INFO Channel 00/0 : 0[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019288:1019588 [0] NCCL INFO Channel 08/0 : 8[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926597:926900 [3] NCCL INFO Channel 13/0 : 51[3] -> 52[4] via P2P/IPC
+ip-26-0-172-73:926598:926894 [4] NCCL INFO Channel 04/0 : 52[4] -> 53[5] via P2P/IPC
+ip-26-0-172-73:926596:926899 [2] NCCL INFO Channel 01/0 : 50[2] -> 51[3] via P2P/IPC
+ip-26-0-172-73:926599:926895 [5] NCCL INFO Channel 02/0 : 53[5] -> 54[6] via P2P/IPC
+ip-26-0-172-147:618587:618887 [3] NCCL INFO Channel 05/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-172-147:618588:618888 [4] NCCL INFO Channel 14/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-172-57:942615:942912 [4] NCCL INFO Channel 07/0 : 44[4] -> 45[5] via P2P/IPC
+ip-26-0-172-57:942617:942916 [6] NCCL INFO Channel 01/0 : 46[6] -> 47[7] via P2P/IPC
+ip-26-0-172-57:942614:942915 [3] NCCL INFO Channel 09/0 : 43[3] -> 44[4] via P2P/IPC
+ip-26-0-172-57:942613:942917 [2] NCCL INFO Channel 09/0 : 42[2] -> 43[3] via P2P/IPC
+ip-26-0-172-57:942616:942913 [5] NCCL INFO Channel 05/0 : 45[5] -> 46[6] via P2P/IPC
+ip-26-0-172-57:942615:942912 [4] NCCL INFO Channel 09/0 : 44[4] -> 45[5] via P2P/IPC
+ip-26-0-172-57:942617:942916 [6] NCCL INFO Channel 02/0 : 46[6] -> 47[7] via P2P/IPC
+ip-26-0-172-57:942614:942915 [3] NCCL INFO Channel 10/0 : 43[3] -> 44[4] via P2P/IPC
+ip-26-0-172-57:942613:942917 [2] NCCL INFO Channel 10/0 : 42[2] -> 43[3] via P2P/IPC
+ip-26-0-172-57:942616:942913 [5] NCCL INFO Channel 07/0 : 45[5] -> 46[6] via P2P/IPC
+ip-26-0-172-147:618587:618887 [3] NCCL INFO Channel 06/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-172-147:618588:618888 [4] NCCL INFO Channel 15/0 : 28[4] -> 29[5] via P2P/IPC
+ip-26-0-172-73:926600:926893 [6] NCCL INFO Channel 11/0 : 54[6] -> 55[7] via P2P/IPC
+ip-26-0-172-147:618590:618889 [6] NCCL INFO Channel 06/0 : 22[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:94347:94646 [5] NCCL INFO Channel 05/0 : 37[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:94347:94646 [5] NCCL INFO Channel 05/0 : 5[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926598:926894 [4] NCCL INFO Channel 06/0 : 52[4] -> 53[5] via P2P/IPC
+ip-26-0-172-73:926596:926899 [2] NCCL INFO Channel 02/0 : 50[2] -> 51[3] via P2P/IPC
+ip-26-0-172-73:926597:926900 [3] NCCL INFO Channel 14/0 : 51[3] -> 52[4] via P2P/IPC
+ip-26-0-172-73:926599:926895 [5] NCCL INFO Channel 03/0 : 53[5] -> 54[6] via P2P/IPC
+ip-26-0-172-147:618590:618889 [6] NCCL INFO Channel 14/0 : 14[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880685:880977 [6] NCCL INFO Channel 06/0 : 22[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618590:618889 [6] NCCL INFO Channel 14/0 : 30[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618587:618887 [3] NCCL INFO Channel 07/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-172-73:926600:926893 [6] NCCL INFO Channel 12/0 : 54[6] -> 55[7] via P2P/IPC
+ip-26-0-172-116:1019294:1019589 [6] NCCL INFO Channel 14/0 : 30[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618589:618883 [5] NCCL INFO Channel 05/0 : 21[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618587:618887 [3] NCCL INFO Channel 09/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-172-57:942615:942912 [4] NCCL INFO Channel 10/0 : 44[4] -> 45[5] via P2P/IPC
+ip-26-0-172-57:942617:942916 [6] NCCL INFO Channel 03/0 : 46[6] -> 47[7] via P2P/IPC
+ip-26-0-172-57:942614:942915 [3] NCCL INFO Channel 11/0 : 43[3] -> 44[4] via P2P/IPC
+ip-26-0-172-57:942613:942917 [2] NCCL INFO Channel 12/0 : 42[2] -> 43[3] via P2P/IPC
+ip-26-0-172-57:942616:942913 [5] NCCL INFO Channel 09/0 : 45[5] -> 46[6] via P2P/IPC
+ip-26-0-172-57:942615:942912 [4] NCCL INFO Channel 11/0 : 44[4] -> 45[5] via P2P/IPC
+ip-26-0-172-57:942617:942916 [6] NCCL INFO Channel 04/0 : 46[6] -> 47[7] via P2P/IPC
+ip-26-0-172-57:942614:942915 [3] NCCL INFO Channel 13/0 : 43[3] -> 44[4] via P2P/IPC
+ip-26-0-172-57:942613:942917 [2] NCCL INFO Channel 13/0 : 42[2] -> 43[3] via P2P/IPC
+ip-26-0-172-57:942616:942913 [5] NCCL INFO Channel 10/0 : 45[5] -> 46[6] via P2P/IPC
+ip-26-0-172-147:618589:618883 [5] NCCL INFO Channel 13/0 : 13[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926598:926894 [4] NCCL INFO Channel 07/0 : 52[4] -> 53[5] via P2P/IPC
+ip-26-0-172-73:926596:926899 [2] NCCL INFO Channel 04/0 : 50[2] -> 51[3] via P2P/IPC
+ip-26-0-172-73:926597:926900 [3] NCCL INFO Channel 15/0 : 51[3] -> 52[4] via P2P/IPC
+ip-26-0-172-73:926599:926895 [5] NCCL INFO Channel 04/0 : 53[5] -> 54[6] via P2P/IPC
+ip-26-0-172-147:618589:618883 [5] NCCL INFO Channel 13/0 : 29[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880684:880981 [5] NCCL INFO Channel 05/0 : 21[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618587:618887 [3] NCCL INFO Channel 10/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-172-116:1019293:1019592 [5] NCCL INFO Channel 13/0 : 29[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926600:926893 [6] NCCL INFO Channel 13/0 : 54[6] -> 55[7] via P2P/IPC
+ip-26-0-172-147:618587:618887 [3] NCCL INFO Channel 11/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-172-57:942615:942912 [4] NCCL INFO Channel 12/0 : 44[4] -> 45[5] via P2P/IPC
+ip-26-0-172-57:942617:942916 [6] NCCL INFO Channel 05/0 : 46[6] -> 47[7] via P2P/IPC
+ip-26-0-172-57:942614:942915 [3] NCCL INFO Channel 14/0 : 43[3] -> 44[4] via P2P/IPC
+ip-26-0-172-57:942613:942917 [2] NCCL INFO Channel 14/0 : 42[2] -> 43[3] via P2P/IPC
+ip-26-0-172-57:942616:942913 [5] NCCL INFO Channel 11/0 : 45[5] -> 46[6] via P2P/IPC
+ip-26-0-172-57:942615:942912 [4] NCCL INFO Channel 14/0 : 44[4] -> 45[5] via P2P/IPC
+ip-26-0-172-57:942617:942916 [6] NCCL INFO Channel 06/0 : 46[6] -> 47[7] via P2P/IPC
+ip-26-0-172-57:942613:942917 [2] NCCL INFO Channel 15/0 : 42[2] -> 43[3] via P2P/IPC
+ip-26-0-172-57:942614:942915 [3] NCCL INFO Channel 15/0 : 43[3] -> 44[4] via P2P/IPC
+ip-26-0-172-73:926598:926894 [4] NCCL INFO Channel 09/0 : 52[4] -> 53[5] via P2P/IPC
+ip-26-0-172-73:926596:926899 [2] NCCL INFO Channel 05/0 : 50[2] -> 51[3] via P2P/IPC
+ip-26-0-172-57:942616:942913 [5] NCCL INFO Channel 12/0 : 45[5] -> 46[6] via P2P/IPC
+ip-26-0-172-73:926599:926895 [5] NCCL INFO Channel 05/0 : 53[5] -> 54[6] via P2P/IPC
+ip-26-0-172-57:942615:942912 [4] NCCL INFO Channel 15/0 : 44[4] -> 45[5] via P2P/IPC
+ip-26-0-172-57:942617:942916 [6] NCCL INFO Channel 09/0 : 46[6] -> 47[7] via P2P/IPC
+ip-26-0-172-57:942616:942913 [5] NCCL INFO Channel 13/0 : 45[5] -> 46[6] via P2P/IPC
+ip-26-0-172-147:618587:618887 [3] NCCL INFO Channel 13/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-172-73:926600:926893 [6] NCCL INFO Channel 14/0 : 54[6] -> 55[7] via P2P/IPC
+ip-26-0-172-57:942612:942914 [1] NCCL INFO Channel 09/0 : 33[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942613:942917 [2] NCCL INFO Channel 10/0 : 34[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942612:942914 [1] NCCL INFO Channel 01/0 : 41[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942613:942917 [2] NCCL INFO Channel 02/0 : 42[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942612:942914 [1] NCCL INFO Channel 09/0 : 41[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942617:942916 [6] NCCL INFO Channel 10/0 : 46[6] -> 47[7] via P2P/IPC
+ip-26-0-172-57:942613:942917 [2] NCCL INFO Channel 10/0 : 42[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942616:942913 [5] NCCL INFO Channel 15/0 : 45[5] -> 46[6] via P2P/IPC
+ip-26-0-172-147:618587:618887 [3] NCCL INFO Channel 14/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-172-252:688201:688500 [1] NCCL INFO Channel 01/0 : 17[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942614:942915 [3] NCCL INFO Channel 11/0 : 35[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926596:926899 [2] NCCL INFO Channel 06/0 : 50[2] -> 51[3] via P2P/IPC
+ip-26-0-172-73:926598:926894 [4] NCCL INFO Channel 10/0 : 52[4] -> 53[5] via P2P/IPC
+ip-26-0-172-73:926599:926895 [5] NCCL INFO Channel 07/0 : 53[5] -> 54[6] via P2P/IPC
+ip-26-0-172-252:688202:688503 [2] NCCL INFO Channel 02/0 : 18[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942614:942915 [3] NCCL INFO Channel 03/0 : 43[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688201:688500 [1] NCCL INFO Channel 01/0 : 33[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688202:688503 [2] NCCL INFO Channel 02/0 : 34[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618591:618885 [7] NCCL INFO Channel 07/0 : 23[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942617:942916 [6] NCCL INFO Channel 11/0 : 46[6] -> 47[7] via P2P/IPC
+ip-26-0-172-147:618587:618887 [3] NCCL INFO Channel 15/0 : 27[3] -> 28[4] via P2P/IPC
+ip-26-0-172-57:942614:942915 [3] NCCL INFO Channel 11/0 : 43[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618584:618882 [0] NCCL INFO Channel 00/0 : 16[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942617:942916 [6] NCCL INFO Channel 12/0 : 46[6] -> 47[7] via P2P/IPC
+ip-26-0-172-142:880680:880984 [1] NCCL INFO Channel 01/0 : 33[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618591:618885 [7] NCCL INFO Channel 15/0 : 15[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942615:942912 [4] NCCL INFO Channel 12/0 : 36[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618584:618882 [0] NCCL INFO Channel 08/0 : 8[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880686:880979 [7] NCCL INFO Channel 07/0 : 23[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942615:942912 [4] NCCL INFO Channel 04/0 : 44[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618591:618885 [7] NCCL INFO Channel 15/0 : 31[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618584:618882 [0] NCCL INFO Channel 08/0 : 24[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880679:880980 [0] NCCL INFO Channel 00/0 : 16[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942615:942912 [4] NCCL INFO Channel 12/0 : 44[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688203:688501 [3] NCCL INFO Channel 03/0 : 19[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942617:942916 [6] NCCL INFO Channel 13/0 : 46[6] -> 47[7] via P2P/IPC
+ip-26-0-172-73:926596:926899 [2] NCCL INFO Channel 07/0 : 50[2] -> 51[3] via P2P/IPC
+ip-26-0-172-73:926599:926895 [5] NCCL INFO Channel 09/0 : 53[5] -> 54[6] via P2P/IPC
+ip-26-0-172-147:618586:618886 [2] NCCL INFO Channel 02/0 : 18[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688203:688501 [3] NCCL INFO Channel 03/0 : 35[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926598:926894 [4] NCCL INFO Channel 11/0 : 52[4] -> 53[5] via P2P/IPC
+ip-26-0-172-147:618588:618888 [4] NCCL INFO Channel 04/0 : 20[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019295:1019591 [7] NCCL INFO Channel 15/0 : 31[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019288:1019588 [0] NCCL INFO Channel 08/0 : 24[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618587:618887 [3] NCCL INFO Channel 03/0 : 19[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942617:942916 [6] NCCL INFO Channel 14/0 : 46[6] -> 47[7] via P2P/IPC
+ip-26-0-172-147:618586:618886 [2] NCCL INFO Channel 10/0 : 10[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688204:688502 [4] NCCL INFO Channel 04/0 : 20[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880681:880983 [2] NCCL INFO Channel 02/0 : 18[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618586:618886 [2] NCCL INFO Channel 10/0 : 26[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688204:688502 [4] NCCL INFO Channel 04/0 : 36[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618588:618888 [4] NCCL INFO Channel 12/0 : 12[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618587:618887 [3] NCCL INFO Channel 11/0 : 11[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880683:880982 [4] NCCL INFO Channel 04/0 : 20[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618588:618888 [4] NCCL INFO Channel 12/0 : 28[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880682:880978 [3] NCCL INFO Channel 03/0 : 19[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618587:618887 [3] NCCL INFO Channel 11/0 : 27[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880681:880983 [2] NCCL INFO Channel 02/0 : 34[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019290:1019595 [2] NCCL INFO Channel 10/0 : 26[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926596:926899 [2] NCCL INFO Channel 09/0 : 50[2] -> 51[3] via P2P/IPC
+ip-26-0-172-73:926599:926895 [5] NCCL INFO Channel 10/0 : 53[5] -> 54[6] via P2P/IPC
+ip-26-0-172-73:926598:926894 [4] NCCL INFO Channel 12/0 : 52[4] -> 53[5] via P2P/IPC
+ip-26-0-172-142:880683:880982 [4] NCCL INFO Channel 04/0 : 36[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942616:942913 [5] NCCL INFO Channel 13/0 : 37[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019292:1019590 [4] NCCL INFO Channel 12/0 : 28[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880682:880978 [3] NCCL INFO Channel 03/0 : 35[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019291:1019594 [3] NCCL INFO Channel 11/0 : 27[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942616:942913 [5] NCCL INFO Channel 05/0 : 45[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942617:942916 [6] NCCL INFO Channel 14/0 : 38[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942616:942913 [5] NCCL INFO Channel 13/0 : 45[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942617:942916 [6] NCCL INFO Channel 06/0 : 46[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942617:942916 [6] NCCL INFO Channel 14/0 : 46[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688205:688505 [5] NCCL INFO Channel 05/0 : 21[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688206:688506 [6] NCCL INFO Channel 06/0 : 22[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926599:926895 [5] NCCL INFO Channel 11/0 : 53[5] -> 54[6] via P2P/IPC
+ip-26-0-172-73:926596:926899 [2] NCCL INFO Channel 10/0 : 50[2] -> 51[3] via P2P/IPC
+ip-26-0-172-252:688205:688505 [5] NCCL INFO Channel 05/0 : 37[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926598:926894 [4] NCCL INFO Channel 14/0 : 52[4] -> 53[5] via P2P/IPC
+ip-26-0-172-252:688206:688506 [6] NCCL INFO Channel 06/0 : 38[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880685:880977 [6] NCCL INFO Channel 06/0 : 38[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926599:926895 [5] NCCL INFO Channel 12/0 : 53[5] -> 54[6] via P2P/IPC
+ip-26-0-172-73:926596:926899 [2] NCCL INFO Channel 12/0 : 50[2] -> 51[3] via P2P/IPC
+ip-26-0-172-73:926598:926894 [4] NCCL INFO Channel 15/0 : 52[4] -> 53[5] via P2P/IPC
+ip-26-0-172-142:880684:880981 [5] NCCL INFO Channel 05/0 : 37[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926599:926895 [5] NCCL INFO Channel 13/0 : 53[5] -> 54[6] via P2P/IPC
+ip-26-0-172-73:926596:926899 [2] NCCL INFO Channel 13/0 : 50[2] -> 51[3] via P2P/IPC
+ip-26-0-172-57:942618:942918 [7] NCCL INFO Channel 15/0 : 39[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942611:942911 [0] NCCL INFO Channel 08/0 : 32[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688207:688504 [7] NCCL INFO Channel 07/0 : 23[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688200:688507 [0] NCCL INFO Channel 00/0 : 16[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942618:942918 [7] NCCL INFO Channel 07/0 : 47[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942611:942911 [0] NCCL INFO Channel 00/0 : 40[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942618:942918 [7] NCCL INFO Channel 15/0 : 47[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942611:942911 [0] NCCL INFO Channel 08/0 : 40[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688200:688507 [0] NCCL INFO Channel 00/0 : 32[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688207:688504 [7] NCCL INFO Channel 07/0 : 39[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926599:926895 [5] NCCL INFO Channel 15/0 : 53[5] -> 54[6] via P2P/IPC
+ip-26-0-172-73:926596:926899 [2] NCCL INFO Channel 14/0 : 50[2] -> 51[3] via P2P/IPC
+ip-26-0-172-142:880679:880980 [0] NCCL INFO Channel 00/0 : 32[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926594:926898 [0] NCCL INFO Channel 00/0 : 40[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926601:926896 [7] NCCL INFO Channel 07/0 : 47[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926594:926898 [0] NCCL INFO Channel 08/0 : 40[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880686:880979 [7] NCCL INFO Channel 07/0 : 39[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926601:926896 [7] NCCL INFO Channel 15/0 : 47[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926596:926899 [2] NCCL INFO Channel 15/0 : 50[2] -> 51[3] via P2P/IPC
+ip-26-0-172-73:926594:926898 [0] NCCL INFO Channel 00/0 : 48[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926601:926896 [7] NCCL INFO Channel 07/0 : 55[7] -> 63[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926598:926894 [4] NCCL INFO Channel 04/0 : 44[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926598:926894 [4] NCCL INFO Channel 12/0 : 44[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926600:926893 [6] NCCL INFO Channel 06/0 : 46[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942618:942918 [7] NCCL INFO Channel 15/0 : 31[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926599:926895 [5] NCCL INFO Channel 05/0 : 45[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926598:926894 [4] NCCL INFO Channel 04/0 : 52[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926600:926893 [6] NCCL INFO Channel 14/0 : 46[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926599:926895 [5] NCCL INFO Channel 13/0 : 45[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926600:926893 [6] NCCL INFO Channel 06/0 : 54[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926599:926895 [5] NCCL INFO Channel 05/0 : 53[5] -> 61[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926595:926897 [1] NCCL INFO Channel 01/0 : 41[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926597:926900 [3] NCCL INFO Channel 03/0 : 43[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926596:926899 [2] NCCL INFO Channel 02/0 : 42[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926595:926897 [1] NCCL INFO Channel 09/0 : 41[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942615:942912 [4] NCCL INFO Channel 12/0 : 28[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926596:926899 [2] NCCL INFO Channel 10/0 : 42[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926597:926900 [3] NCCL INFO Channel 11/0 : 43[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926595:926897 [1] NCCL INFO Channel 01/0 : 49[1] -> 57[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926596:926899 [2] NCCL INFO Channel 02/0 : 50[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926598:926894 [4] NCCL INFO Channel 04/0 : 36[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926597:926900 [3] NCCL INFO Channel 03/0 : 51[3] -> 59[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942616:942913 [5] NCCL INFO Channel 13/0 : 29[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942617:942916 [6] NCCL INFO Channel 14/0 : 30[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942615:942912 [4] NCCL INFO Channel 12/0 : 44[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618588:618888 [4] NCCL INFO Channel 12/0 : 60[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926599:926895 [5] NCCL INFO Channel 05/0 : 37[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926600:926893 [6] NCCL INFO Channel 06/0 : 38[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-7:811254:811557 [6] NCCL INFO Channel 14/0 : 30[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618588:618888 [4] NCCL INFO Channel 12/0 : 28[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926598:926894 [4] NCCL INFO Channel 04/0 : 52[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942612:942914 [1] NCCL INFO Channel 09/0 : 25[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-7:811253:811558 [5] NCCL INFO Channel 13/0 : 29[5] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-7:811252:811560 [4] NCCL INFO Channel 12/0 : 28[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688204:688502 [4] NCCL INFO Channel 04/0 : 4[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942617:942916 [6] NCCL INFO Channel 14/0 : 46[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926595:926897 [1] NCCL INFO Channel 01/0 : 33[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-7:811254:811557 [6] NCCL INFO Channel 14/0 : 62[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-7:811253:811558 [5] NCCL INFO Channel 13/0 : 61[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-7:811252:811560 [4] NCCL INFO Channel 12/0 : 60[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942614:942915 [3] NCCL INFO Channel 11/0 : 27[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942616:942913 [5] NCCL INFO Channel 13/0 : 45[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942613:942917 [2] NCCL INFO Channel 10/0 : 26[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618589:618883 [5] NCCL INFO Channel 13/0 : 61[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618590:618889 [6] NCCL INFO Channel 14/0 : 62[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688204:688502 [4] NCCL INFO Channel 04/0 : 36[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926596:926899 [2] NCCL INFO Channel 02/0 : 34[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926597:926900 [3] NCCL INFO Channel 03/0 : 35[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-7:811249:811559 [1] NCCL INFO Channel 09/0 : 25[1] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926600:926893 [6] NCCL INFO Channel 06/0 : 54[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926599:926895 [5] NCCL INFO Channel 05/0 : 53[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-7:811250:811555 [2] NCCL INFO Channel 10/0 : 26[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618589:618883 [5] NCCL INFO Channel 13/0 : 29[5] -> 61[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618590:618889 [6] NCCL INFO Channel 14/0 : 30[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688206:688506 [6] NCCL INFO Channel 06/0 : 6[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942612:942914 [1] NCCL INFO Channel 09/0 : 41[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-7:811249:811559 [1] NCCL INFO Channel 09/0 : 57[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618585:618884 [1] NCCL INFO Channel 09/0 : 57[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688205:688505 [5] NCCL INFO Channel 05/0 : 5[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926595:926897 [1] NCCL INFO Channel 01/0 : 49[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688206:688506 [6] NCCL INFO Channel 06/0 : 38[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-7:811250:811555 [2] NCCL INFO Channel 10/0 : 58[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926596:926899 [2] NCCL INFO Channel 02/0 : 50[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688205:688505 [5] NCCL INFO Channel 05/0 : 37[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688204:688502 [4] NCCL INFO Channel 04/0 : 52[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942613:942917 [2] NCCL INFO Channel 10/0 : 42[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942614:942915 [3] NCCL INFO Channel 11/0 : 43[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-7:811252:811560 [4] NCCL INFO Channel 04/0 : 60[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:94346:94644 [4] NCCL INFO Channel 12/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618587:618887 [3] NCCL INFO Channel 11/0 : 59[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688202:688503 [2] NCCL INFO Channel 02/0 : 2[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-7:811251:811556 [3] NCCL INFO Channel 11/0 : 27[3] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618585:618884 [1] NCCL INFO Channel 09/0 : 25[1] -> 57[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618586:618886 [2] NCCL INFO Channel 10/0 : 58[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618588:618888 [4] NCCL INFO Channel 12/0 : 44[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618589:618883 [5] NCCL INFO Channel 13/0 : 45[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688204:688502 [4] NCCL INFO Channel 04/0 : 36[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-7:811253:811558 [5] NCCL INFO Channel 05/0 : 61[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-7:811254:811557 [6] NCCL INFO Channel 06/0 : 62[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618590:618889 [6] NCCL INFO Channel 14/0 : 46[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926597:926900 [3] NCCL INFO Channel 03/0 : 51[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688201:688500 [1] NCCL INFO Channel 01/0 : 1[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688202:688503 [2] NCCL INFO Channel 02/0 : 34[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:94347:94646 [5] NCCL INFO Channel 13/0 : 13[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688205:688505 [5] NCCL INFO Channel 05/0 : 53[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-7:811251:811556 [3] NCCL INFO Channel 11/0 : 59[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618587:618887 [3] NCCL INFO Channel 11/0 : 27[3] -> 59[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688206:688506 [6] NCCL INFO Channel 06/0 : 54[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618588:618888 [4] NCCL INFO Channel 12/0 : 28[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618586:618886 [2] NCCL INFO Channel 10/0 : 26[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:94348:94650 [6] NCCL INFO Channel 14/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618589:618883 [5] NCCL INFO Channel 13/0 : 29[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618590:618889 [6] NCCL INFO Channel 14/0 : 30[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688201:688500 [1] NCCL INFO Channel 01/0 : 33[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688203:688501 [3] NCCL INFO Channel 03/0 : 3[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688205:688505 [5] NCCL INFO Channel 05/0 : 37[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688206:688506 [6] NCCL INFO Channel 06/0 : 38[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926598:926894 [4] NCCL INFO Channel 04/0 : 60[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618585:618884 [1] NCCL INFO Channel 09/0 : 41[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:94344:94648 [2] NCCL INFO Channel 10/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-7:811249:811559 [1] NCCL INFO Channel 01/0 : 57[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688203:688501 [3] NCCL INFO Channel 03/0 : 35[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688202:688503 [2] NCCL INFO Channel 02/0 : 50[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618585:618884 [1] NCCL INFO Channel 09/0 : 25[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926598:926894 [4] NCCL INFO Channel 04/0 : 52[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-7:811251:811556 [3] NCCL INFO Channel 03/0 : 59[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688202:688503 [2] NCCL INFO Channel 02/0 : 34[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688201:688500 [1] NCCL INFO Channel 01/0 : 49[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618586:618886 [2] NCCL INFO Channel 10/0 : 42[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880683:880982 [4] NCCL INFO Channel 04/0 : 28[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926598:926894 [4] NCCL INFO Channel 12/0 : 52[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:94343:94647 [1] NCCL INFO Channel 09/0 : 9[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-7:811250:811555 [2] NCCL INFO Channel 02/0 : 58[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942617:942916 [6] NCCL INFO Channel 06/0 : 54[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688204:688502 [4] NCCL INFO Channel 12/0 : 44[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618587:618887 [3] NCCL INFO Channel 11/0 : 43[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618586:618886 [2] NCCL INFO Channel 10/0 : 26[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880683:880982 [4] NCCL INFO Channel 04/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942616:942913 [5] NCCL INFO Channel 05/0 : 53[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942615:942912 [4] NCCL INFO Channel 04/0 : 52[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926600:926893 [6] NCCL INFO Channel 06/0 : 62[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926599:926895 [5] NCCL INFO Channel 05/0 : 61[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942617:942916 [6] NCCL INFO Channel 14/0 : 54[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688201:688500 [1] NCCL INFO Channel 01/0 : 33[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:94345:94649 [3] NCCL INFO Channel 11/0 : 11[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618587:618887 [3] NCCL INFO Channel 11/0 : 27[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880683:880982 [4] NCCL INFO Channel 12/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926600:926893 [6] NCCL INFO Channel 06/0 : 54[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942616:942913 [5] NCCL INFO Channel 13/0 : 53[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942615:942912 [4] NCCL INFO Channel 12/0 : 52[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688203:688501 [3] NCCL INFO Channel 03/0 : 51[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019292:1019590 [4] NCCL INFO Channel 04/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019294:1019589 [6] NCCL INFO Channel 06/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942612:942914 [1] NCCL INFO Channel 01/0 : 49[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942617:942916 [6] NCCL INFO Channel 14/0 : 46[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688205:688505 [5] NCCL INFO Channel 13/0 : 45[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618590:618889 [6] NCCL INFO Channel 06/0 : 30[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618588:618888 [4] NCCL INFO Channel 04/0 : 28[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688206:688506 [6] NCCL INFO Channel 14/0 : 46[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880685:880977 [6] NCCL INFO Channel 06/0 : 30[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618589:618883 [5] NCCL INFO Channel 05/0 : 29[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926599:926895 [5] NCCL INFO Channel 05/0 : 53[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926596:926899 [2] NCCL INFO Channel 02/0 : 58[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019293:1019592 [5] NCCL INFO Channel 05/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942616:942913 [5] NCCL INFO Channel 13/0 : 45[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942615:942912 [4] NCCL INFO Channel 12/0 : 44[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688203:688501 [3] NCCL INFO Channel 03/0 : 35[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880685:880977 [6] NCCL INFO Channel 06/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618585:618884 [1] NCCL INFO Channel 01/0 : 25[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019292:1019590 [4] NCCL INFO Channel 12/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019294:1019589 [6] NCCL INFO Channel 14/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019293:1019592 [5] NCCL INFO Channel 13/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926600:926893 [6] NCCL INFO Channel 14/0 : 54[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942612:942914 [1] NCCL INFO Channel 09/0 : 49[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926599:926895 [5] NCCL INFO Channel 13/0 : 53[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926596:926899 [2] NCCL INFO Channel 02/0 : 50[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688202:688503 [2] NCCL INFO Channel 10/0 : 42[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880681:880983 [2] NCCL INFO Channel 02/0 : 26[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019289:1019593 [1] NCCL INFO Channel 01/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618586:618886 [2] NCCL INFO Channel 02/0 : 26[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942613:942917 [2] NCCL INFO Channel 02/0 : 50[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926595:926897 [1] NCCL INFO Channel 01/0 : 57[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-7:811255:811561 [7] NCCL INFO Channel 15/0 : 31[7] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880685:880977 [6] NCCL INFO Channel 14/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926596:926899 [2] NCCL INFO Channel 10/0 : 50[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926601:926896 [7] NCCL INFO Channel 07/0 : 39[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926595:926897 [1] NCCL INFO Channel 01/0 : 49[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618591:618885 [7] NCCL INFO Channel 15/0 : 63[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019292:1019590 [4] NCCL INFO Channel 12/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019294:1019589 [6] NCCL INFO Channel 14/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019293:1019592 [5] NCCL INFO Channel 13/0 : 13[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942612:942914 [1] NCCL INFO Channel 09/0 : 41[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618587:618887 [3] NCCL INFO Channel 03/0 : 27[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942618:942918 [7] NCCL INFO Channel 15/0 : 47[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880681:880983 [2] NCCL INFO Channel 02/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-7:811255:811561 [7] NCCL INFO Channel 15/0 : 63[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942614:942915 [3] NCCL INFO Channel 03/0 : 51[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926595:926897 [1] NCCL INFO Channel 09/0 : 49[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688201:688500 [1] NCCL INFO Channel 09/0 : 41[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926597:926900 [3] NCCL INFO Channel 03/0 : 59[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019289:1019593 [1] NCCL INFO Channel 09/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618591:618885 [7] NCCL INFO Channel 15/0 : 31[7] -> 63[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019291:1019594 [3] NCCL INFO Channel 03/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019290:1019595 [2] NCCL INFO Channel 02/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880681:880983 [2] NCCL INFO Channel 10/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880680:880984 [1] NCCL INFO Channel 01/0 : 25[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942613:942917 [2] NCCL INFO Channel 10/0 : 50[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942614:942915 [3] NCCL INFO Channel 11/0 : 51[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926597:926900 [3] NCCL INFO Channel 03/0 : 51[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019289:1019593 [1] NCCL INFO Channel 09/0 : 9[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019290:1019595 [2] NCCL INFO Channel 10/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019291:1019594 [3] NCCL INFO Channel 11/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880684:880981 [5] NCCL INFO Channel 05/0 : 29[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942613:942917 [2] NCCL INFO Channel 10/0 : 42[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688203:688501 [3] NCCL INFO Channel 11/0 : 43[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019291:1019594 [3] NCCL INFO Channel 11/0 : 11[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880682:880978 [3] NCCL INFO Channel 03/0 : 27[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926601:926896 [7] NCCL INFO Channel 07/0 : 55[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942614:942915 [3] NCCL INFO Channel 11/0 : 43[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880684:880981 [5] NCCL INFO Channel 05/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926597:926900 [3] NCCL INFO Channel 11/0 : 51[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688207:688504 [7] NCCL INFO Channel 07/0 : 7[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019290:1019595 [2] NCCL INFO Channel 10/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880680:880984 [1] NCCL INFO Channel 01/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880682:880978 [3] NCCL INFO Channel 03/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688207:688504 [7] NCCL INFO Channel 07/0 : 39[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618591:618885 [7] NCCL INFO Channel 15/0 : 47[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880684:880981 [5] NCCL INFO Channel 13/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-7:811255:811561 [7] NCCL INFO Channel 07/0 : 63[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880680:880984 [1] NCCL INFO Channel 09/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880682:880978 [3] NCCL INFO Channel 11/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618591:618885 [7] NCCL INFO Channel 15/0 : 31[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942611:942911 [0] NCCL INFO Channel 08/0 : 24[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688207:688504 [7] NCCL INFO Channel 07/0 : 55[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:94349:94645 [7] NCCL INFO Channel 15/0 : 15[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-7:811248:811554 [0] NCCL INFO Channel 08/0 : 24[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926594:926898 [0] NCCL INFO Channel 00/0 : 32[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688207:688504 [7] NCCL INFO Channel 07/0 : 39[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942611:942911 [0] NCCL INFO Channel 08/0 : 40[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-7:811248:811554 [0] NCCL INFO Channel 08/0 : 56[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618584:618882 [0] NCCL INFO Channel 08/0 : 56[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942618:942918 [7] NCCL INFO Channel 07/0 : 55[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942618:942918 [7] NCCL INFO Channel 15/0 : 55[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926594:926898 [0] NCCL INFO Channel 00/0 : 48[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942618:942918 [7] NCCL INFO Channel 15/0 : 47[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688200:688507 [0] NCCL INFO Channel 00/0 : 0[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926601:926896 [7] NCCL INFO Channel 07/0 : 63[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926601:926896 [7] NCCL INFO Channel 07/0 : 55[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688200:688507 [0] NCCL INFO Channel 00/0 : 32[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880686:880979 [7] NCCL INFO Channel 07/0 : 31[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926601:926896 [7] NCCL INFO Channel 15/0 : 55[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688207:688504 [7] NCCL INFO Channel 15/0 : 47[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880686:880979 [7] NCCL INFO Channel 07/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688207:688504 [7] NCCL INFO Channel 01/0 : 39[7] -> 32[0] via P2P/IPC
+ip-26-0-172-142:880686:880979 [7] NCCL INFO Channel 15/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688200:688507 [0] NCCL INFO Channel 00/0 : 48[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:94342:94643 [0] NCCL INFO Channel 08/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688200:688507 [0] NCCL INFO Channel 00/0 : 32[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926601:926896 [7] NCCL INFO Channel 01/0 : 55[7] -> 48[0] via P2P/IPC
+ip-26-0-172-57:942618:942918 [7] NCCL INFO Channel 01/0 : 47[7] -> 40[0] via P2P/IPC
+ip-26-0-172-147:618591:618885 [7] NCCL INFO Channel 07/0 : 31[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019295:1019591 [7] NCCL INFO Channel 07/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-7:811255:811561 [7] NCCL INFO Channel 01/0 : 63[7] -> 56[0] via P2P/IPC
+ip-26-0-172-116:1019295:1019591 [7] NCCL INFO Channel 15/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926594:926898 [0] NCCL INFO Channel 00/0 : 56[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618591:618885 [7] NCCL INFO Channel 01/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-172-252:688200:688507 [0] NCCL INFO Channel 08/0 : 40[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019295:1019591 [7] NCCL INFO Channel 15/0 : 15[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880679:880980 [0] NCCL INFO Channel 00/0 : 24[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926594:926898 [0] NCCL INFO Channel 00/0 : 48[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880679:880980 [0] NCCL INFO Channel 00/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926594:926898 [0] NCCL INFO Channel 08/0 : 48[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880686:880979 [7] NCCL INFO Channel 01/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-172-142:880679:880980 [0] NCCL INFO Channel 08/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:94347:94646 [5] NCCL INFO Channel 00/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-225:94349:94645 [7] NCCL INFO Channel 01/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-116:1019295:1019591 [7] NCCL INFO Channel 01/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-160-225:94348:94650 [6] NCCL INFO Channel 00/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-173-7:811252:811560 [4] NCCL INFO Channel 00/0 : 60[4] -> 59[3] via P2P/IPC
+ip-26-0-172-252:688207:688504 [7] NCCL INFO Channel 02/0 : 39[7] -> 32[0] via P2P/IPC
+ip-26-0-172-73:926601:926896 [7] NCCL INFO Channel 02/0 : 55[7] -> 48[0] via P2P/IPC
+ip-26-0-172-73:926598:926894 [4] NCCL INFO Channel 00/0 : 52[4] -> 51[3] via P2P/IPC
+ip-26-0-172-142:880683:880982 [4] NCCL INFO Channel 00/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-172-116:1019295:1019591 [7] NCCL INFO Channel 02/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-173-7:811255:811561 [7] NCCL INFO Channel 02/0 : 63[7] -> 56[0] via P2P/IPC
+ip-26-0-172-147:618591:618885 [7] NCCL INFO Channel 02/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-172-147:618588:618888 [4] NCCL INFO Channel 00/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-172-142:880686:880979 [7] NCCL INFO Channel 02/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-160-225:94347:94646 [5] NCCL INFO Channel 04/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-225:94349:94645 [7] NCCL INFO Channel 02/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-73:926600:926893 [6] NCCL INFO Channel 00/0 : 54[6] -> 53[5] via P2P/IPC
+ip-26-0-172-252:688207:688504 [7] NCCL INFO Channel 03/0 : 39[7] -> 32[0] via P2P/IPC
+ip-26-0-160-225:94348:94650 [6] NCCL INFO Channel 05/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-173-7:811254:811557 [6] NCCL INFO Channel 00/0 : 62[6] -> 61[5] via P2P/IPC
+ip-26-0-173-7:811252:811560 [4] NCCL INFO Channel 03/0 : 60[4] -> 59[3] via P2P/IPC
+ip-26-0-172-73:926601:926896 [7] NCCL INFO Channel 03/0 : 55[7] -> 48[0] via P2P/IPC
+ip-26-0-172-73:926598:926894 [4] NCCL INFO Channel 03/0 : 52[4] -> 51[3] via P2P/IPC
+ip-26-0-172-116:1019295:1019591 [7] NCCL INFO Channel 03/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-172-252:688205:688505 [5] NCCL INFO Channel 00/0 : 37[5] -> 36[4] via P2P/IPC
+ip-26-0-173-7:811249:811559 [1] NCCL INFO Channel 00/0 : 57[1] -> 56[0] via P2P/IPC
+ip-26-0-172-142:880686:880979 [7] NCCL INFO Channel 03/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-172-147:618590:618889 [6] NCCL INFO Channel 00/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-173-7:811255:811561 [7] NCCL INFO Channel 03/0 : 63[7] -> 56[0] via P2P/IPC
+ip-26-0-172-252:688206:688506 [6] NCCL INFO Channel 00/0 : 38[6] -> 37[5] via P2P/IPC
+ip-26-0-172-142:880683:880982 [4] NCCL INFO Channel 03/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-172-142:880684:880981 [5] NCCL INFO Channel 00/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-172-147:618591:618885 [7] NCCL INFO Channel 03/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-172-147:618588:618888 [4] NCCL INFO Channel 03/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-172-73:926596:926899 [2] NCCL INFO Channel 00/0 : 50[2] -> 49[1] via P2P/IPC
+ip-26-0-172-57:942615:942912 [4] NCCL INFO Channel 00/0 : 44[4] -> 43[3] via P2P/IPC
+ip-26-0-172-116:1019293:1019592 [5] NCCL INFO Channel 00/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-173-7:811254:811557 [6] NCCL INFO Channel 05/0 : 62[6] -> 61[5] via P2P/IPC
+ip-26-0-172-147:618586:618886 [2] NCCL INFO Channel 00/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-172-73:926600:926893 [6] NCCL INFO Channel 05/0 : 54[6] -> 53[5] via P2P/IPC
+ip-26-0-172-252:688207:688504 [7] NCCL INFO Channel 04/0 : 39[7] -> 32[0] via P2P/IPC
+ip-26-0-172-142:880686:880979 [7] NCCL INFO Channel 04/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-173-7:811252:811560 [4] NCCL INFO Channel 05/0 : 60[4] -> 59[3] via P2P/IPC
+ip-26-0-173-7:811250:811555 [2] NCCL INFO Channel 00/0 : 58[2] -> 57[1] via P2P/IPC
+ip-26-0-172-73:926601:926896 [7] NCCL INFO Channel 04/0 : 55[7] -> 48[0] via P2P/IPC
+ip-26-0-172-57:942612:942914 [1] NCCL INFO Channel 00/0 : 41[1] -> 40[0] via P2P/IPC
+ip-26-0-172-142:880683:880982 [4] NCCL INFO Channel 05/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-172-252:688202:688503 [2] NCCL INFO Channel 00/0 : 34[2] -> 33[1] via P2P/IPC
+ip-26-0-172-73:926598:926894 [4] NCCL INFO Channel 05/0 : 52[4] -> 51[3] via P2P/IPC
+ip-26-0-172-116:1019295:1019591 [7] NCCL INFO Channel 04/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-172-142:880684:880981 [5] NCCL INFO Channel 04/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-172-57:942615:942912 [4] NCCL INFO Channel 03/0 : 44[4] -> 43[3] via P2P/IPC
+ip-26-0-172-142:880681:880983 [2] NCCL INFO Channel 00/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-160-225:94348:94650 [6] NCCL INFO Channel 08/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-252:688204:688502 [4] NCCL INFO Channel 00/0 : 36[4] -> 35[3] via P2P/IPC
+ip-26-0-172-142:880686:880979 [7] NCCL INFO Channel 05/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-172-252:688205:688505 [5] NCCL INFO Channel 04/0 : 37[5] -> 36[4] via P2P/IPC
+ip-26-0-172-147:618589:618883 [5] NCCL INFO Channel 00/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-172-147:618590:618889 [6] NCCL INFO Channel 05/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-160-225:94349:94645 [7] NCCL INFO Channel 03/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-225:94344:94648 [2] NCCL INFO Channel 00/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-147:618585:618884 [1] NCCL INFO Channel 00/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-172-252:688206:688506 [6] NCCL INFO Channel 05/0 : 38[6] -> 37[5] via P2P/IPC
+ip-26-0-172-116:1019293:1019592 [5] NCCL INFO Channel 04/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-172-73:926595:926897 [1] NCCL INFO Channel 00/0 : 49[1] -> 48[0] via P2P/IPC
+ip-26-0-172-142:880685:880977 [6] NCCL INFO Channel 00/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-172-73:926596:926899 [2] NCCL INFO Channel 01/0 : 50[2] -> 49[1] via P2P/IPC
+ip-26-0-172-252:688201:688500 [1] NCCL INFO Channel 00/0 : 33[1] -> 32[0] via P2P/IPC
+ip-26-0-172-147:618591:618885 [7] NCCL INFO Channel 04/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-172-73:926600:926893 [6] NCCL INFO Channel 08/0 : 54[6] -> 53[5] via P2P/IPC
+ip-26-0-160-225:94347:94646 [5] NCCL INFO Channel 08/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-147:618588:618888 [4] NCCL INFO Channel 05/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-172-252:688207:688504 [7] NCCL INFO Channel 05/0 : 39[7] -> 32[0] via P2P/IPC
+ip-26-0-173-7:811253:811558 [5] NCCL INFO Channel 00/0 : 61[5] -> 60[4] via P2P/IPC
+ip-26-0-172-57:942612:942914 [1] NCCL INFO Channel 08/0 : 41[1] -> 40[0] via P2P/IPC
+ip-26-0-172-57:942615:942912 [4] NCCL INFO Channel 05/0 : 44[4] -> 43[3] via P2P/IPC
+ip-26-0-172-147:618586:618886 [2] NCCL INFO Channel 01/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-172-142:880683:880982 [4] NCCL INFO Channel 06/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-173-7:811253:811558 [5] NCCL INFO Channel 04/0 : 61[5] -> 60[4] via P2P/IPC
+ip-26-0-172-57:942616:942913 [5] NCCL INFO Channel 00/0 : 45[5] -> 44[4] via P2P/IPC
+ip-26-0-172-142:880684:880981 [5] NCCL INFO Channel 08/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-172-142:880681:880983 [2] NCCL INFO Channel 01/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-172-116:1019295:1019591 [7] NCCL INFO Channel 05/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-172-116:1019289:1019593 [1] NCCL INFO Channel 00/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-160-225:94348:94650 [6] NCCL INFO Channel 13/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-225:94343:94647 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-252:688202:688503 [2] NCCL INFO Channel 01/0 : 34[2] -> 33[1] via P2P/IPC
+ip-26-0-172-142:880680:880984 [1] NCCL INFO Channel 00/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-172-142:880686:880979 [7] NCCL INFO Channel 06/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-172-73:926598:926894 [4] NCCL INFO Channel 06/0 : 52[4] -> 51[3] via P2P/IPC
+ip-26-0-173-7:811253:811558 [5] NCCL INFO Channel 08/0 : 61[5] -> 60[4] via P2P/IPC
+ip-26-0-172-73:926599:926895 [5] NCCL INFO Channel 00/0 : 53[5] -> 52[4] via P2P/IPC
+ip-26-0-172-116:1019293:1019592 [5] NCCL INFO Channel 08/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-172-252:688204:688502 [4] NCCL INFO Channel 03/0 : 36[4] -> 35[3] via P2P/IPC
+ip-26-0-160-225:94349:94645 [7] NCCL INFO Channel 04/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-173-7:811254:811557 [6] NCCL INFO Channel 08/0 : 62[6] -> 61[5] via P2P/IPC
+ip-26-0-172-116:1019292:1019590 [4] NCCL INFO Channel 00/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-173-7:811249:811559 [1] NCCL INFO Channel 08/0 : 57[1] -> 56[0] via P2P/IPC
+ip-26-0-160-225:94344:94648 [2] NCCL INFO Channel 01/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-225:94346:94644 [4] NCCL INFO Channel 00/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-147:618589:618883 [5] NCCL INFO Channel 04/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-173-7:811255:811561 [7] NCCL INFO Channel 04/0 : 63[7] -> 56[0] via P2P/IPC
+ip-26-0-172-147:618590:618889 [6] NCCL INFO Channel 08/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-172-252:688206:688506 [6] NCCL INFO Channel 08/0 : 38[6] -> 37[5] via P2P/IPC
+ip-26-0-173-7:811250:811555 [2] NCCL INFO Channel 01/0 : 58[2] -> 57[1] via P2P/IPC
+ip-26-0-173-7:811253:811558 [5] NCCL INFO Channel 12/0 : 61[5] -> 60[4] via P2P/IPC
+ip-26-0-172-147:618585:618884 [1] NCCL INFO Channel 08/0 : 25[1] -> 24[0] via P2P/IPC
+ip-26-0-172-252:688201:688500 [1] NCCL INFO Channel 08/0 : 33[1] -> 32[0] via P2P/IPC
+ip-26-0-172-73:926595:926897 [1] NCCL INFO Channel 08/0 : 49[1] -> 48[0] via P2P/IPC
+ip-26-0-172-73:926596:926899 [2] NCCL INFO Channel 08/0 : 50[2] -> 49[1] via P2P/IPC
+ip-26-0-172-73:926600:926893 [6] NCCL INFO Channel 13/0 : 54[6] -> 53[5] via P2P/IPC
+ip-26-0-172-142:880685:880977 [6] NCCL INFO Channel 05/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-172-57:942615:942912 [4] NCCL INFO Channel 06/0 : 44[4] -> 43[3] via P2P/IPC
+ip-26-0-160-225:94347:94646 [5] NCCL INFO Channel 12/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-57:942613:942917 [2] NCCL INFO Channel 00/0 : 42[2] -> 41[1] via P2P/IPC
+ip-26-0-172-252:688205:688505 [5] NCCL INFO Channel 08/0 : 37[5] -> 36[4] via P2P/IPC
+ip-26-0-172-147:618591:618885 [7] NCCL INFO Channel 05/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-172-57:942617:942916 [6] NCCL INFO Channel 00/0 : 46[6] -> 45[5] via P2P/IPC
+ip-26-0-172-147:618588:618888 [4] NCCL INFO Channel 06/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-173-7:811252:811560 [4] NCCL INFO Channel 06/0 : 60[4] -> 59[3] via P2P/IPC
+ip-26-0-172-142:880683:880982 [4] NCCL INFO Channel 07/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-172-73:926598:926894 [4] NCCL INFO Channel 07/0 : 52[4] -> 51[3] via P2P/IPC
+ip-26-0-172-147:618586:618886 [2] NCCL INFO Channel 08/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-172-252:688204:688502 [4] NCCL INFO Channel 05/0 : 36[4] -> 35[3] via P2P/IPC
+ip-26-0-173-7:811254:811557 [6] NCCL INFO Channel 13/0 : 62[6] -> 61[5] via P2P/IPC
+ip-26-0-172-57:942616:942913 [5] NCCL INFO Channel 04/0 : 45[5] -> 44[4] via P2P/IPC
+ip-26-0-160-225:94343:94647 [1] NCCL INFO Channel 08/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-142:880681:880983 [2] NCCL INFO Channel 08/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-172-142:880684:880981 [5] NCCL INFO Channel 12/0 : 21[5] -> 20[4] via P2P/IPC
+ip-26-0-172-73:926599:926895 [5] NCCL INFO Channel 04/0 : 53[5] -> 52[4] via P2P/IPC
+ip-26-0-172-142:880686:880979 [7] NCCL INFO Channel 07/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-172-252:688206:688506 [6] NCCL INFO Channel 13/0 : 38[6] -> 37[5] via P2P/IPC
+ip-26-0-160-225:94349:94645 [7] NCCL INFO Channel 05/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-142:880680:880984 [1] NCCL INFO Channel 08/0 : 17[1] -> 16[0] via P2P/IPC
+ip-26-0-160-225:94344:94648 [2] NCCL INFO Channel 08/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-225:94346:94644 [4] NCCL INFO Channel 03/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-73:926596:926899 [2] NCCL INFO Channel 09/0 : 50[2] -> 49[1] via P2P/IPC
+ip-26-0-172-57:942613:942917 [2] NCCL INFO Channel 01/0 : 42[2] -> 41[1] via P2P/IPC
+ip-26-0-172-57:942615:942912 [4] NCCL INFO Channel 07/0 : 44[4] -> 43[3] via P2P/IPC
+ip-26-0-172-57:942617:942916 [6] NCCL INFO Channel 05/0 : 46[6] -> 45[5] via P2P/IPC
+ip-26-0-172-147:618589:618883 [5] NCCL INFO Channel 08/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-172-57:942618:942918 [7] NCCL INFO Channel 02/0 : 47[7] -> 40[0] via P2P/IPC
+ip-26-0-172-147:618590:618889 [6] NCCL INFO Channel 13/0 : 30[6] -> 29[5] via P2P/IPC
+ip-26-0-172-252:688205:688505 [5] NCCL INFO Channel 12/0 : 37[5] -> 36[4] via P2P/IPC
+ip-26-0-172-116:1019294:1019589 [6] NCCL INFO Channel 00/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-172-142:880685:880977 [6] NCCL INFO Channel 08/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-172-252:688202:688503 [2] NCCL INFO Channel 08/0 : 34[2] -> 33[1] via P2P/IPC
+ip-26-0-172-73:926598:926894 [4] NCCL INFO Channel 08/0 : 52[4] -> 51[3] via P2P/IPC
+ip-26-0-172-73:926599:926895 [5] NCCL INFO Channel 08/0 : 53[5] -> 52[4] via P2P/IPC
+ip-26-0-172-252:688204:688502 [4] NCCL INFO Channel 06/0 : 36[4] -> 35[3] via P2P/IPC
+ip-26-0-172-73:926601:926896 [7] NCCL INFO Channel 05/0 : 55[7] -> 48[0] via P2P/IPC
+ip-26-0-172-147:618591:618885 [7] NCCL INFO Channel 06/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-172-57:942616:942913 [5] NCCL INFO Channel 08/0 : 45[5] -> 44[4] via P2P/IPC
+ip-26-0-172-147:618588:618888 [4] NCCL INFO Channel 07/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-172-147:618586:618886 [2] NCCL INFO Channel 09/0 : 26[2] -> 25[1] via P2P/IPC
+ip-26-0-172-116:1019293:1019592 [5] NCCL INFO Channel 12/0 : 13[5] -> 12[4] via P2P/IPC
+ip-26-0-160-225:94349:94645 [7] NCCL INFO Channel 06/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-225:94344:94648 [2] NCCL INFO Channel 09/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-225:94346:94644 [4] NCCL INFO Channel 05/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-142:880683:880982 [4] NCCL INFO Channel 08/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-172-142:880686:880979 [7] NCCL INFO Channel 09/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-172-116:1019294:1019589 [6] NCCL INFO Channel 05/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-173-7:811250:811555 [2] NCCL INFO Channel 08/0 : 58[2] -> 57[1] via P2P/IPC
+ip-26-0-172-73:926598:926894 [4] NCCL INFO Channel 11/0 : 52[4] -> 51[3] via P2P/IPC
+ip-26-0-172-57:942613:942917 [2] NCCL INFO Channel 08/0 : 42[2] -> 41[1] via P2P/IPC
+ip-26-0-172-57:942617:942916 [6] NCCL INFO Channel 08/0 : 46[6] -> 45[5] via P2P/IPC
+ip-26-0-172-142:880685:880977 [6] NCCL INFO Channel 13/0 : 22[6] -> 21[5] via P2P/IPC
+ip-26-0-172-73:926599:926895 [5] NCCL INFO Channel 12/0 : 53[5] -> 52[4] via P2P/IPC
+ip-26-0-172-73:926601:926896 [7] NCCL INFO Channel 06/0 : 55[7] -> 48[0] via P2P/IPC
+ip-26-0-172-252:688204:688502 [4] NCCL INFO Channel 07/0 : 36[4] -> 35[3] via P2P/IPC
+ip-26-0-172-57:942615:942912 [4] NCCL INFO Channel 08/0 : 44[4] -> 43[3] via P2P/IPC
+ip-26-0-172-57:942618:942918 [7] NCCL INFO Channel 03/0 : 47[7] -> 40[0] via P2P/IPC
+ip-26-0-172-147:618589:618883 [5] NCCL INFO Channel 12/0 : 29[5] -> 28[4] via P2P/IPC
+ip-26-0-172-252:688202:688503 [2] NCCL INFO Channel 09/0 : 34[2] -> 33[1] via P2P/IPC
+ip-26-0-173-7:811252:811560 [4] NCCL INFO Channel 07/0 : 60[4] -> 59[3] via P2P/IPC
+ip-26-0-172-57:942616:942913 [5] NCCL INFO Channel 12/0 : 45[5] -> 44[4] via P2P/IPC
+ip-26-0-173-7:811255:811561 [7] NCCL INFO Channel 05/0 : 63[7] -> 56[0] via P2P/IPC
+ip-26-0-160-225:94349:94645 [7] NCCL INFO Channel 07/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-225:94346:94644 [4] NCCL INFO Channel 06/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-173-7:811250:811555 [2] NCCL INFO Channel 09/0 : 58[2] -> 57[1] via P2P/IPC
+ip-26-0-172-142:880683:880982 [4] NCCL INFO Channel 11/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-172-142:880686:880979 [7] NCCL INFO Channel 10/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-172-116:1019294:1019589 [6] NCCL INFO Channel 08/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-172-147:618588:618888 [4] NCCL INFO Channel 08/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-172-252:688207:688504 [7] NCCL INFO Channel 06/0 : 39[7] -> 32[0] via P2P/IPC
+ip-26-0-172-116:1019295:1019591 [7] NCCL INFO Channel 06/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-172-252:688204:688502 [4] NCCL INFO Channel 08/0 : 36[4] -> 35[3] via P2P/IPC
+ip-26-0-172-73:926598:926894 [4] NCCL INFO Channel 13/0 : 52[4] -> 51[3] via P2P/IPC
+ip-26-0-172-147:618584:618882 [0] NCCL INFO Channel 08/0 : 24[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880681:880983 [2] NCCL INFO Channel 09/0 : 18[2] -> 17[1] via P2P/IPC
+ip-26-0-173-7:811252:811560 [4] NCCL INFO Channel 08/0 : 60[4] -> 59[3] via P2P/IPC
+ip-26-0-172-73:926601:926896 [7] NCCL INFO Channel 07/0 : 55[7] -> 48[0] via P2P/IPC
+ip-26-0-160-225:94349:94645 [7] NCCL INFO Channel 09/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-73:926597:926900 [3] NCCL INFO Channel 00/0 : 51[3] -> 50[2] via P2P/IPC
+ip-26-0-160-225:94346:94644 [4] NCCL INFO Channel 07/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-147:618588:618888 [4] NCCL INFO Channel 11/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-173-7:811255:811561 [7] NCCL INFO Channel 06/0 : 63[7] -> 56[0] via P2P/IPC
+ip-26-0-172-142:880683:880982 [4] NCCL INFO Channel 13/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-172-147:618584:618882 [0] NCCL INFO Channel 08/0 : 40[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-7:811248:811554 [0] NCCL INFO Channel 00/0 : 56[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019294:1019589 [6] NCCL INFO Channel 13/0 : 14[6] -> 13[5] via P2P/IPC
+ip-26-0-172-142:880686:880979 [7] NCCL INFO Channel 11/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-172-147:618584:618882 [0] NCCL INFO Channel 08/0 : 24[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942613:942917 [2] NCCL INFO Channel 09/0 : 42[2] -> 41[1] via P2P/IPC
+ip-26-0-172-57:942617:942916 [6] NCCL INFO Channel 13/0 : 46[6] -> 45[5] via P2P/IPC
+ip-26-0-172-57:942615:942912 [4] NCCL INFO Channel 11/0 : 44[4] -> 43[3] via P2P/IPC
+ip-26-0-172-57:942618:942918 [7] NCCL INFO Channel 04/0 : 47[7] -> 40[0] via P2P/IPC
+ip-26-0-172-73:926598:926894 [4] NCCL INFO Channel 14/0 : 52[4] -> 51[3] via P2P/IPC
+ip-26-0-172-73:926601:926896 [7] NCCL INFO Channel 09/0 : 55[7] -> 48[0] via P2P/IPC
+ip-26-0-172-116:1019295:1019591 [7] NCCL INFO Channel 07/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-172-116:1019290:1019595 [2] NCCL INFO Channel 00/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-172-57:942611:942911 [0] NCCL INFO Channel 00/0 : 48[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:94349:94645 [7] NCCL INFO Channel 10/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-173-7:811252:811560 [4] NCCL INFO Channel 11/0 : 60[4] -> 59[3] via P2P/IPC
+ip-26-0-172-147:618588:618888 [4] NCCL INFO Channel 13/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-160-225:94346:94644 [4] NCCL INFO Channel 08/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-57:942611:942911 [0] NCCL INFO Channel 08/0 : 48[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-7:811255:811561 [7] NCCL INFO Channel 07/0 : 63[7] -> 56[0] via P2P/IPC
+ip-26-0-160-225:94345:94649 [3] NCCL INFO Channel 00/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-57:942611:942911 [0] NCCL INFO Channel 08/0 : 40[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880683:880982 [4] NCCL INFO Channel 14/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-172-142:880686:880979 [7] NCCL INFO Channel 12/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-172-73:926598:926894 [4] NCCL INFO Channel 15/0 : 52[4] -> 51[3] via P2P/IPC
+ip-26-0-172-73:926601:926896 [7] NCCL INFO Channel 10/0 : 55[7] -> 48[0] via P2P/IPC
+ip-26-0-172-147:618587:618887 [3] NCCL INFO Channel 00/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-172-147:618588:618888 [4] NCCL INFO Channel 14/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-172-73:926597:926900 [3] NCCL INFO Channel 02/0 : 51[3] -> 50[2] via P2P/IPC
+ip-26-0-160-225:94349:94645 [7] NCCL INFO Channel 11/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-225:94346:94644 [4] NCCL INFO Channel 11/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-252:688207:688504 [7] NCCL INFO Channel 07/0 : 39[7] -> 32[0] via P2P/IPC
+ip-26-0-160-225:94345:94649 [3] NCCL INFO Channel 02/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-142:880683:880982 [4] NCCL INFO Channel 15/0 : 20[4] -> 19[3] via P2P/IPC
+ip-26-0-172-142:880686:880979 [7] NCCL INFO Channel 13/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-172-57:942615:942912 [4] NCCL INFO Channel 13/0 : 44[4] -> 43[3] via P2P/IPC
+ip-26-0-172-57:942618:942918 [7] NCCL INFO Channel 05/0 : 47[7] -> 40[0] via P2P/IPC
+ip-26-0-172-57:942614:942915 [3] NCCL INFO Channel 00/0 : 43[3] -> 42[2] via P2P/IPC
+ip-26-0-172-73:926601:926896 [7] NCCL INFO Channel 11/0 : 55[7] -> 48[0] via P2P/IPC
+ip-26-0-172-147:618587:618887 [3] NCCL INFO Channel 02/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-172-142:880682:880978 [3] NCCL INFO Channel 00/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-172-147:618588:618888 [4] NCCL INFO Channel 15/0 : 28[4] -> 27[3] via P2P/IPC
+ip-26-0-172-252:688207:688504 [7] NCCL INFO Channel 09/0 : 39[7] -> 32[0] via P2P/IPC
+ip-26-0-160-225:94349:94645 [7] NCCL INFO Channel 12/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-225:94346:94644 [4] NCCL INFO Channel 13/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-225:94345:94649 [3] NCCL INFO Channel 08/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-142:880686:880979 [7] NCCL INFO Channel 14/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-172-73:926597:926900 [3] NCCL INFO Channel 08/0 : 51[3] -> 50[2] via P2P/IPC
+ip-26-0-172-252:688207:688504 [7] NCCL INFO Channel 10/0 : 39[7] -> 32[0] via P2P/IPC
+ip-26-0-172-116:1019295:1019591 [7] NCCL INFO Channel 09/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-172-147:618587:618887 [3] NCCL INFO Channel 08/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-172-116:1019289:1019593 [1] NCCL INFO Channel 08/0 : 9[1] -> 8[0] via P2P/IPC
+ip-26-0-172-252:688204:688502 [4] NCCL INFO Channel 11/0 : 36[4] -> 35[3] via P2P/IPC
+ip-26-0-172-57:942615:942912 [4] NCCL INFO Channel 14/0 : 44[4] -> 43[3] via P2P/IPC
+ip-26-0-172-252:688207:688504 [7] NCCL INFO Channel 11/0 : 39[7] -> 32[0] via P2P/IPC
+ip-26-0-172-116:1019290:1019595 [2] NCCL INFO Channel 01/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-172-57:942618:942918 [7] NCCL INFO Channel 06/0 : 47[7] -> 40[0] via P2P/IPC
+ip-26-0-172-57:942614:942915 [3] NCCL INFO Channel 02/0 : 43[3] -> 42[2] via P2P/IPC
+ip-26-0-160-225:94349:94645 [7] NCCL INFO Channel 13/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-225:94346:94644 [4] NCCL INFO Channel 14/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-147:618584:618882 [0] NCCL INFO Channel 00/0 : 24[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019288:1019588 [0] NCCL INFO Channel 00/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688204:688502 [4] NCCL INFO Channel 13/0 : 36[4] -> 35[3] via P2P/IPC
+ip-26-0-172-116:1019292:1019590 [4] NCCL INFO Channel 03/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-172-252:688207:688504 [7] NCCL INFO Channel 12/0 : 39[7] -> 32[0] via P2P/IPC
+ip-26-0-172-116:1019288:1019588 [0] NCCL INFO Channel 08/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618591:618885 [7] NCCL INFO Channel 07/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-172-116:1019288:1019588 [0] NCCL INFO Channel 08/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942615:942912 [4] NCCL INFO Channel 15/0 : 44[4] -> 43[3] via P2P/IPC
+ip-26-0-172-116:1019295:1019591 [7] NCCL INFO Channel 10/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-172-57:942614:942915 [3] NCCL INFO Channel 08/0 : 43[3] -> 42[2] via P2P/IPC
+ip-26-0-172-57:942618:942918 [7] NCCL INFO Channel 07/0 : 47[7] -> 40[0] via P2P/IPC
+ip-26-0-172-252:688204:688502 [4] NCCL INFO Channel 14/0 : 36[4] -> 35[3] via P2P/IPC
+ip-26-0-173-7:811252:811560 [4] NCCL INFO Channel 13/0 : 60[4] -> 59[3] via P2P/IPC
+ip-26-0-172-252:688207:688504 [7] NCCL INFO Channel 13/0 : 39[7] -> 32[0] via P2P/IPC
+ip-26-0-172-116:1019291:1019594 [3] NCCL INFO Channel 00/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-173-7:811255:811561 [7] NCCL INFO Channel 09/0 : 63[7] -> 56[0] via P2P/IPC
+ip-26-0-172-116:1019290:1019595 [2] NCCL INFO Channel 08/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-172-73:926601:926896 [7] NCCL INFO Channel 12/0 : 55[7] -> 48[0] via P2P/IPC
+ip-26-0-173-7:811252:811560 [4] NCCL INFO Channel 14/0 : 60[4] -> 59[3] via P2P/IPC
+ip-26-0-172-147:618587:618887 [3] NCCL INFO Channel 10/0 : 27[3] -> 26[2] via P2P/IPC
+ip-26-0-160-225:94349:94645 [7] NCCL INFO Channel 14/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-57:942614:942915 [3] NCCL INFO Channel 10/0 : 43[3] -> 42[2] via P2P/IPC
+ip-26-0-172-57:942618:942918 [7] NCCL INFO Channel 09/0 : 47[7] -> 40[0] via P2P/IPC
+ip-26-0-172-252:688203:688501 [3] NCCL INFO Channel 00/0 : 35[3] -> 34[2] via P2P/IPC
+ip-26-0-160-225:94345:94649 [3] NCCL INFO Channel 10/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-116:1019290:1019595 [2] NCCL INFO Channel 09/0 : 10[2] -> 9[1] via P2P/IPC
+ip-26-0-172-252:688204:688502 [4] NCCL INFO Channel 15/0 : 36[4] -> 35[3] via P2P/IPC
+ip-26-0-160-225:94349:94645 [7] NCCL INFO Channel 15/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-252:688207:688504 [7] NCCL INFO Channel 14/0 : 39[7] -> 32[0] via P2P/IPC
+ip-26-0-173-7:811252:811560 [4] NCCL INFO Channel 15/0 : 60[4] -> 59[3] via P2P/IPC
+ip-26-0-172-252:688203:688501 [3] NCCL INFO Channel 02/0 : 35[3] -> 34[2] via P2P/IPC
+ip-26-0-172-142:880686:880979 [7] NCCL INFO Channel 15/0 : 23[7] -> 16[0] via P2P/IPC
+ip-26-0-172-116:1019295:1019591 [7] NCCL INFO Channel 11/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-172-142:880682:880978 [3] NCCL INFO Channel 02/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-172-252:688207:688504 [7] NCCL INFO Channel 15/0 : 39[7] -> 32[0] via P2P/IPC
+ip-26-0-172-73:926597:926900 [3] NCCL INFO Channel 10/0 : 51[3] -> 50[2] via P2P/IPC
+ip-26-0-172-116:1019292:1019590 [4] NCCL INFO Channel 05/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-172-147:618591:618885 [7] NCCL INFO Channel 09/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-173-7:811251:811556 [3] NCCL INFO Channel 00/0 : 59[3] -> 58[2] via P2P/IPC
+ip-26-0-160-225:94346:94644 [4] NCCL INFO Channel 15/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-173-7:811255:811561 [7] NCCL INFO Channel 10/0 : 63[7] -> 56[0] via P2P/IPC
+ip-26-0-172-116:1019295:1019591 [7] NCCL INFO Channel 12/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-172-73:926601:926896 [7] NCCL INFO Channel 13/0 : 55[7] -> 48[0] via P2P/IPC
+ip-26-0-172-116:1019292:1019590 [4] NCCL INFO Channel 06/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-172-116:1019295:1019591 [7] NCCL INFO Channel 13/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-172-116:1019292:1019590 [4] NCCL INFO Channel 07/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-172-116:1019291:1019594 [3] NCCL INFO Channel 02/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-172-57:942618:942918 [7] NCCL INFO Channel 10/0 : 47[7] -> 40[0] via P2P/IPC
+ip-26-0-172-116:1019295:1019591 [7] NCCL INFO Channel 14/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-172-252:688203:688501 [3] NCCL INFO Channel 08/0 : 35[3] -> 34[2] via P2P/IPC
+ip-26-0-172-116:1019292:1019590 [4] NCCL INFO Channel 08/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-172-116:1019291:1019594 [3] NCCL INFO Channel 08/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-172-142:880682:880978 [3] NCCL INFO Channel 08/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-173-7:811251:811556 [3] NCCL INFO Channel 02/0 : 59[3] -> 58[2] via P2P/IPC
+ip-26-0-173-7:811255:811561 [7] NCCL INFO Channel 11/0 : 63[7] -> 56[0] via P2P/IPC
+ip-26-0-172-116:1019295:1019591 [7] NCCL INFO Channel 15/0 : 15[7] -> 8[0] via P2P/IPC
+ip-26-0-172-116:1019292:1019590 [4] NCCL INFO Channel 11/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-172-116:1019291:1019594 [3] NCCL INFO Channel 10/0 : 11[3] -> 10[2] via P2P/IPC
+ip-26-0-172-116:1019292:1019590 [4] NCCL INFO Channel 13/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-172-116:1019292:1019590 [4] NCCL INFO Channel 14/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-172-116:1019292:1019590 [4] NCCL INFO Channel 15/0 : 12[4] -> 11[3] via P2P/IPC
+ip-26-0-172-252:688203:688501 [3] NCCL INFO Channel 10/0 : 35[3] -> 34[2] via P2P/IPC
+ip-26-0-172-142:880682:880978 [3] NCCL INFO Channel 10/0 : 19[3] -> 18[2] via P2P/IPC
+ip-26-0-173-7:811255:811561 [7] NCCL INFO Channel 12/0 : 63[7] -> 56[0] via P2P/IPC
+ip-26-0-173-7:811251:811556 [3] NCCL INFO Channel 08/0 : 59[3] -> 58[2] via P2P/IPC
+ip-26-0-172-147:618591:618885 [7] NCCL INFO Channel 10/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-172-57:942618:942918 [7] NCCL INFO Channel 11/0 : 47[7] -> 40[0] via P2P/IPC
+ip-26-0-172-73:926601:926896 [7] NCCL INFO Channel 14/0 : 55[7] -> 48[0] via P2P/IPC
+ip-26-0-173-7:811255:811561 [7] NCCL INFO Channel 13/0 : 63[7] -> 56[0] via P2P/IPC
+ip-26-0-173-7:811251:811556 [3] NCCL INFO Channel 10/0 : 59[3] -> 58[2] via P2P/IPC
+ip-26-0-172-57:942618:942918 [7] NCCL INFO Channel 12/0 : 47[7] -> 40[0] via P2P/IPC
+ip-26-0-172-73:926601:926896 [7] NCCL INFO Channel 15/0 : 55[7] -> 48[0] via P2P/IPC
+ip-26-0-173-7:811255:811561 [7] NCCL INFO Channel 14/0 : 63[7] -> 56[0] via P2P/IPC
+ip-26-0-172-57:942618:942918 [7] NCCL INFO Channel 13/0 : 47[7] -> 40[0] via P2P/IPC
+ip-26-0-172-147:618591:618885 [7] NCCL INFO Channel 11/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-173-7:811255:811561 [7] NCCL INFO Channel 15/0 : 63[7] -> 56[0] via P2P/IPC
+ip-26-0-172-57:942618:942918 [7] NCCL INFO Channel 14/0 : 47[7] -> 40[0] via P2P/IPC
+ip-26-0-172-147:618591:618885 [7] NCCL INFO Channel 12/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-172-57:942618:942918 [7] NCCL INFO Channel 15/0 : 47[7] -> 40[0] via P2P/IPC
+ip-26-0-172-147:618591:618885 [7] NCCL INFO Channel 13/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-172-116:1019295:1019591 [7] NCCL INFO Channel 00/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-172-116:1019295:1019591 [7] NCCL INFO Channel 06/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-172-147:618591:618885 [7] NCCL INFO Channel 14/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-172-116:1019295:1019591 [7] NCCL INFO Channel 08/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-172-147:618591:618885 [7] NCCL INFO Channel 15/0 : 31[7] -> 24[0] via P2P/IPC
+ip-26-0-172-116:1019295:1019591 [7] NCCL INFO Channel 14/0 : 15[7] -> 14[6] via P2P/IPC
+ip-26-0-173-7:811255:811561 [7] NCCL INFO Channel 00/0 : 63[7] -> 62[6] via P2P/IPC
+ip-26-0-172-252:688207:688504 [7] NCCL INFO Channel 00/0 : 39[7] -> 38[6] via P2P/IPC
+ip-26-0-172-57:942618:942918 [7] NCCL INFO Channel 00/0 : 47[7] -> 46[6] via P2P/IPC
+ip-26-0-173-7:811255:811561 [7] NCCL INFO Channel 06/0 : 63[7] -> 62[6] via P2P/IPC
+ip-26-0-172-147:618591:618885 [7] NCCL INFO Channel 00/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-172-73:926601:926896 [7] NCCL INFO Channel 00/0 : 55[7] -> 54[6] via P2P/IPC
+ip-26-0-160-225:94349:94645 [7] NCCL INFO Channel 00/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-252:688207:688504 [7] NCCL INFO Channel 06/0 : 39[7] -> 38[6] via P2P/IPC
+ip-26-0-172-142:880686:880979 [7] NCCL INFO Channel 00/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-173-7:811255:811561 [7] NCCL INFO Channel 08/0 : 63[7] -> 62[6] via P2P/IPC
+ip-26-0-160-225:94349:94645 [7] NCCL INFO Channel 06/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-57:942618:942918 [7] NCCL INFO Channel 06/0 : 47[7] -> 46[6] via P2P/IPC
+ip-26-0-172-147:618591:618885 [7] NCCL INFO Channel 06/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-172-142:880686:880979 [7] NCCL INFO Channel 06/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-172-252:688207:688504 [7] NCCL INFO Channel 08/0 : 39[7] -> 38[6] via P2P/IPC
+ip-26-0-172-57:942618:942918 [7] NCCL INFO Channel 08/0 : 47[7] -> 46[6] via P2P/IPC
+ip-26-0-172-73:926601:926896 [7] NCCL INFO Channel 06/0 : 55[7] -> 54[6] via P2P/IPC
+ip-26-0-172-147:618591:618885 [7] NCCL INFO Channel 08/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-160-225:94349:94645 [7] NCCL INFO Channel 08/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-73:926601:926896 [7] NCCL INFO Channel 08/0 : 55[7] -> 54[6] via P2P/IPC
+ip-26-0-172-142:880686:880979 [7] NCCL INFO Channel 08/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-173-7:811255:811561 [7] NCCL INFO Channel 14/0 : 63[7] -> 62[6] via P2P/IPC
+ip-26-0-172-252:688207:688504 [7] NCCL INFO Channel 14/0 : 39[7] -> 38[6] via P2P/IPC
+ip-26-0-160-225:94349:94645 [7] NCCL INFO Channel 14/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-57:942618:942918 [7] NCCL INFO Channel 14/0 : 47[7] -> 46[6] via P2P/IPC
+ip-26-0-172-147:618591:618885 [7] NCCL INFO Channel 14/0 : 31[7] -> 30[6] via P2P/IPC
+ip-26-0-172-73:926601:926896 [7] NCCL INFO Channel 14/0 : 55[7] -> 54[6] via P2P/IPC
+ip-26-0-172-142:880686:880979 [7] NCCL INFO Channel 14/0 : 23[7] -> 22[6] via P2P/IPC
+ip-26-0-173-7:811253:811558 [5] NCCL INFO Connected all trees
+ip-26-0-173-7:811253:811558 [5] NCCL INFO NVLS comm 0x9a582e0 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-225:94349:94645 [7] NCCL INFO Connected all trees
+ip-26-0-160-225:94349:94645 [7] NCCL INFO NVLS comm 0x8e4ac60 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-225:94343:94647 [1] NCCL INFO Connected all trees
+ip-26-0-160-225:94343:94647 [1] NCCL INFO NVLS comm 0x874a1e0 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-225:94345:94649 [3] NCCL INFO Connected all trees
+ip-26-0-173-7:811252:811560 [4] NCCL INFO Connected all trees
+ip-26-0-160-225:94345:94649 [3] NCCL INFO NVLS comm 0xa67f5a0 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-173-7:811252:811560 [4] NCCL INFO NVLS comm 0x9103640 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-173-7:811254:811557 [6] NCCL INFO Connected all trees
+ip-26-0-173-7:811254:811557 [6] NCCL INFO NVLS comm 0x8df0080 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-225:94347:94646 [5] NCCL INFO Connected all trees
+ip-26-0-160-225:94347:94646 [5] NCCL INFO NVLS comm 0x95790b0 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-225:94346:94644 [4] NCCL INFO Connected all trees
+ip-26-0-160-225:94346:94644 [4] NCCL INFO NVLS comm 0xa4e59b0 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-173-7:811251:811556 [3] NCCL INFO Connected all trees
+ip-26-0-173-7:811251:811556 [3] NCCL INFO NVLS comm 0x8fe5b80 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-173-7:811250:811555 [2] NCCL INFO Connected all trees
+ip-26-0-173-7:811250:811555 [2] NCCL INFO NVLS comm 0x99e6c20 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-225:94342:94643 [0] NCCL INFO Connected all trees
+ip-26-0-160-225:94342:94643 [0] NCCL INFO NVLS comm 0x9322400 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-173-7:811249:811559 [1] NCCL INFO Connected all trees
+ip-26-0-173-7:811249:811559 [1] NCCL INFO NVLS comm 0x8be9fa0 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-173-7:811248:811554 [0] NCCL INFO Connected all trees
+ip-26-0-173-7:811248:811554 [0] NCCL INFO NVLS comm 0xa5d9ba0 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-225:94348:94650 [6] NCCL INFO Connected all trees
+ip-26-0-160-225:94348:94650 [6] NCCL INFO NVLS comm 0x9c46720 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-225:94344:94648 [2] NCCL INFO Connected all trees
+ip-26-0-160-225:94344:94648 [2] NCCL INFO NVLS comm 0x949ee50 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-173-7:811255:811561 [7] NCCL INFO Connected all trees
+ip-26-0-173-7:811255:811561 [7] NCCL INFO NVLS comm 0xa0349b0 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-57:942615:942912 [4] NCCL INFO Connected all trees
+ip-26-0-172-57:942615:942912 [4] NCCL INFO NVLS comm 0xa394dc0 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-73:926598:926894 [4] NCCL INFO Connected all trees
+ip-26-0-172-73:926598:926894 [4] NCCL INFO NVLS comm 0x9ba3630 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-252:688205:688505 [5] NCCL INFO Connected all trees
+ip-26-0-172-252:688205:688505 [5] NCCL INFO NVLS comm 0x90c5ab0 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-73:926599:926895 [5] NCCL INFO Connected all trees
+ip-26-0-172-57:942614:942915 [3] NCCL INFO Connected all trees
+ip-26-0-172-73:926599:926895 [5] NCCL INFO NVLS comm 0xa251780 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-57:942614:942915 [3] NCCL INFO NVLS comm 0x8c1a590 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-252:688203:688501 [3] NCCL INFO Connected all trees
+ip-26-0-172-252:688203:688501 [3] NCCL INFO NVLS comm 0x8a56220 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-252:688202:688503 [2] NCCL INFO Connected all trees
+ip-26-0-172-252:688202:688503 [2] NCCL INFO NVLS comm 0x94e7bb0 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-57:942613:942917 [2] NCCL INFO Connected all trees
+ip-26-0-172-57:942613:942917 [2] NCCL INFO NVLS comm 0x9dacfa0 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-73:926597:926900 [3] NCCL INFO Connected all trees
+ip-26-0-172-73:926597:926900 [3] NCCL INFO NVLS comm 0x8f37a80 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-73:926596:926899 [2] NCCL INFO Connected all trees
+ip-26-0-172-73:926596:926899 [2] NCCL INFO NVLS comm 0x90ace20 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-252:688204:688502 [4] NCCL INFO Connected all trees
+ip-26-0-172-57:942616:942913 [5] NCCL INFO Connected all trees
+ip-26-0-172-252:688201:688500 [1] NCCL INFO Connected all trees
+ip-26-0-172-252:688204:688502 [4] NCCL INFO NVLS comm 0x9756c40 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-57:942616:942913 [5] NCCL INFO NVLS comm 0x89176f0 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-252:688201:688500 [1] NCCL INFO NVLS comm 0x96030f0 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-73:926601:926896 [7] NCCL INFO Connected all trees
+ip-26-0-172-73:926601:926896 [7] NCCL INFO NVLS comm 0x8839590 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-57:942617:942916 [6] NCCL INFO Connected all trees
+ip-26-0-172-73:926600:926893 [6] NCCL INFO Connected all trees
+ip-26-0-172-57:942617:942916 [6] NCCL INFO NVLS comm 0x99f7c90 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-73:926600:926893 [6] NCCL INFO NVLS comm 0xa25f780 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-252:688200:688507 [0] NCCL INFO Connected all trees
+ip-26-0-172-252:688200:688507 [0] NCCL INFO NVLS comm 0x965a660 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-73:926594:926898 [0] NCCL INFO Connected all trees
+ip-26-0-172-57:942612:942914 [1] NCCL INFO Connected all trees
+ip-26-0-172-73:926595:926897 [1] NCCL INFO Connected all trees
+ip-26-0-172-73:926594:926898 [0] NCCL INFO NVLS comm 0x9665ce0 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-57:942612:942914 [1] NCCL INFO NVLS comm 0x8de9ae0 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-73:926595:926897 [1] NCCL INFO NVLS comm 0x875b3b0 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-57:942611:942911 [0] NCCL INFO Connected all trees
+ip-26-0-172-57:942611:942911 [0] NCCL INFO NVLS comm 0xa02b330 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-57:942618:942918 [7] NCCL INFO Connected all trees
+ip-26-0-172-57:942618:942918 [7] NCCL INFO NVLS comm 0x92847f0 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-116:1019290:1019595 [2] NCCL INFO Connected all trees
+ip-26-0-172-116:1019290:1019595 [2] NCCL INFO NVLS comm 0xa2736a0 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-147:618588:618888 [4] NCCL INFO Connected all trees
+ip-26-0-172-147:618588:618888 [4] NCCL INFO NVLS comm 0x8ab30c0 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-147:618587:618887 [3] NCCL INFO Connected all trees
+ip-26-0-172-252:688207:688504 [7] NCCL INFO Connected all trees
+ip-26-0-172-147:618587:618887 [3] NCCL INFO NVLS comm 0x8dedc20 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-252:688207:688504 [7] NCCL INFO NVLS comm 0x91ccc70 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-147:618586:618886 [2] NCCL INFO Connected all trees
+ip-26-0-172-147:618586:618886 [2] NCCL INFO NVLS comm 0x8da0660 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-147:618589:618883 [5] NCCL INFO Connected all trees
+ip-26-0-172-116:1019291:1019594 [3] NCCL INFO Connected all trees
+ip-26-0-172-147:618589:618883 [5] NCCL INFO NVLS comm 0x9a820e0 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-116:1019291:1019594 [3] NCCL INFO NVLS comm 0xa5fbcf0 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-116:1019293:1019592 [5] NCCL INFO Connected all trees
+ip-26-0-172-116:1019293:1019592 [5] NCCL INFO NVLS comm 0xa68afb0 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-116:1019292:1019590 [4] NCCL INFO Connected all trees
+ip-26-0-172-142:880684:880981 [5] NCCL INFO Connected all trees
+ip-26-0-172-116:1019292:1019590 [4] NCCL INFO NVLS comm 0xa5e9420 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-142:880684:880981 [5] NCCL INFO NVLS comm 0x8d065a0 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-142:880681:880983 [2] NCCL INFO Connected all trees
+ip-26-0-172-252:688206:688506 [6] NCCL INFO Connected all trees
+ip-26-0-172-142:880681:880983 [2] NCCL INFO NVLS comm 0x92c70b0 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-252:688206:688506 [6] NCCL INFO NVLS comm 0xa6af5e0 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-142:880683:880982 [4] NCCL INFO Connected all trees
+ip-26-0-172-116:1019295:1019591 [7] NCCL INFO Connected all trees
+ip-26-0-172-116:1019288:1019588 [0] NCCL INFO Connected all trees
+ip-26-0-172-147:618590:618889 [6] NCCL INFO Connected all trees
+ip-26-0-172-142:880683:880982 [4] NCCL INFO NVLS comm 0x9ee12e0 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-116:1019295:1019591 [7] NCCL INFO NVLS comm 0x9c544f0 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-116:1019288:1019588 [0] NCCL INFO NVLS comm 0x8e0d8a0 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-147:618590:618889 [6] NCCL INFO NVLS comm 0xa40dfa0 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-147:618584:618882 [0] NCCL INFO Connected all trees
+ip-26-0-172-147:618591:618885 [7] NCCL INFO Connected all trees
+ip-26-0-172-142:880682:880978 [3] NCCL INFO Connected all trees
+ip-26-0-172-147:618584:618882 [0] NCCL INFO NVLS comm 0x9d5f5f0 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-147:618591:618885 [7] NCCL INFO NVLS comm 0x9627a20 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-142:880686:880979 [7] NCCL INFO Connected all trees
+ip-26-0-172-142:880682:880978 [3] NCCL INFO NVLS comm 0x9765da0 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-142:880686:880979 [7] NCCL INFO NVLS comm 0x99b27a0 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-225:94348:94650 [6] NCCL INFO Channel 01/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:94346:94644 [4] NCCL INFO Channel 01/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:94342:94643 [0] NCCL INFO Channel 01/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:94344:94648 [2] NCCL INFO Channel 01/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:94348:94650 [6] NCCL INFO Channel 03/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:94346:94644 [4] NCCL INFO Channel 03/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:94344:94648 [2] NCCL INFO Channel 03/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:94342:94643 [0] NCCL INFO Channel 03/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:94348:94650 [6] NCCL INFO Channel 05/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:94346:94644 [4] NCCL INFO Channel 05/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:94342:94643 [0] NCCL INFO Channel 05/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:94344:94648 [2] NCCL INFO Channel 05/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:94348:94650 [6] NCCL INFO Channel 07/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:94343:94647 [1] NCCL INFO Channel 01/0 : 1[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:94346:94644 [4] NCCL INFO Channel 07/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:94349:94645 [7] NCCL INFO Channel 01/0 : 7[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:94342:94643 [0] NCCL INFO Channel 07/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:94344:94648 [2] NCCL INFO Channel 07/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:94346:94644 [4] NCCL INFO Channel 09/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:94348:94650 [6] NCCL INFO Channel 09/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:94343:94647 [1] NCCL INFO Channel 03/0 : 1[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:94349:94645 [7] NCCL INFO Channel 03/0 : 7[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:94342:94643 [0] NCCL INFO Channel 09/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:94344:94648 [2] NCCL INFO Channel 09/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:94345:94649 [3] NCCL INFO Channel 01/0 : 3[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:94346:94644 [4] NCCL INFO Channel 11/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:94347:94646 [5] NCCL INFO Channel 01/0 : 5[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:94348:94650 [6] NCCL INFO Channel 11/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:94343:94647 [1] NCCL INFO Channel 05/0 : 1[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:94342:94643 [0] NCCL INFO Channel 11/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:94344:94648 [2] NCCL INFO Channel 11/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:94349:94645 [7] NCCL INFO Channel 05/0 : 7[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:94346:94644 [4] NCCL INFO Channel 13/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:94345:94649 [3] NCCL INFO Channel 03/0 : 3[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:94348:94650 [6] NCCL INFO Channel 13/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:94347:94646 [5] NCCL INFO Channel 03/0 : 5[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:94343:94647 [1] NCCL INFO Channel 07/0 : 1[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:94342:94643 [0] NCCL INFO Channel 13/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:94349:94645 [7] NCCL INFO Channel 07/0 : 7[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:94344:94648 [2] NCCL INFO Channel 13/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:94346:94644 [4] NCCL INFO Channel 15/0 : 4[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:94345:94649 [3] NCCL INFO Channel 05/0 : 3[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:94348:94650 [6] NCCL INFO Channel 15/0 : 6[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:94347:94646 [5] NCCL INFO Channel 05/0 : 5[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:94343:94647 [1] NCCL INFO Channel 11/0 : 1[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:94342:94643 [0] NCCL INFO Channel 15/0 : 0[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:94344:94648 [2] NCCL INFO Channel 15/0 : 2[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:94349:94645 [7] NCCL INFO Channel 09/0 : 7[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:94345:94649 [3] NCCL INFO Channel 07/0 : 3[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:94347:94646 [5] NCCL INFO Channel 07/0 : 5[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019289:1019593 [1] NCCL INFO Connected all trees
+ip-26-0-160-225:94343:94647 [1] NCCL INFO Channel 13/0 : 1[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:94349:94645 [7] NCCL INFO Channel 11/0 : 7[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:94345:94649 [3] NCCL INFO Channel 09/0 : 3[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:94347:94646 [5] NCCL INFO Channel 09/0 : 5[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:94343:94647 [1] NCCL INFO Channel 15/0 : 1[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:94349:94645 [7] NCCL INFO Channel 13/0 : 7[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:94345:94649 [3] NCCL INFO Channel 13/0 : 3[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:94347:94646 [5] NCCL INFO Channel 11/0 : 5[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:94345:94649 [3] NCCL INFO Channel 15/0 : 3[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:94347:94646 [5] NCCL INFO Channel 15/0 : 5[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019289:1019593 [1] NCCL INFO NVLS comm 0x8887c60 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-142:880679:880980 [0] NCCL INFO Connected all trees
+ip-26-0-172-116:1019294:1019589 [6] NCCL INFO Connected all trees
+ip-26-0-172-142:880679:880980 [0] NCCL INFO NVLS comm 0xa3a4d80 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-116:1019294:1019589 [6] NCCL INFO NVLS comm 0x8886e00 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-142:880685:880977 [6] NCCL INFO Connected all trees
+ip-26-0-172-142:880685:880977 [6] NCCL INFO NVLS comm 0x8bbe030 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-173-7:811253:811558 [5] NCCL INFO Channel 00/0 : 53[5] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-7:811253:811558 [5] NCCL INFO Channel 02/0 : 53[5] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-7:811254:811557 [6] NCCL INFO Channel 00/0 : 54[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-7:811253:811558 [5] NCCL INFO Channel 04/0 : 53[5] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-7:811252:811560 [4] NCCL INFO Channel 00/0 : 52[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-7:811253:811558 [5] NCCL INFO Channel 06/0 : 53[5] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-7:811254:811557 [6] NCCL INFO Channel 02/0 : 54[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-7:811248:811554 [0] NCCL INFO Channel 02/0 : 48[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-7:811253:811558 [5] NCCL INFO Channel 08/0 : 53[5] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-7:811255:811561 [7] NCCL INFO Channel 00/0 : 55[7] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-7:811248:811554 [0] NCCL INFO Channel 04/0 : 48[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-7:811252:811560 [4] NCCL INFO Channel 02/0 : 52[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-7:811254:811557 [6] NCCL INFO Channel 04/0 : 54[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-7:811251:811556 [3] NCCL INFO Channel 00/0 : 51[3] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-7:811249:811559 [1] NCCL INFO Channel 00/0 : 49[1] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-7:811253:811558 [5] NCCL INFO Channel 10/0 : 53[5] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-7:811255:811561 [7] NCCL INFO Channel 02/0 : 55[7] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-7:811250:811555 [2] NCCL INFO Channel 00/0 : 50[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-7:811248:811554 [0] NCCL INFO Channel 06/0 : 48[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-7:811254:811557 [6] NCCL INFO Channel 08/0 : 54[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-7:811252:811560 [4] NCCL INFO Channel 06/0 : 52[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-7:811251:811556 [3] NCCL INFO Channel 02/0 : 51[3] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-7:811253:811558 [5] NCCL INFO Channel 12/0 : 53[5] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-7:811249:811559 [1] NCCL INFO Channel 02/0 : 49[1] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-7:811255:811561 [7] NCCL INFO Channel 04/0 : 55[7] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-7:811250:811555 [2] NCCL INFO Channel 04/0 : 50[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-7:811248:811554 [0] NCCL INFO Channel 08/0 : 48[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-7:811252:811560 [4] NCCL INFO Channel 08/0 : 52[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-7:811254:811557 [6] NCCL INFO Channel 10/0 : 54[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-7:811253:811558 [5] NCCL INFO Channel 14/0 : 53[5] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-7:811251:811556 [3] NCCL INFO Channel 04/0 : 51[3] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-7:811249:811559 [1] NCCL INFO Channel 04/0 : 49[1] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-7:811255:811561 [7] NCCL INFO Channel 06/0 : 55[7] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-7:811250:811555 [2] NCCL INFO Channel 06/0 : 50[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-7:811248:811554 [0] NCCL INFO Channel 10/0 : 48[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-7:811252:811560 [4] NCCL INFO Channel 10/0 : 52[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-7:811254:811557 [6] NCCL INFO Channel 12/0 : 54[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-7:811249:811559 [1] NCCL INFO Channel 06/0 : 49[1] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-7:811251:811556 [3] NCCL INFO Channel 06/0 : 51[3] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-7:811255:811561 [7] NCCL INFO Channel 08/0 : 55[7] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-7:811250:811555 [2] NCCL INFO Channel 08/0 : 50[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-7:811248:811554 [0] NCCL INFO Channel 12/0 : 48[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-7:811252:811560 [4] NCCL INFO Channel 12/0 : 52[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-7:811254:811557 [6] NCCL INFO Channel 14/0 : 54[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-7:811249:811559 [1] NCCL INFO Channel 08/0 : 49[1] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-7:811251:811556 [3] NCCL INFO Channel 08/0 : 51[3] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-7:811255:811561 [7] NCCL INFO Channel 10/0 : 55[7] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-7:811250:811555 [2] NCCL INFO Channel 10/0 : 50[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-7:811248:811554 [0] NCCL INFO Channel 14/0 : 48[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-7:811252:811560 [4] NCCL INFO Channel 14/0 : 52[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-7:811249:811559 [1] NCCL INFO Channel 10/0 : 49[1] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-7:811251:811556 [3] NCCL INFO Channel 10/0 : 51[3] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-7:811255:811561 [7] NCCL INFO Channel 12/0 : 55[7] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-7:811250:811555 [2] NCCL INFO Channel 12/0 : 50[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-7:811249:811559 [1] NCCL INFO Channel 12/0 : 49[1] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-7:811251:811556 [3] NCCL INFO Channel 12/0 : 51[3] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-7:811255:811561 [7] NCCL INFO Channel 14/0 : 55[7] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-7:811250:811555 [2] NCCL INFO Channel 14/0 : 50[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-7:811249:811559 [1] NCCL INFO Channel 14/0 : 49[1] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-7:811251:811556 [3] NCCL INFO Channel 14/0 : 51[3] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618585:618884 [1] NCCL INFO Connected all trees
+ip-26-0-172-147:618585:618884 [1] NCCL INFO NVLS comm 0xa2724f0 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-142:880680:880984 [1] NCCL INFO Connected all trees
+ip-26-0-172-142:880680:880984 [1] NCCL INFO NVLS comm 0x91fe360 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-73:926599:926895 [5] NCCL INFO Channel 00/0 : 45[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926595:926897 [1] NCCL INFO Channel 00/0 : 41[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926597:926900 [3] NCCL INFO Channel 00/0 : 43[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926601:926896 [7] NCCL INFO Channel 00/0 : 47[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926599:926895 [5] NCCL INFO Channel 01/0 : 45[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926597:926900 [3] NCCL INFO Channel 01/0 : 43[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926601:926896 [7] NCCL INFO Channel 01/0 : 47[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926599:926895 [5] NCCL INFO Channel 02/0 : 45[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926595:926897 [1] NCCL INFO Channel 02/0 : 41[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926597:926900 [3] NCCL INFO Channel 02/0 : 43[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926594:926898 [0] NCCL INFO Channel 01/0 : 40[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926601:926896 [7] NCCL INFO Channel 02/0 : 47[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926599:926895 [5] NCCL INFO Channel 03/0 : 45[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926595:926897 [1] NCCL INFO Channel 03/0 : 41[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926598:926894 [4] NCCL INFO Channel 00/0 : 44[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926600:926893 [6] NCCL INFO Channel 00/0 : 46[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926597:926900 [3] NCCL INFO Channel 04/0 : 43[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926599:926895 [5] NCCL INFO Channel 04/0 : 45[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926595:926897 [1] NCCL INFO Channel 04/0 : 41[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926594:926898 [0] NCCL INFO Channel 02/0 : 40[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926601:926896 [7] NCCL INFO Channel 03/0 : 47[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926596:926899 [2] NCCL INFO Channel 00/0 : 42[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926598:926894 [4] NCCL INFO Channel 01/0 : 44[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926600:926893 [6] NCCL INFO Channel 01/0 : 46[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926597:926900 [3] NCCL INFO Channel 05/0 : 43[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926599:926895 [5] NCCL INFO Channel 06/0 : 45[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926595:926897 [1] NCCL INFO Channel 05/0 : 41[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926594:926898 [0] NCCL INFO Channel 03/0 : 40[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926601:926896 [7] NCCL INFO Channel 04/0 : 47[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926596:926899 [2] NCCL INFO Channel 01/0 : 42[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926597:926900 [3] NCCL INFO Channel 06/0 : 43[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926600:926893 [6] NCCL INFO Channel 02/0 : 46[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926598:926894 [4] NCCL INFO Channel 02/0 : 44[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926595:926897 [1] NCCL INFO Channel 06/0 : 41[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926599:926895 [5] NCCL INFO Channel 07/0 : 45[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926594:926898 [0] NCCL INFO Channel 04/0 : 40[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926601:926896 [7] NCCL INFO Channel 05/0 : 47[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926600:926893 [6] NCCL INFO Channel 03/0 : 46[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926596:926899 [2] NCCL INFO Channel 03/0 : 42[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926597:926900 [3] NCCL INFO Channel 07/0 : 43[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926598:926894 [4] NCCL INFO Channel 03/0 : 44[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926595:926897 [1] NCCL INFO Channel 07/0 : 41[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926599:926895 [5] NCCL INFO Channel 08/0 : 45[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926594:926898 [0] NCCL INFO Channel 05/0 : 40[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926601:926896 [7] NCCL INFO Channel 06/0 : 47[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926596:926899 [2] NCCL INFO Channel 04/0 : 42[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926597:926900 [3] NCCL INFO Channel 08/0 : 43[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926600:926893 [6] NCCL INFO Channel 04/0 : 46[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926595:926897 [1] NCCL INFO Channel 08/0 : 41[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926598:926894 [4] NCCL INFO Channel 05/0 : 44[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926599:926895 [5] NCCL INFO Channel 09/0 : 45[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926601:926896 [7] NCCL INFO Channel 08/0 : 47[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926594:926898 [0] NCCL INFO Channel 06/0 : 40[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926596:926899 [2] NCCL INFO Channel 05/0 : 42[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926597:926900 [3] NCCL INFO Channel 09/0 : 43[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926598:926894 [4] NCCL INFO Channel 06/0 : 44[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926595:926897 [1] NCCL INFO Channel 10/0 : 41[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926600:926893 [6] NCCL INFO Channel 05/0 : 46[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926601:926896 [7] NCCL INFO Channel 09/0 : 47[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926599:926895 [5] NCCL INFO Channel 10/0 : 45[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926594:926898 [0] NCCL INFO Channel 07/0 : 40[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926596:926899 [2] NCCL INFO Channel 06/0 : 42[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926597:926900 [3] NCCL INFO Channel 10/0 : 43[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926598:926894 [4] NCCL INFO Channel 07/0 : 44[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926595:926897 [1] NCCL INFO Channel 11/0 : 41[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926601:926896 [7] NCCL INFO Channel 10/0 : 47[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926594:926898 [0] NCCL INFO Channel 09/0 : 40[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926596:926899 [2] NCCL INFO Channel 07/0 : 42[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926599:926895 [5] NCCL INFO Channel 11/0 : 45[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926597:926900 [3] NCCL INFO Channel 12/0 : 43[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926600:926893 [6] NCCL INFO Channel 07/0 : 46[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926598:926894 [4] NCCL INFO Channel 08/0 : 44[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926595:926897 [1] NCCL INFO Channel 12/0 : 41[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926594:926898 [0] NCCL INFO Channel 10/0 : 40[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926596:926899 [2] NCCL INFO Channel 08/0 : 42[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926597:926900 [3] NCCL INFO Channel 13/0 : 43[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926599:926895 [5] NCCL INFO Channel 12/0 : 45[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926598:926894 [4] NCCL INFO Channel 09/0 : 44[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926601:926896 [7] NCCL INFO Channel 11/0 : 47[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926595:926897 [1] NCCL INFO Channel 13/0 : 41[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926600:926893 [6] NCCL INFO Channel 08/0 : 46[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926594:926898 [0] NCCL INFO Channel 11/0 : 40[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926597:926900 [3] NCCL INFO Channel 14/0 : 43[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926596:926899 [2] NCCL INFO Channel 09/0 : 42[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926595:926897 [1] NCCL INFO Channel 14/0 : 41[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926599:926895 [5] NCCL INFO Channel 14/0 : 45[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926594:926898 [0] NCCL INFO Channel 12/0 : 40[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942611:942911 [0] NCCL INFO Channel 01/0 : 32[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926601:926896 [7] NCCL INFO Channel 12/0 : 47[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926598:926894 [4] NCCL INFO Channel 10/0 : 44[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926600:926893 [6] NCCL INFO Channel 09/0 : 46[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942617:942916 [6] NCCL INFO Channel 01/0 : 38[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926597:926900 [3] NCCL INFO Channel 15/0 : 43[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926596:926899 [2] NCCL INFO Channel 11/0 : 42[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926595:926897 [1] NCCL INFO Channel 15/0 : 41[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942611:942911 [0] NCCL INFO Channel 03/0 : 32[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942615:942912 [4] NCCL INFO Channel 01/0 : 36[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942617:942916 [6] NCCL INFO Channel 03/0 : 38[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926599:926895 [5] NCCL INFO Channel 15/0 : 45[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926594:926898 [0] NCCL INFO Channel 13/0 : 40[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926601:926896 [7] NCCL INFO Channel 13/0 : 47[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926598:926894 [4] NCCL INFO Channel 11/0 : 44[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942611:942911 [0] NCCL INFO Channel 05/0 : 32[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926596:926899 [2] NCCL INFO Channel 12/0 : 42[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926597:926900 [3] NCCL INFO Channel 00/0 : 51[3] -> 59[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926595:926897 [1] NCCL INFO Channel 00/0 : 49[1] -> 57[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926600:926893 [6] NCCL INFO Channel 10/0 : 46[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926599:926895 [5] NCCL INFO Channel 00/0 : 53[5] -> 61[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926594:926898 [0] NCCL INFO Channel 14/0 : 40[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942615:942912 [4] NCCL INFO Channel 03/0 : 36[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942617:942916 [6] NCCL INFO Channel 05/0 : 38[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942613:942917 [2] NCCL INFO Channel 01/0 : 34[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926601:926896 [7] NCCL INFO Channel 14/0 : 47[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926598:926894 [4] NCCL INFO Channel 13/0 : 44[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942611:942911 [0] NCCL INFO Channel 07/0 : 32[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926597:926900 [3] NCCL INFO Channel 02/0 : 51[3] -> 59[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926595:926897 [1] NCCL INFO Channel 02/0 : 49[1] -> 57[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926596:926899 [2] NCCL INFO Channel 13/0 : 42[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926599:926895 [5] NCCL INFO Channel 02/0 : 53[5] -> 61[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926600:926893 [6] NCCL INFO Channel 11/0 : 46[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942615:942912 [4] NCCL INFO Channel 05/0 : 36[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942617:942916 [6] NCCL INFO Channel 07/0 : 38[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926594:926898 [0] NCCL INFO Channel 15/0 : 40[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926601:926896 [7] NCCL INFO Channel 00/0 : 55[7] -> 63[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926598:926894 [4] NCCL INFO Channel 14/0 : 44[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942613:942917 [2] NCCL INFO Channel 03/0 : 34[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942616:942913 [5] NCCL INFO Channel 01/0 : 37[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942611:942911 [0] NCCL INFO Channel 09/0 : 32[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942614:942915 [3] NCCL INFO Channel 01/0 : 35[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926595:926897 [1] NCCL INFO Channel 04/0 : 49[1] -> 57[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926596:926899 [2] NCCL INFO Channel 14/0 : 42[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942618:942918 [7] NCCL INFO Channel 01/0 : 39[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942612:942914 [1] NCCL INFO Channel 01/0 : 33[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926597:926900 [3] NCCL INFO Channel 04/0 : 51[3] -> 59[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926594:926898 [0] NCCL INFO Channel 02/0 : 48[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926599:926895 [5] NCCL INFO Channel 04/0 : 53[5] -> 61[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926600:926893 [6] NCCL INFO Channel 12/0 : 46[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926598:926894 [4] NCCL INFO Channel 15/0 : 44[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926601:926896 [7] NCCL INFO Channel 02/0 : 55[7] -> 63[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926595:926897 [1] NCCL INFO Channel 06/0 : 49[1] -> 57[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926596:926899 [2] NCCL INFO Channel 15/0 : 42[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942615:942912 [4] NCCL INFO Channel 07/0 : 36[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942617:942916 [6] NCCL INFO Channel 09/0 : 38[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942613:942917 [2] NCCL INFO Channel 05/0 : 34[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942611:942911 [0] NCCL INFO Channel 11/0 : 32[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942616:942913 [5] NCCL INFO Channel 03/0 : 37[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942614:942915 [3] NCCL INFO Channel 03/0 : 35[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942618:942918 [7] NCCL INFO Channel 03/0 : 39[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926597:926900 [3] NCCL INFO Channel 06/0 : 51[3] -> 59[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926594:926898 [0] NCCL INFO Channel 04/0 : 48[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926599:926895 [5] NCCL INFO Channel 06/0 : 53[5] -> 61[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942612:942914 [1] NCCL INFO Channel 03/0 : 33[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942617:942916 [6] NCCL INFO Channel 11/0 : 38[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942613:942917 [2] NCCL INFO Channel 07/0 : 34[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926601:926896 [7] NCCL INFO Channel 04/0 : 55[7] -> 63[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926595:926897 [1] NCCL INFO Channel 08/0 : 49[1] -> 57[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926596:926899 [2] NCCL INFO Channel 00/0 : 50[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926600:926893 [6] NCCL INFO Channel 13/0 : 46[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926598:926894 [4] NCCL INFO Channel 00/0 : 52[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926597:926900 [3] NCCL INFO Channel 08/0 : 51[3] -> 59[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942611:942911 [0] NCCL INFO Channel 13/0 : 32[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926594:926898 [0] NCCL INFO Channel 06/0 : 48[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926599:926895 [5] NCCL INFO Channel 08/0 : 53[5] -> 61[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942614:942915 [3] NCCL INFO Channel 05/0 : 35[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942616:942913 [5] NCCL INFO Channel 05/0 : 37[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942612:942914 [1] NCCL INFO Channel 05/0 : 33[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942615:942912 [4] NCCL INFO Channel 09/0 : 36[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942618:942918 [7] NCCL INFO Channel 05/0 : 39[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942617:942916 [6] NCCL INFO Channel 13/0 : 38[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926595:926897 [1] NCCL INFO Channel 10/0 : 49[1] -> 57[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942613:942917 [2] NCCL INFO Channel 09/0 : 34[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926596:926899 [2] NCCL INFO Channel 04/0 : 50[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942611:942911 [0] NCCL INFO Channel 15/0 : 32[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926601:926896 [7] NCCL INFO Channel 06/0 : 55[7] -> 63[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926597:926900 [3] NCCL INFO Channel 10/0 : 51[3] -> 59[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926598:926894 [4] NCCL INFO Channel 02/0 : 52[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926600:926893 [6] NCCL INFO Channel 15/0 : 46[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926599:926895 [5] NCCL INFO Channel 10/0 : 53[5] -> 61[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926594:926898 [0] NCCL INFO Channel 08/0 : 48[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942614:942915 [3] NCCL INFO Channel 07/0 : 35[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926595:926897 [1] NCCL INFO Channel 12/0 : 49[1] -> 57[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926596:926899 [2] NCCL INFO Channel 06/0 : 50[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926601:926896 [7] NCCL INFO Channel 08/0 : 55[7] -> 63[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942612:942914 [1] NCCL INFO Channel 07/0 : 33[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942616:942913 [5] NCCL INFO Channel 07/0 : 37[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942613:942917 [2] NCCL INFO Channel 11/0 : 34[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942611:942911 [0] NCCL INFO Channel 01/0 : 40[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942617:942916 [6] NCCL INFO Channel 15/0 : 38[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942615:942912 [4] NCCL INFO Channel 11/0 : 36[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942618:942918 [7] NCCL INFO Channel 07/0 : 39[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926597:926900 [3] NCCL INFO Channel 12/0 : 51[3] -> 59[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926598:926894 [4] NCCL INFO Channel 06/0 : 52[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926594:926898 [0] NCCL INFO Channel 10/0 : 48[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926600:926893 [6] NCCL INFO Channel 00/0 : 54[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926599:926895 [5] NCCL INFO Channel 12/0 : 53[5] -> 61[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926595:926897 [1] NCCL INFO Channel 14/0 : 49[1] -> 57[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926596:926899 [2] NCCL INFO Channel 08/0 : 50[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926601:926896 [7] NCCL INFO Channel 10/0 : 55[7] -> 63[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942614:942915 [3] NCCL INFO Channel 09/0 : 35[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942612:942914 [1] NCCL INFO Channel 11/0 : 33[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942616:942913 [5] NCCL INFO Channel 09/0 : 37[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926597:926900 [3] NCCL INFO Channel 14/0 : 51[3] -> 59[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926598:926894 [4] NCCL INFO Channel 08/0 : 52[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926594:926898 [0] NCCL INFO Channel 12/0 : 48[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926600:926893 [6] NCCL INFO Channel 02/0 : 54[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926599:926895 [5] NCCL INFO Channel 14/0 : 53[5] -> 61[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942611:942911 [0] NCCL INFO Channel 02/0 : 40[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942617:942916 [6] NCCL INFO Channel 00/0 : 46[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942615:942912 [4] NCCL INFO Channel 13/0 : 36[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942613:942917 [2] NCCL INFO Channel 13/0 : 34[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926596:926899 [2] NCCL INFO Channel 10/0 : 50[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926601:926896 [7] NCCL INFO Channel 12/0 : 55[7] -> 63[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926598:926894 [4] NCCL INFO Channel 10/0 : 52[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942614:942915 [3] NCCL INFO Channel 13/0 : 35[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926594:926898 [0] NCCL INFO Channel 14/0 : 48[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942618:942918 [7] NCCL INFO Channel 09/0 : 39[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942612:942914 [1] NCCL INFO Channel 13/0 : 33[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942616:942913 [5] NCCL INFO Channel 11/0 : 37[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942611:942911 [0] NCCL INFO Channel 03/0 : 40[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942615:942912 [4] NCCL INFO Channel 15/0 : 36[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926600:926893 [6] NCCL INFO Channel 04/0 : 54[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926596:926899 [2] NCCL INFO Channel 12/0 : 50[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926601:926896 [7] NCCL INFO Channel 14/0 : 55[7] -> 63[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926598:926894 [4] NCCL INFO Channel 12/0 : 52[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942617:942916 [6] NCCL INFO Channel 01/0 : 46[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942614:942915 [3] NCCL INFO Channel 15/0 : 35[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926600:926893 [6] NCCL INFO Channel 08/0 : 54[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926598:926894 [4] NCCL INFO Channel 14/0 : 52[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926596:926899 [2] NCCL INFO Channel 14/0 : 50[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926600:926893 [6] NCCL INFO Channel 10/0 : 54[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942611:942911 [0] NCCL INFO Channel 04/0 : 40[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942618:942918 [7] NCCL INFO Channel 11/0 : 39[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942613:942917 [2] NCCL INFO Channel 15/0 : 34[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942616:942913 [5] NCCL INFO Channel 15/0 : 37[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942615:942912 [4] NCCL INFO Channel 00/0 : 44[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942612:942914 [1] NCCL INFO Channel 15/0 : 33[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942617:942916 [6] NCCL INFO Channel 02/0 : 46[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926600:926893 [6] NCCL INFO Channel 12/0 : 54[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942614:942915 [3] NCCL INFO Channel 00/0 : 43[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942611:942911 [0] NCCL INFO Channel 05/0 : 40[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926600:926893 [6] NCCL INFO Channel 14/0 : 54[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942613:942917 [2] NCCL INFO Channel 00/0 : 42[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942615:942912 [4] NCCL INFO Channel 01/0 : 44[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942617:942916 [6] NCCL INFO Channel 03/0 : 46[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942616:942913 [5] NCCL INFO Channel 00/0 : 45[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942618:942918 [7] NCCL INFO Channel 13/0 : 39[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942612:942914 [1] NCCL INFO Channel 00/0 : 41[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942611:942911 [0] NCCL INFO Channel 06/0 : 40[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-7:811251:811556 [3] NCCL INFO Channel 01/0 : 27[3] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942614:942915 [3] NCCL INFO Channel 01/0 : 43[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942613:942917 [2] NCCL INFO Channel 01/0 : 42[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-7:811253:811558 [5] NCCL INFO Channel 01/0 : 29[5] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-7:811249:811559 [1] NCCL INFO Channel 01/0 : 25[1] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942617:942916 [6] NCCL INFO Channel 04/0 : 46[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942615:942912 [4] NCCL INFO Channel 02/0 : 44[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942616:942913 [5] NCCL INFO Channel 01/0 : 45[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942612:942914 [1] NCCL INFO Channel 02/0 : 41[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942611:942911 [0] NCCL INFO Channel 07/0 : 40[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942618:942918 [7] NCCL INFO Channel 00/0 : 47[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-7:811251:811556 [3] NCCL INFO Channel 03/0 : 27[3] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-7:811248:811554 [0] NCCL INFO Channel 01/0 : 24[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942613:942917 [2] NCCL INFO Channel 03/0 : 42[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942614:942915 [3] NCCL INFO Channel 02/0 : 43[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942617:942916 [6] NCCL INFO Channel 05/0 : 46[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-7:811253:811558 [5] NCCL INFO Channel 03/0 : 29[5] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-7:811249:811559 [1] NCCL INFO Channel 03/0 : 25[1] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-7:811255:811561 [7] NCCL INFO Channel 01/0 : 31[7] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942615:942912 [4] NCCL INFO Channel 03/0 : 44[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942612:942914 [1] NCCL INFO Channel 03/0 : 41[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942611:942911 [0] NCCL INFO Channel 09/0 : 40[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-7:811248:811554 [0] NCCL INFO Channel 03/0 : 24[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-7:811250:811555 [2] NCCL INFO Channel 01/0 : 26[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-7:811251:811556 [3] NCCL INFO Channel 05/0 : 27[3] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942616:942913 [5] NCCL INFO Channel 02/0 : 45[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942614:942915 [3] NCCL INFO Channel 04/0 : 43[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942618:942918 [7] NCCL INFO Channel 01/0 : 47[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942613:942917 [2] NCCL INFO Channel 04/0 : 42[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942617:942916 [6] NCCL INFO Channel 07/0 : 46[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942612:942914 [1] NCCL INFO Channel 04/0 : 41[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-7:811253:811558 [5] NCCL INFO Channel 05/0 : 29[5] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-7:811249:811559 [1] NCCL INFO Channel 05/0 : 25[1] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-7:811248:811554 [0] NCCL INFO Channel 05/0 : 24[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-7:811255:811561 [7] NCCL INFO Channel 03/0 : 31[7] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-7:811252:811560 [4] NCCL INFO Channel 01/0 : 28[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942615:942912 [4] NCCL INFO Channel 05/0 : 44[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942611:942911 [0] NCCL INFO Channel 10/0 : 40[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942616:942913 [5] NCCL INFO Channel 03/0 : 45[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-7:811250:811555 [2] NCCL INFO Channel 03/0 : 26[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-7:811251:811556 [3] NCCL INFO Channel 07/0 : 27[3] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942613:942917 [2] NCCL INFO Channel 05/0 : 42[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942614:942915 [3] NCCL INFO Channel 05/0 : 43[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942617:942916 [6] NCCL INFO Channel 08/0 : 46[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942618:942918 [7] NCCL INFO Channel 02/0 : 47[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942612:942914 [1] NCCL INFO Channel 05/0 : 41[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942611:942911 [0] NCCL INFO Channel 11/0 : 40[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942615:942912 [4] NCCL INFO Channel 06/0 : 44[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-7:811249:811559 [1] NCCL INFO Channel 07/0 : 25[1] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-7:811253:811558 [5] NCCL INFO Channel 07/0 : 29[5] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942616:942913 [5] NCCL INFO Channel 04/0 : 45[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-7:811248:811554 [0] NCCL INFO Channel 07/0 : 24[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-7:811254:811557 [6] NCCL INFO Channel 01/0 : 30[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-7:811255:811561 [7] NCCL INFO Channel 05/0 : 31[7] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-7:811252:811560 [4] NCCL INFO Channel 03/0 : 28[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-7:811250:811555 [2] NCCL INFO Channel 05/0 : 26[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-7:811251:811556 [3] NCCL INFO Channel 09/0 : 27[3] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942613:942917 [2] NCCL INFO Channel 06/0 : 42[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942614:942915 [3] NCCL INFO Channel 06/0 : 43[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-7:811248:811554 [0] NCCL INFO Channel 09/0 : 24[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-7:811253:811558 [5] NCCL INFO Channel 09/0 : 29[5] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942617:942916 [6] NCCL INFO Channel 09/0 : 46[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-7:811249:811559 [1] NCCL INFO Channel 11/0 : 25[1] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942612:942914 [1] NCCL INFO Channel 06/0 : 41[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942611:942911 [0] NCCL INFO Channel 12/0 : 40[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942618:942918 [7] NCCL INFO Channel 03/0 : 47[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942615:942912 [4] NCCL INFO Channel 07/0 : 44[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942616:942913 [5] NCCL INFO Channel 06/0 : 45[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-7:811254:811557 [6] NCCL INFO Channel 03/0 : 30[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-7:811255:811561 [7] NCCL INFO Channel 07/0 : 31[7] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-7:811250:811555 [2] NCCL INFO Channel 07/0 : 26[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-7:811251:811556 [3] NCCL INFO Channel 13/0 : 27[3] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-7:811252:811560 [4] NCCL INFO Channel 05/0 : 28[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-7:811248:811554 [0] NCCL INFO Channel 11/0 : 24[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-7:811253:811558 [5] NCCL INFO Channel 11/0 : 29[5] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-7:811249:811559 [1] NCCL INFO Channel 13/0 : 25[1] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942613:942917 [2] NCCL INFO Channel 07/0 : 42[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942614:942915 [3] NCCL INFO Channel 07/0 : 43[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942617:942916 [6] NCCL INFO Channel 10/0 : 46[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942612:942914 [1] NCCL INFO Channel 07/0 : 41[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-7:811254:811557 [6] NCCL INFO Channel 05/0 : 30[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-7:811252:811560 [4] NCCL INFO Channel 07/0 : 28[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-7:811255:811561 [7] NCCL INFO Channel 09/0 : 31[7] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942611:942911 [0] NCCL INFO Channel 13/0 : 40[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942618:942918 [7] NCCL INFO Channel 04/0 : 47[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942615:942912 [4] NCCL INFO Channel 08/0 : 44[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942613:942917 [2] NCCL INFO Channel 08/0 : 42[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942614:942915 [3] NCCL INFO Channel 08/0 : 43[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942616:942913 [5] NCCL INFO Channel 07/0 : 45[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942617:942916 [6] NCCL INFO Channel 11/0 : 46[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942612:942914 [1] NCCL INFO Channel 08/0 : 41[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-7:811251:811556 [3] NCCL INFO Channel 15/0 : 27[3] -> 59[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-7:811250:811555 [2] NCCL INFO Channel 09/0 : 26[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-7:811248:811554 [0] NCCL INFO Channel 13/0 : 24[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-7:811253:811558 [5] NCCL INFO Channel 15/0 : 29[5] -> 61[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-7:811249:811559 [1] NCCL INFO Channel 15/0 : 25[1] -> 57[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-7:811254:811557 [6] NCCL INFO Channel 07/0 : 30[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942611:942911 [0] NCCL INFO Channel 14/0 : 40[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-7:811255:811561 [7] NCCL INFO Channel 11/0 : 31[7] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942618:942918 [7] NCCL INFO Channel 05/0 : 47[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942615:942912 [4] NCCL INFO Channel 09/0 : 44[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-7:811251:811556 [3] NCCL INFO Channel 01/0 : 59[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-7:811248:811554 [0] NCCL INFO Channel 15/0 : 24[0] -> 56[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-7:811250:811555 [2] NCCL INFO Channel 11/0 : 26[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-7:811253:811558 [5] NCCL INFO Channel 01/0 : 61[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942614:942915 [3] NCCL INFO Channel 09/0 : 43[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942613:942917 [2] NCCL INFO Channel 09/0 : 42[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942612:942914 [1] NCCL INFO Channel 10/0 : 41[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-7:811249:811559 [1] NCCL INFO Channel 01/0 : 57[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-7:811254:811557 [6] NCCL INFO Channel 09/0 : 30[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942617:942916 [6] NCCL INFO Channel 12/0 : 46[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942611:942911 [0] NCCL INFO Channel 15/0 : 40[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942616:942913 [5] NCCL INFO Channel 08/0 : 45[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942618:942918 [7] NCCL INFO Channel 06/0 : 47[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942615:942912 [4] NCCL INFO Channel 10/0 : 44[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-7:811255:811561 [7] NCCL INFO Channel 13/0 : 31[7] -> 63[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-7:811251:811556 [3] NCCL INFO Channel 03/0 : 59[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942614:942915 [3] NCCL INFO Channel 10/0 : 43[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942612:942914 [1] NCCL INFO Channel 11/0 : 41[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942613:942917 [2] NCCL INFO Channel 11/0 : 42[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942617:942916 [6] NCCL INFO Channel 13/0 : 46[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942616:942913 [5] NCCL INFO Channel 09/0 : 45[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-7:811248:811554 [0] NCCL INFO Channel 01/0 : 56[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-7:811250:811555 [2] NCCL INFO Channel 13/0 : 26[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-7:811253:811558 [5] NCCL INFO Channel 03/0 : 61[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-7:811249:811559 [1] NCCL INFO Channel 03/0 : 57[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-7:811254:811557 [6] NCCL INFO Channel 11/0 : 30[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-7:811251:811556 [3] NCCL INFO Channel 05/0 : 59[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-7:811255:811561 [7] NCCL INFO Channel 01/0 : 63[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942615:942912 [4] NCCL INFO Channel 11/0 : 44[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942618:942918 [7] NCCL INFO Channel 08/0 : 47[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942614:942915 [3] NCCL INFO Channel 12/0 : 43[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-7:811248:811554 [0] NCCL INFO Channel 03/0 : 56[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942613:942917 [2] NCCL INFO Channel 12/0 : 42[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942612:942914 [1] NCCL INFO Channel 12/0 : 41[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942617:942916 [6] NCCL INFO Channel 15/0 : 46[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942616:942913 [5] NCCL INFO Channel 10/0 : 45[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942615:942912 [4] NCCL INFO Channel 13/0 : 44[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942618:942918 [7] NCCL INFO Channel 09/0 : 47[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942614:942915 [3] NCCL INFO Channel 13/0 : 43[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-7:811250:811555 [2] NCCL INFO Channel 15/0 : 26[2] -> 58[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-7:811253:811558 [5] NCCL INFO Channel 05/0 : 61[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942613:942917 [2] NCCL INFO Channel 13/0 : 42[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942612:942914 [1] NCCL INFO Channel 13/0 : 41[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-7:811249:811559 [1] NCCL INFO Channel 05/0 : 57[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-7:811251:811556 [3] NCCL INFO Channel 07/0 : 59[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-7:811254:811557 [6] NCCL INFO Channel 13/0 : 30[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-7:811248:811554 [0] NCCL INFO Channel 05/0 : 56[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-7:811255:811561 [7] NCCL INFO Channel 03/0 : 63[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-7:811253:811558 [5] NCCL INFO Channel 07/0 : 61[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-7:811250:811555 [2] NCCL INFO Channel 01/0 : 58[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942615:942912 [4] NCCL INFO Channel 14/0 : 44[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942616:942913 [5] NCCL INFO Channel 11/0 : 45[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942614:942915 [3] NCCL INFO Channel 14/0 : 43[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942618:942918 [7] NCCL INFO Channel 10/0 : 47[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926594:926898 [0] NCCL INFO Channel 02/0 : 32[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942612:942914 [1] NCCL INFO Channel 14/0 : 41[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942613:942917 [2] NCCL INFO Channel 14/0 : 42[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-7:811251:811556 [3] NCCL INFO Channel 09/0 : 59[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-7:811248:811554 [0] NCCL INFO Channel 07/0 : 56[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-7:811249:811559 [1] NCCL INFO Channel 07/0 : 57[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-7:811254:811557 [6] NCCL INFO Channel 15/0 : 30[6] -> 62[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-7:811255:811561 [7] NCCL INFO Channel 05/0 : 63[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926594:926898 [0] NCCL INFO Channel 04/0 : 32[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942618:942918 [7] NCCL INFO Channel 11/0 : 47[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942615:942912 [4] NCCL INFO Channel 15/0 : 44[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942614:942915 [3] NCCL INFO Channel 15/0 : 43[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-7:811250:811555 [2] NCCL INFO Channel 03/0 : 58[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-7:811253:811558 [5] NCCL INFO Channel 09/0 : 61[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942616:942913 [5] NCCL INFO Channel 12/0 : 45[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942612:942914 [1] NCCL INFO Channel 15/0 : 41[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942613:942917 [2] NCCL INFO Channel 15/0 : 42[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926594:926898 [0] NCCL INFO Channel 06/0 : 32[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-7:811251:811556 [3] NCCL INFO Channel 13/0 : 59[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-7:811248:811554 [0] NCCL INFO Channel 09/0 : 56[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-7:811249:811559 [1] NCCL INFO Channel 11/0 : 57[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-7:811254:811557 [6] NCCL INFO Channel 01/0 : 62[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942618:942918 [7] NCCL INFO Channel 12/0 : 47[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942616:942913 [5] NCCL INFO Channel 14/0 : 45[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926594:926898 [0] NCCL INFO Channel 08/0 : 32[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-7:811255:811561 [7] NCCL INFO Channel 07/0 : 63[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-7:811250:811555 [2] NCCL INFO Channel 05/0 : 58[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-7:811253:811558 [5] NCCL INFO Channel 11/0 : 61[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-7:811251:811556 [3] NCCL INFO Channel 15/0 : 59[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-7:811248:811554 [0] NCCL INFO Channel 11/0 : 56[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942618:942918 [7] NCCL INFO Channel 13/0 : 47[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942616:942913 [5] NCCL INFO Channel 15/0 : 45[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926600:926893 [6] NCCL INFO Channel 00/0 : 38[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926594:926898 [0] NCCL INFO Channel 10/0 : 32[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-7:811254:811557 [6] NCCL INFO Channel 03/0 : 62[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-7:811249:811559 [1] NCCL INFO Channel 13/0 : 57[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942618:942918 [7] NCCL INFO Channel 14/0 : 47[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-7:811255:811561 [7] NCCL INFO Channel 09/0 : 63[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-7:811250:811555 [2] NCCL INFO Channel 07/0 : 58[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-7:811248:811554 [0] NCCL INFO Channel 13/0 : 56[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-7:811253:811558 [5] NCCL INFO Channel 15/0 : 61[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926594:926898 [0] NCCL INFO Channel 12/0 : 32[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926600:926893 [6] NCCL INFO Channel 02/0 : 38[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-7:811254:811557 [6] NCCL INFO Channel 05/0 : 62[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-7:811249:811559 [1] NCCL INFO Channel 15/0 : 57[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-7:811248:811554 [0] NCCL INFO Channel 15/0 : 56[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-7:811250:811555 [2] NCCL INFO Channel 09/0 : 58[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-7:811255:811561 [7] NCCL INFO Channel 11/0 : 63[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-7:811254:811557 [6] NCCL INFO Channel 07/0 : 62[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926594:926898 [0] NCCL INFO Channel 14/0 : 32[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926600:926893 [6] NCCL INFO Channel 04/0 : 38[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-7:811250:811555 [2] NCCL INFO Channel 11/0 : 58[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-7:811255:811561 [7] NCCL INFO Channel 13/0 : 63[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-7:811254:811557 [6] NCCL INFO Channel 09/0 : 62[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926596:926899 [2] NCCL INFO Channel 00/0 : 34[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926598:926894 [4] NCCL INFO Channel 00/0 : 36[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-7:811250:811555 [2] NCCL INFO Channel 13/0 : 58[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-7:811254:811557 [6] NCCL INFO Channel 11/0 : 62[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926597:926900 [3] NCCL INFO Channel 00/0 : 35[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926595:926897 [1] NCCL INFO Channel 00/0 : 33[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926600:926893 [6] NCCL INFO Channel 08/0 : 38[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-7:811250:811555 [2] NCCL INFO Channel 15/0 : 58[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926596:926899 [2] NCCL INFO Channel 04/0 : 34[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926595:926897 [1] NCCL INFO Channel 02/0 : 33[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926598:926894 [4] NCCL INFO Channel 02/0 : 36[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926597:926900 [3] NCCL INFO Channel 02/0 : 35[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-7:811254:811557 [6] NCCL INFO Channel 13/0 : 62[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926600:926893 [6] NCCL INFO Channel 10/0 : 38[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926599:926895 [5] NCCL INFO Channel 00/0 : 37[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926601:926896 [7] NCCL INFO Channel 00/0 : 39[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926596:926899 [2] NCCL INFO Channel 06/0 : 34[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926595:926897 [1] NCCL INFO Channel 04/0 : 33[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-7:811254:811557 [6] NCCL INFO Channel 15/0 : 62[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926598:926894 [4] NCCL INFO Channel 06/0 : 36[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926597:926900 [3] NCCL INFO Channel 04/0 : 35[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926599:926895 [5] NCCL INFO Channel 02/0 : 37[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926600:926893 [6] NCCL INFO Channel 12/0 : 38[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926596:926899 [2] NCCL INFO Channel 08/0 : 34[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926595:926897 [1] NCCL INFO Channel 06/0 : 33[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926601:926896 [7] NCCL INFO Channel 02/0 : 39[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926598:926894 [4] NCCL INFO Channel 08/0 : 36[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926597:926900 [3] NCCL INFO Channel 06/0 : 35[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926599:926895 [5] NCCL INFO Channel 04/0 : 37[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926600:926893 [6] NCCL INFO Channel 14/0 : 38[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926596:926899 [2] NCCL INFO Channel 10/0 : 34[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926595:926897 [1] NCCL INFO Channel 08/0 : 33[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926601:926896 [7] NCCL INFO Channel 04/0 : 39[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926598:926894 [4] NCCL INFO Channel 10/0 : 36[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926597:926900 [3] NCCL INFO Channel 08/0 : 35[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926599:926895 [5] NCCL INFO Channel 06/0 : 37[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926595:926897 [1] NCCL INFO Channel 10/0 : 33[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926601:926896 [7] NCCL INFO Channel 06/0 : 39[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926596:926899 [2] NCCL INFO Channel 12/0 : 34[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926598:926894 [4] NCCL INFO Channel 12/0 : 36[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926597:926900 [3] NCCL INFO Channel 10/0 : 35[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926599:926895 [5] NCCL INFO Channel 08/0 : 37[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926595:926897 [1] NCCL INFO Channel 12/0 : 33[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926601:926896 [7] NCCL INFO Channel 08/0 : 39[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926598:926894 [4] NCCL INFO Channel 14/0 : 36[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926596:926899 [2] NCCL INFO Channel 14/0 : 34[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926599:926895 [5] NCCL INFO Channel 10/0 : 37[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926597:926900 [3] NCCL INFO Channel 12/0 : 35[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926595:926897 [1] NCCL INFO Channel 14/0 : 33[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926601:926896 [7] NCCL INFO Channel 10/0 : 39[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926599:926895 [5] NCCL INFO Channel 12/0 : 37[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926597:926900 [3] NCCL INFO Channel 14/0 : 35[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926601:926896 [7] NCCL INFO Channel 12/0 : 39[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926599:926895 [5] NCCL INFO Channel 14/0 : 37[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926601:926896 [7] NCCL INFO Channel 14/0 : 39[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-7:811252:811560 [4] NCCL INFO Channel 09/0 : 28[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-7:811252:811560 [4] NCCL INFO Channel 11/0 : 28[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-7:811252:811560 [4] NCCL INFO Channel 13/0 : 28[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-7:811252:811560 [4] NCCL INFO Channel 15/0 : 28[4] -> 60[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-7:811252:811560 [4] NCCL INFO Channel 01/0 : 60[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-7:811252:811560 [4] NCCL INFO Channel 03/0 : 60[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-7:811252:811560 [4] NCCL INFO Channel 05/0 : 60[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-7:811252:811560 [4] NCCL INFO Channel 07/0 : 60[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-7:811252:811560 [4] NCCL INFO Channel 09/0 : 60[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-7:811252:811560 [4] NCCL INFO Channel 11/0 : 60[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-7:811252:811560 [4] NCCL INFO Channel 13/0 : 60[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-7:811252:811560 [4] NCCL INFO Channel 15/0 : 60[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688207:688504 [7] NCCL INFO Channel 01/0 : 39[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688207:688504 [7] NCCL INFO Channel 03/0 : 39[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688203:688501 [3] NCCL INFO Channel 01/0 : 35[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688207:688504 [7] NCCL INFO Channel 05/0 : 39[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688201:688500 [1] NCCL INFO Channel 01/0 : 33[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688207:688504 [7] NCCL INFO Channel 07/0 : 39[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688203:688501 [3] NCCL INFO Channel 03/0 : 35[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688201:688500 [1] NCCL INFO Channel 03/0 : 33[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688207:688504 [7] NCCL INFO Channel 09/0 : 39[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688203:688501 [3] NCCL INFO Channel 05/0 : 35[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688201:688500 [1] NCCL INFO Channel 05/0 : 33[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688207:688504 [7] NCCL INFO Channel 11/0 : 39[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688204:688502 [4] NCCL INFO Channel 01/0 : 36[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688203:688501 [3] NCCL INFO Channel 07/0 : 35[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688201:688500 [1] NCCL INFO Channel 07/0 : 33[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688200:688507 [0] NCCL INFO Channel 01/0 : 32[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688207:688504 [7] NCCL INFO Channel 13/0 : 39[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688206:688506 [6] NCCL INFO Channel 01/0 : 38[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688205:688505 [5] NCCL INFO Channel 01/0 : 37[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688204:688502 [4] NCCL INFO Channel 03/0 : 36[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688203:688501 [3] NCCL INFO Channel 09/0 : 35[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688201:688500 [1] NCCL INFO Channel 11/0 : 33[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688200:688507 [0] NCCL INFO Channel 03/0 : 32[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688205:688505 [5] NCCL INFO Channel 03/0 : 37[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688206:688506 [6] NCCL INFO Channel 03/0 : 38[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688202:688503 [2] NCCL INFO Channel 01/0 : 34[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688204:688502 [4] NCCL INFO Channel 05/0 : 36[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688203:688501 [3] NCCL INFO Channel 13/0 : 35[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688201:688500 [1] NCCL INFO Channel 13/0 : 33[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688205:688505 [5] NCCL INFO Channel 05/0 : 37[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688200:688507 [0] NCCL INFO Channel 05/0 : 32[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688202:688503 [2] NCCL INFO Channel 03/0 : 34[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688206:688506 [6] NCCL INFO Channel 05/0 : 38[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942618:942918 [7] NCCL INFO Channel 01/0 : 31[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688207:688504 [7] NCCL INFO Channel 00/0 : 23[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688204:688502 [4] NCCL INFO Channel 07/0 : 36[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688203:688501 [3] NCCL INFO Channel 15/0 : 35[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688201:688500 [1] NCCL INFO Channel 15/0 : 33[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688205:688505 [5] NCCL INFO Channel 07/0 : 37[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688202:688503 [2] NCCL INFO Channel 05/0 : 34[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688200:688507 [0] NCCL INFO Channel 07/0 : 32[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688206:688506 [6] NCCL INFO Channel 07/0 : 38[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942618:942918 [7] NCCL INFO Channel 03/0 : 31[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688207:688504 [7] NCCL INFO Channel 02/0 : 23[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688204:688502 [4] NCCL INFO Channel 09/0 : 36[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688205:688505 [5] NCCL INFO Channel 09/0 : 37[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688202:688503 [2] NCCL INFO Channel 07/0 : 34[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942618:942918 [7] NCCL INFO Channel 05/0 : 31[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688206:688506 [6] NCCL INFO Channel 09/0 : 38[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688200:688507 [0] NCCL INFO Channel 09/0 : 32[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688207:688504 [7] NCCL INFO Channel 04/0 : 23[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688204:688502 [4] NCCL INFO Channel 11/0 : 36[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942618:942918 [7] NCCL INFO Channel 07/0 : 31[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688202:688503 [2] NCCL INFO Channel 09/0 : 34[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688205:688505 [5] NCCL INFO Channel 11/0 : 37[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688206:688506 [6] NCCL INFO Channel 11/0 : 38[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688207:688504 [7] NCCL INFO Channel 06/0 : 23[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688203:688501 [3] NCCL INFO Channel 00/0 : 19[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688204:688502 [4] NCCL INFO Channel 13/0 : 36[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688200:688507 [0] NCCL INFO Channel 11/0 : 32[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688201:688500 [1] NCCL INFO Channel 00/0 : 17[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942612:942914 [1] NCCL INFO Channel 01/0 : 25[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942614:942915 [3] NCCL INFO Channel 01/0 : 27[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688205:688505 [5] NCCL INFO Channel 15/0 : 37[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688202:688503 [2] NCCL INFO Channel 11/0 : 34[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688206:688506 [6] NCCL INFO Channel 13/0 : 38[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942618:942918 [7] NCCL INFO Channel 09/0 : 31[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688207:688504 [7] NCCL INFO Channel 08/0 : 23[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942614:942915 [3] NCCL INFO Channel 03/0 : 27[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942612:942914 [1] NCCL INFO Channel 03/0 : 25[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942618:942918 [7] NCCL INFO Channel 11/0 : 31[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688203:688501 [3] NCCL INFO Channel 02/0 : 19[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688204:688502 [4] NCCL INFO Channel 15/0 : 36[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688200:688507 [0] NCCL INFO Channel 13/0 : 32[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688201:688500 [1] NCCL INFO Channel 02/0 : 17[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942612:942914 [1] NCCL INFO Channel 05/0 : 25[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942614:942915 [3] NCCL INFO Channel 05/0 : 27[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942618:942918 [7] NCCL INFO Channel 13/0 : 31[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688202:688503 [2] NCCL INFO Channel 13/0 : 34[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688206:688506 [6] NCCL INFO Channel 15/0 : 38[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688207:688504 [7] NCCL INFO Channel 10/0 : 23[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688203:688501 [3] NCCL INFO Channel 04/0 : 19[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688200:688507 [0] NCCL INFO Channel 15/0 : 32[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688201:688500 [1] NCCL INFO Channel 04/0 : 17[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942612:942914 [1] NCCL INFO Channel 07/0 : 25[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688207:688504 [7] NCCL INFO Channel 12/0 : 23[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688202:688503 [2] NCCL INFO Channel 15/0 : 34[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688205:688505 [5] NCCL INFO Channel 00/0 : 21[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688201:688500 [1] NCCL INFO Channel 06/0 : 17[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688203:688501 [3] NCCL INFO Channel 06/0 : 19[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942616:942913 [5] NCCL INFO Channel 01/0 : 29[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942614:942915 [3] NCCL INFO Channel 07/0 : 27[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688207:688504 [7] NCCL INFO Channel 14/0 : 23[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688205:688505 [5] NCCL INFO Channel 02/0 : 21[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942612:942914 [1] NCCL INFO Channel 11/0 : 25[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942616:942913 [5] NCCL INFO Channel 03/0 : 29[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688201:688500 [1] NCCL INFO Channel 08/0 : 17[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688204:688502 [4] NCCL INFO Channel 00/0 : 20[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688203:688501 [3] NCCL INFO Channel 08/0 : 19[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688207:688504 [7] NCCL INFO Channel 00/0 : 39[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688206:688506 [6] NCCL INFO Channel 00/0 : 22[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942615:942912 [4] NCCL INFO Channel 01/0 : 28[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942614:942915 [3] NCCL INFO Channel 09/0 : 27[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942612:942914 [1] NCCL INFO Channel 13/0 : 25[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688205:688505 [5] NCCL INFO Channel 04/0 : 21[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688200:688507 [0] NCCL INFO Channel 02/0 : 16[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688201:688500 [1] NCCL INFO Channel 10/0 : 17[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942616:942913 [5] NCCL INFO Channel 05/0 : 29[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942617:942916 [6] NCCL INFO Channel 01/0 : 30[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942611:942911 [0] NCCL INFO Channel 01/0 : 24[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942615:942912 [4] NCCL INFO Channel 03/0 : 28[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942612:942914 [1] NCCL INFO Channel 15/0 : 25[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942614:942915 [3] NCCL INFO Channel 13/0 : 27[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688203:688501 [3] NCCL INFO Channel 10/0 : 19[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688204:688502 [4] NCCL INFO Channel 02/0 : 20[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688202:688503 [2] NCCL INFO Channel 00/0 : 18[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942613:942917 [2] NCCL INFO Channel 01/0 : 26[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942616:942913 [5] NCCL INFO Channel 07/0 : 29[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942617:942916 [6] NCCL INFO Channel 03/0 : 30[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688207:688504 [7] NCCL INFO Channel 02/0 : 39[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688206:688506 [6] NCCL INFO Channel 02/0 : 22[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688205:688505 [5] NCCL INFO Channel 06/0 : 21[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688200:688507 [0] NCCL INFO Channel 04/0 : 16[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688201:688500 [1] NCCL INFO Channel 12/0 : 17[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688203:688501 [3] NCCL INFO Channel 12/0 : 19[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942611:942911 [0] NCCL INFO Channel 03/0 : 24[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942615:942912 [4] NCCL INFO Channel 05/0 : 28[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942614:942915 [3] NCCL INFO Channel 15/0 : 27[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688204:688502 [4] NCCL INFO Channel 06/0 : 20[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688202:688503 [2] NCCL INFO Channel 04/0 : 18[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688207:688504 [7] NCCL INFO Channel 04/0 : 39[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942616:942913 [5] NCCL INFO Channel 09/0 : 29[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942613:942917 [2] NCCL INFO Channel 03/0 : 26[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942617:942916 [6] NCCL INFO Channel 05/0 : 30[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942615:942912 [4] NCCL INFO Channel 07/0 : 28[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942611:942911 [0] NCCL INFO Channel 05/0 : 24[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688206:688506 [6] NCCL INFO Channel 04/0 : 22[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688205:688505 [5] NCCL INFO Channel 08/0 : 21[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688201:688500 [1] NCCL INFO Channel 14/0 : 17[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688200:688507 [0] NCCL INFO Channel 06/0 : 16[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688203:688501 [3] NCCL INFO Channel 14/0 : 19[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688204:688502 [4] NCCL INFO Channel 08/0 : 20[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942616:942913 [5] NCCL INFO Channel 11/0 : 29[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688202:688503 [2] NCCL INFO Channel 06/0 : 18[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688207:688504 [7] NCCL INFO Channel 06/0 : 39[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942617:942916 [6] NCCL INFO Channel 07/0 : 30[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942613:942917 [2] NCCL INFO Channel 05/0 : 26[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942615:942912 [4] NCCL INFO Channel 09/0 : 28[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688206:688506 [6] NCCL INFO Channel 08/0 : 22[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688205:688505 [5] NCCL INFO Channel 10/0 : 21[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942611:942911 [0] NCCL INFO Channel 07/0 : 24[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688201:688500 [1] NCCL INFO Channel 00/0 : 33[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942616:942913 [5] NCCL INFO Channel 15/0 : 29[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942617:942916 [6] NCCL INFO Channel 09/0 : 30[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688200:688507 [0] NCCL INFO Channel 08/0 : 16[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688203:688501 [3] NCCL INFO Channel 00/0 : 35[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688204:688502 [4] NCCL INFO Channel 10/0 : 20[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942613:942917 [2] NCCL INFO Channel 07/0 : 26[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942615:942912 [4] NCCL INFO Channel 11/0 : 28[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688207:688504 [7] NCCL INFO Channel 08/0 : 39[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688202:688503 [2] NCCL INFO Channel 08/0 : 18[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688206:688506 [6] NCCL INFO Channel 10/0 : 22[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942611:942911 [0] NCCL INFO Channel 09/0 : 24[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688205:688505 [5] NCCL INFO Channel 12/0 : 21[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942617:942916 [6] NCCL INFO Channel 11/0 : 30[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688201:688500 [1] NCCL INFO Channel 02/0 : 33[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688200:688507 [0] NCCL INFO Channel 10/0 : 16[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688203:688501 [3] NCCL INFO Channel 02/0 : 35[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688204:688502 [4] NCCL INFO Channel 12/0 : 20[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942613:942917 [2] NCCL INFO Channel 09/0 : 26[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942615:942912 [4] NCCL INFO Channel 13/0 : 28[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942611:942911 [0] NCCL INFO Channel 11/0 : 24[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942617:942916 [6] NCCL INFO Channel 13/0 : 30[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688207:688504 [7] NCCL INFO Channel 10/0 : 39[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688206:688506 [6] NCCL INFO Channel 12/0 : 22[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688205:688505 [5] NCCL INFO Channel 14/0 : 21[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688202:688503 [2] NCCL INFO Channel 10/0 : 18[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942615:942912 [4] NCCL INFO Channel 15/0 : 28[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942613:942917 [2] NCCL INFO Channel 11/0 : 26[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942611:942911 [0] NCCL INFO Channel 13/0 : 24[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942617:942916 [6] NCCL INFO Channel 15/0 : 30[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688201:688500 [1] NCCL INFO Channel 04/0 : 33[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688200:688507 [0] NCCL INFO Channel 12/0 : 16[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688203:688501 [3] NCCL INFO Channel 04/0 : 35[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688204:688502 [4] NCCL INFO Channel 14/0 : 20[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942613:942917 [2] NCCL INFO Channel 13/0 : 26[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942611:942911 [0] NCCL INFO Channel 15/0 : 24[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688207:688504 [7] NCCL INFO Channel 12/0 : 39[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688206:688506 [6] NCCL INFO Channel 14/0 : 22[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688205:688505 [5] NCCL INFO Channel 00/0 : 37[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688202:688503 [2] NCCL INFO Channel 12/0 : 18[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688201:688500 [1] NCCL INFO Channel 06/0 : 33[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942613:942917 [2] NCCL INFO Channel 15/0 : 26[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688203:688501 [3] NCCL INFO Channel 06/0 : 35[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688200:688507 [0] NCCL INFO Channel 14/0 : 16[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688204:688502 [4] NCCL INFO Channel 00/0 : 36[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688207:688504 [7] NCCL INFO Channel 14/0 : 39[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688206:688506 [6] NCCL INFO Channel 00/0 : 38[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688205:688505 [5] NCCL INFO Channel 02/0 : 37[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688201:688500 [1] NCCL INFO Channel 08/0 : 33[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688202:688503 [2] NCCL INFO Channel 14/0 : 18[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688203:688501 [3] NCCL INFO Channel 08/0 : 35[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688200:688507 [0] NCCL INFO Channel 02/0 : 32[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688204:688502 [4] NCCL INFO Channel 02/0 : 36[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688206:688506 [6] NCCL INFO Channel 02/0 : 38[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688205:688505 [5] NCCL INFO Channel 04/0 : 37[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688201:688500 [1] NCCL INFO Channel 10/0 : 33[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688202:688503 [2] NCCL INFO Channel 00/0 : 34[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688203:688501 [3] NCCL INFO Channel 10/0 : 35[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688204:688502 [4] NCCL INFO Channel 06/0 : 36[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688200:688507 [0] NCCL INFO Channel 04/0 : 32[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688206:688506 [6] NCCL INFO Channel 04/0 : 38[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688205:688505 [5] NCCL INFO Channel 06/0 : 37[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688202:688503 [2] NCCL INFO Channel 04/0 : 34[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688201:688500 [1] NCCL INFO Channel 12/0 : 33[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688200:688507 [0] NCCL INFO Channel 06/0 : 32[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688204:688502 [4] NCCL INFO Channel 08/0 : 36[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688206:688506 [6] NCCL INFO Channel 08/0 : 38[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688203:688501 [3] NCCL INFO Channel 12/0 : 35[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688205:688505 [5] NCCL INFO Channel 08/0 : 37[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688202:688503 [2] NCCL INFO Channel 06/0 : 34[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688201:688500 [1] NCCL INFO Channel 14/0 : 33[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688200:688507 [0] NCCL INFO Channel 08/0 : 32[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688204:688502 [4] NCCL INFO Channel 10/0 : 36[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926601:926896 [7] NCCL INFO Channel 00/0 : 55[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688206:688506 [6] NCCL INFO Channel 10/0 : 38[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688203:688501 [3] NCCL INFO Channel 14/0 : 35[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688202:688503 [2] NCCL INFO Channel 08/0 : 34[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688205:688505 [5] NCCL INFO Channel 10/0 : 37[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688204:688502 [4] NCCL INFO Channel 12/0 : 36[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688200:688507 [0] NCCL INFO Channel 10/0 : 32[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926601:926896 [7] NCCL INFO Channel 02/0 : 55[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688206:688506 [6] NCCL INFO Channel 12/0 : 38[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688205:688505 [5] NCCL INFO Channel 12/0 : 37[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688202:688503 [2] NCCL INFO Channel 10/0 : 34[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926601:926896 [7] NCCL INFO Channel 04/0 : 55[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688204:688502 [4] NCCL INFO Channel 14/0 : 36[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688205:688505 [5] NCCL INFO Channel 14/0 : 37[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688206:688506 [6] NCCL INFO Channel 14/0 : 38[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688200:688507 [0] NCCL INFO Channel 12/0 : 32[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688202:688503 [2] NCCL INFO Channel 12/0 : 34[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926601:926896 [7] NCCL INFO Channel 06/0 : 55[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688202:688503 [2] NCCL INFO Channel 14/0 : 34[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688200:688507 [0] NCCL INFO Channel 14/0 : 32[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926595:926897 [1] NCCL INFO Channel 00/0 : 49[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926601:926896 [7] NCCL INFO Channel 08/0 : 55[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926597:926900 [3] NCCL INFO Channel 00/0 : 51[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926595:926897 [1] NCCL INFO Channel 02/0 : 49[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926601:926896 [7] NCCL INFO Channel 10/0 : 55[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926597:926900 [3] NCCL INFO Channel 02/0 : 51[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926595:926897 [1] NCCL INFO Channel 04/0 : 49[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926601:926896 [7] NCCL INFO Channel 12/0 : 55[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926597:926900 [3] NCCL INFO Channel 04/0 : 51[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926595:926897 [1] NCCL INFO Channel 06/0 : 49[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926601:926896 [7] NCCL INFO Channel 14/0 : 55[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926598:926894 [4] NCCL INFO Channel 00/0 : 52[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926599:926895 [5] NCCL INFO Channel 00/0 : 53[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926600:926893 [6] NCCL INFO Channel 00/0 : 54[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926597:926900 [3] NCCL INFO Channel 06/0 : 51[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926595:926897 [1] NCCL INFO Channel 08/0 : 49[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926594:926898 [0] NCCL INFO Channel 02/0 : 48[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926596:926899 [2] NCCL INFO Channel 00/0 : 50[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926599:926895 [5] NCCL INFO Channel 02/0 : 53[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926598:926894 [4] NCCL INFO Channel 02/0 : 52[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926597:926900 [3] NCCL INFO Channel 08/0 : 51[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926595:926897 [1] NCCL INFO Channel 10/0 : 49[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926600:926893 [6] NCCL INFO Channel 02/0 : 54[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926594:926898 [0] NCCL INFO Channel 04/0 : 48[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926596:926899 [2] NCCL INFO Channel 04/0 : 50[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926599:926895 [5] NCCL INFO Channel 04/0 : 53[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926597:926900 [3] NCCL INFO Channel 10/0 : 51[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926595:926897 [1] NCCL INFO Channel 12/0 : 49[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926594:926898 [0] NCCL INFO Channel 06/0 : 48[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926598:926894 [4] NCCL INFO Channel 06/0 : 52[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926596:926899 [2] NCCL INFO Channel 06/0 : 50[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926600:926893 [6] NCCL INFO Channel 04/0 : 54[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926599:926895 [5] NCCL INFO Channel 06/0 : 53[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926597:926900 [3] NCCL INFO Channel 12/0 : 51[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926595:926897 [1] NCCL INFO Channel 14/0 : 49[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926594:926898 [0] NCCL INFO Channel 08/0 : 48[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926600:926893 [6] NCCL INFO Channel 08/0 : 54[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926596:926899 [2] NCCL INFO Channel 08/0 : 50[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926599:926895 [5] NCCL INFO Channel 08/0 : 53[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926598:926894 [4] NCCL INFO Channel 08/0 : 52[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926597:926900 [3] NCCL INFO Channel 14/0 : 51[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926594:926898 [0] NCCL INFO Channel 10/0 : 48[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926600:926893 [6] NCCL INFO Channel 10/0 : 54[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926596:926899 [2] NCCL INFO Channel 10/0 : 50[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926598:926894 [4] NCCL INFO Channel 10/0 : 52[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926599:926895 [5] NCCL INFO Channel 10/0 : 53[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926594:926898 [0] NCCL INFO Channel 12/0 : 48[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926600:926893 [6] NCCL INFO Channel 12/0 : 54[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926596:926899 [2] NCCL INFO Channel 12/0 : 50[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926594:926898 [0] NCCL INFO Channel 14/0 : 48[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926599:926895 [5] NCCL INFO Channel 12/0 : 53[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926598:926894 [4] NCCL INFO Channel 12/0 : 52[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926600:926893 [6] NCCL INFO Channel 14/0 : 54[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926596:926899 [2] NCCL INFO Channel 14/0 : 50[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926599:926895 [5] NCCL INFO Channel 14/0 : 53[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926598:926894 [4] NCCL INFO Channel 14/0 : 52[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019291:1019594 [3] NCCL INFO Channel 01/0 : 3[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019291:1019594 [3] NCCL INFO Channel 03/0 : 3[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019293:1019592 [5] NCCL INFO Channel 01/0 : 5[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019291:1019594 [3] NCCL INFO Channel 05/0 : 3[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019293:1019592 [5] NCCL INFO Channel 03/0 : 5[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019289:1019593 [1] NCCL INFO Channel 01/0 : 1[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019291:1019594 [3] NCCL INFO Channel 07/0 : 3[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019292:1019590 [4] NCCL INFO Channel 01/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019295:1019591 [7] NCCL INFO Channel 01/0 : 7[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019293:1019592 [5] NCCL INFO Channel 05/0 : 5[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019294:1019589 [6] NCCL INFO Channel 01/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019289:1019593 [1] NCCL INFO Channel 03/0 : 1[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019292:1019590 [4] NCCL INFO Channel 03/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019291:1019594 [3] NCCL INFO Channel 09/0 : 3[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019295:1019591 [7] NCCL INFO Channel 03/0 : 7[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019293:1019592 [5] NCCL INFO Channel 07/0 : 5[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019294:1019589 [6] NCCL INFO Channel 03/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019288:1019588 [0] NCCL INFO Channel 01/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019289:1019593 [1] NCCL INFO Channel 05/0 : 1[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019292:1019590 [4] NCCL INFO Channel 05/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019291:1019594 [3] NCCL INFO Channel 13/0 : 3[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019295:1019591 [7] NCCL INFO Channel 05/0 : 7[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019293:1019592 [5] NCCL INFO Channel 09/0 : 5[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019290:1019595 [2] NCCL INFO Channel 01/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019294:1019589 [6] NCCL INFO Channel 05/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019288:1019588 [0] NCCL INFO Channel 03/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019289:1019593 [1] NCCL INFO Channel 07/0 : 1[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019295:1019591 [7] NCCL INFO Channel 07/0 : 7[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019292:1019590 [4] NCCL INFO Channel 07/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019293:1019592 [5] NCCL INFO Channel 11/0 : 5[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019291:1019594 [3] NCCL INFO Channel 15/0 : 3[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019294:1019589 [6] NCCL INFO Channel 07/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019290:1019595 [2] NCCL INFO Channel 03/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019289:1019593 [1] NCCL INFO Channel 11/0 : 1[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019288:1019588 [0] NCCL INFO Channel 05/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019295:1019591 [7] NCCL INFO Channel 09/0 : 7[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019292:1019590 [4] NCCL INFO Channel 09/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019293:1019592 [5] NCCL INFO Channel 15/0 : 5[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019291:1019594 [3] NCCL INFO Channel 00/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019294:1019589 [6] NCCL INFO Channel 09/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019290:1019595 [2] NCCL INFO Channel 05/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019289:1019593 [1] NCCL INFO Channel 13/0 : 1[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019288:1019588 [0] NCCL INFO Channel 07/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019295:1019591 [7] NCCL INFO Channel 11/0 : 7[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019292:1019590 [4] NCCL INFO Channel 11/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019293:1019592 [5] NCCL INFO Channel 00/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019291:1019594 [3] NCCL INFO Channel 01/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019294:1019589 [6] NCCL INFO Channel 11/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019290:1019595 [2] NCCL INFO Channel 07/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019295:1019591 [7] NCCL INFO Channel 13/0 : 7[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019288:1019588 [0] NCCL INFO Channel 09/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019293:1019592 [5] NCCL INFO Channel 01/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019292:1019590 [4] NCCL INFO Channel 13/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019289:1019593 [1] NCCL INFO Channel 15/0 : 1[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019294:1019589 [6] NCCL INFO Channel 13/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019291:1019594 [3] NCCL INFO Channel 02/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019290:1019595 [2] NCCL INFO Channel 09/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019295:1019591 [7] NCCL INFO Channel 00/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019288:1019588 [0] NCCL INFO Channel 11/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019292:1019590 [4] NCCL INFO Channel 15/0 : 4[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019293:1019592 [5] NCCL INFO Channel 02/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019294:1019589 [6] NCCL INFO Channel 15/0 : 6[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019289:1019593 [1] NCCL INFO Channel 00/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019291:1019594 [3] NCCL INFO Channel 04/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019295:1019591 [7] NCCL INFO Channel 01/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019290:1019595 [2] NCCL INFO Channel 11/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019292:1019590 [4] NCCL INFO Channel 00/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019288:1019588 [0] NCCL INFO Channel 13/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019293:1019592 [5] NCCL INFO Channel 03/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019294:1019589 [6] NCCL INFO Channel 00/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019289:1019593 [1] NCCL INFO Channel 02/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019295:1019591 [7] NCCL INFO Channel 02/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019291:1019594 [3] NCCL INFO Channel 05/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019290:1019595 [2] NCCL INFO Channel 13/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019292:1019590 [4] NCCL INFO Channel 01/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019293:1019592 [5] NCCL INFO Channel 04/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019288:1019588 [0] NCCL INFO Channel 15/0 : 0[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019294:1019589 [6] NCCL INFO Channel 01/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019295:1019591 [7] NCCL INFO Channel 03/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019289:1019593 [1] NCCL INFO Channel 03/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019291:1019594 [3] NCCL INFO Channel 06/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019290:1019595 [2] NCCL INFO Channel 15/0 : 2[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019292:1019590 [4] NCCL INFO Channel 02/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019293:1019592 [5] NCCL INFO Channel 06/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019288:1019588 [0] NCCL INFO Channel 01/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019294:1019589 [6] NCCL INFO Channel 02/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019295:1019591 [7] NCCL INFO Channel 04/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019289:1019593 [1] NCCL INFO Channel 04/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019291:1019594 [3] NCCL INFO Channel 07/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019292:1019590 [4] NCCL INFO Channel 03/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019293:1019592 [5] NCCL INFO Channel 07/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019290:1019595 [2] NCCL INFO Channel 00/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019294:1019589 [6] NCCL INFO Channel 03/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019288:1019588 [0] NCCL INFO Channel 02/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019295:1019591 [7] NCCL INFO Channel 05/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019289:1019593 [1] NCCL INFO Channel 05/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019291:1019594 [3] NCCL INFO Channel 08/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019292:1019590 [4] NCCL INFO Channel 05/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019293:1019592 [5] NCCL INFO Channel 08/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019294:1019589 [6] NCCL INFO Channel 04/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019288:1019588 [0] NCCL INFO Channel 03/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019290:1019595 [2] NCCL INFO Channel 01/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019295:1019591 [7] NCCL INFO Channel 06/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019292:1019590 [4] NCCL INFO Channel 06/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019289:1019593 [1] NCCL INFO Channel 06/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019293:1019592 [5] NCCL INFO Channel 09/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019291:1019594 [3] NCCL INFO Channel 09/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019294:1019589 [6] NCCL INFO Channel 05/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019288:1019588 [0] NCCL INFO Channel 04/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019295:1019591 [7] NCCL INFO Channel 08/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019290:1019595 [2] NCCL INFO Channel 03/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019292:1019590 [4] NCCL INFO Channel 07/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019293:1019592 [5] NCCL INFO Channel 10/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019289:1019593 [1] NCCL INFO Channel 07/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019291:1019594 [3] NCCL INFO Channel 10/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019294:1019589 [6] NCCL INFO Channel 07/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019288:1019588 [0] NCCL INFO Channel 05/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019295:1019591 [7] NCCL INFO Channel 09/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019290:1019595 [2] NCCL INFO Channel 04/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019292:1019590 [4] NCCL INFO Channel 08/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019293:1019592 [5] NCCL INFO Channel 11/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019289:1019593 [1] NCCL INFO Channel 08/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019291:1019594 [3] NCCL INFO Channel 12/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019294:1019589 [6] NCCL INFO Channel 08/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019288:1019588 [0] NCCL INFO Channel 06/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019295:1019591 [7] NCCL INFO Channel 10/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019292:1019590 [4] NCCL INFO Channel 09/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019290:1019595 [2] NCCL INFO Channel 05/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019293:1019592 [5] NCCL INFO Channel 12/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019289:1019593 [1] NCCL INFO Channel 10/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019291:1019594 [3] NCCL INFO Channel 13/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019294:1019589 [6] NCCL INFO Channel 09/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019288:1019588 [0] NCCL INFO Channel 07/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019295:1019591 [7] NCCL INFO Channel 11/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019292:1019590 [4] NCCL INFO Channel 10/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019293:1019592 [5] NCCL INFO Channel 14/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019289:1019593 [1] NCCL INFO Channel 11/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019291:1019594 [3] NCCL INFO Channel 14/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019294:1019589 [6] NCCL INFO Channel 10/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019290:1019595 [2] NCCL INFO Channel 06/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019295:1019591 [7] NCCL INFO Channel 12/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019292:1019590 [4] NCCL INFO Channel 11/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019288:1019588 [0] NCCL INFO Channel 09/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019293:1019592 [5] NCCL INFO Channel 15/0 : 13[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019289:1019593 [1] NCCL INFO Channel 12/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019294:1019589 [6] NCCL INFO Channel 11/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019291:1019594 [3] NCCL INFO Channel 15/0 : 11[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019290:1019595 [2] NCCL INFO Channel 07/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019295:1019591 [7] NCCL INFO Channel 13/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019292:1019590 [4] NCCL INFO Channel 13/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019288:1019588 [0] NCCL INFO Channel 10/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019289:1019593 [1] NCCL INFO Channel 13/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019294:1019589 [6] NCCL INFO Channel 12/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019290:1019595 [2] NCCL INFO Channel 08/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019292:1019590 [4] NCCL INFO Channel 14/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019295:1019591 [7] NCCL INFO Channel 14/0 : 15[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019288:1019588 [0] NCCL INFO Channel 11/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019289:1019593 [1] NCCL INFO Channel 14/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019294:1019589 [6] NCCL INFO Channel 13/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019290:1019595 [2] NCCL INFO Channel 09/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:94347:94646 [5] NCCL INFO Channel 00/0 : 37[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019292:1019590 [4] NCCL INFO Channel 15/0 : 12[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019289:1019593 [1] NCCL INFO Channel 15/0 : 9[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019294:1019589 [6] NCCL INFO Channel 15/0 : 14[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019288:1019588 [0] NCCL INFO Channel 12/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019290:1019595 [2] NCCL INFO Channel 11/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:94345:94649 [3] NCCL INFO Channel 00/0 : 35[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:94347:94646 [5] NCCL INFO Channel 02/0 : 37[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019288:1019588 [0] NCCL INFO Channel 13/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019290:1019595 [2] NCCL INFO Channel 12/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:94345:94649 [3] NCCL INFO Channel 02/0 : 35[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:94347:94646 [5] NCCL INFO Channel 04/0 : 37[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019288:1019588 [0] NCCL INFO Channel 14/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:94345:94649 [3] NCCL INFO Channel 04/0 : 35[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019290:1019595 [2] NCCL INFO Channel 13/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:94347:94646 [5] NCCL INFO Channel 06/0 : 37[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:94343:94647 [1] NCCL INFO Channel 00/0 : 33[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:94349:94645 [7] NCCL INFO Channel 00/0 : 39[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:94348:94650 [6] NCCL INFO Channel 00/0 : 38[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:94345:94649 [3] NCCL INFO Channel 06/0 : 35[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019288:1019588 [0] NCCL INFO Channel 15/0 : 8[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:94347:94646 [5] NCCL INFO Channel 08/0 : 37[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:94346:94644 [4] NCCL INFO Channel 00/0 : 36[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:94343:94647 [1] NCCL INFO Channel 02/0 : 33[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019290:1019595 [2] NCCL INFO Channel 14/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:94345:94649 [3] NCCL INFO Channel 08/0 : 35[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:94348:94650 [6] NCCL INFO Channel 02/0 : 38[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:94349:94645 [7] NCCL INFO Channel 02/0 : 39[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019290:1019595 [2] NCCL INFO Channel 15/0 : 10[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:94347:94646 [5] NCCL INFO Channel 10/0 : 37[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:94343:94647 [1] NCCL INFO Channel 04/0 : 33[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:94346:94644 [4] NCCL INFO Channel 02/0 : 36[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:94345:94649 [3] NCCL INFO Channel 10/0 : 35[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:94348:94650 [6] NCCL INFO Channel 04/0 : 38[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:94349:94645 [7] NCCL INFO Channel 04/0 : 39[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:94347:94646 [5] NCCL INFO Channel 12/0 : 37[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:94343:94647 [1] NCCL INFO Channel 06/0 : 33[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:94342:94643 [0] NCCL INFO Channel 02/0 : 32[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:94345:94649 [3] NCCL INFO Channel 12/0 : 35[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:94346:94644 [4] NCCL INFO Channel 06/0 : 36[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:94348:94650 [6] NCCL INFO Channel 08/0 : 38[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:94349:94645 [7] NCCL INFO Channel 06/0 : 39[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:94347:94646 [5] NCCL INFO Channel 14/0 : 37[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:94343:94647 [1] NCCL INFO Channel 08/0 : 33[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:94342:94643 [0] NCCL INFO Channel 04/0 : 32[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:94344:94648 [2] NCCL INFO Channel 00/0 : 34[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:94345:94649 [3] NCCL INFO Channel 14/0 : 35[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:94346:94644 [4] NCCL INFO Channel 08/0 : 36[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:94347:94646 [5] NCCL INFO Channel 00/0 : 5[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:94349:94645 [7] NCCL INFO Channel 08/0 : 39[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:94343:94647 [1] NCCL INFO Channel 10/0 : 33[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:94348:94650 [6] NCCL INFO Channel 10/0 : 38[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:94342:94643 [0] NCCL INFO Channel 06/0 : 32[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:94344:94648 [2] NCCL INFO Channel 04/0 : 34[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:94345:94649 [3] NCCL INFO Channel 00/0 : 3[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:94347:94646 [5] NCCL INFO Channel 02/0 : 5[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:94343:94647 [1] NCCL INFO Channel 12/0 : 33[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:94346:94644 [4] NCCL INFO Channel 10/0 : 36[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:94349:94645 [7] NCCL INFO Channel 10/0 : 39[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:94348:94650 [6] NCCL INFO Channel 12/0 : 38[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:94342:94643 [0] NCCL INFO Channel 08/0 : 32[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:94345:94649 [3] NCCL INFO Channel 02/0 : 3[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:94344:94648 [2] NCCL INFO Channel 06/0 : 34[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:94347:94646 [5] NCCL INFO Channel 04/0 : 5[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:94343:94647 [1] NCCL INFO Channel 14/0 : 33[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:94346:94644 [4] NCCL INFO Channel 12/0 : 36[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:94349:94645 [7] NCCL INFO Channel 12/0 : 39[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:94345:94649 [3] NCCL INFO Channel 04/0 : 3[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:94348:94650 [6] NCCL INFO Channel 14/0 : 38[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:94344:94648 [2] NCCL INFO Channel 08/0 : 34[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:94342:94643 [0] NCCL INFO Channel 10/0 : 32[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:94347:94646 [5] NCCL INFO Channel 06/0 : 5[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:94343:94647 [1] NCCL INFO Channel 00/0 : 1[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:94346:94644 [4] NCCL INFO Channel 14/0 : 36[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:94349:94645 [7] NCCL INFO Channel 14/0 : 39[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:94345:94649 [3] NCCL INFO Channel 06/0 : 3[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:94344:94648 [2] NCCL INFO Channel 10/0 : 34[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:94347:94646 [5] NCCL INFO Channel 08/0 : 5[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:94348:94650 [6] NCCL INFO Channel 00/0 : 6[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:94343:94647 [1] NCCL INFO Channel 02/0 : 1[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:94342:94643 [0] NCCL INFO Channel 12/0 : 32[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:94346:94644 [4] NCCL INFO Channel 00/0 : 4[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:94345:94649 [3] NCCL INFO Channel 08/0 : 3[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:94349:94645 [7] NCCL INFO Channel 00/0 : 7[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:94344:94648 [2] NCCL INFO Channel 12/0 : 34[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:94347:94646 [5] NCCL INFO Channel 10/0 : 5[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:94343:94647 [1] NCCL INFO Channel 04/0 : 1[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:94348:94650 [6] NCCL INFO Channel 02/0 : 6[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:94346:94644 [4] NCCL INFO Channel 02/0 : 4[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:94342:94643 [0] NCCL INFO Channel 14/0 : 32[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:94345:94649 [3] NCCL INFO Channel 10/0 : 3[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:94349:94645 [7] NCCL INFO Channel 02/0 : 7[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:94344:94648 [2] NCCL INFO Channel 14/0 : 34[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:94343:94647 [1] NCCL INFO Channel 06/0 : 1[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:94347:94646 [5] NCCL INFO Channel 12/0 : 5[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:94348:94650 [6] NCCL INFO Channel 04/0 : 6[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:94346:94644 [4] NCCL INFO Channel 06/0 : 4[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:94345:94649 [3] NCCL INFO Channel 12/0 : 3[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:94344:94648 [2] NCCL INFO Channel 00/0 : 2[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:94342:94643 [0] NCCL INFO Channel 02/0 : 0[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:94349:94645 [7] NCCL INFO Channel 04/0 : 7[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:94343:94647 [1] NCCL INFO Channel 08/0 : 1[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:94347:94646 [5] NCCL INFO Channel 14/0 : 5[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:94348:94650 [6] NCCL INFO Channel 08/0 : 6[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:94346:94644 [4] NCCL INFO Channel 08/0 : 4[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:94345:94649 [3] NCCL INFO Channel 14/0 : 3[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:94344:94648 [2] NCCL INFO Channel 04/0 : 2[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:94349:94645 [7] NCCL INFO Channel 06/0 : 7[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:94343:94647 [1] NCCL INFO Channel 10/0 : 1[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:94342:94643 [0] NCCL INFO Channel 04/0 : 0[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:94348:94650 [6] NCCL INFO Channel 10/0 : 6[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:94346:94644 [4] NCCL INFO Channel 10/0 : 4[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:94344:94648 [2] NCCL INFO Channel 06/0 : 2[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:94349:94645 [7] NCCL INFO Channel 08/0 : 7[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:94343:94647 [1] NCCL INFO Channel 12/0 : 1[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:94346:94644 [4] NCCL INFO Channel 12/0 : 4[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:94348:94650 [6] NCCL INFO Channel 12/0 : 6[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:94342:94643 [0] NCCL INFO Channel 06/0 : 0[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:94344:94648 [2] NCCL INFO Channel 08/0 : 2[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:94349:94645 [7] NCCL INFO Channel 10/0 : 7[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:94343:94647 [1] NCCL INFO Channel 14/0 : 1[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:94346:94644 [4] NCCL INFO Channel 14/0 : 4[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:94342:94643 [0] NCCL INFO Channel 08/0 : 0[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:94344:94648 [2] NCCL INFO Channel 10/0 : 2[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:94348:94650 [6] NCCL INFO Channel 14/0 : 6[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:94349:94645 [7] NCCL INFO Channel 12/0 : 7[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:94342:94643 [0] NCCL INFO Channel 10/0 : 0[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:94344:94648 [2] NCCL INFO Channel 12/0 : 2[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:94349:94645 [7] NCCL INFO Channel 14/0 : 7[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:94342:94643 [0] NCCL INFO Channel 12/0 : 0[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:94344:94648 [2] NCCL INFO Channel 14/0 : 2[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:94342:94643 [0] NCCL INFO Channel 14/0 : 0[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880685:880977 [6] NCCL INFO Channel 00/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880685:880977 [6] NCCL INFO Channel 01/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880683:880982 [4] NCCL INFO Channel 00/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880685:880977 [6] NCCL INFO Channel 02/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880681:880983 [2] NCCL INFO Channel 00/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880683:880982 [4] NCCL INFO Channel 01/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618591:618885 [7] NCCL INFO Channel 00/0 : 23[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880679:880980 [0] NCCL INFO Channel 01/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880685:880977 [6] NCCL INFO Channel 03/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880682:880978 [3] NCCL INFO Channel 00/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880681:880983 [2] NCCL INFO Channel 01/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880683:880982 [4] NCCL INFO Channel 02/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618585:618884 [1] NCCL INFO Channel 00/0 : 17[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880686:880979 [7] NCCL INFO Channel 00/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618591:618885 [7] NCCL INFO Channel 02/0 : 23[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880679:880980 [0] NCCL INFO Channel 02/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880685:880977 [6] NCCL INFO Channel 04/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618585:618884 [1] NCCL INFO Channel 02/0 : 17[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880682:880978 [3] NCCL INFO Channel 01/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880680:880984 [1] NCCL INFO Channel 00/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880683:880982 [4] NCCL INFO Channel 03/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880681:880983 [2] NCCL INFO Channel 03/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618591:618885 [7] NCCL INFO Channel 04/0 : 23[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880684:880981 [5] NCCL INFO Channel 00/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618587:618887 [3] NCCL INFO Channel 00/0 : 19[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618585:618884 [1] NCCL INFO Channel 04/0 : 17[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618589:618883 [5] NCCL INFO Channel 00/0 : 21[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880679:880980 [0] NCCL INFO Channel 03/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880686:880979 [7] NCCL INFO Channel 01/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880685:880977 [6] NCCL INFO Channel 05/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618591:618885 [7] NCCL INFO Channel 06/0 : 23[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880682:880978 [3] NCCL INFO Channel 02/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880680:880984 [1] NCCL INFO Channel 02/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880683:880982 [4] NCCL INFO Channel 05/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880681:880983 [2] NCCL INFO Channel 04/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618585:618884 [1] NCCL INFO Channel 06/0 : 17[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618587:618887 [3] NCCL INFO Channel 02/0 : 19[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618589:618883 [5] NCCL INFO Channel 02/0 : 21[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880679:880980 [0] NCCL INFO Channel 04/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880684:880981 [5] NCCL INFO Channel 01/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880685:880977 [6] NCCL INFO Channel 07/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880686:880979 [7] NCCL INFO Channel 02/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880682:880978 [3] NCCL INFO Channel 04/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880683:880982 [4] NCCL INFO Channel 06/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618591:618885 [7] NCCL INFO Channel 08/0 : 23[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618585:618884 [1] NCCL INFO Channel 08/0 : 17[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880681:880983 [2] NCCL INFO Channel 05/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880680:880984 [1] NCCL INFO Channel 03/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880679:880980 [0] NCCL INFO Channel 05/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618587:618887 [3] NCCL INFO Channel 04/0 : 19[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618589:618883 [5] NCCL INFO Channel 04/0 : 21[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880682:880978 [3] NCCL INFO Channel 05/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880685:880977 [6] NCCL INFO Channel 08/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880686:880979 [7] NCCL INFO Channel 03/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880684:880981 [5] NCCL INFO Channel 02/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618584:618882 [0] NCCL INFO Channel 02/0 : 16[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618588:618888 [4] NCCL INFO Channel 00/0 : 20[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880683:880982 [4] NCCL INFO Channel 07/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618585:618884 [1] NCCL INFO Channel 10/0 : 17[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880681:880983 [2] NCCL INFO Channel 06/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618591:618885 [7] NCCL INFO Channel 10/0 : 23[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880680:880984 [1] NCCL INFO Channel 04/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618590:618889 [6] NCCL INFO Channel 00/0 : 22[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618586:618886 [2] NCCL INFO Channel 00/0 : 18[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618587:618887 [3] NCCL INFO Channel 06/0 : 19[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618589:618883 [5] NCCL INFO Channel 06/0 : 21[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880679:880980 [0] NCCL INFO Channel 06/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618588:618888 [4] NCCL INFO Channel 02/0 : 20[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618585:618884 [1] NCCL INFO Channel 12/0 : 17[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880682:880978 [3] NCCL INFO Channel 06/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618584:618882 [0] NCCL INFO Channel 04/0 : 16[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880685:880977 [6] NCCL INFO Channel 09/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618591:618885 [7] NCCL INFO Channel 12/0 : 23[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880686:880979 [7] NCCL INFO Channel 04/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618587:618887 [3] NCCL INFO Channel 08/0 : 19[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618590:618889 [6] NCCL INFO Channel 02/0 : 22[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880684:880981 [5] NCCL INFO Channel 03/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618586:618886 [2] NCCL INFO Channel 04/0 : 18[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880683:880982 [4] NCCL INFO Channel 08/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880681:880983 [2] NCCL INFO Channel 07/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880679:880980 [0] NCCL INFO Channel 07/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880680:880984 [1] NCCL INFO Channel 05/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880682:880978 [3] NCCL INFO Channel 07/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880685:880977 [6] NCCL INFO Channel 10/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618589:618883 [5] NCCL INFO Channel 08/0 : 21[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618588:618888 [4] NCCL INFO Channel 06/0 : 20[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618585:618884 [1] NCCL INFO Channel 14/0 : 17[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618584:618882 [0] NCCL INFO Channel 06/0 : 16[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618591:618885 [7] NCCL INFO Channel 14/0 : 23[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618587:618887 [3] NCCL INFO Channel 10/0 : 19[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880686:880979 [7] NCCL INFO Channel 05/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880683:880982 [4] NCCL INFO Channel 09/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618590:618889 [6] NCCL INFO Channel 04/0 : 22[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880681:880983 [2] NCCL INFO Channel 08/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880684:880981 [5] NCCL INFO Channel 04/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618586:618886 [2] NCCL INFO Channel 06/0 : 18[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880679:880980 [0] NCCL INFO Channel 09/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618588:618888 [4] NCCL INFO Channel 08/0 : 20[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880680:880984 [1] NCCL INFO Channel 06/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618589:618883 [5] NCCL INFO Channel 10/0 : 21[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880682:880978 [3] NCCL INFO Channel 08/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618584:618882 [0] NCCL INFO Channel 08/0 : 16[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880685:880977 [6] NCCL INFO Channel 11/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880686:880979 [7] NCCL INFO Channel 06/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618587:618887 [3] NCCL INFO Channel 12/0 : 19[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618590:618889 [6] NCCL INFO Channel 08/0 : 22[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618586:618886 [2] NCCL INFO Channel 08/0 : 18[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880683:880982 [4] NCCL INFO Channel 10/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618589:618883 [5] NCCL INFO Channel 12/0 : 21[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618588:618888 [4] NCCL INFO Channel 10/0 : 20[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880681:880983 [2] NCCL INFO Channel 09/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618584:618882 [0] NCCL INFO Channel 10/0 : 16[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880679:880980 [0] NCCL INFO Channel 10/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618587:618887 [3] NCCL INFO Channel 14/0 : 19[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880684:880981 [5] NCCL INFO Channel 06/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618590:618889 [6] NCCL INFO Channel 10/0 : 22[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880682:880978 [3] NCCL INFO Channel 09/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880680:880984 [1] NCCL INFO Channel 07/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880685:880977 [6] NCCL INFO Channel 12/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880686:880979 [7] NCCL INFO Channel 08/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880683:880982 [4] NCCL INFO Channel 11/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880681:880983 [2] NCCL INFO Channel 11/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880679:880980 [0] NCCL INFO Channel 11/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880684:880981 [5] NCCL INFO Channel 07/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618588:618888 [4] NCCL INFO Channel 12/0 : 20[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618589:618883 [5] NCCL INFO Channel 14/0 : 21[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618586:618886 [2] NCCL INFO Channel 10/0 : 18[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618584:618882 [0] NCCL INFO Channel 12/0 : 16[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618590:618889 [6] NCCL INFO Channel 12/0 : 22[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880682:880978 [3] NCCL INFO Channel 10/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618588:618888 [4] NCCL INFO Channel 14/0 : 20[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880680:880984 [1] NCCL INFO Channel 08/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880685:880977 [6] NCCL INFO Channel 13/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618586:618886 [2] NCCL INFO Channel 12/0 : 18[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880683:880982 [4] NCCL INFO Channel 13/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880681:880983 [2] NCCL INFO Channel 12/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880686:880979 [7] NCCL INFO Channel 09/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880679:880980 [0] NCCL INFO Channel 12/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618590:618889 [6] NCCL INFO Channel 14/0 : 22[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618584:618882 [0] NCCL INFO Channel 14/0 : 16[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880684:880981 [5] NCCL INFO Channel 08/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880682:880978 [3] NCCL INFO Channel 12/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618586:618886 [2] NCCL INFO Channel 14/0 : 18[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880685:880977 [6] NCCL INFO Channel 15/0 : 14[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880680:880984 [1] NCCL INFO Channel 10/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880686:880979 [7] NCCL INFO Channel 10/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880681:880983 [2] NCCL INFO Channel 13/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880679:880980 [0] NCCL INFO Channel 13/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880683:880982 [4] NCCL INFO Channel 14/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880682:880978 [3] NCCL INFO Channel 13/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880684:880981 [5] NCCL INFO Channel 09/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880685:880977 [6] NCCL INFO Channel 00/0 : 22[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880680:880984 [1] NCCL INFO Channel 11/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880686:880979 [7] NCCL INFO Channel 11/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880681:880983 [2] NCCL INFO Channel 14/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880679:880980 [0] NCCL INFO Channel 14/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880683:880982 [4] NCCL INFO Channel 15/0 : 12[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880682:880978 [3] NCCL INFO Channel 14/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880684:880981 [5] NCCL INFO Channel 10/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880685:880977 [6] NCCL INFO Channel 02/0 : 22[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880680:880984 [1] NCCL INFO Channel 12/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880686:880979 [7] NCCL INFO Channel 12/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880679:880980 [0] NCCL INFO Channel 15/0 : 8[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880681:880983 [2] NCCL INFO Channel 15/0 : 10[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880683:880982 [4] NCCL INFO Channel 00/0 : 20[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880682:880978 [3] NCCL INFO Channel 15/0 : 11[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880685:880977 [6] NCCL INFO Channel 04/0 : 22[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880684:880981 [5] NCCL INFO Channel 11/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880680:880984 [1] NCCL INFO Channel 13/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880686:880979 [7] NCCL INFO Channel 13/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880679:880980 [0] NCCL INFO Channel 02/0 : 16[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880681:880983 [2] NCCL INFO Channel 00/0 : 18[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880683:880982 [4] NCCL INFO Channel 02/0 : 20[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880682:880978 [3] NCCL INFO Channel 00/0 : 19[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880685:880977 [6] NCCL INFO Channel 08/0 : 22[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880686:880979 [7] NCCL INFO Channel 14/0 : 15[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880684:880981 [5] NCCL INFO Channel 12/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880679:880980 [0] NCCL INFO Channel 04/0 : 16[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880680:880984 [1] NCCL INFO Channel 14/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880681:880983 [2] NCCL INFO Channel 04/0 : 18[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880683:880982 [4] NCCL INFO Channel 06/0 : 20[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880682:880978 [3] NCCL INFO Channel 02/0 : 19[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880685:880977 [6] NCCL INFO Channel 10/0 : 22[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880686:880979 [7] NCCL INFO Channel 00/0 : 23[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880679:880980 [0] NCCL INFO Channel 06/0 : 16[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880680:880984 [1] NCCL INFO Channel 15/0 : 9[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880684:880981 [5] NCCL INFO Channel 14/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880681:880983 [2] NCCL INFO Channel 06/0 : 18[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880682:880978 [3] NCCL INFO Channel 04/0 : 19[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880683:880982 [4] NCCL INFO Channel 08/0 : 20[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880685:880977 [6] NCCL INFO Channel 12/0 : 22[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880686:880979 [7] NCCL INFO Channel 02/0 : 23[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880679:880980 [0] NCCL INFO Channel 08/0 : 16[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880680:880984 [1] NCCL INFO Channel 00/0 : 17[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880681:880983 [2] NCCL INFO Channel 08/0 : 18[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880682:880978 [3] NCCL INFO Channel 06/0 : 19[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880683:880982 [4] NCCL INFO Channel 10/0 : 20[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880684:880981 [5] NCCL INFO Channel 15/0 : 13[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880685:880977 [6] NCCL INFO Channel 14/0 : 22[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880686:880979 [7] NCCL INFO Channel 04/0 : 23[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880679:880980 [0] NCCL INFO Channel 10/0 : 16[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880680:880984 [1] NCCL INFO Channel 02/0 : 17[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880681:880983 [2] NCCL INFO Channel 10/0 : 18[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880682:880978 [3] NCCL INFO Channel 08/0 : 19[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880683:880982 [4] NCCL INFO Channel 12/0 : 20[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880686:880979 [7] NCCL INFO Channel 06/0 : 23[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880679:880980 [0] NCCL INFO Channel 12/0 : 16[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880680:880984 [1] NCCL INFO Channel 04/0 : 17[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880684:880981 [5] NCCL INFO Channel 00/0 : 21[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880681:880983 [2] NCCL INFO Channel 12/0 : 18[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880682:880978 [3] NCCL INFO Channel 10/0 : 19[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880683:880982 [4] NCCL INFO Channel 14/0 : 20[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880686:880979 [7] NCCL INFO Channel 08/0 : 23[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880679:880980 [0] NCCL INFO Channel 14/0 : 16[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019294:1019589 [6] NCCL INFO Channel 01/0 : 14[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880680:880984 [1] NCCL INFO Channel 06/0 : 17[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880684:880981 [5] NCCL INFO Channel 02/0 : 21[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880682:880978 [3] NCCL INFO Channel 12/0 : 19[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880681:880983 [2] NCCL INFO Channel 14/0 : 18[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880686:880979 [7] NCCL INFO Channel 10/0 : 23[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019294:1019589 [6] NCCL INFO Channel 03/0 : 14[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880680:880984 [1] NCCL INFO Channel 08/0 : 17[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880682:880978 [3] NCCL INFO Channel 14/0 : 19[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618590:618889 [6] NCCL INFO Channel 01/0 : 14[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019294:1019589 [6] NCCL INFO Channel 05/0 : 14[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880684:880981 [5] NCCL INFO Channel 04/0 : 21[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880686:880979 [7] NCCL INFO Channel 12/0 : 23[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880685:880977 [6] NCCL INFO Channel 00/0 : 22[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019292:1019590 [4] NCCL INFO Channel 01/0 : 12[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880680:880984 [1] NCCL INFO Channel 10/0 : 17[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618590:618889 [6] NCCL INFO Channel 03/0 : 14[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880686:880979 [7] NCCL INFO Channel 14/0 : 23[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880684:880981 [5] NCCL INFO Channel 06/0 : 21[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880685:880977 [6] NCCL INFO Channel 02/0 : 22[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019294:1019589 [6] NCCL INFO Channel 07/0 : 14[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019288:1019588 [0] NCCL INFO Channel 01/0 : 8[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880680:880984 [1] NCCL INFO Channel 12/0 : 17[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618590:618889 [6] NCCL INFO Channel 05/0 : 14[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019292:1019590 [4] NCCL INFO Channel 03/0 : 12[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019294:1019589 [6] NCCL INFO Channel 09/0 : 14[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019288:1019588 [0] NCCL INFO Channel 03/0 : 8[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880684:880981 [5] NCCL INFO Channel 08/0 : 21[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880685:880977 [6] NCCL INFO Channel 04/0 : 22[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880683:880982 [4] NCCL INFO Channel 00/0 : 20[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880679:880980 [0] NCCL INFO Channel 02/0 : 16[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618588:618888 [4] NCCL INFO Channel 01/0 : 12[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019290:1019595 [2] NCCL INFO Channel 01/0 : 10[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019292:1019590 [4] NCCL INFO Channel 05/0 : 12[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019294:1019589 [6] NCCL INFO Channel 11/0 : 14[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880680:880984 [1] NCCL INFO Channel 14/0 : 17[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618584:618882 [0] NCCL INFO Channel 01/0 : 8[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618590:618889 [6] NCCL INFO Channel 07/0 : 14[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618588:618888 [4] NCCL INFO Channel 03/0 : 12[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618584:618882 [0] NCCL INFO Channel 03/0 : 8[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019288:1019588 [0] NCCL INFO Channel 05/0 : 8[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019291:1019594 [3] NCCL INFO Channel 01/0 : 11[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880685:880977 [6] NCCL INFO Channel 08/0 : 22[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019292:1019590 [4] NCCL INFO Channel 07/0 : 12[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880684:880981 [5] NCCL INFO Channel 10/0 : 21[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880679:880980 [0] NCCL INFO Channel 04/0 : 16[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880683:880982 [4] NCCL INFO Channel 02/0 : 20[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880681:880983 [2] NCCL INFO Channel 00/0 : 18[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618590:618889 [6] NCCL INFO Channel 09/0 : 14[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618586:618886 [2] NCCL INFO Channel 01/0 : 10[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019294:1019589 [6] NCCL INFO Channel 13/0 : 14[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019290:1019595 [2] NCCL INFO Channel 03/0 : 10[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019288:1019588 [0] NCCL INFO Channel 07/0 : 8[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019292:1019590 [4] NCCL INFO Channel 09/0 : 12[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019291:1019594 [3] NCCL INFO Channel 03/0 : 11[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880682:880978 [3] NCCL INFO Channel 00/0 : 19[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880685:880977 [6] NCCL INFO Channel 10/0 : 22[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880679:880980 [0] NCCL INFO Channel 06/0 : 16[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880684:880981 [5] NCCL INFO Channel 12/0 : 21[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880683:880982 [4] NCCL INFO Channel 06/0 : 20[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618584:618882 [0] NCCL INFO Channel 05/0 : 8[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618588:618888 [4] NCCL INFO Channel 05/0 : 12[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618587:618887 [3] NCCL INFO Channel 01/0 : 11[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618590:618889 [6] NCCL INFO Channel 11/0 : 14[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019295:1019591 [7] NCCL INFO Channel 01/0 : 15[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019294:1019589 [6] NCCL INFO Channel 15/0 : 14[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019288:1019588 [0] NCCL INFO Channel 09/0 : 8[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019290:1019595 [2] NCCL INFO Channel 05/0 : 10[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880682:880978 [3] NCCL INFO Channel 02/0 : 19[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880681:880983 [2] NCCL INFO Channel 04/0 : 18[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880685:880977 [6] NCCL INFO Channel 12/0 : 22[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618584:618882 [0] NCCL INFO Channel 07/0 : 8[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618588:618888 [4] NCCL INFO Channel 07/0 : 12[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618587:618887 [3] NCCL INFO Channel 03/0 : 11[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019292:1019590 [4] NCCL INFO Channel 11/0 : 12[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019291:1019594 [3] NCCL INFO Channel 05/0 : 11[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019289:1019593 [1] NCCL INFO Channel 01/0 : 9[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019295:1019591 [7] NCCL INFO Channel 03/0 : 15[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880679:880980 [0] NCCL INFO Channel 08/0 : 16[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880683:880982 [4] NCCL INFO Channel 08/0 : 20[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880684:880981 [5] NCCL INFO Channel 14/0 : 21[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618586:618886 [2] NCCL INFO Channel 03/0 : 10[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880686:880979 [7] NCCL INFO Channel 00/0 : 23[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880682:880978 [3] NCCL INFO Channel 04/0 : 19[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618590:618889 [6] NCCL INFO Channel 13/0 : 14[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618584:618882 [0] NCCL INFO Channel 09/0 : 8[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618591:618885 [7] NCCL INFO Channel 01/0 : 15[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019288:1019588 [0] NCCL INFO Channel 11/0 : 8[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019290:1019595 [2] NCCL INFO Channel 07/0 : 10[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019292:1019590 [4] NCCL INFO Channel 13/0 : 12[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019291:1019594 [3] NCCL INFO Channel 07/0 : 11[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019289:1019593 [1] NCCL INFO Channel 03/0 : 9[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880681:880983 [2] NCCL INFO Channel 06/0 : 18[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019295:1019591 [7] NCCL INFO Channel 05/0 : 15[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880685:880977 [6] NCCL INFO Channel 14/0 : 22[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618587:618887 [3] NCCL INFO Channel 05/0 : 11[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618588:618888 [4] NCCL INFO Channel 09/0 : 12[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019288:1019588 [0] NCCL INFO Channel 13/0 : 8[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880679:880980 [0] NCCL INFO Channel 10/0 : 16[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880680:880984 [1] NCCL INFO Channel 00/0 : 17[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618585:618884 [1] NCCL INFO Channel 01/0 : 9[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880683:880982 [4] NCCL INFO Channel 10/0 : 20[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019292:1019590 [4] NCCL INFO Channel 15/0 : 12[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019291:1019594 [3] NCCL INFO Channel 09/0 : 11[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019290:1019595 [2] NCCL INFO Channel 09/0 : 10[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618590:618889 [6] NCCL INFO Channel 15/0 : 14[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019289:1019593 [1] NCCL INFO Channel 05/0 : 9[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880682:880978 [3] NCCL INFO Channel 06/0 : 19[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618584:618882 [0] NCCL INFO Channel 11/0 : 8[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019288:1019588 [0] NCCL INFO Channel 15/0 : 8[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019295:1019591 [7] NCCL INFO Channel 07/0 : 15[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880686:880979 [7] NCCL INFO Channel 02/0 : 23[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618586:618886 [2] NCCL INFO Channel 05/0 : 10[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880681:880983 [2] NCCL INFO Channel 08/0 : 18[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618591:618885 [7] NCCL INFO Channel 03/0 : 15[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880679:880980 [0] NCCL INFO Channel 12/0 : 16[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618587:618887 [3] NCCL INFO Channel 07/0 : 11[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880680:880984 [1] NCCL INFO Channel 02/0 : 17[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618588:618888 [4] NCCL INFO Channel 11/0 : 12[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880683:880982 [4] NCCL INFO Channel 12/0 : 20[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019291:1019594 [3] NCCL INFO Channel 13/0 : 11[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019289:1019593 [1] NCCL INFO Channel 07/0 : 9[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019295:1019591 [7] NCCL INFO Channel 09/0 : 15[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019290:1019595 [2] NCCL INFO Channel 11/0 : 10[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618585:618884 [1] NCCL INFO Channel 03/0 : 9[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880682:880978 [3] NCCL INFO Channel 08/0 : 19[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618590:618889 [6] NCCL INFO Channel 01/0 : 30[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880686:880979 [7] NCCL INFO Channel 04/0 : 23[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618584:618882 [0] NCCL INFO Channel 13/0 : 8[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880681:880983 [2] NCCL INFO Channel 10/0 : 18[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618586:618886 [2] NCCL INFO Channel 07/0 : 10[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880679:880980 [0] NCCL INFO Channel 14/0 : 16[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019291:1019594 [3] NCCL INFO Channel 15/0 : 11[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019289:1019593 [1] NCCL INFO Channel 11/0 : 9[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880680:880984 [1] NCCL INFO Channel 04/0 : 17[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688206:688506 [6] NCCL INFO Channel 00/0 : 6[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019295:1019591 [7] NCCL INFO Channel 11/0 : 15[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019293:1019592 [5] NCCL INFO Channel 01/0 : 13[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880683:880982 [4] NCCL INFO Channel 14/0 : 20[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618591:618885 [7] NCCL INFO Channel 05/0 : 15[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618587:618887 [3] NCCL INFO Channel 09/0 : 11[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618588:618888 [4] NCCL INFO Channel 13/0 : 12[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618585:618884 [1] NCCL INFO Channel 05/0 : 9[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618590:618889 [6] NCCL INFO Channel 03/0 : 30[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618584:618882 [0] NCCL INFO Channel 15/0 : 8[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019290:1019595 [2] NCCL INFO Channel 13/0 : 10[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019289:1019593 [1] NCCL INFO Channel 13/0 : 9[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019295:1019591 [7] NCCL INFO Channel 13/0 : 15[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688206:688506 [6] NCCL INFO Channel 02/0 : 6[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618587:618887 [3] NCCL INFO Channel 13/0 : 11[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618588:618888 [4] NCCL INFO Channel 15/0 : 12[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019293:1019592 [5] NCCL INFO Channel 03/0 : 13[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019290:1019595 [2] NCCL INFO Channel 15/0 : 10[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019289:1019593 [1] NCCL INFO Channel 15/0 : 9[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019293:1019592 [5] NCCL INFO Channel 05/0 : 13[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618591:618885 [7] NCCL INFO Channel 07/0 : 15[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688206:688506 [6] NCCL INFO Channel 04/0 : 6[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618585:618884 [1] NCCL INFO Channel 07/0 : 9[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618584:618882 [0] NCCL INFO Channel 01/0 : 24[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618590:618889 [6] NCCL INFO Channel 05/0 : 30[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618586:618886 [2] NCCL INFO Channel 09/0 : 10[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019293:1019592 [5] NCCL INFO Channel 07/0 : 13[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618587:618887 [3] NCCL INFO Channel 15/0 : 11[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618589:618883 [5] NCCL INFO Channel 01/0 : 13[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618588:618888 [4] NCCL INFO Channel 01/0 : 28[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618591:618885 [7] NCCL INFO Channel 09/0 : 15[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618585:618884 [1] NCCL INFO Channel 11/0 : 9[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019293:1019592 [5] NCCL INFO Channel 09/0 : 13[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688204:688502 [4] NCCL INFO Channel 00/0 : 4[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688206:688506 [6] NCCL INFO Channel 08/0 : 6[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618590:618889 [6] NCCL INFO Channel 07/0 : 30[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618586:618886 [2] NCCL INFO Channel 11/0 : 10[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019293:1019592 [5] NCCL INFO Channel 11/0 : 13[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618587:618887 [3] NCCL INFO Channel 01/0 : 27[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688200:688507 [0] NCCL INFO Channel 02/0 : 0[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618584:618882 [0] NCCL INFO Channel 03/0 : 24[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618589:618883 [5] NCCL INFO Channel 03/0 : 13[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618588:618888 [4] NCCL INFO Channel 03/0 : 28[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688204:688502 [4] NCCL INFO Channel 02/0 : 4[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019293:1019592 [5] NCCL INFO Channel 15/0 : 13[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618585:618884 [1] NCCL INFO Channel 13/0 : 9[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618591:618885 [7] NCCL INFO Channel 11/0 : 15[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618590:618889 [6] NCCL INFO Channel 09/0 : 30[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688206:688506 [6] NCCL INFO Channel 10/0 : 6[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688200:688507 [0] NCCL INFO Channel 04/0 : 0[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618586:618886 [2] NCCL INFO Channel 13/0 : 10[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618587:618887 [3] NCCL INFO Channel 03/0 : 27[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688202:688503 [2] NCCL INFO Channel 00/0 : 2[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618584:618882 [0] NCCL INFO Channel 05/0 : 24[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688204:688502 [4] NCCL INFO Channel 06/0 : 4[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688206:688506 [6] NCCL INFO Channel 12/0 : 6[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688200:688507 [0] NCCL INFO Channel 06/0 : 0[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618588:618888 [4] NCCL INFO Channel 05/0 : 28[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688203:688501 [3] NCCL INFO Channel 00/0 : 3[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688202:688503 [2] NCCL INFO Channel 04/0 : 2[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618589:618883 [5] NCCL INFO Channel 05/0 : 13[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618585:618884 [1] NCCL INFO Channel 15/0 : 9[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688204:688502 [4] NCCL INFO Channel 08/0 : 4[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688206:688506 [6] NCCL INFO Channel 14/0 : 6[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618591:618885 [7] NCCL INFO Channel 13/0 : 15[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618590:618889 [6] NCCL INFO Channel 11/0 : 30[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618587:618887 [3] NCCL INFO Channel 05/0 : 27[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618586:618886 [2] NCCL INFO Channel 15/0 : 10[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618584:618882 [0] NCCL INFO Channel 07/0 : 24[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618588:618888 [4] NCCL INFO Channel 07/0 : 28[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688203:688501 [3] NCCL INFO Channel 02/0 : 3[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688202:688503 [2] NCCL INFO Channel 06/0 : 2[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688200:688507 [0] NCCL INFO Channel 08/0 : 0[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688204:688502 [4] NCCL INFO Channel 10/0 : 4[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618585:618884 [1] NCCL INFO Channel 01/0 : 25[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688206:688506 [6] NCCL INFO Channel 00/0 : 38[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618590:618889 [6] NCCL INFO Channel 13/0 : 30[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618587:618887 [3] NCCL INFO Channel 07/0 : 27[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618589:618883 [5] NCCL INFO Channel 07/0 : 13[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618591:618885 [7] NCCL INFO Channel 01/0 : 31[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618586:618886 [2] NCCL INFO Channel 01/0 : 26[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618584:618882 [0] NCCL INFO Channel 09/0 : 24[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618588:618888 [4] NCCL INFO Channel 09/0 : 28[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618585:618884 [1] NCCL INFO Channel 03/0 : 25[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688203:688501 [3] NCCL INFO Channel 04/0 : 3[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688202:688503 [2] NCCL INFO Channel 08/0 : 2[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688200:688507 [0] NCCL INFO Channel 10/0 : 0[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688207:688504 [7] NCCL INFO Channel 00/0 : 7[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618590:618889 [6] NCCL INFO Channel 15/0 : 30[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688206:688506 [6] NCCL INFO Channel 02/0 : 38[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688204:688502 [4] NCCL INFO Channel 12/0 : 4[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688201:688500 [1] NCCL INFO Channel 00/0 : 1[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618587:618887 [3] NCCL INFO Channel 09/0 : 27[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618584:618882 [0] NCCL INFO Channel 11/0 : 24[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618586:618886 [2] NCCL INFO Channel 03/0 : 26[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688203:688501 [3] NCCL INFO Channel 06/0 : 3[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688202:688503 [2] NCCL INFO Channel 10/0 : 2[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618591:618885 [7] NCCL INFO Channel 03/0 : 31[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618585:618884 [1] NCCL INFO Channel 05/0 : 25[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618588:618888 [4] NCCL INFO Channel 11/0 : 28[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688207:688504 [7] NCCL INFO Channel 02/0 : 7[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618589:618883 [5] NCCL INFO Channel 09/0 : 13[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688206:688506 [6] NCCL INFO Channel 04/0 : 38[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688204:688502 [4] NCCL INFO Channel 14/0 : 4[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618587:618887 [3] NCCL INFO Channel 13/0 : 27[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618584:618882 [0] NCCL INFO Channel 13/0 : 24[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019294:1019589 [6] NCCL INFO Channel 01/0 : 30[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688200:688507 [0] NCCL INFO Channel 12/0 : 0[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618586:618886 [2] NCCL INFO Channel 05/0 : 26[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688201:688500 [1] NCCL INFO Channel 02/0 : 1[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688203:688501 [3] NCCL INFO Channel 08/0 : 3[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618585:618884 [1] NCCL INFO Channel 07/0 : 25[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618591:618885 [7] NCCL INFO Channel 05/0 : 31[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688202:688503 [2] NCCL INFO Channel 12/0 : 2[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618588:618888 [4] NCCL INFO Channel 13/0 : 28[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688207:688504 [7] NCCL INFO Channel 04/0 : 7[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618589:618883 [5] NCCL INFO Channel 11/0 : 13[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618587:618887 [3] NCCL INFO Channel 15/0 : 27[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618584:618882 [0] NCCL INFO Channel 15/0 : 24[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688206:688506 [6] NCCL INFO Channel 08/0 : 38[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019294:1019589 [6] NCCL INFO Channel 03/0 : 30[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688204:688502 [4] NCCL INFO Channel 00/0 : 36[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618586:618886 [2] NCCL INFO Channel 07/0 : 26[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618585:618884 [1] NCCL INFO Channel 11/0 : 25[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019294:1019589 [6] NCCL INFO Channel 05/0 : 30[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618591:618885 [7] NCCL INFO Channel 07/0 : 31[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942617:942916 [6] NCCL INFO Channel 01/0 : 46[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618588:618888 [4] NCCL INFO Channel 15/0 : 28[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688201:688500 [1] NCCL INFO Channel 04/0 : 1[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688200:688507 [0] NCCL INFO Channel 14/0 : 0[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688203:688501 [3] NCCL INFO Channel 10/0 : 3[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688202:688503 [2] NCCL INFO Channel 14/0 : 2[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688207:688504 [7] NCCL INFO Channel 06/0 : 7[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618589:618883 [5] NCCL INFO Channel 15/0 : 13[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688206:688506 [6] NCCL INFO Channel 10/0 : 38[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618590:618889 [6] NCCL INFO Channel 01/0 : 62[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688204:688502 [4] NCCL INFO Channel 02/0 : 36[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942617:942916 [6] NCCL INFO Channel 03/0 : 46[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019294:1019589 [6] NCCL INFO Channel 07/0 : 30[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880685:880977 [6] NCCL INFO Channel 00/0 : 38[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618586:618886 [2] NCCL INFO Channel 09/0 : 26[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688201:688500 [1] NCCL INFO Channel 06/0 : 1[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942617:942916 [6] NCCL INFO Channel 05/0 : 46[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880682:880978 [3] NCCL INFO Channel 10/0 : 19[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618585:618884 [1] NCCL INFO Channel 13/0 : 25[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688202:688503 [2] NCCL INFO Channel 00/0 : 34[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688200:688507 [0] NCCL INFO Channel 02/0 : 32[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019291:1019594 [3] NCCL INFO Channel 01/0 : 27[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880686:880979 [7] NCCL INFO Channel 06/0 : 23[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618591:618885 [7] NCCL INFO Channel 09/0 : 31[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688207:688504 [7] NCCL INFO Channel 08/0 : 7[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880681:880983 [2] NCCL INFO Channel 12/0 : 18[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618589:618883 [5] NCCL INFO Channel 01/0 : 29[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688206:688506 [6] NCCL INFO Channel 12/0 : 38[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880680:880984 [1] NCCL INFO Channel 06/0 : 17[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618590:618889 [6] NCCL INFO Channel 03/0 : 62[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688204:688502 [4] NCCL INFO Channel 06/0 : 36[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688203:688501 [3] NCCL INFO Channel 12/0 : 3[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688205:688505 [5] NCCL INFO Channel 00/0 : 5[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880685:880977 [6] NCCL INFO Channel 02/0 : 38[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618586:618886 [2] NCCL INFO Channel 11/0 : 26[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688201:688500 [1] NCCL INFO Channel 08/0 : 1[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942617:942916 [6] NCCL INFO Channel 07/0 : 46[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880682:880978 [3] NCCL INFO Channel 12/0 : 19[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618585:618884 [1] NCCL INFO Channel 15/0 : 25[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019288:1019588 [0] NCCL INFO Channel 01/0 : 24[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019294:1019589 [6] NCCL INFO Channel 09/0 : 30[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880686:880979 [7] NCCL INFO Channel 08/0 : 23[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618591:618885 [7] NCCL INFO Channel 11/0 : 31[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880684:880981 [5] NCCL INFO Channel 00/0 : 21[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880681:880983 [2] NCCL INFO Channel 14/0 : 18[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880680:880984 [1] NCCL INFO Channel 08/0 : 17[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688202:688503 [2] NCCL INFO Channel 04/0 : 34[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880685:880977 [6] NCCL INFO Channel 04/0 : 38[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688200:688507 [0] NCCL INFO Channel 04/0 : 32[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688207:688504 [7] NCCL INFO Channel 10/0 : 7[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880682:880978 [3] NCCL INFO Channel 14/0 : 19[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688206:688506 [6] NCCL INFO Channel 14/0 : 38[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688204:688502 [4] NCCL INFO Channel 08/0 : 36[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019291:1019594 [3] NCCL INFO Channel 03/0 : 27[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880679:880980 [0] NCCL INFO Channel 02/0 : 32[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618589:618883 [5] NCCL INFO Channel 03/0 : 29[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618590:618889 [6] NCCL INFO Channel 05/0 : 62[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688201:688500 [1] NCCL INFO Channel 10/0 : 1[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688205:688505 [5] NCCL INFO Channel 02/0 : 5[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688203:688501 [3] NCCL INFO Channel 14/0 : 3[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019292:1019590 [4] NCCL INFO Channel 01/0 : 28[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019294:1019589 [6] NCCL INFO Channel 11/0 : 30[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880686:880979 [7] NCCL INFO Channel 10/0 : 23[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618586:618886 [2] NCCL INFO Channel 13/0 : 26[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688202:688503 [2] NCCL INFO Channel 06/0 : 34[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019288:1019588 [0] NCCL INFO Channel 03/0 : 24[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880683:880982 [4] NCCL INFO Channel 00/0 : 36[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880684:880981 [5] NCCL INFO Channel 02/0 : 21[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618591:618885 [7] NCCL INFO Channel 13/0 : 31[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942617:942916 [6] NCCL INFO Channel 09/0 : 46[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880680:880984 [1] NCCL INFO Channel 10/0 : 17[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618589:618883 [5] NCCL INFO Channel 05/0 : 29[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618590:618889 [6] NCCL INFO Channel 07/0 : 62[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880685:880977 [6] NCCL INFO Channel 08/0 : 38[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618584:618882 [0] NCCL INFO Channel 01/0 : 56[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618587:618887 [3] NCCL INFO Channel 01/0 : 59[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618586:618886 [2] NCCL INFO Channel 15/0 : 26[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880679:880980 [0] NCCL INFO Channel 04/0 : 32[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880686:880979 [7] NCCL INFO Channel 12/0 : 23[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688200:688507 [0] NCCL INFO Channel 06/0 : 32[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019291:1019594 [3] NCCL INFO Channel 05/0 : 27[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880683:880982 [4] NCCL INFO Channel 02/0 : 36[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880684:880981 [5] NCCL INFO Channel 04/0 : 21[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688207:688504 [7] NCCL INFO Channel 12/0 : 7[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942611:942911 [0] NCCL INFO Channel 01/0 : 40[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880680:880984 [1] NCCL INFO Channel 12/0 : 17[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688204:688502 [4] NCCL INFO Channel 10/0 : 36[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942614:942915 [3] NCCL INFO Channel 01/0 : 43[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880685:880977 [6] NCCL INFO Channel 10/0 : 38[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880679:880980 [0] NCCL INFO Channel 06/0 : 32[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618589:618883 [5] NCCL INFO Channel 07/0 : 29[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618590:618889 [6] NCCL INFO Channel 09/0 : 62[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019292:1019590 [4] NCCL INFO Channel 03/0 : 28[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019294:1019589 [6] NCCL INFO Channel 13/0 : 30[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880681:880983 [2] NCCL INFO Channel 00/0 : 34[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618587:618887 [3] NCCL INFO Channel 03/0 : 59[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618588:618888 [4] NCCL INFO Channel 01/0 : 60[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019288:1019588 [0] NCCL INFO Channel 05/0 : 24[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880686:880979 [7] NCCL INFO Channel 14/0 : 23[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019289:1019593 [1] NCCL INFO Channel 01/0 : 25[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019291:1019594 [3] NCCL INFO Channel 07/0 : 27[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880682:880978 [3] NCCL INFO Channel 00/0 : 35[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688205:688505 [5] NCCL INFO Channel 04/0 : 5[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688203:688501 [3] NCCL INFO Channel 00/0 : 35[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688201:688500 [1] NCCL INFO Channel 12/0 : 1[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688202:688503 [2] NCCL INFO Channel 08/0 : 34[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942617:942916 [6] NCCL INFO Channel 11/0 : 46[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880683:880982 [4] NCCL INFO Channel 06/0 : 36[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688200:688507 [0] NCCL INFO Channel 08/0 : 32[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688207:688504 [7] NCCL INFO Channel 14/0 : 7[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942611:942911 [0] NCCL INFO Channel 03/0 : 40[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942615:942912 [4] NCCL INFO Channel 01/0 : 44[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880684:880981 [5] NCCL INFO Channel 06/0 : 21[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618589:618883 [5] NCCL INFO Channel 09/0 : 29[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618590:618889 [6] NCCL INFO Channel 11/0 : 62[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618584:618882 [0] NCCL INFO Channel 03/0 : 56[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688204:688502 [4] NCCL INFO Channel 12/0 : 36[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942614:942915 [3] NCCL INFO Channel 03/0 : 43[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880680:880984 [1] NCCL INFO Channel 14/0 : 17[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618585:618884 [1] NCCL INFO Channel 01/0 : 57[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:94348:94650 [6] NCCL INFO Channel 01/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880685:880977 [6] NCCL INFO Channel 12/0 : 38[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880679:880980 [0] NCCL INFO Channel 08/0 : 32[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880681:880983 [2] NCCL INFO Channel 04/0 : 34[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880682:880978 [3] NCCL INFO Channel 02/0 : 35[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019292:1019590 [4] NCCL INFO Channel 05/0 : 28[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880683:880982 [4] NCCL INFO Channel 08/0 : 36[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942617:942916 [6] NCCL INFO Channel 13/0 : 46[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880684:880981 [5] NCCL INFO Channel 08/0 : 21[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688205:688505 [5] NCCL INFO Channel 06/0 : 5[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688201:688500 [1] NCCL INFO Channel 14/0 : 1[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942612:942914 [1] NCCL INFO Channel 01/0 : 41[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942611:942911 [0] NCCL INFO Channel 05/0 : 40[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880685:880977 [6] NCCL INFO Channel 14/0 : 38[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618587:618887 [3] NCCL INFO Channel 05/0 : 59[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618588:618888 [4] NCCL INFO Channel 03/0 : 60[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688202:688503 [2] NCCL INFO Channel 10/0 : 34[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688203:688501 [3] NCCL INFO Channel 02/0 : 35[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688206:688506 [6] NCCL INFO Channel 00/0 : 54[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942615:942912 [4] NCCL INFO Channel 03/0 : 44[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880679:880980 [0] NCCL INFO Channel 10/0 : 32[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618589:618883 [5] NCCL INFO Channel 11/0 : 29[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618590:618889 [6] NCCL INFO Channel 13/0 : 62[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:94348:94650 [6] NCCL INFO Channel 03/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880682:880978 [3] NCCL INFO Channel 04/0 : 35[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618585:618884 [1] NCCL INFO Channel 03/0 : 57[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618584:618882 [0] NCCL INFO Channel 05/0 : 56[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019294:1019589 [6] NCCL INFO Channel 15/0 : 30[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019295:1019591 [7] NCCL INFO Channel 01/0 : 31[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019288:1019588 [0] NCCL INFO Channel 07/0 : 24[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880681:880983 [2] NCCL INFO Channel 06/0 : 34[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019289:1019593 [1] NCCL INFO Channel 03/0 : 25[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019291:1019594 [3] NCCL INFO Channel 09/0 : 27[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880684:880981 [5] NCCL INFO Channel 10/0 : 21[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880683:880982 [4] NCCL INFO Channel 10/0 : 36[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880686:880979 [7] NCCL INFO Channel 00/0 : 39[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019290:1019595 [2] NCCL INFO Channel 01/0 : 26[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942617:942916 [6] NCCL INFO Channel 15/0 : 46[6] -> 30[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880680:880984 [1] NCCL INFO Channel 00/0 : 33[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688207:688504 [7] NCCL INFO Channel 00/0 : 39[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688200:688507 [0] NCCL INFO Channel 10/0 : 32[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688204:688502 [4] NCCL INFO Channel 14/0 : 36[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942614:942915 [3] NCCL INFO Channel 05/0 : 43[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880679:880980 [0] NCCL INFO Channel 12/0 : 32[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688201:688500 [1] NCCL INFO Channel 00/0 : 33[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688205:688505 [5] NCCL INFO Channel 08/0 : 5[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880682:880978 [3] NCCL INFO Channel 06/0 : 35[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618591:618885 [7] NCCL INFO Channel 01/0 : 63[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618587:618887 [3] NCCL INFO Channel 07/0 : 59[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:94348:94650 [6] NCCL INFO Channel 05/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880681:880983 [2] NCCL INFO Channel 08/0 : 34[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618588:618888 [4] NCCL INFO Channel 05/0 : 60[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019292:1019590 [4] NCCL INFO Channel 07/0 : 28[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880683:880982 [4] NCCL INFO Channel 12/0 : 36[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942612:942914 [1] NCCL INFO Channel 03/0 : 41[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942611:942911 [0] NCCL INFO Channel 07/0 : 40[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880684:880981 [5] NCCL INFO Channel 12/0 : 21[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942615:942912 [4] NCCL INFO Channel 05/0 : 44[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942618:942918 [7] NCCL INFO Channel 01/0 : 47[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880686:880979 [7] NCCL INFO Channel 02/0 : 39[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688202:688503 [2] NCCL INFO Channel 12/0 : 34[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688203:688501 [3] NCCL INFO Channel 04/0 : 35[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688206:688506 [6] NCCL INFO Channel 02/0 : 54[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688207:688504 [7] NCCL INFO Channel 02/0 : 39[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019295:1019591 [7] NCCL INFO Channel 03/0 : 31[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019289:1019593 [1] NCCL INFO Channel 05/0 : 25[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019291:1019594 [3] NCCL INFO Channel 13/0 : 27[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019288:1019588 [0] NCCL INFO Channel 09/0 : 24[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942614:942915 [3] NCCL INFO Channel 07/0 : 43[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880680:880984 [1] NCCL INFO Channel 02/0 : 33[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880679:880980 [0] NCCL INFO Channel 14/0 : 32[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618590:618889 [6] NCCL INFO Channel 15/0 : 62[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618589:618883 [5] NCCL INFO Channel 15/0 : 29[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:94348:94650 [6] NCCL INFO Channel 07/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019290:1019595 [2] NCCL INFO Channel 03/0 : 26[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880682:880978 [3] NCCL INFO Channel 08/0 : 35[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618585:618884 [1] NCCL INFO Channel 05/0 : 57[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019292:1019590 [4] NCCL INFO Channel 09/0 : 28[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880681:880983 [2] NCCL INFO Channel 10/0 : 34[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618584:618882 [0] NCCL INFO Channel 07/0 : 56[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880683:880982 [4] NCCL INFO Channel 14/0 : 36[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618586:618886 [2] NCCL INFO Channel 01/0 : 58[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942613:942917 [2] NCCL INFO Channel 01/0 : 42[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880684:880981 [5] NCCL INFO Channel 14/0 : 21[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942611:942911 [0] NCCL INFO Channel 09/0 : 40[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880686:880979 [7] NCCL INFO Channel 04/0 : 39[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688201:688500 [1] NCCL INFO Channel 02/0 : 33[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688205:688505 [5] NCCL INFO Channel 10/0 : 5[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942612:942914 [1] NCCL INFO Channel 05/0 : 41[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942615:942912 [4] NCCL INFO Channel 07/0 : 44[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880680:880984 [1] NCCL INFO Channel 04/0 : 33[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688200:688507 [0] NCCL INFO Channel 12/0 : 32[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942618:942918 [7] NCCL INFO Channel 03/0 : 47[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942614:942915 [3] NCCL INFO Channel 09/0 : 43[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880682:880978 [3] NCCL INFO Channel 10/0 : 35[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688203:688501 [3] NCCL INFO Channel 06/0 : 35[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:94348:94650 [6] NCCL INFO Channel 09/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019295:1019591 [7] NCCL INFO Channel 05/0 : 31[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880681:880983 [2] NCCL INFO Channel 12/0 : 34[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618591:618885 [7] NCCL INFO Channel 03/0 : 63[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618587:618887 [3] NCCL INFO Channel 09/0 : 59[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688202:688503 [2] NCCL INFO Channel 14/0 : 34[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688207:688504 [7] NCCL INFO Channel 04/0 : 39[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688206:688506 [6] NCCL INFO Channel 04/0 : 54[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:94346:94644 [4] NCCL INFO Channel 01/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880686:880979 [7] NCCL INFO Channel 06/0 : 39[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618588:618888 [4] NCCL INFO Channel 07/0 : 60[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942611:942911 [0] NCCL INFO Channel 11/0 : 40[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942613:942917 [2] NCCL INFO Channel 03/0 : 42[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880680:880984 [1] NCCL INFO Channel 06/0 : 33[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618590:618889 [6] NCCL INFO Channel 01/0 : 30[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019288:1019588 [0] NCCL INFO Channel 11/0 : 24[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019291:1019594 [3] NCCL INFO Channel 15/0 : 27[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019289:1019593 [1] NCCL INFO Channel 07/0 : 25[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880682:880978 [3] NCCL INFO Channel 12/0 : 35[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618585:618884 [1] NCCL INFO Channel 07/0 : 57[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618584:618882 [0] NCCL INFO Channel 09/0 : 56[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688201:688500 [1] NCCL INFO Channel 04/0 : 33[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:94348:94650 [6] NCCL INFO Channel 11/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019292:1019590 [4] NCCL INFO Channel 11/0 : 28[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880681:880983 [2] NCCL INFO Channel 14/0 : 34[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618586:618886 [2] NCCL INFO Channel 03/0 : 58[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688205:688505 [5] NCCL INFO Channel 12/0 : 5[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:94346:94644 [4] NCCL INFO Channel 03/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019290:1019595 [2] NCCL INFO Channel 05/0 : 26[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880686:880979 [7] NCCL INFO Channel 08/0 : 39[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618591:618885 [7] NCCL INFO Channel 05/0 : 63[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618587:618887 [3] NCCL INFO Channel 13/0 : 59[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688200:688507 [0] NCCL INFO Channel 14/0 : 32[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688204:688502 [4] NCCL INFO Channel 00/0 : 52[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942612:942914 [1] NCCL INFO Channel 07/0 : 41[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880680:880984 [1] NCCL INFO Channel 08/0 : 33[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618588:618888 [4] NCCL INFO Channel 09/0 : 60[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688203:688501 [3] NCCL INFO Channel 08/0 : 35[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688207:688504 [7] NCCL INFO Channel 06/0 : 39[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688206:688506 [6] NCCL INFO Channel 08/0 : 54[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942615:942912 [4] NCCL INFO Channel 09/0 : 44[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880682:880978 [3] NCCL INFO Channel 14/0 : 35[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688201:688500 [1] NCCL INFO Channel 06/0 : 33[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:94348:94650 [6] NCCL INFO Channel 13/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:94346:94644 [4] NCCL INFO Channel 05/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019295:1019591 [7] NCCL INFO Channel 07/0 : 31[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942614:942915 [3] NCCL INFO Channel 13/0 : 43[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880684:880981 [5] NCCL INFO Channel 00/0 : 37[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019288:1019588 [0] NCCL INFO Channel 13/0 : 24[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942618:942918 [7] NCCL INFO Channel 05/0 : 47[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942611:942911 [0] NCCL INFO Channel 13/0 : 40[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880686:880979 [7] NCCL INFO Channel 10/0 : 39[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019289:1019593 [1] NCCL INFO Channel 11/0 : 25[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019292:1019590 [4] NCCL INFO Channel 13/0 : 28[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019293:1019592 [5] NCCL INFO Channel 01/0 : 29[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942613:942917 [2] NCCL INFO Channel 05/0 : 42[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880680:880984 [1] NCCL INFO Channel 10/0 : 33[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618590:618889 [6] NCCL INFO Channel 03/0 : 30[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:94344:94648 [2] NCCL INFO Channel 01/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019290:1019595 [2] NCCL INFO Channel 07/0 : 26[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880684:880981 [5] NCCL INFO Channel 02/0 : 37[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618585:618884 [1] NCCL INFO Channel 11/0 : 57[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618584:618882 [0] NCCL INFO Channel 11/0 : 56[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880686:880979 [7] NCCL INFO Channel 12/0 : 39[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618586:618886 [2] NCCL INFO Channel 05/0 : 58[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688205:688505 [5] NCCL INFO Channel 14/0 : 5[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880680:880984 [1] NCCL INFO Channel 12/0 : 33[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618591:618885 [7] NCCL INFO Channel 07/0 : 63[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618587:618887 [3] NCCL INFO Channel 15/0 : 59[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688204:688502 [4] NCCL INFO Channel 02/0 : 52[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:94346:94644 [4] NCCL INFO Channel 07/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942612:942914 [1] NCCL INFO Channel 11/0 : 41[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880684:880981 [5] NCCL INFO Channel 04/0 : 37[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618588:618888 [4] NCCL INFO Channel 11/0 : 60[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688207:688504 [7] NCCL INFO Channel 08/0 : 39[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688203:688501 [3] NCCL INFO Channel 10/0 : 35[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688206:688506 [6] NCCL INFO Channel 10/0 : 54[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:94348:94650 [6] NCCL INFO Channel 15/0 : 14[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019295:1019591 [7] NCCL INFO Channel 09/0 : 31[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942615:942912 [4] NCCL INFO Channel 11/0 : 44[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942614:942915 [3] NCCL INFO Channel 15/0 : 43[3] -> 27[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880686:880979 [7] NCCL INFO Channel 14/0 : 39[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618590:618889 [6] NCCL INFO Channel 05/0 : 30[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688201:688500 [1] NCCL INFO Channel 08/0 : 33[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:94342:94643 [0] NCCL INFO Channel 01/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019288:1019588 [0] NCCL INFO Channel 15/0 : 24[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942611:942911 [0] NCCL INFO Channel 15/0 : 40[0] -> 24[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880680:880984 [1] NCCL INFO Channel 14/0 : 33[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618584:618882 [0] NCCL INFO Channel 13/0 : 56[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618585:618884 [1] NCCL INFO Channel 13/0 : 57[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:94344:94648 [2] NCCL INFO Channel 03/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019292:1019590 [4] NCCL INFO Channel 15/0 : 28[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942613:942917 [2] NCCL INFO Channel 07/0 : 42[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942618:942918 [7] NCCL INFO Channel 07/0 : 47[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618586:618886 [2] NCCL INFO Channel 07/0 : 58[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:94346:94644 [4] NCCL INFO Channel 09/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019289:1019593 [1] NCCL INFO Channel 13/0 : 25[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019293:1019592 [5] NCCL INFO Channel 03/0 : 29[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688205:688505 [5] NCCL INFO Channel 00/0 : 37[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688204:688502 [4] NCCL INFO Channel 06/0 : 52[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688202:688503 [2] NCCL INFO Channel 00/0 : 50[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:94342:94643 [0] NCCL INFO Channel 03/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:94344:94648 [2] NCCL INFO Channel 05/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019290:1019595 [2] NCCL INFO Channel 09/0 : 26[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880684:880981 [5] NCCL INFO Channel 06/0 : 37[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688200:688507 [0] NCCL INFO Channel 02/0 : 48[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019295:1019591 [7] NCCL INFO Channel 11/0 : 31[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880684:880981 [5] NCCL INFO Channel 08/0 : 37[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688207:688504 [7] NCCL INFO Channel 10/0 : 39[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688206:688506 [6] NCCL INFO Channel 12/0 : 54[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:94346:94644 [4] NCCL INFO Channel 11/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942612:942914 [1] NCCL INFO Channel 13/0 : 41[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942616:942913 [5] NCCL INFO Channel 01/0 : 45[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880684:880981 [5] NCCL INFO Channel 10/0 : 37[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618587:618887 [3] NCCL INFO Channel 01/0 : 27[3] -> 59[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688201:688500 [1] NCCL INFO Channel 10/0 : 33[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942615:942912 [4] NCCL INFO Channel 13/0 : 44[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880684:880981 [5] NCCL INFO Channel 12/0 : 37[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618591:618885 [7] NCCL INFO Channel 09/0 : 63[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618588:618888 [4] NCCL INFO Channel 13/0 : 60[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688203:688501 [3] NCCL INFO Channel 12/0 : 35[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019293:1019592 [5] NCCL INFO Channel 05/0 : 29[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942613:942917 [2] NCCL INFO Channel 09/0 : 42[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880684:880981 [5] NCCL INFO Channel 14/0 : 37[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618590:618889 [6] NCCL INFO Channel 07/0 : 30[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019289:1019593 [1] NCCL INFO Channel 15/0 : 25[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019295:1019591 [7] NCCL INFO Channel 13/0 : 31[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019290:1019595 [2] NCCL INFO Channel 11/0 : 26[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942618:942918 [7] NCCL INFO Channel 09/0 : 47[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618589:618883 [5] NCCL INFO Channel 01/0 : 61[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019293:1019592 [5] NCCL INFO Channel 07/0 : 29[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942612:942914 [1] NCCL INFO Channel 15/0 : 41[1] -> 25[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618584:618882 [0] NCCL INFO Channel 15/0 : 56[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942615:942912 [4] NCCL INFO Channel 15/0 : 44[4] -> 28[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942616:942913 [5] NCCL INFO Channel 03/0 : 45[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618585:618884 [1] NCCL INFO Channel 15/0 : 57[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:94344:94648 [2] NCCL INFO Channel 07/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618586:618886 [2] NCCL INFO Channel 09/0 : 58[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688204:688502 [4] NCCL INFO Channel 08/0 : 52[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688205:688505 [5] NCCL INFO Channel 02/0 : 37[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:94342:94643 [0] NCCL INFO Channel 05/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:94346:94644 [4] NCCL INFO Channel 13/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019290:1019595 [2] NCCL INFO Channel 13/0 : 26[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618587:618887 [3] NCCL INFO Channel 03/0 : 27[3] -> 59[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688202:688503 [2] NCCL INFO Channel 04/0 : 50[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688200:688507 [0] NCCL INFO Channel 04/0 : 48[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019293:1019592 [5] NCCL INFO Channel 09/0 : 29[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618588:618888 [4] NCCL INFO Channel 15/0 : 60[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688207:688504 [7] NCCL INFO Channel 12/0 : 39[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688206:688506 [6] NCCL INFO Channel 14/0 : 54[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942613:942917 [2] NCCL INFO Channel 11/0 : 42[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618591:618885 [7] NCCL INFO Channel 11/0 : 63[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688201:688500 [1] NCCL INFO Channel 12/0 : 33[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688203:688501 [3] NCCL INFO Channel 14/0 : 35[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942618:942918 [7] NCCL INFO Channel 11/0 : 47[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618590:618889 [6] NCCL INFO Channel 09/0 : 30[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688204:688502 [4] NCCL INFO Channel 10/0 : 52[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688205:688505 [5] NCCL INFO Channel 04/0 : 37[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942616:942913 [5] NCCL INFO Channel 05/0 : 45[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618589:618883 [5] NCCL INFO Channel 03/0 : 61[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942613:942917 [2] NCCL INFO Channel 13/0 : 42[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618584:618882 [0] NCCL INFO Channel 01/0 : 24[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618585:618884 [1] NCCL INFO Channel 01/0 : 25[1] -> 57[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942618:942918 [7] NCCL INFO Channel 13/0 : 47[7] -> 31[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019290:1019595 [2] NCCL INFO Channel 15/0 : 26[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942616:942913 [5] NCCL INFO Channel 07/0 : 45[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:94342:94643 [0] NCCL INFO Channel 07/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:94346:94644 [4] NCCL INFO Channel 15/0 : 12[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:94344:94648 [2] NCCL INFO Channel 09/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019293:1019592 [5] NCCL INFO Channel 11/0 : 29[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618587:618887 [3] NCCL INFO Channel 05/0 : 27[3] -> 59[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688202:688503 [2] NCCL INFO Channel 06/0 : 50[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618588:618888 [4] NCCL INFO Channel 01/0 : 28[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688200:688507 [0] NCCL INFO Channel 06/0 : 48[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688207:688504 [7] NCCL INFO Channel 14/0 : 39[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688206:688506 [6] NCCL INFO Channel 00/0 : 38[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618586:618886 [2] NCCL INFO Channel 11/0 : 58[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618591:618885 [7] NCCL INFO Channel 13/0 : 63[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688201:688500 [1] NCCL INFO Channel 14/0 : 33[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618590:618889 [6] NCCL INFO Channel 11/0 : 30[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618589:618883 [5] NCCL INFO Channel 05/0 : 61[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688205:688505 [5] NCCL INFO Channel 06/0 : 37[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688204:688502 [4] NCCL INFO Channel 12/0 : 52[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942613:942917 [2] NCCL INFO Channel 15/0 : 42[2] -> 26[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618585:618884 [1] NCCL INFO Channel 03/0 : 25[1] -> 57[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688202:688503 [2] NCCL INFO Channel 08/0 : 50[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:94345:94649 [3] NCCL INFO Channel 01/0 : 11[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019293:1019592 [5] NCCL INFO Channel 15/0 : 29[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942616:942913 [5] NCCL INFO Channel 09/0 : 45[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618584:618882 [0] NCCL INFO Channel 03/0 : 24[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618587:618887 [3] NCCL INFO Channel 07/0 : 27[3] -> 59[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688206:688506 [6] NCCL INFO Channel 02/0 : 38[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688200:688507 [0] NCCL INFO Channel 08/0 : 48[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:94342:94643 [0] NCCL INFO Channel 09/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618588:618888 [4] NCCL INFO Channel 03/0 : 28[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:94344:94648 [2] NCCL INFO Channel 11/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942616:942913 [5] NCCL INFO Channel 11/0 : 45[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:94345:94649 [3] NCCL INFO Channel 03/0 : 11[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688205:688505 [5] NCCL INFO Channel 08/0 : 37[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618585:618884 [1] NCCL INFO Channel 05/0 : 25[1] -> 57[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618586:618886 [2] NCCL INFO Channel 13/0 : 58[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618590:618889 [6] NCCL INFO Channel 13/0 : 30[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688204:688502 [4] NCCL INFO Channel 14/0 : 52[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:94342:94643 [0] NCCL INFO Channel 11/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618591:618885 [7] NCCL INFO Channel 01/0 : 31[7] -> 63[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618589:618883 [5] NCCL INFO Channel 07/0 : 61[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618584:618882 [0] NCCL INFO Channel 05/0 : 24[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688203:688501 [3] NCCL INFO Channel 00/0 : 51[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688202:688503 [2] NCCL INFO Channel 10/0 : 50[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618587:618887 [3] NCCL INFO Channel 09/0 : 27[3] -> 59[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618588:618888 [4] NCCL INFO Channel 05/0 : 28[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688206:688506 [6] NCCL INFO Channel 04/0 : 38[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942616:942913 [5] NCCL INFO Channel 15/0 : 45[5] -> 29[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688200:688507 [0] NCCL INFO Channel 10/0 : 48[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:94343:94647 [1] NCCL INFO Channel 01/0 : 9[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:94344:94648 [2] NCCL INFO Channel 13/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618585:618884 [1] NCCL INFO Channel 07/0 : 25[1] -> 57[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:94349:94645 [7] NCCL INFO Channel 01/0 : 15[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:94345:94649 [3] NCCL INFO Channel 05/0 : 11[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688205:688505 [5] NCCL INFO Channel 10/0 : 37[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688204:688502 [4] NCCL INFO Channel 00/0 : 36[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:94342:94643 [0] NCCL INFO Channel 13/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688202:688503 [2] NCCL INFO Channel 12/0 : 50[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688207:688504 [7] NCCL INFO Channel 00/0 : 55[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688206:688506 [6] NCCL INFO Channel 08/0 : 38[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618586:618886 [2] NCCL INFO Channel 15/0 : 58[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688203:688501 [3] NCCL INFO Channel 02/0 : 51[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688200:688507 [0] NCCL INFO Channel 12/0 : 48[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688201:688500 [1] NCCL INFO Channel 00/0 : 49[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618590:618889 [6] NCCL INFO Channel 15/0 : 30[6] -> 62[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688205:688505 [5] NCCL INFO Channel 12/0 : 37[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688204:688502 [4] NCCL INFO Channel 02/0 : 36[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618584:618882 [0] NCCL INFO Channel 07/0 : 24[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:94343:94647 [1] NCCL INFO Channel 03/0 : 9[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:94344:94648 [2] NCCL INFO Channel 15/0 : 10[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618591:618885 [7] NCCL INFO Channel 03/0 : 31[7] -> 63[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618587:618887 [3] NCCL INFO Channel 13/0 : 27[3] -> 59[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618588:618888 [4] NCCL INFO Channel 07/0 : 28[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618589:618883 [5] NCCL INFO Channel 09/0 : 61[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:94345:94649 [3] NCCL INFO Channel 07/0 : 11[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:94342:94643 [0] NCCL INFO Channel 15/0 : 8[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:94349:94645 [7] NCCL INFO Channel 03/0 : 15[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688202:688503 [2] NCCL INFO Channel 14/0 : 50[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618585:618884 [1] NCCL INFO Channel 11/0 : 25[1] -> 57[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:94343:94647 [1] NCCL INFO Channel 05/0 : 9[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618586:618886 [2] NCCL INFO Channel 01/0 : 26[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:94345:94649 [3] NCCL INFO Channel 09/0 : 11[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688206:688506 [6] NCCL INFO Channel 10/0 : 38[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688207:688504 [7] NCCL INFO Channel 02/0 : 55[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:94349:94645 [7] NCCL INFO Channel 05/0 : 15[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688203:688501 [3] NCCL INFO Channel 04/0 : 51[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688205:688505 [5] NCCL INFO Channel 14/0 : 37[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618584:618882 [0] NCCL INFO Channel 09/0 : 24[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688201:688500 [1] NCCL INFO Channel 02/0 : 49[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688204:688502 [4] NCCL INFO Channel 06/0 : 36[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618591:618885 [7] NCCL INFO Channel 05/0 : 31[7] -> 63[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618588:618888 [4] NCCL INFO Channel 09/0 : 28[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688200:688507 [0] NCCL INFO Channel 14/0 : 48[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:94343:94647 [1] NCCL INFO Channel 07/0 : 9[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618587:618887 [3] NCCL INFO Channel 15/0 : 27[3] -> 59[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618589:618883 [5] NCCL INFO Channel 11/0 : 61[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618585:618884 [1] NCCL INFO Channel 13/0 : 25[1] -> 57[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688202:688503 [2] NCCL INFO Channel 00/0 : 34[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618586:618886 [2] NCCL INFO Channel 03/0 : 26[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:94345:94649 [3] NCCL INFO Channel 13/0 : 11[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:94349:94645 [7] NCCL INFO Channel 07/0 : 15[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688206:688506 [6] NCCL INFO Channel 12/0 : 38[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:94343:94647 [1] NCCL INFO Channel 11/0 : 9[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688207:688504 [7] NCCL INFO Channel 04/0 : 55[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:94345:94649 [3] NCCL INFO Channel 15/0 : 11[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618584:618882 [0] NCCL INFO Channel 11/0 : 24[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:94349:94645 [7] NCCL INFO Channel 09/0 : 15[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688203:688501 [3] NCCL INFO Channel 06/0 : 51[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688204:688502 [4] NCCL INFO Channel 08/0 : 36[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688201:688500 [1] NCCL INFO Channel 04/0 : 49[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688200:688507 [0] NCCL INFO Channel 02/0 : 32[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688202:688503 [2] NCCL INFO Channel 04/0 : 34[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:94343:94647 [1] NCCL INFO Channel 13/0 : 9[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688206:688506 [6] NCCL INFO Channel 14/0 : 38[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-7:811254:811557 [6] NCCL INFO Channel 00/0 : 62[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618590:618889 [6] NCCL INFO Channel 01/0 : 46[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688207:688504 [7] NCCL INFO Channel 06/0 : 55[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618585:618884 [1] NCCL INFO Channel 15/0 : 25[1] -> 57[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:94347:94646 [5] NCCL INFO Channel 01/0 : 13[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618588:618888 [4] NCCL INFO Channel 11/0 : 28[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618591:618885 [7] NCCL INFO Channel 07/0 : 31[7] -> 63[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618586:618886 [2] NCCL INFO Channel 05/0 : 26[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618589:618883 [5] NCCL INFO Channel 15/0 : 61[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:94349:94645 [7] NCCL INFO Channel 11/0 : 15[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618584:618882 [0] NCCL INFO Channel 13/0 : 24[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:94343:94647 [1] NCCL INFO Channel 15/0 : 9[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-7:811254:811557 [6] NCCL INFO Channel 02/0 : 62[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688203:688501 [3] NCCL INFO Channel 08/0 : 51[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688205:688505 [5] NCCL INFO Channel 00/0 : 53[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688204:688502 [4] NCCL INFO Channel 10/0 : 36[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688200:688507 [0] NCCL INFO Channel 04/0 : 32[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688201:688500 [1] NCCL INFO Channel 06/0 : 49[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688202:688503 [2] NCCL INFO Channel 06/0 : 34[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:94347:94646 [5] NCCL INFO Channel 03/0 : 13[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618590:618889 [6] NCCL INFO Channel 03/0 : 46[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688207:688504 [7] NCCL INFO Channel 08/0 : 55[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:94349:94645 [7] NCCL INFO Channel 13/0 : 15[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-7:811251:811556 [3] NCCL INFO Channel 00/0 : 59[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618588:618888 [4] NCCL INFO Channel 13/0 : 28[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618586:618886 [2] NCCL INFO Channel 07/0 : 26[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618591:618885 [7] NCCL INFO Channel 09/0 : 31[7] -> 63[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618584:618882 [0] NCCL INFO Channel 15/0 : 24[0] -> 56[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618589:618883 [5] NCCL INFO Channel 01/0 : 29[5] -> 61[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688203:688501 [3] NCCL INFO Channel 10/0 : 51[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:94347:94646 [5] NCCL INFO Channel 05/0 : 13[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618587:618887 [3] NCCL INFO Channel 01/0 : 43[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618590:618889 [6] NCCL INFO Channel 05/0 : 46[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-7:811254:811557 [6] NCCL INFO Channel 04/0 : 62[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-7:811251:811556 [3] NCCL INFO Channel 02/0 : 59[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688205:688505 [5] NCCL INFO Channel 02/0 : 53[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688204:688502 [4] NCCL INFO Channel 12/0 : 36[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688200:688507 [0] NCCL INFO Channel 06/0 : 32[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688207:688504 [7] NCCL INFO Channel 10/0 : 55[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688201:688500 [1] NCCL INFO Channel 08/0 : 49[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688202:688503 [2] NCCL INFO Channel 08/0 : 34[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688203:688501 [3] NCCL INFO Channel 12/0 : 51[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:94347:94646 [5] NCCL INFO Channel 07/0 : 13[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618588:618888 [4] NCCL INFO Channel 15/0 : 28[4] -> 60[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-7:811254:811557 [6] NCCL INFO Channel 08/0 : 62[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-7:811251:811556 [3] NCCL INFO Channel 04/0 : 59[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618586:618886 [2] NCCL INFO Channel 09/0 : 26[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688204:688502 [4] NCCL INFO Channel 14/0 : 36[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:94347:94646 [5] NCCL INFO Channel 09/0 : 13[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618591:618885 [7] NCCL INFO Channel 11/0 : 31[7] -> 63[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688205:688505 [5] NCCL INFO Channel 04/0 : 53[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618585:618884 [1] NCCL INFO Channel 01/0 : 41[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618589:618883 [5] NCCL INFO Channel 03/0 : 29[5] -> 61[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618587:618887 [3] NCCL INFO Channel 03/0 : 43[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618590:618889 [6] NCCL INFO Channel 07/0 : 46[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-7:811254:811557 [6] NCCL INFO Channel 10/0 : 62[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-7:811249:811559 [1] NCCL INFO Channel 00/0 : 57[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-7:811251:811556 [3] NCCL INFO Channel 06/0 : 59[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880685:880977 [6] NCCL INFO Channel 00/0 : 30[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688200:688507 [0] NCCL INFO Channel 08/0 : 32[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688207:688504 [7] NCCL INFO Channel 12/0 : 55[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688201:688500 [1] NCCL INFO Channel 10/0 : 49[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:94347:94646 [5] NCCL INFO Channel 11/0 : 13[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618591:618885 [7] NCCL INFO Channel 13/0 : 31[7] -> 63[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688202:688503 [2] NCCL INFO Channel 10/0 : 34[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688203:688501 [3] NCCL INFO Channel 14/0 : 51[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688206:688506 [6] NCCL INFO Channel 01/0 : 46[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688205:688505 [5] NCCL INFO Channel 06/0 : 53[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-7:811254:811557 [6] NCCL INFO Channel 12/0 : 62[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-7:811249:811559 [1] NCCL INFO Channel 02/0 : 57[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880685:880977 [6] NCCL INFO Channel 02/0 : 30[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:94347:94646 [5] NCCL INFO Channel 15/0 : 13[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618586:618886 [2] NCCL INFO Channel 11/0 : 26[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618589:618883 [5] NCCL INFO Channel 05/0 : 29[5] -> 61[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618585:618884 [1] NCCL INFO Channel 03/0 : 41[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618587:618887 [3] NCCL INFO Channel 05/0 : 43[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618590:618889 [6] NCCL INFO Channel 09/0 : 46[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688207:688504 [7] NCCL INFO Channel 14/0 : 55[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618584:618882 [0] NCCL INFO Channel 01/0 : 40[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-7:811251:811556 [3] NCCL INFO Channel 08/0 : 59[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-7:811248:811554 [0] NCCL INFO Channel 02/0 : 56[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618588:618888 [4] NCCL INFO Channel 01/0 : 44[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-7:811254:811557 [6] NCCL INFO Channel 14/0 : 62[6] -> 54[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880685:880977 [6] NCCL INFO Channel 04/0 : 30[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688202:688503 [2] NCCL INFO Channel 12/0 : 34[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-7:811249:811559 [1] NCCL INFO Channel 04/0 : 57[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-7:811251:811556 [3] NCCL INFO Channel 10/0 : 59[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-7:811248:811554 [0] NCCL INFO Channel 04/0 : 56[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-7:811252:811560 [4] NCCL INFO Channel 00/0 : 60[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688205:688505 [5] NCCL INFO Channel 08/0 : 53[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618589:618883 [5] NCCL INFO Channel 07/0 : 29[5] -> 61[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688202:688503 [2] NCCL INFO Channel 14/0 : 34[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880685:880977 [6] NCCL INFO Channel 08/0 : 30[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-7:811249:811559 [1] NCCL INFO Channel 06/0 : 57[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-7:811251:811556 [3] NCCL INFO Channel 12/0 : 59[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618585:618884 [1] NCCL INFO Channel 05/0 : 41[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618586:618886 [2] NCCL INFO Channel 13/0 : 26[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618587:618887 [3] NCCL INFO Channel 07/0 : 43[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618590:618889 [6] NCCL INFO Channel 11/0 : 46[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618584:618882 [0] NCCL INFO Channel 03/0 : 40[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688205:688505 [5] NCCL INFO Channel 10/0 : 53[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618588:618888 [4] NCCL INFO Channel 03/0 : 44[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618591:618885 [7] NCCL INFO Channel 01/0 : 47[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880685:880977 [6] NCCL INFO Channel 10/0 : 30[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-7:811248:811554 [0] NCCL INFO Channel 06/0 : 56[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-7:811252:811560 [4] NCCL INFO Channel 02/0 : 60[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618589:618883 [5] NCCL INFO Channel 09/0 : 29[5] -> 61[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688203:688501 [3] NCCL INFO Channel 00/0 : 35[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688207:688504 [7] NCCL INFO Channel 00/0 : 39[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-7:811249:811559 [1] NCCL INFO Channel 08/0 : 57[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-7:811255:811561 [7] NCCL INFO Channel 00/0 : 63[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-7:811251:811556 [3] NCCL INFO Channel 14/0 : 59[3] -> 51[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880685:880977 [6] NCCL INFO Channel 12/0 : 30[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618585:618884 [1] NCCL INFO Channel 07/0 : 41[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618586:618886 [2] NCCL INFO Channel 15/0 : 26[2] -> 58[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688205:688505 [5] NCCL INFO Channel 12/0 : 53[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-7:811248:811554 [0] NCCL INFO Channel 08/0 : 56[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-7:811252:811560 [4] NCCL INFO Channel 06/0 : 60[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-7:811249:811559 [1] NCCL INFO Channel 10/0 : 57[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-7:811255:811561 [7] NCCL INFO Channel 02/0 : 63[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-7:811248:811554 [0] NCCL INFO Channel 10/0 : 56[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880685:880977 [6] NCCL INFO Channel 14/0 : 30[6] -> 22[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618587:618887 [3] NCCL INFO Channel 09/0 : 43[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618590:618889 [6] NCCL INFO Channel 13/0 : 46[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618584:618882 [0] NCCL INFO Channel 05/0 : 40[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618588:618888 [4] NCCL INFO Channel 05/0 : 44[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688207:688504 [7] NCCL INFO Channel 02/0 : 39[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618591:618885 [7] NCCL INFO Channel 03/0 : 47[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618589:618883 [5] NCCL INFO Channel 11/0 : 29[5] -> 61[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688203:688501 [3] NCCL INFO Channel 02/0 : 35[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688205:688505 [5] NCCL INFO Channel 14/0 : 53[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618585:618884 [1] NCCL INFO Channel 11/0 : 41[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880685:880977 [6] NCCL INFO Channel 00/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688207:688504 [7] NCCL INFO Channel 04/0 : 39[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688200:688507 [0] NCCL INFO Channel 10/0 : 32[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688205:688505 [5] NCCL INFO Channel 00/0 : 37[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618587:618887 [3] NCCL INFO Channel 13/0 : 43[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618584:618882 [0] NCCL INFO Channel 07/0 : 40[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618590:618889 [6] NCCL INFO Channel 15/0 : 46[6] -> 30[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880681:880983 [2] NCCL INFO Channel 00/0 : 26[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688202:688503 [2] NCCL INFO Channel 01/0 : 42[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688203:688501 [3] NCCL INFO Channel 04/0 : 35[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618589:618883 [5] NCCL INFO Channel 15/0 : 29[5] -> 61[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618588:618888 [4] NCCL INFO Channel 07/0 : 44[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618591:618885 [7] NCCL INFO Channel 05/0 : 47[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618585:618884 [1] NCCL INFO Channel 13/0 : 41[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688207:688504 [7] NCCL INFO Channel 06/0 : 39[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618586:618886 [2] NCCL INFO Channel 01/0 : 42[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880685:880977 [6] NCCL INFO Channel 01/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618587:618887 [3] NCCL INFO Channel 15/0 : 43[3] -> 27[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688201:688500 [1] NCCL INFO Channel 12/0 : 49[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688200:688507 [0] NCCL INFO Channel 12/0 : 32[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688205:688505 [5] NCCL INFO Channel 02/0 : 37[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688202:688503 [2] NCCL INFO Channel 03/0 : 42[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880681:880983 [2] NCCL INFO Channel 04/0 : 26[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618584:618882 [0] NCCL INFO Channel 09/0 : 40[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688207:688504 [7] NCCL INFO Channel 08/0 : 39[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618590:618889 [6] NCCL INFO Channel 01/0 : 30[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688204:688502 [4] NCCL INFO Channel 01/0 : 44[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880685:880977 [6] NCCL INFO Channel 02/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880683:880982 [4] NCCL INFO Channel 00/0 : 28[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618585:618884 [1] NCCL INFO Channel 15/0 : 41[1] -> 25[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880681:880983 [2] NCCL INFO Channel 06/0 : 26[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618588:618888 [4] NCCL INFO Channel 09/0 : 44[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688201:688500 [1] NCCL INFO Channel 14/0 : 49[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688205:688505 [5] NCCL INFO Channel 04/0 : 37[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618591:618885 [7] NCCL INFO Channel 07/0 : 47[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688203:688501 [3] NCCL INFO Channel 06/0 : 35[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618587:618887 [3] NCCL INFO Channel 01/0 : 27[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618586:618886 [2] NCCL INFO Channel 03/0 : 42[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688200:688507 [0] NCCL INFO Channel 14/0 : 32[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926596:926899 [2] NCCL INFO Channel 00/0 : 58[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688202:688503 [2] NCCL INFO Channel 05/0 : 42[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880685:880977 [6] NCCL INFO Channel 03/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880683:880982 [4] NCCL INFO Channel 02/0 : 28[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880681:880983 [2] NCCL INFO Channel 08/0 : 26[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618584:618882 [0] NCCL INFO Channel 11/0 : 40[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618590:618889 [6] NCCL INFO Channel 03/0 : 30[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926598:926894 [4] NCCL INFO Channel 00/0 : 60[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618585:618884 [1] NCCL INFO Channel 01/0 : 25[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688207:688504 [7] NCCL INFO Channel 10/0 : 39[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688204:688502 [4] NCCL INFO Channel 03/0 : 44[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688206:688506 [6] NCCL INFO Channel 03/0 : 46[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880685:880977 [6] NCCL INFO Channel 04/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688201:688500 [1] NCCL INFO Channel 00/0 : 33[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688205:688505 [5] NCCL INFO Channel 06/0 : 37[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926596:926899 [2] NCCL INFO Channel 04/0 : 58[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618589:618883 [5] NCCL INFO Channel 01/0 : 45[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926598:926894 [4] NCCL INFO Channel 02/0 : 60[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618588:618888 [4] NCCL INFO Channel 11/0 : 44[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618586:618886 [2] NCCL INFO Channel 05/0 : 42[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618587:618887 [3] NCCL INFO Channel 03/0 : 27[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618591:618885 [7] NCCL INFO Channel 09/0 : 47[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880683:880982 [4] NCCL INFO Channel 06/0 : 28[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880681:880983 [2] NCCL INFO Channel 10/0 : 26[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618584:618882 [0] NCCL INFO Channel 13/0 : 40[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618590:618889 [6] NCCL INFO Channel 05/0 : 30[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618585:618884 [1] NCCL INFO Channel 03/0 : 25[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926596:926899 [2] NCCL INFO Channel 06/0 : 58[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880685:880977 [6] NCCL INFO Channel 05/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618589:618883 [5] NCCL INFO Channel 03/0 : 45[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926600:926893 [6] NCCL INFO Channel 00/0 : 62[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926598:926894 [4] NCCL INFO Channel 06/0 : 60[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880681:880983 [2] NCCL INFO Channel 12/0 : 26[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880683:880982 [4] NCCL INFO Channel 08/0 : 28[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880685:880977 [6] NCCL INFO Channel 07/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618587:618887 [3] NCCL INFO Channel 05/0 : 27[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618586:618886 [2] NCCL INFO Channel 07/0 : 42[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618591:618885 [7] NCCL INFO Channel 11/0 : 47[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618588:618888 [4] NCCL INFO Channel 13/0 : 44[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926596:926899 [2] NCCL INFO Channel 08/0 : 58[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618584:618882 [0] NCCL INFO Channel 15/0 : 40[0] -> 24[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880681:880983 [2] NCCL INFO Channel 14/0 : 26[2] -> 18[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926598:926894 [4] NCCL INFO Channel 08/0 : 60[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926600:926893 [6] NCCL INFO Channel 02/0 : 62[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618585:618884 [1] NCCL INFO Channel 05/0 : 25[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618590:618889 [6] NCCL INFO Channel 07/0 : 30[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880683:880982 [4] NCCL INFO Channel 10/0 : 28[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926596:926899 [2] NCCL INFO Channel 10/0 : 58[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926600:926893 [6] NCCL INFO Channel 04/0 : 62[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618589:618883 [5] NCCL INFO Channel 05/0 : 45[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880685:880977 [6] NCCL INFO Channel 08/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926598:926894 [4] NCCL INFO Channel 10/0 : 60[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880679:880980 [0] NCCL INFO Channel 02/0 : 24[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618587:618887 [3] NCCL INFO Channel 07/0 : 27[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618586:618886 [2] NCCL INFO Channel 09/0 : 42[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926600:926893 [6] NCCL INFO Channel 08/0 : 62[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880681:880983 [2] NCCL INFO Channel 00/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880683:880982 [4] NCCL INFO Channel 12/0 : 28[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880685:880977 [6] NCCL INFO Channel 09/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880679:880980 [0] NCCL INFO Channel 04/0 : 24[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618591:618885 [7] NCCL INFO Channel 13/0 : 47[7] -> 31[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618588:618888 [4] NCCL INFO Channel 15/0 : 44[4] -> 28[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618584:618882 [0] NCCL INFO Channel 01/0 : 24[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618585:618884 [1] NCCL INFO Channel 07/0 : 25[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618590:618889 [6] NCCL INFO Channel 09/0 : 30[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926596:926899 [2] NCCL INFO Channel 12/0 : 58[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926598:926894 [4] NCCL INFO Channel 12/0 : 60[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618589:618883 [5] NCCL INFO Channel 07/0 : 45[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926601:926896 [7] NCCL INFO Channel 00/0 : 63[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618587:618887 [3] NCCL INFO Channel 09/0 : 27[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618586:618886 [2] NCCL INFO Channel 11/0 : 42[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880681:880983 [2] NCCL INFO Channel 01/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880683:880982 [4] NCCL INFO Channel 14/0 : 28[4] -> 20[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880685:880977 [6] NCCL INFO Channel 10/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880679:880980 [0] NCCL INFO Channel 06/0 : 24[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926600:926893 [6] NCCL INFO Channel 10/0 : 62[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618591:618885 [7] NCCL INFO Channel 01/0 : 31[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880681:880983 [2] NCCL INFO Channel 03/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926598:926894 [4] NCCL INFO Channel 14/0 : 60[4] -> 52[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926596:926899 [2] NCCL INFO Channel 14/0 : 58[2] -> 50[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880683:880982 [4] NCCL INFO Channel 00/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926601:926896 [7] NCCL INFO Channel 02/0 : 63[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880685:880977 [6] NCCL INFO Channel 11/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926600:926893 [6] NCCL INFO Channel 12/0 : 62[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880679:880980 [0] NCCL INFO Channel 08/0 : 24[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880686:880979 [7] NCCL INFO Channel 00/0 : 31[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618585:618884 [1] NCCL INFO Channel 11/0 : 25[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618588:618888 [4] NCCL INFO Channel 01/0 : 28[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618590:618889 [6] NCCL INFO Channel 11/0 : 30[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618584:618882 [0] NCCL INFO Channel 03/0 : 24[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618589:618883 [5] NCCL INFO Channel 09/0 : 45[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618587:618887 [3] NCCL INFO Channel 13/0 : 27[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618586:618886 [2] NCCL INFO Channel 13/0 : 42[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926596:926899 [2] NCCL INFO Channel 00/0 : 50[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926598:926894 [4] NCCL INFO Channel 00/0 : 52[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880681:880983 [2] NCCL INFO Channel 04/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880683:880982 [4] NCCL INFO Channel 01/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880685:880977 [6] NCCL INFO Channel 12/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880679:880980 [0] NCCL INFO Channel 10/0 : 24[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618591:618885 [7] NCCL INFO Channel 03/0 : 31[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926600:926893 [6] NCCL INFO Channel 14/0 : 62[6] -> 54[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926601:926896 [7] NCCL INFO Channel 04/0 : 63[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880681:880983 [2] NCCL INFO Channel 05/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880686:880979 [7] NCCL INFO Channel 02/0 : 31[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618585:618884 [1] NCCL INFO Channel 13/0 : 25[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880683:880982 [4] NCCL INFO Channel 02/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618584:618882 [0] NCCL INFO Channel 05/0 : 24[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618590:618889 [6] NCCL INFO Channel 13/0 : 30[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618588:618888 [4] NCCL INFO Channel 03/0 : 28[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880682:880978 [3] NCCL INFO Channel 00/0 : 27[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880685:880977 [6] NCCL INFO Channel 13/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926598:926894 [4] NCCL INFO Channel 01/0 : 52[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618587:618887 [3] NCCL INFO Channel 15/0 : 27[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618586:618886 [2] NCCL INFO Channel 15/0 : 42[2] -> 26[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926600:926893 [6] NCCL INFO Channel 00/0 : 54[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618589:618883 [5] NCCL INFO Channel 11/0 : 45[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618591:618885 [7] NCCL INFO Channel 05/0 : 31[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880679:880980 [0] NCCL INFO Channel 12/0 : 24[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880681:880983 [2] NCCL INFO Channel 06/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880686:880979 [7] NCCL INFO Channel 04/0 : 31[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926596:926899 [2] NCCL INFO Channel 01/0 : 50[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926601:926896 [7] NCCL INFO Channel 06/0 : 63[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618585:618884 [1] NCCL INFO Channel 15/0 : 25[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880683:880982 [4] NCCL INFO Channel 03/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926598:926894 [4] NCCL INFO Channel 02/0 : 52[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618584:618882 [0] NCCL INFO Channel 07/0 : 24[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926600:926893 [6] NCCL INFO Channel 01/0 : 54[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880684:880981 [5] NCCL INFO Channel 00/0 : 29[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618590:618889 [6] NCCL INFO Channel 15/0 : 30[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618588:618888 [4] NCCL INFO Channel 05/0 : 28[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618591:618885 [7] NCCL INFO Channel 07/0 : 31[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880685:880977 [6] NCCL INFO Channel 15/0 : 22[6] -> 14[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926596:926899 [2] NCCL INFO Channel 03/0 : 50[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926601:926896 [7] NCCL INFO Channel 08/0 : 63[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618586:618886 [2] NCCL INFO Channel 01/0 : 26[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618589:618883 [5] NCCL INFO Channel 15/0 : 45[5] -> 29[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880682:880978 [3] NCCL INFO Channel 02/0 : 27[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926598:926894 [4] NCCL INFO Channel 03/0 : 52[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880679:880980 [0] NCCL INFO Channel 14/0 : 24[0] -> 16[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926600:926893 [6] NCCL INFO Channel 02/0 : 54[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880681:880983 [2] NCCL INFO Channel 07/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880683:880982 [4] NCCL INFO Channel 05/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880686:880979 [7] NCCL INFO Channel 06/0 : 31[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618588:618888 [4] NCCL INFO Channel 07/0 : 28[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618584:618882 [0] NCCL INFO Channel 09/0 : 24[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926598:926894 [4] NCCL INFO Channel 05/0 : 52[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926596:926899 [2] NCCL INFO Channel 04/0 : 50[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926601:926896 [7] NCCL INFO Channel 10/0 : 63[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880684:880981 [5] NCCL INFO Channel 02/0 : 29[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926600:926893 [6] NCCL INFO Channel 03/0 : 54[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880682:880978 [3] NCCL INFO Channel 04/0 : 27[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942612:942914 [1] NCCL INFO Channel 00/0 : 49[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880679:880980 [0] NCCL INFO Channel 01/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880681:880983 [2] NCCL INFO Channel 08/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880683:880982 [4] NCCL INFO Channel 06/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880686:880979 [7] NCCL INFO Channel 08/0 : 31[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618586:618886 [2] NCCL INFO Channel 03/0 : 26[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618591:618885 [7] NCCL INFO Channel 09/0 : 31[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618589:618883 [5] NCCL INFO Channel 01/0 : 29[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926594:926898 [0] NCCL INFO Channel 02/0 : 56[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926595:926897 [1] NCCL INFO Channel 00/0 : 57[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880684:880981 [5] NCCL INFO Channel 04/0 : 29[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618588:618888 [4] NCCL INFO Channel 09/0 : 28[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019291:1019594 [3] NCCL INFO Channel 00/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880682:880978 [3] NCCL INFO Channel 06/0 : 27[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880679:880980 [0] NCCL INFO Channel 02/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942612:942914 [1] NCCL INFO Channel 02/0 : 49[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880680:880984 [1] NCCL INFO Channel 00/0 : 25[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926596:926899 [2] NCCL INFO Channel 05/0 : 50[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926598:926894 [4] NCCL INFO Channel 06/0 : 52[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926601:926896 [7] NCCL INFO Channel 12/0 : 63[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926600:926893 [6] NCCL INFO Channel 04/0 : 54[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618591:618885 [7] NCCL INFO Channel 11/0 : 31[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019289:1019593 [1] NCCL INFO Channel 00/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926594:926898 [0] NCCL INFO Channel 04/0 : 56[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926595:926897 [1] NCCL INFO Channel 02/0 : 57[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618589:618883 [5] NCCL INFO Channel 03/0 : 29[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942617:942916 [6] NCCL INFO Channel 00/0 : 54[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880681:880983 [2] NCCL INFO Channel 09/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618587:618887 [3] NCCL INFO Channel 00/0 : 27[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618584:618882 [0] NCCL INFO Channel 11/0 : 24[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942614:942915 [3] NCCL INFO Channel 00/0 : 51[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880683:880982 [4] NCCL INFO Channel 07/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618586:618886 [2] NCCL INFO Channel 05/0 : 26[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880686:880979 [7] NCCL INFO Channel 10/0 : 31[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618585:618884 [1] NCCL INFO Channel 00/0 : 25[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880679:880980 [0] NCCL INFO Channel 03/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880684:880981 [5] NCCL INFO Channel 06/0 : 29[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880682:880978 [3] NCCL INFO Channel 08/0 : 27[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019291:1019594 [3] NCCL INFO Channel 01/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880680:880984 [1] NCCL INFO Channel 02/0 : 25[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019289:1019593 [1] NCCL INFO Channel 02/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926596:926899 [2] NCCL INFO Channel 06/0 : 50[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942612:942914 [1] NCCL INFO Channel 03/0 : 49[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926598:926894 [4] NCCL INFO Channel 07/0 : 52[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942617:942916 [6] NCCL INFO Channel 01/0 : 54[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926601:926896 [7] NCCL INFO Channel 14/0 : 63[7] -> 55[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926600:926893 [6] NCCL INFO Channel 05/0 : 54[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618588:618888 [4] NCCL INFO Channel 11/0 : 28[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618591:618885 [7] NCCL INFO Channel 13/0 : 31[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880681:880983 [2] NCCL INFO Channel 11/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019291:1019594 [3] NCCL INFO Channel 02/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880683:880982 [4] NCCL INFO Channel 08/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019294:1019589 [6] NCCL INFO Channel 00/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880686:880979 [7] NCCL INFO Channel 12/0 : 31[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019289:1019593 [1] NCCL INFO Channel 03/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926595:926897 [1] NCCL INFO Channel 04/0 : 57[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926594:926898 [0] NCCL INFO Channel 06/0 : 56[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926596:926899 [2] NCCL INFO Channel 07/0 : 50[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618589:618883 [5] NCCL INFO Channel 05/0 : 29[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926598:926894 [4] NCCL INFO Channel 08/0 : 52[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618584:618882 [0] NCCL INFO Channel 13/0 : 24[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942614:942915 [3] NCCL INFO Channel 01/0 : 51[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618586:618886 [2] NCCL INFO Channel 07/0 : 26[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019291:1019594 [3] NCCL INFO Channel 04/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880679:880980 [0] NCCL INFO Channel 04/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880682:880978 [3] NCCL INFO Channel 10/0 : 27[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618587:618887 [3] NCCL INFO Channel 02/0 : 27[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019294:1019589 [6] NCCL INFO Channel 01/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880684:880981 [5] NCCL INFO Channel 08/0 : 29[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618585:618884 [1] NCCL INFO Channel 02/0 : 25[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618590:618889 [6] NCCL INFO Channel 00/0 : 30[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618588:618888 [4] NCCL INFO Channel 13/0 : 28[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019289:1019593 [1] NCCL INFO Channel 04/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880680:880984 [1] NCCL INFO Channel 04/0 : 25[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880681:880983 [2] NCCL INFO Channel 12/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926600:926893 [6] NCCL INFO Channel 07/0 : 54[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880683:880982 [4] NCCL INFO Channel 09/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880686:880979 [7] NCCL INFO Channel 14/0 : 31[7] -> 23[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926594:926898 [0] NCCL INFO Channel 08/0 : 56[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926601:926896 [7] NCCL INFO Channel 00/0 : 55[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942612:942914 [1] NCCL INFO Channel 04/0 : 49[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942617:942916 [6] NCCL INFO Channel 02/0 : 54[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926595:926897 [1] NCCL INFO Channel 06/0 : 57[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926596:926899 [2] NCCL INFO Channel 08/0 : 50[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942614:942915 [3] NCCL INFO Channel 02/0 : 51[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926598:926894 [4] NCCL INFO Channel 09/0 : 52[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942617:942916 [6] NCCL INFO Channel 03/0 : 54[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618589:618883 [5] NCCL INFO Channel 07/0 : 29[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019291:1019594 [3] NCCL INFO Channel 05/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618584:618882 [0] NCCL INFO Channel 15/0 : 24[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019294:1019589 [6] NCCL INFO Channel 02/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618586:618886 [2] NCCL INFO Channel 09/0 : 26[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019289:1019593 [1] NCCL INFO Channel 05/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880679:880980 [0] NCCL INFO Channel 05/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618587:618887 [3] NCCL INFO Channel 04/0 : 27[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942612:942914 [1] NCCL INFO Channel 05/0 : 49[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880684:880981 [5] NCCL INFO Channel 10/0 : 29[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926597:926900 [3] NCCL INFO Channel 00/0 : 59[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942614:942915 [3] NCCL INFO Channel 04/0 : 51[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880682:880978 [3] NCCL INFO Channel 12/0 : 27[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942618:942918 [7] NCCL INFO Channel 00/0 : 55[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880681:880983 [2] NCCL INFO Channel 13/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880680:880984 [1] NCCL INFO Channel 06/0 : 25[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019291:1019594 [3] NCCL INFO Channel 06/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942617:942916 [6] NCCL INFO Channel 04/0 : 54[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019294:1019589 [6] NCCL INFO Channel 03/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618585:618884 [1] NCCL INFO Channel 04/0 : 25[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618588:618888 [4] NCCL INFO Channel 15/0 : 28[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618590:618889 [6] NCCL INFO Channel 02/0 : 30[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926600:926893 [6] NCCL INFO Channel 08/0 : 54[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926595:926897 [1] NCCL INFO Channel 08/0 : 57[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618589:618883 [5] NCCL INFO Channel 09/0 : 29[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880686:880979 [7] NCCL INFO Channel 00/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880683:880982 [4] NCCL INFO Channel 10/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926601:926896 [7] NCCL INFO Channel 01/0 : 55[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926596:926899 [2] NCCL INFO Channel 09/0 : 50[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618587:618887 [3] NCCL INFO Channel 06/0 : 27[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942612:942914 [1] NCCL INFO Channel 06/0 : 49[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880679:880980 [0] NCCL INFO Channel 06/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926594:926898 [0] NCCL INFO Channel 10/0 : 56[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618586:618886 [2] NCCL INFO Channel 11/0 : 26[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019289:1019593 [1] NCCL INFO Channel 06/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942614:942915 [3] NCCL INFO Channel 05/0 : 51[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880684:880981 [5] NCCL INFO Channel 12/0 : 29[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926598:926894 [4] NCCL INFO Channel 10/0 : 52[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618585:618884 [1] NCCL INFO Channel 06/0 : 25[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019291:1019594 [3] NCCL INFO Channel 07/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942617:942916 [6] NCCL INFO Channel 05/0 : 54[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942618:942918 [7] NCCL INFO Channel 01/0 : 55[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880682:880978 [3] NCCL INFO Channel 14/0 : 27[3] -> 19[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926599:926895 [5] NCCL INFO Channel 00/0 : 61[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926597:926900 [3] NCCL INFO Channel 02/0 : 59[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019295:1019591 [7] NCCL INFO Channel 00/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019294:1019589 [6] NCCL INFO Channel 04/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618590:618889 [6] NCCL INFO Channel 04/0 : 30[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618589:618883 [5] NCCL INFO Channel 11/0 : 29[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942612:942914 [1] NCCL INFO Channel 07/0 : 49[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880681:880983 [2] NCCL INFO Channel 14/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618591:618885 [7] NCCL INFO Channel 00/0 : 31[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618587:618887 [3] NCCL INFO Channel 08/0 : 27[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942614:942915 [3] NCCL INFO Channel 06/0 : 51[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880680:880984 [1] NCCL INFO Channel 08/0 : 25[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926595:926897 [1] NCCL INFO Channel 10/0 : 57[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926596:926899 [2] NCCL INFO Channel 11/0 : 50[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926600:926893 [6] NCCL INFO Channel 09/0 : 54[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618585:618884 [1] NCCL INFO Channel 08/0 : 25[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019289:1019593 [1] NCCL INFO Channel 07/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880679:880980 [0] NCCL INFO Channel 07/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880683:880982 [4] NCCL INFO Channel 11/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880686:880979 [7] NCCL INFO Channel 01/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926601:926896 [7] NCCL INFO Channel 02/0 : 55[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618586:618886 [2] NCCL INFO Channel 13/0 : 26[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019291:1019594 [3] NCCL INFO Channel 08/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880684:880981 [5] NCCL INFO Channel 14/0 : 29[5] -> 21[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880682:880978 [3] NCCL INFO Channel 00/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926594:926898 [0] NCCL INFO Channel 12/0 : 56[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618590:618889 [6] NCCL INFO Channel 08/0 : 30[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618589:618883 [5] NCCL INFO Channel 15/0 : 29[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019295:1019591 [7] NCCL INFO Channel 01/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019294:1019589 [6] NCCL INFO Channel 05/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880681:880983 [2] NCCL INFO Channel 15/0 : 18[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926598:926894 [4] NCCL INFO Channel 11/0 : 52[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019289:1019593 [1] NCCL INFO Channel 08/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926597:926900 [3] NCCL INFO Channel 04/0 : 59[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926599:926895 [5] NCCL INFO Channel 02/0 : 61[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942617:942916 [6] NCCL INFO Channel 07/0 : 54[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942611:942911 [0] NCCL INFO Channel 01/0 : 48[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942618:942918 [7] NCCL INFO Channel 02/0 : 55[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942612:942914 [1] NCCL INFO Channel 08/0 : 49[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942615:942912 [4] NCCL INFO Channel 00/0 : 52[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618587:618887 [3] NCCL INFO Channel 10/0 : 27[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618585:618884 [1] NCCL INFO Channel 10/0 : 25[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618591:618885 [7] NCCL INFO Channel 02/0 : 31[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880680:880984 [1] NCCL INFO Channel 10/0 : 25[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618586:618886 [2] NCCL INFO Channel 15/0 : 26[2] -> 10[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019291:1019594 [3] NCCL INFO Channel 09/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880679:880980 [0] NCCL INFO Channel 09/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618590:618889 [6] NCCL INFO Channel 10/0 : 30[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019294:1019589 [6] NCCL INFO Channel 07/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019295:1019591 [7] NCCL INFO Channel 02/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880683:880982 [4] NCCL INFO Channel 13/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618584:618882 [0] NCCL INFO Channel 02/0 : 24[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019289:1019593 [1] NCCL INFO Channel 10/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942614:942915 [3] NCCL INFO Channel 07/0 : 51[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880686:880979 [7] NCCL INFO Channel 02/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019291:1019594 [3] NCCL INFO Channel 10/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942617:942916 [6] NCCL INFO Channel 08/0 : 54[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880684:880981 [5] NCCL INFO Channel 00/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880682:880978 [3] NCCL INFO Channel 01/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880680:880984 [1] NCCL INFO Channel 12/0 : 25[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618587:618887 [3] NCCL INFO Channel 12/0 : 27[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618585:618884 [1] NCCL INFO Channel 12/0 : 25[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618591:618885 [7] NCCL INFO Channel 04/0 : 31[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618590:618889 [6] NCCL INFO Channel 12/0 : 30[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618584:618882 [0] NCCL INFO Channel 04/0 : 24[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019288:1019588 [0] NCCL INFO Channel 01/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942618:942918 [7] NCCL INFO Channel 03/0 : 55[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942611:942911 [0] NCCL INFO Channel 02/0 : 48[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942612:942914 [1] NCCL INFO Channel 10/0 : 49[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618588:618888 [4] NCCL INFO Channel 00/0 : 28[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942615:942912 [4] NCCL INFO Channel 01/0 : 52[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880679:880980 [0] NCCL INFO Channel 10/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618587:618887 [3] NCCL INFO Channel 14/0 : 27[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942614:942915 [3] NCCL INFO Channel 08/0 : 51[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942617:942916 [6] NCCL INFO Channel 09/0 : 54[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880683:880982 [4] NCCL INFO Channel 14/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880686:880979 [7] NCCL INFO Channel 03/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019294:1019589 [6] NCCL INFO Channel 08/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019295:1019591 [7] NCCL INFO Channel 03/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019292:1019590 [4] NCCL INFO Channel 00/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880682:880978 [3] NCCL INFO Channel 02/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019289:1019593 [1] NCCL INFO Channel 11/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019291:1019594 [3] NCCL INFO Channel 12/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019288:1019588 [0] NCCL INFO Channel 02/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880684:880981 [5] NCCL INFO Channel 01/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880680:880984 [1] NCCL INFO Channel 14/0 : 25[1] -> 17[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618585:618884 [1] NCCL INFO Channel 14/0 : 25[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880679:880980 [0] NCCL INFO Channel 11/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618591:618885 [7] NCCL INFO Channel 06/0 : 31[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618590:618889 [6] NCCL INFO Channel 14/0 : 30[6] -> 22[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942616:942913 [5] NCCL INFO Channel 00/0 : 53[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880683:880982 [4] NCCL INFO Channel 15/0 : 20[4] -> 12[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618584:618882 [0] NCCL INFO Channel 06/0 : 24[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942618:942918 [7] NCCL INFO Channel 04/0 : 55[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942611:942911 [0] NCCL INFO Channel 03/0 : 48[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618588:618888 [4] NCCL INFO Channel 02/0 : 28[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880686:880979 [7] NCCL INFO Channel 04/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019294:1019589 [6] NCCL INFO Channel 09/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019292:1019590 [4] NCCL INFO Channel 01/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942612:942914 [1] NCCL INFO Channel 11/0 : 49[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942615:942912 [4] NCCL INFO Channel 02/0 : 52[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942614:942915 [3] NCCL INFO Channel 09/0 : 51[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942617:942916 [6] NCCL INFO Channel 10/0 : 54[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880682:880978 [3] NCCL INFO Channel 04/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618591:618885 [7] NCCL INFO Channel 08/0 : 31[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942613:942917 [2] NCCL INFO Channel 00/0 : 50[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942616:942913 [5] NCCL INFO Channel 01/0 : 53[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880684:880981 [5] NCCL INFO Channel 02/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880679:880980 [0] NCCL INFO Channel 12/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618584:618882 [0] NCCL INFO Channel 08/0 : 24[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880680:880984 [1] NCCL INFO Channel 00/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618588:618888 [4] NCCL INFO Channel 06/0 : 28[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618589:618883 [5] NCCL INFO Channel 00/0 : 29[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880686:880979 [7] NCCL INFO Channel 05/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618586:618886 [2] NCCL INFO Channel 00/0 : 26[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880682:880978 [3] NCCL INFO Channel 05/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019295:1019591 [7] NCCL INFO Channel 04/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019289:1019593 [1] NCCL INFO Channel 12/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019288:1019588 [0] NCCL INFO Channel 03/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019291:1019594 [3] NCCL INFO Channel 13/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942611:942911 [0] NCCL INFO Channel 04/0 : 48[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942618:942918 [7] NCCL INFO Channel 05/0 : 55[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019293:1019592 [5] NCCL INFO Channel 00/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942612:942914 [1] NCCL INFO Channel 12/0 : 49[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019290:1019595 [2] NCCL INFO Channel 00/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880684:880981 [5] NCCL INFO Channel 03/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880679:880980 [0] NCCL INFO Channel 13/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880680:880984 [1] NCCL INFO Channel 02/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942615:942912 [4] NCCL INFO Channel 03/0 : 52[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880686:880979 [7] NCCL INFO Channel 06/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019294:1019589 [6] NCCL INFO Channel 10/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942614:942915 [3] NCCL INFO Channel 10/0 : 51[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942617:942916 [6] NCCL INFO Channel 11/0 : 54[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880682:880978 [3] NCCL INFO Channel 06/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019292:1019590 [4] NCCL INFO Channel 02/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942613:942917 [2] NCCL INFO Channel 01/0 : 50[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942611:942911 [0] NCCL INFO Channel 05/0 : 48[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880684:880981 [5] NCCL INFO Channel 04/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942616:942913 [5] NCCL INFO Channel 02/0 : 53[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942618:942918 [7] NCCL INFO Channel 06/0 : 55[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880679:880980 [0] NCCL INFO Channel 14/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880680:880984 [1] NCCL INFO Channel 03/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019295:1019591 [7] NCCL INFO Channel 05/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942612:942914 [1] NCCL INFO Channel 13/0 : 49[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880682:880978 [3] NCCL INFO Channel 07/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880686:880979 [7] NCCL INFO Channel 08/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019291:1019594 [3] NCCL INFO Channel 14/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019288:1019588 [0] NCCL INFO Channel 04/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019289:1019593 [1] NCCL INFO Channel 13/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019293:1019592 [5] NCCL INFO Channel 01/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880684:880981 [5] NCCL INFO Channel 06/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019290:1019595 [2] NCCL INFO Channel 01/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019294:1019589 [6] NCCL INFO Channel 11/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942615:942912 [4] NCCL INFO Channel 05/0 : 52[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942614:942915 [3] NCCL INFO Channel 12/0 : 51[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942617:942916 [6] NCCL INFO Channel 12/0 : 54[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880679:880980 [0] NCCL INFO Channel 15/0 : 16[0] -> 8[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019292:1019590 [4] NCCL INFO Channel 03/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880680:880984 [1] NCCL INFO Channel 04/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019295:1019591 [7] NCCL INFO Channel 06/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880682:880978 [3] NCCL INFO Channel 08/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880686:880979 [7] NCCL INFO Channel 09/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880684:880981 [5] NCCL INFO Channel 07/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880680:880984 [1] NCCL INFO Channel 05/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942611:942911 [0] NCCL INFO Channel 06/0 : 48[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942613:942917 [2] NCCL INFO Channel 03/0 : 50[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942612:942914 [1] NCCL INFO Channel 14/0 : 49[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942618:942918 [7] NCCL INFO Channel 08/0 : 55[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942616:942913 [5] NCCL INFO Channel 03/0 : 53[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942615:942912 [4] NCCL INFO Channel 06/0 : 52[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019288:1019588 [0] NCCL INFO Channel 05/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942614:942915 [3] NCCL INFO Channel 13/0 : 51[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942617:942916 [6] NCCL INFO Channel 13/0 : 54[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019293:1019592 [5] NCCL INFO Channel 02/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880682:880978 [3] NCCL INFO Channel 09/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019291:1019594 [3] NCCL INFO Channel 15/0 : 19[3] -> 11[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019289:1019593 [1] NCCL INFO Channel 14/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880686:880979 [7] NCCL INFO Channel 10/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019294:1019589 [6] NCCL INFO Channel 12/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019292:1019590 [4] NCCL INFO Channel 05/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019290:1019595 [2] NCCL INFO Channel 03/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019295:1019591 [7] NCCL INFO Channel 08/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880684:880981 [5] NCCL INFO Channel 08/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880680:880984 [1] NCCL INFO Channel 06/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880682:880978 [3] NCCL INFO Channel 10/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942611:942911 [0] NCCL INFO Channel 07/0 : 48[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942613:942917 [2] NCCL INFO Channel 04/0 : 50[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942612:942914 [1] NCCL INFO Channel 15/0 : 49[1] -> 41[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019288:1019588 [0] NCCL INFO Channel 06/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019293:1019592 [5] NCCL INFO Channel 03/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880686:880979 [7] NCCL INFO Channel 11/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880684:880981 [5] NCCL INFO Channel 09/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880680:880984 [1] NCCL INFO Channel 07/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880682:880978 [3] NCCL INFO Channel 12/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942618:942918 [7] NCCL INFO Channel 09/0 : 55[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019289:1019593 [1] NCCL INFO Channel 15/0 : 17[1] -> 9[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019291:1019594 [3] NCCL INFO Channel 01/0 : 11[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942615:942912 [4] NCCL INFO Channel 07/0 : 52[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942617:942916 [6] NCCL INFO Channel 15/0 : 54[6] -> 46[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942616:942913 [5] NCCL INFO Channel 04/0 : 53[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019292:1019590 [4] NCCL INFO Channel 06/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019294:1019589 [6] NCCL INFO Channel 13/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019295:1019591 [7] NCCL INFO Channel 09/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880686:880979 [7] NCCL INFO Channel 12/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880680:880984 [1] NCCL INFO Channel 08/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880684:880981 [5] NCCL INFO Channel 10/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880682:880978 [3] NCCL INFO Channel 13/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942614:942915 [3] NCCL INFO Channel 14/0 : 51[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942611:942911 [0] NCCL INFO Channel 09/0 : 48[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942613:942917 [2] NCCL INFO Channel 05/0 : 50[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942612:942914 [1] NCCL INFO Channel 01/0 : 41[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942618:942918 [7] NCCL INFO Channel 10/0 : 55[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019288:1019588 [0] NCCL INFO Channel 07/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019293:1019592 [5] NCCL INFO Channel 04/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019290:1019595 [2] NCCL INFO Channel 04/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019289:1019593 [1] NCCL INFO Channel 01/0 : 9[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019291:1019594 [3] NCCL INFO Channel 03/0 : 11[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880686:880979 [7] NCCL INFO Channel 13/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880680:880984 [1] NCCL INFO Channel 10/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942615:942912 [4] NCCL INFO Channel 08/0 : 52[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942617:942916 [6] NCCL INFO Channel 01/0 : 46[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019292:1019590 [4] NCCL INFO Channel 07/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019294:1019589 [6] NCCL INFO Channel 15/0 : 22[6] -> 14[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019295:1019591 [7] NCCL INFO Channel 10/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019293:1019592 [5] NCCL INFO Channel 06/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880682:880978 [3] NCCL INFO Channel 14/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880684:880981 [5] NCCL INFO Channel 11/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880686:880979 [7] NCCL INFO Channel 14/0 : 23[7] -> 15[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942611:942911 [0] NCCL INFO Channel 10/0 : 48[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942614:942915 [3] NCCL INFO Channel 15/0 : 51[3] -> 43[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942612:942914 [1] NCCL INFO Channel 03/0 : 41[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942613:942917 [2] NCCL INFO Channel 06/0 : 50[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942616:942913 [5] NCCL INFO Channel 06/0 : 53[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942618:942918 [7] NCCL INFO Channel 11/0 : 55[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942615:942912 [4] NCCL INFO Channel 09/0 : 52[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019290:1019595 [2] NCCL INFO Channel 05/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019288:1019588 [0] NCCL INFO Channel 09/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019291:1019594 [3] NCCL INFO Channel 05/0 : 11[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019289:1019593 [1] NCCL INFO Channel 03/0 : 9[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019292:1019590 [4] NCCL INFO Channel 08/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019294:1019589 [6] NCCL INFO Channel 01/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019295:1019591 [7] NCCL INFO Channel 11/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880680:880984 [1] NCCL INFO Channel 11/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942617:942916 [6] NCCL INFO Channel 03/0 : 46[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880682:880978 [3] NCCL INFO Channel 15/0 : 19[3] -> 11[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880684:880981 [5] NCCL INFO Channel 12/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942611:942911 [0] NCCL INFO Channel 11/0 : 48[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942614:942915 [3] NCCL INFO Channel 01/0 : 43[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019293:1019592 [5] NCCL INFO Channel 07/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019288:1019588 [0] NCCL INFO Channel 10/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019289:1019593 [1] NCCL INFO Channel 05/0 : 9[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019291:1019594 [3] NCCL INFO Channel 07/0 : 11[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019290:1019595 [2] NCCL INFO Channel 06/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942612:942914 [1] NCCL INFO Channel 05/0 : 41[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880680:880984 [1] NCCL INFO Channel 12/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942618:942918 [7] NCCL INFO Channel 12/0 : 55[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942613:942917 [2] NCCL INFO Channel 07/0 : 50[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942615:942912 [4] NCCL INFO Channel 10/0 : 52[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942617:942916 [6] NCCL INFO Channel 05/0 : 46[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942616:942913 [5] NCCL INFO Channel 07/0 : 53[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880684:880981 [5] NCCL INFO Channel 14/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880680:880984 [1] NCCL INFO Channel 13/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019294:1019589 [6] NCCL INFO Channel 03/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019292:1019590 [4] NCCL INFO Channel 09/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019295:1019591 [7] NCCL INFO Channel 12/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019293:1019592 [5] NCCL INFO Channel 08/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942611:942911 [0] NCCL INFO Channel 12/0 : 48[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942614:942915 [3] NCCL INFO Channel 03/0 : 43[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880684:880981 [5] NCCL INFO Channel 15/0 : 21[5] -> 13[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880680:880984 [1] NCCL INFO Channel 14/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019288:1019588 [0] NCCL INFO Channel 11/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019289:1019593 [1] NCCL INFO Channel 07/0 : 9[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942612:942914 [1] NCCL INFO Channel 07/0 : 41[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019291:1019594 [3] NCCL INFO Channel 09/0 : 11[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880680:880984 [1] NCCL INFO Channel 15/0 : 17[1] -> 9[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942615:942912 [4] NCCL INFO Channel 11/0 : 52[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942617:942916 [6] NCCL INFO Channel 07/0 : 46[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942613:942917 [2] NCCL INFO Channel 08/0 : 50[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942618:942918 [7] NCCL INFO Channel 13/0 : 55[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019290:1019595 [2] NCCL INFO Channel 07/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019294:1019589 [6] NCCL INFO Channel 05/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942616:942913 [5] NCCL INFO Channel 08/0 : 53[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942614:942915 [3] NCCL INFO Channel 05/0 : 43[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019292:1019590 [4] NCCL INFO Channel 10/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942612:942914 [1] NCCL INFO Channel 11/0 : 41[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019295:1019591 [7] NCCL INFO Channel 13/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019293:1019592 [5] NCCL INFO Channel 09/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019289:1019593 [1] NCCL INFO Channel 11/0 : 9[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019291:1019594 [3] NCCL INFO Channel 13/0 : 11[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019288:1019588 [0] NCCL INFO Channel 12/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942615:942912 [4] NCCL INFO Channel 13/0 : 52[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926596:926899 [2] NCCL INFO Channel 12/0 : 50[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019294:1019589 [6] NCCL INFO Channel 07/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926595:926897 [1] NCCL INFO Channel 12/0 : 57[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019292:1019590 [4] NCCL INFO Channel 11/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926600:926893 [6] NCCL INFO Channel 10/0 : 54[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019290:1019595 [2] NCCL INFO Channel 08/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926601:926896 [7] NCCL INFO Channel 03/0 : 55[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926594:926898 [0] NCCL INFO Channel 14/0 : 56[0] -> 48[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926598:926894 [4] NCCL INFO Channel 13/0 : 52[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942613:942917 [2] NCCL INFO Channel 09/0 : 50[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942617:942916 [6] NCCL INFO Channel 09/0 : 46[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926597:926900 [3] NCCL INFO Channel 06/0 : 59[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942614:942915 [3] NCCL INFO Channel 07/0 : 43[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926599:926895 [5] NCCL INFO Channel 04/0 : 61[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926596:926899 [2] NCCL INFO Channel 13/0 : 50[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942612:942914 [1] NCCL INFO Channel 13/0 : 41[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942616:942913 [5] NCCL INFO Channel 09/0 : 53[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942618:942918 [7] NCCL INFO Channel 14/0 : 55[7] -> 47[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926595:926897 [1] NCCL INFO Channel 14/0 : 57[1] -> 49[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942611:942911 [0] NCCL INFO Channel 13/0 : 48[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942615:942912 [4] NCCL INFO Channel 14/0 : 52[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926600:926893 [6] NCCL INFO Channel 11/0 : 54[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019295:1019591 [7] NCCL INFO Channel 14/0 : 23[7] -> 15[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019293:1019592 [5] NCCL INFO Channel 10/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926601:926896 [7] NCCL INFO Channel 04/0 : 55[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019289:1019593 [1] NCCL INFO Channel 13/0 : 9[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019291:1019594 [3] NCCL INFO Channel 15/0 : 11[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019288:1019588 [0] NCCL INFO Channel 13/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019294:1019589 [6] NCCL INFO Channel 09/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926594:926898 [0] NCCL INFO Channel 01/0 : 48[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019292:1019590 [4] NCCL INFO Channel 13/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926598:926894 [4] NCCL INFO Channel 14/0 : 52[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926597:926900 [3] NCCL INFO Channel 08/0 : 59[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926596:926899 [2] NCCL INFO Channel 14/0 : 50[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926595:926897 [1] NCCL INFO Channel 00/0 : 49[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942613:942917 [2] NCCL INFO Channel 11/0 : 50[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926599:926895 [5] NCCL INFO Channel 06/0 : 61[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942617:942916 [6] NCCL INFO Channel 11/0 : 46[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942614:942915 [3] NCCL INFO Channel 09/0 : 43[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926600:926893 [6] NCCL INFO Channel 12/0 : 54[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926601:926896 [7] NCCL INFO Channel 05/0 : 55[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019290:1019595 [2] NCCL INFO Channel 09/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926594:926898 [0] NCCL INFO Channel 02/0 : 48[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926598:926894 [4] NCCL INFO Channel 15/0 : 52[4] -> 44[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019295:1019591 [7] NCCL INFO Channel 01/0 : 15[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019293:1019592 [5] NCCL INFO Channel 11/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926597:926900 [3] NCCL INFO Channel 10/0 : 59[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926596:926899 [2] NCCL INFO Channel 15/0 : 50[2] -> 42[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926595:926897 [1] NCCL INFO Channel 02/0 : 49[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942612:942914 [1] NCCL INFO Channel 15/0 : 41[1] -> 33[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942618:942918 [7] NCCL INFO Channel 01/0 : 47[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926600:926893 [6] NCCL INFO Channel 13/0 : 54[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926599:926895 [5] NCCL INFO Channel 08/0 : 61[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942616:942913 [5] NCCL INFO Channel 10/0 : 53[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926601:926896 [7] NCCL INFO Channel 06/0 : 55[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942611:942911 [0] NCCL INFO Channel 14/0 : 48[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942615:942912 [4] NCCL INFO Channel 15/0 : 52[4] -> 44[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926594:926898 [0] NCCL INFO Channel 03/0 : 48[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019289:1019593 [1] NCCL INFO Channel 15/0 : 9[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019294:1019589 [6] NCCL INFO Channel 11/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019288:1019588 [0] NCCL INFO Channel 14/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942613:942917 [2] NCCL INFO Channel 12/0 : 50[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926597:926900 [3] NCCL INFO Channel 12/0 : 59[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019292:1019590 [4] NCCL INFO Channel 14/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926595:926897 [1] NCCL INFO Channel 03/0 : 49[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019290:1019595 [2] NCCL INFO Channel 11/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926600:926893 [6] NCCL INFO Channel 15/0 : 54[6] -> 46[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019295:1019591 [7] NCCL INFO Channel 03/0 : 15[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019293:1019592 [5] NCCL INFO Channel 12/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926599:926895 [5] NCCL INFO Channel 10/0 : 61[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926601:926896 [7] NCCL INFO Channel 08/0 : 55[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926594:926898 [0] NCCL INFO Channel 04/0 : 48[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942614:942915 [3] NCCL INFO Channel 13/0 : 43[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942617:942916 [6] NCCL INFO Channel 13/0 : 46[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926595:926897 [1] NCCL INFO Channel 04/0 : 49[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926597:926900 [3] NCCL INFO Channel 14/0 : 59[3] -> 51[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942618:942918 [7] NCCL INFO Channel 03/0 : 47[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926601:926896 [7] NCCL INFO Channel 09/0 : 55[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942616:942913 [5] NCCL INFO Channel 11/0 : 53[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926594:926898 [0] NCCL INFO Channel 05/0 : 48[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019294:1019589 [6] NCCL INFO Channel 13/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926595:926897 [1] NCCL INFO Channel 05/0 : 49[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926599:926895 [5] NCCL INFO Channel 12/0 : 61[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019292:1019590 [4] NCCL INFO Channel 15/0 : 20[4] -> 12[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926597:926900 [3] NCCL INFO Channel 00/0 : 51[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019288:1019588 [0] NCCL INFO Channel 15/0 : 16[0] -> 8[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926601:926896 [7] NCCL INFO Channel 10/0 : 55[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926594:926898 [0] NCCL INFO Channel 06/0 : 48[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926595:926897 [1] NCCL INFO Channel 06/0 : 49[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926597:926900 [3] NCCL INFO Channel 01/0 : 51[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926599:926895 [5] NCCL INFO Channel 14/0 : 61[5] -> 53[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926601:926896 [7] NCCL INFO Channel 11/0 : 55[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942611:942911 [0] NCCL INFO Channel 15/0 : 48[0] -> 40[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926594:926898 [0] NCCL INFO Channel 07/0 : 48[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019290:1019595 [2] NCCL INFO Channel 12/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942615:942912 [4] NCCL INFO Channel 01/0 : 44[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926595:926897 [1] NCCL INFO Channel 07/0 : 49[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019295:1019591 [7] NCCL INFO Channel 05/0 : 15[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019293:1019592 [5] NCCL INFO Channel 14/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942613:942917 [2] NCCL INFO Channel 13/0 : 50[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926599:926895 [5] NCCL INFO Channel 00/0 : 53[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926597:926900 [3] NCCL INFO Channel 02/0 : 51[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926601:926896 [7] NCCL INFO Channel 12/0 : 55[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019294:1019589 [6] NCCL INFO Channel 15/0 : 14[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942614:942915 [3] NCCL INFO Channel 15/0 : 43[3] -> 35[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942617:942916 [6] NCCL INFO Channel 15/0 : 46[6] -> 38[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942618:942918 [7] NCCL INFO Channel 05/0 : 47[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942616:942913 [5] NCCL INFO Channel 12/0 : 53[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926594:926898 [0] NCCL INFO Channel 09/0 : 48[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019292:1019590 [4] NCCL INFO Channel 01/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942611:942911 [0] NCCL INFO Channel 01/0 : 40[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926595:926897 [1] NCCL INFO Channel 08/0 : 49[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019288:1019588 [0] NCCL INFO Channel 01/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926597:926900 [3] NCCL INFO Channel 04/0 : 51[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926599:926895 [5] NCCL INFO Channel 01/0 : 53[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926594:926898 [0] NCCL INFO Channel 10/0 : 48[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926601:926896 [7] NCCL INFO Channel 13/0 : 55[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926595:926897 [1] NCCL INFO Channel 10/0 : 49[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926597:926900 [3] NCCL INFO Channel 05/0 : 51[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926594:926898 [0] NCCL INFO Channel 11/0 : 48[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926599:926895 [5] NCCL INFO Channel 02/0 : 53[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942615:942912 [4] NCCL INFO Channel 03/0 : 44[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942613:942917 [2] NCCL INFO Channel 14/0 : 50[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926595:926897 [1] NCCL INFO Channel 11/0 : 49[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926601:926896 [7] NCCL INFO Channel 14/0 : 55[7] -> 47[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019295:1019591 [7] NCCL INFO Channel 07/0 : 15[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019293:1019592 [5] NCCL INFO Channel 15/0 : 21[5] -> 13[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942618:942918 [7] NCCL INFO Channel 07/0 : 47[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926597:926900 [3] NCCL INFO Channel 06/0 : 51[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019290:1019595 [2] NCCL INFO Channel 13/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019292:1019590 [4] NCCL INFO Channel 03/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942616:942913 [5] NCCL INFO Channel 14/0 : 53[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942611:942911 [0] NCCL INFO Channel 03/0 : 40[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926594:926898 [0] NCCL INFO Channel 12/0 : 48[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019288:1019588 [0] NCCL INFO Channel 03/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926599:926895 [5] NCCL INFO Channel 03/0 : 53[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926595:926897 [1] NCCL INFO Channel 12/0 : 49[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926597:926900 [3] NCCL INFO Channel 07/0 : 51[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926594:926898 [0] NCCL INFO Channel 13/0 : 48[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926595:926897 [1] NCCL INFO Channel 13/0 : 49[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926599:926895 [5] NCCL INFO Channel 04/0 : 53[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926597:926900 [3] NCCL INFO Channel 08/0 : 51[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942615:942912 [4] NCCL INFO Channel 05/0 : 44[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926594:926898 [0] NCCL INFO Channel 14/0 : 48[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019295:1019591 [7] NCCL INFO Channel 09/0 : 15[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942613:942917 [2] NCCL INFO Channel 15/0 : 50[2] -> 42[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942618:942918 [7] NCCL INFO Channel 09/0 : 47[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926595:926897 [1] NCCL INFO Channel 14/0 : 49[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019293:1019592 [5] NCCL INFO Channel 01/0 : 13[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942611:942911 [0] NCCL INFO Channel 05/0 : 40[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942616:942913 [5] NCCL INFO Channel 15/0 : 53[5] -> 45[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926597:926900 [3] NCCL INFO Channel 09/0 : 51[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926594:926898 [0] NCCL INFO Channel 15/0 : 48[0] -> 40[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926599:926895 [5] NCCL INFO Channel 06/0 : 53[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019292:1019590 [4] NCCL INFO Channel 05/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942615:942912 [4] NCCL INFO Channel 07/0 : 44[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926595:926897 [1] NCCL INFO Channel 15/0 : 49[1] -> 41[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942618:942918 [7] NCCL INFO Channel 11/0 : 47[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926597:926900 [3] NCCL INFO Channel 10/0 : 51[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926599:926895 [5] NCCL INFO Channel 07/0 : 53[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942613:942917 [2] NCCL INFO Channel 01/0 : 42[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926597:926900 [3] NCCL INFO Channel 12/0 : 51[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942611:942911 [0] NCCL INFO Channel 07/0 : 40[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926599:926895 [5] NCCL INFO Channel 08/0 : 53[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942616:942913 [5] NCCL INFO Channel 01/0 : 45[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942615:942912 [4] NCCL INFO Channel 09/0 : 44[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926597:926900 [3] NCCL INFO Channel 13/0 : 51[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926599:926895 [5] NCCL INFO Channel 09/0 : 53[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926599:926895 [5] NCCL INFO Channel 10/0 : 53[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926597:926900 [3] NCCL INFO Channel 14/0 : 51[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019290:1019595 [2] NCCL INFO Channel 14/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926597:926900 [3] NCCL INFO Channel 15/0 : 51[3] -> 43[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926599:926895 [5] NCCL INFO Channel 11/0 : 53[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019288:1019588 [0] NCCL INFO Channel 05/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926599:926895 [5] NCCL INFO Channel 12/0 : 53[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019295:1019591 [7] NCCL INFO Channel 11/0 : 15[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019293:1019592 [5] NCCL INFO Channel 03/0 : 13[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019292:1019590 [4] NCCL INFO Channel 07/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926599:926895 [5] NCCL INFO Channel 14/0 : 53[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942618:942918 [7] NCCL INFO Channel 13/0 : 47[7] -> 39[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926599:926895 [5] NCCL INFO Channel 15/0 : 53[5] -> 45[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942613:942917 [2] NCCL INFO Channel 03/0 : 42[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942611:942911 [0] NCCL INFO Channel 09/0 : 40[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019288:1019588 [0] NCCL INFO Channel 07/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019290:1019595 [2] NCCL INFO Channel 15/0 : 18[2] -> 10[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019295:1019591 [7] NCCL INFO Channel 13/0 : 15[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019292:1019590 [4] NCCL INFO Channel 09/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942615:942912 [4] NCCL INFO Channel 11/0 : 44[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942616:942913 [5] NCCL INFO Channel 03/0 : 45[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019293:1019592 [5] NCCL INFO Channel 05/0 : 13[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942613:942917 [2] NCCL INFO Channel 05/0 : 42[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019288:1019588 [0] NCCL INFO Channel 09/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019290:1019595 [2] NCCL INFO Channel 01/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942611:942911 [0] NCCL INFO Channel 11/0 : 40[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942615:942912 [4] NCCL INFO Channel 13/0 : 44[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942616:942913 [5] NCCL INFO Channel 05/0 : 45[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942613:942917 [2] NCCL INFO Channel 07/0 : 42[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019292:1019590 [4] NCCL INFO Channel 11/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019293:1019592 [5] NCCL INFO Channel 07/0 : 13[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942611:942911 [0] NCCL INFO Channel 13/0 : 40[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019288:1019588 [0] NCCL INFO Channel 11/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019290:1019595 [2] NCCL INFO Channel 03/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942615:942912 [4] NCCL INFO Channel 15/0 : 44[4] -> 36[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019292:1019590 [4] NCCL INFO Channel 13/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019293:1019592 [5] NCCL INFO Channel 09/0 : 13[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019288:1019588 [0] NCCL INFO Channel 13/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019290:1019595 [2] NCCL INFO Channel 05/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942616:942913 [5] NCCL INFO Channel 07/0 : 45[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942613:942917 [2] NCCL INFO Channel 09/0 : 42[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942611:942911 [0] NCCL INFO Channel 15/0 : 40[0] -> 32[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019292:1019590 [4] NCCL INFO Channel 15/0 : 12[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019293:1019592 [5] NCCL INFO Channel 11/0 : 13[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019288:1019588 [0] NCCL INFO Channel 15/0 : 8[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019290:1019595 [2] NCCL INFO Channel 07/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942616:942913 [5] NCCL INFO Channel 09/0 : 45[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942613:942917 [2] NCCL INFO Channel 11/0 : 42[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019293:1019592 [5] NCCL INFO Channel 15/0 : 13[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019290:1019595 [2] NCCL INFO Channel 09/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942616:942913 [5] NCCL INFO Channel 11/0 : 45[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942613:942917 [2] NCCL INFO Channel 13/0 : 42[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019290:1019595 [2] NCCL INFO Channel 11/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942616:942913 [5] NCCL INFO Channel 15/0 : 45[5] -> 37[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942613:942917 [2] NCCL INFO Channel 15/0 : 42[2] -> 34[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019290:1019595 [2] NCCL INFO Channel 13/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019290:1019595 [2] NCCL INFO Channel 15/0 : 10[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-7:811252:811560 [4] NCCL INFO Channel 08/0 : 60[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618591:618885 [7] NCCL INFO Channel 10/0 : 31[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-7:811249:811559 [1] NCCL INFO Channel 12/0 : 57[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618584:618882 [0] NCCL INFO Channel 10/0 : 24[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-7:811248:811554 [0] NCCL INFO Channel 12/0 : 56[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-7:811255:811561 [7] NCCL INFO Channel 04/0 : 63[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618588:618888 [4] NCCL INFO Channel 08/0 : 28[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618589:618883 [5] NCCL INFO Channel 02/0 : 29[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-7:811252:811560 [4] NCCL INFO Channel 10/0 : 60[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618586:618886 [2] NCCL INFO Channel 04/0 : 26[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-7:811249:811559 [1] NCCL INFO Channel 14/0 : 57[1] -> 49[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618591:618885 [7] NCCL INFO Channel 12/0 : 31[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-7:811248:811554 [0] NCCL INFO Channel 14/0 : 56[0] -> 48[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618584:618882 [0] NCCL INFO Channel 12/0 : 24[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-7:811255:811561 [7] NCCL INFO Channel 06/0 : 63[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618588:618888 [4] NCCL INFO Channel 10/0 : 28[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618589:618883 [5] NCCL INFO Channel 04/0 : 29[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688203:688501 [3] NCCL INFO Channel 08/0 : 35[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-7:811252:811560 [4] NCCL INFO Channel 12/0 : 60[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618591:618885 [7] NCCL INFO Channel 14/0 : 31[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618586:618886 [2] NCCL INFO Channel 06/0 : 26[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688207:688504 [7] NCCL INFO Channel 12/0 : 39[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688202:688503 [2] NCCL INFO Channel 07/0 : 42[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-7:811255:811561 [7] NCCL INFO Channel 08/0 : 63[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618584:618882 [0] NCCL INFO Channel 14/0 : 24[0] -> 16[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688204:688502 [4] NCCL INFO Channel 05/0 : 44[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688206:688506 [6] NCCL INFO Channel 05/0 : 46[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-7:811252:811560 [4] NCCL INFO Channel 14/0 : 60[4] -> 52[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618588:618888 [4] NCCL INFO Channel 12/0 : 28[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618589:618883 [5] NCCL INFO Channel 06/0 : 29[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688201:688500 [1] NCCL INFO Channel 02/0 : 33[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-7:811250:811555 [2] NCCL INFO Channel 00/0 : 58[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618586:618886 [2] NCCL INFO Channel 08/0 : 26[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688205:688505 [5] NCCL INFO Channel 08/0 : 37[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-7:811255:811561 [7] NCCL INFO Channel 10/0 : 63[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618588:618888 [4] NCCL INFO Channel 14/0 : 28[4] -> 20[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618589:618883 [5] NCCL INFO Channel 08/0 : 29[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688203:688501 [3] NCCL INFO Channel 10/0 : 35[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-7:811250:811555 [2] NCCL INFO Channel 04/0 : 58[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618586:618886 [2] NCCL INFO Channel 10/0 : 26[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688202:688503 [2] NCCL INFO Channel 09/0 : 42[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688207:688504 [7] NCCL INFO Channel 14/0 : 39[7] -> 23[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-7:811255:811561 [7] NCCL INFO Channel 12/0 : 63[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618589:618883 [5] NCCL INFO Channel 10/0 : 29[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688204:688502 [4] NCCL INFO Channel 07/0 : 44[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688206:688506 [6] NCCL INFO Channel 07/0 : 46[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-7:811250:811555 [2] NCCL INFO Channel 06/0 : 58[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-7:811253:811558 [5] NCCL INFO Channel 00/0 : 61[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618586:618886 [2] NCCL INFO Channel 12/0 : 26[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688201:688500 [1] NCCL INFO Channel 04/0 : 33[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688205:688505 [5] NCCL INFO Channel 10/0 : 37[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688203:688501 [3] NCCL INFO Channel 12/0 : 35[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-7:811255:811561 [7] NCCL INFO Channel 14/0 : 63[7] -> 55[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618589:618883 [5] NCCL INFO Channel 12/0 : 29[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688202:688503 [2] NCCL INFO Channel 11/0 : 42[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-7:811250:811555 [2] NCCL INFO Channel 08/0 : 58[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618586:618886 [2] NCCL INFO Channel 14/0 : 26[2] -> 18[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688204:688502 [4] NCCL INFO Channel 09/0 : 44[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-7:811253:811558 [5] NCCL INFO Channel 02/0 : 61[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618589:618883 [5] NCCL INFO Channel 14/0 : 29[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688206:688506 [6] NCCL INFO Channel 09/0 : 46[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-7:811253:811558 [5] NCCL INFO Channel 04/0 : 61[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688200:688507 [0] NCCL INFO Channel 01/0 : 40[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688205:688505 [5] NCCL INFO Channel 12/0 : 37[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688201:688500 [1] NCCL INFO Channel 06/0 : 33[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688203:688501 [3] NCCL INFO Channel 14/0 : 35[3] -> 19[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-7:811250:811555 [2] NCCL INFO Channel 10/0 : 58[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688202:688503 [2] NCCL INFO Channel 13/0 : 42[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-7:811253:811558 [5] NCCL INFO Channel 06/0 : 61[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688204:688502 [4] NCCL INFO Channel 11/0 : 44[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-7:811250:811555 [2] NCCL INFO Channel 12/0 : 58[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688206:688506 [6] NCCL INFO Channel 11/0 : 46[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-7:811253:811558 [5] NCCL INFO Channel 08/0 : 61[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688205:688505 [5] NCCL INFO Channel 14/0 : 37[5] -> 21[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688201:688500 [1] NCCL INFO Channel 08/0 : 33[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-7:811250:811555 [2] NCCL INFO Channel 14/0 : 58[2] -> 50[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688200:688507 [0] NCCL INFO Channel 03/0 : 40[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688202:688503 [2] NCCL INFO Channel 15/0 : 42[2] -> 34[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-7:811253:811558 [5] NCCL INFO Channel 10/0 : 61[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688204:688502 [4] NCCL INFO Channel 13/0 : 44[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-7:811253:811558 [5] NCCL INFO Channel 12/0 : 61[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688206:688506 [6] NCCL INFO Channel 13/0 : 46[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-7:811253:811558 [5] NCCL INFO Channel 14/0 : 61[5] -> 53[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688201:688500 [1] NCCL INFO Channel 10/0 : 33[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688207:688504 [7] NCCL INFO Channel 01/0 : 47[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688204:688502 [4] NCCL INFO Channel 15/0 : 44[4] -> 36[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688206:688506 [6] NCCL INFO Channel 15/0 : 46[6] -> 38[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688200:688507 [0] NCCL INFO Channel 05/0 : 40[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688201:688500 [1] NCCL INFO Channel 12/0 : 33[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688207:688504 [7] NCCL INFO Channel 03/0 : 47[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688200:688507 [0] NCCL INFO Channel 07/0 : 40[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688203:688501 [3] NCCL INFO Channel 01/0 : 43[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688201:688500 [1] NCCL INFO Channel 14/0 : 33[1] -> 17[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688207:688504 [7] NCCL INFO Channel 05/0 : 47[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688200:688507 [0] NCCL INFO Channel 09/0 : 40[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688203:688501 [3] NCCL INFO Channel 03/0 : 43[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688205:688505 [5] NCCL INFO Channel 01/0 : 45[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688207:688504 [7] NCCL INFO Channel 07/0 : 47[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688200:688507 [0] NCCL INFO Channel 11/0 : 40[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688203:688501 [3] NCCL INFO Channel 05/0 : 43[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688207:688504 [7] NCCL INFO Channel 09/0 : 47[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688205:688505 [5] NCCL INFO Channel 03/0 : 45[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688200:688507 [0] NCCL INFO Channel 13/0 : 40[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688203:688501 [3] NCCL INFO Channel 07/0 : 43[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688207:688504 [7] NCCL INFO Channel 11/0 : 47[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688200:688507 [0] NCCL INFO Channel 15/0 : 40[0] -> 32[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688205:688505 [5] NCCL INFO Channel 05/0 : 45[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688203:688501 [3] NCCL INFO Channel 09/0 : 43[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688207:688504 [7] NCCL INFO Channel 13/0 : 47[7] -> 39[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688201:688500 [1] NCCL INFO Channel 01/0 : 41[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688205:688505 [5] NCCL INFO Channel 07/0 : 45[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688203:688501 [3] NCCL INFO Channel 13/0 : 43[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688201:688500 [1] NCCL INFO Channel 03/0 : 41[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688205:688505 [5] NCCL INFO Channel 09/0 : 45[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688203:688501 [3] NCCL INFO Channel 15/0 : 43[3] -> 35[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688205:688505 [5] NCCL INFO Channel 11/0 : 45[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688201:688500 [1] NCCL INFO Channel 05/0 : 41[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688201:688500 [1] NCCL INFO Channel 07/0 : 41[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688205:688505 [5] NCCL INFO Channel 15/0 : 45[5] -> 37[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688201:688500 [1] NCCL INFO Channel 11/0 : 41[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688201:688500 [1] NCCL INFO Channel 13/0 : 41[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688201:688500 [1] NCCL INFO Channel 15/0 : 41[1] -> 33[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:94346:94644 [4] NCCL INFO Connected NVLS tree
+ip-26-0-160-225:94346:94644 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-225:94346:94644 [4] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-160-225:94346:94644 [4] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-160-225:94343:94647 [1] NCCL INFO Connected NVLS tree
+ip-26-0-160-225:94343:94647 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-225:94343:94647 [1] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-160-225:94343:94647 [1] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-160-225:94345:94649 [3] NCCL INFO Connected NVLS tree
+ip-26-0-160-225:94345:94649 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-225:94345:94649 [3] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-160-225:94345:94649 [3] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-160-225:94349:94645 [7] NCCL INFO Connected NVLS tree
+ip-26-0-160-225:94349:94645 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-225:94349:94645 [7] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-160-225:94349:94645 [7] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-160-225:94344:94648 [2] NCCL INFO Connected NVLS tree
+ip-26-0-160-225:94344:94648 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-225:94344:94648 [2] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-160-225:94344:94648 [2] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-160-225:94347:94646 [5] NCCL INFO Connected NVLS tree
+ip-26-0-160-225:94347:94646 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-225:94347:94646 [5] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-160-225:94347:94646 [5] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-173-7:811250:811555 [2] NCCL INFO Connected NVLS tree
+ip-26-0-173-7:811250:811555 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-173-7:811250:811555 [2] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-173-7:811250:811555 [2] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-173-7:811252:811560 [4] NCCL INFO Connected NVLS tree
+ip-26-0-173-7:811252:811560 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-173-7:811252:811560 [4] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-173-7:811252:811560 [4] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-173-7:811248:811554 [0] NCCL INFO Connected NVLS tree
+ip-26-0-173-7:811248:811554 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-173-7:811248:811554 [0] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-173-7:811248:811554 [0] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-173-7:811254:811557 [6] NCCL INFO Connected NVLS tree
+ip-26-0-173-7:811254:811557 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-173-7:811254:811557 [6] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-173-7:811254:811557 [6] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-173-7:811249:811559 [1] NCCL INFO Connected NVLS tree
+ip-26-0-173-7:811249:811559 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-173-7:811249:811559 [1] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-173-7:811249:811559 [1] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-173-7:811251:811556 [3] NCCL INFO Connected NVLS tree
+ip-26-0-173-7:811251:811556 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-173-7:811251:811556 [3] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-173-7:811251:811556 [3] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-173-7:811255:811561 [7] NCCL INFO Connected NVLS tree
+ip-26-0-173-7:811255:811561 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-173-7:811255:811561 [7] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-173-7:811255:811561 [7] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-172-252:688206:688506 [6] NCCL INFO Connected NVLS tree
+ip-26-0-172-252:688206:688506 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-252:688206:688506 [6] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-172-252:688206:688506 [6] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-173-7:811253:811558 [5] NCCL INFO Connected NVLS tree
+ip-26-0-173-7:811253:811558 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-173-7:811253:811558 [5] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-173-7:811253:811558 [5] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-172-57:942618:942918 [7] NCCL INFO Connected NVLS tree
+ip-26-0-172-57:942618:942918 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-57:942618:942918 [7] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-172-57:942618:942918 [7] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-172-142:880686:880979 [7] NCCL INFO Connected NVLS tree
+ip-26-0-172-142:880686:880979 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-142:880686:880979 [7] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-172-142:880686:880979 [7] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-172-142:880684:880981 [5] NCCL INFO Connected NVLS tree
+ip-26-0-172-142:880684:880981 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-142:880684:880981 [5] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-172-142:880684:880981 [5] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-172-73:926595:926897 [1] NCCL INFO Connected NVLS tree
+ip-26-0-172-73:926595:926897 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-73:926595:926897 [1] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-172-73:926595:926897 [1] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-172-73:926601:926896 [7] NCCL INFO Connected NVLS tree
+ip-26-0-172-73:926601:926896 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-73:926601:926896 [7] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-172-73:926601:926896 [7] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-173-7:811253:811558 [5] NCCL INFO comm 0x9a582e0 rank 61 nranks 64 cudaDev 5 nvmlDev 5 busId a8000 commId 0x6ca6bcab597fda79 - Init COMPLETE
+ip-26-0-173-7:811249:811559 [1] NCCL INFO comm 0x8be9fa0 rank 57 nranks 64 cudaDev 1 nvmlDev 1 busId 64000 commId 0x6ca6bcab597fda79 - Init COMPLETE
+ip-26-0-173-7:811252:811560 [4] NCCL INFO comm 0x9103640 rank 60 nranks 64 cudaDev 4 nvmlDev 4 busId 97000 commId 0x6ca6bcab597fda79 - Init COMPLETE
+ip-26-0-173-7:811255:811561 [7] NCCL INFO comm 0xa0349b0 rank 63 nranks 64 cudaDev 7 nvmlDev 7 busId ca000 commId 0x6ca6bcab597fda79 - Init COMPLETE
+ip-26-0-173-7:811251:811556 [3] NCCL INFO comm 0x8fe5b80 rank 59 nranks 64 cudaDev 3 nvmlDev 3 busId 86000 commId 0x6ca6bcab597fda79 - Init COMPLETE
+ip-26-0-173-7:811248:811554 [0] NCCL INFO comm 0xa5d9ba0 rank 56 nranks 64 cudaDev 0 nvmlDev 0 busId 53000 commId 0x6ca6bcab597fda79 - Init COMPLETE
+ip-26-0-173-7:811254:811557 [6] NCCL INFO comm 0x8df0080 rank 62 nranks 64 cudaDev 6 nvmlDev 6 busId b9000 commId 0x6ca6bcab597fda79 - Init COMPLETE
+ip-26-0-173-7:811250:811555 [2] NCCL INFO comm 0x99e6c20 rank 58 nranks 64 cudaDev 2 nvmlDev 2 busId 75000 commId 0x6ca6bcab597fda79 - Init COMPLETE
+ip-26-0-172-57:942612:942914 [1] NCCL INFO Connected NVLS tree
+ip-26-0-172-57:942612:942914 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-57:942612:942914 [1] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-172-57:942612:942914 [1] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-172-73:926598:926894 [4] NCCL INFO Connected NVLS tree
+ip-26-0-172-73:926598:926894 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-73:926598:926894 [4] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-172-73:926598:926894 [4] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-172-142:880680:880984 [1] NCCL INFO Connected NVLS tree
+ip-26-0-172-142:880680:880984 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-142:880680:880984 [1] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-172-142:880680:880984 [1] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-172-73:926596:926899 [2] NCCL INFO Connected NVLS tree
+ip-26-0-172-73:926596:926899 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-73:926596:926899 [2] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-172-73:926596:926899 [2] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-172-252:688202:688503 [2] NCCL INFO Connected NVLS tree
+ip-26-0-172-252:688202:688503 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-252:688202:688503 [2] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-172-252:688202:688503 [2] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-172-73:926597:926900 [3] NCCL INFO Connected NVLS tree
+ip-26-0-172-73:926597:926900 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-73:926597:926900 [3] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-172-73:926597:926900 [3] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-172-73:926600:926893 [6] NCCL INFO Connected NVLS tree
+ip-26-0-172-73:926600:926893 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-73:926600:926893 [6] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-172-73:926600:926893 [6] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-172-73:926599:926895 [5] NCCL INFO Connected NVLS tree
+ip-26-0-172-73:926599:926895 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-73:926599:926895 [5] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-172-73:926599:926895 [5] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-172-57:942616:942913 [5] NCCL INFO Connected NVLS tree
+ip-26-0-172-57:942616:942913 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-57:942616:942913 [5] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-172-57:942616:942913 [5] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-172-116:1019293:1019592 [5] NCCL INFO Connected NVLS tree
+ip-26-0-172-116:1019293:1019592 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-116:1019293:1019592 [5] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-172-116:1019293:1019592 [5] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-172-73:926594:926898 [0] NCCL INFO Connected NVLS tree
+ip-26-0-172-73:926594:926898 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-73:926594:926898 [0] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-172-73:926594:926898 [0] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-172-142:880682:880978 [3] NCCL INFO Connected NVLS tree
+ip-26-0-172-142:880682:880978 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-142:880682:880978 [3] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-172-142:880682:880978 [3] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-172-57:942614:942915 [3] NCCL INFO Connected NVLS tree
+ip-26-0-172-57:942614:942915 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-57:942614:942915 [3] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-172-57:942614:942915 [3] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-172-252:688200:688507 [0] NCCL INFO Connected NVLS tree
+ip-26-0-172-252:688200:688507 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-252:688200:688507 [0] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-172-252:688200:688507 [0] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-172-57:942613:942917 [2] NCCL INFO Connected NVLS tree
+ip-26-0-172-57:942613:942917 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-57:942613:942917 [2] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-172-57:942613:942917 [2] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-172-116:1019295:1019591 [7] NCCL INFO Connected NVLS tree
+ip-26-0-172-116:1019295:1019591 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-116:1019295:1019591 [7] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-172-116:1019295:1019591 [7] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-172-142:880681:880983 [2] NCCL INFO Connected NVLS tree
+ip-26-0-172-142:880681:880983 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-142:880681:880983 [2] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-172-142:880681:880983 [2] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-172-252:688207:688504 [7] NCCL INFO Connected NVLS tree
+ip-26-0-172-252:688207:688504 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-252:688207:688504 [7] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-172-252:688207:688504 [7] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-172-252:688201:688500 [1] NCCL INFO Connected NVLS tree
+ip-26-0-172-252:688201:688500 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-252:688201:688500 [1] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-172-252:688201:688500 [1] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-172-57:942615:942912 [4] NCCL INFO Connected NVLS tree
+ip-26-0-172-57:942615:942912 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-57:942615:942912 [4] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-172-57:942615:942912 [4] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-172-252:688204:688502 [4] NCCL INFO Connected NVLS tree
+ip-26-0-172-252:688204:688502 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-252:688204:688502 [4] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-172-252:688204:688502 [4] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-172-142:880683:880982 [4] NCCL INFO Connected NVLS tree
+ip-26-0-172-142:880683:880982 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-142:880683:880982 [4] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-172-142:880683:880982 [4] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-172-252:688203:688501 [3] NCCL INFO Connected NVLS tree
+ip-26-0-172-252:688203:688501 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-252:688203:688501 [3] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-172-252:688203:688501 [3] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-172-57:942611:942911 [0] NCCL INFO Connected NVLS tree
+ip-26-0-172-57:942611:942911 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-57:942611:942911 [0] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-172-57:942611:942911 [0] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-172-57:942617:942916 [6] NCCL INFO Connected NVLS tree
+ip-26-0-172-57:942617:942916 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-57:942617:942916 [6] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-172-57:942617:942916 [6] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-172-73:926599:926895 [5] NCCL INFO comm 0xa251780 rank 53 nranks 64 cudaDev 5 nvmlDev 5 busId a8000 commId 0x6ca6bcab597fda79 - Init COMPLETE
+ip-26-0-172-73:926597:926900 [3] NCCL INFO comm 0x8f37a80 rank 51 nranks 64 cudaDev 3 nvmlDev 3 busId 86000 commId 0x6ca6bcab597fda79 - Init COMPLETE
+ip-26-0-172-73:926598:926894 [4] NCCL INFO comm 0x9ba3630 rank 52 nranks 64 cudaDev 4 nvmlDev 4 busId 97000 commId 0x6ca6bcab597fda79 - Init COMPLETE
+ip-26-0-172-73:926600:926893 [6] NCCL INFO comm 0xa25f780 rank 54 nranks 64 cudaDev 6 nvmlDev 6 busId b9000 commId 0x6ca6bcab597fda79 - Init COMPLETE
+ip-26-0-172-73:926601:926896 [7] NCCL INFO comm 0x8839590 rank 55 nranks 64 cudaDev 7 nvmlDev 7 busId ca000 commId 0x6ca6bcab597fda79 - Init COMPLETE
+ip-26-0-172-73:926594:926898 [0] NCCL INFO comm 0x9665ce0 rank 48 nranks 64 cudaDev 0 nvmlDev 0 busId 53000 commId 0x6ca6bcab597fda79 - Init COMPLETE
+ip-26-0-172-73:926596:926899 [2] NCCL INFO comm 0x90ace20 rank 50 nranks 64 cudaDev 2 nvmlDev 2 busId 75000 commId 0x6ca6bcab597fda79 - Init COMPLETE
+ip-26-0-172-73:926595:926897 [1] NCCL INFO comm 0x875b3b0 rank 49 nranks 64 cudaDev 1 nvmlDev 1 busId 64000 commId 0x6ca6bcab597fda79 - Init COMPLETE
+ip-26-0-172-252:688205:688505 [5] NCCL INFO Connected NVLS tree
+ip-26-0-172-252:688205:688505 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-252:688205:688505 [5] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-172-252:688205:688505 [5] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-160-225:94342:94643 [0] NCCL INFO Connected NVLS tree
+ip-26-0-160-225:94342:94643 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-225:94342:94643 [0] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-160-225:94342:94643 [0] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-172-147:618589:618883 [5] NCCL INFO Connected NVLS tree
+ip-26-0-172-147:618589:618883 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-147:618589:618883 [5] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-172-147:618589:618883 [5] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-172-57:942616:942913 [5] NCCL INFO comm 0x89176f0 rank 45 nranks 64 cudaDev 5 nvmlDev 5 busId a8000 commId 0x6ca6bcab597fda79 - Init COMPLETE
+ip-26-0-172-57:942618:942918 [7] NCCL INFO comm 0x92847f0 rank 47 nranks 64 cudaDev 7 nvmlDev 7 busId ca000 commId 0x6ca6bcab597fda79 - Init COMPLETE
+ip-26-0-172-57:942617:942916 [6] NCCL INFO comm 0x99f7c90 rank 46 nranks 64 cudaDev 6 nvmlDev 6 busId b9000 commId 0x6ca6bcab597fda79 - Init COMPLETE
+ip-26-0-172-57:942613:942917 [2] NCCL INFO comm 0x9dacfa0 rank 42 nranks 64 cudaDev 2 nvmlDev 2 busId 75000 commId 0x6ca6bcab597fda79 - Init COMPLETE
+ip-26-0-172-57:942615:942912 [4] NCCL INFO comm 0xa394dc0 rank 44 nranks 64 cudaDev 4 nvmlDev 4 busId 97000 commId 0x6ca6bcab597fda79 - Init COMPLETE
+ip-26-0-172-57:942612:942914 [1] NCCL INFO comm 0x8de9ae0 rank 41 nranks 64 cudaDev 1 nvmlDev 1 busId 64000 commId 0x6ca6bcab597fda79 - Init COMPLETE
+ip-26-0-172-57:942611:942911 [0] NCCL INFO comm 0xa02b330 rank 40 nranks 64 cudaDev 0 nvmlDev 0 busId 53000 commId 0x6ca6bcab597fda79 - Init COMPLETE
+ip-26-0-172-57:942614:942915 [3] NCCL INFO comm 0x8c1a590 rank 43 nranks 64 cudaDev 3 nvmlDev 3 busId 86000 commId 0x6ca6bcab597fda79 - Init COMPLETE
+ip-26-0-160-225:94348:94650 [6] NCCL INFO Connected NVLS tree
+ip-26-0-160-225:94348:94650 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-225:94348:94650 [6] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-160-225:94348:94650 [6] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-172-142:880685:880977 [6] NCCL INFO Connected NVLS tree
+ip-26-0-172-142:880685:880977 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-142:880685:880977 [6] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-172-142:880685:880977 [6] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-160-225:94347:94646 [5] NCCL INFO comm 0x95790b0 rank 5 nranks 64 cudaDev 5 nvmlDev 5 busId a8000 commId 0x6ca6bcab597fda79 - Init COMPLETE
+ip-26-0-160-225:94343:94647 [1] NCCL INFO comm 0x874a1e0 rank 1 nranks 64 cudaDev 1 nvmlDev 1 busId 64000 commId 0x6ca6bcab597fda79 - Init COMPLETE
+ip-26-0-160-225:94342:94643 [0] NCCL INFO comm 0x9322400 rank 0 nranks 64 cudaDev 0 nvmlDev 0 busId 53000 commId 0x6ca6bcab597fda79 - Init COMPLETE
+ip-26-0-160-225:94349:94645 [7] NCCL INFO comm 0x8e4ac60 rank 7 nranks 64 cudaDev 7 nvmlDev 7 busId ca000 commId 0x6ca6bcab597fda79 - Init COMPLETE
+ip-26-0-160-225:94348:94650 [6] NCCL INFO comm 0x9c46720 rank 6 nranks 64 cudaDev 6 nvmlDev 6 busId b9000 commId 0x6ca6bcab597fda79 - Init COMPLETE
+ip-26-0-160-225:94346:94644 [4] NCCL INFO comm 0xa4e59b0 rank 4 nranks 64 cudaDev 4 nvmlDev 4 busId 97000 commId 0x6ca6bcab597fda79 - Init COMPLETE
+ip-26-0-160-225:94344:94648 [2] NCCL INFO comm 0x949ee50 rank 2 nranks 64 cudaDev 2 nvmlDev 2 busId 75000 commId 0x6ca6bcab597fda79 - Init COMPLETE
+ip-26-0-160-225:94345:94649 [3] NCCL INFO comm 0xa67f5a0 rank 3 nranks 64 cudaDev 3 nvmlDev 3 busId 86000 commId 0x6ca6bcab597fda79 - Init COMPLETE
+ip-26-0-172-252:688206:688506 [6] NCCL INFO comm 0xa6af5e0 rank 38 nranks 64 cudaDev 6 nvmlDev 6 busId b9000 commId 0x6ca6bcab597fda79 - Init COMPLETE
+ip-26-0-172-252:688202:688503 [2] NCCL INFO comm 0x94e7bb0 rank 34 nranks 64 cudaDev 2 nvmlDev 2 busId 75000 commId 0x6ca6bcab597fda79 - Init COMPLETE
+ip-26-0-172-252:688200:688507 [0] NCCL INFO comm 0x965a660 rank 32 nranks 64 cudaDev 0 nvmlDev 0 busId 53000 commId 0x6ca6bcab597fda79 - Init COMPLETE
+ip-26-0-172-252:688201:688500 [1] NCCL INFO comm 0x96030f0 rank 33 nranks 64 cudaDev 1 nvmlDev 1 busId 64000 commId 0x6ca6bcab597fda79 - Init COMPLETE
+ip-26-0-172-252:688207:688504 [7] NCCL INFO comm 0x91ccc70 rank 39 nranks 64 cudaDev 7 nvmlDev 7 busId ca000 commId 0x6ca6bcab597fda79 - Init COMPLETE
+ip-26-0-172-252:688204:688502 [4] NCCL INFO comm 0x9756c40 rank 36 nranks 64 cudaDev 4 nvmlDev 4 busId 97000 commId 0x6ca6bcab597fda79 - Init COMPLETE
+ip-26-0-172-252:688203:688501 [3] NCCL INFO comm 0x8a56220 rank 35 nranks 64 cudaDev 3 nvmlDev 3 busId 86000 commId 0x6ca6bcab597fda79 - Init COMPLETE
+ip-26-0-172-252:688205:688505 [5] NCCL INFO comm 0x90c5ab0 rank 37 nranks 64 cudaDev 5 nvmlDev 5 busId a8000 commId 0x6ca6bcab597fda79 - Init COMPLETE
+ip-26-0-172-147:618587:618887 [3] NCCL INFO Connected NVLS tree
+ip-26-0-172-147:618587:618887 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-147:618587:618887 [3] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-172-147:618587:618887 [3] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-172-147:618585:618884 [1] NCCL INFO Connected NVLS tree
+ip-26-0-172-147:618585:618884 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-147:618585:618884 [1] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-172-147:618585:618884 [1] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-172-116:1019294:1019589 [6] NCCL INFO Connected NVLS tree
+ip-26-0-172-116:1019294:1019589 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-116:1019294:1019589 [6] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-172-116:1019294:1019589 [6] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-172-147:618590:618889 [6] NCCL INFO Connected NVLS tree
+ip-26-0-172-147:618590:618889 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-147:618590:618889 [6] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-172-147:618590:618889 [6] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-172-116:1019288:1019588 [0] NCCL INFO Connected NVLS tree
+ip-26-0-172-116:1019288:1019588 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-116:1019288:1019588 [0] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-172-116:1019288:1019588 [0] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-172-116:1019291:1019594 [3] NCCL INFO Connected NVLS tree
+ip-26-0-172-116:1019291:1019594 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-116:1019291:1019594 [3] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-172-116:1019291:1019594 [3] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-172-147:618591:618885 [7] NCCL INFO Connected NVLS tree
+ip-26-0-172-147:618591:618885 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-147:618591:618885 [7] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-172-147:618591:618885 [7] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-172-116:1019290:1019595 [2] NCCL INFO Connected NVLS tree
+ip-26-0-172-116:1019290:1019595 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-116:1019290:1019595 [2] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-172-116:1019290:1019595 [2] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-172-116:1019289:1019593 [1] NCCL INFO Connected NVLS tree
+ip-26-0-172-116:1019289:1019593 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-116:1019289:1019593 [1] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-172-116:1019289:1019593 [1] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-172-116:1019292:1019590 [4] NCCL INFO Connected NVLS tree
+ip-26-0-172-116:1019292:1019590 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-116:1019292:1019590 [4] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-172-116:1019292:1019590 [4] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-172-147:618588:618888 [4] NCCL INFO Connected NVLS tree
+ip-26-0-172-147:618588:618888 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-147:618588:618888 [4] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-172-147:618588:618888 [4] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-172-147:618586:618886 [2] NCCL INFO Connected NVLS tree
+ip-26-0-172-147:618586:618886 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-147:618586:618886 [2] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-172-147:618586:618886 [2] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-172-142:880679:880980 [0] NCCL INFO Connected NVLS tree
+ip-26-0-172-142:880679:880980 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-142:880679:880980 [0] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-172-142:880679:880980 [0] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-172-147:618584:618882 [0] NCCL INFO Connected NVLS tree
+ip-26-0-172-147:618584:618882 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-147:618584:618882 [0] NCCL INFO threadThresholds 8/8/64 | 512/8/64 | 512 | 512
+ip-26-0-172-147:618584:618882 [0] NCCL INFO 16 coll channels, 16 nvls channels, 16 p2p channels, 2 p2p channels per peer
+ip-26-0-172-116:1019288:1019588 [0] NCCL INFO comm 0x8e0d8a0 rank 8 nranks 64 cudaDev 0 nvmlDev 0 busId 53000 commId 0x6ca6bcab597fda79 - Init COMPLETE
+ip-26-0-172-116:1019294:1019589 [6] NCCL INFO comm 0x8886e00 rank 14 nranks 64 cudaDev 6 nvmlDev 6 busId b9000 commId 0x6ca6bcab597fda79 - Init COMPLETE
+ip-26-0-172-116:1019293:1019592 [5] NCCL INFO comm 0xa68afb0 rank 13 nranks 64 cudaDev 5 nvmlDev 5 busId a8000 commId 0x6ca6bcab597fda79 - Init COMPLETE
+ip-26-0-172-116:1019292:1019590 [4] NCCL INFO comm 0xa5e9420 rank 12 nranks 64 cudaDev 4 nvmlDev 4 busId 97000 commId 0x6ca6bcab597fda79 - Init COMPLETE
+ip-26-0-172-116:1019295:1019591 [7] NCCL INFO comm 0x9c544f0 rank 15 nranks 64 cudaDev 7 nvmlDev 7 busId ca000 commId 0x6ca6bcab597fda79 - Init COMPLETE
+ip-26-0-172-116:1019289:1019593 [1] NCCL INFO comm 0x8887c60 rank 9 nranks 64 cudaDev 1 nvmlDev 1 busId 64000 commId 0x6ca6bcab597fda79 - Init COMPLETE
+ip-26-0-172-116:1019290:1019595 [2] NCCL INFO comm 0xa2736a0 rank 10 nranks 64 cudaDev 2 nvmlDev 2 busId 75000 commId 0x6ca6bcab597fda79 - Init COMPLETE
+ip-26-0-172-116:1019291:1019594 [3] NCCL INFO comm 0xa5fbcf0 rank 11 nranks 64 cudaDev 3 nvmlDev 3 busId 86000 commId 0x6ca6bcab597fda79 - Init COMPLETE
+ip-26-0-172-142:880680:880984 [1] NCCL INFO comm 0x91fe360 rank 17 nranks 64 cudaDev 1 nvmlDev 1 busId 64000 commId 0x6ca6bcab597fda79 - Init COMPLETE
+ip-26-0-172-142:880679:880980 [0] NCCL INFO comm 0xa3a4d80 rank 16 nranks 64 cudaDev 0 nvmlDev 0 busId 53000 commId 0x6ca6bcab597fda79 - Init COMPLETE
+ip-26-0-172-142:880683:880982 [4] NCCL INFO comm 0x9ee12e0 rank 20 nranks 64 cudaDev 4 nvmlDev 4 busId 97000 commId 0x6ca6bcab597fda79 - Init COMPLETE
+ip-26-0-172-142:880684:880981 [5] NCCL INFO comm 0x8d065a0 rank 21 nranks 64 cudaDev 5 nvmlDev 5 busId a8000 commId 0x6ca6bcab597fda79 - Init COMPLETE
+ip-26-0-172-142:880685:880977 [6] NCCL INFO comm 0x8bbe030 rank 22 nranks 64 cudaDev 6 nvmlDev 6 busId b9000 commId 0x6ca6bcab597fda79 - Init COMPLETE
+ip-26-0-172-142:880681:880983 [2] NCCL INFO comm 0x92c70b0 rank 18 nranks 64 cudaDev 2 nvmlDev 2 busId 75000 commId 0x6ca6bcab597fda79 - Init COMPLETE
+ip-26-0-172-142:880686:880979 [7] NCCL INFO comm 0x99b27a0 rank 23 nranks 64 cudaDev 7 nvmlDev 7 busId ca000 commId 0x6ca6bcab597fda79 - Init COMPLETE
+ip-26-0-172-142:880682:880978 [3] NCCL INFO comm 0x9765da0 rank 19 nranks 64 cudaDev 3 nvmlDev 3 busId 86000 commId 0x6ca6bcab597fda79 - Init COMPLETE
+ip-26-0-172-147:618591:618885 [7] NCCL INFO comm 0x9627a20 rank 31 nranks 64 cudaDev 7 nvmlDev 7 busId ca000 commId 0x6ca6bcab597fda79 - Init COMPLETE
+ip-26-0-172-147:618590:618889 [6] NCCL INFO comm 0xa40dfa0 rank 30 nranks 64 cudaDev 6 nvmlDev 6 busId b9000 commId 0x6ca6bcab597fda79 - Init COMPLETE
+ip-26-0-172-147:618588:618888 [4] NCCL INFO comm 0x8ab30c0 rank 28 nranks 64 cudaDev 4 nvmlDev 4 busId 97000 commId 0x6ca6bcab597fda79 - Init COMPLETE
+ip-26-0-172-147:618589:618883 [5] NCCL INFO comm 0x9a820e0 rank 29 nranks 64 cudaDev 5 nvmlDev 5 busId a8000 commId 0x6ca6bcab597fda79 - Init COMPLETE
+ip-26-0-172-147:618586:618886 [2] NCCL INFO comm 0x8da0660 rank 26 nranks 64 cudaDev 2 nvmlDev 2 busId 75000 commId 0x6ca6bcab597fda79 - Init COMPLETE
+ip-26-0-172-147:618587:618887 [3] NCCL INFO comm 0x8dedc20 rank 27 nranks 64 cudaDev 3 nvmlDev 3 busId 86000 commId 0x6ca6bcab597fda79 - Init COMPLETE
+ip-26-0-172-147:618584:618882 [0] NCCL INFO comm 0x9d5f5f0 rank 24 nranks 64 cudaDev 0 nvmlDev 0 busId 53000 commId 0x6ca6bcab597fda79 - Init COMPLETE
+ip-26-0-172-147:618585:618884 [1] NCCL INFO comm 0xa2724f0 rank 25 nranks 64 cudaDev 1 nvmlDev 1 busId 64000 commId 0x6ca6bcab597fda79 - Init COMPLETE
+ip-26-0-160-225:94343:94681 [1] NCCL INFO Using network Libfabric
+ip-26-0-160-225:94342:94680 [0] NCCL INFO Using network Libfabric
+ip-26-0-160-225:94347:94685 [5] NCCL INFO Using network Libfabric
+ip-26-0-160-225:94349:94683 [7] NCCL INFO Using network Libfabric
+ip-26-0-160-225:94344:94686 [2] NCCL INFO Using network Libfabric
+ip-26-0-160-225:94346:94682 [4] NCCL INFO Using network Libfabric
+ip-26-0-160-225:94348:94684 [6] NCCL INFO Using network Libfabric
+ip-26-0-160-225:94345:94687 [3] NCCL INFO Using network Libfabric
+ip-26-0-160-225:94344:94686 [2] NCCL INFO comm 0x94b6540 rank 2 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0x3ef1a0376dc31f7 - Init START
+ip-26-0-160-225:94343:94681 [1] NCCL INFO comm 0x87612b0 rank 1 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0x3ef1a0376dc31f7 - Init START
+ip-26-0-160-225:94348:94684 [6] NCCL INFO comm 0x9c5c6f0 rank 6 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0x3ef1a0376dc31f7 - Init START
+ip-26-0-160-225:94346:94682 [4] NCCL INFO comm 0xa4faf60 rank 4 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0x3ef1a0376dc31f7 - Init START
+ip-26-0-160-225:94345:94687 [3] NCCL INFO comm 0xa6960e0 rank 3 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0x3ef1a0376dc31f7 - Init START
+ip-26-0-160-225:94349:94683 [7] NCCL INFO comm 0x8e619f0 rank 7 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0x3ef1a0376dc31f7 - Init START
+ip-26-0-160-225:94347:94685 [5] NCCL INFO comm 0x9590450 rank 5 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0x3ef1a0376dc31f7 - Init START
+NCCL version 2.18.6+cuda12.1
+ip-26-0-172-252:688200:688538 [0] NCCL INFO Using network Libfabric
+ip-26-0-172-252:688204:688540 [4] NCCL INFO Using network Libfabric
+ip-26-0-172-252:688206:688539 [6] NCCL INFO Using network Libfabric
+ip-26-0-172-252:688207:688543 [7] NCCL INFO Using network Libfabric
+ip-26-0-172-252:688201:688542 [1] NCCL INFO Using network Libfabric
+ip-26-0-172-252:688202:688544 [2] NCCL INFO Using network Libfabric
+ip-26-0-172-252:688203:688541 [3] NCCL INFO Using network Libfabric
+ip-26-0-172-252:688205:688545 [5] NCCL INFO Using network Libfabric
+ip-26-0-172-252:688200:688538 [0] NCCL INFO comm 0x9672c10 rank 0 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0xb7c059e1a5ed57c - Init START
+ip-26-0-172-252:688201:688542 [1] NCCL INFO comm 0x9619490 rank 1 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0xb7c059e1a5ed57c - Init START
+ip-26-0-172-252:688207:688543 [7] NCCL INFO comm 0x91e3790 rank 7 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0xb7c059e1a5ed57c - Init START
+ip-26-0-172-252:688204:688540 [4] NCCL INFO comm 0x976f3c0 rank 4 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0xb7c059e1a5ed57c - Init START
+ip-26-0-172-252:688206:688539 [6] NCCL INFO comm 0xa6c7860 rank 6 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0xb7c059e1a5ed57c - Init START
+ip-26-0-172-252:688202:688544 [2] NCCL INFO comm 0x94fef70 rank 2 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0xb7c059e1a5ed57c - Init START
+ip-26-0-172-252:688203:688541 [3] NCCL INFO comm 0x8a6ed40 rank 3 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0xb7c059e1a5ed57c - Init START
+ip-26-0-172-252:688205:688545 [5] NCCL INFO comm 0x90dc4c0 rank 5 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0xb7c059e1a5ed57c - Init START
+ip-26-0-172-252:688203:688541 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-252:688204:688540 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-252:688201:688542 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-252:688202:688544 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-252:688206:688539 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-252:688205:688545 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-252:688200:688538 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-252:688207:688543 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+NCCL version 2.18.6+cuda12.1
+ip-26-0-172-73:926594:926931 [0] NCCL INFO Using network Libfabric
+ip-26-0-172-73:926600:926933 [6] NCCL INFO Using network Libfabric
+ip-26-0-172-73:926601:926934 [7] NCCL INFO Using network Libfabric
+ip-26-0-172-73:926599:926932 [5] NCCL INFO Using network Libfabric
+ip-26-0-172-73:926596:926936 [2] NCCL INFO Using network Libfabric
+ip-26-0-172-73:926595:926937 [1] NCCL INFO Using network Libfabric
+ip-26-0-172-73:926598:926938 [4] NCCL INFO Using network Libfabric
+ip-26-0-172-73:926597:926935 [3] NCCL INFO Using network Libfabric
+NCCL version 2.18.6+cuda12.1
+ip-26-0-172-73:926598:926938 [4] NCCL INFO comm 0x9bba460 rank 4 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0x553420158d220013 - Init START
+ip-26-0-172-73:926597:926935 [3] NCCL INFO comm 0x8f50bb0 rank 3 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0x553420158d220013 - Init START
+ip-26-0-172-73:926596:926936 [2] NCCL INFO comm 0x90c2ad0 rank 2 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0x553420158d220013 - Init START
+ip-26-0-172-73:926595:926937 [1] NCCL INFO comm 0x8771e90 rank 1 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0x553420158d220013 - Init START
+ip-26-0-172-73:926594:926931 [0] NCCL INFO comm 0x967ca20 rank 0 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0x553420158d220013 - Init START
+ip-26-0-172-73:926601:926934 [7] NCCL INFO comm 0x884d470 rank 7 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0x553420158d220013 - Init START
+ip-26-0-172-73:926599:926932 [5] NCCL INFO comm 0xa2685e0 rank 5 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0x553420158d220013 - Init START
+ip-26-0-172-73:926600:926933 [6] NCCL INFO comm 0xa278570 rank 6 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0x553420158d220013 - Init START
+ip-26-0-172-73:926596:926936 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-73:926598:926938 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-73:926594:926931 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-73:926595:926937 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-73:926600:926933 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-73:926601:926934 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-73:926597:926935 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-73:926599:926932 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+NCCL version 2.18.6+cuda12.1
+ip-26-0-172-147:618584:618919 [0] NCCL INFO Using network Libfabric
+ip-26-0-172-147:618591:618921 [7] NCCL INFO Using network Libfabric
+ip-26-0-172-147:618587:618924 [3] NCCL INFO Using network Libfabric
+ip-26-0-172-147:618585:618926 [1] NCCL INFO Using network Libfabric
+ip-26-0-172-147:618586:618923 [2] NCCL INFO Using network Libfabric
+ip-26-0-172-147:618590:618925 [6] NCCL INFO Using network Libfabric
+ip-26-0-172-147:618588:618920 [4] NCCL INFO Using network Libfabric
+ip-26-0-172-142:880684:881017 [5] NCCL INFO Using network Libfabric
+ip-26-0-172-142:880679:881015 [0] NCCL INFO Using network Libfabric
+ip-26-0-172-142:880681:881019 [2] NCCL INFO Using network Libfabric
+ip-26-0-172-142:880683:881016 [4] NCCL INFO Using network Libfabric
+ip-26-0-172-142:880685:881018 [6] NCCL INFO Using network Libfabric
+ip-26-0-172-142:880680:881022 [1] NCCL INFO Using network Libfabric
+ip-26-0-172-142:880686:881021 [7] NCCL INFO Using network Libfabric
+ip-26-0-172-147:618589:618922 [5] NCCL INFO Using network Libfabric
+NCCL version 2.18.6+cuda12.1
+ip-26-0-172-142:880682:881020 [3] NCCL INFO Using network Libfabric
+ip-26-0-173-7:811248:811591 [0] NCCL INFO Using network Libfabric
+ip-26-0-172-147:618585:618926 [1] NCCL INFO comm 0xa287aa0 rank 1 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0xe0cb904ba16faf13 - Init START
+ip-26-0-172-147:618589:618922 [5] NCCL INFO comm 0x9a98e60 rank 5 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0xe0cb904ba16faf13 - Init START
+ip-26-0-172-147:618588:618920 [4] NCCL INFO comm 0x8ac5d80 rank 4 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0xe0cb904ba16faf13 - Init START
+ip-26-0-172-147:618584:618919 [0] NCCL INFO comm 0x9d77ab0 rank 0 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0xe0cb904ba16faf13 - Init START
+ip-26-0-172-147:618586:618923 [2] NCCL INFO comm 0x8db71a0 rank 2 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0xe0cb904ba16faf13 - Init START
+ip-26-0-172-147:618590:618925 [6] NCCL INFO comm 0xa4275b0 rank 6 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0xe0cb904ba16faf13 - Init START
+ip-26-0-172-147:618591:618921 [7] NCCL INFO comm 0x963e350 rank 7 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0xe0cb904ba16faf13 - Init START
+ip-26-0-172-147:618587:618924 [3] NCCL INFO comm 0x8e05020 rank 3 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0xe0cb904ba16faf13 - Init START
+ip-26-0-173-7:811254:811594 [6] NCCL INFO Using network Libfabric
+ip-26-0-172-147:618585:618926 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-173-7:811249:811596 [1] NCCL INFO Using network Libfabric
+ip-26-0-172-147:618584:618919 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-173-7:811250:811597 [2] NCCL INFO Using network Libfabric
+ip-26-0-172-147:618589:618922 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-147:618591:618921 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-147:618590:618925 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-147:618588:618920 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-147:618587:618924 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-147:618586:618923 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-173-7:811252:811592 [4] NCCL INFO Using network Libfabric
+ip-26-0-173-7:811253:811593 [5] NCCL INFO Using network Libfabric
+ip-26-0-173-7:811251:811598 [3] NCCL INFO Using network Libfabric
+ip-26-0-172-142:880683:881016 [4] NCCL INFO comm 0x9ef9b70 rank 4 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0x68a686842a70745a - Init START
+ip-26-0-172-142:880682:881020 [3] NCCL INFO comm 0x977b900 rank 3 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0x68a686842a70745a - Init START
+ip-26-0-172-142:880684:881017 [5] NCCL INFO comm 0x8d1d420 rank 5 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0x68a686842a70745a - Init START
+ip-26-0-172-142:880681:881019 [2] NCCL INFO comm 0x92db6c0 rank 2 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0x68a686842a70745a - Init START
+ip-26-0-172-142:880685:881018 [6] NCCL INFO comm 0x8bd2f30 rank 6 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0x68a686842a70745a - Init START
+ip-26-0-172-142:880680:881022 [1] NCCL INFO comm 0x9213170 rank 1 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0x68a686842a70745a - Init START
+ip-26-0-172-142:880686:881021 [7] NCCL INFO comm 0x99c84d0 rank 7 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0x68a686842a70745a - Init START
+ip-26-0-173-7:811255:811595 [7] NCCL INFO Using network Libfabric
+ip-26-0-172-142:880679:881015 [0] NCCL INFO comm 0xa3bc420 rank 0 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0x68a686842a70745a - Init START
+ip-26-0-172-142:880684:881017 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-142:880683:881016 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-142:880682:881020 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-142:880681:881019 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-142:880685:881018 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-142:880686:881021 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-142:880680:881022 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-142:880679:881015 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-173-7:811254:811594 [6] NCCL INFO comm 0x8e04770 rank 6 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0xb6d3276631c21f81 - Init START
+ip-26-0-173-7:811255:811595 [7] NCCL INFO comm 0xa04ac50 rank 7 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0xb6d3276631c21f81 - Init START
+ip-26-0-173-7:811252:811592 [4] NCCL INFO comm 0x911a140 rank 4 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0xb6d3276631c21f81 - Init START
+ip-26-0-173-7:811253:811593 [5] NCCL INFO comm 0x9a6e690 rank 5 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0xb6d3276631c21f81 - Init START
+ip-26-0-173-7:811251:811598 [3] NCCL INFO comm 0x8ffae80 rank 3 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0xb6d3276631c21f81 - Init START
+ip-26-0-173-7:811249:811596 [1] NCCL INFO comm 0x8bffbd0 rank 1 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0xb6d3276631c21f81 - Init START
+ip-26-0-173-7:811248:811591 [0] NCCL INFO comm 0xa5f0580 rank 0 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0xb6d3276631c21f81 - Init START
+ip-26-0-173-7:811250:811597 [2] NCCL INFO comm 0x99fce90 rank 2 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0xb6d3276631c21f81 - Init START
+ip-26-0-173-7:811255:811595 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-173-7:811254:811594 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-173-7:811252:811592 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-173-7:811253:811593 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-173-7:811250:811597 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-173-7:811249:811596 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-173-7:811251:811598 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-173-7:811248:811591 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+NCCL version 2.18.6+cuda12.1
+ip-26-0-172-116:1019288:1019623 [0] NCCL INFO Using network Libfabric
+ip-26-0-172-116:1019294:1019626 [6] NCCL INFO Using network Libfabric
+ip-26-0-172-116:1019292:1019624 [4] NCCL INFO Using network Libfabric
+ip-26-0-172-116:1019293:1019625 [5] NCCL INFO Using network Libfabric
+ip-26-0-172-116:1019295:1019627 [7] NCCL INFO Using network Libfabric
+ip-26-0-172-116:1019289:1019629 [1] NCCL INFO Using network Libfabric
+NCCL version 2.18.6+cuda12.1
+ip-26-0-172-116:1019291:1019630 [3] NCCL INFO Using network Libfabric
+ip-26-0-172-116:1019290:1019628 [2] NCCL INFO Using network Libfabric
+ip-26-0-172-116:1019293:1019625 [5] NCCL INFO comm 0xa6a1880 rank 5 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0xe87bb15d3c873f23 - Init START
+ip-26-0-172-116:1019292:1019624 [4] NCCL INFO comm 0xa6017e0 rank 4 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0xe87bb15d3c873f23 - Init START
+ip-26-0-172-116:1019290:1019628 [2] NCCL INFO comm 0xa28b4a0 rank 2 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0xe87bb15d3c873f23 - Init START
+ip-26-0-172-116:1019291:1019630 [3] NCCL INFO comm 0xa6143c0 rank 3 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0xe87bb15d3c873f23 - Init START
+ip-26-0-172-116:1019294:1019626 [6] NCCL INFO comm 0x889d980 rank 6 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0xe87bb15d3c873f23 - Init START
+ip-26-0-172-116:1019289:1019629 [1] NCCL INFO comm 0x889ef90 rank 1 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0xe87bb15d3c873f23 - Init START
+ip-26-0-172-116:1019295:1019627 [7] NCCL INFO comm 0x9c6b820 rank 7 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0xe87bb15d3c873f23 - Init START
+ip-26-0-172-116:1019288:1019623 [0] NCCL INFO comm 0x8e255e0 rank 0 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0xe87bb15d3c873f23 - Init START
+ip-26-0-172-116:1019294:1019626 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-116:1019293:1019625 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-116:1019291:1019630 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-116:1019290:1019628 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-116:1019288:1019623 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-116:1019295:1019627 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-116:1019289:1019629 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-116:1019292:1019624 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-57:942611:942947 [0] NCCL INFO Using network Libfabric
+ip-26-0-172-57:942615:942948 [4] NCCL INFO Using network Libfabric
+ip-26-0-172-57:942617:942949 [6] NCCL INFO Using network Libfabric
+ip-26-0-172-57:942613:942950 [2] NCCL INFO Using network Libfabric
+ip-26-0-172-57:942614:942953 [3] NCCL INFO Using network Libfabric
+ip-26-0-172-57:942618:942951 [7] NCCL INFO Using network Libfabric
+ip-26-0-172-57:942612:942954 [1] NCCL INFO Using network Libfabric
+ip-26-0-172-57:942616:942952 [5] NCCL INFO Using network Libfabric
+ip-26-0-172-57:942617:942949 [6] NCCL INFO comm 0x9a0fe60 rank 6 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0xcab7e2994d17fa8a - Init START
+ip-26-0-172-57:942616:942952 [5] NCCL INFO comm 0x892e730 rank 5 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0xcab7e2994d17fa8a - Init START
+ip-26-0-172-57:942614:942953 [3] NCCL INFO comm 0x8c33510 rank 3 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0xcab7e2994d17fa8a - Init START
+ip-26-0-172-57:942615:942948 [4] NCCL INFO comm 0xa3a8410 rank 4 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0xcab7e2994d17fa8a - Init START
+ip-26-0-172-57:942613:942950 [2] NCCL INFO comm 0x9dc3d50 rank 2 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0xcab7e2994d17fa8a - Init START
+ip-26-0-172-57:942618:942951 [7] NCCL INFO comm 0x929bad0 rank 7 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0xcab7e2994d17fa8a - Init START
+ip-26-0-172-57:942612:942954 [1] NCCL INFO comm 0x8e01660 rank 1 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0xcab7e2994d17fa8a - Init START
+ip-26-0-172-57:942611:942947 [0] NCCL INFO comm 0xa0430a0 rank 0 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0xcab7e2994d17fa8a - Init START
+ip-26-0-172-57:942618:942951 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-57:942617:942949 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-57:942614:942953 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-57:942611:942947 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-57:942615:942948 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-57:942613:942950 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-57:942612:942954 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-57:942616:942952 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-160-225:94342:94680 [0] NCCL INFO comm 0x93142a0 rank 0 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0x3ef1a0376dc31f7 - Init START
+ip-26-0-160-225:94343:94681 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-160-225:94345:94687 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-160-225:94347:94685 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-160-225:94342:94680 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/349
+ip-26-0-160-225:94349:94683 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-160-225:94346:94682 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-160-225:94344:94686 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-160-225:94348:94684 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-160-225:94342:94680 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-160-225:94342:94680 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-172-147:618586:618923 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-172-147:618586:618923 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-172-252:688202:688544 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-172-252:688202:688544 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-172-116:1019289:1019629 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-172-116:1019289:1019629 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-160-225:94343:94681 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-160-225:94343:94681 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-160-225:94346:94682 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-160-225:94346:94682 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-173-7:811251:811598 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-173-7:811251:811598 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-172-252:688207:688543 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-172-252:688207:688543 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-172-73:926598:926938 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-172-73:926598:926938 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-160-225:94347:94685 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-160-225:94347:94685 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-172-252:688205:688545 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-172-252:688205:688545 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-172-116:1019295:1019627 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-172-116:1019295:1019627 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-172-252:688203:688541 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-172-252:688203:688541 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-160-225:94345:94687 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-160-225:94345:94687 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-172-252:688200:688538 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-172-252:688200:688538 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-160-225:94349:94683 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-160-225:94349:94683 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-172-142:880682:881020 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-172-142:880682:881020 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-160-225:94348:94684 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-160-225:94348:94684 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-172-116:1019291:1019630 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-172-116:1019291:1019630 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-172-116:1019293:1019625 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-172-116:1019293:1019625 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-173-7:811250:811597 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-173-7:811250:811597 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-172-116:1019292:1019624 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-172-116:1019288:1019623 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-172-116:1019288:1019623 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-172-116:1019292:1019624 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-160-225:94344:94686 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-160-225:94344:94686 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-172-73:926599:926932 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-172-73:926599:926932 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-160-225:94342:94680 [0] NCCL INFO Channel 00/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-225:94344:94686 [2] NCCL INFO Trees [0] 3/-1/-1->2->1 [1] 3/-1/-1->2->1 [2] 3/-1/-1->2->1 [3] 3/-1/-1->2->1 [4] 3/-1/-1->2->1 [5] 3/-1/-1->2->1 [6] 3/-1/-1->2->1 [7] 3/-1/-1->2->1 [8] 3/-1/-1->2->1 [9] 3/-1/-1->2->1 [10] 3/-1/-1->2->1 [11] 3/-1/-1->2->1 [12] 3/-1/-1->2->1 [13] 3/-1/-1->2->1 [14] 3/-1/-1->2->1 [15] 3/-1/-1->2->1 [16] 3/-1/-1->2->1 [17] 3/-1/-1->2->1 [18] 3/-1/-1->2->1 [19] 3/-1/-1->2->1 [20] 3/-1/-1->2->1 [21] 3/-1/-1->2->1 [22] 3/-1/-1->2->1 [23] 3/-1/-1->2->1
+ip-26-0-160-225:94342:94680 [0] NCCL INFO Channel 01/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-225:94344:94686 [2] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-160-225:94342:94680 [0] NCCL INFO Channel 02/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-225:94343:94681 [1] NCCL INFO Trees [0] 2/-1/-1->1->0 [1] 2/-1/-1->1->0 [2] 2/-1/-1->1->0 [3] 2/-1/-1->1->0 [4] 2/-1/-1->1->0 [5] 2/-1/-1->1->0 [6] 2/-1/-1->1->0 [7] 2/-1/-1->1->0 [8] 2/-1/-1->1->0 [9] 2/-1/-1->1->0 [10] 2/-1/-1->1->0 [11] 2/-1/-1->1->0 [12] 2/-1/-1->1->0 [13] 2/-1/-1->1->0 [14] 2/-1/-1->1->0 [15] 2/-1/-1->1->0 [16] 2/-1/-1->1->0 [17] 2/-1/-1->1->0 [18] 2/-1/-1->1->0 [19] 2/-1/-1->1->0 [20] 2/-1/-1->1->0 [21] 2/-1/-1->1->0 [22] 2/-1/-1->1->0 [23] 2/-1/-1->1->0
+ip-26-0-160-225:94342:94680 [0] NCCL INFO Channel 03/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-225:94343:94681 [1] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-160-225:94342:94680 [0] NCCL INFO Channel 04/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-225:94342:94680 [0] NCCL INFO Channel 05/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-225:94342:94680 [0] NCCL INFO Channel 06/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-225:94342:94680 [0] NCCL INFO Channel 07/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-225:94342:94680 [0] NCCL INFO Channel 08/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-225:94342:94680 [0] NCCL INFO Channel 09/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-225:94345:94687 [3] NCCL INFO Trees [0] 4/-1/-1->3->2 [1] 4/-1/-1->3->2 [2] 4/-1/-1->3->2 [3] 4/-1/-1->3->2 [4] 4/-1/-1->3->2 [5] 4/-1/-1->3->2 [6] 4/-1/-1->3->2 [7] 4/-1/-1->3->2 [8] 4/-1/-1->3->2 [9] 4/-1/-1->3->2 [10] 4/-1/-1->3->2 [11] 4/-1/-1->3->2 [12] 4/-1/-1->3->2 [13] 4/-1/-1->3->2 [14] 4/-1/-1->3->2 [15] 4/-1/-1->3->2 [16] 4/-1/-1->3->2 [17] 4/-1/-1->3->2 [18] 4/-1/-1->3->2 [19] 4/-1/-1->3->2 [20] 4/-1/-1->3->2 [21] 4/-1/-1->3->2 [22] 4/-1/-1->3->2 [23] 4/-1/-1->3->2
+ip-26-0-160-225:94342:94680 [0] NCCL INFO Channel 10/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-225:94349:94683 [7] NCCL INFO Trees [0] -1/-1/-1->7->6 [1] -1/-1/-1->7->6 [2] -1/-1/-1->7->6 [3] -1/-1/-1->7->6 [4] -1/-1/-1->7->6 [5] -1/-1/-1->7->6 [6] -1/-1/-1->7->6 [7] -1/-1/-1->7->6 [8] -1/-1/-1->7->6 [9] -1/-1/-1->7->6 [10] -1/-1/-1->7->6 [11] -1/-1/-1->7->6 [12] -1/-1/-1->7->6 [13] -1/-1/-1->7->6 [14] -1/-1/-1->7->6 [15] -1/-1/-1->7->6 [16] -1/-1/-1->7->6 [17] -1/-1/-1->7->6 [18] -1/-1/-1->7->6 [19] -1/-1/-1->7->6 [20] -1/-1/-1->7->6 [21] -1/-1/-1->7->6 [22] -1/-1/-1->7->6 [23] -1/-1/-1->7->6
+ip-26-0-160-225:94342:94680 [0] NCCL INFO Channel 11/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-225:94345:94687 [3] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-160-225:94342:94680 [0] NCCL INFO Channel 12/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-225:94349:94683 [7] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-160-225:94342:94680 [0] NCCL INFO Channel 13/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-225:94348:94684 [6] NCCL INFO Trees [0] 7/-1/-1->6->5 [1] 7/-1/-1->6->5 [2] 7/-1/-1->6->5 [3] 7/-1/-1->6->5 [4] 7/-1/-1->6->5 [5] 7/-1/-1->6->5 [6] 7/-1/-1->6->5 [7] 7/-1/-1->6->5 [8] 7/-1/-1->6->5 [9] 7/-1/-1->6->5 [10] 7/-1/-1->6->5 [11] 7/-1/-1->6->5 [12] 7/-1/-1->6->5 [13] 7/-1/-1->6->5 [14] 7/-1/-1->6->5 [15] 7/-1/-1->6->5 [16] 7/-1/-1->6->5 [17] 7/-1/-1->6->5 [18] 7/-1/-1->6->5 [19] 7/-1/-1->6->5 [20] 7/-1/-1->6->5 [21] 7/-1/-1->6->5 [22] 7/-1/-1->6->5 [23] 7/-1/-1->6->5
+ip-26-0-160-225:94342:94680 [0] NCCL INFO Channel 14/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-225:94348:94684 [6] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-160-225:94342:94680 [0] NCCL INFO Channel 15/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-225:94347:94685 [5] NCCL INFO Trees [0] 6/-1/-1->5->4 [1] 6/-1/-1->5->4 [2] 6/-1/-1->5->4 [3] 6/-1/-1->5->4 [4] 6/-1/-1->5->4 [5] 6/-1/-1->5->4 [6] 6/-1/-1->5->4 [7] 6/-1/-1->5->4 [8] 6/-1/-1->5->4 [9] 6/-1/-1->5->4 [10] 6/-1/-1->5->4 [11] 6/-1/-1->5->4 [12] 6/-1/-1->5->4 [13] 6/-1/-1->5->4 [14] 6/-1/-1->5->4 [15] 6/-1/-1->5->4 [16] 6/-1/-1->5->4 [17] 6/-1/-1->5->4 [18] 6/-1/-1->5->4 [19] 6/-1/-1->5->4 [20] 6/-1/-1->5->4 [21] 6/-1/-1->5->4 [22] 6/-1/-1->5->4 [23] 6/-1/-1->5->4
+ip-26-0-160-225:94346:94682 [4] NCCL INFO Trees [0] 5/-1/-1->4->3 [1] 5/-1/-1->4->3 [2] 5/-1/-1->4->3 [3] 5/-1/-1->4->3 [4] 5/-1/-1->4->3 [5] 5/-1/-1->4->3 [6] 5/-1/-1->4->3 [7] 5/-1/-1->4->3 [8] 5/-1/-1->4->3 [9] 5/-1/-1->4->3 [10] 5/-1/-1->4->3 [11] 5/-1/-1->4->3 [12] 5/-1/-1->4->3 [13] 5/-1/-1->4->3 [14] 5/-1/-1->4->3 [15] 5/-1/-1->4->3 [16] 5/-1/-1->4->3 [17] 5/-1/-1->4->3 [18] 5/-1/-1->4->3 [19] 5/-1/-1->4->3 [20] 5/-1/-1->4->3 [21] 5/-1/-1->4->3 [22] 5/-1/-1->4->3 [23] 5/-1/-1->4->3
+ip-26-0-160-225:94342:94680 [0] NCCL INFO Channel 16/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-225:94342:94680 [0] NCCL INFO Channel 17/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-225:94346:94682 [4] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-160-225:94347:94685 [5] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-160-225:94342:94680 [0] NCCL INFO Channel 18/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-225:94342:94680 [0] NCCL INFO Channel 19/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-225:94342:94680 [0] NCCL INFO Channel 20/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-225:94342:94680 [0] NCCL INFO Channel 21/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-225:94342:94680 [0] NCCL INFO Channel 22/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-225:94342:94680 [0] NCCL INFO Channel 23/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-225:94342:94680 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] 1/-1/-1->0->-1 [7] 1/-1/-1->0->-1 [8] 1/-1/-1->0->-1 [9] 1/-1/-1->0->-1 [10] 1/-1/-1->0->-1 [11] 1/-1/-1->0->-1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] 1/-1/-1->0->-1 [19] 1/-1/-1->0->-1 [20] 1/-1/-1->0->-1 [21] 1/-1/-1->0->-1 [22] 1/-1/-1->0->-1 [23] 1/-1/-1->0->-1
+ip-26-0-160-225:94342:94680 [0] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-172-252:688206:688539 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-172-252:688206:688539 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-172-252:688201:688542 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-172-252:688201:688542 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-172-252:688204:688540 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-172-252:688204:688540 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-172-252:688204:688540 [4] NCCL INFO Trees [0] 5/-1/-1->4->3 [1] 5/-1/-1->4->3 [2] 5/-1/-1->4->3 [3] 5/-1/-1->4->3 [4] 5/-1/-1->4->3 [5] 5/-1/-1->4->3 [6] 5/-1/-1->4->3 [7] 5/-1/-1->4->3 [8] 5/-1/-1->4->3 [9] 5/-1/-1->4->3 [10] 5/-1/-1->4->3 [11] 5/-1/-1->4->3 [12] 5/-1/-1->4->3 [13] 5/-1/-1->4->3 [14] 5/-1/-1->4->3 [15] 5/-1/-1->4->3 [16] 5/-1/-1->4->3 [17] 5/-1/-1->4->3 [18] 5/-1/-1->4->3 [19] 5/-1/-1->4->3 [20] 5/-1/-1->4->3 [21] 5/-1/-1->4->3 [22] 5/-1/-1->4->3 [23] 5/-1/-1->4->3
+ip-26-0-172-252:688204:688540 [4] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-172-252:688206:688539 [6] NCCL INFO Trees [0] 7/-1/-1->6->5 [1] 7/-1/-1->6->5 [2] 7/-1/-1->6->5 [3] 7/-1/-1->6->5 [4] 7/-1/-1->6->5 [5] 7/-1/-1->6->5 [6] 7/-1/-1->6->5 [7] 7/-1/-1->6->5 [8] 7/-1/-1->6->5 [9] 7/-1/-1->6->5 [10] 7/-1/-1->6->5 [11] 7/-1/-1->6->5 [12] 7/-1/-1->6->5 [13] 7/-1/-1->6->5 [14] 7/-1/-1->6->5 [15] 7/-1/-1->6->5 [16] 7/-1/-1->6->5 [17] 7/-1/-1->6->5 [18] 7/-1/-1->6->5 [19] 7/-1/-1->6->5 [20] 7/-1/-1->6->5 [21] 7/-1/-1->6->5 [22] 7/-1/-1->6->5 [23] 7/-1/-1->6->5
+ip-26-0-172-252:688206:688539 [6] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-172-252:688200:688538 [0] NCCL INFO Channel 00/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-252:688205:688545 [5] NCCL INFO Trees [0] 6/-1/-1->5->4 [1] 6/-1/-1->5->4 [2] 6/-1/-1->5->4 [3] 6/-1/-1->5->4 [4] 6/-1/-1->5->4 [5] 6/-1/-1->5->4 [6] 6/-1/-1->5->4 [7] 6/-1/-1->5->4 [8] 6/-1/-1->5->4 [9] 6/-1/-1->5->4 [10] 6/-1/-1->5->4 [11] 6/-1/-1->5->4 [12] 6/-1/-1->5->4 [13] 6/-1/-1->5->4 [14] 6/-1/-1->5->4 [15] 6/-1/-1->5->4 [16] 6/-1/-1->5->4 [17] 6/-1/-1->5->4 [18] 6/-1/-1->5->4 [19] 6/-1/-1->5->4 [20] 6/-1/-1->5->4 [21] 6/-1/-1->5->4 [22] 6/-1/-1->5->4 [23] 6/-1/-1->5->4
+ip-26-0-172-252:688200:688538 [0] NCCL INFO Channel 01/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-252:688200:688538 [0] NCCL INFO Channel 02/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-252:688207:688543 [7] NCCL INFO Trees [0] -1/-1/-1->7->6 [1] -1/-1/-1->7->6 [2] -1/-1/-1->7->6 [3] -1/-1/-1->7->6 [4] -1/-1/-1->7->6 [5] -1/-1/-1->7->6 [6] -1/-1/-1->7->6 [7] -1/-1/-1->7->6 [8] -1/-1/-1->7->6 [9] -1/-1/-1->7->6 [10] -1/-1/-1->7->6 [11] -1/-1/-1->7->6 [12] -1/-1/-1->7->6 [13] -1/-1/-1->7->6 [14] -1/-1/-1->7->6 [15] -1/-1/-1->7->6 [16] -1/-1/-1->7->6 [17] -1/-1/-1->7->6 [18] -1/-1/-1->7->6 [19] -1/-1/-1->7->6 [20] -1/-1/-1->7->6 [21] -1/-1/-1->7->6 [22] -1/-1/-1->7->6 [23] -1/-1/-1->7->6
+ip-26-0-172-252:688205:688545 [5] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-172-252:688200:688538 [0] NCCL INFO Channel 03/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-252:688201:688542 [1] NCCL INFO Trees [0] 2/-1/-1->1->0 [1] 2/-1/-1->1->0 [2] 2/-1/-1->1->0 [3] 2/-1/-1->1->0 [4] 2/-1/-1->1->0 [5] 2/-1/-1->1->0 [6] 2/-1/-1->1->0 [7] 2/-1/-1->1->0 [8] 2/-1/-1->1->0 [9] 2/-1/-1->1->0 [10] 2/-1/-1->1->0 [11] 2/-1/-1->1->0 [12] 2/-1/-1->1->0 [13] 2/-1/-1->1->0 [14] 2/-1/-1->1->0 [15] 2/-1/-1->1->0 [16] 2/-1/-1->1->0 [17] 2/-1/-1->1->0 [18] 2/-1/-1->1->0 [19] 2/-1/-1->1->0 [20] 2/-1/-1->1->0 [21] 2/-1/-1->1->0 [22] 2/-1/-1->1->0 [23] 2/-1/-1->1->0
+ip-26-0-172-252:688207:688543 [7] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-172-252:688200:688538 [0] NCCL INFO Channel 04/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-252:688201:688542 [1] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-172-252:688200:688538 [0] NCCL INFO Channel 05/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-252:688200:688538 [0] NCCL INFO Channel 06/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-252:688200:688538 [0] NCCL INFO Channel 07/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-252:688203:688541 [3] NCCL INFO Trees [0] 4/-1/-1->3->2 [1] 4/-1/-1->3->2 [2] 4/-1/-1->3->2 [3] 4/-1/-1->3->2 [4] 4/-1/-1->3->2 [5] 4/-1/-1->3->2 [6] 4/-1/-1->3->2 [7] 4/-1/-1->3->2 [8] 4/-1/-1->3->2 [9] 4/-1/-1->3->2 [10] 4/-1/-1->3->2 [11] 4/-1/-1->3->2 [12] 4/-1/-1->3->2 [13] 4/-1/-1->3->2 [14] 4/-1/-1->3->2 [15] 4/-1/-1->3->2 [16] 4/-1/-1->3->2 [17] 4/-1/-1->3->2 [18] 4/-1/-1->3->2 [19] 4/-1/-1->3->2 [20] 4/-1/-1->3->2 [21] 4/-1/-1->3->2 [22] 4/-1/-1->3->2 [23] 4/-1/-1->3->2
+ip-26-0-172-252:688200:688538 [0] NCCL INFO Channel 08/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-252:688200:688538 [0] NCCL INFO Channel 09/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-252:688200:688538 [0] NCCL INFO Channel 10/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-252:688203:688541 [3] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-172-252:688200:688538 [0] NCCL INFO Channel 11/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-252:688202:688544 [2] NCCL INFO Trees [0] 3/-1/-1->2->1 [1] 3/-1/-1->2->1 [2] 3/-1/-1->2->1 [3] 3/-1/-1->2->1 [4] 3/-1/-1->2->1 [5] 3/-1/-1->2->1 [6] 3/-1/-1->2->1 [7] 3/-1/-1->2->1 [8] 3/-1/-1->2->1 [9] 3/-1/-1->2->1 [10] 3/-1/-1->2->1 [11] 3/-1/-1->2->1 [12] 3/-1/-1->2->1 [13] 3/-1/-1->2->1 [14] 3/-1/-1->2->1 [15] 3/-1/-1->2->1 [16] 3/-1/-1->2->1 [17] 3/-1/-1->2->1 [18] 3/-1/-1->2->1 [19] 3/-1/-1->2->1 [20] 3/-1/-1->2->1 [21] 3/-1/-1->2->1 [22] 3/-1/-1->2->1 [23] 3/-1/-1->2->1
+ip-26-0-172-252:688200:688538 [0] NCCL INFO Channel 12/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-252:688200:688538 [0] NCCL INFO Channel 13/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-252:688202:688544 [2] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-172-252:688200:688538 [0] NCCL INFO Channel 14/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-252:688200:688538 [0] NCCL INFO Channel 15/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-252:688200:688538 [0] NCCL INFO Channel 16/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-252:688200:688538 [0] NCCL INFO Channel 17/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-252:688200:688538 [0] NCCL INFO Channel 18/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-252:688200:688538 [0] NCCL INFO Channel 19/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-252:688200:688538 [0] NCCL INFO Channel 20/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-252:688200:688538 [0] NCCL INFO Channel 21/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-252:688200:688538 [0] NCCL INFO Channel 22/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-252:688200:688538 [0] NCCL INFO Channel 23/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-252:688200:688538 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] 1/-1/-1->0->-1 [7] 1/-1/-1->0->-1 [8] 1/-1/-1->0->-1 [9] 1/-1/-1->0->-1 [10] 1/-1/-1->0->-1 [11] 1/-1/-1->0->-1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] 1/-1/-1->0->-1 [19] 1/-1/-1->0->-1 [20] 1/-1/-1->0->-1 [21] 1/-1/-1->0->-1 [22] 1/-1/-1->0->-1 [23] 1/-1/-1->0->-1
+ip-26-0-172-252:688200:688538 [0] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-172-147:618590:618925 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-172-147:618590:618925 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-172-142:880685:881018 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-172-142:880685:881018 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-172-116:1019294:1019626 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-172-116:1019294:1019626 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-160-225:94348:94684 [6] NCCL INFO Channel 00/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-225:94348:94684 [6] NCCL INFO Channel 01/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-225:94345:94687 [3] NCCL INFO Channel 00/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-225:94348:94684 [6] NCCL INFO Channel 02/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-225:94346:94682 [4] NCCL INFO Channel 00/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-225:94342:94680 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-225:94345:94687 [3] NCCL INFO Channel 01/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-225:94348:94684 [6] NCCL INFO Channel 03/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-225:94346:94682 [4] NCCL INFO Channel 01/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-225:94342:94680 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-116:1019290:1019628 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-172-116:1019290:1019628 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-172-116:1019290:1019628 [2] NCCL INFO Trees [0] 3/-1/-1->2->1 [1] 3/-1/-1->2->1 [2] 3/-1/-1->2->1 [3] 3/-1/-1->2->1 [4] 3/-1/-1->2->1 [5] 3/-1/-1->2->1 [6] 3/-1/-1->2->1 [7] 3/-1/-1->2->1 [8] 3/-1/-1->2->1 [9] 3/-1/-1->2->1 [10] 3/-1/-1->2->1 [11] 3/-1/-1->2->1 [12] 3/-1/-1->2->1 [13] 3/-1/-1->2->1 [14] 3/-1/-1->2->1 [15] 3/-1/-1->2->1 [16] 3/-1/-1->2->1 [17] 3/-1/-1->2->1 [18] 3/-1/-1->2->1 [19] 3/-1/-1->2->1 [20] 3/-1/-1->2->1 [21] 3/-1/-1->2->1 [22] 3/-1/-1->2->1 [23] 3/-1/-1->2->1
+ip-26-0-160-225:94345:94687 [3] NCCL INFO Channel 02/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-116:1019290:1019628 [2] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-172-116:1019288:1019623 [0] NCCL INFO Channel 00/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-116:1019288:1019623 [0] NCCL INFO Channel 01/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-116:1019288:1019623 [0] NCCL INFO Channel 02/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-116:1019288:1019623 [0] NCCL INFO Channel 03/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-116:1019295:1019627 [7] NCCL INFO Trees [0] -1/-1/-1->7->6 [1] -1/-1/-1->7->6 [2] -1/-1/-1->7->6 [3] -1/-1/-1->7->6 [4] -1/-1/-1->7->6 [5] -1/-1/-1->7->6 [6] -1/-1/-1->7->6 [7] -1/-1/-1->7->6 [8] -1/-1/-1->7->6 [9] -1/-1/-1->7->6 [10] -1/-1/-1->7->6 [11] -1/-1/-1->7->6 [12] -1/-1/-1->7->6 [13] -1/-1/-1->7->6 [14] -1/-1/-1->7->6 [15] -1/-1/-1->7->6 [16] -1/-1/-1->7->6 [17] -1/-1/-1->7->6 [18] -1/-1/-1->7->6 [19] -1/-1/-1->7->6 [20] -1/-1/-1->7->6 [21] -1/-1/-1->7->6 [22] -1/-1/-1->7->6 [23] -1/-1/-1->7->6
+ip-26-0-172-116:1019288:1019623 [0] NCCL INFO Channel 04/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-116:1019288:1019623 [0] NCCL INFO Channel 05/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-116:1019288:1019623 [0] NCCL INFO Channel 06/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-116:1019294:1019626 [6] NCCL INFO Trees [0] 7/-1/-1->6->5 [1] 7/-1/-1->6->5 [2] 7/-1/-1->6->5 [3] 7/-1/-1->6->5 [4] 7/-1/-1->6->5 [5] 7/-1/-1->6->5 [6] 7/-1/-1->6->5 [7] 7/-1/-1->6->5 [8] 7/-1/-1->6->5 [9] 7/-1/-1->6->5 [10] 7/-1/-1->6->5 [11] 7/-1/-1->6->5 [12] 7/-1/-1->6->5 [13] 7/-1/-1->6->5 [14] 7/-1/-1->6->5 [15] 7/-1/-1->6->5 [16] 7/-1/-1->6->5 [17] 7/-1/-1->6->5 [18] 7/-1/-1->6->5 [19] 7/-1/-1->6->5 [20] 7/-1/-1->6->5 [21] 7/-1/-1->6->5 [22] 7/-1/-1->6->5 [23] 7/-1/-1->6->5
+ip-26-0-172-116:1019288:1019623 [0] NCCL INFO Channel 07/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-116:1019293:1019625 [5] NCCL INFO Trees [0] 6/-1/-1->5->4 [1] 6/-1/-1->5->4 [2] 6/-1/-1->5->4 [3] 6/-1/-1->5->4 [4] 6/-1/-1->5->4 [5] 6/-1/-1->5->4 [6] 6/-1/-1->5->4 [7] 6/-1/-1->5->4 [8] 6/-1/-1->5->4 [9] 6/-1/-1->5->4 [10] 6/-1/-1->5->4 [11] 6/-1/-1->5->4 [12] 6/-1/-1->5->4 [13] 6/-1/-1->5->4 [14] 6/-1/-1->5->4 [15] 6/-1/-1->5->4 [16] 6/-1/-1->5->4 [17] 6/-1/-1->5->4 [18] 6/-1/-1->5->4 [19] 6/-1/-1->5->4 [20] 6/-1/-1->5->4 [21] 6/-1/-1->5->4 [22] 6/-1/-1->5->4 [23] 6/-1/-1->5->4
+ip-26-0-172-116:1019295:1019627 [7] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-172-116:1019294:1019626 [6] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-172-116:1019288:1019623 [0] NCCL INFO Channel 08/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-116:1019292:1019624 [4] NCCL INFO Trees [0] 5/-1/-1->4->3 [1] 5/-1/-1->4->3 [2] 5/-1/-1->4->3 [3] 5/-1/-1->4->3 [4] 5/-1/-1->4->3 [5] 5/-1/-1->4->3 [6] 5/-1/-1->4->3 [7] 5/-1/-1->4->3 [8] 5/-1/-1->4->3 [9] 5/-1/-1->4->3 [10] 5/-1/-1->4->3 [11] 5/-1/-1->4->3 [12] 5/-1/-1->4->3 [13] 5/-1/-1->4->3 [14] 5/-1/-1->4->3 [15] 5/-1/-1->4->3 [16] 5/-1/-1->4->3 [17] 5/-1/-1->4->3 [18] 5/-1/-1->4->3 [19] 5/-1/-1->4->3 [20] 5/-1/-1->4->3 [21] 5/-1/-1->4->3 [22] 5/-1/-1->4->3 [23] 5/-1/-1->4->3
+ip-26-0-172-116:1019293:1019625 [5] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-172-116:1019288:1019623 [0] NCCL INFO Channel 09/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-116:1019288:1019623 [0] NCCL INFO Channel 10/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-116:1019292:1019624 [4] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-172-116:1019291:1019630 [3] NCCL INFO Trees [0] 4/-1/-1->3->2 [1] 4/-1/-1->3->2 [2] 4/-1/-1->3->2 [3] 4/-1/-1->3->2 [4] 4/-1/-1->3->2 [5] 4/-1/-1->3->2 [6] 4/-1/-1->3->2 [7] 4/-1/-1->3->2 [8] 4/-1/-1->3->2 [9] 4/-1/-1->3->2 [10] 4/-1/-1->3->2 [11] 4/-1/-1->3->2 [12] 4/-1/-1->3->2 [13] 4/-1/-1->3->2 [14] 4/-1/-1->3->2 [15] 4/-1/-1->3->2 [16] 4/-1/-1->3->2 [17] 4/-1/-1->3->2 [18] 4/-1/-1->3->2 [19] 4/-1/-1->3->2 [20] 4/-1/-1->3->2 [21] 4/-1/-1->3->2 [22] 4/-1/-1->3->2 [23] 4/-1/-1->3->2
+ip-26-0-172-116:1019288:1019623 [0] NCCL INFO Channel 11/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-116:1019288:1019623 [0] NCCL INFO Channel 12/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-116:1019288:1019623 [0] NCCL INFO Channel 13/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-116:1019288:1019623 [0] NCCL INFO Channel 14/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-116:1019288:1019623 [0] NCCL INFO Channel 15/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-116:1019288:1019623 [0] NCCL INFO Channel 16/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-116:1019288:1019623 [0] NCCL INFO Channel 17/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-116:1019288:1019623 [0] NCCL INFO Channel 18/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-116:1019288:1019623 [0] NCCL INFO Channel 19/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-116:1019288:1019623 [0] NCCL INFO Channel 20/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-116:1019288:1019623 [0] NCCL INFO Channel 21/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-116:1019288:1019623 [0] NCCL INFO Channel 22/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-116:1019288:1019623 [0] NCCL INFO Channel 23/24 :    0   1   2   3   4   5   6   7
+ip-26-0-160-225:94348:94684 [6] NCCL INFO Channel 04/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-116:1019288:1019623 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] 1/-1/-1->0->-1 [7] 1/-1/-1->0->-1 [8] 1/-1/-1->0->-1 [9] 1/-1/-1->0->-1 [10] 1/-1/-1->0->-1 [11] 1/-1/-1->0->-1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] 1/-1/-1->0->-1 [19] 1/-1/-1->0->-1 [20] 1/-1/-1->0->-1 [21] 1/-1/-1->0->-1 [22] 1/-1/-1->0->-1 [23] 1/-1/-1->0->-1
+ip-26-0-172-116:1019289:1019629 [1] NCCL INFO Trees [0] 2/-1/-1->1->0 [1] 2/-1/-1->1->0 [2] 2/-1/-1->1->0 [3] 2/-1/-1->1->0 [4] 2/-1/-1->1->0 [5] 2/-1/-1->1->0 [6] 2/-1/-1->1->0 [7] 2/-1/-1->1->0 [8] 2/-1/-1->1->0 [9] 2/-1/-1->1->0 [10] 2/-1/-1->1->0 [11] 2/-1/-1->1->0 [12] 2/-1/-1->1->0 [13] 2/-1/-1->1->0 [14] 2/-1/-1->1->0 [15] 2/-1/-1->1->0 [16] 2/-1/-1->1->0 [17] 2/-1/-1->1->0 [18] 2/-1/-1->1->0 [19] 2/-1/-1->1->0 [20] 2/-1/-1->1->0 [21] 2/-1/-1->1->0 [22] 2/-1/-1->1->0 [23] 2/-1/-1->1->0
+ip-26-0-172-116:1019288:1019623 [0] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-172-116:1019291:1019630 [3] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-160-225:94346:94682 [4] NCCL INFO Channel 02/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-225:94342:94680 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-225:94345:94687 [3] NCCL INFO Channel 03/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-225:94348:94684 [6] NCCL INFO Channel 05/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-225:94346:94682 [4] NCCL INFO Channel 03/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-225:94342:94680 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-225:94345:94687 [3] NCCL INFO Channel 04/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-225:94348:94684 [6] NCCL INFO Channel 06/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-252:688200:688538 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-225:94346:94682 [4] NCCL INFO Channel 04/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-225:94342:94680 [0] NCCL INFO Channel 04/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-252:688200:688538 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-225:94345:94687 [3] NCCL INFO Channel 05/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-225:94348:94684 [6] NCCL INFO Channel 07/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-225:94346:94682 [4] NCCL INFO Channel 05/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-252:688200:688538 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-225:94342:94680 [0] NCCL INFO Channel 05/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-252:688204:688540 [4] NCCL INFO Channel 00/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-225:94345:94687 [3] NCCL INFO Channel 06/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-225:94348:94684 [6] NCCL INFO Channel 08/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-252:688200:688538 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-225:94346:94682 [4] NCCL INFO Channel 06/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-252:688204:688540 [4] NCCL INFO Channel 01/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-225:94342:94680 [0] NCCL INFO Channel 06/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-252:688200:688538 [0] NCCL INFO Channel 04/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-225:94345:94687 [3] NCCL INFO Channel 07/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-225:94348:94684 [6] NCCL INFO Channel 09/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-252:688204:688540 [4] NCCL INFO Channel 02/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-252:688200:688538 [0] NCCL INFO Channel 05/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-225:94346:94682 [4] NCCL INFO Channel 07/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-225:94342:94680 [0] NCCL INFO Channel 07/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-252:688204:688540 [4] NCCL INFO Channel 03/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-252:688200:688538 [0] NCCL INFO Channel 06/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-142:880680:881022 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-172-142:880680:881022 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-160-225:94345:94687 [3] NCCL INFO Channel 08/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-225:94348:94684 [6] NCCL INFO Channel 10/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-225:94346:94682 [4] NCCL INFO Channel 08/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-225:94342:94680 [0] NCCL INFO Channel 08/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-252:688201:688542 [1] NCCL INFO Channel 00/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-225:94349:94683 [7] NCCL INFO Channel 00/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-225:94343:94681 [1] NCCL INFO Channel 00/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-252:688204:688540 [4] NCCL INFO Channel 04/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-252:688200:688538 [0] NCCL INFO Channel 07/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-225:94347:94685 [5] NCCL INFO Channel 00/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-252:688206:688539 [6] NCCL INFO Channel 00/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-225:94345:94687 [3] NCCL INFO Channel 09/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-225:94348:94684 [6] NCCL INFO Channel 11/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-225:94346:94682 [4] NCCL INFO Channel 09/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-225:94342:94680 [0] NCCL INFO Channel 09/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-252:688201:688542 [1] NCCL INFO Channel 01/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-252:688204:688540 [4] NCCL INFO Channel 05/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-252:688200:688538 [0] NCCL INFO Channel 08/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-252:688206:688539 [6] NCCL INFO Channel 01/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-225:94343:94681 [1] NCCL INFO Channel 01/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-225:94349:94683 [7] NCCL INFO Channel 01/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-225:94347:94685 [5] NCCL INFO Channel 01/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-225:94345:94687 [3] NCCL INFO Channel 10/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-225:94348:94684 [6] NCCL INFO Channel 12/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-225:94344:94686 [2] NCCL INFO Channel 00/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-225:94346:94682 [4] NCCL INFO Channel 10/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-252:688201:688542 [1] NCCL INFO Channel 02/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-252:688204:688540 [4] NCCL INFO Channel 06/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-252:688200:688538 [0] NCCL INFO Channel 09/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-252:688206:688539 [6] NCCL INFO Channel 02/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-225:94342:94680 [0] NCCL INFO Channel 10/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-225:94343:94681 [1] NCCL INFO Channel 02/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-225:94349:94683 [7] NCCL INFO Channel 02/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-225:94347:94685 [5] NCCL INFO Channel 02/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-225:94345:94687 [3] NCCL INFO Channel 11/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-225:94348:94684 [6] NCCL INFO Channel 13/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-225:94344:94686 [2] NCCL INFO Channel 01/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-225:94346:94682 [4] NCCL INFO Channel 11/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-252:688201:688542 [1] NCCL INFO Channel 03/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-252:688204:688540 [4] NCCL INFO Channel 07/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-252:688200:688538 [0] NCCL INFO Channel 10/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-252:688206:688539 [6] NCCL INFO Channel 03/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-225:94342:94680 [0] NCCL INFO Channel 11/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-225:94343:94681 [1] NCCL INFO Channel 03/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-225:94349:94683 [7] NCCL INFO Channel 03/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-225:94347:94685 [5] NCCL INFO Channel 03/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-225:94345:94687 [3] NCCL INFO Channel 12/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-225:94348:94684 [6] NCCL INFO Channel 14/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-225:94344:94686 [2] NCCL INFO Channel 02/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-252:688201:688542 [1] NCCL INFO Channel 04/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-252:688204:688540 [4] NCCL INFO Channel 08/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-225:94346:94682 [4] NCCL INFO Channel 12/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-252:688200:688538 [0] NCCL INFO Channel 11/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-252:688206:688539 [6] NCCL INFO Channel 04/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-225:94343:94681 [1] NCCL INFO Channel 04/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-225:94342:94680 [0] NCCL INFO Channel 12/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-225:94349:94683 [7] NCCL INFO Channel 04/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-225:94347:94685 [5] NCCL INFO Channel 04/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-225:94345:94687 [3] NCCL INFO Channel 13/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-225:94346:94682 [4] NCCL INFO Channel 13/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-225:94343:94681 [1] NCCL INFO Channel 05/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-252:688204:688540 [4] NCCL INFO Channel 09/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-252:688201:688542 [1] NCCL INFO Channel 05/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-252:688200:688538 [0] NCCL INFO Channel 12/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-252:688206:688539 [6] NCCL INFO Channel 05/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-225:94345:94687 [3] NCCL INFO Channel 14/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-225:94342:94680 [0] NCCL INFO Channel 13/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-225:94346:94682 [4] NCCL INFO Channel 14/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-225:94343:94681 [1] NCCL INFO Channel 06/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-225:94344:94686 [2] NCCL INFO Channel 03/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-225:94345:94687 [3] NCCL INFO Channel 15/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-142:880679:881015 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-172-252:688204:688540 [4] NCCL INFO Channel 10/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-142:880679:881015 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-172-252:688206:688539 [6] NCCL INFO Channel 06/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-252:688201:688542 [1] NCCL INFO Channel 06/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-252:688200:688538 [0] NCCL INFO Channel 13/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-225:94342:94680 [0] NCCL INFO Channel 14/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-225:94346:94682 [4] NCCL INFO Channel 15/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-225:94343:94681 [1] NCCL INFO Channel 07/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-225:94344:94686 [2] NCCL INFO Channel 04/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-225:94345:94687 [3] NCCL INFO Channel 16/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-252:688203:688541 [3] NCCL INFO Channel 00/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-225:94342:94680 [0] NCCL INFO Channel 15/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-225:94346:94682 [4] NCCL INFO Channel 16/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-225:94343:94681 [1] NCCL INFO Channel 08/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-225:94344:94686 [2] NCCL INFO Channel 05/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-225:94345:94687 [3] NCCL INFO Channel 17/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-252:688204:688540 [4] NCCL INFO Channel 11/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-252:688200:688538 [0] NCCL INFO Channel 14/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-252:688206:688539 [6] NCCL INFO Channel 07/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-252:688201:688542 [1] NCCL INFO Channel 07/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-252:688203:688541 [3] NCCL INFO Channel 01/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-225:94349:94683 [7] NCCL INFO Channel 05/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-225:94342:94680 [0] NCCL INFO Channel 16/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-225:94346:94682 [4] NCCL INFO Channel 17/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-225:94343:94681 [1] NCCL INFO Channel 09/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-225:94344:94686 [2] NCCL INFO Channel 06/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-225:94345:94687 [3] NCCL INFO Channel 18/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-252:688204:688540 [4] NCCL INFO Channel 12/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-252:688200:688538 [0] NCCL INFO Channel 15/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-252:688206:688539 [6] NCCL INFO Channel 08/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-252:688201:688542 [1] NCCL INFO Channel 08/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-225:94349:94683 [7] NCCL INFO Channel 06/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-225:94342:94680 [0] NCCL INFO Channel 17/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-252:688203:688541 [3] NCCL INFO Channel 02/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-225:94346:94682 [4] NCCL INFO Channel 18/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-225:94343:94681 [1] NCCL INFO Channel 10/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-225:94344:94686 [2] NCCL INFO Channel 07/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-225:94345:94687 [3] NCCL INFO Channel 19/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-252:688202:688544 [2] NCCL INFO Channel 00/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-225:94349:94683 [7] NCCL INFO Channel 07/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-252:688204:688540 [4] NCCL INFO Channel 13/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-225:94346:94682 [4] NCCL INFO Channel 19/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-225:94342:94680 [0] NCCL INFO Channel 18/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-252:688200:688538 [0] NCCL INFO Channel 16/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-252:688206:688539 [6] NCCL INFO Channel 09/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-252:688201:688542 [1] NCCL INFO Channel 09/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-225:94343:94681 [1] NCCL INFO Channel 11/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-225:94344:94686 [2] NCCL INFO Channel 08/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-225:94345:94687 [3] NCCL INFO Channel 20/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-252:688203:688541 [3] NCCL INFO Channel 03/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-225:94348:94684 [6] NCCL INFO Channel 15/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-252:688202:688544 [2] NCCL INFO Channel 01/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-225:94349:94683 [7] NCCL INFO Channel 08/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-252:688204:688540 [4] NCCL INFO Channel 14/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-225:94346:94682 [4] NCCL INFO Channel 20/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-252:688200:688538 [0] NCCL INFO Channel 17/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-252:688206:688539 [6] NCCL INFO Channel 10/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-225:94342:94680 [0] NCCL INFO Channel 19/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-225:94343:94681 [1] NCCL INFO Channel 12/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-252:688201:688542 [1] NCCL INFO Channel 10/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-225:94347:94685 [5] NCCL INFO Channel 05/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-225:94344:94686 [2] NCCL INFO Channel 09/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-252:688203:688541 [3] NCCL INFO Channel 04/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-225:94345:94687 [3] NCCL INFO Channel 21/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-225:94348:94684 [6] NCCL INFO Channel 16/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-252:688202:688544 [2] NCCL INFO Channel 02/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-252:688207:688543 [7] NCCL INFO Channel 00/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-225:94349:94683 [7] NCCL INFO Channel 09/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-252:688204:688540 [4] NCCL INFO Channel 15/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-252:688201:688542 [1] NCCL INFO Channel 11/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-225:94346:94682 [4] NCCL INFO Channel 21/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-252:688200:688538 [0] NCCL INFO Channel 18/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-252:688206:688539 [6] NCCL INFO Channel 11/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-225:94342:94680 [0] NCCL INFO Channel 20/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-225:94343:94681 [1] NCCL INFO Channel 13/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-225:94344:94686 [2] NCCL INFO Channel 10/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-225:94347:94685 [5] NCCL INFO Channel 06/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-225:94345:94687 [3] NCCL INFO Channel 22/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-252:688203:688541 [3] NCCL INFO Channel 05/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-252:688202:688544 [2] NCCL INFO Channel 03/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-225:94348:94684 [6] NCCL INFO Channel 17/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-252:688207:688543 [7] NCCL INFO Channel 01/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-225:94349:94683 [7] NCCL INFO Channel 10/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-225:94346:94682 [4] NCCL INFO Channel 22/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-252:688204:688540 [4] NCCL INFO Channel 16/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-252:688201:688542 [1] NCCL INFO Channel 12/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-252:688200:688538 [0] NCCL INFO Channel 19/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-252:688206:688539 [6] NCCL INFO Channel 12/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-225:94343:94681 [1] NCCL INFO Channel 14/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-225:94342:94680 [0] NCCL INFO Channel 21/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-225:94344:94686 [2] NCCL INFO Channel 11/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-225:94345:94687 [3] NCCL INFO Channel 23/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-252:688203:688541 [3] NCCL INFO Channel 06/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-225:94347:94685 [5] NCCL INFO Channel 07/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-252:688205:688545 [5] NCCL INFO Channel 00/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-225:94348:94684 [6] NCCL INFO Channel 18/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-252:688202:688544 [2] NCCL INFO Channel 04/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-252:688207:688543 [7] NCCL INFO Channel 02/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-225:94349:94683 [7] NCCL INFO Channel 11/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-225:94346:94682 [4] NCCL INFO Channel 23/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-252:688204:688540 [4] NCCL INFO Channel 17/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-252:688201:688542 [1] NCCL INFO Channel 13/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-225:94347:94685 [5] NCCL INFO Channel 08/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-225:94343:94681 [1] NCCL INFO Channel 15/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-225:94344:94686 [2] NCCL INFO Channel 12/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-252:688200:688538 [0] NCCL INFO Channel 20/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-225:94342:94680 [0] NCCL INFO Channel 22/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-252:688206:688539 [6] NCCL INFO Channel 13/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-252:688203:688541 [3] NCCL INFO Channel 07/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-252:688205:688545 [5] NCCL INFO Channel 01/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-225:94348:94684 [6] NCCL INFO Channel 19/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-252:688202:688544 [2] NCCL INFO Channel 05/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-252:688207:688543 [7] NCCL INFO Channel 03/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-225:94349:94683 [7] NCCL INFO Channel 12/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-225:94347:94685 [5] NCCL INFO Channel 09/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-225:94343:94681 [1] NCCL INFO Channel 16/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-225:94344:94686 [2] NCCL INFO Channel 13/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-225:94342:94680 [0] NCCL INFO Channel 23/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-73:926596:926936 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-172-252:688201:688542 [1] NCCL INFO Channel 14/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-252:688204:688540 [4] NCCL INFO Channel 18/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-73:926596:926936 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-172-252:688200:688538 [0] NCCL INFO Channel 21/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-252:688206:688539 [6] NCCL INFO Channel 14/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-252:688203:688541 [3] NCCL INFO Channel 08/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-252:688205:688545 [5] NCCL INFO Channel 02/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-252:688202:688544 [2] NCCL INFO Channel 06/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-252:688207:688543 [7] NCCL INFO Channel 04/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-225:94349:94683 [7] NCCL INFO Channel 13/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-225:94348:94684 [6] NCCL INFO Channel 20/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-225:94343:94681 [1] NCCL INFO Channel 17/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-252:688200:688538 [0] NCCL INFO Channel 22/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-160-225:94344:94686 [2] NCCL INFO Channel 14/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-252:688207:688543 [7] NCCL INFO Channel 05/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-225:94349:94683 [7] NCCL INFO Channel 14/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-225:94348:94684 [6] NCCL INFO Channel 21/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-225:94343:94681 [1] NCCL INFO Channel 18/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-252:688201:688542 [1] NCCL INFO Channel 15/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-252:688204:688540 [4] NCCL INFO Channel 19/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-252:688207:688543 [7] NCCL INFO Channel 06/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-252:688202:688544 [2] NCCL INFO Channel 07/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-225:94347:94685 [5] NCCL INFO Channel 10/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-225:94344:94686 [2] NCCL INFO Channel 15/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-225:94349:94683 [7] NCCL INFO Channel 15/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-225:94343:94681 [1] NCCL INFO Channel 19/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-225:94348:94684 [6] NCCL INFO Channel 22/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-252:688206:688539 [6] NCCL INFO Channel 15/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-252:688203:688541 [3] NCCL INFO Channel 09/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-252:688201:688542 [1] NCCL INFO Channel 16/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-252:688206:688539 [6] NCCL INFO Channel 16/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-252:688200:688538 [0] NCCL INFO Channel 23/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-252:688205:688545 [5] NCCL INFO Channel 03/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-252:688203:688541 [3] NCCL INFO Channel 10/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-160-225:94347:94685 [5] NCCL INFO Channel 11/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-252:688201:688542 [1] NCCL INFO Channel 17/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-225:94349:94683 [7] NCCL INFO Channel 16/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-225:94344:94686 [2] NCCL INFO Channel 16/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-252:688204:688540 [4] NCCL INFO Channel 20/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-225:94343:94681 [1] NCCL INFO Channel 20/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-225:94347:94685 [5] NCCL INFO Channel 12/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-225:94349:94683 [7] NCCL INFO Channel 17/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-252:688206:688539 [6] NCCL INFO Channel 17/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-160-225:94344:94686 [2] NCCL INFO Channel 17/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-252:688203:688541 [3] NCCL INFO Channel 11/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-252:688201:688542 [1] NCCL INFO Channel 18/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-252:688205:688545 [5] NCCL INFO Channel 04/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-225:94343:94681 [1] NCCL INFO Channel 21/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-225:94349:94683 [7] NCCL INFO Channel 18/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-225:94347:94685 [5] NCCL INFO Channel 13/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-252:688204:688540 [4] NCCL INFO Channel 21/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-225:94344:94686 [2] NCCL INFO Channel 18/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-225:94343:94681 [1] NCCL INFO Channel 22/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-252:688202:688544 [2] NCCL INFO Channel 08/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-252:688206:688539 [6] NCCL INFO Channel 18/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-252:688203:688541 [3] NCCL INFO Channel 12/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-252:688201:688542 [1] NCCL INFO Channel 19/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-252:688205:688545 [5] NCCL INFO Channel 05/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-225:94349:94683 [7] NCCL INFO Channel 19/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-225:94347:94685 [5] NCCL INFO Channel 14/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-225:94344:94686 [2] NCCL INFO Channel 19/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-173-7:811248:811591 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-173-7:811248:811591 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-160-225:94343:94681 [1] NCCL INFO Channel 23/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-252:688204:688540 [4] NCCL INFO Channel 22/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-160-225:94349:94683 [7] NCCL INFO Channel 20/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-252:688207:688543 [7] NCCL INFO Channel 07/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-225:94347:94685 [5] NCCL INFO Channel 15/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-225:94344:94686 [2] NCCL INFO Channel 20/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-252:688202:688544 [2] NCCL INFO Channel 09/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-252:688203:688541 [3] NCCL INFO Channel 13/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-252:688201:688542 [1] NCCL INFO Channel 20/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-252:688206:688539 [6] NCCL INFO Channel 19/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-252:688205:688545 [5] NCCL INFO Channel 06/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-225:94349:94683 [7] NCCL INFO Channel 21/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-225:94347:94685 [5] NCCL INFO Channel 16/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-225:94344:94686 [2] NCCL INFO Channel 21/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-225:94348:94684 [6] NCCL INFO Channel 23/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-252:688204:688540 [4] NCCL INFO Channel 23/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-252:688207:688543 [7] NCCL INFO Channel 08/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-225:94349:94683 [7] NCCL INFO Channel 22/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-225:94347:94685 [5] NCCL INFO Channel 17/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-225:94344:94686 [2] NCCL INFO Channel 22/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-252:688203:688541 [3] NCCL INFO Channel 14/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-252:688202:688544 [2] NCCL INFO Channel 10/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-252:688201:688542 [1] NCCL INFO Channel 21/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-252:688205:688545 [5] NCCL INFO Channel 07/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-225:94349:94683 [7] NCCL INFO Channel 23/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-225:94347:94685 [5] NCCL INFO Channel 18/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-225:94344:94686 [2] NCCL INFO Channel 23/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-252:688206:688539 [6] NCCL INFO Channel 20/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-252:688207:688543 [7] NCCL INFO Channel 09/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-160-225:94347:94685 [5] NCCL INFO Channel 19/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-252:688203:688541 [3] NCCL INFO Channel 15/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-252:688202:688544 [2] NCCL INFO Channel 11/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-252:688201:688542 [1] NCCL INFO Channel 22/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-252:688205:688545 [5] NCCL INFO Channel 08/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-252:688206:688539 [6] NCCL INFO Channel 21/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-252:688207:688543 [7] NCCL INFO Channel 10/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-252:688203:688541 [3] NCCL INFO Channel 16/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-252:688201:688542 [1] NCCL INFO Channel 23/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-252:688202:688544 [2] NCCL INFO Channel 12/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-252:688205:688545 [5] NCCL INFO Channel 09/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-160-225:94347:94685 [5] NCCL INFO Channel 20/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-252:688206:688539 [6] NCCL INFO Channel 22/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-252:688207:688543 [7] NCCL INFO Channel 11/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-116:1019289:1019629 [1] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-172-116:1019288:1019623 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-116:1019294:1019626 [6] NCCL INFO Channel 00/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-116:1019288:1019623 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-116:1019288:1019623 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-116:1019294:1019626 [6] NCCL INFO Channel 01/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-116:1019295:1019627 [7] NCCL INFO Channel 00/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-116:1019294:1019626 [6] NCCL INFO Channel 02/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-116:1019294:1019626 [6] NCCL INFO Channel 03/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-116:1019291:1019630 [3] NCCL INFO Channel 00/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-252:688203:688541 [3] NCCL INFO Channel 17/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-252:688202:688544 [2] NCCL INFO Channel 13/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-116:1019288:1019623 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-116:1019294:1019626 [6] NCCL INFO Channel 04/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-116:1019291:1019630 [3] NCCL INFO Channel 01/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-116:1019288:1019623 [0] NCCL INFO Channel 04/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-116:1019294:1019626 [6] NCCL INFO Channel 05/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-116:1019291:1019630 [3] NCCL INFO Channel 02/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-116:1019288:1019623 [0] NCCL INFO Channel 05/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-116:1019294:1019626 [6] NCCL INFO Channel 06/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-116:1019291:1019630 [3] NCCL INFO Channel 03/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-116:1019288:1019623 [0] NCCL INFO Channel 06/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-116:1019294:1019626 [6] NCCL INFO Channel 07/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-252:688205:688545 [5] NCCL INFO Channel 10/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-116:1019291:1019630 [3] NCCL INFO Channel 04/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-116:1019292:1019624 [4] NCCL INFO Channel 00/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-116:1019288:1019623 [0] NCCL INFO Channel 07/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-116:1019294:1019626 [6] NCCL INFO Channel 08/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-116:1019291:1019630 [3] NCCL INFO Channel 05/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-116:1019292:1019624 [4] NCCL INFO Channel 01/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-116:1019288:1019623 [0] NCCL INFO Channel 08/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-116:1019294:1019626 [6] NCCL INFO Channel 09/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-252:688206:688539 [6] NCCL INFO Channel 23/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-252:688207:688543 [7] NCCL INFO Channel 12/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-116:1019291:1019630 [3] NCCL INFO Channel 06/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-116:1019288:1019623 [0] NCCL INFO Channel 09/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-116:1019292:1019624 [4] NCCL INFO Channel 02/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-252:688203:688541 [3] NCCL INFO Channel 18/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-252:688202:688544 [2] NCCL INFO Channel 14/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-252:688205:688545 [5] NCCL INFO Channel 11/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-252:688207:688543 [7] NCCL INFO Channel 13/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-252:688203:688541 [3] NCCL INFO Channel 19/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-252:688202:688544 [2] NCCL INFO Channel 15/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-252:688205:688545 [5] NCCL INFO Channel 12/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-252:688207:688543 [7] NCCL INFO Channel 14/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-116:1019295:1019627 [7] NCCL INFO Channel 01/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-252:688203:688541 [3] NCCL INFO Channel 20/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-252:688202:688544 [2] NCCL INFO Channel 16/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-116:1019295:1019627 [7] NCCL INFO Channel 02/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-252:688205:688545 [5] NCCL INFO Channel 13/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-252:688207:688543 [7] NCCL INFO Channel 15/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-116:1019295:1019627 [7] NCCL INFO Channel 03/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-252:688203:688541 [3] NCCL INFO Channel 21/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-252:688202:688544 [2] NCCL INFO Channel 17/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-252:688205:688545 [5] NCCL INFO Channel 14/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-252:688207:688543 [7] NCCL INFO Channel 16/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-116:1019295:1019627 [7] NCCL INFO Channel 04/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-252:688203:688541 [3] NCCL INFO Channel 22/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-116:1019290:1019628 [2] NCCL INFO Channel 00/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-252:688202:688544 [2] NCCL INFO Channel 18/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-252:688205:688545 [5] NCCL INFO Channel 15/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-252:688207:688543 [7] NCCL INFO Channel 17/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-116:1019295:1019627 [7] NCCL INFO Channel 05/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-252:688203:688541 [3] NCCL INFO Channel 23/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-252:688202:688544 [2] NCCL INFO Channel 19/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-116:1019290:1019628 [2] NCCL INFO Channel 01/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-252:688205:688545 [5] NCCL INFO Channel 16/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-252:688207:688543 [7] NCCL INFO Channel 18/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-116:1019289:1019629 [1] NCCL INFO Channel 00/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-252:688202:688544 [2] NCCL INFO Channel 20/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-252:688207:688543 [7] NCCL INFO Channel 19/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-252:688205:688545 [5] NCCL INFO Channel 17/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-116:1019294:1019626 [6] NCCL INFO Channel 10/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-116:1019291:1019630 [3] NCCL INFO Channel 07/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-116:1019288:1019623 [0] NCCL INFO Channel 10/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-116:1019292:1019624 [4] NCCL INFO Channel 03/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-116:1019295:1019627 [7] NCCL INFO Channel 06/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-252:688202:688544 [2] NCCL INFO Channel 21/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-252:688207:688543 [7] NCCL INFO Channel 20/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-252:688205:688545 [5] NCCL INFO Channel 18/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-116:1019290:1019628 [2] NCCL INFO Channel 02/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-116:1019289:1019629 [1] NCCL INFO Channel 01/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-252:688202:688544 [2] NCCL INFO Channel 22/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-252:688207:688543 [7] NCCL INFO Channel 21/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-252:688205:688545 [5] NCCL INFO Channel 19/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-252:688202:688544 [2] NCCL INFO Channel 23/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-116:1019294:1019626 [6] NCCL INFO Channel 11/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-116:1019291:1019630 [3] NCCL INFO Channel 08/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-116:1019288:1019623 [0] NCCL INFO Channel 11/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-116:1019292:1019624 [4] NCCL INFO Channel 04/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-116:1019295:1019627 [7] NCCL INFO Channel 07/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-252:688207:688543 [7] NCCL INFO Channel 22/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-252:688205:688545 [5] NCCL INFO Channel 20/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-116:1019290:1019628 [2] NCCL INFO Channel 03/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-116:1019289:1019629 [1] NCCL INFO Channel 02/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-252:688207:688543 [7] NCCL INFO Channel 23/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-147:618587:618924 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-172-147:618587:618924 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-172-116:1019291:1019630 [3] NCCL INFO Channel 09/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-116:1019288:1019623 [0] NCCL INFO Channel 12/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-116:1019294:1019626 [6] NCCL INFO Channel 12/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-116:1019292:1019624 [4] NCCL INFO Channel 05/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-116:1019295:1019627 [7] NCCL INFO Channel 08/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-116:1019290:1019628 [2] NCCL INFO Channel 04/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-116:1019289:1019629 [1] NCCL INFO Channel 03/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-116:1019288:1019623 [0] NCCL INFO Channel 13/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-116:1019291:1019630 [3] NCCL INFO Channel 10/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-116:1019292:1019624 [4] NCCL INFO Channel 06/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-116:1019295:1019627 [7] NCCL INFO Channel 09/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-252:688205:688545 [5] NCCL INFO Channel 21/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-116:1019290:1019628 [2] NCCL INFO Channel 05/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-116:1019289:1019629 [1] NCCL INFO Channel 04/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-116:1019294:1019626 [6] NCCL INFO Channel 13/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-147:618589:618922 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-172-147:618589:618922 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-172-116:1019293:1019625 [5] NCCL INFO Channel 00/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-116:1019288:1019623 [0] NCCL INFO Channel 14/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-116:1019295:1019627 [7] NCCL INFO Channel 10/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-116:1019291:1019630 [3] NCCL INFO Channel 11/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-116:1019292:1019624 [4] NCCL INFO Channel 07/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-116:1019290:1019628 [2] NCCL INFO Channel 06/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-116:1019294:1019626 [6] NCCL INFO Channel 14/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-116:1019289:1019629 [1] NCCL INFO Channel 05/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-116:1019293:1019625 [5] NCCL INFO Channel 01/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-116:1019288:1019623 [0] NCCL INFO Channel 15/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-116:1019295:1019627 [7] NCCL INFO Channel 11/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-116:1019291:1019630 [3] NCCL INFO Channel 12/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-116:1019292:1019624 [4] NCCL INFO Channel 08/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-116:1019290:1019628 [2] NCCL INFO Channel 07/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-116:1019294:1019626 [6] NCCL INFO Channel 15/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-116:1019289:1019629 [1] NCCL INFO Channel 06/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-116:1019293:1019625 [5] NCCL INFO Channel 02/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-252:688205:688545 [5] NCCL INFO Channel 22/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-116:1019288:1019623 [0] NCCL INFO Channel 16/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-116:1019295:1019627 [7] NCCL INFO Channel 12/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-116:1019292:1019624 [4] NCCL INFO Channel 09/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-116:1019291:1019630 [3] NCCL INFO Channel 13/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-116:1019290:1019628 [2] NCCL INFO Channel 08/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-116:1019294:1019626 [6] NCCL INFO Channel 16/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-116:1019289:1019629 [1] NCCL INFO Channel 07/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-116:1019293:1019625 [5] NCCL INFO Channel 03/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-57:942612:942954 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-172-57:942612:942954 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-172-116:1019288:1019623 [0] NCCL INFO Channel 17/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-116:1019295:1019627 [7] NCCL INFO Channel 13/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-116:1019292:1019624 [4] NCCL INFO Channel 10/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-73:926601:926934 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-172-116:1019291:1019630 [3] NCCL INFO Channel 14/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-116:1019290:1019628 [2] NCCL INFO Channel 09/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-116:1019294:1019626 [6] NCCL INFO Channel 17/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-116:1019289:1019629 [1] NCCL INFO Channel 08/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-73:926601:926934 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-173-7:811255:811595 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-172-116:1019293:1019625 [5] NCCL INFO Channel 04/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-173-7:811255:811595 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-172-116:1019288:1019623 [0] NCCL INFO Channel 18/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-116:1019295:1019627 [7] NCCL INFO Channel 14/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-116:1019292:1019624 [4] NCCL INFO Channel 11/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-116:1019291:1019630 [3] NCCL INFO Channel 15/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-116:1019290:1019628 [2] NCCL INFO Channel 10/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-116:1019294:1019626 [6] NCCL INFO Channel 18/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-116:1019289:1019629 [1] NCCL INFO Channel 09/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-116:1019293:1019625 [5] NCCL INFO Channel 05/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-116:1019288:1019623 [0] NCCL INFO Channel 19/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-116:1019295:1019627 [7] NCCL INFO Channel 15/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-116:1019292:1019624 [4] NCCL INFO Channel 12/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-116:1019291:1019630 [3] NCCL INFO Channel 16/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-116:1019290:1019628 [2] NCCL INFO Channel 11/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-116:1019294:1019626 [6] NCCL INFO Channel 19/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-116:1019289:1019629 [1] NCCL INFO Channel 10/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-252:688205:688545 [5] NCCL INFO Channel 23/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-116:1019293:1019625 [5] NCCL INFO Channel 06/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-116:1019288:1019623 [0] NCCL INFO Channel 20/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-116:1019295:1019627 [7] NCCL INFO Channel 16/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-116:1019292:1019624 [4] NCCL INFO Channel 13/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-147:618584:618919 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-172-116:1019291:1019630 [3] NCCL INFO Channel 17/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-116:1019290:1019628 [2] NCCL INFO Channel 12/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-116:1019294:1019626 [6] NCCL INFO Channel 20/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-116:1019289:1019629 [1] NCCL INFO Channel 11/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-116:1019293:1019625 [5] NCCL INFO Channel 07/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-116:1019288:1019623 [0] NCCL INFO Channel 21/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-116:1019295:1019627 [7] NCCL INFO Channel 17/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-116:1019292:1019624 [4] NCCL INFO Channel 14/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-147:618584:618919 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-172-116:1019291:1019630 [3] NCCL INFO Channel 18/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-116:1019290:1019628 [2] NCCL INFO Channel 13/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-116:1019294:1019626 [6] NCCL INFO Channel 21/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-116:1019289:1019629 [1] NCCL INFO Channel 12/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-116:1019295:1019627 [7] NCCL INFO Channel 18/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-116:1019293:1019625 [5] NCCL INFO Channel 08/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-116:1019288:1019623 [0] NCCL INFO Channel 22/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-116:1019292:1019624 [4] NCCL INFO Channel 15/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-116:1019291:1019630 [3] NCCL INFO Channel 19/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-116:1019290:1019628 [2] NCCL INFO Channel 14/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-160-225:94347:94685 [5] NCCL INFO Channel 21/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-116:1019295:1019627 [7] NCCL INFO Channel 19/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-116:1019293:1019625 [5] NCCL INFO Channel 09/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-142:880686:881021 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-172-142:880686:881021 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-172-116:1019291:1019630 [3] NCCL INFO Channel 20/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-116:1019290:1019628 [2] NCCL INFO Channel 15/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-116:1019295:1019627 [7] NCCL INFO Channel 20/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-116:1019294:1019626 [6] NCCL INFO Channel 22/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-116:1019293:1019625 [5] NCCL INFO Channel 10/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-116:1019291:1019630 [3] NCCL INFO Channel 21/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-116:1019290:1019628 [2] NCCL INFO Channel 16/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-116:1019295:1019627 [7] NCCL INFO Channel 21/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-116:1019294:1019626 [6] NCCL INFO Channel 23/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-116:1019293:1019625 [5] NCCL INFO Channel 11/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-116:1019291:1019630 [3] NCCL INFO Channel 22/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-116:1019295:1019627 [7] NCCL INFO Channel 22/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-116:1019290:1019628 [2] NCCL INFO Channel 17/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-116:1019293:1019625 [5] NCCL INFO Channel 12/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-116:1019288:1019623 [0] NCCL INFO Channel 23/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-116:1019291:1019630 [3] NCCL INFO Channel 23/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-116:1019295:1019627 [7] NCCL INFO Channel 23/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-116:1019290:1019628 [2] NCCL INFO Channel 18/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-116:1019293:1019625 [5] NCCL INFO Channel 13/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-73:926600:926933 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-172-73:926600:926933 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-172-116:1019290:1019628 [2] NCCL INFO Channel 19/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-116:1019293:1019625 [5] NCCL INFO Channel 14/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-116:1019289:1019629 [1] NCCL INFO Channel 13/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-116:1019292:1019624 [4] NCCL INFO Channel 16/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-116:1019290:1019628 [2] NCCL INFO Channel 20/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-116:1019293:1019625 [5] NCCL INFO Channel 15/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-116:1019289:1019629 [1] NCCL INFO Channel 14/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-116:1019292:1019624 [4] NCCL INFO Channel 17/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-116:1019290:1019628 [2] NCCL INFO Channel 21/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-116:1019293:1019625 [5] NCCL INFO Channel 16/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-116:1019292:1019624 [4] NCCL INFO Channel 18/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-116:1019289:1019629 [1] NCCL INFO Channel 15/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-73:926597:926935 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-172-73:926597:926935 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-172-116:1019290:1019628 [2] NCCL INFO Channel 22/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-116:1019292:1019624 [4] NCCL INFO Channel 19/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-116:1019293:1019625 [5] NCCL INFO Channel 17/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-116:1019289:1019629 [1] NCCL INFO Channel 16/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-73:926595:926937 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-172-73:926595:926937 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-172-116:1019292:1019624 [4] NCCL INFO Channel 20/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-116:1019293:1019625 [5] NCCL INFO Channel 18/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-116:1019289:1019629 [1] NCCL INFO Channel 17/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-116:1019290:1019628 [2] NCCL INFO Channel 23/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-73:926594:926931 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-172-73:926594:926931 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-172-73:926594:926931 [0] NCCL INFO Channel 00/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-73:926594:926931 [0] NCCL INFO Channel 01/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-73:926594:926931 [0] NCCL INFO Channel 02/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-73:926594:926931 [0] NCCL INFO Channel 03/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-73:926594:926931 [0] NCCL INFO Channel 04/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-73:926594:926931 [0] NCCL INFO Channel 05/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-73:926594:926931 [0] NCCL INFO Channel 06/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-73:926601:926934 [7] NCCL INFO Trees [0] -1/-1/-1->7->6 [1] -1/-1/-1->7->6 [2] -1/-1/-1->7->6 [3] -1/-1/-1->7->6 [4] -1/-1/-1->7->6 [5] -1/-1/-1->7->6 [6] -1/-1/-1->7->6 [7] -1/-1/-1->7->6 [8] -1/-1/-1->7->6 [9] -1/-1/-1->7->6 [10] -1/-1/-1->7->6 [11] -1/-1/-1->7->6 [12] -1/-1/-1->7->6 [13] -1/-1/-1->7->6 [14] -1/-1/-1->7->6 [15] -1/-1/-1->7->6 [16] -1/-1/-1->7->6 [17] -1/-1/-1->7->6 [18] -1/-1/-1->7->6 [19] -1/-1/-1->7->6 [20] -1/-1/-1->7->6 [21] -1/-1/-1->7->6 [22] -1/-1/-1->7->6 [23] -1/-1/-1->7->6
+ip-26-0-172-73:926594:926931 [0] NCCL INFO Channel 07/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-73:926594:926931 [0] NCCL INFO Channel 08/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-73:926601:926934 [7] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-172-73:926594:926931 [0] NCCL INFO Channel 09/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-73:926594:926931 [0] NCCL INFO Channel 10/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-73:926594:926931 [0] NCCL INFO Channel 11/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-73:926594:926931 [0] NCCL INFO Channel 12/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-73:926600:926933 [6] NCCL INFO Trees [0] 7/-1/-1->6->5 [1] 7/-1/-1->6->5 [2] 7/-1/-1->6->5 [3] 7/-1/-1->6->5 [4] 7/-1/-1->6->5 [5] 7/-1/-1->6->5 [6] 7/-1/-1->6->5 [7] 7/-1/-1->6->5 [8] 7/-1/-1->6->5 [9] 7/-1/-1->6->5 [10] 7/-1/-1->6->5 [11] 7/-1/-1->6->5 [12] 7/-1/-1->6->5 [13] 7/-1/-1->6->5 [14] 7/-1/-1->6->5 [15] 7/-1/-1->6->5 [16] 7/-1/-1->6->5 [17] 7/-1/-1->6->5 [18] 7/-1/-1->6->5 [19] 7/-1/-1->6->5 [20] 7/-1/-1->6->5 [21] 7/-1/-1->6->5 [22] 7/-1/-1->6->5 [23] 7/-1/-1->6->5
+ip-26-0-172-73:926594:926931 [0] NCCL INFO Channel 13/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-73:926594:926931 [0] NCCL INFO Channel 14/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-73:926600:926933 [6] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-172-73:926594:926931 [0] NCCL INFO Channel 15/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-73:926594:926931 [0] NCCL INFO Channel 16/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-73:926594:926931 [0] NCCL INFO Channel 17/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-73:926594:926931 [0] NCCL INFO Channel 18/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-73:926594:926931 [0] NCCL INFO Channel 19/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-73:926595:926937 [1] NCCL INFO Trees [0] 2/-1/-1->1->0 [1] 2/-1/-1->1->0 [2] 2/-1/-1->1->0 [3] 2/-1/-1->1->0 [4] 2/-1/-1->1->0 [5] 2/-1/-1->1->0 [6] 2/-1/-1->1->0 [7] 2/-1/-1->1->0 [8] 2/-1/-1->1->0 [9] 2/-1/-1->1->0 [10] 2/-1/-1->1->0 [11] 2/-1/-1->1->0 [12] 2/-1/-1->1->0 [13] 2/-1/-1->1->0 [14] 2/-1/-1->1->0 [15] 2/-1/-1->1->0 [16] 2/-1/-1->1->0 [17] 2/-1/-1->1->0 [18] 2/-1/-1->1->0 [19] 2/-1/-1->1->0 [20] 2/-1/-1->1->0 [21] 2/-1/-1->1->0 [22] 2/-1/-1->1->0 [23] 2/-1/-1->1->0
+ip-26-0-172-73:926594:926931 [0] NCCL INFO Channel 20/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-73:926594:926931 [0] NCCL INFO Channel 21/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-73:926595:926937 [1] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-172-73:926594:926931 [0] NCCL INFO Channel 22/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-73:926594:926931 [0] NCCL INFO Channel 23/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-73:926598:926938 [4] NCCL INFO Trees [0] 5/-1/-1->4->3 [1] 5/-1/-1->4->3 [2] 5/-1/-1->4->3 [3] 5/-1/-1->4->3 [4] 5/-1/-1->4->3 [5] 5/-1/-1->4->3 [6] 5/-1/-1->4->3 [7] 5/-1/-1->4->3 [8] 5/-1/-1->4->3 [9] 5/-1/-1->4->3 [10] 5/-1/-1->4->3 [11] 5/-1/-1->4->3 [12] 5/-1/-1->4->3 [13] 5/-1/-1->4->3 [14] 5/-1/-1->4->3 [15] 5/-1/-1->4->3 [16] 5/-1/-1->4->3 [17] 5/-1/-1->4->3 [18] 5/-1/-1->4->3 [19] 5/-1/-1->4->3 [20] 5/-1/-1->4->3 [21] 5/-1/-1->4->3 [22] 5/-1/-1->4->3 [23] 5/-1/-1->4->3
+ip-26-0-172-73:926597:926935 [3] NCCL INFO Trees [0] 4/-1/-1->3->2 [1] 4/-1/-1->3->2 [2] 4/-1/-1->3->2 [3] 4/-1/-1->3->2 [4] 4/-1/-1->3->2 [5] 4/-1/-1->3->2 [6] 4/-1/-1->3->2 [7] 4/-1/-1->3->2 [8] 4/-1/-1->3->2 [9] 4/-1/-1->3->2 [10] 4/-1/-1->3->2 [11] 4/-1/-1->3->2 [12] 4/-1/-1->3->2 [13] 4/-1/-1->3->2 [14] 4/-1/-1->3->2 [15] 4/-1/-1->3->2 [16] 4/-1/-1->3->2 [17] 4/-1/-1->3->2 [18] 4/-1/-1->3->2 [19] 4/-1/-1->3->2 [20] 4/-1/-1->3->2 [21] 4/-1/-1->3->2 [22] 4/-1/-1->3->2 [23] 4/-1/-1->3->2
+ip-26-0-172-73:926598:926938 [4] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-172-73:926599:926932 [5] NCCL INFO Trees [0] 6/-1/-1->5->4 [1] 6/-1/-1->5->4 [2] 6/-1/-1->5->4 [3] 6/-1/-1->5->4 [4] 6/-1/-1->5->4 [5] 6/-1/-1->5->4 [6] 6/-1/-1->5->4 [7] 6/-1/-1->5->4 [8] 6/-1/-1->5->4 [9] 6/-1/-1->5->4 [10] 6/-1/-1->5->4 [11] 6/-1/-1->5->4 [12] 6/-1/-1->5->4 [13] 6/-1/-1->5->4 [14] 6/-1/-1->5->4 [15] 6/-1/-1->5->4 [16] 6/-1/-1->5->4 [17] 6/-1/-1->5->4 [18] 6/-1/-1->5->4 [19] 6/-1/-1->5->4 [20] 6/-1/-1->5->4 [21] 6/-1/-1->5->4 [22] 6/-1/-1->5->4 [23] 6/-1/-1->5->4
+ip-26-0-172-73:926597:926935 [3] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-172-73:926594:926931 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] 1/-1/-1->0->-1 [7] 1/-1/-1->0->-1 [8] 1/-1/-1->0->-1 [9] 1/-1/-1->0->-1 [10] 1/-1/-1->0->-1 [11] 1/-1/-1->0->-1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] 1/-1/-1->0->-1 [19] 1/-1/-1->0->-1 [20] 1/-1/-1->0->-1 [21] 1/-1/-1->0->-1 [22] 1/-1/-1->0->-1 [23] 1/-1/-1->0->-1
+ip-26-0-172-73:926599:926932 [5] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-172-73:926594:926931 [0] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-172-73:926596:926936 [2] NCCL INFO Trees [0] 3/-1/-1->2->1 [1] 3/-1/-1->2->1 [2] 3/-1/-1->2->1 [3] 3/-1/-1->2->1 [4] 3/-1/-1->2->1 [5] 3/-1/-1->2->1 [6] 3/-1/-1->2->1 [7] 3/-1/-1->2->1 [8] 3/-1/-1->2->1 [9] 3/-1/-1->2->1 [10] 3/-1/-1->2->1 [11] 3/-1/-1->2->1 [12] 3/-1/-1->2->1 [13] 3/-1/-1->2->1 [14] 3/-1/-1->2->1 [15] 3/-1/-1->2->1 [16] 3/-1/-1->2->1 [17] 3/-1/-1->2->1 [18] 3/-1/-1->2->1 [19] 3/-1/-1->2->1 [20] 3/-1/-1->2->1 [21] 3/-1/-1->2->1 [22] 3/-1/-1->2->1 [23] 3/-1/-1->2->1
+ip-26-0-172-73:926596:926936 [2] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-172-147:618585:618926 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-172-147:618585:618926 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-172-116:1019293:1019625 [5] NCCL INFO Channel 19/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-116:1019292:1019624 [4] NCCL INFO Channel 21/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-116:1019289:1019629 [1] NCCL INFO Channel 18/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-225:94347:94685 [5] NCCL INFO Channel 22/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-116:1019293:1019625 [5] NCCL INFO Channel 20/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-116:1019292:1019624 [4] NCCL INFO Channel 22/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-116:1019289:1019629 [1] NCCL INFO Channel 19/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-160-225:94347:94685 [5] NCCL INFO Channel 23/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-173-7:811252:811592 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-173-7:811252:811592 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-172-116:1019289:1019629 [1] NCCL INFO Channel 20/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-116:1019292:1019624 [4] NCCL INFO Channel 23/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-116:1019293:1019625 [5] NCCL INFO Channel 21/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-116:1019289:1019629 [1] NCCL INFO Channel 21/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-116:1019293:1019625 [5] NCCL INFO Channel 22/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-147:618591:618921 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-172-147:618591:618921 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-172-147:618588:618920 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-172-147:618588:618920 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-172-147:618588:618920 [4] NCCL INFO Trees [0] 5/-1/-1->4->3 [1] 5/-1/-1->4->3 [2] 5/-1/-1->4->3 [3] 5/-1/-1->4->3 [4] 5/-1/-1->4->3 [5] 5/-1/-1->4->3 [6] 5/-1/-1->4->3 [7] 5/-1/-1->4->3 [8] 5/-1/-1->4->3 [9] 5/-1/-1->4->3 [10] 5/-1/-1->4->3 [11] 5/-1/-1->4->3 [12] 5/-1/-1->4->3 [13] 5/-1/-1->4->3 [14] 5/-1/-1->4->3 [15] 5/-1/-1->4->3 [16] 5/-1/-1->4->3 [17] 5/-1/-1->4->3 [18] 5/-1/-1->4->3 [19] 5/-1/-1->4->3 [20] 5/-1/-1->4->3 [21] 5/-1/-1->4->3 [22] 5/-1/-1->4->3 [23] 5/-1/-1->4->3
+ip-26-0-172-147:618588:618920 [4] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-172-147:618584:618919 [0] NCCL INFO Channel 00/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-147:618584:618919 [0] NCCL INFO Channel 01/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-147:618584:618919 [0] NCCL INFO Channel 02/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-147:618584:618919 [0] NCCL INFO Channel 03/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-147:618591:618921 [7] NCCL INFO Trees [0] -1/-1/-1->7->6 [1] -1/-1/-1->7->6 [2] -1/-1/-1->7->6 [3] -1/-1/-1->7->6 [4] -1/-1/-1->7->6 [5] -1/-1/-1->7->6 [6] -1/-1/-1->7->6 [7] -1/-1/-1->7->6 [8] -1/-1/-1->7->6 [9] -1/-1/-1->7->6 [10] -1/-1/-1->7->6 [11] -1/-1/-1->7->6 [12] -1/-1/-1->7->6 [13] -1/-1/-1->7->6 [14] -1/-1/-1->7->6 [15] -1/-1/-1->7->6 [16] -1/-1/-1->7->6 [17] -1/-1/-1->7->6 [18] -1/-1/-1->7->6 [19] -1/-1/-1->7->6 [20] -1/-1/-1->7->6 [21] -1/-1/-1->7->6 [22] -1/-1/-1->7->6 [23] -1/-1/-1->7->6
+ip-26-0-172-147:618584:618919 [0] NCCL INFO Channel 04/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-147:618586:618923 [2] NCCL INFO Trees [0] 3/-1/-1->2->1 [1] 3/-1/-1->2->1 [2] 3/-1/-1->2->1 [3] 3/-1/-1->2->1 [4] 3/-1/-1->2->1 [5] 3/-1/-1->2->1 [6] 3/-1/-1->2->1 [7] 3/-1/-1->2->1 [8] 3/-1/-1->2->1 [9] 3/-1/-1->2->1 [10] 3/-1/-1->2->1 [11] 3/-1/-1->2->1 [12] 3/-1/-1->2->1 [13] 3/-1/-1->2->1 [14] 3/-1/-1->2->1 [15] 3/-1/-1->2->1 [16] 3/-1/-1->2->1 [17] 3/-1/-1->2->1 [18] 3/-1/-1->2->1 [19] 3/-1/-1->2->1 [20] 3/-1/-1->2->1 [21] 3/-1/-1->2->1 [22] 3/-1/-1->2->1 [23] 3/-1/-1->2->1
+ip-26-0-172-147:618585:618926 [1] NCCL INFO Trees [0] 2/-1/-1->1->0 [1] 2/-1/-1->1->0 [2] 2/-1/-1->1->0 [3] 2/-1/-1->1->0 [4] 2/-1/-1->1->0 [5] 2/-1/-1->1->0 [6] 2/-1/-1->1->0 [7] 2/-1/-1->1->0 [8] 2/-1/-1->1->0 [9] 2/-1/-1->1->0 [10] 2/-1/-1->1->0 [11] 2/-1/-1->1->0 [12] 2/-1/-1->1->0 [13] 2/-1/-1->1->0 [14] 2/-1/-1->1->0 [15] 2/-1/-1->1->0 [16] 2/-1/-1->1->0 [17] 2/-1/-1->1->0 [18] 2/-1/-1->1->0 [19] 2/-1/-1->1->0 [20] 2/-1/-1->1->0 [21] 2/-1/-1->1->0 [22] 2/-1/-1->1->0 [23] 2/-1/-1->1->0
+ip-26-0-172-147:618587:618924 [3] NCCL INFO Trees [0] 4/-1/-1->3->2 [1] 4/-1/-1->3->2 [2] 4/-1/-1->3->2 [3] 4/-1/-1->3->2 [4] 4/-1/-1->3->2 [5] 4/-1/-1->3->2 [6] 4/-1/-1->3->2 [7] 4/-1/-1->3->2 [8] 4/-1/-1->3->2 [9] 4/-1/-1->3->2 [10] 4/-1/-1->3->2 [11] 4/-1/-1->3->2 [12] 4/-1/-1->3->2 [13] 4/-1/-1->3->2 [14] 4/-1/-1->3->2 [15] 4/-1/-1->3->2 [16] 4/-1/-1->3->2 [17] 4/-1/-1->3->2 [18] 4/-1/-1->3->2 [19] 4/-1/-1->3->2 [20] 4/-1/-1->3->2 [21] 4/-1/-1->3->2 [22] 4/-1/-1->3->2 [23] 4/-1/-1->3->2
+ip-26-0-172-147:618584:618919 [0] NCCL INFO Channel 05/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-147:618586:618923 [2] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-172-147:618589:618922 [5] NCCL INFO Trees [0] 6/-1/-1->5->4 [1] 6/-1/-1->5->4 [2] 6/-1/-1->5->4 [3] 6/-1/-1->5->4 [4] 6/-1/-1->5->4 [5] 6/-1/-1->5->4 [6] 6/-1/-1->5->4 [7] 6/-1/-1->5->4 [8] 6/-1/-1->5->4 [9] 6/-1/-1->5->4 [10] 6/-1/-1->5->4 [11] 6/-1/-1->5->4 [12] 6/-1/-1->5->4 [13] 6/-1/-1->5->4 [14] 6/-1/-1->5->4 [15] 6/-1/-1->5->4 [16] 6/-1/-1->5->4 [17] 6/-1/-1->5->4 [18] 6/-1/-1->5->4 [19] 6/-1/-1->5->4 [20] 6/-1/-1->5->4 [21] 6/-1/-1->5->4 [22] 6/-1/-1->5->4 [23] 6/-1/-1->5->4
+ip-26-0-172-147:618591:618921 [7] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-172-147:618590:618925 [6] NCCL INFO Trees [0] 7/-1/-1->6->5 [1] 7/-1/-1->6->5 [2] 7/-1/-1->6->5 [3] 7/-1/-1->6->5 [4] 7/-1/-1->6->5 [5] 7/-1/-1->6->5 [6] 7/-1/-1->6->5 [7] 7/-1/-1->6->5 [8] 7/-1/-1->6->5 [9] 7/-1/-1->6->5 [10] 7/-1/-1->6->5 [11] 7/-1/-1->6->5 [12] 7/-1/-1->6->5 [13] 7/-1/-1->6->5 [14] 7/-1/-1->6->5 [15] 7/-1/-1->6->5 [16] 7/-1/-1->6->5 [17] 7/-1/-1->6->5 [18] 7/-1/-1->6->5 [19] 7/-1/-1->6->5 [20] 7/-1/-1->6->5 [21] 7/-1/-1->6->5 [22] 7/-1/-1->6->5 [23] 7/-1/-1->6->5
+ip-26-0-172-147:618584:618919 [0] NCCL INFO Channel 06/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-147:618587:618924 [3] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-172-147:618585:618926 [1] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-172-147:618590:618925 [6] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-172-147:618589:618922 [5] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-172-147:618584:618919 [0] NCCL INFO Channel 07/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-147:618584:618919 [0] NCCL INFO Channel 08/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-147:618584:618919 [0] NCCL INFO Channel 09/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-147:618584:618919 [0] NCCL INFO Channel 10/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-147:618584:618919 [0] NCCL INFO Channel 11/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-147:618584:618919 [0] NCCL INFO Channel 12/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-147:618584:618919 [0] NCCL INFO Channel 13/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-147:618584:618919 [0] NCCL INFO Channel 14/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-147:618584:618919 [0] NCCL INFO Channel 15/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-147:618584:618919 [0] NCCL INFO Channel 16/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-147:618584:618919 [0] NCCL INFO Channel 17/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-147:618584:618919 [0] NCCL INFO Channel 18/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-147:618584:618919 [0] NCCL INFO Channel 19/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-147:618584:618919 [0] NCCL INFO Channel 20/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-147:618584:618919 [0] NCCL INFO Channel 21/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-147:618584:618919 [0] NCCL INFO Channel 22/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-147:618584:618919 [0] NCCL INFO Channel 23/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-147:618584:618919 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] 1/-1/-1->0->-1 [7] 1/-1/-1->0->-1 [8] 1/-1/-1->0->-1 [9] 1/-1/-1->0->-1 [10] 1/-1/-1->0->-1 [11] 1/-1/-1->0->-1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] 1/-1/-1->0->-1 [19] 1/-1/-1->0->-1 [20] 1/-1/-1->0->-1 [21] 1/-1/-1->0->-1 [22] 1/-1/-1->0->-1 [23] 1/-1/-1->0->-1
+ip-26-0-172-147:618584:618919 [0] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-173-7:811253:811593 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-173-7:811253:811593 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-172-116:1019293:1019625 [5] NCCL INFO Channel 23/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-116:1019289:1019629 [1] NCCL INFO Channel 22/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-173-7:811249:811596 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-173-7:811249:811596 [1] NCCL INFO NVLS multicast support is available on dev 1
+ip-26-0-172-116:1019289:1019629 [1] NCCL INFO Channel 23/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-57:942617:942949 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-172-57:942617:942949 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-173-7:811254:811594 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-173-7:811254:811594 [6] NCCL INFO NVLS multicast support is available on dev 6
+ip-26-0-173-7:811254:811594 [6] NCCL INFO Trees [0] 7/-1/-1->6->5 [1] 7/-1/-1->6->5 [2] 7/-1/-1->6->5 [3] 7/-1/-1->6->5 [4] 7/-1/-1->6->5 [5] 7/-1/-1->6->5 [6] 7/-1/-1->6->5 [7] 7/-1/-1->6->5 [8] 7/-1/-1->6->5 [9] 7/-1/-1->6->5 [10] 7/-1/-1->6->5 [11] 7/-1/-1->6->5 [12] 7/-1/-1->6->5 [13] 7/-1/-1->6->5 [14] 7/-1/-1->6->5 [15] 7/-1/-1->6->5 [16] 7/-1/-1->6->5 [17] 7/-1/-1->6->5 [18] 7/-1/-1->6->5 [19] 7/-1/-1->6->5 [20] 7/-1/-1->6->5 [21] 7/-1/-1->6->5 [22] 7/-1/-1->6->5 [23] 7/-1/-1->6->5
+ip-26-0-173-7:811254:811594 [6] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-173-7:811248:811591 [0] NCCL INFO Channel 00/24 :    0   1   2   3   4   5   6   7
+ip-26-0-173-7:811253:811593 [5] NCCL INFO Trees [0] 6/-1/-1->5->4 [1] 6/-1/-1->5->4 [2] 6/-1/-1->5->4 [3] 6/-1/-1->5->4 [4] 6/-1/-1->5->4 [5] 6/-1/-1->5->4 [6] 6/-1/-1->5->4 [7] 6/-1/-1->5->4 [8] 6/-1/-1->5->4 [9] 6/-1/-1->5->4 [10] 6/-1/-1->5->4 [11] 6/-1/-1->5->4 [12] 6/-1/-1->5->4 [13] 6/-1/-1->5->4 [14] 6/-1/-1->5->4 [15] 6/-1/-1->5->4 [16] 6/-1/-1->5->4 [17] 6/-1/-1->5->4 [18] 6/-1/-1->5->4 [19] 6/-1/-1->5->4 [20] 6/-1/-1->5->4 [21] 6/-1/-1->5->4 [22] 6/-1/-1->5->4 [23] 6/-1/-1->5->4
+ip-26-0-173-7:811248:811591 [0] NCCL INFO Channel 01/24 :    0   1   2   3   4   5   6   7
+ip-26-0-173-7:811253:811593 [5] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-173-7:811248:811591 [0] NCCL INFO Channel 02/24 :    0   1   2   3   4   5   6   7
+ip-26-0-173-7:811255:811595 [7] NCCL INFO Trees [0] -1/-1/-1->7->6 [1] -1/-1/-1->7->6 [2] -1/-1/-1->7->6 [3] -1/-1/-1->7->6 [4] -1/-1/-1->7->6 [5] -1/-1/-1->7->6 [6] -1/-1/-1->7->6 [7] -1/-1/-1->7->6 [8] -1/-1/-1->7->6 [9] -1/-1/-1->7->6 [10] -1/-1/-1->7->6 [11] -1/-1/-1->7->6 [12] -1/-1/-1->7->6 [13] -1/-1/-1->7->6 [14] -1/-1/-1->7->6 [15] -1/-1/-1->7->6 [16] -1/-1/-1->7->6 [17] -1/-1/-1->7->6 [18] -1/-1/-1->7->6 [19] -1/-1/-1->7->6 [20] -1/-1/-1->7->6 [21] -1/-1/-1->7->6 [22] -1/-1/-1->7->6 [23] -1/-1/-1->7->6
+ip-26-0-173-7:811248:811591 [0] NCCL INFO Channel 03/24 :    0   1   2   3   4   5   6   7
+ip-26-0-173-7:811252:811592 [4] NCCL INFO Trees [0] 5/-1/-1->4->3 [1] 5/-1/-1->4->3 [2] 5/-1/-1->4->3 [3] 5/-1/-1->4->3 [4] 5/-1/-1->4->3 [5] 5/-1/-1->4->3 [6] 5/-1/-1->4->3 [7] 5/-1/-1->4->3 [8] 5/-1/-1->4->3 [9] 5/-1/-1->4->3 [10] 5/-1/-1->4->3 [11] 5/-1/-1->4->3 [12] 5/-1/-1->4->3 [13] 5/-1/-1->4->3 [14] 5/-1/-1->4->3 [15] 5/-1/-1->4->3 [16] 5/-1/-1->4->3 [17] 5/-1/-1->4->3 [18] 5/-1/-1->4->3 [19] 5/-1/-1->4->3 [20] 5/-1/-1->4->3 [21] 5/-1/-1->4->3 [22] 5/-1/-1->4->3 [23] 5/-1/-1->4->3
+ip-26-0-173-7:811255:811595 [7] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-173-7:811248:811591 [0] NCCL INFO Channel 04/24 :    0   1   2   3   4   5   6   7
+ip-26-0-173-7:811249:811596 [1] NCCL INFO Trees [0] 2/-1/-1->1->0 [1] 2/-1/-1->1->0 [2] 2/-1/-1->1->0 [3] 2/-1/-1->1->0 [4] 2/-1/-1->1->0 [5] 2/-1/-1->1->0 [6] 2/-1/-1->1->0 [7] 2/-1/-1->1->0 [8] 2/-1/-1->1->0 [9] 2/-1/-1->1->0 [10] 2/-1/-1->1->0 [11] 2/-1/-1->1->0 [12] 2/-1/-1->1->0 [13] 2/-1/-1->1->0 [14] 2/-1/-1->1->0 [15] 2/-1/-1->1->0 [16] 2/-1/-1->1->0 [17] 2/-1/-1->1->0 [18] 2/-1/-1->1->0 [19] 2/-1/-1->1->0 [20] 2/-1/-1->1->0 [21] 2/-1/-1->1->0 [22] 2/-1/-1->1->0 [23] 2/-1/-1->1->0
+ip-26-0-173-7:811248:811591 [0] NCCL INFO Channel 05/24 :    0   1   2   3   4   5   6   7
+ip-26-0-173-7:811252:811592 [4] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-173-7:811248:811591 [0] NCCL INFO Channel 06/24 :    0   1   2   3   4   5   6   7
+ip-26-0-173-7:811248:811591 [0] NCCL INFO Channel 07/24 :    0   1   2   3   4   5   6   7
+ip-26-0-173-7:811248:811591 [0] NCCL INFO Channel 08/24 :    0   1   2   3   4   5   6   7
+ip-26-0-173-7:811248:811591 [0] NCCL INFO Channel 09/24 :    0   1   2   3   4   5   6   7
+ip-26-0-173-7:811248:811591 [0] NCCL INFO Channel 10/24 :    0   1   2   3   4   5   6   7
+ip-26-0-173-7:811249:811596 [1] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-173-7:811248:811591 [0] NCCL INFO Channel 11/24 :    0   1   2   3   4   5   6   7
+ip-26-0-173-7:811248:811591 [0] NCCL INFO Channel 12/24 :    0   1   2   3   4   5   6   7
+ip-26-0-173-7:811248:811591 [0] NCCL INFO Channel 13/24 :    0   1   2   3   4   5   6   7
+ip-26-0-173-7:811248:811591 [0] NCCL INFO Channel 14/24 :    0   1   2   3   4   5   6   7
+ip-26-0-173-7:811248:811591 [0] NCCL INFO Channel 15/24 :    0   1   2   3   4   5   6   7
+ip-26-0-173-7:811248:811591 [0] NCCL INFO Channel 16/24 :    0   1   2   3   4   5   6   7
+ip-26-0-173-7:811248:811591 [0] NCCL INFO Channel 17/24 :    0   1   2   3   4   5   6   7
+ip-26-0-173-7:811248:811591 [0] NCCL INFO Channel 18/24 :    0   1   2   3   4   5   6   7
+ip-26-0-173-7:811248:811591 [0] NCCL INFO Channel 19/24 :    0   1   2   3   4   5   6   7
+ip-26-0-173-7:811248:811591 [0] NCCL INFO Channel 20/24 :    0   1   2   3   4   5   6   7
+ip-26-0-173-7:811248:811591 [0] NCCL INFO Channel 21/24 :    0   1   2   3   4   5   6   7
+ip-26-0-173-7:811248:811591 [0] NCCL INFO Channel 22/24 :    0   1   2   3   4   5   6   7
+ip-26-0-173-7:811248:811591 [0] NCCL INFO Channel 23/24 :    0   1   2   3   4   5   6   7
+ip-26-0-173-7:811250:811597 [2] NCCL INFO Trees [0] 3/-1/-1->2->1 [1] 3/-1/-1->2->1 [2] 3/-1/-1->2->1 [3] 3/-1/-1->2->1 [4] 3/-1/-1->2->1 [5] 3/-1/-1->2->1 [6] 3/-1/-1->2->1 [7] 3/-1/-1->2->1 [8] 3/-1/-1->2->1 [9] 3/-1/-1->2->1 [10] 3/-1/-1->2->1 [11] 3/-1/-1->2->1 [12] 3/-1/-1->2->1 [13] 3/-1/-1->2->1 [14] 3/-1/-1->2->1 [15] 3/-1/-1->2->1 [16] 3/-1/-1->2->1 [17] 3/-1/-1->2->1 [18] 3/-1/-1->2->1 [19] 3/-1/-1->2->1 [20] 3/-1/-1->2->1 [21] 3/-1/-1->2->1 [22] 3/-1/-1->2->1 [23] 3/-1/-1->2->1
+ip-26-0-173-7:811251:811598 [3] NCCL INFO Trees [0] 4/-1/-1->3->2 [1] 4/-1/-1->3->2 [2] 4/-1/-1->3->2 [3] 4/-1/-1->3->2 [4] 4/-1/-1->3->2 [5] 4/-1/-1->3->2 [6] 4/-1/-1->3->2 [7] 4/-1/-1->3->2 [8] 4/-1/-1->3->2 [9] 4/-1/-1->3->2 [10] 4/-1/-1->3->2 [11] 4/-1/-1->3->2 [12] 4/-1/-1->3->2 [13] 4/-1/-1->3->2 [14] 4/-1/-1->3->2 [15] 4/-1/-1->3->2 [16] 4/-1/-1->3->2 [17] 4/-1/-1->3->2 [18] 4/-1/-1->3->2 [19] 4/-1/-1->3->2 [20] 4/-1/-1->3->2 [21] 4/-1/-1->3->2 [22] 4/-1/-1->3->2 [23] 4/-1/-1->3->2
+ip-26-0-173-7:811248:811591 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] 1/-1/-1->0->-1 [7] 1/-1/-1->0->-1 [8] 1/-1/-1->0->-1 [9] 1/-1/-1->0->-1 [10] 1/-1/-1->0->-1 [11] 1/-1/-1->0->-1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] 1/-1/-1->0->-1 [19] 1/-1/-1->0->-1 [20] 1/-1/-1->0->-1 [21] 1/-1/-1->0->-1 [22] 1/-1/-1->0->-1 [23] 1/-1/-1->0->-1
+ip-26-0-173-7:811248:811591 [0] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-173-7:811250:811597 [2] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-173-7:811251:811598 [3] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-172-142:880683:881016 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-172-142:880683:881016 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-172-57:942616:942952 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-172-57:942616:942952 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-172-73:926600:926933 [6] NCCL INFO Channel 00/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-73:926598:926938 [4] NCCL INFO Channel 00/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-57:942614:942953 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-172-73:926601:926934 [7] NCCL INFO Channel 00/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-57:942614:942953 [3] NCCL INFO NVLS multicast support is available on dev 3
+ip-26-0-172-73:926594:926931 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-73:926599:926932 [5] NCCL INFO Channel 00/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-73:926595:926937 [1] NCCL INFO Channel 00/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-73:926596:926936 [2] NCCL INFO Channel 00/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-73:926600:926933 [6] NCCL INFO Channel 01/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-73:926598:926938 [4] NCCL INFO Channel 01/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-73:926601:926934 [7] NCCL INFO Channel 01/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-73:926594:926931 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-73:926599:926932 [5] NCCL INFO Channel 01/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-73:926596:926936 [2] NCCL INFO Channel 01/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-73:926595:926937 [1] NCCL INFO Channel 01/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-73:926600:926933 [6] NCCL INFO Channel 02/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-73:926598:926938 [4] NCCL INFO Channel 02/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-57:942611:942947 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-172-73:926601:926934 [7] NCCL INFO Channel 02/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-73:926594:926931 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-73:926599:926932 [5] NCCL INFO Channel 02/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-73:926596:926936 [2] NCCL INFO Channel 02/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-73:926595:926937 [1] NCCL INFO Channel 02/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-73:926600:926933 [6] NCCL INFO Channel 03/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-73:926598:926938 [4] NCCL INFO Channel 03/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-73:926601:926934 [7] NCCL INFO Channel 03/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-73:926594:926931 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-73:926599:926932 [5] NCCL INFO Channel 03/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-73:926596:926936 [2] NCCL INFO Channel 03/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-57:942611:942947 [0] NCCL INFO NVLS multicast support is available on dev 0
+ip-26-0-172-73:926595:926937 [1] NCCL INFO Channel 03/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-73:926600:926933 [6] NCCL INFO Channel 04/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-73:926598:926938 [4] NCCL INFO Channel 04/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-73:926601:926934 [7] NCCL INFO Channel 04/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-73:926594:926931 [0] NCCL INFO Channel 04/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-73:926599:926932 [5] NCCL INFO Channel 04/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-73:926596:926936 [2] NCCL INFO Channel 04/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-73:926595:926937 [1] NCCL INFO Channel 04/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-73:926600:926933 [6] NCCL INFO Channel 05/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-73:926598:926938 [4] NCCL INFO Channel 05/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-73:926601:926934 [7] NCCL INFO Channel 05/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-73:926594:926931 [0] NCCL INFO Channel 05/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-73:926600:926933 [6] NCCL INFO Channel 06/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-73:926596:926936 [2] NCCL INFO Channel 05/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-73:926598:926938 [4] NCCL INFO Channel 06/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-73:926597:926935 [3] NCCL INFO Channel 00/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-73:926600:926933 [6] NCCL INFO Channel 07/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-73:926596:926936 [2] NCCL INFO Channel 06/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-73:926598:926938 [4] NCCL INFO Channel 07/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-73:926597:926935 [3] NCCL INFO Channel 01/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-73:926600:926933 [6] NCCL INFO Channel 08/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-73:926595:926937 [1] NCCL INFO Channel 05/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-142:880681:881019 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-172-73:926601:926934 [7] NCCL INFO Channel 06/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-73:926600:926933 [6] NCCL INFO Channel 09/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-73:926594:926931 [0] NCCL INFO Channel 06/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-73:926599:926932 [5] NCCL INFO Channel 05/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-73:926595:926937 [1] NCCL INFO Channel 06/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-142:880681:881019 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-172-73:926596:926936 [2] NCCL INFO Channel 07/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-73:926598:926938 [4] NCCL INFO Channel 08/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-73:926601:926934 [7] NCCL INFO Channel 07/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-73:926600:926933 [6] NCCL INFO Channel 10/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-73:926594:926931 [0] NCCL INFO Channel 07/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-73:926599:926932 [5] NCCL INFO Channel 06/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-73:926595:926937 [1] NCCL INFO Channel 07/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-73:926596:926936 [2] NCCL INFO Channel 08/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-73:926598:926938 [4] NCCL INFO Channel 09/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-73:926600:926933 [6] NCCL INFO Channel 11/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-73:926601:926934 [7] NCCL INFO Channel 08/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-73:926599:926932 [5] NCCL INFO Channel 07/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-73:926594:926931 [0] NCCL INFO Channel 08/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-73:926595:926937 [1] NCCL INFO Channel 08/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-73:926596:926936 [2] NCCL INFO Channel 09/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-142:880684:881017 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-172-73:926597:926935 [3] NCCL INFO Channel 02/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-142:880684:881017 [5] NCCL INFO NVLS multicast support is available on dev 5
+ip-26-0-172-142:880683:881016 [4] NCCL INFO Trees [0] 5/-1/-1->4->3 [1] 5/-1/-1->4->3 [2] 5/-1/-1->4->3 [3] 5/-1/-1->4->3 [4] 5/-1/-1->4->3 [5] 5/-1/-1->4->3 [6] 5/-1/-1->4->3 [7] 5/-1/-1->4->3 [8] 5/-1/-1->4->3 [9] 5/-1/-1->4->3 [10] 5/-1/-1->4->3 [11] 5/-1/-1->4->3 [12] 5/-1/-1->4->3 [13] 5/-1/-1->4->3 [14] 5/-1/-1->4->3 [15] 5/-1/-1->4->3 [16] 5/-1/-1->4->3 [17] 5/-1/-1->4->3 [18] 5/-1/-1->4->3 [19] 5/-1/-1->4->3 [20] 5/-1/-1->4->3 [21] 5/-1/-1->4->3 [22] 5/-1/-1->4->3 [23] 5/-1/-1->4->3
+ip-26-0-172-142:880685:881018 [6] NCCL INFO Trees [0] 7/-1/-1->6->5 [1] 7/-1/-1->6->5 [2] 7/-1/-1->6->5 [3] 7/-1/-1->6->5 [4] 7/-1/-1->6->5 [5] 7/-1/-1->6->5 [6] 7/-1/-1->6->5 [7] 7/-1/-1->6->5 [8] 7/-1/-1->6->5 [9] 7/-1/-1->6->5 [10] 7/-1/-1->6->5 [11] 7/-1/-1->6->5 [12] 7/-1/-1->6->5 [13] 7/-1/-1->6->5 [14] 7/-1/-1->6->5 [15] 7/-1/-1->6->5 [16] 7/-1/-1->6->5 [17] 7/-1/-1->6->5 [18] 7/-1/-1->6->5 [19] 7/-1/-1->6->5 [20] 7/-1/-1->6->5 [21] 7/-1/-1->6->5 [22] 7/-1/-1->6->5 [23] 7/-1/-1->6->5
+ip-26-0-172-142:880683:881016 [4] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-172-142:880685:881018 [6] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-172-142:880684:881017 [5] NCCL INFO Trees [0] 6/-1/-1->5->4 [1] 6/-1/-1->5->4 [2] 6/-1/-1->5->4 [3] 6/-1/-1->5->4 [4] 6/-1/-1->5->4 [5] 6/-1/-1->5->4 [6] 6/-1/-1->5->4 [7] 6/-1/-1->5->4 [8] 6/-1/-1->5->4 [9] 6/-1/-1->5->4 [10] 6/-1/-1->5->4 [11] 6/-1/-1->5->4 [12] 6/-1/-1->5->4 [13] 6/-1/-1->5->4 [14] 6/-1/-1->5->4 [15] 6/-1/-1->5->4 [16] 6/-1/-1->5->4 [17] 6/-1/-1->5->4 [18] 6/-1/-1->5->4 [19] 6/-1/-1->5->4 [20] 6/-1/-1->5->4 [21] 6/-1/-1->5->4 [22] 6/-1/-1->5->4 [23] 6/-1/-1->5->4
+ip-26-0-172-142:880684:881017 [5] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-172-142:880686:881021 [7] NCCL INFO Trees [0] -1/-1/-1->7->6 [1] -1/-1/-1->7->6 [2] -1/-1/-1->7->6 [3] -1/-1/-1->7->6 [4] -1/-1/-1->7->6 [5] -1/-1/-1->7->6 [6] -1/-1/-1->7->6 [7] -1/-1/-1->7->6 [8] -1/-1/-1->7->6 [9] -1/-1/-1->7->6 [10] -1/-1/-1->7->6 [11] -1/-1/-1->7->6 [12] -1/-1/-1->7->6 [13] -1/-1/-1->7->6 [14] -1/-1/-1->7->6 [15] -1/-1/-1->7->6 [16] -1/-1/-1->7->6 [17] -1/-1/-1->7->6 [18] -1/-1/-1->7->6 [19] -1/-1/-1->7->6 [20] -1/-1/-1->7->6 [21] -1/-1/-1->7->6 [22] -1/-1/-1->7->6 [23] -1/-1/-1->7->6
+ip-26-0-172-142:880679:881015 [0] NCCL INFO Channel 00/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-142:880681:881019 [2] NCCL INFO Trees [0] 3/-1/-1->2->1 [1] 3/-1/-1->2->1 [2] 3/-1/-1->2->1 [3] 3/-1/-1->2->1 [4] 3/-1/-1->2->1 [5] 3/-1/-1->2->1 [6] 3/-1/-1->2->1 [7] 3/-1/-1->2->1 [8] 3/-1/-1->2->1 [9] 3/-1/-1->2->1 [10] 3/-1/-1->2->1 [11] 3/-1/-1->2->1 [12] 3/-1/-1->2->1 [13] 3/-1/-1->2->1 [14] 3/-1/-1->2->1 [15] 3/-1/-1->2->1 [16] 3/-1/-1->2->1 [17] 3/-1/-1->2->1 [18] 3/-1/-1->2->1 [19] 3/-1/-1->2->1 [20] 3/-1/-1->2->1 [21] 3/-1/-1->2->1 [22] 3/-1/-1->2->1 [23] 3/-1/-1->2->1
+ip-26-0-172-142:880682:881020 [3] NCCL INFO Trees [0] 4/-1/-1->3->2 [1] 4/-1/-1->3->2 [2] 4/-1/-1->3->2 [3] 4/-1/-1->3->2 [4] 4/-1/-1->3->2 [5] 4/-1/-1->3->2 [6] 4/-1/-1->3->2 [7] 4/-1/-1->3->2 [8] 4/-1/-1->3->2 [9] 4/-1/-1->3->2 [10] 4/-1/-1->3->2 [11] 4/-1/-1->3->2 [12] 4/-1/-1->3->2 [13] 4/-1/-1->3->2 [14] 4/-1/-1->3->2 [15] 4/-1/-1->3->2 [16] 4/-1/-1->3->2 [17] 4/-1/-1->3->2 [18] 4/-1/-1->3->2 [19] 4/-1/-1->3->2 [20] 4/-1/-1->3->2 [21] 4/-1/-1->3->2 [22] 4/-1/-1->3->2 [23] 4/-1/-1->3->2
+ip-26-0-172-142:880679:881015 [0] NCCL INFO Channel 01/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-142:880681:881019 [2] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-172-142:880679:881015 [0] NCCL INFO Channel 02/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-142:880682:881020 [3] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-172-142:880679:881015 [0] NCCL INFO Channel 03/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-142:880679:881015 [0] NCCL INFO Channel 04/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-142:880679:881015 [0] NCCL INFO Channel 05/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-142:880679:881015 [0] NCCL INFO Channel 06/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-142:880680:881022 [1] NCCL INFO Trees [0] 2/-1/-1->1->0 [1] 2/-1/-1->1->0 [2] 2/-1/-1->1->0 [3] 2/-1/-1->1->0 [4] 2/-1/-1->1->0 [5] 2/-1/-1->1->0 [6] 2/-1/-1->1->0 [7] 2/-1/-1->1->0 [8] 2/-1/-1->1->0 [9] 2/-1/-1->1->0 [10] 2/-1/-1->1->0 [11] 2/-1/-1->1->0 [12] 2/-1/-1->1->0 [13] 2/-1/-1->1->0 [14] 2/-1/-1->1->0 [15] 2/-1/-1->1->0 [16] 2/-1/-1->1->0 [17] 2/-1/-1->1->0 [18] 2/-1/-1->1->0 [19] 2/-1/-1->1->0 [20] 2/-1/-1->1->0 [21] 2/-1/-1->1->0 [22] 2/-1/-1->1->0 [23] 2/-1/-1->1->0
+ip-26-0-172-142:880679:881015 [0] NCCL INFO Channel 07/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-142:880686:881021 [7] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-172-142:880679:881015 [0] NCCL INFO Channel 08/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-142:880679:881015 [0] NCCL INFO Channel 09/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-142:880679:881015 [0] NCCL INFO Channel 10/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-142:880679:881015 [0] NCCL INFO Channel 11/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-142:880679:881015 [0] NCCL INFO Channel 12/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-142:880680:881022 [1] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-172-142:880679:881015 [0] NCCL INFO Channel 13/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-142:880679:881015 [0] NCCL INFO Channel 14/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-142:880679:881015 [0] NCCL INFO Channel 15/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-142:880679:881015 [0] NCCL INFO Channel 16/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-142:880679:881015 [0] NCCL INFO Channel 17/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-142:880679:881015 [0] NCCL INFO Channel 18/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-142:880679:881015 [0] NCCL INFO Channel 19/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-142:880679:881015 [0] NCCL INFO Channel 20/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-142:880679:881015 [0] NCCL INFO Channel 21/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-142:880679:881015 [0] NCCL INFO Channel 22/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-142:880679:881015 [0] NCCL INFO Channel 23/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-142:880679:881015 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] 1/-1/-1->0->-1 [7] 1/-1/-1->0->-1 [8] 1/-1/-1->0->-1 [9] 1/-1/-1->0->-1 [10] 1/-1/-1->0->-1 [11] 1/-1/-1->0->-1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] 1/-1/-1->0->-1 [19] 1/-1/-1->0->-1 [20] 1/-1/-1->0->-1 [21] 1/-1/-1->0->-1 [22] 1/-1/-1->0->-1 [23] 1/-1/-1->0->-1
+ip-26-0-172-142:880679:881015 [0] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-172-73:926600:926933 [6] NCCL INFO Channel 12/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-73:926598:926938 [4] NCCL INFO Channel 10/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-73:926596:926936 [2] NCCL INFO Channel 10/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-73:926601:926934 [7] NCCL INFO Channel 09/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-73:926599:926932 [5] NCCL INFO Channel 08/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-73:926600:926933 [6] NCCL INFO Channel 13/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-73:926598:926938 [4] NCCL INFO Channel 11/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-73:926596:926936 [2] NCCL INFO Channel 11/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-73:926601:926934 [7] NCCL INFO Channel 10/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-73:926599:926932 [5] NCCL INFO Channel 09/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-73:926600:926933 [6] NCCL INFO Channel 14/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-73:926598:926938 [4] NCCL INFO Channel 12/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-73:926596:926936 [2] NCCL INFO Channel 12/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-73:926601:926934 [7] NCCL INFO Channel 11/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-73:926599:926932 [5] NCCL INFO Channel 10/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-147:618591:618921 [7] NCCL INFO Channel 00/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-147:618590:618925 [6] NCCL INFO Channel 00/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-147:618591:618921 [7] NCCL INFO Channel 01/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-147:618590:618925 [6] NCCL INFO Channel 01/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-147:618591:618921 [7] NCCL INFO Channel 02/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-147:618590:618925 [6] NCCL INFO Channel 02/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-147:618591:618921 [7] NCCL INFO Channel 03/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-147:618591:618921 [7] NCCL INFO Channel 04/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-147:618590:618925 [6] NCCL INFO Channel 03/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-147:618591:618921 [7] NCCL INFO Channel 05/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-147:618590:618925 [6] NCCL INFO Channel 04/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-147:618591:618921 [7] NCCL INFO Channel 06/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-147:618591:618921 [7] NCCL INFO Channel 07/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-147:618591:618921 [7] NCCL INFO Channel 08/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-147:618590:618925 [6] NCCL INFO Channel 05/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-147:618590:618925 [6] NCCL INFO Channel 06/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-147:618590:618925 [6] NCCL INFO Channel 07/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-147:618589:618922 [5] NCCL INFO Channel 00/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-147:618590:618925 [6] NCCL INFO Channel 08/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-147:618589:618922 [5] NCCL INFO Channel 01/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-147:618590:618925 [6] NCCL INFO Channel 09/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-147:618589:618922 [5] NCCL INFO Channel 02/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-147:618585:618926 [1] NCCL INFO Channel 00/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-147:618590:618925 [6] NCCL INFO Channel 10/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-147:618587:618924 [3] NCCL INFO Channel 00/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-147:618591:618921 [7] NCCL INFO Channel 09/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-73:926600:926933 [6] NCCL INFO Channel 15/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-73:926596:926936 [2] NCCL INFO Channel 13/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-73:926598:926938 [4] NCCL INFO Channel 13/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-147:618587:618924 [3] NCCL INFO Channel 01/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-147:618591:618921 [7] NCCL INFO Channel 10/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-73:926601:926934 [7] NCCL INFO Channel 12/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-73:926599:926932 [5] NCCL INFO Channel 11/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-147:618589:618922 [5] NCCL INFO Channel 03/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-73:926600:926933 [6] NCCL INFO Channel 16/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-73:926596:926936 [2] NCCL INFO Channel 14/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-73:926598:926938 [4] NCCL INFO Channel 14/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-147:618587:618924 [3] NCCL INFO Channel 02/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-147:618591:618921 [7] NCCL INFO Channel 11/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-73:926601:926934 [7] NCCL INFO Channel 13/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-73:926599:926932 [5] NCCL INFO Channel 12/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-147:618589:618922 [5] NCCL INFO Channel 04/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-147:618587:618924 [3] NCCL INFO Channel 03/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-147:618588:618920 [4] NCCL INFO Channel 00/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-147:618591:618921 [7] NCCL INFO Channel 12/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-147:618589:618922 [5] NCCL INFO Channel 05/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-73:926600:926933 [6] NCCL INFO Channel 17/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-73:926596:926936 [2] NCCL INFO Channel 15/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-73:926594:926931 [0] NCCL INFO Channel 09/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-73:926598:926938 [4] NCCL INFO Channel 15/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-73:926595:926937 [1] NCCL INFO Channel 09/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-73:926597:926935 [3] NCCL INFO Channel 03/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-73:926601:926934 [7] NCCL INFO Channel 14/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-73:926599:926932 [5] NCCL INFO Channel 13/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-147:618588:618920 [4] NCCL INFO Channel 01/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-147:618587:618924 [3] NCCL INFO Channel 04/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-147:618591:618921 [7] NCCL INFO Channel 13/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-147:618589:618922 [5] NCCL INFO Channel 06/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-73:926600:926933 [6] NCCL INFO Channel 18/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-73:926598:926938 [4] NCCL INFO Channel 16/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-73:926596:926936 [2] NCCL INFO Channel 16/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-73:926594:926931 [0] NCCL INFO Channel 10/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-73:926595:926937 [1] NCCL INFO Channel 10/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-73:926597:926935 [3] NCCL INFO Channel 04/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-73:926599:926932 [5] NCCL INFO Channel 14/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-73:926601:926934 [7] NCCL INFO Channel 15/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-147:618585:618926 [1] NCCL INFO Channel 01/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-147:618588:618920 [4] NCCL INFO Channel 02/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-147:618587:618924 [3] NCCL INFO Channel 05/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-147:618591:618921 [7] NCCL INFO Channel 14/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-147:618589:618922 [5] NCCL INFO Channel 07/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-73:926598:926938 [4] NCCL INFO Channel 17/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-73:926600:926933 [6] NCCL INFO Channel 19/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-73:926596:926936 [2] NCCL INFO Channel 17/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-73:926595:926937 [1] NCCL INFO Channel 11/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-73:926594:926931 [0] NCCL INFO Channel 11/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-73:926597:926935 [3] NCCL INFO Channel 05/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-73:926601:926934 [7] NCCL INFO Channel 16/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-73:926599:926932 [5] NCCL INFO Channel 15/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-147:618585:618926 [1] NCCL INFO Channel 02/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-147:618587:618924 [3] NCCL INFO Channel 06/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-147:618588:618920 [4] NCCL INFO Channel 03/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-147:618591:618921 [7] NCCL INFO Channel 15/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-147:618590:618925 [6] NCCL INFO Channel 11/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-147:618589:618922 [5] NCCL INFO Channel 08/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-73:926600:926933 [6] NCCL INFO Channel 20/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-73:926598:926938 [4] NCCL INFO Channel 18/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-73:926596:926936 [2] NCCL INFO Channel 18/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-73:926597:926935 [3] NCCL INFO Channel 06/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-73:926594:926931 [0] NCCL INFO Channel 12/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-73:926595:926937 [1] NCCL INFO Channel 12/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-73:926601:926934 [7] NCCL INFO Channel 17/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-73:926599:926932 [5] NCCL INFO Channel 16/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-147:618585:618926 [1] NCCL INFO Channel 03/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-147:618588:618920 [4] NCCL INFO Channel 04/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-147:618587:618924 [3] NCCL INFO Channel 07/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-147:618591:618921 [7] NCCL INFO Channel 16/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-147:618589:618922 [5] NCCL INFO Channel 09/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-147:618590:618925 [6] NCCL INFO Channel 12/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-73:926600:926933 [6] NCCL INFO Channel 21/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-73:926598:926938 [4] NCCL INFO Channel 19/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-73:926596:926936 [2] NCCL INFO Channel 19/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-73:926594:926931 [0] NCCL INFO Channel 13/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-73:926597:926935 [3] NCCL INFO Channel 07/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-73:926595:926937 [1] NCCL INFO Channel 13/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-73:926601:926934 [7] NCCL INFO Channel 18/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-73:926599:926932 [5] NCCL INFO Channel 17/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-147:618584:618919 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-147:618585:618926 [1] NCCL INFO Channel 04/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-147:618588:618920 [4] NCCL INFO Channel 05/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-147:618587:618924 [3] NCCL INFO Channel 08/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-147:618591:618921 [7] NCCL INFO Channel 17/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-147:618589:618922 [5] NCCL INFO Channel 10/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-147:618590:618925 [6] NCCL INFO Channel 13/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-73:926600:926933 [6] NCCL INFO Channel 22/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-73:926598:926938 [4] NCCL INFO Channel 20/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-73:926596:926936 [2] NCCL INFO Channel 20/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-73:926594:926931 [0] NCCL INFO Channel 14/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-73:926597:926935 [3] NCCL INFO Channel 08/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-73:926595:926937 [1] NCCL INFO Channel 14/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-73:926601:926934 [7] NCCL INFO Channel 19/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-73:926599:926932 [5] NCCL INFO Channel 18/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-147:618584:618919 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-147:618585:618926 [1] NCCL INFO Channel 05/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-147:618588:618920 [4] NCCL INFO Channel 06/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-147:618587:618924 [3] NCCL INFO Channel 09/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-147:618591:618921 [7] NCCL INFO Channel 18/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-147:618589:618922 [5] NCCL INFO Channel 11/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-147:618590:618925 [6] NCCL INFO Channel 14/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-173-7:811255:811595 [7] NCCL INFO Channel 00/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-173-7:811252:811592 [4] NCCL INFO Channel 00/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-173-7:811254:811594 [6] NCCL INFO Channel 00/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-173-7:811255:811595 [7] NCCL INFO Channel 01/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-173-7:811248:811591 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-173-7:811252:811592 [4] NCCL INFO Channel 01/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-173-7:811254:811594 [6] NCCL INFO Channel 01/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-73:926598:926938 [4] NCCL INFO Channel 21/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-73:926600:926933 [6] NCCL INFO Channel 23/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-173-7:811255:811595 [7] NCCL INFO Channel 02/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-173-7:811248:811591 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-173-7:811252:811592 [4] NCCL INFO Channel 02/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-173-7:811254:811594 [6] NCCL INFO Channel 02/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-173-7:811255:811595 [7] NCCL INFO Channel 03/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-173-7:811248:811591 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-173-7:811252:811592 [4] NCCL INFO Channel 03/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-73:926597:926935 [3] NCCL INFO Channel 09/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-73:926594:926931 [0] NCCL INFO Channel 15/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-73:926596:926936 [2] NCCL INFO Channel 21/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-173-7:811254:811594 [6] NCCL INFO Channel 03/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-73:926601:926934 [7] NCCL INFO Channel 20/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-73:926595:926937 [1] NCCL INFO Channel 15/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-173-7:811255:811595 [7] NCCL INFO Channel 04/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-73:926599:926932 [5] NCCL INFO Channel 19/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-173-7:811248:811591 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-173-7:811252:811592 [4] NCCL INFO Channel 04/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-173-7:811254:811594 [6] NCCL INFO Channel 04/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-173-7:811255:811595 [7] NCCL INFO Channel 05/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-173-7:811248:811591 [0] NCCL INFO Channel 04/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-173-7:811252:811592 [4] NCCL INFO Channel 05/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-173-7:811254:811594 [6] NCCL INFO Channel 05/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-173-7:811255:811595 [7] NCCL INFO Channel 06/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-173-7:811248:811591 [0] NCCL INFO Channel 05/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-173-7:811252:811592 [4] NCCL INFO Channel 06/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-173-7:811254:811594 [6] NCCL INFO Channel 06/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-173-7:811253:811593 [5] NCCL INFO Channel 00/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-173-7:811255:811595 [7] NCCL INFO Channel 07/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-173-7:811248:811591 [0] NCCL INFO Channel 06/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-173-7:811249:811596 [1] NCCL INFO Channel 00/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-173-7:811251:811598 [3] NCCL INFO Channel 00/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-173-7:811252:811592 [4] NCCL INFO Channel 07/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-173-7:811254:811594 [6] NCCL INFO Channel 07/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-173-7:811250:811597 [2] NCCL INFO Channel 00/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-173-7:811253:811593 [5] NCCL INFO Channel 01/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-173-7:811255:811595 [7] NCCL INFO Channel 08/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-173-7:811248:811591 [0] NCCL INFO Channel 07/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-173-7:811249:811596 [1] NCCL INFO Channel 01/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-173-7:811252:811592 [4] NCCL INFO Channel 08/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-173-7:811251:811598 [3] NCCL INFO Channel 01/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-173-7:811253:811593 [5] NCCL INFO Channel 02/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-173-7:811254:811594 [6] NCCL INFO Channel 08/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-173-7:811250:811597 [2] NCCL INFO Channel 01/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-173-7:811255:811595 [7] NCCL INFO Channel 09/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-173-7:811248:811591 [0] NCCL INFO Channel 08/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-173-7:811249:811596 [1] NCCL INFO Channel 02/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-173-7:811252:811592 [4] NCCL INFO Channel 09/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-173-7:811251:811598 [3] NCCL INFO Channel 02/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-173-7:811253:811593 [5] NCCL INFO Channel 03/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-173-7:811254:811594 [6] NCCL INFO Channel 09/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-173-7:811250:811597 [2] NCCL INFO Channel 02/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-147:618584:618919 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-147:618585:618926 [1] NCCL INFO Channel 06/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-173-7:811255:811595 [7] NCCL INFO Channel 10/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-147:618588:618920 [4] NCCL INFO Channel 07/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-147:618587:618924 [3] NCCL INFO Channel 10/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-147:618591:618921 [7] NCCL INFO Channel 19/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-147:618589:618922 [5] NCCL INFO Channel 12/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-147:618590:618925 [6] NCCL INFO Channel 15/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-173-7:811248:811591 [0] NCCL INFO Channel 09/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-173-7:811249:811596 [1] NCCL INFO Channel 03/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-147:618586:618923 [2] NCCL INFO Channel 00/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-173-7:811252:811592 [4] NCCL INFO Channel 10/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-173-7:811251:811598 [3] NCCL INFO Channel 03/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-173-7:811253:811593 [5] NCCL INFO Channel 04/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-173-7:811254:811594 [6] NCCL INFO Channel 10/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-73:926597:926935 [3] NCCL INFO Channel 10/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-173-7:811250:811597 [2] NCCL INFO Channel 03/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-73:926598:926938 [4] NCCL INFO Channel 22/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-73:926595:926937 [1] NCCL INFO Channel 16/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-73:926594:926931 [0] NCCL INFO Channel 16/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-73:926601:926934 [7] NCCL INFO Channel 21/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-73:926596:926936 [2] NCCL INFO Channel 22/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-73:926599:926932 [5] NCCL INFO Channel 20/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-173-7:811248:811591 [0] NCCL INFO Channel 10/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-147:618584:618919 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-173-7:811255:811595 [7] NCCL INFO Channel 11/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-147:618585:618926 [1] NCCL INFO Channel 07/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-147:618587:618924 [3] NCCL INFO Channel 11/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-147:618588:618920 [4] NCCL INFO Channel 08/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-147:618591:618921 [7] NCCL INFO Channel 20/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-173-7:811249:811596 [1] NCCL INFO Channel 04/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-147:618590:618925 [6] NCCL INFO Channel 16/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-147:618589:618922 [5] NCCL INFO Channel 13/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-147:618586:618923 [2] NCCL INFO Channel 01/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-173-7:811252:811592 [4] NCCL INFO Channel 11/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-173-7:811251:811598 [3] NCCL INFO Channel 04/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-73:926597:926935 [3] NCCL INFO Channel 11/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-73:926598:926938 [4] NCCL INFO Channel 23/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-73:926595:926937 [1] NCCL INFO Channel 17/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-57:942615:942948 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-172-57:942615:942948 [4] NCCL INFO NVLS multicast support is available on dev 4
+ip-26-0-172-73:926594:926931 [0] NCCL INFO Channel 17/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-73:926601:926934 [7] NCCL INFO Channel 22/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-173-7:811250:811597 [2] NCCL INFO Channel 04/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-73:926596:926936 [2] NCCL INFO Channel 23/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-73:926599:926932 [5] NCCL INFO Channel 21/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-173-7:811255:811595 [7] NCCL INFO Channel 12/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-173-7:811251:811598 [3] NCCL INFO Channel 05/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-173-7:811253:811593 [5] NCCL INFO Channel 05/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-173-7:811250:811597 [2] NCCL INFO Channel 05/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-173-7:811255:811595 [7] NCCL INFO Channel 13/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-173-7:811251:811598 [3] NCCL INFO Channel 06/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-173-7:811253:811593 [5] NCCL INFO Channel 06/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-73:926597:926935 [3] NCCL INFO Channel 12/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-173-7:811250:811597 [2] NCCL INFO Channel 06/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-73:926595:926937 [1] NCCL INFO Channel 18/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-73:926601:926934 [7] NCCL INFO Channel 23/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-73:926594:926931 [0] NCCL INFO Channel 18/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-73:926599:926932 [5] NCCL INFO Channel 22/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-173-7:811248:811591 [0] NCCL INFO Channel 11/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-173-7:811255:811595 [7] NCCL INFO Channel 14/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-147:618589:618922 [5] NCCL INFO Channel 14/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-173-7:811251:811598 [3] NCCL INFO Channel 07/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-147:618588:618920 [4] NCCL INFO Channel 09/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-147:618591:618921 [7] NCCL INFO Channel 21/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-173-7:811253:811593 [5] NCCL INFO Channel 07/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-147:618586:618923 [2] NCCL INFO Channel 02/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-73:926597:926935 [3] NCCL INFO Channel 13/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-73:926595:926937 [1] NCCL INFO Channel 19/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-73:926594:926931 [0] NCCL INFO Channel 19/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-173-7:811250:811597 [2] NCCL INFO Channel 07/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-73:926599:926932 [5] NCCL INFO Channel 23/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-173-7:811248:811591 [0] NCCL INFO Channel 12/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-147:618589:618922 [5] NCCL INFO Channel 15/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-147:618588:618920 [4] NCCL INFO Channel 10/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-173-7:811255:811595 [7] NCCL INFO Channel 15/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-147:618591:618921 [7] NCCL INFO Channel 22/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-147:618584:618919 [0] NCCL INFO Channel 04/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-147:618585:618926 [1] NCCL INFO Channel 08/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-173-7:811251:811598 [3] NCCL INFO Channel 08/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-147:618586:618923 [2] NCCL INFO Channel 03/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-173-7:811253:811593 [5] NCCL INFO Channel 08/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-73:926597:926935 [3] NCCL INFO Channel 14/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-73:926595:926937 [1] NCCL INFO Channel 20/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-73:926594:926931 [0] NCCL INFO Channel 20/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-173-7:811250:811597 [2] NCCL INFO Channel 08/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-147:618590:618925 [6] NCCL INFO Channel 17/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-173-7:811248:811591 [0] NCCL INFO Channel 13/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-173-7:811255:811595 [7] NCCL INFO Channel 16/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-173-7:811253:811593 [5] NCCL INFO Channel 09/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-173-7:811251:811598 [3] NCCL INFO Channel 09/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-173-7:811250:811597 [2] NCCL INFO Channel 09/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-147:618588:618920 [4] NCCL INFO Channel 11/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-173-7:811248:811591 [0] NCCL INFO Channel 14/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-147:618589:618922 [5] NCCL INFO Channel 16/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-147:618591:618921 [7] NCCL INFO Channel 23/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-147:618585:618926 [1] NCCL INFO Channel 09/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-147:618584:618919 [0] NCCL INFO Channel 05/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-147:618586:618923 [2] NCCL INFO Channel 04/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-147:618590:618925 [6] NCCL INFO Channel 18/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-173-7:811255:811595 [7] NCCL INFO Channel 17/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-173-7:811253:811593 [5] NCCL INFO Channel 10/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-173-7:811251:811598 [3] NCCL INFO Channel 10/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-173-7:811250:811597 [2] NCCL INFO Channel 10/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-173-7:811248:811591 [0] NCCL INFO Channel 15/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-147:618588:618920 [4] NCCL INFO Channel 12/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-173-7:811253:811593 [5] NCCL INFO Channel 11/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-173-7:811255:811595 [7] NCCL INFO Channel 18/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-147:618589:618922 [5] NCCL INFO Channel 17/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-173-7:811251:811598 [3] NCCL INFO Channel 11/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-147:618585:618926 [1] NCCL INFO Channel 10/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-147:618584:618919 [0] NCCL INFO Channel 06/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-147:618586:618923 [2] NCCL INFO Channel 05/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-173-7:811250:811597 [2] NCCL INFO Channel 11/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-147:618590:618925 [6] NCCL INFO Channel 19/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-147:618587:618924 [3] NCCL INFO Channel 12/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-173-7:811248:811591 [0] NCCL INFO Channel 16/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-173-7:811254:811594 [6] NCCL INFO Channel 11/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-73:926595:926937 [1] NCCL INFO Channel 21/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-73:926597:926935 [3] NCCL INFO Channel 15/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-73:926594:926931 [0] NCCL INFO Channel 21/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-147:618588:618920 [4] NCCL INFO Channel 13/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-173-7:811255:811595 [7] NCCL INFO Channel 19/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-173-7:811253:811593 [5] NCCL INFO Channel 12/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-173-7:811251:811598 [3] NCCL INFO Channel 12/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-147:618589:618922 [5] NCCL INFO Channel 18/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-147:618585:618926 [1] NCCL INFO Channel 11/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-147:618584:618919 [0] NCCL INFO Channel 07/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-173-7:811249:811596 [1] NCCL INFO Channel 05/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-147:618586:618923 [2] NCCL INFO Channel 06/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-173-7:811252:811592 [4] NCCL INFO Channel 12/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-173-7:811250:811597 [2] NCCL INFO Channel 12/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-147:618590:618925 [6] NCCL INFO Channel 20/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-147:618587:618924 [3] NCCL INFO Channel 13/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-173-7:811248:811591 [0] NCCL INFO Channel 17/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-173-7:811254:811594 [6] NCCL INFO Channel 12/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-173-7:811255:811595 [7] NCCL INFO Channel 20/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-147:618588:618920 [4] NCCL INFO Channel 14/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-173-7:811253:811593 [5] NCCL INFO Channel 13/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-173-7:811251:811598 [3] NCCL INFO Channel 13/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-147:618585:618926 [1] NCCL INFO Channel 12/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-147:618589:618922 [5] NCCL INFO Channel 19/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-147:618584:618919 [0] NCCL INFO Channel 08/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-147:618586:618923 [2] NCCL INFO Channel 07/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-147:618587:618924 [3] NCCL INFO Channel 14/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-173-7:811249:811596 [1] NCCL INFO Channel 06/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-147:618590:618925 [6] NCCL INFO Channel 21/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-173-7:811252:811592 [4] NCCL INFO Channel 13/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-173-7:811250:811597 [2] NCCL INFO Channel 13/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-173-7:811248:811591 [0] NCCL INFO Channel 18/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-173-7:811254:811594 [6] NCCL INFO Channel 13/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-173-7:811255:811595 [7] NCCL INFO Channel 21/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-147:618588:618920 [4] NCCL INFO Channel 15/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-173-7:811253:811593 [5] NCCL INFO Channel 14/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-147:618585:618926 [1] NCCL INFO Channel 13/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-147:618589:618922 [5] NCCL INFO Channel 20/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-147:618584:618919 [0] NCCL INFO Channel 09/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-173-7:811251:811598 [3] NCCL INFO Channel 14/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-147:618586:618923 [2] NCCL INFO Channel 08/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-147:618587:618924 [3] NCCL INFO Channel 15/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-147:618590:618925 [6] NCCL INFO Channel 22/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-173-7:811249:811596 [1] NCCL INFO Channel 07/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-173-7:811252:811592 [4] NCCL INFO Channel 14/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-173-7:811248:811591 [0] NCCL INFO Channel 19/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-173-7:811250:811597 [2] NCCL INFO Channel 14/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-173-7:811254:811594 [6] NCCL INFO Channel 14/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-73:926597:926935 [3] NCCL INFO Channel 16/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-147:618588:618920 [4] NCCL INFO Channel 16/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-173-7:811255:811595 [7] NCCL INFO Channel 22/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-147:618589:618922 [5] NCCL INFO Channel 21/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-173-7:811253:811593 [5] NCCL INFO Channel 15/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-147:618585:618926 [1] NCCL INFO Channel 14/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-147:618584:618919 [0] NCCL INFO Channel 10/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-147:618586:618923 [2] NCCL INFO Channel 09/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-173-7:811251:811598 [3] NCCL INFO Channel 15/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-147:618587:618924 [3] NCCL INFO Channel 16/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-147:618590:618925 [6] NCCL INFO Channel 23/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-173-7:811249:811596 [1] NCCL INFO Channel 08/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-173-7:811248:811591 [0] NCCL INFO Channel 20/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-173-7:811252:811592 [4] NCCL INFO Channel 15/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-73:926595:926937 [1] NCCL INFO Channel 22/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-173-7:811250:811597 [2] NCCL INFO Channel 15/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-73:926594:926931 [0] NCCL INFO Channel 22/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-173-7:811254:811594 [6] NCCL INFO Channel 15/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-142:880686:881021 [7] NCCL INFO Channel 00/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-142:880680:881022 [1] NCCL INFO Channel 00/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-147:618588:618920 [4] NCCL INFO Channel 17/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-147:618589:618922 [5] NCCL INFO Channel 22/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-147:618585:618926 [1] NCCL INFO Channel 15/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-173-7:811255:811595 [7] NCCL INFO Channel 23/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-147:618584:618919 [0] NCCL INFO Channel 11/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-147:618586:618923 [2] NCCL INFO Channel 10/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-173-7:811253:811593 [5] NCCL INFO Channel 16/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-147:618587:618924 [3] NCCL INFO Channel 17/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-173-7:811251:811598 [3] NCCL INFO Channel 16/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-142:880681:881019 [2] NCCL INFO Channel 00/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-173-7:811249:811596 [1] NCCL INFO Channel 09/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-173-7:811252:811592 [4] NCCL INFO Channel 16/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-173-7:811248:811591 [0] NCCL INFO Channel 21/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-173-7:811250:811597 [2] NCCL INFO Channel 16/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-142:880682:881020 [3] NCCL INFO Channel 00/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-173-7:811254:811594 [6] NCCL INFO Channel 16/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-142:880686:881021 [7] NCCL INFO Channel 01/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-142:880680:881022 [1] NCCL INFO Channel 01/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-147:618588:618920 [4] NCCL INFO Channel 18/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-147:618589:618922 [5] NCCL INFO Channel 23/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-147:618585:618926 [1] NCCL INFO Channel 16/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-147:618584:618919 [0] NCCL INFO Channel 12/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-147:618586:618923 [2] NCCL INFO Channel 11/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-147:618587:618924 [3] NCCL INFO Channel 18/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-142:880681:881019 [2] NCCL INFO Channel 01/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-173-7:811253:811593 [5] NCCL INFO Channel 17/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-173-7:811251:811598 [3] NCCL INFO Channel 17/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-173-7:811249:811596 [1] NCCL INFO Channel 10/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-173-7:811252:811592 [4] NCCL INFO Channel 17/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-173-7:811248:811591 [0] NCCL INFO Channel 22/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-173-7:811250:811597 [2] NCCL INFO Channel 17/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-142:880682:881020 [3] NCCL INFO Channel 01/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-142:880686:881021 [7] NCCL INFO Channel 02/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-173-7:811254:811594 [6] NCCL INFO Channel 17/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-57:942618:942951 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-172-142:880680:881022 [1] NCCL INFO Channel 02/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-147:618588:618920 [4] NCCL INFO Channel 19/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-73:926597:926935 [3] NCCL INFO Channel 17/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-57:942618:942951 [7] NCCL INFO NVLS multicast support is available on dev 7
+ip-26-0-172-147:618585:618926 [1] NCCL INFO Channel 17/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-147:618584:618919 [0] NCCL INFO Channel 13/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-147:618586:618923 [2] NCCL INFO Channel 12/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-147:618587:618924 [3] NCCL INFO Channel 19/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-142:880685:881018 [6] NCCL INFO Channel 00/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-142:880681:881019 [2] NCCL INFO Channel 02/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-173-7:811253:811593 [5] NCCL INFO Channel 18/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-173-7:811251:811598 [3] NCCL INFO Channel 18/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-173-7:811249:811596 [1] NCCL INFO Channel 11/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-147:618588:618920 [4] NCCL INFO Channel 20/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-173-7:811252:811592 [4] NCCL INFO Channel 18/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-173-7:811250:811597 [2] NCCL INFO Channel 18/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-173-7:811248:811591 [0] NCCL INFO Channel 23/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-173-7:811254:811594 [6] NCCL INFO Channel 18/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-142:880682:881020 [3] NCCL INFO Channel 02/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-142:880686:881021 [7] NCCL INFO Channel 03/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-73:926595:926937 [1] NCCL INFO Channel 23/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-73:926594:926931 [0] NCCL INFO Channel 23/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-142:880680:881022 [1] NCCL INFO Channel 03/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-142:880685:881018 [6] NCCL INFO Channel 01/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-173-7:811253:811593 [5] NCCL INFO Channel 19/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-142:880681:881019 [2] NCCL INFO Channel 03/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-173-7:811251:811598 [3] NCCL INFO Channel 19/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-173-7:811249:811596 [1] NCCL INFO Channel 12/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-173-7:811252:811592 [4] NCCL INFO Channel 19/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-173-7:811250:811597 [2] NCCL INFO Channel 19/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-173-7:811254:811594 [6] NCCL INFO Channel 19/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-142:880686:881021 [7] NCCL INFO Channel 04/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-142:880682:881020 [3] NCCL INFO Channel 03/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-147:618585:618926 [1] NCCL INFO Channel 18/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-142:880680:881022 [1] NCCL INFO Channel 04/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-147:618584:618919 [0] NCCL INFO Channel 14/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-147:618586:618923 [2] NCCL INFO Channel 13/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-147:618587:618924 [3] NCCL INFO Channel 20/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-142:880684:881017 [5] NCCL INFO Channel 00/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-173-7:811253:811593 [5] NCCL INFO Channel 20/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-173-7:811251:811598 [3] NCCL INFO Channel 20/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-142:880685:881018 [6] NCCL INFO Channel 02/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-173-7:811249:811596 [1] NCCL INFO Channel 13/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-173-7:811252:811592 [4] NCCL INFO Channel 20/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-173-7:811250:811597 [2] NCCL INFO Channel 20/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-142:880681:881019 [2] NCCL INFO Channel 04/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-173-7:811254:811594 [6] NCCL INFO Channel 20/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-142:880686:881021 [7] NCCL INFO Channel 05/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-142:880682:881020 [3] NCCL INFO Channel 04/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-142:880680:881022 [1] NCCL INFO Channel 05/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-173-7:811253:811593 [5] NCCL INFO Channel 21/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-173-7:811251:811598 [3] NCCL INFO Channel 21/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-173-7:811249:811596 [1] NCCL INFO Channel 14/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-147:618588:618920 [4] NCCL INFO Channel 21/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-173-7:811252:811592 [4] NCCL INFO Channel 21/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-142:880684:881017 [5] NCCL INFO Channel 01/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-173-7:811250:811597 [2] NCCL INFO Channel 21/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-173-7:811254:811594 [6] NCCL INFO Channel 21/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-142:880685:881018 [6] NCCL INFO Channel 03/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-142:880681:881019 [2] NCCL INFO Channel 05/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-142:880679:881015 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-73:926597:926935 [3] NCCL INFO Channel 18/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-142:880686:881021 [7] NCCL INFO Channel 06/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-173-7:811253:811593 [5] NCCL INFO Channel 22/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-173-7:811249:811596 [1] NCCL INFO Channel 15/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-142:880682:881020 [3] NCCL INFO Channel 05/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-142:880680:881022 [1] NCCL INFO Channel 06/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-173-7:811251:811598 [3] NCCL INFO Channel 22/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-173-7:811252:811592 [4] NCCL INFO Channel 22/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-173-7:811250:811597 [2] NCCL INFO Channel 22/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-173-7:811254:811594 [6] NCCL INFO Channel 22/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-57:942613:942950 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-172-57:942613:942950 [2] NCCL INFO NVLS multicast support is available on dev 2
+ip-26-0-172-142:880684:881017 [5] NCCL INFO Channel 02/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-142:880685:881018 [6] NCCL INFO Channel 04/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-57:942613:942950 [2] NCCL INFO Trees [0] 3/-1/-1->2->1 [1] 3/-1/-1->2->1 [2] 3/-1/-1->2->1 [3] 3/-1/-1->2->1 [4] 3/-1/-1->2->1 [5] 3/-1/-1->2->1 [6] 3/-1/-1->2->1 [7] 3/-1/-1->2->1 [8] 3/-1/-1->2->1 [9] 3/-1/-1->2->1 [10] 3/-1/-1->2->1 [11] 3/-1/-1->2->1 [12] 3/-1/-1->2->1 [13] 3/-1/-1->2->1 [14] 3/-1/-1->2->1 [15] 3/-1/-1->2->1 [16] 3/-1/-1->2->1 [17] 3/-1/-1->2->1 [18] 3/-1/-1->2->1 [19] 3/-1/-1->2->1 [20] 3/-1/-1->2->1 [21] 3/-1/-1->2->1 [22] 3/-1/-1->2->1 [23] 3/-1/-1->2->1
+ip-26-0-172-147:618585:618926 [1] NCCL INFO Channel 19/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-57:942613:942950 [2] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-172-147:618584:618919 [0] NCCL INFO Channel 15/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-57:942611:942947 [0] NCCL INFO Channel 00/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-57:942614:942953 [3] NCCL INFO Trees [0] 4/-1/-1->3->2 [1] 4/-1/-1->3->2 [2] 4/-1/-1->3->2 [3] 4/-1/-1->3->2 [4] 4/-1/-1->3->2 [5] 4/-1/-1->3->2 [6] 4/-1/-1->3->2 [7] 4/-1/-1->3->2 [8] 4/-1/-1->3->2 [9] 4/-1/-1->3->2 [10] 4/-1/-1->3->2 [11] 4/-1/-1->3->2 [12] 4/-1/-1->3->2 [13] 4/-1/-1->3->2 [14] 4/-1/-1->3->2 [15] 4/-1/-1->3->2 [16] 4/-1/-1->3->2 [17] 4/-1/-1->3->2 [18] 4/-1/-1->3->2 [19] 4/-1/-1->3->2 [20] 4/-1/-1->3->2 [21] 4/-1/-1->3->2 [22] 4/-1/-1->3->2 [23] 4/-1/-1->3->2
+ip-26-0-172-57:942611:942947 [0] NCCL INFO Channel 01/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-57:942614:942953 [3] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-172-57:942611:942947 [0] NCCL INFO Channel 02/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-57:942611:942947 [0] NCCL INFO Channel 03/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-57:942611:942947 [0] NCCL INFO Channel 04/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-57:942611:942947 [0] NCCL INFO Channel 05/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-142:880681:881019 [2] NCCL INFO Channel 06/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-57:942611:942947 [0] NCCL INFO Channel 06/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-57:942611:942947 [0] NCCL INFO Channel 07/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-57:942611:942947 [0] NCCL INFO Channel 08/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-57:942615:942948 [4] NCCL INFO Trees [0] 5/-1/-1->4->3 [1] 5/-1/-1->4->3 [2] 5/-1/-1->4->3 [3] 5/-1/-1->4->3 [4] 5/-1/-1->4->3 [5] 5/-1/-1->4->3 [6] 5/-1/-1->4->3 [7] 5/-1/-1->4->3 [8] 5/-1/-1->4->3 [9] 5/-1/-1->4->3 [10] 5/-1/-1->4->3 [11] 5/-1/-1->4->3 [12] 5/-1/-1->4->3 [13] 5/-1/-1->4->3 [14] 5/-1/-1->4->3 [15] 5/-1/-1->4->3 [16] 5/-1/-1->4->3 [17] 5/-1/-1->4->3 [18] 5/-1/-1->4->3 [19] 5/-1/-1->4->3 [20] 5/-1/-1->4->3 [21] 5/-1/-1->4->3 [22] 5/-1/-1->4->3 [23] 5/-1/-1->4->3
+ip-26-0-172-57:942611:942947 [0] NCCL INFO Channel 09/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-147:618586:618923 [2] NCCL INFO Channel 14/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-57:942617:942949 [6] NCCL INFO Trees [0] 7/-1/-1->6->5 [1] 7/-1/-1->6->5 [2] 7/-1/-1->6->5 [3] 7/-1/-1->6->5 [4] 7/-1/-1->6->5 [5] 7/-1/-1->6->5 [6] 7/-1/-1->6->5 [7] 7/-1/-1->6->5 [8] 7/-1/-1->6->5 [9] 7/-1/-1->6->5 [10] 7/-1/-1->6->5 [11] 7/-1/-1->6->5 [12] 7/-1/-1->6->5 [13] 7/-1/-1->6->5 [14] 7/-1/-1->6->5 [15] 7/-1/-1->6->5 [16] 7/-1/-1->6->5 [17] 7/-1/-1->6->5 [18] 7/-1/-1->6->5 [19] 7/-1/-1->6->5 [20] 7/-1/-1->6->5 [21] 7/-1/-1->6->5 [22] 7/-1/-1->6->5 [23] 7/-1/-1->6->5
+ip-26-0-172-57:942618:942951 [7] NCCL INFO Trees [0] -1/-1/-1->7->6 [1] -1/-1/-1->7->6 [2] -1/-1/-1->7->6 [3] -1/-1/-1->7->6 [4] -1/-1/-1->7->6 [5] -1/-1/-1->7->6 [6] -1/-1/-1->7->6 [7] -1/-1/-1->7->6 [8] -1/-1/-1->7->6 [9] -1/-1/-1->7->6 [10] -1/-1/-1->7->6 [11] -1/-1/-1->7->6 [12] -1/-1/-1->7->6 [13] -1/-1/-1->7->6 [14] -1/-1/-1->7->6 [15] -1/-1/-1->7->6 [16] -1/-1/-1->7->6 [17] -1/-1/-1->7->6 [18] -1/-1/-1->7->6 [19] -1/-1/-1->7->6 [20] -1/-1/-1->7->6 [21] -1/-1/-1->7->6 [22] -1/-1/-1->7->6 [23] -1/-1/-1->7->6
+ip-26-0-172-147:618587:618924 [3] NCCL INFO Channel 21/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-57:942615:942948 [4] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-172-57:942611:942947 [0] NCCL INFO Channel 10/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-57:942617:942949 [6] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-172-57:942611:942947 [0] NCCL INFO Channel 11/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-57:942618:942951 [7] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-172-57:942611:942947 [0] NCCL INFO Channel 12/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-57:942611:942947 [0] NCCL INFO Channel 13/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-57:942611:942947 [0] NCCL INFO Channel 14/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-57:942612:942954 [1] NCCL INFO Trees [0] 2/-1/-1->1->0 [1] 2/-1/-1->1->0 [2] 2/-1/-1->1->0 [3] 2/-1/-1->1->0 [4] 2/-1/-1->1->0 [5] 2/-1/-1->1->0 [6] 2/-1/-1->1->0 [7] 2/-1/-1->1->0 [8] 2/-1/-1->1->0 [9] 2/-1/-1->1->0 [10] 2/-1/-1->1->0 [11] 2/-1/-1->1->0 [12] 2/-1/-1->1->0 [13] 2/-1/-1->1->0 [14] 2/-1/-1->1->0 [15] 2/-1/-1->1->0 [16] 2/-1/-1->1->0 [17] 2/-1/-1->1->0 [18] 2/-1/-1->1->0 [19] 2/-1/-1->1->0 [20] 2/-1/-1->1->0 [21] 2/-1/-1->1->0 [22] 2/-1/-1->1->0 [23] 2/-1/-1->1->0
+ip-26-0-172-57:942611:942947 [0] NCCL INFO Channel 15/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-57:942611:942947 [0] NCCL INFO Channel 16/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-57:942611:942947 [0] NCCL INFO Channel 17/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-142:880679:881015 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-57:942611:942947 [0] NCCL INFO Channel 18/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-57:942611:942947 [0] NCCL INFO Channel 19/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-57:942616:942952 [5] NCCL INFO Trees [0] 6/-1/-1->5->4 [1] 6/-1/-1->5->4 [2] 6/-1/-1->5->4 [3] 6/-1/-1->5->4 [4] 6/-1/-1->5->4 [5] 6/-1/-1->5->4 [6] 6/-1/-1->5->4 [7] 6/-1/-1->5->4 [8] 6/-1/-1->5->4 [9] 6/-1/-1->5->4 [10] 6/-1/-1->5->4 [11] 6/-1/-1->5->4 [12] 6/-1/-1->5->4 [13] 6/-1/-1->5->4 [14] 6/-1/-1->5->4 [15] 6/-1/-1->5->4 [16] 6/-1/-1->5->4 [17] 6/-1/-1->5->4 [18] 6/-1/-1->5->4 [19] 6/-1/-1->5->4 [20] 6/-1/-1->5->4 [21] 6/-1/-1->5->4 [22] 6/-1/-1->5->4 [23] 6/-1/-1->5->4
+ip-26-0-172-57:942611:942947 [0] NCCL INFO Channel 20/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-57:942612:942954 [1] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-172-57:942611:942947 [0] NCCL INFO Channel 21/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-57:942611:942947 [0] NCCL INFO Channel 22/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-57:942611:942947 [0] NCCL INFO Channel 23/24 :    0   1   2   3   4   5   6   7
+ip-26-0-172-57:942616:942952 [5] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-172-57:942611:942947 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] 1/-1/-1->0->-1 [3] 1/-1/-1->0->-1 [4] 1/-1/-1->0->-1 [5] 1/-1/-1->0->-1 [6] 1/-1/-1->0->-1 [7] 1/-1/-1->0->-1 [8] 1/-1/-1->0->-1 [9] 1/-1/-1->0->-1 [10] 1/-1/-1->0->-1 [11] 1/-1/-1->0->-1 [12] 1/-1/-1->0->-1 [13] 1/-1/-1->0->-1 [14] 1/-1/-1->0->-1 [15] 1/-1/-1->0->-1 [16] 1/-1/-1->0->-1 [17] 1/-1/-1->0->-1 [18] 1/-1/-1->0->-1 [19] 1/-1/-1->0->-1 [20] 1/-1/-1->0->-1 [21] 1/-1/-1->0->-1 [22] 1/-1/-1->0->-1 [23] 1/-1/-1->0->-1
+ip-26-0-172-57:942611:942947 [0] NCCL INFO P2P Chunksize set to 524288
+ip-26-0-173-7:811253:811593 [5] NCCL INFO Channel 23/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-173-7:811249:811596 [1] NCCL INFO Channel 16/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-173-7:811251:811598 [3] NCCL INFO Channel 23/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-173-7:811252:811592 [4] NCCL INFO Channel 23/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-147:618588:618920 [4] NCCL INFO Channel 22/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-173-7:811250:811597 [2] NCCL INFO Channel 23/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-142:880683:881016 [4] NCCL INFO Channel 00/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-142:880682:881020 [3] NCCL INFO Channel 06/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-142:880686:881021 [7] NCCL INFO Channel 07/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-173-7:811254:811594 [6] NCCL INFO Channel 23/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-142:880680:881022 [1] NCCL INFO Channel 07/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-142:880684:881017 [5] NCCL INFO Channel 03/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-142:880685:881018 [6] NCCL INFO Channel 05/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-142:880681:881019 [2] NCCL INFO Channel 07/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-142:880679:881015 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-142:880683:881016 [4] NCCL INFO Channel 01/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-142:880682:881020 [3] NCCL INFO Channel 07/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-142:880686:881021 [7] NCCL INFO Channel 08/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-142:880680:881022 [1] NCCL INFO Channel 08/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-147:618585:618926 [1] NCCL INFO Channel 20/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-173-7:811249:811596 [1] NCCL INFO Channel 17/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-142:880684:881017 [5] NCCL INFO Channel 04/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-142:880685:881018 [6] NCCL INFO Channel 06/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-142:880681:881019 [2] NCCL INFO Channel 08/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-142:880679:881015 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-142:880683:881016 [4] NCCL INFO Channel 02/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-147:618584:618919 [0] NCCL INFO Channel 16/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-147:618586:618923 [2] NCCL INFO Channel 15/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-142:880686:881021 [7] NCCL INFO Channel 09/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-142:880682:881020 [3] NCCL INFO Channel 08/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-147:618587:618924 [3] NCCL INFO Channel 22/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-142:880680:881022 [1] NCCL INFO Channel 09/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-147:618588:618920 [4] NCCL INFO Channel 23/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-142:880684:881017 [5] NCCL INFO Channel 05/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-142:880685:881018 [6] NCCL INFO Channel 07/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-142:880681:881019 [2] NCCL INFO Channel 09/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-142:880679:881015 [0] NCCL INFO Channel 04/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-142:880683:881016 [4] NCCL INFO Channel 03/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-142:880686:881021 [7] NCCL INFO Channel 10/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-142:880682:881020 [3] NCCL INFO Channel 09/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-142:880680:881022 [1] NCCL INFO Channel 10/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-142:880684:881017 [5] NCCL INFO Channel 06/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-142:880685:881018 [6] NCCL INFO Channel 08/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-147:618585:618926 [1] NCCL INFO Channel 21/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-73:926597:926935 [3] NCCL INFO Channel 19/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-147:618584:618919 [0] NCCL INFO Channel 17/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-147:618586:618923 [2] NCCL INFO Channel 16/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-142:880681:881019 [2] NCCL INFO Channel 10/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-142:880679:881015 [0] NCCL INFO Channel 05/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-147:618587:618924 [3] NCCL INFO Channel 23/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-142:880683:881016 [4] NCCL INFO Channel 04/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-142:880686:881021 [7] NCCL INFO Channel 11/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-142:880682:881020 [3] NCCL INFO Channel 10/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-142:880680:881022 [1] NCCL INFO Channel 11/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-173-7:811249:811596 [1] NCCL INFO Channel 18/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-142:880684:881017 [5] NCCL INFO Channel 07/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-142:880685:881018 [6] NCCL INFO Channel 09/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-142:880679:881015 [0] NCCL INFO Channel 06/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-142:880681:881019 [2] NCCL INFO Channel 11/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-142:880683:881016 [4] NCCL INFO Channel 05/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-142:880686:881021 [7] NCCL INFO Channel 12/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-142:880682:881020 [3] NCCL INFO Channel 11/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-142:880680:881022 [1] NCCL INFO Channel 12/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-142:880684:881017 [5] NCCL INFO Channel 08/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-142:880685:881018 [6] NCCL INFO Channel 10/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-142:880679:881015 [0] NCCL INFO Channel 07/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-142:880681:881019 [2] NCCL INFO Channel 12/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-142:880683:881016 [4] NCCL INFO Channel 06/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-142:880686:881021 [7] NCCL INFO Channel 13/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-142:880682:881020 [3] NCCL INFO Channel 12/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-142:880680:881022 [1] NCCL INFO Channel 13/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-142:880684:881017 [5] NCCL INFO Channel 09/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-142:880685:881018 [6] NCCL INFO Channel 11/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-142:880679:881015 [0] NCCL INFO Channel 08/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-142:880681:881019 [2] NCCL INFO Channel 13/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-142:880683:881016 [4] NCCL INFO Channel 07/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-142:880686:881021 [7] NCCL INFO Channel 14/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-142:880682:881020 [3] NCCL INFO Channel 13/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-142:880680:881022 [1] NCCL INFO Channel 14/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-142:880684:881017 [5] NCCL INFO Channel 10/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-147:618585:618926 [1] NCCL INFO Channel 22/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-142:880685:881018 [6] NCCL INFO Channel 12/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-142:880679:881015 [0] NCCL INFO Channel 09/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-142:880681:881019 [2] NCCL INFO Channel 14/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-142:880683:881016 [4] NCCL INFO Channel 08/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-142:880686:881021 [7] NCCL INFO Channel 15/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-142:880682:881020 [3] NCCL INFO Channel 14/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-142:880680:881022 [1] NCCL INFO Channel 15/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-142:880684:881017 [5] NCCL INFO Channel 11/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-142:880685:881018 [6] NCCL INFO Channel 13/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-142:880679:881015 [0] NCCL INFO Channel 10/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-142:880681:881019 [2] NCCL INFO Channel 15/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-142:880686:881021 [7] NCCL INFO Channel 16/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-142:880683:881016 [4] NCCL INFO Channel 09/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-142:880680:881022 [1] NCCL INFO Channel 16/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-142:880682:881020 [3] NCCL INFO Channel 15/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-147:618584:618919 [0] NCCL INFO Channel 18/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-147:618585:618926 [1] NCCL INFO Channel 23/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-142:880684:881017 [5] NCCL INFO Channel 12/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-142:880685:881018 [6] NCCL INFO Channel 14/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-142:880679:881015 [0] NCCL INFO Channel 11/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-142:880681:881019 [2] NCCL INFO Channel 16/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-142:880686:881021 [7] NCCL INFO Channel 17/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-142:880683:881016 [4] NCCL INFO Channel 10/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-142:880682:881020 [3] NCCL INFO Channel 16/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-142:880680:881022 [1] NCCL INFO Channel 17/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-73:926597:926935 [3] NCCL INFO Channel 20/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-142:880684:881017 [5] NCCL INFO Channel 13/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-142:880685:881018 [6] NCCL INFO Channel 15/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-142:880679:881015 [0] NCCL INFO Channel 12/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-142:880681:881019 [2] NCCL INFO Channel 17/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-142:880686:881021 [7] NCCL INFO Channel 18/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-142:880683:881016 [4] NCCL INFO Channel 11/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-142:880682:881020 [3] NCCL INFO Channel 17/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-142:880680:881022 [1] NCCL INFO Channel 18/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-147:618586:618923 [2] NCCL INFO Channel 17/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-142:880684:881017 [5] NCCL INFO Channel 14/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-142:880685:881018 [6] NCCL INFO Channel 16/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-142:880679:881015 [0] NCCL INFO Channel 13/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-142:880681:881019 [2] NCCL INFO Channel 18/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-142:880683:881016 [4] NCCL INFO Channel 12/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-142:880686:881021 [7] NCCL INFO Channel 19/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-142:880680:881022 [1] NCCL INFO Channel 19/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-142:880682:881020 [3] NCCL INFO Channel 18/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-173-7:811249:811596 [1] NCCL INFO Channel 19/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-142:880684:881017 [5] NCCL INFO Channel 15/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-142:880685:881018 [6] NCCL INFO Channel 17/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-142:880679:881015 [0] NCCL INFO Channel 14/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-142:880681:881019 [2] NCCL INFO Channel 19/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-142:880686:881021 [7] NCCL INFO Channel 20/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-142:880683:881016 [4] NCCL INFO Channel 13/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-142:880680:881022 [1] NCCL INFO Channel 20/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-142:880682:881020 [3] NCCL INFO Channel 19/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-142:880684:881017 [5] NCCL INFO Channel 16/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-142:880685:881018 [6] NCCL INFO Channel 18/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-142:880679:881015 [0] NCCL INFO Channel 15/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-142:880681:881019 [2] NCCL INFO Channel 20/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-142:880686:881021 [7] NCCL INFO Channel 21/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-142:880683:881016 [4] NCCL INFO Channel 14/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-142:880682:881020 [3] NCCL INFO Channel 20/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-142:880680:881022 [1] NCCL INFO Channel 21/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-142:880684:881017 [5] NCCL INFO Channel 17/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-142:880679:881015 [0] NCCL INFO Channel 16/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-142:880685:881018 [6] NCCL INFO Channel 19/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-142:880681:881019 [2] NCCL INFO Channel 21/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-142:880686:881021 [7] NCCL INFO Channel 22/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-142:880683:881016 [4] NCCL INFO Channel 15/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-142:880682:881020 [3] NCCL INFO Channel 21/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-142:880680:881022 [1] NCCL INFO Channel 22/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-147:618586:618923 [2] NCCL INFO Channel 18/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-142:880684:881017 [5] NCCL INFO Channel 18/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-142:880685:881018 [6] NCCL INFO Channel 20/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-142:880679:881015 [0] NCCL INFO Channel 17/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-142:880681:881019 [2] NCCL INFO Channel 22/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-142:880686:881021 [7] NCCL INFO Channel 23/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-142:880683:881016 [4] NCCL INFO Channel 16/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-142:880682:881020 [3] NCCL INFO Channel 22/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-142:880680:881022 [1] NCCL INFO Channel 23/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-142:880684:881017 [5] NCCL INFO Channel 19/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-142:880685:881018 [6] NCCL INFO Channel 21/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-142:880679:881015 [0] NCCL INFO Channel 18/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-142:880681:881019 [2] NCCL INFO Channel 23/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-142:880683:881016 [4] NCCL INFO Channel 17/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-142:880682:881020 [3] NCCL INFO Channel 23/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-142:880684:881017 [5] NCCL INFO Channel 20/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-142:880685:881018 [6] NCCL INFO Channel 22/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-142:880679:881015 [0] NCCL INFO Channel 19/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-142:880683:881016 [4] NCCL INFO Channel 18/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-147:618584:618919 [0] NCCL INFO Channel 19/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-142:880684:881017 [5] NCCL INFO Channel 21/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-142:880685:881018 [6] NCCL INFO Channel 23/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-142:880679:881015 [0] NCCL INFO Channel 20/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-142:880683:881016 [4] NCCL INFO Channel 19/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-73:926597:926935 [3] NCCL INFO Channel 21/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-173-7:811249:811596 [1] NCCL INFO Channel 20/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-142:880684:881017 [5] NCCL INFO Channel 22/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-142:880679:881015 [0] NCCL INFO Channel 21/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-142:880683:881016 [4] NCCL INFO Channel 20/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-57:942615:942948 [4] NCCL INFO Channel 00/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-57:942615:942948 [4] NCCL INFO Channel 01/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-142:880684:881017 [5] NCCL INFO Channel 23/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-142:880683:881016 [4] NCCL INFO Channel 21/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-142:880679:881015 [0] NCCL INFO Channel 22/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-147:618586:618923 [2] NCCL INFO Channel 19/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-57:942613:942950 [2] NCCL INFO Channel 00/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-57:942615:942948 [4] NCCL INFO Channel 02/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-57:942615:942948 [4] NCCL INFO Channel 03/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-57:942613:942950 [2] NCCL INFO Channel 01/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-57:942618:942951 [7] NCCL INFO Channel 00/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-173-7:811249:811596 [1] NCCL INFO Channel 21/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-57:942613:942950 [2] NCCL INFO Channel 02/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-57:942615:942948 [4] NCCL INFO Channel 04/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-57:942618:942951 [7] NCCL INFO Channel 01/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-57:942612:942954 [1] NCCL INFO Channel 00/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-142:880679:881015 [0] NCCL INFO Channel 23/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-142:880683:881016 [4] NCCL INFO Channel 22/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-57:942613:942950 [2] NCCL INFO Channel 03/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-57:942615:942948 [4] NCCL INFO Channel 05/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-57:942618:942951 [7] NCCL INFO Channel 02/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-57:942612:942954 [1] NCCL INFO Channel 01/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-57:942613:942950 [2] NCCL INFO Channel 04/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-57:942615:942948 [4] NCCL INFO Channel 06/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-57:942612:942954 [1] NCCL INFO Channel 02/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-57:942618:942951 [7] NCCL INFO Channel 03/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-57:942613:942950 [2] NCCL INFO Channel 05/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-57:942618:942951 [7] NCCL INFO Channel 04/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-57:942612:942954 [1] NCCL INFO Channel 03/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-57:942617:942949 [6] NCCL INFO Channel 00/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-57:942613:942950 [2] NCCL INFO Channel 06/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-57:942618:942951 [7] NCCL INFO Channel 05/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-57:942612:942954 [1] NCCL INFO Channel 04/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-57:942617:942949 [6] NCCL INFO Channel 01/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-57:942613:942950 [2] NCCL INFO Channel 07/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-57:942618:942951 [7] NCCL INFO Channel 06/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-57:942612:942954 [1] NCCL INFO Channel 05/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-57:942617:942949 [6] NCCL INFO Channel 02/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-57:942613:942950 [2] NCCL INFO Channel 08/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-57:942618:942951 [7] NCCL INFO Channel 07/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-57:942612:942954 [1] NCCL INFO Channel 06/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-57:942617:942949 [6] NCCL INFO Channel 03/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-173-7:811249:811596 [1] NCCL INFO Channel 22/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-57:942613:942950 [2] NCCL INFO Channel 09/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-57:942618:942951 [7] NCCL INFO Channel 08/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-57:942612:942954 [1] NCCL INFO Channel 07/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-57:942617:942949 [6] NCCL INFO Channel 04/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-73:926597:926935 [3] NCCL INFO Channel 22/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-57:942613:942950 [2] NCCL INFO Channel 10/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-57:942618:942951 [7] NCCL INFO Channel 09/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-57:942612:942954 [1] NCCL INFO Channel 08/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-57:942617:942949 [6] NCCL INFO Channel 05/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-57:942613:942950 [2] NCCL INFO Channel 11/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-57:942615:942948 [4] NCCL INFO Channel 07/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-57:942618:942951 [7] NCCL INFO Channel 10/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-57:942612:942954 [1] NCCL INFO Channel 09/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-57:942617:942949 [6] NCCL INFO Channel 06/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-57:942615:942948 [4] NCCL INFO Channel 08/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-57:942618:942951 [7] NCCL INFO Channel 11/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-57:942616:942952 [5] NCCL INFO Channel 00/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-57:942612:942954 [1] NCCL INFO Channel 10/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-57:942617:942949 [6] NCCL INFO Channel 07/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-57:942615:942948 [4] NCCL INFO Channel 09/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-57:942618:942951 [7] NCCL INFO Channel 12/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-57:942616:942952 [5] NCCL INFO Channel 01/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-57:942612:942954 [1] NCCL INFO Channel 11/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-57:942617:942949 [6] NCCL INFO Channel 08/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-147:618584:618919 [0] NCCL INFO Channel 20/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-57:942615:942948 [4] NCCL INFO Channel 10/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-57:942618:942951 [7] NCCL INFO Channel 13/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-57:942616:942952 [5] NCCL INFO Channel 02/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-57:942612:942954 [1] NCCL INFO Channel 12/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-57:942617:942949 [6] NCCL INFO Channel 09/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-57:942614:942953 [3] NCCL INFO Channel 00/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-147:618586:618923 [2] NCCL INFO Channel 20/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-57:942615:942948 [4] NCCL INFO Channel 11/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-57:942618:942951 [7] NCCL INFO Channel 14/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-57:942616:942952 [5] NCCL INFO Channel 03/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-142:880683:881016 [4] NCCL INFO Channel 23/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-57:942612:942954 [1] NCCL INFO Channel 13/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-57:942617:942949 [6] NCCL INFO Channel 10/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-57:942614:942953 [3] NCCL INFO Channel 01/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-57:942615:942948 [4] NCCL INFO Channel 12/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-57:942618:942951 [7] NCCL INFO Channel 15/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-57:942616:942952 [5] NCCL INFO Channel 04/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-57:942612:942954 [1] NCCL INFO Channel 14/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-57:942617:942949 [6] NCCL INFO Channel 11/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-57:942614:942953 [3] NCCL INFO Channel 02/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-57:942613:942950 [2] NCCL INFO Channel 12/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-57:942615:942948 [4] NCCL INFO Channel 13/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-57:942618:942951 [7] NCCL INFO Channel 16/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-57:942616:942952 [5] NCCL INFO Channel 05/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-173-7:811249:811596 [1] NCCL INFO Channel 23/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-57:942612:942954 [1] NCCL INFO Channel 15/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-57:942617:942949 [6] NCCL INFO Channel 12/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-57:942614:942953 [3] NCCL INFO Channel 03/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-57:942613:942950 [2] NCCL INFO Channel 13/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-57:942615:942948 [4] NCCL INFO Channel 14/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-57:942618:942951 [7] NCCL INFO Channel 17/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-57:942616:942952 [5] NCCL INFO Channel 06/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-57:942612:942954 [1] NCCL INFO Channel 16/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-57:942617:942949 [6] NCCL INFO Channel 13/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-57:942614:942953 [3] NCCL INFO Channel 04/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-57:942613:942950 [2] NCCL INFO Channel 14/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-57:942615:942948 [4] NCCL INFO Channel 15/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-57:942618:942951 [7] NCCL INFO Channel 18/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-57:942616:942952 [5] NCCL INFO Channel 07/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-73:926597:926935 [3] NCCL INFO Channel 23/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-57:942612:942954 [1] NCCL INFO Channel 17/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-57:942617:942949 [6] NCCL INFO Channel 14/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-57:942614:942953 [3] NCCL INFO Channel 05/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-57:942613:942950 [2] NCCL INFO Channel 15/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-57:942615:942948 [4] NCCL INFO Channel 16/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-57:942618:942951 [7] NCCL INFO Channel 19/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-57:942616:942952 [5] NCCL INFO Channel 08/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-57:942611:942947 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-57:942612:942954 [1] NCCL INFO Channel 18/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-57:942614:942953 [3] NCCL INFO Channel 06/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-57:942617:942949 [6] NCCL INFO Channel 15/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-57:942613:942950 [2] NCCL INFO Channel 16/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-57:942615:942948 [4] NCCL INFO Channel 17/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-57:942618:942951 [7] NCCL INFO Channel 20/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-57:942616:942952 [5] NCCL INFO Channel 09/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-57:942611:942947 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-57:942612:942954 [1] NCCL INFO Channel 19/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-57:942614:942953 [3] NCCL INFO Channel 07/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-57:942617:942949 [6] NCCL INFO Channel 16/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-57:942613:942950 [2] NCCL INFO Channel 17/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-57:942615:942948 [4] NCCL INFO Channel 18/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-57:942618:942951 [7] NCCL INFO Channel 21/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-57:942616:942952 [5] NCCL INFO Channel 10/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-57:942611:942947 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-57:942612:942954 [1] NCCL INFO Channel 20/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-57:942614:942953 [3] NCCL INFO Channel 08/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-57:942617:942949 [6] NCCL INFO Channel 17/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-57:942613:942950 [2] NCCL INFO Channel 18/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-57:942615:942948 [4] NCCL INFO Channel 19/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-57:942618:942951 [7] NCCL INFO Channel 22/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-57:942616:942952 [5] NCCL INFO Channel 11/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-57:942611:942947 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-57:942612:942954 [1] NCCL INFO Channel 21/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-57:942614:942953 [3] NCCL INFO Channel 09/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-57:942617:942949 [6] NCCL INFO Channel 18/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-57:942613:942950 [2] NCCL INFO Channel 19/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-57:942615:942948 [4] NCCL INFO Channel 20/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-57:942618:942951 [7] NCCL INFO Channel 23/0 : 7[7] -> 0[0] via P2P/IPC
+ip-26-0-172-57:942616:942952 [5] NCCL INFO Channel 12/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-57:942611:942947 [0] NCCL INFO Channel 04/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-57:942612:942954 [1] NCCL INFO Channel 22/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-57:942614:942953 [3] NCCL INFO Channel 10/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-57:942617:942949 [6] NCCL INFO Channel 19/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-57:942613:942950 [2] NCCL INFO Channel 20/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-57:942615:942948 [4] NCCL INFO Channel 21/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-57:942616:942952 [5] NCCL INFO Channel 13/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-57:942611:942947 [0] NCCL INFO Channel 05/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-57:942612:942954 [1] NCCL INFO Channel 23/0 : 1[1] -> 2[2] via P2P/IPC
+ip-26-0-172-57:942617:942949 [6] NCCL INFO Channel 20/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-57:942614:942953 [3] NCCL INFO Channel 11/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-57:942613:942950 [2] NCCL INFO Channel 21/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-57:942615:942948 [4] NCCL INFO Channel 22/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-57:942616:942952 [5] NCCL INFO Channel 14/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-57:942611:942947 [0] NCCL INFO Channel 06/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-147:618584:618919 [0] NCCL INFO Channel 21/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-57:942617:942949 [6] NCCL INFO Channel 21/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-57:942614:942953 [3] NCCL INFO Channel 12/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-57:942613:942950 [2] NCCL INFO Channel 22/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-57:942615:942948 [4] NCCL INFO Channel 23/0 : 4[4] -> 5[5] via P2P/IPC
+ip-26-0-172-57:942616:942952 [5] NCCL INFO Channel 15/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-57:942611:942947 [0] NCCL INFO Channel 07/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-57:942617:942949 [6] NCCL INFO Channel 22/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-57:942614:942953 [3] NCCL INFO Channel 13/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-147:618586:618923 [2] NCCL INFO Channel 21/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-57:942613:942950 [2] NCCL INFO Channel 23/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-57:942616:942952 [5] NCCL INFO Channel 16/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-57:942611:942947 [0] NCCL INFO Channel 08/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-57:942617:942949 [6] NCCL INFO Channel 23/0 : 6[6] -> 7[7] via P2P/IPC
+ip-26-0-172-57:942614:942953 [3] NCCL INFO Channel 14/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-57:942616:942952 [5] NCCL INFO Channel 17/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-57:942611:942947 [0] NCCL INFO Channel 09/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-57:942614:942953 [3] NCCL INFO Channel 15/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-57:942616:942952 [5] NCCL INFO Channel 18/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-57:942611:942947 [0] NCCL INFO Channel 10/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-57:942614:942953 [3] NCCL INFO Channel 16/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-57:942616:942952 [5] NCCL INFO Channel 19/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-57:942611:942947 [0] NCCL INFO Channel 11/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-147:618584:618919 [0] NCCL INFO Channel 22/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-57:942614:942953 [3] NCCL INFO Channel 17/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-57:942616:942952 [5] NCCL INFO Channel 20/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-57:942611:942947 [0] NCCL INFO Channel 12/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-57:942614:942953 [3] NCCL INFO Channel 18/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-57:942616:942952 [5] NCCL INFO Channel 21/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-57:942611:942947 [0] NCCL INFO Channel 13/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-57:942614:942953 [3] NCCL INFO Channel 19/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-147:618586:618923 [2] NCCL INFO Channel 22/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-57:942616:942952 [5] NCCL INFO Channel 22/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-57:942611:942947 [0] NCCL INFO Channel 14/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-57:942614:942953 [3] NCCL INFO Channel 20/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-57:942616:942952 [5] NCCL INFO Channel 23/0 : 5[5] -> 6[6] via P2P/IPC
+ip-26-0-172-57:942611:942947 [0] NCCL INFO Channel 15/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-57:942614:942953 [3] NCCL INFO Channel 21/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-57:942611:942947 [0] NCCL INFO Channel 16/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-57:942614:942953 [3] NCCL INFO Channel 22/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-147:618584:618919 [0] NCCL INFO Channel 23/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-57:942611:942947 [0] NCCL INFO Channel 17/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-147:618586:618923 [2] NCCL INFO Channel 23/0 : 2[2] -> 3[3] via P2P/IPC
+ip-26-0-172-57:942614:942953 [3] NCCL INFO Channel 23/0 : 3[3] -> 4[4] via P2P/IPC
+ip-26-0-172-57:942611:942947 [0] NCCL INFO Channel 18/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-57:942611:942947 [0] NCCL INFO Channel 19/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-57:942611:942947 [0] NCCL INFO Channel 20/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-57:942611:942947 [0] NCCL INFO Channel 21/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-57:942611:942947 [0] NCCL INFO Channel 22/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-57:942611:942947 [0] NCCL INFO Channel 23/0 : 0[0] -> 1[1] via P2P/IPC
+ip-26-0-172-116:1019295:1019627 [7] NCCL INFO Connected all rings
+ip-26-0-172-116:1019295:1019627 [7] NCCL INFO Channel 00/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-116:1019295:1019627 [7] NCCL INFO Channel 01/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-116:1019295:1019627 [7] NCCL INFO Channel 02/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-116:1019295:1019627 [7] NCCL INFO Channel 03/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-116:1019295:1019627 [7] NCCL INFO Channel 04/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-116:1019295:1019627 [7] NCCL INFO Channel 05/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-116:1019295:1019627 [7] NCCL INFO Channel 06/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-116:1019295:1019627 [7] NCCL INFO Channel 07/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-225:94343:94681 [1] NCCL INFO Connected all rings
+ip-26-0-172-116:1019295:1019627 [7] NCCL INFO Channel 08/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-252:688201:688542 [1] NCCL INFO Connected all rings
+ip-26-0-160-225:94349:94683 [7] NCCL INFO Connected all rings
+ip-26-0-160-225:94349:94683 [7] NCCL INFO Channel 00/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-116:1019295:1019627 [7] NCCL INFO Channel 09/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-116:1019295:1019627 [7] NCCL INFO Channel 10/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-225:94342:94680 [0] NCCL INFO Connected all rings
+ip-26-0-160-225:94349:94683 [7] NCCL INFO Channel 01/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-252:688207:688543 [7] NCCL INFO Connected all rings
+ip-26-0-172-252:688207:688543 [7] NCCL INFO Channel 00/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-116:1019295:1019627 [7] NCCL INFO Channel 11/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-225:94349:94683 [7] NCCL INFO Channel 02/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-252:688207:688543 [7] NCCL INFO Channel 01/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-73:926600:926933 [6] NCCL INFO Connected all rings
+ip-26-0-160-225:94349:94683 [7] NCCL INFO Channel 03/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-225:94344:94686 [2] NCCL INFO Connected all rings
+ip-26-0-172-116:1019295:1019627 [7] NCCL INFO Channel 12/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-252:688207:688543 [7] NCCL INFO Channel 02/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-225:94347:94685 [5] NCCL INFO Connected all rings
+ip-26-0-160-225:94345:94687 [3] NCCL INFO Connected all rings
+ip-26-0-172-147:618590:618925 [6] NCCL INFO Connected all rings
+ip-26-0-160-225:94348:94684 [6] NCCL INFO Connected all rings
+ip-26-0-172-252:688203:688541 [3] NCCL INFO Connected all rings
+ip-26-0-172-252:688207:688543 [7] NCCL INFO Channel 03/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-225:94349:94683 [7] NCCL INFO Channel 04/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-252:688202:688544 [2] NCCL INFO Connected all rings
+ip-26-0-160-225:94346:94682 [4] NCCL INFO Connected all rings
+ip-26-0-160-225:94349:94683 [7] NCCL INFO Channel 05/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-252:688207:688543 [7] NCCL INFO Channel 04/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-116:1019295:1019627 [7] NCCL INFO Channel 13/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-225:94349:94683 [7] NCCL INFO Channel 06/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-252:688206:688539 [6] NCCL INFO Connected all rings
+ip-26-0-172-252:688207:688543 [7] NCCL INFO Channel 05/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-225:94349:94683 [7] NCCL INFO Channel 07/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-252:688207:688543 [7] NCCL INFO Channel 06/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-225:94349:94683 [7] NCCL INFO Channel 08/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-252:688200:688538 [0] NCCL INFO Connected all rings
+ip-26-0-160-225:94349:94683 [7] NCCL INFO Channel 09/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-225:94349:94683 [7] NCCL INFO Channel 10/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-252:688207:688543 [7] NCCL INFO Channel 07/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-225:94349:94683 [7] NCCL INFO Channel 11/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-225:94349:94683 [7] NCCL INFO Channel 12/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-116:1019291:1019630 [3] NCCL INFO Connected all rings
+ip-26-0-160-225:94349:94683 [7] NCCL INFO Channel 13/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-225:94349:94683 [7] NCCL INFO Channel 14/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-252:688204:688540 [4] NCCL INFO Connected all rings
+ip-26-0-160-225:94349:94683 [7] NCCL INFO Channel 15/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-225:94349:94683 [7] NCCL INFO Channel 16/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-252:688207:688543 [7] NCCL INFO Channel 08/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-116:1019288:1019623 [0] NCCL INFO Connected all rings
+ip-26-0-160-225:94349:94683 [7] NCCL INFO Channel 17/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-225:94349:94683 [7] NCCL INFO Channel 18/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-116:1019292:1019624 [4] NCCL INFO Connected all rings
+ip-26-0-172-252:688207:688543 [7] NCCL INFO Channel 09/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-225:94349:94683 [7] NCCL INFO Channel 19/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-252:688207:688543 [7] NCCL INFO Channel 10/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-116:1019295:1019627 [7] NCCL INFO Channel 14/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-225:94343:94681 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-225:94343:94681 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-225:94344:94686 [2] NCCL INFO Channel 00/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-225:94343:94681 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-252:688205:688545 [5] NCCL INFO Connected all rings
+ip-26-0-160-225:94345:94687 [3] NCCL INFO Channel 00/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-225:94344:94686 [2] NCCL INFO Channel 01/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-252:688207:688543 [7] NCCL INFO Channel 11/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-225:94343:94681 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-225:94345:94687 [3] NCCL INFO Channel 01/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-225:94344:94686 [2] NCCL INFO Channel 02/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-252:688207:688543 [7] NCCL INFO Channel 12/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-225:94343:94681 [1] NCCL INFO Channel 04/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-225:94345:94687 [3] NCCL INFO Channel 02/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-225:94344:94686 [2] NCCL INFO Channel 03/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-225:94346:94682 [4] NCCL INFO Channel 00/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-252:688207:688543 [7] NCCL INFO Channel 13/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-225:94349:94683 [7] NCCL INFO Channel 20/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-225:94345:94687 [3] NCCL INFO Channel 03/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-225:94343:94681 [1] NCCL INFO Channel 05/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-225:94344:94686 [2] NCCL INFO Channel 04/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-225:94346:94682 [4] NCCL INFO Channel 01/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-252:688207:688543 [7] NCCL INFO Channel 14/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-225:94349:94683 [7] NCCL INFO Channel 21/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-225:94345:94687 [3] NCCL INFO Channel 04/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-225:94343:94681 [1] NCCL INFO Channel 06/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-225:94344:94686 [2] NCCL INFO Channel 05/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-225:94346:94682 [4] NCCL INFO Channel 02/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-252:688207:688543 [7] NCCL INFO Channel 15/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-116:1019295:1019627 [7] NCCL INFO Channel 15/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-225:94349:94683 [7] NCCL INFO Channel 22/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-225:94345:94687 [3] NCCL INFO Channel 05/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-225:94343:94681 [1] NCCL INFO Channel 07/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-225:94344:94686 [2] NCCL INFO Channel 06/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-225:94346:94682 [4] NCCL INFO Channel 03/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-252:688207:688543 [7] NCCL INFO Channel 16/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-225:94349:94683 [7] NCCL INFO Channel 23/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-225:94345:94687 [3] NCCL INFO Channel 06/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-225:94343:94681 [1] NCCL INFO Channel 08/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-225:94344:94686 [2] NCCL INFO Channel 07/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-225:94346:94682 [4] NCCL INFO Channel 04/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-252:688207:688543 [7] NCCL INFO Channel 17/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-73:926594:926931 [0] NCCL INFO Connected all rings
+ip-26-0-160-225:94343:94681 [1] NCCL INFO Channel 09/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-225:94345:94687 [3] NCCL INFO Channel 07/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-225:94344:94686 [2] NCCL INFO Channel 08/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-225:94346:94682 [4] NCCL INFO Channel 05/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-252:688207:688543 [7] NCCL INFO Channel 18/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-225:94343:94681 [1] NCCL INFO Channel 10/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-225:94345:94687 [3] NCCL INFO Channel 08/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-225:94344:94686 [2] NCCL INFO Channel 09/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-225:94346:94682 [4] NCCL INFO Channel 06/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-252:688207:688543 [7] NCCL INFO Channel 19/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-225:94343:94681 [1] NCCL INFO Channel 11/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-225:94344:94686 [2] NCCL INFO Channel 10/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-225:94345:94687 [3] NCCL INFO Channel 09/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-225:94346:94682 [4] NCCL INFO Channel 07/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-73:926599:926932 [5] NCCL INFO Connected all rings
+ip-26-0-172-252:688207:688543 [7] NCCL INFO Channel 20/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-225:94343:94681 [1] NCCL INFO Channel 12/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-225:94344:94686 [2] NCCL INFO Channel 11/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-225:94345:94687 [3] NCCL INFO Channel 10/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-225:94346:94682 [4] NCCL INFO Channel 08/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-225:94344:94686 [2] NCCL INFO Channel 12/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-225:94343:94681 [1] NCCL INFO Channel 13/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-225:94345:94687 [3] NCCL INFO Channel 11/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-225:94346:94682 [4] NCCL INFO Channel 09/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-252:688207:688543 [7] NCCL INFO Channel 21/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-252:688207:688543 [7] NCCL INFO Channel 22/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-225:94344:94686 [2] NCCL INFO Channel 13/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-225:94343:94681 [1] NCCL INFO Channel 14/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-116:1019295:1019627 [7] NCCL INFO Channel 16/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-225:94345:94687 [3] NCCL INFO Channel 12/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-225:94344:94686 [2] NCCL INFO Channel 14/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-225:94343:94681 [1] NCCL INFO Channel 15/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-252:688207:688543 [7] NCCL INFO Channel 23/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-225:94345:94687 [3] NCCL INFO Channel 13/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-116:1019293:1019625 [5] NCCL INFO Connected all rings
+ip-26-0-160-225:94344:94686 [2] NCCL INFO Channel 15/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-225:94343:94681 [1] NCCL INFO Channel 16/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-225:94345:94687 [3] NCCL INFO Channel 14/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-225:94346:94682 [4] NCCL INFO Channel 10/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-225:94344:94686 [2] NCCL INFO Channel 16/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-225:94343:94681 [1] NCCL INFO Channel 17/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-225:94345:94687 [3] NCCL INFO Channel 15/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-225:94346:94682 [4] NCCL INFO Channel 11/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-225:94344:94686 [2] NCCL INFO Channel 17/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-225:94343:94681 [1] NCCL INFO Channel 18/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-225:94345:94687 [3] NCCL INFO Channel 16/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-116:1019290:1019628 [2] NCCL INFO Connected all rings
+ip-26-0-160-225:94346:94682 [4] NCCL INFO Channel 12/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-225:94348:94684 [6] NCCL INFO Channel 00/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-225:94344:94686 [2] NCCL INFO Channel 18/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-225:94343:94681 [1] NCCL INFO Channel 19/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-225:94345:94687 [3] NCCL INFO Channel 17/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-252:688203:688541 [3] NCCL INFO Channel 00/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-252:688202:688544 [2] NCCL INFO Channel 00/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-225:94346:94682 [4] NCCL INFO Channel 13/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-225:94347:94685 [5] NCCL INFO Channel 00/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-225:94348:94684 [6] NCCL INFO Channel 01/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-225:94344:94686 [2] NCCL INFO Channel 19/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-225:94343:94681 [1] NCCL INFO Channel 20/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-225:94345:94687 [3] NCCL INFO Channel 18/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-252:688203:688541 [3] NCCL INFO Channel 01/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-225:94346:94682 [4] NCCL INFO Channel 14/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-225:94348:94684 [6] NCCL INFO Channel 02/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-225:94344:94686 [2] NCCL INFO Channel 20/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-225:94347:94685 [5] NCCL INFO Channel 01/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-225:94343:94681 [1] NCCL INFO Channel 21/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-116:1019295:1019627 [7] NCCL INFO Channel 17/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-225:94345:94687 [3] NCCL INFO Channel 19/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-252:688201:688542 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-252:688203:688541 [3] NCCL INFO Channel 02/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-252:688201:688542 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-252:688203:688541 [3] NCCL INFO Channel 03/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-225:94346:94682 [4] NCCL INFO Channel 15/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-225:94348:94684 [6] NCCL INFO Channel 03/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-225:94347:94685 [5] NCCL INFO Channel 02/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-225:94344:94686 [2] NCCL INFO Channel 21/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-225:94343:94681 [1] NCCL INFO Channel 22/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-225:94345:94687 [3] NCCL INFO Channel 20/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-252:688206:688539 [6] NCCL INFO Channel 00/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-252:688201:688542 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-252:688203:688541 [3] NCCL INFO Channel 04/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-252:688206:688539 [6] NCCL INFO Channel 01/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-252:688201:688542 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-225:94346:94682 [4] NCCL INFO Channel 16/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-225:94348:94684 [6] NCCL INFO Channel 04/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-225:94347:94685 [5] NCCL INFO Channel 03/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-225:94344:94686 [2] NCCL INFO Channel 22/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-225:94343:94681 [1] NCCL INFO Channel 23/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-225:94345:94687 [3] NCCL INFO Channel 21/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-252:688203:688541 [3] NCCL INFO Channel 05/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-252:688206:688539 [6] NCCL INFO Channel 02/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-252:688201:688542 [1] NCCL INFO Channel 04/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-225:94345:94687 [3] NCCL INFO Channel 22/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-225:94346:94682 [4] NCCL INFO Channel 17/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-225:94348:94684 [6] NCCL INFO Channel 05/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-252:688204:688540 [4] NCCL INFO Channel 00/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-252:688202:688544 [2] NCCL INFO Channel 01/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-252:688203:688541 [3] NCCL INFO Channel 06/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-252:688204:688540 [4] NCCL INFO Channel 01/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-252:688202:688544 [2] NCCL INFO Channel 02/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-116:1019295:1019627 [7] NCCL INFO Channel 18/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-252:688203:688541 [3] NCCL INFO Channel 07/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-252:688204:688540 [4] NCCL INFO Channel 02/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-252:688202:688544 [2] NCCL INFO Channel 03/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-252:688203:688541 [3] NCCL INFO Channel 08/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-252:688204:688540 [4] NCCL INFO Channel 03/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-116:1019294:1019626 [6] NCCL INFO Connected all rings
+ip-26-0-172-252:688202:688544 [2] NCCL INFO Channel 04/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-225:94344:94686 [2] NCCL INFO Channel 23/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-173-7:811252:811592 [4] NCCL INFO Connected all rings
+ip-26-0-172-116:1019289:1019629 [1] NCCL INFO Connected all rings
+ip-26-0-172-252:688206:688539 [6] NCCL INFO Channel 03/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-116:1019295:1019627 [7] NCCL INFO Channel 19/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-225:94347:94685 [5] NCCL INFO Channel 04/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-252:688204:688540 [4] NCCL INFO Channel 04/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-252:688203:688541 [3] NCCL INFO Channel 09/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-252:688206:688539 [6] NCCL INFO Channel 04/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-252:688204:688540 [4] NCCL INFO Channel 05/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-116:1019295:1019627 [7] NCCL INFO Channel 20/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-252:688203:688541 [3] NCCL INFO Channel 10/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-252:688206:688539 [6] NCCL INFO Channel 05/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-252:688204:688540 [4] NCCL INFO Channel 06/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-252:688203:688541 [3] NCCL INFO Channel 11/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-252:688206:688539 [6] NCCL INFO Channel 06/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-116:1019295:1019627 [7] NCCL INFO Channel 21/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-252:688204:688540 [4] NCCL INFO Channel 07/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-252:688203:688541 [3] NCCL INFO Channel 12/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-252:688206:688539 [6] NCCL INFO Channel 07/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-225:94347:94685 [5] NCCL INFO Channel 05/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-252:688204:688540 [4] NCCL INFO Channel 08/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-252:688203:688541 [3] NCCL INFO Channel 13/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-116:1019295:1019627 [7] NCCL INFO Channel 22/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-252:688206:688539 [6] NCCL INFO Channel 08/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-252:688205:688545 [5] NCCL INFO Channel 00/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-252:688204:688540 [4] NCCL INFO Channel 09/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-252:688203:688541 [3] NCCL INFO Channel 14/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-252:688206:688539 [6] NCCL INFO Channel 09/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-252:688205:688545 [5] NCCL INFO Channel 01/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-116:1019295:1019627 [7] NCCL INFO Channel 23/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-252:688204:688540 [4] NCCL INFO Channel 10/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-252:688203:688541 [3] NCCL INFO Channel 15/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-252:688206:688539 [6] NCCL INFO Channel 10/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-252:688205:688545 [5] NCCL INFO Channel 02/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-225:94346:94682 [4] NCCL INFO Channel 18/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-225:94345:94687 [3] NCCL INFO Channel 23/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-252:688204:688540 [4] NCCL INFO Channel 11/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-252:688203:688541 [3] NCCL INFO Channel 16/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-252:688201:688542 [1] NCCL INFO Channel 05/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-252:688206:688539 [6] NCCL INFO Channel 11/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-252:688205:688545 [5] NCCL INFO Channel 03/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-252:688202:688544 [2] NCCL INFO Channel 05/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-252:688204:688540 [4] NCCL INFO Channel 12/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-252:688203:688541 [3] NCCL INFO Channel 17/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-252:688201:688542 [1] NCCL INFO Channel 06/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-252:688206:688539 [6] NCCL INFO Channel 12/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-252:688205:688545 [5] NCCL INFO Channel 04/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-147:618588:618920 [4] NCCL INFO Connected all rings
+ip-26-0-160-225:94347:94685 [5] NCCL INFO Channel 06/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-225:94348:94684 [6] NCCL INFO Channel 06/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-252:688204:688540 [4] NCCL INFO Channel 13/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-252:688203:688541 [3] NCCL INFO Channel 18/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-252:688202:688544 [2] NCCL INFO Channel 06/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-225:94346:94682 [4] NCCL INFO Channel 19/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-252:688206:688539 [6] NCCL INFO Channel 13/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-252:688204:688540 [4] NCCL INFO Channel 14/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-252:688203:688541 [3] NCCL INFO Channel 19/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-252:688202:688544 [2] NCCL INFO Channel 07/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-252:688206:688539 [6] NCCL INFO Channel 14/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-252:688204:688540 [4] NCCL INFO Channel 15/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-252:688203:688541 [3] NCCL INFO Channel 20/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-252:688202:688544 [2] NCCL INFO Channel 08/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-252:688206:688539 [6] NCCL INFO Channel 15/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-252:688203:688541 [3] NCCL INFO Channel 21/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-73:926595:926937 [1] NCCL INFO Connected all rings
+ip-26-0-172-252:688202:688544 [2] NCCL INFO Channel 09/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-252:688204:688540 [4] NCCL INFO Channel 16/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-225:94346:94682 [4] NCCL INFO Channel 20/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-225:94346:94682 [4] NCCL INFO Channel 21/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-252:688201:688542 [1] NCCL INFO Channel 07/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-147:618589:618922 [5] NCCL INFO Connected all rings
+ip-26-0-172-252:688206:688539 [6] NCCL INFO Channel 16/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-252:688203:688541 [3] NCCL INFO Channel 22/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-252:688204:688540 [4] NCCL INFO Channel 17/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-252:688202:688544 [2] NCCL INFO Channel 10/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-252:688201:688542 [1] NCCL INFO Channel 08/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-252:688205:688545 [5] NCCL INFO Channel 05/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-252:688206:688539 [6] NCCL INFO Channel 17/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-252:688203:688541 [3] NCCL INFO Channel 23/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-252:688204:688540 [4] NCCL INFO Channel 18/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-252:688202:688544 [2] NCCL INFO Channel 11/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-252:688201:688542 [1] NCCL INFO Channel 09/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-252:688205:688545 [5] NCCL INFO Channel 06/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-252:688206:688539 [6] NCCL INFO Channel 18/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-252:688204:688540 [4] NCCL INFO Channel 19/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-252:688202:688544 [2] NCCL INFO Channel 12/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-252:688201:688542 [1] NCCL INFO Channel 10/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-252:688205:688545 [5] NCCL INFO Channel 07/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-252:688206:688539 [6] NCCL INFO Channel 19/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-252:688204:688540 [4] NCCL INFO Channel 20/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-252:688202:688544 [2] NCCL INFO Channel 13/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-225:94347:94685 [5] NCCL INFO Channel 07/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-225:94348:94684 [6] NCCL INFO Channel 07/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-252:688201:688542 [1] NCCL INFO Channel 11/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-252:688206:688539 [6] NCCL INFO Channel 20/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-252:688205:688545 [5] NCCL INFO Channel 08/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-252:688204:688540 [4] NCCL INFO Channel 21/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-252:688202:688544 [2] NCCL INFO Channel 14/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-252:688201:688542 [1] NCCL INFO Channel 12/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-252:688206:688539 [6] NCCL INFO Channel 21/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-252:688205:688545 [5] NCCL INFO Channel 09/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-252:688204:688540 [4] NCCL INFO Channel 22/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-252:688202:688544 [2] NCCL INFO Channel 15/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-252:688201:688542 [1] NCCL INFO Channel 13/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-252:688206:688539 [6] NCCL INFO Channel 22/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-252:688205:688545 [5] NCCL INFO Channel 10/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-252:688204:688540 [4] NCCL INFO Channel 23/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-173-7:811255:811595 [7] NCCL INFO Connected all rings
+ip-26-0-172-252:688201:688542 [1] NCCL INFO Channel 14/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-173-7:811255:811595 [7] NCCL INFO Channel 00/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-252:688206:688539 [6] NCCL INFO Channel 23/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-252:688205:688545 [5] NCCL INFO Channel 11/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-252:688201:688542 [1] NCCL INFO Channel 15/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-252:688205:688545 [5] NCCL INFO Channel 12/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-252:688202:688544 [2] NCCL INFO Channel 16/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-252:688201:688542 [1] NCCL INFO Channel 16/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-225:94346:94682 [4] NCCL INFO Channel 22/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-252:688205:688545 [5] NCCL INFO Channel 13/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-252:688202:688544 [2] NCCL INFO Channel 17/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-252:688201:688542 [1] NCCL INFO Channel 17/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-142:880681:881019 [2] NCCL INFO Connected all rings
+ip-26-0-172-116:1019291:1019630 [3] NCCL INFO Channel 00/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-252:688205:688545 [5] NCCL INFO Channel 14/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-252:688202:688544 [2] NCCL INFO Channel 18/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-116:1019291:1019630 [3] NCCL INFO Channel 01/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-116:1019291:1019630 [3] NCCL INFO Channel 02/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-252:688201:688542 [1] NCCL INFO Channel 18/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-116:1019291:1019630 [3] NCCL INFO Channel 03/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-225:94347:94685 [5] NCCL INFO Channel 08/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-225:94348:94684 [6] NCCL INFO Channel 08/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-116:1019291:1019630 [3] NCCL INFO Channel 04/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-252:688205:688545 [5] NCCL INFO Channel 15/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-252:688202:688544 [2] NCCL INFO Channel 19/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-252:688201:688542 [1] NCCL INFO Channel 19/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-116:1019294:1019626 [6] NCCL INFO Channel 00/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-116:1019291:1019630 [3] NCCL INFO Channel 05/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-116:1019294:1019626 [6] NCCL INFO Channel 01/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-116:1019291:1019630 [3] NCCL INFO Channel 06/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-225:94346:94682 [4] NCCL INFO Channel 23/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-116:1019294:1019626 [6] NCCL INFO Channel 02/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-116:1019291:1019630 [3] NCCL INFO Channel 07/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-225:94347:94685 [5] NCCL INFO Channel 09/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-252:688202:688544 [2] NCCL INFO Channel 20/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-252:688205:688545 [5] NCCL INFO Channel 16/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-252:688201:688542 [1] NCCL INFO Channel 20/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-116:1019291:1019630 [3] NCCL INFO Channel 08/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-116:1019294:1019626 [6] NCCL INFO Channel 03/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-116:1019292:1019624 [4] NCCL INFO Channel 00/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-116:1019293:1019625 [5] NCCL INFO Channel 00/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-252:688201:688542 [1] NCCL INFO Channel 21/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-252:688202:688544 [2] NCCL INFO Channel 21/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-252:688205:688545 [5] NCCL INFO Channel 17/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-225:94348:94684 [6] NCCL INFO Channel 09/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-173-7:811255:811595 [7] NCCL INFO Channel 01/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-116:1019292:1019624 [4] NCCL INFO Channel 01/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-116:1019294:1019626 [6] NCCL INFO Channel 04/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-116:1019293:1019625 [5] NCCL INFO Channel 01/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-116:1019289:1019629 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-116:1019290:1019628 [2] NCCL INFO Channel 00/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-116:1019292:1019624 [4] NCCL INFO Channel 02/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-116:1019294:1019626 [6] NCCL INFO Channel 05/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-252:688201:688542 [1] NCCL INFO Channel 22/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-252:688202:688544 [2] NCCL INFO Channel 22/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-252:688205:688545 [5] NCCL INFO Channel 18/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-116:1019291:1019630 [3] NCCL INFO Channel 09/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-116:1019293:1019625 [5] NCCL INFO Channel 02/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-116:1019289:1019629 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-116:1019290:1019628 [2] NCCL INFO Channel 01/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-225:94347:94685 [5] NCCL INFO Channel 10/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-116:1019292:1019624 [4] NCCL INFO Channel 03/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-116:1019294:1019626 [6] NCCL INFO Channel 06/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-225:94348:94684 [6] NCCL INFO Channel 10/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-116:1019291:1019630 [3] NCCL INFO Channel 10/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-116:1019293:1019625 [5] NCCL INFO Channel 03/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-116:1019289:1019629 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-116:1019290:1019628 [2] NCCL INFO Channel 02/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-116:1019292:1019624 [4] NCCL INFO Channel 04/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-116:1019294:1019626 [6] NCCL INFO Channel 07/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-116:1019291:1019630 [3] NCCL INFO Channel 11/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-252:688201:688542 [1] NCCL INFO Channel 23/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-252:688202:688544 [2] NCCL INFO Channel 23/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-252:688205:688545 [5] NCCL INFO Channel 19/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-116:1019293:1019625 [5] NCCL INFO Channel 04/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-116:1019289:1019629 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-116:1019292:1019624 [4] NCCL INFO Channel 05/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-116:1019290:1019628 [2] NCCL INFO Channel 03/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-116:1019294:1019626 [6] NCCL INFO Channel 08/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-116:1019291:1019630 [3] NCCL INFO Channel 12/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-116:1019293:1019625 [5] NCCL INFO Channel 05/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-116:1019292:1019624 [4] NCCL INFO Channel 06/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-116:1019290:1019628 [2] NCCL INFO Channel 04/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-116:1019294:1019626 [6] NCCL INFO Channel 09/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-116:1019289:1019629 [1] NCCL INFO Channel 04/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-116:1019291:1019630 [3] NCCL INFO Channel 13/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-116:1019294:1019626 [6] NCCL INFO Channel 10/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-116:1019289:1019629 [1] NCCL INFO Channel 05/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-116:1019291:1019630 [3] NCCL INFO Channel 14/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-252:688205:688545 [5] NCCL INFO Channel 20/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-116:1019292:1019624 [4] NCCL INFO Channel 07/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-116:1019290:1019628 [2] NCCL INFO Channel 05/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-116:1019289:1019629 [1] NCCL INFO Channel 06/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-116:1019294:1019626 [6] NCCL INFO Channel 11/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-116:1019291:1019630 [3] NCCL INFO Channel 15/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-116:1019292:1019624 [4] NCCL INFO Channel 08/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-116:1019290:1019628 [2] NCCL INFO Channel 06/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-252:688205:688545 [5] NCCL INFO Channel 21/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-116:1019289:1019629 [1] NCCL INFO Channel 07/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-116:1019294:1019626 [6] NCCL INFO Channel 12/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-116:1019291:1019630 [3] NCCL INFO Channel 16/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-252:688205:688545 [5] NCCL INFO Channel 22/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-116:1019292:1019624 [4] NCCL INFO Channel 09/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-252:688205:688545 [5] NCCL INFO Channel 23/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-116:1019290:1019628 [2] NCCL INFO Channel 07/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-116:1019289:1019629 [1] NCCL INFO Channel 08/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-116:1019294:1019626 [6] NCCL INFO Channel 13/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-116:1019291:1019630 [3] NCCL INFO Channel 17/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-116:1019292:1019624 [4] NCCL INFO Channel 10/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-116:1019290:1019628 [2] NCCL INFO Channel 08/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-116:1019294:1019626 [6] NCCL INFO Channel 14/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-116:1019289:1019629 [1] NCCL INFO Channel 09/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-116:1019291:1019630 [3] NCCL INFO Channel 18/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-116:1019290:1019628 [2] NCCL INFO Channel 09/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-116:1019292:1019624 [4] NCCL INFO Channel 11/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-116:1019294:1019626 [6] NCCL INFO Channel 15/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-116:1019289:1019629 [1] NCCL INFO Channel 10/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-116:1019291:1019630 [3] NCCL INFO Channel 19/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-116:1019290:1019628 [2] NCCL INFO Channel 10/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-116:1019292:1019624 [4] NCCL INFO Channel 12/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-116:1019289:1019629 [1] NCCL INFO Channel 11/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-116:1019294:1019626 [6] NCCL INFO Channel 16/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-116:1019291:1019630 [3] NCCL INFO Channel 20/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-116:1019290:1019628 [2] NCCL INFO Channel 11/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-116:1019292:1019624 [4] NCCL INFO Channel 13/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-116:1019289:1019629 [1] NCCL INFO Channel 12/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-116:1019294:1019626 [6] NCCL INFO Channel 17/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-116:1019291:1019630 [3] NCCL INFO Channel 21/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-116:1019290:1019628 [2] NCCL INFO Channel 12/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-116:1019292:1019624 [4] NCCL INFO Channel 14/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-116:1019293:1019625 [5] NCCL INFO Channel 06/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-116:1019289:1019629 [1] NCCL INFO Channel 13/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-116:1019294:1019626 [6] NCCL INFO Channel 18/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-142:880679:881015 [0] NCCL INFO Connected all rings
+ip-26-0-172-116:1019291:1019630 [3] NCCL INFO Channel 22/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-116:1019290:1019628 [2] NCCL INFO Channel 13/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-116:1019292:1019624 [4] NCCL INFO Channel 15/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-116:1019293:1019625 [5] NCCL INFO Channel 07/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-116:1019289:1019629 [1] NCCL INFO Channel 14/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-116:1019294:1019626 [6] NCCL INFO Channel 19/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-116:1019291:1019630 [3] NCCL INFO Channel 23/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-116:1019290:1019628 [2] NCCL INFO Channel 14/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-116:1019292:1019624 [4] NCCL INFO Channel 16/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-225:94347:94685 [5] NCCL INFO Channel 11/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-116:1019293:1019625 [5] NCCL INFO Channel 08/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-116:1019289:1019629 [1] NCCL INFO Channel 15/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-116:1019294:1019626 [6] NCCL INFO Channel 20/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-116:1019290:1019628 [2] NCCL INFO Channel 15/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-116:1019292:1019624 [4] NCCL INFO Channel 17/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-116:1019293:1019625 [5] NCCL INFO Channel 09/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-116:1019289:1019629 [1] NCCL INFO Channel 16/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-116:1019294:1019626 [6] NCCL INFO Channel 21/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-116:1019290:1019628 [2] NCCL INFO Channel 16/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-116:1019292:1019624 [4] NCCL INFO Channel 18/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-116:1019293:1019625 [5] NCCL INFO Channel 10/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-116:1019289:1019629 [1] NCCL INFO Channel 17/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-116:1019294:1019626 [6] NCCL INFO Channel 22/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-225:94347:94685 [5] NCCL INFO Channel 12/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-116:1019290:1019628 [2] NCCL INFO Channel 17/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-73:926601:926934 [7] NCCL INFO Connected all rings
+ip-26-0-172-73:926601:926934 [7] NCCL INFO Channel 00/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-116:1019292:1019624 [4] NCCL INFO Channel 19/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-225:94348:94684 [6] NCCL INFO Channel 11/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-116:1019293:1019625 [5] NCCL INFO Channel 11/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-116:1019289:1019629 [1] NCCL INFO Channel 18/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-116:1019294:1019626 [6] NCCL INFO Channel 23/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-116:1019290:1019628 [2] NCCL INFO Channel 18/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-116:1019292:1019624 [4] NCCL INFO Channel 20/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-116:1019293:1019625 [5] NCCL INFO Channel 12/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-116:1019289:1019629 [1] NCCL INFO Channel 19/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-173-7:811255:811595 [7] NCCL INFO Channel 02/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-116:1019290:1019628 [2] NCCL INFO Channel 19/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-116:1019292:1019624 [4] NCCL INFO Channel 21/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-116:1019293:1019625 [5] NCCL INFO Channel 13/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-116:1019289:1019629 [1] NCCL INFO Channel 20/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-116:1019290:1019628 [2] NCCL INFO Channel 20/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-116:1019292:1019624 [4] NCCL INFO Channel 22/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-225:94347:94685 [5] NCCL INFO Channel 13/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-116:1019292:1019624 [4] NCCL INFO Channel 23/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-116:1019290:1019628 [2] NCCL INFO Channel 21/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-116:1019293:1019625 [5] NCCL INFO Channel 14/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-116:1019289:1019629 [1] NCCL INFO Channel 21/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-116:1019290:1019628 [2] NCCL INFO Channel 22/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-116:1019293:1019625 [5] NCCL INFO Channel 15/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-116:1019289:1019629 [1] NCCL INFO Channel 22/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-73:926601:926934 [7] NCCL INFO Channel 01/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-173-7:811253:811593 [5] NCCL INFO Connected all rings
+ip-26-0-160-225:94347:94685 [5] NCCL INFO Channel 14/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-173-7:811254:811594 [6] NCCL INFO Connected all rings
+ip-26-0-172-116:1019290:1019628 [2] NCCL INFO Channel 23/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-116:1019293:1019625 [5] NCCL INFO Channel 16/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-116:1019289:1019629 [1] NCCL INFO Channel 23/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-73:926596:926936 [2] NCCL INFO Connected all rings
+ip-26-0-172-73:926601:926934 [7] NCCL INFO Channel 02/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-116:1019293:1019625 [5] NCCL INFO Channel 17/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-173-7:811255:811595 [7] NCCL INFO Channel 03/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-225:94347:94685 [5] NCCL INFO Channel 15/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-225:94347:94685 [5] NCCL INFO Channel 16/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-116:1019293:1019625 [5] NCCL INFO Channel 18/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-73:926598:926938 [4] NCCL INFO Connected all rings
+ip-26-0-172-73:926601:926934 [7] NCCL INFO Channel 03/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-57:942616:942952 [5] NCCL INFO Connected all rings
+ip-26-0-160-225:94348:94684 [6] NCCL INFO Channel 12/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-73:926601:926934 [7] NCCL INFO Channel 04/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-173-7:811251:811598 [3] NCCL INFO Connected all rings
+ip-26-0-172-73:926601:926934 [7] NCCL INFO Channel 05/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-142:880685:881018 [6] NCCL INFO Connected all rings
+ip-26-0-173-7:811255:811595 [7] NCCL INFO Channel 04/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-73:926601:926934 [7] NCCL INFO Channel 06/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-225:94342:94680 [0] NCCL INFO Connected all trees
+ip-26-0-160-225:94347:94685 [5] NCCL INFO Channel 17/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-225:94342:94680 [0] NCCL INFO NVLS comm 0x93142a0 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-225:94348:94684 [6] NCCL INFO Channel 13/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-73:926601:926934 [7] NCCL INFO Channel 07/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-116:1019293:1019625 [5] NCCL INFO Channel 19/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-173-7:811249:811596 [1] NCCL INFO Connected all rings
+ip-26-0-172-73:926601:926934 [7] NCCL INFO Channel 08/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-173-7:811255:811595 [7] NCCL INFO Channel 05/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-73:926597:926935 [3] NCCL INFO Connected all rings
+ip-26-0-172-57:942613:942950 [2] NCCL INFO Connected all rings
+ip-26-0-172-73:926601:926934 [7] NCCL INFO Channel 09/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-73:926601:926934 [7] NCCL INFO Channel 10/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-73:926601:926934 [7] NCCL INFO Channel 11/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-73:926600:926933 [6] NCCL INFO Channel 00/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-73:926601:926934 [7] NCCL INFO Channel 12/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-173-7:811255:811595 [7] NCCL INFO Channel 06/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-73:926600:926933 [6] NCCL INFO Channel 01/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-73:926601:926934 [7] NCCL INFO Channel 13/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-73:926600:926933 [6] NCCL INFO Channel 02/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-73:926601:926934 [7] NCCL INFO Channel 14/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-116:1019293:1019625 [5] NCCL INFO Channel 20/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-73:926600:926933 [6] NCCL INFO Channel 03/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-73:926601:926934 [7] NCCL INFO Channel 15/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-73:926600:926933 [6] NCCL INFO Channel 04/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-73:926601:926934 [7] NCCL INFO Channel 16/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-73:926600:926933 [6] NCCL INFO Channel 05/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-73:926601:926934 [7] NCCL INFO Channel 17/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-147:618585:618926 [1] NCCL INFO Connected all rings
+ip-26-0-172-73:926600:926933 [6] NCCL INFO Channel 06/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-73:926601:926934 [7] NCCL INFO Channel 18/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-73:926600:926933 [6] NCCL INFO Channel 07/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-73:926601:926934 [7] NCCL INFO Channel 19/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-73:926600:926933 [6] NCCL INFO Channel 08/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-225:94347:94685 [5] NCCL INFO Channel 18/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-73:926601:926934 [7] NCCL INFO Channel 20/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-73:926600:926933 [6] NCCL INFO Channel 09/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-225:94348:94684 [6] NCCL INFO Channel 14/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-73:926601:926934 [7] NCCL INFO Channel 21/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-73:926600:926933 [6] NCCL INFO Channel 10/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-173-7:811250:811597 [2] NCCL INFO Connected all rings
+ip-26-0-172-73:926601:926934 [7] NCCL INFO Channel 22/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-73:926600:926933 [6] NCCL INFO Channel 11/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-147:618587:618924 [3] NCCL INFO Connected all rings
+ip-26-0-172-116:1019293:1019625 [5] NCCL INFO Channel 21/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-173-7:811248:811591 [0] NCCL INFO Connected all rings
+ip-26-0-172-73:926601:926934 [7] NCCL INFO Channel 23/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-73:926600:926933 [6] NCCL INFO Channel 12/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-225:94347:94685 [5] NCCL INFO Channel 19/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-173-7:811255:811595 [7] NCCL INFO Channel 07/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-73:926600:926933 [6] NCCL INFO Channel 13/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-173-7:811255:811595 [7] NCCL INFO Channel 08/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-73:926600:926933 [6] NCCL INFO Channel 14/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-147:618586:618923 [2] NCCL INFO Connected all rings
+ip-26-0-173-7:811255:811595 [7] NCCL INFO Channel 09/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-73:926600:926933 [6] NCCL INFO Channel 15/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-173-7:811255:811595 [7] NCCL INFO Channel 10/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-160-225:94348:94684 [6] NCCL INFO Channel 15/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-173-7:811255:811595 [7] NCCL INFO Channel 11/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-73:926600:926933 [6] NCCL INFO Channel 16/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-142:880686:881021 [7] NCCL INFO Connected all rings
+ip-26-0-173-7:811255:811595 [7] NCCL INFO Channel 12/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-142:880686:881021 [7] NCCL INFO Channel 00/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-73:926600:926933 [6] NCCL INFO Channel 17/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-73:926595:926937 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-173-7:811255:811595 [7] NCCL INFO Channel 13/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-73:926600:926933 [6] NCCL INFO Channel 18/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-73:926595:926937 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-73:926599:926932 [5] NCCL INFO Channel 00/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-173-7:811255:811595 [7] NCCL INFO Channel 14/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-73:926600:926933 [6] NCCL INFO Channel 19/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-225:94347:94685 [5] NCCL INFO Channel 20/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-73:926595:926937 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-73:926599:926932 [5] NCCL INFO Channel 01/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-73:926595:926937 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-173-7:811255:811595 [7] NCCL INFO Channel 15/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-73:926599:926932 [5] NCCL INFO Channel 02/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-73:926595:926937 [1] NCCL INFO Channel 04/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-73:926599:926932 [5] NCCL INFO Channel 03/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-173-7:811255:811595 [7] NCCL INFO Channel 16/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-73:926595:926937 [1] NCCL INFO Channel 05/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-73:926596:926936 [2] NCCL INFO Channel 00/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-73:926597:926935 [3] NCCL INFO Channel 00/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-73:926599:926932 [5] NCCL INFO Channel 04/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-73:926595:926937 [1] NCCL INFO Channel 06/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-73:926596:926936 [2] NCCL INFO Channel 01/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-173-7:811255:811595 [7] NCCL INFO Channel 17/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-73:926599:926932 [5] NCCL INFO Channel 05/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-73:926597:926935 [3] NCCL INFO Channel 01/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-225:94348:94684 [6] NCCL INFO Channel 16/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-73:926595:926937 [1] NCCL INFO Channel 07/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-73:926596:926936 [2] NCCL INFO Channel 02/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-73:926599:926932 [5] NCCL INFO Channel 06/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-173-7:811255:811595 [7] NCCL INFO Channel 18/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-73:926595:926937 [1] NCCL INFO Channel 08/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-73:926596:926936 [2] NCCL INFO Channel 03/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-73:926599:926932 [5] NCCL INFO Channel 07/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-73:926598:926938 [4] NCCL INFO Channel 00/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-73:926595:926937 [1] NCCL INFO Channel 09/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-73:926596:926936 [2] NCCL INFO Channel 04/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-173-7:811255:811595 [7] NCCL INFO Channel 19/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-73:926599:926932 [5] NCCL INFO Channel 08/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-73:926598:926938 [4] NCCL INFO Channel 01/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-73:926595:926937 [1] NCCL INFO Channel 10/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-73:926596:926936 [2] NCCL INFO Channel 05/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-142:880683:881016 [4] NCCL INFO Connected all rings
+ip-26-0-172-73:926599:926932 [5] NCCL INFO Channel 09/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-73:926598:926938 [4] NCCL INFO Channel 02/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-142:880686:881021 [7] NCCL INFO Channel 01/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-73:926596:926936 [2] NCCL INFO Channel 06/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-173-7:811255:811595 [7] NCCL INFO Channel 20/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-73:926595:926937 [1] NCCL INFO Channel 11/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-73:926599:926932 [5] NCCL INFO Channel 10/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-73:926598:926938 [4] NCCL INFO Channel 03/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-73:926596:926936 [2] NCCL INFO Channel 07/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-73:926595:926937 [1] NCCL INFO Channel 12/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-225:94347:94685 [5] NCCL INFO Channel 21/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-142:880684:881017 [5] NCCL INFO Connected all rings
+ip-26-0-172-116:1019293:1019625 [5] NCCL INFO Channel 22/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-73:926598:926938 [4] NCCL INFO Channel 04/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-173-7:811255:811595 [7] NCCL INFO Channel 21/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-73:926599:926932 [5] NCCL INFO Channel 11/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-73:926596:926936 [2] NCCL INFO Channel 08/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-73:926595:926937 [1] NCCL INFO Channel 13/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-73:926600:926933 [6] NCCL INFO Channel 20/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-73:926598:926938 [4] NCCL INFO Channel 05/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-173-7:811255:811595 [7] NCCL INFO Channel 22/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-73:926599:926932 [5] NCCL INFO Channel 12/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-73:926596:926936 [2] NCCL INFO Channel 09/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-73:926595:926937 [1] NCCL INFO Channel 14/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-73:926600:926933 [6] NCCL INFO Channel 21/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-73:926598:926938 [4] NCCL INFO Channel 06/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-73:926599:926932 [5] NCCL INFO Channel 13/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-173-7:811255:811595 [7] NCCL INFO Channel 23/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-142:880682:881020 [3] NCCL INFO Connected all rings
+ip-26-0-172-73:926596:926936 [2] NCCL INFO Channel 10/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-73:926595:926937 [1] NCCL INFO Channel 15/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-73:926600:926933 [6] NCCL INFO Channel 22/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-73:926598:926938 [4] NCCL INFO Channel 07/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-73:926599:926932 [5] NCCL INFO Channel 14/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-225:94348:94684 [6] NCCL INFO Channel 17/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-73:926596:926936 [2] NCCL INFO Channel 11/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-73:926595:926937 [1] NCCL INFO Channel 16/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-73:926600:926933 [6] NCCL INFO Channel 23/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-147:618584:618919 [0] NCCL INFO Connected all rings
+ip-26-0-172-73:926598:926938 [4] NCCL INFO Channel 08/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-73:926599:926932 [5] NCCL INFO Channel 15/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-57:942617:942949 [6] NCCL INFO Connected all rings
+ip-26-0-172-142:880686:881021 [7] NCCL INFO Channel 02/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-73:926596:926936 [2] NCCL INFO Channel 12/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-73:926595:926937 [1] NCCL INFO Channel 17/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-73:926597:926935 [3] NCCL INFO Channel 02/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-73:926598:926938 [4] NCCL INFO Channel 09/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-73:926599:926932 [5] NCCL INFO Channel 16/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-173-7:811253:811593 [5] NCCL INFO Channel 00/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-173-7:811253:811593 [5] NCCL INFO Channel 01/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-173-7:811254:811594 [6] NCCL INFO Channel 00/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-160-225:94347:94685 [5] NCCL INFO Channel 22/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-73:926597:926935 [3] NCCL INFO Channel 03/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-73:926598:926938 [4] NCCL INFO Channel 10/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-173-7:811253:811593 [5] NCCL INFO Channel 02/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-173-7:811254:811594 [6] NCCL INFO Channel 01/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-142:880680:881022 [1] NCCL INFO Connected all rings
+ip-26-0-172-142:880686:881021 [7] NCCL INFO Channel 03/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-57:942615:942948 [4] NCCL INFO Connected all rings
+ip-26-0-173-7:811253:811593 [5] NCCL INFO Channel 03/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-173-7:811249:811596 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-173-7:811254:811594 [6] NCCL INFO Channel 02/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-73:926596:926936 [2] NCCL INFO Channel 13/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-73:926595:926937 [1] NCCL INFO Channel 18/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-73:926599:926932 [5] NCCL INFO Channel 17/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-173-7:811252:811592 [4] NCCL INFO Channel 00/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-142:880686:881021 [7] NCCL INFO Channel 04/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-173-7:811253:811593 [5] NCCL INFO Channel 04/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-173-7:811249:811596 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-173-7:811254:811594 [6] NCCL INFO Channel 03/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-173-7:811252:811592 [4] NCCL INFO Channel 01/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-173-7:811253:811593 [5] NCCL INFO Channel 05/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-147:618591:618921 [7] NCCL INFO Connected all rings
+ip-26-0-172-147:618591:618921 [7] NCCL INFO Channel 00/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-173-7:811249:811596 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-173-7:811254:811594 [6] NCCL INFO Channel 04/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-173-7:811252:811592 [4] NCCL INFO Channel 02/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-142:880686:881021 [7] NCCL INFO Channel 05/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-73:926597:926935 [3] NCCL INFO Channel 04/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-73:926598:926938 [4] NCCL INFO Channel 11/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-73:926599:926932 [5] NCCL INFO Channel 18/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-73:926595:926937 [1] NCCL INFO Channel 19/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-73:926596:926936 [2] NCCL INFO Channel 14/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-160-225:94348:94684 [6] NCCL INFO Channel 18/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-173-7:811250:811597 [2] NCCL INFO Channel 00/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-173-7:811253:811593 [5] NCCL INFO Channel 06/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-173-7:811249:811596 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-173-7:811254:811594 [6] NCCL INFO Channel 05/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-173-7:811252:811592 [4] NCCL INFO Channel 03/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-147:618591:618921 [7] NCCL INFO Channel 01/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-173-7:811250:811597 [2] NCCL INFO Channel 01/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-173-7:811249:811596 [1] NCCL INFO Channel 04/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-173-7:811253:811593 [5] NCCL INFO Channel 07/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-173-7:811254:811594 [6] NCCL INFO Channel 06/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-173-7:811252:811592 [4] NCCL INFO Channel 04/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-147:618591:618921 [7] NCCL INFO Channel 02/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-173-7:811250:811597 [2] NCCL INFO Channel 02/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-173-7:811253:811593 [5] NCCL INFO Channel 08/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-173-7:811249:811596 [1] NCCL INFO Channel 05/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-173-7:811254:811594 [6] NCCL INFO Channel 07/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-173-7:811252:811592 [4] NCCL INFO Channel 05/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-73:926597:926935 [3] NCCL INFO Channel 05/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-73:926598:926938 [4] NCCL INFO Channel 12/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-73:926599:926932 [5] NCCL INFO Channel 19/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-73:926596:926936 [2] NCCL INFO Channel 15/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-73:926595:926937 [1] NCCL INFO Channel 20/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-173-7:811250:811597 [2] NCCL INFO Channel 03/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-147:618591:618921 [7] NCCL INFO Channel 03/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-173-7:811253:811593 [5] NCCL INFO Channel 09/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-173-7:811249:811596 [1] NCCL INFO Channel 06/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-225:94347:94685 [5] NCCL INFO Channel 23/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-173-7:811254:811594 [6] NCCL INFO Channel 08/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-173-7:811252:811592 [4] NCCL INFO Channel 06/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-142:880686:881021 [7] NCCL INFO Channel 06/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-173-7:811250:811597 [2] NCCL INFO Channel 04/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-142:880686:881021 [7] NCCL INFO Channel 07/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-73:926595:926937 [1] NCCL INFO Channel 21/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-173-7:811253:811593 [5] NCCL INFO Channel 10/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-173-7:811249:811596 [1] NCCL INFO Channel 07/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-147:618591:618921 [7] NCCL INFO Channel 04/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-173-7:811254:811594 [6] NCCL INFO Channel 09/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-173-7:811252:811592 [4] NCCL INFO Channel 07/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-73:926595:926937 [1] NCCL INFO Channel 22/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-73:926595:926937 [1] NCCL INFO Channel 23/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-173-7:811250:811597 [2] NCCL INFO Channel 05/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-147:618591:618921 [7] NCCL INFO Channel 05/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-173-7:811253:811593 [5] NCCL INFO Channel 11/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-173-7:811249:811596 [1] NCCL INFO Channel 08/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-173-7:811254:811594 [6] NCCL INFO Channel 10/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-173-7:811252:811592 [4] NCCL INFO Channel 08/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-225:94348:94684 [6] NCCL INFO Channel 19/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-147:618591:618921 [7] NCCL INFO Channel 06/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-173-7:811250:811597 [2] NCCL INFO Channel 06/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-57:942614:942953 [3] NCCL INFO Connected all rings
+ip-26-0-173-7:811253:811593 [5] NCCL INFO Channel 12/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-173-7:811249:811596 [1] NCCL INFO Channel 09/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-173-7:811252:811592 [4] NCCL INFO Channel 09/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-173-7:811254:811594 [6] NCCL INFO Channel 11/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-147:618591:618921 [7] NCCL INFO Channel 07/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-173-7:811250:811597 [2] NCCL INFO Channel 07/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-147:618590:618925 [6] NCCL INFO Channel 00/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-173-7:811253:811593 [5] NCCL INFO Channel 13/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-173-7:811249:811596 [1] NCCL INFO Channel 10/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-173-7:811252:811592 [4] NCCL INFO Channel 10/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-173-7:811254:811594 [6] NCCL INFO Channel 12/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-147:618591:618921 [7] NCCL INFO Channel 08/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-173-7:811250:811597 [2] NCCL INFO Channel 08/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-147:618590:618925 [6] NCCL INFO Channel 01/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-147:618591:618921 [7] NCCL INFO Channel 09/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-173-7:811253:811593 [5] NCCL INFO Channel 14/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-173-7:811249:811596 [1] NCCL INFO Channel 11/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-173-7:811254:811594 [6] NCCL INFO Channel 13/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-173-7:811252:811592 [4] NCCL INFO Channel 11/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-173-7:811250:811597 [2] NCCL INFO Channel 09/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-147:618590:618925 [6] NCCL INFO Channel 02/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-147:618591:618921 [7] NCCL INFO Channel 10/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-173-7:811253:811593 [5] NCCL INFO Channel 15/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-173-7:811249:811596 [1] NCCL INFO Channel 12/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-173-7:811252:811592 [4] NCCL INFO Channel 12/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-173-7:811254:811594 [6] NCCL INFO Channel 14/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-173-7:811250:811597 [2] NCCL INFO Channel 10/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-73:926597:926935 [3] NCCL INFO Channel 06/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-73:926598:926938 [4] NCCL INFO Channel 13/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-142:880686:881021 [7] NCCL INFO Channel 08/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-73:926596:926936 [2] NCCL INFO Channel 16/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-73:926599:926932 [5] NCCL INFO Channel 20/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-160-225:94348:94684 [6] NCCL INFO Channel 20/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-173-7:811253:811593 [5] NCCL INFO Channel 16/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-173-7:811249:811596 [1] NCCL INFO Channel 13/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-173-7:811252:811592 [4] NCCL INFO Channel 13/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-173-7:811254:811594 [6] NCCL INFO Channel 15/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-73:926597:926935 [3] NCCL INFO Channel 07/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-73:926598:926938 [4] NCCL INFO Channel 14/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-147:618590:618925 [6] NCCL INFO Channel 03/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-173-7:811250:811597 [2] NCCL INFO Channel 11/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-173-7:811251:811598 [3] NCCL INFO Channel 00/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-73:926599:926932 [5] NCCL INFO Channel 21/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-73:926596:926936 [2] NCCL INFO Channel 17/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-142:880686:881021 [7] NCCL INFO Channel 09/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-73:926597:926935 [3] NCCL INFO Channel 08/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-147:618590:618925 [6] NCCL INFO Channel 04/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-173-7:811253:811593 [5] NCCL INFO Channel 17/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-173-7:811249:811596 [1] NCCL INFO Channel 14/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-173-7:811254:811594 [6] NCCL INFO Channel 16/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-173-7:811252:811592 [4] NCCL INFO Channel 14/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-173-7:811250:811597 [2] NCCL INFO Channel 12/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-173-7:811251:811598 [3] NCCL INFO Channel 01/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-147:618590:618925 [6] NCCL INFO Channel 05/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-142:880686:881021 [7] NCCL INFO Channel 10/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-173-7:811253:811593 [5] NCCL INFO Channel 18/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-173-7:811249:811596 [1] NCCL INFO Channel 15/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-173-7:811254:811594 [6] NCCL INFO Channel 17/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-173-7:811252:811592 [4] NCCL INFO Channel 15/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-173-7:811250:811597 [2] NCCL INFO Channel 13/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-173-7:811251:811598 [3] NCCL INFO Channel 02/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-147:618590:618925 [6] NCCL INFO Channel 06/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-142:880686:881021 [7] NCCL INFO Channel 11/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-147:618590:618925 [6] NCCL INFO Channel 07/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-173-7:811253:811593 [5] NCCL INFO Channel 19/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-173-7:811249:811596 [1] NCCL INFO Channel 16/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-173-7:811254:811594 [6] NCCL INFO Channel 18/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-142:880681:881019 [2] NCCL INFO Channel 00/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-173-7:811252:811592 [4] NCCL INFO Channel 16/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-173-7:811251:811598 [3] NCCL INFO Channel 03/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-116:1019293:1019625 [5] NCCL INFO Channel 23/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-173-7:811250:811597 [2] NCCL INFO Channel 14/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-147:618590:618925 [6] NCCL INFO Channel 08/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-142:880686:881021 [7] NCCL INFO Channel 12/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-173-7:811253:811593 [5] NCCL INFO Channel 20/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-173-7:811254:811594 [6] NCCL INFO Channel 19/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-173-7:811249:811596 [1] NCCL INFO Channel 17/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-173-7:811252:811592 [4] NCCL INFO Channel 17/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-173-7:811251:811598 [3] NCCL INFO Channel 04/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-173-7:811250:811597 [2] NCCL INFO Channel 15/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-147:618590:618925 [6] NCCL INFO Channel 09/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-147:618591:618921 [7] NCCL INFO Channel 11/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-173-7:811253:811593 [5] NCCL INFO Channel 21/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-173-7:811254:811594 [6] NCCL INFO Channel 20/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-142:880686:881021 [7] NCCL INFO Channel 13/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-142:880681:881019 [2] NCCL INFO Channel 01/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-173-7:811251:811598 [3] NCCL INFO Channel 05/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-173-7:811249:811596 [1] NCCL INFO Channel 18/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-173-7:811252:811592 [4] NCCL INFO Channel 18/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-173-7:811250:811597 [2] NCCL INFO Channel 16/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-142:880686:881021 [7] NCCL INFO Channel 14/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-142:880681:881019 [2] NCCL INFO Channel 02/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-173-7:811253:811593 [5] NCCL INFO Channel 22/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-173-7:811251:811598 [3] NCCL INFO Channel 06/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-173-7:811249:811596 [1] NCCL INFO Channel 19/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-173-7:811252:811592 [4] NCCL INFO Channel 19/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-173-7:811250:811597 [2] NCCL INFO Channel 17/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-73:926598:926938 [4] NCCL INFO Channel 15/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-142:880686:881021 [7] NCCL INFO Channel 15/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-142:880681:881019 [2] NCCL INFO Channel 03/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-142:880685:881018 [6] NCCL INFO Channel 00/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-173-7:811254:811594 [6] NCCL INFO Channel 21/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-173-7:811253:811593 [5] NCCL INFO Channel 23/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-173-7:811251:811598 [3] NCCL INFO Channel 07/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-173-7:811252:811592 [4] NCCL INFO Channel 20/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-173-7:811249:811596 [1] NCCL INFO Channel 20/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-173-7:811250:811597 [2] NCCL INFO Channel 18/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-142:880686:881021 [7] NCCL INFO Channel 16/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-142:880681:881019 [2] NCCL INFO Channel 04/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-142:880685:881018 [6] NCCL INFO Channel 01/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-173-7:811254:811594 [6] NCCL INFO Channel 22/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-173-7:811251:811598 [3] NCCL INFO Channel 08/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-173-7:811252:811592 [4] NCCL INFO Channel 21/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-173-7:811249:811596 [1] NCCL INFO Channel 21/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-173-7:811250:811597 [2] NCCL INFO Channel 19/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-147:618588:618920 [4] NCCL INFO Channel 00/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-142:880686:881021 [7] NCCL INFO Channel 17/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-142:880681:881019 [2] NCCL INFO Channel 05/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-142:880685:881018 [6] NCCL INFO Channel 02/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-173-7:811254:811594 [6] NCCL INFO Channel 23/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-173-7:811251:811598 [3] NCCL INFO Channel 09/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-173-7:811252:811592 [4] NCCL INFO Channel 22/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-173-7:811249:811596 [1] NCCL INFO Channel 22/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-173-7:811250:811597 [2] NCCL INFO Channel 20/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-252:688207:688543 [7] NCCL INFO Connected all trees
+ip-26-0-172-252:688207:688543 [7] NCCL INFO NVLS comm 0x91e3790 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-147:618591:618921 [7] NCCL INFO Channel 12/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-147:618588:618920 [4] NCCL INFO Channel 01/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-173-7:811251:811598 [3] NCCL INFO Channel 10/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-173-7:811249:811596 [1] NCCL INFO Channel 23/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-173-7:811252:811592 [4] NCCL INFO Channel 23/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-142:880686:881021 [7] NCCL INFO Channel 18/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-173-7:811250:811597 [2] NCCL INFO Channel 21/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-142:880681:881019 [2] NCCL INFO Channel 06/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-142:880685:881018 [6] NCCL INFO Channel 03/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-147:618591:618921 [7] NCCL INFO Channel 13/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-73:926599:926932 [5] NCCL INFO Channel 22/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-73:926596:926936 [2] NCCL INFO Channel 18/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-147:618588:618920 [4] NCCL INFO Channel 02/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-142:880686:881021 [7] NCCL INFO Channel 19/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-142:880681:881019 [2] NCCL INFO Channel 07/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-142:880685:881018 [6] NCCL INFO Channel 04/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-147:618590:618925 [6] NCCL INFO Channel 10/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-147:618591:618921 [7] NCCL INFO Channel 14/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-147:618588:618920 [4] NCCL INFO Channel 03/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-73:926597:926935 [3] NCCL INFO Channel 09/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-225:94348:94684 [6] NCCL INFO Channel 21/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-147:618590:618925 [6] NCCL INFO Channel 11/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-147:618587:618924 [3] NCCL INFO Channel 00/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-147:618591:618921 [7] NCCL INFO Channel 15/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-147:618589:618922 [5] NCCL INFO Channel 00/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-142:880686:881021 [7] NCCL INFO Channel 20/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-142:880681:881019 [2] NCCL INFO Channel 08/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-147:618588:618920 [4] NCCL INFO Channel 04/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-147:618590:618925 [6] NCCL INFO Channel 12/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-142:880686:881021 [7] NCCL INFO Channel 21/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-147:618587:618924 [3] NCCL INFO Channel 01/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-142:880681:881019 [2] NCCL INFO Channel 09/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-57:942612:942954 [1] NCCL INFO Connected all rings
+ip-26-0-172-142:880685:881018 [6] NCCL INFO Channel 05/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-73:926599:926932 [5] NCCL INFO Channel 23/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-147:618588:618920 [4] NCCL INFO Channel 05/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-147:618590:618925 [6] NCCL INFO Channel 13/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-142:880686:881021 [7] NCCL INFO Channel 22/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-147:618587:618924 [3] NCCL INFO Channel 02/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-142:880681:881019 [2] NCCL INFO Channel 10/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-142:880685:881018 [6] NCCL INFO Channel 06/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-147:618588:618920 [4] NCCL INFO Channel 06/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-147:618590:618925 [6] NCCL INFO Channel 14/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-147:618587:618924 [3] NCCL INFO Channel 03/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-147:618588:618920 [4] NCCL INFO Channel 07/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-147:618590:618925 [6] NCCL INFO Channel 15/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-147:618587:618924 [3] NCCL INFO Channel 04/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-147:618588:618920 [4] NCCL INFO Channel 08/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-73:926598:926938 [4] NCCL INFO Channel 16/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-147:618587:618924 [3] NCCL INFO Channel 05/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-147:618590:618925 [6] NCCL INFO Channel 16/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-147:618588:618920 [4] NCCL INFO Channel 09/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-173-7:811250:811597 [2] NCCL INFO Channel 22/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-173-7:811251:811598 [3] NCCL INFO Channel 11/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-147:618587:618924 [3] NCCL INFO Channel 06/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-147:618590:618925 [6] NCCL INFO Channel 17/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-142:880680:881022 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-147:618591:618921 [7] NCCL INFO Channel 16/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-142:880681:881019 [2] NCCL INFO Channel 11/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-147:618589:618922 [5] NCCL INFO Channel 01/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-147:618588:618920 [4] NCCL INFO Channel 10/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-142:880680:881022 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-142:880681:881019 [2] NCCL INFO Channel 12/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-147:618587:618924 [3] NCCL INFO Channel 07/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-147:618590:618925 [6] NCCL INFO Channel 18/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-147:618591:618921 [7] NCCL INFO Channel 17/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-147:618589:618922 [5] NCCL INFO Channel 02/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-147:618585:618926 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-142:880686:881021 [7] NCCL INFO Channel 23/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-142:880680:881022 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-142:880681:881019 [2] NCCL INFO Channel 13/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-147:618588:618920 [4] NCCL INFO Channel 11/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-142:880685:881018 [6] NCCL INFO Channel 07/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-147:618587:618924 [3] NCCL INFO Channel 08/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-147:618590:618925 [6] NCCL INFO Channel 19/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-147:618591:618921 [7] NCCL INFO Channel 18/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-147:618589:618922 [5] NCCL INFO Channel 03/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-142:880680:881022 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-147:618585:618926 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-142:880681:881019 [2] NCCL INFO Channel 14/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-142:880685:881018 [6] NCCL INFO Channel 08/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-147:618588:618920 [4] NCCL INFO Channel 12/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-73:926598:926938 [4] NCCL INFO Channel 17/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-73:926596:926936 [2] NCCL INFO Channel 19/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-147:618590:618925 [6] NCCL INFO Channel 20/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-142:880680:881022 [1] NCCL INFO Channel 04/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-147:618591:618921 [7] NCCL INFO Channel 19/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-147:618589:618922 [5] NCCL INFO Channel 04/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-142:880681:881019 [2] NCCL INFO Channel 15/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-142:880683:881016 [4] NCCL INFO Channel 00/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-142:880685:881018 [6] NCCL INFO Channel 09/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-147:618587:618924 [3] NCCL INFO Channel 09/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-142:880684:881017 [5] NCCL INFO Channel 00/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-147:618585:618926 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-142:880682:881020 [3] NCCL INFO Channel 00/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-173-7:811250:811597 [2] NCCL INFO Channel 23/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-147:618586:618923 [2] NCCL INFO Channel 00/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-147:618588:618920 [4] NCCL INFO Channel 13/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-142:880680:881022 [1] NCCL INFO Channel 05/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-142:880681:881019 [2] NCCL INFO Channel 16/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-147:618590:618925 [6] NCCL INFO Channel 21/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-142:880683:881016 [4] NCCL INFO Channel 01/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-147:618591:618921 [7] NCCL INFO Channel 20/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-142:880685:881018 [6] NCCL INFO Channel 10/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-142:880684:881017 [5] NCCL INFO Channel 01/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-142:880682:881020 [3] NCCL INFO Channel 01/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-147:618589:618922 [5] NCCL INFO Channel 05/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-173-7:811251:811598 [3] NCCL INFO Channel 12/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-147:618587:618924 [3] NCCL INFO Channel 10/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-147:618585:618926 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-147:618586:618923 [2] NCCL INFO Channel 01/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-147:618588:618920 [4] NCCL INFO Channel 14/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-142:880680:881022 [1] NCCL INFO Channel 06/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-142:880683:881016 [4] NCCL INFO Channel 02/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-142:880685:881018 [6] NCCL INFO Channel 11/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-142:880682:881020 [3] NCCL INFO Channel 02/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-142:880684:881017 [5] NCCL INFO Channel 02/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-147:618590:618925 [6] NCCL INFO Channel 22/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-147:618591:618921 [7] NCCL INFO Channel 21/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-147:618589:618922 [5] NCCL INFO Channel 06/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-147:618587:618924 [3] NCCL INFO Channel 11/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-147:618585:618926 [1] NCCL INFO Channel 04/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-142:880680:881022 [1] NCCL INFO Channel 07/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-142:880683:881016 [4] NCCL INFO Channel 03/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-142:880682:881020 [3] NCCL INFO Channel 03/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-142:880685:881018 [6] NCCL INFO Channel 12/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-142:880684:881017 [5] NCCL INFO Channel 03/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-147:618586:618923 [2] NCCL INFO Channel 02/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-73:926597:926935 [3] NCCL INFO Channel 10/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-147:618588:618920 [4] NCCL INFO Channel 15/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-142:880681:881019 [2] NCCL INFO Channel 17/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-73:926597:926935 [3] NCCL INFO Channel 11/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-147:618590:618925 [6] NCCL INFO Channel 23/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-73:926597:926935 [3] NCCL INFO Channel 12/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-147:618591:618921 [7] NCCL INFO Channel 22/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-142:880680:881022 [1] NCCL INFO Channel 08/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-147:618589:618922 [5] NCCL INFO Channel 07/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-73:926597:926935 [3] NCCL INFO Channel 13/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-142:880683:881016 [4] NCCL INFO Channel 04/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-142:880682:881020 [3] NCCL INFO Channel 04/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-142:880685:881018 [6] NCCL INFO Channel 13/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-147:618587:618924 [3] NCCL INFO Channel 12/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-142:880684:881017 [5] NCCL INFO Channel 04/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-147:618585:618926 [1] NCCL INFO Channel 05/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-73:926597:926935 [3] NCCL INFO Channel 14/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-142:880681:881019 [2] NCCL INFO Channel 18/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-147:618586:618923 [2] NCCL INFO Channel 03/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-147:618588:618920 [4] NCCL INFO Channel 16/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-73:926597:926935 [3] NCCL INFO Channel 15/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-73:926597:926935 [3] NCCL INFO Channel 16/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-73:926597:926935 [3] NCCL INFO Channel 17/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-142:880680:881022 [1] NCCL INFO Channel 09/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-142:880683:881016 [4] NCCL INFO Channel 05/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-142:880685:881018 [6] NCCL INFO Channel 14/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-142:880684:881017 [5] NCCL INFO Channel 05/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-142:880681:881019 [2] NCCL INFO Channel 19/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-142:880682:881020 [3] NCCL INFO Channel 05/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-73:926597:926935 [3] NCCL INFO Channel 18/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-147:618587:618924 [3] NCCL INFO Channel 13/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-147:618586:618923 [2] NCCL INFO Channel 04/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-142:880680:881022 [1] NCCL INFO Channel 10/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-142:880685:881018 [6] NCCL INFO Channel 15/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-142:880683:881016 [4] NCCL INFO Channel 06/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-173-7:811251:811598 [3] NCCL INFO Channel 13/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-147:618588:618920 [4] NCCL INFO Channel 17/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-142:880684:881017 [5] NCCL INFO Channel 06/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-142:880681:881019 [2] NCCL INFO Channel 20/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-147:618587:618924 [3] NCCL INFO Channel 14/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-173-7:811251:811598 [3] NCCL INFO Channel 14/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-142:880682:881020 [3] NCCL INFO Channel 06/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-147:618586:618923 [2] NCCL INFO Channel 05/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-173-7:811251:811598 [3] NCCL INFO Channel 15/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-147:618588:618920 [4] NCCL INFO Channel 18/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-147:618585:618926 [1] NCCL INFO Channel 06/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-173-7:811251:811598 [3] NCCL INFO Channel 16/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-147:618587:618924 [3] NCCL INFO Channel 15/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-147:618586:618923 [2] NCCL INFO Channel 06/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-142:880680:881022 [1] NCCL INFO Channel 11/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-142:880683:881016 [4] NCCL INFO Channel 07/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-142:880685:881018 [6] NCCL INFO Channel 16/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-142:880684:881017 [5] NCCL INFO Channel 07/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-142:880681:881019 [2] NCCL INFO Channel 21/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-142:880682:881020 [3] NCCL INFO Channel 07/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-147:618588:618920 [4] NCCL INFO Channel 19/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-147:618585:618926 [1] NCCL INFO Channel 07/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-160-225:94348:94684 [6] NCCL INFO Channel 22/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-147:618587:618924 [3] NCCL INFO Channel 16/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-147:618586:618923 [2] NCCL INFO Channel 07/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-57:942611:942947 [0] NCCL INFO Connected all rings
+ip-26-0-172-147:618588:618920 [4] NCCL INFO Channel 20/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-142:880680:881022 [1] NCCL INFO Channel 12/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-147:618585:618926 [1] NCCL INFO Channel 08/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-142:880683:881016 [4] NCCL INFO Channel 08/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-142:880685:881018 [6] NCCL INFO Channel 17/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-142:880684:881017 [5] NCCL INFO Channel 08/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-142:880681:881019 [2] NCCL INFO Channel 22/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-142:880682:881020 [3] NCCL INFO Channel 08/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-147:618587:618924 [3] NCCL INFO Channel 17/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-147:618586:618923 [2] NCCL INFO Channel 08/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-73:926598:926938 [4] NCCL INFO Channel 18/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-73:926596:926936 [2] NCCL INFO Channel 20/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-147:618588:618920 [4] NCCL INFO Channel 21/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-147:618585:618926 [1] NCCL INFO Channel 09/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-147:618587:618924 [3] NCCL INFO Channel 18/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-142:880680:881022 [1] NCCL INFO Channel 13/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-147:618586:618923 [2] NCCL INFO Channel 09/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-142:880683:881016 [4] NCCL INFO Channel 09/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-142:880685:881018 [6] NCCL INFO Channel 18/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-142:880684:881017 [5] NCCL INFO Channel 09/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-142:880681:881019 [2] NCCL INFO Channel 23/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-142:880682:881020 [3] NCCL INFO Channel 09/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-147:618588:618920 [4] NCCL INFO Channel 22/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-147:618591:618921 [7] NCCL INFO Channel 23/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-147:618585:618926 [1] NCCL INFO Channel 10/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-147:618589:618922 [5] NCCL INFO Channel 08/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-147:618587:618924 [3] NCCL INFO Channel 19/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-147:618586:618923 [2] NCCL INFO Channel 10/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-142:880680:881022 [1] NCCL INFO Channel 14/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-142:880683:881016 [4] NCCL INFO Channel 10/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-142:880685:881018 [6] NCCL INFO Channel 19/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-142:880684:881017 [5] NCCL INFO Channel 10/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-142:880682:881020 [3] NCCL INFO Channel 10/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-147:618588:618920 [4] NCCL INFO Channel 23/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-147:618585:618926 [1] NCCL INFO Channel 11/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-147:618589:618922 [5] NCCL INFO Channel 09/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-147:618587:618924 [3] NCCL INFO Channel 20/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-142:880680:881022 [1] NCCL INFO Channel 15/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-142:880683:881016 [4] NCCL INFO Channel 11/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-142:880684:881017 [5] NCCL INFO Channel 11/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-142:880685:881018 [6] NCCL INFO Channel 20/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-142:880682:881020 [3] NCCL INFO Channel 11/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-147:618586:618923 [2] NCCL INFO Channel 11/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-142:880683:881016 [4] NCCL INFO Channel 12/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-142:880680:881022 [1] NCCL INFO Channel 16/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-142:880685:881018 [6] NCCL INFO Channel 21/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-142:880684:881017 [5] NCCL INFO Channel 12/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-142:880682:881020 [3] NCCL INFO Channel 12/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-147:618589:618922 [5] NCCL INFO Channel 10/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-147:618585:618926 [1] NCCL INFO Channel 12/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-142:880683:881016 [4] NCCL INFO Channel 13/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-142:880680:881022 [1] NCCL INFO Channel 17/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-142:880684:881017 [5] NCCL INFO Channel 13/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-142:880685:881018 [6] NCCL INFO Channel 22/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-142:880682:881020 [3] NCCL INFO Channel 13/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-73:926597:926935 [3] NCCL INFO Channel 19/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-142:880683:881016 [4] NCCL INFO Channel 14/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-142:880680:881022 [1] NCCL INFO Channel 18/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-142:880684:881017 [5] NCCL INFO Channel 14/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-142:880685:881018 [6] NCCL INFO Channel 23/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-142:880682:881020 [3] NCCL INFO Channel 14/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-147:618587:618924 [3] NCCL INFO Channel 21/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-147:618586:618923 [2] NCCL INFO Channel 12/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-57:942618:942951 [7] NCCL INFO Connected all rings
+ip-26-0-160-225:94348:94684 [6] NCCL INFO Channel 23/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-57:942618:942951 [7] NCCL INFO Channel 00/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-142:880683:881016 [4] NCCL INFO Channel 15/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-142:880680:881022 [1] NCCL INFO Channel 19/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-142:880684:881017 [5] NCCL INFO Channel 15/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-142:880682:881020 [3] NCCL INFO Channel 15/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-73:926598:926938 [4] NCCL INFO Channel 19/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-57:942618:942951 [7] NCCL INFO Channel 01/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-147:618589:618922 [5] NCCL INFO Channel 11/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-147:618585:618926 [1] NCCL INFO Channel 13/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-147:618586:618923 [2] NCCL INFO Channel 13/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-147:618587:618924 [3] NCCL INFO Channel 22/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-173-7:811251:811598 [3] NCCL INFO Channel 17/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-142:880683:881016 [4] NCCL INFO Channel 16/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-142:880680:881022 [1] NCCL INFO Channel 20/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-142:880684:881017 [5] NCCL INFO Channel 16/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-142:880682:881020 [3] NCCL INFO Channel 16/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-57:942618:942951 [7] NCCL INFO Channel 02/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-73:926596:926936 [2] NCCL INFO Channel 21/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-116:1019295:1019627 [7] NCCL INFO Connected all trees
+ip-26-0-172-116:1019295:1019627 [7] NCCL INFO NVLS comm 0x9c6b820 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-147:618589:618922 [5] NCCL INFO Channel 12/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-57:942618:942951 [7] NCCL INFO Channel 03/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-147:618585:618926 [1] NCCL INFO Channel 14/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-57:942618:942951 [7] NCCL INFO Channel 04/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-147:618586:618923 [2] NCCL INFO Channel 14/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-147:618587:618924 [3] NCCL INFO Channel 23/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-142:880680:881022 [1] NCCL INFO Channel 21/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-142:880683:881016 [4] NCCL INFO Channel 17/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-142:880684:881017 [5] NCCL INFO Channel 17/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-142:880682:881020 [3] NCCL INFO Channel 17/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-73:926597:926935 [3] NCCL INFO Channel 20/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-57:942618:942951 [7] NCCL INFO Channel 05/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-147:618589:618922 [5] NCCL INFO Channel 13/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-147:618585:618926 [1] NCCL INFO Channel 15/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-73:926598:926938 [4] NCCL INFO Channel 20/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-57:942618:942951 [7] NCCL INFO Channel 06/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-147:618586:618923 [2] NCCL INFO Channel 15/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-142:880680:881022 [1] NCCL INFO Channel 22/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-142:880683:881016 [4] NCCL INFO Channel 18/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-142:880684:881017 [5] NCCL INFO Channel 18/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-142:880682:881020 [3] NCCL INFO Channel 18/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-57:942618:942951 [7] NCCL INFO Channel 07/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-73:926596:926936 [2] NCCL INFO Channel 22/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-147:618589:618922 [5] NCCL INFO Channel 14/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-147:618585:618926 [1] NCCL INFO Channel 16/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-147:618586:618923 [2] NCCL INFO Channel 16/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-57:942618:942951 [7] NCCL INFO Channel 08/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-73:926597:926935 [3] NCCL INFO Channel 21/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-57:942618:942951 [7] NCCL INFO Channel 09/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-73:926598:926938 [4] NCCL INFO Channel 21/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-57:942618:942951 [7] NCCL INFO Channel 10/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-73:926596:926936 [2] NCCL INFO Channel 23/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-147:618589:618922 [5] NCCL INFO Channel 15/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-147:618586:618923 [2] NCCL INFO Channel 17/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-147:618585:618926 [1] NCCL INFO Channel 17/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-57:942618:942951 [7] NCCL INFO Channel 11/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-142:880683:881016 [4] NCCL INFO Channel 19/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-142:880680:881022 [1] NCCL INFO Channel 23/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-142:880682:881020 [3] NCCL INFO Channel 19/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-142:880684:881017 [5] NCCL INFO Channel 19/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-57:942618:942951 [7] NCCL INFO Channel 12/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-173-7:811251:811598 [3] NCCL INFO Channel 18/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-57:942618:942951 [7] NCCL INFO Channel 13/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-57:942618:942951 [7] NCCL INFO Channel 14/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-57:942616:942952 [5] NCCL INFO Channel 00/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-147:618589:618922 [5] NCCL INFO Channel 16/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-147:618586:618923 [2] NCCL INFO Channel 18/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-147:618585:618926 [1] NCCL INFO Channel 18/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-73:926597:926935 [3] NCCL INFO Channel 22/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-142:880683:881016 [4] NCCL INFO Channel 20/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-142:880684:881017 [5] NCCL INFO Channel 20/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-142:880682:881020 [3] NCCL INFO Channel 20/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-142:880683:881016 [4] NCCL INFO Channel 21/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-57:942616:942952 [5] NCCL INFO Channel 01/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-57:942616:942952 [5] NCCL INFO Channel 02/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-57:942616:942952 [5] NCCL INFO Channel 03/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-147:618586:618923 [2] NCCL INFO Channel 19/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-73:926598:926938 [4] NCCL INFO Channel 22/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-57:942616:942952 [5] NCCL INFO Channel 04/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-142:880683:881016 [4] NCCL INFO Channel 22/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-147:618586:618923 [2] NCCL INFO Channel 20/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-147:618585:618926 [1] NCCL INFO Channel 19/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-142:880684:881017 [5] NCCL INFO Channel 21/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-142:880682:881020 [3] NCCL INFO Channel 21/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-142:880683:881016 [4] NCCL INFO Channel 23/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-57:942618:942951 [7] NCCL INFO Channel 15/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-57:942616:942952 [5] NCCL INFO Channel 05/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-73:926597:926935 [3] NCCL INFO Channel 23/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-57:942618:942951 [7] NCCL INFO Channel 16/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-57:942616:942952 [5] NCCL INFO Channel 06/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-147:618586:618923 [2] NCCL INFO Channel 21/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-57:942618:942951 [7] NCCL INFO Channel 17/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-57:942616:942952 [5] NCCL INFO Channel 07/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-57:942614:942953 [3] NCCL INFO Channel 00/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-57:942618:942951 [7] NCCL INFO Channel 18/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-73:926601:926934 [7] NCCL INFO Connected all trees
+ip-26-0-172-57:942613:942950 [2] NCCL INFO Channel 00/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-57:942616:942952 [5] NCCL INFO Channel 08/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-73:926601:926934 [7] NCCL INFO NVLS comm 0x884d470 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-57:942614:942953 [3] NCCL INFO Channel 01/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-147:618589:618922 [5] NCCL INFO Channel 17/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-57:942618:942951 [7] NCCL INFO Channel 19/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-57:942613:942950 [2] NCCL INFO Channel 01/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-142:880684:881017 [5] NCCL INFO Channel 22/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-142:880682:881020 [3] NCCL INFO Channel 22/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-57:942616:942952 [5] NCCL INFO Channel 09/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-57:942614:942953 [3] NCCL INFO Channel 02/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-57:942618:942951 [7] NCCL INFO Channel 20/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-57:942613:942950 [2] NCCL INFO Channel 02/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-57:942616:942952 [5] NCCL INFO Channel 10/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-147:618586:618923 [2] NCCL INFO Channel 22/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-57:942612:942954 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-57:942614:942953 [3] NCCL INFO Channel 03/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-57:942618:942951 [7] NCCL INFO Channel 21/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-147:618589:618922 [5] NCCL INFO Channel 18/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-57:942613:942950 [2] NCCL INFO Channel 03/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-57:942616:942952 [5] NCCL INFO Channel 11/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-73:926598:926938 [4] NCCL INFO Channel 23/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-57:942612:942954 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-57:942614:942953 [3] NCCL INFO Channel 04/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-57:942618:942951 [7] NCCL INFO Channel 22/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-57:942617:942949 [6] NCCL INFO Channel 00/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-57:942613:942950 [2] NCCL INFO Channel 04/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-57:942616:942952 [5] NCCL INFO Channel 12/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-147:618585:618926 [1] NCCL INFO Channel 20/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-147:618586:618923 [2] NCCL INFO Channel 23/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-147:618589:618922 [5] NCCL INFO Channel 19/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-57:942615:942948 [4] NCCL INFO Channel 00/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-57:942612:942954 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-57:942614:942953 [3] NCCL INFO Channel 05/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-142:880684:881017 [5] NCCL INFO Channel 23/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-142:880682:881020 [3] NCCL INFO Channel 23/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-57:942618:942951 [7] NCCL INFO Channel 23/0 : 7[7] -> 6[6] via P2P/IPC
+ip-26-0-172-57:942617:942949 [6] NCCL INFO Channel 01/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-57:942613:942950 [2] NCCL INFO Channel 05/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-57:942616:942952 [5] NCCL INFO Channel 13/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-57:942615:942948 [4] NCCL INFO Channel 01/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-57:942612:942954 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-57:942614:942953 [3] NCCL INFO Channel 06/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-57:942613:942950 [2] NCCL INFO Channel 06/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-57:942617:942949 [6] NCCL INFO Channel 02/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-57:942616:942952 [5] NCCL INFO Channel 14/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-147:618585:618926 [1] NCCL INFO Channel 21/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-147:618589:618922 [5] NCCL INFO Channel 20/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-57:942612:942954 [1] NCCL INFO Channel 04/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-57:942615:942948 [4] NCCL INFO Channel 02/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-57:942614:942953 [3] NCCL INFO Channel 07/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-57:942617:942949 [6] NCCL INFO Channel 03/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-57:942613:942950 [2] NCCL INFO Channel 07/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-57:942616:942952 [5] NCCL INFO Channel 15/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-57:942612:942954 [1] NCCL INFO Channel 05/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-57:942614:942953 [3] NCCL INFO Channel 08/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-57:942615:942948 [4] NCCL INFO Channel 03/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-147:618591:618921 [7] NCCL INFO Connected all trees
+ip-26-0-172-147:618591:618921 [7] NCCL INFO NVLS comm 0x963e350 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-147:618589:618922 [5] NCCL INFO Channel 21/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-147:618585:618926 [1] NCCL INFO Channel 22/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-57:942615:942948 [4] NCCL INFO Channel 04/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-57:942617:942949 [6] NCCL INFO Channel 04/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-57:942616:942952 [5] NCCL INFO Channel 16/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-57:942613:942950 [2] NCCL INFO Channel 08/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-57:942614:942953 [3] NCCL INFO Channel 09/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-57:942612:942954 [1] NCCL INFO Channel 06/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-57:942615:942948 [4] NCCL INFO Channel 05/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-57:942617:942949 [6] NCCL INFO Channel 05/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-173-7:811251:811598 [3] NCCL INFO Channel 19/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-147:618589:618922 [5] NCCL INFO Channel 22/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-147:618585:618926 [1] NCCL INFO Channel 23/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-57:942613:942950 [2] NCCL INFO Channel 09/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-57:942616:942952 [5] NCCL INFO Channel 17/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-57:942614:942953 [3] NCCL INFO Channel 10/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-57:942612:942954 [1] NCCL INFO Channel 07/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-57:942615:942948 [4] NCCL INFO Channel 06/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-57:942613:942950 [2] NCCL INFO Channel 10/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-57:942616:942952 [5] NCCL INFO Channel 18/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-57:942614:942953 [3] NCCL INFO Channel 11/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-57:942612:942954 [1] NCCL INFO Channel 08/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-57:942617:942949 [6] NCCL INFO Channel 06/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-57:942615:942948 [4] NCCL INFO Channel 07/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-57:942616:942952 [5] NCCL INFO Channel 19/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-57:942613:942950 [2] NCCL INFO Channel 11/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-57:942614:942953 [3] NCCL INFO Channel 12/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-57:942612:942954 [1] NCCL INFO Channel 09/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-147:618589:618922 [5] NCCL INFO Channel 23/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-57:942617:942949 [6] NCCL INFO Channel 07/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-57:942615:942948 [4] NCCL INFO Channel 08/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-57:942616:942952 [5] NCCL INFO Channel 20/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-57:942614:942953 [3] NCCL INFO Channel 13/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-57:942613:942950 [2] NCCL INFO Channel 12/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-57:942612:942954 [1] NCCL INFO Channel 10/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-57:942617:942949 [6] NCCL INFO Channel 08/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-57:942615:942948 [4] NCCL INFO Channel 09/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-57:942614:942953 [3] NCCL INFO Channel 14/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-57:942616:942952 [5] NCCL INFO Channel 21/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-57:942613:942950 [2] NCCL INFO Channel 13/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-57:942612:942954 [1] NCCL INFO Channel 11/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-57:942617:942949 [6] NCCL INFO Channel 09/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-57:942615:942948 [4] NCCL INFO Channel 10/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-57:942614:942953 [3] NCCL INFO Channel 15/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-57:942613:942950 [2] NCCL INFO Channel 14/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-57:942616:942952 [5] NCCL INFO Channel 22/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-57:942612:942954 [1] NCCL INFO Channel 12/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-57:942617:942949 [6] NCCL INFO Channel 10/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-57:942614:942953 [3] NCCL INFO Channel 16/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-57:942615:942948 [4] NCCL INFO Channel 11/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-57:942613:942950 [2] NCCL INFO Channel 15/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-57:942616:942952 [5] NCCL INFO Channel 23/0 : 5[5] -> 4[4] via P2P/IPC
+ip-26-0-172-57:942612:942954 [1] NCCL INFO Channel 13/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-57:942617:942949 [6] NCCL INFO Channel 11/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-57:942614:942953 [3] NCCL INFO Channel 17/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-57:942615:942948 [4] NCCL INFO Channel 12/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-57:942613:942950 [2] NCCL INFO Channel 16/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-57:942612:942954 [1] NCCL INFO Channel 14/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-57:942614:942953 [3] NCCL INFO Channel 18/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-57:942617:942949 [6] NCCL INFO Channel 12/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-57:942613:942950 [2] NCCL INFO Channel 17/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-57:942612:942954 [1] NCCL INFO Channel 15/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-57:942612:942954 [1] NCCL INFO Channel 16/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-57:942613:942950 [2] NCCL INFO Channel 18/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-173-7:811251:811598 [3] NCCL INFO Channel 20/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-57:942614:942953 [3] NCCL INFO Channel 19/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-57:942617:942949 [6] NCCL INFO Channel 13/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-57:942612:942954 [1] NCCL INFO Channel 17/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-57:942613:942950 [2] NCCL INFO Channel 19/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-57:942614:942953 [3] NCCL INFO Channel 20/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-57:942615:942948 [4] NCCL INFO Channel 13/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-57:942617:942949 [6] NCCL INFO Channel 14/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-57:942612:942954 [1] NCCL INFO Channel 18/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-57:942613:942950 [2] NCCL INFO Channel 20/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-57:942614:942953 [3] NCCL INFO Channel 21/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-57:942615:942948 [4] NCCL INFO Channel 14/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-57:942617:942949 [6] NCCL INFO Channel 15/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-57:942612:942954 [1] NCCL INFO Channel 19/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-57:942613:942950 [2] NCCL INFO Channel 21/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-57:942614:942953 [3] NCCL INFO Channel 22/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-57:942615:942948 [4] NCCL INFO Channel 15/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-57:942617:942949 [6] NCCL INFO Channel 16/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-57:942612:942954 [1] NCCL INFO Channel 20/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-57:942614:942953 [3] NCCL INFO Channel 23/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-57:942613:942950 [2] NCCL INFO Channel 22/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-57:942615:942948 [4] NCCL INFO Channel 16/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-57:942617:942949 [6] NCCL INFO Channel 17/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-57:942612:942954 [1] NCCL INFO Channel 21/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-57:942613:942950 [2] NCCL INFO Channel 23/0 : 2[2] -> 1[1] via P2P/IPC
+ip-26-0-172-57:942615:942948 [4] NCCL INFO Channel 17/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-57:942617:942949 [6] NCCL INFO Channel 18/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-57:942612:942954 [1] NCCL INFO Channel 22/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-57:942615:942948 [4] NCCL INFO Channel 18/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-57:942617:942949 [6] NCCL INFO Channel 19/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-57:942612:942954 [1] NCCL INFO Channel 23/0 : 1[1] -> 0[0] via P2P/IPC
+ip-26-0-172-57:942617:942949 [6] NCCL INFO Channel 20/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-57:942615:942948 [4] NCCL INFO Channel 19/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-57:942617:942949 [6] NCCL INFO Channel 21/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-57:942615:942948 [4] NCCL INFO Channel 20/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-225:94343:94681 [1] NCCL INFO Connected all trees
+ip-26-0-160-225:94343:94681 [1] NCCL INFO NVLS comm 0x87612b0 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-173-7:811251:811598 [3] NCCL INFO Channel 21/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-160-225:94345:94687 [3] NCCL INFO Connected all trees
+ip-26-0-172-57:942617:942949 [6] NCCL INFO Channel 22/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-57:942615:942948 [4] NCCL INFO Channel 21/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-160-225:94345:94687 [3] NCCL INFO NVLS comm 0xa6960e0 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-116:1019288:1019623 [0] NCCL INFO Connected all trees
+ip-26-0-172-116:1019288:1019623 [0] NCCL INFO NVLS comm 0x8e255e0 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-173-7:811251:811598 [3] NCCL INFO Channel 22/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-173-7:811251:811598 [3] NCCL INFO Channel 23/0 : 3[3] -> 2[2] via P2P/IPC
+ip-26-0-172-57:942617:942949 [6] NCCL INFO Channel 23/0 : 6[6] -> 5[5] via P2P/IPC
+ip-26-0-172-57:942615:942948 [4] NCCL INFO Channel 22/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-57:942615:942948 [4] NCCL INFO Channel 23/0 : 4[4] -> 3[3] via P2P/IPC
+ip-26-0-172-73:926594:926931 [0] NCCL INFO Connected all trees
+ip-26-0-172-73:926594:926931 [0] NCCL INFO NVLS comm 0x967ca20 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-142:880686:881021 [7] NCCL INFO Connected all trees
+ip-26-0-172-142:880686:881021 [7] NCCL INFO NVLS comm 0x99c84d0 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-252:688200:688538 [0] NCCL INFO Connected all trees
+ip-26-0-172-252:688200:688538 [0] NCCL INFO NVLS comm 0x9672c10 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-173-7:811255:811595 [7] NCCL INFO Connected all trees
+ip-26-0-173-7:811255:811595 [7] NCCL INFO NVLS comm 0xa04ac50 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-225:94344:94686 [2] NCCL INFO Connected all trees
+ip-26-0-160-225:94344:94686 [2] NCCL INFO NVLS comm 0x94b6540 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-173-7:811248:811591 [0] NCCL INFO Connected all trees
+ip-26-0-173-7:811248:811591 [0] NCCL INFO NVLS comm 0xa5f0580 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-225:94349:94683 [7] NCCL INFO Connected all trees
+ip-26-0-160-225:94349:94683 [7] NCCL INFO NVLS comm 0x8e619f0 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-252:688202:688544 [2] NCCL INFO Connected all trees
+ip-26-0-172-252:688202:688544 [2] NCCL INFO NVLS comm 0x94fef70 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-116:1019290:1019628 [2] NCCL INFO Connected all trees
+ip-26-0-172-116:1019290:1019628 [2] NCCL INFO NVLS comm 0xa28b4a0 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-142:880679:881015 [0] NCCL INFO Connected all trees
+ip-26-0-172-142:880679:881015 [0] NCCL INFO NVLS comm 0xa3bc420 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-225:94346:94682 [4] NCCL INFO Connected all trees
+ip-26-0-160-225:94346:94682 [4] NCCL INFO NVLS comm 0xa4faf60 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-252:688205:688545 [5] NCCL INFO Connected all trees
+ip-26-0-172-252:688205:688545 [5] NCCL INFO NVLS comm 0x90dc4c0 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-57:942611:942947 [0] NCCL INFO Connected all trees
+ip-26-0-172-57:942611:942947 [0] NCCL INFO NVLS comm 0xa0430a0 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-225:94348:94684 [6] NCCL INFO Connected all trees
+ip-26-0-160-225:94348:94684 [6] NCCL INFO NVLS comm 0x9c5c6f0 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-252:688204:688540 [4] NCCL INFO Connected all trees
+ip-26-0-172-252:688204:688540 [4] NCCL INFO NVLS comm 0x976f3c0 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-252:688201:688542 [1] NCCL INFO Connected all trees
+ip-26-0-160-225:94347:94685 [5] NCCL INFO Connected all trees
+ip-26-0-172-252:688201:688542 [1] NCCL INFO NVLS comm 0x9619490 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-225:94347:94685 [5] NCCL INFO NVLS comm 0x9590450 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-252:688203:688541 [3] NCCL INFO Connected all trees
+ip-26-0-172-116:1019291:1019630 [3] NCCL INFO Connected all trees
+ip-26-0-172-252:688203:688541 [3] NCCL INFO NVLS comm 0x8a6ed40 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-116:1019291:1019630 [3] NCCL INFO NVLS comm 0xa6143c0 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-116:1019289:1019629 [1] NCCL INFO Connected all trees
+ip-26-0-172-116:1019289:1019629 [1] NCCL INFO NVLS comm 0x889ef90 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-252:688206:688539 [6] NCCL INFO Connected all trees
+ip-26-0-172-252:688206:688539 [6] NCCL INFO NVLS comm 0xa6c7860 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-147:618584:618919 [0] NCCL INFO Connected all trees
+ip-26-0-172-147:618584:618919 [0] NCCL INFO NVLS comm 0x9d77ab0 headRank 0 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-116:1019293:1019625 [5] NCCL INFO Connected all trees
+ip-26-0-172-116:1019293:1019625 [5] NCCL INFO NVLS comm 0xa6a1880 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-116:1019292:1019624 [4] NCCL INFO Connected all trees
+ip-26-0-172-116:1019292:1019624 [4] NCCL INFO NVLS comm 0xa6017e0 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-57:942618:942951 [7] NCCL INFO Connected all trees
+ip-26-0-172-57:942618:942951 [7] NCCL INFO NVLS comm 0x929bad0 headRank 7 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-116:1019294:1019626 [6] NCCL INFO Connected all trees
+ip-26-0-172-116:1019294:1019626 [6] NCCL INFO NVLS comm 0x889d980 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-73:926600:926933 [6] NCCL INFO Connected all trees
+ip-26-0-172-73:926600:926933 [6] NCCL INFO NVLS comm 0xa278570 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-173-7:811254:811594 [6] NCCL INFO Connected all trees
+ip-26-0-173-7:811254:811594 [6] NCCL INFO NVLS comm 0x8e04770 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-160-225:94342:94680 [0] NCCL INFO Connected NVLS tree
+ip-26-0-160-225:94342:94680 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-225:94349:94683 [7] NCCL INFO Connected NVLS tree
+ip-26-0-160-225:94349:94683 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-225:94349:94683 [7] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-160-225:94349:94683 [7] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-160-225:94342:94680 [0] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-160-225:94342:94680 [0] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-160-225:94346:94682 [4] NCCL INFO Connected NVLS tree
+ip-26-0-160-225:94346:94682 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-225:94346:94682 [4] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-160-225:94346:94682 [4] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-160-225:94343:94681 [1] NCCL INFO Connected NVLS tree
+ip-26-0-160-225:94343:94681 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-225:94343:94681 [1] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-160-225:94343:94681 [1] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-160-225:94348:94684 [6] NCCL INFO Connected NVLS tree
+ip-26-0-160-225:94348:94684 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-225:94348:94684 [6] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-160-225:94348:94684 [6] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-160-225:94347:94685 [5] NCCL INFO Connected NVLS tree
+ip-26-0-160-225:94347:94685 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-225:94347:94685 [5] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-160-225:94347:94685 [5] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-160-225:94344:94686 [2] NCCL INFO Connected NVLS tree
+ip-26-0-160-225:94344:94686 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-225:94344:94686 [2] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-160-225:94344:94686 [2] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-160-225:94345:94687 [3] NCCL INFO Connected NVLS tree
+ip-26-0-160-225:94345:94687 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-225:94345:94687 [3] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-160-225:94345:94687 [3] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-173-7:811249:811596 [1] NCCL INFO Connected all trees
+ip-26-0-173-7:811249:811596 [1] NCCL INFO NVLS comm 0x8bffbd0 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-173-7:811253:811593 [5] NCCL INFO Connected all trees
+ip-26-0-173-7:811253:811593 [5] NCCL INFO NVLS comm 0x9a6e690 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-252:688206:688539 [6] NCCL INFO Connected NVLS tree
+ip-26-0-172-252:688206:688539 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-252:688204:688540 [4] NCCL INFO Connected NVLS tree
+ip-26-0-172-252:688206:688539 [6] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-172-252:688204:688540 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-252:688206:688539 [6] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-252:688204:688540 [4] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-172-252:688204:688540 [4] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-252:688200:688538 [0] NCCL INFO Connected NVLS tree
+ip-26-0-172-252:688200:688538 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-252:688200:688538 [0] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-172-252:688200:688538 [0] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-252:688203:688541 [3] NCCL INFO Connected NVLS tree
+ip-26-0-172-252:688203:688541 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-252:688201:688542 [1] NCCL INFO Connected NVLS tree
+ip-26-0-172-252:688203:688541 [3] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-172-252:688203:688541 [3] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-252:688201:688542 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-252:688201:688542 [1] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-172-252:688201:688542 [1] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-252:688207:688543 [7] NCCL INFO Connected NVLS tree
+ip-26-0-172-252:688207:688543 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-252:688207:688543 [7] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-172-252:688207:688543 [7] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-252:688202:688544 [2] NCCL INFO Connected NVLS tree
+ip-26-0-172-252:688202:688544 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-252:688202:688544 [2] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-172-252:688202:688544 [2] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-252:688205:688545 [5] NCCL INFO Connected NVLS tree
+ip-26-0-172-252:688205:688545 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-252:688205:688545 [5] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-172-252:688205:688545 [5] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-160-225:94343:94681 [1] NCCL INFO comm 0x87612b0 rank 1 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0x3ef1a0376dc31f7 - Init COMPLETE
+ip-26-0-160-225:94349:94683 [7] NCCL INFO comm 0x8e619f0 rank 7 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0x3ef1a0376dc31f7 - Init COMPLETE
+ip-26-0-160-225:94345:94687 [3] NCCL INFO comm 0xa6960e0 rank 3 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0x3ef1a0376dc31f7 - Init COMPLETE
+ip-26-0-160-225:94348:94684 [6] NCCL INFO comm 0x9c5c6f0 rank 6 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0x3ef1a0376dc31f7 - Init COMPLETE
+ip-26-0-160-225:94347:94685 [5] NCCL INFO comm 0x9590450 rank 5 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0x3ef1a0376dc31f7 - Init COMPLETE
+ip-26-0-160-225:94344:94686 [2] NCCL INFO comm 0x94b6540 rank 2 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0x3ef1a0376dc31f7 - Init COMPLETE
+ip-26-0-160-225:94342:94680 [0] NCCL INFO comm 0x93142a0 rank 0 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0x3ef1a0376dc31f7 - Init COMPLETE
+ip-26-0-160-225:94346:94682 [4] NCCL INFO comm 0xa4faf60 rank 4 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0x3ef1a0376dc31f7 - Init COMPLETE
+ip-26-0-172-142:880681:881019 [2] NCCL INFO Connected all trees
+ip-26-0-172-142:880681:881019 [2] NCCL INFO NVLS comm 0x92db6c0 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-252:688204:688540 [4] NCCL INFO comm 0x976f3c0 rank 4 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0xb7c059e1a5ed57c - Init COMPLETE
+ip-26-0-172-252:688200:688538 [0] NCCL INFO comm 0x9672c10 rank 0 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0xb7c059e1a5ed57c - Init COMPLETE
+ip-26-0-172-252:688203:688541 [3] NCCL INFO comm 0x8a6ed40 rank 3 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0xb7c059e1a5ed57c - Init COMPLETE
+ip-26-0-172-252:688206:688539 [6] NCCL INFO comm 0xa6c7860 rank 6 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0xb7c059e1a5ed57c - Init COMPLETE
+ip-26-0-172-252:688202:688544 [2] NCCL INFO comm 0x94fef70 rank 2 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0xb7c059e1a5ed57c - Init COMPLETE
+ip-26-0-172-252:688207:688543 [7] NCCL INFO comm 0x91e3790 rank 7 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0xb7c059e1a5ed57c - Init COMPLETE
+ip-26-0-172-252:688201:688542 [1] NCCL INFO comm 0x9619490 rank 1 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0xb7c059e1a5ed57c - Init COMPLETE
+ip-26-0-172-252:688205:688545 [5] NCCL INFO comm 0x90dc4c0 rank 5 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0xb7c059e1a5ed57c - Init COMPLETE
+ip-26-0-172-142:880680:881022 [1] NCCL INFO Connected all trees
+ip-26-0-172-142:880680:881022 [1] NCCL INFO NVLS comm 0x9213170 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-116:1019294:1019626 [6] NCCL INFO Connected NVLS tree
+ip-26-0-172-116:1019292:1019624 [4] NCCL INFO Connected NVLS tree
+ip-26-0-172-116:1019294:1019626 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-116:1019292:1019624 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-116:1019292:1019624 [4] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-172-116:1019292:1019624 [4] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-116:1019294:1019626 [6] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-172-116:1019294:1019626 [6] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-116:1019288:1019623 [0] NCCL INFO Connected NVLS tree
+ip-26-0-172-116:1019288:1019623 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-116:1019290:1019628 [2] NCCL INFO Connected NVLS tree
+ip-26-0-172-116:1019290:1019628 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-116:1019290:1019628 [2] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-172-116:1019290:1019628 [2] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-116:1019288:1019623 [0] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-172-116:1019288:1019623 [0] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-116:1019293:1019625 [5] NCCL INFO Connected NVLS tree
+ip-26-0-172-116:1019293:1019625 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-116:1019293:1019625 [5] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-172-116:1019293:1019625 [5] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-116:1019289:1019629 [1] NCCL INFO Connected NVLS tree
+ip-26-0-172-116:1019289:1019629 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-116:1019289:1019629 [1] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-172-116:1019289:1019629 [1] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-116:1019291:1019630 [3] NCCL INFO Connected NVLS tree
+ip-26-0-172-116:1019291:1019630 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-116:1019291:1019630 [3] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-172-116:1019291:1019630 [3] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-116:1019295:1019627 [7] NCCL INFO Connected NVLS tree
+ip-26-0-172-116:1019295:1019627 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-116:1019295:1019627 [7] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-172-116:1019295:1019627 [7] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-57:942612:942954 [1] NCCL INFO Connected all trees
+ip-26-0-172-57:942612:942954 [1] NCCL INFO NVLS comm 0x8e01660 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-173-7:811251:811598 [3] NCCL INFO Connected all trees
+ip-26-0-173-7:811251:811598 [3] NCCL INFO NVLS comm 0x8ffae80 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-116:1019290:1019628 [2] NCCL INFO comm 0xa28b4a0 rank 2 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0xe87bb15d3c873f23 - Init COMPLETE
+ip-26-0-172-116:1019288:1019623 [0] NCCL INFO comm 0x8e255e0 rank 0 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0xe87bb15d3c873f23 - Init COMPLETE
+ip-26-0-172-116:1019294:1019626 [6] NCCL INFO comm 0x889d980 rank 6 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0xe87bb15d3c873f23 - Init COMPLETE
+ip-26-0-172-116:1019289:1019629 [1] NCCL INFO comm 0x889ef90 rank 1 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0xe87bb15d3c873f23 - Init COMPLETE
+ip-26-0-172-116:1019292:1019624 [4] NCCL INFO comm 0xa6017e0 rank 4 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0xe87bb15d3c873f23 - Init COMPLETE
+ip-26-0-172-116:1019293:1019625 [5] NCCL INFO comm 0xa6a1880 rank 5 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0xe87bb15d3c873f23 - Init COMPLETE
+ip-26-0-172-116:1019295:1019627 [7] NCCL INFO comm 0x9c6b820 rank 7 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0xe87bb15d3c873f23 - Init COMPLETE
+ip-26-0-172-116:1019291:1019630 [3] NCCL INFO comm 0xa6143c0 rank 3 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0xe87bb15d3c873f23 - Init COMPLETE
+ip-26-0-173-7:811250:811597 [2] NCCL INFO Connected all trees
+ip-26-0-173-7:811250:811597 [2] NCCL INFO NVLS comm 0x99fce90 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-173-7:811252:811592 [4] NCCL INFO Connected all trees
+ip-26-0-173-7:811252:811592 [4] NCCL INFO NVLS comm 0x911a140 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-147:618585:618926 [1] NCCL INFO Connected all trees
+ip-26-0-172-147:618585:618926 [1] NCCL INFO NVLS comm 0xa287aa0 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-147:618587:618924 [3] NCCL INFO Connected all trees
+ip-26-0-172-147:618587:618924 [3] NCCL INFO NVLS comm 0x8e05020 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-142:880683:881016 [4] NCCL INFO Connected all trees
+ip-26-0-172-142:880683:881016 [4] NCCL INFO NVLS comm 0x9ef9b70 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-57:942613:942950 [2] NCCL INFO Connected all trees
+ip-26-0-172-57:942613:942950 [2] NCCL INFO NVLS comm 0x9dc3d50 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-142:880684:881017 [5] NCCL INFO Connected all trees
+ip-26-0-172-142:880684:881017 [5] NCCL INFO NVLS comm 0x8d1d420 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-147:618588:618920 [4] NCCL INFO Connected all trees
+ip-26-0-172-147:618588:618920 [4] NCCL INFO NVLS comm 0x8ac5d80 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-147:618586:618923 [2] NCCL INFO Connected all trees
+ip-26-0-172-147:618586:618923 [2] NCCL INFO NVLS comm 0x8db71a0 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-142:880685:881018 [6] NCCL INFO Connected all trees
+ip-26-0-172-142:880685:881018 [6] NCCL INFO NVLS comm 0x8bd2f30 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-142:880682:881020 [3] NCCL INFO Connected all trees
+ip-26-0-172-142:880682:881020 [3] NCCL INFO NVLS comm 0x977b900 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-57:942616:942952 [5] NCCL INFO Connected all trees
+ip-26-0-172-57:942616:942952 [5] NCCL INFO NVLS comm 0x892e730 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-147:618589:618922 [5] NCCL INFO Connected all trees
+ip-26-0-172-147:618589:618922 [5] NCCL INFO NVLS comm 0x9a98e60 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-147:618590:618925 [6] NCCL INFO Connected all trees
+ip-26-0-172-147:618590:618925 [6] NCCL INFO NVLS comm 0xa4275b0 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-57:942614:942953 [3] NCCL INFO Connected all trees
+ip-26-0-172-57:942614:942953 [3] NCCL INFO NVLS comm 0x8c33510 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-73:926596:926936 [2] NCCL INFO Connected all trees
+ip-26-0-172-73:926596:926936 [2] NCCL INFO NVLS comm 0x90c2ad0 headRank 2 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-173-7:811248:811591 [0] NCCL INFO Connected NVLS tree
+ip-26-0-173-7:811248:811591 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-173-7:811252:811592 [4] NCCL INFO Connected NVLS tree
+ip-26-0-173-7:811252:811592 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-173-7:811252:811592 [4] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-173-7:811252:811592 [4] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-173-7:811254:811594 [6] NCCL INFO Connected NVLS tree
+ip-26-0-173-7:811248:811591 [0] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-173-7:811248:811591 [0] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-173-7:811254:811594 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-173-7:811254:811594 [6] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-173-7:811254:811594 [6] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-173-7:811250:811597 [2] NCCL INFO Connected NVLS tree
+ip-26-0-173-7:811250:811597 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-173-7:811250:811597 [2] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-173-7:811250:811597 [2] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-173-7:811255:811595 [7] NCCL INFO Connected NVLS tree
+ip-26-0-173-7:811253:811593 [5] NCCL INFO Connected NVLS tree
+ip-26-0-173-7:811255:811595 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-173-7:811253:811593 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-173-7:811255:811595 [7] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-173-7:811255:811595 [7] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-173-7:811253:811593 [5] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-173-7:811253:811593 [5] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-173-7:811249:811596 [1] NCCL INFO Connected NVLS tree
+ip-26-0-173-7:811249:811596 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-173-7:811249:811596 [1] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-173-7:811249:811596 [1] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-173-7:811251:811598 [3] NCCL INFO Connected NVLS tree
+ip-26-0-173-7:811251:811598 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-173-7:811251:811598 [3] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-173-7:811251:811598 [3] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-57:942615:942948 [4] NCCL INFO Connected all trees
+ip-26-0-172-57:942615:942948 [4] NCCL INFO NVLS comm 0xa3a8410 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-57:942617:942949 [6] NCCL INFO Connected all trees
+ip-26-0-172-57:942617:942949 [6] NCCL INFO NVLS comm 0x9a0fe60 headRank 6 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-173-7:811250:811597 [2] NCCL INFO comm 0x99fce90 rank 2 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0xb6d3276631c21f81 - Init COMPLETE
+ip-26-0-173-7:811252:811592 [4] NCCL INFO comm 0x911a140 rank 4 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0xb6d3276631c21f81 - Init COMPLETE
+ip-26-0-173-7:811254:811594 [6] NCCL INFO comm 0x8e04770 rank 6 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0xb6d3276631c21f81 - Init COMPLETE
+ip-26-0-173-7:811248:811591 [0] NCCL INFO comm 0xa5f0580 rank 0 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0xb6d3276631c21f81 - Init COMPLETE
+ip-26-0-173-7:811249:811596 [1] NCCL INFO comm 0x8bffbd0 rank 1 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0xb6d3276631c21f81 - Init COMPLETE
+ip-26-0-173-7:811253:811593 [5] NCCL INFO comm 0x9a6e690 rank 5 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0xb6d3276631c21f81 - Init COMPLETE
+ip-26-0-173-7:811255:811595 [7] NCCL INFO comm 0xa04ac50 rank 7 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0xb6d3276631c21f81 - Init COMPLETE
+ip-26-0-173-7:811251:811598 [3] NCCL INFO comm 0x8ffae80 rank 3 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0xb6d3276631c21f81 - Init COMPLETE
+ip-26-0-172-142:880683:881016 [4] NCCL INFO Connected NVLS tree
+ip-26-0-172-142:880683:881016 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-142:880683:881016 [4] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-172-142:880683:881016 [4] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-142:880685:881018 [6] NCCL INFO Connected NVLS tree
+ip-26-0-172-142:880685:881018 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-142:880685:881018 [6] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-172-142:880685:881018 [6] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-142:880679:881015 [0] NCCL INFO Connected NVLS tree
+ip-26-0-172-142:880679:881015 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-142:880679:881015 [0] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-172-142:880679:881015 [0] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-142:880684:881017 [5] NCCL INFO Connected NVLS tree
+ip-26-0-172-142:880684:881017 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-142:880680:881022 [1] NCCL INFO Connected NVLS tree
+ip-26-0-172-142:880684:881017 [5] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-172-142:880680:881022 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-142:880684:881017 [5] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-142:880680:881022 [1] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-172-142:880686:881021 [7] NCCL INFO Connected NVLS tree
+ip-26-0-172-142:880680:881022 [1] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-142:880682:881020 [3] NCCL INFO Connected NVLS tree
+ip-26-0-172-142:880686:881021 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-142:880682:881020 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-142:880686:881021 [7] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-172-142:880686:881021 [7] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-142:880682:881020 [3] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-172-142:880682:881020 [3] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-142:880681:881019 [2] NCCL INFO Connected NVLS tree
+ip-26-0-172-142:880681:881019 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-142:880681:881019 [2] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-172-142:880681:881019 [2] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-73:926599:926932 [5] NCCL INFO Connected all trees
+ip-26-0-172-73:926599:926932 [5] NCCL INFO NVLS comm 0xa2685e0 headRank 5 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-147:618589:618922 [5] NCCL INFO Connected NVLS tree
+ip-26-0-172-147:618589:618922 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-147:618589:618922 [5] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-172-147:618589:618922 [5] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-147:618588:618920 [4] NCCL INFO Connected NVLS tree
+ip-26-0-172-147:618588:618920 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-147:618588:618920 [4] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-172-147:618588:618920 [4] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-147:618591:618921 [7] NCCL INFO Connected NVLS tree
+ip-26-0-172-147:618591:618921 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-147:618591:618921 [7] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-172-147:618591:618921 [7] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-147:618590:618925 [6] NCCL INFO Connected NVLS tree
+ip-26-0-172-147:618584:618919 [0] NCCL INFO Connected NVLS tree
+ip-26-0-172-147:618590:618925 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-147:618590:618925 [6] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-172-147:618590:618925 [6] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-147:618584:618919 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-147:618584:618919 [0] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-172-147:618584:618919 [0] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-147:618586:618923 [2] NCCL INFO Connected NVLS tree
+ip-26-0-172-147:618585:618926 [1] NCCL INFO Connected NVLS tree
+ip-26-0-172-147:618586:618923 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-147:618585:618926 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-147:618586:618923 [2] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-172-147:618586:618923 [2] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-147:618585:618926 [1] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-172-147:618585:618926 [1] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-147:618587:618924 [3] NCCL INFO Connected NVLS tree
+ip-26-0-172-147:618587:618924 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-147:618587:618924 [3] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-172-147:618587:618924 [3] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-73:926597:926935 [3] NCCL INFO Connected all trees
+ip-26-0-172-73:926597:926935 [3] NCCL INFO NVLS comm 0x8f50bb0 headRank 3 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-73:926595:926937 [1] NCCL INFO Connected all trees
+ip-26-0-172-142:880683:881016 [4] NCCL INFO comm 0x9ef9b70 rank 4 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0x68a686842a70745a - Init COMPLETE
+ip-26-0-172-142:880685:881018 [6] NCCL INFO comm 0x8bd2f30 rank 6 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0x68a686842a70745a - Init COMPLETE
+ip-26-0-172-142:880679:881015 [0] NCCL INFO comm 0xa3bc420 rank 0 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0x68a686842a70745a - Init COMPLETE
+ip-26-0-172-142:880681:881019 [2] NCCL INFO comm 0x92db6c0 rank 2 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0x68a686842a70745a - Init COMPLETE
+ip-26-0-172-142:880684:881017 [5] NCCL INFO comm 0x8d1d420 rank 5 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0x68a686842a70745a - Init COMPLETE
+ip-26-0-172-142:880680:881022 [1] NCCL INFO comm 0x9213170 rank 1 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0x68a686842a70745a - Init COMPLETE
+ip-26-0-172-142:880682:881020 [3] NCCL INFO comm 0x977b900 rank 3 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0x68a686842a70745a - Init COMPLETE
+ip-26-0-172-142:880686:881021 [7] NCCL INFO comm 0x99c84d0 rank 7 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0x68a686842a70745a - Init COMPLETE
+ip-26-0-172-73:926595:926937 [1] NCCL INFO NVLS comm 0x8771e90 headRank 1 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-73:926598:926938 [4] NCCL INFO Connected all trees
+ip-26-0-172-73:926598:926938 [4] NCCL INFO NVLS comm 0x9bba460 headRank 4 nHeads 8 buffSize 4194304 memSize 2097152 nvlsPerRankSize 201326592 nvlsTotalSize 1610612736
+ip-26-0-172-147:618586:618923 [2] NCCL INFO comm 0x8db71a0 rank 2 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0xe0cb904ba16faf13 - Init COMPLETE
+ip-26-0-172-147:618590:618925 [6] NCCL INFO comm 0xa4275b0 rank 6 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0xe0cb904ba16faf13 - Init COMPLETE
+ip-26-0-172-147:618584:618919 [0] NCCL INFO comm 0x9d77ab0 rank 0 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0xe0cb904ba16faf13 - Init COMPLETE
+ip-26-0-172-147:618588:618920 [4] NCCL INFO comm 0x8ac5d80 rank 4 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0xe0cb904ba16faf13 - Init COMPLETE
+ip-26-0-172-147:618591:618921 [7] NCCL INFO comm 0x963e350 rank 7 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0xe0cb904ba16faf13 - Init COMPLETE
+ip-26-0-172-147:618589:618922 [5] NCCL INFO comm 0x9a98e60 rank 5 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0xe0cb904ba16faf13 - Init COMPLETE
+ip-26-0-172-147:618585:618926 [1] NCCL INFO comm 0xa287aa0 rank 1 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0xe0cb904ba16faf13 - Init COMPLETE
+ip-26-0-172-147:618587:618924 [3] NCCL INFO comm 0x8e05020 rank 3 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0xe0cb904ba16faf13 - Init COMPLETE
+ip-26-0-172-57:942617:942949 [6] NCCL INFO Connected NVLS tree
+ip-26-0-172-57:942617:942949 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-57:942617:942949 [6] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-172-57:942617:942949 [6] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-57:942615:942948 [4] NCCL INFO Connected NVLS tree
+ip-26-0-172-57:942615:942948 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-57:942615:942948 [4] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-172-57:942615:942948 [4] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-57:942611:942947 [0] NCCL INFO Connected NVLS tree
+ip-26-0-172-57:942611:942947 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-57:942611:942947 [0] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-172-57:942611:942947 [0] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-57:942616:942952 [5] NCCL INFO Connected NVLS tree
+ip-26-0-172-57:942616:942952 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-57:942616:942952 [5] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-172-57:942616:942952 [5] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-57:942613:942950 [2] NCCL INFO Connected NVLS tree
+ip-26-0-172-57:942613:942950 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-57:942614:942953 [3] NCCL INFO Connected NVLS tree
+ip-26-0-172-57:942613:942950 [2] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-172-57:942614:942953 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-57:942613:942950 [2] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-57:942614:942953 [3] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-172-57:942614:942953 [3] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-57:942612:942954 [1] NCCL INFO Connected NVLS tree
+ip-26-0-172-57:942612:942954 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-57:942612:942954 [1] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-172-57:942612:942954 [1] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-57:942618:942951 [7] NCCL INFO Connected NVLS tree
+ip-26-0-172-57:942618:942951 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-57:942618:942951 [7] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-172-57:942618:942951 [7] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-57:942617:942949 [6] NCCL INFO comm 0x9a0fe60 rank 6 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0xcab7e2994d17fa8a - Init COMPLETE
+ip-26-0-172-57:942611:942947 [0] NCCL INFO comm 0xa0430a0 rank 0 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0xcab7e2994d17fa8a - Init COMPLETE
+ip-26-0-172-57:942615:942948 [4] NCCL INFO comm 0xa3a8410 rank 4 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0xcab7e2994d17fa8a - Init COMPLETE
+ip-26-0-172-57:942618:942951 [7] NCCL INFO comm 0x929bad0 rank 7 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0xcab7e2994d17fa8a - Init COMPLETE
+ip-26-0-172-57:942613:942950 [2] NCCL INFO comm 0x9dc3d50 rank 2 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0xcab7e2994d17fa8a - Init COMPLETE
+ip-26-0-172-57:942614:942953 [3] NCCL INFO comm 0x8c33510 rank 3 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0xcab7e2994d17fa8a - Init COMPLETE
+ip-26-0-172-57:942616:942952 [5] NCCL INFO comm 0x892e730 rank 5 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0xcab7e2994d17fa8a - Init COMPLETE
+ip-26-0-172-57:942612:942954 [1] NCCL INFO comm 0x8e01660 rank 1 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0xcab7e2994d17fa8a - Init COMPLETE
+ip-26-0-172-73:926600:926933 [6] NCCL INFO Connected NVLS tree
+ip-26-0-172-73:926600:926933 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-73:926600:926933 [6] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-172-73:926600:926933 [6] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-73:926598:926938 [4] NCCL INFO Connected NVLS tree
+ip-26-0-172-73:926598:926938 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-73:926598:926938 [4] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-172-73:926598:926938 [4] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-73:926596:926936 [2] NCCL INFO Connected NVLS tree
+ip-26-0-172-73:926596:926936 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-73:926596:926936 [2] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-172-73:926596:926936 [2] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-73:926599:926932 [5] NCCL INFO Connected NVLS tree
+ip-26-0-172-73:926599:926932 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-73:926599:926932 [5] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-172-73:926599:926932 [5] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-73:926595:926937 [1] NCCL INFO Connected NVLS tree
+ip-26-0-172-73:926595:926937 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-73:926595:926937 [1] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-172-73:926595:926937 [1] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-73:926601:926934 [7] NCCL INFO Connected NVLS tree
+ip-26-0-172-73:926601:926934 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-73:926601:926934 [7] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-172-73:926601:926934 [7] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-73:926597:926935 [3] NCCL INFO Connected NVLS tree
+ip-26-0-172-73:926597:926935 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-73:926597:926935 [3] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-172-73:926597:926935 [3] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-73:926594:926931 [0] NCCL INFO Connected NVLS tree
+ip-26-0-172-73:926594:926931 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-73:926594:926931 [0] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-172-73:926594:926931 [0] NCCL INFO 24 coll channels, 16 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-73:926600:926933 [6] NCCL INFO comm 0xa278570 rank 6 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0x553420158d220013 - Init COMPLETE
+ip-26-0-172-73:926598:926938 [4] NCCL INFO comm 0x9bba460 rank 4 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0x553420158d220013 - Init COMPLETE
+ip-26-0-172-73:926596:926936 [2] NCCL INFO comm 0x90c2ad0 rank 2 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0x553420158d220013 - Init COMPLETE
+ip-26-0-172-73:926594:926931 [0] NCCL INFO comm 0x967ca20 rank 0 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0x553420158d220013 - Init COMPLETE
+ip-26-0-172-73:926601:926934 [7] NCCL INFO comm 0x884d470 rank 7 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0x553420158d220013 - Init COMPLETE
+ip-26-0-172-73:926599:926932 [5] NCCL INFO comm 0xa2685e0 rank 5 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0x553420158d220013 - Init COMPLETE
+ip-26-0-172-73:926595:926937 [1] NCCL INFO comm 0x8771e90 rank 1 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0x553420158d220013 - Init COMPLETE
+ip-26-0-172-73:926597:926935 [3] NCCL INFO comm 0x8f50bb0 rank 3 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0x553420158d220013 - Init COMPLETE
+01/07/2025 05:27:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: Config:
+01/07/2025 05:27:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: Config(general=GeneralArgs(project='debug',
+01/07/2025 05:27:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                            run='3.57G_dp8_tp1_pp8_acc16_mbs2_seq4096_zero1_tpmodeRED_vocab131k',
+01/07/2025 05:27:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                            seed=42,
+01/07/2025 05:27:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                            step=None,
+01/07/2025 05:27:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                            consumed_train_samples=None,
+01/07/2025 05:27:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                            benchmark_csv_path=PosixPath('benchmark/results/bench_final2.csv'),
+01/07/2025 05:27:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                            ignore_sanity_checks=True),
+01/07/2025 05:27:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:        parallelism=ParallelismArgs(dp=8,
+01/07/2025 05:27:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                    pp=8,
+01/07/2025 05:27:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                    tp=1,
+01/07/2025 05:27:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                    pp_engine=<nanotron.parallel.pipeline_parallel.engine.OneForwardOneBackwardPipelineEngine object at 0x7fc887ef4c40>,
+01/07/2025 05:27:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                    tp_mode=<TensorParallelLinearMode.REDUCE_SCATTER: 2>,
+01/07/2025 05:27:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                    tp_linear_async_communication=True,
+01/07/2025 05:27:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                    recompute_layer=False,
+01/07/2025 05:27:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                    tp_recompute_allgather=True,
+01/07/2025 05:27:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                    expert_parallel_size=1),
+01/07/2025 05:27:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:        model=ModelArgs(model_config=LlamaConfig(bos_token_id=0,
+01/07/2025 05:27:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 eos_token_id=0,
+01/07/2025 05:27:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 hidden_act='silu',
+01/07/2025 05:27:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 hidden_size=3072,
+01/07/2025 05:27:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 initializer_range=0.02,
+01/07/2025 05:27:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 intermediate_size=8192,
+01/07/2025 05:27:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 is_llama_config=True,
+01/07/2025 05:27:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 max_position_embeddings=4096,
+01/07/2025 05:27:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 num_attention_heads=32,
+01/07/2025 05:27:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 num_hidden_layers=28,
+01/07/2025 05:27:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 num_key_value_heads=32,
+01/07/2025 05:27:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 pad_token_id=None,
+01/07/2025 05:27:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 pretraining_tp=1,
+01/07/2025 05:27:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 rms_norm_eps=1e-05,
+01/07/2025 05:27:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 rope_scaling=None,
+01/07/2025 05:27:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 rope_theta=10000.0,
+01/07/2025 05:27:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 rope_interleaved=False,
+01/07/2025 05:27:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 tie_word_embeddings=True,
+01/07/2025 05:27:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 use_cache=True,
+01/07/2025 05:27:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                 vocab_size=131072),
+01/07/2025 05:27:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                        init_method=RandomInit(std=0.02),
+01/07/2025 05:27:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                        dtype=torch.bfloat16,
+01/07/2025 05:27:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                        make_vocab_size_divisible_by=1,
+01/07/2025 05:27:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                        ddp_bucket_cap_mb=25),
+01/07/2025 05:27:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:        tokenizer=TokenizerArgs(tokenizer_name_or_path='robot-test/dummy-tokenizer-wordlevel',
+01/07/2025 05:27:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                tokenizer_revision=None,
+01/07/2025 05:27:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                tokenizer_max_length=None),
+01/07/2025 05:27:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:        checkpoints=CheckpointsArgs(checkpoints_path=PosixPath('checkpoints'),
+01/07/2025 05:27:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                    checkpoint_interval=10000,
+01/07/2025 05:27:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                    save_initial_state=False,
+01/07/2025 05:27:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                    save_final_state=False,
+01/07/2025 05:27:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                    resume_checkpoint_path=None,
+01/07/2025 05:27:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                    checkpoints_path_is_shared_file_system=False),
+01/07/2025 05:27:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:        logging=LoggingArgs(log_level='info',
+01/07/2025 05:27:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                            log_level_replica='info',
+01/07/2025 05:27:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                            iteration_step_info_interval=1),
+01/07/2025 05:27:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:        tokens=TokensArgs(sequence_length=4096,
+01/07/2025 05:27:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                          train_steps=100,
+01/07/2025 05:27:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                          micro_batch_size=2,
+01/07/2025 05:27:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                          batch_accumulation_per_replica=16,
+01/07/2025 05:27:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                          val_check_interval=100,
+01/07/2025 05:27:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                          limit_val_batches=0,
+01/07/2025 05:27:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                          limit_test_batches=0),
+01/07/2025 05:27:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:        optimizer=OptimizerArgs(optimizer_factory=AdamWOptimizerArgs(adam_eps=1e-08,
+01/07/2025 05:27:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                                     adam_beta1=0.9,
+01/07/2025 05:27:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                                     adam_beta2=0.95,
+01/07/2025 05:27:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                                     torch_adam_is_fused=True,
+01/07/2025 05:27:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                                     name='adamW'),
+01/07/2025 05:27:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                zero_stage=1,
+01/07/2025 05:27:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                weight_decay=0.01,
+01/07/2025 05:27:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                clip_grad=1.0,
+01/07/2025 05:27:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                accumulate_grad_in_fp32=True,
+01/07/2025 05:27:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                learning_rate_scheduler=LRSchedulerArgs(learning_rate=0.0003,
+01/07/2025 05:27:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                                        lr_warmup_steps=2,
+01/07/2025 05:27:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                                        lr_warmup_style='linear',
+01/07/2025 05:27:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                                        lr_decay_style='cosine',
+01/07/2025 05:27:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                                        lr_decay_steps=13,
+01/07/2025 05:27:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                                        lr_decay_starting_step=None,
+01/07/2025 05:27:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                                        min_decay_lr=1e-05)),
+01/07/2025 05:27:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:        data_stages=[DatasetStageArgs(name='Stable Training Stage',
+01/07/2025 05:27:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                      start_training_step=1,
+01/07/2025 05:27:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                      data=DataArgs(dataset=None,
+01/07/2025 05:27:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                    seed=42,
+01/07/2025 05:27:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:                                                    num_loading_workers=1))],
+01/07/2025 05:27:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:        profiler=None,
+01/07/2025 05:27:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:        lighteval=None,
+01/07/2025 05:27:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:        s3_upload=None)
+01/07/2025 05:27:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: Model Config:
+01/07/2025 05:27:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: LlamaConfig(bos_token_id=0,
+01/07/2025 05:27:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             eos_token_id=0,
+01/07/2025 05:27:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             hidden_act='silu',
+01/07/2025 05:27:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             hidden_size=3072,
+01/07/2025 05:27:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             initializer_range=0.02,
+01/07/2025 05:27:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             intermediate_size=8192,
+01/07/2025 05:27:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             is_llama_config=True,
+01/07/2025 05:27:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             max_position_embeddings=4096,
+01/07/2025 05:27:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             num_attention_heads=32,
+01/07/2025 05:27:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             num_hidden_layers=28,
+01/07/2025 05:27:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             num_key_value_heads=32,
+01/07/2025 05:27:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             pad_token_id=None,
+01/07/2025 05:27:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             pretraining_tp=1,
+01/07/2025 05:27:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             rms_norm_eps=1e-05,
+01/07/2025 05:27:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             rope_scaling=None,
+01/07/2025 05:27:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             rope_theta=10000.0,
+01/07/2025 05:27:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             rope_interleaved=False,
+01/07/2025 05:27:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             tie_word_embeddings=True,
+01/07/2025 05:27:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             use_cache=True,
+01/07/2025 05:27:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:             vocab_size=131072)
+01/07/2025 05:27:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: Building model..
+01/07/2025 05:27:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: Initialize RoPE Theta = 10000.0
+01/07/2025 05:27:18 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: Setting PP block ranks...
+01/07/2025 05:27:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.token_position_embeddings | PP: 0/8 | Block rank: 0
+01/07/2025 05:27:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.0 | PP: 0/8 | Block rank: 1
+01/07/2025 05:27:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.1 | PP: 0/8 | Block rank: 2
+01/07/2025 05:27:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.2 | PP: 0/8 | Block rank: 3
+01/07/2025 05:27:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.3 | PP: 0/8 | Block rank: 4
+01/07/2025 05:27:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.4 | PP: 1/8 | Block rank: 0
+01/07/2025 05:27:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.5 | PP: 1/8 | Block rank: 1
+01/07/2025 05:27:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.6 | PP: 1/8 | Block rank: 2
+01/07/2025 05:27:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.7 | PP: 1/8 | Block rank: 3
+01/07/2025 05:27:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.8 | PP: 2/8 | Block rank: 0
+01/07/2025 05:27:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.9 | PP: 2/8 | Block rank: 1
+01/07/2025 05:27:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.10 | PP: 2/8 | Block rank: 2
+01/07/2025 05:27:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.11 | PP: 2/8 | Block rank: 3
+01/07/2025 05:27:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.12 | PP: 3/8 | Block rank: 0
+01/07/2025 05:27:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.13 | PP: 3/8 | Block rank: 1
+01/07/2025 05:27:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.14 | PP: 3/8 | Block rank: 2
+01/07/2025 05:27:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.15 | PP: 3/8 | Block rank: 3
+01/07/2025 05:27:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.16 | PP: 4/8 | Block rank: 0
+01/07/2025 05:27:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.17 | PP: 4/8 | Block rank: 1
+01/07/2025 05:27:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.18 | PP: 4/8 | Block rank: 2
+01/07/2025 05:27:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.19 | PP: 4/8 | Block rank: 3
+01/07/2025 05:27:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.20 | PP: 5/8 | Block rank: 0
+01/07/2025 05:27:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.21 | PP: 5/8 | Block rank: 1
+01/07/2025 05:27:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.22 | PP: 5/8 | Block rank: 2
+01/07/2025 05:27:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.23 | PP: 5/8 | Block rank: 3
+01/07/2025 05:27:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.24 | PP: 6/8 | Block rank: 0
+01/07/2025 05:27:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.25 | PP: 6/8 | Block rank: 1
+01/07/2025 05:27:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.26 | PP: 6/8 | Block rank: 2
+01/07/2025 05:27:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.decoder.27 | PP: 6/8 | Block rank: 3
+01/07/2025 05:27:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.final_layer_norm | PP: 7/8 | Block rank: 0
+01/07/2025 05:27:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.lm_head | PP: 7/8 | Block rank: 1
+01/07/2025 05:27:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: model.cast_to_fp32 | PP: 7/8 | Block rank: 2
+01/07/2025 05:27:19 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: module_name: loss | PP: 7/8 | Block rank: 3
+NCCL version 2.18.6+cuda12.1
+NCCL version 2.18.6+cuda12.1
+NCCL version 2.18.6+cuda12.1
+NCCL version 2.18.6+cuda12.1
+NCCL version 2.18.6+cuda12.1
+NCCL version 2.18.6+cuda12.1
+NCCL version 2.18.6+cuda12.1
+NCCL version 2.18.6+cuda12.1
+NCCL version 2.18.6+cuda12.1
+NCCL version 2.18.6+cuda12.1
+NCCL version 2.18.6+cuda12.1
+NCCL version 2.18.6+cuda12.1
+NCCL version 2.18.6+cuda12.1
+NCCL version 2.18.6+cuda12.1
+NCCL version 2.18.6+cuda12.1
+NCCL version 2.18.6+cuda12.1
+NCCL version 2.18.6+cuda12.1
+NCCL version 2.18.6+cuda12.1
+NCCL version 2.18.6+cuda12.1
+NCCL version 2.18.6+cuda12.1
+NCCL version 2.18.6+cuda12.1
+NCCL version 2.18.6+cuda12.1
+NCCL version 2.18.6+cuda12.1
+NCCL version 2.18.6+cuda12.1
+NCCL version 2.18.6+cuda12.1
+NCCL version 2.18.6+cuda12.1
+NCCL version 2.18.6+cuda12.1
+NCCL version 2.18.6+cuda12.1
+NCCL version 2.18.6+cuda12.1
+NCCL version 2.18.6+cuda12.1NCCL version 2.18.6+cuda12.1
+
+NCCL version 2.18.6+cuda12.1
+NCCL version 2.18.6+cuda12.1
+NCCL version 2.18.6+cuda12.1
+NCCL version 2.18.6+cuda12.1
+NCCL version 2.18.6+cuda12.1
+NCCL version 2.18.6+cuda12.1
+NCCL version 2.18.6+cuda12.1
+NCCL version 2.18.6+cuda12.1
+NCCL version 2.18.6+cuda12.1
+NCCL version 2.18.6+cuda12.1
+NCCL version 2.18.6+cuda12.1
+ip-26-0-172-73:926595:926986 [1] NCCL INFO Using network Libfabric
+ip-26-0-172-73:926600:926988 [6] NCCL INFO Using network Libfabric
+ip-26-0-172-142:880685:881067 [6] NCCL INFO Using network Libfabric
+ip-26-0-172-57:942617:943002 [6] NCCL INFO Using network Libfabric
+ip-26-0-172-116:1019293:1019681 [5] NCCL INFO Using network Libfabric
+ip-26-0-172-142:880682:881070 [3] NCCL INFO Using network Libfabric
+ip-26-0-172-73:926601:926987 [7] NCCL INFO Using network Libfabric
+ip-26-0-172-116:1019292:1019679 [4] NCCL INFO Using network Libfabric
+ip-26-0-172-57:942612:943006 [1] NCCL INFO Using network Libfabric
+ip-26-0-172-142:880681:881073 [2] NCCL INFO Using network Libfabric
+ip-26-0-172-147:618588:618973 [4] NCCL INFO Using network Libfabric
+ip-26-0-172-73:926597:926992 [3] NCCL INFO Using network Libfabric
+ip-26-0-172-73:926596:926995 [2] NCCL INFO Using network Libfabric
+ip-26-0-172-57:942615:943007 [4] NCCL INFO Using network Libfabric
+ip-26-0-172-142:880683:881075 [4] NCCL INFO Using network Libfabric
+ip-26-0-172-147:618586:618976 [2] NCCL INFO Using network Libfabric
+ip-26-0-172-73:926595:926986 [1] NCCL INFO comm 0xa4702b0 rank 0 nranks 1 cudaDev 1 nvmlDev 1 busId 64000 commId 0xefd6094e8078b1b5 - Init START
+ip-26-0-172-147:618584:618980 [0] NCCL INFO Using network Libfabric
+ip-26-0-172-116:1019291:1019684 [3] NCCL INFO Using network Libfabric
+ip-26-0-172-73:926600:926988 [6] NCCL INFO comm 0xb523290 rank 0 nranks 1 cudaDev 6 nvmlDev 6 busId b9000 commId 0xfd8c01f769e5d9c3 - Init START
+ip-26-0-172-73:926595:926986 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-147:618585:618981 [1] NCCL INFO Using network Libfabric
+ip-26-0-172-116:1019295:1019688 [7] NCCL INFO Using network Libfabric
+ip-26-0-172-116:1019288:1019689 [0] NCCL INFO Using network Libfabric
+ip-26-0-172-147:618589:618982 [5] NCCL INFO Using network Libfabric
+ip-26-0-172-116:1019294:1019686 [6] NCCL INFO Using network Libfabric
+ip-26-0-172-116:1019290:1019691 [2] NCCL INFO Using network Libfabric
+ip-26-0-172-142:880684:881078 [5] NCCL INFO Using network Libfabric
+ip-26-0-172-142:880685:881067 [6] NCCL INFO comm 0x9e7b6c0 rank 0 nranks 1 cudaDev 6 nvmlDev 6 busId b9000 commId 0x228a516b849b7aa0 - Init START
+ip-26-0-172-252:688207:688594 [7] NCCL INFO Using network Libfabric
+ip-26-0-172-73:926600:926988 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-73:926594:926994 [0] NCCL INFO Using network Libfabric
+ip-26-0-172-147:618591:618983 [7] NCCL INFO Using network Libfabric
+ip-26-0-172-57:942616:943005 [5] NCCL INFO Using network Libfabric
+ip-26-0-172-57:942617:943002 [6] NCCL INFO comm 0xacb7bf0 rank 0 nranks 1 cudaDev 6 nvmlDev 6 busId b9000 commId 0xce0197adacec94c - Init START
+ip-26-0-172-142:880679:881080 [0] NCCL INFO Using network Libfabric
+ip-26-0-172-142:880680:881074 [1] NCCL INFO Using network Libfabric
+ip-26-0-172-116:1019289:1019690 [1] NCCL INFO Using network Libfabric
+ip-26-0-172-116:1019293:1019681 [5] NCCL INFO comm 0xc39f4c0 rank 0 nranks 1 cudaDev 5 nvmlDev 5 busId a8000 commId 0x94b8b9b1172ac7f7 - Init START
+ip-26-0-172-142:880682:881070 [3] NCCL INFO comm 0xaa24520 rank 0 nranks 1 cudaDev 3 nvmlDev 3 busId 86000 commId 0x1bb35e59b95df540 - Init START
+ip-26-0-172-252:688206:688597 [6] NCCL INFO Using network Libfabric
+ip-26-0-172-252:688201:688600 [1] NCCL INFO Using network Libfabric
+ip-26-0-172-73:926601:926987 [7] NCCL INFO comm 0xa54a640 rank 0 nranks 1 cudaDev 7 nvmlDev 7 busId ca000 commId 0xf3340c6bc170e5e0 - Init START
+ip-26-0-172-73:926599:926998 [5] NCCL INFO Using network Libfabric
+ip-26-0-172-73:926598:926997 [4] NCCL INFO Using network Libfabric
+ip-26-0-172-147:618588:618973 [4] NCCL INFO comm 0x9d6df70 rank 0 nranks 1 cudaDev 4 nvmlDev 4 busId 97000 commId 0xa14c5a4547fa1c34 - Init START
+ip-26-0-172-57:942614:943012 [3] NCCL INFO Using network Libfabric
+ip-26-0-172-57:942617:943002 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-57:942618:943010 [7] NCCL INFO Using network Libfabric
+ip-26-0-172-142:880685:881067 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-116:1019292:1019679 [4] NCCL INFO comm 0xc2fe140 rank 0 nranks 1 cudaDev 4 nvmlDev 4 busId 97000 commId 0x665eadf333a30be4 - Init START
+ip-26-0-172-142:880686:881081 [7] NCCL INFO Using network Libfabric
+ip-26-0-172-252:688203:688593 [3] NCCL INFO Using network Libfabric
+ip-26-0-172-57:942612:943006 [1] NCCL INFO comm 0xa0aa950 rank 0 nranks 1 cudaDev 1 nvmlDev 1 busId 64000 commId 0x6610956bd2f6c2fb - Init START
+ip-26-0-172-142:880682:881070 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-73:926601:926987 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-73:926597:926992 [3] NCCL INFO comm 0xac4c0d0 rank 0 nranks 1 cudaDev 3 nvmlDev 3 busId 86000 commId 0xa38bdf5b995818a7 - Init START
+ip-26-0-172-147:618584:618980 [0] NCCL INFO comm 0xba75700 rank 0 nranks 1 cudaDev 0 nvmlDev 0 busId 53000 commId 0xa4a1c3e91a5f4ee6 - Init START
+ip-26-0-172-73:926596:926995 [2] NCCL INFO comm 0xa3672b0 rank 0 nranks 1 cudaDev 2 nvmlDev 2 busId 75000 commId 0x608f8efdfec306d7 - Init START
+ip-26-0-172-57:942615:943007 [4] NCCL INFO comm 0xc0a4f70 rank 0 nranks 1 cudaDev 4 nvmlDev 4 busId 97000 commId 0x91c0847960e5f6ba - Init START
+ip-26-0-172-116:1019293:1019681 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-252:688204:688601 [4] NCCL INFO Using network Libfabric
+ip-26-0-172-116:1019292:1019679 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-142:880681:881073 [2] NCCL INFO comm 0xafd8200 rank 0 nranks 1 cudaDev 2 nvmlDev 2 busId 75000 commId 0x64df22d9c7d488ab - Init START
+ip-26-0-172-142:880683:881075 [4] NCCL INFO comm 0xbbf8100 rank 0 nranks 1 cudaDev 4 nvmlDev 4 busId 97000 commId 0xdbf3daa2bdee12e3 - Init START
+ip-26-0-172-147:618588:618973 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-147:618586:618976 [2] NCCL INFO comm 0xa062940 rank 0 nranks 1 cudaDev 2 nvmlDev 2 busId 75000 commId 0xcb2298495ca2a828 - Init START
+ip-26-0-172-57:942613:943014 [2] NCCL INFO Using network Libfabric
+ip-26-0-172-57:942612:943006 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-142:880683:881075 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-142:880681:881073 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-73:926597:926992 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-147:618589:618982 [5] NCCL INFO comm 0xb79cec0 rank 0 nranks 1 cudaDev 5 nvmlDev 5 busId a8000 commId 0xc6c21fed3f236879 - Init START
+ip-26-0-172-57:942615:943007 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-73:926596:926995 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-147:618585:618981 [1] NCCL INFO comm 0xb5314d0 rank 0 nranks 1 cudaDev 1 nvmlDev 1 busId 64000 commId 0xbad86dca646ea35b - Init START
+ip-26-0-172-116:1019288:1019689 [0] NCCL INFO comm 0xa173310 rank 0 nranks 1 cudaDev 0 nvmlDev 0 busId 53000 commId 0x18bc386c861e257d - Init START
+ip-26-0-172-116:1019290:1019691 [2] NCCL INFO comm 0xb532d30 rank 0 nranks 1 cudaDev 2 nvmlDev 2 busId 75000 commId 0x2c933182693940d7 - Init START
+ip-26-0-172-252:688207:688594 [7] NCCL INFO comm 0xa4899d0 rank 0 nranks 1 cudaDev 7 nvmlDev 7 busId ca000 commId 0xca84283b6fe59ad8 - Init START
+ip-26-0-172-252:688202:688605 [2] NCCL INFO Using network Libfabric
+ip-26-0-172-142:880684:881078 [5] NCCL INFO comm 0x9fc7470 rank 0 nranks 1 cudaDev 5 nvmlDev 5 busId a8000 commId 0x2ea4053c729e895e - Init START
+ip-26-0-172-57:942611:943015 [0] NCCL INFO Using network Libfabric
+ip-26-0-172-73:926594:926994 [0] NCCL INFO comm 0xb37c500 rank 0 nranks 1 cudaDev 0 nvmlDev 0 busId 53000 commId 0xf65ea4046937c873 - Init START
+ip-26-0-172-147:618584:618980 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-147:618590:618984 [6] NCCL INFO Using network Libfabric
+ip-26-0-172-147:618591:618983 [7] NCCL INFO comm 0xb33b780 rank 0 nranks 1 cudaDev 7 nvmlDev 7 busId ca000 commId 0x43383332916e059c - Init START
+ip-26-0-172-147:618586:618976 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-147:618589:618982 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-147:618585:618981 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-116:1019295:1019688 [7] NCCL INFO comm 0xb9681c0 rank 0 nranks 1 cudaDev 7 nvmlDev 7 busId ca000 commId 0x1a310b0d9fdd31f4 - Init START
+ip-26-0-172-116:1019294:1019686 [6] NCCL INFO comm 0xa59e630 rank 0 nranks 1 cudaDev 6 nvmlDev 6 busId b9000 commId 0x88675990041435c9 - Init START
+ip-26-0-172-116:1019291:1019684 [3] NCCL INFO comm 0xb8bc290 rank 0 nranks 1 cudaDev 3 nvmlDev 3 busId 86000 commId 0x86ea12526dc12129 - Init START
+ip-26-0-172-252:688201:688600 [1] NCCL INFO comm 0xa8c27e0 rank 0 nranks 1 cudaDev 1 nvmlDev 1 busId 64000 commId 0x589239cd9357e3d2 - Init START
+ip-26-0-172-252:688206:688597 [6] NCCL INFO comm 0xc3c4480 rank 0 nranks 1 cudaDev 6 nvmlDev 6 busId b9000 commId 0x585543c159cd3d5c - Init START
+ip-26-0-172-116:1019288:1019689 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-57:942616:943005 [5] NCCL INFO comm 0xa628600 rank 0 nranks 1 cudaDev 5 nvmlDev 5 busId a8000 commId 0xa1770233b27f3f86 - Init START
+ip-26-0-172-116:1019290:1019691 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-142:880679:881080 [0] NCCL INFO comm 0xc15d520 rank 0 nranks 1 cudaDev 0 nvmlDev 0 busId 53000 commId 0xa7e021c6170379b7 - Init START
+ip-26-0-172-116:1019294:1019686 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-116:1019295:1019688 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-116:1019291:1019684 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-73:926598:926997 [4] NCCL INFO comm 0xae614e0 rank 0 nranks 1 cudaDev 4 nvmlDev 4 busId 97000 commId 0xc353ea69253ec4a4 - Init START
+ip-26-0-172-116:1019289:1019690 [1] NCCL INFO comm 0x9b46270 rank 0 nranks 1 cudaDev 1 nvmlDev 1 busId 64000 commId 0xf7413587374e582c - Init START
+ip-26-0-172-73:926594:926994 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-147:618591:618983 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-57:942614:943012 [3] NCCL INFO comm 0xa931fd0 rank 0 nranks 1 cudaDev 3 nvmlDev 3 busId 86000 commId 0xcd152b1d84ad05fc - Init START
+ip-26-0-172-142:880680:881074 [1] NCCL INFO comm 0xaf10cc0 rank 0 nranks 1 cudaDev 1 nvmlDev 1 busId 64000 commId 0x9b9386d3316183bc - Init START
+ip-26-0-172-252:688205:688604 [5] NCCL INFO Using network Libfabric
+ip-26-0-172-252:688207:688594 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-142:880686:881081 [7] NCCL INFO comm 0xac72420 rank 0 nranks 1 cudaDev 7 nvmlDev 7 busId ca000 commId 0x3380cb88e352cfa - Init START
+ip-26-0-172-142:880684:881078 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-116:1019289:1019690 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-73:926598:926997 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-57:942616:943005 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-73:926599:926998 [5] NCCL INFO comm 0xbf652c0 rank 0 nranks 1 cudaDev 5 nvmlDev 5 busId a8000 commId 0x5ef227f7006300f4 - Init START
+ip-26-0-172-147:618587:618986 [3] NCCL INFO Using network Libfabric
+ip-26-0-172-252:688203:688593 [3] NCCL INFO comm 0xa76a540 rank 0 nranks 1 cudaDev 3 nvmlDev 3 busId 86000 commId 0xf66732327b4ac4f5 - Init START
+ip-26-0-172-252:688201:688600 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-252:688206:688597 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-142:880679:881080 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-252:688204:688601 [4] NCCL INFO comm 0xaa188f0 rank 0 nranks 1 cudaDev 4 nvmlDev 4 busId 97000 commId 0xf08c6d2609d51c06 - Init START
+ip-26-0-172-252:688200:688603 [0] NCCL INFO Using network Libfabric
+ip-26-0-172-142:880680:881074 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-252:688203:688593 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-57:942618:943010 [7] NCCL INFO comm 0xaf9c830 rank 0 nranks 1 cudaDev 7 nvmlDev 7 busId ca000 commId 0x1475a1e50e772b6e - Init START
+ip-26-0-172-142:880686:881081 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-57:942614:943012 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-73:926599:926998 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-57:942613:943014 [2] NCCL INFO comm 0xbac0ef0 rank 0 nranks 1 cudaDev 2 nvmlDev 2 busId 75000 commId 0xf529a97f75bd2e80 - Init START
+ip-26-0-172-252:688204:688601 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-57:942618:943010 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-57:942613:943014 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-252:688202:688605 [2] NCCL INFO comm 0xb1fdf50 rank 0 nranks 1 cudaDev 2 nvmlDev 2 busId 75000 commId 0xde9c4b90351391a0 - Init START
+ip-26-0-172-57:942611:943015 [0] NCCL INFO comm 0xb2ecde0 rank 0 nranks 1 cudaDev 0 nvmlDev 0 busId 53000 commId 0xcdc1cbe6ccd8d7eb - Init START
+ip-26-0-172-252:688202:688605 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-252:688205:688604 [5] NCCL INFO comm 0xadd8480 rank 0 nranks 1 cudaDev 5 nvmlDev 5 busId a8000 commId 0x31c8d99302663941 - Init START
+ip-26-0-172-57:942611:943015 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-147:618590:618984 [6] NCCL INFO comm 0xb6cdef0 rank 0 nranks 1 cudaDev 6 nvmlDev 6 busId b9000 commId 0x4638be3954c599a5 - Init START
+ip-26-0-172-147:618587:618986 [3] NCCL INFO comm 0xab050e0 rank 0 nranks 1 cudaDev 3 nvmlDev 3 busId 86000 commId 0x2aba9a2bb9559f82 - Init START
+ip-26-0-172-147:618590:618984 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-147:618587:618986 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-252:688205:688604 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-252:688200:688603 [0] NCCL INFO comm 0xa91c990 rank 0 nranks 1 cudaDev 0 nvmlDev 0 busId 53000 commId 0x3de5d39e61f58567 - Init START
+ip-26-0-172-252:688200:688603 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+NCCL version 2.18.6+cuda12.1
+NCCL version 2.18.6+cuda12.1
+NCCL version 2.18.6+cuda12.1
+NCCL version 2.18.6+cuda12.1
+NCCL version 2.18.6+cuda12.1
+NCCL version 2.18.6+cuda12.1
+NCCL version 2.18.6+cuda12.1
+NCCL version 2.18.6+cuda12.1
+NCCL version 2.18.6+cuda12.1
+NCCL version 2.18.6+cuda12.1
+NCCL version 2.18.6+cuda12.1
+NCCL version 2.18.6+cuda12.1
+NCCL version 2.18.6+cuda12.1
+NCCL version 2.18.6+cuda12.1
+ip-26-0-173-7:811254:811665 [6] NCCL INFO Using network Libfabric
+ip-26-0-160-225:94343:94771 [1] NCCL INFO Using network Libfabric
+ip-26-0-160-225:94342:94765 [0] NCCL INFO Using network Libfabric
+ip-26-0-173-7:811248:811666 [0] NCCL INFO Using network Libfabric
+ip-26-0-160-225:94346:94773 [4] NCCL INFO Using network Libfabric
+ip-26-0-173-7:811255:811662 [7] NCCL INFO Using network Libfabric
+ip-26-0-160-225:94348:94774 [6] NCCL INFO Using network Libfabric
+ip-26-0-160-225:94347:94772 [5] NCCL INFO Using network Libfabric
+ip-26-0-173-7:811250:811669 [2] NCCL INFO Using network Libfabric
+ip-26-0-173-7:811251:811671 [3] NCCL INFO Using network Libfabric
+ip-26-0-173-7:811253:811670 [5] NCCL INFO Using network Libfabric
+ip-26-0-173-7:811254:811665 [6] NCCL INFO comm 0x9697c70 rank 0 nranks 1 cudaDev 6 nvmlDev 6 busId b9000 commId 0x1ea75edfed6c6dc6 - Init START
+ip-26-0-173-7:811248:811666 [0] NCCL INFO comm 0xb8ca7c0 rank 0 nranks 1 cudaDev 0 nvmlDev 0 busId 53000 commId 0x8987331d9e95edae - Init START
+ip-26-0-160-225:94342:94765 [0] NCCL INFO comm 0xa833970 rank 0 nranks 1 cudaDev 0 nvmlDev 0 busId 53000 commId 0x9318754846b66106 - Init START
+ip-26-0-160-225:94343:94771 [1] NCCL INFO comm 0x9c72b30 rank 0 nranks 1 cudaDev 1 nvmlDev 1 busId 64000 commId 0x646dc9f2cbea7601 - Init START
+ip-26-0-173-7:811249:811675 [1] NCCL INFO Using network Libfabric
+ip-26-0-173-7:811254:811665 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-173-7:811255:811662 [7] NCCL INFO comm 0xa8dc9b0 rank 0 nranks 1 cudaDev 7 nvmlDev 7 busId ca000 commId 0x26d6c3aee09acd13 - Init START
+ip-26-0-173-7:811248:811666 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-160-225:94346:94773 [4] NCCL INFO comm 0xba0b090 rank 0 nranks 1 cudaDev 4 nvmlDev 4 busId 97000 commId 0x39ab8cb2ec22973c - Init START
+ip-26-0-160-225:94342:94765 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/349
+ip-26-0-160-225:94343:94771 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-173-7:811252:811674 [4] NCCL INFO Using network Libfabric
+ip-26-0-160-225:94348:94774 [6] NCCL INFO comm 0xbbc1ca0 rank 0 nranks 1 cudaDev 6 nvmlDev 6 busId b9000 commId 0x85028ee211c854d1 - Init START
+ip-26-0-160-225:94345:94779 [3] NCCL INFO Using network Libfabric
+ip-26-0-160-225:94344:94778 [2] NCCL INFO Using network Libfabric
+ip-26-0-160-225:94346:94773 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-173-7:811255:811662 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-160-225:94347:94772 [5] NCCL INFO comm 0xaaa1f70 rank 0 nranks 1 cudaDev 5 nvmlDev 5 busId a8000 commId 0x45b2c163d1e10c8f - Init START
+ip-26-0-160-225:94348:94774 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-173-7:811251:811671 [3] NCCL INFO comm 0x988f950 rank 0 nranks 1 cudaDev 3 nvmlDev 3 busId 86000 commId 0x14c32b7a3c51460a - Init START
+ip-26-0-173-7:811250:811669 [2] NCCL INFO comm 0xa2918a0 rank 0 nranks 1 cudaDev 2 nvmlDev 2 busId 75000 commId 0x1ccdf40322c55beb - Init START
+ip-26-0-173-7:811250:811669 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-160-225:94347:94772 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-173-7:811251:811671 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-173-7:811249:811675 [1] NCCL INFO comm 0x9492570 rank 0 nranks 1 cudaDev 1 nvmlDev 1 busId 64000 commId 0x1847af43c68ddd96 - Init START
+ip-26-0-173-7:811253:811670 [5] NCCL INFO comm 0xad45780 rank 0 nranks 1 cudaDev 5 nvmlDev 5 busId a8000 commId 0xade52ce07c1afbd3 - Init START
+ip-26-0-160-225:94349:94776 [7] NCCL INFO Using network Libfabric
+ip-26-0-173-7:811252:811674 [4] NCCL INFO comm 0x99ac3a0 rank 0 nranks 1 cudaDev 4 nvmlDev 4 busId 97000 commId 0xaef0ef16b8fac6e4 - Init START
+ip-26-0-173-7:811249:811675 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-173-7:811253:811670 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-160-225:94345:94779 [3] NCCL INFO comm 0xbba6cb0 rank 0 nranks 1 cudaDev 3 nvmlDev 3 busId 86000 commId 0xc622d7481d9e9256 - Init START
+ip-26-0-173-7:811252:811674 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-160-225:94345:94779 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-160-225:94344:94778 [2] NCCL INFO comm 0xa9c7db0 rank 0 nranks 1 cudaDev 2 nvmlDev 2 busId 75000 commId 0x7ff16fe4a011ddc8 - Init START
+ip-26-0-160-225:94344:94778 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-160-225:94349:94776 [7] NCCL INFO comm 0xadc7670 rank 0 nranks 1 cudaDev 7 nvmlDev 7 busId ca000 commId 0x6fc54540cf43cf0e - Init START
+ip-26-0-160-225:94349:94776 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-73:926599:926998 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-172-73:926599:926998 [5] NCCL INFO Channel 00/32 :    0
+ip-26-0-172-73:926599:926998 [5] NCCL INFO Channel 01/32 :    0
+ip-26-0-172-73:926599:926998 [5] NCCL INFO Channel 02/32 :    0
+ip-26-0-172-73:926599:926998 [5] NCCL INFO Channel 03/32 :    0
+ip-26-0-172-73:926599:926998 [5] NCCL INFO Channel 04/32 :    0
+ip-26-0-172-73:926599:926998 [5] NCCL INFO Channel 05/32 :    0
+ip-26-0-172-73:926599:926998 [5] NCCL INFO Channel 06/32 :    0
+ip-26-0-172-73:926599:926998 [5] NCCL INFO Channel 07/32 :    0
+ip-26-0-172-73:926599:926998 [5] NCCL INFO Channel 08/32 :    0
+ip-26-0-172-73:926599:926998 [5] NCCL INFO Channel 09/32 :    0
+ip-26-0-172-73:926599:926998 [5] NCCL INFO Channel 10/32 :    0
+ip-26-0-172-73:926599:926998 [5] NCCL INFO Channel 11/32 :    0
+ip-26-0-172-73:926599:926998 [5] NCCL INFO Channel 12/32 :    0
+ip-26-0-172-73:926599:926998 [5] NCCL INFO Channel 13/32 :    0
+ip-26-0-172-73:926599:926998 [5] NCCL INFO Channel 14/32 :    0
+ip-26-0-172-73:926599:926998 [5] NCCL INFO Channel 15/32 :    0
+ip-26-0-172-73:926599:926998 [5] NCCL INFO Channel 16/32 :    0
+ip-26-0-172-73:926599:926998 [5] NCCL INFO Channel 17/32 :    0
+ip-26-0-172-73:926599:926998 [5] NCCL INFO Channel 18/32 :    0
+ip-26-0-172-73:926599:926998 [5] NCCL INFO Channel 19/32 :    0
+ip-26-0-172-73:926599:926998 [5] NCCL INFO Channel 20/32 :    0
+ip-26-0-172-73:926599:926998 [5] NCCL INFO Channel 21/32 :    0
+ip-26-0-172-73:926599:926998 [5] NCCL INFO Channel 22/32 :    0
+ip-26-0-172-73:926599:926998 [5] NCCL INFO Channel 23/32 :    0
+ip-26-0-172-73:926599:926998 [5] NCCL INFO Channel 24/32 :    0
+ip-26-0-172-73:926599:926998 [5] NCCL INFO Channel 25/32 :    0
+ip-26-0-172-73:926599:926998 [5] NCCL INFO Channel 26/32 :    0
+ip-26-0-172-73:926599:926998 [5] NCCL INFO Channel 27/32 :    0
+ip-26-0-172-73:926599:926998 [5] NCCL INFO Channel 28/32 :    0
+ip-26-0-172-73:926599:926998 [5] NCCL INFO Channel 29/32 :    0
+ip-26-0-172-73:926599:926998 [5] NCCL INFO Channel 30/32 :    0
+ip-26-0-172-73:926599:926998 [5] NCCL INFO Channel 31/32 :    0
+ip-26-0-172-73:926599:926998 [5] NCCL INFO Trees [0] -1/-1/-1->0->-1 [1] -1/-1/-1->0->-1 [2] -1/-1/-1->0->-1 [3] -1/-1/-1->0->-1 [4] -1/-1/-1->0->-1 [5] -1/-1/-1->0->-1 [6] -1/-1/-1->0->-1 [7] -1/-1/-1->0->-1 [8] -1/-1/-1->0->-1 [9] -1/-1/-1->0->-1 [10] -1/-1/-1->0->-1 [11] -1/-1/-1->0->-1 [12] -1/-1/-1->0->-1 [13] -1/-1/-1->0->-1 [14] -1/-1/-1->0->-1 [15] -1/-1/-1->0->-1 [16] -1/-1/-1->0->-1 [17] -1/-1/-1->0->-1 [18] -1/-1/-1->0->-1 [19] -1/-1/-1->0->-1 [20] -1/-1/-1->0->-1 [21] -1/-1/-1->0->-1 [22] -1/-1/-1->0->-1 [23] -1/-1/-1->0->-1 [24] -1/-1/-1->0->-1 [25] -1/-1/-1->0->-1 [26] -1/-1/-1->0->-1 [27] -1/-1/-1->0->-1 [28] -1/-1/-1->0->-1 [29] -1/-1/-1->0->-1 [30] -1/-1/-1->0->-1 [31] -1/-1/-1->0->-1
+ip-26-0-172-73:926599:926998 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-73:926599:926998 [5] NCCL INFO Connected all rings
+ip-26-0-172-73:926599:926998 [5] NCCL INFO Connected all trees
+ip-26-0-172-73:926599:926998 [5] NCCL INFO 32 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-252:688205:688604 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-172-252:688205:688604 [5] NCCL INFO Channel 00/32 :    0
+ip-26-0-172-252:688205:688604 [5] NCCL INFO Channel 01/32 :    0
+ip-26-0-172-252:688205:688604 [5] NCCL INFO Channel 02/32 :    0
+ip-26-0-172-252:688205:688604 [5] NCCL INFO Channel 03/32 :    0
+ip-26-0-172-252:688205:688604 [5] NCCL INFO Channel 04/32 :    0
+ip-26-0-172-252:688205:688604 [5] NCCL INFO Channel 05/32 :    0
+ip-26-0-172-252:688205:688604 [5] NCCL INFO Channel 06/32 :    0
+ip-26-0-172-252:688205:688604 [5] NCCL INFO Channel 07/32 :    0
+ip-26-0-172-252:688205:688604 [5] NCCL INFO Channel 08/32 :    0
+ip-26-0-172-252:688205:688604 [5] NCCL INFO Channel 09/32 :    0
+ip-26-0-172-252:688205:688604 [5] NCCL INFO Channel 10/32 :    0
+ip-26-0-172-252:688205:688604 [5] NCCL INFO Channel 11/32 :    0
+ip-26-0-172-252:688205:688604 [5] NCCL INFO Channel 12/32 :    0
+ip-26-0-172-252:688205:688604 [5] NCCL INFO Channel 13/32 :    0
+ip-26-0-172-252:688205:688604 [5] NCCL INFO Channel 14/32 :    0
+ip-26-0-172-252:688205:688604 [5] NCCL INFO Channel 15/32 :    0
+ip-26-0-172-252:688205:688604 [5] NCCL INFO Channel 16/32 :    0
+ip-26-0-172-252:688205:688604 [5] NCCL INFO Channel 17/32 :    0
+ip-26-0-172-252:688205:688604 [5] NCCL INFO Channel 18/32 :    0
+ip-26-0-172-252:688205:688604 [5] NCCL INFO Channel 19/32 :    0
+ip-26-0-172-252:688205:688604 [5] NCCL INFO Channel 20/32 :    0
+ip-26-0-172-252:688205:688604 [5] NCCL INFO Channel 21/32 :    0
+ip-26-0-172-252:688205:688604 [5] NCCL INFO Channel 22/32 :    0
+ip-26-0-172-252:688205:688604 [5] NCCL INFO Channel 23/32 :    0
+ip-26-0-172-252:688205:688604 [5] NCCL INFO Channel 24/32 :    0
+ip-26-0-172-252:688205:688604 [5] NCCL INFO Channel 25/32 :    0
+ip-26-0-172-252:688205:688604 [5] NCCL INFO Channel 26/32 :    0
+ip-26-0-172-252:688205:688604 [5] NCCL INFO Channel 27/32 :    0
+ip-26-0-172-252:688205:688604 [5] NCCL INFO Channel 28/32 :    0
+ip-26-0-172-252:688205:688604 [5] NCCL INFO Channel 29/32 :    0
+ip-26-0-172-252:688205:688604 [5] NCCL INFO Channel 30/32 :    0
+ip-26-0-172-252:688205:688604 [5] NCCL INFO Channel 31/32 :    0
+ip-26-0-172-252:688205:688604 [5] NCCL INFO Trees [0] -1/-1/-1->0->-1 [1] -1/-1/-1->0->-1 [2] -1/-1/-1->0->-1 [3] -1/-1/-1->0->-1 [4] -1/-1/-1->0->-1 [5] -1/-1/-1->0->-1 [6] -1/-1/-1->0->-1 [7] -1/-1/-1->0->-1 [8] -1/-1/-1->0->-1 [9] -1/-1/-1->0->-1 [10] -1/-1/-1->0->-1 [11] -1/-1/-1->0->-1 [12] -1/-1/-1->0->-1 [13] -1/-1/-1->0->-1 [14] -1/-1/-1->0->-1 [15] -1/-1/-1->0->-1 [16] -1/-1/-1->0->-1 [17] -1/-1/-1->0->-1 [18] -1/-1/-1->0->-1 [19] -1/-1/-1->0->-1 [20] -1/-1/-1->0->-1 [21] -1/-1/-1->0->-1 [22] -1/-1/-1->0->-1 [23] -1/-1/-1->0->-1 [24] -1/-1/-1->0->-1 [25] -1/-1/-1->0->-1 [26] -1/-1/-1->0->-1 [27] -1/-1/-1->0->-1 [28] -1/-1/-1->0->-1 [29] -1/-1/-1->0->-1 [30] -1/-1/-1->0->-1 [31] -1/-1/-1->0->-1
+ip-26-0-172-252:688205:688604 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-252:688205:688604 [5] NCCL INFO Connected all rings
+ip-26-0-172-252:688205:688604 [5] NCCL INFO Connected all trees
+ip-26-0-172-252:688205:688604 [5] NCCL INFO 32 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-73:926599:926998 [5] NCCL INFO comm 0xbf652c0 rank 0 nranks 1 cudaDev 5 nvmlDev 5 busId a8000 commId 0x5ef227f7006300f4 - Init COMPLETE
+ip-26-0-172-116:1019288:1019689 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-172-116:1019288:1019689 [0] NCCL INFO Channel 00/32 :    0
+ip-26-0-172-116:1019288:1019689 [0] NCCL INFO Channel 01/32 :    0
+ip-26-0-172-116:1019288:1019689 [0] NCCL INFO Channel 02/32 :    0
+ip-26-0-172-116:1019288:1019689 [0] NCCL INFO Channel 03/32 :    0
+ip-26-0-172-116:1019288:1019689 [0] NCCL INFO Channel 04/32 :    0
+ip-26-0-172-116:1019288:1019689 [0] NCCL INFO Channel 05/32 :    0
+ip-26-0-172-116:1019288:1019689 [0] NCCL INFO Channel 06/32 :    0
+ip-26-0-172-116:1019288:1019689 [0] NCCL INFO Channel 07/32 :    0
+ip-26-0-172-116:1019288:1019689 [0] NCCL INFO Channel 08/32 :    0
+ip-26-0-172-116:1019288:1019689 [0] NCCL INFO Channel 09/32 :    0
+ip-26-0-172-116:1019288:1019689 [0] NCCL INFO Channel 10/32 :    0
+ip-26-0-172-116:1019288:1019689 [0] NCCL INFO Channel 11/32 :    0
+ip-26-0-172-116:1019288:1019689 [0] NCCL INFO Channel 12/32 :    0
+ip-26-0-172-116:1019288:1019689 [0] NCCL INFO Channel 13/32 :    0
+ip-26-0-172-116:1019288:1019689 [0] NCCL INFO Channel 14/32 :    0
+ip-26-0-172-116:1019288:1019689 [0] NCCL INFO Channel 15/32 :    0
+ip-26-0-172-116:1019288:1019689 [0] NCCL INFO Channel 16/32 :    0
+ip-26-0-172-116:1019288:1019689 [0] NCCL INFO Channel 17/32 :    0
+ip-26-0-172-116:1019288:1019689 [0] NCCL INFO Channel 18/32 :    0
+ip-26-0-172-116:1019288:1019689 [0] NCCL INFO Channel 19/32 :    0
+ip-26-0-172-116:1019288:1019689 [0] NCCL INFO Channel 20/32 :    0
+ip-26-0-172-116:1019288:1019689 [0] NCCL INFO Channel 21/32 :    0
+ip-26-0-172-116:1019288:1019689 [0] NCCL INFO Channel 22/32 :    0
+ip-26-0-172-116:1019288:1019689 [0] NCCL INFO Channel 23/32 :    0
+ip-26-0-172-116:1019288:1019689 [0] NCCL INFO Channel 24/32 :    0
+ip-26-0-172-116:1019288:1019689 [0] NCCL INFO Channel 25/32 :    0
+ip-26-0-172-116:1019288:1019689 [0] NCCL INFO Channel 26/32 :    0
+ip-26-0-172-116:1019288:1019689 [0] NCCL INFO Channel 27/32 :    0
+ip-26-0-172-116:1019288:1019689 [0] NCCL INFO Channel 28/32 :    0
+ip-26-0-172-116:1019288:1019689 [0] NCCL INFO Channel 29/32 :    0
+ip-26-0-172-116:1019288:1019689 [0] NCCL INFO Channel 30/32 :    0
+ip-26-0-172-116:1019288:1019689 [0] NCCL INFO Channel 31/32 :    0
+ip-26-0-172-116:1019288:1019689 [0] NCCL INFO Trees [0] -1/-1/-1->0->-1 [1] -1/-1/-1->0->-1 [2] -1/-1/-1->0->-1 [3] -1/-1/-1->0->-1 [4] -1/-1/-1->0->-1 [5] -1/-1/-1->0->-1 [6] -1/-1/-1->0->-1 [7] -1/-1/-1->0->-1 [8] -1/-1/-1->0->-1 [9] -1/-1/-1->0->-1 [10] -1/-1/-1->0->-1 [11] -1/-1/-1->0->-1 [12] -1/-1/-1->0->-1 [13] -1/-1/-1->0->-1 [14] -1/-1/-1->0->-1 [15] -1/-1/-1->0->-1 [16] -1/-1/-1->0->-1 [17] -1/-1/-1->0->-1 [18] -1/-1/-1->0->-1 [19] -1/-1/-1->0->-1 [20] -1/-1/-1->0->-1 [21] -1/-1/-1->0->-1 [22] -1/-1/-1->0->-1 [23] -1/-1/-1->0->-1 [24] -1/-1/-1->0->-1 [25] -1/-1/-1->0->-1 [26] -1/-1/-1->0->-1 [27] -1/-1/-1->0->-1 [28] -1/-1/-1->0->-1 [29] -1/-1/-1->0->-1 [30] -1/-1/-1->0->-1 [31] -1/-1/-1->0->-1
+ip-26-0-172-116:1019288:1019689 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-116:1019288:1019689 [0] NCCL INFO Connected all rings
+ip-26-0-172-116:1019288:1019689 [0] NCCL INFO Connected all trees
+ip-26-0-172-116:1019288:1019689 [0] NCCL INFO 32 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-173-7:811255:811662 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-173-7:811255:811662 [7] NCCL INFO Channel 00/32 :    0
+ip-26-0-173-7:811255:811662 [7] NCCL INFO Channel 01/32 :    0
+ip-26-0-173-7:811255:811662 [7] NCCL INFO Channel 02/32 :    0
+ip-26-0-173-7:811255:811662 [7] NCCL INFO Channel 03/32 :    0
+ip-26-0-173-7:811255:811662 [7] NCCL INFO Channel 04/32 :    0
+ip-26-0-173-7:811255:811662 [7] NCCL INFO Channel 05/32 :    0
+ip-26-0-173-7:811255:811662 [7] NCCL INFO Channel 06/32 :    0
+ip-26-0-173-7:811255:811662 [7] NCCL INFO Channel 07/32 :    0
+ip-26-0-173-7:811255:811662 [7] NCCL INFO Channel 08/32 :    0
+ip-26-0-173-7:811255:811662 [7] NCCL INFO Channel 09/32 :    0
+ip-26-0-173-7:811255:811662 [7] NCCL INFO Channel 10/32 :    0
+ip-26-0-173-7:811255:811662 [7] NCCL INFO Channel 11/32 :    0
+ip-26-0-173-7:811255:811662 [7] NCCL INFO Channel 12/32 :    0
+ip-26-0-173-7:811255:811662 [7] NCCL INFO Channel 13/32 :    0
+ip-26-0-173-7:811255:811662 [7] NCCL INFO Channel 14/32 :    0
+ip-26-0-173-7:811255:811662 [7] NCCL INFO Channel 15/32 :    0
+ip-26-0-173-7:811255:811662 [7] NCCL INFO Channel 16/32 :    0
+ip-26-0-173-7:811255:811662 [7] NCCL INFO Channel 17/32 :    0
+ip-26-0-173-7:811255:811662 [7] NCCL INFO Channel 18/32 :    0
+ip-26-0-173-7:811255:811662 [7] NCCL INFO Channel 19/32 :    0
+ip-26-0-173-7:811255:811662 [7] NCCL INFO Channel 20/32 :    0
+ip-26-0-173-7:811255:811662 [7] NCCL INFO Channel 21/32 :    0
+ip-26-0-173-7:811255:811662 [7] NCCL INFO Channel 22/32 :    0
+ip-26-0-173-7:811255:811662 [7] NCCL INFO Channel 23/32 :    0
+ip-26-0-173-7:811255:811662 [7] NCCL INFO Channel 24/32 :    0
+ip-26-0-173-7:811255:811662 [7] NCCL INFO Channel 25/32 :    0
+ip-26-0-173-7:811255:811662 [7] NCCL INFO Channel 26/32 :    0
+ip-26-0-173-7:811255:811662 [7] NCCL INFO Channel 27/32 :    0
+ip-26-0-173-7:811255:811662 [7] NCCL INFO Channel 28/32 :    0
+ip-26-0-173-7:811255:811662 [7] NCCL INFO Channel 29/32 :    0
+ip-26-0-173-7:811255:811662 [7] NCCL INFO Channel 30/32 :    0
+ip-26-0-173-7:811255:811662 [7] NCCL INFO Channel 31/32 :    0
+ip-26-0-173-7:811255:811662 [7] NCCL INFO Trees [0] -1/-1/-1->0->-1 [1] -1/-1/-1->0->-1 [2] -1/-1/-1->0->-1 [3] -1/-1/-1->0->-1 [4] -1/-1/-1->0->-1 [5] -1/-1/-1->0->-1 [6] -1/-1/-1->0->-1 [7] -1/-1/-1->0->-1 [8] -1/-1/-1->0->-1 [9] -1/-1/-1->0->-1 [10] -1/-1/-1->0->-1 [11] -1/-1/-1->0->-1 [12] -1/-1/-1->0->-1 [13] -1/-1/-1->0->-1 [14] -1/-1/-1->0->-1 [15] -1/-1/-1->0->-1 [16] -1/-1/-1->0->-1 [17] -1/-1/-1->0->-1 [18] -1/-1/-1->0->-1 [19] -1/-1/-1->0->-1 [20] -1/-1/-1->0->-1 [21] -1/-1/-1->0->-1 [22] -1/-1/-1->0->-1 [23] -1/-1/-1->0->-1 [24] -1/-1/-1->0->-1 [25] -1/-1/-1->0->-1 [26] -1/-1/-1->0->-1 [27] -1/-1/-1->0->-1 [28] -1/-1/-1->0->-1 [29] -1/-1/-1->0->-1 [30] -1/-1/-1->0->-1 [31] -1/-1/-1->0->-1
+ip-26-0-173-7:811255:811662 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-173-7:811255:811662 [7] NCCL INFO Connected all rings
+ip-26-0-173-7:811255:811662 [7] NCCL INFO Connected all trees
+ip-26-0-173-7:811255:811662 [7] NCCL INFO 32 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-252:688205:688604 [5] NCCL INFO comm 0xadd8480 rank 0 nranks 1 cudaDev 5 nvmlDev 5 busId a8000 commId 0x31c8d99302663941 - Init COMPLETE
+ip-26-0-172-73:926596:926995 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-172-73:926596:926995 [2] NCCL INFO Channel 00/32 :    0
+ip-26-0-172-73:926596:926995 [2] NCCL INFO Channel 01/32 :    0
+ip-26-0-172-73:926596:926995 [2] NCCL INFO Channel 02/32 :    0
+ip-26-0-172-73:926596:926995 [2] NCCL INFO Channel 03/32 :    0
+ip-26-0-172-73:926596:926995 [2] NCCL INFO Channel 04/32 :    0
+ip-26-0-172-73:926596:926995 [2] NCCL INFO Channel 05/32 :    0
+ip-26-0-172-73:926596:926995 [2] NCCL INFO Channel 06/32 :    0
+ip-26-0-172-73:926596:926995 [2] NCCL INFO Channel 07/32 :    0
+ip-26-0-172-73:926596:926995 [2] NCCL INFO Channel 08/32 :    0
+ip-26-0-172-73:926596:926995 [2] NCCL INFO Channel 09/32 :    0
+ip-26-0-172-73:926596:926995 [2] NCCL INFO Channel 10/32 :    0
+ip-26-0-172-73:926596:926995 [2] NCCL INFO Channel 11/32 :    0
+ip-26-0-172-73:926596:926995 [2] NCCL INFO Channel 12/32 :    0
+ip-26-0-172-73:926596:926995 [2] NCCL INFO Channel 13/32 :    0
+ip-26-0-172-73:926596:926995 [2] NCCL INFO Channel 14/32 :    0
+ip-26-0-172-73:926596:926995 [2] NCCL INFO Channel 15/32 :    0
+ip-26-0-172-73:926596:926995 [2] NCCL INFO Channel 16/32 :    0
+ip-26-0-172-73:926596:926995 [2] NCCL INFO Channel 17/32 :    0
+ip-26-0-172-73:926596:926995 [2] NCCL INFO Channel 18/32 :    0
+ip-26-0-172-73:926596:926995 [2] NCCL INFO Channel 19/32 :    0
+ip-26-0-172-73:926596:926995 [2] NCCL INFO Channel 20/32 :    0
+ip-26-0-172-73:926596:926995 [2] NCCL INFO Channel 21/32 :    0
+ip-26-0-172-73:926596:926995 [2] NCCL INFO Channel 22/32 :    0
+ip-26-0-172-73:926596:926995 [2] NCCL INFO Channel 23/32 :    0
+ip-26-0-172-73:926596:926995 [2] NCCL INFO Channel 24/32 :    0
+ip-26-0-172-73:926596:926995 [2] NCCL INFO Channel 25/32 :    0
+ip-26-0-172-73:926596:926995 [2] NCCL INFO Channel 26/32 :    0
+ip-26-0-172-73:926596:926995 [2] NCCL INFO Channel 27/32 :    0
+ip-26-0-172-73:926596:926995 [2] NCCL INFO Channel 28/32 :    0
+ip-26-0-172-73:926596:926995 [2] NCCL INFO Channel 29/32 :    0
+ip-26-0-172-73:926596:926995 [2] NCCL INFO Channel 30/32 :    0
+ip-26-0-172-73:926596:926995 [2] NCCL INFO Channel 31/32 :    0
+ip-26-0-172-73:926596:926995 [2] NCCL INFO Trees [0] -1/-1/-1->0->-1 [1] -1/-1/-1->0->-1 [2] -1/-1/-1->0->-1 [3] -1/-1/-1->0->-1 [4] -1/-1/-1->0->-1 [5] -1/-1/-1->0->-1 [6] -1/-1/-1->0->-1 [7] -1/-1/-1->0->-1 [8] -1/-1/-1->0->-1 [9] -1/-1/-1->0->-1 [10] -1/-1/-1->0->-1 [11] -1/-1/-1->0->-1 [12] -1/-1/-1->0->-1 [13] -1/-1/-1->0->-1 [14] -1/-1/-1->0->-1 [15] -1/-1/-1->0->-1 [16] -1/-1/-1->0->-1 [17] -1/-1/-1->0->-1 [18] -1/-1/-1->0->-1 [19] -1/-1/-1->0->-1 [20] -1/-1/-1->0->-1 [21] -1/-1/-1->0->-1 [22] -1/-1/-1->0->-1 [23] -1/-1/-1->0->-1 [24] -1/-1/-1->0->-1 [25] -1/-1/-1->0->-1 [26] -1/-1/-1->0->-1 [27] -1/-1/-1->0->-1 [28] -1/-1/-1->0->-1 [29] -1/-1/-1->0->-1 [30] -1/-1/-1->0->-1 [31] -1/-1/-1->0->-1
+ip-26-0-172-73:926596:926995 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-73:926596:926995 [2] NCCL INFO Connected all rings
+ip-26-0-172-73:926596:926995 [2] NCCL INFO Connected all trees
+ip-26-0-172-73:926596:926995 [2] NCCL INFO 32 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-116:1019288:1019689 [0] NCCL INFO comm 0xa173310 rank 0 nranks 1 cudaDev 0 nvmlDev 0 busId 53000 commId 0x18bc386c861e257d - Init COMPLETE
+ip-26-0-172-73:926600:926988 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-172-73:926600:926988 [6] NCCL INFO Channel 00/32 :    0
+ip-26-0-172-73:926600:926988 [6] NCCL INFO Channel 01/32 :    0
+ip-26-0-172-73:926600:926988 [6] NCCL INFO Channel 02/32 :    0
+ip-26-0-172-73:926600:926988 [6] NCCL INFO Channel 03/32 :    0
+ip-26-0-172-73:926600:926988 [6] NCCL INFO Channel 04/32 :    0
+ip-26-0-172-73:926600:926988 [6] NCCL INFO Channel 05/32 :    0
+ip-26-0-172-73:926600:926988 [6] NCCL INFO Channel 06/32 :    0
+ip-26-0-172-73:926600:926988 [6] NCCL INFO Channel 07/32 :    0
+ip-26-0-172-73:926600:926988 [6] NCCL INFO Channel 08/32 :    0
+ip-26-0-172-73:926600:926988 [6] NCCL INFO Channel 09/32 :    0
+ip-26-0-172-73:926600:926988 [6] NCCL INFO Channel 10/32 :    0
+ip-26-0-172-73:926600:926988 [6] NCCL INFO Channel 11/32 :    0
+ip-26-0-172-73:926600:926988 [6] NCCL INFO Channel 12/32 :    0
+ip-26-0-172-73:926600:926988 [6] NCCL INFO Channel 13/32 :    0
+ip-26-0-172-73:926600:926988 [6] NCCL INFO Channel 14/32 :    0
+ip-26-0-172-73:926600:926988 [6] NCCL INFO Channel 15/32 :    0
+ip-26-0-172-73:926600:926988 [6] NCCL INFO Channel 16/32 :    0
+ip-26-0-172-73:926600:926988 [6] NCCL INFO Channel 17/32 :    0
+ip-26-0-172-73:926600:926988 [6] NCCL INFO Channel 18/32 :    0
+ip-26-0-172-73:926600:926988 [6] NCCL INFO Channel 19/32 :    0
+ip-26-0-172-73:926600:926988 [6] NCCL INFO Channel 20/32 :    0
+ip-26-0-172-73:926600:926988 [6] NCCL INFO Channel 21/32 :    0
+ip-26-0-172-73:926600:926988 [6] NCCL INFO Channel 22/32 :    0
+ip-26-0-172-73:926600:926988 [6] NCCL INFO Channel 23/32 :    0
+ip-26-0-172-73:926600:926988 [6] NCCL INFO Channel 24/32 :    0
+ip-26-0-172-73:926600:926988 [6] NCCL INFO Channel 25/32 :    0
+ip-26-0-172-73:926600:926988 [6] NCCL INFO Channel 26/32 :    0
+ip-26-0-172-73:926600:926988 [6] NCCL INFO Channel 27/32 :    0
+ip-26-0-172-73:926600:926988 [6] NCCL INFO Channel 28/32 :    0
+ip-26-0-172-73:926600:926988 [6] NCCL INFO Channel 29/32 :    0
+ip-26-0-172-73:926600:926988 [6] NCCL INFO Channel 30/32 :    0
+ip-26-0-172-73:926600:926988 [6] NCCL INFO Channel 31/32 :    0
+ip-26-0-172-73:926600:926988 [6] NCCL INFO Trees [0] -1/-1/-1->0->-1 [1] -1/-1/-1->0->-1 [2] -1/-1/-1->0->-1 [3] -1/-1/-1->0->-1 [4] -1/-1/-1->0->-1 [5] -1/-1/-1->0->-1 [6] -1/-1/-1->0->-1 [7] -1/-1/-1->0->-1 [8] -1/-1/-1->0->-1 [9] -1/-1/-1->0->-1 [10] -1/-1/-1->0->-1 [11] -1/-1/-1->0->-1 [12] -1/-1/-1->0->-1 [13] -1/-1/-1->0->-1 [14] -1/-1/-1->0->-1 [15] -1/-1/-1->0->-1 [16] -1/-1/-1->0->-1 [17] -1/-1/-1->0->-1 [18] -1/-1/-1->0->-1 [19] -1/-1/-1->0->-1 [20] -1/-1/-1->0->-1 [21] -1/-1/-1->0->-1 [22] -1/-1/-1->0->-1 [23] -1/-1/-1->0->-1 [24] -1/-1/-1->0->-1 [25] -1/-1/-1->0->-1 [26] -1/-1/-1->0->-1 [27] -1/-1/-1->0->-1 [28] -1/-1/-1->0->-1 [29] -1/-1/-1->0->-1 [30] -1/-1/-1->0->-1 [31] -1/-1/-1->0->-1
+ip-26-0-172-73:926600:926988 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-142:880679:881080 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-172-142:880679:881080 [0] NCCL INFO Channel 00/32 :    0
+ip-26-0-172-142:880679:881080 [0] NCCL INFO Channel 01/32 :    0
+ip-26-0-172-142:880679:881080 [0] NCCL INFO Channel 02/32 :    0
+ip-26-0-172-142:880679:881080 [0] NCCL INFO Channel 03/32 :    0
+ip-26-0-172-142:880679:881080 [0] NCCL INFO Channel 04/32 :    0
+ip-26-0-172-142:880679:881080 [0] NCCL INFO Channel 05/32 :    0
+ip-26-0-172-142:880679:881080 [0] NCCL INFO Channel 06/32 :    0
+ip-26-0-172-142:880679:881080 [0] NCCL INFO Channel 07/32 :    0
+ip-26-0-172-142:880679:881080 [0] NCCL INFO Channel 08/32 :    0
+ip-26-0-172-142:880679:881080 [0] NCCL INFO Channel 09/32 :    0
+ip-26-0-172-142:880679:881080 [0] NCCL INFO Channel 10/32 :    0
+ip-26-0-172-142:880679:881080 [0] NCCL INFO Channel 11/32 :    0
+ip-26-0-172-142:880679:881080 [0] NCCL INFO Channel 12/32 :    0
+ip-26-0-172-142:880679:881080 [0] NCCL INFO Channel 13/32 :    0
+ip-26-0-172-142:880679:881080 [0] NCCL INFO Channel 14/32 :    0
+ip-26-0-172-142:880679:881080 [0] NCCL INFO Channel 15/32 :    0
+ip-26-0-172-142:880679:881080 [0] NCCL INFO Channel 16/32 :    0
+ip-26-0-172-142:880679:881080 [0] NCCL INFO Channel 17/32 :    0
+ip-26-0-172-142:880679:881080 [0] NCCL INFO Channel 18/32 :    0
+ip-26-0-172-142:880679:881080 [0] NCCL INFO Channel 19/32 :    0
+ip-26-0-172-142:880679:881080 [0] NCCL INFO Channel 20/32 :    0
+ip-26-0-172-142:880679:881080 [0] NCCL INFO Channel 21/32 :    0
+ip-26-0-172-142:880679:881080 [0] NCCL INFO Channel 22/32 :    0
+ip-26-0-172-142:880679:881080 [0] NCCL INFO Channel 23/32 :    0
+ip-26-0-172-142:880679:881080 [0] NCCL INFO Channel 24/32 :    0
+ip-26-0-172-142:880679:881080 [0] NCCL INFO Channel 25/32 :    0
+ip-26-0-172-142:880679:881080 [0] NCCL INFO Channel 26/32 :    0
+ip-26-0-172-142:880679:881080 [0] NCCL INFO Channel 27/32 :    0
+ip-26-0-172-142:880679:881080 [0] NCCL INFO Channel 28/32 :    0
+ip-26-0-172-142:880679:881080 [0] NCCL INFO Channel 29/32 :    0
+ip-26-0-172-142:880679:881080 [0] NCCL INFO Channel 30/32 :    0
+ip-26-0-172-142:880679:881080 [0] NCCL INFO Channel 31/32 :    0
+ip-26-0-172-142:880679:881080 [0] NCCL INFO Trees [0] -1/-1/-1->0->-1 [1] -1/-1/-1->0->-1 [2] -1/-1/-1->0->-1 [3] -1/-1/-1->0->-1 [4] -1/-1/-1->0->-1 [5] -1/-1/-1->0->-1 [6] -1/-1/-1->0->-1 [7] -1/-1/-1->0->-1 [8] -1/-1/-1->0->-1 [9] -1/-1/-1->0->-1 [10] -1/-1/-1->0->-1 [11] -1/-1/-1->0->-1 [12] -1/-1/-1->0->-1 [13] -1/-1/-1->0->-1 [14] -1/-1/-1->0->-1 [15] -1/-1/-1->0->-1 [16] -1/-1/-1->0->-1 [17] -1/-1/-1->0->-1 [18] -1/-1/-1->0->-1 [19] -1/-1/-1->0->-1 [20] -1/-1/-1->0->-1 [21] -1/-1/-1->0->-1 [22] -1/-1/-1->0->-1 [23] -1/-1/-1->0->-1 [24] -1/-1/-1->0->-1 [25] -1/-1/-1->0->-1 [26] -1/-1/-1->0->-1 [27] -1/-1/-1->0->-1 [28] -1/-1/-1->0->-1 [29] -1/-1/-1->0->-1 [30] -1/-1/-1->0->-1 [31] -1/-1/-1->0->-1
+ip-26-0-172-142:880679:881080 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-73:926600:926988 [6] NCCL INFO Connected all rings
+ip-26-0-172-73:926600:926988 [6] NCCL INFO Connected all trees
+ip-26-0-172-73:926600:926988 [6] NCCL INFO 32 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-142:880679:881080 [0] NCCL INFO Connected all rings
+ip-26-0-172-142:880679:881080 [0] NCCL INFO Connected all trees
+ip-26-0-172-142:880679:881080 [0] NCCL INFO 32 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-160-225:94343:94771 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-160-225:94343:94771 [1] NCCL INFO Channel 00/32 :    0
+ip-26-0-160-225:94343:94771 [1] NCCL INFO Channel 01/32 :    0
+ip-26-0-160-225:94343:94771 [1] NCCL INFO Channel 02/32 :    0
+ip-26-0-160-225:94343:94771 [1] NCCL INFO Channel 03/32 :    0
+ip-26-0-160-225:94343:94771 [1] NCCL INFO Channel 04/32 :    0
+ip-26-0-160-225:94343:94771 [1] NCCL INFO Channel 05/32 :    0
+ip-26-0-160-225:94343:94771 [1] NCCL INFO Channel 06/32 :    0
+ip-26-0-160-225:94343:94771 [1] NCCL INFO Channel 07/32 :    0
+ip-26-0-160-225:94343:94771 [1] NCCL INFO Channel 08/32 :    0
+ip-26-0-160-225:94343:94771 [1] NCCL INFO Channel 09/32 :    0
+ip-26-0-160-225:94343:94771 [1] NCCL INFO Channel 10/32 :    0
+ip-26-0-160-225:94343:94771 [1] NCCL INFO Channel 11/32 :    0
+ip-26-0-160-225:94343:94771 [1] NCCL INFO Channel 12/32 :    0
+ip-26-0-160-225:94343:94771 [1] NCCL INFO Channel 13/32 :    0
+ip-26-0-160-225:94343:94771 [1] NCCL INFO Channel 14/32 :    0
+ip-26-0-160-225:94343:94771 [1] NCCL INFO Channel 15/32 :    0
+ip-26-0-160-225:94343:94771 [1] NCCL INFO Channel 16/32 :    0
+ip-26-0-160-225:94343:94771 [1] NCCL INFO Channel 17/32 :    0
+ip-26-0-160-225:94343:94771 [1] NCCL INFO Channel 18/32 :    0
+ip-26-0-160-225:94343:94771 [1] NCCL INFO Channel 19/32 :    0
+ip-26-0-160-225:94343:94771 [1] NCCL INFO Channel 20/32 :    0
+ip-26-0-160-225:94343:94771 [1] NCCL INFO Channel 21/32 :    0
+ip-26-0-160-225:94343:94771 [1] NCCL INFO Channel 22/32 :    0
+ip-26-0-160-225:94343:94771 [1] NCCL INFO Channel 23/32 :    0
+ip-26-0-160-225:94343:94771 [1] NCCL INFO Channel 24/32 :    0
+ip-26-0-160-225:94343:94771 [1] NCCL INFO Channel 25/32 :    0
+ip-26-0-160-225:94343:94771 [1] NCCL INFO Channel 26/32 :    0
+ip-26-0-160-225:94343:94771 [1] NCCL INFO Channel 27/32 :    0
+ip-26-0-160-225:94343:94771 [1] NCCL INFO Channel 28/32 :    0
+ip-26-0-160-225:94343:94771 [1] NCCL INFO Channel 29/32 :    0
+ip-26-0-160-225:94343:94771 [1] NCCL INFO Channel 30/32 :    0
+ip-26-0-160-225:94343:94771 [1] NCCL INFO Channel 31/32 :    0
+ip-26-0-160-225:94343:94771 [1] NCCL INFO Trees [0] -1/-1/-1->0->-1 [1] -1/-1/-1->0->-1 [2] -1/-1/-1->0->-1 [3] -1/-1/-1->0->-1 [4] -1/-1/-1->0->-1 [5] -1/-1/-1->0->-1 [6] -1/-1/-1->0->-1 [7] -1/-1/-1->0->-1 [8] -1/-1/-1->0->-1 [9] -1/-1/-1->0->-1 [10] -1/-1/-1->0->-1 [11] -1/-1/-1->0->-1 [12] -1/-1/-1->0->-1 [13] -1/-1/-1->0->-1 [14] -1/-1/-1->0->-1 [15] -1/-1/-1->0->-1 [16] -1/-1/-1->0->-1 [17] -1/-1/-1->0->-1 [18] -1/-1/-1->0->-1 [19] -1/-1/-1->0->-1 [20] -1/-1/-1->0->-1 [21] -1/-1/-1->0->-1 [22] -1/-1/-1->0->-1 [23] -1/-1/-1->0->-1 [24] -1/-1/-1->0->-1 [25] -1/-1/-1->0->-1 [26] -1/-1/-1->0->-1 [27] -1/-1/-1->0->-1 [28] -1/-1/-1->0->-1 [29] -1/-1/-1->0->-1 [30] -1/-1/-1->0->-1 [31] -1/-1/-1->0->-1
+ip-26-0-160-225:94343:94771 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-225:94343:94771 [1] NCCL INFO Connected all rings
+ip-26-0-160-225:94343:94771 [1] NCCL INFO Connected all trees
+ip-26-0-160-225:94343:94771 [1] NCCL INFO 32 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-173-7:811254:811665 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-173-7:811254:811665 [6] NCCL INFO Channel 00/32 :    0
+ip-26-0-173-7:811254:811665 [6] NCCL INFO Channel 01/32 :    0
+ip-26-0-173-7:811254:811665 [6] NCCL INFO Channel 02/32 :    0
+ip-26-0-173-7:811254:811665 [6] NCCL INFO Channel 03/32 :    0
+ip-26-0-173-7:811254:811665 [6] NCCL INFO Channel 04/32 :    0
+ip-26-0-173-7:811254:811665 [6] NCCL INFO Channel 05/32 :    0
+ip-26-0-173-7:811254:811665 [6] NCCL INFO Channel 06/32 :    0
+ip-26-0-173-7:811254:811665 [6] NCCL INFO Channel 07/32 :    0
+ip-26-0-173-7:811254:811665 [6] NCCL INFO Channel 08/32 :    0
+ip-26-0-173-7:811254:811665 [6] NCCL INFO Channel 09/32 :    0
+ip-26-0-173-7:811254:811665 [6] NCCL INFO Channel 10/32 :    0
+ip-26-0-173-7:811254:811665 [6] NCCL INFO Channel 11/32 :    0
+ip-26-0-173-7:811254:811665 [6] NCCL INFO Channel 12/32 :    0
+ip-26-0-173-7:811254:811665 [6] NCCL INFO Channel 13/32 :    0
+ip-26-0-173-7:811254:811665 [6] NCCL INFO Channel 14/32 :    0
+ip-26-0-173-7:811254:811665 [6] NCCL INFO Channel 15/32 :    0
+ip-26-0-173-7:811254:811665 [6] NCCL INFO Channel 16/32 :    0
+ip-26-0-173-7:811254:811665 [6] NCCL INFO Channel 17/32 :    0
+ip-26-0-173-7:811254:811665 [6] NCCL INFO Channel 18/32 :    0
+ip-26-0-173-7:811254:811665 [6] NCCL INFO Channel 19/32 :    0
+ip-26-0-173-7:811254:811665 [6] NCCL INFO Channel 20/32 :    0
+ip-26-0-173-7:811254:811665 [6] NCCL INFO Channel 21/32 :    0
+ip-26-0-173-7:811254:811665 [6] NCCL INFO Channel 22/32 :    0
+ip-26-0-173-7:811254:811665 [6] NCCL INFO Channel 23/32 :    0
+ip-26-0-173-7:811254:811665 [6] NCCL INFO Channel 24/32 :    0
+ip-26-0-173-7:811254:811665 [6] NCCL INFO Channel 25/32 :    0
+ip-26-0-173-7:811254:811665 [6] NCCL INFO Channel 26/32 :    0
+ip-26-0-173-7:811254:811665 [6] NCCL INFO Channel 27/32 :    0
+ip-26-0-173-7:811254:811665 [6] NCCL INFO Channel 28/32 :    0
+ip-26-0-173-7:811254:811665 [6] NCCL INFO Channel 29/32 :    0
+ip-26-0-173-7:811254:811665 [6] NCCL INFO Channel 30/32 :    0
+ip-26-0-173-7:811254:811665 [6] NCCL INFO Channel 31/32 :    0
+ip-26-0-173-7:811254:811665 [6] NCCL INFO Trees [0] -1/-1/-1->0->-1 [1] -1/-1/-1->0->-1 [2] -1/-1/-1->0->-1 [3] -1/-1/-1->0->-1 [4] -1/-1/-1->0->-1 [5] -1/-1/-1->0->-1 [6] -1/-1/-1->0->-1 [7] -1/-1/-1->0->-1 [8] -1/-1/-1->0->-1 [9] -1/-1/-1->0->-1 [10] -1/-1/-1->0->-1 [11] -1/-1/-1->0->-1 [12] -1/-1/-1->0->-1 [13] -1/-1/-1->0->-1 [14] -1/-1/-1->0->-1 [15] -1/-1/-1->0->-1 [16] -1/-1/-1->0->-1 [17] -1/-1/-1->0->-1 [18] -1/-1/-1->0->-1 [19] -1/-1/-1->0->-1 [20] -1/-1/-1->0->-1 [21] -1/-1/-1->0->-1 [22] -1/-1/-1->0->-1 [23] -1/-1/-1->0->-1 [24] -1/-1/-1->0->-1 [25] -1/-1/-1->0->-1 [26] -1/-1/-1->0->-1 [27] -1/-1/-1->0->-1 [28] -1/-1/-1->0->-1 [29] -1/-1/-1->0->-1 [30] -1/-1/-1->0->-1 [31] -1/-1/-1->0->-1
+ip-26-0-173-7:811254:811665 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-173-7:811254:811665 [6] NCCL INFO Connected all rings
+ip-26-0-173-7:811254:811665 [6] NCCL INFO Connected all trees
+ip-26-0-173-7:811254:811665 [6] NCCL INFO 32 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-116:1019295:1019688 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-172-116:1019295:1019688 [7] NCCL INFO Channel 00/32 :    0
+ip-26-0-172-116:1019295:1019688 [7] NCCL INFO Channel 01/32 :    0
+ip-26-0-172-116:1019295:1019688 [7] NCCL INFO Channel 02/32 :    0
+ip-26-0-172-116:1019295:1019688 [7] NCCL INFO Channel 03/32 :    0
+ip-26-0-172-116:1019295:1019688 [7] NCCL INFO Channel 04/32 :    0
+ip-26-0-172-116:1019295:1019688 [7] NCCL INFO Channel 05/32 :    0
+ip-26-0-172-116:1019295:1019688 [7] NCCL INFO Channel 06/32 :    0
+ip-26-0-172-116:1019295:1019688 [7] NCCL INFO Channel 07/32 :    0
+ip-26-0-172-116:1019295:1019688 [7] NCCL INFO Channel 08/32 :    0
+ip-26-0-172-116:1019295:1019688 [7] NCCL INFO Channel 09/32 :    0
+ip-26-0-172-116:1019295:1019688 [7] NCCL INFO Channel 10/32 :    0
+ip-26-0-172-116:1019295:1019688 [7] NCCL INFO Channel 11/32 :    0
+ip-26-0-172-116:1019295:1019688 [7] NCCL INFO Channel 12/32 :    0
+ip-26-0-172-116:1019295:1019688 [7] NCCL INFO Channel 13/32 :    0
+ip-26-0-173-7:811255:811662 [7] NCCL INFO comm 0xa8dc9b0 rank 0 nranks 1 cudaDev 7 nvmlDev 7 busId ca000 commId 0x26d6c3aee09acd13 - Init COMPLETE
+ip-26-0-172-116:1019295:1019688 [7] NCCL INFO Channel 14/32 :    0
+ip-26-0-172-116:1019295:1019688 [7] NCCL INFO Channel 15/32 :    0
+ip-26-0-172-116:1019295:1019688 [7] NCCL INFO Channel 16/32 :    0
+ip-26-0-172-116:1019295:1019688 [7] NCCL INFO Channel 17/32 :    0
+ip-26-0-172-116:1019295:1019688 [7] NCCL INFO Channel 18/32 :    0
+ip-26-0-172-116:1019295:1019688 [7] NCCL INFO Channel 19/32 :    0
+ip-26-0-172-116:1019295:1019688 [7] NCCL INFO Channel 20/32 :    0
+ip-26-0-172-116:1019295:1019688 [7] NCCL INFO Channel 21/32 :    0
+ip-26-0-172-116:1019295:1019688 [7] NCCL INFO Channel 22/32 :    0
+ip-26-0-172-116:1019295:1019688 [7] NCCL INFO Channel 23/32 :    0
+ip-26-0-172-116:1019295:1019688 [7] NCCL INFO Channel 24/32 :    0
+ip-26-0-172-116:1019295:1019688 [7] NCCL INFO Channel 25/32 :    0
+ip-26-0-172-116:1019295:1019688 [7] NCCL INFO Channel 26/32 :    0
+ip-26-0-172-116:1019295:1019688 [7] NCCL INFO Channel 27/32 :    0
+ip-26-0-172-116:1019295:1019688 [7] NCCL INFO Channel 28/32 :    0
+ip-26-0-172-116:1019295:1019688 [7] NCCL INFO Channel 29/32 :    0
+ip-26-0-172-116:1019295:1019688 [7] NCCL INFO Channel 30/32 :    0
+ip-26-0-172-116:1019295:1019688 [7] NCCL INFO Channel 31/32 :    0
+ip-26-0-172-116:1019295:1019688 [7] NCCL INFO Trees [0] -1/-1/-1->0->-1 [1] -1/-1/-1->0->-1 [2] -1/-1/-1->0->-1 [3] -1/-1/-1->0->-1 [4] -1/-1/-1->0->-1 [5] -1/-1/-1->0->-1 [6] -1/-1/-1->0->-1 [7] -1/-1/-1->0->-1 [8] -1/-1/-1->0->-1 [9] -1/-1/-1->0->-1 [10] -1/-1/-1->0->-1 [11] -1/-1/-1->0->-1 [12] -1/-1/-1->0->-1 [13] -1/-1/-1->0->-1 [14] -1/-1/-1->0->-1 [15] -1/-1/-1->0->-1 [16] -1/-1/-1->0->-1 [17] -1/-1/-1->0->-1 [18] -1/-1/-1->0->-1 [19] -1/-1/-1->0->-1 [20] -1/-1/-1->0->-1 [21] -1/-1/-1->0->-1 [22] -1/-1/-1->0->-1 [23] -1/-1/-1->0->-1 [24] -1/-1/-1->0->-1 [25] -1/-1/-1->0->-1 [26] -1/-1/-1->0->-1 [27] -1/-1/-1->0->-1 [28] -1/-1/-1->0->-1 [29] -1/-1/-1->0->-1 [30] -1/-1/-1->0->-1 [31] -1/-1/-1->0->-1
+ip-26-0-172-116:1019295:1019688 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-116:1019295:1019688 [7] NCCL INFO Connected all rings
+ip-26-0-172-116:1019295:1019688 [7] NCCL INFO Connected all trees
+ip-26-0-172-116:1019295:1019688 [7] NCCL INFO 32 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-173-7:811251:811671 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-173-7:811251:811671 [3] NCCL INFO Channel 00/32 :    0
+ip-26-0-173-7:811251:811671 [3] NCCL INFO Channel 01/32 :    0
+ip-26-0-173-7:811251:811671 [3] NCCL INFO Channel 02/32 :    0
+ip-26-0-173-7:811251:811671 [3] NCCL INFO Channel 03/32 :    0
+ip-26-0-173-7:811251:811671 [3] NCCL INFO Channel 04/32 :    0
+ip-26-0-173-7:811251:811671 [3] NCCL INFO Channel 05/32 :    0
+ip-26-0-173-7:811251:811671 [3] NCCL INFO Channel 06/32 :    0
+ip-26-0-173-7:811251:811671 [3] NCCL INFO Channel 07/32 :    0
+ip-26-0-173-7:811251:811671 [3] NCCL INFO Channel 08/32 :    0
+ip-26-0-173-7:811251:811671 [3] NCCL INFO Channel 09/32 :    0
+ip-26-0-173-7:811251:811671 [3] NCCL INFO Channel 10/32 :    0
+ip-26-0-173-7:811251:811671 [3] NCCL INFO Channel 11/32 :    0
+ip-26-0-173-7:811251:811671 [3] NCCL INFO Channel 12/32 :    0
+ip-26-0-173-7:811251:811671 [3] NCCL INFO Channel 13/32 :    0
+ip-26-0-173-7:811251:811671 [3] NCCL INFO Channel 14/32 :    0
+ip-26-0-173-7:811251:811671 [3] NCCL INFO Channel 15/32 :    0
+ip-26-0-173-7:811251:811671 [3] NCCL INFO Channel 16/32 :    0
+ip-26-0-173-7:811251:811671 [3] NCCL INFO Channel 17/32 :    0
+ip-26-0-173-7:811251:811671 [3] NCCL INFO Channel 18/32 :    0
+ip-26-0-173-7:811251:811671 [3] NCCL INFO Channel 19/32 :    0
+ip-26-0-173-7:811251:811671 [3] NCCL INFO Channel 20/32 :    0
+ip-26-0-173-7:811251:811671 [3] NCCL INFO Channel 21/32 :    0
+ip-26-0-173-7:811251:811671 [3] NCCL INFO Channel 22/32 :    0
+ip-26-0-173-7:811251:811671 [3] NCCL INFO Channel 23/32 :    0
+ip-26-0-173-7:811251:811671 [3] NCCL INFO Channel 24/32 :    0
+ip-26-0-173-7:811251:811671 [3] NCCL INFO Channel 25/32 :    0
+ip-26-0-173-7:811251:811671 [3] NCCL INFO Channel 26/32 :    0
+ip-26-0-173-7:811251:811671 [3] NCCL INFO Channel 27/32 :    0
+ip-26-0-173-7:811251:811671 [3] NCCL INFO Channel 28/32 :    0
+ip-26-0-172-147:618584:618980 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-173-7:811251:811671 [3] NCCL INFO Channel 29/32 :    0
+ip-26-0-173-7:811251:811671 [3] NCCL INFO Channel 30/32 :    0
+ip-26-0-173-7:811251:811671 [3] NCCL INFO Channel 31/32 :    0
+ip-26-0-173-7:811251:811671 [3] NCCL INFO Trees [0] -1/-1/-1->0->-1 [1] -1/-1/-1->0->-1 [2] -1/-1/-1->0->-1 [3] -1/-1/-1->0->-1 [4] -1/-1/-1->0->-1 [5] -1/-1/-1->0->-1 [6] -1/-1/-1->0->-1 [7] -1/-1/-1->0->-1 [8] -1/-1/-1->0->-1 [9] -1/-1/-1->0->-1 [10] -1/-1/-1->0->-1 [11] -1/-1/-1->0->-1 [12] -1/-1/-1->0->-1 [13] -1/-1/-1->0->-1 [14] -1/-1/-1->0->-1 [15] -1/-1/-1->0->-1 [16] -1/-1/-1->0->-1 [17] -1/-1/-1->0->-1 [18] -1/-1/-1->0->-1 [19] -1/-1/-1->0->-1 [20] -1/-1/-1->0->-1 [21] -1/-1/-1->0->-1 [22] -1/-1/-1->0->-1 [23] -1/-1/-1->0->-1 [24] -1/-1/-1->0->-1 [25] -1/-1/-1->0->-1 [26] -1/-1/-1->0->-1 [27] -1/-1/-1->0->-1 [28] -1/-1/-1->0->-1 [29] -1/-1/-1->0->-1 [30] -1/-1/-1->0->-1 [31] -1/-1/-1->0->-1
+ip-26-0-173-7:811251:811671 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-147:618584:618980 [0] NCCL INFO Channel 00/32 :    0
+ip-26-0-172-147:618584:618980 [0] NCCL INFO Channel 01/32 :    0
+ip-26-0-172-147:618584:618980 [0] NCCL INFO Channel 02/32 :    0
+ip-26-0-172-147:618584:618980 [0] NCCL INFO Channel 03/32 :    0
+ip-26-0-172-147:618584:618980 [0] NCCL INFO Channel 04/32 :    0
+ip-26-0-172-147:618584:618980 [0] NCCL INFO Channel 05/32 :    0
+ip-26-0-172-147:618584:618980 [0] NCCL INFO Channel 06/32 :    0
+ip-26-0-172-147:618584:618980 [0] NCCL INFO Channel 07/32 :    0
+ip-26-0-172-147:618584:618980 [0] NCCL INFO Channel 08/32 :    0
+ip-26-0-172-147:618584:618980 [0] NCCL INFO Channel 09/32 :    0
+ip-26-0-172-147:618584:618980 [0] NCCL INFO Channel 10/32 :    0
+ip-26-0-172-147:618584:618980 [0] NCCL INFO Channel 11/32 :    0
+ip-26-0-172-147:618584:618980 [0] NCCL INFO Channel 12/32 :    0
+ip-26-0-172-147:618584:618980 [0] NCCL INFO Channel 13/32 :    0
+ip-26-0-172-147:618584:618980 [0] NCCL INFO Channel 14/32 :    0
+ip-26-0-172-147:618584:618980 [0] NCCL INFO Channel 15/32 :    0
+ip-26-0-172-147:618584:618980 [0] NCCL INFO Channel 16/32 :    0
+ip-26-0-172-147:618584:618980 [0] NCCL INFO Channel 17/32 :    0
+ip-26-0-172-147:618584:618980 [0] NCCL INFO Channel 18/32 :    0
+ip-26-0-172-147:618584:618980 [0] NCCL INFO Channel 19/32 :    0
+ip-26-0-172-147:618584:618980 [0] NCCL INFO Channel 20/32 :    0
+ip-26-0-172-147:618584:618980 [0] NCCL INFO Channel 21/32 :    0
+ip-26-0-172-147:618584:618980 [0] NCCL INFO Channel 22/32 :    0
+ip-26-0-172-147:618584:618980 [0] NCCL INFO Channel 23/32 :    0
+ip-26-0-172-147:618584:618980 [0] NCCL INFO Channel 24/32 :    0
+ip-26-0-172-147:618584:618980 [0] NCCL INFO Channel 25/32 :    0
+ip-26-0-172-147:618584:618980 [0] NCCL INFO Channel 26/32 :    0
+ip-26-0-172-147:618584:618980 [0] NCCL INFO Channel 27/32 :    0
+ip-26-0-172-147:618584:618980 [0] NCCL INFO Channel 28/32 :    0
+ip-26-0-172-147:618584:618980 [0] NCCL INFO Channel 29/32 :    0
+ip-26-0-172-147:618584:618980 [0] NCCL INFO Channel 30/32 :    0
+ip-26-0-172-147:618584:618980 [0] NCCL INFO Channel 31/32 :    0
+ip-26-0-172-147:618584:618980 [0] NCCL INFO Trees [0] -1/-1/-1->0->-1 [1] -1/-1/-1->0->-1 [2] -1/-1/-1->0->-1 [3] -1/-1/-1->0->-1 [4] -1/-1/-1->0->-1 [5] -1/-1/-1->0->-1 [6] -1/-1/-1->0->-1 [7] -1/-1/-1->0->-1 [8] -1/-1/-1->0->-1 [9] -1/-1/-1->0->-1 [10] -1/-1/-1->0->-1 [11] -1/-1/-1->0->-1 [12] -1/-1/-1->0->-1 [13] -1/-1/-1->0->-1 [14] -1/-1/-1->0->-1 [15] -1/-1/-1->0->-1 [16] -1/-1/-1->0->-1 [17] -1/-1/-1->0->-1 [18] -1/-1/-1->0->-1 [19] -1/-1/-1->0->-1 [20] -1/-1/-1->0->-1 [21] -1/-1/-1->0->-1 [22] -1/-1/-1->0->-1 [23] -1/-1/-1->0->-1 [24] -1/-1/-1->0->-1 [25] -1/-1/-1->0->-1 [26] -1/-1/-1->0->-1 [27] -1/-1/-1->0->-1 [28] -1/-1/-1->0->-1 [29] -1/-1/-1->0->-1 [30] -1/-1/-1->0->-1 [31] -1/-1/-1->0->-1
+ip-26-0-172-147:618584:618980 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-173-7:811251:811671 [3] NCCL INFO Connected all rings
+ip-26-0-173-7:811251:811671 [3] NCCL INFO Connected all trees
+ip-26-0-173-7:811251:811671 [3] NCCL INFO 32 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-147:618584:618980 [0] NCCL INFO Connected all rings
+ip-26-0-172-147:618584:618980 [0] NCCL INFO Connected all trees
+ip-26-0-172-147:618584:618980 [0] NCCL INFO 32 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-73:926594:926994 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-172-116:1019289:1019690 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-172-116:1019289:1019690 [1] NCCL INFO Channel 00/32 :    0
+ip-26-0-172-116:1019289:1019690 [1] NCCL INFO Channel 01/32 :    0
+ip-26-0-172-116:1019289:1019690 [1] NCCL INFO Channel 02/32 :    0
+ip-26-0-172-116:1019289:1019690 [1] NCCL INFO Channel 03/32 :    0
+ip-26-0-172-116:1019289:1019690 [1] NCCL INFO Channel 04/32 :    0
+ip-26-0-172-116:1019289:1019690 [1] NCCL INFO Channel 05/32 :    0
+ip-26-0-172-116:1019289:1019690 [1] NCCL INFO Channel 06/32 :    0
+ip-26-0-172-116:1019289:1019690 [1] NCCL INFO Channel 07/32 :    0
+ip-26-0-172-116:1019289:1019690 [1] NCCL INFO Channel 08/32 :    0
+ip-26-0-172-116:1019289:1019690 [1] NCCL INFO Channel 09/32 :    0
+ip-26-0-172-116:1019289:1019690 [1] NCCL INFO Channel 10/32 :    0
+ip-26-0-172-116:1019289:1019690 [1] NCCL INFO Channel 11/32 :    0
+ip-26-0-172-116:1019289:1019690 [1] NCCL INFO Channel 12/32 :    0
+ip-26-0-172-116:1019289:1019690 [1] NCCL INFO Channel 13/32 :    0
+ip-26-0-172-116:1019289:1019690 [1] NCCL INFO Channel 14/32 :    0
+ip-26-0-172-116:1019289:1019690 [1] NCCL INFO Channel 15/32 :    0
+ip-26-0-172-116:1019289:1019690 [1] NCCL INFO Channel 16/32 :    0
+ip-26-0-172-116:1019289:1019690 [1] NCCL INFO Channel 17/32 :    0
+ip-26-0-172-116:1019289:1019690 [1] NCCL INFO Channel 18/32 :    0
+ip-26-0-172-116:1019289:1019690 [1] NCCL INFO Channel 19/32 :    0
+ip-26-0-172-116:1019289:1019690 [1] NCCL INFO Channel 20/32 :    0
+ip-26-0-172-116:1019289:1019690 [1] NCCL INFO Channel 21/32 :    0
+ip-26-0-172-116:1019289:1019690 [1] NCCL INFO Channel 22/32 :    0
+ip-26-0-172-116:1019289:1019690 [1] NCCL INFO Channel 23/32 :    0
+ip-26-0-172-116:1019289:1019690 [1] NCCL INFO Channel 24/32 :    0
+ip-26-0-172-116:1019289:1019690 [1] NCCL INFO Channel 25/32 :    0
+ip-26-0-172-116:1019289:1019690 [1] NCCL INFO Channel 26/32 :    0
+ip-26-0-172-116:1019289:1019690 [1] NCCL INFO Channel 27/32 :    0
+ip-26-0-172-116:1019289:1019690 [1] NCCL INFO Channel 28/32 :    0
+ip-26-0-172-116:1019289:1019690 [1] NCCL INFO Channel 29/32 :    0
+ip-26-0-172-116:1019289:1019690 [1] NCCL INFO Channel 30/32 :    0
+ip-26-0-172-116:1019289:1019690 [1] NCCL INFO Channel 31/32 :    0
+ip-26-0-172-116:1019289:1019690 [1] NCCL INFO Trees [0] -1/-1/-1->0->-1 [1] -1/-1/-1->0->-1 [2] -1/-1/-1->0->-1 [3] -1/-1/-1->0->-1 [4] -1/-1/-1->0->-1 [5] -1/-1/-1->0->-1 [6] -1/-1/-1->0->-1 [7] -1/-1/-1->0->-1 [8] -1/-1/-1->0->-1 [9] -1/-1/-1->0->-1 [10] -1/-1/-1->0->-1 [11] -1/-1/-1->0->-1 [12] -1/-1/-1->0->-1 [13] -1/-1/-1->0->-1 [14] -1/-1/-1->0->-1 [15] -1/-1/-1->0->-1 [16] -1/-1/-1->0->-1 [17] -1/-1/-1->0->-1 [18] -1/-1/-1->0->-1 [19] -1/-1/-1->0->-1 [20] -1/-1/-1->0->-1 [21] -1/-1/-1->0->-1 [22] -1/-1/-1->0->-1 [23] -1/-1/-1->0->-1 [24] -1/-1/-1->0->-1 [25] -1/-1/-1->0->-1 [26] -1/-1/-1->0->-1 [27] -1/-1/-1->0->-1 [28] -1/-1/-1->0->-1 [29] -1/-1/-1->0->-1 [30] -1/-1/-1->0->-1 [31] -1/-1/-1->0->-1
+ip-26-0-172-116:1019289:1019690 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-73:926596:926995 [2] NCCL INFO comm 0xa3672b0 rank 0 nranks 1 cudaDev 2 nvmlDev 2 busId 75000 commId 0x608f8efdfec306d7 - Init COMPLETE
+ip-26-0-172-116:1019289:1019690 [1] NCCL INFO Connected all rings
+ip-26-0-172-116:1019289:1019690 [1] NCCL INFO Connected all trees
+ip-26-0-172-116:1019289:1019690 [1] NCCL INFO 32 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-73:926594:926994 [0] NCCL INFO Channel 00/32 :    0
+ip-26-0-172-73:926594:926994 [0] NCCL INFO Channel 01/32 :    0
+ip-26-0-172-73:926594:926994 [0] NCCL INFO Channel 02/32 :    0
+ip-26-0-172-73:926594:926994 [0] NCCL INFO Channel 03/32 :    0
+ip-26-0-172-73:926594:926994 [0] NCCL INFO Channel 04/32 :    0
+ip-26-0-172-73:926594:926994 [0] NCCL INFO Channel 05/32 :    0
+ip-26-0-172-73:926594:926994 [0] NCCL INFO Channel 06/32 :    0
+ip-26-0-172-73:926594:926994 [0] NCCL INFO Channel 07/32 :    0
+ip-26-0-172-73:926594:926994 [0] NCCL INFO Channel 08/32 :    0
+ip-26-0-172-73:926594:926994 [0] NCCL INFO Channel 09/32 :    0
+ip-26-0-172-73:926594:926994 [0] NCCL INFO Channel 10/32 :    0
+ip-26-0-172-73:926594:926994 [0] NCCL INFO Channel 11/32 :    0
+ip-26-0-172-73:926594:926994 [0] NCCL INFO Channel 12/32 :    0
+ip-26-0-172-73:926594:926994 [0] NCCL INFO Channel 13/32 :    0
+ip-26-0-172-73:926594:926994 [0] NCCL INFO Channel 14/32 :    0
+ip-26-0-172-73:926594:926994 [0] NCCL INFO Channel 15/32 :    0
+ip-26-0-172-73:926594:926994 [0] NCCL INFO Channel 16/32 :    0
+ip-26-0-172-73:926594:926994 [0] NCCL INFO Channel 17/32 :    0
+ip-26-0-172-73:926594:926994 [0] NCCL INFO Channel 18/32 :    0
+ip-26-0-172-73:926594:926994 [0] NCCL INFO Channel 19/32 :    0
+ip-26-0-172-73:926594:926994 [0] NCCL INFO Channel 20/32 :    0
+ip-26-0-172-73:926594:926994 [0] NCCL INFO Channel 21/32 :    0
+ip-26-0-172-73:926594:926994 [0] NCCL INFO Channel 22/32 :    0
+ip-26-0-172-73:926594:926994 [0] NCCL INFO Channel 23/32 :    0
+ip-26-0-172-73:926594:926994 [0] NCCL INFO Channel 24/32 :    0
+ip-26-0-172-73:926594:926994 [0] NCCL INFO Channel 25/32 :    0
+ip-26-0-172-73:926594:926994 [0] NCCL INFO Channel 26/32 :    0
+ip-26-0-172-73:926594:926994 [0] NCCL INFO Channel 27/32 :    0
+ip-26-0-172-73:926594:926994 [0] NCCL INFO Channel 28/32 :    0
+ip-26-0-172-73:926594:926994 [0] NCCL INFO Channel 29/32 :    0
+ip-26-0-172-73:926594:926994 [0] NCCL INFO Channel 30/32 :    0
+ip-26-0-172-73:926594:926994 [0] NCCL INFO Channel 31/32 :    0
+ip-26-0-172-73:926594:926994 [0] NCCL INFO Trees [0] -1/-1/-1->0->-1 [1] -1/-1/-1->0->-1 [2] -1/-1/-1->0->-1 [3] -1/-1/-1->0->-1 [4] -1/-1/-1->0->-1 [5] -1/-1/-1->0->-1 [6] -1/-1/-1->0->-1 [7] -1/-1/-1->0->-1 [8] -1/-1/-1->0->-1 [9] -1/-1/-1->0->-1 [10] -1/-1/-1->0->-1 [11] -1/-1/-1->0->-1 [12] -1/-1/-1->0->-1 [13] -1/-1/-1->0->-1 [14] -1/-1/-1->0->-1 [15] -1/-1/-1->0->-1 [16] -1/-1/-1->0->-1 [17] -1/-1/-1->0->-1 [18] -1/-1/-1->0->-1 [19] -1/-1/-1->0->-1 [20] -1/-1/-1->0->-1 [21] -1/-1/-1->0->-1 [22] -1/-1/-1->0->-1 [23] -1/-1/-1->0->-1 [24] -1/-1/-1->0->-1 [25] -1/-1/-1->0->-1 [26] -1/-1/-1->0->-1 [27] -1/-1/-1->0->-1 [28] -1/-1/-1->0->-1 [29] -1/-1/-1->0->-1 [30] -1/-1/-1->0->-1 [31] -1/-1/-1->0->-1
+ip-26-0-172-73:926594:926994 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-173-7:811248:811666 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-173-7:811248:811666 [0] NCCL INFO Channel 00/32 :    0
+ip-26-0-173-7:811248:811666 [0] NCCL INFO Channel 01/32 :    0
+ip-26-0-173-7:811248:811666 [0] NCCL INFO Channel 02/32 :    0
+ip-26-0-173-7:811248:811666 [0] NCCL INFO Channel 03/32 :    0
+ip-26-0-173-7:811248:811666 [0] NCCL INFO Channel 04/32 :    0
+ip-26-0-173-7:811248:811666 [0] NCCL INFO Channel 05/32 :    0
+ip-26-0-173-7:811248:811666 [0] NCCL INFO Channel 06/32 :    0
+ip-26-0-173-7:811248:811666 [0] NCCL INFO Channel 07/32 :    0
+ip-26-0-173-7:811248:811666 [0] NCCL INFO Channel 08/32 :    0
+ip-26-0-173-7:811248:811666 [0] NCCL INFO Channel 09/32 :    0
+ip-26-0-173-7:811248:811666 [0] NCCL INFO Channel 10/32 :    0
+ip-26-0-173-7:811248:811666 [0] NCCL INFO Channel 11/32 :    0
+ip-26-0-173-7:811248:811666 [0] NCCL INFO Channel 12/32 :    0
+ip-26-0-173-7:811248:811666 [0] NCCL INFO Channel 13/32 :    0
+ip-26-0-173-7:811248:811666 [0] NCCL INFO Channel 14/32 :    0
+ip-26-0-173-7:811248:811666 [0] NCCL INFO Channel 15/32 :    0
+ip-26-0-173-7:811248:811666 [0] NCCL INFO Channel 16/32 :    0
+ip-26-0-173-7:811248:811666 [0] NCCL INFO Channel 17/32 :    0
+ip-26-0-173-7:811248:811666 [0] NCCL INFO Channel 18/32 :    0
+ip-26-0-173-7:811248:811666 [0] NCCL INFO Channel 19/32 :    0
+ip-26-0-173-7:811248:811666 [0] NCCL INFO Channel 20/32 :    0
+ip-26-0-173-7:811248:811666 [0] NCCL INFO Channel 21/32 :    0
+ip-26-0-173-7:811248:811666 [0] NCCL INFO Channel 22/32 :    0
+ip-26-0-173-7:811248:811666 [0] NCCL INFO Channel 23/32 :    0
+ip-26-0-173-7:811248:811666 [0] NCCL INFO Channel 24/32 :    0
+ip-26-0-173-7:811248:811666 [0] NCCL INFO Channel 25/32 :    0
+ip-26-0-173-7:811248:811666 [0] NCCL INFO Channel 26/32 :    0
+ip-26-0-173-7:811248:811666 [0] NCCL INFO Channel 27/32 :    0
+ip-26-0-173-7:811248:811666 [0] NCCL INFO Channel 28/32 :    0
+ip-26-0-173-7:811248:811666 [0] NCCL INFO Channel 29/32 :    0
+ip-26-0-173-7:811248:811666 [0] NCCL INFO Channel 30/32 :    0
+ip-26-0-173-7:811248:811666 [0] NCCL INFO Channel 31/32 :    0
+ip-26-0-172-73:926594:926994 [0] NCCL INFO Connected all rings
+ip-26-0-172-73:926594:926994 [0] NCCL INFO Connected all trees
+ip-26-0-172-73:926594:926994 [0] NCCL INFO 32 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-173-7:811248:811666 [0] NCCL INFO Trees [0] -1/-1/-1->0->-1 [1] -1/-1/-1->0->-1 [2] -1/-1/-1->0->-1 [3] -1/-1/-1->0->-1 [4] -1/-1/-1->0->-1 [5] -1/-1/-1->0->-1 [6] -1/-1/-1->0->-1 [7] -1/-1/-1->0->-1 [8] -1/-1/-1->0->-1 [9] -1/-1/-1->0->-1 [10] -1/-1/-1->0->-1 [11] -1/-1/-1->0->-1 [12] -1/-1/-1->0->-1 [13] -1/-1/-1->0->-1 [14] -1/-1/-1->0->-1 [15] -1/-1/-1->0->-1 [16] -1/-1/-1->0->-1 [17] -1/-1/-1->0->-1 [18] -1/-1/-1->0->-1 [19] -1/-1/-1->0->-1 [20] -1/-1/-1->0->-1 [21] -1/-1/-1->0->-1 [22] -1/-1/-1->0->-1 [23] -1/-1/-1->0->-1 [24] -1/-1/-1->0->-1 [25] -1/-1/-1->0->-1 [26] -1/-1/-1->0->-1 [27] -1/-1/-1->0->-1 [28] -1/-1/-1->0->-1 [29] -1/-1/-1->0->-1 [30] -1/-1/-1->0->-1 [31] -1/-1/-1->0->-1
+ip-26-0-173-7:811248:811666 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-173-7:811248:811666 [0] NCCL INFO Connected all rings
+ip-26-0-173-7:811248:811666 [0] NCCL INFO Connected all trees
+ip-26-0-173-7:811248:811666 [0] NCCL INFO 32 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-252:688200:688603 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-172-116:1019292:1019679 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-172-116:1019292:1019679 [4] NCCL INFO Channel 00/32 :    0
+ip-26-0-172-116:1019292:1019679 [4] NCCL INFO Channel 01/32 :    0
+ip-26-0-172-116:1019292:1019679 [4] NCCL INFO Channel 02/32 :    0
+ip-26-0-172-116:1019292:1019679 [4] NCCL INFO Channel 03/32 :    0
+ip-26-0-172-116:1019292:1019679 [4] NCCL INFO Channel 04/32 :    0
+ip-26-0-172-116:1019292:1019679 [4] NCCL INFO Channel 05/32 :    0
+ip-26-0-172-116:1019292:1019679 [4] NCCL INFO Channel 06/32 :    0
+ip-26-0-172-116:1019292:1019679 [4] NCCL INFO Channel 07/32 :    0
+ip-26-0-172-116:1019292:1019679 [4] NCCL INFO Channel 08/32 :    0
+ip-26-0-172-116:1019292:1019679 [4] NCCL INFO Channel 09/32 :    0
+ip-26-0-172-116:1019292:1019679 [4] NCCL INFO Channel 10/32 :    0
+ip-26-0-172-116:1019292:1019679 [4] NCCL INFO Channel 11/32 :    0
+ip-26-0-172-116:1019292:1019679 [4] NCCL INFO Channel 12/32 :    0
+ip-26-0-172-116:1019292:1019679 [4] NCCL INFO Channel 13/32 :    0
+ip-26-0-172-116:1019292:1019679 [4] NCCL INFO Channel 14/32 :    0
+ip-26-0-172-116:1019292:1019679 [4] NCCL INFO Channel 15/32 :    0
+ip-26-0-172-116:1019292:1019679 [4] NCCL INFO Channel 16/32 :    0
+ip-26-0-172-116:1019292:1019679 [4] NCCL INFO Channel 17/32 :    0
+ip-26-0-172-116:1019292:1019679 [4] NCCL INFO Channel 18/32 :    0
+ip-26-0-172-116:1019292:1019679 [4] NCCL INFO Channel 19/32 :    0
+ip-26-0-172-116:1019292:1019679 [4] NCCL INFO Channel 20/32 :    0
+ip-26-0-172-116:1019292:1019679 [4] NCCL INFO Channel 21/32 :    0
+ip-26-0-172-116:1019292:1019679 [4] NCCL INFO Channel 22/32 :    0
+ip-26-0-172-116:1019292:1019679 [4] NCCL INFO Channel 23/32 :    0
+ip-26-0-172-116:1019292:1019679 [4] NCCL INFO Channel 24/32 :    0
+ip-26-0-172-116:1019292:1019679 [4] NCCL INFO Channel 25/32 :    0
+ip-26-0-172-116:1019292:1019679 [4] NCCL INFO Channel 26/32 :    0
+ip-26-0-172-116:1019292:1019679 [4] NCCL INFO Channel 27/32 :    0
+ip-26-0-172-116:1019292:1019679 [4] NCCL INFO Channel 28/32 :    0
+ip-26-0-172-116:1019292:1019679 [4] NCCL INFO Channel 29/32 :    0
+ip-26-0-172-116:1019292:1019679 [4] NCCL INFO Channel 30/32 :    0
+ip-26-0-172-116:1019292:1019679 [4] NCCL INFO Channel 31/32 :    0
+ip-26-0-172-116:1019292:1019679 [4] NCCL INFO Trees [0] -1/-1/-1->0->-1 [1] -1/-1/-1->0->-1 [2] -1/-1/-1->0->-1 [3] -1/-1/-1->0->-1 [4] -1/-1/-1->0->-1 [5] -1/-1/-1->0->-1 [6] -1/-1/-1->0->-1 [7] -1/-1/-1->0->-1 [8] -1/-1/-1->0->-1 [9] -1/-1/-1->0->-1 [10] -1/-1/-1->0->-1 [11] -1/-1/-1->0->-1 [12] -1/-1/-1->0->-1 [13] -1/-1/-1->0->-1 [14] -1/-1/-1->0->-1 [15] -1/-1/-1->0->-1 [16] -1/-1/-1->0->-1 [17] -1/-1/-1->0->-1 [18] -1/-1/-1->0->-1 [19] -1/-1/-1->0->-1 [20] -1/-1/-1->0->-1 [21] -1/-1/-1->0->-1 [22] -1/-1/-1->0->-1 [23] -1/-1/-1->0->-1 [24] -1/-1/-1->0->-1 [25] -1/-1/-1->0->-1 [26] -1/-1/-1->0->-1 [27] -1/-1/-1->0->-1 [28] -1/-1/-1->0->-1 [29] -1/-1/-1->0->-1 [30] -1/-1/-1->0->-1 [31] -1/-1/-1->0->-1
+ip-26-0-172-116:1019292:1019679 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-252:688200:688603 [0] NCCL INFO Channel 00/32 :    0
+ip-26-0-172-252:688200:688603 [0] NCCL INFO Channel 01/32 :    0
+ip-26-0-172-252:688200:688603 [0] NCCL INFO Channel 02/32 :    0
+ip-26-0-172-252:688200:688603 [0] NCCL INFO Channel 03/32 :    0
+ip-26-0-172-252:688200:688603 [0] NCCL INFO Channel 04/32 :    0
+ip-26-0-172-252:688200:688603 [0] NCCL INFO Channel 05/32 :    0
+ip-26-0-172-252:688200:688603 [0] NCCL INFO Channel 06/32 :    0
+ip-26-0-172-252:688200:688603 [0] NCCL INFO Channel 07/32 :    0
+ip-26-0-172-252:688200:688603 [0] NCCL INFO Channel 08/32 :    0
+ip-26-0-172-252:688200:688603 [0] NCCL INFO Channel 09/32 :    0
+ip-26-0-172-252:688200:688603 [0] NCCL INFO Channel 10/32 :    0
+ip-26-0-172-252:688200:688603 [0] NCCL INFO Channel 11/32 :    0
+ip-26-0-172-252:688200:688603 [0] NCCL INFO Channel 12/32 :    0
+ip-26-0-172-252:688200:688603 [0] NCCL INFO Channel 13/32 :    0
+ip-26-0-172-252:688200:688603 [0] NCCL INFO Channel 14/32 :    0
+ip-26-0-172-252:688200:688603 [0] NCCL INFO Channel 15/32 :    0
+ip-26-0-172-252:688200:688603 [0] NCCL INFO Channel 16/32 :    0
+ip-26-0-172-252:688200:688603 [0] NCCL INFO Channel 17/32 :    0
+ip-26-0-172-252:688200:688603 [0] NCCL INFO Channel 18/32 :    0
+ip-26-0-172-252:688200:688603 [0] NCCL INFO Channel 19/32 :    0
+ip-26-0-172-252:688200:688603 [0] NCCL INFO Channel 20/32 :    0
+ip-26-0-172-252:688200:688603 [0] NCCL INFO Channel 21/32 :    0
+ip-26-0-172-252:688200:688603 [0] NCCL INFO Channel 22/32 :    0
+ip-26-0-172-252:688200:688603 [0] NCCL INFO Channel 23/32 :    0
+ip-26-0-172-252:688200:688603 [0] NCCL INFO Channel 24/32 :    0
+ip-26-0-172-252:688200:688603 [0] NCCL INFO Channel 25/32 :    0
+ip-26-0-172-252:688200:688603 [0] NCCL INFO Channel 26/32 :    0
+ip-26-0-172-252:688200:688603 [0] NCCL INFO Channel 27/32 :    0
+ip-26-0-172-252:688200:688603 [0] NCCL INFO Channel 28/32 :    0
+ip-26-0-172-252:688200:688603 [0] NCCL INFO Channel 29/32 :    0
+ip-26-0-172-252:688200:688603 [0] NCCL INFO Channel 30/32 :    0
+ip-26-0-172-252:688200:688603 [0] NCCL INFO Channel 31/32 :    0
+ip-26-0-172-252:688200:688603 [0] NCCL INFO Trees [0] -1/-1/-1->0->-1 [1] -1/-1/-1->0->-1 [2] -1/-1/-1->0->-1 [3] -1/-1/-1->0->-1 [4] -1/-1/-1->0->-1 [5] -1/-1/-1->0->-1 [6] -1/-1/-1->0->-1 [7] -1/-1/-1->0->-1 [8] -1/-1/-1->0->-1 [9] -1/-1/-1->0->-1 [10] -1/-1/-1->0->-1 [11] -1/-1/-1->0->-1 [12] -1/-1/-1->0->-1 [13] -1/-1/-1->0->-1 [14] -1/-1/-1->0->-1 [15] -1/-1/-1->0->-1 [16] -1/-1/-1->0->-1 [17] -1/-1/-1->0->-1 [18] -1/-1/-1->0->-1 [19] -1/-1/-1->0->-1 [20] -1/-1/-1->0->-1 [21] -1/-1/-1->0->-1 [22] -1/-1/-1->0->-1 [23] -1/-1/-1->0->-1 [24] -1/-1/-1->0->-1 [25] -1/-1/-1->0->-1 [26] -1/-1/-1->0->-1 [27] -1/-1/-1->0->-1 [28] -1/-1/-1->0->-1 [29] -1/-1/-1->0->-1 [30] -1/-1/-1->0->-1 [31] -1/-1/-1->0->-1
+ip-26-0-172-252:688200:688603 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-116:1019292:1019679 [4] NCCL INFO Connected all rings
+ip-26-0-172-116:1019292:1019679 [4] NCCL INFO Connected all trees
+ip-26-0-172-116:1019292:1019679 [4] NCCL INFO 32 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-252:688200:688603 [0] NCCL INFO Connected all rings
+ip-26-0-172-252:688200:688603 [0] NCCL INFO Connected all trees
+ip-26-0-172-252:688200:688603 [0] NCCL INFO 32 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-73:926601:926987 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-172-73:926601:926987 [7] NCCL INFO Channel 00/32 :    0
+ip-26-0-172-73:926601:926987 [7] NCCL INFO Channel 01/32 :    0
+ip-26-0-172-73:926601:926987 [7] NCCL INFO Channel 02/32 :    0
+ip-26-0-172-73:926601:926987 [7] NCCL INFO Channel 03/32 :    0
+ip-26-0-172-73:926601:926987 [7] NCCL INFO Channel 04/32 :    0
+ip-26-0-172-73:926601:926987 [7] NCCL INFO Channel 05/32 :    0
+ip-26-0-172-73:926601:926987 [7] NCCL INFO Channel 06/32 :    0
+ip-26-0-172-73:926601:926987 [7] NCCL INFO Channel 07/32 :    0
+ip-26-0-172-73:926601:926987 [7] NCCL INFO Channel 08/32 :    0
+ip-26-0-172-73:926601:926987 [7] NCCL INFO Channel 09/32 :    0
+ip-26-0-172-73:926601:926987 [7] NCCL INFO Channel 10/32 :    0
+ip-26-0-172-73:926601:926987 [7] NCCL INFO Channel 11/32 :    0
+ip-26-0-172-73:926601:926987 [7] NCCL INFO Channel 12/32 :    0
+ip-26-0-172-73:926601:926987 [7] NCCL INFO Channel 13/32 :    0
+ip-26-0-172-73:926601:926987 [7] NCCL INFO Channel 14/32 :    0
+ip-26-0-172-73:926601:926987 [7] NCCL INFO Channel 15/32 :    0
+ip-26-0-172-73:926601:926987 [7] NCCL INFO Channel 16/32 :    0
+ip-26-0-172-73:926601:926987 [7] NCCL INFO Channel 17/32 :    0
+ip-26-0-172-73:926601:926987 [7] NCCL INFO Channel 18/32 :    0
+ip-26-0-172-73:926601:926987 [7] NCCL INFO Channel 19/32 :    0
+ip-26-0-172-73:926601:926987 [7] NCCL INFO Channel 20/32 :    0
+ip-26-0-172-73:926601:926987 [7] NCCL INFO Channel 21/32 :    0
+ip-26-0-172-73:926601:926987 [7] NCCL INFO Channel 22/32 :    0
+ip-26-0-172-73:926601:926987 [7] NCCL INFO Channel 23/32 :    0
+ip-26-0-172-73:926601:926987 [7] NCCL INFO Channel 24/32 :    0
+ip-26-0-172-73:926601:926987 [7] NCCL INFO Channel 25/32 :    0
+ip-26-0-172-73:926601:926987 [7] NCCL INFO Channel 26/32 :    0
+ip-26-0-172-73:926601:926987 [7] NCCL INFO Channel 27/32 :    0
+ip-26-0-172-73:926601:926987 [7] NCCL INFO Channel 28/32 :    0
+ip-26-0-172-73:926601:926987 [7] NCCL INFO Channel 29/32 :    0
+ip-26-0-172-73:926601:926987 [7] NCCL INFO Channel 30/32 :    0
+ip-26-0-172-73:926601:926987 [7] NCCL INFO Channel 31/32 :    0
+ip-26-0-172-73:926601:926987 [7] NCCL INFO Trees [0] -1/-1/-1->0->-1 [1] -1/-1/-1->0->-1 [2] -1/-1/-1->0->-1 [3] -1/-1/-1->0->-1 [4] -1/-1/-1->0->-1 [5] -1/-1/-1->0->-1 [6] -1/-1/-1->0->-1 [7] -1/-1/-1->0->-1 [8] -1/-1/-1->0->-1 [9] -1/-1/-1->0->-1 [10] -1/-1/-1->0->-1 [11] -1/-1/-1->0->-1 [12] -1/-1/-1->0->-1 [13] -1/-1/-1->0->-1 [14] -1/-1/-1->0->-1 [15] -1/-1/-1->0->-1 [16] -1/-1/-1->0->-1 [17] -1/-1/-1->0->-1 [18] -1/-1/-1->0->-1 [19] -1/-1/-1->0->-1 [20] -1/-1/-1->0->-1 [21] -1/-1/-1->0->-1 [22] -1/-1/-1->0->-1 [23] -1/-1/-1->0->-1 [24] -1/-1/-1->0->-1 [25] -1/-1/-1->0->-1 [26] -1/-1/-1->0->-1 [27] -1/-1/-1->0->-1 [28] -1/-1/-1->0->-1 [29] -1/-1/-1->0->-1 [30] -1/-1/-1->0->-1 [31] -1/-1/-1->0->-1
+ip-26-0-172-73:926601:926987 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-116:1019291:1019684 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-172-73:926600:926988 [6] NCCL INFO comm 0xb523290 rank 0 nranks 1 cudaDev 6 nvmlDev 6 busId b9000 commId 0xfd8c01f769e5d9c3 - Init COMPLETE
+ip-26-0-172-116:1019291:1019684 [3] NCCL INFO Channel 00/32 :    0
+ip-26-0-172-116:1019291:1019684 [3] NCCL INFO Channel 01/32 :    0
+ip-26-0-172-116:1019291:1019684 [3] NCCL INFO Channel 02/32 :    0
+ip-26-0-172-116:1019291:1019684 [3] NCCL INFO Channel 03/32 :    0
+ip-26-0-172-116:1019291:1019684 [3] NCCL INFO Channel 04/32 :    0
+ip-26-0-172-116:1019291:1019684 [3] NCCL INFO Channel 05/32 :    0
+ip-26-0-172-116:1019291:1019684 [3] NCCL INFO Channel 06/32 :    0
+ip-26-0-172-116:1019291:1019684 [3] NCCL INFO Channel 07/32 :    0
+ip-26-0-172-116:1019291:1019684 [3] NCCL INFO Channel 08/32 :    0
+ip-26-0-172-116:1019291:1019684 [3] NCCL INFO Channel 09/32 :    0
+ip-26-0-172-116:1019291:1019684 [3] NCCL INFO Channel 10/32 :    0
+ip-26-0-172-116:1019291:1019684 [3] NCCL INFO Channel 11/32 :    0
+ip-26-0-172-116:1019291:1019684 [3] NCCL INFO Channel 12/32 :    0
+ip-26-0-172-116:1019291:1019684 [3] NCCL INFO Channel 13/32 :    0
+ip-26-0-172-116:1019291:1019684 [3] NCCL INFO Channel 14/32 :    0
+ip-26-0-172-116:1019291:1019684 [3] NCCL INFO Channel 15/32 :    0
+ip-26-0-172-116:1019291:1019684 [3] NCCL INFO Channel 16/32 :    0
+ip-26-0-172-116:1019291:1019684 [3] NCCL INFO Channel 17/32 :    0
+ip-26-0-172-116:1019291:1019684 [3] NCCL INFO Channel 18/32 :    0
+ip-26-0-172-116:1019291:1019684 [3] NCCL INFO Channel 19/32 :    0
+ip-26-0-172-116:1019291:1019684 [3] NCCL INFO Channel 20/32 :    0
+ip-26-0-172-116:1019291:1019684 [3] NCCL INFO Channel 21/32 :    0
+ip-26-0-172-116:1019291:1019684 [3] NCCL INFO Channel 22/32 :    0
+ip-26-0-172-116:1019291:1019684 [3] NCCL INFO Channel 23/32 :    0
+ip-26-0-172-116:1019291:1019684 [3] NCCL INFO Channel 24/32 :    0
+ip-26-0-172-116:1019291:1019684 [3] NCCL INFO Channel 25/32 :    0
+ip-26-0-172-116:1019291:1019684 [3] NCCL INFO Channel 26/32 :    0
+ip-26-0-172-116:1019291:1019684 [3] NCCL INFO Channel 27/32 :    0
+ip-26-0-172-116:1019291:1019684 [3] NCCL INFO Channel 28/32 :    0
+ip-26-0-172-116:1019291:1019684 [3] NCCL INFO Channel 29/32 :    0
+ip-26-0-172-116:1019291:1019684 [3] NCCL INFO Channel 30/32 :    0
+ip-26-0-172-116:1019291:1019684 [3] NCCL INFO Channel 31/32 :    0
+ip-26-0-172-116:1019291:1019684 [3] NCCL INFO Trees [0] -1/-1/-1->0->-1 [1] -1/-1/-1->0->-1 [2] -1/-1/-1->0->-1 [3] -1/-1/-1->0->-1 [4] -1/-1/-1->0->-1 [5] -1/-1/-1->0->-1 [6] -1/-1/-1->0->-1 [7] -1/-1/-1->0->-1 [8] -1/-1/-1->0->-1 [9] -1/-1/-1->0->-1 [10] -1/-1/-1->0->-1 [11] -1/-1/-1->0->-1 [12] -1/-1/-1->0->-1 [13] -1/-1/-1->0->-1 [14] -1/-1/-1->0->-1 [15] -1/-1/-1->0->-1 [16] -1/-1/-1->0->-1 [17] -1/-1/-1->0->-1 [18] -1/-1/-1->0->-1 [19] -1/-1/-1->0->-1 [20] -1/-1/-1->0->-1 [21] -1/-1/-1->0->-1 [22] -1/-1/-1->0->-1 [23] -1/-1/-1->0->-1 [24] -1/-1/-1->0->-1 [25] -1/-1/-1->0->-1 [26] -1/-1/-1->0->-1 [27] -1/-1/-1->0->-1 [28] -1/-1/-1->0->-1 [29] -1/-1/-1->0->-1 [30] -1/-1/-1->0->-1 [31] -1/-1/-1->0->-1
+ip-26-0-172-116:1019291:1019684 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-73:926601:926987 [7] NCCL INFO Connected all rings
+ip-26-0-172-73:926601:926987 [7] NCCL INFO Connected all trees
+ip-26-0-172-73:926601:926987 [7] NCCL INFO 32 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-73:926595:926986 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-172-116:1019291:1019684 [3] NCCL INFO Connected all rings
+ip-26-0-172-116:1019291:1019684 [3] NCCL INFO Connected all trees
+ip-26-0-172-116:1019291:1019684 [3] NCCL INFO 32 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-73:926595:926986 [1] NCCL INFO Channel 00/32 :    0
+ip-26-0-172-73:926595:926986 [1] NCCL INFO Channel 01/32 :    0
+ip-26-0-172-73:926595:926986 [1] NCCL INFO Channel 02/32 :    0
+ip-26-0-172-73:926595:926986 [1] NCCL INFO Channel 03/32 :    0
+ip-26-0-172-73:926595:926986 [1] NCCL INFO Channel 04/32 :    0
+ip-26-0-172-73:926595:926986 [1] NCCL INFO Channel 05/32 :    0
+ip-26-0-172-73:926595:926986 [1] NCCL INFO Channel 06/32 :    0
+ip-26-0-172-73:926595:926986 [1] NCCL INFO Channel 07/32 :    0
+ip-26-0-172-73:926595:926986 [1] NCCL INFO Channel 08/32 :    0
+ip-26-0-172-73:926595:926986 [1] NCCL INFO Channel 09/32 :    0
+ip-26-0-172-73:926595:926986 [1] NCCL INFO Channel 10/32 :    0
+ip-26-0-172-73:926595:926986 [1] NCCL INFO Channel 11/32 :    0
+ip-26-0-172-73:926595:926986 [1] NCCL INFO Channel 12/32 :    0
+ip-26-0-172-73:926595:926986 [1] NCCL INFO Channel 13/32 :    0
+ip-26-0-172-73:926595:926986 [1] NCCL INFO Channel 14/32 :    0
+ip-26-0-172-73:926595:926986 [1] NCCL INFO Channel 15/32 :    0
+ip-26-0-172-73:926595:926986 [1] NCCL INFO Channel 16/32 :    0
+ip-26-0-172-73:926595:926986 [1] NCCL INFO Channel 17/32 :    0
+ip-26-0-172-73:926595:926986 [1] NCCL INFO Channel 18/32 :    0
+ip-26-0-172-73:926595:926986 [1] NCCL INFO Channel 19/32 :    0
+ip-26-0-172-73:926595:926986 [1] NCCL INFO Channel 20/32 :    0
+ip-26-0-172-73:926595:926986 [1] NCCL INFO Channel 21/32 :    0
+ip-26-0-172-73:926595:926986 [1] NCCL INFO Channel 22/32 :    0
+ip-26-0-172-73:926595:926986 [1] NCCL INFO Channel 23/32 :    0
+ip-26-0-172-73:926595:926986 [1] NCCL INFO Channel 24/32 :    0
+ip-26-0-172-73:926595:926986 [1] NCCL INFO Channel 25/32 :    0
+ip-26-0-172-73:926595:926986 [1] NCCL INFO Channel 26/32 :    0
+ip-26-0-172-73:926595:926986 [1] NCCL INFO Channel 27/32 :    0
+ip-26-0-172-73:926595:926986 [1] NCCL INFO Channel 28/32 :    0
+ip-26-0-172-73:926595:926986 [1] NCCL INFO Channel 29/32 :    0
+ip-26-0-172-73:926595:926986 [1] NCCL INFO Channel 30/32 :    0
+ip-26-0-172-73:926595:926986 [1] NCCL INFO Channel 31/32 :    0
+ip-26-0-172-73:926595:926986 [1] NCCL INFO Trees [0] -1/-1/-1->0->-1 [1] -1/-1/-1->0->-1 [2] -1/-1/-1->0->-1 [3] -1/-1/-1->0->-1 [4] -1/-1/-1->0->-1 [5] -1/-1/-1->0->-1 [6] -1/-1/-1->0->-1 [7] -1/-1/-1->0->-1 [8] -1/-1/-1->0->-1 [9] -1/-1/-1->0->-1 [10] -1/-1/-1->0->-1 [11] -1/-1/-1->0->-1 [12] -1/-1/-1->0->-1 [13] -1/-1/-1->0->-1 [14] -1/-1/-1->0->-1 [15] -1/-1/-1->0->-1 [16] -1/-1/-1->0->-1 [17] -1/-1/-1->0->-1 [18] -1/-1/-1->0->-1 [19] -1/-1/-1->0->-1 [20] -1/-1/-1->0->-1 [21] -1/-1/-1->0->-1 [22] -1/-1/-1->0->-1 [23] -1/-1/-1->0->-1 [24] -1/-1/-1->0->-1 [25] -1/-1/-1->0->-1 [26] -1/-1/-1->0->-1 [27] -1/-1/-1->0->-1 [28] -1/-1/-1->0->-1 [29] -1/-1/-1->0->-1 [30] -1/-1/-1->0->-1 [31] -1/-1/-1->0->-1
+ip-26-0-172-73:926595:926986 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-73:926595:926986 [1] NCCL INFO Connected all rings
+ip-26-0-172-73:926595:926986 [1] NCCL INFO Connected all trees
+ip-26-0-172-73:926595:926986 [1] NCCL INFO 32 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-73:926598:926997 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-172-73:926598:926997 [4] NCCL INFO Channel 00/32 :    0
+ip-26-0-172-73:926598:926997 [4] NCCL INFO Channel 01/32 :    0
+ip-26-0-172-73:926598:926997 [4] NCCL INFO Channel 02/32 :    0
+ip-26-0-172-73:926598:926997 [4] NCCL INFO Channel 03/32 :    0
+ip-26-0-172-73:926598:926997 [4] NCCL INFO Channel 04/32 :    0
+ip-26-0-172-73:926598:926997 [4] NCCL INFO Channel 05/32 :    0
+ip-26-0-172-73:926598:926997 [4] NCCL INFO Channel 06/32 :    0
+ip-26-0-172-73:926598:926997 [4] NCCL INFO Channel 07/32 :    0
+ip-26-0-172-73:926598:926997 [4] NCCL INFO Channel 08/32 :    0
+ip-26-0-172-73:926598:926997 [4] NCCL INFO Channel 09/32 :    0
+ip-26-0-172-73:926598:926997 [4] NCCL INFO Channel 10/32 :    0
+ip-26-0-172-73:926598:926997 [4] NCCL INFO Channel 11/32 :    0
+ip-26-0-172-73:926598:926997 [4] NCCL INFO Channel 12/32 :    0
+ip-26-0-172-73:926598:926997 [4] NCCL INFO Channel 13/32 :    0
+ip-26-0-172-73:926598:926997 [4] NCCL INFO Channel 14/32 :    0
+ip-26-0-172-73:926598:926997 [4] NCCL INFO Channel 15/32 :    0
+ip-26-0-172-73:926598:926997 [4] NCCL INFO Channel 16/32 :    0
+ip-26-0-172-73:926598:926997 [4] NCCL INFO Channel 17/32 :    0
+ip-26-0-172-73:926598:926997 [4] NCCL INFO Channel 18/32 :    0
+ip-26-0-172-73:926598:926997 [4] NCCL INFO Channel 19/32 :    0
+ip-26-0-172-73:926598:926997 [4] NCCL INFO Channel 20/32 :    0
+ip-26-0-172-73:926598:926997 [4] NCCL INFO Channel 21/32 :    0
+ip-26-0-172-73:926598:926997 [4] NCCL INFO Channel 22/32 :    0
+ip-26-0-172-73:926598:926997 [4] NCCL INFO Channel 23/32 :    0
+ip-26-0-172-73:926598:926997 [4] NCCL INFO Channel 24/32 :    0
+ip-26-0-172-73:926598:926997 [4] NCCL INFO Channel 25/32 :    0
+ip-26-0-172-73:926598:926997 [4] NCCL INFO Channel 26/32 :    0
+ip-26-0-172-73:926598:926997 [4] NCCL INFO Channel 27/32 :    0
+ip-26-0-172-73:926598:926997 [4] NCCL INFO Channel 28/32 :    0
+ip-26-0-172-73:926598:926997 [4] NCCL INFO Channel 29/32 :    0
+ip-26-0-172-73:926598:926997 [4] NCCL INFO Channel 30/32 :    0
+ip-26-0-172-73:926598:926997 [4] NCCL INFO Channel 31/32 :    0
+ip-26-0-172-73:926598:926997 [4] NCCL INFO Trees [0] -1/-1/-1->0->-1 [1] -1/-1/-1->0->-1 [2] -1/-1/-1->0->-1 [3] -1/-1/-1->0->-1 [4] -1/-1/-1->0->-1 [5] -1/-1/-1->0->-1 [6] -1/-1/-1->0->-1 [7] -1/-1/-1->0->-1 [8] -1/-1/-1->0->-1 [9] -1/-1/-1->0->-1 [10] -1/-1/-1->0->-1 [11] -1/-1/-1->0->-1 [12] -1/-1/-1->0->-1 [13] -1/-1/-1->0->-1 [14] -1/-1/-1->0->-1 [15] -1/-1/-1->0->-1 [16] -1/-1/-1->0->-1 [17] -1/-1/-1->0->-1 [18] -1/-1/-1->0->-1 [19] -1/-1/-1->0->-1 [20] -1/-1/-1->0->-1 [21] -1/-1/-1->0->-1 [22] -1/-1/-1->0->-1 [23] -1/-1/-1->0->-1 [24] -1/-1/-1->0->-1 [25] -1/-1/-1->0->-1 [26] -1/-1/-1->0->-1 [27] -1/-1/-1->0->-1 [28] -1/-1/-1->0->-1 [29] -1/-1/-1->0->-1 [30] -1/-1/-1->0->-1 [31] -1/-1/-1->0->-1
+ip-26-0-172-73:926598:926997 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-73:926598:926997 [4] NCCL INFO Connected all rings
+ip-26-0-172-73:926598:926997 [4] NCCL INFO Connected all trees
+ip-26-0-172-252:688203:688593 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-172-73:926598:926997 [4] NCCL INFO 32 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-252:688203:688593 [3] NCCL INFO Channel 00/32 :    0
+ip-26-0-172-252:688203:688593 [3] NCCL INFO Channel 01/32 :    0
+ip-26-0-172-252:688203:688593 [3] NCCL INFO Channel 02/32 :    0
+ip-26-0-172-252:688203:688593 [3] NCCL INFO Channel 03/32 :    0
+ip-26-0-172-252:688203:688593 [3] NCCL INFO Channel 04/32 :    0
+ip-26-0-172-252:688203:688593 [3] NCCL INFO Channel 05/32 :    0
+ip-26-0-172-252:688203:688593 [3] NCCL INFO Channel 06/32 :    0
+ip-26-0-172-252:688203:688593 [3] NCCL INFO Channel 07/32 :    0
+ip-26-0-172-252:688203:688593 [3] NCCL INFO Channel 08/32 :    0
+ip-26-0-172-252:688203:688593 [3] NCCL INFO Channel 09/32 :    0
+ip-26-0-172-252:688203:688593 [3] NCCL INFO Channel 10/32 :    0
+ip-26-0-172-252:688203:688593 [3] NCCL INFO Channel 11/32 :    0
+ip-26-0-172-252:688203:688593 [3] NCCL INFO Channel 12/32 :    0
+ip-26-0-172-252:688203:688593 [3] NCCL INFO Channel 13/32 :    0
+ip-26-0-172-252:688203:688593 [3] NCCL INFO Channel 14/32 :    0
+ip-26-0-172-252:688203:688593 [3] NCCL INFO Channel 15/32 :    0
+ip-26-0-172-252:688203:688593 [3] NCCL INFO Channel 16/32 :    0
+ip-26-0-172-252:688203:688593 [3] NCCL INFO Channel 17/32 :    0
+ip-26-0-172-252:688203:688593 [3] NCCL INFO Channel 18/32 :    0
+ip-26-0-172-252:688203:688593 [3] NCCL INFO Channel 19/32 :    0
+ip-26-0-172-252:688203:688593 [3] NCCL INFO Channel 20/32 :    0
+ip-26-0-172-252:688203:688593 [3] NCCL INFO Channel 21/32 :    0
+ip-26-0-172-252:688203:688593 [3] NCCL INFO Channel 22/32 :    0
+ip-26-0-172-252:688203:688593 [3] NCCL INFO Channel 23/32 :    0
+ip-26-0-172-252:688203:688593 [3] NCCL INFO Channel 24/32 :    0
+ip-26-0-172-252:688203:688593 [3] NCCL INFO Channel 25/32 :    0
+ip-26-0-172-252:688203:688593 [3] NCCL INFO Channel 26/32 :    0
+ip-26-0-172-252:688203:688593 [3] NCCL INFO Channel 27/32 :    0
+ip-26-0-172-252:688203:688593 [3] NCCL INFO Channel 28/32 :    0
+ip-26-0-172-252:688203:688593 [3] NCCL INFO Channel 29/32 :    0
+ip-26-0-172-252:688203:688593 [3] NCCL INFO Channel 30/32 :    0
+ip-26-0-172-252:688203:688593 [3] NCCL INFO Channel 31/32 :    0
+ip-26-0-172-252:688203:688593 [3] NCCL INFO Trees [0] -1/-1/-1->0->-1 [1] -1/-1/-1->0->-1 [2] -1/-1/-1->0->-1 [3] -1/-1/-1->0->-1 [4] -1/-1/-1->0->-1 [5] -1/-1/-1->0->-1 [6] -1/-1/-1->0->-1 [7] -1/-1/-1->0->-1 [8] -1/-1/-1->0->-1 [9] -1/-1/-1->0->-1 [10] -1/-1/-1->0->-1 [11] -1/-1/-1->0->-1 [12] -1/-1/-1->0->-1 [13] -1/-1/-1->0->-1 [14] -1/-1/-1->0->-1 [15] -1/-1/-1->0->-1 [16] -1/-1/-1->0->-1 [17] -1/-1/-1->0->-1 [18] -1/-1/-1->0->-1 [19] -1/-1/-1->0->-1 [20] -1/-1/-1->0->-1 [21] -1/-1/-1->0->-1 [22] -1/-1/-1->0->-1 [23] -1/-1/-1->0->-1 [24] -1/-1/-1->0->-1 [25] -1/-1/-1->0->-1 [26] -1/-1/-1->0->-1 [27] -1/-1/-1->0->-1 [28] -1/-1/-1->0->-1 [29] -1/-1/-1->0->-1 [30] -1/-1/-1->0->-1 [31] -1/-1/-1->0->-1
+ip-26-0-172-252:688203:688593 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-225:94345:94779 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-160-225:94345:94779 [3] NCCL INFO Channel 00/32 :    0
+ip-26-0-160-225:94345:94779 [3] NCCL INFO Channel 01/32 :    0
+ip-26-0-160-225:94345:94779 [3] NCCL INFO Channel 02/32 :    0
+ip-26-0-160-225:94345:94779 [3] NCCL INFO Channel 03/32 :    0
+ip-26-0-160-225:94345:94779 [3] NCCL INFO Channel 04/32 :    0
+ip-26-0-160-225:94345:94779 [3] NCCL INFO Channel 05/32 :    0
+ip-26-0-160-225:94345:94779 [3] NCCL INFO Channel 06/32 :    0
+ip-26-0-160-225:94345:94779 [3] NCCL INFO Channel 07/32 :    0
+ip-26-0-160-225:94345:94779 [3] NCCL INFO Channel 08/32 :    0
+ip-26-0-160-225:94345:94779 [3] NCCL INFO Channel 09/32 :    0
+ip-26-0-160-225:94345:94779 [3] NCCL INFO Channel 10/32 :    0
+ip-26-0-160-225:94345:94779 [3] NCCL INFO Channel 11/32 :    0
+ip-26-0-160-225:94345:94779 [3] NCCL INFO Channel 12/32 :    0
+ip-26-0-160-225:94345:94779 [3] NCCL INFO Channel 13/32 :    0
+ip-26-0-160-225:94345:94779 [3] NCCL INFO Channel 14/32 :    0
+ip-26-0-160-225:94345:94779 [3] NCCL INFO Channel 15/32 :    0
+ip-26-0-160-225:94345:94779 [3] NCCL INFO Channel 16/32 :    0
+ip-26-0-160-225:94345:94779 [3] NCCL INFO Channel 17/32 :    0
+ip-26-0-160-225:94345:94779 [3] NCCL INFO Channel 18/32 :    0
+ip-26-0-160-225:94345:94779 [3] NCCL INFO Channel 19/32 :    0
+ip-26-0-160-225:94345:94779 [3] NCCL INFO Channel 20/32 :    0
+ip-26-0-160-225:94345:94779 [3] NCCL INFO Channel 21/32 :    0
+ip-26-0-160-225:94345:94779 [3] NCCL INFO Channel 22/32 :    0
+ip-26-0-160-225:94345:94779 [3] NCCL INFO Channel 23/32 :    0
+ip-26-0-160-225:94345:94779 [3] NCCL INFO Channel 24/32 :    0
+ip-26-0-160-225:94345:94779 [3] NCCL INFO Channel 25/32 :    0
+ip-26-0-160-225:94345:94779 [3] NCCL INFO Channel 26/32 :    0
+ip-26-0-160-225:94345:94779 [3] NCCL INFO Channel 27/32 :    0
+ip-26-0-160-225:94345:94779 [3] NCCL INFO Channel 28/32 :    0
+ip-26-0-160-225:94345:94779 [3] NCCL INFO Channel 29/32 :    0
+ip-26-0-160-225:94345:94779 [3] NCCL INFO Channel 30/32 :    0
+ip-26-0-160-225:94345:94779 [3] NCCL INFO Channel 31/32 :    0
+ip-26-0-160-225:94345:94779 [3] NCCL INFO Trees [0] -1/-1/-1->0->-1 [1] -1/-1/-1->0->-1 [2] -1/-1/-1->0->-1 [3] -1/-1/-1->0->-1 [4] -1/-1/-1->0->-1 [5] -1/-1/-1->0->-1 [6] -1/-1/-1->0->-1 [7] -1/-1/-1->0->-1 [8] -1/-1/-1->0->-1 [9] -1/-1/-1->0->-1 [10] -1/-1/-1->0->-1 [11] -1/-1/-1->0->-1 [12] -1/-1/-1->0->-1 [13] -1/-1/-1->0->-1 [14] -1/-1/-1->0->-1 [15] -1/-1/-1->0->-1 [16] -1/-1/-1->0->-1 [17] -1/-1/-1->0->-1 [18] -1/-1/-1->0->-1 [19] -1/-1/-1->0->-1 [20] -1/-1/-1->0->-1 [21] -1/-1/-1->0->-1 [22] -1/-1/-1->0->-1 [23] -1/-1/-1->0->-1 [24] -1/-1/-1->0->-1 [25] -1/-1/-1->0->-1 [26] -1/-1/-1->0->-1 [27] -1/-1/-1->0->-1 [28] -1/-1/-1->0->-1 [29] -1/-1/-1->0->-1 [30] -1/-1/-1->0->-1 [31] -1/-1/-1->0->-1
+ip-26-0-172-252:688203:688593 [3] NCCL INFO Connected all rings
+ip-26-0-172-252:688203:688593 [3] NCCL INFO Connected all trees
+ip-26-0-160-225:94345:94779 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-252:688203:688593 [3] NCCL INFO 32 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-73:926597:926992 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-160-225:94345:94779 [3] NCCL INFO Connected all rings
+ip-26-0-160-225:94345:94779 [3] NCCL INFO Connected all trees
+ip-26-0-160-225:94345:94779 [3] NCCL INFO 32 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-73:926597:926992 [3] NCCL INFO Channel 00/32 :    0
+ip-26-0-172-73:926597:926992 [3] NCCL INFO Channel 01/32 :    0
+ip-26-0-172-73:926597:926992 [3] NCCL INFO Channel 02/32 :    0
+ip-26-0-172-73:926597:926992 [3] NCCL INFO Channel 03/32 :    0
+ip-26-0-172-73:926597:926992 [3] NCCL INFO Channel 04/32 :    0
+ip-26-0-172-73:926597:926992 [3] NCCL INFO Channel 05/32 :    0
+ip-26-0-172-73:926597:926992 [3] NCCL INFO Channel 06/32 :    0
+ip-26-0-172-73:926597:926992 [3] NCCL INFO Channel 07/32 :    0
+ip-26-0-172-73:926597:926992 [3] NCCL INFO Channel 08/32 :    0
+ip-26-0-172-73:926597:926992 [3] NCCL INFO Channel 09/32 :    0
+ip-26-0-172-73:926597:926992 [3] NCCL INFO Channel 10/32 :    0
+ip-26-0-172-73:926597:926992 [3] NCCL INFO Channel 11/32 :    0
+ip-26-0-172-73:926597:926992 [3] NCCL INFO Channel 12/32 :    0
+ip-26-0-172-73:926597:926992 [3] NCCL INFO Channel 13/32 :    0
+ip-26-0-172-73:926597:926992 [3] NCCL INFO Channel 14/32 :    0
+ip-26-0-172-73:926597:926992 [3] NCCL INFO Channel 15/32 :    0
+ip-26-0-172-73:926597:926992 [3] NCCL INFO Channel 16/32 :    0
+ip-26-0-172-73:926597:926992 [3] NCCL INFO Channel 17/32 :    0
+ip-26-0-172-73:926597:926992 [3] NCCL INFO Channel 18/32 :    0
+ip-26-0-172-73:926597:926992 [3] NCCL INFO Channel 19/32 :    0
+ip-26-0-172-73:926597:926992 [3] NCCL INFO Channel 20/32 :    0
+ip-26-0-172-73:926597:926992 [3] NCCL INFO Channel 21/32 :    0
+ip-26-0-172-73:926597:926992 [3] NCCL INFO Channel 22/32 :    0
+ip-26-0-172-73:926597:926992 [3] NCCL INFO Channel 23/32 :    0
+ip-26-0-172-73:926597:926992 [3] NCCL INFO Channel 24/32 :    0
+ip-26-0-172-73:926597:926992 [3] NCCL INFO Channel 25/32 :    0
+ip-26-0-172-73:926597:926992 [3] NCCL INFO Channel 26/32 :    0
+ip-26-0-172-73:926597:926992 [3] NCCL INFO Channel 27/32 :    0
+ip-26-0-172-73:926597:926992 [3] NCCL INFO Channel 28/32 :    0
+ip-26-0-172-73:926597:926992 [3] NCCL INFO Channel 29/32 :    0
+ip-26-0-172-73:926597:926992 [3] NCCL INFO Channel 30/32 :    0
+ip-26-0-172-73:926597:926992 [3] NCCL INFO Channel 31/32 :    0
+ip-26-0-172-73:926597:926992 [3] NCCL INFO Trees [0] -1/-1/-1->0->-1 [1] -1/-1/-1->0->-1 [2] -1/-1/-1->0->-1 [3] -1/-1/-1->0->-1 [4] -1/-1/-1->0->-1 [5] -1/-1/-1->0->-1 [6] -1/-1/-1->0->-1 [7] -1/-1/-1->0->-1 [8] -1/-1/-1->0->-1 [9] -1/-1/-1->0->-1 [10] -1/-1/-1->0->-1 [11] -1/-1/-1->0->-1 [12] -1/-1/-1->0->-1 [13] -1/-1/-1->0->-1 [14] -1/-1/-1->0->-1 [15] -1/-1/-1->0->-1 [16] -1/-1/-1->0->-1 [17] -1/-1/-1->0->-1 [18] -1/-1/-1->0->-1 [19] -1/-1/-1->0->-1 [20] -1/-1/-1->0->-1 [21] -1/-1/-1->0->-1 [22] -1/-1/-1->0->-1 [23] -1/-1/-1->0->-1 [24] -1/-1/-1->0->-1 [25] -1/-1/-1->0->-1 [26] -1/-1/-1->0->-1 [27] -1/-1/-1->0->-1 [28] -1/-1/-1->0->-1 [29] -1/-1/-1->0->-1 [30] -1/-1/-1->0->-1 [31] -1/-1/-1->0->-1
+ip-26-0-172-73:926597:926992 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-73:926597:926992 [3] NCCL INFO Connected all rings
+ip-26-0-172-73:926597:926992 [3] NCCL INFO Connected all trees
+ip-26-0-172-73:926597:926992 [3] NCCL INFO 32 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-147:618587:618986 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-172-147:618587:618986 [3] NCCL INFO Channel 00/32 :    0
+ip-26-0-172-147:618587:618986 [3] NCCL INFO Channel 01/32 :    0
+ip-26-0-172-147:618587:618986 [3] NCCL INFO Channel 02/32 :    0
+ip-26-0-172-147:618587:618986 [3] NCCL INFO Channel 03/32 :    0
+ip-26-0-172-147:618587:618986 [3] NCCL INFO Channel 04/32 :    0
+ip-26-0-172-147:618587:618986 [3] NCCL INFO Channel 05/32 :    0
+ip-26-0-172-147:618587:618986 [3] NCCL INFO Channel 06/32 :    0
+ip-26-0-172-147:618587:618986 [3] NCCL INFO Channel 07/32 :    0
+ip-26-0-172-147:618587:618986 [3] NCCL INFO Channel 08/32 :    0
+ip-26-0-172-147:618587:618986 [3] NCCL INFO Channel 09/32 :    0
+ip-26-0-172-147:618587:618986 [3] NCCL INFO Channel 10/32 :    0
+ip-26-0-172-147:618587:618986 [3] NCCL INFO Channel 11/32 :    0
+ip-26-0-172-147:618587:618986 [3] NCCL INFO Channel 12/32 :    0
+ip-26-0-172-147:618587:618986 [3] NCCL INFO Channel 13/32 :    0
+ip-26-0-172-147:618587:618986 [3] NCCL INFO Channel 14/32 :    0
+ip-26-0-172-147:618587:618986 [3] NCCL INFO Channel 15/32 :    0
+ip-26-0-172-147:618587:618986 [3] NCCL INFO Channel 16/32 :    0
+ip-26-0-172-147:618587:618986 [3] NCCL INFO Channel 17/32 :    0
+ip-26-0-172-147:618587:618986 [3] NCCL INFO Channel 18/32 :    0
+ip-26-0-172-147:618587:618986 [3] NCCL INFO Channel 19/32 :    0
+ip-26-0-172-147:618587:618986 [3] NCCL INFO Channel 20/32 :    0
+ip-26-0-172-147:618587:618986 [3] NCCL INFO Channel 21/32 :    0
+ip-26-0-172-147:618587:618986 [3] NCCL INFO Channel 22/32 :    0
+ip-26-0-172-147:618587:618986 [3] NCCL INFO Channel 23/32 :    0
+ip-26-0-172-147:618587:618986 [3] NCCL INFO Channel 24/32 :    0
+ip-26-0-172-147:618587:618986 [3] NCCL INFO Channel 25/32 :    0
+ip-26-0-172-147:618587:618986 [3] NCCL INFO Channel 26/32 :    0
+ip-26-0-172-147:618587:618986 [3] NCCL INFO Channel 27/32 :    0
+ip-26-0-172-147:618587:618986 [3] NCCL INFO Channel 28/32 :    0
+ip-26-0-172-147:618587:618986 [3] NCCL INFO Channel 29/32 :    0
+ip-26-0-172-147:618587:618986 [3] NCCL INFO Channel 30/32 :    0
+ip-26-0-172-147:618587:618986 [3] NCCL INFO Channel 31/32 :    0
+ip-26-0-172-147:618587:618986 [3] NCCL INFO Trees [0] -1/-1/-1->0->-1 [1] -1/-1/-1->0->-1 [2] -1/-1/-1->0->-1 [3] -1/-1/-1->0->-1 [4] -1/-1/-1->0->-1 [5] -1/-1/-1->0->-1 [6] -1/-1/-1->0->-1 [7] -1/-1/-1->0->-1 [8] -1/-1/-1->0->-1 [9] -1/-1/-1->0->-1 [10] -1/-1/-1->0->-1 [11] -1/-1/-1->0->-1 [12] -1/-1/-1->0->-1 [13] -1/-1/-1->0->-1 [14] -1/-1/-1->0->-1 [15] -1/-1/-1->0->-1 [16] -1/-1/-1->0->-1 [17] -1/-1/-1->0->-1 [18] -1/-1/-1->0->-1 [19] -1/-1/-1->0->-1 [20] -1/-1/-1->0->-1 [21] -1/-1/-1->0->-1 [22] -1/-1/-1->0->-1 [23] -1/-1/-1->0->-1 [24] -1/-1/-1->0->-1 [25] -1/-1/-1->0->-1 [26] -1/-1/-1->0->-1 [27] -1/-1/-1->0->-1 [28] -1/-1/-1->0->-1 [29] -1/-1/-1->0->-1 [30] -1/-1/-1->0->-1 [31] -1/-1/-1->0->-1
+ip-26-0-172-147:618587:618986 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-147:618589:618982 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-172-147:618589:618982 [5] NCCL INFO Channel 00/32 :    0
+ip-26-0-172-147:618589:618982 [5] NCCL INFO Channel 01/32 :    0
+ip-26-0-172-147:618589:618982 [5] NCCL INFO Channel 02/32 :    0
+ip-26-0-172-116:1019290:1019691 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-172-147:618589:618982 [5] NCCL INFO Channel 03/32 :    0
+ip-26-0-172-147:618589:618982 [5] NCCL INFO Channel 04/32 :    0
+ip-26-0-172-147:618589:618982 [5] NCCL INFO Channel 05/32 :    0
+ip-26-0-172-147:618589:618982 [5] NCCL INFO Channel 06/32 :    0
+ip-26-0-172-147:618589:618982 [5] NCCL INFO Channel 07/32 :    0
+ip-26-0-172-147:618589:618982 [5] NCCL INFO Channel 08/32 :    0
+ip-26-0-172-147:618589:618982 [5] NCCL INFO Channel 09/32 :    0
+ip-26-0-172-147:618589:618982 [5] NCCL INFO Channel 10/32 :    0
+ip-26-0-172-147:618589:618982 [5] NCCL INFO Channel 11/32 :    0
+ip-26-0-172-147:618589:618982 [5] NCCL INFO Channel 12/32 :    0
+ip-26-0-172-147:618589:618982 [5] NCCL INFO Channel 13/32 :    0
+ip-26-0-172-147:618589:618982 [5] NCCL INFO Channel 14/32 :    0
+ip-26-0-172-147:618589:618982 [5] NCCL INFO Channel 15/32 :    0
+ip-26-0-172-147:618589:618982 [5] NCCL INFO Channel 16/32 :    0
+ip-26-0-172-147:618589:618982 [5] NCCL INFO Channel 17/32 :    0
+ip-26-0-172-147:618589:618982 [5] NCCL INFO Channel 18/32 :    0
+ip-26-0-172-147:618589:618982 [5] NCCL INFO Channel 19/32 :    0
+ip-26-0-172-147:618589:618982 [5] NCCL INFO Channel 20/32 :    0
+ip-26-0-172-147:618589:618982 [5] NCCL INFO Channel 21/32 :    0
+ip-26-0-172-147:618589:618982 [5] NCCL INFO Channel 22/32 :    0
+ip-26-0-172-147:618589:618982 [5] NCCL INFO Channel 23/32 :    0
+ip-26-0-172-147:618589:618982 [5] NCCL INFO Channel 24/32 :    0
+ip-26-0-172-147:618589:618982 [5] NCCL INFO Channel 25/32 :    0
+ip-26-0-172-147:618589:618982 [5] NCCL INFO Channel 26/32 :    0
+ip-26-0-172-147:618589:618982 [5] NCCL INFO Channel 27/32 :    0
+ip-26-0-172-147:618589:618982 [5] NCCL INFO Channel 28/32 :    0
+ip-26-0-172-147:618589:618982 [5] NCCL INFO Channel 29/32 :    0
+ip-26-0-172-147:618589:618982 [5] NCCL INFO Channel 30/32 :    0
+ip-26-0-172-147:618589:618982 [5] NCCL INFO Channel 31/32 :    0
+ip-26-0-172-147:618589:618982 [5] NCCL INFO Trees [0] -1/-1/-1->0->-1 [1] -1/-1/-1->0->-1 [2] -1/-1/-1->0->-1 [3] -1/-1/-1->0->-1 [4] -1/-1/-1->0->-1 [5] -1/-1/-1->0->-1 [6] -1/-1/-1->0->-1 [7] -1/-1/-1->0->-1 [8] -1/-1/-1->0->-1 [9] -1/-1/-1->0->-1 [10] -1/-1/-1->0->-1 [11] -1/-1/-1->0->-1 [12] -1/-1/-1->0->-1 [13] -1/-1/-1->0->-1 [14] -1/-1/-1->0->-1 [15] -1/-1/-1->0->-1 [16] -1/-1/-1->0->-1 [17] -1/-1/-1->0->-1 [18] -1/-1/-1->0->-1 [19] -1/-1/-1->0->-1 [20] -1/-1/-1->0->-1 [21] -1/-1/-1->0->-1 [22] -1/-1/-1->0->-1 [23] -1/-1/-1->0->-1 [24] -1/-1/-1->0->-1 [25] -1/-1/-1->0->-1 [26] -1/-1/-1->0->-1 [27] -1/-1/-1->0->-1 [28] -1/-1/-1->0->-1 [29] -1/-1/-1->0->-1 [30] -1/-1/-1->0->-1 [31] -1/-1/-1->0->-1
+ip-26-0-172-147:618589:618982 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-147:618587:618986 [3] NCCL INFO Connected all rings
+ip-26-0-172-147:618587:618986 [3] NCCL INFO Connected all trees
+ip-26-0-172-147:618587:618986 [3] NCCL INFO 32 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-147:618589:618982 [5] NCCL INFO Connected all rings
+ip-26-0-172-147:618589:618982 [5] NCCL INFO Connected all trees
+ip-26-0-172-147:618589:618982 [5] NCCL INFO 32 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-252:688204:688601 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-172-252:688204:688601 [4] NCCL INFO Channel 00/32 :    0
+ip-26-0-172-252:688204:688601 [4] NCCL INFO Channel 01/32 :    0
+ip-26-0-172-252:688204:688601 [4] NCCL INFO Channel 02/32 :    0
+ip-26-0-172-252:688204:688601 [4] NCCL INFO Channel 03/32 :    0
+ip-26-0-172-252:688204:688601 [4] NCCL INFO Channel 04/32 :    0
+ip-26-0-172-252:688204:688601 [4] NCCL INFO Channel 05/32 :    0
+ip-26-0-172-252:688204:688601 [4] NCCL INFO Channel 06/32 :    0
+ip-26-0-172-252:688204:688601 [4] NCCL INFO Channel 07/32 :    0
+ip-26-0-172-252:688204:688601 [4] NCCL INFO Channel 08/32 :    0
+ip-26-0-172-252:688204:688601 [4] NCCL INFO Channel 09/32 :    0
+ip-26-0-172-252:688204:688601 [4] NCCL INFO Channel 10/32 :    0
+ip-26-0-172-252:688204:688601 [4] NCCL INFO Channel 11/32 :    0
+ip-26-0-172-252:688204:688601 [4] NCCL INFO Channel 12/32 :    0
+ip-26-0-172-252:688204:688601 [4] NCCL INFO Channel 13/32 :    0
+ip-26-0-172-252:688204:688601 [4] NCCL INFO Channel 14/32 :    0
+ip-26-0-172-252:688204:688601 [4] NCCL INFO Channel 15/32 :    0
+ip-26-0-172-252:688204:688601 [4] NCCL INFO Channel 16/32 :    0
+ip-26-0-172-252:688204:688601 [4] NCCL INFO Channel 17/32 :    0
+ip-26-0-172-252:688204:688601 [4] NCCL INFO Channel 18/32 :    0
+ip-26-0-172-252:688204:688601 [4] NCCL INFO Channel 19/32 :    0
+ip-26-0-172-252:688204:688601 [4] NCCL INFO Channel 20/32 :    0
+ip-26-0-172-252:688204:688601 [4] NCCL INFO Channel 21/32 :    0
+ip-26-0-172-252:688204:688601 [4] NCCL INFO Channel 22/32 :    0
+ip-26-0-172-252:688204:688601 [4] NCCL INFO Channel 23/32 :    0
+ip-26-0-172-252:688204:688601 [4] NCCL INFO Channel 24/32 :    0
+ip-26-0-172-252:688204:688601 [4] NCCL INFO Channel 25/32 :    0
+ip-26-0-172-252:688204:688601 [4] NCCL INFO Channel 26/32 :    0
+ip-26-0-172-252:688204:688601 [4] NCCL INFO Channel 27/32 :    0
+ip-26-0-172-252:688204:688601 [4] NCCL INFO Channel 28/32 :    0
+ip-26-0-172-252:688204:688601 [4] NCCL INFO Channel 29/32 :    0
+ip-26-0-172-252:688204:688601 [4] NCCL INFO Channel 30/32 :    0
+ip-26-0-172-252:688204:688601 [4] NCCL INFO Channel 31/32 :    0
+ip-26-0-172-252:688204:688601 [4] NCCL INFO Trees [0] -1/-1/-1->0->-1 [1] -1/-1/-1->0->-1 [2] -1/-1/-1->0->-1 [3] -1/-1/-1->0->-1 [4] -1/-1/-1->0->-1 [5] -1/-1/-1->0->-1 [6] -1/-1/-1->0->-1 [7] -1/-1/-1->0->-1 [8] -1/-1/-1->0->-1 [9] -1/-1/-1->0->-1 [10] -1/-1/-1->0->-1 [11] -1/-1/-1->0->-1 [12] -1/-1/-1->0->-1 [13] -1/-1/-1->0->-1 [14] -1/-1/-1->0->-1 [15] -1/-1/-1->0->-1 [16] -1/-1/-1->0->-1 [17] -1/-1/-1->0->-1 [18] -1/-1/-1->0->-1 [19] -1/-1/-1->0->-1 [20] -1/-1/-1->0->-1 [21] -1/-1/-1->0->-1 [22] -1/-1/-1->0->-1 [23] -1/-1/-1->0->-1 [24] -1/-1/-1->0->-1 [25] -1/-1/-1->0->-1 [26] -1/-1/-1->0->-1 [27] -1/-1/-1->0->-1 [28] -1/-1/-1->0->-1 [29] -1/-1/-1->0->-1 [30] -1/-1/-1->0->-1 [31] -1/-1/-1->0->-1
+ip-26-0-172-252:688204:688601 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-173-7:811254:811665 [6] NCCL INFO comm 0x9697c70 rank 0 nranks 1 cudaDev 6 nvmlDev 6 busId b9000 commId 0x1ea75edfed6c6dc6 - Init COMPLETE
+ip-26-0-172-252:688204:688601 [4] NCCL INFO Connected all rings
+ip-26-0-172-252:688204:688601 [4] NCCL INFO Connected all trees
+ip-26-0-172-252:688204:688601 [4] NCCL INFO 32 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-116:1019290:1019691 [2] NCCL INFO Channel 00/32 :    0
+ip-26-0-172-116:1019290:1019691 [2] NCCL INFO Channel 01/32 :    0
+ip-26-0-172-116:1019290:1019691 [2] NCCL INFO Channel 02/32 :    0
+ip-26-0-172-116:1019290:1019691 [2] NCCL INFO Channel 03/32 :    0
+ip-26-0-172-116:1019290:1019691 [2] NCCL INFO Channel 04/32 :    0
+ip-26-0-172-116:1019290:1019691 [2] NCCL INFO Channel 05/32 :    0
+ip-26-0-172-116:1019290:1019691 [2] NCCL INFO Channel 06/32 :    0
+ip-26-0-172-116:1019290:1019691 [2] NCCL INFO Channel 07/32 :    0
+ip-26-0-172-116:1019290:1019691 [2] NCCL INFO Channel 08/32 :    0
+ip-26-0-172-116:1019290:1019691 [2] NCCL INFO Channel 09/32 :    0
+ip-26-0-172-116:1019290:1019691 [2] NCCL INFO Channel 10/32 :    0
+ip-26-0-172-116:1019290:1019691 [2] NCCL INFO Channel 11/32 :    0
+ip-26-0-172-116:1019290:1019691 [2] NCCL INFO Channel 12/32 :    0
+ip-26-0-172-116:1019290:1019691 [2] NCCL INFO Channel 13/32 :    0
+ip-26-0-172-116:1019290:1019691 [2] NCCL INFO Channel 14/32 :    0
+ip-26-0-172-116:1019290:1019691 [2] NCCL INFO Channel 15/32 :    0
+ip-26-0-172-116:1019290:1019691 [2] NCCL INFO Channel 16/32 :    0
+ip-26-0-172-116:1019290:1019691 [2] NCCL INFO Channel 17/32 :    0
+ip-26-0-172-116:1019290:1019691 [2] NCCL INFO Channel 18/32 :    0
+ip-26-0-172-116:1019290:1019691 [2] NCCL INFO Channel 19/32 :    0
+ip-26-0-172-116:1019290:1019691 [2] NCCL INFO Channel 20/32 :    0
+ip-26-0-172-116:1019290:1019691 [2] NCCL INFO Channel 21/32 :    0
+ip-26-0-172-116:1019290:1019691 [2] NCCL INFO Channel 22/32 :    0
+ip-26-0-172-116:1019290:1019691 [2] NCCL INFO Channel 23/32 :    0
+ip-26-0-172-116:1019290:1019691 [2] NCCL INFO Channel 24/32 :    0
+ip-26-0-172-116:1019290:1019691 [2] NCCL INFO Channel 25/32 :    0
+ip-26-0-172-116:1019290:1019691 [2] NCCL INFO Channel 26/32 :    0
+ip-26-0-172-116:1019290:1019691 [2] NCCL INFO Channel 27/32 :    0
+ip-26-0-172-116:1019290:1019691 [2] NCCL INFO Channel 28/32 :    0
+ip-26-0-172-116:1019290:1019691 [2] NCCL INFO Channel 29/32 :    0
+ip-26-0-172-116:1019290:1019691 [2] NCCL INFO Channel 30/32 :    0
+ip-26-0-172-116:1019290:1019691 [2] NCCL INFO Channel 31/32 :    0
+ip-26-0-172-116:1019290:1019691 [2] NCCL INFO Trees [0] -1/-1/-1->0->-1 [1] -1/-1/-1->0->-1 [2] -1/-1/-1->0->-1 [3] -1/-1/-1->0->-1 [4] -1/-1/-1->0->-1 [5] -1/-1/-1->0->-1 [6] -1/-1/-1->0->-1 [7] -1/-1/-1->0->-1 [8] -1/-1/-1->0->-1 [9] -1/-1/-1->0->-1 [10] -1/-1/-1->0->-1 [11] -1/-1/-1->0->-1 [12] -1/-1/-1->0->-1 [13] -1/-1/-1->0->-1 [14] -1/-1/-1->0->-1 [15] -1/-1/-1->0->-1 [16] -1/-1/-1->0->-1 [17] -1/-1/-1->0->-1 [18] -1/-1/-1->0->-1 [19] -1/-1/-1->0->-1 [20] -1/-1/-1->0->-1 [21] -1/-1/-1->0->-1 [22] -1/-1/-1->0->-1 [23] -1/-1/-1->0->-1 [24] -1/-1/-1->0->-1 [25] -1/-1/-1->0->-1 [26] -1/-1/-1->0->-1 [27] -1/-1/-1->0->-1 [28] -1/-1/-1->0->-1 [29] -1/-1/-1->0->-1 [30] -1/-1/-1->0->-1 [31] -1/-1/-1->0->-1
+ip-26-0-172-116:1019290:1019691 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-116:1019290:1019691 [2] NCCL INFO Connected all rings
+ip-26-0-172-116:1019290:1019691 [2] NCCL INFO Connected all trees
+ip-26-0-172-116:1019290:1019691 [2] NCCL INFO 32 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-73:926594:926994 [0] NCCL INFO comm 0xb37c500 rank 0 nranks 1 cudaDev 0 nvmlDev 0 busId 53000 commId 0xf65ea4046937c873 - Init COMPLETE
+ip-26-0-172-116:1019295:1019688 [7] NCCL INFO comm 0xb9681c0 rank 0 nranks 1 cudaDev 7 nvmlDev 7 busId ca000 commId 0x1a310b0d9fdd31f4 - Init COMPLETE
+ip-26-0-172-73:926601:926987 [7] NCCL INFO comm 0xa54a640 rank 0 nranks 1 cudaDev 7 nvmlDev 7 busId ca000 commId 0xf3340c6bc170e5e0 - Init COMPLETE
+ip-26-0-172-116:1019294:1019686 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-172-116:1019294:1019686 [6] NCCL INFO Channel 00/32 :    0
+ip-26-0-172-116:1019294:1019686 [6] NCCL INFO Channel 01/32 :    0
+ip-26-0-172-116:1019294:1019686 [6] NCCL INFO Channel 02/32 :    0
+ip-26-0-172-116:1019294:1019686 [6] NCCL INFO Channel 03/32 :    0
+ip-26-0-172-116:1019294:1019686 [6] NCCL INFO Channel 04/32 :    0
+ip-26-0-172-116:1019294:1019686 [6] NCCL INFO Channel 05/32 :    0
+ip-26-0-172-116:1019294:1019686 [6] NCCL INFO Channel 06/32 :    0
+ip-26-0-172-116:1019294:1019686 [6] NCCL INFO Channel 07/32 :    0
+ip-26-0-172-116:1019294:1019686 [6] NCCL INFO Channel 08/32 :    0
+ip-26-0-172-116:1019294:1019686 [6] NCCL INFO Channel 09/32 :    0
+ip-26-0-172-116:1019294:1019686 [6] NCCL INFO Channel 10/32 :    0
+ip-26-0-172-116:1019294:1019686 [6] NCCL INFO Channel 11/32 :    0
+ip-26-0-172-116:1019294:1019686 [6] NCCL INFO Channel 12/32 :    0
+ip-26-0-172-116:1019294:1019686 [6] NCCL INFO Channel 13/32 :    0
+ip-26-0-172-116:1019294:1019686 [6] NCCL INFO Channel 14/32 :    0
+ip-26-0-172-116:1019294:1019686 [6] NCCL INFO Channel 15/32 :    0
+ip-26-0-172-116:1019294:1019686 [6] NCCL INFO Channel 16/32 :    0
+ip-26-0-172-116:1019294:1019686 [6] NCCL INFO Channel 17/32 :    0
+ip-26-0-172-116:1019294:1019686 [6] NCCL INFO Channel 18/32 :    0
+ip-26-0-172-116:1019294:1019686 [6] NCCL INFO Channel 19/32 :    0
+ip-26-0-172-116:1019294:1019686 [6] NCCL INFO Channel 20/32 :    0
+ip-26-0-172-116:1019294:1019686 [6] NCCL INFO Channel 21/32 :    0
+ip-26-0-172-116:1019294:1019686 [6] NCCL INFO Channel 22/32 :    0
+ip-26-0-172-116:1019294:1019686 [6] NCCL INFO Channel 23/32 :    0
+ip-26-0-172-116:1019294:1019686 [6] NCCL INFO Channel 24/32 :    0
+ip-26-0-172-116:1019294:1019686 [6] NCCL INFO Channel 25/32 :    0
+ip-26-0-172-116:1019294:1019686 [6] NCCL INFO Channel 26/32 :    0
+ip-26-0-172-116:1019294:1019686 [6] NCCL INFO Channel 27/32 :    0
+ip-26-0-172-116:1019294:1019686 [6] NCCL INFO Channel 28/32 :    0
+ip-26-0-172-116:1019294:1019686 [6] NCCL INFO Channel 29/32 :    0
+ip-26-0-172-116:1019294:1019686 [6] NCCL INFO Channel 30/32 :    0
+ip-26-0-172-116:1019294:1019686 [6] NCCL INFO Channel 31/32 :    0
+ip-26-0-172-116:1019294:1019686 [6] NCCL INFO Trees [0] -1/-1/-1->0->-1 [1] -1/-1/-1->0->-1 [2] -1/-1/-1->0->-1 [3] -1/-1/-1->0->-1 [4] -1/-1/-1->0->-1 [5] -1/-1/-1->0->-1 [6] -1/-1/-1->0->-1 [7] -1/-1/-1->0->-1 [8] -1/-1/-1->0->-1 [9] -1/-1/-1->0->-1 [10] -1/-1/-1->0->-1 [11] -1/-1/-1->0->-1 [12] -1/-1/-1->0->-1 [13] -1/-1/-1->0->-1 [14] -1/-1/-1->0->-1 [15] -1/-1/-1->0->-1 [16] -1/-1/-1->0->-1 [17] -1/-1/-1->0->-1 [18] -1/-1/-1->0->-1 [19] -1/-1/-1->0->-1 [20] -1/-1/-1->0->-1 [21] -1/-1/-1->0->-1 [22] -1/-1/-1->0->-1 [23] -1/-1/-1->0->-1 [24] -1/-1/-1->0->-1 [25] -1/-1/-1->0->-1 [26] -1/-1/-1->0->-1 [27] -1/-1/-1->0->-1 [28] -1/-1/-1->0->-1 [29] -1/-1/-1->0->-1 [30] -1/-1/-1->0->-1 [31] -1/-1/-1->0->-1
+ip-26-0-172-116:1019294:1019686 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-116:1019294:1019686 [6] NCCL INFO Connected all rings
+ip-26-0-172-116:1019294:1019686 [6] NCCL INFO Connected all trees
+ip-26-0-172-116:1019294:1019686 [6] NCCL INFO 32 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-116:1019289:1019690 [1] NCCL INFO comm 0x9b46270 rank 0 nranks 1 cudaDev 1 nvmlDev 1 busId 64000 commId 0xf7413587374e582c - Init COMPLETE
+ip-26-0-172-73:926595:926986 [1] NCCL INFO comm 0xa4702b0 rank 0 nranks 1 cudaDev 1 nvmlDev 1 busId 64000 commId 0xefd6094e8078b1b5 - Init COMPLETE
+ip-26-0-172-73:926598:926997 [4] NCCL INFO comm 0xae614e0 rank 0 nranks 1 cudaDev 4 nvmlDev 4 busId 97000 commId 0xc353ea69253ec4a4 - Init COMPLETE
+ip-26-0-160-225:94349:94776 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-172-147:618591:618983 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-172-147:618591:618983 [7] NCCL INFO Channel 00/32 :    0
+ip-26-0-172-147:618591:618983 [7] NCCL INFO Channel 01/32 :    0
+ip-26-0-172-147:618591:618983 [7] NCCL INFO Channel 02/32 :    0
+ip-26-0-172-147:618591:618983 [7] NCCL INFO Channel 03/32 :    0
+ip-26-0-172-147:618591:618983 [7] NCCL INFO Channel 04/32 :    0
+ip-26-0-172-147:618591:618983 [7] NCCL INFO Channel 05/32 :    0
+ip-26-0-172-147:618591:618983 [7] NCCL INFO Channel 06/32 :    0
+ip-26-0-172-147:618591:618983 [7] NCCL INFO Channel 07/32 :    0
+ip-26-0-172-147:618591:618983 [7] NCCL INFO Channel 08/32 :    0
+ip-26-0-172-147:618591:618983 [7] NCCL INFO Channel 09/32 :    0
+ip-26-0-172-147:618591:618983 [7] NCCL INFO Channel 10/32 :    0
+ip-26-0-172-147:618591:618983 [7] NCCL INFO Channel 11/32 :    0
+ip-26-0-172-147:618591:618983 [7] NCCL INFO Channel 12/32 :    0
+ip-26-0-172-147:618591:618983 [7] NCCL INFO Channel 13/32 :    0
+ip-26-0-172-147:618591:618983 [7] NCCL INFO Channel 14/32 :    0
+ip-26-0-172-147:618591:618983 [7] NCCL INFO Channel 15/32 :    0
+ip-26-0-172-147:618591:618983 [7] NCCL INFO Channel 16/32 :    0
+ip-26-0-172-147:618591:618983 [7] NCCL INFO Channel 17/32 :    0
+ip-26-0-172-147:618591:618983 [7] NCCL INFO Channel 18/32 :    0
+ip-26-0-172-147:618591:618983 [7] NCCL INFO Channel 19/32 :    0
+ip-26-0-172-147:618591:618983 [7] NCCL INFO Channel 20/32 :    0
+ip-26-0-172-147:618591:618983 [7] NCCL INFO Channel 21/32 :    0
+ip-26-0-172-147:618591:618983 [7] NCCL INFO Channel 22/32 :    0
+ip-26-0-172-147:618591:618983 [7] NCCL INFO Channel 23/32 :    0
+ip-26-0-172-147:618591:618983 [7] NCCL INFO Channel 24/32 :    0
+ip-26-0-172-147:618591:618983 [7] NCCL INFO Channel 25/32 :    0
+ip-26-0-172-147:618591:618983 [7] NCCL INFO Channel 26/32 :    0
+ip-26-0-172-147:618591:618983 [7] NCCL INFO Channel 27/32 :    0
+ip-26-0-172-147:618591:618983 [7] NCCL INFO Channel 28/32 :    0
+ip-26-0-172-147:618591:618983 [7] NCCL INFO Channel 29/32 :    0
+ip-26-0-172-147:618591:618983 [7] NCCL INFO Channel 30/32 :    0
+ip-26-0-172-147:618591:618983 [7] NCCL INFO Channel 31/32 :    0
+ip-26-0-160-225:94349:94776 [7] NCCL INFO Channel 00/32 :    0
+ip-26-0-160-225:94349:94776 [7] NCCL INFO Channel 01/32 :    0
+ip-26-0-160-225:94349:94776 [7] NCCL INFO Channel 02/32 :    0
+ip-26-0-160-225:94349:94776 [7] NCCL INFO Channel 03/32 :    0
+ip-26-0-160-225:94349:94776 [7] NCCL INFO Channel 04/32 :    0
+ip-26-0-160-225:94349:94776 [7] NCCL INFO Channel 05/32 :    0
+ip-26-0-160-225:94349:94776 [7] NCCL INFO Channel 06/32 :    0
+ip-26-0-160-225:94349:94776 [7] NCCL INFO Channel 07/32 :    0
+ip-26-0-160-225:94349:94776 [7] NCCL INFO Channel 08/32 :    0
+ip-26-0-160-225:94349:94776 [7] NCCL INFO Channel 09/32 :    0
+ip-26-0-160-225:94349:94776 [7] NCCL INFO Channel 10/32 :    0
+ip-26-0-172-147:618591:618983 [7] NCCL INFO Trees [0] -1/-1/-1->0->-1 [1] -1/-1/-1->0->-1 [2] -1/-1/-1->0->-1 [3] -1/-1/-1->0->-1 [4] -1/-1/-1->0->-1 [5] -1/-1/-1->0->-1 [6] -1/-1/-1->0->-1 [7] -1/-1/-1->0->-1 [8] -1/-1/-1->0->-1 [9] -1/-1/-1->0->-1 [10] -1/-1/-1->0->-1 [11] -1/-1/-1->0->-1 [12] -1/-1/-1->0->-1 [13] -1/-1/-1->0->-1 [14] -1/-1/-1->0->-1 [15] -1/-1/-1->0->-1 [16] -1/-1/-1->0->-1 [17] -1/-1/-1->0->-1 [18] -1/-1/-1->0->-1 [19] -1/-1/-1->0->-1 [20] -1/-1/-1->0->-1 [21] -1/-1/-1->0->-1 [22] -1/-1/-1->0->-1 [23] -1/-1/-1->0->-1 [24] -1/-1/-1->0->-1 [25] -1/-1/-1->0->-1 [26] -1/-1/-1->0->-1 [27] -1/-1/-1->0->-1 [28] -1/-1/-1->0->-1 [29] -1/-1/-1->0->-1 [30] -1/-1/-1->0->-1 [31] -1/-1/-1->0->-1
+ip-26-0-172-147:618591:618983 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-225:94349:94776 [7] NCCL INFO Channel 11/32 :    0
+ip-26-0-160-225:94349:94776 [7] NCCL INFO Channel 12/32 :    0
+ip-26-0-160-225:94349:94776 [7] NCCL INFO Channel 13/32 :    0
+ip-26-0-160-225:94349:94776 [7] NCCL INFO Channel 14/32 :    0
+ip-26-0-160-225:94349:94776 [7] NCCL INFO Channel 15/32 :    0
+ip-26-0-160-225:94349:94776 [7] NCCL INFO Channel 16/32 :    0
+ip-26-0-160-225:94349:94776 [7] NCCL INFO Channel 17/32 :    0
+ip-26-0-160-225:94349:94776 [7] NCCL INFO Channel 18/32 :    0
+ip-26-0-160-225:94349:94776 [7] NCCL INFO Channel 19/32 :    0
+ip-26-0-160-225:94349:94776 [7] NCCL INFO Channel 20/32 :    0
+ip-26-0-160-225:94349:94776 [7] NCCL INFO Channel 21/32 :    0
+ip-26-0-160-225:94349:94776 [7] NCCL INFO Channel 22/32 :    0
+ip-26-0-160-225:94349:94776 [7] NCCL INFO Channel 23/32 :    0
+ip-26-0-160-225:94349:94776 [7] NCCL INFO Channel 24/32 :    0
+ip-26-0-160-225:94349:94776 [7] NCCL INFO Channel 25/32 :    0
+ip-26-0-160-225:94349:94776 [7] NCCL INFO Channel 26/32 :    0
+ip-26-0-160-225:94349:94776 [7] NCCL INFO Channel 27/32 :    0
+ip-26-0-160-225:94349:94776 [7] NCCL INFO Channel 28/32 :    0
+ip-26-0-160-225:94349:94776 [7] NCCL INFO Channel 29/32 :    0
+ip-26-0-160-225:94349:94776 [7] NCCL INFO Channel 30/32 :    0
+ip-26-0-160-225:94349:94776 [7] NCCL INFO Channel 31/32 :    0
+ip-26-0-160-225:94349:94776 [7] NCCL INFO Trees [0] -1/-1/-1->0->-1 [1] -1/-1/-1->0->-1 [2] -1/-1/-1->0->-1 [3] -1/-1/-1->0->-1 [4] -1/-1/-1->0->-1 [5] -1/-1/-1->0->-1 [6] -1/-1/-1->0->-1 [7] -1/-1/-1->0->-1 [8] -1/-1/-1->0->-1 [9] -1/-1/-1->0->-1 [10] -1/-1/-1->0->-1 [11] -1/-1/-1->0->-1 [12] -1/-1/-1->0->-1 [13] -1/-1/-1->0->-1 [14] -1/-1/-1->0->-1 [15] -1/-1/-1->0->-1 [16] -1/-1/-1->0->-1 [17] -1/-1/-1->0->-1 [18] -1/-1/-1->0->-1 [19] -1/-1/-1->0->-1 [20] -1/-1/-1->0->-1 [21] -1/-1/-1->0->-1 [22] -1/-1/-1->0->-1 [23] -1/-1/-1->0->-1 [24] -1/-1/-1->0->-1 [25] -1/-1/-1->0->-1 [26] -1/-1/-1->0->-1 [27] -1/-1/-1->0->-1 [28] -1/-1/-1->0->-1 [29] -1/-1/-1->0->-1 [30] -1/-1/-1->0->-1 [31] -1/-1/-1->0->-1
+ip-26-0-160-225:94349:94776 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-116:1019292:1019679 [4] NCCL INFO comm 0xc2fe140 rank 0 nranks 1 cudaDev 4 nvmlDev 4 busId 97000 commId 0x665eadf333a30be4 - Init COMPLETE
+ip-26-0-172-147:618591:618983 [7] NCCL INFO Connected all rings
+ip-26-0-172-147:618591:618983 [7] NCCL INFO Connected all trees
+ip-26-0-172-147:618591:618983 [7] NCCL INFO 32 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-116:1019293:1019681 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-172-116:1019293:1019681 [5] NCCL INFO Channel 00/32 :    0
+ip-26-0-172-116:1019293:1019681 [5] NCCL INFO Channel 01/32 :    0
+ip-26-0-172-116:1019293:1019681 [5] NCCL INFO Channel 02/32 :    0
+ip-26-0-172-116:1019293:1019681 [5] NCCL INFO Channel 03/32 :    0
+ip-26-0-172-116:1019293:1019681 [5] NCCL INFO Channel 04/32 :    0
+ip-26-0-172-116:1019293:1019681 [5] NCCL INFO Channel 05/32 :    0
+ip-26-0-172-116:1019293:1019681 [5] NCCL INFO Channel 06/32 :    0
+ip-26-0-172-116:1019293:1019681 [5] NCCL INFO Channel 07/32 :    0
+ip-26-0-172-116:1019293:1019681 [5] NCCL INFO Channel 08/32 :    0
+ip-26-0-172-116:1019293:1019681 [5] NCCL INFO Channel 09/32 :    0
+ip-26-0-172-116:1019293:1019681 [5] NCCL INFO Channel 10/32 :    0
+ip-26-0-172-116:1019293:1019681 [5] NCCL INFO Channel 11/32 :    0
+ip-26-0-172-116:1019293:1019681 [5] NCCL INFO Channel 12/32 :    0
+ip-26-0-172-116:1019293:1019681 [5] NCCL INFO Channel 13/32 :    0
+ip-26-0-172-116:1019293:1019681 [5] NCCL INFO Channel 14/32 :    0
+ip-26-0-172-116:1019293:1019681 [5] NCCL INFO Channel 15/32 :    0
+ip-26-0-172-116:1019293:1019681 [5] NCCL INFO Channel 16/32 :    0
+ip-26-0-172-116:1019293:1019681 [5] NCCL INFO Channel 17/32 :    0
+ip-26-0-172-116:1019293:1019681 [5] NCCL INFO Channel 18/32 :    0
+ip-26-0-172-116:1019293:1019681 [5] NCCL INFO Channel 19/32 :    0
+ip-26-0-172-116:1019293:1019681 [5] NCCL INFO Channel 20/32 :    0
+ip-26-0-172-116:1019293:1019681 [5] NCCL INFO Channel 21/32 :    0
+ip-26-0-172-116:1019293:1019681 [5] NCCL INFO Channel 22/32 :    0
+ip-26-0-172-116:1019293:1019681 [5] NCCL INFO Channel 23/32 :    0
+ip-26-0-172-116:1019293:1019681 [5] NCCL INFO Channel 24/32 :    0
+ip-26-0-172-116:1019293:1019681 [5] NCCL INFO Channel 25/32 :    0
+ip-26-0-172-116:1019293:1019681 [5] NCCL INFO Channel 26/32 :    0
+ip-26-0-160-225:94349:94776 [7] NCCL INFO Connected all rings
+ip-26-0-160-225:94349:94776 [7] NCCL INFO Connected all trees
+ip-26-0-172-116:1019293:1019681 [5] NCCL INFO Channel 27/32 :    0
+ip-26-0-172-116:1019293:1019681 [5] NCCL INFO Channel 28/32 :    0
+ip-26-0-172-116:1019293:1019681 [5] NCCL INFO Channel 29/32 :    0
+ip-26-0-172-116:1019293:1019681 [5] NCCL INFO Channel 30/32 :    0
+ip-26-0-172-116:1019293:1019681 [5] NCCL INFO Channel 31/32 :    0
+ip-26-0-160-225:94349:94776 [7] NCCL INFO 32 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-116:1019293:1019681 [5] NCCL INFO Trees [0] -1/-1/-1->0->-1 [1] -1/-1/-1->0->-1 [2] -1/-1/-1->0->-1 [3] -1/-1/-1->0->-1 [4] -1/-1/-1->0->-1 [5] -1/-1/-1->0->-1 [6] -1/-1/-1->0->-1 [7] -1/-1/-1->0->-1 [8] -1/-1/-1->0->-1 [9] -1/-1/-1->0->-1 [10] -1/-1/-1->0->-1 [11] -1/-1/-1->0->-1 [12] -1/-1/-1->0->-1 [13] -1/-1/-1->0->-1 [14] -1/-1/-1->0->-1 [15] -1/-1/-1->0->-1 [16] -1/-1/-1->0->-1 [17] -1/-1/-1->0->-1 [18] -1/-1/-1->0->-1 [19] -1/-1/-1->0->-1 [20] -1/-1/-1->0->-1 [21] -1/-1/-1->0->-1 [22] -1/-1/-1->0->-1 [23] -1/-1/-1->0->-1 [24] -1/-1/-1->0->-1 [25] -1/-1/-1->0->-1 [26] -1/-1/-1->0->-1 [27] -1/-1/-1->0->-1 [28] -1/-1/-1->0->-1 [29] -1/-1/-1->0->-1 [30] -1/-1/-1->0->-1 [31] -1/-1/-1->0->-1
+ip-26-0-172-116:1019293:1019681 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-116:1019293:1019681 [5] NCCL INFO Connected all rings
+ip-26-0-172-116:1019293:1019681 [5] NCCL INFO Connected all trees
+ip-26-0-172-116:1019293:1019681 [5] NCCL INFO 32 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-142:880679:881080 [0] NCCL INFO comm 0xc15d520 rank 0 nranks 1 cudaDev 0 nvmlDev 0 busId 53000 commId 0xa7e021c6170379b7 - Init COMPLETE
+ip-26-0-173-7:811251:811671 [3] NCCL INFO comm 0x988f950 rank 0 nranks 1 cudaDev 3 nvmlDev 3 busId 86000 commId 0x14c32b7a3c51460a - Init COMPLETE
+ip-26-0-172-142:880682:881070 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-172-73:926597:926992 [3] NCCL INFO comm 0xac4c0d0 rank 0 nranks 1 cudaDev 3 nvmlDev 3 busId 86000 commId 0xa38bdf5b995818a7 - Init COMPLETE
+ip-26-0-172-142:880682:881070 [3] NCCL INFO Channel 00/32 :    0
+ip-26-0-172-142:880682:881070 [3] NCCL INFO Channel 01/32 :    0
+ip-26-0-172-142:880682:881070 [3] NCCL INFO Channel 02/32 :    0
+ip-26-0-172-142:880682:881070 [3] NCCL INFO Channel 03/32 :    0
+ip-26-0-172-142:880682:881070 [3] NCCL INFO Channel 04/32 :    0
+ip-26-0-172-142:880682:881070 [3] NCCL INFO Channel 05/32 :    0
+ip-26-0-172-142:880682:881070 [3] NCCL INFO Channel 06/32 :    0
+ip-26-0-172-142:880682:881070 [3] NCCL INFO Channel 07/32 :    0
+ip-26-0-172-142:880682:881070 [3] NCCL INFO Channel 08/32 :    0
+ip-26-0-172-142:880682:881070 [3] NCCL INFO Channel 09/32 :    0
+ip-26-0-172-142:880682:881070 [3] NCCL INFO Channel 10/32 :    0
+ip-26-0-172-142:880682:881070 [3] NCCL INFO Channel 11/32 :    0
+ip-26-0-172-142:880682:881070 [3] NCCL INFO Channel 12/32 :    0
+ip-26-0-172-142:880682:881070 [3] NCCL INFO Channel 13/32 :    0
+ip-26-0-172-142:880682:881070 [3] NCCL INFO Channel 14/32 :    0
+ip-26-0-172-142:880682:881070 [3] NCCL INFO Channel 15/32 :    0
+ip-26-0-172-142:880682:881070 [3] NCCL INFO Channel 16/32 :    0
+ip-26-0-172-142:880682:881070 [3] NCCL INFO Channel 17/32 :    0
+ip-26-0-172-142:880682:881070 [3] NCCL INFO Channel 18/32 :    0
+ip-26-0-172-142:880682:881070 [3] NCCL INFO Channel 19/32 :    0
+ip-26-0-172-142:880682:881070 [3] NCCL INFO Channel 20/32 :    0
+ip-26-0-172-142:880682:881070 [3] NCCL INFO Channel 21/32 :    0
+ip-26-0-172-142:880682:881070 [3] NCCL INFO Channel 22/32 :    0
+ip-26-0-172-142:880682:881070 [3] NCCL INFO Channel 23/32 :    0
+ip-26-0-172-142:880682:881070 [3] NCCL INFO Channel 24/32 :    0
+ip-26-0-172-142:880682:881070 [3] NCCL INFO Channel 25/32 :    0
+ip-26-0-172-142:880682:881070 [3] NCCL INFO Channel 26/32 :    0
+ip-26-0-172-142:880682:881070 [3] NCCL INFO Channel 27/32 :    0
+ip-26-0-172-142:880682:881070 [3] NCCL INFO Channel 28/32 :    0
+ip-26-0-172-142:880682:881070 [3] NCCL INFO Channel 29/32 :    0
+ip-26-0-172-142:880682:881070 [3] NCCL INFO Channel 30/32 :    0
+ip-26-0-172-142:880682:881070 [3] NCCL INFO Channel 31/32 :    0
+ip-26-0-172-142:880682:881070 [3] NCCL INFO Trees [0] -1/-1/-1->0->-1 [1] -1/-1/-1->0->-1 [2] -1/-1/-1->0->-1 [3] -1/-1/-1->0->-1 [4] -1/-1/-1->0->-1 [5] -1/-1/-1->0->-1 [6] -1/-1/-1->0->-1 [7] -1/-1/-1->0->-1 [8] -1/-1/-1->0->-1 [9] -1/-1/-1->0->-1 [10] -1/-1/-1->0->-1 [11] -1/-1/-1->0->-1 [12] -1/-1/-1->0->-1 [13] -1/-1/-1->0->-1 [14] -1/-1/-1->0->-1 [15] -1/-1/-1->0->-1 [16] -1/-1/-1->0->-1 [17] -1/-1/-1->0->-1 [18] -1/-1/-1->0->-1 [19] -1/-1/-1->0->-1 [20] -1/-1/-1->0->-1 [21] -1/-1/-1->0->-1 [22] -1/-1/-1->0->-1 [23] -1/-1/-1->0->-1 [24] -1/-1/-1->0->-1 [25] -1/-1/-1->0->-1 [26] -1/-1/-1->0->-1 [27] -1/-1/-1->0->-1 [28] -1/-1/-1->0->-1 [29] -1/-1/-1->0->-1 [30] -1/-1/-1->0->-1 [31] -1/-1/-1->0->-1
+ip-26-0-172-142:880682:881070 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-173-7:811249:811675 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-173-7:811249:811675 [1] NCCL INFO Channel 00/32 :    0
+ip-26-0-173-7:811249:811675 [1] NCCL INFO Channel 01/32 :    0
+ip-26-0-173-7:811249:811675 [1] NCCL INFO Channel 02/32 :    0
+ip-26-0-173-7:811249:811675 [1] NCCL INFO Channel 03/32 :    0
+ip-26-0-173-7:811249:811675 [1] NCCL INFO Channel 04/32 :    0
+ip-26-0-173-7:811249:811675 [1] NCCL INFO Channel 05/32 :    0
+ip-26-0-173-7:811249:811675 [1] NCCL INFO Channel 06/32 :    0
+ip-26-0-173-7:811249:811675 [1] NCCL INFO Channel 07/32 :    0
+ip-26-0-173-7:811249:811675 [1] NCCL INFO Channel 08/32 :    0
+ip-26-0-173-7:811249:811675 [1] NCCL INFO Channel 09/32 :    0
+ip-26-0-173-7:811249:811675 [1] NCCL INFO Channel 10/32 :    0
+ip-26-0-173-7:811249:811675 [1] NCCL INFO Channel 11/32 :    0
+ip-26-0-173-7:811249:811675 [1] NCCL INFO Channel 12/32 :    0
+ip-26-0-173-7:811249:811675 [1] NCCL INFO Channel 13/32 :    0
+ip-26-0-173-7:811249:811675 [1] NCCL INFO Channel 14/32 :    0
+ip-26-0-173-7:811249:811675 [1] NCCL INFO Channel 15/32 :    0
+ip-26-0-173-7:811249:811675 [1] NCCL INFO Channel 16/32 :    0
+ip-26-0-173-7:811249:811675 [1] NCCL INFO Channel 17/32 :    0
+ip-26-0-173-7:811249:811675 [1] NCCL INFO Channel 18/32 :    0
+ip-26-0-173-7:811249:811675 [1] NCCL INFO Channel 19/32 :    0
+ip-26-0-173-7:811249:811675 [1] NCCL INFO Channel 20/32 :    0
+ip-26-0-173-7:811249:811675 [1] NCCL INFO Channel 21/32 :    0
+ip-26-0-173-7:811249:811675 [1] NCCL INFO Channel 22/32 :    0
+ip-26-0-173-7:811249:811675 [1] NCCL INFO Channel 23/32 :    0
+ip-26-0-173-7:811249:811675 [1] NCCL INFO Channel 24/32 :    0
+ip-26-0-173-7:811249:811675 [1] NCCL INFO Channel 25/32 :    0
+ip-26-0-173-7:811249:811675 [1] NCCL INFO Channel 26/32 :    0
+ip-26-0-173-7:811249:811675 [1] NCCL INFO Channel 27/32 :    0
+ip-26-0-173-7:811249:811675 [1] NCCL INFO Channel 28/32 :    0
+ip-26-0-173-7:811249:811675 [1] NCCL INFO Channel 29/32 :    0
+ip-26-0-173-7:811249:811675 [1] NCCL INFO Channel 30/32 :    0
+ip-26-0-173-7:811249:811675 [1] NCCL INFO Channel 31/32 :    0
+ip-26-0-173-7:811249:811675 [1] NCCL INFO Trees [0] -1/-1/-1->0->-1 [1] -1/-1/-1->0->-1 [2] -1/-1/-1->0->-1 [3] -1/-1/-1->0->-1 [4] -1/-1/-1->0->-1 [5] -1/-1/-1->0->-1 [6] -1/-1/-1->0->-1 [7] -1/-1/-1->0->-1 [8] -1/-1/-1->0->-1 [9] -1/-1/-1->0->-1 [10] -1/-1/-1->0->-1 [11] -1/-1/-1->0->-1 [12] -1/-1/-1->0->-1 [13] -1/-1/-1->0->-1 [14] -1/-1/-1->0->-1 [15] -1/-1/-1->0->-1 [16] -1/-1/-1->0->-1 [17] -1/-1/-1->0->-1 [18] -1/-1/-1->0->-1 [19] -1/-1/-1->0->-1 [20] -1/-1/-1->0->-1 [21] -1/-1/-1->0->-1 [22] -1/-1/-1->0->-1 [23] -1/-1/-1->0->-1 [24] -1/-1/-1->0->-1 [25] -1/-1/-1->0->-1 [26] -1/-1/-1->0->-1 [27] -1/-1/-1->0->-1 [28] -1/-1/-1->0->-1 [29] -1/-1/-1->0->-1 [30] -1/-1/-1->0->-1 [31] -1/-1/-1->0->-1
+ip-26-0-173-7:811249:811675 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-142:880682:881070 [3] NCCL INFO Connected all rings
+ip-26-0-172-142:880682:881070 [3] NCCL INFO Connected all trees
+ip-26-0-172-142:880682:881070 [3] NCCL INFO 32 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-116:1019291:1019684 [3] NCCL INFO comm 0xb8bc290 rank 0 nranks 1 cudaDev 3 nvmlDev 3 busId 86000 commId 0x86ea12526dc12129 - Init COMPLETE
+ip-26-0-173-7:811253:811670 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-173-7:811249:811675 [1] NCCL INFO Connected all rings
+ip-26-0-173-7:811249:811675 [1] NCCL INFO Connected all trees
+ip-26-0-173-7:811249:811675 [1] NCCL INFO 32 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-142:880685:881067 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-173-7:811253:811670 [5] NCCL INFO Channel 00/32 :    0
+ip-26-0-173-7:811253:811670 [5] NCCL INFO Channel 01/32 :    0
+ip-26-0-173-7:811253:811670 [5] NCCL INFO Channel 02/32 :    0
+ip-26-0-173-7:811253:811670 [5] NCCL INFO Channel 03/32 :    0
+ip-26-0-173-7:811253:811670 [5] NCCL INFO Channel 04/32 :    0
+ip-26-0-173-7:811253:811670 [5] NCCL INFO Channel 05/32 :    0
+ip-26-0-173-7:811253:811670 [5] NCCL INFO Channel 06/32 :    0
+ip-26-0-173-7:811253:811670 [5] NCCL INFO Channel 07/32 :    0
+ip-26-0-173-7:811253:811670 [5] NCCL INFO Channel 08/32 :    0
+ip-26-0-173-7:811253:811670 [5] NCCL INFO Channel 09/32 :    0
+ip-26-0-173-7:811253:811670 [5] NCCL INFO Channel 10/32 :    0
+ip-26-0-173-7:811253:811670 [5] NCCL INFO Channel 11/32 :    0
+ip-26-0-173-7:811253:811670 [5] NCCL INFO Channel 12/32 :    0
+ip-26-0-173-7:811253:811670 [5] NCCL INFO Channel 13/32 :    0
+ip-26-0-173-7:811253:811670 [5] NCCL INFO Channel 14/32 :    0
+ip-26-0-173-7:811253:811670 [5] NCCL INFO Channel 15/32 :    0
+ip-26-0-173-7:811253:811670 [5] NCCL INFO Channel 16/32 :    0
+ip-26-0-173-7:811253:811670 [5] NCCL INFO Channel 17/32 :    0
+ip-26-0-173-7:811253:811670 [5] NCCL INFO Channel 18/32 :    0
+ip-26-0-173-7:811253:811670 [5] NCCL INFO Channel 19/32 :    0
+ip-26-0-173-7:811253:811670 [5] NCCL INFO Channel 20/32 :    0
+ip-26-0-173-7:811253:811670 [5] NCCL INFO Channel 21/32 :    0
+ip-26-0-173-7:811253:811670 [5] NCCL INFO Channel 22/32 :    0
+ip-26-0-173-7:811253:811670 [5] NCCL INFO Channel 23/32 :    0
+ip-26-0-173-7:811253:811670 [5] NCCL INFO Channel 24/32 :    0
+ip-26-0-173-7:811253:811670 [5] NCCL INFO Channel 25/32 :    0
+ip-26-0-173-7:811253:811670 [5] NCCL INFO Channel 26/32 :    0
+ip-26-0-172-142:880685:881067 [6] NCCL INFO Channel 00/32 :    0
+ip-26-0-172-142:880685:881067 [6] NCCL INFO Channel 01/32 :    0
+ip-26-0-172-142:880685:881067 [6] NCCL INFO Channel 02/32 :    0
+ip-26-0-172-142:880685:881067 [6] NCCL INFO Channel 03/32 :    0
+ip-26-0-172-142:880685:881067 [6] NCCL INFO Channel 04/32 :    0
+ip-26-0-172-142:880685:881067 [6] NCCL INFO Channel 05/32 :    0
+ip-26-0-173-7:811253:811670 [5] NCCL INFO Channel 27/32 :    0
+ip-26-0-173-7:811253:811670 [5] NCCL INFO Channel 28/32 :    0
+ip-26-0-173-7:811253:811670 [5] NCCL INFO Channel 29/32 :    0
+ip-26-0-173-7:811253:811670 [5] NCCL INFO Channel 30/32 :    0
+ip-26-0-173-7:811253:811670 [5] NCCL INFO Channel 31/32 :    0
+ip-26-0-172-142:880685:881067 [6] NCCL INFO Channel 06/32 :    0
+ip-26-0-172-142:880685:881067 [6] NCCL INFO Channel 07/32 :    0
+ip-26-0-172-142:880685:881067 [6] NCCL INFO Channel 08/32 :    0
+ip-26-0-172-142:880685:881067 [6] NCCL INFO Channel 09/32 :    0
+ip-26-0-172-142:880685:881067 [6] NCCL INFO Channel 10/32 :    0
+ip-26-0-172-142:880685:881067 [6] NCCL INFO Channel 11/32 :    0
+ip-26-0-172-142:880685:881067 [6] NCCL INFO Channel 12/32 :    0
+ip-26-0-172-142:880685:881067 [6] NCCL INFO Channel 13/32 :    0
+ip-26-0-172-142:880685:881067 [6] NCCL INFO Channel 14/32 :    0
+ip-26-0-172-142:880685:881067 [6] NCCL INFO Channel 15/32 :    0
+ip-26-0-172-142:880685:881067 [6] NCCL INFO Channel 16/32 :    0
+ip-26-0-172-142:880685:881067 [6] NCCL INFO Channel 17/32 :    0
+ip-26-0-172-142:880685:881067 [6] NCCL INFO Channel 18/32 :    0
+ip-26-0-172-142:880685:881067 [6] NCCL INFO Channel 19/32 :    0
+ip-26-0-172-142:880685:881067 [6] NCCL INFO Channel 20/32 :    0
+ip-26-0-172-142:880685:881067 [6] NCCL INFO Channel 21/32 :    0
+ip-26-0-172-142:880685:881067 [6] NCCL INFO Channel 22/32 :    0
+ip-26-0-172-142:880685:881067 [6] NCCL INFO Channel 23/32 :    0
+ip-26-0-172-142:880685:881067 [6] NCCL INFO Channel 24/32 :    0
+ip-26-0-172-142:880685:881067 [6] NCCL INFO Channel 25/32 :    0
+ip-26-0-172-142:880685:881067 [6] NCCL INFO Channel 26/32 :    0
+ip-26-0-173-7:811253:811670 [5] NCCL INFO Trees [0] -1/-1/-1->0->-1 [1] -1/-1/-1->0->-1 [2] -1/-1/-1->0->-1 [3] -1/-1/-1->0->-1 [4] -1/-1/-1->0->-1 [5] -1/-1/-1->0->-1 [6] -1/-1/-1->0->-1 [7] -1/-1/-1->0->-1 [8] -1/-1/-1->0->-1 [9] -1/-1/-1->0->-1 [10] -1/-1/-1->0->-1 [11] -1/-1/-1->0->-1 [12] -1/-1/-1->0->-1 [13] -1/-1/-1->0->-1 [14] -1/-1/-1->0->-1 [15] -1/-1/-1->0->-1 [16] -1/-1/-1->0->-1 [17] -1/-1/-1->0->-1 [18] -1/-1/-1->0->-1 [19] -1/-1/-1->0->-1 [20] -1/-1/-1->0->-1 [21] -1/-1/-1->0->-1 [22] -1/-1/-1->0->-1 [23] -1/-1/-1->0->-1 [24] -1/-1/-1->0->-1 [25] -1/-1/-1->0->-1 [26] -1/-1/-1->0->-1 [27] -1/-1/-1->0->-1 [28] -1/-1/-1->0->-1 [29] -1/-1/-1->0->-1 [30] -1/-1/-1->0->-1 [31] -1/-1/-1->0->-1
+ip-26-0-173-7:811253:811670 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-142:880685:881067 [6] NCCL INFO Channel 27/32 :    0
+ip-26-0-172-142:880685:881067 [6] NCCL INFO Channel 28/32 :    0
+ip-26-0-172-142:880685:881067 [6] NCCL INFO Channel 29/32 :    0
+ip-26-0-172-142:880685:881067 [6] NCCL INFO Channel 30/32 :    0
+ip-26-0-172-142:880685:881067 [6] NCCL INFO Channel 31/32 :    0
+ip-26-0-172-142:880685:881067 [6] NCCL INFO Trees [0] -1/-1/-1->0->-1 [1] -1/-1/-1->0->-1 [2] -1/-1/-1->0->-1 [3] -1/-1/-1->0->-1 [4] -1/-1/-1->0->-1 [5] -1/-1/-1->0->-1 [6] -1/-1/-1->0->-1 [7] -1/-1/-1->0->-1 [8] -1/-1/-1->0->-1 [9] -1/-1/-1->0->-1 [10] -1/-1/-1->0->-1 [11] -1/-1/-1->0->-1 [12] -1/-1/-1->0->-1 [13] -1/-1/-1->0->-1 [14] -1/-1/-1->0->-1 [15] -1/-1/-1->0->-1 [16] -1/-1/-1->0->-1 [17] -1/-1/-1->0->-1 [18] -1/-1/-1->0->-1 [19] -1/-1/-1->0->-1 [20] -1/-1/-1->0->-1 [21] -1/-1/-1->0->-1 [22] -1/-1/-1->0->-1 [23] -1/-1/-1->0->-1 [24] -1/-1/-1->0->-1 [25] -1/-1/-1->0->-1 [26] -1/-1/-1->0->-1 [27] -1/-1/-1->0->-1 [28] -1/-1/-1->0->-1 [29] -1/-1/-1->0->-1 [30] -1/-1/-1->0->-1 [31] -1/-1/-1->0->-1
+ip-26-0-172-142:880685:881067 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-173-7:811253:811670 [5] NCCL INFO Connected all rings
+ip-26-0-173-7:811253:811670 [5] NCCL INFO Connected all trees
+ip-26-0-173-7:811253:811670 [5] NCCL INFO 32 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-142:880685:881067 [6] NCCL INFO Connected all rings
+ip-26-0-172-142:880685:881067 [6] NCCL INFO Connected all trees
+ip-26-0-172-142:880685:881067 [6] NCCL INFO 32 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-173-7:811248:811666 [0] NCCL INFO comm 0xb8ca7c0 rank 0 nranks 1 cudaDev 0 nvmlDev 0 busId 53000 commId 0x8987331d9e95edae - Init COMPLETE
+ip-26-0-160-225:94342:94765 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-160-225:94342:94765 [0] NCCL INFO Channel 00/32 :    0
+ip-26-0-160-225:94342:94765 [0] NCCL INFO Channel 01/32 :    0
+ip-26-0-160-225:94342:94765 [0] NCCL INFO Channel 02/32 :    0
+ip-26-0-160-225:94342:94765 [0] NCCL INFO Channel 03/32 :    0
+ip-26-0-160-225:94342:94765 [0] NCCL INFO Channel 04/32 :    0
+ip-26-0-160-225:94342:94765 [0] NCCL INFO Channel 05/32 :    0
+ip-26-0-160-225:94342:94765 [0] NCCL INFO Channel 06/32 :    0
+ip-26-0-160-225:94342:94765 [0] NCCL INFO Channel 07/32 :    0
+ip-26-0-160-225:94342:94765 [0] NCCL INFO Channel 08/32 :    0
+ip-26-0-160-225:94342:94765 [0] NCCL INFO Channel 09/32 :    0
+ip-26-0-160-225:94342:94765 [0] NCCL INFO Channel 10/32 :    0
+ip-26-0-160-225:94342:94765 [0] NCCL INFO Channel 11/32 :    0
+ip-26-0-160-225:94342:94765 [0] NCCL INFO Channel 12/32 :    0
+ip-26-0-160-225:94342:94765 [0] NCCL INFO Channel 13/32 :    0
+ip-26-0-160-225:94342:94765 [0] NCCL INFO Channel 14/32 :    0
+ip-26-0-160-225:94342:94765 [0] NCCL INFO Channel 15/32 :    0
+ip-26-0-160-225:94342:94765 [0] NCCL INFO Channel 16/32 :    0
+ip-26-0-160-225:94342:94765 [0] NCCL INFO Channel 17/32 :    0
+ip-26-0-160-225:94342:94765 [0] NCCL INFO Channel 18/32 :    0
+ip-26-0-160-225:94342:94765 [0] NCCL INFO Channel 19/32 :    0
+ip-26-0-160-225:94342:94765 [0] NCCL INFO Channel 20/32 :    0
+ip-26-0-160-225:94342:94765 [0] NCCL INFO Channel 21/32 :    0
+ip-26-0-160-225:94342:94765 [0] NCCL INFO Channel 22/32 :    0
+ip-26-0-160-225:94342:94765 [0] NCCL INFO Channel 23/32 :    0
+ip-26-0-160-225:94342:94765 [0] NCCL INFO Channel 24/32 :    0
+ip-26-0-160-225:94342:94765 [0] NCCL INFO Channel 25/32 :    0
+ip-26-0-160-225:94342:94765 [0] NCCL INFO Channel 26/32 :    0
+ip-26-0-160-225:94342:94765 [0] NCCL INFO Channel 27/32 :    0
+ip-26-0-160-225:94342:94765 [0] NCCL INFO Channel 28/32 :    0
+ip-26-0-160-225:94342:94765 [0] NCCL INFO Channel 29/32 :    0
+ip-26-0-160-225:94342:94765 [0] NCCL INFO Channel 30/32 :    0
+ip-26-0-160-225:94342:94765 [0] NCCL INFO Channel 31/32 :    0
+ip-26-0-172-142:880680:881074 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-160-225:94342:94765 [0] NCCL INFO Trees [0] -1/-1/-1->0->-1 [1] -1/-1/-1->0->-1 [2] -1/-1/-1->0->-1 [3] -1/-1/-1->0->-1 [4] -1/-1/-1->0->-1 [5] -1/-1/-1->0->-1 [6] -1/-1/-1->0->-1 [7] -1/-1/-1->0->-1 [8] -1/-1/-1->0->-1 [9] -1/-1/-1->0->-1 [10] -1/-1/-1->0->-1 [11] -1/-1/-1->0->-1 [12] -1/-1/-1->0->-1 [13] -1/-1/-1->0->-1 [14] -1/-1/-1->0->-1 [15] -1/-1/-1->0->-1 [16] -1/-1/-1->0->-1 [17] -1/-1/-1->0->-1 [18] -1/-1/-1->0->-1 [19] -1/-1/-1->0->-1 [20] -1/-1/-1->0->-1 [21] -1/-1/-1->0->-1 [22] -1/-1/-1->0->-1 [23] -1/-1/-1->0->-1 [24] -1/-1/-1->0->-1 [25] -1/-1/-1->0->-1 [26] -1/-1/-1->0->-1 [27] -1/-1/-1->0->-1 [28] -1/-1/-1->0->-1 [29] -1/-1/-1->0->-1 [30] -1/-1/-1->0->-1 [31] -1/-1/-1->0->-1
+ip-26-0-160-225:94342:94765 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-142:880680:881074 [1] NCCL INFO Channel 00/32 :    0
+ip-26-0-172-142:880680:881074 [1] NCCL INFO Channel 01/32 :    0
+ip-26-0-172-142:880680:881074 [1] NCCL INFO Channel 02/32 :    0
+ip-26-0-172-142:880680:881074 [1] NCCL INFO Channel 03/32 :    0
+ip-26-0-172-142:880680:881074 [1] NCCL INFO Channel 04/32 :    0
+ip-26-0-172-142:880680:881074 [1] NCCL INFO Channel 05/32 :    0
+ip-26-0-172-142:880680:881074 [1] NCCL INFO Channel 06/32 :    0
+ip-26-0-172-142:880680:881074 [1] NCCL INFO Channel 07/32 :    0
+ip-26-0-172-142:880680:881074 [1] NCCL INFO Channel 08/32 :    0
+ip-26-0-172-142:880680:881074 [1] NCCL INFO Channel 09/32 :    0
+ip-26-0-172-142:880680:881074 [1] NCCL INFO Channel 10/32 :    0
+ip-26-0-172-142:880680:881074 [1] NCCL INFO Channel 11/32 :    0
+ip-26-0-172-142:880680:881074 [1] NCCL INFO Channel 12/32 :    0
+ip-26-0-172-142:880680:881074 [1] NCCL INFO Channel 13/32 :    0
+ip-26-0-172-142:880680:881074 [1] NCCL INFO Channel 14/32 :    0
+ip-26-0-172-142:880680:881074 [1] NCCL INFO Channel 15/32 :    0
+ip-26-0-172-142:880680:881074 [1] NCCL INFO Channel 16/32 :    0
+ip-26-0-172-142:880680:881074 [1] NCCL INFO Channel 17/32 :    0
+ip-26-0-172-142:880680:881074 [1] NCCL INFO Channel 18/32 :    0
+ip-26-0-172-142:880680:881074 [1] NCCL INFO Channel 19/32 :    0
+ip-26-0-172-142:880680:881074 [1] NCCL INFO Channel 20/32 :    0
+ip-26-0-172-142:880680:881074 [1] NCCL INFO Channel 21/32 :    0
+ip-26-0-172-142:880680:881074 [1] NCCL INFO Channel 22/32 :    0
+ip-26-0-172-142:880680:881074 [1] NCCL INFO Channel 23/32 :    0
+ip-26-0-172-142:880680:881074 [1] NCCL INFO Channel 24/32 :    0
+ip-26-0-172-142:880680:881074 [1] NCCL INFO Channel 25/32 :    0
+ip-26-0-172-142:880680:881074 [1] NCCL INFO Channel 26/32 :    0
+ip-26-0-172-142:880680:881074 [1] NCCL INFO Channel 27/32 :    0
+ip-26-0-172-142:880680:881074 [1] NCCL INFO Channel 28/32 :    0
+ip-26-0-172-142:880680:881074 [1] NCCL INFO Channel 29/32 :    0
+ip-26-0-172-142:880680:881074 [1] NCCL INFO Channel 30/32 :    0
+ip-26-0-172-142:880680:881074 [1] NCCL INFO Channel 31/32 :    0
+ip-26-0-172-142:880680:881074 [1] NCCL INFO Trees [0] -1/-1/-1->0->-1 [1] -1/-1/-1->0->-1 [2] -1/-1/-1->0->-1 [3] -1/-1/-1->0->-1 [4] -1/-1/-1->0->-1 [5] -1/-1/-1->0->-1 [6] -1/-1/-1->0->-1 [7] -1/-1/-1->0->-1 [8] -1/-1/-1->0->-1 [9] -1/-1/-1->0->-1 [10] -1/-1/-1->0->-1 [11] -1/-1/-1->0->-1 [12] -1/-1/-1->0->-1 [13] -1/-1/-1->0->-1 [14] -1/-1/-1->0->-1 [15] -1/-1/-1->0->-1 [16] -1/-1/-1->0->-1 [17] -1/-1/-1->0->-1 [18] -1/-1/-1->0->-1 [19] -1/-1/-1->0->-1 [20] -1/-1/-1->0->-1 [21] -1/-1/-1->0->-1 [22] -1/-1/-1->0->-1 [23] -1/-1/-1->0->-1 [24] -1/-1/-1->0->-1 [25] -1/-1/-1->0->-1 [26] -1/-1/-1->0->-1 [27] -1/-1/-1->0->-1 [28] -1/-1/-1->0->-1 [29] -1/-1/-1->0->-1 [30] -1/-1/-1->0->-1 [31] -1/-1/-1->0->-1
+ip-26-0-172-142:880680:881074 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-225:94342:94765 [0] NCCL INFO Connected all rings
+ip-26-0-160-225:94342:94765 [0] NCCL INFO Connected all trees
+ip-26-0-160-225:94342:94765 [0] NCCL INFO 32 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-147:618590:618984 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-172-142:880680:881074 [1] NCCL INFO Connected all rings
+ip-26-0-172-142:880680:881074 [1] NCCL INFO Connected all trees
+ip-26-0-172-142:880680:881074 [1] NCCL INFO 32 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-147:618590:618984 [6] NCCL INFO Channel 00/32 :    0
+ip-26-0-172-147:618590:618984 [6] NCCL INFO Channel 01/32 :    0
+ip-26-0-172-147:618590:618984 [6] NCCL INFO Channel 02/32 :    0
+ip-26-0-172-147:618590:618984 [6] NCCL INFO Channel 03/32 :    0
+ip-26-0-172-147:618590:618984 [6] NCCL INFO Channel 04/32 :    0
+ip-26-0-172-147:618590:618984 [6] NCCL INFO Channel 05/32 :    0
+ip-26-0-172-147:618590:618984 [6] NCCL INFO Channel 06/32 :    0
+ip-26-0-172-147:618590:618984 [6] NCCL INFO Channel 07/32 :    0
+ip-26-0-172-147:618590:618984 [6] NCCL INFO Channel 08/32 :    0
+ip-26-0-172-147:618590:618984 [6] NCCL INFO Channel 09/32 :    0
+ip-26-0-172-147:618590:618984 [6] NCCL INFO Channel 10/32 :    0
+ip-26-0-172-147:618590:618984 [6] NCCL INFO Channel 11/32 :    0
+ip-26-0-172-147:618590:618984 [6] NCCL INFO Channel 12/32 :    0
+ip-26-0-172-147:618590:618984 [6] NCCL INFO Channel 13/32 :    0
+ip-26-0-172-147:618590:618984 [6] NCCL INFO Channel 14/32 :    0
+ip-26-0-172-147:618590:618984 [6] NCCL INFO Channel 15/32 :    0
+ip-26-0-172-147:618590:618984 [6] NCCL INFO Channel 16/32 :    0
+ip-26-0-172-147:618590:618984 [6] NCCL INFO Channel 17/32 :    0
+ip-26-0-172-147:618590:618984 [6] NCCL INFO Channel 18/32 :    0
+ip-26-0-172-147:618590:618984 [6] NCCL INFO Channel 19/32 :    0
+ip-26-0-172-147:618590:618984 [6] NCCL INFO Channel 20/32 :    0
+ip-26-0-172-147:618590:618984 [6] NCCL INFO Channel 21/32 :    0
+ip-26-0-172-147:618590:618984 [6] NCCL INFO Channel 22/32 :    0
+ip-26-0-172-147:618590:618984 [6] NCCL INFO Channel 23/32 :    0
+ip-26-0-172-147:618590:618984 [6] NCCL INFO Channel 24/32 :    0
+ip-26-0-172-147:618590:618984 [6] NCCL INFO Channel 25/32 :    0
+ip-26-0-172-147:618590:618984 [6] NCCL INFO Channel 26/32 :    0
+ip-26-0-172-147:618590:618984 [6] NCCL INFO Channel 27/32 :    0
+ip-26-0-172-147:618590:618984 [6] NCCL INFO Channel 28/32 :    0
+ip-26-0-172-147:618590:618984 [6] NCCL INFO Channel 29/32 :    0
+ip-26-0-172-147:618590:618984 [6] NCCL INFO Channel 30/32 :    0
+ip-26-0-172-147:618590:618984 [6] NCCL INFO Channel 31/32 :    0
+ip-26-0-172-147:618590:618984 [6] NCCL INFO Trees [0] -1/-1/-1->0->-1 [1] -1/-1/-1->0->-1 [2] -1/-1/-1->0->-1 [3] -1/-1/-1->0->-1 [4] -1/-1/-1->0->-1 [5] -1/-1/-1->0->-1 [6] -1/-1/-1->0->-1 [7] -1/-1/-1->0->-1 [8] -1/-1/-1->0->-1 [9] -1/-1/-1->0->-1 [10] -1/-1/-1->0->-1 [11] -1/-1/-1->0->-1 [12] -1/-1/-1->0->-1 [13] -1/-1/-1->0->-1 [14] -1/-1/-1->0->-1 [15] -1/-1/-1->0->-1 [16] -1/-1/-1->0->-1 [17] -1/-1/-1->0->-1 [18] -1/-1/-1->0->-1 [19] -1/-1/-1->0->-1 [20] -1/-1/-1->0->-1 [21] -1/-1/-1->0->-1 [22] -1/-1/-1->0->-1 [23] -1/-1/-1->0->-1 [24] -1/-1/-1->0->-1 [25] -1/-1/-1->0->-1 [26] -1/-1/-1->0->-1 [27] -1/-1/-1->0->-1 [28] -1/-1/-1->0->-1 [29] -1/-1/-1->0->-1 [30] -1/-1/-1->0->-1 [31] -1/-1/-1->0->-1
+ip-26-0-172-147:618590:618984 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-173-7:811250:811669 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-173-7:811250:811669 [2] NCCL INFO Channel 00/32 :    0
+ip-26-0-173-7:811250:811669 [2] NCCL INFO Channel 01/32 :    0
+ip-26-0-173-7:811250:811669 [2] NCCL INFO Channel 02/32 :    0
+ip-26-0-173-7:811250:811669 [2] NCCL INFO Channel 03/32 :    0
+ip-26-0-173-7:811250:811669 [2] NCCL INFO Channel 04/32 :    0
+ip-26-0-173-7:811250:811669 [2] NCCL INFO Channel 05/32 :    0
+ip-26-0-173-7:811250:811669 [2] NCCL INFO Channel 06/32 :    0
+ip-26-0-173-7:811250:811669 [2] NCCL INFO Channel 07/32 :    0
+ip-26-0-173-7:811250:811669 [2] NCCL INFO Channel 08/32 :    0
+ip-26-0-173-7:811250:811669 [2] NCCL INFO Channel 09/32 :    0
+ip-26-0-173-7:811250:811669 [2] NCCL INFO Channel 10/32 :    0
+ip-26-0-173-7:811250:811669 [2] NCCL INFO Channel 11/32 :    0
+ip-26-0-173-7:811250:811669 [2] NCCL INFO Channel 12/32 :    0
+ip-26-0-173-7:811250:811669 [2] NCCL INFO Channel 13/32 :    0
+ip-26-0-173-7:811250:811669 [2] NCCL INFO Channel 14/32 :    0
+ip-26-0-173-7:811250:811669 [2] NCCL INFO Channel 15/32 :    0
+ip-26-0-173-7:811250:811669 [2] NCCL INFO Channel 16/32 :    0
+ip-26-0-173-7:811250:811669 [2] NCCL INFO Channel 17/32 :    0
+ip-26-0-173-7:811250:811669 [2] NCCL INFO Channel 18/32 :    0
+ip-26-0-173-7:811250:811669 [2] NCCL INFO Channel 19/32 :    0
+ip-26-0-173-7:811250:811669 [2] NCCL INFO Channel 20/32 :    0
+ip-26-0-173-7:811250:811669 [2] NCCL INFO Channel 21/32 :    0
+ip-26-0-173-7:811250:811669 [2] NCCL INFO Channel 22/32 :    0
+ip-26-0-173-7:811250:811669 [2] NCCL INFO Channel 23/32 :    0
+ip-26-0-173-7:811250:811669 [2] NCCL INFO Channel 24/32 :    0
+ip-26-0-173-7:811250:811669 [2] NCCL INFO Channel 25/32 :    0
+ip-26-0-173-7:811250:811669 [2] NCCL INFO Channel 26/32 :    0
+ip-26-0-173-7:811250:811669 [2] NCCL INFO Channel 27/32 :    0
+ip-26-0-173-7:811250:811669 [2] NCCL INFO Channel 28/32 :    0
+ip-26-0-173-7:811250:811669 [2] NCCL INFO Channel 29/32 :    0
+ip-26-0-173-7:811250:811669 [2] NCCL INFO Channel 30/32 :    0
+ip-26-0-173-7:811250:811669 [2] NCCL INFO Channel 31/32 :    0
+ip-26-0-173-7:811250:811669 [2] NCCL INFO Trees [0] -1/-1/-1->0->-1 [1] -1/-1/-1->0->-1 [2] -1/-1/-1->0->-1 [3] -1/-1/-1->0->-1 [4] -1/-1/-1->0->-1 [5] -1/-1/-1->0->-1 [6] -1/-1/-1->0->-1 [7] -1/-1/-1->0->-1 [8] -1/-1/-1->0->-1 [9] -1/-1/-1->0->-1 [10] -1/-1/-1->0->-1 [11] -1/-1/-1->0->-1 [12] -1/-1/-1->0->-1 [13] -1/-1/-1->0->-1 [14] -1/-1/-1->0->-1 [15] -1/-1/-1->0->-1 [16] -1/-1/-1->0->-1 [17] -1/-1/-1->0->-1 [18] -1/-1/-1->0->-1 [19] -1/-1/-1->0->-1 [20] -1/-1/-1->0->-1 [21] -1/-1/-1->0->-1 [22] -1/-1/-1->0->-1 [23] -1/-1/-1->0->-1 [24] -1/-1/-1->0->-1 [25] -1/-1/-1->0->-1 [26] -1/-1/-1->0->-1 [27] -1/-1/-1->0->-1 [28] -1/-1/-1->0->-1 [29] -1/-1/-1->0->-1 [30] -1/-1/-1->0->-1 [31] -1/-1/-1->0->-1
+ip-26-0-173-7:811250:811669 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-225:94343:94771 [1] NCCL INFO comm 0x9c72b30 rank 0 nranks 1 cudaDev 1 nvmlDev 1 busId 64000 commId 0x646dc9f2cbea7601 - Init COMPLETE
+ip-26-0-172-147:618590:618984 [6] NCCL INFO Connected all rings
+ip-26-0-172-147:618590:618984 [6] NCCL INFO Connected all trees
+ip-26-0-172-147:618590:618984 [6] NCCL INFO 32 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-252:688202:688605 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-173-7:811250:811669 [2] NCCL INFO Connected all rings
+ip-26-0-173-7:811250:811669 [2] NCCL INFO Connected all trees
+ip-26-0-173-7:811250:811669 [2] NCCL INFO 32 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-116:1019290:1019691 [2] NCCL INFO comm 0xb532d30 rank 0 nranks 1 cudaDev 2 nvmlDev 2 busId 75000 commId 0x2c933182693940d7 - Init COMPLETE
+ip-26-0-172-252:688202:688605 [2] NCCL INFO Channel 00/32 :    0
+ip-26-0-172-252:688202:688605 [2] NCCL INFO Channel 01/32 :    0
+ip-26-0-172-252:688202:688605 [2] NCCL INFO Channel 02/32 :    0
+ip-26-0-172-252:688202:688605 [2] NCCL INFO Channel 03/32 :    0
+ip-26-0-172-252:688202:688605 [2] NCCL INFO Channel 04/32 :    0
+ip-26-0-172-252:688202:688605 [2] NCCL INFO Channel 05/32 :    0
+ip-26-0-172-252:688202:688605 [2] NCCL INFO Channel 06/32 :    0
+ip-26-0-172-252:688202:688605 [2] NCCL INFO Channel 07/32 :    0
+ip-26-0-172-252:688202:688605 [2] NCCL INFO Channel 08/32 :    0
+ip-26-0-172-252:688202:688605 [2] NCCL INFO Channel 09/32 :    0
+ip-26-0-172-252:688202:688605 [2] NCCL INFO Channel 10/32 :    0
+ip-26-0-172-252:688202:688605 [2] NCCL INFO Channel 11/32 :    0
+ip-26-0-172-252:688202:688605 [2] NCCL INFO Channel 12/32 :    0
+ip-26-0-172-252:688202:688605 [2] NCCL INFO Channel 13/32 :    0
+ip-26-0-172-252:688202:688605 [2] NCCL INFO Channel 14/32 :    0
+ip-26-0-172-252:688202:688605 [2] NCCL INFO Channel 15/32 :    0
+ip-26-0-172-252:688202:688605 [2] NCCL INFO Channel 16/32 :    0
+ip-26-0-172-252:688202:688605 [2] NCCL INFO Channel 17/32 :    0
+ip-26-0-172-252:688202:688605 [2] NCCL INFO Channel 18/32 :    0
+ip-26-0-172-252:688202:688605 [2] NCCL INFO Channel 19/32 :    0
+ip-26-0-172-252:688202:688605 [2] NCCL INFO Channel 20/32 :    0
+ip-26-0-172-252:688202:688605 [2] NCCL INFO Channel 21/32 :    0
+ip-26-0-172-252:688202:688605 [2] NCCL INFO Channel 22/32 :    0
+ip-26-0-172-252:688202:688605 [2] NCCL INFO Channel 23/32 :    0
+ip-26-0-172-252:688202:688605 [2] NCCL INFO Channel 24/32 :    0
+ip-26-0-172-252:688202:688605 [2] NCCL INFO Channel 25/32 :    0
+ip-26-0-172-252:688202:688605 [2] NCCL INFO Channel 26/32 :    0
+ip-26-0-172-252:688202:688605 [2] NCCL INFO Channel 27/32 :    0
+ip-26-0-172-252:688202:688605 [2] NCCL INFO Channel 28/32 :    0
+ip-26-0-172-252:688202:688605 [2] NCCL INFO Channel 29/32 :    0
+ip-26-0-172-252:688202:688605 [2] NCCL INFO Channel 30/32 :    0
+ip-26-0-172-252:688202:688605 [2] NCCL INFO Channel 31/32 :    0
+ip-26-0-172-252:688202:688605 [2] NCCL INFO Trees [0] -1/-1/-1->0->-1 [1] -1/-1/-1->0->-1 [2] -1/-1/-1->0->-1 [3] -1/-1/-1->0->-1 [4] -1/-1/-1->0->-1 [5] -1/-1/-1->0->-1 [6] -1/-1/-1->0->-1 [7] -1/-1/-1->0->-1 [8] -1/-1/-1->0->-1 [9] -1/-1/-1->0->-1 [10] -1/-1/-1->0->-1 [11] -1/-1/-1->0->-1 [12] -1/-1/-1->0->-1 [13] -1/-1/-1->0->-1 [14] -1/-1/-1->0->-1 [15] -1/-1/-1->0->-1 [16] -1/-1/-1->0->-1 [17] -1/-1/-1->0->-1 [18] -1/-1/-1->0->-1 [19] -1/-1/-1->0->-1 [20] -1/-1/-1->0->-1 [21] -1/-1/-1->0->-1 [22] -1/-1/-1->0->-1 [23] -1/-1/-1->0->-1 [24] -1/-1/-1->0->-1 [25] -1/-1/-1->0->-1 [26] -1/-1/-1->0->-1 [27] -1/-1/-1->0->-1 [28] -1/-1/-1->0->-1 [29] -1/-1/-1->0->-1 [30] -1/-1/-1->0->-1 [31] -1/-1/-1->0->-1
+ip-26-0-172-252:688202:688605 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-116:1019294:1019686 [6] NCCL INFO comm 0xa59e630 rank 0 nranks 1 cudaDev 6 nvmlDev 6 busId b9000 commId 0x88675990041435c9 - Init COMPLETE
+ip-26-0-172-252:688202:688605 [2] NCCL INFO Connected all rings
+ip-26-0-172-252:688202:688605 [2] NCCL INFO Connected all trees
+ip-26-0-172-252:688202:688605 [2] NCCL INFO 32 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-160-225:94347:94772 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-172-147:618584:618980 [0] NCCL INFO comm 0xba75700 rank 0 nranks 1 cudaDev 0 nvmlDev 0 busId 53000 commId 0xa4a1c3e91a5f4ee6 - Init COMPLETE
+ip-26-0-172-252:688200:688603 [0] NCCL INFO comm 0xa91c990 rank 0 nranks 1 cudaDev 0 nvmlDev 0 busId 53000 commId 0x3de5d39e61f58567 - Init COMPLETE
+ip-26-0-172-252:688206:688597 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-172-252:688206:688597 [6] NCCL INFO Channel 00/32 :    0
+ip-26-0-172-252:688206:688597 [6] NCCL INFO Channel 01/32 :    0
+ip-26-0-172-252:688206:688597 [6] NCCL INFO Channel 02/32 :    0
+ip-26-0-172-252:688206:688597 [6] NCCL INFO Channel 03/32 :    0
+ip-26-0-172-252:688206:688597 [6] NCCL INFO Channel 04/32 :    0
+ip-26-0-172-252:688206:688597 [6] NCCL INFO Channel 05/32 :    0
+ip-26-0-172-252:688206:688597 [6] NCCL INFO Channel 06/32 :    0
+ip-26-0-172-252:688206:688597 [6] NCCL INFO Channel 07/32 :    0
+ip-26-0-172-252:688206:688597 [6] NCCL INFO Channel 08/32 :    0
+ip-26-0-172-252:688206:688597 [6] NCCL INFO Channel 09/32 :    0
+ip-26-0-172-252:688206:688597 [6] NCCL INFO Channel 10/32 :    0
+ip-26-0-172-252:688206:688597 [6] NCCL INFO Channel 11/32 :    0
+ip-26-0-172-252:688206:688597 [6] NCCL INFO Channel 12/32 :    0
+ip-26-0-172-252:688206:688597 [6] NCCL INFO Channel 13/32 :    0
+ip-26-0-172-252:688206:688597 [6] NCCL INFO Channel 14/32 :    0
+ip-26-0-172-252:688206:688597 [6] NCCL INFO Channel 15/32 :    0
+ip-26-0-172-252:688206:688597 [6] NCCL INFO Channel 16/32 :    0
+ip-26-0-172-252:688206:688597 [6] NCCL INFO Channel 17/32 :    0
+ip-26-0-172-252:688206:688597 [6] NCCL INFO Channel 18/32 :    0
+ip-26-0-172-252:688206:688597 [6] NCCL INFO Channel 19/32 :    0
+ip-26-0-172-252:688206:688597 [6] NCCL INFO Channel 20/32 :    0
+ip-26-0-172-252:688206:688597 [6] NCCL INFO Channel 21/32 :    0
+ip-26-0-172-252:688206:688597 [6] NCCL INFO Channel 22/32 :    0
+ip-26-0-172-252:688206:688597 [6] NCCL INFO Channel 23/32 :    0
+ip-26-0-172-252:688206:688597 [6] NCCL INFO Channel 24/32 :    0
+ip-26-0-172-252:688206:688597 [6] NCCL INFO Channel 25/32 :    0
+ip-26-0-172-252:688206:688597 [6] NCCL INFO Channel 26/32 :    0
+ip-26-0-172-252:688206:688597 [6] NCCL INFO Channel 27/32 :    0
+ip-26-0-172-252:688206:688597 [6] NCCL INFO Channel 28/32 :    0
+ip-26-0-172-252:688206:688597 [6] NCCL INFO Channel 29/32 :    0
+ip-26-0-172-252:688206:688597 [6] NCCL INFO Channel 30/32 :    0
+ip-26-0-172-252:688206:688597 [6] NCCL INFO Channel 31/32 :    0
+ip-26-0-172-252:688206:688597 [6] NCCL INFO Trees [0] -1/-1/-1->0->-1 [1] -1/-1/-1->0->-1 [2] -1/-1/-1->0->-1 [3] -1/-1/-1->0->-1 [4] -1/-1/-1->0->-1 [5] -1/-1/-1->0->-1 [6] -1/-1/-1->0->-1 [7] -1/-1/-1->0->-1 [8] -1/-1/-1->0->-1 [9] -1/-1/-1->0->-1 [10] -1/-1/-1->0->-1 [11] -1/-1/-1->0->-1 [12] -1/-1/-1->0->-1 [13] -1/-1/-1->0->-1 [14] -1/-1/-1->0->-1 [15] -1/-1/-1->0->-1 [16] -1/-1/-1->0->-1 [17] -1/-1/-1->0->-1 [18] -1/-1/-1->0->-1 [19] -1/-1/-1->0->-1 [20] -1/-1/-1->0->-1 [21] -1/-1/-1->0->-1 [22] -1/-1/-1->0->-1 [23] -1/-1/-1->0->-1 [24] -1/-1/-1->0->-1 [25] -1/-1/-1->0->-1 [26] -1/-1/-1->0->-1 [27] -1/-1/-1->0->-1 [28] -1/-1/-1->0->-1 [29] -1/-1/-1->0->-1 [30] -1/-1/-1->0->-1 [31] -1/-1/-1->0->-1
+ip-26-0-172-252:688206:688597 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-173-7:811252:811674 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-173-7:811252:811674 [4] NCCL INFO Channel 00/32 :    0
+ip-26-0-173-7:811252:811674 [4] NCCL INFO Channel 01/32 :    0
+ip-26-0-173-7:811252:811674 [4] NCCL INFO Channel 02/32 :    0
+ip-26-0-173-7:811252:811674 [4] NCCL INFO Channel 03/32 :    0
+ip-26-0-173-7:811252:811674 [4] NCCL INFO Channel 04/32 :    0
+ip-26-0-173-7:811252:811674 [4] NCCL INFO Channel 05/32 :    0
+ip-26-0-173-7:811252:811674 [4] NCCL INFO Channel 06/32 :    0
+ip-26-0-173-7:811252:811674 [4] NCCL INFO Channel 07/32 :    0
+ip-26-0-173-7:811252:811674 [4] NCCL INFO Channel 08/32 :    0
+ip-26-0-173-7:811252:811674 [4] NCCL INFO Channel 09/32 :    0
+ip-26-0-173-7:811252:811674 [4] NCCL INFO Channel 10/32 :    0
+ip-26-0-173-7:811252:811674 [4] NCCL INFO Channel 11/32 :    0
+ip-26-0-173-7:811252:811674 [4] NCCL INFO Channel 12/32 :    0
+ip-26-0-173-7:811252:811674 [4] NCCL INFO Channel 13/32 :    0
+ip-26-0-173-7:811252:811674 [4] NCCL INFO Channel 14/32 :    0
+ip-26-0-173-7:811252:811674 [4] NCCL INFO Channel 15/32 :    0
+ip-26-0-173-7:811252:811674 [4] NCCL INFO Channel 16/32 :    0
+ip-26-0-173-7:811252:811674 [4] NCCL INFO Channel 17/32 :    0
+ip-26-0-173-7:811252:811674 [4] NCCL INFO Channel 18/32 :    0
+ip-26-0-173-7:811252:811674 [4] NCCL INFO Channel 19/32 :    0
+ip-26-0-173-7:811252:811674 [4] NCCL INFO Channel 20/32 :    0
+ip-26-0-173-7:811252:811674 [4] NCCL INFO Channel 21/32 :    0
+ip-26-0-173-7:811252:811674 [4] NCCL INFO Channel 22/32 :    0
+ip-26-0-173-7:811252:811674 [4] NCCL INFO Channel 23/32 :    0
+ip-26-0-173-7:811252:811674 [4] NCCL INFO Channel 24/32 :    0
+ip-26-0-173-7:811252:811674 [4] NCCL INFO Channel 25/32 :    0
+ip-26-0-173-7:811252:811674 [4] NCCL INFO Channel 26/32 :    0
+ip-26-0-173-7:811252:811674 [4] NCCL INFO Channel 27/32 :    0
+ip-26-0-173-7:811252:811674 [4] NCCL INFO Channel 28/32 :    0
+ip-26-0-173-7:811252:811674 [4] NCCL INFO Channel 29/32 :    0
+ip-26-0-173-7:811252:811674 [4] NCCL INFO Channel 30/32 :    0
+ip-26-0-173-7:811252:811674 [4] NCCL INFO Channel 31/32 :    0
+ip-26-0-173-7:811252:811674 [4] NCCL INFO Trees [0] -1/-1/-1->0->-1 [1] -1/-1/-1->0->-1 [2] -1/-1/-1->0->-1 [3] -1/-1/-1->0->-1 [4] -1/-1/-1->0->-1 [5] -1/-1/-1->0->-1 [6] -1/-1/-1->0->-1 [7] -1/-1/-1->0->-1 [8] -1/-1/-1->0->-1 [9] -1/-1/-1->0->-1 [10] -1/-1/-1->0->-1 [11] -1/-1/-1->0->-1 [12] -1/-1/-1->0->-1 [13] -1/-1/-1->0->-1 [14] -1/-1/-1->0->-1 [15] -1/-1/-1->0->-1 [16] -1/-1/-1->0->-1 [17] -1/-1/-1->0->-1 [18] -1/-1/-1->0->-1 [19] -1/-1/-1->0->-1 [20] -1/-1/-1->0->-1 [21] -1/-1/-1->0->-1 [22] -1/-1/-1->0->-1 [23] -1/-1/-1->0->-1 [24] -1/-1/-1->0->-1 [25] -1/-1/-1->0->-1 [26] -1/-1/-1->0->-1 [27] -1/-1/-1->0->-1 [28] -1/-1/-1->0->-1 [29] -1/-1/-1->0->-1 [30] -1/-1/-1->0->-1 [31] -1/-1/-1->0->-1
+ip-26-0-173-7:811252:811674 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-252:688206:688597 [6] NCCL INFO Connected all rings
+ip-26-0-172-252:688206:688597 [6] NCCL INFO Connected all trees
+ip-26-0-160-225:94347:94772 [5] NCCL INFO Channel 00/32 :    0
+ip-26-0-160-225:94347:94772 [5] NCCL INFO Channel 01/32 :    0
+ip-26-0-160-225:94347:94772 [5] NCCL INFO Channel 02/32 :    0
+ip-26-0-160-225:94347:94772 [5] NCCL INFO Channel 03/32 :    0
+ip-26-0-160-225:94347:94772 [5] NCCL INFO Channel 04/32 :    0
+ip-26-0-160-225:94347:94772 [5] NCCL INFO Channel 05/32 :    0
+ip-26-0-160-225:94347:94772 [5] NCCL INFO Channel 06/32 :    0
+ip-26-0-160-225:94347:94772 [5] NCCL INFO Channel 07/32 :    0
+ip-26-0-160-225:94347:94772 [5] NCCL INFO Channel 08/32 :    0
+ip-26-0-160-225:94347:94772 [5] NCCL INFO Channel 09/32 :    0
+ip-26-0-160-225:94347:94772 [5] NCCL INFO Channel 10/32 :    0
+ip-26-0-160-225:94347:94772 [5] NCCL INFO Channel 11/32 :    0
+ip-26-0-160-225:94347:94772 [5] NCCL INFO Channel 12/32 :    0
+ip-26-0-160-225:94347:94772 [5] NCCL INFO Channel 13/32 :    0
+ip-26-0-160-225:94347:94772 [5] NCCL INFO Channel 14/32 :    0
+ip-26-0-160-225:94347:94772 [5] NCCL INFO Channel 15/32 :    0
+ip-26-0-160-225:94347:94772 [5] NCCL INFO Channel 16/32 :    0
+ip-26-0-160-225:94347:94772 [5] NCCL INFO Channel 17/32 :    0
+ip-26-0-160-225:94347:94772 [5] NCCL INFO Channel 18/32 :    0
+ip-26-0-160-225:94347:94772 [5] NCCL INFO Channel 19/32 :    0
+ip-26-0-160-225:94347:94772 [5] NCCL INFO Channel 20/32 :    0
+ip-26-0-160-225:94347:94772 [5] NCCL INFO Channel 21/32 :    0
+ip-26-0-160-225:94347:94772 [5] NCCL INFO Channel 22/32 :    0
+ip-26-0-172-252:688206:688597 [6] NCCL INFO 32 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-160-225:94347:94772 [5] NCCL INFO Channel 23/32 :    0
+ip-26-0-160-225:94347:94772 [5] NCCL INFO Channel 24/32 :    0
+ip-26-0-160-225:94347:94772 [5] NCCL INFO Channel 25/32 :    0
+ip-26-0-160-225:94347:94772 [5] NCCL INFO Channel 26/32 :    0
+ip-26-0-160-225:94347:94772 [5] NCCL INFO Channel 27/32 :    0
+ip-26-0-160-225:94347:94772 [5] NCCL INFO Channel 28/32 :    0
+ip-26-0-160-225:94347:94772 [5] NCCL INFO Channel 29/32 :    0
+ip-26-0-160-225:94347:94772 [5] NCCL INFO Channel 30/32 :    0
+ip-26-0-160-225:94347:94772 [5] NCCL INFO Channel 31/32 :    0
+ip-26-0-160-225:94347:94772 [5] NCCL INFO Trees [0] -1/-1/-1->0->-1 [1] -1/-1/-1->0->-1 [2] -1/-1/-1->0->-1 [3] -1/-1/-1->0->-1 [4] -1/-1/-1->0->-1 [5] -1/-1/-1->0->-1 [6] -1/-1/-1->0->-1 [7] -1/-1/-1->0->-1 [8] -1/-1/-1->0->-1 [9] -1/-1/-1->0->-1 [10] -1/-1/-1->0->-1 [11] -1/-1/-1->0->-1 [12] -1/-1/-1->0->-1 [13] -1/-1/-1->0->-1 [14] -1/-1/-1->0->-1 [15] -1/-1/-1->0->-1 [16] -1/-1/-1->0->-1 [17] -1/-1/-1->0->-1 [18] -1/-1/-1->0->-1 [19] -1/-1/-1->0->-1 [20] -1/-1/-1->0->-1 [21] -1/-1/-1->0->-1 [22] -1/-1/-1->0->-1 [23] -1/-1/-1->0->-1 [24] -1/-1/-1->0->-1 [25] -1/-1/-1->0->-1 [26] -1/-1/-1->0->-1 [27] -1/-1/-1->0->-1 [28] -1/-1/-1->0->-1 [29] -1/-1/-1->0->-1 [30] -1/-1/-1->0->-1 [31] -1/-1/-1->0->-1
+ip-26-0-160-225:94347:94772 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-173-7:811252:811674 [4] NCCL INFO Connected all rings
+ip-26-0-173-7:811252:811674 [4] NCCL INFO Connected all trees
+ip-26-0-173-7:811252:811674 [4] NCCL INFO 32 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-116:1019293:1019681 [5] NCCL INFO comm 0xc39f4c0 rank 0 nranks 1 cudaDev 5 nvmlDev 5 busId a8000 commId 0x94b8b9b1172ac7f7 - Init COMPLETE
+ip-26-0-160-225:94347:94772 [5] NCCL INFO Connected all rings
+ip-26-0-160-225:94347:94772 [5] NCCL INFO Connected all trees
+ip-26-0-160-225:94347:94772 [5] NCCL INFO 32 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-252:688201:688600 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-172-252:688201:688600 [1] NCCL INFO Channel 00/32 :    0
+ip-26-0-172-252:688201:688600 [1] NCCL INFO Channel 01/32 :    0
+ip-26-0-172-252:688201:688600 [1] NCCL INFO Channel 02/32 :    0
+ip-26-0-172-252:688201:688600 [1] NCCL INFO Channel 03/32 :    0
+ip-26-0-172-252:688201:688600 [1] NCCL INFO Channel 04/32 :    0
+ip-26-0-172-252:688201:688600 [1] NCCL INFO Channel 05/32 :    0
+ip-26-0-172-252:688201:688600 [1] NCCL INFO Channel 06/32 :    0
+ip-26-0-172-252:688201:688600 [1] NCCL INFO Channel 07/32 :    0
+ip-26-0-172-252:688201:688600 [1] NCCL INFO Channel 08/32 :    0
+ip-26-0-172-252:688201:688600 [1] NCCL INFO Channel 09/32 :    0
+ip-26-0-172-252:688201:688600 [1] NCCL INFO Channel 10/32 :    0
+ip-26-0-172-252:688201:688600 [1] NCCL INFO Channel 11/32 :    0
+ip-26-0-172-252:688201:688600 [1] NCCL INFO Channel 12/32 :    0
+ip-26-0-172-252:688201:688600 [1] NCCL INFO Channel 13/32 :    0
+ip-26-0-172-252:688201:688600 [1] NCCL INFO Channel 14/32 :    0
+ip-26-0-172-252:688201:688600 [1] NCCL INFO Channel 15/32 :    0
+ip-26-0-172-252:688201:688600 [1] NCCL INFO Channel 16/32 :    0
+ip-26-0-172-252:688201:688600 [1] NCCL INFO Channel 17/32 :    0
+ip-26-0-172-252:688201:688600 [1] NCCL INFO Channel 18/32 :    0
+ip-26-0-172-252:688201:688600 [1] NCCL INFO Channel 19/32 :    0
+ip-26-0-172-252:688201:688600 [1] NCCL INFO Channel 20/32 :    0
+ip-26-0-172-252:688201:688600 [1] NCCL INFO Channel 21/32 :    0
+ip-26-0-172-252:688201:688600 [1] NCCL INFO Channel 22/32 :    0
+ip-26-0-172-252:688201:688600 [1] NCCL INFO Channel 23/32 :    0
+ip-26-0-160-225:94348:94774 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-172-252:688201:688600 [1] NCCL INFO Channel 24/32 :    0
+ip-26-0-172-252:688201:688600 [1] NCCL INFO Channel 25/32 :    0
+ip-26-0-172-252:688201:688600 [1] NCCL INFO Channel 26/32 :    0
+ip-26-0-172-252:688201:688600 [1] NCCL INFO Channel 27/32 :    0
+ip-26-0-172-252:688201:688600 [1] NCCL INFO Channel 28/32 :    0
+ip-26-0-172-252:688201:688600 [1] NCCL INFO Channel 29/32 :    0
+ip-26-0-172-252:688201:688600 [1] NCCL INFO Channel 30/32 :    0
+ip-26-0-172-252:688201:688600 [1] NCCL INFO Channel 31/32 :    0
+ip-26-0-172-252:688201:688600 [1] NCCL INFO Trees [0] -1/-1/-1->0->-1 [1] -1/-1/-1->0->-1 [2] -1/-1/-1->0->-1 [3] -1/-1/-1->0->-1 [4] -1/-1/-1->0->-1 [5] -1/-1/-1->0->-1 [6] -1/-1/-1->0->-1 [7] -1/-1/-1->0->-1 [8] -1/-1/-1->0->-1 [9] -1/-1/-1->0->-1 [10] -1/-1/-1->0->-1 [11] -1/-1/-1->0->-1 [12] -1/-1/-1->0->-1 [13] -1/-1/-1->0->-1 [14] -1/-1/-1->0->-1 [15] -1/-1/-1->0->-1 [16] -1/-1/-1->0->-1 [17] -1/-1/-1->0->-1 [18] -1/-1/-1->0->-1 [19] -1/-1/-1->0->-1 [20] -1/-1/-1->0->-1 [21] -1/-1/-1->0->-1 [22] -1/-1/-1->0->-1 [23] -1/-1/-1->0->-1 [24] -1/-1/-1->0->-1 [25] -1/-1/-1->0->-1 [26] -1/-1/-1->0->-1 [27] -1/-1/-1->0->-1 [28] -1/-1/-1->0->-1 [29] -1/-1/-1->0->-1 [30] -1/-1/-1->0->-1 [31] -1/-1/-1->0->-1
+ip-26-0-172-252:688201:688600 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-225:94348:94774 [6] NCCL INFO Channel 00/32 :    0
+ip-26-0-160-225:94348:94774 [6] NCCL INFO Channel 01/32 :    0
+ip-26-0-160-225:94348:94774 [6] NCCL INFO Channel 02/32 :    0
+ip-26-0-160-225:94348:94774 [6] NCCL INFO Channel 03/32 :    0
+ip-26-0-160-225:94348:94774 [6] NCCL INFO Channel 04/32 :    0
+ip-26-0-160-225:94348:94774 [6] NCCL INFO Channel 05/32 :    0
+ip-26-0-160-225:94348:94774 [6] NCCL INFO Channel 06/32 :    0
+ip-26-0-160-225:94348:94774 [6] NCCL INFO Channel 07/32 :    0
+ip-26-0-160-225:94348:94774 [6] NCCL INFO Channel 08/32 :    0
+ip-26-0-160-225:94348:94774 [6] NCCL INFO Channel 09/32 :    0
+ip-26-0-160-225:94348:94774 [6] NCCL INFO Channel 10/32 :    0
+ip-26-0-160-225:94348:94774 [6] NCCL INFO Channel 11/32 :    0
+ip-26-0-160-225:94348:94774 [6] NCCL INFO Channel 12/32 :    0
+ip-26-0-160-225:94348:94774 [6] NCCL INFO Channel 13/32 :    0
+ip-26-0-160-225:94348:94774 [6] NCCL INFO Channel 14/32 :    0
+ip-26-0-160-225:94348:94774 [6] NCCL INFO Channel 15/32 :    0
+ip-26-0-160-225:94348:94774 [6] NCCL INFO Channel 16/32 :    0
+ip-26-0-160-225:94348:94774 [6] NCCL INFO Channel 17/32 :    0
+ip-26-0-160-225:94348:94774 [6] NCCL INFO Channel 18/32 :    0
+ip-26-0-160-225:94348:94774 [6] NCCL INFO Channel 19/32 :    0
+ip-26-0-160-225:94348:94774 [6] NCCL INFO Channel 20/32 :    0
+ip-26-0-160-225:94348:94774 [6] NCCL INFO Channel 21/32 :    0
+ip-26-0-160-225:94348:94774 [6] NCCL INFO Channel 22/32 :    0
+ip-26-0-160-225:94348:94774 [6] NCCL INFO Channel 23/32 :    0
+ip-26-0-160-225:94348:94774 [6] NCCL INFO Channel 24/32 :    0
+ip-26-0-160-225:94348:94774 [6] NCCL INFO Channel 25/32 :    0
+ip-26-0-160-225:94348:94774 [6] NCCL INFO Channel 26/32 :    0
+ip-26-0-160-225:94348:94774 [6] NCCL INFO Channel 27/32 :    0
+ip-26-0-160-225:94348:94774 [6] NCCL INFO Channel 28/32 :    0
+ip-26-0-160-225:94348:94774 [6] NCCL INFO Channel 29/32 :    0
+ip-26-0-160-225:94348:94774 [6] NCCL INFO Channel 30/32 :    0
+ip-26-0-160-225:94348:94774 [6] NCCL INFO Channel 31/32 :    0
+ip-26-0-160-225:94348:94774 [6] NCCL INFO Trees [0] -1/-1/-1->0->-1 [1] -1/-1/-1->0->-1 [2] -1/-1/-1->0->-1 [3] -1/-1/-1->0->-1 [4] -1/-1/-1->0->-1 [5] -1/-1/-1->0->-1 [6] -1/-1/-1->0->-1 [7] -1/-1/-1->0->-1 [8] -1/-1/-1->0->-1 [9] -1/-1/-1->0->-1 [10] -1/-1/-1->0->-1 [11] -1/-1/-1->0->-1 [12] -1/-1/-1->0->-1 [13] -1/-1/-1->0->-1 [14] -1/-1/-1->0->-1 [15] -1/-1/-1->0->-1 [16] -1/-1/-1->0->-1 [17] -1/-1/-1->0->-1 [18] -1/-1/-1->0->-1 [19] -1/-1/-1->0->-1 [20] -1/-1/-1->0->-1 [21] -1/-1/-1->0->-1 [22] -1/-1/-1->0->-1 [23] -1/-1/-1->0->-1 [24] -1/-1/-1->0->-1 [25] -1/-1/-1->0->-1 [26] -1/-1/-1->0->-1 [27] -1/-1/-1->0->-1 [28] -1/-1/-1->0->-1 [29] -1/-1/-1->0->-1 [30] -1/-1/-1->0->-1 [31] -1/-1/-1->0->-1
+ip-26-0-160-225:94348:94774 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-252:688203:688593 [3] NCCL INFO comm 0xa76a540 rank 0 nranks 1 cudaDev 3 nvmlDev 3 busId 86000 commId 0xf66732327b4ac4f5 - Init COMPLETE
+ip-26-0-172-252:688204:688601 [4] NCCL INFO comm 0xaa188f0 rank 0 nranks 1 cudaDev 4 nvmlDev 4 busId 97000 commId 0xf08c6d2609d51c06 - Init COMPLETE
+ip-26-0-172-252:688201:688600 [1] NCCL INFO Connected all rings
+ip-26-0-172-252:688201:688600 [1] NCCL INFO Connected all trees
+ip-26-0-172-252:688201:688600 [1] NCCL INFO 32 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-160-225:94348:94774 [6] NCCL INFO Connected all rings
+ip-26-0-160-225:94348:94774 [6] NCCL INFO Connected all trees
+ip-26-0-160-225:94348:94774 [6] NCCL INFO 32 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-160-225:94344:94778 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-160-225:94344:94778 [2] NCCL INFO Channel 00/32 :    0
+ip-26-0-160-225:94344:94778 [2] NCCL INFO Channel 01/32 :    0
+ip-26-0-160-225:94344:94778 [2] NCCL INFO Channel 02/32 :    0
+ip-26-0-160-225:94344:94778 [2] NCCL INFO Channel 03/32 :    0
+ip-26-0-160-225:94344:94778 [2] NCCL INFO Channel 04/32 :    0
+ip-26-0-160-225:94344:94778 [2] NCCL INFO Channel 05/32 :    0
+ip-26-0-160-225:94344:94778 [2] NCCL INFO Channel 06/32 :    0
+ip-26-0-160-225:94344:94778 [2] NCCL INFO Channel 07/32 :    0
+ip-26-0-160-225:94344:94778 [2] NCCL INFO Channel 08/32 :    0
+ip-26-0-160-225:94344:94778 [2] NCCL INFO Channel 09/32 :    0
+ip-26-0-160-225:94344:94778 [2] NCCL INFO Channel 10/32 :    0
+ip-26-0-160-225:94344:94778 [2] NCCL INFO Channel 11/32 :    0
+ip-26-0-160-225:94344:94778 [2] NCCL INFO Channel 12/32 :    0
+ip-26-0-160-225:94344:94778 [2] NCCL INFO Channel 13/32 :    0
+ip-26-0-160-225:94344:94778 [2] NCCL INFO Channel 14/32 :    0
+ip-26-0-160-225:94344:94778 [2] NCCL INFO Channel 15/32 :    0
+ip-26-0-160-225:94344:94778 [2] NCCL INFO Channel 16/32 :    0
+ip-26-0-160-225:94344:94778 [2] NCCL INFO Channel 17/32 :    0
+ip-26-0-160-225:94344:94778 [2] NCCL INFO Channel 18/32 :    0
+ip-26-0-160-225:94344:94778 [2] NCCL INFO Channel 19/32 :    0
+ip-26-0-160-225:94344:94778 [2] NCCL INFO Channel 20/32 :    0
+ip-26-0-160-225:94344:94778 [2] NCCL INFO Channel 21/32 :    0
+ip-26-0-160-225:94344:94778 [2] NCCL INFO Channel 22/32 :    0
+ip-26-0-160-225:94344:94778 [2] NCCL INFO Channel 23/32 :    0
+ip-26-0-160-225:94344:94778 [2] NCCL INFO Channel 24/32 :    0
+ip-26-0-160-225:94344:94778 [2] NCCL INFO Channel 25/32 :    0
+ip-26-0-160-225:94344:94778 [2] NCCL INFO Channel 26/32 :    0
+ip-26-0-160-225:94344:94778 [2] NCCL INFO Channel 27/32 :    0
+ip-26-0-160-225:94344:94778 [2] NCCL INFO Channel 28/32 :    0
+ip-26-0-160-225:94344:94778 [2] NCCL INFO Channel 29/32 :    0
+ip-26-0-160-225:94344:94778 [2] NCCL INFO Channel 30/32 :    0
+ip-26-0-160-225:94344:94778 [2] NCCL INFO Channel 31/32 :    0
+ip-26-0-160-225:94344:94778 [2] NCCL INFO Trees [0] -1/-1/-1->0->-1 [1] -1/-1/-1->0->-1 [2] -1/-1/-1->0->-1 [3] -1/-1/-1->0->-1 [4] -1/-1/-1->0->-1 [5] -1/-1/-1->0->-1 [6] -1/-1/-1->0->-1 [7] -1/-1/-1->0->-1 [8] -1/-1/-1->0->-1 [9] -1/-1/-1->0->-1 [10] -1/-1/-1->0->-1 [11] -1/-1/-1->0->-1 [12] -1/-1/-1->0->-1 [13] -1/-1/-1->0->-1 [14] -1/-1/-1->0->-1 [15] -1/-1/-1->0->-1 [16] -1/-1/-1->0->-1 [17] -1/-1/-1->0->-1 [18] -1/-1/-1->0->-1 [19] -1/-1/-1->0->-1 [20] -1/-1/-1->0->-1 [21] -1/-1/-1->0->-1 [22] -1/-1/-1->0->-1 [23] -1/-1/-1->0->-1 [24] -1/-1/-1->0->-1 [25] -1/-1/-1->0->-1 [26] -1/-1/-1->0->-1 [27] -1/-1/-1->0->-1 [28] -1/-1/-1->0->-1 [29] -1/-1/-1->0->-1 [30] -1/-1/-1->0->-1 [31] -1/-1/-1->0->-1
+ip-26-0-172-252:688207:688594 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-160-225:94344:94778 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-252:688207:688594 [7] NCCL INFO Channel 00/32 :    0
+ip-26-0-172-252:688207:688594 [7] NCCL INFO Channel 01/32 :    0
+ip-26-0-172-252:688207:688594 [7] NCCL INFO Channel 02/32 :    0
+ip-26-0-172-252:688207:688594 [7] NCCL INFO Channel 03/32 :    0
+ip-26-0-172-252:688207:688594 [7] NCCL INFO Channel 04/32 :    0
+ip-26-0-172-252:688207:688594 [7] NCCL INFO Channel 05/32 :    0
+ip-26-0-172-252:688207:688594 [7] NCCL INFO Channel 06/32 :    0
+ip-26-0-172-252:688207:688594 [7] NCCL INFO Channel 07/32 :    0
+ip-26-0-172-252:688207:688594 [7] NCCL INFO Channel 08/32 :    0
+ip-26-0-172-252:688207:688594 [7] NCCL INFO Channel 09/32 :    0
+ip-26-0-172-252:688207:688594 [7] NCCL INFO Channel 10/32 :    0
+ip-26-0-172-252:688207:688594 [7] NCCL INFO Channel 11/32 :    0
+ip-26-0-172-252:688207:688594 [7] NCCL INFO Channel 12/32 :    0
+ip-26-0-172-252:688207:688594 [7] NCCL INFO Channel 13/32 :    0
+ip-26-0-172-252:688207:688594 [7] NCCL INFO Channel 14/32 :    0
+ip-26-0-172-252:688207:688594 [7] NCCL INFO Channel 15/32 :    0
+ip-26-0-172-252:688207:688594 [7] NCCL INFO Channel 16/32 :    0
+ip-26-0-172-252:688207:688594 [7] NCCL INFO Channel 17/32 :    0
+ip-26-0-172-252:688207:688594 [7] NCCL INFO Channel 18/32 :    0
+ip-26-0-172-252:688207:688594 [7] NCCL INFO Channel 19/32 :    0
+ip-26-0-172-252:688207:688594 [7] NCCL INFO Channel 20/32 :    0
+ip-26-0-172-252:688207:688594 [7] NCCL INFO Channel 21/32 :    0
+ip-26-0-172-252:688207:688594 [7] NCCL INFO Channel 22/32 :    0
+ip-26-0-172-252:688207:688594 [7] NCCL INFO Channel 23/32 :    0
+ip-26-0-172-252:688207:688594 [7] NCCL INFO Channel 24/32 :    0
+ip-26-0-172-252:688207:688594 [7] NCCL INFO Channel 25/32 :    0
+ip-26-0-172-252:688207:688594 [7] NCCL INFO Channel 26/32 :    0
+ip-26-0-172-252:688207:688594 [7] NCCL INFO Channel 27/32 :    0
+ip-26-0-172-252:688207:688594 [7] NCCL INFO Channel 28/32 :    0
+ip-26-0-172-252:688207:688594 [7] NCCL INFO Channel 29/32 :    0
+ip-26-0-172-252:688207:688594 [7] NCCL INFO Channel 30/32 :    0
+ip-26-0-172-252:688207:688594 [7] NCCL INFO Channel 31/32 :    0
+ip-26-0-172-252:688207:688594 [7] NCCL INFO Trees [0] -1/-1/-1->0->-1 [1] -1/-1/-1->0->-1 [2] -1/-1/-1->0->-1 [3] -1/-1/-1->0->-1 [4] -1/-1/-1->0->-1 [5] -1/-1/-1->0->-1 [6] -1/-1/-1->0->-1 [7] -1/-1/-1->0->-1 [8] -1/-1/-1->0->-1 [9] -1/-1/-1->0->-1 [10] -1/-1/-1->0->-1 [11] -1/-1/-1->0->-1 [12] -1/-1/-1->0->-1 [13] -1/-1/-1->0->-1 [14] -1/-1/-1->0->-1 [15] -1/-1/-1->0->-1 [16] -1/-1/-1->0->-1 [17] -1/-1/-1->0->-1 [18] -1/-1/-1->0->-1 [19] -1/-1/-1->0->-1 [20] -1/-1/-1->0->-1 [21] -1/-1/-1->0->-1 [22] -1/-1/-1->0->-1 [23] -1/-1/-1->0->-1 [24] -1/-1/-1->0->-1 [25] -1/-1/-1->0->-1 [26] -1/-1/-1->0->-1 [27] -1/-1/-1->0->-1 [28] -1/-1/-1->0->-1 [29] -1/-1/-1->0->-1 [30] -1/-1/-1->0->-1 [31] -1/-1/-1->0->-1
+ip-26-0-172-252:688207:688594 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-173-7:811253:811670 [5] NCCL INFO comm 0xad45780 rank 0 nranks 1 cudaDev 5 nvmlDev 5 busId a8000 commId 0xade52ce07c1afbd3 - Init COMPLETE
+ip-26-0-160-225:94344:94778 [2] NCCL INFO Connected all rings
+ip-26-0-160-225:94344:94778 [2] NCCL INFO Connected all trees
+ip-26-0-160-225:94344:94778 [2] NCCL INFO 32 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-252:688207:688594 [7] NCCL INFO Connected all rings
+ip-26-0-172-252:688207:688594 [7] NCCL INFO Connected all trees
+ip-26-0-172-252:688207:688594 [7] NCCL INFO 32 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-147:618588:618973 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-172-147:618588:618973 [4] NCCL INFO Channel 00/32 :    0
+ip-26-0-172-147:618588:618973 [4] NCCL INFO Channel 01/32 :    0
+ip-26-0-172-147:618588:618973 [4] NCCL INFO Channel 02/32 :    0
+ip-26-0-172-147:618588:618973 [4] NCCL INFO Channel 03/32 :    0
+ip-26-0-172-147:618588:618973 [4] NCCL INFO Channel 04/32 :    0
+ip-26-0-172-147:618588:618973 [4] NCCL INFO Channel 05/32 :    0
+ip-26-0-172-147:618588:618973 [4] NCCL INFO Channel 06/32 :    0
+ip-26-0-172-147:618588:618973 [4] NCCL INFO Channel 07/32 :    0
+ip-26-0-172-147:618588:618973 [4] NCCL INFO Channel 08/32 :    0
+ip-26-0-172-147:618588:618973 [4] NCCL INFO Channel 09/32 :    0
+ip-26-0-172-147:618588:618973 [4] NCCL INFO Channel 10/32 :    0
+ip-26-0-172-147:618588:618973 [4] NCCL INFO Channel 11/32 :    0
+ip-26-0-172-147:618588:618973 [4] NCCL INFO Channel 12/32 :    0
+ip-26-0-172-147:618588:618973 [4] NCCL INFO Channel 13/32 :    0
+ip-26-0-172-147:618588:618973 [4] NCCL INFO Channel 14/32 :    0
+ip-26-0-172-147:618588:618973 [4] NCCL INFO Channel 15/32 :    0
+ip-26-0-172-147:618588:618973 [4] NCCL INFO Channel 16/32 :    0
+ip-26-0-172-147:618588:618973 [4] NCCL INFO Channel 17/32 :    0
+ip-26-0-172-147:618588:618973 [4] NCCL INFO Channel 18/32 :    0
+ip-26-0-172-147:618588:618973 [4] NCCL INFO Channel 19/32 :    0
+ip-26-0-172-147:618588:618973 [4] NCCL INFO Channel 20/32 :    0
+ip-26-0-172-147:618588:618973 [4] NCCL INFO Channel 21/32 :    0
+ip-26-0-172-147:618588:618973 [4] NCCL INFO Channel 22/32 :    0
+ip-26-0-172-147:618588:618973 [4] NCCL INFO Channel 23/32 :    0
+ip-26-0-172-147:618588:618973 [4] NCCL INFO Channel 24/32 :    0
+ip-26-0-172-147:618588:618973 [4] NCCL INFO Channel 25/32 :    0
+ip-26-0-172-147:618588:618973 [4] NCCL INFO Channel 26/32 :    0
+ip-26-0-172-147:618588:618973 [4] NCCL INFO Channel 27/32 :    0
+ip-26-0-172-147:618588:618973 [4] NCCL INFO Channel 28/32 :    0
+ip-26-0-172-147:618588:618973 [4] NCCL INFO Channel 29/32 :    0
+ip-26-0-172-147:618588:618973 [4] NCCL INFO Channel 30/32 :    0
+ip-26-0-172-147:618588:618973 [4] NCCL INFO Channel 31/32 :    0
+ip-26-0-172-147:618588:618973 [4] NCCL INFO Trees [0] -1/-1/-1->0->-1 [1] -1/-1/-1->0->-1 [2] -1/-1/-1->0->-1 [3] -1/-1/-1->0->-1 [4] -1/-1/-1->0->-1 [5] -1/-1/-1->0->-1 [6] -1/-1/-1->0->-1 [7] -1/-1/-1->0->-1 [8] -1/-1/-1->0->-1 [9] -1/-1/-1->0->-1 [10] -1/-1/-1->0->-1 [11] -1/-1/-1->0->-1 [12] -1/-1/-1->0->-1 [13] -1/-1/-1->0->-1 [14] -1/-1/-1->0->-1 [15] -1/-1/-1->0->-1 [16] -1/-1/-1->0->-1 [17] -1/-1/-1->0->-1 [18] -1/-1/-1->0->-1 [19] -1/-1/-1->0->-1 [20] -1/-1/-1->0->-1 [21] -1/-1/-1->0->-1 [22] -1/-1/-1->0->-1 [23] -1/-1/-1->0->-1 [24] -1/-1/-1->0->-1 [25] -1/-1/-1->0->-1 [26] -1/-1/-1->0->-1 [27] -1/-1/-1->0->-1 [28] -1/-1/-1->0->-1 [29] -1/-1/-1->0->-1 [30] -1/-1/-1->0->-1 [31] -1/-1/-1->0->-1
+ip-26-0-172-147:618588:618973 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-147:618588:618973 [4] NCCL INFO Connected all rings
+ip-26-0-172-147:618588:618973 [4] NCCL INFO Connected all trees
+ip-26-0-172-147:618588:618973 [4] NCCL INFO 32 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-160-225:94345:94779 [3] NCCL INFO comm 0xbba6cb0 rank 0 nranks 1 cudaDev 3 nvmlDev 3 busId 86000 commId 0xc622d7481d9e9256 - Init COMPLETE
+ip-26-0-173-7:811249:811675 [1] NCCL INFO comm 0x9492570 rank 0 nranks 1 cudaDev 1 nvmlDev 1 busId 64000 commId 0x1847af43c68ddd96 - Init COMPLETE
+ip-26-0-173-7:811250:811669 [2] NCCL INFO comm 0xa2918a0 rank 0 nranks 1 cudaDev 2 nvmlDev 2 busId 75000 commId 0x1ccdf40322c55beb - Init COMPLETE
+ip-26-0-160-225:94349:94776 [7] NCCL INFO comm 0xadc7670 rank 0 nranks 1 cudaDev 7 nvmlDev 7 busId ca000 commId 0x6fc54540cf43cf0e - Init COMPLETE
+ip-26-0-172-147:618589:618982 [5] NCCL INFO comm 0xb79cec0 rank 0 nranks 1 cudaDev 5 nvmlDev 5 busId a8000 commId 0xc6c21fed3f236879 - Init COMPLETE
+ip-26-0-172-142:880683:881075 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-172-142:880683:881075 [4] NCCL INFO Channel 00/32 :    0
+ip-26-0-172-142:880683:881075 [4] NCCL INFO Channel 01/32 :    0
+ip-26-0-172-142:880683:881075 [4] NCCL INFO Channel 02/32 :    0
+ip-26-0-172-142:880683:881075 [4] NCCL INFO Channel 03/32 :    0
+ip-26-0-172-142:880683:881075 [4] NCCL INFO Channel 04/32 :    0
+ip-26-0-172-142:880683:881075 [4] NCCL INFO Channel 05/32 :    0
+ip-26-0-172-142:880683:881075 [4] NCCL INFO Channel 06/32 :    0
+ip-26-0-172-142:880683:881075 [4] NCCL INFO Channel 07/32 :    0
+ip-26-0-172-142:880683:881075 [4] NCCL INFO Channel 08/32 :    0
+ip-26-0-172-142:880683:881075 [4] NCCL INFO Channel 09/32 :    0
+ip-26-0-172-142:880683:881075 [4] NCCL INFO Channel 10/32 :    0
+ip-26-0-172-142:880683:881075 [4] NCCL INFO Channel 11/32 :    0
+ip-26-0-172-142:880683:881075 [4] NCCL INFO Channel 12/32 :    0
+ip-26-0-172-142:880683:881075 [4] NCCL INFO Channel 13/32 :    0
+ip-26-0-172-142:880683:881075 [4] NCCL INFO Channel 14/32 :    0
+ip-26-0-172-142:880683:881075 [4] NCCL INFO Channel 15/32 :    0
+ip-26-0-172-142:880683:881075 [4] NCCL INFO Channel 16/32 :    0
+ip-26-0-172-142:880683:881075 [4] NCCL INFO Channel 17/32 :    0
+ip-26-0-172-142:880683:881075 [4] NCCL INFO Channel 18/32 :    0
+ip-26-0-172-142:880683:881075 [4] NCCL INFO Channel 19/32 :    0
+ip-26-0-172-142:880683:881075 [4] NCCL INFO Channel 20/32 :    0
+ip-26-0-172-142:880683:881075 [4] NCCL INFO Channel 21/32 :    0
+ip-26-0-172-142:880683:881075 [4] NCCL INFO Channel 22/32 :    0
+ip-26-0-172-142:880683:881075 [4] NCCL INFO Channel 23/32 :    0
+ip-26-0-172-142:880683:881075 [4] NCCL INFO Channel 24/32 :    0
+ip-26-0-172-142:880683:881075 [4] NCCL INFO Channel 25/32 :    0
+ip-26-0-172-142:880683:881075 [4] NCCL INFO Channel 26/32 :    0
+ip-26-0-172-142:880683:881075 [4] NCCL INFO Channel 27/32 :    0
+ip-26-0-172-142:880683:881075 [4] NCCL INFO Channel 28/32 :    0
+ip-26-0-172-142:880683:881075 [4] NCCL INFO Channel 29/32 :    0
+ip-26-0-172-142:880683:881075 [4] NCCL INFO Channel 30/32 :    0
+ip-26-0-172-142:880683:881075 [4] NCCL INFO Channel 31/32 :    0
+ip-26-0-172-142:880683:881075 [4] NCCL INFO Trees [0] -1/-1/-1->0->-1 [1] -1/-1/-1->0->-1 [2] -1/-1/-1->0->-1 [3] -1/-1/-1->0->-1 [4] -1/-1/-1->0->-1 [5] -1/-1/-1->0->-1 [6] -1/-1/-1->0->-1 [7] -1/-1/-1->0->-1 [8] -1/-1/-1->0->-1 [9] -1/-1/-1->0->-1 [10] -1/-1/-1->0->-1 [11] -1/-1/-1->0->-1 [12] -1/-1/-1->0->-1 [13] -1/-1/-1->0->-1 [14] -1/-1/-1->0->-1 [15] -1/-1/-1->0->-1 [16] -1/-1/-1->0->-1 [17] -1/-1/-1->0->-1 [18] -1/-1/-1->0->-1 [19] -1/-1/-1->0->-1 [20] -1/-1/-1->0->-1 [21] -1/-1/-1->0->-1 [22] -1/-1/-1->0->-1 [23] -1/-1/-1->0->-1 [24] -1/-1/-1->0->-1 [25] -1/-1/-1->0->-1 [26] -1/-1/-1->0->-1 [27] -1/-1/-1->0->-1 [28] -1/-1/-1->0->-1 [29] -1/-1/-1->0->-1 [30] -1/-1/-1->0->-1 [31] -1/-1/-1->0->-1
+ip-26-0-172-142:880683:881075 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-142:880683:881075 [4] NCCL INFO Connected all rings
+ip-26-0-172-142:880683:881075 [4] NCCL INFO Connected all trees
+ip-26-0-172-142:880683:881075 [4] NCCL INFO 32 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-173-7:811252:811674 [4] NCCL INFO comm 0x99ac3a0 rank 0 nranks 1 cudaDev 4 nvmlDev 4 busId 97000 commId 0xaef0ef16b8fac6e4 - Init COMPLETE
+ip-26-0-172-147:618587:618986 [3] NCCL INFO comm 0xab050e0 rank 0 nranks 1 cudaDev 3 nvmlDev 3 busId 86000 commId 0x2aba9a2bb9559f82 - Init COMPLETE
+ip-26-0-172-252:688206:688597 [6] NCCL INFO comm 0xc3c4480 rank 0 nranks 1 cudaDev 6 nvmlDev 6 busId b9000 commId 0x585543c159cd3d5c - Init COMPLETE
+ip-26-0-172-252:688202:688605 [2] NCCL INFO comm 0xb1fdf50 rank 0 nranks 1 cudaDev 2 nvmlDev 2 busId 75000 commId 0xde9c4b90351391a0 - Init COMPLETE
+ip-26-0-160-225:94342:94765 [0] NCCL INFO comm 0xa833970 rank 0 nranks 1 cudaDev 0 nvmlDev 0 busId 53000 commId 0x9318754846b66106 - Init COMPLETE
+ip-26-0-172-147:618586:618976 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-172-147:618586:618976 [2] NCCL INFO Channel 00/32 :    0
+ip-26-0-172-147:618586:618976 [2] NCCL INFO Channel 01/32 :    0
+ip-26-0-172-147:618586:618976 [2] NCCL INFO Channel 02/32 :    0
+ip-26-0-172-142:880682:881070 [3] NCCL INFO comm 0xaa24520 rank 0 nranks 1 cudaDev 3 nvmlDev 3 busId 86000 commId 0x1bb35e59b95df540 - Init COMPLETE
+ip-26-0-172-147:618586:618976 [2] NCCL INFO Channel 03/32 :    0
+ip-26-0-172-147:618586:618976 [2] NCCL INFO Channel 04/32 :    0
+ip-26-0-172-147:618586:618976 [2] NCCL INFO Channel 05/32 :    0
+ip-26-0-172-147:618586:618976 [2] NCCL INFO Channel 06/32 :    0
+ip-26-0-172-147:618586:618976 [2] NCCL INFO Channel 07/32 :    0
+ip-26-0-172-147:618586:618976 [2] NCCL INFO Channel 08/32 :    0
+ip-26-0-172-147:618586:618976 [2] NCCL INFO Channel 09/32 :    0
+ip-26-0-172-147:618586:618976 [2] NCCL INFO Channel 10/32 :    0
+ip-26-0-172-147:618586:618976 [2] NCCL INFO Channel 11/32 :    0
+ip-26-0-172-147:618586:618976 [2] NCCL INFO Channel 12/32 :    0
+ip-26-0-172-147:618586:618976 [2] NCCL INFO Channel 13/32 :    0
+ip-26-0-172-147:618586:618976 [2] NCCL INFO Channel 14/32 :    0
+ip-26-0-172-147:618586:618976 [2] NCCL INFO Channel 15/32 :    0
+ip-26-0-172-147:618586:618976 [2] NCCL INFO Channel 16/32 :    0
+ip-26-0-172-147:618586:618976 [2] NCCL INFO Channel 17/32 :    0
+ip-26-0-172-147:618586:618976 [2] NCCL INFO Channel 18/32 :    0
+ip-26-0-172-147:618586:618976 [2] NCCL INFO Channel 19/32 :    0
+ip-26-0-172-147:618586:618976 [2] NCCL INFO Channel 20/32 :    0
+ip-26-0-172-147:618586:618976 [2] NCCL INFO Channel 21/32 :    0
+ip-26-0-172-147:618586:618976 [2] NCCL INFO Channel 22/32 :    0
+ip-26-0-172-147:618586:618976 [2] NCCL INFO Channel 23/32 :    0
+ip-26-0-172-147:618586:618976 [2] NCCL INFO Channel 24/32 :    0
+ip-26-0-172-147:618586:618976 [2] NCCL INFO Channel 25/32 :    0
+ip-26-0-172-147:618586:618976 [2] NCCL INFO Channel 26/32 :    0
+ip-26-0-172-147:618586:618976 [2] NCCL INFO Channel 27/32 :    0
+ip-26-0-172-147:618586:618976 [2] NCCL INFO Channel 28/32 :    0
+ip-26-0-172-147:618586:618976 [2] NCCL INFO Channel 29/32 :    0
+ip-26-0-172-147:618586:618976 [2] NCCL INFO Channel 30/32 :    0
+ip-26-0-172-147:618586:618976 [2] NCCL INFO Channel 31/32 :    0
+ip-26-0-172-147:618586:618976 [2] NCCL INFO Trees [0] -1/-1/-1->0->-1 [1] -1/-1/-1->0->-1 [2] -1/-1/-1->0->-1 [3] -1/-1/-1->0->-1 [4] -1/-1/-1->0->-1 [5] -1/-1/-1->0->-1 [6] -1/-1/-1->0->-1 [7] -1/-1/-1->0->-1 [8] -1/-1/-1->0->-1 [9] -1/-1/-1->0->-1 [10] -1/-1/-1->0->-1 [11] -1/-1/-1->0->-1 [12] -1/-1/-1->0->-1 [13] -1/-1/-1->0->-1 [14] -1/-1/-1->0->-1 [15] -1/-1/-1->0->-1 [16] -1/-1/-1->0->-1 [17] -1/-1/-1->0->-1 [18] -1/-1/-1->0->-1 [19] -1/-1/-1->0->-1 [20] -1/-1/-1->0->-1 [21] -1/-1/-1->0->-1 [22] -1/-1/-1->0->-1 [23] -1/-1/-1->0->-1 [24] -1/-1/-1->0->-1 [25] -1/-1/-1->0->-1 [26] -1/-1/-1->0->-1 [27] -1/-1/-1->0->-1 [28] -1/-1/-1->0->-1 [29] -1/-1/-1->0->-1 [30] -1/-1/-1->0->-1 [31] -1/-1/-1->0->-1
+ip-26-0-172-147:618586:618976 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-225:94346:94773 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-160-225:94346:94773 [4] NCCL INFO Channel 00/32 :    0
+ip-26-0-160-225:94346:94773 [4] NCCL INFO Channel 01/32 :    0
+ip-26-0-160-225:94346:94773 [4] NCCL INFO Channel 02/32 :    0
+ip-26-0-160-225:94346:94773 [4] NCCL INFO Channel 03/32 :    0
+ip-26-0-160-225:94346:94773 [4] NCCL INFO Channel 04/32 :    0
+ip-26-0-160-225:94346:94773 [4] NCCL INFO Channel 05/32 :    0
+ip-26-0-160-225:94346:94773 [4] NCCL INFO Channel 06/32 :    0
+ip-26-0-160-225:94346:94773 [4] NCCL INFO Channel 07/32 :    0
+ip-26-0-160-225:94346:94773 [4] NCCL INFO Channel 08/32 :    0
+ip-26-0-160-225:94346:94773 [4] NCCL INFO Channel 09/32 :    0
+ip-26-0-160-225:94346:94773 [4] NCCL INFO Channel 10/32 :    0
+ip-26-0-160-225:94346:94773 [4] NCCL INFO Channel 11/32 :    0
+ip-26-0-160-225:94346:94773 [4] NCCL INFO Channel 12/32 :    0
+ip-26-0-160-225:94346:94773 [4] NCCL INFO Channel 13/32 :    0
+ip-26-0-160-225:94346:94773 [4] NCCL INFO Channel 14/32 :    0
+ip-26-0-160-225:94346:94773 [4] NCCL INFO Channel 15/32 :    0
+ip-26-0-160-225:94346:94773 [4] NCCL INFO Channel 16/32 :    0
+ip-26-0-160-225:94346:94773 [4] NCCL INFO Channel 17/32 :    0
+ip-26-0-160-225:94346:94773 [4] NCCL INFO Channel 18/32 :    0
+ip-26-0-160-225:94346:94773 [4] NCCL INFO Channel 19/32 :    0
+ip-26-0-160-225:94346:94773 [4] NCCL INFO Channel 20/32 :    0
+ip-26-0-160-225:94346:94773 [4] NCCL INFO Channel 21/32 :    0
+ip-26-0-160-225:94346:94773 [4] NCCL INFO Channel 22/32 :    0
+ip-26-0-160-225:94346:94773 [4] NCCL INFO Channel 23/32 :    0
+ip-26-0-160-225:94346:94773 [4] NCCL INFO Channel 24/32 :    0
+ip-26-0-160-225:94346:94773 [4] NCCL INFO Channel 25/32 :    0
+ip-26-0-160-225:94346:94773 [4] NCCL INFO Channel 26/32 :    0
+ip-26-0-160-225:94346:94773 [4] NCCL INFO Channel 27/32 :    0
+ip-26-0-160-225:94346:94773 [4] NCCL INFO Channel 28/32 :    0
+ip-26-0-160-225:94346:94773 [4] NCCL INFO Channel 29/32 :    0
+ip-26-0-160-225:94346:94773 [4] NCCL INFO Channel 30/32 :    0
+ip-26-0-160-225:94346:94773 [4] NCCL INFO Channel 31/32 :    0
+ip-26-0-160-225:94346:94773 [4] NCCL INFO Trees [0] -1/-1/-1->0->-1 [1] -1/-1/-1->0->-1 [2] -1/-1/-1->0->-1 [3] -1/-1/-1->0->-1 [4] -1/-1/-1->0->-1 [5] -1/-1/-1->0->-1 [6] -1/-1/-1->0->-1 [7] -1/-1/-1->0->-1 [8] -1/-1/-1->0->-1 [9] -1/-1/-1->0->-1 [10] -1/-1/-1->0->-1 [11] -1/-1/-1->0->-1 [12] -1/-1/-1->0->-1 [13] -1/-1/-1->0->-1 [14] -1/-1/-1->0->-1 [15] -1/-1/-1->0->-1 [16] -1/-1/-1->0->-1 [17] -1/-1/-1->0->-1 [18] -1/-1/-1->0->-1 [19] -1/-1/-1->0->-1 [20] -1/-1/-1->0->-1 [21] -1/-1/-1->0->-1 [22] -1/-1/-1->0->-1 [23] -1/-1/-1->0->-1 [24] -1/-1/-1->0->-1 [25] -1/-1/-1->0->-1 [26] -1/-1/-1->0->-1 [27] -1/-1/-1->0->-1 [28] -1/-1/-1->0->-1 [29] -1/-1/-1->0->-1 [30] -1/-1/-1->0->-1 [31] -1/-1/-1->0->-1
+ip-26-0-160-225:94346:94773 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-147:618586:618976 [2] NCCL INFO Connected all rings
+ip-26-0-172-147:618586:618976 [2] NCCL INFO Connected all trees
+ip-26-0-160-225:94347:94772 [5] NCCL INFO comm 0xaaa1f70 rank 0 nranks 1 cudaDev 5 nvmlDev 5 busId a8000 commId 0x45b2c163d1e10c8f - Init COMPLETE
+ip-26-0-172-147:618586:618976 [2] NCCL INFO 32 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-160-225:94346:94773 [4] NCCL INFO Connected all rings
+ip-26-0-160-225:94346:94773 [4] NCCL INFO Connected all trees
+ip-26-0-160-225:94346:94773 [4] NCCL INFO 32 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-252:688201:688600 [1] NCCL INFO comm 0xa8c27e0 rank 0 nranks 1 cudaDev 1 nvmlDev 1 busId 64000 commId 0x589239cd9357e3d2 - Init COMPLETE
+ip-26-0-172-142:880681:881073 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-172-147:618591:618983 [7] NCCL INFO comm 0xb33b780 rank 0 nranks 1 cudaDev 7 nvmlDev 7 busId ca000 commId 0x43383332916e059c - Init COMPLETE
+ip-26-0-172-252:688207:688594 [7] NCCL INFO comm 0xa4899d0 rank 0 nranks 1 cudaDev 7 nvmlDev 7 busId ca000 commId 0xca84283b6fe59ad8 - Init COMPLETE
+ip-26-0-172-147:618585:618981 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-172-147:618585:618981 [1] NCCL INFO Channel 00/32 :    0
+ip-26-0-172-147:618585:618981 [1] NCCL INFO Channel 01/32 :    0
+ip-26-0-172-147:618585:618981 [1] NCCL INFO Channel 02/32 :    0
+ip-26-0-172-147:618585:618981 [1] NCCL INFO Channel 03/32 :    0
+ip-26-0-172-147:618585:618981 [1] NCCL INFO Channel 04/32 :    0
+ip-26-0-172-147:618585:618981 [1] NCCL INFO Channel 05/32 :    0
+ip-26-0-172-147:618585:618981 [1] NCCL INFO Channel 06/32 :    0
+ip-26-0-172-147:618585:618981 [1] NCCL INFO Channel 07/32 :    0
+ip-26-0-172-147:618585:618981 [1] NCCL INFO Channel 08/32 :    0
+ip-26-0-172-147:618585:618981 [1] NCCL INFO Channel 09/32 :    0
+ip-26-0-172-147:618585:618981 [1] NCCL INFO Channel 10/32 :    0
+ip-26-0-172-147:618585:618981 [1] NCCL INFO Channel 11/32 :    0
+ip-26-0-172-147:618585:618981 [1] NCCL INFO Channel 12/32 :    0
+ip-26-0-172-147:618585:618981 [1] NCCL INFO Channel 13/32 :    0
+ip-26-0-172-147:618585:618981 [1] NCCL INFO Channel 14/32 :    0
+ip-26-0-172-147:618585:618981 [1] NCCL INFO Channel 15/32 :    0
+ip-26-0-172-147:618585:618981 [1] NCCL INFO Channel 16/32 :    0
+ip-26-0-172-147:618585:618981 [1] NCCL INFO Channel 17/32 :    0
+ip-26-0-172-147:618585:618981 [1] NCCL INFO Channel 18/32 :    0
+ip-26-0-172-147:618585:618981 [1] NCCL INFO Channel 19/32 :    0
+ip-26-0-172-147:618585:618981 [1] NCCL INFO Channel 20/32 :    0
+ip-26-0-172-147:618585:618981 [1] NCCL INFO Channel 21/32 :    0
+ip-26-0-172-147:618585:618981 [1] NCCL INFO Channel 22/32 :    0
+ip-26-0-172-147:618585:618981 [1] NCCL INFO Channel 23/32 :    0
+ip-26-0-172-147:618585:618981 [1] NCCL INFO Channel 24/32 :    0
+ip-26-0-172-147:618585:618981 [1] NCCL INFO Channel 25/32 :    0
+ip-26-0-172-147:618585:618981 [1] NCCL INFO Channel 26/32 :    0
+ip-26-0-172-147:618585:618981 [1] NCCL INFO Channel 27/32 :    0
+ip-26-0-172-147:618585:618981 [1] NCCL INFO Channel 28/32 :    0
+ip-26-0-172-147:618585:618981 [1] NCCL INFO Channel 29/32 :    0
+ip-26-0-172-147:618585:618981 [1] NCCL INFO Channel 30/32 :    0
+ip-26-0-172-147:618585:618981 [1] NCCL INFO Channel 31/32 :    0
+ip-26-0-172-147:618585:618981 [1] NCCL INFO Trees [0] -1/-1/-1->0->-1 [1] -1/-1/-1->0->-1 [2] -1/-1/-1->0->-1 [3] -1/-1/-1->0->-1 [4] -1/-1/-1->0->-1 [5] -1/-1/-1->0->-1 [6] -1/-1/-1->0->-1 [7] -1/-1/-1->0->-1 [8] -1/-1/-1->0->-1 [9] -1/-1/-1->0->-1 [10] -1/-1/-1->0->-1 [11] -1/-1/-1->0->-1 [12] -1/-1/-1->0->-1 [13] -1/-1/-1->0->-1 [14] -1/-1/-1->0->-1 [15] -1/-1/-1->0->-1 [16] -1/-1/-1->0->-1 [17] -1/-1/-1->0->-1 [18] -1/-1/-1->0->-1 [19] -1/-1/-1->0->-1 [20] -1/-1/-1->0->-1 [21] -1/-1/-1->0->-1 [22] -1/-1/-1->0->-1 [23] -1/-1/-1->0->-1 [24] -1/-1/-1->0->-1 [25] -1/-1/-1->0->-1 [26] -1/-1/-1->0->-1 [27] -1/-1/-1->0->-1 [28] -1/-1/-1->0->-1 [29] -1/-1/-1->0->-1 [30] -1/-1/-1->0->-1 [31] -1/-1/-1->0->-1
+ip-26-0-172-147:618585:618981 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-142:880685:881067 [6] NCCL INFO comm 0x9e7b6c0 rank 0 nranks 1 cudaDev 6 nvmlDev 6 busId b9000 commId 0x228a516b849b7aa0 - Init COMPLETE
+ip-26-0-172-142:880681:881073 [2] NCCL INFO Channel 00/32 :    0
+ip-26-0-172-142:880681:881073 [2] NCCL INFO Channel 01/32 :    0
+ip-26-0-172-142:880681:881073 [2] NCCL INFO Channel 02/32 :    0
+ip-26-0-172-142:880681:881073 [2] NCCL INFO Channel 03/32 :    0
+ip-26-0-172-142:880681:881073 [2] NCCL INFO Channel 04/32 :    0
+ip-26-0-172-142:880681:881073 [2] NCCL INFO Channel 05/32 :    0
+ip-26-0-172-142:880681:881073 [2] NCCL INFO Channel 06/32 :    0
+ip-26-0-172-142:880681:881073 [2] NCCL INFO Channel 07/32 :    0
+ip-26-0-172-142:880681:881073 [2] NCCL INFO Channel 08/32 :    0
+ip-26-0-172-142:880681:881073 [2] NCCL INFO Channel 09/32 :    0
+ip-26-0-172-142:880681:881073 [2] NCCL INFO Channel 10/32 :    0
+ip-26-0-172-142:880681:881073 [2] NCCL INFO Channel 11/32 :    0
+ip-26-0-172-142:880681:881073 [2] NCCL INFO Channel 12/32 :    0
+ip-26-0-172-142:880681:881073 [2] NCCL INFO Channel 13/32 :    0
+ip-26-0-172-142:880681:881073 [2] NCCL INFO Channel 14/32 :    0
+ip-26-0-172-142:880681:881073 [2] NCCL INFO Channel 15/32 :    0
+ip-26-0-172-142:880681:881073 [2] NCCL INFO Channel 16/32 :    0
+ip-26-0-172-142:880681:881073 [2] NCCL INFO Channel 17/32 :    0
+ip-26-0-172-142:880681:881073 [2] NCCL INFO Channel 18/32 :    0
+ip-26-0-172-142:880681:881073 [2] NCCL INFO Channel 19/32 :    0
+ip-26-0-172-142:880681:881073 [2] NCCL INFO Channel 20/32 :    0
+ip-26-0-172-142:880681:881073 [2] NCCL INFO Channel 21/32 :    0
+ip-26-0-172-142:880681:881073 [2] NCCL INFO Channel 22/32 :    0
+ip-26-0-172-142:880681:881073 [2] NCCL INFO Channel 23/32 :    0
+ip-26-0-172-142:880681:881073 [2] NCCL INFO Channel 24/32 :    0
+ip-26-0-172-142:880681:881073 [2] NCCL INFO Channel 25/32 :    0
+ip-26-0-172-142:880681:881073 [2] NCCL INFO Channel 26/32 :    0
+ip-26-0-172-142:880681:881073 [2] NCCL INFO Channel 27/32 :    0
+ip-26-0-172-142:880681:881073 [2] NCCL INFO Channel 28/32 :    0
+ip-26-0-172-142:880681:881073 [2] NCCL INFO Channel 29/32 :    0
+ip-26-0-172-142:880681:881073 [2] NCCL INFO Channel 30/32 :    0
+ip-26-0-172-142:880681:881073 [2] NCCL INFO Channel 31/32 :    0
+ip-26-0-172-147:618585:618981 [1] NCCL INFO Connected all rings
+ip-26-0-172-147:618585:618981 [1] NCCL INFO Connected all trees
+ip-26-0-172-147:618585:618981 [1] NCCL INFO 32 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-142:880681:881073 [2] NCCL INFO Trees [0] -1/-1/-1->0->-1 [1] -1/-1/-1->0->-1 [2] -1/-1/-1->0->-1 [3] -1/-1/-1->0->-1 [4] -1/-1/-1->0->-1 [5] -1/-1/-1->0->-1 [6] -1/-1/-1->0->-1 [7] -1/-1/-1->0->-1 [8] -1/-1/-1->0->-1 [9] -1/-1/-1->0->-1 [10] -1/-1/-1->0->-1 [11] -1/-1/-1->0->-1 [12] -1/-1/-1->0->-1 [13] -1/-1/-1->0->-1 [14] -1/-1/-1->0->-1 [15] -1/-1/-1->0->-1 [16] -1/-1/-1->0->-1 [17] -1/-1/-1->0->-1 [18] -1/-1/-1->0->-1 [19] -1/-1/-1->0->-1 [20] -1/-1/-1->0->-1 [21] -1/-1/-1->0->-1 [22] -1/-1/-1->0->-1 [23] -1/-1/-1->0->-1 [24] -1/-1/-1->0->-1 [25] -1/-1/-1->0->-1 [26] -1/-1/-1->0->-1 [27] -1/-1/-1->0->-1 [28] -1/-1/-1->0->-1 [29] -1/-1/-1->0->-1 [30] -1/-1/-1->0->-1 [31] -1/-1/-1->0->-1
+ip-26-0-172-142:880681:881073 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-142:880681:881073 [2] NCCL INFO Connected all rings
+ip-26-0-172-142:880681:881073 [2] NCCL INFO Connected all trees
+ip-26-0-172-142:880681:881073 [2] NCCL INFO 32 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-160-225:94348:94774 [6] NCCL INFO comm 0xbbc1ca0 rank 0 nranks 1 cudaDev 6 nvmlDev 6 busId b9000 commId 0x85028ee211c854d1 - Init COMPLETE
+ip-26-0-172-147:618590:618984 [6] NCCL INFO comm 0xb6cdef0 rank 0 nranks 1 cudaDev 6 nvmlDev 6 busId b9000 commId 0x4638be3954c599a5 - Init COMPLETE
+ip-26-0-160-225:94344:94778 [2] NCCL INFO comm 0xa9c7db0 rank 0 nranks 1 cudaDev 2 nvmlDev 2 busId 75000 commId 0x7ff16fe4a011ddc8 - Init COMPLETE
+ip-26-0-172-147:618588:618973 [4] NCCL INFO comm 0x9d6df70 rank 0 nranks 1 cudaDev 4 nvmlDev 4 busId 97000 commId 0xa14c5a4547fa1c34 - Init COMPLETE
+ip-26-0-172-142:880680:881074 [1] NCCL INFO comm 0xaf10cc0 rank 0 nranks 1 cudaDev 1 nvmlDev 1 busId 64000 commId 0x9b9386d3316183bc - Init COMPLETE
+ip-26-0-172-142:880686:881081 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-172-142:880686:881081 [7] NCCL INFO Channel 00/32 :    0
+ip-26-0-172-142:880686:881081 [7] NCCL INFO Channel 01/32 :    0
+ip-26-0-172-142:880686:881081 [7] NCCL INFO Channel 02/32 :    0
+ip-26-0-172-142:880686:881081 [7] NCCL INFO Channel 03/32 :    0
+ip-26-0-172-142:880686:881081 [7] NCCL INFO Channel 04/32 :    0
+ip-26-0-172-142:880686:881081 [7] NCCL INFO Channel 05/32 :    0
+ip-26-0-172-142:880686:881081 [7] NCCL INFO Channel 06/32 :    0
+ip-26-0-172-142:880686:881081 [7] NCCL INFO Channel 07/32 :    0
+ip-26-0-172-142:880686:881081 [7] NCCL INFO Channel 08/32 :    0
+ip-26-0-172-142:880686:881081 [7] NCCL INFO Channel 09/32 :    0
+ip-26-0-172-142:880686:881081 [7] NCCL INFO Channel 10/32 :    0
+ip-26-0-172-142:880686:881081 [7] NCCL INFO Channel 11/32 :    0
+ip-26-0-172-142:880686:881081 [7] NCCL INFO Channel 12/32 :    0
+ip-26-0-172-142:880686:881081 [7] NCCL INFO Channel 13/32 :    0
+ip-26-0-172-142:880686:881081 [7] NCCL INFO Channel 14/32 :    0
+ip-26-0-172-142:880686:881081 [7] NCCL INFO Channel 15/32 :    0
+ip-26-0-172-142:880686:881081 [7] NCCL INFO Channel 16/32 :    0
+ip-26-0-172-142:880686:881081 [7] NCCL INFO Channel 17/32 :    0
+ip-26-0-172-142:880686:881081 [7] NCCL INFO Channel 18/32 :    0
+ip-26-0-172-142:880686:881081 [7] NCCL INFO Channel 19/32 :    0
+ip-26-0-172-142:880686:881081 [7] NCCL INFO Channel 20/32 :    0
+ip-26-0-172-142:880686:881081 [7] NCCL INFO Channel 21/32 :    0
+ip-26-0-172-142:880686:881081 [7] NCCL INFO Channel 22/32 :    0
+ip-26-0-172-142:880686:881081 [7] NCCL INFO Channel 23/32 :    0
+ip-26-0-172-142:880686:881081 [7] NCCL INFO Channel 24/32 :    0
+ip-26-0-172-142:880686:881081 [7] NCCL INFO Channel 25/32 :    0
+ip-26-0-172-142:880686:881081 [7] NCCL INFO Channel 26/32 :    0
+ip-26-0-172-142:880686:881081 [7] NCCL INFO Channel 27/32 :    0
+ip-26-0-172-142:880686:881081 [7] NCCL INFO Channel 28/32 :    0
+ip-26-0-172-142:880686:881081 [7] NCCL INFO Channel 29/32 :    0
+ip-26-0-172-142:880686:881081 [7] NCCL INFO Channel 30/32 :    0
+ip-26-0-172-142:880686:881081 [7] NCCL INFO Channel 31/32 :    0
+ip-26-0-172-142:880686:881081 [7] NCCL INFO Trees [0] -1/-1/-1->0->-1 [1] -1/-1/-1->0->-1 [2] -1/-1/-1->0->-1 [3] -1/-1/-1->0->-1 [4] -1/-1/-1->0->-1 [5] -1/-1/-1->0->-1 [6] -1/-1/-1->0->-1 [7] -1/-1/-1->0->-1 [8] -1/-1/-1->0->-1 [9] -1/-1/-1->0->-1 [10] -1/-1/-1->0->-1 [11] -1/-1/-1->0->-1 [12] -1/-1/-1->0->-1 [13] -1/-1/-1->0->-1 [14] -1/-1/-1->0->-1 [15] -1/-1/-1->0->-1 [16] -1/-1/-1->0->-1 [17] -1/-1/-1->0->-1 [18] -1/-1/-1->0->-1 [19] -1/-1/-1->0->-1 [20] -1/-1/-1->0->-1 [21] -1/-1/-1->0->-1 [22] -1/-1/-1->0->-1 [23] -1/-1/-1->0->-1 [24] -1/-1/-1->0->-1 [25] -1/-1/-1->0->-1 [26] -1/-1/-1->0->-1 [27] -1/-1/-1->0->-1 [28] -1/-1/-1->0->-1 [29] -1/-1/-1->0->-1 [30] -1/-1/-1->0->-1 [31] -1/-1/-1->0->-1
+ip-26-0-172-142:880686:881081 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-142:880686:881081 [7] NCCL INFO Connected all rings
+ip-26-0-172-142:880686:881081 [7] NCCL INFO Connected all trees
+ip-26-0-172-142:880686:881081 [7] NCCL INFO 32 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-147:618586:618976 [2] NCCL INFO comm 0xa062940 rank 0 nranks 1 cudaDev 2 nvmlDev 2 busId 75000 commId 0xcb2298495ca2a828 - Init COMPLETE
+ip-26-0-160-225:94346:94773 [4] NCCL INFO comm 0xba0b090 rank 0 nranks 1 cudaDev 4 nvmlDev 4 busId 97000 commId 0x39ab8cb2ec22973c - Init COMPLETE
+ip-26-0-172-142:880683:881075 [4] NCCL INFO comm 0xbbf8100 rank 0 nranks 1 cudaDev 4 nvmlDev 4 busId 97000 commId 0xdbf3daa2bdee12e3 - Init COMPLETE
+ip-26-0-172-142:880684:881078 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-172-147:618585:618981 [1] NCCL INFO comm 0xb5314d0 rank 0 nranks 1 cudaDev 1 nvmlDev 1 busId 64000 commId 0xbad86dca646ea35b - Init COMPLETE
+ip-26-0-172-142:880684:881078 [5] NCCL INFO Channel 00/32 :    0
+ip-26-0-172-142:880684:881078 [5] NCCL INFO Channel 01/32 :    0
+ip-26-0-172-142:880684:881078 [5] NCCL INFO Channel 02/32 :    0
+ip-26-0-172-142:880684:881078 [5] NCCL INFO Channel 03/32 :    0
+ip-26-0-172-142:880684:881078 [5] NCCL INFO Channel 04/32 :    0
+ip-26-0-172-142:880684:881078 [5] NCCL INFO Channel 05/32 :    0
+ip-26-0-172-142:880684:881078 [5] NCCL INFO Channel 06/32 :    0
+ip-26-0-172-142:880684:881078 [5] NCCL INFO Channel 07/32 :    0
+ip-26-0-172-142:880684:881078 [5] NCCL INFO Channel 08/32 :    0
+ip-26-0-172-142:880684:881078 [5] NCCL INFO Channel 09/32 :    0
+ip-26-0-172-142:880684:881078 [5] NCCL INFO Channel 10/32 :    0
+ip-26-0-172-142:880684:881078 [5] NCCL INFO Channel 11/32 :    0
+ip-26-0-172-142:880684:881078 [5] NCCL INFO Channel 12/32 :    0
+ip-26-0-172-142:880684:881078 [5] NCCL INFO Channel 13/32 :    0
+ip-26-0-172-142:880684:881078 [5] NCCL INFO Channel 14/32 :    0
+ip-26-0-172-142:880684:881078 [5] NCCL INFO Channel 15/32 :    0
+ip-26-0-172-142:880684:881078 [5] NCCL INFO Channel 16/32 :    0
+ip-26-0-172-142:880684:881078 [5] NCCL INFO Channel 17/32 :    0
+ip-26-0-172-142:880684:881078 [5] NCCL INFO Channel 18/32 :    0
+ip-26-0-172-142:880684:881078 [5] NCCL INFO Channel 19/32 :    0
+ip-26-0-172-142:880684:881078 [5] NCCL INFO Channel 20/32 :    0
+ip-26-0-172-142:880684:881078 [5] NCCL INFO Channel 21/32 :    0
+ip-26-0-172-142:880684:881078 [5] NCCL INFO Channel 22/32 :    0
+ip-26-0-172-142:880684:881078 [5] NCCL INFO Channel 23/32 :    0
+ip-26-0-172-142:880684:881078 [5] NCCL INFO Channel 24/32 :    0
+ip-26-0-172-142:880684:881078 [5] NCCL INFO Channel 25/32 :    0
+ip-26-0-172-142:880684:881078 [5] NCCL INFO Channel 26/32 :    0
+ip-26-0-172-142:880684:881078 [5] NCCL INFO Channel 27/32 :    0
+ip-26-0-172-142:880684:881078 [5] NCCL INFO Channel 28/32 :    0
+ip-26-0-172-142:880684:881078 [5] NCCL INFO Channel 29/32 :    0
+ip-26-0-172-142:880684:881078 [5] NCCL INFO Channel 30/32 :    0
+ip-26-0-172-142:880684:881078 [5] NCCL INFO Channel 31/32 :    0
+ip-26-0-172-142:880684:881078 [5] NCCL INFO Trees [0] -1/-1/-1->0->-1 [1] -1/-1/-1->0->-1 [2] -1/-1/-1->0->-1 [3] -1/-1/-1->0->-1 [4] -1/-1/-1->0->-1 [5] -1/-1/-1->0->-1 [6] -1/-1/-1->0->-1 [7] -1/-1/-1->0->-1 [8] -1/-1/-1->0->-1 [9] -1/-1/-1->0->-1 [10] -1/-1/-1->0->-1 [11] -1/-1/-1->0->-1 [12] -1/-1/-1->0->-1 [13] -1/-1/-1->0->-1 [14] -1/-1/-1->0->-1 [15] -1/-1/-1->0->-1 [16] -1/-1/-1->0->-1 [17] -1/-1/-1->0->-1 [18] -1/-1/-1->0->-1 [19] -1/-1/-1->0->-1 [20] -1/-1/-1->0->-1 [21] -1/-1/-1->0->-1 [22] -1/-1/-1->0->-1 [23] -1/-1/-1->0->-1 [24] -1/-1/-1->0->-1 [25] -1/-1/-1->0->-1 [26] -1/-1/-1->0->-1 [27] -1/-1/-1->0->-1 [28] -1/-1/-1->0->-1 [29] -1/-1/-1->0->-1 [30] -1/-1/-1->0->-1 [31] -1/-1/-1->0->-1
+ip-26-0-172-142:880684:881078 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-142:880684:881078 [5] NCCL INFO Connected all rings
+ip-26-0-172-142:880684:881078 [5] NCCL INFO Connected all trees
+ip-26-0-172-142:880684:881078 [5] NCCL INFO 32 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-142:880681:881073 [2] NCCL INFO comm 0xafd8200 rank 0 nranks 1 cudaDev 2 nvmlDev 2 busId 75000 commId 0x64df22d9c7d488ab - Init COMPLETE
+ip-26-0-172-142:880686:881081 [7] NCCL INFO comm 0xac72420 rank 0 nranks 1 cudaDev 7 nvmlDev 7 busId ca000 commId 0x3380cb88e352cfa - Init COMPLETE
+ip-26-0-172-142:880684:881078 [5] NCCL INFO comm 0x9fc7470 rank 0 nranks 1 cudaDev 5 nvmlDev 5 busId a8000 commId 0x2ea4053c729e895e - Init COMPLETE
+ip-26-0-172-57:942615:943007 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-172-57:942615:943007 [4] NCCL INFO Channel 00/32 :    0
+ip-26-0-172-57:942615:943007 [4] NCCL INFO Channel 01/32 :    0
+ip-26-0-172-57:942615:943007 [4] NCCL INFO Channel 02/32 :    0
+ip-26-0-172-57:942615:943007 [4] NCCL INFO Channel 03/32 :    0
+ip-26-0-172-57:942615:943007 [4] NCCL INFO Channel 04/32 :    0
+ip-26-0-172-57:942615:943007 [4] NCCL INFO Channel 05/32 :    0
+ip-26-0-172-57:942615:943007 [4] NCCL INFO Channel 06/32 :    0
+ip-26-0-172-57:942615:943007 [4] NCCL INFO Channel 07/32 :    0
+ip-26-0-172-57:942615:943007 [4] NCCL INFO Channel 08/32 :    0
+ip-26-0-172-57:942615:943007 [4] NCCL INFO Channel 09/32 :    0
+ip-26-0-172-57:942615:943007 [4] NCCL INFO Channel 10/32 :    0
+ip-26-0-172-57:942615:943007 [4] NCCL INFO Channel 11/32 :    0
+ip-26-0-172-57:942615:943007 [4] NCCL INFO Channel 12/32 :    0
+ip-26-0-172-57:942615:943007 [4] NCCL INFO Channel 13/32 :    0
+ip-26-0-172-57:942615:943007 [4] NCCL INFO Channel 14/32 :    0
+ip-26-0-172-57:942615:943007 [4] NCCL INFO Channel 15/32 :    0
+ip-26-0-172-57:942615:943007 [4] NCCL INFO Channel 16/32 :    0
+ip-26-0-172-57:942615:943007 [4] NCCL INFO Channel 17/32 :    0
+ip-26-0-172-57:942615:943007 [4] NCCL INFO Channel 18/32 :    0
+ip-26-0-172-57:942615:943007 [4] NCCL INFO Channel 19/32 :    0
+ip-26-0-172-57:942615:943007 [4] NCCL INFO Channel 20/32 :    0
+ip-26-0-172-57:942615:943007 [4] NCCL INFO Channel 21/32 :    0
+ip-26-0-172-57:942615:943007 [4] NCCL INFO Channel 22/32 :    0
+ip-26-0-172-57:942615:943007 [4] NCCL INFO Channel 23/32 :    0
+ip-26-0-172-57:942615:943007 [4] NCCL INFO Channel 24/32 :    0
+ip-26-0-172-57:942615:943007 [4] NCCL INFO Channel 25/32 :    0
+ip-26-0-172-57:942615:943007 [4] NCCL INFO Channel 26/32 :    0
+ip-26-0-172-57:942615:943007 [4] NCCL INFO Channel 27/32 :    0
+ip-26-0-172-57:942615:943007 [4] NCCL INFO Channel 28/32 :    0
+ip-26-0-172-57:942615:943007 [4] NCCL INFO Channel 29/32 :    0
+ip-26-0-172-57:942615:943007 [4] NCCL INFO Channel 30/32 :    0
+ip-26-0-172-57:942615:943007 [4] NCCL INFO Channel 31/32 :    0
+ip-26-0-172-57:942615:943007 [4] NCCL INFO Trees [0] -1/-1/-1->0->-1 [1] -1/-1/-1->0->-1 [2] -1/-1/-1->0->-1 [3] -1/-1/-1->0->-1 [4] -1/-1/-1->0->-1 [5] -1/-1/-1->0->-1 [6] -1/-1/-1->0->-1 [7] -1/-1/-1->0->-1 [8] -1/-1/-1->0->-1 [9] -1/-1/-1->0->-1 [10] -1/-1/-1->0->-1 [11] -1/-1/-1->0->-1 [12] -1/-1/-1->0->-1 [13] -1/-1/-1->0->-1 [14] -1/-1/-1->0->-1 [15] -1/-1/-1->0->-1 [16] -1/-1/-1->0->-1 [17] -1/-1/-1->0->-1 [18] -1/-1/-1->0->-1 [19] -1/-1/-1->0->-1 [20] -1/-1/-1->0->-1 [21] -1/-1/-1->0->-1 [22] -1/-1/-1->0->-1 [23] -1/-1/-1->0->-1 [24] -1/-1/-1->0->-1 [25] -1/-1/-1->0->-1 [26] -1/-1/-1->0->-1 [27] -1/-1/-1->0->-1 [28] -1/-1/-1->0->-1 [29] -1/-1/-1->0->-1 [30] -1/-1/-1->0->-1 [31] -1/-1/-1->0->-1
+ip-26-0-172-57:942615:943007 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-57:942615:943007 [4] NCCL INFO Connected all rings
+ip-26-0-172-57:942615:943007 [4] NCCL INFO Connected all trees
+ip-26-0-172-57:942615:943007 [4] NCCL INFO 32 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-57:942616:943005 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-172-57:942616:943005 [5] NCCL INFO Channel 00/32 :    0
+ip-26-0-172-57:942616:943005 [5] NCCL INFO Channel 01/32 :    0
+ip-26-0-172-57:942616:943005 [5] NCCL INFO Channel 02/32 :    0
+ip-26-0-172-57:942616:943005 [5] NCCL INFO Channel 03/32 :    0
+ip-26-0-172-57:942616:943005 [5] NCCL INFO Channel 04/32 :    0
+ip-26-0-172-57:942616:943005 [5] NCCL INFO Channel 05/32 :    0
+ip-26-0-172-57:942616:943005 [5] NCCL INFO Channel 06/32 :    0
+ip-26-0-172-57:942616:943005 [5] NCCL INFO Channel 07/32 :    0
+ip-26-0-172-57:942616:943005 [5] NCCL INFO Channel 08/32 :    0
+ip-26-0-172-57:942616:943005 [5] NCCL INFO Channel 09/32 :    0
+ip-26-0-172-57:942616:943005 [5] NCCL INFO Channel 10/32 :    0
+ip-26-0-172-57:942616:943005 [5] NCCL INFO Channel 11/32 :    0
+ip-26-0-172-57:942616:943005 [5] NCCL INFO Channel 12/32 :    0
+ip-26-0-172-57:942616:943005 [5] NCCL INFO Channel 13/32 :    0
+ip-26-0-172-57:942616:943005 [5] NCCL INFO Channel 14/32 :    0
+ip-26-0-172-57:942616:943005 [5] NCCL INFO Channel 15/32 :    0
+ip-26-0-172-57:942616:943005 [5] NCCL INFO Channel 16/32 :    0
+ip-26-0-172-57:942616:943005 [5] NCCL INFO Channel 17/32 :    0
+ip-26-0-172-57:942616:943005 [5] NCCL INFO Channel 18/32 :    0
+ip-26-0-172-57:942616:943005 [5] NCCL INFO Channel 19/32 :    0
+ip-26-0-172-57:942616:943005 [5] NCCL INFO Channel 20/32 :    0
+ip-26-0-172-57:942616:943005 [5] NCCL INFO Channel 21/32 :    0
+ip-26-0-172-57:942616:943005 [5] NCCL INFO Channel 22/32 :    0
+ip-26-0-172-57:942616:943005 [5] NCCL INFO Channel 23/32 :    0
+ip-26-0-172-57:942616:943005 [5] NCCL INFO Channel 24/32 :    0
+ip-26-0-172-57:942616:943005 [5] NCCL INFO Channel 25/32 :    0
+ip-26-0-172-57:942616:943005 [5] NCCL INFO Channel 26/32 :    0
+ip-26-0-172-57:942616:943005 [5] NCCL INFO Channel 27/32 :    0
+ip-26-0-172-57:942616:943005 [5] NCCL INFO Channel 28/32 :    0
+ip-26-0-172-57:942616:943005 [5] NCCL INFO Channel 29/32 :    0
+ip-26-0-172-57:942616:943005 [5] NCCL INFO Channel 30/32 :    0
+ip-26-0-172-57:942616:943005 [5] NCCL INFO Channel 31/32 :    0
+ip-26-0-172-57:942616:943005 [5] NCCL INFO Trees [0] -1/-1/-1->0->-1 [1] -1/-1/-1->0->-1 [2] -1/-1/-1->0->-1 [3] -1/-1/-1->0->-1 [4] -1/-1/-1->0->-1 [5] -1/-1/-1->0->-1 [6] -1/-1/-1->0->-1 [7] -1/-1/-1->0->-1 [8] -1/-1/-1->0->-1 [9] -1/-1/-1->0->-1 [10] -1/-1/-1->0->-1 [11] -1/-1/-1->0->-1 [12] -1/-1/-1->0->-1 [13] -1/-1/-1->0->-1 [14] -1/-1/-1->0->-1 [15] -1/-1/-1->0->-1 [16] -1/-1/-1->0->-1 [17] -1/-1/-1->0->-1 [18] -1/-1/-1->0->-1 [19] -1/-1/-1->0->-1 [20] -1/-1/-1->0->-1 [21] -1/-1/-1->0->-1 [22] -1/-1/-1->0->-1 [23] -1/-1/-1->0->-1 [24] -1/-1/-1->0->-1 [25] -1/-1/-1->0->-1 [26] -1/-1/-1->0->-1 [27] -1/-1/-1->0->-1 [28] -1/-1/-1->0->-1 [29] -1/-1/-1->0->-1 [30] -1/-1/-1->0->-1 [31] -1/-1/-1->0->-1
+ip-26-0-172-57:942616:943005 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-57:942616:943005 [5] NCCL INFO Connected all rings
+ip-26-0-172-57:942616:943005 [5] NCCL INFO Connected all trees
+ip-26-0-172-57:942616:943005 [5] NCCL INFO 32 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-57:942618:943010 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-172-57:942618:943010 [7] NCCL INFO Channel 00/32 :    0
+ip-26-0-172-57:942618:943010 [7] NCCL INFO Channel 01/32 :    0
+ip-26-0-172-57:942618:943010 [7] NCCL INFO Channel 02/32 :    0
+ip-26-0-172-57:942618:943010 [7] NCCL INFO Channel 03/32 :    0
+ip-26-0-172-57:942618:943010 [7] NCCL INFO Channel 04/32 :    0
+ip-26-0-172-57:942618:943010 [7] NCCL INFO Channel 05/32 :    0
+ip-26-0-172-57:942618:943010 [7] NCCL INFO Channel 06/32 :    0
+ip-26-0-172-57:942618:943010 [7] NCCL INFO Channel 07/32 :    0
+ip-26-0-172-57:942618:943010 [7] NCCL INFO Channel 08/32 :    0
+ip-26-0-172-57:942618:943010 [7] NCCL INFO Channel 09/32 :    0
+ip-26-0-172-57:942618:943010 [7] NCCL INFO Channel 10/32 :    0
+ip-26-0-172-57:942618:943010 [7] NCCL INFO Channel 11/32 :    0
+ip-26-0-172-57:942618:943010 [7] NCCL INFO Channel 12/32 :    0
+ip-26-0-172-57:942618:943010 [7] NCCL INFO Channel 13/32 :    0
+ip-26-0-172-57:942618:943010 [7] NCCL INFO Channel 14/32 :    0
+ip-26-0-172-57:942618:943010 [7] NCCL INFO Channel 15/32 :    0
+ip-26-0-172-57:942618:943010 [7] NCCL INFO Channel 16/32 :    0
+ip-26-0-172-57:942618:943010 [7] NCCL INFO Channel 17/32 :    0
+ip-26-0-172-57:942618:943010 [7] NCCL INFO Channel 18/32 :    0
+ip-26-0-172-57:942618:943010 [7] NCCL INFO Channel 19/32 :    0
+ip-26-0-172-57:942618:943010 [7] NCCL INFO Channel 20/32 :    0
+ip-26-0-172-57:942618:943010 [7] NCCL INFO Channel 21/32 :    0
+ip-26-0-172-57:942618:943010 [7] NCCL INFO Channel 22/32 :    0
+ip-26-0-172-57:942618:943010 [7] NCCL INFO Channel 23/32 :    0
+ip-26-0-172-57:942618:943010 [7] NCCL INFO Channel 24/32 :    0
+ip-26-0-172-57:942618:943010 [7] NCCL INFO Channel 25/32 :    0
+ip-26-0-172-57:942618:943010 [7] NCCL INFO Channel 26/32 :    0
+ip-26-0-172-57:942618:943010 [7] NCCL INFO Channel 27/32 :    0
+ip-26-0-172-57:942618:943010 [7] NCCL INFO Channel 28/32 :    0
+ip-26-0-172-57:942618:943010 [7] NCCL INFO Channel 29/32 :    0
+ip-26-0-172-57:942618:943010 [7] NCCL INFO Channel 30/32 :    0
+ip-26-0-172-57:942618:943010 [7] NCCL INFO Channel 31/32 :    0
+ip-26-0-172-57:942618:943010 [7] NCCL INFO Trees [0] -1/-1/-1->0->-1 [1] -1/-1/-1->0->-1 [2] -1/-1/-1->0->-1 [3] -1/-1/-1->0->-1 [4] -1/-1/-1->0->-1 [5] -1/-1/-1->0->-1 [6] -1/-1/-1->0->-1 [7] -1/-1/-1->0->-1 [8] -1/-1/-1->0->-1 [9] -1/-1/-1->0->-1 [10] -1/-1/-1->0->-1 [11] -1/-1/-1->0->-1 [12] -1/-1/-1->0->-1 [13] -1/-1/-1->0->-1 [14] -1/-1/-1->0->-1 [15] -1/-1/-1->0->-1 [16] -1/-1/-1->0->-1 [17] -1/-1/-1->0->-1 [18] -1/-1/-1->0->-1 [19] -1/-1/-1->0->-1 [20] -1/-1/-1->0->-1 [21] -1/-1/-1->0->-1 [22] -1/-1/-1->0->-1 [23] -1/-1/-1->0->-1 [24] -1/-1/-1->0->-1 [25] -1/-1/-1->0->-1 [26] -1/-1/-1->0->-1 [27] -1/-1/-1->0->-1 [28] -1/-1/-1->0->-1 [29] -1/-1/-1->0->-1 [30] -1/-1/-1->0->-1 [31] -1/-1/-1->0->-1
+ip-26-0-172-57:942618:943010 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-57:942618:943010 [7] NCCL INFO Connected all rings
+ip-26-0-172-57:942618:943010 [7] NCCL INFO Connected all trees
+ip-26-0-172-57:942618:943010 [7] NCCL INFO 32 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-57:942611:943015 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-172-57:942611:943015 [0] NCCL INFO Channel 00/32 :    0
+ip-26-0-172-57:942611:943015 [0] NCCL INFO Channel 01/32 :    0
+ip-26-0-172-57:942611:943015 [0] NCCL INFO Channel 02/32 :    0
+ip-26-0-172-57:942611:943015 [0] NCCL INFO Channel 03/32 :    0
+ip-26-0-172-57:942611:943015 [0] NCCL INFO Channel 04/32 :    0
+ip-26-0-172-57:942611:943015 [0] NCCL INFO Channel 05/32 :    0
+ip-26-0-172-57:942611:943015 [0] NCCL INFO Channel 06/32 :    0
+ip-26-0-172-57:942611:943015 [0] NCCL INFO Channel 07/32 :    0
+ip-26-0-172-57:942611:943015 [0] NCCL INFO Channel 08/32 :    0
+ip-26-0-172-57:942611:943015 [0] NCCL INFO Channel 09/32 :    0
+ip-26-0-172-57:942611:943015 [0] NCCL INFO Channel 10/32 :    0
+ip-26-0-172-57:942611:943015 [0] NCCL INFO Channel 11/32 :    0
+ip-26-0-172-57:942611:943015 [0] NCCL INFO Channel 12/32 :    0
+ip-26-0-172-57:942611:943015 [0] NCCL INFO Channel 13/32 :    0
+ip-26-0-172-57:942611:943015 [0] NCCL INFO Channel 14/32 :    0
+ip-26-0-172-57:942611:943015 [0] NCCL INFO Channel 15/32 :    0
+ip-26-0-172-57:942611:943015 [0] NCCL INFO Channel 16/32 :    0
+ip-26-0-172-57:942611:943015 [0] NCCL INFO Channel 17/32 :    0
+ip-26-0-172-57:942611:943015 [0] NCCL INFO Channel 18/32 :    0
+ip-26-0-172-57:942611:943015 [0] NCCL INFO Channel 19/32 :    0
+ip-26-0-172-57:942611:943015 [0] NCCL INFO Channel 20/32 :    0
+ip-26-0-172-57:942611:943015 [0] NCCL INFO Channel 21/32 :    0
+ip-26-0-172-57:942611:943015 [0] NCCL INFO Channel 22/32 :    0
+ip-26-0-172-57:942611:943015 [0] NCCL INFO Channel 23/32 :    0
+ip-26-0-172-57:942611:943015 [0] NCCL INFO Channel 24/32 :    0
+ip-26-0-172-57:942611:943015 [0] NCCL INFO Channel 25/32 :    0
+ip-26-0-172-57:942611:943015 [0] NCCL INFO Channel 26/32 :    0
+ip-26-0-172-57:942611:943015 [0] NCCL INFO Channel 27/32 :    0
+ip-26-0-172-57:942611:943015 [0] NCCL INFO Channel 28/32 :    0
+ip-26-0-172-57:942611:943015 [0] NCCL INFO Channel 29/32 :    0
+ip-26-0-172-57:942611:943015 [0] NCCL INFO Channel 30/32 :    0
+ip-26-0-172-57:942611:943015 [0] NCCL INFO Channel 31/32 :    0
+ip-26-0-172-57:942611:943015 [0] NCCL INFO Trees [0] -1/-1/-1->0->-1 [1] -1/-1/-1->0->-1 [2] -1/-1/-1->0->-1 [3] -1/-1/-1->0->-1 [4] -1/-1/-1->0->-1 [5] -1/-1/-1->0->-1 [6] -1/-1/-1->0->-1 [7] -1/-1/-1->0->-1 [8] -1/-1/-1->0->-1 [9] -1/-1/-1->0->-1 [10] -1/-1/-1->0->-1 [11] -1/-1/-1->0->-1 [12] -1/-1/-1->0->-1 [13] -1/-1/-1->0->-1 [14] -1/-1/-1->0->-1 [15] -1/-1/-1->0->-1 [16] -1/-1/-1->0->-1 [17] -1/-1/-1->0->-1 [18] -1/-1/-1->0->-1 [19] -1/-1/-1->0->-1 [20] -1/-1/-1->0->-1 [21] -1/-1/-1->0->-1 [22] -1/-1/-1->0->-1 [23] -1/-1/-1->0->-1 [24] -1/-1/-1->0->-1 [25] -1/-1/-1->0->-1 [26] -1/-1/-1->0->-1 [27] -1/-1/-1->0->-1 [28] -1/-1/-1->0->-1 [29] -1/-1/-1->0->-1 [30] -1/-1/-1->0->-1 [31] -1/-1/-1->0->-1
+ip-26-0-172-57:942611:943015 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-57:942611:943015 [0] NCCL INFO Connected all rings
+ip-26-0-172-57:942611:943015 [0] NCCL INFO Connected all trees
+ip-26-0-172-57:942611:943015 [0] NCCL INFO 32 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-57:942615:943007 [4] NCCL INFO comm 0xc0a4f70 rank 0 nranks 1 cudaDev 4 nvmlDev 4 busId 97000 commId 0x91c0847960e5f6ba - Init COMPLETE
+ip-26-0-172-57:942613:943014 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-172-57:942615:943026 [4] NCCL INFO Using network Libfabric
+ip-26-0-172-252:688204:688623 [4] NCCL INFO Using network Libfabric
+ip-26-0-160-225:94346:94798 [4] NCCL INFO Using network Libfabric
+ip-26-0-172-57:942613:943014 [2] NCCL INFO Channel 00/32 :    0
+ip-26-0-172-57:942613:943014 [2] NCCL INFO Channel 01/32 :    0
+ip-26-0-172-57:942613:943014 [2] NCCL INFO Channel 02/32 :    0
+ip-26-0-172-57:942613:943014 [2] NCCL INFO Channel 03/32 :    0
+ip-26-0-172-57:942613:943014 [2] NCCL INFO Channel 04/32 :    0
+ip-26-0-172-57:942613:943014 [2] NCCL INFO Channel 05/32 :    0
+ip-26-0-172-57:942613:943014 [2] NCCL INFO Channel 06/32 :    0
+ip-26-0-172-57:942613:943014 [2] NCCL INFO Channel 07/32 :    0
+ip-26-0-172-57:942613:943014 [2] NCCL INFO Channel 08/32 :    0
+ip-26-0-172-57:942613:943014 [2] NCCL INFO Channel 09/32 :    0
+ip-26-0-172-57:942613:943014 [2] NCCL INFO Channel 10/32 :    0
+ip-26-0-172-57:942613:943014 [2] NCCL INFO Channel 11/32 :    0
+ip-26-0-172-57:942613:943014 [2] NCCL INFO Channel 12/32 :    0
+ip-26-0-172-57:942613:943014 [2] NCCL INFO Channel 13/32 :    0
+ip-26-0-172-57:942613:943014 [2] NCCL INFO Channel 14/32 :    0
+ip-26-0-172-57:942613:943014 [2] NCCL INFO Channel 15/32 :    0
+ip-26-0-172-57:942613:943014 [2] NCCL INFO Channel 16/32 :    0
+ip-26-0-172-57:942613:943014 [2] NCCL INFO Channel 17/32 :    0
+ip-26-0-172-57:942613:943014 [2] NCCL INFO Channel 18/32 :    0
+ip-26-0-172-57:942613:943014 [2] NCCL INFO Channel 19/32 :    0
+ip-26-0-172-57:942613:943014 [2] NCCL INFO Channel 20/32 :    0
+ip-26-0-172-57:942613:943014 [2] NCCL INFO Channel 21/32 :    0
+ip-26-0-172-57:942613:943014 [2] NCCL INFO Channel 22/32 :    0
+ip-26-0-172-57:942613:943014 [2] NCCL INFO Channel 23/32 :    0
+ip-26-0-172-57:942613:943014 [2] NCCL INFO Channel 24/32 :    0
+ip-26-0-172-57:942613:943014 [2] NCCL INFO Channel 25/32 :    0
+ip-26-0-172-57:942613:943014 [2] NCCL INFO Channel 26/32 :    0
+ip-26-0-172-57:942613:943014 [2] NCCL INFO Channel 27/32 :    0
+ip-26-0-172-57:942613:943014 [2] NCCL INFO Channel 28/32 :    0
+ip-26-0-172-57:942613:943014 [2] NCCL INFO Channel 29/32 :    0
+ip-26-0-172-57:942613:943014 [2] NCCL INFO Channel 30/32 :    0
+ip-26-0-172-57:942613:943014 [2] NCCL INFO Channel 31/32 :    0
+ip-26-0-172-116:1019292:1019710 [4] NCCL INFO Using network Libfabric
+ip-26-0-172-57:942613:943014 [2] NCCL INFO Trees [0] -1/-1/-1->0->-1 [1] -1/-1/-1->0->-1 [2] -1/-1/-1->0->-1 [3] -1/-1/-1->0->-1 [4] -1/-1/-1->0->-1 [5] -1/-1/-1->0->-1 [6] -1/-1/-1->0->-1 [7] -1/-1/-1->0->-1 [8] -1/-1/-1->0->-1 [9] -1/-1/-1->0->-1 [10] -1/-1/-1->0->-1 [11] -1/-1/-1->0->-1 [12] -1/-1/-1->0->-1 [13] -1/-1/-1->0->-1 [14] -1/-1/-1->0->-1 [15] -1/-1/-1->0->-1 [16] -1/-1/-1->0->-1 [17] -1/-1/-1->0->-1 [18] -1/-1/-1->0->-1 [19] -1/-1/-1->0->-1 [20] -1/-1/-1->0->-1 [21] -1/-1/-1->0->-1 [22] -1/-1/-1->0->-1 [23] -1/-1/-1->0->-1 [24] -1/-1/-1->0->-1 [25] -1/-1/-1->0->-1 [26] -1/-1/-1->0->-1 [27] -1/-1/-1->0->-1 [28] -1/-1/-1->0->-1 [29] -1/-1/-1->0->-1 [30] -1/-1/-1->0->-1 [31] -1/-1/-1->0->-1
+ip-26-0-172-57:942613:943014 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-73:926598:927015 [4] NCCL INFO Using network Libfabric
+ip-26-0-172-147:618588:619004 [4] NCCL INFO Using network Libfabric
+ip-26-0-173-7:811252:811692 [4] NCCL INFO Using network Libfabric
+ip-26-0-172-142:880683:881099 [4] NCCL INFO Using network Libfabric
+ip-26-0-172-57:942616:943005 [5] NCCL INFO comm 0xa628600 rank 0 nranks 1 cudaDev 5 nvmlDev 5 busId a8000 commId 0xa1770233b27f3f86 - Init COMPLETE
+ip-26-0-172-57:942613:943014 [2] NCCL INFO Connected all rings
+ip-26-0-172-57:942613:943014 [2] NCCL INFO Connected all trees
+ip-26-0-172-57:942613:943014 [2] NCCL INFO 32 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-160-225:94346:94798 [4] NCCL INFO comm 0xba0dd40 rank 0 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0x886b815c42972199 - Init START
+ip-26-0-173-7:811252:811692 [4] NCCL INFO comm 0x99af050 rank 7 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0x886b815c42972199 - Init START
+ip-26-0-172-116:1019292:1019710 [4] NCCL INFO comm 0xb8a8ed0 rank 1 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0x886b815c42972199 - Init START
+ip-26-0-172-142:880683:881099 [4] NCCL INFO comm 0xb1a2ce0 rank 2 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0x886b815c42972199 - Init START
+ip-26-0-172-57:942615:943026 [4] NCCL INFO comm 0xb6503d0 rank 5 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0x886b815c42972199 - Init START
+ip-26-0-172-73:926598:927015 [4] NCCL INFO comm 0xae64190 rank 6 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0x886b815c42972199 - Init START
+ip-26-0-172-147:618588:619004 [4] NCCL INFO comm 0x9d70c20 rank 3 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0x886b815c42972199 - Init START
+ip-26-0-172-252:688204:688623 [4] NCCL INFO comm 0xaa1b5a0 rank 4 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0x886b815c42972199 - Init START
+ip-26-0-172-73:926598:927015 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-173-7:811252:811692 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-160-225:94346:94798 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-116:1019292:1019710 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-57:942615:943026 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-252:688204:688623 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-142:880683:881099 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-147:618588:619004 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-142:880684:881100 [5] NCCL INFO Using network Libfabric
+ip-26-0-160-225:94347:94800 [5] NCCL INFO Using network Libfabric
+ip-26-0-172-147:618589:619005 [5] NCCL INFO Using network Libfabric
+ip-26-0-172-57:942616:943029 [5] NCCL INFO Using network Libfabric
+ip-26-0-173-7:811253:811693 [5] NCCL INFO Using network Libfabric
+ip-26-0-172-73:926599:927016 [5] NCCL INFO Using network Libfabric
+ip-26-0-172-116:1019293:1019711 [5] NCCL INFO Using network Libfabric
+ip-26-0-172-252:688205:688624 [5] NCCL INFO Using network Libfabric
+ip-26-0-160-225:94347:94800 [5] NCCL INFO comm 0xaaa4c20 rank 0 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0xa25ee5d67de91a57 - Init START
+ip-26-0-173-7:811253:811693 [5] NCCL INFO comm 0xa3027b0 rank 7 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0xa25ee5d67de91a57 - Init START
+ip-26-0-172-116:1019293:1019711 [5] NCCL INFO comm 0xb94aca0 rank 1 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0xa25ee5d67de91a57 - Init START
+ip-26-0-172-142:880684:881100 [5] NCCL INFO comm 0x9fca120 rank 2 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0xa25ee5d67de91a57 - Init START
+ip-26-0-172-147:618589:619005 [5] NCCL INFO comm 0xad42db0 rank 3 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0xa25ee5d67de91a57 - Init START
+ip-26-0-172-252:688205:688624 [5] NCCL INFO comm 0xa383660 rank 4 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0xa25ee5d67de91a57 - Init START
+ip-26-0-172-57:942616:943029 [5] NCCL INFO comm 0x9bd20e0 rank 5 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0xa25ee5d67de91a57 - Init START
+ip-26-0-172-73:926599:927016 [5] NCCL INFO comm 0xb5102b0 rank 6 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0xa25ee5d67de91a57 - Init START
+ip-26-0-172-73:926599:927016 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-160-225:94347:94800 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-173-7:811253:811693 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-116:1019293:1019711 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-57:942616:943029 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-147:618589:619005 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-252:688205:688624 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-142:880684:881100 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-57:942617:943002 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-172-57:942617:943002 [6] NCCL INFO Channel 00/32 :    0
+ip-26-0-172-57:942617:943002 [6] NCCL INFO Channel 01/32 :    0
+ip-26-0-172-57:942617:943002 [6] NCCL INFO Channel 02/32 :    0
+ip-26-0-172-57:942617:943002 [6] NCCL INFO Channel 03/32 :    0
+ip-26-0-172-57:942617:943002 [6] NCCL INFO Channel 04/32 :    0
+ip-26-0-172-57:942617:943002 [6] NCCL INFO Channel 05/32 :    0
+ip-26-0-172-57:942617:943002 [6] NCCL INFO Channel 06/32 :    0
+ip-26-0-172-57:942617:943002 [6] NCCL INFO Channel 07/32 :    0
+ip-26-0-172-57:942617:943002 [6] NCCL INFO Channel 08/32 :    0
+ip-26-0-172-57:942617:943002 [6] NCCL INFO Channel 09/32 :    0
+ip-26-0-172-57:942617:943002 [6] NCCL INFO Channel 10/32 :    0
+ip-26-0-172-57:942617:943002 [6] NCCL INFO Channel 11/32 :    0
+ip-26-0-172-57:942617:943002 [6] NCCL INFO Channel 12/32 :    0
+ip-26-0-172-57:942617:943002 [6] NCCL INFO Channel 13/32 :    0
+ip-26-0-172-57:942617:943002 [6] NCCL INFO Channel 14/32 :    0
+ip-26-0-172-57:942617:943002 [6] NCCL INFO Channel 15/32 :    0
+ip-26-0-172-57:942617:943002 [6] NCCL INFO Channel 16/32 :    0
+ip-26-0-172-57:942617:943002 [6] NCCL INFO Channel 17/32 :    0
+ip-26-0-172-57:942617:943002 [6] NCCL INFO Channel 18/32 :    0
+ip-26-0-172-57:942617:943002 [6] NCCL INFO Channel 19/32 :    0
+ip-26-0-172-57:942617:943002 [6] NCCL INFO Channel 20/32 :    0
+ip-26-0-172-57:942617:943002 [6] NCCL INFO Channel 21/32 :    0
+ip-26-0-172-57:942617:943002 [6] NCCL INFO Channel 22/32 :    0
+ip-26-0-172-57:942617:943002 [6] NCCL INFO Channel 23/32 :    0
+ip-26-0-172-57:942617:943002 [6] NCCL INFO Channel 24/32 :    0
+ip-26-0-172-57:942617:943002 [6] NCCL INFO Channel 25/32 :    0
+ip-26-0-172-57:942617:943002 [6] NCCL INFO Channel 26/32 :    0
+ip-26-0-172-57:942617:943002 [6] NCCL INFO Channel 27/32 :    0
+ip-26-0-172-57:942617:943002 [6] NCCL INFO Channel 28/32 :    0
+ip-26-0-172-57:942617:943002 [6] NCCL INFO Channel 29/32 :    0
+ip-26-0-172-57:942617:943002 [6] NCCL INFO Channel 30/32 :    0
+ip-26-0-172-57:942617:943002 [6] NCCL INFO Channel 31/32 :    0
+ip-26-0-172-57:942617:943002 [6] NCCL INFO Trees [0] -1/-1/-1->0->-1 [1] -1/-1/-1->0->-1 [2] -1/-1/-1->0->-1 [3] -1/-1/-1->0->-1 [4] -1/-1/-1->0->-1 [5] -1/-1/-1->0->-1 [6] -1/-1/-1->0->-1 [7] -1/-1/-1->0->-1 [8] -1/-1/-1->0->-1 [9] -1/-1/-1->0->-1 [10] -1/-1/-1->0->-1 [11] -1/-1/-1->0->-1 [12] -1/-1/-1->0->-1 [13] -1/-1/-1->0->-1 [14] -1/-1/-1->0->-1 [15] -1/-1/-1->0->-1 [16] -1/-1/-1->0->-1 [17] -1/-1/-1->0->-1 [18] -1/-1/-1->0->-1 [19] -1/-1/-1->0->-1 [20] -1/-1/-1->0->-1 [21] -1/-1/-1->0->-1 [22] -1/-1/-1->0->-1 [23] -1/-1/-1->0->-1 [24] -1/-1/-1->0->-1 [25] -1/-1/-1->0->-1 [26] -1/-1/-1->0->-1 [27] -1/-1/-1->0->-1 [28] -1/-1/-1->0->-1 [29] -1/-1/-1->0->-1 [30] -1/-1/-1->0->-1 [31] -1/-1/-1->0->-1
+ip-26-0-172-57:942617:943002 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-225:94342:94802 [0] NCCL INFO Using network Libfabric
+ip-26-0-172-116:1019288:1019712 [0] NCCL INFO Using network Libfabric
+ip-26-0-172-73:926594:927017 [0] NCCL INFO Using network Libfabric
+ip-26-0-172-142:880679:881101 [0] NCCL INFO Using network Libfabric
+ip-26-0-173-7:811248:811694 [0] NCCL INFO Using network Libfabric
+ip-26-0-172-147:618584:619006 [0] NCCL INFO Using network Libfabric
+ip-26-0-172-252:688200:688625 [0] NCCL INFO Using network Libfabric
+ip-26-0-160-225:94349:94804 [7] NCCL INFO Using network Libfabric
+ip-26-0-172-142:880686:881102 [7] NCCL INFO Using network Libfabric
+ip-26-0-172-252:688207:688626 [7] NCCL INFO Using network Libfabric
+ip-26-0-172-116:1019295:1019713 [7] NCCL INFO Using network Libfabric
+ip-26-0-172-73:926601:927018 [7] NCCL INFO Using network Libfabric
+ip-26-0-173-7:811255:811695 [7] NCCL INFO Using network Libfabric
+ip-26-0-172-147:618591:619007 [7] NCCL INFO Using network Libfabric
+ip-26-0-173-7:811248:811694 [0] NCCL INFO comm 0xb8c6260 rank 7 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0x242f1fc2efc03fb0 - Init START
+ip-26-0-160-225:94342:94802 [0] NCCL INFO comm 0xa836620 rank 0 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0x242f1fc2efc03fb0 - Init START
+ip-26-0-172-116:1019288:1019712 [0] NCCL INFO comm 0xa175fc0 rank 1 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0x242f1fc2efc03fb0 - Init START
+ip-26-0-172-142:880679:881101 [0] NCCL INFO comm 0xb7085d0 rank 2 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0x242f1fc2efc03fb0 - Init START
+ip-26-0-172-147:618584:619006 [0] NCCL INFO comm 0xb0208b0 rank 3 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0x242f1fc2efc03fb0 - Init START
+ip-26-0-172-252:688200:688625 [0] NCCL INFO comm 0xa91f640 rank 4 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0x242f1fc2efc03fb0 - Init START
+ip-26-0-172-73:926594:927017 [0] NCCL INFO comm 0xa9281d0 rank 6 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0x242f1fc2efc03fb0 - Init START
+ip-26-0-172-73:926594:927017 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-160-225:94342:94802 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/349
+ip-26-0-173-7:811248:811694 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-116:1019288:1019712 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-142:880679:881101 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-252:688200:688625 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-147:618584:619006 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-173-7:811255:811695 [7] NCCL INFO comm 0xa8df660 rank 7 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0xca6ca50e1a87f46c - Init START
+ip-26-0-160-225:94349:94804 [7] NCCL INFO comm 0xa3736b0 rank 0 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0xca6ca50e1a87f46c - Init START
+ip-26-0-172-116:1019295:1019713 [7] NCCL INFO comm 0xaf12270 rank 1 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0xca6ca50e1a87f46c - Init START
+ip-26-0-172-142:880686:881102 [7] NCCL INFO comm 0xac750d0 rank 2 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0xca6ca50e1a87f46c - Init START
+ip-26-0-172-147:618591:619007 [7] NCCL INFO comm 0xa8e6270 rank 3 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0xca6ca50e1a87f46c - Init START
+ip-26-0-172-73:926601:927018 [7] NCCL INFO comm 0x9af5e90 rank 6 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0xca6ca50e1a87f46c - Init START
+ip-26-0-172-252:688207:688626 [7] NCCL INFO comm 0xa48c680 rank 4 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0xca6ca50e1a87f46c - Init START
+ip-26-0-172-73:926601:927018 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-160-225:94349:94804 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-173-7:811255:811695 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-116:1019295:1019713 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-252:688207:688626 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-147:618591:619007 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-142:880686:881102 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-160-225:94344:94806 [2] NCCL INFO Using network Libfabric
+ip-26-0-172-147:618586:619008 [2] NCCL INFO Using network Libfabric
+ip-26-0-172-142:880681:881103 [2] NCCL INFO Using network Libfabric
+ip-26-0-172-252:688202:688627 [2] NCCL INFO Using network Libfabric
+ip-26-0-172-116:1019290:1019714 [2] NCCL INFO Using network Libfabric
+ip-26-0-173-7:811250:811696 [2] NCCL INFO Using network Libfabric
+ip-26-0-172-73:926596:927019 [2] NCCL INFO Using network Libfabric
+ip-26-0-160-225:94344:94806 [2] NCCL INFO comm 0xa9caa60 rank 0 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0x347557a4b53fac7e - Init START
+ip-26-0-173-7:811250:811696 [2] NCCL INFO comm 0xa294550 rank 7 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0x347557a4b53fac7e - Init START
+ip-26-0-172-116:1019290:1019714 [2] NCCL INFO comm 0xb5359e0 rank 1 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0x347557a4b53fac7e - Init START
+ip-26-0-172-142:880681:881103 [2] NCCL INFO comm 0xa582810 rank 2 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0x347557a4b53fac7e - Init START
+ip-26-0-172-147:618586:619008 [2] NCCL INFO comm 0xa0655f0 rank 3 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0x347557a4b53fac7e - Init START
+ip-26-0-172-252:688202:688627 [2] NCCL INFO comm 0xa7a9a50 rank 4 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0x347557a4b53fac7e - Init START
+ip-26-0-172-73:926596:927019 [2] NCCL INFO comm 0xa369f60 rank 6 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0x347557a4b53fac7e - Init START
+ip-26-0-172-73:926596:927019 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-160-225:94344:94806 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-173-7:811250:811696 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-116:1019290:1019714 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-142:880681:881103 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-147:618586:619008 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-252:688202:688627 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-147:618590:619009 [6] NCCL INFO Using network Libfabric
+ip-26-0-172-73:926600:927020 [6] NCCL INFO Using network Libfabric
+ip-26-0-172-252:688206:688628 [6] NCCL INFO Using network Libfabric
+ip-26-0-160-225:94348:94808 [6] NCCL INFO Using network Libfabric
+ip-26-0-173-7:811254:811697 [6] NCCL INFO Using network Libfabric
+ip-26-0-172-116:1019294:1019715 [6] NCCL INFO Using network Libfabric
+ip-26-0-172-142:880685:881104 [6] NCCL INFO Using network Libfabric
+ip-26-0-172-252:688203:688629 [3] NCCL INFO Using network Libfabric
+ip-26-0-172-147:618587:619010 [3] NCCL INFO Using network Libfabric
+ip-26-0-172-142:880682:881105 [3] NCCL INFO Using network Libfabric
+ip-26-0-172-116:1019291:1019716 [3] NCCL INFO Using network Libfabric
+ip-26-0-160-225:94345:94810 [3] NCCL INFO Using network Libfabric
+ip-26-0-173-7:811251:811698 [3] NCCL INFO Using network Libfabric
+ip-26-0-172-73:926597:927021 [3] NCCL INFO Using network Libfabric
+ip-26-0-173-7:811254:811697 [6] NCCL INFO comm 0x969a920 rank 7 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0x45d17d92b31ef879 - Init START
+ip-26-0-160-225:94348:94808 [6] NCCL INFO comm 0xb16cac0 rank 0 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0x45d17d92b31ef879 - Init START
+ip-26-0-172-116:1019294:1019715 [6] NCCL INFO comm 0x9b47f70 rank 1 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0x45d17d92b31ef879 - Init START
+ip-26-0-172-142:880685:881104 [6] NCCL INFO comm 0x9e7e370 rank 2 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0x45d17d92b31ef879 - Init START
+ip-26-0-172-73:926600:927020 [6] NCCL INFO comm 0xb525f40 rank 6 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0x45d17d92b31ef879 - Init START
+ip-26-0-172-252:688206:688628 [6] NCCL INFO comm 0xb96fa70 rank 4 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0x45d17d92b31ef879 - Init START
+ip-26-0-172-147:618590:619009 [6] NCCL INFO comm 0xb6d0ba0 rank 3 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0x45d17d92b31ef879 - Init START
+ip-26-0-172-73:926600:927020 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-116:1019294:1019715 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-160-225:94348:94808 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-173-7:811254:811697 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-142:880685:881104 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-147:618590:619009 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-252:688206:688628 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-173-7:811251:811698 [3] NCCL INFO comm 0x9892600 rank 7 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0xa72f98a98e82f37d - Init START
+ip-26-0-160-225:94345:94810 [3] NCCL INFO comm 0xbba9960 rank 0 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0xa72f98a98e82f37d - Init START
+ip-26-0-172-116:1019291:1019716 [3] NCCL INFO comm 0xb8bef40 rank 1 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0xa72f98a98e82f37d - Init START
+ip-26-0-172-142:880682:881105 [3] NCCL INFO comm 0xaa271d0 rank 2 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0xa72f98a98e82f37d - Init START
+ip-26-0-172-147:618587:619010 [3] NCCL INFO comm 0xa0b1130 rank 3 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0xa72f98a98e82f37d - Init START
+ip-26-0-172-57:942617:943002 [6] NCCL INFO Connected all rings
+ip-26-0-172-57:942617:943002 [6] NCCL INFO Connected all trees
+ip-26-0-172-252:688203:688629 [3] NCCL INFO comm 0x9d167c0 rank 4 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0xa72f98a98e82f37d - Init START
+ip-26-0-172-57:942617:943002 [6] NCCL INFO 32 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-73:926597:927021 [3] NCCL INFO comm 0xa1f68d0 rank 6 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0xa72f98a98e82f37d - Init START
+ip-26-0-172-57:942614:943012 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-172-57:942614:943012 [3] NCCL INFO Channel 00/32 :    0
+ip-26-0-172-57:942614:943012 [3] NCCL INFO Channel 01/32 :    0
+ip-26-0-172-57:942614:943012 [3] NCCL INFO Channel 02/32 :    0
+ip-26-0-172-57:942614:943012 [3] NCCL INFO Channel 03/32 :    0
+ip-26-0-172-57:942614:943012 [3] NCCL INFO Channel 04/32 :    0
+ip-26-0-172-57:942614:943012 [3] NCCL INFO Channel 05/32 :    0
+ip-26-0-172-57:942614:943012 [3] NCCL INFO Channel 06/32 :    0
+ip-26-0-172-57:942614:943012 [3] NCCL INFO Channel 07/32 :    0
+ip-26-0-172-57:942614:943012 [3] NCCL INFO Channel 08/32 :    0
+ip-26-0-172-57:942614:943012 [3] NCCL INFO Channel 09/32 :    0
+ip-26-0-172-57:942614:943012 [3] NCCL INFO Channel 10/32 :    0
+ip-26-0-172-57:942614:943012 [3] NCCL INFO Channel 11/32 :    0
+ip-26-0-172-57:942614:943012 [3] NCCL INFO Channel 12/32 :    0
+ip-26-0-172-57:942614:943012 [3] NCCL INFO Channel 13/32 :    0
+ip-26-0-172-57:942614:943012 [3] NCCL INFO Channel 14/32 :    0
+ip-26-0-172-57:942614:943012 [3] NCCL INFO Channel 15/32 :    0
+ip-26-0-172-57:942614:943012 [3] NCCL INFO Channel 16/32 :    0
+ip-26-0-172-57:942614:943012 [3] NCCL INFO Channel 17/32 :    0
+ip-26-0-172-57:942614:943012 [3] NCCL INFO Channel 18/32 :    0
+ip-26-0-172-57:942614:943012 [3] NCCL INFO Channel 19/32 :    0
+ip-26-0-172-57:942614:943012 [3] NCCL INFO Channel 20/32 :    0
+ip-26-0-172-57:942614:943012 [3] NCCL INFO Channel 21/32 :    0
+ip-26-0-172-57:942614:943012 [3] NCCL INFO Channel 22/32 :    0
+ip-26-0-172-57:942614:943012 [3] NCCL INFO Channel 23/32 :    0
+ip-26-0-172-57:942614:943012 [3] NCCL INFO Channel 24/32 :    0
+ip-26-0-172-57:942614:943012 [3] NCCL INFO Channel 25/32 :    0
+ip-26-0-172-57:942614:943012 [3] NCCL INFO Channel 26/32 :    0
+ip-26-0-172-57:942614:943012 [3] NCCL INFO Channel 27/32 :    0
+ip-26-0-172-57:942614:943012 [3] NCCL INFO Channel 28/32 :    0
+ip-26-0-172-57:942614:943012 [3] NCCL INFO Channel 29/32 :    0
+ip-26-0-172-57:942614:943012 [3] NCCL INFO Channel 30/32 :    0
+ip-26-0-172-57:942614:943012 [3] NCCL INFO Channel 31/32 :    0
+ip-26-0-172-57:942614:943012 [3] NCCL INFO Trees [0] -1/-1/-1->0->-1 [1] -1/-1/-1->0->-1 [2] -1/-1/-1->0->-1 [3] -1/-1/-1->0->-1 [4] -1/-1/-1->0->-1 [5] -1/-1/-1->0->-1 [6] -1/-1/-1->0->-1 [7] -1/-1/-1->0->-1 [8] -1/-1/-1->0->-1 [9] -1/-1/-1->0->-1 [10] -1/-1/-1->0->-1 [11] -1/-1/-1->0->-1 [12] -1/-1/-1->0->-1 [13] -1/-1/-1->0->-1 [14] -1/-1/-1->0->-1 [15] -1/-1/-1->0->-1 [16] -1/-1/-1->0->-1 [17] -1/-1/-1->0->-1 [18] -1/-1/-1->0->-1 [19] -1/-1/-1->0->-1 [20] -1/-1/-1->0->-1 [21] -1/-1/-1->0->-1 [22] -1/-1/-1->0->-1 [23] -1/-1/-1->0->-1 [24] -1/-1/-1->0->-1 [25] -1/-1/-1->0->-1 [26] -1/-1/-1->0->-1 [27] -1/-1/-1->0->-1 [28] -1/-1/-1->0->-1 [29] -1/-1/-1->0->-1 [30] -1/-1/-1->0->-1 [31] -1/-1/-1->0->-1
+ip-26-0-172-57:942614:943012 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-57:942611:943015 [0] NCCL INFO comm 0xb2ecde0 rank 0 nranks 1 cudaDev 0 nvmlDev 0 busId 53000 commId 0xcdc1cbe6ccd8d7eb - Init COMPLETE
+ip-26-0-172-57:942614:943012 [3] NCCL INFO Connected all rings
+ip-26-0-172-57:942614:943012 [3] NCCL INFO Connected all trees
+ip-26-0-172-57:942614:943012 [3] NCCL INFO 32 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-57:942618:943010 [7] NCCL INFO comm 0xaf9c830 rank 0 nranks 1 cudaDev 7 nvmlDev 7 busId ca000 commId 0x1475a1e50e772b6e - Init COMPLETE
+ip-26-0-172-57:942611:943034 [0] NCCL INFO Using network Libfabric
+ip-26-0-172-57:942612:943006 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-172-57:942612:943006 [1] NCCL INFO Channel 00/32 :    0
+ip-26-0-172-57:942612:943006 [1] NCCL INFO Channel 01/32 :    0
+ip-26-0-172-57:942612:943006 [1] NCCL INFO Channel 02/32 :    0
+ip-26-0-172-57:942612:943006 [1] NCCL INFO Channel 03/32 :    0
+ip-26-0-172-57:942612:943006 [1] NCCL INFO Channel 04/32 :    0
+ip-26-0-172-57:942612:943006 [1] NCCL INFO Channel 05/32 :    0
+ip-26-0-172-57:942612:943006 [1] NCCL INFO Channel 06/32 :    0
+ip-26-0-172-57:942612:943006 [1] NCCL INFO Channel 07/32 :    0
+ip-26-0-172-57:942612:943006 [1] NCCL INFO Channel 08/32 :    0
+ip-26-0-172-57:942612:943006 [1] NCCL INFO Channel 09/32 :    0
+ip-26-0-172-57:942612:943006 [1] NCCL INFO Channel 10/32 :    0
+ip-26-0-172-57:942612:943006 [1] NCCL INFO Channel 11/32 :    0
+ip-26-0-172-57:942612:943006 [1] NCCL INFO Channel 12/32 :    0
+ip-26-0-172-57:942612:943006 [1] NCCL INFO Channel 13/32 :    0
+ip-26-0-172-57:942612:943006 [1] NCCL INFO Channel 14/32 :    0
+ip-26-0-172-57:942612:943006 [1] NCCL INFO Channel 15/32 :    0
+ip-26-0-172-57:942612:943006 [1] NCCL INFO Channel 16/32 :    0
+ip-26-0-172-57:942612:943006 [1] NCCL INFO Channel 17/32 :    0
+ip-26-0-172-57:942612:943006 [1] NCCL INFO Channel 18/32 :    0
+ip-26-0-172-57:942612:943006 [1] NCCL INFO Channel 19/32 :    0
+ip-26-0-172-57:942612:943006 [1] NCCL INFO Channel 20/32 :    0
+ip-26-0-172-57:942612:943006 [1] NCCL INFO Channel 21/32 :    0
+ip-26-0-172-57:942612:943006 [1] NCCL INFO Channel 22/32 :    0
+ip-26-0-172-57:942612:943006 [1] NCCL INFO Channel 23/32 :    0
+ip-26-0-172-57:942612:943006 [1] NCCL INFO Channel 24/32 :    0
+ip-26-0-172-57:942612:943006 [1] NCCL INFO Channel 25/32 :    0
+ip-26-0-172-57:942612:943006 [1] NCCL INFO Channel 26/32 :    0
+ip-26-0-172-57:942612:943006 [1] NCCL INFO Channel 27/32 :    0
+ip-26-0-172-57:942612:943006 [1] NCCL INFO Channel 28/32 :    0
+ip-26-0-172-57:942612:943006 [1] NCCL INFO Channel 29/32 :    0
+ip-26-0-172-57:942612:943006 [1] NCCL INFO Channel 30/32 :    0
+ip-26-0-172-57:942612:943006 [1] NCCL INFO Channel 31/32 :    0
+ip-26-0-172-73:926597:927021 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-57:942612:943006 [1] NCCL INFO Trees [0] -1/-1/-1->0->-1 [1] -1/-1/-1->0->-1 [2] -1/-1/-1->0->-1 [3] -1/-1/-1->0->-1 [4] -1/-1/-1->0->-1 [5] -1/-1/-1->0->-1 [6] -1/-1/-1->0->-1 [7] -1/-1/-1->0->-1 [8] -1/-1/-1->0->-1 [9] -1/-1/-1->0->-1 [10] -1/-1/-1->0->-1 [11] -1/-1/-1->0->-1 [12] -1/-1/-1->0->-1 [13] -1/-1/-1->0->-1 [14] -1/-1/-1->0->-1 [15] -1/-1/-1->0->-1 [16] -1/-1/-1->0->-1 [17] -1/-1/-1->0->-1 [18] -1/-1/-1->0->-1 [19] -1/-1/-1->0->-1 [20] -1/-1/-1->0->-1 [21] -1/-1/-1->0->-1 [22] -1/-1/-1->0->-1 [23] -1/-1/-1->0->-1 [24] -1/-1/-1->0->-1 [25] -1/-1/-1->0->-1 [26] -1/-1/-1->0->-1 [27] -1/-1/-1->0->-1 [28] -1/-1/-1->0->-1 [29] -1/-1/-1->0->-1 [30] -1/-1/-1->0->-1 [31] -1/-1/-1->0->-1
+ip-26-0-172-57:942612:943006 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-225:94345:94810 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-173-7:811251:811698 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-116:1019291:1019716 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-57:942618:943035 [7] NCCL INFO Using network Libfabric
+ip-26-0-172-142:880682:881105 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-252:688203:688629 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-57:942612:943006 [1] NCCL INFO Connected all rings
+ip-26-0-172-57:942612:943006 [1] NCCL INFO Connected all trees
+ip-26-0-172-57:942612:943006 [1] NCCL INFO 32 coll channels, 0 nvls channels, 32 p2p channels, 32 p2p channels per peer
+ip-26-0-172-147:618587:619010 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-57:942611:943034 [0] NCCL INFO comm 0xb2efa90 rank 5 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0x242f1fc2efc03fb0 - Init START
+ip-26-0-172-57:942611:943034 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-57:942618:943035 [7] NCCL INFO comm 0xa546420 rank 5 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0xca6ca50e1a87f46c - Init START
+ip-26-0-172-57:942618:943035 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-57:942613:943014 [2] NCCL INFO comm 0xbac0ef0 rank 0 nranks 1 cudaDev 2 nvmlDev 2 busId 75000 commId 0xf529a97f75bd2e80 - Init COMPLETE
+ip-26-0-172-57:942613:943038 [2] NCCL INFO Using network Libfabric
+ip-26-0-172-57:942613:943038 [2] NCCL INFO comm 0xb06cc90 rank 5 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0x347557a4b53fac7e - Init START
+ip-26-0-172-57:942613:943038 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-57:942617:943002 [6] NCCL INFO comm 0xacb7bf0 rank 0 nranks 1 cudaDev 6 nvmlDev 6 busId b9000 commId 0xce0197adacec94c - Init COMPLETE
+ip-26-0-172-57:942614:943012 [3] NCCL INFO comm 0xa931fd0 rank 0 nranks 1 cudaDev 3 nvmlDev 3 busId 86000 commId 0xcd152b1d84ad05fc - Init COMPLETE
+ip-26-0-172-57:942617:943039 [6] NCCL INFO Using network Libfabric
+ip-26-0-172-57:942614:943040 [3] NCCL INFO Using network Libfabric
+ip-26-0-172-57:942617:943039 [6] NCCL INFO comm 0xacba8a0 rank 5 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0x45d17d92b31ef879 - Init START
+ip-26-0-172-57:942617:943039 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-57:942614:943040 [3] NCCL INFO comm 0x9edc950 rank 5 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0xa72f98a98e82f37d - Init START
+ip-26-0-172-57:942614:943040 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-57:942612:943006 [1] NCCL INFO comm 0xa0aa950 rank 0 nranks 1 cudaDev 1 nvmlDev 1 busId 64000 commId 0x6610956bd2f6c2fb - Init COMPLETE
+ip-26-0-160-225:94343:94812 [1] NCCL INFO Using network Libfabric
+ip-26-0-172-57:942612:943041 [1] NCCL INFO Using network Libfabric
+ip-26-0-172-252:688201:688630 [1] NCCL INFO Using network Libfabric
+ip-26-0-172-73:926595:927022 [1] NCCL INFO Using network Libfabric
+ip-26-0-172-147:618585:619011 [1] NCCL INFO Using network Libfabric
+ip-26-0-172-116:1019289:1019717 [1] NCCL INFO Using network Libfabric
+ip-26-0-172-142:880680:881106 [1] NCCL INFO Using network Libfabric
+ip-26-0-173-7:811249:811699 [1] NCCL INFO Using network Libfabric
+ip-26-0-160-225:94343:94812 [1] NCCL INFO comm 0x9c757e0 rank 0 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0x502faa1c75e98308 - Init START
+ip-26-0-173-7:811249:811699 [1] NCCL INFO comm 0x9495220 rank 7 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0x502faa1c75e98308 - Init START
+ip-26-0-172-116:1019289:1019717 [1] NCCL INFO comm 0x9b48f20 rank 1 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0x502faa1c75e98308 - Init START
+ip-26-0-172-142:880680:881106 [1] NCCL INFO comm 0xa4bbcf0 rank 2 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0x502faa1c75e98308 - Init START
+ip-26-0-172-147:618585:619011 [1] NCCL INFO comm 0xb534180 rank 3 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0x502faa1c75e98308 - Init START
+ip-26-0-172-252:688201:688630 [1] NCCL INFO comm 0xa8c5490 rank 4 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0x502faa1c75e98308 - Init START
+ip-26-0-172-57:942612:943041 [1] NCCL INFO comm 0xa0ad600 rank 5 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0x502faa1c75e98308 - Init START
+ip-26-0-172-73:926595:927022 [1] NCCL INFO comm 0x9a19f50 rank 6 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0x502faa1c75e98308 - Init START
+ip-26-0-172-73:926595:927022 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-160-225:94343:94812 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-173-7:811249:811699 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-116:1019289:1019717 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-142:880680:881106 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-147:618585:619011 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-252:688201:688630 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-57:942612:943041 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-173-7:811253:811693 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-172-142:880683:881099 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-172-147:618588:619004 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-172-57:942615:943026 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-173-7:811252:811692 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-160-225:94346:94798 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-172-73:926598:927015 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-173-7:811255:811695 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-172-73:926599:927016 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-172-252:688204:688623 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-172-73:926601:927018 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-172-252:688205:688624 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-172-147:618589:619005 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-173-7:811250:811696 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-172-252:688200:688625 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-160-225:94349:94804 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-160-225:94344:94806 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-173-7:811248:811694 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-160-225:94342:94802 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-172-142:880680:881106 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-172-73:926600:927020 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-172-142:880679:881101 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-173-7:811249:811699 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-172-57:942618:943035 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-160-225:94347:94800 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-172-116:1019293:1019711 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-160-225:94348:94808 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-173-7:811251:811698 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-172-73:926595:927022 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-172-252:688203:688629 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-172-252:688202:688627 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-173-7:811254:811697 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-172-73:926597:927021 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-160-225:94343:94812 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-172-73:926594:927017 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-172-57:942616:943029 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-160-225:94345:94810 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-172-73:926596:927019 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-172-252:688206:688628 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-172-252:688207:688626 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-172-252:688201:688630 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-172-142:880684:881100 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-172-142:880684:881100 [5] NCCL INFO Trees [0] 1/3/-1->2->4 [1] 1/3/-1->2->4 [2] -1/-1/-1->2->1 [3] -1/-1/-1->2->1
+ip-26-0-172-142:880684:881100 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-147:618589:619005 [5] NCCL INFO Trees [0] -1/-1/-1->3->2 [1] -1/-1/-1->3->2 [2] 5/1/-1->3->7 [3] 5/1/-1->3->7
+ip-26-0-172-147:618589:619005 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-252:688205:688624 [5] NCCL INFO Trees [0] 2/6/-1->4->0 [1] 2/6/-1->4->0 [2] -1/-1/-1->4->5 [3] -1/-1/-1->4->5
+ip-26-0-172-252:688205:688624 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-57:942616:943029 [5] NCCL INFO Trees [0] -1/-1/-1->5->6 [1] -1/-1/-1->5->6 [2] 6/4/-1->5->3 [3] 6/4/-1->5->3
+ip-26-0-172-57:942616:943029 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-73:926599:927016 [5] NCCL INFO Trees [0] 5/7/-1->6->4 [1] 5/7/-1->6->4 [2] -1/-1/-1->6->5 [3] -1/-1/-1->6->5
+ip-26-0-172-73:926599:927016 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-225:94347:94800 [5] NCCL INFO Channel 00/04 :    0   1   2   3   4   5   6   7
+ip-26-0-160-225:94347:94800 [5] NCCL INFO Channel 01/04 :    0   1   2   3   4   5   6   7
+ip-26-0-160-225:94347:94800 [5] NCCL INFO Channel 02/04 :    0   1   2   3   4   5   6   7
+ip-26-0-160-225:94347:94800 [5] NCCL INFO Channel 03/04 :    0   1   2   3   4   5   6   7
+ip-26-0-160-225:94347:94800 [5] NCCL INFO Trees [0] 4/-1/-1->0->-1 [1] 4/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-160-225:94347:94800 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-173-7:811253:811693 [5] NCCL INFO Trees [0] -1/-1/-1->7->6 [1] -1/-1/-1->7->6 [2] 3/-1/-1->7->-1 [3] 3/-1/-1->7->-1
+ip-26-0-173-7:811253:811693 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-116:1019292:1019710 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-172-116:1019293:1019711 [5] NCCL INFO Trees [0] -1/-1/-1->1->2 [1] -1/-1/-1->1->2 [2] 2/0/-1->1->3 [3] 2/0/-1->1->3
+ip-26-0-172-116:1019293:1019711 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-57:942612:943041 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-172-142:880686:881102 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-172-116:1019292:1019710 [4] NCCL INFO Trees [0] -1/-1/-1->1->2 [1] -1/-1/-1->1->2 [2] 2/0/-1->1->3 [3] 2/0/-1->1->3
+ip-26-0-172-116:1019292:1019710 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-142:880683:881099 [4] NCCL INFO Trees [0] 1/3/-1->2->4 [1] 1/3/-1->2->4 [2] -1/-1/-1->2->1 [3] -1/-1/-1->2->1
+ip-26-0-172-142:880683:881099 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-147:618588:619004 [4] NCCL INFO Trees [0] -1/-1/-1->3->2 [1] -1/-1/-1->3->2 [2] 5/1/-1->3->7 [3] 5/1/-1->3->7
+ip-26-0-172-147:618588:619004 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-252:688204:688623 [4] NCCL INFO Trees [0] 2/6/-1->4->0 [1] 2/6/-1->4->0 [2] -1/-1/-1->4->5 [3] -1/-1/-1->4->5
+ip-26-0-172-252:688204:688623 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-57:942615:943026 [4] NCCL INFO Trees [0] -1/-1/-1->5->6 [1] -1/-1/-1->5->6 [2] 6/4/-1->5->3 [3] 6/4/-1->5->3
+ip-26-0-172-57:942615:943026 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-73:926598:927015 [4] NCCL INFO Trees [0] 5/7/-1->6->4 [1] 5/7/-1->6->4 [2] -1/-1/-1->6->5 [3] -1/-1/-1->6->5
+ip-26-0-172-73:926598:927015 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-225:94346:94798 [4] NCCL INFO Channel 00/04 :    0   1   2   3   4   5   6   7
+ip-26-0-160-225:94346:94798 [4] NCCL INFO Channel 01/04 :    0   1   2   3   4   5   6   7
+ip-26-0-160-225:94346:94798 [4] NCCL INFO Channel 02/04 :    0   1   2   3   4   5   6   7
+ip-26-0-160-225:94346:94798 [4] NCCL INFO Channel 03/04 :    0   1   2   3   4   5   6   7
+ip-26-0-173-7:811252:811692 [4] NCCL INFO Trees [0] -1/-1/-1->7->6 [1] -1/-1/-1->7->6 [2] 3/-1/-1->7->-1 [3] 3/-1/-1->7->-1
+ip-26-0-173-7:811252:811692 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-225:94346:94798 [4] NCCL INFO Trees [0] 4/-1/-1->0->-1 [1] 4/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-160-225:94346:94798 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-147:618585:619011 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-172-142:880681:881103 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-172-142:880682:881105 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-172-147:618584:619006 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-172-147:618591:619007 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-172-147:618590:619009 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-172-57:942617:943039 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-172-147:618587:619010 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-172-142:880685:881104 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-172-57:942614:943040 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-172-147:618586:619008 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-172-57:942611:943034 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-172-57:942613:943038 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-172-116:1019288:1019712 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-172-116:1019288:1019712 [0] NCCL INFO Trees [0] -1/-1/-1->1->2 [1] -1/-1/-1->1->2 [2] 2/0/-1->1->3 [3] 2/0/-1->1->3
+ip-26-0-172-116:1019288:1019712 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-142:880679:881101 [0] NCCL INFO Trees [0] 1/3/-1->2->4 [1] 1/3/-1->2->4 [2] -1/-1/-1->2->1 [3] -1/-1/-1->2->1
+ip-26-0-172-142:880679:881101 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-147:618584:619006 [0] NCCL INFO Trees [0] -1/-1/-1->3->2 [1] -1/-1/-1->3->2 [2] 5/1/-1->3->7 [3] 5/1/-1->3->7
+ip-26-0-172-147:618584:619006 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-252:688200:688625 [0] NCCL INFO Trees [0] 2/6/-1->4->0 [1] 2/6/-1->4->0 [2] -1/-1/-1->4->5 [3] -1/-1/-1->4->5
+ip-26-0-172-252:688200:688625 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-57:942611:943034 [0] NCCL INFO Trees [0] -1/-1/-1->5->6 [1] -1/-1/-1->5->6 [2] 6/4/-1->5->3 [3] 6/4/-1->5->3
+ip-26-0-172-57:942611:943034 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-225:94342:94802 [0] NCCL INFO Channel 00/04 :    0   1   2   3   4   5   6   7
+ip-26-0-160-225:94342:94802 [0] NCCL INFO Channel 01/04 :    0   1   2   3   4   5   6   7
+ip-26-0-160-225:94342:94802 [0] NCCL INFO Channel 02/04 :    0   1   2   3   4   5   6   7
+ip-26-0-160-225:94342:94802 [0] NCCL INFO Channel 03/04 :    0   1   2   3   4   5   6   7
+ip-26-0-160-225:94342:94802 [0] NCCL INFO Trees [0] 4/-1/-1->0->-1 [1] 4/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-160-225:94342:94802 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-173-7:811248:811694 [0] NCCL INFO Trees [0] -1/-1/-1->7->6 [1] -1/-1/-1->7->6 [2] 3/-1/-1->7->-1 [3] 3/-1/-1->7->-1
+ip-26-0-173-7:811248:811694 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-73:926594:927017 [0] NCCL INFO Trees [0] 5/7/-1->6->4 [1] 5/7/-1->6->4 [2] -1/-1/-1->6->5 [3] -1/-1/-1->6->5
+ip-26-0-172-73:926594:927017 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-116:1019295:1019713 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-172-116:1019295:1019713 [7] NCCL INFO Trees [0] -1/-1/-1->1->2 [1] -1/-1/-1->1->2 [2] 2/0/-1->1->3 [3] 2/0/-1->1->3
+ip-26-0-172-116:1019295:1019713 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-142:880686:881102 [7] NCCL INFO Trees [0] 1/3/-1->2->4 [1] 1/3/-1->2->4 [2] -1/-1/-1->2->1 [3] -1/-1/-1->2->1
+ip-26-0-172-142:880686:881102 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-147:618591:619007 [7] NCCL INFO Trees [0] -1/-1/-1->3->2 [1] -1/-1/-1->3->2 [2] 5/1/-1->3->7 [3] 5/1/-1->3->7
+ip-26-0-172-147:618591:619007 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-252:688207:688626 [7] NCCL INFO Trees [0] 2/6/-1->4->0 [1] 2/6/-1->4->0 [2] -1/-1/-1->4->5 [3] -1/-1/-1->4->5
+ip-26-0-172-252:688207:688626 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-57:942618:943035 [7] NCCL INFO Trees [0] -1/-1/-1->5->6 [1] -1/-1/-1->5->6 [2] 6/4/-1->5->3 [3] 6/4/-1->5->3
+ip-26-0-172-57:942618:943035 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-173-7:811255:811695 [7] NCCL INFO Trees [0] -1/-1/-1->7->6 [1] -1/-1/-1->7->6 [2] 3/-1/-1->7->-1 [3] 3/-1/-1->7->-1
+ip-26-0-173-7:811255:811695 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-73:926601:927018 [7] NCCL INFO Trees [0] 5/7/-1->6->4 [1] 5/7/-1->6->4 [2] -1/-1/-1->6->5 [3] -1/-1/-1->6->5
+ip-26-0-172-73:926601:927018 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-225:94349:94804 [7] NCCL INFO Channel 00/04 :    0   1   2   3   4   5   6   7
+ip-26-0-160-225:94349:94804 [7] NCCL INFO Channel 01/04 :    0   1   2   3   4   5   6   7
+ip-26-0-160-225:94349:94804 [7] NCCL INFO Channel 02/04 :    0   1   2   3   4   5   6   7
+ip-26-0-160-225:94349:94804 [7] NCCL INFO Channel 03/04 :    0   1   2   3   4   5   6   7
+ip-26-0-160-225:94349:94804 [7] NCCL INFO Trees [0] 4/-1/-1->0->-1 [1] 4/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-160-225:94349:94804 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-116:1019294:1019715 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-172-116:1019294:1019715 [6] NCCL INFO Trees [0] -1/-1/-1->1->2 [1] -1/-1/-1->1->2 [2] 2/0/-1->1->3 [3] 2/0/-1->1->3
+ip-26-0-172-116:1019294:1019715 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-142:880685:881104 [6] NCCL INFO Trees [0] 1/3/-1->2->4 [1] 1/3/-1->2->4 [2] -1/-1/-1->2->1 [3] -1/-1/-1->2->1
+ip-26-0-172-142:880685:881104 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-147:618590:619009 [6] NCCL INFO Trees [0] -1/-1/-1->3->2 [1] -1/-1/-1->3->2 [2] 5/1/-1->3->7 [3] 5/1/-1->3->7
+ip-26-0-172-147:618590:619009 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-252:688206:688628 [6] NCCL INFO Trees [0] 2/6/-1->4->0 [1] 2/6/-1->4->0 [2] -1/-1/-1->4->5 [3] -1/-1/-1->4->5
+ip-26-0-172-252:688206:688628 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-57:942617:943039 [6] NCCL INFO Trees [0] -1/-1/-1->5->6 [1] -1/-1/-1->5->6 [2] 6/4/-1->5->3 [3] 6/4/-1->5->3
+ip-26-0-172-57:942617:943039 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-73:926600:927020 [6] NCCL INFO Trees [0] 5/7/-1->6->4 [1] 5/7/-1->6->4 [2] -1/-1/-1->6->5 [3] -1/-1/-1->6->5
+ip-26-0-172-73:926600:927020 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-173-7:811254:811697 [6] NCCL INFO Trees [0] -1/-1/-1->7->6 [1] -1/-1/-1->7->6 [2] 3/-1/-1->7->-1 [3] 3/-1/-1->7->-1
+ip-26-0-173-7:811254:811697 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-225:94348:94808 [6] NCCL INFO Channel 00/04 :    0   1   2   3   4   5   6   7
+ip-26-0-160-225:94348:94808 [6] NCCL INFO Channel 01/04 :    0   1   2   3   4   5   6   7
+ip-26-0-160-225:94348:94808 [6] NCCL INFO Channel 02/04 :    0   1   2   3   4   5   6   7
+ip-26-0-160-225:94348:94808 [6] NCCL INFO Channel 03/04 :    0   1   2   3   4   5   6   7
+ip-26-0-160-225:94348:94808 [6] NCCL INFO Trees [0] 4/-1/-1->0->-1 [1] 4/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-160-225:94348:94808 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-116:1019290:1019714 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-172-116:1019290:1019714 [2] NCCL INFO Trees [0] -1/-1/-1->1->2 [1] -1/-1/-1->1->2 [2] 2/0/-1->1->3 [3] 2/0/-1->1->3
+ip-26-0-172-116:1019290:1019714 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-142:880681:881103 [2] NCCL INFO Trees [0] 1/3/-1->2->4 [1] 1/3/-1->2->4 [2] -1/-1/-1->2->1 [3] -1/-1/-1->2->1
+ip-26-0-172-142:880681:881103 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-147:618586:619008 [2] NCCL INFO Trees [0] -1/-1/-1->3->2 [1] -1/-1/-1->3->2 [2] 5/1/-1->3->7 [3] 5/1/-1->3->7
+ip-26-0-172-147:618586:619008 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-252:688202:688627 [2] NCCL INFO Trees [0] 2/6/-1->4->0 [1] 2/6/-1->4->0 [2] -1/-1/-1->4->5 [3] -1/-1/-1->4->5
+ip-26-0-172-252:688202:688627 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-57:942613:943038 [2] NCCL INFO Trees [0] -1/-1/-1->5->6 [1] -1/-1/-1->5->6 [2] 6/4/-1->5->3 [3] 6/4/-1->5->3
+ip-26-0-172-57:942613:943038 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-73:926596:927019 [2] NCCL INFO Trees [0] 5/7/-1->6->4 [1] 5/7/-1->6->4 [2] -1/-1/-1->6->5 [3] -1/-1/-1->6->5
+ip-26-0-172-73:926596:927019 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-173-7:811250:811696 [2] NCCL INFO Trees [0] -1/-1/-1->7->6 [1] -1/-1/-1->7->6 [2] 3/-1/-1->7->-1 [3] 3/-1/-1->7->-1
+ip-26-0-173-7:811250:811696 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-225:94344:94806 [2] NCCL INFO Channel 00/04 :    0   1   2   3   4   5   6   7
+ip-26-0-160-225:94344:94806 [2] NCCL INFO Channel 01/04 :    0   1   2   3   4   5   6   7
+ip-26-0-160-225:94344:94806 [2] NCCL INFO Channel 02/04 :    0   1   2   3   4   5   6   7
+ip-26-0-160-225:94344:94806 [2] NCCL INFO Channel 03/04 :    0   1   2   3   4   5   6   7
+ip-26-0-160-225:94344:94806 [2] NCCL INFO Trees [0] 4/-1/-1->0->-1 [1] 4/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-160-225:94344:94806 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-116:1019291:1019716 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-172-116:1019291:1019716 [3] NCCL INFO Trees [0] -1/-1/-1->1->2 [1] -1/-1/-1->1->2 [2] 2/0/-1->1->3 [3] 2/0/-1->1->3
+ip-26-0-172-116:1019291:1019716 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-142:880682:881105 [3] NCCL INFO Trees [0] 1/3/-1->2->4 [1] 1/3/-1->2->4 [2] -1/-1/-1->2->1 [3] -1/-1/-1->2->1
+ip-26-0-172-142:880682:881105 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-147:618587:619010 [3] NCCL INFO Trees [0] -1/-1/-1->3->2 [1] -1/-1/-1->3->2 [2] 5/1/-1->3->7 [3] 5/1/-1->3->7
+ip-26-0-172-147:618587:619010 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-252:688203:688629 [3] NCCL INFO Trees [0] 2/6/-1->4->0 [1] 2/6/-1->4->0 [2] -1/-1/-1->4->5 [3] -1/-1/-1->4->5
+ip-26-0-172-252:688203:688629 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-57:942614:943040 [3] NCCL INFO Trees [0] -1/-1/-1->5->6 [1] -1/-1/-1->5->6 [2] 6/4/-1->5->3 [3] 6/4/-1->5->3
+ip-26-0-172-57:942614:943040 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-73:926597:927021 [3] NCCL INFO Trees [0] 5/7/-1->6->4 [1] 5/7/-1->6->4 [2] -1/-1/-1->6->5 [3] -1/-1/-1->6->5
+ip-26-0-172-73:926597:927021 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-173-7:811251:811698 [3] NCCL INFO Trees [0] -1/-1/-1->7->6 [1] -1/-1/-1->7->6 [2] 3/-1/-1->7->-1 [3] 3/-1/-1->7->-1
+ip-26-0-173-7:811251:811698 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-225:94345:94810 [3] NCCL INFO Channel 00/04 :    0   1   2   3   4   5   6   7
+ip-26-0-160-225:94345:94810 [3] NCCL INFO Channel 01/04 :    0   1   2   3   4   5   6   7
+ip-26-0-160-225:94345:94810 [3] NCCL INFO Channel 02/04 :    0   1   2   3   4   5   6   7
+ip-26-0-160-225:94345:94810 [3] NCCL INFO Channel 03/04 :    0   1   2   3   4   5   6   7
+ip-26-0-160-225:94345:94810 [3] NCCL INFO Trees [0] 4/-1/-1->0->-1 [1] 4/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-160-225:94345:94810 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-225:94347:94800 [5] NCCL INFO Channel 00/0 : 7[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:94347:94800 [5] NCCL INFO Channel 01/0 : 7[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:94347:94800 [5] NCCL INFO Channel 02/0 : 7[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:94347:94800 [5] NCCL INFO Channel 03/0 : 7[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:94347:94800 [5] NCCL INFO Channel 00/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:94347:94800 [5] NCCL INFO Channel 01/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:94347:94800 [5] NCCL INFO Channel 02/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:94347:94800 [5] NCCL INFO Channel 03/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-7:811252:811692 [4] NCCL INFO Channel 00/0 : 6[4] -> 7[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-7:811252:811692 [4] NCCL INFO Channel 01/0 : 6[4] -> 7[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-7:811252:811692 [4] NCCL INFO Channel 02/0 : 6[4] -> 7[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-7:811252:811692 [4] NCCL INFO Channel 03/0 : 6[4] -> 7[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-7:811252:811692 [4] NCCL INFO Channel 00/0 : 7[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880684:881100 [5] NCCL INFO Channel 00/0 : 1[5] -> 2[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-7:811252:811692 [4] NCCL INFO Channel 01/0 : 7[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880684:881100 [5] NCCL INFO Channel 01/0 : 1[5] -> 2[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-7:811252:811692 [4] NCCL INFO Channel 02/0 : 7[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-7:811252:811692 [4] NCCL INFO Channel 03/0 : 7[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880684:881100 [5] NCCL INFO Channel 02/0 : 1[5] -> 2[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880684:881100 [5] NCCL INFO Channel 03/0 : 1[5] -> 2[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880684:881100 [5] NCCL INFO Channel 00/0 : 2[5] -> 3[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880684:881100 [5] NCCL INFO Channel 01/0 : 2[5] -> 3[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880684:881100 [5] NCCL INFO Channel 02/0 : 2[5] -> 3[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880684:881100 [5] NCCL INFO Channel 03/0 : 2[5] -> 3[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019289:1019717 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-160-225:94346:94798 [4] NCCL INFO Channel 00/0 : 7[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:94346:94798 [4] NCCL INFO Channel 01/0 : 7[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:94346:94798 [4] NCCL INFO Channel 02/0 : 7[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:94346:94798 [4] NCCL INFO Channel 03/0 : 7[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019293:1019711 [5] NCCL INFO Channel 00/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:94346:94798 [4] NCCL INFO Channel 00/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019293:1019711 [5] NCCL INFO Channel 01/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618589:619005 [5] NCCL INFO Channel 00/0 : 2[5] -> 3[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:94346:94798 [4] NCCL INFO Channel 01/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019293:1019711 [5] NCCL INFO Channel 02/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618589:619005 [5] NCCL INFO Channel 01/0 : 2[5] -> 3[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:94346:94798 [4] NCCL INFO Channel 02/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019293:1019711 [5] NCCL INFO Channel 03/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:94346:94798 [4] NCCL INFO Channel 03/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618589:619005 [5] NCCL INFO Channel 02/0 : 2[5] -> 3[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019293:1019711 [5] NCCL INFO Channel 00/0 : 1[5] -> 2[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880683:881099 [4] NCCL INFO Channel 00/0 : 1[4] -> 2[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-7:811253:811693 [5] NCCL INFO Channel 00/0 : 6[5] -> 7[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618589:619005 [5] NCCL INFO Channel 03/0 : 2[5] -> 3[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019293:1019711 [5] NCCL INFO Channel 01/0 : 1[5] -> 2[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618589:619005 [5] NCCL INFO Channel 00/0 : 3[5] -> 4[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-7:811253:811693 [5] NCCL INFO Channel 01/0 : 6[5] -> 7[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019293:1019711 [5] NCCL INFO Channel 02/0 : 1[5] -> 2[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880683:881099 [4] NCCL INFO Channel 01/0 : 1[4] -> 2[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618589:619005 [5] NCCL INFO Channel 01/0 : 3[5] -> 4[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019293:1019711 [5] NCCL INFO Channel 03/0 : 1[5] -> 2[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-7:811253:811693 [5] NCCL INFO Channel 02/0 : 6[5] -> 7[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880683:881099 [4] NCCL INFO Channel 02/0 : 1[4] -> 2[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618589:619005 [5] NCCL INFO Channel 02/0 : 3[5] -> 4[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880683:881099 [4] NCCL INFO Channel 03/0 : 1[4] -> 2[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-7:811253:811693 [5] NCCL INFO Channel 03/0 : 6[5] -> 7[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618589:619005 [5] NCCL INFO Channel 03/0 : 3[5] -> 4[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880683:881099 [4] NCCL INFO Channel 00/0 : 2[4] -> 3[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-7:811253:811693 [5] NCCL INFO Channel 00/0 : 7[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880683:881099 [4] NCCL INFO Channel 01/0 : 2[4] -> 3[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-7:811253:811693 [5] NCCL INFO Channel 01/0 : 7[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926599:927016 [5] NCCL INFO Channel 00/0 : 5[5] -> 6[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942616:943029 [5] NCCL INFO Channel 00/0 : 4[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880683:881099 [4] NCCL INFO Channel 02/0 : 2[4] -> 3[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-7:811253:811693 [5] NCCL INFO Channel 02/0 : 7[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926599:927016 [5] NCCL INFO Channel 01/0 : 5[5] -> 6[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-7:811253:811693 [5] NCCL INFO Channel 03/0 : 7[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942616:943029 [5] NCCL INFO Channel 01/0 : 4[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880683:881099 [4] NCCL INFO Channel 03/0 : 2[4] -> 3[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926599:927016 [5] NCCL INFO Channel 02/0 : 5[5] -> 6[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618588:619004 [4] NCCL INFO Channel 00/0 : 2[4] -> 3[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942616:943029 [5] NCCL INFO Channel 02/0 : 4[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926599:927016 [5] NCCL INFO Channel 03/0 : 5[5] -> 6[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942616:943029 [5] NCCL INFO Channel 03/0 : 4[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618588:619004 [4] NCCL INFO Channel 01/0 : 2[4] -> 3[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926599:927016 [5] NCCL INFO Channel 00/0 : 6[5] -> 7[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019289:1019717 [1] NCCL INFO Trees [0] -1/-1/-1->1->2 [1] -1/-1/-1->1->2 [2] 2/0/-1->1->3 [3] 2/0/-1->1->3
+ip-26-0-172-116:1019289:1019717 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-57:942616:943029 [5] NCCL INFO Channel 00/0 : 5[5] -> 6[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926599:927016 [5] NCCL INFO Channel 01/0 : 6[5] -> 7[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618588:619004 [4] NCCL INFO Channel 02/0 : 2[4] -> 3[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880680:881106 [1] NCCL INFO Trees [0] 1/3/-1->2->4 [1] 1/3/-1->2->4 [2] -1/-1/-1->2->1 [3] -1/-1/-1->2->1
+ip-26-0-172-142:880680:881106 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-57:942616:943029 [5] NCCL INFO Channel 01/0 : 5[5] -> 6[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618585:619011 [1] NCCL INFO Trees [0] -1/-1/-1->3->2 [1] -1/-1/-1->3->2 [2] 5/1/-1->3->7 [3] 5/1/-1->3->7
+ip-26-0-172-147:618585:619011 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-73:926599:927016 [5] NCCL INFO Channel 02/0 : 6[5] -> 7[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688201:688630 [1] NCCL INFO Trees [0] 2/6/-1->4->0 [1] 2/6/-1->4->0 [2] -1/-1/-1->4->5 [3] -1/-1/-1->4->5
+ip-26-0-172-252:688205:688624 [5] NCCL INFO Channel 00/0 : 3[5] -> 4[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688201:688630 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-57:942616:943029 [5] NCCL INFO Channel 02/0 : 5[5] -> 6[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618588:619004 [4] NCCL INFO Channel 03/0 : 2[4] -> 3[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926599:927016 [5] NCCL INFO Channel 03/0 : 6[5] -> 7[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942612:943041 [1] NCCL INFO Trees [0] -1/-1/-1->5->6 [1] -1/-1/-1->5->6 [2] 6/4/-1->5->3 [3] 6/4/-1->5->3
+ip-26-0-172-57:942612:943041 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-147:618588:619004 [4] NCCL INFO Channel 00/0 : 3[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688205:688624 [5] NCCL INFO Channel 01/0 : 3[5] -> 4[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:94343:94812 [1] NCCL INFO Channel 00/04 :    0   1   2   3   4   5   6   7
+ip-26-0-160-225:94343:94812 [1] NCCL INFO Channel 01/04 :    0   1   2   3   4   5   6   7
+ip-26-0-160-225:94343:94812 [1] NCCL INFO Channel 02/04 :    0   1   2   3   4   5   6   7
+ip-26-0-160-225:94343:94812 [1] NCCL INFO Channel 03/04 :    0   1   2   3   4   5   6   7
+ip-26-0-160-225:94343:94812 [1] NCCL INFO Trees [0] 4/-1/-1->0->-1 [1] 4/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-160-225:94343:94812 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-57:942615:943026 [4] NCCL INFO Channel 00/0 : 4[4] -> 5[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926595:927022 [1] NCCL INFO Trees [0] 5/7/-1->6->4 [1] 5/7/-1->6->4 [2] -1/-1/-1->6->5 [3] -1/-1/-1->6->5
+ip-26-0-172-73:926595:927022 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-173-7:811249:811699 [1] NCCL INFO Trees [0] -1/-1/-1->7->6 [1] -1/-1/-1->7->6 [2] 3/-1/-1->7->-1 [3] 3/-1/-1->7->-1
+ip-26-0-173-7:811249:811699 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-147:618588:619004 [4] NCCL INFO Channel 01/0 : 3[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688205:688624 [5] NCCL INFO Channel 02/0 : 3[5] -> 4[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942616:943029 [5] NCCL INFO Channel 03/0 : 5[5] -> 6[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942615:943026 [4] NCCL INFO Channel 01/0 : 4[4] -> 5[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618588:619004 [4] NCCL INFO Channel 02/0 : 3[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688205:688624 [5] NCCL INFO Channel 03/0 : 3[5] -> 4[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942615:943026 [4] NCCL INFO Channel 02/0 : 4[4] -> 5[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618588:619004 [4] NCCL INFO Channel 03/0 : 3[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942615:943026 [4] NCCL INFO Channel 03/0 : 4[4] -> 5[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688205:688624 [5] NCCL INFO Channel 00/0 : 4[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942615:943026 [4] NCCL INFO Channel 00/0 : 5[4] -> 6[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688205:688624 [5] NCCL INFO Channel 01/0 : 4[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942615:943026 [4] NCCL INFO Channel 01/0 : 5[4] -> 6[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688205:688624 [5] NCCL INFO Channel 02/0 : 4[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942615:943026 [4] NCCL INFO Channel 02/0 : 5[4] -> 6[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688205:688624 [5] NCCL INFO Channel 03/0 : 4[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942615:943026 [4] NCCL INFO Channel 03/0 : 5[4] -> 6[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926598:927015 [4] NCCL INFO Channel 00/0 : 5[4] -> 6[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926598:927015 [4] NCCL INFO Channel 01/0 : 5[4] -> 6[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926598:927015 [4] NCCL INFO Channel 02/0 : 5[4] -> 6[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926598:927015 [4] NCCL INFO Channel 03/0 : 5[4] -> 6[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926598:927015 [4] NCCL INFO Channel 00/0 : 6[4] -> 7[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926598:927015 [4] NCCL INFO Channel 01/0 : 6[4] -> 7[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926598:927015 [4] NCCL INFO Channel 02/0 : 6[4] -> 7[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926598:927015 [4] NCCL INFO Channel 03/0 : 6[4] -> 7[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688204:688623 [4] NCCL INFO Channel 00/0 : 3[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688204:688623 [4] NCCL INFO Channel 01/0 : 3[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019292:1019710 [4] NCCL INFO Channel 00/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688204:688623 [4] NCCL INFO Channel 02/0 : 3[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019292:1019710 [4] NCCL INFO Channel 01/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688204:688623 [4] NCCL INFO Channel 03/0 : 3[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019292:1019710 [4] NCCL INFO Channel 02/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688204:688623 [4] NCCL INFO Channel 00/0 : 4[4] -> 5[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019292:1019710 [4] NCCL INFO Channel 03/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688204:688623 [4] NCCL INFO Channel 01/0 : 4[4] -> 5[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019292:1019710 [4] NCCL INFO Channel 00/0 : 1[4] -> 2[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688204:688623 [4] NCCL INFO Channel 02/0 : 4[4] -> 5[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019292:1019710 [4] NCCL INFO Channel 01/0 : 1[4] -> 2[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688204:688623 [4] NCCL INFO Channel 03/0 : 4[4] -> 5[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019292:1019710 [4] NCCL INFO Channel 02/0 : 1[4] -> 2[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019292:1019710 [4] NCCL INFO Channel 03/0 : 1[4] -> 2[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618584:619006 [0] NCCL INFO Channel 00/0 : 2[0] -> 3[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618584:619006 [0] NCCL INFO Channel 01/0 : 2[0] -> 3[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688200:688625 [0] NCCL INFO Channel 00/0 : 3[0] -> 4[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618584:619006 [0] NCCL INFO Channel 02/0 : 2[0] -> 3[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618584:619006 [0] NCCL INFO Channel 03/0 : 2[0] -> 3[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688200:688625 [0] NCCL INFO Channel 01/0 : 3[0] -> 4[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:94342:94802 [0] NCCL INFO Channel 00/0 : 7[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618584:619006 [0] NCCL INFO Channel 00/0 : 3[0] -> 4[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688200:688625 [0] NCCL INFO Channel 02/0 : 3[0] -> 4[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618584:619006 [0] NCCL INFO Channel 01/0 : 3[0] -> 4[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:94342:94802 [0] NCCL INFO Channel 01/0 : 7[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688200:688625 [0] NCCL INFO Channel 03/0 : 3[0] -> 4[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618584:619006 [0] NCCL INFO Channel 02/0 : 3[0] -> 4[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688200:688625 [0] NCCL INFO Channel 00/0 : 4[0] -> 5[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:94342:94802 [0] NCCL INFO Channel 02/0 : 7[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618584:619006 [0] NCCL INFO Channel 03/0 : 3[0] -> 4[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688200:688625 [0] NCCL INFO Channel 01/0 : 4[0] -> 5[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:94342:94802 [0] NCCL INFO Channel 03/0 : 7[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688200:688625 [0] NCCL INFO Channel 02/0 : 4[0] -> 5[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:94342:94802 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-7:811248:811694 [0] NCCL INFO Channel 00/0 : 6[0] -> 7[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688200:688625 [0] NCCL INFO Channel 03/0 : 4[0] -> 5[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:94342:94802 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:94342:94802 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-7:811248:811694 [0] NCCL INFO Channel 01/0 : 6[0] -> 7[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:94342:94802 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-7:811248:811694 [0] NCCL INFO Channel 02/0 : 6[0] -> 7[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880679:881101 [0] NCCL INFO Channel 00/0 : 1[0] -> 2[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-7:811248:811694 [0] NCCL INFO Channel 03/0 : 6[0] -> 7[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880679:881101 [0] NCCL INFO Channel 01/0 : 1[0] -> 2[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-7:811248:811694 [0] NCCL INFO Channel 00/0 : 7[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880679:881101 [0] NCCL INFO Channel 02/0 : 1[0] -> 2[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-7:811248:811694 [0] NCCL INFO Channel 01/0 : 7[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880679:881101 [0] NCCL INFO Channel 03/0 : 1[0] -> 2[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-7:811248:811694 [0] NCCL INFO Channel 02/0 : 7[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880679:881101 [0] NCCL INFO Channel 00/0 : 2[0] -> 3[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-7:811248:811694 [0] NCCL INFO Channel 03/0 : 7[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880679:881101 [0] NCCL INFO Channel 01/0 : 2[0] -> 3[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880679:881101 [0] NCCL INFO Channel 02/0 : 2[0] -> 3[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880679:881101 [0] NCCL INFO Channel 03/0 : 2[0] -> 3[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:94349:94804 [7] NCCL INFO Channel 00/0 : 7[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:94349:94804 [7] NCCL INFO Channel 01/0 : 7[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:94349:94804 [7] NCCL INFO Channel 02/0 : 7[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:94349:94804 [7] NCCL INFO Channel 03/0 : 7[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:94349:94804 [7] NCCL INFO Channel 00/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:94349:94804 [7] NCCL INFO Channel 01/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942611:943034 [0] NCCL INFO Channel 00/0 : 4[0] -> 5[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:94349:94804 [7] NCCL INFO Channel 02/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-7:811255:811695 [7] NCCL INFO Channel 00/0 : 6[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942611:943034 [0] NCCL INFO Channel 01/0 : 4[0] -> 5[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:94349:94804 [7] NCCL INFO Channel 03/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942611:943034 [0] NCCL INFO Channel 02/0 : 4[0] -> 5[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-7:811255:811695 [7] NCCL INFO Channel 01/0 : 6[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880686:881102 [7] NCCL INFO Channel 00/0 : 1[7] -> 2[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942611:943034 [0] NCCL INFO Channel 03/0 : 4[0] -> 5[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-7:811255:811695 [7] NCCL INFO Channel 02/0 : 6[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942611:943034 [0] NCCL INFO Channel 00/0 : 5[0] -> 6[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880686:881102 [7] NCCL INFO Channel 01/0 : 1[7] -> 2[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-7:811255:811695 [7] NCCL INFO Channel 03/0 : 6[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942611:943034 [0] NCCL INFO Channel 01/0 : 5[0] -> 6[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880686:881102 [7] NCCL INFO Channel 02/0 : 1[7] -> 2[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-7:811255:811695 [7] NCCL INFO Channel 00/0 : 7[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618591:619007 [7] NCCL INFO Channel 00/0 : 2[7] -> 3[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942611:943034 [0] NCCL INFO Channel 02/0 : 5[0] -> 6[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880686:881102 [7] NCCL INFO Channel 03/0 : 1[7] -> 2[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-7:811255:811695 [7] NCCL INFO Channel 01/0 : 7[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942611:943034 [0] NCCL INFO Channel 03/0 : 5[0] -> 6[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618591:619007 [7] NCCL INFO Channel 01/0 : 2[7] -> 3[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019295:1019713 [7] NCCL INFO Channel 00/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880686:881102 [7] NCCL INFO Channel 00/0 : 2[7] -> 3[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-7:811255:811695 [7] NCCL INFO Channel 02/0 : 7[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618591:619007 [7] NCCL INFO Channel 02/0 : 2[7] -> 3[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880686:881102 [7] NCCL INFO Channel 01/0 : 2[7] -> 3[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019295:1019713 [7] NCCL INFO Channel 01/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-7:811255:811695 [7] NCCL INFO Channel 03/0 : 7[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618591:619007 [7] NCCL INFO Channel 03/0 : 2[7] -> 3[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880686:881102 [7] NCCL INFO Channel 02/0 : 2[7] -> 3[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926601:927018 [7] NCCL INFO Channel 00/0 : 5[7] -> 6[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019295:1019713 [7] NCCL INFO Channel 02/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618591:619007 [7] NCCL INFO Channel 00/0 : 3[7] -> 4[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880686:881102 [7] NCCL INFO Channel 03/0 : 2[7] -> 3[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942618:943035 [7] NCCL INFO Channel 00/0 : 4[7] -> 5[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019295:1019713 [7] NCCL INFO Channel 03/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618591:619007 [7] NCCL INFO Channel 01/0 : 3[7] -> 4[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926601:927018 [7] NCCL INFO Channel 01/0 : 5[7] -> 6[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019295:1019713 [7] NCCL INFO Channel 00/0 : 1[7] -> 2[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618591:619007 [7] NCCL INFO Channel 02/0 : 3[7] -> 4[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942618:943035 [7] NCCL INFO Channel 01/0 : 4[7] -> 5[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926601:927018 [7] NCCL INFO Channel 02/0 : 5[7] -> 6[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019295:1019713 [7] NCCL INFO Channel 01/0 : 1[7] -> 2[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618591:619007 [7] NCCL INFO Channel 03/0 : 3[7] -> 4[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942618:943035 [7] NCCL INFO Channel 02/0 : 4[7] -> 5[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926601:927018 [7] NCCL INFO Channel 03/0 : 5[7] -> 6[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019288:1019712 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942618:943035 [7] NCCL INFO Channel 03/0 : 4[7] -> 5[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926601:927018 [7] NCCL INFO Channel 00/0 : 6[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019295:1019713 [7] NCCL INFO Channel 02/0 : 1[7] -> 2[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942618:943035 [7] NCCL INFO Channel 00/0 : 5[7] -> 6[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019295:1019713 [7] NCCL INFO Channel 03/0 : 1[7] -> 2[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019288:1019712 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926601:927018 [7] NCCL INFO Channel 01/0 : 6[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942618:943035 [7] NCCL INFO Channel 01/0 : 5[7] -> 6[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019288:1019712 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926601:927018 [7] NCCL INFO Channel 02/0 : 6[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942618:943035 [7] NCCL INFO Channel 02/0 : 5[7] -> 6[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019288:1019712 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926601:927018 [7] NCCL INFO Channel 03/0 : 6[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942618:943035 [7] NCCL INFO Channel 03/0 : 5[7] -> 6[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019288:1019712 [0] NCCL INFO Channel 00/0 : 1[0] -> 2[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019288:1019712 [0] NCCL INFO Channel 01/0 : 1[0] -> 2[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019288:1019712 [0] NCCL INFO Channel 02/0 : 1[0] -> 2[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019288:1019712 [0] NCCL INFO Channel 03/0 : 1[0] -> 2[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926594:927017 [0] NCCL INFO Channel 00/0 : 5[0] -> 6[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926594:927017 [0] NCCL INFO Channel 01/0 : 5[0] -> 6[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926594:927017 [0] NCCL INFO Channel 02/0 : 5[0] -> 6[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688207:688626 [7] NCCL INFO Channel 00/0 : 3[7] -> 4[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926594:927017 [0] NCCL INFO Channel 03/0 : 5[0] -> 6[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926594:927017 [0] NCCL INFO Channel 00/0 : 6[0] -> 7[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688207:688626 [7] NCCL INFO Channel 01/0 : 3[7] -> 4[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926594:927017 [0] NCCL INFO Channel 01/0 : 6[0] -> 7[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688207:688626 [7] NCCL INFO Channel 02/0 : 3[7] -> 4[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926594:927017 [0] NCCL INFO Channel 02/0 : 6[0] -> 7[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688207:688626 [7] NCCL INFO Channel 03/0 : 3[7] -> 4[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926594:927017 [0] NCCL INFO Channel 03/0 : 6[0] -> 7[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688207:688626 [7] NCCL INFO Channel 00/0 : 4[7] -> 5[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688207:688626 [7] NCCL INFO Channel 01/0 : 4[7] -> 5[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688207:688626 [7] NCCL INFO Channel 02/0 : 4[7] -> 5[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688207:688626 [7] NCCL INFO Channel 03/0 : 4[7] -> 5[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880685:881104 [6] NCCL INFO Channel 00/0 : 1[6] -> 2[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880685:881104 [6] NCCL INFO Channel 01/0 : 1[6] -> 2[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880685:881104 [6] NCCL INFO Channel 02/0 : 1[6] -> 2[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880685:881104 [6] NCCL INFO Channel 03/0 : 1[6] -> 2[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-7:811254:811697 [6] NCCL INFO Channel 00/0 : 6[6] -> 7[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880685:881104 [6] NCCL INFO Channel 00/0 : 2[6] -> 3[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-7:811254:811697 [6] NCCL INFO Channel 01/0 : 6[6] -> 7[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880685:881104 [6] NCCL INFO Channel 01/0 : 2[6] -> 3[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-7:811254:811697 [6] NCCL INFO Channel 02/0 : 6[6] -> 7[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880685:881104 [6] NCCL INFO Channel 02/0 : 2[6] -> 3[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-7:811254:811697 [6] NCCL INFO Channel 03/0 : 6[6] -> 7[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880685:881104 [6] NCCL INFO Channel 03/0 : 2[6] -> 3[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-7:811254:811697 [6] NCCL INFO Channel 00/0 : 7[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:94348:94808 [6] NCCL INFO Channel 00/0 : 7[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-7:811254:811697 [6] NCCL INFO Channel 01/0 : 7[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:94348:94808 [6] NCCL INFO Channel 01/0 : 7[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-7:811254:811697 [6] NCCL INFO Channel 02/0 : 7[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:94348:94808 [6] NCCL INFO Channel 02/0 : 7[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-7:811254:811697 [6] NCCL INFO Channel 03/0 : 7[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:94348:94808 [6] NCCL INFO Channel 03/0 : 7[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:94348:94808 [6] NCCL INFO Channel 00/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:94348:94808 [6] NCCL INFO Channel 01/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:94348:94808 [6] NCCL INFO Channel 02/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:94348:94808 [6] NCCL INFO Channel 03/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618590:619009 [6] NCCL INFO Channel 00/0 : 2[6] -> 3[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618590:619009 [6] NCCL INFO Channel 01/0 : 2[6] -> 3[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019294:1019715 [6] NCCL INFO Channel 00/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688206:688628 [6] NCCL INFO Channel 00/0 : 3[6] -> 4[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618590:619009 [6] NCCL INFO Channel 02/0 : 2[6] -> 3[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019294:1019715 [6] NCCL INFO Channel 01/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618590:619009 [6] NCCL INFO Channel 03/0 : 2[6] -> 3[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688206:688628 [6] NCCL INFO Channel 01/0 : 3[6] -> 4[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019294:1019715 [6] NCCL INFO Channel 02/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618590:619009 [6] NCCL INFO Channel 00/0 : 3[6] -> 4[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019294:1019715 [6] NCCL INFO Channel 03/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688206:688628 [6] NCCL INFO Channel 02/0 : 3[6] -> 4[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618590:619009 [6] NCCL INFO Channel 01/0 : 3[6] -> 4[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019294:1019715 [6] NCCL INFO Channel 00/0 : 1[6] -> 2[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688206:688628 [6] NCCL INFO Channel 03/0 : 3[6] -> 4[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019294:1019715 [6] NCCL INFO Channel 01/0 : 1[6] -> 2[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618590:619009 [6] NCCL INFO Channel 02/0 : 3[6] -> 4[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688206:688628 [6] NCCL INFO Channel 00/0 : 4[6] -> 5[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019294:1019715 [6] NCCL INFO Channel 02/0 : 1[6] -> 2[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618590:619009 [6] NCCL INFO Channel 03/0 : 3[6] -> 4[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688206:688628 [6] NCCL INFO Channel 01/0 : 4[6] -> 5[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019294:1019715 [6] NCCL INFO Channel 03/0 : 1[6] -> 2[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942617:943039 [6] NCCL INFO Channel 00/0 : 4[6] -> 5[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688206:688628 [6] NCCL INFO Channel 02/0 : 4[6] -> 5[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942617:943039 [6] NCCL INFO Channel 01/0 : 4[6] -> 5[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688206:688628 [6] NCCL INFO Channel 03/0 : 4[6] -> 5[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942617:943039 [6] NCCL INFO Channel 02/0 : 4[6] -> 5[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942617:943039 [6] NCCL INFO Channel 03/0 : 4[6] -> 5[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942617:943039 [6] NCCL INFO Channel 00/0 : 5[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942617:943039 [6] NCCL INFO Channel 01/0 : 5[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942617:943039 [6] NCCL INFO Channel 02/0 : 5[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942617:943039 [6] NCCL INFO Channel 03/0 : 5[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926600:927020 [6] NCCL INFO Channel 00/0 : 5[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926600:927020 [6] NCCL INFO Channel 01/0 : 5[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926600:927020 [6] NCCL INFO Channel 02/0 : 5[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926600:927020 [6] NCCL INFO Channel 03/0 : 5[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926600:927020 [6] NCCL INFO Channel 00/0 : 6[6] -> 7[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926600:927020 [6] NCCL INFO Channel 01/0 : 6[6] -> 7[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926600:927020 [6] NCCL INFO Channel 02/0 : 6[6] -> 7[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926600:927020 [6] NCCL INFO Channel 03/0 : 6[6] -> 7[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-7:811250:811696 [2] NCCL INFO Channel 00/0 : 6[2] -> 7[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-7:811250:811696 [2] NCCL INFO Channel 01/0 : 6[2] -> 7[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-7:811250:811696 [2] NCCL INFO Channel 02/0 : 6[2] -> 7[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-7:811250:811696 [2] NCCL INFO Channel 03/0 : 6[2] -> 7[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-7:811250:811696 [2] NCCL INFO Channel 00/0 : 7[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:94344:94806 [2] NCCL INFO Channel 00/0 : 7[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-7:811250:811696 [2] NCCL INFO Channel 01/0 : 7[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:94344:94806 [2] NCCL INFO Channel 01/0 : 7[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-7:811250:811696 [2] NCCL INFO Channel 02/0 : 7[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:94344:94806 [2] NCCL INFO Channel 02/0 : 7[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-7:811250:811696 [2] NCCL INFO Channel 03/0 : 7[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:94344:94806 [2] NCCL INFO Channel 03/0 : 7[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:94344:94806 [2] NCCL INFO Channel 00/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:94344:94806 [2] NCCL INFO Channel 01/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:94344:94806 [2] NCCL INFO Channel 02/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:94344:94806 [2] NCCL INFO Channel 03/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688202:688627 [2] NCCL INFO Channel 00/0 : 3[2] -> 4[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618586:619008 [2] NCCL INFO Channel 00/0 : 2[2] -> 3[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688202:688627 [2] NCCL INFO Channel 01/0 : 3[2] -> 4[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618586:619008 [2] NCCL INFO Channel 01/0 : 2[2] -> 3[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688202:688627 [2] NCCL INFO Channel 02/0 : 3[2] -> 4[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019290:1019714 [2] NCCL INFO Channel 00/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618586:619008 [2] NCCL INFO Channel 02/0 : 2[2] -> 3[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688202:688627 [2] NCCL INFO Channel 03/0 : 3[2] -> 4[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618586:619008 [2] NCCL INFO Channel 03/0 : 2[2] -> 3[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688202:688627 [2] NCCL INFO Channel 00/0 : 4[2] -> 5[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618586:619008 [2] NCCL INFO Channel 00/0 : 3[2] -> 4[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688202:688627 [2] NCCL INFO Channel 01/0 : 4[2] -> 5[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019290:1019714 [2] NCCL INFO Channel 01/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688202:688627 [2] NCCL INFO Channel 02/0 : 4[2] -> 5[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942613:943038 [2] NCCL INFO Channel 00/0 : 4[2] -> 5[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618586:619008 [2] NCCL INFO Channel 01/0 : 3[2] -> 4[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019290:1019714 [2] NCCL INFO Channel 02/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688202:688627 [2] NCCL INFO Channel 03/0 : 4[2] -> 5[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618586:619008 [2] NCCL INFO Channel 02/0 : 3[2] -> 4[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019290:1019714 [2] NCCL INFO Channel 03/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942613:943038 [2] NCCL INFO Channel 01/0 : 4[2] -> 5[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618586:619008 [2] NCCL INFO Channel 03/0 : 3[2] -> 4[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019290:1019714 [2] NCCL INFO Channel 00/0 : 1[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880681:881103 [2] NCCL INFO Channel 00/0 : 1[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942613:943038 [2] NCCL INFO Channel 02/0 : 4[2] -> 5[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019290:1019714 [2] NCCL INFO Channel 01/0 : 1[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880681:881103 [2] NCCL INFO Channel 01/0 : 1[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942613:943038 [2] NCCL INFO Channel 03/0 : 4[2] -> 5[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019290:1019714 [2] NCCL INFO Channel 02/0 : 1[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880681:881103 [2] NCCL INFO Channel 02/0 : 1[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942613:943038 [2] NCCL INFO Channel 00/0 : 5[2] -> 6[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019290:1019714 [2] NCCL INFO Channel 03/0 : 1[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880681:881103 [2] NCCL INFO Channel 03/0 : 1[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942613:943038 [2] NCCL INFO Channel 01/0 : 5[2] -> 6[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880681:881103 [2] NCCL INFO Channel 00/0 : 2[2] -> 3[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:94345:94810 [3] NCCL INFO Channel 00/0 : 7[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880681:881103 [2] NCCL INFO Channel 01/0 : 2[2] -> 3[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942613:943038 [2] NCCL INFO Channel 02/0 : 5[2] -> 6[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880681:881103 [2] NCCL INFO Channel 02/0 : 2[2] -> 3[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942613:943038 [2] NCCL INFO Channel 03/0 : 5[2] -> 6[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:94345:94810 [3] NCCL INFO Channel 01/0 : 7[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880681:881103 [2] NCCL INFO Channel 03/0 : 2[2] -> 3[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:94345:94810 [3] NCCL INFO Channel 02/0 : 7[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:94345:94810 [3] NCCL INFO Channel 03/0 : 7[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:94345:94810 [3] NCCL INFO Channel 00/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-7:811251:811698 [3] NCCL INFO Channel 00/0 : 6[3] -> 7[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:94345:94810 [3] NCCL INFO Channel 01/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-7:811251:811698 [3] NCCL INFO Channel 01/0 : 6[3] -> 7[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:94345:94810 [3] NCCL INFO Channel 02/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-7:811251:811698 [3] NCCL INFO Channel 02/0 : 6[3] -> 7[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:94345:94810 [3] NCCL INFO Channel 03/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-7:811251:811698 [3] NCCL INFO Channel 03/0 : 6[3] -> 7[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-7:811251:811698 [3] NCCL INFO Channel 00/0 : 7[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-7:811251:811698 [3] NCCL INFO Channel 01/0 : 7[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-7:811251:811698 [3] NCCL INFO Channel 02/0 : 7[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-7:811251:811698 [3] NCCL INFO Channel 03/0 : 7[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688203:688629 [3] NCCL INFO Channel 00/0 : 3[3] -> 4[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688203:688629 [3] NCCL INFO Channel 01/0 : 3[3] -> 4[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688203:688629 [3] NCCL INFO Channel 02/0 : 3[3] -> 4[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688203:688629 [3] NCCL INFO Channel 03/0 : 3[3] -> 4[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926596:927019 [2] NCCL INFO Channel 00/0 : 5[2] -> 6[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688203:688629 [3] NCCL INFO Channel 00/0 : 4[3] -> 5[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926596:927019 [2] NCCL INFO Channel 01/0 : 5[2] -> 6[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688203:688629 [3] NCCL INFO Channel 01/0 : 4[3] -> 5[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926596:927019 [2] NCCL INFO Channel 02/0 : 5[2] -> 6[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688203:688629 [3] NCCL INFO Channel 02/0 : 4[3] -> 5[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926596:927019 [2] NCCL INFO Channel 03/0 : 5[2] -> 6[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688203:688629 [3] NCCL INFO Channel 03/0 : 4[3] -> 5[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019291:1019716 [3] NCCL INFO Channel 00/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926596:927019 [2] NCCL INFO Channel 00/0 : 6[2] -> 7[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926596:927019 [2] NCCL INFO Channel 01/0 : 6[2] -> 7[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019291:1019716 [3] NCCL INFO Channel 01/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926596:927019 [2] NCCL INFO Channel 02/0 : 6[2] -> 7[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019291:1019716 [3] NCCL INFO Channel 02/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926596:927019 [2] NCCL INFO Channel 03/0 : 6[2] -> 7[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019291:1019716 [3] NCCL INFO Channel 03/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019291:1019716 [3] NCCL INFO Channel 00/0 : 1[3] -> 2[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019291:1019716 [3] NCCL INFO Channel 01/0 : 1[3] -> 2[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019291:1019716 [3] NCCL INFO Channel 02/0 : 1[3] -> 2[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019291:1019716 [3] NCCL INFO Channel 03/0 : 1[3] -> 2[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880682:881105 [3] NCCL INFO Channel 00/0 : 1[3] -> 2[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880682:881105 [3] NCCL INFO Channel 01/0 : 1[3] -> 2[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880682:881105 [3] NCCL INFO Channel 02/0 : 1[3] -> 2[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880682:881105 [3] NCCL INFO Channel 03/0 : 1[3] -> 2[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880682:881105 [3] NCCL INFO Channel 00/0 : 2[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880682:881105 [3] NCCL INFO Channel 01/0 : 2[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880682:881105 [3] NCCL INFO Channel 02/0 : 2[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880682:881105 [3] NCCL INFO Channel 03/0 : 2[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618587:619010 [3] NCCL INFO Channel 00/0 : 2[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942614:943040 [3] NCCL INFO Channel 00/0 : 4[3] -> 5[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618587:619010 [3] NCCL INFO Channel 01/0 : 2[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618587:619010 [3] NCCL INFO Channel 02/0 : 2[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942614:943040 [3] NCCL INFO Channel 01/0 : 4[3] -> 5[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618587:619010 [3] NCCL INFO Channel 03/0 : 2[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618587:619010 [3] NCCL INFO Channel 00/0 : 3[3] -> 4[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618587:619010 [3] NCCL INFO Channel 01/0 : 3[3] -> 4[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618587:619010 [3] NCCL INFO Channel 02/0 : 3[3] -> 4[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618587:619010 [3] NCCL INFO Channel 03/0 : 3[3] -> 4[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942614:943040 [3] NCCL INFO Channel 02/0 : 4[3] -> 5[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926597:927021 [3] NCCL INFO Channel 00/0 : 5[3] -> 6[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926597:927021 [3] NCCL INFO Channel 01/0 : 5[3] -> 6[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926597:927021 [3] NCCL INFO Channel 02/0 : 5[3] -> 6[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926597:927021 [3] NCCL INFO Channel 03/0 : 5[3] -> 6[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926597:927021 [3] NCCL INFO Channel 00/0 : 6[3] -> 7[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942614:943040 [3] NCCL INFO Channel 03/0 : 4[3] -> 5[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926597:927021 [3] NCCL INFO Channel 01/0 : 6[3] -> 7[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926597:927021 [3] NCCL INFO Channel 02/0 : 6[3] -> 7[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926597:927021 [3] NCCL INFO Channel 03/0 : 6[3] -> 7[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942614:943040 [3] NCCL INFO Channel 00/0 : 5[3] -> 6[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942614:943040 [3] NCCL INFO Channel 01/0 : 5[3] -> 6[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-7:811249:811699 [1] NCCL INFO Channel 00/0 : 6[1] -> 7[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-7:811249:811699 [1] NCCL INFO Channel 01/0 : 6[1] -> 7[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-7:811249:811699 [1] NCCL INFO Channel 02/0 : 6[1] -> 7[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-7:811249:811699 [1] NCCL INFO Channel 03/0 : 6[1] -> 7[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-7:811249:811699 [1] NCCL INFO Channel 00/0 : 7[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-7:811249:811699 [1] NCCL INFO Channel 01/0 : 7[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-7:811249:811699 [1] NCCL INFO Channel 02/0 : 7[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:94343:94812 [1] NCCL INFO Channel 00/0 : 7[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-7:811249:811699 [1] NCCL INFO Channel 03/0 : 7[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880680:881106 [1] NCCL INFO Channel 00/0 : 1[1] -> 2[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:94343:94812 [1] NCCL INFO Channel 01/0 : 7[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880680:881106 [1] NCCL INFO Channel 01/0 : 1[1] -> 2[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:94343:94812 [1] NCCL INFO Channel 02/0 : 7[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:94343:94812 [1] NCCL INFO Channel 03/0 : 7[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880680:881106 [1] NCCL INFO Channel 02/0 : 1[1] -> 2[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:94343:94812 [1] NCCL INFO Channel 00/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880680:881106 [1] NCCL INFO Channel 03/0 : 1[1] -> 2[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:94343:94812 [1] NCCL INFO Channel 01/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880680:881106 [1] NCCL INFO Channel 00/0 : 2[1] -> 3[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688201:688630 [1] NCCL INFO Channel 00/0 : 3[1] -> 4[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:94343:94812 [1] NCCL INFO Channel 02/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880680:881106 [1] NCCL INFO Channel 01/0 : 2[1] -> 3[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942614:943040 [3] NCCL INFO Channel 02/0 : 5[3] -> 6[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688201:688630 [1] NCCL INFO Channel 01/0 : 3[1] -> 4[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:94343:94812 [1] NCCL INFO Channel 03/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942614:943040 [3] NCCL INFO Channel 03/0 : 5[3] -> 6[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880680:881106 [1] NCCL INFO Channel 02/0 : 2[1] -> 3[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688201:688630 [1] NCCL INFO Channel 02/0 : 3[1] -> 4[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880680:881106 [1] NCCL INFO Channel 03/0 : 2[1] -> 3[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688201:688630 [1] NCCL INFO Channel 03/0 : 3[1] -> 4[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688201:688630 [1] NCCL INFO Channel 00/0 : 4[1] -> 5[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688201:688630 [1] NCCL INFO Channel 01/0 : 4[1] -> 5[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618585:619011 [1] NCCL INFO Channel 00/0 : 2[1] -> 3[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688201:688630 [1] NCCL INFO Channel 02/0 : 4[1] -> 5[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618585:619011 [1] NCCL INFO Channel 01/0 : 2[1] -> 3[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688201:688630 [1] NCCL INFO Channel 03/0 : 4[1] -> 5[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618585:619011 [1] NCCL INFO Channel 02/0 : 2[1] -> 3[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618585:619011 [1] NCCL INFO Channel 03/0 : 2[1] -> 3[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618585:619011 [1] NCCL INFO Channel 00/0 : 3[1] -> 4[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618585:619011 [1] NCCL INFO Channel 01/0 : 3[1] -> 4[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618585:619011 [1] NCCL INFO Channel 02/0 : 3[1] -> 4[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618585:619011 [1] NCCL INFO Channel 03/0 : 3[1] -> 4[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926595:927022 [1] NCCL INFO Channel 00/0 : 5[1] -> 6[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926595:927022 [1] NCCL INFO Channel 01/0 : 5[1] -> 6[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926595:927022 [1] NCCL INFO Channel 02/0 : 5[1] -> 6[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926595:927022 [1] NCCL INFO Channel 03/0 : 5[1] -> 6[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926595:927022 [1] NCCL INFO Channel 00/0 : 6[1] -> 7[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926595:927022 [1] NCCL INFO Channel 01/0 : 6[1] -> 7[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926595:927022 [1] NCCL INFO Channel 02/0 : 6[1] -> 7[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926595:927022 [1] NCCL INFO Channel 03/0 : 6[1] -> 7[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942612:943041 [1] NCCL INFO Channel 00/0 : 4[1] -> 5[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942612:943041 [1] NCCL INFO Channel 01/0 : 4[1] -> 5[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019289:1019717 [1] NCCL INFO Channel 00/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942612:943041 [1] NCCL INFO Channel 02/0 : 4[1] -> 5[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019289:1019717 [1] NCCL INFO Channel 01/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942612:943041 [1] NCCL INFO Channel 03/0 : 4[1] -> 5[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019289:1019717 [1] NCCL INFO Channel 02/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019289:1019717 [1] NCCL INFO Channel 03/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942612:943041 [1] NCCL INFO Channel 00/0 : 5[1] -> 6[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019289:1019717 [1] NCCL INFO Channel 00/0 : 1[1] -> 2[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019289:1019717 [1] NCCL INFO Channel 01/0 : 1[1] -> 2[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019289:1019717 [1] NCCL INFO Channel 02/0 : 1[1] -> 2[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019289:1019717 [1] NCCL INFO Channel 03/0 : 1[1] -> 2[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942612:943041 [1] NCCL INFO Channel 01/0 : 5[1] -> 6[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942612:943041 [1] NCCL INFO Channel 02/0 : 5[1] -> 6[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942612:943041 [1] NCCL INFO Channel 03/0 : 5[1] -> 6[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880684:881100 [5] NCCL INFO Connected all rings
+ip-26-0-172-142:880684:881100 [5] NCCL INFO Channel 00/0 : 2[5] -> 4[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880684:881100 [5] NCCL INFO Channel 01/0 : 2[5] -> 4[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926599:927016 [5] NCCL INFO Connected all rings
+ip-26-0-172-73:926599:927016 [5] NCCL INFO Channel 00/0 : 4[5] -> 6[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926599:927016 [5] NCCL INFO Channel 01/0 : 4[5] -> 6[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926598:927015 [4] NCCL INFO Connected all rings
+ip-26-0-172-73:926598:927015 [4] NCCL INFO Channel 00/0 : 4[4] -> 6[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926598:927015 [4] NCCL INFO Channel 01/0 : 4[4] -> 6[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-7:811252:811692 [4] NCCL INFO Connected all rings
+ip-26-0-173-7:811252:811692 [4] NCCL INFO Channel 02/0 : 3[4] -> 7[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-7:811253:811693 [5] NCCL INFO Connected all rings
+ip-26-0-173-7:811252:811692 [4] NCCL INFO Channel 03/0 : 3[4] -> 7[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-7:811252:811692 [4] NCCL INFO Channel 02/0 : 7[4] -> 3[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-7:811253:811693 [5] NCCL INFO Channel 02/0 : 3[5] -> 7[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-7:811252:811692 [4] NCCL INFO Channel 03/0 : 7[4] -> 3[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-7:811253:811693 [5] NCCL INFO Channel 03/0 : 3[5] -> 7[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-7:811253:811693 [5] NCCL INFO Channel 02/0 : 7[5] -> 3[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-7:811253:811693 [5] NCCL INFO Channel 03/0 : 7[5] -> 3[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942615:943026 [4] NCCL INFO Connected all rings
+ip-26-0-172-57:942615:943026 [4] NCCL INFO Channel 02/0 : 3[4] -> 5[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942615:943026 [4] NCCL INFO Channel 03/0 : 3[4] -> 5[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880683:881099 [4] NCCL INFO Connected all rings
+ip-26-0-172-57:942616:943029 [5] NCCL INFO Connected all rings
+ip-26-0-172-142:880683:881099 [4] NCCL INFO Channel 00/0 : 2[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942616:943029 [5] NCCL INFO Channel 02/0 : 3[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880683:881099 [4] NCCL INFO Channel 01/0 : 2[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942616:943029 [5] NCCL INFO Channel 03/0 : 3[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019293:1019711 [5] NCCL INFO Connected all rings
+ip-26-0-172-116:1019293:1019711 [5] NCCL INFO Channel 02/0 : 1[5] -> 3[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019293:1019711 [5] NCCL INFO Channel 03/0 : 1[5] -> 3[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688205:688624 [5] NCCL INFO Connected all rings
+ip-26-0-172-252:688205:688624 [5] NCCL INFO Channel 00/0 : 2[5] -> 4[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688205:688624 [5] NCCL INFO Channel 01/0 : 2[5] -> 4[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688205:688624 [5] NCCL INFO Channel 00/0 : 4[5] -> 6[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688205:688624 [5] NCCL INFO Channel 01/0 : 4[5] -> 6[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880684:881100 [5] NCCL INFO Channel 00/0 : 4[5] -> 2[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880684:881100 [5] NCCL INFO Channel 01/0 : 4[5] -> 2[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688205:688624 [5] NCCL INFO Channel 00/0 : 0[5] -> 4[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926599:927016 [5] NCCL INFO Channel 00/0 : 6[5] -> 4[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688205:688624 [5] NCCL INFO Channel 01/0 : 0[5] -> 4[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926599:927016 [5] NCCL INFO Channel 01/0 : 6[5] -> 4[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688205:688624 [5] NCCL INFO Channel 00/0 : 4[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688205:688624 [5] NCCL INFO Channel 01/0 : 4[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:94347:94800 [5] NCCL INFO Connected all rings
+ip-26-0-160-225:94347:94800 [5] NCCL INFO Channel 00/0 : 4[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:94347:94800 [5] NCCL INFO Channel 01/0 : 4[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:94347:94800 [5] NCCL INFO Channel 00/0 : 0[5] -> 4[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:94347:94800 [5] NCCL INFO Channel 01/0 : 0[5] -> 4[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:94347:94800 [5] NCCL INFO Channel 02/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688205:688624 [5] NCCL INFO Channel 00/0 : 6[5] -> 4[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:94347:94800 [5] NCCL INFO Channel 03/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688205:688624 [5] NCCL INFO Channel 01/0 : 6[5] -> 4[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688205:688624 [5] NCCL INFO Channel 00/0 : 4[5] -> 2[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688205:688624 [5] NCCL INFO Channel 01/0 : 4[5] -> 2[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926599:927016 [5] NCCL INFO Channel 00/0 : 7[5] -> 6[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926599:927016 [5] NCCL INFO Channel 01/0 : 7[5] -> 6[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688205:688624 [5] NCCL INFO Channel 02/0 : 5[5] -> 4[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880684:881100 [5] NCCL INFO Channel 00/0 : 3[5] -> 2[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688205:688624 [5] NCCL INFO Channel 03/0 : 5[5] -> 4[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926599:927016 [5] NCCL INFO Channel 00/0 : 6[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:94346:94798 [4] NCCL INFO Connected all rings
+ip-26-0-172-142:880684:881100 [5] NCCL INFO Channel 01/0 : 3[5] -> 2[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926599:927016 [5] NCCL INFO Channel 01/0 : 6[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880684:881100 [5] NCCL INFO Channel 00/0 : 2[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926599:927016 [5] NCCL INFO Channel 02/0 : 6[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:94346:94798 [4] NCCL INFO Channel 00/0 : 4[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880684:881100 [5] NCCL INFO Channel 01/0 : 2[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926599:927016 [5] NCCL INFO Channel 03/0 : 6[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:94346:94798 [4] NCCL INFO Channel 01/0 : 4[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880684:881100 [5] NCCL INFO Channel 02/0 : 2[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:94346:94798 [4] NCCL INFO Channel 00/0 : 0[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880684:881100 [5] NCCL INFO Channel 03/0 : 2[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:94346:94798 [4] NCCL INFO Channel 01/0 : 0[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618589:619005 [5] NCCL INFO Connected all rings
+ip-26-0-172-147:618588:619004 [4] NCCL INFO Connected all rings
+ip-26-0-172-147:618589:619005 [5] NCCL INFO Channel 02/0 : 1[5] -> 3[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618589:619005 [5] NCCL INFO Channel 03/0 : 1[5] -> 3[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618588:619004 [4] NCCL INFO Channel 02/0 : 1[4] -> 3[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618589:619005 [5] NCCL INFO Channel 02/0 : 3[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618588:619004 [4] NCCL INFO Channel 03/0 : 1[4] -> 3[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618589:619005 [5] NCCL INFO Channel 03/0 : 3[5] -> 5[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618588:619004 [4] NCCL INFO Channel 02/0 : 3[4] -> 5[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618588:619004 [4] NCCL INFO Channel 03/0 : 3[4] -> 5[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019293:1019711 [5] NCCL INFO Channel 02/0 : 3[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019293:1019711 [5] NCCL INFO Channel 03/0 : 3[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618589:619005 [5] NCCL INFO Channel 02/0 : 7[5] -> 3[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942616:943029 [5] NCCL INFO Channel 02/0 : 5[5] -> 3[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618589:619005 [5] NCCL INFO Channel 03/0 : 7[5] -> 3[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942616:943029 [5] NCCL INFO Channel 03/0 : 5[5] -> 3[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618589:619005 [5] NCCL INFO Channel 02/0 : 3[5] -> 7[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942615:943026 [4] NCCL INFO Channel 02/0 : 5[4] -> 3[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942615:943026 [4] NCCL INFO Channel 03/0 : 5[4] -> 3[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618589:619005 [5] NCCL INFO Channel 03/0 : 3[5] -> 7[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618589:619005 [5] NCCL INFO Channel 02/0 : 5[5] -> 3[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618589:619005 [5] NCCL INFO Channel 03/0 : 5[5] -> 3[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618589:619005 [5] NCCL INFO Channel 02/0 : 3[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618589:619005 [5] NCCL INFO Channel 03/0 : 3[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942616:943029 [5] NCCL INFO Channel 00/0 : 6[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942616:943029 [5] NCCL INFO Channel 01/0 : 6[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618589:619005 [5] NCCL INFO Channel 00/0 : 3[5] -> 2[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019293:1019711 [5] NCCL INFO Channel 00/0 : 2[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942616:943029 [5] NCCL INFO Channel 02/0 : 6[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618589:619005 [5] NCCL INFO Channel 01/0 : 3[5] -> 2[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942616:943029 [5] NCCL INFO Channel 03/0 : 6[5] -> 5[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019293:1019711 [5] NCCL INFO Channel 01/0 : 2[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942616:943029 [5] NCCL INFO Channel 02/0 : 5[5] -> 4[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019293:1019711 [5] NCCL INFO Channel 02/0 : 2[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942616:943029 [5] NCCL INFO Channel 03/0 : 5[5] -> 4[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019293:1019711 [5] NCCL INFO Channel 03/0 : 2[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019293:1019711 [5] NCCL INFO Channel 02/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019293:1019711 [5] NCCL INFO Channel 03/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-7:811253:811693 [5] NCCL INFO Channel 00/0 : 7[5] -> 6[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-7:811253:811693 [5] NCCL INFO Channel 01/0 : 7[5] -> 6[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688204:688623 [4] NCCL INFO Connected all rings
+ip-26-0-172-252:688204:688623 [4] NCCL INFO Channel 00/0 : 2[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688204:688623 [4] NCCL INFO Channel 01/0 : 2[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688204:688623 [4] NCCL INFO Channel 00/0 : 4[4] -> 6[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688204:688623 [4] NCCL INFO Channel 01/0 : 4[4] -> 6[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880683:881099 [4] NCCL INFO Channel 00/0 : 4[4] -> 2[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880683:881099 [4] NCCL INFO Channel 01/0 : 4[4] -> 2[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688204:688623 [4] NCCL INFO Channel 00/0 : 0[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926598:927015 [4] NCCL INFO Channel 00/0 : 6[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688204:688623 [4] NCCL INFO Channel 01/0 : 0[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926598:927015 [4] NCCL INFO Channel 01/0 : 6[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688204:688623 [4] NCCL INFO Channel 00/0 : 4[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688204:688623 [4] NCCL INFO Channel 01/0 : 4[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688204:688623 [4] NCCL INFO Channel 00/0 : 6[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:94346:94798 [4] NCCL INFO Channel 02/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688204:688623 [4] NCCL INFO Channel 01/0 : 6[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:94346:94798 [4] NCCL INFO Channel 03/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688204:688623 [4] NCCL INFO Channel 00/0 : 4[4] -> 2[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688204:688623 [4] NCCL INFO Channel 01/0 : 4[4] -> 2[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926598:927015 [4] NCCL INFO Channel 00/0 : 7[4] -> 6[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926598:927015 [4] NCCL INFO Channel 01/0 : 7[4] -> 6[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926598:927015 [4] NCCL INFO Channel 00/0 : 6[4] -> 5[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926598:927015 [4] NCCL INFO Channel 01/0 : 6[4] -> 5[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880683:881099 [4] NCCL INFO Channel 00/0 : 3[4] -> 2[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688204:688623 [4] NCCL INFO Channel 02/0 : 5[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926598:927015 [4] NCCL INFO Channel 02/0 : 6[4] -> 5[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880683:881099 [4] NCCL INFO Channel 01/0 : 3[4] -> 2[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688204:688623 [4] NCCL INFO Channel 03/0 : 5[4] -> 4[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926598:927015 [4] NCCL INFO Channel 03/0 : 6[4] -> 5[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880683:881099 [4] NCCL INFO Channel 00/0 : 2[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880683:881099 [4] NCCL INFO Channel 01/0 : 2[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880683:881099 [4] NCCL INFO Channel 02/0 : 2[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880683:881099 [4] NCCL INFO Channel 03/0 : 2[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-7:811255:811695 [7] NCCL INFO Connected all rings
+ip-26-0-173-7:811255:811695 [7] NCCL INFO Channel 02/0 : 3[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-7:811255:811695 [7] NCCL INFO Channel 03/0 : 3[7] -> 7[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-7:811255:811695 [7] NCCL INFO Channel 02/0 : 7[7] -> 3[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-7:811255:811695 [7] NCCL INFO Channel 03/0 : 7[7] -> 3[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-7:811248:811694 [0] NCCL INFO Connected all rings
+ip-26-0-173-7:811248:811694 [0] NCCL INFO Channel 02/0 : 3[0] -> 7[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-7:811248:811694 [0] NCCL INFO Channel 03/0 : 3[0] -> 7[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-7:811248:811694 [0] NCCL INFO Channel 02/0 : 7[0] -> 3[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926601:927018 [7] NCCL INFO Connected all rings
+ip-26-0-173-7:811248:811694 [0] NCCL INFO Channel 03/0 : 7[0] -> 3[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926601:927018 [7] NCCL INFO Channel 00/0 : 4[7] -> 6[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926601:927018 [7] NCCL INFO Channel 01/0 : 4[7] -> 6[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942618:943035 [7] NCCL INFO Connected all rings
+ip-26-0-172-57:942618:943035 [7] NCCL INFO Channel 02/0 : 3[7] -> 5[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942618:943035 [7] NCCL INFO Channel 03/0 : 3[7] -> 5[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942611:943034 [0] NCCL INFO Connected all rings
+ip-26-0-172-57:942611:943034 [0] NCCL INFO Channel 02/0 : 3[0] -> 5[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942611:943034 [0] NCCL INFO Channel 03/0 : 3[0] -> 5[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019292:1019710 [4] NCCL INFO Connected all rings
+ip-26-0-172-116:1019292:1019710 [4] NCCL INFO Channel 02/0 : 1[4] -> 3[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019292:1019710 [4] NCCL INFO Channel 03/0 : 1[4] -> 3[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019292:1019710 [4] NCCL INFO Channel 02/0 : 3[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019292:1019710 [4] NCCL INFO Channel 03/0 : 3[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618588:619004 [4] NCCL INFO Channel 02/0 : 7[4] -> 3[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618588:619004 [4] NCCL INFO Channel 03/0 : 7[4] -> 3[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618588:619004 [4] NCCL INFO Channel 02/0 : 3[4] -> 7[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618588:619004 [4] NCCL INFO Channel 03/0 : 3[4] -> 7[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:94342:94802 [0] NCCL INFO Connected all rings
+ip-26-0-160-225:94342:94802 [0] NCCL INFO Channel 00/0 : 4[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618588:619004 [4] NCCL INFO Channel 02/0 : 5[4] -> 3[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:94342:94802 [0] NCCL INFO Channel 01/0 : 4[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618588:619004 [4] NCCL INFO Channel 03/0 : 5[4] -> 3[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-7:811252:811692 [4] NCCL INFO Channel 00/0 : 7[4] -> 6[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:94342:94802 [0] NCCL INFO Channel 00/0 : 0[0] -> 4[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-7:811252:811692 [4] NCCL INFO Channel 01/0 : 7[4] -> 6[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618588:619004 [4] NCCL INFO Channel 02/0 : 3[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:94342:94802 [0] NCCL INFO Channel 01/0 : 0[0] -> 4[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618588:619004 [4] NCCL INFO Channel 03/0 : 3[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942615:943026 [4] NCCL INFO Channel 00/0 : 6[4] -> 5[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618588:619004 [4] NCCL INFO Channel 00/0 : 3[4] -> 2[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942615:943026 [4] NCCL INFO Channel 01/0 : 6[4] -> 5[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019292:1019710 [4] NCCL INFO Channel 00/0 : 2[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618588:619004 [4] NCCL INFO Channel 01/0 : 3[4] -> 2[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942615:943026 [4] NCCL INFO Channel 02/0 : 6[4] -> 5[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019292:1019710 [4] NCCL INFO Channel 01/0 : 2[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942615:943026 [4] NCCL INFO Channel 03/0 : 6[4] -> 5[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019292:1019710 [4] NCCL INFO Channel 02/0 : 2[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942615:943026 [4] NCCL INFO Channel 02/0 : 5[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019292:1019710 [4] NCCL INFO Channel 03/0 : 2[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942615:943026 [4] NCCL INFO Channel 03/0 : 5[4] -> 4[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019292:1019710 [4] NCCL INFO Channel 02/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019292:1019710 [4] NCCL INFO Channel 03/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926594:927017 [0] NCCL INFO Connected all rings
+ip-26-0-160-225:94348:94808 [6] NCCL INFO Connected all rings
+ip-26-0-172-73:926594:927017 [0] NCCL INFO Channel 00/0 : 4[0] -> 6[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926594:927017 [0] NCCL INFO Channel 01/0 : 4[0] -> 6[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:94348:94808 [6] NCCL INFO Channel 00/0 : 4[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:94348:94808 [6] NCCL INFO Channel 01/0 : 4[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:94348:94808 [6] NCCL INFO Channel 00/0 : 0[6] -> 4[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-7:811254:811697 [6] NCCL INFO Connected all rings
+ip-26-0-160-225:94348:94808 [6] NCCL INFO Channel 01/0 : 0[6] -> 4[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-7:811254:811697 [6] NCCL INFO Channel 02/0 : 3[6] -> 7[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-7:811254:811697 [6] NCCL INFO Channel 03/0 : 3[6] -> 7[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-7:811254:811697 [6] NCCL INFO Channel 02/0 : 7[6] -> 3[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-7:811254:811697 [6] NCCL INFO Channel 03/0 : 7[6] -> 3[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688207:688626 [7] NCCL INFO Connected all rings
+ip-26-0-172-252:688207:688626 [7] NCCL INFO Channel 00/0 : 2[7] -> 4[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688207:688626 [7] NCCL INFO Channel 01/0 : 2[7] -> 4[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688207:688626 [7] NCCL INFO Channel 00/0 : 4[7] -> 6[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688207:688626 [7] NCCL INFO Channel 01/0 : 4[7] -> 6[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926601:927018 [7] NCCL INFO Channel 00/0 : 6[7] -> 4[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926601:927018 [7] NCCL INFO Channel 01/0 : 6[7] -> 4[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926600:927020 [6] NCCL INFO Connected all rings
+ip-26-0-172-73:926600:927020 [6] NCCL INFO Channel 00/0 : 4[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926600:927020 [6] NCCL INFO Channel 01/0 : 4[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688206:688628 [6] NCCL INFO Connected all rings
+ip-26-0-172-252:688206:688628 [6] NCCL INFO Channel 00/0 : 2[6] -> 4[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688206:688628 [6] NCCL INFO Channel 01/0 : 2[6] -> 4[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688206:688628 [6] NCCL INFO Channel 00/0 : 4[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688206:688628 [6] NCCL INFO Channel 01/0 : 4[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926600:927020 [6] NCCL INFO Channel 00/0 : 6[6] -> 4[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926600:927020 [6] NCCL INFO Channel 01/0 : 6[6] -> 4[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942617:943039 [6] NCCL INFO Connected all rings
+ip-26-0-172-57:942617:943039 [6] NCCL INFO Channel 02/0 : 3[6] -> 5[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942617:943039 [6] NCCL INFO Channel 03/0 : 3[6] -> 5[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688200:688625 [0] NCCL INFO Connected all rings
+ip-26-0-172-252:688200:688625 [0] NCCL INFO Channel 00/0 : 2[0] -> 4[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688200:688625 [0] NCCL INFO Channel 01/0 : 2[0] -> 4[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688200:688625 [0] NCCL INFO Channel 00/0 : 4[0] -> 6[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688200:688625 [0] NCCL INFO Channel 01/0 : 4[0] -> 6[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926594:927017 [0] NCCL INFO Channel 00/0 : 6[0] -> 4[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926594:927017 [0] NCCL INFO Channel 01/0 : 6[0] -> 4[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618584:619006 [0] NCCL INFO Connected all rings
+ip-26-0-172-147:618584:619006 [0] NCCL INFO Channel 02/0 : 1[0] -> 3[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618584:619006 [0] NCCL INFO Channel 03/0 : 1[0] -> 3[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618584:619006 [0] NCCL INFO Channel 02/0 : 3[0] -> 5[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618584:619006 [0] NCCL INFO Channel 03/0 : 3[0] -> 5[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942611:943034 [0] NCCL INFO Channel 02/0 : 5[0] -> 3[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942611:943034 [0] NCCL INFO Channel 03/0 : 5[0] -> 3[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880686:881102 [7] NCCL INFO Connected all rings
+ip-26-0-172-142:880686:881102 [7] NCCL INFO Channel 00/0 : 2[7] -> 4[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880686:881102 [7] NCCL INFO Channel 01/0 : 2[7] -> 4[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880686:881102 [7] NCCL INFO Channel 00/0 : 4[7] -> 2[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688207:688626 [7] NCCL INFO Channel 00/0 : 0[7] -> 4[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-7:811250:811696 [2] NCCL INFO Connected all rings
+ip-26-0-172-142:880686:881102 [7] NCCL INFO Channel 01/0 : 4[7] -> 2[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688207:688626 [7] NCCL INFO Channel 01/0 : 0[7] -> 4[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-7:811250:811696 [2] NCCL INFO Channel 02/0 : 3[2] -> 7[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688207:688626 [7] NCCL INFO Channel 00/0 : 4[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688207:688626 [7] NCCL INFO Channel 01/0 : 4[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-7:811250:811696 [2] NCCL INFO Channel 03/0 : 3[2] -> 7[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-7:811250:811696 [2] NCCL INFO Channel 02/0 : 7[2] -> 3[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-7:811250:811696 [2] NCCL INFO Channel 03/0 : 7[2] -> 3[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-7:811251:811698 [3] NCCL INFO Connected all rings
+ip-26-0-173-7:811251:811698 [3] NCCL INFO Channel 02/0 : 3[3] -> 7[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-7:811251:811698 [3] NCCL INFO Channel 03/0 : 3[3] -> 7[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-7:811251:811698 [3] NCCL INFO Channel 02/0 : 7[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-7:811251:811698 [3] NCCL INFO Channel 03/0 : 7[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:94344:94806 [2] NCCL INFO Connected all rings
+ip-26-0-160-225:94344:94806 [2] NCCL INFO Channel 00/0 : 4[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:94344:94806 [2] NCCL INFO Channel 01/0 : 4[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:94344:94806 [2] NCCL INFO Channel 00/0 : 0[2] -> 4[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:94344:94806 [2] NCCL INFO Channel 01/0 : 0[2] -> 4[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-7:811249:811699 [1] NCCL INFO Connected all rings
+ip-26-0-173-7:811249:811699 [1] NCCL INFO Channel 02/0 : 3[1] -> 7[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-7:811249:811699 [1] NCCL INFO Channel 03/0 : 3[1] -> 7[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-7:811249:811699 [1] NCCL INFO Channel 02/0 : 7[1] -> 3[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-7:811249:811699 [1] NCCL INFO Channel 03/0 : 7[1] -> 3[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926596:927019 [2] NCCL INFO Connected all rings
+ip-26-0-172-73:926596:927019 [2] NCCL INFO Channel 00/0 : 4[2] -> 6[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618590:619009 [6] NCCL INFO Connected all rings
+ip-26-0-172-73:926596:927019 [2] NCCL INFO Channel 01/0 : 4[2] -> 6[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618590:619009 [6] NCCL INFO Channel 02/0 : 1[6] -> 3[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618590:619009 [6] NCCL INFO Channel 03/0 : 1[6] -> 3[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618590:619009 [6] NCCL INFO Channel 02/0 : 3[6] -> 5[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618590:619009 [6] NCCL INFO Channel 03/0 : 3[6] -> 5[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880679:881101 [0] NCCL INFO Connected all rings
+ip-26-0-172-142:880679:881101 [0] NCCL INFO Channel 00/0 : 2[0] -> 4[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880679:881101 [0] NCCL INFO Channel 01/0 : 2[0] -> 4[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942617:943039 [6] NCCL INFO Channel 02/0 : 5[6] -> 3[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942617:943039 [6] NCCL INFO Channel 03/0 : 5[6] -> 3[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880679:881101 [0] NCCL INFO Channel 00/0 : 4[0] -> 2[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688200:688625 [0] NCCL INFO Channel 00/0 : 0[0] -> 4[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880679:881101 [0] NCCL INFO Channel 01/0 : 4[0] -> 2[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688200:688625 [0] NCCL INFO Channel 01/0 : 0[0] -> 4[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688200:688625 [0] NCCL INFO Channel 00/0 : 4[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688200:688625 [0] NCCL INFO Channel 01/0 : 4[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688200:688625 [0] NCCL INFO Channel 00/0 : 6[0] -> 4[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:94342:94802 [0] NCCL INFO Channel 02/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688200:688625 [0] NCCL INFO Channel 01/0 : 6[0] -> 4[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:94342:94802 [0] NCCL INFO Channel 03/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688200:688625 [0] NCCL INFO Channel 00/0 : 4[0] -> 2[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688200:688625 [0] NCCL INFO Channel 01/0 : 4[0] -> 2[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926594:927017 [0] NCCL INFO Channel 00/0 : 7[0] -> 6[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926594:927017 [0] NCCL INFO Channel 01/0 : 7[0] -> 6[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942613:943038 [2] NCCL INFO Connected all rings
+ip-26-0-172-252:688200:688625 [0] NCCL INFO Channel 02/0 : 5[0] -> 4[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880679:881101 [0] NCCL INFO Channel 00/0 : 3[0] -> 2[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926594:927017 [0] NCCL INFO Channel 00/0 : 6[0] -> 5[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688200:688625 [0] NCCL INFO Channel 03/0 : 5[0] -> 4[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942613:943038 [2] NCCL INFO Channel 02/0 : 3[2] -> 5[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880679:881101 [0] NCCL INFO Channel 01/0 : 3[0] -> 2[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926594:927017 [0] NCCL INFO Channel 01/0 : 6[0] -> 5[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942613:943038 [2] NCCL INFO Channel 03/0 : 3[2] -> 5[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880679:881101 [0] NCCL INFO Channel 00/0 : 2[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926594:927017 [0] NCCL INFO Channel 02/0 : 6[0] -> 5[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880679:881101 [0] NCCL INFO Channel 01/0 : 2[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926594:927017 [0] NCCL INFO Channel 03/0 : 6[0] -> 5[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880679:881101 [0] NCCL INFO Channel 02/0 : 2[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880679:881101 [0] NCCL INFO Channel 03/0 : 2[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926595:927022 [1] NCCL INFO Connected all rings
+ip-26-0-172-73:926597:927021 [3] NCCL INFO Connected all rings
+ip-26-0-172-73:926595:927022 [1] NCCL INFO Channel 00/0 : 4[1] -> 6[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926597:927021 [3] NCCL INFO Channel 00/0 : 4[3] -> 6[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926595:927022 [1] NCCL INFO Channel 01/0 : 4[1] -> 6[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926597:927021 [3] NCCL INFO Channel 01/0 : 4[3] -> 6[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880685:881104 [6] NCCL INFO Connected all rings
+ip-26-0-172-142:880685:881104 [6] NCCL INFO Channel 00/0 : 2[6] -> 4[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880685:881104 [6] NCCL INFO Channel 01/0 : 2[6] -> 4[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880685:881104 [6] NCCL INFO Channel 00/0 : 4[6] -> 2[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688206:688628 [6] NCCL INFO Channel 00/0 : 0[6] -> 4[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880685:881104 [6] NCCL INFO Channel 01/0 : 4[6] -> 2[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688206:688628 [6] NCCL INFO Channel 01/0 : 0[6] -> 4[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688206:688628 [6] NCCL INFO Channel 00/0 : 4[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688206:688628 [6] NCCL INFO Channel 01/0 : 4[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688206:688628 [6] NCCL INFO Channel 00/0 : 6[6] -> 4[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:94348:94808 [6] NCCL INFO Channel 02/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688206:688628 [6] NCCL INFO Channel 01/0 : 6[6] -> 4[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:94348:94808 [6] NCCL INFO Channel 03/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688206:688628 [6] NCCL INFO Channel 00/0 : 4[6] -> 2[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688206:688628 [6] NCCL INFO Channel 01/0 : 4[6] -> 2[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926600:927020 [6] NCCL INFO Channel 00/0 : 7[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688206:688628 [6] NCCL INFO Channel 02/0 : 5[6] -> 4[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880685:881104 [6] NCCL INFO Channel 00/0 : 3[6] -> 2[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926600:927020 [6] NCCL INFO Channel 01/0 : 7[6] -> 6[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688206:688628 [6] NCCL INFO Channel 03/0 : 5[6] -> 4[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880685:881104 [6] NCCL INFO Channel 01/0 : 3[6] -> 2[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926600:927020 [6] NCCL INFO Channel 00/0 : 6[6] -> 5[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880685:881104 [6] NCCL INFO Channel 00/0 : 2[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926600:927020 [6] NCCL INFO Channel 01/0 : 6[6] -> 5[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880685:881104 [6] NCCL INFO Channel 01/0 : 2[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926600:927020 [6] NCCL INFO Channel 02/0 : 6[6] -> 5[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880685:881104 [6] NCCL INFO Channel 02/0 : 2[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926600:927020 [6] NCCL INFO Channel 03/0 : 6[6] -> 5[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880685:881104 [6] NCCL INFO Channel 03/0 : 2[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942612:943041 [1] NCCL INFO Connected all rings
+ip-26-0-172-57:942612:943041 [1] NCCL INFO Channel 02/0 : 3[1] -> 5[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942612:943041 [1] NCCL INFO Channel 03/0 : 3[1] -> 5[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618591:619007 [7] NCCL INFO Connected all rings
+ip-26-0-172-147:618591:619007 [7] NCCL INFO Channel 02/0 : 1[7] -> 3[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618591:619007 [7] NCCL INFO Channel 03/0 : 1[7] -> 3[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618591:619007 [7] NCCL INFO Channel 02/0 : 3[7] -> 5[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618591:619007 [7] NCCL INFO Channel 03/0 : 3[7] -> 5[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942618:943035 [7] NCCL INFO Channel 02/0 : 5[7] -> 3[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942618:943035 [7] NCCL INFO Channel 03/0 : 5[7] -> 3[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618586:619008 [2] NCCL INFO Connected all rings
+ip-26-0-172-147:618586:619008 [2] NCCL INFO Channel 02/0 : 1[2] -> 3[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618586:619008 [2] NCCL INFO Channel 03/0 : 1[2] -> 3[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688202:688627 [2] NCCL INFO Connected all rings
+ip-26-0-172-147:618586:619008 [2] NCCL INFO Channel 02/0 : 3[2] -> 5[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618586:619008 [2] NCCL INFO Channel 03/0 : 3[2] -> 5[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688202:688627 [2] NCCL INFO Channel 00/0 : 2[2] -> 4[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688202:688627 [2] NCCL INFO Channel 01/0 : 2[2] -> 4[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688202:688627 [2] NCCL INFO Channel 00/0 : 4[2] -> 6[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688202:688627 [2] NCCL INFO Channel 01/0 : 4[2] -> 6[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942613:943038 [2] NCCL INFO Channel 02/0 : 5[2] -> 3[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942613:943038 [2] NCCL INFO Channel 03/0 : 5[2] -> 3[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926596:927019 [2] NCCL INFO Channel 00/0 : 6[2] -> 4[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926596:927019 [2] NCCL INFO Channel 01/0 : 6[2] -> 4[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942614:943040 [3] NCCL INFO Connected all rings
+ip-26-0-172-57:942614:943040 [3] NCCL INFO Channel 02/0 : 3[3] -> 5[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942614:943040 [3] NCCL INFO Channel 03/0 : 3[3] -> 5[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688201:688630 [1] NCCL INFO Connected all rings
+ip-26-0-172-252:688201:688630 [1] NCCL INFO Channel 00/0 : 2[1] -> 4[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688201:688630 [1] NCCL INFO Channel 01/0 : 2[1] -> 4[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688203:688629 [3] NCCL INFO Connected all rings
+ip-26-0-172-252:688201:688630 [1] NCCL INFO Channel 00/0 : 4[1] -> 6[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688201:688630 [1] NCCL INFO Channel 01/0 : 4[1] -> 6[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688203:688629 [3] NCCL INFO Channel 00/0 : 2[3] -> 4[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688203:688629 [3] NCCL INFO Channel 01/0 : 2[3] -> 4[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618587:619010 [3] NCCL INFO Connected all rings
+ip-26-0-172-252:688203:688629 [3] NCCL INFO Channel 00/0 : 4[3] -> 6[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618587:619010 [3] NCCL INFO Channel 02/0 : 1[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688203:688629 [3] NCCL INFO Channel 01/0 : 4[3] -> 6[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618587:619010 [3] NCCL INFO Channel 03/0 : 1[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618587:619010 [3] NCCL INFO Channel 02/0 : 3[3] -> 5[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926595:927022 [1] NCCL INFO Channel 00/0 : 6[1] -> 4[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618587:619010 [3] NCCL INFO Channel 03/0 : 3[3] -> 5[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926595:927022 [1] NCCL INFO Channel 01/0 : 6[1] -> 4[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926597:927021 [3] NCCL INFO Channel 00/0 : 6[3] -> 4[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926597:927021 [3] NCCL INFO Channel 01/0 : 6[3] -> 4[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942614:943040 [3] NCCL INFO Channel 02/0 : 5[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942614:943040 [3] NCCL INFO Channel 03/0 : 5[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:94349:94804 [7] NCCL INFO Connected all rings
+ip-26-0-160-225:94349:94804 [7] NCCL INFO Channel 00/0 : 4[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:94349:94804 [7] NCCL INFO Channel 01/0 : 4[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:94349:94804 [7] NCCL INFO Channel 00/0 : 0[7] -> 4[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:94349:94804 [7] NCCL INFO Channel 01/0 : 0[7] -> 4[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:94349:94804 [7] NCCL INFO Channel 02/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:94349:94804 [7] NCCL INFO Channel 03/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688207:688626 [7] NCCL INFO Channel 00/0 : 6[7] -> 4[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688207:688626 [7] NCCL INFO Channel 01/0 : 6[7] -> 4[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688207:688626 [7] NCCL INFO Channel 00/0 : 4[7] -> 2[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688207:688626 [7] NCCL INFO Channel 01/0 : 4[7] -> 2[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688207:688626 [7] NCCL INFO Channel 02/0 : 5[7] -> 4[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688207:688626 [7] NCCL INFO Channel 03/0 : 5[7] -> 4[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880686:881102 [7] NCCL INFO Channel 00/0 : 3[7] -> 2[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880686:881102 [7] NCCL INFO Channel 01/0 : 3[7] -> 2[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880686:881102 [7] NCCL INFO Channel 00/0 : 2[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880686:881102 [7] NCCL INFO Channel 01/0 : 2[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880686:881102 [7] NCCL INFO Channel 02/0 : 2[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880686:881102 [7] NCCL INFO Channel 03/0 : 2[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926601:927018 [7] NCCL INFO Channel 00/0 : 7[7] -> 6[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926601:927018 [7] NCCL INFO Channel 01/0 : 7[7] -> 6[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926601:927018 [7] NCCL INFO Channel 00/0 : 6[7] -> 5[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926601:927018 [7] NCCL INFO Channel 01/0 : 6[7] -> 5[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926601:927018 [7] NCCL INFO Channel 02/0 : 6[7] -> 5[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926601:927018 [7] NCCL INFO Channel 03/0 : 6[7] -> 5[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880681:881103 [2] NCCL INFO Connected all rings
+ip-26-0-172-147:618585:619011 [1] NCCL INFO Connected all rings
+ip-26-0-172-142:880681:881103 [2] NCCL INFO Channel 00/0 : 2[2] -> 4[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618585:619011 [1] NCCL INFO Channel 02/0 : 1[1] -> 3[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880681:881103 [2] NCCL INFO Channel 01/0 : 2[2] -> 4[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618585:619011 [1] NCCL INFO Channel 03/0 : 1[1] -> 3[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618585:619011 [1] NCCL INFO Channel 02/0 : 3[1] -> 5[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618585:619011 [1] NCCL INFO Channel 03/0 : 3[1] -> 5[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880681:881103 [2] NCCL INFO Channel 00/0 : 4[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880681:881103 [2] NCCL INFO Channel 01/0 : 4[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942612:943041 [1] NCCL INFO Channel 02/0 : 5[1] -> 3[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942612:943041 [1] NCCL INFO Channel 03/0 : 5[1] -> 3[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688202:688627 [2] NCCL INFO Channel 00/0 : 0[2] -> 4[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688202:688627 [2] NCCL INFO Channel 01/0 : 0[2] -> 4[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688202:688627 [2] NCCL INFO Channel 00/0 : 4[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688202:688627 [2] NCCL INFO Channel 01/0 : 4[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688202:688627 [2] NCCL INFO Channel 00/0 : 6[2] -> 4[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:94344:94806 [2] NCCL INFO Channel 02/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688202:688627 [2] NCCL INFO Channel 01/0 : 6[2] -> 4[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:94344:94806 [2] NCCL INFO Channel 03/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688202:688627 [2] NCCL INFO Channel 00/0 : 4[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688202:688627 [2] NCCL INFO Channel 01/0 : 4[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926596:927019 [2] NCCL INFO Channel 00/0 : 7[2] -> 6[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688202:688627 [2] NCCL INFO Channel 02/0 : 5[2] -> 4[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926596:927019 [2] NCCL INFO Channel 01/0 : 7[2] -> 6[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880681:881103 [2] NCCL INFO Channel 00/0 : 3[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926596:927019 [2] NCCL INFO Channel 00/0 : 6[2] -> 5[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688202:688627 [2] NCCL INFO Channel 03/0 : 5[2] -> 4[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880681:881103 [2] NCCL INFO Channel 01/0 : 3[2] -> 2[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926596:927019 [2] NCCL INFO Channel 01/0 : 6[2] -> 5[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880681:881103 [2] NCCL INFO Channel 00/0 : 2[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926596:927019 [2] NCCL INFO Channel 02/0 : 6[2] -> 5[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880681:881103 [2] NCCL INFO Channel 01/0 : 2[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926596:927019 [2] NCCL INFO Channel 03/0 : 6[2] -> 5[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880681:881103 [2] NCCL INFO Channel 02/0 : 2[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880681:881103 [2] NCCL INFO Channel 03/0 : 2[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880680:881106 [1] NCCL INFO Connected all rings
+ip-26-0-172-142:880680:881106 [1] NCCL INFO Channel 00/0 : 2[1] -> 4[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880680:881106 [1] NCCL INFO Channel 01/0 : 2[1] -> 4[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880680:881106 [1] NCCL INFO Channel 00/0 : 4[1] -> 2[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880680:881106 [1] NCCL INFO Channel 01/0 : 4[1] -> 2[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688201:688630 [1] NCCL INFO Channel 00/0 : 0[1] -> 4[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688201:688630 [1] NCCL INFO Channel 01/0 : 0[1] -> 4[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688201:688630 [1] NCCL INFO Channel 00/0 : 4[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688201:688630 [1] NCCL INFO Channel 01/0 : 4[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-7:811253:811693 [5] NCCL INFO Connected all trees
+ip-26-0-173-7:811253:811693 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-173-7:811253:811693 [5] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-173-7:811253:811693 [5] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-173-7:811253:811693 [5] NCCL INFO comm 0xa3027b0 rank 7 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0xa25ee5d67de91a57 - Init COMPLETE
+ip-26-0-173-7:811252:811692 [4] NCCL INFO Connected all trees
+ip-26-0-173-7:811252:811692 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-173-7:811252:811692 [4] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-173-7:811252:811692 [4] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-160-225:94343:94812 [1] NCCL INFO Connected all rings
+ip-26-0-160-225:94343:94812 [1] NCCL INFO Channel 00/0 : 4[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:94343:94812 [1] NCCL INFO Channel 01/0 : 4[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:94343:94812 [1] NCCL INFO Channel 00/0 : 0[1] -> 4[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:94343:94812 [1] NCCL INFO Channel 01/0 : 0[1] -> 4[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:94343:94812 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:94343:94812 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688201:688630 [1] NCCL INFO Channel 00/0 : 6[1] -> 4[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688201:688630 [1] NCCL INFO Channel 01/0 : 6[1] -> 4[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688201:688630 [1] NCCL INFO Channel 00/0 : 4[1] -> 2[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688201:688630 [1] NCCL INFO Channel 01/0 : 4[1] -> 2[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926595:927022 [1] NCCL INFO Channel 00/0 : 7[1] -> 6[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688201:688630 [1] NCCL INFO Channel 02/0 : 5[1] -> 4[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926595:927022 [1] NCCL INFO Channel 01/0 : 7[1] -> 6[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880680:881106 [1] NCCL INFO Channel 00/0 : 3[1] -> 2[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688201:688630 [1] NCCL INFO Channel 03/0 : 5[1] -> 4[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926595:927022 [1] NCCL INFO Channel 00/0 : 6[1] -> 5[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880680:881106 [1] NCCL INFO Channel 01/0 : 3[1] -> 2[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926595:927022 [1] NCCL INFO Channel 01/0 : 6[1] -> 5[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880680:881106 [1] NCCL INFO Channel 00/0 : 2[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926595:927022 [1] NCCL INFO Channel 02/0 : 6[1] -> 5[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880680:881106 [1] NCCL INFO Channel 01/0 : 2[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926595:927022 [1] NCCL INFO Channel 03/0 : 6[1] -> 5[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880680:881106 [1] NCCL INFO Channel 02/0 : 2[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880680:881106 [1] NCCL INFO Channel 03/0 : 2[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-7:811252:811692 [4] NCCL INFO comm 0x99af050 rank 7 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0x886b815c42972199 - Init COMPLETE
+ip-26-0-172-73:926599:927016 [5] NCCL INFO Connected all trees
+ip-26-0-172-73:926599:927016 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-73:926599:927016 [5] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-172-73:926599:927016 [5] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-73:926599:927016 [5] NCCL INFO comm 0xb5102b0 rank 6 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0xa25ee5d67de91a57 - Init COMPLETE
+ip-26-0-172-73:926598:927015 [4] NCCL INFO Connected all trees
+ip-26-0-172-73:926598:927015 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-73:926598:927015 [4] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-172-73:926598:927015 [4] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-147:618589:619005 [5] NCCL INFO Connected all trees
+ip-26-0-172-147:618589:619005 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-147:618589:619005 [5] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-172-147:618589:619005 [5] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-73:926598:927015 [4] NCCL INFO comm 0xae64190 rank 6 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0x886b815c42972199 - Init COMPLETE
+ip-26-0-172-57:942616:943029 [5] NCCL INFO Connected all trees
+ip-26-0-172-57:942616:943029 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-57:942616:943029 [5] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-172-57:942616:943029 [5] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-147:618589:619005 [5] NCCL INFO comm 0xad42db0 rank 3 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0xa25ee5d67de91a57 - Init COMPLETE
+ip-26-0-172-252:688205:688624 [5] NCCL INFO Connected all trees
+ip-26-0-172-252:688205:688624 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-252:688205:688624 [5] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-172-252:688205:688624 [5] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-57:942616:943029 [5] NCCL INFO comm 0x9bd20e0 rank 5 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0xa25ee5d67de91a57 - Init COMPLETE
+ip-26-0-172-252:688205:688624 [5] NCCL INFO comm 0xa383660 rank 4 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0xa25ee5d67de91a57 - Init COMPLETE
+ip-26-0-172-57:942615:943026 [4] NCCL INFO Connected all trees
+ip-26-0-172-57:942615:943026 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-57:942615:943026 [4] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-172-57:942615:943026 [4] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-252:688204:688623 [4] NCCL INFO Connected all trees
+ip-26-0-172-252:688204:688623 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-252:688204:688623 [4] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-172-252:688204:688623 [4] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-57:942615:943026 [4] NCCL INFO comm 0xb6503d0 rank 5 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0x886b815c42972199 - Init COMPLETE
+ip-26-0-172-252:688204:688623 [4] NCCL INFO comm 0xaa1b5a0 rank 4 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0x886b815c42972199 - Init COMPLETE
+ip-26-0-172-142:880682:881105 [3] NCCL INFO Connected all rings
+ip-26-0-172-142:880682:881105 [3] NCCL INFO Channel 00/0 : 2[3] -> 4[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880682:881105 [3] NCCL INFO Channel 01/0 : 2[3] -> 4[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880682:881105 [3] NCCL INFO Channel 00/0 : 4[3] -> 2[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880682:881105 [3] NCCL INFO Channel 01/0 : 4[3] -> 2[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688203:688629 [3] NCCL INFO Channel 00/0 : 0[3] -> 4[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688203:688629 [3] NCCL INFO Channel 01/0 : 0[3] -> 4[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688203:688629 [3] NCCL INFO Channel 00/0 : 4[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688203:688629 [3] NCCL INFO Channel 01/0 : 4[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:94345:94810 [3] NCCL INFO Connected all rings
+ip-26-0-160-225:94345:94810 [3] NCCL INFO Channel 00/0 : 4[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:94345:94810 [3] NCCL INFO Channel 01/0 : 4[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:94345:94810 [3] NCCL INFO Channel 00/0 : 0[3] -> 4[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:94345:94810 [3] NCCL INFO Channel 01/0 : 0[3] -> 4[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:94345:94810 [3] NCCL INFO Channel 02/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:94345:94810 [3] NCCL INFO Channel 03/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688203:688629 [3] NCCL INFO Channel 00/0 : 6[3] -> 4[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688203:688629 [3] NCCL INFO Channel 01/0 : 6[3] -> 4[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688203:688629 [3] NCCL INFO Channel 00/0 : 4[3] -> 2[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688203:688629 [3] NCCL INFO Channel 01/0 : 4[3] -> 2[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688203:688629 [3] NCCL INFO Channel 02/0 : 5[3] -> 4[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880682:881105 [3] NCCL INFO Channel 00/0 : 3[3] -> 2[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688203:688629 [3] NCCL INFO Channel 03/0 : 5[3] -> 4[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926597:927021 [3] NCCL INFO Channel 00/0 : 7[3] -> 6[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880682:881105 [3] NCCL INFO Channel 01/0 : 3[3] -> 2[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926597:927021 [3] NCCL INFO Channel 01/0 : 7[3] -> 6[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880682:881105 [3] NCCL INFO Channel 00/0 : 2[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926597:927021 [3] NCCL INFO Channel 00/0 : 6[3] -> 5[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880682:881105 [3] NCCL INFO Channel 01/0 : 2[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926597:927021 [3] NCCL INFO Channel 01/0 : 6[3] -> 5[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926597:927021 [3] NCCL INFO Channel 02/0 : 6[3] -> 5[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880682:881105 [3] NCCL INFO Channel 02/0 : 2[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926597:927021 [3] NCCL INFO Channel 03/0 : 6[3] -> 5[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880682:881105 [3] NCCL INFO Channel 03/0 : 2[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019295:1019713 [7] NCCL INFO Connected all rings
+ip-26-0-172-116:1019295:1019713 [7] NCCL INFO Channel 02/0 : 1[7] -> 3[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019295:1019713 [7] NCCL INFO Channel 03/0 : 1[7] -> 3[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019295:1019713 [7] NCCL INFO Channel 02/0 : 3[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019295:1019713 [7] NCCL INFO Channel 03/0 : 3[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618591:619007 [7] NCCL INFO Channel 02/0 : 7[7] -> 3[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618591:619007 [7] NCCL INFO Channel 03/0 : 7[7] -> 3[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618591:619007 [7] NCCL INFO Channel 02/0 : 3[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618591:619007 [7] NCCL INFO Channel 03/0 : 3[7] -> 7[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618591:619007 [7] NCCL INFO Channel 02/0 : 5[7] -> 3[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618591:619007 [7] NCCL INFO Channel 03/0 : 5[7] -> 3[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-7:811255:811695 [7] NCCL INFO Channel 00/0 : 7[7] -> 6[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618591:619007 [7] NCCL INFO Channel 02/0 : 3[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-7:811255:811695 [7] NCCL INFO Channel 01/0 : 7[7] -> 6[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618591:619007 [7] NCCL INFO Channel 03/0 : 3[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942618:943035 [7] NCCL INFO Channel 00/0 : 6[7] -> 5[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019295:1019713 [7] NCCL INFO Channel 00/0 : 2[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942618:943035 [7] NCCL INFO Channel 01/0 : 6[7] -> 5[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019295:1019713 [7] NCCL INFO Channel 01/0 : 2[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942618:943035 [7] NCCL INFO Channel 02/0 : 6[7] -> 5[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019295:1019713 [7] NCCL INFO Channel 02/0 : 2[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942618:943035 [7] NCCL INFO Channel 03/0 : 6[7] -> 5[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019295:1019713 [7] NCCL INFO Channel 03/0 : 2[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942618:943035 [7] NCCL INFO Channel 02/0 : 5[7] -> 4[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618591:619007 [7] NCCL INFO Channel 00/0 : 3[7] -> 2[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019295:1019713 [7] NCCL INFO Channel 02/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942618:943035 [7] NCCL INFO Channel 03/0 : 5[7] -> 4[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618591:619007 [7] NCCL INFO Channel 01/0 : 3[7] -> 2[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019295:1019713 [7] NCCL INFO Channel 03/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019294:1019715 [6] NCCL INFO Connected all rings
+ip-26-0-172-116:1019294:1019715 [6] NCCL INFO Channel 02/0 : 1[6] -> 3[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019294:1019715 [6] NCCL INFO Channel 03/0 : 1[6] -> 3[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019294:1019715 [6] NCCL INFO Channel 02/0 : 3[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019294:1019715 [6] NCCL INFO Channel 03/0 : 3[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618590:619009 [6] NCCL INFO Channel 02/0 : 7[6] -> 3[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618590:619009 [6] NCCL INFO Channel 03/0 : 7[6] -> 3[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618590:619009 [6] NCCL INFO Channel 02/0 : 3[6] -> 7[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618590:619009 [6] NCCL INFO Channel 03/0 : 3[6] -> 7[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618590:619009 [6] NCCL INFO Channel 02/0 : 5[6] -> 3[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618590:619009 [6] NCCL INFO Channel 03/0 : 5[6] -> 3[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-7:811254:811697 [6] NCCL INFO Channel 00/0 : 7[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618590:619009 [6] NCCL INFO Channel 02/0 : 3[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-7:811254:811697 [6] NCCL INFO Channel 01/0 : 7[6] -> 6[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618590:619009 [6] NCCL INFO Channel 03/0 : 3[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618590:619009 [6] NCCL INFO Channel 00/0 : 3[6] -> 2[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019294:1019715 [6] NCCL INFO Channel 00/0 : 2[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618590:619009 [6] NCCL INFO Channel 01/0 : 3[6] -> 2[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019294:1019715 [6] NCCL INFO Channel 01/0 : 2[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019294:1019715 [6] NCCL INFO Channel 02/0 : 2[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019294:1019715 [6] NCCL INFO Channel 03/0 : 2[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019294:1019715 [6] NCCL INFO Channel 02/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942617:943039 [6] NCCL INFO Channel 00/0 : 6[6] -> 5[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019294:1019715 [6] NCCL INFO Channel 03/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942617:943039 [6] NCCL INFO Channel 01/0 : 6[6] -> 5[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942617:943039 [6] NCCL INFO Channel 02/0 : 6[6] -> 5[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942617:943039 [6] NCCL INFO Channel 03/0 : 6[6] -> 5[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942617:943039 [6] NCCL INFO Channel 02/0 : 5[6] -> 4[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942617:943039 [6] NCCL INFO Channel 03/0 : 5[6] -> 4[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019288:1019712 [0] NCCL INFO Connected all rings
+ip-26-0-172-116:1019288:1019712 [0] NCCL INFO Channel 02/0 : 1[0] -> 3[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019288:1019712 [0] NCCL INFO Channel 03/0 : 1[0] -> 3[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019288:1019712 [0] NCCL INFO Channel 02/0 : 3[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019288:1019712 [0] NCCL INFO Channel 03/0 : 3[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618584:619006 [0] NCCL INFO Channel 02/0 : 7[0] -> 3[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618584:619006 [0] NCCL INFO Channel 03/0 : 7[0] -> 3[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618584:619006 [0] NCCL INFO Channel 02/0 : 3[0] -> 7[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618584:619006 [0] NCCL INFO Channel 03/0 : 3[0] -> 7[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618584:619006 [0] NCCL INFO Channel 02/0 : 5[0] -> 3[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618584:619006 [0] NCCL INFO Channel 03/0 : 5[0] -> 3[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618584:619006 [0] NCCL INFO Channel 02/0 : 3[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-7:811248:811694 [0] NCCL INFO Channel 00/0 : 7[0] -> 6[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618584:619006 [0] NCCL INFO Channel 03/0 : 3[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-7:811248:811694 [0] NCCL INFO Channel 01/0 : 7[0] -> 6[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618584:619006 [0] NCCL INFO Channel 00/0 : 3[0] -> 2[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019288:1019712 [0] NCCL INFO Channel 00/0 : 2[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942611:943034 [0] NCCL INFO Channel 00/0 : 6[0] -> 5[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618584:619006 [0] NCCL INFO Channel 01/0 : 3[0] -> 2[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019288:1019712 [0] NCCL INFO Channel 01/0 : 2[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942611:943034 [0] NCCL INFO Channel 01/0 : 6[0] -> 5[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942611:943034 [0] NCCL INFO Channel 02/0 : 6[0] -> 5[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019288:1019712 [0] NCCL INFO Channel 02/0 : 2[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019288:1019712 [0] NCCL INFO Channel 03/0 : 2[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942611:943034 [0] NCCL INFO Channel 03/0 : 6[0] -> 5[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942611:943034 [0] NCCL INFO Channel 02/0 : 5[0] -> 4[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019288:1019712 [0] NCCL INFO Channel 02/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019288:1019712 [0] NCCL INFO Channel 03/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942611:943034 [0] NCCL INFO Channel 03/0 : 5[0] -> 4[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618587:619010 [3] NCCL INFO Channel 02/0 : 7[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618587:619010 [3] NCCL INFO Channel 03/0 : 7[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618587:619010 [3] NCCL INFO Channel 02/0 : 3[3] -> 7[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618587:619010 [3] NCCL INFO Channel 03/0 : 3[3] -> 7[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019291:1019716 [3] NCCL INFO Connected all rings
+ip-26-0-172-116:1019291:1019716 [3] NCCL INFO Channel 02/0 : 1[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019291:1019716 [3] NCCL INFO Channel 03/0 : 1[3] -> 3[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019291:1019716 [3] NCCL INFO Channel 02/0 : 3[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019291:1019716 [3] NCCL INFO Channel 03/0 : 3[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618587:619010 [3] NCCL INFO Channel 02/0 : 5[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618587:619010 [3] NCCL INFO Channel 03/0 : 5[3] -> 3[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618587:619010 [3] NCCL INFO Channel 02/0 : 3[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618587:619010 [3] NCCL INFO Channel 03/0 : 3[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618587:619010 [3] NCCL INFO Channel 00/0 : 3[3] -> 2[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-7:811251:811698 [3] NCCL INFO Channel 00/0 : 7[3] -> 6[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019291:1019716 [3] NCCL INFO Channel 00/0 : 2[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618587:619010 [3] NCCL INFO Channel 01/0 : 3[3] -> 2[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019291:1019716 [3] NCCL INFO Channel 01/0 : 2[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-7:811251:811698 [3] NCCL INFO Channel 01/0 : 7[3] -> 6[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019291:1019716 [3] NCCL INFO Channel 02/0 : 2[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019291:1019716 [3] NCCL INFO Channel 03/0 : 2[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019291:1019716 [3] NCCL INFO Channel 02/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019291:1019716 [3] NCCL INFO Channel 03/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942614:943040 [3] NCCL INFO Channel 00/0 : 6[3] -> 5[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942614:943040 [3] NCCL INFO Channel 01/0 : 6[3] -> 5[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942614:943040 [3] NCCL INFO Channel 02/0 : 6[3] -> 5[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942614:943040 [3] NCCL INFO Channel 03/0 : 6[3] -> 5[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942614:943040 [3] NCCL INFO Channel 02/0 : 5[3] -> 4[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942614:943040 [3] NCCL INFO Channel 03/0 : 5[3] -> 4[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019290:1019714 [2] NCCL INFO Connected all rings
+ip-26-0-172-116:1019290:1019714 [2] NCCL INFO Channel 02/0 : 1[2] -> 3[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019290:1019714 [2] NCCL INFO Channel 03/0 : 1[2] -> 3[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019290:1019714 [2] NCCL INFO Channel 02/0 : 3[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019290:1019714 [2] NCCL INFO Channel 03/0 : 3[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618586:619008 [2] NCCL INFO Channel 02/0 : 7[2] -> 3[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618586:619008 [2] NCCL INFO Channel 03/0 : 7[2] -> 3[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618586:619008 [2] NCCL INFO Channel 02/0 : 3[2] -> 7[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618586:619008 [2] NCCL INFO Channel 03/0 : 3[2] -> 7[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618586:619008 [2] NCCL INFO Channel 02/0 : 5[2] -> 3[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618586:619008 [2] NCCL INFO Channel 03/0 : 5[2] -> 3[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-7:811250:811696 [2] NCCL INFO Channel 00/0 : 7[2] -> 6[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618586:619008 [2] NCCL INFO Channel 02/0 : 3[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-7:811250:811696 [2] NCCL INFO Channel 01/0 : 7[2] -> 6[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618586:619008 [2] NCCL INFO Channel 03/0 : 3[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942613:943038 [2] NCCL INFO Channel 00/0 : 6[2] -> 5[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618586:619008 [2] NCCL INFO Channel 00/0 : 3[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019290:1019714 [2] NCCL INFO Channel 00/0 : 2[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942613:943038 [2] NCCL INFO Channel 01/0 : 6[2] -> 5[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019290:1019714 [2] NCCL INFO Channel 01/0 : 2[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618586:619008 [2] NCCL INFO Channel 01/0 : 3[2] -> 2[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019290:1019714 [2] NCCL INFO Channel 02/0 : 2[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019290:1019714 [2] NCCL INFO Channel 03/0 : 2[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019290:1019714 [2] NCCL INFO Channel 02/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019290:1019714 [2] NCCL INFO Channel 03/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942613:943038 [2] NCCL INFO Channel 02/0 : 6[2] -> 5[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942613:943038 [2] NCCL INFO Channel 03/0 : 6[2] -> 5[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942613:943038 [2] NCCL INFO Channel 02/0 : 5[2] -> 4[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942613:943038 [2] NCCL INFO Channel 03/0 : 5[2] -> 4[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019289:1019717 [1] NCCL INFO Connected all rings
+ip-26-0-172-116:1019289:1019717 [1] NCCL INFO Channel 02/0 : 1[1] -> 3[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019289:1019717 [1] NCCL INFO Channel 03/0 : 1[1] -> 3[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019289:1019717 [1] NCCL INFO Channel 02/0 : 3[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019289:1019717 [1] NCCL INFO Channel 03/0 : 3[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618585:619011 [1] NCCL INFO Channel 02/0 : 7[1] -> 3[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618585:619011 [1] NCCL INFO Channel 03/0 : 7[1] -> 3[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618585:619011 [1] NCCL INFO Channel 02/0 : 3[1] -> 7[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618585:619011 [1] NCCL INFO Channel 03/0 : 3[1] -> 7[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618585:619011 [1] NCCL INFO Channel 02/0 : 5[1] -> 3[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618585:619011 [1] NCCL INFO Channel 03/0 : 5[1] -> 3[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618585:619011 [1] NCCL INFO Channel 02/0 : 3[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-7:811249:811699 [1] NCCL INFO Channel 00/0 : 7[1] -> 6[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618585:619011 [1] NCCL INFO Channel 03/0 : 3[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-7:811249:811699 [1] NCCL INFO Channel 01/0 : 7[1] -> 6[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618585:619011 [1] NCCL INFO Channel 00/0 : 3[1] -> 2[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942612:943041 [1] NCCL INFO Channel 00/0 : 6[1] -> 5[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019289:1019717 [1] NCCL INFO Channel 00/0 : 2[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618585:619011 [1] NCCL INFO Channel 01/0 : 3[1] -> 2[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019289:1019717 [1] NCCL INFO Channel 01/0 : 2[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942612:943041 [1] NCCL INFO Channel 01/0 : 6[1] -> 5[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942612:943041 [1] NCCL INFO Channel 02/0 : 6[1] -> 5[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019289:1019717 [1] NCCL INFO Channel 02/0 : 2[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942612:943041 [1] NCCL INFO Channel 03/0 : 6[1] -> 5[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019289:1019717 [1] NCCL INFO Channel 03/0 : 2[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942612:943041 [1] NCCL INFO Channel 02/0 : 5[1] -> 4[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019289:1019717 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942612:943041 [1] NCCL INFO Channel 03/0 : 5[1] -> 4[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019289:1019717 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:94349:94804 [7] NCCL INFO Connected all trees
+ip-26-0-160-225:94349:94804 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-225:94349:94804 [7] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-160-225:94349:94804 [7] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-160-225:94349:94804 [7] NCCL INFO comm 0xa3736b0 rank 0 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0xca6ca50e1a87f46c - Init COMPLETE
+ip-26-0-172-147:618588:619004 [4] NCCL INFO Connected all trees
+ip-26-0-172-147:618588:619004 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-147:618588:619004 [4] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-172-147:618588:619004 [4] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-160-225:94346:94798 [4] NCCL INFO Connected all trees
+ip-26-0-160-225:94346:94798 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-225:94346:94798 [4] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-160-225:94346:94798 [4] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-147:618588:619004 [4] NCCL INFO comm 0x9d70c20 rank 3 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0x886b815c42972199 - Init COMPLETE
+ip-26-0-160-225:94348:94808 [6] NCCL INFO Connected all trees
+ip-26-0-160-225:94348:94808 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-225:94348:94808 [6] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-160-225:94348:94808 [6] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-160-225:94347:94800 [5] NCCL INFO Connected all trees
+ip-26-0-160-225:94347:94800 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-225:94347:94800 [5] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-160-225:94347:94800 [5] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-160-225:94346:94798 [4] NCCL INFO comm 0xba0dd40 rank 0 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0x886b815c42972199 - Init COMPLETE
+ip-26-0-160-225:94348:94808 [6] NCCL INFO comm 0xb16cac0 rank 0 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0x45d17d92b31ef879 - Init COMPLETE
+ip-26-0-173-7:811255:811695 [7] NCCL INFO Connected all trees
+ip-26-0-173-7:811255:811695 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-173-7:811255:811695 [7] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-173-7:811255:811695 [7] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-142:880684:881100 [5] NCCL INFO Connected all trees
+ip-26-0-172-142:880684:881100 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-142:880684:881100 [5] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-172-142:880684:881100 [5] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-160-225:94347:94800 [5] NCCL INFO comm 0xaaa4c20 rank 0 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0xa25ee5d67de91a57 - Init COMPLETE
+ip-26-0-173-7:811254:811697 [6] NCCL INFO Connected all trees
+ip-26-0-173-7:811254:811697 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-173-7:811254:811697 [6] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-173-7:811254:811697 [6] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-173-7:811255:811695 [7] NCCL INFO comm 0xa8df660 rank 7 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0xca6ca50e1a87f46c - Init COMPLETE
+ip-26-0-172-142:880684:881100 [5] NCCL INFO comm 0x9fca120 rank 2 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0xa25ee5d67de91a57 - Init COMPLETE
+ip-26-0-173-7:811254:811697 [6] NCCL INFO comm 0x969a920 rank 7 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0x45d17d92b31ef879 - Init COMPLETE
+ip-26-0-173-7:811251:811698 [3] NCCL INFO Connected all trees
+ip-26-0-173-7:811251:811698 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-173-7:811251:811698 [3] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-173-7:811251:811698 [3] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-173-7:811251:811698 [3] NCCL INFO comm 0x9892600 rank 7 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0xa72f98a98e82f37d - Init COMPLETE
+ip-26-0-160-225:94342:94802 [0] NCCL INFO Connected all trees
+ip-26-0-160-225:94342:94802 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-225:94342:94802 [0] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-160-225:94342:94802 [0] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-116:1019293:1019711 [5] NCCL INFO Connected all trees
+ip-26-0-172-116:1019293:1019711 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-116:1019293:1019711 [5] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-172-116:1019293:1019711 [5] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-173-7:811248:811694 [0] NCCL INFO Connected all trees
+ip-26-0-173-7:811248:811694 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-173-7:811248:811694 [0] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-173-7:811248:811694 [0] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-160-225:94342:94802 [0] NCCL INFO comm 0xa836620 rank 0 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0x242f1fc2efc03fb0 - Init COMPLETE
+ip-26-0-172-142:880683:881099 [4] NCCL INFO Connected all trees
+ip-26-0-172-142:880683:881099 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-142:880683:881099 [4] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-172-142:880683:881099 [4] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-73:926601:927018 [7] NCCL INFO Connected all trees
+ip-26-0-172-73:926601:927018 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-73:926601:927018 [7] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-172-73:926601:927018 [7] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-116:1019293:1019711 [5] NCCL INFO comm 0xb94aca0 rank 1 nranks 8 cudaDev 5 nvmlDev 5 busId a8000 commId 0xa25ee5d67de91a57 - Init COMPLETE
+ip-26-0-173-7:811248:811694 [0] NCCL INFO comm 0xb8c6260 rank 7 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0x242f1fc2efc03fb0 - Init COMPLETE
+ip-26-0-172-142:880683:881099 [4] NCCL INFO comm 0xb1a2ce0 rank 2 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0x886b815c42972199 - Init COMPLETE
+ip-26-0-172-73:926601:927018 [7] NCCL INFO comm 0x9af5e90 rank 6 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0xca6ca50e1a87f46c - Init COMPLETE
+ip-26-0-172-116:1019292:1019710 [4] NCCL INFO Connected all trees
+ip-26-0-172-116:1019292:1019710 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-116:1019292:1019710 [4] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-172-116:1019292:1019710 [4] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-116:1019292:1019710 [4] NCCL INFO comm 0xb8a8ed0 rank 1 nranks 8 cudaDev 4 nvmlDev 4 busId 97000 commId 0x886b815c42972199 - Init COMPLETE
+ip-26-0-160-225:94343:94812 [1] NCCL INFO Connected all trees
+ip-26-0-160-225:94343:94812 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-225:94343:94812 [1] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-160-225:94343:94812 [1] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-142:880686:881102 [7] NCCL INFO Connected all trees
+ip-26-0-172-142:880686:881102 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-142:880686:881102 [7] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-172-142:880686:881102 [7] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-173-7:811250:811696 [2] NCCL INFO Connected all trees
+ip-26-0-173-7:811250:811696 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-173-7:811250:811696 [2] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-173-7:811250:811696 [2] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-160-225:94343:94812 [1] NCCL INFO comm 0x9c757e0 rank 0 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0x502faa1c75e98308 - Init COMPLETE
+ip-26-0-160-225:94344:94806 [2] NCCL INFO Connected all trees
+ip-26-0-160-225:94344:94806 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-225:94344:94806 [2] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-160-225:94344:94806 [2] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-73:926600:927020 [6] NCCL INFO Connected all trees
+ip-26-0-172-73:926600:927020 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-73:926600:927020 [6] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-172-73:926600:927020 [6] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-142:880686:881102 [7] NCCL INFO comm 0xac750d0 rank 2 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0xca6ca50e1a87f46c - Init COMPLETE
+ip-26-0-173-7:811250:811696 [2] NCCL INFO comm 0xa294550 rank 7 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0x347557a4b53fac7e - Init COMPLETE
+ip-26-0-172-73:926594:927017 [0] NCCL INFO Connected all trees
+ip-26-0-172-73:926594:927017 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-73:926594:927017 [0] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-172-73:926594:927017 [0] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-57:942617:943039 [6] NCCL INFO Connected all trees
+ip-26-0-172-57:942617:943039 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-57:942617:943039 [6] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-172-57:942617:943039 [6] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-160-225:94345:94810 [3] NCCL INFO Connected all trees
+ip-26-0-160-225:94345:94810 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-225:94345:94810 [3] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-160-225:94345:94810 [3] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-173-7:811249:811699 [1] NCCL INFO Connected all trees
+ip-26-0-173-7:811249:811699 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-173-7:811249:811699 [1] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-173-7:811249:811699 [1] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-73:926600:927020 [6] NCCL INFO comm 0xb525f40 rank 6 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0x45d17d92b31ef879 - Init COMPLETE
+ip-26-0-160-225:94344:94806 [2] NCCL INFO comm 0xa9caa60 rank 0 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0x347557a4b53fac7e - Init COMPLETE
+ip-26-0-172-57:942617:943039 [6] NCCL INFO comm 0xacba8a0 rank 5 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0x45d17d92b31ef879 - Init COMPLETE
+ip-26-0-172-116:1019295:1019713 [7] NCCL INFO Connected all trees
+ip-26-0-172-116:1019295:1019713 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-116:1019295:1019713 [7] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-172-116:1019295:1019713 [7] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-160-225:94345:94810 [3] NCCL INFO comm 0xbba9960 rank 0 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0xa72f98a98e82f37d - Init COMPLETE
+ip-26-0-172-73:926594:927017 [0] NCCL INFO comm 0xa9281d0 rank 6 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0x242f1fc2efc03fb0 - Init COMPLETE
+ip-26-0-172-73:926597:927021 [3] NCCL INFO Connected all trees
+ip-26-0-172-73:926597:927021 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-73:926597:927021 [3] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-172-73:926597:927021 [3] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-57:942618:943035 [7] NCCL INFO Connected all trees
+ip-26-0-172-57:942618:943035 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-57:942618:943035 [7] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-172-57:942618:943035 [7] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-142:880685:881104 [6] NCCL INFO Connected all trees
+ip-26-0-172-142:880685:881104 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-142:880685:881104 [6] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-172-142:880685:881104 [6] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-173-7:811249:811699 [1] NCCL INFO comm 0x9495220 rank 7 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0x502faa1c75e98308 - Init COMPLETE
+ip-26-0-172-252:688207:688626 [7] NCCL INFO Connected all trees
+ip-26-0-172-252:688207:688626 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-252:688207:688626 [7] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-172-252:688207:688626 [7] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-116:1019295:1019713 [7] NCCL INFO comm 0xaf12270 rank 1 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0xca6ca50e1a87f46c - Init COMPLETE
+ip-26-0-172-142:880679:881101 [0] NCCL INFO Connected all trees
+ip-26-0-172-142:880679:881101 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-142:880679:881101 [0] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-172-142:880679:881101 [0] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-57:942618:943035 [7] NCCL INFO comm 0xa546420 rank 5 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0xca6ca50e1a87f46c - Init COMPLETE
+ip-26-0-172-73:926597:927021 [3] NCCL INFO comm 0xa1f68d0 rank 6 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0xa72f98a98e82f37d - Init COMPLETE
+ip-26-0-172-147:618591:619007 [7] NCCL INFO Connected all trees
+ip-26-0-172-147:618591:619007 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-147:618591:619007 [7] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-172-147:618591:619007 [7] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-142:880685:881104 [6] NCCL INFO comm 0x9e7e370 rank 2 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0x45d17d92b31ef879 - Init COMPLETE
+ip-26-0-172-252:688207:688626 [7] NCCL INFO comm 0xa48c680 rank 4 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0xca6ca50e1a87f46c - Init COMPLETE
+ip-26-0-172-142:880679:881101 [0] NCCL INFO comm 0xb7085d0 rank 2 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0x242f1fc2efc03fb0 - Init COMPLETE
+ip-26-0-172-73:926596:927019 [2] NCCL INFO Connected all trees
+ip-26-0-172-73:926596:927019 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-73:926596:927019 [2] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-172-73:926596:927019 [2] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-116:1019294:1019715 [6] NCCL INFO Connected all trees
+ip-26-0-172-116:1019294:1019715 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-116:1019294:1019715 [6] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-172-116:1019294:1019715 [6] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-116:1019288:1019712 [0] NCCL INFO Connected all trees
+ip-26-0-172-116:1019288:1019712 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-116:1019288:1019712 [0] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-172-116:1019288:1019712 [0] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-147:618591:619007 [7] NCCL INFO comm 0xa8e6270 rank 3 nranks 8 cudaDev 7 nvmlDev 7 busId ca000 commId 0xca6ca50e1a87f46c - Init COMPLETE
+ip-26-0-172-73:926596:927019 [2] NCCL INFO comm 0xa369f60 rank 6 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0x347557a4b53fac7e - Init COMPLETE
+ip-26-0-172-147:618590:619009 [6] NCCL INFO Connected all trees
+ip-26-0-172-147:618590:619009 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-147:618590:619009 [6] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-172-147:618590:619009 [6] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-116:1019291:1019716 [3] NCCL INFO Connected all trees
+ip-26-0-172-116:1019291:1019716 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-116:1019291:1019716 [3] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-172-116:1019291:1019716 [3] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-73:926595:927022 [1] NCCL INFO Connected all trees
+ip-26-0-172-73:926595:927022 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-73:926595:927022 [1] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-172-73:926595:927022 [1] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-116:1019294:1019715 [6] NCCL INFO comm 0x9b47f70 rank 1 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0x45d17d92b31ef879 - Init COMPLETE
+ip-26-0-172-57:942611:943034 [0] NCCL INFO Connected all trees
+ip-26-0-172-57:942611:943034 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-57:942611:943034 [0] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-172-57:942611:943034 [0] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-116:1019288:1019712 [0] NCCL INFO comm 0xa175fc0 rank 1 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0x242f1fc2efc03fb0 - Init COMPLETE
+ip-26-0-172-116:1019291:1019716 [3] NCCL INFO comm 0xb8bef40 rank 1 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0xa72f98a98e82f37d - Init COMPLETE
+ip-26-0-172-73:926595:927022 [1] NCCL INFO comm 0x9a19f50 rank 6 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0x502faa1c75e98308 - Init COMPLETE
+ip-26-0-172-147:618590:619009 [6] NCCL INFO comm 0xb6d0ba0 rank 3 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0x45d17d92b31ef879 - Init COMPLETE
+ip-26-0-172-57:942611:943034 [0] NCCL INFO comm 0xb2efa90 rank 5 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0x242f1fc2efc03fb0 - Init COMPLETE
+ip-26-0-172-116:1019290:1019714 [2] NCCL INFO Connected all trees
+ip-26-0-172-116:1019290:1019714 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-116:1019290:1019714 [2] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-172-116:1019290:1019714 [2] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-252:688206:688628 [6] NCCL INFO Connected all trees
+ip-26-0-172-252:688206:688628 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-252:688206:688628 [6] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-172-252:688206:688628 [6] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-252:688200:688625 [0] NCCL INFO Connected all trees
+ip-26-0-172-252:688200:688625 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-252:688200:688625 [0] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-172-252:688200:688625 [0] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-57:942614:943040 [3] NCCL INFO Connected all trees
+ip-26-0-172-57:942614:943040 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-57:942614:943040 [3] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-172-57:942614:943040 [3] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-116:1019289:1019717 [1] NCCL INFO Connected all trees
+ip-26-0-172-116:1019289:1019717 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-116:1019289:1019717 [1] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-172-116:1019289:1019717 [1] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-116:1019290:1019714 [2] NCCL INFO comm 0xb5359e0 rank 1 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0x347557a4b53fac7e - Init COMPLETE
+ip-26-0-172-147:618584:619006 [0] NCCL INFO Connected all trees
+ip-26-0-172-147:618584:619006 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-147:618584:619006 [0] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-172-147:618584:619006 [0] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-142:880682:881105 [3] NCCL INFO Connected all trees
+ip-26-0-172-142:880682:881105 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-142:880682:881105 [3] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-172-142:880682:881105 [3] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-252:688203:688629 [3] NCCL INFO Connected all trees
+ip-26-0-172-252:688203:688629 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-252:688203:688629 [3] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-172-252:688203:688629 [3] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-252:688206:688628 [6] NCCL INFO comm 0xb96fa70 rank 4 nranks 8 cudaDev 6 nvmlDev 6 busId b9000 commId 0x45d17d92b31ef879 - Init COMPLETE
+ip-26-0-172-252:688200:688625 [0] NCCL INFO comm 0xa91f640 rank 4 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0x242f1fc2efc03fb0 - Init COMPLETE
+ip-26-0-172-116:1019289:1019717 [1] NCCL INFO comm 0x9b48f20 rank 1 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0x502faa1c75e98308 - Init COMPLETE
+ip-26-0-172-57:942614:943040 [3] NCCL INFO comm 0x9edc950 rank 5 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0xa72f98a98e82f37d - Init COMPLETE
+ip-26-0-172-147:618584:619006 [0] NCCL INFO comm 0xb0208b0 rank 3 nranks 8 cudaDev 0 nvmlDev 0 busId 53000 commId 0x242f1fc2efc03fb0 - Init COMPLETE
+ip-26-0-172-142:880682:881105 [3] NCCL INFO comm 0xaa271d0 rank 2 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0xa72f98a98e82f37d - Init COMPLETE
+ip-26-0-172-252:688203:688629 [3] NCCL INFO comm 0x9d167c0 rank 4 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0xa72f98a98e82f37d - Init COMPLETE
+ip-26-0-172-147:618587:619010 [3] NCCL INFO Connected all trees
+ip-26-0-172-147:618587:619010 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-147:618587:619010 [3] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-172-147:618587:619010 [3] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-142:880680:881106 [1] NCCL INFO Connected all trees
+ip-26-0-172-142:880680:881106 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-142:880680:881106 [1] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-172-142:880680:881106 [1] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-57:942613:943038 [2] NCCL INFO Connected all trees
+ip-26-0-172-57:942613:943038 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-57:942613:943038 [2] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-172-57:942613:943038 [2] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-147:618587:619010 [3] NCCL INFO comm 0xa0b1130 rank 3 nranks 8 cudaDev 3 nvmlDev 3 busId 86000 commId 0xa72f98a98e82f37d - Init COMPLETE
+ip-26-0-172-252:688202:688627 [2] NCCL INFO Connected all trees
+ip-26-0-172-252:688202:688627 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-252:688202:688627 [2] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-172-252:688202:688627 [2] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-147:618586:619008 [2] NCCL INFO Connected all trees
+ip-26-0-172-147:618586:619008 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-147:618586:619008 [2] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-172-147:618586:619008 [2] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-57:942612:943041 [1] NCCL INFO Connected all trees
+ip-26-0-172-57:942612:943041 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-57:942612:943041 [1] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-172-57:942612:943041 [1] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-142:880680:881106 [1] NCCL INFO comm 0xa4bbcf0 rank 2 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0x502faa1c75e98308 - Init COMPLETE
+ip-26-0-172-252:688201:688630 [1] NCCL INFO Connected all trees
+ip-26-0-172-252:688201:688630 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-252:688201:688630 [1] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-172-252:688201:688630 [1] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-142:880681:881103 [2] NCCL INFO Connected all trees
+ip-26-0-172-142:880681:881103 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-142:880681:881103 [2] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-172-142:880681:881103 [2] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-57:942613:943038 [2] NCCL INFO comm 0xb06cc90 rank 5 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0x347557a4b53fac7e - Init COMPLETE
+ip-26-0-172-252:688202:688627 [2] NCCL INFO comm 0xa7a9a50 rank 4 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0x347557a4b53fac7e - Init COMPLETE
+ip-26-0-172-147:618586:619008 [2] NCCL INFO comm 0xa0655f0 rank 3 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0x347557a4b53fac7e - Init COMPLETE
+ip-26-0-172-147:618585:619011 [1] NCCL INFO Connected all trees
+ip-26-0-172-147:618585:619011 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-147:618585:619011 [1] NCCL INFO threadThresholds 8/8/64 | 64/8/64 | 512 | 512
+ip-26-0-172-147:618585:619011 [1] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-57:942612:943041 [1] NCCL INFO comm 0xa0ad600 rank 5 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0x502faa1c75e98308 - Init COMPLETE
+ip-26-0-172-252:688201:688630 [1] NCCL INFO comm 0xa8c5490 rank 4 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0x502faa1c75e98308 - Init COMPLETE
+ip-26-0-172-142:880681:881103 [2] NCCL INFO comm 0xa582810 rank 2 nranks 8 cudaDev 2 nvmlDev 2 busId 75000 commId 0x347557a4b53fac7e - Init COMPLETE
+ip-26-0-172-147:618585:619011 [1] NCCL INFO comm 0xb534180 rank 3 nranks 8 cudaDev 1 nvmlDev 1 busId 64000 commId 0x502faa1c75e98308 - Init COMPLETE
+01/07/2025 05:27:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: Total number of parameters: 3.98G (7584.33MiB)
+01/07/2025 05:27:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: Local number of parameters: 856M (1632.05MiB)
+01/07/2025 05:27:21 [INFO|DP=0|PP=4|TP=0|ip-26-0-172-252]: Local number of parameters: 453M (864.05MiB)
+01/07/2025 05:27:21 [INFO|DP=0|PP=6|TP=0|ip-26-0-172-73]: Local number of parameters: 453M (864.05MiB)
+01/07/2025 05:27:21 [INFO|DP=0|PP=7|TP=0|ip-26-0-173-7]: Local number of parameters: 403M (768.01MiB)
+01/07/2025 05:27:21 [INFO|DP=0|PP=2|TP=0|ip-26-0-172-142]: Local number of parameters: 453M (864.05MiB)
+01/07/2025 05:27:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [After model building] Memory usage: 1632.06MiB. Peak allocated: 5408.00MiB Peak reserved: 11554.00MiB
+01/07/2025 05:27:21 [INFO|DP=0|PP=4|TP=0|ip-26-0-172-252]: [After model building] Memory usage: 864.06MiB. Peak allocated: 5408.00MiB Peak reserved: 11554.00MiB
+01/07/2025 05:27:21 [INFO|DP=0|PP=6|TP=0|ip-26-0-172-73]: [After model building] Memory usage: 864.06MiB. Peak allocated: 5408.00MiB Peak reserved: 11554.00MiB
+01/07/2025 05:27:21 [INFO|DP=0|PP=2|TP=0|ip-26-0-172-142]: [After model building] Memory usage: 864.06MiB. Peak allocated: 5408.00MiB Peak reserved: 10530.00MiB
+01/07/2025 05:27:21 [INFO|DP=0|PP=7|TP=0|ip-26-0-173-7]: [After model building] Memory usage: 768.02MiB. Peak allocated: 5408.00MiB Peak reserved: 11554.00MiB
+01/07/2025 05:27:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: No checkpoint path provided.
+01/07/2025 05:27:21 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: Parametrizing model parameters using StandardParametrizator
+01/07/2025 05:27:21 [INFO|DP=0|PP=1|TP=0|ip-26-0-172-116]: Local number of parameters: 453M (864.05MiB)
+01/07/2025 05:27:21 [INFO|DP=0|PP=5|TP=0|ip-26-0-172-57]: Local number of parameters: 453M (864.05MiB)
+01/07/2025 05:27:21 [INFO|DP=0|PP=3|TP=0|ip-26-0-172-147]: Local number of parameters: 453M (864.05MiB)
+01/07/2025 05:27:21 [INFO|DP=0|PP=1|TP=0|ip-26-0-172-116]: [After model building] Memory usage: 864.06MiB. Peak allocated: 5408.00MiB Peak reserved: 11554.00MiB
+01/07/2025 05:27:21 [INFO|DP=0|PP=5|TP=0|ip-26-0-172-57]: [After model building] Memory usage: 864.06MiB. Peak allocated: 5408.00MiB Peak reserved: 11554.00MiB
+01/07/2025 05:27:21 [INFO|DP=0|PP=3|TP=0|ip-26-0-172-147]: [After model building] Memory usage: 864.06MiB. Peak allocated: 5408.00MiB Peak reserved: 11554.00MiB
+ip-26-0-160-225:94344:94834 [2] NCCL INFO Using network Libfabric
+ip-26-0-160-225:94343:94840 [1] NCCL INFO Using network Libfabric
+ip-26-0-160-225:94342:94838 [0] NCCL INFO Using network Libfabric
+ip-26-0-160-225:94347:94842 [5] NCCL INFO Using network Libfabric
+ip-26-0-160-225:94346:94843 [4] NCCL INFO Using network Libfabric
+ip-26-0-173-7:811250:811718 [2] NCCL INFO Using network Libfabric
+ip-26-0-160-225:94348:94845 [6] NCCL INFO Using network Libfabric
+ip-26-0-173-7:811248:811719 [0] NCCL INFO Using network Libfabric
+ip-26-0-173-7:811253:811720 [5] NCCL INFO Using network Libfabric
+ip-26-0-160-225:94345:94847 [3] NCCL INFO Using network Libfabric
+ip-26-0-160-225:94349:94848 [7] NCCL INFO Using network Libfabric
+ip-26-0-173-7:811254:811721 [6] NCCL INFO Using network Libfabric
+ip-26-0-173-7:811252:811723 [4] NCCL INFO Using network Libfabric
+ip-26-0-173-7:811251:811724 [3] NCCL INFO Using network Libfabric
+ip-26-0-173-7:811255:811725 [7] NCCL INFO Using network Libfabric
+ip-26-0-173-7:811248:811719 [0] NCCL INFO comm 0xbaa41f0 rank 1 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0xbff43a723081b14 - Init START
+ip-26-0-173-7:811250:811718 [2] NCCL INFO comm 0xaeafb40 rank 1 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0x5ce82daf32c3aa8d - Init START
+ip-26-0-160-225:94342:94838 [0] NCCL INFO comm 0xb2b6050 rank 0 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0xbff43a723081b14 - Init START
+ip-26-0-160-225:94344:94834 [2] NCCL INFO comm 0xb44ce30 rank 0 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0x5ce82daf32c3aa8d - Init START
+ip-26-0-160-225:94342:94838 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/349
+ip-26-0-173-7:811248:811719 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-160-225:94344:94834 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-160-225:94347:94842 [5] NCCL INFO comm 0xb526b40 rank 0 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0x5f53f5fc2baeae2 - Init START
+ip-26-0-173-7:811250:811718 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-173-7:811249:811722 [1] NCCL INFO Using network Libfabric
+ip-26-0-173-7:811253:811720 [5] NCCL INFO comm 0xaf21380 rank 1 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0x5f53f5fc2baeae2 - Init START
+ip-26-0-160-225:94348:94845 [6] NCCL INFO comm 0xbbef680 rank 0 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0x4d13c7f185f0b9cd - Init START
+ip-26-0-160-225:94347:94842 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-160-225:94346:94843 [4] NCCL INFO comm 0xc48ff40 rank 0 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0xf8c2d10e47b238ab - Init START
+ip-26-0-173-7:811254:811721 [6] NCCL INFO comm 0xa2b7900 rank 1 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0x4d13c7f185f0b9cd - Init START
+ip-26-0-160-225:94345:94847 [3] NCCL INFO comm 0xc628db0 rank 0 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0x957e4f4cd9e3e0ff - Init START
+ip-26-0-160-225:94348:94845 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-173-7:811252:811723 [4] NCCL INFO comm 0xa5cb730 rank 1 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0xf8c2d10e47b238ab - Init START
+ip-26-0-173-7:811253:811720 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-173-7:811251:811724 [3] NCCL INFO comm 0xa4ae480 rank 1 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0x957e4f4cd9e3e0ff - Init START
+ip-26-0-160-225:94346:94843 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-173-7:811254:811721 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-160-225:94345:94847 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-173-7:811252:811723 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-160-225:94349:94848 [7] NCCL INFO comm 0xadf7a60 rank 0 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0x76201d02e3efd9f9 - Init START
+ip-26-0-173-7:811251:811724 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-173-7:811255:811725 [7] NCCL INFO comm 0xb4fbe30 rank 1 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0x76201d02e3efd9f9 - Init START
+ip-26-0-160-225:94349:94848 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-173-7:811255:811725 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-160-225:94343:94840 [1] NCCL INFO comm 0xa6f7d10 rank 0 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0xfaaf189a91460e79 - Init START
+ip-26-0-173-7:811249:811722 [1] NCCL INFO comm 0xa0b1810 rank 1 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0xfaaf189a91460e79 - Init START
+ip-26-0-160-225:94343:94840 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-173-7:811249:811722 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-173-7:811251:811724 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-160-225:94346:94843 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-173-7:811248:811719 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+ip-26-0-160-225:94344:94834 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-160-225:94345:94847 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-160-225:94345:94847 [3] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-160-225:94345:94847 [3] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-160-225:94345:94847 [3] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-160-225:94345:94847 [3] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-160-225:94345:94847 [3] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-160-225:94345:94847 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-173-7:811251:811724 [3] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-173-7:811251:811724 [3] NCCL INFO P2P Chunksize set to 131072
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+ip-26-0-173-7:811250:811718 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+ip-26-0-173-7:811250:811718 [2] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-173-7:811250:811718 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-225:94344:94834 [2] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-160-225:94344:94834 [2] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-160-225:94344:94834 [2] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-160-225:94344:94834 [2] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-160-225:94344:94834 [2] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-160-225:94344:94834 [2] NCCL INFO P2P Chunksize set to 131072
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+ip-26-0-173-7:811255:811725 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+ip-26-0-160-225:94349:94848 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+ip-26-0-160-225:94349:94848 [7] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-160-225:94349:94848 [7] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-160-225:94349:94848 [7] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-160-225:94349:94848 [7] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-160-225:94349:94848 [7] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-160-225:94349:94848 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-173-7:811255:811725 [7] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-173-7:811255:811725 [7] NCCL INFO P2P Chunksize set to 131072
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+ip-26-0-173-7:811252:811723 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-173-7:811252:811723 [4] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-173-7:811252:811723 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-225:94346:94843 [4] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-160-225:94346:94843 [4] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-160-225:94346:94843 [4] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-160-225:94346:94843 [4] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-160-225:94346:94843 [4] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-160-225:94346:94843 [4] NCCL INFO P2P Chunksize set to 131072
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+ip-26-0-160-225:94343:94840 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-173-7:811253:811720 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-173-7:811254:811721 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+ip-26-0-173-7:811249:811722 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+ip-26-0-160-225:94343:94840 [1] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-160-225:94343:94840 [1] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-160-225:94343:94840 [1] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-160-225:94343:94840 [1] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-160-225:94343:94840 [1] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-160-225:94343:94840 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-173-7:811249:811722 [1] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-173-7:811249:811722 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-225:94342:94838 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+ip-26-0-160-225:94342:94838 [0] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-160-225:94342:94838 [0] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-160-225:94342:94838 [0] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-160-225:94342:94838 [0] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-160-225:94342:94838 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-160-225:94342:94838 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-173-7:811248:811719 [0] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-173-7:811248:811719 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-225:94348:94845 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+ip-26-0-160-225:94348:94845 [6] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-160-225:94348:94845 [6] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-160-225:94348:94845 [6] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-160-225:94348:94845 [6] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-160-225:94348:94845 [6] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-160-225:94348:94845 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-173-7:811254:811721 [6] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-173-7:811254:811721 [6] NCCL INFO P2P Chunksize set to 131072
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+ip-26-0-160-225:94347:94842 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+ip-26-0-160-225:94347:94842 [5] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-160-225:94347:94842 [5] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-160-225:94347:94842 [5] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-160-225:94347:94842 [5] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-160-225:94347:94842 [5] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-160-225:94347:94842 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-173-7:811253:811720 [5] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-173-7:811253:811720 [5] NCCL INFO P2P Chunksize set to 131072
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+ip-26-0-173-7:811251:811724 [3] NCCL INFO Channel 00/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-7:811251:811724 [3] NCCL INFO Channel 01/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-7:811251:811724 [3] NCCL INFO Channel 02/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-7:811251:811724 [3] NCCL INFO Channel 03/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-7:811251:811724 [3] NCCL INFO Channel 00/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-7:811251:811724 [3] NCCL INFO Channel 01/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-7:811251:811724 [3] NCCL INFO Channel 02/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:94345:94847 [3] NCCL INFO Channel 00/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-7:811251:811724 [3] NCCL INFO Channel 03/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:94345:94847 [3] NCCL INFO Channel 01/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:94345:94847 [3] NCCL INFO Channel 02/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:94345:94847 [3] NCCL INFO Channel 03/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:94345:94847 [3] NCCL INFO Channel 00/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:94345:94847 [3] NCCL INFO Channel 01/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:94345:94847 [3] NCCL INFO Channel 02/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:94345:94847 [3] NCCL INFO Channel 03/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+ip-26-0-173-7:811250:811718 [2] NCCL INFO Channel 00/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-7:811250:811718 [2] NCCL INFO Channel 01/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-7:811250:811718 [2] NCCL INFO Channel 02/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-7:811250:811718 [2] NCCL INFO Channel 03/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-7:811250:811718 [2] NCCL INFO Channel 00/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-7:811250:811718 [2] NCCL INFO Channel 01/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-7:811250:811718 [2] NCCL INFO Channel 02/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-7:811250:811718 [2] NCCL INFO Channel 03/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:94344:94834 [2] NCCL INFO Channel 00/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:94344:94834 [2] NCCL INFO Channel 01/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:94344:94834 [2] NCCL INFO Channel 02/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:94344:94834 [2] NCCL INFO Channel 03/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:94344:94834 [2] NCCL INFO Channel 00/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:94344:94834 [2] NCCL INFO Channel 01/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+ip-26-0-160-225:94344:94834 [2] NCCL INFO Channel 02/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:94344:94834 [2] NCCL INFO Channel 03/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:94349:94848 [7] NCCL INFO Channel 00/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:94349:94848 [7] NCCL INFO Channel 01/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:94349:94848 [7] NCCL INFO Channel 02/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:94349:94848 [7] NCCL INFO Channel 03/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:94349:94848 [7] NCCL INFO Channel 00/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:94349:94848 [7] NCCL INFO Channel 01/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:94349:94848 [7] NCCL INFO Channel 02/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:94349:94848 [7] NCCL INFO Channel 03/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-7:811255:811725 [7] NCCL INFO Channel 00/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-7:811255:811725 [7] NCCL INFO Channel 01/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-7:811255:811725 [7] NCCL INFO Channel 02/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-7:811255:811725 [7] NCCL INFO Channel 03/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-7:811255:811725 [7] NCCL INFO Channel 00/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-7:811255:811725 [7] NCCL INFO Channel 01/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-7:811255:811725 [7] NCCL INFO Channel 02/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-7:811255:811725 [7] NCCL INFO Channel 03/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-7:811252:811723 [4] NCCL INFO Channel 00/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-7:811252:811723 [4] NCCL INFO Channel 01/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-7:811252:811723 [4] NCCL INFO Channel 02/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-7:811252:811723 [4] NCCL INFO Channel 03/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-7:811252:811723 [4] NCCL INFO Channel 00/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-7:811252:811723 [4] NCCL INFO Channel 01/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-7:811252:811723 [4] NCCL INFO Channel 02/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-7:811252:811723 [4] NCCL INFO Channel 03/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:94346:94843 [4] NCCL INFO Channel 00/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:94346:94843 [4] NCCL INFO Channel 01/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:94346:94843 [4] NCCL INFO Channel 02/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:94346:94843 [4] NCCL INFO Channel 03/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:94346:94843 [4] NCCL INFO Channel 00/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:94346:94843 [4] NCCL INFO Channel 01/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:94346:94843 [4] NCCL INFO Channel 02/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:94346:94843 [4] NCCL INFO Channel 03/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-7:811249:811722 [1] NCCL INFO Channel 00/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-7:811249:811722 [1] NCCL INFO Channel 01/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-7:811249:811722 [1] NCCL INFO Channel 02/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:94343:94840 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-7:811249:811722 [1] NCCL INFO Channel 03/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:94343:94840 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-7:811249:811722 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-7:811249:811722 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:94343:94840 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-7:811249:811722 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:94343:94840 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-7:811249:811722 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:94343:94840 [1] NCCL INFO Channel 00/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:94343:94840 [1] NCCL INFO Channel 01/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:94343:94840 [1] NCCL INFO Channel 02/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:94343:94840 [1] NCCL INFO Channel 03/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:94342:94838 [0] NCCL INFO Channel 00/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:94342:94838 [0] NCCL INFO Channel 01/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:94342:94838 [0] NCCL INFO Channel 02/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:94342:94838 [0] NCCL INFO Channel 03/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:94342:94838 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:94342:94838 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:94342:94838 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:94342:94838 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-7:811248:811719 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-7:811248:811719 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-7:811248:811719 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-7:811248:811719 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-7:811248:811719 [0] NCCL INFO Channel 00/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-7:811248:811719 [0] NCCL INFO Channel 01/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-7:811248:811719 [0] NCCL INFO Channel 02/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-7:811248:811719 [0] NCCL INFO Channel 03/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-7:811254:811721 [6] NCCL INFO Channel 00/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-7:811254:811721 [6] NCCL INFO Channel 01/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-7:811254:811721 [6] NCCL INFO Channel 02/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-7:811254:811721 [6] NCCL INFO Channel 03/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-7:811254:811721 [6] NCCL INFO Channel 00/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:94348:94845 [6] NCCL INFO Channel 00/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-7:811254:811721 [6] NCCL INFO Channel 01/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-7:811254:811721 [6] NCCL INFO Channel 02/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:94348:94845 [6] NCCL INFO Channel 01/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-7:811254:811721 [6] NCCL INFO Channel 03/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:94348:94845 [6] NCCL INFO Channel 02/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:94348:94845 [6] NCCL INFO Channel 03/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:94348:94845 [6] NCCL INFO Channel 00/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:94348:94845 [6] NCCL INFO Channel 01/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:94348:94845 [6] NCCL INFO Channel 02/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:94348:94845 [6] NCCL INFO Channel 03/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:94347:94842 [5] NCCL INFO Channel 00/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:94347:94842 [5] NCCL INFO Channel 01/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:94347:94842 [5] NCCL INFO Channel 02/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:94347:94842 [5] NCCL INFO Channel 03/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:94347:94842 [5] NCCL INFO Channel 00/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:94347:94842 [5] NCCL INFO Channel 01/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:94347:94842 [5] NCCL INFO Channel 02/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:94347:94842 [5] NCCL INFO Channel 03/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-7:811253:811720 [5] NCCL INFO Channel 00/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-7:811253:811720 [5] NCCL INFO Channel 01/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-7:811253:811720 [5] NCCL INFO Channel 02/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-7:811253:811720 [5] NCCL INFO Channel 03/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-7:811253:811720 [5] NCCL INFO Channel 00/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-7:811253:811720 [5] NCCL INFO Channel 01/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-7:811253:811720 [5] NCCL INFO Channel 02/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-7:811253:811720 [5] NCCL INFO Channel 03/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:94345:94847 [3] NCCL INFO Connected all rings
+ip-26-0-160-225:94345:94847 [3] NCCL INFO Connected all trees
+ip-26-0-160-225:94345:94847 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-225:94345:94847 [3] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-160-225:94345:94847 [3] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-173-7:811251:811724 [3] NCCL INFO Connected all rings
+ip-26-0-173-7:811251:811724 [3] NCCL INFO Connected all trees
+ip-26-0-173-7:811251:811724 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-173-7:811251:811724 [3] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-173-7:811251:811724 [3] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-160-225:94345:94847 [3] NCCL INFO comm 0xc628db0 rank 0 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0x957e4f4cd9e3e0ff - Init COMPLETE
+ip-26-0-160-225:94344:94834 [2] NCCL INFO Connected all rings
+ip-26-0-160-225:94344:94834 [2] NCCL INFO Connected all trees
+ip-26-0-160-225:94344:94834 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-225:94344:94834 [2] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-160-225:94344:94834 [2] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-173-7:811251:811724 [3] NCCL INFO comm 0xa4ae480 rank 1 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0x957e4f4cd9e3e0ff - Init COMPLETE
+ip-26-0-173-7:811250:811718 [2] NCCL INFO Connected all rings
+ip-26-0-173-7:811250:811718 [2] NCCL INFO Connected all trees
+ip-26-0-173-7:811250:811718 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-173-7:811250:811718 [2] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-173-7:811250:811718 [2] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-160-225:94344:94834 [2] NCCL INFO comm 0xb44ce30 rank 0 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0x5ce82daf32c3aa8d - Init COMPLETE
+ip-26-0-173-7:811250:811718 [2] NCCL INFO comm 0xaeafb40 rank 1 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0x5ce82daf32c3aa8d - Init COMPLETE
+ip-26-0-160-225:94346:94843 [4] NCCL INFO Connected all rings
+ip-26-0-160-225:94346:94843 [4] NCCL INFO Connected all trees
+ip-26-0-160-225:94346:94843 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-225:94346:94843 [4] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-160-225:94346:94843 [4] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-173-7:811252:811723 [4] NCCL INFO Connected all rings
+ip-26-0-173-7:811252:811723 [4] NCCL INFO Connected all trees
+ip-26-0-173-7:811252:811723 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-173-7:811252:811723 [4] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-173-7:811252:811723 [4] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-160-225:94349:94848 [7] NCCL INFO Connected all rings
+ip-26-0-160-225:94349:94848 [7] NCCL INFO Connected all trees
+ip-26-0-160-225:94349:94848 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-225:94349:94848 [7] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-160-225:94349:94848 [7] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-173-7:811255:811725 [7] NCCL INFO Connected all rings
+ip-26-0-173-7:811255:811725 [7] NCCL INFO Connected all trees
+ip-26-0-173-7:811255:811725 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-173-7:811255:811725 [7] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-173-7:811255:811725 [7] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-173-7:811249:811722 [1] NCCL INFO Connected all rings
+ip-26-0-173-7:811249:811722 [1] NCCL INFO Connected all trees
+ip-26-0-173-7:811249:811722 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-173-7:811249:811722 [1] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-173-7:811249:811722 [1] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-160-225:94346:94843 [4] NCCL INFO comm 0xc48ff40 rank 0 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0xf8c2d10e47b238ab - Init COMPLETE
+ip-26-0-160-225:94343:94840 [1] NCCL INFO Connected all rings
+ip-26-0-160-225:94343:94840 [1] NCCL INFO Connected all trees
+ip-26-0-160-225:94343:94840 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-225:94343:94840 [1] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-160-225:94343:94840 [1] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-160-225:94348:94845 [6] NCCL INFO Connected all rings
+ip-26-0-160-225:94348:94845 [6] NCCL INFO Connected all trees
+ip-26-0-160-225:94348:94845 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-225:94348:94845 [6] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-160-225:94348:94845 [6] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-173-7:811248:811719 [0] NCCL INFO Connected all rings
+ip-26-0-173-7:811248:811719 [0] NCCL INFO Connected all trees
+ip-26-0-173-7:811248:811719 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-173-7:811248:811719 [0] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-173-7:811252:811723 [4] NCCL INFO comm 0xa5cb730 rank 1 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0xf8c2d10e47b238ab - Init COMPLETE
+ip-26-0-173-7:811248:811719 [0] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-160-225:94349:94848 [7] NCCL INFO comm 0xadf7a60 rank 0 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0x76201d02e3efd9f9 - Init COMPLETE
+ip-26-0-173-7:811255:811725 [7] NCCL INFO comm 0xb4fbe30 rank 1 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0x76201d02e3efd9f9 - Init COMPLETE
+ip-26-0-160-225:94347:94842 [5] NCCL INFO Connected all rings
+ip-26-0-160-225:94347:94842 [5] NCCL INFO Connected all trees
+ip-26-0-160-225:94347:94842 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-225:94347:94842 [5] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-160-225:94347:94842 [5] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-173-7:811249:811722 [1] NCCL INFO comm 0xa0b1810 rank 1 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0xfaaf189a91460e79 - Init COMPLETE
+ip-26-0-173-7:811254:811721 [6] NCCL INFO Connected all rings
+ip-26-0-173-7:811254:811721 [6] NCCL INFO Connected all trees
+ip-26-0-173-7:811254:811721 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-173-7:811254:811721 [6] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-173-7:811254:811721 [6] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-173-7:811253:811720 [5] NCCL INFO Connected all rings
+ip-26-0-173-7:811253:811720 [5] NCCL INFO Connected all trees
+ip-26-0-173-7:811253:811720 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-173-7:811253:811720 [5] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-173-7:811253:811720 [5] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-160-225:94343:94840 [1] NCCL INFO comm 0xa6f7d10 rank 0 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0xfaaf189a91460e79 - Init COMPLETE
+ip-26-0-160-225:94348:94845 [6] NCCL INFO comm 0xbbef680 rank 0 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0x4d13c7f185f0b9cd - Init COMPLETE
+ip-26-0-173-7:811248:811719 [0] NCCL INFO comm 0xbaa41f0 rank 1 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0xbff43a723081b14 - Init COMPLETE
+ip-26-0-160-225:94342:94838 [0] NCCL INFO Connected all rings
+ip-26-0-160-225:94342:94838 [0] NCCL INFO Connected all trees
+ip-26-0-160-225:94342:94838 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-225:94342:94838 [0] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-160-225:94342:94838 [0] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-173-7:811254:811721 [6] NCCL INFO comm 0xa2b7900 rank 1 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0x4d13c7f185f0b9cd - Init COMPLETE
+ip-26-0-160-225:94347:94842 [5] NCCL INFO comm 0xb526b40 rank 0 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0x5f53f5fc2baeae2 - Init COMPLETE
+ip-26-0-173-7:811253:811720 [5] NCCL INFO comm 0xaf21380 rank 1 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0x5f53f5fc2baeae2 - Init COMPLETE
+ip-26-0-160-225:94342:94838 [0] NCCL INFO comm 0xb2b6050 rank 0 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0xbff43a723081b14 - Init COMPLETE
+01/07/2025 05:27:23 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [Optimizer Building] Using LearningRateForSP as learning rate
+01/07/2025 05:27:23 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [ZeRO sharding] Size of optimizer params per rank:
+01/07/2025 05:27:23 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [ZeRO sharding] DP Rank 0 has 107M out of 856M (12.50%) params' optimizer states
+01/07/2025 05:27:23 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [ZeRO sharding] DP Rank 1 has 107M out of 856M (12.50%) params' optimizer states
+01/07/2025 05:27:23 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [ZeRO sharding] DP Rank 2 has 107M out of 856M (12.50%) params' optimizer states
+01/07/2025 05:27:23 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [ZeRO sharding] DP Rank 3 has 107M out of 856M (12.50%) params' optimizer states
+01/07/2025 05:27:23 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [ZeRO sharding] DP Rank 4 has 107M out of 856M (12.50%) params' optimizer states
+01/07/2025 05:27:23 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [ZeRO sharding] DP Rank 5 has 107M out of 856M (12.50%) params' optimizer states
+01/07/2025 05:27:23 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [ZeRO sharding] DP Rank 6 has 107M out of 856M (12.50%) params' optimizer states
+01/07/2025 05:27:23 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [ZeRO sharding] DP Rank 7 has 107M out of 856M (12.50%) params' optimizer states
+01/07/2025 05:27:23 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [Training Plan] Stage Stable Training Stage has 99 remaining training steps and has consumed 0 samples
+01/07/2025 05:27:23 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: Using dummy data generator
+01/07/2025 05:27:23 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [Training Plan] There are 1 training stages 
+01/07/2025 05:27:23 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [Stage Stable Training Stage] start from step 1 
+01/07/2025 05:27:23 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: 
+01/07/2025 05:27:23 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: [Start training] datetime: 2025-01-07 05:27:23.358729 | mbs: 2 | grad_accum: 16 | global_batch_size: 256 | sequence_length: 4096 | train_steps: 100 | start_iteration_step: 0 | consumed_train_samples: 0
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+01/07/2025 05:27:23 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: Resuming training from stage Stable Training Stage, it has trained for 0 samples and has 99 remaining train steps
+01/07/2025 05:27:23 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:  Memory usage: 5304.17MiB. Peak allocated 5408.00MiB. Peak reserved: 14820.00MiB
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+ip-26-0-160-225:94346:94870 [4] NCCL INFO Using network Libfabric
+ip-26-0-172-116:1019292:1019740 [4] NCCL INFO Using network Libfabric
+ip-26-0-160-225:94346:94870 [4] NCCL INFO comm 0x19782b00 rank 0 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0x6d6d1ffd58c5f489 - Init START
+ip-26-0-172-116:1019292:1019740 [4] NCCL INFO comm 0xc7a4460 rank 1 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0x6d6d1ffd58c5f489 - Init START
+ip-26-0-160-225:94346:94870 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-116:1019292:1019740 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+ip-26-0-160-225:94343:94872 [1] NCCL INFO Using network Libfabric
+ip-26-0-172-116:1019289:1019741 [1] NCCL INFO Using network Libfabric
+ip-26-0-160-225:94343:94872 [1] NCCL INFO comm 0x179e7f70 rank 0 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0xbe0c1a0d6ca210f3 - Init START
+ip-26-0-172-116:1019289:1019741 [1] NCCL INFO comm 0xaa43b20 rank 1 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0xbe0c1a0d6ca210f3 - Init START
+ip-26-0-160-225:94343:94872 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-116:1019289:1019741 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+ip-26-0-160-225:94348:94874 [6] NCCL INFO Using network Libfabric
+ip-26-0-172-116:1019294:1019742 [6] NCCL INFO Using network Libfabric
+ip-26-0-172-116:1019294:1019742 [6] NCCL INFO comm 0xaa47af0 rank 1 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0x7ee2d881c1c563fa - Init START
+ip-26-0-160-225:94348:94874 [6] NCCL INFO comm 0x18ee5380 rank 0 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0x7ee2d881c1c563fa - Init START
+ip-26-0-160-225:94348:94874 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-116:1019294:1019742 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+ip-26-0-160-225:94345:94877 [3] NCCL INFO Using network Libfabric
+ip-26-0-172-116:1019291:1019743 [3] NCCL INFO Using network Libfabric
+ip-26-0-160-225:94345:94877 [3] NCCL INFO comm 0x1991db00 rank 0 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0x65095958e54fbc61 - Init START
+ip-26-0-172-116:1019291:1019743 [3] NCCL INFO comm 0xc7b7d10 rank 1 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0x65095958e54fbc61 - Init START
+ip-26-0-160-225:94345:94877 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-116:1019291:1019743 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+ip-26-0-160-225:94344:94879 [2] NCCL INFO Using network Libfabric
+ip-26-0-172-116:1019290:1019744 [2] NCCL INFO Using network Libfabric
+ip-26-0-172-116:1019290:1019744 [2] NCCL INFO comm 0xc3c3b50 rank 1 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0x2810fd433447c5e8 - Init START
+ip-26-0-160-225:94344:94879 [2] NCCL INFO comm 0x1873ec80 rank 0 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0x2810fd433447c5e8 - Init START
+ip-26-0-160-225:94344:94879 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-116:1019290:1019744 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+ip-26-0-160-225:94347:94881 [5] NCCL INFO Using network Libfabric
+ip-26-0-172-116:1019293:1019745 [5] NCCL INFO Using network Libfabric
+ip-26-0-160-225:94347:94881 [5] NCCL INFO comm 0x18818600 rank 0 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0x93b96afa7c628803 - Init START
+ip-26-0-172-116:1019293:1019745 [5] NCCL INFO comm 0xc7605d0 rank 1 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0x93b96afa7c628803 - Init START
+ip-26-0-160-225:94347:94881 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-116:1019293:1019745 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+ip-26-0-160-225:94342:94883 [0] NCCL INFO Using network Libfabric
+ip-26-0-172-116:1019288:1019746 [0] NCCL INFO Using network Libfabric
+ip-26-0-172-116:1019288:1019746 [0] NCCL INFO comm 0xb0737f0 rank 1 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0xa7249dbeffa3c890 - Init START
+ip-26-0-160-225:94342:94883 [0] NCCL INFO comm 0x185ab2b0 rank 0 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0xa7249dbeffa3c890 - Init START
+ip-26-0-160-225:94342:94883 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/349
+ip-26-0-172-116:1019288:1019746 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-116:1019292:1019740 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-172-116:1019289:1019741 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-160-225:94346:94870 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-160-225:94346:94870 [4] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-160-225:94346:94870 [4] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-160-225:94346:94870 [4] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-160-225:94346:94870 [4] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-160-225:94346:94870 [4] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-160-225:94346:94870 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-116:1019292:1019740 [4] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-172-116:1019292:1019740 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-116:1019294:1019742 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+ip-26-0-160-225:94349:94887 [7] NCCL INFO Using network Libfabric
+ip-26-0-172-116:1019295:1019750 [7] NCCL INFO Using network Libfabric
+ip-26-0-160-225:94349:94887 [7] NCCL INFO comm 0x180e9140 rank 0 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0xfcf5b612f18dc520 - Init START
+ip-26-0-172-116:1019295:1019750 [7] NCCL INFO comm 0xbd2b600 rank 1 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0xfcf5b612f18dc520 - Init START
+ip-26-0-160-225:94349:94887 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-116:1019295:1019750 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-160-225:94343:94872 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-160-225:94343:94872 [1] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-160-225:94343:94872 [1] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-160-225:94343:94872 [1] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-160-225:94343:94872 [1] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-160-225:94343:94872 [1] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-160-225:94343:94872 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-116:1019289:1019741 [1] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-172-116:1019289:1019741 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-225:94346:94870 [4] NCCL INFO Channel 00/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:94346:94870 [4] NCCL INFO Channel 01/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:94346:94870 [4] NCCL INFO Channel 02/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:94346:94870 [4] NCCL INFO Channel 03/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:94346:94870 [4] NCCL INFO Channel 00/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:94346:94870 [4] NCCL INFO Channel 01/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:94346:94870 [4] NCCL INFO Channel 02/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:94346:94870 [4] NCCL INFO Channel 03/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019292:1019740 [4] NCCL INFO Channel 00/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019292:1019740 [4] NCCL INFO Channel 01/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019292:1019740 [4] NCCL INFO Channel 02/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019292:1019740 [4] NCCL INFO Channel 03/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019292:1019740 [4] NCCL INFO Channel 00/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019292:1019740 [4] NCCL INFO Channel 01/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019292:1019740 [4] NCCL INFO Channel 02/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019292:1019740 [4] NCCL INFO Channel 03/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-160-225:94343:94872 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:94343:94872 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:94343:94872 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:94343:94872 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:94343:94872 [1] NCCL INFO Channel 00/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:94343:94872 [1] NCCL INFO Channel 01/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:94343:94872 [1] NCCL INFO Channel 02/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:94343:94872 [1] NCCL INFO Channel 03/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019289:1019741 [1] NCCL INFO Channel 00/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019289:1019741 [1] NCCL INFO Channel 01/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019289:1019741 [1] NCCL INFO Channel 02/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019289:1019741 [1] NCCL INFO Channel 03/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019289:1019741 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019289:1019741 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019289:1019741 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019289:1019741 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-160-225:94348:94874 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-160-225:94348:94874 [6] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-160-225:94348:94874 [6] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-160-225:94348:94874 [6] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-160-225:94348:94874 [6] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-160-225:94348:94874 [6] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-160-225:94348:94874 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-116:1019294:1019742 [6] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-172-116:1019294:1019742 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-116:1019291:1019743 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-160-225:94345:94877 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-160-225:94345:94877 [3] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-160-225:94345:94877 [3] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-160-225:94345:94877 [3] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-160-225:94345:94877 [3] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-160-225:94345:94877 [3] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-160-225:94345:94877 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-116:1019291:1019743 [3] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-172-116:1019291:1019743 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-116:1019294:1019742 [6] NCCL INFO Channel 00/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019294:1019742 [6] NCCL INFO Channel 01/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019294:1019742 [6] NCCL INFO Channel 02/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019294:1019742 [6] NCCL INFO Channel 03/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019294:1019742 [6] NCCL INFO Channel 00/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019294:1019742 [6] NCCL INFO Channel 01/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019294:1019742 [6] NCCL INFO Channel 02/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019294:1019742 [6] NCCL INFO Channel 03/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:94348:94874 [6] NCCL INFO Channel 00/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:94348:94874 [6] NCCL INFO Channel 01/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:94348:94874 [6] NCCL INFO Channel 02/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:94348:94874 [6] NCCL INFO Channel 03/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:94348:94874 [6] NCCL INFO Channel 00/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:94348:94874 [6] NCCL INFO Channel 01/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:94348:94874 [6] NCCL INFO Channel 02/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-160-225:94348:94874 [6] NCCL INFO Channel 03/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019290:1019744 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-160-225:94344:94879 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-160-225:94344:94879 [2] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-160-225:94344:94879 [2] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-160-225:94344:94879 [2] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-160-225:94344:94879 [2] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-160-225:94344:94879 [2] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-160-225:94344:94879 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-116:1019290:1019744 [2] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-172-116:1019290:1019744 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-225:94347:94881 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-172-116:1019288:1019746 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-172-116:1019293:1019745 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-172-116:1019293:1019745 [5] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-172-116:1019293:1019745 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-225:94347:94881 [5] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-160-225:94347:94881 [5] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-160-225:94347:94881 [5] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-160-225:94347:94881 [5] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-160-225:94347:94881 [5] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-160-225:94347:94881 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-116:1019295:1019750 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-160-225:94345:94877 [3] NCCL INFO Channel 00/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:94345:94877 [3] NCCL INFO Channel 01/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:94345:94877 [3] NCCL INFO Channel 02/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:94345:94877 [3] NCCL INFO Channel 03/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:94345:94877 [3] NCCL INFO Channel 00/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:94345:94877 [3] NCCL INFO Channel 01/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:94345:94877 [3] NCCL INFO Channel 02/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:94345:94877 [3] NCCL INFO Channel 03/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019291:1019743 [3] NCCL INFO Channel 00/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019291:1019743 [3] NCCL INFO Channel 01/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019291:1019743 [3] NCCL INFO Channel 02/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019291:1019743 [3] NCCL INFO Channel 03/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019291:1019743 [3] NCCL INFO Channel 00/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019291:1019743 [3] NCCL INFO Channel 01/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019291:1019743 [3] NCCL INFO Channel 02/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019291:1019743 [3] NCCL INFO Channel 03/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-160-225:94342:94883 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-160-225:94342:94883 [0] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-160-225:94342:94883 [0] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-160-225:94342:94883 [0] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-160-225:94342:94883 [0] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-160-225:94342:94883 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-160-225:94342:94883 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-116:1019288:1019746 [0] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-172-116:1019288:1019746 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-225:94344:94879 [2] NCCL INFO Channel 00/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:94344:94879 [2] NCCL INFO Channel 01/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:94344:94879 [2] NCCL INFO Channel 02/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:94344:94879 [2] NCCL INFO Channel 03/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:94344:94879 [2] NCCL INFO Channel 00/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019290:1019744 [2] NCCL INFO Channel 00/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:94344:94879 [2] NCCL INFO Channel 01/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019290:1019744 [2] NCCL INFO Channel 01/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:94344:94879 [2] NCCL INFO Channel 02/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:94344:94879 [2] NCCL INFO Channel 03/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019290:1019744 [2] NCCL INFO Channel 02/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019290:1019744 [2] NCCL INFO Channel 03/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019290:1019744 [2] NCCL INFO Channel 00/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019290:1019744 [2] NCCL INFO Channel 01/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019290:1019744 [2] NCCL INFO Channel 02/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019290:1019744 [2] NCCL INFO Channel 03/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-160-225:94349:94887 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-160-225:94349:94887 [7] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-160-225:94349:94887 [7] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-160-225:94349:94887 [7] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-160-225:94349:94887 [7] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-160-225:94349:94887 [7] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-160-225:94349:94887 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-116:1019295:1019750 [7] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-172-116:1019295:1019750 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-160-225:94347:94881 [5] NCCL INFO Channel 00/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:94347:94881 [5] NCCL INFO Channel 01/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:94347:94881 [5] NCCL INFO Channel 02/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:94347:94881 [5] NCCL INFO Channel 03/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:94347:94881 [5] NCCL INFO Channel 00/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:94347:94881 [5] NCCL INFO Channel 01/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:94347:94881 [5] NCCL INFO Channel 02/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019293:1019745 [5] NCCL INFO Channel 00/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-160-225:94347:94881 [5] NCCL INFO Channel 03/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019293:1019745 [5] NCCL INFO Channel 01/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019293:1019745 [5] NCCL INFO Channel 02/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019293:1019745 [5] NCCL INFO Channel 03/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019293:1019745 [5] NCCL INFO Channel 00/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019293:1019745 [5] NCCL INFO Channel 01/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019293:1019745 [5] NCCL INFO Channel 02/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019293:1019745 [5] NCCL INFO Channel 03/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019288:1019746 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019288:1019746 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019288:1019746 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019288:1019746 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019288:1019746 [0] NCCL INFO Channel 00/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019288:1019746 [0] NCCL INFO Channel 01/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019288:1019746 [0] NCCL INFO Channel 02/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:94342:94883 [0] NCCL INFO Channel 00/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019288:1019746 [0] NCCL INFO Channel 03/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:94342:94883 [0] NCCL INFO Channel 01/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:94342:94883 [0] NCCL INFO Channel 02/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:94342:94883 [0] NCCL INFO Channel 03/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:94342:94883 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:94342:94883 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:94342:94883 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:94342:94883 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-160-225:94349:94887 [7] NCCL INFO Channel 00/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:94349:94887 [7] NCCL INFO Channel 01/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:94349:94887 [7] NCCL INFO Channel 02/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:94349:94887 [7] NCCL INFO Channel 03/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:94349:94887 [7] NCCL INFO Channel 00/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:94349:94887 [7] NCCL INFO Channel 01/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:94349:94887 [7] NCCL INFO Channel 02/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:94349:94887 [7] NCCL INFO Channel 03/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019295:1019750 [7] NCCL INFO Channel 00/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019295:1019750 [7] NCCL INFO Channel 01/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019295:1019750 [7] NCCL INFO Channel 02/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019295:1019750 [7] NCCL INFO Channel 03/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019295:1019750 [7] NCCL INFO Channel 00/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019295:1019750 [7] NCCL INFO Channel 01/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019295:1019750 [7] NCCL INFO Channel 02/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019295:1019750 [7] NCCL INFO Channel 03/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-160-225:94343:94872 [1] NCCL INFO Connected all rings
+ip-26-0-160-225:94343:94872 [1] NCCL INFO Connected all trees
+ip-26-0-160-225:94343:94872 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-225:94343:94872 [1] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-160-225:94343:94872 [1] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-160-225:94346:94870 [4] NCCL INFO Connected all rings
+ip-26-0-160-225:94346:94870 [4] NCCL INFO Connected all trees
+ip-26-0-160-225:94346:94870 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-225:94346:94870 [4] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-160-225:94346:94870 [4] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-160-225:94348:94874 [6] NCCL INFO Connected all rings
+ip-26-0-160-225:94348:94874 [6] NCCL INFO Connected all trees
+ip-26-0-160-225:94348:94874 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-225:94348:94874 [6] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-160-225:94348:94874 [6] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-116:1019294:1019742 [6] NCCL INFO Connected all rings
+ip-26-0-172-116:1019294:1019742 [6] NCCL INFO Connected all trees
+ip-26-0-172-116:1019294:1019742 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-116:1019294:1019742 [6] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-172-116:1019294:1019742 [6] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-160-225:94343:94872 [1] NCCL INFO comm 0x179e7f70 rank 0 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0xbe0c1a0d6ca210f3 - Init COMPLETE
+ip-26-0-160-225:94343:94902 [1] NCCL INFO Channel 02/1 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA/Shared
+ip-26-0-172-116:1019292:1019740 [4] NCCL INFO Connected all rings
+ip-26-0-172-116:1019292:1019740 [4] NCCL INFO Connected all trees
+ip-26-0-172-116:1019292:1019740 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-116:1019292:1019740 [4] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-160-225:94343:94902 [1] NCCL INFO Channel 03/1 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA/Shared
+ip-26-0-172-116:1019292:1019740 [4] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-116:1019294:1019742 [6] NCCL INFO comm 0xaa47af0 rank 1 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0x7ee2d881c1c563fa - Init COMPLETE
+ip-26-0-172-116:1019294:1019765 [6] NCCL INFO Channel 02/1 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA/Shared
+ip-26-0-172-116:1019294:1019765 [6] NCCL INFO Channel 03/1 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA/Shared
+ip-26-0-160-225:94348:94874 [6] NCCL INFO comm 0x18ee5380 rank 0 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0x7ee2d881c1c563fa - Init COMPLETE
+ip-26-0-160-225:94346:94870 [4] NCCL INFO comm 0x19782b00 rank 0 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0x6d6d1ffd58c5f489 - Init COMPLETE
+ip-26-0-160-225:94348:94903 [6] NCCL INFO Channel 02/1 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA/Shared
+ip-26-0-160-225:94348:94903 [6] NCCL INFO Channel 03/1 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA/Shared
+ip-26-0-160-225:94346:94904 [4] NCCL INFO Channel 02/1 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA/Shared
+ip-26-0-160-225:94346:94904 [4] NCCL INFO Channel 03/1 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA/Shared
+ip-26-0-172-116:1019289:1019741 [1] NCCL INFO Connected all rings
+ip-26-0-172-116:1019289:1019741 [1] NCCL INFO Connected all trees
+ip-26-0-172-116:1019289:1019741 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-116:1019289:1019741 [1] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-172-116:1019289:1019741 [1] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-116:1019292:1019740 [4] NCCL INFO comm 0xc7a4460 rank 1 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0x6d6d1ffd58c5f489 - Init COMPLETE
+ip-26-0-172-116:1019292:1019766 [4] NCCL INFO Channel 02/1 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA/Shared
+ip-26-0-172-116:1019292:1019766 [4] NCCL INFO Channel 03/1 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA/Shared
+ip-26-0-172-116:1019289:1019741 [1] NCCL INFO comm 0xaa43b20 rank 1 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0xbe0c1a0d6ca210f3 - Init COMPLETE
+ip-26-0-172-116:1019289:1019767 [1] NCCL INFO Channel 02/1 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA/Shared
+ip-26-0-172-116:1019289:1019767 [1] NCCL INFO Channel 03/1 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA/Shared
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+ip-26-0-160-225:94345:94877 [3] NCCL INFO Connected all rings
+ip-26-0-160-225:94345:94877 [3] NCCL INFO Connected all trees
+ip-26-0-160-225:94345:94877 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-225:94345:94877 [3] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-160-225:94345:94877 [3] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-116:1019291:1019743 [3] NCCL INFO Connected all rings
+ip-26-0-172-116:1019291:1019743 [3] NCCL INFO Connected all trees
+ip-26-0-172-116:1019291:1019743 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-116:1019291:1019743 [3] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-172-116:1019291:1019743 [3] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-160-225:94344:94879 [2] NCCL INFO Connected all rings
+ip-26-0-160-225:94344:94879 [2] NCCL INFO Connected all trees
+ip-26-0-160-225:94344:94879 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-225:94344:94879 [2] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-160-225:94344:94879 [2] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-160-225:94345:94877 [3] NCCL INFO comm 0x1991db00 rank 0 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0x65095958e54fbc61 - Init COMPLETE
+ip-26-0-160-225:94345:94905 [3] NCCL INFO Channel 02/1 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA/Shared
+ip-26-0-160-225:94345:94905 [3] NCCL INFO Channel 03/1 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA/Shared
+ip-26-0-160-225:94347:94881 [5] NCCL INFO Connected all rings
+ip-26-0-160-225:94347:94881 [5] NCCL INFO Connected all trees
+ip-26-0-160-225:94347:94881 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-225:94347:94881 [5] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-160-225:94347:94881 [5] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-116:1019291:1019743 [3] NCCL INFO comm 0xc7b7d10 rank 1 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0x65095958e54fbc61 - Init COMPLETE
+ip-26-0-160-225:94344:94879 [2] NCCL INFO comm 0x1873ec80 rank 0 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0x2810fd433447c5e8 - Init COMPLETE
+ip-26-0-172-116:1019291:1019768 [3] NCCL INFO Channel 02/1 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA/Shared
+ip-26-0-172-116:1019291:1019768 [3] NCCL INFO Channel 03/1 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA/Shared
+ip-26-0-160-225:94344:94906 [2] NCCL INFO Channel 02/1 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA/Shared
+ip-26-0-160-225:94344:94906 [2] NCCL INFO Channel 03/1 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA/Shared
+ip-26-0-172-116:1019290:1019744 [2] NCCL INFO Connected all rings
+ip-26-0-172-116:1019290:1019744 [2] NCCL INFO Connected all trees
+ip-26-0-172-116:1019290:1019744 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-116:1019290:1019744 [2] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-172-116:1019290:1019744 [2] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-160-225:94347:94881 [5] NCCL INFO comm 0x18818600 rank 0 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0x93b96afa7c628803 - Init COMPLETE
+ip-26-0-160-225:94347:94907 [5] NCCL INFO Channel 02/1 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA/Shared
+ip-26-0-160-225:94347:94907 [5] NCCL INFO Channel 03/1 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA/Shared
+ip-26-0-172-116:1019293:1019745 [5] NCCL INFO Connected all rings
+ip-26-0-172-116:1019293:1019745 [5] NCCL INFO Connected all trees
+ip-26-0-172-116:1019293:1019745 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-116:1019293:1019745 [5] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-172-116:1019293:1019745 [5] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-116:1019290:1019744 [2] NCCL INFO comm 0xc3c3b50 rank 1 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0x2810fd433447c5e8 - Init COMPLETE
+ip-26-0-172-116:1019290:1019769 [2] NCCL INFO Channel 02/1 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA/Shared
+ip-26-0-172-116:1019290:1019769 [2] NCCL INFO Channel 03/1 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA/Shared
+ip-26-0-172-116:1019293:1019745 [5] NCCL INFO comm 0xc7605d0 rank 1 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0x93b96afa7c628803 - Init COMPLETE
+ip-26-0-172-116:1019293:1019770 [5] NCCL INFO Channel 02/1 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA/Shared
+ip-26-0-172-116:1019293:1019770 [5] NCCL INFO Channel 03/1 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA/Shared
+ip-26-0-172-116:1019288:1019746 [0] NCCL INFO Connected all rings
+ip-26-0-172-116:1019288:1019746 [0] NCCL INFO Connected all trees
+ip-26-0-172-116:1019288:1019746 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-116:1019288:1019746 [0] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-172-116:1019288:1019746 [0] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-160-225:94342:94883 [0] NCCL INFO Connected all rings
+ip-26-0-160-225:94342:94883 [0] NCCL INFO Connected all trees
+ip-26-0-160-225:94342:94883 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-225:94342:94883 [0] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-160-225:94342:94883 [0] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-116:1019295:1019750 [7] NCCL INFO Connected all rings
+ip-26-0-172-116:1019295:1019750 [7] NCCL INFO Connected all trees
+ip-26-0-172-116:1019295:1019750 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-116:1019295:1019750 [7] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-172-116:1019295:1019750 [7] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-116:1019288:1019746 [0] NCCL INFO comm 0xb0737f0 rank 1 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0xa7249dbeffa3c890 - Init COMPLETE
+ip-26-0-160-225:94342:94883 [0] NCCL INFO comm 0x185ab2b0 rank 0 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0xa7249dbeffa3c890 - Init COMPLETE
+ip-26-0-172-116:1019288:1019771 [0] NCCL INFO Channel 02/1 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA/Shared
+ip-26-0-172-116:1019288:1019771 [0] NCCL INFO Channel 03/1 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA/Shared
+ip-26-0-160-225:94342:94908 [0] NCCL INFO Channel 02/1 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA/Shared
+ip-26-0-160-225:94342:94908 [0] NCCL INFO Channel 03/1 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA/Shared
+ip-26-0-172-116:1019295:1019750 [7] NCCL INFO comm 0xbd2b600 rank 1 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0xfcf5b612f18dc520 - Init COMPLETE
+ip-26-0-172-116:1019295:1019772 [7] NCCL INFO Channel 02/1 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA/Shared
+ip-26-0-172-116:1019295:1019772 [7] NCCL INFO Channel 03/1 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA/Shared
+ip-26-0-160-225:94349:94887 [7] NCCL INFO Connected all rings
+ip-26-0-160-225:94349:94887 [7] NCCL INFO Connected all trees
+ip-26-0-160-225:94349:94887 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-160-225:94349:94887 [7] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-160-225:94349:94887 [7] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-160-225:94349:94887 [7] NCCL INFO comm 0x180e9140 rank 0 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0xfcf5b612f18dc520 - Init COMPLETE
+ip-26-0-160-225:94349:94909 [7] NCCL INFO Channel 02/1 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA/Shared
+ip-26-0-160-225:94349:94909 [7] NCCL INFO Channel 03/1 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA/Shared
+ip-26-0-160-225:94348:94919 [6] NCCL INFO Channel 02/1 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA/Shared
+ip-26-0-160-225:94348:94919 [6] NCCL INFO Channel 03/1 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA/Shared
+ip-26-0-160-225:94346:94928 [4] NCCL INFO Channel 02/1 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA/Shared
+ip-26-0-160-225:94346:94928 [4] NCCL INFO Channel 03/1 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA/Shared
+ip-26-0-160-225:94343:94937 [1] NCCL INFO Channel 02/1 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA/Shared
+ip-26-0-160-225:94343:94937 [1] NCCL INFO Channel 03/1 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA/Shared
+ip-26-0-160-225:94345:94946 [3] NCCL INFO Channel 02/1 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA/Shared
+ip-26-0-160-225:94345:94946 [3] NCCL INFO Channel 03/1 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA/Shared
+ip-26-0-160-225:94347:94955 [5] NCCL INFO Channel 02/1 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA/Shared
+ip-26-0-160-225:94347:94955 [5] NCCL INFO Channel 03/1 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA/Shared
+ip-26-0-160-225:94344:94964 [2] NCCL INFO Channel 02/1 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA/Shared
+ip-26-0-160-225:94344:94964 [2] NCCL INFO Channel 03/1 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA/Shared
+ip-26-0-160-225:94342:94973 [0] NCCL INFO Channel 02/1 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA/Shared
+ip-26-0-160-225:94342:94973 [0] NCCL INFO Channel 03/1 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA/Shared
+ip-26-0-160-225:94349:94982 [7] NCCL INFO Channel 02/1 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA/Shared
+ip-26-0-160-225:94349:94982 [7] NCCL INFO Channel 03/1 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA/Shared
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+ip-26-0-172-116:1019294:1019776 [6] NCCL INFO Using network Libfabric
+ip-26-0-172-142:880685:881134 [6] NCCL INFO Using network Libfabric
+ip-26-0-172-142:880685:881134 [6] NCCL INFO comm 0xad81100 rank 1 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0x525c205cc8134fe3 - Init START
+ip-26-0-172-116:1019294:1019776 [6] NCCL INFO comm 0x1364f540 rank 0 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0x525c205cc8134fe3 - Init START
+ip-26-0-172-116:1019294:1019776 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-142:880685:881134 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+ip-26-0-172-116:1019289:1019778 [1] NCCL INFO Using network Libfabric
+ip-26-0-172-142:880680:881135 [1] NCCL INFO Using network Libfabric
+ip-26-0-172-142:880680:881135 [1] NCCL INFO comm 0xb34b180 rank 1 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0x6b45b633dd439f28 - Init START
+ip-26-0-172-116:1019289:1019778 [1] NCCL INFO comm 0x1364b770 rank 0 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0x6b45b633dd439f28 - Init START
+ip-26-0-172-142:880680:881135 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-116:1019289:1019778 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+ip-26-0-172-116:1019292:1019780 [4] NCCL INFO Using network Libfabric
+ip-26-0-172-142:880683:881136 [4] NCCL INFO Using network Libfabric
+ip-26-0-172-142:880683:881136 [4] NCCL INFO comm 0xc0a1680 rank 1 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0xa6165201324a543d - Init START
+ip-26-0-172-116:1019292:1019780 [4] NCCL INFO comm 0x153adb70 rank 0 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0xa6165201324a543d - Init START
+ip-26-0-172-142:880683:881136 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-116:1019292:1019780 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-142:880685:881134 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-172-116:1019294:1019776 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-172-116:1019294:1019776 [6] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-172-116:1019294:1019776 [6] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-172-142:880685:881134 [6] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-172-142:880685:881134 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-116:1019294:1019776 [6] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-172-116:1019294:1019776 [6] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-172-116:1019294:1019776 [6] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-172-116:1019294:1019776 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-142:880680:881135 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-172-116:1019289:1019778 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-172-116:1019289:1019778 [1] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-172-116:1019289:1019778 [1] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-172-116:1019289:1019778 [1] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-172-116:1019289:1019778 [1] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-172-116:1019289:1019778 [1] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-172-116:1019289:1019778 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-142:880680:881135 [1] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-172-142:880680:881135 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-142:880683:881136 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+ip-26-0-172-116:1019291:1019787 [3] NCCL INFO Using network Libfabric
+ip-26-0-172-116:1019290:1019788 [2] NCCL INFO Using network Libfabric
+ip-26-0-172-142:880682:881142 [3] NCCL INFO Using network Libfabric
+ip-26-0-172-142:880681:881143 [2] NCCL INFO Using network Libfabric
+ip-26-0-172-142:880681:881143 [2] NCCL INFO comm 0xb481ae0 rank 1 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0x7356f1a08d3799bc - Init START
+ip-26-0-172-116:1019290:1019788 [2] NCCL INFO comm 0x1503aa00 rank 0 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0x7356f1a08d3799bc - Init START
+ip-26-0-172-142:880682:881142 [3] NCCL INFO comm 0xb91c700 rank 1 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0xc898011f09db575d - Init START
+ip-26-0-172-116:1019291:1019787 [3] NCCL INFO comm 0x153c2530 rank 0 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0xc898011f09db575d - Init START
+ip-26-0-172-142:880681:881143 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-116:1019290:1019788 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-142:880682:881142 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-116:1019291:1019787 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-116:1019294:1019776 [6] NCCL INFO Channel 00/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019294:1019776 [6] NCCL INFO Channel 01/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019294:1019776 [6] NCCL INFO Channel 02/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019294:1019776 [6] NCCL INFO Channel 03/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019294:1019776 [6] NCCL INFO Channel 00/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019294:1019776 [6] NCCL INFO Channel 01/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019294:1019776 [6] NCCL INFO Channel 02/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019294:1019776 [6] NCCL INFO Channel 03/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880685:881134 [6] NCCL INFO Channel 00/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880685:881134 [6] NCCL INFO Channel 01/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880685:881134 [6] NCCL INFO Channel 02/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880685:881134 [6] NCCL INFO Channel 03/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880685:881134 [6] NCCL INFO Channel 00/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880685:881134 [6] NCCL INFO Channel 01/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880685:881134 [6] NCCL INFO Channel 02/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880685:881134 [6] NCCL INFO Channel 03/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-116:1019292:1019780 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-172-116:1019292:1019780 [4] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-172-116:1019292:1019780 [4] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-172-116:1019292:1019780 [4] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-172-116:1019292:1019780 [4] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-172-142:880683:881136 [4] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-172-142:880683:881136 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-116:1019292:1019780 [4] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-172-116:1019292:1019780 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-142:880680:881135 [1] NCCL INFO Channel 00/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880680:881135 [1] NCCL INFO Channel 01/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880680:881135 [1] NCCL INFO Channel 02/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880680:881135 [1] NCCL INFO Channel 03/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880680:881135 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880680:881135 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880680:881135 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880680:881135 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019289:1019778 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019289:1019778 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019289:1019778 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019289:1019778 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019289:1019778 [1] NCCL INFO Channel 00/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019289:1019778 [1] NCCL INFO Channel 01/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019289:1019778 [1] NCCL INFO Channel 02/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-116:1019289:1019778 [1] NCCL INFO Channel 03/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+ip-26-0-172-116:1019293:1019792 [5] NCCL INFO Using network Libfabric
+ip-26-0-172-142:880684:881146 [5] NCCL INFO Using network Libfabric
+ip-26-0-172-142:880684:881146 [5] NCCL INFO comm 0xb05e6d0 rank 1 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0xf33f0cb6a1d78c81 - Init START
+ip-26-0-172-116:1019293:1019792 [5] NCCL INFO comm 0x1544fc70 rank 0 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0xf33f0cb6a1d78c81 - Init START
+ip-26-0-172-116:1019293:1019792 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-142:880684:881146 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-142:880683:881136 [4] NCCL INFO Channel 00/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880683:881136 [4] NCCL INFO Channel 01/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880683:881136 [4] NCCL INFO Channel 02/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880683:881136 [4] NCCL INFO Channel 03/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880683:881136 [4] NCCL INFO Channel 00/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880683:881136 [4] NCCL INFO Channel 01/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880683:881136 [4] NCCL INFO Channel 02/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880683:881136 [4] NCCL INFO Channel 03/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019292:1019780 [4] NCCL INFO Channel 00/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019292:1019780 [4] NCCL INFO Channel 01/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019292:1019780 [4] NCCL INFO Channel 02/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019292:1019780 [4] NCCL INFO Channel 03/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019292:1019780 [4] NCCL INFO Channel 00/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019292:1019780 [4] NCCL INFO Channel 01/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019292:1019780 [4] NCCL INFO Channel 02/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-116:1019292:1019780 [4] NCCL INFO Channel 03/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880681:881143 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-172-142:880682:881142 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-172-116:1019291:1019787 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-172-116:1019291:1019787 [3] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-172-116:1019291:1019787 [3] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-172-116:1019291:1019787 [3] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-172-116:1019291:1019787 [3] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-172-116:1019291:1019787 [3] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-172-116:1019291:1019787 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-142:880682:881142 [3] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-172-142:880682:881142 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-116:1019290:1019788 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-172-116:1019290:1019788 [2] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-172-116:1019290:1019788 [2] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-172-116:1019290:1019788 [2] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-172-116:1019290:1019788 [2] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-172-116:1019290:1019788 [2] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-172-116:1019290:1019788 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-142:880681:881143 [2] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-172-142:880681:881143 [2] NCCL INFO P2P Chunksize set to 131072
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+ip-26-0-172-142:880684:881146 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-172-116:1019288:1019799 [0] NCCL INFO Using network Libfabric
+ip-26-0-172-142:880679:881151 [0] NCCL INFO Using network Libfabric
+ip-26-0-172-142:880679:881151 [0] NCCL INFO comm 0xc595560 rank 1 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0x59152c3b439684bf - Init START
+ip-26-0-172-116:1019288:1019799 [0] NCCL INFO comm 0x13c79070 rank 0 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0x59152c3b439684bf - Init START
+ip-26-0-172-142:880679:881151 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-116:1019288:1019799 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+ip-26-0-172-116:1019295:1019801 [7] NCCL INFO Using network Libfabric
+ip-26-0-172-142:880686:881152 [7] NCCL INFO Using network Libfabric
+ip-26-0-172-142:880686:881152 [7] NCCL INFO comm 0xbab76c0 rank 1 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0xb6ad2873cfe08d3b - Init START
+ip-26-0-172-116:1019295:1019801 [7] NCCL INFO comm 0x14a1a1f0 rank 0 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0xb6ad2873cfe08d3b - Init START
+ip-26-0-172-116:1019295:1019801 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-142:880686:881152 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-142:880682:881142 [3] NCCL INFO Channel 00/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880682:881142 [3] NCCL INFO Channel 01/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880682:881142 [3] NCCL INFO Channel 02/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880682:881142 [3] NCCL INFO Channel 03/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880682:881142 [3] NCCL INFO Channel 00/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880682:881142 [3] NCCL INFO Channel 01/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880682:881142 [3] NCCL INFO Channel 02/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880682:881142 [3] NCCL INFO Channel 03/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019291:1019787 [3] NCCL INFO Channel 00/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019291:1019787 [3] NCCL INFO Channel 01/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019291:1019787 [3] NCCL INFO Channel 02/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019291:1019787 [3] NCCL INFO Channel 03/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019291:1019787 [3] NCCL INFO Channel 00/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019291:1019787 [3] NCCL INFO Channel 01/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019291:1019787 [3] NCCL INFO Channel 02/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019291:1019787 [3] NCCL INFO Channel 03/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-116:1019293:1019792 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-172-116:1019293:1019792 [5] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-172-116:1019293:1019792 [5] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-172-116:1019293:1019792 [5] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-172-116:1019293:1019792 [5] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-172-116:1019293:1019792 [5] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-172-116:1019293:1019792 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-142:880684:881146 [5] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-172-142:880684:881146 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-142:880681:881143 [2] NCCL INFO Channel 00/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880681:881143 [2] NCCL INFO Channel 01/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880681:881143 [2] NCCL INFO Channel 02/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880681:881143 [2] NCCL INFO Channel 03/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880681:881143 [2] NCCL INFO Channel 00/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880681:881143 [2] NCCL INFO Channel 01/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880681:881143 [2] NCCL INFO Channel 02/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880681:881143 [2] NCCL INFO Channel 03/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019290:1019788 [2] NCCL INFO Channel 00/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019290:1019788 [2] NCCL INFO Channel 01/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019290:1019788 [2] NCCL INFO Channel 02/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019290:1019788 [2] NCCL INFO Channel 03/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019290:1019788 [2] NCCL INFO Channel 00/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019290:1019788 [2] NCCL INFO Channel 01/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019290:1019788 [2] NCCL INFO Channel 02/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-116:1019290:1019788 [2] NCCL INFO Channel 03/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880684:881146 [5] NCCL INFO Channel 00/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880684:881146 [5] NCCL INFO Channel 01/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880684:881146 [5] NCCL INFO Channel 02/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019293:1019792 [5] NCCL INFO Channel 00/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880684:881146 [5] NCCL INFO Channel 03/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019293:1019792 [5] NCCL INFO Channel 01/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880684:881146 [5] NCCL INFO Channel 00/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019293:1019792 [5] NCCL INFO Channel 02/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880684:881146 [5] NCCL INFO Channel 01/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019293:1019792 [5] NCCL INFO Channel 03/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880684:881146 [5] NCCL INFO Channel 02/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019293:1019792 [5] NCCL INFO Channel 00/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880684:881146 [5] NCCL INFO Channel 03/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019293:1019792 [5] NCCL INFO Channel 01/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019293:1019792 [5] NCCL INFO Channel 02/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-116:1019293:1019792 [5] NCCL INFO Channel 03/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880679:881151 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-172-116:1019288:1019799 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-172-142:880679:881151 [0] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-172-142:880679:881151 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-116:1019288:1019799 [0] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-172-116:1019288:1019799 [0] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-172-116:1019288:1019799 [0] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-172-116:1019288:1019799 [0] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-172-116:1019288:1019799 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-172-116:1019288:1019799 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-142:880686:881152 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-172-116:1019295:1019801 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-172-116:1019295:1019801 [7] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-172-116:1019295:1019801 [7] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-172-116:1019295:1019801 [7] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-172-116:1019295:1019801 [7] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-172-116:1019295:1019801 [7] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-172-116:1019295:1019801 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-142:880686:881152 [7] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-172-142:880686:881152 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-142:880679:881151 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880679:881151 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880679:881151 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880679:881151 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880679:881151 [0] NCCL INFO Channel 00/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880679:881151 [0] NCCL INFO Channel 01/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880679:881151 [0] NCCL INFO Channel 02/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880679:881151 [0] NCCL INFO Channel 03/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019288:1019799 [0] NCCL INFO Channel 00/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019288:1019799 [0] NCCL INFO Channel 01/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019288:1019799 [0] NCCL INFO Channel 02/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019288:1019799 [0] NCCL INFO Channel 03/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019288:1019799 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019288:1019799 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019288:1019799 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-116:1019288:1019799 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880685:881134 [6] NCCL INFO Connected all rings
+ip-26-0-172-142:880685:881134 [6] NCCL INFO Connected all trees
+ip-26-0-172-142:880685:881134 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-142:880685:881134 [6] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-172-142:880685:881134 [6] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-142:880686:881152 [7] NCCL INFO Channel 00/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880686:881152 [7] NCCL INFO Channel 01/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880686:881152 [7] NCCL INFO Channel 02/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880686:881152 [7] NCCL INFO Channel 03/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880686:881152 [7] NCCL INFO Channel 00/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880686:881152 [7] NCCL INFO Channel 01/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880686:881152 [7] NCCL INFO Channel 02/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880686:881152 [7] NCCL INFO Channel 03/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019295:1019801 [7] NCCL INFO Channel 00/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019295:1019801 [7] NCCL INFO Channel 01/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019295:1019801 [7] NCCL INFO Channel 02/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019295:1019801 [7] NCCL INFO Channel 03/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019295:1019801 [7] NCCL INFO Channel 00/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019295:1019801 [7] NCCL INFO Channel 01/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880685:881134 [6] NCCL INFO comm 0xad81100 rank 1 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0x525c205cc8134fe3 - Init COMPLETE
+ip-26-0-172-116:1019295:1019801 [7] NCCL INFO Channel 02/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-116:1019295:1019801 [7] NCCL INFO Channel 03/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880685:881159 [6] NCCL INFO Channel 02/1 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA/Shared
+ip-26-0-172-142:880685:881159 [6] NCCL INFO Channel 03/1 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA/Shared
+ip-26-0-172-116:1019294:1019776 [6] NCCL INFO Connected all rings
+ip-26-0-172-116:1019294:1019776 [6] NCCL INFO Connected all trees
+ip-26-0-172-116:1019294:1019776 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-116:1019294:1019776 [6] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-172-116:1019294:1019776 [6] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-142:880683:881136 [4] NCCL INFO Connected all rings
+ip-26-0-172-142:880683:881136 [4] NCCL INFO Connected all trees
+ip-26-0-172-142:880683:881136 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-142:880683:881136 [4] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-172-142:880683:881136 [4] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-116:1019294:1019776 [6] NCCL INFO comm 0x1364f540 rank 0 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0x525c205cc8134fe3 - Init COMPLETE
+ip-26-0-172-116:1019294:1019808 [6] NCCL INFO Channel 02/1 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA/Shared
+ip-26-0-172-116:1019294:1019808 [6] NCCL INFO Channel 03/1 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA/Shared
+ip-26-0-172-142:880680:881135 [1] NCCL INFO Connected all rings
+ip-26-0-172-142:880680:881135 [1] NCCL INFO Connected all trees
+ip-26-0-172-142:880680:881135 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-142:880680:881135 [1] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-172-142:880680:881135 [1] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-142:880683:881136 [4] NCCL INFO comm 0xc0a1680 rank 1 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0xa6165201324a543d - Init COMPLETE
+ip-26-0-172-142:880683:881160 [4] NCCL INFO Channel 02/1 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA/Shared
+ip-26-0-172-142:880683:881160 [4] NCCL INFO Channel 03/1 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA/Shared
+ip-26-0-172-116:1019292:1019780 [4] NCCL INFO Connected all rings
+ip-26-0-172-116:1019292:1019780 [4] NCCL INFO Connected all trees
+ip-26-0-172-116:1019292:1019780 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-116:1019292:1019780 [4] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-172-116:1019292:1019780 [4] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-116:1019289:1019778 [1] NCCL INFO Connected all rings
+ip-26-0-172-116:1019289:1019778 [1] NCCL INFO Connected all trees
+ip-26-0-172-116:1019289:1019778 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-116:1019289:1019778 [1] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-172-116:1019289:1019778 [1] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-142:880680:881135 [1] NCCL INFO comm 0xb34b180 rank 1 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0x6b45b633dd439f28 - Init COMPLETE
+ip-26-0-172-142:880680:881161 [1] NCCL INFO Channel 02/1 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA/Shared
+ip-26-0-172-142:880680:881161 [1] NCCL INFO Channel 03/1 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA/Shared
+ip-26-0-172-142:880682:881142 [3] NCCL INFO Connected all rings
+ip-26-0-172-142:880682:881142 [3] NCCL INFO Connected all trees
+ip-26-0-172-142:880682:881142 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-142:880682:881142 [3] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-172-142:880682:881142 [3] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-116:1019292:1019780 [4] NCCL INFO comm 0x153adb70 rank 0 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0xa6165201324a543d - Init COMPLETE
+ip-26-0-172-116:1019292:1019809 [4] NCCL INFO Channel 02/1 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA/Shared
+ip-26-0-172-116:1019292:1019809 [4] NCCL INFO Channel 03/1 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA/Shared
+ip-26-0-172-116:1019289:1019778 [1] NCCL INFO comm 0x1364b770 rank 0 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0x6b45b633dd439f28 - Init COMPLETE
+ip-26-0-172-116:1019289:1019810 [1] NCCL INFO Channel 02/1 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA/Shared
+ip-26-0-172-116:1019289:1019810 [1] NCCL INFO Channel 03/1 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA/Shared
+ip-26-0-172-142:880681:881143 [2] NCCL INFO Connected all rings
+ip-26-0-172-142:880681:881143 [2] NCCL INFO Connected all trees
+ip-26-0-172-142:880681:881143 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-142:880681:881143 [2] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-172-142:880681:881143 [2] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-142:880682:881142 [3] NCCL INFO comm 0xb91c700 rank 1 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0xc898011f09db575d - Init COMPLETE
+ip-26-0-172-142:880682:881162 [3] NCCL INFO Channel 02/1 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA/Shared
+ip-26-0-172-142:880682:881162 [3] NCCL INFO Channel 03/1 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA/Shared
+ip-26-0-172-116:1019291:1019787 [3] NCCL INFO Connected all rings
+ip-26-0-172-116:1019291:1019787 [3] NCCL INFO Connected all trees
+ip-26-0-172-116:1019291:1019787 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-116:1019291:1019787 [3] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-172-116:1019291:1019787 [3] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-142:880681:881143 [2] NCCL INFO comm 0xb481ae0 rank 1 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0x7356f1a08d3799bc - Init COMPLETE
+ip-26-0-172-142:880681:881163 [2] NCCL INFO Channel 02/1 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA/Shared
+ip-26-0-172-142:880681:881163 [2] NCCL INFO Channel 03/1 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA/Shared
+ip-26-0-172-116:1019290:1019788 [2] NCCL INFO Connected all rings
+ip-26-0-172-116:1019290:1019788 [2] NCCL INFO Connected all trees
+ip-26-0-172-116:1019290:1019788 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-116:1019290:1019788 [2] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-172-116:1019290:1019788 [2] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-142:880684:881146 [5] NCCL INFO Connected all rings
+ip-26-0-172-142:880684:881146 [5] NCCL INFO Connected all trees
+ip-26-0-172-142:880684:881146 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-142:880684:881146 [5] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-172-142:880684:881146 [5] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-116:1019291:1019787 [3] NCCL INFO comm 0x153c2530 rank 0 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0xc898011f09db575d - Init COMPLETE
+ip-26-0-172-116:1019291:1019811 [3] NCCL INFO Channel 02/1 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA/Shared
+ip-26-0-172-116:1019291:1019811 [3] NCCL INFO Channel 03/1 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA/Shared
+ip-26-0-172-142:880684:881146 [5] NCCL INFO comm 0xb05e6d0 rank 1 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0xf33f0cb6a1d78c81 - Init COMPLETE
+ip-26-0-172-116:1019290:1019788 [2] NCCL INFO comm 0x1503aa00 rank 0 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0x7356f1a08d3799bc - Init COMPLETE
+ip-26-0-172-142:880684:881164 [5] NCCL INFO Channel 02/1 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA/Shared
+ip-26-0-172-142:880684:881164 [5] NCCL INFO Channel 03/1 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA/Shared
+ip-26-0-172-116:1019290:1019812 [2] NCCL INFO Channel 02/1 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA/Shared
+ip-26-0-172-116:1019290:1019812 [2] NCCL INFO Channel 03/1 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA/Shared
+ip-26-0-172-116:1019293:1019792 [5] NCCL INFO Connected all rings
+ip-26-0-172-116:1019293:1019792 [5] NCCL INFO Connected all trees
+ip-26-0-172-116:1019293:1019792 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-116:1019293:1019792 [5] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-172-116:1019293:1019792 [5] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-116:1019293:1019792 [5] NCCL INFO comm 0x1544fc70 rank 0 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0xf33f0cb6a1d78c81 - Init COMPLETE
+ip-26-0-172-116:1019293:1019813 [5] NCCL INFO Channel 02/1 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA/Shared
+ip-26-0-172-116:1019293:1019813 [5] NCCL INFO Channel 03/1 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA/Shared
+ip-26-0-172-142:880679:881151 [0] NCCL INFO Connected all rings
+ip-26-0-172-142:880679:881151 [0] NCCL INFO Connected all trees
+ip-26-0-172-142:880679:881151 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-142:880679:881151 [0] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-172-142:880679:881151 [0] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-116:1019288:1019799 [0] NCCL INFO Connected all rings
+ip-26-0-172-116:1019288:1019799 [0] NCCL INFO Connected all trees
+ip-26-0-172-116:1019288:1019799 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-116:1019288:1019799 [0] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-172-116:1019288:1019799 [0] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-142:880679:881151 [0] NCCL INFO comm 0xc595560 rank 1 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0x59152c3b439684bf - Init COMPLETE
+ip-26-0-172-142:880679:881165 [0] NCCL INFO Channel 02/1 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA/Shared
+ip-26-0-172-142:880679:881165 [0] NCCL INFO Channel 03/1 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA/Shared
+ip-26-0-172-116:1019288:1019799 [0] NCCL INFO comm 0x13c79070 rank 0 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0x59152c3b439684bf - Init COMPLETE
+ip-26-0-172-116:1019288:1019814 [0] NCCL INFO Channel 02/1 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA/Shared
+ip-26-0-172-116:1019288:1019814 [0] NCCL INFO Channel 03/1 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA/Shared
+ip-26-0-172-142:880686:881152 [7] NCCL INFO Connected all rings
+ip-26-0-172-142:880686:881152 [7] NCCL INFO Connected all trees
+ip-26-0-172-142:880686:881152 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-142:880686:881152 [7] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-172-142:880686:881152 [7] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-142:880686:881152 [7] NCCL INFO comm 0xbab76c0 rank 1 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0xb6ad2873cfe08d3b - Init COMPLETE
+ip-26-0-172-142:880686:881166 [7] NCCL INFO Channel 02/1 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA/Shared
+ip-26-0-172-142:880686:881166 [7] NCCL INFO Channel 03/1 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA/Shared
+ip-26-0-172-116:1019295:1019801 [7] NCCL INFO Connected all rings
+ip-26-0-172-116:1019295:1019801 [7] NCCL INFO Connected all trees
+ip-26-0-172-116:1019295:1019801 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-116:1019295:1019801 [7] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-172-116:1019295:1019801 [7] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-116:1019295:1019801 [7] NCCL INFO comm 0x14a1a1f0 rank 0 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0xb6ad2873cfe08d3b - Init COMPLETE
+ip-26-0-172-116:1019295:1019815 [7] NCCL INFO Channel 02/1 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA/Shared
+ip-26-0-172-116:1019295:1019815 [7] NCCL INFO Channel 03/1 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA/Shared
+ip-26-0-172-116:1019294:1019824 [6] NCCL INFO Channel 02/1 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA/Shared
+ip-26-0-172-116:1019294:1019824 [6] NCCL INFO Channel 03/1 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA/Shared
+ip-26-0-172-116:1019289:1019833 [1] NCCL INFO Channel 02/1 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA/Shared
+ip-26-0-172-116:1019289:1019833 [1] NCCL INFO Channel 03/1 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA/Shared
+ip-26-0-172-116:1019292:1019842 [4] NCCL INFO Channel 02/1 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA/Shared
+ip-26-0-172-116:1019292:1019842 [4] NCCL INFO Channel 03/1 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA/Shared
+ip-26-0-172-116:1019291:1019851 [3] NCCL INFO Channel 02/1 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA/Shared
+ip-26-0-172-116:1019291:1019851 [3] NCCL INFO Channel 03/1 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA/Shared
+ip-26-0-172-116:1019290:1019860 [2] NCCL INFO Channel 02/1 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA/Shared
+ip-26-0-172-116:1019290:1019860 [2] NCCL INFO Channel 03/1 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA/Shared
+ip-26-0-172-116:1019293:1019869 [5] NCCL INFO Channel 02/1 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA/Shared
+ip-26-0-172-116:1019293:1019869 [5] NCCL INFO Channel 03/1 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA/Shared
+ip-26-0-172-116:1019288:1019878 [0] NCCL INFO Channel 02/1 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA/Shared
+ip-26-0-172-116:1019288:1019878 [0] NCCL INFO Channel 03/1 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA/Shared
+ip-26-0-172-116:1019295:1019887 [7] NCCL INFO Channel 02/1 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA/Shared
+ip-26-0-172-116:1019295:1019887 [7] NCCL INFO Channel 03/1 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA/Shared
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+ip-26-0-172-142:880685:881170 [6] NCCL INFO Using network Libfabric
+ip-26-0-172-147:618590:619040 [6] NCCL INFO Using network Libfabric
+ip-26-0-172-147:618590:619040 [6] NCCL INFO comm 0xc5cc610 rank 1 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0x2c6f4b901609cd27 - Init START
+ip-26-0-172-142:880685:881170 [6] NCCL INFO comm 0x139838b0 rank 0 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0x2c6f4b901609cd27 - Init START
+ip-26-0-172-147:618590:619040 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-142:880685:881170 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+ip-26-0-172-142:880683:881172 [4] NCCL INFO Using network Libfabric
+ip-26-0-172-147:618588:619041 [4] NCCL INFO Using network Libfabric
+ip-26-0-172-147:618588:619041 [4] NCCL INFO comm 0xabfc930 rank 1 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0x554c6402e5aec2e4 - Init START
+ip-26-0-172-142:880683:881172 [4] NCCL INFO comm 0x14ca7bc0 rank 0 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0x554c6402e5aec2e4 - Init START
+ip-26-0-172-147:618588:619041 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-142:880683:881172 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+ip-26-0-172-142:880680:881174 [1] NCCL INFO Using network Libfabric
+ip-26-0-172-147:618585:619042 [1] NCCL INFO Using network Libfabric
+ip-26-0-172-147:618585:619042 [1] NCCL INFO comm 0xc345e10 rank 1 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0x7b41a802d809a123 - Init START
+ip-26-0-172-142:880680:881174 [1] NCCL INFO comm 0x13fc18f0 rank 0 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0x7b41a802d809a123 - Init START
+ip-26-0-172-147:618585:619042 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-142:880680:881174 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+ip-26-0-172-142:880682:881176 [3] NCCL INFO Using network Libfabric
+ip-26-0-172-147:618587:619043 [3] NCCL INFO Using network Libfabric
+ip-26-0-172-147:618587:619043 [3] NCCL INFO comm 0xaf43950 rank 1 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0x2b15454c0616ac9b - Init START
+ip-26-0-172-142:880682:881176 [3] NCCL INFO comm 0x14528240 rank 0 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0x2b15454c0616ac9b - Init START
+ip-26-0-172-147:618587:619043 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-142:880682:881176 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-147:618590:619040 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+ip-26-0-172-142:880681:881178 [2] NCCL INFO Using network Libfabric
+ip-26-0-172-147:618586:619044 [2] NCCL INFO Using network Libfabric
+ip-26-0-172-142:880685:881170 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-172-147:618586:619044 [2] NCCL INFO comm 0xaf62530 rank 1 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0xb44b452b89dfaf88 - Init START
+ip-26-0-172-142:880681:881178 [2] NCCL INFO comm 0x14089840 rank 0 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0xb44b452b89dfaf88 - Init START
+ip-26-0-172-147:618586:619044 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-142:880681:881178 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-142:880685:881170 [6] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-172-142:880685:881170 [6] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-172-142:880685:881170 [6] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-172-142:880685:881170 [6] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-172-142:880685:881170 [6] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-172-142:880685:881170 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-147:618590:619040 [6] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-172-147:618590:619040 [6] NCCL INFO P2P Chunksize set to 131072
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+ip-26-0-172-142:880684:881183 [5] NCCL INFO Using network Libfabric
+ip-26-0-172-147:618589:619047 [5] NCCL INFO Using network Libfabric
+ip-26-0-172-147:618589:619047 [5] NCCL INFO comm 0xbc3efa0 rank 1 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0xce17eb5cc20165b4 - Init START
+ip-26-0-172-142:880684:881183 [5] NCCL INFO comm 0x13aca8b0 rank 0 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0xce17eb5cc20165b4 - Init START
+ip-26-0-172-142:880684:881183 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-147:618589:619047 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-142:880685:881170 [6] NCCL INFO Channel 00/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880685:881170 [6] NCCL INFO Channel 01/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880685:881170 [6] NCCL INFO Channel 02/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880685:881170 [6] NCCL INFO Channel 03/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880685:881170 [6] NCCL INFO Channel 00/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880685:881170 [6] NCCL INFO Channel 01/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880685:881170 [6] NCCL INFO Channel 02/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-142:880685:881170 [6] NCCL INFO Channel 03/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618590:619040 [6] NCCL INFO Channel 00/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618590:619040 [6] NCCL INFO Channel 01/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618590:619040 [6] NCCL INFO Channel 02/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618590:619040 [6] NCCL INFO Channel 03/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618590:619040 [6] NCCL INFO Channel 00/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618590:619040 [6] NCCL INFO Channel 01/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618590:619040 [6] NCCL INFO Channel 02/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618590:619040 [6] NCCL INFO Channel 03/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618585:619042 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-172-147:618588:619041 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-172-142:880683:881172 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-172-142:880683:881172 [4] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-172-142:880683:881172 [4] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-172-142:880683:881172 [4] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-172-147:618588:619041 [4] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-172-147:618588:619041 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-142:880683:881172 [4] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-172-142:880683:881172 [4] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-172-142:880683:881172 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-142:880680:881174 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-172-142:880680:881174 [1] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-172-142:880680:881174 [1] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-172-142:880680:881174 [1] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-172-142:880680:881174 [1] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-172-142:880680:881174 [1] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-172-142:880680:881174 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-147:618585:619042 [1] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-172-147:618585:619042 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-147:618587:619043 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-172-147:618588:619041 [4] NCCL INFO Channel 00/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618588:619041 [4] NCCL INFO Channel 01/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618588:619041 [4] NCCL INFO Channel 02/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618588:619041 [4] NCCL INFO Channel 03/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618588:619041 [4] NCCL INFO Channel 00/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618588:619041 [4] NCCL INFO Channel 01/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618588:619041 [4] NCCL INFO Channel 02/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618588:619041 [4] NCCL INFO Channel 03/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880683:881172 [4] NCCL INFO Channel 00/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880683:881172 [4] NCCL INFO Channel 01/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880683:881172 [4] NCCL INFO Channel 02/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880683:881172 [4] NCCL INFO Channel 03/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880683:881172 [4] NCCL INFO Channel 00/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880683:881172 [4] NCCL INFO Channel 01/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880683:881172 [4] NCCL INFO Channel 02/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-142:880683:881172 [4] NCCL INFO Channel 03/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618586:619044 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-172-147:618589:619047 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+ip-26-0-172-142:880679:881189 [0] NCCL INFO Using network Libfabric
+ip-26-0-172-147:618584:619053 [0] NCCL INFO Using network Libfabric
+ip-26-0-172-147:618584:619053 [0] NCCL INFO comm 0xbf223b0 rank 1 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0x7d96b932922a0b23 - Init START
+ip-26-0-172-142:880679:881189 [0] NCCL INFO comm 0x1520d2b0 rank 0 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0x7d96b932922a0b23 - Init START
+ip-26-0-172-142:880679:881189 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-147:618584:619053 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-147:618585:619042 [1] NCCL INFO Channel 00/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618585:619042 [1] NCCL INFO Channel 01/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618585:619042 [1] NCCL INFO Channel 02/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618585:619042 [1] NCCL INFO Channel 03/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618585:619042 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618585:619042 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618585:619042 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618585:619042 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880680:881174 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880680:881174 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880680:881174 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880680:881174 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880680:881174 [1] NCCL INFO Channel 00/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880680:881174 [1] NCCL INFO Channel 01/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880680:881174 [1] NCCL INFO Channel 02/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880680:881174 [1] NCCL INFO Channel 03/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-142:880681:881178 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-172-147:618586:619044 [2] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-172-147:618586:619044 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-142:880681:881178 [2] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-172-142:880681:881178 [2] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-172-142:880681:881178 [2] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-172-142:880681:881178 [2] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-172-142:880681:881178 [2] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-172-142:880681:881178 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-142:880682:881176 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-172-142:880682:881176 [3] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-172-142:880682:881176 [3] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-172-142:880682:881176 [3] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-172-142:880682:881176 [3] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-172-142:880682:881176 [3] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-172-142:880682:881176 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-147:618587:619043 [3] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-172-147:618587:619043 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-142:880684:881183 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-172-147:618589:619047 [5] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-172-147:618589:619047 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-142:880684:881183 [5] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-172-142:880684:881183 [5] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-172-142:880684:881183 [5] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-172-142:880684:881183 [5] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-172-142:880684:881183 [5] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-172-142:880684:881183 [5] NCCL INFO P2P Chunksize set to 131072
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+ip-26-0-172-142:880686:881196 [7] NCCL INFO Using network Libfabric
+ip-26-0-172-147:618591:619059 [7] NCCL INFO Using network Libfabric
+ip-26-0-172-147:618591:619059 [7] NCCL INFO comm 0xb6ff4d0 rank 1 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0xf9647465a95971d6 - Init START
+ip-26-0-172-142:880686:881196 [7] NCCL INFO comm 0x14778ab0 rank 0 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0xf9647465a95971d6 - Init START
+ip-26-0-172-147:618591:619059 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-142:880686:881196 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-147:618586:619044 [2] NCCL INFO Channel 00/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618586:619044 [2] NCCL INFO Channel 01/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618586:619044 [2] NCCL INFO Channel 02/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618586:619044 [2] NCCL INFO Channel 03/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618586:619044 [2] NCCL INFO Channel 00/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880681:881178 [2] NCCL INFO Channel 00/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618586:619044 [2] NCCL INFO Channel 01/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880681:881178 [2] NCCL INFO Channel 01/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618586:619044 [2] NCCL INFO Channel 02/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618586:619044 [2] NCCL INFO Channel 03/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880681:881178 [2] NCCL INFO Channel 02/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880681:881178 [2] NCCL INFO Channel 03/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880681:881178 [2] NCCL INFO Channel 00/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880681:881178 [2] NCCL INFO Channel 01/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880681:881178 [2] NCCL INFO Channel 02/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-142:880681:881178 [2] NCCL INFO Channel 03/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618587:619043 [3] NCCL INFO Channel 00/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618587:619043 [3] NCCL INFO Channel 01/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618587:619043 [3] NCCL INFO Channel 02/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618587:619043 [3] NCCL INFO Channel 03/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618587:619043 [3] NCCL INFO Channel 00/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618587:619043 [3] NCCL INFO Channel 01/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618587:619043 [3] NCCL INFO Channel 02/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880682:881176 [3] NCCL INFO Channel 00/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618587:619043 [3] NCCL INFO Channel 03/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880682:881176 [3] NCCL INFO Channel 01/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880682:881176 [3] NCCL INFO Channel 02/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880682:881176 [3] NCCL INFO Channel 03/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880682:881176 [3] NCCL INFO Channel 00/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880682:881176 [3] NCCL INFO Channel 01/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880682:881176 [3] NCCL INFO Channel 02/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-142:880682:881176 [3] NCCL INFO Channel 03/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618589:619047 [5] NCCL INFO Channel 00/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618589:619047 [5] NCCL INFO Channel 01/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618589:619047 [5] NCCL INFO Channel 02/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618589:619047 [5] NCCL INFO Channel 03/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618589:619047 [5] NCCL INFO Channel 00/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618589:619047 [5] NCCL INFO Channel 01/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618589:619047 [5] NCCL INFO Channel 02/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618589:619047 [5] NCCL INFO Channel 03/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880684:881183 [5] NCCL INFO Channel 00/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880684:881183 [5] NCCL INFO Channel 01/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880684:881183 [5] NCCL INFO Channel 02/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880684:881183 [5] NCCL INFO Channel 03/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880684:881183 [5] NCCL INFO Channel 00/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880684:881183 [5] NCCL INFO Channel 01/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880684:881183 [5] NCCL INFO Channel 02/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-142:880684:881183 [5] NCCL INFO Channel 03/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618584:619053 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-172-142:880679:881189 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-172-142:880679:881189 [0] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-172-142:880679:881189 [0] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-172-142:880679:881189 [0] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-172-142:880679:881189 [0] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-172-142:880679:881189 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-172-142:880679:881189 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-147:618584:619053 [0] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-172-147:618584:619053 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-147:618591:619059 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-172-142:880686:881196 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-172-142:880686:881196 [7] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-172-142:880686:881196 [7] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-172-142:880686:881196 [7] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-172-142:880686:881196 [7] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-172-142:880686:881196 [7] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-172-142:880686:881196 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-147:618591:619059 [7] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-172-147:618591:619059 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-147:618584:619053 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618584:619053 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618584:619053 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618584:619053 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880679:881189 [0] NCCL INFO Channel 00/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618584:619053 [0] NCCL INFO Channel 00/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880679:881189 [0] NCCL INFO Channel 01/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618584:619053 [0] NCCL INFO Channel 01/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880679:881189 [0] NCCL INFO Channel 02/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618584:619053 [0] NCCL INFO Channel 02/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880679:881189 [0] NCCL INFO Channel 03/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618584:619053 [0] NCCL INFO Channel 03/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880679:881189 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880679:881189 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880679:881189 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-142:880679:881189 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618590:619040 [6] NCCL INFO Connected all rings
+ip-26-0-172-147:618590:619040 [6] NCCL INFO Connected all trees
+ip-26-0-172-147:618590:619040 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-147:618590:619040 [6] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-172-147:618590:619040 [6] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-147:618590:619040 [6] NCCL INFO comm 0xc5cc610 rank 1 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0x2c6f4b901609cd27 - Init COMPLETE
+ip-26-0-172-147:618590:619065 [6] NCCL INFO Channel 02/1 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA/Shared
+ip-26-0-172-147:618590:619065 [6] NCCL INFO Channel 03/1 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA/Shared
+ip-26-0-172-142:880686:881196 [7] NCCL INFO Channel 00/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880686:881196 [7] NCCL INFO Channel 01/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880686:881196 [7] NCCL INFO Channel 02/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880686:881196 [7] NCCL INFO Channel 03/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880686:881196 [7] NCCL INFO Channel 00/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880686:881196 [7] NCCL INFO Channel 01/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880686:881196 [7] NCCL INFO Channel 02/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880686:881196 [7] NCCL INFO Channel 03/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618591:619059 [7] NCCL INFO Channel 00/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618591:619059 [7] NCCL INFO Channel 01/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618591:619059 [7] NCCL INFO Channel 02/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618591:619059 [7] NCCL INFO Channel 03/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618591:619059 [7] NCCL INFO Channel 00/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618591:619059 [7] NCCL INFO Channel 01/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618591:619059 [7] NCCL INFO Channel 02/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618591:619059 [7] NCCL INFO Channel 03/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-142:880685:881170 [6] NCCL INFO Connected all rings
+ip-26-0-172-142:880685:881170 [6] NCCL INFO Connected all trees
+ip-26-0-172-142:880685:881170 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-142:880685:881170 [6] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-172-142:880685:881170 [6] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-147:618588:619041 [4] NCCL INFO Connected all rings
+ip-26-0-172-147:618588:619041 [4] NCCL INFO Connected all trees
+ip-26-0-172-147:618588:619041 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-147:618588:619041 [4] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-172-147:618588:619041 [4] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-142:880685:881170 [6] NCCL INFO comm 0x139838b0 rank 0 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0x2c6f4b901609cd27 - Init COMPLETE
+ip-26-0-172-142:880685:881202 [6] NCCL INFO Channel 02/1 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA/Shared
+ip-26-0-172-142:880685:881202 [6] NCCL INFO Channel 03/1 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA/Shared
+ip-26-0-172-147:618585:619042 [1] NCCL INFO Connected all rings
+ip-26-0-172-147:618585:619042 [1] NCCL INFO Connected all trees
+ip-26-0-172-147:618585:619042 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-147:618585:619042 [1] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-172-147:618585:619042 [1] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-147:618588:619041 [4] NCCL INFO comm 0xabfc930 rank 1 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0x554c6402e5aec2e4 - Init COMPLETE
+ip-26-0-172-147:618588:619066 [4] NCCL INFO Channel 02/1 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA/Shared
+ip-26-0-172-147:618588:619066 [4] NCCL INFO Channel 03/1 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA/Shared
+ip-26-0-172-147:618585:619042 [1] NCCL INFO comm 0xc345e10 rank 1 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0x7b41a802d809a123 - Init COMPLETE
+ip-26-0-172-147:618585:619067 [1] NCCL INFO Channel 02/1 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA/Shared
+ip-26-0-172-147:618585:619067 [1] NCCL INFO Channel 03/1 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA/Shared
+ip-26-0-172-147:618586:619044 [2] NCCL INFO Connected all rings
+ip-26-0-172-147:618586:619044 [2] NCCL INFO Connected all trees
+ip-26-0-172-147:618586:619044 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-147:618586:619044 [2] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-172-147:618586:619044 [2] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-142:880683:881172 [4] NCCL INFO Connected all rings
+ip-26-0-172-142:880683:881172 [4] NCCL INFO Connected all trees
+ip-26-0-172-142:880683:881172 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-142:880683:881172 [4] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-172-142:880683:881172 [4] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-147:618587:619043 [3] NCCL INFO Connected all rings
+ip-26-0-172-147:618587:619043 [3] NCCL INFO Connected all trees
+ip-26-0-172-147:618587:619043 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-147:618587:619043 [3] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-172-147:618587:619043 [3] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-142:880680:881174 [1] NCCL INFO Connected all rings
+ip-26-0-172-142:880680:881174 [1] NCCL INFO Connected all trees
+ip-26-0-172-142:880680:881174 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-142:880680:881174 [1] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-172-142:880680:881174 [1] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-142:880683:881172 [4] NCCL INFO comm 0x14ca7bc0 rank 0 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0x554c6402e5aec2e4 - Init COMPLETE
+ip-26-0-172-142:880683:881203 [4] NCCL INFO Channel 02/1 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA/Shared
+ip-26-0-172-142:880683:881203 [4] NCCL INFO Channel 03/1 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA/Shared
+ip-26-0-172-147:618586:619044 [2] NCCL INFO comm 0xaf62530 rank 1 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0xb44b452b89dfaf88 - Init COMPLETE
+ip-26-0-172-147:618586:619068 [2] NCCL INFO Channel 02/1 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA/Shared
+ip-26-0-172-147:618586:619068 [2] NCCL INFO Channel 03/1 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA/Shared
+ip-26-0-172-147:618587:619043 [3] NCCL INFO comm 0xaf43950 rank 1 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0x2b15454c0616ac9b - Init COMPLETE
+ip-26-0-172-147:618587:619069 [3] NCCL INFO Channel 02/1 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA/Shared
+ip-26-0-172-147:618587:619069 [3] NCCL INFO Channel 03/1 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA/Shared
+ip-26-0-172-147:618589:619047 [5] NCCL INFO Connected all rings
+ip-26-0-172-147:618589:619047 [5] NCCL INFO Connected all trees
+ip-26-0-172-147:618589:619047 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-147:618589:619047 [5] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-172-147:618589:619047 [5] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-142:880680:881174 [1] NCCL INFO comm 0x13fc18f0 rank 0 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0x7b41a802d809a123 - Init COMPLETE
+ip-26-0-172-142:880680:881204 [1] NCCL INFO Channel 02/1 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA/Shared
+ip-26-0-172-142:880680:881204 [1] NCCL INFO Channel 03/1 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA/Shared
+ip-26-0-172-142:880681:881178 [2] NCCL INFO Connected all rings
+ip-26-0-172-142:880681:881178 [2] NCCL INFO Connected all trees
+ip-26-0-172-142:880681:881178 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-142:880681:881178 [2] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-172-142:880681:881178 [2] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-142:880682:881176 [3] NCCL INFO Connected all rings
+ip-26-0-172-142:880682:881176 [3] NCCL INFO Connected all trees
+ip-26-0-172-142:880682:881176 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-142:880682:881176 [3] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-172-142:880682:881176 [3] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-142:880684:881183 [5] NCCL INFO Connected all rings
+ip-26-0-172-142:880684:881183 [5] NCCL INFO Connected all trees
+ip-26-0-172-142:880684:881183 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-142:880684:881183 [5] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-172-142:880684:881183 [5] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-147:618589:619047 [5] NCCL INFO comm 0xbc3efa0 rank 1 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0xce17eb5cc20165b4 - Init COMPLETE
+ip-26-0-172-147:618589:619070 [5] NCCL INFO Channel 02/1 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA/Shared
+ip-26-0-172-147:618589:619070 [5] NCCL INFO Channel 03/1 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA/Shared
+ip-26-0-172-142:880681:881178 [2] NCCL INFO comm 0x14089840 rank 0 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0xb44b452b89dfaf88 - Init COMPLETE
+ip-26-0-172-142:880681:881205 [2] NCCL INFO Channel 02/1 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA/Shared
+ip-26-0-172-142:880681:881205 [2] NCCL INFO Channel 03/1 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA/Shared
+ip-26-0-172-142:880682:881176 [3] NCCL INFO comm 0x14528240 rank 0 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0x2b15454c0616ac9b - Init COMPLETE
+ip-26-0-172-142:880682:881206 [3] NCCL INFO Channel 02/1 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA/Shared
+ip-26-0-172-142:880682:881206 [3] NCCL INFO Channel 03/1 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA/Shared
+ip-26-0-172-142:880684:881183 [5] NCCL INFO comm 0x13aca8b0 rank 0 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0xce17eb5cc20165b4 - Init COMPLETE
+ip-26-0-172-142:880684:881207 [5] NCCL INFO Channel 02/1 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA/Shared
+ip-26-0-172-142:880684:881207 [5] NCCL INFO Channel 03/1 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA/Shared
+ip-26-0-172-142:880679:881189 [0] NCCL INFO Connected all rings
+ip-26-0-172-142:880679:881189 [0] NCCL INFO Connected all trees
+ip-26-0-172-142:880679:881189 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-142:880679:881189 [0] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-172-142:880679:881189 [0] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-147:618584:619053 [0] NCCL INFO Connected all rings
+ip-26-0-172-147:618584:619053 [0] NCCL INFO Connected all trees
+ip-26-0-172-147:618584:619053 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-147:618584:619053 [0] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-172-147:618584:619053 [0] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-142:880679:881189 [0] NCCL INFO comm 0x1520d2b0 rank 0 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0x7d96b932922a0b23 - Init COMPLETE
+ip-26-0-172-142:880679:881208 [0] NCCL INFO Channel 02/1 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA/Shared
+ip-26-0-172-142:880679:881208 [0] NCCL INFO Channel 03/1 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA/Shared
+ip-26-0-172-147:618584:619053 [0] NCCL INFO comm 0xbf223b0 rank 1 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0x7d96b932922a0b23 - Init COMPLETE
+ip-26-0-172-147:618584:619071 [0] NCCL INFO Channel 02/1 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA/Shared
+ip-26-0-172-147:618584:619071 [0] NCCL INFO Channel 03/1 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA/Shared
+ip-26-0-172-147:618591:619059 [7] NCCL INFO Connected all rings
+ip-26-0-172-147:618591:619059 [7] NCCL INFO Connected all trees
+ip-26-0-172-147:618591:619059 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-147:618591:619059 [7] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-172-147:618591:619059 [7] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-147:618591:619059 [7] NCCL INFO comm 0xb6ff4d0 rank 1 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0xf9647465a95971d6 - Init COMPLETE
+ip-26-0-172-147:618591:619072 [7] NCCL INFO Channel 02/1 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA/Shared
+ip-26-0-172-147:618591:619072 [7] NCCL INFO Channel 03/1 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA/Shared
+ip-26-0-172-142:880686:881196 [7] NCCL INFO Connected all rings
+ip-26-0-172-142:880686:881196 [7] NCCL INFO Connected all trees
+ip-26-0-172-142:880686:881196 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-142:880686:881196 [7] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-172-142:880686:881196 [7] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-142:880686:881196 [7] NCCL INFO comm 0x14778ab0 rank 0 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0xf9647465a95971d6 - Init COMPLETE
+ip-26-0-172-142:880686:881209 [7] NCCL INFO Channel 02/1 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA/Shared
+ip-26-0-172-142:880686:881209 [7] NCCL INFO Channel 03/1 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA/Shared
+ip-26-0-172-142:880685:881218 [6] NCCL INFO Channel 02/1 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA/Shared
+ip-26-0-172-142:880685:881218 [6] NCCL INFO Channel 03/1 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA/Shared
+ip-26-0-172-142:880683:881227 [4] NCCL INFO Channel 02/1 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA/Shared
+ip-26-0-172-142:880683:881227 [4] NCCL INFO Channel 03/1 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA/Shared
+ip-26-0-172-142:880682:881244 [3] NCCL INFO Channel 02/1 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA/Shared
+ip-26-0-172-142:880681:881245 [2] NCCL INFO Channel 02/1 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA/Shared
+ip-26-0-172-142:880682:881244 [3] NCCL INFO Channel 03/1 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA/Shared
+ip-26-0-172-142:880681:881245 [2] NCCL INFO Channel 03/1 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA/Shared
+ip-26-0-172-142:880684:881254 [5] NCCL INFO Channel 02/1 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA/Shared
+ip-26-0-172-142:880684:881254 [5] NCCL INFO Channel 03/1 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA/Shared
+ip-26-0-172-142:880679:881264 [0] NCCL INFO Channel 02/1 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA/Shared
+ip-26-0-172-142:880679:881264 [0] NCCL INFO Channel 03/1 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA/Shared
+ip-26-0-172-142:880680:881273 [1] NCCL INFO Channel 02/1 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA/Shared
+ip-26-0-172-142:880680:881273 [1] NCCL INFO Channel 03/1 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA/Shared
+ip-26-0-172-142:880686:881282 [7] NCCL INFO Channel 02/1 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA/Shared
+ip-26-0-172-142:880686:881282 [7] NCCL INFO Channel 03/1 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA/Shared
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+ip-26-0-172-147:618590:619077 [6] NCCL INFO Using network Libfabric
+ip-26-0-172-252:688206:688662 [6] NCCL INFO Using network Libfabric
+ip-26-0-172-252:688206:688662 [6] NCCL INFO comm 0xc800b60 rank 1 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0xbb6950b11ee120f2 - Init START
+ip-26-0-172-147:618590:619077 [6] NCCL INFO comm 0x151d31f0 rank 0 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0xbb6950b11ee120f2 - Init START
+ip-26-0-172-147:618590:619077 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-252:688206:688662 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-252:688206:688662 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+ip-26-0-172-147:618588:619081 [4] NCCL INFO Using network Libfabric
+ip-26-0-172-147:618585:619080 [1] NCCL INFO Using network Libfabric
+ip-26-0-172-252:688201:688664 [1] NCCL INFO Using network Libfabric
+ip-26-0-172-252:688204:688665 [4] NCCL INFO Using network Libfabric
+ip-26-0-172-252:688201:688664 [1] NCCL INFO comm 0xb6d73d0 rank 1 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0xd3620b1555668162 - Init START
+ip-26-0-172-147:618585:619080 [1] NCCL INFO comm 0x15035980 rank 0 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0xd3620b1555668162 - Init START
+ip-26-0-172-147:618588:619081 [4] NCCL INFO comm 0x138735b0 rank 0 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0xe35f0d1ddc70c8ac - Init START
+ip-26-0-172-252:688204:688665 [4] NCCL INFO comm 0xb918890 rank 1 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0xe35f0d1ddc70c8ac - Init START
+ip-26-0-172-147:618585:619080 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-252:688201:688664 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-147:618588:619081 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-252:688204:688665 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-147:618590:619077 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-172-147:618590:619077 [6] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-172-147:618590:619077 [6] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-172-147:618590:619077 [6] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-172-147:618590:619077 [6] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-172-147:618590:619077 [6] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-172-147:618590:619077 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-252:688206:688662 [6] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-172-252:688206:688662 [6] NCCL INFO P2P Chunksize set to 131072
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+ip-26-0-172-147:618586:619085 [2] NCCL INFO Using network Libfabric
+ip-26-0-172-252:688202:688668 [2] NCCL INFO Using network Libfabric
+ip-26-0-172-252:688202:688668 [2] NCCL INFO comm 0xb638870 rank 1 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0x94c2144e0212d222 - Init START
+ip-26-0-172-147:618586:619085 [2] NCCL INFO comm 0x13b675c0 rank 0 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0x94c2144e0212d222 - Init START
+ip-26-0-172-147:618586:619085 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-252:688202:688668 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+ip-26-0-172-147:618589:619087 [5] NCCL INFO Using network Libfabric
+ip-26-0-172-252:688205:688669 [5] NCCL INFO Using network Libfabric
+ip-26-0-172-252:688205:688669 [5] NCCL INFO comm 0xb282390 rank 1 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0xcd9768f6a4298843 - Init START
+ip-26-0-172-147:618589:619087 [5] NCCL INFO comm 0x14847a80 rank 0 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0xcd9768f6a4298843 - Init START
+ip-26-0-172-147:618589:619087 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-252:688205:688669 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-147:618590:619077 [6] NCCL INFO Channel 00/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618590:619077 [6] NCCL INFO Channel 01/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618590:619077 [6] NCCL INFO Channel 02/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618590:619077 [6] NCCL INFO Channel 03/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618590:619077 [6] NCCL INFO Channel 00/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618590:619077 [6] NCCL INFO Channel 01/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618590:619077 [6] NCCL INFO Channel 02/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-147:618590:619077 [6] NCCL INFO Channel 03/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688206:688662 [6] NCCL INFO Channel 00/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688206:688662 [6] NCCL INFO Channel 01/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688206:688662 [6] NCCL INFO Channel 02/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688206:688662 [6] NCCL INFO Channel 03/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688206:688662 [6] NCCL INFO Channel 00/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688206:688662 [6] NCCL INFO Channel 01/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688206:688662 [6] NCCL INFO Channel 02/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688206:688662 [6] NCCL INFO Channel 03/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+ip-26-0-172-147:618587:619089 [3] NCCL INFO Using network Libfabric
+ip-26-0-172-252:688203:688670 [3] NCCL INFO Using network Libfabric
+ip-26-0-172-252:688203:688670 [3] NCCL INFO comm 0xac123f0 rank 1 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0x599ed5a91849038c - Init START
+ip-26-0-172-147:618587:619089 [3] NCCL INFO comm 0x13bb64b0 rank 0 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0x599ed5a91849038c - Init START
+ip-26-0-172-147:618587:619089 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-252:688203:688670 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+ip-26-0-172-147:618584:619091 [0] NCCL INFO Using network Libfabric
+ip-26-0-172-252:688200:688671 [0] NCCL INFO Using network Libfabric
+ip-26-0-172-252:688200:688671 [0] NCCL INFO comm 0xb7aad80 rank 1 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0xc76d1c86b2edff9b - Init START
+ip-26-0-172-147:618584:619091 [0] NCCL INFO comm 0x14b26940 rank 0 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0xc76d1c86b2edff9b - Init START
+ip-26-0-172-147:618584:619091 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-252:688200:688671 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-147:618588:619081 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-172-252:688204:688665 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-172-252:688204:688665 [4] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-172-252:688204:688665 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-147:618588:619081 [4] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-172-147:618588:619081 [4] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-172-147:618588:619081 [4] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-172-147:618588:619081 [4] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-172-147:618588:619081 [4] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-172-147:618588:619081 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-252:688201:688664 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-172-147:618585:619080 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-172-147:618585:619080 [1] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-172-147:618585:619080 [1] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-172-147:618585:619080 [1] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-172-147:618585:619080 [1] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-172-147:618585:619080 [1] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-172-147:618585:619080 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-252:688201:688664 [1] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-172-252:688201:688664 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-252:688202:688668 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-172-252:688204:688665 [4] NCCL INFO Channel 00/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688204:688665 [4] NCCL INFO Channel 01/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688204:688665 [4] NCCL INFO Channel 02/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688204:688665 [4] NCCL INFO Channel 03/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688204:688665 [4] NCCL INFO Channel 00/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688204:688665 [4] NCCL INFO Channel 01/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688204:688665 [4] NCCL INFO Channel 02/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618588:619081 [4] NCCL INFO Channel 00/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688204:688665 [4] NCCL INFO Channel 03/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618588:619081 [4] NCCL INFO Channel 01/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618588:619081 [4] NCCL INFO Channel 02/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618588:619081 [4] NCCL INFO Channel 03/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618588:619081 [4] NCCL INFO Channel 00/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618588:619081 [4] NCCL INFO Channel 01/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618588:619081 [4] NCCL INFO Channel 02/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-147:618588:619081 [4] NCCL INFO Channel 03/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688205:688669 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-172-252:688201:688664 [1] NCCL INFO Channel 00/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688201:688664 [1] NCCL INFO Channel 01/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688201:688664 [1] NCCL INFO Channel 02/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688201:688664 [1] NCCL INFO Channel 03/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688201:688664 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688201:688664 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688201:688664 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688201:688664 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618585:619080 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618585:619080 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618585:619080 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618585:619080 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618585:619080 [1] NCCL INFO Channel 00/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618585:619080 [1] NCCL INFO Channel 01/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618585:619080 [1] NCCL INFO Channel 02/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618585:619080 [1] NCCL INFO Channel 03/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-147:618587:619089 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-172-252:688203:688670 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-172-252:688203:688670 [3] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-172-252:688203:688670 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-147:618587:619089 [3] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-172-147:618587:619089 [3] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-172-147:618587:619089 [3] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-172-147:618587:619089 [3] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-172-147:618587:619089 [3] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-172-147:618587:619089 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-147:618586:619085 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-172-147:618586:619085 [2] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-172-147:618586:619085 [2] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-172-147:618586:619085 [2] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-172-147:618586:619085 [2] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-172-147:618586:619085 [2] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-172-147:618586:619085 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-252:688202:688668 [2] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-172-252:688202:688668 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-147:618589:619087 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-172-147:618589:619087 [5] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-172-147:618589:619087 [5] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-172-147:618589:619087 [5] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-172-147:618589:619087 [5] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-172-147:618589:619087 [5] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-172-147:618589:619087 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-252:688205:688669 [5] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-172-252:688205:688669 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-252:688200:688671 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+ip-26-0-172-147:618591:619104 [7] NCCL INFO Using network Libfabric
+ip-26-0-172-252:688207:688682 [7] NCCL INFO Using network Libfabric
+ip-26-0-172-252:688207:688682 [7] NCCL INFO comm 0xb523520 rank 1 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0x89a7e73c0af2d52c - Init START
+ip-26-0-172-147:618591:619104 [7] NCCL INFO comm 0x143ed100 rank 0 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0x89a7e73c0af2d52c - Init START
+ip-26-0-172-147:618591:619104 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-252:688207:688682 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-147:618584:619091 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-172-147:618584:619091 [0] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-172-147:618584:619091 [0] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-172-147:618584:619091 [0] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-172-147:618584:619091 [0] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-172-147:618584:619091 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-172-252:688200:688671 [0] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-172-252:688200:688671 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-147:618584:619091 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-252:688203:688670 [3] NCCL INFO Channel 00/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688203:688670 [3] NCCL INFO Channel 01/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688203:688670 [3] NCCL INFO Channel 02/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688203:688670 [3] NCCL INFO Channel 03/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688203:688670 [3] NCCL INFO Channel 00/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688203:688670 [3] NCCL INFO Channel 01/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688203:688670 [3] NCCL INFO Channel 02/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688203:688670 [3] NCCL INFO Channel 03/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618587:619089 [3] NCCL INFO Channel 00/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618587:619089 [3] NCCL INFO Channel 01/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618587:619089 [3] NCCL INFO Channel 02/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618587:619089 [3] NCCL INFO Channel 03/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618587:619089 [3] NCCL INFO Channel 00/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618587:619089 [3] NCCL INFO Channel 01/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618587:619089 [3] NCCL INFO Channel 02/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-147:618587:619089 [3] NCCL INFO Channel 03/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688202:688668 [2] NCCL INFO Channel 00/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688202:688668 [2] NCCL INFO Channel 01/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688202:688668 [2] NCCL INFO Channel 02/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688202:688668 [2] NCCL INFO Channel 03/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688202:688668 [2] NCCL INFO Channel 00/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688202:688668 [2] NCCL INFO Channel 01/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688202:688668 [2] NCCL INFO Channel 02/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688202:688668 [2] NCCL INFO Channel 03/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618586:619085 [2] NCCL INFO Channel 00/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618586:619085 [2] NCCL INFO Channel 01/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618586:619085 [2] NCCL INFO Channel 02/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618586:619085 [2] NCCL INFO Channel 03/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618586:619085 [2] NCCL INFO Channel 00/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618586:619085 [2] NCCL INFO Channel 01/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618586:619085 [2] NCCL INFO Channel 02/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-147:618586:619085 [2] NCCL INFO Channel 03/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688205:688669 [5] NCCL INFO Channel 00/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688205:688669 [5] NCCL INFO Channel 01/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688205:688669 [5] NCCL INFO Channel 02/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688205:688669 [5] NCCL INFO Channel 03/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688205:688669 [5] NCCL INFO Channel 00/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688205:688669 [5] NCCL INFO Channel 01/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688205:688669 [5] NCCL INFO Channel 02/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688205:688669 [5] NCCL INFO Channel 03/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618589:619087 [5] NCCL INFO Channel 00/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618589:619087 [5] NCCL INFO Channel 01/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618589:619087 [5] NCCL INFO Channel 02/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618589:619087 [5] NCCL INFO Channel 03/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618589:619087 [5] NCCL INFO Channel 00/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618589:619087 [5] NCCL INFO Channel 01/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618589:619087 [5] NCCL INFO Channel 02/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-147:618589:619087 [5] NCCL INFO Channel 03/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688207:688682 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-172-252:688200:688671 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688200:688671 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688200:688671 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688200:688671 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688200:688671 [0] NCCL INFO Channel 00/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688200:688671 [0] NCCL INFO Channel 01/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688200:688671 [0] NCCL INFO Channel 02/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688200:688671 [0] NCCL INFO Channel 03/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618591:619104 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-172-147:618591:619104 [7] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-172-147:618591:619104 [7] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-172-147:618591:619104 [7] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-172-252:688207:688682 [7] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-172-252:688207:688682 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-147:618591:619104 [7] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-172-147:618591:619104 [7] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-172-147:618591:619104 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-147:618584:619091 [0] NCCL INFO Channel 00/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618584:619091 [0] NCCL INFO Channel 01/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618584:619091 [0] NCCL INFO Channel 02/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618584:619091 [0] NCCL INFO Channel 03/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618584:619091 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618584:619091 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618584:619091 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-147:618584:619091 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688206:688662 [6] NCCL INFO Connected all rings
+ip-26-0-172-252:688206:688662 [6] NCCL INFO Connected all trees
+ip-26-0-172-252:688206:688662 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-252:688206:688662 [6] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-172-252:688206:688662 [6] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-252:688206:688662 [6] NCCL INFO comm 0xc800b60 rank 1 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0xbb6950b11ee120f2 - Init COMPLETE
+ip-26-0-172-252:688206:688687 [6] NCCL INFO Channel 02/1 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA/Shared
+ip-26-0-172-252:688206:688687 [6] NCCL INFO Channel 03/1 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA/Shared
+ip-26-0-172-147:618590:619077 [6] NCCL INFO Connected all rings
+ip-26-0-172-147:618590:619077 [6] NCCL INFO Connected all trees
+ip-26-0-172-147:618590:619077 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-147:618590:619077 [6] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-172-147:618590:619077 [6] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-252:688207:688682 [7] NCCL INFO Channel 00/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688207:688682 [7] NCCL INFO Channel 01/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688207:688682 [7] NCCL INFO Channel 02/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688207:688682 [7] NCCL INFO Channel 03/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688207:688682 [7] NCCL INFO Channel 00/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688207:688682 [7] NCCL INFO Channel 01/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688207:688682 [7] NCCL INFO Channel 02/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688207:688682 [7] NCCL INFO Channel 03/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618591:619104 [7] NCCL INFO Channel 00/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618591:619104 [7] NCCL INFO Channel 01/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618591:619104 [7] NCCL INFO Channel 02/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618591:619104 [7] NCCL INFO Channel 03/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618591:619104 [7] NCCL INFO Channel 00/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618591:619104 [7] NCCL INFO Channel 01/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618591:619104 [7] NCCL INFO Channel 02/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618591:619104 [7] NCCL INFO Channel 03/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-147:618590:619077 [6] NCCL INFO comm 0x151d31f0 rank 0 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0xbb6950b11ee120f2 - Init COMPLETE
+ip-26-0-172-147:618590:619109 [6] NCCL INFO Channel 02/1 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA/Shared
+ip-26-0-172-147:618590:619109 [6] NCCL INFO Channel 03/1 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA/Shared
+ip-26-0-172-252:688204:688665 [4] NCCL INFO Connected all rings
+ip-26-0-172-252:688204:688665 [4] NCCL INFO Connected all trees
+ip-26-0-172-252:688204:688665 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-252:688204:688665 [4] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-172-252:688204:688665 [4] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-147:618588:619081 [4] NCCL INFO Connected all rings
+ip-26-0-172-147:618588:619081 [4] NCCL INFO Connected all trees
+ip-26-0-172-147:618588:619081 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-147:618588:619081 [4] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-172-147:618588:619081 [4] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-252:688204:688665 [4] NCCL INFO comm 0xb918890 rank 1 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0xe35f0d1ddc70c8ac - Init COMPLETE
+ip-26-0-172-252:688204:688688 [4] NCCL INFO Channel 02/1 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA/Shared
+ip-26-0-172-252:688204:688688 [4] NCCL INFO Channel 03/1 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA/Shared
+ip-26-0-172-252:688201:688664 [1] NCCL INFO Connected all rings
+ip-26-0-172-252:688201:688664 [1] NCCL INFO Connected all trees
+ip-26-0-172-252:688201:688664 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-252:688201:688664 [1] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-172-252:688201:688664 [1] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-147:618588:619081 [4] NCCL INFO comm 0x138735b0 rank 0 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0xe35f0d1ddc70c8ac - Init COMPLETE
+ip-26-0-172-147:618588:619110 [4] NCCL INFO Channel 02/1 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA/Shared
+ip-26-0-172-147:618588:619110 [4] NCCL INFO Channel 03/1 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA/Shared
+ip-26-0-172-147:618585:619080 [1] NCCL INFO Connected all rings
+ip-26-0-172-147:618585:619080 [1] NCCL INFO Connected all trees
+ip-26-0-172-147:618585:619080 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-147:618585:619080 [1] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-172-147:618585:619080 [1] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-252:688201:688664 [1] NCCL INFO comm 0xb6d73d0 rank 1 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0xd3620b1555668162 - Init COMPLETE
+ip-26-0-172-252:688201:688689 [1] NCCL INFO Channel 02/1 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA/Shared
+ip-26-0-172-252:688201:688689 [1] NCCL INFO Channel 03/1 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA/Shared
+ip-26-0-172-147:618585:619080 [1] NCCL INFO comm 0x15035980 rank 0 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0xd3620b1555668162 - Init COMPLETE
+ip-26-0-172-147:618585:619111 [1] NCCL INFO Channel 02/1 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA/Shared
+ip-26-0-172-147:618585:619111 [1] NCCL INFO Channel 03/1 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA/Shared
+ip-26-0-172-147:618587:619089 [3] NCCL INFO Connected all rings
+ip-26-0-172-147:618587:619089 [3] NCCL INFO Connected all trees
+ip-26-0-172-147:618587:619089 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-147:618587:619089 [3] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-172-147:618587:619089 [3] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-252:688203:688670 [3] NCCL INFO Connected all rings
+ip-26-0-172-252:688203:688670 [3] NCCL INFO Connected all trees
+ip-26-0-172-252:688203:688670 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-252:688203:688670 [3] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-172-252:688203:688670 [3] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-147:618586:619085 [2] NCCL INFO Connected all rings
+ip-26-0-172-147:618586:619085 [2] NCCL INFO Connected all trees
+ip-26-0-172-147:618586:619085 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-147:618586:619085 [2] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-172-147:618586:619085 [2] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-252:688202:688668 [2] NCCL INFO Connected all rings
+ip-26-0-172-252:688202:688668 [2] NCCL INFO Connected all trees
+ip-26-0-172-252:688202:688668 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-252:688202:688668 [2] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-172-147:618587:619089 [3] NCCL INFO comm 0x13bb64b0 rank 0 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0x599ed5a91849038c - Init COMPLETE
+ip-26-0-172-252:688202:688668 [2] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-252:688203:688670 [3] NCCL INFO comm 0xac123f0 rank 1 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0x599ed5a91849038c - Init COMPLETE
+ip-26-0-172-147:618587:619112 [3] NCCL INFO Channel 02/1 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA/Shared
+ip-26-0-172-252:688203:688690 [3] NCCL INFO Channel 02/1 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA/Shared
+ip-26-0-172-147:618587:619112 [3] NCCL INFO Channel 03/1 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA/Shared
+ip-26-0-172-252:688203:688690 [3] NCCL INFO Channel 03/1 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA/Shared
+ip-26-0-172-147:618589:619087 [5] NCCL INFO Connected all rings
+ip-26-0-172-147:618589:619087 [5] NCCL INFO Connected all trees
+ip-26-0-172-147:618589:619087 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-147:618589:619087 [5] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-172-147:618589:619087 [5] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-147:618586:619085 [2] NCCL INFO comm 0x13b675c0 rank 0 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0x94c2144e0212d222 - Init COMPLETE
+ip-26-0-172-147:618586:619113 [2] NCCL INFO Channel 02/1 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA/Shared
+ip-26-0-172-147:618586:619113 [2] NCCL INFO Channel 03/1 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA/Shared
+ip-26-0-172-252:688205:688669 [5] NCCL INFO Connected all rings
+ip-26-0-172-252:688205:688669 [5] NCCL INFO Connected all trees
+ip-26-0-172-252:688205:688669 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-252:688205:688669 [5] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-172-252:688205:688669 [5] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-252:688202:688668 [2] NCCL INFO comm 0xb638870 rank 1 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0x94c2144e0212d222 - Init COMPLETE
+ip-26-0-172-252:688202:688691 [2] NCCL INFO Channel 02/1 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA/Shared
+ip-26-0-172-252:688202:688691 [2] NCCL INFO Channel 03/1 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA/Shared
+ip-26-0-172-147:618589:619087 [5] NCCL INFO comm 0x14847a80 rank 0 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0xcd9768f6a4298843 - Init COMPLETE
+ip-26-0-172-147:618589:619114 [5] NCCL INFO Channel 02/1 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA/Shared
+ip-26-0-172-147:618589:619114 [5] NCCL INFO Channel 03/1 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA/Shared
+ip-26-0-172-252:688205:688669 [5] NCCL INFO comm 0xb282390 rank 1 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0xcd9768f6a4298843 - Init COMPLETE
+ip-26-0-172-252:688205:688692 [5] NCCL INFO Channel 02/1 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA/Shared
+ip-26-0-172-252:688205:688692 [5] NCCL INFO Channel 03/1 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA/Shared
+ip-26-0-172-252:688200:688671 [0] NCCL INFO Connected all rings
+ip-26-0-172-252:688200:688671 [0] NCCL INFO Connected all trees
+ip-26-0-172-252:688200:688671 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-252:688200:688671 [0] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-172-252:688200:688671 [0] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-147:618584:619091 [0] NCCL INFO Connected all rings
+ip-26-0-172-147:618584:619091 [0] NCCL INFO Connected all trees
+ip-26-0-172-147:618584:619091 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-147:618584:619091 [0] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-172-147:618584:619091 [0] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-252:688200:688671 [0] NCCL INFO comm 0xb7aad80 rank 1 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0xc76d1c86b2edff9b - Init COMPLETE
+ip-26-0-172-252:688200:688693 [0] NCCL INFO Channel 02/1 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA/Shared
+ip-26-0-172-252:688200:688693 [0] NCCL INFO Channel 03/1 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA/Shared
+ip-26-0-172-147:618584:619091 [0] NCCL INFO comm 0x14b26940 rank 0 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0xc76d1c86b2edff9b - Init COMPLETE
+ip-26-0-172-147:618584:619115 [0] NCCL INFO Channel 02/1 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA/Shared
+ip-26-0-172-147:618584:619115 [0] NCCL INFO Channel 03/1 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA/Shared
+ip-26-0-172-252:688207:688682 [7] NCCL INFO Connected all rings
+ip-26-0-172-252:688207:688682 [7] NCCL INFO Connected all trees
+ip-26-0-172-252:688207:688682 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-252:688207:688682 [7] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-172-252:688207:688682 [7] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-147:618590:619124 [6] NCCL INFO Channel 02/1 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA/Shared
+ip-26-0-172-147:618590:619124 [6] NCCL INFO Channel 03/1 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA/Shared
+ip-26-0-172-252:688207:688682 [7] NCCL INFO comm 0xb523520 rank 1 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0x89a7e73c0af2d52c - Init COMPLETE
+ip-26-0-172-252:688207:688695 [7] NCCL INFO Channel 02/1 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA/Shared
+ip-26-0-172-252:688207:688695 [7] NCCL INFO Channel 03/1 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA/Shared
+ip-26-0-172-147:618591:619104 [7] NCCL INFO Connected all rings
+ip-26-0-172-147:618591:619104 [7] NCCL INFO Connected all trees
+ip-26-0-172-147:618591:619104 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-147:618591:619104 [7] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-172-147:618591:619104 [7] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-147:618591:619104 [7] NCCL INFO comm 0x143ed100 rank 0 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0x89a7e73c0af2d52c - Init COMPLETE
+ip-26-0-172-147:618591:619125 [7] NCCL INFO Channel 02/1 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA/Shared
+ip-26-0-172-147:618591:619125 [7] NCCL INFO Channel 03/1 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA/Shared
+ip-26-0-172-147:618588:619134 [4] NCCL INFO Channel 02/1 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA/Shared
+ip-26-0-172-147:618588:619134 [4] NCCL INFO Channel 03/1 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA/Shared
+ip-26-0-172-147:618585:619143 [1] NCCL INFO Channel 02/1 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA/Shared
+ip-26-0-172-147:618585:619143 [1] NCCL INFO Channel 03/1 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA/Shared
+ip-26-0-172-147:618587:619152 [3] NCCL INFO Channel 02/1 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA/Shared
+ip-26-0-172-147:618587:619152 [3] NCCL INFO Channel 03/1 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA/Shared
+ip-26-0-172-147:618589:619169 [5] NCCL INFO Channel 02/1 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA/Shared
+ip-26-0-172-147:618589:619169 [5] NCCL INFO Channel 03/1 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA/Shared
+ip-26-0-172-147:618586:619170 [2] NCCL INFO Channel 02/1 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA/Shared
+ip-26-0-172-147:618586:619170 [2] NCCL INFO Channel 03/1 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA/Shared
+ip-26-0-172-147:618584:619179 [0] NCCL INFO Channel 02/1 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA/Shared
+ip-26-0-172-147:618584:619179 [0] NCCL INFO Channel 03/1 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA/Shared
+ip-26-0-172-147:618591:619188 [7] NCCL INFO Channel 02/1 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA/Shared
+ip-26-0-172-147:618591:619188 [7] NCCL INFO Channel 03/1 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA/Shared
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+ip-26-0-172-252:688206:688698 [6] NCCL INFO Using network Libfabric
+ip-26-0-172-57:942617:943075 [6] NCCL INFO Using network Libfabric
+ip-26-0-172-57:942617:943075 [6] NCCL INFO comm 0xbbb29f0 rank 1 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0xc569ed238b729028 - Init START
+ip-26-0-172-252:688206:688698 [6] NCCL INFO comm 0x15476080 rank 0 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0xc569ed238b729028 - Init START
+ip-26-0-172-57:942617:943075 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-252:688206:688698 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-57:942617:943075 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+ip-26-0-172-252:688204:688700 [4] NCCL INFO Using network Libfabric
+ip-26-0-172-57:942615:943076 [4] NCCL INFO Using network Libfabric
+ip-26-0-172-57:942615:943076 [4] NCCL INFO comm 0xc54bfb0 rank 1 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0xccb518c3d93de71d - Init START
+ip-26-0-172-252:688204:688700 [4] NCCL INFO comm 0x1451d9b0 rank 0 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0xccb518c3d93de71d - Init START
+ip-26-0-172-252:688204:688700 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-57:942615:943076 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-252:688206:688698 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-172-252:688206:688698 [6] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-172-252:688206:688698 [6] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-172-252:688206:688698 [6] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-172-252:688206:688698 [6] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-172-252:688206:688698 [6] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-172-252:688206:688698 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-57:942617:943075 [6] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-172-57:942617:943075 [6] NCCL INFO P2P Chunksize set to 131072
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+ip-26-0-172-252:688201:688704 [1] NCCL INFO Using network Libfabric
+ip-26-0-172-57:942612:943079 [1] NCCL INFO Using network Libfabric
+ip-26-0-172-57:942612:943079 [1] NCCL INFO comm 0xb143f20 rank 1 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0x8321d0b002906d74 - Init START
+ip-26-0-172-252:688201:688704 [1] NCCL INFO comm 0x143c3ec0 rank 0 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0x8321d0b002906d74 - Init START
+ip-26-0-172-57:942612:943079 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-252:688201:688704 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-252:688206:688698 [6] NCCL INFO Channel 00/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688206:688698 [6] NCCL INFO Channel 01/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688206:688698 [6] NCCL INFO Channel 02/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688206:688698 [6] NCCL INFO Channel 03/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688206:688698 [6] NCCL INFO Channel 00/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688206:688698 [6] NCCL INFO Channel 01/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688206:688698 [6] NCCL INFO Channel 02/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-252:688206:688698 [6] NCCL INFO Channel 03/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942617:943075 [6] NCCL INFO Channel 00/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942617:943075 [6] NCCL INFO Channel 01/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942617:943075 [6] NCCL INFO Channel 02/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942617:943075 [6] NCCL INFO Channel 03/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942617:943075 [6] NCCL INFO Channel 00/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942617:943075 [6] NCCL INFO Channel 01/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942617:943075 [6] NCCL INFO Channel 02/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942617:943075 [6] NCCL INFO Channel 03/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+ip-26-0-172-252:688203:688707 [3] NCCL INFO Using network Libfabric
+ip-26-0-172-57:942614:943080 [3] NCCL INFO Using network Libfabric
+ip-26-0-172-57:942614:943080 [3] NCCL INFO comm 0xadd7480 rank 1 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0x8ac20c8c94c0ca03 - Init START
+ip-26-0-172-252:688203:688707 [3] NCCL INFO comm 0x1381c3c0 rank 0 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0x8ac20c8c94c0ca03 - Init START
+ip-26-0-172-252:688203:688707 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-57:942614:943080 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+ip-26-0-172-252:688202:688709 [2] NCCL INFO Using network Libfabric
+ip-26-0-172-252:688202:688709 [2] NCCL INFO comm 0x142ada40 rank 0 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0xab4f75a87632c1ad - Init START
+ip-26-0-172-252:688202:688709 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-57:942613:943081 [2] NCCL INFO Using network Libfabric
+ip-26-0-172-57:942613:943081 [2] NCCL INFO comm 0xc1053e0 rank 1 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0xab4f75a87632c1ad - Init START
+ip-26-0-172-57:942615:943076 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-172-57:942613:943081 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-57:942612:943079 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+ip-26-0-172-252:688205:688711 [5] NCCL INFO Using network Libfabric
+ip-26-0-172-57:942616:943082 [5] NCCL INFO Using network Libfabric
+ip-26-0-172-252:688205:688711 [5] NCCL INFO comm 0x13e89e00 rank 0 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0x4cde90429a839879 - Init START
+ip-26-0-172-252:688205:688711 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-57:942616:943082 [5] NCCL INFO comm 0xaacee10 rank 1 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0x4cde90429a839879 - Init START
+ip-26-0-172-57:942616:943082 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-252:688204:688700 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-172-252:688204:688700 [4] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-172-252:688204:688700 [4] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-172-57:942615:943076 [4] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-172-57:942615:943076 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-252:688204:688700 [4] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-172-252:688204:688700 [4] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-172-252:688204:688700 [4] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-172-252:688204:688700 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-252:688201:688704 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-172-252:688201:688704 [1] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-172-252:688201:688704 [1] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-172-252:688201:688704 [1] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-172-252:688201:688704 [1] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-172-252:688201:688704 [1] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-172-252:688201:688704 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-57:942612:943079 [1] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-172-57:942612:943079 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-57:942615:943076 [4] NCCL INFO Channel 00/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942615:943076 [4] NCCL INFO Channel 01/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942615:943076 [4] NCCL INFO Channel 02/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942615:943076 [4] NCCL INFO Channel 03/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942615:943076 [4] NCCL INFO Channel 00/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942615:943076 [4] NCCL INFO Channel 01/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942615:943076 [4] NCCL INFO Channel 02/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942615:943076 [4] NCCL INFO Channel 03/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688204:688700 [4] NCCL INFO Channel 00/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688204:688700 [4] NCCL INFO Channel 01/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688204:688700 [4] NCCL INFO Channel 02/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688204:688700 [4] NCCL INFO Channel 03/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688204:688700 [4] NCCL INFO Channel 00/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688204:688700 [4] NCCL INFO Channel 01/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688204:688700 [4] NCCL INFO Channel 02/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-252:688204:688700 [4] NCCL INFO Channel 03/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942613:943081 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-172-252:688201:688704 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688201:688704 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688201:688704 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688201:688704 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688201:688704 [1] NCCL INFO Channel 00/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688201:688704 [1] NCCL INFO Channel 01/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688201:688704 [1] NCCL INFO Channel 02/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942612:943079 [1] NCCL INFO Channel 00/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-252:688201:688704 [1] NCCL INFO Channel 03/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942612:943079 [1] NCCL INFO Channel 01/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942612:943079 [1] NCCL INFO Channel 02/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942612:943079 [1] NCCL INFO Channel 03/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942612:943079 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942612:943079 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942612:943079 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942612:943079 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942614:943080 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-172-252:688203:688707 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-172-252:688203:688707 [3] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-172-252:688203:688707 [3] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-172-252:688203:688707 [3] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-172-252:688203:688707 [3] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-172-252:688203:688707 [3] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-172-252:688203:688707 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-57:942614:943080 [3] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-172-57:942614:943080 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-252:688202:688709 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-172-252:688202:688709 [2] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-172-252:688202:688709 [2] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-172-252:688202:688709 [2] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-172-252:688202:688709 [2] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-172-252:688202:688709 [2] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-172-252:688202:688709 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-57:942613:943081 [2] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-172-57:942613:943081 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-57:942616:943082 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+ip-26-0-172-252:688200:688721 [0] NCCL INFO Using network Libfabric
+ip-26-0-172-57:942611:943092 [0] NCCL INFO Using network Libfabric
+ip-26-0-172-57:942611:943092 [0] NCCL INFO comm 0xc1ea7c0 rank 1 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0x50153b1217aaf917 - Init START
+ip-26-0-172-252:688200:688721 [0] NCCL INFO comm 0x14420f30 rank 0 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0x50153b1217aaf917 - Init START
+ip-26-0-172-252:688200:688721 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-57:942611:943092 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-252:688205:688711 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-172-252:688205:688711 [5] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-172-252:688205:688711 [5] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-172-252:688205:688711 [5] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-172-252:688205:688711 [5] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-172-252:688205:688711 [5] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-172-252:688205:688711 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-57:942616:943082 [5] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-172-57:942616:943082 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-252:688203:688707 [3] NCCL INFO Channel 00/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688203:688707 [3] NCCL INFO Channel 01/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688203:688707 [3] NCCL INFO Channel 02/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688203:688707 [3] NCCL INFO Channel 03/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688203:688707 [3] NCCL INFO Channel 00/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688203:688707 [3] NCCL INFO Channel 01/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688203:688707 [3] NCCL INFO Channel 02/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-252:688203:688707 [3] NCCL INFO Channel 03/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942614:943080 [3] NCCL INFO Channel 00/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942614:943080 [3] NCCL INFO Channel 01/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942614:943080 [3] NCCL INFO Channel 02/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942614:943080 [3] NCCL INFO Channel 03/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942614:943080 [3] NCCL INFO Channel 00/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942614:943080 [3] NCCL INFO Channel 01/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942614:943080 [3] NCCL INFO Channel 02/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942614:943080 [3] NCCL INFO Channel 03/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942613:943081 [2] NCCL INFO Channel 00/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942613:943081 [2] NCCL INFO Channel 01/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942613:943081 [2] NCCL INFO Channel 02/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942613:943081 [2] NCCL INFO Channel 03/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942613:943081 [2] NCCL INFO Channel 00/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942613:943081 [2] NCCL INFO Channel 01/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942613:943081 [2] NCCL INFO Channel 02/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688202:688709 [2] NCCL INFO Channel 00/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942613:943081 [2] NCCL INFO Channel 03/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688202:688709 [2] NCCL INFO Channel 01/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688202:688709 [2] NCCL INFO Channel 02/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688202:688709 [2] NCCL INFO Channel 03/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688202:688709 [2] NCCL INFO Channel 00/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688202:688709 [2] NCCL INFO Channel 01/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688202:688709 [2] NCCL INFO Channel 02/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-252:688202:688709 [2] NCCL INFO Channel 03/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942611:943092 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+ip-26-0-172-252:688207:688725 [7] NCCL INFO Using network Libfabric
+ip-26-0-172-57:942618:943095 [7] NCCL INFO Using network Libfabric
+ip-26-0-172-57:942618:943095 [7] NCCL INFO comm 0xb4401f0 rank 1 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0x7ad0d54d5fb22f18 - Init START
+ip-26-0-172-252:688207:688725 [7] NCCL INFO comm 0x13f8e930 rank 0 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0x7ad0d54d5fb22f18 - Init START
+ip-26-0-172-57:942618:943095 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-252:688207:688725 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-252:688200:688721 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-172-57:942616:943082 [5] NCCL INFO Channel 00/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688205:688711 [5] NCCL INFO Channel 00/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688205:688711 [5] NCCL INFO Channel 01/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942616:943082 [5] NCCL INFO Channel 01/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688205:688711 [5] NCCL INFO Channel 02/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942616:943082 [5] NCCL INFO Channel 02/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688205:688711 [5] NCCL INFO Channel 03/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942616:943082 [5] NCCL INFO Channel 03/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688205:688711 [5] NCCL INFO Channel 00/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942616:943082 [5] NCCL INFO Channel 00/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688205:688711 [5] NCCL INFO Channel 01/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942616:943082 [5] NCCL INFO Channel 01/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688205:688711 [5] NCCL INFO Channel 02/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942616:943082 [5] NCCL INFO Channel 02/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688205:688711 [5] NCCL INFO Channel 03/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942616:943082 [5] NCCL INFO Channel 03/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-252:688200:688721 [0] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-172-252:688200:688721 [0] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-172-252:688200:688721 [0] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-172-252:688200:688721 [0] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-172-252:688200:688721 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-172-252:688200:688721 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-57:942611:943092 [0] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-172-57:942611:943092 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-57:942611:943092 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688200:688721 [0] NCCL INFO Channel 00/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942611:943092 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688200:688721 [0] NCCL INFO Channel 01/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942611:943092 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688200:688721 [0] NCCL INFO Channel 02/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942611:943092 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688200:688721 [0] NCCL INFO Channel 03/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942611:943092 [0] NCCL INFO Channel 00/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688200:688721 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942611:943092 [0] NCCL INFO Channel 01/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688200:688721 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688200:688721 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942611:943092 [0] NCCL INFO Channel 02/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-252:688200:688721 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942611:943092 [0] NCCL INFO Channel 03/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942618:943095 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-172-252:688207:688725 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-172-252:688207:688725 [7] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-172-252:688207:688725 [7] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-172-252:688207:688725 [7] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-172-252:688207:688725 [7] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-172-252:688207:688725 [7] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-172-252:688207:688725 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-57:942618:943095 [7] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-172-57:942618:943095 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-57:942617:943075 [6] NCCL INFO Connected all rings
+ip-26-0-172-57:942617:943075 [6] NCCL INFO Connected all trees
+ip-26-0-172-57:942617:943075 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-57:942617:943075 [6] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-172-57:942617:943075 [6] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-57:942617:943075 [6] NCCL INFO comm 0xbbb29f0 rank 1 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0xc569ed238b729028 - Init COMPLETE
+ip-26-0-172-57:942617:943100 [6] NCCL INFO Channel 02/1 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA/Shared
+ip-26-0-172-57:942617:943100 [6] NCCL INFO Channel 03/1 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA/Shared
+ip-26-0-172-57:942615:943076 [4] NCCL INFO Connected all rings
+ip-26-0-172-57:942615:943076 [4] NCCL INFO Connected all trees
+ip-26-0-172-57:942615:943076 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-57:942615:943076 [4] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-172-57:942615:943076 [4] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-252:688206:688698 [6] NCCL INFO Connected all rings
+ip-26-0-172-252:688206:688698 [6] NCCL INFO Connected all trees
+ip-26-0-172-252:688206:688698 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-252:688206:688698 [6] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-172-252:688206:688698 [6] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-57:942615:943076 [4] NCCL INFO comm 0xc54bfb0 rank 1 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0xccb518c3d93de71d - Init COMPLETE
+ip-26-0-172-57:942615:943101 [4] NCCL INFO Channel 02/1 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA/Shared
+ip-26-0-172-57:942615:943101 [4] NCCL INFO Channel 03/1 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA/Shared
+ip-26-0-172-252:688206:688698 [6] NCCL INFO comm 0x15476080 rank 0 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0xc569ed238b729028 - Init COMPLETE
+ip-26-0-172-252:688206:688730 [6] NCCL INFO Channel 02/1 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA/Shared
+ip-26-0-172-252:688206:688730 [6] NCCL INFO Channel 03/1 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA/Shared
+ip-26-0-172-57:942612:943079 [1] NCCL INFO Connected all rings
+ip-26-0-172-57:942612:943079 [1] NCCL INFO Connected all trees
+ip-26-0-172-57:942612:943079 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-57:942612:943079 [1] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-172-57:942612:943079 [1] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-57:942618:943095 [7] NCCL INFO Channel 00/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942618:943095 [7] NCCL INFO Channel 01/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942618:943095 [7] NCCL INFO Channel 02/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942618:943095 [7] NCCL INFO Channel 03/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942618:943095 [7] NCCL INFO Channel 00/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942618:943095 [7] NCCL INFO Channel 01/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942618:943095 [7] NCCL INFO Channel 02/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942618:943095 [7] NCCL INFO Channel 03/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688207:688725 [7] NCCL INFO Channel 00/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688207:688725 [7] NCCL INFO Channel 01/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688207:688725 [7] NCCL INFO Channel 02/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688207:688725 [7] NCCL INFO Channel 03/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688207:688725 [7] NCCL INFO Channel 00/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688207:688725 [7] NCCL INFO Channel 01/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688207:688725 [7] NCCL INFO Channel 02/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688207:688725 [7] NCCL INFO Channel 03/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-252:688204:688700 [4] NCCL INFO Connected all rings
+ip-26-0-172-252:688204:688700 [4] NCCL INFO Connected all trees
+ip-26-0-172-252:688204:688700 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-252:688204:688700 [4] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-172-252:688204:688700 [4] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-57:942612:943079 [1] NCCL INFO comm 0xb143f20 rank 1 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0x8321d0b002906d74 - Init COMPLETE
+ip-26-0-172-57:942612:943102 [1] NCCL INFO Channel 02/1 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA/Shared
+ip-26-0-172-57:942612:943102 [1] NCCL INFO Channel 03/1 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA/Shared
+ip-26-0-172-252:688204:688700 [4] NCCL INFO comm 0x1451d9b0 rank 0 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0xccb518c3d93de71d - Init COMPLETE
+ip-26-0-172-252:688201:688704 [1] NCCL INFO Connected all rings
+ip-26-0-172-252:688201:688704 [1] NCCL INFO Connected all trees
+ip-26-0-172-252:688201:688704 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-252:688201:688704 [1] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-172-252:688201:688704 [1] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-252:688204:688731 [4] NCCL INFO Channel 02/1 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA/Shared
+ip-26-0-172-252:688204:688731 [4] NCCL INFO Channel 03/1 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA/Shared
+ip-26-0-172-252:688201:688704 [1] NCCL INFO comm 0x143c3ec0 rank 0 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0x8321d0b002906d74 - Init COMPLETE
+ip-26-0-172-252:688201:688732 [1] NCCL INFO Channel 02/1 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA/Shared
+ip-26-0-172-252:688201:688732 [1] NCCL INFO Channel 03/1 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA/Shared
+ip-26-0-172-57:942613:943081 [2] NCCL INFO Connected all rings
+ip-26-0-172-57:942613:943081 [2] NCCL INFO Connected all trees
+ip-26-0-172-57:942613:943081 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-57:942613:943081 [2] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-172-57:942613:943081 [2] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-252:688203:688707 [3] NCCL INFO Connected all rings
+ip-26-0-172-252:688203:688707 [3] NCCL INFO Connected all trees
+ip-26-0-172-252:688203:688707 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-252:688202:688709 [2] NCCL INFO Connected all rings
+ip-26-0-172-252:688202:688709 [2] NCCL INFO Connected all trees
+ip-26-0-172-252:688202:688709 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-252:688203:688707 [3] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-172-252:688203:688707 [3] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-252:688202:688709 [2] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-172-252:688202:688709 [2] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-57:942614:943080 [3] NCCL INFO Connected all rings
+ip-26-0-172-57:942614:943080 [3] NCCL INFO Connected all trees
+ip-26-0-172-57:942614:943080 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-57:942614:943080 [3] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-172-57:942614:943080 [3] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-57:942616:943082 [5] NCCL INFO Connected all rings
+ip-26-0-172-57:942616:943082 [5] NCCL INFO Connected all trees
+ip-26-0-172-57:942616:943082 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-57:942616:943082 [5] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-172-57:942616:943082 [5] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-57:942613:943081 [2] NCCL INFO comm 0xc1053e0 rank 1 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0xab4f75a87632c1ad - Init COMPLETE
+ip-26-0-172-57:942613:943103 [2] NCCL INFO Channel 02/1 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA/Shared
+ip-26-0-172-57:942613:943103 [2] NCCL INFO Channel 03/1 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA/Shared
+ip-26-0-172-252:688205:688711 [5] NCCL INFO Connected all rings
+ip-26-0-172-252:688205:688711 [5] NCCL INFO Connected all trees
+ip-26-0-172-252:688205:688711 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-252:688205:688711 [5] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-172-252:688205:688711 [5] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-57:942614:943080 [3] NCCL INFO comm 0xadd7480 rank 1 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0x8ac20c8c94c0ca03 - Init COMPLETE
+ip-26-0-172-252:688203:688707 [3] NCCL INFO comm 0x1381c3c0 rank 0 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0x8ac20c8c94c0ca03 - Init COMPLETE
+ip-26-0-172-252:688202:688709 [2] NCCL INFO comm 0x142ada40 rank 0 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0xab4f75a87632c1ad - Init COMPLETE
+ip-26-0-172-57:942614:943104 [3] NCCL INFO Channel 02/1 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA/Shared
+ip-26-0-172-252:688203:688733 [3] NCCL INFO Channel 02/1 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA/Shared
+ip-26-0-172-57:942614:943104 [3] NCCL INFO Channel 03/1 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA/Shared
+ip-26-0-172-252:688203:688733 [3] NCCL INFO Channel 03/1 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA/Shared
+ip-26-0-172-252:688202:688734 [2] NCCL INFO Channel 02/1 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA/Shared
+ip-26-0-172-252:688202:688734 [2] NCCL INFO Channel 03/1 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA/Shared
+ip-26-0-172-57:942616:943082 [5] NCCL INFO comm 0xaacee10 rank 1 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0x4cde90429a839879 - Init COMPLETE
+ip-26-0-172-57:942616:943105 [5] NCCL INFO Channel 02/1 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA/Shared
+ip-26-0-172-57:942616:943105 [5] NCCL INFO Channel 03/1 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA/Shared
+ip-26-0-172-252:688205:688711 [5] NCCL INFO comm 0x13e89e00 rank 0 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0x4cde90429a839879 - Init COMPLETE
+ip-26-0-172-252:688205:688735 [5] NCCL INFO Channel 02/1 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA/Shared
+ip-26-0-172-252:688205:688735 [5] NCCL INFO Channel 03/1 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA/Shared
+ip-26-0-172-252:688200:688721 [0] NCCL INFO Connected all rings
+ip-26-0-172-252:688200:688721 [0] NCCL INFO Connected all trees
+ip-26-0-172-252:688200:688721 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-252:688200:688721 [0] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-172-252:688200:688721 [0] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-57:942611:943092 [0] NCCL INFO Connected all rings
+ip-26-0-172-57:942611:943092 [0] NCCL INFO Connected all trees
+ip-26-0-172-57:942611:943092 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-57:942611:943092 [0] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-172-57:942611:943092 [0] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-252:688200:688721 [0] NCCL INFO comm 0x14420f30 rank 0 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0x50153b1217aaf917 - Init COMPLETE
+ip-26-0-172-252:688200:688736 [0] NCCL INFO Channel 02/1 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA/Shared
+ip-26-0-172-252:688200:688736 [0] NCCL INFO Channel 03/1 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA/Shared
+ip-26-0-172-57:942611:943092 [0] NCCL INFO comm 0xc1ea7c0 rank 1 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0x50153b1217aaf917 - Init COMPLETE
+ip-26-0-172-57:942611:943106 [0] NCCL INFO Channel 02/1 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA/Shared
+ip-26-0-172-57:942611:943106 [0] NCCL INFO Channel 03/1 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA/Shared
+ip-26-0-172-57:942618:943095 [7] NCCL INFO Connected all rings
+ip-26-0-172-57:942618:943095 [7] NCCL INFO Connected all trees
+ip-26-0-172-57:942618:943095 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-57:942618:943095 [7] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-172-57:942618:943095 [7] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-252:688206:688745 [6] NCCL INFO Channel 02/1 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA/Shared
+ip-26-0-172-252:688206:688745 [6] NCCL INFO Channel 03/1 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA/Shared
+ip-26-0-172-57:942618:943095 [7] NCCL INFO comm 0xb4401f0 rank 1 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0x7ad0d54d5fb22f18 - Init COMPLETE
+ip-26-0-172-57:942618:943107 [7] NCCL INFO Channel 02/1 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA/Shared
+ip-26-0-172-57:942618:943107 [7] NCCL INFO Channel 03/1 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA/Shared
+ip-26-0-172-252:688207:688725 [7] NCCL INFO Connected all rings
+ip-26-0-172-252:688207:688725 [7] NCCL INFO Connected all trees
+ip-26-0-172-252:688207:688725 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-252:688207:688725 [7] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-172-252:688207:688725 [7] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-252:688207:688725 [7] NCCL INFO comm 0x13f8e930 rank 0 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0x7ad0d54d5fb22f18 - Init COMPLETE
+ip-26-0-172-252:688207:688746 [7] NCCL INFO Channel 02/1 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA/Shared
+ip-26-0-172-252:688207:688746 [7] NCCL INFO Channel 03/1 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA/Shared
+ip-26-0-172-252:688204:688756 [4] NCCL INFO Channel 02/1 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA/Shared
+ip-26-0-172-252:688204:688756 [4] NCCL INFO Channel 03/1 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA/Shared
+ip-26-0-172-252:688201:688765 [1] NCCL INFO Channel 02/1 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA/Shared
+ip-26-0-172-252:688201:688765 [1] NCCL INFO Channel 03/1 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA/Shared
+ip-26-0-172-252:688203:688774 [3] NCCL INFO Channel 02/1 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA/Shared
+ip-26-0-172-252:688203:688774 [3] NCCL INFO Channel 03/1 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA/Shared
+ip-26-0-172-252:688205:688783 [5] NCCL INFO Channel 02/1 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA/Shared
+ip-26-0-172-252:688205:688783 [5] NCCL INFO Channel 03/1 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA/Shared
+ip-26-0-172-252:688200:688792 [0] NCCL INFO Channel 02/1 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA/Shared
+ip-26-0-172-252:688200:688792 [0] NCCL INFO Channel 03/1 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA/Shared
+ip-26-0-172-252:688202:688801 [2] NCCL INFO Channel 02/1 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA/Shared
+ip-26-0-172-252:688202:688801 [2] NCCL INFO Channel 03/1 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA/Shared
+ip-26-0-172-252:688207:688810 [7] NCCL INFO Channel 02/1 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA/Shared
+ip-26-0-172-252:688207:688810 [7] NCCL INFO Channel 03/1 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA/Shared
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+ip-26-0-172-57:942617:943110 [6] NCCL INFO Using network Libfabric
+ip-26-0-172-73:926600:927060 [6] NCCL INFO Using network Libfabric
+ip-26-0-172-73:926600:927060 [6] NCCL INFO comm 0xc41cce0 rank 1 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0x7956c28ea09ccafe - Init START
+ip-26-0-172-57:942617:943110 [6] NCCL INFO comm 0x147bc6c0 rank 0 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0x7956c28ea09ccafe - Init START
+ip-26-0-172-73:926600:927060 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-57:942617:943110 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+ip-26-0-172-57:942615:943112 [4] NCCL INFO Using network Libfabric
+ip-26-0-172-73:926598:927061 [4] NCCL INFO Using network Libfabric
+ip-26-0-172-73:926598:927061 [4] NCCL INFO comm 0xbd5d850 rank 1 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0xaf906d987317197a - Init START
+ip-26-0-172-57:942615:943112 [4] NCCL INFO comm 0x15154940 rank 0 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0xaf906d987317197a - Init START
+ip-26-0-172-73:926598:927061 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-57:942615:943112 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-73:926600:927060 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+ip-26-0-172-57:942612:943114 [1] NCCL INFO Using network Libfabric
+ip-26-0-172-73:926595:927062 [1] NCCL INFO Using network Libfabric
+ip-26-0-172-73:926595:927062 [1] NCCL INFO comm 0xa916c70 rank 1 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0x8356292740d1f9f7 - Init START
+ip-26-0-172-73:926595:927062 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-57:942612:943114 [1] NCCL INFO comm 0x13bb07f0 rank 0 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0x8356292740d1f9f7 - Init START
+ip-26-0-172-57:942612:943114 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-57:942617:943110 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-172-57:942617:943110 [6] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-172-57:942617:943110 [6] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-172-57:942617:943110 [6] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-172-57:942617:943110 [6] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-172-73:926600:927060 [6] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-172-73:926600:927060 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-57:942617:943110 [6] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-172-57:942617:943110 [6] NCCL INFO P2P Chunksize set to 131072
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+ip-26-0-172-57:942613:943118 [2] NCCL INFO Using network Libfabric
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+ip-26-0-172-73:926596:927065 [2] NCCL INFO Using network Libfabric
+ip-26-0-172-73:926596:927065 [2] NCCL INFO comm 0xb265010 rank 1 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0x13a84d53f90c5817 - Init START
+ip-26-0-172-57:942613:943118 [2] NCCL INFO comm 0x14b70a00 rank 0 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0x13a84d53f90c5817 - Init START
+ip-26-0-172-57:942613:943118 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-73:926596:927065 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-57:942614:943120 [3] NCCL INFO Using network Libfabric
+ip-26-0-172-73:926597:927066 [3] NCCL INFO Using network Libfabric
+ip-26-0-172-73:926597:927066 [3] NCCL INFO comm 0xb0f67c0 rank 1 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0x50d5179981d932a3 - Init START
+ip-26-0-172-57:942614:943120 [3] NCCL INFO comm 0x139e09b0 rank 0 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0x50d5179981d932a3 - Init START
+ip-26-0-172-73:926597:927066 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-57:942614:943120 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-73:926598:927061 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+ip-26-0-172-57:942616:943122 [5] NCCL INFO Using network Libfabric
+ip-26-0-172-73:926599:927067 [5] NCCL INFO Using network Libfabric
+ip-26-0-172-73:926599:927067 [5] NCCL INFO comm 0xc410d20 rank 1 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0xe4118250b310bcbb - Init START
+ip-26-0-172-57:942616:943122 [5] NCCL INFO comm 0x136d7f80 rank 0 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0xe4118250b310bcbb - Init START
+ip-26-0-172-73:926599:927067 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-57:942616:943122 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-57:942617:943110 [6] NCCL INFO Channel 00/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942617:943110 [6] NCCL INFO Channel 01/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942617:943110 [6] NCCL INFO Channel 02/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942617:943110 [6] NCCL INFO Channel 03/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926600:927060 [6] NCCL INFO Channel 00/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942617:943110 [6] NCCL INFO Channel 00/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926600:927060 [6] NCCL INFO Channel 01/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942617:943110 [6] NCCL INFO Channel 01/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926600:927060 [6] NCCL INFO Channel 02/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942617:943110 [6] NCCL INFO Channel 02/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942617:943110 [6] NCCL INFO Channel 03/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926600:927060 [6] NCCL INFO Channel 03/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926600:927060 [6] NCCL INFO Channel 00/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926600:927060 [6] NCCL INFO Channel 01/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926600:927060 [6] NCCL INFO Channel 02/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926600:927060 [6] NCCL INFO Channel 03/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-57:942615:943112 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-172-57:942615:943112 [4] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-172-57:942615:943112 [4] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-172-57:942615:943112 [4] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-172-57:942615:943112 [4] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-172-57:942615:943112 [4] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-172-57:942615:943112 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-73:926598:927061 [4] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-172-73:926598:927061 [4] NCCL INFO P2P Chunksize set to 131072
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+ip-26-0-172-57:942611:943126 [0] NCCL INFO Using network Libfabric
+ip-26-0-172-73:926594:927071 [0] NCCL INFO Using network Libfabric
+ip-26-0-172-73:926594:927071 [0] NCCL INFO comm 0xb8281b0 rank 1 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0xa8fffdd7955f55da - Init START
+ip-26-0-172-57:942611:943126 [0] NCCL INFO comm 0x14df0740 rank 0 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0xa8fffdd7955f55da - Init START
+ip-26-0-172-73:926594:927071 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-57:942611:943126 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-57:942615:943112 [4] NCCL INFO Channel 00/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942615:943112 [4] NCCL INFO Channel 01/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942615:943112 [4] NCCL INFO Channel 02/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942615:943112 [4] NCCL INFO Channel 03/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942615:943112 [4] NCCL INFO Channel 00/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942615:943112 [4] NCCL INFO Channel 01/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942615:943112 [4] NCCL INFO Channel 02/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-57:942615:943112 [4] NCCL INFO Channel 03/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926598:927061 [4] NCCL INFO Channel 00/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926598:927061 [4] NCCL INFO Channel 01/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926598:927061 [4] NCCL INFO Channel 02/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926598:927061 [4] NCCL INFO Channel 03/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926598:927061 [4] NCCL INFO Channel 00/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926598:927061 [4] NCCL INFO Channel 01/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926598:927061 [4] NCCL INFO Channel 02/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926598:927061 [4] NCCL INFO Channel 03/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926595:927062 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-172-73:926597:927066 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-172-73:926596:927065 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-172-57:942612:943114 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-172-57:942612:943114 [1] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-172-57:942612:943114 [1] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-172-57:942612:943114 [1] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-172-57:942612:943114 [1] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-172-57:942612:943114 [1] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-172-57:942612:943114 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-73:926595:927062 [1] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-172-73:926595:927062 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-73:926599:927067 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-172-57:942614:943120 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-172-57:942614:943120 [3] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-172-57:942614:943120 [3] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-172-57:942614:943120 [3] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-172-57:942614:943120 [3] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-172-57:942614:943120 [3] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-172-57:942614:943120 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-73:926597:927066 [3] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-172-73:926597:927066 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-73:926594:927071 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-172-57:942616:943122 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-172-57:942616:943122 [5] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-172-57:942616:943122 [5] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-172-57:942616:943122 [5] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-172-57:942616:943122 [5] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-172-57:942616:943122 [5] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-172-57:942616:943122 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-73:926599:927067 [5] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-172-73:926599:927067 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-57:942613:943118 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-172-57:942613:943118 [2] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-172-57:942613:943118 [2] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-172-57:942613:943118 [2] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-172-57:942613:943118 [2] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-172-57:942613:943118 [2] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-172-57:942613:943118 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-73:926596:927065 [2] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-172-73:926596:927065 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-73:926595:927062 [1] NCCL INFO Channel 00/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926595:927062 [1] NCCL INFO Channel 01/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926595:927062 [1] NCCL INFO Channel 02/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926595:927062 [1] NCCL INFO Channel 03/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926595:927062 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926595:927062 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926595:927062 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942612:943114 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926595:927062 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942612:943114 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942612:943114 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942612:943114 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942612:943114 [1] NCCL INFO Channel 00/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942612:943114 [1] NCCL INFO Channel 01/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942612:943114 [1] NCCL INFO Channel 02/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942612:943114 [1] NCCL INFO Channel 03/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-57:942611:943126 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-172-57:942611:943126 [0] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-172-57:942611:943126 [0] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-172-57:942611:943126 [0] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-172-57:942611:943126 [0] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-172-57:942611:943126 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-172-57:942611:943126 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-73:926594:927071 [0] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-172-73:926594:927071 [0] NCCL INFO P2P Chunksize set to 131072
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+ip-26-0-172-57:942618:943139 [7] NCCL INFO Using network Libfabric
+ip-26-0-172-73:926601:927082 [7] NCCL INFO Using network Libfabric
+ip-26-0-172-73:926601:927082 [7] NCCL INFO comm 0xa9f6390 rank 1 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0xf5008e58e498bbee - Init START
+ip-26-0-172-57:942618:943139 [7] NCCL INFO comm 0x1404b430 rank 0 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0xf5008e58e498bbee - Init START
+ip-26-0-172-57:942618:943139 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-73:926601:927082 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-57:942614:943120 [3] NCCL INFO Channel 00/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926597:927066 [3] NCCL INFO Channel 00/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942614:943120 [3] NCCL INFO Channel 01/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942614:943120 [3] NCCL INFO Channel 02/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942614:943120 [3] NCCL INFO Channel 03/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926597:927066 [3] NCCL INFO Channel 01/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942614:943120 [3] NCCL INFO Channel 00/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942614:943120 [3] NCCL INFO Channel 01/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942614:943120 [3] NCCL INFO Channel 02/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926597:927066 [3] NCCL INFO Channel 02/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942614:943120 [3] NCCL INFO Channel 03/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926597:927066 [3] NCCL INFO Channel 03/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926597:927066 [3] NCCL INFO Channel 00/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926597:927066 [3] NCCL INFO Channel 01/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926597:927066 [3] NCCL INFO Channel 02/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926597:927066 [3] NCCL INFO Channel 03/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-57:942616:943122 [5] NCCL INFO Channel 00/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926599:927067 [5] NCCL INFO Channel 00/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942616:943122 [5] NCCL INFO Channel 01/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926599:927067 [5] NCCL INFO Channel 01/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942616:943122 [5] NCCL INFO Channel 02/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926599:927067 [5] NCCL INFO Channel 02/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942616:943122 [5] NCCL INFO Channel 03/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926599:927067 [5] NCCL INFO Channel 03/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942616:943122 [5] NCCL INFO Channel 00/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926599:927067 [5] NCCL INFO Channel 00/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942616:943122 [5] NCCL INFO Channel 01/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926599:927067 [5] NCCL INFO Channel 01/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942616:943122 [5] NCCL INFO Channel 02/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926599:927067 [5] NCCL INFO Channel 02/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942616:943122 [5] NCCL INFO Channel 03/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926599:927067 [5] NCCL INFO Channel 03/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-57:942613:943118 [2] NCCL INFO Channel 00/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942613:943118 [2] NCCL INFO Channel 01/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926596:927065 [2] NCCL INFO Channel 00/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942613:943118 [2] NCCL INFO Channel 02/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926596:927065 [2] NCCL INFO Channel 01/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942613:943118 [2] NCCL INFO Channel 03/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926596:927065 [2] NCCL INFO Channel 02/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942613:943118 [2] NCCL INFO Channel 00/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926596:927065 [2] NCCL INFO Channel 03/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942613:943118 [2] NCCL INFO Channel 01/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926596:927065 [2] NCCL INFO Channel 00/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942613:943118 [2] NCCL INFO Channel 02/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926596:927065 [2] NCCL INFO Channel 01/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-57:942613:943118 [2] NCCL INFO Channel 03/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926596:927065 [2] NCCL INFO Channel 02/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926596:927065 [2] NCCL INFO Channel 03/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926594:927071 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926594:927071 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926594:927071 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926594:927071 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926594:927071 [0] NCCL INFO Channel 00/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926594:927071 [0] NCCL INFO Channel 01/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926594:927071 [0] NCCL INFO Channel 02/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926594:927071 [0] NCCL INFO Channel 03/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942611:943126 [0] NCCL INFO Channel 00/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942611:943126 [0] NCCL INFO Channel 01/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942611:943126 [0] NCCL INFO Channel 02/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942611:943126 [0] NCCL INFO Channel 03/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942611:943126 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942611:943126 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942611:943126 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-57:942611:943126 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926601:927082 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-172-57:942618:943139 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-172-57:942618:943139 [7] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-172-57:942618:943139 [7] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-172-57:942618:943139 [7] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-172-57:942618:943139 [7] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-172-57:942618:943139 [7] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-172-57:942618:943139 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-73:926601:927082 [7] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-172-73:926601:927082 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-73:926600:927060 [6] NCCL INFO Connected all rings
+ip-26-0-172-73:926600:927060 [6] NCCL INFO Connected all trees
+ip-26-0-172-73:926600:927060 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-73:926600:927060 [6] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-172-73:926600:927060 [6] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-73:926598:927061 [4] NCCL INFO Connected all rings
+ip-26-0-172-73:926598:927061 [4] NCCL INFO Connected all trees
+ip-26-0-172-73:926598:927061 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-73:926598:927061 [4] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-172-73:926598:927061 [4] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-73:926600:927060 [6] NCCL INFO comm 0xc41cce0 rank 1 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0x7956c28ea09ccafe - Init COMPLETE
+ip-26-0-172-73:926600:927085 [6] NCCL INFO Channel 02/1 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA/Shared
+ip-26-0-172-73:926600:927085 [6] NCCL INFO Channel 03/1 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA/Shared
+ip-26-0-172-73:926598:927061 [4] NCCL INFO comm 0xbd5d850 rank 1 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0xaf906d987317197a - Init COMPLETE
+ip-26-0-172-73:926598:927086 [4] NCCL INFO Channel 02/1 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA/Shared
+ip-26-0-172-73:926598:927086 [4] NCCL INFO Channel 03/1 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA/Shared
+ip-26-0-172-57:942617:943110 [6] NCCL INFO Connected all rings
+ip-26-0-172-57:942617:943110 [6] NCCL INFO Connected all trees
+ip-26-0-172-57:942617:943110 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-57:942617:943110 [6] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-172-57:942617:943110 [6] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-57:942615:943112 [4] NCCL INFO Connected all rings
+ip-26-0-172-57:942615:943112 [4] NCCL INFO Connected all trees
+ip-26-0-172-57:942615:943112 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-57:942615:943112 [4] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-172-57:942615:943112 [4] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-57:942617:943110 [6] NCCL INFO comm 0x147bc6c0 rank 0 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0x7956c28ea09ccafe - Init COMPLETE
+ip-26-0-172-57:942617:943143 [6] NCCL INFO Channel 02/1 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA/Shared
+ip-26-0-172-57:942617:943143 [6] NCCL INFO Channel 03/1 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA/Shared
+ip-26-0-172-57:942618:943139 [7] NCCL INFO Channel 00/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942618:943139 [7] NCCL INFO Channel 01/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942618:943139 [7] NCCL INFO Channel 02/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942618:943139 [7] NCCL INFO Channel 03/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942618:943139 [7] NCCL INFO Channel 00/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942618:943139 [7] NCCL INFO Channel 01/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926601:927082 [7] NCCL INFO Channel 00/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942618:943139 [7] NCCL INFO Channel 02/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926601:927082 [7] NCCL INFO Channel 01/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942618:943139 [7] NCCL INFO Channel 03/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926601:927082 [7] NCCL INFO Channel 02/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926601:927082 [7] NCCL INFO Channel 03/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926601:927082 [7] NCCL INFO Channel 00/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926601:927082 [7] NCCL INFO Channel 01/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926601:927082 [7] NCCL INFO Channel 02/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926601:927082 [7] NCCL INFO Channel 03/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-57:942615:943112 [4] NCCL INFO comm 0x15154940 rank 0 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0xaf906d987317197a - Init COMPLETE
+ip-26-0-172-57:942615:943144 [4] NCCL INFO Channel 02/1 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA/Shared
+ip-26-0-172-57:942615:943144 [4] NCCL INFO Channel 03/1 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA/Shared
+ip-26-0-172-73:926595:927062 [1] NCCL INFO Connected all rings
+ip-26-0-172-73:926595:927062 [1] NCCL INFO Connected all trees
+ip-26-0-172-73:926595:927062 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-73:926595:927062 [1] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-172-73:926595:927062 [1] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-57:942612:943114 [1] NCCL INFO Connected all rings
+ip-26-0-172-57:942612:943114 [1] NCCL INFO Connected all trees
+ip-26-0-172-57:942612:943114 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-57:942612:943114 [1] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-172-57:942612:943114 [1] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-73:926595:927062 [1] NCCL INFO comm 0xa916c70 rank 1 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0x8356292740d1f9f7 - Init COMPLETE
+ip-26-0-172-73:926595:927087 [1] NCCL INFO Channel 02/1 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA/Shared
+ip-26-0-172-73:926595:927087 [1] NCCL INFO Channel 03/1 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA/Shared
+ip-26-0-172-57:942612:943114 [1] NCCL INFO comm 0x13bb07f0 rank 0 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0x8356292740d1f9f7 - Init COMPLETE
+ip-26-0-172-57:942612:943145 [1] NCCL INFO Channel 02/1 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA/Shared
+ip-26-0-172-57:942612:943145 [1] NCCL INFO Channel 03/1 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA/Shared
+ip-26-0-172-73:926597:927066 [3] NCCL INFO Connected all rings
+ip-26-0-172-73:926597:927066 [3] NCCL INFO Connected all trees
+ip-26-0-172-73:926597:927066 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-73:926597:927066 [3] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-172-73:926597:927066 [3] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-57:942614:943120 [3] NCCL INFO Connected all rings
+ip-26-0-172-57:942614:943120 [3] NCCL INFO Connected all trees
+ip-26-0-172-57:942614:943120 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-57:942614:943120 [3] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-172-57:942614:943120 [3] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-73:926597:927066 [3] NCCL INFO comm 0xb0f67c0 rank 1 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0x50d5179981d932a3 - Init COMPLETE
+ip-26-0-172-73:926597:927088 [3] NCCL INFO Channel 02/1 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA/Shared
+ip-26-0-172-73:926597:927088 [3] NCCL INFO Channel 03/1 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA/Shared
+ip-26-0-172-73:926599:927067 [5] NCCL INFO Connected all rings
+ip-26-0-172-73:926599:927067 [5] NCCL INFO Connected all trees
+ip-26-0-172-73:926599:927067 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-73:926599:927067 [5] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-172-73:926599:927067 [5] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-57:942614:943120 [3] NCCL INFO comm 0x139e09b0 rank 0 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0x50d5179981d932a3 - Init COMPLETE
+ip-26-0-172-57:942614:943146 [3] NCCL INFO Channel 02/1 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA/Shared
+ip-26-0-172-57:942614:943146 [3] NCCL INFO Channel 03/1 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA/Shared
+ip-26-0-172-73:926596:927065 [2] NCCL INFO Connected all rings
+ip-26-0-172-73:926596:927065 [2] NCCL INFO Connected all trees
+ip-26-0-172-73:926596:927065 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-73:926596:927065 [2] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-172-73:926596:927065 [2] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-57:942616:943122 [5] NCCL INFO Connected all rings
+ip-26-0-172-57:942616:943122 [5] NCCL INFO Connected all trees
+ip-26-0-172-57:942616:943122 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-57:942616:943122 [5] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-172-57:942616:943122 [5] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-73:926599:927067 [5] NCCL INFO comm 0xc410d20 rank 1 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0xe4118250b310bcbb - Init COMPLETE
+ip-26-0-172-73:926599:927089 [5] NCCL INFO Channel 02/1 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA/Shared
+ip-26-0-172-73:926599:927089 [5] NCCL INFO Channel 03/1 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA/Shared
+ip-26-0-172-57:942613:943118 [2] NCCL INFO Connected all rings
+ip-26-0-172-57:942613:943118 [2] NCCL INFO Connected all trees
+ip-26-0-172-57:942613:943118 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-57:942613:943118 [2] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-172-57:942613:943118 [2] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-73:926596:927065 [2] NCCL INFO comm 0xb265010 rank 1 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0x13a84d53f90c5817 - Init COMPLETE
+ip-26-0-172-73:926596:927090 [2] NCCL INFO Channel 02/1 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA/Shared
+ip-26-0-172-73:926596:927090 [2] NCCL INFO Channel 03/1 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA/Shared
+ip-26-0-172-57:942616:943122 [5] NCCL INFO comm 0x136d7f80 rank 0 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0xe4118250b310bcbb - Init COMPLETE
+ip-26-0-172-57:942616:943147 [5] NCCL INFO Channel 02/1 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA/Shared
+ip-26-0-172-57:942616:943147 [5] NCCL INFO Channel 03/1 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA/Shared
+ip-26-0-172-57:942613:943118 [2] NCCL INFO comm 0x14b70a00 rank 0 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0x13a84d53f90c5817 - Init COMPLETE
+ip-26-0-172-57:942613:943148 [2] NCCL INFO Channel 02/1 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA/Shared
+ip-26-0-172-57:942613:943148 [2] NCCL INFO Channel 03/1 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA/Shared
+ip-26-0-172-73:926594:927071 [0] NCCL INFO Connected all rings
+ip-26-0-172-73:926594:927071 [0] NCCL INFO Connected all trees
+ip-26-0-172-73:926594:927071 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-73:926594:927071 [0] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-172-73:926594:927071 [0] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-57:942611:943126 [0] NCCL INFO Connected all rings
+ip-26-0-172-57:942611:943126 [0] NCCL INFO Connected all trees
+ip-26-0-172-57:942611:943126 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-57:942611:943126 [0] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-172-57:942611:943126 [0] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-73:926594:927071 [0] NCCL INFO comm 0xb8281b0 rank 1 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0xa8fffdd7955f55da - Init COMPLETE
+ip-26-0-172-73:926594:927091 [0] NCCL INFO Channel 02/1 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA/Shared
+ip-26-0-172-73:926594:927091 [0] NCCL INFO Channel 03/1 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA/Shared
+ip-26-0-172-57:942611:943126 [0] NCCL INFO comm 0x14df0740 rank 0 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0xa8fffdd7955f55da - Init COMPLETE
+ip-26-0-172-57:942611:943149 [0] NCCL INFO Channel 02/1 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA/Shared
+ip-26-0-172-57:942611:943149 [0] NCCL INFO Channel 03/1 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA/Shared
+ip-26-0-172-57:942617:943158 [6] NCCL INFO Channel 02/1 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA/Shared
+ip-26-0-172-57:942617:943158 [6] NCCL INFO Channel 03/1 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA/Shared
+ip-26-0-172-57:942615:943167 [4] NCCL INFO Channel 02/1 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA/Shared
+ip-26-0-172-57:942615:943167 [4] NCCL INFO Channel 03/1 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA/Shared
+ip-26-0-172-73:926601:927082 [7] NCCL INFO Connected all rings
+ip-26-0-172-73:926601:927082 [7] NCCL INFO Connected all trees
+ip-26-0-172-73:926601:927082 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-73:926601:927082 [7] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-172-73:926601:927082 [7] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-73:926601:927082 [7] NCCL INFO comm 0xa9f6390 rank 1 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0xf5008e58e498bbee - Init COMPLETE
+ip-26-0-172-73:926601:927092 [7] NCCL INFO Channel 02/1 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA/Shared
+ip-26-0-172-73:926601:927092 [7] NCCL INFO Channel 03/1 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA/Shared
+ip-26-0-172-57:942618:943139 [7] NCCL INFO Connected all rings
+ip-26-0-172-57:942618:943139 [7] NCCL INFO Connected all trees
+ip-26-0-172-57:942618:943139 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-57:942618:943139 [7] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-172-57:942618:943139 [7] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-57:942612:943176 [1] NCCL INFO Channel 02/1 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA/Shared
+ip-26-0-172-57:942612:943176 [1] NCCL INFO Channel 03/1 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA/Shared
+ip-26-0-172-57:942618:943139 [7] NCCL INFO comm 0x1404b430 rank 0 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0xf5008e58e498bbee - Init COMPLETE
+ip-26-0-172-57:942618:943177 [7] NCCL INFO Channel 02/1 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA/Shared
+ip-26-0-172-57:942618:943177 [7] NCCL INFO Channel 03/1 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA/Shared
+ip-26-0-172-57:942614:943186 [3] NCCL INFO Channel 02/1 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA/Shared
+ip-26-0-172-57:942614:943186 [3] NCCL INFO Channel 03/1 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA/Shared
+ip-26-0-172-57:942616:943195 [5] NCCL INFO Channel 02/1 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA/Shared
+ip-26-0-172-57:942616:943195 [5] NCCL INFO Channel 03/1 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA/Shared
+ip-26-0-172-57:942613:943204 [2] NCCL INFO Channel 02/1 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA/Shared
+ip-26-0-172-57:942613:943204 [2] NCCL INFO Channel 03/1 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA/Shared
+ip-26-0-172-57:942611:943213 [0] NCCL INFO Channel 02/1 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA/Shared
+ip-26-0-172-57:942611:943213 [0] NCCL INFO Channel 03/1 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA/Shared
+ip-26-0-172-57:942618:943222 [7] NCCL INFO Channel 02/1 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA/Shared
+ip-26-0-172-57:942618:943222 [7] NCCL INFO Channel 03/1 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA/Shared
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+ip-26-0-172-73:926600:927095 [6] NCCL INFO Using network Libfabric
+ip-26-0-173-7:811254:811773 [6] NCCL INFO Using network Libfabric
+ip-26-0-173-7:811254:811773 [6] NCCL INFO comm 0xb59d8f0 rank 1 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0x46cdb14851677f96 - Init START
+ip-26-0-172-73:926600:927095 [6] NCCL INFO comm 0x15026b70 rank 0 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0x46cdb14851677f96 - Init START
+ip-26-0-173-7:811254:811773 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-73:926600:927095 [6] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+ip-26-0-172-73:926598:927097 [4] NCCL INFO Using network Libfabric
+ip-26-0-173-7:811252:811774 [4] NCCL INFO Using network Libfabric
+ip-26-0-173-7:811252:811774 [4] NCCL INFO comm 0xb8b25f0 rank 1 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0x2f780d840a209ef2 - Init START
+ip-26-0-172-73:926598:927097 [4] NCCL INFO comm 0x14966230 rank 0 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0x2f780d840a209ef2 - Init START
+ip-26-0-173-7:811252:811774 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-73:926598:927097 [4] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-173-7:811254:811773 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-173-7:811252:811774 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+ip-26-0-172-73:926595:927099 [1] NCCL INFO Using network Libfabric
+ip-26-0-173-7:811249:811775 [1] NCCL INFO Using network Libfabric
+ip-26-0-173-7:811249:811775 [1] NCCL INFO comm 0xb398d80 rank 1 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0x3369bb1417fe388d - Init START
+ip-26-0-172-73:926595:927099 [1] NCCL INFO comm 0x135204b0 rank 0 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0x3369bb1417fe388d - Init START
+ip-26-0-173-7:811249:811775 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-73:926595:927099 [1] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-73:926600:927095 [6] NCCL INFO Setting affinity for GPU 6 to ffffffff,ffff0000,00000000
+ip-26-0-172-73:926600:927095 [6] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-172-73:926600:927095 [6] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-173-7:811254:811773 [6] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-173-7:811254:811773 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-73:926600:927095 [6] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-172-73:926600:927095 [6] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-172-73:926600:927095 [6] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-172-73:926600:927095 [6] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-73:926598:927097 [4] NCCL INFO Setting affinity for GPU 4 to ffffffff,ffff0000,00000000
+ip-26-0-172-73:926598:927097 [4] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-172-73:926598:927097 [4] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-172-73:926598:927097 [4] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-173-7:811252:811774 [4] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-173-7:811252:811774 [4] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-73:926598:927097 [4] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-172-73:926598:927097 [4] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-172-73:926598:927097 [4] NCCL INFO P2P Chunksize set to 131072
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+ip-26-0-172-73:926597:927105 [3] NCCL INFO Using network Libfabric
+ip-26-0-173-7:811251:811780 [3] NCCL INFO Using network Libfabric
+ip-26-0-173-7:811251:811780 [3] NCCL INFO comm 0xb793130 rank 1 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0xc0be7100b59ca4f3 - Init START
+ip-26-0-173-7:811251:811780 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-73:926597:927105 [3] NCCL INFO comm 0x13cfd0f0 rank 0 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0xc0be7100b59ca4f3 - Init START
+ip-26-0-172-73:926597:927105 [3] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+ip-26-0-172-73:926599:927107 [5] NCCL INFO Using network Libfabric
+ip-26-0-173-7:811253:811781 [5] NCCL INFO Using network Libfabric
+ip-26-0-173-7:811253:811781 [5] NCCL INFO comm 0xc207e60 rank 1 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0x57be41670ad0e398 - Init START
+ip-26-0-172-73:926599:927107 [5] NCCL INFO comm 0x15016730 rank 0 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0x57be41670ad0e398 - Init START
+ip-26-0-173-7:811253:811781 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-73:926599:927107 [5] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+ip-26-0-172-73:926596:927109 [2] NCCL INFO Using network Libfabric
+ip-26-0-173-7:811250:811782 [2] NCCL INFO Using network Libfabric
+ip-26-0-173-7:811250:811782 [2] NCCL INFO comm 0xc194bc0 rank 1 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0x840b226f2838c652 - Init START
+ip-26-0-172-73:926596:927109 [2] NCCL INFO comm 0x13e6e940 rank 0 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0x840b226f2838c652 - Init START
+ip-26-0-173-7:811250:811782 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-73:926596:927109 [2] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-173-7:811254:811773 [6] NCCL INFO Channel 00/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926600:927095 [6] NCCL INFO Channel 00/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-7:811254:811773 [6] NCCL INFO Channel 01/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926600:927095 [6] NCCL INFO Channel 01/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-7:811254:811773 [6] NCCL INFO Channel 02/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926600:927095 [6] NCCL INFO Channel 02/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-7:811254:811773 [6] NCCL INFO Channel 03/0 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926600:927095 [6] NCCL INFO Channel 03/0 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-7:811254:811773 [6] NCCL INFO Channel 00/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926600:927095 [6] NCCL INFO Channel 00/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-7:811254:811773 [6] NCCL INFO Channel 01/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926600:927095 [6] NCCL INFO Channel 01/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-7:811254:811773 [6] NCCL INFO Channel 02/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-7:811254:811773 [6] NCCL INFO Channel 03/0 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926600:927095 [6] NCCL INFO Channel 02/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-172-73:926600:927095 [6] NCCL INFO Channel 03/0 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA
+ip-26-0-173-7:811252:811774 [4] NCCL INFO Channel 00/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-7:811252:811774 [4] NCCL INFO Channel 01/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-7:811252:811774 [4] NCCL INFO Channel 02/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-7:811252:811774 [4] NCCL INFO Channel 03/0 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-7:811252:811774 [4] NCCL INFO Channel 00/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-7:811252:811774 [4] NCCL INFO Channel 01/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-7:811252:811774 [4] NCCL INFO Channel 02/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-7:811252:811774 [4] NCCL INFO Channel 03/0 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926598:927097 [4] NCCL INFO Channel 00/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926598:927097 [4] NCCL INFO Channel 01/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926598:927097 [4] NCCL INFO Channel 02/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926598:927097 [4] NCCL INFO Channel 03/0 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926598:927097 [4] NCCL INFO Channel 00/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926598:927097 [4] NCCL INFO Channel 01/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926598:927097 [4] NCCL INFO Channel 02/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-172-73:926598:927097 [4] NCCL INFO Channel 03/0 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA
+ip-26-0-173-7:811249:811775 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-172-73:926595:927099 [1] NCCL INFO Setting affinity for GPU 1 to ffff,ffffffff
+ip-26-0-172-73:926595:927099 [1] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-172-73:926595:927099 [1] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-172-73:926595:927099 [1] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-173-7:811249:811775 [1] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-173-7:811249:811775 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-73:926595:927099 [1] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-172-73:926595:927099 [1] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-172-73:926595:927099 [1] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-173-7:811251:811780 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-173-7:811250:811782 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-172-73:926597:927105 [3] NCCL INFO Setting affinity for GPU 3 to ffff,ffffffff
+ip-26-0-172-73:926597:927105 [3] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-172-73:926597:927105 [3] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-172-73:926597:927105 [3] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-172-73:926597:927105 [3] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-172-73:926597:927105 [3] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-172-73:926597:927105 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-173-7:811251:811780 [3] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-173-7:811251:811780 [3] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-173-7:811253:811781 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+ip-26-0-172-73:926594:927116 [0] NCCL INFO Using network Libfabric
+ip-26-0-173-7:811248:811788 [0] NCCL INFO Using network Libfabric
+ip-26-0-173-7:811248:811788 [0] NCCL INFO comm 0xcdaab90 rank 1 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0xf44e045fbd0f3601 - Init START
+ip-26-0-172-73:926594:927116 [0] NCCL INFO comm 0x1442e1f0 rank 0 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0xf44e045fbd0f3601 - Init START
+ip-26-0-173-7:811248:811788 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-73:926594:927116 [0] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-173-7:811249:811775 [1] NCCL INFO Channel 00/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-7:811249:811775 [1] NCCL INFO Channel 01/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-7:811249:811775 [1] NCCL INFO Channel 02/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-7:811249:811775 [1] NCCL INFO Channel 03/0 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-7:811249:811775 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-7:811249:811775 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-7:811249:811775 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-173-7:811249:811775 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926596:927109 [2] NCCL INFO Setting affinity for GPU 2 to ffff,ffffffff
+ip-26-0-172-73:926595:927099 [1] NCCL INFO Channel 00/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926595:927099 [1] NCCL INFO Channel 01/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926595:927099 [1] NCCL INFO Channel 02/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926595:927099 [1] NCCL INFO Channel 03/0 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926595:927099 [1] NCCL INFO Channel 00/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926595:927099 [1] NCCL INFO Channel 01/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926595:927099 [1] NCCL INFO Channel 02/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926595:927099 [1] NCCL INFO Channel 03/0 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA
+ip-26-0-172-73:926596:927109 [2] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-172-73:926596:927109 [2] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-172-73:926596:927109 [2] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-172-73:926596:927109 [2] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-172-73:926596:927109 [2] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-172-73:926596:927109 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-173-7:811250:811782 [2] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-173-7:811250:811782 [2] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-73:926599:927107 [5] NCCL INFO Setting affinity for GPU 5 to ffffffff,ffff0000,00000000
+ip-26-0-172-73:926599:927107 [5] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-172-73:926599:927107 [5] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-172-73:926599:927107 [5] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-172-73:926599:927107 [5] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-173-7:811253:811781 [5] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-173-7:811253:811781 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-73:926599:927107 [5] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-172-73:926599:927107 [5] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-173-7:811251:811780 [3] NCCL INFO Channel 00/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-7:811251:811780 [3] NCCL INFO Channel 01/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-7:811251:811780 [3] NCCL INFO Channel 02/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-7:811251:811780 [3] NCCL INFO Channel 03/0 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-7:811251:811780 [3] NCCL INFO Channel 00/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-7:811251:811780 [3] NCCL INFO Channel 01/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-7:811251:811780 [3] NCCL INFO Channel 02/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-173-7:811251:811780 [3] NCCL INFO Channel 03/0 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926597:927105 [3] NCCL INFO Channel 00/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926597:927105 [3] NCCL INFO Channel 01/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926597:927105 [3] NCCL INFO Channel 02/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926597:927105 [3] NCCL INFO Channel 03/0 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926597:927105 [3] NCCL INFO Channel 00/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926597:927105 [3] NCCL INFO Channel 01/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926597:927105 [3] NCCL INFO Channel 02/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+ip-26-0-172-73:926597:927105 [3] NCCL INFO Channel 03/0 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+ip-26-0-172-73:926601:927122 [7] NCCL INFO Using network Libfabric
+ip-26-0-173-7:811255:811793 [7] NCCL INFO Using network Libfabric
+ip-26-0-173-7:811255:811793 [7] NCCL INFO comm 0xc7e4b50 rank 1 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0x5660b1c9c80a4b4e - Init START
+ip-26-0-172-73:926601:927122 [7] NCCL INFO comm 0x135fc070 rank 0 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0x5660b1c9c80a4b4e - Init START
+ip-26-0-173-7:811255:811793 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-172-73:926601:927122 [7] NCCL INFO NCCL_TOPO_FILE set by environment to /proc/self/fd/280
+ip-26-0-173-7:811250:811782 [2] NCCL INFO Channel 00/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-7:811250:811782 [2] NCCL INFO Channel 01/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-7:811250:811782 [2] NCCL INFO Channel 02/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-7:811250:811782 [2] NCCL INFO Channel 03/0 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-7:811250:811782 [2] NCCL INFO Channel 00/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-7:811250:811782 [2] NCCL INFO Channel 01/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-7:811250:811782 [2] NCCL INFO Channel 02/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-7:811250:811782 [2] NCCL INFO Channel 03/0 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926596:927109 [2] NCCL INFO Channel 00/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926596:927109 [2] NCCL INFO Channel 01/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926596:927109 [2] NCCL INFO Channel 02/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926596:927109 [2] NCCL INFO Channel 03/0 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926596:927109 [2] NCCL INFO Channel 00/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926596:927109 [2] NCCL INFO Channel 01/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926596:927109 [2] NCCL INFO Channel 02/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-172-73:926596:927109 [2] NCCL INFO Channel 03/0 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA
+ip-26-0-173-7:811253:811781 [5] NCCL INFO Channel 00/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-7:811253:811781 [5] NCCL INFO Channel 01/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-7:811253:811781 [5] NCCL INFO Channel 02/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-7:811253:811781 [5] NCCL INFO Channel 03/0 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-7:811253:811781 [5] NCCL INFO Channel 00/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-7:811253:811781 [5] NCCL INFO Channel 01/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-7:811253:811781 [5] NCCL INFO Channel 02/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-7:811253:811781 [5] NCCL INFO Channel 03/0 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926599:927107 [5] NCCL INFO Channel 00/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926599:927107 [5] NCCL INFO Channel 01/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926599:927107 [5] NCCL INFO Channel 02/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926599:927107 [5] NCCL INFO Channel 03/0 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926599:927107 [5] NCCL INFO Channel 00/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926599:927107 [5] NCCL INFO Channel 01/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926599:927107 [5] NCCL INFO Channel 02/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-172-73:926599:927107 [5] NCCL INFO Channel 03/0 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA
+ip-26-0-173-7:811248:811788 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-172-73:926594:927116 [0] NCCL INFO Setting affinity for GPU 0 to ffff,ffffffff
+ip-26-0-173-7:811248:811788 [0] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-173-7:811248:811788 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-172-73:926594:927116 [0] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-172-73:926594:927116 [0] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-172-73:926594:927116 [0] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-172-73:926594:927116 [0] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-172-73:926594:927116 [0] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-172-73:926594:927116 [0] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-173-7:811255:811793 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-172-73:926601:927122 [7] NCCL INFO Setting affinity for GPU 7 to ffffffff,ffff0000,00000000
+ip-26-0-172-73:926601:927122 [7] NCCL INFO Channel 00/04 :    0   1
+ip-26-0-172-73:926601:927122 [7] NCCL INFO Channel 01/04 :    0   1
+ip-26-0-172-73:926601:927122 [7] NCCL INFO Channel 02/04 :    0   1
+ip-26-0-172-73:926601:927122 [7] NCCL INFO Channel 03/04 :    0   1
+ip-26-0-172-73:926601:927122 [7] NCCL INFO Trees [0] 1/-1/-1->0->-1 [1] 1/-1/-1->0->-1 [2] -1/-1/-1->0->1 [3] -1/-1/-1->0->1
+ip-26-0-172-73:926601:927122 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-173-7:811255:811793 [7] NCCL INFO Trees [0] -1/-1/-1->1->0 [1] -1/-1/-1->1->0 [2] 0/-1/-1->1->-1 [3] 0/-1/-1->1->-1
+ip-26-0-173-7:811255:811793 [7] NCCL INFO P2P Chunksize set to 131072
+ip-26-0-173-7:811248:811788 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-7:811248:811788 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-7:811248:811788 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-7:811248:811788 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-7:811248:811788 [0] NCCL INFO Channel 00/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-7:811248:811788 [0] NCCL INFO Channel 01/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-7:811248:811788 [0] NCCL INFO Channel 02/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-7:811248:811788 [0] NCCL INFO Channel 03/0 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926594:927116 [0] NCCL INFO Channel 00/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926594:927116 [0] NCCL INFO Channel 01/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926594:927116 [0] NCCL INFO Channel 02/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926594:927116 [0] NCCL INFO Channel 03/0 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926594:927116 [0] NCCL INFO Channel 00/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926594:927116 [0] NCCL INFO Channel 01/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926594:927116 [0] NCCL INFO Channel 02/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-172-73:926594:927116 [0] NCCL INFO Channel 03/0 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA
+ip-26-0-173-7:811254:811773 [6] NCCL INFO Connected all rings
+ip-26-0-173-7:811254:811773 [6] NCCL INFO Connected all trees
+ip-26-0-173-7:811254:811773 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-173-7:811254:811773 [6] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-173-7:811254:811773 [6] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-173-7:811254:811773 [6] NCCL INFO comm 0xb59d8f0 rank 1 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0x46cdb14851677f96 - Init COMPLETE
+ip-26-0-173-7:811254:811798 [6] NCCL INFO Channel 02/1 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA/Shared
+ip-26-0-173-7:811254:811798 [6] NCCL INFO Channel 03/1 : 0[6] -> 1[6] [receive] via NET/Libfabric/6/GDRDMA/Shared
+ip-26-0-173-7:811252:811774 [4] NCCL INFO Connected all rings
+ip-26-0-173-7:811252:811774 [4] NCCL INFO Connected all trees
+ip-26-0-173-7:811252:811774 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-173-7:811252:811774 [4] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-173-7:811252:811774 [4] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-173-7:811255:811793 [7] NCCL INFO Channel 00/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926601:927122 [7] NCCL INFO Channel 00/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-7:811255:811793 [7] NCCL INFO Channel 01/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-7:811255:811793 [7] NCCL INFO Channel 02/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926601:927122 [7] NCCL INFO Channel 01/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-7:811255:811793 [7] NCCL INFO Channel 03/0 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926601:927122 [7] NCCL INFO Channel 02/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-7:811255:811793 [7] NCCL INFO Channel 00/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926601:927122 [7] NCCL INFO Channel 03/0 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-7:811255:811793 [7] NCCL INFO Channel 01/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926601:927122 [7] NCCL INFO Channel 00/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-7:811255:811793 [7] NCCL INFO Channel 02/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926601:927122 [7] NCCL INFO Channel 01/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-173-7:811255:811793 [7] NCCL INFO Channel 03/0 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926601:927122 [7] NCCL INFO Channel 02/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926601:927122 [7] NCCL INFO Channel 03/0 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA
+ip-26-0-172-73:926598:927097 [4] NCCL INFO Connected all rings
+ip-26-0-172-73:926598:927097 [4] NCCL INFO Connected all trees
+ip-26-0-172-73:926598:927097 [4] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-73:926598:927097 [4] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-172-73:926598:927097 [4] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-73:926600:927095 [6] NCCL INFO Connected all rings
+ip-26-0-172-73:926600:927095 [6] NCCL INFO Connected all trees
+ip-26-0-172-73:926600:927095 [6] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-73:926600:927095 [6] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-172-73:926600:927095 [6] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-173-7:811252:811774 [4] NCCL INFO comm 0xb8b25f0 rank 1 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0x2f780d840a209ef2 - Init COMPLETE
+ip-26-0-173-7:811252:811799 [4] NCCL INFO Channel 02/1 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA/Shared
+ip-26-0-173-7:811252:811799 [4] NCCL INFO Channel 03/1 : 0[4] -> 1[4] [receive] via NET/Libfabric/4/GDRDMA/Shared
+ip-26-0-172-73:926598:927097 [4] NCCL INFO comm 0x14966230 rank 0 nranks 2 cudaDev 4 nvmlDev 4 busId 97000 commId 0x2f780d840a209ef2 - Init COMPLETE
+ip-26-0-172-73:926598:927127 [4] NCCL INFO Channel 02/1 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA/Shared
+ip-26-0-172-73:926598:927127 [4] NCCL INFO Channel 03/1 : 0[4] -> 1[4] [send] via NET/Libfabric/4/GDRDMA/Shared
+ip-26-0-172-73:926600:927095 [6] NCCL INFO comm 0x15026b70 rank 0 nranks 2 cudaDev 6 nvmlDev 6 busId b9000 commId 0x46cdb14851677f96 - Init COMPLETE
+ip-26-0-172-73:926600:927128 [6] NCCL INFO Channel 02/1 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA/Shared
+ip-26-0-172-73:926600:927128 [6] NCCL INFO Channel 03/1 : 0[6] -> 1[6] [send] via NET/Libfabric/6/GDRDMA/Shared
+ip-26-0-173-7:811251:811780 [3] NCCL INFO Connected all rings
+ip-26-0-173-7:811251:811780 [3] NCCL INFO Connected all trees
+ip-26-0-173-7:811251:811780 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-173-7:811251:811780 [3] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-173-7:811251:811780 [3] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-173-7:811249:811775 [1] NCCL INFO Connected all rings
+ip-26-0-173-7:811249:811775 [1] NCCL INFO Connected all trees
+ip-26-0-173-7:811249:811775 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-173-7:811249:811775 [1] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-173-7:811249:811775 [1] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-73:926595:927099 [1] NCCL INFO Connected all rings
+ip-26-0-172-73:926595:927099 [1] NCCL INFO Connected all trees
+ip-26-0-172-73:926595:927099 [1] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-73:926595:927099 [1] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-172-73:926595:927099 [1] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-173-7:811251:811780 [3] NCCL INFO comm 0xb793130 rank 1 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0xc0be7100b59ca4f3 - Init COMPLETE
+ip-26-0-173-7:811251:811800 [3] NCCL INFO Channel 02/1 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA/Shared
+ip-26-0-173-7:811251:811800 [3] NCCL INFO Channel 03/1 : 0[3] -> 1[3] [receive] via NET/Libfabric/3/GDRDMA/Shared
+ip-26-0-173-7:811249:811775 [1] NCCL INFO comm 0xb398d80 rank 1 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0x3369bb1417fe388d - Init COMPLETE
+ip-26-0-173-7:811250:811782 [2] NCCL INFO Connected all rings
+ip-26-0-173-7:811250:811782 [2] NCCL INFO Connected all trees
+ip-26-0-173-7:811249:811801 [1] NCCL INFO Channel 02/1 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA/Shared
+ip-26-0-173-7:811250:811782 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-173-7:811250:811782 [2] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-173-7:811250:811782 [2] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-173-7:811249:811801 [1] NCCL INFO Channel 03/1 : 0[1] -> 1[1] [receive] via NET/Libfabric/1/GDRDMA/Shared
+ip-26-0-173-7:811253:811781 [5] NCCL INFO Connected all rings
+ip-26-0-173-7:811253:811781 [5] NCCL INFO Connected all trees
+ip-26-0-173-7:811253:811781 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-173-7:811253:811781 [5] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-173-7:811253:811781 [5] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-73:926597:927105 [3] NCCL INFO Connected all rings
+ip-26-0-172-73:926597:927105 [3] NCCL INFO Connected all trees
+ip-26-0-172-73:926597:927105 [3] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-73:926597:927105 [3] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-172-73:926597:927105 [3] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-73:926596:927109 [2] NCCL INFO Connected all rings
+ip-26-0-172-73:926596:927109 [2] NCCL INFO Connected all trees
+ip-26-0-172-73:926596:927109 [2] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-73:926596:927109 [2] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-172-73:926596:927109 [2] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-73:926595:927099 [1] NCCL INFO comm 0x135204b0 rank 0 nranks 2 cudaDev 1 nvmlDev 1 busId 64000 commId 0x3369bb1417fe388d - Init COMPLETE
+ip-26-0-172-73:926595:927129 [1] NCCL INFO Channel 02/1 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA/Shared
+ip-26-0-172-73:926595:927129 [1] NCCL INFO Channel 03/1 : 0[1] -> 1[1] [send] via NET/Libfabric/1/GDRDMA/Shared
+ip-26-0-172-73:926599:927107 [5] NCCL INFO Connected all rings
+ip-26-0-172-73:926599:927107 [5] NCCL INFO Connected all trees
+ip-26-0-172-73:926599:927107 [5] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-73:926599:927107 [5] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-172-73:926599:927107 [5] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-173-7:811250:811782 [2] NCCL INFO comm 0xc194bc0 rank 1 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0x840b226f2838c652 - Init COMPLETE
+ip-26-0-173-7:811253:811781 [5] NCCL INFO comm 0xc207e60 rank 1 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0x57be41670ad0e398 - Init COMPLETE
+ip-26-0-173-7:811253:811803 [5] NCCL INFO Channel 02/1 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA/Shared
+ip-26-0-173-7:811253:811803 [5] NCCL INFO Channel 03/1 : 0[5] -> 1[5] [receive] via NET/Libfabric/5/GDRDMA/Shared
+ip-26-0-173-7:811250:811802 [2] NCCL INFO Channel 02/1 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA/Shared
+ip-26-0-173-7:811250:811802 [2] NCCL INFO Channel 03/1 : 0[2] -> 1[2] [receive] via NET/Libfabric/2/GDRDMA/Shared
+ip-26-0-172-73:926597:927105 [3] NCCL INFO comm 0x13cfd0f0 rank 0 nranks 2 cudaDev 3 nvmlDev 3 busId 86000 commId 0xc0be7100b59ca4f3 - Init COMPLETE
+ip-26-0-172-73:926599:927107 [5] NCCL INFO comm 0x15016730 rank 0 nranks 2 cudaDev 5 nvmlDev 5 busId a8000 commId 0x57be41670ad0e398 - Init COMPLETE
+ip-26-0-172-73:926597:927130 [3] NCCL INFO Channel 02/1 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA/Shared
+ip-26-0-172-73:926597:927130 [3] NCCL INFO Channel 03/1 : 0[3] -> 1[3] [send] via NET/Libfabric/3/GDRDMA/Shared
+ip-26-0-172-73:926596:927109 [2] NCCL INFO comm 0x13e6e940 rank 0 nranks 2 cudaDev 2 nvmlDev 2 busId 75000 commId 0x840b226f2838c652 - Init COMPLETE
+ip-26-0-172-73:926599:927131 [5] NCCL INFO Channel 02/1 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA/Shared
+ip-26-0-172-73:926599:927131 [5] NCCL INFO Channel 03/1 : 0[5] -> 1[5] [send] via NET/Libfabric/5/GDRDMA/Shared
+ip-26-0-172-73:926596:927132 [2] NCCL INFO Channel 02/1 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA/Shared
+ip-26-0-172-73:926596:927132 [2] NCCL INFO Channel 03/1 : 0[2] -> 1[2] [send] via NET/Libfabric/2/GDRDMA/Shared
+ip-26-0-173-7:811248:811788 [0] NCCL INFO Connected all rings
+ip-26-0-173-7:811248:811788 [0] NCCL INFO Connected all trees
+ip-26-0-173-7:811248:811788 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-173-7:811248:811788 [0] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-173-7:811248:811788 [0] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-73:926594:927116 [0] NCCL INFO Connected all rings
+ip-26-0-172-73:926594:927116 [0] NCCL INFO Connected all trees
+ip-26-0-172-73:926594:927116 [0] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-73:926594:927116 [0] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-172-73:926594:927116 [0] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-173-7:811248:811788 [0] NCCL INFO comm 0xcdaab90 rank 1 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0xf44e045fbd0f3601 - Init COMPLETE
+ip-26-0-173-7:811248:811804 [0] NCCL INFO Channel 02/1 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA/Shared
+ip-26-0-173-7:811248:811804 [0] NCCL INFO Channel 03/1 : 0[0] -> 1[0] [receive] via NET/Libfabric/0/GDRDMA/Shared
+ip-26-0-172-73:926594:927116 [0] NCCL INFO comm 0x1442e1f0 rank 0 nranks 2 cudaDev 0 nvmlDev 0 busId 53000 commId 0xf44e045fbd0f3601 - Init COMPLETE
+ip-26-0-172-73:926594:927133 [0] NCCL INFO Channel 02/1 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA/Shared
+ip-26-0-172-73:926594:927133 [0] NCCL INFO Channel 03/1 : 0[0] -> 1[0] [send] via NET/Libfabric/0/GDRDMA/Shared
+ip-26-0-172-73:926598:927142 [4] NCCL INFO Channel 02/1 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA/Shared
+ip-26-0-172-73:926598:927142 [4] NCCL INFO Channel 03/1 : 1[4] -> 0[4] [receive] via NET/Libfabric/4/GDRDMA/Shared
+ip-26-0-172-73:926601:927122 [7] NCCL INFO Connected all rings
+ip-26-0-172-73:926601:927122 [7] NCCL INFO Connected all trees
+ip-26-0-172-73:926601:927122 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-172-73:926601:927122 [7] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-172-73:926601:927122 [7] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-173-7:811255:811793 [7] NCCL INFO Connected all rings
+ip-26-0-173-7:811255:811793 [7] NCCL INFO Connected all trees
+ip-26-0-173-7:811255:811793 [7] NCCL INFO NCCL_PROTO set by environment to simple
+ip-26-0-173-7:811255:811793 [7] NCCL INFO threadThresholds 8/8/64 | 16/8/64 | 512 | 512
+ip-26-0-173-7:811255:811793 [7] NCCL INFO 4 coll channels, 0 nvls channels, 4 p2p channels, 2 p2p channels per peer
+ip-26-0-172-73:926595:927151 [1] NCCL INFO Channel 02/1 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA/Shared
+ip-26-0-172-73:926595:927151 [1] NCCL INFO Channel 03/1 : 1[1] -> 0[1] [receive] via NET/Libfabric/1/GDRDMA/Shared
+ip-26-0-172-73:926601:927122 [7] NCCL INFO comm 0x135fc070 rank 0 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0x5660b1c9c80a4b4e - Init COMPLETE
+ip-26-0-172-73:926601:927152 [7] NCCL INFO Channel 02/1 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA/Shared
+ip-26-0-172-73:926601:927152 [7] NCCL INFO Channel 03/1 : 0[7] -> 1[7] [send] via NET/Libfabric/7/GDRDMA/Shared
+ip-26-0-173-7:811255:811793 [7] NCCL INFO comm 0xc7e4b50 rank 1 nranks 2 cudaDev 7 nvmlDev 7 busId ca000 commId 0x5660b1c9c80a4b4e - Init COMPLETE
+ip-26-0-173-7:811255:811805 [7] NCCL INFO Channel 02/1 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA/Shared
+ip-26-0-173-7:811255:811805 [7] NCCL INFO Channel 03/1 : 0[7] -> 1[7] [receive] via NET/Libfabric/7/GDRDMA/Shared
+ip-26-0-172-73:926597:927161 [3] NCCL INFO Channel 02/1 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA/Shared
+ip-26-0-172-73:926597:927161 [3] NCCL INFO Channel 03/1 : 1[3] -> 0[3] [receive] via NET/Libfabric/3/GDRDMA/Shared
+ip-26-0-172-73:926596:927177 [2] NCCL INFO Channel 02/1 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA/Shared
+ip-26-0-172-73:926596:927177 [2] NCCL INFO Channel 03/1 : 1[2] -> 0[2] [receive] via NET/Libfabric/2/GDRDMA/Shared
+ip-26-0-172-73:926599:927179 [5] NCCL INFO Channel 02/1 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA/Shared
+ip-26-0-172-73:926599:927179 [5] NCCL INFO Channel 03/1 : 1[5] -> 0[5] [receive] via NET/Libfabric/5/GDRDMA/Shared
+ip-26-0-172-73:926600:927189 [6] NCCL INFO Channel 02/1 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA/Shared
+ip-26-0-172-73:926600:927189 [6] NCCL INFO Channel 03/1 : 1[6] -> 0[6] [receive] via NET/Libfabric/6/GDRDMA/Shared
+ip-26-0-172-73:926594:927198 [0] NCCL INFO Channel 02/1 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA/Shared
+ip-26-0-172-73:926594:927198 [0] NCCL INFO Channel 03/1 : 1[0] -> 0[0] [receive] via NET/Libfabric/0/GDRDMA/Shared
+ip-26-0-172-73:926601:927208 [7] NCCL INFO Channel 02/1 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA/Shared
+ip-26-0-172-73:926601:927208 [7] NCCL INFO Channel 03/1 : 1[7] -> 0[7] [receive] via NET/Libfabric/7/GDRDMA/Shared
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+ip-26-0-173-7:811254:811889 [6] NCCL INFO Channel 02/1 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA/Shared
+ip-26-0-173-7:811254:811889 [6] NCCL INFO Channel 03/1 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA/Shared
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+ip-26-0-173-7:811249:811890 [1] NCCL INFO Channel 02/1 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA/Shared
+ip-26-0-173-7:811249:811890 [1] NCCL INFO Channel 03/1 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA/Shared
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+ip-26-0-173-7:811252:811891 [4] NCCL INFO Channel 02/1 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA/Shared
+ip-26-0-173-7:811252:811891 [4] NCCL INFO Channel 03/1 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA/Shared
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+ip-26-0-173-7:811253:811892 [5] NCCL INFO Channel 02/1 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA/Shared
+ip-26-0-173-7:811253:811892 [5] NCCL INFO Channel 03/1 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA/Shared
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+ip-26-0-173-7:811251:811893 [3] NCCL INFO Channel 02/1 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA/Shared
+ip-26-0-173-7:811251:811893 [3] NCCL INFO Channel 03/1 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA/Shared
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+ip-26-0-173-7:811250:811894 [2] NCCL INFO Channel 02/1 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA/Shared
+ip-26-0-173-7:811250:811894 [2] NCCL INFO Channel 03/1 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA/Shared
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+ip-26-0-173-7:811248:811895 [0] NCCL INFO Channel 02/1 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA/Shared
+ip-26-0-173-7:811248:811895 [0] NCCL INFO Channel 03/1 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA/Shared
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+ip-26-0-173-7:811255:811897 [7] NCCL INFO Channel 02/1 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA/Shared
+ip-26-0-173-7:811255:811897 [7] NCCL INFO Channel 03/1 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA/Shared
+ip-26-0-172-73:926600:927212 [6] NCCL INFO Channel 02/1 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA/Shared
+ip-26-0-172-73:926600:927212 [6] NCCL INFO Channel 03/1 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA/Shared
+ip-26-0-172-73:926595:927213 [1] NCCL INFO Channel 02/1 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA/Shared
+ip-26-0-172-73:926595:927213 [1] NCCL INFO Channel 03/1 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA/Shared
+ip-26-0-172-73:926598:927214 [4] NCCL INFO Channel 02/1 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA/Shared
+ip-26-0-172-73:926598:927214 [4] NCCL INFO Channel 03/1 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA/Shared
+ip-26-0-172-73:926599:927215 [5] NCCL INFO Channel 02/1 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA/Shared
+ip-26-0-172-73:926599:927215 [5] NCCL INFO Channel 03/1 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA/Shared
+ip-26-0-172-73:926597:927216 [3] NCCL INFO Channel 02/1 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA/Shared
+ip-26-0-172-73:926597:927216 [3] NCCL INFO Channel 03/1 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA/Shared
+ip-26-0-172-73:926596:927217 [2] NCCL INFO Channel 02/1 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA/Shared
+ip-26-0-172-73:926596:927217 [2] NCCL INFO Channel 03/1 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA/Shared
+ip-26-0-172-73:926594:927219 [0] NCCL INFO Channel 02/1 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA/Shared
+ip-26-0-172-73:926594:927219 [0] NCCL INFO Channel 03/1 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA/Shared
+ip-26-0-172-73:926601:927220 [7] NCCL INFO Channel 02/1 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA/Shared
+ip-26-0-172-73:926601:927220 [7] NCCL INFO Channel 03/1 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA/Shared
+ip-26-0-172-57:942617:943232 [6] NCCL INFO Channel 02/1 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA/Shared
+ip-26-0-172-57:942617:943232 [6] NCCL INFO Channel 03/1 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA/Shared
+ip-26-0-172-57:942612:943233 [1] NCCL INFO Channel 02/1 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA/Shared
+ip-26-0-172-57:942612:943233 [1] NCCL INFO Channel 03/1 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA/Shared
+ip-26-0-172-57:942615:943234 [4] NCCL INFO Channel 02/1 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA/Shared
+ip-26-0-172-57:942615:943234 [4] NCCL INFO Channel 03/1 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA/Shared
+ip-26-0-172-57:942616:943235 [5] NCCL INFO Channel 02/1 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA/Shared
+ip-26-0-172-57:942616:943235 [5] NCCL INFO Channel 03/1 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA/Shared
+ip-26-0-172-57:942613:943236 [2] NCCL INFO Channel 02/1 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA/Shared
+ip-26-0-172-57:942613:943236 [2] NCCL INFO Channel 03/1 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA/Shared
+ip-26-0-172-57:942614:943237 [3] NCCL INFO Channel 02/1 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA/Shared
+ip-26-0-172-57:942614:943237 [3] NCCL INFO Channel 03/1 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA/Shared
+ip-26-0-172-57:942611:943238 [0] NCCL INFO Channel 02/1 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA/Shared
+ip-26-0-172-57:942611:943238 [0] NCCL INFO Channel 03/1 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA/Shared
+ip-26-0-172-57:942618:943239 [7] NCCL INFO Channel 02/1 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA/Shared
+ip-26-0-172-57:942618:943239 [7] NCCL INFO Channel 03/1 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA/Shared
+ip-26-0-172-252:688206:688822 [6] NCCL INFO Channel 02/1 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA/Shared
+ip-26-0-172-252:688206:688822 [6] NCCL INFO Channel 03/1 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA/Shared
+ip-26-0-172-252:688204:688823 [4] NCCL INFO Channel 02/1 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA/Shared
+ip-26-0-172-252:688204:688823 [4] NCCL INFO Channel 03/1 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA/Shared
+ip-26-0-172-252:688201:688824 [1] NCCL INFO Channel 02/1 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA/Shared
+ip-26-0-172-252:688201:688824 [1] NCCL INFO Channel 03/1 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA/Shared
+ip-26-0-172-252:688205:688825 [5] NCCL INFO Channel 02/1 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA/Shared
+ip-26-0-172-252:688205:688825 [5] NCCL INFO Channel 03/1 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA/Shared
+ip-26-0-172-252:688202:688826 [2] NCCL INFO Channel 02/1 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA/Shared
+ip-26-0-172-252:688202:688826 [2] NCCL INFO Channel 03/1 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA/Shared
+ip-26-0-172-252:688203:688827 [3] NCCL INFO Channel 02/1 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA/Shared
+ip-26-0-172-252:688203:688827 [3] NCCL INFO Channel 03/1 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA/Shared
+ip-26-0-172-252:688200:688828 [0] NCCL INFO Channel 02/1 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA/Shared
+ip-26-0-172-252:688200:688828 [0] NCCL INFO Channel 03/1 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA/Shared
+ip-26-0-172-252:688207:688830 [7] NCCL INFO Channel 02/1 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA/Shared
+ip-26-0-172-252:688207:688830 [7] NCCL INFO Channel 03/1 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA/Shared
+ip-26-0-172-147:618590:619206 [6] NCCL INFO Channel 02/1 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA/Shared
+ip-26-0-172-147:618590:619206 [6] NCCL INFO Channel 03/1 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA/Shared
+ip-26-0-172-147:618585:619207 [1] NCCL INFO Channel 02/1 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA/Shared
+ip-26-0-172-147:618585:619207 [1] NCCL INFO Channel 03/1 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA/Shared
+ip-26-0-172-147:618588:619208 [4] NCCL INFO Channel 02/1 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA/Shared
+ip-26-0-172-147:618588:619208 [4] NCCL INFO Channel 03/1 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA/Shared
+ip-26-0-172-147:618589:619209 [5] NCCL INFO Channel 02/1 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA/Shared
+ip-26-0-172-147:618589:619209 [5] NCCL INFO Channel 03/1 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA/Shared
+ip-26-0-172-147:618586:619210 [2] NCCL INFO Channel 02/1 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA/Shared
+ip-26-0-172-147:618586:619210 [2] NCCL INFO Channel 03/1 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA/Shared
+ip-26-0-172-147:618587:619211 [3] NCCL INFO Channel 02/1 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA/Shared
+ip-26-0-172-147:618587:619211 [3] NCCL INFO Channel 03/1 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA/Shared
+ip-26-0-172-147:618584:619212 [0] NCCL INFO Channel 02/1 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA/Shared
+ip-26-0-172-147:618584:619212 [0] NCCL INFO Channel 03/1 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA/Shared
+ip-26-0-172-147:618591:619213 [7] NCCL INFO Channel 02/1 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA/Shared
+ip-26-0-172-147:618591:619213 [7] NCCL INFO Channel 03/1 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA/Shared
+ip-26-0-172-142:880685:881303 [6] NCCL INFO Channel 02/1 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA/Shared
+ip-26-0-172-142:880685:881303 [6] NCCL INFO Channel 03/1 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA/Shared
+ip-26-0-172-142:880683:881304 [4] NCCL INFO Channel 02/1 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA/Shared
+ip-26-0-172-142:880683:881304 [4] NCCL INFO Channel 03/1 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA/Shared
+ip-26-0-172-142:880680:881305 [1] NCCL INFO Channel 02/1 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA/Shared
+ip-26-0-172-142:880680:881305 [1] NCCL INFO Channel 03/1 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA/Shared
+ip-26-0-172-142:880684:881306 [5] NCCL INFO Channel 02/1 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA/Shared
+ip-26-0-172-142:880684:881306 [5] NCCL INFO Channel 03/1 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA/Shared
+ip-26-0-172-142:880681:881307 [2] NCCL INFO Channel 02/1 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA/Shared
+ip-26-0-172-142:880681:881307 [2] NCCL INFO Channel 03/1 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA/Shared
+ip-26-0-172-142:880679:881308 [0] NCCL INFO Channel 02/1 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA/Shared
+ip-26-0-172-142:880679:881308 [0] NCCL INFO Channel 03/1 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA/Shared
+ip-26-0-172-142:880682:881309 [3] NCCL INFO Channel 02/1 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA/Shared
+ip-26-0-172-142:880682:881309 [3] NCCL INFO Channel 03/1 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA/Shared
+ip-26-0-172-142:880686:881310 [7] NCCL INFO Channel 02/1 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA/Shared
+ip-26-0-172-142:880686:881310 [7] NCCL INFO Channel 03/1 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA/Shared
+ip-26-0-172-116:1019294:1019913 [6] NCCL INFO Channel 02/1 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA/Shared
+ip-26-0-172-116:1019294:1019913 [6] NCCL INFO Channel 03/1 : 1[6] -> 0[6] [send] via NET/Libfabric/6/GDRDMA/Shared
+ip-26-0-172-116:1019289:1019914 [1] NCCL INFO Channel 02/1 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA/Shared
+ip-26-0-172-116:1019289:1019914 [1] NCCL INFO Channel 03/1 : 1[1] -> 0[1] [send] via NET/Libfabric/1/GDRDMA/Shared
+ip-26-0-172-116:1019292:1019915 [4] NCCL INFO Channel 02/1 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA/Shared
+ip-26-0-172-116:1019292:1019915 [4] NCCL INFO Channel 03/1 : 1[4] -> 0[4] [send] via NET/Libfabric/4/GDRDMA/Shared
+ip-26-0-172-116:1019293:1019916 [5] NCCL INFO Channel 02/1 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA/Shared
+ip-26-0-172-116:1019293:1019916 [5] NCCL INFO Channel 03/1 : 1[5] -> 0[5] [send] via NET/Libfabric/5/GDRDMA/Shared
+ip-26-0-172-116:1019290:1019917 [2] NCCL INFO Channel 02/1 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA/Shared
+ip-26-0-172-116:1019290:1019917 [2] NCCL INFO Channel 03/1 : 1[2] -> 0[2] [send] via NET/Libfabric/2/GDRDMA/Shared
+ip-26-0-172-116:1019288:1019918 [0] NCCL INFO Channel 02/1 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA/Shared
+ip-26-0-172-116:1019288:1019918 [0] NCCL INFO Channel 03/1 : 1[0] -> 0[0] [send] via NET/Libfabric/0/GDRDMA/Shared
+ip-26-0-172-116:1019291:1019919 [3] NCCL INFO Channel 02/1 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA/Shared
+ip-26-0-172-116:1019291:1019919 [3] NCCL INFO Channel 03/1 : 1[3] -> 0[3] [send] via NET/Libfabric/3/GDRDMA/Shared
+ip-26-0-172-116:1019295:1019920 [7] NCCL INFO Channel 02/1 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA/Shared
+ip-26-0-172-116:1019295:1019920 [7] NCCL INFO Channel 03/1 : 1[7] -> 0[7] [send] via NET/Libfabric/7/GDRDMA/Shared
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+01/07/2025 05:27:46 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:  Memory usage: 5371.24MiB. Peak allocated 37549.69MiB. Peak reserved: 39456.00MiB
+01/07/2025 05:27:47 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:  Memory usage: 6188.26MiB. Peak allocated 9656.36MiB. Peak reserved: 41090.00MiB
+01/07/2025 05:27:47 [INFO|DP=0|PP=7|TP=0|ip-26-0-173-7]: iteration: 1 / 100 | consumed_tokens: 1.05M | elapsed_time_per_iteration_ms: 21.9K | tokens_per_sec: 48K | tokens_per_sec_per_gpu: 750 | global_batch_size: 256 | lm_loss: 12.1 | lr: 0.00015 | model_tflops_per_gpu: 19.2 | hardware_tflops_per_gpu: 19.2 | grad_norm: 1.91 | cuda_memory_allocated: 3.09G | cuda_max_memory_reserved: 13.7G | hd_total_memory_tb: 312G | hd_used_memory_tb: 148G | hd_free_memory_tb: 164G
+01/07/2025 05:27:48 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:  Memory usage: 6188.26MiB. Peak allocated 38366.72MiB. Peak reserved: 41090.00MiB
+01/07/2025 05:27:48 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:  Memory usage: 6188.26MiB. Peak allocated 9656.36MiB. Peak reserved: 41090.00MiB
+01/07/2025 05:27:48 [INFO|DP=0|PP=7|TP=0|ip-26-0-173-7]: iteration: 2 / 100 | consumed_tokens: 2.1M | elapsed_time_per_iteration_ms: 1.53K | tokens_per_sec: 688K | tokens_per_sec_per_gpu: 10.7K | global_batch_size: 256 | lm_loss: 12.1 | lr: 0.0003 | model_tflops_per_gpu: 276 | hardware_tflops_per_gpu: 276 | grad_norm: 1.91 | cuda_memory_allocated: 3.09G | cuda_max_memory_reserved: 13.7G | hd_total_memory_tb: 312G | hd_used_memory_tb: 148G | hd_free_memory_tb: 164G
+01/07/2025 05:27:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:  Memory usage: 6188.26MiB. Peak allocated 38366.72MiB. Peak reserved: 41090.00MiB
+01/07/2025 05:27:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:  Memory usage: 6188.26MiB. Peak allocated 9656.36MiB. Peak reserved: 41090.00MiB
+01/07/2025 05:27:50 [INFO|DP=0|PP=7|TP=0|ip-26-0-173-7]: iteration: 3 / 100 | consumed_tokens: 3.15M | elapsed_time_per_iteration_ms: 1.52K | tokens_per_sec: 691K | tokens_per_sec_per_gpu: 10.8K | global_batch_size: 256 | lm_loss: 12.1 | lr: 0.000296 | model_tflops_per_gpu: 277 | hardware_tflops_per_gpu: 277 | grad_norm: 1.86 | cuda_memory_allocated: 3.09G | cuda_max_memory_reserved: 13.7G | hd_total_memory_tb: 312G | hd_used_memory_tb: 148G | hd_free_memory_tb: 164G
+01/07/2025 05:27:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]:  Memory usage: 6188.26MiB. Peak allocated 38366.72MiB. Peak reserved: 41090.00MiB
+01/07/2025 05:27:51 [INFO|DP=0|PP=7|TP=0|ip-26-0-173-7]: iteration: 4 / 100 | consumed_tokens: 4.19M | elapsed_time_per_iteration_ms: 1.52K | tokens_per_sec: 691K | tokens_per_sec_per_gpu: 10.8K | global_batch_size: 256 | lm_loss: 12.1 | lr: 0.000283 | model_tflops_per_gpu: 277 | hardware_tflops_per_gpu: 277 | grad_norm: 1.47 | cuda_memory_allocated: 3.09G | cuda_max_memory_reserved: 13.7G | hd_total_memory_tb: 312G | hd_used_memory_tb: 148G | hd_free_memory_tb: 164G
+01/07/2025 05:27:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: | job_id   | name                                                           | nodes | seq_len | mbs | batch_accum | gbs | mTFLOPs | hTFLOPs | tok/s/gpu | AllReduce (GB/s) | AllGather (GB/s) | ReduceScatter (GB/s) | AR Intra-node (GB/s) | AG Intra-node (GB/s) | RS Intra-node (GB/s) | Mem Alloc (GB) | Mem Res (GB) | dp | pp | tp | pp_engine                           | tp_mode        | tp_async_comm | hidden_size | hidden_act | num_layers | num_heads | num_kv_heads | max_pos | vocab_size | tie_word_embeddings | dtype          | zero_stage | ddp_bucket_cap_mb | accumulate_grad_in_fp32 | Total Params | Local Params |
+01/07/2025 05:27:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: | -------- | -------------------------------------------------------------- | ----- | ------- | --- | ----------- | --- | ------- | ------- | --------- | ---------------- | ---------------- | -------------------- | -------------------- | -------------------- | -------------------- | -------------- | ------------ | -- | -- | -- | ----------------------------------- | -------------- | ------------- | ----------- | ---------- | ---------- | --------- | ------------ | ------- | ---------- | ------------------- | -------------- | ---------- | ----------------- | ----------------------- | ------------ | ------------ |
+01/07/2025 05:27:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: | 14098868 | 3.57G_dp8_tp1_pp8_acc16_mbs2_seq4096_zero1_tpmodeRED_vocab131k | 8     | 4096    | 2   | 16          | 256 | 277.22  | 277.22  | 10799.78  | 96.00            | 53.63            | 52.61                | 424.52               | 242.03               | 241.28               | 9.43           | 40.13        | 8  | 8  | 1  | OneForwardOneBackwardPipelineEngine | REDUCE_SCATTER | True          | 3072        | silu       | 28         | 32        | 32           | 4096    | 131072     | True                | torch.bfloat16 | 1          | 25                | True                    | 3.98G        | 856M         |
+01/07/2025 05:27:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: Attempting to write benchmark results to CSV file: benchmark/results/bench_final2.csv
+01/07/2025 05:27:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: Acquired lock for CSV file: benchmark/results/bench_final2.csv
+01/07/2025 05:27:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: Successfully wrote to CSV file: benchmark/results/bench_final2.csv. Releasing lock...
+01/07/2025 05:27:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-160-225]: Throughput logging complete
+ip-26-0-173-7:811248:811610 [32703] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-173-7:811253:811576 [32557] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-173-7:811252:811573 [32648] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-173-7:811251:811577 [32675] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-173-7:811250:811575 [32615] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-173-7:811249:811574 [32632] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-173-7:811248:811571 [32705] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-173-7:811254:811570 [32702] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-173-7:811253:811433 [32559] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-173-7:811252:811431 [32649] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-173-7:811251:811429 [32676] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-173-7:811250:811434 [32616] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-173-7:811249:811432 [32633] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-173-7:811254:811430 [32703] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-173-7:811248:811428 [32706] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-160-225:94342:94701 [32706] NCCL INFO [Service thread] Connection closed by localRank 5
+ip-26-0-160-225:94349:94659 [32612] NCCL INFO [Service thread] Connection closed by localRank 5
+ip-26-0-160-225:94348:94661 [32708] NCCL INFO [Service thread] Connection closed by localRank 5
+ip-26-0-160-225:94343:94660 [32594] NCCL INFO [Service thread] Connection closed by localRank 5
+ip-26-0-160-225:94344:94662 [32731] NCCL INFO [Service thread] Connection closed by localRank 5
+ip-26-0-160-225:94345:94665 [32731] NCCL INFO [Service thread] Connection closed by localRank 5
+ip-26-0-160-225:94342:94663 [32708] NCCL INFO [Service thread] Connection closed by localRank 5
+ip-26-0-160-225:94349:94526 [32613] NCCL INFO [Service thread] Connection closed by localRank 5
+ip-26-0-160-225:94348:94525 [32710] NCCL INFO [Service thread] Connection closed by localRank 5
+ip-26-0-160-225:94345:94528 [32732] NCCL INFO [Service thread] Connection closed by localRank 5
+ip-26-0-160-225:94343:94529 [32595] NCCL INFO [Service thread] Connection closed by localRank 5
+ip-26-0-160-225:94344:94530 [32732] NCCL INFO [Service thread] Connection closed by localRank 5
+ip-26-0-160-225:94342:94531 [32710] NCCL INFO [Service thread] Connection closed by localRank 5
+ip-26-0-172-73:926601:926910 [32754] NCCL INFO [Service thread] Connection closed by localRank 0
+ip-26-0-172-73:926600:926909 [32510] NCCL INFO [Service thread] Connection closed by localRank 0
+ip-26-0-172-73:926598:926915 [32648] NCCL INFO [Service thread] Connection closed by localRank 0
+ip-26-0-172-73:926599:926911 [32579] NCCL INFO [Service thread] Connection closed by localRank 0
+ip-26-0-172-73:926597:926914 [32632] NCCL INFO [Service thread] Connection closed by localRank 0
+ip-26-0-172-73:926595:926912 [32573] NCCL INFO [Service thread] Connection closed by localRank 0
+ip-26-0-172-73:926596:926913 [32600] NCCL INFO [Service thread] Connection closed by localRank 0
+ip-26-0-172-73:926601:926775 [32756] NCCL INFO [Service thread] Connection closed by localRank 0
+ip-26-0-172-73:926599:926776 [32581] NCCL INFO [Service thread] Connection closed by localRank 0
+ip-26-0-172-73:926600:926774 [32511] NCCL INFO [Service thread] Connection closed by localRank 0
+ip-26-0-172-73:926598:926773 [32649] NCCL INFO [Service thread] Connection closed by localRank 0
+ip-26-0-172-73:926597:926778 [32633] NCCL INFO [Service thread] Connection closed by localRank 0
+ip-26-0-172-73:926596:926780 [32602] NCCL INFO [Service thread] Connection closed by localRank 0
+ip-26-0-172-73:926595:926779 [32575] NCCL INFO [Service thread] Connection closed by localRank 0
+ip-26-0-172-116:1019288:1019641 [32701] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-172-116:1019293:1019607 [32701] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-172-116:1019292:1019606 [32602] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-172-116:1019291:1019609 [32657] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-172-116:1019290:1019605 [32742] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-172-116:1019294:1019610 [32576] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-172-116:1019288:1019603 [32704] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-172-116:1019289:1019604 [32560] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-172-116:1019293:1019475 [32703] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-172-116:1019292:1019471 [32604] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-172-116:1019291:1019474 [32659] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-172-116:1019289:1019470 [32562] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-172-116:1019290:1019472 [32743] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-172-116:1019294:1019476 [32578] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-172-116:1019288:1019469 [32706] NCCL INFO [Service thread] Connection closed by localRank 7
+ip-26-0-172-252:688207:688519 [32694] NCCL INFO [Service thread] Connection closed by localRank 5
+ip-26-0-172-252:688201:688521 [32696] NCCL INFO [Service thread] Connection closed by localRank 5
+ip-26-0-172-252:688202:688518 [32650] NCCL INFO [Service thread] Connection closed by localRank 5
+ip-26-0-172-252:688203:688522 [32690] NCCL INFO [Service thread] Connection closed by localRank 5
+ip-26-0-172-252:688204:688517 [32639] NCCL INFO [Service thread] Connection closed by localRank 5
+ip-26-0-172-252:688207:688384 [32696] NCCL INFO [Service thread] Connection closed by localRank 5
+ip-26-0-172-252:688203:688383 [32691] NCCL INFO [Service thread] Connection closed by localRank 5
+ip-26-0-172-252:688202:688388 [32652] NCCL INFO [Service thread] Connection closed by localRank 5
+ip-26-0-172-252:688201:688382 [32698] NCCL INFO [Service thread] Connection closed by localRank 5
+ip-26-0-172-252:688204:688387 [32640] NCCL INFO [Service thread] Connection closed by localRank 5
+ip-26-0-172-147:618586:618898 [32573] NCCL INFO [Service thread] Connection closed by localRank 4
+ip-26-0-172-147:618590:618905 [32551] NCCL INFO [Service thread] Connection closed by localRank 4
+ip-26-0-172-147:618585:618901 [32692] NCCL INFO [Service thread] Connection closed by localRank 4
+ip-26-0-172-147:618591:618904 [32678] NCCL INFO [Service thread] Connection closed by localRank 4
+ip-26-0-172-147:618587:618903 [32700] NCCL INFO [Service thread] Connection closed by localRank 4
+ip-26-0-172-147:618590:618768 [32553] NCCL INFO [Service thread] Connection closed by localRank 4
+ip-26-0-172-147:618585:618769 [32694] NCCL INFO [Service thread] Connection closed by localRank 4
+ip-26-0-172-147:618586:618765 [32575] NCCL INFO [Service thread] Connection closed by localRank 4
+ip-26-0-172-147:618591:618764 [32679] NCCL INFO [Service thread] Connection closed by localRank 4
+ip-26-0-172-147:618587:618770 [32702] NCCL INFO [Service thread] Connection closed by localRank 4
+ip-26-0-172-116:1019289:1019604 [32560] NCCL INFO [Service thread] Connection closed by localRank 5
+ip-26-0-172-116:1019292:1019606 [32602] NCCL INFO [Service thread] Connection closed by localRank 5
+ip-26-0-172-116:1019289:1019470 [32562] NCCL INFO [Service thread] Connection closed by localRank 5
+ip-26-0-172-116:1019292:1019471 [32604] NCCL INFO [Service thread] Connection closed by localRank 5
+ip-26-0-172-147:618590:618905 [32551] NCCL INFO [Service thread] Connection closed by localRank 0
+ip-26-0-172-147:618590:618768 [32553] NCCL INFO [Service thread] Connection closed by localRank 0
+[2025-01-07 05:28:12,235] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 94342 closing signal SIGTERM
+[2025-01-07 05:28:12,235] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 94343 closing signal SIGTERM
+[2025-01-07 05:28:12,235] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 94344 closing signal SIGTERM
+[2025-01-07 05:28:12,235] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 94345 closing signal SIGTERM
+[2025-01-07 05:28:12,235] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 94348 closing signal SIGTERM
+[2025-01-07 05:28:12,235] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 94349 closing signal SIGTERM
+[2025-01-07 05:28:15,832] torch.distributed.elastic.multiprocessing.api: [ERROR] failed (exitcode: 1) local_rank: 4 (pid: 94346) of binary: /fsx/nouamane/miniconda/envs/2-1-cu121/bin/python
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 264, in launch_agent
+    raise ChildFailedError(
+torch.distributed.elastic.multiprocessing.errors.ChildFailedError: 
+============================================================
+/fsx/nouamane/projects/nanotron/run_train.py FAILED
+------------------------------------------------------------
+Failures:
+[1]:
+  time      : 2025-01-07_05:28:12
+  host      : ip-26-0-160-225.ec2.internal
+  rank      : 5 (local_rank: 5)
+  exitcode  : 1 (pid: 94347)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+------------------------------------------------------------
+Root Cause (first observed failure):
+[0]:
+  time      : 2025-01-07_05:28:12
+  host      : ip-26-0-160-225.ec2.internal
+  rank      : 4 (local_rank: 4)
+  exitcode  : 1 (pid: 94346)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+============================================================
+srun: error: ip-26-0-160-225: task 0: Exited with exit code 1
+srun: launch/slurm: _step_signal: Terminating StepId=14098868.0
+[2025-01-07 05:28:16,223] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-172-57.ec2.internal_942539_0' has failed to send a keep-alive heartbeat to the rendezvous '14098868' due to an error of type RendezvousConnectionError.
+[2025-01-07 05:28:16,228] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-07 05:28:16,228] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 618585 closing signal SIGTERM
+[2025-01-07 05:28:16,228] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 618586 closing signal SIGTERM
+[2025-01-07 05:28:16,228] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-07 05:28:16,228] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-07 05:28:16,228] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 618587 closing signal SIGTERM
+[2025-01-07 05:28:16,228] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-07 05:28:16,228] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 618589 closing signal SIGTERM
+[2025-01-07 05:28:16,228] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 618590 closing signal SIGTERM
+[2025-01-07 05:28:16,230] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-07 05:28:16,228] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 942612 closing signal SIGTERM
+[2025-01-07 05:28:16,229] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 880681 closing signal SIGTERM
+[2025-01-07 05:28:16,228] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 618591 closing signal SIGTERM
+[2025-01-07 05:28:16,229] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 942613 closing signal SIGTERM
+[2025-01-07 05:28:16,230] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 688200 closing signal SIGTERM
+[2025-01-07 05:28:16,229] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 942614 closing signal SIGTERM
+[2025-01-07 05:28:16,230] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 688201 closing signal SIGTERM
+[2025-01-07 05:28:16,230] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 688202 closing signal SIGTERM
+[2025-01-07 05:28:16,229] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 942615 closing signal SIGTERM
+[2025-01-07 05:28:16,229] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 942616 closing signal SIGTERM
+[2025-01-07 05:28:16,229] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 942617 closing signal SIGTERM
+[2025-01-07 05:28:16,230] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 688203 closing signal SIGTERM
+[2025-01-07 05:28:16,229] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 880682 closing signal SIGTERM
+[2025-01-07 05:28:16,230] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 688204 closing signal SIGTERM
+[2025-01-07 05:28:16,229] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 880683 closing signal SIGTERM
+[2025-01-07 05:28:16,229] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-07 05:28:16,230] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 688206 closing signal SIGTERM
+[2025-01-07 05:28:16,229] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 880684 closing signal SIGTERM
+[2025-01-07 05:28:16,229] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 926595 closing signal SIGTERM
+[2025-01-07 05:28:16,227] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-07 05:28:16,229] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 880686 closing signal SIGTERM
+[2025-01-07 05:28:16,229] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 926596 closing signal SIGTERM
+[2025-01-07 05:28:16,228] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1019289 closing signal SIGTERM
+[2025-01-07 05:28:16,229] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 926597 closing signal SIGTERM
+[2025-01-07 05:28:16,229] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 926599 closing signal SIGTERM
+[2025-01-07 05:28:16,228] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1019290 closing signal SIGTERM
+[2025-01-07 05:28:16,229] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 926600 closing signal SIGTERM
+[2025-01-07 05:28:16,229] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 926601 closing signal SIGTERM
+[2025-01-07 05:28:16,228] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1019291 closing signal SIGTERM
+[2025-01-07 05:28:16,228] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1019292 closing signal SIGTERM
+[2025-01-07 05:28:16,228] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1019293 closing signal SIGTERM
+[2025-01-07 05:28:16,228] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1019294 closing signal SIGTERM
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 811176 got signal: 15
+[2025-01-07 05:28:16,382] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-172-73.ec2.internal_926522_0' has failed to send a keep-alive heartbeat to the rendezvous '14098868' due to an error of type RendezvousConnectionError.
+srun: error: ip-26-0-173-7: task 7: Exited with exit code 1
+[2025-01-07 05:28:17,072] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-172-147.ec2.internal_618511_0' has failed to send a keep-alive heartbeat to the rendezvous '14098868' due to an error of type RendezvousConnectionError.
+[2025-01-07 05:28:17,107] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-172-116.ec2.internal_1019216_0' has failed to send a keep-alive heartbeat to the rendezvous '14098868' due to an error of type RendezvousConnectionError.
+[2025-01-07 05:28:17,133] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-172-252.ec2.internal_688130_0' has failed to send a keep-alive heartbeat to the rendezvous '14098868' due to an error of type RendezvousConnectionError.
+[2025-01-07 05:28:17,198] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-172-142.ec2.internal_880607_0' has failed to send a keep-alive heartbeat to the rendezvous '14098868' due to an error of type RendezvousConnectionError.
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 688130 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 880607 got signal: 15
+srun: error: ip-26-0-172-252: task 6: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 942539 got signal: 15
+srun: error: ip-26-0-172-142: task 4: Exited with exit code 1
+srun: error: ip-26-0-172-57: task 1: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 1019216 got signal: 15
+srun: error: ip-26-0-172-116: task 3: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 618511 got signal: 15
+srun: error: ip-26-0-172-147: task 5: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 926522 got signal: 15
+srun: error: ip-26-0-172-73: task 2: Exited with exit code 1
diff --git a/logs/14099236-bench_3.57G_dp16_tp4_pp4_acc8_mbs2_seq4096_zero1_tpmodeRED_vocab131k.out b/logs/14099236-bench_3.57G_dp16_tp4_pp4_acc8_mbs2_seq4096_zero1_tpmodeRED_vocab131k.out
new file mode 100644
index 0000000000000000000000000000000000000000..1b2dd38f59afc603b2e547e1bd13597fafe7f490
--- /dev/null
+++ b/logs/14099236-bench_3.57G_dp16_tp4_pp4_acc8_mbs2_seq4096_zero1_tpmodeRED_vocab131k.out
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0b0e265d6a959404f19903b6a3c3c5fa2a4de2947478c1ca57ee2415ed70a6f6
+size 16931408
diff --git a/logs/14113491-bench_1.34G_dp64_tp1_pp1_acc2_mbs2_seq4096_zero1_tpmodeRED_vocab131k.out b/logs/14113491-bench_1.34G_dp64_tp1_pp1_acc2_mbs2_seq4096_zero1_tpmodeRED_vocab131k.out
new file mode 100644
index 0000000000000000000000000000000000000000..0510b9edb826417b64296499b704ec71e9be3f1c
--- /dev/null
+++ b/logs/14113491-bench_1.34G_dp64_tp1_pp1_acc2_mbs2_seq4096_zero1_tpmodeRED_vocab131k.out
@@ -0,0 +1,1109 @@
++ echo 'Running script: /var/spool/slurmd/job14113491/slurm_script'
+Running script: /var/spool/slurmd/job14113491/slurm_script
++ '[' -z 14113491 ']'
++ source /etc/profile.d/modules.sh
+++ . /usr/share/modules/init/bash
++++ unset _mlshdbg
++++ '[' 0 = 1 ']'
++++ unset _mlre _mlIFS
++++ '[' -n x ']'
++++ _mlIFS=' 	
+'
++++ IFS=' '
++++ '[' -n '' ']'
+++++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash autoinit
++++ _mlcode='module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+MODULEPATH_modshare=/etc/environment-modules/modules:1:/usr/share/modules/\$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1; export MODULEPATH_modshare;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+LOADEDMODULES=; export LOADEDMODULES;
+MODULEPATH=/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/\$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles; export MODULEPATH;
+test 0;'
++++ _mlret=0
++++ '[' -n x ']'
++++ IFS=' 	
+'
++++ unset _mlIFS
++++ unset _mlre _mlv _mlrv
++++ '[' 0 -eq 0 ']'
++++ eval 'module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+MODULEPATH_modshare=/etc/environment-modules/modules:1:/usr/share/modules/\$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1; export MODULEPATH_modshare;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+LOADEDMODULES=; export LOADEDMODULES;
+MODULEPATH=/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/\$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles; export MODULEPATH;
+test 0;'
+++++ MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl
+++++ export MODULES_CMD
+++++ ENV=/usr/share/modules/init/profile.sh
+++++ export ENV
+++++ MODULEPATH_modshare='/etc/environment-modules/modules:1:/usr/share/modules/$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1'
+++++ export MODULEPATH_modshare
+++++ BASH_ENV=/usr/share/modules/init/bash
+++++ export BASH_ENV
+++++ MODULESHOME=/usr/share/modules
+++++ export MODULESHOME
+++++ LOADEDMODULES=
+++++ export LOADEDMODULES
+++++ MODULEPATH='/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles'
+++++ export MODULEPATH
+++++ test 0
++++ '[' 0 = 1 ']'
++++ '[' -t 2 ']'
++++ export -f module
++++ export -f switchml
++++ '[' 5 -ge 3 ']'
++++ [[ ehxB =~ i ]]
++++ [[ ! :/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin: =~ :/usr/bin: ]]
++++ '[' '!' -n '' ']'
++++ MANPATH=:
++++ export MANPATH
+++++ manpath
++++ [[ ! :/fsx/nouamane/miniconda/envs/2-1-cu121/man:/fsx/nouamane/miniconda/envs/2-1-cu121/share/man:/fsx/nouamane/miniconda/man:/fsx/nouamane/miniconda/share/man:/usr/local/man:/usr/local/share/man:/usr/share/man:/opt/slurm/share/man:/admin/home/nouamane/.local/share/man:/admin/home/nouamane/.fzf/man:: =~ :/usr/share/man: ]]
++++ unset _mlcode _mlret
++++ '[' -n '' ']'
++ module load cuda/12.1
++ unset _mlshdbg
++ '[' 0 = 1 ']'
++ unset _mlre _mlIFS
++ '[' -n x ']'
++ _mlIFS=' 	
+'
++ IFS=' '
++ '[' -n '' ']'
+++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash load cuda/12.1
++ eval 'CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include;' export 'CPATH;
+LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:;' export 'LD_LIBRARY_PATH;
+FI_EFA_FORK_SAFE=1;' export 'FI_EFA_FORK_SAFE;
+MANPATH=/usr/local/cuda-12.1/share/man::;' export 'MANPATH;
+LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64;' export 'LIBRARY_PATH;
+_LMFILES_=/usr/share/modules/modulefiles/cuda/12.1;' export '_LMFILES_;
+LOADEDMODULES=cuda/12.1;' export 'LOADEDMODULES;
+MPI_PATH=/opt/amazon/openmpi;' export 'MPI_PATH;
+NCCL_HOME_modshare=/opt/nccl/build:1;' export 'NCCL_HOME_modshare;
+NCCL_PROTO=simple;' export 'NCCL_PROTO;
+MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1;' export 'MANPATH_modshare;
+LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1;' export 'LIBRARY_PATH_modshare;
+NCCL_SOCKET_IFNAME=enp;' export 'NCCL_SOCKET_IFNAME;
+AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl;' export 'AWS_OFI_NCCL_HOME;
+NCCL_HOME=/opt/nccl/build;' export 'NCCL_HOME;
+FI_PROVIDER=efa;' export 'FI_PROVIDER;
+AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1;' export 'AWS_OFI_NCCL_HOME_modshare;
+CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1;' export 'CPATH_modshare;
+LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2::1:/opt/nccl/build/lib:1:/opt/aws-ofi-nccl/lib:2:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/efa/lib:1;' export 'LD_LIBRARY_PATH_modshare;
+FI_EFA_ENABLE_SHM_TRANSFER=1;' export 'FI_EFA_ENABLE_SHM_TRANSFER;
+_LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1;' export '_LMFILES__modshare;
+LOADEDMODULES_modshare=cuda/12.1:1;' export 'LOADEDMODULES_modshare;
+MPI_PATH_modshare=/opt/amazon/openmpi:1;' export 'MPI_PATH_modshare;
+PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin;' export 'PATH;
+CUDA_HOME=/usr/local/cuda-12.1;' export 'CUDA_HOME;
+PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/opt/slurm/bin:1:/usr/sbin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1;' export 'PATH_modshare;
+test' '0;'
+++ CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include
+++ export CPATH
+++ LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:
+++ export LD_LIBRARY_PATH
+++ FI_EFA_FORK_SAFE=1
+++ export FI_EFA_FORK_SAFE
+++ MANPATH=/usr/local/cuda-12.1/share/man::
+++ export MANPATH
+++ LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64
+++ export LIBRARY_PATH
+++ _LMFILES_=/usr/share/modules/modulefiles/cuda/12.1
+++ export _LMFILES_
+++ LOADEDMODULES=cuda/12.1
+++ export LOADEDMODULES
+++ MPI_PATH=/opt/amazon/openmpi
+++ export MPI_PATH
+++ NCCL_HOME_modshare=/opt/nccl/build:1
+++ export NCCL_HOME_modshare
+++ NCCL_PROTO=simple
+++ export NCCL_PROTO
+++ MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1
+++ export MANPATH_modshare
+++ LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1
+++ export LIBRARY_PATH_modshare
+++ NCCL_SOCKET_IFNAME=enp
+++ export NCCL_SOCKET_IFNAME
+++ AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl
+++ export AWS_OFI_NCCL_HOME
+++ NCCL_HOME=/opt/nccl/build
+++ export NCCL_HOME
+++ FI_PROVIDER=efa
+++ export FI_PROVIDER
+++ AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1
+++ export AWS_OFI_NCCL_HOME_modshare
+++ CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1
+++ export CPATH_modshare
+++ LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2::1:/opt/nccl/build/lib:1:/opt/aws-ofi-nccl/lib:2:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/efa/lib:1
+++ export LD_LIBRARY_PATH_modshare
+++ FI_EFA_ENABLE_SHM_TRANSFER=1
+++ export FI_EFA_ENABLE_SHM_TRANSFER
+++ _LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1
+++ export _LMFILES__modshare
+++ LOADEDMODULES_modshare=cuda/12.1:1
+++ export LOADEDMODULES_modshare
+++ MPI_PATH_modshare=/opt/amazon/openmpi:1
+++ export MPI_PATH_modshare
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
+++ export PATH
+++ CUDA_HOME=/usr/local/cuda-12.1
+++ export CUDA_HOME
+++ PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/opt/slurm/bin:1:/usr/sbin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1
+++ export PATH_modshare
+++ test 0
++ _mlstatus=0
++ '[' -n x ']'
++ IFS=' 	
+'
++ unset _mlre _mlv _mlrv _mlIFS
++ '[' -n '' ']'
++ unset _mlshdbg
++ return 0
++ source /fsx/nouamane/miniconda/bin/activate
+++ _CONDA_ROOT=/fsx/nouamane/miniconda
+++ . /fsx/nouamane/miniconda/etc/profile.d/conda.sh
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ '[' -z x ']'
+++ conda activate
+++ local cmd=activate
+++ case "$cmd" in
+++ __conda_activate activate
+++ '[' -n '' ']'
+++ local ask_conda
++++ PS1=
++++ __conda_exe shell.posix activate
++++ /fsx/nouamane/miniconda/bin/conda shell.posix activate
+Error while loading conda entry point: conda-libmamba-solver (module 'libmambapy' has no attribute 'QueryFormat')
+++ ask_conda='. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''2'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_1='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/etc/conda/activate.d/libxml2_activate.sh"'
+++ eval '. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''2'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_1='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/etc/conda/activate.d/libxml2_activate.sh"'
++++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh
+++++ test -n ''
+++++ unset XML_CATALOG_FILES
+++++ unset xml_catalog_files_libxml2
++++ PS1='(base) '
++++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
++++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
++++ export CONDA_PREFIX=/fsx/nouamane/miniconda
++++ CONDA_PREFIX=/fsx/nouamane/miniconda
++++ export CONDA_SHLVL=2
++++ CONDA_SHLVL=2
++++ export CONDA_DEFAULT_ENV=base
++++ CONDA_DEFAULT_ENV=base
++++ export 'CONDA_PROMPT_MODIFIER=(base) '
++++ CONDA_PROMPT_MODIFIER='(base) '
++++ export CONDA_PREFIX_1=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ CONDA_PREFIX_1=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ . /fsx/nouamane/miniconda/etc/conda/activate.d/libxml2_activate.sh
+++++ test -n ''
+++++ xml_catalog_files_libxml2=
+++++ XML_CATALOG_FILES=
+++++ conda_catalog_files=
+++++ ifs_libxml2=' 	
+'
+++++ IFS=' '
+++++ rem=/fsx/nouamane/miniconda
+++++ for pre in ${rem}
+++++ test '' = /fsx/nouamane/miniconda
+++++ conda_catalog_files=/fsx/nouamane/miniconda
+++++ rem=
+++++ IFS=' 	
+'
+++++ conda_catalog_files='file:///fsx/nouamane/miniconda/etc/xml/catalog file:///etc/xml/catalog'
+++++ export 'XML_CATALOG_FILES=file:///fsx/nouamane/miniconda/etc/xml/catalog file:///etc/xml/catalog'
+++++ XML_CATALOG_FILES='file:///fsx/nouamane/miniconda/etc/xml/catalog file:///etc/xml/catalog'
+++++ unset conda_catalog_files ifs_libxml2 rem
+++ __conda_hashr
+++ '[' -n '' ']'
+++ '[' -n '' ']'
+++ hash -r
++ conda activate 2-1-cu121
++ local cmd=activate
++ case "$cmd" in
++ __conda_activate activate 2-1-cu121
++ '[' -n '' ']'
++ local ask_conda
+++ PS1='(base) '
+++ __conda_exe shell.posix activate 2-1-cu121
+++ /fsx/nouamane/miniconda/bin/conda shell.posix activate 2-1-cu121
+Error while loading conda entry point: conda-libmamba-solver (module 'libmambapy' has no attribute 'QueryFormat')
++ ask_conda='. "/fsx/nouamane/miniconda/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_2='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
++ eval '. "/fsx/nouamane/miniconda/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_2='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
+++ . /fsx/nouamane/miniconda/etc/conda/deactivate.d/libxml2_deactivate.sh
++++ test -n ''
++++ unset XML_CATALOG_FILES
++++ unset xml_catalog_files_libxml2
+++ PS1='(2-1-cu121) '
+++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
+++ export CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ export CONDA_SHLVL=3
+++ CONDA_SHLVL=3
+++ export CONDA_DEFAULT_ENV=2-1-cu121
+++ CONDA_DEFAULT_ENV=2-1-cu121
+++ export 'CONDA_PROMPT_MODIFIER=(2-1-cu121) '
+++ CONDA_PROMPT_MODIFIER='(2-1-cu121) '
+++ export CONDA_PREFIX_2=/fsx/nouamane/miniconda
+++ CONDA_PREFIX_2=/fsx/nouamane/miniconda
+++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ export _CE_M=
+++ _CE_M=
+++ export _CE_CONDA=
+++ _CE_CONDA=
+++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh
++++ test -n ''
++++ xml_catalog_files_libxml2=
++++ XML_CATALOG_FILES=
++++ conda_catalog_files=
++++ ifs_libxml2=' 	
+'
++++ IFS=' '
++++ rem=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ for pre in ${rem}
++++ test '' = /fsx/nouamane/miniconda/envs/2-1-cu121
++++ conda_catalog_files=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ rem=
++++ IFS=' 	
+'
++++ conda_catalog_files='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ export 'XML_CATALOG_FILES=file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ XML_CATALOG_FILES='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ unset conda_catalog_files ifs_libxml2 rem
++ __conda_hashr
++ '[' -n '' ']'
++ '[' -n '' ']'
++ hash -r
++ export PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
++ PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
++ '[' -z '' ']'
+++ scontrol show hostnames 'ip-26-0-167-[217,245],ip-26-0-168-[30,34,52,95,120,238]'
++ export 'NODELIST=ip-26-0-167-217
+ip-26-0-167-245
+ip-26-0-168-30
+ip-26-0-168-34
+ip-26-0-168-52
+ip-26-0-168-95
+ip-26-0-168-120
+ip-26-0-168-238'
++ NODELIST='ip-26-0-167-217
+ip-26-0-167-245
+ip-26-0-168-30
+ip-26-0-168-34
+ip-26-0-168-52
+ip-26-0-168-95
+ip-26-0-168-120
+ip-26-0-168-238'
+++ head -n1
+++ scontrol show hostnames 'ip-26-0-167-[217,245],ip-26-0-168-[30,34,52,95,120,238]'
++ export MASTER_NODE=ip-26-0-167-217
++ MASTER_NODE=ip-26-0-167-217
++ export MASTER_PORT=12356
++ MASTER_PORT=12356
++ export NNODES=8
++ NNODES=8
++ export GPUS_PER_NODE=8
++ GPUS_PER_NODE=8
++ export WORLD_SIZE=64
++ WORLD_SIZE=64
++ export CUDA_DEVICE_MAX_CONNECTIONS=1
++ CUDA_DEVICE_MAX_CONNECTIONS=1
++ export NCCL_DEBUG=WARN
++ NCCL_DEBUG=WARN
++ export NANOTRON_BENCHMARK=1
++ NANOTRON_BENCHMARK=1
++ export WANDB_MODE=disabled
++ WANDB_MODE=disabled
++ export TORCH_NCCL_ASYNC_ERROR_HANDLING=1
++ TORCH_NCCL_ASYNC_ERROR_HANDLING=1
++ export TORCH_DISTRIBUTED_DEBUG=DETAIL
++ TORCH_DISTRIBUTED_DEBUG=DETAIL
++ '[' -z '' ']'
++ echo '=== GPU Topology ==='
+=== GPU Topology ===
++ nvidia-smi topo -m
+	GPU0	GPU1	GPU2	GPU3	GPU4	GPU5	GPU6	GPU7	CPU Affinity	NUMA Affinity	GPU NUMA ID
+GPU0	 X 	NV18	NV18	NV18	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU1	NV18	 X 	NV18	NV18	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU2	NV18	NV18	 X 	NV18	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU3	NV18	NV18	NV18	 X 	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU4	NV18	NV18	NV18	NV18	 X 	NV18	NV18	NV18	48-95	1		N/A
+GPU5	NV18	NV18	NV18	NV18	NV18	 X 	NV18	NV18	48-95	1		N/A
+GPU6	NV18	NV18	NV18	NV18	NV18	NV18	 X 	NV18	48-95	1		N/A
+GPU7	NV18	NV18	NV18	NV18	NV18	NV18	NV18	 X 	48-95	1		N/A
+
+Legend:
+
+  X    = Self
+  SYS  = Connection traversing PCIe as well as the SMP interconnect between NUMA nodes (e.g., QPI/UPI)
+  NODE = Connection traversing PCIe as well as the interconnect between PCIe Host Bridges within a NUMA node
+  PHB  = Connection traversing PCIe as well as a PCIe Host Bridge (typically the CPU)
+  PXB  = Connection traversing multiple PCIe bridges (without traversing the PCIe Host Bridge)
+  PIX  = Connection traversing at most a single PCIe bridge
+  NV#  = Connection traversing a bonded set of # NVLinks
++ echo ==================
+==================
++ export SRUN_ALLOC_ARGS=
++ SRUN_ALLOC_ARGS=
++ echo 'Master node: ip-26-0-167-217'
+Master node: ip-26-0-167-217
++ echo 'All nodes: ip-26-0-167-217
+ip-26-0-167-245
+ip-26-0-168-30
+ip-26-0-168-34
+ip-26-0-168-52
+ip-26-0-168-95
+ip-26-0-168-120
+ip-26-0-168-238'
+All nodes: ip-26-0-167-217
+ip-26-0-167-245
+ip-26-0-168-30
+ip-26-0-168-34
+ip-26-0-168-52
+ip-26-0-168-95
+ip-26-0-168-120
+ip-26-0-168-238
++ echo 'World size: 64'
+World size: 64
++ '[' -n '' ']'
++ srun --wait=0 --kill-on-bad-exit=1 torchrun --nnodes=8 --nproc_per_node=8 --rdzv_id=14113491 --rdzv_backend=c10d --rdzv_endpoint=ip-26-0-167-217:12356 --max_restarts 0 --rdzv_conf timeout=60 /fsx/nouamane/projects/nanotron/run_train.py --config-file benchmark/configs/config_1.34G_dp64_tp1_pp1_acc2_mbs2_seq4096_zero1_tpmodeRED_vocab131k.yaml
+[2025-01-07 18:04:46,271] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-07 18:04:46,299] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-07 18:04:46,309] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-07 18:04:46,271] torch.distributed.run: [WARNING] 
+[2025-01-07 18:04:46,271] torch.distributed.run: [WARNING] *****************************************
+[2025-01-07 18:04:46,271] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-07 18:04:46,271] torch.distributed.run: [WARNING] *****************************************
+[2025-01-07 18:04:46,336] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-07 18:04:46,299] torch.distributed.run: [WARNING] 
+[2025-01-07 18:04:46,299] torch.distributed.run: [WARNING] *****************************************
+[2025-01-07 18:04:46,299] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-07 18:04:46,299] torch.distributed.run: [WARNING] *****************************************
+[2025-01-07 18:04:46,347] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-07 18:04:46,309] torch.distributed.run: [WARNING] 
+[2025-01-07 18:04:46,309] torch.distributed.run: [WARNING] *****************************************
+[2025-01-07 18:04:46,309] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-07 18:04:46,309] torch.distributed.run: [WARNING] *****************************************
+[2025-01-07 18:04:46,363] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-07 18:04:46,337] torch.distributed.run: [WARNING] 
+[2025-01-07 18:04:46,337] torch.distributed.run: [WARNING] *****************************************
+[2025-01-07 18:04:46,337] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-07 18:04:46,337] torch.distributed.run: [WARNING] *****************************************
+[2025-01-07 18:04:46,347] torch.distributed.run: [WARNING] 
+[2025-01-07 18:04:46,347] torch.distributed.run: [WARNING] *****************************************
+[2025-01-07 18:04:46,347] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-07 18:04:46,347] torch.distributed.run: [WARNING] *****************************************
+[2025-01-07 18:04:46,408] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-07 18:04:46,408] torch.distributed.run: [WARNING] 
+[2025-01-07 18:04:46,408] torch.distributed.run: [WARNING] *****************************************
+[2025-01-07 18:04:46,408] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-07 18:04:46,408] torch.distributed.run: [WARNING] *****************************************
+[2025-01-07 18:04:46,363] torch.distributed.run: [WARNING] 
+[2025-01-07 18:04:46,363] torch.distributed.run: [WARNING] *****************************************
+[2025-01-07 18:04:46,363] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-07 18:04:46,363] torch.distributed.run: [WARNING] *****************************************
+[2025-01-07 18:04:46,693] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-07 18:04:46,693] torch.distributed.run: [WARNING] 
+[2025-01-07 18:04:46,693] torch.distributed.run: [WARNING] *****************************************
+[2025-01-07 18:04:46,693] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-07 18:04:46,693] torch.distributed.run: [WARNING] *****************************************
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+01/07/2025 18:05:31 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]: Config:
+01/07/2025 18:05:31 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]: Config(general=GeneralArgs(project='debug',
+01/07/2025 18:05:31 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]:                            run='1.34G_dp64_tp1_pp1_acc2_mbs2_seq4096_zero1_tpmodeRED_vocab131k',
+01/07/2025 18:05:31 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]:                            seed=42,
+01/07/2025 18:05:31 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]:                            step=None,
+01/07/2025 18:05:31 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]:                            consumed_train_samples=None,
+01/07/2025 18:05:31 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]:                            benchmark_csv_path=PosixPath('benchmark/results/bench_final2.csv'),
+01/07/2025 18:05:31 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]:                            ignore_sanity_checks=True),
+01/07/2025 18:05:31 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]:        parallelism=ParallelismArgs(dp=64,
+01/07/2025 18:05:31 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]:                                    pp=1,
+01/07/2025 18:05:31 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]:                                    tp=1,
+01/07/2025 18:05:31 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]:                                    pp_engine=<nanotron.parallel.pipeline_parallel.engine.OneForwardOneBackwardPipelineEngine object at 0x7f1a95bdcc70>,
+01/07/2025 18:05:31 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]:                                    tp_mode=<TensorParallelLinearMode.REDUCE_SCATTER: 2>,
+01/07/2025 18:05:31 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]:                                    tp_linear_async_communication=True,
+01/07/2025 18:05:31 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]:                                    recompute_layer=False,
+01/07/2025 18:05:31 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]:                                    tp_recompute_allgather=True,
+01/07/2025 18:05:31 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]:                                    expert_parallel_size=1),
+01/07/2025 18:05:31 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]:        model=ModelArgs(model_config=LlamaConfig(bos_token_id=0,
+01/07/2025 18:05:31 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]:                                                 eos_token_id=0,
+01/07/2025 18:05:31 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]:                                                 hidden_act='silu',
+01/07/2025 18:05:31 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]:                                                 hidden_size=2048,
+01/07/2025 18:05:31 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]:                                                 initializer_range=0.02,
+01/07/2025 18:05:31 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]:                                                 intermediate_size=8192,
+01/07/2025 18:05:31 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]:                                                 is_llama_config=True,
+01/07/2025 18:05:31 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]:                                                 max_position_embeddings=4096,
+01/07/2025 18:05:31 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]:                                                 num_attention_heads=32,
+01/07/2025 18:05:31 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]:                                                 num_hidden_layers=16,
+01/07/2025 18:05:31 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]:                                                 num_key_value_heads=32,
+01/07/2025 18:05:31 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]:                                                 pad_token_id=None,
+01/07/2025 18:05:31 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]:                                                 pretraining_tp=1,
+01/07/2025 18:05:31 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]:                                                 rms_norm_eps=1e-05,
+01/07/2025 18:05:31 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]:                                                 rope_scaling=None,
+01/07/2025 18:05:31 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]:                                                 rope_theta=10000.0,
+01/07/2025 18:05:31 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]:                                                 rope_interleaved=False,
+01/07/2025 18:05:31 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]:                                                 tie_word_embeddings=True,
+01/07/2025 18:05:31 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]:                                                 use_cache=True,
+01/07/2025 18:05:31 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]:                                                 vocab_size=131072),
+01/07/2025 18:05:31 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]:                        init_method=RandomInit(std=0.02),
+01/07/2025 18:05:31 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]:                        dtype=torch.bfloat16,
+01/07/2025 18:05:31 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]:                        make_vocab_size_divisible_by=1,
+01/07/2025 18:05:31 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]:                        ddp_bucket_cap_mb=25),
+01/07/2025 18:05:31 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]:        tokenizer=TokenizerArgs(tokenizer_name_or_path='robot-test/dummy-tokenizer-wordlevel',
+01/07/2025 18:05:31 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]:                                tokenizer_revision=None,
+01/07/2025 18:05:31 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]:                                tokenizer_max_length=None),
+01/07/2025 18:05:31 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]:        checkpoints=CheckpointsArgs(checkpoints_path=PosixPath('checkpoints'),
+01/07/2025 18:05:31 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]:                                    checkpoint_interval=10000,
+01/07/2025 18:05:31 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]:                                    save_initial_state=False,
+01/07/2025 18:05:31 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]:                                    save_final_state=False,
+01/07/2025 18:05:31 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]:                                    resume_checkpoint_path=None,
+01/07/2025 18:05:31 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]:                                    checkpoints_path_is_shared_file_system=False),
+01/07/2025 18:05:31 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]:        logging=LoggingArgs(log_level='info',
+01/07/2025 18:05:31 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]:                            log_level_replica='info',
+01/07/2025 18:05:31 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]:                            iteration_step_info_interval=1),
+01/07/2025 18:05:31 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]:        tokens=TokensArgs(sequence_length=4096,
+01/07/2025 18:05:31 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]:                          train_steps=100,
+01/07/2025 18:05:31 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]:                          micro_batch_size=2,
+01/07/2025 18:05:31 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]:                          batch_accumulation_per_replica=2,
+01/07/2025 18:05:31 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]:                          val_check_interval=100,
+01/07/2025 18:05:31 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]:                          limit_val_batches=0,
+01/07/2025 18:05:31 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]:                          limit_test_batches=0),
+01/07/2025 18:05:31 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]:        optimizer=OptimizerArgs(optimizer_factory=AdamWOptimizerArgs(adam_eps=1e-08,
+01/07/2025 18:05:31 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]:                                                                     adam_beta1=0.9,
+01/07/2025 18:05:31 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]:                                                                     adam_beta2=0.95,
+01/07/2025 18:05:31 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]:                                                                     torch_adam_is_fused=True,
+01/07/2025 18:05:31 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]:                                                                     name='adamW'),
+01/07/2025 18:05:31 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]:                                zero_stage=1,
+01/07/2025 18:05:31 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]:                                weight_decay=0.01,
+01/07/2025 18:05:31 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]:                                clip_grad=1.0,
+01/07/2025 18:05:31 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]:                                accumulate_grad_in_fp32=True,
+01/07/2025 18:05:31 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]:                                learning_rate_scheduler=LRSchedulerArgs(learning_rate=0.0003,
+01/07/2025 18:05:31 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]:                                                                        lr_warmup_steps=2,
+01/07/2025 18:05:31 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]:                                                                        lr_warmup_style='linear',
+01/07/2025 18:05:31 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]:                                                                        lr_decay_style='cosine',
+01/07/2025 18:05:31 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]:                                                                        lr_decay_steps=13,
+01/07/2025 18:05:31 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]:                                                                        lr_decay_starting_step=None,
+01/07/2025 18:05:31 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]:                                                                        min_decay_lr=1e-05)),
+01/07/2025 18:05:31 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]:        data_stages=[DatasetStageArgs(name='Stable Training Stage',
+01/07/2025 18:05:31 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]:                                      start_training_step=1,
+01/07/2025 18:05:31 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]:                                      data=DataArgs(dataset=None,
+01/07/2025 18:05:31 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]:                                                    seed=42,
+01/07/2025 18:05:31 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]:                                                    num_loading_workers=1))],
+01/07/2025 18:05:31 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]:        profiler=None,
+01/07/2025 18:05:31 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]:        lighteval=None,
+01/07/2025 18:05:31 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]:        s3_upload=None)
+01/07/2025 18:05:31 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]: Model Config:
+01/07/2025 18:05:31 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]: LlamaConfig(bos_token_id=0,
+01/07/2025 18:05:31 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]:             eos_token_id=0,
+01/07/2025 18:05:31 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]:             hidden_act='silu',
+01/07/2025 18:05:31 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]:             hidden_size=2048,
+01/07/2025 18:05:31 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]:             initializer_range=0.02,
+01/07/2025 18:05:31 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]:             intermediate_size=8192,
+01/07/2025 18:05:31 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]:             is_llama_config=True,
+01/07/2025 18:05:31 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]:             max_position_embeddings=4096,
+01/07/2025 18:05:31 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]:             num_attention_heads=32,
+01/07/2025 18:05:31 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]:             num_hidden_layers=16,
+01/07/2025 18:05:31 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]:             num_key_value_heads=32,
+01/07/2025 18:05:31 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]:             pad_token_id=None,
+01/07/2025 18:05:31 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]:             pretraining_tp=1,
+01/07/2025 18:05:31 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]:             rms_norm_eps=1e-05,
+01/07/2025 18:05:31 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]:             rope_scaling=None,
+01/07/2025 18:05:31 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]:             rope_theta=10000.0,
+01/07/2025 18:05:31 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]:             rope_interleaved=False,
+01/07/2025 18:05:31 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]:             tie_word_embeddings=True,
+01/07/2025 18:05:31 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]:             use_cache=True,
+01/07/2025 18:05:31 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]:             vocab_size=131072)
+01/07/2025 18:05:31 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]: Building model..
+01/07/2025 18:05:31 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]: Initialize RoPE Theta = 10000.0
+01/07/2025 18:05:31 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]: Setting PP block ranks...
+01/07/2025 18:05:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]: module_name: model.token_position_embeddings | PP: 0/1 | Block rank: 0
+01/07/2025 18:05:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]: module_name: model.decoder.0 | PP: 0/1 | Block rank: 1
+01/07/2025 18:05:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]: module_name: model.decoder.1 | PP: 0/1 | Block rank: 2
+01/07/2025 18:05:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]: module_name: model.decoder.2 | PP: 0/1 | Block rank: 3
+01/07/2025 18:05:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]: module_name: model.decoder.3 | PP: 0/1 | Block rank: 4
+01/07/2025 18:05:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]: module_name: model.decoder.4 | PP: 0/1 | Block rank: 5
+01/07/2025 18:05:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]: module_name: model.decoder.5 | PP: 0/1 | Block rank: 6
+01/07/2025 18:05:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]: module_name: model.decoder.6 | PP: 0/1 | Block rank: 7
+01/07/2025 18:05:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]: module_name: model.decoder.7 | PP: 0/1 | Block rank: 8
+01/07/2025 18:05:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]: module_name: model.decoder.8 | PP: 0/1 | Block rank: 9
+01/07/2025 18:05:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]: module_name: model.decoder.9 | PP: 0/1 | Block rank: 10
+01/07/2025 18:05:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]: module_name: model.decoder.10 | PP: 0/1 | Block rank: 11
+01/07/2025 18:05:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]: module_name: model.decoder.11 | PP: 0/1 | Block rank: 12
+01/07/2025 18:05:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]: module_name: model.decoder.12 | PP: 0/1 | Block rank: 13
+01/07/2025 18:05:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]: module_name: model.decoder.13 | PP: 0/1 | Block rank: 14
+01/07/2025 18:05:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]: module_name: model.decoder.14 | PP: 0/1 | Block rank: 15
+01/07/2025 18:05:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]: module_name: model.decoder.15 | PP: 0/1 | Block rank: 16
+01/07/2025 18:05:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]: module_name: model.final_layer_norm | PP: 0/1 | Block rank: 17
+01/07/2025 18:05:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]: module_name: model.lm_head | PP: 0/1 | Block rank: 18
+01/07/2025 18:05:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]: module_name: model.cast_to_fp32 | PP: 0/1 | Block rank: 19
+01/07/2025 18:05:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]: module_name: loss | PP: 0/1 | Block rank: 20
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+01/07/2025 18:05:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]: Total number of parameters: 1.34G (2560.13MiB)
+01/07/2025 18:05:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]: Local number of parameters: 1.34G (2560.13MiB)
+01/07/2025 18:05:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]: [After model building] Memory usage: 2560.15MiB. Peak allocated: 5408.00MiB Peak reserved: 10530.00MiB
+01/07/2025 18:05:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]: No checkpoint path provided.
+01/07/2025 18:05:33 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]: Parametrizing model parameters using StandardParametrizator
+01/07/2025 18:05:44 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]: [Optimizer Building] Using LearningRateForSP as learning rate
+01/07/2025 18:05:44 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]: [ZeRO sharding] Size of optimizer params per rank:
+01/07/2025 18:05:44 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]: [ZeRO sharding] DP Rank 0 has 21M out of 1.34G (1.56%) params' optimizer states
+01/07/2025 18:05:44 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]: [ZeRO sharding] DP Rank 1 has 21M out of 1.34G (1.56%) params' optimizer states
+01/07/2025 18:05:44 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]: [ZeRO sharding] DP Rank 2 has 21M out of 1.34G (1.56%) params' optimizer states
+01/07/2025 18:05:44 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]: [ZeRO sharding] DP Rank 3 has 21M out of 1.34G (1.56%) params' optimizer states
+01/07/2025 18:05:44 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]: [ZeRO sharding] DP Rank 4 has 21M out of 1.34G (1.56%) params' optimizer states
+01/07/2025 18:05:44 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]: [ZeRO sharding] DP Rank 5 has 21M out of 1.34G (1.56%) params' optimizer states
+01/07/2025 18:05:44 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]: [ZeRO sharding] DP Rank 6 has 21M out of 1.34G (1.56%) params' optimizer states
+01/07/2025 18:05:44 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]: [ZeRO sharding] DP Rank 7 has 21M out of 1.34G (1.56%) params' optimizer states
+01/07/2025 18:05:44 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]: [ZeRO sharding] DP Rank 8 has 21M out of 1.34G (1.56%) params' optimizer states
+01/07/2025 18:05:44 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]: [ZeRO sharding] DP Rank 9 has 21M out of 1.34G (1.56%) params' optimizer states
+01/07/2025 18:05:44 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]: [ZeRO sharding] DP Rank 10 has 21M out of 1.34G (1.56%) params' optimizer states
+01/07/2025 18:05:44 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]: [ZeRO sharding] DP Rank 11 has 21M out of 1.34G (1.56%) params' optimizer states
+01/07/2025 18:05:44 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]: [ZeRO sharding] DP Rank 12 has 21M out of 1.34G (1.56%) params' optimizer states
+01/07/2025 18:05:44 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]: [ZeRO sharding] DP Rank 13 has 21M out of 1.34G (1.56%) params' optimizer states
+01/07/2025 18:05:44 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]: [ZeRO sharding] DP Rank 14 has 21M out of 1.34G (1.56%) params' optimizer states
+01/07/2025 18:05:44 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]: [ZeRO sharding] DP Rank 15 has 21M out of 1.34G (1.56%) params' optimizer states
+01/07/2025 18:05:44 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]: [ZeRO sharding] DP Rank 16 has 21M out of 1.34G (1.56%) params' optimizer states
+01/07/2025 18:05:44 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]: [ZeRO sharding] DP Rank 17 has 21M out of 1.34G (1.56%) params' optimizer states
+01/07/2025 18:05:44 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]: [ZeRO sharding] DP Rank 18 has 21M out of 1.34G (1.56%) params' optimizer states
+01/07/2025 18:05:44 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]: [ZeRO sharding] DP Rank 19 has 21M out of 1.34G (1.56%) params' optimizer states
+01/07/2025 18:05:44 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]: [ZeRO sharding] DP Rank 20 has 21M out of 1.34G (1.56%) params' optimizer states
+01/07/2025 18:05:44 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]: [ZeRO sharding] DP Rank 21 has 21M out of 1.34G (1.56%) params' optimizer states
+01/07/2025 18:05:44 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]: [ZeRO sharding] DP Rank 22 has 21M out of 1.34G (1.56%) params' optimizer states
+01/07/2025 18:05:44 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]: [ZeRO sharding] DP Rank 23 has 21M out of 1.34G (1.56%) params' optimizer states
+01/07/2025 18:05:44 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]: [ZeRO sharding] DP Rank 24 has 21M out of 1.34G (1.56%) params' optimizer states
+01/07/2025 18:05:44 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]: [ZeRO sharding] DP Rank 25 has 21M out of 1.34G (1.56%) params' optimizer states
+01/07/2025 18:05:44 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]: [ZeRO sharding] DP Rank 26 has 21M out of 1.34G (1.56%) params' optimizer states
+01/07/2025 18:05:44 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]: [ZeRO sharding] DP Rank 27 has 21M out of 1.34G (1.56%) params' optimizer states
+01/07/2025 18:05:44 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]: [ZeRO sharding] DP Rank 28 has 21M out of 1.34G (1.56%) params' optimizer states
+01/07/2025 18:05:44 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]: [ZeRO sharding] DP Rank 29 has 21M out of 1.34G (1.56%) params' optimizer states
+01/07/2025 18:05:44 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]: [ZeRO sharding] DP Rank 30 has 21M out of 1.34G (1.56%) params' optimizer states
+01/07/2025 18:05:44 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]: [ZeRO sharding] DP Rank 31 has 21M out of 1.34G (1.56%) params' optimizer states
+01/07/2025 18:05:44 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]: [ZeRO sharding] DP Rank 32 has 21M out of 1.34G (1.56%) params' optimizer states
+01/07/2025 18:05:44 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]: [ZeRO sharding] DP Rank 33 has 21M out of 1.34G (1.56%) params' optimizer states
+01/07/2025 18:05:44 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]: [ZeRO sharding] DP Rank 34 has 21M out of 1.34G (1.56%) params' optimizer states
+01/07/2025 18:05:44 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]: [ZeRO sharding] DP Rank 35 has 21M out of 1.34G (1.56%) params' optimizer states
+01/07/2025 18:05:44 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]: [ZeRO sharding] DP Rank 36 has 21M out of 1.34G (1.56%) params' optimizer states
+01/07/2025 18:05:44 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]: [ZeRO sharding] DP Rank 37 has 21M out of 1.34G (1.56%) params' optimizer states
+01/07/2025 18:05:44 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]: [ZeRO sharding] DP Rank 38 has 21M out of 1.34G (1.56%) params' optimizer states
+01/07/2025 18:05:44 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]: [ZeRO sharding] DP Rank 39 has 21M out of 1.34G (1.56%) params' optimizer states
+01/07/2025 18:05:44 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]: [ZeRO sharding] DP Rank 40 has 21M out of 1.34G (1.56%) params' optimizer states
+01/07/2025 18:05:44 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]: [ZeRO sharding] DP Rank 41 has 21M out of 1.34G (1.56%) params' optimizer states
+01/07/2025 18:05:44 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]: [ZeRO sharding] DP Rank 42 has 21M out of 1.34G (1.56%) params' optimizer states
+01/07/2025 18:05:44 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]: [ZeRO sharding] DP Rank 43 has 21M out of 1.34G (1.56%) params' optimizer states
+01/07/2025 18:05:44 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]: [ZeRO sharding] DP Rank 44 has 21M out of 1.34G (1.56%) params' optimizer states
+01/07/2025 18:05:44 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]: [ZeRO sharding] DP Rank 45 has 21M out of 1.34G (1.56%) params' optimizer states
+01/07/2025 18:05:44 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]: [ZeRO sharding] DP Rank 46 has 21M out of 1.34G (1.56%) params' optimizer states
+01/07/2025 18:05:44 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]: [ZeRO sharding] DP Rank 47 has 21M out of 1.34G (1.56%) params' optimizer states
+01/07/2025 18:05:44 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]: [ZeRO sharding] DP Rank 48 has 21M out of 1.34G (1.56%) params' optimizer states
+01/07/2025 18:05:44 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]: [ZeRO sharding] DP Rank 49 has 21M out of 1.34G (1.56%) params' optimizer states
+01/07/2025 18:05:44 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]: [ZeRO sharding] DP Rank 50 has 21M out of 1.34G (1.56%) params' optimizer states
+01/07/2025 18:05:44 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]: [ZeRO sharding] DP Rank 51 has 21M out of 1.34G (1.56%) params' optimizer states
+01/07/2025 18:05:44 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]: [ZeRO sharding] DP Rank 52 has 21M out of 1.34G (1.56%) params' optimizer states
+01/07/2025 18:05:44 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]: [ZeRO sharding] DP Rank 53 has 21M out of 1.34G (1.56%) params' optimizer states
+01/07/2025 18:05:44 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]: [ZeRO sharding] DP Rank 54 has 21M out of 1.34G (1.56%) params' optimizer states
+01/07/2025 18:05:44 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]: [ZeRO sharding] DP Rank 55 has 21M out of 1.34G (1.56%) params' optimizer states
+01/07/2025 18:05:44 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]: [ZeRO sharding] DP Rank 56 has 21M out of 1.34G (1.56%) params' optimizer states
+01/07/2025 18:05:44 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]: [ZeRO sharding] DP Rank 57 has 21M out of 1.34G (1.56%) params' optimizer states
+01/07/2025 18:05:44 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]: [ZeRO sharding] DP Rank 58 has 21M out of 1.34G (1.56%) params' optimizer states
+01/07/2025 18:05:44 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]: [ZeRO sharding] DP Rank 59 has 21M out of 1.34G (1.56%) params' optimizer states
+01/07/2025 18:05:44 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]: [ZeRO sharding] DP Rank 60 has 21M out of 1.34G (1.56%) params' optimizer states
+01/07/2025 18:05:44 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]: [ZeRO sharding] DP Rank 61 has 21M out of 1.34G (1.56%) params' optimizer states
+01/07/2025 18:05:44 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]: [ZeRO sharding] DP Rank 62 has 21M out of 1.34G (1.56%) params' optimizer states
+01/07/2025 18:05:44 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]: [ZeRO sharding] DP Rank 63 has 21M out of 1.34G (1.56%) params' optimizer states
+01/07/2025 18:05:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]: [Training Plan] Stage Stable Training Stage has 99 remaining training steps and has consumed 0 samples
+01/07/2025 18:05:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]: Using dummy data generator
+01/07/2025 18:05:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]: [Training Plan] There are 1 training stages 
+01/07/2025 18:05:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]: [Stage Stable Training Stage] start from step 1 
+01/07/2025 18:05:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]: 
+01/07/2025 18:05:45 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]: [Start training] datetime: 2025-01-07 18:05:45.387954 | mbs: 2 | grad_accum: 2 | global_batch_size: 256 | sequence_length: 4096 | train_steps: 100 | start_iteration_step: 0 | consumed_train_samples: 0
+01/07/2025 18:05:47 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]: Resuming training from stage Stable Training Stage, it has trained for 0 samples and has 99 remaining train steps
+01/07/2025 18:05:47 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]:  Memory usage: 7760.42MiB. Peak allocated 7760.42MiB. Peak reserved: 15652.00MiB
+01/07/2025 18:05:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]:  Memory usage: 7832.57MiB. Peak allocated 29417.96MiB. Peak reserved: 35582.00MiB
+01/07/2025 18:05:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]: iteration: 1 / 100 | consumed_tokens: 1.05M | elapsed_time_per_iteration_ms: 2.5K | tokens_per_sec: 419K | tokens_per_sec_per_gpu: 6.55K | global_batch_size: 256 | lm_loss: 12.2 | lr: 0.00015 | model_tflops_per_gpu: 63.3 | hardware_tflops_per_gpu: 63.3 | grad_norm: 1.14 | cuda_memory_allocated: 8.38G | cuda_max_memory_reserved: 37.4G | hd_total_memory_tb: 312G | hd_used_memory_tb: 69G | hd_free_memory_tb: 243G
+01/07/2025 18:05:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]:  Memory usage: 7992.65MiB. Peak allocated 13152.93MiB. Peak reserved: 35630.00MiB
+01/07/2025 18:05:49 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]:  Memory usage: 7992.65MiB. Peak allocated 29578.04MiB. Peak reserved: 35646.00MiB
+01/07/2025 18:05:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]: iteration: 2 / 100 | consumed_tokens: 2.1M | elapsed_time_per_iteration_ms: 679 | tokens_per_sec: 1.54M | tokens_per_sec_per_gpu: 24.1K | global_batch_size: 256 | lm_loss: 12.2 | lr: 0.0003 | model_tflops_per_gpu: 233 | hardware_tflops_per_gpu: 233 | grad_norm: 1.14 | cuda_memory_allocated: 8.38G | cuda_max_memory_reserved: 37.4G | hd_total_memory_tb: 312G | hd_used_memory_tb: 69G | hd_free_memory_tb: 243G
+01/07/2025 18:05:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]:  Memory usage: 7992.65MiB. Peak allocated 13152.94MiB. Peak reserved: 35646.00MiB
+01/07/2025 18:05:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]:  Memory usage: 7992.65MiB. Peak allocated 29578.04MiB. Peak reserved: 35646.00MiB
+01/07/2025 18:05:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]: iteration: 3 / 100 | consumed_tokens: 3.15M | elapsed_time_per_iteration_ms: 537 | tokens_per_sec: 1.95M | tokens_per_sec_per_gpu: 30.5K | global_batch_size: 256 | lm_loss: 12.2 | lr: 0.000296 | model_tflops_per_gpu: 295 | hardware_tflops_per_gpu: 295 | grad_norm: 1.13 | cuda_memory_allocated: 8.38G | cuda_max_memory_reserved: 37.4G | hd_total_memory_tb: 312G | hd_used_memory_tb: 69G | hd_free_memory_tb: 243G
+01/07/2025 18:05:50 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]:  Memory usage: 7992.65MiB. Peak allocated 13152.94MiB. Peak reserved: 35646.00MiB
+01/07/2025 18:05:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]:  Memory usage: 7992.65MiB. Peak allocated 29578.04MiB. Peak reserved: 35646.00MiB
+01/07/2025 18:05:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]: iteration: 4 / 100 | consumed_tokens: 4.19M | elapsed_time_per_iteration_ms: 531 | tokens_per_sec: 1.97M | tokens_per_sec_per_gpu: 30.8K | global_batch_size: 256 | lm_loss: 12.2 | lr: 0.000283 | model_tflops_per_gpu: 298 | hardware_tflops_per_gpu: 298 | grad_norm: 1.06 | cuda_memory_allocated: 8.38G | cuda_max_memory_reserved: 37.4G | hd_total_memory_tb: 312G | hd_used_memory_tb: 69G | hd_free_memory_tb: 243G
+01/07/2025 18:05:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]: | job_id   | name                                                           | nodes | seq_len | mbs | batch_accum | gbs | mTFLOPs | hTFLOPs | tok/s/gpu | AllReduce (GB/s) | AllGather (GB/s) | ReduceScatter (GB/s) | AR Intra-node (GB/s) | AG Intra-node (GB/s) | RS Intra-node (GB/s) | Mem Alloc (GB) | Mem Res (GB) | dp | pp | tp | pp_engine                           | tp_mode        | tp_async_comm | hidden_size | hidden_act | num_layers | num_heads | num_kv_heads | max_pos | vocab_size | tie_word_embeddings | dtype          | zero_stage | ddp_bucket_cap_mb | accumulate_grad_in_fp32 | Total Params | Local Params |
+01/07/2025 18:05:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]: | -------- | -------------------------------------------------------------- | ----- | ------- | --- | ----------- | --- | ------- | ------- | --------- | ---------------- | ---------------- | -------------------- | -------------------- | -------------------- | -------------------- | -------------- | ------------ | -- | -- | -- | ----------------------------------- | -------------- | ------------- | ----------- | ---------- | ---------- | --------- | ------------ | ------- | ---------- | ------------------- | -------------- | ---------- | ----------------- | ----------------------- | ------------ | ------------ |
+01/07/2025 18:05:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]: | 14113491 | 1.34G_dp64_tp1_pp1_acc2_mbs2_seq4096_zero1_tpmodeRED_vocab131k | 8     | 4096    | 2   | 2           | 256 | 298.05  | 298.05  | 30841.90  | 94.44            | 52.55            | 52.48                | 425.07               | 242.00               | 242.38               | 12.84          | 34.81        | 64 | 1  | 1  | OneForwardOneBackwardPipelineEngine | REDUCE_SCATTER | True          | 2048        | silu       | 16         | 32        | 32           | 4096    | 131072     | True                | torch.bfloat16 | 1          | 25                | True                    | 1.34G        | 1.34G        |
+01/07/2025 18:05:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]: Attempting to write benchmark results to CSV file: benchmark/results/bench_final2.csv
+01/07/2025 18:05:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]: Acquired lock for CSV file: benchmark/results/bench_final2.csv
+01/07/2025 18:05:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]: Successfully wrote to CSV file: benchmark/results/bench_final2.csv. Releasing lock...
+01/07/2025 18:05:51 [INFO|DP=0|PP=0|TP=0|ip-26-0-167-217]: Throughput logging complete
+[2025-01-07 18:06:08,700] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 166314 closing signal SIGTERM
+[2025-01-07 18:06:08,700] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 166315 closing signal SIGTERM
+[2025-01-07 18:06:08,700] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 166316 closing signal SIGTERM
+[2025-01-07 18:06:08,700] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 166317 closing signal SIGTERM
+[2025-01-07 18:06:08,700] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 166318 closing signal SIGTERM
+[2025-01-07 18:06:08,700] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 166319 closing signal SIGTERM
+[2025-01-07 18:06:08,700] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 166320 closing signal SIGTERM
+[2025-01-07 18:06:12,997] torch.distributed.elastic.multiprocessing.api: [ERROR] failed (exitcode: 1) local_rank: 0 (pid: 166313) of binary: /fsx/nouamane/miniconda/envs/2-1-cu121/bin/python
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 264, in launch_agent
+    raise ChildFailedError(
+torch.distributed.elastic.multiprocessing.errors.ChildFailedError: 
+============================================================
+/fsx/nouamane/projects/nanotron/run_train.py FAILED
+------------------------------------------------------------
+Failures:
+  <NO_OTHER_FAILURES>
+------------------------------------------------------------
+Root Cause (first observed failure):
+[0]:
+  time      : 2025-01-07_18:06:08
+  host      : ip-26-0-168-95.ec2.internal
+  rank      : 56 (local_rank: 0)
+  exitcode  : 1 (pid: 166313)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+============================================================
+srun: error: ip-26-0-168-95: task 5: Exited with exit code 1
+srun: launch/slurm: _step_signal: Terminating StepId=14113491.0
+slurmstepd: error: *** STEP 14113491.0 ON ip-26-0-167-217 CANCELLED AT 2025-01-07T18:06:13 ***
+[2025-01-07 18:06:13,320] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-07 18:06:13,320] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-07 18:06:13,320] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 2201576 closing signal SIGTERM
+[2025-01-07 18:06:13,320] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-07 18:06:13,320] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 222556 closing signal SIGTERM
+[2025-01-07 18:06:13,320] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-07 18:06:13,320] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 222557 closing signal SIGTERM
+[2025-01-07 18:06:13,321] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 91395 closing signal SIGTERM
+[2025-01-07 18:06:13,321] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-07 18:06:13,321] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 91396 closing signal SIGTERM
+[2025-01-07 18:06:13,320] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 222558 closing signal SIGTERM
+[2025-01-07 18:06:13,321] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 91397 closing signal SIGTERM
+[2025-01-07 18:06:13,321] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 834260 closing signal SIGTERM
+[2025-01-07 18:06:13,321] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 91398 closing signal SIGTERM
+[2025-01-07 18:06:13,322] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 28588 closing signal SIGTERM
+[2025-01-07 18:06:13,320] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 222559 closing signal SIGTERM
+[2025-01-07 18:06:13,321] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 91399 closing signal SIGTERM
+[2025-01-07 18:06:13,321] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 834261 closing signal SIGTERM
+[2025-01-07 18:06:13,322] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 28589 closing signal SIGTERM
+[2025-01-07 18:06:13,320] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 222560 closing signal SIGTERM
+[2025-01-07 18:06:13,321] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 91400 closing signal SIGTERM
+[2025-01-07 18:06:13,321] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 91401 closing signal SIGTERM
+[2025-01-07 18:06:13,322] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 28590 closing signal SIGTERM
+[2025-01-07 18:06:13,321] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 222561 closing signal SIGTERM
+[2025-01-07 18:06:13,321] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 91402 closing signal SIGTERM
+[2025-01-07 18:06:13,321] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 834262 closing signal SIGTERM
+[2025-01-07 18:06:13,322] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 28591 closing signal SIGTERM
+[2025-01-07 18:06:13,321] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 222562 closing signal SIGTERM
+[2025-01-07 18:06:13,321] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 834263 closing signal SIGTERM
+[2025-01-07 18:06:13,322] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 28593 closing signal SIGTERM
+[2025-01-07 18:06:13,321] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-07 18:06:13,321] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 834264 closing signal SIGTERM
+[2025-01-07 18:06:13,321] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 222563 closing signal SIGTERM
+[2025-01-07 18:06:13,321] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 834265 closing signal SIGTERM
+[2025-01-07 18:06:13,321] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 72627 closing signal SIGTERM
+[2025-01-07 18:06:13,321] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 834266 closing signal SIGTERM
+[2025-01-07 18:06:13,321] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 72628 closing signal SIGTERM
+[2025-01-07 18:06:13,321] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 834267 closing signal SIGTERM
+[2025-01-07 18:06:13,321] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 72629 closing signal SIGTERM
+[2025-01-07 18:06:13,321] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 72630 closing signal SIGTERM
+[2025-01-07 18:06:13,321] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 72631 closing signal SIGTERM
+[2025-01-07 18:06:13,321] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 72632 closing signal SIGTERM
+[2025-01-07 18:06:13,321] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 72633 closing signal SIGTERM
+[2025-01-07 18:06:13,321] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 72634 closing signal SIGTERM
+[2025-01-07 18:06:13,321] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-07 18:06:13,321] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 152261 closing signal SIGTERM
+[2025-01-07 18:06:13,321] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 152262 closing signal SIGTERM
+[2025-01-07 18:06:13,321] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 152263 closing signal SIGTERM
+[2025-01-07 18:06:13,321] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 152264 closing signal SIGTERM
+[2025-01-07 18:06:13,322] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 152266 closing signal SIGTERM
+[2025-01-07 18:06:13,322] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 152267 closing signal SIGTERM
+[2025-01-07 18:06:13,322] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 152268 closing signal SIGTERM
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 2201497 got signal: 15
+[2025-01-07 18:06:13,650] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-168-238.ec2.internal_72555_0' has failed to send a keep-alive heartbeat to the rendezvous '14113491' due to an error of type RendezvousConnectionError.
+[2025-01-07 18:06:13,661] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-168-120.ec2.internal_152188_0' has failed to send a keep-alive heartbeat to the rendezvous '14113491' due to an error of type RendezvousConnectionError.
+srun: error: ip-26-0-167-217: task 0: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 28514 got signal: 15
+srun: error: ip-26-0-168-30: task 2: Exited with exit code 1
+[2025-01-07 18:06:17,725] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-168-52.ec2.internal_834187_0' has failed to send a keep-alive heartbeat to the rendezvous '14113491' due to an error of type RendezvousConnectionError.
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 834187 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 152188 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 91323 got signal: 15
+[2025-01-07 18:06:18,565] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-167-245.ec2.internal_222484_0' has failed to send a keep-alive heartbeat to the rendezvous '14113491' due to an error of type RendezvousConnectionError.
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 222484 got signal: 15
+srun: error: ip-26-0-168-52: task 4: Exited with exit code 1
+[2025-01-07 18:06:18,652] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-168-238.ec2.internal_72555_0' has failed to send a keep-alive heartbeat to the rendezvous '14113491' due to an error of type RendezvousConnectionError.
+srun: error: ip-26-0-168-120: task 6: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 72555 got signal: 15
+srun: error: ip-26-0-168-34: task 3: Exited with exit code 1
+srun: error: ip-26-0-167-245: task 1: Exited with exit code 1
+srun: error: ip-26-0-168-238: task 7: Exited with exit code 1
diff --git a/logs/14280977-bench_469G_dp2_tp8_pp8_acc128_mbs1_seq4096_zero0_tpmodeRED_vocab131k_gqa8.out b/logs/14280977-bench_469G_dp2_tp8_pp8_acc128_mbs1_seq4096_zero0_tpmodeRED_vocab131k_gqa8.out
new file mode 100644
index 0000000000000000000000000000000000000000..95aeb2e74fb31ba226c8f8f07577be4da86b4cf8
--- /dev/null
+++ b/logs/14280977-bench_469G_dp2_tp8_pp8_acc128_mbs1_seq4096_zero0_tpmodeRED_vocab131k_gqa8.out
@@ -0,0 +1,4183 @@
++ echo 'Running script: /var/spool/slurmd/job14280977/slurm_script'
+Running script: /var/spool/slurmd/job14280977/slurm_script
++ '[' -z 14280977 ']'
++ source /etc/profile.d/modules.sh
+++ . /usr/share/modules/init/bash
++++ unset _mlshdbg
++++ '[' 0 = 1 ']'
++++ unset _mlre _mlIFS
++++ '[' -n x ']'
++++ _mlIFS=' 	
+'
++++ IFS=' '
++++ '[' -n '' ']'
+++++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash autoinit
++++ _mlcode='module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+MODULEPATH_modshare=/etc/environment-modules/modules:1:/usr/share/modules/\$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1; export MODULEPATH_modshare;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+LOADEDMODULES=; export LOADEDMODULES;
+MODULEPATH=/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/\$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles; export MODULEPATH;
+test 0;'
++++ _mlret=0
++++ '[' -n x ']'
++++ IFS=' 	
+'
++++ unset _mlIFS
++++ unset _mlre _mlv _mlrv
++++ '[' 0 -eq 0 ']'
++++ eval 'module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+MODULEPATH_modshare=/etc/environment-modules/modules:1:/usr/share/modules/\$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1; export MODULEPATH_modshare;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+LOADEDMODULES=; export LOADEDMODULES;
+MODULEPATH=/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/\$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles; export MODULEPATH;
+test 0;'
+++++ MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl
+++++ export MODULES_CMD
+++++ ENV=/usr/share/modules/init/profile.sh
+++++ export ENV
+++++ MODULEPATH_modshare='/etc/environment-modules/modules:1:/usr/share/modules/$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1'
+++++ export MODULEPATH_modshare
+++++ BASH_ENV=/usr/share/modules/init/bash
+++++ export BASH_ENV
+++++ MODULESHOME=/usr/share/modules
+++++ export MODULESHOME
+++++ LOADEDMODULES=
+++++ export LOADEDMODULES
+++++ MODULEPATH='/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles'
+++++ export MODULEPATH
+++++ test 0
++++ '[' 0 = 1 ']'
++++ '[' -t 2 ']'
++++ export -f module
++++ export -f switchml
++++ '[' 5 -ge 3 ']'
++++ [[ ehxB =~ i ]]
++++ [[ ! :/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin: =~ :/usr/bin: ]]
++++ '[' '!' -n '' ']'
++++ MANPATH=:
++++ export MANPATH
+++++ manpath
++++ [[ ! :/fsx/nouamane/miniconda/envs/2-1-cu121/man:/fsx/nouamane/miniconda/envs/2-1-cu121/share/man:/fsx/nouamane/miniconda/man:/fsx/nouamane/miniconda/share/man:/usr/local/man:/usr/local/share/man:/usr/share/man:/opt/slurm/share/man:/admin/home/nouamane/.local/share/man:/admin/home/nouamane/.fzf/man:: =~ :/usr/share/man: ]]
++++ unset _mlcode _mlret
++++ '[' -n '' ']'
++ module load cuda/12.1
++ unset _mlshdbg
++ '[' 0 = 1 ']'
++ unset _mlre _mlIFS
++ '[' -n x ']'
++ _mlIFS=' 	
+'
++ IFS=' '
++ '[' -n '' ']'
+++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash load cuda/12.1
++ eval 'CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include;' export 'CPATH;
+LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:;' export 'LD_LIBRARY_PATH;
+FI_EFA_FORK_SAFE=1;' export 'FI_EFA_FORK_SAFE;
+MANPATH=/usr/local/cuda-12.1/share/man::;' export 'MANPATH;
+LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64;' export 'LIBRARY_PATH;
+_LMFILES_=/usr/share/modules/modulefiles/cuda/12.1;' export '_LMFILES_;
+LOADEDMODULES=cuda/12.1;' export 'LOADEDMODULES;
+MPI_PATH=/opt/amazon/openmpi;' export 'MPI_PATH;
+NCCL_HOME_modshare=/opt/nccl/build:1;' export 'NCCL_HOME_modshare;
+NCCL_PROTO=simple;' export 'NCCL_PROTO;
+MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1;' export 'MANPATH_modshare;
+LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1;' export 'LIBRARY_PATH_modshare;
+NCCL_SOCKET_IFNAME=enp;' export 'NCCL_SOCKET_IFNAME;
+AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl;' export 'AWS_OFI_NCCL_HOME;
+NCCL_HOME=/opt/nccl/build;' export 'NCCL_HOME;
+FI_PROVIDER=efa;' export 'FI_PROVIDER;
+AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1;' export 'AWS_OFI_NCCL_HOME_modshare;
+CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1;' export 'CPATH_modshare;
+LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2::1:/opt/nccl/build/lib:1:/opt/aws-ofi-nccl/lib:2:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/efa/lib:1;' export 'LD_LIBRARY_PATH_modshare;
+FI_EFA_ENABLE_SHM_TRANSFER=1;' export 'FI_EFA_ENABLE_SHM_TRANSFER;
+_LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1;' export '_LMFILES__modshare;
+LOADEDMODULES_modshare=cuda/12.1:1;' export 'LOADEDMODULES_modshare;
+MPI_PATH_modshare=/opt/amazon/openmpi:1;' export 'MPI_PATH_modshare;
+PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin;' export 'PATH;
+CUDA_HOME=/usr/local/cuda-12.1;' export 'CUDA_HOME;
+PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/opt/slurm/bin:1:/usr/sbin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1;' export 'PATH_modshare;
+test' '0;'
+++ CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include
+++ export CPATH
+++ LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:
+++ export LD_LIBRARY_PATH
+++ FI_EFA_FORK_SAFE=1
+++ export FI_EFA_FORK_SAFE
+++ MANPATH=/usr/local/cuda-12.1/share/man::
+++ export MANPATH
+++ LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64
+++ export LIBRARY_PATH
+++ _LMFILES_=/usr/share/modules/modulefiles/cuda/12.1
+++ export _LMFILES_
+++ LOADEDMODULES=cuda/12.1
+++ export LOADEDMODULES
+++ MPI_PATH=/opt/amazon/openmpi
+++ export MPI_PATH
+++ NCCL_HOME_modshare=/opt/nccl/build:1
+++ export NCCL_HOME_modshare
+++ NCCL_PROTO=simple
+++ export NCCL_PROTO
+++ MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1
+++ export MANPATH_modshare
+++ LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1
+++ export LIBRARY_PATH_modshare
+++ NCCL_SOCKET_IFNAME=enp
+++ export NCCL_SOCKET_IFNAME
+++ AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl
+++ export AWS_OFI_NCCL_HOME
+++ NCCL_HOME=/opt/nccl/build
+++ export NCCL_HOME
+++ FI_PROVIDER=efa
+++ export FI_PROVIDER
+++ AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1
+++ export AWS_OFI_NCCL_HOME_modshare
+++ CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1
+++ export CPATH_modshare
+++ LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2::1:/opt/nccl/build/lib:1:/opt/aws-ofi-nccl/lib:2:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/efa/lib:1
+++ export LD_LIBRARY_PATH_modshare
+++ FI_EFA_ENABLE_SHM_TRANSFER=1
+++ export FI_EFA_ENABLE_SHM_TRANSFER
+++ _LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1
+++ export _LMFILES__modshare
+++ LOADEDMODULES_modshare=cuda/12.1:1
+++ export LOADEDMODULES_modshare
+++ MPI_PATH_modshare=/opt/amazon/openmpi:1
+++ export MPI_PATH_modshare
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
+++ export PATH
+++ CUDA_HOME=/usr/local/cuda-12.1
+++ export CUDA_HOME
+++ PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/sbin:1:/opt/slurm/bin:1:/usr/sbin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1
+++ export PATH_modshare
+++ test 0
++ _mlstatus=0
++ '[' -n x ']'
++ IFS=' 	
+'
++ unset _mlre _mlv _mlrv _mlIFS
++ '[' -n '' ']'
++ unset _mlshdbg
++ return 0
++ source /fsx/nouamane/miniconda/bin/activate
+++ _CONDA_ROOT=/fsx/nouamane/miniconda
+++ . /fsx/nouamane/miniconda/etc/profile.d/conda.sh
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ '[' -z x ']'
+++ conda activate
+++ local cmd=activate
+++ case "$cmd" in
+++ __conda_activate activate
+++ '[' -n '' ']'
+++ local ask_conda
++++ PS1=
++++ __conda_exe shell.posix activate
++++ /fsx/nouamane/miniconda/bin/conda shell.posix activate
+Error while loading conda entry point: conda-libmamba-solver (module 'libmambapy' has no attribute 'QueryFormat')
+++ ask_conda='. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''2'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_1='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/etc/conda/activate.d/libxml2_activate.sh"'
+++ eval '. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''2'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_1='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/etc/conda/activate.d/libxml2_activate.sh"'
++++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh
+++++ test -n ''
+++++ unset XML_CATALOG_FILES
+++++ unset xml_catalog_files_libxml2
++++ PS1='(base) '
++++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
++++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
++++ export CONDA_PREFIX=/fsx/nouamane/miniconda
++++ CONDA_PREFIX=/fsx/nouamane/miniconda
++++ export CONDA_SHLVL=2
++++ CONDA_SHLVL=2
++++ export CONDA_DEFAULT_ENV=base
++++ CONDA_DEFAULT_ENV=base
++++ export 'CONDA_PROMPT_MODIFIER=(base) '
++++ CONDA_PROMPT_MODIFIER='(base) '
++++ export CONDA_PREFIX_1=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ CONDA_PREFIX_1=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ . /fsx/nouamane/miniconda/etc/conda/activate.d/libxml2_activate.sh
+++++ test -n ''
+++++ xml_catalog_files_libxml2=
+++++ XML_CATALOG_FILES=
+++++ conda_catalog_files=
+++++ ifs_libxml2=' 	
+'
+++++ IFS=' '
+++++ rem=/fsx/nouamane/miniconda
+++++ for pre in ${rem}
+++++ test '' = /fsx/nouamane/miniconda
+++++ conda_catalog_files=/fsx/nouamane/miniconda
+++++ rem=
+++++ IFS=' 	
+'
+++++ conda_catalog_files='file:///fsx/nouamane/miniconda/etc/xml/catalog file:///etc/xml/catalog'
+++++ export 'XML_CATALOG_FILES=file:///fsx/nouamane/miniconda/etc/xml/catalog file:///etc/xml/catalog'
+++++ XML_CATALOG_FILES='file:///fsx/nouamane/miniconda/etc/xml/catalog file:///etc/xml/catalog'
+++++ unset conda_catalog_files ifs_libxml2 rem
+++ __conda_hashr
+++ '[' -n '' ']'
+++ '[' -n '' ']'
+++ hash -r
++ conda activate 2-1-cu121
++ local cmd=activate
++ case "$cmd" in
++ __conda_activate activate 2-1-cu121
++ '[' -n '' ']'
++ local ask_conda
+++ PS1='(base) '
+++ __conda_exe shell.posix activate 2-1-cu121
+++ /fsx/nouamane/miniconda/bin/conda shell.posix activate 2-1-cu121
+Error while loading conda entry point: conda-libmamba-solver (module 'libmambapy' has no attribute 'QueryFormat')
++ ask_conda='. "/fsx/nouamane/miniconda/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_2='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
++ eval '. "/fsx/nouamane/miniconda/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_2='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
+++ . /fsx/nouamane/miniconda/etc/conda/deactivate.d/libxml2_deactivate.sh
++++ test -n ''
++++ unset XML_CATALOG_FILES
++++ unset xml_catalog_files_libxml2
+++ PS1='(2-1-cu121) '
+++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
+++ export CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ export CONDA_SHLVL=3
+++ CONDA_SHLVL=3
+++ export CONDA_DEFAULT_ENV=2-1-cu121
+++ CONDA_DEFAULT_ENV=2-1-cu121
+++ export 'CONDA_PROMPT_MODIFIER=(2-1-cu121) '
+++ CONDA_PROMPT_MODIFIER='(2-1-cu121) '
+++ export CONDA_PREFIX_2=/fsx/nouamane/miniconda
+++ CONDA_PREFIX_2=/fsx/nouamane/miniconda
+++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ export _CE_M=
+++ _CE_M=
+++ export _CE_CONDA=
+++ _CE_CONDA=
+++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh
++++ test -n ''
++++ xml_catalog_files_libxml2=
++++ XML_CATALOG_FILES=
++++ conda_catalog_files=
++++ ifs_libxml2=' 	
+'
++++ IFS=' '
++++ rem=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ for pre in ${rem}
++++ test '' = /fsx/nouamane/miniconda/envs/2-1-cu121
++++ conda_catalog_files=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ rem=
++++ IFS=' 	
+'
++++ conda_catalog_files='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ export 'XML_CATALOG_FILES=file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ XML_CATALOG_FILES='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ unset conda_catalog_files ifs_libxml2 rem
++ __conda_hashr
++ '[' -n '' ']'
++ '[' -n '' ']'
++ hash -r
++ export PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
++ PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/316e524257c2ea23b755332b0a72c50cf23e1b00/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
++ '[' -z '' ']'
+++ scontrol show hostnames 'ip-26-0-162-[46,79,180,233],ip-26-0-165-[24,38,59,131],ip-26-0-167-[217,245],ip-26-0-168-[30,34,52,95,120,238]'
++ export 'NODELIST=ip-26-0-162-46
+ip-26-0-162-79
+ip-26-0-162-180
+ip-26-0-162-233
+ip-26-0-165-24
+ip-26-0-165-38
+ip-26-0-165-59
+ip-26-0-165-131
+ip-26-0-167-217
+ip-26-0-167-245
+ip-26-0-168-30
+ip-26-0-168-34
+ip-26-0-168-52
+ip-26-0-168-95
+ip-26-0-168-120
+ip-26-0-168-238'
++ NODELIST='ip-26-0-162-46
+ip-26-0-162-79
+ip-26-0-162-180
+ip-26-0-162-233
+ip-26-0-165-24
+ip-26-0-165-38
+ip-26-0-165-59
+ip-26-0-165-131
+ip-26-0-167-217
+ip-26-0-167-245
+ip-26-0-168-30
+ip-26-0-168-34
+ip-26-0-168-52
+ip-26-0-168-95
+ip-26-0-168-120
+ip-26-0-168-238'
+++ head -n1
+++ scontrol show hostnames 'ip-26-0-162-[46,79,180,233],ip-26-0-165-[24,38,59,131],ip-26-0-167-[217,245],ip-26-0-168-[30,34,52,95,120,238]'
++ export MASTER_NODE=ip-26-0-162-46
++ MASTER_NODE=ip-26-0-162-46
++ export MASTER_PORT=12356
++ MASTER_PORT=12356
++ export NNODES=16
++ NNODES=16
++ export GPUS_PER_NODE=8
++ GPUS_PER_NODE=8
++ export WORLD_SIZE=128
++ WORLD_SIZE=128
++ export CUDA_DEVICE_MAX_CONNECTIONS=1
++ CUDA_DEVICE_MAX_CONNECTIONS=1
++ export NCCL_DEBUG=WARN
++ NCCL_DEBUG=WARN
++ export NANOTRON_BENCHMARK=1
++ NANOTRON_BENCHMARK=1
++ export WANDB_MODE=disabled
++ WANDB_MODE=disabled
++ export TORCH_NCCL_ASYNC_ERROR_HANDLING=1
++ TORCH_NCCL_ASYNC_ERROR_HANDLING=1
++ export TORCH_DISTRIBUTED_DEBUG=DETAIL
++ TORCH_DISTRIBUTED_DEBUG=DETAIL
++ '[' -z '' ']'
++ echo '=== GPU Topology ==='
+=== GPU Topology ===
++ nvidia-smi topo -m
+	GPU0	GPU1	GPU2	GPU3	GPU4	GPU5	GPU6	GPU7	CPU Affinity	NUMA Affinity	GPU NUMA ID
+GPU0	 X 	NV18	NV18	NV18	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU1	NV18	 X 	NV18	NV18	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU2	NV18	NV18	 X 	NV18	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU3	NV18	NV18	NV18	 X 	NV18	NV18	NV18	NV18	0-47	0		N/A
+GPU4	NV18	NV18	NV18	NV18	 X 	NV18	NV18	NV18	48-95	1		N/A
+GPU5	NV18	NV18	NV18	NV18	NV18	 X 	NV18	NV18	48-95	1		N/A
+GPU6	NV18	NV18	NV18	NV18	NV18	NV18	 X 	NV18	48-95	1		N/A
+GPU7	NV18	NV18	NV18	NV18	NV18	NV18	NV18	 X 	48-95	1		N/A
+
+Legend:
+
+  X    = Self
+  SYS  = Connection traversing PCIe as well as the SMP interconnect between NUMA nodes (e.g., QPI/UPI)
+  NODE = Connection traversing PCIe as well as the interconnect between PCIe Host Bridges within a NUMA node
+  PHB  = Connection traversing PCIe as well as a PCIe Host Bridge (typically the CPU)
+  PXB  = Connection traversing multiple PCIe bridges (without traversing the PCIe Host Bridge)
+  PIX  = Connection traversing at most a single PCIe bridge
+  NV#  = Connection traversing a bonded set of # NVLinks
++ echo ==================
+==================
++ export SRUN_ALLOC_ARGS=
++ SRUN_ALLOC_ARGS=
++ echo 'Master node: ip-26-0-162-46'
+Master node: ip-26-0-162-46
++ echo 'All nodes: ip-26-0-162-46
+ip-26-0-162-79
+ip-26-0-162-180
+ip-26-0-162-233
+ip-26-0-165-24
+ip-26-0-165-38
+ip-26-0-165-59
+ip-26-0-165-131
+ip-26-0-167-217
+ip-26-0-167-245
+ip-26-0-168-30
+ip-26-0-168-34
+ip-26-0-168-52
+ip-26-0-168-95
+ip-26-0-168-120
+ip-26-0-168-238'
+All nodes: ip-26-0-162-46
+ip-26-0-162-79
+ip-26-0-162-180
+ip-26-0-162-233
+ip-26-0-165-24
+ip-26-0-165-38
+ip-26-0-165-59
+ip-26-0-165-131
+ip-26-0-167-217
+ip-26-0-167-245
+ip-26-0-168-30
+ip-26-0-168-34
+ip-26-0-168-52
+ip-26-0-168-95
+ip-26-0-168-120
+ip-26-0-168-238
++ echo 'World size: 128'
+World size: 128
++ '[' -n '' ']'
++ srun --wait=0 --kill-on-bad-exit=1 torchrun --nnodes=16 --nproc_per_node=8 --rdzv_id=14280977 --rdzv_backend=c10d --rdzv_endpoint=ip-26-0-162-46:12356 --max_restarts 0 --rdzv_conf timeout=60 /fsx/nouamane/projects/nanotron/run_train.py --config-file benchmark/configs/config_469G_dp2_tp8_pp8_acc128_mbs1_seq4096_zero0_tpmodeRED_vocab131k_gqa8.yaml
+[2025-01-08 11:13:35,272] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-08 11:13:35,272] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-08 11:13:35,272] torch.distributed.run: [WARNING] 
+[2025-01-08 11:13:35,272] torch.distributed.run: [WARNING] *****************************************
+[2025-01-08 11:13:35,272] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-08 11:13:35,272] torch.distributed.run: [WARNING] *****************************************
+[2025-01-08 11:13:35,272] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-08 11:13:35,272] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-08 11:13:35,274] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-08 11:13:35,274] torch.distributed.run: [WARNING] 
+[2025-01-08 11:13:35,274] torch.distributed.run: [WARNING] *****************************************
+[2025-01-08 11:13:35,274] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-08 11:13:35,274] torch.distributed.run: [WARNING] *****************************************
+[2025-01-08 11:13:35,274] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-08 11:13:35,274] torch.distributed.run: [WARNING] 
+[2025-01-08 11:13:35,274] torch.distributed.run: [WARNING] *****************************************
+[2025-01-08 11:13:35,274] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-08 11:13:35,274] torch.distributed.run: [WARNING] *****************************************
+[2025-01-08 11:13:35,277] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-08 11:13:35,277] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-08 11:13:35,277] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-08 11:13:35,278] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-08 11:13:35,282] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-08 11:13:35,283] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-08 11:13:35,283] torch.distributed.run: [WARNING] 
+[2025-01-08 11:13:35,283] torch.distributed.run: [WARNING] *****************************************
+[2025-01-08 11:13:35,283] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-08 11:13:35,283] torch.distributed.run: [WARNING] *****************************************
+[2025-01-08 11:13:35,284] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-08 11:13:35,287] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-08 11:13:35,310] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-08 11:13:35,272] torch.distributed.run: [WARNING] 
+[2025-01-08 11:13:35,272] torch.distributed.run: [WARNING] *****************************************
+[2025-01-08 11:13:35,272] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-08 11:13:35,272] torch.distributed.run: [WARNING] *****************************************
+[2025-01-08 11:13:35,272] torch.distributed.run: [WARNING] 
+[2025-01-08 11:13:35,272] torch.distributed.run: [WARNING] *****************************************
+[2025-01-08 11:13:35,272] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-08 11:13:35,272] torch.distributed.run: [WARNING] *****************************************
+[2025-01-08 11:13:35,277] torch.distributed.run: [WARNING] 
+[2025-01-08 11:13:35,277] torch.distributed.run: [WARNING] *****************************************
+[2025-01-08 11:13:35,277] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-08 11:13:35,277] torch.distributed.run: [WARNING] *****************************************
+[2025-01-08 11:13:35,277] torch.distributed.run: [WARNING] 
+[2025-01-08 11:13:35,277] torch.distributed.run: [WARNING] *****************************************
+[2025-01-08 11:13:35,277] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-08 11:13:35,277] torch.distributed.run: [WARNING] *****************************************
+[2025-01-08 11:13:35,277] torch.distributed.run: [WARNING] 
+[2025-01-08 11:13:35,277] torch.distributed.run: [WARNING] *****************************************
+[2025-01-08 11:13:35,277] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-08 11:13:35,277] torch.distributed.run: [WARNING] *****************************************
+[2025-01-08 11:13:35,272] torch.distributed.run: [WARNING] 
+[2025-01-08 11:13:35,272] torch.distributed.run: [WARNING] *****************************************
+[2025-01-08 11:13:35,272] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-08 11:13:35,272] torch.distributed.run: [WARNING] *****************************************
+[2025-01-08 11:13:35,278] torch.distributed.run: [WARNING] 
+[2025-01-08 11:13:35,278] torch.distributed.run: [WARNING] *****************************************
+[2025-01-08 11:13:35,278] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-08 11:13:35,278] torch.distributed.run: [WARNING] *****************************************
+[2025-01-08 11:13:35,284] torch.distributed.run: [WARNING] 
+[2025-01-08 11:13:35,284] torch.distributed.run: [WARNING] *****************************************
+[2025-01-08 11:13:35,284] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-08 11:13:35,284] torch.distributed.run: [WARNING] *****************************************
+[2025-01-08 11:13:35,282] torch.distributed.run: [WARNING] 
+[2025-01-08 11:13:35,282] torch.distributed.run: [WARNING] *****************************************
+[2025-01-08 11:13:35,282] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-08 11:13:35,282] torch.distributed.run: [WARNING] *****************************************
+[2025-01-08 11:13:35,287] torch.distributed.run: [WARNING] 
+[2025-01-08 11:13:35,287] torch.distributed.run: [WARNING] *****************************************
+[2025-01-08 11:13:35,287] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-08 11:13:35,287] torch.distributed.run: [WARNING] *****************************************
+[2025-01-08 11:13:35,310] torch.distributed.run: [WARNING] 
+[2025-01-08 11:13:35,310] torch.distributed.run: [WARNING] *****************************************
+[2025-01-08 11:13:35,310] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-08 11:13:35,310] torch.distributed.run: [WARNING] *****************************************
+[2025-01-08 11:13:35,424] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-01-08 11:13:35,425] torch.distributed.run: [WARNING] 
+[2025-01-08 11:13:35,425] torch.distributed.run: [WARNING] *****************************************
+[2025-01-08 11:13:35,425] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-01-08 11:13:35,425] torch.distributed.run: [WARNING] *****************************************
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+01/08/2025 11:14:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: Config:
+01/08/2025 11:14:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: Config(general=GeneralArgs(project='debug',
+01/08/2025 11:14:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                            run='469G_dp2_tp8_pp8_acc128_mbs1_seq4096_zero0_tpmodeRED_vocab131k_gqa8',
+01/08/2025 11:14:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                            seed=42,
+01/08/2025 11:14:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                            step=None,
+01/08/2025 11:14:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                            consumed_train_samples=None,
+01/08/2025 11:14:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                            benchmark_csv_path=PosixPath('benchmark/results/bench_final2.csv'),
+01/08/2025 11:14:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                            ignore_sanity_checks=True),
+01/08/2025 11:14:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:        parallelism=ParallelismArgs(dp=2,
+01/08/2025 11:14:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                                    pp=8,
+01/08/2025 11:14:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                                    tp=8,
+01/08/2025 11:14:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                                    pp_engine=<nanotron.parallel.pipeline_parallel.engine.OneForwardOneBackwardPipelineEngine object at 0x7f6dab47cbb0>,
+01/08/2025 11:14:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                                    tp_mode=<TensorParallelLinearMode.REDUCE_SCATTER: 2>,
+01/08/2025 11:14:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                                    tp_linear_async_communication=True,
+01/08/2025 11:14:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                                    recompute_layer=False,
+01/08/2025 11:14:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                                    tp_recompute_allgather=True,
+01/08/2025 11:14:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                                    expert_parallel_size=1),
+01/08/2025 11:14:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:        model=ModelArgs(model_config=LlamaConfig(bos_token_id=0,
+01/08/2025 11:14:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                                                 eos_token_id=0,
+01/08/2025 11:14:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                                                 hidden_act='silu',
+01/08/2025 11:14:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                                                 hidden_size=16384,
+01/08/2025 11:14:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                                                 initializer_range=0.02,
+01/08/2025 11:14:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                                                 intermediate_size=53248,
+01/08/2025 11:14:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                                                 is_llama_config=True,
+01/08/2025 11:14:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                                                 max_position_embeddings=4096,
+01/08/2025 11:14:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                                                 num_attention_heads=128,
+01/08/2025 11:14:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                                                 num_hidden_layers=126,
+01/08/2025 11:14:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                                                 num_key_value_heads=8,
+01/08/2025 11:14:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                                                 pad_token_id=None,
+01/08/2025 11:14:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                                                 pretraining_tp=1,
+01/08/2025 11:14:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                                                 rms_norm_eps=1e-05,
+01/08/2025 11:14:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                                                 rope_scaling=None,
+01/08/2025 11:14:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                                                 rope_theta=10000.0,
+01/08/2025 11:14:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                                                 rope_interleaved=False,
+01/08/2025 11:14:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                                                 tie_word_embeddings=False,
+01/08/2025 11:14:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                                                 use_cache=True,
+01/08/2025 11:14:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                                                 vocab_size=131072),
+01/08/2025 11:14:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                        init_method=RandomInit(std=0.02),
+01/08/2025 11:14:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                        dtype=torch.bfloat16,
+01/08/2025 11:14:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                        make_vocab_size_divisible_by=1,
+01/08/2025 11:14:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                        ddp_bucket_cap_mb=25),
+01/08/2025 11:14:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:        tokenizer=TokenizerArgs(tokenizer_name_or_path='robot-test/dummy-tokenizer-wordlevel',
+01/08/2025 11:14:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                                tokenizer_revision=None,
+01/08/2025 11:14:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                                tokenizer_max_length=None),
+01/08/2025 11:14:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:        checkpoints=CheckpointsArgs(checkpoints_path=PosixPath('checkpoints'),
+01/08/2025 11:14:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                                    checkpoint_interval=10000,
+01/08/2025 11:14:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                                    save_initial_state=False,
+01/08/2025 11:14:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                                    save_final_state=False,
+01/08/2025 11:14:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                                    resume_checkpoint_path=None,
+01/08/2025 11:14:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                                    checkpoints_path_is_shared_file_system=False),
+01/08/2025 11:14:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:        logging=LoggingArgs(log_level='info',
+01/08/2025 11:14:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                            log_level_replica='info',
+01/08/2025 11:14:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                            iteration_step_info_interval=1),
+01/08/2025 11:14:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:        tokens=TokensArgs(sequence_length=4096,
+01/08/2025 11:14:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                          train_steps=100,
+01/08/2025 11:14:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                          micro_batch_size=1,
+01/08/2025 11:14:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                          batch_accumulation_per_replica=128,
+01/08/2025 11:14:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                          val_check_interval=100,
+01/08/2025 11:14:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                          limit_val_batches=0,
+01/08/2025 11:14:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                          limit_test_batches=0),
+01/08/2025 11:14:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:        optimizer=OptimizerArgs(optimizer_factory=AdamWOptimizerArgs(adam_eps=1e-08,
+01/08/2025 11:14:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                                                                     adam_beta1=0.9,
+01/08/2025 11:14:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                                                                     adam_beta2=0.95,
+01/08/2025 11:14:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                                                                     torch_adam_is_fused=True,
+01/08/2025 11:14:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                                                                     name='adamW'),
+01/08/2025 11:14:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                                zero_stage=0,
+01/08/2025 11:14:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                                weight_decay=0.01,
+01/08/2025 11:14:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                                clip_grad=1.0,
+01/08/2025 11:14:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                                accumulate_grad_in_fp32=True,
+01/08/2025 11:14:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                                learning_rate_scheduler=LRSchedulerArgs(learning_rate=0.0003,
+01/08/2025 11:14:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                                                                        lr_warmup_steps=2,
+01/08/2025 11:14:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                                                                        lr_warmup_style='linear',
+01/08/2025 11:14:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                                                                        lr_decay_style='cosine',
+01/08/2025 11:14:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                                                                        lr_decay_steps=13,
+01/08/2025 11:14:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                                                                        lr_decay_starting_step=None,
+01/08/2025 11:14:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                                                                        min_decay_lr=1e-05)),
+01/08/2025 11:14:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:        data_stages=[DatasetStageArgs(name='Stable Training Stage',
+01/08/2025 11:14:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                                      start_training_step=1,
+01/08/2025 11:14:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                                      data=DataArgs(dataset=None,
+01/08/2025 11:14:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                                                    seed=42,
+01/08/2025 11:14:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:                                                    num_loading_workers=1))],
+01/08/2025 11:14:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:        profiler=None,
+01/08/2025 11:14:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:        lighteval=None,
+01/08/2025 11:14:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:        s3_upload=None)
+01/08/2025 11:14:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: Model Config:
+01/08/2025 11:14:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: LlamaConfig(bos_token_id=0,
+01/08/2025 11:14:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:             eos_token_id=0,
+01/08/2025 11:14:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:             hidden_act='silu',
+01/08/2025 11:14:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:             hidden_size=16384,
+01/08/2025 11:14:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:             initializer_range=0.02,
+01/08/2025 11:14:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:             intermediate_size=53248,
+01/08/2025 11:14:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:             is_llama_config=True,
+01/08/2025 11:14:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:             max_position_embeddings=4096,
+01/08/2025 11:14:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:             num_attention_heads=128,
+01/08/2025 11:14:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:             num_hidden_layers=126,
+01/08/2025 11:14:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:             num_key_value_heads=8,
+01/08/2025 11:14:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:             pad_token_id=None,
+01/08/2025 11:14:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:             pretraining_tp=1,
+01/08/2025 11:14:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:             rms_norm_eps=1e-05,
+01/08/2025 11:14:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:             rope_scaling=None,
+01/08/2025 11:14:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:             rope_theta=10000.0,
+01/08/2025 11:14:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:             rope_interleaved=False,
+01/08/2025 11:14:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:             tie_word_embeddings=False,
+01/08/2025 11:14:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:             use_cache=True,
+01/08/2025 11:14:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]:             vocab_size=131072)
+01/08/2025 11:14:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: Building model..
+01/08/2025 11:14:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: Initialize RoPE Theta = 10000.0
+01/08/2025 11:14:24 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: Setting PP block ranks...
+01/08/2025 11:14:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: module_name: model.token_position_embeddings | PP: 0/8 | Block rank: 0
+01/08/2025 11:14:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: module_name: model.decoder.0 | PP: 0/8 | Block rank: 1
+01/08/2025 11:14:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: module_name: model.decoder.1 | PP: 0/8 | Block rank: 2
+01/08/2025 11:14:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: module_name: model.decoder.2 | PP: 0/8 | Block rank: 3
+01/08/2025 11:14:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: module_name: model.decoder.3 | PP: 0/8 | Block rank: 4
+01/08/2025 11:14:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: module_name: model.decoder.4 | PP: 0/8 | Block rank: 5
+01/08/2025 11:14:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: module_name: model.decoder.5 | PP: 0/8 | Block rank: 6
+01/08/2025 11:14:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: module_name: model.decoder.6 | PP: 0/8 | Block rank: 7
+01/08/2025 11:14:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: module_name: model.decoder.7 | PP: 0/8 | Block rank: 8
+01/08/2025 11:14:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: module_name: model.decoder.8 | PP: 0/8 | Block rank: 9
+01/08/2025 11:14:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: module_name: model.decoder.9 | PP: 0/8 | Block rank: 10
+01/08/2025 11:14:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: module_name: model.decoder.10 | PP: 0/8 | Block rank: 11
+01/08/2025 11:14:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: module_name: model.decoder.11 | PP: 0/8 | Block rank: 12
+01/08/2025 11:14:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: module_name: model.decoder.12 | PP: 0/8 | Block rank: 13
+01/08/2025 11:14:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: module_name: model.decoder.13 | PP: 0/8 | Block rank: 14
+01/08/2025 11:14:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: module_name: model.decoder.14 | PP: 0/8 | Block rank: 15
+01/08/2025 11:14:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: module_name: model.decoder.15 | PP: 0/8 | Block rank: 16
+01/08/2025 11:14:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: module_name: model.decoder.16 | PP: 1/8 | Block rank: 0
+01/08/2025 11:14:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: module_name: model.decoder.17 | PP: 1/8 | Block rank: 1
+01/08/2025 11:14:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: module_name: model.decoder.18 | PP: 1/8 | Block rank: 2
+01/08/2025 11:14:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: module_name: model.decoder.19 | PP: 1/8 | Block rank: 3
+01/08/2025 11:14:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: module_name: model.decoder.20 | PP: 1/8 | Block rank: 4
+01/08/2025 11:14:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: module_name: model.decoder.21 | PP: 1/8 | Block rank: 5
+01/08/2025 11:14:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: module_name: model.decoder.22 | PP: 1/8 | Block rank: 6
+01/08/2025 11:14:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: module_name: model.decoder.23 | PP: 1/8 | Block rank: 7
+01/08/2025 11:14:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: module_name: model.decoder.24 | PP: 1/8 | Block rank: 8
+01/08/2025 11:14:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: module_name: model.decoder.25 | PP: 1/8 | Block rank: 9
+01/08/2025 11:14:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: module_name: model.decoder.26 | PP: 1/8 | Block rank: 10
+01/08/2025 11:14:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: module_name: model.decoder.27 | PP: 1/8 | Block rank: 11
+01/08/2025 11:14:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: module_name: model.decoder.28 | PP: 1/8 | Block rank: 12
+01/08/2025 11:14:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: module_name: model.decoder.29 | PP: 1/8 | Block rank: 13
+01/08/2025 11:14:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: module_name: model.decoder.30 | PP: 1/8 | Block rank: 14
+01/08/2025 11:14:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: module_name: model.decoder.31 | PP: 1/8 | Block rank: 15
+01/08/2025 11:14:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: module_name: model.decoder.32 | PP: 2/8 | Block rank: 0
+01/08/2025 11:14:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: module_name: model.decoder.33 | PP: 2/8 | Block rank: 1
+01/08/2025 11:14:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: module_name: model.decoder.34 | PP: 2/8 | Block rank: 2
+01/08/2025 11:14:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: module_name: model.decoder.35 | PP: 2/8 | Block rank: 3
+01/08/2025 11:14:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: module_name: model.decoder.36 | PP: 2/8 | Block rank: 4
+01/08/2025 11:14:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: module_name: model.decoder.37 | PP: 2/8 | Block rank: 5
+01/08/2025 11:14:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: module_name: model.decoder.38 | PP: 2/8 | Block rank: 6
+01/08/2025 11:14:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: module_name: model.decoder.39 | PP: 2/8 | Block rank: 7
+01/08/2025 11:14:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: module_name: model.decoder.40 | PP: 2/8 | Block rank: 8
+01/08/2025 11:14:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: module_name: model.decoder.41 | PP: 2/8 | Block rank: 9
+01/08/2025 11:14:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: module_name: model.decoder.42 | PP: 2/8 | Block rank: 10
+01/08/2025 11:14:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: module_name: model.decoder.43 | PP: 2/8 | Block rank: 11
+01/08/2025 11:14:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: module_name: model.decoder.44 | PP: 2/8 | Block rank: 12
+01/08/2025 11:14:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: module_name: model.decoder.45 | PP: 2/8 | Block rank: 13
+01/08/2025 11:14:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: module_name: model.decoder.46 | PP: 2/8 | Block rank: 14
+01/08/2025 11:14:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: module_name: model.decoder.47 | PP: 2/8 | Block rank: 15
+01/08/2025 11:14:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: module_name: model.decoder.48 | PP: 3/8 | Block rank: 0
+01/08/2025 11:14:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: module_name: model.decoder.49 | PP: 3/8 | Block rank: 1
+01/08/2025 11:14:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: module_name: model.decoder.50 | PP: 3/8 | Block rank: 2
+01/08/2025 11:14:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: module_name: model.decoder.51 | PP: 3/8 | Block rank: 3
+01/08/2025 11:14:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: module_name: model.decoder.52 | PP: 3/8 | Block rank: 4
+01/08/2025 11:14:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: module_name: model.decoder.53 | PP: 3/8 | Block rank: 5
+01/08/2025 11:14:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: module_name: model.decoder.54 | PP: 3/8 | Block rank: 6
+01/08/2025 11:14:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: module_name: model.decoder.55 | PP: 3/8 | Block rank: 7
+01/08/2025 11:14:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: module_name: model.decoder.56 | PP: 3/8 | Block rank: 8
+01/08/2025 11:14:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: module_name: model.decoder.57 | PP: 3/8 | Block rank: 9
+01/08/2025 11:14:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: module_name: model.decoder.58 | PP: 3/8 | Block rank: 10
+01/08/2025 11:14:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: module_name: model.decoder.59 | PP: 3/8 | Block rank: 11
+01/08/2025 11:14:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: module_name: model.decoder.60 | PP: 3/8 | Block rank: 12
+01/08/2025 11:14:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: module_name: model.decoder.61 | PP: 3/8 | Block rank: 13
+01/08/2025 11:14:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: module_name: model.decoder.62 | PP: 3/8 | Block rank: 14
+01/08/2025 11:14:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: module_name: model.decoder.63 | PP: 3/8 | Block rank: 15
+01/08/2025 11:14:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: module_name: model.decoder.64 | PP: 4/8 | Block rank: 0
+01/08/2025 11:14:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: module_name: model.decoder.65 | PP: 4/8 | Block rank: 1
+01/08/2025 11:14:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: module_name: model.decoder.66 | PP: 4/8 | Block rank: 2
+01/08/2025 11:14:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: module_name: model.decoder.67 | PP: 4/8 | Block rank: 3
+01/08/2025 11:14:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: module_name: model.decoder.68 | PP: 4/8 | Block rank: 4
+01/08/2025 11:14:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: module_name: model.decoder.69 | PP: 4/8 | Block rank: 5
+01/08/2025 11:14:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: module_name: model.decoder.70 | PP: 4/8 | Block rank: 6
+01/08/2025 11:14:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: module_name: model.decoder.71 | PP: 4/8 | Block rank: 7
+01/08/2025 11:14:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: module_name: model.decoder.72 | PP: 4/8 | Block rank: 8
+01/08/2025 11:14:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: module_name: model.decoder.73 | PP: 4/8 | Block rank: 9
+01/08/2025 11:14:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: module_name: model.decoder.74 | PP: 4/8 | Block rank: 10
+01/08/2025 11:14:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: module_name: model.decoder.75 | PP: 4/8 | Block rank: 11
+01/08/2025 11:14:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: module_name: model.decoder.76 | PP: 4/8 | Block rank: 12
+01/08/2025 11:14:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: module_name: model.decoder.77 | PP: 4/8 | Block rank: 13
+01/08/2025 11:14:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: module_name: model.decoder.78 | PP: 4/8 | Block rank: 14
+01/08/2025 11:14:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: module_name: model.decoder.79 | PP: 4/8 | Block rank: 15
+01/08/2025 11:14:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: module_name: model.decoder.80 | PP: 5/8 | Block rank: 0
+01/08/2025 11:14:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: module_name: model.decoder.81 | PP: 5/8 | Block rank: 1
+01/08/2025 11:14:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: module_name: model.decoder.82 | PP: 5/8 | Block rank: 2
+01/08/2025 11:14:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: module_name: model.decoder.83 | PP: 5/8 | Block rank: 3
+01/08/2025 11:14:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: module_name: model.decoder.84 | PP: 5/8 | Block rank: 4
+01/08/2025 11:14:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: module_name: model.decoder.85 | PP: 5/8 | Block rank: 5
+01/08/2025 11:14:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: module_name: model.decoder.86 | PP: 5/8 | Block rank: 6
+01/08/2025 11:14:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: module_name: model.decoder.87 | PP: 5/8 | Block rank: 7
+01/08/2025 11:14:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: module_name: model.decoder.88 | PP: 5/8 | Block rank: 8
+01/08/2025 11:14:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: module_name: model.decoder.89 | PP: 5/8 | Block rank: 9
+01/08/2025 11:14:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: module_name: model.decoder.90 | PP: 5/8 | Block rank: 10
+01/08/2025 11:14:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: module_name: model.decoder.91 | PP: 5/8 | Block rank: 11
+01/08/2025 11:14:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: module_name: model.decoder.92 | PP: 5/8 | Block rank: 12
+01/08/2025 11:14:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: module_name: model.decoder.93 | PP: 5/8 | Block rank: 13
+01/08/2025 11:14:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: module_name: model.decoder.94 | PP: 5/8 | Block rank: 14
+01/08/2025 11:14:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: module_name: model.decoder.95 | PP: 5/8 | Block rank: 15
+01/08/2025 11:14:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: module_name: model.decoder.96 | PP: 6/8 | Block rank: 0
+01/08/2025 11:14:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: module_name: model.decoder.97 | PP: 6/8 | Block rank: 1
+01/08/2025 11:14:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: module_name: model.decoder.98 | PP: 6/8 | Block rank: 2
+01/08/2025 11:14:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: module_name: model.decoder.99 | PP: 6/8 | Block rank: 3
+01/08/2025 11:14:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: module_name: model.decoder.100 | PP: 6/8 | Block rank: 4
+01/08/2025 11:14:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: module_name: model.decoder.101 | PP: 6/8 | Block rank: 5
+01/08/2025 11:14:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: module_name: model.decoder.102 | PP: 6/8 | Block rank: 6
+01/08/2025 11:14:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: module_name: model.decoder.103 | PP: 6/8 | Block rank: 7
+01/08/2025 11:14:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: module_name: model.decoder.104 | PP: 6/8 | Block rank: 8
+01/08/2025 11:14:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: module_name: model.decoder.105 | PP: 6/8 | Block rank: 9
+01/08/2025 11:14:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: module_name: model.decoder.106 | PP: 6/8 | Block rank: 10
+01/08/2025 11:14:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: module_name: model.decoder.107 | PP: 6/8 | Block rank: 11
+01/08/2025 11:14:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: module_name: model.decoder.108 | PP: 6/8 | Block rank: 12
+01/08/2025 11:14:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: module_name: model.decoder.109 | PP: 6/8 | Block rank: 13
+01/08/2025 11:14:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: module_name: model.decoder.110 | PP: 6/8 | Block rank: 14
+01/08/2025 11:14:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: module_name: model.decoder.111 | PP: 6/8 | Block rank: 15
+01/08/2025 11:14:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: module_name: model.decoder.112 | PP: 7/8 | Block rank: 0
+01/08/2025 11:14:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: module_name: model.decoder.113 | PP: 7/8 | Block rank: 1
+01/08/2025 11:14:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: module_name: model.decoder.114 | PP: 7/8 | Block rank: 2
+01/08/2025 11:14:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: module_name: model.decoder.115 | PP: 7/8 | Block rank: 3
+01/08/2025 11:14:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: module_name: model.decoder.116 | PP: 7/8 | Block rank: 4
+01/08/2025 11:14:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: module_name: model.decoder.117 | PP: 7/8 | Block rank: 5
+01/08/2025 11:14:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: module_name: model.decoder.118 | PP: 7/8 | Block rank: 6
+01/08/2025 11:14:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: module_name: model.decoder.119 | PP: 7/8 | Block rank: 7
+01/08/2025 11:14:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: module_name: model.decoder.120 | PP: 7/8 | Block rank: 8
+01/08/2025 11:14:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: module_name: model.decoder.121 | PP: 7/8 | Block rank: 9
+01/08/2025 11:14:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: module_name: model.decoder.122 | PP: 7/8 | Block rank: 10
+01/08/2025 11:14:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: module_name: model.decoder.123 | PP: 7/8 | Block rank: 11
+01/08/2025 11:14:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: module_name: model.decoder.124 | PP: 7/8 | Block rank: 12
+01/08/2025 11:14:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: module_name: model.decoder.125 | PP: 7/8 | Block rank: 13
+01/08/2025 11:14:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: module_name: model.final_layer_norm | PP: 7/8 | Block rank: 14
+01/08/2025 11:14:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: module_name: model.lm_head | PP: 7/8 | Block rank: 15
+01/08/2025 11:14:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: module_name: model.cast_to_fp32 | PP: 7/8 | Block rank: 16
+01/08/2025 11:14:25 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: module_name: loss | PP: 7/8 | Block rank: 17
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+01/08/2025 11:14:27 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: Total number of parameters: 406G (774335.25MiB)
+01/08/2025 11:14:27 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: Local number of parameters: 6.64G (12673.00MiB)
+01/08/2025 11:14:27 [INFO|DP=0|PP=4|TP=0|ip-26-0-167-217]: Local number of parameters: 6.38G (12161.00MiB)
+01/08/2025 11:14:27 [INFO|DP=0|PP=2|TP=0|ip-26-0-165-131]: Local number of parameters: 6.38G (12161.00MiB)
+01/08/2025 11:14:27 [INFO|DP=0|PP=6|TP=0|ip-26-0-168-30]: Local number of parameters: 6.38G (12161.00MiB)
+01/08/2025 11:14:27 [INFO|DP=0|PP=1|TP=0|ip-26-0-162-46]: Local number of parameters: 6.38G (12161.00MiB)
+01/08/2025 11:14:27 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: [After model building] Memory usage: 12673.03MiB. Peak allocated: 12673.09MiB Peak reserved: 13222.00MiB
+01/08/2025 11:14:27 [INFO|DP=0|PP=2|TP=0|ip-26-0-165-131]: [After model building] Memory usage: 12161.03MiB. Peak allocated: 12161.09MiB Peak reserved: 12726.00MiB
+01/08/2025 11:14:27 [INFO|DP=0|PP=4|TP=0|ip-26-0-167-217]: [After model building] Memory usage: 12161.03MiB. Peak allocated: 12161.09MiB Peak reserved: 12726.00MiB
+01/08/2025 11:14:27 [INFO|DP=0|PP=1|TP=0|ip-26-0-162-46]: [After model building] Memory usage: 12161.03MiB. Peak allocated: 12161.09MiB Peak reserved: 12726.00MiB
+01/08/2025 11:14:27 [INFO|DP=0|PP=6|TP=0|ip-26-0-168-30]: [After model building] Memory usage: 12161.03MiB. Peak allocated: 12161.09MiB Peak reserved: 12726.00MiB
+01/08/2025 11:14:27 [INFO|DP=0|PP=3|TP=0|ip-26-0-165-38]: Local number of parameters: 6.38G (12161.00MiB)
+01/08/2025 11:14:27 [INFO|DP=0|PP=5|TP=0|ip-26-0-168-120]: Local number of parameters: 6.38G (12161.00MiB)
+01/08/2025 11:14:27 [INFO|DP=0|PP=7|TP=0|ip-26-0-168-52]: Local number of parameters: 5.85G (11152.91MiB)
+01/08/2025 11:14:27 [INFO|DP=0|PP=3|TP=0|ip-26-0-165-38]: [After model building] Memory usage: 12161.03MiB. Peak allocated: 12161.09MiB Peak reserved: 12726.00MiB
+01/08/2025 11:14:27 [INFO|DP=0|PP=5|TP=0|ip-26-0-168-120]: [After model building] Memory usage: 12161.03MiB. Peak allocated: 12161.09MiB Peak reserved: 12670.00MiB
+01/08/2025 11:14:27 [INFO|DP=0|PP=7|TP=0|ip-26-0-168-52]: [After model building] Memory usage: 11152.93MiB. Peak allocated: 11152.99MiB Peak reserved: 11718.00MiB
+01/08/2025 11:14:27 [INFO|DP=0|PP=0|TP=1|ip-26-0-162-180]: Local number of parameters: 6.64G (12673.00MiB)
+01/08/2025 11:14:27 [INFO|DP=0|PP=0|TP=2|ip-26-0-162-180]: Local number of parameters: 6.64G (12673.00MiB)
+01/08/2025 11:14:27 [INFO|DP=0|PP=0|TP=5|ip-26-0-162-180]: Local number of parameters: 6.64G (12673.00MiB)
+01/08/2025 11:14:27 [INFO|DP=0|PP=0|TP=3|ip-26-0-162-180]: Local number of parameters: 6.64G (12673.00MiB)
+01/08/2025 11:14:27 [INFO|DP=0|PP=0|TP=1|ip-26-0-162-180]: [After model building] Memory usage: 12673.03MiB. Peak allocated: 12673.09MiB Peak reserved: 13222.00MiB
+01/08/2025 11:14:27 [INFO|DP=0|PP=0|TP=2|ip-26-0-162-180]: [After model building] Memory usage: 12673.03MiB. Peak allocated: 12673.09MiB Peak reserved: 13222.00MiB
+01/08/2025 11:14:27 [INFO|DP=0|PP=0|TP=4|ip-26-0-162-180]: Local number of parameters: 6.64G (12673.00MiB)
+01/08/2025 11:14:27 [INFO|DP=0|PP=0|TP=6|ip-26-0-162-180]: Local number of parameters: 6.64G (12673.00MiB)
+01/08/2025 11:14:27 [INFO|DP=0|PP=0|TP=5|ip-26-0-162-180]: [After model building] Memory usage: 12673.03MiB. Peak allocated: 12673.09MiB Peak reserved: 13222.00MiB
+01/08/2025 11:14:27 [INFO|DP=0|PP=0|TP=6|ip-26-0-162-180]: [After model building] Memory usage: 12673.03MiB. Peak allocated: 12673.09MiB Peak reserved: 13222.00MiB
+01/08/2025 11:14:27 [INFO|DP=0|PP=0|TP=3|ip-26-0-162-180]: [After model building] Memory usage: 12673.03MiB. Peak allocated: 12673.09MiB Peak reserved: 13222.00MiB
+01/08/2025 11:14:27 [INFO|DP=0|PP=0|TP=4|ip-26-0-162-180]: [After model building] Memory usage: 12673.03MiB. Peak allocated: 12673.09MiB Peak reserved: 13222.00MiB
+01/08/2025 11:14:27 [INFO|DP=0|PP=0|TP=7|ip-26-0-162-180]: Local number of parameters: 6.64G (12673.00MiB)
+01/08/2025 11:14:27 [INFO|DP=0|PP=0|TP=7|ip-26-0-162-180]: [After model building] Memory usage: 12673.03MiB. Peak allocated: 12673.09MiB Peak reserved: 13222.00MiB
+01/08/2025 11:14:27 [INFO|DP=0|PP=4|TP=1|ip-26-0-167-217]: Local number of parameters: 6.38G (12161.00MiB)
+01/08/2025 11:14:27 [INFO|DP=0|PP=2|TP=1|ip-26-0-165-131]: Local number of parameters: 6.38G (12161.00MiB)
+01/08/2025 11:14:27 [INFO|DP=0|PP=1|TP=1|ip-26-0-162-46]: Local number of parameters: 6.38G (12161.00MiB)
+01/08/2025 11:14:27 [INFO|DP=0|PP=6|TP=1|ip-26-0-168-30]: Local number of parameters: 6.38G (12161.00MiB)
+01/08/2025 11:14:27 [INFO|DP=0|PP=3|TP=1|ip-26-0-165-38]: Local number of parameters: 6.38G (12161.00MiB)
+01/08/2025 11:14:27 [INFO|DP=0|PP=2|TP=3|ip-26-0-165-131]: Local number of parameters: 6.38G (12161.00MiB)
+01/08/2025 11:14:27 [INFO|DP=0|PP=7|TP=1|ip-26-0-168-52]: Local number of parameters: 5.85G (11152.91MiB)
+01/08/2025 11:14:27 [INFO|DP=0|PP=1|TP=3|ip-26-0-162-46]: Local number of parameters: 6.38G (12161.00MiB)
+01/08/2025 11:14:27 [INFO|DP=0|PP=5|TP=1|ip-26-0-168-120]: Local number of parameters: 6.38G (12161.00MiB)
+01/08/2025 11:14:27 [INFO|DP=0|PP=4|TP=3|ip-26-0-167-217]: Local number of parameters: 6.38G (12161.00MiB)
+01/08/2025 11:14:27 [INFO|DP=0|PP=2|TP=1|ip-26-0-165-131]: [After model building] Memory usage: 12161.03MiB. Peak allocated: 12161.09MiB Peak reserved: 12726.00MiB
+01/08/2025 11:14:27 [INFO|DP=0|PP=6|TP=3|ip-26-0-168-30]: Local number of parameters: 6.38G (12161.00MiB)
+01/08/2025 11:14:27 [INFO|DP=0|PP=4|TP=1|ip-26-0-167-217]: [After model building] Memory usage: 12161.03MiB. Peak allocated: 12161.09MiB Peak reserved: 12726.00MiB
+01/08/2025 11:14:27 [INFO|DP=0|PP=1|TP=1|ip-26-0-162-46]: [After model building] Memory usage: 12161.03MiB. Peak allocated: 12161.09MiB Peak reserved: 12726.00MiB
+01/08/2025 11:14:27 [INFO|DP=0|PP=7|TP=3|ip-26-0-168-52]: Local number of parameters: 5.85G (11152.91MiB)
+01/08/2025 11:14:27 [INFO|DP=0|PP=5|TP=3|ip-26-0-168-120]: Local number of parameters: 6.38G (12161.00MiB)
+01/08/2025 11:14:27 [INFO|DP=0|PP=6|TP=1|ip-26-0-168-30]: [After model building] Memory usage: 12161.03MiB. Peak allocated: 12161.09MiB Peak reserved: 12726.00MiB
+01/08/2025 11:14:27 [INFO|DP=0|PP=3|TP=3|ip-26-0-165-38]: Local number of parameters: 6.38G (12161.00MiB)
+01/08/2025 11:14:27 [INFO|DP=0|PP=3|TP=1|ip-26-0-165-38]: [After model building] Memory usage: 12161.03MiB. Peak allocated: 12161.09MiB Peak reserved: 12726.00MiB
+01/08/2025 11:14:27 [INFO|DP=0|PP=6|TP=5|ip-26-0-168-30]: Local number of parameters: 6.38G (12161.00MiB)
+01/08/2025 11:14:27 [INFO|DP=0|PP=2|TP=5|ip-26-0-165-131]: Local number of parameters: 6.38G (12161.00MiB)
+01/08/2025 11:14:27 [INFO|DP=0|PP=2|TP=3|ip-26-0-165-131]: [After model building] Memory usage: 12161.03MiB. Peak allocated: 12161.09MiB Peak reserved: 12726.00MiB
+01/08/2025 11:14:27 [INFO|DP=0|PP=1|TP=5|ip-26-0-162-46]: Local number of parameters: 6.38G (12161.00MiB)
+01/08/2025 11:14:27 [INFO|DP=0|PP=1|TP=3|ip-26-0-162-46]: [After model building] Memory usage: 12161.03MiB. Peak allocated: 12161.09MiB Peak reserved: 12726.00MiB
+01/08/2025 11:14:27 [INFO|DP=0|PP=7|TP=1|ip-26-0-168-52]: [After model building] Memory usage: 11152.93MiB. Peak allocated: 11152.99MiB Peak reserved: 11718.00MiB
+01/08/2025 11:14:27 [INFO|DP=0|PP=4|TP=5|ip-26-0-167-217]: Local number of parameters: 6.38G (12161.00MiB)
+01/08/2025 11:14:27 [INFO|DP=0|PP=3|TP=5|ip-26-0-165-38]: Local number of parameters: 6.38G (12161.00MiB)
+01/08/2025 11:14:27 [INFO|DP=0|PP=5|TP=1|ip-26-0-168-120]: [After model building] Memory usage: 12161.03MiB. Peak allocated: 12161.09MiB Peak reserved: 12670.00MiB
+01/08/2025 11:14:27 [INFO|DP=0|PP=4|TP=3|ip-26-0-167-217]: [After model building] Memory usage: 12161.03MiB. Peak allocated: 12161.09MiB Peak reserved: 12726.00MiB
+01/08/2025 11:14:27 [INFO|DP=0|PP=7|TP=5|ip-26-0-168-52]: Local number of parameters: 5.85G (11152.91MiB)
+01/08/2025 11:14:27 [INFO|DP=0|PP=6|TP=3|ip-26-0-168-30]: [After model building] Memory usage: 12161.03MiB. Peak allocated: 12161.09MiB Peak reserved: 12726.00MiB
+01/08/2025 11:14:27 [INFO|DP=0|PP=7|TP=3|ip-26-0-168-52]: [After model building] Memory usage: 11152.93MiB. Peak allocated: 11152.99MiB Peak reserved: 11718.00MiB
+01/08/2025 11:14:27 [INFO|DP=0|PP=5|TP=5|ip-26-0-168-120]: Local number of parameters: 6.38G (12161.00MiB)
+01/08/2025 11:14:27 [INFO|DP=0|PP=3|TP=3|ip-26-0-165-38]: [After model building] Memory usage: 12161.03MiB. Peak allocated: 12161.09MiB Peak reserved: 12726.00MiB
+01/08/2025 11:14:27 [INFO|DP=0|PP=4|TP=4|ip-26-0-167-217]: Local number of parameters: 6.38G (12161.00MiB)
+01/08/2025 11:14:27 [INFO|DP=0|PP=5|TP=3|ip-26-0-168-120]: [After model building] Memory usage: 12161.03MiB. Peak allocated: 12161.09MiB Peak reserved: 12670.00MiB
+01/08/2025 11:14:27 [INFO|DP=0|PP=6|TP=5|ip-26-0-168-30]: [After model building] Memory usage: 12161.03MiB. Peak allocated: 12161.09MiB Peak reserved: 12726.00MiB
+01/08/2025 11:14:27 [INFO|DP=0|PP=2|TP=4|ip-26-0-165-131]: Local number of parameters: 6.38G (12161.00MiB)
+01/08/2025 11:14:27 [INFO|DP=0|PP=3|TP=4|ip-26-0-165-38]: Local number of parameters: 6.38G (12161.00MiB)
+01/08/2025 11:14:27 [INFO|DP=0|PP=2|TP=5|ip-26-0-165-131]: [After model building] Memory usage: 12161.03MiB. Peak allocated: 12161.09MiB Peak reserved: 12726.00MiB
+01/08/2025 11:14:27 [INFO|DP=0|PP=1|TP=4|ip-26-0-162-46]: Local number of parameters: 6.38G (12161.00MiB)
+01/08/2025 11:14:27 [INFO|DP=0|PP=4|TP=5|ip-26-0-167-217]: [After model building] Memory usage: 12161.03MiB. Peak allocated: 12161.09MiB Peak reserved: 12726.00MiB
+01/08/2025 11:14:27 [INFO|DP=0|PP=1|TP=5|ip-26-0-162-46]: [After model building] Memory usage: 12161.03MiB. Peak allocated: 12161.09MiB Peak reserved: 12726.00MiB
+01/08/2025 11:14:27 [INFO|DP=0|PP=3|TP=5|ip-26-0-165-38]: [After model building] Memory usage: 12161.03MiB. Peak allocated: 12161.09MiB Peak reserved: 12726.00MiB
+01/08/2025 11:14:27 [INFO|DP=0|PP=7|TP=5|ip-26-0-168-52]: [After model building] Memory usage: 11152.93MiB. Peak allocated: 11152.99MiB Peak reserved: 11718.00MiB
+01/08/2025 11:14:27 [INFO|DP=0|PP=4|TP=4|ip-26-0-167-217]: [After model building] Memory usage: 12161.03MiB. Peak allocated: 12161.09MiB Peak reserved: 12726.00MiB
+01/08/2025 11:14:27 [INFO|DP=0|PP=2|TP=4|ip-26-0-165-131]: [After model building] Memory usage: 12161.03MiB. Peak allocated: 12161.09MiB Peak reserved: 12726.00MiB
+01/08/2025 11:14:27 [INFO|DP=0|PP=6|TP=2|ip-26-0-168-30]: Local number of parameters: 6.38G (12161.00MiB)
+01/08/2025 11:14:27 [INFO|DP=0|PP=3|TP=4|ip-26-0-165-38]: [After model building] Memory usage: 12161.03MiB. Peak allocated: 12161.09MiB Peak reserved: 12726.00MiB
+01/08/2025 11:14:27 [INFO|DP=0|PP=4|TP=2|ip-26-0-167-217]: Local number of parameters: 6.38G (12161.00MiB)
+01/08/2025 11:14:27 [INFO|DP=0|PP=5|TP=5|ip-26-0-168-120]: [After model building] Memory usage: 12161.03MiB. Peak allocated: 12161.09MiB Peak reserved: 12670.00MiB
+01/08/2025 11:14:27 [INFO|DP=0|PP=2|TP=2|ip-26-0-165-131]: Local number of parameters: 6.38G (12161.00MiB)
+01/08/2025 11:14:27 [INFO|DP=0|PP=3|TP=2|ip-26-0-165-38]: Local number of parameters: 6.38G (12161.00MiB)
+01/08/2025 11:14:27 [INFO|DP=0|PP=1|TP=2|ip-26-0-162-46]: Local number of parameters: 6.38G (12161.00MiB)
+01/08/2025 11:14:27 [INFO|DP=0|PP=1|TP=4|ip-26-0-162-46]: [After model building] Memory usage: 12161.03MiB. Peak allocated: 12161.09MiB Peak reserved: 12726.00MiB
+01/08/2025 11:14:27 [INFO|DP=0|PP=5|TP=2|ip-26-0-168-120]: Local number of parameters: 6.38G (12161.00MiB)
+01/08/2025 11:14:27 [INFO|DP=0|PP=7|TP=2|ip-26-0-168-52]: Local number of parameters: 5.85G (11152.91MiB)
+01/08/2025 11:14:27 [INFO|DP=0|PP=3|TP=7|ip-26-0-165-38]: Local number of parameters: 6.38G (12161.00MiB)
+01/08/2025 11:14:27 [INFO|DP=0|PP=4|TP=7|ip-26-0-167-217]: Local number of parameters: 6.38G (12161.00MiB)
+01/08/2025 11:14:27 [INFO|DP=0|PP=6|TP=7|ip-26-0-168-30]: Local number of parameters: 6.38G (12161.00MiB)
+01/08/2025 11:14:27 [INFO|DP=0|PP=2|TP=7|ip-26-0-165-131]: Local number of parameters: 6.38G (12161.00MiB)
+01/08/2025 11:14:27 [INFO|DP=0|PP=6|TP=2|ip-26-0-168-30]: [After model building] Memory usage: 12161.03MiB. Peak allocated: 12161.09MiB Peak reserved: 12726.00MiB
+01/08/2025 11:14:27 [INFO|DP=0|PP=4|TP=2|ip-26-0-167-217]: [After model building] Memory usage: 12161.03MiB. Peak allocated: 12161.09MiB Peak reserved: 12726.00MiB
+01/08/2025 11:14:27 [INFO|DP=0|PP=3|TP=2|ip-26-0-165-38]: [After model building] Memory usage: 12161.03MiB. Peak allocated: 12161.09MiB Peak reserved: 12726.00MiB
+01/08/2025 11:14:27 [INFO|DP=0|PP=2|TP=2|ip-26-0-165-131]: [After model building] Memory usage: 12161.03MiB. Peak allocated: 12161.09MiB Peak reserved: 12726.00MiB
+01/08/2025 11:14:27 [INFO|DP=0|PP=5|TP=7|ip-26-0-168-120]: Local number of parameters: 6.38G (12161.00MiB)
+01/08/2025 11:14:27 [INFO|DP=0|PP=1|TP=7|ip-26-0-162-46]: Local number of parameters: 6.38G (12161.00MiB)
+01/08/2025 11:14:27 [INFO|DP=0|PP=1|TP=2|ip-26-0-162-46]: [After model building] Memory usage: 12161.03MiB. Peak allocated: 12161.09MiB Peak reserved: 12726.00MiB
+01/08/2025 11:14:27 [INFO|DP=0|PP=7|TP=7|ip-26-0-168-52]: Local number of parameters: 5.85G (11152.91MiB)
+01/08/2025 11:14:27 [INFO|DP=0|PP=5|TP=2|ip-26-0-168-120]: [After model building] Memory usage: 12161.03MiB. Peak allocated: 12161.09MiB Peak reserved: 12670.00MiB
+01/08/2025 11:14:27 [INFO|DP=0|PP=7|TP=2|ip-26-0-168-52]: [After model building] Memory usage: 11152.93MiB. Peak allocated: 11152.99MiB Peak reserved: 11718.00MiB
+01/08/2025 11:14:27 [INFO|DP=0|PP=4|TP=7|ip-26-0-167-217]: [After model building] Memory usage: 12161.03MiB. Peak allocated: 12161.09MiB Peak reserved: 12726.00MiB
+01/08/2025 11:14:27 [INFO|DP=0|PP=3|TP=7|ip-26-0-165-38]: [After model building] Memory usage: 12161.03MiB. Peak allocated: 12161.09MiB Peak reserved: 12726.00MiB
+01/08/2025 11:14:27 [INFO|DP=0|PP=6|TP=7|ip-26-0-168-30]: [After model building] Memory usage: 12161.03MiB. Peak allocated: 12161.09MiB Peak reserved: 12726.00MiB
+01/08/2025 11:14:27 [INFO|DP=0|PP=5|TP=7|ip-26-0-168-120]: [After model building] Memory usage: 12161.03MiB. Peak allocated: 12161.09MiB Peak reserved: 12670.00MiB
+01/08/2025 11:14:27 [INFO|DP=0|PP=2|TP=7|ip-26-0-165-131]: [After model building] Memory usage: 12161.03MiB. Peak allocated: 12161.09MiB Peak reserved: 12726.00MiB
+01/08/2025 11:14:27 [INFO|DP=0|PP=6|TP=4|ip-26-0-168-30]: Local number of parameters: 6.38G (12161.00MiB)
+01/08/2025 11:14:27 [INFO|DP=0|PP=1|TP=7|ip-26-0-162-46]: [After model building] Memory usage: 12161.03MiB. Peak allocated: 12161.09MiB Peak reserved: 12726.00MiB
+01/08/2025 11:14:27 [INFO|DP=0|PP=7|TP=7|ip-26-0-168-52]: [After model building] Memory usage: 11152.93MiB. Peak allocated: 11152.99MiB Peak reserved: 11718.00MiB
+01/08/2025 11:14:27 [INFO|DP=0|PP=4|TP=6|ip-26-0-167-217]: Local number of parameters: 6.38G (12161.00MiB)
+01/08/2025 11:14:27 [INFO|DP=0|PP=6|TP=6|ip-26-0-168-30]: Local number of parameters: 6.38G (12161.00MiB)
+01/08/2025 11:14:27 [INFO|DP=0|PP=1|TP=6|ip-26-0-162-46]: Local number of parameters: 6.38G (12161.00MiB)
+01/08/2025 11:14:27 [INFO|DP=0|PP=7|TP=4|ip-26-0-168-52]: Local number of parameters: 5.85G (11152.91MiB)
+01/08/2025 11:14:27 [INFO|DP=0|PP=2|TP=6|ip-26-0-165-131]: Local number of parameters: 6.38G (12161.00MiB)
+01/08/2025 11:14:27 [INFO|DP=0|PP=7|TP=6|ip-26-0-168-52]: Local number of parameters: 5.85G (11152.91MiB)
+01/08/2025 11:14:27 [INFO|DP=0|PP=5|TP=4|ip-26-0-168-120]: Local number of parameters: 6.38G (12161.00MiB)
+01/08/2025 11:14:27 [INFO|DP=0|PP=5|TP=6|ip-26-0-168-120]: Local number of parameters: 6.38G (12161.00MiB)
+01/08/2025 11:14:27 [INFO|DP=0|PP=3|TP=6|ip-26-0-165-38]: Local number of parameters: 6.38G (12161.00MiB)
+01/08/2025 11:14:27 [INFO|DP=0|PP=6|TP=4|ip-26-0-168-30]: [After model building] Memory usage: 12161.03MiB. Peak allocated: 12161.09MiB Peak reserved: 12726.00MiB
+01/08/2025 11:14:27 [INFO|DP=0|PP=4|TP=6|ip-26-0-167-217]: [After model building] Memory usage: 12161.03MiB. Peak allocated: 12161.09MiB Peak reserved: 12726.00MiB
+01/08/2025 11:14:27 [INFO|DP=0|PP=6|TP=6|ip-26-0-168-30]: [After model building] Memory usage: 12161.03MiB. Peak allocated: 12161.09MiB Peak reserved: 12726.00MiB
+01/08/2025 11:14:27 [INFO|DP=0|PP=1|TP=6|ip-26-0-162-46]: [After model building] Memory usage: 12161.03MiB. Peak allocated: 12161.09MiB Peak reserved: 12726.00MiB
+01/08/2025 11:14:27 [INFO|DP=0|PP=7|TP=4|ip-26-0-168-52]: [After model building] Memory usage: 11152.93MiB. Peak allocated: 11152.99MiB Peak reserved: 11718.00MiB
+01/08/2025 11:14:27 [INFO|DP=0|PP=2|TP=6|ip-26-0-165-131]: [After model building] Memory usage: 12161.03MiB. Peak allocated: 12161.09MiB Peak reserved: 12726.00MiB
+01/08/2025 11:14:27 [INFO|DP=0|PP=7|TP=6|ip-26-0-168-52]: [After model building] Memory usage: 11152.93MiB. Peak allocated: 11152.99MiB Peak reserved: 11718.00MiB
+01/08/2025 11:14:27 [INFO|DP=0|PP=5|TP=6|ip-26-0-168-120]: [After model building] Memory usage: 12161.03MiB. Peak allocated: 12161.09MiB Peak reserved: 12670.00MiB
+01/08/2025 11:14:27 [INFO|DP=0|PP=3|TP=6|ip-26-0-165-38]: [After model building] Memory usage: 12161.03MiB. Peak allocated: 12161.09MiB Peak reserved: 12726.00MiB
+01/08/2025 11:14:27 [INFO|DP=0|PP=5|TP=4|ip-26-0-168-120]: [After model building] Memory usage: 12161.03MiB. Peak allocated: 12161.09MiB Peak reserved: 12670.00MiB
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+01/08/2025 11:14:28 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: No checkpoint path provided.
+01/08/2025 11:14:28 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: Parametrizing model parameters using StandardParametrizator
+01/08/2025 11:14:28 [INFO|DP=0|PP=0|TP=0|ip-26-0-162-180]: [Optimizer Building] Using LearningRateForSP as learning rate
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 95, in __init__
+    big_flat_buffer = torch.empty(length, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 23.75 GiB. GPU 5 has a total capacty of 79.33 GiB of which 22.23 GiB is free. Including non-PyTorch memory, this process has 57.09 GiB memory in use. Of the allocated memory 47.50 GiB is allocated by PyTorch, and 544.98 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 95, in __init__
+    big_flat_buffer = torch.empty(length, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 23.75 GiB. GPU 6 has a total capacty of 79.33 GiB of which 22.23 GiB is free. Including non-PyTorch memory, this process has 57.09 GiB memory in use. Of the allocated memory 47.50 GiB is allocated by PyTorch, and 544.98 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 95, in __init__
+    big_flat_buffer = torch.empty(length, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 23.75 GiB. GPU 1 has a total capacty of 79.33 GiB of which 22.23 GiB is free. Including non-PyTorch memory, this process has 57.09 GiB memory in use. Of the allocated memory 47.50 GiB is allocated by PyTorch, and 544.98 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 95, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    big_flat_buffer = torch.empty(length, dtype=torch.float, device="cuda")
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 23.75 GiB. GPU 2 has a total capacty of 79.33 GiB of which 22.27 GiB is free. Including non-PyTorch memory, this process has 57.05 GiB memory in use. Of the allocated memory 47.50 GiB is allocated by PyTorch, and 544.98 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    trainer = DistributedTrainer(config_file)
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 95, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    big_flat_buffer = torch.empty(length, dtype=torch.float, device="cuda")
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 23.75 GiB. GPU 4 has a total capacty of 79.33 GiB of which 22.21 GiB is free. Including non-PyTorch memory, this process has 57.10 GiB memory in use. Of the allocated memory 47.50 GiB is allocated by PyTorch, and 544.98 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 95, in __init__
+    big_flat_buffer = torch.empty(length, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 23.75 GiB. GPU 0 has a total capacty of 79.33 GiB of which 22.30 GiB is free. Including non-PyTorch memory, this process has 57.02 GiB memory in use. Of the allocated memory 47.50 GiB is allocated by PyTorch, and 544.98 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 95, in __init__
+    big_flat_buffer = torch.empty(length, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 23.75 GiB. GPU 5 has a total capacty of 79.33 GiB of which 22.19 GiB is free. Including non-PyTorch memory, this process has 57.13 GiB memory in use. Of the allocated memory 47.50 GiB is allocated by PyTorch, and 544.98 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 95, in __init__
+    big_flat_buffer = torch.empty(length, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 23.75 GiB. GPU 3 has a total capacty of 79.33 GiB of which 22.07 GiB is free. Including non-PyTorch memory, this process has 57.24 GiB memory in use. Of the allocated memory 47.50 GiB is allocated by PyTorch, and 544.98 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    Traceback (most recent call last):
+gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 95, in __init__
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    big_flat_buffer = torch.empty(length, dtype=torch.float, device="cuda")
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 23.75 GiB. GPU 0 has a total capacty of 79.33 GiB of which 22.30 GiB is free. Including non-PyTorch memory, this process has 57.02 GiB memory in use. Of the allocated memory 47.50 GiB is allocated by PyTorch, and 544.98 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    optimizer = optimizer_builder(named_param_groups)
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 95, in __init__
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 95, in __init__
+    big_flat_buffer = torch.empty(length, dtype=torch.float, device="cuda")
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 23.75 GiB. GPU 0 has a total capacty of 79.33 GiB of which 22.26 GiB is free. Including non-PyTorch memory, this process has 57.06 GiB memory in use. Of the allocated memory 47.50 GiB is allocated by PyTorch, and 544.98 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 95, in __init__
+    big_flat_buffer = torch.empty(length, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 23.75 GiB. GPU 6 has a total capacty of 79.33 GiB of which 22.27 GiB is free. Including non-PyTorch memory, this process has 57.05 GiB memory in use. Of the allocated memory 47.50 GiB is allocated by PyTorch, and 544.98 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    trainer = DistributedTrainer(config_file)
+    big_flat_buffer = torch.empty(length, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 23.75 GiB. GPU 5 has a total capacty of 79.33 GiB of which 22.35 GiB is free. Including non-PyTorch memory, this process has 56.97 GiB memory in use. Of the allocated memory 47.50 GiB is allocated by PyTorch, and 544.98 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    optimizer = optimizer_builder(named_param_groups)
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    result = OptimizerFromGradientAccumulator(
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+        optimizer = optimizer_builder(named_param_groups)gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    trainer = DistributedTrainer(config_file)
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 95, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 95, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    big_flat_buffer = torch.empty(length, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 23.75 GiB. GPU 1 has a total capacty of 79.33 GiB of which 22.27 GiB is free. Including non-PyTorch memory, this process has 57.05 GiB memory in use. Of the allocated memory 47.50 GiB is allocated by PyTorch, and 544.98 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    big_flat_buffer = torch.empty(length, dtype=torch.float, device="cuda")
+    optimizer = optimizer_builder(named_param_groups)
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 23.75 GiB. GPU 2 has a total capacty of 79.33 GiB of which 22.23 GiB is free. Including non-PyTorch memory, this process has 57.09 GiB memory in use. Of the allocated memory 47.50 GiB is allocated by PyTorch, and 544.98 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 95, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 95, in __init__
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    result = OptimizerFromGradientAccumulator(
+    big_flat_buffer = torch.empty(length, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 23.75 GiB. GPU 7 has a total capacty of 79.33 GiB of which 22.43 GiB is free. Including non-PyTorch memory, this process has 56.89 GiB memory in use. Of the allocated memory 47.50 GiB is allocated by PyTorch, and 544.98 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    big_flat_buffer = torch.empty(length, dtype=torch.float, device="cuda")
+    result = OptimizerFromGradientAccumulator(
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 23.75 GiB. GPU 5 has a total capacty of 79.33 GiB of which 22.31 GiB is free. Including non-PyTorch memory, this process has 57.01 GiB memory in use. Of the allocated memory 47.50 GiB is allocated by PyTorch, and 544.98 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 95, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    big_flat_buffer = torch.empty(length, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 23.75 GiB. GPU 5 has a total capacty of 79.33 GiB of which 22.27 GiB is free. Including non-PyTorch memory, this process has 57.05 GiB memory in use. Of the allocated memory 47.50 GiB is allocated by PyTorch, and 544.98 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 95, in __init__
+    big_flat_buffer = torch.empty(length, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 23.75 GiB. GPU 3 has a total capacty of 79.33 GiB of which 22.19 GiB is free. Including non-PyTorch memory, this process has 57.13 GiB memory in use. Of the allocated memory 47.50 GiB is allocated by PyTorch, and 544.98 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    trainer = DistributedTrainer(config_file)
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 95, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    big_flat_buffer = torch.empty(length, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 23.75 GiB. GPU 2 has a total capacty of 79.33 GiB of which 22.31 GiB is free. Including non-PyTorch memory, this process has 57.01 GiB memory in use. Of the allocated memory 47.50 GiB is allocated by PyTorch, and 544.98 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 95, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 95, in __init__
+    big_flat_buffer = torch.empty(length, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 23.75 GiB. GPU 0 has a total capacty of 79.33 GiB of which 22.26 GiB is free. Including non-PyTorch memory, this process has 57.06 GiB memory in use. Of the allocated memory 47.50 GiB is allocated by PyTorch, and 544.98 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    big_flat_buffer = torch.empty(length, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 24.75 GiB. GPU 4 has a total capacty of 79.33 GiB of which 20.31 GiB is free. Including non-PyTorch memory, this process has 59.01 GiB memory in use. Of the allocated memory 49.50 GiB is allocated by PyTorch, and 528.98 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 95, in __init__
+    optimizer = optimizer_builder(named_param_groups)
+    big_flat_buffer = torch.empty(length, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 23.75 GiB. GPU 1 has a total capacty of 79.33 GiB of which 22.31 GiB is free. Including non-PyTorch memory, this process has 57.01 GiB memory in use. Of the allocated memory 47.50 GiB is allocated by PyTorch, and 544.98 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+Traceback (most recent call last):
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 95, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    trainer = DistributedTrainer(config_file)
+    big_flat_buffer = torch.empty(length, dtype=torch.float, device="cuda")
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 23.75 GiB. GPU 7 has a total capacty of 79.33 GiB of which 22.50 GiB is free. Including non-PyTorch memory, this process has 56.81 GiB memory in use. Of the allocated memory 47.50 GiB is allocated by PyTorch, and 544.98 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    result = OptimizerFromGradientAccumulator(
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+Traceback (most recent call last):
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    result = OptimizerFromGradientAccumulator(
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 95, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 95, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+        result = OptimizerFromGradientAccumulator(big_flat_buffer = torch.empty(length, dtype=torch.float, device="cuda")
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 23.75 GiB. GPU 0 has a total capacty of 79.33 GiB of which 22.22 GiB is free. Including non-PyTorch memory, this process has 57.10 GiB memory in use. Of the allocated memory 47.50 GiB is allocated by PyTorch, and 544.98 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    big_flat_buffer = torch.empty(length, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 23.75 GiB. GPU 3 has a total capacty of 79.33 GiB of which 22.19 GiB is free. Including non-PyTorch memory, this process has 57.13 GiB memory in use. Of the allocated memory 47.50 GiB is allocated by PyTorch, and 544.98 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 95, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    optimizer = optimizer_builder(named_param_groups)
+    big_flat_buffer = torch.empty(length, dtype=torch.float, device="cuda")
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 24.75 GiB. GPU 5 has a total capacty of 79.33 GiB of which 20.40 GiB is free. Including non-PyTorch memory, this process has 58.92 GiB memory in use. Of the allocated memory 49.50 GiB is allocated by PyTorch, and 528.98 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 95, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 95, in __init__
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    big_flat_buffer = torch.empty(length, dtype=torch.float, device="cuda")
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 23.75 GiB. GPU 6 has a total capacty of 79.33 GiB of which 22.19 GiB is free. Including non-PyTorch memory, this process has 57.13 GiB memory in use. Of the allocated memory 47.50 GiB is allocated by PyTorch, and 544.98 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    big_flat_buffer = torch.empty(length, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 23.75 GiB. GPU 5 has a total capacty of 79.33 GiB of which 22.23 GiB is free. Including non-PyTorch memory, this process has 57.09 GiB memory in use. Of the allocated memory 47.50 GiB is allocated by PyTorch, and 544.98 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 95, in __init__
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    big_flat_buffer = torch.empty(length, dtype=torch.float, device="cuda")
+    optimizer = optimizer_builder(named_param_groups)
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 23.75 GiB. GPU 1 has a total capacty of 79.33 GiB of which 22.31 GiB is free. Including non-PyTorch memory, this process has 57.01 GiB memory in use. Of the allocated memory 47.50 GiB is allocated by PyTorch, and 544.98 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 95, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 95, in __init__
+    big_flat_buffer = torch.empty(length, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 23.75 GiB. GPU 6 has a total capacty of 79.33 GiB of which 22.31 GiB is free. Including non-PyTorch memory, this process has 57.01 GiB memory in use. Of the allocated memory 47.50 GiB is allocated by PyTorch, and 544.98 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    big_flat_buffer = torch.empty(length, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 23.75 GiB. GPU 3 has a total capacty of 79.33 GiB of which 22.15 GiB is free. Including non-PyTorch memory, this process has 57.17 GiB memory in use. Of the allocated memory 47.50 GiB is allocated by PyTorch, and 544.98 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(    
+optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 95, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    big_flat_buffer = torch.empty(length, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 23.75 GiB. GPU 0 has a total capacty of 79.33 GiB of which 22.34 GiB is free. Including non-PyTorch memory, this process has 56.98 GiB memory in use. Of the allocated memory 47.50 GiB is allocated by PyTorch, and 544.98 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 95, in __init__
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    big_flat_buffer = torch.empty(length, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 23.75 GiB. GPU 3 has a total capacty of 79.33 GiB of which 22.11 GiB is free. Including non-PyTorch memory, this process has 57.21 GiB memory in use. Of the allocated memory 47.50 GiB is allocated by PyTorch, and 544.98 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 95, in __init__
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    big_flat_buffer = torch.empty(length, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 23.75 GiB. GPU 7 has a total capacty of 79.33 GiB of which 22.39 GiB is free. Including non-PyTorch memory, this process has 56.93 GiB memory in use. Of the allocated memory 47.50 GiB is allocated by PyTorch, and 544.98 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 95, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    big_flat_buffer = torch.empty(length, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 23.75 GiB. GPU 4 has a total capacty of 79.33 GiB of which 22.29 GiB is free. Including non-PyTorch memory, this process has 57.03 GiB memory in use. Of the allocated memory 47.50 GiB is allocated by PyTorch, and 544.98 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 95, in __init__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    big_flat_buffer = torch.empty(length, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 23.75 GiB. GPU 6 has a total capacty of 79.33 GiB of which 22.27 GiB is free. Including non-PyTorch memory, this process has 57.05 GiB memory in use. Of the allocated memory 47.50 GiB is allocated by PyTorch, and 544.98 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 95, in __init__
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    big_flat_buffer = torch.empty(length, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 23.75 GiB. GPU 2 has a total capacty of 79.33 GiB of which 22.19 GiB is free. Including non-PyTorch memory, this process has 57.13 GiB memory in use. Of the allocated memory 47.50 GiB is allocated by PyTorch, and 544.98 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    trainer = DistributedTrainer(config_file)
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    result = OptimizerFromGradientAccumulator(
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 95, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    big_flat_buffer = torch.empty(length, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 23.75 GiB. GPU 1 has a total capacty of 79.33 GiB of which 22.19 GiB is free. Including non-PyTorch memory, this process has 57.13 GiB memory in use. Of the allocated memory 47.50 GiB is allocated by PyTorch, and 544.98 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 95, in __init__
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    big_flat_buffer = torch.empty(length, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 23.75 GiB. GPU 7 has a total capacty of 79.33 GiB of which 22.46 GiB is free. Including non-PyTorch memory, this process has 56.85 GiB memory in use. Of the allocated memory 47.50 GiB is allocated by PyTorch, and 544.98 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 95, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    big_flat_buffer = torch.empty(length, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 23.75 GiB. GPU 4 has a total capacty of 79.33 GiB of which 22.25 GiB is free. Including non-PyTorch memory, this process has 57.06 GiB memory in use. Of the allocated memory 47.50 GiB is allocated by PyTorch, and 544.98 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 95, in __init__
+    big_flat_buffer = torch.empty(length, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 23.75 GiB. GPU 4 has a total capacty of 79.33 GiB of which 22.18 GiB is free. Including non-PyTorch memory, this process has 57.14 GiB memory in use. Of the allocated memory 47.50 GiB is allocated by PyTorch, and 544.98 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 95, in __init__
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    big_flat_buffer = torch.empty(length, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 23.75 GiB. GPU 7 has a total capacty of 79.33 GiB of which 22.50 GiB is free. Including non-PyTorch memory, this process has 56.81 GiB memory in use. Of the allocated memory 47.50 GiB is allocated by PyTorch, and 544.98 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    optimizer = optimizer_builder(named_param_groups)
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 95, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 95, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    big_flat_buffer = torch.empty(length, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 23.75 GiB. GPU 4 has a total capacty of 79.33 GiB of which 22.21 GiB is free. Including non-PyTorch memory, this process has 57.10 GiB memory in use. Of the allocated memory 47.50 GiB is allocated by PyTorch, and 544.98 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    big_flat_buffer = torch.empty(length, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 23.75 GiB. GPU 2 has a total capacty of 79.33 GiB of which 22.31 GiB is free. Including non-PyTorch memory, this process has 57.01 GiB memory in use. Of the allocated memory 47.50 GiB is allocated by PyTorch, and 544.98 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 95, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    big_flat_buffer = torch.empty(length, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 23.75 GiB. GPU 6 has a total capacty of 79.33 GiB of which 22.31 GiB is free. Including non-PyTorch memory, this process has 57.01 GiB memory in use. Of the allocated memory 47.50 GiB is allocated by PyTorch, and 544.98 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    trainer = DistributedTrainer(config_file)
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    result = OptimizerFromGradientAccumulator(
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 95, in __init__
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    big_flat_buffer = torch.empty(length, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 23.75 GiB. GPU 6 has a total capacty of 79.33 GiB of which 22.23 GiB is free. Including non-PyTorch memory, this process has 57.09 GiB memory in use. Of the allocated memory 47.50 GiB is allocated by PyTorch, and 544.98 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 95, in __init__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    big_flat_buffer = torch.empty(length, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 23.75 GiB. GPU 3 has a total capacty of 79.33 GiB of which 22.11 GiB is free. Including non-PyTorch memory, this process has 57.21 GiB memory in use. Of the allocated memory 47.50 GiB is allocated by PyTorch, and 544.98 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 95, in __init__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    big_flat_buffer = torch.empty(length, dtype=torch.float, device="cuda")
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())    
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 23.75 GiB. GPU 2 has a total capacty of 79.33 GiB of which 22.27 GiB is free. Including non-PyTorch memory, this process has 57.05 GiB memory in use. Of the allocated memory 47.50 GiB is allocated by PyTorch, and 544.98 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    trainer = DistributedTrainer(config_file)
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 95, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+Traceback (most recent call last):
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    big_flat_buffer = torch.empty(length, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 23.75 GiB. GPU 2 has a total capacty of 79.33 GiB of which 22.23 GiB is free. Including non-PyTorch memory, this process has 57.09 GiB memory in use. Of the allocated memory 47.50 GiB is allocated by PyTorch, and 544.98 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+        self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(optimizer = optimizer_builder(named_param_groups)
+
+    trainer = DistributedTrainer(config_file)
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    result = OptimizerFromGradientAccumulator(
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    trainer = DistributedTrainer(config_file)
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 95, in __init__
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    result = OptimizerFromGradientAccumulator(
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 95, in __init__
+    big_flat_buffer = torch.empty(length, dtype=torch.float, device="cuda")
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 23.75 GiB. GPU 5 has a total capacty of 79.33 GiB of which 22.31 GiB is free. Including non-PyTorch memory, this process has 57.01 GiB memory in use. Of the allocated memory 47.50 GiB is allocated by PyTorch, and 544.98 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    result = OptimizerFromGradientAccumulator(
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 95, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    big_flat_buffer = torch.empty(length, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 23.75 GiB. GPU 0 has a total capacty of 79.33 GiB of which 22.34 GiB is free. Including non-PyTorch memory, this process has 56.98 GiB memory in use. Of the allocated memory 47.50 GiB is allocated by PyTorch, and 544.98 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    result = OptimizerFromGradientAccumulator(
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    big_flat_buffer = torch.empty(length, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 23.75 GiB. GPU 1 has a total capacty of 79.33 GiB of which 22.23 GiB is free. Including non-PyTorch memory, this process has 57.09 GiB memory in use. Of the allocated memory 47.50 GiB is allocated by PyTorch, and 544.98 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 95, in __init__
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+        result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 95, in __init__
+gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    big_flat_buffer = torch.empty(length, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 23.75 GiB. GPU 3 has a total capacty of 79.33 GiB of which 22.23 GiB is free. Including non-PyTorch memory, this process has 57.09 GiB memory in use. Of the allocated memory 47.50 GiB is allocated by PyTorch, and 544.98 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    big_flat_buffer = torch.empty(length, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 23.75 GiB. GPU 4 has a total capacty of 79.33 GiB of which 22.25 GiB is free. Including non-PyTorch memory, this process has 57.06 GiB memory in use. Of the allocated memory 47.50 GiB is allocated by PyTorch, and 544.98 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 95, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 95, in __init__
+    big_flat_buffer = torch.empty(length, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 24.75 GiB. GPU 7 has a total capacty of 79.33 GiB of which 20.60 GiB is free. Including non-PyTorch memory, this process has 58.72 GiB memory in use. Of the allocated memory 49.50 GiB is allocated by PyTorch, and 528.98 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    trainer = DistributedTrainer(config_file)
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    big_flat_buffer = torch.empty(length, dtype=torch.float, device="cuda")
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 24.75 GiB. GPU 1 has a total capacty of 79.33 GiB of which 20.40 GiB is free. Including non-PyTorch memory, this process has 58.92 GiB memory in use. Of the allocated memory 49.50 GiB is allocated by PyTorch, and 528.98 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 95, in __init__
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 95, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 95, in __init__
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    big_flat_buffer = torch.empty(length, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 23.75 GiB. GPU 1 has a total capacty of 79.33 GiB of which 22.27 GiB is free. Including non-PyTorch memory, this process has 57.05 GiB memory in use. Of the allocated memory 47.50 GiB is allocated by PyTorch, and 544.98 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    big_flat_buffer = torch.empty(length, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 23.75 GiB. GPU 1 has a total capacty of 79.33 GiB of which 22.35 GiB is free. Including non-PyTorch memory, this process has 56.97 GiB memory in use. Of the allocated memory 47.50 GiB is allocated by PyTorch, and 544.98 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    big_flat_buffer = torch.empty(length, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 23.75 GiB. GPU 2 has a total capacty of 79.33 GiB of which 22.35 GiB is free. Including non-PyTorch memory, this process has 56.97 GiB memory in use. Of the allocated memory 47.50 GiB is allocated by PyTorch, and 544.98 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    optimizer = optimizer_builder(named_param_groups)
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 95, in __init__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    big_flat_buffer = torch.empty(length, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 24.75 GiB. GPU 2 has a total capacty of 79.33 GiB of which 20.32 GiB is free. Including non-PyTorch memory, this process has 58.99 GiB memory in use. Of the allocated memory 49.50 GiB is allocated by PyTorch, and 528.98 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 95, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    big_flat_buffer = torch.empty(length, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 23.75 GiB. GPU 7 has a total capacty of 79.33 GiB of which 22.54 GiB is free. Including non-PyTorch memory, this process has 56.78 GiB memory in use. Of the allocated memory 47.50 GiB is allocated by PyTorch, and 544.98 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 95, in __init__
+    trainer = DistributedTrainer(config_file)
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    big_flat_buffer = torch.empty(length, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 24.75 GiB. GPU 6 has a total capacty of 79.33 GiB of which 20.32 GiB is free. Including non-PyTorch memory, this process has 58.99 GiB memory in use. Of the allocated memory 49.50 GiB is allocated by PyTorch, and 528.98 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    optimizer = optimizer_builder(named_param_groups)
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 95, in __init__
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    big_flat_buffer = torch.empty(length, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 23.75 GiB. GPU 0 has a total capacty of 79.33 GiB of which 22.38 GiB is free. Including non-PyTorch memory, this process has 56.94 GiB memory in use. Of the allocated memory 47.50 GiB is allocated by PyTorch, and 544.98 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 95, in __init__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    big_flat_buffer = torch.empty(length, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 24.75 GiB. GPU 3 has a total capacty of 79.33 GiB of which 20.29 GiB is free. Including non-PyTorch memory, this process has 59.03 GiB memory in use. Of the allocated memory 49.50 GiB is allocated by PyTorch, and 528.98 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 95, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    big_flat_buffer = torch.empty(length, dtype=torch.float, device="cuda")
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 95, in __init__
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 24.75 GiB. GPU 5 has a total capacty of 79.33 GiB of which 20.99 GiB is free. Including non-PyTorch memory, this process has 58.33 GiB memory in use. Of the allocated memory 49.50 GiB is allocated by PyTorch, and 528.98 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    big_flat_buffer = torch.empty(length, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 23.75 GiB. GPU 7 has a total capacty of 79.33 GiB of which 22.46 GiB is free. Including non-PyTorch memory, this process has 56.85 GiB memory in use. Of the allocated memory 47.50 GiB is allocated by PyTorch, and 544.98 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 95, in __init__
+    big_flat_buffer = torch.empty(length, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 24.75 GiB. GPU 2 has a total capacty of 79.33 GiB of which 20.99 GiB is free. Including non-PyTorch memory, this process has 58.33 GiB memory in use. Of the allocated memory 49.50 GiB is allocated by PyTorch, and 528.98 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 95, in __init__
+    big_flat_buffer = torch.empty(length, dtype=torch.float, device="cuda")
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 24.75 GiB. GPU 3 has a total capacty of 79.33 GiB of which 20.87 GiB is free. Including non-PyTorch memory, this process has 58.45 GiB memory in use. Of the allocated memory 49.50 GiB is allocated by PyTorch, and 528.98 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 95, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    big_flat_buffer = torch.empty(length, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 24.75 GiB. GPU 0 has a total capacty of 79.33 GiB of which 20.36 GiB is free. Including non-PyTorch memory, this process has 58.96 GiB memory in use. Of the allocated memory 49.50 GiB is allocated by PyTorch, and 528.98 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 95, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    big_flat_buffer = torch.empty(length, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 23.75 GiB. GPU 3 has a total capacty of 79.33 GiB of which 22.15 GiB is free. Including non-PyTorch memory, this process has 57.17 GiB memory in use. Of the allocated memory 47.50 GiB is allocated by PyTorch, and 544.98 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+Traceback (most recent call last):
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 95, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 95, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    big_flat_buffer = torch.empty(length, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 23.75 GiB. GPU 7 has a total capacty of 79.33 GiB of which 22.43 GiB is free. Including non-PyTorch memory, this process has 56.89 GiB memory in use. Of the allocated memory 47.50 GiB is allocated by PyTorch, and 544.98 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    big_flat_buffer = torch.empty(length, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 23.75 GiB. GPU 4 has a total capacty of 79.33 GiB of which 22.33 GiB is free. Including non-PyTorch memory, this process has 56.99 GiB memory in use. Of the allocated memory 47.50 GiB is allocated by PyTorch, and 544.98 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 95, in __init__
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    big_flat_buffer = torch.empty(length, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 23.75 GiB. GPU 5 has a total capacty of 79.33 GiB of which 22.27 GiB is free. Including non-PyTorch memory, this process has 57.05 GiB memory in use. Of the allocated memory 47.50 GiB is allocated by PyTorch, and 544.98 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 95, in __init__
+    big_flat_buffer = torch.empty(length, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 23.75 GiB. GPU 6 has a total capacty of 79.33 GiB of which 22.35 GiB is free. Including non-PyTorch memory, this process has 56.97 GiB memory in use. Of the allocated memory 47.50 GiB is allocated by PyTorch, and 544.98 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 95, in __init__
+    big_flat_buffer = torch.empty(length, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 24.75 GiB. GPU 4 has a total capacty of 79.33 GiB of which 20.97 GiB is free. Including non-PyTorch memory, this process has 58.35 GiB memory in use. Of the allocated memory 49.50 GiB is allocated by PyTorch, and 528.98 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 95, in __init__
+    big_flat_buffer = torch.empty(length, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 23.75 GiB. GPU 4 has a total capacty of 79.33 GiB of which 22.29 GiB is free. Including non-PyTorch memory, this process has 57.03 GiB memory in use. Of the allocated memory 47.50 GiB is allocated by PyTorch, and 544.98 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 95, in __init__
+    big_flat_buffer = torch.empty(length, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 23.75 GiB. GPU 4 has a total capacty of 79.33 GiB of which 22.29 GiB is free. Including non-PyTorch memory, this process has 57.03 GiB memory in use. Of the allocated memory 47.50 GiB is allocated by PyTorch, and 544.98 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 95, in __init__
+    big_flat_buffer = torch.empty(length, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 24.75 GiB. GPU 7 has a total capacty of 79.33 GiB of which 21.18 GiB is free. Including non-PyTorch memory, this process has 58.13 GiB memory in use. Of the allocated memory 49.50 GiB is allocated by PyTorch, and 528.98 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 95, in __init__
+    big_flat_buffer = torch.empty(length, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 24.75 GiB. GPU 1 has a total capacty of 79.33 GiB of which 20.99 GiB is free. Including non-PyTorch memory, this process has 58.33 GiB memory in use. Of the allocated memory 49.50 GiB is allocated by PyTorch, and 528.98 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 95, in __init__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    big_flat_buffer = torch.empty(length, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 24.75 GiB. GPU 6 has a total capacty of 79.33 GiB of which 20.99 GiB is free. Including non-PyTorch memory, this process has 58.33 GiB memory in use. Of the allocated memory 49.50 GiB is allocated by PyTorch, and 528.98 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 95, in __init__
+    big_flat_buffer = torch.empty(length, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 24.75 GiB. GPU 0 has a total capacty of 79.33 GiB of which 21.02 GiB is free. Including non-PyTorch memory, this process has 58.30 GiB memory in use. Of the allocated memory 49.50 GiB is allocated by PyTorch, and 528.98 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 95, in __init__
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    big_flat_buffer = torch.empty(length, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 23.75 GiB. GPU 1 has a total capacty of 79.33 GiB of which 22.35 GiB is free. Including non-PyTorch memory, this process has 56.97 GiB memory in use. Of the allocated memory 47.50 GiB is allocated by PyTorch, and 544.98 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 95, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 95, in __init__
+    big_flat_buffer = torch.empty(length, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 23.75 GiB. GPU 2 has a total capacty of 79.33 GiB of which 22.27 GiB is free. Including non-PyTorch memory, this process has 57.05 GiB memory in use. Of the allocated memory 47.50 GiB is allocated by PyTorch, and 544.98 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    big_flat_buffer = torch.empty(length, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 23.75 GiB. GPU 5 has a total capacty of 79.33 GiB of which 22.35 GiB is free. Including non-PyTorch memory, this process has 56.97 GiB memory in use. Of the allocated memory 47.50 GiB is allocated by PyTorch, and 544.98 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 95, in __init__
+    big_flat_buffer = torch.empty(length, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 23.75 GiB. GPU 0 has a total capacty of 79.33 GiB of which 22.30 GiB is free. Including non-PyTorch memory, this process has 57.02 GiB memory in use. Of the allocated memory 47.50 GiB is allocated by PyTorch, and 544.98 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 95, in __init__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    big_flat_buffer = torch.empty(length, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 23.75 GiB. GPU 0 has a total capacty of 79.33 GiB of which 22.34 GiB is free. Including non-PyTorch memory, this process has 56.98 GiB memory in use. Of the allocated memory 47.50 GiB is allocated by PyTorch, and 544.98 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 95, in __init__
+    big_flat_buffer = torch.empty(length, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 23.75 GiB. GPU 4 has a total capacty of 79.33 GiB of which 22.25 GiB is free. Including non-PyTorch memory, this process has 57.06 GiB memory in use. Of the allocated memory 47.50 GiB is allocated by PyTorch, and 544.98 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    trainer = DistributedTrainer(config_file)
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+Traceback (most recent call last):
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    trainer = DistributedTrainer(config_file)
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 95, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    optimizer = optimizer_builder(named_param_groups)    
+big_flat_buffer = torch.empty(length, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 95, in __init__
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 23.75 GiB. GPU 7 has a total capacty of 79.33 GiB of which 22.54 GiB is free. Including non-PyTorch memory, this process has 56.78 GiB memory in use. Of the allocated memory 47.50 GiB is allocated by PyTorch, and 544.98 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 95, in __init__
+    result = OptimizerFromGradientAccumulator(
+    big_flat_buffer = torch.empty(length, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 23.75 GiB. GPU 3 has a total capacty of 79.33 GiB of which 22.11 GiB is free. Including non-PyTorch memory, this process has 57.21 GiB memory in use. Of the allocated memory 47.50 GiB is allocated by PyTorch, and 544.98 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    big_flat_buffer = torch.empty(length, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 23.75 GiB. GPU 1 has a total capacty of 79.33 GiB of which 22.23 GiB is free. Including non-PyTorch memory, this process has 57.09 GiB memory in use. Of the allocated memory 47.50 GiB is allocated by PyTorch, and 544.98 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+        gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 95, in __init__
+    big_flat_buffer = torch.empty(length, dtype=torch.float, device="cuda")
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 23.75 GiB. GPU 3 has a total capacty of 79.33 GiB of which 22.23 GiB is free. Including non-PyTorch memory, this process has 57.09 GiB memory in use. Of the allocated memory 47.50 GiB is allocated by PyTorch, and 544.98 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 95, in __init__
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    big_flat_buffer = torch.empty(length, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 23.75 GiB. GPU 6 has a total capacty of 79.33 GiB of which 22.27 GiB is free. Including non-PyTorch memory, this process has 57.05 GiB memory in use. Of the allocated memory 47.50 GiB is allocated by PyTorch, and 544.98 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 95, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    big_flat_buffer = torch.empty(length, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 23.75 GiB. GPU 6 has a total capacty of 79.33 GiB of which 22.31 GiB is free. Including non-PyTorch memory, this process has 57.01 GiB memory in use. Of the allocated memory 47.50 GiB is allocated by PyTorch, and 544.98 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 95, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    big_flat_buffer = torch.empty(length, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 23.75 GiB. GPU 2 has a total capacty of 79.33 GiB of which 22.31 GiB is free. Including non-PyTorch memory, this process has 57.01 GiB memory in use. Of the allocated memory 47.50 GiB is allocated by PyTorch, and 544.98 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 95, in __init__
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    big_flat_buffer = torch.empty(length, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 23.75 GiB. GPU 5 has a total capacty of 79.33 GiB of which 22.23 GiB is free. Including non-PyTorch memory, this process has 57.09 GiB memory in use. Of the allocated memory 47.50 GiB is allocated by PyTorch, and 544.98 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 95, in __init__
+    big_flat_buffer = torch.empty(length, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 23.75 GiB. GPU 7 has a total capacty of 79.33 GiB of which 22.43 GiB is free. Including non-PyTorch memory, this process has 56.89 GiB memory in use. Of the allocated memory 47.50 GiB is allocated by PyTorch, and 544.98 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 95, in __init__
+    big_flat_buffer = torch.empty(length, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 23.75 GiB. GPU 0 has a total capacty of 79.33 GiB of which 22.26 GiB is free. Including non-PyTorch memory, this process has 57.06 GiB memory in use. Of the allocated memory 47.50 GiB is allocated by PyTorch, and 544.98 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 95, in __init__
+    big_flat_buffer = torch.empty(length, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 23.75 GiB. GPU 2 has a total capacty of 79.33 GiB of which 22.40 GiB is free. Including non-PyTorch memory, this process has 56.92 GiB memory in use. Of the allocated memory 47.50 GiB is allocated by PyTorch, and 488.98 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+Traceback (most recent call last):
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    trainer = DistributedTrainer(config_file)
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    optimizer = optimizer_builder(named_param_groups)
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 95, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+    big_flat_buffer = torch.empty(length, dtype=torch.float, device="cuda")
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 95, in __init__
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 23.75 GiB. GPU 6 has a total capacty of 79.33 GiB of which 22.40 GiB is free. Including non-PyTorch memory, this process has 56.92 GiB memory in use. Of the allocated memory 47.50 GiB is allocated by PyTorch, and 488.98 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 95, in __init__
+    big_flat_buffer = torch.empty(length, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 23.75 GiB. GPU 6 has a total capacty of 79.33 GiB of which 22.23 GiB is free. Including non-PyTorch memory, this process has 57.09 GiB memory in use. Of the allocated memory 47.50 GiB is allocated by PyTorch, and 544.98 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    big_flat_buffer = torch.empty(length, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 23.75 GiB. GPU 0 has a total capacty of 79.33 GiB of which 22.43 GiB is free. Including non-PyTorch memory, this process has 56.88 GiB memory in use. Of the allocated memory 47.50 GiB is allocated by PyTorch, and 488.98 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 95, in __init__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    big_flat_buffer = torch.empty(length, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 23.75 GiB. GPU 3 has a total capacty of 79.33 GiB of which 22.21 GiB is free. Including non-PyTorch memory, this process has 57.11 GiB memory in use. Of the allocated memory 47.50 GiB is allocated by PyTorch, and 488.98 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 95, in __init__
+    big_flat_buffer = torch.empty(length, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 23.75 GiB. GPU 2 has a total capacty of 79.33 GiB of which 22.23 GiB is free. Including non-PyTorch memory, this process has 57.09 GiB memory in use. Of the allocated memory 47.50 GiB is allocated by PyTorch, and 544.98 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 95, in __init__
+    big_flat_buffer = torch.empty(length, dtype=torch.float, device="cuda")
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 23.75 GiB. GPU 3 has a total capacty of 79.33 GiB of which 22.19 GiB is free. Including non-PyTorch memory, this process has 57.13 GiB memory in use. Of the allocated memory 47.50 GiB is allocated by PyTorch, and 544.98 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 95, in __init__
+    big_flat_buffer = torch.empty(length, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 23.75 GiB. GPU 5 has a total capacty of 79.33 GiB of which 22.31 GiB is free. Including non-PyTorch memory, this process has 57.01 GiB memory in use. Of the allocated memory 47.50 GiB is allocated by PyTorch, and 544.98 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 95, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    big_flat_buffer = torch.empty(length, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 23.75 GiB. GPU 1 has a total capacty of 79.33 GiB of which 22.31 GiB is free. Including non-PyTorch memory, this process has 57.01 GiB memory in use. Of the allocated memory 47.50 GiB is allocated by PyTorch, and 544.98 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 95, in __init__
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    big_flat_buffer = torch.empty(length, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 23.75 GiB. GPU 7 has a total capacty of 79.33 GiB of which 22.50 GiB is free. Including non-PyTorch memory, this process has 56.81 GiB memory in use. Of the allocated memory 47.50 GiB is allocated by PyTorch, and 544.98 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 95, in __init__
+    big_flat_buffer = torch.empty(length, dtype=torch.float, device="cuda")
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 23.75 GiB. GPU 4 has a total capacty of 79.33 GiB of which 22.21 GiB is free. Including non-PyTorch memory, this process has 57.10 GiB memory in use. Of the allocated memory 47.50 GiB is allocated by PyTorch, and 544.98 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 95, in __init__
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    big_flat_buffer = torch.empty(length, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 23.75 GiB. GPU 7 has a total capacty of 79.33 GiB of which 22.52 GiB is free. Including non-PyTorch memory, this process has 56.80 GiB memory in use. Of the allocated memory 47.50 GiB is allocated by PyTorch, and 488.98 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 95, in __init__
+    big_flat_buffer = torch.empty(length, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 23.75 GiB. GPU 4 has a total capacty of 79.33 GiB of which 22.39 GiB is free. Including non-PyTorch memory, this process has 56.93 GiB memory in use. Of the allocated memory 47.50 GiB is allocated by PyTorch, and 488.98 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 95, in __init__
+    big_flat_buffer = torch.empty(length, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 23.75 GiB. GPU 5 has a total capacty of 79.33 GiB of which 22.32 GiB is free. Including non-PyTorch memory, this process has 56.99 GiB memory in use. Of the allocated memory 47.50 GiB is allocated by PyTorch, and 488.98 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 249, in <module>
+    trainer = DistributedTrainer(config_file)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 184, in __init__
+    self.optimizer, self.grad_accumulator = init_optimizer_and_grad_accumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 402, in init_optimizer_and_grad_accumulator
+    optimizer = optimizer_builder(named_param_groups)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 364, in grad_optimizer_builder
+    result = OptimizerFromGradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/optimizer_from_gradient_accumulator.py", line 32, in __init__
+    gradient_accumulator = gradient_accumulator_builder(name_to_param.items())
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/helpers.py", line 365, in <lambda>
+    gradient_accumulator_builder=lambda named_params: FP32GradientAccumulator(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/optim/gradient_accumulator.py", line 95, in __init__
+    big_flat_buffer = torch.empty(length, dtype=torch.float, device="cuda")
+torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 23.75 GiB. GPU 1 has a total capacty of 79.33 GiB of which 22.32 GiB is free. Including non-PyTorch memory, this process has 56.99 GiB memory in use. Of the allocated memory 47.50 GiB is allocated by PyTorch, and 488.98 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting max_split_size_mb to avoid fragmentation.  See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[E TraceUtils.h:35] Store is down while updating #101 with key NCCL_0_trace_start
+[E TraceUtils.h:35] Store is down while updating #104 with key NCCL_0_trace_start
+[E TraceUtils.h:35] Store is down while updating #56 with key NCCL_6_trace_start
+[E TraceUtils.h:35] Store is down while updating #57 with key NCCL_5_trace_start
+[E TraceUtils.h:35] Store is down while updating #56 with key NCCL_4_trace_start
+[E TraceUtils.h:35] Store is down while updating #56 with key NCCL_0_trace_start
+[E TraceUtils.h:35] Store is down while updating #56 with key NCCL_7_trace_start
+[E TraceUtils.h:35] Store is down while updating #56 with key NCCL_3_trace_start
+[E TraceUtils.h:35] Store is down while updating #56 with key NCCL_5_trace_start
+[E TraceUtils.h:35] Store is down while updating #56 with key NCCL_2_trace_start
+[E TraceUtils.h:35] Store is down while updating #56 with key NCCL_4_trace_start
+[E TraceUtils.h:35] Store is down while updating #99 with key NCCL_1_trace_start
+[E TraceUtils.h:35] Store is down while updating #104 with key NCCL_1_trace_start
+[E TraceUtils.h:35] Store is down while updating #107 with key NCCL_1_trace_start
+[E TraceUtils.h:35] Store is down while updating #102 with key NCCL_1_trace_start
+[E TraceUtils.h:35] Store is down while updating #104 with key NCCL_0_trace_end
+[E TraceUtils.h:35] Store is down while updating #116 with key NCCL_1_trace_start
+[E TraceUtils.h:35] Store is down while updating #107 with key NCCL_0_trace_start
+[E TraceUtils.h:35] Store is down while updating #113 with key NCCL_1_trace_start
+[E TraceUtils.h:35] Store is down while updating #114 with key NCCL_0_trace_start
+[E TraceUtils.h:35] Store is down while updating #105 with key NCCL_1_trace_start
+[E TraceUtils.h:35] Store is down while updating #116 with key NCCL_0_trace_start
+01/08/2025 11:14:30 [ERROR|DP=1|PP=7|TP=5|ip-26-0-168-95]: [RECV META] Rank 7: Failed to receive metadata from rank 6: [7] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '6:7', but store->get('6:7') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/08/2025 11:14:30 [WARNING|DP=1|PP=7|TP=5|ip-26-0-168-95]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 6: [7] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '6:7', but store->get('6:7') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/08/2025 11:14:30 [ERROR|DP=1|PP=7|TP=0|ip-26-0-168-95]: [RECV META] Rank 7: Failed to receive metadata from rank 6: [7] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '6:7', but store->get('6:7') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/08/2025 11:14:30 [ERROR|DP=1|PP=7|TP=6|ip-26-0-168-95]: [RECV META] Rank 7: Failed to receive metadata from rank 6: [7] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '6:7', but store->get('6:7') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/08/2025 11:14:30 [WARNING|DP=1|PP=7|TP=5|ip-26-0-168-95]: No progress made in communication iteration, attempt 1/1
+01/08/2025 11:14:30 [ERROR|DP=1|PP=7|TP=4|ip-26-0-168-95]: [RECV META] Rank 7: Failed to receive metadata from rank 6: [7] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '6:7', but store->get('6:7') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/08/2025 11:14:30 [WARNING|DP=1|PP=7|TP=0|ip-26-0-168-95]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 6: [7] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '6:7', but store->get('6:7') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/08/2025 11:14:30 [ERROR|DP=1|PP=7|TP=5|ip-26-0-168-95]: Max retries reached, giving up on communication
+01/08/2025 11:14:30 [WARNING|DP=1|PP=7|TP=6|ip-26-0-168-95]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 6: [7] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '6:7', but store->get('6:7') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/08/2025 11:14:30 [ERROR|DP=1|PP=7|TP=3|ip-26-0-168-95]: [RECV META] Rank 7: Failed to receive metadata from rank 6: [7] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '6:7', but store->get('6:7') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/08/2025 11:14:30 [ERROR|DP=1|PP=7|TP=7|ip-26-0-168-95]: [RECV META] Rank 7: Failed to receive metadata from rank 6: [7] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '6:7', but store->get('6:7') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/08/2025 11:14:30 [WARNING|DP=1|PP=7|TP=0|ip-26-0-168-95]: No progress made in communication iteration, attempt 1/1
+01/08/2025 11:14:30 [WARNING|DP=1|PP=7|TP=6|ip-26-0-168-95]: No progress made in communication iteration, attempt 1/1
+01/08/2025 11:14:30 [ERROR|DP=1|PP=7|TP=0|ip-26-0-168-95]: Max retries reached, giving up on communication
+01/08/2025 11:14:30 [WARNING|DP=1|PP=7|TP=4|ip-26-0-168-95]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 6: [7] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '6:7', but store->get('6:7') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/08/2025 11:14:30 [ERROR|DP=1|PP=7|TP=6|ip-26-0-168-95]: Max retries reached, giving up on communication
+01/08/2025 11:14:30 [WARNING|DP=1|PP=7|TP=3|ip-26-0-168-95]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 6: [7] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '6:7', but store->get('6:7') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/08/2025 11:14:30 [ERROR|DP=1|PP=7|TP=2|ip-26-0-168-95]: [RECV META] Rank 7: Failed to receive metadata from rank 6: [7] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '6:7', but store->get('6:7') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/08/2025 11:14:30 [WARNING|DP=1|PP=7|TP=7|ip-26-0-168-95]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 6: [7] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '6:7', but store->get('6:7') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/08/2025 11:14:30 [WARNING|DP=1|PP=7|TP=4|ip-26-0-168-95]: No progress made in communication iteration, attempt 1/1
+01/08/2025 11:14:30 [WARNING|DP=1|PP=7|TP=3|ip-26-0-168-95]: No progress made in communication iteration, attempt 1/1
+01/08/2025 11:14:30 [WARNING|DP=1|PP=7|TP=7|ip-26-0-168-95]: No progress made in communication iteration, attempt 1/1
+01/08/2025 11:14:30 [ERROR|DP=1|PP=7|TP=4|ip-26-0-168-95]: Max retries reached, giving up on communication
+01/08/2025 11:14:30 [WARNING|DP=1|PP=7|TP=2|ip-26-0-168-95]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 6: [7] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '6:7', but store->get('6:7') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/08/2025 11:14:30 [ERROR|DP=1|PP=7|TP=3|ip-26-0-168-95]: Max retries reached, giving up on communication
+01/08/2025 11:14:30 [ERROR|DP=1|PP=7|TP=7|ip-26-0-168-95]: Max retries reached, giving up on communication
+01/08/2025 11:14:30 [WARNING|DP=1|PP=7|TP=2|ip-26-0-168-95]: No progress made in communication iteration, attempt 1/1
+01/08/2025 11:14:30 [ERROR|DP=0|PP=7|TP=5|ip-26-0-168-52]: [RECV META] Rank 7: Failed to receive metadata from rank 6: [7] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '6:7', but store->get('6:7') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/08/2025 11:14:30 [ERROR|DP=1|PP=7|TP=2|ip-26-0-168-95]: Max retries reached, giving up on communication
+01/08/2025 11:14:30 [ERROR|DP=0|PP=7|TP=7|ip-26-0-168-52]: [RECV META] Rank 7: Failed to receive metadata from rank 6: [7] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '6:7', but store->get('6:7') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/08/2025 11:14:30 [WARNING|DP=0|PP=7|TP=5|ip-26-0-168-52]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 6: [7] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '6:7', but store->get('6:7') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/08/2025 11:14:30 [WARNING|DP=0|PP=7|TP=7|ip-26-0-168-52]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 6: [7] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '6:7', but store->get('6:7') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/08/2025 11:14:30 [ERROR|DP=0|PP=7|TP=3|ip-26-0-168-52]: [RECV META] Rank 7: Failed to receive metadata from rank 6: [7] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '6:7', but store->get('6:7') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/08/2025 11:14:30 [WARNING|DP=0|PP=7|TP=5|ip-26-0-168-52]: No progress made in communication iteration, attempt 1/1
+01/08/2025 11:14:30 [WARNING|DP=0|PP=7|TP=7|ip-26-0-168-52]: No progress made in communication iteration, attempt 1/1
+01/08/2025 11:14:30 [ERROR|DP=0|PP=7|TP=5|ip-26-0-168-52]: Max retries reached, giving up on communication
+01/08/2025 11:14:30 [ERROR|DP=0|PP=7|TP=7|ip-26-0-168-52]: Max retries reached, giving up on communication
+01/08/2025 11:14:30 [WARNING|DP=0|PP=7|TP=3|ip-26-0-168-52]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 6: [7] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '6:7', but store->get('6:7') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/08/2025 11:14:30 [WARNING|DP=0|PP=7|TP=3|ip-26-0-168-52]: No progress made in communication iteration, attempt 1/1
+01/08/2025 11:14:30 [ERROR|DP=0|PP=7|TP=3|ip-26-0-168-52]: Max retries reached, giving up on communication
+01/08/2025 11:14:30 [ERROR|DP=0|PP=7|TP=6|ip-26-0-168-52]: [RECV META] Rank 7: Failed to receive metadata from rank 6: [7] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '6:7', but store->get('6:7') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/08/2025 11:14:30 [ERROR|DP=0|PP=7|TP=1|ip-26-0-168-52]: [RECV META] Rank 7: Failed to receive metadata from rank 6: [7] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '6:7', but store->get('6:7') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/08/2025 11:14:30 [WARNING|DP=0|PP=7|TP=6|ip-26-0-168-52]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 6: [7] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '6:7', but store->get('6:7') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/08/2025 11:14:30 [ERROR|DP=0|PP=7|TP=2|ip-26-0-168-52]: [RECV META] Rank 7: Failed to receive metadata from rank 6: [7] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '6:7', but store->get('6:7') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/08/2025 11:14:30 [WARNING|DP=0|PP=7|TP=6|ip-26-0-168-52]: No progress made in communication iteration, attempt 1/1
+01/08/2025 11:14:30 [ERROR|DP=0|PP=7|TP=4|ip-26-0-168-52]: [RECV META] Rank 7: Failed to receive metadata from rank 6: [7] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '6:7', but store->get('6:7') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/08/2025 11:14:30 [WARNING|DP=0|PP=7|TP=2|ip-26-0-168-52]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 6: [7] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '6:7', but store->get('6:7') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/08/2025 11:14:30 [ERROR|DP=0|PP=7|TP=6|ip-26-0-168-52]: Max retries reached, giving up on communication
+01/08/2025 11:14:30 [WARNING|DP=0|PP=7|TP=1|ip-26-0-168-52]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 6: [7] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '6:7', but store->get('6:7') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/08/2025 11:14:30 [WARNING|DP=0|PP=7|TP=2|ip-26-0-168-52]: No progress made in communication iteration, attempt 1/1
+01/08/2025 11:14:30 [WARNING|DP=0|PP=7|TP=4|ip-26-0-168-52]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 6: [7] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '6:7', but store->get('6:7') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/08/2025 11:14:30 [WARNING|DP=0|PP=7|TP=1|ip-26-0-168-52]: No progress made in communication iteration, attempt 1/1
+01/08/2025 11:14:30 [ERROR|DP=0|PP=7|TP=2|ip-26-0-168-52]: Max retries reached, giving up on communication
+01/08/2025 11:14:30 [ERROR|DP=0|PP=7|TP=1|ip-26-0-168-52]: Max retries reached, giving up on communication
+01/08/2025 11:14:30 [WARNING|DP=0|PP=7|TP=4|ip-26-0-168-52]: No progress made in communication iteration, attempt 1/1
+01/08/2025 11:14:30 [ERROR|DP=0|PP=7|TP=4|ip-26-0-168-52]: Max retries reached, giving up on communication
+01/08/2025 11:14:30 [ERROR|DP=1|PP=7|TP=1|ip-26-0-168-95]: [RECV META] Rank 7: Failed to receive metadata from rank 6: [7] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '6:7', but store->get('6:7') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/08/2025 11:14:30 [WARNING|DP=1|PP=7|TP=1|ip-26-0-168-95]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 6: [7] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '6:7', but store->get('6:7') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/08/2025 11:14:30 [WARNING|DP=1|PP=7|TP=1|ip-26-0-168-95]: No progress made in communication iteration, attempt 1/1
+01/08/2025 11:14:30 [ERROR|DP=1|PP=7|TP=1|ip-26-0-168-95]: Max retries reached, giving up on communication
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 460, in train
+Traceback (most recent call last):
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    return f(*args, **kwargs)
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 460, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 497, in training_step
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 460, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 497, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 320, in train_batch_iter
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 460, in train
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 320, in train_batch_iter
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 497, in training_step
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+        trainer.train(dataloader)trainer.train(dataloader)
+
+    trainer.train(dataloader)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    
+trainer.train(dataloader)    
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 460, in train
+    trainer.train(dataloader)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 497, in training_step
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 320, in train_batch_iter
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 497, in training_step
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 460, in train
+    output = model(**micro_batch)
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 497, in training_step
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 460, in train
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 320, in train_batch_iter
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 1519, in forward
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 320, in train_batch_iter
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 460, in train
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return f(*args, **kwargs)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 460, in train
+return f(*args, **kwargs)    
+return f(*args, **kwargs)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 460, in train
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 460, in train
+return f(*args, **kwargs)    
+return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 460, in train
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 460, in train
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 460, in train
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 320, in train_batch_iter
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    output = model(**micro_batch)    
+output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    return f(*args, **kwargs)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 1519, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 460, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 497, in training_step
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)    
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 497, in training_step
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    else self._run_ddp_forward(*inputs, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 1355, in _run_ddp_forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 320, in train_batch_iter
+    return self._call_impl(*args, **kwargs)
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 497, in training_step
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 1519, in forward
+    else self._run_ddp_forward(*inputs, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 1355, in _run_ddp_forward
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 320, in train_batch_iter
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self.module(*inputs, **kwargs)  # type: ignore[index]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    else self._run_ddp_forward(*inputs, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 1355, in _run_ddp_forward
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 1519, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 497, in training_step
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 497, in training_step
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)    
+outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 497, in training_step
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 497, in training_step
+    output = model(**micro_batch)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+return self.module(*inputs, **kwargs)  # type: ignore[index]
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 1519, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 497, in training_step
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 497, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 320, in train_batch_iter
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 320, in train_batch_iter
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 320, in train_batch_iter
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 320, in train_batch_iter
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 320, in train_batch_iter
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 320, in train_batch_iter
+    return self.module(*inputs, **kwargs)  # type: ignore[index]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 320, in train_batch_iter
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    else self._run_ddp_forward(*inputs, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 1355, in _run_ddp_forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    else self._run_ddp_forward(*inputs, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 1355, in _run_ddp_forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 1519, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+        output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+
+        return self._call_impl(*args, **kwargs)
+return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 1519, in forward
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self.module(*inputs, **kwargs)  # type: ignore[index]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = model(**micro_batch)    
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1012, in forward
+output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    output = model(**micro_batch)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1012, in forward
+    return self.module(*inputs, **kwargs)  # type: ignore[index]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 1519, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1012, in forward
+    else self._run_ddp_forward(*inputs, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 1355, in _run_ddp_forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    else self._run_ddp_forward(*inputs, **kwargs)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 1355, in _run_ddp_forward
+return self.module(*inputs, **kwargs)  # type: ignore[index]
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    else self._run_ddp_forward(*inputs, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 1355, in _run_ddp_forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self.module(*inputs, **kwargs)  # type: ignore[index]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1012, in forward
+    sharded_logits = self.model(
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+        return self._call_impl(*args, **kwargs)return self._call_impl(*args, **kwargs)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self.module(*inputs, **kwargs)  # type: ignore[index]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1012, in forward
+    return forward_call(*args, **kwargs)
+        sharded_logits = self.model(return forward_call(*args, **kwargs)
+
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1012, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 1519, in forward
+        return forward_call(*args, **kwargs)return forward_call(*args, **kwargs)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 1519, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 1519, in forward
+return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1012, in forward
+    return self._call_impl(*args, **kwargs)    
+return self._call_impl(*args, **kwargs)  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 891, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 891, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 1519, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 1519, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 1519, in forward
+        return forward_call(*args, **kwargs)return self._call_impl(*args, **kwargs)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 891, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 1519, in forward
+    else self._run_ddp_forward(*inputs, **kwargs)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1012, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 1355, in _run_ddp_forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 891, in forward
+    else self._run_ddp_forward(*inputs, **kwargs)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 1355, in _run_ddp_forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+        else self._run_ddp_forward(*inputs, **kwargs)else self._run_ddp_forward(*inputs, **kwargs)
+
+    sharded_logits = self.model(
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 1355, in _run_ddp_forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 1355, in _run_ddp_forward
+else self._run_ddp_forward(*inputs, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 1355, in _run_ddp_forward
+    else self._run_ddp_forward(*inputs, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 1355, in _run_ddp_forward
+    sharded_logits = self.model(  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return self.module(*inputs, **kwargs)  # type: ignore[index]
+        return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self.module(*inputs, **kwargs)  # type: ignore[index]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    else self._run_ddp_forward(*inputs, **kwargs)
+    hidden_states = encoder_block(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 1355, in _run_ddp_forward
+    hidden_states = encoder_block(hidden_states=hidden_states)["hidden_states"]
+        return self.module(*inputs, **kwargs)  # type: ignore[index]return self.module(*inputs, **kwargs)  # type: ignore[index]
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self.module(*inputs, **kwargs)  # type: ignore[index]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 891, in forward
+        hidden_states = encoder_block(hidden_states=hidden_states)["hidden_states"]hidden_states = encoder_block(hidden_states=hidden_states)["hidden_states"]
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 891, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+        return self._call_impl(*args, **kwargs)return self._call_impl(*args, **kwargs)
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self.module(*inputs, **kwargs)  # type: ignore[index]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1012, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1012, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1012, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1012, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1012, in forward
+    return self._call_impl(*args, **kwargs)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self.module(*inputs, **kwargs)  # type: ignore[index]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+    return forward_call(*args, **kwargs)    
+return forward_call(*args, **kwargs)  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 891, in forward
+    sharded_logits = self.model(
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+hidden_states = encoder_block(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+    hidden_states = encoder_block(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 891, in forward
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+    sharded_logits = self.model(
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    hidden_states = encoder_block(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+    return self._call_impl(*args, **kwargs)
+    activation = pipeline_state.activations_buffer.popleft()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+IndexError: pop from an empty deque
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1012, in forward
+    hidden_states = encoder_block(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 891, in forward
+return forward_call(*args, **kwargs)
+            return forward_call(*args, **kwargs)  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 891, in forward
+return forward_call(*args, **kwargs)return self._call_impl(*args, **kwargs)
+
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1012, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 891, in forward
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 891, in forward
+    activation = pipeline_state.activations_buffer.popleft()
+IndexError: pop from an empty deque
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+    activation = pipeline_state.activations_buffer.popleft()
+IndexError: pop from an empty deque
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+    activation = pipeline_state.activations_buffer.popleft()
+IndexError: pop from an empty deque
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 891, in forward
+    activation = pipeline_state.activations_buffer.popleft()
+IndexError: pop from an empty deque
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+    activation = pipeline_state.activations_buffer.popleft()
+IndexError: pop from an empty deque
+    activation = pipeline_state.activations_buffer.popleft()
+IndexError: pop from an empty deque
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+    sharded_logits = self.model(  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+        return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    hidden_states = encoder_block(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states = encoder_block(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    hidden_states = encoder_block(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 891, in forward
+    hidden_states = encoder_block(hidden_states=hidden_states)["hidden_states"]
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+hidden_states = encoder_block(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+      File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    activation = pipeline_state.activations_buffer.popleft()
+    return self._call_impl(*args, **kwargs)
+IndexError: pop from an empty deque
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+    return forward_call(*args, **kwargs)
+    return forward_call(*args, **kwargs)  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 891, in forward
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+    hidden_states = encoder_block(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+      File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+new_kwargs[name] = recv_from_pipeline_state_buffer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+    activation = pipeline_state.activations_buffer.popleft()
+    activation = pipeline_state.activations_buffer.popleft()
+    IndexErroractivation = pipeline_state.activations_buffer.popleft()
+: IndexErrorpop from an empty deque: 
+IndexErrorpop from an empty deque    
+: pop from an empty dequereturn forward_call(*args, **kwargs)
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+        new_kwargs[name] = recv_from_pipeline_state_buffer(hidden_states = encoder_block(hidden_states=hidden_states)["hidden_states"]
+
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    activation = pipeline_state.activations_buffer.popleft()
+IndexError: pop from an empty deque
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+    activation = pipeline_state.activations_buffer.popleft()    
+return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+IndexError: pop from an empty deque
+    activation = pipeline_state.activations_buffer.popleft()
+IndexError: pop from an empty deque
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+    activation = pipeline_state.activations_buffer.popleft()
+IndexError: pop from an empty deque
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+01/08/2025 11:14:31 [ERROR|DP=0|PP=7|TP=0|ip-26-0-168-52]: [RECV META] Rank 7: Failed to receive metadata from rank 6: [7] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '6:7', but store->get('6:7') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/08/2025 11:14:31 [WARNING|DP=0|PP=7|TP=0|ip-26-0-168-52]: recv_activation failed, will retry later. Error: Failed to receive metadata from rank 6: [7] is setting up NCCL communicator and retrieving ncclUniqueId from [0] via c10d key-value store by key '6:7', but store->get('6:7') got error: Connection reset by peer. This may indicate a possible application crash on rank 0 or a network set up issue.
+01/08/2025 11:14:31 [WARNING|DP=0|PP=7|TP=0|ip-26-0-168-52]: No progress made in communication iteration, attempt 1/1
+01/08/2025 11:14:31 [ERROR|DP=0|PP=7|TP=0|ip-26-0-168-52]: Max retries reached, giving up on communication
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 253, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 460, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 497, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 320, in train_batch_iter
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 52, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 1519, in forward
+    else self._run_ddp_forward(*inputs, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/parallel/distributed.py", line 1355, in _run_ddp_forward
+    return self.module(*inputs, **kwargs)  # type: ignore[index]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1012, in forward
+    sharded_logits = self.model(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 891, in forward
+    return self.forward_with_hidden_states(input_ids=input_ids, input_mask=input_mask)[0]
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 903, in forward_with_hidden_states
+    hidden_states = encoder_block(hidden_states=hidden_states)["hidden_states"]
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 135, in forward
+    new_kwargs[name] = recv_from_pipeline_state_buffer(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/functional.py", line 122, in recv_from_pipeline_state_buffer
+    activation = pipeline_state.activations_buffer.popleft()
+IndexError: pop from an empty deque
+[2025-01-08 11:14:47,609] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 94596 closing signal SIGTERM
+[2025-01-08 11:14:47,609] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 94597 closing signal SIGTERM
+[2025-01-08 11:14:47,609] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 94599 closing signal SIGTERM
+[2025-01-08 11:14:47,609] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 94600 closing signal SIGTERM
+[2025-01-08 11:14:47,609] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 94601 closing signal SIGTERM
+[2025-01-08 11:14:47,609] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 94602 closing signal SIGTERM
+[2025-01-08 11:14:47,609] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 94603 closing signal SIGTERM
+[2025-01-08 11:14:50,791] torch.distributed.elastic.multiprocessing.api: [ERROR] failed (exitcode: 1) local_rank: 2 (pid: 94598) of binary: /fsx/nouamane/miniconda/envs/2-1-cu121/bin/python
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 264, in launch_agent
+    raise ChildFailedError(
+torch.distributed.elastic.multiprocessing.errors.ChildFailedError: 
+============================================================
+/fsx/nouamane/projects/nanotron/run_train.py FAILED
+------------------------------------------------------------
+Failures:
+  <NO_OTHER_FAILURES>
+------------------------------------------------------------
+Root Cause (first observed failure):
+[0]:
+  time      : 2025-01-08_11:14:47
+  host      : ip-26-0-162-180.ec2.internal
+  rank      : 2 (local_rank: 2)
+  exitcode  : 1 (pid: 94598)
+  error_file: <N/A>
+  traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
+============================================================
+srun: error: ip-26-0-162-180: task 2: Exited with exit code 1
+srun: launch/slurm: _step_signal: Terminating StepId=14280977.0
+slurmstepd: error: *** STEP 14280977.0 ON ip-26-0-162-46 CANCELLED AT 2025-01-08T11:14:51 ***
+[2025-01-08 11:14:51,325] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-08 11:14:51,326] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 332842 closing signal SIGTERM
+[2025-01-08 11:14:51,326] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-08 11:14:51,326] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 180403 closing signal SIGTERM
+[2025-01-08 11:14:51,326] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 332843 closing signal SIGTERM
+[2025-01-08 11:14:51,326] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 180404 closing signal SIGTERM
+[2025-01-08 11:14:51,326] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 332844 closing signal SIGTERM
+[2025-01-08 11:14:51,326] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 180405 closing signal SIGTERM
+[2025-01-08 11:14:51,326] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 180406 closing signal SIGTERM
+[2025-01-08 11:14:51,326] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 332845 closing signal SIGTERM
+[2025-01-08 11:14:51,326] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 180407 closing signal SIGTERM
+[2025-01-08 11:14:51,326] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 332846 closing signal SIGTERM
+[2025-01-08 11:14:51,326] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 332847 closing signal SIGTERM
+[2025-01-08 11:14:51,326] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 180409 closing signal SIGTERM
+[2025-01-08 11:14:51,325] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-08 11:14:51,326] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 332848 closing signal SIGTERM
+[2025-01-08 11:14:51,326] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 180410 closing signal SIGTERM
+[2025-01-08 11:14:51,326] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-08 11:14:51,325] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 405963 closing signal SIGTERM
+[2025-01-08 11:14:51,326] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 405964 closing signal SIGTERM
+[2025-01-08 11:14:51,326] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 237250 closing signal SIGTERM
+[2025-01-08 11:14:51,326] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-08 11:14:51,326] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 405965 closing signal SIGTERM
+[2025-01-08 11:14:51,326] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-08 11:14:51,326] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 405966 closing signal SIGTERM
+[2025-01-08 11:14:51,326] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 237251 closing signal SIGTERM
+[2025-01-08 11:14:51,326] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 405967 closing signal SIGTERM
+[2025-01-08 11:14:51,326] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 291973 closing signal SIGTERM
+[2025-01-08 11:14:51,326] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 67730 closing signal SIGTERM
+[2025-01-08 11:14:51,327] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 237252 closing signal SIGTERM
+[2025-01-08 11:14:51,326] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-08 11:14:51,326] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 405969 closing signal SIGTERM
+[2025-01-08 11:14:51,326] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 291974 closing signal SIGTERM
+[2025-01-08 11:14:51,326] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 291975 closing signal SIGTERM
+[2025-01-08 11:14:51,326] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-08 11:14:51,327] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 67731 closing signal SIGTERM
+[2025-01-08 11:14:51,327] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 237254 closing signal SIGTERM
+[2025-01-08 11:14:51,326] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 113292 closing signal SIGTERM
+[2025-01-08 11:14:51,326] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 291977 closing signal SIGTERM
+[2025-01-08 11:14:51,326] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-08 11:14:51,327] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 67732 closing signal SIGTERM
+[2025-01-08 11:14:51,326] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 495446 closing signal SIGTERM
+[2025-01-08 11:14:51,327] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 237255 closing signal SIGTERM
+[2025-01-08 11:14:51,326] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 291978 closing signal SIGTERM
+[2025-01-08 11:14:51,326] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 113293 closing signal SIGTERM
+[2025-01-08 11:14:51,327] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 67733 closing signal SIGTERM
+[2025-01-08 11:14:51,326] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 495448 closing signal SIGTERM
+[2025-01-08 11:14:51,326] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 479062 closing signal SIGTERM
+[2025-01-08 11:14:51,327] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 237256 closing signal SIGTERM
+[2025-01-08 11:14:51,326] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 113294 closing signal SIGTERM
+[2025-01-08 11:14:51,326] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 291979 closing signal SIGTERM
+[2025-01-08 11:14:51,326] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 291980 closing signal SIGTERM
+[2025-01-08 11:14:51,327] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 67734 closing signal SIGTERM
+[2025-01-08 11:14:51,327] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 67735 closing signal SIGTERM
+[2025-01-08 11:14:51,327] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-08 11:14:51,326] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 495449 closing signal SIGTERM
+[2025-01-08 11:14:51,326] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 113295 closing signal SIGTERM
+[2025-01-08 11:14:51,326] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 479063 closing signal SIGTERM
+[2025-01-08 11:14:51,327] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 67736 closing signal SIGTERM
+[2025-01-08 11:14:51,326] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 495450 closing signal SIGTERM
+[2025-01-08 11:14:51,327] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 495451 closing signal SIGTERM
+[2025-01-08 11:14:51,327] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-08 11:14:51,326] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 113296 closing signal SIGTERM
+[2025-01-08 11:14:51,326] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 479064 closing signal SIGTERM
+[2025-01-08 11:14:51,327] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 495452 closing signal SIGTERM
+[2025-01-08 11:14:51,327] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-08 11:14:51,327] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 207764 closing signal SIGTERM
+[2025-01-08 11:14:51,326] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 113297 closing signal SIGTERM
+[2025-01-08 11:14:51,327] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-08 11:14:51,326] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 479065 closing signal SIGTERM
+[2025-01-08 11:14:51,327] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-08 11:14:51,327] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 63392 closing signal SIGTERM
+[2025-01-08 11:14:51,326] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 479067 closing signal SIGTERM
+[2025-01-08 11:14:51,327] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 201324 closing signal SIGTERM
+[2025-01-08 11:14:51,327] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 238873 closing signal SIGTERM
+[2025-01-08 11:14:51,327] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 238874 closing signal SIGTERM
+[2025-01-08 11:14:51,327] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 63394 closing signal SIGTERM
+[2025-01-08 11:14:51,327] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 207765 closing signal SIGTERM
+[2025-01-08 11:14:51,327] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 479068 closing signal SIGTERM
+[2025-01-08 11:14:51,327] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 238875 closing signal SIGTERM
+[2025-01-08 11:14:51,327] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 41391 closing signal SIGTERM
+[2025-01-08 11:14:51,327] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 238876 closing signal SIGTERM
+[2025-01-08 11:14:51,327] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 63395 closing signal SIGTERM
+[2025-01-08 11:14:51,327] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 207766 closing signal SIGTERM
+[2025-01-08 11:14:51,327] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 201325 closing signal SIGTERM
+[2025-01-08 11:14:51,327] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 238877 closing signal SIGTERM
+[2025-01-08 11:14:51,327] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 41392 closing signal SIGTERM
+[2025-01-08 11:14:51,327] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 63396 closing signal SIGTERM
+[2025-01-08 11:14:51,327] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 63397 closing signal SIGTERM
+[2025-01-08 11:14:51,327] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 238879 closing signal SIGTERM
+[2025-01-08 11:14:51,327] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 207767 closing signal SIGTERM
+[2025-01-08 11:14:51,327] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 41393 closing signal SIGTERM
+[2025-01-08 11:14:51,327] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 63398 closing signal SIGTERM
+[2025-01-08 11:14:51,327] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 201326 closing signal SIGTERM
+[2025-01-08 11:14:51,327] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 238880 closing signal SIGTERM
+[2025-01-08 11:14:51,327] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 41394 closing signal SIGTERM
+[2025-01-08 11:14:51,327] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 201327 closing signal SIGTERM
+[2025-01-08 11:14:51,328] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 41395 closing signal SIGTERM
+[2025-01-08 11:14:51,328] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 41396 closing signal SIGTERM
+[2025-01-08 11:14:51,327] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 207769 closing signal SIGTERM
+[2025-01-08 11:14:51,327] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 201328 closing signal SIGTERM
+[2025-01-08 11:14:51,327] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-01-08 11:14:51,328] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 41397 closing signal SIGTERM
+[2025-01-08 11:14:51,328] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 207770 closing signal SIGTERM
+[2025-01-08 11:14:51,328] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 41398 closing signal SIGTERM
+[2025-01-08 11:14:51,327] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 201329 closing signal SIGTERM
+[2025-01-08 11:14:51,327] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 317669 closing signal SIGTERM
+[2025-01-08 11:14:51,327] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 317670 closing signal SIGTERM
+[2025-01-08 11:14:51,327] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 201330 closing signal SIGTERM
+[2025-01-08 11:14:51,327] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 317671 closing signal SIGTERM
+[2025-01-08 11:14:51,327] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 317672 closing signal SIGTERM
+[2025-01-08 11:14:51,327] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 317673 closing signal SIGTERM
+[2025-01-08 11:14:51,327] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 317674 closing signal SIGTERM
+[2025-01-08 11:14:51,327] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 317675 closing signal SIGTERM
+[2025-01-08 11:14:51,327] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 317676 closing signal SIGTERM
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 495372 got signal: 15
+srun: error: ip-26-0-167-245: task 9: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 237176 got signal: 15
+srun: error: ip-26-0-165-131: task 7: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 478990 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 405890 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 180330 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 207692 got signal: 15
+srun: error: ip-26-0-162-233: task 3: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 67657 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 291902 got signal: 15
+srun: error: ip-26-0-165-24: task 4: Exited with exit code 1
+srun: error: ip-26-0-165-59: task 6: Exited with exit code 1
+srun: error: ip-26-0-165-38: task 5: Exited with exit code 1
+srun: error: ip-26-0-162-79: task 1: Exited with exit code 1
+srun: error: ip-26-0-168-30: task 10: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 238802 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 63318 got signal: 15
+srun: error: ip-26-0-168-238: task 15: Exited with exit code 1
+srun: error: ip-26-0-167-217: task 8: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 332769 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 201251 got signal: 15
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 317597 got signal: 15
+srun: error: ip-26-0-162-46: task 0: Exited with exit code 1
+srun: error: ip-26-0-168-120: task 14: Exited with exit code 1
+[2025-01-08 11:14:56,617] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-168-95.ec2.internal_113219_0' has failed to send a keep-alive heartbeat to the rendezvous '14280977' due to an error of type RendezvousConnectionError.
+srun: error: ip-26-0-168-34: task 11: Exited with exit code 1
+[2025-01-08 11:14:57,571] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-168-52.ec2.internal_41319_0' has failed to send a keep-alive heartbeat to the rendezvous '14280977' due to an error of type RendezvousConnectionError.
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 113219 got signal: 15
+srun: error: ip-26-0-168-95: task 13: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 41319 got signal: 15
+srun: error: ip-26-0-168-52: task 12: Exited with exit code 1
diff --git a/logs/14599514-bench_3.57G_dp1_tp1_pp32_acc31_mbs3_seq4096_zero0_tpmodeRED_vocab131k_cache.out b/logs/14599514-bench_3.57G_dp1_tp1_pp32_acc31_mbs3_seq4096_zero0_tpmodeRED_vocab131k_cache.out
new file mode 100644
index 0000000000000000000000000000000000000000..40fbb927177bb247d728ae385bdc92f62e8d0b54
--- /dev/null
+++ b/logs/14599514-bench_3.57G_dp1_tp1_pp32_acc31_mbs3_seq4096_zero0_tpmodeRED_vocab131k_cache.out
@@ -0,0 +1,1058 @@
++ echo 'Running script: /var/spool/slurmd/job14599514/slurm_script'
+Running script: /var/spool/slurmd/job14599514/slurm_script
++ '[' -z 14599514 ']'
++ source /etc/profile.d/modules.sh
+++ . /usr/share/modules/init/bash
++++ unset _mlshdbg
++++ '[' 0 = 1 ']'
++++ unset _mlre _mlIFS
++++ '[' -n x ']'
++++ _mlIFS=' 	
+'
++++ IFS=' '
++++ '[' -n '' ']'
+++++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash autoinit
++++ _mlcode='module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+MODULEPATH_modshare=/etc/environment-modules/modules:1:/usr/share/modules/\$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1; export MODULEPATH_modshare;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+LOADEDMODULES=; export LOADEDMODULES;
+MODULEPATH=/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/\$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles; export MODULEPATH;
+test 0;'
++++ _mlret=0
++++ '[' -n x ']'
++++ IFS=' 	
+'
++++ unset _mlIFS
++++ unset _mlre _mlv _mlrv
++++ '[' 0 -eq 0 ']'
++++ eval 'module() {
+   unset _mlshdbg;
+   if [ "${MODULES_SILENT_SHELL_DEBUG:-0}" = '\''1'\'' ]; then
+      case "$-" in
+         *v*x*) set +vx; _mlshdbg='\''vx'\'' ;;
+         *v*) set +v; _mlshdbg='\''v'\'' ;;
+         *x*) set +x; _mlshdbg='\''x'\'' ;;
+         *) _mlshdbg='\'''\'' ;;
+      esac;
+   fi;
+   unset _mlre _mlIFS;
+   if [ -n "${IFS+x}" ]; then
+      _mlIFS=$IFS;
+   fi;
+   IFS='\'' '\'';
+   for _mlv in ${MODULES_RUN_QUARANTINE:-}; do
+      if [ "${_mlv}" = "${_mlv##*[!A-Za-z0-9_]}" -a "${_mlv}" = "${_mlv#[0-9]}" ]; then
+         if [ -n "`eval '\''echo ${'\''$_mlv'\''+x}'\''`" ]; then
+            _mlre="${_mlre:-}${_mlv}_modquar='\''`eval '\''echo ${'\''$_mlv'\''}'\''`'\'' ";
+         fi;
+         _mlrv="MODULES_RUNENV_${_mlv}";
+         _mlre="${_mlre:-}${_mlv}='\''`eval '\''echo ${'\''$_mlrv'\'':-}'\''`'\'' ";
+      fi;
+   done;
+   if [ -n "${_mlre:-}" ]; then
+      eval `eval ${_mlre}/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash '\''"$@"'\''`;
+   else
+      eval `/usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash "$@"`;
+   fi;
+   _mlstatus=$?;
+   if [ -n "${_mlIFS+x}" ]; then
+      IFS=$_mlIFS;
+   else
+      unset IFS;
+   fi;
+   unset _mlre _mlv _mlrv _mlIFS;
+   if [ -n "${_mlshdbg:-}" ]; then
+      set -$_mlshdbg;
+   fi;
+   unset _mlshdbg;
+   return $_mlstatus;
+};
+MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl; export MODULES_CMD;
+ENV=/usr/share/modules/init/profile.sh; export ENV;
+MODULEPATH_modshare=/etc/environment-modules/modules:1:/usr/share/modules/\$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1; export MODULEPATH_modshare;
+BASH_ENV=/usr/share/modules/init/bash; export BASH_ENV;
+MODULESHOME=/usr/share/modules; export MODULESHOME;
+LOADEDMODULES=; export LOADEDMODULES;
+MODULEPATH=/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/\$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles; export MODULEPATH;
+test 0;'
+++++ MODULES_CMD=/usr/lib/x86_64-linux-gnu/modulecmd.tcl
+++++ export MODULES_CMD
+++++ ENV=/usr/share/modules/init/profile.sh
+++++ export ENV
+++++ MODULEPATH_modshare='/etc/environment-modules/modules:1:/usr/share/modules/$MODULE_VERSION/modulefiles:1:/usr/share/modules/modulefiles:1:/usr/share/modules/versions:1'
+++++ export MODULEPATH_modshare
+++++ BASH_ENV=/usr/share/modules/init/bash
+++++ export BASH_ENV
+++++ MODULESHOME=/usr/share/modules
+++++ export MODULESHOME
+++++ LOADEDMODULES=
+++++ export LOADEDMODULES
+++++ MODULEPATH='/etc/environment-modules/modules:/usr/share/modules/versions:/usr/share/modules/$MODULE_VERSION/modulefiles:/usr/share/modules/modulefiles'
+++++ export MODULEPATH
+++++ test 0
++++ '[' 0 = 1 ']'
++++ '[' -t 2 ']'
++++ export -f module
++++ export -f switchml
++++ '[' 5 -ge 3 ']'
++++ [[ ehxB =~ i ]]
++++ [[ ! :/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/f5f18731406b73244e0558ee7716d77c8096d150/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin: =~ :/usr/bin: ]]
++++ '[' '!' -n '' ']'
++++ MANPATH=:
++++ export MANPATH
+++++ manpath
++++ [[ ! :/fsx/nouamane/miniconda/envs/2-1-cu121/man:/fsx/nouamane/miniconda/envs/2-1-cu121/share/man:/fsx/nouamane/miniconda/man:/fsx/nouamane/miniconda/share/man:/usr/local/man:/usr/local/share/man:/usr/share/man:/opt/slurm/share/man:/admin/home/nouamane/.local/share/man:/admin/home/nouamane/.fzf/man:: =~ :/usr/share/man: ]]
++++ unset _mlcode _mlret
++++ '[' -n '' ']'
++ module load cuda/12.1
++ unset _mlshdbg
++ '[' 0 = 1 ']'
++ unset _mlre _mlIFS
++ '[' -n x ']'
++ _mlIFS=' 	
+'
++ IFS=' '
++ '[' -n '' ']'
+++ /usr/bin/tclsh8.6 /usr/lib/x86_64-linux-gnu/modulecmd.tcl bash load cuda/12.1
++ eval 'CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include;' export 'CPATH;
+LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:;' export 'LD_LIBRARY_PATH;
+FI_EFA_FORK_SAFE=1;' export 'FI_EFA_FORK_SAFE;
+MANPATH=/usr/local/cuda-12.1/share/man::;' export 'MANPATH;
+LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64;' export 'LIBRARY_PATH;
+_LMFILES_=/usr/share/modules/modulefiles/cuda/12.1;' export '_LMFILES_;
+LOADEDMODULES=cuda/12.1;' export 'LOADEDMODULES;
+MPI_PATH=/opt/amazon/openmpi;' export 'MPI_PATH;
+NCCL_HOME_modshare=/opt/nccl/build:1;' export 'NCCL_HOME_modshare;
+NCCL_PROTO=simple;' export 'NCCL_PROTO;
+MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1;' export 'MANPATH_modshare;
+LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1;' export 'LIBRARY_PATH_modshare;
+NCCL_SOCKET_IFNAME=enp;' export 'NCCL_SOCKET_IFNAME;
+AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl;' export 'AWS_OFI_NCCL_HOME;
+NCCL_HOME=/opt/nccl/build;' export 'NCCL_HOME;
+FI_PROVIDER=efa;' export 'FI_PROVIDER;
+AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1;' export 'AWS_OFI_NCCL_HOME_modshare;
+CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1;' export 'CPATH_modshare;
+LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2::1:/opt/nccl/build/lib:1:/opt/aws-ofi-nccl/lib:2:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/efa/lib:1;' export 'LD_LIBRARY_PATH_modshare;
+FI_EFA_ENABLE_SHM_TRANSFER=1;' export 'FI_EFA_ENABLE_SHM_TRANSFER;
+_LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1;' export '_LMFILES__modshare;
+LOADEDMODULES_modshare=cuda/12.1:1;' export 'LOADEDMODULES_modshare;
+MPI_PATH_modshare=/opt/amazon/openmpi:1;' export 'MPI_PATH_modshare;
+PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/f5f18731406b73244e0558ee7716d77c8096d150/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin;' export 'PATH;
+CUDA_HOME=/usr/local/cuda-12.1;' export 'CUDA_HOME;
+PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/admin/home/nouamane/.cursor-server/bin/f5f18731406b73244e0558ee7716d77c8096d150/bin/remote-cli:1:/sbin:1:/opt/slurm/bin:1:/usr/sbin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1;' export 'PATH_modshare;
+test' '0;'
+++ CPATH=/opt/nccl/build/include:/usr/local/cuda-12.1/include
+++ export CPATH
+++ LD_LIBRARY_PATH=/opt/nccl/build/lib:/opt/aws-ofi-nccl/lib:/opt/amazon/efa/lib:/opt/amazon/openmpi/lib:/usr/local/cuda-12.1/efa/lib:/usr/local/cuda-12.1/lib:/usr/local/cuda-12.1/lib64:/usr/local/cuda-12.1:/usr/local/cuda-12.1/extras/CUPTI/lib64:/usr/local/cuda-12.1/targets/x86_64-linux/lib:
+++ export LD_LIBRARY_PATH
+++ FI_EFA_FORK_SAFE=1
+++ export FI_EFA_FORK_SAFE
+++ MANPATH=/usr/local/cuda-12.1/share/man::
+++ export MANPATH
+++ LIBRARY_PATH=/opt/aws-ofi-nccl/lib:/opt/nccl/build/lib:/usr/local/cuda-12.1/lib64
+++ export LIBRARY_PATH
+++ _LMFILES_=/usr/share/modules/modulefiles/cuda/12.1
+++ export _LMFILES_
+++ LOADEDMODULES=cuda/12.1
+++ export LOADEDMODULES
+++ MPI_PATH=/opt/amazon/openmpi
+++ export MPI_PATH
+++ NCCL_HOME_modshare=/opt/nccl/build:1
+++ export NCCL_HOME_modshare
+++ NCCL_PROTO=simple
+++ export NCCL_PROTO
+++ MANPATH_modshare=:1:/usr/local/cuda-12.1/share/man:1
+++ export MANPATH_modshare
+++ LIBRARY_PATH_modshare=/opt/aws-ofi-nccl/lib:1:/opt/nccl/build/lib:1:/usr/local/cuda-12.1/lib64:1
+++ export LIBRARY_PATH_modshare
+++ NCCL_SOCKET_IFNAME=enp
+++ export NCCL_SOCKET_IFNAME
+++ AWS_OFI_NCCL_HOME=/opt/aws-ofi-nccl
+++ export AWS_OFI_NCCL_HOME
+++ NCCL_HOME=/opt/nccl/build
+++ export NCCL_HOME
+++ FI_PROVIDER=efa
+++ export FI_PROVIDER
+++ AWS_OFI_NCCL_HOME_modshare=/opt/aws-ofi-nccl:1
+++ export AWS_OFI_NCCL_HOME_modshare
+++ CPATH_modshare=/usr/local/cuda-12.1/include:1:/opt/nccl/build/include:1
+++ export CPATH_modshare
+++ LD_LIBRARY_PATH_modshare=/opt/amazon/efa/lib:3:/opt/amazon/openmpi/lib:2:/usr/local/cuda-12.1/lib:1:/usr/local/cuda-12.1/extras/CUPTI/lib64:2:/usr/local/cuda-12.1/targets/x86_64-linux/lib:2::1:/opt/nccl/build/lib:1:/opt/aws-ofi-nccl/lib:2:/usr/local/cuda-12.1/lib64:2:/usr/local/cuda-12.1:2:/usr/local/cuda-12.1/efa/lib:1
+++ export LD_LIBRARY_PATH_modshare
+++ FI_EFA_ENABLE_SHM_TRANSFER=1
+++ export FI_EFA_ENABLE_SHM_TRANSFER
+++ _LMFILES__modshare=/usr/share/modules/modulefiles/cuda/12.1:1
+++ export _LMFILES__modshare
+++ LOADEDMODULES_modshare=cuda/12.1:1
+++ export LOADEDMODULES_modshare
+++ MPI_PATH_modshare=/opt/amazon/openmpi:1
+++ export MPI_PATH_modshare
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/f5f18731406b73244e0558ee7716d77c8096d150/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
+++ export PATH
+++ CUDA_HOME=/usr/local/cuda-12.1
+++ export CUDA_HOME
+++ PATH_modshare=/usr/local/cuda-12.1/efa/test-cuda-12.1:1:/usr/bin:1:/admin/home/nouamane/.local/bin:1:/usr/local/bin:1:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:1:/bin:1:/snap/bin:1:/fsx/nouamane/miniconda/condabin:1:/admin/home/nouamane/.cursor-server/bin/f5f18731406b73244e0558ee7716d77c8096d150/bin/remote-cli:1:/sbin:1:/opt/slurm/bin:1:/usr/sbin:1:/usr/games:1:/usr/local/sbin:1:/usr/local/cuda-12.1/bin:1:/admin/home/nouamane/.fzf/bin:1:/usr/local/games:1
+++ export PATH_modshare
+++ test 0
++ _mlstatus=0
++ '[' -n x ']'
++ IFS=' 	
+'
++ unset _mlre _mlv _mlrv _mlIFS
++ '[' -n '' ']'
++ unset _mlshdbg
++ return 0
++ source /fsx/nouamane/miniconda/bin/activate
+++ _CONDA_ROOT=/fsx/nouamane/miniconda
+++ . /fsx/nouamane/miniconda/etc/profile.d/conda.sh
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ '[' -z x ']'
+++ conda activate
+++ local cmd=activate
+++ case "$cmd" in
+++ __conda_activate activate
+++ '[' -n '' ']'
+++ local ask_conda
++++ PS1=
++++ __conda_exe shell.posix activate
++++ /fsx/nouamane/miniconda/bin/conda shell.posix activate
+Error while loading conda entry point: conda-libmamba-solver (module 'libmambapy' has no attribute 'QueryFormat')
+++ ask_conda='. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/f5f18731406b73244e0558ee7716d77c8096d150/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''2'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_1='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/etc/conda/activate.d/libxml2_activate.sh"'
+++ eval '. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(base) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/f5f18731406b73244e0558ee7716d77c8096d150/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda'\''
+export CONDA_SHLVL='\''2'\''
+export CONDA_DEFAULT_ENV='\''base'\''
+export CONDA_PROMPT_MODIFIER='\''(base) '\''
+export CONDA_PREFIX_1='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/etc/conda/activate.d/libxml2_activate.sh"'
++++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/deactivate.d/libxml2_deactivate.sh
+++++ test -n ''
+++++ unset XML_CATALOG_FILES
+++++ unset xml_catalog_files_libxml2
++++ PS1='(base) '
++++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/f5f18731406b73244e0558ee7716d77c8096d150/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
++++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/f5f18731406b73244e0558ee7716d77c8096d150/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
++++ export CONDA_PREFIX=/fsx/nouamane/miniconda
++++ CONDA_PREFIX=/fsx/nouamane/miniconda
++++ export CONDA_SHLVL=2
++++ CONDA_SHLVL=2
++++ export CONDA_DEFAULT_ENV=base
++++ CONDA_DEFAULT_ENV=base
++++ export 'CONDA_PROMPT_MODIFIER=(base) '
++++ CONDA_PROMPT_MODIFIER='(base) '
++++ export CONDA_PREFIX_1=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ CONDA_PREFIX_1=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
++++ export _CE_M=
++++ _CE_M=
++++ export _CE_CONDA=
++++ _CE_CONDA=
++++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
++++ . /fsx/nouamane/miniconda/etc/conda/activate.d/libxml2_activate.sh
+++++ test -n ''
+++++ xml_catalog_files_libxml2=
+++++ XML_CATALOG_FILES=
+++++ conda_catalog_files=
+++++ ifs_libxml2=' 	
+'
+++++ IFS=' '
+++++ rem=/fsx/nouamane/miniconda
+++++ for pre in ${rem}
+++++ test '' = /fsx/nouamane/miniconda
+++++ conda_catalog_files=/fsx/nouamane/miniconda
+++++ rem=
+++++ IFS=' 	
+'
+++++ conda_catalog_files='file:///fsx/nouamane/miniconda/etc/xml/catalog file:///etc/xml/catalog'
+++++ export 'XML_CATALOG_FILES=file:///fsx/nouamane/miniconda/etc/xml/catalog file:///etc/xml/catalog'
+++++ XML_CATALOG_FILES='file:///fsx/nouamane/miniconda/etc/xml/catalog file:///etc/xml/catalog'
+++++ unset conda_catalog_files ifs_libxml2 rem
+++ __conda_hashr
+++ '[' -n '' ']'
+++ '[' -n '' ']'
+++ hash -r
++ conda activate 2-1-cu121
++ local cmd=activate
++ case "$cmd" in
++ __conda_activate activate 2-1-cu121
++ '[' -n '' ']'
++ local ask_conda
+++ PS1='(base) '
+++ __conda_exe shell.posix activate 2-1-cu121
+++ /fsx/nouamane/miniconda/bin/conda shell.posix activate 2-1-cu121
+Error while loading conda entry point: conda-libmamba-solver (module 'libmambapy' has no attribute 'QueryFormat')
++ ask_conda='. "/fsx/nouamane/miniconda/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/f5f18731406b73244e0558ee7716d77c8096d150/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_2='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
++ eval '. "/fsx/nouamane/miniconda/etc/conda/deactivate.d/libxml2_deactivate.sh"
+PS1='\''(2-1-cu121) '\''
+export PATH='\''/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/f5f18731406b73244e0558ee7716d77c8096d150/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin'\''
+export CONDA_PREFIX='\''/fsx/nouamane/miniconda/envs/2-1-cu121'\''
+export CONDA_SHLVL='\''3'\''
+export CONDA_DEFAULT_ENV='\''2-1-cu121'\''
+export CONDA_PROMPT_MODIFIER='\''(2-1-cu121) '\''
+export CONDA_PREFIX_2='\''/fsx/nouamane/miniconda'\''
+export CONDA_EXE='\''/fsx/nouamane/miniconda/bin/conda'\''
+export _CE_M='\'''\''
+export _CE_CONDA='\'''\''
+export CONDA_PYTHON_EXE='\''/fsx/nouamane/miniconda/bin/python'\''
+. "/fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh"'
+++ . /fsx/nouamane/miniconda/etc/conda/deactivate.d/libxml2_deactivate.sh
++++ test -n ''
++++ unset XML_CATALOG_FILES
++++ unset xml_catalog_files_libxml2
+++ PS1='(2-1-cu121) '
+++ export PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/f5f18731406b73244e0558ee7716d77c8096d150/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
+++ PATH=/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/f5f18731406b73244e0558ee7716d77c8096d150/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
+++ export CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ CONDA_PREFIX=/fsx/nouamane/miniconda/envs/2-1-cu121
+++ export CONDA_SHLVL=3
+++ CONDA_SHLVL=3
+++ export CONDA_DEFAULT_ENV=2-1-cu121
+++ CONDA_DEFAULT_ENV=2-1-cu121
+++ export 'CONDA_PROMPT_MODIFIER=(2-1-cu121) '
+++ CONDA_PROMPT_MODIFIER='(2-1-cu121) '
+++ export CONDA_PREFIX_2=/fsx/nouamane/miniconda
+++ CONDA_PREFIX_2=/fsx/nouamane/miniconda
+++ export CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ CONDA_EXE=/fsx/nouamane/miniconda/bin/conda
+++ export _CE_M=
+++ _CE_M=
+++ export _CE_CONDA=
+++ _CE_CONDA=
+++ export CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ CONDA_PYTHON_EXE=/fsx/nouamane/miniconda/bin/python
+++ . /fsx/nouamane/miniconda/envs/2-1-cu121/etc/conda/activate.d/libxml2_activate.sh
++++ test -n ''
++++ xml_catalog_files_libxml2=
++++ XML_CATALOG_FILES=
++++ conda_catalog_files=
++++ ifs_libxml2=' 	
+'
++++ IFS=' '
++++ rem=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ for pre in ${rem}
++++ test '' = /fsx/nouamane/miniconda/envs/2-1-cu121
++++ conda_catalog_files=/fsx/nouamane/miniconda/envs/2-1-cu121
++++ rem=
++++ IFS=' 	
+'
++++ conda_catalog_files='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ export 'XML_CATALOG_FILES=file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ XML_CATALOG_FILES='file:///fsx/nouamane/miniconda/envs/2-1-cu121/etc/xml/catalog file:///etc/xml/catalog'
++++ unset conda_catalog_files ifs_libxml2 rem
++ __conda_hashr
++ '[' -n '' ']'
++ '[' -n '' ']'
++ hash -r
++ export PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/f5f18731406b73244e0558ee7716d77c8096d150/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
++ PATH=/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/usr/local/cuda-12.1/efa/test-cuda-12.1:/usr/local/cuda-12.1/bin:/fsx/nouamane/miniconda/envs/2-1-cu121/bin:/fsx/nouamane/miniconda/condabin:/admin/home/nouamane/.cursor-server/bin/f5f18731406b73244e0558ee7716d77c8096d150/bin/remote-cli:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin:/opt/slurm/bin:/admin/home/nouamane/.local/bin:/admin/home/nouamane/.fzf/bin
++ '[' -z '' ']'
+++ scontrol show hostnames 'ip-26-0-171-[102,249],ip-26-0-172-116,ip-26-0-173-121'
++ export 'NODELIST=ip-26-0-171-102
+ip-26-0-171-249
+ip-26-0-172-116
+ip-26-0-173-121'
++ NODELIST='ip-26-0-171-102
+ip-26-0-171-249
+ip-26-0-172-116
+ip-26-0-173-121'
+++ head -n1
+++ scontrol show hostnames 'ip-26-0-171-[102,249],ip-26-0-172-116,ip-26-0-173-121'
++ export MASTER_NODE=ip-26-0-171-102
++ MASTER_NODE=ip-26-0-171-102
++ export MASTER_PORT=12356
++ MASTER_PORT=12356
++ export NNODES=4
++ NNODES=4
++ export GPUS_PER_NODE=8
++ GPUS_PER_NODE=8
++ export WORLD_SIZE=32
++ WORLD_SIZE=32
++ export CUDA_DEVICE_MAX_CONNECTIONS=1
++ CUDA_DEVICE_MAX_CONNECTIONS=1
++ export NCCL_DEBUG=WARN
++ NCCL_DEBUG=WARN
++ export NANOTRON_BENCHMARK=1
++ NANOTRON_BENCHMARK=1
++ export WANDB_MODE=disabled
++ WANDB_MODE=disabled
++ export TORCH_NCCL_ASYNC_ERROR_HANDLING=1
++ TORCH_NCCL_ASYNC_ERROR_HANDLING=1
++ export TORCH_DISTRIBUTED_DEBUG=DETAIL
++ TORCH_DISTRIBUTED_DEBUG=DETAIL
++ '[' -z '' ']'
++ echo '=== GPU Topology ==='
+=== GPU Topology ===
++ nvidia-smi topo -m
+	GPU0	GPU1	GPU2	GPU3	GPU4	GPU5	GPU6	GPU7	CPU Affinity	NUMA Affinity	GPU NUMA ID
+GPU0	 X 	NV18	NV18	NV18	NV18	NV18	NV18	NV18	0-43	0		N/A
+GPU1	NV18	 X 	NV18	NV18	NV18	NV18	NV18	NV18	0-43	0		N/A
+GPU2	NV18	NV18	 X 	NV18	NV18	NV18	NV18	NV18	0-43	0		N/A
+GPU3	NV18	NV18	NV18	 X 	NV18	NV18	NV18	NV18	0-43	0		N/A
+GPU4	NV18	NV18	NV18	NV18	 X 	NV18	NV18	NV18	52-95	1		N/A
+GPU5	NV18	NV18	NV18	NV18	NV18	 X 	NV18	NV18	52-95	1		N/A
+GPU6	NV18	NV18	NV18	NV18	NV18	NV18	 X 	NV18	52-95	1		N/A
+GPU7	NV18	NV18	NV18	NV18	NV18	NV18	NV18	 X 	52-95	1		N/A
+
+Legend:
+
+  X    = Self
+  SYS  = Connection traversing PCIe as well as the SMP interconnect between NUMA nodes (e.g., QPI/UPI)
+  NODE = Connection traversing PCIe as well as the interconnect between PCIe Host Bridges within a NUMA node
+  PHB  = Connection traversing PCIe as well as a PCIe Host Bridge (typically the CPU)
+  PXB  = Connection traversing multiple PCIe bridges (without traversing the PCIe Host Bridge)
+  PIX  = Connection traversing at most a single PCIe bridge
+  NV#  = Connection traversing a bonded set of # NVLinks
++ echo ==================
+==================
++ export SRUN_ALLOC_ARGS=
++ SRUN_ALLOC_ARGS=
++ echo 'Master node: ip-26-0-171-102'
+Master node: ip-26-0-171-102
++ echo 'All nodes: ip-26-0-171-102
+ip-26-0-171-249
+ip-26-0-172-116
+ip-26-0-173-121'
+All nodes: ip-26-0-171-102
+ip-26-0-171-249
+ip-26-0-172-116
+ip-26-0-173-121
++ echo 'World size: 32'
+World size: 32
++ '[' -n '' ']'
++ srun --wait=0 --kill-on-bad-exit=1 torchrun --nnodes=4 --nproc_per_node=8 --rdzv_id=14599514 --rdzv_backend=c10d --rdzv_endpoint=ip-26-0-171-102:12356 --max_restarts 0 --rdzv_conf timeout=60 /fsx/nouamane/projects/nanotron/run_train.py --config-file benchmark/configs/config_3.57G_dp1_tp1_pp32_acc31_mbs3_seq4096_zero0_tpmodeRED_vocab131k_cache.yaml
+[2025-02-08 15:01:34,862] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-02-08 15:01:34,861] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-02-08 15:01:34,874] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-02-08 15:01:34,884] torch.distributed.run: [WARNING] master_addr is only used for static rdzv_backend and when rdzv_endpoint is not specified.
+[2025-02-08 15:01:34,884] torch.distributed.run: [WARNING] 
+[2025-02-08 15:01:34,884] torch.distributed.run: [WARNING] *****************************************
+[2025-02-08 15:01:34,884] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-02-08 15:01:34,884] torch.distributed.run: [WARNING] *****************************************
+[2025-02-08 15:01:34,862] torch.distributed.run: [WARNING] 
+[2025-02-08 15:01:34,862] torch.distributed.run: [WARNING] *****************************************
+[2025-02-08 15:01:34,862] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-02-08 15:01:34,862] torch.distributed.run: [WARNING] *****************************************
+[2025-02-08 15:01:34,862] torch.distributed.run: [WARNING] 
+[2025-02-08 15:01:34,862] torch.distributed.run: [WARNING] *****************************************
+[2025-02-08 15:01:34,862] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-02-08 15:01:34,862] torch.distributed.run: [WARNING] *****************************************
+[2025-02-08 15:01:34,875] torch.distributed.run: [WARNING] 
+[2025-02-08 15:01:34,875] torch.distributed.run: [WARNING] *****************************************
+[2025-02-08 15:01:34,875] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to avoid your system being overloaded, please further tune the variable for optimal performance in your application as needed. 
+[2025-02-08 15:01:34,875] torch.distributed.run: [WARNING] *****************************************
+NCCL version 2.18.5+cuda12.2
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]: Config:
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]: Config(general=GeneralArgs(project='debug',
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                            run='3.57G_dp1_tp1_pp32_acc31_mbs3_seq4096_zero0_tpmodeRED_vocab131k_cache',
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                            seed=42,
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                            step=None,
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                            consumed_train_samples=None,
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                            benchmark_csv_path=PosixPath('benchmark/results/bench_final2.csv'),
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                            ignore_sanity_checks=True),
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:        parallelism=ParallelismArgs(dp=1,
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                                    pp=32,
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                                    tp=1,
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                                    pp_engine=<nanotron.parallel.pipeline_parallel.engine.OneForwardOneBackwardPipelineEngine object at 0x7fdba55fc970>,
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                                    tp_mode=<TensorParallelLinearMode.REDUCE_SCATTER: 2>,
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                                    tp_linear_async_communication=True,
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                                    recompute_layer=False,
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                                    tp_recompute_allgather=True,
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                                    expert_parallel_size=1),
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:        model=ModelArgs(model_config=LlamaConfig(bos_token_id=0,
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                                                 eos_token_id=0,
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                                                 hidden_act='silu',
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                                                 hidden_size=3072,
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                                                 initializer_range=0.02,
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                                                 intermediate_size=8192,
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                                                 is_llama_config=True,
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                                                 max_position_embeddings=4096,
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                                                 num_attention_heads=32,
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                                                 num_hidden_layers=28,
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                                                 num_key_value_heads=32,
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                                                 pad_token_id=None,
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                                                 pretraining_tp=1,
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                                                 rms_norm_eps=1e-05,
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                                                 rope_scaling=None,
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                                                 rope_theta=10000.0,
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                                                 rope_interleaved=False,
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                                                 tie_word_embeddings=True,
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                                                 use_cache=True,
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                                                 vocab_size=131072),
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                        init_method=RandomInit(std=0.02),
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                        dtype=torch.bfloat16,
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                        make_vocab_size_divisible_by=1,
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                        ddp_bucket_cap_mb=25),
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:        tokenizer=TokenizerArgs(tokenizer_name_or_path='robot-test/dummy-tokenizer-wordlevel',
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                                tokenizer_revision=None,
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                                tokenizer_max_length=None),
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:        checkpoints=CheckpointsArgs(checkpoints_path=PosixPath('checkpoints'),
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                                    checkpoint_interval=10000,
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                                    save_initial_state=False,
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                                    save_final_state=False,
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                                    resume_checkpoint_path=None,
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                                    checkpoints_path_is_shared_file_system=False),
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:        logging=LoggingArgs(log_level='info',
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                            log_level_replica='info',
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                            iteration_step_info_interval=1),
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:        tokens=TokensArgs(sequence_length=4096,
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                          train_steps=100,
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                          micro_batch_size=3,
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                          batch_accumulation_per_replica=31,
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                          val_check_interval=100,
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                          limit_val_batches=0,
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                          limit_test_batches=0),
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:        optimizer=OptimizerArgs(optimizer_factory=AdamWOptimizerArgs(adam_eps=1e-08,
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                                                                     adam_beta1=0.9,
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                                                                     adam_beta2=0.95,
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                                                                     torch_adam_is_fused=True,
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                                                                     name='adamW'),
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                                zero_stage=0,
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                                weight_decay=0.01,
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                                clip_grad=1.0,
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                                accumulate_grad_in_fp32=True,
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                                learning_rate_scheduler=LRSchedulerArgs(learning_rate=0.0003,
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                                                                        lr_warmup_steps=2,
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                                                                        lr_warmup_style='linear',
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                                                                        lr_decay_style='cosine',
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                                                                        lr_decay_steps=13,
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                                                                        lr_decay_starting_step=None,
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                                                                        min_decay_lr=1e-05)),
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:        data_stages=[DatasetStageArgs(name='Stable Training Stage',
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                                      start_training_step=1,
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                                      data=DataArgs(dataset=None,
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                                                    seed=42,
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:                                                    num_loading_workers=1))],
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:        profiler=None,
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:        lighteval=None,
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:        s3_upload=None)
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]: Model Config:
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]: LlamaConfig(bos_token_id=0,
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:             eos_token_id=0,
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:             hidden_act='silu',
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:             hidden_size=3072,
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:             initializer_range=0.02,
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:             intermediate_size=8192,
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:             is_llama_config=True,
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:             max_position_embeddings=4096,
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:             num_attention_heads=32,
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:             num_hidden_layers=28,
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:             num_key_value_heads=32,
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:             pad_token_id=None,
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:             pretraining_tp=1,
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:             rms_norm_eps=1e-05,
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:             rope_scaling=None,
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:             rope_theta=10000.0,
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:             rope_interleaved=False,
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:             tie_word_embeddings=True,
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:             use_cache=True,
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:             vocab_size=131072)
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]: Building model..
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]: Initialize RoPE Theta = 10000.0
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]: Setting PP block ranks...
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]: module_name: model.token_position_embeddings | PP: 0/32 | Block rank: 0
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]: module_name: model.decoder.0 | PP: 0/32 | Block rank: 1
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]: module_name: model.decoder.1 | PP: 1/32 | Block rank: 0
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]: module_name: model.decoder.2 | PP: 2/32 | Block rank: 0
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]: module_name: model.decoder.3 | PP: 3/32 | Block rank: 0
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]: module_name: model.decoder.4 | PP: 4/32 | Block rank: 0
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]: module_name: model.decoder.5 | PP: 5/32 | Block rank: 0
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]: module_name: model.decoder.6 | PP: 6/32 | Block rank: 0
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]: module_name: model.decoder.7 | PP: 7/32 | Block rank: 0
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]: module_name: model.decoder.8 | PP: 8/32 | Block rank: 0
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]: module_name: model.decoder.9 | PP: 9/32 | Block rank: 0
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]: module_name: model.decoder.10 | PP: 10/32 | Block rank: 0
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]: module_name: model.decoder.11 | PP: 11/32 | Block rank: 0
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]: module_name: model.decoder.12 | PP: 12/32 | Block rank: 0
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]: module_name: model.decoder.13 | PP: 13/32 | Block rank: 0
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]: module_name: model.decoder.14 | PP: 14/32 | Block rank: 0
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]: module_name: model.decoder.15 | PP: 15/32 | Block rank: 0
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]: module_name: model.decoder.16 | PP: 16/32 | Block rank: 0
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]: module_name: model.decoder.17 | PP: 17/32 | Block rank: 0
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]: module_name: model.decoder.18 | PP: 18/32 | Block rank: 0
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]: module_name: model.decoder.19 | PP: 19/32 | Block rank: 0
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]: module_name: model.decoder.20 | PP: 20/32 | Block rank: 0
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]: module_name: model.decoder.21 | PP: 21/32 | Block rank: 0
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]: module_name: model.decoder.22 | PP: 22/32 | Block rank: 0
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]: module_name: model.decoder.23 | PP: 23/32 | Block rank: 0
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]: module_name: model.decoder.24 | PP: 24/32 | Block rank: 0
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]: module_name: model.decoder.25 | PP: 25/32 | Block rank: 0
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]: module_name: model.decoder.26 | PP: 26/32 | Block rank: 0
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]: module_name: model.decoder.27 | PP: 27/32 | Block rank: 0
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]: module_name: model.final_layer_norm | PP: 28/32 | Block rank: 0
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]: module_name: model.lm_head | PP: 28/32 | Block rank: 1
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]: module_name: model.cast_to_fp32 | PP: 29/32 | Block rank: 0
+02/08/2025 15:01:59 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]: module_name: loss | PP: 30/32 | Block rank: 0
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+NCCL version 2.18.5+cuda12.2
+02/08/2025 15:02:06 [INFO|DP=0|PP=15|TP=0|ip-26-0-171-249]: Local number of parameters: 113M (216.01MiB)
+02/08/2025 15:02:06 [INFO|DP=0|PP=13|TP=0|ip-26-0-171-249]: Local number of parameters: 113M (216.01MiB)
+02/08/2025 15:02:06 [INFO|DP=0|PP=12|TP=0|ip-26-0-171-249]: Local number of parameters: 113M (216.01MiB)
+02/08/2025 15:02:06 [INFO|DP=0|PP=31|TP=0|ip-26-0-173-121]: Local number of parameters: 0 (0.00MiB)
+02/08/2025 15:02:06 [INFO|DP=0|PP=29|TP=0|ip-26-0-173-121]: Local number of parameters: 0 (0.00MiB)
+02/08/2025 15:02:06 [INFO|DP=0|PP=9|TP=0|ip-26-0-171-249]: Local number of parameters: 113M (216.01MiB)
+02/08/2025 15:02:06 [INFO|DP=0|PP=14|TP=0|ip-26-0-171-249]: Local number of parameters: 113M (216.01MiB)
+02/08/2025 15:02:06 [INFO|DP=0|PP=10|TP=0|ip-26-0-171-249]: Local number of parameters: 113M (216.01MiB)
+02/08/2025 15:02:06 [INFO|DP=0|PP=24|TP=0|ip-26-0-173-121]: Local number of parameters: 113M (216.01MiB)
+02/08/2025 15:02:06 [INFO|DP=0|PP=30|TP=0|ip-26-0-173-121]: Local number of parameters: 0 (0.00MiB)
+02/08/2025 15:02:06 [INFO|DP=0|PP=27|TP=0|ip-26-0-173-121]: Local number of parameters: 113M (216.01MiB)
+02/08/2025 15:02:06 [INFO|DP=0|PP=8|TP=0|ip-26-0-171-249]: Local number of parameters: 113M (216.01MiB)
+02/08/2025 15:02:06 [INFO|DP=0|PP=26|TP=0|ip-26-0-173-121]: Local number of parameters: 113M (216.01MiB)
+02/08/2025 15:02:06 [INFO|DP=0|PP=25|TP=0|ip-26-0-173-121]: Local number of parameters: 113M (216.01MiB)
+02/08/2025 15:02:06 [INFO|DP=0|PP=28|TP=0|ip-26-0-173-121]: Local number of parameters: 403M (768.01MiB)
+02/08/2025 15:02:06 [INFO|DP=0|PP=11|TP=0|ip-26-0-171-249]: Local number of parameters: 113M (216.01MiB)
+02/08/2025 15:02:06 [INFO|DP=0|PP=13|TP=0|ip-26-0-171-249]: [After model building] Memory usage: 216.02MiB. Peak allocated: 216.03MiB Peak reserved: 218.00MiB
+02/08/2025 15:02:06 [INFO|DP=0|PP=12|TP=0|ip-26-0-171-249]: [After model building] Memory usage: 216.02MiB. Peak allocated: 216.03MiB Peak reserved: 218.00MiB
+02/08/2025 15:02:06 [INFO|DP=0|PP=15|TP=0|ip-26-0-171-249]: [After model building] Memory usage: 216.02MiB. Peak allocated: 216.03MiB Peak reserved: 218.00MiB
+02/08/2025 15:02:06 [INFO|DP=0|PP=9|TP=0|ip-26-0-171-249]: [After model building] Memory usage: 216.02MiB. Peak allocated: 216.03MiB Peak reserved: 218.00MiB
+02/08/2025 15:02:06 [INFO|DP=0|PP=31|TP=0|ip-26-0-173-121]: [After model building] Memory usage: 0.01MiB. Peak allocated: 0.02MiB Peak reserved: 2.00MiB
+02/08/2025 15:02:06 [INFO|DP=0|PP=14|TP=0|ip-26-0-171-249]: [After model building] Memory usage: 216.02MiB. Peak allocated: 216.03MiB Peak reserved: 218.00MiB
+02/08/2025 15:02:06 [INFO|DP=0|PP=24|TP=0|ip-26-0-173-121]: [After model building] Memory usage: 216.02MiB. Peak allocated: 216.03MiB Peak reserved: 218.00MiB
+02/08/2025 15:02:06 [INFO|DP=0|PP=30|TP=0|ip-26-0-173-121]: [After model building] Memory usage: 0.01MiB. Peak allocated: 0.02MiB Peak reserved: 2.00MiB
+02/08/2025 15:02:06 [INFO|DP=0|PP=27|TP=0|ip-26-0-173-121]: [After model building] Memory usage: 216.02MiB. Peak allocated: 216.03MiB Peak reserved: 218.00MiB
+02/08/2025 15:02:06 [INFO|DP=0|PP=29|TP=0|ip-26-0-173-121]: [After model building] Memory usage: 0.01MiB. Peak allocated: 0.02MiB Peak reserved: 2.00MiB
+02/08/2025 15:02:06 [INFO|DP=0|PP=10|TP=0|ip-26-0-171-249]: [After model building] Memory usage: 216.02MiB. Peak allocated: 216.03MiB Peak reserved: 218.00MiB
+02/08/2025 15:02:06 [INFO|DP=0|PP=26|TP=0|ip-26-0-173-121]: [After model building] Memory usage: 216.02MiB. Peak allocated: 216.03MiB Peak reserved: 218.00MiB
+02/08/2025 15:02:06 [INFO|DP=0|PP=8|TP=0|ip-26-0-171-249]: [After model building] Memory usage: 216.02MiB. Peak allocated: 216.03MiB Peak reserved: 218.00MiB
+02/08/2025 15:02:06 [INFO|DP=0|PP=25|TP=0|ip-26-0-173-121]: [After model building] Memory usage: 216.02MiB. Peak allocated: 216.03MiB Peak reserved: 218.00MiB
+02/08/2025 15:02:06 [INFO|DP=0|PP=28|TP=0|ip-26-0-173-121]: [After model building] Memory usage: 768.02MiB. Peak allocated: 768.03MiB Peak reserved: 770.00MiB
+02/08/2025 15:02:06 [INFO|DP=0|PP=22|TP=0|ip-26-0-172-116]: Local number of parameters: 113M (216.01MiB)
+02/08/2025 15:02:06 [INFO|DP=0|PP=17|TP=0|ip-26-0-172-116]: Local number of parameters: 113M (216.01MiB)
+02/08/2025 15:02:06 [INFO|DP=0|PP=4|TP=0|ip-26-0-171-102]: Local number of parameters: 113M (216.01MiB)
+02/08/2025 15:02:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]: Total number of parameters: 3.98G (7584.33MiB)
+02/08/2025 15:02:06 [INFO|DP=0|PP=6|TP=0|ip-26-0-171-102]: Local number of parameters: 113M (216.01MiB)
+02/08/2025 15:02:06 [INFO|DP=0|PP=16|TP=0|ip-26-0-172-116]: Local number of parameters: 113M (216.01MiB)
+02/08/2025 15:02:06 [INFO|DP=0|PP=2|TP=0|ip-26-0-171-102]: Local number of parameters: 113M (216.01MiB)
+02/08/2025 15:02:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]: Local number of parameters: 516M (984.01MiB)
+02/08/2025 15:02:06 [INFO|DP=0|PP=7|TP=0|ip-26-0-171-102]: Local number of parameters: 113M (216.01MiB)
+02/08/2025 15:02:06 [INFO|DP=0|PP=3|TP=0|ip-26-0-171-102]: Local number of parameters: 113M (216.01MiB)
+02/08/2025 15:02:06 [INFO|DP=0|PP=5|TP=0|ip-26-0-171-102]: Local number of parameters: 113M (216.01MiB)
+02/08/2025 15:02:06 [INFO|DP=0|PP=18|TP=0|ip-26-0-172-116]: Local number of parameters: 113M (216.01MiB)
+02/08/2025 15:02:06 [INFO|DP=0|PP=20|TP=0|ip-26-0-172-116]: Local number of parameters: 113M (216.01MiB)
+02/08/2025 15:02:06 [INFO|DP=0|PP=11|TP=0|ip-26-0-171-249]: [After model building] Memory usage: 216.02MiB. Peak allocated: 216.03MiB Peak reserved: 218.00MiB
+02/08/2025 15:02:06 [INFO|DP=0|PP=21|TP=0|ip-26-0-172-116]: Local number of parameters: 113M (216.01MiB)
+02/08/2025 15:02:06 [INFO|DP=0|PP=19|TP=0|ip-26-0-172-116]: Local number of parameters: 113M (216.01MiB)
+02/08/2025 15:02:06 [INFO|DP=0|PP=1|TP=0|ip-26-0-171-102]: Local number of parameters: 113M (216.01MiB)
+02/08/2025 15:02:06 [INFO|DP=0|PP=4|TP=0|ip-26-0-171-102]: [After model building] Memory usage: 216.02MiB. Peak allocated: 216.03MiB Peak reserved: 218.00MiB
+02/08/2025 15:02:06 [INFO|DP=0|PP=22|TP=0|ip-26-0-172-116]: [After model building] Memory usage: 216.02MiB. Peak allocated: 216.03MiB Peak reserved: 218.00MiB
+02/08/2025 15:02:06 [INFO|DP=0|PP=17|TP=0|ip-26-0-172-116]: [After model building] Memory usage: 216.02MiB. Peak allocated: 216.03MiB Peak reserved: 218.00MiB
+02/08/2025 15:02:06 [INFO|DP=0|PP=6|TP=0|ip-26-0-171-102]: [After model building] Memory usage: 216.02MiB. Peak allocated: 216.03MiB Peak reserved: 218.00MiB
+02/08/2025 15:02:06 [INFO|DP=0|PP=2|TP=0|ip-26-0-171-102]: [After model building] Memory usage: 216.02MiB. Peak allocated: 216.03MiB Peak reserved: 218.00MiB
+02/08/2025 15:02:06 [INFO|DP=0|PP=16|TP=0|ip-26-0-172-116]: [After model building] Memory usage: 216.02MiB. Peak allocated: 216.03MiB Peak reserved: 218.00MiB
+02/08/2025 15:02:06 [INFO|DP=0|PP=3|TP=0|ip-26-0-171-102]: [After model building] Memory usage: 216.02MiB. Peak allocated: 216.03MiB Peak reserved: 218.00MiB
+02/08/2025 15:02:06 [INFO|DP=0|PP=18|TP=0|ip-26-0-172-116]: [After model building] Memory usage: 216.02MiB. Peak allocated: 216.03MiB Peak reserved: 218.00MiB
+02/08/2025 15:02:06 [INFO|DP=0|PP=5|TP=0|ip-26-0-171-102]: [After model building] Memory usage: 216.02MiB. Peak allocated: 216.03MiB Peak reserved: 218.00MiB
+02/08/2025 15:02:06 [INFO|DP=0|PP=7|TP=0|ip-26-0-171-102]: [After model building] Memory usage: 216.02MiB. Peak allocated: 216.03MiB Peak reserved: 218.00MiB
+02/08/2025 15:02:06 [INFO|DP=0|PP=20|TP=0|ip-26-0-172-116]: [After model building] Memory usage: 216.02MiB. Peak allocated: 216.03MiB Peak reserved: 218.00MiB
+02/08/2025 15:02:06 [INFO|DP=0|PP=21|TP=0|ip-26-0-172-116]: [After model building] Memory usage: 216.02MiB. Peak allocated: 216.03MiB Peak reserved: 218.00MiB
+02/08/2025 15:02:06 [INFO|DP=0|PP=23|TP=0|ip-26-0-172-116]: Local number of parameters: 113M (216.01MiB)
+02/08/2025 15:02:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]: [After model building] Memory usage: 984.02MiB. Peak allocated: 984.03MiB Peak reserved: 986.00MiB
+02/08/2025 15:02:06 [INFO|DP=0|PP=19|TP=0|ip-26-0-172-116]: [After model building] Memory usage: 216.02MiB. Peak allocated: 216.03MiB Peak reserved: 218.00MiB
+02/08/2025 15:02:06 [INFO|DP=0|PP=1|TP=0|ip-26-0-171-102]: [After model building] Memory usage: 216.02MiB. Peak allocated: 216.03MiB Peak reserved: 218.00MiB
+02/08/2025 15:02:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]: No checkpoint path provided.
+02/08/2025 15:02:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]: Parametrizing model parameters using StandardParametrizator
+02/08/2025 15:02:06 [INFO|DP=0|PP=23|TP=0|ip-26-0-172-116]: [After model building] Memory usage: 216.02MiB. Peak allocated: 216.03MiB Peak reserved: 218.00MiB
+02/08/2025 15:02:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]: [Optimizer Building] Using LearningRateForSP as learning rate
+02/08/2025 15:02:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]: Before merge_named_param_groups
+02/08/2025 15:02:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:  Memory usage: 984.02MiB. Peak allocated 984.03MiB. Peak reserved: 986.00MiB
+02/08/2025 15:02:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]: After merge_named_param_groups
+02/08/2025 15:02:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:  Memory usage: 984.02MiB. Peak allocated 984.02MiB. Peak reserved: 986.00MiB
+02/08/2025 15:02:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]: Before fp32_grad_optimizer_builder
+02/08/2025 15:02:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:  Memory usage: 984.02MiB. Peak allocated 984.02MiB. Peak reserved: 986.00MiB
+02/08/2025 15:02:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]: Before basic_optimizer_builder
+02/08/2025 15:02:06 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:  Memory usage: 4920.07MiB. Peak allocated 4920.07MiB. Peak reserved: 4926.00MiB
+02/08/2025 15:02:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]: After basic_optimizer_builder
+02/08/2025 15:02:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:  Memory usage: 4920.07MiB. Peak allocated 4920.07MiB. Peak reserved: 4926.00MiB
+02/08/2025 15:02:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]: After fp32_grad_optimizer_builder
+02/08/2025 15:02:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:  Memory usage: 4920.07MiB. Peak allocated 4920.07MiB. Peak reserved: 4926.00MiB
+02/08/2025 15:02:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]: [Training Plan] Stage Stable Training Stage has 99 remaining training steps and has consumed 0 samples
+02/08/2025 15:02:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]: Using dummy data generator
+02/08/2025 15:02:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]: [Training Plan] There are 1 training stages 
+02/08/2025 15:02:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]: [Stage Stable Training Stage] start from step 1 
+02/08/2025 15:02:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]: 
+02/08/2025 15:02:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]: [Start training] datetime: 2025-02-08 15:02:07.120368 | mbs: 3 | grad_accum: 31 | global_batch_size: 93 | sequence_length: 4096 | train_steps: 100 | start_iteration_step: 0 | consumed_train_samples: 0
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+02/08/2025 15:02:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]: Resuming training from stage Stable Training Stage, it has trained for 0 samples and has 99 remaining train steps
+02/08/2025 15:02:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]: Before TBI
+02/08/2025 15:02:07 [INFO|DP=0|PP=0|TP=0|ip-26-0-171-102]:  Memory usage: 4920.07MiB. Peak allocated 4920.07MiB. Peak reserved: 4926.00MiB
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+[W ProcessGroupNCCL.cpp:1856] Warning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (function operator())
+/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/__init__.py:251: UserWarning: 0NCCL_AVOID_RECORD_STREAMS=1 has no effect for point-to-point collectives. (Triggered internally at ../torch/csrc/distributed/c10d/ProcessGroupNCCL.cpp:1856.)
+  Variable._execution_engine.run_backward(  # Calls into the C++ engine to run the backward pass
+Traceback (most recent call last):
+  File "/fsx/nouamane/projects/nanotron/run_train.py", line 254, in <module>
+    trainer.train(dataloader)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 466, in train
+    outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 508, in training_step
+    outputs = self.pipeline_engine.train_batch_iter(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 301, in train_batch_iter
+    output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 58, in forward
+    output = model(**micro_batch)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1016, in forward
+    loss = self.loss(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 160, in forward
+    output = self.pp_block(**new_kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 969, in forward
+    loss = sharded_cross_entropy(
+  File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 122, in sharded_cross_entropy
+    return _ShardedCrossEntropy.apply(sharded_logits, target, group)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+    return super().apply(*args, **kwargs)  # type: ignore[misc]
+RuntimeError: Output 0 of RecvTensorFromPipelineBufferBackward is a view and its base or another view of its base has been modified inplace. This view was created inside a custom Function (or because an input was returned as-is) and the autograd logic to handle view+inplace would override the custom backward associated with the custom Function, leading to incorrect gradients. This behavior is forbidden. You can fix this by cloning the output of the custom Function.
+[2025-02-08 15:02:52,213] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3995316 closing signal SIGTERM
+[2025-02-08 15:02:52,213] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3995317 closing signal SIGTERM
+[2025-02-08 15:02:52,213] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3995318 closing signal SIGTERM
+[2025-02-08 15:02:52,213] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3995319 closing signal SIGTERM
+[2025-02-08 15:02:52,213] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3995320 closing signal SIGTERM
+[2025-02-08 15:02:52,213] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3995321 closing signal SIGTERM
+[2025-02-08 15:02:52,214] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3995323 closing signal SIGTERM
+[2025-02-08 15:03:09,597] torch.distributed.elastic.multiprocessing.api: [ERROR] failed (exitcode: 1) local_rank: 6 (pid: 3995322) of binary: /fsx/nouamane/miniconda/envs/2-1-cu121/bin/python
+[2025-02-08 15:03:09,634] torch.distributed.elastic.multiprocessing.errors.error_handler: [ERROR] no error file defined for parent, to copy child error file (/tmp/torchelastic_u9jjqgf7/14599514_e29xjrrq/attempt_0/6/error.json)
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 264, in launch_agent
+    raise ChildFailedError(
+torch.distributed.elastic.multiprocessing.errors.ChildFailedError: 
+============================================================
+/fsx/nouamane/projects/nanotron/run_train.py FAILED
+------------------------------------------------------------
+Failures:
+  <NO_OTHER_FAILURES>
+------------------------------------------------------------
+Root Cause (first observed failure):
+[0]:
+  time      : 2025-02-08_15:02:40
+  host      : ip-26-0-173-121.ec2.internal
+  rank      : 30 (local_rank: 6)
+  exitcode  : 1 (pid: 3995322)
+  error_file: /tmp/torchelastic_u9jjqgf7/14599514_e29xjrrq/attempt_0/6/error.json
+  traceback : Traceback (most recent call last):
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+      return f(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 466, in train
+      outputs, loss_avg = self.training_step(dataloader=self.current_dataloader)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/trainer.py", line 508, in training_step
+      outputs = self.pipeline_engine.train_batch_iter(
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 301, in train_batch_iter
+      output = self.forward(context=context, state=state, micro_batch=micro_batch, model=model)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/engine.py", line 58, in forward
+      output = model(**micro_batch)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 1016, in forward
+      loss = self.loss(
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/pipeline_parallel/block.py", line 160, in forward
+      output = self.pp_block(**new_kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1518, in _wrapped_call_impl
+      return self._call_impl(*args, **kwargs)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1527, in _call_impl
+      return forward_call(*args, **kwargs)
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/models/llama.py", line 969, in forward
+      loss = sharded_cross_entropy(
+    File "/fsx/nouamane/projects/nanotron/src/nanotron/parallel/tensor_parallel/functional.py", line 122, in sharded_cross_entropy
+      return _ShardedCrossEntropy.apply(sharded_logits, target, group)
+    File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/autograd/function.py", line 539, in apply
+      return super().apply(*args, **kwargs)  # type: ignore[misc]
+  RuntimeError: Output 0 of RecvTensorFromPipelineBufferBackward is a view and its base or another view of its base has been modified inplace. This view was created inside a custom Function (or because an input was returned as-is) and the autograd logic to handle view+inplace would override the custom backward associated with the custom Function, leading to incorrect gradients. This behavior is forbidden. You can fix this by cloning the output of the custom Function.
+  
+============================================================
+srun: error: ip-26-0-173-121: task 3: Exited with exit code 1
+srun: launch/slurm: _step_signal: Terminating StepId=14599514.0
+slurmstepd: error: *** STEP 14599514.0 ON ip-26-0-171-102 CANCELLED AT 2025-02-08T15:03:10 ***
+[2025-02-08 15:03:10,964] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-02-08 15:03:10,965] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 182227 closing signal SIGTERM
+[2025-02-08 15:03:10,965] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 182228 closing signal SIGTERM
+[2025-02-08 15:03:10,965] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 182229 closing signal SIGTERM
+[2025-02-08 15:03:10,965] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 182230 closing signal SIGTERM
+[2025-02-08 15:03:10,965] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 182231 closing signal SIGTERM
+[2025-02-08 15:03:10,965] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 182232 closing signal SIGTERM
+[2025-02-08 15:03:10,965] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 182233 closing signal SIGTERM
+[2025-02-08 15:03:10,965] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 182234 closing signal SIGTERM
+[2025-02-08 15:03:10,964] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-02-08 15:03:10,966] torch.distributed.elastic.agent.server.api: [WARNING] Received Signals.SIGTERM death signal, shutting down workers
+[2025-02-08 15:03:10,966] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3667827 closing signal SIGTERM
+[2025-02-08 15:03:10,966] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3667828 closing signal SIGTERM
+[2025-02-08 15:03:10,967] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3667829 closing signal SIGTERM
+[2025-02-08 15:03:10,967] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3667830 closing signal SIGTERM
+[2025-02-08 15:03:10,969] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3667831 closing signal SIGTERM
+[2025-02-08 15:03:10,969] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3667832 closing signal SIGTERM
+[2025-02-08 15:03:10,969] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3667833 closing signal SIGTERM
+[2025-02-08 15:03:10,969] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 3667834 closing signal SIGTERM
+[2025-02-08 15:03:10,964] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1207087 closing signal SIGTERM
+[2025-02-08 15:03:10,964] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1207088 closing signal SIGTERM
+[2025-02-08 15:03:10,964] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1207089 closing signal SIGTERM
+[2025-02-08 15:03:10,964] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1207090 closing signal SIGTERM
+[2025-02-08 15:03:10,964] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1207091 closing signal SIGTERM
+[2025-02-08 15:03:10,965] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1207092 closing signal SIGTERM
+[2025-02-08 15:03:10,965] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1207093 closing signal SIGTERM
+[2025-02-08 15:03:10,966] torch.distributed.elastic.multiprocessing.api: [WARNING] Sending process 1207094 closing signal SIGTERM
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 182154 got signal: 15
+srun: error: ip-26-0-171-102: task 0: Exited with exit code 1
+[2025-02-08 15:03:32,167] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-172-116.ec2.internal_3667755_0' has failed to send a keep-alive heartbeat to the rendezvous '14599514' due to an error of type RendezvousConnectionError.
+[2025-02-08 15:03:32,197] torch.distributed.elastic.rendezvous.dynamic_rendezvous: [WARNING] The node 'ip-26-0-171-249.ec2.internal_1207016_0' has failed to send a keep-alive heartbeat to the rendezvous '14599514' due to an error of type RendezvousConnectionError.
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 1207016 got signal: 15
+srun: error: ip-26-0-171-249: task 1: Exited with exit code 1
+Traceback (most recent call last):
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/bin/torchrun", line 8, in <module>
+    sys.exit(main())
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
+    return f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 806, in main
+    run(args)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/run.py", line 797, in run
+    elastic_launch(
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
+    return launch_agent(self._config, self._entrypoint, list(args))
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 255, in launch_agent
+    result = agent.run()
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/metrics/api.py", line 124, in wrapper
+    result = f(*args, **kwargs)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 736, in run
+    result = self._invoke_run(role)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/agent/server/api.py", line 877, in _invoke_run
+    time.sleep(monitor_interval)
+  File "/fsx/nouamane/miniconda/envs/2-1-cu121/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py", line 62, in _terminate_process_handler
+    raise SignalException(f"Process {os.getpid()} got signal: {sigval}", sigval=sigval)
+torch.distributed.elastic.multiprocessing.api.SignalException: Process 3667755 got signal: 15
+srun: error: ip-26-0-172-116: task 2: Exited with exit code 1